Steatohepatitis Impairs T cell-directed Immunotherapies Against Liver Tumors in Mice (original) (raw)

. Author manuscript; available in PMC: 2022 Jan 1.

Published in final edited form as: Gastroenterology. 2020 Oct 1;160(1):331–345.e6. doi: 10.1053/j.gastro.2020.09.031

Abstract

Background & Aims:

Non-alcoholic steatohepatitis causes loss of hepatic CD4+ T cells and promotes tumor growth. The liver is the most common site of distant metastases from a variety of malignancies, many of which respond to immunotherapy. We investigated the effects of steatohepatitis on the efficacy of immunotherapeutic agents against liver tumors in mice.

Methods:

Steatohepatitis was induced by feeding C57BL/6NCrl or BALB/c AnNCr mice a methionine and choline-deficient diet (MCD) or a choline-deficient L-amino acid-defined diet (CDAA). Mice were given intrahepatic or subcutaneous injections of B16 melanoma and CT26 colon cancer cells, followed by given intravenous injections of mRNA vaccine (M30) or intraperitoneal injections of an antibody against OX40 (aOX40) on days 3, 7, and 10 after injection of the tumor cells. We measured tumor growth and analyzed immune cells in tumor tissues by flow cytometry. Mice were given N-acetylcysteine (NAC) to prevent loss of CD4+ T cells from liver.

Results:

Administration of M30 and aOX40 inhibited growth of tumors from intrahepatic injections of B16 or CT26 cells in mice on regular diet. However, M30 and/or aOX40 did not slow growth of liver tumors from B16 or CT26 cells in mice with diet-induced steatohepatitis (MCD or CDAA). Steatohepatitis did not affect the ability of M30 to slow growth of subcutaneous B16 tumors. In mice with steatohepatitis given NAC, which prevents loss of CD4+ T cells, M30 and aOX40 were able slow growth of hepatic tumors. Flow cytometry analysis of liver tumors revealed reduced CD4+ T cells and effector memory cells in mice with vs without steatohepatitis

Conclusions:

Steatohepatitis reduces the abilities of immunotherapeutic agents such as M30 and aOX40 to inhibit tumor liver growth by reducing tumor infiltration by CD4+ T cells and effector memory cells. NAC restores T-cell numbers in tumors and increases the ability of M30 and aOX40 to slow tumor growth in mice.

Keywords: NAFLD, NASH, metastasis, anti-tumor immune response

Graphical Abstract

graphic file with name nihms-1633962-f0001.jpg

Introduction

Non-alcoholic fatty liver disease (NAFLD), ranging from simple steatosis to nonalcoholic steatohepatitis (NASH), affects nearly one in four adults worldwide.1, 2 In the United States, the incidence of NAFLD is projected to increase by over 20% by 2030.3 NASH, a serious form of NAFLD, is associated with hepatic inflammation and progression to fibrosis and cirrhosis.4 NAFLD patients carry a high risk for developing primary liver malignancies such as hepatocellular carcinoma (HCC). In addition, the liver represents the most common site for distant metastasis, which represent nearly 95% of all liver tumors.5 While immunotherapy has shown promise for some liver malignancies, the hepatic tumor microenvironment (TME) may affect response to immune checkpoint inhibitor therapy.6 The role of NAFLD on anti-tumor immunotherapy efficacy remains unclear.

We previously reported that NAFLD/NASH promotes reactive oxygen species (ROS)-dependent cell death of hepatic CD4+ T cells, which impairs anti-tumor surveillance and promotes HCC.7, 8 CD4+ T cells are essential to tumor control by preventing tumor initiation and mediating clearance of premalignant and tumor cells.9, 10 Importantly, adoptive transfer of tumor specific CD4+ T cells caused complete tumor eradication in a patient bearing cholangiocarcinoma.11 Network analysis identified CD4+ T cells as initiators of anti-tumor response and linked CD4+ T cell subsets to favorable responses to immunotherapy.12 These reports suggest that NAFLD-associated CD4+ T cell loss may impair immunotherapy targeting liver tumors.

Previously, we studied an RNA-based vaccine to treat patients with malignant melanoma.13–15 This vaccine encodes tumor specific neo-epitopes that are part of the patient’s tumor “mutanome”.15 Mouse and human studies demonstrated that the majority of the neo-epitopes were recognized by CD4+ T cells.15, 16 B16-M30 RNA encodes for a highly immunogenic neo-epitope of the B16 murine melanoma cell line and induces strong CD4+ T cell-dependent antitumor response.14

In this study, we tested the B16-M30 vaccination (M30) in mice bearing intrahepatic B16 tumors. We investigated the effects of two models of diet-induced steatohepatitis on anti-tumor efficacy. Administration of M30 impeded liver tumor growth in healthy livers but not under steatohepatitis conditions. These findings were liver-specific, because the ability of M30 to slow subcutaneous B16 tumors.

To confirm our findings, we studied the effect of an OX40 receptor agonistic antibody, known to cause stimulation and proliferation of T cells on CT26 colorectal cancer (CRC) liver tumors.17 Anti-OX40 (aOX40) reduced tumor burden of CRC liver tumors in healthy mouse liver environment but had minimal effects in mice with steatohepatitis. Administration of N-Acetylcysteine (NAC) restored anti-tumor efficacy of both M30 and aOX40 in mice with diet-induced steatohepatitis. Our study demonstrates that steatohepatitis impairs anti-tumor efficacy of immunotherapy against liver tumors.

