- Email: [email protected]
Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability
Article
Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability Graphical Abstract
Authors
nl U GDP
[email protected]
g
Transitioning
GDP
GDP
GTP
GTP Maintenance of Inhibition
Inhibited GATOR1 FLCNFNIP2 GTP GDP
GDP
Maintenance of Activation
GDP
g
GDP
nl
in
Transitioning
U
ck
Activated
oc
Lo
ki
ng
GTP
Kuang Shen, Abigail Choe, David M. Sabatini
Correspondence
in
oc
ki
RagC
ck
RagA
GDP
Lo
ng
GDP
Amino acid deficiency
GDP
Amino acid sufficiency
Highlights d
The Rag GTPase dimer uses a ‘‘locking mechanism’’ to regulate its nucleotide state
d
The binding of GTP to one subunit prevents GTP from associating with the other
d
Unexpected GTP binding in the locked conformation leads to stimulation of hydrolysis
d
The locking mechanism is necessary for the normal response of mTORC1 to amino acids
Shen et al., 2017, Molecular Cell 68, 1–14 November 2, 2017 ª 2017 Elsevier Inc. https://doi.org/10.1016/j.molcel.2017.09.026
In Brief The Rag GTPase heterodimer transmits amino acid signals to mTORC1. It is distinct from canonical small signaling GTPases because of its obligate heterodimeric nature. Shen et al. used the kinetic method to analyze communication between Rag GTPase subunits, revealing a unique intersubunit locking mechanism that controls its nucleotide state.
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Molecular Cell
Article Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability Kuang Shen,1,2,3,4 Abigail Choe,1 and David M. Sabatini1,2,3,4,5,* 1Whitehead Institute for Biomedical Research and Massachusetts Institute of Technology, Department of Biology, 9 Cambridge Center, Cambridge, MA 02142, USA 2Howard Hughes Medical Institute, Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 3Koch Institute for Integrative Cancer Research, 77 Massachusetts Avenue, Cambridge, MA 02139, USA 4Broad Institute of Harvard and Massachusetts Institute of Technology, 7 Cambridge Center, Cambridge, MA 02142, USA 5Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.molcel.2017.09.026
SUMMARY
mTOR complex I (mTORC1) is a central growth regulator that senses amino acids through a pathway that converges on the Rag GTPases, an obligate heterodimer of two related GTPases. Despite their central role in amino acid sensing, it is unknown why the Rag GTPases are heterodimeric and whether their subunits communicate with each other. Here, we find that the binding of guanosine triphosphate (GTP) to one subunit inhibits the binding and induces the hydrolysis of GTP by the other. This intersubunit communication pushes the Rag GTPases into either of two stable configurations, which represent active ‘‘on’’ or ‘‘off’’ states that interconvert via transient intermediates. Subunit coupling confers on the mTORC1 pathway its capacity to respond rapidly to the amino acid level. Thus, the dynamic response of mTORC1 requires intersubunit communication by the Rag GTPases, providing a rationale for why they exist as a dimer and revealing a distinct mode of control for a GTP-binding protein.
INTRODUCTION mTOR complex I (mTORC1) is a central regulator of cell growth and proliferation. Diverse inputs, including amino acids, growth factors, and energy levels, converge on mTORC1 to regulate its kinase activity and signal to downstream effectors (Efeyan et al., 2015; Gonza´lez and Hall, 2017; Jewell et al., 2013; Shaw and Cantley, 2006). Upon activation, mTORC1 triggers anabolic processes, such as protein synthesis and ribosomal biogenesis, and inhibits catabolic ones, such as autophagy (Dibble and Manning, 2013). Dysregulation of mTORC1 promotes aberrant growth and contributes to numerous human diseases, including cancer and neurological disorders like epilepsy (Wong, 2013). mTORC1 activation requires two steps controlled by distinct
small GTPases: translocation of mTORC1 to the lysosomal surface and stimulation of its kinase activity. In the presence of amino acids, the Rag GTPases recruit mTORC1 to the lysosome (Sancak et al., 2008; Wang et al., 2005), where, if growth factors are present, the Rheb GTPase turns on its kinase activity (Inoki et al., 2003; Li et al., 2004; Menon et al., 2014; Saito et al., 2005; Saucedo et al., 2003; Stocker et al., 2003). Thus, these two GTPases form an AND gate that ensures that mTORC1 is active only under favorable growth conditions. Within the Ras-like superfamily of GTPases, the Rag GTPases belong to the TRAFAC (translation factor-related) class due to a conserved threonine residue on the lateral side of the nucleotide-binding pocket (Thr35 in Ras; Thr42 and Thr96 in RagA and RagC, respectively) (Leipe et al., 2002; Shan, 2016). However, unlike other characterized members of the superfamily, the Rag GTPases form obligate heterodimers consisting of RagA or RagB bound to RagC or RagD, and so each functional unit can bind two guanine nucleotides (Nakashima et al., 1999; €rmann et al., 1995). Distinct from the signal recognition parSchu ticle (SRP) and its receptor (SR), which toggle between monomeric and dimeric forms during the protein-targeting cycle, there is no evidence that the Rag GTPase heterodimer ever dissociates into stable monomers within cells. Several multi-protein complexes regulate the Rag GTPases, including Ragulator, which localizes them to the lysosomal surface and modulates their nucleotide loading (Sancak et al., 2010), and the GATOR1 (Bar-Peled et al., 2013; Panchaud et al., 2013) and FLCN-FNIP2 complexes (Petit et al., 2013; Tsun et al., 2013), which are GTPase-activating proteins (GAPs) for RagA/B and RagC/D, respectively. Previous studies in cells found that the Rag heterodimer is active and capable of interacting with mTORC1 when RagA/B is bound to guanosine triphosphate (GTP) and RagC/D to guanosine diphosphate (GDP) (GTPRagA-RagCGDP). In contrast, the opposite nucleotide configuration (GDPRagA-RagCGTP) weakens the interaction with mTORC1 so that it does not localize to lysosomes and the pathway is suppressed (Sancak et al., 2008). It is unclear whether, out of the four possible nucleotide states, these are the only two in which the Rag GTPase heterodimer can exist. Moreover, in general, why the Rag GTPases are heterodimeric and whether intersubunit communication occurs and, if so, Molecular Cell 68, 1–14, November 2, 2017 ª 2017 Elsevier Inc. 1
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
RagC
GNP
GNP
Purification Strategy
C
pET Duet-1 (AmpR) ORF1: RagA-His; ORF2: RagC
Nucleotide binding domain (NBD)
Q Sepharose
C-terminal roadblock domain (CRD)
D
RagA-RagC
150
80− MW (kD)
B RagA
OD280 (mAu)
A
Nucleotides
100
Ni-NTA
50
monoQ
[Rag] MW (kD)
F
α-32P-GNP RagA-RagC (trace) (varying amount)
Binding (au)
UV light (254 nm)
Crosslinking assay
chilled surface
50 100 Elution Volume (ml)
46−
RagC×GTP
32−
RagA×GTP
G
46−
RagC×GDP
32−
RagA×GDP
4 105
RagC RagA
4 105
2 105
150
[Rag] MW (kD)
0
Binding (au)
Equilibrium binding assay
RagC RagA
2 105
GTP binding 0
0.01
Autoradiography imaging
0.01
RagA-RagC (μM)
I
RagC×GTP
100 RagA×GTP
varying time
1
106
0
200
0
50
0.10
0.05
0.00
100
Time (min)
K
400 [RagA-RagC] (nM)
1.0
kobsd (min-1)
Binding (au)
RagC×GTP
106
5.0 105
RagA×GTP
0.0
50
RagA RagC
0.2
0.1
0.0
100
P
Time MW (kD)
GDP association 0
46−
RagC×GTP
32−
RagA×GTP
Unlabeled GNP (excess)
1
Crosslinking assay
2 105 RagC RagA
1 105
0
GTP dissociation 0
5000 Time (min)
Summary of binding kinetics (4°C)
-1
-1
kon (M s ) Kd (nM) koff, calc (s-1) † -1
R
GTP
GDP
3
3
GDP GTP 3.5×10 (.2) 5.1×10 (.2) 4.1×103(.2) 2.9×103(.3) 47(12)
76(7)
11(5)
41(6)
1.6×10-4
3.9×10-4
4.5×10-5
1.2×10-4
RagC×GDP
32−
RagA×GDP
105 RagC RagA
5 104
GDP dissociation 0
50
100
-4
-7
150
Time (min)
Free energy profile for GDP binding 19
kon 0 -2
-7
46−
S
RagC
RagA
400
Time
0
10000
ΔG (kcal/mol)
Q
Binding (au)
Binding (au)
varying time
200
RagA-RagC (nM)
-4
koff, measured (s ) 1.2×10 (.6) 5.3×10 (.3) 9.8×10 (.6) 5.6×10 (.4) 1300×↓ 46×↓
koff
Kd empty
Free energy profile for GTP binding
bound
?
22 19
kon 0 -2
koff
Kd empty
intermediate
bound
GTP
GDP GDP
RagA RagC
ΔG (kcal/mol)
RagA-RagC
O
α-32P-GNP (trace)
MW (kD)
Off-rate measurement
400
RagA-RagC (nM)
Time (min)
N
200
200 nM
400 nM 300 nM 100 nM RagA-RagC
0
GTP association 0
M
1.5 106 RagC×GTP
300 RagA×GTP
RagA RagC
200 nM
400 nM 300 nM 100 nM RagA-RagC
RagA×GTP RagC×GTP
10
0.15
L
RagA×GTP
2 106
RagC×GTP
Crosslinking assay
1
RagA-RagC (μM)
J
Time
α-32P-GNP (trace)
0.1
kobsd (min-1)
On-rate measurement RagA-RagC (fixed amount)
GDP binding 0
1
Binding (au)
H
0.1
RagA
25−
0
E
RagC 46− 32−
Void
Gel filtration
58−
GTP
GTP
?
(legend on next page)
2 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
how it interfaces with the actions of the extrinsic regulators are all open questions. Here, we used kinetic analyses to quantitatively dissect the functional cycle of the Rag GTPases. We find that the Rag heterodimer uses a previously undescribed mode of regulation for GTPases—extensive intersubunit communication—to transmit upstream signals to mTORC1. This mechanism has two key properties: the binding of GTP to one of the Rag subunits inhibits the association of GTP to the other, and if an unwanted binding event does occur, the prebound GTP triggers the hydrolysis of the later-bound GTP, which maintains the original state by destabilizing the GTPRagA-RagCGTP configuration. We find that intersubunit communication is critical for the mTORC1 pathway to rapidly respond to the presence and absence of amino acids, providing a possible explanation for the evolutionary advantage of the heterodimeric architecture of the Rag GTPases. RESULTS Binding of GTP to the Rag GTPase Heterodimer Induces Conformational Changes that Inhibit Its Dissociation To examine potential communication between the subunits of the Rag GTPase heterodimer, it was necessary to develop an assay in which the nucleotide state of each could be quantitatively assessed in the context of the intact complex. Each Rag subunit has an N-terminal nucleotide-binding domain (NBD) with a similar structural fold to Ras and a C-terminal roadblock domain (CRD) that mediates heterodimerization (Figure 1A) (De Franceschi et al., 2014; Gong et al., 2011; Jeong et al., 2012; Levine et al., 2013). Expression of individual subunits in bacteria did not yield soluble proteins (data not shown), probably due to the exposure of hydrophobic residues on the CRD that mediates heterodimer formation. In contrast, co-expression of codon-
optimized, C-terminally His-tagged RagA and tagless RagC in the pETDuet-1 vector (Figure 1B) generated a stable complex that could be isolated at the high purity needed for biochemical studies (Figures 1C and 1D). The fact that RagA and RagC have substantially different molecular weights and that UV light can covalently crosslink nucleotides to GTPases (Yue and Schimmel, 1977) enabled us to develop an assay for measuring, within intact complexes, the binding of radiolabeled nucleotides to each subunit at the same time (Figure 1E; STAR Methods). At the end of the binding reaction, we expose samples to UV light and use SDS-PAGE followed by autoradiography to quantify the amounts of radiolabeled nucleotides bound to each Rag subunit (Figure 1F). As expected, unlabeled GTP competed with the binding of radiolabeled guanine nucleotides to RagA and RagC (Figure S1A), and in heterodimers in which RagA or RagC had a mutation that mimics the Ras(S17N) mutant that does not bind GTP (RagA T21N; RagC S75N) (Feig and Cooper, 1988; Sancak et al., 2008), GTP bound only to the wild-type subunit (Figures S1B and S1C). Within the heterodimer, RagC binds GTP with a Kd of 11 nM, 4-fold tighter than RagA, and both bind GDP slightly less well than GTP (Figures 1F and 1G; summarized in Figure 1Q). These Kd values are orders of magnitudes lower than the cellular concentration of GTP (0.7 mM), suggesting that mechanisms must exist to prevent both subunits from being loaded with GTP as the active heterodimer is in the GTPRagA-RagCGDP state. To understand nucleotide binding in more detail, we examined pre-steady-state kinetics by measuring the nucleotide association and dissociation rates of the Rag GTPases (Fersht, 1999; Feuerstein et al., 1987; Nowak and Goody, 1988; Simon et al., 1996). For the on-rate measurement (Figure 1H), we monitored nucleotide association to Rag subunits over time (Figures 1I
Figure 1. Quantitative Measures of the Nucleotide-Binding Properties of the Rag GTPases (A) General architecture and domain organization of the Rag GTPase heterodimer. GNP, guanosine di/triphosphate. (B) Generation of recombinant Rag GTPases in bacteria. (C) Elution profile of the Rag GTPase heterodimer on a HiLoad 16/60 Superdex 200 gel-filtration column. After EDTA stripping, the Rag heterodimer separates from its bound nucleotides and emerges as a single peak as indicated. (D) SDS-PAGE and Coomassie blue staining analysis of purified Rags. (E) Experimental setup of the crosslinking assay for measuring the nucleotide binding affinity of each Rag subunit. (F and G) GTP (F) and GDP (G) binding to the Rag GTPases. Top: nucleotide bound to RagA (lower bands, molecular weight [MW] = 37 kDa) and RagC (upper bands, MW = 44 kDa) visualized by autoradiography. Bottom: quantification of the radioactive signals from the top panels. Data were fit to a single-site binding equation to obtain the dissociation constant Kd. (H) Experimental setup of the crosslinking assay for measuring the on rates of nucleotides to the Rag GTPases. (I) Reaction time course of GTP bound to Rag subunits during an on-rate measurement. Four Rag concentrations (100, 200, 300, and 400 nM) are shown from top to bottom, where time points were taken to monitor the reaction process. At higher Rag concentrations, reactions reach equilibrium faster. (J and K) Quantification of the observed rate constants for GTP association to RagA (J) and RagC (K) at different Rag concentrations. A single exponential rise function was used to fit the time points shown in (I). Curve colors match with the corresponding Rag concentrations in (I). (L and M) On rates for GTP (L) and GDP (M) to the Rag GTPase heterodimer. Linear regression of the observed rate constants (values obtained from J and K for GTP association and Figures S1E and S1F for GDP association) against Rag concentration gives the slope (kon) for RagA (red) and RagC (blue). (N) Experimental setup of the crosslinking assay for measuring the off rates of nucleotides from the Rag GTPases. (O and P) Reaction time course of radioactively labeled nucleotides remaining bound to Rag subunits during the off-rate measurement of GTP (O) and GDP (P) dissociation. Top: SDS-PAGE resolves nucleotide bound to RagA (lower bands) and RagC (upper bands) in the presence of unlabeled nucleotides as a chase. Bottom: quantification. Data were fit to a single exponential decay to extract the off rates (koff). (Q) Summary of binding kinetics of nucleotides to the Rag GTPases at 4 C. Gray numbers in parenthesis denote the SDs of the reported values calculated from at least three independent experiments. (R and S) Free energy profiles and models for GDP (R) and GTP (S) binding to the Rag GTPases. In the case of GDP binding (R), the matched predicted and measured koff values suggest a single energy barrier. In the case of GTP binding (S), a second barrier is required to explain the discrepancy between the predicted and measured koff values.
Molecular Cell 68, 1–14, November 2, 2017 3
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
A
B
Rag mutants and properties
Binding preference RagC RagA RagA(D130N)-RagC Xanthine Guanine RagA-RagC(D181N) Guanine Xanthine Mutants
C +XTP
D
Sequential chase assay
[RagA(D130N)-RagC] RagC×GTP
RagA(D130N) -RagC
[RagA(D130N)-RagC]
α-32P-GTP (trace) XNP (excess)
RagC×GTP
+XDP
+XTP
RagC×GTP
+XDP
RagC×GTP
Pre-bound GTP
1 hour
Sequential binding assay 2 106
α-32P-GTP (trace) 1 hour
Crosslinking assay
+XTP +XDP
1 106 Binding (au)
Binding (au)
RagA(D130N) XNP -RagC (1.2 eq of Rag)
1 106
+XTP +XDP
5 105
Crosslinking assay 0
0
RagA-RagC
F
Cold GNP (1.2 eq of Rag)
Stabilize GNP
RagA-RagC
or RagA-RagC
α-32P-GNP (trace)
GNP
0.06
0.04
Varying time Crosslinking assay
J
Cold XNP RagA(D130N)(1.2 eq of Rag) RagC Stabilize XNP
RagA(D130N)-RagC
α-32P-GTP (trace)
0.08
0.04 0.02
GTP association 0
Crosslinking assay
M
GTP binding to RagA
Locked
Summary of half-site on-rates (4°C) RagA-RagCN Prebound N= GTP GDP
kon (M-1 s-1)
0.05
K
+XTP +XDP
0.06
0.00
Varying time
0.01 0.1 1 RagA(D130N)-RagC (μM)
N RagA-RagC GDP GTP
GTP 1.1×10-4 s-1 890(65) 1.1×10-4 s-1 460(35) GDP
580(67) 1100(70) 229(23)
980(99)
GDP association 1.0
2.0
RagA-RagC (μM)
kobsd (min-1)
Differential on-rate for GTP
kobsd (min-1)
I
0.10
0.00 0.0
0.5 1.0 RagA-RagC (μM)
H
RagA-RagCGDP GDP RagA-RagC RagA-RagCGTP GTP RagA-RagC
0.15
GTP association
0.02
0.00 0.0
G
RagA-RagCGDP GDP RagA-RagC RagA-RagCGTP GTP RagA-RagC
kobsd (min-1)
Half-site binding assay
kobsd (min-1)
E
0.01 0.1 1 RagA(D130N)-RagC (μM)
Summary of on-rates for GTP (4°C) Prebound N= XTP -1
-1
kon (M s )
0.02
0.00
1 2 3 RagA(D130N)-RagC (μM)
L
+XTP +XDP
0.04
XDP
RagA-RagC(D181N)N 103(13) 231(25) N
RagA(D130N)-RagC 309(13) 609(40)
GTP association 0
1 2 3 RagA-RagC(D181N) (μM)
Prevention of GTP binding to RagC
GTP GTP RagC
RagA
GDP
GDP
open
GTP
GDP
closed
GTP
GDP
closed
Figure 2. Intersubunit Crosstalk Maintains the Singly GTP-Loaded State of the Rag GTPase Heterodimer (A) Experimental setup for a sequential binding assay to probe the effect of xanthine nucleotide-loaded RagA on guanine nucleotide binding to RagC. Rag mutants with modified hydrogen-bonding pattern decouple nucleotide binding to the subunits. (B) XTP loading on RagA(D130N) inhibits GTP binding to RagC. Top: GTP binding to RagC when RagA(D130N) is prebound to XTP or XDP. Bottom: quantification. When RagA(D130N) binds XTP, RagC loads a significantly smaller amount of GTP than when it binds XDP. (C) Experimental setup of a sequential chase assay to probe the effect of later-bound nucleotides on the first-bound nucleotide. (D) Later-bound XTP to RagA(D130N) has no effect on the amount of GTP prebound to RagC. Top: prebound GTP to RagC chased by XTP or XDP binding on RagA(D130N). Bottom: quantification. Later-bound XTP or XDP on RagA(D130N) has no effect on the prebound GTP. (E) Experimental setup of a half-site binding assay to measure the on rates of nucleotides to singly loaded Rag GTPases. Unlabeled nucleotides were added at sub-equal molar amounts compared to the nucleotide-binding pockets to generate singly loaded Rag GTPases. (F) Half-site on-rate measurements for GTP association when the Rags are singly loaded with GTP (filled circle) or GDP (open circle). The subunit in bold is the subunit associated with the labeled GTP in the assay to measure the half-site on rate. For example, RagA-RagCGDP (open red circle) shows how fast RagA associates with labeled GTP when RagC is preloaded with GDP. The slope gives the half-site on rates. Notice that concentration dependence is lost in the case of GTP. (G) Half-site on-rate measurements for GDP association when the Rags are singly loaded with GTP (filled circle) or GDP (open circle). As before, subunits in bold indicates which subunit is binding the labeled GDP in the assay to measure the half-site on rate. GDP associates with the Rag GTPases more slowly if one subunit is preloaded with GTP. (legend continued on next page)
4 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
and S1D). Linear regression of the observed reaction rates (kobsd; Figures 1J and 1K for GTP and Figures S1E and S1F for GDP) against Rag concentrations generated the secondary reaction rate constants kon, which reveal that RagA and RagC associate with guanine nucleotides at very similar rates (Figure 1L for GTP and Figure 1M for GDP; summarized in Figure 1Q). We measured the off rate in a pulse-chase setup (Figure 1N), in which preloaded labeled GTP on the Rags was chased by an excess of unlabeled GTP (koff; Figure 1O for GTP and Figure 1P for GDP). Despite the similar binding affinities and on rates, GTP dissociates from the Rags much more slowly than GDP, with a 600- to 4,000-fold difference (Figure 1Q). Careful examination of the kinetics data revealed a surprising discrepancy (Figure 1Q). If the binding of nucleotides to the Rag GTPases involves a single transition state, then the dissociation constant, and the on and off rates, should conform to the equation Kd = kon/koff. This prediction holds true for GDP, where the calculated koff matches well with the measured koff (Figure 1Q), suggesting that GDP binding does indeed overcome one transition state (Figure 1R). However, the measured koff for GTP is 1,300- and 46-fold slower than the calculated one for RagA and RagC, respectively (Figure 1Q; GTP koff, calc versus koff, measured). This discrepancy suggests that the binding of GTP, but not GDP, induces conformational changes in the heterodimer beyond the local ones on switch I and switch II that are common to small GTP-binding proteins (Hall et al., 2002; Vetter et al., 1999) and that these prevent GTP dissociation (Figure 1S). To our knowledge this second transition state has not been observed in monomeric GTPases, suggesting that it is a result of the heterodimeric architecture of the Rag GTPases. Binding of a Single GTP to the Rag GTPase Heterodimer Inhibits Binding of a Second GTP To understand the consequences of GTP binding to the Rag heterodimer, it was necessary to characterize its singly GTP-loaded state in vitro by combining the crosslinking assay with Rag mutants that preferentially bind xanthine instead of guanine nucleotides (RagA D130N; RagC D181N) (Figure 2A) (Hwang and Miller, 1987; Sweet and Gerace, 1996). Surprisingly, when RagA(D130N) in the RagA(D130N)-RagC heterodimer was bound to xanthosine 50 -triphosphate (XTP) instead of xanthosine 50 -diphosphate (XDP), the wild-type RagC partner bound less GTP (Figure 2B). In the reverse direction, this observation also holds for the RagA-RagC(D181N) heterodimer (Figure S2A), suggesting that nucleoside triphosphate binding to one subunit (RagA or RagC) inhibits nucleoside triphosphate binding to the other (RagC or RagA). As a control experiment, we switched the sequence of binding (Figure 2C); once RagC was already
loaded with GTP, the subsequent binding of XTP or XDP to RagA(D130N) had no observable effect on the amount of preloaded GTP (Figure 2D), indicating that the binding of the first nucleoside triphosphate has dominant consequences. These findings suggest that the binding of GTP to either RagA or RagC could alter the conformation of the entire heterodimer to induce a ‘‘locked state’’ that suppresses the association of GTP to the other subunit (negative cooperativity). To verify this possibility, we directly measured the on rate of GTP to singly loaded wild-type Rag GTPase heterodimers using a half-site binding assay (Figures 2E–2H and S2B–S2E). We reasoned that if the binding of the first GTP inhibits the association of the second GTP, then we should observe a slower on rate. This is indeed the case: when RagC is bound to GDP, RagA associates GTP at 890 M1s1 (Figures 2F and 2H). When RagC is loaded with GTP, however, the on rate no longer exhibits concentration dependence (Figure 2F) and gives a first-order rate constant of 1.1 3 104s1. Similar results were obtained using the xanthine-specific mutants described above (Figures 2I–2L and S2F–S2I); when the xanthine-specific subunit prebinds XTP instead of XDP, the on rate of GTP to the guanine-specific subunit is slower (Figure 2J for RagC and Figure 2K for RagA; summarized in Figure 2L). These results are consistent with the Rag heterodimer being in a conformation that is not permissive to the direct binding of the substrate when it is singly loaded with GTP. In this locked state, the transient opening of the binding pocket becomes rate limiting, making the ‘‘on rate’’ first order and thus independent of Rag concentration (Figures S2J and S2K) (Fersht, 1999). We extended the half-site on-rate measurement to GDP association and found, to a lesser extent, a decrease in on rates when one subunit was preloaded with GTP (Figures 2G and 2H). However, the observed rate for GDP association is still dependent on the Rag concentration, likely because GDP is smaller so that even in the locked state the Rags cannot fully exclude it from the partially shielded nucleotide-binding site. To directly probe the conformations of the Rag GTPases in different nucleotide-loading states, we utilized a lysine modification assay (Figure S2L), where the extent of lysine modification by sulfo-N-hydroxysuccinimide (NHS)-acetate correlates with its solvent accessibility (Guo et al., 2008). When both subunits bind GDP, the Rag GTPases have a modification profile very similar to that of empty Rags (log2 ratio 0; Figure S2M). However, in the presence of 1:1 GTP/GDP, several lysine residues show a significantly lower level of modification (Figure S2M). For example, Lys20 of RagA and Lys74 of RagC, both localizing within the nucleotide-binding pocket, are much less modified than without GTP, consistent with a local conformational change
(H) Summary of half-site on rates for wild-type Rag GTPases at 4 C. Gray numbers in parentheses denote the SDs of the reported values calculated from at least three independent experiments. (I) Experimental setup for measuring the on rates of nucleotides to the guanine-specific subunit when the xanthine-specific subunit is preloaded. (J) On rate of GTP association to RagC when RagA(D130N) was preloaded with XTP (filled circle) or XDP (open circle). (K) On rate of GTP association to RagA when RagC(D181N) was preloaded with XTP (filled circle) or XDP (open circle). (L) Summary of differential on rates for GTP at 4 C. Gray numbers in parenthesis denote the SDs of the reported values calculated from at least three independent experiments. (M) Illustration of the locked conformation induced by the binding of the first GTP to RagA and its inhibition of the association of the second GTP with RagC. The reverse case is shown in Figure S2Q.
