|Home | About | Journals | Submit | Contact Us | Français|
Oral and oro-pharyngeal squamous cell carcinomas (OSCC) exhibit surface breach, and recent studies have demonstrated bacterial contamination of primary and metastatic OSCC. Increasing concentrations of inflammatory products, such as interleukin (IL)-6 and vascular endothelial growth factor (VEGF), correlate with, and contribute to, cancer progression, but their regulation in OSCC is poorly understood. We hypothesized that monocyte-lineage cells and bacterial contamination may contribute important inflammatory products that can support OSCC progression. We found that relative to non-specific chronic mucositis, oral carcinoma-in-situ/superficially-invasive OSCC contained more monocyte-lineage cells. In vitro, we used lipopolysaccharide (LPS) to model bacterial contamination, and evaluated the effects of oral and oropharyngeal (O)SCC-monocyte interactions and of LPS on OSCC cells and on the production of IL-6 and VEGF. OSCC cell lines varied in constitutive cytokine and chemokine production, and OSCC-monocyte interactions in the absence of LPS stimulated IL-6 and VEGF occasionally, while LPS-OSCC-monocyte interactions were always strongly stimulatory. Importantly, LPS independently stimulated some OSCC lines to secrete monocyte-dendritic cell chemoattractants CCL2 and/or CCL20, as well as IL-6 and/or VEGF. While very little constitutive Y705-STAT3 phosphorylation (pY705-STAT3) was detectable in HNSCC lines, IL-6 rapidly induced pY705-STAT3 in OSCC lines that produced little IL-6 constitutively. Supernatants from LPS-OSCC-monocyte co-cultures always rapidly and strongly activated STAT3, which was partly due to IL-6. We conclude that monocytes and microbial contamination have the potential to contribute to OSCC progression, as STAT3 activation in OSCC cells depends on soluble factors, which are consistently available through LPS-OSCC-monocyte interactions.
Progressing head and neck squamous cell carcinomas (HNSCC), which include upper aero-digestive tract oral and oropharyngeal (O)SCC, are biologically heterogeneous, and the factors contributing to their growth are poorly understood. There is frequent loss of surface integrity in OSCC, with Gram-positive and Gram-negative oral and enteric bacteria found in ~75% of cervical lymph nodes containing metastatic OSCC . More recently, viable bacteria were isolated from both superficial and deep portions of OSCC , revealing that the tumor microenvironment is well suited for bacterial survival. Bacterial products, such as cell wall product of Gram-negative bacteria lipopolysaccharide (LPS), play a significant role in a variety of pathologic conditions due to bacterial association with vast mucosal and skin surfaces, and LPS stimulates many cell types . The role of LPS in HNSCC pathogenesis is not known.
The majority of OSCC are associated with dense infiltrates of leukocytes, including monocytes, macrophages and dendritic cells (DC) [4, 5]. Monocytes and myeloid DC can be recruited out of the peripheral blood by the chemokine CCL2 (MCP-1) , and SCC cells were shown to attract monocytes by making CCL2 . Monocytes can differentiate into macrophages or myeloid DC, depending upon available soluble factors . In advanced upper aero-digestive tract SCC, the presence of high numbers of CD68+ cells (which includes monocytes, macrophages, and a subset of DC) was reported to be strongly associated with lymph node metastasis , while the numbers of tumor-associated DC, in general, were lower in more advanced lesions . However, little is known about monocyte-lineage cells in pre-invasive or superficially-invasive OSCC, or about the regulation of factors that contribute to monocyte-lineage cell attraction and differentiation in OSCC.
Similar to other carcinomas, a variety of factors are produced in HNSCC [7, 10–12], and the importance of IL-6 is well documented. High serum IL-6 levels directly correlate with poor prognosis in cancer patients [11, 13]. IL-6 activates the signal transducer and activator of transcription-3 (STAT3) in keratinocytes , HNSCC and other normal and cancer cell types, leading to protection from apoptosis and to proliferation [15–17]. Among the effects of IL-6 on monocyte-lineage cells, of particular interest are that it up-regulates monocyte-DC chemoattractant CCL2  and promotes monocyte differentiation into macrophages. Further, IL-6-induced STAT3 activation in macrophages decreases activation of STAT1 required for classical IFN-gamma responses , thus preventing anti-tumor macrophage cellular cytotoxicity, which depends on IFN-gamma . In addition, maturation of DC exposed to tumor cells with high STAT3 activity is compromised , and DC maturation is important for effective immune responses. Finally, IL-6 can also stimulate production of vascular endothelial growth factor (VEGF) , the expression of which is regulated by STAT3 [22, 23]. VEGF is produced in most cancers and is responsible for tumor angiogenesis and growth , and it also inhibits precursor cell differentiation into DC . Overall, IL-6 and VEGF are strongly implicated in cancer progression as potent cancer-supporting factors, but the mechanisms by which these cytokines are regulated in cancer require further investigation.
