PKC-ζ mediated reduction of the extracellular vesicles-associated TGF-β1 overcomes radiotherapy resistance in breast cancer

Female BALB/c (6–8 weeks old) were purchased from Vital River Laboratory Animal Technology (Beijing, China). Foxp3-GFP mice were kindly provided by Prof. Yangxin Fu (University of Texas, Southwestern Medical Center, Texas, USA). All animal experiments were performed according to the institutional ethical guidelines on animal care and the protocols used for this study were approved by the Animal Care and Use Committee at the Institute of Biophysics, Chinese Academy of Sciences. Murine breast cancer 4T1 cell line was obtained from ATCC and cultured in 5% CO₂ and maintained in RPMI 1640 medium supplemented with 10% FBS (VivaCell, Isreal) 100 U/ml penicillin, and 100 mg/ml streptomycin.

For in vitro experiment, 4T1 cells were cultured in RPMI 1640 media supplemented with 10% EV-depleted FBS. Supernatant was collected and centrifuged at 500 g for 10 min followed by a step of 3000 g for 20 min at 4 °C to pellet cells and debris. The supernatant was collected without disturbing the cell/debris pellet and was transfered to an ultracentrifuge tube. Then the supernatant was centrifuged at 100,000 g for 70 min at 10 °C and the EV pellets were collected. The pellets were resuspended in a small volume of PBS. For tumor tissues, the harvested tumors were dissected and cut into small pieces, followed by culture in RPMI 1640 media supplemented with 10% EV-depleted FBS for 48 h. Supernatant of tumor pieces was collected for the EVs purification. The EVs was characterized by Transmission Electron Macroscopy (FEI, Tecnai Spirit, 120 kV, USA) and quantified by BCA protein assay kit.

TGF-β1 in supernatant or EV was detected by ELISA (DY1679, R&D Systems, Minneapolis, MN). In brief, the 96-well microplate was coated with the Capture antibody overnight at 4 degree. Cells were washed by filling each well with Washing buffer. The plate was blocked by Block buffer for 1 h. TGF-β1 was activated by HCl and added to each well and incubate 2 h at room temperature. Cells were washed and the detection antibodies were added and incubated for 2 h. Streptavidin-HRP was added to each well. Washing wells and adding Substrate solution to each well were followed by adding Stop solution. The optical density of each well was determined immediately at 450 nm.

The gene mRNA expression matrix and clinical follow-up information of Breast Invasive Carcinoma patients with information of radiation therapy were obtained from the cBioPortal database (http://​www.​cbioportal.​org). The association between TGF-β1 expression and the overall survival was determined using the Kaplan–Meier survival analysis with the 'survival' package (version 4.1.2) in R statistics software, and the Log-rank test was used to detect significant differences. Immune cell infiltration was obtained from the ImmuCellAI database (http://​bioinfo.​life.​hust.​edu.​cn/​ImmuCellAI#!/) and the clinical information was also acquired from the cBioPortal database by using the patient ID.

Cells were lysed by RIPA lysis buffer and the protein concentration was determined by BCA protein assay. Cell lysates were separated by SDS-PAGE gels and transferred to nitrocellulose membranes. The membranes were then blotted with the indicated antibodies (anti-TGF-β1 antibody, ab179695; anti-PKCζ antibody, ab108970; anti-p-PKCζ antibody, ab76129; anti-β-actin antibody, ab8226. Abcam, Cambridge, UK).

Phorbol myristate acetate (PMA) and Calphostin C (CAL) were purchased from Merck and used at the indicated concentrations. 4T1 cells were treated by PMA or CAL for 48 h. Supernatant was collected and EV was purified.

gRNA targeting sequence were designed as the sequences described above using CRISPR design tool (https://​zlab.​bio/​guide-design-resources) and the sgRNA oligos with BbsI restriction site were prepared by annealing. The gRNA oligos were constructed into pX458M and EZ-Guide plasmids, respectively. Digest pX458M-gRNA1 and EZ-Guide-gRNA2 plasmid using XhoI and HindIII Restriction Enzyme (New England Biolabs, Beijing, China) and the two plasmids were ligated by T4 DNA ligase. pX458M-gRNA1 + gRNA2 plasmid were transformed into 4T1 cells by Lipofectamine 3000 Reagent (Thermo Fisher Scientific, Waltham, MA). After 48 h transfection, GFP + cells were sorting into 96-well plate by FACS Influx (BD, Franklin Lake, NJ) and cells were identified by qPCR for TGFB mRNA and ELISA for TGF-β1 protein.