Molecular Cell 68, 1–14, November 2, 2017 5
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
KH2PO4
Quench
C
Time
Pi
GTP
0.15
D
20 nM RagA-RagC 10 nM 5 nM 2 nM 1 nM 0.5 nM
0.10 0.05
0.0003
kobsd (min-1)
Varying time
B
Fraction GTP hydrolyzed
Single turnover GTP hydrolysis
RagA-RagC γ-32P-GTP (varying amount) (trace)
Expansion direction
A
0.00 0
20 nM RagA-RagC
TLC
200 400 Time (min)
0.0002
0.0001
0.0000
600
0
5
10
15 20 RagA-RagC (nM)
100
1000
Autoradiography
F
G
Time
γ-32P-GTP (trace) + Cold GTP (varying amount)
RagA-RagC (fixed amount) Varying time
Pi
GTP
KH2PO4
0.10 0.05
2 μM GTP 5 μM 10 μM
0.00
2 μM GTP
Quench
H
0.15
0
500 1000 Time (min)
0.003
kobsd (min -1)
Multiple turnover GTP hydrolysis
Fraction GTP hydrolyzed
E
20 μM 50 μM 100 μM
0.002
0.001
0.000
1500
0
50 GTP (μM)
100
TLC
Autoradiography
J
RagA-RagC(S75N) RagA(T21N)-RagC WT
0.002
50 RagA-RagC (μM)
Half-site GTP hydrolysis
Cold GNP RagA-RagC (fixed amount) (1.2 eq of Rags) Stabilize
γ-32P-GTP (trace)
0.002
0.000
100
M
0.3
0
50 GTP (μM)
O
No nucleotides +GTP +GDP +GppNHp
0.2
400
Observed hydrolysis rates (per min) No NT +GTP +GDP +GppNHp
R
200 Time (min)
1.1×104(.3) 1.7×104(.4)
Half-site hydrolysis chase γ-32P-GTP (trace) Cold GNP (excess)
Initial
P
Chased
0.10
0.05
+Cold GNP
0.00 0
200 Time (min)
+Buffer +GTP +GDP +GppNHp
400
Observed chased rates (per min) +Buffer +GTP +GDP +GppNHp
Unexpected GTP binding to RagC
1.2×104(.2) 3000×↑
RagA(T21N)-RagC 1.9×10-3(.2) 5.2×10-3(.2)
Chased time points (- - -) Quench and TLC analysis
Q
2.2×10-4 1.0×10-3 2.7×10-4 7.7×10-4
4.0(.8)
Multiple Turnover
1.6×104(.5) 8.4×103(2)
Initial time points (- - -) Quench and TLC analysis 0
-3
2.2×10 (.2) 3.0×10 (.1) 14×↑
Start reaction
0.1
Single Turnover
RagA-RagC(S75N) 3.7×10-3(.4) 3.7×10-3(.4)
RagA-RagC (fixed amount)
0.0
Quench and TLC analysis
100
KM (nM)
Multiple Turnover
-4
Wildtype
Varying time
N
kcat (min-1)
Multiple Turnover
Fraction Pi
L
0
Summary of GTP hydrolysis kinetics (25°C)
Single Turnover
Single Turnover 0.000
K
RagA-RagC(S75N) RagA(T21N)-RagC WT
0.004
Fraction Pi
kobsd (min-1)
0.004
kobsd (min-1)
I
2.2×10-4 2.0×10-4 2.2×10-4 2.1×10-4
Stimulated hydrolysis
GTP GTP RagA
GDP
GTP
GT P
GTP
GDP
RagC
Figure 3. Directional GTP Hydrolysis Maintains the Functional State of the Rag GTPases (A and E) Experimental setup for the single- (A) and multiple-turnover (E) GTP hydrolysis assays. A saturating amount of the Rag GTPases was incubated with a trace amount of labeled GTP to ensure single-turnover conditions (A). A saturating amount of GTP was incubated with the Rag GTPases to ensure multipleturnover conditions (E). (B and F) Thin-layer chromatography to separate radioactive phosphate from unhydrolyzed GTP at different time points. 20 nM Rag GTPases hydrolyzed a trace amount of labeled GTP in a single-turnover assay (B). 2 mM Rag GTPases hydrolyzed 2 mM GTP in a multiple-turnover assay (F). (C and G) Quantification and linear regression of the fraction of hydrolyzed GTP yields the observed rate constants (kobsd) at different Rag (C) or GTP (G) concentrations. (D and H) Michaelis-Menten fittings of kobsd against Rag or GTP concentration yield kcat and KM for the single-turnover (D) and multiple-turnover (H) reactions. (I and J) Single-turnover (I) and multiple-turnover (J) GTP hydrolysis reactions for RagA-RagC(S75N) (red circle) and RagA(T21N)-RagC (blue circle). Dashed curves show the activity of wild-type Rags as a reference. (legend continued on next page)
6 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
upon GTP binding that is prevalent among Ras-family GTPases. Another region that displays lower levels of modification is Lys230 of RagA and Lys258 of RagC (Figures S2N and S2O). Based on a homologous structure (Figure S2P), these residues localize to the hinge region connecting the NBD and CRD (Gong et al., 2011; Jeong et al., 2012). The significant reduction in the modification of these residues indicates that the hinge region is more shielded than in the GDP-bound state, suggesting that GTP binding induces a more compact conformation. Thus, in aggregate, these data indicate that when one subunit binds GTP, the Rag heterodimer rearranges into a more compact (‘‘locked’’) conformation that can account for the second transition state on the free-energy plot (Figure 2M) and decouples kon and koff. Binding of a Single GTP to the Rag Heterodimer Induces Hydrolysis of a Second GTP Despite the existence of the locked state, we wondered what would happen if both Rag subunits did manage to bind GTP. For example, if in the active heterodimer (GTPRagA-RagCGDP) RagC loses its GDP and misloads with GTP, is there a mechanism to promote hydrolysis of the newly bound GTP, or do both subunits have an equal chance of hydrolyzing their GTPs? If the latter were the case, the heterodimer could flip from the active to inactive state, which might be undesirable. We monitored labeled GTP hydrolysis using thin-layer chromatography (TLC) in both single- and multiple-turnover setups (Figures 3A and 3E) (Peluso et al., 2001). In single-turnover assays, a trace amount of GTP is incubated with an excess of the Rag GTPases, so that only a single subunit in each heterodimer will have the opportunity to hydrolyze GTP once (Figure 3A). In multiple-turnover assays, the Rag GTPases are saturated with excess GTP, so that both subunits can hydrolyze GTP for multiple rounds (Figure 3E). These two setups yielded completely different kinetic parameters (Figures 3B–3D, 3F– 3H, S3B, and S3C; summarized in Figure 3K). The Rag GTPases hydrolyzed GTP 14-fold faster in the multiple-turnover assay than in the single-turnover assay (Figure 3K, kcat), suggesting that the GTPRagA-RagCGTP state is unstable. Moreover, the KM for GTP was also 3,000-fold higher in the multiple-turnover assay than in the single-turnover assay (Figure 3K), showing that the association of the second GTP is kinetically unfavorable, as predicted by the locking model proposed above. In the same assays, heterodimers containing RagA or RagC mutants with
impaired GTP binding (cf. Figures S1B and S1C) showed no differences between single- and multiple-turnover assays (Figures 3I–3K and S3D–S3G). We also probed how the order of GTP binding to the Rag heterodimer impacts hydrolysis by its subunits. We first checked the hydrolysis of the later bound GTP using a half-site hydrolysis setup (Figure 3L), where we loaded half of the nucleotide-binding pockets in the Rag heterodimer with unlabeled GTP or GDP, so that labeled GTP can only occupy and be hydrolyzed by the remaining site. When the Rag GTPases were singly loaded with GTP or GppNHp, the hydrolysis rate on the other subunit was 4-fold faster than when loaded with GDP (Figures 3M and 3N). To measure the hydrolysis rate of the prebound GTP, we switched the reaction order so that labeled GTP was first loaded on the Rag GTPases, and then an excess of unlabeled GTP or GDP was added as a chase to occupy the remaining site (Figure 3O). In this scenario, no acceleration in the reaction rate was observed regardless of the chase. The prebound GTP was hydrolyzed at a constant, slow speed, similar to that in a single-turnover setup (Figures 3P and 3Q), suggesting the later-loaded GTP has no effect on the prebound GTP. Thus, the prebound GTP hydrolyzes itself at a slow rate while causing accelerated hydrolysis at the other nucleotide-binding pocket. Combining these results, we conclude that a single GTP-binding event on one subunit is necessary and sufficient to drive the heterodimer into a locked state that inhibits the association and promotes the hydrolysis of GTP by the other subunit (Figure 3R). Thus, ‘‘intersubunit communication’’ maintains the Rag heterodimer in its active or inactive state, ensuring the stability of each and pointing to the necessity of regulators to interconvert between them. Rag GAPs Interconvert between the Functional States of the Rag GTPase Heterodimer The consequences of intersubunit communication and the slow intrinsic GTP dissociation and hydrolysis rates of the Rags should make transitions between the two stable states intrinsically challenging. These findings clearly point to a key role for extrinsic regulators in modulating the Rag GTPases in response to nutrients. Indeed, two GAPs have been discovered for the Rags: GATOR1 (Bar-Peled et al., 2013; Panchaud et al., 2013) and FLCN-FNIP2 (Petit et al., 2013; Tsun et al., 2013) for RagA/B and RagC/D, respectively (Figure 4A). To test the hypothesis
(K) Summary of GTP hydrolysis kinetics for the Rag GTPases at 25 C. Gray numbers in parenthesis denote the SDs of the reported values calculated from at least three independent experiments. (L) Experimental setup for a half-site GTP hydrolysis reaction. (M) Half-site GTP hydrolysis in the presence of GTP (red), GDP (blue), or GppNHp (orange). Radioactively labeled GTP is hydrolyzed on one subunit of the Rag GTPases when the other is preloaded with nucleotides. Preloaded GTP and GppNHp trigger hydrolysis by the other subunit. Control (black) contains no preloaded nucleotides. (N) Summary of the observed rate constants for half-site GTP hydrolysis reactions in (M). (O) Experimental setup for a half-site GTP hydrolysis chase. Labeled GTP was preloaded and hydrolyzed on one subunit of the Rag GTPases before an excess amount of cold nucleotides was added to occupy the remaining binding sites. (P) Half-site GTP hydrolysis chased by GTP (red), GDP (blue), or GppNHp (orange). No difference in the slope was observed. Pure buffer was added to the control (black). (Q) Summary of the observed rate constants for half-site GTP hydrolysis chase in (P). (R) Illustration of the effect of the locked conformation on GTP hydrolysis. GTP binding to RagA triggers GTP hydrolysis by RagC. The reverse case is shown in Figure S3H.
Molecular Cell 68, 1–14, November 2, 2017 7
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
B
46−
Nprl2
FLCN
46− 32−
0.2 0.1
RagA-RagC
0.0 0
25−
Rags only 10 nM GATOR1 20 nM 50 nM 100 nM 250 nM
0.1
0
200 Time (min)
Rags only 100 nM FLCN-FNIP2 200 nM 500 nM 1 μM 2 μM
F 0.4
0.2
G
RagA-RagC(Q120L)
0.0 0
200 Time (min)
0.000
400
400
H
RagA-RagC RagA-RagC(Q120L) RagA(Q66L)-RagC RagA(Q66L)-RagC(Q120L)
0.005
0
100 200 GATOR1 (nM)
0.4
0.2 0.1 0.0 0.0
0.2
RagA-RagC
200 Time (min)
100 Single Turnover 10-2
400
+FLCN-FNIP2 +GATOR1 No GAP
10-4 RagA- RagAQ66L- RagA- RagAQ66LRagC RagC RagCQ120LRagCQ120L
300
Multiple Turnover
Single Turnover +FLCN-FNIP2
0.3
0.4
0
Single Turnover +GATOR1
0.010
RagA(Q66L)-RagC
0.0
Fraction Pi
kobsd (min-1)
0.2
E
kobsd (min-1)
Fraction GTP hydrolyzed
0.3
Rags only 10 nM FLCN-FNIP2 20 nM 50 nM 100 nM 200 nM 500 nM
0.0
200 400 Time (min)
32−
D
Fraction Pi
Nprl3
58−
FNIP2
kcat (min-1)
MW (kD)
80−
135− 100− 80− 58−
Fraction GTP hydrolyzed
Depdc5
190− 135− 100−
0.6
C
Rags only 10 nM GATOR1 20 nM 50 nM 100 nM 250 nM
0.3
kcat (min-1)
A
10-1
+FLCN-FNIP2 +GATOR1 No GAP
10-2 10-3
RagA- RagAQ66L- RagA- RagAQ66LRagC RagC RagCQ120LRagCQ120L
RagA-RagC RagA-RagC(Q120L) RagA(Q66L)-RagC RagA(Q66L)-RagC(Q120L)
0.5 1.0 1.5 FLCN-FNIP2 (μM)
2.0
Figure 4. Regulators of the Rag GTPase Heterodimer Control Its Nucleotide-Loading State (A) SDS-PAGE followed by Coomassie blue staining of recombinant GATOR1 and FLCN-FNIP2 complexes indicating their subunits. (B and C) Single-turnover reactions of GATOR1- (B) and FLCN-FNIP-stimulated (C) GTP hydrolysis of wild-type Rags. GATOR1 and FLCN-FNIP dose-dependently increase GTP hydrolysis by wild-type Rags. (D) GATOR1-stimulated GTP hydrolysis of RagA(Q66L)-RagC mutant in a single-turnover condition. The leucine substitution in RagA abolishes the stimulation by GATOR1. (E) Concentration dependence of GATOR1-stimulated GTP hydrolysis under single-turnover conditions for wild-type Rags (black), RagA(Q66L)-RagC (blue), RagA-RagC(Q120L) (orange), and RagA(Q66L)-RagC(Q120L) (red). (F) FLCN-FNIP2-stimulated GTP hydrolysis of RagA-RagC(Q120L) mutant in a single-turnover condition. The leucine substitution in RagC abolishes the stimulation by FLCN-FNIP2. (G) Concentration dependence of FLCN-FNIP2-stimulated GTP hydrolysis under single-turnover conditions for wild-type Rags (black), RagA(Q66L)-RagC (blue), RagA-RagC(Q120L) (orange), and RagA(Q66L)-RagC(Q120L) (red). (H) Summary of hydrolysis kinetics for GATOR1- or FLCN-FNIP2-stimulated reactions.
that these GAPs accelerate state conversion of the Rag heterodimer to match the temporal requirements of mTORC1 signaling, we measured the kinetic parameters for GAP-stimulated GTP hydrolysis reactions. For wild-type Rags, GATOR1 and FLCN-FNIP2 stimulate GTP hydrolysis in a dose-dependent manner (Figures 4B and 4C; summarized in Figure 4K). The rate enhancement of GATOR1 or FLCN-FNIP2 is 45-fold (Figure 4B) and 1,000-fold (Figure 4C) in a single-turnover reaction and 20-fold (Figure S4E) and 50-fold (Figure S4M) in a multipleturnover reaction, respectively. The stimulated hydrolysis is much faster than the intrinsic hydrolysis of the Rag GTPases, which brings the half-life of the activated state from over 8 hr down to 12 min (37 C) and of the inactive state to 30 s. These rates are consistent with the timescales at which mTORC1 signaling responds to amino acids. Therefore, we concluded that GATOR1 and FLCN-FNIP2 resolve the locked
8 Molecular Cell 68, 1–14, November 2, 2017
states (activated or inactive), facilitating the Rags for stateconversion. In Ras-like GTPases, a glutamine residue (Gln61 in Ras) serves as the general base for GTP hydrolysis and chelates the nucleophilic-attacking water to stabilize the transition state (Pai et al., 1990; Scheffzek et al., 1997). We analyzed the corresponding glutamine mutants of the Rag GTPases in the GAP-stimulated GTP hydrolysis reaction. As expected, GATOR1 acts specifically on RagA, because RagA(Q66L)-RagC (Figures 4D and 4E), but not RagA-RagC(Q120L) (Figure S4C), loses GATOR1-dependent stimulation. In contrast, RagA-RagC(Q120L) (Figure 4F and 4G), but not RagA(Q66L)-RagC (Figure S4J), no longer responds to FLCN-FNIP2. These results suggest that the conserved glutamine residues strongly participate in GAP-stimulated GTP hydrolysis and that the two GAPs likely reposition this residue on the corresponding Rag to facilitate GTP hydrolysis.
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Point Mutations in Switch I of the Rag GTPases Disrupt Intersubunit Communication As discussed above, intersubunit communication by the Rag heterodimer provides a mechanism for maintaining its subunits loaded with opposite nucleotides. To understand the biological relevance of this mechanism, we sought mutants that disrupt intersubunit communication. By systematically characterizing the biochemical properties of recurring RagC mutations found in follicular lymphoma, we identified two interesting mutants, RagC(L91P) and RagC(K84T), in switch I (Figure 5A) (Long et al., 2016; Okosun et al., 2016). The affinity of guanine nucleotides to these mutants closely resembles that of the wild-type proteins (Figures 5B, 5D, and S5A–S5E), suggesting that they maintain an intact nucleotide-binding pocket. However, in the RagA-RagC(L91P) heterodimer, the off rate of GTP for RagC increases by 10-fold, while that for GDP remains similar (Figures 5C, 5D, and S5F). These results suggest that the binding of GTP to RagC(L91P) does not drive the Rag heterodimer into the locked conformation that should have inhibited its dissociation. If this were true, the previously observed properties of the locked state should diminish: preloaded GTP should not inhibit the second GTP from associating and, if it does, trigger its hydrolysis. We used the half-site setup to measure the on rate of the second GTP (Figure 2E). For wild-type Rags with normal intersubunit communication, the second GTP association does not have concentration dependence (Figure 5E, dashed line; cf. Figure 2F), but for the RagA-RagC(L91P) mutant, the concentration dependence is partially restored for both subunits (Figures 5E and 5G), suggesting that the first GTP binding does not effectively drive the heterodimer containing this mutant into the closed conformation. Moreover, the rate enhancement in a multiple-turnover GTP hydrolysis reaction (Figure 5I), but not in a single-turnover reaction (Figure 5H), is partially suppressed for RagA-RagC(L91P), suggesting that the prebound GTP does not efficiently trigger hydrolysis of the later-bound GTP. In addition, the half-site hydrolysis stimulation by GTP is impaired for this mutant (Figures S5M and S5N). Therefore, we conclude that the L91P mutation in RagC partially disrupts intersubunit communication (Figure 5K). Tests of the analogous mutant in RagA, RagA(R37P), and of the RagC(K84T) mutant, led to similar conclusions (Figures 5D–5J), suggesting that mutations in either Rag subunit can disrupt the locking process. Rag Mutants Deficient in Intersubunit Communication Impair Regulation of mTORC1 by Amino Acids We took advantage of the mutants defective in intersubunit communication to determine its importance in mTORC1 signaling in cells. We first focused on the interaction of the Rag heterodimer with mTORC1, which is strongest when RagA is loaded with GTP and RagC with GDP. As expected, heterodimers containing the RagA(Q66L) mutant that does not respond to GATOR1-stimulated GTP hydrolysis or the RagC(S75N) mutant that favors GDP binding interacted better with mTORC1 than the wild-type proteins, while mutants having the opposite effects bound worse (Figures 6A and S6A) (Sancak et al., 2008). Interestingly, the communication-defective RagA mutant (RagA R37P), despite its capability to bind GTP, bound as poorly
to mTORC1 as the RagA(T21N) mutant that is deficient in GTPbinding, suggesting that locking is required for RagA to maintain the GTP-loaded state needed to associate with mTORC1 (Figure 6A). In contrast, heterodimers containing the communication-defective RagC mutants L91P or K84T interacted better with mTORC1 than those with wild-type RagC, indicating that in the absence of the locking mechanism, RagC likely reverts to the GDP-bound state that promotes the interaction with mTORC1 (Figure 6A). Thus, mutants that do not affect nucleotide binding but prevent intersubunit communication have strong effects on the Rag-mTORC1 interaction by preventing the Rag heterodimer from achieving the locked states. We next examined the effects of the mutants on the phosphorylation of the mTORC1 substrate S6K1 in response to amino acid stimulation and starvation. In HEK293T cells expressing the control protein Metap2 or the wild-type Rags, amino acid stimulation and deprivation regulated mTORC1 activity, as monitored by the phosphorylation of its substrate, S6K1 (Figures 6B and 6C). As expected, mutants that alter the nucleotide-binding preferences of the Rags strongly impacted mTORC1 signaling; the RagA(T21N) and RagA(Q66L) mutants rendered the pathway unresponsive to amino acids (Figure 6B), while RagC(S75N) prevented the turning off of mTORC1 signaling normally caused by amino acid starvation (Figure 6C). Mutants that block intersubunit communication also caused severe signaling defects. The expression of RagA(R37P) prevented mTORC1 activation in response to amino acid stimulation (Figure 6B), while that of RagC(L91P) and RagC(K84T) suppressed the inhibition of mTORC1 signaling that normally occurs upon amino acid deprivation (Figure 6C). Moreover, in cells expressing the communication-defective mutants, the response rate of mTORC1 signaling to amino acid stimulation or starvation was also blunted (Figures 6D and 6E). These observations correlate well with the biochemical and protein interaction assays and reveal that mTORC1 signaling in response to amino acids is distorted if the Rag heterodimer cannot achieve the locked state (Figure 6F). DISCUSSION As the convergence point of diverse amino acid signals, the heterodimeric Rag GTPases have a central role in controlling the localization and activation of mTORC1. Here, we used kinetic approaches to analyze the intersubunit communication between the subunits within the Rag heterodimer and found that its unique architecture allows for crosstalk between the subunits that is essential for the normal function of the mTORC1 pathway. Specifically, when one subunit binds GTP, it triggers a conformational change that makes it dominant over the other and drives the Rag heterodimer into a locked conformation. This state suppresses the association of a second GTP, and, even if binding does occur, induces its faster hydrolysis. This intersubunit communication is a unique mode of regulation for GTPases and informs how the Rag GTPases transmit amino acid sufficiency to mTORC1. At the molecular level, how do the Rag GTPases attain the locked conformation? Combining insights from our biochemical analyses with those from the previously solved structures of the
Molecular Cell 68, 1–14, November 2, 2017 9
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
32−
RagA×GTP
2 105
E kobsd (min-1) kobsd (min-1)
H
** **
0.02
GTP association
I
0.0002 WT RagA(R37P)-RagC RagA-RagC(L91P) RagA-RagC(K84T)
0
20 40 60 80 Rag GTPases (nM)
GDP
GDP RagA
GTP
4000
Wildtype
47
76
11
41
RagC(L91P)
61
60
19
47
RagC(K84T)
26
65
7.6
45
RagA(R37P)
61
101
6.5
47
Summary of half-site on-rates (4°C) RagA-RagCN GDP Prebound N= GTP
* *
RagC(L91P)
170
1060
140
1250
ns
kon (M-1 s-1)
RagC(K84T)
400
1970
570
1710
RagA(R37P)
120
650
120
620
2.0
Summary of GTP hydrolysis kinetics (25°C)
J
0.003 0.002
0
GDP
50 GTP (μM)
Loop closing (Intrasubunit) GTP GT
100
kcat (min-1)
KM (nM)
Single Multiple Turnover Turnover
Single Multiple Turnover Turnover
Wildtype
2.2×10-4 3.0×10-3
4.0
12000
RagA-RagC(L91P)
2.6×10-4 1.6×10-3↓
3.3
8100
RagA-RagC(K84T)
2.1×10-4 3.9×10-3
6.1
9700
RagA(R37P)-RagC 2.0×10-4 1.8×10-3↓
5.5
7600
0.001 0.000
1.1×10-4 s-1 460
GTP association
GTP association 0.5 1.0 1.5 RagA-RagC (μM)
N RagA-RagC GDP GTP
ns
WT RagA(R37P)-RagC RagA-RagC(L91P) RagA-RagC(K84T)
0.004
100
-4 RagC(K84T) 7.0×10-8 3.0×10-4 1.6×10-5 ↑ 6.6×10
Wildtype 1.1×10-4 s-1 890
GTP attaching
GTP
2000 Time (min)
0.01
0.00 0.0
0.5 1.0 RagA-RagC (μM)
0.0001
0
RagA(R37P)-RagCGTP GTP RagA(R37P)-RagC
ns ns
0.0003
0.0000
K
0.02
**
0.00 0.0
0
G
**
0.04
Kd (nM)
GTP dissociation
F
RagA-RagC(K84T)GTP RagA-RagC(K84T) RagA-RagC(L91P)GTP GTP RagA-RagC(L91P)
1.2×10-7 8.0×10-4 9.8×10-7 6.4×10-4
-4 RagC(L91P) 9.0×10-7 4.0×10-4 1.0×10-5 ↑ 4.8×10
4 104
0.01 0.1 1 RagA-RagC(L91P) (μM)
GTP
0.06
Wildtype
koff (s-1)
RagC(L91P) RagA
RagC GDP GTP
-4 -6 -4 RagA(R37P) 8.9×10-7 ↑ 7.2×10 5.3×10 6.0×10
GTP binding
0
32−
Summary of binding kinetics (4°C) RagA GTP GDP
RagC(L91P) ×GTP RagA×GTP
46−
104
8
RagC(L91P) RagA
D
Time
MW (kD)
RagC(L91P) ×GTP
105
4 Binding (au)
46−
Binding (au)
MW (kD)
Leu91Hs (Leu39Sc)
Lys84Hs (Asn32Sc)
C
[RagA-RagC(L91P)]
kobsd (min-1)
B
Gtr2pGppNHp PDB: 3R7W
kobsd (min-1)
A
Subunit rotation (Intersubunit) GTP
GDP
GDP
RagC open
RagC(L91P) ?
RagA(R37P)
RagA(T21N)
closed
GTP
GDP
GDP RagA
GDP
GTP
GDP
GDP
GTP TP
GTP
RagC open
RagC(S75N)
RagC(L91P) RagC(K84T)
RagA(R37P) ?
closed
Figure 5. Mutations in Switch I of the Rag GTPases Impair Intersubunit Communication (A) Crystal structure of Gtr2p (RagC homolog in yeast) bound to GppNHp (PDB: 3R7W). The switch I region is shown in orange. Upon GppNHp binding, switch I forms a lid on top the nucleotide-binding pocket. Leu91Hs and Lys84Hs are shown in spheres. (B) GTP binding to RagA-RagC(L91P). Top: nucleotide bound to RagA and RagC(L91P). Bottom: quantification. Data were fit to a single-site binding equation to obtain the Kd. In comparison, wild-type binding is shown as dashed curves. RagA-RagC(L91P) and wild-type Rags bind GTP in a similar fashion. (C) Reaction time course of radioactively labeled GTP remaining bound to RagA-RagC(L91P) during the off-rate measurement. Top: GTP bound to RagA and RagC(L91P) after using unlabeled GTP as a chase. Bottom: quantification. Data were fit to a single exponential decay to obtain the koff. The wild-type off rate is shown as dashed curves for reference. GTP dissociates from RagA-RagC(L91P) faster than from wild-type Rags. (D) Summary of koff and Kd values for RagA-RagC(L91P), RagA-RagC(K84T), and RagA(R37P)-RagC. Wild-type data are reproduced from Figure 1Q for ease of comparison. (E and F) Half-site on rates for RagA-RagC(L91P) (squares) and RagA-RagC(K84T) (diamonds) (E) and RagA(R37P)-RagC (circles) (F). The subunits in bold are those with which the labeled GTP is associating in the assay to measure the half-site on rate. The slope gives the half-site on rate. Wild-type half-site on rates are shown as dashed lines for comparisons. Notice that the concentration dependence for GTP association is partially restored for all the mutants. *p < 0.05; **p < 0.01; ns, not significant from zero. (G) Summary of half-site on rates of GTP to RagA-RagC(L91P), RagA-RagC(K84T), and RagA(R37P)-RagC. On rates for wild-type Rags are reproduced from Figure 2H for ease of comparison. (H and I) Single- (H) and multiple-turnover (I) GTP hydrolysis assays for RagA-RagC(L91P) (blue circles), RagA-RagC(K84T) (brown circles), and RagA(R37P)RagC (red circles). Wild-type curves are shown as dashed lines. (J) Summary of hydrolysis kinetics for RagA-RagC(L91P), RagA-RagC(K84T), and RagA(R37P)-RagC. Wild-type data are reproduced from Figure 3K for ease of comparison. (K) Illustration of the effects of the switch I mutations on intersubunit crosstalk within the Rag heterodimer.
