CD80 Expression on Tumor Cells Alters Tumor Microenvironment and Efficacy of Cancer Immunotherapy by CTLA-4 Blockade

Simple Summary The recent discovery of immune checkpoint inhibitors constituted a breakthrough in cancer treatment, but most patients are resistant to this therapy. Although the co-stimulatory molecule cluster of differentiation (CD) 80 has been detected in several types of tumor cells, its role in the tumor microenvironment and its sensitivity to immune checkpoint blockade are unclear. We, therefore, introduced a clinically relevant mouse tumor model with deactivated CD80. The deactivation promoted a “hot” tumor microenvironment and enhanced the sensitivity to immune checkpoint blockade with antibody against the cytotoxic T-lymphocyte antigen 4 (CTLA-4). This study contributed to the research into predictive markers to select patients who are suitable for immune checkpoint blockade therapy and suggested the development of a novel cancer immunotherapy based on a tumor-cell-targeted CD80 blockade. Abstract Cluster of differentiation (CD) 80 is mainly expressed in immune cells but can also be found in several types of cancer cells. This molecule may either activate or inhibit immune reactions. Here, we determined the immunosuppressive role of CD80 in the tumor microenvironment by CRISPR/Cas9-mediated deactivation of the corresponding gene in the mouse oncogenic TC-1 cell line. The tumor cells with deactivated CD80 (TC-1/dCD80-1) were more immunogenic than parental cells and induced tumors that gained sensitivity to cytotoxic T-lymphocyte antigen 4 (CTLA-4) blockade, as compared with the TC-1 cells. In vivo depletion experiments showed that the deactivation of CD80 switched the pro-tumorigenic effect of macrophages observed in TC-1-induced tumors into an anti-tumorigenic effect in TC-1/dCD80-1 tumors and induced the pro-tumorigenic activity of CD4+ cells. Moreover, the frequency of lymphoid and myeloid cells and the CTLA-4 expression by T helper (Th)17 cells were increased in TC-1/dCD80-1- compared with that in the TC-1-induced tumors. CTLA-4 blockade downregulated the frequencies of most immune cell types and upregulated the frequency of M2 macrophages in the TC-1 tumors, while it increased the frequency of lymphoid cells in TC-1/dCD80-1-induced tumors. Furthermore, the anti-CTLA-4 therapy enhanced the frequency of CD8+ T cells as well as CD4+ T cells, especially for a Th1 subset. Regulatory T cells (Treg) formed the most abundant CD4+ T cell subset in untreated tumors. The anti-CTLA-4 treatment downregulated the frequency of Treg cells with limited immunosuppressive potential in the TC-1 tumors, whereas it enriched this type of Treg cells and decreased the Treg cells with high immunosuppressive potential in TC-1/dCD80-1-induced tumors. The immunosuppressive role of tumor-cell-expressed CD80 should be considered in research into biomarkers for the prediction of cancer patients’ sensitivity to immune checkpoint inhibitors and for the development of a tumor-cell-specific CD80 blockade.


Introduction
The costimulatory molecule cluster of differentiation (CD) 80, which can be expressed on antigen-presenting cells (APCs) or tumor cells, interacts with both costimulatory (CD28) we developed an experimental model of mouse tumors with functional or deactivated CD80 and compared their immunogenicity and sensitivity to immunotherapy. Tumors with CD80 deactivation were more immunogenic and prone to CTLA-4 blockade.

CD80 Deactivation Reduced Tumor Growth
As CD80 expression on tumor cells can influence their immunogenicity, we sought to determine whether CD80 expression on TC-1 cells [34] affects tumor growth, the tumor microenvironment, and the sensitivity to immune checkpoint blockade. Therefore, we generated TC-1 cells with a deactivated CD80 molecule (TC-1/dCD80) using the CRISPR/Cas9 system. As the clones obtained after CD80 deactivation were heterogeneous in their CD80 expression, we selected three clones with apparently reduced CD80 surface expression compared with TC-1 cells (Figure 1) for an analysis of immunogenicity ( Figure 2A). Deactivation of CD80 in the TC-1 cell line significantly reduced the growth of tumors induced with all three clones. While TC-1/dCD80 clone 1 (TC-1/dCD80-1) formed tumors in all inoculated mice, clones TC-1/dCD80-2 and -3 generated tumors in 80% and 60% of mice, respectively, and their growth was highly delayed. We, therefore, used the TC-1/dCD80-1 clone in our further study of the tumor microenvironment and immunotherapy response.
Cancers 2021, 13,1935 3 of 17 Regarding the various interacting partners of CD80, with their broad range of effects on anti-tumor immune responses, the role of CD80 in the tumor microenvironment and its impact on the efficacy of cancer therapy need to be elucidated. In the present study, we developed an experimental model of mouse tumors with functional or deactivated CD80 and compared their immunogenicity and sensitivity to immunotherapy. Tumors with CD80 deactivation were more immunogenic and prone to CTLA-4 blockade.