Materials and Methods

Mice and cell lines

6–10 weeks old C57BL/6NCrl or BALB/c AnNCr (Charles River Wilmington, USA) or B6(Cg)-Tyr/J mice (The Jackson Laboratory, Bar Harbor, USA18) were utilized for mouse experiments. Steatohepatitis was induced by feeding mice a methionine–choline-deficient (MCD) diet (Research Diets, New Brunswick, NJ, USA) or a choline-deficient L-amino acid defined (CDAA) diet (Dyets, Bethlehem, PA, USA). B16F10-luciferase-expressing melanoma cell line (B16F10-Luc) was described previously.14 Cells were cultured in RPMI1640 GlutaMax (+10% FCS + 1% So-Pyruvate +1% 1M HEPES +1% NEAA, all ThermoFisher, Grand Island, NY, USA). CT26 colon carcinoma cells were obtained from ATCC and cultured as recommended.

Synthetic RNA

In vitro transcribed RNA and Liposomes13 were provided by BioNTech RNA Pharmaceuticals GmbH (Mainz, Germany). RNA encoding the MHC class II restricted neoantigen B16-M3019 has been described previously.14 In brief, RNA was coding for 27 amino acids of the murine Kif18b protein harboring the K739N alteration in the center (nucleotide sequence: CCAGCAAGCCCAGCTTCCAGGAATTCGTCGACTGGGAGAAC GTGTCCCCCGAGCTGAACTCTACCGACCAGCCCTTCCTG). The antigen sequence was flanked by an MHC class I derived signal peptide (SP), glycine serine linker (L) and the MHC class I trafficking domain (MITD) in the order 3’ SP-L-B16-M30-L-MITD 5’ as described by Kreiter et al.20 The control was RNA encoding for the backbone only: 3’ SP-L-L-MITD 5’.

In vivo antibodies

Agonistic _InVivo_MAb aOX40 antibody (CD134, clone OX-86) or _InVivo_MAb rat IgG1 isotype control, anti-horseradish peroxidase (clone HRPN) was obtained from BioXCell (West Lebanon, NH, USA). For the depletion experiments, 100 μg of anti-CD4 antibody (clone GK1.5, BioXCell, West Lebanon, LH, USA) or anti-CD8 antibody (clone 2.43, BioXCell) or InVivoMAb rat IgG2b isotype control, anti-keyhole limpet hemocyanin (clone LTF-2, BioXCell) was administered one week before tumor inoculation and then once weekly to achieve depletion of CD4+ or CD8+ T cells. Successful eradication was confirmed by flow cytometry.

Animal studies

Age and gender matched mice were fed regular chow or MCD/CDAA diet started six to seven weeks of age and kept on specific diet until sacrifice. 10mg/ml NAC (Sigma Aldrich, St. Louis, MO, USA) per drinking water was added where indicated. Intrahepatic tumor injections were performed 2 weeks after start of the MCD diet or 22 weeks after start of the CDAA diet. A mouse model using intrahepatic injection has been described in detail elsewhere.21 Briefly, single cell suspensions of murine tumor cells were prepared. For intrahepatic tumors, 20 μL of 2.5×105 tumor cells were mixed with Matrigel (Corning, NY, USA) and injected into the liver. For RNA-vaccine M30 studies, mice were injected intravenously (i.v.) with 40μg RNA per mouse. 100ug aOx40 was injected intraperitoneally (i.p.) on day 3, 7 and 10 after intrahepatic injection. Bioluminescence imaging was used to monitor intrahepatic tumor growth of B16F10 luciferase expressing tumors. To reduce background signal, only white fur albino strain B6(Cg)-Tyr/J mice were used for intrahepatic injection experiments.22 In vivo bioluminescence using Xenogen’s IVIS imaging system was performed as previously described23. Subcutaneous injections were performed one week after the start of MCD diet administration. To establish subcutaneous tumors, 1×106 B16F10Luc melanoma or CT26 colon carcinoma cells were re-suspended in PBS and tumor cell solution was injected in the left lateral flank. Tumor size was measured by caliper and tumor volume was calculated by formula (width × width × length)/2. Mice with similar tumor volume of approximately 100mm3 were randomized after 1 week and i.v. injection with M30 RNA vaccine (or control RNA) or i.p. injection of aOX40 (or IgG control), respectively, was started. Subcutaneous tumors were blindly measured by caliper every 2–3 days. All experiments were conducted according to local institutional guidelines and approved by the Animal Care and Use Committee of the National Institutes of Health, Bethesda, USA.

