Co-administered internalizing RGD peptide boosts anti-PD-L1 therapy in hepatocellular carcinoma

Abstract

Background & Aims

Immune checkpoint inhibitor (ICI) therapy has significantly improved the treatment of solid tumors such as hepatocellular carcinoma (HCC); however, most patients fail to respond. Here, we examined whether co-administration of the tumor-penetrating internalizing (i)RGD peptide, which selectively increases tumor vascular permeability in a neuropilin-1–dependent manner, enhances intratumoral delivery and therapeutic efficacy of αPD-L1 in mouse models of HCC.

Methods

αPD-L1, with or without iRGD, was administered intravenously to mice bearing endogenous HCCs (TGFα/c-myc and diethylnitrosamine [DEN]/carbon tetrachloride [CCl₄] models). Tumor growth was monitored by MRI. Immune cell composition and activation were analyzed by flow cytometry. RNA sequencing was performed on whole tumors and isolated intratumoral CD45⁺ immune cells.

Results

While αPD-L1 monotherapy had minimal impact on tumor progression, combination treatment with iRGD significantly improved therapeutic efficacy, resulting in markedly reduced tumor growth (mean difference −198.2%, p <0.0001 in TGFα/c-myc and −88.8%, p = 0.0159 in DEN/CCl₄-induced HCC mice) and increased objective response rates from 0 to 33% (90% confidence interval 14.6–58.6) in TGFα/c-myc mice and to 80% (90% confidence interval 39.6–95.8) in DEN/CCl₄-induced HCC mice. Flow cytometry revealed reduced PD-1^high CD8⁺ T cells and enhanced expression of activation markers (Ki67, CD44, IFN-γ) in the combination group. RNA sequencing of CD45⁺ cells and whole-tumor transcriptomes indicated decreased immunosuppression and increased vascular permeability in mice receiving the combination therapy. Immunoblot analysis showed enhanced accumulation of αPD-L1 in tumors following iRGD co-administration.

Conclusions

iRGD co-administration significantly improves the therapeutic efficacy of αPD-L1 in HCC mouse models by increasing intratumoral αPD-L1 delivery and more effectively alleviating the immunosuppressive tumor microenvironment. This non-conjugated, systemic approach holds strong translational potential to enhance ICI responses in patients with HCC.

Impact and implications

Immune checkpoint inhibitors (ICIs) show limited efficacy in hepatocellular carcinoma (HCC), to which poor intratumoral drug delivery is likely to contribute. This study demonstrates that intravenous co-administration of the tumor-penetrating peptide iRGD with anti–PD-L1 significantly improves antibody distribution and therapeutic response without increasing toxicity in preclinical HCC models. These findings are highly relevant for clinicians and researchers aiming to enhance ICI effectiveness in solid tumors. Given the ongoing clinical evaluation of iRGD, this simple, non-conjugated strategy offers a feasible and rapidly translatable approach to overcome a key limitation of current ICI therapy and improve outcomes for patients with HCC.

Graphical abstract

Highlights

• •Response rates to anti–PD-L1 immune checkpoint therapy for HCC remain limited.
• •Co-administration of iRGD markedly increases responses to anti–PD-L1 therapy in HCC mouse models.
• •iRGD enhances tumor-selective drug delivery, translating into improved immune checkpoint blockade efficacy.
• •Combination treatment leads to remodeling of the tumor microenvironment, including activation of CD8⁺ T cells.
• •iRGD co-administration shows potential to improve efficacy of anti–PD-L1 monotherapy for HCC without increasing side effects.

Introduction

The immunosuppressive tumor microenvironment (TME) plays a major role during hepatocarcinogenesis and tumor progression. Consequently, restoring antitumor immune responses through immune checkpoint blockade has become an important therapeutic strategy. In particular, programmed cell death protein 1 (PD-1) and programmed cell death ligand 1 (PD-L1) immune checkpoint inhibitors (ICIs) have provided significant survival benefits for patients with advanced cancers in recent years. However, objective response rates (ORRs) across many solid tumor types remain low, at approximately 20%.¹ Combination strategies involving multiple ICIs or ICIs with other treatment modalities can increase ORRs to ∼17–30%,[2], [3], [4], [5], [6], [7] but often at the cost of increased toxicity.

Hepatocellular carcinoma (HCC) is the most common histological subtype of liver cancer, the third-leading cause of cancer-related mortality,⁸ and shows rapidly rising incidence rates. In patients with non-resectable HCC, combinations of atezolizumab anti-PD-L1 (αPD-L1)/bevacizumab (anti-vascular endothelial growth factor, αVEGF) and durvalumab (αPD-L1)/tremelimumab (anti-cytotoxic T-lymphocyte-associated protein 4 [αCTLA-4]) have become the new first-line standard of care for advanced HCC, achieving ORRs of ∼20-30%.^2,3 In addition, durvalumab has been approved as first-line monotherapy for HCC in Europe and Japan for patients with contraindications to atezolizumab+bevacizumab.⁹

Poor bioavailability of ICIs and, consequently, the ineffective binding of antibodies to their targets within tumors are primary drivers of ICI resistance.¹⁰ In order to reactivate functionally exhausted anti-tumor CD8⁺ T cells within tumors, intravenously administered therapeutic antibodies such as αPD-L1 must extravasate from tumor vessels and penetrate the extravascular tumor tissue. High molecular weight compounds such as antibodies show poor vascular and tissue penetration, reducing their efficacy.^11,12 Moreover, a recent study comparing tumor accumulation of αPD-1 and αPD-L1 in mouse tumor models demonstrated that considerably higher doses of αPD-L1 are required to achieve intratumoral antibody concentrations and therapeutic effects comparable to those of αPD-1.¹³ Thus, the efficacy of αPD-L1–based ICI therapy appears to depend critically on achieving sufficiently high intratumoral antibody concentrations, and low intratumoral levels following αPD-L1 administration may contribute to poor clinical response rates.

