Hematopoietic Progenitor Kinase-1 Structure in a Domain-Swapped Dimer

Short Article

Hematopoietic Progenitor Kinase-1 Structure in a Domain-Swapped Dimer Highlights d

The active and inactive HPK1 kinase in apo and ligand bound forms are characterized

d

Structures show rare domain-swapped dimer involving helical activation segment

Authors Ping Wu, Christopher J. Sneeringer, Keith E. Pitts, ..., Jane L. Grogan, Timothy P. Heffron, Weiru Wang

Correspondence

d

Dimerization observed in solution suggesting potential biologically relevance

[email protected]

d

Ligand complex structure provides insights for structurebased drug design

Enhancement of antigen-specific T cell immunity has shown significant therapeutic benefit in infectious diseases and cancer. Hematopoietic progenitor kinase-1 (HPK1) is one of the intracellular regulators that dampens T cell receptor signaling. Wu et al. studied the molecular structure of HPK1 kinase with bound ligand, which provides insights for structure-based drug design.

Wu et al., 2019, Structure 27, 125–133 January 2, 2019 ª 2018 Elsevier Ltd. https://doi.org/10.1016/j.str.2018.10.025

In Brief

Structure

Short Article Hematopoietic Progenitor Kinase-1 Structure in a Domain-Swapped Dimer Ping Wu,1 Christopher J. Sneeringer,2 Keith E. Pitts,2 Eric S. Day,3 Bryan K. Chan,4 Binqing Wei,4 Isabelle Lehoux,5 Kyle Mortara,5 Hong Li,6 Jiansheng Wu,6 Yvonne Franke,5 John G. Moffat,2 Jane L. Grogan,7 Timothy P. Heffron,4 and Weiru Wang1,8,* 1Department

of Structural Biology, Genentech, South San Francisco, CA 94080, USA of Biochemical Pharmacology, Genentech, South San Francisco, CA 94080, USA 3Department of Late Stage Pharmaceutical Development, Genentech, South San Francisco, CA 94080, USA 4Department of Discovery Chemistry, Genentech, South San Francisco, CA 94080, USA 5Department of Biomolecular Resources, Genentech, South San Francisco, CA 94080, USA 6Department of Protein Chemistry, Genentech, South San Francisco, CA 94080, USA 7Department of Cancer Immunology, Genentech, South San Francisco, CA 94080, USA 8Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.str.2018.10.025 2Department

SUMMARY

Enhancement of antigen-specific T cell immunity has shown significant therapeutic benefit in infectious diseases and cancer. Hematopoietic progenitor kinase-1 (HPK1) is a negative-feedback regulator of T cell receptor signaling, which dampens T cell proliferation and effector function. A recent report showed that a catalytic dead mutant of HPK1 phenocopies augmented T cell responses observed in HPK1knockout mice, indicating that kinase activity is critical for function. We evaluated active and inactive mutants and determined crystal structures of HPK1 kinase domain (HPK1-KD) in apo and ligand bound forms. In all structures HPK1-KD displays a rare domain-swapped dimer, in which the activation segment comprises a well-conserved dimer interface. Biophysical measurements show formation of dimer in solution. The activation segment adopts an a-helical structure which exhibits distinct orientations in active and inactive states. This face-to-face configuration suggests that the domain-swapped dimer may possess alternative selectivity for certain substrates of HPK1 under relevant cellular context.

INTRODUCTION Augmentation of T cell anti-tumor immunity has achieved transformational clinical success across a variety of tumor types (Chen and Mellman, 2017). The key to current cancer immunotherapy is to target surface immunoreceptors such as CTLA4 and PD1/PD-L1 with protein therapeutics to directly activate T cells within tumors (Buchbinder and Desai, 2016) or providing exogenous T cells to a patient with so-called CAR-T cell therapy. These successes have garnered attention and many seek to improve the response rates of approved agents through combination with additional drugs, including those that might over-

come specific negative regulation of immune function (Mahoney et al., 2015). Among particular targets under study is a small set of kinases, intracellular targets requiring small-molecule intervention, that have been identified as having potential relevance for cancer immunotherapy (Adams et al., 2015; Gross et al., 2015; Weinmann, 2016). Hematopoietic progenitor kinase-1 (HPK1), a serine/threonine Ste20-related protein kinase whose expression is restricted to the hematopoietic compartment (e.g., T cells, B cells, and dendritic cells), is one such kinase (Kiefer et al., 1996; Sawasdikosol et al., 2012). Whereas the vast majority of kinases are positive regulators of cellular function, HPK1 is a negative regulator of T cell receptor (TCR) signaling. As illustrated in Figure 1A, upon TCR activation, cytosolic HPK1 is recruited to the plasma membrane where it is phosphorylated at residue Tyr381, creating a docking site for SLP76 (Lasserre et al., 2011; Shui et al., 2007). Full activation of HPK1 kinase activity requires phosphorylation of residues Thr165 and Ser171 in the activation segment (AS). It has been shown that Ser171 can be phosphorylated subsequently to TCR activation in a PKD1-dependent manner (Arnold et al., 2005; Tsuji et al., 2001), whereas Thr165 appears to be an autophosphorylation site. Activated HPK1 phosphorylates SLP76 at residue Ser376 leading to destabilization of the TCR signaling complex and impediment of downstream T cell activation and proliferation (Di Bartolo et al., 2007; Lasserre et al., 2011; Shui et al., 2007). The HPK1-dependent negative regulation of T cell signaling can be alternatively triggered through the prostaglandin E2 receptor in a PKA-dependent manner (Sawasdikosol et al., 2007). Further studies using an HPK1 overexpression system suggested its role in activation-induced cell death (Arnold et al., 2001; Brenner et al., 2005, 2007) and in JNK activation (Hu et al., 1996; Kiefer et al., 1996), although the detailed mechanisms are less well understood. HPK1 has additional functions in regulating B cell receptor signaling in an analogous way to T cells (Sauer et al., 2001; Tsuji et al., 2001; Wang et al., 2012). In addition a putative role in limiting dendritic cell activation to TLR4 has been suggested (Alzabin et al., 2009). HPK1 protein consists of multiple domains: an N-terminal kinase domain, a C-terminal citron homology domain, and an intrinsically disordered central domain containing four proline-rich (PR) Structure 27, 125–133, January 2, 2019 ª 2018 Elsevier Ltd. 125

RESULTS

Figure 1. HPK1 Function as a Negative Regulator of TCR Signaling (A) TCR engagement triggers sequential events at membrane proximal lipid rafts, including activation of tyrosine kinases Lck and Zap70 and recruitment of adaptor protein SLP76 to LAT (Liou et al., 2000). Association of additional signaling proteins to SLP76 gives rise to a signalosome complex leading to T cell activation, proliferation, and cytokine production (Brownlie and Zamoyska, 2013; Malissen and Bongrand, 2015). Upon TCR activation, HPK1 is recruited to the TCR complex by adaptor proteins such as Gads and Grb2 (Ling et al., 2001; Liou et al., 2000; Liu et al., 2000). Lck/Zap70 phosphorylates HPK1 at residue Tyr381, which creates a docking site for SLP76 (Lasserre et al., 2011; Shui et al., 2007). Full activation of HPK1 kinase activity leads to phosphorylation of SLP76, recruitment of 14-3-3 and subsequent decomposition of the TCR complex, therefore dampens TCR signaling (Di Bartolo et al., 2007; Lasserre et al., 2011; Shui et al., 2007). (B) Sequence domain structure of HPK1.

