STAT1 is regulated by TRIM24 and promotes immunosuppression in head and neck squamous carcinoma cells, but enhances T cell antitumour immunity in the tumour microenvironment

Background Head and neck squamous cell carcinoma (HNSCC) is a significant problem and is frequently resistant to current treatments. STAT1 is important in anti-tumour immune responses against HNSCC. However, the role of STAT1 expression by tumour cells and its regulation during HNSCC is unclear. Methods We determined the effects of STAT1 inhibition on tumour development and immunity in CAL27 and UMSCC22A HNSCC cell lines in vitro and in a HNSCC carcinogen-induced model in vivo. Results STAT1 siRNA knockdown in human HNSCC cells impaired their proliferation and expression of the immunosuppressive marker PD-L1. Stat1-deficient mice displayed increased oral lesion incidence and multiplicity during tumour carcinogenesis in vivo. Immunosuppressive markers PD-1 in CD8+ T cells and PD-L1 in monocytic MDSCs and macrophages were reduced in oral tumours and draining lymph nodes of tumour-bearing Stat1-deficient mice. However, STAT1 was required for anti-tumour functions of T cells during HNSCC in vivo. Finally, we identified TRIM24 to be a negative regulator of STAT1 that plays a similar tumorigenic function to STAT1 in vitro and thus may be a potential target when treating HNSCC. Conclusion Our findings indicate that STAT1 activity plays an important role in tumorigenicity and immunosuppression during HNSCC development.


BACKGROUND
Head and neck squamous cell carcinoma (HNSCC) is the sixth most common form of cancer, affecting nearly one million people per year [1,2]. With the annual burden of HNSCC expected to increase, there exists a dire need to identify and elucidate the context-dependent function of molecular targets that can be therapeutically exploited in cancer cells and associated cells of the HNSCC tumour microenvironment [3]. Signal transducer and activator of transcription 1 (STAT1) mediates the signalling of type I, II and III interferons (IFNs) and growth factors like epidermal growth factor and is involved in several biological processes, including apoptosis [4], inflammation [5] and cell cycle modulation [6,7]. Not surprisingly, due to the wide range of biological activity mediated by STAT1, its activity is heavily altered in many HNSCC cases [8]. Interestingly, STAT1 downregulation has been implicated in oral carcinogenesis and immune escape, while STAT1 phosphorylation has been associated with increased survival [9][10][11].
Recent studies conducted by our group identified the critical anti-tumour role of STAT1 in the host immune response to HNSCC [12]. We used an orthotopic model of metastatic HNSCC cells injected into Stat1-deficient (Stat1 −/− ) or Stat1-sufficient (Stat1 +/+ ) mice to show that mice deficient in Stat1 are highly susceptible to tumour growth and metastasis. This was associated with an exhausted T cell phenotype with increased programmed cell death protein 1 (PD-1) expression, and increased accumulation of immunosuppressive myeloid-derived suppressor cell (MDSC) populations in the tumours of Stat1-deficient mice. These results indicate that host derived STAT1 is critical for anti-tumour immune responses against HNSCC. However, questions remain regarding the role of STAT1 expression by oral epithelial and cancer cells during stages of HNSCC tumour development.
Although STAT1 has long been considered to be a tumour suppressor by promoting HNSCC cell death [10], recent reports show that STAT1 activation also mediates programmed cell death ligand 1 (PD-L1) expression in HNSCC cells thereby promoting tumour immunosuppression [13,14]. STAT1 activation has also been shown to mediate resistance to cisplatin therapy [15] and radioresistance in HNSCC cells [16]. A limitation to these studies that have investigated the impact of STAT1 activation in HNSCC cells are that the methods mostly used are in vitro cell culture models, xenograft models into nude mice or correlative studies from HNSCC clinical samples. These approaches generally do not address (i) the influence of STAT1 on the tumour microenvironment, (ii) the interaction of STAT1 expressing HNSCC cells with immune cells, stromal cells, and fibroblasts and (iii) the impact of STAT1 activation by oral epithelial cells on HNSCC tumour initiation and promotion. This represents a major gap in our knowledge of STAT1 activation by HNSCC cells in the tumour microenvironment, which is essential to advance our mechanistic understanding of the role of STAT1 in HNSCC, and design therapies that exploit the STAT1 pathway in HNSCC treatment.
