Pembrolizumab Interferes with the Differentiation of Human FOXP3 +−Induced T Regulatory Cells, but Not with FOXP3 Stability, through Activation of mTOR This information is current as of September 28, 2021. Varun Sasidharan Nair, Salman M. Toor, Ghina Taouk, Gerald Pfister, Khalid Ouararhni, Nehad M. Alajez and Eyad Elkord J Immunol published online 4 December 2019 http://www.jimmunol.org/content/early/2019/12/10/jimmun Downloaded from ol.1900575

Supplementary http://www.jimmunol.org/content/suppl/2019/12/03/jimmunol.190057 http://www.jimmunol.org/ Material 5.DCSupplemental

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 28, 2021 • Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2019 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published December 11, 2019, doi:10.4049/jimmunol.1900575 The Journal of Immunology

Pembrolizumab Interferes with the Differentiation of Human FOXP3+–Induced T Regulatory Cells, but Not with FOXP3 Stability, through Activation of mTOR

Varun Sasidharan Nair,* Salman M. Toor,* Ghina Taouk,* Gerald Pfister,† Khalid Ouararhni,* Nehad M. Alajez,* and Eyad Elkord*,‡

Programmed cell death 1 (PD-1) is critical for T regulatory cells (Tregs) to maintain peripheral tolerance to self-antigens. In the tumor microenvironment, interaction between PD-1 and its ligands supports tumor immune evasion. Pembrolizumab blocks in- teractions of PD-1 with its ligands, enhancing antitumor and clinical responses. We and others have reported that pembrolizumab does not affect function or phenotype of thymic-derived Tregs; however, little is known about its effect on extrathymic differen-

tiation of peripheral Tregs. In this study, we investigated the effect of pembrolizumab on in vitro–induced Tregs (iTregs). Our work Downloaded from showed that PD-1 blockade interferes with iTreg differentiation and has no potential effect on the stability of FOXP3 after differentiation. Additionally, we found that both nontreated and pembrolizumab-treated iTregs were suppressive. However, pembrolizumab-treated iTregs were relatively less suppressive in higher Treg ratios and failed to produce IL-10 compared with their nontreated counterparts. Different methods including transcriptomic analyses confirmed that the downregulation of FOXP3 was mediated by activating mTOR and STAT1 and inhibiting MAPK pathways, shifting the iTreg polarization in favor of Th1 and

Th17 subsets. To confirm the role of mTOR activation, we found that rapamycin diminished the effect of pembrolizumab-mediated http://www.jimmunol.org/ downregulation of FOXP3. Ingenuity pathway analysis revealed that pembrolizumab-treated iTregs showed upregulation of genes promoting DNA repair and immune cell trafficking, in addition to downregulation of genes supporting cellular assembly and organization. To our knowledge, this is the first study to show that pembrolizumab interferes with differentiation of human FOXP3+ iTregs and to disclose some of the molecular pathways involved. The Journal of Immunology, 2020, 204: 000–000.

orkhead box protein 3+ T regulatory cells (Tregs) include their cognate Ags. However, the composition of FOXP3+ Treg thymic-derived Tregs (tTregs)/natural Tregs (1), which subsets within the tumor microenvironment (TME) or in circula- originate in thymus, and peripheral Tregs (pTregs) (2, 3), tion of cancer patients is poorly understood (4). pTregs have an

F by guest on September 28, 2021 which are induced in periphery from naive T cells after encountering essential role in the adaptive immune system to promote tolerance in the gut (5) but have not been extensively described in context of tumorigenesis. FOXP3 can be transiently induced upon TCR *Cancer Research Center, Qatar Biomedical Research Institute, Hamad Bin Khalifa stimulation in the presence of TGF-b in CD4+ T cells in vitro to University, Qatar Foundation, Doha, Qatar; †Flow Cytometry Core, Qatar Biomedical Research Institute, Hamad Bin Khalifa University, Qatar Foundation, Doha, Qatar; generate induced Tregs (iTregs) (6). Apart from Foxp3, Tregs also and ‡Institute of Cancer Sciences, University of Manchester, Manchester M20 4BX, express programmed cell death protein 1 (PD-1), a coinhibitory United Kingdom receptor expressed on activated B cells and T cells (7, 8). In an- ORCIDs: 0000-0002-1546-1091 (N.M.A.); 0000-0002-3868-0318 (E.E.). imal studies, it has been reported that PD-1 is dispensable for Received for publication May 20, 2019. Accepted for publication October 27, 2019. tTreg development but critical for extrathymic differentiation of This work was supported by a start-up grant (VR04) to E.E. from the Qatar Biomed- pTregs in vivo (9). Moreover, blockade of PD-1/programmed ical Research Institute, Qatar Foundation. death–ligand 1 (PD-L1) pathway could suppress pTreg differen- V.S.N. performed experimental work, data analyses, and wrote the manuscript. S.M.T. tiation (9). We have previously reported that Tregs accumulate in assisted in experimental work, data analyses, and reviewing the manuscript. G.T. as- sisted in experimental work. G.P. performed cell sorting and data analyses. K.O. helped breast and colorectal TME and show higher PD-1 and CTLA4 in RNA-Seq and reviewing the manuscript. N.M.A. performed bioinformatics and data expression, which might have negative effects on disease prog- analyses and reviewed the manuscript. E.E. conceived the idea, designed the study, nosis (10, 11). Additionally, accumulation of PD-1hiFOXP3+ supervised the project, analyzed and interpreted data, and wrote and revised the man- uscript. All authors were involved in the final approval of the manuscript. Tregs that may enhance tumor progression has been reported in Address correspondence and reprint requests to Dr. Eyad Elkord, Cancer Research the TME of prostate cancer patients (12). These reports rationalize Center, Qatar Biomedical Research Institute, Hamad Bin Khalifa University, Qatar the importance of targeting PD-1 on Tregs to improve current Foundation, Post Office Box: 34110, Doha, Qatar. E-mail address: [email protected] immunotherapeutic protocols. or [email protected] Pembrolizumab is a humanized IgG4 k mAb, which selectively The online version of this article contains supplemental material. inhibits PD-1 in the TME (13). Pembrolizumab recently received Abbreviations used in this article: CTV, CellTrace Violet; FC, fold change; FVD, Food and Drug Administration approval for the treatment of pa- Fixable Viability Dye; gDNA, genomic DNA; IPA, Ingenuity Pathways Analysis; iTreg, induced Treg; MFI, mean fluorescence intensity; mTOR, mammalian target of tients with advanced solid tumors that show mismatch repair de- rapamycin; PD-1, programmed cell death protein 1; PD-L1, programmed death– ficiency and high microsatellite instability (14). Many clinical ligand 1; +0 pembrolizumab, with pembrolizumab from day 0; +2 pembrolizumab, with pembrolizumab from day 2; p-H2AX, phospho-histone 2A family member X; trials are undergoing to investigate the therapeutic benefits of pTreg, peripheral Treg; RNA-Seq, RNA-sequencing; TME, tumor microenviron- pembrolizumab in different cancers in combination therapies. In ment; Treg, T regulatory cell; TSDR, Treg-specific demethylation region; tTreg, two recent studies by our group, we have reported that pem- thymic-derived Treg. brolizumab does not affect Treg-related markers (15) and has no + Copyright Ó 2019 by The American Association of Immunologists, Inc. 0022-1767/19/$37.50 effect on the suppressive activity of FOXP3 Tregs in vitro (16);

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1900575 2 EFFECT OF PEMBROLIZUMAB ON HUMAN-INDUCED Tregs however, the effect of PD-1 blockade on pTregs or iTregs are RNA isolation still unclear. In this study, we investigated the effect of Total RNA was isolated from iTreg, cultured in the absence or presence pembrolizumab on iTregs differentiated from PBMC isolated of pembrolizumab for 7 d, using Monarch Total RNA Miniprep Kit (New from healthy donors. Our results showed that pembrolizumab England BioLabs, Ipswich, MA) following the manufacturer’s protocol. significantly interferes with the differentiation of iTregs by The concentrations of nucleic acids were measured using NanoDrop 2000c 2 shifting the axis of differentiation toward Th1- and Th17-related (Thermo Fisher Scientific) and then aliquoted and stored at 80˚C. condition through upregulation of mammalian target of rapamycin Real-time quantitative RT-PCR (mTOR), MAPK inhibitory genes, and p-STAT1. These data ad- vance our current knowledge on the importance of PD-1 in hu- cDNA was synthesized from total RNA using QuantiTect Reverse Tran- scription Kit (Qiagen, Hilden, Germany). The real-time PCR reactions were man iTreg polarization and also on the mechanism of action of performed using PowerUP SYBR Green Master Mix (Applied Biosystems, pembrolizumab on human iTreg differentiation. Further investi- CA) on QuantStudio 6/7 Flex Real-Time PCR System (Applied Biosystems). gations are required to explore the role of pembrolizumab in other PCR amplification was done for PD-1 and FOXP3 genes and normalized to b-actin. Quantification of relative gene expression was determined using Th lineages and also to elucidate their correlations with other 2 2 DDCT method. The amplification specificity was checked using melting immune checkpoints. curves and gel electrophoresis. Sequences of the primers are as follows: PD- 1forward,59-CCAGGATGGTTCTTAGACTCCC-39; PD-1 reverse, 59-TT- TAGCACGAAGCTCTCCGAT-39; FOXP3 forward, 59-GTGGCCCGGAT- Materials and Methods GTGAGAAG-39; FOXP3 reverse, 59-GGAGCCCTTGTCGGATGATG-39; Samples b-actin forward, 59-AGAGCTACGAGCTGCCTGAC-39;andb-actin reverse, 59-AGCACTGTGTTGGCGTACAG-39. Blood samples were collected from healthy donors and PBMC were Downloaded from isolated by density-gradient centrifugation using Histopaque-1077 (Sigma- In vitro suppression assay Aldrich, Irvine, U.K.). This study was performed under an ethical approval from Qatar Biomedical Research Institute, Doha, Qatar (Protocol no. CellTrace Violet (CTV; Thermo Fisher Scientific)–based suppression as- 2018-018). All experiments were performed in accordance with relevant says were performed using nontreated and pembrolizumab-treated iTregs guidelines and regulations. (cultured for 7 d) as suppressors. CD4+CD252 responder cells were sorted from the same donors from which iTregs were generated. For Flow cytometry and sorting sorting responder cells, PBMC were stained with 7AAD, anti–CD8-FITC, anti–CD4-PE, and anti–CD25-allophycocyanin. Sorted 7AAD2CD4+CD252 http://www.jimmunol.org/ Frozen PBMC were thawed and stained with anti–CD8-FITC (clone HIT8a; T cells were labeled with CTV and used as responders. CTV-labeled BD Biosciences, San Jose, CA), anti–CD4-PE (clone RPA-T4; BD responder cells (25,000 per well) were cocultured with different ra- Biosciences), anti-CD127–Alexa Fluor 647 (clone HIL-7R-M21; BD tios (1:0, 1:1/32, 1:1/16, 1:1/8, 1:1/4, 1:1/2, and 1:1) of nontreated or Biosciences), anti–CD45RA-PE/Dazzle (clone HI100; BioLegend, San pembrolizumab-treated iTregs, in the presence of Dynabeads Human Diego, CA), and anti–CD25- allophycocyanin-Cyanine7 (clone M-A251; T-Activator CD3/CD28 at a ratio of 1:1 (responder cells/beads) in 96-well BioLegend) after blocking with Fc receptor–blocking reagent (Miltenyi 2 2 flat bottom tissue culture plates. After 72 h culture, cells were harvested Biotec, Bergisch Gladbach, Germany). CD4+CD8 CD45RA+CD25 CD127+ and analyzed by flow cytometry. Proliferation and suppression were naive CD4+ T cells were sorted using the BD FACSARIA III Cell Sorter measured using FlowJo Software. The effector cells were identified by (BD Biosciences) after excluding dead cells by 7-aminoactinomycin D +/low selecting CTV population. Proliferation was measured by calculating (7-AAD; BioLegend). The purity for each cell type after sorting was mostly