A tumor-specific increase in the proportion of a drug with an intratumoral mode of action, such as αPD-L1, should improve its efficacy without increasing side effects. A number of studies in tumor-bearing mice (as well as the results from the first-in-man study in patients with pancreatic cancer) indicate that tumor-specific drug delivery and accordingly anti-tumor efficacy can be achieved by using a bifunctional nine-amino acid circular tumor-homing peptide termed iRGD (also termed certepetide).[14], [15], [16], [17], [18], [19] iRGD contains an RGD motif, an internal C- end Rule (CendR) motif (R/KXXR/K), and a proteolytic cleavage recognition site. Intravenously injected iRGD targets and specifically increases the permeability of the tumor vasculature by initially binding to αvβ3 and αvβ5 integrins that are selectively expressed on tumor blood vessels, via its RGD motif.¹⁸ It is then proteolytically cleaved to produce a peptide with a C-terminal exposed CendR motif. The cleaved peptide loses affinity for integrins and binds to and activates neuropilin-1 (NRP-1) in the tumor vasculature,¹⁸ a cell surface receptor that plays a fundamental role in mediating vascular permeability signals.²⁰ Various iRGD-mediated drug delivery strategies have been investigated to enhance intratumoral transport, including direct drug conjugation with iRGD and co-administration with unconjugated drugs.^14,15,18,19,[21], [22], [23] Among the two modes of tumor delivery, iRGD conjugates may show enhanced penetration of tissue barriers due to multivalent peptide presentation and receptor clustering, whereas the co-administration mode can simply be combined with standard-of-care drug therapy.^23,24

Intravenously co-administered iRGD is currently being evaluated in clinical trials in patients with other tumor entities, such as pancreatic and colorectal cancers, for its ability to enhance standard-of-care therapies. A first-in-human study in patients with pancreatic ductal adenocarcinoma demonstrated that iRGD is well tolerated and showed promising signs of improved therapeutic efficacy when combined with nab-paclitaxel and gemcitabine.²⁵

Improving the currently low response rates to ICI therapy is a major goal of contemporary oncological research. In mice with subcutaneous gastric cancer, intraperitoneal co-injection of iRGD with anti–PD-1 failed to effectively target tumors or demonstrate therapeutic efficacy, whereas an anti–PD-1 conjugate carrying two iRGD molecules was effective, acting as a T-cell engager rather than by reversing T-cell exhaustion.²² However, intraperitoneal administration may not be optimal for the co-administration mode of iRGD-mediated transport, as intraperitoneally delivered iRGD poorly targets subcutaneous tumors.²⁶ Therefore, in this study we investigated whether intravenous co-administration of iRGD could enhance tumor targeting and therapeutic efficacy of αPD-L1 therapy in two mouse models of HCC with endogenously formed tumors, as well as the underlying mechanisms.

Materials and methods

Animals

All animal experiments have been approved by the local Ethics Animal Review Broad Hessen under approval numbers FK/1139, FK/2007 and were carried out in accordance with recommendations of the Animal Protection Agency of the Federal State of Hessen (Regierungspräsidium Darmstadt, Germany). All the animals were fed and housed in groups of five animals per cage in the central animal research facility of the University Hospital Frankfurt. All mice used in the study were matched in age and exclusively male, because of their higher susceptibility to develop HCC compared to female mice.

Breeding of genetically modified mouse strains was carried out internally in the specific pathogen-free facility of the central animal research facility of University Hospital Frankfurt. Wild-type mice were purchased by Charles River Laboratories (Wilmington, MA, USA) and transferred to the experimental animal facility.

Generation of transgenic mice and visualization of HCC

Male transforming growth factor-α (TGFα)/c-myc mice were generated by crossing homozygous metallothionein/TGFα and albumin/c-myc mice on a CD13B6CBA background.^27,28 After weaning, hepatocarcinogenesis in mice was accelerated by adding ZnCl₂ to drinking water. The endogenously formed HCCs were detected and monitored by contrast-enhanced MRI.

DEN/CCl₄-induced HCC mouse model

Five-week-old male C57Bl/6J mice received a single injection of a low dose (1 mg/kg i.p.) of diethylnitrosamine (DEN) (Merck, Darmstadt, Germany), followed by the injection of carbon tetrachloride (CCl₄) (0.2 ml/kg, in an identical volume of corn oil, i.p.) from the age of 10 weeks twice per week for 14 weeks. The endogenously formed HCCs were detected and monitored by Gd-EOB-DTPA-enhanced MRI in a 3T MRI scanner.

Contrast-enhanced MRI

To diagnose endogenously formed HCC lesions in the liver, the animals were anesthetized and examined by gadoxetic acid (Gd-EOB-DTPA, Bayer, Leverkusen, Germany)-enhanced MRI.^28,29 For this purpose, anesthetized animals were injected with 150 μl of 10 μM Gd-EOB-DTPA contrast agent i.v. into the tail vein and subsequently placed in a hand wrist coil of a Siemens Magnetom Trio 3T MRI scanner (Siemens Medical Solution Health Service, Eschborn, Germany). Animals were examined with a T1 weighted 3D gradient echo sequence consisting of two localizer sequences followed by a 12-minute measurement sequence. For MRI operation the Siemens software Prisma_fit – NUMARIS/4, version syngo MR E11 was used. The diameters of the tumors were measured using the program Centricity RIS 4.1i Plus, version 4.1 (GE Healthcare, Little Chalfont, UK). Largest diameter (L) and width (W) determined by MRI were used to calculate the tumor volume (V_t) according to formula 1 below. ORRs for the different treatments were calculated using clinical RECIST criteria for each tumor: complete response = -100% tumor growth, partial response ≤- 30% tumor growth, stable disease >-30% tumor growth, progressive disease >25% tumor growth. Tumor burden per animal was calculated as the sum of all tumor volumes.

$$V_t=\frac{L\ast W^2\ast \pi }{6}$$ (1)

In vivo treatments of the mice with peptides and antibodies

Mice were enrolled following contrast-enhanced MRI once tumor diameters reached 0.5 cm after 18 weeks of ZnCl₂ supplementation in the drinking water. One day after the initial MRI, all animals were randomly assigned to treatment groups. Dosing volume was 150 μl per i.v. injection into the tail vein every 72 h. Tumor growth was monitored by contrast-enhanced MRI on day 12 and day 21 of the treatment period. Response to therapy was assessed using RECIST.^30,31 On day 22 all animals were subjected to terminal anesthesia and perfused via the left ventricle. Some animals were sacrificed after two antibody/peptide injections to dissect and dissociate tumors in order to analyze them by flow cytometry.