Identifying cell-specific regulators of STAT1 activation in HNSCC cells can facilitate the development of novel approaches to HNSCC treatment. In this study, we identify a novel potential negative regulator of STAT1 in HNSCC cells-tripartite motif (Trim) 24 protein, also known as transcriptional intermediary factor 1. TRIM24 is a multifunctional protein that binds and regulates nuclear receptors, such as retinoic acid and oestrogen receptors, and associates with chromatin-and heterochromatin-associated factors that result in inactivation of Stat1 transcription [17]. Thus, TRIM24 presents itself as a potential target for modulating STAT1 expression in the tumour microenvironment. Although TRIM24 has been shown to promote colorectal, ovarian and other cancers, [18][19][20][21], the effect of the TRIM24-STAT1 axis on HNSCC development still remains unclear.
To enhance our understanding of STAT1 activation and regulation in HNSCC initiation, growth and anti-tumour immune responses, we determined the effects of STAT1 small interfering RNA (siRNA) knockdown on the human papillomavirus (HPV)negative oral cancer cell lines CAL27 and UM-SCC22A. We also used the 4-nitroquinoline-1-oxide (4NQO) carcinogen-induced HNSCC model in STAT1-competent and STAT1-deficient mice to determine the effect of STAT1 on tumour development and immune responses to HNSCC in vivo [22]. This 4NQO mouse model recapitulates the mutational profile and immune landscape of non-HPV-associated human HNSCC and mirrors its development from early dysplastic to invasive stages [23]. Specifically, high incidence mutations observed in human HNSCC, such as in Trp53, Notch1, Fat1, Lama3 and Syne2, were demonstrated to be associated with 4NQO-induced mouse tumours, while immune infiltration to tumours was maintained [23]. Indeed, STAT1 and IFN-γ activity are upregulated during PD-1 blockade in mice administered 4NQO [24]. Because of the similar mutational landscape of 4NQO-induced mouse oral carcinogenesis to human HNSCC, combined with the use of immunocompetent hosts, this model further allows for study of the immunologic determinants of HNSCC development [25]. Finally, we investigate the impact of TRIM24 siRNA knockdown on STAT1 expression and subsequent cell proliferation, apoptosis and expression of immunosuppressive biomarkers in CAL27 and UM-SCC22A HNSCC cells. The findings of our study shed light on the role of STAT1 expression and regulation by HNSCC cells on tumour development and the immune response during HNSCC.

MATERIALS AND METHODS Mice
Stat1 +/+ and Stat1 −/− male and female BALB/c mice, 6-8 weeks old (N = 20, 10 males, 10 females per group) were used for experimental HNSCC studies. This sample size was selected to achieve 80% power to detect a 2.5-fold difference between groups (corresponding to an effect size of 2). Animals were kept according to regulations maintained by the University Laboratory Animal Resources. Experiments were approved by the Institutional Animal Care and Use Committee (Protocol #2018A00000054) and Institutional Biosafety Committee of The Ohio State University.

siRNA knockdown
A reverse transfection mRNA knockdown cocktail containing STAT1 siRNA (Cell Signaling; Danvers, MA), TRIM24 siRNA (Life Technologies; Carlsbad, CA) or scrambled siRNA (Thermo Fisher; Waltham, MA) and 1 μL of Lipofectamine RNAiMAX in 100 μL OptiMEM serum-free media (Life Technologies; Carlsbad, CA) was prepared and plated at 100 μL per well. Following 15 min of room temperature incubation, cells were plated at 2.5 × 10 5 cells/well in 500 μL antibiotic-free complete media in each well. Cells were allowed to grow for 48 or 72 h before downstream analyses.