CD80 Deactivation Reduced Tumor Growth
As CD80 expression on tumor cells can influence their immunogenicity, we sought to determine whether CD80 expression on TC-1 cells [34] affects tumor growth, the tumor microenvironment, and the sensitivity to immune checkpoint blockade. Therefore, we generated TC-1 cells with a deactivated CD80 molecule (TC-1/dCD80) using the CRISPR/Cas9 system. As the clones obtained after CD80 deactivation were heterogeneous in their CD80 expression, we selected three clones with apparently reduced CD80 surface expression compared with TC-1 cells (Figure 1) for an analysis of immunogenicity ( Figure  2A). Deactivation of CD80 in the TC-1 cell line significantly reduced the growth of tumors induced with all three clones. While TC-1/dCD80 clone 1 (TC-1/dCD80-1) formed tumors in all inoculated mice, clones TC-1/dCD80-2 and -3 generated tumors in 80% and 60% of mice, respectively, and their growth was highly delayed. We, therefore, used the TC-1/dCD80-1 clone in our further study of the tumor microenvironment and immunotherapy response. We first tested various doses of TC-1/dCD80-1 cells to find a dose inducing tumors of a size that was comparable to the tumors induced with 3 × 10 4 TC-1 cells. While the growth of the TC-1/dCD80-1-induced tumors after inoculation of 3 × 10 4 and 1 × 10 5 cells was slower than that of TC-1-induced tumors, the 3 × 10 5 dose significantly enhanced the growth of TC-1/dCD80-1-induced tumors and provided tumors of a similar size to those induced with 3 × 10 4 TC-1 cells ( Figure 2B). We first tested various doses of TC-1/dCD80-1 cells to find a dose inducing tumors of a size that was comparable to the tumors induced with 3 × 10 4 TC-1 cells. While the growth of the TC-1/dCD80-1-induced tumors after inoculation of 3 × 10 4 and 1 × 10 5 cells was slower than that of TC-1-induced tumors, the 3 × 10 5 dose significantly enhanced the growth of TC-1/dCD80-1-induced tumors and provided tumors of a similar size to those induced with 3 × 10 4 TC-1 cells ( Figure 2B).

