Recombinant adenovirus expressing the fusion protein PD1PVR improves CD8⁺ T cell-mediated antitumor efficacy with long-term tumor-specific immune surveillance in hepatocellular carcinoma

Abstract

Purpose

Treatment-associated upregulation of suppressive checkpoints and a lack of costimulatory signals compromise the antitumor efficacy of oncolytic virus immunotherapy. Therefore, we aimed to identify highly effective therapeutic targets to provide a proof-of-principle for immune checkpoint together with oncolytic virus-mediated viro-immunotherapy for cancer.

Methods

A fusion protein containing both the extracellular domain of programmed death-1 (PD-1) and the poliovirus receptor (PVR) was designed. Next, the corresponding expression fragment was inserted into the genome of a replication-competent adenovirus to generate Ad5sPD1PVR. The infection, expression, replication and oncolysis of Ad5sPD1PVR were investigated in hepatocellular carcinoma (HCC) cell lines. Immune activation and the antitumor efficacy of Ad5sPD1PVR were examined in HCC tumor models including a humanized immunocompetent mouse model.

Results

Ad5sPD1PVR effectively infected and replicated in HCC cells and secreted sPD1PVR. In a H22 ascitic HCC mouse model, intraperitoneal injection of Ad5sPD1PVR markedly recruited lymphocytes and activated antitumor immune responses. Ad5sPD1PVR exerted a profound antitumor effect on ascitic HCC. Furthermore, we found that Ad5sPD1PVR-H expressing sPD1PVR of human origin exhibited potent antitumor effects in a HCC humanized mouse model. We also found that CD8⁺ T cells mediated the antitumor effects and long-term tumor-specific immune surveillance induced by Ad5sPD1PVR. Finally, when combined with fludarabine, the antitumor efficacy of Ad5sPD1PVR was found to be further improved in the ascitic HCC model.

Conclusions

From our data we conclude that the newly designed recombinant Ad5sPD1PVR virus significantly enhances CD8⁺ T cell-mediated antitumor efficacy with long-term tumor-specific immune surveillance in hepatocellular carcinoma, and that fludarabine is a promising therapeutic partner for Ad5sPD1PVR.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-021-00633-w.

Introduction

Hepatocellular carcinoma (HCC) accounts for 90% of primary liver cancers globally and is the second most common cause of cancer-related death [1, 2]. Although several treatments, such as surgery, radiotherapy and chemotherapy, have been developed, the prognosis of HCC remains unsatisfactory, primarily because of its high recurrence rate [1]. Recent studies have shown that immune checkpoint therapy, virotherapy and adoptive cell therapy exhibit good efficacies and have attracted attention to tumor immunotherapy, bringing hope to patients with HCC [3–5].

One way by which tumors can evade immune surveillance and clearance is through the formation of inhibitory tumor immune microenvironments (TMEs) [6]. In TMEs, checkpoint receptors function as inhibitory molecules that are upregulated after T cell activation, and tumor cells make full use of these molecules to suppress T cells, thereby inhibiting antitumor immune responses [7]. Blocking immune checkpoints has been hailed as a hope in curing cancer [8]. Clinical trials have indicated that blocking the immune checkpoint proteins PD-1/PD-L1, B7/CTLA-4 and PVR/TIGIT can effectively elicit antitumor immune responses and prolong the overall survival of patients [9, 10]. TIGIT (T cell immunoglobulin and immunoreceptor tyrosine based inhibitory motif [ITIM] domain) was found to function as a critical checkpoint molecule involved in immune negative regulation, and to be specifically upregulated on the surface of activated NK and T cells [11]. Several regulatory pathways mediated by TIGIT have been proposed: (1) binding to its cognate ligand PVR (CD155) expressed on tumor/APC cells upregulates IL-10 and downregulates IL-12 expression by tumor/APC cells leading to immunosuppression; (2) binding to costimulatory receptor CD226 disrupts the dimerization of CD226 resulting in immune suppression; (3) competitive binding to its cognate ligand PVR with higher affinity than the binding of CD226 with PVR leading to insufficient costimulatory signaling via the PVR/CD226 pathway [12]. Recently, it has been found that PD-1 and TIGIT expression is correlated in most tumor-specific CD8⁺ T cells and, thus, that blocking both PD-1 and TIGIT may enhance the antitumor immune responses of CD8⁺ T cells [13, 14]. Although immune checkpoint blockade is effective, it also has some drawbacks, including low universality [15, 16], off-target effects and accidental damage to normal tissues [8, 17]. At present, the total effective rate of clinical anti-PD-1/PD-L1 therapy does not exceed 30% [18] mainly due to lack of lymphocyte infiltration [6, 19]. In addition, anti-PD-1/PD-L1 drugs have shown considerable clinical side effects, leading to autoimmune tissue injury [20, 21]. Solving these two problems has become a major aim in tumor immunotherapy [22, 23].

Increasing evidence suggests that the induction of tumor-specific T cells is a prerequisite for immune checkpoint blockade [17, 24, 25]. In recent years, oncolytic viruses have attracted attention because they selectively infect tumor cells, thus inducing antitumor immune responses, and because of their flexibility for genetic modification. These advantages may address the deficiencies in immune checkpoint blockade therapy. Oncolytic virotherapy is used to induce immunogenic cell death (ICD) and type I interferon (IFN) signaling in tumor cells and to recruit more lymphocytes that infiltrate into the tumor environment and thereby turn “cold” tumors “hot”, ultimately enhancing antitumor immune responses [26]. In 2015, the US Food and Drug Administration (FDA) approved T-vec for the treatment of advanced metastatic melanoma [27]. Recently, oncolytic viruses encoding soluble proteins have been used to design combined therapies with oncolytic viruses and immune checkpoint blockers [28, 29], immune-stimulatory cytokines [30, 31] or bispecific T cell engagers [32, 33] to optimize ICD or enhance dendritic cell activation and cross-presentation [34] for the enhancement of antitumor immune responses. Thus, oncolytic viruses may serve as ideal engineering platforms for combination immunotherapy [35, 36].

