Retrospective study on the correlation between CXCL13, immune infiltration, and tertiary lymphoid structures in cutaneous squamous cell carcinoma

TME deconvolution

A total of 120 RNA-sequencing data were downloaded from the Gene Expression Omnibus (GEO) with the accession code GSE199070. Then, 53 samples comprising 40 primary cSCC, one metastatic cSCC, and 12 normal sun-exposed skin (NS) were kept for cell composition deconvolution analysis. Abundance scores for 11 cell types (T cells, CD8⁺ T cells, cytotoxic T cells, B cells, NK cells, monocytes, macrophages, myeloid dendritic cells, neutrophils, endothelial cells, cancer-associated fibroblasts) in the TME were quantified using MCP-counter (Becht et al., 2016). Since the lack of cSCC datasets with both transcriptomic and clinical information, survival analysis for CXCL13 was performed in head and neck squamous cell carcinomas using the GEPIA website (Lu et al., 2024).

Patients and samples

This was a retrospective study, and written informed consent was obtained from patients. This study included 63 consecutive patients diagnosed with primary cSCC, confirmed through pathological examination, at Shanghai Skin Disease Hospital (Shanghai, China) between June 2017 and June 2019. All patients had not received chemotherapy, immunotherapy, or radiotherapy prior to surgery. The study was approved by the Ethics Committee of Shanghai Dermatology Hospital (2024-50). Comprehensive clinical and histopathological data were anonymized and systematically collected for each patient to ensure confidentiality.

Pathological examination

Tumor tissues from all patients were surgically excised, fixed in formalin, and embedded in paraffin before being sectioned. Histological evaluation was conducted by two experienced pathologists following the World Health Organization (WHO) grading criteria. Hematoxylin and eosin (HE) staining was employed to assess tumor infiltration depth, subcutaneous extension, lymphatic and vascular metastasis, as well as peripheral nerve invasion.

Immunohistochemical staining

All tissue samples were initially heated at 60 °C for 30 min to enhance tissue adherence to the sections. Subsequently, samples were deparaffinized in xylene, hydrated through a series of graded alcohol baths, and treated with 3% H2O2 for 15 min to quench endogenous peroxidase activity. Antigen retrieval was then achieved by incubation in EDTA (pH 9.0) at sub-boiling temperature, followed by gradual cooling to room temperature. Sections were further blocked with 5% BSA solution at room temperature for 30 min. Primary antibodies were applied and incubated overnight at 4 °C. The following day, sections were incubated with secondary antibodies for 15 min at 37 °C, followed by DAB staining for 3 to 8 min. Staining was halted by washing with ddH2O. Finally, counterstaining with hematoxylin, dehydration through graded alcohols, and mounting of the sections were performed.

Immunohistochemical evaluation

The CXCL13 score was quantified as the CXCL13-positive area ratio using ImageJ software. Based on the median CXCL13 score (0.932), patients were stratified into CXCL13 high-expression (CXCL13high) and low-expression (CXCL13low) groups. TLS was defined as aggregates of immune cells, predominantly composed of B cells surrounded by T cells. Immunohistochemical (IHC) staining for CD3 and CD20 was used to identify TLS-positive structures, characterized by CD3⁺ T cells encircling CD20⁺ B cells. TLS density was calculated as the medium number across 3 randomly selected high-power fields (HPFs) for each section. All assessments were independently performed by two pathologists. Significant discrepancies were resolved through review by two additional pathologists to reach consensus. Final results were further verified by an independent pathologist to ensure accuracy.

Immunofluorescence

We used CD3 and CD20 staining to identify TLS. After deparaffinization in xylene and rehydration in alcohol, samples were treated with 3% H2O2, followed by antigen retrieval using EDTA (pH 9.0). Samples were blocked with 3% BSA, then incubated overnight at 4 °C with CD3 and CD20 primary antibodies (1:500; Servicebio, Hubei, China). The next day, after PBS washes, Alexa Fluor 488-labeled secondary antibodies (1:400; Servicebio, Hubei, China) were applied for 15 min at 37 °C. DAPI staining followed for 10 min in the dark, with autofluorescence quenching and sealing of coverslips in an anti-fade mounting medium.