Isolation of liver mononuclear cells (MNCs) and tumor infiltrating lymphocytes (TIL)

Isolation of MNCs and TILs from tumor bearing liver has been previously described.24

Flow cytometric analysis

Cells were surface labelled with the indicated antibodies for 15min at 4°C. Foxp3/transcription factor staining buffer set (eBioscience) was used for intracellular staining according to the manufacturer’s instructions. Flow cytometry was performed on a BD LSRFortessa or a Beckman Coulter Cytoflex LX platform and results were analyzed using FlowJo software version 10.4.2 (TreeStar). Dead cells were excluded by using live/dead fixable near-IR dead cell staining kit (ThermoFisher scientific). The following antibodies were used for flow cytometry analysis: anti-TCRβ-BV510 (clone H57–587, Biolegend), anti-CD3-PE (clone 17A2, Biolegend), anti-CD4-PE (clone RM4–5, Biolegend), anti-CD4-Alexa Fluor 700 (clone GK1.5 or clone RM4–5 for depletion experiments, Biolegend), anti-CD8-BV510 (clone 53–6.7 Biolegend), anti-CD8-BV421 (clone 53–6.7 for depletion experiments Biolegend), anti-CD11b-BV421 (clone M1/70, Biolegend), anti-Ly6G-Alexa Fluor 700 (clone 1A8, Biolegend), anti-Ly6C-APC (clone HK1.4, Biolegend),), anti-CD44-PE/Cy7 (clone 1M7, Biolegend) and anti-CD62L PerCP/Cy5 (clone MEL-14, Biolegend). The following markers were used for identifying different immune cell subsets: CD3+CD4hi for hepatic CD4+ T cells, CD3+CD8+ for CD8+ T cells. CD3+CD4+(or CD8+) CD44+CD62L− for effector memory T cells. CD11b+Gr1+ for MDSCs, CD11b+Ly6G−Ly6Chi for M-MDSCs, CD11b+Ly6G+Ly6Clo for G-MDSCs, CD11b+F4/80+ for macrophages. Data are reported as frequency of live lymphocytes or cells per gram tissue, calculated from frequency of lymphocytes multiplied by absolute count and divided by weight of tissue.

Statistics

Means were compared by using Student’s t-test or one-Way ANOVA for hypothesis testing to compare individual or corresponding groups (Tukey’s multiple comparison test). Mann–Whitney U or Kruskall-Wallis test (Dunn’s multiple comparison test) were applied if data sets failed the Pearson omnibus normality test (or Shapiro-Wilk normality test if N is too small) (alpha = 0.05). Error bars reflect standard error of the mean SEM. All analyses were two-tailed and carried out using GraphPad Prism 8.1.2. not significant (ns): P>0.05, *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Results

M30 RNA vaccine controls tumor growth of intrahepatic B16 melanoma

B16 melanoma liver tumors were established by intrahepatic injection of B16-F10-Luc tumor cells. Tumor-bearing mice were injected with M30 RNA vaccine (M30) or nonimmunogenic RNA as a control (Fig. 1A). B16-F10-Luc cells express luciferase enabling us to monitor intrahepatic tumor growth using non-invasive bioluminescence imaging. M30 injected mice demonstrated slow and stable tumor growth while in control RNA injected mice tumor growth increased rapidly (Fig. 1B). Significantly lower tumor burden in the vaccine group was observed from day 15 on (Fig. 1C). Consistent with bioluminescence imaging, a significant reduction of hepatic tumor burden was found in M30 injected mice at the experimental endpoint (Fig. 1D & E). These findings were independent of gender (Fig. S1A & S1B). Our results confirm that M30 effectively controls tumor growth in healthy livers.

Figure 1: M30-RNA vaccine controls tumor growth of B16 melanoma liver tumors.

Figure 1:

A. Experimental setup: Mouse model of B6(Cg)-Tyrc-2J/J albino mice with intrahepatic injection of B16 luciferase-expressing melanoma cells and RNA vaccination (at indicated time points).

B. Tumor growth of intrahepatic B16 tumors using bioluminescence imaging. Representative images of 4 mice per group are shown. 2 groups: control (n=7), M30 RNA-vaccination (n=8), both groups regular diet.

C. Quantification of in vitro bioluminescence assay displayed as tumor growth curve.

D. Tumor burden (same groups as in B) at d27 presented as tumor-to-liver weight-ratio.

E. Representative photos of intrahepatic melanoma tumors. 2 groups: control (upper panel), M30 RNA-vaccination (lower panel), both reg diet.

Steatohepatitis impairs efficacy of M30 RNA-vaccine against intrahepatic B16, but not extrahepatic tumors.

Next, the efficacy of M30 against liver tumors was tested in mice with MCD diet-induced steatohepatitis (Fig. 2A). In line with the previous experiment, tumor growth kinetics showed that M30 administration effectively controlled tumor growth in regular diet fed mice (Fig. 2B & 2C). However, mice with steatohepatitis showed accelerated tumor growth compared to mice on regular diet. Over the course of M30 administration, an increase in tumor growth was found in MCD diet fed mice despite injections with M30 vaccine, demonstrating that the vaccine loses efficacy in MCD diet-induced steatohepatitis (Fig. 2B & 2C). This phenomenon was also observed at the endpoint of the experiment, where there were no significant differences found in the tumor burden (Fig 2D & 2E). These results indicate that M30 fails to control tumor growth in diet-induced steatohepatitis livers.

Figure 2: M30-RNA vaccine loses efficacy for B16 liver tumors in mice with MCD diet-induced steatohepatitis but not in subcutaneous tumors.

Figure 2:

A. Experimental setup: Mouse model of B6(Cg)-Tyrc-2J/J albino mice and MCD diet-induced steatohepatitis with intrahepatic injection of B16 luciferase-expressing melanoma cells and RNA vaccination (at indicated time points).

B. Tumor growth of intrahepatic B16 tumors presented as bioluminescence imaging showing radiance of liver tumors over time. 4 groups: control - MCD diet, M30 -MCD diet, control – reg. diet, M30 – reg. diet.

