An arginase1- and PD-L1-derived peptide-based vaccine for myeloproliferative neoplasms: A first-in-man clinical trial

Introduction Arginase-1 (ARG1) and Programed death ligand-1 (PD-L1) play a vital role in immunosuppression in myeloproliferative neoplasms (MPNs) and directly inhibit T-cell activation and proliferation. We previously identified spontaneous T-cell responses towards PD-L1 and ARG1 derived peptide epitopes in patients with MPNs. In the present First-in-Man study we tested dual vaccinations of ARG1- derived and PD-L1-derived peptides, combined with Montanide ISA-51 as adjuvant, in patients with Janus Kinase 2 (JAK2) V617F-mutated MPN. Methods Safety and efficacy of vaccination with ARG1- derived and PD-L1-derived peptides with montanide as an adjuvant was tested in 9 patients with MPN The primary end point was safety and toxicity evaluation. The secondary end point was assessment of the immune response to the vaccination epitope (www.clinicaltrials.gov identifier NCT04051307). Results The study included 9 patients with JAK2-mutant MPN of which 8 received all 24 planned vaccines within a 9-month treatment period. Patients reported only grade 1 and 2 vaccine related adverse events. No alterations in peripheral blood counts were identified, and serial measurements of the JAK2V617F allelic burden showed that none of the patients achieved a molecular response during the treatment period. The vaccines induced strong immune responses against both ARG1 and PD-L1- derived epitopes in the peripheral blood of all patients, and vaccine-specific skin-infiltrating lymphocytes from 5/6 patients could be expanded in vitro after a delayed-type hypersensitivity test. In two patients we also detected both ARG1- and PD-L1-specific T cells in bone marrow samples at the end of trial. Intracellular cytokine staining revealed IFNγ and TNFγ producing CD4+- and CD8+- T cells specific against both vaccine epitopes. Throughout the study, the peripheral CD8/CD4 ratio increased significantly, and the CD8+ TEMRA subpopulation was enlarged. We also identified a significant decrease in PD-L1 mRNA expression in CD14+ myeloid cells in the peripheral blood in all treated patients and a decrease in ARG1 mRNA expression in bone marrow of 6 out of 7 evaluated patients. Conclusion Overall, the ARG1- and PD-L1-derived vaccines were safe and tolerable and induced strong T-cell responses in all patients. These results warrant further studies of the vaccine in other settings or in combination with additional immune-activating treatments.


