|Home | About | Journals | Submit | Contact Us | Français|
In order to determine the phenotype and function of myeloid DCs from human cutaneous squamous cell carcinoma (SCC), we studied their surface marker expression and allo-stimulatory potential ex vivo. There were abundant CD11c+ myeloid DCs, as well as TNF and inducible nitric oxide synthase (iNOS)-producing DCs (TIP-DCs), in and around SCC tumor nests. Although myeloid DCs from SCC, adjacent non-tumor bearing skin, and normal skin, were phenotypically similar by flow cytometry, and there was a pronounced genomic signature of mature DCs in SCC, they showed different T cell stimulatory potential in an allo-MLR. Myeloid DCs from SCC were less potent stimulators of allogeneic T cell proliferation than DCs from non-tumor bearing skin. Culture with a DC-maturing cytokine cocktail (IL-1β, IL-6, TNF, and PGE2) enhanced stimulatory potential in DCs from non-tumor-bearing skin, while SCC associated DCs remained poor stimulators of T cell proliferation. The microenvironment associated with SCC showed expression of TGFβ, IL-10 and VEGF-A, factors capable of suppressing DC function. These findings indicate that CD11c+/HLA-DRhi DCs from SCC are mature, but are not potent stimulators of T cell proliferation compared with phenotypically similar DCs isolated from non-tumor-bearing skin. Identification of mechanisms responsible for suppression of tumor-associated DCs may provide insight into evasion of immunosurveillance by SCC.
Squamous cell carcinoma (SCC) is the second most common human cancer affecting over 250,000 individuals in the United States in 2007 (Weinberg et al., 2007). In most cases, cutaneous SCC is treated successfully by surgery, however it has the potential to behave aggressively and accounts for the majority of ~10,000 non-melanoma skin cancer deaths in the US each year (Thomas et al., 2007). Aggressive behavior in SCC is associated with immune compromise in the host manifested by increased incidence in solid organ transplant recipients and individuals with chronic lymphocytic leukemia (CLL) (Vanacker et al., 2008). Based on the potential for host immunity to regulate behavior in SCC, we evaluated the immune microenvironment and function of tumor-associated myeloid DCs.
DCs are professional antigen presenting cells that mediate innate and adaptive immunity, and may play a key role in anti-tumor response (Steinman, 2006). Mature DCs are identified by expression of HLA-DR, CD86, CD83, CD80, CD40 and other markers, and their ability to stimulate naïve T cell proliferation. There are three major subsets of cutaneous DCs in humans: myeloid DCs, plasmacytoid DCs (pDCs) and Langerhans cells (Zaba et al., 2008b). There have been reports in recent years describing tumor-associated DCs and how the numbers and types of DCs present may serve as a prognostic indicator, including colorectal, cervical carcinoma, esophageal SCC, nasopharyngeal carcinoma, urinary bladder carcinoma and gastric carcinoma. For example, there was a significant increase in the number of immature DCs in late stage ovarian carcinoma-associated ascites compared to peripheral blood, and there was a significantly greater number of ascitic pDCs than myeloid DCs in many of these patients (Siegal et al., 1999; Wertel et al., 2008a; Wertel et al., 2008b). These observations are typically indicative of a poor prognosis for these patients (Zou et al., 2001). An increased presence of pDCs has also been correlated with a poor prognosis for patients with SCC of the head and neck (SCC of the oral and nasal mucosa) (Hartmann et al., 2003; O'Donnell et al., 2007).
In vitro tumor antigen- or tumor lysate-pulsed monocyte-derived myeloid DCs (MoDCs) can be generated for use as a specific anti-tumor vaccine (Lee et al., 2002). For example, in a clinical trial of 16 patients with stage IV melanoma, two complete, three partial and one minor response was reported after treatment with melanoma tumor or peptide pulsed MoDCs (Nestle et al., 1998). This strategy has been variably successful in different cancer settings, but supports the anti-tumor potential of these DCs (Dhodapkar et al., 2002; Jeong et al., 2007; Nestle et al., 2001; Pospisilova et al., 2002).