We performed multiplex immunofluorescence (mIF) for CD4 (1:500, Abcam, Cambridge, UK), CD8 (1:100, Abcam, Cambridge, UK), α-SMA (1:8000, Servicebio, Hubei, China), and CXCL13 (1:200, CST). After deparaffinization in xylene and rehydration in graded alcohol, antigen retrieval was done with EDTA (pH 9.0), followed by overnight incubation with CD4 primary antibodies at 4 °C. The next day, samples were treated with CD4 secondary antibodies (1:500, Servicebio, Hubei, China) for 50 min at room temperature, then with 647 reagent for 10 min in the dark at room temperature. The same procedure was repeated for α-SMA (1:500, Servicebio, Hubei, China) and CD8⁺ CXCL13 (CD8, 1:200; CXCL13, 1:300, Servicebio, Hubei, China), including respective primary and secondary antibody incubations. Finally, a DAPI staining solution and an autofluorescence quench agent were applied before mounting. Imaging was performed using an epifluorescence microscope (Nikon Eclipse C1, Nikon, Tokyo, Japan). CXCL13 expression was quantified as the CXCL13-positive area ratio, while CD4, CD8, and α-SMA expression were quantified as the positive cell ratio.

Analysis of CXCL13-producing cells

The single-cell RNA-sequencing data for skin from the Deeply Integrated human Single-Cell Omics data (DISCO, https://www.immunesinglecell.org) was used to investigate the CXCL13-producing cells. This skin atlas encompassed 18 different skin disorders, including cSCC, and more than 40 cell types were annotated.

Statistical analysis

IBM SPSS statistics version 22.0 (IBM Corp., Armonk, NY, USA) was used for data analysis, and GraphPad Prism 9.5 was used for graphic plotting. Heatmap and correlation plots were visualized in R. Clinical characteristics of patients were expressed as counts and percentages. Pearson correlation analysis was used to evaluate the relationships between CXCL13 and TLS density, as well as between CXCL13 and CD4, CD8, and α-SMA expression. Fisher’s exact test was used to analyze the association between CXCL13 and clinicopathological characteristics of cSCC patients. P value <0.05 was considered statistically significant.

CXCL13 is positively correlated with the immune infiltration

A total of 53 samples from the GEO database were kept for cell composition inference, which includes 40 primary cSCC, one metastatic cSCC, and 12 NS. Among the primary 40 cSCC, eight were well differentiated, 15 were moderately differentiated, 13 were poorly differentiated, and 17 cases were not categorized. Results showed that compared to the NS, cSCC exhibited significant enrichment of T cells, CD8⁺ T cells, cytotoxic cells, B cells, NK cells, monocytes, and macrophages, while myeloid dendritic cells, neutrophils, endothelial cells, and cancer-associated fibroblasts showed decreased presence in primary cSCC (Fig. 1A). CXCL13 was notably enriched in primary cSCC and correlated with increased levels of T cells (r = 0.4951, p = 0.0002), CD8⁺ T cells (r = 0.4881, p = 0.0002), cytotoxic cells (r = 0.4212, p = 0.0017), B cells (r = 0.3050, p = 0.0264), NK cells (r = 0.5111, p = 0.0001), monocytes (r = 0.4194, p = 0.0018), and macrophages (r = 0.4194, p = 0.0018) (Fig. 1B). Furthermore, survival analysis using in 517 cases of head and neck squamous cell carcinomas showed that patients with higher CXCL13 expressions were associated with better overall survival (p = 0.0032) (Fig. 1C). Collectively, this result indicates that CXCL13 is associated with immune infiltration in cSCC.

