Boosting immune surveillance by low-dose PI3K inhibitor facilitates early intervention of breast cancer

Abstract

Prevention of estrogen receptor-negative (ER-) breast cancer is an unmet challenge, although tamoxifen and aromatase inhibitors can successfully decrease the incidence of ER-positive (ER+) breast cancer. PI3K pathway activation has been detected in tamoxifen-resistant ER- breast lesions of patients. Here, we further ratified that the PI3K pathway is significantly activated in premalignant ER- breast lesions compared with paired normal tissues of patients, which prompted our assessment of targeting PI3K on inhibition of ER- mammary tumor initiation and progression. Both genetic knockdown of PIK3CA or intervention with low-doses of a PI3K inhibitor (GDC-0941) prevented the dysplasia phenotype of semi-transformed human ER- mammary epithelial cells in 3-dimensional culture in vitro. Importantly, low-dose GDC-0941 treatment significantly delayed mammary tumor initiation in the MMTV-neu mouse model without exhibiting discernable adverse effects. Interestingly, increased CD8⁺/GZMB⁺ T-cells were detected in mammary tissue after GDC-0941 treatment, suggesting enhanced immune surveillance. Mechanistically, elevated expression of potent T-cell chemo-attractants, including CCL5 and CXCL10, were detected both in vitro and in vivo after GDC-0941 treatment. Furthermore, inhibition of PI3K significantly increased T-cell recruitment in a CCL5/CXCL10-dependent manner. In human ER- breast cancer, PI3K activation is correlated with significantly reduced CCL5, CXCL10 and CD8A expression, suggesting that the decreased CD8⁺ T-cell recruitment and escape of immune surveillance may contribute to ER- breast cancer development. In summary, our study indicates that low-dose PI3K inhibitor treatment may intervene early stage ER- breast cancer development by enhancing immune surveillance via CCL5/CXCL10.

Introduction

Elucidating the molecular mechanisms involved in early stage breast cancer development can facilitate early detection and prevention, which are critical for reducing the morbidity and mortality of breast cancer. Large-scale clinical trials on breast cancer prevention showed that selective estrogen receptor modulators (e.g., tamoxifen) and aromatase inhibitors could reduce incidence of estrogen receptor-positive (ER+) breast tumor in high-risk populations by approximately 50% in various age groups [1,2]. However, no effective agents are available for prevention of estrogen receptor negative (ER-) breast cancer [3-5], which has poorer clinical outcomes compared with ER+ breast cancer. Therefore, there is an urgent need to develop prevention and early intervention strategies for ER- breast cancer.

The phosphatidylinositol-3-kinase (PI3K)/Akt intracellular signaling pathway is involved in various cellular functions, e.g., proliferation, apoptosis, and metabolism, which play critical roles in cancer [6]. In many cancers, this pathway is overactive, leading to reduced apoptosis and accelerated proliferation [7]. Furthermore, several studies have shown that the PI3K/Akt pathway is frequently upregulated in breast cancer [8-10]. PI3K/Akt activation is strongly associated with poor prognosis in patients with HER2+/ER- breast cancer [11-13]. In our previous study, profiling of key signaling proteins and pathways by reverse phase protein array uncovered high levels of PI3K/Akt activation in breast tissue samples from women with atypical hyperplasia (ADH) and ductal carcinoma in situ (DCIS) compared with normal tissue, and we detected the PI3K/Akt signaling pathway activation in the early stage of breast disease [14]. These findings suggest that PI3K/Akt activation in mammary atypia could drive initiation and progression of ER- breast cancer. Several PI3K inhibitors-including GDC-0941-have been clinically tested for cancer treatment with tolerable toxicity [15]. We postulate that low doses of PI3K/Akt-targeting agents may be applicable for safe and effective prevention of ER- breast cancer.

Cancer immune surveillance refers to the ability of immune cells to recognize and eliminate premalignant and malignant cells before they can cause harm, thus immune surveillance is critical in cancer prevention. Trafficking of immune cells in tumors is mediated by chemokines, which are also altered during tumor initiation and progression [16]. C-X-C motif chemokines, including CXCL9, CXCL10, and CXCL11, as well as CC chemokines such as CCL3, CCL4, and CCL5, show tight correlations with T-cell infiltration in the tumor microenvironment (TME) and are associated with better clinical outcomes in some cancer types [17]. Some studies have shown a relationship between PI3K/Akt pathway activation in cancer cells and suppression of T-cells in the TME [18,19]. Therefore, we envisage that targeting the PI3K/Akt pathway using low-dose PI3K inhibitor might also influence T-cell mediated immune surveillance that prevent early stage ER- breast cancer development. The purpose of this study was to determine the effect of low doses of a PI3K inhibitor on chemoprevention of ER- breast cancer.

Materials and methods

Cell lines and cell culture

MCF10A cell line was purchased from American Type Culture Collection (ATCC) and cultured in DMEM/F12 (Caisson No. DFL13) supplied with 5% horse serum (Thermo Fisher Scientific, 16050122), 20 ng/mL EGF, 100 ng/mL cholera toxin, 0.5 μg/mL hydrocortisone, 10 μg/mL insulin, 50 units/mL penicillin, and 50 µg/mL streptomycin as previously described [2]. McNeuA and N202 cell lines were established from a mammary tumor originating from a female MMTV-neu transgenic mouse and cultured in DMEM with 4.5 g/L glucose (Corning No. 10-013-CV) supplied with 10% FBS (Thermo Fisher Scientific, SH3007103) [20]. HCC-1569 cell line was purchased from ATCC and cultured in RPMI-1640 (Caisson No. RPL09) supplied with 10% FBS. All cell lines had been tested for mycoplasma contamination.

