Super-enhancer receives signals from the extracellular matrix to induce PD-L1-mediated immune evasion via integrin/BRAF/TAK1/ERK/ETV4 signaling

Objective: PD-L1 and PD-L2 expression levels determine immune evasion and the therapeutic efficacy of immune checkpoint blockade. The factors that drive inducible PD-L1 expression have been extensively studied, but mechanisms that result in constitutive PD-L1 expression in cancer cells are largely unknown. Methods: DNA elements were deleted in cells by CRISPR/Cas9-mediated knockout. Protein function was inhibited by chemical inhibitors. Protein levels were examined by Western blot, mRNA levels were examined by real-time RT-PCR, and surface protein expression was determined by cellular immunofluorescence and flow cytometry. Immune evasion was examined by in vitro T cell-mediated killing. Results: We determined the core regions (chr9: 5, 496, 378–5, 499, 663) of a previously identified PD-L1L2-super-enhancer (SE). Through systematic analysis, we found that the E26 transformation-specific (ETS) variant transcription factor (ETV4) bound to this core DNA region but not to DNA surrounding PD-L1L2SE. Genetic knockout of ETV4 dramatically reduced the expressions of both PD-L1 and PD-L2. ETV4 transcription was dependent on ERK activation, and BRAF/TAK1-induced ERK activation was dependent on extracellular signaling from αvβ3 integrin, which profoundly affected ETV4 transcription and PD-L1/L2 expression. Genetic silencing or pharmacological inhibition of components of the PD-L1L2-SE-associated pathway rendered cancer cells susceptible to T cell-mediated killing. Conclusions: We identified a pathway originating from the extracellular matrix that signaled via integrin/BRAF/TAK1/ERK/ETV4 to PD-L1L2-SE to induce PD-L1-mediated immune evasion. These results provided new insights into PD-L1L2-SE activation and pathways associated with immune checkpoint regulation in cancer.