motifs (Figure 1B). The PR motifs mediate interaction with SH3 domain-containing proteins, such as Grb2 and GADS (Boomer and Tan, 2005). The central domain also contains a caspase-3 cleavage site and full-length HPK1 is cleaved in apoptotic cells (Arnold et al., 2001). This suggests that HPK1 exerts differential functions in vivo depending on the cellular context and either as the full-length or as individual domains. Furthermore HPK1 might play an enzymatic and a scaffolding role. However, mice deficient of the HPK1 gene (HPK1.KO) have enhanced TCR signaling and T cell responses, and recent data comparing HPK1.KO with HPK1-kinase dead knockin mice (Hernandez et al., 2018) showed that this effect is solely due to the kinase activity. Thus a chemical inhibitor of the kinase function of HPK1 could potentially enhance T cells and boost anti-tumor immunity. Here we report the crystal structures of HPK1 kinase domain in apoenzyme and ligand bound forms. The structures consist a domain-swapped dimer that is agnostic to the variables in crystallization conditions. Using analytical ultra-centrifugation (AUC), we show that HPK1 forms reversible dimers in solution. The structure of HPK1 in complex with an ATP-competitive small-molecule inhibitor provides critical insights for rational drug design. 126 Structure 27, 125–133, January 2, 2019

Construct Design of HPK1 Kinase Domain We compared HPK1 with p21-activated kinase 1 (PAK1) and macrophage stimulating 1), two well-characterized members of Ste20 kinases to generate expression constructs for HPK1 kinase domain. Based on sequence homology we designed a series of constructs with various N and C termini (Figure S1A). The constructs were tested for overexpression in bacterial (E. coli) and insect cell expression systems. The constructs spanning amino acid residues Asp2-Asn293 (short) and Asp2Arg319 (long) achieved sufficient levels of expression in baculovirus-infected insect cells (T. ni and sf9). We generated mutant constructs mimicking the phosphorylated active state (Thr165Glu/Ser171Glu, TSEE) and inactive state (Ser171Ala, SA), of the kinase along with the wild-type sequence. The various proteins were purified to homogeneity and crystallized. In silico secondary structure prediction suggested low complexity in the ancillary C-terminal sequence of the long construct. Consistent with this prediction, subsequent crystallography results revealed that residues beyond Asn293 are disordered. In addition, the short construct generally yielded better diffracting crystals and therefore was the choice for following studies. We then tested kinase activity of the mutant and wild-type short constructs in a SLP76 phosphorylation activity assay and demonstrated that the activity profile was consistent with the mutational status (Figure 2A). We observed similar ATP Km among these proteins (Figure S1B) suggesting ATP kinetics is basically the same. The active kinase mutant exhibited the highest activity. The inactive kinase mimetic mutant exhibited some residual (signal/noise of
5) activity, about 5% of active mutant. As a positive control, a commercially available wild-type HPK1 kinase domain protein was tested in the same assay. This protein showed about 60% the activity of HPK1-TSEE. Structure of Active HPK1 Kinase Domain with an ATP Analog We determined the crystal structure of HPK1-TSEE in complex with a non-hydrolyzable ATP analog, AMPPNP, at 2.10 A˚ resolution. Each asymmetric unit is composed of a non-crystallographic symmetry (NCS)-related dimer. The overall structure of each protomer exhibits a bilobed architecture, typical of protein kinases (Figure 2B). A ‘‘hinge’’ region (Glu92-Gly97) connects the N-terminal lobe (Asp2-Met91) to the C-terminal lobe (Ser98Asn293) and provides flexibility for relative movement of the two lobes. The ATP binding site is situated at the interface between the N and C lobes (Figure 2B). HPK1 N lobe consists of a five-stranded anti-parallel b sheet (b1-b5) flanked by two helices—a small helix (aA) caps the apical face of the b sheet and a larger helix (aC) lies on the opposite side of the b sheet while in contact with the C-terminal lobe. The glycine-rich loop (G loop) connecting strands b1 and b2 is well defined in this structure, likely due to the presence of AMPPNP/Mg2+. A short loop (Lys49-Pro52) connecting b3 and aC is disordered as indicated by poor electron density. The C-terminal lobe of HPK1-TSEE core structure appears highly compact and rigid, predominantly consisting a helices. The AS (Asp155-Gly174) is folded in a threeturn a-helical structure at its N terminus (Figure 2B). The AS normally adopts extended loop conformations in most kinase

Figure 2. HPK1 Kinase Domain Structure (A) Kinase activity of HPK1 domain variants. Bars are defined as the average (n = 16) fluorescence intensity from 90-minute reactions using DELFIA. ‘‘SGNL’’ is (100 pM) HPK enzyme reacted with (100 nM) biotin-SLP76, in the presence of (50 mM) ATP. ‘‘BKGD" is all reaction components minus the enzyme. The error bars are standard deviations calculated from the average. (B) A ribbons diagram of HPK1-TSEE in complex with AMPPNP/Mg2+. The protein backbone trace is shown in ribbons, color spectrum by residue number with blue at N terminus and red at C terminus. Two phosphorylation sites in the activation segment, T165 and S171, are shown in spheres. (C) Upper panel is a close-up view of the active site with AMPPNP/Mg2+ bound. AMPPNP is shown in purple sticks. Mg2+ ions are shown in cyan spheres. Key residues interacting with AMPPNP are shown in thick sticks. H bond and salt bridge interactions are indicated by red dotted lines. The lower panel shows the domain-swapped dimer. (D) Superimposition of HPK1-TSEE (cyan) and HPK1-SA (brown). (E) A close-up view of the activation segment and surrounding structure in HPK1-SA.

structures; some exhibit a high degree of flexibility and undergo phosphorylation-dependent conformational changes. A distinct a-helical secondary structure in this region is unusual although has been observed in a small number of kinases such as SPAK

(Taylor et al., 2015) and GLK (Marcotte et al., 2017). AS helices typically hinder the aC helix from adopting the active conformation leading to inactive states of the kinase. The HPK1-TSEE structure, however, shows features resembling a catalytically Structure 27, 125–133, January 2, 2019 127

Figure 3. The Ligand Bound HPK1-SA Structure (A) The chemical structure of GNE-1858. (B) Dose-dependent inhibition of GNE-1858 of HPK1 variants. (C) The upper panel is a close-up view of ligand binding interactions. GNE-1858 is shown in yellow sticks. Water molecules are shown in red spheres. Key residues are shown in thick sticks. H bond interactions are indicated by red dotted lines. The number for a dotted line indicates distance in A˚. The lower panel shows the domain-swapped dimer of this structure.

competent enzyme as observed in the active PKA structure (Zheng et al., 1993), which is consistent with substrate binding. AMPPNP and cofactor Mg2+ ions are well resolved in this structure (Figures 2C and S1C). The adenine base is anchored in an aliphatic cleft composed of hydrophobic residues Leu23, Val31, Met91, and Leu144, and the hinge backbone, making two hydrogen bonds with hinge residues Glu92 and Cys94. The ribose moiety forms hydrogen bonds with Asp101 and Ala141. The a-, b-, and g-phosphate groups participate in metal ion coordination together with residues Asp137, Asn142, and Asp155. The catalytic lysine (Lys46) adopts the active conformation by making a hydrogen bond with the a-phosphate and a salt bridge with Glu62 from the aC helix. The phosphate groups bind under the G loop as expected (Figure 2C). In addition, the regulatory spine (R spine) (Taylor and Kornev, 2011) residues His135, Phe156, Leu66, and Tyr77 are assembled in a ‘‘kinase-active’’ conformation. The two Glu mutations in the AS are folded into the helix and participate in the NCS interface. They do not make interactions specifically promoting the active state. The AS helix supports the extensive interface constituent of the aforementioned NCS dimer. Such dimer interaction appears to cooperatively stabilize the AS helical conformation. We suspected that this apparent crystal-packing phenomenon might hint at an analogous protein-protein interaction that occurs in solution. This prompted us to carry out solution studies described later in this report. The Apo-HPK1 Structure in an Inactive State The HPK1-SA protein exhibits reduced kinase activity compared with the TSEE mutant in the SLP76 phosphorylation assay (Figure 2A), indicating that the Ser171 side chain may be required for activity, and the mutation could shift the HPK1-SA protein toward the inactive structural states. To gain understanding of kinase inactive states of HPK1, we determined the SA mutant structure at 1.90 A˚ resolution. The overall structure of HPK1SA (Figure S1E) resembles HPK1-TSEE (Figure 2B) with an identical bilobed architecture and similar secondary structure 128 Structure 27, 125–133, January 2, 2019