Carcinogen-induced HNSCC model
Stat1 +/+ and Stat1 −/− (n = 20; 6-8 weeks old) BALB/c mice were administered the oral carcinogen 4NQO (Sigma Aldrich; St. Louis, MO) at a concentration 100 μg/mL in their drinking water for 16 weeks, followed by drinking water for 8 weeks [28,29]. Control mice of each genotype were administered drinking water for the entire duration of the experiment. At 24 weeks following initial 4NQO treatment, mice were sacrificed. Tongue lesions and tumours were counted macroscopically and microscopically.

Flow cytometry
Cells were stained with extracellular antibodies prior to fixation in 4% neutral-buffered formalin and permeabilisation. Fluorochrome-conjugated antibodies against CD4, CD8, CD11b, Ly6C, Ly6G, PD-1 and PD-L1 were used for extracellular staining. Intracellular staining was performed with antibodies against Gzmb, IFN-γ, pSTAT1 Y701 (Biolegend) and TRIM24 (AbClonal; Woburn MA). Mouse IgG1κ and Rat IgG1κ were used as isotype controls for Gzmb and IFN-γ, respectively. TRIM24 primary antibodies were further incubated with goat anti-rabbit FITC-conjugated secondary antibodies (BD Biosciences; San Jose, CA) prior to acquisition. Other samples were treated with propidium iodide (PI) and FITC-conjugated Annexin V for 15 min for apoptosis analysis. Samples were analysed using a FACS Celesta flow cytometer (BD Biosciences; San Jose, CA). Analysis was performed using the FlowJo software (FlowJo; Ashland, OR).

Statistics
Analysis was performed in a blinded fashion. Statistical analyses were performed using GraphPad Prism v9.0 (GraphPad Software; San Diego, CA). Analysis of variance or Student's t test (two-sided) was used to determine statistically significant differences between groups.

STAT1 promotes expression of PD-L1 and proliferative biomarkers in HNSCC cells
To explore the role of STAT1 in HNSCC, we performed siRNA mediated knockdown of Stat1 expression in the human HNSCC cells. We used the CAL27 and UM-SCC22A cell lines because they are HPV negative HNSCC cells that express PD-L1 and STAT1. We first confirmed downregulation of Stat1 mRNA transcripts and total STAT1 protein in both cell lines using RT-qPCR and western blot (Fig. 1a, b). Further, in CAL27 cells, STAT1 siRNA knockdown decreased STAT1 phosphorylation compared to scrambled RNA treated cells. We did not observe any differences in STAT1 phosphorylation between STAT1 siRNA and scrambled RNA treated UM-SCC22A cells (Fig. 1a, b). Next, we determined the effects of STAT1 knockdown on immune suppression, apoptosis and biomarkers of cell proliferation. We first examined the effects of Stat1 silencing on HNSCC cell PD-L1 expression by flow cytometry. We observed that PD-L1 surface expression was significantly downregulated following Stat1 knockdown in both cell lines, indicating an essential role for STAT1 in promoting PD-L1 surface expression on HNSCC cells (Fig. 1c, e). Notably, despite the low constitutive expression of PD-L1 by UM-SCC22A cells, STAT1 knockdown eliminated the expression of this immunosuppressive receptor on this HNSCC cell line (Fig. 1e). Our data demonstrate that STAT1 promotes PD-L1 expression on HPV negative HNSCC cells in vitro. We further characterised the relationship between STAT1 and PD-L1 expression using the publicly available cancer genome atlas (TCGA) database of HNSCC patients. We observed a striking positive correlation between STAT1 and PDL1 in tumours of HNSCC patients (Fig. 1g), further supporting our hypothesis of PD-L1 regulation by STAT1 in HNSCC.
To examine the effects of STAT1 on apoptosis, we utilised Annexin-V/PI staining on cell populations 48 and 72 h following STAT1 siRNA knockdown. We observed a significantly higher percentage of early apoptotic Annexin-V + I − cells in STAT1deficient compared to STAT1-sufficient CAL27 and UM-SCC22A cell lines (Fig. 1d, f). Finally, we examined the effects of STAT1 on biomarkers of cell proliferation (Ki67 and Mcm3) by RT-qPCR [30]. We observed decreased expression of both markers following STAT1 knockdown in CAL27 (Fig. 1h). In UM-SCC22A, we observed a similar downregulation of these proliferation markers at 48 h following STAT1 inhibition (Fig. 1i). Taken together, these data indicate that STAT1 expression by HNSCC cells promotes PDL1 expression, increases proliferative potential, and inhibits apoptosis, thereby enhancing the tumorigenicity of HNSCC cells.
