Abstract
Despite therapeutic advancements, there has been little change in the survival of patients with head and neck squamous cell carcinoma (HNSCC). Recent results suggest that cancer-associated fibroblasts (CAF) drive progression of this disease. Here, we report that autophagy is upregulated in HNSCC-associated CAFs, where it is responsible for key pathogenic contributions in this disease. Autophagy is fundamentally involved in cell degradation, but there is emerging evidence that suggests it is also important for cellular secretion. Thus, we hypothesized that autophagy-dependent secretion of tumor-promoting factors by HNSCC-associated CAFs may explain their role in malignant development. In support of this hypothesis, we observed a reduction in CAF-facilitated HNSCC progression after blocking CAF autophagy. Studies of cell growth media conditioned after autophagy blockade revealed levels of secreted IL6, IL8, and other cytokines were modulated by autophagy. Notably, when HNSCC cells were cocultured with normal fibroblasts, they upregulated autophagy through IL6, IL8, and basic fibroblast growth factor. In a mouse xenograft model of HNSCC, pharmacologic inhibition of Vps34, a key mediator of autophagy, enhanced the antitumor efficacy of cisplatin. Our results establish an oncogenic function for secretory autophagy in HNSCC stromal cells that promotes malignant progression. Cancer Res; 77(23); 6679–91. ©2017 AACR.
Introduction
As the fifth most common cancer worldwide, head and neck squamous cell carcinoma (HNSCC) is a leading global health burden (1). HNSCC causes significant morbidity and mortality, and is associated with intensive treatment protocols and a 5-year survival rate of less than 50% (2). Despite therapeutic advancements, the survival rate has remained relatively unchanged for the last 40 years. A better understanding of the biology of the disease is necessary for improving these dismal outcomes.
In the past decade, the tumor microenvironment's role in promoting cancer progression and resistance to therapy has gathered great attention (3). Cancer-associated fibroblasts (CAF) form the predominant nonmalignant cell type in the HNSCC microenvironment. Studies in our laboratory and others have demonstrated the critical role CAFs play in facilitating HNSCC progression (4). CAFs promote HNSCC progression by secreting growth factors, remodeling the extracellular matrix, and potentiating therapy resistance (5). HNSCC cells symbiotically communicate with CAFs through secreted factors (6); however, little is understood of the activation and biology of CAFs.
The survival-promoting pathway of autophagy is upregulated in many cancer types, including HNSCC. Autophagy serves to capture and degrade intracellular components for homeostasis, metabolism, and cell survival. Initiation of autophagy centers around the phosphorylation of the Beclin-1/Vps34 complex, which initiates the nucleation of an autophagosome (7). Autophagy-related gene product (Atg8) mammalian homolog LC3 is cleaved and lipidated (the lipidated form denoted as LC3-II), and is incorporated into the autophagosome membrane, serving as the most widely recognized marker of the autophagosome (8). SQSTM1, or p62, acts as a molecular shuttle to target ubiquinated cargo to the autophagosome (9). The autophagosome is trafficked to the lysosome. LC3-II in the autophagosome is rapidly degraded upon fusion with the lysosome. To accurately determine the levels of LC3-II undergoing rapid degradation, it is essential to inhibit the degradative enzymes in the lysosome, thus inhibiting autophagic flux. Chloroquine neutralizes the acidity in the lysosome, inactivating the enzymes and preventing the degradation of the autophagosome (10, 11).
There are a number of ongoing clinical trials investigating the role of autophagy inhibition in several cancer types, both as a stand-alone therapy and as an adjuvant to conventional therapeutic regimens. Accumulating evidence indicates the improvement of anticancer regimens by adjuvant autophagy inhibition (9). However, these studies are limited by the high doses of hydroxychloroquine, an orally available modification of chloroquine, with no clinically viable alternatives (12). Promisingly, advances in the understanding of the autophagic pathway over the last decade have resulted in small-molecule inhibitors that specifically target autophagy, such as SAR405, which targets Vps34 (13).
Although paradigmatically a degradation pathway, a growing appreciation for a novel role of autophagy in cellular secretion has been reported (14). Secretory autophagy is involved in the export of a variety of cellular cargoes. This includes inflammatory mediators, such as IL1β, IL6, IL8, and IL18 (15, 16); granule contents from Paneth, mast, goblet, and endothelial cells (17–20); as well as release of intracellular pathogens (21). In addition, high levels of basal autophagy correlate with a unique secretory profile when compared with cancer cells with lower levels of autophagy (22).
We set out to investigate the role of autophagy in HNSCC patient-derived CAFs and discovered a high basal level of autophagy compared with normal fibroblasts (NF) derived from the same anatomic location. Mitigating autophagy significantly reduced CAF-induced HNSCC progression. We further assessed the role of autophagy on HNSCC tumor growth. Treatment with a Vps34 inhibitor, SAR405, attenuated xenograft growth. Furthermore, autophagy inhibition potentiated the effects of standard-of-care therapy.
Patients and Methods
Cells and reagents
HNSCC and tonsil or uvulopalatoplasty explants from cancer-free patients were collected with written consent from patients under the auspices of the University of Kansas Medical Center Biospecimen Repository Core Facility. All protocols for collection and use were approved by the Human Subject Committee at the University of Kansas Medical Center (Kansas City, KS). Primary fibroblast explants were established using our previously described protocol (4), and all fibroblast lines used were cultured for no more than 12 passages. In all experiments, results presented are from fibroblasts derived from a minimum of two patient explants.
Well-characterized HNSCC cell lines [UM-SCC-1 (a gift from Dr. Tom Carey, University of Michigan, Ann Arbor, MI), OSC19 (a gift from Theresa Whiteside, University of Pittsburgh, Pittsburgh, PA), HN5 (a gift from Dr. Jeff Myers, The University of Texas MD Anderson Cancer Center, Houston, TX)] were used in this study (23). Established cell lines were authenticated by short tandem repeat profiling at Johns Hopkins in 2015 using the Promega GenePrint 10 kit and analyzed using GeneMapper v4.0 software. All cells were maintained in DMEM (Corning) with 10% heat-inactivated FBS (Sigma-Aldrich) without antibiotics. Cells were incubated at 37°C in the presence of 5% CO2.
Chloroquine diphosphate salt, 4-nitroquinoline N-oxide (4-NQO), IL6, IL8, and basic fibroblast growth factor (bFGF) were obtained from Sigma-Aldrich. SAR405 was obtained through APExBIO. Cisplatin was obtained from Fresenius Kabi.
The following antibodies were used: LC3 A/B (#12741), Beclin-1 (#4122), phospho-p70 S6K (Thr389; #9205), Stat3 (#9139), and phospho-Stat3 (Tyr705; 9145) from Cell Signaling Technology; β-tubulin from Sigma-Aldrich; p62 (SQSTM1, M01) from Abnova; vimentin (#6260) from Santa Cruz Biotechnology; and neutralizing antibodies IL6 (6708), IL8 (6217), and secondary anti-rabbit IgG DyLight 680 (#35568), anti-rabbit IgG Dylight 488 (#35553), and anti-mouse IgG Dylight 800 (#35521) from Thermo Fisher Scientific. Hoescht 33342 was used as nuclear counterstain (Thermo Fisher Scientific).