the percentage of CTV loss after 3 d. Percentage of suppression was by guest on September 28, 2021 .95%. Other Abs used in this study are anti–TIM-3–BV711 (clone 7D3; calculated using the formula. BD Biosciences), anti–PD-1–PerCP (clone EH 12.1; BD Biosciences), and anti–GARP-FITC (clone G14D9; eBioscience, San Diego, CA). Intracellular Western blotting staining was performed using anti–FOXP3-PE-Cyanine7 (clone PCH101; BD Biosciences), anti–Helios-FITC (clone 22F6; BioLegend), anti–p-STAT1- The expression levels of STAT1, p-STAT1, and FOXP3 were measured using eFluor 660 (clone KIKSI0803; eBioscience), anti–p-STAT3 eFluor 450 Western blotting. Briefly, PBMC cultured in iTreg conditions in the absence (clone LUVNKLA; eBioscience), and anti–CD152 (CTLA4)-PerCP- or presence of pembrolizumab were lysed on ice using 1X RIPA Lysis eFluor710 (clone 14D3; eBioscience) using FOXP3/Transcription Factor Buffer (Thermo Fisher Scientific) containing protease inhibitor mixture Staining Buffer Set (eBioscience). Dead cells were excluded using Fixable (Sigma-Aldrich). Protein concentrations were measured using Pierce BCA Viability Dye (FVD) eFluor 780 (eBioscience). Protein Assay Kit (Thermo Fisher Scientific) as per the manufacturer’s For intracellular cytokine staining, cells cultured in iTreg condition instructions. The absorbance was measured using FLUOstar V Microplate were harvested every day (days 1–7), and incubated with 10 ng/ml Reader at 660 nm (BMG LABTECH, Ortenberg, Germany). Thirty phorbol 12-myristate-3-acetate (PMA, Sigma-Aldrich), 1 mg/ml ionomycin micrograms of proteins from each condition were resolved in NuPAGE (Sigma-Aldrich), and 1 mg/ml GolgiPlug (BD Biosciences) for 4 h in 4–12% Bis-Tris Gel (Thermo Fisher Scientific) and blotted on nitrocel- complete culture media. Cells were then collected and stained with anti- lulose membrane (Amersham Biosciences, Little Chalfont, U.K.). Mem- CD4–Alexa Fluor 700. Intracellular staining was performed using anti– brane blocking was done by 5% skimmed milk in TBS buffer containing IFN-g–PE (clone 4S.B3; BD Biosciences), anti–IL-4–allophycocyanin TBST. The membranes were then incubated overnight at 4˚C with primary (clone 8D4-8; eBioscience), anti–IL-17–Brilliant Violet 711 (clone Abs including b-actin (Cell Signaling Technology, Danvers, MA), poly- BL168; BioLegend) and anti–IL-10–FITC (clone JES3-19F1; BioLegend) clonal anti-STAT1 (Cell Signaling Technology), monoclonal anti–p-STAT1 using Transcription Factor Buffer Set (BD Pharmingen). Dead cells were (Clone 58D6; Cell Signaling Technology), and monoclonal anti-FOXP3 excluded by using FVD. All data were acquired using LSRFortessa X-20 (Clone 236A/E7; eBioscience). All primary Abs were prepared at 1:1000 (BD Biosciences) and were analyzed using FlowJo v.10.0 Software (Tree dilution in 5% skimmed milk or BSA in TBST. The membranes were then Star, Ashland, OR). incubated with HRP-conjugated donkey anti-rabbit or Goat anti-mouse IgG (Thermo Fisher Scientific) secondary Abs (dilution 1:10,000) at iTreg generation room temperature for 2 h. Detection was performed using SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scien- Sorted pure cells were cultured in RPMI 1640 (Life Technologies, MD), tific) and images were acquired using Molecular Imager ChemiDoc supplemented with 10% FBS (HyClone; GE Healthcare Life Sciences, Salt XRS+ with Image Lab Software (Bio-Rad, Hercules, CA). The den- Lake City, UT), 1% penicillin, streptomycin (HyClone), and 50 mM 2-ME + sitometric analyses were performed using ImageJ Software (National (Sigma-Aldrich). For iTreg generation, FACS-sorted naive CD4 T cells Institutes of Health, Bethesda, MD). were stimulated with Dynabeads Human T-Activator CD3/CD28 (Thermo Fisher Scientific, Waltham, MA) at a ratio of 1:1 (cells/beads), 100 U/ml RNA concentration and quality assessment recombinant human IL-2 (eBioscience), and 10 ng/ml rTGF-b (PeproTech, Rocky Hill, NJ) in the absence or presence of pembrolizumab (Keytruda; The quality and quantity of extracted RNA were measured on Agilent Merck, Kenilworth, NJ) at a final concentration of 2 mg/ml. Pembrolizumab 2100 Bioanalyzer (Agilent Technologies) using on-chip electrophoresis was replenished every 2 d. In some culture conditions, pembrolizumab was and Agilent RNA 6000 Nano Kit as per the manufacturer’s instruc- added from day 2 onwards. Cells were incubated at 37˚C and 5% CO2 and tions. Samples with RNA Integrity Number .7 were used for library analyzed in subsequent days. preparation. The Journal of Immunology 3