Global STAT1 deficiency in a 4NQO model of HNSCC results in increased incidence of carcinogenesis Previous work from our group demonstrated that in STAT1-sufficient murine HNSCC cells orthotopically injected into Stat1 +/+ and Stat1 −/− mice, tumour growth is significantly enhanced when Stat1 is deleted in host tissue [12]. Therefore, to determine the impact of tumour-derived STAT1 on HNSCC tumour development in vivo, we used the oral carcinogen 4NQO to induce oral carcinogenesis in both Stat1 +/+ and Stat1 −/− mice (Fig. 2a). We chose this model because it recapitulates the initiation, promotion and progression stages of carcinogenesis in clinical cases of non-HPV associated HNSCC. First, we observed increased mortality in carcinogen induced Stat1 −/− compared to Stat1 +/+ mice (Fig. 2b). At terminal sacrifice, we also noted higher incidence and multiplicity of both lesions and tumours in Stat1 −/− mice ( Fig. 2c-e). Our findings were confirmed by H&E staining of the carcinogen-induced tongue tissue (Fig. 2f). To determine whether global STAT1 deficiency affects proliferation and apoptosis of tumour cells in vivo, we measured Ki-67 and caspase 3 expression in tumours by RT-qPCR and immunohistochemistry. We also measured Aurka and Bcl2 gene expression levels. Overall, proliferative and apoptotic markers were not affected by STAT1 deficiency in vivo ( Fig. 2g-i). Taken together, our results indicate that despite the pro-tumorigenic properties of STAT1 in HNSCC tumour cells, global STAT1 deficiency results in higher tumour burdens in the carcinogen-induced model of HNSCC carcinogenesis. Although these results mimic our observations in the orthotopic model, the difference in tumour development between Stat1 +/+ and Stat1 −/− mice in the 4NQO model was not as high as in the orthotopic model. These results demonstrate that, although STAT1 expression by immune cells is essential for an antitumour response, STAT1 expression by tumour cells might influence tumour development to a lesser extent than in immune cells, but in a manner that suggests a tumour promoting role in vivo.
STAT1 promotes PD-1/PD-L1 expression in the tumour microenvironment during HNSCC Given the observed role for STAT1 in PD-L1 expression by HNSCC cells in vitro, we next explored the impact of STAT1 on PD-1 and PD-L1 expression in the tumour microenvironment during HNSCC carcinogenesis in vivo. We observed significantly reduced gene expression and cell protein expression of Pd1 at the tongue tissues of carcinogen induced Stat1 −/− mice, indicating decreased PD-1 immune checkpoint activity (Fig. 3a). We had previously observed that PD-1 was strongly induced in T-lymphocytes of tumour-bearing Stat1 −/− mice orthotopically injected with Stat1 competent aggressive HNSCC [12]. Therefore, we examined PD-1 expression in T lymphocytes of carcinogen-induced Stat1 +/+ and Stat1 −/− mice. In contrast with Stat1 −/− orthotopic tumour-bearing mice, PD-1 expression was reduced in CD8 + T cells in the tumours, lymph nodes and spleens of carcinogen-induced Stat1 −/− mice compared to Stat1 +/+ mice (Fig. 3b, d). PD-1 expression in CD4 + T cells was reduced in spleens, but not tumours and draining lymph nodes of carcinogen-induced Stat1 −/− mice compared to Stat1 +/+ mice (Fig. 3c, e). Combined with previous studies, these data suggest that during HNSCC, induction of PD-1 expression on T cells is partly dependent on STAT1 expression by tumour cells.