The following primer sequences were used: mTOR (F: GGCCGACTCAGTAGCAT; R: CGGGCACTCTGCTCTTT); SOX2 (F: ACGGAGCTGAAGCCGCC; R: CTTGACGCGGTCCGGGCT); β-ACTIN (F: AGGGGCCGGACTCGTCATACT; R: GGCGGCACCACCATGTACCCT)—all obtained from Thermo Fisher Scientific.
Control (#44236), siBECN1 (#29797), and siFGF-2 (#39446) siRNAs were obtained from Santa Cruz Biotechnology.
Electron microscopy
Tissues were processed at University of Kansas Medical Center Electron Microscopy Research Lab facility. Tissue samples were fixed in 2.5% glutaraldehyde in 0.1 mol/L cacodylate buffer and postfixed in 1.5% osmium tetraoxide. Samples were embedded in resin–propylene oxide and allowed to cure. Tissues were sectioned using a diatome diamond knife on a Leica UC7 ultra microtome at 80-nm thickness. Sections were collected on 200 mesh copper grids, and samples imaged using a JEOL JEM-1400 transmission electron microscope at 100 KV.
Immunoblotting
Whole-cell lysates were extracted using RIPA lysis buffer and a mixture of protease and phosphatase inhibitors (Minitab; Roche). Lysates were sonicated on ice, debris removed by centrifugation, and supernatants stored at −80°C. SDS-polyacrylamide 12% gels were used to separate proteins, and proteins were transferred to nitrocellulose membranes. Membranes were blocked with Odyssey blocking buffer (LI-COR) in a 1:1 mixture with PBS 1% Tween-20 (PBST). Primary antibodies were incubated overnight in 1:1 blocking buffer to PBST. Primary antibodies were detected using DyLight conjugated secondary antibodies. Protein bands were detected using LI-COR Odyssey protein imaging system and quantified using ImageJ software (v. 1.50i).
Immunofluorescence imaging
Cells were plated at a low confluence (1 × 104 cells per well) in 8-well chamber slides (Thermo Fisher Scientific). Methanol (70%) was used to fix cells. Triton-X (0.5%) in PBS was used as a permeabilization buffer. Cells were blocked with a 2% BSA solution. For paraffin sections, paraffin was removed by xylene, and the section was rehydrated by ethanol titration. Antigen was retrieved using sodium citrate solution. Both cells and tissue sections were then incubated overnight in a primary antibody (1:100 concentration in 2% BSA) at 4°C. DyLight (488-anti-rabbit; 550-anti-mouse) conjugated secondary antibodies were used, and Hoescht staining was done following the manufacturer's instructions for nuclear detection. Slides were mounted with coverslip in VECTASHIELD mounting media. Images were captured on a Nikon Eclipse TE2000 inverted microscope with a Photometrics CoolSNAP HQ2 camera. LC3 puncta per cell of at least 30 cells in each experimental arm were identified by blinded observer at 20× magnification.
Conditioned media collection
CAFs (3 × 105 cells/well in 60-mm dish) were plated in 10% FBS DMEM, and were treated with 20 μmol/L chloroquine for 6 hours or vehicle control (H2O). Following drug treatment, cells were washed two times with serum-free media, and then conditioned media was collected over 24 hours in serum-free DMEM. Following conditioned media collection, cell lysates were harvested for immunoblotting to confirm autophagy inhibition by assessing LC3 levels. Conditioned media supernatants were clarified by centrifugation and stored for no more than 2 weeks at 4°C.
CAFs (3 × 105 cells/well in 60-mm dish) were plated in 10% FBS DMEM and were transfected with 100 nmol/L siBeclin-1, siATG7, or siControl (Santa Cruz Biotechnology) containing Lipofectamine 2000 (Thermo Fisher Scientific) liposomes in Opti-MEM for 4 hours. Media was changed to 10% FBS DMEM overnight, and conditioned media collection began the next day for 24 hours in serum-free DMEM. Following conditioned media collection, cell lysates were harvested for immunoblotting to confirm siBECN knockdown. Conditioned media was clarified by centrifugation and stored for no more than 2 weeks at 4°C.
Proliferation
HNSCC cells were seeded in triplicate (2,000 cells/well, 96-well plate). After cells had adhered, various experimental conditions were applied for 72-hour duration. Cell viability was assessed using CyQUANT proliferation kit (Life Technologies) according to the manufacturer's instructions. For irradiation experiments, plates were exposed to gamma radiation (J.L. Shepherd and Associates Mark I Model 68A cesium-137 source irradiator; dose rate = 2.9 Gy/min).
Invasion and migration
Cell invasion and migration was assessed using Transwell Boyden chamber system. HNSCC cells were seeded in 8-μm pore inserts for migration. For invasion, a layer of diluted (2 mg/mL) growth factor–reduced Matrigel (Corning) in DMEM was placed in the insert. HNSCC cells in serum-free media were seeded onto Matrigel. The inserts were placed in triplicate holding wells containing treatment conditions for 24 hours. Cells were also plated in experimental conditions in parallel to assess viability using CyQUANT. The number of cells that moved to other side of membrane was counted after fixation and staining with Hema 3 kit (Thermo Fisher Scientific). The number of invading or migrating cells was normalized to cell viability.
Cytokine array
Cytokine array (C5) was obtained from RayBiotech, and conditioned media from CAFs were analyzed following the manufacturer's instructions.
PCR
HNSCC cells were carboxyfluorescein diacetate succinimidyl ester (CFSE; Thermo Fisher Scientific) labeled following the manufacturer's protocol. CFSE-labeled HNSCC cells were cocultured in a 1:1 ratio with unlabeled NFs for 72 hours. Cells were harvested by trypsinization and FACS sorted using BD FACSAria III. RNA was extracted from harvested cells using TRIzol reagent (Thermo Fisher Scientific) following the manufacturer's instructions. RNA was subjected to DNAse digestion prior to cDNA preparation using the SuperScript First-Strand Synthesis System (Invitrogen). PCR products were resolved on agarose gel and imaged. Densitometric analyses were performed with ImageJ software (v1.50i). PCR array (PAHS-176ZD, Qiagen) was used to identify differences between NFs and cocultured NFs, and read using CFX96 real-time system (Bio-Rad).
Coculture proliferation assay
HNSCC cells were CFSE (Thermo Fisher Scientific) labeled following the manufacturer's protocol. CFSE-labeled HNSCC cells were cocultured in 1:1 ratio with unlabeled CAFs for 72 hours. Cells were harvested by trypsinization, and labeled HNSCC cells were counted using Attune NxT Flow Cytometer (Life Technologies).
In vivo experiments
All experiments were approved by the Institutional Review Board at the University of Kansas Medical Center (Kansas City, KS). To assess biomarker modulation by chloroquine, 100 μL of HNSCC (UM-SCC-1, 0.5 × 106) alone or admixed with CAFs (0.5 × 106) were injected into the right flank of athymic male mice (n = 3/group). After tumors were allowed to form, chloroquine was administered by oral gavage (162 mg/kg) for 3 days (24). Tissue was processed for electron microscopy.