Library preparation temperature. Samples were then acquired on BD LSRFortessa X-20, and data were analyzed using FlowJo v.10.0 software. RNA was quantified using Qubit Instrument (Thermo Fisher Scientific) and RNA BR Assay Kit (Thermo Fisher Scientific). One hundred nanograms Cytokine analyses of RNA was used as an input for library preparation using TruSeq Stranded + mRNA Library Preparation Kit (Illumina, San Diego, CA) as per the Following7dofculturingnaiveCD4 T cells in iTreg condition manufacturer’s instructions. Briefly, mRNA was purified by using poly-T (with/without pembrolizumab), the supernatants were collected, spun 3 oligo beads and then was fragmented into small pieces under high tem- down, and frozen in 1 protease inhibitor mixture (Sigma-Aldrich) perature using divalent cations. The mRNA fragments were immedi- until subsequent analyses. The concentrations of cytokines in supernatants ately reverse-transcribed to first-stranded cDNA using random hexamers. were measured by using Bio-Plex 200 System (Bio-Rad Laboratories) Second strand was synthesized by incorporating dUTP instead of dTTP. using Bio-Plex Pro Human Cytokine 17-plex Kit (Bio-Rad) according The adaptors were ligated to the dscDNA followed by a single “A” to the manufacturer’s protocol. Each sample was assessed in duplicate. The nucleotide adenylation at 39 end of blunt fragments. The yield of DNA cytokine panel composed of IL-1b, -2, -4, -5, -6, -7, -8, -10, -12, -13, -17A, libraries was quantified using Qubit dsDNA HS Assay Kit (Thermo Fisher G-CSF, GM-CSF, IFN-g, monocyte chemotactic protein-1, MIP-1b, Scientific) and size distribution of the DNA libraries was determined using and TNF. the Agilent 2100 Bioanalyzer DNA1000 Chip (Agilent Technologies). Statistical analyses The clusters were generated by cBot (Illumina) and sequencing was done on HiSeq 4000 System with 300 bp paired-end using HiSeq 3000/4000 All statistical analyses were performed using GraphPad Prism 8.0 Software SBS Kit (Illumina). (GraphPad Software, San Diego, CA). The Shapiro–Wilk normality test followed by paired t test/Wilcoxon matched-pairs signed rank tests were RNA-sequencing data analysis used to examine the differences within groups. The p values are repre- , , , . Paired-end reads were aligned to the hg19 human reference genome in CLC sented as follows: ***p 0.001, **p 0.01, *p 0.05. A p value 0.05 is considered statistically nonsignificant. The data are represented as Downloaded from Genomics Workbench 12 (Qiagen). The abundance of transcript expression 6 was measured as the score of transcripts per million mapped reads in CLC mean SEM. Genomics Workbench 12. Abundance data were subsequently subjected to differential gene expression and clustering analyses using 2.0-fold change Results , (FC) and 0.05 p value cutoff. Pembrolizumab masks PD-1 receptor without altering the Gene set enrichment analyses and modeling of transcriptomic level and downregulates FOXP3 in iTregs gene interactions The mechanism of action of pembrolizumab is known to block the http://www.jimmunol.org/ Differentially expressed genes were imported into the Ingenuity Path- interactions of PD-1 with its ligands but the detailed mechanism is ways Analysis (IPA) Software (Ingenuity Systems; www.ingenuity.com) still not clear (22). Moreover, it has been reported that PD-1 is to obtain functional regulatory networks and canonical pathways using indispensable for Tregs to maintain immune homeostasis (23), upstream regulator analysis, downstream effects analysis, mechanistic whereas little studies have shown its role in human iTreg. We have networks, and causal network analysis prediction algorithms. IPA uses precise database to paradigm functional regulatory networks from a list recently reported that pembrolizumab does not affect the stability of individual genes and determines a statistical score for each network or function of tTreg, but rather blocks the interaction of PD-1 with according to the fit of the network to the set of focus genes. The score is its ligands (16). In this study, we cultured sorted naive CD4+ the negative log of P and denotes the likelihood of the focus genes in T cells under iTreg-polarizing conditions for up to 14 d with or the network being found together by chance. The biological functions by guest on September 28, 2021 without pembrolizumab (Fig. 1A). We found that pembrolizumab assigned to each network are ranked according to the significance of + that biological function to the network (17, 18). masks PD-1 receptor on CD4 T cells and significantly down- regulates the level of FOXP3+ Tregs compared with untreated FOXP3 demethylation analyses iTreg, as determined by flow cytometry (Fig. 1B, 1C). Addition- Cells cultured under iTreg-polarizing condition for 7 d in the absence or ally, we checked the mean fluorescence intensity (MFI) of FOXP3 presence of pembrolizumab were stained with anti-CD4-PE and intracel- in the presence and absence of pembrolizumab to confirm the + hi lular anti-FOXP3-PE-Cy7 Abs. CD4 FOXP3 cells were then sorted from downregulation and found that MFI was also reduced in FVD eFluor 780live cell pool using the BD FACSARIA III Cell Sorter (BD Biosciences). The genomic DNA (gDNA) was extracted from pembrolizumab-treated samples (Fig. 1B). Next, we checked the fixed/sorted cells using FFPE RNA/DNA Purification Plus Kit (Norgen mRNA expression of PD-1 after pembrolizumab treatment and Biotek, Thorold, ON, Canada) according to manufacturer’s instruc- found that there is no change in the transcriptomic level of PD-1 tions. Extracted gDNA was converted by EZ DNA Methylation-Gold following pembrolizumab treatment (Fig. 1D). These data show Kit (Zymo Research, Irvine, CA). The bisulfite-treated DNA was then that pembrolizumab masks PD-1 receptor without altering its subjected to PCR for the amplification of Treg-specific demethylation region (TSDR) and FOXP3 promoter using Hot Start TaKaRa Taq DNA transcriptomic level. We also checked the overall viability of Polymerase (TaKaRa Bio, Shiga, Japan). The PCR products were then nontreated and pembrolizumab-treated iTregs and found that they cloned into the pGEMT-Easy Vector (Promega, Madison, WI) using were similar (data not shown). Down regulation of FOXP3 ex- DNA Ligation Kit, Mighty Mix (TaKaRa Bio). The colonies were se- pression in iTregs by pembrolizumab was further confirmed by quenced with M13 reverse primer using 3130X Genetic Analyzer (Applied Biosystems), as previously described (19). Sequencing data Western blotting (Fig. 1E) and qRT-PCR (Fig. 1F). Moreover, we were analyzed using Bisulfite Sequencing DNA Methylation Analysis investigated the kinetics of FOXP3 expression from day 2 to day Software (Jacobs University, Bremen, Germany). Sequences of the 7 and found that FOXP3 started to downregulate from day 5 of primers are as follows: TSDR forward, 59-TGTTTGGGGGTAGAGG- postculture in pembrolizumab-treated samples (data not shown). AT-39 (20); TSDR reverse, 59-TATCACCCCACCTAAACCAA-39 (20); FOXP3 promoter forward, 59-TGGTGAAGTGGATTGATAGAAAAGG- Pembrolizumab treatment relatively reduces the suppressive 39 (21); and FOXP3 promoter reverse, 59-TATAAAAACCCCCCCCC- activity of iTregs ACC-39 (21). It has been reported that the suppressive capacity of iTregs are DNA damage detection assay comparable to thymic-generated Tregs (24). In this study, we Flow cytometric detection of DNA damage was performed by staining cells checked whether pembrolizumab treatment can affect the sup- with anti–phospho-histone 2A family member X (p-H2AX) and DAPI. pressive function of iTregs. CTV-labeled CD4+CD252 responder iTregs cultured with and without pembrolizumab were stained with anti– cells were cocultured with nontreated or pembrolizumab-treated p-H2AX–Alexa Fluor 647 Ab (clone N1-431; BD Biosciences) for 20 min at 4˚C. The cells were then fixed and permeabilized using intracellular iTregs in the presence of Dynabeads Human T-Activator CD3/CD28. fixation and permeabilization concentrate (eBioscience) and incubated After 3 d of culture, the extents of proliferation of responders were with 1 mg/ml DAPI (Thermo Fisher Scientific) for 15 min at room measured by flow cytometry. We found that both nontreated and 4 EFFECT OF PEMBROLIZUMAB ON HUMAN-INDUCED Tregs Downloaded from http://www.jimmunol.org/