arrangement. However, the G loop is completely disordered likely due to lack of ligand binding. The aC helix displays an unusual bend at the N terminus that is reminiscent of the disordered loop observed in the HPK1-TSEE structure. Together, they suggest localized intrinsic flexibility in this part, although the functional relevance of such flexibility needs further investigation. In the C lobe, a major difference between the TSEE and SA structures is in the AS (Figure 2D). Similar to TSEE, the HPK1-SA AS is folded into an a helix, although taking a different orientation and inserting itself between the aC helix and the ATP binding site. Unlike TSEE, the HPK1-SA structure bears features of catalytically incompetent kinase (Figure 1E), which is consistent with the diminished activity. The AS helix blocks the salt bridge between Lys46 and Glu62. The aC helix is pushed away from the active site, resulting in a wider opening of the N, C lobe cleft. And the DFG motif adopts an ‘‘out’’ conformation therefore disrupts the regulatory spine. The low-level residue kinase activity associated with HPK1-SA protein suggests that it predominantly populates the inactive states represented in this crystal structure while having a small probability of adopting the active state structure as observed for HPK1-TSEE. The HPK1-SA AS helix also appears to contribute to the bending in the aC helix structure. Despite the large difference in orientation, the AS in HPK1-SA still participates in an NCS-related dimer interface. An Inhibitor Binds to the Active Site and Inhibits SPL76 Phosphorylation We screened the Genentech/Roche compound library and identified a diverse set of ATP-competitive inhibitors of HPK1. Here we report the biochemical and structural characterization of a screen hit, GNE-1858, containing an aminopyridine core scaffold (Figure 3A). Other hits will be described in subsequent publications. GNE-1858 potently inhibits all three HPK1 kinase domain variants under investigation in a dose-dependent manner (Figure 3B). The potencies against wild-type and the active mimetic mutants are identical with half maximal inhibitory concentration (IC50) values of 1.9 nM in the SLP76 phosphorylation assay. GNE-1858 also inhibited the residual kinase activity of HPK1SA at 4.5 nM (Figure 3B). We generated co-crystals of GNE1858 and HPK1-SA by soaking and determined a structure at 2.25 A˚ resolution. The compound is bound in the ATP binding

Figure 4. Characterization of HPK1 KD Dimer Interactions (A) HPK1-SA dimer. Two HPK1 molecules are colored in cyan and magenta. Each molecule is rendered in cartoon. The semi-transparent envelope is the molecule surface. (B) A close-up view of interactions at the dimer interface. (C) SEC of HPK1-TSEE showing monomer and dimer species. The inset is an SDS-PAGE image. (D) g(s*) distributions of 2.4 mg/mL (blue trace), 0.9 mg/mL (red trace), and 0.3 mg/mL (black trace) HPK1-TSEE over the same reduced sedimentation time for each concentration, normalized for loading concentration. (E) Global fit to monomer-dimer equilibrium plus a non-interacting component model from the same three protein concentrations as in (D).

site (Figures 3C and S1D). The amino azaindazole core scaffold binds to the hinge and forms two hydrogen bonds with the hinge backbone. The other exposed nitrogen of the azaindazole moiety makes water-mediated interactions with Asp155 and the backbone of the DFG motif at the N terminus of the AS, resulting in a DFG-in conformation. The pyrimidine occupies the ribose binding pocket and projects two pyrrolidine moieties into the surrounding. The N-linked pyrrolidine binds under the G loop, with the difluoro group making van der Waals contacts with Val31 and Gly24. The G loop appears to be stabilized by these interactions. The C-linked pyrrolidine is exposed to the bulk solvent. Not surprisingly, the co-structure with GNE-1858 displays similar structural features that resemble the inactive mutant described above for apo HPK1-SA. HPK1 Kinase Domain Forms a Domain-Swapped Dimer in Crystal We found that HPK1 kinase domain forms similarly arranged dimers in all crystals obtained under different experimental conditions (Figures S2A and S2D). Intriguingly, the dimer interface invariably involves a large portion of the AS and the following aEF helix (face-to-face dimer; Figure 4A). The buried surface

areas are 4,190 and 5,250 A˚2 and shape complementarity scores, Sc (Lawrence and Colman, 1993), are 0.76 and 0.77 for HPK1-TSEE and HPK1-SA, respectively. Although we cannot rule out the possibility of crystallization influence, this extensive and highly complementary interface could potentially be indicative of a significant and authentic protein-protein interaction. The dimer appears nearly 2-fold symmetrical. A prominent feature at the dimer interface is that the aEF helix of one protomer (aEFA, superscripts ‘‘A’’ or ‘‘B’’ denote protomers in the dimer) docks onto the other protomer by binding between two helices, aFB and aGB. Helices aEF, aF, and aG are common motifs in protein kinases. aEF normally folds back to the C lobe and binds between aF and aG in a similar manner as shown in the MAP4K4 example in Figures S2B and S2C. There appears to be a pocket between aF and aG (FG pocket) that can bind aEF either in cis or in trans. The in trans exchange of aEF motif represents a phenomenon known as domain-swap. Such domainswap in HPK1 is facilitated by the protruding AS (Figure 4A). Although the AS helix orientations are different between active and inactive mutant proteins as described above, the specific interactions at the dimer interface are remarkably conserved (Figures 4B and S2D). Notably, the loop connecting ASA and aEFA anchors in the FGB pocket of the partner protomer with a number of specific interactions. Starting from Tyr177A, the side chain is rigidified by stacking against Pro176A. Tyr177A side chain donates a hydrogen bond to the backbone oxygen of Val218B. Trp178A anchors deep in the FGB pocket making van der Waals contact with Glu207B, Pro213B, and Ser200B. Pro181A and Trp199B form another Pro-Trp pair stacking interaction. Finally, Glu182A of aEFA helix forms a salt bridge with Arg262B. HPK1 Dimer Detected in Solution Given repeated observation of dimers in crystallography, we investigated if HPK1 forms dimers in solution. Indeed, we Structure 27, 125–133, January 2, 2019 129

Table 1. Crystallography statistics HPK1_SA

HPK1_SA + GNE-1858

HKP1_TSEE + AMPPNP

PDB code

6CQE

6CQF

6CQD

Space group

P21212 a=76.5A˚, b=87.5A˚, c=99.0A˚,

C2221

C2

Resolution

a=b=g=90 1.90 A˚

a=91.9A˚, b=98.8A˚, c=77.1A˚, a=b=g=90 2.25 A˚

a=146.9A˚, b=51.9A˚, c=107.5A˚, a=g=0 , b=131.0 2.12 A˚

Total measured reflections

54162 (5338)a

16438 (1346)a

113955 (1147)a

Completeness (%)

100 (100)

96.4 (80.6)

99.9 (98.8)

Redundancy

6.0 (5.8)

5.8 (4.6)

3.3 (3.4)

I/s

23.0 (2.1)

27.4 (1.9)

8.8 (2.2)

Rsymb

0.070 (0.808) 50-1.90 A˚

0.059 (0.693)

0.075 (0.490)

Resolution range

50 - 2.25 A˚

50-2.12 A˚

Rcrystc / Rfreed

0.205/0.258

0.208/0.255

0.203/0.250

Non-hydrogen atoms

4931

2306

4855

Water molecules

383

34

218

Average B

22.8 A˚

79.2 A˚

49.6 A˚

r.m.s.d. bond lengths

0.007 A˚

0.005 A˚

0.005 A˚

Unit cell



r.m.s.d. angles

1.074

Ramachandran

0.940/0.045/0.006/0

0.954



0.878/0.118/0.004/0

0.847 0.903/0.091/0.006/0

a

Values in parentheses are of the highest resolution shell b Rsym = SjIhi - Ihj/SIhi , where Ihi is the scaled intensity of the ith symmetry-related observation of reflection h and Ih is the mean value. c Rcryst = ShjFoh - Fchj /ShFoh, where Foh and Fch are the observed and calculated structure factor amplitudes for reflection h. d Value of Rfree is calculated for 5% randomly chosen reflections not included in the refinement.