10 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
A
B
Rag mutants and abbreviations
Mutation property Abbre. Wildtype RagA-RagC W D Lower binding affinity to GTP Slower GTP hydrolysis T Switch I mutation 1 S1 Switch I mutation 2 S2
cDNA transfected
FLAG-metap2 HA-RagA FLAG-RagC
RagA T21N Q66L R37P -
Amino acid stimulation
cDNA transfection
Change media to RPMI(5% aa) 2 hours
+ W D T S1 W W W W W W W W W W D T S1 S2
‘-aa’
HA-metap2 cDNA transfected HA-RagA HA-RagC amino acids
IP: FLAG
– – – –
+
+
–
+
WT WT
T21N WT
Q66L R37P WT WT
–
–
–
+
+
+
–
+
–
WT WT
+
–
WT WT WT S75N Q120L L91P
+
–
+
–
+
–
+
WT K84T –
+ Short exp.
P -T389-S6K1
Long exp.
S6K1
+95% aa 15 mins
Raptor
HA-metap2
‘+aa’
HA-RagC
mTOR IP: FLAG
FLAG-S6K1 and +
HEK 293T cells
RagC S75N Q120L L91P K84T
Cell lysate
HA-RagA
HA-RagA
FLAG-metap2
Raptor
FLAG-RagC Raptor
C
mTOR Cell lysate
Amino acid starvation
HA-RagA
HEK 293T cells
FLAG-metap2
cDNA transfection
Protein level (au)
FLAG-RagC
Change media to RPMI(100% aa)
Raptor mTOR
2
1 hour
HA-metap2 cDNA transfected HA-RagA HA-RagC amino acids
‘+aa’
1
2
3
4
5 6 7 Lane
8
–
+
–
WT WT
T21N Q66L WT WT
+
+ –
–
+
–
FLAG-S6K1 and + R37P WT + –
+
WT WT –
+
P -T389-S6K1
–
WT WT WT S75N Q120L L91P +
–
+
–
+
–
WT K84T +
–
Long exp.
S6K1
Change to RPMI(5% aa)
9 10
+
+
Short exp
IP: FLAG
1 0
– – –
HA-metap2
30 mins
HA-RagC
‘-aa’ Cell lysate
HA-RagA Raptor
R37P
GTP GTP RagA
GDP RagC
T R
C
(L
91
P)
P)
W
37 R
250
ag
200
A(
100 150 Time (min)
ag
50
R
(L
91
P)
T
P)
W
37
0.02
0.00
0
R
L91P or K84T
(min-1)
1 106 0
R
Activated
2 106
0.00
R
250
0.04
WT RagA(R37P) RagC(L91P)
C
200
A(
100 150 Time (min)
3
106
ag
50
ag
0
4
106
ON→OFF
Signal (au)
OFF→ON
1 106
0.05
k
Signal (au)
2 106
Kinetics for amino acid starvation
0.10
(min-1)
WT RagA(R37P) RagC(L91P)
3 106
0
F
E
Kinetics for amino acid stimulation 4 106
k
D
Inhibited
GTP GDP RagA
GTP RagC
Figure 6. Communication-Defective Rag Mutants Impair mTORC1 Signaling in Cells (A) Coimmunoprecipitation of Rag GTPase heterodimers containing communication-defective mutants with mTORC1 subunits mTOR and Raptor. Communication-defective mutants mimic mutants incapable of binding GTP in terms of their interactions with mTORC1. (B and C) Effects of the communication-defective mutants on mTORC1 signaling in response to amino acid stimulation (B) or starvation (C). mTORC1 signaling is detected by the phosphorylation of Thr389 of S6K1 in response to amino acid stimulation (B) or starvation (C). mTORC1 signaling is impaired in cells expressing communication-defective mutants. (D and E) mTORC1 activity during a time course of amino acid stimulation (D) or starvation (E). Communication-defective mutants slow the response of the mTORC1 pathway to changes in amino acid levels. (F) Illustration of the effects of the communication-defective mutants on mTORC1 signaling in response to amino acids.
homologous yeast Gtr proteins (Gong et al., 2011; Jeong et al., 2012), we hypothesize the following chain of events. The binding of the GTP to one subunit of the heterodimer triggers local conformational changes in switch I and switch II, which form a lid over its nucleotide-binding pocket. Unlike Ras, whose switch I adopts a loop conformation upon GTP binding, we predict from the Gtr1 structure that switch I of the Rag GTPases forms a short a helix
(Figures 7A and 7B). This helix is in the right position to touch the other subunit, possibly precluding rearrangement of its switch I and thus causing slower association and a higher hydrolysis rate for the second GTP. The proposed reliance of the locking state on cooperation between the switch I regions of the Rag subunits could explain why we can impair it by introducing mutations that promote switch I flexibility.
Molecular Cell 68, 1–14, November 2, 2017 11
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
A
RasGMPPNP
B
RasGDP
I
I
Gtr2pGMPPNP
Gtr2pGDP
II
I
II
II
II
I
Ras (overlap)
Gtr2p (overlap)
C
nl U
Transitioning
g
GDP
in
oc
ki
RagC
ck
RagA
Amino acid deficiency
GDP
Lo
ng
GDP
GDP GTP Maintenance of Inhibition
GDP
GTP
Inhibited GATOR1 FLCNFNIP2 GTP GDP
GDP
Maintenance of Activation
GDP
nl
in g
Transitioning GDP
U
ck
Activated
oc
Lo
ki
ng
GTP
GDP
Amino acid sufficiency
Figure 7. Proposed Locking Mechanism for the Regulation of the Rag GTPases (A and B) Comparison of the conformations of Ras (A) and Gtr2p (B) upon GTP or GDP binding. The switch I regions of Ras and Gtr2p are shown in red or orange. Switch II is shown in cyan or green. Gtr2p (the RagC homolog in yeast) shows more dramatic conformational changes than Ras in switch I and switch II when it is associated with the GTP analog. (C) Functional cycle of the Rag heterodimer in response to amino acid levels. Two stable states, GTPRagA-RagCGDP and GDPRagA-RagCGTP, generate constant outputs for mTORC1 in the presence or absence of amino acids, respectively. The GDPRagA-RagCGDP configuration allows for interconversion between the active and inhibited states.
Beyond its requirement for intersubunit communication and normal mTORC1 signaling, there are several other reasons why the heterodimeric architecture of Rag GTPases may have evolved. First, unlike conventional signaling GTPases that exist in two states, on (GTP bound) and off (GDP bound), the Rag GTPases can exist in four states. Although previous work suggests that two are stable (Sancak et al., 2008) (GTPRagARagCGDP and GDPRagA-RagCGTP), a third (GDPRagA-RagCGDP) must be an intermediate given the importance of the GAPs for interconverting between them. Moreover, the off state of the Rag
12 Molecular Cell 68, 1–14, November 2, 2017
heterodimer requires RagC to bind GTP, which suppresses its association with mTORC1 regardless of the identity of the nucleotide bound to RagA (Figure S7). It is likely that a switch I rearrangement in RagC either directly displaces mTORC1 or shields the residues responsible for it to associate. Therefore, the off state of the Rag GTPase heterodimer is likely different from that of canonical GTPases, in that it is ‘‘active’’ because mTORC1 is actively rejected due to physical hindrance rather simply being unable to bind with it. We speculate that this provides an additional mechanism to prevent the mis-activation of mTORC1. Therefore, the heterodimeric nature of the Rag GTPases creates more states that should allow for the more precise control of mTORC1 signaling. We hypothesized that a role for the fourth state (GTPRagA-RagCGTP) will eventually emerge. Second, a heterodimer allows for more binding sites to integrate the many regulators and effectors that the Rag GTPases must contact. Residing in what is likely a crowded environment on the lysosomal surface, the Rag GTPases need to interact with several large multi-protein factors, including, but probably not limited to, Ragulator (Sancak et al., 2010), SLC38A9 (Jung et al., 2015; Rebsamen et al., 2015; Wang et al., 2015), GATOR1 (Bar-Peled et al., 2013; Panchaud et al., 2013), FLCN-FNIP2 (Petit et al., 2013; Tsun et al., 2013), and mTORC1 (Sancak et al., 2008). Considering that the size of a canonical GTPase domain is only 20 kDa, it is physically impossible for it to interface with so many factors at the same time, and a heterodimer provides additional binding sites for this to occur. In addition, the existence of the locking mechanism allows for the control of the nucleotide-loading state of one subunit to determine the functional state of the entire heterodimer. Compared to Ras, the Rag GTPases have slower rates of GTP dissociation and hydrolysis (Frech et al., 1994). These properties enable the Rags to hold on to the nucleotide-loading state in which they reside (Figure 7C), probably because of a requirement for generating a constant output. In order to switch between functional states, two GAPs are needed. By better positioning the critical glutamine residue that chelates the nucleophilic-attacking water, GATOR1 and FLCN-FNIP2 strongly stimulate the GTP hydrolysis rates of RagA and RagC, respectively, which drives the heterodimer from a locked on or off state into the GDPRagA-RagCGDP intermediate (Figure 7C). At this point, if a GTP molecule binds to the same subunit, then it will be hydrolyzed immediately because of the proximity to the corresponding GAP. On the other hand, if GTP binds to the other subunit, which drives the Rags into the opposite functional state, then no stimulated hydrolysis will occur because of the specificity of GATOR1 and FLCN-FNIP2 (cf. Figure 5). The existence of intersubunit communication by the Rag heterodimer and of many Rag GTPase regulators hints at the importance of ensuring faithful and efficient mTORC1 signaling. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d
KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
d d
d
EXPERIMENTAL MODEL AND SUBJECT DETAILS METHOD DETAILS B Purification of the Rag GTPase Heterodimer (I) B Purification of the Rag GTPase Heterodimer (II) B Purification of the GATOR1 and FLCN-FNIP2 Complexes B Equilibrium Binding Assays B On-Rate Measurements B Off-Rate Measurements B Single Turnover GTP Hydrolysis Experiments B Multiple Turnover GTP Hydrolysis Experiments B Free Energy Diagram B Preparation of Cell Lysates and Immunoprecipitates B Lysine Modification Assay QUANTIFICATION AND STATISTICAL ANALYSIS
Frech, M., Darden, T.A., Pedersen, L.G., Foley, C.K., Charifson, P.S., Anderson, M.W., and Wittinghofer, A. (1994). Role of glutamine-61 in the hydrolysis of GTP by p21H-ras: an experimental and theoretical study. Biochemistry 33, 3237–3244. Gong, R., Li, L., Liu, Y., Wang, P., Yang, H., Wang, L., Cheng, J., Guan, K.-L., and Xu, Y. (2011). Crystal structure of the Gtr1p-Gtr2p complex reveals new insights into the amino acid-induced TORC1 activation. Genes Dev. 25, 1668–1673. Gonza´lez, A., and Hall, M.N. (2017). Nutrient sensing and TOR signaling in yeast and mammals. EMBO J. 36, 397–408. Guo, X., Bandyopadhyay, P., Schilling, B., Young, M.M., Fujii, N., Aynechi, T., Guy, R.K., Kuntz, I.D., and Gibson, B.W. (2008). Partial acetylation of lysine residues improves intraprotein cross-linking. Anal. Chem. 80, 951–960. Hall, B.E., Bar-Sagi, D., and Nassar, N. (2002). The structural basis for the transition from Ras-GTP to Ras-GDP. Proc. Natl. Acad. Sci. USA 99, 12138–12142.
SUPPLEMENTAL INFORMATION
Hwang, Y.W., and Miller, D.L. (1987). A mutation that alters the nucleotide specificity of elongation factor Tu, a GTP regulatory protein. J. Biol. Chem. 262, 13081–13085.
Supplemental Information includes seven figures and can be found with this article online at https://doi.org/10.1016/j.molcel.2017.09.026.
Inoki, K., Li, Y., Xu, T., and Guan, K.-L. (2003). Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signaling. Genes Dev. 17, 1829–1834.
AUTHOR CONTRIBUTIONS K.S. and D.M.S. initiated the project, and K.S. designed the research plan. K.S. performed all the experiments and analyzed the data with input from A.C. K.S. wrote the manuscript, and D.M.S. edited the manuscript. ACKNOWLEDGMENTS We thank all members of the Sabatini Laboratory for helpful insights. We thank Dr. Shu-ou Shan for critical reading of the manuscript. This work was supported by grants from the NIH (R01 CA103866, R01 CA129105, and R37 AI47389) and the Department of Defense (W81XWH-15-1-0230) to D.M.S. and fellowship support from the Life Sciences Research Foundation to K.S, where he is a Pfizer Fellow. D.M.S. is an investigator of the Howard Hughes Medical Institute. Received: June 5, 2017 Revised: August 14, 2017 Accepted: September 19, 2017 Published: October 19, 2017 REFERENCES Bar-Peled, L., Chantranupong, L., Cherniack, A.D., Chen, W.W., Ottina, K.A., Grabiner, B.C., Spear, E.D., Carter, S.L., Meyerson, M., and Sabatini, D.M. (2013). A tumor suppressor complex with GAP activity for the Rag GTPases that signal amino acid sufficiency to mTORC1. Science 340, 1100–1106. De Franceschi, N., Wild, K., Schlacht, A., Dacks, J.B., Sinning, I., and Filippini, F. (2014). Longin and GAF domains: structural evolution and adaptation to the subcellular trafficking machinery. Traffic 15, 104–121. Dibble, C.C., and Manning, B.D. (2013). Signal integration by mTORC1 coordinates nutrient input with biosynthetic output. Nat. Cell Biol. 15, 555–564. Efeyan, A., Comb, W.C., and Sabatini, D.M. (2015). Nutrient-sensing mechanisms and pathways. Nature 517, 302–310. Feig, L.A., and Cooper, G.M. (1988). Inhibition of NIH 3T3 cell proliferation by a mutant ras protein with preferential affinity for GDP. Mol. Cell. Biol. 8, 3235–3243. Fersht, A. (1999). Structure and Mechanism in Protein Science (Macmillan). Feuerstein, J., Goody, R.S., and Wittinghofer, A. (1987). Preparation and characterization of nucleotide-free and metal ion-free p21 ‘‘apoprotein’’. J. Biol. Chem. 262, 8455–8458.
Jeong, J.-H., Lee, K.-H., Kim, Y.-M., Kim, D.-H., Oh, B.-H., and Kim, Y.-G. (2012). Crystal structure of the Gtr1p(GTP)-Gtr2p(GDP) protein complex reveals large structural rearrangements triggered by GTP-to-GDP conversion. J. Biol. Chem. 287, 29648–29653. Jewell, J.L., Russell, R.C., and Guan, K.-L. (2013). Amino acid signalling upstream of mTOR. Nat. Rev. Mol. Cell Biol. 14, 133–139. Jung, J., Genau, H.M., and Behrends, C. (2015). Amino acid-dependent mTORC1 regulation by the lysosomal membrane protein SLC38A9. Mol. Cell. Biol. 35, 2479–2494. Leipe, D.D., Wolf, Y.I., Koonin, E.V., and Aravind, L. (2002). Classification and evolution of P-loop GTPases and related ATPases. J. Mol. Biol. 317, 41–72. Levine, T.P., Daniels, R.D., Wong, L.H., Gatta, A.T., Gerondopoulos, A., and Barr, F.A. (2013). Discovery of new Longin and Roadblock domains that form platforms for small GTPases in Ragulator and TRAPP-II. Small GTPases 4, 62–69. Li, Y., Inoki, K., and Guan, K.L. (2004). Biochemical and functional characterizations of small GTPase Rheb and TSC2 GAP activity. Mol. Cell. Biol. 24, 7965–7975. Long, P.A., Zimmermann, M.T., Kim, M., Evans, J.M., Xu, X., and Olson, T.M. (2016). De novo RRAGC mutation activates mTORC1 signaling in syndromic fetal dilated cardiomyopathy. Hum. Genet. 135, 909–917. Menon, S., Dibble, C.C., Talbott, G., Hoxhaj, G., Valvezan, A.J., Takahashi, H., Cantley, L.C., and Manning, B.D. (2014). Spatial control of the TSC complex integrates insulin and nutrient regulation of mTORC1 at the lysosome. Cell 156, 771–785. Nakashima, N., Noguchi, E., and Nishimoto, T. (1999). Saccharomyces cerevisiae putative G protein, Gtr1p, which forms complexes with itself and a novel protein designated as Gtr2p, negatively regulates the Ran/Gsp1p G protein cycle through Gtr2p. Genetics 152, 853–867. Nowak, E., and Goody, R.S. (1988). Kinetics of adenosine 50 -triphosphate and adenosine 50 -diphosphate interaction with G-actin. Biochemistry 27, 8613–8617. Okosun, J., Wolfson, R.L., Wang, J., Araf, S., Wilkins, L., Castellano, B.M., Escudero-Ibarz, L., Al Seraihi, A.F., Richter, J., Bernhart, S.H., et al. (2016). Recurrent mTORC1-activating RRAGC mutations in follicular lymphoma. Nat. Genet. 48, 183–188. Pai, E.F., Krengel, U., Petsko, G.A., Goody, R.S., Kabsch, W., and Wittinghofer, A. (1990). Refined crystal structure of the triphosphate conformation of H-ras p21 at 1.35 A resolution: implications for the mechanism of GTP hydrolysis. EMBO J. 9, 2351–2359.
Molecular Cell 68, 1–14, November 2, 2017 13
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Panchaud, N., Pe´li-Gulli, M.-P., and De Virgilio, C. (2013). Amino acid deprivation inhibits TORC1 through a GTPase-activating protein complex for the Rag family GTPase Gtr1. Sci. Signal. 6, ra42. Peluso, P., Shan, S.-O., Nock, S., Herschlag, D., and Walter, P. (2001). Role of SRP RNA in the GTPase cycles of Ffh and FtsY. Biochemistry 40, 15224–15233. Petit, C.S., Roczniak-Ferguson, A., and Ferguson, S.M. (2013). Recruitment of folliculin to lysosomes supports the amino acid-dependent activation of Rag GTPases. J. Cell Biol. 202, 1107–1122. Rebsamen, M., Pochini, L., Stasyk, T., de Araujo, M.E.G., Galluccio, M., Kandasamy, R.K., Snijder, B., Fauster, A., Rudashevskaya, E.L., Bruckner, M., et al. (2015). SLC38A9 is a component of the lysosomal amino acid sensing machinery that controls mTORC1. Nature 519, 477–481. Saito, K., Araki, Y., Kontani, K., Nishina, H., and Katada, T. (2005). Novel role of the small GTPase Rheb: its implication in endocytic pathway independent of the activation of mammalian target of rapamycin. J. Biochem. 137, 423–430. Sancak, Y., Peterson, T.R., Shaul, Y.D., Lindquist, R.A., Thoreen, C.C., BarPeled, L., and Sabatini, D.M. (2008). The Rag GTPases bind raptor and mediate amino acid signaling to mTORC1. Science 320, 1496–1501. Sancak, Y., Bar-Peled, L., Zoncu, R., Markhard, A.L., Nada, S., and Sabatini, D.M. (2010). Ragulator-Rag complex targets mTORC1 to the lysosomal surface and is necessary for its activation by amino acids. Cell 141, 290–303. Saucedo, L.J., Gao, X., Chiarelli, D.A., Li, L., Pan, D., and Edgar, B.A. (2003). Rheb promotes cell growth as a component of the insulin/TOR signalling network. Nat. Cell Biol. 5, 566–571. €ller, L., Lautwein, A., Scheffzek, K., Ahmadian, M.R., Kabsch, W., Wiesmu Schmitz, F., and Wittinghofer, A. (1997). The Ras-RasGAP complex: structural basis for GTPase activation and its loss in oncogenic Ras mutants. Science 277, 333–338. €rmann, A., Brauers, A., Massmann, S., Becker, W., and Joost, H.G. Schu (1995). Cloning of a novel family of mammalian GTP-binding proteins (RagA, RagBs, RagB1) with remote similarity to the Ras-related GTPases. J. Biol. Chem. 270, 28982–28988.
14 Molecular Cell 68, 1–14, November 2, 2017
Shan, S.-O. (2016). ATPase and GTPase tangos drive intracellular protein transport. Trends Biochem. Sci. 41, 1050–1060. Shaw, R.J., and Cantley, L.C. (2006). Ras, PI(3)K and mTOR signalling controls tumour cell growth. Nature 441, 424–430. Simon, I., Zerial, M., and Goody, R.S. (1996). Kinetics of interaction of Rab5 and Rab7 with nucleotides and magnesium ions. J. Biol. Chem. 271, 20470–20478. Stocker, H., Radimerski, T., Schindelholz, B., Wittwer, F., Belawat, P., Daram, P., Breuer, S., Thomas, G., and Hafen, E. (2003). Rheb is an essential regulator of S6K in controlling cell growth in Drosophila. Nat. Cell Biol. 5, 559–565. Sweet, D.J., and Gerace, L. (1996). A GTPase distinct from Ran is involved in nuclear protein import. J. Cell Biol. 133, 971–983. Tsun, Z.-Y., Bar-Peled, L., Chantranupong, L., Zoncu, R., Wang, T., Kim, C., Spooner, E., and Sabatini, D.M. (2013). The folliculin tumor suppressor is a GAP for the RagC/D GTPases that signal amino acid levels to mTORC1. Mol. Cell 52, 495–505. Vetter, I.R., Nowak, C., Nishimoto, T., Kuhlmann, J., and Wittinghofer, A. (1999). Structure of a Ran-binding domain complexed with Ran bound to a GTP analogue: implications for nuclear transport. Nature 398, 39–46. Wang, Y., Nakashima, N., Sekiguchi, T., and Nishimoto, T. (2005). Saccharomyces cerevisiae GTPase complex: Gtr1p-Gtr2p regulates cell-proliferation through Saccharomyces cerevisiae Ran-binding protein, Yrb2p. Biochem. Biophys. Res. Commun. 336, 639–645. Wang, S., Tsun, Z.-Y., Wolfson, R.L., Shen, K., Wyant, G.A., Plovanich, M.E., Yuan, E.D., Jones, T.D., Chantranupong, L., Comb, W., et al. (2015). Lysosomal amino acid transporter SLC38A9 signals arginine sufficiency to mTORC1. Science 347, 188–194. Wong, M. (2013). A critical review of mTOR inhibitors and epilepsy: from basic science to clinical trials. Expert Rev. Neurother. 13, 657–669. Yue, V.T., and Schimmel, P.R. (1977). Direct and specific photochemical cross-linking of adenosine 50 -triphosphate to an aminoacyl-tRNA synthetase. Biochemistry 16, 4678–4684.
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
STAR+METHODS KEY RESOURCES TABLE
REAGENT or RESOURCE
SOURCE
IDENTIFIER
Rabbit anti-S6K1
Cell Signaling Technology
2708; RRID: AB_390722
Rabbit anti-mTOR
Cell Signaling Technology
2972; RRID: AB_330978
Rabbit anti-Raptor
EMD Millipore
09-217; RRID: AB_612103
Antibodies
Rabbit anti-p-T389-S6K1
Cell Signaling Technology
9205; RRID: AB_330944
Rabbit anti-HA
Bethyl
A190-208A; RRID: AB_67466
Rabbit anti-Flag
Cell Signaling Technology
14793; RRID: AB_2572291
Anti-rabbit, HRP-linked antibody
Cell Signaling Technology
7074; RRID: AB_2099233
NEB
C2527
HEPES
Sigma-Aldrich
H3375
Potassium acetate
Sigma-Aldrich
P1190
Magnesium chloride
Sigma-Aldrich
M8266
Sodium chloride
Sigma-Aldrich
S7563
DTT
Sigma-Aldrich
D5545
Chaps
Sigma-Aldrich
C5070
Triton
Sigma-Aldrich
T9284
Imidazole
Sigma-Aldrich
I5513
GTP
Sigma-Aldrich
G8877
GDP
Sigma-Aldrich
G7127
TCEP
Sigma-Aldrich
C4706
ATP
Sigma-Aldrich
A7699
a-32P-GTP
PerkinElmer
BLU006H250UC
g-32P-GTP
PerkinElmer
BLU004Z250UC
a-32P-GDP
Homemade
N/A
DMEM
Sigma-Aldrich
D5523
RPMI
Sigma-Aldrich
R5886
Human: HEK293T
ATCC
CRL-3216
Human: FreeStyle 293-F
Thermo Fisher Scientific
R79007
[pSK109] pETDuet RagA-RagC
This study
Addgene 99641
[pSK151] pETDuet RagA(D130N)-RagC
This study
Addgene 99642
[pSK121] pETDuet RagA-RagC(D181N)
This study
Addgene 99643
[pSK152] pETDuet RagA(T21N)-RagC
This study
Addgene 99644
Bacterial and Virus Strains BL21(DE3) Chemicals, Peptides, and Recombinant Proteins
Experimental Models: Cell Lines
Recombinant DNA
[pSK170] pETDuet RagA-RagC(S75N)
This study
Addgene 99645
[pSK330] pETDuet RagA(R37P)-RagC
This study
Addgene 99646
[pSK210] pETDuet RagA-RagC(L91P)
This study
Addgene 99647
[pSK224] pETDuet RagA-RagC(K84T)
This study
Addgene 99648
[pSK235] pK RagA-RagC
This study
Addgene 99649
[pSK251] pK RagA(T21N)-RagC
This study
Addgene 99650
[pSK252] pK RagA-RagC(S75N)
This study
Addgene 99699
[pSK245] pK RagA(Q66L)-RagC
This study
Addgene 99700 (Continued on next page)
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e1
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Continued REAGENT or RESOURCE
SOURCE
IDENTIFIER
[pSK246] pK RagA-RagC(Q120L)
This study
Addgene 99710
[pSK247] pK RagA(Q66L)-RagC(Q120L)
This study
Addgene 99702
[pSK249] pK RagA(R37P)-RagC
This study
Addgene 99706
[pSK269] pK RagA-RagC(L91P)
This study
Addgene 99707
[pSK332] pK RagA-RagC(K84T)
This study
Addgene 99708
[pSK6] pRK5 HA-RagA
This study
Addgene 99710
[pSK323] pRK5 HA-RagA(T21N)
This study
Addgene 99711
[pSK324] pRK5 HA-RagA(Q66L)
This study
Addgene 99712
[pSK325] pRK5 HA-RagA(R37P)
This study
Addgene 99713
[pSK184] pRK5 Flag-RagA
This study
Addgene 99714
[pSK192] pRK5 Flag-RagA(T21N)
This study
Addgene 99715
[pSK194] pRK5 Flag-RagA(Q66L)
This study
Addgene 99716
[pSK327] pRK5 Flag-RagA(R37P)
This study
Addgene 99717
[pSK8] pRK5 HA-RagC
This study
Addgene 99718
[pSK320] pRK5 HA-RagC(S75N)
This study
Addgene 99719
[pSK321] pRK5 HA-RagC(Q120L)
This study
Addgene 99720
[pSK322] pRK5 HA-RagC(L91P)
This study
Addgene 99721
[pSK328] pRK5 HA-RagC(K84T)
This study
Addgene 99722
[pSK189] pRK5 Flag-RagC
This study
Addgene 99723
[pSK196] pRK5 Flag-RagC(S75N)
This study
Addgene 99724
[pSK197] pRK5 Flag-RagC(Q120L)
This study
Addgene 99725
[pSK326] pRK5 Flag-RagC(L91P)
This study
Addgene 99726
[pSK329] pRK5 Flag-RagC(K84T)
This study
Addgene 99727
[pSK163] pRK5 HA-Nprl2
Bar-Peled et al., 2013
Addgene 99709
[pSK164] pRK5 HA-Nprl3
Bar-Peled et al., 2013
Addgene 46330
[pSK165] pRK5 Flag-Depdc5
Bar-Peled et al., 2013
Addgene 46340
[pSK310] pRK5 HA-FLCN
Tsun et al., 2013
Addgene 72297
[pSK313] pRK5 Flag-FNIP2
Tsun et al., 2013
Addgene 72294
Software and Algorithms Prism
GraphPad Software
N/A
ImageJ
N/A
N/A
Flag-M2 affinity resin
Sigma-Aldrich
A2220
Ni-NTA affinity resin
EMD Millipore
30230
Q-Sepharose FF resin
GE Healthcare
17051001
Cellulose 300 PEI thin layer chromatography plates
Sorbtech
1122126
Codon optimized sequence for RagA and RagC
Genescript
N/A
Other
CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, David M. Sabatini ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS HEK293T cells were obtained from ATCC and maintained in an incubator sett at 37 C and 5% CO2. HEK293T cells were cultured in DMEM with 10% IFS supplemented with 2 mM glutamine, penicillin (100 IU/ml), and streptomycin (100 mg/ml).