RNA interference and stable cell line generation

The shRNAs for PIK3CA (V3LHS-364628, V3LHS-364671) were obtained from MD Anderson Cancer Center’s shRNA core facility. The sequence for clone V3LHS-364628 and V3LHS-364671 were 5’-TCTTGAGTAACACTTACGA-3’ and 5’-TTACCACACTGCTGAACCA-3’, respectively. The shRNA gene knockdown lentiviral vectors and its packaging plasmid (psPAX2) as well as envelope plasmid (pMD2G) were transfected into 293T cells together using the LipoD293 reagent (SignaGen, No. SL100668) according to the manufacturer’s instructions. After 48 h, the lentivirus-containing media was collected, filtered with a 0.45-μm filter, and used to infect MCF10A cells in the presence of 10 μg/mL polybrene for 24 h. The infected cells were selected by 2 μg/mL puromycin (InvivoGen, 58-58-2) for 5 days based on the drug selection genes of the introduced vectors.

Cell proliferation assay

Cells were seeded at 1000 cells/well in sextuplicate in 96-well cell culture plates. 20 μL of 5 mg/mL MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Thermo Fisher Scientific, M6496) in PBS were added into each well and incubated for 2 h at 37°C in the dark. After 2 h of incubation, medium with MTT was removed and replaced with 100 μL of DMSO and vibrated for 5 min on a shaking table to solubilize the intracellular purple formazan, which was analyzed at the absorbance of 570 nm and 620 nm with a microtiter plate reader (BioTek).

Cell apoptosis assay

Cell apoptosis was detected using FITC Annexin V apoptosis detection kit (BD Biosciences, No. 556547). Briefly, cells were treated with vehicle or GDC-0941 (0.3 μM or 3 μM) overnight, washed with cold PBS and resuspended in 1X Binding buffer. 5 μL of FITC Annexin V and 5 μL of propidium iodide (PI) were added in each sample, incubated for 15 min at room temperature in the dark, and then analyzed by Flow Cytometry. Early (Annexin V⁺, PI^-) and late apoptotic cells (Annexin V⁺, PI⁺) were counted as total cells that underwent apoptosis.

Western blotting analysis

Western blotting was performed as previously described [21]. The following Primary antibodies were used: ErbB2 (Cell Signaling, No. 4290), p110α (Cell Signaling, No. 4255), phospho-Akt (Cell Signaling, No. 4060S), (pan) Akt (Cell Signaling, No. 4691), β-actin (Sigma, No. AC-15). The signal was detected by ECL (Amersham) following the manufacturer’s instructions. Images shown in figures are representative of 3 independent experiments.

Three-dimensional cell culture assay

Three dimensional (3D) culture assay was performed following the protocol as previously described [22]. Cells were seeded at 1000 cells/well in culture slides (BD Falcon, No. 354108) and cultured in medium containing 2% Matrigel (BD Biosciences), and the medium was replaced every 3 days. At day 4, the 10A.B2 cells were treated with either vehicle or GDC-0941 (LC Laboratories) for another 5 days. Phase-contrast images were captured at day 9 and day 14. At day 9, acini were stained with anti-cleaved caspase-3 (Cell Signaling, No. D175), laminin 5 (Millipore, No. MAB1947), and DAPI (Abcam, No. ab104139). At day 14, acini were stained with anti-Ki-67, laminin 5, and DAPI. The images were taken at 60× magnification using a Zeiss confocal microscope. ImageJ software was used to measure acinar size.

Immunofluorescence staining of cells

Immunofluorescence staining on cells in 3D culture was performed as previously described [14]. 3D cultures were fixed with 4% PFA, permeabilized with 0.3% Triton X-100, washed 3 times with PBS/glycine buffer, and blocked with IF buffer (10% goat serum, 0.2% Triton X-100, and 0.05% Tween 20) for 1 h. Antibodies were diluted in IF buffer and incubated overnight at 4°C followed by incubation with Goat Anti-Rabbit 488 and Alexa Fluor Goat Anti-Mouse 594 (Invitrogen, No. O-11038, A-11005, 1:300 dilution) at 37°C for 1 h. Samples were covered with anti-fade mounting medium with DAPI (Abcam) and analyzed with a Zeiss LSM 880 laser scanning confocal microscope.

Animal experiments

All animal experiments were carried out in accordance with protocols approved by the Institutional Animal Care and Use Committee of The University of Texas MD Anderson Cancer Center. Female MMTV-neu mice at 10 weeks of age were treated with GDC-0941 at 50 mg/kg (n=29) or vehicle (n=29, 0.5% hydroxypropyl methylcellulose with 0.1% Tween 80) by oral gavage once daily until tumors were palpable. The median tumor-free survival time was referred as T (50). 8-, 10- and 15-weeks after treatment, 4-5 mice were euthanized in each group and the fourth pair of normal-looking mammary fat pads (MFPs), bones, and bone marrow were isolated. All sample tissues were collected for histological analyses, flow cytometry, and RT-PCR. FVB mice were treated with oral gavage of vehicle (n=3) or GDC-0941 at 20 mg/kg (n=3) and 100 mg/kg (n=4) for two weeks, and then examined toxicity of GDC-0941.

