The local Ethics Committee of the Saarland (Ärztekammer des Saarlandes, Saarbrücken, Germany) in accordance with the Declaration of Helsinki approved the experiments with human material used in this study, and a written consent form was provided by the study participants. The TMA cohort consists of patients derived from Russia and Germany between 1992 and 2015. Data on clinical outcome and HPV status were published previously (15, 16). Briefly, DNA was isolated from formalin-fixed paraffin-embedded (FFPE) tissue sections by the QIAamp DNA FFPE Tissue Kit (Qiagen, Hilden, Germany) following the manufacturer’s protocol, and the HPV PCR was conducted using the GP5+/6+ primers as described previously (15, 16). HPV status was further determined by p16^INK4A immunohistochemistry (IHC), a surrogate marker for active HPV oncoprotein–driven transformation, using a published protocol (15, 16). HPV status was considered as positive in case of positive PCR and IHC with subsequent genotyping using INNO-LiPA genotyping extra II (Fujirebio Germany GmbH, Hannover, Germany), revealing HPV16, HPV18, and HPV59 in 27 (93.1%), one (3.45%), and one (3.45%) of the 29 cases, respectively. Sections of all cases were reviewed by two experienced uropathologists, and histological subtypes and tumor grade were defined according to the 2016 WHO classification and the eighth editions of TNM classification of malignant tumors. Specimens were stratified as non-invasive (pTis and pT1a) and invasive (pT1b, pT2, pT3, and pT4), not metastasizing (cN0 and pN0) and metastasizing (pN1, pN2, and pN3). Basic cohort description and patient/disease characteristics are listed in Supplementary Table 1 . Of the 74 patients in total, respective tissue was punched for the individual TMA reflecting tumor center (TMA TC), invasion front (IF), lymph node metastases (TMA LM), and adjacent normal tissue (TMA NO) as duplicates (TMA TC, IF, and NO) and triplicates (TMA LM) with the total number of evaluable specimens indicated for each staining.

Three HPV-positive PeCa cell lines were generated previously from a primary carcinoma and lymph node metastases, including the particularly rare case of one primarius-derived (named P2) and one metastasis-derived (named L2) cell line originating from the same patient and one further metastasis-derived cell line of an additional patient (named L3). The cell lines were authenticated by Multiplexion in 2018 using the originating biopsies obtained from patients that underwent penectomy and metachronous radical inguinal lymph node dissection for metastatic squamous cell carcinoma of the penis at the University Hospital Schleswig-Holstein, thus representing a validated cell culture system (37). The HPV status of the cell lines was investigated recently (16). Cells were cultivated in PeCa medium {1:1 mixture of keratinocyte growth medium 2 (KGM2) containing all supplements [C-20011, bovine pituitary extract at 0.004 ml/ml, epidermal growth factor (EGF) at 0.125 ng/ml, insulin at 5 µg/ml, hydrocortisone at 0.33 µg/ml, epinephrine at 0.39 µg/ml, transferrin at 10 µg/ml, calcium chloride at 0.06 M; PromoCell, Heidelberg, Germany] and RPMI 1640 containing 10% heat-inactivated fetal bovine serum (FCS), 1% sodium pyruvate, and 1% penicillin and streptomycin (R10+/+, Merck, Schnelldorf, Germany)}. Normal foreskin keratinocytes (NFKs) and human foreskin fibroblasts (HFFs) were isolated from foreskin tissue (Saarland University Medical Center), tested negative for HPV using PCR, and expanded in KGM2 (C-20011, PromoCell) and Dulbecco's Modified Eagle's Medium (DMEM) with 10% heat-inactivated FCS, 1% sodium pyruvate, and 1% penicillin and streptomycin (D10+/+), respectively. NFKs were cultured in PeCa medium for experiments. We conducted mycoplasma-specific PCRs on a regularly basis of once per month. The human vulva carcinoma cell line A431 (DSMZ, Braunschweig, Germany, ACC-91, obtained 2018, RRID : CVCL_0037) was cultured in R10+/+. Cell lines were used below passage 20, NFKs up to passage 4, and HFFs up to passage 7. Organotypic three-dimensional (3D) cultures were generated using HFF (5 × 10⁵ cells, passages 3–5) embedded in 1 ml of rat collagen [as described previously (16)] in 24-well plate in D10+/+. The day after, medium was exchanged to PeCa medium for 1 h before 7 × 10⁵ PeCa cells were seeded on top of the collagen-fibroblast matrix. The next day, the cultures were transferred onto a metal grid in six-well plates to allow multilayered growth at the air-liquid interface. Fourteen days later, supernatants were harvested, and organotypic 3D cultures were fixed in 4% paraformaldehyde (Merck) and embedded in paraffin.

