Global pannexin 1 deletion increases tumor‐infiltrating lymphocytes in the BRAF/Pten mouse melanoma model

Immunotherapies for malignant melanoma seek to boost the anti‐tumoral response of CD8+ T cells, but have a limited patient response rate, in part due to limited tumoral immune cell infiltration. Genetic or pharmacological inhibition of the pannexin 1 (PANX1) channel‐forming protein is known to decrease melanoma cell tumorigenic properties in vitro and ex vivo. Here, we crossed Panx1 knockout (Panx1 −/−) mice with the inducible melanoma model Braf CA, Pten loxP, Tyr::CreER T2 (BPC). We found that deleting the Panx1 gene in mice does not reduce BRAF(V600E)/Pten‐driven primary tumor formation or improve survival. However, tumors in BPC‐Panx1 −/− mice exhibited a significant increase in the infiltration of CD8+ T lymphocytes, with no changes in the expression of early T‐cell activation marker CD69, lymphocyte activation gene 3 protein (LAG‐3) checkpoint receptor, or programmed cell death ligand‐1 (PD‐L1) in tumors when compared to the BPC‐Panx1 +/+ genotype. Our results suggest that, although Panx1 deletion does not overturn the aggressive BRAF/Pten‐driven melanoma progression in vivo, it does increase the infiltration of effector immune T‐cell populations in the tumor microenvironment. We propose that PANX1‐targeted therapy could be explored as a strategy to increase tumor‐infiltrating lymphocytes to boost anti‐tumor immunity.

Depending on the diagnosis and the mutational landscape of the tumor biopsies (e.g., BRAF mutation status), targeted therapy with mutant kinase inhibitors or immunotherapies with monoclonal antibodies against immune checkpoint inhibitors (e.g., anti-CTLA4, anti-PD-1 antibodies) are currently being used for melanoma treatment [3].As research into the PI3K (phosphoinositide 3-kinase)/AKT (protein kinase B)/NF-jB pathway deepens, its role in melanoma has also attracted significant attention.Dysregulating mutations of AKT family members are a common occurrence in melanoma, being identified in as much as 43-60% of melanoma cases [4].PTEN, a protein phosphatase that typically restrains the cellgrowth signaling cascade of PI3K/AKT/mTOR, is altered in 38% of primary melanoma patients and 58% of those with metastatic disease [5].Notably, alterations in both PTEN and the BRAF genes also coexist, potentially leading to the simultaneous dysregulation of both the mitogen-activated protein kinases (MAPK) and the PI3K pathways.Consequently, there is a possibility that PI3K inhibitors may offer some therapeutic benefit to patients with PTEN and/or AKT-mutant melanomas [6].On the other hand, the use of immune checkpoint inhibitors enhances antitumoral T cell-mediated immune responses, but unfortunately, about ~60% of melanoma patients do not respond to this type of therapy [7][8][9].Regardless of the treatment options, their effectiveness is usually dampened due to the insufficient effector activity of tumor-infiltrating lymphocytes (TILs) [10], recurrence, and the development of multiple resistance mechanisms [11,12]; therefore, innovative therapy concepts are still warranted.
Pannexin1 (PANX1) is a channel-forming membrane glycoprotein that mediates the passage of nucleotides, ions, and other metabolites involved in intercellular communication [13].We have previously demonstrated that PANX1 expression correlates with tumor cell aggressiveness in isogenic mouse melanoma cell lines, and its knockdown reduces melanoma tumor size and metastasis to the liver in chick chorioallantoic membrane (CAM) xenografts [14].More recently, we showed that PANX1 is highly expressed in human melanomas, and genetic or pharmacological targeting of PANX1 decreases the tumorigenic properties of melanoma cells [15].Mechanistically, PANX1 regulates melanoma cell growth and metabolism through direct interaction with b-catenin and modulation of the Wnt signaling pathway [16].All these findings place PANX1 as an attractive target for preclinical melanoma research, but it is still unclear whether the silencing of this gene would impact melanoma development in vivo.
In addition, PANX1 influences the inflammatory response [17][18][19][20][21][22] and is expressed in different subtypes of immune cells like macrophages and lymphocytes [17,23,24].Previous evidence shows that PANX1mediated adenosine-5 0 -triphosphate (ATP) permeability enables activation and functional enhancement of T cells [21,22].On the other hand, the release of ATP through caspase-cleaved PANX1 channels in apoptotic leukemic lymphocytes has been shown to induce the recruitment of inflammatory cells (e.g., monocytes) [17].Paradoxically, the caspase-dependent PANX1 channel opening in dying macrophages and lymphocytes releases metabolites that suppress, in vivo, proinflammatory signals of myeloid cells in the surrounding tissue [13].Considering the broad PANX1 expression and different cell types present in the tumor microenvironment (TME), no studies have specifically addressed whether PANX1 modulates the recruitment of immune cells in the melanoma TME [18].
Previous histological and pathological examinations of a constitutive global Panx1 knockout mouse (Panx1 À/À , Panx1KO) (developed by Genentech, San Francisco, CA, USA) [20] showed no overt phenotypes.However, using the same Panx1 À/À mouse strain, we have found differences in dorsal skin thickness and delayed wound healing [14,25].Moreover, our studies show that in Panx1 wildtype mice, normal melanocytes express low levels of PANX1 which are upregulated in melanoma cells [14].To gain new insights into the roles of PANX1 in the context of skin cancer, we set out to test the combination of this Panx1 À/À mouse line with a cutaneous melanoma mouse model.
In this work, we crossed the global Panx1 À/À mouse strain with the inducible melanoma model Braf CA/+ , Pten LoxP/Loxp , Tyr::CreER T2 (abbreviated here as BPC) that harbors two known oncogenic driver mutations [26][27][28].Due to the role of PANX1 in regulating tumor growth and the immune response [17][18][19][20][21][22], we sought to evaluate the effects of the Panx1 germline deletion on melanoma progression and the tumor immune infiltration of this new hybrid mouse model.
Our results showed that Panx1 global deletion did not reduce the strong BRAF/Pten-driven melanoma progression but increased the tumor infiltration of effector immune T-cell populations.Interestingly, PANX1deficient BPC mice exhibited sex-driven morphological differences in spleen size without an apparent influence on the tumor burden.We anticipate that targeting PANX1 in melanoma may increase the homing of effector T lymphocytes to melanoma tumors which in the future could be harnessed to improve the effect of immunotherapies.