Total RNA was isolated from cells by Trizol (Invitrogen, Carlsbad, CA). The mRNA was reversely transcribed to cDNA by M-MLV reverse transcriptase (Invitrogen, Carlsbad, CA). qPCR was performed using SYBR Green qPCR SuperMix (Invitrogen, Carlsbad, CA) to detect the expression of TGF-β1 and PKCs mRNA. Gene expression was normalized to GAPDH expression and presented as fold-change compared to the Control experiment.

Anti-CD3 antibody was coated into 96-well U plate overnight. Naïve splenocytes were added into each well. EVs were isolated from cells treated by different reagents and quantified by BCA protein assay. Different EVs were diluted by same times and added to naïve splenocytes with anti-CD28 antibody. After 72 h, cells were collected, stained with fluorescent antibodies and detected by Flow cytometry.

Immunohistochemical studies were done in 5-µm sections of paraffin-embedded tumor tissues using antibodies for TGF-β1 to determine its content in the tumors. The slides were incubated in citrate buffer for 20 min in a steamer and endogenous peroxidase was blocked by incubation with 3% H₂O₂ for 20 min at room temperature. The anti-TGFβ1 antibody (ab179695, Abcam, Cambridge, UK) was used in a dilution of 1:500. The slides were then stained with the secondary antibody in a dilution of 1:500. To determine the protein expression, stained slides were examined under fluorescence microscopy.

4T1 cells at 80% confluence were transfected with siRNA (sc-36254, Santa Cruz) by Lipofectamine 3000 Reagent (Thermo Fisher Scientific, Waltham, MA) for 24 h. Cells were treated by PMA or radiation for indicated time.

Spleen of GFP-Foxp3 transgenic BALB/c mice was harvested and single cell suspension was prepared. Cells were treated with anti-CD3ε/CD28 function antibody and EV or TGF-beta1 or 1D11 (BE0057, InVivoMab, West Lebanon, NH) for 72 h. Percentage of CD4⁺GFP⁺ cells in CD4⁺ cells was detected by BD FACSCalibur.

The following putative structures of mouse PKCs were downloaded from the AlphaFold Protein Structure Database (https://​alphafold.​ebi.​ac.​uk/​ and used in our study: KPCA (UniProtKB: P20444), KPCB (UniProtKB: P68404), KPCD (UniProtKB: P28867), KPCE (UniProtKB: P16054), KPCG (UniProtKB: P63318), KPCI (UniProtKB: Q62074), KPCL (UniProtKB: P23298), KPCT (UniProtKB: Q02111), KPCZ (UniProtKB: Q02956). Geometrical alignments, as well as visualization, were performed with PyMOL version 2.1.0.

Before radiation administration, 4T1 cells were treated with Naringenin of 200 uM concentration for 30 min. X-Ray of 2, 4 or 8 Gy dose was used to treat 4T1 cells, respectively. After 2 h, cells were collected and washed with PBS 3 times. TSQ was dissolved in Lock’s buffer (pH 7.4) and cells were stained with 150 nM TSQ for 1 min. After 3 times washing, cells were added into 96-well plate and were examined under a fluorescence microplate reader (Excitation, 345 nm; Emission 495 nm) (SpectraMax M4 Multi-Mode Microplate Reader, Molecular Devices, San Jose, CA). The instrument measures the intensity of the reradiated light and expresses the result in Relative Fluorescence Units (RFU) using SoftMax Pro Software (Standard Edition 7.1).

4T1 cells were subcutaneously injected at 5 × 10⁴ cells per mouse. Mice were randomized to treatment groups. Radiotherapy was administrated with a dose of 10 Gy when tumor volume reached 70–80 mm³. 100 mg/kg naringenin was administrated daily by intragastric administration for 30 days. 5 mg/kg 1D11 was injected intraperitoneally 3 times per week for 3 weeks. Tumor volumes were measured twice a week and calculated as length*width*height/2. The survival days of tumor-bearing mice were recorded. All animal experiments were performed according to the institutional ethical guide lines on animal care and the protocols used for this study were approved by the Animal Care and Use Committee at the Institute of Biophysics, Chinese Academy of Sciences.