Statistical analysis

Data are presented as means ± SEM or means ± SD as indicated for each data set. Statistical analysis was performed using GraphPad Prism version 10.4.1. Comparison between two groups was performed by Mann-Whitney or unpaired two-tailed Student’s t test depending on the results for D'Agostino-Pearson omnibus normality test. One-way or two-way ANOVA was used for multiple comparisons as indicated. Differences in survival curves were analyzed using Log-rank (Mantel-Cox) test.

For analytical methods see the supplementary information.

Results

Co-administered iRGD augments the therapeutic efficacy of αPD-L1 in two mouse models with endogenously formed HCC

In order to investigate whether iRGD improves the therapeutic efficacy of αPD-L1 therapy in HCC, we examined the effect of iRGD on tumor progression in conditional double transgenic TGFα/c-myc mice with HCC (Fig. 1A). Upon induction of transgene expression by zinc in the drinking water, these mice develop HCCs, which were monitored by contrast-enhanced MRI. Mice with liver tumors with diameters of approximately 4-5 mm were randomly assigned to the treatment groups (Fig. 1A). All tumors were monitored individually in the follow-up MRI during the treatments. Intravenously injected αPD-L1 or iRGD alone had no significant effects on the progression of tumors in TGFα/c-myc mice (Fig. 1B and S1A-C). In contrast, combination therapy with iRGD and αPD-L1 effectively reduced tumor progression (Fig. 1B and S1D). Larger tumors at the beginning of the therapy showed less therapeutic response compared to smaller tumors (Fig. S1E). Longitudinal monitoring of individual tumors in the differently treated mice, along with quantification of relative tumor growth, revealed that approximately half of the tumors in the mice receiving the combination therapy decreased in size compared to baseline (Fig. S1D). This trend was also reflected in the relative change in tumor burden per animal within the combination-treated group (Fig. S1F). In contrast, all tumors from control-, αPD-L1-, iRGD-, or isotype antibody–treated HCC mice showed increased volumes at the end of treatment compared with their baseline tumor volumes (Fig. 1C, S1A-C and S2). Because tumor growth kinetics varied within individual animals, the tumor growth curves in the αPD-L1+iRGD treated mice (Fig. S1G) and the total tumor burden represented as the sum of all tumor volumes per animal (Fig. S1H), were also determined. Sixty percent of the animals in the combination therapy group exhibited a partial response or stable disease according to clinical RECIST criteria, resulting in an ORR of 33% (90% CI, 14.6–58.6), whereas none of the animals in the other treatment groups showed an objective response (Fig 1D). Animals treated with iRGD plus αPD-L1 showed improved survival compared to animals treated with αPD-L1 alone (Fig. 1E). iRGD did not cause changes in the body weight of the animals and there were no other signs of toxicity associated with the injection of iRGD per se (Fig. S1I). These data indicate that co-injection of iRGD considerably improved the therapeutic efficacy of αPD-L1 therapy in this HCC mouse model and increased the proportion of HCCs responding to ICI therapy.

Fig. 1.

Co-administration of iRGD improved the therapeutic response to αPD-L1 therapy in two mouse models with endogenous hepatocarcinogenesis.

(A) Design of the experiments in TGFα/c-myc mice. (B) iRGD improved the therapeutic efficacy of αPD-L1 therapy in TGFα/c-myc HCC mice. TGFα/c-myc mice with HCC were treated with isotype control antibody (10 mg/kg BW, n = 4), αPD-L1 (10 mg/kg BW, n = 8) or αPD-L1/iRGD (10 µmol/kg) combination therapy (n = 7) for 22 days. Tumor burden of the mice was determined by MRI at days 0, 12 and 22. Values are means ± SEM. Statistical analysis was performed using unpaired t-test. (C) Only αPD-L1/iRGD combination therapy significantly reduced the relative tumor growth in TGFα/c-myc HCC mice. Relative tumor growth of the HCCs in the mice treated with 0.9 % (v/v) NaCl (n = 8), isotype control antibody (n = 4), iRGD (n = 5), αPD-L1 (n = 8), αPD-L1/iRGD combination therapy (n = 7). The tumor burden at 22 days was related to the tumor size at day 0 for each tumor. p values were calculated using two-way ANOVA with Tukey’s multiple comparison test. ORR for the different treatments were calculated using clinical RECIST criteria (PR ≤-30% tumor growth). (E) Improved survival of TGFα/c-myc HCC mice treated with combination of αPD-L1/iRGD. Kaplan-Mayer diagram shows survival end point of reaching termination criteria (tumor volumes >100 mm³). p values were determined using the Log-rank (Mantel-Cox) test: αPD-L1 p = 0.0608; αPD-L1/iRGD p = 0.0080. (F) Design of the experiments in DEN/CCl₄-treated mice. (G-I) Co-administered iRGD improved the therapeutic efficacy of αPD-L1 therapy in DEN/CCl₄ HCC mice. DEN/CCl₄-treated mice with HCC were treated with isotype control antibody (n = 8), αPD-L1 (n = 9) or αPD-L1/iRGD combination therapy (n = 8) for 22 days. (G) Tumor burden during the treatments (control: n = 8, αPD-L1: n = 9, αPD-L1/iRGD: n = 8). Values are means ± SEM. Statistical analysis was performed using unpaired t-test. (H) and (I): Relative tumor growth after 12 (H) and 22 days (I) (control: n = 6, αPD-L1: n = 4, αPD-L1/iRGD: n = 4) of treatment. Data are displayed as mean ± SEM. p values were calculated using one-way ANOVA. (J, K): Animals treated with αPD-L1/iRGD combination therapy showed an elevated ORR, PR and SD after 12 days (J) and after 22 days (K) of treatment in comparison to αPD-L1 alone. αPD-L1, anti–programmed cell death ligand 1; BW, body weight; CCl₄, carbon tetrachloride; DEN, diethylnitrosamine; HCC, hepatocellular carcinoma; iRGD, internalizing RGD peptide; ORR, objective response rate; PR, partial response; SD, stable disease; TGFα, transforming growth factor alpha.