Introduction
Cancer immunotherapy can be broadly divided into 4 categories, including immune checkpoint inhibitors (ICIs), chimeric antigen receptor T-cell therapy, cytokines, and vaccines. Unlike conventional therapies such as surgery, chemotherapy, and radiotherapy, cancer immunotherapy activates the body's immune system to fight cancer cells, and has become an innovative tool used to treat more than a dozen types of cancers 1-3 . Immune checkpoints are essential for maintaining immune homeostasis and preventing autoimmunity; nevertheless, they are often activated to suppress nascent antitumor immune responses 4,5 . Among the identified immune checkpoints, cytotoxic T-lymphocyte antigen-4 (CTLA-4) and programmed cell death protein 1 (PD-1) are 2 receptors that have shown promising therapeutic outcomes 6 . To escape immune cell-mediated killing, a variety of cancer cells present the PD-1 ligand 1 (PD-L1) on their surface to restrict full activation of T cells they encounter 7 . Although anti-PD-1/PD-L1 drugs provide a new breakthrough treatment with promising longer-term efficacy, their benefits to the overall population of cancer patients are quite low due to low responses and immune-related adverse events 8,9 . Thus, identifying the underlying systems that are activated by cancer cells to achieve expression of inhibitory checkpoint molecules will provide new insights that will improve responses and reduce side effects.
PD-L1, also known as cluster of differentiation 274 (CD274) or B7 homolog 1 (B7-H1), is encoded by the CD274 gene. Programmed cell death 1 ligand 2 (PD-L2), which is encoded by CD273 is homologous to PD-L1 and shows higher native binding affinity to PD-1, when compared with PD-L1 10,11 . In normal healthy conditions, PD-L1 and PD-L2 are primarily expressed on antigen-presenting cells and activated immune cells to turn-off immune responses and re-establish immune homeostasis after activation 12 . However, during oncogenesis, cancer cells organize different programs to activate PD-L1 and PD-L2 expressions 13 . For example, a multitude of cytokines in the tumor microenvironment such as IFNγ, TNFα, and IL-6 can initiate or increase PD-L1 expression 14,15 . Moreover, oncogenic pathways such as RAS, MYC, and STAT3 signaling can induce constitutive PD-L1 expression 14,15 . PD-L1 and PD-L2 expressions in cancer determine responses to ICIs, and also affect the establishment of intrinsic and acquired resistance to ICIs. Thus, fully understanding the molecular mechanisms of activation by cancer cells to induce PD-L1 and PD-L2 expressions is important in the development of accurate and individualized ICI treatments. Recently, we identified a super-enhancer (SE) named PD-L1L2-SE, which was essential for constitutive PD-L1 and PD-L2 expressions 16 . Deleting PD-L1L2-SE in cancer cells is sufficient to robustly reduce PD-L1 and PD-L2 expressions, which subsequently makes the cells sensitive to T cell-mediated killing 16 . However, the extracellular signals and critical transcription factors (TFs) that activate PD-L1L2-SE in cancer cells remain to be determined.
The ETS-TFs have a conserved ETS domain that recognizes a purine-rich core DNA sequence 17,18 . There are 28 members of the ETS family in the human genome, with most being involved in crucial biological processes such as development, differentiation, cell death, and angiogenesis. Furthermore, many ETS TFs are implicated in cancer initiation and progression. One ETS subfamily named polyomavirus enhancer activator 3 (PEA3) includes 3 homologous members: ETV1, ETV4, and ETV5 19 . Among these, ETV4 has been widely reported to play oncogenic roles in numerous cancers such as papillary thyroid carcinoma, renal cell carcinoma, gastric cancer, lung cancer, and breast cancer 20 . However, the molecular mechanism through which ETV4 promotes carcinogenesis remains largely unknown. It is also unclear whether ETV4 is involved in immune evasion and SE activation. Here, by precisely mapping the identified PD-L1L2-SE, we discovered that ETV4 was a critical TF that induced PD-L1L2-SE-mediated PD-L1 and PD-L2 expressions. ETV4 was transcriptionally activated by αvβ3 integrin through the TAK1/BRAF/ERK pathway. Importantly, cancer cells with genetic knockout or pharmacological inhibition of components of this pathway were sensitive to T cell-mediated killing.

Western blot assays
Cells were lysed in radioimmunoprecipitation assay buffer (WB3110; New Cell & Molecular Biotech, China) supplemented with protease and phosphatase inhibitors (P002; New Cell & Molecular Biotech, China). Protein concentrations were quantified using a BCA assay kit (WB6501; New Cell & Molecular Biotech, China), and the assays were read on a Beckman Coulter DU-800 (M200) spectrophotometer (Beckman, Brea, CA, USA). Equal amounts of protein were subjected to SDS-PAGE, and then transferred onto polyvinylidene fluoride membranes (GE Healthcare, Chicago, IL, USA). After blocking in 5% nonfat milk in TBST for 2 h at room temperature, the membranes were incubated with the following primary antibodies overnight at 4 °C: anti-PD-L1 (ab213524; Abcam, Cambridge, UK), anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (sc-365062; Santa Cruz Biotechnology), anti-HSP70 (sc-32239; Santa Cruz Biotechnology), and anti-HSP90 (sc-13119; Santa Cruz Biotechnology). After incubation with horseradish peroxidase-conjugated secondary antibodies for 2 h at room temperature, the blots were visualized using an ECL kit (BL520B; Biosharp, Sakai, Japan). Finally, the images were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Flow cytometry
Cells were seeded into a 6-well plate, and then treated with inhibitors for 48 h. The treated cells were collected and washed for 30 min in fluorescence-activated cell sorting (FACS) buffer [0.5% bovine serum albumin (BSA) and 2% FBS in phosphate-buffered saline (PBS)]. Phycoerythrin (PE)-conjugated anti-PD-L1 antibody and allophycocyanin (APC)-conjugated anti-PD-L2 antibody were added to the suspended cells for 30 min at 4 °C. After incubation, the cells were washed with FACS buffer (0.5% BSA in PBS) 3 times and detected using a BD Calibur (BD Biosciences, Franklin Lakes, NJ, USA) flow cytometer.

