Dynamics of the Tec‐family tyrosine kinase SH3 domains

Abstract

The Src Homology 3 (SH3) domain is an important regulatory domain found in many signaling proteins. X‐ray crystallography and NMR structures of SH3 domains are generally conserved but other studies indicate that protein flexibility and dynamics are not. We previously reported that based on hydrogen exchange mass spectrometry (HX MS) studies, there is variable flexibility and dynamics among the SH3 domains of the Src‐family tyrosine kinases and related proteins. Here we have extended our studies to the SH3 domains of the Tec family tyrosine kinases (Itk, Btk, Tec, Txk, Bmx). The SH3 domains of members of this family augment the variety in dynamics observed in previous SH3 domains. Txk and Bmx SH3 were found to be highly dynamic in solution by HX MS and Bmx was unstructured by NMR. Itk and Btk SH3 underwent a clear EX1 cooperative unfolding event, which was localized using pepsin digestion and mass spectrometry after hydrogen exchange labeling. The unfolding was localized to peptide regions that had been previously identified in the Src‐family and related protein SH3 domains, yet the kinetics of unfolding were not. Sequence alignment does not provide an easy explanation for the observed dynamics behavior, yet the similarity of location of EX1 unfolding suggests that higher‐order structural properties may play a role. While the exact reason for such dynamics is not clear, such motions can be exploited in intra‐ and intermolecular binding assays of proteins containing the domains.

Abbreviations

ACN

acetonitrile

Amp

ampicillin

ATP

adenosine triphosphate

BiFC

bimolecular fluorescence complementation

Bmx

bone marrow tyrosine kinase gene in chromosome X protein

Btk

Bruton's tyrosine kinase

D₂O

deuterium oxide

DTT

dithiothreitol

E. coli

Escherichia coli

FA

formic acid

FWHM

full width at half maximum

GST

glutathione‐S‐transferase

GuHCl

guanidine hydrochloride

HIV

human immunodeficiency virus

HPLC

high performance liquid chromotagraphy

HX MS

hydrogen exchange mass spectrometry

IPTG

isopropyl β‐D‐1‐thiogalactopyranoside

Itk

interleukin‐2 (IL‐2) inducible T cell kinase

LB

lysogeny broth; Nef, negative regulatory factor

NMR

nuclear magnetic resonance

nRTK

non‐receptor tyrosine kinases

PDB

protein data bank

PH

pleckstrin homology

Rlk, Txk

resting lymphocyte kinase

RMSD

root‐mean‐square deviation

SFK

Src family kinases

SH3

Src homology 3

t_1/2

half‐life

Tec

tyrosine kinase expressed in hepatocellular carcinoma

TFA

trifluoroacetic acid

TH

Tec homology

TK

tyrosine kinase

Tris

tris(hydroxymethyl)aminomethane

UND

undeuterated

UPLC

ultra performance liquid chromotagraphy

XLA

X‐linked agammaglobulinemia

Introduction

The Src Homology 3 (SH3) domain is an important modular regulatory module found in many proteins, with over 1000 unique examples documented in the literature.1, 2, 3, 4, 5, 6, 7 SH3 domains mediate protein–protein interactions with binding partners containing polyproline class II helices characterized by a PxxP motif, where x represents any residue. Some specific SH3 domains show discrimination for variations on the PxxP motif (PxxPxR/K being a common example) as well as several other non‐consensus sequences.6, 8 The SH3 structure represents an ancient fold found in both prokaryotes and eukaryotes alike.9 SH3 domains typically consist of 60 amino acids folded into a β‐barrel‐like hydrophobic core comprised of two orthogonally stacked β‐sheets. Two variable loops, termed the RT and n‐Src loops, are situated within the fold and mediate the orientation of target protein binding. The omnipresence and generally modest binding affinities associated with SH3:target protein interactions (typical K _D values in the μM range3, 6) raise important questions about how SH3 domains achieve specificity, particularly in cells expressing multiple SH3 domains and signaling partners.