RNA extraction and RT-PCR

Total RNA of MFP cells was extracted using Trizol LS (Molecular Research Center, No. TS120) according to the manufacturer’s protocol. Total RNA was isolated using Trizol reagent (Molecular Research Center, No. TR118) from the tissue samples. RNA was reverse transcribed into cDNA using iScript cDNA Synthesis Kit (Bio-Rad, No. 1708891). Real-time PCR was performed in 96-well plates on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad) using iTaq Universal SYBR Green Supermix (Bio-Rad, No. 1725121). Targeted mRNA expression in each sample was normalized to endogenous control (18S). The relative mRNA expression was quantified by 2-ΔΔCt method. Each experiment was repeated at least 3 times. The following primers were used to detect corresponding mRNAs in the RT-PCR analyses: CCL5 FQP, 5’-TTTGCCTACCTCTCCCTCG-3’; CCL5 RQP, 5’-CGACTGCAAGATTGGAGCACT-3’; CXCL10 FQP, 5’-CCAAGTGCTGCCGTCATTTTC-3’; CXCL10 RQP, 5’-GGCTCGCAGGGATGATTTCAA-3’; 18S FQP, 5’-GTAACCCGTTGAACCCCATT-3’; and 18S RQP, 5’-CCATCCAATCGGTAGTAGCG-3’.

Immunohistochemistry and immunofluorescent staining of tissue

Immunohistochemical staining was conducted as previously described [23]. The slides were incubated at 4°C overnight with the following primary antibodies: phospho-Akt (Cell Signaling, No. 4060S, 1:200), Ki-67 (Abcam, No. ab15580, 1:1000), CD45 (BioLegend, No. 103102, 1:100), CD3 (Abcam, No. ab16669, 1:100), CD8 (Abcam, No. ab217344, 1:100), and GZMB (Abcam, No. ab4059, 1:100). Positive control and negative control slides were included in each staining. The stained MFP samples were scored using the immunoreactive score according to multiplication of staining intensity scores and percentage of positive cell scores. Immunohistochemical staining and quantification were performed in a double-blind manner. For immunofluorescent staining, slides were incubated with Alexa Fluor secondary antibodies (Invitrogen) and analyzed by Zeiss LSM 880 laser scanning confocal microscope.

Human lymphocyte generation

Human leukocyte concentrate (buffy coat) was collected from healthy blood donor volunteers by MD Anderson Blood Donor Center. Written informed consent was obtained from all participating blood donors and the use of anonymized leftover specimens for scientific purposes was approved by the Ethics Committee of the MD Anderson Cancer Center. For the generation of lymphocytes, peripheral blood of healthy individuals was applied to a Ficoll gradient (BD), lymphocytes were collected and cultured in RPMI-1640 culture medium (10% FBS, 2 mM glutamine, 1% sodium pyruvate, 1% HEPES, 1% MEM non-essential amino acids, 50 μM 2-mercaptoethanol).

T-cell recruitment assay

T-cell recruitment assays were performed as previously described [24] with minor modifications. Briefly, splenocytes of FVB mice were collected and were activated with CD3/CD28 (BioLegend, No. 100238, 102102) for 24 h. N202 cells pretreated with either vehicle or 0.3 μM GDC-0941 for 48 h were seeded onto 24-well plates, and activated splenocytes were then added on 3-μm hanging cell culture inserts (Millicell, No. MCMP24H48) placed in a 24-well plate. After overnight incubation, splenocytes that migrated to the 24 wells were collected, counted, and stained with fixable viability dye (Invitrogen, No. 65-0863-14) followed by APC/Cyanine7 anti-mouse CD45, FITC anti-CD3, PerCP/Cyanine5.5 anti-CD8, PE anti-CD4, and APC anti-IFN-γ (BioLegend, No. 103116, 100204, 100734, 100408, 505810) antibodies and analyzed by FACS Canto II flow cytometer. For chemokine blockade, anti-CCL5, anti-CXCL10 antibodies and IgG control (R&D systems, No. AF478, AF466, AF007) were added after GDC-0941 treatment in N202 cells, and T-cell recruitment assay was performed as stated above.

Terminal deoxynucleotidyl transferase dUTP nick end labeling staining

Paraffin-embedded tissue slices were dewaxed, rehydrated, fixed, washed with PBS and permeabilized with 20 μg/ml Proteinase K solution at room temperature. After washing with PBS, the slides were dipped in a terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) reaction mixture (Promega, No. G7360) for 60 min at 37°C in a humidified chamber. The reaction was stopped with SSC, washed with PBS, and blocked with 0.3% hydrogen peroxide for 3 min. Subsequently, the slides were incubated with 100 μl streptavidin HRP for 30 min at room temperature. The slides were stained with DAB and visualized with a light microscope.

Bioinformatics analysis

A 32-gene transitional signature of the PI3K/Akt/mTOR pathway was adapted from previous publication [25]. Gene expression data of CCL5, CXCL10, CD8A, and the 32-genes were obtained from different sources accordingly. Gene expression data of matched breast tissues, DCIS, and IDC is available from GSE14548 (https://www.ncbi.nlm.nih.gov/geo/query). The data of atypical ductal breast hyperplasia, DCIS, and IDC can be download from supporting information from the publication [26]. The TCGA ER- breast cancer gene expression data (175 cases) is downloaded from TCGA data portal (https://tcga-data.nci.nih.gov/tcga/dataAccessMatrix.htm). Another ER- breast cancer gene expression data from 123 patients is obtained from GSE20685 [27]. The correlations between CCL5 vs CD8A and CXCL10 vs CD8A were measured using Pearson correlation coefficient r. The 32-gene signature of PI3K pathway was calculated using Z-score of the 32 genes. The ER- patients were divided to PI3K-score high (top one third of the highest PI3K signature score) and PI3K-score low (bottom one third of the lowest PI3K signature score) based on the 32-gene signature.

Statistical analysis

Statistics were analyzed either by 1-way ANOVA or Student t test where applicable. Tumor-free survival analyses were performed using the Kaplan-Meier method and Wilcoxon test. Statistical analysis was performed using SPSS (16.0; SPSS, Inc.) and GraphPad Prism (Prism 8.0; GraphPad Software Inc.) packages. Error bars represent means ± SD. A P value < 0.05 was considered significant. All quantitative experiments were repeated in at least 3 independent tests.

