- Email: [email protected]
Application of tyrosine-tryptophan fluorescence resonance energy transfer in monitoring protein size changes
Analytical Biochemistry 557 (2018) 142–150
Contents lists available at ScienceDirect
Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio
Application of tyrosine-tryptophan fluorescence resonance energy transfer in monitoring protein size changes
T
Kenneth B. Davis, Zihan Zhang, Elizaveta A. Karpova, Jun Zhang∗ Department of Chemistry, College of Arts and Sciences, University of Alabama at Birmingham, CH266, 901 14th Street South, Birmingham, AL, 35294-1240, USA
A R T I C LE I N FO
A B S T R A C T
Keywords: FRET Protein unfolding Intrinsically disordered proteins Fluorescence Ligand binding FirbY-W
Monitoring protein size changes has versatile applications in studying protein folding/unfolding, conformational rearrangements, and ligand binding. Traditionally, FRET has been used to obtain this information. However, the use of FRET often requires covalent attachment of exogenous fluorophores. Although intrinsic FRET also exists between tyrosine and tryptophan residues, it has been underused because of tyrosinate formation and spectroscopic overlap. Herein, we clarified the concern of tyrosinate formation and mathematically deconvoluted tyrosine/tryptophan fluorescence spectra. We define a new parameter called FirbY-W (fluorescence intensity ratio between tyrosine and tryptophan) to reflect protein sizes. We demonstrate its applications in studying protein unfolding using several model proteins. In all the cases, our method offers superior sensitivity, data quality, and robustness compared with traditional techniques. The unique power of our method is in its ability to detect elusive conformational changes of intrinsically disordered proteins (IDP). The lack of structure makes IDPs unsuitable for CD or tryptophan fluorescence characterization. Using histone mRNA stem-loop binding protein (SLBP) as an example of disordered proteins, we showed that our method is capable of detecting conformational changes caused by phosphorylation, which are effectively invisible for traditional spectroscopic methods. Our method can also be used to detect RNA binding of disordered proteins.
Introduction Many biochemical questions, such as protein folding, ligand binding or conformational changes, can be ultimately simplified as measuring intra- and inter-atomic distances. One approach to measuring the desired distances is fluorescence resonance energy transfer (FRET). However, application of FRET to protein systems usually requires introducing exogenous donor/acceptor pairs [1]. This introduction is cumbersome and may introduce unnecessary artifacts into the system. Therefore, circular dichroism (CD), infrared spectroscopy, or tryptophan fluorescence are often used, each with shortcomings [2–6]. Specifically, CD and infrared are mainly sensitive to protein secondary structure but not tertiary structure. Tryptophan fluorescence intensity and emission maximum can be useful, assuming that folding or binding events perturb tryptophan microenvironments. The validity of this assumption, however, depends on the location of tryptophan residues. In
addition, due to the fact that spectroscopic signals are proportional to sample concentration, small errors in pipetting may ultimately result in inaccurate measurements. This situation is further exacerbated by protein aggregation, which is not uncommon in many biochemical studies. Moreover, studying intrinsically disordered proteins (IDPs) is challenging to most spectroscopic methods due to the lack of secondary or tertiary structure. IDPs account for over 30% of the human proteome and play versatile biological roles [7]. Despite being unstructured, IDPs do not sample all possible conformations stochastically. Instead, IDP conformation sampling is biased or can be shifted by phosphorylation or binding to their specific partners [8]. Detecting this elusive conformation redistribution is the key to understanding regulation of IDPs. In this study, we developed a new method that makes use of protein tyrosine-tryptophan FRET to monitor protein size changes. Although tyrosine-tryptophan FRET was observed decades ago, it has remained
Abbreviations: FRET, fluorescence resonance energy transfer; NMR, nuclear magnetic resonance; CD, circular dichroism; IDP, intrinsically disordered proteins; Nop9, nucleolar protein 9; Nop15, nucleolar protein 15; Nob1, 20S-pre-rRNA D-site endonuclease NOB1; SLBP, histone mRNA stem-loop binding protein; SRP19, signal recognition particle 19 kDa protein; DCL1, dicer-like protein 1; snRNP70, U1 small nuclear ribonucleoprotein 70 kDa; U1A, U1 small nuclear ribonucleoprotein A ∗ Corresponding author. Department of Chemistry, College of Arts and Sciences, Chemistry Building, Room 266, 901 14th Street South, Birmingham, AL 352941240, USA. E-mail address: [email protected] (J. Zhang). https://doi.org/10.1016/j.ab.2018.07.022 Received 22 April 2018; Received in revised form 19 July 2018; Accepted 23 July 2018 Available online 25 July 2018 0003-2697/ © 2018 Elsevier Inc. All rights reserved.
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
an underused method for measuring atomic distances for several reasons and the problems of using tyrosine fluorescence were summarized by Dr. Lakowicz [9,10]. One reason is that the tyrosine resonance energy is transferred to tryptophan residues, making tyrosine fluorescence invisible in most folded proteins. Another reason is the formation of tyrosinate which is the deprotonated product of the tyrosine sidechain hydroxyl group. Tyrosinate emission maximum is at 350 nm, where tryptophan fluoresces [11–13]. The final reason is that the spectral overlap between tyrosine and tryptophan further complicates the data interpretation. Herein we clarified the concern of tyrosinate and solved the problems in using tyrosine-tryptophan FRET. We used the fluorescence intensity ratio between tyrosine and tryptophan (FirbY-W) as an index of molecular size or folding status. Because tryptophan fluorescence intensity essentially serves as an internal control, FirbY-W is not affected by sample concentration. We showed that FirbY-W is extremely stable in tracking protein unfolding or binding compared to CD or tryptophan fluorescence. Moreover, our method is able to sensitively detect the elusive conformational changes of intrinsically disordered proteins, which usually escape the scrutiny of commonly used biophysical spectroscopic methods. We also demonstrated the application of our method in detecting binding events of intrinsically disordered proteins.
ln2 a − v ⎞⎤ I (v ) = Im⋅exp ⎡− 2 ⋅ln2 ⎛ ⎢ ln ρ − vm ⎠ ⎥ a ⎝ ⎣ ⎦ ⎜
⎟
(1)
where vm is the spectral maximum position; Im is the maximal amplitude; v is the wave number. ρ is the band asymmetry parameter defined by ρ= (vm − v−)/(v+ − vm) . v+ and v− are the positions of the two halfmaximal amplitudes; a is the function-limiting point position defined as ρ ⋅ (v+ − v−) a = vm + . vm , v+ and v− define the shape of the tryptophan ρ2 − 1 fluorescence spectrum and have a linear relationship. For tryptophan, the linear relationship is parameterized as:
v+ = 0.831⋅vm + 7070 (cm−1)
(2)
(cm−1)
(3)
v− = 1.177⋅vm − 7780
Currently, the analytical representation of the tyrosine fluorescence spectrum is still unknown. Finding an analytical equation will significantly simply our mathematical deconvolution process. The aforementioned log-normal function was initially proposed to describe the shape of absorption spectra of complex molecules and its mirror-symmetric form was later utilized to describe fluorescence spectra [20,21]. To test whether the log-normal function can be used to describe the tyrosine fluorescence spectrum, we parameterized Equations (2) and (3) to fit the experimental tyrosine spectra shown in Fig. 1, as the constant terms in these equations delineate the shape of the log-normal function. Using grid searching, we found that the following optimized empirical parameters to delineate the tyrosine fluorescence spectrum:
Results Tyrosinate does not form in commonly used buffers or salts The formation of tyrosinate complicates the use of tyrosine fluorescence in spectroscopic measurements due to its spectral overlap with tryptophan emission. Tyrosinate fluorescence is observed in few proteins without tryptophan, such as bovine testes calmodulin, alphapurothionin, and beta-purothionin [11–13]. Tyrosinate fluorescence is also observed at high pH (Fig. 1A and B) when the hydroxyl group of tyrosine is ionized, or in the presence of 2 M acetate [14,15]. Although acetate of this concentration is unlikely to be used in general, it is still unknown whether tyrosinate forms in commonly encountered biochemical solutions. To this end, we collected tyrosine fluorescence spectra in commonly used buffers (100 mM Citrate pH 5.5, 100 mM HEPES pH 7.0, 100 mM MES pH 6.0, 100 mM Phosphate pH 7.0, 100 mM Tris-HCl pH 7.5, 200 mM Arg/Glu pH 7.0), salts (4 M NaCl, 4 M KCl) or denaturants (8 M urea, 8 M guanidinium chloride) using Nop15 and U1A as model proteins (Fig. 1C and D). Nop1581−191 and U1A have no tryptophan, but have 7 and 3 tyrosine residues, respectively. We determined that all the spectra collected for these two proteins at different conditions are essentially identical (Fig. 1C and D). No tyrosinate fluorescence (emission maximum around 350 nm) is observed at these conditions. Our results also clearly indicate that the shape of the tyrosine fluorescence spectrum is constant in the above solutions. Taken together, we conclude that tyrosinate fluorescence is not formed under often-used conditions.
v+ = 0.831⋅vm + 7028 (cm−1)
(4)
v− = 1.177⋅vm − 8023 (cm−1)
(5)
As shown by Fig. 1, the tyrosine fluorescence peak maximum (302 nm) or its shape is not affected by buffer conditions or folding status of a protein. This suggests that Equations (1), (4) and (5) can be generalized to describe the tyrosine fluorescence spectra of other proteins. For most proteins, tyrosine fluorescence has negligible influence on the tryptophan fluorescence emission maximum (340–350 nm). As exemplified by Fig. 2A, tyrosine fluorescence intensity at 340–350 nm is only 10–20% of its emission maximum for Nop15. Considering that tyrosine fluorescence is usually 10%–30% of that of tryptophan in a FRET spectrum, perturbation of tyrosine fluorescence on tryptophan emission maximum accounts for less than 5% of tryptophan maximum and is unable to shift the tryptophan emission maximum. This suggests that the pure tryptophan emission maximum can be accurately read from FRET spectra. With the optimized parameters for tyrosine and tryptophan fluorescence spectra, the tyrosine emission maximum (302 nm), and the tryptophan emission maximum directly read from the FRET spectrum, we successfully deconvoluted the tyrosine-tryptophan fluorescence spectrum (Fig. 2B). The robustness of this deconvolution is bolstered by the negligible residual error. As expected, tryptophan fluorescence contributes to over 97% of its maximum and its contribution to the tyrosine region can be approached with reasonable accuracy by assuming tryptophan contributes 100% of the spectrum maximum (0.04042 versus 0.0416). This results in only a 0.5% error in estimation of tyrosine fluorescence intensity. A more thorough and precise deconvolution of tyrosine and tryptophan was carried out using the method of least squares (See the Method Section). However, we determined that there were no significant differences between the simplified (Equation (1)) and the more complex (least squares) methodologies.
