PD-1 blockade attenuates immunosuppressive myeloid cells due to inhibition of CD47/SIRPα axis in HPV negative head and neck squamous cell carcinoma.

Myeloid-derived suppressor cells (MDSCs) and tumor associated macrophages (TAMs) play key roles in the tumor immune suppressive network and tumor progression. However, precise roles of programmed death-1 (PD-1) in immunological functions of MDSCs and TAMs in head and neck squamous cell carcinoma (HNSCC) have not been clearly elucidated. In the present study, we show that PD-1 and PD-L1 levels were significantly higher in human HNSCC specimen than in normal oral mucosa. MDSCs and TAMs were characterized in mice and human HNSCC specimen, correlated well with PD-1 and PD-L1 expression. αPD-1 treatment was well tolerated and significantly reduced tumor growth in the HNSCC mouse model along with significant reduction in MDSCs and TAMs in immune organs and tumors. Molecular analysis suggests a reduction in the CD47/SIRPα pathway by PD-1 blockade, which regulates MDSCs, TAMs, dendritic cell as well as effector T cells. Hence, these data identify that PD-1/PD-L1 axis is significantly increased in human and mouse HNSCC. Adoptive αPD-1 immunotherapy may provide a novel therapeutic approach to modulate the micro- and macro-environment in HNSCC.


Proteome profiler antibody array analysis
The proteome profiler TM mouse cytokine array panel an array kit (ARY006, R&D Systems) consists of 40 different cytokine and chemokine antibodies spotted in duplicate onto a membrane. Tissue lysis from 5 pooled Tgfbr1 flox/flox ; Pten flox/flox K14Cre ERtam-/tongue mucosa samples, 5 Tgfbr1/Pten 2cKO tongue mucosa samples and 5 Tgfbr1/Pten 2cKO tongue squamous cell carcinoma samples were used according to the manufacturer's instructions. Briefly, after blocking with 10% bovine serum albumin in Tris-buffered saline, the membranes were incubated with 250 μg (50 μg from each individual sample) of tissue lysis buffer in the presence of the supplied antibody cocktail. Following washes and incubation with a Streptavidin-HRP buffer, positive signals were revealed using ECL reagent. Positive (six spots) and negative controls (2 spots) were located in the corner of each array kit. A chemiluminescence substrate (Millipore, Billerica, MA, USA) was used to detect protein expression and the data were captured by multiple exposures to Kodak BioMax Light film. The arrays were scanned into a computer. For quantitation, after background subtraction, the average optical intensity of duplicate spots for each cytokine was normalized to the average of positive controls on the same chip using the Image J software (NIH, Bethesda, MD, USA).

Flow cytometry analysis
To obtain single cell suspensions, tumour tissues of mice (K14-Cre ERtam+/-; Tgfbr1 flox/flox /Pten flox/flox ) with or without PD-1 blockade antibody RMP1-14 treatment were processed using a gentle Macs dissociator and a murine tumour dissociation kit (Miltenyi Biotec). Single cell suspensions from spleens, lymph node, and blood were prepared according to a standardized protocol [3]. Wild type controls with the same dose of tamoxifen were used for flow cytometry analysis. These cells were labelled with FITC-conjugated anti-mouse CD4, CD8 and CD11b, PEconjugated anti-mouse PD-1 and Gr-1 (all from Becton Dickinson, Mountain View, CA); Percp-Cy5.5-conjugated anti-mouse F4/80, FITC-conjugated anti-mouse CD11c, PE-conjugated anti-mouse MHC-II, CD40, CD80 and CD86 (all from eBioscience, San Diego, CA); and isotype-matched IgG controls (eBioscience, San Diego, CA). The cells were analyzed on a FACScalibur flow cytometer equipped with CellQuest software, and gated by the side scatter and forward scatter filters. (Becton Dickinson, Mountain View, CA). Live cells were gated by 7AAD (Invitrogen) and populations were phenotyped as described above.

Cell culture and RNAi
HNSCC cell lines CAL27 and FaDu were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and genotype confirmed using STR sequence. Cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM)/F12, 10% fetal bovine serum (FBS), at 5% CO2 and 37°C in a humidified incubator. For the TGFBR1 and PTEN knockdown experiment, 2 TGFBR1 siRNA (siRNA5 and SiRNA6),

SUPPLEMENTARY MATERIALS AND METHODS
www.impactjournals.com/oncotarget/ Oncotarget, Supplementary Materials 2015 2 PTEN siRNA (siRNA5 and siRNA6) were transfected into appropriate cells using HiPerfect transfection reagent (Qiagen, Germantown, MD) with a final concentration of 5nM as previously described. All-star negative controls (Qiagen), confirmed to have no interference with other miRNAs, were used as negative controls. MAPK1 siRNA and Cell Death siRNA (Qiagen) were used as positive controls. For inhibition efficiency and target mRNA transcription studies, RNA was extracted 24 h after transfection and validated by qPCR. For protein extraction, cells were lysed 48 h after transfection. A combined knock down of TGFBR1 (siRNA5) and PTEN siRNA (siRNA6) were performed after validation of knock down efficiency as previous reported [1,2].