observed dimeric species in the final step of protein purification. Size-exclusion chromatography (SEC) separated two peaks with sizes corresponding to monomeric and dimeric HPK1 (Figure 4C). Concentration series of HPK1 proteins were analyzed by sedimentation velocity AUC to investigate this phenomenon in detail. It is characteristic of an associating system, in comparison of the distributions of the apparent sedimentation coefficient, g(s*), taken over the same reduced sedimentation time for each concentration shows an increase in the weight average sedimentation coefficient with increasing concentration (Figures 4D and S3A). Therefore the data were globally fit to a monomer-dimer equilibrium model using SedAnal (Stafford and Sherwood, 2004). Fitting to a model for monomer-dimer equilibrium with an additional non-interacting component explains the data well (Figures 4E and S3B). The molecular weight of the monomer, the sedimentation coefficients of the monomer, the dimer and the noninteracting component, as well as the KD for the monomerdimer equilibrium were floated as variables in the fit. The molecular weight of the dimer is fixed in the model to be twice the molecular weight of the monomer. The results are collected in Table S1. The reasonableness of the returned results was confirmed by calculating the predicted sedimentation coefficients of the dimer from the present structures using HydroPRO (Ortega et al., 2011). The results are also included in Table S1. These results confirm that HPK1-TSEE forms reversible dimers in solution with KD
60 mM, with approximately 5%–10% of a higher S component that does not participate in the reaction. 130 Structure 27, 125–133, January 2, 2019

DISCUSSION HPK1 is a Ste20-related protein kinase whose expression is restricted to the hematopoietic compartment (e.g., T cells, B cells, and dendritic cells). It is a critical negative regulator of TCR and BCR signaling; however, the mechanism by which it regulates other pathways across other cell types such as dendritic cells remain elusive. Here we report the crystal structures of HPK1 kinase domain in various states of activation. Our results reveal several special features in this key regulatory protein. First, the AS adopts a rigid a-helical structure for both active and inactive mutants. Second, it displays a dynamic domain-swapped face-toface dimer that is rarely observed in other kinases. Furthermore, the dimer persists in ATP bound active kinase and inhibitor bound inactive kinase. To our knowledge, this work is the first documentation of HPK1 kinase domain structure in literature. A remarkable feature observed in the HPK1-TSEE and HPK1-SA structures is the highly conserved homodimer interface, persistent in an otherwise different crystal-packing environment under variable crystallization conditions (Table 1). A common factor is protein concentration in crystallization always exceeded the KD determined by AUC. HPK1 dimerization, therefore, appears rather agnostic to the chemical environment but more of a concentration-driven phenomenon, which is characteristic of typical protein-protein interactions. The large area of the dimer interface likely reflects the dimer observed by AUC and SEC. To investigate the correspondence between the dimers in crystal and in solution, we attempted to generate dimer-deficient mutants guided by structures. However, all constructs had protein instability issues.

Dimerization-induced allosteric modulation of protein kinases has been increasingly appreciated in the literature over the past decade. Such interactions are shown to play critical roles in regulating kinase enzymatic activity or the activity of ancillary proteins. Prominent examples include RAF kinase activation by homo- or heterodimerization among isoforms, which in turn activates the MAPK signaling pathway (Rajakulendran et al., 2009); and IRE1 transphosphorylation by face-to-face dimerization or allosteric activation of RNase by back-to-back dimerization leading to unfolded protein response upon ER stress (Lee et al., 2008). Our HPK1 structures represent a distinct type of kinase dimer that involves domain-swap of the AS and aEF helices. While highly unusual, similarly domain-swapped kinase dimers have been observed in a small number of kinases, including Ste20 members OSR1 (Lee et al., 2009), SPAK (Taylor et al., 2015), and GLK (Marcotte et al., 2017). The function of such domain-swap has not yet been revealed. Pike et al. (2008) described this phenomenon as ‘‘activation segment exchange’’ and proposed that it could be a self-regulatory mechanism by which the kinase trans-autophosphorylates. The proposal implies that the homodimer represents an enzyme-substrate complex, as reported for PAK1 (Wang et al., 2011). While this may be a reasonable hypothesis for some kinases, it does not explain phosphorylation of T165 in HPK1 for the following reasons: (1) the crystal structures show T165 being distal from the ATP g-phosphate; (2) dimerization takes place regardless of the phosphorylation state of the AS; (3) the AUC experiment on HPK1-SA and HPK1-TSEE resulted in a similar dimerization constant. Intuitively, the product (HPK1pThr165) should leave the enzyme instead of keeping the enzyme-product complex intact with same affinity as the enzyme-substrate complex. The emerging trend of face-toface domain-swapped dimer among Ste20 family members might suggest that the dimerization plays a functional role. Comparing HPK1 with OSR1, SPAK, and GLK (Figure S4), reveals that aEF docking into the FG pocket is a common theme. HPK1 appears to involve more extensive area of interaction, and the AS is more rigid and involved in the dimer interface. A caveat is the dimerization affinity being relatively low,
60 mM. While biological relevance of the dimer remains poorly understood, we hypothesize that functional roles of HPK1 dimerization, if any, might depend on clustering of higher order protein complexes where HPK1 is involved. Full-length HPK1 encompasses a caspase-3 cleavage site (DDVD385) in the unstructured central domain. Human Jurkat T cell endogenous HPK1 was efficiently cleaved at Asp385 in a caspase-3-dependent manner during Fas-ligation-induced apoptosis (Chen et al., 1999). Interestingly, HPK1 cleavage appeared to enhance its kinase activity. Caspase-3-dependent processing of endogenous HPK1 was also shown in mouse progenitor cells (Arnold et al., 2007). These observations suggest that HPK1 can exist and function as either a multi-domain protein or separate kinase and citron homology domains in response to changes in the cellular environment. Together, current knowledge on HPK1 suggests that the dimer may play more complex regulatory roles other than autophosphorylation. It is noteworthy that AS helix orientation seems to drive the transition between the inactive and active states. HPK1 regulates different pathways via phosphorylation of multiple

substrate proteins. It is conceivable that the dimer might confer alternative substrate specificity, providing an additional level of regulation, and further differentiate functionally in response to signaling events. Further studies are needed to understand the function and regulation of monomer or dimer species of HPK1. The following methods are described in detail in the Supplemental Information: (1) cloning, expression, and purification of HPK1 kinase domain; (2) expression and purification of SLP76 protein; (3) SLP76 phosphorylation assays; (4) AUC; (5) crystallization, data collection, and structure determination. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d d d d

d d

KEY RESOURCES TABLE CONTACT FOR REAGENT AND RESOURCE SHARING EXPERIMENTAL MODEL AND SUBJECT DETAILS METHOD DETAILS B Chemical Synthesis of GNE-1858 B Analytical Ultracentrifugation B Cloning, Expression and Purification of HPK1 Kinase Domain B Expression and Purification of SLP76 Protein B SLP76 Phosphorylation Assays B Crystallization, Data Collection and Structure Determination QUANTIFICATION AND STATISTICAL ANALYSIS DATA AND SOFTWARE AVAILABILITY

SUPPLEMENTAL INFORMATION Supplemental Information includes four figures and one table can be found with this article online at https://doi.org/10.1016/j.str.2018.10.025. ACKNOWLEDGMENTS We thank Dr. Edward Kraft for assistance on protein expression. We thank Dr. James Kiefer for helpful discussion. We thank the staff at the Advanced Light Source (ALS) and at the Advanced Photon Source (APS) for assistance on data collection. Use of ALS and APS is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the US Department of Energy under contract no. DE-AC02-05CH11231 and no. DE-AC02-06CH11357, respectively. The crystal structures of HPK1-STEE, HPK1-SA, and HPK1-SA/G1858 complex are deposited into the PDB, accession codes PDB: 6CQD, 6CQE, and 6CQF, respectively. AUTHOR CONTRIBUTIONS W.W., P.W., E.S.D., Y.F., and J.M. conceived the experiments. W.W., E.S.D., J.L.G., and T.P.H. wrote the main manuscript. I.L., K.M., F.Y., H.L., J.W., and P.W. performed cloning, protein expression, purification, and crystallization, and wrote part of the methods. W.W. determined the crystal structures and wrote part of the methods. W.W., B.W., B.K.C., and T.P.H. analyzed the structures. C.J.S., K.E.P., and J.M. performed biochemical assays and wrote part of the methods. E.S.D. performed analytical ultra-centrifugation experiments and wrote part of the methods. All authors reviewed the manuscript. DECLARATION OF INTERESTS The authors declare competing financial interests. All authors are employees of Genentech a member of the Roche Group.