To assess autophagy inhibition in combination with cisplatin, 100 μL of admixed HNSCC (UM-SCC-1, 0.5 × 106) and CAFs (0.5 × 106) were injected into the right flank of athymic female mice. Mice (n = 9/group) were treated with cisplatin (3 mg/kg i.p. 1×/week), chloroquine (162 mg/kg oral gavage 5 days/week), or SAR405 (50 μL intratumoral injection of 10 μmol/L SAR405 in PBS, concentration determined on the basis of in vitro IC50, 5 days/week). Tumor diameters were measured by a blinded observer using vernier calipers in two perpendicular dimensions as described previously (4). Tumors were excised and processed for electron microscopy.
To assess progression of autophagy in developing tumor, 4-NQO (100 ppm in sterile drinking water ad libitum; ref. 25) was administered for 16 weeks to C3H mice. Mice were then given sterile drinking water for 3 weeks, and tongues were excised.
The Cancer Genome Atlas data analysis
The Cancer Genome Atlas (TCGA) head and neck cancer (HNSC) cohort gene expression RNA sequencing (RNA-seq) data were downloaded using UCSC Xena browser (http://xena.ucsc.edu). Expression levels of BECN1 or MAP1LC3B were designated as high or low in relation to median expression of gene-level transcription estimates [log2(x+1) transformed RSEM normalized count]. This was matched to clinical survivorship data from TCGA HNSC phenotype data downloaded from UCSC Xena.
Statistical analysis
Data are reported as mean ± SEM. Nonparametric, two-tailed Mann–Whitney U tests were used to assess significance in all experiments and Kruskal–Wallis test for comparison of multiple groups. For in vivo study, one-way ANOVA test was employed to assess the level of significance in tumor volumes between treatment arms. For TCGA survivorship comparison, log-rank (Mantel–Cox) test assessed differences between curves. All statistical calculations were performed on GraphPad Prism software (version 6.03), with significance determined by P < 0.05.
Results
CAFs demonstrate high level of basal autophagy
Our laboratory and others have identified the significant role CAFs play to enhance HNSCC progression (4). CAF-induced progression was significantly greater than NF-induced progression. To better understand the underlying biology of CAFs, we assessed CAFs compared with NFs by electron microscopy and identified a significantly enhanced vesicular architecture of the CAFs compared with normal fibroblasts (Fig. 1A; low magnification in Supplementary Fig. S1A). The vesicular electron-dense morphology led us to question whether CAFs had a heightened level of basal autophagy compared with NFs (26). As such, we assessed autophagy marker LC3, which is enzymatically conjugated to phosphatidylethanolamine during autophagic flux to LC3-II, and the autophagy shuttling protein, p62 (8). To evaluate basal autophagy, comparative LC3-II levels were assessed between NF and CAFs with and without the autophagic flux inhibitor chloroquine. By immunoblotting (Fig. 1B; Supplementary Fig. S1B), we identified that CAFs have significantly greater LC3-II (P = 0.0286), although p62 was a bit more variable in expression. LC3-II expression was validated by visualizing and quantification of autophagic puncta by immunofluorescence of LC3 (P = 0.0094; Fig. 1C; low magnification images in Supplementary Fig. S1C). Therefore, we concluded that CAFs have an increased rate of basal autophagy compared with NFs.
CAFs have greater basal autophagic flux than NFs. A, Electron microscopy exhibits highly vesicular architecture of CAFs with heterogeneous electron dense and electron poor organelles compared with NFs. Scale bars, 0.5 μm. Graph depicts percent autophagosomes/fibroblasts relative to NFs. Autophagosomes were counted in a total of 36 fibroblasts from each group including four explants, each from HNSCC or cancer-free subjects. Error bars, ± SEM. B, Representative immunoblot of CAFs compared with NFs with and without chloroquine (CQ; 20 μmol/L for 6 hours) for LC3 protein conversion and p62. Graph depicts percent cumulative density of LC3 levels in chloroquine-treated lanes relative to NF in four explants, each from HNSCC or cancer-free subjects. LC3-II levels were normalized to β-tubulin levels. Error bars, ± SEM. C, Representative immunofluorescent of LC3 (green) puncta and Hoechst nuclear stain (blue), comparing NFs with CAFs with and without chloroquine (80 μmol/L for 2 hours; magnification, ×60). Cumulative results of LC3 puncta per cell counted by a blinded observer of at least 30 cells each of NFs and CAFs. The experiment was repeated three times using three explants, each from HNSCC or cancer-free subjects. Error bars, ± SEM.
CAFs have greater basal autophagic flux than NFs. A, Electron microscopy exhibits highly vesicular architecture of CAFs with heterogeneous electron dense and electron poor organelles compared with NFs. Scale bars, 0.5 μm. Graph depicts percent autophagosomes/fibroblasts relative to NFs. Autophagosomes were counted in a total of 36 fibroblasts from each group including four explants, each from HNSCC or cancer-free subjects. Error bars, ± SEM. B, Representative immunoblot of CAFs compared with NFs with and without chloroquine (CQ; 20 μmol/L for 6 hours) for LC3 protein conversion and p62. Graph depicts percent cumulative density of LC3 levels in chloroquine-treated lanes relative to NF in four explants, each from HNSCC or cancer-free subjects. LC3-II levels were normalized to β-tubulin levels. Error bars, ± SEM. C, Representative immunofluorescent of LC3 (green) puncta and Hoechst nuclear stain (blue), comparing NFs with CAFs with and without chloroquine (80 μmol/L for 2 hours; magnification, ×60). Cumulative results of LC3 puncta per cell counted by a blinded observer of at least 30 cells each of NFs and CAFs. The experiment was repeated three times using three explants, each from HNSCC or cancer-free subjects. Error bars, ± SEM.
Mitigation of CAF autophagy decreases HNSCC progression in vitro
We hypothesized CAF autophagy facilitates tumor progression. As we previously identified a role of CAF-secreted factors on HNSCC progression, we began by establishing autophagy-inhibited conditioned media. To do this, we inhibited autophagy using chloroquine, which inhibits lysosomal degradation of the autophagosome (27). Following chloroquine inhibition, cells were washed to remove excess extracellular chloroquine, and conditioned media were then collected. To confirm autophagy inhibition over the entirety of the conditioned media collection, CAF lysates were harvested following collection and assessed for LC3-II conversion by immunoblot as an indicator of autophagic flux inhibition (Supplementary Fig. S2A). As serum starvation could potentially induce autophagy in CAFs, we evaluated the extent of autophagy induction upon 24-hour serum starvation compared with baseline autophagy in serum-containing media. Our data demonstrate that 24 hours of serum starvation had little effect on the extent of autophagy in CAFs compared with that in 10% serum-containing media. The comparison of 10% FBS DMEM (complete media) and 24 hours of serum starvation on autophagic flux is demonstrated in Supplementary Fig. S2B. In addition, to confirm there was negligible remaining chloroquine in the conditioned media, we devised a control where chloroquine was administered for only 5 minutes, washed off, and then conditioned media was harvested in a similar fashion. This 5-minute treatment did not induce LC3-II conversion observed by immunoblot and did not influence HNSCC migration compared with vehicle control (Supplementary Fig. S2C and S2D).
