In this study, 5 datasets (Table 1) were used for differential expression analysis using online tool GEO2R (https://www.ncbi.nlm.nih.gov/geo/geo2r/) and R (version 4.1.1, The R Foundation, https://www.r-project.org/) package limma. Based on the percentile of C1R expression, we divided the samples in the GSE53625 dataset into three groups (40%, 40%–80%, $>$80%) and the prognostic value of C1R mRNA was assessed using the Survfit function in the R software’s survival package.

The raw data (GSE188900) of scRNA-seq experiments was analyzed and integrated by R packages Seurat and Harmony. For quality control, the standard used is the same as the reference [18]. R package SingelR and CellMarker database [19], cell markers [18, 20, 21, 22, 23] were used to annotate cell types, and R package CopyKat was used to identify malignant epithelial cells [24]. Only aneuploid cells were considered malignant epithelial cells, while other cells were considered non-malignant. Gene ontology (GO) and gene set variation analysis (GSVA) were used to enrich the pathways in each group or cell.

Using the median C1R expression value, tumor samples in GSE44021 were divided into low-expression and high-expression groups. GSEA was subsequently used to evaluate different functions in two groups with different C1R expression.

Based on esophageal carcinoma (ESCA) data in TCGA, online tool TIMER2.0 (http://timer.cistrome.org/) was used for the systematic analysis of the infiltration of various immune cells in ESCA [25]. Correlation modules of TIMER2.0 were used to reveal the co-expression of C1R and tumor-infiltrating lymphocyte (TILs) gene markers, including gene markers for macrophages, myeloid cells, and cancer-associated fibroblasts (CAFs). TIMER2.0 did not distinguish ESCA data into ESCC or EAC.

The human ESCC TMA (HEsoS180Su11, Shanghai Outdo Biotech Co., Ltd., Shanghai, China) used, contained a total of 180 tissues, specifically, 72 pairs of ESCC tissues and matched paratumor tissues, and 36 non-matched ESCC tissues. The TMA was deparaffinized, rehydrated, subjected to antigen retrieval, and then incubated with diluted primary antibodies to C1r (1:2000, Proteintech, Rosemont, IL, USA) at 4 °C overnight. After blocking and incubating with secondary antibody, EnVision™ FLEX+ and Autostainer Link 48 (Dako, Glostrup, Denmark) were used for signal development. Finally, the TMA was counterstained with hematoxylin solution for 1 min and in 0.25% hydrochloric-alcohol solution for 10 sec.

Staining intensity was classified into four degrees (0–3) of intensity: no staining, weak staining, moderate staining, strong staining. The number of positive cells within a field was divided into five degrees (0, 5%; 1, 5%–24%; 2, 25%–49%; 3, 50%–74%; 4, $\ge$75%). Finally, the IHC staining score was calculated using the formula: staining intensity score $\times$ positive-staining score. According to this IHC staining scoring scheme, 106 ESCC samples (two of the tumor samples were missing from the TMA) were divided into a high C1r expression group (staining score $>$6) and the low C1r expression group (staining score $\le$6).

The ESCC cell lines (TE-1, Eca-109, and KYSE-150) and normal human esophageal epithelial cell line (HEEC) were separately obtained from the Cell Bank of Type Culture Collection of the Chinese Academy of Science (Shanghai, China) and Sciencell (Carlsbad, CA, USA). To identify the cell lines, the short tandem repeat (STR) was conducted. All cells were maintained in RPMI 1640 medium (Gibco, Waltham, MA, USA) containing 10% fetal bovine serum (Every Green, Zhejiang, China), 100 U/mL penicillin, and 100 µg/mL streptomycin (HyClone, Austria) in a 5% CO${}_2$ humidified incubator at 37 °C. MycoBlue Mycoplasma Detector (Vazyme, Nanjing, China) was used to detect mycoplasma contamination.

Lipofectamine™ 2000 (Invitrogen, Waltham, MA, USA) was used to transiently transfect cells with C1r-specific small interfering RNAs (C1r siRNAs) or negative control siRNA (NC siRNA). The siRNAs used were listed in Table 2 [14] and were obtained from GenePharma (Shanghai, China). Commercially available C1r knockdown lentivirus (C1r-specific short hairpin RNA, C1r shRNA) or the control (CTRL), and C1r overexpression lentivirus (C1r OE) or the negative control (NC) were designed and produced by Genechem (Shanghai, China). TE-1 or Eca-109 cells were infected with these lentiviruses and selected with puromycin (Solarbio, Beijing, China). The shRNAs used were listed in Table 3.