C. Quantification of in vitro bioluminescence assay displayed as tumor growth curve. Data represent

D. Tumor burden at d20 presented as tumor-to-liver weight-ratio.

E. Representative pictures of intrahepatic melanoma tumors). 4 groups: Control-MCD diet, M30-MCD diet, Control – reg. diet, M30 - reg. diet.

F. Experimental setup (left): M30-vaccine in C57BL/6 mice fed MCD diet and given s.c. injection of B16 melanoma cells and RNA vaccination (at indicated time points).

G. Subcutaneous tumor model. Tumor growth over time shown as tumor volume over time.

H. Subcutaneous tumor model. Weight of subcutaneous tumors comparing groups: control RNA, M30 (both MCD diet).

To investigate the systemic effects of steatohepatitis on immunotherapy, we used a subcutaneous tumor model. Similar to mice fed with normal diet (Fig. S1C–E)14, MCD diet-induced steatohepatitis did not affect the efficacy of M30 on subcutaneous B16 tumors. A significantly smaller tumor volume was found in the M30 group throughout the experiment (Fig. 2F & 2G). Consistently, the end-point tumor sizes and weights were significantly lower in the M30 RNA group compared to control RNA (Fig. 2H). This indicates that steatohepatitis drives liver-specific impairment of T cell-based immunotherapies.

We confirmed our findings in a second model of diet-induced steatohepatitis (Fig. 3A). We selected the better characterized CDAA diet. Mice fed CDAA diet for 22 weeks showed a significant increase in body weight (Fig 3B) compared to regular chow. B16 cells were once again injected intrahepatically and administration of M30 against control RNA was tested. Consistent with the findings in the MCD diet, efficacy of M30 vaccine was completely abrogated in the CDAA-diet mice measured by both in vivo imaging over time (Fig. 3C & 3D) and by endpoint tumor burden (Fig. 3E & 3F)

Figure 3: M30-RNA vaccine also loses efficacy for B16 liver tumors in mice with a second diet-induced steatohepatitis model (CDAA diet).

Figure 3:

A. Experimental setup: Mouse model of CDAA diet-induced steatohepatitis. B6(Cg)-Tyrc-2J/J albino mice were given CDAA diet for 22 weeks before intrahepatic injection of B16 luciferase-expressing melanoma cells and RNA vaccination (at indicated time points).

B. Mouse weight of B6(Cg)-Tyrc-2J/J albino mice at day of tumor inoculation after 22 weeks of reg diet vs CDAA diet (both n=17).

C. Tumor growth presented as bioluminescence imaging of intrahepatic B16 tumors showing radiance of liver tumors over time (left). 4 groups: Control RNA - CDAA diet, M30 - CDAA diet, control RNA – reg. diet, M30 – reg. diet.

D. Quantification of in vitro bioluminescence assay displayed as tumor growth curve (right). Data represent

E. Tumor burden at d20 presented as tumor-to-liver weight-ratio.

F. Representative pictures of intrahepatic melanoma tumors. 4 groups: Control RNA - MCD diet, M30 - MCD diet, control RNA– reg. diet, M30 – reg - diet.

N-Acetylcysteine administration restores efficacy of M30 RNA vaccine against liver tumors in diet-induced steatohepatitis liver

We have previously shown that NAC reduces reactive oxygen species (ROS) and reverses the CD4+ T cell loss in NAFLD livers.7 Therefore, we tested whether preventing CD4+ T cell loss using NAC could restore the M30 efficacy in NAFLD livers (Fig. 4A). M30 or NAC single agent administration did not significantly affect tumor growth measured by bioluminescence in MCD fed tumor bearing mice (Fig. 4B & 4C). Conversely, combining NAC and M30 caused a significant reduction of tumor growth (Fig. 4B & 4C) and end-point tumor size in MCD fed tumor bearing mice (Fig. 4D & 4E). CD4+ T cell dependency of NAC-rescue of M30 efficacy against intrahepatic B16 tumors in steatohepatitis mice was tested. Depletion of CD4+ T cells was confirmed using flow cytometry (Fig. S1F). Efficacy of M30 and NAC treated mice was abrogated by depletion of CD4+ T cells, confirming M30 dependence on CD4+ T (Fig. 4F, Fig. S1G). Of note, injection of anti-CD4 significantly reduced overall survival resulting in premature death of some of the mice in the CD4 depleted group (Fig. S1H).

Figure 4: NAC administration rescues M30-RNA vaccine efficacy for B16 liver tumors in mice with steatohepatitis.

Figure 4:

A. Experimental setup: Mouse model of B6(Cg)-Tyrc-2J/J albino mice and MCD diet-induced steatohepatitis ± NAC in the drinking water. Mice received intrahepatic injection of B16 luciferase-expressing melanoma cells and RNA vaccination at indicated time points.

B. Quantification of in vitro bioluminescence assay displayed as tumor growth curve.

C. Tumor growth presented as bioluminescence imaging of intrahepatic B16 tumors showing radiance of liver tumors over time. 4 groups: (all MCD diet): control RNA (n=6), M30 (n=6), control RNA + NAC (n=7), M30 + NAC (n=6).