Melanoma tumor induction, monitoring, and sample collection
All experiments performed on the mice were approved by the Animal Care Committee of the University Council on Animal Care at the University of Western Ontario, London ON, Canada (UWO AUP# 2019-070).3-4-week-old BPC mice were used for tumor induction experiments: Panx1 +/+ (7 females and 8 males) and Panx1 À/À (9 females and 4 males).Cutaneous melanomas were induced with topical application of 4-hydroxytamoxifen (4-HT) (Sigma) on the skin of the lower back as per protocol in Ref. [27].Mouse body weight and tumor dimensions were monitored every 3-4 days as soon as the tumors were visible.Measurements were done with a digital caliper, and tumor volume was calculated using the modified ellipsoid formula V = (W(92) 9 L)/2, where V is tumor volume and L and W are the longest (length) and shortest (width) tumor diameters, respectively.The endpoint was considered when the tumor volume (or total combined tumor volume, in case of multiple tumors per animal) reached 2.0 cm 3 , tumors showed signs of heavy ulceration, or the animals had poor body condition.Euthanasia was carried out using carbon dioxide asphyxiation followed by cervical dislocation.The spleens were weighed and measured using a ruler.Draining inguinal lymph nodes compromised with pigmented lesions were classified as metastatic sites and counted per mouse.Tissue samples were fixed overnight with periodatelysine-paraformaldehyde (PLP) fixative or flash-frozen in liquid nitrogen and stored at À80 °C until processing for RNA extraction.

Tissue processing, RNA extraction, and real-time qPCR
Frozen tissues/tumor samples were ground in liquid nitrogen, and total RNA was extracted using TRIzol TM Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol and purified using the RNeasy Ò Plus Mini Kit (Qiagen, Germantown, MD, USA).RNA concentration was determined using an Epoch Microplate Spectrophotometer (BioTek Ò , Santa Clara, CA, USA).cDNA was generated using the High-Capacity cDNA Synthesis Kit (Thermo Fisher Scientific; REF 4368814, Waltham, MA, USA) in a T100 TM Thermal Cycler (Bio-Rad, Hercules, CA, USA).SYBR Ò Green Real-Time PCR Master Mix (Bio-Rad) was used along with the primers listed in Table 1, where mouse gapdh and 18S were the housekeeping gene controls.Real-time qPCR was performed in a CFX96 Touch TM Real-Time PCR Detection System (Bio-Rad).All the assays involved at least three technical replicates (n = 3), and the relative mRNA expression analysis was determined using the DDC t method calculated in the Bio-Rad CFX Maestro Software, ver 1.1 (Bio-Rad).Between assays, a nontamoxifen-treated skin BPC-Panx1 +/+ sample was used as the control for inter-plate variation.

Image acquisition and analysis
Tile-scan imaging of whole tumor sections (209) was acquired in an Eclipse Ni-E Fluorescence Microscope (Nikon) with an ORCA-flash4.0LT plus (Model C11440) Hamamatsu digital camera and an S Plan Fluor ELWD Ph1 ADM 20x objective (N/A 0.45 and refractive index of 1).Granzyme B imaging was performed in an LSM 800 Confocal Microscope (Carl Zeiss, Oberkochen, Germany) using a Plan-Apochromat LCI Plan-Neofluar 259 (0.8 Imm Korr Water DIC) objective and a 639/1.40Oil DIC objective (Zeiss, Oberkochen, Germany).For the images acquired with the Nikon Fluorescence microscope, the noise component was removed with Denoise AI utility, and the fluorescence background was subtracted with the NIS-ELEMENTS ADVANCED RESEARCH ANALY- SIS software (ver.5.21.02,Nikon, Melville, NY, USA).Separated tile images per channel were converted to binary images, and automated image analysis was performed using CELL PROFILER Software (ver.4.0.7,Broad Institute MIT) [29], using a custom pipeline to determine the percent of nuclei (cells) with positive immunostaining for the marker over the total number of cells normalized to the area of the field of view, and expressed as percent (%) of cells with positive marker per lm 2 .Immunofluorescence (IF) was quantified using images from two serial cryosections of tumors from at least three mice (N = 3) per genotype.