Results

Blocking PIK3CA prevents the dysplasia phenotype of 10A.B2 mammary epithelial cells

Aberrant PI3K/Akt activation has been found in more than 27% of breast cancers [28], including HER2+ breast cancer. Bioinformatic analysis of early stage breast cancer reveals a significantly increased PI3K activation in ductal carcinoma in situ (DCIS) compared with patient-matched normal breast tissue (Figures 1A, S1) [14]. Atypical hyperplasia lesions showed similarly increased PI3K activation as found in DCIS and invasive ductal carcinoma (IDC) stage (Figure 1B), indicating PI3K activation is a very early event in breast cancer initiation. To determine whether PI3K pathway activation contributes to breast cancer initiation, we knocked down PIK3CA using shRNA (shPIK3CA) in the semi-transformed 10A.B2 cell line, a subline of MCF10A human mammary epithelial cells (MEC) that stably overexpress exogenous ErbB2 gene and recapitulate the dysplasia phenotypes of DCIS in three dimensional (3D) cell cultures [22]. Compared with control vector transfected MCF10A (10A.vec) cells, ErbB2-overexpression in 10A.B2 cells led to a dramatic increase of p110α and p-Akt-S473 levels, whereas knocking down of PIK3CA in 10A.B2 cells clearly reduced the levels of p110α and p-Akt-S473 in both 2D and 3D cell cultures (Figure 1C). Knocking down PIK3CA in 10A.B2 cells significantly inhibited cell growth compared with 10A.B2 shctrl cells in 2D culture (P < 0.001; Figure 1D). In 3D cell culture, the 10A.vec cells formed smooth and polarized (laminin 5) spherical acinar structures with a hollow lumen that mimicked normal mammary glands in vivo (Figure 1E, first column from left), while 10A.B2 cells grew into large disorganized acinar structures with a filled lumen due to reduced apoptosis (cleaved caspase 3) and increased proliferation (Ki-67; Figure 1E, second column from left). Knocking down PIK3CA in 10A.B2 cells restored the spherical acini structure with a hollow lumen, similar to that of the 10A.vec cells (Figure 1E, first and second column from right). Immunofluorescent staining of the acini showed insignificant increases in apoptosis (cleaved caspase-3) with PIK3CA knockdown despite restoration of hollow lumens, whereas proliferation (Ki-67) was markedly reduced in PIK3CA-knockdown 10A.B2 cells (Figure 1E). Our further analyses indicated that knocking down PIK3CA not only significantly reduced the percentage of transformed acini but also decreased acini sizes (Figure 1F, 1G).

Figure 1.

Activation of PI3K signaling in early stage breast cancer and knockdown of PIK3CA reduced ER- MEC transformation. (A) PI3K activation signature in matched breast tissue and ductal carcinoma in situ (DCIS) (PMID: 19187537). (B) Comparison of PI3K activation in atypical ductal hyperplasia, DCIS, and IDC in Ma et al breast cancer dataset (PMID: 12714683). (C) Western blotting of ErbB2, p110α, p-Akt, and pan Akt in 10A.vec and 10A.B2 cells, and shctrl and shPIK3CA clones of 10A.B2 cells in 2D and 3D cell culture. (D) Cell growth curve of 10A.B2 cells and shctrl and shPIK3CA clones of 10A.B2 cells. (E) Phase-contrast microscopy images of 10A.vec and 10A.B2 cells and shctrl and shPIK3CA clones of 10A.B2 cells in 3D culture. IF images showing cleaved caspase-3, Ki-67, phospho-Akt, laminin 5, and DAPI staining in acini. Magnification was 60×. (F, G) Quantification of percentage of round and transformed acinar phenotypes (E) and quantification of relative (rel.) acinar size (F). *P < 0.05, **P < 0.01, ***P < 0.001.

Next, we tested whether GDC-0941, a PI3K/Akt inhibitor that has been tested in clinical trials, can reverse transformed phenotype of 10A.B2 MECs in a manner similar to genetic knockdown of PIK3CA. Indeed, GDC-0941 treatment significantly reduced P-Akt-S473 in 10A.B2 cells in a concentration-dependent manner under both 2D and 3D cell culture conditions (Figure 2A). In addition, GDC-0941 significantly inhibited 10A.B2 cell growth (P < 0.001) but had only a marginal effect on 10A.vec cells (Figure 2B). Low-dose (0.3 μM) GDC-0941 didn’t significantly increase total apoptotic (early and late apoptosis) 10A.B2 cells, while high-dose (3 μM) GDC-0941 significantly enhanced apoptosis of 10A.B2 cells (Figure S2A). Both low-dose and high-dose GDC-0941 induced apoptosis in N202 cells, a mammary tumor cell line derived from MMTV-neu mice [20] (Figure S2B). In 3D cell culture, low-dose (0.3 μM) GDC-0941 significantly reversed the disorganized “grape-like” acini phenotype of 10A.B2 cells back to spherical acini similar to the 10A.vec cells, with reduced Ki67-positive proliferating cells but insignificant change of apoptosis (Figure 2C). GDC-0941 treatment also significantly reduced the sizes of the 10A.B2 acinar compared with acini treated with vehicle (Figure 2D, 2E).

Figure 2.