We assessed autophagy-inhibited CAF conditioned media (CAF-CM) and found autophagy inhibition significantly decreased HNSCC migration, invasion, and proliferation (Fig. 2A; Supplementary Fig. S2E). To validate our observations, we inhibited autophagy in CAFs by siRNA knockdown of Beclin-1, an upstream regulator of autophagy (28). Successful knockdown at conditioned media endpoint was confirmed by immunoblot analysis (Fig. 2B; Supplementary Fig. S2F). Again, significant reductions in HNSCC migration, invasion, and proliferation were observed following knockdown of autophagy (Fig. 2C). Furthermore, these findings were validated by siRNA knockdown of Atg7 (29), which demonstrated a significant decrease in HNSCC migration and invasion (Supplementary Fig. S2G and S2H). These findings demonstrate that CAF autophagy–dependent secretion promotes HNSCC progression.
CAF autophagy inhibition significantly decreases CAF-facilitated HNSCC progression in vitro. A, CAFs pretreated with chloroquine (CQ; 20 μmol/L for 6 hours) were washed extensively to remove excess chloroquine and then conditioned media (CM) was collected with and without chloroquine pretreatment. HNSCC (OSC19) migration, invasion, and proliferation are significantly reduced in CAF autophagy–inhibited conditioned media. Graph depicts cumulative results from three independent experiments, including triplicate treatments, using CAFs derived from two patients with HNSCC. Migration and invasion experiments were normalized to cell viability. Error bars, ± SEM. B, Representative immunoblot confirming Beclin-1 knockdown throughout CAF-CM collection. C, Significant reduction observed in HNSCC migration, invasion, and proliferation with Beclin-1 knockdown CAF-CM (siBECN) compared with control siRNA (siCon). Graph depicts combined results of at least three trials per experiments plated in triplicate using at least two different CAF patient samples. Migration and invasion experiments normalized to cell viability. Error bars, ± SEM. VC, vehicle control.
CAF autophagy inhibition significantly decreases CAF-facilitated HNSCC progression in vitro. A, CAFs pretreated with chloroquine (CQ; 20 μmol/L for 6 hours) were washed extensively to remove excess chloroquine and then conditioned media (CM) was collected with and without chloroquine pretreatment. HNSCC (OSC19) migration, invasion, and proliferation are significantly reduced in CAF autophagy–inhibited conditioned media. Graph depicts cumulative results from three independent experiments, including triplicate treatments, using CAFs derived from two patients with HNSCC. Migration and invasion experiments were normalized to cell viability. Error bars, ± SEM. B, Representative immunoblot confirming Beclin-1 knockdown throughout CAF-CM collection. C, Significant reduction observed in HNSCC migration, invasion, and proliferation with Beclin-1 knockdown CAF-CM (siBECN) compared with control siRNA (siCon). Graph depicts combined results of at least three trials per experiments plated in triplicate using at least two different CAF patient samples. Migration and invasion experiments normalized to cell viability. Error bars, ± SEM. VC, vehicle control.
Autocrine and paracrine IL6 and IL8 regulate CAF autophagy
Recently, a growing appreciation for the role of the autophagic machinery in cellular secretion has been reported (30). Our data indicate an autophagy-regulated secretome that influences HNSCC progression in vitro. A cytokine array was used to identify changes in CAF-secreted factors following autophagy inhibition. By siRNA knockdown of Beclin-1, we observed a significant decrease in a number of CAF-secreted cytokines (Fig. 3A), most notably being IL8 and IL6. Reconstitution of IL6 and IL8 in autophagy-impaired CAF-CM restored CAF-CM–induced HNSCC migration (P < 0.0001; Fig. 3B). Therefore, autophagy, in part, regulates the secretion of IL6 and IL8 from CAFs, facilitating HNSCC migration.
CAFs secrete IL6 and IL8 through autophagy, which further induce autophagy in an autocrine pathway. A, Relative density of top four cytokines recognized on cytokine array of CAF-CM with Beclin-1 siRNA knockdown (siBECN) or control siRNA (siCon). B, Reconstitution of HNSCC (UM-SCC-1) migration in Beclin-1 knockdown CAF-CM with recombinant IL6 (10 ng/mL) and IL8 (80 ng/mL); data cumulative of two trials plated in duplicate using different CAF patient samples. C, Representative immunofluorescence of NFs treated with vehicle control (VC; water), IL6 (10 ng/mL), or IL8 (80 ng/mL) for 24 hours with and without chloroquine (CQ; 80 μmol/L for last 2 hours of cytokine treatment; magnification, ×20). Graph depicts cumulative results of LC3 puncta per cell counted by a blinded observer of at least 30 cells per experimental arm in three separate experiments. Error bars, ±SEM. D and E, Representative immunoblot of NFs treated with IL6 (10 ng/mL; D) or IL8 (80 ng/mL; E) for 24 hours with and without chloroquine (20 μmol/L for last 6 hours of treatment).
CAFs secrete IL6 and IL8 through autophagy, which further induce autophagy in an autocrine pathway. A, Relative density of top four cytokines recognized on cytokine array of CAF-CM with Beclin-1 siRNA knockdown (siBECN) or control siRNA (siCon). B, Reconstitution of HNSCC (UM-SCC-1) migration in Beclin-1 knockdown CAF-CM with recombinant IL6 (10 ng/mL) and IL8 (80 ng/mL); data cumulative of two trials plated in duplicate using different CAF patient samples. C, Representative immunofluorescence of NFs treated with vehicle control (VC; water), IL6 (10 ng/mL), or IL8 (80 ng/mL) for 24 hours with and without chloroquine (CQ; 80 μmol/L for last 2 hours of cytokine treatment; magnification, ×20). Graph depicts cumulative results of LC3 puncta per cell counted by a blinded observer of at least 30 cells per experimental arm in three separate experiments. Error bars, ±SEM. D and E, Representative immunoblot of NFs treated with IL6 (10 ng/mL; D) or IL8 (80 ng/mL; E) for 24 hours with and without chloroquine (20 μmol/L for last 6 hours of treatment).
Previous reports have demonstrated a role for IL6 in promoting autophagy (31). As such, we investigated the autocrine effects of IL6 and IL8 on autophagy in NFs. Interestingly, we observed a significant increase in LC3 puncta per cell in NFs treated with IL6 or IL8 (P < 0.0001; Fig. 3C; Supplementary Fig. S3A). This finding was validated by assessing LC3-II levels using immunoblotting (Fig. 3D and E). Both IL6 and IL8 induced LC3-II, indicating an induction of autophagy. It has been well established that HNSCC secretes IL6 and IL8 (32), and this may be one method by which the tumor modulates the surrounding stroma. Neutralizing antibodies to IL6 and IL8 mitigated HNSCC induction of autophagy in NFs (Supplementary Fig. S3B). This induction of autophagy in fibroblasts may occur through paracrine stimulation of IL6 and IL8 by HNSCC and autocrine factors from activated fibroblasts.