A second mouse model of HCC was used to confirm the treatment-boosting effect of co-administered iRGD on αPD-L1 therapy. We chose the DEN-CCl₄-HCC mouse model, in which DEN triggers hepatocarcinogenesis on the background of liver fibrosis, thus well reflecting human HCC, in which HCCs usually form in a chronically inflamed fibrotic/cirrhotic liver.[32], [33], [34] Mice with appropriate HCCs according to contrast-enhanced MRI were again randomly assigned to the treatment groups (Fig. 1F). In this HCC mouse model, intravenously injected αPD-L1 alone significantly reduced the progression of tumors (Fig. 1G and S1J, K), consistent with observations in patients with HCC. In line with the TGFα/c-myc HCC mice results, the combination therapy with iRGD plus αPD-L1 showed a stronger effect on tumor growth of these HCCs compared to the αPD-L1 monotherapy (Fig. 1G, and S1K,L). Quantification of relative tumor growth revealed that HCCs exhibited a stronger treatment response to combination therapy than to αPD-L1 monotherapy (Fig. 1H, I). When the clinical RECIST criteria were used to monitor the treatment response, the proportion of HCCs responding was significantly higher, with an ORR of 80% (90% CI 39.6-95.8) in the combination treated group (Fig. 1J, K). Co-injection of αPD-L1 with iRGD nevertheless further strongly augmented the therapeutic effect of αPD-L1. These data confirm that co-injection of iRGD substantially enhanced both the therapeutic efficacy of αPD-L1 monotherapy in HCC and the objective response.

Co-administered iRGD augments the delivery of antibodies selectively into HCCs in mice with endogenously formed HCC

iRGD is well documented to specifically increase tumor vascular permeability in diverse tumor models, including HCC,[15], [16], [17], [18]^,29 thereby facilitating the extravasation of intravenously co-injected substances – ranging from small molecules to antibodies – and resulting in substantially higher intratumoral concentrations.¹⁸ Therefore, enhanced intratumoral transport of intravenously injected αPD-L1 may explain the improved efficacy of immune checkpoint blockade and the higher response rates observed with combination therapy (αPD-L1 + iRGD) compared with αPD-L1 monotherapy in HCC mice. We examined the effect of co-injected iRGD on tumor accumulation of a biotinylated isotype control antibody that, in contrast to αPD-L1, should not bind to targets within HCCs. iRGD induced an approximately threefold increase in the accumulation of a biotinylated isotype antibody in HCCs compared with tumors from mice injected with the antibody alone (Fig. 2A,B), suggesting that this enhanced tumor accumulation was mediated by iRGD–facilitated transport into the tumors. In contrast, iRGD did not affect the levels of the antibody in the corresponding liver tissue (Fig. 2A), in agreement with tumor specificity of iRGD-induced transport. We also examined the intratumoral concentration of αPD-L1 in DEN/CCl₄ mice with HCCs co-injected with iRGD+αPD-L1 or αPD-L1 (Fig. 2C,D). iRGD led to an approximately twofold increase of the αPD-L1 concentration in the tumors from co-injected mice compared to animals that were treated with αPD-L1 only (Fig. 2C). The effects of iRGD on the tumor levels of the antibodies resembled the effect of iRGD on tumor accumulation of doxorubicin and Evans blue in this HCC mouse model.¹⁶

Fig. 2.

Western blot analysis of tissue homogenates of TGFα/c-myc mouse tumors reveals enhanced antibody import under iRGD co-administration.

(A) Tumor-bearing TGFα/c-myc mice were given two doses of 10 mg/kg BW biotinylated rat IgG2b, κ ± iRGD (4 µmol/kg) by i.v. injection 72 h apart. Afterwards, animals were sacrificed by terminal perfusion and tumors dissected. Values are means ± SEM. p values were calculated using unpaired t-test or Mann-Whitney test according to D'Agostino & Pearson normality test (control: n = 2, rat IgG2b, κ: n = 3, rat IgG2b, κ/iRGD: n = 3) (B) Tissue lysates were analyzed by immunoblotting for the presence of the biotinylated heavy chain of the administered antibody. Pyruvate carboxylase served as loading control. (C) Tumor-bearing DEN/CCl₄ mice were given two doses of 10 mg/kg BW αPD-L1 ± iRGD by i.v. injection 72 h apart. Values are means ± SEM. p values were calculated using unpaired t-test or Mann-Whitney test according to D'Agostino & Pearson normality test (control: n = 2, rat IgG2b, κ: n = 4, rat IgG2b, κ/iRGD: n = 5). (D) Tissue lysates of dissected tumors were again analyzed by SDS-PAGE. TPN signal was analyzed using ImageJ. αPD-L1, anti–programmed cell death ligand 1; BW, body weight; CCl₄, carbon tetrachloride; DEN, diethylnitrosamine; HCC, hepatocellular carcinoma; iRGD, internalizing RGD peptide; TGFα, transforming growth factor alpha; TPN, total protein normalization.

Combination therapy of HCC mice with iRGD and αPD-L1 effectively reinvigorated exhausted CD8⁺ T cells in the HCCs

ICIs mediate anti-cancer effects by mobilizing the host’s immune system, rescuing T cells from co-inhibitory signaling and reviving immune surveillance against cancer cells. To investigate whether iRGD co-administration with αPD-L1 elicited immune changes in the TME compared to αPD-L1 monotherapy, we examined the levels of immune cells in the TME of the HCCs from the differently treated mice at an early time point of treatment. To this end, TGFα/c-myc mice with radiologically visualized HCCs were recruited after the detection of liver tumors ≥0.5 cm in diameter by contrast-enhanced MRI (Fig. 3A). Treatment started 1 day after tumor detection and the mice received a second i.v. injection after 72 h. 24 h after the second injection, mice were sacrificed and tumors resected. Tumors were then dissociated and cell suspensions consecutively analyzed by flow cytometry (Fig. 3A,B). iRGD had no detectable effect on the composition of the different immune cell types in the HCCs (Fig. 3C-G). At this timepoint there was no change in the relative frequencies of memory T cells (Fig. 3G). Determination of the proportion of CD8⁺ T cells with high PD-1 expression (PD-1^hi) in the HCCs of the differently treated HCC mice revealed that the co-injection of iRGD with αPD-L1 clearly reduced the proportion of PD-1^hi CD8⁺ T cells in the HCCs as compared to the other treatment groups, whereas no changes were found between the other treatment groups and the control-treated HCC mice (Fig. 3H).

Fig. 3.

Characterization of the tumor immune microenvironment by flow cytometry of dissociated TGFα/c-myc liver tumors after 96 h of treatment with αPD-L1 ± iRGD reveals early efficacy of iRGD co-administration.

