VRK2 inhibition synergizes with PD-1 blockade to improve T cell responses

Abstract

Therapeutic programmed cell death protein 1 (PD-1) blockade enhances T cell mediated anti-tumor immunity but many patients do not respond and a significant proportion develops inflammatory toxicities. To develop better therapeutics and to understand the signaling pathways downstream of PD-1 we performed phosphoproteomic analysis of PD-1 and identified vaccinia related kinase 2 (VRK2) as a key mediator of PD-1 signaling. Using genetic and pharmacological approaches, we discovered that VRK2 is required for PD-1-induced phosphorylation of the protein p21 activated kinase 2 (PAK2), and for the inhibition of IL-2, IL-8, and IFN-γ secretion. Moving into in vivo syngeneic tumor models, pharmacologic inhibition of VRK2 in combination with PD-1 blockade enhanced tumor clearance through T cell activation. This study suggests that VRK2 is a unique therapeutic target and that combination of VRK2 inhibitors with PD-1 blockade may improve cancer immunotherapy.

Introduction

Despite its clinical utility, the molecular pathways engaged by the checkpoint PD-1 remain poorly defined.PD-1 lacks intrinsic enzymatic activity, and instead recruits other proteins to mediate its inhibitory function. In T cells, T cell receptor (TCR) recognition of antigens presented by major histocompatibility complex (MHC) molecules and PD-1 colocalization with the TCR are absolute requirements for PD-1 function. Following antigen recognition, PD-1 binds to its ligands, programmed death ligand 1 (PD-L1) and programmed death ligand 2 (PD-L2), expressed on tumor cells and antigen presenting cells (APC) ^{1, 2, 3, 4}. Ligand binding leads to tyrosine phosphorylation of the immune tyrosine inhibitory motif (ITIM) and immune tyrosine switch motif (ITSM) within the cytoplasmic tail of PD-1, which subsequently recruits the tyrosine phosphatase Src homology 2 (SH2) domain containing tyrosine phosphatase 2 (SHP2) ⁵. Subsequently, SHP2 dephosphorylates proteins critical for proximal TCR signaling, such as CD3, zeta chain of T cell receptor associated protein kinase 70 (ZAP70), CD28, IL-2 inducible T cell kinase (ITK) and Rap guanine nucleotide exchange factor 1 ^{6, 7, 8, 9, 10}.

Our knowledge regarding the phosphorylation networks triggered by PD-1 and that interfere with proximal and distal TCR signaling is limited. This significant knowledge gap undermines our understanding of PD-1 function in T cells and hampers the development of improved therapeutics targeting the PD-1 axis. To address this, we have previously interrogated the PD-1-regulated phosphoproteome using mass spectrometry ¹¹. Through this approach we identified that the extent of PD-1-regulated phosphorylation networks extends far beyond proximal TCR signaling. In fact, the functional consequences of PD-1 signaling lies at the intersection of phosphorylation networks associated with TCR signaling, cytoskeletal organization, cell cycle, gene expression, and protein translation ¹¹.

In the current work we further analyzed this data and focused on VRK2 as a potential effector of PD-1 signaling, as many of its predicted substrates were phosphorylated upon PD-1 ligation according to the phosphoproteome analysis. VRK2 is a serine/threonine protein kinase that its expression at the RNA level is widespread, and at the protein level is higher in proliferating cells ¹². VRK2 has two isoforms: VRK2A (VRK2), localized in the cytoplasm and anchored to the endoplasmic reticulum, and VRK2B, a shorter isoform generated by alternative splicing, found in the nucleus ¹³. VRK2 phosphorylates nuclear factor of activated T cells 1 (NFAT1) which increases NFAT1-dependent transcription, resulting in increase in COX-2 gene expression and cancer cell invasion¹⁴. Furthermore, VRK2A can modulate signaling pathways independent of its kinase activity, through interaction with C-Jun N-Terminal Kinase 1-Interacting Protein 1 (JIP1), resulting in the reduction of the stress response to hypoxia ¹⁵ and to interleukin-1β ¹⁶. While NFAT1 is an important mediator of lymphocytes activation ^{17, 18}, the effect of VRK2 on lymphocytes has never been explored.