Deconvolution of tyrosine and tryptophan fluorescence Measuring pure tyrosine or tryptophan fluorescence intensities is difficult due to tyrosine and tryptophan spectral overlap. Although tyrosine fluoresces consistently at 302 nm, tryptophan fluorescence peak maximum varies between 320 nm and 355 nm depending on solvent accessibility (Fig. 2A) [16]. More specifically, the degree of tyrosine and tryptophan spectral overlap is dependent on the blue- or redshift of tryptophan fluorescence in response to protein folding or unfolding. Therefore it is essential to deconvolute the overlapping spectra between tryptophan and tyrosine. The fluorescence spectrum of Trp can be analytically expressed by the log-normal function (Equation (1)) [17–19]:
Application of fluorescence intensity ratio between tyrosine and tryptophan (FirbY-W) to monitor unfolding event of structured proteins Tyrosine fluorescence is usually invisible in folded proteins due the 143
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Fig. 1. Tyrosinate is not induced in commonly used buffers, salts or denaturants. Normalized Tyrosine fluorescence spectra at pH 7.5 (A) and pH 11.3 (B). The fluorescence peak around 340 nm stems from tyrosinate in panel B. Normalized fluorescence spectra of Nop15 (C) and U1A (D) were collected in the presence of 100 mM Citrate pH 5.5, 100 mM HEPES pH 7.4, 100 mM MES pH 6.0, 100 mM Sodium Phosphate pH 7.0, 200 mM Arg/Glu pH 8.5, 100 mM Tris-HCl pH 7.5, 1 M NaCl, 1 M KCl, 8 M Urea or 6 M Guanidinium HCl.
and fluorophore flexibility. As shown by Fig. 3, all these proteins display distinguishable tyrosine fluorescence upon protein unfolding by urea or guanidinium. Considering the diversity of model systems selected, we believe that change in FRET efficiency upon protein unfolding is common to most tryptophan- and tyrosine-containing proteins thereby illustrating the robustness of our method. To reflect the changes in FRET efficiency, we defined a parameter termed Fluorescence Intensity Ratio Between Tyrosine and Tryptophan (FirbY-W). FirbY-W is positively correlated to protein size and therefore able to follow the protein unfolding process due to the fact that unfolded proteins generally have larger inter-residue distances. To test our
efficient resonance energy transfer from tyrosine to tryptophan. Once proteins are unfolded, however, concomitant increases in the average tyrosine-tryptophan distance decreases the resonance energy transfer efficiency. Consequently, the energy in tyrosine is retained, leading to an amplification of tyrosine fluorescence. Our method depends on this change in tyrosine-tryptophan FRET efficiency to monitor protein folding status. To confirm that the change is common to other proteins besides the few reported cases [22,23], we collected fluorescence spectra of Nob1, Nop9, Nop1581−191 Y94W and snRNP7092−202 (Fig. 3, Fig. S1). These proteins were selected because they represent different acceptor/donor (tryptophan/tyrosine) contents, spatial distribution,
Fig. 2. Deconvolution of tyrosine and tryptophan fluorescence. (A) Normalized tyrosine-tryptophan FRET spectra of Nop15 collected at different urea concentrations. The spectra were normalized by tryptophan fluorescence maximum. The horizontal and vertical arrows indicate the red-shift of tryptophan peak positions and the buildup of tyrosine fluorescence with increased urea concentration, respectively. Urea concentrations are listed aside in molar. (B) Deconvolution of the tyrosine and tryptophan fluorescence. An example spectrum of Nop15 collected in 7.0 M urea is shown. The equations to fit the tyrosine and tryptophan fluorescence spectrum are described in the text.
144
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Fig. 3. Tyrosine fluorescence is visible once unfolding of (A) Nob1 (PDB ID: 2LCQ), (B) Nop9 (PDB ID: 5SVD), (C) Nop15 (PDB ID: 5T9P) and (D) snRNP70 (PDB ID: 4PKD) as shown in the top row. Fluorescence spectra of the folded and unfolded states are shown in black and red lines, respectively. Nop15 and snRNP70 is unfolded by urea. Nob1 and Nop9 are unfolded by 6 M Guanidinium HCl. Nop1581−191 mutant Y94W was used here. Tyrosine and tryptophan residues are denoted as blue dots and red stars, respectively, along with the primary sequence (the middle row), or shown as blue and red spheres in the crystal structures (the bottom row). The disorder probabilities of the proteins were predicted using the PrDOS server (http://prdos.hgc.jp/cgi-bin/top.cgi). Only the protein regions with crystal structure available are shown in the bottom row. A homologue structure for Nob1 is used. The molecular graphics were prepared by PyMOL. Fig. 4. Protein unfolding of (A) Nop15, (B) snRNP70 and (C) Nop9 tracked by fluorescence intensity ratio between tyrosine and tryptophan (FirbY-W) (top row, ▲), tryptophan fluorescence peak position (middle row, ○) and circular dichroism (bottom row, ●). Circular dichroism signals at 222 nm were plotted. The two-state model was assumed in fitting the unfolding curves. The units for ΔG and m-value are kcal/mol, kcal/(mol M) and M, respectively.
145
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
method, we used FirbY-W to track unfolding of Nop1581−191 Y94W, snRNP7092−202 and Nop946−645 (Fig. 4, Fig. S2). Our results show that in these three cases, FirbY-W increases as a function of protein unfolding. We assumed that all transitions follow the 2-state model. In the case of Nop15, a mutant tryptophan was introduced at position Y94, which is located at a surface β-strand. The folding energy measured by FirbY-W (ΔG = 2.1±0.1 kcal/mol) is comparable to that measured by CD (1.8±0.5 kcal/mol) (Fig. 4A). We note that using FirbY-W yields a smoother transition curve and a smaller error in fitting ΔG when compared to CD. In addition, FirbY-W curves read at 298 nm and 302 nm yield a similar ΔG (1.9±0.3 kcal/mol versus 2.1±0.1 kcal/mol). The similarity in ΔG values obtained at different wavelengths is a strong indication of successful deconvolution of tyrosine and tryptophan fluorescence. In the case of snRNP7092−202, the only tryptophan at the dynamic tail is completely solvent-exposed and distant to all tyrosine located at the structured RNA recognition motif (RRM). As a result, tryptophan emission maximum fails to detect protein unfolding due to the lack of any unfolding-induced red shift (Fig. 4B, the middle row). In contrast, FirbY-W sensitively tracks the unfolding process. It is noteworthy that protein stability measured by FirbY-W is different from that measured by CD (Fig. 4B). This is possibly because the flexible region where the tryptophan is located only weakly interacts with the RRM. We also tested our method in Nop9, a system with multiple tyrosine and tryptophan residues. The FirbY-W denature curve at high guanidinium concentration suggests Nop9 folding is highly cooperative and stable. Using our FirbY-W method, we measured similar unfolding energy values for this denaturation process with a greater signal-to-noise ratio when compared to methodologies using tryptophan peak position or CD (Fig. 4C). In all cases, our FirbY-W method yields smaller error bars compared to CD and Trp emission maximum (Fig. S3, Table S1A and S1B).
phosphor-mimicry (Ser to Glu mutation) can partially mimic the RNAbinding promotion and is more compact than the non-phosphorylated SLBP [27]. Although the idea of tyrosine-tryptophan FRET was utilized in our previous study, systematic testing of the method and rigorous deconvolution of tyrosine and tryptophan fluorescence were not performed, which prevents deeper data analysis. In addition, formation of tyrosinate is particularly a concern for the SLBP system, because carboxylic groups of glutamic acid and aspartic acid, like the SLBP C-terminus, were speculated to facilitate formation of tyrosinate. Furthermore, the study was carried out on the glutamic phosphor-mimicry, which is significantly different from the authentic phosphorylation in modulating RNA binding affinity of SLBP. In this study, we first addressed the concern about tyrosinate formation in SLBP. To remove the interference of tryptophan, we mutated the only tryptophan of SLBP to phenylalanine (W249F). A tyrosinatecontaminated fluorescence spectrum has a signature shoulder around 350 nm. As shown by Fig. S4, the tyrosine fluorescence spectrum of the W249F construct has no shoulder around 350 nm. This result clear showed that tyrosinate formation is not induced by the acidic patch, reinforcing our claim that tyrosinate formation requires certain special protein tertiary structure. As shown by mass spectrometry (Fig. S5), we have also successfully obtained hyper-phosphorylated SLBP, which provides us an opportunity to analyze phosphorylated SLBP directly using FirbY-W to obtain a deeper insight. The minimal functional motif of SLBP contains six tyrosine residues and one tryptophan located at the hydrophobic core in the RNA-bound state. We collected and compared the fluorescence spectra of SLBP in different phosphorylated states. Despite the dramatic difference in RNA-binding affinities, circular dichroism fails to detect any secondary structure difference between phosphorylated and non-phosphorylated SLBP (Fig. 5C). A blue shift is observed in tryptophan fluorescence emission maximum upon phosphorylation, indicating the increased extent of tryptophan bury and residual structure (Fig. 5D). However, denature curves of the residual structure monitored by tryptophan emission maximum have significant noise and a narrow dynamic range for non-phosphorylated SLBP (Fig. 5D). In contrast, FirbY-W analysis yields denature curves with a much lower noise level (Fig. 5E and F). In addition, FirbY-W values of phosphorylated SLBP in general is smaller than unphosphorylated protein, suggesting that phosphorylation may pre-fold SLBP into the conformation that is similar to the RNA-bound state. Indeed, we found that phosphorylated SLBP is more tolerant to urea denaturation. This finding at least partially explains the dramatic RNA-binding enhancement observed before. It is noteworthy that this finding is also in line with the blue-shift of tryptophan fluorescence emission maximum induced by phosphorylation, suggesting that the protein becomes more compact upon phosphorylation (Fig. 5D). However, the transition curve of tryptophan emission maximum is very noisy.
Application of FirbY-W in studying conformational redistribution of intrinsically disordered proteins Despite lack of definite secondary or tertiary structure, intrinsically disordered proteins play various indispensable functions [7]. The conformation distribution in the IDP ensemble is very sensitive to external signaling and important for IDP functions. Although intrinsically disordered regions account for more than 30% of the human proteome, our understanding of IDPs is still limited due to the difficulty in characterizing their physical chemical behavior. Most traditional biophysical spectroscopic methods are insensitive to IDP ensemble redistribution since secondary or tertiary structure is absent in the conformational ensemble. FRET has been used to describe IDP behavior, but introduction of exogenous FRET donor/acceptor pair is usually needed [24]. To test whether our method is applicable to IDPs, the following three disordered proteins were selected: signal recognition particle (SRP19), dicer-like protein 1 (DCL1) and histone mRNA stemloop binding protein (SLBP). As shown by Fig. 5A and E, tyrosine fluorescence shoulder is clearly visible in the three disordered proteins tested here. Visibility of tyrosine fluorescence in IDPs is distinct from what we observed in folded proteins (Fig. 3). We further found that DCL1 and SRP19 have no FirbY-W change at different denaturant concentrations, suggesting the proteins have no residual structure, consistent with previous results [25,26]. Residual structural formation is not absent in every IDP conformational ensemble, nor all possible conformations sampled equally in the ensemble. One pertinent example is the histone mRNA stem-loop binding protein (SLBP). SLBP is essential for histone mRNA processing in response to cell duplication. The minimal functional motif of SLBP (residue 190–276) contains an RNA-binding domain and a C-terminus bearing four serine residues (SNSDSDSD) [27]. Hyper-phosphorylation of the terminus is required for SLBP function and dramatically increases RNA binding affinity [27]. Our previous study also showed that
Application of FirbY-W in studying ligand binding of intrinsically disordered proteins Our previous study shows that RNA binding of SLBP induces protein folding [27]. SLBP190−276 contains a single tryptophan residue at the protein hydrophobic core, surrounded by 6 tyrosine residues. In the folded state induced by RNA binding, the average tyrosine-tryptophan distance is 11.4 Å (Fig. 6A). The protein folding increases FRET efficiency by bringing the tyrosine and tryptophan residues closer. To test whether FirbY-W will sense the RNA binding event, we collected fluorescence spectra of SLBP with and without RNA. As shown in Fig. 6B, upon RNA binding, the tyrosine fluorescence shoulder peak disappears. The binding event reduces FirbY-W by roughly 50%. In the case of SLBP/RNA binding, the tryptophan blue shift can also be used to detect RNA binding. However, this is not true for the tryptophan residues that do not experience changes in micro-environments during ligand binding. For example, RNA binding does not change the 146
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Fig. 5. FirbY-W studying protein size changes of intrinsically disordered proteins. Fluorescence spectra (A) and FirbY-W (B) of disordered DCL1and SRP19. (C) Circular dichroism profiles of phosphorylated SLBP and non-phosphorylated SLBP. (D) Protein unfolding curves monitored by tryptophan fluorescence emission maximum. m-values are 1.1 ± 0.5 kcal/ (mol M) and 1.5 ± 0.9 kcal/(mol M) for phosphorylated and non-phosphorylated SLBP, respectively. (E) Fluorescence spectra of phosphorylated and non-phosphorylated SLBP. (F) Unfolding profile of phosphorylated SLBP and non-phosphorylate SLBP monitored by FirbY-W. For clarity, only the fluorescence spectra at 0 M of urea are shown. The unit for ΔG is kcal/mol. m-values for phosphorylated SLBP and non-phosphorylated SLBP are 1.0 ± 0.2 kcal/ (mol M) and 1.3 ± 0.4 kcal/(mol M), respectively.