Introduction
Philadelphia chromosome-negative myeloproliferative neoplasms (MPNs) include essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis. These three diseases almost exclusively exhibit certain driver mutations: the JAK2V617F point mutation, the calreticulin gene (CALR) frameshift mutations, or the thrombopoietin receptor gene (MPL) mutation. The treatment of MPN is aimed at reducing the risk of thromboembolic episodes through lifestyle intervention, low-dose aspirin, and cytoreductive therapies such as hydroxyurea (HU) and pegylated interferon alpha (IFN-a). Patients with PV and ET have a 5-10% risk of developing acute myeloid leukemia within 20 years (1). Currently, the only available curative treatment modality is allogeneic hematopoietic stem cell transplantation (alloHSC). This treatment is, however, associated with high mortality and is thus only used in patients with high or intermediate risk myelofibrosis (2).
During the last decade cancer immune therapy based on targeting immunosuppressive mechanisms has shown great potential in treatment of solid tumors (3) and several of the key known immunosuppressive pathways such as programmed death ligand-1 (PD-L1) and arginase1 (ARG1) have also been identified in MPN (4)(5)(6). Thus, cancer immune therapies targeting these suppressive mechanisms could be a potential new treatment option for MPN.
PD-L1 is upregulated on either tumor cells or tumor-infiltrating cells in many cancers and is known to diminish T-cell activation through its interaction with programmed death receptor 1 (PD-1) on T cells (7). Antibodies that block the PD-1:PD-L1 axis have been effective in different types of cancers, including hematologic malignancies (7)(8)(9), but these drugs can have severe side effects (10). Prestipino et al. showed that the JAK2V617F mutation in MPNs upregulates PD-L1 by activation of STAT3 and STAT5, transcription factors for the CD274 gene, and thereby mediates immune escape in MPN (5). Other studies have shown that PD-L1 expression is increased in patients with MPN compared to healthy controls, regardless of the driver mutation (11)(12)(13). Arginase-1 (ARG1), on the other hand, is often expressed by regulatory immune cells such myeloid-derived suppressor cells (MDSC) and exerts its suppressive function reducing the availability of L-arginine in the tumor microenvironment and thus causing the downregulation of the CD3z chain, and inhibiting T-cell proliferation (14,15). It has been shown that patients with MPN have higher levels of MDSCs and generally higher ARG1 mRNA expression levels in peripheral blood compared to healthy controls (6).
T cells that specifically target epitopes derived from immune suppressive protein expressed by immune-suppressive cells are defined as anti-Tregs (16). Immunogenic epitopes from multiple immunosuppressive proteins such as indoleamine 2,3-dioxygenase (IDO) (17), ARG1, and PD-L1 have been identified and characterized (18,19). Anti-Tregs appear to be important for immune homeostasis, due to their ability to directly react against regulatory immune cells (16) by suppressing their inhibitory function and promoting a more pro-inflammatory microenvironment (18,20). Accordingly, we reasoned that, by activating anti-Tregs specific for PD-L1 and ARG1 derived epitopes through peptide vaccination, it may be an effective strategy to reinstate immune homeostasis in patients with MPN by immunomodulation of the tumor microenvironment and promotion of the tumor-specific T cell responses.
In 2013, we identified immunogenic peptide epitopes in PD-L1 (21) and spontaneous T-cell responses against these epitopes were identified in peripheral blood mononuclear cells (PBMC) from both healthy controls and patients with cancer (21). These specific T cells were able to kill melanoma cell lines as well as dendritic cells, in a PD-L1-dependent manner (21). and enhanced virus-and cancer-specific T-cell responses in vitro (18,22). Similarly, we identified a 50-amino acid (50-aa) hotspot region of immunogenic epitopes in the ARG1 protein (aa 161-210) (19). Exploration of this hotspot region identified a 38-aa peptide, 'ArgLong2', that activated frequent, strong, and spontaneous ARG1-specific T-cell responses in PBMC samples from heathy donors as well as cancer patients (20,23). We frequently observed spontaneous responses against ARG1 and PD-L1-derived peptides in T cells from patients with MPN (12,24).
Among the immunogenic peptides identified in PD-L1, PD-L1Long1 was tested previously in four clinical trials. One trial was performed in ten patients with multiple myeloma after high dose chemotherapy and autologous hematopoietic stem cell transplantation. Of the 10 patients vaccinated, three showed clinical improvement (25). The second trial included patients with basal cell carcinoma, where a PD-L1Long1 vaccine, in combination with Montanide ISA-51, showed a profound effect on tumor lesions (26). In the third trial, patients with follicular lymphoma were vaccinated with PD-L1Long1 in combination with a PD-L2-derived peptide, and disease remission was detected in follow-up (27). Finally, PD-L1-vaccines were administered in combination with an IDOderived peptide and a PD-1 blocking antibody to patients with metastatic malignant melanoma. The study showed a remarkable clinical effect, with a staggering 80% overall response rate (28). These four studies underlined the potential of a PD-L1-based peptide vaccine in solid as well as hematological cancers.
In the present study we report on a phase I first-in-man study in which ArgLong2 and PD-L1Long1 peptide vaccine combined with the adjuvant Montanide ISA-51 was performed in patients with MPN.

