Computational insights into CRISP3 downregulation in cervical cancer and its cervical lineages pattern

Abstract Background Cysteine-rich secretory protein 3 (CRISP3) emerges as a potential biomarker in the study of many cancers, including cervical cancer (CC). This study aimed to analyze the expression pattern of CRISP3 in CC patients and CC cell lineages, following treatment with the epigenetic drugs: trichostatin A (TSA) and 5-aza-2'-deoxycytidine (5-aza). Methods The differentially expressed genes identified in GSE63514 were used to construct a protein–protein interaction network. CRISP3 was selected for subsequent analyses. We utilized data from the TCGA and GENT2 projects to evaluate the expression profile and clinical behavior of CRISP3. Additionally, we conducted cell culture experiments to analyze the expression profile of CRISP3 in cells. Results Low levels of CRISP3 were observed in squamous cell carcinoma (SCC) and human papillomavirus (HPV)16+, along with being associated with worse overall survival (OS). MIR-1229–3p was analyzed, and its high expression was associated with worse prognostic outcomes. In CC-derived cell lines, we observed low levels of CRISP3 in SiHa, followed by SW756, C33A, HeLa, and higher levels in CaSki. All cells were treated with TSA, 5-aza, or both. In all cell lines, treatment with TSA resulted in increased transcription of CRISP3. Conclusion We identified a significant downregulation of CRISP3 in CC, particularly in cases with HPV16 infection and SCC, which was associated with poorer OS. Preliminary findings suggest that epigenetic treatments with TSA and 5-aza may modulate CRISP3 expression, warranting further research to elucidate its regulatory mechanisms and potential as a prognostic biomarker.


Gr aphical Abstr act Introduction
Cervical cancer ( CC ) is one of the most lethal gynecological neoplasms and the fourth most pr e v alent cause of morbidity among w omen w orldwide.Accor ding to an analysis conducted by the Global Cancer Observatory ( GCO ) , in 2020, there were 604 127 new cases and 341 831 deaths reported due to CC, with most of these numbers coming from low-and middle-income countries [ 1 ].Furthermore, 99.7% of cervical carcinomas are due to high-risk human papillomavirus infection and persistence, considered the primary etiological agent of CC [ 2 ].
Despite advances in preventive measures such as human papillomavirus ( HPV ) v accination, the v ariable global cov er a ge, and the lack of efficacy for those already diagnosed or in advanced stages of the disease underscore the need for continued research into nov el ther a peutic tar gets.In this context, the cysteine-ric h secr etory pr otein 3 ( CRISP3 ) emerges as a potential biomarker.Initially identified as the 28 kDa specific gr anule pr otein ( SGP28 ) , CRISP3 was first isolated in 1996 from exocytosed material of human neutrophils, marking a significant discovery in the study of secr etory pr oteins [ 3 , 4 ].Situated on c hr omosome 6p12.3, the CRISP3 gene encodes an extracellular matrix protein consisting of 245 amino acids.Physiologically, CRISP3 is found in abundance in plasma, sweat, seminal fluid, and exocrine glands such as the pancr eas, saliv ary glands, and pr ostate.Conv ersel y, lo w er le v els of CRISP3 are observed in the e pidid ymis , testicles , colon, o vary, lacrimal glands, and thymus [ 5 , 6 ].CRISP3 is stored intracellularly in eosinophil and neutrophil granules, in glycosylated or nonglycosylated form, and is secreted for innate host defense, succeeding in the degradation of the extracellular matrix through pr oteol ytic enzymes [ 5 , 3 ].Although the exact function of CRISP3 remains unknown, its structural similarity to proteins involved in pathogenic processes, along with its occurrence in exocrine secretions of epithelial tissues, suggests its involvement in immune responses and inflammatory processes, as well as indicating a potential role in the pathophysiology of various diseases [ 6 , 7 ].
In the prostate, CRISP3 levels are reduced; ho w ever , in cancerous contexts of this tissue , o v er expr ession is observed, with significantl y incr eased le v els r anging fr om 20 to 2 000 times mor e [ 8 ].In the context of prostate cancer, the CRISP3 promoter is regulated by the andr ogen r eceptor ( AR ) thr ough an epigenetic mechanism, and CRISP3 expression is induced by the presence of dihydr otestoster one ( DHT ) in AR-positive cells .T his is evidenced by CRISP3 pr omoter activity, whic h was r educed in cells cultur ed in ster oid-fr ee medium and was r estor ed after treatment with DHT [ 9 ].In lung cancer, high le v els of CRISP3 are observed, and even higher le v els ar e seen after c hemother a py tr eatment, corr elating with a metastatic profile [ 10 ].
On the other hand, a study demonstrated CRISP3 in an axis with miR-508-5p and LINC01342, where both CRISP3 and LINC01342 silencing resulted in decreased cell growth capacity, colon y formation, inv asion, and metastasis [ 11 ].In breast neoplasms , CRISP3 o v er expr ession is associated with tumor spread and se v erity [ 12 ].Similarl y, in esopha geal cancer, high le v els of CRISP3 have been associated with cell proliferation and metastases, leading to poor pr ognosis [ 7 ].Imm unostaining in ovarian epithelium, benign tumors, and ovarian cancer ( OC ) specimens did not show variations for CRISP3 protein distribution [ 13 ].Conv ersel y, another study found that transcriptional reduction of CRISP3 is an independent prognostic factor and is associated with lo w er ov er all surviv al ( OS ) of OC patients [ 14 ].In or al squamous cell carcinoma, a decrease in CRISP3 expression was observed, especially in early stages, where loss of CRISP3 DN A cop y number was detected [ 15 ].In the context of CC, although only one stud y ad dr essed CRISP3, ther e was no specific attention dir ected to this gene.In this study, normal tissues, low-and high-grade intraepithelial lesions, as well as squamous cell carcinoma were examined through immunohistochemistry .Notably , only normal tissues sho w ed positive staining for CRISP3 [ 16 ].
To enhance our understanding of the relationship between CRISP3 and CC, acknowledging that there is no fixed pattern suggesting a consistent role for CRISP3 in the tumorigenic process and that it appears to vary depending on the tumor type and stage of tumorigenesis, we e v aluated CRISP3 expr ession in CC tissues and examined its prognostic significance using public databases.We extended our investigation by assessing CRISP3 expression in r epr esentativ e cervical tumor-derived cell lines.