FIGURE 1. Effect of pembrolizumab on PD-1 and FOXP3 expression in iTregs. Naive CD4+ T cells were cultured under iTreg-polarizing conditions in the absence or presence of pembrolizumab. Schematic diagram shows iTreg generation (A). Representative flow cytometric plots show PD-1 and FOXP3 expression in nontreated and pembrolizumab-treated iTregs cultured for 7 and 14 d (positive populations were identified according to fluorescence minus one [FMO] controls) (B). Box plots show the percentage of CD4+FOXP3+ T cells in nontreated and pembrolizumab-treated iTregs cultured for 7 and 14 d (C). Bar plot shows the relative PD-1 expression normalized to b-actin after 7 d of iTreg culture in the absence or presence of pembrolizumab (D). Western blots show expression of FOXP3 in the protein extracts prepared from iTreg cultured with or without pembrolizumab for 7 d and scatter plot shows FOXP3 expression normalized to b-actin from Western blot data of four independent experiments (E). Scatter plot shows mRNA expression of FOXP3 normalized by guest on September 28, 2021 to b-actin after 7 d of iTreg cultured in the absence or presence of pembrolizumab (F). Data represent the mean 6 SEM of four independent experiments. ***p , 0.001, **p , 0.01, *p , 0.05. pembrolizumab-treated iTregs were suppressive (Fig. 2A, 2B); compared with the other two conditions (Fig. 3A, 3B). After 7 d, however, the treated iTregs showed relatively less suppression, FOXP3 was downregulated in all conditions, but the expression was especially at higher Treg ratios, compared with nontreated iTregs significantly lower in +0 pembrolizumab condition compared with (Fig. 2C). Next, we checked the level of intracellular cytokines the other two conditions in continued analyses on days 12 and in iTregs and found that the immunosuppressive cytokine IL-10 14 (Fig. 3B). was highly expressed in nontreated iTregs compared with To exclude the possibility of FOXP3 instability in the presence pembrolizumab-treated counterparts, which did not secrete any of pembrolizumab, we investigated the methylation profile of IL-10 (Fig. 2D). Notably, both iTregs did not produce Th1 cyto- TSDR and proximal promoter region from the gDNA extracted kine (IFN-g), Th2 cytokine (IL-4), and Th17 cytokine (IL-17) from FOXP3+/2 cells cultured in the absence or presence of (Fig. 2D) because they are committed iTregs. These data indicate pembrolizumab in iTreg-polarizing condition. We found that there that PD-1 blockade may partially interfere with the suppressive were no significant differences in demethylation percentages of function of iTregs; this could be because they express less FOXP3 both FOXP3-TSDR and promoter region in the absence or presence and IL-10. of pembrolizumab either in FOXP3+ or in FOXP32 populations (Fig. 3C), confirming that pembrolizumab has no effect on FOXP3 Pembrolizumab interferes with the differentiation of iTregs but stability. Reports showed that mTOR inhibition (25) and MAPK not the stability of FOXP3 activation (26) favor iTreg generation. Our transcriptomic profil- Our results show that FOXP3 was downregulated upon pembrolizumab ing from nontreated and pembrolizumab-treated iTreg data show treatment, but it was not clear whether pembrolizumab affects the stability that mTOR activation genes, including WDR24, LAMTOR5, of FOXP3 or differentiation of iTreg. To check this, we cultured iTregs in TBC1D7, CAB39L, RHEB, LAMTOR5-AS1, EIF4EBP1, AKT1, three different conditions: without pembrolizumab, with pembrolizumab NFKB1, AKT3, LAMTOR4, MLST8, and LAMTOR1, were up- from day 0 (+0 pembrolizumab), and with pembrolizumab from day 2 regulated in pembrolizumab-treated iTregs. However, this upreg- (+2 pembrolizumab) for 7 d. Interestingly, we found that after 7 d of ulation was significantly reversed in iTregs cultured in the culture, the expression of FOXP3 in CD4+ T cells was similar in presence of rapamycin (mTOR inhibitor), Fig. 3D). To confirm the without pembrolizumab (65%) and +2 pembrolizumab (60%) role of mTOR, we cultured iTregs in the presence or absence of (Fig. 3A, 3B). These data show that downregulation of FOXP3 was the rapamycin in combination with pembrolizumab. Interestingly, due to the interference with iTreg differentiation. Additionally, the we found that rapamycin diminished the effect of pembrolizumab- expression of FOXP3 in +0 pembrolizumab was much lower (32%) mediated downregulation of FOXP3 (without pembrolizumab 47% The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 2. Functional analysis of nontreated and pembrolizumab-treated iTregs. CTV-labeled CD4+CD252 responder cells were cultured with non- treated and pembrolizumab-treated iTregs (cultured under iTreg-polarizing conditions for 7 d) in the presence of CD3/CD28 dynabeads at various Teff/Treg ratios, ranging from 1:0 to 1:1/32. Representative histogram plots show CTV loss in responders at different Teff/Treg ratios in nontreated (upper panel) and treated (lower panel) iTregs (A). Line graphs show the percentage of suppression at different Teff/Treg ratios in three donors (B). Bar plots show the overall percentage of suppression by both nontreated and pembrolizumab-treated iTregs at 1:1 and 1:1/2 Teff/Treg ratios (C). Representative flow cytometric plots show the percentage of IL-10+, IFN-g+, IL-4+, and IL-17+ CD4+ T cells at day 7 in the absence or presence of pembrolizumab. Data are from three independent experiments (D). and pembrolizumab + rapamycin 45%, Fig. 3E, 3F). Additionally, FOXP3 is directly associated with STAT1 activation, we cultured genes involved in MAPK inhibitory pathways, including DUSP14, iTregs with or without pembrolizumab in the presence or absence ZNF383, RGS14, DUSP12, IL1B, AIDA, and TSC2, were up- of STAT1 activation inhibitor (fludarabine) for 7 d and checked regulated in pembrolizumab-treated iTregs (Fig. 3G). These data the level of FOXP3. As expected, the pembrolizumab treatment rationalize that pembrolizumab interferes with the differentiation downregulated the expression of FOXP3 from 78% (nontreated) to of iTreg by activating mTOR and inhibiting MAPK pathways 56% (treated). Notably, the blockade of STAT1 activation could but not the stability of FOXP3. not restore pembrolizumab-mediated downregulation of FOXP3 (51%, Fig. 4C). These data indicate that activation of STAT1 could p-STAT1 was upregulated upon pembrolizumab treatment be a dispensable factor for pembrolizumab-mediated downregu- in iTregs lation of FOXP3. Additionally, we also cultured iTregs in the To further investigate the role of pembrolizumab on iTregs, we presence and absence of IFN-g neutralization Ab to check whether investigated the expression of p-STAT1 in pembrolizumab-treated pembrolizumab-mediated upregulation p-STAT1 is directly due to and nontreated iTregs. It has been reported that iTreg differentiation the autocrine function of IFN-g. However, we could not see any could be inhibited by cytokines through STAT1 dependent T cell significant changes in p-STAT1 expression in both iTregs (data not autocrine pathway (27, 28). In this study, we found that p-STAT1 shown). was significantly upregulated in pembrolizumab-treated iTregs as Next, we compared the expression level of p-STAT1 in FOXP3+ and confirmed by flow cytometry (Fig. 4A) and Western blotting FOXP32 subpopulations in the absence or presence of pembrolizumab. (Fig. 4B). However, there was no upregulation in STAT1 (Fig. 4B), Interestingly, we found that p-STAT1 was significantly higher in confirming that the upregulation of p-STAT1 is through the acti- FOXP3+ subpopulation, compared with FOXP32 in both treated and vation of STAT1 without altering the expression of STAT1. To nontreated conditions (Fig. 4D). Moreover, within FOXP3+ subpop- check whether the pembrolizumab-mediated downregulation of ulation, p-STAT1 was significantly higher in pembrolizumab-treated 6 EFFECT OF PEMBROLIZUMAB ON HUMAN-INDUCED Tregs Downloaded from http://www.jimmunol.org/

+

FIGURE 3. Kinetics of FOXP3 expression and methylation profile of TSDR and FOXP3 promoter of iTregs. Naive CD4 T cells were cultured under by guest on September 28, 2021 three different iTreg-polarizing conditions (2 pembro, +0 pembro, and +2 pembro), with “pembro” meaning “pembrolizumab.” Representative flow cytometric plots show frequency of FOXP3+ Tregs at day 7 in the three conditions (A). Line graph shows kinetics of FOXP3 expression at intervals from day 3 to day 14 postculture in the absence or presence of pembrolizumab (B). Methylation profile of 15 CpG sites of TSDR and eight CpG sites of FOXP3 proximal promoter, from gDNA isolated from FOXP3+/2 cells cultured under iTreg conditions in the absence or presence of pembrolizumab for 7 d, was investigated. Representative flow cytometric plots show the percentage of FOXP3+ and FOXP32 Tregs cultured under iTreg conditions 6 pembrolizumab, and bar plots show the demethylation percentage of TSDR and promoter region (C). FOXP3+ and FOXP32 subpopulations from nontreated and pem- brolizumab-treated iTregs were sorted and gDNA was isolated as shown in the Fig. 3C. Data are representative of two individual experiments. Heat map shows the FCs (.2) relative to the mean expression of mTOR activation genes in nontreated and pembrolizumab-treated iTregs cultured in the absence or presence of rapamycin for each individual sample (D). Representative flow cytometric plots show the percentage of FOXP3+ Tregs at day 7 in four different conditions (E). Scatter plots show the percentage of FOXP3+ cells gated on CD4+ populations in different conditions, cultured for 7 d, from four inde- pendent experiments (F). Data are from two individual experiments. Heat map shows the FCs (.2) relative to the mean expression of MAPK inhibition- related genes in nontreated and pembrolizumab-treated iTregs of each individual sample (G). **p , 0.01. condition compared with nontreated condition (Fig. 4D). Addi- (Fig. 5B). After 5 d of culture, the level of IFN-g+ T cells dropped tionally, we checked the level of p-STAT3 and found that it was in pembrolizumab-treated condition (Fig. 5B). Notably, there was not expressed either in nontreated or pembrolizumab-treated no substantial IFN-g increase in nontreated iTregs at any time iTregs (Fig. 4E). These data show that the role of p-STAT1 point (Fig. 5A, 5B). Additionally, we found that there was a sig- in iTreg polarization and the downregulation of FOXP3 upon nificant upregulation of IL-10 at later time points (days 6 and 7) in pembrolizumab treatment could be indirectly dependent on the nontreated iTregs (Fig. 5C, 5D). In contrast, pembrolizumab- upregulation of p-STAT1. treated iTregs did not express IL-10 at any time point (Fig. 5C, 5D). Furthermore, IL-17 or IL-4 expressions were not detectable Th1 and Th17 cytokines were upregulated upon in both pembrolizumab-treated and nontreated iTregs at any time pembrolizumab treatment point (data not shown). Additionally, we investigated the levels of To investigate the role of pembrolizumab in iTreg polarization, we inflammatory cytokines accumulated in the supernatants of iTreg- performed intracellular cytokine kinetic analyses from day 1 to day polarizing conditions. It has been reported that PD-1 blockade 7 on iTregs cultured in the absence or presence of pembrolizumab. upregulates the transcriptomic expression of IFN-g–related genes Interestingly, we found that pembrolizumab treatment increases (29). In this study, we found that pembrolizumab treatment shifts IFN-g production at early time points (days 1–4) compared with iTreg polarization toward inflammatory Th1/Th17 responses, nontreated condition (Fig. 5A, 5B). Although the overall IFN-g which was demonstrated by the increase in IFN-g and IL-17 cy- expression was low at all time points, the percentage of IFN-g+ tokines (Fig. 5E). Moreover, there was no difference in Th2 re- T cells was highest on day 3 in pembrolizumab-treated iTregs sponse, which was evidenced by no significant differences in the The Journal of Immunology 7 Downloaded from http://www.jimmunol.org/

FIGURE 4. Effect of pembrolizumab on p-STAT1 expression in iTregs. Naive CD4+ T cells were cultured under iTreg-polarizing conditions. Repre- sentative flow cytometric plots show expression of p-STAT1 (gated on live CD4+FOXP3+) in nontreated and pembrolizumab-treated iTregs, cultured for 7 and 14 d (A). Western blots show expression of p-STAT1 in protein extracts prepared from iTregs6 pembrolizumab for 7 d. Scatter plots show expression of p-STAT1 and STAT1 from Western blot data of four independent experiments (B). Representative flow cytometric plots show FOXP3 expression (gated on live CD4+ T cells) in nontreated and pembrolizumab-treated iTregs cultured in the absence or presence of fludarabine (10 mM) from three independent experiments (C). Representative flow cytometric and scatter plots show the percentage of p-STAT1+ cells in FOXP3+ and FOXP32 T cells at day