We have previously characterized the immune infiltrate in basal cell carcinoma (BCC) (Kaporis et al., 2007). That study showed that there was an increased number of CD11c+ DCs in the BCC tumor microenvironment compared to normal skin, suggesting a role for DCs in this type of cancer. We have now turned our attention to DCs in cutaneous SCC, as their potential role and function in this human epithelial cancer remains undefined. We studied the DC microenvironment in human SCC and evaluated the populations, numbers, phenotypes and stimulatory capacity of DCs from SCC, peritumoral non-lesional skin (PTNL), and normal skin. We found the following: 1) SCC is associated with appreciable numbers of CD1a+ cells, Langerin+ cells, CD11c+ cells, and BDCA1+ cells; 2) the SCC microenvironment is also associated with TNF and inducible nitric oxide-producing myeloid DCs (TIP-DCs); 3) CD11c+/HLA-DRhi myeloid DCs from SCC express maturity markers at levels similar to those from patient-matched, site-matched PTNL skin, and normal skin from tumor-free subjects; 4) SCC is associated with a mature DC gene expression signature; 5) myeloid DCs from SCC are poor stimulators of allogeneic T cell proliferation compared to myeloid DCs from PTNL skin and normal skin; 6) and the tumor microenvironment is associated with an increased expression of immunosuppressive cytokines.
SCC and normal skin were evaluated for the presence of DCs by immunohistochemistry. Representative immunohistochemistry is shown for each antigen analyzed, in normal skin and SCC and (n=10-12 for each group), and cell counts of all the cases are shown (Figure 1, right). We determined the numbers of DCs within the SCC microenvironment based on those counted within the tumor nodules and in the papillary dermis 100μm immediately around the tumor nodules (juxtatumoral). The numbers of DCs in normal skin were evaluated based on those observed in the epidermis and those found in the normal papillary dermis 100μm deep to the basement membrane.
In SCC, CD1a+ and Langerin+ cells were infiltrating epithelial tumor aggregates at numbers lower than those per unit area of normal epidermis (p=0.016 and p=0.003 respectively) (Figure 1A, and B). There were even less Langerhans cells in juxtatumoral skin compared to SCC (p<0.001). There were greater numbers of CD11c+ myeloid DCs in the juxtatumoral dermis compared to SCC tumor nodules (p<0.001), with relatively few CD11c+ myeloid DCs in SCCs compared to normal skin (p=0.0012). (Figure 1C). There were significantly greater numbers of BDCA-1+ cells observed in both normal dermis (p<0.05) and juxtatumoral dermis compared to SCC (p<0.005) (Figure 1D). There were higher numbers of BDCA-2+ pDCs in the juxtatumoral dermis compared to normal papillary dermis (p<0.002), fewer pDCs infiltrating SCC tumor nodules compared to juxtatumoral dermis (p<0.025), but greater numbers of pDCs were observed in SCC tumor nodules than in normal papillary dermis (p<0.05) (Figure 1E).
TIP-DCs have been described in imiquimod-treated BCC (Stary et al., 2007), and recent studies from our group described the presence of TIP-DCs in psoriasis (Lowes et al., 2005). Treatment of psoriasis patients with efalizumab (anti-CD11a, Raptiva) strongly reduces infiltration by these DCs in patients responding to this agent, suggesting that TIP-DCs may play a role in keratinocyte hyperproliferation. Since SCC is also a condition involving keratinocyte hyperproliferation, we wanted to determine if TIP-DCs were present in the SCC microenvironment. Triple-label immunoflourescence shows that there are CD11c+ (blue) myeloid DCs that also express TNFα (green) and iNOS (red), which appear white in color (Figure 2, representative image).
We wanted to characterize the surface phenotype of CD11c+ myeloid DCs associated with SCC. CD11c+/HLA-DRhi myeloid DCs from normal skin, PTNL skin and SCC were evaluated for expression of co-stimulatory and maturity markers DC-LAMP/CD208, DC-SIGN/CD209, DEC-205/CD205, CD83, CD86, and CD80 by flow cytometry. Representative FACS histograms for selected markers are shown; MFI is mean of patients studied (n=3) (Figure 3).
Expression of DC-LAMP was slightly lower in SCC-derived myeloid DCs (MFI=1068) compared to those from PTNL and normal skin (MFI=1228 and 1492 respectively). DC-SIGN expression was slightly higher in myeloid DCs from SCC (MFI=1938) compared to those from PTNL skin, but lower than those from normal skin (MFI=1759 and 2278 respectively). Myeloid DCs from SCC expressed greater levels of DEC-205 (MFI=32963) than PTNL and normal skin (MFI=27513 and 18616 respectively). SCC-associated myeloid DCs expressed CD83 (MFI=2963) but at slightly lower levels than those from site-matched PTNL and normal skin (MFI=3959 and 3527 respectively). Myeloid DCs from SCC expressed higher levels of CD86 (MFI=788) compared with site matched PTNL and normal skin MFI=566 and 414 respectively). Myeloid DCs from SCC express lower levels of CD80 (MFI=283) than those from site-matched PTNL and normal skin (MFI=387 and 662 respectively). Based on these results it appears that SCC-associated myeloid DCs achieved slightly increased expression of maturation marker DEC-205 and co-stimulatory molecule CD86 than PTNL skin and normal skin, but all three groups have a similar overall phenotype.