An oncolytic adenovirus can selectively replicate in tumors through deletion of the E1B-55kD gene [37]. A genetically modified virus, known as H101, has been proven safe and effective in two clinical trials (ChiCTR-OPN-15006746 and ChiCTR-OPC-15006142) [37]. Previously, we constructed a replicative oncolytic type V adenovirus expressing the extracellular domain of PVR/CD155 (Ad5sPVR), which expressed a soluble domain with a much higher affinity to TIGIT than to CD226. Additionally, we found that Ad5sPVR significantly enhanced antitumor immune responses and led to a significantly better antitumor outcome [38].

Given that blocking both PD-1 and TIGIT can significantly improve the effectiveness of treatment [13, 14], we here designed and constructed a recombinant adenovirus expressing a fusion protein targeting both the PD-L1/PD-1 and TIGIT signaling pathways. This genetically engineered adenovirus expresses a soluble protein, sPD1PVR, which contains the extracellular domains of PD-1 and PVR and, as such, is anticipated to co-block the PD-L1/PD-1 and PVR/TIGIT signaling pathways to diminish T cell dysfunction and activate tumor-specific cytotoxic T lymphocytes (CTLs) through PVR/CD226 costimulatory signaling. We verified the antitumor effects and the underlying mechanisms of Ad5sPD1PVR in a H22 HCC murine model and in a humanized immunocompetent HCC mouse model. In addition, we combined Ad5sPD1PVR with fludarabine, a purine analog which has been shown to promote degradation of IDO1 and to reduce immune suppressive cells such as Tregs and MDSCs [39, 40], to improve therapeutic outcomes.

Materials and methods

Cell culture

The murine hepatocellular carcinoma (HCC) cell line H22 (CCTCC; ID: 3111C0001CCC000309, Wuhan, China) and human HCC cell line HCC-LM3 (CCTCC; ID: 3142C0001000000316, Wuhan, China) were purchased from the China Center for Type Culture Collection. The mouse HCC cell line Hepa1-6 (Cat# CRL-1830) and human embryonic kidney cell line HEK293 (Cat# CRL-1573) were purchased from the ATCC (American Type Culture Collection, Manassas, USA). H22 cells were cultured in Roswell Park Memorial Institute 1640 (RPMI-1640; Cat# 11,875,093, Gibco) medium and the other cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Cat# 11,965,092, Gibico-Thermo Fisher Scientific, USA) containing 10% fetal bovine serum (FBS; Cat# 16,000,044, Gibco), 100 units/ml penicillin and 100 μg/ml streptomycin (Thermo Fisher Scientific, Grand Island, NY, USA). All cells were kept in a humidified incubator with 5% CO₂ at 37 °C.

Construction of recombinant oncolytic adenovirus

The gene fragments sPD1, sPVR and sPD1PVR were designed as shown in Fig. 1B and cloned using a PCR method. Briefly, the sPVR and sPD1 fragments were obtained by PCR using mouse PVR (Cat# MG50259-M, Sino Biological, China) or PD-1 (Cat# MG50124-M, Sino Biological, China) cDNA clones as templates, whereas human sPVR and sPD1 fragments were obtained by PCR using human PVR (Cat# HG10109-CH, Sino Biological, China) and human PD-1 (Cat# HG10377-CH, Sino Biological, China) cDNA clones as templates. The mouse sPD1PVR fragment was cloned by fusing mouse PD-1 and PVR extracellular domains using (GGGGS)3 as a linker by PCR, whereas the human sPD1PVR fragment was cloned by fusing human PD-1 and human PVR extracellular domains and using (GGGGS)3 as a lisnker by PCR. Next, the target gene fragments were inserted in a pShuttle vector as previously described [36]. The recombined virus genomes were cloned by performing long-range (LR) reactions between the pAd/PL-DEST™ vector (V49420, Invitrogen, USA) and the recombined pShuttle vectors. The recombinant adenoviruses were rescued following the ViraPower™ Adenoviral Expression System manual protocol. Virus amplification was performed in HEK293 cells, and virus purification was performed using iodixanol (R714JV, Shanghai BasalMedia Technologies Co., China) gradient ultracentrifugation. Virus titration was performed by adding serially diluted viruses into a 96-well plate seeded with HEK293 cells (10,000 cells/well). The cells were cultured for 7 days, after which fluorescence was evaluated by microscopy. Virus titers were determined using the following formula: TCID50 = 102 + (S/N-0.5)/ml, pfu/ml = TCID50/ml × 0.7, where S represents the number of fluorescence-positive wells and N represents the number of replicates.

Fig. 1.