D. Tumor burden at d20 presented as tumor-to-liver weight-ratio.

E. Representative pictures of intrahepatic melanoma tumors (right). Groups as in B.

F. NAC-rescue of M30 vaccine efficacy is CD4+ T cell dependent. Experimental setup as in A. An additional group received anti-CD4-depleting antibodies. 3 groups: (all MCD diet): M30+H2O (n=7), M30+NAC (n=7), M30+NAC+anti-CD4 (n=6). Tumor burden at d20 presented as tumor-to-liver weight-ratio.

Steatohepatitis impairs efficacy of agonistic anti-OX40 T cell-directed immunotherapy in a murine liver tumor model.

OX40 activation is another target currently being investigated to boost T cell anti-tumor immune responses.25 As previously reported26, we found that aOX40 lacks efficacy against B16 melanoma in subcutaneous models (data not shown). In order to determine if our above findings are indeed liver-specific, we chose a model of CT26 colorectal cancer liver tumors to test aOX40 efficacy in healthy vs. diet-induced NAFLD mice (Fig. 5A). Similar to mice injected with M30, mice fed regular diet and injected with aOX40 showed a lower tumor burden compared to the control group (Fig. 5B & 5D). In mice fed MCD diet, aOX40 no longer controlled liver tumor growth (Fig. 5C &5D). This effect was again rescued using combination NAC and aOX40, Mice fed with MCD diet had a significant reduction of tumor burden when compared to aOX40 alone or the corresponding IgG control group (Fig. 5E & 5F). These findings indicate a crucial role for CD4+ T cells in the mechanism of action of OX40-directed immunotherapies. Since aOX40 is known to augment T cell differentiation and cytolytic function of both CD4+ and CD8+ cells, we targeted both T cell subsets using depletion experiments in tumor-bearing MCD diet fed mice receiving aOX40 therapy (Fig. 5G). Again, NAC in combination with aOX40 significantly reduced B16 intrahepatic tumor burden in mice on MCD diet (IgG MCD diet vs. aOX40 MCD NAC), but anti-CD4 abrogated the effect (Fig. 5G). Interestingly, CD8+ T cell depletion in mice with combined NAC and M30 administration reversed the therapeutic effect and actually displayed a trend towards a greater tumor burden, even compared to the IgG control group (Fig. 5G). These effects were less pronounced in mice on regular chow (Fig. S2A). Similar to the above findings above using M30 against B16 tumors, aOX40 injection controlled subcutaneous CT26 CRC cell growth and tumor burden in mice receiving either MCD (Fig. 5H–J) or regular diet (Fig. S2B–D). Taken together, these results strongly suggest that NAFLD selectively impairs T cell-based tumor immunotherapies within the liver but exert minimal effects on extrahepatic tumors.

Figure 5: CD4+ and CD8+ T cell-dependent agonistic anti-OX40 immunotherapy loses therapeutic efficacy for CT26 liver tumors in mice with steatohepatitis but not in subcutaneous tumors.

Figure 5:

A. Experimental setup: BALB/c AnNCr mice were fed MCD diet (+/− NAC in the drinking water) to induce steatohepatitis or regular diet. For depletion experiments, some groups received anti-CD4 or anti-CD8-depleting antibodies prior to tumor inoculation and then once weekly. On d0 mice were injected intrahepatically with CT26 murine colon carcinoma cells. Injection with aOX40 antibody or IgG-Control i.p. was performed at indicated time points.

B. Tumor burden at d20 in CT26 tumor model presented as tumor-to-liver weight-ratio. Mice fed regular diet and injected with aOX40 or control antibody (n=9 IgG/regular diet; n=9 aOX40/regular diet).

C. Tumor burden at d20 in CT26 tumor model presented as tumor-to-liver weight-ratio. MCD diet fed mice injected with aOX40 or control antibody (n=9 IgG/MCD diet; n=9 aOX40/MCD diet).

D. Representative pictures of intrahepatic CT26 tumors are shown.

E. Tumor burden at d20 in CT26 tumor model with aOX40 and NAC. Mice were fed MCD diet and given NAC in drinking water and injected intrahepatically with CT26. (All groups MCD diet. n=7 IgG/water; n=7 aOX40/water; n=10 IgG/NAC; n=11 aOX40/NAC).

F. Representative pictures of intrahepatic CT26 tumors are shown.

G. NAC-rescue of aOX40 is CD4+ and CD8+ T cell dependent. Experimental setup as in C. An additional group received anti-CD4 or anti-CD8-depletion antibody (starting 1 week prior to tumor inoculation, then once weekly), respectively. 4 groups: (all MCD diet): IgG/H2O (n=5) aOX40/ H2O (n=5), aOX40/NAC + anti-CD4 (n=8), aOX40/NAC + anti-CD8 (n=8).

H. CT26 subcutaneous model. BALB/c AnNCr mice were fed MCD diet to induce steatohepatitis prior to s.c. injection of CT26 cells. Mice received agonistic aOX40 antibody on d3, d7, d10. Tumor growth shown as tumor volume over time.

I. Weight of subcutaneous CT26 tumors (as in E) at d14 (left) comparing groups IgG control (n=5), aOX40 (n=5) (both MCD diet).

J. Representative pictures of s.c. CT26 tumors after excision are shown. Two animals in the aOX40 group had no detectable tumors.