RIPA lysis buffer containing PMSF (Beyotime, Shanghai, China) was utilized to extract cellular protein. Using the protocol supplied with the SDS-PAGE gel preparation kit (Beyotime), after protein denaturation, 8% gel was used to separate the proteins. Then the gel contents were subsequently transferred to PVDF membranes (Beyotime). Membranes were subsequently blocked, and incubated with primary antibodies to C1r (1:2000, Proteintech), MMP-1 (1:2000, Proteintech), MMP-10 (1:1000, Beyotime), or $\beta$-actin (1:2000, Beyotime) at 4 °C overnight. Following this, the membranes were incubated with horseradish peroxidase-labeled secondary antibody (1:1000, Beyotime). Protein bands were visualized using Electrochemical Luminescence star kit (Beyotime, China), and ChemiDoc™ XRS+ with Image Lab™ software (version 6.1.0, Bio-Rad, Hercules, CA, USA) was used for protein band quantification.

5 $\times$ 10${}^3$ cells/well of ESCC cells were seeded in 96-well plates. Cells were transfected with siRNAs and then further incubated. After transfection for 24, 48, and 72 hr, Cell Counting Kit-8 (CCK-8) reagent (Beyotime) was added. Absorbance at 450 nm could represent the proliferation status of the responsive cells. The proliferation assay of stably transduced cell lines was the same as outlined above except no transfection after cell seeding was conducted.

After reaching 90% confluence in 12-well plates, ESCC cell monolayers were scratched, transfected with siRNA, and cultured in an incubator. At 0, 24, 48, and 72 hr, images were taken with an inverted microscope (Olympus IX71, Olympus Optical, Tokyo, Japan). ImageJ (version 1.51, https://imagej.net/) was used to calculate the migration rate. Similarly, the procedure for the stable cell lines was the same as mentioned above, but there was no transfection step after the cells were seeded.

To assess the impact of C1r on the invasive potential of ESCC cell lines, a Transwell invasion assay was conducted. Matrigel (Corning, Corning, NY, USA) was diluted in serum-free Roswell Park Memorial Institute (RPMI) 1640 medium and added to the upper chambers (Corning). ESCC cells were subjected to a 24 hr starvation period in serum-free RPMI 1640 medium. Subsequently, cells (2 $\times$ 10${}^6$) in 200 µL serum-free RPMI 1640 medium were seeded in the upper chambers, while the 600 µL serum-free RPMI 1640 medium containing 10% FBS was added to lower chambers. Following incubation at 37 °C for 24–36 hr, non-invading cells were removed, and the invaded cells were fixed with 4% paraformaldehyde (Biosharp, Hefei, China) and stained with 0.1% crystal violet for 30 min. Quantification of cells that traversed the Matrigel and migrated to the lower chambers was performed using ImageJ.

For cell-matrix adhesion assay, 10 µg/mL fibronectin (Solarbio) was coated into 96-well plates. After being blocked by 1% bovine serum albumin for 1 hr at 37 °C, 5 $\times$ 10${}^4$ cells/well of ESCC cells in RPMI 1640 medium were seeded in these 96-well plates. After incubating for 2 hr at 37 °C, adherent cells were stained with 0.1% crystal violet at room temperature for 30 min. The intracellular stain was solubilized in 10% acetic acid and absorbance was measured at a wavelength of 595 nm.

A 60% confluent culture of ESCC cells was transfected with indicated siRNAs and then further cultured for 48 hr. Per the instructions of the Annexin V-APC/PI Apoptosis Detection Kit (KeyGEN BioTECH, Nanjing, China), cells were stained, and apoptosis was analyzed by flow cytometry using a NovoCyte™ Flow Cytometer (ACEA Biosciences, Santa Clara, CA, USA).