Immunofluorescence staining
When the cell density reached 60% confluence on a round glass slide in a 24-well plate, the indicated inhibitors were added for 48 h. The samples were fixed with 4% paraformaldehyde for 15 min at room temperature, treated with cell permeabilization buffer (PBS with 0.2% Triton X-100) for 10 min at room temperature, and then blocked with blocking buffer (PBS with 1% BSA) for 2 h at room temperature. Primary antibodies were incubated with the cells overnight at 4 °C in the blocking solution. After washing in PBS with 0.25% FBS, secondary antibodies including anti-rabbit Alexa Fluor 488 or 594 dye conjugate (Jackson Laboratories, Bar Harbor, ME, USA) were added and incubated for 2 h at room temperature. Finally, the cells were incubated with Hoechst dye (33342; Beyotime, Nanjing, China) in the dark at room temperature for 10 min. To measure PD-1 and PD-L1 protein interactions, the indicated cells were fixed in 4% paraformaldehyde at room temperature for 30 min and then incubated with recombinant human PD-1-Fc protein (PKSH033554; Elabscience, Houston, TX, USA) for 2 h. The secondary antibody was anti-human Fc conjugated with PE (ab99761; Abcam). Fluorescence signals were detected and captured using a fluorescent microscope (DMI3000B; Leica, Wetzlar, Germany).

Peripheral blood mononuclear cell (PBMC)mediated killing in vitro
Human PBMCs were isolated from whole blood by density gradient centrifugation using a Ficoll-Paque buffer. Isolated PBMCs were cultured in RPMI-1640 supplemented with 10% FBS and were activated by a combination of anti-CD3 antibody, anti-CD28 antibody, and IL-2. The expanded T cells were collected for killing assays. Alternatively, CD8 + T cells were positively isolated by anti-CD8-FITC microbeads from PBMCs. SUM-159 cells and their genetically modified clones were seeded into 96-well plates. SUM-159 cells were also treated with inhibitors for 24 h as indicated. Activated and expanded T cells were counted and co-cultured with the SUM-159 cells at a 5:1 ratio. After 48 h, the T cells were washed with PBS and collected by centrifugation for RT-PCR. The remaining live cancer cells were photographed and quantified using the CCK-8 assay (Sigma-Aldrich, St. Louis, MO, USA).
The ethics committees of Nantong University approved the study protocol (Approval Number: 2019-45), and the healthy donors provided written informed consent.

Statistical analysis
Two-tailed Student's t-test was used to determine the statistical significance between control and knockout cells. The statistical significance between groups treated with or without inhibitors was evaluated by a two-tailed Student's t-test. All values are presented as the mean ± standard deviation (SD). A value of P < 0.05 was considered statistically significant. *P < 0.05; **P < 0.01; ***P < 0.001.

Mapping the core DNA regulatory element in PD-L1L2-SE
In our previous study, we identified a SE (chr9: 5,495,000-5,505,000, 10 kb) called PD-L1L2-SE that was essential for constitutive PD-L1 and PD-L2 expression and PD-L1mediated immune evasion in cancer. To further characterize the DNA regulatory element in PD-L1L2-SE, we divided it into 3 subdomains (C1, C2, and C3) based on enrichment of the H3K27Ac modification and BRD4 binding (Figure 1A, 1B, and 1C). Establishment of genetic knockouts for each element in SUM-159 cells was confirmed by PCR-based genotyping and sequencing ( Figure 1D and data not shown). The joint DNA sequences after deleting each DNA element are shown in Figure 1D. We then assessed PD-L1 and PD-L2 expressions in C1-, C2-, and C3-deficient cells. RT-PCR showed that mRNA levels of both PD-L1 and PD-L2 were significantly reduced in cells with the C1 domain deletion (chr9: 5,496,378-5,499,663), but not in cells lacking the C2 or C3 elements ( Figure 1E). C1-deficient cells also had greatly reduced PD-L1 protein levels compared with control cells ( Figure 1F). The reduced PD-L1 protein expression in C1-deficient cells was confirmed by flow cytometry ( Figure 1G). Consistently, immunofluorescence also demonstrated reduced PD-L1 expression in C1-deficient cells ( Figure 1H). Together, these data showed that the C1 element (chr9: 5,496,378-5,499,663, ~3 kb) was a core region of PD-L1L2-SE that was essential for PD-L1 and PD-L2 expressions.