Steatohepatitis alters the tumor microenvironment and impairs T cell-directed immunotherapy

In order to further elucidate the mechanism by which steatohepatitis can impair the efficacy of T cell-directed immunotherapy against liver tumors, alterations of immune cells in the liver environment were investigated. T cell subsets (Fig.6 & 7, Fig. S3–6) and myeloid cells (Fig. S3–6) in the liver, TME, and spleens (an approximation of systemic changes) were studied by flow cytometry (gating strategies Fig S3A & S3B). In mice fed a regular diet, a significant increase of CD4+ T cell frequency was found in animals injected with M30 RNA vaccine, while CD8+ T cells were not significantly altered (Fig. 6A). Regulatory T cells (Tregs), on the other hand, were significantly reduced in frequency in M30 groups (Fig. 6B). Tumor associated myeloid cells are often reshaped to obtain immune suppressive function and promote tumor progression such as myeloid-derived suppressor cells (MDSCs).The myeloid cell population was examined and a reduction of all MDSCs and specifically granulocytic MDSCs (G-MDSCs, defined as Ly6Clow and Ly6G+, Fig S3B) was found in the M30 group (Fig. S3C). This data indicates that M30 increases CD4+ T cell frequency but reduces both the Treg-fraction and tumor-promoting myeloid populations. These changes generate a TME which successfully controls liver tumor growth.

Figure 6: Steatohepatitis causes a selective loss of intrahepatic CD4+ T lymphocytes and abrogates M30 RNA-vaccine mediated changes in the liver tumor microenvironment.

Figure 6:

A. Effect of M30-RNA vaccine (n=8) vs control RNA (n=7, both regular diet) on the frequency of intrahepatic CD4+ and CD8+ T cells in a mouse model of B16 liver tumors (experimental setup Fig. 1A).

B. Effect of M30-RNA vaccine (n=8) vs control RNA (n=7, both regular diet) on the frequency of Treg cells in a mouse model of B16 liver tumors.

C-E. Effect of MCD diet on M30-RNA vaccine-mediated changes of CD4+ T cell numbers presented as cells per gram tissue in the liver (C), tumor (D) and spleen (E). Experimental setup as in Fig. 2A. Four groups: control RNA - MCD diet, M30 - MCD diet, control RNA – reg. diet, M30 – reg. diet.

F-H. Effect of NAC administration on M30-RNA vaccine-mediated changes of CD4+ T cell numbers in the liver (F), tumor (G) and spleen (H) in mice fed with MCD diet. Experimental setup as in Fig. 4A. 4 groups (all MCD diet): control RNA (n=6), M30 (n=6), control RNA + NAC (n=7), M30 + NAC (n=5).

I-L. Effect of CDAA diet on M30-RNA vaccine-mediated changes of CD4+ T cell numbers in the liver (I), tumor (K) and spleen (L). Experimental setup as in Fig. 3A. Four groups: control RNA - CDAA diet, M30 - CDAA diet, control RNA – reg. diet, M30 - reg diet.

Figure 7: The effect of diet-induced steatohepatitis on intrahepatic and tumor-infiltrating T cell subsets in CT26 liver tumor-bearing mice receiving aOX40.

Figure 7:

A-B. Effect of MCD diet on aOX40-mediated changes of CD4+ T cell numbers in the liver(A) and tumor(B). Experimental setup as in Fig. 5A. 4 groups: IgG control -MCD diet (n=19), aOX40 - MCD diet (n=13), IgG control – reg. diet (n=16), aOX40 – reg. diet (n=21).

C-D. Effect of NAC on aOX40-mediated changes of tumor CD4+ (C) and CD8+ T cells

(D) in mice fed MCD diet. Experimental setup as in Fig. 5A. (All groups MCD diet. n=7 IgG/water; n=7 aOX40/water; n=10 IgG/NAC; n=11 aOX40/NAC).

E-F. Effect of MCD diet on aOX40-mediated changes of CD4+ (E) and CD8+ effector memory T cell numbers (F) in the tumor. Experimental setup as in Fig. 5A, groups as in A. Numbers of tumor-infiltrating CD4+ effector memory (CD3+CD4+CD44+CD62L−) cells (E) and CD8+ effector memory (CD3+CD4+CD44+CD62L−) cells (F) are shown as cells per gram tumor.

G-H. Effect of NAC on aOX40-mediated changes of liver (G) and tumor (H) CD4+ effector memory T cells in mice fed MCD diet. Experimental setup as in Fig. 5A. (All groups MCD diet. n=7 IgG/water; n=7 aOX40/water; n=10 IgG/NAC; n=11 aOX40/NAC).

Mice with MCD diet-induced NASH showed significantly reduced numbers of liver and tumor CD4+ T cells, whereas numbers in spleens were not significantly different (Fig.6C, 6D & 6E). Unlike in mice fed with normal diet, M30 failed to significantly increase the CD4+ T cell frequency in liver tissue of MCD diet fed mice (Fig. S4A). Numbers of CD8+ T cells in liver, tumor and spleen were not significantly affected in mice fed MCD diet (Reg Control vs MCD Control, Fig. S4D, S4E & S4F). M30 treatment resulted in increased numbers of tumor infiltrating CD4+ and CD8+T cells only in regular diet fed mice, an effect not to be seen in MCD diet induced NASH mice (Fig. 6D & S1E). Splenic CD4+ T cells did not show this trend (Fig. 6E), indicating a local process rather than systemic effects.