Single-cell suspensions were prepared. Samples were stained (20–30 min) with the following antibodies: anti-CD45, anti-CD3, anti-CD4, anti-CD8α, anti-CD25 antibodies. For intracellular staining, cells were fixed, permeabilized overnight at 4 °C (Fixation/Permeabilization Concentrate and Diluent kit, eBioscience, San Diego. CA) and subsequently stained using anti-Foxp3 antibody for 30 min. All experiments were performed on BD FACSCalibur or BD LSRFortessa and data was analyzed with FlowJo 7.6.1.

As RT has been reported to cause the increases of the intratumoral TGF-β1 level and the infiltration of Tregs in the TME [17, 18], TGF-β1 therefore is considered to be an endogenous factor for RT resistance. To investigate how much TGF-β1 is secreted in the form associating with EVs in clinical settings, the EVs of tumor samples were isolated from six NSCLC patients, respectively. As expected, up to 72% of TGF-β1 was secreted in the EV-associated form in these tumors from the NSCLC patients (Fig. 1A).

Given the clinical data indicate that the TGF-β1 level is increased in the irradiated tumors [8, 27], we then wondered if the increased TGF-β1 could augment Tregs infiltration in tumors after irradiation. In line with the previous finding, the irradiation was found to promote more Tregs infiltration in 4T1 tumors than the untreated control (Fig. 1B, C). Curiously, compared to RT alone, the combination of TGF-β1 neutralizing antibody 1D11 only partially suppressed radiation-induced Tregs infiltration (Fig. 1C). A delayed tumor growth was observed after RT and 1D11 combination treatment (Fig. 1D), while the survival benefit was comparable to the mice received RT alone (Fig. 1E, F). These results suggested that 1D11 did not effectively block the function of the TGF-β1_EV to improve the effect of radiotherapy.

In order to verify the functions of the TGF-β1_EV secreted by breast cancer cells, we isolated EVs from 4T1 cells and characterized. The results showed that the average diameter of the EVs was about 40 nm (Fig. 2A), and approximately 82% of TGF-β1 secreted by 4T1 cells was associated with the EVs (Fig. 2B). Consistent with the finding in 4T1 cells, up to 80% of the TGF-β1 was secreted with EVs from 4T1 tumor tissues of balb/c mice (Fig. 2C). Notably, we found that the TGF-β1_EV was mainly in the form of latent-TGF-β1, while the TGF-β1 left in the EV-free supernatant derived from 4T1 cells was mainly in the form of active-TGF-β1 (Fig. 2D). The latent TGF-β1_EV transmits signals after endocytosis [13, 28], which probably explained why 1D11 or other TGF-β1 inhibitors to bind ligands or receptors could not effectively block the function of the TGF-β1_EV.

To further determine if the TGF-β1_EV affect T cells differentiation, we generated the TGF-β1 knockout (KO) 4T1 cell line (Fig. 2E, F). A significant increase of Tregs population in response to the EVs from wild type (WT) 4T1 cells were observed. The TGF-β1 knockout impeded the ability of the EVs to induce Tregs population (Fig. 2G). However, 1D11 only had a partial effect on the TGF-β1_EV-induced Tregs infiltration (Fig. 2I). We further demonstrated that the EVs from 4T1 cells directly promoted the differentiation of splenic CD4⁺ T cells into Tregs (Additional file 1: Fig. S1A). In addition, the TGF-β1_EV did not show any effects on the proliferation and apoptosis of Treg cells (Additional file 1: Fig. S1B, C). Interestingly, the PHK67-labeled TGF-β1_EV was rapidly engulfed by CD4⁺CD25⁺ lymphocytes and resulted in the intracellular accumulation (Fig. 2H). These data suggested that the tumor-derived EVs containing TGF-β1 is capable of inducing CD4⁺ T cells differentiation into Tregs, which cannot be effectively blocked by the TGF-β1 neutralizing antibody.

