Mona/Gads SH3C Binding to Hematopoietic Progenitor Kinase 1 (HPK1) Combines an Atypical SH3 Binding Motif, R/KXXK, with a Classical PXXP Motif Embedded in a Polyproline Type II (PPII) Helix*

Hematopoietic progenitor kinase 1 (HPK1) is implicated in signaling downstream of the T cell receptor. Its non-catalytic, C-terminal half contains several prolinerich motifs, which have been shown to interact with different SH3 domain-containing adaptor proteins in vitro. One of these, Mona/Gads, was also shown to bind HPK1 in mouse T cells in vivo. The region of HPK1 that binds to the Mona/Gads C-terminal SH3 domain has been mapped and shows only very limited similarity to a recently identified high affinity binding motif in SLP-76, another T-cell adaptor. Using isothermal titration calorimetry and x-ray crystallography, the binding of the HPK1 motif to Mona/Gads SH3C has now been characterized in molecular detail. The results indicate that although charge interactions through an RXXK motif are essential for complex formation, a PXXP motif in HPK1 strongly complements binding. This unexpected binding mode therefore differs considerably from the previously described interaction of Mona/Gads SH3C with SLP-76. The crystal structure of the complex highlights the great versatility of SH3 domains, which allows interactions with very different proteins. This currently limits our ability to categorize SH3 binding properties by simple rules.

Hematopoietic progenitor kinase 1 (HPK1 1 ; Human Genome Organization gene symbol MAP4K1) is a member of the germinal center kinase (GCK) family within the large superfamily of STE20 and p21-activated kinases (Refs. 1 and 2). As its name indicates, HPK1 expression is prominent in hematopoietic cells. Initial cloning reports implicated HPK1 in the activation of stress kinases (SAPKs/JNKs) (3,4). Since then, HPK1 activity has been associated with lymphocyte antigen receptor signaling (BCR, TCR; Refs. [5][6][7][8][9] and in signaling induced by transforming growth factor-␤ (10,11). It has also been shown that physiological concentrations of prostaglandin E2 raise HPK1 kinase activity in T cells and myeloid cells by a yet undefined mechanism, possibly leading to the suppression of fos gene transcription (12). In addition, a single report suggests a role for HPK1 in erythropoietin-induced signaling of two cell lines (13), but this finding remains unconfirmed for primary erythroid progenitor cells.
HPK1 is presumed to act as a MAP4K (MAP kinase kinase kinase kinase). Its potential substrates include the mixed lineage kinase MLK3, the mitogenic kinase MEKK1, the transforming growth factor-␤ activated kinase TAK1 (8), and also c-Jun. 2 Furthermore, HPK1 is activated by caspase cleavage (14,15) and functions as a regulator of NFB signaling (15,16).
Despite these reports, the in vivo functions and signaling pathways of HPK1 remain only marginally defined. HPK1 (Ϫ/Ϫ) mice have been generated by targeted gene disruption but exhibit no obvious phenotype. 3 This may not be entirely surprising, given that HPK1 is just one member of a family of similar kinases with hematopoietic expression (1,2). Close relatives with overlapping expression patterns include GCK, kinase homologous to SPS1/STE20 (syn. GCKR), HPK1/GCKlike kinase (named NIK in mice) and germinal center-like kinase. The targeted disruption of multiple GCK family kinase genes in mice may eventually shed more light on HPK1 functions.
Physical interactions of HPK1 with other intracellular proteins in vitro and in vivo have been reported in numerous studies. The T-cell adaptor SLP-76 and its B-cell relative BLNK, also known as SLP-65 or BASH, are functionally important interaction partners of HPK1 and crucial in the signal transmission of TCR and BCR, respectively (5,7,17,18). After receptor activation, SLP-76 and BLNK bind via their SH2 domains to the phosphorylated tyrosine 397 of human HPK1 (18). Other protein complexes are formed via four proline-rich motifs in the non-catalytic region of HPK1. SH3 domain-containing adaptors, including Crk, CRKL, Grb2, Nck, Mona/Gads (syn. Grap-2, GrpL, Grf40, GRID), and HIP-55 have been shown to bind HPK1 (5, 6, 9, 19 -22). Of these, Mona/Gads * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The binding to HPK1 was clearly demonstrated in mouse T cells by co-immunoprecipitation of the endogenous proteins (5). The same study implicated the fourth proline-rich motif in the non-catalytic part of HPK1 as the binding region for Mona/ Gads but did not provide details regarding the HPK1 residues crucial for high affinity binding. We and others have recently shown that the SH3C domain of Mona/Gads can interact with nanomolar affinity with a SLP-76 motif, which lacks a typical PXXP motif (23,24).