Importantly, in mice on MCD diet, NAC administration rescued M30-induced increase of CD4+ T cells in numbers and frequency in TILs, showed less effect on the surrounding liver tissue and had no effect in the spleen (Fig. 6F–H, Fig. S4G–S4I) NAC increased tumor CD8+ T cell numbers but had no effect on CD8+ T cells in liver and spleen (Fig.S4J–S4L). MDSCs and macrophages in the liver of MCD diet fed mice were significantly increased compared to regular chow fed mice (Fig. S4M). NAC did not significantly affect MDSCs and macrophages in the MCD diet groups (Fig S4N).

Similar findings were noted in a CDAA-induced NASH model including a significant loss in numbers of hepatic CD4+ T cells and CD8+ T cells (Fig. 6I & Fig. S5A). Among TILs, only mice fed regular diet showed a significant increase in CD4+ and CD8+ T cell numbers after M30 treatment, again an effect not seen in the CDAA induced NASH mice. Splenic lymphocytes were still showing increased T cell numbers with borderline significance in the CDAA diet groups after M30 treatment (Fig.6K & 6L & Fig S5B & S5C). In contrast to mice with MCD diet induced NASH, mice fed CDAA diet showed reduced numbers of all MDSCs, MDSC subgroups and macrophages (Fig.S5D).

Next we studied immune cells in the CT26 tumor model with aOX40 administration. Again, similar to C57BL/6 mice, hepatic CD4+ T cells were significantly reduced in BALB/c mice on MCD diet (Fig. 7A). Tumor infiltrating CD4+ cells were reduced in mice fed MCD diet and aOX40 administration resulted in the highest CD4+ T cell numbers in TILs in regular diet fed mice (Fig. 7B). Tumor CD8+ T cell numbers were not significantly different between groups in this model and liver CD8+ T cells showed actually higher numbers in the MCD group after aOX40 treatment (Fig. S6A). Although administration of NAC in mice with steatohepatitis did not have a strong effect on overall T cell numbers in the liver (Fig. S6B), NAC alone and combined NAC and aOX40 significantly increased CD4+ and CD8+ T cells in the tumor (Fig. 7C & Fig. 7D). OX40 has been shown to be involved in effector and memory T cells responses27. Thus, we studied CD44high/CD62L− effector memory T cells in this model. Efficacy of aOX40 was strongly associated with increased CD4+ and CD8+ effector memory T cells in TILs of regular mice after aOX40 administration (Fig.7E & 7F). In mice with steatohepatitis, tumor infiltrating CD4+ effector memory T cells were reduced, and aOX40 administration did not cause a significant increase of effector CD4+ and CD8+ T cell numbers in the tumor (Fig.7E & 7F). Liver and spleen CD4+ effector memory cells were increased in MCD mice after aOX40 administration whereas CD8+ effector memory T cells were not significantly different between normal and steatohepatitis conditions (Fig. S6C). Furthermore, specifically liver CD4+ effector memory T cells (Fig. 7G) as well as CD4+ and CD8+ effector memory T cells in the tumor (Fig.7H & S6D) were significantly increased in the combination group using NAC and aOX40 compared to MCD fed control mice. Liver CD8+ effector memory cells were actually higher in numbers in MCD fed mice treated with aOX40 without NAC (Fig S6E). Splenic CD4+ effector memory T cells were higher in NAC treated mice regardless of aOX40 administration and splenic CD8+ effector memory cells were not significantly different (Fig.S6F). These findings suggest, that specifically liver CD4+ effector memory T cells and tumor infiltrating CD4+ and CD8+ effector memory T cells are an important mediator of aOX40 dependent anti-tumor effects in the liver and can be restored by NAC in mice with steatohepatitis.

In summary, these results suggest that reduced hepatic T cells and specifically CD4+ T cells in diet-induced steatohepatitis livers impairs immunotherapy efficacy. Furthermore, preventing CD4+ T cell loss in steatohepatitis livers restores antitumor efficacy of T cell-directed immunotherapy against liver tumors. Efficient T cell-based immunotherapy with M30 and aOX40 is associated with increased numbers of CD4+ and CD8+ T cells in tumor infiltrating lymphocytes. Tumor effector memory T cells are specifically associated with effective control of liver tumors in the aOX40 immunotherapy model.

Discussion

The incidence of cancer patients with comorbid NAFLD/NASH is expected to increase. However, the effect of NAFLD/NASH on the efficacy of immunotherapy for liver tumors remains unknown. Here we studied the efficacy of two different immunotherapies in two different diet-induced steatohepatitis models and provide robust evidence that steatohepatitis impairs the efficacy of immunotherapy directed against liver tumors.

The liver is the most common site for distant metastases and 95% of liver tumors are metastatic lesions from other cancers.28 Patients with liver metastases secondary to melanoma or non-small cell lung cancer (NSCLC) have been reported to experience reduced treatment response, progression free survival, and CD8+ T cell infiltration.6 It is suggested that body mass index correlates with outcomes of NSCLC patients treated with checkpoint inhibitors.29 We have previously demonstrated that NAFLD promotes liver tumorigenesis through ROS-mediated death of CD4+ T cells.7, 8 We performed a comprehensive analysis of the effects of steatohepatitis on immunotherapy against liver tumors. We demonstrate that steatohepatitis alters the TME resulting in poor response to CD4+ T cell-based immunotherapies.