Selection of datasets and analysis of differential genes expression
The gene expression datasets under analysis were sourced from the Gene Expression Omnibus database ( https://www.ncbi.nlm.nih.gov/ geo/ ) .Specifically focusing on datasets associated with CC, an exhaustiv e searc h was conducted within the database, and after meticulous screening, we identified the micr oarr ay dataset GSE63514 [ 17 ].Employing GEO2R ( www.ncbi.nlm.nih.gov/geo/geo2r) , we screened and isolated differ entiall y expr essed genes ( DEGs ) based on a significance thr eshold, wher e Benjamini-Hoc hber g < 0.05, and | logFC |≥1.5 ( normal vs. tumor ) .A box plot ( s upplementary Fig. S1 , see online supplementary material ) was constructed with GEO2R to standardize and assess the dataset samples.Additionall y, to visuall y r epr esent the DEGs, a volcano plot was generated ( Fig. 1 A ) , providing a graphical overview of the gene expression alterations.From this analysis, the probe that display ed the lo w est adjusted P -value ( p.adj ) w as singled out for an in-depth investigation into its associations with the clinicopathological features of patients diagnosed with CC, utilizing data from The Cancer Genome Atlas ( TCGA ) study.

Function and pathway enrichment analysis and pr otein-pr otein interaction network
The top 100 DEGs were used in the STRING database ( https:// www.string-db.org/) to gener ate the pr otein-pr otein inter action ( PPI ) network and enrichment pathway [ 18 ].A confidence interaction score of 0.4 w as emplo y ed as the significance threshold.Subsequently, the PPI network was visualized utilizing Cytoscape software v3.10.1 ( www.cytosca pe.or g/) [ 19 ].