The genomic signature of DC genes was analyzed in our samples, since the surface phenotype of myeloid DC in SCCs was relatively mature and similar to DCs in adjacent PTNL skin. The fold change for DC genes in four lists was evaluated, as described in material and methods. Figure 4 shows the mean fold change of DC gene sets for SCC vs. normal skin (black bars) and PTNL vs. normal skin (gray bars). Mature DC genes showed increased mean expression in SCC (p<0.004). In PTNL vs. normal skin, all gene lists showed increased expression (p<0.05). The significance of those findings was also tested using Gene Set Enrichment Analysis (GSEA) (Table 1). DC gene expression was significantly enriched in PTNL skin compared to normal skin (Hohenkirk_DC_Up, p=0.024), while in SCC, enrichment of mature DC genes is increased (NES =1.25) but did not reach statistical significance (p=0.12). Overall, these data indicate DC gene activation in and around human SCC.
Since CD11c+/HLA-DRhi DCs from SCC and PTNL skin expressed similar levels of maturity markers we measured the capacity of these cells to stimulate allogeneic T cell proliferation in an allo-MLR. DCs were isolated from SCC and adjacent PTNL skin and co-cultured with allogeneic T cells for 8 days. Representative dot plots for CFSE vs. CD3 are shown for controls and each condition, and the gate for proliferating cells is shown (Figure 5A). A summary of all MLR data is shown in Figure 5B, and results tabulated in Table 2. On day 8, allogeneic T cell proliferation determined by CFSE dilution showed that SCC-derived myeloid DCs were 2-3 fold weaker stimulators of allogeneic T cell proliferation compared to DCs derived from matched PTNL, but they were similar to normal skin (Figure 5A). Treatment with a maturation cocktail containing cytokines including: IL1β, IL6, TNFα and PGE2, increased stimulatory potential of peritumoral DCs slightly, but was less effective in enhancing stimulatory potential of DCs from SCC than normal skin (Figure 5, right lower three panels).
Since myeloid DCs from SCC display maturity markers, yet lack appreciable allostimulatory potential, the level of expression of soluble immunosuppressive factors, IL-10, TGF-β and VEGF-A, was evaluated in the tumor microenvironment compared to PTNL and normal skin. The mean IL-10 mRNA expression was increased in both SCC and PTNL skin compared to normal skin. However, these increases were not significant (p=0.086 and p=0.199, respectively) (Figure 6, left panel). The mean expression of TGF-β mRNA was significantly increased in SCC compared with both PTNL and normal skin (p<0.027 and p<0.001, respectively) (Figure 6, middle panel) and in PTNL compared to normal skin (p<0.001). There was also a significantly higher mean level of expression of VEGF-A mRNA in both SCC and PTNL skin compared to normal skin (p<0.001 for both) and the mean level of VEGF-A mRNA was significantly increased in PTNL skin compared to SCC (p<0.041) (Figure 6, right panel).
Several findings stand out from the results reported here: 1) human cutaneous SCC was associated with CD11c+ and BDCA1+ cells in the juxtatumoral dermis; 2) CD1a+ and Langerin+ Langerhans cells were found in epithelial tumor nests, but at lower concentrations per unit area than normal epithelium; 3) the SCC microenvironment was associated with CD11c+ TIP DCs; 4) CD11c+/HLA-DRhi myeloid DCs from SCC and site-matched peritumoral skin express fundamentally similar levels of maturation markers; 5) expression array analysis indicated a mature DC pathway signature in SCCs; 6) myeloid DC from SCC were poor stimulators of allogeneic T cell proliferation compared with DCs from adjacent PTNL skin; 7) treatment with maturation cytokines enhanced stimulatory potential of DCs from normal skin and peritumoral skin but had less effect on DCs from SCC; and 8) the SCC tumor microenvironment was associated with an increased expression of immunosuppressive cytokines.
Thus, our initial characterization of the myeloid DCs in SCC demonstrated that while these DCs were as phenotypically mature as normal skin DCs, they were impaired in their ability to upregulate their allostimulatory capacity compared to those obtained from normal skin. The difference in allostimulation is most noticeable when the myeloid DCs from both SCC and normal skin were treated with maturing cytokines. A similar observation was described by Zaba et al (Zaba et al., 2007b) and is due to the “steady-state” myeloid DCs found in normal skin not being fully mature, but may become so when immunologically or otherwise challenged. We also demonstrated that although located only 2-5mm from the tumor margin, DCs derived from PTNL skin were consistently better stimulators of allogeneic T cells than those derived from SCC. Hence, although myeloid DCs from SCC and PTNL skin expressed similarly intermediate level of maturity, they differed in allogeneic T cell stimulatory capacity.
