Phosphorylation of PD-1-Y248 is a marker of PD-1-mediated inhibitory function in human T cells

PD-1 is a target of cancer immunotherapy but responses are limited to a fraction of patients. Identifying patients with T cells subjected to PD-1-mediated inhibition will allow selection of suitable candidates for PD-1-blocking therapy and will improve the therapeutic success. We sought to develop an approach to detect PD-1-mediated inhibitory signaling. The cytoplasmic tail of PD-1 contains an immunoreceptor tyrosine-based inhibitory motif (ITIM) encompassing Y223 and an immunoreceptor tyrosine-based switch motif (ITSM) encompassing Y248, which is indispensable for interaction of SHP-2 and delivery of PD-1 inhibitory function. We generated an antibody specific for phosphorylated PD-1-Y248 and examined PD-1pY248+ (pPD-1) expression in human T cells. pPD-1 was upregulated by TCR/CD3 + CD28 stimulation and simultaneous PD-1 ligation. pPD-1+CD8+ T cells were identified in human peripheral blood and had impaired effector function. pPD-1+ T cells were also detected in tumor-draining lymph nodes of tumor bearing mice and in biopsies of patients with glioblastoma multiform. Detection of pPD-1+ T cells might serve as a biomarker for identification of T cells subjected to PD-1-mediated immunosuppression.

A key mechanism by which cancer limits the host immune response is by upregulating PD-1 ligands in the tumor microenvironment (TME) which mediate ligation of PD-1 on tumor-specific CD8 + T-cells, leading to the development of T EX cells that are incapable of generating anti-tumor responses 14 . Under these conditions, therapeutic targeting of PD-1 pathway induces expansion of oligoclonal CD8 + TILs that recognize tumor neoantigens 15 , and selectively expands CD8 + memory T cells 16 . Despite the significant progress with PD-1-based immunotherapy, durable clinical responses are only observed in a fraction of patients. Thus, a central question in the field of immunotherapy is how to predict the clinical response to PD-1 blockade in order to identify suitable candidates for PD-1 immunotherapy. Studies have evaluated several markers for potential roles in shaping, or predicting therapeutic responses including PD-L1 and PD-1 expression pattern, genetic mutations and neoantigens in cancer cells, cancer epigenetics, effector T cell landscape, and microbiota but results remain inconclusive 14,[17][18][19] . Although most studies have paid attention to the inhibition of CD8 + T cells by PD-1, recently it was reported that functional systemic immunity of CD4 + T cells is required for recovery of CD8 + T cell responses after blockade of the PD-1 pathway indicating that CD4 + T cells are key targets of PD-1-mediated inhibition 20 . Together these www.nature.com/scientificreports www.nature.com/scientificreports/ extensive studies indicate that identifying which patients will respond to anti-PD-1 therapy and developing criteria for selection of suitable candidates will significantly improve the therapeutic success.
We sought to develop an approach that allows the detection of PD-1-mediated inhibitory signaling. To this end, we generated an antibody specific for phosphorylated Y248 of PD-1, which is required for interaction with SHP-2 and delivery of PD-1-mediated inhibition [21][22][23] , and examined expression and function of PD-1pY248 + T cells. We found that pPD-1 is upregulated in primary human CD4 + and CD8 + T cells during stimulation via CD3 and CD28 and PD-1 co-ligation with PD-L1. A fraction of pPD-1 + CD8 + T cells is identified in the peripheral blood of healthy individuals and displays impaired production of effector cytokines. pPD-1 + CD8 + T cells were also detected in tumor-draining lymph nodes of tumor-bearing mice. Furthermore, pPD-1 + cells were identified in biopsies of patients with glioblastoma. Our results indicate that pPD-1 identifies T cells subjected to PD-1-mediated immunosuppression, and can be detected in the context of cancer. Detection of pPD-1 + T cells in biopsies of cancer patients might serve as a biomarker for identification of patients with T cells subjected to PD-1-mediated immunosuppression, who would be suitable candidates for PD-1-checkpoint immunotherapy.