ETV4 drives PD-L1L2-SE-mediated PD-L1/L2 expression
To investigate TFs that were important for PD-L1L1-SEmediated PD-L1 expression, we systematically analyzed TFs that bound to the BRD4-enriched region of the C1 and C2 elements. Among all potential TFs that could bind, ETV4 had the highest score among all C1 bound TFs, while ETV4 did not bind the C2 element (Figure 2A). These results suggested that ETV4 might be important for PD-L1L2-SE-mediated PD-L1 expression. Because our previous study showed that PD-L1L2-SE was specifically activated in SUM-159 and MDA-MB-231 cells, but not MCF7 cells 16  s g V e c s g C 1 s g C 2 s g C 3 s g V e c s g C 1 s g C 2 s g C 3 (red) in control and C1-deficient cells. Nuclei were stained with 4ʹ,6-diamidino-2-phenylindole. *P < 0.05; **P < 0.01; ***P < 0.001.
To further examine the role of ETV4 in PD-L1 expression, we next established an ETV4 knockout cell line using CRISPR-Cas9 technology ( Figure 2D). Genetic disruption of ETV4 significantly reduced mRNA levels of PD-L1 and PD-L2 in SUM-159 cells ( Figure 2E). The decreased expression of PD-L1 after genetically silencing ETV4 was also confirmed by Western blot (Figure 2F). Consistently, the surface expressions of PD-L1 and PD-L2 were significantly reduced in ETV4 knockout cells as examined by flow cytometry ( Figure  2G and 2H). Taken together, these data suggested that ETV4 was essential for PD-L1L2-SE-mediated PD-L1/L2 expression in cancer cells.

ERK is required for ETV4 expression and PD-L1L2-SE activation
Because ETV4 is a TF, we next investigated which upstream pathways activated ETV4 and subsequently activated PD-L1/ L2 expression. Four inhibitors (targeting the ERK, P38, JNK, and NFκB pathways) were used to examine the possible roles of these pathways in regulating PD-L1 and PD-L2 expressions. Notably, inhibition of ERK (AZD8330) but not P38  Figure 3H). Together, these results showed that ERK was required for PD-L1/L2 expression and ETV4 activation.

