ABSTRACT
Human papillomaviruses (HPVs) infect squamous epithelia and cause several important cancers. Immune evasion is critical for viral persistence. Fibroblasts in the stromal microenvironment provide growth signals and cytokines that are required for proper epithelial differentiation, maintenance, and immune responses and are critical in the development of many cancers. In this study, we examined the role of epithelial-stromal interactions in the HPV16 life cycle using organotypic (raft) cultures as a model. Rafts were created using uninfected human foreskin keratinocytes (HFKs) and HFKs containing either wild-type HPV16 or HPV16 with a stop mutation to prevent the expression of the viral oncogene E5. Microarray analysis revealed significant changes in gene expression patterns in the stroma in response to HPV16, some of which were E5 dependent. Interferon (IFN)-stimulated genes (ISGs) and extracellular matrix remodeling genes were suppressed, the most prominent pathways affected. STAT1, IFNAR1, IRF3, and IRF7 were knocked down in stromal fibroblasts using lentiviral short hairpin RNA (shRNA) transduction. HPV late gene expression and viral copy number in the epithelium were increased when the stromal IFN pathway was disrupted, indicating that the stroma helps control the late phase of the HPV life cycle in the epithelium. Increased late gene expression correlated with increased late keratinocyte differentiation but not decreased IFN signaling in the epithelium. These studies show HPV16 has a paracrine effect on stromal innate immunity, reveal a new role for E5 as a stromal innate immune suppressor, and suggest that stromal IFN signaling may influence keratinocyte differentiation.
IMPORTANCE The persistence of high-risk human papillomavirus (HPV) infections is the key risk factor for developing HPV-associated cancers. The ability of HPV to evade host immunity is a critical component of its ability to persist. The environment surrounding a tumor is increasingly understood to be critical in cancer development, including immune evasion. Our studies show that HPV can suppress the expression of immune-related genes in neighboring fibroblasts in a three-dimensional (3D) model of human epithelium. This finding is significant, because it indicates that HPV can control innate immunity not only in the infected cell but also in the microenvironment. In addition, the ability of HPV to regulate stromal gene expression depends in part on the viral oncogene E5, revealing a new function for this protein as an immune evasion factor.
INTRODUCTION
Human papillomaviruses (HPVs) infect and can cause malignancy in keratinocytes of differentiating stratified squamous epithelia (1). Genital HPV is the most common sexually transmitted infectious agent (2), and although HPV-induced cancers are widespread, most infections are benign and self-limiting (3, 4). Infections by high-risk HPVs (such as HPV16) typically last for 12 to 18 months and are usually cleared through cell-mediated immune responses (5–7), but in a minority of cases, HPV infections can last for decades, resulting in the accumulation of genetic changes needed for cancer development (6). Through the activities of viral oncogenes, HPV actively suppresses both innate and adaptive immune responses, facilitating viral persistence (8, 9).
Uninfected keratinocytes constitutively express low levels of a variety of cytokines (10), but HPV-containing keratinocytes exhibit significantly reduced expression of many inflammatory mediators, including interleukin 1 (IL-1), IL-6, and tumor necrosis factor alpha (TNF-α) (11–14). One important cytokine system targeted by HPV is the interferon (IFN) response (11, 15, 16). Type 1 IFNs (including IFN-α, -β, and others) are produced by most cell types in response to viral infections (17). Pattern recognition receptors (PRRs) detect viral infection and activate IFN response factors (IRFs) that drive expression of type I IFNs (17). Secreted type I IFNs bind to the IFN (alpha, beta, and omega) receptor (IFNAR) in both autocrine and paracrine manners and promote phosphorylation of signal transducer and activator of transcription (STAT1/2), which translocates to the nucleus to promote the transcription of a wide array of IFN-stimulated genes (ISGs) (18). ISGs include components of the antigen processing and presentation pathway, factors that intrinsically limit viral replication, and the IFNs themselves to propagate the response (18). Thus, IFNs induce a cell-intrinsic antiviral state both in infected and in neighboring cells (17). In addition to responding to viral infection, constitutive, low-level IFN production has physiological roles in maintaining signaling from other cytokines and ensuring immune readiness (19). For example, IFN-κ is a type I IFN expressed constitutively in unstimulated keratinocytes that can limit aspects of the HPV life cycle (20, 21).