The engineered 225-IgA2m(1)-N166G-P221R-C331S-N337T-I338L-T339S-dC471-dY472 further named 225-IgA2.0 was produced using the variable regions of the m225 antibody and the engineered IgA2m(1) constant region as previously described (38). ChromPure human Serum IgA purchased from the Jackson ImmunoResearch Labs (#009-000-011, RRID : AB_2337047) served as IgA isotype control.

Target cells (2 × 10⁴) per well were labeled for 30 min at 37°C with 10 μM Calcein-AM (Fisher Scientific, C3099) in suspension. Effector cells were isolated as previously described (39). After washing three times, isolated effector cells were added with an effector-to-target cell ratio (E:T ratio) of 40:1. Antibodies were added to the microtiter plates in triplicates as indicated. For maximum release, labeled target cells were treated with 1% Triton X-100 (Sigma-Aldrich) or left untreated for basal release. After 3 h of incubation at 37°C, plates were centrifuged; supernatants were transferred into black 96-well plates with clear flat bottom (Sigma-Aldrich), and fluorescence was measured in the Victor II plate reader (PerkinElmer, Waltham, MA, USA). Percentage of cellular cytotoxicity was calculated using the equation: % specific lysis = (experimental release − basal release)/(maximal release − basal release). Antibody-independent cytotoxicity (effectors without target antibodies) or effector-independent cytotoxicity (target antibodies without effectors) was not observed.

For CD147 surface expression analysis, cells were seeded and grown to 80% confluence, trypsinized, and stained with mouse anti-human CD147-specific antibody (50 µg/ml; #306202, RRID : AB_314586, BioLegend, Amsterdam, Netherlands) per 150,000 cells for 30 min on ice. After washing, bound antibody was stained with mouse-immunoglobulin G kappa chain (IgGκ) phycoerythrin (PE)-conjugated binding protein (sc-516141, Santa Cruz, Dallas, TX, USA) and measured with a FACS Canto II (BD Biosciences). Relative fluorescence intensity (RFI) was calculated using the following equation: mean fluorescence intensity specific Ab/mean fluorescence intensity control Ab. Soluble CD147 was quantified in 3D culture supernatants using Human CD147/EMMPRIN ELISA (#ELH-CD147-1, RayBiotech, Peachtree Corners, USA) according to the manufacturer’s instructions.

formalin-fixed paraffin-embedded (FFPE) tissue (TMA) slides were stained by IHC. Antigen retrieval was performed by heating the sections in 1 mM citrate buffer (pH 6.0) at 95˚C for 10 min, and endogenous peroxidase activity was blocked with 3% H₂O₂/tris buffered saline (TBS) for 10 min [for 3,3`-Diaminobenzidine (DAB) staining]. formalin-fixed paraffin-embedded (FFPE) sections were incubated with the S100A8-specific (#NBP1-42076, rabbit anti-human S100A8, RRID : AB_2184111, Novus Biologicals, Cambridge, UK), S100A9-specific (#sc-20173, rabbit-anti-human S100A9, RRID : AB_2184420, Santa Cruz, Heidelberg, Germany), or CD147/Emmprin-specific (BioLegend, Cat# 306202, RRID : AB_314586) antibodies overnight followed by Alkaline phosphatase/horseradish peroxidase (AP/HRP-conjugated anti-rabbit/mouse antibody incubation and developed with 3,3`-Diaminobenzidine (DAB) or HRP substrates [ImmPRESS HRP/Peroxidase Reagent Kit and VECTOR Red Alkaline Phosphatase Substrate Kit (Alkaline Phosphatase Anti-Rabbit/Mouse IgG), both Vector, Burlingame, CA, USA]. After counterstaining with Hematoxylin Nuclear Counterstain (Gill’s Formula) (Vector), the slides were covered with Vectamount and documented. Staining was documented using a Leica DMI6000 (RRID : SCR_018713) with LAS X software (RRID : SCR_013673).