CD80 Deactivation Altered Immune Reactions and Sensitivity to CTLA-4 Blockade
To investigate the impact of the CD80 deactivation in tumor cells on anti-tumor immune reactions, we depleted CD4 + , CD8 + , natural killer (NK) 1.1 + cells, or macrophages in mice bearing TC-1-and TC-1/dCD80-1-induced tumors ( Figure 3). The CD80 deactivation switched the pro-tumor role of macrophages observed in TC-1-induced tumors ( Figure  3A) to anti-tumor abilities in TC-1/dCD80-1-induced tumors ( Figure 3B). Moreover, the depletion of CD4 + cells did not markedly affect the growth of TC-1-induced tumors (Figure 3A) but did reduce the growth of TC-1/dCD80-1-induced tumors ( Figure 3B). CD8 + cells had an anti-tumor effect regardless of CD80 expression on tumor cells. Depletion of NK1.1 + cells significantly enhanced the growth of TC-1-induced tumors, whereas it did not have a significant impact on the growth of TC-1/dCD80-1-induced tumors.
CTLA-4 and PD-L1 blockades considerably decreased the infiltration of most immune cells into TC-1-induced tumors, but this effect was significant only for NKT cells ( Figure 5B). On the contrary, CTLA-4 blockade substantially increased the infiltration of lymphoid cells into TC-1/dCD80-1-induced tumors, such as CD4 + or CD8 + T cells, γδT cells, and NK cells, and simultaneously decreased the infiltration of pDCs. Moreover, CTLA-4 blockade enhanced the infiltration of eosinophils (CD45 + , CD11b + , CD11c − , Ly6G − , SSC hi ) into tumors, regardless of CD80 expression on tumor cells, and increased the frequency of MHC-II − M2 macrophages within the subpopulation of TAMs in TC-1-induced tumors.
Taken together, CD80 deactivation increased numbers of both lymphoid and myeloid cells infiltrating tumors and was associated with further enhancement of lymphoid cells after treatment with the CTLA-4 antibody.
In order to assess which T cell subsets were the main target of CTLA-4 blockade in the tumor microenvironment, we next evaluated the level of CTLA-4 expression on T cells. In non-treated tumors, CTLA-4 was mainly expressed on Th2 cells, Th17 cells, and Treg cells, whereas its expression was negligible on Th1, γδT, and CD8 + T cells (CD45 + , CD3 + , γδTCR − , CD8 + ; Figure 6B). In TC-1/dCD80-1-induced tumors, CTLA-4 expression was significantly upregulated on Th17 cells, which might be associated with the enhanced sensitivity of these tumors to CTLA-4 blockade.  In order to assess which T cell subsets were the main target of CTLA-4 blockade in the tumor microenvironment, we next evaluated the level of CTLA-4 expression on T cells. In non-treated tumors, CTLA-4 was mainly expressed on Th2 cells, Th17 cells, and Treg cells, whereas its expression was negligible on Th1, γδT, and CD8 + T cells (CD45 + , CD3 + , γδTCR − , CD8 + ; Figure 6B). In TC-1/dCD80-1-induced tumors, CTLA-4 expression was significantly upregulated on Th17 cells, which might be associated with the enhanced sensitivity of these tumors to CTLA-4 blockade.
Next, we evaluated the effect of CD80 deactivation in tumor cells and immune checkpoint blockade on the expression of Treg markers by the automatically generated Treg subpopulations ( Figure S2). Although CD80 deactivation did not have a significant impact on the expression of most markers expressed by the Treg clusters, CD73 expression was significantly downregulated in subpopulation 2 in TC-1/dCD80-1-induced tumors compared with the same subpopulation in the TC-1-induced tumors. Immunotherapy significantly affected the expression of several markers in the Treg subsets, such as the expression of CTLA-4, GITR, CD73, and GrzB in TC-1-induced tumors and the expression of GITR in the TC-1/dCD80-1-induced tumors. These data suggest that CD80 expression in tumor cells and immune checkpoint blockade might have altered immunosuppressive potential within distinct Treg cell subsets. ulated the frequency of Treg cells with subpopulation 3 and markedly upregulated subpopulations 2 and 4 in TC-1-induced tumors. The proportion of Treg subpopulation 3 was increased in TC-1/dCD80-1-induced tumors treated with CTLA-4 blockade, which resulted in a significant difference in this subpopulation in comparison with the TC-1-induced tumors treated with the same antibody. These data show that CD80 deactivation in tumor cells reduced the immunosuppressive potential of Treg cells after tumor treatment with CTLA-4 blockade.  Figure S2). Although CD80 deactivation did not have a significant impact on the expression of most markers expressed by the Treg clusters, CD73 expression was significantly downregulated in subpopulation 2 in TC-1/dCD80-1-induced tumors compared with the same subpopulation in the TC-1-induced tumors. Immunotherapy significantly affected the expression of several markers in the Treg subsets, such as the expression of CTLA-4, GITR, CD73, and GrzB in TC-1-induced tumors and the expression of GITR in the TC-1/dCD80-1-induced tumors. These data suggest that CD80 expression in tumor cells and immune checkpoint blockade might have altered immunosuppressive potential within distinct Treg cell subsets.