In this study, the interaction of HPK1 with Mona/Gads was initially analyzed by isothermal titration calorimetry (ITC). Subsequently, a suitable fragment of HPK1 in complex with Mona/Gads SH3C was crystallized and studied by x-ray crystallography. The structure of the Mona/Gads SH3C-HPK1 peptide complex was solved to 1.5-Å resolution. The data show that, different from the interaction of Mona/Gads SH3C with SLP-76, a PXXP motif, which is part of a polyproline type II helix, is essential for complex formation of murine Mona/Gads SH3C and HPK1. Alanine scanning and crystal contacts define the residues PXVPXRXXK as key sites for complex formation.

EXPERIMENTAL PROCEDURES
Peptide synthesis as well as expression and purification of murine Mona/Gads SH3C were done as described previously (24). The pGEX-2T expression vector for GST-Grb2 SH3N (human amino acids 3-57) is a gift of O. Janssen. Expression and glutathione-Sepharose purification of this fusion protein was done under the same conditions as for Mona/ Gads SH3C. The functionality of the construct had been previously confirmed in precipitation experiments using Sos-containing cell lysate.
ITC was done essentially as described previously (24). Briefly, ITC measurements were performed on a VP-ITC MicroCalorimeter (Micro-Cal, Northhampton, MA). 1.43 ml of 0.05 mM affinity-purified GST-SH3 domain in ITC buffer (25 mM HEPES-KOH, pH 7.5, 100 mM potassium acetate, 5 mM magnesium acetate) was clarified for 10 min at 20,800 ϫ g and degassed for 10 min with Thermovac (MicroCal) before being transferred into the sample chamber. Synthetic peptides were diluted to 0.5 mM in ITC buffer, clarified, and degassed, and ϳ300 l was loaded into the syringe. During the measurements, 5-and 10-l aliquots of peptide solution were titrated once and 28 times, respectively, into the sample chamber at an equilibrium temperature of 25°C. The heat of dilution was negligible. Total heat generated and K d values were calculated in ORIGIN (V5.0).
For crystallization, the Mona/Gads SH3C construct at 15 mg/ml was mixed with the 16-residue HPK1 peptide P5 at a 1:3 molar ratio. Crystals were obtained with the sitting drop vapor diffusion method at 20°C overnight under equilibration conditions of 0.1 M HEPES, pH 7.5, 2 M (NH 4 ) 2 SO 4 (Index Screen, Hampton Research). These were cryoprotected through transfer into mother liquor containing 20% glycerol and vitrified. Single wavelength data were collected on BM14 at the European Synchrotron Radiation Facility, Grenoble, France (see also Table  II). The structure was solved by molecular replacement using AMoRe (25) with 1OEB.pdb as the search model. The atomic model was built using the program O (26) and refined with Refmac (27); see Table II for details. Generation of structure and density representations, as well as calculation of buried surface areas and surface representations, were done as described previously (24).

RESULTS AND DISCUSSION
Characterization of the Sequence Motif in HPK1 Required for Mona/Gads SH3C Binding-Until now, the SH3C domain of Mona/Gads is the only domain of this adaptor known to mediate protein complex formation independent of tyrosine phosphorylation. Previously, we have shown that the Mona/Gads SH3C domain binds with very high affinity to a PX 2 DRX 2 KP motif from SLP-76 and that all of the specified residues contribute considerably to complex formation (24). However, HPK1, which also binds to Mona/Gads independently of tyrosine phosphorylation, lacks such a PX 2 DRX 2 KP motif (Fig. 1A). Instead, the implicated region in HPK1 contains only an RXXK motif in the mouse protein and a KXXK motif in the human protein. Additionally, HPK1 proteins have a cluster of proline residues N-terminal to the R/KXXK motif, which is not present in SLP-76, and could possibly contribute to Mona/Gads SH3C binding. To test this hypothesis, peptides corresponding to 26 amino acids of the relevant region in mouse and human HPK1 (Table I, P1 and P2) were analyzed for binding by ITC. Both peptides bound to Mona/Gads SH3C with affinities of 4 and 21 M, respectively. These are typical affinities for known SH3 domain-peptide interactions.