Materials and methods
This phase I-II clinical vaccination trial was initiated at Herlev and Gentofte Hospital, Capital Region of Denmark. The trial aimed to determine the immunogenicity, clinical efficacy, and safety of a dual vaccination with ArgLong2 and PD-L1Long1 peptides, combined with Montanide ISA-51 as adjuvant, in patients with MPN. Patients were enrolled in the trial in the Departments of Hematology at Herlev and Gentofte Hospital, Copenhagen, and Zealand University Hospital, Roskilde, Denmark. The trial started on 7 October 2019 and the last patient received the last vaccine on the January 26 th , 2021.
All participants provided written informed consent before trial enrollment. The protocol was approved by the Ethics Committee of the Capital Region of Denmark, the National Board of Health, and the Danish Data Protection Agency, and it was registered at https://www. clinicaltrials.gov (NCT04051307; date of registration: August 9, 2019).
We intended to include 24 patients in the trial and planned for the possibility of including an additional 24 patients if two or more patients from the first cohort showed a clinical response. However, due to the COVID-19 pandemic this could not be met. The main inclusion criteria were a diagnosis of ET or PV, according to WHO criteria (29). A full list of the inclusion and exclusion criteria is shown in Supplementary Figure 1. Patients could receive concurrent treatments with IFN-a, HU, or Anagrelide (ANA) in any combination, but no other anti-neoplastic or anti-MPN treatments were permitted.
To evaluate clinical responses, we applied the response criteria for PV and ET (30). A 10% reduction of the allele burden, based on a validated, in-house qPCR method, was defined as a response.
The two vaccines were administered at the same time. Briefly, the peptides were individually dissolved in 500 µg sterile water/DMSO and emulsified with 500 µl Montanide ISA-51, just prior to administration. The vaccines were administered subcutaneously, one in each shoulder, every two weeks. Patients received six treatments (i.e., 12 vaccinations) over 12 weeks. After these first six treatments, treatment was paused for 1-3 months; then, six more treatments were given, again at two-week intervals. An additional six treatments could be scheduled for patients that showed a response. The entire treatment plan is shown in Supplementary Figure 2.
Adverse events, safety, and toxicity evaluations Adverse events were assessed according to the Common Terminology Criteria for Adverse Events, version 5.0. All patients were evaluated prior to inclusion with a medical examination by the treating physician. This included spleen palpation (no ultasonography nor computed tomography was performed), blood sample analyses (Hemoglobin, leukocyte differentiation count, platelets, IgG, IgA, IgM, Hematocrit, Bilirubin, Potassium, Sodium, Creatinine, albumin, uric acid, lactate dehydrogenase, Alkaline Phosphatase, Alanine transaminase, amylase, bilirubin, D-dimer, ionized calcium, C-Reactive Protein, Thyrotropin, thyroxin, Luteinizing Hormone, Adrenocorticotropic Hormone, Cortisol, Hepatitis B, hepatitis C (IgG), HIV, HTLV-1(IgG), IgG and IgM for Cytomegalovirus (CMV), Epstein-Barr Virus (EBV) and toxoplasmosis) and an electrocardiogram. According to the treatment plan, patients were evaluated at inclusion, during treatment pause, and at the end of the trial (EOT). At every vaccine treatment, an investigator recorded the patient's symptoms, and when necessary, conducted a medical examination and evaluation. Bone marrow biopsies were acquired from patients before trial entry and at end of the trial. Histopathological evaluation of nonblinded biopsies was performed by a trained hematopathologist.

DNA analyses with digital droplet polymerase chain reaction
We analyzed patient DNA samples for mutations with the droplet digital PCR (ddPCR) method, on a QX100 system (Bio-Rad, Hercules, California, USA), according to manufacturer instructions. Briefly, we mixed 10 µl of 2x digital PCR Supermix for probes (Bio-rad), 2 µl primer/probe mix, 3 µl nuclease-free water, and 5 µl DNA (20 ng/µl), for a total volume of 20 µl. Droplets were generated on a QX100 Droplet Generator System (Bio-rad). The polymerase chain reaction (PCR) was performed with an initial stage at 95°C for 10 min, followed by 43 cycles of 94°C for 30 sec and 57°C for 60 sec, then a final stage at 98°C for 10 min. PCR was carried out on an Applied Biosystems Veriti 96-Well Thermal Cycler (Thermo Fisher, Waltham, MA, USA). Droplets were subsequently quantified on a QX100 Droplet Reader (Bio-rad) and analyzed with Quantasoft ™ Analysis Pro software.

Isolation of bone marrow and peripheral blood
For peripheral blood mononuclear cell (PBMC) isolation, blood samples were obtained and cryopreserved as previously reported (31) at baseline, after three vaccinations, after six vaccinations (during the treatment pause), after seven vaccinations, after nine vaccinations, and at EOT. Heparinized bone marrow aspirations (10 mL in a heparinized tube) were obtained at baseline and at the end of the trial. Ortho-Lysing Buffer diluted 10× in H 2 0 was added to half of the sample, followed by centrifugation and incubation for 15 minutes in the dark. The other half of the sample was handled and cryopreserved following the same procedure as for PBMCs.

