Tissue sections were obtained from 100 formalin-fixed paraffin-embedded (FFPE) blocks, previously collected from surgically treated cervical lesions, and stored in the Norwich University Hospital material archives. The cervical lesions were obtained from high-risk HPV-positive women with abnormal cytology on cervical screening. In cases of high-grade (HG) cervical intraepithelial neoplasia (CIN)—at different stages (see below)—or invasive squamous cell carcinoma (SCC), tumor specimens were obtained by large loop excision of the transformation zone (LLETZ) or by hysterectomy. CIN is classified on a scale from one to three. CIN1 refers to abnormal cells affecting about one-third of the thickness of the epithelium, whereas CIN2 and 3 define abnormal cells present in one- to two-thirds or more than two-thirds of the epithelium, respectively ( Table 1 ).

All procedures were performed in accordance with the ethical standards of the institutional Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

All human tissues derived from biopsied cervical lesions were classified according to the 8^th Edition of the American Joint Committee on Cancer (AJCC-TNM) Staging Manual (TNM8) (Brierley et al., 2017).

Hematoxylin and eosin (H&E)-stained slides were reviewed to confirm the diagnosis and to assess the cytological and histomorphological features of each specimen.

HPV detection was performed using the cobas^® HPV Test (Roche Molecular Systems). Positive results were subcategorized into HPV16, HPV18, and other hrHPV types. p16 was used as a surrogate for hrHPV infection ( Table 1 ).

Serial 5-μm sections from FFPE tissues were processed using the automated immunostainer Leica Bond III (Leica Biosystems). The primary antibodies employed are reported in Supplementary Table S1. Immunohistochemical expression of PAD4, PAD2, and anti-citrulline was evaluated by histo (H) score. For each histological section, the staining intensity was scored as 0 (negative), 1 (weak), 2 (moderate), and 3 (strong). The H-score was calculated by multiplying the intensity score (0-3) and the percentage of positive cells (0-100%), with a maximum of 300.

CaSki cells (ATCC CRL-1550™) were cultured in RPMI-1640 medium and HeLa cells (ATCC CCL-2™) in Dulbecco’s Modified Eagle’s medium, both supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM glutamine, 1 mM sodium pyruvate, 100 U/mL penicillin, and 100 µg/mL streptomycin sulfate (Sigma-Aldrich). HPV-negative normal oral keratinocytes (NOKs) were kindly provided by Frank Rösl (Germany) and were cultured as previously described (Yang et al., 2019).

CaSki and HeLa cells were transiently transfected with small interfering RNAs (siRNAs) using a Neon™ Transfection System (Life Technologies) according to the manufacturer’s instructions (1005 V, 35 ms pulse width, two impulses). The following siRNAs were used: control siRNA (siCTRL; 1027292) was purchased from Qiagen; siRNAs against HPV16 E6/E7#1,HPV16 E6/E7#2, HPV18 E6/E7#1, and HPV18 E6/E7#2 were synthesized by Eurofins Genomics. The indicated siRNAs were previously characterized (Lo Cigno et al., 2020b), and the specific sequences are reported in Supplementary Table S2 .

BB-Cl-amidine (BB-Cl-A; HY-111347A) (Ledet et al., 2018) was purchased by MedChemExpress and dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich) at stock concentrations of 25 mM.

CaSki viability upon exposure to increasing concentrations of BB-Cl-A or DMSO was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT) method (Toscani et al., 2021).

Total RNA was extracted using TRI Reagent (Sigma-Aldrich), and 1 μg was retrotranscribed using the Revert-Aid H-Minus First Strand cDNA Synthesis Kit (Thermo Fisher Scientific), according to the manufacturer’s instructions. Comparison of mRNA expression between samples was performed by SYBR green-based RT-qPCR using Mx3000P apparatus (Stratagene). The housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used to normalize for variation in cDNA levels. The primer sequences used are reported in Supplementary Table S2 .

Protein extracts were prepared in RIPA buffer and subjected to immunoblotting. The primary antibodies used are reported in Supplementary Table S1. Scanning densitometry was performed using Image Lab (version 6.0.1; Bio-Rad).

Equal amounts of protein were diluted with 80% trichloroacetic acid and incubated with Rh–PG (final concentration 0.1 mM; Cayman Chemical) for 30 min. The reaction was quenched with 100 mM L-citrulline (Sigma-Aldrich), centrifuged at 21,100×g for 10 min, washed with ice-cold acetone, resuspended in PBS supplemented with L-arginine and analyzed through gel electrophoresis. Gels were imaged (excitation = 532 nm, emission = 580 nm) using a ChemiDoc MP Imaging System (Bio-Rad Laboratories), stained with brilliant blue G-colloidal solution (Sigma-Aldrich).