The essential Mona/Gads SH3C binding region in mouse HPK1 was further mapped by analyzing shorter overlapping HPK1 fragments (Table I, P3-P8). A representative example of an ITC measurement is shown in Fig. 1B. P7, a 14-amino-acid motif, was the shortest sequence with an affinity significantly below 10 M. Further truncation by 3 amino acids (P8) resulted in a 4-5-fold decreased affinity. An alanine scan through P8 subsequently defined the motif PXVPXRX 2 K as essential for mouse HPK1 binding to Mona/Gads SH3C (Table I,   The dashed box indicates the crucial arginine in SLP-76, which is also present in mouse HPK1 but replaced by a lysine in human HPK1. HPK1 and SLP-76 residues known to make contacts with the Mona/Gads SH3C domain in crystal structures are indicated by arrowheads. Note that the proline residue, which aligns in all four peptide primary structures (depicted in red), is not involved in contacting the Mona/Gads SH3C in the crystallized SH3-peptide complexes. B, representative ITC measurement of Mona/Gads SH3C binding to a 16-amino-acid HPK1 peptide (P5; sequence in Table I).
As discussed above, HPK1 has several proline-rich stretches in the non-catalytic part of the protein. In one study, using human HPK1, a region different from the site implicated in the mouse HPK1 has been suggested as the preferential binding site for Mona/Gads SH3C (6). A corresponding peptide (P20) was therefore also tested for binding to Mona/Gads SH3C by ITC. However, the affinity of this peptide was considerably lower than that of P1, and no further studies were performed with it.
The Mona/Gads SH3C binding motif identified in mouse HPK1 clearly overlaps with the PXXPXR core consensus motif for binding of the N-terminal SH3 domain of the Mona/Gadsrelative Grb2 as defined previously in several studies (Ref. 28 and references therein). Grb2 appears to be ubiquitously expressed and might therefore compete with Mona/Gads for some binding partners. Consequently, the affinity of the Grb2 SH3N domain for two HPK1-derived peptides (Table IB, P5 and P8) was also determined. Surprisingly only marginal binding was detected, although an ITC measurement with another PXX-PXR-containing peptide (Table IB, P21) clearly showed that the expressed Grb2 SH3N domain is functional. The significant differences between the HPK1 sequence that binds to Mona/ Gads SH3C and the well characterized SH3C binding motif in SLP-76 prompted us to further analyze the Mona/Gads SH3C-HPK1 peptide interaction by x-ray crystallography.
Structure of the Mona/Gads SH3C Domain in Complex with an HPK1-derived Peptide-Diffracting crystals were obtained from a complex of Mona/Gads SH3C with HPK1-derived peptide P5. Data and refinement statistics are summarized in Table II. The structure of the complex was solved to 1.5-Å resolution by molecular replacement. A schematic overview of the SH3 domain in complex with the HPK1 peptide is shown in Fig. 2. The structure of the SH3C domain itself remains unaffected by HPK1 peptide binding, retaining all the structural characteristics described previously (24). The typical ␤-barrel SH3 domain presents the same binding pocket to the HPK1 peptide as in the case of the SLP-76 peptide (stereo view shown in Fig. 3A), although the interactions between domain and peptide in this structure vary considerably from those established in the previous complex (24). Sixteen structured residues comprise the HPK1 peptide in its bound conformation (distinguished from the SH3 domain residues by * in the subsequent text). The distinct structural characteristics within the peptide are a 3 10 helix from residues Arg-9* to Lys-12* and a polyproline type II (PPII) helix of residues Gln-2* to Pro-7* (Figs. 2 and 3B). Although the 3 10 helix is also present in the high affinity SLP-76 peptide described previously, the PPII helix within the HPK1 peptide is clearly novel for a Mona/Gads SH3C interaction.
