The conformational state of the BTK substrate PLCγ contributes to Ibrutinib resistance.

Abstract

Mutations in PLCγ, a substrate of the tyrosine kinase BTK, are often found in patients who develop resistance to the BTK inhibitor Ibrutinib. However, the mechanisms by which these PLCγ mutations cause Ibrutinib resistance are unclear. Under normal signaling conditions, BTK mediated phosphorylation of Y783 within the PLCγ cSH2-linker promotes the intramolecular association of this site with the adjacent cSH2 domain resulting in active PLCγ. Thus, the cSH2linker region in the center of the regulatory gamma specific array (γSA) of PLCγ is a key feature controlling PLCγ activity. Even in the unphosphorylated state this linker exists in a conformational equilibrium between free and bound to the cSH2 domain. The position of this equilibrium is optimized within the properly regulated PLCγ enzyme but may be altered in the context of mutations. We therefore assessed the conformational status of four resistance associated mutations within the PLCγ γSA and find that they each alter the conformational equilibrium of the γSA leading to a shift toward active PLCγ. Interestingly, two distinct modes of mutation induced activation are revealed by this panel of Ibrutinib resistance mutations. These findings, along with the recently determined structure of fully autoinhibited PLCγ, provide new insight into the nature of the conformational change that occurs within the γSA regulatory region to affect PLCγ activation. Improving our mechanistic understanding of how B cell signaling escapes Ibrutinib treatment via mutations in PLCγ will aid in the development of strategies to counter drug resistance.

Graphical Abstract

Introduction

Ibrutinib (Imbruvica), is a covalent, irreversible inhibitor that targets the B-cell non-receptor tyrosine kinase, Bruton’s tyrosine kinase (BTK). Ibrutinib and related second-generation inhibitors are emerging as frontline drugs for the treatment of B cell cancers such as chronic lymphocytic leukemia (CLL), Mantle cell lymphoma (MCL) and Marginal zone lymphoma (MZL) (1–3). Despite the success of Ibrutinib, development of drug resistance is a recurring problem (4, 5). The majority of the Ibrutinib resistance mutations that have been identified in patients occur within BTK (6). These BTK resistance mutations either interfere with effective Ibrutinib binding or activate BTK (4, 7). In addition, 30% of patients administered Ibrutinib develop resistance mutations in the substrate of BTK, Phospholipase C gamma 2 (PLCγ2) (4, 8). The mechanisms by which PLCγ mutations confer Ibrutinib resistance are less clear.

PLCγs are enzymes that catalyze the hydrolysis of phosphatidylinositol 4, 5-bisphosphate (PIP₂) to diacylglycerol and inositol 1, 4, 5-trisphosphate (9). PLCγ2, the major isoform expressed in B cells, as well as its ubiquitously expressed isoform PLCγ1 (10), are multi-domain proteins with a PLCγ core composed of an N-terminal Pleckstrin homology (PH) domain, EF hands, catalytic X and Y domains and a C-terminal C2 domain (Fig. 1a). A defining feature of the gamma isoform of PLCs is the presence of the gamma specific array (γSA), which is inserted between the catalytic X and Y domains (9). The γSA consists of a split PH (sPH) domain, two Src homology 2 (SH2) domains (N-terminal SH2 (nSH2) and C-terminal SH2, (cSH2)), a long linker (cSH2-linker) that contains the regulatory phosphorylation site, and a Src homology 3 (SH3) domain (Fig. 1a). Recent crystal and cryoEM structures of PLCγ1 in the autoinhibited conformation (11, 12) show that the γSA module assembles over the PLC core (Fig. 1b). Critical interactions between the cSH2 and sPH domains with the PLCγ core C2 and catalytic domains, respectively, maintain the autoinhibited state of the enzyme by blocking productive interactions between PLCγ and the PIP₂ substrate in the lipid bilayer (Fig. 1b). It is noted that the long cSH2-linker (residues 766 – 790) between the cSH2 and the SH3 domain is not defined in the autoinhibited structures (Fig. 1b).

Figure 1:

Domain organization of PLCγ. (a) Full-length PLCγ has a PLC core (dark blue) consisting of a PH domain, EF hands, the catalytic X and Y domains and a C2 domain. The gamma specific array (γSA) is inserted between the X and Y domains of the core and consists of a split PH domain (sPH, green), two SH2 domains (N-terminal SH2, nSH2 (brown) and C-terminal SH2, cSH2 (purple)) and an SH3 domain (cyan). The linker between the cSH2 domain and the SH3 domain (cSH2-linker, light blue) contains a conserved tyrosine phosphorylation site (Y783) associated with activation of PLCγ1. The Ibrutinib resistance mutations (PLCγ1 numbering) are indicated within the γSA. (b) Crystal structure of full-length PLCγ1 (PDB ID: 6PBC) in the autoinhibited conformation. The color scheme for the domains is the same as in (a). The γSA (ribbon) rests above the PLCγ core (dark blue surface). The cSH2-linker is absent from the crystal structure of full-length PLCγ1 and is indicated as a light blue dotted line. The Ibrutinib resistance mutations are indicated on the structure (sticks and transparent spheres). (c) Crystal structure of the nSH2-cSH2-linker phosphorylated on Y783 (PDB ID: 4EY0). The nSH2 and cSH2 domains are surface rendered and the linker region is depicted as a ribbon with the side chain of pY783 in red. Binding of the phosphorylated cSH2-linker to the cSH2 domain would sterically block binding of cSH2 to the C2 domain. (d) Close up of the residues mutated in Ibrutinib resistance patients (PLCγ1 numbering): (i) P686 and R687, (ii) S729 and E730, (iii) Y509. (e) Constructs that are used in this study.