CaSki cells were seeded in 6-well plates and treated with BB-Cl-A (3 μM) for 24 or 48 h. After treatment, cells were pelleted down and fixed with 70% methanol for 30 min at 4°C. After washing with PBS twice, cells were incubated with a DNA staining solution consisting of propidium iodide (PI; Sigma-Aldrich) and RNase (Merck Millipore) for 15 min at 37°C in the dark. The proportion of cells in each phase of the cell cycle was determined by DNA content stained with PI using a BD FACSCanto II flow cytometer (BD Biosciences). Data obtained were analyzed with FlowJo software (BD Biosciences).

To distinguish apoptotic from necrotic cells, double staining was performed for exposed phosphatidylserine and PI exclusion using the annexin V-FITC Apoptosis Detection Kit (Calbiochem). Experiments were performed according to the manufacturer’s instructions. Briefly, CaSki cells were seeded in 6-well plates and treated with BB-Cl-A (3 μM) for 24 or 48 h. After treatment, cells were washed with PBS twice, trypsinized, and then resuspended in a binding buffer (10 mM HEPES/NaOH, pH 7.4, 140 mM NaCl, 2.5 mM CaCl₂). Annexin V-FITC was added to a final concentration of 100 ng/mL, and the cells were incubated in the dark for 10 min, then washed again in PBS, and resuspended in 300 μL of the binding buffer. In total, 40 μg/mL of PI was added to each sample before the flow cytometric analyses. Cells were analyzed using a BD FACSCanto II flow cytometer (BD Biosciences). Data obtained were analyzed with FlowJo software (BD Biosciences). Unstained cells and cells only stained with annexin V-FITC or PI were used as controls to establish compensation and quadrants. Cells were gated according to their light-scatter properties to exclude cell debris.

All statistical tests were performed using GraphPad Prism version 7.04 for Windows (GraphPad Software). Data are presented as means ± SEM or medians ± interquartile. Means were compared using an unpaired t-test, meanwhile, medians were compared using a one-way analysis of variance (ANOVA) with Bonferroni’s post-tests. The two-tailed Pearson correlation was employed to assess th e correlation between PAD4 and citrulline expression in immunohistochemistry. Differences were considered statistically significant at P < 0.05 (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

To evaluate the role of citrullination in the context of hrHPV-related lesions, we assessed the citrullination profiles of FFPE biopsies from samples negative for intraepithelial lesion or malignancy (NILM) vs. cervical intraepithelial neoplasia (CIN) at different stages — from CIN (CIN1 > CIN2 > CIN3, see Materials and Methods) to invasive SCC. Information about patients’ age, menopausal status (yes/no), and follow-up is presented in Table 1 . The mean age of the cohort was 47 years, ranging from 22 to 63 years, with 12.2% of women in a menopausal state. HrHPV and p16 positivity are also reported and agree with previous findings (Kombe Kombe et al., 2021; Shafique et al., 2023). A recurrence of carcinoma has been reported after 2 years in 1 out of 14 SCC cases with available follow-up data.

As shown in Figure 1A and Supplementary Table S3 , disease progression was paralleled by a significant increase in total protein citrullination levels in cells with aberrant proliferative capacity, and this effect is more pronounced in CIN2 and CIN3 lesions, while in the SCC group, there was significant variability in citrulline staining. Specifically, keratinocytes with high and abnormal levels of citrullinated proteins in the cytoplasm were localized throughout the mucosal layers, whereas in the normal epithelium, they were predominantly found in the basal and parabasal layers ( Supplementary Table S3 ).

Since PADs are the enzymes catalyzing citrullination, we next sought to determine their expression according to tumor stage. To rule out bias due to allelic loss in tumors, we first examined PAD expression in a squamous cervical carcinoma data set curated by TCGA (Gao et al., 2013; Ciriello et al., 2015). We found the PAD genomes rarely mutated—less than 1.2% out of 278 tumors, considering all PAD isoforms ( Figure 1B )—indicating that loss of PAD expression is a rare event.

Next, we assessed cervical cancer vs. NILM for protein expression levels of PADs 2 and 4, the two PAD isozymes most broadly expressed in human cancer (Yuzhalin, 2019; Sarnik and Makowska, 2022). In NILM, PAD4 expression was limited to the nuclei in basal and parabasal layers of the squamous epithelium and in sparse stromal and inflammatory cells within the stromal compartment ( Figure 1C ). On the other hand, PAD4 staining of both the atypical squamous epithelium and the stromal compartment significantly increased as the cancer progressed to more advanced stages ( Figure 1C ). Interestingly, this increase was significantly correlated with enhanced citrullinated protein expression levels in CIN2 and CIN3 lesions but not CIN1 and SCC ( Supplementary Table S3 ). No signal was observed in negative controls that were incubated with either the primary or secondary antibody alone (data not shown). Of note, koilocytes in low-grade CIN exhibited a very weak PAD4 immunoreactivity in comparison with high-grade lesions, which suggests differential expression of this marker according to the integration status of HPV in squamous cells. Moreover, mature squamous cells in the upper epithelial layers were PAD4 negative, which implies that PAD4 expression can be detected only in immature squamous phenotypes.

