HOXA9 Transcriptionally Promotes Apoptosis and Represses Autophagy by Targeting NF-κB in Cutaneous Squamous Cell Carcinoma

Tumor suppressor HOXA9 has been identified to promote apoptosis in cutaneous squamous cell carcinoma (cSCC). However, the mechanism of such pro-apoptotic role of HOXA9 remains obscure. KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis of RNA-seq data showed that NF-κB, apoptosis and autophagy pathways are significantly regulated after HOXA9 knockdown. HOXA9 transcriptionally regulates RELA, the p65 subunit of NF-κB. Loss of HOXA9 in cSCC significantly upregulates RELA expression and thus enhances NF-κB pathway. Interestingly, RELA transcriptionally promotes not only anti-apoptotic factor BCL-XL but also autophagic genes including ATG1, ATG3, and ATG12. Our results reveal an enhanced NF-κB signaling network regulated by HOXA9, which contributes to repressed apoptosis and activated autophagy in cSCC development and may represent an intervention target for cSCC therapy.


Introduction
Cutaneous squamous cell carcinoma (cSCC) is the second most common type of cancer with an annual incidence of over one million globally [1][2][3]. Chronic ultraviolet (UV) exposure was suggested to be responsible for DNA damage of normal keratinocytes in the epidermis, which leads to the development of skin cancers including cSCC [2,4]. However, the detailed underlying molecular mechanisms for this transition still need to be fully elucidated.
Apoptosis is one type of programmed cell death, which is initiated by intrinsic or extrinsic signals and finely-tuned by many factors including BCL-2 family proteins [5]. BCL-2 family members, including both anti-apoptotic and pro-apoptotic BCL-2 family proteins, govern mitochondrial outer membrane permeabilization (MOMP) to regulate apoptosis. When apoptotic cell death program starts, MOMP releases cytochrome c to cytosol and activates caspase-family proteases for initiating apoptotic events. Tumors develop effective mechanisms to resist apoptosis and thus become one of the hallmark features of cancer [6]. The lead member BCL-2 and its homologue BCL-XL (BCL2L1) are featured by the presence of four conserved BH domains, three of which, BH1, BH2, and BH3 domains, are responsible for their antiapoptotic role mainly by interactions with other proapoptotic members of the BCL-2 family [7,8]. Inhibiting the activities of anti-apoptotic BCL-2 members by drugs [9] or neutralizing by BH3 peptides [10] to prime cancer cells to death has been regarded as effective therapeutic modalities. In another way, transcriptionally or post-transcriptionally repressing the expression levels of anti-apoptotic BCL-2 members in cancer is also promising. However, the mechanism regulating the expression of BCL-2 family members still remain unclear.

RNA-Seq
Total RNA Sequencing was done by Novogene using the Illumina HiSeq 2500. Tophat2 was used for the alignment (http://ccb.jhu.edu/software/tophat) and human assembly GRCh37/hg19 was used as the reference genome. Significant differentially-expressed genes were detected by HTSeq (http://www-huber.embl.de/HTSeq). On the website of DAVID Bioinformatics Resources 6.8 (http://david.abcc.ncifcrf.gov/), we performed Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of the differentially expressed genes.

Xenograft Mouse Model
Briefly, male nude mice (BALB/C-nu/nu, 4 weeks old) were used in this study. 1.0 × 10 7 cells were subcutaneously injected into the left and right flanks of the above nude mice. After implanted xenografts grows to around 100 mm 3 in size, pre-incubated 3 nM siRNAs including siNC and siHOXA9 were respectively injected into the xenografts and the injection was repeated every three day. The handling persons were "blind" to the different treatments. Tumor growth curve was generated by measuring the tumor diameters every day. At the end of experiment (21 days after the first siRNA injection), the xenografts were dissected and snap-frozen immediately for later RNA and protein extractions or paraffin-embedded for IHC staining.

Cell Proliferation Assay
4000/well cells were seeded on 96-well plates and transfected with siRNAs. Cell proliferation was done after 0, 24, 48, and 72 h by cell counting kit (TransGen Biotech, Beijing, China) at 450 nm.

Apoptosis Assay
Cells on 60-mm dishes were transfected with siNC, siHOXA9_01, or siHOXA9_02 and cultivated for 36 h. Cell apoptosis was detected using TransDetect Annexin V-FITC/PI cell apoptosis detection

Statistical Analysis
Statistical analysis was done using an unpaired t-test or one-way ANOVA tests (SPSS v20). All statistical tests incorporated two-tailed tests and homogeneity of variance tests and were thought to indicate significant difference if * p < 0.05, ** p < 0.01, or *** p < 0.001.