Here we demonstrate that monocyte-lineage cells are very common already in carcinoma-in-situ/superficially invasive OSCC (collectively referred to as OSCC). We show that in vitro, interactions between OSCC cells, monocytes and LPS consistently stimulate production of IL-6 and VEGF, irrespective of constitutive cytokine production by OSCC cells. Importantly, LPS can directly stimulate a subset of OSCC cell lines to produce cancer-supporting factors and chemoattractants. Finally, STAT3 was rapidly and consistently activated in all OSCC lines by supernatants derived from LPS-OSCC-monocyte co-cultures, and IL-6 was partially responsible for STAT3 activation. Our data suggest that bacterial contamination and the presence of monocyte-lineage cells in the OSCC environment may contribute to cancer progression.
Study protocols involving human subjects were approved by the University of Iowa Institutional Review Board.
We used OSCC cell lines Cal27 (tongue; ATCC, Rockville, MD), FaDu (pharynx; ATCC), UMSCC19, UMSCC47 (tongue; Dr. T. Carey, U. Michigan), 1483 (oral cavity; gift from Dr. P. Sacks, NYU), telomerase-immortalized tonsillar keratinocytes tertAd7cl41 and normal pharyngeal keratinocytes HTE1163 passages 4–7 (gifts from Dr. A. Klingelhutz and Dr. J. Lee, U. Iowa). HTE1163 and tertAd7cl41 were grown in Keratinocyte Serum Free Medium (KSFM) with 0.2 ng/ml epidermal growth factor and 30 μg/ml bovine pituitary extract (Invitrogen-GIBCO, Grand Island, NY). Cal27 and FaDu were grown in DMEM with 10% heat-inactivated FBS (HyClone, Logan UT); 1483, UMSCC19 and UMSCC47 were grown in DMEM/F12 (Invitrogen-GIBCO) with 10% FBS. Cell lines repeatedly tested negative for mycoplasma (ATCC). For all experiments, cell lines were plated in culture media with gentamycin for attachment (6–12 h). Experiments were conducted in serum-free X-Vivo15 with gentamycin (Cambrex Bio Science Inc., Walkersville, MD), unless specified otherwise. Once attached, all lines grew well in X-Vivo-15.
Monocytes were first enriched from normal donor peripheral blood with RosetteSep™ Monocyte Enrichment Cocktail and centrifugation through Ficoll-Paque Plus (StemCell Technologies, Vancouver BC). After 40–50 min incubation on plastic at 37°C, remaining unwanted cells were washed out. Purified monocytes were 95–99% CD14+, 99% CD11c+.
All primary antibodies were murine IgG1, IgG2a or IgG2b. Control antibodies were MIgG1-, 2a-, 2b-FITC or -PE (Southern Biotech Associates, Inc., Birmingham AL), anti-CD11c-PE clone BU15 (mIgG1); anti-CD14-Cy5-PE clone RMO52 (mIgG1) (Immunotech, Marseille, France).
Anti-CD68 (KP-1, LabVision, Fremont, CA and monocyte-specific PG-M1, Dako, Carpinteria, CA); anti-CD1a (O10, Immunotech); mouse IgG controls (LabVision).
Cell preparations for flow cytometry and analysis were performed as described previously [26, 27]. Briefly, cells were incubated on ice for 20–30 min with fluorochrome-labeled or biotinylated antibodies, washed, and the cells pre-incubated with biotinylated antibodies were then stained with avidin-fluorochrome conjugates (Southern Biotech Associates). Cells were fixed in 2% paraformaldehyde and analyzed on BD FACS Calibur (Becton Dickinson, Franklin Lakes, NJ).