Structure 27, 125–133, January 2, 2019 131

Received: March 25, 2018 Revised: July 27, 2018 Accepted: October 25, 2018 Published: November 29, 2018 REFERENCES Adams, J.L., Smothers, J., Srinivasan, R., and Hoos, A. (2015). Big opportunities for small molecules in immuno-oncology. Nat. Rev. Drug Discov. 14, 603–622. Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010). PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, 213–221. Alzabin, S., Bhardwaj, N., Kiefer, F., Sawasdikosol, S., and Burakoff, S. (2009). Hematopoietic progenitor kinase 1 is a negative regulator of dendritic cell activation. J. Immunol. 182, 6187–6194. Arnold, K., Bordoli, L., Kopp, J., and Schwede, T. (2006). The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics 22, 195–201. Arnold, R., Frey, C.R., Muller, W., Brenner, D., Krammer, P.H., and Kiefer, F. (2007). Sustained JNK signaling by proteolytically processed HPK1 mediates IL-3 independent survival during monocytic differentiation. Cell Death Differ. 14, 568–575. Arnold, R., Liou, J., Drexler, H.C., Weiss, A., and Kiefer, F. (2001). Caspasemediated cleavage of hematopoietic progenitor kinase 1 (HPK1) converts an activator of NFkappaB into an inhibitor of NFkappaB. J. Biol. Chem. 276, 14675–14684. Arnold, R., Patzak, I.M., Neuhaus, B., Vancauwenbergh, S., Veillette, A., Van Lint, J., and Kiefer, F. (2005). Activation of hematopoietic progenitor kinase 1 involves relocation, autophosphorylation, and transphosphorylation by protein kinase D1. Mol. Cell. Biol. 25, 2364–2383. Boomer, J.S., and Tan, T.H. (2005). Functional interactions of HPK1 with adaptor proteins. J. Cell. Biochem. 95, 34–44. Brenner, D., Golks, A., Becker, M., Muller, W., Frey, C.R., Novak, R., Melamed, D., Kiefer, F., Krammer, P.H., and Arnold, R. (2007). Caspase-cleaved HPK1 induces CD95L-independent activation-induced cell death in T and B lymphocytes. Blood 110, 3968–3977. Brenner, D., Golks, A., Kiefer, F., Krammer, P.H., and Arnold, R. (2005). Activation or suppression of NFkappaB by HPK1 determines sensitivity to activation-induced cell death. EMBO J. 24, 4279–4290. Brownlie, R.J., and Zamoyska, R. (2013). T cell receptor signalling networks: branched, diversified and bounded. Nat. Rev. Immunol. 13, 257–269. Buchbinder, E.I., and Desai, A. (2016). CTLA-4 and PD-1 pathways: similarities, differences, and implications of their inhibition. Am. J. Clin. Oncol. 39, 98–106. Chen, D.S., and Mellman, I. (2017). Elements of cancer immunity and the cancer-immune set point. Nature 541, 321–330. Chen, Y.R., Meyer, C.F., Ahmed, B., Yao, Z., and Tan, T.H. (1999). Caspasemediated cleavage and functional changes of hematopoietic progenitor kinase 1 (HPK1). Oncogene 18, 7370–7377. Di Bartolo, V., Montagne, B., Salek, M., Jungwirth, B., Carrette, F., Fourtane, J., Sol-Foulon, N., Michel, F., Schwartz, O., Lehmann, W.D., et al. (2007). A novel pathway down-modulating T cell activation involves HPK-1-dependent recruitment of 14-3-3 proteins on SLP-76. J. Exp. Med. 204, 681–691. Emsley, P., and Cowtan, K. (2004). Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132. Gross, S., Rahal, R., Stransky, N., Lengauer, C., and Hoeflich, K.P. (2015). Targeting cancer with kinase inhibitors. J. Clin. Invest. 125, 1780–1789. Hayes, D., Laue, T., and Philo, J. (1995). Program Sednterp: Sedimentation Interpretation Program (Alliance Protein Laboratories). Hernandez, S., Qing, J., Thibodeau, R.H., Du, X., Park, S., Lee, H., Xu, M., Oh, S., Navarro, A., Roose-Girma, M., et al. (2018). The kinase activity of hemato-

132 Structure 27, 125–133, January 2, 2019

poietic progenitor kinase 1 (HPK1) is essential for the regulation of T cell function. Cell Rep. 25, 80–94. Hu, M.C., Qiu, W.R., Wang, X., Meyer, C.F., and Tan, T.H. (1996). Human HPK1, a novel human hematopoietic progenitor kinase that activates the JNK/SAPK kinase cascade. Genes Dev. 10, 2251–2264. Kabsch, W. (2010). XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132. Kiefer, F., Tibbles, L.A., Anafi, M., Janssen, A., Zanke, B.W., Lassam, N., Pawson, T., Woodgett, J.R., and Iscove, N.N. (1996). HPK1, a hematopoietic protein kinase activating the SAPK/JNK pathway. EMBO J. 15, 7013–7025. Lasserre, R., Cuche, C., Blecher-Gonen, R., Libman, E., Biquand, E., Danckaert, A., Yablonski, D., Alcover, A., and Di Bartolo, V. (2011). Release of serine/threonine-phosphorylated adaptors from signaling microclusters down-regulates T cell activation. J. Cell Biol. 195, 839–853. Lawrence, M.C., and Colman, P.M. (1993). Shape complementarity at protein/ protein interfaces. J. Mol. Biol. 234, 946–950. Lee, K.P., Dey, M., Neculai, D., Cao, C., Dever, T.E., and Sicheri, F. (2008). Structure of the dual enzyme Ire1 reveals the basis for catalysis and regulation in nonconventional RNA splicing. Cell 132, 89–100. Lee, S.J., Cobb, M.H., and Goldsmith, E.J. (2009). Crystal structure of domainswapped STE20 OSR1 kinase domain. Protein Sci. 18, 304–313. Ling, P., Meyer, C.F., Redmond, L.P., Shui, J.W., Davis, B., Rich, R.R., Hu, M.C., Wange, R.L., and Tan, T.H. (2001). Involvement of hematopoietic progenitor kinase 1 in T cell receptor signaling. J. Biol. Chem. 276, 18908–18914. Liou, J., Kiefer, F., Dang, A., Hashimoto, A., Cobb, M.H., Kurosaki, T., and Weiss, A. (2000). HPK1 is activated by lymphocyte antigen receptors and negatively regulates AP-1. Immunity 12, 399–408. Liu, S.K., Smith, C.A., Arnold, R., Kiefer, F., and McGlade, C.J. (2000). The adaptor protein Gads (Grb2-related adaptor downstream of Shc) is implicated in coupling hemopoietic progenitor kinase-1 to the activated TCR. J. Immunol. 165, 1417–1426. Mahoney, K.M., Rennert, P.D., and Freeman, G.J. (2015). Combination cancer immunotherapy and new immunomodulatory targets. Nat. Rev. Drug Discov. 14, 561–584. Malissen, B., and Bongrand, P. (2015). Early T cell activation: integrating biochemical, structural, and biophysical cues. Annu. Rev. Immunol. 33, 539–561. Marcotte, D., Rushe, M., M Arduini, R., Lukacs, C., Atkins, K., Sun, X., Little, K., Cullivan, M., Paramasivam, M., Patterson, T.A., et al. (2017). Germinal-center kinase-like kinase co-crystal structure reveals a swapped activation loop and C-terminal extension. Protein Sci. 26, 152–162. McCoy, A.J., Grosse-Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C., and Read, R.J. (2007). Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674. Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240–255. Ortega, A., Amoros, D., and Garcia de la Torre, J. (2011). Prediction of hydrodynamic and other solution properties of rigid proteins from atomic- and residue-level models. Biophys. J. 101, 892–898. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. In Methods in Enzymology: Macromolecular Crystallography, Part A, C.W. Carter, ed. (Academic Press), pp. 307–326. Pike, A.C., Rellos, P., Niesen, F.H., Turnbull, A., Oliver, A.W., Parker, S.A., Turk, B.E., Pearl, L.H., and Knapp, S. (2008). Activation segment dimerization: a mechanism for kinase autophosphorylation of non-consensus sites. EMBO J. 27, 704–714. Rajakulendran, T., Sahmi, M., Lefrancois, M., Sicheri, F., and Therrien, M. (2009). A dimerization-dependent mechanism drives RAF catalytic activation. Nature 461, 542–545. Sauer, K., Liou, J., Singh, S.B., Yablonski, D., Weiss, A., and Perlmutter, R.M. (2001). Hematopoietic progenitor kinase 1 associates physically and functionally with the adaptor proteins B cell linker protein and SLP-76 in lymphocytes. J. Biol. Chem. 276, 45207–45216.