HNSCC factors induce fibroblast autophagy
With CAFs having a significantly greater level of basal autophagy than NFs, we sought to identify whether HNSCC promotes CAF autophagy. We cocultured NFs with HNSCC and observed an increase in LC3 puncta per cell (Fig. 4A; Supplementary Fig. S4A). Other studies in our laboratory have identified a role of HNSCC-secreted bFGF in inducing CAF activation through FGFR signaling. Therefore, we hypothesized that bFGF may be a causative factor in the induction of CAF autophagy. There was a significant increase in LC3 puncta in NFs treated with bFGF (P < 0.0001; Fig. 4B; Supplementary Fig. S4B). This indicated HNSCC secreted bFGF, as well as IL6 and IL8 promoted CAF autophagy. Therefore, to confirm this induction, we cocultured HNSCC and NF with siRNA to bFGF in the HNSCC cells. We demonstrate that inhibiting HNSCC-secreted bFGF limits HNSCC-induced fibroblast autophagy (Fig. 4C; Supplementary Fig. S4C). In addition, treatment of NFs with recombinant bFGF also increased IL6 and IL8 secretion (Supplementary Fig. S4D).
HNSCC induces CAF autophagy through paracrine secretion of bFGF. A, Representative immunofluorescence of NFs in a 1:1 coculture with HNSCC (HN5) with and without chloroquine (CQ; 80 μmol/L for 2 hours) with cytokeratin 14 HNSCC label (red), LC3 (green), and Hoescht (blue) nuclear stain (magnification, ×20). Graph depicts cumulative results of LC3 puncta per cell counted by a blinded observer of chloroquine-treated wells in at least 20 cells per group. Results are cumulative of three experiments using two different NF patient samples and presented relative to NFs. B, Representative immunofluorescence of NFs with and without bFGF (100 ng/mL for 24 hours) with chloroquine flux inhibition (80 μmol/L for final 2 hours of bFGF treatment), LC3 (green), Hoescht nuclear (blue; magnification, ×20). Graph depicts cumulative results of LC3 puncta per cell counted by blinded observer of NF ± bFGF + chloroquine in at least 20 cells per group, and results are cumulative of three experiments using two different NF patient samples. C, Representative immunofluorescence of NFs, or NFs cocultured with either control siRNA–transfected HNSCC (HN5; siCon) or bFGF siRNA–transfected HNSCC (sibFGF) in a 1:1 ratio. Chloroquine (80 μmol/L for 2 hours) was used to inhibit flux. LC3 (green) and Hoescht (blue) are visualized at ×20 magnification. Graph depicts cumulative results of LC3 puncta counted per cell in chloroquine-treated wells of at least 39 cells per group and presented relative to NFs alone. D, Representative immunoblot of bFGF (100 ng/mL for 24 hours) with and without chloroquine (20 μmol/L for 6 hours)-treated NFs of LC3 and phospho-p70S6K (Thr389). VC, vehicle control.
HNSCC induces CAF autophagy through paracrine secretion of bFGF. A, Representative immunofluorescence of NFs in a 1:1 coculture with HNSCC (HN5) with and without chloroquine (CQ; 80 μmol/L for 2 hours) with cytokeratin 14 HNSCC label (red), LC3 (green), and Hoescht (blue) nuclear stain (magnification, ×20). Graph depicts cumulative results of LC3 puncta per cell counted by a blinded observer of chloroquine-treated wells in at least 20 cells per group. Results are cumulative of three experiments using two different NF patient samples and presented relative to NFs. B, Representative immunofluorescence of NFs with and without bFGF (100 ng/mL for 24 hours) with chloroquine flux inhibition (80 μmol/L for final 2 hours of bFGF treatment), LC3 (green), Hoescht nuclear (blue; magnification, ×20). Graph depicts cumulative results of LC3 puncta per cell counted by blinded observer of NF ± bFGF + chloroquine in at least 20 cells per group, and results are cumulative of three experiments using two different NF patient samples. C, Representative immunofluorescence of NFs, or NFs cocultured with either control siRNA–transfected HNSCC (HN5; siCon) or bFGF siRNA–transfected HNSCC (sibFGF) in a 1:1 ratio. Chloroquine (80 μmol/L for 2 hours) was used to inhibit flux. LC3 (green) and Hoescht (blue) are visualized at ×20 magnification. Graph depicts cumulative results of LC3 puncta counted per cell in chloroquine-treated wells of at least 39 cells per group and presented relative to NFs alone. D, Representative immunoblot of bFGF (100 ng/mL for 24 hours) with and without chloroquine (20 μmol/L for 6 hours)-treated NFs of LC3 and phospho-p70S6K (Thr389). VC, vehicle control.
To further determine the mechanism of fibroblast autophagy by HNSCC, we used PCR-based microarray analyses of various molecular regulators, including transcription factors, to identify signaling pathways upregulated when NFs were cocultured with HNSCC. SOX2 and FGFR2 were highly upregulated in cocultured NFs compared with NFs alone (Supplementary Fig. S4E). Intriguingly, SOX2 transcriptionally represses mTOR, thereby inducing autophagy (33). SOX2 is a downstream target of bFGF. Therefore, we interrogated the role of FGFR signaling in HNSCC-mediated SOX2 induction in NFs using the pan-FGFR inhibitor AZD-4547 (34). We validated the microarray finding by RT-PCR and observed an induction of SOX2 in NFs upon coculture with HNSCC, which corresponded with a decrease in mTOR mRNA levels (Supplementary Fig. S4F and S4G). Furthermore, FGFR inhibition restored SOX2 levels (Supplementary Fig. S4G). Furthermore, we found that stimulation of fibroblasts with recombinant bFGF decreased phospho-p70S6K levels, which is a signaling intermediate downstream of mTOR (Fig. 4D). This corresponded with an increase in LC3-II cleavage in NFs. HNSCC conditioned media activation of phospho-STAT3, a transcription factor downstream of FGFR, and LC3-II protein levels were inhibited with FGFR inhibition (Supplementary Fig. S4H). These results suggest that HNSCC induces autophagy in CAFs through paracrine signaling, including bFGF, IL6, and IL8.
Mitigation of HNSCC autophagy decreases HNSCC progression in vitro
In HNSCC, increased levels of the autophagy marker LC3 is associated with reduced overall survival (35). However, no studies have examined the therapeutic potential of autophagy inhibition in HNSCC. Chloroquine and its clinically relevant derivative hydroxychloroquine (36) have been assessed in a number of cancer types, but not HNSCC. We identified that chloroquine significantly inhibited the proliferation, migration (P < 0.0001), and invasion (P = 0.0004) of HNSCC cells (Fig. 5A–C; Supplementary Fig. S5A). Furthermore, chloroquine potentiated the effects of cisplatin in combination treatment (Fig. 5D). In addition, the combination of chloroquine and radiation (3 Gy) also demonstrated a potentiated effect (Supplementary Fig. S5B). These data indicate the therapeutic potential of direct autophagy inhibition in HNSCC.