Quantitative real-time PCR analysis
Quantitative real-time RT-PCR analysis was performed to evaluate the expression of representative cytokine and chemokine in Tgfbr1 flox/flox ; Pten flox/flox K14Cre ERtam-/tongue mucosa, Tgfbr1/Pten 2cKO tongue mucosa and Tgfbr1/Pten 2cKO tongue squamous cell carcinoma. RNA was extracted and reverse-transcribed into cDNA with random hexamers and Maloney murine leukemia virus reverse transcriptase (Takara, Kyoto, Japan). Real-time analysis was performed using Cd54, Csf1, Cxcl1, Cxcl2 and Ccl2-specific primers and probes with the ABI 7500 Sequence Detection System instrument and software (Applied Biosystems, Foster City, CA). Experiments were performed in triplicate for each sample.

Western blotting
Tumours that developed in Tgfbr1/Pten 2cKO mice were carefully dissected (n = 6, respectively). Cultured cells and tissue were lysed in a T-PER buffer containing1% phosphatase inhibitors and complete mini cocktail (Roche). The detailed immunoblotting procedures were described previously [4]. A total amount of 30 μg protein from each sample was denatured and then subjected to 12% SDS-polyacrylamide gel electrophoresis followed by transfer onto polyvinylidene fluoride membranes (Millipore Corporation, Bil-lerica, MA). Nest, the blots were using an enhanced chemiluminescence detection kit (West Pico, Thermo). GAPDH was detected on the same membrane and used as a loading control.

Human HNSCC tissue array
The School and Hospital of Stomatology of Wuhan University Medical Ethics Committee approved this study, and informed consent was obtained from the patients before they underwent surgery. Custom made tissue microarray including 86 HNSCC, 12 oral epithelial dysplasia and 32 normal oral mucosa tissue samples were used for immunohistochemistry staining as previous described [1].

Immunohistochemistry and immunofluorescence
Tumors from Tgfbr1/Pten 2cKO mice were dissected and fixed as previously described [4], and slides were stained with the appropriate antibody using a standard immunohistochemical staining protocol as previously described [5]. The bound antibodies were visualized using an appropriate biotin-conjugated secondary antibody and the Vectastain ABC Elite kit (Vector, Burlingame, CA, USA), using 3, 30-diaminobenzidine as the substrate (Sigma). For immunofluorescence, slides were hydrated in alcohol, washed three times in PBS, retrieved using sodium citrate in a pressure cooker, blocked with 2.5% bovine serum album in PBS buffer for 1 hour at 37°C and were then incubated with primary antibody overnight at 4°C. The next day, slides were incubated with fluorochrome conjugated secondary antibodies (Alexa 594 anti-rabbit and Alexa 488 anti-mouse; Invitrogen) and mounted in Vectashield with 4', 6-diamidino-2phenylindole (DAPI; Vector Laboratories). Fluorescence images were then captured using a CLSM-310, Zeiss fluorescence microscope.

Scoring system, hierarchical clustering and data visualization
Whole slices were scanned using an Aperio ScanScope CS scanner (Vista, CA, USA) with a background substrate for each slice, and quantified using Aperio Quantification software (Version 9.1) for membrane, nuclear, or pixel quantification [6]. An area of interest was selected either in the epithelial or the cancerous area for scanning and quantification. The histoscore of membrane and nuclear staining was calculated as a percentage of different positive cells using the formula (3+) × 3 + (2+) × 2 + (1+) × 1. The histoscore of pixel quantification was calculated as the total intensity/ total cell number [1]. The threshold for scanning of different positive cells was set according to the standard controls provided by Aperio [1,7].The expression scores were converted into scaled values centered on zero in Microsoft excel. Next, the hierarchical analysis was achieved using the Cluster 3.0 with average linkage based on Pearson's correlation coefficient [8]. Java TreeView 1.0.5 was used to visualize the results [9]. Finally, we arranged the clustered data and tissue samples on the horizontal axis and vertical axis respectively. Biomarkers with a close relationship are located next to each other.

Statistical analysis
Data analyses were performed using Graph Pad Prism version 5.0 for Windows (Graph Pad Software Inc, La Jolla, CA). One-way ANOVA followed by the post-Tukey multiple comparison tests were used to analyze the differences in protein levels, RNA levels and positive cells among each group. A unpaired t test was used to analyze immunohistochemical staining of the difference between the PD-1 blockade group and vehicle group. Two-tailed Pearson's statistics was used for the correlated expression of PD-1, PD-L1, CD11b, CD33, CD68 and CD163 after the confirmation of the sample with Gaussian distribution.
The mean values ± SEM with a difference of P < 0.05 were considered statistically significant.