Generation of a novel recombinant adenovirus Ad5sPD1PVR, and replication and oncolytic capability of Ad5sPD1PVR in HCC cells. (A) Two schemes showing the hypothesized antitumor mechanisms of Ad5sPD1PVR: insufficient immune activation (left) and sustained immune activation with sPD1PVR (right). (B) Scheme showing the construction of the recombinant type V adenovirus expressing a soluble protein containing both the PD-1 and PVR extracellular domains. (C) H22 cells were infected with the recombinant oncolytic adenovirus at a multiplicity of infection (MOI) of 20 for 72 h. The supernatant was harvested, after which the expression of sPD1PVR was detected by Western blotting using an anti-His antibody. The molecular weight of sPD1PVR is estimated to be 100–110 kDa. (D) H22 and LM3 cells were infected with either Ad5sPD1PVR or the control virus Ad5con at MOIs of 20 and 2, respectively, after which the cells were harvested at various time points as indicated. Total DNA was extracted, and viral copies were detected by Q-PCR. The means ± SD of triplicates are shown. Similar results were obtained in two independent experiments. (E) H22 and LM3 cells were infected with either Ad5sPD1PVR or the control virus Ad5con at a series of MOIs as indicated, and 72 h later cell viabilities were measured using MTT assays. Similar results were obtained in three independent experiments. Significance was determined using one-way ANOVA with repeated measures (SPSS). Ad5sPD1, recombinant adenovirus encoding the extracellular domain of PD-1; Ad5sPVR, recombinant adenovirus encoding the the extracellular domain of PVR; Ad5sPD1PVR, recombinant adenovirus encoding fusion protein sPD1PVR; signal, signal peptide; EXO, extracellular domain; his, his tag; ITR, inverted terminal repeat; PA, poly (A)

Western blot analysis

Cells were seeded in 6-well plates at a density of 5 × 10⁵ cells per well and infected with the recombinant adenovirus (MOI = 20) for 72 h. Next, supernatants or cell lysates were harvested and protein concentrations were quantified. Subsequently, the samples were equally loaded and separated by SDS-PAGE, after which the proteins were electrophoretically transferred to polyvinylidene fluoride membranes (PVDF; Cat# K5MA6539B, Merck Millipore, MA, USA), followed by blocking with 5% nonfat milk for 1 h at room temperature. Then, the membranes were incubated with a specific primary mouse anti-his antibody (Cat# A00186, GenScript Biotech, Nanjing, China) at 4 ℃ overnight. Next, an appropriate secondary horseradish peroxidase (HRP)-conjugated rabbit anti-mouse IgG (Cat# A0216, Beyotime, China) was added for 1 h at room temperature. Finally, luminescence signals were determined using enhanced chemiluminescence reagent (Millipore, Billerica, MA, USA) in a chemiluminescence imaging system (ChampChemi 610, Sage Creation Science, Beijing, China).

Quantitative PCR analysis of virus copy numbers and mRNA expression

H22 and LM3 cells were seeded in 24-well plates at 10 × 10⁴ cells per well and kept in a humidified incubator with 5% CO₂ at 37 °C overnight. Next the cells were infected with Ad5con, Ad5sPD1, Ad5sPVR or Ad5sPD1PVR, respectively, after which the cells were harvested at serial time points as indicated (6, 24, 36, 48, 60 and 72 h after infection). Total DNA was extracted for viral copy number determination by Q-PCR using a TIANamp Genomic DNA Kit (Cat# DP304-03, TIANGEN, Beijing, China). Q-PCR was performed using a SYBR Green PCR Master Mix (Cat# 04,913,914,001, Roche, USA) on a ViiA 7 Real-Time PCR System (Applied Biosystems, Foster, CA, USA). Virus copy numbers were determined using a standard curve method.

Cell viability assays

Tumor cells were seeded in a 96-well plate at 1 × 10⁴ cells per well overnight, after which the cells were infected with Ad5con, Ad5sPD1, Ad5sPVR or Ad5sPD1PVR at a series of MOIs as indicated, for 72 h. Then, 100 μl (5 mg/ml) of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; Cat# IM0280, Solarbio, Beijing, China) was added to each well for 3 h at 37 °C after which the supernatants were removed and 100 μl isopropyl alcohol (Nanjing Chemical Reagent Co., China) was added for another 20 min with shaking to dissolve crystals. The absorbance (A) at 570 nm was measured using a multimode microplate reader (SMP500-13,497-JWYK, Molecular Devices, USA). Cell viability was calculated using the following formula: cell viability (%) = (A_treatment − A_blank)/(A_control − A_blank) × 100%.

Alternatively, tumor cells were seeded in a 96-well plate at 1 × 10⁴ cells per well and infected with adenoviruses at a series of MOIs as indicated, for 72 h. Then 10 μl CCK8 (Cat# C0037, Beyotime, China) was added to each well. 1 h later, the plate was subjected to a multimode microplate reader and the absorbance (A) was measured at 450 nm, after which cell viability was calculated similar to the MTT method.

FACS analysis

Either peripheral blood or ascites were harvested and washed with PBS with or without erythrocyte lysis. Then, the cells were incubated with anti-CD3-APC (Cat# 17–0,032,082, eBioscience, USA), anti-CD4-FITC (Cat# 557,307, BD, USA), anti-CD8a-PerCP-Cy5.5 (Cat# 45–0081-82, eBioscience, USA) or anti-NK1.1-FITC (Cat# 553,164, BD, USA) antibody for 30 min at room temperature. Next, the cells were subjected to FACS analysis using a FACSCalibur flow cytometer (BD Bioscience) and data were analyzed using FlowJo software (v.7.6.5, Tree Star, Ashland, OR, USA).