In this work we found that the ability of PD-1 to inhibit cytokine secretion from lymphocytes was reduced in the absence of VRK2 and interfering with VRK2 function in vivo augmented the anti-tumoral effect of PD-1 blockade. These data suggest that cancer patients might benefit from combining VRK2 inhibitors with anti-PD1 blocking antibody.

RESULTS

Prediction analysis of PD-1 phosphoproteome identifies VRK2 as potential downstream effector

As previously reported, we analyzed the phosphoproteome of PD-1-activated T cells and carried motif prediction analysis and database search for known substrates to determine which kinases phosphorylated proteins that were significantly differentially phosphorylated following PD-1 ligation ¹¹. To validate if the predicted kinases mediate signaling downstream of PD-1, we generated shRNA knock-down Jurkat T cell lines of the most represented kinases (Fig. 1A in Data in Brief). All the cell lines produced IL-2 following overnight stimulation with aCD3+aCD28 (Fig. 1B in Data in Brief). Next, we measured IL-2 production in response to TCR stimulation in the presence of PD-1 ligation to determine the consequences of kinase knockdown on PD-1 function (Fig. 1C in Data in Brief). Interestingly, compared to the parental Jurkat T cells, the vaccinia-related kinase 2 (VRK2) knockdown cells were the least susceptible to PD-1 mediated inhibition of IL-2 production (Fig. 1C in Data in Brief). Moreover, VRK2 was predicted to phosphorylate 26% of all differentially phosphorylated serine and threonine substrates downstream of PD-1 (Fig. 1).

Figure 1. In silico prediction analysis of kinase/substrate interactions downstream of PD-1.

Treeplot representing the proportion of targeted phosphosites by each kinase (rectangle size) and percent IL-2 inhibition following plate-bound stimulation of the different kinase knock-down Jurkat T cells lines (color intensity). Kinases within the white rectangles were not included in the shRNA knock-down screen; nt not tested.

VRK2 is required for PD-1 mediated inhibition of cytokine secretion

To determine whether VRK2 knockdown abrogates primary T cell responses to PD-1 ligation, human primary T cells were transfected with siRNA targeting VRK2 and similarly to Jurkat T cells, PD-1 ligation was not able to efficiently inhibit IL-2 (Fig. 2A), IFN-γ (Fig. 2B), and IL-8 (Fig. 2C) secretion.

Figure 2. VRK2 is required for PD-1 inhibitory functions.

(A) IL-2, (B) IFN-γ, and IL-8 (C) inhibition in response to plate-bound stimulation in the presence of PD-L2 following siRNA knockdown of VRK2 in primary human T cells. Data from n=5–6 independent experiments, *p≤0.05. (D) Schematic of PD-1, TCR, VRK2 and PAK2 interaction sequence. Western blot analysis (E) and quantification (F) of PAK2 S197 phosphorylation in Jurkat T cells left untreated, following shRNA knockdown of VRK2 (shVRK2) and in Jukrat T cells treated with AZD-7762 inhibitor. Statistical analysis was performed using unpaired Student’s t test, where *p≤0.05.

The disrupted phosphorylation landscape of proteins involved in immune synapse formation following PD-1 ligation and the great proportion of VRK2 targets identified in our kinase screen, suggested that VRK2 may target key proteins regulating the actin cytoskeletal dynamics. Indeed, the phosphoprotemic analysis showed an further increase in phosphorylation of PAK2 ¹¹, a protein involved in cytoskeletal organization, that is a potential target of VRK2 according to the in silico analysis ¹¹ (Fig. 2D). Therefore, we sought to validate our in silico predictions in Jurkat T cells with and without pharmacological VRK2 inhibition as well as in shRNA VRK2 knockdown Jurkat T cell line. As demonstrated previously ¹¹, PD-1 ligation increased S197 phosphorylation of PAK2 but not PAK1 or PAK3 (Fig. 2E), and, as expected, shRNA knockdown of VRK2 as well as pharmacologic inhibition with a VRK2 inhibitor, AZD-7762 ¹⁹, fully inhibited PAK2 S197 phosphorylation following PD-1 ligation (Fig. 2E and 2F). Collectively, these findings suggest that VRK2 may be also a downstream effector of PD-1 signaling, targeting serine and threonine phosphosites.