Collectively, these results argue in favor of PAD4-mediated citrullination as a critical event in cervical tumorigenesis. In contrast, PAD2, whose expression is restricted to the glandular epithelium and absent in the squamous epithelium in both normal mucosa and hrHPV-related lesions ( Supplementary Figure S1 ), does not seem to be involved in disease progression.

To investigate whether HPV trigger PAD-mediated citrullination to promote viral fitness, we first evaluate the protein citrullination profile in CaSki cells, which contain about 600 copies of HPV16 genomes per cell (Mincheva et al., 1987). Protein lysates obtained from either siRNA E6/E7- or siRNA CTRL-transfected CaSki were exposed to the citrulline-specific probe Rh-PG (Bicker et al., 2012). Even though the expression of the viral proteins was almost completely suppressed upon gene silencing ( Figure 2A ), as demonstrated by immunoblotting with antibodies raised against HPV E6 and E7, we did not notice any significant differences between the citrullination profiles of cells lacking E6 and E7 (siE6/E7) and their controls (siCTRL) ( Figure 2B , left panel), neither the overall amount of proteins was modified ( Figure 2B , middle panel). The same results were obtained when membranes were probed with an anti-cyclic citrullinated peptide (CCP) antibody ( Figure 2B , right panel). We detected the same citrullination pattern in HeLa cells transfected with siRNA E6/E7 or siRNA CTRL ( Supplementary Figures S2A, B ), confirming that the overall citrullination profile is not markedly affected by the absence of hrHPV E6/E7. Furthermore, RT-qPCR analysis of RNAs from the same cells revealed that PADI genes were all expressed in CaSki and HeLa, but not profoundly modulated by E6 and E7 compared to the controls ( Figure 2C , Supplementary Figure S2C ). When we analyzed PAD protein expression, we only observed a slight downregulation of PAD2 in CaSki and HeLa cells lacking E6 and E7 (24% and 22%, respectively), while PAD3 and PAD4 were not significantly modulated. Other PAD isoforms (PAD1 and 6) were expressed at undetectable levels in CaSki and HeLa cells and did not vary upon E6/7 gene silencing ( Figure 2D , Supplementary Figure S2D ).

To conclusively elucidate the impact of PAD-mediated citrullination on hrHPV pathogenesis, we took advantage of the cell-permeable pan-PAD inhibitor BB-Cl-amidine (BB-Cl-A). However, before assessing BB-Cl-A activity on CaSki cells, we performed a standard MTT viability assay to rule out the possibility that the drug may have cytotoxic effects. Indeed, our screening indicated that the cytotoxicity of BB-Cl-A was low or undetectable at concentrations of up to 3 µM, as ~90% of the cells were viable after 48 h of treatment ( Figure 3A ).

We then evaluated E6 and E7 protein expression levels in CaSki cells treated or not with BB-Cl-A. As shown in Figure 3B , we observed a dramatic downregulation of both E6 and E7 expression in BB-Cl-A-treated cells, with a more pronounced effect after 24 h. Conversely, an expected increase in p53 protein expression, a known target of E6 (Mittal and Banks, 2017), was detected at 24 and 48 h post-BB-Cl-A treatment (15 and 2.4 folds at 24 and 48 h vs. vehicle control, respectively). Fittingly, at the same time points, we also recorded a 9.2 (24 h) and 4.5 (48 h) fold increase of the cyclin-dependent kinase inhibitor p21 (Harris and Levine, 2005), a p53, and E7 target gene. Overall, these results demonstrate that the inhibitory activity of the BB-Cl-A compound against hrHPV E6 and E7 is associated with the upregulation of p53 and p21 expression. This effect is specific, as p53 and p21 were not modulated by BB-Cl-A in HPV-negative NOKs ( Supplementary Figure S3 ).

Finally, since p53 target genes, such as p21, are involved in the regulation of cell cycle progression (Xiong et al., 1993), we sought to determine the effect of BB-Cl-A treatment on the cell cycle. Flow cytometry analyses showed that upon 48 h of treatment, BB-Cl-A significantly increased the population of CaSki cells in the sub-G₀ phase, while decreasing in the G₀-G₁ phase, compared to control, indicating that PAD inhibition leads to cell cycle arrest ( Figure 3C ). To further strengthen this observation and given that apoptosis is one of the major mechanisms by which hosts evade viral infections, including HPV (Shimada et al., 2020; Gusho and Laimins, 2022), we assessed the ability of BB-Cl-A to drive cell-death pathways in CaSki cells by dual staining with annexin V and PI. As shown in Figure 3D , at 48 hpt (hours post-treatment), a significant increase in apoptosis (annexin V+/PI−) was detected in cells treated with BB-Cl-A compared to vehicle-treated cells (33.9% vs. 12.73%, respectively), while necrosis (annexin V+/PI+) barely occurred in every condition tested.

Altogether, our findings indicate that the inhibition of protein citrullination by BB-Cl-A halts cell cycle progression at the sub-G₀ phase and drives CaSki cells to apoptosis, suggesting that BB Cl-A is a promising pharmacological agent to treat HPV-infected cells.