STAT1 is required for anti-tumour function by T-lymphocytes during HNSCC carcinogenesis
Despite the increased PD-1/PD-L1 expression observed in carcinogen induced Stat1 competent mice, tumour development was still lower than in Stat1 −/− mice. Therefore, we next examined potential STAT1 mediated anti-tumour mechanisms. Specifically, we measured IFN-γ, GranzymeB, and CXCL9 [31,32], markers known to be regulated by STAT1 and are associated with T cell anti-tumour activity, effector responses and chemotaxis of anti-tumour effector T cells. At the primary tumour site, we found that Gzmb, Ifng, and Cxcl9 transcripts were significantly reduced in carcinogen-induced Stat1 −/− compared to Stat1 +/+ mice (Fig. 4a). ELISA analysis of tongue lysates and immunohistochemical staining of tongue lesion sites further corroborated the decreased IFN-γ, GranzymeB and Cxcl9 protein expression in Stat1 −/− mice (Fig. 4b-d). Given the decreased anti-tumour activity in the STAT1-deficient tumour microenvironment, we analysed T-cell populations of the oral draining lymph nodes and spleens of experimental mice. At the lymph nodes, Gzmb expression by CD4 + and CD8 + lymphocytes were decreased in Stat1 −/− mice (Fig. 4e-g). IFN-γ production was decreased in splenic CD8 + T cells and NK cells of the lymph nodes and spleen of carcinogen-induced Stat1 −/− mice compared to carcinogen-induced Stat1 +/+ mice (Fig. 4h-j and Supplementary Fig. 1). These results demonstrate the importance of STAT1 in maintaining an effective anti-tumour immune response and immunosurveillance during HNSCC, and explain potential mechanisms underlying the increased tumour burden observed in Stat1 −/− mice compared to their Stat1 +/+ counterparts.

TRIM24 is selectively expressed in HNSCC cells and negatively regulates STAT1
We next investigated potential STAT1 modulators that could selectively impact STAT1 signalling in HNSCC cells. TRIM24 has previously been shown to be over expressed in advanced HNSCC and is associated with HNSCC metastasis [19]. Pathway analysis of RNA sequencing data from tumours from Stat1 +/+ and Stat1 −/− mice orthotopically injected with LY2 HNSCC cells also identified TRIM24 as the top predicted negative transcriptional regulator of STAT1 during HNSCC in our data set (z score -5.804, p value 1.32 × 10 −16 ) (Fig. 5a, b). Mutational analysis of these orthotopically injected tumours in Stat1 +/+ and Stat1 −/− mice from sequencing data are shown in Supplementary Fig. 2. Analysis of 519 HNSCC patients from the cancer genome atlas (TCGA) did not show a real inverse correlation between TRIM24 and STAT1 expression (r = −0.1764, p = 7.05 × 10 −5 ) (Fig. 5c). However, in STAT1-knockdown CAL27 HNSCC cells, gene and protein expression of TRIM24 is slightly, but significantly increased (Fig. 5d, e). Interestingly, in our orthotopic HNSCC model, we found that TRIM24 expression is high in HNSCC cells but not in tumour-infiltrating T lymphocytes (Fig. 5f, g), suggesting the potential of targeting TRIM24 to modulate STAT1 signalling specifically in HNSCC cells but not T cells. Therefore, we investigated the effects of TRIM24 in human HNSCC cells using TRIM24 specific siRNA knockdown. We first verified knockdown of 14 15 56.4 76.8 TRIM24 gene and protein expression by RT-qPCR and Western blot (Fig. 5h, i). Next, we determined the effect of TRIM24 on STAT1 expression, and we observed a marked increase in STAT1 transcripts (Fig. 5h), total STAT1 and phosphorylated STAT1 protein ( Fig. 5i and Supplementary Fig. 3). These data strongly implicate TRIM24 as a negative regulator of STAT1 production and activity in HNSCC cells. These results were further corroborated in UMSCC22A cells, although p-STAT1 protein expression was unchanged in TRIM24 knockdown cells (Fig. 5j, k).