PI3K inhibitor GDC-0941 prevents disorganized acini formation of early ER- MECs. (A) Western blotting of ErbB2, p-Akt, and pan Akt in 10A.vec and 10A.B2 cells treated with GDC-0941 at various doses. (B) Cell growth curve of 10A.vec and 10A.B2 cells treated with GDC-0941. DMSO was used as vehicle control. (C) Microscopy images of 10A.vec and 10A.B2 cells grown in 3D culture for 9 days and 14 days, treated with DMSO or GDC-0941. Immunofluorescent images showing Ki-67 (on day 14), cleaved caspase-3 (on day 9), p-Akt, laminin 5, and DAPI staining in 10A.vec and 10A.B2 acini treated with vehicle or GDC-0941. Magnification was 60×. (D, E) Quantification of percentage of round and transformed acinar phenotypes (D) and quantification of relative (rel.) acinar size (E).

Together, these data indicated that PI3K/Akt activation is critical in early transformation of 10A.B2 MECs, and blocking PI3K/Akt activation by PIK3CA knockdown or low dose inhibitor rescues, at least partially, the dysplasia phenotype.

Targeting PI3K/Akt by GDC-0941 prolonged tumor-free survival of MMTV-neu mice

Encouraged by the in vitro data above, we further tested whether GDC-0941 administration would inhibit mammary tumor initiation in the wild-type neu (neu is the rat homologue of the human ERBB2/HER2 gene) transgenic mouse model of MMTV-neu 202 Mul/J (denoted as MMTV-neu) [29]. MMTV-neu mice develop ER- mammary intraepithelial neoplasia (MIN) at about 10 to 18 weeks and invasive ductal carcinoma (IDC) lesions at approximately 30 weeks, respectively [2]. The well-tolerated GDC-0941 dose in clinic is 330 mg/day, equivalent to 75-150 mg/kg/day in mice [30]. Our dose toxicity evaluations, including body weight, hematology and blood chemistry analyses, showed that GDC-0941 was well tolerated at the dosages of 20 and 100 mg/kg/day with no discernable toxicity (Figure 3A, 3B; Table S1). For potential clinical prevention application, we need to apply a lower dose GDC-0941 that minimize potential side effects while preserving most of its biological activities. We treated MMTV-neu mice (n=29 per group) with 50 mg/kg/day low dose GDC-0941 starting at 10 weeks of age. 50 mg/kg/day GDC-0941 didn’t show significant impact on the body weight of treated mice compared with that of vehicle treated mice after 6-week’s treatment (Figure S3A), suggesting that the dose was well tolerated. Mammary fat pads (MFPs) from multiple vehicle- and GDC-0941-treated mice were collected after 10-week’s treatment, and immunohistochemical (IHC) staining was performed to examine the impacts of GDC-0941 treatment on PI3K pathway activation, mammary epithelial cell proliferation, and apoptosis. The levels of p-Akt-S473 and Ki-67 in mammary epithelial cells from the GDC-0941-treated group were significantly reduced compared with that of the vehicle-treated group (Figure 3C), indicating that low-dose GDC-0941 effectively inhibited PI3K pathway and mammary epithelial cell proliferation in vivo. On the other hand, no significant differences of apoptotic cells were detected in the MFPs between the two groups (Figure S3B). To examine whether low dose GDC-0941 treatment would delay mammary tumor initiation, mice in the vehicle- and GDC-0941-treated groups were evaluated for tumor-free survival. Briefly, mouse was treated with vehicle or GDC-0941 starting at week 10, and the treatment continued until palpable tumor was detected in mouse. By the week of 39, all the mice in the vehicle group (n=15) developed mammary tumors, while 11 out of the 15 mice in the GDC-0941-treated group remained tumor-free (Figure 3D). Indeed, low dose GDC-0941 treatment significantly extended the median tumor-free survival of the mice from 31.6 weeks (vehicle group) to 48.1 weeks (GDC-0941 group) (Figure 3D). Altogether, these data demonstrated that targeting PI3K/Akt with low dose GDC-0941 delayed tumor initiation and increased tumor-free survival in MMTV-neu mouse model of ER- mammary tumors.

Figure 3.

PI3K inhibitor GDC-0941 prolongs tumor-free survival in MMTV-neu mouse model. A. Ratio of total body weight in FVB mice treated with oral gavage of vehicle (n=3) or GDC-0941 at 20 mg/kg (n=3) and 100 mg/kg (n=4) for 2 weeks. B. Percentage of liver/body weight in FVB mice treated with oral gavage of vehicle (n=3) or GDC-0941 at 20 mg/kg (n=3) and 100 mg/kg (n=4) for 2 weeks. C. Representative images and quantification from immunohistochemistry staining for p-Akt and Ki-67 in MFPs of 10-week’s vehicle- or GDC-0941-treated MMTV-neu mice. The respective scores are shown at the bottom. Magnification was 40×. D. Kaplan-Meier tumor-free survival curve in MMTV-neu mice treated with either vehicle (n=15, red) or GDC-0941 (n=15, blue) starting at week 10 (pointed by red triangle). Treatment was stopped when palpable tumor was detected in mouse.

Low-dose GDC-0941 increases T-cell infiltration in mammary tissues of MMTV-neu mice