HPV16 encodes E6 and E7, which are the major viral oncogenes that promote HPV-associated cancers (1, 22). Additionally, high-risk HPV types such as HPV16 encode a minor oncogenic protein, E5, which is a small, hydrophobic, and weakly immortalizing membrane protein (23). Although relatively divergent between HPV types, E5 has coevolved with E6 and E7 and thus is presumed to play a significant role in the viral life cycle in cooperation with the other viral oncogenes (24). E5 has no effect on maintenance of viral genomes in undifferentiated cells, but loss of E5 results in reduced viral DNA amplification and the expression of late genes upon differentiation (25, 26), although this effect is not seen in all HPV types (27). The best understood activity of E5 is to enhance signaling from the epidermal growth factor receptor (EGFR) (28, 29). E5 interacts with or regulates several growth factor systems, including transforming growth factor beta (TGF-β), hepatocyte growth factor (HGF), and keratinocyte growth factor (KGF) (30–34). Other functions have been reported, including inhibition of MHC-I expression (23). Thus, E5 appears to be important for regulating the response of HPV-containing keratinocytes to signals from the environment. However, the specific molecular contribution of E5 to the viral life cycle remains unclear.
Work by many investigators has uncovered a complex array of interactions between epithelial cells and fibroblasts, endothelial cells, and immune cells in the tissue stromal microenvironment (35–38). Interactions between epithelia and stroma are important for immune responses (39), and signals provided by stromal fibroblasts are increasingly recognized as critical elements in cancer development (38, 40–42). Fibroblasts are the major cell type in dermal stroma; they participate in many signaling pathways, extracellular matrix (ECM) maintenance, and other functions to ensure tissue homeostasis and produce factors important for inflammation, wound healing, and angiogenesis (reviewed in references 37 and 43). Fibroblasts are also necessary for proper keratinocyte differentiation in vivo (44–46), which is in turn necessary for the HPV life cycle (1). Cancer-associated fibroblasts (CAFs) are fibroblasts activated by paracrine mediators produced by cancer cells. CAFs respond to the cancer cells by expressing growth factors that promote cancer growth, metastasis, angiogenesis, and immune evasion (41, 47). HPVs infect keratinocytes exclusively, but many of the genes regulated by HPV oncogenes are secreted factors such as growth factors, cytokines, and IFNs that have the potential to affect stromal cells (48–50). Paracrine factors produced by stromal fibroblasts have the potential to impact the growth and invasiveness of HPV-containing epithelia (39, 51, 52), and changes in stromal fibroblast gene expression have been observed during progression of low- to high-grade cervical lesions (53, 54). However, little is understood about how fibroblasts might regulate the HPV life cycle, or vice versa, and how HPV might affect epithelial-stromal communication in general.
Feeder fibroblasts provide important two-way communication of signals to support keratinocyte growth and differentiation in many culture systems (46, 55–60). Organotypic keratinocyte cultures (raft cultures) (Fig. 1a) require fibroblasts for proper tissue morphogenesis and differentiation (61–63). The organotypic culture system is capable of supporting the full HPV life cycle in vitro because it can create conditions necessary for terminal differentiation of keratinocytes by recapitulating epithelial differentiation in three dimensions, essential to complete papillomavirus replication and virion morphogenesis (64, 65). Because organotypic raft cultures consist only of two cell types (epithelial keratinocytes and stromal fibroblasts), these cultures can serve as a model of epithelial-stromal interactions. In this study, we used organotypic raft cultures to test the hypothesis that episomal HPV in the keratinocytes of the epithelial layer can regulate gene expression patterns in stromal fibroblasts. We found that the presence of HPV16 in the epithelium impacted several pathways in the stroma, but the IFN-regulated genes were most strongly affected. The E5 oncogene was necessary for a portion of these effects. Finally, we discovered that the IFN system in stromal fibroblasts is able to dampen the expression of viral genes in the epithelium, with a preferential impact on late viral gene expression.
Organotypic cultures contain both epithelial and stromal compartments. (a) Organotypic raft cultures are created by mixing type I collagen with fibroblasts to create a dermal equivalent. Keratinocytes are seeded and allowed to come to confluence. The whole construct is lifted to the air-liquid interface on a wire grid. When exposed to the air, keratinocytes will stratify and differentiate to form a fully stratified epithelium. (b) In this study, rafts composed of human fibroblasts with HFKs, HPV16+ keratinocytes, or E5 Stop cells were cultured for 14 days. RNA from the fibroblasts in the stroma was subjected to microarray analysis. (c) H&E sections from representative raft cultures from this study. The stromal and epithelial compartments are labeled on the HFK image. White arrows in the HPV16+ image indicate the presence of koilocytes, which are largely absent in samples from the other cell types.