The widespread importance and ubiquity of SH3 domains in diverse cellular processes have sparked extensive studies of both structure and function. Structural alignments of the hundreds of SH3 domains contained in the Protein Data Bank (PDB) reveal conservation of the basic backbone topology described above. Despite apparent similarities in tertiary structure, SH3 domain stability and dynamics vary widely.10 For example, the Drk N‐terminal SH3 domain, which has been studied extensively by NMR, was found to exist in equilibrium between a folded and unfolded state under non‐denaturing conditions at ratio of 2:1 with an exchange rate constant of 2.2 s⁻¹.11 Previously, our group probed the solution dynamics of a number of SH3 domains with hydrogen exchange mass spectrometry (HX MS) and found unpredictable dynamics and partial unfolding at near‐physiological conditions.5, 12, 13, 14 In one study, the dynamics of the SH3 domains of eight members of the Src‐family of non‐receptor tyrosine kinases (nRTK) along with four other biologically relevant SH3 domains were probed side‐by‐side.13 This small sample of SH3 domains displayed a surprisingly wide range of dynamic properties, which was not expected given both their sequence and structural homology, particularly among the members from the Src‐family kinases. Remarkable variation in the timescales (from 1 second to 20 minutes) of partial cooperative unfolding events was observed across the SH3 domains [reviewed in Ref. 13]. These observations suggest that the specificity of SH3 domain interactions may be dictated in part by their dynamic behavior, a property not predictable from sequence. We have now extended our studies to the SH3 domains of Tec‐family tyrosine kinases and compare these new results with those previously obtained for the SH3 domains from Src‐family kinases. Overall we found dynamics and partial unfolding in some, but not all, of the Tec‐family SH3 domains. Similar to the previous SH3 domains we analyzed, these unfolding events could not be predicted from sequence or tertiary structure. We interpret our findings in light of the known structures for members of this family and compare the results to the previous studies, both in terms of the magnitude and locations of the dynamics and sequence similarity.

Results

Sequence comparison of Src and Tec‐family kinase SH3 domains

In this study, we examined the solution dynamics of the SH3 domains derived from the Tec kinase family using HX MS. Tec family kinases are expressed in hematopoietic tissues and are well known to regulate T‐and B‐lymphocyte differentiation and function.15, 16, 17, 18 There are five members of the Tec‐family: Tec (tyrosine kinase expressed in hepatocellular carcinoma), Itk [interleukin‐2 (IL‐2) inducible T cell kinase], Btk (Bruton's tyrosine kinase), Rlk/Txk (Resting lymphocyte kinase), and Bmx (Bone Marrow Kinase gene on the X chromosome). Like the Src‐family kinases, members of the Tec family contain an SH3 domain (although the Bmx SH3 is non‐canonical based on sequence homology), an SH2 domain, and a kinase domain. A recent crystal structure of the SH3‐SH2‐Kinase domain fragment of the Tec family kinase, Btk, also suggests that the Tec family SH3 and SH2 domains serve a regulatory role much like that of the Src family.19 Unlike the Src kinases, Tec‐family kinases (excluding Txk) contain pleckstrin homology (PH) and Tec homology (TH) domains within their N‐terminal regions, which contribute to membrane targeting and overall regulation of kinase activity.18, 20 The sequence similarity (Fig. 1) between Tec‐family and Src‐family SH3 domains is moderate, with the most conservation in the β strands that form the backbone of the structure.

Figure 1.

Primary and secondary structure comparisons for SH3 domains. (A) Sequence alignment of SH3 domains. Pink and green boxes represent protein domains within the Tec and Src family tyrosine kinases, respectively. The secondary structure is detailed at the top of the figure, and the residues in each β‐strand are enclosed by black boxes. As the sequence of Amphiphysin II is longer in the n‐Src and distal loops, the extra amino acids [(a) NPEEQ; and (b) DWNQHKEL] were removed to maintain structural continuity of the alignment. The residue numbers listed at the beginning and end of each sequence were taken from the domain's position in the full length protein. Blue letters indicate residues that are involved in cooperative unfolding in Tec‐family proteins (this study) or Src‐family and related proteins.13 (B) A comparison of cartoon representations of the structures of Itk SH3 (2RN8), Btk SH3 (1AWX), and Tec SH3 (1GL5) all light pink with that of Src SH3 (1QWE) in green are shown to the right of the sequence alignment.

Hydrogen deuterium exchange mass spectrometry

Proteins labeled in solution with D₂O gain mass that can be monitored with mass spectrometry [recently reviewed in Ref. 20,21]. Exchange of backbone amide hydrogens for deuterium in solution occurs under two kinetic limits, termed EX1 and EX2,23, 24, 25 which can be distinguished in mass spectra.14, 26, 27, 28 In the more commonly observed EX2 exchange regime, protein stability is high and brief visits to exchange competent states are most often immediately lost in favor of refolding before successful deuteration. The rate of refolding is therefore much larger than the rate of deuteration. Proteins exchanging under primarily EX2 kinetic regimes produce mass spectra showing a gradual mass increase with time in D₂O [examples in Ref. 25]. In the other regime, EX1 kinetics, protein refolding and loss of exchange competence is slow relative to the rate of deuteration. In the regions of the protein that undergo EX1, exchange occurs in a cooperative fashion, i.e., all competent positions become deuterated at once. Subsequently, two isotope envelopes appear in mass spectra of proteins exchanging with EX1 kinetics: one low‐mass envelope where no labeling has occurred and another higher‐mass envelope where labeling of all positions in the unfolding regions is complete. In the most clear cases, the two isotope envelopes are fully resolved from each other, whereas in other cases they are not, depending on how many residues are involved in the EX1‐driven event.26 To analyze EX1 spectra where there are clearly resolved isotope envelopes, the intensity of the unfolded (higher mass) species is monitored with time to extract the rate constant and half‐life (t_1/2) for unfolding. In the case of non‐resolved isotope envelopes, the width of the total isotope distribution at half maximum (FWHM) is monitored as a function of exchange time and a peak‐width plot26 of these data provides the half‐life of unfolding. Within a single protein there can be both EX2 and EX1 exchange, with some residues undergoing EX1 kinetics and others EX2 kinetics. Proteolytic digestion of the protein in an exchange‐quenched state after deuteration but before analysis, the so‐called fragment separation method,21, 29, 30, 31 can be used to localize which parts of the protein exchange under which regime. SH3 domains derived from the Src and Abl kinase families represent some of the best‐studied examples of EX1 kinetics.14 These domains display slow cooperative unfolding events in their unbound states, and this cooperative unfolding is completely quenched upon target protein interaction.

