2.1 Collection of tissue samples

Total 68 pairs of TNBC tissues and adjacent normal tissues were acquired from TNBC patients of Fuzhou First Hospital Affiliated to Fujian Medical University between August 2019 and February 2021. None of patients received radiotherapy or chemotherapy before surgery, and patients with other diseases were excluded from this investigation. This study got permission from the Ethics Committee of Fuzhou First Hospital Affiliated to Fujian Medical University, and all participants have signed informed consents. Moreover, TNBC patients were assigned into COL3A1 low or high-expression group on the basis of the median value of COL3A1 expression in tumor tissues.

2.2 Cell culture

A human normal breast epithelial cell line (MCF10A) and human TNBC cell lines (BT-549, MDA-MB-231, and MDA-MB-468) were bought from American Type Culture Collection (ATCC). Dulbecco's modified Eagle's medium (DMEM; Invitrogen) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin was utilized for cell culture. All cells were maintained at 37°C in an incubator with 5% CO₂.

2.3 Cell transfection

The small interfering RNA against COL3A1 (si-COL3A1#1, si-COL3A1#2, and si-COL3A1#3), si-negative control (NC), pcDNA3.1-PD-L1 (including full length of PD-L1), and pcDNA3.1-NC were bought from Genepharma. Then these plasmids were transfected into BT-549 and MDA-MB-231 cells through Lipofectamine 3000 (Invitrogen) for 48 h.

2.4 Quantitative real-time polymerase chain reaction (qRT-PCR)

Total RNAs of TNBC tissues were obtained using RNAiso Plus reagent (Takara), which were then used for synthesis of complementary DNAs (cDNAs) according to instructions of a PrimeScript RT reagent kit (Takara). After that, cDNAs were applied to perform qRT-PCR through the SYBR Green PCR kit (Takara). Reaction procedures for PCR were exhibited as follows: 5 min at 94°C, 35 cycles of 20 s at 94°C, 40 s at 60°C, and 2 min at 72°C. Primer sequences (Sangon) were listed as follows: COL3A1-F, TTGAAGGAGGATGTTCCCATCT; COL3A1-R, ACAGACACATATTTGGCATGGTT; GAPDH-F, CTGACTTCAACAGCGACACC; GAPDH-R, CCCTGTTGCTGTAGCCAAAT. The 2^−ΔΔct method was used for calculating relative mRNA expression of COL3A1 normalized to GAPDH.

2.5 CCK-8 assay

Proliferation of MDA-MB-231 and BT-549 cells was tested based on instructions of a CCK-8 kit (Boster Biological Technology). In detail, transfected cells (5 × 10³ cells/well) were put into 96-well plates, which were maintained for 0, 24, 48, and 72 h. Then, 10 μl CCK-8 solution was appended to each well to incubate for 1 h. After plates were washed with phosphate buffered solution, a microplate reader (MG LABTECH) was employed to monitor the optical density at 450 nm.

2.6 5-Ethynyl-2′-deoxyuridine (EdU) proliferation assay

The ability of MDA-MB-231 and BT-549 cells to proliferate was evaluated using EdU proliferation assay (RiboBio). Briefly, cells were subjected to EdU (50 μm) to incubate for 2 h. Afterward, these cells were stained using Apolo and 4′,6-diamidi-no-2-phenylindole and a fluorescence microscopy was used for detection of EdU-positive cells.

2.7 Flow cytometry analysis

Transfected MDA-MB-231 and BT-549 cells (1 × 10⁵ cells/well) were plated in six-well plates, followed by incubation with 5 μl Annexin V-FITC for 10 min and 10 μl PI for 15 min in the dark. Finally, a flow cytometer (BD Biosciences) was applied to assess cell apoptosis rate.

Moreover, transfected cells were cocultured with CD8⁺ T cells, after which flow cytometry was used for detecting the percentage of CD8⁺ T cells or apoptosis of CD8⁺ T cells.

2.8 Transwell assay

The transwell chamber (8.0 μm pore size; Millipore) coated with Matrigel (BD Biosciences) was utilized for evaluating the invasive capacity of transfected MDA-MB-231 and BT-549 cells. In short, the upper chamber was filled with 100 μl cell suspension suspended in FBS-free DMEM. Then, 600 μl DMEM containing 10% FBS was appended to the lower chamber. Following 48 h of incubation, cells in the lower chamber were fixed using methanol and then stained using crystal violet. An optical microscope was utilized for counting the number of stained cells.

Besides, the cell migratory ability was assessed via the upper chamber without Matrigel and other procedures were the same as the evaluation of cell invasion.

2.9 Lactate dehydrogenase (LDH) cytotoxicity assay

A LDH assay kit (Beyotime) was used for measuring LDH leakage. This assay was executed in 96-well plates, and a microplate reader (Thermo Fisher Scientific) was employed to measure the absorbance at 490 nm.