e2 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
FreeStyle 293-F cells were obtained from Thermo Fisher and maintained in a Multitron Pro shaker setting at 37 C, 125 rpm, 8% CO2, and 80% humidity. FreeStyle 293-F cells were cultured in FreeStyle 293 Expression Medium supplemented with penicillin (100 IU/ml) and streptomycin (100 mg/ml). METHOD DETAILS Purification of the Rag GTPase Heterodimer (I) To co-express the Rag GTPase dimer in bacteria, a pETDuet-1 vector encoding codon optimized, C-terminally His-tagged RagA and tagless RagC (pSK109) was transformed into BL21(DE3) competent cells and plated on a LB-agar plate containing 100 mg/ml ampicillin. A single colony was inoculated in 50 mL LB media containing 100 mg/ml ampicillin and shaken at 37 C overnight. The saturated overnight culture was diluted 1:100 into four liters of LB to continue growth. Optical density (OD) of the culture was monitored at 600 nm, and the temperature was lowered to 18 C when OD600 reached 0.4 – 0.6 (early log-phase). Half an hour later when the culture was cooled down to 25 C, the Rag dimer was induced by 0.5 mM IPTG and expressed overnight at 18 C (20 hr). Unless specified below, all the purification steps were carried out at 4 C or on ice to stabilize the Rag GTPases. The bacterial pellets were resuspended in 100 mL of Resuspension Buffer (50 mM NaHEPES, pH 7.4; 100 mM NaCl; 2 mM MgCl2; 2 mM DTT; 0.5 mM PMSF; 0.05% Triton; 100 mM GDP; 100 mg/ml lysozyme; 10 mg/ml DNase; and four tablets of protease inhibitor cocktail). The mixture was stirred at 4 C for 30 min to ensure homogeneous resuspension, before passing through the microfluidizer five times at the pressure of 18,000 psi. The cell lysate was cleared by centrifugation at 40,000 3 g for one hour. The supernatant containing soluble Rag GTPase heterodimer was taken for the next step. The cleared cell lysate was first cleaned up using a hand-packed anion exchange column, taking advantage of the negatively charged RagC subunit. 15 mL of Q-Sepharose Fast Flow slurry was packed and equilibrated in Buffer A (50 mM NaHEPES, pH 7.4; 50 mM NaCl; 2 mM MgCl2; 2 mM DTT; 100 mM GDP; 0.05% Triton). The cleared cell lysate was then applied to the column using an isocratic flow pump at a rate of 2 mL/min. After sample loading, the column was washed by Buffer A for 20 column volumes, and the bound proteins were eluted by a linear salt gradient of Buffer A to Buffer A’ (50 mM NaHEPES, pH 7.4; 500 mM NaCl; 2 mM MgCl2; 2 mM DTT; 100 mM GDP; 0.05% Triton). Fractions containing the Rag dimer were pooled together and dialyzed three times against Buffer B (50 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; 0.5 mM TCEP; 100 mM GDP; 0.05% Triton) in preparation for Ni-NTA purification. Dialyzed fractions from Q-Sepharose were incubated with 5 mL of Ni-NTA resin that was pre-equilibrated in Buffer C (50 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; 0.5 mM TCEP; 100 mM ATP; 100 mM GDP; 0.05% Triton). Ni-NTA column was hand-packed after an hour of incubation, and washed by 20 column volumes of Buffer C and 10 column volumes of Buffer C’ (Buffer C supplemented with 30 mM imidazole and 300 mM NaCl). The Rag GTPase dimer was then eluted by Buffer C’’ (Buffer C supplemented with 250 mM imidazole). The elutant was concentrated in a 30 kDa molecular-weight cut-off concentrator, and ultracentrifuged at 100,000 3 g for 30 min, to remove all the aggregates in preparation for FPLC purification. For FPLC purification, a monoQ column was first equilibrated in 90% Buffer D (50 mM NaHEPES, pH 7.4; 2 mM MgCl2; 2 mM DTT) and 10% Buffer D’ (50 mM NaHEPES, pH 7.4; 1 M NaCl; 2 mM MgCl2; 2 mM DTT), before the elutant from Ni-NTA was applied. A linear gradient from 12% D’ to 30% D’ was used to elute the bound proteins, and a single peak was collected at around 22% D’, which corresponds to the Rag GTPase heterodimer. The peak was concentrated, and any bound nucleotides were stripped by 10 mM EDTA for an hour at room temperature. The apo Rag heterodimer was then applied to a HiLoad 16/60 Superdex 200 gelfiltration column pre-equilibrated in Buffer E (50 mM NaHEPES, pH 7.4; 100 mM KOAc; 1 mM EDTA; 2 mM DTT). The peak corresponding to the Rag heterodimer was collected and concentrated. The final product was aliquoted and flash frozen in the presence of 5% glycerol, and stored at 80 C. The final yield is 1 mg Rag heterodimer per liter of LB culture. The Rag GTPases purified using this construct are suitable for equilibrium binding assays, on- and off-rate measurements, and single turnover GTP hydrolysis assays. Because of the trace amount of contaminating GTPases which hydrolyzes GTP at a much faster rate, the product here cannot be used for multiple turnover GTP hydrolysis assay. Purification of the Rag GTPase Heterodimer (II) To completely remove any contaminating GTPases, a pCOLADuet-1 vector (KanR) encoding codon optimized, N-terminally His10-R8SUMO-tagged RagA and tagless RagC (pSK235) was expressed using a similar protocol as above. 8 L of bacteria culture were lysed and cleared in 160 mL of Resuspension Buffer (substitute DTT with 0.5 mM TCEP). The cleared supernatant was first cleaned up using a hand-packed Ni-NTA column using a similar protocol as above. The elutant was concentrated and ultracentrifuged for FPLC. A monoS column was first equilibrated in 85% Buffer D and 15% Buffer D’, before the elutant from Ni-NTA was applied. A linear gradient from 20% D’ to 100% D’ was used to elute the Rag GTPases. The peak was concentrated, and the fusion protein was digested overnight by the 3C protease to cleave the His10-R8-SUMO-tag. The digested mixture was applied to monoS again to remove the tag, and the apo Rag heterodimer was then applied to monoQ followed by HiLoad 16/60 Superdex 200 gel-filtration column. The final yield is 50 mg Rag dimer per liter of LB culture. Despite the low yield, the Rag GTPases purified using this construct are free of contaminating GTPases and are suitable for multiple turnover GTP hydrolysis assays.
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e3
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Purification of the GATOR1 and FLCN-FNIP2 Complexes To generate the GAPs for the Rag GTPases, one million 293 FreeStyle cells were inoculated into 2 L of FreeStyle media. Twenty-four hours later, cells were transfected using PEI with the following combination of pRK5 constructs: For GATOR1: 420 mg FLAG-Depdc5, 840 mg HA-Nprl2, and 840 mg HA-Nprl3; For FLCN-FNIP2: 600 mg FLAG-FNIP2 and 1.5 mg HA-FLCN. Thirty-six hours after transfection, cells were spun down and washed once with ice-cold PBS. Cells were lysed in 100 mL Triton lysis buffer (40 mM NaHEPES, pH 7.4; 100 mM NaCl; 5 mM MgCl2; 100 mM ATP; 10 mM Na4P2O7; 10 mM Na b-glycerol phosphate; 1% Triton; and two tablets of protease inhibitor cocktail), and were dounced to homogenize. After clearing up the insoluble fractions by centrifuging at 40,000 3 g for 30 min, 4 mL of pre-equilibrated FLAG-M2 beads were added to lysate. Recombinant protein complexes were immunoprecipitated for 3 hr at 4 C, before packing into a column. The resin was washed with 10 column volumes of Triton lysis buffer, followed by 20 column volumes of Triton lysis buffer supplemented with 300 mM NaCl. Before elution, the resin was buffer exchanged into CHAPS buffer (40 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; and 0.1% CHAPS). FLAG-tagged GATOR1 or FLCN-FNIP2 was then eluted from the FLAG-M2 beads with 3 3 FLAG peptide dissolved in CHAPS buffer for one hour at 4 C. The elutant was concentrated using a 100 kDa molecular-weight cut-off filter and ultracentrifuged at 100,000 3 g for 30 min in a TLA-100.3 rotor, to remove all the aggregates in preparation for FPLC purification. To further improve the purity and confirm the integrity of subunit assembly, the protein complexes were subsequently loaded onto a HiLoad 16/60 Superdex 200 gel-filtration column pre-equilibrated in Buffer GFB (50 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; 2 mM DTT; and 0.1% CHAPS). The peak corresponding to the desired complex was collected, flash frozen in GFB supplemented with 10% glycerol and stored at 80 C. Equilibrium Binding Assays Before use, the newly thawed Rag GTPase heterodimer was ultracentrifuged at 200,000 3 g for an hour and its concentration was redetermined, to completely remove any aggregates caused by the freeze-thaw cycle. Unless otherwise specified, all the biochemical assays below were carried out in Assay Buffer (50 mM NaHEPES, pH 7.4; 100 mM KOAc; 2 mM MgCl2; 2 mM DTT; and 0.1% CHAPS). To determine the binding affinity of nucleotides to individual Rag subunits, increasing concentrations of Rag GTPases, usually ranging from 5 nM to 2 mM on a log-scale, were mixed with trace amount (1 nM) of a-32P-GTP or a-32P-GDP (Perkin-Elmer). The reaction mixtures were incubated at 4 C until equilibrium was reached, usually for four hours, before directly spotted onto a chilled metal block covered with Parafilm. 0.3 Joule of 254 nm UV light was applied to the droplets, to induce non-specific, zero distance crosslinking between the nucleotide and the subunit it bound to. The resulting products were analyzed by 12% Tris-Glycine gel. After fixing and drying, the gel was exposed to a phosphor-imaging screen, and the intensities of the radioactive signals (Iobsd) were visualized by a Typhoon scanner and quantified using the ImageQuant software. The signals were fit to a single-site binding equation (Equation I) to extract the Kd for both subunits. Iobsd = Amplitude 3
½Rag + Background; Kd + ½Rag
(I)
in which Amplitude = Radioactive Intensity 3 Crosslinking efficiency 3 Imaging efficiency. For samples taken from the same batch of experiment and run on the same gel, these three terms can be assumed approximately equal, and the Amplitude can be directly compared within. However, these relative terms may vary for mutants or in different assays, so the absolute values of the radioactive signal cannot be compared across different experiments. Notice that the dissociation constant Kd is independent of the Amplitude and thus is irrelevant to a change in any of the relative terms. The sequential binding (Figure 2A) and sequential chase (Figure 2C) assays were performed in a similar fashion. For the sequential binding assay (Figure 2A), before radioactively labeled GTP was added to the reaction, the Rag GTPase mutants were pre-incubated with 1.2 equal molar of XTP or XDP at each concentration, to ensure that the xanthine-specific site was occupied. For the sequential chase assay (Figure 2C), 5 equal molar of XTP or XDP was added to the reaction mixture after the guanine-specific site was preloaded with radioactively labeled GTP. On-Rate Measurements To determine the on-rates of nucleotides to individual Rag subunits, at time zero, different concentrations of Rag GTPases, e.g., 100/ 200/300/400 nM, were mixed with trace amount (1 nM) of a-32P-GTP or a-32P-GDP to initiate the reaction. Small aliquots of the mixture were taken out at different time points for crosslinking analysis, to trace the amount of radioactively labeled nucleotides that had been associated to the Rag GTPases, until equilibrium was reached. For example, in Figure 1I, eleven time points were taken between time zero and two hours. For each concentration of the Rag GTPases, the time points were blotted on the same gel and quantified as above. The observed reaction constants (kobsd) were extracted using a single exponential rise function (Equation II): Iobsd = Amplitude 3 1 ekobsd t + Background; (II) in which Amplitude = Radioactive Intensity 3 Crosslinking efficiency 3 Imaging efficiency.
e4 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
The kobsds were then linear-fit against the concentration of the Rag GTPases to obtain the on-rate (kon) using Equation III: kobsd = kon 3 ½Rag + koff
(III)
The half-site on-rate (Figure 2E) was determined in a similar fashion, except that wild-type Rag GTPases were pre-incubated with 1.2 equal molar of unlabeled GTP or GDP at each concentration prior to the on-rate measurement. Because each functional unit of the Rag GTPases contain two nucleotide binding sites and the fact that the second GTP binding is more difficult than the first, we can assume that in the presence of 1.2 equal molar of unlabeled nucleotide, the majority of the Rag GTPases was singly occupied, while sufficient amount of unoccupied nucleotide binding pockets remained for the later-added radioactively labeled nucleotide to be associated with. Therefore, the half-site on-rate is a measure of the association rate of the radioactively labeled nucleotide to the singly-loaded Rag GTPases. The differential on-rate (Figure 2I) was determined using xanthine-specific Rag mutants, which were pre-incubated with 1.2 equal molar of unlabeled XTP or XDP at each concentration, before the radioactively labeled GTP was added to initiate the on-rate measurement. Because of the modified hydrogen bonding pattern, xanthine-specific Rag mutants should be singly loaded with unlabeled XTP or XDP, so the differential on-rate measures the association of the radioactively labeled GTP to the guanine-specific site of the Rag mutants. Notice that in these mutants, the effect of pre-loaded XTP to inhibit GTP association is not as pronounced as that of GTP in the half-site on-rate measurement for wild-type Rags, which suggests that intersubunit communication may be partially defective in the xanthine-specific mutants. Off-Rate Measurements To determine the dissociation rates of nucleotides from individual Rag subunits, we used a pulse-chase setup. A trace amount (1 nM) of a-32P-GTP or a-32P-GDP was first incubated with 1 mM Rag GTPases for two hours, so that a certain amount of radioactively labeled nucleotides were associated with each subunit. At time zero, a large excess (1 mM) of unlabeled nucleotides were added to the reaction, to ensure that once the radioactively labeled nucleotides fall out of the nucleotide binding pocket, the chance of its re-association is small. Small aliquots of the mixture were taken out at different time points for crosslinking analysis, to trace the amount of radioactively labeled nucleotides remaining bound to the Rag GTPases. The off-rates (koff) were extracted by fitting the intensities of the radioactive signals to a single exponential decay function using Equation IV: Iobsd = Amplitude 3 ekoff t + Background;
(IV)
in which Amplitude = Radioactive Intensity 3 Crosslinking efficiency 3 Imaging efficiency. Notice that the koff is independent of the Amplitude and thus is irrelevant to a change in any of the relative terms, usually caused by batch-to-batch variation. Single Turnover GTP Hydrolysis Experiments To determine the hydrolysis rate of the Rag GTPases under single turnover conditions, increasing amounts of Rag GTPases, from 0.5 nM to 20 nM for wild-type Rags, was incubated with 0.1 nM of g-32P-GTP. Aliquots of the reaction mixture were taken at different time points and quenched using 0.75 M KH2PO4 (pH 3.3). These points were taken within the time window that less than 20% of the GTP was hydrolyzed. For example, six time points were taken for wild-type Rags at 1 min, 1, 2, 4, 6, and 9 hr, as shown in Figure 3B. The time points were then applied to cellulose 300 PEI thin layer chromatography (TLC) plates, which were expanded in TLC buffer (1 M formic acid and 0.5 M LiCl) for 20 min. The plates were dried and imaged with phosphorimaging screens. The intensities of the radioactive signals were read out by a Typhoon scanner and quantified using the ImageQuant software. Linear regression was used to fit the fraction of phosphate generated by GTP hydrolysis against time, giving the observed rate constants (kobsd). The kobsds were then fit to a Michaelis-Menten equation (Equation V), to extract the kcat and KM values for the single turnover hydrolysis reaction. kobsd = kcat 3
½Rag0 KM + ½Rag0
(V)
Equation V was derived as following. Consider the reaction in a single turnover setup: k1
k2
Rag + GTP 4 ½Rag,GTP/½Rag,GDP + Pi;
(V.1)
in which k1 and k-1 represent the association and dissociation constants, and k2 represents the catalytic rate constant for the chemistry step. Notice that in the single turnover setup, the amount of Rag GTPases is in excess of that of radioactively labeled GTP, so that the enzyme turnover, i.e., dissociation of GDP from the Rag, was omitted. Intermediary species are generated at the speed of: d½Rag,GTP = k1 ½Rag½GTP ðk1 + k2 Þ½Rag,GTP; dt
(V.2)
and
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e5
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
d½Pi d½Rag,GDP = = k2 ½Rag,GTP: dt dt
(V.3)
½Rag0 = ½Rag + ½Rag,GTP + ½Rag,GDP;
(V.4)
½GTP0 = ½GTP + ½Rag,GTP + ½Rag,GDP;
(V.5)
Based on conservation of matter:
and
in which [Rag]0 and [GTP]0 represent the total concentration of the Rag GTPases and GTP at the beginning of the reaction. In a single turnover setup, where (1) the concentration of the Rag GTPases is much higher than that of GTP, and (2) the reaction is monitored within 20% of hydrolysis, we can approximate the equations above to: ½Rag0 z½Rag;
(V.6)
½GTP0 z½GTP + ½Rag,GTP:
(V.7)
d½Rag,GTP = k1 ½Rag0 ½GTP0 k1 ½GTP0 + k1 + k2 ½Rag,GTP; dt
(V.8)
and
Substitute (V.6) and (V.7) into (V.2), we obtain:
the solution of which is: ½Rag,GTP =
k1 ½Rag0 ½GTP0 1 eðk1 ½Rag0 + k1 + k2 Þt ; k1 ½Rag0 + k1 + k2
(V.9)
under the boundary condition that at time zero, [Rag,GTP] = 0. Notice that this equation can only be applied at the early stage of the reaction, before 20% of the GTP has been hydrolyzed. When t > 0, the exponential term quickly diminishes to zero, so the concentration of the intermediate species [Rag,GTP] can be reduced to: ½Rag,GTP =
k1 ½Rag0 ½GTP0 ½Rag0 ½GTP0 = : k1 ½Rag0 + k1 + k2 ½Rag0 + k1k+ k2
(V.10)
1
Define KM hðk1 + k2 Þ=k1 and substitute (V.10) into (V.3), and solve for the concentration of phosphate, we get: ½Pi = k2 3
½Rag0 ½GTP0 t: ½Rag0 + KM
(V.11)
Because the readout from TLC is the fraction of GTP that is hydrolyzed, we have: Fraction of hydrolyzed GTP =
½Pi ½Rag0 = k2 3 thkobsd t; ½GTP0 ½Rag0 + KM
(V.12)
in which kobsd = k2 3½Rag0 =ð½Rag0 + KM Þ is the observed rate constant at a given Rag concentration. Notice that the TLC assay cannot differentiate GTP hydrolysis from individual subunits of the wild-type Rag GTPases. Therefore, the measured kcat and Km are the weighted average of the kcat and Km values for RagA and RagC. Considering the similar on-rates of GTP to RagA and RagC, and that RagA(T21N)-RagC and RagA-RagC(S75N) behave similarly in both single and multiple turnover reactions (Figures 3I–3K), the true values of kcat and Km for RagA and RagC should be similar and close to the measured values. For the half-site GTP hydrolysis reaction in Figure 3L, 10 mM of wild-type RagA-RagC was pre-equilibrated with 12 mM of unlabeled nucleotides (1.2 equal molar). After the Rags were singly loaded, 0.1 nM of g-32P-GTP was added to the mixture to start the reaction. Time points were taken and analyzed as above, and the observed hydrolysis rates were obtained by applying linear regression to the radioactive signals against time. For the half-site hydrolysis chase reaction in Figure 3O, 0.1 nM of g-32P-GTP was first incubated with 200 nM of wild-type RagARagC to start the reaction. Six time points were taken, before 100 mM of unlabeled nucleotides were added to the reaction mixture, after which more time points were taken. The observed hydrolysis rates were obtained by applying linear regression to the radioactive signals against time. For GATOR1- and FLCN-FNIP2-stimulated GTP hydrolysis reaction under single turnover conditions, increasing amounts of GATOR1 or FLCN-FNIP2 were included in the reaction that measures the hydrolysis rate of 0.1 nM of g-32P-GTP by 50 nM wild-type or mutant Rag GTPases. Time points were taken at each GATOR1 or FLCN-FNIP2 concentration to trace the reaction, and analyzed as above.
e6 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Multiple Turnover GTP Hydrolysis Experiments To determine the hydrolysis rates of the Rag GTPases under multiple turnover conditions, a fixed amount of Rag GTPases (2 mM for wild-type Rags), was incubated with increasing amounts of GTP doped with a trace amount of g-32P-GTP. The total GTP concentration used to determine the kinetic parameters for multiple turnover GTP hydrolysis of wild-type Rags are 2, 5, 10, 20, 50, and 100 mM (Figure 3G). The reactions were treated and analyzed similarly as above. Linear regression was used to fit the fraction of phosphate generated by GTP hydrolysis against time, generating the observed rate constants (kobsd). The kobsds were then fit to a Michaelis-Menten equation (Equation VI), to extract the kcat and KM values for the multiple turnover hydrolysis reaction. kobsd = kcat 3
½GTP0 KM + ½GTP0
(VI)
Equation VI was derived as following. Consider the reaction in a multiple turnover setup: k1
k3
k2
Rag + GTP 4 ½Rag,GTP/½Rag,GDP + Pi/Rag + GDP;
(VI.1)
in which k1 and k-1 represent the association and dissociation constants; k2 represents the catalytic rate for the chemistry step; k3 represents the dissociation constant of GDP. Because GDP dissociation is fast as determined by the off-rate measurement and GTP concentration is high, the intermediary species Rag,GDP can be omitted. Therefore, at the initial stage of the reaction (early steady-state), the intermediates are generated at the speed of: d½Rag,GTP = k1 ½Rag½GTP ðk1 + k2 Þ½Rag,GTP; dt
(VI.2)
d½Pi = k2 ½Rag,GTP: dt
(VI.3)
and
Using the steady-state assumption ðd½Rag,GTP=dtÞ = 0 and conservation of matter: ½Rag0 = ½Rag + ½Rag,GTP;
(VI.4)
the rate of GTP being hydrolyzed can be written as: ½Pi = k2 3
½Rag0 ½GTP0 t: ½GTP0 + k1k1+ k2
(VI.5)
Let KM hðk1 + k2 Þ=k1 and since the readout from TLC is the fraction of GTP that is hydrolyzed, we have: Fraction of hydrolyzed GTP =
½Pi ½Rag0 ½Rag0 = k2 3 thkobsd 3 t; ½GTP0 ½GTP0 + KM ½GTP0
(VI.6)
in which kobsd = kcat 3 ½GTP0 =ðKM + ½GTP0 Þ is the observed rate constant at a given GTP concentration. Similarly, the measured kcat and Km here are the weighted average of the kcat and Km values for RagA and RagC, which in principle should be close to one another. For GATOR1- and FLCN-FNIP2-stimulated GTP hydrolysis reactions under multiple turnover conditions, increasing amounts of GATOR1 or FLCN-FNIP2 were included in the reaction that monitors the hydrolysis rate of 250 mM GTP by 2 mM wild-type or mutant Rag GTPases. Time points were taken at each GATOR1 or FLCN-FNIP2 concentration to trace the reaction, and analyzed as above. Free Energy Diagram The free energy diagrams in Figures 1R and 1S were constructed based on the experimentally determined reaction rates and equilibrium constants using the standard state of 1 mM. For free-energy change at equilibrium, we used Equation VII: DG = RT ln K;
Equation VII
in which R = 1.987 cal mol1 K1, T = 277 K, and K is the equilibrium constant measured experimentally. For activation barrier, we used Equation VIII: DGz = RT ln
kh ; kB T
Equation VIII
in which R = 1.987 cal mol1 K1, T = 277 K, kB = 3.33 3 1024 cal K1, h = 1.583 3 1034 cal sec, and k is the experimentally measured rate constant. Preparation of Cell Lysates and Immunoprecipitates To determine the response of mTORC1 to amino acid stimulation or starvation under the control of Rag mutants, two million HEK293T cells were plated onto a 10-cm dish. 24 hr later, the cells were transfected with 2 ng of Flag-S6K1 and 100 ng of each Rag mutants.
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e7
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
36 hr later, amino acid stimulation or starvation was performed based on the procedure shown in Figures 6B and 6C. Cells were rinsed once with ice-cold PBS and lysed with Triton lysis buffer (40 mM NaHEPES, pH 7.4; 5 mM MgCl2; 100 mM ATP; 10 mM Na4P2O7; 10 mM Na b-glycerol phosphate; 1% Triton; and one tablet of protease inhibitor cocktail per 25 mL of buffer). The lysates were cleared by centrifugation at 15,000 rpm at 4 C in a microcentrifuge for 10 min. For anti-FLAG immunoprecipitations, the FLAGM2 beads were pre-equilibrated in Triton lysis buffer. 30 mL of a 50/50 slurry of the FLAG-M2 affinity beads was then added to cleared lysates and incubated at 4 C for two hours. Following immunoprecipitation, the beads were washed one time with Triton lysis buffer and 3 times with Triton lysis buffer supplemented with 500 mM NaCl. Immunoprecipitated proteins were denatured by the addition of 50 mL of 2.5 3 SDS buffer, resolved by SDS-PAGE, and analyzed by immunoblotting. western blots were quantified using ImageJ software. For the time-course experiments (Figures 6D and 6E), HEK293T cells were treated and lysed at precise time points, and the pS6K1 level was visualized as the indicator of mTORC1 activity. Western blots were quantified using ImageJ software. Lysine Modification Assay To detect the conformational changes of Rag GTPases in different nucleotide loading state, we adapted a lysine modification assay. Aliquots of 0.5 mg/ml wild-type Rag GTPase were first incubated with 200 mM GDP, 100 mM GDP and 100 mM GTP, or buffer on ice for two hours. 4 mM Sulfo-NHS-acetate was then added to the mixture, to irreversibly modify the amine group on the lysine side chain. The reactions were quenched by 0.2 M Tris-HCl (pH 8.0) after 10 min. The products were digested by chymotrypsin and the peptides were identified by mass spectrometry. Total ion current (TIC) for the peptide that contains modified lysine residue was used to quantify the modification efficiency. QUANTIFICATION AND STATISTICAL ANALYSIS All the kinetic assays were repeated at least three times, and the results were reported with Mean ± SEM.
e8 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability Graphical Abstract
Authors
nl U GDP
[email protected]
g
Transitioning
GDP
GDP
GTP
GTP Maintenance of Inhibition
Inhibited GATOR1 FLCNFNIP2 GTP GDP
GDP
Maintenance of Activation
GDP
g
GDP
nl
in
Transitioning
U
ck
Activated
oc
Lo
ki
ng
GTP
Kuang Shen, Abigail Choe, David M. Sabatini
Correspondence
in
oc
ki
RagC
ck
RagA
GDP
Lo
ng
GDP
Amino acid deficiency
GDP
Amino acid sufficiency
Highlights d
The Rag GTPase dimer uses a ‘‘locking mechanism’’ to regulate its nucleotide state
d
The binding of GTP to one subunit prevents GTP from associating with the other
d
Unexpected GTP binding in the locked conformation leads to stimulation of hydrolysis
d
The locking mechanism is necessary for the normal response of mTORC1 to amino acids
Shen et al., 2017, Molecular Cell 68, 1–14 November 2, 2017 ª 2017 Elsevier Inc. https://doi.org/10.1016/j.molcel.2017.09.026
In Brief The Rag GTPase heterodimer transmits amino acid signals to mTORC1. It is distinct from canonical small signaling GTPases because of its obligate heterodimeric nature. Shen et al. used the kinetic method to analyze communication between Rag GTPase subunits, revealing a unique intersubunit locking mechanism that controls its nucleotide state.
