1. Caspase-4 is involved in Non-Small Cell Lung Cancer.
Recently, we demonstrated that NSCLC patients had higher circulating levels of caspase-4 than healthy subjects (12). In this study we found that NSCLC tissues had a positive histological score of caspase-4 which was higher in the tumor masses than in non-cancerous tissues (Figure 1A). In particular, as already observed by using a different technical approach (12), more than 70% of patients presented tissue positivity to caspase-4 (Table I). Similarly, transcriptional levels of CASP4 mRNA were higher in the tumor mass than in healthy (non-cancerous) tissues of NSCLC patients (Figure 1B). To note, peritumoral areas had intermediate levels of caspase-4 mRNA compared to healthy (non-cancerous) and tumoral areas (Figure 1B). These data were confirmed by means of western blotting (Figure 1C and D), which showed that lung cancer tissues were characterized by the presence of the cleaved form of casapse-4 (25-30 kDa) than healthy tissues. Based on these data, we hypothesized that the presence of caspase-4 could be related to the TNM stage of the pathology. Interestingly, the cleaved form of caspase-4, herein identified as tumor-associated caspase-4, statistically increased from stage I up to stage III (Figure 1E, stage I to III). To correlate the malignancy stage to the levels of tissue-associated caspase-4, we chose a cut-off value of the protein, as already reported (12) and correlated it to the survival rate. The survival rate of NSCLC patients at stage I who presented higher levels (>377 pg/ml) of tissue-associated caspase-4 was lower (median=0.967 years) than patients at stage I who presented lower (<377 pg/ml) levels (median=3.02 years) (Figure 1F). Similarly, NSCLC patients at stage II had lower median survival rate when tumor-associated caspase-4 levels were higher than 377 pg/ml (Figure 1G). However, in this last case we were not able to reach statistical significance due to the low number of patients. Moreover, stage III NSCLC patients had worse survival rate (median= 0.65 years) when tumor-associated caspase-4 was >377 pg/ml compared to patients who had lower levels (median= not revealed; n=2) (Figure 1H). Again, it has to be pointed out that the number of patients at stage III were low.
Caspase-4 in humans and caspase-11 in mice have been widely associated to the non-canonical pathway of the inflammasome (8; 9). Therefore, we went on by analyzing the levels of tissue IL-1β and IL-1α, two cytokines derived by the inflammasome activation (8, 11, 14). We found that, despite what reported in literature (15, 16), NSCLC patients did not have differential levels of tissue IL-1β (Figure 1I). Instead, we found that IL-1α was significantly increased in the tumor tissues than non-cancerous tissues (Figure 1J). Because we already demonstrated a correlation between IL-1α and caspase-4 (4, 17), we found that patients who presented levels of IL-1α >1.39pg/mg protein (cut-off value chosen according to ROC analysis) in the tissue (black line, Figure 1K) had similar survival rate as patients who had IL-1α >1.39pg/mg protein and caspase-4>377 pg/ml (Figure 1K, red line), compared to patients who had lower levels of both caspase-4 and IL-1α (Figure 1K, green and blue lines).
Taken together these data suggest that tumor tissues are characterized by high levels of caspase-4 and IL-1α which are associated to lower survival rate of NSCLC patients.
2. Higher levels of tumor-associated caspase-4 are present in the lung of NSCLC patients with poor survival rate.
To better understand the role of caspase-4 in lung cancer, we took advantage of a public database (www.cbioportal.org) where we analyzed what already reported in the literature regarding this enzyme. We found that 2% of patients in the public database presented a genomic alteration of CASP4 gene, intended as mutation or amplification or higher gene copy number. In this pattern we analyzed positive correlation between CASP4 and other genes according to Spearman correlation coefficient >0.25 taking into consideration both adenocarcinoma and squamous carcinoma patients reported in the database. In adenocarcinoma patients, CASP4 positively correlated to 9 genes involved in gene expression (i.e. C-MYCT1, SRA1, GTF2B, EEF1A1, SP110, CNOT8), 186 genes involved in inflammation (i.e. CASP1-5, CARD16, CD63, HLA, IRF1, TLRs), 106 genes involved in cell proliferation (i.e. C-MYCT1, SRA1, RHOA, CDK7, RAB32, CD74, RRAS) and 17 genes involved in cell death (i.e. CASP5, ANXA5, BAK1, RIPK2, MLKL) (Figure 2A). Similarly, in squamous carcinoma CASP4 positively correlated to 47 genes involved in gene expression (i.e. CREM, BATF, BZW1, HDAC9, STAT1-4), 303 genes involved in inflammation (i.e. AIM2, CASP5, CARD6-16-17-19, CCLs, ICOS, TLRs), 266 genes involved in cell proliferation (i.e. FAM107B, FGF7, IRF1-8-9, RASGRP3) and 63 genes involved in cell death (i.e. CASP8, ANXA3-5, FAS, GZMs, RIPK3) (Figure 2B). Among these, we found that caspase-5 and K-Ras correlated to CASP4 in both histotypes (Figure 2A and B, common areas in the Venn diagram).