Materials and Methods
Generation of PD-1pY248 antibody. The PD-1 peptide H2N-CVPEQTE[pY]ATIVF-Ahx-KKK conjugated to KLH was emulsified in complete Freud's adjuvant was used as immunogen to raise the polyclonal rabbit antiserum 1.2 (Rockland). Rabbit sera was screened by ELISA and was selected for reactivity on PD-1 pY248 peptide and lack of reactivity to unphosphorylated peptide. Affinity purified antibody was used for western blot. For use in flow cytometry, the affinity purified antibody was conjugated using Antibody Conjugation kits for R-Phycoerythrin (R-PE) (Lightening-Link antibody labelling kit, Innova Biosciences 703-0000), PE-Cy7 (abcam ab102903) or Alexa Fluor 488 (Lightening-Link antibody labelling kit, Innova Biosciences 707-0010).
Jurkat T cells were stably transfected with PD-1, and stable lines were generated by culture with 5 µg/ml blasticidin. Before use in experiments, Jurkat T cells were rested overnight at 37 °C in RPMI-1640 containing 2% FBS and primary human or mouse T cells were rested under the same conditions for 1 hour. For pervanadate treatment, Jurkat-PD-1 T cells (5 × 10 6 cells/sample) were washed twice with PBS and resuspended in 800 ul of per-warmed (37 °C) PBS. Pervanadate was prepared by mixing 5 ml 1 mM sodium orthovanadate (Na 3 VO 4 ) with 5 ml 0.1% hydrogen peroxide (H 2 O 2 ) (both made in PBS) and incubating 15 min at RT. A total of 200 ul of the H 2 O 2 /Na 3 VO 4 mixture were added to the cells and incubated at 37 °C for the indicated time intervals. Reaction was stopped by adding 0.5 ml cold PBS and placing on ice. Cells were washed in cold PBS and lysed in lysis buffer containing 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 10% glycerol and 1% NP-40 supplemented with 2 mM sodium orthovanadate, 1 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride (PMSF), and protease Inhibitor Cocktail (Thermo Scientific). Cell lysates were resolved by SDS-PAGE and then analyzed by Western blotting. When pervanadate-treated cells were used for flow cytometry, after incubation with pervanadate for the indicated time intervals, cells were resuspended in FACS buffer (PBS 1x supplemented with 10% FBS) and washed twice. Subsequently 1 × 10 6 cells per sample were fixed using formaldehyde (1.5%) for 10 min at RT. After fixation, www.nature.com/scientificreports www.nature.com/scientificreports/   Statistics. Statistical significance was determined by two-tailed Student's t test. Statistical significance for comparison among three or more groups was determined by ANOVA (*p value < 0.05; **p value < 0.01; ***p value < 0.001).
It has been reported that SHP-2 may interact with both ITIM and ITSM of PD-1 24 but association of SHP-2 with ITSM is required for PD-1 inhibitory function 21,22 . We generated an antibody (pPD-1 1.2) specific for phosphorylated ITIM Y248 in PD-1 cytoplasmic tail by using as immunogen a phosphotyrosine peptide of PD-1 ITSM, which is conserved between mouse and human (Fig. 1A). We have previously determined that TCR proximal Src family kinases can mediate PD-1 phosphorylation required for interaction with SHP-2 25 . To confirm specificity for phosphorylated PD-1, we co-transfected COS cells with human PD-1 cDNA together with kinase active or kinase inactive Fyn. PD-1 phosphorylation was detected in the presence of kinase active but not kinase inactive Fyn (Fig. 1B). To confirm specificity of pPD-1 1.2 antibody for PD-1pY248, we used cDNA of human PD-1 in which Y223 or Y248 was mutagenized to phenylalanine. Co-expression of kinase active Fyn together with human PD-1 WT or PD-1Y223F but not PD-1Y248F resulted in detection of PD-1 phosphorylation by pPD-1 1.2 Ab (Fig. 1C) confirming specificity for PD-1 Y248.

PD-1 phosphorylation is induced in primary human T cells during TCR/CD3 and CD28 stimulation and PD-1 ligation by PD-L1.
To examine regulation of PD-1 phosphorylation in primary human T cells, we used CD3 + T cells isolated from PBMC of healthy volunteer donors. Expression of PD-1, pPD-1, PD-L1 and CD80 was examined on CD4 + and CD8 + T cells during stimulation in the presence or in the absence of PD-1 ligation. Low level PD-1 expression and PD-1 phosphorylation was detected prior to stimulation in vitro ( Fig. 2A). Stimulation of CD4 + T cells via TCR/CD3 and CD28 by using aCD3/CD28/IgG-coated beads upregulated expression of PD-1 and induced expression of PD-L1 and phosphorylation of PD-1Y248 whereas expression of CD80 was not detected (Fig. 2B). Co-ligation of PD-1 during stimulation of CD4 + T cells via TCR/CD3 and CD28 increased the intensity of pPD-1-Y248 expression (Fig. 2D) compared to stimulation without PD-1 ligation induced by aCD3/CD28 stimulation ( Fig. 2B; Supplementary Fig. S1). A similar pattern of PD-1 expression and phosphorylation was observed in CD8 + T cells ( Fig. 2A,B,D; Supplementary Fig. S1). Gating on cells that obtained PD-1 expression after stimulation with or without PD-1 ligation, showed that pPD-1 was expressed in the PD-1 + population of CD4 + or CD8 + subsets (Supplementary Fig. S2). Upregulation of PD-1 and PD-L1 and phosphorylation of PD-1 required signaling via TCR/CD3 and CD28 and were not induced by culture with PD-L1-Ig alone (Fig. 2C). These results indicate that engagement of the PD-1 pathway can be detected in human T cells by phosphorylation of PD-1Y248 using a site-specific antibody. Notably, compared to CD4 + T cells, CD8 + T cells expressed constitutively higher levels of PD-L1 prior to in vitro stimulation and a higher degree of baseline PD-1 phosphorylation (Fig. 2A).

T cells expressing PD-1Y248 phosphorylation are detected in the peripheral blood of healthy individuals.
Because our studies showed that a small fraction of primary human T cells constitutively expressed PD-1pY248 ( Fig. 2A), we sought to further characterize this population and its function. Using peripheral blood mononuclear cells from healthy donors we determined that primary T cells in the memory compartment expressed PD-1 (Fig. 3), consistent with previous observations 26 . Within this population, pPD-1 + cells could be detected within CD4 + but mostly within CD8 + T fraction, predominantly in the CD45RO + CCR7 + subset and to a significantly lower extent in the CD45RO + CCR7 − subset ( Fig. 3; Supplementary Fig. S3), which represent central memory (T CM ) and effector memory (T EM ) subsets respectively 27 . Although the overall expression of pPD-1 in these non-stimulated peripheral blood T cells is low compared to in vitro stimulated T cells, it is noteworthy that higher pPD-1 expression level was detected in CD45RO intermediate rather than CD45RO hi cells, suggesting that pPD-1 + cells are possibly in the process of reducing CD45RO expression (Fig. 3).
Previously, it was determined that the expression level of PD-1 in peripheral blood of healthy individuals did not correlate with properties of T EX cells or impaired cytokine production during stimulation in vitro 26 . Because www.nature.com/scientificreports www.nature.com/scientificreports/ phosphorylation of PD-1-Y248 is mandatory for PD-1-mediated inhibition 21-23 , we examined whether expression of pPD-1 in peripheral blood CD8 + T cells might correlate with impaired effector function. For this purpose, we assessed intracellular expression of IFN-γ and TNF-α in PD-1 + and pPD-1 + CD8 + T cells after stimulation in vitro. Although PD-1 high cells did not display diminished production of these effector cytokines, pPD-1 + CD8 + T cells had impaired expression of IFN-γ and TNF-α in response to stimulation in vitro (Fig. 4). Thus, PD-1pY248 is a marker of CD8 + T cells with impaired capacity to mount immune responses.