PD-1 blockade synergizes with VRK2 inhibition to improve anti-tumor T cell responses

Our in vitro results, which suggest that VRK2 may be an effector of PD-1 signaling, led us to hypothesize that in vivo pharmacologic inhibition of VRK2 may augment T cell responses by directly targeting the PD-1 axis and potentially increasing T cell activation additively with PD-1 blockade in a combination therapy. We took advantage of the MC38 syngeneic mouse tumor model to determine whether VRK2 inhibition alone or in combination with PD-1 blockade would improve anti-tumor immune response (Fig. 3A). Since there are no VRK2 specific inhibitors, we used AZD-7762, a selective inhibitor of both VRK2 and the checkpoint kinases (CHK) ¹⁹. This agent did not cause additional cytotoxicity when cultured in vitro with MC38 cells (Fig. 2B in Data in Brief). AZD-7762 treatment alone led to significantly decreased tumor volume compared to vehicle or aPD-1 antibody treatment (Fig. 3B). To exclude the possibility that AZD-7762 acts primarily by inhibiting tumor growth via CHK in the tumor cells, we also determined tumor growth kinetics in response to treatment with the two other CHK-specific inhibitors, Prexasertib (LY2606368) ²⁰ and PF-477736 ²¹. Compared to these drugs, AZD-7762 treatment led to marked decrease in MC38 tumor volume (Fig. 2C in Data in Brief) indicating that its activity is not solely mediated via CHK inhibition. Furthermore, the combination of AZD-7762 with PD-1 blockade augmented anti-tumor responses compared to PD-1 blockade alone or AZD-7762 treatment alone (Fig. 3B). This anti-tumor response was consequential to increased proportion of CD8⁺ T cells in tumor draining lymph nodes (tdLN) of treated mice (Fig. 3C). Furthermore, the combination therapy led to significantly increased proportion of activated CD44⁺CD69⁺ CD4⁺ and CD8⁺ T cells in the tdLN, elevated proportion of CD44⁺ PD-1⁺ T cells in both spleen and tdLN (Fig. 2D in Data in Brief), and significantly increased LAG-3⁺ CD8⁺ T cells (Fig. 3C).

Figure 3. VRK2 inhibition synergizes with PD-1 blockade to augment anti-tumor T cell responses.

(A) Graphical representation of MC38 mouse tumor protocol used in this study. MC38 mouse tumor cells were implanted subcutaneously in the right hind flank of mice, and once tumor volume reached 100mm³ (day 0), treatment with AZD-7762 was started and continued until day 17. Anti-PD-1 antibodies were given on day 0 and day 7. (B) MC38 tumor growth curves following administration of the indicated treatments. The data represents mean ± SEM of tumor volume measured in nine mice of three independent experiments. (C) Proportion of CD4⁺ and CD8⁺ T cells in tdLN expressing CD69, PD-1, LAG-3 and CD44. (D) CD4⁺ and CD8⁺ T cell depletion (T^DEP) was performed by intraperitoneal injection of anti-CD4 and anti-CD8 antibodies on days 3 and 5. (E) Graphical representation of MC38 mouse tumor protocol used in this study. MC38 mouse tumor cells were implanted subcutaneously in the right hind flank of mice, and once tumor volume reached 100mm³ (day 0), treatment with AZD-7762 was started and continued until day 17. T cell depletion was induced on day 3 and day 5. (F) MC38 tumor growth curves in T cell depleted mice measured in three mice of two independent experiments. (G) Graphical representation of MC38 mouse tumor protocol used in this study. MC38 mouse tumor cells were implanted subcutaneously in the right hind flank of mice, and once tumor volume reached 100mm³ (day 0), T cell were adoptively transferred as indicated. A second cell transfer therapy was given on day 4. Anti-PD-1 antibodies were administrated on day 0 and day 7. (H) MC38 tumor growth curves in mice, measured in three to six mice of two independent experiments. The effects of different treatments on tumor volume in (B), (F), and (H) were compared by repeated measures two-way ANOVA and Tukey’s multiple comparisons test with individual variances computed for each comparison.