In order to understand the impact of caspase-5 in NSCLC, we used a caspase-5 antibody to test via an ELISA assay the presence of this enzyme in tumor tissues compared to caspase-4, which was detected by means of ELISA by using a specific antibody, different from what is commercially available, actually under patent (RM2014A000080 and PCT/IB2015/051262). ROC analyses showed that the detection of caspase-4 in lung tumor masses was an excellent diagnostic tool for both adenocarcinoma (Figure 2C, red line) and squamous carcinoma (Figure 2D, red line) compared to caspase-5 (Figure 2C and 2D, green line). These data are in support to what already published (12), but further highlights that caspase-5, differently than caspase-4, is not associated to the cancerous pattern as it is also present in non-cancerous tissues. This effect was better described by the value of the AUC which was 0.5739 in adenocarcinoma (Figure 2C, green line) and 0.5385 in squamous carcinoma tissues (Figure 2D, green line). In support, adenocarcinoma patients who had tumor-associated levels of caspase-4 higher than 377 pg/ml (n=42/53, 79,3%) survived less (median=0.925 years) than patients with lower levels (n=11/53, 20,7%; median survival =3.03 years) (Figure 2E). Similar scenario was observed for squamous carcinoma patients. Higher tumor-associated caspase-4 (n=15, 88.2%) was associated to lower survival rate (median=0.86 years) than those who had lower levels (n=2, 11.8%; median= undefined) (Figure 2F). These data, despite the low number of patients in Figure 2F, most likely imply that tumor-associated caspase-4, but not caspase-5, is correlated to lung carcinogenesis.
3. Higher levels of tumor-associated caspase-4 are present in the lung of PD-L1 negative NSCLC patients.
NSCLC patients could be classified according to mutations and PD-L1 positivity (2). Therefore, we stratified NSCLC patients as PD-L1 positive vs PD-L1 negative according to the levels of tumor-associated caspase-4. We found that NSCLC patients who had higher levels of tumor-associated caspase-4 (>377 pg/ml, cut-off) but were negative for PD-L1 had lower survival rate (median survival=0.96 years; % of survival rate at 1 year≈30%) (Figure 3A, blue line) (Figure 3B, n=52/75, 69.3%) than patients (9.3%) who had higher levels of tumor-associated caspase-4 but PD-L1 positive (median survival=undefined; % survival rate at 1 year≈80%) (Figure 3A, black line; Figure 3B, n=7/75, 11.1%). However, patients with higher levels of tumor-associated caspase-4 but positive for PD-L1 (Figure 3A, black line) showed lower survival rate (p=0.033) than patients with low caspase-4 but positive for PD-L1 (Figure 3A, green line, n=1/75, 1.3%, median survival=undefined). Nevertheless, it has to be noted that there was a group of patients who were PD-L1 negative and presented lower tumor-associated caspase-4 (Figure 3A, red line; Figure 3B, n=15/75, 20%, % survival at 1 year=85%) whose median survival rate was of 2.98 years. These latter group survived more than patients with high levels of caspase-4 and PD-L1 negative (Figure 3A, blue line), survival rate that was still lower than patients who had PD-L1 positivity and high caspase-4 (Figure 3A, black line) and PD-L1 positivity and low caspase-4 (Figure 3A, green line). These data imply that a group of patients (Figure 3B, n=15/75, 20%), whose survival is still low, is independent from caspase-4 and PD-L1. Though, it is noteworthy that the survival of this group at 1 year is ≈80% (Figure 3A, red line) compared to the group with high caspase-4 and PD-L1 negative (Figure 3A, blue line), who instead, presented a survival rate at 1 year of ≈30%. Based on these analyses, among patients who presented with high levels of caspase-4 (Figure 3B, red bar), those who were negative for PD-L1 represented 88.1% (Figure 3C) compared to PD-L1 positive patients (Figure 3C, 11.9%), implying a further stratification of NSCLC patients. Moreover, according to chi squared test, caspase-4 and PD-L1 were independent biomarkers as also observed in Figure 3B and 3C where caspase-4 positive and PD-L1 negative patients were the majority of patients in our database.