ELISpot assay

For in vivo immune activation assays, H22 cells were injected intraperitoneally (ip.) into male C57BL/6 mice (2 × 10⁶ cells/mouse). Then, the mice were treated with recombinant adenoviruses (5 × 10⁸ pfu each mouse) ip. on day 8 and 12 after H22 cell inoculation. Ascitic cells were harvested on day 16. The immune status of the tumor microenvironment (TME) was determined using an IFN-γ enzyme-linked immunosorbent spot (ELISpot) Assay Kit (Cat# 3321-2AW-Plus, Mabtech, NackaStrand, Sweden) according to the manufacturer’s instructions. Briefly, a 96-well plate was precoated with murine IFN-γ capture antibody overnight at 4 °C, after which 2 × 10⁴ ascitic cells were added into the wells (1.5 × 10⁵ cells/well) and incubated for another 24 h at 37 °C in a humidified incubator with 5% CO₂. Next, the cells were removed and the plate was washed with PBS three times, followed by incubation with a biotinylated anti-IFN-γ antibody for 2 h at room temperature. The plate was subsequently washed with PBS three times and incubated with streptavidin-ALP for 1 h, after which reaction substrates were added and, when spots emerged, the reaction was stopped with tap water. Spot numbers and sizes were determined using an ELISpot reader system (Autoimmune Diagnostika GmbH, Germany). The average spot size and intensity represents the immune activity.

ELISA

For in vivo immune activation experiments, male C57BL/6 mice were injected ip. with 5 × 10⁶ H22 cells and next treated ip. with recombinant adenoviruses, 5 × 10⁸ pfu/injection on day 8 and day 12, respectively. Mice treated with the same dose of saline were used as controls. Ascites were harvested on day 16 for further evaluation. The concentration of IFN-γ in the ascitic fluids was quantified using a mouse IFN-γ ELISA kit in accordance with the manufacturer’s instructions (BD Biosciences, Franklin Lakes, NJ, USA).

To quantify the amount of soluble protein, 100 μl supernatant from ascites was added into a microplate precoated with a His-tag antibody (A00186, GenScript Biotech Corp., Nanjing, China) and incubated for 2 h at 37 °C. Then, the plate was washed and further incubated with an anti-PD-1 (Cat# 50,124-R010, Sino Biological, Inc., Beijing) and anti-CD155 (Cat# 50,259-R001, Sino Biological, Inc., Beijing) antibody for another 2 h at 37 °C. Next, the plate was rinsed with PBS and incubated with HRP-conjugated streptavidin for 1 h. Finally, the substrate tetramethyl benzidine (TMB) was added and the absorbance was measured at 450 nm.

Mice and tumor inoculation

Six- to eight-weeks-old male C57BL/6 mice were purchased from Nanjing Biomedical Research Institute of Nanjing University (Nanjing, China) and maintained under specific pathogen-free (SPF) conditions (temperature 22 ± 2 °C, humidity 60 ± 10%, 12/12 h light/dark cycle) at the Medical School of Nanjing University. All animal experimental procedures used were approved by the Animal Care Committee of Nanjing University in accordance with the Institutional Animal Care and Use Committee guidelines. Mouse care and handling procedures were according to the NIH Guide for the Care and Use of Laboratory Animals. This project was approved by the Ethics Committee of the Medical School of Nanjing University.

To determine the antitumor efficacies of the recombinant viruses, 5 × 10⁶ H22 cells (diluted in 100 μl saline) were injectec intraperitoneally (ip.) into male C57BL/6 mice. Then, the mice received 5 × 10⁸ pfu of the recombinant adenovirus ip. on days 8, 12 and 16 after H22 cell inoculation. Mice treated with either the same dose of Ad5con or saline were used as controls. Ascites fluids were harvested on day 16 for further evaluation. On days 90 and 150 after treatment, cured mice were rechallenged with H22 cells (5 × 10⁶ cells/mouse). Naïve mice were used as negative controls. To deplete NK cells and CD8⁺ T cells, 500 μg anti-NK mAb (clone PK136, BE0036, Bio X Cell, USA) or anti-CD8α mAb (clone YTS 169.4, BE0017, Bio X Cell, USA) was ip. injected twice a week. NK cell or CD8⁺ T cell depletion in peripheral blood was evaluated using fluorescence-activated cell sorting (FACS) analysis. Mice were killed by cervical dislocation when their condition was determined to be moribund.

To determine the antitumor efficacy of the recombinant viruses in a human HCC mouse model, NOD-Prkdc^scid Il2rg^null (NCG) mice (from the Model Animal Research Institute of Nanjing University) were injected with 5 × 10⁶ LM3 cells (diluted in 100 μl saline) subcutaneously. Healthy human peripheral blood mononuclear cells (PBMCs) were separated with ficoll, after which they (2 × 10⁶ cells/mouse) were intravenously injected into the NCG mice 7 days after LM3 cell injection. Ad5sPD1PVR and Ad5con were intratumorally injected on days 8, 10, 12 and 15. Mouse body weights, tumor volumes and survival were determined. The tumor volumes were calculated as tumor length × tumor width² / 2. Mice were killed by cervical dislocation when the tumor volume reached 2 cm³, or when mice appeared moribund.

To detect inflammatory gene expression, 6 to 8-week-old male C57BL/6 J mice received intraperitoneal (ip.) injections of 2 × 10⁶ H22 cells to develop an ascites tumor model. H22-bearing mice were randomized into four groups on day 8. On days 8, 12 and 16, the mice received ip. injections of recombinant viruses, and saline was used as a control. On day 12, ascites were collected for inflammatory gene expression analysis. To this end, cellular RNA was extracted with TRIzol (Cat# 15,596–026, Thermo Fisher Scientific, USA) and reverse transcribed using a PrimeScript™ RT Master Mix (Cat# DRR036A, TaKaRa, Shiga, Japan). Then, SYBR Green was used according to the manufacturer’s protocol, and PCR was performed on a Real-Time PCR system (Viia7, Thermo Fisher Scientific, USA). Gene expression was calculated using the comparative threshold cycle (CT) method and normalized to GAPDH expression. The primer sequences used are provided in the supplementary information (Table S1).