Data Availability
All data generated or analyzed in this study are included in this published article and its Supplementary data, which are also available from the corresponding author on request.

HOXA9 Is Predicted to Regulate Apoptotic-and Autophagic-Genes in cSCC
HOXA9 has been identified as a tumor suppressor in cSCC by inhibiting glycolysis while promoting apoptosis [15]. Yet, the exact roles of HOXA9 in regulating apoptosis process is still unclear. To understand how HOXA9 regulates the molecular events related with apoptosis and other cellular processes in cSCC cells, we repeated the bioinformatic analysis of the previous transcriptome sequencing after HOXA9 knockdown [15]. Gene Ontology (GO) analysis with the list of significantly up-regulated genes revealed that the top-ranked lists of enriched Gene Ontology categories includes "Positive regulation of apoptotic process", "Apoptotic process", "Regulation of apoptotic process", "Regulation of extrinsic apoptotic signaling pathway via death domain receptors", "Positive regulation of programmed cell death", "Positive regulation of NF-kappaB transcription factor activity", "Positive regulation of I-kappaB kinase/NF-kappaB signaling", "Macroautophagy" and "Positive regulation of transcription, DNA-templated", etc. (p < 0.05, Table 1, Supplementary Data 1). Markedly, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis indicated that molecular signaling pathways including "NF-kappaB signaling pathway", "Apoptosis" and "Autophagy" are significantly influenced (p < 0.05, Table 1, Supplementary Data 1). Among the genes affected in the above three pathways, significantly-upregulated genes including NF-κB, BCL2L1 (BCL-XL), ULK1 (ATG1), ATG3, and ATG12 were functionally relevant to apoptosis or autophagy. Table 1. Transcriptomic analysis of HOXA9-regulated genes by RNA-Seq in cutaneous squamous cell carcinoma (cSCC) cells. Over-represented categories by GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis of differently-expressed genes. BP: biological process. The "NF-kappaB signaling pathway", "Apoptosis" and "Regulation of autophagy" were highlighted.

HOXA9 Promotes Apoptosis While Inhibits Autophagy
To verify the apoptosis-and autophagy-regulatory roles of HOXA9 supported by the above bioinformatic analysis of transcriptomic sequencing, we performed assays to determine the variations of apoptosis after HOXA9 expression alterations. We did gain-of-and loss-of-functional tests and confirmed that depletion of HOXA9 enhances the proliferation and decreases the apoptosis of cSCC cells as shown by repressed cleaved CASPASE3 generation and apoptotic cells (Figure 1a-c), while overexpression of HOXA9 indeed plays tumor-suppressive roles by inhibiting the proliferation and promoting apoptosis in cSCC cells (Figure 1d-f), which is consistent with the previous report [15].

HOXA9 Promotes Apoptosis while Inhibits Autophagy
To verify the apoptosis-and autophagy-regulatory roles of HOXA9 supported by the above bioinformatic analysis of transcriptomic sequencing, we performed assays to determine the variations of apoptosis after HOXA9 expression alterations. We did gain-of-and loss-of-functional The status of autophagy in response to HOXA9 variation was also checked. Knockdown of HOXA9 enhances autophagy as shown by the increased LC3B II modification, P62 expression ( Figure 1a) and LC3B immunofluorescence (Figure 1g), while overexpression of HOXA9 significantly inhibited autophagy ( Figure 1d). Thus, we concluded that loss of HOXA9 inhibits apoptosis but promotes autophagy in cSCC cells.
3.3. HOXA9 Negatively Regulated the Expression of RELA, BCL-XL, ULK1, ATG3, and ATG12 To further validate the pro-apoptotic and anti-autophagic functions of HOXA9, RELA (encoding P65 subunit of NF-κB), BCL-XL, ULK1, ATG3, and ATG12 were selected from the significantly varied genes owing to their critical roles in apoptosis and autophagy. qRT-PCR and western blot detection confirmed that the upregulation of RELA, BCL-XL, ULK1, ATG3, and ATG12 in response to HOXA9 knockdown (Figure 2a,b). Conversely, overexpression of HOXA9 inhibited the expression of RELA, BCL-XL, ULK1, ATG3, and ATG12 (Figure 2c,d). Collectively, HOXA9 regulates apoptosis and autophagy by negatively regulating anti-apoptotic and pro-autophagic genes. proliferation and promoting apoptosis in cSCC cells (Figure 1d-f), which is consistent with the previous report [15]. The status of autophagy in response to HOXA9 variation was also checked. Knockdown of HOXA9 enhances autophagy as shown by the increased LC3B II modification, P62 expression ( Figure  1a) and LC3B immunofluorescence (Figure 1g), while overexpression of HOXA9 significantly inhibited autophagy (Figure 1d). Thus, we concluded that loss of HOXA9 inhibits apoptosis but promotes autophagy in cSCC cells.