Next, we sought to confirm that AZD-7762 acts primarily on the T cell compartment to exert its anti-tumor activity. To this end, prior to the AZD-7762 administration we depleted CD4⁺ and CD8⁺ T cells (Fig. 3D) and determined tumor growth rate (Fig. 3E). Importantly, T cell depletion fully reversed the anti-tumor activity of AZD-7762, confirming that its therapeutic activity is mediated by the T cell compartment (Fig. 3F). To definitely attribute the limited tumor growth specifically to VRK2 and more directly to T cells, we used an adoptive cell transfer of wildtype or VRK2 knockout T cells into MC38 tumor bearing wildtype mice (Fig. 3G). We found that transfer of VRK2 deficient T cells alone was sufficient to limit tumor growth (Fig. 3H).

Tumor growth in mice treated with PD-1 blockade in combination with adoptively transferred VRK2 deficient cells were marginally smaller compared to mice treated just with anti-PD-1 antibodies (Fig. 3H). These findings reveal VRK2 as potential therapeutic target downstream of PD-1 and suggest that combining PD-1 blockade with VRK2 inhibition can augment anti-tumor T cell responses.

Discussion

Reinvigorating T cell-mediated anti-tumor immunity is a prerequisite for the development of successful immunotherapies. Therapeutic blockade of PD-1 has proven to be successful in many cancers ²². However, not all patients have durable responses to these therapies and a significant proportion of patient experiences immune toxicity ²³, emphasizing the critical role of PD-1 in immune tolerance. To develop more successful therapeutics targeting the PD-1 axis, it is imperative that we understand the underlying molecular mechanism and signaling pathways engaged by PD-1. Previously, we carried an unbiased phosphoproteomic screen of PD-1-triggered pathways following ligation with PD-L2 in Jurkat T cells ¹¹ and revealed multiple molecular tiers of PD-1 regulation, extending beyond the known TCR proximal signaling events. These molecular networks converged into three specific functional groups encompassing leukocyte activation, cellular adhesion and control of gene expression, ultimately leading to attenuated T cell function ¹¹.

In an attempt to look for potential complementary treatments to PD-1 blockade based on the phosphoproteomic screen, we determined the putative kinases that might phosphorylate PD-1 targeted phosphosites and confirmed their function in vitro. One quarter of the S and T sites were substrates for VRK2. Knockdown of this kinase led to markedly diminished PD-1 inhibition of cytokines secretion, suggesting that VRK2 may act as a downstream effector of PD-1. In fact, our kinase screen predicted that VRK2 mediated PAK2 S197 phosphorylation downstream of PD-1 ligation. We validated these predictions by western blot and confirmed that PAK2 S197 is phosphorylated by VRK2 downstream of PD-1. Importantly, we have also demonstrated that PAK2 is phosphorylated upon TCR activation and the knock-down of VRK2 also abrogates TCR-induced PAK2 phosphorylation (Fig. 2D and 2E), which corroborates previously published data ²⁴. Functionally, PAK2 is necessary for TCR signaling ²⁵, and thus it seems that PAK2 phosphorylation mediates both TCR and PD-1 signaling, despite their apparently opposite functions. Indeed, SHP2, the most notable mediator of PD-1 signaling and function ⁵, is also partially responsible for TCR-induced signaling ²⁶. Thus, these opposite roles are not unique to PAK2. Since both TCR and PD-1 induce cytoskeleton remodeling for modulation of the immunological synapse ^{6, 27}, and PAK2 mediates cytoskeleton remodeling ²⁵, PAK2 may serve as a mediator for both receptors.