Sawasdikosol, S., Pyarajan, S., Alzabin, S., Matejovic, G., and Burakoff, S.J. (2007). Prostaglandin E2 activates HPK1 kinase activity via a PKA-dependent pathway. J. Biol. Chem. 282, 34693–34699. Sawasdikosol, S., Zha, R., Yang, B., and Burakoff, S. (2012). HPK1 as a novel target for cancer immunotherapy. Immunol. Res. 54, 262–265. Shui, J.W., Boomer, J.S., Han, J., Xu, J., Dement, G.A., Zhou, G., and Tan, T.H. (2007). Hematopoietic progenitor kinase 1 negatively regulates T cell receptor signaling and T cell-mediated immune responses. Nat. Immunol. 8, 84–91. Stafford, W.F., and Sherwood, P.J. (2004). Analysis of heterologous interacting systems by sedimentation velocity: curve fitting algorithms for estimation of sedimentation coefficients, equilibrium and kinetic constants. Biophys. Chem. 108, 231–243. Taylor, C.A., Juang, Y.C., Earnest, S., Sengupta, S., Goldsmith, E.J., and Cobb, M.H. (2015). Domain-swapping switch point in ste20 protein kinase SPAK. Biochemistry 54, 5063–5071. Taylor, S.S., and Kornev, A.P. (2011). Protein kinases: evolution of dynamic regulatory proteins. Trends Biochem. Sci. 36, 65–77. Tsuji, S., Okamoto, M., Yamada, K., Okamoto, N., Goitsuka, R., Arnold, R., Kiefer, F., and Kitamura, D. (2001). B cell adaptor containing src homology

2 domain (BASH) links B cell receptor signaling to the activation of hematopoietic progenitor kinase 1. J. Exp. Med. 194, 529–539. Wang, J., Wu, J.W., and Wang, Z.X. (2011). Structural insights into the autoactivation mechanism of p21-activated protein kinase. Structure 19, 1752–1761. Wang, X., Li, J.P., Kuo, H.K., Chiu, L.L., Dement, G.A., Lan, J.L., Chen, D.Y., Yang, C.Y., Hu, H., and Tan, T.H. (2012). Down-regulation of B cell receptor signaling by hematopoietic progenitor kinase 1 (HPK1)-mediated phosphorylation and ubiquitination of activated B cell linker protein (BLNK). J. Biol. Chem. 287, 11037–11048. Weinmann, H. (2016). Cancer immunotherapy: selected targets and smallmolecule modulators. ChemMedChem 11, 450–466. Winn, M.D., Ballard, C.C., Cowtan, K.D., Dodson, E.J., Emsley, P., Evans, P.R., Keegan, R.M., Krissinel, E.B., Leslie, A.G., McCoy, A., et al. (2011). Overview of the CCP4 suite and current developments. Acta Crystallogr. D Biol. Crystallogr. 67, 235–242. Zheng, J., Trafny, E.A., Knighton, D.R., Xuong, N.H., Taylor, S.S., Ten Eyck, L.F., and Sowadski, J.M. (1993). 2.2 A refined crystal structure of the catalytic subunit of cAMP-dependent protein kinase complexed with MnATP and a peptide inhibitor. Acta Crystallogr. D Biol. Crystallogr. 49, 362–365.

Structure 27, 125–133, January 2, 2019 133

STAR+METHODS KEY RESOURCES TABLE

REAGENT or RESOURCE

SOURCE

IDENTIFIER

Expression Systems, LLC

Cat # 91-001

Bacterial and Virus Strains Virus strain: BestBac 1.0 Experimental Models: Cell Lines Sf9 cells in serum-free ESF921

Expression Systems, LLC

Cat # 94-001F

Cellfectin

Invitrogen

Cat # 10362-010

Chemicals, Peptides and Recombinant Proteins Wild type HPK1 kinase domain

SignalChem

Cat #M23-116-100

SLP76

This paper

N/A

HisTrap FF column

GE Healthcare Life Sciences

Cat # 17525501

Hitrap SP-HP column

GE Healthcare Life Sciences

Cat # 17115101

JB Kinase screen kit

Jena Bioscience

Cat # CS-204-L

Silver Bullet additive screen

Hampton Research

Cat # HR2-096

HPK1-STEE

This paper

PDB: 6CQD

HPK1-SA

This paper

PDB: 6CQE

HPK1-SA/G1858

This paper

PDB: 6CQF

MST1

Protein Data Bank

PDB: 3COM

OSR1

(Lee et al., 2009)

PDB:3DAK

SPAK

(Taylor et al., 2015)

PDB:5D9H

GLK

(Marcotte et al., 2017)

PDB:5J5T

pAcgp67

BD Biosciences

Cat # 554759

BestBac linearized viral DNA

Expression Systems, LLC

Cat # 91-001

Deposited Data

Recombinant DNA

Software and Algorithms HKL2000

(Otwinowski and Minor, 1997)

http://www.hkl-xray.com/

XDS

(Kabsch, 2010)

http://xds.mpimf-heidelberg.mpg.de/

Phenix

(Adams et al., 2010)

https://www.phenix-online.org/

Phaser

(McCoy et al., 2007)

(included in CCP4 and PHENIX)

CCP4 program suite

(Winn et al., 2011)

http://www.ccp4.ac.uk/

COOT

(Emsley and Cowtan, 2004)

(included in CCP4 and PHENIX)

REFMAC5

(Murshudov et al., 1997)

(included in CCP4)

SWISS-MODEL

(Arnold et al., 2006)

https://swissmodel.expasy.org/

SedAnal

(Stafford and Sherwood, 2004)

http://www.sedanal.org/

HydroPRO

(Ortega et al., 2011)

http://leonardo.inf.um.es/macromol/ programs/programs.htm

SEDNTERP v1.09

(Hayes et al., 1995)

http://www.jphilo.mailway.com/download.htm

PyMOL

Schrodinger LLC

https://pymol.org

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 Contect Weiru Wang ([email protected])