However, there was an abundant genomic signature of both immature and mature DCs in SCC lesions, so we consider that the tumor may be secreting factors causing this paralysis. Candidates include IL-10 and TGF-β, both of which were elevated in lesional tissue and sun-damaged skin adjacent to the tumor. Both these cytokines may have an immunosuppressive role (Mimura et al., 2007; Steinbrink et al., 1999). VEGF-A, while not increased in SCC, was increased in peritumoral skin compared to normal skin, and has been reported to decrease the stimulatory function of mature MoDCs (Mimura et al., 2007). This observation is noteworthy, because it may help explain why myeloid DCs from peritumoral skin still significantly lack stimulatory capacity compared to those from normal skin even when treated with maturing cytokines. VEGF-A isoforms are designated based on their peptide lengths, the most well studied being VEGF-A121, VEGF-A165 and VEGF-A189, all of which have pro-angiogenic effects. However, Ladomery et al has described the expression of another class of VEGF-A isoforms derived from alternative 3’ splicing of the mRNAs that correspond to the above-mentioned VEGF-A proteins (Ladomery et al., 2007). These splice variants produce peptides of nearly identical lengths (i.e. VEGF-A121b, VEGF-A165b and VEGF-A189b), but have anti-angiogenic effects. The splice variant mRNAs for VEGF-A165 and VEGF-A165b produce peptides that are identical except for the last 6 C-terminal amino acids. Since the commercially available RT-PCR primer-probe set we used may identify both pro- and anti-angiogenic isoforms of VEGF-A, it may be that the increased expression of VEGF-A in normal skin represents the anti-angiogenic isoform and the level of expression of VEGF-A in SCC represents the level of the pro-angiogenic isoform. Bates et al address this distinction for a number of normal tissues in their study (Bates et al., 2002). We are currently working to define differential VEGF-A isoform expression in the SCC microenvironment. Thus tumors may develop many mechanisms by which to slow or disable myeloid DC anti-tumoral activity.
In addition to tumor-derived factors that may impair local DC function, there are other leukocytes that appear to promote tumor growth, including immature DCs, pDCs, regulatory DCs, and regulatory T cells (Berger et al., 2002; Dauer et al., 2003; Enk et al., 1997; Talmadge et al., 2007). There is also a heterogeneous population of myeloid cells that may have tumor promoting functions, termed myeloid-derived suppressor cells (MDSC). This group of cells includes granulocytic, macrophage, DC and early myeloid progenitors (Apolloniet al., 2000; Kusmartsev and Gabrilovich, 2005; Young and Lathers, 1999; Zea et al., 2005). In humans they are identified as CD3−, CD11b+, CD14−, CD19−, CCD33+, CD56−, and HLA-DR−. MDSCs can inhibit T cell activation and antigen-specific proliferation (Movahedi et al., 2008). Angiogenic factors such as plasminogen activator inhibitor 1 (PAI-1) correlated with an increased number of immature DCs and MDSCs in the blood of patients with advanced malignancies (Osada et al., 2008). The mechanisms responsible for immunosuppressive functions of MDSC include production of reactive oxygen species (ROS), nitric oxide (NO), L-arginine metabolism, and cytokines. MDSCs have been shown to prevent the development of antigen-specific T cells by decreasing the proximal concentration of L-arginine via the arginase and nitric oxide synthase pathways (Bronte et al., 2003). VEGF has also been directly correlated with myeloid DC expansion (Ohm and Carbone, 2001). Therefore, not only can tumor cells modulate the immune system to escape immunosurveillance, but they can also take control of immune cells and use them to actively interfere with continued efforts by the immune system to destroy it.
In our study, we have observed the presence of TIP-DCs around the SCC tumor nests. We have previously described TIP-DCs in the common inflammatory skin disease psoriasis (Lowes et al., 2005). These cells were first described in a murine model of Listeria monocytogenes infection (Serbina et al., 2003), and their presence in the spleen of infected mice was CCR2-dependant. We have since determined that these TIP-DCs are part of a group of myeloid dermal DCs that we have termed “inflammatory” DCs. These cells are characterized by the following surface markers: CD11c+, HLA-DRhi, and they are BDCA-1/CD1c negative, unlike the resident myeloid DCs found in normal skin in steady state which are BDCA-1/CD1c+ (Zaba et al., 2007b; Zaba et al., 2008b). These inflammatory DCs are able to produce cytokines and inflammatory mediators such as TNF, iNOS, IL-20, and IL-23 (Wang et al., 2006; Zaba et al., 2007a; Zaba et al., 2008a; Zaba et al., 2008b). They are also able to induce allogeneic T cell proliferation, and Th1 and Th17 cell polarization (Zaba et al., 2008a). Thus, while capable of classic antigen-presenting function ex vivo, we hypothesize that the main role of these inflammatory DCs in the skin may be that of pro-inflammatory cytokine production critical in the pathogenesis of psoriasis.