Five cases of non-specific chronic mucositis and 15 cases of carcinoma-in-situ or superficially-invasive SCC were retrieved from the archives of the Surgical Oral Pathology Laboratory and the UI College of Dentistry. The hematoxylin- and eosin-stained sections were reviewed by Board-certified oral and maxillofacial pathologist (ZBK) to verify uniform application of criteria. Five-μm sections of the paraffin-embedded formalin-fixed tissue were mounted on aminosilane-coated slides (Newcomer Supply, Inc. Middleton, WI), de-paraffinized and re-hydrated. For IHC, DAKO Envision Doublestain System (DAKO, Carpinteria, CA) was used according to manufacturer instructions. Briefly, antibody pairs (anti-CD1a and anti-CD68, please see “Primary Antibodies”) were pre-tested on control tissue to establish the best protocol. After a peroxidase block, sections were stained with primary antibody or control, followed by polymer-HRP and DAB substrate. After a double-stain block, second primary antibody or control were added, followed by polymer-AP and Fast Red substrate. All incubations were separated by washing with Tris buffer. Additional controls were adjacent sections, stained with each of the two antibodies separately.
ELISAs for IL-6 (Pierce Endogen, Rockford, IL), VEGF, CCL-2 (MCP-1), and CCL-20 (MIP-3 alpha) [R&D Systems, Minneapolis, MN, except hrCCL20 (Peprotech, Inc., Rockford, IL)] were performed according to manufacturer instructions. Briefly, Nunc MaxiSorp™ 96-well plates were coated with cytokine-specific antibodies, blocked, and incubated sequentially with sample supernatants (in triplicate), biotinylated cytokine-specific antibodies, avidin-conjugated HRP and tetramethyl benzidine (TMB) substrate (BioFX Laboratories, Inc., Owings Mills, MD). OD at 450–650 nm (Powerwave X, Bio-Tek Instruments, Inc., Winooski, Vermont) was converted into concentration using a standard curve.
OSCC or keratinocytes were added to monocytes (in 12-well plates) at 1–2 × 105 cells per well (1:1 ratio). As both monocytes and OSCC/keratinocytes can produce some of the factors measured, a 1:1 ratio provides each population with optimal chance to interact. After 4–5 h OSCC and keratinocytes attached, and the media were replaced with 2 ml/well X-Vivo-15/gentamycin. Co-cultures were incubated for 3 days (7.5% CO2, 37°C). Cells and supernatants were analyzed, respectively, by flow cytometry and ELISA.
Keratinocyte and OSCC lines were plated in 12- or 6-well plates at 2 × 105 cells/ml. After cells adhered, media were replaced by X-Vivo-15 with gentamycin. Cells were stimulated with escalating doses of E. coli 026:B6 LPS (L2654, 2.5% protein, 1,500,000 EU/mg LPS, Sigma-Aldrich, St. Louis, MO) at 0, 2, 20, and 200 ng/ml. LPS stocks of 5 mg/ml in pyrogen-free water for injections were serially diluted in X-Vivo-15 to the desired concentrations. In other experiments, cells were stimulated with human recombinant IL-6 (Peproptech, Inc., Rockford, IL) (0, 10 and 20 ng/ml, encompassing the highest IL-6 levels found in the supernatants, and taking into account 1:1 dilution of supernatants for STAT3 activation experiments), or 1:1 fresh media:supernatants from respective co-culture experiments, in the presence of IL-6-neutralizing antibodies (1 μg/ml, clone 6708, MIgG1, R&D Systems), IL-6R-neutralizing antibodies (goat IgG AB-227-NA, 10 μg/ml, R&D Systems) or isotype/species/concentration-matched controls. For all experiments testing STAT3 activation, adherent cells were rinsed with warm media prior to stimulation. Cells were harvested at 30 min for Western blotting. In addition, supernatants from LPS-treated cultures and controls were harvested after 48 h for ELISA to verify functional responses.