Delayed-type hypersensitivity and skininfiltrating lymphocytes
To assess the presence of PD-L1 and ARG1-specific lymphocytes, we performed a delayed-type hypersensitivity (DTH) test. DTH tests were assessed at baseline and at EOT. Briefly, we administered intradermal injections of the two peptides, without the adjuvant, at the lower back. The peptides were dissolved in sterile water and DMSO. We also administered a control injection, which included water and DMSO, but no peptide. At 48 h after the injection, skin reactions (induration) were measured; additionally, at the sites of ArgLong2-and PD-L1long1 injections, punch biopsies were acquired and cut into fragments. To identify fragments that contained skin-infiltrating lymphocytes (SKILs), fragments were cultured in 24-well plates in RPMI-1640 with 10% human serum and 100 U/mL interleukin-2 (IL-2), with penicillin, streptomycin, and fungizone, for 3-5 weeks to allow SKIL outgrowth. Every second or third day, half the medium was replaced with fresh medium containing IL-2, penicillin, streptomycin, and fungizone. After 3-5 weeks, SKILs were harvested, and a fraction was tested in ELISPOT assays. The remaining SKILs were cryopreserved.

Interferon-g ELISPOT assays in vitro and ex vivo
Immune responses were evaluated with in vitro and ex vivo IFN-g ELISPOT assays. For in vitro ELISPOT assays, PBMCs were thawed and stimulated with the target epitope. The next day, PBMCs were stimulated with IL-2 (120 U/mL) and incubated for 12-14 days. The PBMCs were then counted and plated in ELISPOT wells. Cells were restimulated with or without the target peptide. All conditions were performed in triplicates. For ex vivo IFN-g ELISPOT assays, cells were thawed, rested, then plated on ELISPOT plates. Cells were stimulated with or without the target epitope for 24-48 h to ensure antigen presentation.
In vitro ELISPOT assays were performed with a cell density of 2.5 ×10 5 cells/well. Ex vivo ELISPOTs were performed with cell densities of 9 ×10 5 cells/well, for PBMCs, and 6.8×10 5 cells/well, for bone marrow mononuclear cells. Plates were analyzed with the ImmunoSpot Series 2.0 Analyzer (CTL, Shaker Heights, Ohio). Results were generated by subtracting the background obtained with negative controls. A detailed description of our setup was described previously (32). Statistical significance of ELISPOT responses was analyzed by the DFR method (33). For non-triplicate samples, responses were evaluated empirically and defined as true if the number of spots observed in the peptide stimulated wells were at least double of the spot counts in the control wells.

RT-qPCR analysis of arginase-1 and PD-L1 expression
Analysis was performed on samples derived from patient PBMCs and bone marrow-derived MNCs isolated at baseline and end-of-trial. For the PBMCs, CD14 Microbeads (Miltenyi Biotec) were utilized to separate CD14+ and CD14-cells. Total RNA purification was performed using the RNEasy Plus Mini Kit (Qiagen) according to the manufacturer's protocol. RNA concentration was measured on a NanoDrop2000 Spectrophotometer (Thermo Scientific). cDNA was synthesized using the iScript ™ cDNA Synthesis Kit (Bio-Rad) based on 500 ng or 400 ng total RNA (PBMC-derived cells and bone marrow, respectively). RT-qPCR analysis was performed using the TaqMan Gene Expression Assay on a Roche LightCycler 480 instrument. The assay was performed in technical triplicates for all primers and the subsequent data analysis was performed as previously described using the DDCt-method (34) with normalization of PD-L1 (Primer ID Hs01125296_m1) and ARG1 (Primer ID Hs00163660_m1) expression to the housekeeping gene POL2RA (Primed ID Hs00172187_m1) and to the baseline sample. Controls lacking reverse transcriptase during cDNA creation were included for primer validation.

Phenotyping of PMBCs and bone marrow mononuclear cells
PBMCs collected at three time-points and bone marrow collected at 2 time-points were thawed and washed in preheated phosphate-buffered saline. Fc-receptors were blocked by incubating with human IgG (50 mg/ml), and dead cells were stained with the Fixable Near-IR Dead cell stain kit (Thermo-Fisher). After mixing, the cells were stained in the dark at 4°C for 20 min with fluorochrome-labeled antibodies (Supplementary Figure 3). Next, the cells were washed and analyzed with a NovoCyte Quanteon Flow Cytometer (Agilent, Santa Clara, CA). The gating strategy is described in Supplementary Figure 4. Data were analyzed with NovoExpress 1.5.1 software. All gates at baseline were applied to all timepoints. Illustrations were created with Graphpad Prism v 8.0 (GraphPad Software. Inc.).

Statistical analysis
ELISPOT responses were analyzed with the distribution free resampling (DFR) method (33). DFR analyses were performed with the R statistical analysis program. Immune subsets at different time points were compared with the Wilcoxon matched-pairs signedrank test. P values ≤0.05 were considered significant. Immune subset analyses were performed in Graphpad Prism v 8.0 (GraphPad Software. Inc.).