Graphical, correlation, and statistical analyses were performed using GraphPad Prism 9.02 (GraphPad Software, San Diego, CA, USA, RRID : SCR_002798). The annual percentage change (HRPC) was calculated using the following formula: HRPC = {Exp(slope) − 1} × 100. Group data are reported as mean ± SEM. Data of multiple experiments were illustrated as a box and whisker plot showing individual results with minimum and maximum values. Significance was determined by two-way or one-way ANOVA repeated measures test with Tukey’s correction or Fisher’s exact test as indicated, and odds ratios were calculated using the Baptista-Pike method. Significance was accepted when p-values were ≤ 0.05.

The expression of both Calprotectin subunits S100A8 and S100A9 was examined in clinical PeCa specimens using three TMAs reflecting the tumor center (TMA TC, n = 63), the invasion front (TMA IF, n = 57), and the corresponding lymph node metastases (TMA LM, n = 22). The HPV⁺ status was defined as positive for GP PCR and p16^INK4A IHC (15). Slides were stained by IHC and immune reactive scores (IRSs) were determined according to Remmele and Stegner ( Figures 1A, B ). Tumor nests and infiltrating immune cells were intensively stained, whereas, in some cases, the tumor margin was spared ( Supplementary Figures 1 , 2 ). Staining for both subunits was limited to suprabasal rather superficial layers in the squamous epithelium of non-malignant tissue ( Supplementary Figures 1 , 2 ). Specimens with a throughout S100A8⁺/S100A9⁺-positive epithelium or PeCa tissue were identified for each TMA. The majority of PeCa specimens were positive for S100A8 and S100A9 regarding TMA TC (S100A8, 92.1%; S100A9, 91.5%), TMA IF (S100A8, 91.2%; S100A9, 65.5%), and especially of the TMA LM (S100A8, 100.0%; S100A9, 100.0%) ( Figures 1C, D ). Notably, the majority of S100A8⁺S100A9⁺ specimens were HPV⁻ for the TMA TC (S100A8, 63.5%; S100A9, 62.7%) and IF (S100A8, 61.4%; S100A9, 61.8%), which flips regarding TMA LM (S100A8, 70.0%; S100A9, 80.0%). There was a remarkable and significant increase of positive IRS regarding the tumor center compared to the invasion front for both HPV⁺ (p = 0.0093) and HPV⁻ (p = 0.0013) PeCa for the Calprotectin subunit S100A9 but not for S100A8 ( Figures 1E, F ). The expression of both Calprotectin subunits suggests a TME actively recruiting myeloid cells that once trapped polarizes them into MDSCs (1–3). Of those, especially neutrophil-MDSCs have been repeatedly reported as a negative prognostic marker for patients with dismal prognoses (2, 7, 40–43).