7 in nontreated and pembrolizumab-treated iTregs (D). Data represent the mean 6 SEM of four independent experiments. Representative flow cytometric by guest on September 28, 2021 plots from four independent experiments show the expression of p-STAT3 (gated on live CD4+ T cells) at day 7 in nontreated and pembrolizumab-treated iTregs (E). **p , 0.01, *p , 0.05. secretions of IL-4 and IL-5 in pembrolizumab-treated samples RNA-sequencing expression profiling of iTregs cultured in the (Fig. 5E). Additionally, there were no significant differences in absence or presence of pembrolizumab IL-10 production in the accumulative supernatants during the To characterize the transcriptomic alterations of iTregs in the course of iTreg differentiation between pembrolizumab-treated absence or presence of pembrolizumab, we subjected the iTregs and nontreated iTregs (Fig. 5E). This could be explained by the from both conditions (2/+ pembrolizumab) to whole transcriptomic intracellular cytokine kinetic data, which showed that IL-10 was sequencing from two independent experiments. The hierarchical only secreted at later time points (days 6 and 7). Taken together, clustering of differentially expressed transcripts showed a distinct + these data show that naive CD4 T cells cultured in the presence cluster of nontreated and pembrolizumab-treated iTregs (Fig. 6A). of pembrolizumab secrete more Th1- and Th17-related cytokines A total of 2963 upregulated and 4683 downregulated transcripts during iTreg differentiation. were identified in pembrolizumab-treated, compared with non- . TIM-3 is downregulated upon pembrolizumab treatment treated, iTregs using 2.0-FC cutoff. Canonical pathway analyses on both upregulated and downregulated genes revealed that genes in iTregs involved in DNA damage/repair (homologous recombination of Reports showed that both PD-1 (30) and TIM-3 (31) expressions cells, DNA recombination, and repair of genes) and immune cell are induced upon TCR activation transiently and decreased in their trafficking pathways (cell movement of mononuclear leukocytes/ absence. Next, we intended to investigate whether pembrolizumab leukocytes/lymphocytes, migration of lymphocytes/mononuclear has any role in the expression of TIM-3 in iTreg-polarized con- leukocytes, and transmigration of leukocytes/lymphocytes/ dition. TIM-3 expression was upregulated upon TCR stimulation mononuclear leukocytes) were upregulated, whereas genes in- in nontreated and pembrolizumab-treated conditions (Supplemental volved in cellular assembly and organization (formation of Fig. 1A). Interestingly, we found that both expression and MFI of cytoskeleton/vesicles/autophagosomes/actin filaments/filaments/ TIM-3 (Supplemental Fig. 1A) and PD-1 (Supplemental Fig. 1B) actin stress fibers, development of cytoplasm, cell-cell adhe- were downregulated upon pembrolizumab treatment at all time sion, and organization of cytoplasm) were downregulated in points tested. These data show that, in addition to PD-1, pembrolizumab-treated iTregs (Fig. 6B, 6C). Selected upregulated pembrolizumab also downregulated TIM-3 in iTreg-polarized transcripts including AUNIP, WFS1, RAD51, G0S2, LAMTOR5, conditions. Further investigations are warranted to reveal the CKLF, DUSP14, and S100p, and downregulated gene BRK1 are role of pembrolizumab on TIM-3 expression and also on the role shown in Fig. 6D. Genes involved in DNA repair mechanisms of TIM-3 in iTreg differentiation. (RAD51, AUNIP, HIST1H3B, HIST1H2AJ, C4orf27, C9orf142, 8 EFFECT OF PEMBROLIZUMAB ON HUMAN-INDUCED Tregs Downloaded from http://www.jimmunol.org/

FIGURE 5. + Detailed kinetic analysis and multiplex data of cytokine secretion. Naive CD4 T cells were cultured under iTreg-polarizing conditions. For by guest on September 28, 2021 intracellular cytokine staining, cells were collected at different days (1–7) and stained with anti–IFN-g and IL-10. Representative flow cytometric plots show the percentage of IFN-g+ (A) and IL-10+ (C) CD4+ T cells from day 1 to day 7 postculture in the absence or presence of pembrolizumab. Line graphs show the overall percentages of IFN-g+ (B) and IL-10+ (D) CD4+ T cells. Data are from three independent experiments with five replicates. Supernatants were collected at day 7 and cytokine levels were measured by Bio-Plex System. The scatter plots show mean cytokine concentration in picograms per milliliter of IFN-g, IL-17, IL-4, IL-5, and IL-10 from supernatants cultured in the presence or absence of pembrolizumab (E). Data are from nine individual experiments. **p , 0.01, *p , 0.05.

ZSWIM7, and GMNN) and cell cycle regulation (CKS1B_2, APOM, MRGBP, and UBD were upregulated (.2-FC, Fig. 7D) HMG20B, DSN1, CDC20, MAPK3, SLC8B1, REEP4, and and antiapoptotic genes including HSPA1B and CARD6 were FTSJ2) were upregulated (.2-FC, Fig. 7A). To confirm the up- downregulated (.2-FC, Fig. 7D) in pembrolizumab-treated regulation of DNA damage-related genes upon pembrolizumab iTregs. Meanwhile, there was no detectable difference in canoni- treatment, we did flow cytometric functional analysis using cal antiapoptotic genes from Bcl-2 family (data not shown). No- anti–p-H2AX and DAPI staining. Interestingly we found tably, the proliferation-related genes including S100P, ANAPC2, that p-H2AX+DAPI+ level was significantly upregulated in IL1A, IL1B, PLK1, EMP1, and CD86 were upregulated in pembrolizumab-treated iTregs compared with nontreated iTregs pembrolizumab-treated iTregs (.2-FC; Fig. 7D). This is in line (Fig. 7B). Additionally, our network analyses indicate that DNA with previously published data where pembrolizumab augmented damage and repair pathways were upregulated in iTregs upon CD3+ T cell proliferation (32). Next, we checked Th1-, Th2-, pembrolizumab treatment (Supplemental Fig. 2). These data show Th17-, and Treg-related genes and found that there was no change that pembrolizumab plays a role in DNA damage in iTreg-polarizing in the transcriptional levels of Treg-, Th2-, and Th17-related conditions. Moreover, immune cell trafficking genes (CKLF, FCAR, genes between pembrolizumab-treated and nontreated iTregs AIF1, SIRPA, and ICAM1) were upregulated (.2-FC,Fig.7C)in (Fig. 7E, 7F). However, Th1-related genes including IFNGR2, pembrolizumab-treated iTregs compared with nontreated iTregs. ETV5, TBX21, TICAM2, and CKLF were upregulated in Altogether, our data show that pembrolizumab treatment leads to pembrolizumab-treated iTregs (.2-FC;Fig.7E).Next,we the upregulation of genes related to immune trafficking and DNA checked the expression of immune checkpoints and ligands (ICs repair in iTregs. and ICLs), including PD-1, LAG-3, CTLA4, PD-L1, VISTA, TIGIT, and galactin-9. We did not find any significant differences Upregulation of proapoptotic and downregulation of in FC between pembrolizumab-treated and nontreated iTregs antiapoptotic genes in pembrolizumab-treated iTregs (Fig. 7G). Additionally, by using flow cytometry we checked the Next, we checked the differential expression of genes involved expression of Treg signature gene, CTLA4, in nontreated and in proliferation and apoptosis. Interestingly, we found that the pembrolizumab-treated iTregs. In agreement with the RNA- proapoptotic genes including WFS1, G0S2, SMURF2P1, TFPT, sequencing (RNA-Seq) data (Fig. 7G), flow cytometric data The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/