BALB/c nude mice (female, 4–6 weeks old, weighing 14 g–16 g) were provided by the Experimental Animal Study Center of North Sichuan Medical College. Considering different efficiencies of shRNAs, C1r shRNA1 was ultimately used. To establish a xenograft mouse model (n = 5 in each group), TE-1 cells with C1r knockdown (TE-1 CTRL/TE-1 C1r shRNA1, 3 $\times$ 10${}^6$ cells in 100 µL phosphate buffered saline (PBS)) or Eca-109 cells with C1r overexpression (Eca-109 NC/Eca-109 C1r OE, 1.5 $\times$ 10${}^6$ cells in 100 µL PBS) were subcutaneously injected into the right armpit of nude mice. After 25 days (TE-1 CTRL or TE-1 C1r shRNA1) and 35 days (Eca-109 NC or Eca-109 C1r OE), the tumors were completely dissected and photographed, and the tumor volume and weight were measured.

For data analysis, SPSS 23.0 (IBM Corp., Chicago, IL, USA) statistical software was used. The data are presented as mean $\pm$ standard deviation (SD) and all assays were repeated in triplicate. Graphical representations were carried out with GraphPad Prism 8 software (GraphPad Software, San Diego, CA, USA). Depending on the experimental design, homogeneity of variance, and data distribution characteristics, different statistical tests (one-way ANOVA, two-way ANOVA, t-test) were used to analyze the significance of differences between groups, with p 0.05, meaning statistically significant. Dunnett’s test was used for adjusting multiple comparisons. The relationship between the clinicopathological variables and C1r expression was assessed by $\chi$${}^2$ test.

Multiple differentially expressed complement components in four GEO datasets were integrated and displayed through heatmaps. According to results obtained using GEO2R, the expression of C1QA, C1QB, C1QC, C1R, C1S, C2, and CFB genes showed a significant trend of high expression in ESCC tissues (p 0.05). In contrast, the expression of terminal pathway components such as C6 and C7 was reduced (Fig. 1A). Further, high expression of the C1R gene was significantly associated with a shorter overall survival (Fig. 1B).

Fig. 1.

C1R is highly expressed in ESCC tissues. (A) Heatmaps showed the differential expression of various complement genes in normal tissues and ESCC tissues. (B) Kaplan-Meier curves indicated the correlation of C1R with overall survival. * adjusted p-value 0.05, ** adjusted p-value 0.01, *** adjusted p-value 0.001, **** adjusted p-value 0.0001.

In scRNA-seq experiments, 22,317 cells were divided into 9 cell types (Supplementary Fig. 1A,B). To further examine the expression of C1R in epithelial cells, epithelial cells were further separated into malignant and non-malignant epithelial cells (Supplementary Fig. 1C,D). Cluster 1, 5, and 8 were annotated as malignant epithelial cells, while other clusters were annotated as non-malignant epithelial cells (Fig. 2A). We found that the malignant epithelial cells highly expressed the C1R gene (Fig. 2B). Cluster 5 highly expressed C1R, KRT14, CD74, HLA-DRA, and HLA-DRB1, which is an epithelial-immune dual feature of malignant cells previously reported in nasopharyngeal carcinoma (Fig. 2C,D) [26]. The results of GSVA revealed that Cluster 5 was enriched in antibody processing and presentation, NOD-like receptor and p53 signaling pathways. In Clusters 1 and 5, apoptosis, Toll-like receptor signaling pathway, and MAPK signaling pathways were enriched, while Cluster 8 was enriched in the TGF beta signaling pathway and pathways that regulate autophagy (Fig. 2E). The results of GO analysis revealed that compared with other malignant cell clusters, Cluster 5, which has a high expression of C1R, was also enriched in genes associated with epithelial cell proliferation, regulation of apoptotic signaling, cell-substrate junction, and DNA-binding transcription factor binding (Fig. 2F).

Fig. 2.

C1R is highly expressed in malignant epithelial cells in ESCC. (A) A uniform manifold approximation and projection (UMAP) plot showed subclusters of epithelial cells. (B) A Dot plot showed the expression level for complement component genes in malignant cells and non-malignant cells. (C) Density map indicated the expression level of C1R and KRT14. (D) Up and down-regulated genes within tumor clusters, or between non-malignant epithelial cells and malignant epithelial cells, were shown. (E) A circle heatmap generated by gene set variation analysis (GSVA) indicated the differences in enriched pathways of the four epithelial cell subtypes. (F) The results of gene ontology (GO) enrichment were shown using a dot plot.