After removing supernatants, cells were treated with radio-immunoprecipitation assay (RIPA) buffer (50 mM Tris, 150 mM NaCl, 1% Triton X-100, and 0.5% deoxycholate) with 50 mg/ml leupeptin, 10 mg/ml aprotonin, 2 mM EDTA, and 1 mM orthovanadate, and scraped on ice. Lysates were cleared by centrifugation at 8,000 × g, 4°C and supernatants tested by Western blotting. Protein concentration was determined using the Micro bicinchoninic acid Protein Assay Reagent Kit (Pierce, Rockford, IL). Normalized lysates were boiled in electrophoresis sodium dodecyl sulfate sample buffer with ME, run on 10% SDS-polyacrylamide gel electrophoresis gels (Bio-Rad Lab, Hercules, CA) and transferred to 0.45 μm nitrocellulose membranes (Whatman, Inc., Sanford, ME). Membranes were blocked overnight in TBS, pH 7.6 with 0.05% Tween-20 (TBS-T) and 3% dry milk. Membranes were probed overnight at 4°C with manufacturer-recommended concentration of 155 ng/ml anti-phospho-Y705-STAT3 rabbit antibody (Cell Signaling Technology, Beverly, MA), washed with TBS-T, incubated with 160 ng/ml goat-anti-rabbit antibody-HRP (Jackson ImmunoResearch, West Grove, PA) followed by ECL substrate (Amersham, Piscataway, NJ). Blot-exposed film (KODAK Biomax, Rochester NY) was analyzed using the Foto/Analyst Luminary System (Fotodyne, Inc., Hartland, WI). Membranes blotted for pY705-STAT3 were stripped with Stripping Buffer (Pierce, Rockford, IL) and re-blotted with anti-STAT3 rabbit antibodies at 6.25 ng/ml (Cell Signaling Technology). Alternatively, duplicate protein samples were run on parallel gels and transferred to two membranes, which were blotted, respectively, for pY705-STAT3 or total STAT3.
Up to seven high power fields (400×) of each section were evaluated for numbers of CD1a+CD68± and CD1a−CD68+ cells within the epithelial areas and separately, within the connective tissue areas. The average number of cells per three most intensely inflamed high power fields (HPF) in five samples of chronic mucositis (CM) and 15 samples of carcinoma-in-situ (CIS)/superficially-invasive SCC were compared using the nonparametric Wilcoxon Rank Sum test, and significance probabilities were calculated using exact methods.
For each cell line, the Spearman rank correlation was used to assess whether there was an increasing or decreasing trend in level of response with an increasing LPS dose, using exact tables . The one-way ANOVA was used to compare the four different conditions for each cell line in OSCC-monocyte co-culture experiments (media, monocytes, LPS, MO + LPS). The specific pairwise group comparisons of interest were the comparison of control response with each of the other three treatment groups; adjustment for multiple comparisons was made using the Bonferroni method in conjunction with an overall 0.05 level of Type I error. Standard regression techniques were used to conduct a formal test for interaction, in order to address the question of whether the combined treatment was associated with a greater mean response than treatment with either LPS or monocytes alone. Because sample sizes were modest, the overall test for group differences was confirmed using the nonparametric Kruskal–Wallis test, calculating significance probabilities using exact methods; the significance of results was confirmed in all instances.
Examination of carcinoma-in-situ/superficially-invasive OSCC typically revealed in numerous leukocytes, especially mononuclear cells. Because monocytes can differentiate into myeloid DC or macrophages, we used two-color IHC for a more complete assessment of monocyte-lineage cells in the epithelial and connective tissue components of carcinoma-in-situ/superficially invasive OSCC. We found CD68dim/brightCD1a− monocytes and macrophages and CD1a+CD68± myeloid DC populations (Fig. 1). Our analysis revealed several patterns in the relative proportions of the infiltrating populations. In some cases, CD1a−CD68+ monocytes and macrophages were predominant, while CD1a+ DC were sparse (Fig. 1A). In other cases, all three cell types were numerous (Fig. 1B). Alternatively, CD1a+ DC predominated in some cases (Fig. 1C), and occasionally the entire infiltrate was relatively mild, but never absent. In addition, patterns from one area to another in the same lesion often varied.
Because of frequent variability in the distribution of infiltrating cells in the same lesion, we performed quantitative analysis of most intensely leukocyte-infiltrated areas in OSCC, as an indication of the maximal local response. To assess what impact malignant epithelial change had on the monocyte-lineage cell involvement in mucosal inflammation, we compared the maximal response in OSCC to that in non-specific chronic oral mucositis, i.e. chronic inflammation of the oral mucosa without evidence of pre-cancer or cancer, or a specific infectious or autoimmune process. Figure 1D shows that as a group, monocytes, macrophages and myeloid DC were significantly more common in the connective tissue of OSCC than in the connective tissue of chronic mucositis specimens (P < 0.0072). Similarly, the malignant epithelium in OSCC often contained more infiltrating cells of the monocyte lineage, although the differences from chronic mucositis did not reach statistical significance (P < 0.069). Together, these results indicate that as a group, monocyte-lineage cells represent a major component of the OSCC environment in carcinoma-in-situ/superficially-invasive lesions, and that the presence of malignant epithelial change may enhance their recruitment and/or retention, while the relative proportions of the three populations vary.