Statistical analysis
All the statistical analyses were performed using GraphPad Prism Ò ver.8 (San Diego, CA, USA).Unless otherwise indicated, the results are expressed as the mean AE standard deviation (SD), and the biological (N ) and technical replicates (n) for a given experiment are indicated in each figure.Animal numbers were estimated by a priori power analysis (G*POWER ver.3.1 [30]) based on a pilot study of spontaneous tumor incidence between BPC-Panx1 +/À (heterozygous) Table 1.List of primers used for real-time qPCR analysis.

Target gene
Forward primer (5 0 -3 0 ) Reverse primer (5 0 -3 0 ) and BPC-Panx1 À/À mice with an effect size of 1.0 and a power of 0.8.For multiple comparisons involving the genotype and sex of mice, two-way ANOVA was performed, followed by a Tukey-Kramer post hoc test for unequal sample sizes.For real-time qPCR, log 2 (relative normalized expression) was used as data for statistical comparisons.Outliers in the gene expression data were detected using the ROUT method with a false discovery rate of < 1% and excluded from analysis.Data normality was verified with the D'Agostino-Pearson test.One-way ANOVA followed by Sidak's post hoc test was done to analyze relevant mean comparisons in the mRNA expression between tissues.For comparisons, IF quantification data (% cells with positive marker per lm 2 ) per mouse were transformed using a Box-Cox transformation [31] and the means of the two groups (genotypes) were compared using a nested two-way ANOVA (nested t-test).Statistical significance was considered when P < 0.05.

Germline Panx1 deletion does not reduce melanoma progression of Braf(V600E)/Pten (del) mice
To explore the in vivo effects of the Panx1 global deletion on melanoma progression, we crossed Braf CA/+ , Pten LoxP/Loxp , Tyr::CreER T2 (BPC) mice with the global Panx1 knockout (Panx1 À/À , Panx1KO) mouse line, reported on Ref. [20,27].In the BPC model, a 4hydroxytamoxifen (4-HT)-inducible Cre recombinase is used for the melanocyte-specific expression of mutant BRaf V600E -a constitutively active protein serine kinase-and inactivation of the Pten phosphatase to produce primary cutaneous malignant melanomas [11].However, as in Ref. [23], we observed that spontaneous melanomas occurred in approximately 70% of the BPC colony in the absence of tamoxifen exposure.Thus, we started the 4-HT induction experiments in mice between 3 and 4 weeks of age that looked healthy and had no spontaneous melanoma formation.
Our results showed that regardless of the Panx1 genotype, melanocytic pigmented lesions arose from the skin and quickly grew to form cutaneous melanomas as early as 3 weeks after topical application of 4-HT.Furthermore, although we applied 4-HT to a small zone in the dorsal skin of the lower back, we found that, in both animal groups, multiple tumor lesions grew not only in the application region but also in distant sites (Fig. 1A).As a result, all the visible melanoma lesions found in the dorsal and ventral/lateral skin of the animals were considered to determine tumor burden (Fig. 1B) and endpoint conditions.Overall, we observed that BPC-Panx1 À/À mice had similar tumor incidence (Fig. 1B), survival (Fig. 1C), and mouse weight among sexes (Fig. 1D) compared to Panx1-wildtype counterparts.However, although the total combined tumor growth over time (Fig. 1E) was similar between genotypes, analysis of estimated tumor growth parameters showed sex differences of the tumor growth in BPC-Panx1 +/+ mice (Fig. 1F,G).BPC-Panx1-wildtype male mice had a slightly lower (though not significant) tumor growth rate (Fig. 1G, two-way ANOVA, sex factor: F 1,23 = 4.64, P < 0.05) with a significant increase (P = 0.188) in tumor doubling time (Fig. 1F, two-way ANOVA, sex factor: F 1,23 = 7.86, P < 0.05) compared to female counterparts.However, this sex-driven difference was not significant in BPC-Panx1-deficient mice and overall tumor growth did not differ between genotypes.These results indicated that Panx1 deletion in this mouse model did not reduce BRAF V600E /Pten-driven melanoma development.
In general, most melanoma lesions were presented with a combination of melanotic (black) and amelanotic (white) regions (Fig. 2A).However, a few exceptions (< 30% of animals per genotype) also had completely amelanotic tumors primarily distant from the tamoxifen application site.A classification of the tumors with the presence or complete absence of black pigmentation rendered no significant differences in their number among genotypes (Fig. 2A).All mice, regardless of the genotype, developed fast-growing tumors that, in most cases, were ulcerated by the time of endpoint (Fig. 2B).Since our previous reports indicated that PANX1 is highly expressed in human melanoma tumors compared to normal skin, we performed real-time qPCR in matched samples of non-tamoxifentreated (adjacent) skin and tamoxifen-induced tumors.Panx1 mRNA expression was found to be 3.2-fold increased (Fig. 2C) in tumors of BPC-Panx1 +/+ mice compared to skin, while undetectable in any tissue sample from BPC-Panx1 À/À mice, as expected.Moreover, mRNA expression of the S100a1 melanoma marker [32,33] had a 5.1-fold increase (P < 0.05) in tumors of BPC-Panx1 À/À compared to the skin (Fig. 2D), indicating that the development of primary tumors occurred independently of Panx1 expression.