We previously identified a CXCL8-dependent mechanism of neutrophil chemotaxis with HPV⁺CD15⁺ characterizing PeCa with a higher staging regarding invasion and metastasis (16). Here, a remarkable amount of PeCa specimens with infiltrating cells intensively stained for S100A8 and S100A9 was observed ( Figure 2A ). While, in the adjacent normal foreskin, only occasionally S100A8⁺S100A9⁺ immune cells were observed, this dramatically changes in PeCa specimens. Specimens with partially high numbers of intensively stained infiltrating and tumor-intersecting immune cells were identified at the invasion front, at the tumor center, and lymph node metastasis ( Figure 2A ). Reconciling previously published data, a mechanism of elevated expression levels of HPV16 oncoproteins that cause a p63-dependent upregulation of CXCL8, a well-known chemokine for neutrophils, was described (16). This mechanism provided a possible explanation for the infiltration of neutrophils observed in PeCa specimens. For this purpose, the same TMAs as used here were stained for p63 and CD15, with the latter being a marker for neutrophils (16). As described previously (16), we detected a high number of CD15⁺ infiltrating and tumor-intersecting cells in PeCa specimens but not in adjacent normal foreskin ( Figure 2B ) with an overlap of S100A8⁺S100A9⁺ and CD15⁺ immune cells ( Supplementary Figure 3 ). Triple-positive (S100A8⁺S100A9⁺CD15⁺) PeCa specimens were with higher-frequency HPV⁻ for the TMA TC and IF, whereas the majority of LM were positive for all four parameters (S100A8⁺S100A9⁺CD15⁺HPV⁺) ( Figure 2C ). S100A8⁺S100A9⁺CD15⁺ PeCa specimens were in tendency rather invasive growing cancers (TMA TC odds ratio = 2.872, TMA IF odds ratio = 3.048) and significantly enriched in the HPV⁺ group of metastasizing PeCa (odds ratio = 8.444, p = 0.0232) ( Figures 2D, E ). Thus, in line with the previous reports (15, 16), our data provide further evidence that patients with PeCa presenting with HPV-related histological subtypes and tumors infiltrated with neutrophils are at a higher risk for the prognosis of more aggressively dedifferentiated growing cancer ( Supplementary Figure 4 ) and of lymph node involvement ( Supplementary Figure 5 ). Predominantly, three histological subtypes—usual SCC (1), warty-basaloid (11), and basaloid (13)—followed by papillary-basaloid (10) and clear cell carcinoma (14) were characterized by a high amount of S100A8⁺S100A9⁺CD15⁺ cases, particularly if HPV⁺ ( Supplementary Figure 6 ), and with lymph node involvement (pN1-3; Supplementary Figure 5 ). Because both S100 subunits mark MDSCs frequently associated with disease progression and treatment failure, these data suggest that the marker profile HPV⁺S100A8⁺S100A9⁺CD15⁺ characterizes patients with PeCa at a higher risk for a worse prognosis.

Because both S100 subunits were notably higher expressed in PeCa specimens, particularly at the tumor center, we questioned if the same might be true for the Calprotectin receptor CD147, especially because the previous data already indicated an increased expression in PeCa (44). Slides of the individual TMA reflecting the tumor center (TMA TC, n = 73), the invasion front (TMA IF, n = 69), and corresponding lymph node metastases (TMA LM, n = 22) and adjacent normal tissue (TMA NO, n = 24) were stained for CD147 expression using IHC, and IRSs were determined ( Figure 3A ). Specimens of the PeCa TMA (TC, IF, and LM) displayed a significantly more intense staining ( Figure 3B ) and a higher scoring than that of the TMA NO (TC vs. NO, p < 0.0001; IF vs. NO, p = 0.0043; LM vs. NO, p = 0.0414) ( Figure 3C ). IRSs of the PeCa specimens of the TMA TC were significantly higher than IRSs of normal tissue in the HPV⁺ and HPV⁻ groups (TC HPV⁺ vs. NO HPV⁺, p = 0.0138; TC HPC⁻ vs. NO HPV⁻, p < 0.0001) and of the TMA IF compared to NO in the HPV⁻ group (p = 0.0122) ( Figure 3D ). CD147⁺ PeCa specimens were significantly enriched in the HPV⁺ group at the TC compared to NO ( Figure 3E ) and IF ( Figure 3F ), as well as for the LM compared to the NO ( Figure 3G ) and IF ( Figure 3H ). All HPV⁺ LMs were positive for CD147 ( Figures 3E, F ). Specimens of the TMA TC, IF, and from the adjacent NO that were positive for CD147⁺ had with a higher-frequency metastases in case of an HPV⁺ than HPV⁻ status (N classification pN1-3) ( Figures 3I–K ). Notably, all HPV⁺ LM specimens were positive for S100A8/A9, CD15, and CD147. These results suggest that CD147⁺ marks HPV⁺ PeCa specimens with a higher risk for lymph node metastases. Finally, our data provide evidence that the Calprotectin-CD147-neutrophil axis plays an important role in disease progression particularly in the event of an HPV⁺ status.

