Bioinformatics analysis

The PTX resistance-related genes were screened from the GEO databases GSE28784 and GSE90564, selecting significantly differentially expressed genes (DEGs) with p < 0.05. The list of human transcription factors was obtained from the HumanTFDB system (http://bioinfo.life.hust.edu.cn/HumanTFDB#!/download). The hTFtarget system (http://bioinfo.life.hust.edu.cn/hTFtarget/#!/) was then utilized to predict downstream target genes of the transcription factor CEBPD. The Jvenn system (https://jvenn.toulouse.inrae.fr/app/example.html) was used to generate a Venn diagram illustrating the intersections of different datasets. Kaplan–Meier Plotter (https://kmplot.com/analysis/) was employed to analyze the prognostic significance of gene expression on overall survival (OS) in breast cancer patients.

Clinical samples

Post-chemotherapy biopsy tissue samples from 41 patients with TNBC were included for this research. Based on the therapeutic response of patients, the samples were divided into complete remission/partial remission (CR/PR) samples (n = 18; including 14 basal-like subtype, 3 mesenchymal subtype, and 1 LAR subtype) and stable disease/progressive disease (SD/PD) samples (n = 23; including 16 basal-like subtype, 4 mesenchymal subtype, and 3 LAR subtype). The two categories of patients showed no significant differences in terms of molecular subtypes (analyzed by Chi-square test). The usage of clinical samples was ratified by the Institute Review Board of the First Hospital of China Medical University. All procedures involving human samples were conducted in compliance with the Declaration of Helsinki.

Immunohistochemistry (IHC)

Harvested tissue samples were fixed, paraffined, and cut into sections for IHC assay. According to the instruction of the kit (ab64261, Abcam Inc., Cambridge, MA, USA), the sections were successively dewaxed, deparaffined, rehydrated, treated with citric acid, blocked with H2O2 for 15 min, and treated with protein blockage reagent for 15 min. After that, the sections were probed by the antibodies against CEBPD (1:500, ab245214, Abcam) and VAMP3 (1:50, 10702–1-AP, Proteintech Group, Inc., Wuhan, Hubei, China) at 4℃ overnight, followed by incubation with biotinylated secondary antibody for 15 min and with streptavidin-peroxidase. After color development by DAB, the sections were counter-stained with hematoxylin. Additionally, the PD-L1 IHC 22C3 pharmDx kit (SKU: SK006, Agilent Technologies, Palo Alto, CA, USA) was utilized to assess the PD-L1 expression in tissue sections.

Protein expression in patient tissues was evaluated using IHC Profiler: An Open Source Plugin for the Quantitative Evaluation and Automated Scoring of Immunohistochemistry Images of Human Tissue Samples [22]. The final IHC score (0 ~ 3) of the sections was calculated as follows: percentage of high positive × 3 + percentage of positive × 2 + percentage of low positive × 1.

Cell culture

Human TNBC cells MDA-MB-231 (HTB-26) and MDA-MB-468 (HTB-132) and CD8+ cytotoxic T cells (PCS-800–017) were procured from ATCC (Manassas, VA, USA). Another TNBC cell line, MDA-MB-453 (BNCC340797), was acquired from BeNa Culture Collection (Beijing, China). The TNBC cells were cultured in L-15 medium along with 10% fetal bovine serum (FBS) in a 37℃ incubator, and the CD8+ T cells were recovered and cultured in Roswell Park Memorial Institute (RPMI)-1640 medium containing 10% FBS.

Construction of PTX-resistant TNBC cells

Approximately 2 × 106 TNBC cells were incubated in culture flasks overnight and treated with 3 nM PTX. When the confluence reached 80%, the cells (2 × 106) were digested, cultured in the flasks, and treated with PTX at higher concentrations. After five such repetitions, the PTX resistance of cells was evaluated by the CCK-8 assay. Those TNBC cells with an over twofold half maximal inhibitory concentration (IC50) of PTX than parental cells were designated as PTX-resistant cells.

Cell counting kit-8 (CCK-8) assay for viability detection

Approximately 5,000 TNBC cells were resuspended in 90 µL culture medium and seeded in 96-well plates. On the next day, PTX reagent (HY-B0015, MedChemExpress, Monmouth Junction, NJ, USA) at an initial concentration of 20 µM was prepared, which was repeatedly diluted at a twofold gradient, with seven varying concentrations obtained. The TNBC cells were treated with PTX reagent at different concentrations for 48 h, followed by addition of 10 µL CCK-8 reagent (ab228554, Abcam) to each well. After 1 h of dark incubation at 37℃, the optical density (OD) value at 460 nm was examined. Cells treated with phosphate-buffered saline (PBS) instead of PTX were set to controls, whose OD460 value was defined as 100%, according to which the viability of PTX-treated cells was evaluated.