Surface colonization of rough oral mucosal lesions is readily apparent histopathologically, and Fig. 1E shows an example of microbial colonies associated with the rough surface of an oral carcinoma-in-situ specimen (top). Figure 1E also shows an example of ulcerated poorly-differentiated advanced T4 OSCC (bottom) that contains dense infiltrates of inflammatory cells admixed with tumor cells and spilling onto the mucosal surface. These observations suggest that microbial products have the potential to play a role not only in advanced OSCC, but also in pre-invasive lesions.
In specimens, we found monocytes often in direct contact with OSCC cells. As monocytes produce IL-6 in response to microbial products, and carcinoma cells may or may not produce IL-6, we evaluated in vitro the outcome of interactions between monocytes and OSCC, in the presence or absence of LPS. As shown in Fig. 2A (top panel), three out of five OSCC cell lines (Cal27, FaDu and1483) produced very little to no IL-6 constitutively, and little if any IL-6 was detectable in unstimulated monocyte cultures. In the absence of LPS, monocyte co-cultures with normal or immortalized keratinocytes or with OSCC lines UMSCC19 and UMSCC47 stimulated IL-6 production, while monocyte-Cal27, monocyte-FaDu and monocyte-1483 interactions generally did not. However, LPS strongly stimulated IL-6 production in all monocyte-OSCC and monocyte-keratinocyte co-cultures, as compared to the levels of constitutive IL-6 production in OSCC cells (“medium”), in the presence of LPS only, or with monocytes only (Fig. 2A, lower panel). The strength of IL-6 induction in LPS-containing co-cultures paralleled the responses in monocyte-OSCC co-cultures without LPS, i.e. the monocyte-OSCC co-cultures that showed no IL-6 induction in the absence of LPS (Cal27, FaDu and 1483) also showed less potent, but nevertheless highly significant and consistent IL-6 induction in the presence of LPS.
Similarly, presence of LPS enhanced VEGF production in all monocyte-keratinocyte and monocyte-OSCC co-cultures, while in the absence of LPS, very small increases in VEGF were seen in most co-cultures with monocytes (Fig. 2B). These data suggest that for all the cell lines tested, the presence of both LPS and monocytes is important for consistent and potent IL-6 and VEGF production, while in all other conditions there is significant heterogeneity among cell lines. These experiments also show that LPS can independently stimulate VEGF production in Cal27 and 1483, as well as IL-6 in Cal27 cells, and possibly had some effect on FaDu cells.
We further investigated OSCC and keratinocyte responses to LPS. As monocyte-lineage cells are common in OSCC, we measured LPS dose-dependent production of important monocyte and DC chemoattractants CCL2 and CCL20, as well as cancer-supporting factors IL-6 and VEGF (Fig. 3A–D). Normal keratinocytes HTE1163 did not significantly change cytokine and chemokine output, but immortalized keratinocytes tertAd7cl41 and three out of five carcinoma cell lines tested (Cal27, FaDu and 1483) each increased production of two or more factors in a dose-dependent fashion. Interestingly, the same three lines also constitutively produced little or no IL-6, and two of these three (Cal27 and 1483) responded to LPS by up-regulating monocyte-DC chemoattractant CCL2. Cal27 consistently and strongly up-regulated all tested factors, while 1483 up-regulated CCL2 and VEGF, but CCL20 and IL-6 were undetectable. FaDu showed small but consistent and statistically significant increases in CCL20 and IL-6 without increasing VEGF, and no CCL2 was detected. Keratinocytes tertAd7cl41 showed modest, but statistically significant increases in CCL20 and VEGF. Importantly, Cal27 and 1483 showed much stronger responses than keratinocytes. Further, with the exception of 1483 and tertAd7cl41, all lines expressed detectable levels of surface CD14, and all cell lines showed very low surface levels of toll-like receptor (TLR)4 (Fig. 4), without apparent correlation to the strength of the OSCC cell responses.