BRAF and TAK1 are upstream kinases of the ERK/ETV4/PD-L1 (PD-L2) pathway
Next, we sought to determine the upstream kinases required for ERK activation and ERK-mediated PD-L1/L2 expression. Using small molecules to inhibit 3 potential kinases, TAK1 [(5z)-7-Oxozeaenol], BRAF (belvarafenib), and AKT (MK2206), we found that inhibiting TAK1 greatly reduced ERK activations in both SUM-159 and MDA-MB-231 cells ( Figure 4A). In addition, ETV4 protein levels were also reduced by inhibiting TAK1 (Figure 4B), and ETV4 mRNA levels were also significantly decreased after TAK1 inhibition ( Figure 4C). These results suggested that the reduced ETV4 protein level upon 5z7 treatment was due to decreased ETV4 transcription. As expected, inhibiting TAK1 significantly reduced PD-L1 and PD-L2 mRNA levels in both SUM-159 and MDA-MB-231 cells (Figure 4D and 4E). Western blot analysis showed that PD-L1 protein levels were dramatically decreased after TAK1 inhibition in both SUM-159 and MDA-MB-231 cells (Figure 4F). Immunofluorescence also showed greatly reduced PD-L1 expression in SUM-159 cells ( Figure 4G). These results strongly indicated that TAK1 was an upstream regulator of ERK activation, ETV4 transcription, and PD-L1 expression.
Moreover, we also found that inhibiting BRAF but not AKT significantly reduced PD-L1 and PD-  immunofluorescence results (Figure 5D). BRAF inhibition also blocked ERK activation (Figure 5E), suggesting that ERK was downstream of BRAF. Moreover, BRAF inhibition significantly reduced ETV4 mRNA expression in both SUM-159 and MDA-MB-231 cells (Figure 5F). The reduced ETV4 mRNA levels following BRAF inhibition is consistent with reduced ETV4 protein levels after BRAF inhibition ( Figure 5G). Together, these results showed that BRAF was required for ERK activation as well as for ETV4 and PD-L1 expressions, and provided compelling support for the  critically important role of TAK1 and BRAF in ERK activation, ETV4 transcription, and PD-L1 expression.
cell types (Figure 6F). Flow cytometry further confirmed reduced surface expression of PD-L1 and PD-L2 after inhibiting integrin signaling ( Figure 6G). Taken together, these findings suggested a critical role for integrins in activating ERK and downstream ETV4-mediated PD-L1/L2 expression.

Targeting the αvβ3 integrin/ERK/ETV4/PD-L1 (PD-L2) pathway blocks cancer cell immune evasion
Although immune checkpoint blockade (ICB) targeting PD-L1/PD-1 with antibodies has achieved impressive clinical success, ICB strategies still need to be improved. First, we measured the interaction between PD-1 and PD-L1 proteins by using recombinant human PD-1-Fc protein as well as antihuman Fc antibodies-conjugated with PE. Figure 7A shows that fluorescence images clearly demonstrated the engagement of PD-1 and PD-L1 in sgC1 and sgETV4 cells was greatly reduced, when compared to that in control sgVec cells ( Figure 7A). Next, we tested whether targeting the identified pathway that drives SE-mediated PD-L1 expression could enhance anti tumor immune reactions, by performing in vitro T cell-mediated killing assays. Figure 7B shows a significant portion of control cells (sgVec) survived during co-culture with activated T cells, while genetic knockout of the C1 element (sgC1) or ETV4 (sgETV4) made the cells sensitive to T cell-mediated killing (Figure 7B and 7C). Consistent with the role of PD-L1, T cells co-cultured with sgC1 or sgETV4 cells were more activated than T cells co-cultured with sgVec cells, as shown by increased IFNγ and granzyme B production ( Figure 7D). Additionally, SUM-159 cells were sensitive to T cell-mediated killing after being treated with the ERK inhibitor, AZD8330, indicating the critical role of ERK signaling in PD-L1 expression (Figure 7E and 7F). Inhibiting TAK1, BRAF, or integrin also rendered SUM-159 cells sensitive to T cell-mediated killing (Figure 7G and 7H).
These results further supported the conclusion that the integrin/TAK1/BRAF/ERK/ETV4 signaling axis was involved in SE-mediated PD-L1 expression. Thus, targeting this pathway could achieve immune cell-mediated cancer cell death. These results therefore suggested that the identified pathway could be a potential target for ICB.