Autophagy inhibition significantly reduces HNSCC progression observed by in vitro models. A, Chloroquine (CQ) reduces HNSCC (UM-SCC-1) proliferation with IC50 = 11.51 μmol/L over 72 hours. B and C, Chloroquine mitigates HNSCC (UM-SCC-1) migration (B) and invasion (C) at IC50 concentration. Migration and invasion normalized to cell viability, and graph depicts three experiments plated in duplicate. D, Combination of chloroquine (IC50) and cisplatin (4 μmol/L) significantly reduces HNSCC (UM-SCC-1) proliferation over 72 hours; graph depicts three experiments plated in triplicate. E, CFSE-labeled HNSCC (HN5) proliferation over 72 hours cocultured in 1:1 ratio with CAFs with and without pretreatment of Beclin-1 siRNA knockdown, and with and without chloroquine (IC50) throughout 72-hour coculture. Graph depicts results of two separate experiments using two different CAF patient samples, plated in duplicate. F, SAR405 reduces HNSCC proliferation with IC50 = 7.92 μmol/L over 72 hours; graph depicts three experiments plated in triplicate. G, Representative immunoblot of increasing doses of SAR405 with and without chloroquine flux inhibition (20 μmol/L for 6 hours). Experiment was repeated twice. H, Representative immunoblot of 1.0 μmol/L SAR405 on CAF with and without chloroquine flux inhibition (20 μmol/L for 6 hours). Experiment was repeated twice. All error bars, ±SEM.
Autophagy inhibition significantly reduces HNSCC progression observed by in vitro models. A, Chloroquine (CQ) reduces HNSCC (UM-SCC-1) proliferation with IC50 = 11.51 μmol/L over 72 hours. B and C, Chloroquine mitigates HNSCC (UM-SCC-1) migration (B) and invasion (C) at IC50 concentration. Migration and invasion normalized to cell viability, and graph depicts three experiments plated in duplicate. D, Combination of chloroquine (IC50) and cisplatin (4 μmol/L) significantly reduces HNSCC (UM-SCC-1) proliferation over 72 hours; graph depicts three experiments plated in triplicate. E, CFSE-labeled HNSCC (HN5) proliferation over 72 hours cocultured in 1:1 ratio with CAFs with and without pretreatment of Beclin-1 siRNA knockdown, and with and without chloroquine (IC50) throughout 72-hour coculture. Graph depicts results of two separate experiments using two different CAF patient samples, plated in duplicate. F, SAR405 reduces HNSCC proliferation with IC50 = 7.92 μmol/L over 72 hours; graph depicts three experiments plated in triplicate. G, Representative immunoblot of increasing doses of SAR405 with and without chloroquine flux inhibition (20 μmol/L for 6 hours). Experiment was repeated twice. H, Representative immunoblot of 1.0 μmol/L SAR405 on CAF with and without chloroquine flux inhibition (20 μmol/L for 6 hours). Experiment was repeated twice. All error bars, ±SEM.
To better delineate the effects of autophagy inhibition in the tumor cells compared with autophagy inhibition in CAFs, we established a coculture assay of both cell types by labeling HNSCC cells with CFSE. Following a 72-hour coculture, HNSCC proliferation was determined by flow cytometry of labeled cells. This allowed for individual assessment of chloroquine treatment and/or Beclin-1 knockdown in CAFs (Fig. 5E). As observed with our previous proliferation study, chloroquine treatment of cancer cells alone significantly reduced HNSCC proliferation (P = 0.0002). Addition of CAFs significantly increased HNSCC proliferation (P = 0.004), which was ameliorated by both knockdown of Beclin-1 in CAFs (P = 0.029) and the use of chloroquine (P = 0.029). This further established the role of CAF autophagy in promoting HNSCC proliferation, which can be therapeutically mitigated.
In the clinic, high doses of chloroquine are required to inhibit autophagy, limiting its clinical utility (37). To overcome this, recent understanding of the autophagic pathway has led to the development of highly specific small-molecule inhibitors of autophagy, such as SAR405 (13), which inhibits Vps34, a class III PI3K unique to autophagy (38).
As such, we investigated the therapeutic potential of SAR405 in HNSCC cell lines. Although a relatively high IC50 (7.92 μmol/L) was required to limit HNSCC proliferation (Fig. 5F; Supplementary Fig. S5C), SAR405 significantly inhibited LC3-II conversion at a low concentration (0.1 μmol/L; Fig. 5G). In addition, SAR405 did not significantly reduce CAF proliferation (Supplementary Fig. S5D); however, CAF autophagy as assessed by LC3-II conversion was significantly limited at low concentration of inhibitor (1 μmol/L; Fig. 5H). Thus, we assessed the therapeutic relevance of limiting HNSCC progression in combination therapy using the standard of care, cisplatin, with SAR405. The low concentration of SAR405 (1 μmol/L) necessary to limit autophagic flux enhanced the effect of cisplatin (4 μmol/L) in combination therapy (P < 0.0001; Supplementary Fig. S5E).
Combination therapy significantly reduces HNSCC tumor volume
Given the promising results observed in vitro, we sought to assess autophagy inhibition in murine models of HNSCC. To better understand autophagy through the progression of this disease, we used a 4-NQO carcinogen–induced model of HNSCC, and assessed the tongues in various stages of disease progression (25). By interrogating this model, we were able to observe that scant LC3 puncta could be identified in the normal tongue epithelium, but there were significant LC3 puncta in dysplastic and cancerous tissues (Fig. 6A). This further supported the therapeutic potential of targeting autophagy in HNSCC.
Autophagy inhibition significantly reduces HNSCC progression in vivo. A, Representative immunofluorescence images of LC3 (green) puncta in 4-NQO–induced HNSCC model progression from normal tongue epithelium, low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion (HSIL), carcinoma in situ (CIS), and invasive squamous cell carcinoma (Invasive SCC) at ×20 magnification (top), and ×60 magnification (bottom). Arrowheads, LC3 puncta accumulation. Blue, nuclear stain. White box, area of 60× image. B, Representative electron microscopy images of sections from CAFs and HNSCCs (UM-SCC-1) injected subcutaneously into nude male mice. Chloroquine (CQ; 162 μg/mL oral gavage) treatment significantly enhanced autophagosome accumulation. C, Autophagosomes per cell were counted by blinded observer from at least 48 cells from two different mice per treatment group. D, Autophagy inhibition potentiates standard-of-care therapy. 1:1 admixture of CAFs and HNSCCs (UM-SCC-1) was injected subcutaneously in nude female mice. Mice were treated with cisplatin (3 mg/kg, i.p. 1×/week), chloroquine (162 mg/kg oral gavage, 5 days/week), or SAR405 (50-μL intratumoral injection of 10 μmol/L SAR405 in PBS; n = 9/group). Tumor volumes were assessed by a blinded observer. E, Autophagosomes counted by blinded observer of at least 13 cells from two different mice per treatment group of vehicle control, SAR405, cisplatin, and combination cisplatin and SAR405. VC, vehicle control.