Statistical analysis

Analysis of variance (ANOVA) was used to compare mean values among three or more groups. In ANOVA, we first obtain a common p value. A significant p value indicates that at least for one pair the mean difference was statistically significant. Post-hoc test (Tukey’s multiple comparisons) was subsequently used to identify significant pair(s). Repeated ANOVA measures were used to compare differences in tumor volume, virus titer and cell viability in different groups. Survival curves were estimated by the Kaplan–Meier method. The log-rank test was used to determine statistical differences between survival curves. All tests were carried out using Prism 6.0 (GraphPad, San Diego, USA), SPSS (Version 23.0, Chicago, USA) or R (Version 3.6.0, Auckland, NZ). A two-sided p-value < 0.05 was considered statistically significant for all analyses. The data are plotted as mean ± standard deviation (SD).

Results

Generation and in vitro identification of Ad5sPD1PVR

Upon MHC-I/TCR binding, PD-1 and TIGIT are often upregulated in activated T cells, leading to immune-suppressive feedback (Fig. 1A left panel). To block both the PD-1/PD-L1 and PVR/TIGIT immune-suppressive checkpoints, we designed a soluble fusion protein, sPD1PVR, containing both the PD-1 (sPD1) and PVR (sPVR) extracellular domains. sPD1PVR anticipatively blocks PD-1/PD-L1 and PVR/TIGIT immune-suppressive checkpoints through its sPD1 and sPVR termini, respectively, while providing sufficient ligands for CD226 costimulation (Fig. 1A right panel). sPD1PVR may act as a bridge between tumor and effector cells by binding to PD-L1 on malignant cells and to TIGIT and/or CD226 on effector lymphocytes, which improves lymphocyte cytotoxicity (Fig. 1A right panel). The fusion gene expressing sPD1PVR was inserted into a shuttle plasmid, followed by recombination processes to obtain Ad5sPD1PVR. Ad5 expressing sPD1 and/or GFP were used as experimental controls (Fig. 1B). We found that the recombinant virus restored the infectivity of malignant cells. By detecting the sPD1PVR fusion protein by Western blotting using an anti-his antibody we found that the fusion protein (with its appropriate molecular weight) was abundantly secreted from the infected tumor cells (Fig. 1C). Thus, the recombinant oncolytic adenovirus Ad5sPD1PVR was successfully constructed. Next, we explored the replication efficiency and oncolytic activity of Ad5sPD1PVR. In H22 and LM3 cells, Ad5sPD1PVR replicated as efficiently as the control viruses Ad5con, Ad5sPD1 and Ad5sPVR (Fig. 1D). Consistent with these results, we found that Ad5sPD1PVR exerted oncolytic activity similar to that of the control viruses in H22 and LM3 cells (Fig. 1E). These results suggest that the fusion protein did not alter viral replication and/or oncolysis.

Ad5sPD1PVR secretes soluble PD1PVR and increases lymphocyte infiltration and activation in an ascitic HCC model

To investigate the antitumor efficacy of Ad5sPD1PVR, we established a HCC model. The intervention regimen is depicted in Fig. 2A. While the survival rates of both the Ad5sPD1 group and the Ad5con group were similar to that of the saline group, the cure rate of the Ad5sPVR group was approximately 10% to 20%. The cure rate in mice treated with Ad5sPD1PVR was significantly improved (to > 40%) (Fig. 2B). During the course of therapy, viral replication and gene expression were monitored in parallel. We found that each recombinant virus replicated effectively and similarly in ascites (Fig. 2C), and that sPD1PVR, sPVR and sPD1 expression after three virus injections was easily detectable in the ascites at a mean of approximately 800, 700 and 900 pg/ml, respectively (Fig. 2D). Moreover, intraperitoneal injections of viruses resulted in increased CD8⁺ T cell infiltrations in ascites (Fig. 2E, right panels). Interestingly, we found that only ascites from mice treated with Ad5sPD1PVR and Ad5sPVR exhibited more CD4⁺ T cell and NK cell infiltrations (Fig. 2E, left and middle panels). Additionally, we determined the IFN-γ levels in ascites after viral injections using enzyme-linked immunosorbent assays (ELISAs). Ad5sPD1PVR treatment increased IFN-γ production in ascites (Fig. 2F). Finally, an ELISpot assay indicated that IFN-γ-producing lymphocytes were markedly increased in ascites of the Ad5sPD1PVR group (Fig. 2G). IFN-γ could be used as an index of immune activation. Taken together, these results indicate that Ad5sPD1PVR significantly enhances lymphocyte infiltration, immune activation and antitumor efficacy in an ascitic HCC model.

Fig. 2.