4. Higher levels of tumor-associated caspase-4 are present in the lung of K-Ras and c-MyC-mutated NSCLC patients.
Another important issue for NSCLC patients is about gene mutation/s (2). In this latter case, we stratified patients who presented EGFR mutation or MET, ROS1 or ALK translocation, herein defined as mutated (Mut+), and compared them to the levels of tumor-associated caspase-4. We found that NSCLC patients who had high levels of tumor-associated caspase-4 (>377 pg/ml) and did not have any gene mutation (Mut-) (Figure 3D, blue line) had lower survival rate (median survival=0.95 years; % of survival rate at 1 year≈35%) (Figure 3E, n=53/74, 71.6%) than patients who had higher levels of tumor-associated caspase-4 and were Mut+ (survival at 1 year≈80%) (Figure 3D, green line; Figure 3E, n=6/74, 8.1%). As observed before, there was a group of patients who were Mut- and had lower tumor-associated caspase-4 (Figure 3D, red line; Figure 3E, n=14/74, 18.9%). This group of patients had a median survival rate of 2.98 years and however, had higher survival rate than patients with higher levels of caspase-4 and Mut- (Figure 3D, blue line). In addition, the survival rate of patients who had Mut+ and high caspase-4 (Figure 3D, green line) was still lower than Mut+ and low caspase-4 (Figure 3D, black line), implying a group of patients (Figure 3E, n=14, 18.9%) where the survival is still low without any relationship with caspase-4. The survival rate of this group at 1 year was of ≈85% compared to the group with high caspase-4 but Mut- (Figure 3D, blue line), that instead presented survival rate at 1 year of ≈35%. Based on these analyses, among caspase-4 positive patients (Figure 3F, red bar) those who were Mut- represented 89.8% (Figure 3F) compared to Mut+ patients (Figure 3F, 10.2%). Moreover, according to chi squared test, caspase-4 and genetic alterations were independent biomarkers as also observed in Figure 3E and F where caspase-4 positive and Mut- patients were the majority of patients in the database.
As shown in the Venn diagram (Figure 2A-B), caspase-4 was associated to genes involved in cell proliferation, such as K-Ras and c-MyC, well-known to be involved in lung carcinogenesis (2, 18, 19). Therefore, we analyzed both K-Ras mutations (G12C, G12D and G12V) and c-MyC expression related to caspase-4 levels in our experimental human samples. We found that NSCLC patients with high levels of tumor-associated caspase-4 and c-MyC positive (c-MyC+) (Figure 3G, black line) had a median survival of 1 year, lower than patients who had lower levels of caspase-4 and c-MyC+ (Figure 3G, green line; median survival=undefined). Patients who presented no positivity to both caspase-4 and c-MyC had longer survival than the other groups (Figure 3G, red line, median survival=3.028). Interestingly, patients with high levels of tumor-associated caspase-4 but who were c-MyC negative (c-MyC-) still had lower survival rate (median survival=2.036 years) (Figure 3G, blue line), although it was higher than patients positive for both Caspase-4 and c-MyC (Figure 3G, black line). The survival rate at 1 year was of 50% for Caspase-4-c-MyC+ patients, of ≈95% for Caspase-4-c-MyC– and of ≈85% for Caspase-4 negative c-MyC+ and Caspase-4 positive c-MyC– patients (Figure 3G). Based on these analyses, patients who presented with high levels of caspase-4 (Figure 3H, red bars) and c-MyC+ were 66.7% in our database (Figure 3I), implying that caspase-4 and c-MyC gene overexpression were strictly associated.
Similarly, we stratified patients according to the positivity to mutated K-Ras (G12C, G12D and G12V). We considered as positive those patients who presented at least one of the three mutations. Very interestingly, the survival rate of caspase-4+ and K-Ras+ patients was significantly reduced (0.97 years) (Figure 3J, black line) compared to patients who were caspase-4- and K-Ras– (Figure 3J, red line). Moreover, both groups of patients who were caspase-4+ and K-Ras– or caspase-4- and K-Ras+ had median survival rate of 2.07 and 3.028 years, respectively, further highlighting the relevance of caspase-4 (Figure 3J and 3K, green and blue line). To date, we found that the majority of patients were caspase-4 + and K-Ras + (Figure 3L, 85.2%), compared to caspase-4 + but K-Ras – patients (Figure 3L, 14.8%).