The cSH2-linker region contains the conserved tyrosine phosphorylation site (PLCγ2 Y759 or PLCγ1 Y783) required for activation of PLCγ (Fig. 1a) (9). While phosphorylation on PLCγ1 Y783 is necessary and sufficient for PLCγ1 activation, mechanistic details of the PLCγ activation process are otherwise poorly defined. It has been previously demonstrated that phosphorylation on Y783 leads to intramolecular association of the phosphorylated cSH2-linker with the cSH2 domain (Fig. 1c) (13). The intramolecular pY783/cSH2 interaction is mutually exclusive with the interaction between cSH2 domain and the C2 domain, and so this interaction will destabilize the autoinhibited state of PLCγ1 It is also possible that disruption of the cSH2/C2 interaction could destabilize the sPH domain interaction with the catalytic domain thereby allowing insertion of the hydrophobic ridge of the PLCγ catalytic domain into PIP₂ lipid bilayers. Additionally, previous biochemical analyses and crystal structures have shown that the cSH2-linker associates with the cSH2 domain even in the absence of phosphorylation on Y783 (13). This finding suggests that, prior to phosphorylation triggered activation, the PLCγ cSH2-linker region samples a conformational equilibrium between cSH2-associated and dissociated states. Association of the cSH2-linker with the cSH2 domain would transiently open the autoinhibitory contact between cSH2 and C2 domain possibly priming the enzyme for full activation.

This emerging model of PLCγ activation suggests that mutations that shift the conformational equilibrium of the cSH2-linker could lead to cellular changes in PLCγ function by altering the optimal conformational equilibrium of the autoinhibited enzyme. Mutations that promote activated PLCγ could bypass BTK inhibition by Ibrutinib and thereby contribute to Ibrutinib resistance. We therefore assessed the conformational state of the PLCγ cSH2-linker in the context of four different Ibrutinib resistance mutations observed in patients: P664S, ΔS707/A708, S707Y and Y495H (14–16). Due to difficulty in bacterial expression of PLCγ2 for structural characterization, the mutations are characterized using PLCγ1 whose numbering will be used throughout this manuscript. The Ibrutinib resistance mutants in PLCγ2 correspond to PLCγ1 P686S, ΔS729/E730, S729Y and Y509H (Fig. 1a,b,d).

Biophysical analysis shows that the resistance mutations in PLCγ studied here have divergent effects on the conformational state of the cSH2-linker. Sequence changes within the cSH2 domain, PLCγ1 P686S, ΔS729/E730 and S729Y decrease the binding of the cSH2-linker to the cSH2 domain. This results in increased exposure of the cSH2-linker thereby increasing the likelihood of phosphorylation on Y783 which favors the intramolecular interaction between cSH2-linker and cSH2 and disfavors the autoinhibited form of PLCγ. Conversely, mutation of Y509H in the sPH domain increases cSH2-linker association with the cSH2 domain. An increase in the population of cSH2 domain bound to cSH2-linker disfavors association of the cSH2 domain with the C2 domain thereby disrupting the autoinhibited conformation and shifting the equilibrium toward active PLCγ1 even in the absence of phosphorylation on Y783. Thus, the conformational status of the cSH2-linker is altered by mutations in the PLCγ cSH2 and sPH domains, which can contribute to activation of PLCγ and bypass inhibition of BTK. An improved understanding of the mechanisms by which these mutations work allows us to develop better treatment options in the face of drug resistance.

Results

NMR spectral changes report on equilibrium between the cSH2-linker and cSH2 domain.

To test the conformational effects of resistance mutations within PLCγ, we compare NMR spectra of wild type and mutant PLCγ1 γSA. Experiments focus on the complete γSA (sP-nSH2-cSH2-linker-SH3-H) and the shorter cSH2-linker-SH3 fragment (Fig. 1e). Partial NMR assignments are available for the PLCγ1 γSA (13) and of the available assignments (~40% of the γSA residues), we were able to transfer 92% to the spectra acquired using our samples (details in methods).

One Ibrutinib resistance mutation, PLCγ2 R665W (corresponding to PLCγ1 R687W), has been previously characterized in the context of both the γSA and nSH2-cSH2-linker fragment by NMR (17). In that work, the backbone amide resonances of G772 and G777 within the cSH2-linker region report on the conformational state of the PLCγ1 cSH2-linker (Fig. 2a,b). We repeated this experiment for the cSH2-linker-SH3 fragment used in our work and confirm previous findings that the R687W mutation of PLCγ1 results in dissociation of the PLCγ1 cSH2-linker from the cSH2 domain as indicated by an upfield shift in the proton dimension of the G772 resonance and a downfield shift in the G777 resonance in a ¹H-¹⁵N TROSY-HSQC spectrum (Supp. Fig. S1a).

Figure 2:

NMR analysis of the interaction between the cSH2-linker and the cSH2 domain within γSA. (a) Representation of the equilibrium between the cSH2-linker bound (left) and unbound (right) to the cSH2 domain. The cSH2 domain is shown in purple and bound to the cSH2-linker (cyan); structure of the cSH2-linker/cSH2 complex (left) is derived from PDB ID: 4EY0. The positions of the NMR probes G772 and G777 on the cSH2-linker are indicated as blue spheres (G777 is missing in the crystal structure, therefore approximate position is shown). The position of Y783 in the cSH2-linker is labeled. The unbound or released state of the cSH2-linker is represented as a dotted line in the model on the right. (b) Previous work (17) established the backbone resonance of G772 and G777 as reporters for the conformational equilibrium between cSH2-linker bound and unbound to the cSH2 domain (see (a)). The previously characterized R687W mutation causes the conformational ensemble to shift toward the unbound state. This is indicated by an upfield shift for the G772 resonance and a downfield shift for the G777 resonance; here and in (c) the wild type resonances are black and mutant are cyan. (c) Overlay of the ¹H-¹⁵N TROSY HSQC spectra of PLCγ1 γSA wild-type and mutants. Based on previous characterization of the G772 and G777 resonances (17), the shifts in the proton dimension observed for each mutant can be assigned to a shift in the conformational equilibrium shown in (a) to either an increase or decrease in the population of cSH2-linker bound to cSH2. P686S (i), ΔS729/E730 (ii), and S729Y (iii), shift the conformational equilibrium toward release of the cSH2-linker. Y509H (iv) shifts the conformational equilibrium of γSA in the opposite direction toward association of the cSH2-linker with the cSH2 domain.

The PLCγ1 P686S and ΔS729/E730 mutations decrease cSH2-linker binding to the cSH2 domain.

Having confirmed the G772 and G777 backbone amide resonances report on the conformational equilibrium of the intramolecular cSH2/cSH2-linker interaction, we next examined two additional Ibrutinib resistance mutations within the cSH2 domain: P686S and ΔS729/E730. P686 is located at the interface where cSH2, sPH, cSH2-linker, and the SH3-sPH linker converge in the autoinhibited structure of PLCγ1 (Fig.1b, d(i)). S729/E730 is located on the other side of the cSH2 domain in the βF strand and lies at the interface between the cSH2 domain and the core C2 domain in the autoinhibited structure of PLCγ1 (Fig.1b, d(ii)).

Comparison of ¹H-¹⁵N TROSY-HSQC spectra of both wild-type PLCγ1 γSA (Fig. 2c(i,ii)) and the cSH2-linker-SH3 fragment (Supp. Fig. S1b(i, ii)) to that of the P686S and ΔS729/E730 variants shows an upfield shift in G772 and a downfield shift in G777 resonances in both proteins. P686 is adjacent to R687 and likely to disrupt the same domain interface within γSA and so it is unsurprising that the P686S mutation causes NMR changes that are very similar to R687W. These data are consistent with a mutation induced shift in the conformational population of γSA (and the smaller cSH2-linker-SH3 fragment) toward the more dissociated cSH2-linker for both P686S and ΔS729/E730 (equilibrium depicted in Fig. 2a).

Chemical shift mapping of the cSH2-linker interaction.

Use of the shorter cSH2-linker-SH3 fragment in these experiments is driven by the fact that resonances corresponding to backbone amides of the PLCγ1 cSH2 domain are not visible in ¹H-¹⁵N TROSY-HSQC spectra of the entire PLCγ1 γSA region likely due to intermediate timescale dynamics (13). Thus, using the assigned cSH2 domain resonances present in spectra of the shorter cSH2-linker-SH3 fragment in combination with resonance assignments for the other domains within γSA, we monitored mutation induced changes in each domain of the γSA region. The chemical shift changes due to the P686S mutation localize to cSH2 (Fig.3a, Supp Fig. S2a). Limited changes are also observed in the sPH domain which could indicate allosteric changes upon mutation in the cSH2 domain. Significant chemical shift changes were not detected in other domains of the γSA region in the context of the P686S mutation. The ΔS729/E730 deletion causes extensive chemical shift changes in the cSH2 domain and the sPH domain, with more limited changes extending into the nSH2 domain and SH3 domains (Fig 3c, Supp Fig. S2b). The data suggest that the deletion causes more significant changes to the autoinhibited γSA structure than does the single P686S mutation (Fig. 3a,c). The entire ¹H-¹⁵N TROSY-HSQC spectrum of the ΔS729/E730 double mutant shows chemical shift dispersion consistent with that of a folded protein and overlays well with the wildtype protein (Supp. Fig. S2b shows the superposition for the cSH2-linker-SH3 fragment) suggesting that the chemical shift changes observed upon deletion within the cSH2 domain reflect conformational adjustments and not non-specific denaturation.

Figure 3:

NMR and HDX-MS analysis of Ibrutinib resistant mutations in the γSA region of PLCγ. For P686S (a,b), ΔS729/E730 (c,d), S729Y (e,f), and Y509H (g,h), changes in chemical shift values for backbone resonances with γSA caused by mutation and differences in deuterium exchange between wild type γSA and mutant are shown on the structure of γSA derived from the full-length autoinhibited PLCγ1 structure (PDB ID: 6PBC). NMR chemical shift changes shown (a,c,e,g) represent the combined changes that occur due to the indicated mutation in the PLCγ1 γSA construct as well as the shorter cSH2-linker-SH3 domain construct. Residues showing chemical shift changes greater than 0.02 ppm are shown as red sticks and transparent spheres. Residues showing no changes are in teal. The location of each mutation is indicated. Observed HDX-MS changes due to each mutation are mapped onto the structure of PLCγ1 γSA (b,d,f,h); regions showing an increase in deuterium exchange is indicated in green and decrease in exchange is mapped in blue. Pink regions are not followed due to mutation and salmon regions indicate no data.

To further test the impact of the PLCγ1 P686S mutation and ΔS729/E730 deletion, we next compared these Ibrutinib resistant variants to wild-type γSA using HDX-MS. The data show that mutation of P686S or deletion of S729/E730 lead to an increase in deuterium uptake in the cSH2 domain and the cSH2-linker supporting the interpretation that these sequence changes shift the conformational equilibrium toward unbound cSH2-linker (Figs. 3b,d and 4a,b). No significant differences in deuterium uptake between the wild-type and P686S mutant are observed outside of the cSH2 domain and cSH2 linker region. In contrast, increased deuterium uptake for the S729/E730 deletion extends into the SH3 and nSH2 domains of γSA consistent with the NMR data acquired for the S729/E730 deletion. As well, the sPH domain of the γSA S729/E730 deletion is slightly more protected from exchange than wild type (Fig. 4b) suggesting that allosteric changes resulting from release of the cSH2-linker propagate across the entire γSA region of PLCγ.

Figure 4:

HDX-MS data for a curated set of peptides that are coincident across all γSA proteins studied (WT, P686S, ΔS729/E730, S729Y, Y509H and ΔsPH fragment). The position of the domains of PLCγ1, as described in Figure 1a, is shown at the left and the amount of time in deuterium is shown left to right for each γSA variant. The relative difference data represent changes between wild type and mutant greater than 1.0 Da (dark blue indicates decreased exchange in the mutant compared to wildtype and dark green indicates increased exchange in the mutant compared to wildtype); modest differences between 0.5 Da and 1.0 Da are shown as light blue (decrease) and light green (increase).

The PLCγ1 S729Y mutation has a modest impact on the γSA.

In addition to deletion of residues from the βF strand of the cSH2 domain, Ibrutinib resistance has also been observed in patients with a single mutation (S729Y) in this region. Using the same NMR and HDX-MS approaches already described, we find that the S729Y mutation results in only small upfield and downfield shifts of the G772 and G777 resonances, respectively (Fig. 2c(iii), Supp. Fig. S1b(iii)). Mapping the chemical shift changes due to the S729Y mutation onto the structure of PLCγ shows perturbations within cSH2 that are a smaller subset of those observed for the S729/E730 deletion (Fig. 3c,e, Supp Fig. S2b,c). Limited spectral changes were also detected in the sPH domain similar to the P686S mutant and, unlike the S729/E730 deletion, chemical shift changes were not detected in the SH3 and nSH2 domains upon mutation of S729Y (Fig. 3a,c,e). HDX-MS analysis of the S729Y mutant γSA shows no significant differences in deuterium uptake compared to the wild-type protein in contrast to the P686S and the S729/E730 deletion (Figs. 3f, 4c). This is consistent with the modest NMR spectral changes observed with the S729Y mutation.

The PLCγ1 Y509H mutation increases association of cSH2-linker with cSH2.

The Y509 residue in the sPH domain lies at the interface of the sPH domain and the catalytic domain in the autoinhibited structure of full-length PLCγ1 (Fig.1b,d(iii)). Comparison of the ¹H-¹⁵N TROSY-HSQC spectrum of wild-type PLCγ1 γSA to that of the Y509H mutant shows minor shifts in the G772 and G777 resonances in a direction that is opposite from the other mutations studied (G772 shifts downfield and G777 moves upfield (Fig. 2c(iv)). This observation suggests that the conformational ensemble of the PLCγ1 γSA Y509H mutant might favor increased association of the cSH2-linker (unphosphorylated) with the cSH2 domain. Mapping the overall chemical shift changes onto the structure of PLCγ γSA shows chemical shift changes occur primarily in the sPH domain upon mutation of Y509 (Fig. 3g, Supp Fig. S2d). The SH3 domain shows no chemical shift perturbations and the nSH2 shows limited chemical shift changes upon mutation of Y509 to His (Fig. 3g). It is important to note that changes in the cSH2 domain could not be monitored in the context of the γSA construct due to significant line broadening as reported previously (13).

To further test the hypothesis that the PLCγ1 Y509H mutation promotes a shift in the equilibrium toward association of the cSH2-linker with the cSH2 domain, we next probed the impact of the Y509H mutation by HDX-MS. Comparison of PLCγ1 wild-type γSA with the Y509H mutant by HDX-MS shows that the Y509H mutation results in decreased deuterium uptake in the βD and βE strands of the cSH2 domain (Figs. 3h, 4d). The observation that the cSH2 domain is protected from exchange upon mutation of Y509H in the sPH domain is consistent with the direction of small chemical shift change observed for the G772 and G777 amide resonances (Fig. 2c(iv)) and provides further evidence that mutation of Y509H shifts the conformational equilibrium of γSA toward association of the cSH2-linker with the cSH2 domain in the absence of phosphorylation. The only additional region that shows differences in exchange is the increased deuterium uptake for the Y509H mutant in the β1 strand of the sPH domain, perhaps due to local effects of the Y509H mutation or a consequence of allosteric changes (Fig. 3h).

Ibrutinib resistance mutations in PLCγ alter the efficiency of Y783 phosphorylation by BTK

Given the chemical shifts changes of G772 and G777 in the cSH2-linker, the spectral changes induced upon mutation within the cSH2 domain, and HDX data indicating increased exposure of cSH2 backbone protons to solvent exchange, we hypothesized that Y783 within the cSH2-linker is more accessible in the P686S, ΔS729/E730, and S729Y mutants and might therefore allow increased phosphorylation on Y783 compared to wild type γSA. In contrast, mutation of Y509H results in small but opposite changes in the G772 and G777 resonances and decreased solvent exchange for regions within the cSH2 domain consistent with stabilization of the cSH2-linker association with the cSH2 domain. Thus, the Y509H mutation may lead to conformation changes that protect Y783 from phosphorylation by an exogenous kinase. To test the effects of γSA mutations on Y783 accessibility, we assessed the ease with which Y783 is phosphorylated in vitro for the panel of Ibrutinib resistance mutants.

In previous work we have shown that the cSH2-linker association with the cSH2 domain regulates phosphorylation on Y783 (18, 19). Examining the remote resistance mutations in the sPH and cSH2 domains using the same approach, we find that the Y509H mutation does not alter the level of Y783 phosphorylation by BTK compared to wild type PLCγ1 SA (Fig. 5a,b). It is likely that since the conformational equilibrium of the cSH2-linker is shifted predominantly to the bound (to the cSH2 domain) state in the isolated γSA fragment (see discussion in the next section), we are not able to detect a further decrease in Y783 phosphorylation levels in the Y509H mutant compared to the wild-type protein given the limited dynamic range/sensitivity of western immunodetection. In contrast, the three variants that shift the conformational equilibrium away from cSH2-linker association with the cSH2 domain (P686S, ΔS729/E730, and S729Y), are all phosphorylated on Y783 to a greater extent than wild type γSA (Fig. 5a,b). These data suggest that three of the four Ibrutinib resistant PLCγ mutants studied here might lead to increased levels of pY783 via cellular kinases allowing mutated PLCγ to bypass activation by BTK (20). Alternatively, the Y509H mutation does not lead to increased pY783 levels but instead appears to stabilize the association between unphosphorylated cSH2-linker and the cSH2 domain. Increased population of the cSH2-linker/cSH2 interaction would interfere with autoinhibitory contacts between cSH2 and the core C2 domain limiting the population of the fully autoinhibited enzyme and shifting the equilibrium of PLCγ toward a more open, active state. Indeed, the location of Y509 at the sPH/catalytic domain interface suggests that the Y509H mutation directly alters this autoinhibitory contact. To further understand how mutation in the sPH domain of PLCγ renders the enzyme resistant to Ibrutinib without increasing pY783, we next examined changes in the conformational equilibrium of the cSH2-linker/cSH2 complex upon wholescale deletion of the sPH domain.

Figure 5:

Kinase assays reveal the relative ease of BTK mediated phosphorylation on Y783 for wild type and mutant PLCγ1. (a) Western blot comparing the phosphorylation on Y783 of PLCγ1 γSA wild-type (WT) and the Y509H, P686S, S729Y and ΔS729/E730 mutants by BTK. Phosphorylation on Y783 was monitored using the anti PLCγ1 pY783 antibody and the total protein levels were monitored by Ponceau S staining. BTK enzyme levels were monitored using the Anti-BTK antibody. (b) Histogram quantifying the western blots shown in (a). The blots were quantified and normalized as described in the Materials and Methods. Data shown are the average of three independent experiments.

Deletion of the split PH domain from the γSA increases cSH2-linker association with the cSH2 domain.

Deletion of the sPH domain has been previously shown to activate full-length PLCγ (21). Disruption of the interaction between the sPH domain and the catalytic domain in the autoinhibited conformation of PLCγ has been proposed as the likely mechanism for this activation. To assess whether the sPH domain has additional regulatory roles, specifically whether this domain influences the cSH2-linker/cSH2 domain interaction, we compared the PLCγ1 γSA fragment with a deletion construct that lacks the sPH domain (ΔsPH, nSH2-cSH2-linker-SH3 fragment, Fig. 1e) by HDX-MS. Deletion of the sPH domain leads to a small decrease in deuterium uptake in the nSH2 domain and a greater decrease in exchange within the cSH2 domain and the cSH2-linker (Fig. 4e). These data suggest that the interaction between cSH2 and cSH2-linker is enhanced when the sPH domain is deleted (Fig. 4e). Thus, in addition to directly blocking the catalytic domain (see Fig. 1b), the presence of the sPH domain also disfavors the interaction between the cSH2linker and the cSH2 domain in the full γSA. This is consistent with the sPH domain playing a role in maintaining the optimal equilibrium between autoinhibited PLCγ and conformations that ‘prime’ the enzyme for activation. Interestingly, the changes observed upon deletion of the sPH domain from the γSA are similar to the changes observed in the Y509H mutant (Fig. 4d and e), suggesting that the Y509H mutation also decouples the sPH domain from the cSH2 domain.

To further explore the coupling between the sPH domain and cSH2 domains of PLCγ, we carried out an NMR titration with increasing concentration of unlabeled sPH domain added to ¹⁵N labeled cSH2 domain in trans. A subset of cSH2 backbone resonances undergo changes in resonance frequency over the course of the titration (Fig. 6a,b) and the residues corresponding to these spectral changes map to a contiguous surface of cSH2 domain (Fig. 6c). Surprisingly, sPH binding to cSH2 does not map the sPH/cSH2 interface evident in the crystal structure of full-length, autoinhibited PLCγ1 (Fig. 6d). Instead, as separate domains in solution, the sPH domain binds to the opposite side of cSH2 making contact to the beta sheet comprised of the end of the βD strand, βE and βF (Fig. 6c,d).

Figure 6:

Interaction of the split PH (sPH) and cSH2 domains. (a) NMR titration of unlabeled sPH domain into ¹⁵N-labeled cSH2. Histogram showing changes in chemical shift values for each cSH2 resonances upon addition of excess sPH domain. The solid line represents the mean and the dashed line is mean plus one standard deviation. Residues corresponding to resonance changes greater than the mean plus one standard deviation are considered significant and mapped onto the cSH2 domain in the autoinhibited γSA structure in (c). (b) Select regions of the ¹H-¹⁵N HSQC spectra are shown to illustrate the resonance changes during the titration of sPH into cSH2. Resonances corresponding to cSH2 domain residues L726 and G727 are labeled. (c) Structure of PLCγ1 cSH2 domain showing subset of residues that exhibit resonance changes upon addition of sPH in yellow. (d,e) NMR chemical shift mapping of the sPH/cSH2 interaction fits only one of two distinct arrangements of the PLCγ γSA. (d) The sPH interaction interface on the cSH2 domain is colored yellow on the surface rendering of cSH2 and shown within a cartoon representation of the fully autoinhibited structure presented in Fig. 1b. (e) Surface rendered cSH2 domain from crystal structure (PDB ID: 4EY0) of the nSH2-cSH2-linker fragment containing pY783 (presented in Fig. 1c) where yellow surface indicates interaction interface with sPH domain. The cartoon representation for this model is the γSA domain arrangement previously described based on SAXS and NMR methods (13). The PLC core was not included in these earlier solution studies. The relative arrangement of the domains is compatible with the restraints placed by linker lengths when the cSH2-linker is bound to the cSH2 domain.

The interaction site between sPH and cSH2 mapped by NMR is reminiscent of an earlier model of the γSA derived from SAXS, solution NMR and biochemical data (13). In that work the γSA was studied in isolation revealing a domain arrangement that is distinct from that of the γSA region within the full-length, autoinhibited PLCγ crystal structure (Fig. 6e). It is immediately evident that the solution-based model of the γSA (Fig. 6e) predicts that the sPH domain contacts the cSH2 domain in the precise location mapped by NMR (Fig. 6a). The differences in the γSA domain arrangement in the two models (Fig. 6d,e) are consistent with a large-scale rearrangement of the γSA domains upon PLCγ activation predicted in previous biochemical studies (21). Moreover, the crystal structure of the nSH2-cSH2-linker fragment containing pY783 (Fig. 1c) is compatible with the SAXS derived model (Fig. 6e) and the path of the cSH2-linker region across the cSH2 domain in the crystal structure of phosphorylated nSH2-cSH2-linker would directly interfere with autoinhibitory contacts between cSH2 and the core C2 domain of PLCγ. Moreover, based on the cSH2-linker length constraints, binding of the cSH2-linker to the cSH2 domain would require the displacement of the SH3 domain from its location in the autoinhibited full-length model. Given the propensity of the cSH2-linker to associate with the cSH2 domain even in the absence of phosphorylation on Y783, it is likely that the conformational equilibrium of the isolated γSA fragment is shifted to the cSH2-linker-bound state in the absence of the competing C2 domain in the full-length protein. Thus, with the benefit of the now available structure of autoinhibited full-length PLCγ1, we hypothesize that the previously described model of the isolated γSA (Fig. 6e) reflects the active conformation of PLCγ γSA rather than the inactive conformation ascribed to this structure at the time the model was developed (13). Future structural characterization of the active form of full-length PLCγ will confirm this hypothesis.

Discussion

A major advance in our understanding of PLCγ regulation was ushered in with the recent structures of full-length autoinhibited $\text{PLCγ}\{\text{Hαφ1χεκ},2019\#886;\Lambda 10,2020\#887\}$. The work we present here now provides a detailed examination of the conformational equilibrium between the fully autoinhibited γSA and a ‘pre-primed’ state that may be sampled prior to cellular activation (Fig. 7a,b). It is likely that the conformational ensemble of the isolated γSA studied here favors the cSH2-linker/cSH2 bound state (Fig. 7b,d) to a greater extent than full-length PLCγ (the C2 domain within full-length will compete for binding to cSH2). Indeed, the deletion of twenty-five amino acids within the cSH2-linker required to successfully crystalize the autoinhibited full-length PLCγ did not change the hydrodynamic radius of PLCγ1 compared to wild-type, consistent with the conformational equilibrium of full-length PLCγ1 lying predominantly on the side of the fully autoinhibited enzyme (Fig. 7a). At the same time, the fact that deletion of the cSH2-linker region was required for crystallization argues that the full-length enzyme dynamically samples a conformational equilibrium in solution (Fig 7a,b). Thus, the study of the isolated γSA has revealed conformational features that are likely more difficult to detect within the full-length PLCγ protein but nevertheless likely at work controlling the optimal distribution across the conformational ensemble to avoid spurious PLCγ mediated signaling.

Figure 7:

Activation model for PLCγ and effects of Ibrutinib resistance mutations. This model builds on the previously published model for activation of PLCγ by FGFR1 (11). (a) Autoinhibited PLCγ based on crystal structure (PDB ID: 6PBC). Domain colors are as in Fig. 1b, cSH2-linker is indicated with a dotted line and the approximate position of Y783 in the cSH2-linker is shown. The yellow line on the cSH2 domain indicates the position of the kinase docking site on the PLCγ cSH2 domain; albeit obscured by the cSH2/C2 interaction of the autoinhibited structure (25). (b) A pre-primed state characterized by dissociation of the cSH2 domain with the C2 domain of the PLC core exposing the kinase docking site (yellow oval). Double headed arrow indicates the transient association of unphosphorylated Y783 with cSH2. The effect of the different Ibrutinib resistance mutations in PLCγ are shown above and below the equilibrium arrows between the preprimed and active state (shown in (d)). (c) On receptor activation, BTK and PLCγ are recruited to the BCR complex via specific binding to phosphotyrosines in the adaptor protein SLP-65 (alternatively named BLNK or BASH) (38). The C-lobe of the BTK kinase domain (or ITK in T cells) docks onto the PLCγ cSH2 domain creating a primed state that can productively phosphorylate PLCγ Y783. (d) Once phosphorylated, the cSH2-linker/cSH2 domain interaction is stabilized and the γSA is released from the PLC core allowing association of the core with lipid membrane and subsequent PLCγ mediated hydrolysis of the PIP₂ substrate. Domain rearrangement within the γSA (to that suggested in the model shown in Fig. 6e) may accompany full activation of PLCγ.

Resistance mutations identified in patients demonstrate that activation of PLCγ can bypass Ibrutinib inhibition of BTK and propagate signals emanating from the BCR (20). Given the importance of Y783 within the PLCγ cSH2-linker in PLCγ activation, it is not surprising that the PLCγ Ibrutinib resistance mutations alter the conformational status of this linker. The changes in the conformational equilibrium of the cSH2-linker observed here in the resistance mutants and their predicted impact on phospholipase activity are completely consistent with previous in vitro and/or in vivo PLC activity measurements (4, 11, 22–24). The activity of the S707Y mutant in response to EGF stimulation was ~9 fold higher than that of the wild-type protein (23). The PLCγ2 S707/A708 double deletion (PLCγ1 ΔS729/E730) mutant had a ~22 fold increase in activity compared to the wild-type protein (23). These activity measurements correlate well with the magnitude of the changes observed here with the double deletion mutant which showed more extensive changes compared to the single S729Y mutation. Moreover, the activity of the S707Y mutant was sensitive to phosphorylation on Y759 (PLCγ1 Y783) unlike what has been observed with the PLCγ1 Y509H mutant (23, 24). Again, these results are consistent with the mechanism of action of these mutants uncovered in this study. The Y509H mutation directly promotes cSH2-linker association with the cSH2 domain (thereby activating PLCγ) and is therefore insensitive to phosphorylation on Y783 whereas the S729Y mutation (which decreases cSH2-linker association with the cSH2 domain) is still reliant on Y783 phosphorylation to promote cSH2-linker association with the cSH2 domain to activate PLCγ. While the activity of the PLCγ1 P686S mutant has not been measured, the activity of the R687W mutant (which is adjacent to P686) has been measured, and shows an increase in activity in vitro and in vivo consistent with our structural predictions (4, 11, 22, 23). Interestingly, additive/synergistic effects of combining certain resistance mutations have also been seen previously supporting the importance of each domain/region in PLCγ regulation (23, 24).

Phosphorylation of Y783 requires a specific docking interaction between the ITK or BTK kinase domain and residues spanning the BG loop in the PLCγ cSH2 domain (19, 25). The docking surface on PLCγ is obscured in the fully autoinhibited state (Fig. 7a) but may become accessible upon release of cSH2 from the C2 domain in the core (Fig. 7b). Additionally, we have previously shown that the T-cell adaptor protein SLP-76 pY173 (equivalent to SLP-65 in B-cells, Fig. 7c) binds to the cSH2 domain of PLCγ and aids in the phosphorylation of Y783 by promoting the unbound state of the cSH2-linker (18). We have therefore defined a ‘pre-primed’ state (Fig. 7b) where the PLCγ γSA is poised to associate with cognate kinase and SLP-65 (Fig. 7b). Transient association of unphosphorylated Y783 with cSH2 may contribute to maintenance of this pre-primed state. The ‘primed’ state (Fig. 7c)) leads to productive Y783 phosphorylation and subsequent conformational changes within γSA resulting in exposure of the hydrophobic ridge to allow active PLCγ to associate with the membrane (Fig. 7d). Regardless of how it is achieved, whether by normal phosphorylation of the cSH2-linker tyrosine that associates with cSH2 domain and limits autoinhibitory interactions with the core or whether a mutation such as Y509H allosterically increases association of the unphosphorylated cSH2-linker with the cSH2 domain, any perturbation of the optimal equilibrium between the autoinhibitory and pre-primed state has the potential to alter PLCγ activity.

In addition to cellular activation by phosphorylation on the cSH2-linker, PLCγ2 can be activated by binding to Rac1 via its sPH domain (26, 27). These previous studies have shown that the PLCγ2 Y495C Ibrutinib resistance mutation (corresponding to PLCγ1 Y509H) has increased sensitivity to activation by Rac1 in vivo. However, Rac1 binding to the sPH domain occurs solely in PLCγ2 and not in PLCγ1, whereas the Y509H or Y495H mutation in the sPH domain is activating in both PLCγ1 and PLCγ2, respectively (24). This suggests a common activation mechanism that could be applicable to both PLCγ proteins upon mutation of Y509/Y495 in the sPH domain. Our data shows that the PLCγ1 Y509H mutation shifts the γSA toward the active PLCγ1 conformation, a mechanism that is likely at work within PLCγ2 as well. Thus, the PLCγ2 Y495H mutation could have pleiotropic effects: shifting the conformational equilibrium of PLCγ2 away from the optimal conformational equilibrium that maintains proper autoinhibition as well as increasing susceptibility to activation by Rac1.

Even though the focus on the conformational equilibrium within γSA likely revealed important features that might be more difficult to study in the full-length PLCγ protein, it is nevertheless important to consider the limitations inherent in working with protein fragments. The crystal structure of full-length PLCγ1 shows that the Y509H mutation lies at the interface of the sPH domain and the catalytic domain (Fig. 1b, d(iii)). It is therefore likely that the Y509H mutation in the context of the full-length protein also destabilizes the interaction of the sPH domain with the catalytic domain. Similarly, the S729Y mutation in the cSH2 domain lies at the interface of the cSH2 domain and the C2 domain, potentially destabilizing this autoinhibitory interaction (Fig. 1b, d(ii). In fact, previous studies have demonstrated that the S729Y mutation in the cSH2 domain decreases the affinity of the cSH2 domain for the PLCγ core when compared to the wild-type cSH2 domain (13). Decreased interaction between the cSH2 domain and the C2 domain in the S729Y mutant would make the cSH2 domain available for docking on the kinase domain of BTK enabling phosphorylation on the cSH2-linker (19, 25), further shifting the conformational equilibrium away from the autoinhibitory conformation. Alternatively, increased docking of the PLCγ substrate with BTK may be responsible for the increased phosphorylation on Y783 observed for this mutant (Fig. 5a,b) despite the relatively weak NMR and HDX changes. Additionally, recent work has shown that PLCγ2 S707Y (equivalent to PLCγ1 S729Y), has increased sensitivity to activation in cells harboring either active or inactive BTK (28). Along the same lines, the PLCγ2 R665W and S707Y mutations (equivalent to the PLCγ1 R687W and S729Y) have been shown to have increased susceptibility to activation by Rac2 (22, 23). Thus, mutations that cause Ibrutinib resistance likely impact multiple domain/domain or protein interactions that act in concert to shift the equilibrium to an active population of PLCγ that escapes Ibrutinib inhibition.

A mechanistic understanding of drug resistance mechanisms has been instrumental in developing methods to combat resistance and treat cancer (29–31). In the current study, the P686S, S729Y and ΔS729/E730 mutations promote cSH2-linker dissociation from the cSH2 domain which in turn could promote phosphorylation on Y783 suggesting that efforts to decrease phosphorylation on Y783 via inhibition of BTK or other upstream kinases such as SYK is a viable treatment strategy in patients that develop these mutations. However, in patients that develop the Y509H mutation, inhibiting phosphorylation on Y783 (via BTK or other upstream kinase inhibition) would not be a viable treatment strategy since the Y509H mutation promotes phosphorylation independent association of the cSH2-linker with the cSH2 domain and PLC activation. Alternate approaches such as directly inhibiting PLCγ or other downstream kinases could be more effective under these circumstances. Moreover, a mechanistic understanding of these PLC mutations highlight the regulatory interfaces within the protein that could be targeted by drug design to modulate PLC activity. Thus, a full understanding of the mechanistic details will guide both the choice of treatment approaches and the development of PLCγ inhibitors to overcome drug resistance.

Materials and Methods

Constructs and reagents:

The bacterial expression constructs for rat PLCγ1 isolated cSH2 domain (H663-E759) and γSA fragment (D485-R936 with a C-terminal 6X-His tag) have been described previously (18). The PLCγ1 nSH2-cSH2-linker-SH3 (NCL3/ΔsPH) fragment (G540-D863 with a C-terminal 6X-His tag), cSH2-linker-SH3 fragment (Q659-D863), isolated nSH2 domain (E548-Q659), Linker-SH3 fragment (N757- D863) and isolated sPH domain (D485-N529/N865-R936 with a C-terminal 6X-His tag) were cloned into the pGEX-2T or −4T vectors. The bacterial expression construct for full-length murine BTK carrying the solubilizing Y617P mutation has been described elsewhere (19). All mutations were made using the site directed mutagenesis kit (Agilent), and the sequences confirmed by sequencing at the Iowa State University DNA facility.

Protein expression and purification:

The expression and purification of PLCγ1 cSH2 domain, cSH2-linker and γSA fragment have been described elsewhere (18, 25). All PLCγ1 constructs were expressed in Escherichia coli BL21 cells (Millipore Sigma). Briefly, the cultures were grown at 37 °C to an O.D. 600 nm of 0.6 to 0.8. The temperature of the culture was then lowered to 20 °C and induced with 1.0 mM IPTG. The culture was harvested 24 hours after induction and the pellets were resuspended in lysis buffer (50 mM KH₂PO₄ (pH 6.4), 150 mM Sodium chloride, 2 mM DTT and 0.02 % Sodium azide and 0.5 mg/ml lysozyme) and stored at −80°C. Cells were lysed by thawing and the action of lysozyme, and 3000 U DNAse I (Sigma) and 1 mM PMSF were added to the lysate, incubated at RT for 20 minutes and then spun at 16,000 rpm for one hour at 4 °C. The supernatant was loaded onto two 5 ml Glutathione Sepharose columns (GE healthcare) and washed with lysis buffer without lysozyme. The proteins were eluted, concentrated and cleaved with Thrombin (γSA was cleaved for 2 h at RT, nSH2-cSH2-linker-SH3 (ΔsPH), cSH2-linker-SH3 and sPH domain overnight at RT). The proteins were further purified by size exclusion chromatography (Hiload Superdex 26/60 200 pg or 75 pg GE Healthcare). The fractions containing pure protein were pooled, concentrated and stored at 4° C. The expression and purification of full-length, kinase active BTK has been described previously (32).