Unfolding in intact SH3 domains

The five Tec‐family kinase SH3 domains—Itk, Btk, Tec, Txk and Bmx (see Supporting Information for amino acid sequences)—were overexpressed in bacteria, purified and exposed to an excess of D₂O. The relative incorporation of deuterium with time was determined by mass spectrometry and the results are shown in Figure 2. EX1 kinetics (illustrative of a cooperative unfolding event, as described in the previous section) were observed for the Itk and Btk SH3 domains, while no evidence of EX1 kinetics was found for the SH3 domains from Txk, Tec, or Bmx. The clear bimodal isotope pattern seen for the Itk SH3 [Fig. 2(A)] revealed an unfolding half‐life (t_1/2) of 6.3 minutes (determined as described in Materials & Methods). Peak‐width analysis26 to monitor EX1 (See Supporting Information Fig. S1), measured at FWHM, yielded a 22.3 Da increase in peak width above baseline for Itk SH3. In contrast, the Btk SH3 domain [Fig. 2(B)] displayed more subtle EX1 kinetics with a similar t_1/2 of 6.8 minutes and a maximum increase in peak width of about 6 Da. For the Tec SH3 domain [Fig. 2(C)], there was detectable peak widening but this was likely not significant as some peak widening occurs as a result of deuterium incorporation itself, due to an isotopic redistribution of labeled protein.26 Our empirically derived requirements for the increase in peak width needed to be considered significant, and therefore indicative of EX1 kinetics, is >4 Da for proteins, and >2 Da for peptides.26 The peak‐width plot of Tec SH3 [Supporting Information Fig. S1(B)] indicated that there might be some EX1 kinetics with a half‐life of approximately 1 minute; however we cannot explicitly state that this is a statistically significant EX1 signature due to the minimal widening observed. The remaining SH3 domains, Txk and Bmx displayed no evidence of EX1 kinetics. Both of these domains were fully deuterated after 10 seconds of labeling, consistent with a lack of stable structure under these conditions. For the sake of comparison, the previous studies14, 26 involving cooperative unfolding in other SH3 domains indicate that Hck SH3 and α‐spectrin undergo the greatest documented increases in peak width for an SH3 domain (measured at FWHM) upon deuteration of 15.4 and 19.1 Da, respectively.

Figure 2.

Raw electrospray mass spectra of deuterium labeled SH3 domains. Labeled from A to E: Itk, Btk, Tec, Txk, Bmx SH3 domains. Each domain was labeled for the time indicated above each spectrum increasing from top to bottom; UND = undeuterated control. All spectra show the +5 charge state, except for Bmx SH3, where the +11 charge state is shown.

Recombinant SH3 protein stability experiments

To independently probe whether or not the Tec‐family SH3 domains in these studies were structured, we employed fluorescence spectroscopy and NMR. For four of the Tec‐family members (Itk, Btk, Tec, Txk), a redshift in the center of each spectrum and a decrease in fluorescence were observed in the presence of 6 M GuHCl, indicative of protein unfolding (Fig. 3). The Itk SH3 domain showed the greatest redshift (6.6 nm) and an intensity decrease of 66%. Btk, Tec, and Txk showed a redshift to a lesser degree, shifting 1.1, 3.9, and 3.1 nm respectively with decreases in intensity of 66, 80, and 68%. In contrast, Bmx showed an increase in intensity upon addition of GuHCl (85%) with no appreciable shift in the center of the spectrum.

Figure 3.

Tryptophan fluorescence spectra for the five Tec SH3 domains. Emission wavelength scans were recorded from 300 to 400 nm using an excitation wavelength of 278 nm. Solid lines indicate spectra obtained under mild buffering conditions (10 mM sodium phosphate buffer, pH 7.1), whereas dashed lines represent spectra obtained under denaturing conditions (10 mM sodium phosphate buffer, 6 M guanidinium chloride, pH 7.1). See Materials and Methods for experimental details.