Construction of cells with stable gene transfection

Short hairpin (sh) RNA of CEBPD (sh-CEBPD), overexpression plasmid of VAMP3 (oe-VAMP3), and the negative controls (NC; including sh-NC and oe-NC) used for lentivirus-based infection were procured from Genomeditech Co., Ltd. (Shanghai, China). In short, the TNBC cells were seeded in culture plates, and the lentivirus infection was performed when the confluence reached 60%. After 48 h, stably transfected cells were screened by the addition of corresponding antibiotics, and the successful gene expression alteration was confirmed by quantitative polymerase chain reaction (qPCR) and western blot (WB) analysis.

Colony formation assay

The TNBC cells were treated with PTX for 48 h and harvested. Approximately 1,000 cells were cultured in six-well plates at 37℃ for two weeks. Thereafter, the cells were fixed with paraformaldehyde and stained with crystal violet (G1062, Solarbio Science & Technology Co., Ltd., Beijing, China) for 15 min. The formed cell colonies were counted under a microscope.

Flow cytometry for cell apoptosis

According to the protocol of the Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) kit (V13242, Thermo Fisher Scientific, Rockford, IL, USA), approximately 1 × 106 PTX-treated TNBC cells were resuspended and seeded in 96-well plates, followed by dark incubation with FITC Annexin V and PI reagent for 15 min. Thereafter, the cells were washed and resuspended for apoptosis detection by a flow cytometer.

Transwell assay for cell mobility

Transwell plates with or without Matrigel precoating were used to analyze invasion and migration of the PTX-treated TNBC cells, respectively. Approximately 5 × 104 cells were resuspended in serum-free medium and seeded into the apical chambers, with serum-containing complete medium filled in basolateral chambers. After 24 h of incubation, the invasive or migratory TNBC cells were fixed, stained with crystal violet, and counted.

Wound healing assay

TNBC cells were cultured in complete medium (containing 10% FBS) until confluency. Using a sterile 200 mL pipette tip, equidistant scratches were made on the cell monolayer, followed by capturing images. Subsequently, the cells were cultured for an additional 48 h in serum-free medium containing PTX. The migration of cells was observed, documented, and the migration rate over 48 h was calculated.

RNA isolation and qPCR analysis

Total RNA from TNBC cells was isolated using the TRIzol reagent (Thermo Fisher Scientific) and used for cDNA synthesis using the PrimeScript RT Master Mix (RR036Q, Takara Holdings Inc., Kyoto, Japan). The qPCR analysis was performed following the protocol of TB Green® Premix Ex Taq II (RR820Q, Takara). Relative gene expression, with β-actin as the endogenous loading, was quantified by the 2−ΔΔCt method. Below are the primer sequences: CEBPD (forward): 5ʹ-TCCGGCAGTTCTTCAAGCAGCT-3ʹ, (reverse): 5ʹ-GAGGTATGGGTCGTTGCTGAGT-3ʹ; VAMP3 (forward): 5ʹ-GCTCTCTGAGTTAGACGACCGT-3ʹ, (reverse): 5ʹ-CCAGAACAGTAATCCCGATTGCC-3ʹ; β-actin (forward): 5ʹ-CACCATTGGCAATGAGCGGTTC-3ʹ, (reverse): 5ʹ-AGGTCTTTGCGGATGTCCACGT-3ʹ.

WB analysis

Total protein was isolated by RIPA buffer (Cat. No. 89900, Thermo Fisher Scientific), followed by protein concentration analysis using the bicinchoninic acid kit (Cat. No. 23225, Thermo Fisher Scientific). Total protein was separated by gel electrophoresis and wet-transferred to polyvinylidene fluoride membranes. The membranes were blocked by 5% non-fat milk, on which the proteins were probed by the antibodies of CEBPD (1:1,000, ab245214, Abcam), VAMP3 (1:500, 10702–1-AP, Proteintech Group, Inc., Wuhan, Hubei, China), and β-actin (1:1,000, #4970, Cell Signaling Technologies, Beverly, MA, USA) at 4℃ overnight, followed by incubation with biotinylated secondary antibody for 15 min. On the next day, the membranes were washed ad incubated with horseradish peroxidase-conjugated secondary antibody (1:5,000, ab6721, Abcam) at room temperature for 1 h. The blot bands were developed by enhanced chemiluminescence (D601039, Sangon Biotech Co., Ltd., Shanghai, China). Relative protein expression was then analyzed with β-actin as control as well.

Chromatin immunoprecipitation (ChIP)-qPCR

As instructed by the Simple ChIP Enzymatic Chromatin IP Kit (9003, Cell Signaling Technology), the PTX-resistant TNBC cells were treated with formaldehyde for 15 min. The protein-chromatin crosslinking was terminated by glycine. The chromatin was digested by micrococcal nuclease and resuspended in ChIP buffer. Thereafter, the CEBPD antibody (1:30, ab245214, Abcam) or rabbit immunoglobulin G was added for IP reaction at 4℃ overnight. The chromatin was eluted and de-crosslinked, and the DNA was purified for qPCR analysis.