PTEN loss and PI3K/Akt activation in tumor cells are associated with immune evasion, reduced T-cell infiltration, and resistance to immunotherapy in certain types of cancer [18,31]. Thus, we explored whether GDC-0941 delayed mammary tumor initiation in MMTV-neu mice not only by inhibiting mammary epithelial cell proliferation (Figure 3C) but also by influencing T-cell infiltration in the TME. We examined immune cell infiltration in MFPs of 10-week’s-veh-or GDC-0941-treated mice by immunofluorescent staining. Indeed, compared with vehicle treated mice, low-dose GDC-0941 treatment significantly increased CD3⁺ total T-lymphocytes and CD8⁺ T lymphocytes (Figure 4A, 4B) in MFPs compared with that of vehicle-treated mice. Additionally, immunohistochemistry also showed increased infiltration of GZMB⁺ cytotoxic T cells in MFPs from GDC-0941 treated mice compared to vehicle control mice (Figure 4C), suggesting enhanced immune surveillance by low dose GDC-0941 treatment. To test whether GDC-0941 would affect T-cells directly, low (0.3 µM) and high (3 µM) doses of GDC-0941 were applied to spleen T-lymphocytes in the absence or presence of the N202 mammary tumor cells. Interestingly, in the absence of tumor cells, low-dose GDC-0941 slightly increased the percentage of CD8⁺ T-cells and enhanced T-cell activation (IFNγ⁺) as well, while high-dose GDC-0941 significantly reduced CD8⁺ T-cell activation (Figure S4). On the other hand, under co-culture with the N202 mammary tumor cells derived from MMTV-neu mice [20], GDC-0941 did not significantly impact on the percentage and activation of CD8⁺ T-cells (Figure 4D, 4E), suggesting that other mechanisms, such as enhanced T-cell recruitment, may contribute to increased T-cell infiltration in GDC-0941-treated MFPs.

Figure 4.

Low dose GDC-0941 treatment increased T-cell infiltration in MFPs of MMTV-neu mice. (A, B) Immunofluorescent staining and quantification of CD3⁺ T cells (A) and CD8⁺ T cells (B) in MFPs of 10-week’s vehicle- and GDC-0941-treated MMTV-neu mice. The arrows pointed to CD3⁺ T cells and CD8⁺ T, respectively. Magnification was 60×. (C) Immunohistochemical staining and quantification of GZMB⁺ cells in MFPs of 15-week’s vehicle- or GDC-0941-treated MMTV-neu mice. Magnification was 20×. (D) Percentage of CD8⁺/CD45⁺ cells of splenocytes after co-culture with 0.3 μM vs 3 μM GDC-0941-pretreated N202 cells. (E) IFN-γ⁺ percentage in CD45⁺ cells of splenocytes after co-culture with 0.3 μM vs 3 μM GDC-0941-pretreated N202 cells. *P < 0.05, **P < 0.01, ***P < 0.001.

GDC-0941 enhances T-cell migration towards tumor cells

PTEN loss and PI3K/Akt activation could result in dysregulation of cytokines and chemokines, which may suppress T-cell migration [18]. To examine whether GDC-0941 modulates T-cell recruitment to tumor cells, N202 mammary tumor cells were treated with low-dose GDC-0941 (0.3 µM) or vehicle, overlaid into the plate well of the trans-well unit and co-cultured with splenocytes loaded in culture inserts of the trans-well plate. The migrated T-cells were then analyzed by flow cytometry (Figure 5A). CD3⁺ T-cells migrated toward N202 tumor cells but not to culture medium; migration of total CD3⁺ T-cells toward tumor cells was significantly increased by GDC-0941 treatment compared to vehicle treatment (Figure 5B). Tumor cells’ recruitment of CD45⁺CD3^- immune cells, including B cells and myeloid cells, are similar between two treatments (Figure 5C). Furthermore, we detected significantly increased migrations of CD4⁺, CD8⁺, and IFN-γ⁺ CD8⁺ T-cells in response to low-dose GDC-0941 treatment of N202 tumor cells (Figure 5D-F). Importantly, GDC-0941 didn’t significantly impact on T-cell migration in the absence of N202 tumor cells (Figure 5B-F), which suggested that low dose GDC-0941 modulate tumor cell-dependent chemotaxis migration of T-cells, but not random migration of T-cells. Next, we tested whether GDC-0941 also increases human lymphocytes migration towards human ER- breast cancer cells by seeding the HER2+ER- HCC-1569 human breast cancer cells in the plate well and human peripheral blood mononuclear cells (PBMCs) in culture inserts of the trans-well plate for migration assay. Indeed, significantly increased migration of human CD3⁺, CD4⁺, CD8⁺, and IFN-γ⁺ CD8⁺ T-cells were detected in trans-wells seeded with GDC-0941-treated HCC-1569 cells (Figure 5G-J), while migration of CD3^- immune cells didn’t change significantly (Figure 5K). Conversely, reduced migration of human CD3⁺, CD4⁺, CD8⁺ T lymphocytes and IFN-γ⁺ CD8⁺ T-cells were detected in PI3K pathway activated 10A.B2 cells compared with that of 10A.vec cells (Figure S5A-D). Altogether, our data indicate that PI3K pathway activation in mouse and human ER- breast cancer cells may contribute to T-cell suppression and inhibiting PI3K activation in tumor cells by low dose GDC-0941 enhances T-cell chemotaxis migration.

Figure 5.

Low dose GDC-0941 enhances tumor cell-dependent T-cell recruitment. A. In vitro model of T-cell recruitment. B. Representative figures and quantification of mouse CD3⁺ T cells migrating to vehicle- and 0.3 μM GDC-0941-treated medium or N202 cells. C-F. Quantification of mouse CD45⁺CD3^-, CD4⁺, CD8⁺ and IFN-γ⁺ CD8⁺ cells migrating to vehicle- and 0.3 μM GDC-0941-treated N202 cells. G. Representative figures and quantification of human CD3⁺ T cells migrating to vehicle- and 0.3 μM GDC-0941-treated HCC-1569 cells. H-K. Quantification of human CD4⁺, CD8⁺, IFN-γ⁺ CD8⁺, and CD45⁺CD3^- cells migrating to vehicle- and 0.3 μM GDC-0941-treated HCC-1569 cells. Each experiment had performed 3-6 times. *P < 0.05, **P < 0.01, ***P < 0.001.