TCGA and GENT2 data analysis
T he cBioP ortal platform ( https:// www.cbioportal.org/) provided access to the Firehose Legacy TCGA public data, from which the mRNA CRISP3 expression profiles of 310 patients were acquired.Among these, 4 cases lacked expression information and were consequentl y excluded fr om the study, r esulting in a cohort of 306 patients eligible for e v aluation.After this exclusion, an integration of the expression data with clinical-pathological information was conducted to investigate the correlation between CRISP3 gene expression and various clinical and pathological parameters.Median gene expression values were computed to stratify the patient groups into low ( ≤median ) and high ( > median ) .Following this classification, a contingency table was constructed, and statistical analysis was performed employing either the chisquared test or Fisher's exact test, as a ppr opriate .T he survival analysis utilized the Kaplan-Meier ( KM ) method and log-rank test, employing the same 'high' and 'low' stratification.For a comprehensi ve quantitati ve examination of CRISP3 expression profiles, an initial assessment was made to ascertain the Gaussian distribution of the data.Subsequently, depending on the data distribution c har acteristics, statistical anal yses suc h as the t-test or Mann-Whitney test were applied for paired variables, while analysis of variance ( ANOVA ) or Kruskal-Wallis tests were employed for comparisons among three or more groups .T he presentation of CRISP3 expression involved log-transformed mRNA expression z-scor es, whic h wer e juxta posed with the ov er all expr ession distribution of all samples ( RNA Seq V2 RSEM ) .Additionally, the beta value ( HM450 ) was utilized to indicate the DNA methylation le v els of individual CpG sites in comparison to the corresponding mRNA expression, as described in previous studies [ 20 , 21 ].GENT2 provides differential expression analysis and prognosis assessment based on normal and tumor tissues, including cell lineages.All data in GENT2 is sourced from TCGA and GEO databases [ 22 ].The k e yw or d CRISP3 was utilized to access mRNA expression data in tissues and CC cells.Expression data files for GPL570 platform ( HG-U133_plus_2 ) and GPL96 platform ( HG-U133A ) were acquir ed, and statistical anal ysis was performed.To e v aluate the difference between normal and tumor tissue, the Mann-Whitney test was employed.

microRNA prediction, and survival association
Detection and analysis of microRNAs ( miRNAs ) that target CRISP3 was performed using the mirTarBase database, ( mirtarbase.mbc.nctu.edu.tw ) which contains miRNAs experimentall y v alidated.In addition, miR W alk ( zmf.umm.uniheidelberg.de/) , w as also emplo y ed to predict miRN As b y anal yzing statisticall y significant inter actions ( P < 0.05 ) [ 23 , 24 ].The prognostic value of selected miRNAs was determined using the TCGA database ( Firehose Legacy study ) .All microRNAs were e v aluated for ov er all surviv al ( OS ) and r ela pse-fr ee surviv al ( RFS ) across all subtypes, including cervix squamous cell carcinomas.For statistical analysis, the KM method with log-rank test comparison was applied by selecting the optimal cutoff from the CC TCGA cohort.

Cell culture and 5-aza-2'-deoxycytidine and trichostatin A treatment
The following cervical cancer-derived cell lines were utilized: C33A ( HPV -negative; A TCC ® CRM-HTB-31™) , SiHa ( HPV16; A TCC ® HTB-35™) , SW756 ( HPV18; ATCC ® CRL-10302TM ) , HeLa ( HPV18; A TCC ® CCL-2™) , and CaSki ( HPV16; A TCC ® CRL-1550™) ; all cell lines were cultured in Minimum Essential Medium ( MEM ) ( GibcoTM, Invitrogen, USA ) with 10% fetal bovine serum ( FBS ) ( Gibco ) .Authentication and mycoplasma testing were conducted for all cell lines used.Maintenance was carried out at 37 • C and 5% CO 2 ( Thermo Electron Corporation, USA ) .Prior to the com-mencement of the expanded experiment across all cell lines, preliminary toxicity assays were conducted for trichostatin A ( TSA ) and 5-aza-2'-deoxycytidine ( 5-AZA ) to determine non-toxic drug concentrations for use in subsequent treatments.Based on these assa ys , uniform drug concentr ations wer e selected for the treatment of SiHa, HeLa, SW756, and CaSki cell lines.Tr eatments wer e administered either as single-agent or in combination.Ho w e v er, for the C33A cell line, which also received the same treatments, the TSA dose was halved in the combined tr eatment r egimen to mitigate toxicity.The treatment dosages were as follows: for SiHa, HeLa, SW756, and CaSki cell lines, the TSA concentration was set at 75 nM, 5-AZA at 10 μM, and for the combined treatment of 5-AZA with TSA, the concentration was 7.5 μM of 5-AZA plus 75 nM TSA.In contrast, the C33A cell line was treated with a reduced TSA concentration of 37.5 nM, with 5-AZA remaining at 10 μM, and for the combined treatment, the concentration was 7.5 μM of 5-AZA plus 37.5 nM TSA.Due to the compound's short half-life in culture media, the 5-AZA medium was prepared and replaced daily for 7 da ys .