FIGURE 6. Differentially expressed genes of iTregs. Hierarchical clustering of two independent experiments on differentially expressed RNA transcripts from RNA-Seq data. Each column represents a sample, and each row represents a transcript. Expression level of each gene in a single sample is depicted by guest on September 28, 2021 according to color scale (A). Top significantly affected (1.5 , Z-score ,22.0) canonical pathways based on IPA analysis. The horizontal bars denote the different pathways based on the Z-scores (B). Tree map (hierarchical heat map) depicting affected functional categories based on upregulated gene transcripts where the major boxes represent a category of DNA replication, recombination, and repair, cellular assembly, and immune cell trafficking(C). Each individual colored rectangle is a particular biological function and the color range indicates its predicted activation state: increasing (orange) or decreasing (blue). Darker colors indicate absolute Z-scores. In this default view, the size of the rectangle is correlated with increased overlap significance. Gene expression analysis by RNA-seq in iTregs are shown by scatter plots. The x- and y-axes showing transcripts per million of 2/+ pembrolizumab-treated iTregs (D). RNA-Seq data were obtained from two independent experiments. confirmed no difference in CTLA4 expression between nontreated of PD-1 could augment proliferation and suppressive function of and pembrolizumab-treated iTregs (Fig. 7H). human natural Tregs (37). Another study on nonresponders from B16 melanoma-bearing mouse model showed that PD-1 blockade Discussion can result in the accumulation of CD4+FOXP32PD-1hi cells, In this study, we demonstrated the importance of PD-1 in the which inhibited antitumor responses (38). In breast and colorectal generation of human iTregs and also the role of pembrolizumab in TME, it was reported that the expression of PD-1 in Tregs was iTreg differentiation. PD-1 has been shown to be transiently up- significantly higher in tumor tissue compared with normal tissue regulated on activated T cells and constitutively expressed on cells (10, 11). A study on tumor tissues showed that pembrolizumab did exhibiting exhausted phenotype (33, 34). A recent study of single- not alter the expression level or phenotype of tTregs (39). In cell analysis on immune cells from melanoma patients treated mouse models, it was reported that PD-1/PD-L1 blockade can with anti–PD-1 showed that the expression of TCF7 on CD8+ augment the response of conventional T cells by suppressing the T cells can predict the clinical response of immune checkpoint differentiation of pTregs (9). In contrast, another group reported blockade (35). In contrast, the expression of exhausted and that PD-1 deficiency neither inhibit the differentiation of pTregs dysfunctional signatures on T cells was associated with lack of nor lead to Th17-mediated autoimmune response, but rather in- response to checkpoint blockade therapy (35). Another study crease the pTreg generation (40). These contradictory reports on melanoma patients revealed that anti–PD-1–induced cellular show that the role of PD-1 in mouse iTreg generation is still immune responses were mediated through an increase of tumor- unclear. Moreover, little is known about the role of PD-1 in infiltrating CD8/Treg ratio by depleting intratumoral Tregs (36). In human iTreg generation and currently there are no studies, to our addition, a recent study on advanced gastric cancer nonresponder knowledge, showing the role of pembrolizumab in iTreg differen- patients showed that the PD-1 blockade could significantly in- tiation. We found that pembrolizumab significantly downregulates crease effector PD-1+ Tregs within the TME, which adversely FOXP3 expression in iTreg-polarizing conditions compared with affected antitumor immunity and resulted in rapid cancer pro- nontreated iTregs (Fig. 1B, 1C, 1E, 1F). These data indicate that gression (37). Moreover, it has been shown that in vitro blockade pembrolizumab masks the PD-1 receptor, which is essential for the 10 EFFECT OF PEMBROLIZUMAB ON HUMAN-INDUCED Tregs Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 7. Transcriptomic profile of iTregs as categorized by their functional characteristics. Heat maps show the FCs (.2) relative to the mean expression of housekeeping controls/DNA repair/cell cycle regulation–related genes (A). Representative flow cytometric and scatter plots show the per- centage of H2AX+DAPI+ cells gated on CD4+ T cells at day 7 in nontreated and pembrolizumab-treated iTregs (B). Data represent the mean 6 SEM of four independent experiments. Heat maps show the FCs (.2) relative to the mean expression of cell trafficking (C)–, proliferation/proapoptosis/antiapoptosis (D)–, Treg/Th1(E)–, Th2/Th17 (F)–, and immune checkpoint/ligand (G)–related genes. Representative flow cytometric plots from four independent ex- periments show the percentage of CTLA4+FOXP3+ cells gated on CD4+ T cells at day 7 (H). *p , 0.05. upregulation of FOXP3 in human iTreg generation. Receptor/ligand cytokines in the absence or presence of pembrolizumab (Fig. 2D). interactions are usually important mechanisms for adequate signal These data showed that PD-1 blockade can partially affect the transduction. In this study, we suggest that pembrolizumab can alter suppressive ability of iTregs, possibly by interfering with IL-10 the transcription profile of iTregs independently of PD-1/PD-L1 secretion. The downregulation of FOXP3 could be explained in two interactions. Our findings could imply that the changes in iTreg different ways: 1) pembrolizumab interferes with the differentia- transcriptome in response to pembrolizumab treatment are not due tion of iTregs or 2) pembrolizumab affects with the stability of to a change in PD-1 expression, but possibly as a result of an altered FOXP3. To elucidate the role of pembrolizumab, we cultured protein structure (41). Similarly, we have recently reported that iTregs in three different conditions as shown in Fig. 3A. We chose atezolizumab (anti–PD-L1) alters the structure of PD-L1 on MDA- day 2 because we found that FOXP3 expression was similar with or MB-231 breast cancer cells by inducing a transition from random without pembrolizumab treatment up to day 3 (Fig. 3B). In this coil and a-helical structure to b-sheet conformation (42). Moreover, study, we found that expression of FOXP3 in 2pembrolizumab it has been reported that the action of pembrolizumab is profoundly and +2 pembrolizumab conditions are similar at all time points dependent on the binding of flexible C9D loop of PD-1, rather than (day 3–day 14, Fig. 3A, 3B). Moreover, FOXP3 was downregulated blocking PD-1/PD-L1 interactions (43). These data suggest that in +0 pembrolizumab condition, compared with 2pembrolizumab pembrolizumab could alter the structure of PD-1, which might have and +2 pembrolizumab from day 7 (Fig. 3A, 3B). These data an impact on the downstream functional pathways. confirm our findings that pembrolizumab interferes with the dif- To check the suppressive function of nontreated and ferentiation of iTreg and not the stability of FOXP3. pembrolizumab-treated iTregs, we performed in vitro sup- Reports showed that the inhibition of mTOR pathways can pression assays and found that both iTregs were suppressive, but induce the differentiation of iTregs and suppressive activity through pembrolizumab-treated iTregs were relatively less suppressive and CD39 (ENTPD1) pathway (25, 44, 45). Our transcriptomic data failed to secrete IL-10 (Fig. 2). Notably, when iTregs became show that multiple genes involved in the activation of mTOR committed at day 7, they did not produce Th1, Th2, and Th17 pathway were upregulated including ENTPD1 in pembrolizumab- The Journal of Immunology 11 treated iTregs (Fig. 3D). To confirm these data, we cultured iTregs data suggest that iTregs become committed by day 6, when they with pembrolizumab in the presence of rapamycin and found that start to produce IL-10. However, our in vitro suppression data the mTOR-related genes were downregulated (Fig. 3D). Inter- showed that pembrolizumab-treated iTregs also exhibit sup- estingly, rapamycin in the presence of pembrolizumab restored pressive activity (Fig. 2A–C). This could be mediated through the level of FOXP3 in iTregs to a similar level of that in non- other mechanisms including IL-35, TGF-b, or other cell-cell treated samples (Fig. 3E, 3F). Moreover, the MAPK inhibitory contact-dependent mechanisms. Recent study on IL-35– and genes, including DUSP12 and DUSP14, were upregulated in IL-10–producing Tregs revealed that IL-35–producing Tregs pembrolizumab-treated iTregs (Fig. 3G). The MAPK pathways are a distinct suppressor population from IL-10–producing are shown to be essential for the differentiation of iTregs (26). subsets and these cytokines are reciprocally expressed by these Altogether, our data show that pembrolizumab upregulates genes Treg populations (55). Moreover, IL-10– and IL-35–producing responsible for the mTOR activation and MAPK inhibition, and Tregs have different transcription factor dependencies and have thereby interferes with iTreg differentiation. different geographic locations (55). However, it would be in- PD-1 transcriptomic data exhibited that there is no difference in teresting to investigate the actual suppressive mechanism of the expression of PD-1 in nontreated and pembrolizumab-treated pembrolizumab-treated iTregs. iTregs. Moreover, both iTregs show similar level of PD-1 ex- Our transcriptomic data confirmed that Th1-related genes, no- pression (Fig. 1D). Notably, the flow cytometry data show that tably TBX21, were upregulated in pembrolizumab-treated iTregs pembrolizumab completely block the expression of PD-1 (Fig. 7E). Previous reports showed that T-bet was upregulated (Fig. 1B), confirming that pembrolizumab might mask the PD-1 specifically in Th1 cells, which binds to and activates IFN-g (56). epitope without altering its transcriptional level. However, the Our gene network results show that TBX21 plays an important Downloaded from pembrolizumab-mediated downregulation of FOXP3 at both tran- role to create an inflammatory environment in pembrolizumab- scriptional (Fig. 1F) and translational levels (Fig. 1B, 1E) shows treated iTregs (Supplemental Fig. 3). However, there was no that pembrolizumab not only masks the epitope but it also af- change in other lineage markers related to Th2, Th17, and Tregs fects the downstream signaling pathways mediated by PD-1. It (Fig. 7E, 7F). These results suggest that pembrolizumab interferes has been reported that the demethylation status of TSDR and with the differentiation of iTregs by shifting the polarization in

proximal promoter are considered as hallmarks for committed favor of Th1 and Th17. http://www.jimmunol.org/ Tregs with stable FOXP3 expression (20, 21, 46). In iTregs, it PD-1 blockade can upregulate another immune checkpoint, was reported that CpG islands of TSDR region was fully TIM-3, as a compensatory mechanism in head and neck squa- methylated (20, 47) and proximal promoter was partially mous cell carcinoma immunotherapy (57–59). These data methylated (21, 48). Our results also confirmed that TSDR is prompted us to investigate the effects of pembrolizumab on fully methylated and proximal promoter is partially methylated TIM-3expressioniniTregs.Notably,wefoundthatTIM-3is in FOXP3+ iTregs cultured in the absence or presence of pem- also downregulated upon pembrolizumab treatment even at brolizumab (Fig. 3C). Moreover, there was no significant dif- early time points. Our data show that TIM-3 was upregulated ference in the methylation pattern of both TSDR and promoter upon TCR activation like PD-1 in iTreg, and PD-1 blockade in the absence or presence of pembrolizumab, which exclude can interfere with TIM-3 upregulation (Supplemental Fig. 1). by guest on September 28, 2021 the possibility of pembrolizumab affecting the stability of Further investigations are required to confirm the role of TIM-3 FOXP3 (Fig. 3C). in iTreg polarization. It was reported that STAT1 is activated in non–small-cell lung It has been reported that PD-1/PD-L1blockade therapy works carcinoma, classical Hodgkin’s lymphoma, and melanoma upon in three ways to eliminate tumor cells: 1) activation of immune pembrolizumab treatment with IFN-g–induced signature (29, 49, cells with in the TME, 2) immune cell trafficking, and 3) rec- 50). Additionally, some reports showed that STAT1 can negatively ognize and kill tumor cells (60). Our pathway analyses showed regulate the generation of Foxp3+ Tregs (27, 51). In this study, we that genes related to immune cell trafficking were upregulated in found that p-STAT1 was upregulated in pembrolizumab-treated pembrolizumab-treated iTregs (Figs. 6, 7C). Moreover, RAD51, iTregs (Fig. 4). Moreover, we found that p-STAT1 was signifi- a key protein that plays a central role in DNA damage and repair cantly upregulated in pembrolizumab-treated FOXP3+ sub pop- (61), along with other DNA repair/cell cycle regulatory and ulation compared with nontreated FOXP3+ sub population proapoptotic genes were upregulated in pembrolizumab-treated (Fig.4D).Wealsocheckedwhether pembrolizumab-mediated iTregs (Figs. 6, 7A–C). These data suggest that pembrolizumab FOXP3 downregulation could be reversed by blocking the treatment could upregulate immune cell trafficking signals which STAT1 activation. Notably, we could not see any difference in in turn interrupt the cellular assembly by downregulating cellular FOXP3 expression with and without fludarabine treatment organization-related genes and finally leading to DNA damage (Fig. 4C). These data show that activation of STAT1 is not di- and apoptosis. This study could have important implications on rectly linked with the downregulation of FOXP3. A report showed understanding the effect of pembrolizumab on the differentiation that PD-1 blockade can shift Ag-induced cellular reactivity to- of peripherally induced Tregs in the TME. Pembrolizumab might ward Th1 and Th17 proinflammatory responses by silencing Th2 interfere with the generation of pTregs, which could augment the responses, which was evident by the upregulation of IFN-g and therapeutic efficacy of PD-1 blockade. However, the importance IL-17andalsodownregulationofIL-5andIL-13(52).Ourde- of these findings in patients treated with pembrolizumab requires tailed kinetic analysis and multiplex data of cytokine secretion further investigations as the TME is much different from in vitro showed that pembrolizumab-treated CD4+ T cells secrete higher models. levels of Th1 and Th17 cytokines during iTreg differentiation compared with nontreated iTregs (Fig. 5). Moreover, the higher Acknowledgments levels of IL-10 in nontreated iTregs suggest that pembrolizumab We thank the Genomics Core Facility at Qatar Biomedical Research Insti- interferes with IL-10 production. It has been reported that tute for RNA-Seq and Sanger sequencing. committed Tregs express minimal amounts of effector cytokines such as IFN-g, IL-4, and IL-17 (53) and produce higher levels of Disclosures suppressive cytokines such as IL-10 (54). Taken together, our The authors have no financial conflicts of interest. 12 EFFECT OF PEMBROLIZUMAB ON HUMAN-INDUCED Tregs