HOXA9 Negatively Regulated the Expression of RELA, BCL-XL, ULK1, ATG3, and ATG12
To further validate the pro-apoptotic and anti-autophagic functions of HOXA9, RELA (encoding P65 subunit of NF-κB), BCL-XL, ULK1, ATG3, and ATG12 were selected from the significantly varied genes owing to their critical roles in apoptosis and autophagy. qRT-PCR and western blot detection confirmed that the upregulation of RELA, BCL-XL, ULK1, ATG3, and ATG12 in response to HOXA9 knockdown (Figure 2a,b). Conversely, overexpression of HOXA9 inhibited the expression of RELA, BCL-XL, ULK1, ATG3, and ATG12 (Figure 2c,d). Collectively, HOXA9 regulates apoptosis and autophagy by negatively regulating anti-apoptotic and pro-autophagic genes.  The qRT-PCR data were normalized to GAPDH gene expression. In western blots, GAPDH was used as a loading control. The bands of BCL-XL and ATG12 were densimetrically quantified (n = 3). (e) Predicted binding site of HOXA9 (diamond) at the promoter of RELA (p65) or binding sites of RELA for on the promoter regions of BCL-XL, ULK1, ATG3, and ATG12 by rVista (https://rvista.dcode.org/). (f) The binding enrichment of HOXA9 at the RELA locus or RELA at the loci of BCL-XL, ULK1, ATG3, and ATG12 was detected by ChIP-qPCR after knockdown of HOXA9. One-Way ANOVA and Dunnett's multiple comparison test. Means ± s.d., * p < 0.05, ** p < 0.01, *** p < 0.001.

HOXA9 Inhibits the Epigenetic Activities of NF-κB
HOXA9 is a known transcription factor functioning in transcriptionally regulating gene expression [16]. To explore how HOXA9 regulates the above apoptosis-and autophagy-related genes, the predicted promoter regions (+1000 bp to −1000 bp relative to the transcription start site [TSS]) of RELA, BCL-XL, ULK1, ATG3, and ATG12 were analyzed according to a minimum consensus binding site (ATAA) via the UCSC genome browser (http://genome.ucsc.edu/) [17] and rVista (https://rvista.dcode.org/) [18]. Interestingly, HOXA9 binding site was only predicted to be present on the promoter of RELA, but not on BCL-XL, ULK1, ATG3, and ATG12 loci (Figure 2e). Seeing the expression levels of BCL-XL, ULK1, ATG3, and ATG12 are responsive to HOXA9 overexpression and knockdown, it promoted that they are potentially regulated by RELA. As anticipated, NF-κB banding site (aGGGgaTTTCCaxxx) was predicted to be presented on the promoter regions of BCL-XL, ULK1, ATG3, and ATG12 genes (Figure 2e). Based on the above predictions, chromatin immunoprecipitation (ChIP)-qPCR detection was performed and confirmed that depletion of HOXA9 by RNA interference led to marked decrease in the binding enrichment of HOXA9 at the binding sites of RELA locus (Figure 2f) while the binding enrichments of RELA on BCL-XL, ULK1, ATG3, and ATG12 loci are significantly enhanced (Figure 2f), which is consistent the upregulated expression of RELA, BCL-XL, ULK1, ATG3, and ATG12 after HOXA9 knockdown (Figure 2a,b). The above results clearly proved that HOXA9 negatively regulates BCL-XL, ULK1, ATG3, and ATG12 through transcriptionally-repressive regulation of RELA.