Stary et al were the first to show that TIP-DCs were present in a tumor model, in BCCs after treatment with the TLR7/8 agonist imiquimod (Stary et al., 2007). Although they did not show staining of these tumors for these markers before imiquimod, it was presumed that the treatment induced these TIP-DCs. In an effort to further characterize the CD11c+ myeloid DCs surrounding the SCCs, we performed immunofluorescence for TNF and iNOS, expecting an absence of such mediators. It was a surprise to find abundant dermal TNF and iNOS-producing CD11c+ cells. We have previously demonstrated increased iNOS mRNA in SCCs (Haider et al., 2006).
This finding led us to consider the role of these TIP-DCs, and their relationship to the myeloid DCs we obtained for functional assays. The gating strategy for obtaining cells from the SCC tissue for T cell stimulating capability was CD11c+/HLA-DR+, and it is not possible to retrospectively assess if some of the myeloid DCs obtained from the single cell suspensions were making TNF-α and iNOS. We will address this question in future studies. These TIP-DCs could be part of the host inflammatory response to the tumor, to try and induce a pro-inflammatory environment and stimulate regression. Current dogma is that certain inflammatory leukocytes are good prognostic indicators, reviewed by Talmadge et al (Talmadge, 2007). For example, mature DCs, IFN-γ primed macrophages, and Th1 cells, are all leukocyte populations that may contribute to tumor regression. The human clinical situations where TIP-DCs have been described to date are both in the setting of reduced cancer: in psoriasis, where there is little cutaneous malignancy, and imiquimod-induced regressing BCCs. We have not quantitated these cells or examined sufficient SCCs to conclude this, but it is possible that TIP-DCs are also trying to induce tumor regression, but are impaired in their efforts by immunosuppressive factors produced by the tumor.
Alternatively, it is possible that the TIP-DCs fall into the group of MDSC, as one of their defining mediators is iNOS and subsequently NO. This short-lived mediator can have immunosuppressive effects such as inhibiting the proliferation of activated helper T cells (Huang et al., 1998; van der Veen et al., 2000). However, against classifying the TIP-DCs as myeloid DCs is the fact that TIP-DCs are likely to be HLA-DR+, and MDCS were defined as HLA-DR−. Future investigations will further evaluate the phenotype and function of these TIP-DCs, and how they fit into our working model of resident and “inflammatory” dermal DCs (Zaba et al., 2008b), as well as their contribution to tumor growth or suppression.
Some of the major causes for the lack of immunogenicity in cancer are associated with hypoactive DCs and the presence of regulatory T cells (Ishibashiet al., 2006; Kono et al., 2006; Larkin et al., 2006). The cytokine milieu in the tumor microenvironment may help drive these associative factors, but appears to depend on the concentration of each cytokine at any given time. Additionally, expression of the Th17-associated cytokine, IL-22, may contribute to the hyperproliferation of keratinocytes in SCC as has been described in psoriasis (Chan et al., 2006; Fitch et al., 2007; Lowes et al., 2008; Ma et al., 2008; Nickoloff, 2007). Taken together, all of these conditions pose a formidable challenge for standard chemotheraputic and biotheraputic modalities in the treatment of malignant cancers of the skin. However, a better understanding of the cytokine milieu, DC function and tumor evasion of immune responses may eventually lead to better immunomodulatory drugs and immune-based vaccine therapies applicable to carcinomas not amenable to standard treatments.
Approval from the Weill Cornell Medical College Institutional Review Board and written informed consent was obtained before enrolling patients to participate in this study, and the study was performed with strict adherence to the Declaration of Helsinki Principles.
Cutaneous SCC samples were obtained during Mohs micrographic surgery. Tumors were obtained from head and neck, trunk and extremities. Additionally, normal specimens were obtained via 3-mm punch biopsies from non-sun exposed areas of patients without skin cancer. Some of these samples may have been used in prior studies as normal controls (Guttman-Yassky et al., 2007).