A multitude of interactions defines the HPK1 peptide-Mona/ Gads SH3C domain interface. Starting from the N terminus of the HPK1 peptide (P5), the first 3 residues do not contribute to binding, although Pro-3* engages in a very weak hydrophobic interaction with Tyr-8. Pro-4* appears to be the first residue vital for peptide docking (Fig. 3C). The pyrrolidine ring of this residue rests in a hydrophobic groove formed by Tyr-52 and Tyr-8 stacking against the phenyl ring of the former and positioned at a 90°angle to the latter. This interaction is further reinforced by a hydrogen bond from the backbone carboxyl oxygen of Pro-4* to the hydroxyl of the Tyr-52 side chain. Leu-5* exhibits only a single hydrogen bond docking the peptide main chain to N␦ 2 of Asn-51.
The next hydrophobic groove along the path of the HPK1 peptide on the domain surface is formed by residues Tyr-52, Pro-59, Phe-10, and Trp-36. Tight hydrophobic interactions in this region maintain the side chains of Val-6*, and to a lesser degree Pro-7*, firmly locked in place, making their presence indispensable for tight peptide binding. The pyrrolidine ring of Pro-8* stacks against the aliphatic portion of the Glu-11* side chain. Hydrogen bonds from the backbone oxygen of Pro-8* to the nitrogen of the indole ring of Trp-36 and the backbone amine of Glu-11* dock the peptide main chain onto the SH3 surface and onto itself, stabilizing the 3 10 helix of P5, which emerges past this point.
The charged interactions of amino acids Arg-9* and Lys-12*, which flank the 3 10 helix, involve a tight network of hydrogen bonds with residues of the Mona/Gads SH3C. Specifically, N⑀ and N 1 of Arg-9* form hydrogen bonds to O⑀ 1 and O⑀ 2 of Glu-17, and Lys-12* N forms hydrogen bonds to Glu-14 O⑀ 1 , Glu-17 O⑀ 2 , and Asp-16 O␦ 1 in a manner identical to that observed in the binding of the SLP-76 peptide to the same interaction surface (24). In addition, Lys-12* further stacks against the indole ring of Trp-36. Lys-10* does not contribute to SH3 domain binding at all, whereas the aliphatic portion of the Glu-11* side chain stacks between Lys-16 and Pro-8*. All residues of the 3 10 helix contribute hydrogen bonds toward the formation of the secondary structure element itself. The 4 C-terminal residues of P5 do not participate in protein docking and have no direct interaction with any of the Mona/Gads residues in the vicinity. The observed intra-peptide interactions therein only provide structure to this end of the peptide.
Data obtained from binding experiments with SH3-and HPK1-derived Ala mutant peptides in solution (Table I, ITC data) and the contact points between SH3 and HPK1 peptide identified in the crystal structure (summarized in Fig. 1A, indicated by arrows) agree in clearly delineating the residues PXVPXRXXK in mouse HPK1 as important for Mona/Gads SH3C binding. This motif differs considerably from the previously identified contact motif PXIDRXXKPXL derived from the SLP-76 protein (24), although the RXXK core motif is present in both mouse proteins. Nevertheless, it is evident from the human HPK1 sequence (Fig. 1A) that the arginine can be substituted with a lysine, leaving only a single lysine residue as a strictly conserved position. It is therefore not possible to effectively predict additional candidate binding partners of Mona/Gads SH3C.
Comparison of the Mona SH3C-HPK1 Peptide Complex with Other SH3 Domain Complexes-We have previously reported the atomic structure of a SLP-76-derived peptide in complex with Mona/Gads SH3C (24). The HPK1-and SLP-76-derived SH3C binding peptides both share the 3 10 helical secondary structure, and accordingly, the respective RXXK motifs are positioned in similar spatial arrangements. Strikingly, although the SLP-76 peptide lacks a PPII helical segment, its C-terminal region occupies approximately the same space as the PPII helix on the hydrophobic surface region of the SH3C (Fig. 3A). The buried surface areas at the SH3-peptide inter- where h is the unique reflection index, I j,h is the intensity of the symmetry-related reflection, and ͗I h ͘ is the mean intensity.  faces are quite similar, 1048 Å 2 for the bound HPK1 peptide as compared with 1143 Å 2 for the complexed SLP-76 peptide and therefore do not provide a simple explanation for the considerable differences in binding affinity of the SLP-76 versus the HPK1 peptide (0.181 and 2.4 M, respectively; Table I, P5, and Ref. 24. Stable conformational characteristics of the HPK1 peptide can be identified by analysis of the comparative mean square atomic displacement values (͗u 2 ͘), which provide an indication of the positional stability of each atom in the crystal structure (Fig. 4A). Temperature factors (B ϭ ͗u 2 ͘8 2 /3) of P5 atoms fluctuate between 9.6 and 30.9 Å 2 . This variability is introduced by the terminal residues of the peptide, which are not stabilized by binding to any docking regions on the face of the Mona/Gads domain. The core of P5, defined by the bulk of the two secondary structure arrangements present, one 3 10 and one PPII helix, is extremely stable. Residues 4* to 12*, which are directly involved with SH3 domain binding, exhibit average B values for main chain atoms of 12.6 Å 2 with the side chain atoms showing only marginally higher values averaging 13.8 Å 2 .