GDC-0941 enhances T-cell migration via CCL5/CXCL10

Next, we investigated how inhibition of PI3K/Akt in tumor cells by low dose GDC-0941 enhances T-cell chemotaxis. We examined mRNA expression levels of well-known T-cell-attracting chemokines, such as CCL5, CXCL9, and CXCL10, in response to GDC-0941 treatment in ER- murine mammary tumor cells, such as N202 and McNeuA generated from MMTV-neu transgenic mice, as well as human breast cancer cells. RT-PCR analysis detected increased mRNA levels of ccl5 and cxcl10 in murine mammary tumor cells (N202 and McNeuA) and human breast cancer cell HCC-1569 after inhibition of PI3K/Akt by GDC-0941 (0.3 µM) treatment in vitro (Figures 6A, S6A, 6B, S7A). Consistently, GDC-0941 treatment in vivo also increased ccl5 and cxcl10 mRNA expressions in MFPs of MMTV-neu mice compared with those of vehicle-treated mice (Figure 6B). Both in vitro and vivo data suggest that GDC-0941 treatment can upregulate T-cell-attracting chemokines ccl5 and cxcl10 expressions in mammary tumor cells.

Figure 6.

Low dose GDC-0941 increases CCL5/CXCL10 expression in tumor cells and enhances CCL5/CXCL10 dependent T-cell chemotaxis migration. A. mRNA expression of ccl5 and cxcl10 after vehicle and 2 h, 6 h, and 12 h of 0.3 μM GDC-0941 treatment in N202 cells. B. mRNA expression of ccl5 and cxcl10 after vehicle and GDC-0941 treatment in MFP in 20-week-old MMTV-neu mice. C-F. Quantification of mouse CD3⁺, CD4⁺, CD8⁺ and IFN-γ⁺ CD8⁺ cells under IgG control or anti-CCL5 antibody or anti-CXCL10 antibody treatment followed by GDC-0941 in N202 cells. Each experiment had performed 3 times. *P < 0.05, **P < 0.01, ***P < 0.001.

To test whether CCL5 and CXCL10 mediate GDC-0941 treatment-induced T-cell chemotactic migration towards mammary tumor cells, CCL5 and CXCL10 blocking antibodies were applied individually or in combination to GDC-0941-pretreated N202 cells under co-cultured with T-cells. Indeed, blockade of either CCL5 or CXCL10 alone significantly inhibited GDC-0941 treatment-induced CD3⁺, CD4⁺, CD8⁺, and IFNγ⁺ CD8⁺ T-cell migrations to N202 tumor cells (Figure 6C-F). Combination of anti-CCL5 and anti-CXCL10 antibodies inhibited T-cell migration with a slightly better trend (Figure 6C-F). Similarly, anti-human CCL5 blocking antibody also inhibited human CD3⁺, CD4⁺, CD8⁺, and IFNγ⁺ CD8⁺ T-cell migration to GDC-0941-treated HCC-1569 cells (Figure S7B-E). However, CXCL10 antibody did not show much effect on human T-cell migration (Figure S7B-E). We found that the HCC-1569 cells without GDC-0941 treatment already express CXCL10 at a very high level (data not shown), it is possible that the CXCL10 antibody can’t sufficiently block CXCL10 or/and CXCL10 is not a major mediator of GDC-0941-induced human PBMC T-cell migration. To further test the role of CXCL10 in human T-cell recruitment, we knocked down CXCL10 in HCC-1569 cells by siRNAs (Figure S8A). Knocking down of CXCL10 in HCC-1569 cells indeed abolished enhanced human T-cell migration in response to GDC-0941 treatment with minimal effects on migration of other immune cells in human PBMC (Figure S8B-F). Taken together, these data indicate that PI3K/Akt inhibition in mammary tumor cells upregulates critical T-cell-attracting chemokines, including CCL5 and CXCL10, which mediate T-cell chemotatic migration.

PI3K activation in human ER- breast cancer is associated with reduced CD8A, CCL5 and CXCL10

Above data suggest that PI3K activation suppresses CCL5 and CXCL10 expressions resulting in decreased T-cell recruitment and the escape of immune surveillance. To examine the relevance of our finding in human ER- breast cancer patients, we examined whether PI3K activation signature may correlate with CCL5, CXCL10, and CD8A expressions in human ER- breast cancer using dataset from the Cancer Genome Atlas (TCGA). Indeed, elevated PI3K activation based on a 32-gene expression signature [25], is associated with significantly reduced expressions of CD8A, CCL5, and CXCL10 (Figure 7A). Moreover, CD8A expression shows strong positive correlations with CCL5 and CXCL10 expressions, consistent with their critical roles in T-cell recruitment (Figure 7B). We further validated these findings in another ER- breast cancer patient cohort (GSE20685) which yielded similar results, although PI3K activation signature only showed a trend of negative association with CXCL10 expression which exhibited a relatively weak correlation with CD8A expression (Figure 7C, 7D). Altogether, these data support that PI3K activation is associated with significantly reduced expression of T-cell-attracting chemokines and CD8⁺ T-cell infiltration, which could contribute to immune evasion in ER- breast cancer (Figure 8).

Figure 7.

PI3K activation signature is associated with reduced CD8A, CCL5, and CXCL10 in human ER- breast cancer. A. Comparison of CD8A, CCL5, and CXCL10 expression between PI3K signature high (n=58) versus PI3K signature low (n=58) ER- breast cancers from TCGA breast cancer dataset. (t-test, two tails). B. Pearson correlations between CD8A vs CCL5, CD8A vs CXCL10 in TCGA ER- breast cancer. C. Comparison of CD8A, CCL5, and CXCL10 expression between PI3K signature high (n=41) versus PI3K signature low (n=41) ER- breast cancers from ER-breast cancer dataset (GSE20685). (t-test, two tails). D. Pearson correlations between CD8A vs CCL5, CD8A vs CXCL10 in GSE20685 ER- breast cancer dataset.