tryptophan emission maximum of snRNP70 (Fig. 6D). As suggested by deconvolution of the Trp and Tyr fluorescence spectra, the Trp emission maximum is at 350 nm in the free and RNA-bound states (Fig. S6). However, FirbY-W can still sense the RNA binding as shown by Fig. 6D and Fig. S6. FirbY-W changes from 0.53 to 1.29 upon RNA binding. It is noteworthy that RNA binding decreases FirbY-W of snRNP70, in contrast to the FirbY-W increase upon SLBP/RNA binding. This opposite effect on snRNP70 FirbY-W can be interpreted by a closer structural analysis (Fig. 6C). The only tryptophan of snRNP70 located at the disordered C-terminus has transient opportunities to approach the six tyrosine residues, which allows tyrosine-tryptophan FRET. However, upon RNA binding the disordered C-terminus becomes ordered and positions the tryptophan residues away from the tyrosine residues, with an average distance of 20.4 Å.
Discussion Despite the prevalence of tyrosine-tryptophan FRET in proteins, it is underused in probing protein dimensions, due to the concerns about tyrosinate fluorescence, the invisibility of tyrosine fluorescence in folded proteins, and spectral overlap of tyrosine and tryptophan fluorescence. Here we show that tyrosinate fluorescence is not observed in commonly used buffers, salts or denaturants. Tyrosinate fluorescence was reported at high concentration of acetate and high pH (> 10) [15] which are rarely used in biochemistry studies. Under physiological pH and salt concentrations, tyrosinate fluorescence is only observed in few proteins, such as bovine testes calmodulin, alpha-purothionin and betapurothionin [12,13]. Interestingly the tyrosinate fluorescence disappears once these proteins are unfolded [12]. Therefore it is likely that
Fig. 6. Application of FirbY-W in studying protein/RNA binding. Tyrosine and tryptophan distribution in RNA complex of SLBP (PDB ID: 4TV0) (A) and snRNP70 (PDB ID: 4PKD) (C). The graphic were prepared using PyMOL. Blue spheres denote the C-alpha atoms of tyrosine residues and red ones for tryptophan residues. RNA is shown in orange cartoon and protein in green cartoon. Normalized fluorescence spectra of SLBP and snRNP70 are shown in panel B and D, respectively.
147
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
screening method for protein/RNA interaction, it is not suitable for quantitative determination of RNA binding affinity, as the sensitivity limit of Tyr-Trp FRET is around μM, which is much higher than dissociation constants (Kd) of many protein/RNA binding events. FirbY-W also has the potential to probe protein:protein interaction if the binding event incurs change of FRET between Tyr and Trp residues. In this situation, the fluorescence spectrum of the complex will not be a linear combination of each individual binder's spectrum.
the specific tertiary structure of these proteins positions tyrosine in a special environment that facilitates tyrosinate formation. Tyrosinate fluorescence observed in these samples should not be a concern for other protein studies. In this study we used FirbY-W as an index to reflect the overall dimension of proteins. On one hand, FirbY-W reflects the average protein size; on the other, FirbY-W eliminates data noise caused by sample concentration deviations, which usually stems from pipetting error or protein aggregation. For example, we found that Nop9 has slight protein aggregation at low concentration of guanidinium, which results in vibration of CD signals (Fig. 4). However, FirbY-W is tolerant to the slight protein aggregation problem. Since tyrosine and tryptophan fluorescence is collected in the same spectrum scan, FirbY-W is also tolerant to fluorimeter instability. For proteins with tryptophan residues completely exposed to solvent as exemplified by snRNP70, it is impossible to use tryptophan fluorescence to track protein unfolding. However, FirbY-W can still sense protein unfolding because FirbY-W directly detects global size changes of proteins instead of local microenvironments of tryptophan residues. Even Trp emission maximum can be used to track protein unfolding, as in Nop15 and Nop9, it suffers from a narrow dynamic range from 340 nm to 355 nm. This narrow dynamic range is vulnerable to fluorescence intensity vibration which frequently happens around emission maximum, to the low spectrum resolution which is at the nanometer scale. These two factors cause instability of unfolding curves tracked by Trp emission maximum. In addition, our method also provides information on tryptophan fluorescence emission maximum. For these reasons, FirbY-W is superior to CD or fluorescence emission maximum in protein unfolding studies. The reported Foster distances for tyrosine/tryptophan FRET range from 9 Å to 18 Å, the dimension of most protein domains [9,28–31]. This means that FirbY-W is sensitive to folding/unfolding events of most proteins. One exception would be in proteins where all tyrosine residues are adjacent to all tryptophan residues in the primary sequence. However, as exemplified by Nop15 and DCL1 in our study, mutant Trp can be introduced for measurement of FirbY-W. Although it is still difficult to predict whether FirbY-W is applicable for a protein, experimental testing can be easily performed by comparing the FRET spectra with and without denaturant as shown by Fig. 3. Currently it is still challenging to use FirbY-W to calculate donor-acceptor distance due to the fact that most proteins contain more than one tyrosine and/or tryptophan residue, but application of FirbY-W to monitor protein folding/ unfolding can be generalized to most protein systems. The unique power of our method is its ability to track behavior of intrinsically disordered proteins. Monitoring dynamic behavior of IDPs is challenging and is only feasible for a few experimental methods such as NMR and FRET. FRET study of IDPs usually necessitates the introduction of exogenous fluorophores. The lack of secondary structure and tertiary structure in IDPs makes canonical biophysical methods, such as CD or tryptophan fluorescence, problematic. Using model disordered proteins DCL1 and SRP19, we showed that our method is able to determine whether a protein is unstructured based on its response to urea concentration change. Moreover, application of FirbY-W to SLBP shows that our method is capable of detecting the subtle phosphorylation-triggered conformation redistribution that escapes the detection of circular dichroism. This is probably because phosphorylation does not affect the secondary structure content of SLBP, but reduces the charge repulsion and promotes protein folding. In addition, we also show that FirbY-W responds to RNA binding of SLBP and snRNP70. Considering that binding-induced folding is common to many IDPs and the process will change IDP size, FirbY-W can be applied to study other IDP-involved binding processes [32]. Although RNA has no fluorescence from 280 nm to 600 nm, it is noteworthy that UV absorbance of RNA around 260–280 nm can significantly weaken the incident light for fluorescence excitation and consequently decreases available energy for FRET. At high concentrations (> 1 Au), RNA can dramatically decrease signal of FRET spectra. Although FirbY-W can be a convenient
Experimental procedures Protein purification Wild type Nop1581−191, Nop1581−191 Y94W, SLBP190−276, Nop946−645 and Nob1 were purified as reported [27,33,34]. SLBP was in vitro phosphorylated by casein kinase II (New England Biolabs) at room temperature in 50 mM Tris-HCl, 10 mM MgCl2, 2 mM DTT and 1 mM ATP. Casein kinase II specifically phosphorylates S269, S271, S273 and S275. DCL1, snRNP7092−202 and SRP19 were purified using the following procedure. DCL1 encoding DNA was synthesized by Genscript and a tryptophan residue was inserted at the protein C-terminus. DNA encoding snRNP7092−202 and SRP19 were ordered from Addgene (https://www.addgene.org/) and Harvard PlasmID Database (https://plasmid.med.harvard.edu/PLASMID/Home.xhtml), respectively. All these protein-encoding DNA were cloned into pSMT3 (Memorial Sloan Kettering Cancer Center). All protein sequences of the constructs used in this study are shown in Fig. S1. Transformed E. coli BL21-CodonPlus (DE3) cells were grown in LB media at 37 °C. 0.5 mM IPTG was added when OD600 reached 0.6. The culture was continued at 22 °C overnight. Cell pellets were re-suspended in 25 mM HEPES, pH 7.5, 1 M NaCl, 0.2 mM TCEP, 25 mM imidazole, 1 mM PMSF, 1 mg/ml lysozyme and lysed by sonication, followed by centrifugation (14,000 rpm for 40 min) to remove cell debris. The supernatant was applied to 5 ml HisPur Ni-NTA resin (Thermo Scientific), washed with 200 ml 25 mM HEPES, pH 7.5, 1 M NaCl, 0.2 mM TCEP, 25 mM imidazole, and eluted with 25 mM HEPES, pH 7.5, 500 mM NaCl, 1 mM TCEP, 250 mM imidazole. The N-terminal SUMO-tagged protein was cleaved overnight with 1 μg/ml of Ulp1 at 4 °C. The cleaved sample was diluted 5 fold using 20 mM HEPES, pH 7.5, 20 mM NaCl, 1 mM TCEP and loaded onto a 5-ml HiTrap Heparin column (GE Healthcare). The sample was eluted with a linear gradient from 0 to 2 M NaCl in 20 mM HEPES, 0.2 mM TCEP. DCL1 was eluted with 400 mM NaCl; snRNP70 and SRP19 were eluted with 860 mM NaCl. The fractions containing target proteins were pooled and further purified using a HiLoad 16/60 Superdex75 column (GE Healthcare) equilibrated with 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.2 mM TCEP. The identities of the proteins were confirmed by mass spectrometry, and the purities were > 95% based on SDS-PAGE. Preparation of RNA sample Histone mRNA stem-loop RNA for SLBP is a product of Dharmacon, Inc. (5′-GGC CAA AGG CCC UUU UCA GGG CCA CCC A-3′). The stemloop RNA for snRNP70 (GGG AGA UAC CAU GAU CAC GAA GGU GGU UUU CCC) was prepared by in vitro T7 transcription, which is carried out at 37 °C for 8 h in 100 mM Tris-HCl, pH 8.5, 20 mM MgCl2, 3 mM TCEP, 2 mM spermidine, 3% (w/v) PEG 8000, 0.01% (v/v) Triton X100, 4 mM NTPs, 2 units of inorganic pyrophosphatase, 0.6 μM doublestranded DNA template, and 0.06 mg/ml T7 RNA polymerase. Transcription product was purified by 15% polyacrylamide gel (30 cm × 40 cm x 1.6 mm) in the presence of 7 M urea and 1X TBE. The gels were pre-run for 30 min to a temperature of 50 °C, and RNA samples were resolved under a constant power of 75 W for 6 h, during which RNAs travel about two thirds of the gel. The excised gel with target RNA was electronically eluted overnight in 10 mM Tris-HCl, pH 8, and 1 mM EDTA. RNAs were incubated at 90 °C for 2 min and snap148
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
CD data collection
cooled on ice to refold the RNA. The homogeneity of refolded RNA was confirmed by single bands on 10% polyacrylamide native TBE gels. 10 μM proteins were mixed with binding RNA into 1:1 ratio.