Frozen tissue sections were stained as previously described (Kaporis et al., 2007). We stained sections with hematoxylin (Fisher, Fair Lawn, NJ) and eosin (Shandon, Pittsburgh, PA) and with purified mouse anti-human monoclonal antibodies to CD1a and CD11c, (BD-Pharmingen, San Diego, CA; all 1:100), Langerin, (Immunotech, Marseille, France 1:50), blood derived dendritic cell antigen (BDCA)-1/CD1c and BDCA-2/CD303 (clones AD5-8E7 and AC144 respectively, Miltenyi Biotech, Auburn, CA., both 1:100). Biotin-labeled horse anti-mouse antibody (Vector Laboratories, Burlingame, CA) was amplified with avidin–biotin complex (Vector Laboratories, Burlingame, CA) and developed with chromogen 3-amino-9-ethylcarbazole (Sigma Aldrich, St Louis, MO). Counterstaining was performed with light green (Sigma Aldrich, St Louis, MO). Appropriate isotype controls were used.
Tumor-bearing skin and non-tumor-bearing skin adjacent to excised tumor (n=10-19), and normal skin from volunteers (n=10) were evaluated for the expression of a panel of surface markers by immunohistochemistry. The regions examined within SCC samples were designated as “tumor” and “juxtatumoral dermis” (the dermis immediately surrounding the tumor). The regions examined within normal skin samples were designated as normal epidermis and normal (papillary) dermis. Normal papillary dermis and juxtatumoral dermis were compared in this study based on the structural similarities between these regions. Positive cells were counted manually, and area measures were computed by computer-assisted image analysis, National Institutes of Health software (NIH IMAGE 6.1 and IMAGE J). Cell counts per unit area (μm2 × 100000) were determined within normal epidermis, normal papillary dermis (100 μm deep to the epidermis), SCC epithelial aggregates and juxtatumoral dermis (100 μm circumferential to tumor nodules). In all cases, tumor sections with representative response were selected for analysis.
Frozen sections of squamous cell carcinoma were fixed in 100% acetone for five minutes (n=5). Non-specific staining was blocked with 5% normal goat serum and 5% normal chicken serum (both from Vector, Burlingame, CA). Tissue was then stained overnight at 4°C with two primary antibodies: (1) murine anti-TNF-FITC (clone 6401.1111, 1:10, BD Biosciences, San Jose, CA), and (2) iNOS affinity purified rabbit antibody (N-20, 1:20, Santa Cruz, Santa Cruz, CA). The following day, slides were washed and labeled with two secondary antibody conjugates for 30 minutes at room temperature: (1) goat anti-FITC-Alexa-488, and (2) anti-rabbit Alexa-594. Sections were blocked with 10% mouse serum for 30 minutes and stained overnight with murine anti-CD11c (B-ly6, 1:100, BD Pharmingen, San Diego, CA) labeled with Zenon Alexa Fluor-647 Mouse IgG1. All amplification/ detection systems were from Invitrogen/Molecular Probes (Eugene, OR). Images were acquired using appropriate filters of a Zeiss Axioplan 2 widefield fluorescence microscope fitted with a PlanNeofluar 10X/ 0.30 N.A. objective lens and a Hamamatsu Orca ER CCD camera (Hamamatsu, Bridgewater, NJ) controlled by MetaVue software (MDS Analytical Technologies, Downington, PA). TNF was visualized using a FITC filter set, iNOS with a Texas red filter set, and CD11c with a Cy5 filter set. Images in the figure are presented both as single monochrome images, and in pseudocolor (green, red and blue) located above the merged image, so that one can appreciate the localization of markers on similar or different cells. Red and green overlapping cells appear yellow in color; red and blue appear purple; and green and blue appear aqua; cells labeled with all three stains appear white. Inset shows triple-labeled cell at higher magnification (4 fold increase in image size). Dermal collagen fibers gave green auto-fluorescence. Size bar = 100μm.
SCC tumors and sight-matched peritumoral skin were obtained at surgery, and normal skin was obtained as the discarded product of abdominoplastic surgery. Subcutaneous fat was excised, and the remaining tissue was washed three times with PBS and then once with PBS containing 10% gentamicin reagent solution (Invitrogen, Carlsbad, CA). Both the dermal and epidermal layers were heavily scored with a scalpel then transferred to fresh RPMI 1640 supplemented with 10% pooled human serum (Mediatech Inc., Manassas, VA), 0.1% gentamicin reagent solution (Invitrogen, Carlsbad, CA), 1% penicillin-streptomycin solution (Sigma Aldrich, St Louis, MO) and 1% 1 M HEPES buffer (Sigma Aldrich, St Louis, MO). The tissue was incubated for 72 hours in 5% CO2 at 37°C. The supernatant was collected and filtered with 40μm cell strainers (BD Biosciences, San Jose, CA) yielding a single cell suspension. Cells were then either used immediately (for mixed leukocyte reaction [MLR]) or frozen in RPMI 1640 (Invitrogen, Carlsbad, CA) with 50% pooled human serum and 10% DMSO (ATCC) for FACS.