Discussion
ICB targeting PD-L1/PD-1 has been used in a broad spectrum of cancers to renormalize and reset antitumor immunity 21 .
Although its overall therapeutic effect is promising and some patients show long-term remission, ICB faces several major challenges such as low response and adverse effects 22 . Antibody-based ICB can effectively block PD-L1 on the cell surface, but intracellular PD-L1 likely cannot be targeted. These issues should be seriously considered, as several recent reports have demonstrated a nuclear function of PD-L1 [23][24][25] .
In addition, disrupting PD-L1 on both cancer and normal cells also blocks the beneficial functions of PD-L1 and leads to side effects. Thus, it is ideal to target a unique pathway through which cancer cells, but not normal cells, induce PD-L1 expression. Previously, we found that PD-L1L2-SE was specifically activated in cancer cells, where it was required for PD-L1 expression 16 . Here, we further mapped the core region of PD-L1L2-SE and showed that the TF ETV4 was important for PD-L1 expression (Figure 1 and Figure 2). Furthermore, ETV4 and PD-L1 were activated by the extracellular matrix (ECM) through the integrin/BRAF/TAK1/ERK pathway (Figures 3-6). Disrupting this pathway by either genetic knockouts or inhibitors made cancer cells sensitive to T cell-mediated killing (Figure 7). These findings indicated that PD-L1L2-SE itself as well as its related pathways might be potential pharmacological targets to specifically disrupt PD-L1 in cancer cells. The ECM and its resident cells such as endothelial, mesenchymal, and immune cells comprise the tumor microenvironment 26 . It is now recognized that the tumor microenvironment plays a pivotal role in tumor growth and metastasis, and also determines responses to antitumor therapies including ICB 27 . In many malignancies, the ECM from tumor cells and cancer-associated fibroblasts compromise up to 60% of the tumor mass 28 . Various ECM molecules such as fibrillar collagens, fibronectin, elastin, hyaluronan, and laminins are frequently observed in solid tumors 29 . Cancer cell-matrix adhesion is mediated by interactions between membrane-associated integrins and the arginine-glycineaspartic (RGD) motif displayed on many ECM proteins. In the present study, we found that Cyclo (RGDfK), which is a potent inhibitor of αvβ3 integrin, significantly diminished PD-L1 and ETV4 expressions as well as immune evasion (Figure 6). This is consistent with a recent report that showed αvβ3-integrin regulated PD-L1 expression and was involved in immune evasion 30 . Although αvβ3-integrin is required for SE-mediated PD-L1/L2 expression, it might be insufficient to activate PD-L1L2-SE because αvβ3-integrin activation is widely observed in cultured cell lines, while PD-L1L2-SE-induced PD-L1/L2 expression is less frequently observed. Therefore, besides αvβ3-integrin, other factors or downstream pathways might collaboratively induce PD-L1L2-SE. In our investigation of pathways downstream of αvβ3-integrin, we unexpectedly found that both TAK1 and BRAF were required for ETV4 activation and PD-L1 expression (Figure 3 and Figure 4). MEK inhibition in combination with anti-PD-L1 promotes T cell infiltration and anti-tumor activity, however the effect of MEK inhibition on PD-L1 expression is inconsistent 31,32 . Here, we found that ERK activation was required for ETV4 expression and PD-L1L2-SE activation ( Figure 5). This may partially explain the synergic effects between MEK inhibitors and anti-PD-L1 antibodies. TAK1 and BRAF are both mitogen-activated protein kinase kinases (MAP3Ks). Inhibition of TAK1 or BRAF could efficiently block ERK activation and downstream PD-L1 expression. It is interesting that cancer cells need both MAP3Ks to activate SE-mediated PD-L1 expression. However, further studies need to address the distinct roles of TAK1 and BRAF in PD-L1L2-SE activation.
By analyzing the TFs that bind to a core region of PD-L1L2-SE, we found a potentially important TF (ETV4) for PD-L1L2-SE-mediated PD-L1 expression. Molecular studies have confirmed the role of ETV4 in PD-L1 expression (Figure 2). ETV4 is a TF that is known to be involved in tumor development and progression 20 . Our results suggested an important role for ETV4 in immune evasion by activating PD-L1. Although we observed higher ETV4 protein levels in SUM159 and MDA-MB-231 cells compared with MCF7 cells, MCF7 cells expressed ETV4 but not PD-L1. Thus, it was possible that other TFs cooperatively worked with ETV4 to activate PD-L1L2-SE.