Inhibition of RELA Decreased the Expression of BCL-XL, ULK1, ATG3, and ATG12
To verify if HOXA9 knockdown could lead to increased NF-κB expression and enhanced NF-κB activity, IF staining of RELA protein was performed. Total RELA protein expression was strongly increased and enhanced RELA translocation to nucleus was apparently detected, which is comparable to TNF-α induction (Figure 3a). To validate if BCL-XL, ULK1, ATG3, and ATG12 are regulated by RELA, an inhibitor of NF-κB transcriptional activity, JSH-23, was used to treat cSCC cells. As predicted, JSH-23 treatment significantly repressed the expression of BCL-XL, ULK1, ATG3, and ATG12 in both mRNA and protein levels (Figure 3b,c). Interestingly, knockdown of HOXA9 could partially rescue RELA expression and thus promote BCL-XL and ATG genes expression in mRNA and protein levels, which verifies the expression level is critical for RELA in regulating downstream genes expression (Figure 3d,e). Thus, our results demonstrated that NF-κB directly promotes the expression levels of downstream BCL-XL, ULK1, ATG3, and ATG12 genes.

Loss of HOXA9 Inhibits Apoptotic geNes While Enhances Autophagic Genes In Vivo
To evaluate the pro-apoptotic and anti-autophagic roles of HOXA9 in vivo, a xenograft tumor model was established in immunocompromised mice. Knockdown of HOXA9 strongly promoted tumor growth and the tumor sizes were apparently larger compared to siNC group at the end of the experimental period (Figure 4a). HOXA9 depletion in the siHOXA9 group was verified by western blot, qPCR and IHC staining (Figure 4b,c). Furthermore, HOXA9 depletion led to the significant up-regulation of RELA, BCL-XL, ULK1, ATG3, and ATG12 in both protein and RNA levels as shown in western blot, qPCR detection and IHC staining (Figure 4c-e). Collectively, our in vivo experiments clearly indicated that HOXA9 promotes apoptosis and represses autophagy by negatively regulating NF-κB and its downstream anti-apoptotic and pro-autophagic genes including BCL-XL, ULK1, ATG3, and ATG12.

Loss of HOXA9 Inhibits Apoptotic geNes while Enhances Autophagic Genes In Vivo
To evaluate the pro-apoptotic and anti-autophagic roles of HOXA9 in vivo, a xenograft tumor model was established in immunocompromised mice. Knockdown of HOXA9 strongly promoted tumor growth and the tumor sizes were apparently larger compared to siNC group at the end of the experimental period (Figure 4a). HOXA9 depletion in the siHOXA9 group was verified by western blot, qPCR and IHC staining (Figure 4b,c). Furthermore, HOXA9 depletion led to the significant upregulation of RELA, BCL-XL, ULK1, ATG3, and ATG12 in both protein and RNA levels as shown in western blot, qPCR detection and IHC staining (Figure 4c-e). Collectively, our in vivo experiments clearly indicated that HOXA9 promotes apoptosis and represses autophagy by negatively regulating NF-κB and its downstream anti-apoptotic and pro-autophagic genes including BCL-XL, ULK1, ATG3, and ATG12.  In cSCC tumors, loss of HOXA9 up-regulates NF-κB and its downstream anti-apoptotic gene BCL-XL and pro-autophagic genes of ULK1, ATG3, and ATG12, which contributes to the repressed apoptosis, enhanced autophagy and promotes cSCC progression. * p < 0.05, ** p < 0.01, or *** p < 0.001.

Discussion
As one type of programmed cell deaths, apoptosis is characterized by a series of morphological changes including nuclear condensation, DNA fragmentation and formation of apoptotic bodies. Autophagy is generally regarded as a pro-survival cellular process by controlling the turnover of In cSCC tumors, loss of HOXA9 up-regulates NF-κB and its downstream anti-apoptotic gene BCL-XL and pro-autophagic genes of ULK1, ATG3, and ATG12, which contributes to the repressed apoptosis, enhanced autophagy and promotes cSCC progression. * p < 0.05, ** p < 0.01.