Discussion
Immune checkpoint blockade is a promising approach to cancer treatment despite its lack of efficacy in most patients. Markers for the prediction of immune checkpoint blockade efficacy are, therefore, being intensively investigated [35]. CD80 is expressed in various human tumors, such as melanoma, colon adenoma, and gastric adenocarcinoma [4,36,37], as well as in oncogenic cell lines, for instance, cell lines derived from melanoma, colorectal carcinoma, Burkitt's lymphoma, and gastrointestinal cancer [6,38]. CD80 expression has also been reported in the mouse oncogenic TC-1 cell line used in this study [34]. As CD80 may either stimulate or inhibit the immune response [1,2], we tested the role of CD80 expression on TC-1 cells in the oncogenicity and efficacy of CTLA-4 and/or PD-L1 blockades. Previously, it has been shown that CD80 silencing, as well as overexpression, in tumor cells could inhibit their oncogenicity [6]. In line with that report, CD80 deactivation in TC-1 cells reduced tumor formation and growth in our study. However, the role of CD80 may differ in various tumor types. In contrast to the aforementioned study, deactivation of CD80 expression or its neutralization by an anti-CD80 antibody has promoted the expansion of colonic neoplasia in mice and reduced T cell cytotoxicity [37].
Here, we showed that CD80 deactivation in tumor cells affected the pro-/anti-tumorigenic role of distinct immune cell populations. For instance, NK1.1 + cells lost the anti-tumorigenic function in mice bearing TC-1/dCD80-1-compared with those with TC-1-induced tumors. Our data correspond to the observation of an NK cell-mediated reduction of CD80 high tumor growth [5]. Surprisingly, we observed that CD80 deactivation in tumor cells markedly enhanced infiltration by NK and NKT cells, although the depletion of NK1.1 + cells did not affect TC-1/dCD80-1-induced tumor growth. Next, macrophages gained an antitumorigenic function after CD80 deactivation in TC-1 cells, which is in agreement with the enhanced frequency of M1 macrophages in TC-1/dCD80-1-induced tumors. In line with this result, the blockade of the CTLA-4/CD80 axis with ipilimumab in melanoma patients has induced a switch in macrophage polarization from the M2 to the M1 phenotype [39]. As CD4 + T cells and activated NKT cells have been reported to promote macrophage polarization into the M1 phenotype [40][41][42], enhanced infiltration of these cells might alter macrophage polarization in TC-1/dCD80-1-induced tumors. Furthermore, CD4 + cells became pro-tumorigenic in TC-1/dCD80-1-induced tumors, while CD8 + cells controlled tumor growth regardless of CD80 expression. It has been reported that CD4 + T cells preferentially express CTLA-4 and, therefore, have higher immunosuppressive potential compared with that of CD8 + T cells [43]. Moreover, CD80 deactivation in tumor cells promoted tumor infiltration with APCs in our model. CD28 signaling induced by APC-mediated co-stimulation has been shown to enhance CTLA-4 expression, predominantly on the Th17 cell subset [44]. In our study, the APCs in the TC-1/dCD80-1-induced tumors might enhance T cell activation and CTLA-4 expression on Th17 cells.
CTLA-4 blockade did not significantly reduce the growth of TC-1-induced tumors, while it did markedly inhibit the growth of TC-1/dCD80-1-induced tumors. Furthermore, PD-L1 blockade was ineffective as a single therapy, and it did not enhance the effect of CTLA-4 blockade in our study. However, the supportive effects of anti-CTLA-4 or anti-PD-L1 therapies on treatment with cisplatin or a synthetic fusion protein vaccine, inducing an immune response against the HPV16 E7 oncoprotein, respectively, have been previously shown in TC-1-induced tumors [45,46]. Similarly to the TC-1/dCD80-1-induced tumors, anti-CTLA-4 administered as a single therapy reduced the progression of another tumor type, i.e., mouse bladder cancer induced by the MB49 cell line with low CD80 expression [6,47].
Unlike in our study, simultaneous blockade of CTLA-4/CD80 and PD-1/PD-L1 axes has been more efficient than a single therapy in several studies [48][49][50]. Paradoxically, anti-PD-L1 administered as a single therapy may enhance CTLA-4/CD80-mediated immunosuppression in some patients due to the disruption of the tumor-suppressive CD80 and PD-L1 in cis interaction [9]. However, the interaction of mouse CTLA-4 with CD80 has been reported to outcompete the tumor-suppressive CD80 and PD-L1 in cis interaction [51]. This corresponds to our previous observation showing that deactivation of PD-L1 in the TC-1 cell line markedly reduced the tumorigenicity of these cells, which implies the pro-tumorigenic role of the PD-L1 molecule, rather than CD80/PD-L1-mediated tumor suppression [52]. We, therefore, assume that PD-L1 on tumor cells did not effectively control the CTLA-4/CD80 axis in TC-1-and TC-1/dCD80-1-induced tumors.