2.10 Western blot analysis

RIPA lysis buffer (Beyotime) was utilized for extracting proteins from TNBC cells and tissues, after which protein concentration was measured via a BCA Protein Assay Kit (Beyotime). Next, 10% SDS-PAGE was applied to separate these proteins. Then, these proteins were transferred to polyvinylidene difluoride membranes (Millipore), followed by block using 5% skim milk. Subsequently, primary antibodies (Abcam) including anti-COL3A1 (1:1000, ab184993), anticleaved caspase-3 (1:500, ab32042), anti-Bcl-2 (1:1000, ab32124), anti-PD-L1 (1:1000, ab205921), and anti-β-actin (1:200, ab115777) were added to immerse membranes overnight. After being washed using TBST, the secondary antibody (1:2000, ab6721) was added to react for 1 h. Eventually, the enhanced chemiluminenscent detection system was employed to examine protein blots. Image J software (NIH) was used for analyzing the intensity of protein blots.

2.11 Immunohistochemistry (IHC) staining

TNBC tissues fixed by 4% paraformaldehyde were embedded by paraffin, after which these tissues were cut into segments (4 μm), dewaxed, and rehydrated. Following antigen repair and seal of nonspecific sites, these sections reacted with primary antibodies including anti-COL3A1 (1:500, ab6310; Abcam), anti-Ki67 (1:200, ab16667; Abcam), and anticleaved caspase-3 (1:400, #9661; CST) overnight. After that, the secondary antibody (1:500, ab6112; Abcam) was added, followed by addition of 3,3′-diaminobenzidine substrate solution. At last, hematoxylin was used for staining these samples and staining areas were observed through a microscope.

2.12 In vivo experiment

Procedures of animal experiments have got permission from the Institutional Animal Care and Use Committee of Fuzhou First Hospital Affiliated to Fujian Medical University, and conformed to the guide for care and use of laboratory animals published by the NIH. BALB/C nude mice (6-week-old, female) from Chinese Academy of Sciences were kept at 25°C under the condition of 50% humidity and a 12h light/dark cycle. For stable transfection, lentiviral sh-NC (LV-sh-NC) and lentiviral sh-COL3A1 (LV-sh-COL3A1) were bought from Genechem. All mice were assigned into two groups (n = 5 mice/group). MDA-MB-231 cells (5 × 10⁶ cells in 0.1 ml/mouse) carrying LV-sh-NC/LV-sh-COL3A1 were introduced into the flank of mice through subcutaneous injection (Fan et al., 2021). The tumor size of mice was gauged every week after injection. Tumor volume was measured with the formula V = (shortest diameter)² × (longest diameter) × 0.5. Following 4 weeks of injection, mice anesthetized using pentobarbital sodium were killed and tumors were separated from mice.

2.13 Statistical analysis

GraphPad Prism 7.0 (GraphPad Software Inc.) was employed to perform statistical analysis. Data from three independent experiments were displayed as mean ± standard deviation. Difference evaluation between two groups was carried out using Student's t test. Difference estimation among multiple groups was performed via one-way analysis of variance, followed by Tukey's multiple comparisons test. Statistical significance was identified as any difference with p < .05.

3.1 COL3A1 upregulation is observed in TNBC tissues and cells

First of all, we analyzed COL3A1 expression using the Gene Expression Profiling Interactive Analysis (GEPIA) database, and discovered that COL3A1 expression in tumor tissues was remarkably higher than that in adjacent normal tissues (Figure 1a, p < .05). Based on data from qRT-PCR, we also observed high expression of COL3A1 in tumor tissues compared to adjacent normal tissues (Figure 1b, p < .01). Afterward, we verified correlation between COL3A1 expression and survival of TNBC patients. Indeed, COL3A1 upregulation resulted in poor survival data of TNBC patients (Figure 1c, p < .01). Furthermore, protein expression of COL3A1 was determined via western blot and IHC staining, and found that the protein expression COL3A1 in tumor tissues was boosted compared to normal tissues (Figure 1d,e, p < .05).

When we assessed protein expression of COL3A1 in human normal breast epithelial cells (MCF10A) and TNBC cells, COL3A1 upregulation was observed in BT-549, MDA-MB-231, and MDA-MB-468 cells as opposed to MCF10A cells (Figure 1f, p < .01). MDA-MB-231 and BT-549 cells were used to perform subsequent experiments since they showed higher expression of COL3A1 than MDA-MB-468 cells.