In vivo replication, sPD1PVR secretion, antitumor immune activation and lymphocyte infiltration capacity of Ad5sPD1PVR in a H22 ascitic HCC model. (A) Schematic diagram for recombinant adenovirus therapy. On day 0, male C57BL/6 mice were ip. injected with H22 cells (5 × 10⁶ cells/mouse), after which the mice were randomly divided into five groups: saline, Ad5con, Ad5sPD1, Ad5sPVR and Ad5sPD1PVR. On days 8, 12 and 16, mice in the saline group, Ad5con group, Ad5sPD1 group, Ad5sPVR group and Ad5sPD1PVR group were ip. injected with saline, the control virus, Ad5con (5 × 10⁸ pfu), Ad5sPD1 (5 × 10⁸ pfu), Ad5sPVR (5 × 10⁸ pfu) and Ad5sPD1PVR (5 × 10⁸ pfu), respectively. On day 16, before ip. injection of saline or virus, ascites fluid was collected. (B) Survival curves for mice treated with adenovirus. (C) Viral replication in ascites infected with adenovirus. Before ip. injection of saline or virus, ascites fluid was collected, and viral copy numbers were determined by Q-PCR on day 16. (D) The concentration of sPVR in mouse ascites was determined by ELISA. (E) CD4⁺ T, CD8⁺ T and NK cells in ascites were detected by FACS. (F) IFN-γ in ascites was detected by ELISA. (G) Immune activity in the TME was measured on an ELISpot, and the number of IFN-γ spots in each group was calculated. Significance was determined using one-way ANOVA with Tukey multiple comparison test. Data are shown as the mean ± SD. ns, not significant; * p < 0.05, ** p < 0.01, *** p < 0.001

CD8⁺ T cells mediate the antitumor activity of Ad5sPD1PVR and long-term tumor-specific immune surveillance in an ascitic HCC model

Having shown that Ad5sPD1PVR significantly increased both T cell and NK cell infiltration in ascites, we further clarified exactly which lymphocyte subtype participates in Ad5sPD1PVR-induced antitumor immune responses. To this end, NK cells or CD8⁺ T cells were depleted in mice via intraperitoneal injection of either anti-NK or anti-CD8 depletion antibodies. The specific depletion of CD8⁺ T cells achieved ranged from approximately 18.2% in control mice to approximately 2.28% (Fig. 3B upper panel). The specific depletion of NK cells achieved ranged from approximately 14.5% in control mice to approximately 1.36% (Fig. 3B lower panel). The antitumor effect of Ad5sPD1PVR was completely abrogated in mice with CD8⁺ T cell depletion, whereas the activity was retained or even mildly increased in mice with NK cell depletion (Fig. 3C). To exclude possible impacts of either CD8⁺ T cells or NK cells on viral propagation and/or sPD1PVR expression, which would alter the therapeutic effect, the viral copy number (Fig. 3D) and sPD1/PVR production (Fig. 3E) in ascites were dynamically monitored. We found that neither CD8⁺ T cell depletion nor NK cell depletion affected viral replication or sPD1PVR production in ascites. However, IFN-γ production was significantly reduced in CD8⁺ T cell-depleted but not NK cell-depleted ascites on day 12 (Fig. 3F), indicating that CD8⁺ T cells are the primary source of IFN-γ in ascites from mice treated with Ad5sPD1PVR. Then, the cured mice received two rounds of tumor rechallenge (Fig. 3G), and no ascites was observed, whereas all the naïve mice were susceptible to H22 challenge (Fig. 3H and I). To further elucidate whether Ad5sPD1PVR-induced antitumor immunity is tumor-specific, the cured mice were rechallenged subcutaneously with either the same H22 cells or with Hepa1-6 HCC cells (Fig. J). We found that Hepa1-6 cells rapidly developed solid tumors, whereas H22 cells did not grow in the cured mice (Fig. 3K and L). These data suggest that Ad5sPD1PVR induces CD8⁺ T cell-mediated tumor-specific activity with long-term immune surveillance in a HCC model.

Fig. 3.

Effect and mechanism of recombinant adenovirus Ad5sPD1PVR on the antitumor immune response in the H22 ascitic model. (A) Schematic diagram for recombinant adenovirus therapy. On day 0, male C57BL/6 mice were injected ip. with H22 cells (5 × 10⁶ cells/mouse), after which the mice were randomly divided into four groups: saline group, Ad5sPD1PVR group, Ad5sPD1PVR + anti-NK1.1 group and Ad5sPD1PVR + anti-CD8α group. On days 8, 12 and 16, mice in the saline group, Ad5sPD1PVR group, Ad5sPD1PVR + anti-NK1.1 group and Ad5sPD1PVR + anti-CD8α group were ip. injected with saline or Ad5sPD1PVR (5 × 10⁸ pfu). On days 8, 12 and 16, mice in the Ad5sPD1PVR + anti-NK1.1 and Ad5sPD1PVR + anti-CD8α groups were ip. injected with anti-CD8α and anti-NK1.1 (5 × 10⁸ pfu) twice a week. On days 12 and 16, before ip. injection of saline or virus, ascites fluid was collected for further study. (B) Mouse PBMCs were separated, and the depletion efficiency of the anti-NK1.1 and anti-CD8α antibodies was determined by flow cytometry 4 days after administration of the depletion antibodies. (C) Survival curves of mice from the Ad5sPD1PVR group with or without NK cell or CD8⁺ T cell depletion. Mice treated with PBS or Ad5con were used as controls. Ascites was harvested on day 12 and day 16. (D) Viral replication in ascites infected with adenovirus. Before ip. injection of saline or virus, ascites fluid was collected, and viral copy numbers were determined by RT-PCR on day 16. (E) The level of sPVR in ascites was determined by ELISA, and (F) the level of IFN-γ in ascites was determined by ELISA. Data are shown as the mean ± SD. (G) A schematic regimen showing rechallenge with H22 cells in cured mice. (H) The cured mice received the first round of H22 cells (5 × 10⁶ cells/mouse) on day 90 and (I) received the second round of H22 cells on day 150. Naïve mice received the same number of H22 cells and were used as controls. (J) The cured mice were subcutaneously rechallenged with either the same HCC cells (H22; 5 × 10⁶) or with other HCC cells (Hepa1-6; 5 × 10⁶). (K) Survival was analyzed by Kaplan–Meier curves and log-rank tests, and (L) tumor volumes were measured every three days. Significance was determined using one-way ANOVA with Tukey multiple comparison test. ns, not significant; * p < 0.05, ** p < 0.01, *** p < 0.001