In our study, the depletion of CD8 + cells abrogated the effect of CTLA-4 blockade, whereas the depletion of CD4 + cells synergized with the therapy in TC-1/dCD80-1-induced tumors. This result corresponds to published data showing that the efficacy of anti-CTLA-4 treatment has been dependent on activated tumor-infiltrating CD8 + T cells [53]. Moreover, distinct levels of PD-1, a marker of T cell activation, on CD8 + T cells [54,55], might also support the enhanced efficacy of CTLA-4 blockade in TC-1/dCD80-1-compared with that in TC-1-induced tumors. Furthermore, we did not observe a significant impact of the depletion of NK1.1 + cells or macrophages on the efficacy of CTLA-4 blockade although this treatment has been shown to induce the activation and degranulation of tumor-infiltrating NK cells [56].
CTLA-4 or PD-L1 blockade downregulated the frequency of immune cells in TC-1induced tumors, while anti-CTLA-4 treatment upregulated the frequency of most lymphoid cells in TC-1/dCD80-1-induced tumors. Enhanced infiltration of immune cells into tumors treated with immune checkpoint inhibitors has been previously reported [57]. As the efficacy of immune checkpoint blockade is dependent on tumor-infiltrating immune cells and tumor-specific T cell responses [58], increased infiltration of TC-1/dCD80-1-induced tumors by immune cells might contribute to the sensitivity of these tumors to anti-CTLA-4 therapy.
The CD4 + T cell subsets regulate tumor growth differently [59]. Th1 cells in tumors protect the host against tumor growth, whereas Th17 and Treg cells are usually associated with progression of the disease, and Th2 cells do not correlate with clinical outcomes in many cases [59][60][61][62][63]. The anti-CTLA-4 treatment upregulated the frequency of Th1, Th2, Th17, and Treg cells in TC-1/dCD80-1-induced tumors. This effect was not elicited in TC-1induced tumors. The highly increased Th1 cells probably contributed to the anti-tumor effect of CTLA-4 blockade in TC-1/dCD80-1-induced tumors. Similarly, CTLA-4 blockade has been previously reported to enrich Th1 and Th2 subsets in mouse as well as human tumors and enhance IFN-γ production by T cells [64][65][66].
Treg cells comprised the most abundant CD4 + T cell subset in both TC-1-and TC-1/dCD80-induced tumors. Treg subpopulations 1 and 2, automatically generated by the FlowSOM software, markedly expressed CTLA-4, GITR, and ICOS molecules. High expression of CTLA-4, as well as co-stimulatory receptors GITR and ICOS, has been observed in Treg cells infiltrating mouse and human tumors [26]. CTLA-4 has been shown to be indispensable in Treg-mediated immunosuppression, as CTLA-4 − Treg cells were unable to maintain self-tolerance and immune homeostasis, and Treg-specific CTLA-4 deactivation promoted anti-tumor immunity [19]. Moreover, GITR and ICOS molecules maintain Treg homeostasis, survival, and immunosuppressive functions [67][68][69]. Thus, we suppose that the immunosuppressive potential of subpopulations 1 and 2 was high, whereas the low CTLA-4 expression in subpopulations 3 and 4 implied limited immunosuppressive potential. Furthermore, subpopulation 4 of Treg cells was characterized by a low expression of ICOS and GITR and slightly higher Nrp-1 expression compared with that in the remaining Treg subpopulations. Similarly, Nrp-1 did not cluster with other markers in a study of the phenotypic diversity of Treg cells isolated from skin [70]. Nrp-1 + Treg cells have a strong potential to infiltrate tumors in a vascular endothelial growth factor (VEGF)-dependent manner and inhibit the anti-tumor immune response [25]. As previously noted, VEGF occurs in TC-1-induced tumors [71]. However, the ICOS − Treg subset has been defined as "death prone" [67,68]. Therefore, we presume that the immunosuppressive potential of subpopulation 4 was markedly limited. Our analysis of Treg subpopulations suggests that the decreased immunosuppressive potential of Treg cells in TC-1/dCD80-1-induced tumors after CTLA-4 blockade might also contribute to the anti-tumor effect of anti-CTLA-4 therapy in these tumors.