Braf(V600E)/Pten-driven melanoma metastasis to inguinal lymph nodes is not restricted by Panx1 deletion
Melanoma tumor cells are known to pass through the lymphatic system and enter the bloodstream to metastasize to other organs [34].We observed localized and distant clusters of small black pigmented spots spread in the underside of dorsal skin that was not treated with tamoxifen, suggesting the presence of melanoma microlesions or that in-transit metastasis may have occurred (Fig. 3A).Regional black melanoma metastases were found in both right and left inguinal lymph nodes in both Panx1 genotypes (Fig. 3B-D) but this was not noticeable in major organs (e.g., lungs, liver, or brain).Upon measuring the length and weight of spleens (Fig. 4A-C), a statistically significant sex-dependent difference was found in BPC-Panx1 À/À mice (two-way ANOVA, F 1,20 = 10.91,P < 0.01).Females had an increased spleen weight index (normalized to body weight) compared to males of the BPC-Panx1 À/À cohort but this was not observed in BPC-Panx1 +/+ mice or among genotypes.A positive correlation between spleen weight and length measurements (Fig. 4D) (Pearson's coefficients r ≥ 0.7, P = 0.0247 and P = 0.0087) in both BPC-Panx1 +/+ and BPC-Panx1 À/À mice, respectively, confirmed a similar trend in both parameters.