Together, these results suggest that LPS can independently stimulate production of cancer-supporting factors IL-6 and/or VEGF in a subset of OSCC cells, and can enhance OSCC cell ability to attract monocytes, important for the production of IL-6 and VEGF.
To verify that keratinocytes and OSCC cells could directly benefit from the presence of IL-6, we measured Y705-STAT3 phosphorylation in response to recombinant human IL-6 (Fig. 5A). Little to none constitutively phosphorylated Y705-STAT3 was detected in most cell lines (“medium”), while recombinant IL-6 rapidly induced Y705-STAT3 phosphorylation in four out of five OSCC cell lines (all low-intermediate- or non-producers of IL-6) and in both keratinocyte lines (one of two keratinocyte cell lines is shown). Neutralizing anti-IL-6 and anti-IL-6R antibodies effectively prevented IL-6 from inducing STAT3 activation. In contrast, recombinant IL-6 induced little, if any, STAT3 phosphorylation in the high constitutive IL-6-producer UMSCC19. In summary, four our five OSCC cell lines tested, as well as keratinocytes, were responsive to IL-6 as demonstrated by STAT3 activation.
Next, we evaluated OSCC cell STAT3 activation in response to factors produced in the co-culture experiments shown in Fig. 2, and examined the role of IL-6. Figure 5B shows that supernatants from only two unstimulated OSCC lines (sOSCC) FaDu and UMSCC47 induced readily detectable pY705-STAT3 at 30 min, which did not change significantly upon IL-6 blocking, even though UMSCC47 produces IL-6 constitutively. Importantly, the supernatants from LPS-monocyte-OSCC/keratinocyte co-cultures rapidly and strongly induced pY705-STAT3 in all the respective keratinocyte and OSCC lines, and when IL-6 was neutralized, pY705-STAT3 decreased to varying degrees in Cal27 and 1483 cells. Even though IL-6 was clearly induced in all LPS-monocyte-OSCC/keratinocyte co-cultures (Fig. 2), and all cell lines except UMSCC19 were responsive to IL-6 (5A), tertAd7cl41, FaDu, UMSCC19 and UMSCC47 showed little to no decrease in STAT3 activation when IL-6 in the sOSCC + mo/LPS sups was neutralized (5B). At the same time, 30-min stimulation with LPS had no effect on STAT3 activation in any cell lines (not shown). These results indicate that consistent and rapid STAT3 activation in heterogeneous OSCC lines requires soluble factors, which are produced constitutively by some OSCC, but are most consistently and strongly induced through LPS-monocyte-OSCC interactions, and that IL-6 is one of these factors.
Our studies of oral specimens provide new information on the characteristics of OSCC-associated infiltrates. We focused specifically on under-studied carcinoma-in-situ/superficially-invasive OSCC to determine how prominent monocyte-lineage cells are at these relatively early developmental stages of OSCC. We evaluated monocytes, macrophages and DC as a group for reasons of their relatedness through differentiation, and examined separately the connective tissue and the epithelial component to understand the distribution in better detail. Most importantly, we compared OSCC to a more relevant condition, i.e. non-specific chronic mucositis without malignant change, rather than to normal mucosa, in order to assess the effect of malignant epithelial change on the participation of monocyte-lineage cells in inflammation. Our observations are consistent with previous reports showing that CD68+ cells were significantly more frequent in T1-T4 HNSCC than in normal tissue , and that the numbers of DC associated with advanced HNSCC varies, as detected by IHC for S-100 and p55 , CD1a and S-100 , or CD1a and the DC maturation marker CD83 . Reported studies agree that DC numbers decline with advancing HNSCC stage within the tumor [9, 30] as well as in the peripheral blood .
There are several potential reasons for increased association of monocyte-lineage cells with developing OSCC. One likely reason is the compromised mucosal barrier, because both epithelial cell differentiation/maturation and epithelial continuity are important for effective barrier function . OSCC environment appears to be well suited for bacterial survival, because viable bacteria were isolated from various portions of primary OSCC , as well as from ~75% of cervical lymph nodes with metastatic OSCC from various oral primaries, and 100% of cervical lymph nodes with metastatic OSCC from gingival primaries . In addition, high salivary counts of several oral bacteria were described as potential diagnostic predictors of OSCC . A compromised barrier would also permit other irritants, besides microorganisms and their products, to pass through more readily.