Mice
Animal experiments were performed with female 6 to 8-week-old C57BL/6N mice (Charles River, Sulzfeld, Germany) that were maintained under specific pathogen-free conditions at the animal facility of the Czech Center of Phenogenomics (BIOCEV, Vestec, Czech Republic). The guidelines for the proper treatment of laboratory animals were observed.
The B7-1 Double Nickase Plasmid (m) kit (sc-419570-NIC; Santa Cruz Biotechnology, Dallas, TX, USA) was used to produce TC-1/dCD80 cell clones with deactivated CD80. The transfected cells were selected for 4 days with 6 µg/mL puromycin, which was added to the culture media 2 days after transfection. Next, cells were stained with anti-CD80-FITC (fluorescein isothiocyanate) antibody (clone 16-10A1; BD Pharmingen, San Diego, CA, USA) and single-cell clones with deactivated CD80 were selected by cell sorting into a 96-well plate using a flow cytometer FACSAria Fusion (BD Biosciences, Franklin Lakes, NJ, USA).
Deactivation of the CD80 gene in the TC-1/dCD80-1 clone was verified by sequencing of the target site.

Animal Experiments
Tumor cells were suspended in 150 µL PBS and subcutaneously (s.c.) administered into the back of C57BL/6N mice (five per group) under anesthesia with xylazine (16 mg/kg; Bioveta, Ivanovice na Hane, Czech Republic) and ketamine (100 mg/kg; Bioveta). Tumors were measured three times a week with calipers, and tumor volume was calculated using the formula (π/6) (

Flow Cytometry
Cells obtained from tumors were incubated with Fixable Viability Dye eFluor 455UV (eBioscience) to stain dead cells. Then, the cells were treated with anti-mouse CD16/32 (Fc block, clone 93; BioLegend) and subsequently with antibodies binding surface markers ( Table 1). The washed cells were fixed and permeabilized with the Fixation/Permeabilization working solution (eBioscience, Thermo Fisher Scientific, Waltham, MA, USA). Furthermore, a working solution of the Permeabilization Buffer (eBioscience) was used to stain intracellular and nuclear markers with respective antibodies. Measurement of the stained samples was performed on LSRFortessa (BD Biosciences, San Diego, CA, USA) and CytoFLEX LX (Beckman Coulter, Indianapolis, IN, USA) flow cytometers. FlowJo™ software version 10.7 (BD Biosciences), FlowSOM version 2.6 [73], and R version 4.0.2 were used for data analysis. Gating strategies are depicted in Figures S3 and S4. The values of the parameters were as follows for the calculation of t-SNE: iterations-1000, perplexity-30, learning rate (Eta)-478, gradient algorithm-Barnes-Hut; and, for FlowSOM, number of meta clusters: 4; set seed: 3.

Statistical Analysis
Flow cytometry analysis and animal experiments were evaluated by two-way analysis of variance and Bonferroni's post-test using Prism software, version 7 (GraphPad Software, San Diego, CA, USA). A difference of results was considered significant if p < 0.05.

Conclusions
CD80 expression on TC-1 tumor cells affected the tumor microenvironment and sensitivity to immunotherapy. CD80 deactivation in these cells was associated with a "hotter" microenvironment, decreased tumor growth, and enhanced sensitivity to CTLA-4 blockade. The impact of CD80 expression on tumor cells on the efficacy of CTLA-4 blockade has not been sufficiently investigated yet. Our study implies that CD80 expression on tumor cells should be evaluated further as a possible predictive marker that may assist clinicians in the selection of cancer patients who may be suitable for CTLA-4 blockade cancer therapy. Finally, the development of the tumor-cell-targeted CD80 blockade should be assessed as a novel immunotherapeutic approach.