Patient characteristics and clinical response evaluation
We evaluated bone marrow biopsies and blood samples from 12 patients before inclusion in the vaccination trial. Three patients did not meet the inclusion criteria and were excluded: One patient had patient had unmeasurable JAK2V617F and normal bone marrow and thus did not meet the diagnostic criteria, one patient had progressive disease and proceeded to alloHSC and one patient had additional findings in the bone marrow suggesting a more MDS-like disease and thereby did not meet the diagnostic criteria Thus, nine patients (5 male, 4 female, median age: 57 years, range: 46 to 72) with a median disease duration (time from diagnosis) of 2 years (range: 4 months to 10 years) were enrolled in the clinical study. The vaccines were administered over a period of six to nine months. Among the nine patients, eight received a minimum of 12 vaccines; one patient did not receive one ArgLong2 vaccine. One patient (#3) proceeded with an extra round of vaccinations, due to an apparent drop in the JAK2V617F allele burden, measured with qPCR. This drop could however not be confirmed by ddPCR. The primary end of trial (EOT) was defined as 12 treatments.

Vaccine tolerability and safety
Vaccines were generally well tolerated: the majority of reported adverse events were grades 1 and 2 ( Table 2). Injection site reaction (grade 2) was the most commonly observed adverse event and was reported by all the patients at least once during the study period. In two patients, injection site reactions remained visible for one year after the last vaccine (patients #4 and #5). Only one grade 3 adverse event was registered: one patient (# 9) had a vasovagal reaction immediately after the first PD-L1Long1 vaccination and did not receive the ArgLong2 peptide vaccine during the same hospital visit. However, the patient continued the treatments afterwards and received both vaccines, as scheduled, without similar reactions. There were two peculiar adverse events reported by the same patient during the trial: a change in taste (grade 1 dysgeusia) and a change of body odor (grade 1). The patient reported on this at the time of the last vaccine, but claimed that the changes started already after the first vaccine. The reactions were deemed to be related to the DMSO contained in the vaccine solution. Patient # 1 reported several reactivations of Herpes Simplex virus during the trial (at treatments 3, 5, 7, and 11). The patient had a history of Herpes Simplex infections, but the last reactivation had occurred several years before study inclusion.

Immune responses
To assess the vaccine induced immune responses, we analyzed PBMCs and BMNCs for T-cell responses against the two peptides used in the trial by ex vivo IFN-g ELISPOT assay (Figure 2). We detected a spontaneous response against the ARG1-and the PD-L1-derived peptides in PBMCs from one out of nine and from three out of nine patients at baseline respectively. During treatment, the responses against the ARG1-derived and PD-L1-derived peptides were detected ex vivo in PBMCs from five and eight patients respectively (Figures 2A-D). In patient 5, we did not observe a significant ex vivo PBMC response during treatment, even though the patient's PBMCs displayed a spontaneous T-cell response at baseline. Ex vivo IFN-g ELISPOT was performed using bone marrow mononuclear cells (BMNC) from patients 4 and 6, and cells from both patients showed vaccine-specific T cells after treatment with an apparent increase in responses in EOT samples compared to baseline (Figures 2E, F). Due to low viability of the isolated BMNC, we were not able to perform a similar assay on BMNCs from the remaining seven patients.
In an addition to ex vivo assays, we performed an in vitro IFN-g ELISPOT and intracellular cytokine staining (ICS) on in vitro prestimulated PBMCs to improve the detection of the treatment induced response and phenotypically characterize the vaccine reactive T cell populations (Figure 3; Supplementary Figures 6 D-F). At baseline, we detected spontaneous responses against the ARG1-derived peptide in   Figure 6B). Only PBMCs from patient 8 did not display a spontaneous response against neither the PD-L1 nor ARG1 epitopes at baseline. PBMCs from all patients displayed greatly enhanced T-cell responses against the PD-L1-derived epitope after 3 vaccinations ( Figure 3B; Supplementary Figures 6B, C).
In ICS we chose to analyze PBMCs collected at time points in which the individual patients showed the strongest immune response as detected by in vitro ELISPOT (Figures 3A, B). We detected that the majority of the vaccine specific T cells were producing TNF-a and/or IFN-g in response to peptides used in the vaccination. Responses against ArgLong2 were identified in all nine patients, though primarily in the CD4 + T-cell compartment with lower incidence of CD8 + T-cell responses detected in 6 out of 9 patients (Figures 3C, E). Similarly, the analyses confirmed that all 9 patients generated CD4 + T-cell response against PD-L1-derived peptide and 6 out of 9 patients also displayed a lower magnitude of CD8 + T cell response against the same peptide ( Figures 3D, F).
Induction and expansion of vaccine specific T cell responses was also evaluated by delayed type hypersensitivity (DTH) testing. Six out of nine patients provided informed consent for this analysis. Baseline DTH biopsies collected from four out of six patients could not be expanded to obtain skin infiltrating lymphocytes (SKILs) cultures. In baseline SKILs samples obtained from patients 2 and 9 we could only detect a significant response against the ARG1-derived peptide in IFNg ELISPOT in patient 9 (Supplementary Table 1). At EOT we expanded SKILs from biopsies collected from all six patients and in 5 out 6 analyzed patients, the SKILS exhibited vaccine-specific responses against one or both peptides as determined by IFNg ELISPOT signifying expansion and target epitope dependent homing of the vaccine induced T cells (Supplementary Table 1). None of the patients displayed a visible skin reaction at baseline, however at EOT (corresponding to the expanded SKILs samples), all patients except patient 1 showed skin reactions with redness and swelling to various degrees at all injected DTH sites.