To test if the TGF-β1_EV contributes to Tregs induced radiotherapy resistance in the TME, we employed a 4T1 breast cancer model treated by irradiation (Fig. 3A). In line with the previous reports [18], the area of TGF-β1 positive region in IHC slides was increased from 6.4% (on day 16) to 26.8% (on day 28) after tumor cell injection (Fig. 3B, C). Notably, irradiation further enlarged the area of TGF-β1 positive region from 25% (on day 2) to 45% (on day 14) after RT (Fig. 3B, C). However, the TGF-β1 concentration in peripheral blood was not changed regardless of whether the mice were irradiated (Fig. 3D, E). The irradiation-induced TGF-β1 in tumor was supposed to be also mainly in a form associated with EVs. Indeed, more than 65% of the TGF-β1 was on the EVs in both 4T1 cells and tumor tissues after irradiation (Additional file 1: Fig. S2A, B). Furthermore, compared with the control group, the irradiation promoted 4T1 tumors to release more TGF-β1_EV (Fig. 3F), thus resulting in more Tregs infiltration (Fig. 3G). To verify if the TGF-β1_EV tends to retain in TME, we intratumorally injected the fluorescent-labeled TGF-β1_EV and the free TGF-β1 into 4T1 tumor, respectively. The results demonstrated that the free TGF-β1 diffused away from tumor tissue much faster than the TGF-β1_EV (Fig. 3H). Compared with 1 h post-injection, only 53% of free TGF-β1 remained after 24 h of injection, meanwhile, much higher fluorescent intensity (~ 74%) was retained in tumors after the TGF-β1_EV injection (Fig. 3I). These results suggested that the retained TGF-β1_EV in TME highly possibly induced immunosuppression through promoting the differentiation of CD4⁺ T cells to Tregs.

As PKCs have been reported to be involved in the secretion of EVs, to understand the underlying mechanism of the TGF-β1_EV secretion, we focused on PKCs. As expected, the pan-PKCs agonist PMA promoted the TGF-β1_EV secretion from 4T1 cells. Conversely, the inhibition of PKCs by Calphostin C (CAL) showed an opposite effect (Fig. 4 A, B).

Radiation has been shown to promote the TGF-β1_EV secretion from tumor tissues (Fig. 3F). To identify the subtype of PKCs regulating the TGF-β1_EV secretion, we compared the mRNA expression of various PKC subtypes before and after irradiation. We found that, even after receiving a low dose of radiation as 2 Gy, the mRNA expression of PKC-ζ was elevated more significantly than that of other PKC subtypes (Fig. 4C), suggesting that PKC-ζ could effectively regulates the TGF-β1_EV secretion. We then used the siRNA of PKC-ζ, which was verified for its effectiveness (Additional file 1: Fig. S3), to treat 4T1 cells before treatment of PMA or radiation. The results showed that the suppression of PKC-ζ by siRNA significantly inhibited the TGF-β1_EV secretion induced by PMA (Fig. 4D, E). As expected, the secretions of EVs and TGF-β1_EV from 4T1 cells (Fig. 4F, G) and the phosphorylation level of PKC-ζ in 4T1 cells (Fig. 4J) were dose-dependently enhanced after irradiation. Accordingly, the PKC-ζ siRNA significantly reduced the TGF-β1_EV release induced by irradiation (Fig. 4H, I). These results indicated that PKC-ζ could play an essential role in the TGF-β1_EV secretion.

We have previously demonstrated that a natural flavonoid, naringenin, can decrease total TGF-β1 secretion through regulation of PKCs [22]. Deservedly, naringenin might reduce the secretion of TGF-β1_EV via inhibiting PKC-ζ phosphorylation. Here, our results showed that naringenin had a dose-dependent effect on reducing the level of phosphorylation of PKC-ζ stimulated by irradiation (Fig. 5A) but had no effects on the level of PKC-ζ mRNA expression (Fig. 5B) in 4T1 cells. Furthermore, naringenin markedly attenuated the secretions of the EVs and the TGF-β1_EV promoted by radiation (Fig. 5C, D), thus in turn resulting in the prevention of the TGF-β1_EV-induced Tregs differentiation from CD4⁺ T cells (Fig. 5E). In addition, the in vivo data further demonstrated that naringenin reduced the TGF-β1_EV secretion from the irradiated 4T1 tumors (Fig. 5F), indicating that naringenin could be used as an inhibitor of PKC-ζ phosphorylation to combine with radiotherapy as a promising strategy for breast cancer treatment.