CXCL13 expression and clinicopathological correlations in cSCC

We investigated the correlation between CXCL13 expression and clinicopathological characteristics in 63 patients diagnosed with cSCC. Results revealed that patients exhibiting higher CXCL13 expression levels tended to have better pathological grades, suggesting a potential association between elevated CXCL13 density and favorable prognosis in cSCC. Furthermore, CXCL13 density was found to correlate with patient age, with those in the high CXCL13 group being older than those in the low CXCL13 group. While age itself does not directly impact disease progression, these findings suggest a potential age-related role for CXCL13 in the pathogenesis of cSCC, meriting further investigation. However, our study did not find any statistically significant correlation between CXCL13 expression density and other clinicopathological features such as subcutaneous invasion, lymphovascular invasion, perineural invasion, tumor thickness, gender, and site of sun exposure (Table 1).

Association between CXCL13 expression and TLS was validated in clinical samples

CXCL13 is recognized as a marker for TLS (Petitprez et al., 2020). To explore the relationship between CXCL13 and TLS, we performed IHC analysis, which revealed CXCL13 positivity in 96.8% of all cSCC sections, indicating its ubiquitous presence in cSCC. CXCL13 expression was quantitatively assessed by calculating the positive area ratio at 100 × magnification using ImageJ software. Based on the median percentage of CXCL13-positive cells (0.932), cases were stratified into CXCL13high and CXCL13low groups, comprising 47.6% and 52.4% of cases, respectively (Figs. 2A–2B).

TLS were defined as structures containing CD20⁺ B cells surrounded by CD3⁺ T cells at the periphery. TLS density was determined as the medium count across three randomly for each section (Fig. S1). TLS were identified in 69.8% of the samples (Figs. 2C–2E). Pearson correlation analysis was performed to evaluate the association between CXCL13 expression and TLS (Fig. 2F). The results demonstrated a significant positive correlation between CXCL13 expression levels and TLS presence (r = 0.7126, p < 0.001), providing strong theoretical support for further investigation into the role of CXCL13 in the tumor immune microenvironment.

CXCL13 was mainly produced by fibroblasts and T cells

To elucidate the primary cellular sources of CXCL13 in skin diseases, we conducted an in-depth analysis of the ImmuneSingleCell skin atlas, a comprehensive database encompassing 18 distinct skin disorders, including cSCC, basal cell carcinoma, keloid, and psoriasis (Fig. 3A). This dataset profiles 40 cell types, such as epithelial cells, fibroblasts, CD8⁺ T cells, CD4⁺ T cells, Tregs, NK cells, and other immune cell populations (Fig. 3B). Subsequently, we investigated the expression patterns of CXCL13 within this dataset. Our analysis revealed that CXCL13 is predominantly expressed in CD8⁺ T cells, CD4⁺ T cells, and fibroblasts (Figs. 3C–3D).

To validate these findings, we performed mIF staining on clinical cSCC tissue samples. Fibroblasts were identified using alpha smooth muscle actin (αSMA) as a marker, followed by mIF co-staining with CXCL13 (Kalluri, 2016). Triple staining for CD4, CD8, and CXCL13 confirmed that CXCL13 is actively secreted by both CD8⁺ T cells and CD4⁺ T cells within these samples, consistent with our bioinformatic analysis (Figs. 3E–3F). For quantitative assessment, we selected four tissue sections each for CXCL13 co-staining with CD4⁺ T cells, CD8⁺ T cells, and cancer-associated fibroblasts (CAFs). In each section, five randomly selected HPFs at 100 × magnification were analyzed to quantify the mIF staining results. Correlation analysis demonstrated strong and statistically significant associations between CXCL13 and CD4+ T cells (r = 0.5046, p =0.0233), CD8⁺ T cells (r = 0.9318, p < 0.0001), and CAFs (r = 0.6500, p = 0.0019) (Fig. 3G).

In conclusion, our integrated approach, combining public dataset analysis and clinical validation, demonstrates that CXCL13 is primarily produced by fibroblasts and T cells in skin diseases.