Next, we evaluated the expression of CD147 by HPV-positive PeCa cell lines using organotypic 3D raft cultures. These composite of HFF embedded in extracellular matrix with keratinocytes or tumor cells seeded on top at the air-liquid interface that allows them to grow to a multilayered epithelium. This cell culture model creates an environment that mirrors inter-cellular regulatory networks, thus enabling studies of the expression of CD147 in a more physiological context. IHC staining for CD147 on formalin-fixed paraffin-embedded (FFPE) slides of these 3D cultures revealed a more intense staining of HPV-positive PeCa cell lines than that of NFK ( Figure 4A ). Because this approach does not differentiate between the soluble and membranous form of CD147, we tested the conditioned media from these 3D cultures for the release of soluble CD147 (sCD147). We recently identified sCD147 as upregulated in the same HPV-positive PeCa cell lines using a cytokine array and conditioned media from monolayer culture (16). In line with this, media conditioned by 3D cultures of PeCa cell lines contained notable higher amounts of sCD147 than those of NFK (P2 vs. NFK, p = 0.0210; L3 vs. NFK, p = 0.0005; Figure 4B ). Next, we tested for the expression of surface membranous CD147 (mCD147) expression using indirect immunofluorescence and flow cytometry ( Figures 4C, D ). All three HPV-positive PeCa cell lines expressed higher surface levels of mCD147 than the cell line A431, which was used as HPV-negative control tumor cell line, particularly P2 and L2 (P2 vs. A431, p = 0.0199; L2 vs. A431, p = 0.0002). Notably, the PeCa cell line L2 expressed higher amounts than the cell lines P2 and L3 (L3 vs. L2, p = 0.0019). We described previously that the PeCa cell lines display a different susceptibility to IgA-dependent neutrophil-mediated tumor cell killing (16) and thus questioned whether the amount of mCD147 would be predictive for this. We investigated the capacity of the EGF receptor (EGFR)–directed engineered IgA2 antibody 225-IgA2.0 to engage freshly isolated neutrophils for antibody-dependent cell-mediated cytotoxicity (ADCC) of the PeCa cell lines and A431 cells [ Figure 4E ; re-evaluated from (16)]. The IgA-dependent neutrophil-mediated killing was most effective using the A431 cells, followed by PeCa cell lines P2 (A431 vs. P2, p = 0.0223) and L3 (A431 vs. L3, p = 0.0065) as target cells, whereas the cell line L2 (A431 vs. L2, p = 0.0001) was not lysed. There was a significant difference to the isotype control in case of A431, P2, and L3 (p < 0.0001) but not for L2 ( Figure 4E ). The specific lysis rates for P2 and L3 were significantly higher than for the PeCa cell line L2 (P2 vs. L2, p = 0.0047; L3 vs. L2, p = 0.0156). Thus, mCD147 expression displayed a contrary pattern to the specific lysis rate and correlated negatively with the efficacy of the 225-IgA2.0 to engage neutrophils for tumor cell killing (Pearson, r = −0.9093, p = 0.0454), whereas the soluble variant did not (Pearson, r = −0.6387, p = 0.3613) ( Figure 4F ). Our results suggest that HPV-positive PeCa cell lines expressed higher amounts of CD147 than HPV-negative cells and that CD147 could display a critical factor for neutrophil-centered immunotherapeutic approaches.