Cellular suspensions (SCC and matched PTNL n=3; normal skin n=3) were washed once with PBS and then stained with AquaMarina Live/Dead cell detection according to the manufacturers instructions (Molecular probes, Invitrogen, Carlsbad, CA) and then stained with the following anti-human, mouse monoclonal antibodies: HLA-DR – Alexa Fluor 700 (L243, IgG2a, 1:50; BioLegend, San Diego, CA), CD11c – PE-Cy7 (3.9, IgG1, 1:20; BioLegend, San Diego, CA), dendritic cell lysosome-associated membrane protein - PE (DC-LAMP/CD208-PE) (I10-1112, IgG1, 1:20; BD Pharmingen, San Jose, CA), dendritic cell-specific intercellular adhesion moleule-3-grabbing non-integrin – PerCP-Cy5.5 (DC-SIGN/CD209-PerCP-Cy5.5) (DCN46, IgG2b, 1:20; BD Biosciences, San Jose, CA), DEC-205/CD205-Alexa 647 (MMRI-7, IgG1, 1:33; BD Pharmingen, San Jose, CA), CD80-PE (L307.4, IgG1, 1:20; BD Biosciences, San Jose, CA), CD86-Pacific Blue (IT2.2, IgG2b, 1:50; BioLegend, San Diego, CA), and CD83-APC (Michel-19, IgG1, 1:33; BD-Pharmingen, San Jose, CA). Briefly, cells were stained in a total volume of 100μl for 30 minutes at 4°C, washed with FACSwash (PBS with 0.2% wt/vol BSA, 0.1% wt/vol sodium azide), and resuspended in 1.3% formaldehyde (Thermo Fisher Scientific, Waltham, MA) in FACSwash. Samples were acquired using a flow cytometer (LSR II; BD Biosciences, San Jose, CA) and analyzed with FlowJo software (TreeStar Inc., Ashland, OR). Appropriate isotype controls were used.
Single-cell suspensions of dermal cell crawl-outs were stained with mouse anti–human, CD11c-PE (S-CHL-3, IgG2b, 1:20, BD Biosciences, San Jose, CA) and HLA-DR – Alexa Fluor 700 monoclonal antibodies, and sorted on a FACSAria (BD Biosciences, San Jose, CA) using a low-pressure setting (n=3-4). A population of CD11c+/HLA-DRhi cells was collected and a post-sort collection was performed to confirm its purity. T cells were obtained from a normal volunteer by treatment of while blood with RosetteSep T cell enrichment cocktail as per the manufacturers instructions (Stem Cell Technologies, Vancouver, BC, Canada) then subjected to density centrifugation over Ficoll-Paque Plus (Amersham Biosciences, Piscataway, NJ), and subsequently labeled with 10μm CFSE (Carboxyfluorescein diacetate, succinimidyl ester) using the Vybrant CFDA SE Cell Tracer Kit (Invitrogen, Molecular Probes, Carlsbad, CA) in PBS with 0.1% BSA for 15 minutes at 37oC. We have adapted a modified allogeneic mixed leukocyte reaction (MLR) described in Zaba et al. (Zaba et al., 2007b). Briefly, CFSE labeled T cells were co-cultured with sorted CD11c+/HLA-DRhi cells for 8 days with and without cytokines for maturing DCs ex-vivo (IL-1β, IL-6, TNF-α, and PGE2), at a ratio of 1:50. T cells alone were used as a negative control. T cells cultured with CD3/CD28 coated beads (Dynal, Invitrogen, Carlsbad, CA), or immature MoDCs were used as positive controls. The process for making monocyte-derived DCs was previously described (Dhodapkar et al., 2002; Lee et al., 2002). T cell proliferation was analyzed on day 8 after sorting. The cultures were harvested, stained with 250ng/ml propidium iodide (PI, BD Phamingen, San Jose, CA [catalog #51-66211E]), to label dead cells, and CD3-APC (BD Biosciences, San Jose, CA) for 20 minutes at room temperature. PI-negative cells were gated and then plotted as CFSE versus CD3+ cells, where proliferating cells diluted their content of CFSE and moved to the left of the non-proliferating cells. The CFSE-low cells were quantified as a percentage of proliferating cells in the culture.