Luciferase reporter assay

Binding site of CEBPD to the VAMP3 promoter was predicted from Jaspar (http://jaspar.genereg.net/). Wild-type (WT) VAMP3 or mutant-type (MUT) VAMP3 promoter sequence was inserted into pGL3-Basic luciferase reporter vectors. Human embryonic kidney (HEK) 293 T cells were seeded into 96-well plates. When the cell confluence reached 60%, they were transfected with sh-CEBPD along with the constructed reporter vectors using Lipofectamine 3000 (Thermo Fisher Scientific). After 48 h, the luciferase activity in cells was analyzed by the Dual Luciferase Reporter Gene Assay Kit (11402ES60, YEASEN Biotechnology Co. Ltd., Shanghai, China).

Immunofluorescence staining

Approximately 5,000 PTX-treated cells were seeded in 24-well plates and fixed. After permeabilization with Triton-X-100 for 15 min, the cells were blocked by 5% goat serum for 1 h and incubated with the LC3B antibody (1:2,000, #3868, Cell Signaling Technology) at 4℃ overnight, followed by incubation with Alexa Fluor 488-conjugated secondary (#2975, Cell Signaling Technology) at room temperature for 40 min. The staining was analyzed under fluorescence microscopy.

Isolation of extracellular vesicles (EVs)

Approximately 1 × 105 TNBC cells were seeded in culture flasks and incubated at 37℃ for 24 h. The culture supernatant was collected, in which the EVs were isolated by differential ultracentrifugation. In short, the supernatant was centrifuged at 800 × g for 10 min, at 2,000 × g for 30 min, and then at 16,000 × g for 1 h to isolate the EVs, which were resuspended in PBS for subsequent analyses.

Co-culture of CD8 + T cytotoxic cells and EVs

The CD8+ T cells were activated by CD3+ and CD28+ stimulation. After that, 10 μg TNBC-derived EVs was co-cultured with 2 × 106 activated CD8+ T cells for 48 h. The CD8+ T cells were then harvested, and the culture solution was centrifuged at 800 × g for 10 min to collect the supernatant for subsequent use.

Enzyme-linked immunosorbent assay (ELISA)

Expression of PD-L1 in the culture supernatant of TNBC cells and the secretory cytokines in the co-culture system of CD8+ T cells and EVs was analyzed by ELISA. In short, the collected cell culture supernatant was added with the standard of each cytokine to the detection kit, followed by the addition of detection reagents. The OD450 value was detected using a microplate reader. Standard curves were produced according to the OD450 value of standards to evaluate the levels of cytokines of interests. Following ELISA kits were applied: PD-L1 (ab277712, Abcam), interferon gamma (IFN-γ; DY285B, R&D Systems, Minneapolis, MN, USA), Granzyme B (BMS2027-2, Thermo Fisher Scientific), and Perforin (BMS2306, Thermo Fisher Scientific).

Xenograft mouse models

C57BL/6 J mice were purchased from Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China) and used in protocols approved by the Animal Ethics Committee of the First Hospital of China Medical University. The mice were randomly allocated into four groups, n = 5 in each. Approximately 5 × 106 MDA-MB-231/PTX cells were resuspended in 200 µL PBS and subcutaneously injected into mice. After one week, the growth of xenograft tumors was monitored. The major and minor axes of tumors were measured twice a week, on each Monday and Friday, respectively. The tumor volume (V) was calculated as follow: V = 1/2 × major axis × minor axis2. When the tumor volume reached 100 mm3, PTX treatment was applied at 20 mg/kg via intraperitoneal injection after every tumor volume measurement. Mice were euthanized by intraperitoneal injection of 1% pentobarbital sodium (150 mg/kg) on day 28, and tumor tissues were harvested and weighed.

Xenograft tumor treatment

The tumor tissues were cut up and then soaked in RPMI-1640 medium containing 0.5 mg/mL collagenase IV, 0.1 mg/mL DNase I, 10% FBS, 1% antibiotics for 30 min, followed by centrifugation at 500 × g for 20 min to separate lymphocytes. The CD8+ T cells were then sorted using the Dynabeads FlowComp mouse CD8 kit (11462D, Thermo Fisher Scientific).

Flow cytometry for T cell exhaustion analysis

The CD8+ cytotoxic T cells harvested from the co-cultured system and those from mouse xenograft tumors were subjected to flow cytometry to analyze T cell exhaustion. In short, the CD8+ T cells were resuspended in PBS and seeded in 96-well plates, followed by dark incubation with the antibodies of CD8a (1:100, 12–0081-82, Thermo Fisher Scientific) and Human Tim3 (1:20, 345022, Biolegend, San Diego, CA, USA)/Mouse Tim3 (1:100, MA5-17957, Thermo Fisher Scientific) for 30 min. Exhausted CD8+ T cells (Tim3+CD8+) were then screened using the flow cytometer.

Statistical analysis

All data were analyzed by the Prism 8.0 software (GraphPad, La Jolla, CA, USA) and presented by the mean ± SD. Differences were compared using t-test. When more than two groups were compared, one- or two-way analysis of variance (ANOVA) were used with post-hoc tests using Tukey's or Sidak's multiple comparisons test. P < 0.05 is indicative of statistical significance. IC50 of PTX to cells was analyzed by nonlinear fitting analysis.