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Molecular Cell
Article Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability Kuang Shen,1,2,3,4 Abigail Choe,1 and David M. Sabatini1,2,3,4,5,* 1Whitehead Institute for Biomedical Research and Massachusetts Institute of Technology, Department of Biology, 9 Cambridge Center, Cambridge, MA 02142, USA 2Howard Hughes Medical Institute, Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 3Koch Institute for Integrative Cancer Research, 77 Massachusetts Avenue, Cambridge, MA 02139, USA 4Broad Institute of Harvard and Massachusetts Institute of Technology, 7 Cambridge Center, Cambridge, MA 02142, USA 5Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.molcel.2017.09.026
SUMMARY
mTOR complex I (mTORC1) is a central growth regulator that senses amino acids through a pathway that converges on the Rag GTPases, an obligate heterodimer of two related GTPases. Despite their central role in amino acid sensing, it is unknown why the Rag GTPases are heterodimeric and whether their subunits communicate with each other. Here, we find that the binding of guanosine triphosphate (GTP) to one subunit inhibits the binding and induces the hydrolysis of GTP by the other. This intersubunit communication pushes the Rag GTPases into either of two stable configurations, which represent active ‘‘on’’ or ‘‘off’’ states that interconvert via transient intermediates. Subunit coupling confers on the mTORC1 pathway its capacity to respond rapidly to the amino acid level. Thus, the dynamic response of mTORC1 requires intersubunit communication by the Rag GTPases, providing a rationale for why they exist as a dimer and revealing a distinct mode of control for a GTP-binding protein.
INTRODUCTION mTOR complex I (mTORC1) is a central regulator of cell growth and proliferation. Diverse inputs, including amino acids, growth factors, and energy levels, converge on mTORC1 to regulate its kinase activity and signal to downstream effectors (Efeyan et al., 2015; Gonza´lez and Hall, 2017; Jewell et al., 2013; Shaw and Cantley, 2006). Upon activation, mTORC1 triggers anabolic processes, such as protein synthesis and ribosomal biogenesis, and inhibits catabolic ones, such as autophagy (Dibble and Manning, 2013). Dysregulation of mTORC1 promotes aberrant growth and contributes to numerous human diseases, including cancer and neurological disorders like epilepsy (Wong, 2013). mTORC1 activation requires two steps controlled by distinct
small GTPases: translocation of mTORC1 to the lysosomal surface and stimulation of its kinase activity. In the presence of amino acids, the Rag GTPases recruit mTORC1 to the lysosome (Sancak et al., 2008; Wang et al., 2005), where, if growth factors are present, the Rheb GTPase turns on its kinase activity (Inoki et al., 2003; Li et al., 2004; Menon et al., 2014; Saito et al., 2005; Saucedo et al., 2003; Stocker et al., 2003). Thus, these two GTPases form an AND gate that ensures that mTORC1 is active only under favorable growth conditions. Within the Ras-like superfamily of GTPases, the Rag GTPases belong to the TRAFAC (translation factor-related) class due to a conserved threonine residue on the lateral side of the nucleotide-binding pocket (Thr35 in Ras; Thr42 and Thr96 in RagA and RagC, respectively) (Leipe et al., 2002; Shan, 2016). However, unlike other characterized members of the superfamily, the Rag GTPases form obligate heterodimers consisting of RagA or RagB bound to RagC or RagD, and so each functional unit can bind two guanine nucleotides (Nakashima et al., 1999; €rmann et al., 1995). Distinct from the signal recognition parSchu ticle (SRP) and its receptor (SR), which toggle between monomeric and dimeric forms during the protein-targeting cycle, there is no evidence that the Rag GTPase heterodimer ever dissociates into stable monomers within cells. Several multi-protein complexes regulate the Rag GTPases, including Ragulator, which localizes them to the lysosomal surface and modulates their nucleotide loading (Sancak et al., 2010), and the GATOR1 (Bar-Peled et al., 2013; Panchaud et al., 2013) and FLCN-FNIP2 complexes (Petit et al., 2013; Tsun et al., 2013), which are GTPase-activating proteins (GAPs) for RagA/B and RagC/D, respectively. Previous studies in cells found that the Rag heterodimer is active and capable of interacting with mTORC1 when RagA/B is bound to guanosine triphosphate (GTP) and RagC/D to guanosine diphosphate (GDP) (GTPRagA-RagCGDP). In contrast, the opposite nucleotide configuration (GDPRagA-RagCGTP) weakens the interaction with mTORC1 so that it does not localize to lysosomes and the pathway is suppressed (Sancak et al., 2008). It is unclear whether, out of the four possible nucleotide states, these are the only two in which the Rag GTPase heterodimer can exist. Moreover, in general, why the Rag GTPases are heterodimeric and whether intersubunit communication occurs and, if so, Molecular Cell 68, 1–14, November 2, 2017 ª 2017 Elsevier Inc. 1
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
RagC
GNP
GNP
Purification Strategy
C
pET Duet-1 (AmpR) ORF1: RagA-His; ORF2: RagC
Nucleotide binding domain (NBD)
Q Sepharose
C-terminal roadblock domain (CRD)
D
RagA-RagC
150
80− MW (kD)
B RagA
OD280 (mAu)
A
Nucleotides
100
Ni-NTA
50
monoQ
[Rag] MW (kD)
F
α-32P-GNP RagA-RagC (trace) (varying amount)
Binding (au)
UV light (254 nm)
Crosslinking assay
chilled surface
50 100 Elution Volume (ml)
46−
RagC×GTP
32−
RagA×GTP
G
46−
RagC×GDP
32−
RagA×GDP
4 105
RagC RagA
4 105
2 105
150
[Rag] MW (kD)
0
Binding (au)
Equilibrium binding assay
RagC RagA
2 105
GTP binding 0
0.01
Autoradiography imaging
0.01
RagA-RagC (μM)
I
RagC×GTP
100 RagA×GTP
varying time
1
106
0
200
0
50
0.10
0.05
0.00
100
Time (min)
K
400 [RagA-RagC] (nM)
1.0
kobsd (min-1)
Binding (au)
RagC×GTP
106
5.0 105
RagA×GTP
0.0
50
RagA RagC
0.2
0.1
0.0
100
P
Time MW (kD)
GDP association 0
46−
RagC×GTP
32−
RagA×GTP
Unlabeled GNP (excess)
1
Crosslinking assay
2 105 RagC RagA
1 105
0
GTP dissociation 0
5000 Time (min)
Summary of binding kinetics (4°C)
-1
-1
kon (M s ) Kd (nM) koff, calc (s-1) † -1
R
GTP
GDP
3
3
GDP GTP 3.5×10 (.2) 5.1×10 (.2) 4.1×103(.2) 2.9×103(.3) 47(12)
76(7)
11(5)
41(6)
1.6×10-4
3.9×10-4
4.5×10-5
1.2×10-4
RagC×GDP
32−
RagA×GDP
105 RagC RagA
5 104
GDP dissociation 0
50
100
-4
-7
150
Time (min)
Free energy profile for GDP binding 19
kon 0 -2
-7
46−
S
RagC
RagA
400
Time
0
10000
ΔG (kcal/mol)
Q
Binding (au)
Binding (au)
varying time
200
RagA-RagC (nM)
-4
koff, measured (s ) 1.2×10 (.6) 5.3×10 (.3) 9.8×10 (.6) 5.6×10 (.4) 1300×↓ 46×↓
koff
Kd empty
Free energy profile for GTP binding
bound
?
22 19
kon 0 -2
koff
Kd empty
intermediate
bound
GTP
GDP GDP
RagA RagC
ΔG (kcal/mol)
RagA-RagC
O
α-32P-GNP (trace)
MW (kD)
Off-rate measurement
400
RagA-RagC (nM)
Time (min)
N
200
200 nM
400 nM 300 nM 100 nM RagA-RagC
0
GTP association 0
M
1.5 106 RagC×GTP
300 RagA×GTP
RagA RagC
200 nM
400 nM 300 nM 100 nM RagA-RagC
RagA×GTP RagC×GTP
10
0.15
L
RagA×GTP
2 106
RagC×GTP
Crosslinking assay
1
RagA-RagC (μM)
J
Time
α-32P-GNP (trace)
0.1
kobsd (min-1)
On-rate measurement RagA-RagC (fixed amount)
GDP binding 0
1
Binding (au)
H
0.1
RagA
25−
0
E
RagC 46− 32−
Void
Gel filtration
58−
GTP
GTP
?
(legend on next page)
2 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
how it interfaces with the actions of the extrinsic regulators are all open questions. Here, we used kinetic analyses to quantitatively dissect the functional cycle of the Rag GTPases. We find that the Rag heterodimer uses a previously undescribed mode of regulation for GTPases—extensive intersubunit communication—to transmit upstream signals to mTORC1. This mechanism has two key properties: the binding of GTP to one of the Rag subunits inhibits the association of GTP to the other, and if an unwanted binding event does occur, the prebound GTP triggers the hydrolysis of the later-bound GTP, which maintains the original state by destabilizing the GTPRagA-RagCGTP configuration. We find that intersubunit communication is critical for the mTORC1 pathway to rapidly respond to the presence and absence of amino acids, providing a possible explanation for the evolutionary advantage of the heterodimeric architecture of the Rag GTPases. RESULTS Binding of GTP to the Rag GTPase Heterodimer Induces Conformational Changes that Inhibit Its Dissociation To examine potential communication between the subunits of the Rag GTPase heterodimer, it was necessary to develop an assay in which the nucleotide state of each could be quantitatively assessed in the context of the intact complex. Each Rag subunit has an N-terminal nucleotide-binding domain (NBD) with a similar structural fold to Ras and a C-terminal roadblock domain (CRD) that mediates heterodimerization (Figure 1A) (De Franceschi et al., 2014; Gong et al., 2011; Jeong et al., 2012; Levine et al., 2013). Expression of individual subunits in bacteria did not yield soluble proteins (data not shown), probably due to the exposure of hydrophobic residues on the CRD that mediates heterodimer formation. In contrast, co-expression of codon-
optimized, C-terminally His-tagged RagA and tagless RagC in the pETDuet-1 vector (Figure 1B) generated a stable complex that could be isolated at the high purity needed for biochemical studies (Figures 1C and 1D). The fact that RagA and RagC have substantially different molecular weights and that UV light can covalently crosslink nucleotides to GTPases (Yue and Schimmel, 1977) enabled us to develop an assay for measuring, within intact complexes, the binding of radiolabeled nucleotides to each subunit at the same time (Figure 1E; STAR Methods). At the end of the binding reaction, we expose samples to UV light and use SDS-PAGE followed by autoradiography to quantify the amounts of radiolabeled nucleotides bound to each Rag subunit (Figure 1F). As expected, unlabeled GTP competed with the binding of radiolabeled guanine nucleotides to RagA and RagC (Figure S1A), and in heterodimers in which RagA or RagC had a mutation that mimics the Ras(S17N) mutant that does not bind GTP (RagA T21N; RagC S75N) (Feig and Cooper, 1988; Sancak et al., 2008), GTP bound only to the wild-type subunit (Figures S1B and S1C). Within the heterodimer, RagC binds GTP with a Kd of 11 nM, 4-fold tighter than RagA, and both bind GDP slightly less well than GTP (Figures 1F and 1G; summarized in Figure 1Q). These Kd values are orders of magnitudes lower than the cellular concentration of GTP (0.7 mM), suggesting that mechanisms must exist to prevent both subunits from being loaded with GTP as the active heterodimer is in the GTPRagA-RagCGDP state. To understand nucleotide binding in more detail, we examined pre-steady-state kinetics by measuring the nucleotide association and dissociation rates of the Rag GTPases (Fersht, 1999; Feuerstein et al., 1987; Nowak and Goody, 1988; Simon et al., 1996). For the on-rate measurement (Figure 1H), we monitored nucleotide association to Rag subunits over time (Figures 1I
Figure 1. Quantitative Measures of the Nucleotide-Binding Properties of the Rag GTPases (A) General architecture and domain organization of the Rag GTPase heterodimer. GNP, guanosine di/triphosphate. (B) Generation of recombinant Rag GTPases in bacteria. (C) Elution profile of the Rag GTPase heterodimer on a HiLoad 16/60 Superdex 200 gel-filtration column. After EDTA stripping, the Rag heterodimer separates from its bound nucleotides and emerges as a single peak as indicated. (D) SDS-PAGE and Coomassie blue staining analysis of purified Rags. (E) Experimental setup of the crosslinking assay for measuring the nucleotide binding affinity of each Rag subunit. (F and G) GTP (F) and GDP (G) binding to the Rag GTPases. Top: nucleotide bound to RagA (lower bands, molecular weight [MW] = 37 kDa) and RagC (upper bands, MW = 44 kDa) visualized by autoradiography. Bottom: quantification of the radioactive signals from the top panels. Data were fit to a single-site binding equation to obtain the dissociation constant Kd. (H) Experimental setup of the crosslinking assay for measuring the on rates of nucleotides to the Rag GTPases. (I) Reaction time course of GTP bound to Rag subunits during an on-rate measurement. Four Rag concentrations (100, 200, 300, and 400 nM) are shown from top to bottom, where time points were taken to monitor the reaction process. At higher Rag concentrations, reactions reach equilibrium faster. (J and K) Quantification of the observed rate constants for GTP association to RagA (J) and RagC (K) at different Rag concentrations. A single exponential rise function was used to fit the time points shown in (I). Curve colors match with the corresponding Rag concentrations in (I). (L and M) On rates for GTP (L) and GDP (M) to the Rag GTPase heterodimer. Linear regression of the observed rate constants (values obtained from J and K for GTP association and Figures S1E and S1F for GDP association) against Rag concentration gives the slope (kon) for RagA (red) and RagC (blue). (N) Experimental setup of the crosslinking assay for measuring the off rates of nucleotides from the Rag GTPases. (O and P) Reaction time course of radioactively labeled nucleotides remaining bound to Rag subunits during the off-rate measurement of GTP (O) and GDP (P) dissociation. Top: SDS-PAGE resolves nucleotide bound to RagA (lower bands) and RagC (upper bands) in the presence of unlabeled nucleotides as a chase. Bottom: quantification. Data were fit to a single exponential decay to extract the off rates (koff). (Q) Summary of binding kinetics of nucleotides to the Rag GTPases at 4 C. Gray numbers in parenthesis denote the SDs of the reported values calculated from at least three independent experiments. (R and S) Free energy profiles and models for GDP (R) and GTP (S) binding to the Rag GTPases. In the case of GDP binding (R), the matched predicted and measured koff values suggest a single energy barrier. In the case of GTP binding (S), a second barrier is required to explain the discrepancy between the predicted and measured koff values.
Molecular Cell 68, 1–14, November 2, 2017 3
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
A
B
Rag mutants and properties
Binding preference RagC RagA RagA(D130N)-RagC Xanthine Guanine RagA-RagC(D181N) Guanine Xanthine Mutants
C +XTP
D
Sequential chase assay
[RagA(D130N)-RagC] RagC×GTP
RagA(D130N) -RagC
[RagA(D130N)-RagC]
α-32P-GTP (trace) XNP (excess)
RagC×GTP
+XDP
+XTP
RagC×GTP
+XDP
RagC×GTP
Pre-bound GTP
1 hour
Sequential binding assay 2 106
α-32P-GTP (trace) 1 hour
Crosslinking assay
+XTP +XDP
1 106 Binding (au)
Binding (au)
RagA(D130N) XNP -RagC (1.2 eq of Rag)
1 106
+XTP +XDP
5 105
Crosslinking assay 0
0
RagA-RagC
F
Cold GNP (1.2 eq of Rag)
Stabilize GNP
RagA-RagC
or RagA-RagC
α-32P-GNP (trace)
GNP
0.06
0.04
Varying time Crosslinking assay
J
Cold XNP RagA(D130N)(1.2 eq of Rag) RagC Stabilize XNP
RagA(D130N)-RagC
α-32P-GTP (trace)
0.08
0.04 0.02
GTP association 0
Crosslinking assay
M
GTP binding to RagA
Locked
Summary of half-site on-rates (4°C) RagA-RagCN Prebound N= GTP GDP
kon (M-1 s-1)
0.05
K
+XTP +XDP
0.06
0.00
Varying time
0.01 0.1 1 RagA(D130N)-RagC (μM)
N RagA-RagC GDP GTP
GTP 1.1×10-4 s-1 890(65) 1.1×10-4 s-1 460(35) GDP
580(67) 1100(70) 229(23)
980(99)
GDP association 1.0
2.0
RagA-RagC (μM)
kobsd (min-1)
Differential on-rate for GTP
kobsd (min-1)
I
0.10
0.00 0.0
0.5 1.0 RagA-RagC (μM)
H
RagA-RagCGDP GDP RagA-RagC RagA-RagCGTP GTP RagA-RagC
0.15
GTP association
0.02
0.00 0.0
G
RagA-RagCGDP GDP RagA-RagC RagA-RagCGTP GTP RagA-RagC
kobsd (min-1)
Half-site binding assay
kobsd (min-1)
E
0.01 0.1 1 RagA(D130N)-RagC (μM)
Summary of on-rates for GTP (4°C) Prebound N= XTP -1
-1
kon (M s )
0.02
0.00
1 2 3 RagA(D130N)-RagC (μM)
L
+XTP +XDP
0.04
XDP
RagA-RagC(D181N)N 103(13) 231(25) N
RagA(D130N)-RagC 309(13) 609(40)
GTP association 0
1 2 3 RagA-RagC(D181N) (μM)
Prevention of GTP binding to RagC
GTP GTP RagC
RagA
GDP
GDP
open
GTP
GDP
closed
GTP
GDP
closed
Figure 2. Intersubunit Crosstalk Maintains the Singly GTP-Loaded State of the Rag GTPase Heterodimer (A) Experimental setup for a sequential binding assay to probe the effect of xanthine nucleotide-loaded RagA on guanine nucleotide binding to RagC. Rag mutants with modified hydrogen-bonding pattern decouple nucleotide binding to the subunits. (B) XTP loading on RagA(D130N) inhibits GTP binding to RagC. Top: GTP binding to RagC when RagA(D130N) is prebound to XTP or XDP. Bottom: quantification. When RagA(D130N) binds XTP, RagC loads a significantly smaller amount of GTP than when it binds XDP. (C) Experimental setup of a sequential chase assay to probe the effect of later-bound nucleotides on the first-bound nucleotide. (D) Later-bound XTP to RagA(D130N) has no effect on the amount of GTP prebound to RagC. Top: prebound GTP to RagC chased by XTP or XDP binding on RagA(D130N). Bottom: quantification. Later-bound XTP or XDP on RagA(D130N) has no effect on the prebound GTP. (E) Experimental setup of a half-site binding assay to measure the on rates of nucleotides to singly loaded Rag GTPases. Unlabeled nucleotides were added at sub-equal molar amounts compared to the nucleotide-binding pockets to generate singly loaded Rag GTPases. (F) Half-site on-rate measurements for GTP association when the Rags are singly loaded with GTP (filled circle) or GDP (open circle). The subunit in bold is the subunit associated with the labeled GTP in the assay to measure the half-site on rate. For example, RagA-RagCGDP (open red circle) shows how fast RagA associates with labeled GTP when RagC is preloaded with GDP. The slope gives the half-site on rates. Notice that concentration dependence is lost in the case of GTP. (G) Half-site on-rate measurements for GDP association when the Rags are singly loaded with GTP (filled circle) or GDP (open circle). As before, subunits in bold indicates which subunit is binding the labeled GDP in the assay to measure the half-site on rate. GDP associates with the Rag GTPases more slowly if one subunit is preloaded with GTP. (legend continued on next page)
4 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
and S1D). Linear regression of the observed reaction rates (kobsd; Figures 1J and 1K for GTP and Figures S1E and S1F for GDP) against Rag concentrations generated the secondary reaction rate constants kon, which reveal that RagA and RagC associate with guanine nucleotides at very similar rates (Figure 1L for GTP and Figure 1M for GDP; summarized in Figure 1Q). We measured the off rate in a pulse-chase setup (Figure 1N), in which preloaded labeled GTP on the Rags was chased by an excess of unlabeled GTP (koff; Figure 1O for GTP and Figure 1P for GDP). Despite the similar binding affinities and on rates, GTP dissociates from the Rags much more slowly than GDP, with a 600- to 4,000-fold difference (Figure 1Q). Careful examination of the kinetics data revealed a surprising discrepancy (Figure 1Q). If the binding of nucleotides to the Rag GTPases involves a single transition state, then the dissociation constant, and the on and off rates, should conform to the equation Kd = kon/koff. This prediction holds true for GDP, where the calculated koff matches well with the measured koff (Figure 1Q), suggesting that GDP binding does indeed overcome one transition state (Figure 1R). However, the measured koff for GTP is 1,300- and 46-fold slower than the calculated one for RagA and RagC, respectively (Figure 1Q; GTP koff, calc versus koff, measured). This discrepancy suggests that the binding of GTP, but not GDP, induces conformational changes in the heterodimer beyond the local ones on switch I and switch II that are common to small GTP-binding proteins (Hall et al., 2002; Vetter et al., 1999) and that these prevent GTP dissociation (Figure 1S). To our knowledge this second transition state has not been observed in monomeric GTPases, suggesting that it is a result of the heterodimeric architecture of the Rag GTPases. Binding of a Single GTP to the Rag GTPase Heterodimer Inhibits Binding of a Second GTP To understand the consequences of GTP binding to the Rag heterodimer, it was necessary to characterize its singly GTP-loaded state in vitro by combining the crosslinking assay with Rag mutants that preferentially bind xanthine instead of guanine nucleotides (RagA D130N; RagC D181N) (Figure 2A) (Hwang and Miller, 1987; Sweet and Gerace, 1996). Surprisingly, when RagA(D130N) in the RagA(D130N)-RagC heterodimer was bound to xanthosine 50 -triphosphate (XTP) instead of xanthosine 50 -diphosphate (XDP), the wild-type RagC partner bound less GTP (Figure 2B). In the reverse direction, this observation also holds for the RagA-RagC(D181N) heterodimer (Figure S2A), suggesting that nucleoside triphosphate binding to one subunit (RagA or RagC) inhibits nucleoside triphosphate binding to the other (RagC or RagA). As a control experiment, we switched the sequence of binding (Figure 2C); once RagC was already
loaded with GTP, the subsequent binding of XTP or XDP to RagA(D130N) had no observable effect on the amount of preloaded GTP (Figure 2D), indicating that the binding of the first nucleoside triphosphate has dominant consequences. These findings suggest that the binding of GTP to either RagA or RagC could alter the conformation of the entire heterodimer to induce a ‘‘locked state’’ that suppresses the association of GTP to the other subunit (negative cooperativity). To verify this possibility, we directly measured the on rate of GTP to singly loaded wild-type Rag GTPase heterodimers using a half-site binding assay (Figures 2E–2H and S2B–S2E). We reasoned that if the binding of the first GTP inhibits the association of the second GTP, then we should observe a slower on rate. This is indeed the case: when RagC is bound to GDP, RagA associates GTP at 890 M1s1 (Figures 2F and 2H). When RagC is loaded with GTP, however, the on rate no longer exhibits concentration dependence (Figure 2F) and gives a first-order rate constant of 1.1 3 104s1. Similar results were obtained using the xanthine-specific mutants described above (Figures 2I–2L and S2F–S2I); when the xanthine-specific subunit prebinds XTP instead of XDP, the on rate of GTP to the guanine-specific subunit is slower (Figure 2J for RagC and Figure 2K for RagA; summarized in Figure 2L). These results are consistent with the Rag heterodimer being in a conformation that is not permissive to the direct binding of the substrate when it is singly loaded with GTP. In this locked state, the transient opening of the binding pocket becomes rate limiting, making the ‘‘on rate’’ first order and thus independent of Rag concentration (Figures S2J and S2K) (Fersht, 1999). We extended the half-site on-rate measurement to GDP association and found, to a lesser extent, a decrease in on rates when one subunit was preloaded with GTP (Figures 2G and 2H). However, the observed rate for GDP association is still dependent on the Rag concentration, likely because GDP is smaller so that even in the locked state the Rags cannot fully exclude it from the partially shielded nucleotide-binding site. To directly probe the conformations of the Rag GTPases in different nucleotide-loading states, we utilized a lysine modification assay (Figure S2L), where the extent of lysine modification by sulfo-N-hydroxysuccinimide (NHS)-acetate correlates with its solvent accessibility (Guo et al., 2008). When both subunits bind GDP, the Rag GTPases have a modification profile very similar to that of empty Rags (log2 ratio 0; Figure S2M). However, in the presence of 1:1 GTP/GDP, several lysine residues show a significantly lower level of modification (Figure S2M). For example, Lys20 of RagA and Lys74 of RagC, both localizing within the nucleotide-binding pocket, are much less modified than without GTP, consistent with a local conformational change
(H) Summary of half-site on rates for wild-type Rag GTPases at 4 C. Gray numbers in parentheses denote the SDs of the reported values calculated from at least three independent experiments. (I) Experimental setup for measuring the on rates of nucleotides to the guanine-specific subunit when the xanthine-specific subunit is preloaded. (J) On rate of GTP association to RagC when RagA(D130N) was preloaded with XTP (filled circle) or XDP (open circle). (K) On rate of GTP association to RagA when RagC(D181N) was preloaded with XTP (filled circle) or XDP (open circle). (L) Summary of differential on rates for GTP at 4 C. Gray numbers in parenthesis denote the SDs of the reported values calculated from at least three independent experiments. (M) Illustration of the locked conformation induced by the binding of the first GTP to RagA and its inhibition of the association of the second GTP with RagC. The reverse case is shown in Figure S2Q.