In addition, we stratified patients as caspase-4, c-MyC and K-Ras triple positive. Patients who were positive for the three targets had a survival rate of 0.98 years (Figure 3M, black line), similar to that observed for caspase-4 and K-Ras + (Figure 3J, black line, median survival=0.97 years) and c-MyC+ (Figure 3G, black line, median survival=1 year) patients. This group of patients represented ≈74.1% (Figure 3N-O). Nevertheless, patients who were solely positive for caspase-4 still represented ≈7.4% of the study population (Figure 3N-O).
These data highlight that caspase-4 collaborates with c-MyC and K-Ras leading to a bad prognosis of NSCLC patients.
5. Caspase-11 facilitates lung tumor progression in mice.
Experimental biological samples showed that capase-4 is related to poor survival of NSCLC patients. To better investigate the molecular mechanism, we went on by taking advantage of a mouse model of adenocarcinoma (13). C57Bl/6 mice (wild type) and Caspase-11 knockout (Caspase 11-ko) were i.t. injected with NMU at week 1-2-3, 8-9-10, 12-13-14 and sacrificed at week 16 (Figure 4A). Very importantly and in support to our previous human data, caspase-11 ko mice robustly developed lower lung tumor lesions than wild type mice (Figure 4B). Because LPS, a TLR4 ligand, was supposed to be able to induce caspase-11 activation via a second signal model (20), to rule out the role of TLR4, we used C3H mice who are defective in TLR4 signalling (21, 22). Surprisingly, C3H mice had similar lung tumor lesions as C57Bl/6 wild type mice (Figure 4C), implying that caspase-11-related pro-tumor activity is directly involved in lung tumor progression in mice. Therefore, because we previously demonstrated that the majority of NSCLC patients were caspase-4 and K-Ras+, we used lung tissues of K-Ras LA1 and K-Ras LA1p53R172HB mice who were previously described by Prof. Quaglino’s laboratory (23) as a lung adenocarcinoma model that shows lung tumor lesions starting from 10 weeks of age. K-RasLA1 mice, similar to C57Bl/6 mice injected with NMU, were characterized by cleaved caspase-11 (p25 fragment) (Figure 4D). Moreover, K-Ras LA1p53R172HB mice, who were characterized by mutated K-Ras and p53 and high lung tumor lesions (23), showed not only p25 but also p10 fragment (Figure 4D). Noteworthy, the precursor form of caspase-11 was higher expressed in these latter mice compared to NMU-injected C57Bl/6, control (Ctr) mice (Figure 4D), implying that not only K-Ras mutation, but also p53 genetic alteration could be implied in caspase-11-related pro-tumor activity in mice.
Caspase-11 was described as involved in the non-canonical inflammasome pathway and associated to IL-1-like cytokine release (10, 11, 13, 17, 20). We found that IL-1β was reduced in the BAL of caspase-11 ko mice at longer time point (16 weeks) (Figure 4E) compared to wild type mice. Instead, as also observed in human samples (Figure 1J), IL-1α was significantly higher in the lung of wild type mice than caspase-11 ko mice (Figure 4F) at all-time points (30-56-116 days), implying a major role of IL-1α. To better understand the relevance of IL-1α vs IL-1β in our experimental conditions, we injected mice with a monoclonal antibody against IL-1α or IL-1β. The neutralization of IL-1α significantly reduced the number of hyperplastic cells in the lung of NMU-injected C57Bl/6 mice (Figure 4G, aIL-1α) compared to the control group (CTR) and mice who were similarly treated with a neutralizing antibody against IL-1β. The IgG isotype control group showed similar hyperplastic cell number (0.133±0.07) as CTR mice (0.139±0.06).