Autophagy inhibition significantly reduces HNSCC progression in vivo. A, Representative immunofluorescence images of LC3 (green) puncta in 4-NQO–induced HNSCC model progression from normal tongue epithelium, low-grade squamous intraepithelial lesion (LSIL), high-grade squamous intraepithelial lesion (HSIL), carcinoma in situ (CIS), and invasive squamous cell carcinoma (Invasive SCC) at ×20 magnification (top), and ×60 magnification (bottom). Arrowheads, LC3 puncta accumulation. Blue, nuclear stain. White box, area of 60× image. B, Representative electron microscopy images of sections from CAFs and HNSCCs (UM-SCC-1) injected subcutaneously into nude male mice. Chloroquine (CQ; 162 μg/mL oral gavage) treatment significantly enhanced autophagosome accumulation. C, Autophagosomes per cell were counted by blinded observer from at least 48 cells from two different mice per treatment group. D, Autophagy inhibition potentiates standard-of-care therapy. 1:1 admixture of CAFs and HNSCCs (UM-SCC-1) was injected subcutaneously in nude female mice. Mice were treated with cisplatin (3 mg/kg, i.p. 1×/week), chloroquine (162 mg/kg oral gavage, 5 days/week), or SAR405 (50-μL intratumoral injection of 10 μmol/L SAR405 in PBS; n = 9/group). Tumor volumes were assessed by a blinded observer. E, Autophagosomes counted by blinded observer of at least 13 cells from two different mice per treatment group of vehicle control, SAR405, cisplatin, and combination cisplatin and SAR405. VC, vehicle control.
To assess chloroquine dosing, we conducted a small pilot study of HNSCC alone or HNSCC and CAFs inoculated in a 1:1 admixture subcutaneously in athymic mice. Chloroquine caused the accumulation of autophagosomes in the tumor cells, which was observed to significantly increase in the CAF and HNSCC cohort (P < 0.0001; Fig. 6B and C, and more examples of counted autophagosomes in Supplementary Fig. S6).
To assess autophagy inhibition in combination with current standard of care, cisplatin, HNSCCs were inoculated in a 1:1 admixture subcutaneously in athymic mice. Mice were treated with the autophagy inhibitors chloroquine or SAR405 and/or cisplatin. With the addition of the autophagy inhibitor SAR405 to cisplatin therapy, there was a significant reduction in tumor volume when compared with SAR405 alone, cisplatin alone, or untreated mice (Fig. 6D). Chloroquine and cisplatin combination treatment reduced tumor volume, but the reduction was not as significant as the more specific inhibitor, SAR405. In addition, by electron microscopy, we were able to observe the significant reduction in autophagosomes per cell due to SAR405 therapy (Fig. 6E). This indicates the use of targeted autophagy inhibition will potentiate the efficacy of current therapy.
Therefore, to understand the role of autophagy in patient tissue, we assessed patient sections for autophagosomes. Pathologic IHC identification of autophagy markers in patient tissue has given mixed results, with some studies indicating a favorable outcome with increased expression of autophagic markers and others indicating an unfavorable outcome (35, 39). We identify that LC3 puncta were significantly increased in HNSCC fibroblasts as compared with normal specimens from cancer-free patients (Fig. 7A).
Autophagy is overexpressed in patient fibroblasts, and overexpression of the autophagy initiator BECN1 is correlated with poor survival. A, Representative immunofluorescence of LC3 (green), vimentin (red), or Hoescht (blue) in normal tonsil from cancer-free patients and HNSCCs. Graph depicts LC3 puncta/fibroblast (as determined by vimentin positivity in spindle-shaped cells) of 12 fibroblasts from 10 each of cancer-free and HNSCC patients (120 fibroblasts per group) and normalized to normal tonsil. Error bars, ±SEM. B, LC3 (MAP1LC3B) overexpression does not significantly correlate with survival. Data downloaded from TCGA HNSC cohort and stratified by median MAP1LC3B RNA expression (RSEM). High expression was determined by primary tumor patient samples that had greater expression than median (log2 expression RSEM; high expression n = 283; low expression n = 283). C, BECN1 overexpression correlates with poor patient survival. Data downloaded from TCGA HNSC cohort and stratified by median BECN1 RNA expression (RSEM). High expression was determined by primary tumor patient samples that had greater expression than median (0.00525 log2 expression RSEM; high expression n = 283; low expression n = 283). D, Schematic representation of the mechanism of autophagy induction in CAFs by HNSCC that facilitates HNSCC progression.
Autophagy is overexpressed in patient fibroblasts, and overexpression of the autophagy initiator BECN1 is correlated with poor survival. A, Representative immunofluorescence of LC3 (green), vimentin (red), or Hoescht (blue) in normal tonsil from cancer-free patients and HNSCCs. Graph depicts LC3 puncta/fibroblast (as determined by vimentin positivity in spindle-shaped cells) of 12 fibroblasts from 10 each of cancer-free and HNSCC patients (120 fibroblasts per group) and normalized to normal tonsil. Error bars, ±SEM. B, LC3 (MAP1LC3B) overexpression does not significantly correlate with survival. Data downloaded from TCGA HNSC cohort and stratified by median MAP1LC3B RNA expression (RSEM). High expression was determined by primary tumor patient samples that had greater expression than median (log2 expression RSEM; high expression n = 283; low expression n = 283). C, BECN1 overexpression correlates with poor patient survival. Data downloaded from TCGA HNSC cohort and stratified by median BECN1 RNA expression (RSEM). High expression was determined by primary tumor patient samples that had greater expression than median (0.00525 log2 expression RSEM; high expression n = 283; low expression n = 283). D, Schematic representation of the mechanism of autophagy induction in CAFs by HNSCC that facilitates HNSCC progression.
Finally, we assessed the TCGA database for expression levels of known autophagy regulators and their correlation with patient survival. Although expression of LC3 (MAP1LC3B) did not correlate with differences in patient survival (Fig. 7B), there was a profound decrease in survival when the upstream initiator of autophagy, Beclin-1 (BECN1), was overexpressed (P = 0.0004; Fig. 7C). In vitro, overexpression of Beclin-1 is a potent inducer of autophagy (28). These data, combined with our combination therapy data, indicate that autophagy inhibition with standard of care would likely improve outcomes for patients with HNSCC. In summary, we have identified a role for secretory autophagy in the supporting stromal CAFs, which enhances HNSCC progression (Fig. 7D).
Discussion
Reciprocal communication between cancer cells and the tumor microenvironment sustains and enables cancer progression. For the past decade, the contribution of stromal cell–secreted factors to the progression of cancer has been appreciated (3); however, the underlying secretory machinery involved remains enigmatic. We observed that CAFs in culture appear phenotypically different from NFs from the same anatomic location, displaying a highly vesicular architecture. This led us to question what fundamental biologic mechanisms account for the cancer-promoting secretory profile. In this study, we observe that primary patient-derived CAFs sustain an increased level of basal autophagy as compared with NFs from cancer-free patients. This is the first observation of enhanced fibroblast autophagy from primary CAFs derived from patient stroma, and corroborates recent observations of breast cancer cells inducing autophagy in skin fibroblasts and a Drosophila melanogaster tumor model where microenvironmental autophagy was observed (40, 41).