Molecular Cell 68, 1–14, November 2, 2017 5
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
KH2PO4
Quench
C
Time
Pi
GTP
0.15
D
20 nM RagA-RagC 10 nM 5 nM 2 nM 1 nM 0.5 nM
0.10 0.05
0.0003
kobsd (min-1)
Varying time
B
Fraction GTP hydrolyzed
Single turnover GTP hydrolysis
RagA-RagC γ-32P-GTP (varying amount) (trace)
Expansion direction
A
0.00 0
20 nM RagA-RagC
TLC
200 400 Time (min)
0.0002
0.0001
0.0000
600
0
5
10
15 20 RagA-RagC (nM)
100
1000
Autoradiography
F
G
Time
γ-32P-GTP (trace) + Cold GTP (varying amount)
RagA-RagC (fixed amount) Varying time
Pi
GTP
KH2PO4
0.10 0.05
2 μM GTP 5 μM 10 μM
0.00
2 μM GTP
Quench
H
0.15
0
500 1000 Time (min)
0.003
kobsd (min -1)
Multiple turnover GTP hydrolysis
Fraction GTP hydrolyzed
E
20 μM 50 μM 100 μM
0.002
0.001
0.000
1500
0
50 GTP (μM)
100
TLC
Autoradiography
J
RagA-RagC(S75N) RagA(T21N)-RagC WT
0.002
50 RagA-RagC (μM)
Half-site GTP hydrolysis
Cold GNP RagA-RagC (fixed amount) (1.2 eq of Rags) Stabilize
γ-32P-GTP (trace)
0.002
0.000
100
M
0.3
0
50 GTP (μM)
O
No nucleotides +GTP +GDP +GppNHp
0.2
400
Observed hydrolysis rates (per min) No NT +GTP +GDP +GppNHp
R
200 Time (min)
1.1×104(.3) 1.7×104(.4)
Half-site hydrolysis chase γ-32P-GTP (trace) Cold GNP (excess)
Initial
P
Chased
0.10
0.05
+Cold GNP
0.00 0
200 Time (min)
+Buffer +GTP +GDP +GppNHp
400
Observed chased rates (per min) +Buffer +GTP +GDP +GppNHp
Unexpected GTP binding to RagC
1.2×104(.2) 3000×↑
RagA(T21N)-RagC 1.9×10-3(.2) 5.2×10-3(.2)
Chased time points (- - -) Quench and TLC analysis
Q
2.2×10-4 1.0×10-3 2.7×10-4 7.7×10-4
4.0(.8)
Multiple Turnover
1.6×104(.5) 8.4×103(2)
Initial time points (- - -) Quench and TLC analysis 0
-3
2.2×10 (.2) 3.0×10 (.1) 14×↑
Start reaction
0.1
Single Turnover
RagA-RagC(S75N) 3.7×10-3(.4) 3.7×10-3(.4)
RagA-RagC (fixed amount)
0.0
Quench and TLC analysis
100
KM (nM)
Multiple Turnover
-4
Wildtype
Varying time
N
kcat (min-1)
Multiple Turnover
Fraction Pi
L
0
Summary of GTP hydrolysis kinetics (25°C)
Single Turnover
Single Turnover 0.000
K
RagA-RagC(S75N) RagA(T21N)-RagC WT
0.004
Fraction Pi
kobsd (min-1)
0.004
kobsd (min-1)
I
2.2×10-4 2.0×10-4 2.2×10-4 2.1×10-4
Stimulated hydrolysis
GTP GTP RagA
GDP
GTP
GT P
GTP
GDP
RagC
Figure 3. Directional GTP Hydrolysis Maintains the Functional State of the Rag GTPases (A and E) Experimental setup for the single- (A) and multiple-turnover (E) GTP hydrolysis assays. A saturating amount of the Rag GTPases was incubated with a trace amount of labeled GTP to ensure single-turnover conditions (A). A saturating amount of GTP was incubated with the Rag GTPases to ensure multipleturnover conditions (E). (B and F) Thin-layer chromatography to separate radioactive phosphate from unhydrolyzed GTP at different time points. 20 nM Rag GTPases hydrolyzed a trace amount of labeled GTP in a single-turnover assay (B). 2 mM Rag GTPases hydrolyzed 2 mM GTP in a multiple-turnover assay (F). (C and G) Quantification and linear regression of the fraction of hydrolyzed GTP yields the observed rate constants (kobsd) at different Rag (C) or GTP (G) concentrations. (D and H) Michaelis-Menten fittings of kobsd against Rag or GTP concentration yield kcat and KM for the single-turnover (D) and multiple-turnover (H) reactions. (I and J) Single-turnover (I) and multiple-turnover (J) GTP hydrolysis reactions for RagA-RagC(S75N) (red circle) and RagA(T21N)-RagC (blue circle). Dashed curves show the activity of wild-type Rags as a reference. (legend continued on next page)
6 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
upon GTP binding that is prevalent among Ras-family GTPases. Another region that displays lower levels of modification is Lys230 of RagA and Lys258 of RagC (Figures S2N and S2O). Based on a homologous structure (Figure S2P), these residues localize to the hinge region connecting the NBD and CRD (Gong et al., 2011; Jeong et al., 2012). The significant reduction in the modification of these residues indicates that the hinge region is more shielded than in the GDP-bound state, suggesting that GTP binding induces a more compact conformation. Thus, in aggregate, these data indicate that when one subunit binds GTP, the Rag heterodimer rearranges into a more compact (‘‘locked’’) conformation that can account for the second transition state on the free-energy plot (Figure 2M) and decouples kon and koff. Binding of a Single GTP to the Rag Heterodimer Induces Hydrolysis of a Second GTP Despite the existence of the locked state, we wondered what would happen if both Rag subunits did manage to bind GTP. For example, if in the active heterodimer (GTPRagA-RagCGDP) RagC loses its GDP and misloads with GTP, is there a mechanism to promote hydrolysis of the newly bound GTP, or do both subunits have an equal chance of hydrolyzing their GTPs? If the latter were the case, the heterodimer could flip from the active to inactive state, which might be undesirable. We monitored labeled GTP hydrolysis using thin-layer chromatography (TLC) in both single- and multiple-turnover setups (Figures 3A and 3E) (Peluso et al., 2001). In single-turnover assays, a trace amount of GTP is incubated with an excess of the Rag GTPases, so that only a single subunit in each heterodimer will have the opportunity to hydrolyze GTP once (Figure 3A). In multiple-turnover assays, the Rag GTPases are saturated with excess GTP, so that both subunits can hydrolyze GTP for multiple rounds (Figure 3E). These two setups yielded completely different kinetic parameters (Figures 3B–3D, 3F– 3H, S3B, and S3C; summarized in Figure 3K). The Rag GTPases hydrolyzed GTP 14-fold faster in the multiple-turnover assay than in the single-turnover assay (Figure 3K, kcat), suggesting that the GTPRagA-RagCGTP state is unstable. Moreover, the KM for GTP was also 3,000-fold higher in the multiple-turnover assay than in the single-turnover assay (Figure 3K), showing that the association of the second GTP is kinetically unfavorable, as predicted by the locking model proposed above. In the same assays, heterodimers containing RagA or RagC mutants with
impaired GTP binding (cf. Figures S1B and S1C) showed no differences between single- and multiple-turnover assays (Figures 3I–3K and S3D–S3G). We also probed how the order of GTP binding to the Rag heterodimer impacts hydrolysis by its subunits. We first checked the hydrolysis of the later bound GTP using a half-site hydrolysis setup (Figure 3L), where we loaded half of the nucleotide-binding pockets in the Rag heterodimer with unlabeled GTP or GDP, so that labeled GTP can only occupy and be hydrolyzed by the remaining site. When the Rag GTPases were singly loaded with GTP or GppNHp, the hydrolysis rate on the other subunit was 4-fold faster than when loaded with GDP (Figures 3M and 3N). To measure the hydrolysis rate of the prebound GTP, we switched the reaction order so that labeled GTP was first loaded on the Rag GTPases, and then an excess of unlabeled GTP or GDP was added as a chase to occupy the remaining site (Figure 3O). In this scenario, no acceleration in the reaction rate was observed regardless of the chase. The prebound GTP was hydrolyzed at a constant, slow speed, similar to that in a single-turnover setup (Figures 3P and 3Q), suggesting the later-loaded GTP has no effect on the prebound GTP. Thus, the prebound GTP hydrolyzes itself at a slow rate while causing accelerated hydrolysis at the other nucleotide-binding pocket. Combining these results, we conclude that a single GTP-binding event on one subunit is necessary and sufficient to drive the heterodimer into a locked state that inhibits the association and promotes the hydrolysis of GTP by the other subunit (Figure 3R). Thus, ‘‘intersubunit communication’’ maintains the Rag heterodimer in its active or inactive state, ensuring the stability of each and pointing to the necessity of regulators to interconvert between them. Rag GAPs Interconvert between the Functional States of the Rag GTPase Heterodimer The consequences of intersubunit communication and the slow intrinsic GTP dissociation and hydrolysis rates of the Rags should make transitions between the two stable states intrinsically challenging. These findings clearly point to a key role for extrinsic regulators in modulating the Rag GTPases in response to nutrients. Indeed, two GAPs have been discovered for the Rags: GATOR1 (Bar-Peled et al., 2013; Panchaud et al., 2013) and FLCN-FNIP2 (Petit et al., 2013; Tsun et al., 2013) for RagA/B and RagC/D, respectively (Figure 4A). To test the hypothesis
(K) Summary of GTP hydrolysis kinetics for the Rag GTPases at 25 C. Gray numbers in parenthesis denote the SDs of the reported values calculated from at least three independent experiments. (L) Experimental setup for a half-site GTP hydrolysis reaction. (M) Half-site GTP hydrolysis in the presence of GTP (red), GDP (blue), or GppNHp (orange). Radioactively labeled GTP is hydrolyzed on one subunit of the Rag GTPases when the other is preloaded with nucleotides. Preloaded GTP and GppNHp trigger hydrolysis by the other subunit. Control (black) contains no preloaded nucleotides. (N) Summary of the observed rate constants for half-site GTP hydrolysis reactions in (M). (O) Experimental setup for a half-site GTP hydrolysis chase. Labeled GTP was preloaded and hydrolyzed on one subunit of the Rag GTPases before an excess amount of cold nucleotides was added to occupy the remaining binding sites. (P) Half-site GTP hydrolysis chased by GTP (red), GDP (blue), or GppNHp (orange). No difference in the slope was observed. Pure buffer was added to the control (black). (Q) Summary of the observed rate constants for half-site GTP hydrolysis chase in (P). (R) Illustration of the effect of the locked conformation on GTP hydrolysis. GTP binding to RagA triggers GTP hydrolysis by RagC. The reverse case is shown in Figure S3H.
Molecular Cell 68, 1–14, November 2, 2017 7
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
B
46−
Nprl2
FLCN
46− 32−
0.2 0.1
RagA-RagC
0.0 0
25−
Rags only 10 nM GATOR1 20 nM 50 nM 100 nM 250 nM
0.1
0
200 Time (min)
Rags only 100 nM FLCN-FNIP2 200 nM 500 nM 1 μM 2 μM
F 0.4
0.2
G
RagA-RagC(Q120L)
0.0 0
200 Time (min)
0.000
400
400
H
RagA-RagC RagA-RagC(Q120L) RagA(Q66L)-RagC RagA(Q66L)-RagC(Q120L)
0.005
0
100 200 GATOR1 (nM)
0.4
0.2 0.1 0.0 0.0
0.2
RagA-RagC
200 Time (min)
100 Single Turnover 10-2
400
+FLCN-FNIP2 +GATOR1 No GAP
10-4 RagA- RagAQ66L- RagA- RagAQ66LRagC RagC RagCQ120LRagCQ120L
300
Multiple Turnover
Single Turnover +FLCN-FNIP2
0.3
0.4
0
Single Turnover +GATOR1
0.010
RagA(Q66L)-RagC
0.0
Fraction Pi
kobsd (min-1)
0.2
E
kobsd (min-1)
Fraction GTP hydrolyzed
0.3
Rags only 10 nM FLCN-FNIP2 20 nM 50 nM 100 nM 200 nM 500 nM
0.0
200 400 Time (min)
32−
D
Fraction Pi
Nprl3
58−
FNIP2
kcat (min-1)
MW (kD)
80−
135− 100− 80− 58−
Fraction GTP hydrolyzed
Depdc5
190− 135− 100−
0.6
C
Rags only 10 nM GATOR1 20 nM 50 nM 100 nM 250 nM
0.3
kcat (min-1)
A
10-1
+FLCN-FNIP2 +GATOR1 No GAP
10-2 10-3
RagA- RagAQ66L- RagA- RagAQ66LRagC RagC RagCQ120LRagCQ120L
RagA-RagC RagA-RagC(Q120L) RagA(Q66L)-RagC RagA(Q66L)-RagC(Q120L)
0.5 1.0 1.5 FLCN-FNIP2 (μM)
2.0
Figure 4. Regulators of the Rag GTPase Heterodimer Control Its Nucleotide-Loading State (A) SDS-PAGE followed by Coomassie blue staining of recombinant GATOR1 and FLCN-FNIP2 complexes indicating their subunits. (B and C) Single-turnover reactions of GATOR1- (B) and FLCN-FNIP-stimulated (C) GTP hydrolysis of wild-type Rags. GATOR1 and FLCN-FNIP dose-dependently increase GTP hydrolysis by wild-type Rags. (D) GATOR1-stimulated GTP hydrolysis of RagA(Q66L)-RagC mutant in a single-turnover condition. The leucine substitution in RagA abolishes the stimulation by GATOR1. (E) Concentration dependence of GATOR1-stimulated GTP hydrolysis under single-turnover conditions for wild-type Rags (black), RagA(Q66L)-RagC (blue), RagA-RagC(Q120L) (orange), and RagA(Q66L)-RagC(Q120L) (red). (F) FLCN-FNIP2-stimulated GTP hydrolysis of RagA-RagC(Q120L) mutant in a single-turnover condition. The leucine substitution in RagC abolishes the stimulation by FLCN-FNIP2. (G) Concentration dependence of FLCN-FNIP2-stimulated GTP hydrolysis under single-turnover conditions for wild-type Rags (black), RagA(Q66L)-RagC (blue), RagA-RagC(Q120L) (orange), and RagA(Q66L)-RagC(Q120L) (red). (H) Summary of hydrolysis kinetics for GATOR1- or FLCN-FNIP2-stimulated reactions.
that these GAPs accelerate state conversion of the Rag heterodimer to match the temporal requirements of mTORC1 signaling, we measured the kinetic parameters for GAP-stimulated GTP hydrolysis reactions. For wild-type Rags, GATOR1 and FLCN-FNIP2 stimulate GTP hydrolysis in a dose-dependent manner (Figures 4B and 4C; summarized in Figure 4K). The rate enhancement of GATOR1 or FLCN-FNIP2 is 45-fold (Figure 4B) and 1,000-fold (Figure 4C) in a single-turnover reaction and 20-fold (Figure S4E) and 50-fold (Figure S4M) in a multipleturnover reaction, respectively. The stimulated hydrolysis is much faster than the intrinsic hydrolysis of the Rag GTPases, which brings the half-life of the activated state from over 8 hr down to 12 min (37 C) and of the inactive state to 30 s. These rates are consistent with the timescales at which mTORC1 signaling responds to amino acids. Therefore, we concluded that GATOR1 and FLCN-FNIP2 resolve the locked
8 Molecular Cell 68, 1–14, November 2, 2017
states (activated or inactive), facilitating the Rags for stateconversion. In Ras-like GTPases, a glutamine residue (Gln61 in Ras) serves as the general base for GTP hydrolysis and chelates the nucleophilic-attacking water to stabilize the transition state (Pai et al., 1990; Scheffzek et al., 1997). We analyzed the corresponding glutamine mutants of the Rag GTPases in the GAP-stimulated GTP hydrolysis reaction. As expected, GATOR1 acts specifically on RagA, because RagA(Q66L)-RagC (Figures 4D and 4E), but not RagA-RagC(Q120L) (Figure S4C), loses GATOR1-dependent stimulation. In contrast, RagA-RagC(Q120L) (Figure 4F and 4G), but not RagA(Q66L)-RagC (Figure S4J), no longer responds to FLCN-FNIP2. These results suggest that the conserved glutamine residues strongly participate in GAP-stimulated GTP hydrolysis and that the two GAPs likely reposition this residue on the corresponding Rag to facilitate GTP hydrolysis.
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Point Mutations in Switch I of the Rag GTPases Disrupt Intersubunit Communication As discussed above, intersubunit communication by the Rag heterodimer provides a mechanism for maintaining its subunits loaded with opposite nucleotides. To understand the biological relevance of this mechanism, we sought mutants that disrupt intersubunit communication. By systematically characterizing the biochemical properties of recurring RagC mutations found in follicular lymphoma, we identified two interesting mutants, RagC(L91P) and RagC(K84T), in switch I (Figure 5A) (Long et al., 2016; Okosun et al., 2016). The affinity of guanine nucleotides to these mutants closely resembles that of the wild-type proteins (Figures 5B, 5D, and S5A–S5E), suggesting that they maintain an intact nucleotide-binding pocket. However, in the RagA-RagC(L91P) heterodimer, the off rate of GTP for RagC increases by 10-fold, while that for GDP remains similar (Figures 5C, 5D, and S5F). These results suggest that the binding of GTP to RagC(L91P) does not drive the Rag heterodimer into the locked conformation that should have inhibited its dissociation. If this were true, the previously observed properties of the locked state should diminish: preloaded GTP should not inhibit the second GTP from associating and, if it does, trigger its hydrolysis. We used the half-site setup to measure the on rate of the second GTP (Figure 2E). For wild-type Rags with normal intersubunit communication, the second GTP association does not have concentration dependence (Figure 5E, dashed line; cf. Figure 2F), but for the RagA-RagC(L91P) mutant, the concentration dependence is partially restored for both subunits (Figures 5E and 5G), suggesting that the first GTP binding does not effectively drive the heterodimer containing this mutant into the closed conformation. Moreover, the rate enhancement in a multiple-turnover GTP hydrolysis reaction (Figure 5I), but not in a single-turnover reaction (Figure 5H), is partially suppressed for RagA-RagC(L91P), suggesting that the prebound GTP does not efficiently trigger hydrolysis of the later-bound GTP. In addition, the half-site hydrolysis stimulation by GTP is impaired for this mutant (Figures S5M and S5N). Therefore, we conclude that the L91P mutation in RagC partially disrupts intersubunit communication (Figure 5K). Tests of the analogous mutant in RagA, RagA(R37P), and of the RagC(K84T) mutant, led to similar conclusions (Figures 5D–5J), suggesting that mutations in either Rag subunit can disrupt the locking process. Rag Mutants Deficient in Intersubunit Communication Impair Regulation of mTORC1 by Amino Acids We took advantage of the mutants defective in intersubunit communication to determine its importance in mTORC1 signaling in cells. We first focused on the interaction of the Rag heterodimer with mTORC1, which is strongest when RagA is loaded with GTP and RagC with GDP. As expected, heterodimers containing the RagA(Q66L) mutant that does not respond to GATOR1-stimulated GTP hydrolysis or the RagC(S75N) mutant that favors GDP binding interacted better with mTORC1 than the wild-type proteins, while mutants having the opposite effects bound worse (Figures 6A and S6A) (Sancak et al., 2008). Interestingly, the communication-defective RagA mutant (RagA R37P), despite its capability to bind GTP, bound as poorly
to mTORC1 as the RagA(T21N) mutant that is deficient in GTPbinding, suggesting that locking is required for RagA to maintain the GTP-loaded state needed to associate with mTORC1 (Figure 6A). In contrast, heterodimers containing the communication-defective RagC mutants L91P or K84T interacted better with mTORC1 than those with wild-type RagC, indicating that in the absence of the locking mechanism, RagC likely reverts to the GDP-bound state that promotes the interaction with mTORC1 (Figure 6A). Thus, mutants that do not affect nucleotide binding but prevent intersubunit communication have strong effects on the Rag-mTORC1 interaction by preventing the Rag heterodimer from achieving the locked states. We next examined the effects of the mutants on the phosphorylation of the mTORC1 substrate S6K1 in response to amino acid stimulation and starvation. In HEK293T cells expressing the control protein Metap2 or the wild-type Rags, amino acid stimulation and deprivation regulated mTORC1 activity, as monitored by the phosphorylation of its substrate, S6K1 (Figures 6B and 6C). As expected, mutants that alter the nucleotide-binding preferences of the Rags strongly impacted mTORC1 signaling; the RagA(T21N) and RagA(Q66L) mutants rendered the pathway unresponsive to amino acids (Figure 6B), while RagC(S75N) prevented the turning off of mTORC1 signaling normally caused by amino acid starvation (Figure 6C). Mutants that block intersubunit communication also caused severe signaling defects. The expression of RagA(R37P) prevented mTORC1 activation in response to amino acid stimulation (Figure 6B), while that of RagC(L91P) and RagC(K84T) suppressed the inhibition of mTORC1 signaling that normally occurs upon amino acid deprivation (Figure 6C). Moreover, in cells expressing the communication-defective mutants, the response rate of mTORC1 signaling to amino acid stimulation or starvation was also blunted (Figures 6D and 6E). These observations correlate well with the biochemical and protein interaction assays and reveal that mTORC1 signaling in response to amino acids is distorted if the Rag heterodimer cannot achieve the locked state (Figure 6F). DISCUSSION As the convergence point of diverse amino acid signals, the heterodimeric Rag GTPases have a central role in controlling the localization and activation of mTORC1. Here, we used kinetic approaches to analyze the intersubunit communication between the subunits within the Rag heterodimer and found that its unique architecture allows for crosstalk between the subunits that is essential for the normal function of the mTORC1 pathway. Specifically, when one subunit binds GTP, it triggers a conformational change that makes it dominant over the other and drives the Rag heterodimer into a locked conformation. This state suppresses the association of a second GTP, and, even if binding does occur, induces its faster hydrolysis. This intersubunit communication is a unique mode of regulation for GTPases and informs how the Rag GTPases transmit amino acid sufficiency to mTORC1. At the molecular level, how do the Rag GTPases attain the locked conformation? Combining insights from our biochemical analyses with those from the previously solved structures of the
Molecular Cell 68, 1–14, November 2, 2017 9
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
32−
RagA×GTP
2 105
E kobsd (min-1) kobsd (min-1)
H
** **
0.02
GTP association
I
0.0002 WT RagA(R37P)-RagC RagA-RagC(L91P) RagA-RagC(K84T)
0
20 40 60 80 Rag GTPases (nM)
GDP
GDP RagA
GTP
4000
Wildtype
47
76
11
41
RagC(L91P)
61
60
19
47
RagC(K84T)
26
65
7.6
45
RagA(R37P)
61
101
6.5
47
Summary of half-site on-rates (4°C) RagA-RagCN GDP Prebound N= GTP
* *
RagC(L91P)
170
1060
140
1250
ns
kon (M-1 s-1)
RagC(K84T)
400
1970
570
1710
RagA(R37P)
120
650
120
620
2.0
Summary of GTP hydrolysis kinetics (25°C)
J
0.003 0.002
0
GDP
50 GTP (μM)
Loop closing (Intrasubunit) GTP GT
100
kcat (min-1)
KM (nM)
Single Multiple Turnover Turnover
Single Multiple Turnover Turnover
Wildtype
2.2×10-4 3.0×10-3
4.0
12000
RagA-RagC(L91P)
2.6×10-4 1.6×10-3↓
3.3
8100
RagA-RagC(K84T)
2.1×10-4 3.9×10-3
6.1
9700
RagA(R37P)-RagC 2.0×10-4 1.8×10-3↓
5.5
7600
0.001 0.000
1.1×10-4 s-1 460
GTP association
GTP association 0.5 1.0 1.5 RagA-RagC (μM)
N RagA-RagC GDP GTP
ns
WT RagA(R37P)-RagC RagA-RagC(L91P) RagA-RagC(K84T)
0.004
100
-4 RagC(K84T) 7.0×10-8 3.0×10-4 1.6×10-5 ↑ 6.6×10
Wildtype 1.1×10-4 s-1 890
GTP attaching
GTP
2000 Time (min)
0.01
0.00 0.0
0.5 1.0 RagA-RagC (μM)
0.0001
0
RagA(R37P)-RagCGTP GTP RagA(R37P)-RagC
ns ns
0.0003
0.0000
K
0.02
**
0.00 0.0
0
G
**
0.04
Kd (nM)
GTP dissociation
F
RagA-RagC(K84T)GTP RagA-RagC(K84T) RagA-RagC(L91P)GTP GTP RagA-RagC(L91P)
1.2×10-7 8.0×10-4 9.8×10-7 6.4×10-4
-4 RagC(L91P) 9.0×10-7 4.0×10-4 1.0×10-5 ↑ 4.8×10
4 104
0.01 0.1 1 RagA-RagC(L91P) (μM)
GTP
0.06
Wildtype
koff (s-1)
RagC(L91P) RagA
RagC GDP GTP
-4 -6 -4 RagA(R37P) 8.9×10-7 ↑ 7.2×10 5.3×10 6.0×10
GTP binding
0
32−
Summary of binding kinetics (4°C) RagA GTP GDP
RagC(L91P) ×GTP RagA×GTP
46−
104
8
RagC(L91P) RagA
D
Time
MW (kD)
RagC(L91P) ×GTP
105
4 Binding (au)
46−
Binding (au)
MW (kD)
Leu91Hs (Leu39Sc)
Lys84Hs (Asn32Sc)
C
[RagA-RagC(L91P)]
kobsd (min-1)
B
Gtr2pGppNHp PDB: 3R7W
kobsd (min-1)
A
Subunit rotation (Intersubunit) GTP
GDP
GDP
RagC open
RagC(L91P) ?
RagA(R37P)
RagA(T21N)
closed
GTP
GDP
GDP RagA
GDP
GTP
GDP
GDP
GTP TP
GTP
RagC open
RagC(S75N)
RagC(L91P) RagC(K84T)
RagA(R37P) ?
closed
Figure 5. Mutations in Switch I of the Rag GTPases Impair Intersubunit Communication (A) Crystal structure of Gtr2p (RagC homolog in yeast) bound to GppNHp (PDB: 3R7W). The switch I region is shown in orange. Upon GppNHp binding, switch I forms a lid on top the nucleotide-binding pocket. Leu91Hs and Lys84Hs are shown in spheres. (B) GTP binding to RagA-RagC(L91P). Top: nucleotide bound to RagA and RagC(L91P). Bottom: quantification. Data were fit to a single-site binding equation to obtain the Kd. In comparison, wild-type binding is shown as dashed curves. RagA-RagC(L91P) and wild-type Rags bind GTP in a similar fashion. (C) Reaction time course of radioactively labeled GTP remaining bound to RagA-RagC(L91P) during the off-rate measurement. Top: GTP bound to RagA and RagC(L91P) after using unlabeled GTP as a chase. Bottom: quantification. Data were fit to a single exponential decay to obtain the koff. The wild-type off rate is shown as dashed curves for reference. GTP dissociates from RagA-RagC(L91P) faster than from wild-type Rags. (D) Summary of koff and Kd values for RagA-RagC(L91P), RagA-RagC(K84T), and RagA(R37P)-RagC. Wild-type data are reproduced from Figure 1Q for ease of comparison. (E and F) Half-site on rates for RagA-RagC(L91P) (squares) and RagA-RagC(K84T) (diamonds) (E) and RagA(R37P)-RagC (circles) (F). The subunits in bold are those with which the labeled GTP is associating in the assay to measure the half-site on rate. The slope gives the half-site on rate. Wild-type half-site on rates are shown as dashed lines for comparisons. Notice that the concentration dependence for GTP association is partially restored for all the mutants. *p < 0.05; **p < 0.01; ns, not significant from zero. (G) Summary of half-site on rates of GTP to RagA-RagC(L91P), RagA-RagC(K84T), and RagA(R37P)-RagC. On rates for wild-type Rags are reproduced from Figure 2H for ease of comparison. (H and I) Single- (H) and multiple-turnover (I) GTP hydrolysis assays for RagA-RagC(L91P) (blue circles), RagA-RagC(K84T) (brown circles), and RagA(R37P)RagC (red circles). Wild-type curves are shown as dashed lines. (J) Summary of hydrolysis kinetics for RagA-RagC(L91P), RagA-RagC(K84T), and RagA(R37P)-RagC. Wild-type data are reproduced from Figure 3K for ease of comparison. (K) Illustration of the effects of the switch I mutations on intersubunit crosstalk within the Rag heterodimer.