The Bmx SH3 homology region was unique in that by HX MS and fluorescence spectroscopy, it appeared to have no structure at all. To verify this, we performed NMR measurements of Bmx SH3 alone, in the presence of a binding peptide, and in the context of the Bmx SH2 domain. The ¹H‐¹⁵N HSQC spectra of uniformly ¹⁵N‐labeled Bmx SH3 domain showed poor chemical shift dispersion consistent with that of an unfolded protein [Fig. 4(A)]. Addition of a ten‐fold excess of a proline‐rich peptide ligand to the Bmx SH3 NMR sample caused no significant chemical shift changes in the Bmx SH3 domain ¹H‐¹⁵N HSQC spectrum (data not shown). This result suggests that the recombinant Bmx SH3 domain does not bind the proline‐rich peptide ligand nor does the peptide ligand induce folding of this SH3 domain. In contrast, the Tec family SH3 domains of Itk, Btk, Txk, and Tec and their interaction with proline rich peptides have been studied previously by NMR and the ¹H‐¹⁵N HSQC spectra of all these SH3 domains show chemical shift dispersion consistent with that of folded proteins.32, 33, 34, 35 Comparison of the ¹H‐¹⁵N HSQC spectra of the Bmx SH3 domain with that of the corresponding SH3‐SH2 dual domain [Fig. 4(B)], showed that the residues corresponding to the Bmx SH3 domain still had poor chemical shift dispersion unlike the residues derived from the Bmx SH2 domain which were well dispersed. These results suggested that the Bmx SH3 domain remains unfolded even in the context of the larger Bmx SH3‐SH2 dual domain construct. As there was no observable change in the Bmx SH3 domain spectrum in the context of the larger Bmx SH3‐SH2 domain, the proline rich peptide was not added to the Bmx SH3‐SH2 construct. However, it is certainly possible that the SH3 domain needs both the peptide and SH2 domain to achieve a stable folded structure. Overall, the NMR measurements of Bmx, combined with the fluorescence and HX MS data, strongly indicate that Bmx SH3 is intrinsically unfolded.

Figure 4.

Comparison of the ¹H‐¹⁵N HSQC spectra of uniformly ¹⁵N labeled Bmx SH3 domain and Bmx SH3SH2 dual domain. (A) A ¹H‐¹⁵N HSQC spectrum of uniformly ¹⁵N labeled Bmx SH3 domain (267 μM) shows poor chemical shift dispersion consistent with that of an unfolded protein. (B) An overlay of the ¹H‐¹⁵N HSQC spectra of uniformly ¹⁵N labeled Bmx SH3SH2 dual domain (289 μM, red spectrum) with that of the Bmx SH3 domain (black spectrum) shows that the residues corresponding to the Bmx SH3 domain shows poor chemical shift dispersion unlike the residues corresponding to the Bmx SH2 domain which are well dispersed.

Localizing unfolding in Itk and Btk SH3

In order to localize the EX1 unfolding events that were found in the Btk and Itk SH3 domains, each protein was labeled with deuterium over a range of times, followed by quenching and digestion with pepsin (see Materials and Methods). Individual peptic peptides covered 95.3% of the Itk SH3 sequence and 90.5% of the Btk SH3 sequence (Supporting Information Figs. S2 and S3). Two examples of the raw mass spectra are shown in Figure 5. Both stacked spectra display pronounced peak widening [Fig. 5(A,E)] and sigmoidal‐shaped uptake curves [Fig. 5(B–G)] characteristic of EX1 kinetics. Based on these criteria, EX1 unfolding in both proteins was localized to Itk SH3 domain peptides in the regions E189–Y198 (RT loop/β2) and W208–Y225 (β3/distal loop/β4/3₁₀ helix) and to Btk SH3 domain regions K217–L222 (β1) and L233–F242 (RT loop/β2). The raw mass spectra for the peptides that were used to define the regions of EX1 are shown in Supporting Information Figure S4. These regions of EX1 unfolding were mapped onto the 3D structures, as shown in Figure 5(D,H). For comparative purposes, in Figure 1 we indicate with blue which parts of each SH3 domain (both Tec‐family SH3 domains and others previously tested) showed evidence for EX1 kinetics and localized unfolding. Not surprisingly, the regions from Tec‐ and Src‐family SH3 domains that exhibit EX1 kinetics are in similar locations of the proteins, namely in the RT‐loop and in β3‐distal loop‐β4.

Figure 5.

Localization of cooperative unfolding in Itk and Btk SH3 domains. (A) Electrospray mass spectra for Itk SH3 peptic peptide 209–225 (see Supporting Information Fig. S1 for sequences and uptake plots). Deuterium incorporation curves for regions where bimodal isotope patterns were found for: (B) Itk SH3 residues 209–225; (C) Itk SH3 residues 191–197. The location of the regions shown in (B,C) are shown in panel (D): Itk SH3 (PDB 2RN8), residues 208–225 (red), 189–198 (blue). (E) Electrospray mass spectra for Btk SH3 peptic peptide 233–242 (see supplemental S2 for sequences and uptake plots). Deuterium incorporation curves for regions where bimodal isotope patterns were found for: (F) Btk residues 233–242; (G) Btk residues 217–222. The location of regions shown in (F,G) are shown in panel (H): Btk SH3 (PDB 1AWX): residues 217–222 (green), 233–242 (blue).