(A) Design of the experiments in TGFα/c-myc mice. (B) To differentiate between lymphoid (C, D) and myeloid (E) immune cells, flow cytometry data were gated on CD19 against CD11b. (C-E) Analyses of lymphocyte (C) and T cell (D) subtypes from HCCs in the different treatment groups to determine the effect of PD-L1 immune checkpoint blockade. Tregs (F) and (G) memory T cells were analyzed by gating of their respective surface markers. (H) PD-1^hi CD8 T cells were depleted under αPD-L1/iRGD combination therapy compared to monotherapy. Data are means ± SEM for (C-E) or median (min-max) (F–H). p values were calculated using (H) unpaired t-test or (F, G) Mann-Whitney test according to D'Agostino & Pearson normality test or (C, D, E) two-way ANOVA with Tukey’s multiple comparison test. (control: n = 6, isotype: n = 4, iRGD: n = 6, αPD-L1: n = 7, αPD-L1/iRGD: n = 7). αPD-L1, anti–programmed cell death ligand 1; HCC, hepatocellular carcinoma; iRGD, internalizing RGD peptide; TGFα, transforming growth factor alpha; Tregs, regulatory T cells.

To investigate potential changes occurring after prolonged treatment of the TGFα/c-myc HCC mice with αPD-L1 with and without iRGD in the TME of HCCs, we analyzed the immune cell composition in the HCCs after 22 days of treatment by flow cytometry (Fig. 4A). Treatment started 1 day after tumor detection and the mice were treated by i.v. injection every 72 h. After 22 days, the mice were sacrificed and tumors dissected. Tumors were then dissociated and single cell suspensions subsequently analyzed by flow cytometry (Fig. 4A). Overall, there were no major differences in the immune cell subtypes from the HCCs of the different treatment groups (Fig. 4B-F). Notably, there was an increased ratio of CD8⁺/regulatory T cells in the HCCs of mice treated with αPD-L1+iRGD compared to control-treated mice (Fig. 4G). Moreover, analyses of activation markers of the CD8⁺ T cells revealed an elevated proportion of CD8⁺ T cells with high expression of CD44, Ki-67, and interferon-γ (IFNγ) in the HCCs of the αPD-L1+iRGD-treated compared to αPD-L1 monotherapy-treated mice (Fig. 4J-L). iRGD treatment alone did not alter T cell phenotypes (Fig. S3). These data are consistent with increased CD8⁺ T cell activation in HCCs of combination-treated mice compared with αPD-L1 monotherapy–treated mice. However, no significant changes were found for the proportion of CD8⁺ T cells with high expression of CD69 and granzyme B between the tumors of αPD-L1+iRGD- and αPD-L1-treated animals (Fig. 4M, N).

Fig. 4.

Activation pattern of tumor-infiltrating lymphocytes in the tumors of TGFα/c-myc mice after 22 days of treatment with αPD-L1 ± iRGD.

(A) Experimental design. (B-E) αPD-L1 treatment did not alter the overall composition of CD45⁺ immune cell populations. (D) CD4 and CD8 T-cell populations showed no difference while only (E) dendritic cell frequency was increased under combination therapy. (F) Tregs were not significantly altered, but the ratio of Tregs to CD8 T cells per tumor (G) resulted in a distinct profile for αPD-L1/iRGD therapy. (H–I) Memory T-cell function was determined by CD62L and CD44 expression (H) but was not altered under combinational therapy (I). (J-N) The relative frequencies of CD44⁺ (J), IFNγ⁺ (K), Ki67⁺ (L), CD69⁺ (M) and GrB⁺ T-cells (N) as a percentage of CD8⁺ effector T-cells are displayed. Data are means ± SEM for (B, D, E) or median (min-max) (C, F–N). p values were calculated using (C) unpaired t-test or (I–N) Mann-Whitney test according to D'Agostino & Pearson normality test, (F, G) one-way ANOVA or (B, D, E) two-way ANOVA with Tukey’s multiple comparison test. (control: n = 4, isotype: n = 2, iRGD: n = 5, αPD-L1: n = 5, αPD-L1/iRGD: n = 4). αPD-L1, anti–programmed cell death ligand 1; GrB, granzyme B; IFNγ, interferon gamma; iRGD, internalizing RGD peptide; TGFα, transforming growth factor alpha; Tregs, regulatory T cells.

To provide more evidence that αPD-L1 treatment of the HCC mice ameliorates CD8⁺ T cell anti-tumor immunity in the TME when co-administered with iRGD, we examined the immune cell composition, T cell exhaustion and activation markers in the HCCs of DEN-CCl₄-treated mice with radiologically detected HCCs after 3 weeks of treatment (Fig. S4A). No differences were found with respect to the content of the different immune cell types in the HCCs of the differently treated HCC mice (Fig. S4B–D). CD8⁺ T cells with high expression of PD-1 were reduced (Fig. S4E), whereas CD8⁺ T cells with high expression of CD44 were elevated in the HCCs of mice treated with combination therapy compared to those receiving αPD-L1 monotherapy (Fig. S4F). Moreover, there was a trend towards elevated CD8⁺ T cells with high expression of CD69 in the HCCs of combination-treated mice (Fig. S4G). These data are compatible with an iRGD-induced improvement of αPD-L1-induced amelioration of immune checkpoint blockade.

αPD-L1+iRGD combination therapy induces a differential gene signature in CD45⁺ isolated cells regarding immune activity compared to the cells from HCCs of mice treated with αPD-L1 monotherapy