Tumor immunosuppression has been widely described as pivotal for cancer progression (11, 13, 24). Therefore, we moved on to analyze the immune microenvironment in the lung of NMU-treated tumour-bearing mice. The percentage of myeloid-derived suppressor cells (MDSCs; identified as CD11b+Gr-1high) were significantly reduced in the lungs of NMU-treated caspase-11 ko mice compared with wild type (Figure 4H). In contrast, the percentage of T regulatory cells (Treg: identified as CD4+CD25+FoxP3+ cells) was not altered in caspase-11 ko mice compared to wild type (Figure 4I), implying that the adaptive immunity was not affected by the genetic absence of caspase-11. Therefore, we injected NMU-treated Nude mice with isolated CD4+ cells obtained from wild type or caspase-11 ko mice (Figure 4A). We found that NMU-treated Nude mice, although with a tendency to reduction, not statistically significant, had similar lung tumor lesions as wild type (C57Bl/6) mice (Figure 4J). Interestingly, neither the adoptive transfer of wild type CD4+ T cells nor the adoptive transfer of caspase-11 ko CD4+ T cells altered lung tumor lesions (Figure 4J), implying that caspase-11 has no relevance in the adaptive immunity. In contrast, bone marrow transplantation experiments showed that caspase-11 ko mice receiving wild type bone marrow-derived cells (Figure 4K) had the lowest number of lung tumor lesions compared to wild type mice receiving ko bone marrow-derived cells (Figure 4K; ko into wt vs wt) and wild type lung tumor-bearing mice (Figure 4K; wt into ko vs casp 11 ko). To note, NMU-treated caspase-11 ko mice receiving wild type bone marrow-derived cells had similar levels of tumor lesions as NMU-treated caspase-11 ko mice (Figure 4K, wt into ko vs casp 11 ko), implying that the structural caspase-11 is involved in lung cancer progression in mice. Nevertheless, NMU-treated wild type mice receiving ko bone marrow-derived cells (Figure 4K; ko into wt) still had lower lung tumor lesions compared to NMU-treated wt mice, although the tumor lesions was higher than those in caspase-11 ko mice receiving wild type bone marrow-derived cells (Figure 4K; ko into wt vs wt into ko). Taken altogether, these data suggest that caspase-11 is involved in lung tumor progression in mice influencing the innate immunity most likely via IL-1α.
6. Large subunit of Caspase-4 facilitates tumor cell proliferation.
The above data imply that caspase-4 in the structural cell component of lung tissues can favor tumor formation and progression. Therefore, to go inside the molecular mechanism we transfected A549 cells, mimicking lung epithelial cells, with viral vectors to express the long structure of caspase-4 (mRNA: 74-1205 nucleotides, nt, CARD+LARGE+SMALL subunit, PC4-1), the structure of the protein with CARD+LARGE subunit (mRNA: 74-810 nt, PC4-2), the structure of the protein with LARGE+SMALL subunit (mRNA: 348-1205 nt, PC4-3) and the structure of the protein with the sole LARGE subunit (mRNA: 423-886 nt, PC4-4). Lung epithelial cells transfected with PC4-1, PC4-2 and PC4-3 showed higher proliferation rate than control and empty vector-transfected cells (Figure 5A). Because the three vectors had a common nucleotidic sequence which corresponded to the LARGE subunit and because both human and mouse samples showed p25 fragment of caspase-4 and caspase-11 in lung homogenates, we investigated the role of this subunit. Interestingly, we found that cells transfected with PC4-4, which solely contained nucleotides for the LARGE subunit, robustly proliferated compared to control and empty vector transfected cells, but also to PC4-1, PC4-2 and PC4-3 transfected cells (Figure 5A). Similarly, treatment of cells with the recombinant protein that mimicked the large subunit of caspase-4 increased cell proliferation compared to the control (Figure 5B; white violin plot). Moreover, the addition of NSCLC patient-derived PBMCs did not alter the proliferation of cells (Figure 5B, green bars), further highlighting what was previously observed in mice about the role of caspase-11 in lung structural cells.
In order to understand the molecular mechanism underlying caspase-4-induced cell proliferation, we treated cells with cetuximab, a monoclonal antibody against EGFR. Cell proliferation was not altered when cetuximab was added to caspase-4-treated cells (Figure 5C), implying that EGFR signaling was not involved. Similarly, the inhibition of histone deacetylase (HDAC), highly involved in lung cancer (27), by means of SAHA (Figure 5D) and of DNA methyltransferase by means of 5-AZA (Figure 5E) did not modify caspase-4-induced cell proliferation, implying that caspase-4 is unlikely to be involved in epigenetic modulation/s that lead to tumor cell proliferation (28). In sharp contrast, the inhibition of K-Ras by means of FTI significantly reduced caspase-4-induced cell proliferation (Figure 5F, red bars). In support, we found that caspase-4 and K-Ras co-immunoprecipitated in lung tumor tissues (Supplementary Figure 1). To rule out another proliferative pathway, we treated cells with rapamicin, mTOR inhibitor. Again, we did not find an alteration in caspase-4-induced cell proliferation (Figure 5G, blue bars). These data suggest that the large subunit of caspase-4 induces cell proliferation via K-Ras pathway. In support, we found that caspase-4+ and K-Ras mutated patients had lower survival rate than other NSCLC patients, supporting the biochemical analyses.