CD data from 260 nm to 185 nm were collected using a Jasco J-815 circular dichroism spectropolarimeter at 295 K using a 1-mm cuvette. All protein samples (200 μL, ∼30 μM) were dissolved into 20 mM TrisHCl, pH 7.5, 100 mM NaCl, 0.1 mM TCEP. Corresponding CD signal for buffers were collected and subtracted. CD signals below 200 nm were seriously interfered by high concentration of urea or Guanidinium HCl. Therefore, CD signal at 222 nm were used to track the protein folding event.
Fluorescence data collection Fluorescence data were collected using a Varian Cary Eclipse fluorometer at 295 K with a 5-mm cuvette. All protein samples (600 μL, ∼20 μM) were exchanged into 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1 mM TCEP before urea or Guanidinium HCl was added. The samples were centrifuged at 10,000 g for 10 min at 4 °C before data collection. Excitation wavelength was set to 275 nm and spectra from 280 nm to 400 nm were collected with 5-nm excitation and 5-nm emission slits. The same settings were used to collect buffer fluorescence as reference. After subtraction of background fluorescence, the data were smoothed using a 20-point Savitzky-Golay function and then normalized by tryptophan fluorescence maximum.
Fitting of protein unfolding by the two-state model The two-state model was assumed in fitting the unfolding process of the proteins versus denaturant concentrations. Nonlinear regression was carried out using the following equation:
( exp (
D⋅exp
Deconvolution of tyrosine and tryptophan fluorescence
E (x ) =
For protein systems with more than one tryptophan residue, it is possible that all three classes of tryptophan are present in the sample. For simplicity and clarity, assume tyrosine fluorescence and only one tryptophan fluorescence component are present in the sample. It is convenient to extend the following calculation into the case in which other tryptophan fluorescence components are present. The theoretical spectrum at discrete wavenumber vi can be written as:
)+N )+1
mx − ΔG RT
mx − ΔG RT
(6)
where E is fluorescence or CD observables; x is denaturant concentration; m is the so-called m-value for urea or Guanidinium-HCl; N and D are the signal baselines at the native and denatured states, respectively; ΔG is the unfolding energy of proteins; R is the gas constant and T is the temperature (295 K) at which experiments were carried out. A tilted baseline was observed for the unfolding processes with Guanidinium HCl and a modified two-state equation was used as the following:
I (vi ) = ImY ⋅fY (vi ) + ImW ⋅fW (vi )
( exp (
(ax + D)⋅exp
where ImY and ImW are the maximum peak intensities of tyrosine and tryptophan, respectively; fY and fW are spectrum shape defined by Equation (1) and Equation (2). The deviation between experimental data E (vi ) and theoretical data are expressed as:
E (x ) =
) + (ax + N ) )+1
mx − ΔG RT
mx − ΔG RT
(7)
where N is the total number of discrete data points of the spectrum. The minimal value of Q or the optimal deconvolution is found when the following relationship is satisfied:
where a is the slope of the baseline; N and D are the intercepts on the Xaxis at the native and denatured states, respectively, with other parameters being the same as Equation (6). For the CD method, experimental observables, E(x), are read from 222 nm. For the Trp emission maximum method, experimental observables are the emission positions. For the FirbY-W method, the ratio of the deconvoluted Tyr and Trp fluorescence intensity is used to fit Eq (7). All data fitting were performed using Graphpad Prism.
∂Q =0 ∂Im
Conflicts of interest
By carrying out partial derivation with respect to all spectrum components (ImY , ImW …), the group of equations obtained can be written as:
The authors declare that they have no conflicts of interest with the contents of this article.
N
Q=
∑ (E (vi) − I (vi))2 i
A⋅Im = Y Author contributions
where
JZ designed research; KBD, ZZ, EAK and JZ performed research; JZ and KBD analyzed data; and JZ wrote the paper.
I Im = ⎜⎛ mY ⎞⎟ I ⎝ mW ⎠ N
⎛ ∑ 1 E (vi )⋅fY (vi ) ⎞ Y = ⎜ Ni = ⎟ ∑ E (vi )⋅fW (vi ) ⎝ i=1 ⎠ N
Funding This work was supported in whole by the start-up funding of Chemistry Department, College of Arts and Sciences, University of Alabama at Birmingham.
N
⎛ ∑ fY (vi )⋅fY (vi ) ∑i = 1 fW (vi )⋅fY (vi ) ⎞ A = ⎜ Ni = 1 ⎟ N ∑ f (v )⋅f (v ) ∑i = 1 fW (vi )⋅fW (vi ) ⎝ i=1 Y i W i ⎠ Once an inverse matrix of A , A−1 is obtained, Im can be determined as the following:
Acknowledgments We thank Michael J. Jablonsky for technical support, Dr. Chad Petit for stimulating comments and discussion. This work was suppoDrted by faculty start-up funding from the Chemistry Department, College of Arts and Sciences, UAB.
Im = A−1 Y The fitting was carried out using a Python program, which is available upon request. 149
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Appendix A. Supplementary data
Biophys. J. 81 (3) (2001) 1699–1709. [18] Y.K. Reshetnyak, E.A. Burstein, Decomposition of protein tryptophan fluorescence spectra into log-normal components. II. The statistical proof of discreteness of tryptophan classes in proteins, Biophys. J. 81 (3) (2001) 1710–1734. [19] Y.K. Reshetnyak, Y. Koshevnik, E.A. Burstein, Decomposition of protein tryptophan fluorescence spectra into log-normal components. III. Correlation between fluorescence and microenvironment parameters of individual tryptophan residues, Biophys. J. 81 (3) (2001) 1735–1758. [20] D.B. Siano, D.E. Metzler, Band shapes of the electronic spectra of complex molecules, J. Chem. Phys. 51 (1969) 106. [21] E.A. Burstein, Log-normal description of fluorescence spectra of organic fluorophores, Photochem. Photobiol. 64 (1996) 5. [22] R. Boteva, et al., Dissociation equilibrium of human recombinant interferon gamma, Biochemistry 35 (47) (1996) 14825–14830. [23] M. Eisenhawer, et al., Fluorescence resonance energy transfer shows a close helixhelix distance in the transmembrane M13 procoat protein, Biochemistry 40 (41) (2001) 12321–12328. [24] S. Muller-Spath, et al., From the Cover: charge interactions can dominate the dimensions of intrinsically disordered proteins, Proc. Natl. Acad. Sci. U.S.A. 107 (33) (2010) 14609–14614. [25] I.P. Suarez, et al., Induced folding in RNA recognition by Arabidopsis thaliana DCL1, Nucleic Acids Res. 43 (13) (2015) 6607–6619. [26] T.S. Maity, K.M. Weeks, A threefold RNA-protein interface in the signal recognition particle gates native complex assembly, J. Mol. Biol. 369 (2) (2007) 512–524. [27] J. Zhang, et al., Molecular mechanisms for the regulation of histone mRNA stemloop-binding protein by phosphorylation, Proc. Natl. Acad. Sci. U.S.A. 111 (29) (2014) E2937–E2946. [28] J. Eisinger, Intramolecular energy transfer in adrenocorticotropin, Biochem. 8 (10) (1969) 3902–3908. [29] P.W. Schiller, Fluorescence study on the conformation of a cyclic enkephalin analog in aqueous solution, Biochem. Biophys. Res. Commun. 114 (1) (1983) 268–274. [30] S.F. Pearce, E. Hawrot, Intrinsic fluorescence of binding-site fragments of the nicotinic acetylcholine receptor: perturbations produced upon binding alpha-bungarotoxin, Biochem. 29 (47) (1990) 10649–10659. [31] M.J. Moreno, M. Prieto, Interaction of the peptide hormone adrenocorticotropin, ACTH(1-24), with a membrane model system: a fluorescence study, Photochem. Photobiol. 57 (3) (1993) 431–437. [32] P.E. Wright, H.J. Dyson, Linking folding and binding, Curr. Opin. Struct. Biol. 19 (1) (2009) 31–38. [33] J. Zhang, L.E. Gonzalez, T.M.T. Hall, Structural analysis reveals the flexible C-terminus of Nop15 undergoes rearrangement to recognize a pre-ribosomal RNA folding intermediate, Nucleic Acids Res. 45 (5) (2017) 2829–2837. [34] J. Zhang, et al., Nop9 is a PUF-like protein that prevents premature cleavage to correctly process pre-18S rRNA, Nat. Commun. 7 (2016) 13085.
Supplementary data related to this article can be found at https:// doi.org/10.1016/j.ab.2018.07.022. References [1] B.T. Bajar, et al., A guide to fluorescent protein FRET pairs, Sensors 16 (9) (2016). [2] S.M. Kelly, T.J. Jess, N.C. Price, How to study proteins by circular dichroism, Biochim. Biophys. Acta Protein Proteonomics 1751 (2) (2005) 119–139. [3] J.T. Vivian, P.R. Callis, Mechanisms of tryptophan fluorescence shifts in proteins, Biophys. J. 80 (5) (2001) 2093–2109. [4] P. Pattanaik, et al., Unusual fluorescence of W168 in Plasmodium falciparum triosephosphate isomerase, probed by single-tryptophan mutants, Eur. J. Biochem. 270 (4) (2003) 745–756. [5] C.R. Middaugh, et al., Infrared spectroscopy, Meth. Mol. Biol. 40 (1995) 137–156. [6] J. Zhang, Y.B. Yan, Probing conformational changes of proteins by quantitative second-derivative infrared spectroscopy, Anal. Biochem. 340 (1) (2005) 89–98. [7] B. Xue, A.K. Dunker, V.N. Uversky, Orderly order in protein intrinsic disorder distribution: disorder in 3500 proteomes from viruses and the three domains of life, J. Biomol. Struct. Dynam. 30 (2) (2012) 137–149. [8] V.N. Uversky, A Decade and a Half of Protein Intrinsic Disorder: Biology Still Waits for Physics, Protein science: a publication of the Protein Society, 2013, pp. 693–724 22(6). [9] G. Weber, Fluorescence-polarization spectrum and electronic-energy transfer in tyrosine, tryptophan and related compounds, Biochem. J. 75 (1960) 335–345. [10] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, (2006) (Chapter 16). [11] A.G. Szabo, et al., Tyrosinate fluorescence maxima at 345 nm in proteins lacking tryptophan at pH 7, FEBS Lett. 94 (2) (1978) 249–252. [12] F.G. Prendergast, P.D. Hampton, B. Jones, Characteristics of tyrosinate fluorescence emission in alpha- and beta-purothionins, Biochemistry 23 (26) (1984) 6690–6697. [13] S. Pundak, R.S. Roche, Tyrosine and tyrosinate fluorescence of bovine testes calmodulin: calcium and pH dependence, Biochemistry 23 (7) (1984) 1549–1555. [14] D.M. Rayner, D.T. Krajcarski, A.G. Szabo, Excited state acid–base equilibrium of tyrosine, Can. J. Chem. 56 (9) (1978) 1238–1245. [15] J.B.A.L.,W.R. Ross, K.W. Rousslang, H.R. Wyssbrod, Tyrosine fluorescence and phosphorescence from proteins and polypeptides. In Topics in fluorescence spectroscopy, Biochem. Appl. 3 (1992) 63. [16] M.R. Eftink, Fluorescence techniques for studying protein structure, Meth. Biochem. Anal. 35 (1990) 13. [17] E.A. Burstein, S.M. Abornev, Y.K. Reshetnyak, Decomposition of protein tryptophan fluorescence spectra into log-normal components. I. Decomposition algorithms,
150
Contents lists available at ScienceDirect
Analytical Biochemistry journal homepage: www.elsevier.com/locate/yabio
Application of tyrosine-tryptophan fluorescence resonance energy transfer in monitoring protein size changes
T
Kenneth B. Davis, Zihan Zhang, Elizaveta A. Karpova, Jun Zhang∗ Department of Chemistry, College of Arts and Sciences, University of Alabama at Birmingham, CH266, 901 14th Street South, Birmingham, AL, 35294-1240, USA
A R T I C LE I N FO
A B S T R A C T
Keywords: FRET Protein unfolding Intrinsically disordered proteins Fluorescence Ligand binding FirbY-W
Monitoring protein size changes has versatile applications in studying protein folding/unfolding, conformational rearrangements, and ligand binding. Traditionally, FRET has been used to obtain this information. However, the use of FRET often requires covalent attachment of exogenous fluorophores. Although intrinsic FRET also exists between tyrosine and tryptophan residues, it has been underused because of tyrosinate formation and spectroscopic overlap. Herein, we clarified the concern of tyrosinate formation and mathematically deconvoluted tyrosine/tryptophan fluorescence spectra. We define a new parameter called FirbY-W (fluorescence intensity ratio between tyrosine and tryptophan) to reflect protein sizes. We demonstrate its applications in studying protein unfolding using several model proteins. In all the cases, our method offers superior sensitivity, data quality, and robustness compared with traditional techniques. The unique power of our method is in its ability to detect elusive conformational changes of intrinsically disordered proteins (IDP). The lack of structure makes IDPs unsuitable for CD or tryptophan fluorescence characterization. Using histone mRNA stem-loop binding protein (SLBP) as an example of disordered proteins, we showed that our method is capable of detecting conformational changes caused by phosphorylation, which are effectively invisible for traditional spectroscopic methods. Our method can also be used to detect RNA binding of disordered proteins.