10 Molecular Cell 68, 1–14, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
A
B
Rag mutants and abbreviations
Mutation property Abbre. Wildtype RagA-RagC W D Lower binding affinity to GTP Slower GTP hydrolysis T Switch I mutation 1 S1 Switch I mutation 2 S2
cDNA transfected
FLAG-metap2 HA-RagA FLAG-RagC
RagA T21N Q66L R37P -
Amino acid stimulation
cDNA transfection
Change media to RPMI(5% aa) 2 hours
+ W D T S1 W W W W W W W W W W D T S1 S2
‘-aa’
HA-metap2 cDNA transfected HA-RagA HA-RagC amino acids
IP: FLAG
– – – –
+
+
–
+
WT WT
T21N WT
Q66L R37P WT WT
–
–
–
+
+
+
–
+
–
WT WT
+
–
WT WT WT S75N Q120L L91P
+
–
+
–
+
–
+
WT K84T –
+ Short exp.
P -T389-S6K1
Long exp.
S6K1
+95% aa 15 mins
Raptor
HA-metap2
‘+aa’
HA-RagC
mTOR IP: FLAG
FLAG-S6K1 and +
HEK 293T cells
RagC S75N Q120L L91P K84T
Cell lysate
HA-RagA
HA-RagA
FLAG-metap2
Raptor
FLAG-RagC Raptor
C
mTOR Cell lysate
Amino acid starvation
HA-RagA
HEK 293T cells
FLAG-metap2
cDNA transfection
Protein level (au)
FLAG-RagC
Change media to RPMI(100% aa)
Raptor mTOR
2
1 hour
HA-metap2 cDNA transfected HA-RagA HA-RagC amino acids
‘+aa’
1
2
3
4
5 6 7 Lane
8
–
+
–
WT WT
T21N Q66L WT WT
+
+ –
–
+
–
FLAG-S6K1 and + R37P WT + –
+
WT WT –
+
P -T389-S6K1
–
WT WT WT S75N Q120L L91P +
–
+
–
+
–
WT K84T +
–
Long exp.
S6K1
Change to RPMI(5% aa)
9 10
+
+
Short exp
IP: FLAG
1 0
– – –
HA-metap2
30 mins
HA-RagC
‘-aa’ Cell lysate
HA-RagA Raptor
R37P
GTP GTP RagA
GDP RagC
T R
C
(L
91
P)
P)
W
37 R
250
ag
200
A(
100 150 Time (min)
ag
50
R
(L
91
P)
T
P)
W
37
0.02
0.00
0
R
L91P or K84T
(min-1)
1 106 0
R
Activated
2 106
0.00
R
250
0.04
WT RagA(R37P) RagC(L91P)
C
200
A(
100 150 Time (min)
3
106
ag
50
ag
0
4
106
ON→OFF
Signal (au)
OFF→ON
1 106
0.05
k
Signal (au)
2 106
Kinetics for amino acid starvation
0.10
(min-1)
WT RagA(R37P) RagC(L91P)
3 106
0
F
E
Kinetics for amino acid stimulation 4 106
k
D
Inhibited
GTP GDP RagA
GTP RagC
Figure 6. Communication-Defective Rag Mutants Impair mTORC1 Signaling in Cells (A) Coimmunoprecipitation of Rag GTPase heterodimers containing communication-defective mutants with mTORC1 subunits mTOR and Raptor. Communication-defective mutants mimic mutants incapable of binding GTP in terms of their interactions with mTORC1. (B and C) Effects of the communication-defective mutants on mTORC1 signaling in response to amino acid stimulation (B) or starvation (C). mTORC1 signaling is detected by the phosphorylation of Thr389 of S6K1 in response to amino acid stimulation (B) or starvation (C). mTORC1 signaling is impaired in cells expressing communication-defective mutants. (D and E) mTORC1 activity during a time course of amino acid stimulation (D) or starvation (E). Communication-defective mutants slow the response of the mTORC1 pathway to changes in amino acid levels. (F) Illustration of the effects of the communication-defective mutants on mTORC1 signaling in response to amino acids.
homologous yeast Gtr proteins (Gong et al., 2011; Jeong et al., 2012), we hypothesize the following chain of events. The binding of the GTP to one subunit of the heterodimer triggers local conformational changes in switch I and switch II, which form a lid over its nucleotide-binding pocket. Unlike Ras, whose switch I adopts a loop conformation upon GTP binding, we predict from the Gtr1 structure that switch I of the Rag GTPases forms a short a helix
(Figures 7A and 7B). This helix is in the right position to touch the other subunit, possibly precluding rearrangement of its switch I and thus causing slower association and a higher hydrolysis rate for the second GTP. The proposed reliance of the locking state on cooperation between the switch I regions of the Rag subunits could explain why we can impair it by introducing mutations that promote switch I flexibility.
Molecular Cell 68, 1–14, November 2, 2017 11
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
A
RasGMPPNP
B
RasGDP
I
I
Gtr2pGMPPNP
Gtr2pGDP
II
I
II
II
II
I
Ras (overlap)
Gtr2p (overlap)
C
nl U
Transitioning
g
GDP
in
oc
ki
RagC
ck
RagA
Amino acid deficiency
GDP
Lo
ng
GDP
GDP GTP Maintenance of Inhibition
GDP
GTP
Inhibited GATOR1 FLCNFNIP2 GTP GDP
GDP
Maintenance of Activation
GDP
nl
in g
Transitioning GDP
U
ck
Activated
oc
Lo
ki
ng
GTP
GDP
Amino acid sufficiency
Figure 7. Proposed Locking Mechanism for the Regulation of the Rag GTPases (A and B) Comparison of the conformations of Ras (A) and Gtr2p (B) upon GTP or GDP binding. The switch I regions of Ras and Gtr2p are shown in red or orange. Switch II is shown in cyan or green. Gtr2p (the RagC homolog in yeast) shows more dramatic conformational changes than Ras in switch I and switch II when it is associated with the GTP analog. (C) Functional cycle of the Rag heterodimer in response to amino acid levels. Two stable states, GTPRagA-RagCGDP and GDPRagA-RagCGTP, generate constant outputs for mTORC1 in the presence or absence of amino acids, respectively. The GDPRagA-RagCGDP configuration allows for interconversion between the active and inhibited states.
Beyond its requirement for intersubunit communication and normal mTORC1 signaling, there are several other reasons why the heterodimeric architecture of Rag GTPases may have evolved. First, unlike conventional signaling GTPases that exist in two states, on (GTP bound) and off (GDP bound), the Rag GTPases can exist in four states. Although previous work suggests that two are stable (Sancak et al., 2008) (GTPRagARagCGDP and GDPRagA-RagCGTP), a third (GDPRagA-RagCGDP) must be an intermediate given the importance of the GAPs for interconverting between them. Moreover, the off state of the Rag
12 Molecular Cell 68, 1–14, November 2, 2017
heterodimer requires RagC to bind GTP, which suppresses its association with mTORC1 regardless of the identity of the nucleotide bound to RagA (Figure S7). It is likely that a switch I rearrangement in RagC either directly displaces mTORC1 or shields the residues responsible for it to associate. Therefore, the off state of the Rag GTPase heterodimer is likely different from that of canonical GTPases, in that it is ‘‘active’’ because mTORC1 is actively rejected due to physical hindrance rather simply being unable to bind with it. We speculate that this provides an additional mechanism to prevent the mis-activation of mTORC1. Therefore, the heterodimeric nature of the Rag GTPases creates more states that should allow for the more precise control of mTORC1 signaling. We hypothesized that a role for the fourth state (GTPRagA-RagCGTP) will eventually emerge. Second, a heterodimer allows for more binding sites to integrate the many regulators and effectors that the Rag GTPases must contact. Residing in what is likely a crowded environment on the lysosomal surface, the Rag GTPases need to interact with several large multi-protein factors, including, but probably not limited to, Ragulator (Sancak et al., 2010), SLC38A9 (Jung et al., 2015; Rebsamen et al., 2015; Wang et al., 2015), GATOR1 (Bar-Peled et al., 2013; Panchaud et al., 2013), FLCN-FNIP2 (Petit et al., 2013; Tsun et al., 2013), and mTORC1 (Sancak et al., 2008). Considering that the size of a canonical GTPase domain is only 20 kDa, it is physically impossible for it to interface with so many factors at the same time, and a heterodimer provides additional binding sites for this to occur. In addition, the existence of the locking mechanism allows for the control of the nucleotide-loading state of one subunit to determine the functional state of the entire heterodimer. Compared to Ras, the Rag GTPases have slower rates of GTP dissociation and hydrolysis (Frech et al., 1994). These properties enable the Rags to hold on to the nucleotide-loading state in which they reside (Figure 7C), probably because of a requirement for generating a constant output. In order to switch between functional states, two GAPs are needed. By better positioning the critical glutamine residue that chelates the nucleophilic-attacking water, GATOR1 and FLCN-FNIP2 strongly stimulate the GTP hydrolysis rates of RagA and RagC, respectively, which drives the heterodimer from a locked on or off state into the GDPRagA-RagCGDP intermediate (Figure 7C). At this point, if a GTP molecule binds to the same subunit, then it will be hydrolyzed immediately because of the proximity to the corresponding GAP. On the other hand, if GTP binds to the other subunit, which drives the Rags into the opposite functional state, then no stimulated hydrolysis will occur because of the specificity of GATOR1 and FLCN-FNIP2 (cf. Figure 5). The existence of intersubunit communication by the Rag heterodimer and of many Rag GTPase regulators hints at the importance of ensuring faithful and efficient mTORC1 signaling. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d
KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
d d
d
EXPERIMENTAL MODEL AND SUBJECT DETAILS METHOD DETAILS B Purification of the Rag GTPase Heterodimer (I) B Purification of the Rag GTPase Heterodimer (II) B Purification of the GATOR1 and FLCN-FNIP2 Complexes B Equilibrium Binding Assays B On-Rate Measurements B Off-Rate Measurements B Single Turnover GTP Hydrolysis Experiments B Multiple Turnover GTP Hydrolysis Experiments B Free Energy Diagram B Preparation of Cell Lysates and Immunoprecipitates B Lysine Modification Assay QUANTIFICATION AND STATISTICAL ANALYSIS
Frech, M., Darden, T.A., Pedersen, L.G., Foley, C.K., Charifson, P.S., Anderson, M.W., and Wittinghofer, A. (1994). Role of glutamine-61 in the hydrolysis of GTP by p21H-ras: an experimental and theoretical study. Biochemistry 33, 3237–3244. Gong, R., Li, L., Liu, Y., Wang, P., Yang, H., Wang, L., Cheng, J., Guan, K.-L., and Xu, Y. (2011). Crystal structure of the Gtr1p-Gtr2p complex reveals new insights into the amino acid-induced TORC1 activation. Genes Dev. 25, 1668–1673. Gonza´lez, A., and Hall, M.N. (2017). Nutrient sensing and TOR signaling in yeast and mammals. EMBO J. 36, 397–408. Guo, X., Bandyopadhyay, P., Schilling, B., Young, M.M., Fujii, N., Aynechi, T., Guy, R.K., Kuntz, I.D., and Gibson, B.W. (2008). Partial acetylation of lysine residues improves intraprotein cross-linking. Anal. Chem. 80, 951–960. Hall, B.E., Bar-Sagi, D., and Nassar, N. (2002). The structural basis for the transition from Ras-GTP to Ras-GDP. Proc. Natl. Acad. Sci. USA 99, 12138–12142.
SUPPLEMENTAL INFORMATION
Hwang, Y.W., and Miller, D.L. (1987). A mutation that alters the nucleotide specificity of elongation factor Tu, a GTP regulatory protein. J. Biol. Chem. 262, 13081–13085.
Supplemental Information includes seven figures and can be found with this article online at https://doi.org/10.1016/j.molcel.2017.09.026.
Inoki, K., Li, Y., Xu, T., and Guan, K.-L. (2003). Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signaling. Genes Dev. 17, 1829–1834.
AUTHOR CONTRIBUTIONS K.S. and D.M.S. initiated the project, and K.S. designed the research plan. K.S. performed all the experiments and analyzed the data with input from A.C. K.S. wrote the manuscript, and D.M.S. edited the manuscript. ACKNOWLEDGMENTS We thank all members of the Sabatini Laboratory for helpful insights. We thank Dr. Shu-ou Shan for critical reading of the manuscript. This work was supported by grants from the NIH (R01 CA103866, R01 CA129105, and R37 AI47389) and the Department of Defense (W81XWH-15-1-0230) to D.M.S. and fellowship support from the Life Sciences Research Foundation to K.S, where he is a Pfizer Fellow. D.M.S. is an investigator of the Howard Hughes Medical Institute. Received: June 5, 2017 Revised: August 14, 2017 Accepted: September 19, 2017 Published: October 19, 2017 REFERENCES Bar-Peled, L., Chantranupong, L., Cherniack, A.D., Chen, W.W., Ottina, K.A., Grabiner, B.C., Spear, E.D., Carter, S.L., Meyerson, M., and Sabatini, D.M. (2013). A tumor suppressor complex with GAP activity for the Rag GTPases that signal amino acid sufficiency to mTORC1. Science 340, 1100–1106. De Franceschi, N., Wild, K., Schlacht, A., Dacks, J.B., Sinning, I., and Filippini, F. (2014). Longin and GAF domains: structural evolution and adaptation to the subcellular trafficking machinery. Traffic 15, 104–121. Dibble, C.C., and Manning, B.D. (2013). Signal integration by mTORC1 coordinates nutrient input with biosynthetic output. Nat. Cell Biol. 15, 555–564. Efeyan, A., Comb, W.C., and Sabatini, D.M. (2015). Nutrient-sensing mechanisms and pathways. Nature 517, 302–310. Feig, L.A., and Cooper, G.M. (1988). Inhibition of NIH 3T3 cell proliferation by a mutant ras protein with preferential affinity for GDP. Mol. Cell. Biol. 8, 3235–3243. Fersht, A. (1999). Structure and Mechanism in Protein Science (Macmillan). Feuerstein, J., Goody, R.S., and Wittinghofer, A. (1987). Preparation and characterization of nucleotide-free and metal ion-free p21 ‘‘apoprotein’’. J. Biol. Chem. 262, 8455–8458.
Jeong, J.-H., Lee, K.-H., Kim, Y.-M., Kim, D.-H., Oh, B.-H., and Kim, Y.-G. (2012). Crystal structure of the Gtr1p(GTP)-Gtr2p(GDP) protein complex reveals large structural rearrangements triggered by GTP-to-GDP conversion. J. Biol. Chem. 287, 29648–29653. Jewell, J.L., Russell, R.C., and Guan, K.-L. (2013). Amino acid signalling upstream of mTOR. Nat. Rev. Mol. Cell Biol. 14, 133–139. Jung, J., Genau, H.M., and Behrends, C. (2015). Amino acid-dependent mTORC1 regulation by the lysosomal membrane protein SLC38A9. Mol. Cell. Biol. 35, 2479–2494. Leipe, D.D., Wolf, Y.I., Koonin, E.V., and Aravind, L. (2002). Classification and evolution of P-loop GTPases and related ATPases. J. Mol. Biol. 317, 41–72. Levine, T.P., Daniels, R.D., Wong, L.H., Gatta, A.T., Gerondopoulos, A., and Barr, F.A. (2013). Discovery of new Longin and Roadblock domains that form platforms for small GTPases in Ragulator and TRAPP-II. Small GTPases 4, 62–69. Li, Y., Inoki, K., and Guan, K.L. (2004). Biochemical and functional characterizations of small GTPase Rheb and TSC2 GAP activity. Mol. Cell. Biol. 24, 7965–7975. Long, P.A., Zimmermann, M.T., Kim, M., Evans, J.M., Xu, X., and Olson, T.M. (2016). De novo RRAGC mutation activates mTORC1 signaling in syndromic fetal dilated cardiomyopathy. Hum. Genet. 135, 909–917. Menon, S., Dibble, C.C., Talbott, G., Hoxhaj, G., Valvezan, A.J., Takahashi, H., Cantley, L.C., and Manning, B.D. (2014). Spatial control of the TSC complex integrates insulin and nutrient regulation of mTORC1 at the lysosome. Cell 156, 771–785. Nakashima, N., Noguchi, E., and Nishimoto, T. (1999). Saccharomyces cerevisiae putative G protein, Gtr1p, which forms complexes with itself and a novel protein designated as Gtr2p, negatively regulates the Ran/Gsp1p G protein cycle through Gtr2p. Genetics 152, 853–867. Nowak, E., and Goody, R.S. (1988). Kinetics of adenosine 50 -triphosphate and adenosine 50 -diphosphate interaction with G-actin. Biochemistry 27, 8613–8617. Okosun, J., Wolfson, R.L., Wang, J., Araf, S., Wilkins, L., Castellano, B.M., Escudero-Ibarz, L., Al Seraihi, A.F., Richter, J., Bernhart, S.H., et al. (2016). Recurrent mTORC1-activating RRAGC mutations in follicular lymphoma. Nat. Genet. 48, 183–188. Pai, E.F., Krengel, U., Petsko, G.A., Goody, R.S., Kabsch, W., and Wittinghofer, A. (1990). Refined crystal structure of the triphosphate conformation of H-ras p21 at 1.35 A resolution: implications for the mechanism of GTP hydrolysis. EMBO J. 9, 2351–2359.
Molecular Cell 68, 1–14, November 2, 2017 13
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Panchaud, N., Pe´li-Gulli, M.-P., and De Virgilio, C. (2013). Amino acid deprivation inhibits TORC1 through a GTPase-activating protein complex for the Rag family GTPase Gtr1. Sci. Signal. 6, ra42. Peluso, P., Shan, S.-O., Nock, S., Herschlag, D., and Walter, P. (2001). Role of SRP RNA in the GTPase cycles of Ffh and FtsY. Biochemistry 40, 15224–15233. Petit, C.S., Roczniak-Ferguson, A., and Ferguson, S.M. (2013). Recruitment of folliculin to lysosomes supports the amino acid-dependent activation of Rag GTPases. J. Cell Biol. 202, 1107–1122. Rebsamen, M., Pochini, L., Stasyk, T., de Araujo, M.E.G., Galluccio, M., Kandasamy, R.K., Snijder, B., Fauster, A., Rudashevskaya, E.L., Bruckner, M., et al. (2015). SLC38A9 is a component of the lysosomal amino acid sensing machinery that controls mTORC1. Nature 519, 477–481. Saito, K., Araki, Y., Kontani, K., Nishina, H., and Katada, T. (2005). Novel role of the small GTPase Rheb: its implication in endocytic pathway independent of the activation of mammalian target of rapamycin. J. Biochem. 137, 423–430. Sancak, Y., Peterson, T.R., Shaul, Y.D., Lindquist, R.A., Thoreen, C.C., BarPeled, L., and Sabatini, D.M. (2008). The Rag GTPases bind raptor and mediate amino acid signaling to mTORC1. Science 320, 1496–1501. Sancak, Y., Bar-Peled, L., Zoncu, R., Markhard, A.L., Nada, S., and Sabatini, D.M. (2010). Ragulator-Rag complex targets mTORC1 to the lysosomal surface and is necessary for its activation by amino acids. Cell 141, 290–303. Saucedo, L.J., Gao, X., Chiarelli, D.A., Li, L., Pan, D., and Edgar, B.A. (2003). Rheb promotes cell growth as a component of the insulin/TOR signalling network. Nat. Cell Biol. 5, 566–571. €ller, L., Lautwein, A., Scheffzek, K., Ahmadian, M.R., Kabsch, W., Wiesmu Schmitz, F., and Wittinghofer, A. (1997). The Ras-RasGAP complex: structural basis for GTPase activation and its loss in oncogenic Ras mutants. Science 277, 333–338. €rmann, A., Brauers, A., Massmann, S., Becker, W., and Joost, H.G. Schu (1995). Cloning of a novel family of mammalian GTP-binding proteins (RagA, RagBs, RagB1) with remote similarity to the Ras-related GTPases. J. Biol. Chem. 270, 28982–28988.
14 Molecular Cell 68, 1–14, November 2, 2017
Shan, S.-O. (2016). ATPase and GTPase tangos drive intracellular protein transport. Trends Biochem. Sci. 41, 1050–1060. Shaw, R.J., and Cantley, L.C. (2006). Ras, PI(3)K and mTOR signalling controls tumour cell growth. Nature 441, 424–430. Simon, I., Zerial, M., and Goody, R.S. (1996). Kinetics of interaction of Rab5 and Rab7 with nucleotides and magnesium ions. J. Biol. Chem. 271, 20470–20478. Stocker, H., Radimerski, T., Schindelholz, B., Wittwer, F., Belawat, P., Daram, P., Breuer, S., Thomas, G., and Hafen, E. (2003). Rheb is an essential regulator of S6K in controlling cell growth in Drosophila. Nat. Cell Biol. 5, 559–565. Sweet, D.J., and Gerace, L. (1996). A GTPase distinct from Ran is involved in nuclear protein import. J. Cell Biol. 133, 971–983. Tsun, Z.-Y., Bar-Peled, L., Chantranupong, L., Zoncu, R., Wang, T., Kim, C., Spooner, E., and Sabatini, D.M. (2013). The folliculin tumor suppressor is a GAP for the RagC/D GTPases that signal amino acid levels to mTORC1. Mol. Cell 52, 495–505. Vetter, I.R., Nowak, C., Nishimoto, T., Kuhlmann, J., and Wittinghofer, A. (1999). Structure of a Ran-binding domain complexed with Ran bound to a GTP analogue: implications for nuclear transport. Nature 398, 39–46. Wang, Y., Nakashima, N., Sekiguchi, T., and Nishimoto, T. (2005). Saccharomyces cerevisiae GTPase complex: Gtr1p-Gtr2p regulates cell-proliferation through Saccharomyces cerevisiae Ran-binding protein, Yrb2p. Biochem. Biophys. Res. Commun. 336, 639–645. Wang, S., Tsun, Z.-Y., Wolfson, R.L., Shen, K., Wyant, G.A., Plovanich, M.E., Yuan, E.D., Jones, T.D., Chantranupong, L., Comb, W., et al. (2015). Lysosomal amino acid transporter SLC38A9 signals arginine sufficiency to mTORC1. Science 347, 188–194. Wong, M. (2013). A critical review of mTOR inhibitors and epilepsy: from basic science to clinical trials. Expert Rev. Neurother. 13, 657–669. Yue, V.T., and Schimmel, P.R. (1977). Direct and specific photochemical cross-linking of adenosine 50 -triphosphate to an aminoacyl-tRNA synthetase. Biochemistry 16, 4678–4684.
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
STAR+METHODS KEY RESOURCES TABLE
REAGENT or RESOURCE
SOURCE
IDENTIFIER
Rabbit anti-S6K1
Cell Signaling Technology
2708; RRID: AB_390722
Rabbit anti-mTOR
Cell Signaling Technology
2972; RRID: AB_330978
Rabbit anti-Raptor
EMD Millipore
09-217; RRID: AB_612103
Antibodies
Rabbit anti-p-T389-S6K1
Cell Signaling Technology
9205; RRID: AB_330944
Rabbit anti-HA
Bethyl
A190-208A; RRID: AB_67466
Rabbit anti-Flag
Cell Signaling Technology
14793; RRID: AB_2572291
Anti-rabbit, HRP-linked antibody
Cell Signaling Technology
7074; RRID: AB_2099233
NEB
C2527
HEPES
Sigma-Aldrich
H3375
Potassium acetate
Sigma-Aldrich
P1190
Magnesium chloride
Sigma-Aldrich
M8266
Sodium chloride
Sigma-Aldrich
S7563
DTT
Sigma-Aldrich
D5545
Chaps
Sigma-Aldrich
C5070
Triton
Sigma-Aldrich
T9284
Imidazole
Sigma-Aldrich
I5513
GTP
Sigma-Aldrich
G8877
GDP
Sigma-Aldrich
G7127
TCEP
Sigma-Aldrich
C4706
ATP
Sigma-Aldrich
A7699
a-32P-GTP
PerkinElmer
BLU006H250UC
g-32P-GTP
PerkinElmer
BLU004Z250UC
a-32P-GDP
Homemade
N/A
DMEM
Sigma-Aldrich
D5523
RPMI
Sigma-Aldrich
R5886
Human: HEK293T
ATCC
CRL-3216
Human: FreeStyle 293-F
Thermo Fisher Scientific
R79007
[pSK109] pETDuet RagA-RagC
This study
Addgene 99641
[pSK151] pETDuet RagA(D130N)-RagC
This study
Addgene 99642
[pSK121] pETDuet RagA-RagC(D181N)
This study
Addgene 99643
[pSK152] pETDuet RagA(T21N)-RagC
This study
Addgene 99644
Bacterial and Virus Strains BL21(DE3) Chemicals, Peptides, and Recombinant Proteins
Experimental Models: Cell Lines
Recombinant DNA
[pSK170] pETDuet RagA-RagC(S75N)
This study
Addgene 99645
[pSK330] pETDuet RagA(R37P)-RagC
This study
Addgene 99646
[pSK210] pETDuet RagA-RagC(L91P)
This study
Addgene 99647
[pSK224] pETDuet RagA-RagC(K84T)
This study
Addgene 99648
[pSK235] pK RagA-RagC
This study
Addgene 99649
[pSK251] pK RagA(T21N)-RagC
This study
Addgene 99650
[pSK252] pK RagA-RagC(S75N)
This study
Addgene 99699
[pSK245] pK RagA(Q66L)-RagC
This study
Addgene 99700 (Continued on next page)
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e1
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Continued REAGENT or RESOURCE
SOURCE
IDENTIFIER
[pSK246] pK RagA-RagC(Q120L)
This study
Addgene 99710
[pSK247] pK RagA(Q66L)-RagC(Q120L)
This study
Addgene 99702
[pSK249] pK RagA(R37P)-RagC
This study
Addgene 99706
[pSK269] pK RagA-RagC(L91P)
This study
Addgene 99707
[pSK332] pK RagA-RagC(K84T)
This study
Addgene 99708
[pSK6] pRK5 HA-RagA
This study
Addgene 99710
[pSK323] pRK5 HA-RagA(T21N)
This study
Addgene 99711
[pSK324] pRK5 HA-RagA(Q66L)
This study
Addgene 99712
[pSK325] pRK5 HA-RagA(R37P)
This study
Addgene 99713
[pSK184] pRK5 Flag-RagA
This study
Addgene 99714
[pSK192] pRK5 Flag-RagA(T21N)
This study
Addgene 99715
[pSK194] pRK5 Flag-RagA(Q66L)
This study
Addgene 99716
[pSK327] pRK5 Flag-RagA(R37P)
This study
Addgene 99717
[pSK8] pRK5 HA-RagC
This study
Addgene 99718
[pSK320] pRK5 HA-RagC(S75N)
This study
Addgene 99719
[pSK321] pRK5 HA-RagC(Q120L)
This study
Addgene 99720
[pSK322] pRK5 HA-RagC(L91P)
This study
Addgene 99721
[pSK328] pRK5 HA-RagC(K84T)
This study
Addgene 99722
[pSK189] pRK5 Flag-RagC
This study
Addgene 99723
[pSK196] pRK5 Flag-RagC(S75N)
This study
Addgene 99724
[pSK197] pRK5 Flag-RagC(Q120L)
This study
Addgene 99725
[pSK326] pRK5 Flag-RagC(L91P)
This study
Addgene 99726
[pSK329] pRK5 Flag-RagC(K84T)
This study
Addgene 99727
[pSK163] pRK5 HA-Nprl2
Bar-Peled et al., 2013
Addgene 99709
[pSK164] pRK5 HA-Nprl3
Bar-Peled et al., 2013
Addgene 46330
[pSK165] pRK5 Flag-Depdc5
Bar-Peled et al., 2013
Addgene 46340
[pSK310] pRK5 HA-FLCN
Tsun et al., 2013
Addgene 72297
[pSK313] pRK5 Flag-FNIP2
Tsun et al., 2013
Addgene 72294
Software and Algorithms Prism
GraphPad Software
N/A
ImageJ
N/A
N/A
Flag-M2 affinity resin
Sigma-Aldrich
A2220
Ni-NTA affinity resin
EMD Millipore
30230
Q-Sepharose FF resin
GE Healthcare
17051001
Cellulose 300 PEI thin layer chromatography plates
Sorbtech
1122126
Codon optimized sequence for RagA and RagC
Genescript
N/A
Other
CONTACT FOR REAGENT AND RESOURCE SHARING Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, David M. Sabatini ([email protected]). EXPERIMENTAL MODEL AND SUBJECT DETAILS HEK293T cells were obtained from ATCC and maintained in an incubator sett at 37 C and 5% CO2. HEK293T cells were cultured in DMEM with 10% IFS supplemented with 2 mM glutamine, penicillin (100 IU/ml), and streptomycin (100 mg/ml).