References 29. Ayers, M., J. Lunceford, M. Nebozhyn, E. Murphy, A. Loboda, D. R. Kaufman, A. Albright, J. D. Cheng, S. P. Kang, V. Shankaran, et al. 2017. IFN-g-related 1. Ohkura, N., Y. Kitagawa, and S. Sakaguchi. 2013. Development and mainte- mRNA profile predicts clinical response to PD-1 blockade. J. Clin. Invest. 127: nance of regulatory T cells. Immunity 38: 414–423. 2930–2940. 2. Elkord, E. 2014. Thymus-derived, peripherally derived, and in vitro-induced 30. Chikuma, S., S. Terawaki, T. Hayashi, R. Nabeshima, T. Yoshida, S. Shibayama, T regulatory cells. Front. Immunol. 5: 17. T. Okazaki, and T. Honjo. 2009. PD-1-mediated suppression of IL-2 production 3. Yadav, M., S. Stephan, and J. A. Bluestone. 2013. Peripherally induced tregs - induces CD8+ T cell anergy in vivo. J. Immunol. 182: 6682–6689. role in immune homeostasis and autoimmunity. Front. Immunol. 4: 232. 31. Hastings, W. D., D. E. Anderson, N. Kassam, K. Koguchi, E. A. Greenfield, 4. Chaudhary, B., and E. Elkord. 2016. Regulatory T cells in the tumor micro- S. C. Kent, X. X. Zheng, T. B. Strom, D. A. Hafler, and V. K. Kuchroo. 2009. environment and cancer progression: role and therapeutic targeting. Vaccines TIM-3 is expressed on activated human CD4+ T cells and regulates Th1 and (Basel) 4: e28 Th17 cytokines. Eur. J. Immunol. 39: 2492–2501. 5. Mizrahi, M., and Y. Ilan. 2009. The gut mucosa as a site for induction of reg- 32. Bae, J., D. Keskin, Y.-T. Tai, T. Hideshima, N.C. Munshi, and K. C. Anderson. ulatory T-cells. Curr. Pharm. Des. 15: 1191–1202. 2017. Pembrolizumab (anti-PD-1) treatment increases anti-tumor activities of 6. Wang, J., A. Ioan-Facsinay, E. I. van der Voort, T. W. Huizinga, and R. E. Toes. XBP1/CD138/CS1-specific cytotoxic T lymphocytes against multiple myeloma. 2007. Transient expression of FOXP3 in human activated nonregulatory CD4+ Blood 130: 1867. T cells. Eur. J. Immunol. 37: 129–138. 33. Rowshanravan, B., N. Halliday, and D. M. Sansom. 2018. CTLA-4: a moving 7. Francisco, L. M., P. T. Sage, and A. H. Sharpe. 2010. The PD-1 pathway in target in immunotherapy. Blood 131: 58–67. tolerance and autoimmunity. Immunol. Rev. 236: 219–242. 34. Trautmann, L., L. Janbazian, N. Chomont, E. A. Said, S. Gimmig, B. Bessette, 8. Ishida, Y., Y. Agata, K. Shibahara, and T. Honjo. 1992. Induced expression M.-R. Boulassel, E. Delwart, H. Sepulveda, R. S. Balderas, et al. 2006. Up- of PD-1, a novel member of the immunoglobulin gene superfamily, upon regulation of PD-1 expression on HIV-specific CD8+ T cells leads to reversible programmed cell death. EMBO J. 11: 3887–3895. immune dysfunction. [Published erratum appears in 2006 Nat. Med. 12: 1329.] 9. Chen, X., D. Fosco, D. E. Kline, L. Meng, S. Nishi, P. A. Savage, and J. Kline. Nat. Med. 12: 1198–1202. 2014. PD-1 regulates extrathymic regulatory T-cell differentiation. Eur. J. 35. Sade-Feldman, M., K. Yizhak, S. L. Bjorgaard, J. P. Ray, C. G. de Boer, Immunol. 44: 2603–2616. R. W. Jenkins, D. J. Lieb, J. H. Chen, D. T. Frederick, M. Barzily-Rokni, et al. 10. Syed Khaja, A. S., S. M. Toor, H. El Salhat, B. R. Ali, and E. Elkord. 2017. + + 2018. Defining T cell states associated with response to checkpoint immuno- Intratumoral FOXP3 Helios regulatory T cells upregulating immunosuppres- therapy in melanoma. [Published erratum appears in 2019 Cell 176: 404.] Cell Downloaded from sive molecules are expanded in human colorectal cancer. Front. Immunol. 8: 619. 175: 998–1013.e20. 11.SyedKhaja,A.S.,S.M.Toor,H.ElSalhat,I.Faour,N.UlHaq,B.R.Ali, 36. Wei, S. C., J. H. Levine, A. P. Cogdill, Y. Zhao, N. A. S. Anang, M. C. Andrews, and E. Elkord. 2017. Preferential accumulation of regulatory T cells with P. Sharma, J. Wang, J. A. Wargo, D. Pe’er, and J. P. Allison. 2017. Distinct highly immunosuppressive characteristics in breast tumor microenviron- cellular mechanisms underlie anti-CTLA-4 and anti-PD-1 checkpoint blockade. ment. Oncotarget 8: 33159–33171. Cell 170: 1120–1133.e17. 12. Kumar, S., S. Malik, U. P. Singh, S. Ponnazhagan, K. Scissum-Gunn, U. Manne, 37. Kamada, T., Y. Togashi, C. Tay, D. Ha, A. Sasaki, Y. Nakamura, E. Sato, and M. K. Mishra. 2017. PD-1 expression on Foxp3+ Treg cells modulates CD8+ S. Fukuoka, Y. Tada, A. Tanaka, et al. 2019. PD-1+ regulatory T cells amplified T cell function in prostatic tumor microenvironment. J. Immunol. 198: 155.11. by PD-1 blockade promote hyperprogression of cancer. Proc. Natl. Acad. Sci. http://www.jimmunol.org/ 13. Patnaik, A., S. P. Kang, D. Rasco, K. P. Papadopoulos, J. Elassaiss-Schaap, USA 116: 9999–10008. M. Beeram, R. Drengler, C. Chen, L. Smith, G. Espino, et al. 2015. Phase I study 38. Zappasodi, R., S. Budhu, M. D. Hellmann, M. A. Postow, Y. Senbabaoglu, of pembrolizumab (MK-3475; anti-PD-1 monoclonal antibody) in patients with S. Manne, B. Gasmi, C. Liu, H. Zhong, Y. Li, et al. 2018. Non-conventional advanced solid tumors. Clin. Cancer Res. 21: 4286–4293. inhibitory CD4+Foxp3-PD-1hi T cells as a biomarker of immune checkpoint 14. Boyiadzis, M. M., J. M. Kirkwood, J. L. Marshall, C. C. Pritchard, N. S. Azad, blockade activity. [Published erratum appears in 2018 Cancer Cell 34: 691.] and J. L. Gulley. 2018. Significance and implications of FDA approval of Cancer Cell 33: 1017–1032.e7. pembrolizumab for biomarker-defined disease. J. Immunother. Cancer 6: 35. 39. Ribas, A., D. S. Shin, J. Zaretsky, J. Frederiksen, A. Cornish, E. Avramis, 15. Toor, S. M., A. S. Syed Khaja, I. Alkurd, and E. Elkord. 2018. In-vitro effect of E. Seja, C. Kivork, J. Siebert, P. Kaplan-Lefko, et al. 2016. PD-1 blockade ex- pembrolizumab on different T regulatory cell subsets. Clin. Exp. Immunol. 191: pands intratumoral memory T cells. Cancer Immunol. Res. 4: 194–203. 189–197. 40. Ellestad, K. K., G. Thangavelu, C. L. Ewen, L. Boon, and C. C. Anderson. 2014. 16. Toor, S. M., V. Sasidharan Nair, G. Pfister, and E. Elkord. 2019. Effect of PD-1 is not required for natural or peripherally induced regulatory T cells: severe pembrolizumab on CD4+CD25+, CD4+LAP+ and CD4+TIM‐3+ T cell subsets.