Discussion

Our results illustrate that the Tec‐family SH3 domains display a wide range of dynamic properties, including the presence or absence of EX2 and EX1 kinetics, variability in the rate of deuterium uptake, as well as the degree of EX1 unfolding and half‐life. Both Bmx and Txk were heavily deuterated after just 10 seconds of labeling in D₂O (Fig. 2), indicating that these domains are largely unstructured and highly dynamic in solution. Both fluorescence (Fig. 3) and NMR (Fig. 4) studies also indicated that the Bmx SH3 domain was unstructured. In contrast to the other SH3 domains, the Bmx SH3 fluorescence spectrum did not undergo a redshift in the presence of denaturant and the fluorescence intensity actually increased under denaturing conditions. We speculate that the increase in fluorescence intensity upon addition of GuHCl may be due to an electrostatic shielding of charged resides within the Bmx sequence, allowing it to take on a more compact structure36—burying the two side‐by‐side tryptophan residues within the protein. The human Bmx SH3 domain is actually a non‐canonical SH3 domain in that at the N‐terminus there is a repeat sequence (¹⁸⁵ PSSSTTTLAQYDNESKKNYGSQ²⁰⁶) not found in the mouse Bmx SH3 domain. Considering this repeat sequence, the human Bmx SH3 domain may actually encompass residues Leu‐213 to Glu‐265.37, 38 The construct we used encompasses residues Ala‐184 to Asp‐292 and includes this repeat sequence (see supporting information). This additional N‐terminal sequence may be what is responsible for rendering human Bmx SH3 unfolded as having two copies of the first beta1 strand and the entire RT loop sequence could wreak havoc with the folding as this part of the sequence is central to the SH3 fold.

In contrast to Bmx, the Txk SH3 domain sequence is homologous to other well‐folded Tec‐family SH3 domains that were protected from deuteration (Itk, Btk, and Tec). Despite this homology, the Txk SH3 domain was rapidly deuterated yet showed a reduction in fluorescence intensity upon addition of denaturant, suggesting that some residual structure may be present. Previous NMR studies show that the Txk SH3 domain is folded in solution and interacts with its adjacent proline‐rich region.34 These observations raise the question of why the Txk SH3 domain shows such a rapid HX MS labeling profile. The most likely possibility is that the Txk SH3 domain is highly dynamic on the time scale of HX MS labeling (first labeling point of 10 seconds). Alternatively, of the three well‐folded SH3 domains, the highest sequence similarity to Txk is the Tec SH3 domain and so we inspected differences between Txk and Tec amino acid sequences to identify possible reasons for the observed differences in dynamics. Focusing on the hydrophobic side chains that might be most responsible for stabilizing these small domains in their folded form, we find that three hydrophobic residues of Tec: I183, V185 and G236 create contacts between the N‐ and C‐terminus of Tec SH3 (Supporting Information Fig. S5) that might serve to ‘lock’ the domain in a more stable conformation. The corresponding residues in Txk are: Q86, K88 and E139. The hydrophilic nature of these side chains in Txk might mean that the N‐ and C‐termini of Txk are less stably associated thereby conferring a greater degree of breathing or opening of the domain leading to the level of incorporated deuterium that was observed.

NMR structures of three other members of the Tec family in this study [Itk,32, 39 Btk,40 Tec35] show minimal differences in backbone topology (see also Fig. 1). The largest RMSD is 2.1Å across 56 residues between the Itk and Tec SH3 domains (Itk–Btk RMSD = 1.567 Å across 56 residues, Itk–Tec RMSD = 2.079 Å across 56 residues, Btk–Tec RMSD = 1.584 Å across 56 residues). Despite these similarities in structure, the Itk, Btk and Tec SH3 domains demonstrated substantial differences in protein dynamics via HX MS. Not only do these domains show a variation in the degree of unfolding, they also showed differences in both EX1 unfolding half‐lives and residues involved in the unfolding event. The Tec SH3 domain displayed no evidence of EX1 kinetics or partial unfolding. In contrast, regions of localized EX1 unfolding were found for both the Itk and Btk SH3 domains. Of the two, the Itk SH3 domain displayed a much greater degree of cooperative unfolding, where two populations of labeled proteins could be easily distinguished. The Btk SH3 showed only moderate peak widening, and the regions of localized SH3 unfolding (Fig. 5) were distinct from Itk and previously studied SH3 domains. The Itk SH3 domain, with a change in peak width of 22.3 Da (FWHM) and a t _1/2 of 6.3 minutes, is among a unique group of SH3 domains (along with Hck, Lyn, and α‐spectrin) that undergo widespread unfolding. The residues involved stretched across a peptide covering the β1 and β2 strands and included the 3₁₀‐helix, as well as the β2 strand and part of the RT‐loop in a separate part of the protein [Fig. 5(D)]. The residues involved in the EX1 event localized to the β2‐strand/RT‐loop of Itk and overlap structurally with the peptide in Btk involved in EX1 unfolding. Previous NMR experiments showed that the Btk SH3 domain exists in two conformations, where unfolding is achieved by a hydrophobic ring flip on the order of one per second.40 Our data show that Btk SH3 undergoes an unfolding event with a t _1/2 on the order of 7 minutes, which may be independent of this previously identified ring flip mechanism which is on a completely different (1 sec) time scale.