Although a direct comparison with the previous Mona/Gads SH3C complex with an SLP-76 peptide is not possible due to variability in crystallographic data quality and crystal lattice environment, we can normalize the average B values for the atoms of the peptides implicated in SH3C binding to a common mean value. Comparative values generated thus suggest that the binding stability of the core elements of the respective peptides is similar. However, whereas most residues of the SLP-76 peptide contribute to SH3C binding, terminal residues of P5, amounting to a third of the HPK1 residues present, do not participate in binding. Moreover, the relative stability of the PPII region of P5 cannot be fully accounted for by an induced fit on binding SH3C since its N-terminal part does not participate in complex formation. The PPII element appears, rather, to be inherently stable irrespective to SH3C binding. Fig. 4B shows the large degree of overlap but also some  (P5; right). The different colors are a measure of the flexibility of a given amino acid region within the crystallized complex, ranging from very stable positioning (dark blue) to moderate flexibility (red), with varying intermediates in different shades of blue, gray, and light red (cutoff set at 25 Å 2 ). The crucial RXXK motifs point to the left in both peptides and extend out of the green shaded boxes, which indicate the 3 10 -helical region within the peptides. The PPII helix, which is only present in the HPK1 peptide, is indicated by the pale yellow box. B depicts amino acids of the Mona/Gads SH3C, which are contacted by the HPK1 and/or SLP-76 peptides. Residues involved in both SH3-peptide complexes are shown in blue. Residues that contact the SLP-76 but not the HPK1 peptide are shown in green, and the single residue that is contacted only when the HPK1 peptide is bound is drawn in orange.
differences in the contacts that the Mona/Gads SH3C makes with the HPK1 and SLP-76 peptides, respectively. Fourteen amino acid residues of the SH3 domain contribute in both cases to peptide binding. However, there are also three contacts to residues within the Mona/Gads SH3C, which are unique to the interaction with the SLP-76 peptide (Leu-45, Gly-46, Thr-38). These are part of two of the characteristic SH3 domain ␤-strands, which contribute to the ␤-barrel structure. Binding of the HPK1 peptide involves only one SH3 domain residue, Ser-32 within the n-Src loop, which is not involved in the binding of the SLP-76 peptide.
SH3 domains are highly versatile protein interaction modules. Peptide binding can occur via distinct regions of their domain surfaces (discussed in Ref. 29), but a particular SH3 domain surface area can also accommodate rather distinct peptides. Due to space limitations, only two selected examples of other SH3 domain-peptide interactions are compared here with the Mona/Gads SH3C complexed with the HPK1 peptide. The first is the interaction of the SH3 domain from the endocytosisregulating adaptor protein STAM2 with a peptide derived from the ubiquitin peptidase UBPY (also known as USP8), which binds with moderate affinity (27 M;Ref. 30). We and others have previously shown that Mona/Gads SH3C can bind to the UBPY protein and RXXK-containing peptides derived from UBPY in vitro (23,24), but we have failed to detect a complex of the two proteins within cells. The RXXK-containing UBPY peptide was very recently co-crystallized with the STAM2 SH3 domain. SH3 domain-based superposition of the complexed HPK1 peptide with the shorter UBPY peptide (1UJO.pdb) shows that not only do both peptides contain a 3 10 helix in the RXXK region but that they also occupy overall similar positions on the respective SH3 domain surfaces (Fig. 5A, SH3 domains are omitted for clarity). Why UBPY fails to bind to Mona/Gads SH3C in vivo is not self-evident from a comparison of the crystal structures and needs to be investigated through biochemical and biological studies. The third superimposed peptide is the Sos peptide structure as determined in complex with Grb2 SH3N (1GBQ.pdb). As mentioned above, the Sos peptide contains a PXXPXR motif that is crucial for Grb2 SH3N binding. Such a motif is also found in the HPK1 peptide, but its binding to Grb2 SH3N is nevertheless marginal (Table I). Again, comparison of the bound HPK1 peptide to the Sos peptide shows a high degree of similarity with respect to peptide