Total RNA isolation and RT-PCR reactions were performed as previously described (Kaporis et al., 2007). Briefly, SCC tumor samples, removed at Mohs surgery (n=10-12), and patient-matched site-matched peritumoral non-lesional (PTNL) skin (n=10-12), were obtained at the time of repair after clear margins were achieved. Normal skin was obtained as the waste from abdominoplastic surgery (n=4). All samples were snap-frozen and stored in liquid nitrogen. Individual frozen samples were placed in 1ml of room temperature RLT buffer with 1% β-mecaptoethanol (Qiagen, Valencia, CA) and immediately homogenized at full power for 30 seconds using a PowerGen 1000 homogenizer (Fisher Scientific, Pittsburgh, PA). Homogenates were sonicated on ice for 20 seconds at full power. DNA was removed with on-column DNAse digestion using an RNAse-free DNAse Set (Qiagen, Valencia, CA). RNA was isolated using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to manufacturer's recommendations. Total RNA concentration and purity was evaluated using an Ultraspec 2100 pro-spectrophotometer (Amersham Biosciences, Piscataway, NJ).
The primers and probes for IL-10 (Hs00174086_m1), TGF-β (Hs99999918_m1) and VEGF-A (Hs00900054_m1) were TaqMan RT-PCR individual assays designed by Applied Biosystems (Foster City, CA). The sequences of the primers and probe for HARP are: HARP-forward, CGCTGCTGAACATGCTCAA, HARP reverse, TGTCGAACACCTGCTGGATG; HARP-probe, 6-FAM-TCCCCCTTCTCCTTTGGGCTGG-TAMRA (GenBank accession no. NM-001002). The RT-PCR reaction was performed using 5ng total RNA and EZ PCR Core Reagents (Applied Biosystems, Foster City, CA) according to the manufacturer's directions. The samples were amplified and quantified on an Applied Biosystems PRISM 7900 HT using the following thermal cycler conditions: 2 minutes at 50°C, 30 minutes at 60°C, 5 minutes 95°C; and 40 cycles of 15 seconds at 95°C followed by 60 seconds at 60°C. Each sample and gene was normalized to the human acidic ribosomal protein gene, a housekeeping gene. The data were analyzed and samples quantified by the software provided with the Applied Biosystems PRISM 7900 HT.
SCC microarray data has been previously published (Haider et al., 2007). We wanted to evaluate the fold change for DC genes published in four lists: (1) immature DC upregulated genes (Haider et al., 2007), (2) mature DC up-regulated genes (Haider et al., 2007), (3) Lindstedt (Lindstedt et al., 2002) and (4) Hohenkirk monocyte-derived DCs “up” (Le Naour et al., 2001). These lists were similar: ten of the genes in the Hohenkirk list were in the immature DC list (p< 0.0001), and 9 of the Lindstedt and 11 of the Hohenkirk were in the mature DC list (p< 0.0001 for both) (analyzed in http://www.broad.mit.edu/gsea/index.jsp). We also tested the significance of gene expression findings using Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2007). GSEA is a method that determines whether a previously defined set of genes (in our case the DC gene lists) shows statistically significant differences between two biological states (skin samples, for example SCC vs. normal skin) using rank statistics. If the up-regulated genes in the DC list are ranked high in “SCCs vs. normal fold change”, the ES score will be near 1, if the opposite effect happens, then the value approaches -1. Because we were working with a several lists, a normalized enrichment score (NES) allowed us to compare the enrichment score between lists.
Statistical comparisons of relative cell counts per μm2 ×100000 was performed using an Anova mixed model in log2 scale, with P<0.05 considered significant. Statistical comparisons of mRNA expression levels was performed also using the same Anova mixed model in log2 scale with P<0.05 considered significant.
Research was supported by the Dana Foundation (Human Immunology Consortium Grant), which supports JAC, DM, KCP, LF and AP-K. MJB is supported by National Institutes of Health (NIH) grant T32-HL07423, LCZ is supported by NIH MSTP grant GM07739, MAL is supported by NIH grant 1 K23 AR052404-01A1, and MS-F is partially supported by NIH grant UL1 RR024143 from the National Center for Research Resources (NCRR). We thank plastic surgeons Drs. AN LaBruna and DM Senderoff for their generous donation of abdominoplasty surgical waste. We also appreciate the assistance and advice of the Flow Cytometry Core Facility (Dr. S. Mazel, Dr. X. Fan and C. Bare) and Bio-imaging Resource Center (Dr. A. North) at Rockefeller University. The authors do not have any conflict of interest related to this work.