3.2 Silencing of COL3A1 represses proliferation and induced apoptosis in TNBC cells

In consideration of COL3A1 upregulation in TNBC, we then explored the potential role of COL3A1 via loss-of-function assays. As displayed in Figure 2a, COL3A1 was silenced by transfection of si-COL3A1#1, si-COL3A1#2, and si-COL3A1#3 in MDA-MB-231 and BT-549 cells (p < .05), and si-COL3A1#3 was used for following function experiments. Based on results of CCK-8 and EdU assays, we discovered that the proliferation capacity of MDA-MB-231 and BT-549 cells was evidently suppressed after COL3A1 knockdown (Figure 2b,c, p < .01). According to data of flow cytometry (Figure 2d), we found that the apoptotic rate of MDA-MB-231 and BT-549 cells was markedly elevated following COL3A1 knockdown (p < .001). At the same time, expression levels of apoptosis-related proteins were detected using western blot, which indicated that COL3A1 knockdown facilitated protein expression of cleaved caspase-3 but repressed protein expression of Bcl-2 in MDA-MB-231 and BT-549 cells (Figure 2e, p < .01).

3.3 COL3A1 downregulation impedes migration, invasion, and immune escape of TNBC cells

Afterward, malignant behavior changes of TNBC cells affected by COL3A1 were explored. It was demonstrated that migration and invasion abilities of MDA-MB-231 and BT-549 cells were attenuated by COL3A1 downregulation (Figure 3a,b, p < .01). Moreover, immune escape of cancer cells to CD8⁺ T cells is a key inducement for cancer progression, accompanied by upregulation of PD-L1 expression and exclusion of CD8⁺ T cells (Fang et al., 2021; Liu et al., 2020; Spranger, 2016). Thus, for probing the effect of COL3A1 downregulation on immune escape of TNBC cells, following experiments were implemented. It was demonstrated that COL3A1 knockdown led to the increase of LDH release in MDA-MB-231 and CD8⁺ T cocultured cells or BT-549 cells and CD8⁺ T cocultured cells (Figure 3c, p < .01). Besides, the percentage of CD8⁺ T cells was raised in response to COL3A1 knockdown, and the apoptosis rate of CD8⁺ T cells was reduced in response to COL3A1 knockdown in MDA-MB-231 and BT-549 cells (Figure 3d,e, p < .01). The protein level of PD-L1 was diminished by COL3A1 knockdown in MDA-MB-231 and BT-549 cells (Figure 3f, p < .01). These findings suggested the oncogenic property of COL3A1 in TNBC.

3.4 PD-L1 overexpression partly reverses inhibitory impacts of COL3A1 knockdown on proliferation, migration, invasiveness and immune escape of TNBC cells

To verify whether PD-L1 was implicated in COL3A1 knockdown-mediated inhibitory effect on TNBC progression, PD-L1 was firstly overexpressed through transfection of pcDNA3.1-PD-L1. As expected, PD-L1 expression was notably boosted after transfection of pcDNA3.1-PD-L1 in MDA-MB-231 and BT-549 cells (Figure 4a, p < .01). Subsequently, rescue assays of cell proliferation, apoptosis, migration, invasion, and immune escape were conducted. It was demonstrated that the suppressive effect of COL3A1 knockdown on cell proliferation and the promoting effect of COL3A1 knockdown on cell apoptosis were reversed by PD-L1 overexpression (Figure 4b,c, p < .01). Likewise, migration and invasion repression of MDA-MB-231 and BT-549 cells caused by COL3A1 knockdown was reversed by PD-L1 overexpression (Figure 4d,e, p < .05). In addition, the promoting effects of COL3A1 knockdown on LDH release and the percentage of CD8⁺ T cells as well as the suppressive effect of COL3A1 knockdown on apoptosis rate of CD8⁺ T cells were partly abolished by PD-L1 overexpression in MDA-MB-231 and BT-549 cells (Figure 4f–h, p < .05). Above outcomes indicated that COL3A1 knockdown made impact on TNBC progression via regulating PD-L1 expression.

3.5 Silencing of COL3A1 represses tumor growth in vivo

At last, the influence of COL3A1 on tumor growth was validated in xenograft mice. As manifested in Figure 5a,b, tumor size and tumor volume in the LV-sh-COL3A1 group were smaller than that in the LV-sh-NC group (p < .05). Tumor weight in the LV-sh-COL3A1 group was lower than that in the LV-sh-NC group (Figure 5c, p < .001). In contrast to the LV-sh-NC group, Ki67 expression in tumor tissues was evidently reduced in the LV-sh-COL3A1 group and cleaved caspase-3 expression in tumor tissues was evidently boosted in the LV-sh-COL3A1 group, indicating that COL3A1 downregulation suppressed cell proliferation in tumors (Figure 5d). Besides, regulatory relation between COL3A1 and PD-L1 was verified in tumor tissues, which indicated that protein expression of PD-L1 was diminished after COL3A1 knockdown (Figure 5e, p < .01).