Effects of TRIM24 on HNSCC cells are mediated by STAT1dependent and independent mechanisms Given the co-inhibitory effects of TRIM24 and STAT1 expression in HNSCC cells, (Fig. 6a, b), we further explored the interaction between TRIM24 and STAT1 on HNSCC cell PD-L1 expression, apoptosis and cell proliferation. Interestingly, despite the increase in STAT1 induced by TRIM24 knockdown, we observed that TRIM24 silencing inhibited the tumour promoting effects and enhanced the tumour inhibitory effects of STAT1 activation. Specifically, while we showed STAT1 to be important for PD-L1 expression, we surprisingly saw that TRIM24 knockdown resulted in an even more significant decrease in PD-L1 surface expression in CAL27 cells compared to STAT1 siRNA treated cells (Fig. 6c). Similarly, TRIM24 knockdown significantly decreased the expression of the proliferation markers KI67 and MCM3, which we had showed to be decreased by STAT1 inhibition (Fig. 6d, e and Supplementary Fig. 3). This was also true of the increased expression of the apoptotic markers CASP8 and BCL2 in TRIM24 knockdown CAL27 and UMSCC22A cells (Fig. 6d, e and Supplementary Fig. 3). These results indicate that TRIM24, while a negative regulator of STAT1 in HNSCC cells, inhibits the tumour-promoting effects and enhances the tumour-inhibitory effects of STAT1 activation. Next, we performed double knockdown experiments of STAT1 and TRIM24 to determine mechanisms of TRIM24 effects on HNSCC cells and the involvement of STAT1 activity. We determined the expression of PDL1 on TRIM24 and STAT1 single or double knockdown CAL27 cells. We observed that knockdown of STAT1 abrogated the effects of TRIM24 on PDL1 expression in these cells. On the other hand, knockdown of TRIM24 did not significantly affect PDL1 expression on STAT1 knock down cells (Fig. 6c). This suggests that the effects of TRIM24 on PDL1 expression in CAL27 HNSCC cells are STAT1 dependent. We also examined the impact of TRIM24 and STAT1 double knockdown HNSCC cells on markers of apoptosis and cellular proliferation. We observed similar or enhanced reduction of the proliferation markers Ki67 and MCM3 in double knockdown compared to single knock down CAL27 and UMSCC cells (Fig. 6d, e and Supplementary Fig. 3). Taken together, our results demonstrate a unique interaction between TRIM24 and STAT1 in HNSCC cells. Based on these findings, we propose a model demonstrating the divergent roles of STAT1 signalling in cancer cells and immune cells of the tumour microenvironment (Fig. 6f). We also demonstrate the interactions of TRIM24 as an upstream regulator of STAT1, inhibiting STAT1 expression and phosphorylation in HNSCC cells. Conversely, STAT1 inhibits TRIM24 expression, although to a lesser extent (Fig. 6f). Further we demonstrate that the effects of TRIM24 on immunosuppression, apoptosis and proliferation of HNSCC cell are mediated by STAT1-dependent and independent mechanisms.

DISCUSSION
This study highlights new mechanistic insights into the role of STAT1 expression by HNSCC cells and immune cells, as well as the function of a novel STAT1 negative regulator TRIM24, in HNSCC cells. Using RNAi silencing technology, we found that STAT1 plays an essential role in PD-L1 expression by HNSCC cells. We further found, using the 4NQO murine model of HNSCC, that STAT1 signalling is essential for PD-1/PD-L1 expression in the tumour microenvironment and by immune cells, including the accumulation of immunosuppressive populations during head and neck carcinogenesis. We further demonstrate an important role for STAT1 in anti-tumour immunity by T cells, which supports our previous reports [12]. Lastly, we show a negative co-regulatory function between STAT1 and TRIM24, which together induce PD-L1 expression in HNSCC cells (Fig. 6f).
The results of our study indicate a pro-tumour role for tumour-derived STAT1 expression in HNSCC based on its ability to enhance PD-L1-mediated immunosuppression and proliferation in both UM-SCC22A and CAL27 HNSCC cell lines. The role of STAT1 in regulating apoptosis, however, appears to be contextdependent. A downregulatory effect of STAT1 on the antiapoptotic protein BCL2 has been previously established, although not in HNSCC cells [33]. We show that following STAT1 silencing in CAL27 cells, BCL2 expression is highly upregulated, showing that STAT1 maintains its downregulatory effect on BCL2 among the HNSCC cells tested. On the other hand, STAT1 activity appears to induce caspase-8 expression [34]. However, the non-canonical elevation of Casp8 expression we observed in CAL27 cells following STAT1 inhibition warrants further investigation, as this may be a driver behind tumour survival in cases of HNSCC with high expression of STAT1.