Our studies provide further evidence that partial cooperative unfolding is a conserved dynamic property of many SH3 domains. However, the structural mechanism responsible for this unfolding event is not clear.14 We prepared a phylogenetic tree based on the sequences of all SH3 domains studied by HX MS to date, including the Tec‐family kinase SH3 domains investigated here (Fig. 6). Overall sequence homology is relatively low, despite high and obvious similarities in tertiary structure. For some highly related “sister” proteins such as Lyn and Hck, the exchange dynamics are very similar, with EX1 half‐lives of 20 and 19 minutes, respectively. At the same time, there are several examples of differences in SH3 dynamics despite close proximity in the phylogenic tree. Two pairs of examples are Fyn/Fgr and Yes/Src (each having >70% sequence homology) in which one member undergoes some EX1 kinetics with a half‐life of 5 minutes, while the homologous partner undergoes EX2 exchange kinetics. The sequence homology between the SH3 domains of the Tec members Btk and Itk is 44%, and yet the EX1 dynamics of Itk bears a much closer resemblance to that of Hck (28% homology) or Lyn (27% homology) of the Src‐family in degree of unfolding and peak widening. To complicate things further, the Btk SH3 domain has localized EX1 unfolding in the region of β1 (Fig. 5), which has only been previously observed in the distant relative, α‐spectrin (30% homology). We were unable to monitor any peptides for three residues of Itk SH3 in the β1 area, and so cannot say for certain whether or not that region is participating in EX1 unfolding. Btk also fails to show any significant EX1 folding in the region of β3 and β4 as seen in Itk, Lck, Hck, and Lyn SH3s (Fig. 1). Taken together, secondary structure is not the only factor that contributes to SH3 domain unfolding. As previously speculated,14 tertiary structures driven by side‐chains must play a role in stability, although we are unable to ascertain exactly how.

Figure 6.

Summary of unfolding data for Tec and Src‐family SH3 domains and 4 other select SH3 domains.13 The results are organized into a phylogenic tree (based on the SH3 domain sequence similarity). Members of the Tec‐family and Src‐family are colored pink and green respectively. The degree of EX1 unfolding for each domain is designated unresolved, resolved, or not‐occurring (EX2 only). The bar scale illustrates the half‐life of unfolding on a scale of 0–20 minutes, with the actual value next to the name of the protein. See Figure 1 for corresponding sequences and specific residues involved in EX1 unfolding.

HX MS has been used to assess SH3 domain interactions with potential ligands as changes in EX1 kinetics.41 One of the best‐studied examples of this approach is the interaction of the Hck SH3 domain with the HIV‐1 accessory protein, Nef.12 HIV‐1 Nef binding substantially slows the unfolding half‐life of the Hck SH3 domain,5, 14 an event that correlates with Nef:Hck complex formation and kinase activation. Nef‐dependent activation of Hck contributes to HIV‐1 replication, in that selective inhibitors of this pathway interfere with viral replication in vitro.42, 43, 44 More recent studies have suggested a similar role for the Tec‐family kinase Itk in HIV‐1 pathogenicity as well, although the connection between the virus and Itk were not initially apparent.45 Subsequent cell‐based bimolecular fluorescence complementation (BiFC) assays revealed selective interaction of HIV‐1 Nef with the Tec‐family kinases Itk, Btk, and Bmx.46 Interaction of Nef with these Tec‐family members required their SH3 domains, and was conserved across a wide range of Nef alleles. Nef was also shown to induce Itk activation in cells, and pharmacological inhibition of this pathway blocked HIV infectivity and replication in a Nef‐dependent manner. Based on these results, we explored whether Nef influenced the dynamics of Tec‐family kinase SH3 domains by HX MS. Surprisingly and in contrast to what was has been seen with the Src‐family SH3 domains,47 co‐incubation with recombinant HIV‐1 Nef did not detectably affect Btk SH3 domain dynamics in vitro, despite the strong interactions of these domains in cells (data not shown). These findings suggest that the membrane may have a critical role in mediating this interaction, as the fluorescent complementation signal observed in the cell‐based assay localized almost exclusively to the cell membrane. Furthermore, the cell‐based studies included not only the SH3 domain, but also the N‐terminal PH‐TH region, which may also have an important influence on SH3 domain function.