CD8 mRNA is significantly increased in
Panx1-deficient skin and melanoma tumors Due to PANX1's role in the inflammatory response [17][18][19][20][21][22], we sought to examine changes in the infiltration of immune cells within the TME.We performed a real-time qPCR analysis of gene expression for markers of immune cell populations known to modulate the immune response against melanoma.We included primers specific to protein tyrosine phosphatase, receptor type, C (Ptprc) [also known as cluster of differentiation (CD) 45] for leukocytes; CD3 antigen, epsilon polypeptide (CD3e) for T lymphocytes; CD4 antigen (CD4) for T-helper lymphocytes; CD8 antigen, beta polypeptide (CD8b) for cytotoxic T cells; forkhead box P3 (Foxp3) for regulatory T cells (Treg); and natural cytotoxicity triggering receptor-1 or NCTR1 (Ncr1) of natural killer (NK) cells and the cell surface glycoprotein F4/80 (Agre1) for monocytes/macrophages.The "immune" marker expression was evaluated in melanoma tumors and matched samples of non-tamoxifen-treated dorsal skin devoid of any visible melanoma lesions (referred to as "adjacent skin").These adjacent skin samples were used to account for basal immune infiltration levels or skinresident immune cells [35,36] at the time of sample collection.
When analyzing tumors compared with their matched skin samples, we observed a significant (F 3,20 = 6.24,P < 0.01) 4.3-fold increase expression of CD45 only within the BPC-Panx1 +/+ mice group (Fig. 5A).The F4/80 mRNA expression (F 3,18 = 19.38,P < 0.0001) was 14.0-fold (P < 0.0001) and 7.7-fold (P < 0.01) higher in tumors than the skin in BPC-Panx1 +/+ and BPC-Panx1 À/À mice, respectively (Fig. 5B).On the other hand, transcripts of NCTR1 (F 3,11 = 8.24, P < 0.01) (Fig. 5C) were 9.3-fold upregulated (P < 0.01) in BPC-Panx1 À/À tumors versus adjacent skin.CD3e (Fig. 5D) showed a similar expression regardless of the sample type or Panx1 genotype, while CD4 mRNA expression (F 3,11 = 27.46,P < 0.0001) was significantly augmented in the tumors compared to the skin (Fig. 5E) with a 10.5-fold (P < 0.01) and 18.9-fold (P < 0.001) increase in the BPC-Panx1 À/À and BPC-Panx1 +/+ , respectively, and five-fold upregulated (P < 0.05) in the skin of BPC-Panx1 À/À vs. that of BPC-Panx1 +/+ .Moreover, Foxp3 (F 3,16 = 9.80, P < 0.001) was upregulated (5.9-fold, P < 0.05, and 15.2-fold, P < 0.01) in tumors of BPC-Panx1 +/+ and BPC-Panx1 À/À mice, respectively (Fig. 5G).Notably, CD8 (F 3,16 = 15.39,P < 0.0001) was significantly increased (P < 0.01) in the skin and tumors (9-and 9.5-fold, respectively) of BPC-Panx1 À/À compared to BPC-Panx1 +/+ animals (Fig. 5F).Taken together, these results indicated that Panx1 deficiency did not impair the overall recruitment of immune cells to the primary tumors in this melanoma model but had an effect on Fig. 1.Global Panx1 deletion does not improve survival or reduce tumor progression in the Braf(V600E)/Pten-deficient mouse melanoma model.(A) Representative picture of Tyr::CreER; BRaf CA/+ ; Pten À/À (BPC) mice knockout (À/À) for Panx1 with multiples tumors at endpoint.The region of the tamoxifen application is indicated in the picture.(B) The number of visible cutaneous melanoma lesions at the endpoint was similar (ns, not statistically significant) between both mouse sexes and genotypes as analyzed by a two-way ANOVA.(C) Kaplan-Meier survival curves were similar among BPC-Panx1 wildtype ( +/+ ) or BPC-Panx1 À/À mice (log-rank test, P = 0.7803).Dotted lines denote 95% confidence intervals.(D) Comparison of mice's body weight after tumor induction showed no differences among genotypes within the same sex group.Statistical significance was determined by multiple t-tests per timepoint using the Holm-Sidak method without assuming a consistent SD. (E) The tumor growth curve showed overlap between genotypes, with a group of mice exhibiting slightly delayed tumor growth.Individual lines depict measurements per mouse.(F, G) Tumor growth rate and doubling time were similar among genotypes but significantly different between males and females within the BPC-Panx1 +/+ group.Two-way ANOVA showed no interaction between genotype and sex, but significant P-values for the sex factor in tumor doubling time (F 1,23 = 7.855, P < 0.05) and growth rate (F 1,23 = 4.642, P < 0.05), respectively.Symbols stand for individual data per mouse (BPC-Panx1 +/+ , n = 15, and BPC-Panx1 À/À , n = 13) except for the graph in D, where mean AE SD are shown.Due to unequal sample sizes per sex, the Tukey-Kramer test was used as post hoc test for pairwise comparisons shown in F and G.If not explicitly stated, the error bars depicted in the figures represent the standard deviation (SD).Asterisk (*) shows statistical significance when P < 0.05.increasing the transcript expression of CD8 in skin and tumors of BPC mice.Given the similarity in tumor burden among genotypes but increased CD8 mRNA in the BPC-Panx1 À/À tumors, we investigated whether there were differences in the activation/exhaustion phenotype of TILs and immunosuppression compared to those of the BPC-Panx1 +/+ cohort.We observed a trend of lower transcript expression (not significant) of CD69 (early T-cell activation marker, [37]) in BPC-Panx1 À/À tumors (Fig. 6A).Moreover, no differences in expression of the regulatory immune checkpoint receptor lymphocyte activation gene-3 (LAG-3) [38] or programmed cell death ligand-1 (PD-L1) [39] were found between genotypes (Fig. 6B,C).Finally, due to the association of PANX1 channel function and the pro-inflammatory and tumor promoter cytokine: interleukin-1b (IL-1b), we investigated whether the IL-1b expression was affected within the TME of BPC Panx1-deficient mice.IL-1b mRNA levels (Fig. 6D) were found to be highly variable in the BPC-Panx1 À/À tumors but were not different compared to tumors of the BPC-Panx1 +/+ mice.To confirm the differences found in mRNA markers for T-cell infiltration in tumors between Panx1 genotypes, we performed IF staining on PLP-fixed tumor cryosections to verify the presence of CD4 + , CD8 + , and Foxp3 + lymphocytic cells (Fig. 7A).Overall, CD4 + and CD8 + T cells were seen in small but sparsely distributed clusters closer to the skin epidermis of the primary melanoma tumors, and Foxp3 + T cells showed a more homogeneous distribution within the tumor core irrespective of genotype (Fig. 7A).
Overall, the quantification of the immune cell infiltration of lymphocytes agreed with the previous results on real-time qPCR immune markers.The percent of positive CD4 + and CD8 + T cells was variable among tumor samples and, although not statistically significant, it trended increased in BPC-Panx1 À/À vs BPC-Panx1 +/+ tumors with a median value of 3. Relative normalized expression (log 2 ) Fig. 6.Global Panx1 deletion does not affect the activation phenotype of tumor-infiltrated T lymphocytes.Real-time qPCR analysis of (A) the early T-cell immune activation marker CD69, (B) checkpoint receptor lymphocyte activation gene-3 (LAG-3), (C) immune checkpoint molecule programmed cell death ligand-1 (PD-L1), and (D) IL-1b in tumor samples.Gapdh was used as a reference gene, and the normalized expression is relative to one of the BPC-Panx1 +/+ skin samples.Box plots represent the 95% confidence interval (CI) and the median (inner line), with whiskers showing the maximum and minimum values of the gene expression.Each sample was assayed independently in triplicate (N = 3), and the symbols represent the expression data of each mouse (n = 4 per genotype).A t-test compared the mean of the log 2 of the relative expression per group; ns stands for no statistical significance, P > 0.05.
Collectively, these results confirmed that the deletion of Panx1 did not affect the infiltration of CD4 + and CD8 + T lymphocytes in the tumors, although the cytotoxic effect seemed to be insufficient to prevent the tumor progression in BPC-Panx1 À/À mice.