Discussion
As one type of programmed cell deaths, apoptosis is characterized by a series of morphological changes including nuclear condensation, DNA fragmentation and formation of apoptotic bodies. Autophagy is generally regarded as a pro-survival cellular process by controlling the turnover of intracellular organelles and proteins. In most cases, autophagy stands as a survival strategy to deal with uncomfortable stresses [19]. Stresses often induce both of autophagy and apoptosis in the same cell, generally, autophagy appears earlier than apoptosis. However, if apoptosis is onset, autophagy will be inactivated [20]. The molecular mechanisms under such interplay between apoptosis and autophagy is debating and still under exploration [19].
HOXA9 belonging to HOX gene family is primarily identified to regulate embryonic development, maintain hematopoietic stem cells and play roles either as an oncogene or as a tumor suppressor in various tumors [21][22][23][24]. HOXA9 is previously identified to be significantly downregulated in cSCC, which contributes to the repressed apoptosis of cSCC cells [15]. In this study, bioinformatic analysis of the transcriptomic sequencing data after HOXA9 knockdown indicated that HOXA9 functions in regulating NF-κB pathway, apoptosis and autophagy (Table 1). Apoptosis-related genes including RELA and BCL-XL were significantly upregulated in response to HOXA9 depletion. Further study the upregulation of BCL-XL is not directly regulated by HOXA9, but by RELA, which also directly modulates the expression of ULK1, ATG3, and ATG12. Thus, RELA mediates the HOXA9 s regulation of BCL-XL, ULK1, ATG3. and ATG12.
The NF-κB family includes five DNA-binding proteins which can form homodimers or heterodimers and are critical regulators of cancer development [25]. Generally, upregulation or activation or of NF-κB enhance cell proliferation, inhibit apoptosis, stimulate cell migration and invasion [26]. RELA (p65) bind p50 to form heterodimer and mediates the transcriptional regulation of classical NF-κB signaling pathway to promote cell proliferation, survival, angiogenesis and metastasis [25]. Although HOXA9 has been reported to regulate RELA expression in Non-small Cell Lung Cancer, no report has demonstrated such regulation is direct or indirect [27]. Our work provides evidences to demonstrate the direct binding of HOXA9 on RELA promoter and deepen the understanding of HOXA-RELA regulator axis, which highlights the tumor-suppressive role of HOXA9 in cSCC.
The BCL-2 family includes important apoptosis regulators. The balanced expression between pro-apoptotic members (BAX, BAK, BAD, BIK and BID) and antiapoptotic members (BCL-2, BCL-XL, BCL-W and MCL-1) maintains the homeostasis and survival of cells [28]. Disruption of the above balance, especially the repression of the antiapoptotic members, will lead to apoptosis. BCL-XL contains conserved topology of BCL-2 family and binds with pro-apoptotic ligands to promote pro-survival features [29]. The regulation of BCL-XL by NF-κB has been reported by previous reports [30,31]. In this study, we further identified the transcriptional regulatory binding site that the NF-κB p65 subunit, RELA, directly binds and promote the expression of BCL-XL. ULK1 (ATG1) is critical for the initiation of autophagy by perceiving the nutrient-related cellular status, recruiting ATG proteins to form autophagosomes [32]. ATG3 is an E2-like conjugating enzyme mediating LC3 lipidation while acts as alternative conjugate of ATG12. ATG12-ATG3 complex not only regulates basal autophagy, also connects the late endolysosomal trafficking [33].
As BCL-2 family proteins and ATG proteins are critical factors for apoptosis or autophagy respectively, their turnover mechanisms are of profoundly interest. Also, one open question is how the two critical survival processes, apoptosis and autophagy, are connected and balanced by any molecular switcher or regulator? The HOXA9-NF-κB regulatory axis identified demonstrates how one tumor suppressor, HOXA9, negatively enhances apoptosis and suppresses autophagy through transcriptionally regulating NF-κB. Regulating both apoptotic and autophagic factors at the same time by NF-κB is an important novel finding in this study, which primarily give the answer for the above question.
Collectively, our findings establish a novel HOXA9-NF-κB regulatory axis that functions in regulating NF-κB-mediated apoptosis and autophagy in cSCC. As depicted in Figure 4f, HOXA9 is down-regulated in both cSCC cells [15]. Loss of HOXA9 up-regulates NF-κB and its downstream genes including BCL-XL, ULK1, ATG3, and ATG12, which regulate apoptotic and autophagic pathway and contributes to the survival of cSCC cell and cSCC progression. HOXA9 directly bind the promoter and negatively regulates the expression of NF-κB is critical for HOXA9 s indirect modulation of apoptotic and autophagic pathways. As the mediator, NF-κB directly binds the promoters of BCL-XL, ULK1, ATG3, and ATG12 genes to inhibit apoptosis and enhance autophagy at the same time, which executes the pro-survival function. Our findings highlight the pro-apoptotic and anti-autophagic roles of HOXA9 in skin tissue, support our previous finding that HOXA9 acts as tumor suppressor in cSCC, and emphasize a newly identified HOXA9-NF-κB axis that may provide novel intervention targets for cSCC therapy.