RN A isola tion and re verse tr anscription-pol ymer ase chain reaction
RNAs extracted from the cell cultures were isolated using TRIzol r ea gent ( Invitr ogen ) , following the manufactur er's instructions .T he RNA samples underwent treatment with DNAase RQ1 ( Pr omega Cor p., USA ) .cDN A synthesis w as performed using the Go ScriptTM Re v erse Tr anscription System Kit ( Promega Corp .) .Quantitative reverse transcription polymerase chain reaction ( R Tq-PCR ) w ere conducted using Go Taq ® qPCR Master Mix ( Pr omega Cor p . ) .Primers specific for CRISP3 and the housek ee ping gene gl ycer aldeh yde-3-phosphate deh ydrogenase ( GAPDH ) were synthesized by Thermo Fisher Scientific ( USA ) .The primer sequences ( from 5 to 3 ) were as follows: human CRISP3 forw ar d: TGCTCTGGAAA CCA CTGCAA, human CRISP3 reverse: C AGC AACC AGGAAC AAC AGC , and human GAPDH forw ar d: GA CTGTGGTCATGA GTCCTCCC, human GAPDH r e v erse: CAAGAT-C ATC AGC AATGCCTCC.T he R T-qPCR reactions w ere processed using an ABI Prism 7500 instrument ( Applied Biosystems, USA ) , and the delta-delta CT method was applied for quantification.

Statistics
For statistical analysis, SPSS ( Statistical Package for Social Sciences ) version 25.0 ( IBM Inc., Armonk, NY USA ) or Gr a phP ad v.7 ( California, USA ) was used.The c hi-squar e or Fisher's exact test was applied to compare categorical variables.A Gaussian distribution test was performed on all anal yzed gr oups .T he Mann-Whitney or t-test was used to e v aluate the difference between two groups, and ANOVA or Kruskal-Wallis for evaluation in more than two gr oups.Corr elation anal ysis w as carried out b y Spearman test with a 95% confidence interval ( CI ) .For survival analysis, the curv es wer e performed using the KM method with comparison using the log-rank test; in ad dition, Co x regression was performed, and the result was demonstrated as a hazard ratio ( HR ) with a 95% CI.A significance le v el of 5% was adopted.