Introduction Many biochemical questions, such as protein folding, ligand binding or conformational changes, can be ultimately simplified as measuring intra- and inter-atomic distances. One approach to measuring the desired distances is fluorescence resonance energy transfer (FRET). However, application of FRET to protein systems usually requires introducing exogenous donor/acceptor pairs [1]. This introduction is cumbersome and may introduce unnecessary artifacts into the system. Therefore, circular dichroism (CD), infrared spectroscopy, or tryptophan fluorescence are often used, each with shortcomings [2–6]. Specifically, CD and infrared are mainly sensitive to protein secondary structure but not tertiary structure. Tryptophan fluorescence intensity and emission maximum can be useful, assuming that folding or binding events perturb tryptophan microenvironments. The validity of this assumption, however, depends on the location of tryptophan residues. In
addition, due to the fact that spectroscopic signals are proportional to sample concentration, small errors in pipetting may ultimately result in inaccurate measurements. This situation is further exacerbated by protein aggregation, which is not uncommon in many biochemical studies. Moreover, studying intrinsically disordered proteins (IDPs) is challenging to most spectroscopic methods due to the lack of secondary or tertiary structure. IDPs account for over 30% of the human proteome and play versatile biological roles [7]. Despite being unstructured, IDPs do not sample all possible conformations stochastically. Instead, IDP conformation sampling is biased or can be shifted by phosphorylation or binding to their specific partners [8]. Detecting this elusive conformation redistribution is the key to understanding regulation of IDPs. In this study, we developed a new method that makes use of protein tyrosine-tryptophan FRET to monitor protein size changes. Although tyrosine-tryptophan FRET was observed decades ago, it has remained
Abbreviations: FRET, fluorescence resonance energy transfer; NMR, nuclear magnetic resonance; CD, circular dichroism; IDP, intrinsically disordered proteins; Nop9, nucleolar protein 9; Nop15, nucleolar protein 15; Nob1, 20S-pre-rRNA D-site endonuclease NOB1; SLBP, histone mRNA stem-loop binding protein; SRP19, signal recognition particle 19 kDa protein; DCL1, dicer-like protein 1; snRNP70, U1 small nuclear ribonucleoprotein 70 kDa; U1A, U1 small nuclear ribonucleoprotein A ∗ Corresponding author. Department of Chemistry, College of Arts and Sciences, Chemistry Building, Room 266, 901 14th Street South, Birmingham, AL 352941240, USA. E-mail address: [email protected] (J. Zhang). https://doi.org/10.1016/j.ab.2018.07.022 Received 22 April 2018; Received in revised form 19 July 2018; Accepted 23 July 2018 Available online 25 July 2018 0003-2697/ © 2018 Elsevier Inc. All rights reserved.
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
an underused method for measuring atomic distances for several reasons and the problems of using tyrosine fluorescence were summarized by Dr. Lakowicz [9,10]. One reason is that the tyrosine resonance energy is transferred to tryptophan residues, making tyrosine fluorescence invisible in most folded proteins. Another reason is the formation of tyrosinate which is the deprotonated product of the tyrosine sidechain hydroxyl group. Tyrosinate emission maximum is at 350 nm, where tryptophan fluoresces [11–13]. The final reason is that the spectral overlap between tyrosine and tryptophan further complicates the data interpretation. Herein we clarified the concern of tyrosinate and solved the problems in using tyrosine-tryptophan FRET. We used the fluorescence intensity ratio between tyrosine and tryptophan (FirbY-W) as an index of molecular size or folding status. Because tryptophan fluorescence intensity essentially serves as an internal control, FirbY-W is not affected by sample concentration. We showed that FirbY-W is extremely stable in tracking protein unfolding or binding compared to CD or tryptophan fluorescence. Moreover, our method is able to sensitively detect the elusive conformational changes of intrinsically disordered proteins, which usually escape the scrutiny of commonly used biophysical spectroscopic methods. We also demonstrated the application of our method in detecting binding events of intrinsically disordered proteins.
ln2 a − v ⎞⎤ I (v ) = Im⋅exp ⎡− 2 ⋅ln2 ⎛ ⎢ ln ρ − vm ⎠ ⎥ a ⎝ ⎣ ⎦ ⎜
⎟
(1)
where vm is the spectral maximum position; Im is the maximal amplitude; v is the wave number. ρ is the band asymmetry parameter defined by ρ= (vm − v−)/(v+ − vm) . v+ and v− are the positions of the two halfmaximal amplitudes; a is the function-limiting point position defined as ρ ⋅ (v+ − v−) a = vm + . vm , v+ and v− define the shape of the tryptophan ρ2 − 1 fluorescence spectrum and have a linear relationship. For tryptophan, the linear relationship is parameterized as:
v+ = 0.831⋅vm + 7070 (cm−1)
(2)
(cm−1)
(3)
v− = 1.177⋅vm − 7780
Currently, the analytical representation of the tyrosine fluorescence spectrum is still unknown. Finding an analytical equation will significantly simply our mathematical deconvolution process. The aforementioned log-normal function was initially proposed to describe the shape of absorption spectra of complex molecules and its mirror-symmetric form was later utilized to describe fluorescence spectra [20,21]. To test whether the log-normal function can be used to describe the tyrosine fluorescence spectrum, we parameterized Equations (2) and (3) to fit the experimental tyrosine spectra shown in Fig. 1, as the constant terms in these equations delineate the shape of the log-normal function. Using grid searching, we found that the following optimized empirical parameters to delineate the tyrosine fluorescence spectrum:
Results Tyrosinate does not form in commonly used buffers or salts The formation of tyrosinate complicates the use of tyrosine fluorescence in spectroscopic measurements due to its spectral overlap with tryptophan emission. Tyrosinate fluorescence is observed in few proteins without tryptophan, such as bovine testes calmodulin, alphapurothionin, and beta-purothionin [11–13]. Tyrosinate fluorescence is also observed at high pH (Fig. 1A and B) when the hydroxyl group of tyrosine is ionized, or in the presence of 2 M acetate [14,15]. Although acetate of this concentration is unlikely to be used in general, it is still unknown whether tyrosinate forms in commonly encountered biochemical solutions. To this end, we collected tyrosine fluorescence spectra in commonly used buffers (100 mM Citrate pH 5.5, 100 mM HEPES pH 7.0, 100 mM MES pH 6.0, 100 mM Phosphate pH 7.0, 100 mM Tris-HCl pH 7.5, 200 mM Arg/Glu pH 7.0), salts (4 M NaCl, 4 M KCl) or denaturants (8 M urea, 8 M guanidinium chloride) using Nop15 and U1A as model proteins (Fig. 1C and D). Nop1581−191 and U1A have no tryptophan, but have 7 and 3 tyrosine residues, respectively. We determined that all the spectra collected for these two proteins at different conditions are essentially identical (Fig. 1C and D). No tyrosinate fluorescence (emission maximum around 350 nm) is observed at these conditions. Our results also clearly indicate that the shape of the tyrosine fluorescence spectrum is constant in the above solutions. Taken together, we conclude that tyrosinate fluorescence is not formed under often-used conditions.
v+ = 0.831⋅vm + 7028 (cm−1)
(4)
v− = 1.177⋅vm − 8023 (cm−1)
(5)
As shown by Fig. 1, the tyrosine fluorescence peak maximum (302 nm) or its shape is not affected by buffer conditions or folding status of a protein. This suggests that Equations (1), (4) and (5) can be generalized to describe the tyrosine fluorescence spectra of other proteins. For most proteins, tyrosine fluorescence has negligible influence on the tryptophan fluorescence emission maximum (340–350 nm). As exemplified by Fig. 2A, tyrosine fluorescence intensity at 340–350 nm is only 10–20% of its emission maximum for Nop15. Considering that tyrosine fluorescence is usually 10%–30% of that of tryptophan in a FRET spectrum, perturbation of tyrosine fluorescence on tryptophan emission maximum accounts for less than 5% of tryptophan maximum and is unable to shift the tryptophan emission maximum. This suggests that the pure tryptophan emission maximum can be accurately read from FRET spectra. With the optimized parameters for tyrosine and tryptophan fluorescence spectra, the tyrosine emission maximum (302 nm), and the tryptophan emission maximum directly read from the FRET spectrum, we successfully deconvoluted the tyrosine-tryptophan fluorescence spectrum (Fig. 2B). The robustness of this deconvolution is bolstered by the negligible residual error. As expected, tryptophan fluorescence contributes to over 97% of its maximum and its contribution to the tyrosine region can be approached with reasonable accuracy by assuming tryptophan contributes 100% of the spectrum maximum (0.04042 versus 0.0416). This results in only a 0.5% error in estimation of tyrosine fluorescence intensity. A more thorough and precise deconvolution of tyrosine and tryptophan was carried out using the method of least squares (See the Method Section). However, we determined that there were no significant differences between the simplified (Equation (1)) and the more complex (least squares) methodologies.