The role of tumour-derived STAT1 in promoting of PD-L1 expression, however, is more evident. Previously, our group observed that STAT1-competent HNSCC induced greater expression of both PD-1 and PD-L1 in immune cells of Stat1 −/− mice compared to Stat1 +/+ in both in situ and metastatic orthotopic models of HNSCC [12]. Our current results show that global STAT1 deficiency in a carcinogen-induced model of HNSCC leads to decreased PD-1/PD-L1 expression in the tumour microenvironment and by immune cells. In light of the finding that PD-1/ PD-L1 expression STAT1-deficient immune cells were elevated in the context of STAT1-competent tumours but not STAT1deficient tumours, it appears that STAT1 expression by tumour cells plays a direct role in inducing immune checkpoint signalling between immune cells. Moreover, host immune cells do not require STAT1 to express high levels of PD-1 or PD-L1. Taken together, these findings indicate that PD-1/PD-L1 activity in immune cells is dependent on STAT1 expression by tumour cells during HNSCC but may be independent of STAT1 expression by immune cells.
It was surprising to observe a reduced accumulation of immunosuppressive myeloid populations in carcinogen induced Stat1-deficient mice compared to Stat1-sufficient mice. This contrasts with our previous findings where these populations significantly accumulate in Stat1-deficient mice bearing Stat1 competent aggressive HSNCC tumours [12]. While it has been shown that STAT1 promotes M2 macrophage activity and development [35,36], we previously observed that these populations also accumulate in Stat1-deficient mice. Our observation that these populations are reduced in the bone marrow of carcinogen induced Stat1 −/− mice implies that tumour-derived STAT1 plays a major role. Notably, we observed a significant decrease in Ly6C hi Ly6G− cells, but not Ly6C int Ly6G + cells, in the lymph nodes and spleens of carcinogen induced Stat1-deficient mice, which is another population we previously observed to accumulate in Stat1-deficient mice bearing Stat1-competent aggressive HNSCC. Together, these results highlight the important role of tumour-derived STAT1 in promoting the accumulation of immunosuppressive myeloid populations, especially when these host derived immune cells are Stat1 deficient. Because M-MDSCs are known to facilitate the epithelial-to-mesenchymal transition that precedes metastasis [37], tumour-derived STAT1 may be a predictor of immune suppression and metastasis in HNSCC. Alternatively, MDSC populations have been shown to expand in response to chronic inflammation [38,39], and their deficiency may be due to the failure of Stat1 −/− mice to initiate a robust anti- tumour response during the early stages of head and neck carcinogenesis [40]. Despite the reduction in the immunosuppressive PD-1/PD-L1 axis and MDSC populations observed in carcinogen induced Stat1deficient mice, they still displayed significantly higher mortality and burden of dysplasia and neoplasia compared to Stat1 competent mice. This finding may be explained by our findings of decreased expression of Gzmb and IFN-γ by lymphocytes in Stat1 −/− mice, which are known to facilitate the anti-tumour immune response to HNSCC and inhibit tumour proliferation [41][42][43]. Initially, this observation may seem to contradict the heightened PD-1/PD-L1 signalling which is known to inhibit lymphocyte anti-tumour functions [14]. However, the global reduction of Gzmb in Stat1 −/− mice can be explained by the role STAT1 plays in its production in CD8 T cells [44]. Alternatively, decreased PD-1 expression by T-cells may be resultant of decreased activation rather than a direct result of STAT1 deficiency. Further, absent IFN-γ signal transduction mediated by STAT1 leads to failure to prepare immune cells for sufficient anti-tumour immune activity, as evidenced by decreased IFN-γ expression in the tumour microenvironment. However, we also observed a corresponding decrease in PD-L1 in the tumour microenvironment, which corresponds to the role of IFN-γ and STAT1 signalling in the production of PD-L1. Thus, while immune cell STAT1 activity is strongly associated with HNSCC survival and can be considerably variable among patients [11], our data supports the hypothesis that STAT1 activation by HNSCC cells could simultaneously potentiate immunosuppressive and tumorigenic effects during HNSCC. These data taken together indicate that while impaired STAT1 signalling within lymphoid populations is likely to produce a deleterious effect in HNSCC, selective targeting of STAT1 signalling in HNSCC cells could potentially enhance antitumour immune responses and inhibit tumour cell growth.