autoimmunity despite normal production of regulatory T cells. Eur. J. Immunol. by guest on September 28, 2021 Clin. Exp. Immunol. 196: 345–352. 44: 3560–3572. 17. Bredel, M., D. M. Scholtens, G. R. Harsh, C. Bredel, J. P. Chandler, 41. Lee, H. T., S. H. Lee, and Y. S. Heo. 2019. Molecular interactions of antibody J. J. Renfrow, A. K. Yadav, H. Vogel, A. C. Scheck, R. Tibshirani, and B. I. Sikic. drugs targeting PD-1, PD-L1, and CTLA-4 in immuno-oncology. Molecules 2009. A network model of a cooperative genetic landscape in brain tumors. 24: e1190. JAMA 302: 261–275. 42. Ali, M. H. M., S. M. Toor, F. Rakib, R. Mall, E. Ullah, K. Mroue, P. R. Kolatkar, 18. Kra¨mer, A., J. Green, J. Pollard, Jr., and S. Tugendreich. 2014. Causal analysis K. Al-Saad, and E. Elkord. 2019. Investigation of the effect of PD-L1 blockade approaches in ingenuity pathway analysis. Bioinformatics 30: 523–530. on triple negative breast cancer cells using fourier transform infrared spectros- 19. Sasidharan Nair, V., H. El Salhat, R. Z. Taha, A. John, B. R. Ali, and E. Elkord. copy. Vaccines (Basel) 7: e109. 2018. DNA methylation and repressive H3K9 and H3K27 trimethylation in the 43. Fessas, P., H. Lee, S. Ikemizu, and T. Janowitz. 2017. A molecular and pre- promoter regions of PD-1, CTLA-4, TIM-3, LAG-3, TIGIT, and PD-L1 genes in clinical comparison of the PD-1-targeted T-cell checkpoint inhibitors nivolumab human primary breast cancer. Clin. Epigenetics 10: 78. and pembrolizumab. Semin. Oncol. 44: 136–140. 20. Baron, U., S. Floess, G. Wieczorek, K. Baumann, A. Gru¨tzkau, J. Dong, 44. Chapman, N. M., and H. Chi. 2014. mTOR signaling, Tregs and immune A. Thiel, T. J. Boeld, P. Hoffmann, M. Edinger, et al. 2007. DNA demethylation modulation. Immunotherapy 6: 1295–1311. in the human FOXP3 locus discriminates regulatory T cells from activated 45. Lu, Y., J. Wang, J. Gu, H. Lu, X. Li, X. Qian, X. Liu, X. Wang, F. Zhang, and FOXP3(+) conventional T cells. Eur. J. Immunol. 37: 2378–2389. L. Lu. 2014. Rapamycin regulates iTreg function through CD39 and Runx1 21. Janson, P. C., M. E. Winerdal, P. Marits, M. Tho¨rn, R. Ohlsson, and O. Winqvist. pathways. J. Immunol. Res. 2014: 989434. 2008. FOXP3 promoter demethylation reveals the committed Treg population in 46. Huehn, J., J. K. Polansky, and A. Hamann. 2009. Epigenetic control of FOXP3 humans. PLoS One 3: e1612. expression: the key to a stable regulatory T-cell lineage? Nat. Rev. Immunol. 22. Na, Z., S. P. Yeo, S. R. Bharath, M. W. Bowler, E. Balıkc¸ı, C. I. Wang, and 9: 83–89. H. Song. 2017. Structural basis for blocking PD-1-mediated immune suppression 47. Polansky, J. K., K. Kretschmer, J. Freyer, S. Floess, A. Garbe, U. Baron, S. Olek, by therapeutic antibody pembrolizumab. Cell Res. 27: 147–150. A. Hamann, H. von Boehmer, and J. Huehn. 2008. DNA methylation controls 23. Gianchecchi, E., and A. Fierabracci. 2018. Inhibitory receptors and pathways Foxp3 gene expression. Eur. J. Immunol. 38: 1654–1663. of lymphocytes: the role of PD-1 in Treg development and their involvement in 48. Polansky, J. K., L. Schreiber, C. Thelemann, L. Ludwig, M. Kru¨ger, autoimmunity onset and cancer progression. Front. Immunol. 9: 2374. R. Baumgrass, S. Cording, S. Floess,A.Hamann,andJ.Huehn.2010. 24. Vercoulen, Y., T. Guichelaar, J. Meerding, M. Emmelot, M. Pingen, W. de Jager, Methylation matters: binding of Ets-1 to the demethylated Foxp3 gene T. Mutis, A. Martens, P. Coffer, and B. Prakken. 2011. Human induced contributes to the stabilization of Foxp3 expression in regulatory T cells. CD4+CD25+FOXP3+ regulatory T cells are suppressive in vitro, but fail to J. Mol. Med. (Berl.) 88: 1029–1040. suppress inflammation in vivo. Ann. Rheum. Dis. 70: A53. 49. Shin, D. S., J. M. Zaretsky, H. Escuin-Ordinas, A. Garcia-Diaz, S. Hu-Lieskovan, 25. Volta, V., A. Ernlund, A. Valeta, S. Demaria, and R. J. Schneider. 2015. MTOR A. Kalbasi, C. S. Grasso, W. Hugo, S. Sandoval, D. Y. Torrejon, et al. 2017. inhibition favors the differentiation of human in vitro-induced regulatory T cell Primary resistance to PD-1 blockade mediated by JAK1/2 mutations. Cancer through selective protein synthesis. J. Immunother. Cancer 3(Suppl. 2): P79. Discov. 7: 188–201. 26. Huber, S., J. Schrader, G. Fritz, K. Presser, S. Schmitt, A. Waisman, S. Lu¨th, 50. Shindiapina, P., and L. Alinari. 2018. Pembrolizumab and its role in M. Blessing, J. Herkel, and C. Schramm. 2008. P38 MAP kinase signaling relapsed/refractory classical Hodgkin’s lymphoma: evidence to date and is required for the conversion of CD4+CD25- T cells into iTreg. PLoS One clinical utility. Ther. Adv. Hematol. 9: 89–105. 3: e3302. 51.Knosp,C.A.,C.Schiering,S.Spence,H.P.Carroll,H.J.Nel,M.Osbourn, 27. Chang, J. H., Y. J. Kim, S. H. Han, and C. Y. Kang. 2009. IFN-gamma-STAT1 R. Jackson, O. Lyubomska, B. Malissen, R. Ingram, et al. 2013. Regulation of signal regulates the differentiation of inducible Treg: potential role for Foxp3+ inducible regulatory T cell stability by SOCS2. J. Immunol. 190: 3235–3245. ROS-mediated apoptosis. Eur. J. Immunol. 39: 1241–1251. 52. Dulos, J., G. J. Carven, S. J. van Boxtel, S. Evers, L. J. Driessen-Engels, 28. Ma, H., C. Lu, J. Ziegler, A. Liu, A. Sepulveda, H. Okada, S. Lentzsch, and W. Hobo, M. A. Gorecka, A. F. de Haan, P. Mulders, C. J. Punt, et al. 2012. PD-1 M. Y. Mapara. 2011. Absence of Stat1 in donor CD4+ T cells promotes the blockade augments Th1 and Th17 and suppresses Th2 responses in peripheral expansion of Tregs and reduces graft-versus-host disease in mice. J. Clin. Invest. blood from patients with prostate and advanced melanoma cancer. J. Immunother. 121: 2554–2569. 35: 169–178. The Journal of Immunology 13

53. Kryczek, I., R. Liu, G. Wang, K. Wu, X. Shu, W. Szeliga, L. Vatan, E. Finlayson, radiotherapy and PD-L1 blockade is mediated by TIM-3 upregulation and reg- E. Huang, D. Simeone, et al. 2009. FOXP3 defines regulatory T cells in human ulatory T-cell infiltration. Clin. Cancer Res. 24: 5368–5380. tumor and autoimmune disease. Cancer Res. 69: 3995–4000. 58. Shayan, G., and R. L. Ferris. 2015. PD-1 blockade upregulate TIM-3 expression 54. O’Garra, A., P. L. Vieira, P. Vieira, and A. E. Goldfeld. 2004. IL-10-producing as a compensatory regulation of immune check point receptors in HNSCC TIL. and naturally occurring CD4+ Tregs: limiting collateral damage. J. Clin. Invest. J. Immunother. Cancer 3(Suppl. 2): P196. 114: 1372–1378. 59. Shayan, G., R. Srivastava, J. Li, N. Schmitt, L. P. Kane, and R. L. Ferris. 2016. 55. Wei, X., J. Zhang, Q. Gu, M. Huang, W. Zhang, J. Guo, and X. Zhou. 2017. Adaptive resistance to anti-PD1 therapy by Tim-3 upregulation is mediated by Reciprocal expression of IL-35 and IL-10 defines two distinct effector Treg subsets the PI3K-Akt pathway in head and neck cancer. OncoImmunology 6: e1261779. that are required for maintenance of immune tolerance. Cell Rep. 21: 1853–1869. 60. Kim, J. M., and D. S. Chen. 2016. Immune escape to PD-L1/PD-1 blockade: 56. Szabo, S. J., S. T. Kim, G. L. Costa, X. Zhang, C. G. Fathman, and seven steps to success (or failure). Ann. Oncol. 27: 1492–1504. L. H. Glimcher. 2000. A novel transcription factor, T-bet, directs Th1 lineage 61. Bhattacharya, S., K. Srinivasan, S. Abdisalaam, F. Su, P. Raj, I. Dozmorov, commitment. Cell 100: 655–669. R. Mishra, E. K. Wakeland, S. Ghose, S. Mukherjee, and A. Asaithamby. 2017. 57. Oweida, A., M. K. Hararah, A. Phan, D. Binder, S. Bhatia, S. Lennon, RAD51 interconnects between DNA replication, DNA repair and immunity. S. Bukkapatnam, B. Van Court, N. Uyanga, L. Darragh, et al. 2018. Resistance to Nucleic Acids Res. 45: 4590–4605. Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021