Another possible reason is that OSCC cells can produce chemoattractants. We found, consistent with other studies , that most established OSCC lines we tested in vitro produced monocyte-DC chemoattractants CCL2 and/or CCL20, which is similar to keratinocytes. While CCL2 recruits monocytes and DC out of peripheral blood , CCL20 is responsible for attracting a subset of myeloid DC known as Langerhans cells into the epithelium . Most importantly, three out of five OSCC cell lines enhanced production of one or both chemoattractants in response to LPS. These observations suggest that in vivo, increasing numbers of proliferating OSCC cells, and the enhanced production of chemoattractants in response to LPS by some OSCC, could both contribute to the increase in monocyte and/or DC association with OSCC. It is also possible that OSCC cells produce other monocyte-DC chemoattractants. To our knowledge, this is the first demonstration that LPS can independently stimulate OSCC cells to produce chemoattractants CCL2 and/or CCL20, as well as cancer-supporting factors IL-6 and VEGF.
We used E. coli LPS in vitro as the model Gram-negative bacterial product, because of the abundance of Gram-negative bacteria at mucosal surfaces, the ability of LPS to stimulate cells at very low concentrations, and because the effects of E. coli LPS of high purity are representative of most typical LPS. Further, Gram-negative bacteria are implicated in the pathogenesis of gastric carcinoma [34, 35]. Helicobacter (H.) pylori is strongly associated with gastric carcinoma , and H. pylori LPS is one of the major virulence factors for induction of gastritis . LPS was shown to enhance DNA synthesis in human colon carcinoma cells  and to stimulate integrin receptors, increasing the invasive and metastatic potentials . In addition, a variety of murine cancer cell lines were recently shown to express multiple functional receptors for microbial products . E. coli LPS effects described in immune system cells are mediated primarily by signaling through TLR4, which depends on LPS forming a complex with serum LPS-Binding Protein (LBP), membrane or serum CD14, and extracellular protein MD-2 . CD14, TLR4 and other TLR can be expressed in keratinocytes . Keratinocyte and OSCC lines described here bound the TLR4-specific antibody HTA125 at low levels. Further, all OSCC lines, except 1483, expressed membrane-bound CD14, and the experiments were conducted in low-serum conditions with potentially sufficient LBP and soluble CD14, so that OSCC may well respond to LPS using TLR4. The surface levels of TLR4 expression were very low, which could not be explained by trypsin treatment, because the non-adherent TLR4-positive monocytoid cell line THP-1 showed no decrease in surface HTA125 binding upon treatment with trypsin (not shown). In immune system cells, multiple factors are known to regulate LPS binding and TLR4 signaling, suggesting that the observed lack of correlation between OSCC cell receptor expression levels and responsiveness to LPS is not sufficient to eliminate TLR4 and CD14 as the candidate LPS receptors in OSCC cells.
LPS was shown to induce IL-1 and IL-8 in keratinocytes . In our studies, LPS stimulated immortalized oral keratinocytes tertAd7cl41, but not normal oropharyngeal keratinocytes HTE1163, to produce statistically significant increases in CCL20 and VEGF (Fig. 2), as well as IL-8 (data not shown). Notably, the differences between OSCC lines and normal or immortalized keratinocytes in the constitutive and LPS-induced production of soluble factors were primarily quantitative, rather than qualitative. Two out of three LPS-responsive OSCC lines (Cal27 and 1483) demonstrated strong responses, and the third line (FaDu) showed weak responses similar to keratinocytes tertAd7cl41, while there were two OSCC lines, UMSCC19 and UMSCC47, that did not respond to LPS stimulation, similar to keratinocytes HTE1163. While the number of cell lines tested is small, one possible interpretation of these similarities is that malignant squamous cells may preserve or enhance the abilities to interact with microbial products and/or monocytes characteristic of normal keratinocytes.