Discussion
PANX1 channels have attracted considerable attention due to their multiple roles in inflammation, cell death, and cancer [18,40].We have demonstrated that PANX1 influences the regulation of effector molecules that control melanocyte differentiation (e.g., MITF and b-catenin) and the metabolism of melanoma cells [14][15][16].Furthermore, we have shown promising results after silencing Panx1 expression and reducing the growth and tumorigenicity of melanoma cell lines in vitro.In this work, using the immunocompetent BPC mouse melanoma model, we did not observe a reduction in Braf(V600E)/Pten(del)-driven melanomas or improve mice survival upon germline Panx1 deletion.Of note, our previous in vitro and ex vivo work targeting PANX1 in melanoma employed murine B16-F0, B16-F1, B16-F10, and human BRAF(V600E)mutant melanoma cell lines (A375-P and A375-M2) that do not entirely share the same genomic alterations of the BPC model [15,41].The deletion of Pten is known to have a synergistic effect and dramatically accelerates BRAFV600E-induced melanoma tumors in BPC mice [26,27,42].Therefore, we hypothesize that oncogenic mutations intrinsic to this mouse BPC model may have overridden our anticipated antitumoral effect of Panx1 deletion.Since this is a highly penetrant mouse model, a syngeneic melanoma model may be more suitable to study the effect of Panx1 deletion in melanoma metastasis [43].Besides, the germline deletion of Panx1 does not necessarily imply the same effect previously observed with the use of PANX1 channel blockers (e.g., with carbenoxolone, probenecid, or spironolactone) due to possible compensation with other proteins (e.g., connexins or other paralogs) yet to be tested using this or another in vivo melanoma model.We also observed the formation of micro-metastasis to draining inguinal lymph nodes (a phenotype reported elsewhere [26][27][28]44]) in Panx1deficient mice.Therefore, we concluded that Panx1 deletion does not control the early spread of melanoma.Due to the ubiquitous expression of Panx1, its global deletion should have affected many cell types, including melanoma and immune cells which may resemble the effect of an irreversible depletion of PANX1 function in which adaptative or alternative molecular mechanisms are in play that helped to sustain tumor growth.
Interestingly, we noticed a sex dimorphism in melanoma tumor progression where BPC-Panx1 +/+ female mice seemed to have a more aggressive disease (Fig. 1E,G) than males of the same cohort.Other researchers have noted this outcome in a C57B/6J background, but the origin of this difference has not been clarified yet [45].It has been speculated that this may be due to the presence of alternative steroid receptors in melanoma cells or technical limitations with tamoxifen (an estrogen receptor modulator) differentially influencing the mouse sexes in this melanoma model [45].This sex dependency was lost in BPC-Panx1 À/À mice, suggesting that Panx1 deletion may influence such a sex-driven phenotype.It is known that sexual dimorphism exists in the response of Panx1-deficient mice in ischemia [46] and epilepsy [47], and research on this is still lacking in the melanoma and immune cell recruitment contexts.A recent study has shown that tumor progression and antitumoral T-cell response are sex-dependent in a syngeneic B16-F10/BL6 mouse model; however, in contrast to our observation, such study showed that females had less tumor growth rate and higher CD4 + and CD8 + T-cell infiltrates compared to males [48].Due to the lack of sample availability at the time of the experiment, we did not quantify differences in T-cell subtypes between samples from different sexes.Therefore, it is unclear the nature of the loss of such sex disparity and further investigation may be required to clarify this finding.
To date, there are no reports of major abnormalities or immune-related diseases in the Panx1-deficient mice (reviewed at Ref. [18]).Here, we report for the first time that tumor-bearing BPC-Panx1 À/À female mice have significantly augmented spleen size compared to their male counterparts.Due to the short timeframe for tumor growth (~2 months) and given that the sexual dimorphism was found only in BPC-Panx1 À/À spleens, we speculate that characteristic developmental differences may account for this phenotype in Panx1 À/À mice.Nonetheless, further research is warranted to determine whether global Panx1 deletion alone exerts sex-specific changes in the spleen morphology and whether this causes an immune dysregulation that could impact the tumor response in mice.Tumor-infiltrating lymphocytes (TILs) have already been described in this BPC model [28].However, others have highlighted that this model does not appear to induce a marked immune response due to the lack of additional tumoral somatic mutations/neoepitopes [43].Notably, our results showed trends of increased TILs in BPC-Panx1 À/À tumors which suggests that Panx1 has an influence on the immune infiltration in the TME of this model.
It must be taken into consideration that, although not investigated here, in the TME, there are diverse immunosuppressive cells like myeloid-derived suppressor cells, T regulatory (Treg) cells, tumorassociated fibroblasts, and macrophages.These cells release various soluble factors, including reactive oxygen species (ROS), which effectively hinder the response of other effector cells like natural killer (NK) cells.Moreover, elevated levels of fibroblasts in the TME contribute to increased secretion of metalloproteinases, leading to the further shedding of ligands that could otherwise engage with NK cells [49].Additionally, tumor-associated fibroblasts are known to negatively influence NK cells by inhibiting the upregulation of activating receptors induced by cytokines [50].Furthermore, immunosuppression of NK cells can be exerted by exosomes derived from melanoma [51], which were also not considered in the present study.
Despite our limited analysis on immunosuppression, we found no difference in FoxP3 mRNA expression and Treg cell infiltration between tumors from both mice cohorts but an increase of FoxP3 transcript expression in tumors (Figs 5G and 7D) was evident compared to the skin tissue.This suggests that deletion of Panx1 alone may not influence Treg cell-driven immunosuppression in this tumor model.On the other hand, myeloid-derived suppressor cells, along with high PD-1 expression on the T cells, are known to highly immunosuppress this melanoma model [28].It is unclear whether these or other subsets of CD4 + T cells (Th1 or Th17) and macrophages (M2) may have contributed to immune evasion.Furthermore, we found no differences in gene expression of markers of TIL activation (CD69) [52,53], exhaustion (LAG-3), or tumor immune evasion (PD-L1); however, our strategy could not provide a definitive assessment of the proportion of the activation/exhaustion status within the TIL subset.Nevertheless, it remains to be tested whether the absence of PANX1 in a specific subtype of the immune cell in the tumor could be exploited to counterbalance the immunosuppressive or evasive mechanisms of melanoma.
PANX1 has been proposed as a potential target to regulate anti-tumor immunity in melanoma due to the PANX1-mediated release of proinflammatory cytokine IL-1b (reviewed in Ref. [54]).IL-1b in melanoma elicits inflammation and the expansion of highly immunosuppressive myeloid-derived suppressor cells [55].In this study, we found no significant differences in IL-1b mRNA expression among cohorts (Fig. 6D) suggesting that this immune evasion mechanism may be still in place contributing to tumor progression despite the increased TIL in Panx1 À/À tumors.In addition, using this same Panx1 À/À mouse strain, another group showed that marrow-derived macrophages have no difference in the secretion of IL-1b or inflammasome activation [20].This underscores the need for research into the crosstalk between immune and tumor cells under cell-specific PANX1 suppression/inhibition to better understand the implications of PANX1 targeting in the context of melanoma and immunotherapy.As a limitation of this study, Panx1 gene deletion was not directed to any specific cell compartment; therefore, it is difficult to determine the specific impact of this deletion in the immune and tumor compartments distinctly.A possible solution to this problem could include the use of a syngeneic mouse model where Panx1 deletion can be targeted to specific immune cell types and be studied in a more relevant immunogenic melanoma model like YOVAL1.1 [56].
This transgenic melanoma mouse model was originally developed to harness two significant oncogenic alterations [BRAF V600E and PTEN (loss)], but it is not representative of the larger patient population since in our analysis of the Cutaneous Melanoma TCGA Pan-Cancer data, we found that only 5.0% of cases had co-occurrence of both mutations (Fig. S1A).Furthermore, since PANX1 expression seems independent of genomic alterations of melanoma (Fig. S1B) and is present at all stages of the disease [15], a larger patient population may benefit from PANX1 targeted therapy.Interestingly, Pten-null cancers seem resistant to immune checkpoint inhibitory antibodies, where PTEN loss in tumor cells increases the expression of immunosuppressive cytokines inhibiting trafficking and T cell-mediated killing of tumors [57,58].Hence, our results point toward the direction of targeting PANX1 as a potential alternative to overcome T-cell exclusion in immunosuppressive TMEs.