Based on the median expression of C1R, samples in GSE44021 were divided into two groups. Using GSEA (Fig. 3A), the significant enrichment pathways for the biological function of C1R were cytokine-receptor interactions, focal adhesion, ECM receptor interaction, NOD-like receptor signaling pathway, complement and coagulation cascades, and apoptosis.

Fig. 3.

C1R expression is correlated with extracellular matrix (ECM) receptor regulatory networks and immune infiltration. (A) Gene set enrichment analysis (GSEA) analysis indicated the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways associated with C1R expression. (B) A dot plot indicated the expression levels of matrix metalloproteinases (MMPs) in GSE53625. (C,D) The correlation was shown for C1R expression and MMPs expression. (E) The correlation of C1R expression with immune cell infiltration levels in esophageal carcinoma (ESCA) was shown.

The ECM receptor regulatory network plays an essential role in promoting the adhesion, proliferation, and migration of tumor cells, and influencing the tumor microenvironment (TME). Matrix metalloproteinases (MMPs) are involved in the remodeling of the ECM, and some of these, such as MMP-1, MMP-10, and MMP-13, were highly expressed in ESCC tissues (Fig. 3B). We found that the expression of C1R was significantly co-expressed with MMP-1 and MMP-10 (Fig. 3C,D).

As a part of TME, immune infiltration is essential in promoting malignant progression. In this analysis, C1R expression had a negative correlation with the purity of ESCA (rho = –0.262, p = 3.61 $\times$ 10${}^{-4}$). According to the results obtained, samples exhibited high-expression of C1R had a high degree of infiltrating CD8${}^{+}$ T cells (rho = 0.38, p = 1.48 $\times$ 10${}^{-7}$), M2 macrophages (rho = 0.514, p = 1.67 $\times$ 10${}^{-13}$), CAFs (rho = 0.753, p = 3.76 $\times$ 10${}^{-34}$), and myeloid dendritic cells (rho = 0.551, p = 1.13 $\times$ 10${}^{-15}$) (Fig. 3E).

The abundance of C1r protein in an ESCC TMA was analyzed by IHC. C1r was found to be located in the cytoplasm of cancer cells and normal esophageal mucosal epithelial cells (Fig. 4A,B). IHC staining scores indicated that when compared to normal esophageal mucosal epithelium, C1r protein abundance in ESCC tissues was significantly increased (Fig. 4C. paired t-test). The high expression of C1r in cancer cells was also significantly correlated with histological grade, but not with other patient/tumor variables including age, gender, pathologic type, size, lymph node metastasis, infiltration degree, and TNM stage (Table 4). Representative IHC images for C1r with different histological grade ESCC tumors is shown in Fig. 4D. High expression of C1r was significantly associated with poor prognosis (Fig. 4E). Western blot analysis showed that C1r was expressed in both normal HEEC and three ESCC cell lines. C1r abundance in all three ESCC cell lines was notably higher than that observed in the HEEC cell line, and was highest in TE-1 cells (Fig. 4F).

Fig. 4.

C1r protein is highly expressed in ESCC tissues and ESCC cell lines, and is associated with poor prognosis. (A) Hematoxylin-eosin (HE) staining was used to identify pathological changes in ESCC tissues. (B) The expression patterns of C1r in paired ESCC and paratumor tissues was shown. (C) Immunohistochemistry (IHC) staining score was evaluated in paired ESCC and normal tissues. (D) Representative IHC images showing C1r expression in different histological grades were shown. (E) Kaplan-Meier survival analysis was performed to assess the correlation of C1r with overall survival in the low and high IHC staining score groups. (F) The expression of C1r in ESCC cells and human esophageal epithelial cell line (HEEC) cells was evaluated by western blotting. **** p 0.0001.

The expression of C1r was significantly reduced in C1r knockdown TE-1 cells. Further, the expression of C1r was significantly higher in Eca-109 cells overexpressing recombinant C1r than that documented in negative control cells (Fig. 5A). In C1r knockdown TE-1 cells, the proliferation rate of these cells was significantly reduced. In contrast, the proliferation rate of Eca-109 cells overexpressing recombinant C1r was increased (Fig. 5B).

Fig. 5.