To further investigate the regulatory mechanism of naringenin on the TGF-β1_EV release when it combined with RT, we need to answer what caused the elevation of PKC-ζ phosphorylation after radiation. It has been reported that PKC can be activated by the exposure to oxidants [29]. In rat brain, the phosphorylated activation of PKC has been thought via oxidation of thiols and release of zinc from the cysteine-rich region [30]. We then assumed that radiation could promote superoxide production, which stimulated PKC-ζ activity via the release of zinc from zinc finger domain of PKC-ζ in breast cancer cells. When the sequences of PKCs were aligned, it showed that, unlike other PKCs, the atypical PKCs (ζ and ι) have only one single zinc finger motif of the cysteine-rich region in the conserved C1 domain (Fig. 6A). The atypical PKCs also has the smallest number of cysteines in zinc finger region among all PKC isoforms (Fig. 6B). However, the two zinc finger motifs in either classical or novel PKCs are tightly bound to each other leading the cysteine residues are partially hidden within the cleft of the binding interface, but the individual cysteine-rich region for the atypical PKCs is highly isolated from other residues (Additional file 1: Fig. S4A, B). Considering that the superoxide has a very short half-life to stimulate the autonomous PKC activity [24], we assume that the atypical PKCs including the ζ and ι isoforms are susceptible for superoxide-induced thiol oxidation and following activity increase. The mRNA relative expression of PRKCZ was significantly higher than PRKCI in breast cancer cells, including 4T1 cells (Fig. 6C) and MDA-MB-231 cells (Additional file 1: Fig. S5A), which indicating PKC-ζ may be activated by superoxide more easily than PKC-ι.

Radiation is reported to promote TGF-β1 activation through reactive oxygen species (ROS), which is mostly composed of superoxide radicals (O2˙−) [31]. Here we demonstrated that irradiation induced a time-dependent increasing level of O2˙- (DHE staining) (Fig. 6D), a dose-dependent increase of zinc release (Fig. 6F), and an elevated level of PKC-ζ phosphorylation (Fig. 4J), suggesting a positive correlation among O2˙−, zinc release and PKC-ζ phosphorylation. When 4T1 cells were treated with paraquat (PQ), one of the superoxide generators, a marked elevation of O2˙− was observed. Similar to superoxide dismutase (SOD), naringenin reduced the level of O2˙− induced by irradiation or PQ to a relatively low level in a dose dependent manner (Fig. 6D, E). By reducing of the elevated level of O2˙−, naringenin inhibited zinc release (Fig. 6F, Additional file 1: Fig. S5B) and reduced PKC-ζ phosphorylation level significantly (Fig. 6G, H), demonstrating that naringenin decreases the TGF-β1_EV release via inhibition of the superoxide-Zinc-PKC-ζ-TGF-β1_EV pathway (Fig. 6I). As one of the top transcription factors for PRKCZ gene expression (PROMO analysis), NF-kB showed a radiation induced nucleus translocation, which was modulated by naringenin (Additional file 1: Fig. S6). The results suggested that RT induced PKC-ζ activation and then feedback stimulated PRKCZ mRNA expression via NF-kB nuclear binding.

To evaluate the in vivo therapeutic effects of naringenin combined with RT, the 4T1 tumor model was used (Fig. 7A). Compared with RT alone, RT combined with naringenin significantly decreased radiation-induced Tregs infiltration, resulting in a higher ratio of CD8⁺/Treg (Fig. 7B, C). The reduction of suppressive Treg cells in the TME by naringenin could bring an inhibition of tumor weights (Fig. 7D) and a prolonged survival of mice bearing breast tumors (Fig. 7E, F). When compared to 1D11 antibody, naringenin combined with RT was more effective in delaying the tumor growth (Fig. 7G). After RT, the body weights of mice decreased dramatically. The combination of naringenin could restore somewhat of mice weight on the 19 day, 24 day and 37 day after tumor inoculation (Fig. 7H). Specifically, the weights of three mice among ten were lower than 15 g after the mice received RT combining with 1D11 treatments. In contrast, naringenin quickly recovered the mice weight loss caused by radiation and no mouse weight was lower than 15 g (Fig. 7I). A much shorter time was needed to recover for the mice treated by RT combined with naringenin than that with 1D11 (Fig. 7J). Therefore, the inhibition of PKC-ζ to prevent the TGF-β1_EV secretion could be an optional combination strategy to overcome the radiotherapy resistance.