Notably, pharmacological inhibition of VRK2 augmented the ability of PD-1 blockade to reduce tumor growth. This response was T cell dependent since T cell depletion reversed the anti-tumor activity observed following VRK2 inhibition. Deleting VRK2 specifically in T cells confirmed the role of this kinase in T cell reposes to MC38 tumors. Interestingly, VRK2 can induce cancer cell invasion ¹⁴, thus its inhibition may prevent tumor cell invasion and enhance anti-tumor immune response. However, the effect on cancer cell invasion in the MC38 model was negligible. The pharmacological inhibitor that we used, AZD-7762, induced cardiotoxicity in clinical trials, and thus was abandoned. Currently there are no specific inhibitors for VRK2, as this kinase family is very insensitive to inhibitors due to their structural characteristics ²⁸.

In conclusion, our findings point towards VRK2 as a novel downstream effector of PD-1, and ongoing experiments with more specific VRK2 inhibitors should define its role as a therapeutic target in cancer immunology. Importantly, the concept of double targeting of the same pathway is being evaluated for other pathways in cancer, such as the epidermal growth factor receptor (EGFR) pathway, for which a combination of a blocking antibody (Cetuximab) and tyrosine kinase inhibitors are being explored in clinical trials ²⁹, based on preclinical studies that showed that the combined treatment is additive or even synergistic ³⁰, probably due to incomplete blocking of the pathway by either agents. Thus, our results raise the possibility of a dual anti-PD-1 strategy involving both a blocking antibody and small molecules.

Materials and methods

General reagents

RPMI 1640 medium, DMEM, Dulbecco’s PBS, and FBS were purchased from Life Technologies. Opti-MEMI was purchased from Invitrogen. Ficoll-Paque was purchased from Stem Cell. BCA assay was purchased from Pierce Biotechnology. Puromycin was obtained from Sigma-Aldrich.

Cell culture, transfection, and stimulation

Peripheral blood was purchased from New York blood center. Total CD3⁺ T cells were isolated by density gradient centrifugation (Lymphoprep) and negative selection using the RosetteSep human T cell enrichment cocktail (Stem Cell). Primary T cells were directly used in stimulation assays. In vitro T cell cultures were maintained in complete RPMI, containing 10% FBS, MEM nonessential amino acids, 1mM sodium pyruvate, 100 IU/ml of penicillin, 100 μg/ml streptomycin and GlutaMAX-I. Jurkat T cells were obtained from the American Type Culture Collection and maintained in RPMI 1640 medium supplemented with 10% FBS and 100 U/ml penicillin and streptomycin. MC38 cells were provided by Kerafast and maintained in DMEM medium supplemented with 10% FBS and 100 U/ml penicillin and streptomycin. HEK 293T cells were obtained from the American Type Culture Collection and maintained in 5% CO₂ at 37°C in DMEM media supplemented with 10% FBS and 100 U/ml penicillin and streptomycin. Cells were stimulated with magnetic beads (ratio of 1:5 cells per bead), which were conjugated with the following protein combinations (the ratio in parentheses indicates the relative concentration of each protein): anti-CD3/IgG1 (1:3), anti-CD3/PDL2-Fc/IgG1 (1:2:1), anti-CD3/anti-CD28/IgG1 (1:1:2), or anti-CD3/anti-CD28/PD-L2-Fc (1:1:2).

Antibodies

Anti-CD3 (UCHT1), and PD-L2-Fc were purchased from Acros. IgG1 (isotype control) was purchased from Jackson ImmunoResearch. Anti-CD28 (CD28.2) was purchased from eBioscience. Anti-mouse antibodies were purchased from BioLegend: CD3-AF488 (clone 17A2), CD8-PercpCy5.5 (clone 53–6.7), CD4-PE (clone GK1.5), CD44-BV421 (clone IM7), PD-1-PECy7 (clone RPM1–30), CD62L-BV711 (clone MEL-14), CD69-BV711 (clone H1.2F3), and LAG-3-PE (clone C9B7W). For the in vivo experiments, anti-mouse-PD-1 antibody was purchased from BioXcell (clone RMP1–14).