Results
In the identification gr oup, fr om study GSE63514, a total of 54 675 pr obes wer e examined, corr esponding to individual genes, of which 13 908 sho w ed an adjusted P -value ( p.adj ) < 0.05 ( Fig. 1 A ) .Of these, probe 207 802_at stands out for having the lo w est p.adj value on a logarithmic scale and corresponding to the CRISP3 gene.In the same population, CRISP3 was isolated and gr a phically depicted, showing its significant reduction in cervical tumors ( Fig. 1 B ) .In addition, samples of cervical intraepithelial neoplasia ( CIN ) grades 1 to 3 were included.An apparently progressive decrease in CRISP3 expression appears to occur according to the se v erity of the lesion, although in comparison with normal tissue it is significant only for CIN3 and tumors ( Fig. 1 C ) .
With the objective to investigate the functions related to the deregulation of CRISP3 in the context of uterine colon cancer, we performed a pr otein-pr otein inter action anal ysis using the String database and Cytoscape for network visualization.Among the DEGs co-expressed with CRISP3 are classic markers such as the estrogen receptor ( ESR1 ) and the AR, as well as interleukins and interleukin receptors ( Fig. 2 A ) .Next, we highlight the closest inter actions with CRISP3: wher e CRISP2 is observ ed, whic h shar es affinities as a member of the same protein family; transcobalamin 1 ( TCN1 ) , a member of a family of vitamin B12-binding proteins associated with imm une infiltr ation [ 25 ]; and a BPI plug that contains member 1 of family B ( BPIFB1 ) , known for its involvement in innate immune response to bacterial exposure ( Fig. 2 B ) [ 26 ].In addition to this, we performed a pathway enrichment analysis, fr om whic h we could observ e a significant r ole of these genes in 'epidermal de v elopment', 'TGF-β signaling pathway', 'cell cycle r egulation', 'extr acellular matrix or ganization', 'r egulation of hormonal le v els', 'olefinic compound metabolic pr ocess', 'r egulation of DNA-binding transcription factor activity' and 'inflammatory response' ( Fig. 2 C ) .
After identifying and understanding CRISP3 more profoundly, we aimed to look for possible associations with the clinicopathological c har acteristics of patients with uterine CC.To do this, we accessed the TCGA project's RNA-seq dataset through the cbioPortal.The median expression of CRISP3 le v els of all samples was calculated.P atients wer e then gr ouped into high ( abov e the median ) or low ( below the median ) .We observed significant associations between CRISP3 and patient age ( P = 0.0098 ) , in terms of histological subtype ( P = 0.0022 ) , and in terms of HPV type ( P = 0.0001 ) .No associations were observed between CRISP3 and history of contrace pti ve use, menopausal status, tumor staging, or histological grade ( Table 1 ) .
Still in relation to this same population from the TCGA study, we e v aluated possible differ ences in CRISP3 tr anscriptional le v els depending on clinical pathological data.CRISP3 le v els wer e significantly lo w er in patients with squamous cell carcinoma ( SCC ) , compared with adenocarcinoma ( P < 0.0001; Fig. 3 A ) and other histological subtypes ( P = 0.0046; Fig. 3 A ) .Due to the c har acteristics of the E6 and E7 oncoproteins produced by different types of HPV, we also analyzed whether there could be differences in CRISP3 transcriptional levels.We observed that tumor samples positive for HPV18 have a significantly higher mean expression of CRISP3 than other viral types ( P < 0.05; Fig. 3 B ) .As already demonstrated in the association analysis in Table 1 , we also investigated possible differences depending on age and observed that patients > 50 years of age present a reduction in CRISP3 le v els.Furthermor e, we observ ed a positiv e corr elation between CRISP3 methylation and its transcriptional levels ( P < 0.0001; Fig. 3 D ) .
Finall y, we anal yzed the tr anscriptional pattern of CRISP3 in cell lines derived from CC.We observed that SiHa has a reduced expression of CRISP3, while SW756 exhibits a fold-change of 3.1.Both C33A and HeLa show a CRISP3 expression 8.3 and 16.5 times higher than SiHa, r espectiv el y, while for CaSki, this increase can be up to 112.6 times higher ( Fig. 6 A ) .Treatment with TSA effectively increased the transcriptional levels of CRISP3 in all analyzed cell lines ( Fig. 6 B-F ) .Similarl y, tr eatment with 5-AZA also led to an increase in CRISP3 expression, although a transcriptional reduction was observed in CaSki ( Fig. 6 F ) .The combined treatment of TSA and 5-AZA resulted in enhanced CRISP3 expression across all cell lines, with an ad diti ve effect noted in SW756 cells ( Fig. 6 C ) and HeLa cells ( Fig. 6 E ) .