Several murine dietary fatty liver and steatohepatitis models have been described, all in the past.30, 31 There is no model which can fully mimic the complex biology of patients with distinct disadvantages.32, 32 Mice on CDAA diet gained significant body weight, whereas mice on MCD diet lost weight over time due to the inability to excrete hepatic very-low-density lipoprotein.33 As such, the MCD model is not ideal to study extrahepatic features of steatohepatitis. Nevertheless, we have shown that the loss of CD4+ T cells is liver specific in both models allowing us to test the effects of steatohepatitis on M30 and aOX40.7 The numbers of hepatic MDSCs and macrophages differed between MCD diet and CDAA diet steatohepatitis models, indicating different underlying mechanisms of immune-mediated liver injuries and a different function in tumor bearing mice with underlying NASH.

Recently, increasing knowledge regarding mechanisms of primary, adaptive, and acquired resistance to immunotherapy have been described. These include both tumor cell intrinsic mechanisms such as absence of antigenic proteins, a defect in antigen presentation, genetic T cell exclusion and insensibility of tumors to T cells as well as tumor cell extrinsic mechanisms including T cell exclusion, expression of inhibitory immune checkpoints and immunosuppressive T cells.34 A number of studies also suggest a critical role of the microbiome in controlling the efficacy of immunotherapy.35 To the best of our knowledge, we present the first study demonstrating that steatohepatitis impairs immunotherapy efficacy.

Metabolic changes within the tumor and immune cells in the TME can hinder immunotherapy.36 Central memory and stem-cell like precursor T cells are T cell subsets that exhibit low metabolic activity and contribute to long-lasting immune responses. Effector T cells, on the other hand, are highly metabolically active and use mainly glycolysis to cover their metabolic needs.37, 38 The TME surrounding liver tumors generates low glucose levels and low oxygen tension which hinders highly metabolically active cells. Similar to effector T cells, chimeric antigen receptor T cells used against several malignancies including HCC39, undergo quick turnover and thus often fail to provide durable responses.38 In one study, Metformin, an anti-diabetic drug, successfully prevented apoptosis and exhaustion of tumor infiltrating CD8+ T cells and increased effector-memory T cells population within TILs of solid mouse tumors.40 Specifically, Metformin restored fatty acid oxidation in CD8+ T cells resulting in increased CD8+ T memory cells and improved efficacy of an anti-cancer vaccine.41 These findings are consistent with our finding that administration of NAC resulted in lower rates of CD4+ T cell death in the livers of mice with fatty liver disease.9

Our findings confirm that effective cancer immunotherapy relies on potent immunotherapeutic agents to initiate and activate tumor-specific T cell responses, but also highlight the need for comprehensive considerations of organ specific distribution of immune cells and changes in the TME. Prudent strategies and stratification of patients are needed when using immunotherapy for liver tumors, particularly in patients with underlying NAFLD/NASH. Treatment of NAFLD/NASH may be beneficial to efficacy of immunotherapy for liver metastases. This must be validated in a randomized clinical trial. Future immunotherapy must consider all aspects of the host, including microbiome,42 tumor and organ metabolism, as well as comorbid conditions affecting the tumor bearing organ to ensure optimal response to therapy.

Supplementary Material

1

2

3

4

5

6

7

What you need to know:

Background and Context:

Non-alcoholic steatohepatitis causes loss of hepatic CD4+ T cells and promotes tumor growth. This study investigated the effects of steatohepatitis on the efficacy of immunotherapeutic agents against liver tumors in mice.

New Findings:

Steatohepatitis reduces the abilities of immunotherapeutic agents such as M30 and aOX40 to inhibit liver tumor growth by reducing tumor infiltration by CD4+ T cells and effector memory cells. NAC restores T-cell numbers in tumors and increases the ability of M30 and aOX40 to slow tumor growth in mice

Limitations:

This study was performed in mice with injected tumor cells. Further studies are needed in other models and in humans.

Impact:

These findings might increase the efficacy of immunotherapy for liver cancer.

Lay Summary:

The authors found that fatty liver reduces the efficacy of immunotherapy against liver cancer, but that restoring immune cells to the liver can increase the efficacy of this treatment.

Acknowledgements:

We would like to thank the NIH tetramer facility for providing PBS57/CD1d-tetramer

Grant support: T.F.G. was supported by the Intramural Research Program of the NIH, NCI (ZIA BC 011345). S.W was supported by the Deutsche Forschungsgemeinschaft (WA-4610/1-1).

Abbreviations:

aOX40

agonistic anti-OX40 antibody

CDAA

choline-deficient L-amino acid defined

CRC

colorectal cancer

HCC

hepatocellular carcinoma

M30

M30-RNA vaccine

MCD

methionine–choline-deficient

NAC

N-Acetylcysteine

NAFLD

Nonalcoholic fatty liver disease

NASH

Nonalcoholic steatohepatitis

ROS

reactive oxygen species

TIL

tumor-infiltrating leukocyte

TME

tumor microenvironment

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Conflict of interest: The authors declare no competing interest. Some of the authors are employees at BioNTech AG (Mainz, Germany) as mentioned in the affiliations. U.S. is stock owner of BioNTech AG (Mainz, Germany). U.S., M.V. are inventors on patents and patent applications, which cover parts of this article.

References:

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1

2

3

4

5

6

7