Cytokine secretion

IL-2, IL-8, and IFN-γ concentrations in the supernatant were measured by enzyme linked immunosorbent assay (ELISA) from BioLegend.

Western blot

To determine PAK phosphorylation, 5×10⁶ Jurkat T cells were washed with cold PBS and resuspended in complete RPMI. The cells were stimulated for 5 minutes with Dynabeads at 37°C as indicated above (beads/cells ratio of 3:1). The samples were treated with Calyculin A (CST 9902S) at 100 nM for 25 minutes at 37°C, washed once with cold PBS and centrifugation at 500 × g and lysed with RIPA buffer supplemented with PhosStop (Roche) for 1hr at 4°C. Following 10 second sonication and centrifugation at 12,000 × g for 10 minutes at 4°C, the samples were blotted using Phospho-PAK1 (Ser199/204)/PAK2 (Ser192/197) antibody (Cell Signaling). Blots were imaged and analyzed using Licor Odyssey.

Knocking down PD-1 related kinases

Kinases were stably knocked down in Jurkat T cells by short hairpin RNA using Mission shRNA plasmids (Sigma-Aldrich). Lentiviral particles were generated by transfecting HEK 293T cells with pMD2G, psPAX2, and the shRNA plasmid using SuperFect (Qiagen). T cells were transduced by centrifugation and selected with puromycin. SMARTpool ON-TARGETplus VRK2 and nontargeting control small interfering siRNA (Dharmacon) were used according to the manufacturer’s instruction.

RT-PCR analysis

Total RNA was extracted using the RNeasy Plus Mini Kit (Qiagen). RNA (500 ng) was used for cDNA synthesis using SuperScript II First Strand Synthesis (Invitrogen). Human kinases and HPRT Taqman Primer/Probes were used for all Taqman Gene Expression Assays with the Taqman Universal PCR Master Mix (Applied Biosystems). Quantitative gene expression analyses were performed with Applied Biosystems 7300 Real-Time PCR. Gene expression was analyzed by the ΔΔCt method.

Mice, MC38 tumor inoculation and T cell analysis

One million MC38 cells were implanted subcutaneously in the right hind flank of mice. Tumor growth was monitored using electronic calipers and calculated according to the formula: V = length × width² × 0.52. For T cell phenotypic analysis by flow cytometry, spleens and inguinal lymph nodes were harvested 17 days post-treatment initiation. Splenic and tdLN cells were stained with anti-mouse antibodies for flow cytometry analysis. To deplete T cells, mice received intraperitoneal injection of 200 μg anti-CD4 (BioXcell BE0003) and 200 μg anti-CD8 (BioXcell BE0061) antibodies in PBS, a second dose of the antibodies was administered two days later. To asses cytotoxicity, MC38 cells were thawed, seeded at a density of 3 × 10⁴ in a flat bottom 96-well plate and treated overnight at 37°C and 5% CO₂ with AZD-7762 (MCE HY-10992) at the indicated concentrations. Cell viability was measured with PrestoBlue (Invitrogen).

Statistical analysis

GraphPad Prism software was used for statistical analysis. Unpaired Student’s t test was used to compare differences between the means of two groups and a two-tailed p-value ≤0.05 was considered statistically significant, where * p<0.05. To compare the effects of different treatments on tumor volume, we used repeated measures two-way ANOVA and Tukey’s multiple comparisons test with individual variances computed for each comparison.

Highlights.

• Vaccinia related kinase 2 (VRK2) is a novel downstream effector of program cell death-1 (PD-1) signaling.

• Pharmacologic inhibition of VRK2 in combination with PD-1 blockade enhanced tumor clearance through T cell activation.

• VRK2 is a unique therapeutic target and combination of VRK2 inhibitors with PD-1 blockade may improve cancer immunotherapy.