Discussion
The quest for novel cancer biomarkers is a matter of global importance, with substantial pr ogr ess arising fr om bioinformatics.While CC can be effectiv el y pr e v ented thr ough scr eening and HPV vaccination, the continued incidence of new cases is influenced by v ariable v accination cov er a ge, the impact of anti-vaccine movements, and regional disparities in the adoption of pr e v entiv e pr actices [ 27 , 28 ].It is critical to acknowledge that while vaccination targets the viral types commonly linked to CC, it does not eradicate the possibility of infection with other high-risk HPV types.Mor eov er, social behavior might facilitate the spread of less fre- quent types not cov er ed by the vaccine.In this regard, understanding the patterns and mechanisms by which specific genes drive cancer development is crucial for elucidating oncogenic processes and promoting the development of new targeted therapies.Additionall y, ther e is a significant number of women around the world living with CC who cannot benefit from vaccination, requiring special attention from the scientific community.In this sense we conducted this work with the aim of identifying new biomarkers that may serve as prognostic support in management and ther a peutics of the disease.
Through a computational approach, we identified the differential expression of CRISP3, characterized by significantly reduced le v els in cervical tumors in comparison with normal tissue.Additionally, we conducted an analysis using the GENT2 platform and observed that, in other populations, the expression of CRISP3 is also significantly reduced in tumor samples ( supplementary Fig. S1 ) .T hese findings ma y be slightly supported by the study conducted by Li et al .[ 16 ].In this study, Li et al. selected DEGs for an IHC scr eening, whic h included CRISP2, CRISP3, Ki67, CDKN2A, KRT17, SYCP2, NEFH, DSG1, and PTG in samples of normal tissue, low-grade lesions ( LSIL ) , high-grade lesions ( HSIL ) , and SCC of the uterine cervix [ 16 ].During this screening, CRISP3 was not selected based on the authors' criteria and, ther efor e, was not further explored.Ho w ever, the absence of CRISP3 staining in LSIL, HSIL, and SCC, in contrast to its positivity in normal tissue samples , dra ws our attention.Our study differs from Li et al .'s work and other reports by using a compr ehensiv e bioinformatics a ppr oac h to anal yze gene expr ession data fr om a lar ge TCGA dataset.While other studies may not hav e r eported CRISP3 expr ession due to methodological differences and sample types, our detailed analysis reveals specific patterns of CRISP3 expression across different histological subtypes and in relation to HPV status.Additionally, we provide a transcriptional profile of CRISP3 in different cervical tumor cell lines and its alterations in response to treatment with drugs that regulate epigenetic factors.
In our analysis, we observed that intraepithelial lesions presented a transcriptional reduction of CRISP3 in an apparent progr essiv e decline from normal tissue to high-grade lesions ( CIN3 ) and CC.Confirmation of this hypothesis in the analyzed patient group was challenging, partly due to the small sample size.Nevertheless, the medians of CIN1 and CIN2 are relatively lo w er than those of normal tissue, while the disparity between normal tis-    S1 , see online supplementary material ) identified in our study.After statistical analysis, we observed a high positive correlation with CRISP3 ( r = 0.86; 95% CI: 0.80-0.90;P < 0.0001; supplementary Fig. S3, see online supplementary material ) .Based on this and the proximity of these genes as members of the same famil y, we anal yzed the CRISP2 pr ofile, and a similar pattern to CRISP3 was observ ed, likel y indicating a pr ogr essiv e decrease ( supplementary Fig. S3B ) .Supporting these findings, in Li et al .'s work, the staining for CRISP2 was significantly lower in HSIL and SCC, with high medians in normal tissue and LSIL [ 16 ].
During gestation, a r ecurring observ ation is the decrease in CRISP3 expr ession, an e v ent that corr elates with the ele v ation of human chorionic gonadotropin ( hCG ) hormone levels.Scientific liter atur e suggests that this reduction in CRISP3 expression may be associated with temporary and selective suppression of im-mune function at the implantation site, thereby facilitating blastocyst nesting and subsequent establishment of pregnancy [ 29 ].This phenomenon is of particular interest when considering the documented association between multiparity and increased susceptibility to cervical carcinoma [ 30 ].The conjecture arising from this observation is that decreased levels of CRISP3, induced by hCG ele v ation during gestation, could contribute to an increased risk of de v eloping cervical oncolog ical patholog ies .T he proposed explanation for this mechanism is that temporary attenuation of local immune response could favor an environment for progression of pre-existing infections, particularly those of viral etiology, such as HPV infection, already known for its etiological relationship with CC.