NMR:

Uniformly ¹⁵N labeled PLCγ1 samples were produced by growth in modified M9 minimal media containing ¹⁵N ammonium chloride (1g/L, Cambridge Isotope Laboratories, Inc.) as the sole source of nitrogen. The final NMR sample buffer consists of 50 mM KH₂PO₄ (pH 6.4), 150 mM Sodium chloride, 2 mM DTT and 0.02 % Sodium azide. All NMR spectra were acquired at 298 K for the γSA PLCγ1 fragment or at 303 K for all other PLCγ1 fragments on a Bruker AVIII HD 800 spectrometer equipped with a 5 mm HCN z-gradient cryoprobe operating at a ¹H frequency of 800.35. NMR samples consisted of 150–200 μM ¹⁵N labeled PLCγ1 unless specified otherwise. All data were analyzed using NMRViewJ (33).

NMR Assignments:

Partial assignments for the PLCγ1 γSA have been published previously (13) and a high-resolution figure of the spectrum with peak labels was kindly provided upon request by Paul Driscoll. Assignments are also available for the isolated cSH2 domain (BMRB ID 5318) and the nSH2-cSH2-linker fragment (BMRB ID 27496) of PLCγ1. Assignments for the PLCγ1 γSA and cSH2-linker-SH3 fragments were successfully transferred using both the high-resolution figure and deposited assignments listed above (Supp. Figs. S3 and S4). Additionally, where possible, the transferred assignments were crosschecked using peak overlap between the spectra of the γSA or cSH2-linker-SH3 and the isolated cSH2, isolated nSH2 or isolated SH3 domain fragments. The cSH2 domain resonances are broadened beyond detection in the PLCγ1 γSA as noted previously (13) but can be detected in the shorter cSH2-linker-SH3 fragment. Overall coverage post assignment transfer was as follows: ~37% for PLCγ1 γSA (~47% sPH, ~45% nSH2, ~22% cSH2 and ~51% SH3 domain) and ~46% for the cSH2-linker-SH3 fragment (~47% cSH2 and ~51% SH3 domain).

HDX-MS:

HDX-MS studies were performed using methods modified from those reported previously for BTK (7, 32, 34). In addition to the descriptions below, comprehensive experimental details and parameters are provided within the Supplemental Datafile, in the recommended (35) tabular format. All HDX-MS data have been deposited to the ProteomeXchange Consortium via the PRIDE (36) partner repository with the dataset identifier PXD028923.

To prepare for labeling, all PLCγ1 constructs were diluted from stock concentrations to 22 μM using 25 mM KH₂PO₄ pH 6.4, 150 mM NaCl, 2 mM DTT, H₂O. Deuterium labeling was initiated with a 36-fold dilution into D₂O buffer (36 μL, 25 mM KH₂PO₄ pD 6.43, 150 mM NaCl, 2 mM DTT, 99.9% D₂O). After each labeling time (10 seconds, 1 minute, 10 minutes, 1 hour, and 4 hours) at 20 °C, the labeling reaction was quenched with the addition of 37 μL of ice-cold quenching buffer 150mM potassium phosphate, pH 2.44, H₂O and analyzed immediately.

Deuterated and control samples were digested online with pepsin at 15 °C using a Waters Enzymate pepsin column. The cooling chamber of the UPLC system, which housed all the chromatographic elements, was held at 0.0 ± 0.1 °C for the entire time of the measurements. Peptides were trapped and desalted on a VanGuard Pre-Column trap [2.1 mm × 5 mm, ACQUITY UPLC BEH C18, 1.7 μm, for 3 minutes at 100 μL/min. Peptides were then eluted from the trap using a 5%–35% gradient of acetonitrile over 6 minutes at a flow rate of 100 μL/min, and separated using an ACQUITY UPLC HSS T3, 1.8 μm, 1.0 mm × 50 mm column. The error of determining the deuterium levels was ± 0.25 Da in this experimental setup. To eliminate peptide carryover, a wash solution [1.5 M guanidinium chloride, 0.8% formic acid and 4% acetonitrile] was injected over the Enzymate column during each analytical run. Mass spectra were acquired using a Waters Synapt G2-Si HDMS^E mass spectrometer in ion mobility mode. All comparison experiments were done under identical experimental conditions such that deuterium levels were not corrected for back-exchange and are therefore reported as relative (37).

Peptides were identified using PLGS 3.0.1 using replicates of undeuterated control samples and raw MS data were imported into DynamX 3.0 for initial automated processing followed by manual inspection. The relative amount of deuterium in each peptide was determined by the software by subtracting the centroid mass of the undeuterated form of each peptide from the deuterated form, at each time point, for each condition. These deuterium uptake values were used to generate all uptake graphs and difference maps found in the Supplemental Datafile.

Activity assays:

In vitro kinase assays were performed by incubating 0.1 μM BTK FL-Y617P enzyme with 1.0 μM GST-tagged PLCγ1 γSA wild-type or mutants in a kinase assay buffer (50 mM Hepes pH 7.0, 10 mM MgCl₂, 1 mM DTT, 5 % glycerol, 1 mM Pefabloc, and 200 μM ATP) at room temperature for 10 minutes. The reactions were stopped by the addition of SDS-PAGE loading buffer and the samples were boiled, separated by SDS−PAGE, and Western blotted with the anti-PLCγ1 phospho-Y783 antibody (EMD Millipore) or anti-BTK antibody (BD Biosciences) as described previously (7, 25). The bands were quantified using ImageJ. The phosphorylation levels (Anti-PLCγ1 pY783 blot) were normalized to the total PLCγ1 substrate level (Ponceau stain levels). The PLCγ1 wild-type value was set to 1 and compared to the mutants.

Highlights:

1. Conformational preference of the cSH2-linker modulates PLCγ activity.

2. Wildtype PLCγ adopts an optimal equilibrium between cSH2-linker associated with or dissociated from the adjacent cSH2 domain.

3. Ibrutinib resistance mutations in PLCγ alter the conformational equilibrium of the cSH2linker.

4. In addition to the regulatory phosphorylation of Y783, the sPH domain regulates association of cSH2-linker with cSH2 domain.

5. Mechanistic understanding of drug resistance will aid development of treatment strategies.

Abbreviations:

PLCγ

Phospholipase C gamma

γSA

gamma specific array

BTK

Bruton’s tyrosine kinase

NMR

Nuclear magnetic resonance

HDX-MS

hydrogen/deuterium exchange-mass spectrometry

CLL

chronic lymphocytic leukemia

MCL

Mantle cell lymphoma

MZL

Marginal zone lymphoma

SAXS

Small angle X-ray scattering

PIP2

Phosphatidylinositol 4, 5-bisphosphate