Phenotypic characterization of PBMCs during treatment
We investigated the peripheral blood of the patients for changes in composition of immune cell subsets during the vaccination trial. We performed fluorescence-activated cell sorting (FACS) on isolated PBMC at three time-points: baseline, treatment pause (after 6 treatments), and EOT. Bone marrow aspirations acquired at baseline and at EOT were also analyzed.
The fraction of CD3 + T cells in PBMCs remained unchanged throughout the study period ( Figure 4A). However, the percentage of CD4 + T cells decreased significantly from baseline to EOT, and accordingly the fraction of CD8 + T cells increased significantly during treatment ( Figures 4B, C). The significant increase in the percentage of CD8 + T cells was detectable already after six treatments. We further analyzed the changes in the CD8 + T cell subsets and observed a significant decrease and increase in the CD8 + T cell T EM and T EMRA subpopulations respectively by the end of the treatment (Figures 4F, G). A significant decrease in T EM cells was already detectable at the treatment pause. T EMRA cells increased from baseline to EOT. Additionally, while no changes were seen in the T CM population, a significant decrease in CD8 + T Naïve cells was seen (Figures 4D, E). The CD4 + T-cell subsets remained unchanged, including central memory (T CM ), effector memory (T EM ), T Naive , and T EMRA cells (Supplementary Figure 7). PD-1 expression on CD3 + T cells as well as CD4 + and CD8 + T cell subsets remained unchanged ( Supplementary Figures 8A-C). No changes in levels of regulatory T cells, natural killer (NK) cells, or B cells were observed during treatment (data not shown). CD16 dim CD56 hi NK cells decreased significantly during the trial. MDSCs, defined as HLA -DR -CD33 + CD14 + cells, monocytes, and dendritic cells remained stable during treatment. CD3 + -, CD4 + -, CD8 + -and NK-cells in the bone marrow remained stable during treatment (data not shown).

Changes in ARG1 and PD-L1 expression in treated patients
Changes in the expression of PD-L1 were assessed by flow cytometry and no alterations of PD-L1 expression on the myeloid cells (CD3 -CD19 -CD56cells) were observed during the trial (Supplementary Figure 8D). Changes in PD-L1 and ARG1 expression were also evaluated in PBMCs (n=9) and BMNCs (n=6) of the treated patients using RT-qPCR. Interestingly, a significant decrease in the PD-L1 mRNA expression was seen in CD14 + myeloid cells in the peripheral blood from baseline to end-oftreatment ( Figure 4H). Decrease in PD-L1 expression was also detected in CD14 Neg cell fraction in 6 out of 7 evaluated patient samples (Supplementary Figure 8E). Interestingly, while no ARG1 expression was detected in the peripheral blood in neither CD14 + nor CD14 Neg cell fractions, a clear decrease in the ARG1 expression was detected in BMNCs of 5 out of 6 evaluated patients ( Figure 4I). Contrary to ARG1, expression of PD-L1 was only detected in BMNCs of one patient (#5) and an increase in PD-L1 expression was seen in this sample (Supplementary Figure 8F). These results suggest a potential differential expression of ARG1 and PD-L1 between the peripheral blood and the bone marrow compartments in patients with MPN.