orientation relative to the SH3 domains and also secondary structure in the proline-rich segments. However, the comparison of the structure of Grb2 SH3N in complex with a Sos peptide and the Mona/Gads SH3C-HPK1 peptide structure provides a possible explanation for the low affinity binding of RXXK motifs to the Grb2 SH3N domain. Detailed views of the relevant regions are shown in Fig. 5, B and C. In the Grb2 SH3N structure, both O⑀ 1 and O⑀ 2 of Glu-16 (Glu-17 Mona ) form hydrogen bonds to O␥ 1 of Thr-12 (Leu-13 Mona ). Therefore Grb2 Glu-16 can no longer bind to the incoming arginine of the RXXK motif in the same way as its equivalent Mona/Gads Glu-17 can. The resulting displacement of the arginine to farther along the binding groove forces it to associate with the next available residue, Asp-15 (Asp-16 Mona ), thus rendering the latter unavailable for binding to the lysine of the RXXK motif. Therefore the affinity of this interaction is severely impaired. Furthermore, Ala-13 (Glu-14 Mona ) substitutes for the third hydrogen-bonded partner of the N atom of the lysine within RXXK, although that should not greatly affect binding potential, as shown by the high affinity Mona/Gads SH3C-SLP-76 complex (1OEB.pdb), in which both scenarios are present (24).
Signaling Specificity Generated by SH3 Domain Interactions-Our results, combined with those of many previous studies, show that simple one-to-one SH3 domain interactions with short motifs of associating proteins can generate a considerable degree of binding selectivity. However, with more than 800 SH3 domains in mammalian genomes, the building of SH3 domain-mediated complexes within cells further relies on several other mechanisms to generate timely and functionally appropriate protein interactions. These include obvious means, such as temporal and spatial separation of the expression of potential binding partners, but go far beyond. For example, posttranslational modification of proteins can affect the binding of SH3 domains directly or indirectly (Refs. 31-34 and references therein). Moreover, protein complexes based on the simultaneous interaction of several SH3 and/or other domains do not only allow dramatic increases in binding affinities but may also generate a great degree of binding selectivity.
In the case of Mona/Gads, several such multiprotein complexes could be hypothesized on the basis of its known proteinprotein interactions. For example, after activation of the T-cell receptor, adaptor proteins such as LAT become phosphorylated on multiple tyrosine residues. This could allow simultaneous docking of several Mona/Gads molecules onto a single LAT molecule via their SH2 domains, possibly bringing into close proximity several molecules of HPK1 or SLP-76 that are bound to the Mona/Gads SH3C domains. Alternatively, since HPK1 and SLP-76 are also known to interact (by binding of the SLP-76 SH2 domain to a specific phosphorylated tyrosine on HPK1; Ref. 18), it is conceivable that a complex consisting of one molecule of HPK1 bound to one molecule of SLP-76 and two Mona/Gads molecules may form on LAT in activated in cells. Additional stabilization of such a complex could be achieved through ion-dependent dimerization of Mona/Gads SH3C domains (24). The potential heterodimerization of SLP-76 and HPK1 on LAT via two Mona/Gads molecules could have interesting implications for signal integration by molecular crosstalk. However, in-depth biochemical and biological studies are needed to clarify the actual composition of Mona/Gads-mediated signaling complexes in T-cells and other cells.
Our results further confirm that there is not a set of simple rules that can be applied to genomes of great complexity, such as those of mammals, to reliably predict biologically relevant SH3 domain interactions. Motif scanning approaches are very unlikely to single out functionally important homologies that are based on only 2 amino acid residues such as the crucial basic R/KXXK motif shared between the Mona/Gads SH3C binding sites in the HPK1 and SLP-76 proteins. The challenge is even greater when considering subfamilies of SH3 domains with considerable homology, such as Grb2 SH3C, Mona/Gads SH3C, and Stam2 SH3 (30). Simple extrapolation of results from one species to another may also be limited at times, given existing sequence differences, for example, between the relevant regions in mouse and human HPK1 (Fig. 1A).