To further investigate the ability of co-administered iRGD to improve αPD-L1-induced immune activation in the TME, we isolated CD45⁺ cells from suspended HCC cells of the different mouse treatment groups at the end of therapy and analyzed the transcriptome data by gene set-enrichment analysis (GSEA) (Fig. 5A). Differential expression analysis identified CD44 and IFNγ as upregulated at the mRNA level under combination therapy (Fig. 5B), confirming the protein-level expression patterns observed in T cells by flow cytometry. Moreover, both treatment groups exhibited varying expression of immune-related genes that potentially contribute to increased anti-tumor efficacy. Treatment with iRGD alone did not modulate T-cell activation in flow cytometry experiments, suggesting that the observed effects on immune cells in combination-treated tumors rely on enhanced efficiency of checkpoint inhibition due to increased bioavailability within tumor tissue. αPD-L1+iRGD treatment included upregulation of CXCL10 and IL15 receptor (IL15ra) that indicate increased memory and cytotoxic T cell functions, respectively (Fig. 5B). IL10 and IL17a were significantly downregulated under iRGD+αPD-L1+iRGD therapy (Fig. 5B). Besides their described dual effect on immune function, they were mainly associated with poor prognosis in HCC, and downregulation of IL10 and IL17a could indicate a less immunosuppressive phenotype.³⁵ In contrast, IL12 and IL4 were significantly upregulated, potentially promoting effector T cell activity (Fig. 5B). The data also indicated downregulation of NRP-1 in the immune cells of HCC mice treated with αPD-L1+iRGD. As NRP-1 augments the immunosuppressive function of Tregs and CD14⁺ cells,³⁶ the downregulation of NRP-1 on immune cells is compatible with a recovery of immune responses. These results demonstrate a distinct immune cell profile in the TME of combination-treated HCC mice that may translate into improved efficacy of αPD-L1+iRGD therapy. However, the bulk RNA sequencing (RNAseq) approach does not provide information on cellular origin, limiting the interpretation of the immune modulation of all immune-related marker genes. Also, GSEA suggested T helper 17 (Th17) cell differentiation and C-type lectin signaling pathways to be significantly downregulated in the group of αPD-L1/iRGD-treated HCC mice (Fig. 5C-E). Taken together, these findings suggest that reduced immunosuppression resulting from interactions among immune cells, combined with enhanced memory (CD44^hi) and effector functions of T cells, may play a key role in the improved efficacy of immunotherapy combined with iRGD compared with αPD-L1 monotherapy.

Fig. 5.

Effect of iRGD co-administration on the gene expression pattern of isolated CD45⁺ tumor-infiltrating immune cells from tumors of TGFα/c-myc HCC mice.

(A) Experimental setup for mice treated for 22 days with i.v. injection of αPD-L1 ± iRGD every 72 h. GSEA was performed using RNA sequencing data of isolated CD45⁺ cells from dissociated TGFα/c-myc tumors. (B) Gene expression of relevant marker genes for immune modulation of αPD-L1 and αPD-L1/iRGD-treated mice (normalized expression per row) (C) and NES of gene set-enrichment analysis with most significant up-/downregulated KEGG pathways. (D, E) Enrichment plots for C-type lectin receptor signaling and Th17 differentiation pathways. αPD-L1, anti–programmed cell death ligand 1; GSEA, gene set-enrichment analysis; HCC, hepatocellular carcinoma; iRGD, internalizing RGD peptide; NES, normalized enrichment score; TGFα, transforming growth factor alpha; Th17, T helper 17.

Bulk RNAseq analysis of the ICI+iRGD-treated HCCs of the mice revealed an upregulation of immune-related pathways and a downregulation of adherence junction-, HCC-related pathways

To investigate whether iRGD co-administration was accompanied by an amelioration of the immune suppressive TME and its effects on the tumor, we performed bulk RNAseq analysis of the HCCs of these mice (Fig. 6A). GSEA identified gene clusters significantly altered between the HCCs of control- and iRGD+αPD-L1-treated mice. It revealed upregulation of hematopoietic cell lineage and other immune-related pathways, as well as oxidative phosphorylation in the HCCs of the combination-treated mice compared to control-treated mice. In contrast, PD-1/PD-L1 checkpoint pathways, endothelial cell adherence junctions, inflammation-related pathways, and HCC- and carcinoma-associated pathways were downregulated (Fig. 6B). These data are compatible with an upregulation of immune-related pathways and downregulation of immune checkpoint blockade and carcinoma-related pathways in HCCs of iRGD+αPD-L1-treated mice.

Fig. 6.

Identification of specific and biologically plausible iRGD-related changes of mRNA levels in the HCCs of TGFα/c-myc HCC mice shows downregulation of junction- and adhesion-related mRNAs.

(A) Experimental setup for the analysis of the samples from mice treated for 22 days with i.v. injection of αPD-L1 ± iRGD every 72 h. GSEA was performed using RNA sequencing data of snap frozen whole tumor tissue. (B) Top differentially regulated pathways of HCCs from the mice that received iRGD+αPD-L1 treatment (n = 3 per group), ranked according to the NES. (C) Enrichment curves for pathways related to tight junctions, adherence junctions and gap junctions with respective NES, p values and FDR. (D) Volcano plot for the regulation of cell adhesion-related genes comparing αPD-L1 and control whole tumor transcriptome data. (E) Volcano plot showing a trend toward downregulation of cell adhesion–related genes from the Jam, (proto)cadherin, and claudin families under αPD-L1/iRGD therapy (n = 3 per group). αPD-L1, anti–programmed cell death ligand 1; FDR, false discovery rate; GSEA, gene set-enrichment analysis; HCC, hepatocellular carcinoma; iRGD, internalizing RGD peptide; NES, normalized enrichment score; TGFα, transforming growth factor alpha.

Analysis of genes involved in cell–cell contacts suggested downregulation of mRNAs associated with tight junctions, adherens junctions, and gap junctions in HCCs of mice treated with iRGD + αPD-L1 compared with HCCs from vehicle-treated mice (Fig. 6C). Comparison of the genes related to cell-cell contacts in HCCs of mice treated with αPD-L1+iRGD alone revealed downregulation of these mRNAs (VE-cadherin/Cdh5, Jam, CD99, Pecam1, claudin, occludin) in the HCCs of mice treated with iRGD+αPD- L1, but not in HCCs of mice treated with αPD-L1 monotherapy (Fig. 6D, E, Fig. S5), findings compatible with iRGD-induced changes in endothelial cell-cell contacts.

Discussion

Improving the still unsatisfactory response rates of patients to immune checkpoint therapy remains a major challenge in oncology. In this study, we demonstrate that intravenous co-administration of the tumor-homing peptide iRGD significantly augments the therapeutic efficacy of αPD-L1 and increases the proportion of HCCs responding to immune checkpoint therapy in two models with endogenous hepatocarcinogenesis, different TME and baseline sensitivity to aPD-L1. Treatment with iRGD alone had no effects, indicating that iRGD is a therapeutic booster rather than per se possessing relevant anti-tumor properties. Co-administration of iRGD increased ORR to αPD-L1 therapy from 0% to 33% in TGFα/c-myc HCC mice and to 80% in the mice with DEN/CCl₄-induced HCCs, suggesting that co-administration of iRGD may greatly improve the efficacy of αPD-L1 therapy, while maintaining a tolerable side effect profile. The therapy-boosting effect of iRGD is most likely mediated by increased transport of αPD-L1 across the tumor vasculature into the TME, resulting in enhanced checkpoint blockade activity specifically within tumors.