C1r promotes the proliferation, migration, invasion, cell-matrix adhesion, and inhibits the apoptosis of ESCC cells. (A) The expression of C1r in C1r knockdown TE-1 cells and C1r overexpressing Eca-109 cells was evaluated by western blotting. (B) The impact of C1r on the proliferation of ESCC cells was quantified by Cell Counting Kit-8 (CCK-8) assay. (C) The impact of C1r on the migration and invasion of ESCC cells was evaluated by a wound healing assay and Transwell invasion assays. (D) The impact of C1r on the cell matrix adhesion of ESCC cells was evaluated by crystal violet staining. (E) Apoptosis was measured in C1r knockdown TE-1 cells by flow cytometry. NS p $>$ 0.05, * p 0.05, ** p 0.01, *** p 0.001, **** p 0.0001.

Wound healing assays and Transwell invasion assays were performed separately to determine the effect of C1r expression on the migration and invasion ability of ESCC cells. Results obtained indicated that the migration and invasion in C1r knockdown TE-1 cells decreased significantly. Coordinately, the migration and invasion of Eca-109 cells overexpressing recombinant C1r was found to be increased compared to the negative control group (Fig. 5C).

To better understand the effect of C1r on human ESCC cell extracellular matrix adhesion, a cell-matrix adhesion experiment was conducted using fibronectin. The results showed that fibronectin adhesion was significantly decreased in TE-1 cells with C1r knockdown. In contrast, in Eca-109 cells overexpressing C1r exhibited significantly increased fibronectin adhesion when compared to negative control cells (Fig. 5D) strongly suggesting that C1r promotes cell-matrix adhesion in ESCC cells.

Next, the impact of C1r on apoptosis in human ESCC cell lines was assessed through flow cytometry. C1r knockdown resulted in significant increases in apoptotic (Annexin V/PI-positive) TE-1 cells (Fig. 5E), leading us to conclude that C1r decreases apoptosis in ESCC cells.

The result of GSEA suggested that C1R expression is closely associated with the expression of MMP-1 and MMP-10. Thus, we next examined the expression of MMP-1 and MMP-10 in C1r knockdown and C1r overexpressing ESCC cell lines by western blotting. In response to C1r knockdown in TE-1 cells, the expression of MMP-1 and MMP-10 significantly decreased. Conversely, overexpression of C1r in Eca-109 cells resulted in heightened expression of both MMP-1 and MMP-10 (Fig. 6A,B).

Fig. 6.

C1r influences the expression of MMP-1 and MMP-10 in ESCC cells. (A) The expression of MMP-1 and MMP-10 in C1r-knockdown TE-1 cells was evaluated and quantified. (B) The expression of MMP-1 and MMP-10 in C1r overexpressing Eca-109 cells was evaluated. * p 0.05, ** p 0.01.

To further examine the effect of C1r on the growth of ESCC cells in vivo, TE-1 cells with C1r knockdown (TE-1 CTRL/TE-1 C1r shRNA1) or Eca-109 cells with C1r overexpression (Eca-109 NC/Eca-109 C1r OE) were subcutaneously injected into the armpit of the nude mice. After injection, we observed that all mice grew tumors, and the tumor formation rate in both groups was 100%. However, TE-1 cells in both CTRL and C1r shRNA1 groups exhibited extremely slow growth in nude mice. After complete dissection of the tumor, we observed a significant reduction in tumor weight in the C1r shRNA1 group compared to the CTRL group (Fig. 7A,B). In contrast, the average tumor weight of the C1r overexpression group was 0.5508 $\pm$ 0.1809 g, which was significantly higher than 0.3012 $\pm$ 0.0674 g measured in the negative control group (Fig. 7C,D).

Fig. 7.

C1r promotes ESCC tumor growth in vivo. (A) Cell-based (TE-1 CTRL/TE-1 C1r shRNA1) xenografts in nude mice were completely dissected and photographed 25 days post-implantation into nude mice. (B) The tumor weights of C1r knockdown group and negative control group were confirmed at the end of the experiment. (C) Cell-based (Eca-109 NC/Eca-109 C1r OE) xenografts were completely dissected and photographed 35 days post-implantation into nude mice. (D) Dissected xenografts of C1r overexpression group and control group were weighed at the end of the experiment.