Deconvolution of tyrosine and tryptophan fluorescence Measuring pure tyrosine or tryptophan fluorescence intensities is difficult due to tyrosine and tryptophan spectral overlap. Although tyrosine fluoresces consistently at 302 nm, tryptophan fluorescence peak maximum varies between 320 nm and 355 nm depending on solvent accessibility (Fig. 2A) [16]. More specifically, the degree of tyrosine and tryptophan spectral overlap is dependent on the blue- or redshift of tryptophan fluorescence in response to protein folding or unfolding. Therefore it is essential to deconvolute the overlapping spectra between tryptophan and tyrosine. The fluorescence spectrum of Trp can be analytically expressed by the log-normal function (Equation (1)) [17–19]:
Application of fluorescence intensity ratio between tyrosine and tryptophan (FirbY-W) to monitor unfolding event of structured proteins Tyrosine fluorescence is usually invisible in folded proteins due the 143
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Fig. 1. Tyrosinate is not induced in commonly used buffers, salts or denaturants. Normalized Tyrosine fluorescence spectra at pH 7.5 (A) and pH 11.3 (B). The fluorescence peak around 340 nm stems from tyrosinate in panel B. Normalized fluorescence spectra of Nop15 (C) and U1A (D) were collected in the presence of 100 mM Citrate pH 5.5, 100 mM HEPES pH 7.4, 100 mM MES pH 6.0, 100 mM Sodium Phosphate pH 7.0, 200 mM Arg/Glu pH 8.5, 100 mM Tris-HCl pH 7.5, 1 M NaCl, 1 M KCl, 8 M Urea or 6 M Guanidinium HCl.
and fluorophore flexibility. As shown by Fig. 3, all these proteins display distinguishable tyrosine fluorescence upon protein unfolding by urea or guanidinium. Considering the diversity of model systems selected, we believe that change in FRET efficiency upon protein unfolding is common to most tryptophan- and tyrosine-containing proteins thereby illustrating the robustness of our method. To reflect the changes in FRET efficiency, we defined a parameter termed Fluorescence Intensity Ratio Between Tyrosine and Tryptophan (FirbY-W). FirbY-W is positively correlated to protein size and therefore able to follow the protein unfolding process due to the fact that unfolded proteins generally have larger inter-residue distances. To test our
efficient resonance energy transfer from tyrosine to tryptophan. Once proteins are unfolded, however, concomitant increases in the average tyrosine-tryptophan distance decreases the resonance energy transfer efficiency. Consequently, the energy in tyrosine is retained, leading to an amplification of tyrosine fluorescence. Our method depends on this change in tyrosine-tryptophan FRET efficiency to monitor protein folding status. To confirm that the change is common to other proteins besides the few reported cases [22,23], we collected fluorescence spectra of Nob1, Nop9, Nop1581−191 Y94W and snRNP7092−202 (Fig. 3, Fig. S1). These proteins were selected because they represent different acceptor/donor (tryptophan/tyrosine) contents, spatial distribution,
Fig. 2. Deconvolution of tyrosine and tryptophan fluorescence. (A) Normalized tyrosine-tryptophan FRET spectra of Nop15 collected at different urea concentrations. The spectra were normalized by tryptophan fluorescence maximum. The horizontal and vertical arrows indicate the red-shift of tryptophan peak positions and the buildup of tyrosine fluorescence with increased urea concentration, respectively. Urea concentrations are listed aside in molar. (B) Deconvolution of the tyrosine and tryptophan fluorescence. An example spectrum of Nop15 collected in 7.0 M urea is shown. The equations to fit the tyrosine and tryptophan fluorescence spectrum are described in the text.
144
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Fig. 3. Tyrosine fluorescence is visible once unfolding of (A) Nob1 (PDB ID: 2LCQ), (B) Nop9 (PDB ID: 5SVD), (C) Nop15 (PDB ID: 5T9P) and (D) snRNP70 (PDB ID: 4PKD) as shown in the top row. Fluorescence spectra of the folded and unfolded states are shown in black and red lines, respectively. Nop15 and snRNP70 is unfolded by urea. Nob1 and Nop9 are unfolded by 6 M Guanidinium HCl. Nop1581−191 mutant Y94W was used here. Tyrosine and tryptophan residues are denoted as blue dots and red stars, respectively, along with the primary sequence (the middle row), or shown as blue and red spheres in the crystal structures (the bottom row). The disorder probabilities of the proteins were predicted using the PrDOS server (http://prdos.hgc.jp/cgi-bin/top.cgi). Only the protein regions with crystal structure available are shown in the bottom row. A homologue structure for Nob1 is used. The molecular graphics were prepared by PyMOL. Fig. 4. Protein unfolding of (A) Nop15, (B) snRNP70 and (C) Nop9 tracked by fluorescence intensity ratio between tyrosine and tryptophan (FirbY-W) (top row, ▲), tryptophan fluorescence peak position (middle row, ○) and circular dichroism (bottom row, ●). Circular dichroism signals at 222 nm were plotted. The two-state model was assumed in fitting the unfolding curves. The units for ΔG and m-value are kcal/mol, kcal/(mol M) and M, respectively.
145
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
method, we used FirbY-W to track unfolding of Nop1581−191 Y94W, snRNP7092−202 and Nop946−645 (Fig. 4, Fig. S2). Our results show that in these three cases, FirbY-W increases as a function of protein unfolding. We assumed that all transitions follow the 2-state model. In the case of Nop15, a mutant tryptophan was introduced at position Y94, which is located at a surface β-strand. The folding energy measured by FirbY-W (ΔG = 2.1±0.1 kcal/mol) is comparable to that measured by CD (1.8±0.5 kcal/mol) (Fig. 4A). We note that using FirbY-W yields a smoother transition curve and a smaller error in fitting ΔG when compared to CD. In addition, FirbY-W curves read at 298 nm and 302 nm yield a similar ΔG (1.9±0.3 kcal/mol versus 2.1±0.1 kcal/mol). The similarity in ΔG values obtained at different wavelengths is a strong indication of successful deconvolution of tyrosine and tryptophan fluorescence. In the case of snRNP7092−202, the only tryptophan at the dynamic tail is completely solvent-exposed and distant to all tyrosine located at the structured RNA recognition motif (RRM). As a result, tryptophan emission maximum fails to detect protein unfolding due to the lack of any unfolding-induced red shift (Fig. 4B, the middle row). In contrast, FirbY-W sensitively tracks the unfolding process. It is noteworthy that protein stability measured by FirbY-W is different from that measured by CD (Fig. 4B). This is possibly because the flexible region where the tryptophan is located only weakly interacts with the RRM. We also tested our method in Nop9, a system with multiple tyrosine and tryptophan residues. The FirbY-W denature curve at high guanidinium concentration suggests Nop9 folding is highly cooperative and stable. Using our FirbY-W method, we measured similar unfolding energy values for this denaturation process with a greater signal-to-noise ratio when compared to methodologies using tryptophan peak position or CD (Fig. 4C). In all cases, our FirbY-W method yields smaller error bars compared to CD and Trp emission maximum (Fig. S3, Table S1A and S1B).
phosphor-mimicry (Ser to Glu mutation) can partially mimic the RNAbinding promotion and is more compact than the non-phosphorylated SLBP [27]. Although the idea of tyrosine-tryptophan FRET was utilized in our previous study, systematic testing of the method and rigorous deconvolution of tyrosine and tryptophan fluorescence were not performed, which prevents deeper data analysis. In addition, formation of tyrosinate is particularly a concern for the SLBP system, because carboxylic groups of glutamic acid and aspartic acid, like the SLBP C-terminus, were speculated to facilitate formation of tyrosinate. Furthermore, the study was carried out on the glutamic phosphor-mimicry, which is significantly different from the authentic phosphorylation in modulating RNA binding affinity of SLBP. In this study, we first addressed the concern about tyrosinate formation in SLBP. To remove the interference of tryptophan, we mutated the only tryptophan of SLBP to phenylalanine (W249F). A tyrosinatecontaminated fluorescence spectrum has a signature shoulder around 350 nm. As shown by Fig. S4, the tyrosine fluorescence spectrum of the W249F construct has no shoulder around 350 nm. This result clear showed that tyrosinate formation is not induced by the acidic patch, reinforcing our claim that tyrosinate formation requires certain special protein tertiary structure. As shown by mass spectrometry (Fig. S5), we have also successfully obtained hyper-phosphorylated SLBP, which provides us an opportunity to analyze phosphorylated SLBP directly using FirbY-W to obtain a deeper insight. The minimal functional motif of SLBP contains six tyrosine residues and one tryptophan located at the hydrophobic core in the RNA-bound state. We collected and compared the fluorescence spectra of SLBP in different phosphorylated states. Despite the dramatic difference in RNA-binding affinities, circular dichroism fails to detect any secondary structure difference between phosphorylated and non-phosphorylated SLBP (Fig. 5C). A blue shift is observed in tryptophan fluorescence emission maximum upon phosphorylation, indicating the increased extent of tryptophan bury and residual structure (Fig. 5D). However, denature curves of the residual structure monitored by tryptophan emission maximum have significant noise and a narrow dynamic range for non-phosphorylated SLBP (Fig. 5D). In contrast, FirbY-W analysis yields denature curves with a much lower noise level (Fig. 5E and F). In addition, FirbY-W values of phosphorylated SLBP in general is smaller than unphosphorylated protein, suggesting that phosphorylation may pre-fold SLBP into the conformation that is similar to the RNA-bound state. Indeed, we found that phosphorylated SLBP is more tolerant to urea denaturation. This finding at least partially explains the dramatic RNA-binding enhancement observed before. It is noteworthy that this finding is also in line with the blue-shift of tryptophan fluorescence emission maximum induced by phosphorylation, suggesting that the protein becomes more compact upon phosphorylation (Fig. 5D). However, the transition curve of tryptophan emission maximum is very noisy.
Application of FirbY-W in studying conformational redistribution of intrinsically disordered proteins Despite lack of definite secondary or tertiary structure, intrinsically disordered proteins play various indispensable functions [7]. The conformation distribution in the IDP ensemble is very sensitive to external signaling and important for IDP functions. Although intrinsically disordered regions account for more than 30% of the human proteome, our understanding of IDPs is still limited due to the difficulty in characterizing their physical chemical behavior. Most traditional biophysical spectroscopic methods are insensitive to IDP ensemble redistribution since secondary or tertiary structure is absent in the conformational ensemble. FRET has been used to describe IDP behavior, but introduction of exogenous FRET donor/acceptor pair is usually needed [24]. To test whether our method is applicable to IDPs, the following three disordered proteins were selected: signal recognition particle (SRP19), dicer-like protein 1 (DCL1) and histone mRNA stemloop binding protein (SLBP). As shown by Fig. 5A and E, tyrosine fluorescence shoulder is clearly visible in the three disordered proteins tested here. Visibility of tyrosine fluorescence in IDPs is distinct from what we observed in folded proteins (Fig. 3). We further found that DCL1 and SRP19 have no FirbY-W change at different denaturant concentrations, suggesting the proteins have no residual structure, consistent with previous results [25,26]. Residual structural formation is not absent in every IDP conformational ensemble, nor all possible conformations sampled equally in the ensemble. One pertinent example is the histone mRNA stem-loop binding protein (SLBP). SLBP is essential for histone mRNA processing in response to cell duplication. The minimal functional motif of SLBP (residue 190–276) contains an RNA-binding domain and a C-terminus bearing four serine residues (SNSDSDSD) [27]. Hyper-phosphorylation of the terminus is required for SLBP function and dramatically increases RNA binding affinity [27]. Our previous study also showed that
Application of FirbY-W in studying ligand binding of intrinsically disordered proteins Our previous study shows that RNA binding of SLBP induces protein folding [27]. SLBP190−276 contains a single tryptophan residue at the protein hydrophobic core, surrounded by 6 tyrosine residues. In the folded state induced by RNA binding, the average tyrosine-tryptophan distance is 11.4 Å (Fig. 6A). The protein folding increases FRET efficiency by bringing the tyrosine and tryptophan residues closer. To test whether FirbY-W will sense the RNA binding event, we collected fluorescence spectra of SLBP with and without RNA. As shown in Fig. 6B, upon RNA binding, the tyrosine fluorescence shoulder peak disappears. The binding event reduces FirbY-W by roughly 50%. In the case of SLBP/RNA binding, the tryptophan blue shift can also be used to detect RNA binding. However, this is not true for the tryptophan residues that do not experience changes in micro-environments during ligand binding. For example, RNA binding does not change the 146
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Fig. 5. FirbY-W studying protein size changes of intrinsically disordered proteins. Fluorescence spectra (A) and FirbY-W (B) of disordered DCL1and SRP19. (C) Circular dichroism profiles of phosphorylated SLBP and non-phosphorylated SLBP. (D) Protein unfolding curves monitored by tryptophan fluorescence emission maximum. m-values are 1.1 ± 0.5 kcal/ (mol M) and 1.5 ± 0.9 kcal/(mol M) for phosphorylated and non-phosphorylated SLBP, respectively. (E) Fluorescence spectra of phosphorylated and non-phosphorylated SLBP. (F) Unfolding profile of phosphorylated SLBP and non-phosphorylate SLBP monitored by FirbY-W. For clarity, only the fluorescence spectra at 0 M of urea are shown. The unit for ΔG is kcal/mol. m-values for phosphorylated SLBP and non-phosphorylated SLBP are 1.0 ± 0.2 kcal/ (mol M) and 1.3 ± 0.4 kcal/(mol M), respectively.