This investigation of the Tec SH3 domains represents an expansion upon our previous investigations of dynamics in Src‐family and related SH3 domains by HX MS. While the motions of these domains have proven invaluable as an assay for both inter‐ and intramolecular binding, we are still uncertain as to the exact biophysical reasons for the unfolding phenomena. Mutational analysis might hold the key. We earlier found that a single point mutant of HckSH3, W93A in the n‐Src loop, maintained both the locations and magnitudes of the partial unfolding behavior but altered the rate to be 10 times faster (2–3 instead of 19 minutes) when compared to the wild‐type protein.14 In this case, the mutation occurred at a W residue that is strictly conserved across all the SH3 domains this far studied by HX MS. We suspect there must be a key network of residues and/or hydrophobic packing within the SH3 domain fold that dictates folding, dynamics, and exchange behavior. Further investigations of site‐directed mutants using these SH3 domains could be used to tease out the intricacies of partial unfolding.14

Materials and Methods

Protein expression and purification

All Tec family SH3 domains were human except Itk, which was mouse. The boundaries of each construct and the exact sequence are found in the Supporting Information. Recombinant SH3 domains of Btk, Txk, Tec, and Bmx were overexpressed in and purified from Escherichia coli as C‐terminal His‐tagged constructs as previously described.48 Itk SH3 was overexpressed in and purified from Escherichia coli (strain BL21 (DE3) as a GST‐tagged fusion protein as previously described.49 The purity (>95%) and correct mass were determined by electrospray mass spectrometry.

Fluorescence spectroscopy

Fluorescence spectra were obtained with a Cary Eclipse fluorescence spectrometer at proteins at concentrations of 10 μM in 10 mM sodium phosphate buffer (pH 7.1) for the native state and 10 mM sodium phosphate buffer plus 6 M guanidinium chloride (pH 7.1) for the denatured state. A 400 μL quartz cuvette was used for Itk, Btk, Txk, Tec and a 50 μL quartz cuvette for Bmx, both with 1 cm path length. Emission wavelength scans were recorded from 300 to 400 nm using an excitation wavelength of 278 nm. For each protein in each buffer, three individual scans were recorded and then averaged. Blanks for each buffer system were subtracted from each averaged spectrum to correct for background fluorescence. The centers of spectral mass were calculated for the fluorescence emission wavelength scans according to the equation:

$$〈v_i〉=\frac{{\sum }_i{F}_i{v}_i}{{\sum }_i{F}_i}$$

where F _i is the fluorescence emitted at wavenumber v _i, and the summation is carried out over the full range of the emission spectrum.

NMR

The human Bmx SH3 domain (D183‐E287) and Bmx SH3SH2 dual domain (D183‐S393) were cloned into pGEX 4T‐1. All constructs were verified by sequencing at the Iowa State University DNA synthesis and sequencing facility.

Uniformly ¹⁵N labeled isolated Bmx SH3 domain or Bmx SH3SH2 dual domain were produced in E. coli BL21(DE3) cells as described previously.49 Briefly, the cultures were grown in modified M9 minimal medium containing 1g/L ¹⁵N‐enriched NH₄Cl (Cambridge Isotope Laboratories) to an OD_600nm of 0.8 and induced with 1 mM IPTG at 18°C overnight. The cell pellet was re‐suspended in lysis buffer (50 mM KH₂PO₄ pH 7.2, 75 mM NaCl, 2 mM DTT, 0.02% NaN₃ and 1mg/ml lysozyme) and stored at −80°C. The cell pellets were thawed along with 3000 Units DNase I (Sigma) and 1 mM PMSF. The cleared supernatant was loaded onto a Glutathione agarose resin (Sigma). The column was washed with 200 ml of wash buffer (50 mM KH₂PO₄ pH 7.2, 75 mM NaCl, 2 mM DTT and 0.02% NaN₃) and the proteins were eluted with 150 ml of elution buffer (50 mM KH₂PO₄ pH 7.2, 75 mM NaCl, 2 mM DTT, 0.02% NaN₃ and 10 mM glutathione). The GST‐fusion proteins were concentrated, and then cleaved with Thrombin (Calbiochem) at room temperature for 2 h. The cleaved proteins were passed over the Glutathione resin (to remove cleaved GST) and then further purified on a Sephacryl S‐100 HR gel filtration column (GE Healthcare). The purified proteins were concentrated in NMR buffer which consisted of 50 mM KH₂PO₄ pH 7.2, 75 mM NaCl, 2 mM DTT and 0.02% NaN₃.