Clearly, OSCC or keratinocyte association with both monocytes and LPS consistently stimulated IL-6 and VEGF, in spite of heterogeneity among cells lines under all other conditions tested. The mechanisms, by which monocyte-OSCC interactions, particularly in the presence of LPS, stimulate IL-6 and VEGF production, are under investigation. Because these experiments were conducted over several days, it is possible that differences in cell differentiation, survival and proliferation affected the absolute levels of secreted factors. Indeed, we have found some differences in the phenotypes of the CD11c-positive monocyte-lineage cells at the end of the co-cultures, depending on the specific OSCC cell line (not shown). Nevertheless, it is clear that IL-6 and VEGF are induced in all LPS-monocyte-OSCC co-cultures.
Our results suggest that in the LPS-containing co-cultures, IL-6 may be produced by the OSCC cells (except 1483) and by monocytes, while VEGF appears to have been produced mainly by keratinocytes and OSCC cells. Intracellular staining revealed that in LPS-stimulated co-cultures with low IL-6 producers (Cal27, 1483 and FaDu) monocytes were the key source of IL-6 (A. Lam-ubol, Manuscript in preparation). No matter what the source, any cells present are potentially subject to the effects of IL-6 and VEGF. Both IL-6 and VEGF are known to have profound negative effects on monocyte-DC differentiation and on anti-tumor immune responses [25, 31, 42]. The reported decline in myeloid DC numbers with advancing OSCC [9, 29, 30, 42] is likely due at least in part to increasing concentrations of both IL-6 and VEGF, which interfere with precursor cell differentiation into DC [8, 31], but do not interfere with continued production of IL-6 and VEGF.
STAT3 activation, which is inducible by IL-6, was shown to be important for cancer cell survival and growth and for VEGF production. However, some carcinoma cells do not necessarily respond to IL-6 . For example, one out of five OSCC, the high IL-6-producer UMSCC19, showed little or no STAT3 activation in response to IL-6. Similarly, cervical carcinoma cells that produced high IL-6 levels poorly responded to IL-6 due to the loss of IL-6-binding receptor subunit gp80 . Yet again, OSCC cells proved to be heterogeneous in the conditions under which STAT3 activation occurred, but uniform in the lack of constitutive STAT3 activation within 30 min of placement in fresh medium. Remarkably, all OSCC cell lines were able to undergo STAT3 activation in response to soluble products of LPS-monocyte-OSCC interactions, which may be the basis for the consistent increase in VEGF production in these co-cultures. Because constitutively active STAT3 can by itself act as a transforming agent in epithelial cells , and activated STAT3 mediates protection from apoptosis, leads to proliferation [15–17], and induces VEGF production [22, 23, 45], microbial product-contaminated and monocyte-rich OSCC could have a significant survival and growth advantage, potentially at any stage of OSCC development.
In contrast to many cancers, the degree of HNSCC cell differentiation is not a useful prognostic factor. A multi-center study of over 3000 HNSCC patients with well differentiated vs. poorly differentiated carcinomas revealed very modest differences in survival (33% vs. 27%, respectively) . Our model is that whether well- or poorly-differentiated, successful HNSCC have a “working relationship” with the microbial environment and the innate immune system, and the most rapidly progressing SCC make (or increase production of) multiple cancer-supporting factors directly in response to microbial products, and strongly attract monocytes while influencing their state of differentiation, which further magnifies the production of such factors.
Clearly, there are also implications for in vitro studies of carcinoma biology, as even minor LPS contamination can have profound effects on the experimental outcomes. As the mucosal environment includes a variety of microorganisms, other microbial products may influence OSCC pathogenesis. The results of this study necessitate investigations of the potential for cancer-promoting effects of TLR engagement that may cancel out beneficial anti-cancer effects of immunotherapies that target TLR.
We thank Dr. G. B. Schneider, Rebecca Zacharias and Denise Seabold (College of Dentistry, University of Iowa) for the generous help with Western blotting; Ms. Crystal Fairlie (College of Dentistry, University of Iowa) for assistance with cutting tissue sections, and Dr. O. Rokhlin (Department of Pathology, University of Iowa), Dr. J. McCutcheon (New York University), and Dr. C. T. Lutz (University of Kentucky) for the review and critique of the manuscript. Supported by NIDCR Dental Student Research Fellowship (C. De La Mater), Anandamahidol Foundation (A. Lam-ubol), American Cancer Society Grant #IN-122V administered by The University of Iowa Holden Comprehensive Cancer Center, NIH R01 DE11139, University of Iowa College of Dentistry Seed Grant and the Department of Oral Pathology, Radiology and Medicine at the College of Dentistry, University of Iowa.