tryptophan emission maximum of snRNP70 (Fig. 6D). As suggested by deconvolution of the Trp and Tyr fluorescence spectra, the Trp emission maximum is at 350 nm in the free and RNA-bound states (Fig. S6). However, FirbY-W can still sense the RNA binding as shown by Fig. 6D and Fig. S6. FirbY-W changes from 0.53 to 1.29 upon RNA binding. It is noteworthy that RNA binding decreases FirbY-W of snRNP70, in contrast to the FirbY-W increase upon SLBP/RNA binding. This opposite effect on snRNP70 FirbY-W can be interpreted by a closer structural analysis (Fig. 6C). The only tryptophan of snRNP70 located at the disordered C-terminus has transient opportunities to approach the six tyrosine residues, which allows tyrosine-tryptophan FRET. However, upon RNA binding the disordered C-terminus becomes ordered and positions the tryptophan residues away from the tyrosine residues, with an average distance of 20.4 Å.
Discussion Despite the prevalence of tyrosine-tryptophan FRET in proteins, it is underused in probing protein dimensions, due to the concerns about tyrosinate fluorescence, the invisibility of tyrosine fluorescence in folded proteins, and spectral overlap of tyrosine and tryptophan fluorescence. Here we show that tyrosinate fluorescence is not observed in commonly used buffers, salts or denaturants. Tyrosinate fluorescence was reported at high concentration of acetate and high pH (> 10) [15] which are rarely used in biochemistry studies. Under physiological pH and salt concentrations, tyrosinate fluorescence is only observed in few proteins, such as bovine testes calmodulin, alpha-purothionin and betapurothionin [12,13]. Interestingly the tyrosinate fluorescence disappears once these proteins are unfolded [12]. Therefore it is likely that
Fig. 6. Application of FirbY-W in studying protein/RNA binding. Tyrosine and tryptophan distribution in RNA complex of SLBP (PDB ID: 4TV0) (A) and snRNP70 (PDB ID: 4PKD) (C). The graphic were prepared using PyMOL. Blue spheres denote the C-alpha atoms of tyrosine residues and red ones for tryptophan residues. RNA is shown in orange cartoon and protein in green cartoon. Normalized fluorescence spectra of SLBP and snRNP70 are shown in panel B and D, respectively.
147
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
screening method for protein/RNA interaction, it is not suitable for quantitative determination of RNA binding affinity, as the sensitivity limit of Tyr-Trp FRET is around μM, which is much higher than dissociation constants (Kd) of many protein/RNA binding events. FirbY-W also has the potential to probe protein:protein interaction if the binding event incurs change of FRET between Tyr and Trp residues. In this situation, the fluorescence spectrum of the complex will not be a linear combination of each individual binder's spectrum.
the specific tertiary structure of these proteins positions tyrosine in a special environment that facilitates tyrosinate formation. Tyrosinate fluorescence observed in these samples should not be a concern for other protein studies. In this study we used FirbY-W as an index to reflect the overall dimension of proteins. On one hand, FirbY-W reflects the average protein size; on the other, FirbY-W eliminates data noise caused by sample concentration deviations, which usually stems from pipetting error or protein aggregation. For example, we found that Nop9 has slight protein aggregation at low concentration of guanidinium, which results in vibration of CD signals (Fig. 4). However, FirbY-W is tolerant to the slight protein aggregation problem. Since tyrosine and tryptophan fluorescence is collected in the same spectrum scan, FirbY-W is also tolerant to fluorimeter instability. For proteins with tryptophan residues completely exposed to solvent as exemplified by snRNP70, it is impossible to use tryptophan fluorescence to track protein unfolding. However, FirbY-W can still sense protein unfolding because FirbY-W directly detects global size changes of proteins instead of local microenvironments of tryptophan residues. Even Trp emission maximum can be used to track protein unfolding, as in Nop15 and Nop9, it suffers from a narrow dynamic range from 340 nm to 355 nm. This narrow dynamic range is vulnerable to fluorescence intensity vibration which frequently happens around emission maximum, to the low spectrum resolution which is at the nanometer scale. These two factors cause instability of unfolding curves tracked by Trp emission maximum. In addition, our method also provides information on tryptophan fluorescence emission maximum. For these reasons, FirbY-W is superior to CD or fluorescence emission maximum in protein unfolding studies. The reported Foster distances for tyrosine/tryptophan FRET range from 9 Å to 18 Å, the dimension of most protein domains [9,28–31]. This means that FirbY-W is sensitive to folding/unfolding events of most proteins. One exception would be in proteins where all tyrosine residues are adjacent to all tryptophan residues in the primary sequence. However, as exemplified by Nop15 and DCL1 in our study, mutant Trp can be introduced for measurement of FirbY-W. Although it is still difficult to predict whether FirbY-W is applicable for a protein, experimental testing can be easily performed by comparing the FRET spectra with and without denaturant as shown by Fig. 3. Currently it is still challenging to use FirbY-W to calculate donor-acceptor distance due to the fact that most proteins contain more than one tyrosine and/or tryptophan residue, but application of FirbY-W to monitor protein folding/ unfolding can be generalized to most protein systems. The unique power of our method is its ability to track behavior of intrinsically disordered proteins. Monitoring dynamic behavior of IDPs is challenging and is only feasible for a few experimental methods such as NMR and FRET. FRET study of IDPs usually necessitates the introduction of exogenous fluorophores. The lack of secondary structure and tertiary structure in IDPs makes canonical biophysical methods, such as CD or tryptophan fluorescence, problematic. Using model disordered proteins DCL1 and SRP19, we showed that our method is able to determine whether a protein is unstructured based on its response to urea concentration change. Moreover, application of FirbY-W to SLBP shows that our method is capable of detecting the subtle phosphorylation-triggered conformation redistribution that escapes the detection of circular dichroism. This is probably because phosphorylation does not affect the secondary structure content of SLBP, but reduces the charge repulsion and promotes protein folding. In addition, we also show that FirbY-W responds to RNA binding of SLBP and snRNP70. Considering that binding-induced folding is common to many IDPs and the process will change IDP size, FirbY-W can be applied to study other IDP-involved binding processes [32]. Although RNA has no fluorescence from 280 nm to 600 nm, it is noteworthy that UV absorbance of RNA around 260–280 nm can significantly weaken the incident light for fluorescence excitation and consequently decreases available energy for FRET. At high concentrations (> 1 Au), RNA can dramatically decrease signal of FRET spectra. Although FirbY-W can be a convenient
Experimental procedures Protein purification Wild type Nop1581−191, Nop1581−191 Y94W, SLBP190−276, Nop946−645 and Nob1 were purified as reported [27,33,34]. SLBP was in vitro phosphorylated by casein kinase II (New England Biolabs) at room temperature in 50 mM Tris-HCl, 10 mM MgCl2, 2 mM DTT and 1 mM ATP. Casein kinase II specifically phosphorylates S269, S271, S273 and S275. DCL1, snRNP7092−202 and SRP19 were purified using the following procedure. DCL1 encoding DNA was synthesized by Genscript and a tryptophan residue was inserted at the protein C-terminus. DNA encoding snRNP7092−202 and SRP19 were ordered from Addgene (https://www.addgene.org/) and Harvard PlasmID Database (https://plasmid.med.harvard.edu/PLASMID/Home.xhtml), respectively. All these protein-encoding DNA were cloned into pSMT3 (Memorial Sloan Kettering Cancer Center). All protein sequences of the constructs used in this study are shown in Fig. S1. Transformed E. coli BL21-CodonPlus (DE3) cells were grown in LB media at 37 °C. 0.5 mM IPTG was added when OD600 reached 0.6. The culture was continued at 22 °C overnight. Cell pellets were re-suspended in 25 mM HEPES, pH 7.5, 1 M NaCl, 0.2 mM TCEP, 25 mM imidazole, 1 mM PMSF, 1 mg/ml lysozyme and lysed by sonication, followed by centrifugation (14,000 rpm for 40 min) to remove cell debris. The supernatant was applied to 5 ml HisPur Ni-NTA resin (Thermo Scientific), washed with 200 ml 25 mM HEPES, pH 7.5, 1 M NaCl, 0.2 mM TCEP, 25 mM imidazole, and eluted with 25 mM HEPES, pH 7.5, 500 mM NaCl, 1 mM TCEP, 250 mM imidazole. The N-terminal SUMO-tagged protein was cleaved overnight with 1 μg/ml of Ulp1 at 4 °C. The cleaved sample was diluted 5 fold using 20 mM HEPES, pH 7.5, 20 mM NaCl, 1 mM TCEP and loaded onto a 5-ml HiTrap Heparin column (GE Healthcare). The sample was eluted with a linear gradient from 0 to 2 M NaCl in 20 mM HEPES, 0.2 mM TCEP. DCL1 was eluted with 400 mM NaCl; snRNP70 and SRP19 were eluted with 860 mM NaCl. The fractions containing target proteins were pooled and further purified using a HiLoad 16/60 Superdex75 column (GE Healthcare) equilibrated with 25 mM HEPES, pH 7.5, 500 mM NaCl, 0.2 mM TCEP. The identities of the proteins were confirmed by mass spectrometry, and the purities were > 95% based on SDS-PAGE. Preparation of RNA sample Histone mRNA stem-loop RNA for SLBP is a product of Dharmacon, Inc. (5′-GGC CAA AGG CCC UUU UCA GGG CCA CCC A-3′). The stemloop RNA for snRNP70 (GGG AGA UAC CAU GAU CAC GAA GGU GGU UUU CCC) was prepared by in vitro T7 transcription, which is carried out at 37 °C for 8 h in 100 mM Tris-HCl, pH 8.5, 20 mM MgCl2, 3 mM TCEP, 2 mM spermidine, 3% (w/v) PEG 8000, 0.01% (v/v) Triton X100, 4 mM NTPs, 2 units of inorganic pyrophosphatase, 0.6 μM doublestranded DNA template, and 0.06 mg/ml T7 RNA polymerase. Transcription product was purified by 15% polyacrylamide gel (30 cm × 40 cm x 1.6 mm) in the presence of 7 M urea and 1X TBE. The gels were pre-run for 30 min to a temperature of 50 °C, and RNA samples were resolved under a constant power of 75 W for 6 h, during which RNAs travel about two thirds of the gel. The excised gel with target RNA was electronically eluted overnight in 10 mM Tris-HCl, pH 8, and 1 mM EDTA. RNAs were incubated at 90 °C for 2 min and snap148
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
CD data collection
cooled on ice to refold the RNA. The homogeneity of refolded RNA was confirmed by single bands on 10% polyacrylamide native TBE gels. 10 μM proteins were mixed with binding RNA into 1:1 ratio.