Patients may not respond to ICIs for various reasons, most notably insufficient presentation of cancer antigens to cytotoxic T cells,[37], [38], [39] high expression of immune checkpoint molecules, an immunosuppressive TME,⁴⁰ and low intratumoral ICI levels.¹³ The present study indicates that insufficient tumor delivery and consequently low levels of the ICI antibody within the tumors upon drug administration is an important reason for the low efficacy of αPD-L1 ICI therapy in HCC, resulting in inadequate inhibition of immune checkpoint signaling in the TME. This conclusion is supported by findings that co-administered iRGD increased intratumoral αPD-L1 levels in HCC mice, which was accompanied by both enhanced therapeutic efficacy and improved response rates to immune checkpoint blockade. The apparent tumor-specific effect of iRGD in increasing intratumoral levels of the co-injected antibody is consistent with earlier studies demonstrating the tumor specificity of iRGD-mediated transport.^15,18,29 The administration of iRGD showed no detectable toxicity in the preclinical studies and in the first clinical study with iRGD in patients with pancreatic cancer.²⁵ Thus, co-administration of iRGD may offer the possibility of improving the efficacy of αPD-L1 therapy for advanced HCC without increasing side effects.

Several studies have explored iRGD-conjugated antibodies or nanoparticles to enhance tumor penetration, including a recent report using an intraperitoneally applied (iRGD)₂-anti–PD-1 conjugate that showed effective tumor penetration and strong anti-tumor effects in a mouse model with gastric carcinoma xenografts, whereas intraperitoneal co-administration of iRGD with anti-PD-1 had no effect.²² While conjugates/particles with multivalent presentation of iRGD may show enhanced cellular transcytosis and penetration of tissue barriers,^22,23 intraperitoneally injected iRGD may be ineffective to increase the delivery of cargo to subcutaneous tumors.²⁶ Thus, it is possible that the ability of iRGD to improve delivery of co-injected immune checkpoint inhibitor antibodies into vascularized tumors requires intravenous injection of the peptide. In the present study we employed a non-conjugated, intravenously co-administered iRGD strategy, which aligns with the clinical mode of checkpoint inhibitor administration and avoids creating new molecular entities that may require additional regulatory approval.

iRGD has been shown to induce transcellular transport across the tumor vascular endothelium via an NRP-1-dependent mechanism.^14,15,39 As NRP-1 activation also leads to loosening of VE-cadherin junctions,^41,42 iRGD may also induce paracellular transport across the tumor vascular endothelium. The results obtained from the RNAseq analysis of the HCCs of mice treated with iRGD are compatible with iRGD treatment of the HCC mice inducing downregulation of cell-cell contacts, including of VE-cadherin (Cdh5), i.e. that iRGD/NRP-1 induce loosening of adherence junctions (Fig. S5). As the harvesting of the HCCs from the treated mice was performed 24 h after the injection of the mice with iRGD, the iRGD-induced differences in the cell-cell contacts in the HCCs may have lasted much longer than the blood circulation half-life of iRGD. This would be compatible with the prolonged duration of the iRGD-induced vascular permeability in HCC described recently⁴³ and may explain the ability of iRGD to improve therapeutic efficacy of a drug with an intratumoral mechanism of action and slow extravasation/tumor penetration.

Immune checkpoint blockade leads to reactivation of anti-tumor CD8⁺ T cells, tumor cell killing, a proinflammatory TME, and recruitment of lymphocytes from the blood circulation into the tumors. In the present study, no significant changes in immune cell infiltration in the tumors were found between the differently treated HCC mice, despite the strong anti-tumor effects of αPD-L1+iRGD therapy, suggesting that the anti-tumor activity of αPD-L1+iRGD treatment was mainly due to re-activation of the CD8⁺ T cells that were already present in the tumors. We further found a distinct profile for various interleukin cytokines and their receptors in the transcriptomes of isolated CD45⁺ cells from αPD-L1+iRGD treated HCC mice, which was compatible with a less immunosuppressive TME. The RNAseq data suggested elevated levels of IFNγ, CXCL10 and IL15ra in the tumors of αPD-L1+iRGD-treated HCC mice. CXCL10 expression by macrophages has been identified as a potential promoter of antitumor immunity, as well as of IL-15–mediated cytotoxic T cell functions, thereby suppressing tumor progression.^44,45 Thus, these events may contribute to the increased anti-tumor effects of the αPD-L1+iRGD treatment compared to monotherapy. This differential activation pattern could potentially lead to delayed immune cell recruitment and infiltration, further fostering long-term anti-tumor immunity beyond the observation time of the present study. However, after two doses of treatment, combination therapy-treated mice already showed reduced frequencies of PD-1^hi expressing CD8⁺ T cells compared to HCC mice receiving αPD-L1 monotherapy, indicating an early onset of the effects of the combination therapy. After 22 days of treatment, an elevated proportion of T cell activation markers and a reduced proportion of Tregs were found in the T cell population in the HCCs of mice treated with αPD-L1+iRGD, suggesting that iRGD augmented the anti-tumor effect of αPD-L1 by reactivation of CD8⁺ T cells and reduced immunosuppression. The CD8 T cell memory marker CD44 involved in Th1–Th2 differentiation^46,47 was significantly upregulated on T cells, which potentially plays a crucial role in mediating anti-tumor immunity of combination therapy, since upregulated CD44 levels were observed in both mouse models upon αPD-L1+iRGD treatment. In contrast, Th17 differentiation was downregulated according to our RNAseq data in isolated CD45⁺ cells from HCCs of combination-treated mice. Th17 T cells and Il17a signaling were previously correlated with poor survival in patients with HCC.³⁵

iRGD may exert anti-tumor effects via additional mechanisms that escaped detection in our experiments. iRGD or the CendR peptide formed upon proteolytic cleavage of iRGD may activate NRP-1 in the extravascular tumor tissue. However, effects associated with NRP-1 activation in tumors, such as activation of immune suppressive Tregs or dendritic cells,⁴⁸ were not observed in the present study. In pancreatic cancer mouse models, iRGD may mediate anti-tumor activity through the depletion of Tregs.⁴⁹ Recent single-cell RNAseq data in human HCCs indicate that the majority of NRP-1 localizes to the tumor vascular cells,⁴⁴ which is in agreement with a major role of NRP-1/iRGD in the tumor vascular endothelium in HCC. In the present study we found reduced proportions of Tregs among total immune cells in the tumors of TGFα/c-myc HCC mice after 3 weeks of treatment with αPD-L1+iRGD. Moreover, there were no significant effects of iRGD monotherapy on tumor immune cells or tumor growth, favoring the idea that iRGD exerted anti-Treg effects indirectly by increasing efficacy of αPD-L1 in the TME of HCCs.