CD8 + CXCR5 + central memory human T cells express pPD-1. We are intrigued by the presence of pPD-1 + cells, which displayed impaired effector responses, in the peripheral blood of healthy donors and we sought to examine their properties further. Recent studies determined that a subset of PD-1 + CD8 + T EX cells that are subjected to PD-1-mediated inhibition during chronic viral infection, expresses the chemokine receptor CXCR5 28,29 . Importantly, the proliferative burst induced after PD-1 blockade in such chronically infected mice comes exclusively from this CD8 + T cell subset 29 . In humans, such cell population is detectable during chronic infections (e.g. HIV) but also during acute infection and after immunization, and resides within the T CM CD8 + subset 30 . For these reasons, we examined whether human CXCR5 + in the T CM subset might express pPD-1. Assessment of pPD-1 showed that the vast majority of peripheral blood CXCR5 + CD8 + T cells in the CD45RO + CCR7 + compartment expressed pPD-1 (Fig. 5). Together with our findings that pPD-1 + cells have impaired production of effector cytokines (Fig. 4), these results indicate that pPD-1 might identify a population of CD8 + T cells resembling the quiescent CD8 + CXCR5 + T EX cells that are subjected to PD-1-mediated inhibitory function. Furthermore, these data suggest that pPD-1 might serve as a marker of CD8 + T EX cells.

pPD-1 is expressed in the context of cancer in a mouse tumor model and patients' biopsies.
To examine whether PD-1 phosphorylation is detected in the context of cancer, we used the MC-38 colon adenocarcinoma mouse tumor model. We examined whether pPD-1 was detected in T FC -like CD8 + cells, which are recruited to tumor draining lymph nodes (DLNs), express CD44 and CXCR3 and are subjected to PD-1-mediated inhibition 31 . We found that in tumor-bearing mice, CD44 high CXCR3 + CXCR5 + CD8 + T cells expressing pPD-1 were detected in tumor DLNs but not in distal, non-tumor draining lymph nodes (NDLNs) (Fig. 6). www.nature.com/scientificreports www.nature.com/scientificreports/ To test whether pPD-1 + cells might be detected in the context of human cancer, we used biopsies from patients with glioblastoma. In three different patients, we detected expression of pPD-1 + cells in the tumor microenvironment (Fig. 7). Notably, different levels of pPD-1 expression are detected in these biopsies indicating that pPD-1 expression level might be an informative biomarker of PD-1-mediated inhibition in the context of cancer, which might vary among patients. Further studies will investigate the expression of pPD-1 and its significance in various cancers.  www.nature.com/scientificreports www.nature.com/scientificreports/ Our present studies showed that activation of the PD-1 pathway in human T cells can be identified by phosphorylation of PD-1-Y248, a site indispensable for induction of PD-1 inhibitory function [21][22][23] . Our findings point to the conclusion that assessing PD-1 phosphorylation in patients with chronic infections and cancer might serve as an indicator of PD-1 inhibitory function. For example, identification of pPD-1 + CD8 + T cells in tumor biopsies, might be informative for identifying patients who have evidence of PD-1-mediated inhibitory signaling and would be best candidates to be benefitted by immunotherapy with PD-1 checkpoint blockade. Because circulating CD8 + T cells contain high frequency of neoantigen-specific T cells within the PD-1 + fraction 32 , it will be important to determine whether such T cells reside preferentially within the pPD-1 + population in the peripheral blood of patients with cancer. Pathogen-specific CD8 + T cells subjected to PD-1 inhibitory signaling in chronic infections such as HIV, Hep C, CMV or EBV might also reside within the pPD-1 + CD8 + population. Further studies will investigate the expression and function of pPD-1 + T cells in the context of cancer and chronic infections.
In our studies, we identified expression and function of pPD-1 + CD8 + T cells in the peripheral blood of healthy individuals. PD-1 is a key mediator of peripheral tolerance 33 and aberration in the function of this pathway leads to autoimmunity in several mouse models [34][35][36][37][38] . Similarly, a regulatory polymorphism in PD-1 gene in humans is associated with susceptibility to systemic lupus erythematosus 39 . Thus, pPD-1 detection in T cells of healthy individuals may serve as an indicator of self-tolerance, whereas lack or decrease of pPD-1 expression might serve as an indicator of autoimmunity.
In conclusion, we generated a new tool to assess PD-1-mediated inhibitory signaling in human T cells, which allows the identification of immune cells with impaired effector function in vivo and might serve as a biomarker to facilitate clinically relevant decisions for selection of suitable candidate patients for PD-1 checkpoint immunotherapy.