CD data from 260 nm to 185 nm were collected using a Jasco J-815 circular dichroism spectropolarimeter at 295 K using a 1-mm cuvette. All protein samples (200 μL, ∼30 μM) were dissolved into 20 mM TrisHCl, pH 7.5, 100 mM NaCl, 0.1 mM TCEP. Corresponding CD signal for buffers were collected and subtracted. CD signals below 200 nm were seriously interfered by high concentration of urea or Guanidinium HCl. Therefore, CD signal at 222 nm were used to track the protein folding event.
Fluorescence data collection Fluorescence data were collected using a Varian Cary Eclipse fluorometer at 295 K with a 5-mm cuvette. All protein samples (600 μL, ∼20 μM) were exchanged into 20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1 mM TCEP before urea or Guanidinium HCl was added. The samples were centrifuged at 10,000 g for 10 min at 4 °C before data collection. Excitation wavelength was set to 275 nm and spectra from 280 nm to 400 nm were collected with 5-nm excitation and 5-nm emission slits. The same settings were used to collect buffer fluorescence as reference. After subtraction of background fluorescence, the data were smoothed using a 20-point Savitzky-Golay function and then normalized by tryptophan fluorescence maximum.
Fitting of protein unfolding by the two-state model The two-state model was assumed in fitting the unfolding process of the proteins versus denaturant concentrations. Nonlinear regression was carried out using the following equation:
( exp (
D⋅exp
Deconvolution of tyrosine and tryptophan fluorescence
E (x ) =
For protein systems with more than one tryptophan residue, it is possible that all three classes of tryptophan are present in the sample. For simplicity and clarity, assume tyrosine fluorescence and only one tryptophan fluorescence component are present in the sample. It is convenient to extend the following calculation into the case in which other tryptophan fluorescence components are present. The theoretical spectrum at discrete wavenumber vi can be written as:
)+N )+1
mx − ΔG RT
mx − ΔG RT
(6)
where E is fluorescence or CD observables; x is denaturant concentration; m is the so-called m-value for urea or Guanidinium-HCl; N and D are the signal baselines at the native and denatured states, respectively; ΔG is the unfolding energy of proteins; R is the gas constant and T is the temperature (295 K) at which experiments were carried out. A tilted baseline was observed for the unfolding processes with Guanidinium HCl and a modified two-state equation was used as the following:
I (vi ) = ImY ⋅fY (vi ) + ImW ⋅fW (vi )
( exp (
(ax + D)⋅exp
where ImY and ImW are the maximum peak intensities of tyrosine and tryptophan, respectively; fY and fW are spectrum shape defined by Equation (1) and Equation (2). The deviation between experimental data E (vi ) and theoretical data are expressed as:
E (x ) =
) + (ax + N ) )+1
mx − ΔG RT
mx − ΔG RT
(7)
where N is the total number of discrete data points of the spectrum. The minimal value of Q or the optimal deconvolution is found when the following relationship is satisfied:
where a is the slope of the baseline; N and D are the intercepts on the Xaxis at the native and denatured states, respectively, with other parameters being the same as Equation (6). For the CD method, experimental observables, E(x), are read from 222 nm. For the Trp emission maximum method, experimental observables are the emission positions. For the FirbY-W method, the ratio of the deconvoluted Tyr and Trp fluorescence intensity is used to fit Eq (7). All data fitting were performed using Graphpad Prism.
∂Q =0 ∂Im
Conflicts of interest
By carrying out partial derivation with respect to all spectrum components (ImY , ImW …), the group of equations obtained can be written as:
The authors declare that they have no conflicts of interest with the contents of this article.
N
Q=
∑ (E (vi) − I (vi))2 i
A⋅Im = Y Author contributions
where
JZ designed research; KBD, ZZ, EAK and JZ performed research; JZ and KBD analyzed data; and JZ wrote the paper.
I Im = ⎜⎛ mY ⎞⎟ I ⎝ mW ⎠ N
⎛ ∑ 1 E (vi )⋅fY (vi ) ⎞ Y = ⎜ Ni = ⎟ ∑ E (vi )⋅fW (vi ) ⎝ i=1 ⎠ N
Funding This work was supported in whole by the start-up funding of Chemistry Department, College of Arts and Sciences, University of Alabama at Birmingham.
N
⎛ ∑ fY (vi )⋅fY (vi ) ∑i = 1 fW (vi )⋅fY (vi ) ⎞ A = ⎜ Ni = 1 ⎟ N ∑ f (v )⋅f (v ) ∑i = 1 fW (vi )⋅fW (vi ) ⎝ i=1 Y i W i ⎠ Once an inverse matrix of A , A−1 is obtained, Im can be determined as the following:
Acknowledgments We thank Michael J. Jablonsky for technical support, Dr. Chad Petit for stimulating comments and discussion. This work was suppoDrted by faculty start-up funding from the Chemistry Department, College of Arts and Sciences, UAB.
Im = A−1 Y The fitting was carried out using a Python program, which is available upon request. 149
Analytical Biochemistry 557 (2018) 142–150
K.B. Davis et al.
Appendix A. Supplementary data
Biophys. J. 81 (3) (2001) 1699–1709. [18] Y.K. Reshetnyak, E.A. Burstein, Decomposition of protein tryptophan fluorescence spectra into log-normal components. II. The statistical proof of discreteness of tryptophan classes in proteins, Biophys. J. 81 (3) (2001) 1710–1734. [19] Y.K. Reshetnyak, Y. Koshevnik, E.A. Burstein, Decomposition of protein tryptophan fluorescence spectra into log-normal components. III. Correlation between fluorescence and microenvironment parameters of individual tryptophan residues, Biophys. J. 81 (3) (2001) 1735–1758. [20] D.B. Siano, D.E. Metzler, Band shapes of the electronic spectra of complex molecules, J. Chem. Phys. 51 (1969) 106. [21] E.A. Burstein, Log-normal description of fluorescence spectra of organic fluorophores, Photochem. Photobiol. 64 (1996) 5. [22] R. Boteva, et al., Dissociation equilibrium of human recombinant interferon gamma, Biochemistry 35 (47) (1996) 14825–14830. [23] M. Eisenhawer, et al., Fluorescence resonance energy transfer shows a close helixhelix distance in the transmembrane M13 procoat protein, Biochemistry 40 (41) (2001) 12321–12328. [24] S. Muller-Spath, et al., From the Cover: charge interactions can dominate the dimensions of intrinsically disordered proteins, Proc. Natl. Acad. Sci. U.S.A. 107 (33) (2010) 14609–14614. [25] I.P. Suarez, et al., Induced folding in RNA recognition by Arabidopsis thaliana DCL1, Nucleic Acids Res. 43 (13) (2015) 6607–6619. [26] T.S. Maity, K.M. Weeks, A threefold RNA-protein interface in the signal recognition particle gates native complex assembly, J. Mol. Biol. 369 (2) (2007) 512–524. [27] J. Zhang, et al., Molecular mechanisms for the regulation of histone mRNA stemloop-binding protein by phosphorylation, Proc. Natl. Acad. Sci. U.S.A. 111 (29) (2014) E2937–E2946. [28] J. Eisinger, Intramolecular energy transfer in adrenocorticotropin, Biochem. 8 (10) (1969) 3902–3908. [29] P.W. Schiller, Fluorescence study on the conformation of a cyclic enkephalin analog in aqueous solution, Biochem. Biophys. Res. Commun. 114 (1) (1983) 268–274. [30] S.F. Pearce, E. Hawrot, Intrinsic fluorescence of binding-site fragments of the nicotinic acetylcholine receptor: perturbations produced upon binding alpha-bungarotoxin, Biochem. 29 (47) (1990) 10649–10659. [31] M.J. Moreno, M. Prieto, Interaction of the peptide hormone adrenocorticotropin, ACTH(1-24), with a membrane model system: a fluorescence study, Photochem. Photobiol. 57 (3) (1993) 431–437. [32] P.E. Wright, H.J. Dyson, Linking folding and binding, Curr. Opin. Struct. Biol. 19 (1) (2009) 31–38. [33] J. Zhang, L.E. Gonzalez, T.M.T. Hall, Structural analysis reveals the flexible C-terminus of Nop15 undergoes rearrangement to recognize a pre-ribosomal RNA folding intermediate, Nucleic Acids Res. 45 (5) (2017) 2829–2837. [34] J. Zhang, et al., Nop9 is a PUF-like protein that prevents premature cleavage to correctly process pre-18S rRNA, Nat. Commun. 7 (2016) 13085.
Supplementary data related to this article can be found at https:// doi.org/10.1016/j.ab.2018.07.022. References [1] B.T. Bajar, et al., A guide to fluorescent protein FRET pairs, Sensors 16 (9) (2016). [2] S.M. Kelly, T.J. Jess, N.C. Price, How to study proteins by circular dichroism, Biochim. Biophys. Acta Protein Proteonomics 1751 (2) (2005) 119–139. [3] J.T. Vivian, P.R. Callis, Mechanisms of tryptophan fluorescence shifts in proteins, Biophys. J. 80 (5) (2001) 2093–2109. [4] P. Pattanaik, et al., Unusual fluorescence of W168 in Plasmodium falciparum triosephosphate isomerase, probed by single-tryptophan mutants, Eur. J. Biochem. 270 (4) (2003) 745–756. [5] C.R. Middaugh, et al., Infrared spectroscopy, Meth. Mol. Biol. 40 (1995) 137–156. [6] J. Zhang, Y.B. Yan, Probing conformational changes of proteins by quantitative second-derivative infrared spectroscopy, Anal. Biochem. 340 (1) (2005) 89–98. [7] B. Xue, A.K. Dunker, V.N. Uversky, Orderly order in protein intrinsic disorder distribution: disorder in 3500 proteomes from viruses and the three domains of life, J. Biomol. Struct. Dynam. 30 (2) (2012) 137–149. [8] V.N. Uversky, A Decade and a Half of Protein Intrinsic Disorder: Biology Still Waits for Physics, Protein science: a publication of the Protein Society, 2013, pp. 693–724 22(6). [9] G. Weber, Fluorescence-polarization spectrum and electronic-energy transfer in tyrosine, tryptophan and related compounds, Biochem. J. 75 (1960) 335–345. [10] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, (2006) (Chapter 16). [11] A.G. Szabo, et al., Tyrosinate fluorescence maxima at 345 nm in proteins lacking tryptophan at pH 7, FEBS Lett. 94 (2) (1978) 249–252. [12] F.G. Prendergast, P.D. Hampton, B. Jones, Characteristics of tyrosinate fluorescence emission in alpha- and beta-purothionins, Biochemistry 23 (26) (1984) 6690–6697. [13] S. Pundak, R.S. Roche, Tyrosine and tyrosinate fluorescence of bovine testes calmodulin: calcium and pH dependence, Biochemistry 23 (7) (1984) 1549–1555. [14] D.M. Rayner, D.T. Krajcarski, A.G. Szabo, Excited state acid–base equilibrium of tyrosine, Can. J. Chem. 56 (9) (1978) 1238–1245. [15] J.B.A.L.,W.R. Ross, K.W. Rousslang, H.R. Wyssbrod, Tyrosine fluorescence and phosphorescence from proteins and polypeptides. In Topics in fluorescence spectroscopy, Biochem. Appl. 3 (1992) 63. [16] M.R. Eftink, Fluorescence techniques for studying protein structure, Meth. Biochem. Anal. 35 (1990) 13. [17] E.A. Burstein, S.M. Abornev, Y.K. Reshetnyak, Decomposition of protein tryptophan fluorescence spectra into log-normal components. I. Decomposition algorithms,
150