Conclusions
Our study demonstrates that a Panx1 deletion cannot overcome the aggressive tumorigenic effect of Braf(V600E)/Pten(del) driver mutations, but it increases the tumor T-cell infiltration with a preference in CD8 + cytotoxic T cells.This could be advantageous since a considerable therapeutic benefit is achieved in patients with a pre-existing CD8 + T-cell infiltration within the melanoma microenvironment [59].Future preclinical studies should include the combination strategies in which PANX1 inhibition could be considered as means to stimulate cytotoxic T-cell tumor infiltration to improve the efficacy of immunotherapies.

Fig. 2 .
Fig. 2. Pigmentation or ulceration of Braf(V600E)/Pten(del) melanoma lesions is not influenced by the global Panx1 deletion.(A, B)Representative pictures of pigmented and ulcerated tumors.Fractions (out of the total tumors per mouse) of pigmented or ulcerated tumors were similar among the genotype and sex groups (8 males and 7 females were BPC-Panx1 +/+ and 4 males and 9 females were BPC-Panx1 À/À ).Symbols show individual data per mouse and horizontal lines, and error bars represent mean AE SD.Data were analyzed with two-way ANOVA.(C, D) mRNA expression of Panx1 and melanoma marker s100a1 in tumors compared to matched samples of nontamoxifen-treated skin per mouse.Gapdh and 18s were used as reference genes, and the normalized expression is shown relative to one of the BPC-Panx1 +/+ skin samples.Box plots represent the 95% confidence interval (CI) and the median (inner line), with whiskers representing the maximum and minimum values of the gene expression.Symbols display the individual expression data per mouse (n = 5 and 6 BPC-Panx1 +/+ ; n = 7 and 5 BPC-Panx1 À/À , tumor and skin samples, respectively).Paired t-test was used to compare the log 2 (normalized expression) of Panx1 and one-way ANOVA followed by a Tukey test used to compare the expression of S100a1.All paired samples were assayed by triplicate (N = 3).Statistical significance is shown as P < 0.05(*).