Figure 8.

Schematic showing that PI3K/Akt is elevated in early stage of ER- breast cancer and inhibiting the PI3K/Akt pathway with low-dose GDC-0941 effectively inhibited ER- mammary tumor initiation by both inhibiting MEC proliferation and by increasing T-cell recruitment in a CCL5/CXCL10-dependent manner to enhance immune surveillance.

Discussion

Dysregulation of the PI3K/Akt/mTOR/p70S6K pathway has been found in multiple types of cancer, including breast cancer, and it confers proliferation advantages that promote malignant transformation and tumorigenesis [32-34]. Our bioinformatics analysis in precancerous patient samples revealed that PI3K/Akt activation occurred at very early stage (e.g., hyperplasia) of ER- breast cancer development, suggesting that targeting the PI3K/Akt pathway might prevent hyperplasia transition to early tumor initiation. A growing number of kinase inhibitors have shown remarkable clinical efficacy in cancer treatment and been approved by the US Food and Drug Administration [35]. However, kinase inhibitors have not been keenly considered for cancer prevention due to toxic effects associated with the high doses used for treatment of late stage cancer. Here, we propose a novel approach of using highly specific, low-dose kinase inhibitors with low toxicity to target PI3K/Akt for preventing early-lesion transition to early-stage breast cancer. In a phase 2 clinical trial, GDC-0941 was tested and suppressed tumor cell proliferation [36]. Our data showed that PI3K/Akt inhibition by genetic knockdown or low dose GDC-0941 both significantly reduced disorganized acinar growth of the semi-transformed 10A.B2 human mammary epithelial cells in 3D cell culture (Figures 1E-G, 2C-E). More importantly, low-dose treatment with GDC-0941 in MMTV-neu mice was effective in suppressing Akt phosphorylation, decreasing cell proliferation, delaying mammary tumor initiation, and prolonging tumor-free survival (Figure 3C, 3D) without inducing apparent adverse effects (Figure 3A, 3B; Table S1).

In addition to their well-known roles in promoting cancer cell proliferation and resistance to apoptosis, PTEN loss and PI3K/Akt activation also help tumor cells to evade immune surveillance during tumor evolution. Here, we found that inhibiting PI3K/Akt activation using low-dose GDC-0941 enhanced T-cell infiltration in mammary tissues surrounding DCIS-like lesions of MMTV-neu mice (Figure 4A, 4B). Although high dose GDC-0941 may inhibit T-cell activity in vitro (Figure S4B), low dose GDC-0941 had no significant toxicity in T-cells in vitro and no discernable adverse effects in mice during 20 or more weeks of treatment (Figures 3A, 3B, S4; Table S1). On the other hand, low-dose GDC-0941 treatment facilitated tumor cell-mediated T-cell chemotaxis, which is critical for immune surveillance (Figure 5). These data suggest that enhanced immune surveillance in tumor microenvironment may also contribute to prolonged tumor-free survival in GDC-0941-treated mice on top of targeting tumor cells themselves.

Our screening of chemokines expressed by mammary tumor cells that are involved in GDC-0941-facilitated T-cell infiltration revealed that, two potent chemoattractant for T-cells, CCL5 and CXCL10, were significantly up-regulated after GDC-0941 treatment and contributed to T-cell chemotaxis migration in both mouse and human breast cancer cells (Figures 6A, S6A, 6B, S7A). On the other hand, chemokines involved in myeloid migration, such as CCL2, CXCL2, and CXCL5, did not show significant changes in response to PI3K/Akt inhibition in tumor cells (data not shown), consistent with minimal alterations of CD45⁺CD3^- cell migration by GDC-0941 treatment in vitro (Figure 5C, 5K). A potential mediator of CCL5 and CXCL10 suppression by PI3K/Akt pathway is Erk5 since Erk5 knockout in prostate tissue of a Pten-deficient mouse model of prostate cancer significantly upregulated ccl5 and cxcl10 expressions [19]. Additionally, the crucial roles of ERK5 overexpression in early stage breast cancer development [37] as well as breast cancer invasion/metastasis [38] have been suggested. However, inhibiting PI3K pathway by GDC-0941 didn’t affect p-Erk5 expression (data not shown), suggesting other downstream signals of PI3K pathway are involved in CCL5 and CXCL10 regulation, which warrants future investigation.

A limitation of our study is that only one PI3K inhibitor GDC-0941 was tested in our models. Very recently, two PI3K inhibitors predominantly targeting PI3K-α have been approved by FDA. One of them, alpelisib (BYL719), is approved for treatment of ER+ HER2- breast cancer in combination with endocrine therapy. It merits future testing to determine whether alpelisib has similar effects as that of GDC-0941 in prevention and early intervention of ER- breast cancer.

In summary, we identified PI3K/Akt activation as a potential target for ER- breast cancer prevention and showed that inhibiting the PI3K/Akt pathway by low-dose GDC-0941 effectively suppressed abnormal acini growth of semi-transformed mammary epithelial cells in vitro and delayed ER- mammary tumor initiation in vivo. In addition, low dose of PI3K inhibitor also significantly enhanced T-cell recruitment in a CCL5 and CXCL10-dependent manner (Figure 8). Cancer vaccines have shown efficacy in secondary prevention of HER2+ breast cancer [39]. It is anticipated that a low-dose PI3K inhibitor in combination with cancer vaccines may further improve the immunoprevention efficacy of cancer vaccines in breast cancer. Our data in this study, along with other novel findings [2,14,40], could serve as scientific bases of developing effective prevention strategies for ER- breast cancer in the future.

Disclosure of conflict of interest

None.