e1 Structure 27, 125–133.e1–e4, January 2, 2019

EXPERIMENTAL MODEL AND SUBJECT DETAILS Transfer vectors were co-transfected with BestBac linearized viral DNA (Expression Systems, LLC) into Sf9 cells using Cellfectin (Invitrogen) to produce recombinant baculovirus. The virus was amplified twice to prepare the stock used for protein expression. A 22L Wave bioreactor was inoculated with Sf9 cells (Expression Systems, LLC) at 1x106 cells/ml in serum-free ESF921 (Expression Systems, LLC). The cells were grown to 2x106 cells/ml and infected with the appropriate virus at a ratio of virus/culture of 2.5 mL/L (vol/vol) for Sf9 cells. The Wave reactor was maintained at 27 C and 25 R.P.M. with a fixed angle of 9 and 0.3L/m of 30% oxygen. At 72 hours post infection, the Sf9 cells were harvested by centrifugation at 4000 R.C.F. for 15 minutes and frozen at -80 C. METHOD DETAILS Chemical Synthesis of GNE-1858 To a mixture of 6-chloro-1-isopropyl-1H-pyrazolo[4,3-c]pyridine (826 mg, 4.2 mmol), tris(dibenzylideneacetone)dipalladium(0) (35mg, 0.039 mmol), 2-(di-tert-butylphosphino)-20 ,40 ,60 - triisopropyl-3,6-dimethoxy-1,10 -biphenyl (94 mg, 0.19 mmol) and sodium tert-butoxide (566 mg, 5.9 mmol) in 1,4-dioxane (25 mL) was added a solution of ammonia in 1,4-dioxane (25.2 mL, 0.5 M, 12.6 mmol). The reaction was then degassed by nitrogen bubbling, sealed and stirred at 100 C for 4 h. The reaction mixture was partitioned between ethyl acetate and water. The organic extracts were combined, washed with brine, dried over sodium sulfate and concentrated. The residue was purified by flash column chromatography (0-100% EtOAc in cyclohexane) to give 1-isopropyl-1H-pyrazolo[4,3-c]pyridin-6-amine as a white solid (540 mg, 73%). A mixture of 1-isopropyl-1H-pyrazolo[4,3-c]pyridin-6-amine (485 mg, 2.75 mmol), tert-butyl 3-(2,6-dichloropyrimidin-4-yl) pyrrolidine-1-carboxylate (795 mg, 2.5 mmol), sodium tert-butoxide (290 mg, 3 mmol), Xantphos (221 mg, 0.38 mmol), tris(dibenzylideneactone dipalladium(0) (358 mg, 0.38 mmol) in 1,4-dioxane (10 mL) was stirred under a nitrogen atmosphere at 80 C for 2 h. The reaction was filtered and concentrated. The crude material was purified by flash column chromatography (0-100% EtOAc in heptane) to give (±)-tert-butyl 3-(2-chloro-6-((1-isopropyl-1H-pyrazolo[4,3-c]pyridin-6-yl)amino)pyrimidin-4-yl) pyrrolidine-1-carboxylate (351 mg, 31%). A mixture of (±)-tert-butyl 3-(2-chloro-6-((1-isopropyl-1H-pyrazolo[4,3-c]pyridin-6-yl)amino)pyrimidin-4-yl)pyrrolidine-1carboxylate (0.20 g, 0.44 mmol) and 3,3-difluoropyrolidine (0.13 g, 0.87 mmol) and N,N-diisopropylethylamine (0.4 mL, 2 mmol) in dimethylformamide (5 mL) was stirred overnight at 80 C. The reaction was then diluted with water and extracted with ethyl acetate. The combined extracts were washed with water and brine, dried over sodium sulfate, filtered and concentrated. The crude material was purified by flash column chromatography (0-100% EtOAc in heptane) to give (±)-tert-butyl 3-(2-(3,3-difluoropyrrolidin-1-yl)-6((1-isopropyl-1H-pyrazolo[4,3-c]pyridin-6-yl)amino)pyrimidin-4-yl)pyrrolidine-1-carboxylate. The material was then dissolved in a mixture of 1,4-dioxane (5 mL) and 4M HCl in 1,4-dioxane (5 mL). The mixture was stirred at room temperature for 3 h. The reaction mixture was concentrated. The desired product was purified by reverse phase HPLC to give (±)-N-(2-(3,3-difluoropyrrolidin-1-yl)-6(pyrrolidin-3-yl)pyrimidin-4-yl)-1-isopropyl-1H-pyrazolo[4,3-c]pyridin-6-amine (GNE-1858, 157 mg, 84%). 1H NMR (400 MHz, CDCl3) d 8.72 (d, J = 1.1 Hz, 1H), 8.39 (t, J = 1.1 Hz, 1H), 8.06-7.96 (m, 1H), 7.39 (s, 1H), 6.02 (s, 1H), 4.78-4.71 (m, 1H), 4.02 (t, J = 13.2 Hz, 2H), 3.89 (t, J = 7.3 Hz, 2H), 3.23-3.18 (m, 1H), 3.12 (m, 3H), 2.99-2.93 (m, 1H), 2.53-2.46 (m, 2H), 2.17-2.09 (m, 1H), 2.03-1.94 (m, 1H), 1.62 (d, J = 6.7 Hz, 6H). LCMS (ESI): [M+H]+= 429.2. Analytical Ultracentrifugation A concentration series of HPK1-SA, run at 1.7, 0.4 and 0.2 mg/mL, was analyzed by sedimentation velocity analytical ultracentrifugation using 3 mm path length centerpieces. Approximately 150 absorbance scans at 280 nm were collected for each sample. Comparison of the distributions of the apparent sedimentation coefficient, g(s*), taken over the same reduced sedimentation time for each concentration shows an increase in the weight average sedimentation coefficient with increasing concentration (Figure S3A). This is characteristic of an associating system; therefore the series was globally fit to a monomer – dimer equilibrium model using SedAnal (Stafford and Sherwood, 2004). Data were selected over a reduced sedimentation time slightly expanded relative to that used for g(s*). SEDANAL performs a Levenberg-Marquardt nonlinear least-squares fit between time-differenced sedimentation velocity data and ideal sedimentation velocity profiles generated using the finite element method of Claverie ((Stafford and Sherwood, 2004) and references therein) and user-defined equilibrium mass action equations. Fitting to a model for monomer-dimer equilibrium with an additional non-interacting component explains the data well. The molecular weight of the monomer, the sedimentation coefficients of the monomer, the dimer and the non-interacting component and the KD for the monomer-dimer equilibrium were floated as variables in the fit. The protein concentration in each cell was floated as a local parameter. (Figure S3B). Shown in Figure 4E, each concentration are the first and last ordered sets out of the 25 sets used in the fit. The red points are concentration differences, DC(obs), at constant radius between two absorbance scans taken at different times. The solid lines are concentration differences, DC(calc), calculated from the parameters being fit. The DC(obs) and DC(calc) correspond to the left y-axis in units of absorbance at 280 nm. The plotted deviations between the observed and calculated concentration differences (blue points) are on the right y-axis using the same units and scale but offset to be at the bottom of each plot. The x-axis, radius of the cell in cm, is the same for both y-axes. The results are collected in Table S1. The reasonableness of the returned sedimentation coefficient results was confirmed by calculating the predicted sedimentation coefficients of the dimer from the present structures using HydroPRO (Ortega Structure 27, 125–133.e1–e4, January 2, 2019 e2