Adenovirus expressing sPD1PVR of human origin attenuates tumor growth in a humanized in vivo HCC model

We further constructed a replication-competent recombinant adenovirus expressing sPD1PVR of human origin that contains the extracellular domains of human PD-1 and human PVR (Ad5sPD1PVR-H, Fig. 4A). The sPD1PVR-H protein was produced and secreted from cells infected with Ad5sPD1PVR-H (Fig. 4B). The oncolytic efficacy and viral replication of Ad5sPD1PVR-H were as effective as those of the control virus, indicating that sPD1PVR-H does not affect the oncolytic adenovirus (Fig. 4C and D). We then established a humanized subcutaneous LM3 HCC mouse model in NCG mice through intravenous perfusion of human PBMCs. The intervention regimen is shown in Fig. 4E. Compared to saline and Ad5con, Ad5sPD1PVR-H significantly reduced the tumor burden (Fig. 4F). No obvious therapy-associated side effects, indicated by body weight, were observed (Fig. 4G). These findings confirm that Ad5sPD1PVR of human origin may work efficiently in human cancer.

Fig. 4.

Antitumor effect of Ad5sPD1PVR-H in a subcutaneous LM3 humanized mouse HCC model. (A) Scheme showing the construction of a recombinant adenovirus expressing sPD1PVR of human origin, sPD1PVR-H. (B) LM3 cells were infected with oncolytic adenovirus at a MOI of 2 for 72 h, after which the level of sPD1PVR-H in the supernatant was determined via Western blotting. (C) LM3 cells were infected with the oncolytic adenovirus at a MOI of 2 for 72 h, after which cell viability was measured by MTT assay. (D) LM3 cells were infected with Ad5sPD1PVR-H at a MOI of 2 and harvested at the indicated time points. Adenovirus DNA was extracted, and viral copies were determined by Q-PCR. (E) Schematic diagram for recombinant adenovirus therapy in a solid tumor model. On day 0, NCG mice were subcutaneously inoculated with LM3 cells (5 × 10⁶ cells/mouse) in the right flank, after which the mice were randomly divided into two groups: Ad5con and Ad5sPD1PVR-H. On days 8, 10, 12 and 14, control virus Ad5con (5 × 10⁸ pfu) and Ad5sPD1PVR-H (5 × 10⁸ pfu) were intratumorally injected. On day 8, mice in the Ad5con and Ad5sPD1PVR-H groups were intravenously injected with PBMCs (2 × 10⁶ cells). (F and G) Tumor growth and body weights were measured. Significance was determined using one-way ANOVA with repeated measures (SPSS). Data are shown as the mean ± SD; ns, not significant; * p < 0.05, ** p < 0.01, *** p < 0.001. Ad5sPD1PVR-H, recombinant adenovirus encoding sPD1PVR of human origin; signal, signal peptide; EXO, extracellular domain; his, his tag; ITR, inverted terminal repeat; PA, poly (A)

Fludarabine enhances the antitumor efficacy of Ad5sPD1PVR by reducing protumoral inflammation and immune-suppressive factors in the ascitic HCC model

Although Ad5sPD1PVR achieved a cure rate of over 40% in ascitic HCC-bearing mice, we wondered why the other 60% of the mice responded only partially or not at all. In the H22 ascitic mouse model, we monitored tumor-associated inflammation and immunosuppressive factors in ascites during virotherapy (Fig. 5A). We found that Ad5sPD1PVR treatment significantly upregulated protumoral inflammatory cytokines, such as TNF-α, COX2 and IL-1β, and increased immunosuppressive factors, such as IL-10 and TGF-β (Fig. 5B and C). Interestingly, these virotherapy-upregulated cytokines were found to be suppressed by low-dose fludarabine, a purine analog (Fig. 5B and C), and two critical immune activation cytokines, IL-12 and IFN-γ, were found to be significantly upregulated by fludarabine in mice treated with Ad5sPD1PVR (Fig. 5D). Surprisingly, long-term survival was vigorously increased, from 40 to 90%, in ascitic HCC mice treated with Ad5sPD1PVR and fludarabine (Fig. 5E). Thus, Ad5sPD1PVR upregulates protumoral inflammatory and immunosuppressive cytokines, which can be effectively blocked by low-dose fludarabine, leading to enhanced antitumor responses.

Fig. 5.

Antitumor effect of Ad5sPD1PVR combined with fludarabine in the ascitic HCC model. (A) Schematic diagram for recombinant adenovirus therapy. On day 0, C57BL/6 mice were ip. injected with H22 cells (5 × 10⁶ cells/well) and randomly divided into four groups: saline, flu, Ad5sPD1PVR and flu + Ad5sPD1PVR. On days 8, 12 and 16, the mice were ip. injected with 5 × 10⁸ pfu of Ad5sPD1PVR with or without 0.75 mg/mouse fludarabine at the indicated time points. Control mice were ip. injected with an equal volume of saline. On day 12, before the ip. injection of saline or virus, ascites fluid was collected for further study. (B, C and D) Ascites cells were harvested on day 12, after which COX-2, IFN-β, IL-10, IL-1β, TGF-β, TNF-α and IL-12 mRNA levels were determined by Q-PCR. The IFN-γ concentration was determined by ELISA. (E) Survival was analyzed by Kaplan–Meier curves and log-rank (Mantel-Cox) tests. Significance was determined using one-way ANOVA with Tukey's multiple comparison test. Data are shown as the mean ± SD, ns, not significant; * p < 0.05, ** p < 0.01, *** p < 0.001

Discussion

In many cancer types, the immune-suppressive milieu exhibits a lack of sufficient lymphocyte infiltration and costimulatory signals, together with upregulated immune checkpoints. In this study, we constructed a self-replicative recombinant type V adenovirus expressing the soluble fusion protein sPD1PVR (Ad5sPD1PVR). We found that compared with Ad5sPVR or Ad5sPD1, Ad5sPD1PVR exhibited a significantly improved CD8⁺ T cell-mediated antitumor efficacy with long-term tumor-specific immune surveillance in hepatocellular carcinoma (HCC).