NMR spectra were collected at 298 K on a Bruker AVII 700 spectrometer equipped with a 5 mm HCN z‐gradient cryoprobe operating at a ¹H frequency of 700.13 MHz.

Deuterium exchange reactions

Stock solutions of Tec, Txk and Bmx SH3 domains were each 35 pmol/μL in 100 mM NaCl, 20 mM Tris, 3 mM DTT buffer, (pH: 8.32). Stock solutions of Itk and Btk SH3 were pmol/μL and 20 pmol/μL, respectively, in the same buffer. Deuterium exchange was initiated by dilution of stock solutions 18‐fold with labeling buffer (100 mM NaCl, 20 mM Tris, 3 mM DTT buffer (pD: 8.321), D₂O, 21°C). An aliquot was removed from the exchange reaction and labeling was quenched by adjusting the pH with an equal volume of 150 mM potassium phosphate quench buffer (pH: 2.47), H₂O. Quenched samples were either immediately frozen on dry ice and stored at −80°C until analysis (Tec, Txk and Bmx SH3 domains) or injected immediately after quenching (Itk and Btk SH3 domains). For analysis of the pepsin fragments of Itk and Btk SH3, the stock solution were only diluted 15‐fold with labeling buffer and were quenched as described above.

Mass spectrometry and data analysis

Labeled proteins were rapidly thawed at 0°C and a total of 120 pmol of Tec, Txk, Btk, Bmx, or 18 pmol of Itk were injected onto a C‐18 protein trap and desalted with 4× the injection volume of H₂O containing 0.05% TFA. The proteins were eluted into the mass spectrometer using a 15–98% ACN (0.05% TFA) step in 5.5 minutes at a flow rate of 50 μL/min with a Shimadzu prominence HPLC (LC‐20AD). The HPLC was performed using protonated solvents (H₂O), causing the removal of deuterium from the side chains and amino/carboxyl termini that exchange faster than backbone amide hydrogens.31 The C‐18 trap, associated tubing, and solvent feedstock bottles were submerged in an ice bath to reduce back exchange.31 Mass spectra were obtained using a Waters Micromass LCT Premier Mass Spectrometer with a standard electrospray source, capillary voltage of 3200 V and a cone voltage of 32 V. The deuterium levels were not corrected for back exchange and are reported as relative.50 The relative deuterium incorporation of each SH3 domain analyzed was determined by subtracting the undeuterated mass from the mass of the protein at each labeling time point.

For pepsin digestion experiments of Itk (60 pmol) and Btk SH3 (66 pmol), deuterium exchange was performed as described above. Quenched samples were injected onto a Waters nanoACUITY UPLC with HDX technology51 chilled to 0°C. Digestion was accomplished using a immobilized pepsin column.52 Peptides were separated using an ACUITY 50 mm $\times$ 1 mm HSS T3 1.8 μm C‐18 column in 6 min by a 5‐35% ACN‐H₂O (0.1% FA) gradient with a flow rate of 60 μL/min. Peptides were identified using exact mass and MS^E. All deuterium incorporation was processed using DynamX software (Waters).

To determine peak widths, mass/charge spectra were transformed to a mass‐only scale using MassLynx (Waters). Peak width was measured at full‐width half maximum (FWHM) for each mass spectrum (Fig. 2) by hand for each time point and plotted against the deuterium labeling time.26 For those members that show significant peak widening (Itk, Btk), the intersection method53 was used to determine the unfolding half‐life (t_1/2). In the plots shown in Figure 3, for Btk and Itk the average values between two experiments were plotted; data for only a single experiment are shown for the other domains. Because two distinct isotopic distributions were observed for Itk SH3, the half‐life value could also be obtained by plotting the natural log of the percent of folded molecules versus time.12, 13, 42 The area of the peak representing the folded form [lower mass envelope in Fig. 2(A)] and the total area of both peaks were measured using MassLynx. Two independent labeling experiments were performed on Itk SH3, the unfolding half‐life determined and the values averaged for the 6.3 min value reported.

Sequence alignment and phylogeny

The alignment in Figure 1 was made using Clustal Omega from each SH3 sequence (see Supporting Information) as found in UniProtKB and the ΔSH3 sequence of Bmx.37, 38 As the sequence of Amphiphysin II is longer in the n‐Src and distal loops, the extra amino acids ⁵⁴⁹NPEEQ⁵⁵³ and ⁵⁶⁵DWNQHKEL⁵⁷² were removed to maintain structural continuity of the alignment. The phylogenic tree in Figure 6 was created using Phylogeny.fr and TreeDyn54, 55, 56 based on SH3 sequence homology by loading the SH3 domain sequences into Clustal Omega and then rendering the results. The full SH3 domain sequence of each protein (as indicated in UniProtKB) was used in creating the tree except for Bmx which contains a non‐canonical SH3; the full Bmx construct (See Supporting Information) was used in making the tree.

Supporting information

Supporting Information