Although autophagy is conventionally a degradation pathway, recent reports of a role for autophagy in unconventional cellular secretion (30) prompted us to investigate the role for CAF autophagy in secreting tumor-promoting factors. By collecting CAF conditioned media under autophagy inhibition through both upstream knockdown of Beclin-1 and Atg7, and downstream lysosomal inhibition using chloroquine, we observed significant phenotypic differences in cancer progression in vitro. This indicated that CAF autophagy modulates secreted factors important for tumor progression. Using a cytokine array, we identified IL6 and IL8 as being autophagy modulated in their secretion, consistent with previous observations of autophagy-controlled secretion of these factors in other systems. IL6 and IL8 are elevated systemically in patients with HNSCC (32), have been associated with resistance to targeted therapy (42), and are known to be secreted from stromal fibroblasts found in a number of cancer types (43). This is the first report linking CAF tumor-promoting cytokine secretion with autophagy. Interestingly, IL6 and IL8 are among a plethora of factors associated with the senescence-associated secretory phenotype (44). We observed that primary CAF lines demonstrate rapid proliferation despite autophagy. Our data also demonstrate that knockdown of Beclin-1 mitigated secretory autophagy and consequently reduced the levels of IL6 and IL8 in CAFs. Thus, reduced IL6 and IL8 secretion on inhibition of autophagy coupled with the rapid proliferation of these fibroblasts leads us to the conclusion that this is not a replication-induced senescent phenomenon. Recent reports indicate a role for autophagy-controlled amino acid secretion by microenvironment cells of pancreatic cancer (45). Our data corroborate this finding and unveil a role for tumor microenvironment autophagy in altering secreted factors.
By coculturing NFs from cancer-free patients with HNSCC cells, an increase in fibroblast autophagy was observed. The list of known physiologic autophagy inducers is short (46), with amino acid/glucose starvation and reactive oxygen species being some of the only understood natural inducers. HNSCC metabolically outcompeting CAFs for extracellular nutrients may play a role in inducing CAF autophagy (47). However, assessment occurred in a short 24-hour time frame in high-glucose DMEM, which likely indicates this is not a starvation or nutrient depletion phenomenon. We identified that IL6, IL8, and bFGF, known HNSCC-secreted factors, are all at least in part responsible for CAF autophagy. This corroborates recent findings in a Drosophila tumor model that IL6 secretion from the tumor promotes stromal autophagy (41). We demonstrate HNSCC-secreted bFGF activates STAT3 and induces the transcription of SOX2, which inhibits mTOR transcription (33). As mTOR represses autophagy (48), SOX2 inhibition of mTOR conferred increased autophagy in the fibroblasts in our system. The dynamics of CAF autophagy regulation are even more complex when taken into account with our observation that autophagy-promoting IL6 and IL8 are also secreted through autophagic machinery.
There is extensive evidence in literature demonstrating that both chemotherapeutic drugs and radiation promote cytoprotective autophagy in tumor cells (49). Despite multiple studies in a variety of cancer cell types, investigations into the therapeutic potential of autophagy inhibition in HNSCC is lacking (50). Our results provide evidence for the first time of autophagy inhibition as having therapeutic potential in HNSCC. The largest limitations to autophagy treatment in clinical trials has been the high doses required of chloroquine, which may fail to achieve intratumoral concentrations sufficient to limit autophagy, and a lack of accepted methodology for monitoring autophagy in patients' tumors to confirm successful inhibition (51). As such, a large body of research is investigating potential small-molecule inhibitors of autophagy, and few have been discovered (52). SAR405 is one such inhibitor that specifically targets PI3K class III, of which the only recognized member is Vps34, an upstream autophagy-regulating kinase. We observed that low doses of SAR405 (1.0 μmol/L or less) were sufficient to limit HNSCC and CAF autophagy in vitro. This may provide a feasible therapeutic alternative to chloroquine or hydroxychloroquine for clinical autophagy inhibition. The potentiated effects of HNSCC standard of care, cisplatin, with SAR405 in vivo were profound and give hope for future combinational therapy.
In summary, we demonstrate enhanced basal autophagy in CAFs that facilitates the secretion of tumor-promoting factors, notably IL6 and IL8. HNSCC paracrine secretion of IL6, IL8, and bFGF is at least in part responsible for the promotion of CAF autophagy, which is further maintained through IL6 and IL8 autocrine feedback. Amelioration of HNSCC autophagy by both chloroquine and SAR405 give indication to the potential therapeutic value of combinatorial targeting of autophagy with HNSCC standard of care.
Disclosure of Potential Conflicts of Interest
Y. Shnayder has ownership interest (including patents) in and is a consultant/advisory board member for Hylapharm. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: J. New, O. Tawfik, S. Anant, S.M. Thomas
Development of methodology: J. New, O. Tawfik, W.-X. Ding, S. Anant, S.M. Thomas
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. New, M. Ananth, S. Alvi, M. Thornton, L.R. Werner, O. Tawfik, Y. Shnayder, K. Kakarala, T.T. Tsue, S.M. Thomas
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. New, L. Arnold, M. Ananth, M. Thornton, O. Tawfik, H. Dai, W.-X. Ding, S. Anant, S.M. Thomas
Writing, review, and/or revision of the manuscript: J. New, M. Ananth, S. Alvi, O. Tawfik, H. Dai, Y. Shnayder, K. Kakarala, T.T. Tsue, D.A. Girod, S. Anant, S.M. Thomas
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Arnold, L.R. Werner, S.M. Thomas
Study supervision: Y. Shnayder, D.A. Girod, S.M. Thomas
Acknowledgments
We acknowledge support from the University of Kansas (KU) Cancer Center's Biospecimen Repository Core Facility staff for helping obtain human specimens and for performing histologic work.
Grant Support
This study was supported in part by the University of Kansas Cancer Center under CCSG P30CA168524; an NIH Clinical and Translational Science Award grant (UL1TR000001, formerly UL1RR033179) awarded to the University of Kansas Medical Center and an internal Lied Basic Science Grant Program of the KUMC Research Institute (to S. M. Thomas); grant R01CA182872 (to S. Anant); grants R01AA020518, U01AA024733, P20GM103549, and P30GM118247 (to W.-X. Ding); and the KUMC Biomedical Research Training Program (to J. New). Immunofluorescence microscopy was supported by the Smith Intellectual and Developmental Disabilities Research Center (NIH U54 HD 090216). The Flow Cytometry Core Laboratory is sponsored, in part, by the NIH/NIGMS COBRE grant P30GM103326. The KUMC Electron Microscopy Research Lab facility is supported, in part, by NIH COBRE grants P20GM104936 and S10RR027564.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.