et al., 2011). The results are also included in Table S1. These results confirm that HPK1-SM forms reversible dimers in solution with KD
80 mM and approximately 5 – 10% of the protein exists as a higher S species that does not participate in the reaction. Cloning, Expression and Purification of HPK1 Kinase Domain DNA encoding WT or mutant HPK1protein was cloned into a slightly modified version of pAcgp67 (BD Biosciences). Transfer vectors were co-transfected with BestBac linearized viral DNA (Expression Systems, LLC) into Sf9 cells using Cellfectin (Invitrogen) to produce recombinant baculovirus. The virus was amplified twice to prepare the stock used for protein expression. A 22L Wave bioreactor was inoculated with Sf9 cells (Expression Systems, LLC) at 1x106 cells/ml in serum-free ESF921 (Expression Systems, LLC). The cells were grown to 2x106 cells/ml and infected with the appropriate virus at a ratio of virus/culture of 2.5 mL/L (vol/vol) for Sf9 cells. The Wave reactor was maintained at 27 C and 25 R.P.M. with a fixed angle of 9 and 0.3L/m of 30% oxygen. At 72 hours post infection, the Sf9 cells were harvested by centrifugation at 4000 R.C.F. for 15 minutes and frozen at -80 C. The kinase domain of HPK1 (D2-N293, S171A) with an N-terminal cleavable His6-tag was expressed intracellularly in sf9 cells. Cell paste was harvested by centrifugation, re-suspended in the lysis buffer (50 mM Tris-HCl pH 7.5, 500 mM NaCl, 10 % glycerol and 0.5 mM TCEP), homogenized, and passed through a microfluidizer three times. The lysate, clarified by centrifugation and passed through a 0.45 mm filter, was loaded onto a HisTrap FF column (GE Healthcare), and eluted with a linear imidazole gradient (20-250 mM). Fractions containing HPK1 kinase were pooled and digested with TEV protease to remove the His6-tag. The de-tagged protein was separated from the residual His tagged protein on a second Ni-column then further purified over a Hitrap SP-HP column. Finally the kinase was purified on a S75 column equilibrated with 20 mM Hepes pH7.2, 300 mM NaCl and 0.5 mM TCEP. Pooled HPK1 kinase was concentrated to 18 mg/mL for crystallization. The kinase domain of HPK1 (D2-N293, T165E/S171E) was cloned, expressed and purified using the same protocol described above. Expression and Purification of SLP76 Protein N-terminal His-tagged and Avi-tagged full-length human SLP76 construct was transformed into BL21(DE3) E. coli cells. The cells were grown at 17 C in 8 L TB-Auto media and were harvested after 48 hours of auto-induction. The pellet was re-suspended in a buffer containing 50 mM Tris/HCl, pH 8.0, 0.15 M NaCl, 20 mM Imidazole, 1 mM TCEP. The cells were homogenized by passing through a microfluidizer twice at 1000 bar, followed by centrifugation at 40,000 rpm for 1 hour. Cell lysates were loaded onto a Ni-NTA column (XK16/20, 3mL by GE Healthcare) in a buffer containing 20 mM Tris/HCl, pH 8.0, 0.3 M NaCl, 20 mM Imidazole, 0.5 mM TCEP and the bound SLP76 protein was eluted with 250 mM imidazole in the same buffer. The protein was then further purified by a size-exclusion column (Superdex 200 GE Healthcare) in a buffer containing 50 mM Tris/HCl, pH 7.5, 0.15 M NaCl, 0.5 mM TCEP, 10% Glycerol. The protein was then loaded on a Q HP column (GE Healthcare) equilibrated in a low salt buffer (20 mM Tris, pH 8.0, 10% glycerol, 0.5 mM TCEP) before elution with 1.0 M NaCl in the same buffer. The elution was done by salt gradient at 20-40% in 40 CV. The protein was then dialyzed against a final buffer containing 50 mM Tris, pH 7.5, 0.15 M NaCl, 0.5 mM TCEP, and 10% gelycerol, and concentrated to 1.5 mg/mL. SLP76 Phosphorylation Assays Enzyme reactions were performed in Proxiplate Plus 384-well microplates at 8 mL per well. An Echo acoustic dispenser was used to add inhibitor to the plate; 11-point, 4-fold dose response at 80 nL per well. Buffer conditions were: 50 mM of HEPES pH 7.5, 0.01% Brij 35, 10 mM MgCl2, 2 mM TCEP and 0.01% bovine skin gelatin. Reaction conditions were: 100 pM HPK, 50 mM ATP, 100 nM biotinSLP76, 90 minutes at RT. Detection of phosphorylated biotin-SLP76 was by anti-phospho-S376 antibody, followed by Europium labeled secondary antibody, using a standard DELFIA immunoassay protocol. IC50s were determined by fitting curves to a standard four parameter equation in Prism 7. Wild type kinase domain was purchased from SignalChem (cat #M23-116-100) Crystallization, Data Collection and Structure Determination The HPK1-SA crystallization conditions were found by spars-matrix screening. The hits from JB Kinase screen kit condition G5 was optimized using Silver Bullet additive screen (Hampton Research). The final crystals were obtained by setting up in 24-well Linbro plates using hanging drop vapor diffusion method at room temperature. Each well contains 500 mL of reservoir solution (0.1 M Tris-HCl, pH 8.5, 0.25 M sodium tartrate and 12% PEG 8000). Each drop contains 1 mL of protein solution, 0.5 mL of reservoir solution and 0.5 mL of Silver Bullet condition C11. Micro-seeding helped to consistently produce crystals with desirable size. Crystals usually appeared after overnight incubation at 19 C, and reached optimal size in three to five days. Crystals were briefly treated with cryoprotectant containing reservoir solution with additional 25% ethylene glycol, then flash frozen in liquid nitrogen and were maintained at cryogenic temperature for collection. Apo crystals of HPK1_SA were soaked overnight against 0.1 mM inhibitor compound in crystallization reservoir solution. The crystals were briefly treated with cryo protectant containing the reservoir solution with additional 20% glycerol and 0.1 mM GNE-1858, then flash frozen in liquid nitrogen and were maintained at cryogenic temperature for data collection. A spars-matrix screen was performed with the HPK1-TSEE protein (12.3 mg/mL) incubated with 10 mM MgCl2 and 5 mM AMPPNP. In the final condition, the pre-incubated protein sample was mixed 1:1 with reservoir solution composed of 0.1 M HEPES pH 7.5 and 12% PEG 8000. Crystals were grown at 19 C using setting-drop vapor diffusion method. The crystals were briefly treated with cryo-protectant containing 0.1 M HEPES pH 7.5 and 35% PEG3350, then flash frozen in liquid nitrogen and were maintained at cryogenic temperature for data collection. e3 Structure 27, 125–133.e1–e4, January 2, 2019

The diffraction data of apo HPK1-SA, HPK1-SA/GNE-1858 and HPK1-TSEE/AMPPNP crystals were collected using monochromatic X-rays at the Advanced Photon Source (APS) beam line 22ID (Rayonix M300 CCD detector), beam line 21IDF (Rayonix M225 CCD detector) and Advanced Light Source (ALS) beam line 5.0.2 (PILATUS3 6M detector), respectively. Rotation method was applied to a single crystal for each of the complete data set. The crystals were kept at cryogenic temperature throughout the data collection process. Data reduction was done using the program HKL2000 (Otwinowski and Minor, 1997) for apo HPK1-SA data, and XDS (Kabsch, 2010) and the CCP4 program suite (Winn et al., 2011) for HPK1-SA/GNE-1858 and HPK1-TSEE/AMPPNP data sets. Data reduction statistics are shown in Table S1. The apo HPK1-SA structure was phased by molecular replacement (MR) using program Phaser (McCoy et al., 2007). We generated a homology model of HPK1-SA based on a crystal structure of MST1 (PDB code 3COM) using program SWISS-MODEL (Arnold et al., 2006). The homology model was used as the MR search model. Initial difference electron density indicated significant conformational difference in the activation segment, aC-helix, and C-terminal helix. Manual rebuilding was performed with graphics program COOT (Emsley and Cowtan, 2004). The structure was further refined iteratively using program REFMAC5 (Murshudov et al., 1997) and PHENIX (Adams et al., 2010) using maximum likelihood target functions, anisotropic individual B-factor refinement and TLS refinement, and to achieve final statistics shown in Table S1. The refined structure resembled the overall structure of the homology model with RMSD of Ca atoms of 1.1A˚ in the core region excluding the activation segment and aC-helix. The structures HPK1-SA/GNE-1858 and HPK1-TSEE/AMPPNP were phase by MR using the program Phaser. Apo-HPK1-SA was used as the MR search model. The ligands were built into the difference density with program COOT. The structures were subsequently refined similarly as for apo HPK1-SA structure. The final refinement statistics are shown in Table 1. QUANTIFICATION AND STATISTICAL ANALYSIS Statistics generated from X-ray crystallography data processing, refinement, and structure validation are displayed in Table 1. DATA AND SOFTWARE AVAILABILITY The accession numbers for the crystal structures of HPK1-STEE, HPK1-SA and HPK1-SA/G1858 complex are in this paper are PDB:6CQD, PDB:6CQE and PDB:6CQF, respectively.

Structure 27, 125–133.e1–e4, January 2, 2019 e4