Discussion
We successfully conducted a First-in-Man trial with an ARG1derived peptide vaccine combined with a PD-L1-derived peptide vaccine in patients with MPN. We found that the vaccines were safe as vaccine related adverse events of only grade 1 and 2 were observed. The most common adverse event was development of granulomas at the injection site during treatment, which was observed in all patients and was related to the adjuvant (35). A single grade 3 vasovagal adverse event was reported, but was concluded to be unrelated to the vaccines.
We have previously demonstrated that ARG1-and PD-L1specific T cells can directly target immunosuppressive cells (18,) (20,) (23), and we therefore hypothesized that a boost in the ARG1and PD-L1-specific T-cell responses should lead to a reduction in the number of immunosuppressive cells in MPN which in turn could increase the tumor specific T-cell responses in treated patients. Our previous studies (12,24) have shown that among patients with MPN patients with myelofibrosis display a significantly reduced T-cell response to ARG1-and PD-L1derived peptide epitopes as compared to patients with ET (12,36). This could reflect a dysregulation in the immune system in patients with advanced disease. In the current study, in line with our hypothesis, we observed an increase in specific T-cell responses against both the ARG1-and the PD-L1-derived epitopes in vaccinated patients as detected by ex vivo and in vitro IFN-g ELISPOT assays (Figures 2, 3). Both vaccines generated clear specific immune responses already after 3 vaccinations which appeared to be maintained until the end of the study. Interestingly, the majority of the vaccine specific T-cell responses were detected among the CD4 + T cells for both vaccine epitopes (Figures 3C, D This is in part due to the use of long peptide epitopes (38aa and 19aa) for the T cell stimulation., The processing of long peptides into HLA_I restricted epitopes requires uptake and intracellular processing of long peptides which may delay the presentation of class I epitopes in comparison to class II. As the standard duration of the ICS assay is only five hours, this might not be enough time for optimal processing and presentation of the HLA-I epitopes. A significant overall increase in the number of CD8 + T cells was seen in the peripheral blood during treatment. Among the CD8 + T cell subsets, the terminally differentiated T EMRA -cell population increased, and the CD8 + T EM -cell population significantly decreased suggesting that the vaccines boosted the CD8-mediated effector arm of the immune system.
During the treatment, T cell responses against both vaccination peptides were also observed among the infiltrating T cells in the DTH skin biopsies and bone marrow as detected by ex vivo and in vitro IFN-g ELISPOT assays. This shows that T cells reactive against ARG1-and PD-L1-derived epitopes were able to home to tissues with increased presentation of these epitopes. Interestingly, in addition to induction of vaccine specific T cell responses, we were also able to show a decrease in ARG1 and PD-L1 expression as detected by RT-qPCR in bone marrow and peripheral blood respectively. These findings indicate the immunomodulatory capacity of the ARG1 and PD-L1 specific T cells in vivo. In the present study the data could suggest, that ARG1 specific T cells home to the bone marrow and kill ARG1-expressing target cells. However, it is still unknown if transformed myeloid cells in the bone marrow produce more ARG1 than non-transformed myeloid cells. PD-L1specific CD8 + T cells can directly kill PD-L1-expressing cells (21).
However, it should be taken into account that PD-L1-specific CD4 + T cells may increase the fraction of PD-L1 + cells, due to the local production of pro-inflammatory cytokines (37,38).
We were unable to observe a clinical effect in the treated patients within the trial period. All patients except patient 5 had peripheral blood values that remained stable during treatment. Likewise, bone marrow histology and the JAK2V617F allele burden remained generally unchanged for seven out of nine patients. Two patients (#5 and #8) showed an increasing allele burden over the course of the treatment ( Figure 1A). The lack of clinical response could be explained by the low number of recruited patients for the trial, as we only included 9 of the intended 24 patients. In MPN the measured JAK2V617F mutational burden is very high which translates into a high tumor burden that might be impossible for the cellular immune system to control. It is noteworthy that patient 5 was identified as having MPN through participation in the population study termed the Danish General Suburban Study (GESUS) (39), In January 2018 the patient had a JAK2V617F allele burden of 4%, normal blood cells counts, as well as a normal bone marrow. Accordingly, the patient had JAK2V617F clonal hematopoiesis of indeterminate potential (CHIP). Nineteen months later, the patient was diagnosed with ET, and was in an early MPN-disease stage at the time of inclusion in the present study (after four additional months). Despite the early disease stage, the patient did not exhibit an ex vivo and only a limited in vitro ELISPOT response against either of the vaccine-derived epitopes during the treatment. The JAK2V617F allele burden in this patient increased from 12% to 17% during the trial, and platelet counts increased from 523 to 706 × 10 9 /l. These findings, including PD-L1 expression in the bone marrow (Supplementary Figure 8F), point towards a severe dysregulation of the immune system in this patient. We have earlier shown in another trial of therapeutic cancer vaccines in MPN, that patients with PMF show weaker immune responses to the vaccine compared to patients with ET (31). However, the patient mentioned above demonstrates that an exhausted immune response is not only restricted to patients with advanced MPN disease.