Conv ersel y, in silico r esearc h r e v ealed an incr ease of CRISP3 expression in patients diagnosed with multiple myeloma ( MM ) .In this study, r esearc hers pr oceeded with anal ysis using total RNA extr acted fr om peripher al blood samples, whic h included 18 healthy controls and 12 MM patient samples .T he results obtained thr ough RT-qPCR significantl y confirmed the CRISP3 ov er expr ession in MM [ 31 ].T hus , differ ential expr ession of CRISP3 seems to emerge as a potential biomarker not only in established tumors but also in other human malignancies, reinforcing the idea of studying CRISP3 mor e compr ehensiv el y.Similarl y, we not only identified CRISP3 as a differ entiall y expr essed gene but also as a potential prognostic biomarker.Initially, we assessed its prognostic role independent of histological subtype.Ho w ever, only after stratifying the study population did we identified a significant association betw een lo w le v els of CRISP3 and worst OS in patients with SCC.
Additionally, we conducted an analysis to identify miRNAs with the potential to regulate CRISP3.miR-1229-3p was the only one r etrie v ed fr om the miR TarBase public database and w as significant associated with worse OS and RFS.In addition, we identified the miRNAs miR-508-5p and miR-3614-5p.According to the literatur e, high le v els of miR-508-5p ar e observ ed in patients with lung cancer.In the same study, the authors demonstrate that high levels of CRISP3 and a long non-coding RN A ( lncRN A ) , LINC01342, suggesting that the increase in CRISP3 in these tumors is due to the competitive effect promoted by the lncRNA [ 11 ].In the context of CC, we hypothesize that highly complex epigenetic regulation may occur, including post-tr anscriptional r egulation mediated by miRNAs.Additionally, factors such as DNA methylation and histone acetylation seem to contribute to the downregulation of CRISP3, which led to the motivation for subsequent analyses.
In our in vitro analysis, we observed that HPV16-positive cell lines ( SiHa and SW756 ) exhibit a lo w er fold-change value, while the HPV-negative C33A and HPV18-positive HeLa cells show foldc hange v alues higher than SiHa by 8.3 and 16.5 times, r espectiv el y.Inter estingl y, in our in silico analyses, we found lo w er transcriptional le v els of CRISP3 in patients positiv e for HPV16.Furthermor e, in the in silico anal ysis, we observ ed significantl y decr eased le v els in patients with SCC, and the SiHa cell line, deriv ed fr om squamous car cinoma, also sho w ed a r educed tr anscriptional pr ofile.Ho w e v er, CaSki, also known to be HPV16 positive, exhibited the highest le v els of CRISP3.For this cell line, a cautious e v aluation is w arranted, considering tw o substantial c har acteristics that dr astically alter its behavior: the multiple copies of integrated HPV16 in its genome ( > 600 copies ) and its deriv ation fr om a metastatic SCC.Regar ding treatment, TSA w as able to increase CRISP3 transcription in all analyzed tumor cell lines, suggesting that epigenetic regulation based on histone activity may be present.Inter estingl y, in P athak et al .'s study [ 9 ], the CRISP3 promoter activity was measured using luciferase activity.In PC3 and RWPE-1 cells , both CRISP3-negative , it was demonstrated that the CRISP3 promoter is silenced by histone deacetylation [ 9 ].Treatment with 5-AZA also led to a tr anscriptional incr ease in CRISP3 in the analyzed cell lines, except in CaSki.In our in silico analysis, we found a positiv e corr elation betw een methylation and CRISP3 mRN A, whic h contr adicts the idea of a simple pr omoter r egulation.Howe v er, it is worth noting an analysis by Pongor et al .[ 32 ] that examined global DNA methylation in human small cell lung cancer.
Although not focusing on CRISP3, one of their analyses shows a pattern of both hyper-and h ypo-meth ylation in both the promoter and the gene body, despite the reduced levels in these cell lines [ 32 ].Finally, in cell lines such as SW756 and HeLa, an e v en mor e sophisticated regulation seems to occur, as the combined treatment of TSA and 5-AZA resulted in higher transcriptional activity of CRISP3.Although we observed that HPV18-positive samples sho w ed significantly higher CRISP3 expression, the comparable expression of CRISP3 in HPV-negative cell lines suggests that other molecular factors may be involv ed.Futur e studies should investigate the specific regulatory mechanisms of CRISP3 in relation to different HPV types to clarify this complex relationship.Finally, the regulation of CRISP3 expression may be influenced by various signaling pathways and transcription factors.Pr e vious studies suggest that CRISP3 expression can be modulated by epigenetic factors and specific protein interactions [ 9 ].In our study, we observ ed that tr eatment with epigenetic r egulators can significantl y influence the transcriptional levels of CRISP3.This suggests that the regulation of CRISP3 is tightly controlled by robust epigenetic mechanisms, as well as potential effects of miRNAs .T his insight adds a new dimension to our understanding of CRISP3's role in CC, although additional studies are needed to fully understand the extensive molecular mechanisms regulating this gene.
We acknowledge that our study has limitations, including the lack of additional experimental validation in biological samples and the reliance on gene expression data from public databases.Additionall y, the r elationship between CRISP3 and different HPV types is not fully elucidated, requiring further studies to confirm our findings.