The pre-clinical work and the first clinical trial with iRGD in patients with pancreatic cancer suggest that iRGD is very well tolerated. There are also signs of clinical activity of iRGD in patients with pancreatic cancer, as co-administration of iRGD was found to improve response rates to standard-of-care therapy (gemcitabine, nab-paclitaxel) in these patients without increasing side effects or altering safety profiles.²⁵ It is important that the removal of immune checkpoint signaling occurs only in the tumors to avoid immune-related side effects.⁵⁰ A tumor-specific increase of the delivery of αPD-L1 may allow for a reduction of the administered dose of the antibody and thereby reduce side effects. The targeted structures of iRGD, αv integrins and NRP-1, are overexpressed mainly on endothelial cells and vessels of human tumors, including HCC.^44,[51], [52], [53] Our finding that iRGD considerably improved ORR of mice treated with αPD-L1 monotherapy without detectable side effects in two HCC mouse models with endogenous carcinogenesis, including one with hepatocarcinogenesis on the background of liver fibrosis, thus reflecting human HCC, suggests that iRGD may also work in patients. However, this needs to be tested in clinical trials in patients with HCC.

The toxicity profile of PD-1/PD-L1 monotherapy is more favorable than that of combination therapies and is associated with a better quality of life. Therefore, improving the therapeutic response to PD-1/PD-L1 axis blockade is a highly desirable goal. Co-administration of iRGD with PD-L1 might be highly beneficial, considering that a large proportion of patients with HCC are not eligible for the currently approved combination therapies, the signs of high tolerability of iRGD in the first-in-man study,²⁵ and the considerable improvement of αPD-L1 therapy responses in mouse models in the present study. Thus, a clinical trial testing the ability of iRGD co-administration to increase therapeutic efficacy/ORR of αPD-L1 (durvalumab) monotherapy in patients with advanced HCC is highly warranted.

While these findings are highly encouraging, a limitation of our study is that they are based on two murine HCC models. Although both the DEN/CCl₄ model with underlying fibrosis and the TGFα/c-myc transgenic model capture distinct aspects of human HCC pathogenesis, neither fully reproduces the complexity and heterogeneity of the human disease and the spectrum of ICI resistance mechanisms. Moreover, the mode of action of iRGD in tumors is not yet fully elucidated. The ongoing clinical trials with iRGD in patients with other tumor entities may help to obtain a better understanding of the underlying mode of action of iRGD in humans.

In conclusion, this study demonstrates that intravenous co-administration of iRGD significantly improves αPD-L1 response rate and efficacy by enhancing intratumoral delivery and thereby modulating the immune microenvironment in HCC. This clinically compatible, non-conjugated approach may represent a straightforward and powerful strategy to improve response rates to immune checkpoint therapy in patients with HCC and other solid tumors.

Abbreviations

αPD-L1, anti–PD-L1, CCl₄, carbon tetrachloride; DEN, diethylnitrosamine; GSEA, gene set-enrichment analysis; HCC, hepatocellular carcinoma; ICI, immune checkpoint inhibitor; IFNγ, interferon gamma; iRGD, internalizing RGD peptide; NRP-1, neuropilin 1; ORR, objective response rate, PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; RNAseq, RNA sequencing; TGFα, transforming growth factor alpha; TME, tumor microenvironment.

Authors’ contributions

JHK, LAB, DD and AP designed the study and experiments. JHK and LAB performed animal experiments. JHK, BA, LAB and LG conducted the ex vivo experiments and JHK, LAB and CDT analyzed the data. JHK and LAB had direct access to the mouse data. JHK, CDT and AP performed the statistical analysis. AP drafted the manuscript and wrote the paper, with important contributions of JHK, AW, FF, NMB, OW, TJV, BB and SZ. FF, AW and AP supervised the whole project. All authors contributed intellectual content, reviewed and edited the manuscript and agree to be accountable for all aspects of the work. All authors have read and approved the final version of the manuscript.

Data availability

Data are available upon reasonable request from the corresponding author.

Financial support

The authors acknowledge support by the German Consortium for Translational Cancer Research (DKTK) (to AP and SZ) and by the LOEWE Center Frankfurt Cancer Institute (FCI) funded by the Hessen State Ministry for Higher Education, Research and the Arts [III L 5 - 519/03/03.001 - (0015)] (to FF and AP), DFG ME 3621/2-1 (to NMB), bwHPC (Baden-Würtemberg) and the German Research Foundation (DFG) - grant INST 35/1597-1 FUGG (to CDT). Schematic illustrations created by Biorender.com.

Conflict of interest

(OW): Personal fees from Amgen, Bayer, BMS, Celgene, Daiicgi Sankyo, Eisai, Incyte, Ipsen, Merck, MSD, Novartis, Pierre Fabre, Roche, Servier; honoraria for lectures and/or presentations from Amgen, AstraZeneca, Bayer, BMS, Eisai, Ipsen, MSD, Novartis, Roche, Zentiva; support for attending meetings and/or travel: Abbvie, AstraZeneca, Bayer, BMS, Gilead, Ipsen, Medac, Merck, Pierre Fabre, Roche. (SZ): Consultancy and/or speaker's bureau: Abbvie, Boehringer Ingelheim, Gilead, GSK, Madrigal, MSD, Novo Nordisk, SoBi. (DD): Travel funding AbbVie. The other authors declare no competing interests.

Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the Supplementary data to this article: