Recruitment of phospholipase Cγ1 to the non-structural membrane protein pK15 of Kaposi Sarcoma-associated herpesvirus promotes its Src-dependent phosphorylation

Kaposi Sarcoma-associated herpesvirus (KSHV) causes three human malignancies, Kaposi Sarcoma (KS), Primary Effusion Lymphoma (PEL) and the plasma cell variant of multicentric Castleman’s Disease (MCD), as well as an inflammatory cytokine syndrome (KICS). Its non-structural membrane protein, pK15, is among a limited set of viral proteins expressed in KSHV-infected KS tumor cells. Following its phosphorylation by Src family tyrosine kinases, pK15 recruits phospholipase C gamma 1 (PLCγ1) to activate downstream signaling cascades such as the MEK/ERK, NFkB and PI3K pathway, and thereby contributes to the increased proliferation and migration as well as the spindle cell morphology of KSHV-infected endothelial cells. Here, we show that a phosphorylated Y481EEVL motif in pK15 preferentially binds into the PLCγ1 C-terminal SH2 domain (cSH2), which is involved in conformational changes occurring during the activation of PLCγ1 by receptor tyrosine kinases. We determined the crystal structure of a pK15 12mer peptide containing the phosphorylated pK15 Y481EEVL motif in complex with a shortened PLCγ1 tandem SH2 (tSH2) domain. This structure demonstrates that the pK15 peptide binds to the PLCγ1 cSH2 domain in a position that is normally occupied by the linker region connecting the PLCγ1 cSH2 and SH3 domains. We also show that longer pK15 peptides containing the phosphorylated pK15 Y481EEVL motif can increase the Src-mediated phosphorylation of the PLCγ1 tSH2 region in vitro. This pK15-induced increase in Src-mediated phosphorylation of PLCγ1 can be inhibited with the small pK15-derived peptide which occupies the PLCγ1 cSH2 domain. Our findings thus suggest that pK15 may act as a scaffold protein to promote PLCγ1 activation in a manner similar to the cellular scaffold protein SLP-76, which has been shown to promote PLCγ1 activation in the context of T-cell receptor signaling. Reminiscent of its positional homologue in Epstein-Barr Virus, LMP2A, pK15 may therefore mimic aspects of antigen-receptor signaling. Our findings also suggest that it may be possible to inhibit the recruitment and activation of PLCγ1 pharmacologically.

Physiological PLCγ1 activation occurs following its recruitment to receptor or non-receptor tyrosine kinases (RTKs, TKs), which in turn phosphorylate the PLCγ1 SA domain on tyrosine residue 783. Either the aminoterminal nSH2 domain, or the carboxyterminal cSH2 domain have been found to bind to an activated RTK [46,47]. Activation of PLCγ1 occurs due to the intramolecular binding of the phosphorylated Y 783 to the cSH2 domain, which leads to conformational changes in the protein and the release of the cSH2 domain from the catalytic core of PLCγ1, thus rendering the latter accessible to the substrate PIP2 [43,44,[46][47][48][49].
In contrast to cellular RTKs, pK15 is not a tyrosine kinase, and one of the unanswered questions therefore is how pK15 can activate the phosphorylation of PLCγ1. As a first step towards answering this question, we studied the interaction of pK15 with PLCγ1 in molecular detail. We identified regions of pK15 and PLCγ1 that are crucial for their interaction in cells as well as in vitro, and determined the crystal structure of the PLCγ1 tandem SH2 domain in complex with a phosphorylated peptide representing the C-terminal end of the pK15 cytoplasmic domain. Our results suggest that the interaction of a phosphorylated pK15 with PLCγ1 primes the latter for phosphorylation by src kinase. pK15 may therefore act as a 'scaffold' to promote the phosphorylation of PLCγ1 by cellular RTKs or TKs. We also show that it is possible to inhibit the pK15-dependent phosphorylation of PLCγ1 with a small pK15-derived peptide occupying the PLCγ1 cSH2 domain.

pK15 phosphorylation in KSHV infected cells and its role in the interaction with PLCγ1
We and others have previously shown that the cytoplasmic tail of pK15 is phosphorylated on tyrosine residue Y 481 in cells that were transfected with a K15 expression vector [24,[27][28][29]31,38]. To show that pK15 phosphorylation takes place in KSHV infected cells, we immunoprecipitated all proteins phosphorylated on tyrosine from HEK293 cells carrying either a bacmid containing the wt KSHV genome (BAC36; WT) or a KSHV genome, in which the first seven exons of ORFK15 had been replaced with a kanamycin cassette (ΔK15) [38]. We were able to detect pK15 bands after phosphotyrosine immunoprecipitation in the HEK293 cells carrying the KSHV WT, but not KSHV ΔK15 virus, using an antibody to pK15 (Fig 1C, blot labelled 'pK15'). The abundance of the phosphorylated pK15 is low, as we could not detect it by immunoblotting with an antibody to phosphotyrosines in either the input lysates, or after immunoprecipitation with phosphotyrosine antibody-coated beads (Fig 1C, blots labelled 'pTyr'). Moreover, in HEK293 cells carrying the KSHV WT, pK15 could only be detected ( Fig  1C, input), or immunoprecipitated with an anti-phosphotyrosine antibody (Fig 1C, IP:pTyr), upon the induction of the lytic cycle by RTA and sodium butyrate (indicated by the expression of the early lytic proteins k-bZIP and pORF45 in Fig 1C, input). We obtained a similar result in an immortalised endothelial cell line, HuAR2T, which had been stably infected with virus produced in HEK293 cells carrying either KSHV BAC 36 or KSHV BAC36 ΔK15 [28,38] (Fig  1D). In these endothelial cells, phosphorylated pK15 was already expressed weakly before the induction of the lytic cycle (Fig 1D, Input). This result indicates that pK15 is phosphorylated in KSHV-infected epithelial and endothelial cells, in particular following the reactivation of the lytic replication cycle.
pK15 interacts with many cellular proteins having diverse functions [22,26,29,31,38,50,51]. Among them, the interaction with phospholipase C gamma 1 (PLCγ1) has been shown to be responsible for the invasive and angiogenic phenotype of KSHV-infected endothelial cells as well as for KSHV reactivation from latency [28,29,38]. Considering that one of the SH2 binding sites of pK15 (Y 481 EEV; Fig 1A) is crucial for this interaction [38] and that pK15 is phosphorylated on the tyrosine residue within this motif [29,31,38], it was important to test the effect of pK15 phosphorylation on its binding to PLCγ1. For this, we phosphorylated the purified cytoplasmic domain of pK15 in vitro and measured its binding to transfected or purified recombinant PLCγ1 or PLCγ1 fragments.
We expressed and purified the pK15 cytoplasmic tail (pK15 CT) fused to GST at its N-terminal end in bacteria and phosphorylated it in vitro using a commercially obtained human GST-6xHis Src, or a recombinant chicken 2xStrep Src kinase expressed in insect cells. In parallel, the GST-pK15 CT domain carrying the Y 481 F mutation, or GST protein alone, were subjected to the same procedure. Phosphorylation of pK15 CT WT resulted in a dominant (bottom) and a minor (top) pK15 band that could be detected by Western blot using an antibody to phosphorylated tyrosine residues (Figs 2A-2C, middle panel labelled 'pTyr' and S1), while phosphorylation of pK15 CT Y481F only produced the minor band (top). This minor band results from the phosphorylation of the second SH2-binding site (Y 431 ASL) in the pK15 CT domain, since submitting a GST-pK15 CT domain carrying a Y 431 F mutation to in vitro phosphorylation by recombinant Src kinase failed to produce this band (S1 Fig). Following in vitro phosphorylation, pK15 CT WT and pK15 CT Y481F were used to pull down transiently expressed (Fig 2A) or endogenous ( Fig 2B) PLCγ1 from HEK 293 lysates. We found that, for both transfected and endogenous PLCγ1, in vitro phosphorylation of the cytoplasmic tail of pK15 CT WT led to a substantial increase in its interaction with PLCγ1, while phosphorylation of the mutant pK15 CT Y481F only showed a minimal interaction with PLCγ1 (Fig 2A and 2B).
The PLCγ1 specific array (SA), which is important for the interaction with pK15 [29,38], contains one SH3 and two SH2 domains, with the latter constituting the tandem SH2 (tSH2) domain ( Fig 1B). As shown in Fig 2C, we also transfected the isolated PLCγ1 SA region into HEK 293 cells and measured its binding to the in vitro phosphorylated pK15. Similar to full length PLCγ1 (Fig 2A and 2B), phosphorylated pK15 CT WT bound more strongly to the SA region than the phosphorylated pK15 CT Y481F mutant; a marginal binding to PLCγ1 SA was observed without prior in vitro phosphorylation of pK15 CT WT by Src kinase (Fig 2C).

Role of the PLCγ1 SH2 domains in the interaction with phosphorylated pK15
Having shown that phosphorylation of the distal SH2-binding motif (Y 481 EEVL) in the pK15 cytoplasmic domain is required for its recruitment of PLCγ1, we next investigated the involvement of the two PLCγ1 SH2 domains in this interaction. In a first step, we compared the interaction of phosphorylated pK15 CT with transfected full-length PLCγ1, the PLCγ1 SA domain, the PLCγ1 tSH2 domain, or individual SH2 domains and with PLCγ1 mutants, in which arginine residues at positions 586 in the nSH2 domain and/or position 694 in the cSH2 domain had been substituted to leucines (Fig 1B). These arginine residues form a central part of the phospho-tyrosine binding pockets within the nSH2 or cSH2 domains, into which the phosphorylated pK15 would be expected to bind. were transfected in HEK-293T cells. After 30h cells were lysed and a GST-pulldown assay was performed with GST-fused pK15 CT (WT), the pK15 CT Y 481 F mutant (Y 481 F) or GST that had been bound to glutathione beads and previously phosphorylated by GST-6xHis Src (+) or left unphosphorylated (-). Bound proteins were analysed by WB using antibodies to the S tag on PLCγ1 and pTyr. (B) Untransfected HEK-293T cells were lysed and a GST-pulldown assay with pre-phosphorylated (+) or unphosphorylated (-) GST-fused pK15 CT WT/Y 481 F or GST was performed and analysed by WB as in (A). (C) An expression plasmid for an S-tagged PLCγ1 SA or the empty vector were transfected in HEK-293T cells. After 30h, cells were lysed and a GST-pulldown was performed as in panel A. https://doi.org/10.1371/journal.ppat.1009635.g002 GST pulldown assays with phosphorylated pK15 CT demonstrated that, when full-length PLCγ1 or the PLCγ1 SA domain were transiently expressed in HEK 293T cells, mutating either R586 or R694 in, respectively, the PLCγ1 nSH2 or cSH2 domains led to a moderately reduced binding to phosphorylated GST-pK15 CT in the case of full-length PLCγ1 WT ( Fig 3A) and strongly reduced binding in the case of the PLCγ1 SA domain (Fig 3B). Mutating both SH2 domains simultaneously completely eliminated the binding of phosphorylated pK15 CT to the PLCγ1 SA domain, but not to full length PLCγ1 (Fig 3A and 3B), suggesting that other regions in PLCγ1 located outside the SA domain may also contribute to this interaction with in vitro phosphorylated pK15. We also investigated the individual contribution of the two PLCγ1 SH2 domains to the interaction with phosphorylated pK15 CT by transfecting expression vectors for the nSH2 or cSH2 domain into HEK 293 cells and performing a GST-pulldown assay with in vitro phosphorylated pK15 CT WT or pK15 CT Y481F . In this experiment, the phosphorylated pK15 CT WT bound preferentially to the PLCγ1 cSH2 domain compared to its nSH2 domain ( Fig 3C).
We next quantified the contribution of the PLCγ1 γ-specific array (SA) and its subdomains (tSH2, nSH2 or cSH2) to the interaction with phosphorylated pK15 CT in an AlphaLisa experiment, in which the interaction of phosphorylated purified GST-pK15 CT with recombinant purified His-tagged PLCγ1 domains is measured by adding glutathione-coated donor beads and Ni-derivatised acceptor beads. As the two protein ligands interact, donor and acceptor beads are brought into close proximity and excitation of the donor beads at 680 nm generates a luminescence signal at 615 nm emanating from the acceptor beads. In this experiment, in vitro phosphorylated pK15 CT WT bound more strongly to the PLCγ1 tSH2 domain than to the PLCγ1 SA domain; for both the SA and tSH2 domains the interaction was entirely dependent on the two SH2 domains, since the PLCγ1 R 586 L and R 694 L double mutant (DM) failed to interact with phosphorylated pK15 CT WT (Fig 4A). In the context of both PLCγ1 SA and PLCγ1 tSH2, the nSH2 R 586 L mutant bound pK15 CT WT to a similar extent as wt PLCγ1 SA and PLCγ1 tSH2, while mutating the cSH2 domain (R 694 L) markedly reduced binding ( Fig  4A). In keeping with this observation, only the PLCγ1 cSH2 domain (and not the nSH2 domain) interacted with phosphorylated pK15 CT WT when tested individually, and this interaction was abolished by the R 694 L mutation in the cSH2 domain (Fig 4A), thus suggesting that the PLCγ1 cSH2 domain is crucial for its binding to the phosphorylated pK15 CT domain.
An in vitro phosphorylated pK15 CT N2 only bound weakly to PLCγ1 SA, not to the PLCγ1 nSH2 domain, but did interact with the tSH2 and cSH2 domains ( Fig 4B). This truncated pK15 CT N2 contains pK15 amino acids 410-489 fused to GST [51] and thus lacks the P 387 PLP SH3 binding site in the pK15 cytoplasmic domain (Fig 1A). The fact that in vitro phosphorylated pK15 CT N2 binds strongly to the PLCγ1 tSH2 domain but not to the PLCγ1 SA domain, while pK15 CT WT does interact with the PLCγ1 SA, could suggest an involvement of the PLCγ1 SH3 domain in the interaction with pK15 CT WT . The stronger binding of pK15 CT WT and pK15 CT N2 to the PLCγ1 tSH2 domain than to the longer PLCγ1 SA domain could be explained by the pseudo-cyclic conformation of the latter, whose N-and C-termini are known to be brought together in a non-covalent fashion [46]; this 'closed' conformation may be harder to access for pK15 CT and pK15 CT N2 than the tSH2 domain.

The PLCγ1 cSH2 domain can serve as a dominant negative inhibitor of the K15-PLCγ1 interaction
We previously showed that, when transfected into cells, the isolated cSH2 domain of PLCγ2 can be used to inhibit the K15-PLCγ1 interaction and subsequent PLCγ1 activation in a dominant negative manner [29]. In view of the contribution of the PLCγ1 cSH2 domain to the interaction of phosphorylated pK15 CT with PLCγ1 (Figs 3, 4A and 4B) we therefore explored whether it might have a similar inhibitory effect as PLCγ2 cSH2. Using the AlphaLISA assay we could show that the purified PLCγ1 cSH2 domain can directly inhibit the interaction  between two in vitro phosphorylated pK15 fragments, pK15 CT N1 or pK15 CT N2 , and PLCγ1 SA or PLCγ1 tSH2 in a dose-dependent manner (Fig 4C and 4D). pK15 CT N1 contains pK15 amino acids 365-489 fused to GST [51] and therefore retains the P 387 PLP SH3 binding site in the pK15 cytoplasmic domain, in contrast to pK15 CT N2 (Fig 1A). The inhibitory effect of the PLCγ1 cSH2 domain was more pronounced in the case of pK15 CT N2 (Fig 4C and 4D), suggesting that, in addition to the key contribution of the PLCγ1 cSH2 domain to the interaction with phosphorylated pK15, other regions in the pK15 CT and PLCγ1 SA domain may contribute.
We also investigated whether the isolated PLCγ1 cSH2 domain could interfere with the recruitment of PLCγ1 to pK15 in pK15-expressing cells. Co-transfection of full length pK15 together with a plasmid encoding the isolated PLCγ1 cSH2 domain resulted in a reduced phosphorylation of PLCγ1 on Y 783 , which is normally phosphorylated upon PLCγ1 activation ( Fig  4E). In this experiment, the PLCγ1 cSH2 domain inhibited PLCγ1 phosphorylation to a greater extent than the PLCγ2 cSH2 domain (Fig 4E and 4F).

Affinity of the distal pK15 SH2 binding site for PLCγ1 domains
We next investigated the affinity of a phosphorylated peptide representing the distal pK15 SH2 binding site and its key residues (Y 481 EEVL) for the PLCγ1 tSH2 domain. We used a phosphorylated FAM-labelled peptide representing the last 12 residues (DDLpYEEVLFPRN) of pK15 CT and measured its affinity to the purified PLCγ1 tSH2 domain (aa 545-790), to the R 586 L and R 694 L mutants of the tSH2 domain, and to the individual PLCγ1 SH2 domains (nSH2, aa 545-662; cSH2, aa 668-790) and their mutants (R 586 L and R 694 L), using a diffusion chamber approach. As shown in Fig 5A, the individual SH2 domains bound the phosphorylated pK15 peptide with high affinity (cSH2 domain, K D = 166 ± 30 nM; nSH2 domain, K D = 180 ± 43 nM), and mutation of the key arginine residues R 586 and R 694 in the isolated nSH2 and cSH2 domains, respectively, abolished pK15 peptide binding (Fig 5A), in keeping with the importance of the phosphorylated pK15 Y 481 residue for the interaction with PLCγ1.
The PLCγ1 tSH2 domain bound the phosphorylated pK15 peptide~10 times more weakly than the isolated nSH2 and cSH2 domains (K D = 1500 ± 450 nM), suggesting that the tSH2 domain may be in a 'closed' conformation, which impedes access of the pK15 peptide to the individual SH2 domains. In spite of the similar affinities of the isolated nSH2 and cSH2 domains for the pK15 peptide, their contribution to the affinity of tSH2 was different. The tSH2 R 694 L mutant bound the pK15 peptide with a K D of 6318 ± 475 nM, in agreement with the idea that the cSH2 subdomain is important for the interaction with the phosphorylated pK15 peptide, as also suggested by the results shown in Fig 4. However, the tSH2 R 586 L mutant, with a mutated pTyr recognition site in nSH2, bound the 12mer pK15 peptide better that WT tSH2 (K D = 440 ± 77 nM) ( Fig 5A). This result suggests that binding of the pK15 peptide to nSH2 in the context of the tSH2 fragment inhibits binding of pK15 to cSH2. The structural basis for this effect remains to be investigated.
Next, we used Nuclear Magnetic Resonance (NMR) spectroscopy to further compare the interaction of the phosphorylated pK15 12mer to the isolated nSH2 and cSH2 domains ( Fig  5B-5E). We recorded 1 H-15 N HSQC spectra of each of the two domains in the presence of increasing concentrations of the pK15 peptide. In this spectrum, each N-H group of the protein gives rise to one peak. Thus, besides the side-chain amide groups, the spectrum shows one peak per amino acid. A change in the position of a few NMR peaks of the amide groups of the protein upon addition of the peptide is indicative of binding. We could confirm that both domains bound the K15 peptide with similar affinity, as the NMR peaks of either nSH2 or cSH2 had completely shifted to the frequencies of the corresponding protein-peptide complex after addition of similar amounts of peptide (that is, the saturating concentration of the peptide was the same). The behavior of the NMR peaks during the titration series, however, differed for the two SH2 domains, indicating different binding kinetics. In NMR the way how peaks change their position during a titration experiment depends on the exchange constant k ex , defined as the sum of the dissociation rate k off and the association rate k on of the complex (k ex = k off + k on ). If k ex is more than one order-of-magnitude larger than the difference in the frequencies of the peptide-free and peptide-bound states of the protein (that is, the difference in the position of the peak in the absence of peptide and in the presence of a saturating concentration of peptide), the peak is expected to move continuously from the peptide-free to the peptide-bound position during the titration (fast exchange regime), as seen in Fig 5E. However, if k ex is more than one order-of-magnitude smaller than the difference in the frequencies of the peptide-free and peptide-bound states of the protein, the peak is expected to disappear from the peptide-free position and reappear at the peptide-bound position during the titration (slow exchange regime). At a saturating concentration of peptide, the peak has completely disappeared from the peptide-free position and has gained maximum intensity at the peptidebound position, as in Fig 5D. Thus, the spectra in Fig 5D and 5E demonstrated that either the dissociation rate k off or association rate k on of the nSH2-peptide complex is more than one order of magnitude slower than the corresponding rate of the cSH2-peptide complex. In light of this, it is conceivable that the preference of pK15 CT for cSH2 in vivo may be determined by kinetic rather than thermodynamic factors.

Spatial organization of the PLCγ1 tSH2 domain in complex with the distal pK15 SH2 binding site
To obtain structural insights into this interaction we sought to determine the structure of the PLCγ1 tSH2 domain in complex with the phosphorylated 12mer pK15 peptide. Initial attempts to obtain diffractable crystals of the entire PLCγ1 tSH2 domain (aa 541-790) [46,52] in complex with this 12mer pK15 peptide were unsuccessful. We reasoned that the linker region separating the PLCγ1 cSH2 and SH3 domains ( Fig 1B) could have interfered with the binding of the pK15 12mer peptide to the PLCγ1 cSH2 domain, since the former contains tyrosine residue 783, which is known to interact with the PLCγ1 cSH2 domain [46]. We therefore expressed and purified a truncated PLCγ1 tSH2 domain (PLCγ1 aa 545-772), which lacks this linker segment. We obtained diffraction quality crystals of this truncated tSH2 domain in complex with the 12mer phosphorylated pK15 peptide. The structure was determined by the molecular replacement method and refined to a resolution of 2.1Å (Fig 6). Details of crystallization and structure determination are described in Materials and Methods, and the crystallographic statistics are listed in Table 1. In this complex, the 12mer pK15 peptide is unambiguously bound to the PLCγ1 cSH2 domain (Fig 6A and 6B), but at lower occupancy peptide binding to the nSH2 domain was also observed, in line with the findings shown in Figs 3-5. In the PLCγ1 cSH2 domain the pK15 peptide straddles the triple ß-sheet that divides the two pockets of the SH2 domain ( Fig 6B). Its phosphorylated tyrosine residue (pK15 Y 481 ) is in contact with PLCγ1 R 675 and R 694 in the first of these two pockets (Fig 6E and 6F), while its hydrophobic residues V 484 and L 485 in the canonical pK15 SH2-binding site (YEEVL) are in proximity to the second pocket that is known to accommodate hydrophobic residues (Fig 6D). The pK15 peptide adopts a similar position to that of residues 781-790 in the PLCγ1 linker peptide containing phosphorylated Y 783 , which connects the PLCγ1 tSH2 and SH3 domains (Figs 1B, 6C, and 6F) [46]. In its unphosphorylated form, Y 783 in the PLCγ1 tSH2 domain linker is oriented in the opposite direction in the published structure of the PLCγ1 tSH2 domain [52] (Fig 6B and 6E). We therefore propose that, following phosphorylation of pK15 on Y 481 , the last 12 amino acids of the pK15 cytoplasmic domain engage the PLCγ1 cSH2 domain in a manner similar to the PLCγ1 tSH2-SH3 domain linker, once the latter has been phosphorylated on Y 783 by the activating receptor tyrosine kinase (Figs 1B, 6B, 6E and 6F).

Binding of phosphorylated pK15 CT enhances phosphorylation of PLCγ1 by src kinase
During its recruitment to the phosphorylated C-terminal kinase domains of receptor tyrosine kinases (RTK), PLCγ1 is thought to bind to the RTK via its nSH2 domain, thereby allowing the phosphorylation of PLCγ1 Y 783 , which then in turn engages the cSH2 domain to trigger a conformational change in PLCγ1 that is required for its lipase activity [43,46,52]. However, an alternative model envisages engagement of both the nSH2 and cSH2 domains by the activated RTK [47]. Our observation that phosphorylated pK15 CT, which does not possess kinase activity, engages the PLCγ1 cSH2 domain therefore raised the question how this would impact on the phosphorylation of PLCγ1. The scaffold protein SLP-76 uses its non-canonical SH2 binding site (pY 173 IDR) to interact with the PLCγ1 cSH2 domain [53] and a proline-rich motif (PPVPPQRP) to bind to the PLCγ1 SH3 domain [54]. The interaction between SLP-76 pY 173 IDR and the PLCγ1 cSH2 domain has been hypothesized to promote a conformational change in PLCγ1 that displaces the autoinhibitory cSH2 domain from the PLCγ1 core and thereby leads to an increased exposure and enhanced phosphorylation of PLCγ1 on Y 783 in the cSH2-SH3 linker peptide (Fig 1B) by the TEC family kinase ITK [53]. We therefore asked if pK15 could recruit and activate PLCγ1 in a manner similar to SLP-76.
To address this hypothesis, we tested if phosphorylated fragments of pK15 CT could enhance the phosphorylation of the purified PLCγ1 tSH2 domain on Y 783 by recombinant Src kinase in an in vitro kinase assay. We established conditions in which limiting amounts of recombinant purified Src kinase were added to the purified tSH2 domain so as to allow only a

PLOS PATHOGENS
low level of tSH2 domain phosphorylation, which we detected on Western blots using an antibody to the phosphorylated PLCγ1 Y 783 motif. Using these conditions, we then tested whether increasing amounts of the phosphorylated 12mer pK15 peptide used in the crystallography, diffusion chamber and NMR experiments (Figs 5 and 6), a longer phosphorylated pK15 peptide corresponding to the last 35 amino acids of pK15 CT, or the in vitro phosphorylated pK15 CT N2 could enhance the Src-mediated phosphorylation of the PLCγ1 tSH2 domain. While the 12mer pK15 peptide showed a moderate inhibition in this assay (Fig 7A and 7B), the 35mer pK15 peptide enhanced the Src-mediated phosphorylation of the PLCγ1 tSH2 domain in a dose dependent manner, as shown by the increased intensity of the tSH2 band stained with an antibody to the phosphorylated PLCγ1 Y 783 motif (Fig 7A, upper panel) and the appearance of a more slowly migrating tSH2 band on the Western blot stained with an antibody to the 5x His tag on the recombinant tSH2 protein (Fig 7A, lower panel). Likewise, the in vitro phosphorylated pK15 CT N2 protein, which corresponds to the last 80 amino acids of pK15 CT (see above), also enhanced the Src-mediated phosphorylation of the PLCγ1 tSH2 domain (Fig 7B). This result suggests that, like SLP-76, phosphorylated pK15 might act as a scaffold protein that recruits PLCγ1 via its cSH2 domain and thereby promotes its activation by a tyrosine kinase.

Inhibition of the pK15 CT-mediated enhancement of PLCγ1 phosphorylation using a short pK15-derived peptide
Unlike the C-terminal 80 or 35 amino acids of the pK15 CT domain, the shorter 12mer peptide cannot enhance the Src-mediated phosphorylation of PLCγ1 (Fig 7A and 7B). We therefore explored if this short 12mer peptide, which contains the binding site for the PLCγ1 cSH2 domain (Fig 6) can inhibit the pK15-mediated enhancement of PLCγ1 phosphorylation in a competitive manner. As shown in Fig 7C, the phosphorylated 12mer peptide could inhibit the increased PLCγ1 tSH2 phosphorylation induced by the 35mer peptide. This observation suggests that it may be possible to counteract the pK15-dependent phosphorylation and activation of PLCγ1 with small peptides and possibly also with small pharmacological inhibitors.

Discussion
The role of KSHV as causative agent of Kaposi's sarcoma was established more than twenty years ago [1] and since then substantial evidence has accumulated demonstrating the contribution of different viral proteins and cellular pathways to the disease progression. The ability of KSHV to trigger invasiveness and angiogenesis is important for the development of KS and is associated with higher expression of matrix metalloproteinases (MMPs), activation of cyclooxygenase 2 (Cox-2), mitogen-activated protein kinase kinase kinase kinase 4 (MAP4K4), PI3K/ Akt/mTOR, NF-κB and PLCγ1 pathways [29,38,[55][56][57][58][59]. Also viral factors increasing angiogenesis and invasion have been described: viral microRNAs miR-K6-3p and miR-K6-5p as well as proteins pK1, LANA, vFLIP, vGPCR and pK15 among others [29,38,[59][60][61][62][63]. We have previously shown the importance of the recruitment and activation of PLCγ1 by pK15 during the reactivation of KSHV from latency [28] and the increased migration, proliferation and invasiveness of KSHV-infected endothelial cells [28,29,38]. A better understanding of how pK15 recruits and activates PLCγ1 might therefore provide a basis for the development of small molecule inhibitors that target some of these effects induced by KSHV in endothelial cells.
In this study we first demonstrate that pK15 is phosphorylated on tyrosine residues in KSHV-infected cells (Fig 1). While such tyrosine phosphorylation of pK15 has previously been shown with the help of an overexpressed chimeric K15 protein [24] and the purified recombinant pK15 CT [31], the result shown in Fig 1 demonstrates that phosphorylation of pK15 on

PLOS PATHOGENS
Activation of PLCγ1 by KSHV pK15 tyrosine residues occurs in KSHV-infected epithelial and endothelial cells. We also show that pK15 phosphorylation on Y 481 in the second SH2 binding site (Y 481 EEV) of the pK15 CT enhances binding of pK15 to PLCγ1 (Fig 2). This observation is in line with the fact that pK15 can be phosphorylated on Y 481 by Src family kinases in vitro and that the SH2 binding site is necessary for PLCγ1 recruitment [25,31,37,38].
PLCγ1 contains two SH2 domains, which contribute differentially to its interaction with other proteins. Both its nSH2 and cSH2 domains are necessary for the interaction of PLCγ1 with platelet-derived growth factor receptor (PDGFR) [64], while the nSH2 domain mediates the interaction with the phosphorylated cytoplasmic domains of the epidermal growth factor receptor (EGFR) and fibroblast growth factor receptor (FGFR) [46,52,65]. The cSH2 domain has been reported to also bind directly to the FGFR cytoplasmic domain [47] and to an atypical SH2-binding site motif (pY 173 IDR) of SLP-76 [53]. The prevailing view holds that the cSH2 domain is required for an intramolecular interaction with the phosphorylated Y 783 residue in the linker domain between the cSH2 and SH3 domains of PLCγ1, which relocates the cSH2 domain from an inhibitory interaction with the PLCγ1 core and is thus thought to contribute to conformational changes in PLCγ1 that are required for its enzymatic activity [43,46,66].
In our experiments with the purified phosphorylated pK15 CT domain, PLCγ1 cSH2 seemed to be the main contributor to its interaction with pK15 (Figs 4A, 4B, 5 and 6). When we used the purified, in vitro phosphorylated, GST-fused pK15 CT domain to pull down full length PLCγ1 expressed in mammalian cells, mutations of arginine residues in the phosphotyrosine-binding pocket of individual PLCγ1 SH2 domains reduced their interaction (Fig 3). However, only in the case of the PLCγ1 SA domain did the simultaneous mutation of both the nSH2 and the cSH2 domain completely abolish the interaction with phosphorylated pK15 (Fig  3B), suggesting that PLCγ1 regions outside the SA domain may contribute to the recruitment of PLCγ1 to phosphorylated pK15. Our results obtained with transfected cells are therefore compatible with both the nSH2 and the cSH2 domain contributing to the pK15-PLCγ1 interaction. However, when isolated PLCγ1 domains were expressed in cells, or used as purified recombinant proteins, and tested for the interaction with phosphorylated pK15, the PLCγ1 cSH2 domain bound to pK15 CT more strongly than the nSH2 domain, both in cells and in vitro (Figs 3C, 4A and 4B). Also, the cSH2 domain in isolation inhibited the interaction of pK15 with PLCγ1 (Fig 4C and 4D) as well as pK15-induced PLCγ1 phosphorylation in transfected cells (Fig 4E).
We measured the affinity of a short phosphorylated pK15 peptide, representing the last 12 amino acids of pK15 CT, for the entire PLCγ1 tSH2 domain, as well as for its individual SH2 domains. Our results suggest that the entire PLCγ1 tSH2 domain has a low affinity for this short peptide, while individual SH2 domains display a higher affinity (Fig 5). It therefore appears that the entire PLCγ1 tSH2 domain adopts a conformation that is only poorly accessible to pK15 CT. This hypothesis is supported by our observation that we had to truncate the tSH2 domain linker containing PLCγ1 Y 783 in order to obtain diffraction quality crystals of the tSH2 domain in complex with this 12mer pK15 peptide (Fig 6). A comparison of the previously published structure of the entire PLCγ1 tSH2 domain [46,52] with the structure of the truncated PLCγ1 tSH2 domain in complex with the 12mer pK15 peptide indicates that the Cterminal end of the pK15 cytoplasmic domain may compete with the PLCγ1 cSH2-SH3 linker for binding to the PLCγ1 tSH2 domain (Fig 6B). If we assume that the cSH2-SH3 linker is able to bind to the cSH2 domain better in the tSH2 fragment than in the isolated cSH2 domain, this interpretation would explain the higher affinity of the isolated cSH2 domain compared to the entire tSH2 domain for the phosphorylated last 12 amino acids of pK15 CT (Fig 5A).
While the pK15 12mer peptide showed comparable affinities to the isolated PLCγ1 nSH2 and cSH2 domains (Fig 5A), longer pK15 CT fragments bound much more strongly to the cSH2 domain than to the nSH2 domain (Fig 4A and 4B). This could suggest that other regions in the pK15 CT domain may also contact the PLCγ1 cSH2 and thereby increase the affinity of the phosphorylated pK15 CT domain for the PLCγ1 cSH2 relative to the nSH2 domain. In addition, by NMR we found that the 12mer peptide showed a faster exchange with the cSH2 domain than with the nSH2 domain (Fig 5B-5E), providing the possibility of a kineticallydriven binding regulation in vivo.
We also found that the last 35 amino acids of the cytoplasmic domain of pK15, as well as a longer 80 amino acid segment derived from the C-terminal end of the pK15 cytoplasmic domain, can enhance the phosphorylation of PLCγ1 by Src kinase in vitro (Fig 7). A possible interpretation of this observation is that the pK15 cytoplasmic domain displaces the PLCγ1 cSH2-SH3 linker peptide (Figs 1B and 6B) from the PLCγ1 cSH2 domain and thus renders Y 783 in this PLCγ1 linker peptide more accessible to Src-mediated phosphorylation, as schematically depicted in Fig 8. The fact that the short 12mer pK15 peptide does not increase the Src-mediated phosphorylation of the PLCγ1 tSH2 domain suggests that pK15 CT regions outside the 12mer peptide may contribute to this conformational change. We find it intriguing that a similar mechanism of PLCγ1 activation has been reported for the SLP-76 scaffold protein, which, like pK15, uses its SH2-binding site (pY 173 IDR) to bind to the PLCγ1 cSH2 domain and thereby primes PLCγ1 for phosphorylation by the T-cell receptor associated tyrosine kinase ITK [53]. We therefore speculate that pK15 mimics aspects of intracellular signaling triggered by the activation of the T cell receptor. Interestingly, the positional homologue of the K15 gene in the related gammaherpesvirus Epstein-Barr virus (EBV) is LMP2, and the LMP2A protein is thought to mimic aspects of B-cell receptor signaling [67]. The two nonstructural membrane proteins pK15 and LMP2A of these two human gammaherpesviruses may thus have found similar ways to engage intracellular pathways involving the activation of PLCγ1. Intriguingly, the cagA protein of Helicobacter pylori is also phosphorylated by Src family kinases, allowing phosphorylated cagA to recruit and activate the cellular phosphatase SHP2 [68]. Since we could show that, in the case of KSHV pK15, the increased phosphorylation of PLCγ1 observed in the presence of a pK15 fragment can be inhibited with a small peptide that docks into the PLCγ1 cSH2 domain (Figs 6 and 7), the underlying mechanism may provide a target for the future development of inhibitors against KSHV-related diseases. The structure of this short pK15 peptide in complex with the PLCγ1 tSH2 domain reported here provides the basis for the in silico screening of small molecule inhibitors that would interfere with the pK15-PLCγ1 interaction by docking into the PLCγ1 cSH2 pocket shown in Fig 6. Given the key role of the phosphorylated PLCγ1 Y 783 residue in the PLCγ1 cSH2-SH3 linker peptide during the activation of PLCγ1 by cellular receptor tyrosine kinases (Fig 8), we speculate that small molecules mimicking the effect of the pK15 12mer peptide described here could also prove useful in controlling PLCγ1 activity in other diseases that are caused by an excessive activation of PLCγ1 [43,69].

DNA constructs
pFJ-EA and pFJ K15P WT were a kind gift from J. Jung (University of Southern California). The generation of the pFJ K15P YF mutant was described before [31]. pGEX-6P-1 K15P 347-489 is described in [29]. N-terminal truncation mutants of K15 cytoplasmic tail pGEX 6P-1 K15P N1 and N2 were generated by PCR on pGEX-6P-1 K15P 347-489, using forward primers 5'-GCATTGGATCCATTTATACCCGTGATCAGAATCTGC-3' and 5'-GCATTGGATC CAGCCAGCCGCTGAATGAAG-3' respectively. In both cases the reverse primer 5'-GCATT GAATTCTTAGTTACGCGGAAACAGAACTTCT-3' was used. The resulting products  [43,44,45,46,66]. Below Left: Activation of PLCγ1 by receptor tyrosine kinases (RTK): following autophosphorylation on tyrosine residues, the cytoplasmic RTK domain recruits the PLCγ1 nSH2 domain and phosphorylates PLCγ1 Y 783 , which in turn binds into the cSH2 domain, thus removing it from its proximity to the catalytic core, and thereby activates the enzymatic activity of the latter. Below Right: Activation of PLCγ1 by pK15: as shown in this report, the phosphorylated Y 481 of pK15 (phosphorylated by Src in our experiments) binds into the PLCγ1 cSH2 domain. This presumably leads to a conformational change in PLCγ1, which in turn facilitates the subsequent phosphorylation of PLCγ1 Y 783 by members of the Src kinase family and subsequent activation of PLCγ1 lipase activity.

Protein purification
For production and purification of GST-fused K15 proteins, Rosetta E. coli were grown overnight in LB-medium with ampicillin at 37˚C, 220 rpm, then diluted 1:10 in fresh LB containing ampicillin and incubated until OD 600 0.6. Protein production was induced with 1 mM IPTG for 4 hours at 30˚C. Cells were pelleted at 5.000g 4˚C for 10 minutes, resuspended in PBS (50ml for 1 liter culture) with protease inhibitors and lysed by sonication on ice 5 times for 30 seconds with a 30 second pause after each sonication and 0.5% (v/v) NP-40 was then added. After centrifugation at 30.000g, 4˚C for 10 minutes, the supernatant was transferred to a new tube for the second centrifugation. Cleared lysate was incubated overnight at 4˚C on a roller with Glutathione Sepharose 4 Fast Flow beads (GE Healthcare), which were washed before 3 times in PBS. Afterwards, beads with bound protein were washed 3 times in 1xPBS, 5% (v/v) glycerol, 0.5% (v/v) NP-40, protease inhibitors and the protein was eluted by incubating the beads for 3 hours at 4˚C on a roller in 1xPBS, 10% (v/v) glycerol, 0.5% (v/v) NP-40, 60 mM glutathione, protease inhibitors, pH7.3. The eluted material was centrifuged for 10 minutes at 3.220g and dialyzed against PBS with 1 mM DTT and 100 μM PMSF in a Slide-A-Lyzer Dialysis Cassettes, 3.5K MWCO (Thermo Fisher Scientific).
For expression and purification of PLCγ1 constructs cloned as a thioredoxin fusion protein in pETM22, the plasmids were transformed into E. coli strain BL21(DE3). Cultures were grown in Luria-Bertani (LB) broth at 37˚C to OD 600 of 0.6, then induced with 1 mM IPTG and incubated at 20˚C overnight. 15 N-labeled proteins for NMR studies were prepared by growing the bacteria in M9 minimal medium containing kanamycin (50 μg/ml) and 15 NH 4 Cl (1 g/l, Cambridge Isotope Laboratories). Cells were harvested and resuspended in lysis buffer (1 M NaCl, 50 mM Tris-HCl, 5% glycerol, 10 mM imidazole, 5 M mercaptoethanol, 1 EDTA-free protease inhibitor cocktail tablet (Roche), 100 μg lysozyme (Roth), 50 μg DNAse (NEB), pH 7.6) then lysed by sonication, followed by centrifugation to remove cellular debris. The filtered supernatant was loaded on a HisTrap HP column (GE Healthcare) pre-equilibrated with wash buffer. The thioredoxin tag was cleaved with 3C protease overnight and the cleaved protein was collected in the flow-through of a second Ni 2+ -affinity chromatography run. The eluted sample was subsequently subjected to size-exclusion chromatography on HiLoad 16/600 Superdex 75pg column (GE Healthcare) pre-equilibrated with gel filtration buffer (100 mM MES, 150 mM NaCl, 3 mM TCEP (Tris(2-carboxyethyl)phosphin, Roth), pH 6.5).
PLCγ1 constructs with 6xHis in pTriex-4 and pOPINS vectors were grown as described for GST constructs, but kanamycin was used instead of ampicillin for pOPINS. Bacteria were grown until OD 600 0.8 and the temperature was changed to 30˚C. When the OD 600 reached 1, protein production was induced with 1 mM IPTG for 5 hours at 30˚C, bacteria were pelleted at 6.000g, 4˚C for 10 minutes and the pellet was stored at -80˚C. After pellet resuspension in 40 ml buffer core buffer (50 mM Tris pH 7.4, 500 mM NaCl, 0.1% (v/v) Triton X-100, protease inhibitors), cells were lysed by sonication on ice 5 times for 30 seconds with 30 seconds pause after each sonication and the lysate was cleared by centrifugation at 30.000g, 4˚C for 15 minutes. The cleared lysate was incubated 4 hours at 4˚C on a roller with Ni-NTA Superflow resin (Qiagen), which had been washed 3 times with core buffer before use. The resin with the bound protein was loaded into an Econo-Pac Chromatography Column (BioRad) and the flow through was collected. The column was then washed 3 times with 20 ml core buffer with 20 mM imidazole. Proteins were eluted in 10 ml core buffer with 200 mM imidazole and 10 mM DTT and 1 ml fractions were collected. Purified protein was dialyzed against cold core buffer overnight at 4˚C in a Slide-A-Lyzer Dialysis Cassettes, 3.5K MWCO (Thermo Fisher Scientific).
For 2xStrep-Src production and purification, Drosophila S2 Src cells were grown with 70 rpm agitation in a one-liter culture at 28˚C until cell density reached 6x10 6 . Protein production was induced with 4 μM CdCl 2 for 5 days. Cells were pelleted at 4.000g, 4˚C for 15 minutes, resuspended in 25 ml buffer A1 (10 mM Tris pH 8, 150 mM NaCl, 1 mM EDTA) and lysed in a high-pressure homogenizer at 1.3 KBar, 4˚C. Strep-PLCγ1 cSH2 was produced as described for GST-fused proteins, but the pellets were resuspended in 50 ml buffer A1 and lysed in a high-pressure homogenizer at 1.5 KBar, 4˚C. After clearing the lysates at 75.000g, 4˚C for 30 minutes, supernatants were filtered (0.22 μm) and applied to an 8 ml Strep-Tactin column on an Ä KTA FPLC at a flow rate 2 ml/min with buffer A1. 30 ml of B1 buffer containing 10 mM Tris pH 8, 150 mM NaCl, 1 mM EDTA and 2.5 mM Desthibiotin were used for the elution. Protein containing fractions were pooled together, concentrated and subjected to a gel filtration with GF buffer (100 mM Tris pH 7.5, 100 mM NaCl).
The Drosophila S2 Src cell line used for the purification of the recombinant Src kinase was established by co-transfecting the pT1204 plasmid encoding full length chicken Src with the pT371 plasmid encoding a puromycin resistant gene with Effectene Transfection Reagent (Qiagen) according to the manufacturer's instructions. For this, 5x10 6 cells were seeded in a T25 flask with 5 ml of Schneider's medium (Gibco) and transfected after 24h with 2 μg of pT1204 Src and 0.1 μg of pT371. Resistant cells were selected in the presence of 8 μg/ml puromycin and medium was changed to Insect Xpress medium (Lonza) for protein production.
HEK-293T cells were transfected with Fugene 6 Transfection Reagent (Promega) according to the manufacturer's instructions. 5x10 5 cells were seeded in each well of a 6-well plate (or 2.5x10 5 cells well of a 12-well plate). The next day Fugene was incubated at room temperature with OPTI-MEM for 5 minutes, added to DNA and incubated for another 15 minutes. For each 1 μg DNA, 3 μl Fugene were used.

Western blot analysis and antibodies
For Western blot analysis of cellular lysates, cells were lysed in 80 μl 1xSDS sample buffer (62.5 mM Tris pH 6.8, 10% (v/v) glycerol, 2% (v/v) SDS, 50 mM DTT, 0.01% Bromophenol Blue) per well of a 12-well plate. Lysates were sonicated on ice and cleared in a table top centrifuge at 20.000g, 4˚C for 10 minutes. Samples were analysed by SDS-PAGE with 4% polyacrylamide in the stacking gel and 10% or 12% in the separating gel. For the determination of protein size Precision Plus Protein All Blue Prestained Protein Standards from BioRad was used. After separation, proteins were transferred to a 0.45 μm nitrocellulose membrane (Amersham) in cold transfer buffer (25 mM Tris-Base, 250 mM glycine, 20% (v/v) methanol) at 350mA for 70 minutes. For GST-pulldown assays proteins were stained with Ponceau S for 5 minutes prior 1 hour blocking with PBS-T with 5% non-fat milk (Carl Roth) or TBS-T with 5% IgG-free albumin (Carl Roth). Membranes were incubated at 4˚C overnight with primary antibodies. This was followed by 3 washing steps (10 minutes each) with PBS-T or TBS-T, a one hour incubation at room temperature with secondary HRP-conjugated antibodies, and another 3 washing steps.

Immunoprecipitation of phosphorylated proteins
HEK293 were plated at a density 5x10 5 cells per well of a 6 well plate (2 wells were used per condition) and transfected the next day with K15-encoding or empty PFJ vector using Fugene. HEK293 BAC36 WT/ΔK15 cells were plated at density 3x10 6 per T75 flask and HuART BAC36 at density 5.8x10 6 cells per 15 cm culture dish. After 24 hours virus-infected cells were reactivated and 48 hours after reactivation washed with PBS and lysed in 600 μl (for HEK293) or 500 μl (for HuART) Tris buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% (v/v) NP-40, protease and phosphatase inhibitors). For the transfected cells 300 μl Tris buffer was used per well of a 6-well plate (48 hours after transfection). Lysates were incubated on ice for 30 minutes and cleared by centrifugation for 10 minutes at 20.000g, 4˚C. Phosphotyrosine affinity beads (30 μl per IP) (Cytoskeleton, Inc.) were washed twice in 1 ml of PBST and centrifuged at 800g, 4˚C for 1 minute. The lysates (1.8 mg per IP) were incubated overnight at 4˚C, washed 3 times with 1 ml Tris buffer and the bound proteins were eluted for 5 minutes at room temperature in 30 μl 2x non-reducing SDS sample buffer (125 mM Tris pH6.8, 20% (v/v) glycerol, 4% (v/v) SDS, 0.005% Bromphenol Blue). The samples were centrifuged at 20.000g for 1 minute, supernatants were incubated for 5 minutes at room temperature with 1μl β-mercaptoethanol and analysed by western blot.
For co-immunoprecipitation, 1 μg of pFJ K15P and 1 μg of pTriex-4 PLCγ1 SA were coexpressed in HEK-293T cells. Thirty-two hours after transfection cells were washed once with PBS and lysed in 300 μl IP buffer per well of a 6-well plate. Lysates (500 μl) were incubated overnight at 4˚C with 10μl Red Anti-Flag M2 beads (Sigma Aldrich), which were washed before 3 times with 300 μl IP buffer. After that, beads were subjected to 6 washing steps with 700μl IP buffer, bound proteins were eluted in 15 μl 5x Laemmli sample buffer and analysed by Western blot.
To test the impact of a pK15 CT fragment (pK15 CT N2 ) or phosphorylated pK15 peptides representing the last 12 (DDLpYEEVLFPRN) or 35 (SILRVDGGSAFRIDTAQAATQPTDDL pYEEVLFRN) amino acids of pK15 CT on the Src-mediated phosphorylation of the PLCγ1 tSH2 domain, GST-pK15CT N2 was first phosphorylated with Src in vitro as described in the preceding paragraph. Phosphorylated GST-pK15 CT N2 or the phosphorylated pK15 peptides, suspended in 50 mM Tris pH7.4, 300 mM NaCl, 1 mM DTT, were then added to the purified recombinant tSH2 domain (0.6 μM) together with a limiting amount of recombinant Src kinase (0.085 μM) in kinase buffer (20 mM Tris pH7.4, 20 mM NaCl, 1 mM DTT, 10 mM MgCl2, 0.2 mM ATP, 0.5 mM Sodium orthovanadate, 0.5 mM β-glycerophosphate) and the reaction allowed to proceed for 45 min. The reaction mix was then analysed by Western blot using an antibody to the phosphorylated Y 783 in the linker region of the PLCγ1 tSH2 domain and an antibody to the 5x His tag. To test the inhibitory effect of the phosphorylated 12mer pK15 peptide on the enhancement of PLCγ1 tSH2 domain phosphorylation induced by the pK15 35mer peptide, increasing concentrations (1.25-5.0 μM) of the 12mer peptide were added to a reaction mix containing the purified recombinant tSH2 domain (2.5 μM), the phosphorylated pK15 35mer (2.5 μM), and recombinant Src kinase (0.085 μM). Following a 45 min incubation at 37˚C, the reaction mixture was analysed by Western blot as above.

AlphaLisa
His-tagged PLCγ1 (SA, tSH2 or the isolated SH2 domains) was incubated at room temperature with GST-tagged K15P or its mutants for 1 hour. Both proteins were diluted in Alpha buffer (1xPBS, 0.5% BSA, 0.01% Tween 20) and if indicated, pK15 proteins were phosphorylated by Src kinase before use in the AlphaLisa. Ni chelate acceptor beads and glutathione donor beads (PerkinElmer) were diluted in the Alpha buffer and added sequentially to the proteins for another hour of incubation at room temperature. All manipulations with Alpha beads were performed under subdued light conditions. Final concentration of each protein was 300 nM and of each bead type 4 μg/ml in 25 μl total Alpha reaction volume in the well of an OptiPlate-384 (PerkinElmer). Emission at 615 nm was detected on BioTek Synergy 2 plate reader.

Measuring the affinity of a pK15 peptide for PLCγ1 domains
The 12mer pK15 peptide (DDL(pY)EEVLFPRN) was purchased pre-labelled with 5,6-FAM at the N-terminus from Caslo ApS (Lyngby, Denmark) and was dissolved in buffer (100 mM MES, 150 mM NaCl, 3 mM TCEP (Tris(2-carboxyethyl)phosphin, ROTH), pH 6.5) and diluted to a final concentration of 200 nM for the experiments. To measure its affinity for PLCγ1 domains (tSH2, tSH2 R 586 L mutant, tSH2 R 694 L mutant, nSH2, nSH2 R 586 L mutant, cSH2, cSH2R 694 L mutant) in a microdiffusion chamber the peptide concentration was held constant at 200 nM in each sample, while the concentration of the PLCγ1 domains was varied from 0 nM to 8000 nM. The nSH2, cSH2 and tSH2 domains comprised residues 545-662, 668-790 and 545-790, respectively. All samples were prepared at the same time, equilibrated at room temperature for 30 minutes to reach the steady state and then tested in order, from the highest to the lowest protein content. A 5 μL aliquot of each sample was pipetted onto a microfluidic chip and tested using a Fluidity One-W instrument. Each sample was tested in triplicate.
The values of the hydrodynamic radius R h of the pK15-protein complex in each solution were generated by the Fluidity One-W software; the average values were fitted to a standard binding equation. K D values were determined by non-linear least squares fitting of the data with Equation 1 [71]. y ¼ R h;free þ ðR h;complex À R h;free Þ ðK D þ A þ nxÞ À ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi where y is the hydrodynamic radius of the mixture measured on Fluidity One-W, R h,free is the hydrodynamic radius of free peptide, R h,complex is the hydrodynamic radius of the peptide bound to the PLCγ1 domain, x is the concentration of the PLCγ1 domain, K D is the dissociation constant of the complex, n is the number of binding sites of the PLCγ1 domain and A is the concentration of the peptide.

NMR spectroscopy
NMR experiments were recorded at 298 K on a 600-MHz Bruker Avance III-HD spectrometer equipped with an inverse HCN cryogenic probehead (nitrogen-cooled) and running Topspin 3.2 software. 2D 15 N-HSQC spectra were recorded using States-TPPI for frequency discrimination, with water suppression achieved via a combination of WATERGATE and water flipback pulses to preserve the water magnetization [72,73]. NMR data were processed in Topspin and analysed in CcpNmr Analysis v2.4 [74].

Structure of the PLCγ1 tSH2 domain in complex with a phosphorylated pK15 peptide
A shortened version of the PLCγ1 tSH2 domain (aa 545-772) lacking the linker peptide connecting the PLCγ1 cSH2 and SH3 domains (see Results section) was cloned into a pET28a vector carrying an N-terminal His Tag and a SUMO sequence. Expression was carried out in E. coli and the protein was purified to homogeneity as described above. His-Tag and SUMO were proteolytically removed using SUMO protease (SIGMA) following the manufacturer's instructions, and the cleaved protein purified on a HisTrap column. Fractions containing cleaved protein were further purified by size exclusion chromatography (SEC) using a Superdex 75 26/600 column (GE Healthcare) equilibrated with 50mM Tris pH7.5, 150mM NaCl and 1mM TCEP buffer. The purified protein in SEC buffer was added to lyophilized phosphorylated 12mer pK15 peptide (DDLpYEEVLFPRN) in a 1:8 molar ratio (Protein: peptide) at a final concentration of 3mg/ml of peptide and 7.5mg/ml protein. Crystals were grown at 293 K using the sitting-drop vapor diffusion method in drops containing 1μl protein solution mixed with 1 μL reservoir solution containing 30% PEG 4000, 0.1M TrisHCl pH 8.5 and 0.2M sodium acetate. The crystals were flash frozen in the mother liquor containing 30% ethylene glycol and diffraction data were collected at P13 at DESY-Hamburg. Data were processed, scaled and reduced with XDS [75], Pointless [76] and programs from the CCP4 suite [77]. The phase problem was overcome by the molecular replacement method using 4FBN [63] as search model in Phaser [78]. Model building was performed using Coot [79], and refinement was done using Auto-Buster [80] with repeated validation using MolProbity [81]. Figures were generated using PYMOL (http://www.pymol.org/2/support.html).