e2 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
FreeStyle 293-F cells were obtained from Thermo Fisher and maintained in a Multitron Pro shaker setting at 37 C, 125 rpm, 8% CO2, and 80% humidity. FreeStyle 293-F cells were cultured in FreeStyle 293 Expression Medium supplemented with penicillin (100 IU/ml) and streptomycin (100 mg/ml). METHOD DETAILS Purification of the Rag GTPase Heterodimer (I) To co-express the Rag GTPase dimer in bacteria, a pETDuet-1 vector encoding codon optimized, C-terminally His-tagged RagA and tagless RagC (pSK109) was transformed into BL21(DE3) competent cells and plated on a LB-agar plate containing 100 mg/ml ampicillin. A single colony was inoculated in 50 mL LB media containing 100 mg/ml ampicillin and shaken at 37 C overnight. The saturated overnight culture was diluted 1:100 into four liters of LB to continue growth. Optical density (OD) of the culture was monitored at 600 nm, and the temperature was lowered to 18 C when OD600 reached 0.4 – 0.6 (early log-phase). Half an hour later when the culture was cooled down to 25 C, the Rag dimer was induced by 0.5 mM IPTG and expressed overnight at 18 C (20 hr). Unless specified below, all the purification steps were carried out at 4 C or on ice to stabilize the Rag GTPases. The bacterial pellets were resuspended in 100 mL of Resuspension Buffer (50 mM NaHEPES, pH 7.4; 100 mM NaCl; 2 mM MgCl2; 2 mM DTT; 0.5 mM PMSF; 0.05% Triton; 100 mM GDP; 100 mg/ml lysozyme; 10 mg/ml DNase; and four tablets of protease inhibitor cocktail). The mixture was stirred at 4 C for 30 min to ensure homogeneous resuspension, before passing through the microfluidizer five times at the pressure of 18,000 psi. The cell lysate was cleared by centrifugation at 40,000 3 g for one hour. The supernatant containing soluble Rag GTPase heterodimer was taken for the next step. The cleared cell lysate was first cleaned up using a hand-packed anion exchange column, taking advantage of the negatively charged RagC subunit. 15 mL of Q-Sepharose Fast Flow slurry was packed and equilibrated in Buffer A (50 mM NaHEPES, pH 7.4; 50 mM NaCl; 2 mM MgCl2; 2 mM DTT; 100 mM GDP; 0.05% Triton). The cleared cell lysate was then applied to the column using an isocratic flow pump at a rate of 2 mL/min. After sample loading, the column was washed by Buffer A for 20 column volumes, and the bound proteins were eluted by a linear salt gradient of Buffer A to Buffer A’ (50 mM NaHEPES, pH 7.4; 500 mM NaCl; 2 mM MgCl2; 2 mM DTT; 100 mM GDP; 0.05% Triton). Fractions containing the Rag dimer were pooled together and dialyzed three times against Buffer B (50 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; 0.5 mM TCEP; 100 mM GDP; 0.05% Triton) in preparation for Ni-NTA purification. Dialyzed fractions from Q-Sepharose were incubated with 5 mL of Ni-NTA resin that was pre-equilibrated in Buffer C (50 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; 0.5 mM TCEP; 100 mM ATP; 100 mM GDP; 0.05% Triton). Ni-NTA column was hand-packed after an hour of incubation, and washed by 20 column volumes of Buffer C and 10 column volumes of Buffer C’ (Buffer C supplemented with 30 mM imidazole and 300 mM NaCl). The Rag GTPase dimer was then eluted by Buffer C’’ (Buffer C supplemented with 250 mM imidazole). The elutant was concentrated in a 30 kDa molecular-weight cut-off concentrator, and ultracentrifuged at 100,000 3 g for 30 min, to remove all the aggregates in preparation for FPLC purification. For FPLC purification, a monoQ column was first equilibrated in 90% Buffer D (50 mM NaHEPES, pH 7.4; 2 mM MgCl2; 2 mM DTT) and 10% Buffer D’ (50 mM NaHEPES, pH 7.4; 1 M NaCl; 2 mM MgCl2; 2 mM DTT), before the elutant from Ni-NTA was applied. A linear gradient from 12% D’ to 30% D’ was used to elute the bound proteins, and a single peak was collected at around 22% D’, which corresponds to the Rag GTPase heterodimer. The peak was concentrated, and any bound nucleotides were stripped by 10 mM EDTA for an hour at room temperature. The apo Rag heterodimer was then applied to a HiLoad 16/60 Superdex 200 gelfiltration column pre-equilibrated in Buffer E (50 mM NaHEPES, pH 7.4; 100 mM KOAc; 1 mM EDTA; 2 mM DTT). The peak corresponding to the Rag heterodimer was collected and concentrated. The final product was aliquoted and flash frozen in the presence of 5% glycerol, and stored at 80 C. The final yield is 1 mg Rag heterodimer per liter of LB culture. The Rag GTPases purified using this construct are suitable for equilibrium binding assays, on- and off-rate measurements, and single turnover GTP hydrolysis assays. Because of the trace amount of contaminating GTPases which hydrolyzes GTP at a much faster rate, the product here cannot be used for multiple turnover GTP hydrolysis assay. Purification of the Rag GTPase Heterodimer (II) To completely remove any contaminating GTPases, a pCOLADuet-1 vector (KanR) encoding codon optimized, N-terminally His10-R8SUMO-tagged RagA and tagless RagC (pSK235) was expressed using a similar protocol as above. 8 L of bacteria culture were lysed and cleared in 160 mL of Resuspension Buffer (substitute DTT with 0.5 mM TCEP). The cleared supernatant was first cleaned up using a hand-packed Ni-NTA column using a similar protocol as above. The elutant was concentrated and ultracentrifuged for FPLC. A monoS column was first equilibrated in 85% Buffer D and 15% Buffer D’, before the elutant from Ni-NTA was applied. A linear gradient from 20% D’ to 100% D’ was used to elute the Rag GTPases. The peak was concentrated, and the fusion protein was digested overnight by the 3C protease to cleave the His10-R8-SUMO-tag. The digested mixture was applied to monoS again to remove the tag, and the apo Rag heterodimer was then applied to monoQ followed by HiLoad 16/60 Superdex 200 gel-filtration column. The final yield is 50 mg Rag dimer per liter of LB culture. Despite the low yield, the Rag GTPases purified using this construct are free of contaminating GTPases and are suitable for multiple turnover GTP hydrolysis assays.
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e3
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Purification of the GATOR1 and FLCN-FNIP2 Complexes To generate the GAPs for the Rag GTPases, one million 293 FreeStyle cells were inoculated into 2 L of FreeStyle media. Twenty-four hours later, cells were transfected using PEI with the following combination of pRK5 constructs: For GATOR1: 420 mg FLAG-Depdc5, 840 mg HA-Nprl2, and 840 mg HA-Nprl3; For FLCN-FNIP2: 600 mg FLAG-FNIP2 and 1.5 mg HA-FLCN. Thirty-six hours after transfection, cells were spun down and washed once with ice-cold PBS. Cells were lysed in 100 mL Triton lysis buffer (40 mM NaHEPES, pH 7.4; 100 mM NaCl; 5 mM MgCl2; 100 mM ATP; 10 mM Na4P2O7; 10 mM Na b-glycerol phosphate; 1% Triton; and two tablets of protease inhibitor cocktail), and were dounced to homogenize. After clearing up the insoluble fractions by centrifuging at 40,000 3 g for 30 min, 4 mL of pre-equilibrated FLAG-M2 beads were added to lysate. Recombinant protein complexes were immunoprecipitated for 3 hr at 4 C, before packing into a column. The resin was washed with 10 column volumes of Triton lysis buffer, followed by 20 column volumes of Triton lysis buffer supplemented with 300 mM NaCl. Before elution, the resin was buffer exchanged into CHAPS buffer (40 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; and 0.1% CHAPS). FLAG-tagged GATOR1 or FLCN-FNIP2 was then eluted from the FLAG-M2 beads with 3 3 FLAG peptide dissolved in CHAPS buffer for one hour at 4 C. The elutant was concentrated using a 100 kDa molecular-weight cut-off filter and ultracentrifuged at 100,000 3 g for 30 min in a TLA-100.3 rotor, to remove all the aggregates in preparation for FPLC purification. To further improve the purity and confirm the integrity of subunit assembly, the protein complexes were subsequently loaded onto a HiLoad 16/60 Superdex 200 gel-filtration column pre-equilibrated in Buffer GFB (50 mM NaHEPES, pH 7.4; 150 mM NaCl; 2 mM MgCl2; 2 mM DTT; and 0.1% CHAPS). The peak corresponding to the desired complex was collected, flash frozen in GFB supplemented with 10% glycerol and stored at 80 C. Equilibrium Binding Assays Before use, the newly thawed Rag GTPase heterodimer was ultracentrifuged at 200,000 3 g for an hour and its concentration was redetermined, to completely remove any aggregates caused by the freeze-thaw cycle. Unless otherwise specified, all the biochemical assays below were carried out in Assay Buffer (50 mM NaHEPES, pH 7.4; 100 mM KOAc; 2 mM MgCl2; 2 mM DTT; and 0.1% CHAPS). To determine the binding affinity of nucleotides to individual Rag subunits, increasing concentrations of Rag GTPases, usually ranging from 5 nM to 2 mM on a log-scale, were mixed with trace amount (1 nM) of a-32P-GTP or a-32P-GDP (Perkin-Elmer). The reaction mixtures were incubated at 4 C until equilibrium was reached, usually for four hours, before directly spotted onto a chilled metal block covered with Parafilm. 0.3 Joule of 254 nm UV light was applied to the droplets, to induce non-specific, zero distance crosslinking between the nucleotide and the subunit it bound to. The resulting products were analyzed by 12% Tris-Glycine gel. After fixing and drying, the gel was exposed to a phosphor-imaging screen, and the intensities of the radioactive signals (Iobsd) were visualized by a Typhoon scanner and quantified using the ImageQuant software. The signals were fit to a single-site binding equation (Equation I) to extract the Kd for both subunits. Iobsd = Amplitude 3
½Rag + Background; Kd + ½Rag
(I)
in which Amplitude = Radioactive Intensity 3 Crosslinking efficiency 3 Imaging efficiency. For samples taken from the same batch of experiment and run on the same gel, these three terms can be assumed approximately equal, and the Amplitude can be directly compared within. However, these relative terms may vary for mutants or in different assays, so the absolute values of the radioactive signal cannot be compared across different experiments. Notice that the dissociation constant Kd is independent of the Amplitude and thus is irrelevant to a change in any of the relative terms. The sequential binding (Figure 2A) and sequential chase (Figure 2C) assays were performed in a similar fashion. For the sequential binding assay (Figure 2A), before radioactively labeled GTP was added to the reaction, the Rag GTPase mutants were pre-incubated with 1.2 equal molar of XTP or XDP at each concentration, to ensure that the xanthine-specific site was occupied. For the sequential chase assay (Figure 2C), 5 equal molar of XTP or XDP was added to the reaction mixture after the guanine-specific site was preloaded with radioactively labeled GTP. On-Rate Measurements To determine the on-rates of nucleotides to individual Rag subunits, at time zero, different concentrations of Rag GTPases, e.g., 100/ 200/300/400 nM, were mixed with trace amount (1 nM) of a-32P-GTP or a-32P-GDP to initiate the reaction. Small aliquots of the mixture were taken out at different time points for crosslinking analysis, to trace the amount of radioactively labeled nucleotides that had been associated to the Rag GTPases, until equilibrium was reached. For example, in Figure 1I, eleven time points were taken between time zero and two hours. For each concentration of the Rag GTPases, the time points were blotted on the same gel and quantified as above. The observed reaction constants (kobsd) were extracted using a single exponential rise function (Equation II): Iobsd = Amplitude 3 1 ekobsd t + Background; (II) in which Amplitude = Radioactive Intensity 3 Crosslinking efficiency 3 Imaging efficiency.
e4 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
The kobsds were then linear-fit against the concentration of the Rag GTPases to obtain the on-rate (kon) using Equation III: kobsd = kon 3 ½Rag + koff
(III)
The half-site on-rate (Figure 2E) was determined in a similar fashion, except that wild-type Rag GTPases were pre-incubated with 1.2 equal molar of unlabeled GTP or GDP at each concentration prior to the on-rate measurement. Because each functional unit of the Rag GTPases contain two nucleotide binding sites and the fact that the second GTP binding is more difficult than the first, we can assume that in the presence of 1.2 equal molar of unlabeled nucleotide, the majority of the Rag GTPases was singly occupied, while sufficient amount of unoccupied nucleotide binding pockets remained for the later-added radioactively labeled nucleotide to be associated with. Therefore, the half-site on-rate is a measure of the association rate of the radioactively labeled nucleotide to the singly-loaded Rag GTPases. The differential on-rate (Figure 2I) was determined using xanthine-specific Rag mutants, which were pre-incubated with 1.2 equal molar of unlabeled XTP or XDP at each concentration, before the radioactively labeled GTP was added to initiate the on-rate measurement. Because of the modified hydrogen bonding pattern, xanthine-specific Rag mutants should be singly loaded with unlabeled XTP or XDP, so the differential on-rate measures the association of the radioactively labeled GTP to the guanine-specific site of the Rag mutants. Notice that in these mutants, the effect of pre-loaded XTP to inhibit GTP association is not as pronounced as that of GTP in the half-site on-rate measurement for wild-type Rags, which suggests that intersubunit communication may be partially defective in the xanthine-specific mutants. Off-Rate Measurements To determine the dissociation rates of nucleotides from individual Rag subunits, we used a pulse-chase setup. A trace amount (1 nM) of a-32P-GTP or a-32P-GDP was first incubated with 1 mM Rag GTPases for two hours, so that a certain amount of radioactively labeled nucleotides were associated with each subunit. At time zero, a large excess (1 mM) of unlabeled nucleotides were added to the reaction, to ensure that once the radioactively labeled nucleotides fall out of the nucleotide binding pocket, the chance of its re-association is small. Small aliquots of the mixture were taken out at different time points for crosslinking analysis, to trace the amount of radioactively labeled nucleotides remaining bound to the Rag GTPases. The off-rates (koff) were extracted by fitting the intensities of the radioactive signals to a single exponential decay function using Equation IV: Iobsd = Amplitude 3 ekoff t + Background;
(IV)
in which Amplitude = Radioactive Intensity 3 Crosslinking efficiency 3 Imaging efficiency. Notice that the koff is independent of the Amplitude and thus is irrelevant to a change in any of the relative terms, usually caused by batch-to-batch variation. Single Turnover GTP Hydrolysis Experiments To determine the hydrolysis rate of the Rag GTPases under single turnover conditions, increasing amounts of Rag GTPases, from 0.5 nM to 20 nM for wild-type Rags, was incubated with 0.1 nM of g-32P-GTP. Aliquots of the reaction mixture were taken at different time points and quenched using 0.75 M KH2PO4 (pH 3.3). These points were taken within the time window that less than 20% of the GTP was hydrolyzed. For example, six time points were taken for wild-type Rags at 1 min, 1, 2, 4, 6, and 9 hr, as shown in Figure 3B. The time points were then applied to cellulose 300 PEI thin layer chromatography (TLC) plates, which were expanded in TLC buffer (1 M formic acid and 0.5 M LiCl) for 20 min. The plates were dried and imaged with phosphorimaging screens. The intensities of the radioactive signals were read out by a Typhoon scanner and quantified using the ImageQuant software. Linear regression was used to fit the fraction of phosphate generated by GTP hydrolysis against time, giving the observed rate constants (kobsd). The kobsds were then fit to a Michaelis-Menten equation (Equation V), to extract the kcat and KM values for the single turnover hydrolysis reaction. kobsd = kcat 3
½Rag0 KM + ½Rag0
(V)
Equation V was derived as following. Consider the reaction in a single turnover setup: k1
k2
Rag + GTP 4 ½Rag,GTP/½Rag,GDP + Pi;
(V.1)
in which k1 and k-1 represent the association and dissociation constants, and k2 represents the catalytic rate constant for the chemistry step. Notice that in the single turnover setup, the amount of Rag GTPases is in excess of that of radioactively labeled GTP, so that the enzyme turnover, i.e., dissociation of GDP from the Rag, was omitted. Intermediary species are generated at the speed of: d½Rag,GTP = k1 ½Rag½GTP ðk1 + k2 Þ½Rag,GTP; dt
(V.2)
and
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e5
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
d½Pi d½Rag,GDP = = k2 ½Rag,GTP: dt dt
(V.3)
½Rag0 = ½Rag + ½Rag,GTP + ½Rag,GDP;
(V.4)
½GTP0 = ½GTP + ½Rag,GTP + ½Rag,GDP;
(V.5)
Based on conservation of matter:
and
in which [Rag]0 and [GTP]0 represent the total concentration of the Rag GTPases and GTP at the beginning of the reaction. In a single turnover setup, where (1) the concentration of the Rag GTPases is much higher than that of GTP, and (2) the reaction is monitored within 20% of hydrolysis, we can approximate the equations above to: ½Rag0 z½Rag;
(V.6)
½GTP0 z½GTP + ½Rag,GTP:
(V.7)
d½Rag,GTP = k1 ½Rag0 ½GTP0 k1 ½GTP0 + k1 + k2 ½Rag,GTP; dt
(V.8)
and
Substitute (V.6) and (V.7) into (V.2), we obtain:
the solution of which is: ½Rag,GTP =
k1 ½Rag0 ½GTP0 1 eðk1 ½Rag0 + k1 + k2 Þt ; k1 ½Rag0 + k1 + k2
(V.9)
under the boundary condition that at time zero, [Rag,GTP] = 0. Notice that this equation can only be applied at the early stage of the reaction, before 20% of the GTP has been hydrolyzed. When t > 0, the exponential term quickly diminishes to zero, so the concentration of the intermediate species [Rag,GTP] can be reduced to: ½Rag,GTP =
k1 ½Rag0 ½GTP0 ½Rag0 ½GTP0 = : k1 ½Rag0 + k1 + k2 ½Rag0 + k1k+ k2
(V.10)
1
Define KM hðk1 + k2 Þ=k1 and substitute (V.10) into (V.3), and solve for the concentration of phosphate, we get: ½Pi = k2 3
½Rag0 ½GTP0 t: ½Rag0 + KM
(V.11)
Because the readout from TLC is the fraction of GTP that is hydrolyzed, we have: Fraction of hydrolyzed GTP =
½Pi ½Rag0 = k2 3 thkobsd t; ½GTP0 ½Rag0 + KM
(V.12)
in which kobsd = k2 3½Rag0 =ð½Rag0 + KM Þ is the observed rate constant at a given Rag concentration. Notice that the TLC assay cannot differentiate GTP hydrolysis from individual subunits of the wild-type Rag GTPases. Therefore, the measured kcat and Km are the weighted average of the kcat and Km values for RagA and RagC. Considering the similar on-rates of GTP to RagA and RagC, and that RagA(T21N)-RagC and RagA-RagC(S75N) behave similarly in both single and multiple turnover reactions (Figures 3I–3K), the true values of kcat and Km for RagA and RagC should be similar and close to the measured values. For the half-site GTP hydrolysis reaction in Figure 3L, 10 mM of wild-type RagA-RagC was pre-equilibrated with 12 mM of unlabeled nucleotides (1.2 equal molar). After the Rags were singly loaded, 0.1 nM of g-32P-GTP was added to the mixture to start the reaction. Time points were taken and analyzed as above, and the observed hydrolysis rates were obtained by applying linear regression to the radioactive signals against time. For the half-site hydrolysis chase reaction in Figure 3O, 0.1 nM of g-32P-GTP was first incubated with 200 nM of wild-type RagARagC to start the reaction. Six time points were taken, before 100 mM of unlabeled nucleotides were added to the reaction mixture, after which more time points were taken. The observed hydrolysis rates were obtained by applying linear regression to the radioactive signals against time. For GATOR1- and FLCN-FNIP2-stimulated GTP hydrolysis reaction under single turnover conditions, increasing amounts of GATOR1 or FLCN-FNIP2 were included in the reaction that measures the hydrolysis rate of 0.1 nM of g-32P-GTP by 50 nM wild-type or mutant Rag GTPases. Time points were taken at each GATOR1 or FLCN-FNIP2 concentration to trace the reaction, and analyzed as above.
e6 Molecular Cell 68, 1–14.e1–e8, November 2, 2017
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
Multiple Turnover GTP Hydrolysis Experiments To determine the hydrolysis rates of the Rag GTPases under multiple turnover conditions, a fixed amount of Rag GTPases (2 mM for wild-type Rags), was incubated with increasing amounts of GTP doped with a trace amount of g-32P-GTP. The total GTP concentration used to determine the kinetic parameters for multiple turnover GTP hydrolysis of wild-type Rags are 2, 5, 10, 20, 50, and 100 mM (Figure 3G). The reactions were treated and analyzed similarly as above. Linear regression was used to fit the fraction of phosphate generated by GTP hydrolysis against time, generating the observed rate constants (kobsd). The kobsds were then fit to a Michaelis-Menten equation (Equation VI), to extract the kcat and KM values for the multiple turnover hydrolysis reaction. kobsd = kcat 3
½GTP0 KM + ½GTP0
(VI)
Equation VI was derived as following. Consider the reaction in a multiple turnover setup: k1
k3
k2
Rag + GTP 4 ½Rag,GTP/½Rag,GDP + Pi/Rag + GDP;
(VI.1)
in which k1 and k-1 represent the association and dissociation constants; k2 represents the catalytic rate for the chemistry step; k3 represents the dissociation constant of GDP. Because GDP dissociation is fast as determined by the off-rate measurement and GTP concentration is high, the intermediary species Rag,GDP can be omitted. Therefore, at the initial stage of the reaction (early steady-state), the intermediates are generated at the speed of: d½Rag,GTP = k1 ½Rag½GTP ðk1 + k2 Þ½Rag,GTP; dt
(VI.2)
d½Pi = k2 ½Rag,GTP: dt
(VI.3)
and
Using the steady-state assumption ðd½Rag,GTP=dtÞ = 0 and conservation of matter: ½Rag0 = ½Rag + ½Rag,GTP;
(VI.4)
the rate of GTP being hydrolyzed can be written as: ½Pi = k2 3
½Rag0 ½GTP0 t: ½GTP0 + k1k1+ k2
(VI.5)
Let KM hðk1 + k2 Þ=k1 and since the readout from TLC is the fraction of GTP that is hydrolyzed, we have: Fraction of hydrolyzed GTP =
½Pi ½Rag0 ½Rag0 = k2 3 thkobsd 3 t; ½GTP0 ½GTP0 + KM ½GTP0
(VI.6)
in which kobsd = kcat 3 ½GTP0 =ðKM + ½GTP0 Þ is the observed rate constant at a given GTP concentration. Similarly, the measured kcat and Km here are the weighted average of the kcat and Km values for RagA and RagC, which in principle should be close to one another. For GATOR1- and FLCN-FNIP2-stimulated GTP hydrolysis reactions under multiple turnover conditions, increasing amounts of GATOR1 or FLCN-FNIP2 were included in the reaction that monitors the hydrolysis rate of 250 mM GTP by 2 mM wild-type or mutant Rag GTPases. Time points were taken at each GATOR1 or FLCN-FNIP2 concentration to trace the reaction, and analyzed as above. Free Energy Diagram The free energy diagrams in Figures 1R and 1S were constructed based on the experimentally determined reaction rates and equilibrium constants using the standard state of 1 mM. For free-energy change at equilibrium, we used Equation VII: DG = RT ln K;
Equation VII
in which R = 1.987 cal mol1 K1, T = 277 K, and K is the equilibrium constant measured experimentally. For activation barrier, we used Equation VIII: DGz = RT ln
kh ; kB T
Equation VIII
in which R = 1.987 cal mol1 K1, T = 277 K, kB = 3.33 3 1024 cal K1, h = 1.583 3 1034 cal sec, and k is the experimentally measured rate constant. Preparation of Cell Lysates and Immunoprecipitates To determine the response of mTORC1 to amino acid stimulation or starvation under the control of Rag mutants, two million HEK293T cells were plated onto a 10-cm dish. 24 hr later, the cells were transfected with 2 ng of Flag-S6K1 and 100 ng of each Rag mutants.
Molecular Cell 68, 1–14.e1–e8, November 2, 2017 e7
Please cite this article in press as: Shen et al., Intersubunit Crosstalk in the Rag GTPase Heterodimer Enables mTORC1 to Respond Rapidly to Amino Acid Availability, Molecular Cell (2017), https://doi.org/10.1016/j.molcel.2017.09.026
36 hr later, amino acid stimulation or starvation was performed based on the procedure shown in Figures 6B and 6C. Cells were rinsed once with ice-cold PBS and lysed with Triton lysis buffer (40 mM NaHEPES, pH 7.4; 5 mM MgCl2; 100 mM ATP; 10 mM Na4P2O7; 10 mM Na b-glycerol phosphate; 1% Triton; and one tablet of protease inhibitor cocktail per 25 mL of buffer). The lysates were cleared by centrifugation at 15,000 rpm at 4 C in a microcentrifuge for 10 min. For anti-FLAG immunoprecipitations, the FLAGM2 beads were pre-equilibrated in Triton lysis buffer. 30 mL of a 50/50 slurry of the FLAG-M2 affinity beads was then added to cleared lysates and incubated at 4 C for two hours. Following immunoprecipitation, the beads were washed one time with Triton lysis buffer and 3 times with Triton lysis buffer supplemented with 500 mM NaCl. Immunoprecipitated proteins were denatured by the addition of 50 mL of 2.5 3 SDS buffer, resolved by SDS-PAGE, and analyzed by immunoblotting. western blots were quantified using ImageJ software. For the time-course experiments (Figures 6D and 6E), HEK293T cells were treated and lysed at precise time points, and the pS6K1 level was visualized as the indicator of mTORC1 activity. Western blots were quantified using ImageJ software. Lysine Modification Assay To detect the conformational changes of Rag GTPases in different nucleotide loading state, we adapted a lysine modification assay. Aliquots of 0.5 mg/ml wild-type Rag GTPase were first incubated with 200 mM GDP, 100 mM GDP and 100 mM GTP, or buffer on ice for two hours. 4 mM Sulfo-NHS-acetate was then added to the mixture, to irreversibly modify the amine group on the lysine side chain. The reactions were quenched by 0.2 M Tris-HCl (pH 8.0) after 10 min. The products were digested by chymotrypsin and the peptides were identified by mass spectrometry. Total ion current (TIC) for the peptide that contains modified lysine residue was used to quantify the modification efficiency. QUANTIFICATION AND STATISTICAL ANALYSIS All the kinetic assays were repeated at least three times, and the results were reported with Mean ± SEM.
e8 Molecular Cell 68, 1–14.e1–e8, November 2, 2017