TRIM24 is a transcriptional intermediary factor that we have found to negatively correlate with STAT1 expression. Thus TRIM24 presents itself as an interesting target for modulating STAT1 expression and downstream effects in HNSCC. Previously, TRIM24 has been shown to bind to retinoic acid responsive elements of the STAT1 promoter, thereby repressing STAT1 transcription [17,19]. However, there has been a long-standing debate on whether TRIM24 functions as either a tumour suppressor or oncogene, and the answer appears to be context-dependent [18,45]. Notably, TRIM24 expression has been shown correlate with both locally advanced and metastatic HNSCC, though the mechanisms underlying this effect remain unclear [19,46]. Although we did not observe a strong correlation between STAT1 and TRIM24 based on TCGA data, our results demonstrate that TRIM24 negatively regulates STAT1, and we present novel evidence that STAT1 also negatively regulates TRIM24 in HNSCC cells.
Despite its inhibitory effect on STAT1 expression, we also found TRIM24 to have a significant impact on the immunosuppressive ability of HNSCC cells, indicating a complex regulatory interaction beyond simple co-inhibitory effects. For example, while STAT1 is involved in PD-L1 expression, TRIM24 knockdown in HNSCC cells significantly inhibited PD-L1 expression to an even greater extent than observed in STAT1 knockdown HNSCC cells. This finding was particularly surprising given the observed STAT1 mediated promotion of PD-L1 expression and the downregulatory effect of TRIM24 on STAT1 protein expression and phosphorylation. These results appear to highlight the complex crosstalk between TRIM24 and STAT1 to induce PD-L1 expression in HNSCC. This is further demonstrated by our observation that STAT1 depletion in TRIM24 knockdown cells partially restored PD-L1 expression levels, indicating that TRIM24 effects of PD-L1 expression on HNSCC cells are at least partly mediated by STAT1. Potential transcriptional, translational and posttranslational mechanisms underlying the complex interplay between TRIM24 and STAT1 in mediating immunosuppression in HNSCC cells will require further investigation.
Given that TRIM24 was selectively expressed in tumour cells of the HNSCC microenvironment potentially provides an opportunity for selectively targeting STAT1 and mitigating its tumour promoting effects in HNSCC in vivo. This is particularly significant since STAT1 mediates tumour promoting and inhibitory roles in HNSCC cells despite playing a significant anti-tumour function in immune cells. Further, the tumour inhibitory effects mediated by TRIM24 inhibition were significantly greater than STAT1 inhibition. There is therefore a strong rationale to exploring TRIM24 as a target in HNSCC treatment, which can potentially leverage the STAT1 pathway to improve HNSCC tumour outcomes. It is possible that TRIM24 inhibition could directly or indirectly inhibit the tumorigenicity of HNSCC tumour cells. Additionally, while TRIM24 expression is low in tumour-infiltrating lymphocytes, its downregulation could, in turn, upregulate STAT1 anti-tumour activity. A small molecule inhibitor of TRIM24 has been reported with high potency and favourable pharmacokinetic properties-a therapeutic role in HNSCC could be further explored in future studies [47].
In conclusion, our findings indicate that tumour-derived STAT1 regulates HNSCC tumorigenicity and PD-L1 immune checkpoint signalling, while T-cell-derived STAT1 is necessary for immune activation and anti-tumour immune responses. We further demonstrate that TRIM24 potentiates the tumour promoting function of STAT1 in HNSCC cells.

DATA AVAILABILITY
All raw data supporting the findings of this study are available upon request. Sequencing data are available via the Gene Expression Omnibus (GSE198790).