Conclusion
In the course of our r esearc h, we pinpointed CRISP3 as the predominant gene exhibiting differential expression between normal and neoplastic tissues.Notably, we documented a decline in CRISP3 expression in patients diagnosed with SCC and concurrent HPV16 infection, a phenomenon that was associated with a diminished OS r ate.Additionall y, our explor ation into the modulatory effects of both isolated and synergistic applications of TSA and 5-AZA on CRISP3 expression yielded encouraging outcomes.Despite these advances, further research is imperative to elucidate the intricate r egulatory mec hanisms gov erning CRISP3 and to ascertain its potential role in the ongoing surveillance and management of individuals affected by CC.

Ac kno wledgments
We sincer el y thank Pr of.Dr Luisa Villa for gr aciousl y pr oviding access to her labor atory facilities, whic h made the execution of this study possible.

Figure 1 .
Figure 1.CRISP3 identification and expression profile.( A ) Volcano plot containing the 54 675 probes analyzed in GSE63514 from the identification gr oup.Blue corr esponds to probes with fold-change value < 0 and p.adj value < 0.05 in tumor tissue.Red corresponds to probes with fold-change value > 0 and p.adj value < 0.05 in tumor tissue .T he region with the highest number of overlapping points, in black, corresponds to probes with a P -value without statistical significance .T he black arrow indicates the CRISP3 probe.( B ) CRISP3 expression profile in normal tissue and tumor tissue samples from the identification group.( C ) CRISP3 expression profile in normal tissue, grade 1, 2 and 3 of cervical intraepithelial neoplasia ( CIN ) , and tumor tissue .T he data were tested for normality and subjected to the Mann-Whitney and Kruskal-Wallis U Test with Dunn's multiple comparison test.

F igure 2 .
PPI netw ork and pathw ay enric hment anal ysis .( A ) T he PPI netw ork w as constructed using the STRING database, emplo ying the top 100 deregulated genes identified in the GSE63514 dataset.( B ) A cropped and enlarged version of ( A ) , focusing on genes closely associated with CRISP3.( C ) The enrichment analysis prominently highlights the main altered pathwa ys , pro viding an in-depth insight into the functional implications arising fr om pr otein inter actions.

F igure 3 .
CRISP3 RN A le v els in CC by ( A ) histological subtype categorized in SCC, Adeno, and other subtypes; ( B ) HPV type.Kruskal-Wallis test was applied with Dunn's multiple comparison test; and ( C ) age.Mann-Whitney test was applied.( D ) Correlation between CRISP3 RNA le v els and methylation was accessed using the Spearman test.The analyses are based on TCGA Firehose Legacy CC patients.SCC: squamous cervical car cinomas.Adeno: adenocar cinomas.

Figure 4 .
Figure 4. OS of patients with uterine CC with CRISP3 expression ( A ) regardless of histological subtype and ( B ) SCCs of the uterine cervix subtype.RFS of patients with uterine CC with CRISP3 expression ( C ) regardless of histological subtype and ( D ) only SCCs of the uterine cervix.Patients were categorized according to the best cutoff of CRISP3 mRNA expression.Data from the TCGA Firehose study.

Figure 5 .
Figure 5. Surviv al anal ysis of patients with SCCs of the uterine cervix str atified by has-miR-1229-3p expr ession.( A ) OS and ( B ) RFS. P atients wer e categorized according to the best cutoff of miRNA expression.Data from the TCGA Firehose study.

Table 1 .
Clinicopathological features based on CRISP3 expression in patients with CC from the TCGA-Firehose Legacy study.
The low and high classification was accessed by mRNA expression z-score median value.SCC: squamous cervical carcinomas.Adeno: adenocarcinomas.