The most attractive characteristic of oncolytic viruses is that they turn tumors from “cold” to “hot”. In line with this, we found that Ad5sPD1PVR effectively promoted lymphocyte infiltration into the tumor microenvironment (TME). This function was similar among Ad5con, Ad5sPD1, Ad5sPVR and Ad5sPD1PVR, indicating that increased lymphocyte infiltration is induced by the virus and not by the fusion protein sPD1PVR.

While Ad5sPD1PVR and Ad5sPVR massively upregulated immune activation compared with Ad5con, there were no significant differences between Ad5sPD1 and Ad5con treatment, suggesting that a lack of costimulatory signals predominantly contributes to the immune-suppressive milieu. Consistent with these findings, except for one complete response, Ad5sPD1 treatment did not show any significant antitumor efficacy compared with Ad5con, suggesting that inhibition of a single checkpoint may be insufficient, even when combined with oncolytic virotherapy. However, Ad5sPD1PVR showed a more significant antitumor efficacy than Ad5sPVR, indicating that the inhibition of a checkpoint does help when sufficient costimulatory signals are provided. Similar to our previous study, we found that the recombinant adenovirus expressing the bispecific fusion protein sPD1CD137L (Ad5sPD1CD137L) also significantly enhanced antitumor immune responses compared with Ad5sPD1 or Ad5sCD137L. Both studies validate the necessity of providing costimulatory signals when inhibiting immune checkpoints for effective immunotherapy.

Although Ad5sPD1PVR treatment increased NK cell infiltration in the TME, we found that the antitumor effect of Ad5sPD1PVR was predominantly mediated by CD8⁺ T cells and not by NK cells. CD8⁺ T cell depletion resulted in a sharp decrease in IFN-γ production, suggesting that the primary source of IFN-γ in the TME is CD8⁺ T cells. We also observed tumor-specific and long-term immune surveillance in mice that completely responded to Ad5sPD1PVR, showing that Ad5sPD1PVR can effectively induce tumor-specific immune memory.

The systemic administration of immune checkpoint inhibitors often induces systemic effects and off-target cytokine storms, which lead to undesired damage in patients. Local administration of recombinant Ad5sPD1PVR confines the fusion protein sPD1PVR to the TME. sPD1PVR is anticipated to primarily be sequestered on tumor cells expressing PD-L1 via its PD-1 terminus. As a result, Ad5sPD1PVR may be capable of maintaining a high sPD1PVR concentration in the TME and of improving antitumor immune responses.

Interestingly, we found that Ad5sPD1PVR treatment upregulated the expression of certain inflammatory genes, such as TNF-α, COX-2 and IL-1β, which are pro-tumoral cytokines in HCC [39, 40]. In addition, certain immunosuppressive cytokines, such as IL-10 and TGF-β, were also found to be increased in mice treated with Ad5sPD1PVR. These therapy-upregulated inflammatory and immunosuppressive cytokines likely limited the complete response in half of the mice treated with Ad5sPD1PVR. Fludarabine is a purine analog used in the treatment of leukemia and lymphoma. Previous studies have shown that low-dose fludarabine cannot only down-regulate Tregs and MDSCs, but also promote the degradation of IDO1 by the proteasome [39, 40]. Previously, we showed that fludarabine significantly improved Newcastle disease virus (NDV)-mediated antitumor immune responses [40]. In the present study, fludarabine further enhanced the antitumor efficacty of Ad5sPD1PVR when the aforementioned inflammatory and immunosuppressive cytokines were blocked by low-dose fludarabine, and the cure rate was robustly increased from 50 to 90%. Our results underscore the point that fludarabine can be used as an adjuvant for oncolytic virotherapy [40]. Overall, protumoral inflammation and suppressive feedback in immune activation should be evaluated and considered in oncolytic virus immunotherapy. Combining low-dose fludarabine with an oncolytic virus is a potential therapeutic strategy for cancer patients.

In conclusion, we found that recombinant Ad5sPD1PVR can induce long-term and tumor-specific immune memory in mice. This finding may provide a new option for clinical treatment. Low-dose fludarabine may be helpful in facilitating Ad5sPD1PVR virotherapy.

Supplementary Information

Below is the link to the electronic supplementary material.

Author' contributions

J. WEI conceived and designed the study. J. WEI and J. WU supervised the project. H.Z., Y.Z. and J.D. performed the experiments and analyzed the data. H.Z., J. WU and J. WEI wrote the manuscript. All authors read and approved the final manuscript.

Funding

This work was funded by the National Natural Science Foundation of China (81773255, 81972888, 81700037 and 81472820) and the Primary Research & Development Plan of Jiangsu Province (BE2018701).

Availability of data and material

All data and materials generated during and/or analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no potential conflicts of interest.

Ethics approval and consent to participate

All animal experiments were conducted with approval of the Ethics Committee of Nanjing University Medical School. Animal welfare was closely monitored in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Consent for publication

No consent was involved in this publication.