In future studies, it would be intriguing to vaccinate individuals with CHIP, as these patients have a very low tumor burden compared to patients with overt MPN. Otherwise, it will be necessary to combine the vaccines used in the present trial with other agents like an immune-checkpoint inhibitor. To date, treatments with the PD-1 checkpoint inhibitor, Pembrolizumab, have been unsuccessful in patients with MPN (40). However, a case study of a 71-year-old man with ET treated with the PD-1 checkpoint inhibitor, pembrolizumab, for a PD-L1-positive lung adenocarcinoma reported a decrease in elevated platelet counts and a dramatic decrease in the JAK2V617F allele burden after 17 months of pembrolizumab treatment (41). The immune-induced killing of adenocarcinoma cells in the patient may have re-activated the MPN-specific T-cell activity in the patient as the JAK2V617F mutation is also found in some carcinoma cells. Overall, in latestage MPN, a PD-1:PD-L1 blockade may be hampered by the general deregulation and exhaustion of the immune system (4), which is characterized by a down-regulation of HLA molecules and severe gene dysregulation of proteins involved in inflammation (42, 43). Thus, the increased frequency of vaccine specific T cells combined with the overall increase in CD8 + cells observed in this study might be enhanced by checkpoint blockade. Thus, this combination might evoke a stronger immune response against both tumor cells and regulatory cells. Our group has previously shown that both the JAK2V617F mutation and the CALR frameshift mutation lead to generation of immunogenic peptides, and patients with these mutations spontaneously harbored specific T-cell responses against these epitopes (44)(45)(46)(47). Combination therapy with IFN-a/a2 might also be an appealing option to further boost the immune activation. It should be noted that in our study five patients received IFN-a/a2 and no additional side effects were seen in this group, compared to patients not receiving IFN-a/a2.
In conclusion, we have tested a dual ARG1/PD-L1 peptide vaccination for the first time in humans. We have demonstrated that the combination of the two vaccines was safe and tolerable. We observed an induction or enhancement of vaccine-specific T-cell responses in all nine patients. Vaccine-specific T cells were also detected in the bone marrow after treatment and a decrease in ARG1 and PD-L1 expression was seen in bone marrow and peripheral blood respectively. Immune phenotyping of PBMCs showed a significant increase in the proportion of CD8 + T cells after the vaccination and an indication of activated CD8 + memory T cell subtypes. We did not observe an immediate effect on the JAK2V617F allele burden within the trial period. However, late onset disease regression as evidenced by complete remission after the finalization of the vaccination treatment was recently observed in two patients with follicular lymphoma treated in a similar trial with a dual PD-L1/PD-L2-derived peptide vaccine (27). Hence follow up of patients in the trial will be performed by annual measurements of peripheral blood counts and the JAK2V617F allelic burden, which will allow us to identify any late-onset responders among the vaccinated patients.

Data availability statement
The datasets presented in this article are not readily available because of the Danish Law on data protection and the GDPR rules.
Requests to access the datasets should be directed to the CCIT-DK office.

Ethics statement
The protocol was approved by the Ethics Committee of the Capital Region of Denmark, the National Board of Health, and the Danish Data Protection Agency, and it was registered at www.clinicaltrials.gov (NCT04051307). The patients/participants provided their written informed consent to participate in this study.

Author contributions
All authors contributed to the manuscript in regards to either conceptualizing and designing the trial or performing research or collecting, analyzing and interpretating data. All authors took part in writing or revision of the manuscript.

Funding
This work was supported by Copenhagen University Hospital, Herlev, Copenhagen University, Danish Cancer Society, grant number R149-A10159-B120 and Grants from the Danish Health Authority, Grant number: 05-0400-18 and 05-0400-50. The funders had no role in the study design, collection of data, data analysis, decision to publish or manuscript preparation.