Fig. 3 .
Fig. 3. Panx1 deletion does not prevent the development of Braf/Pten-driven melanoma microlesions or metastasis to inguinal lymph nodes in BPC mice.(A) Representative pictures of melanoma microlesions found underside the dorsal and ventral/lateral skin and located distant (inset A) and nearby (inset B) to the application sites of tamoxifen (e.g., lower back dorsal skin).LN: Inguinal draining lymph nodes.(B) Representative pictures of right and left inguinal lymph nodes showing black pigmentation indicative of melanoma metastasis.(C) Phasecontrast micrograph of an inguinal draining lymph node cryosection exhibiting melanin deposits and pigmented cells (white arrows).Scale bar = 100 lm (D) Quantification of the fraction (out of two inguinal LN per mice) of draining lymph nodes classified with melanoma metastasis based on the presence or absence of visible pigmented lesions.Symbols correspond to the individual data per mouse (n = 8 males and 7 females BPC-Panx1 +/+ ; n = 4 males and 9 females BPC-Panx1 À/À ) where mean AE SD are shown.Data were analyzed with a two-way ANOVA.Statistical significance was considered when P < 0.05.

Fig. 4 .
Fig.4.Sex-dependent differences in spleen size were observed for BPC-Panx1 À/À but not for BPC-Panx1 +/+ mice.(A) Representative pictures of BPC mice spleens.Scale bar = 1 cm (B) Morphological comparison of the spleen measured by length and (C) weight index (spleen weight normalized to body mass) showed a trend of enlarged spleens in female BPC-Panx1 À/À mice.Sex-dependent differences in BPC-Panx1 À/À mice were statistically significant (two-way ANOVA, F 1,20 = 10.91,P < 0.01).A Tukey-Kramer test used as post hoc test for pairwise comparisons showed a significantly (**P < 0.01) higher spleen weight index in BPC-Panx1 À/À females than males, but not in the BPC-Panx1 +/+ cohort.Symbols correspond to data from individual mice (n = 6 males and 6 females BPC-Panx1 +/+ ; n = 4 males and 8 females BPC-Panx1 À/À ) where mean AE SD are shown.(D) A positive correlation was found between the spleen weight and length, confirming the trends of increased spleen size in female BPC-Panx1 À/À mice observed by both measurements.

Fig. 5 .
Fig.5.CD8 T-lymphocyte marker is significantly increased in the non-tamoxifen-treated skin and melanoma tumors of BPC mice lacking PANX1.(A-G) Real-time qPCR analysis of immune cell marker expression in tamoxifen-induced tumors and matched non-tamoxifen-treated skin.Gapdh and 18s were used as reference genes, and the normalized expression was calculated relative to one of the BPC-Panx1 +/+ skin samples.Box plots represent the 95% confidence interval (CI) and the median (inner line), with whiskers showing the maximum and minimum values of the gene expression.The symbols represent the mRNA expression data of each mouse (n = 6 per genotype, except in cases where one or two outliers were excluded) (e.g., n = 5 in F4/80, FoxP3 and n = 4 in NCTR1, BPC-Panx1 À/À tumor; n = 4 in CD3 BPC-Panx1 À/À skin) or there was reduced sample availability (e.g., n = 3 in NCTR1 and CD4 BPC-Panx1 À/À skin).Data are representative of 3 independent assays (N = 3).The log 2 of the expression per group was analyzed by one-way ANOVA followed by a Sidak post hoc test.Only significant P values of relevant comparisons between groups are shown as P < 0.05 (*), <0.01 (**), <0.001 (***), and <0.0001 (****).

982
Molecular Oncology 18 (2024) 969-987 ª 2024 The Authors.Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies.

Fig. 7 .
Fig. 7. Tumor-infiltrating CD4 + , CD8 + T lymphocytes, and granzyme B-producing cells are present in BPC-Panx1 À/À tumors.(A) Representative micrographs (209, insets at 609 magnification) showing immunofluorescence staining of primary melanoma tumor cryosections with immune cell infiltration markers for CD4 + , CD8 + , and FoxP3 + T lymphocytes (shown in cyan).Nuclei (in magenta) were counterstained with Hoechst 33324.Yellow arrowheads indicate stained immune cells.Scale bars = 50 lm.(B-D) Immunofluorescence quantification of CD4-, CD8-, and FoxP3-positive cells for the whole tumor section per animals (E) Representative granzyme B (GzmB) immunofluorescence staining of a BPC-Panx1 À/À tumor (fluorescence channels on top and brightfield channel at the bottom) with the intra-tumoral distribution of GzmB + cells and black pigmented regions in the tumor.Nuclei are shown in blue, and GzmB is shown in green.Yellow arrowheads indicate stained immune cells.Scale bars = 40 lm.(F) Quantification of GzmB immunostaining.The percent of nuclei with positive immune marker staining was normalized per area of the field of view (FOV).Data shown in the nested graphs are representative of one (N = 1) immunofluorescence staining of four mice (n = 4) per genotype for CD4, CD8, and FoxP3 markers and n = 3 mice for GzmB.Lines within graphs depict the median with the interquartile range, and individual measurements per FOV are shown with symbols.Box-Cox transformed data from the plots were compared using a two-tailed nested t-test, and all were not statistically significant, P > 0.05.