RESULTS
HPV16 in the epidermis of raft cultures regulates gene stromal gene expression.The organotypic raft culture system can recapitulate epithelial differentiation in three dimensions, which is essential to complete papillomavirus replication and life cycle (64, 65). A mixture of type I collagen and primary human foreskin fibroblasts forms a dermal equivalent, while keratinocytes seeded on top are equivalent to epidermis (Fig. 1a). When lifted to the air-liquid interface, the keratinocytes exposed to the air stratify and differentiate into a full-thickness epidermis (Fig. 1c). To determine how gene expression is altered in stromal fibroblasts during HPV16 infection in keratinocytes, a microarray analysis was performed on stromal samples from 14-day-old organotypic raft tissue grown with early-passage human foreskin keratinocytes (HFKs) or HFKs stably transfected with the HPV16 genome (66) (Fig. 1b). These HPV16+ cells are immortalized but not transformed, harbor episomally replicating HPV16, and resemble a benign low-grade lesion when grown in raft culture (66). Combinations of early-passage HFKs, HPV16+ cell lines, and human foreskin fibroblasts (HFFs) from different donors were used in replicate cultures to control for the effect of donor variability. Following 14 days of culture, the epithelial and stromal compartments were physically separated, and total RNA from the stroma was subjected to microarray analysis. Although good differentiation and cornification were observed for HFK and HPV16+ rafts, E5 Stop rafts (with HPV16 harboring a stop codon mutation in the E5 open reading frame) (see below) had poor cornification overall (Fig. 1c).
When stromal samples from rafts grown with uninfected HFKs were compared to stromal samples from rafts having HPV16+ keratinocytes, 964 genes were altered at least 1.5-fold. Of these 964 genes, 540 genes were upregulated, while 424 genes were downregulated (see Table S1 in the supplemental material). The online gene ontology analysis tool GOrilla (67) was used to determine the significantly affected pathways in the stromal microenvironment due to HPV infection in epithelia. Further, REViGO (68) was used to analyze the gene ontology (GO) list generated through GOrilla and narrow down the list of genes to a set of nonredundant and significant GO terms for simplified visualization. Tables 1 and 2 summarize the gene ontogeny analysis of the down- and upregulated genes, respectively. Table 3 shows the 25 most upregulated and the 25 most downregulated individual genes in the microarray analysis. We expected that the presence of an oncogenic virus in the epithelium would primarily induce the expression of growth factors or other CAF-associated changes in the fibroblasts. Among the pathways downregulated in the stromal microenvironment due to HPV infection in epithelia were extracellular matrix organization, extracellular structure organization, and locomotion (regulation of cell migration, regulation of locomotion, and regulation of cellular component movement). However, the clearest theme among the downregulated pathways was immune defense responses (Table 1) (defense response to virus, type I interferon signaling pathway, immune system process, immune effector process, and cytokine-mediated signaling pathway) and ECM regulation. Seventeen of the top 25 genes downregulated in stromal fibroblasts by the presence of HPV16 in epithelium were identified as immune system related or interferon-stimulated genes (ISGs), including IFI27, MX1, OAS2, IFI44L, IFIT1, IFIT3, DDX58, and STAT1 (Table 3). The GO terms for upregulated pathways in Table 2 and the upregulated specific genes listed in Table 3 did not fall into a clear theme, although amino acid transport (SLC3A2, SLC38A1, and SLC7A11), locomotion/cell migration/movement of cell or subcellular components (SLC3A2, ITGB8, SLC7A11, CD44, MYO19, TNFRSF10D, and FAT1), and wound healing were some commonalities. GO terms associated with the upregulated genes in the stromal microenvironment because of HPV infection in epithelium included diverse pathways such as cornification, cell death, water homeostasis, wound healing, and locomotion among others (Table 2). The identification of cornification in the upregulated pathways in the stroma has been observed before (69) and may be due to mesenchymal-to-epithelial transition or to residual epithelial cells contaminating the stromal samples (see below). These results indicate that although HPV is confined to the epithelial cells, it can influence stromal gene expression patterns in a paracrine manner in the underlying stroma. Furthermore, the primary effects are downregulation of innate immune responses and altered ECM organization.
Pathway analysis of genes downregulated in HPV16 versus HFK rafts
Pathway analysis of genes upregulated in HPV16 versus HFK rafts
Genes altered in rafts containing wild-type HPV16 versus HFK
We assumed that the viral oncogenes were largely responsible for these changes in stromal gene expression. Loss of E6 or E7 from the viral genome prevents episomal replication and immortalization (70), making it difficult to examine the roles of these genes in the context of the viral genome in our experimental system. However, the oncogene E5 is known to regulate responses of the infected cell to its environment (23), and HPV16 harboring a stop codon mutation in the E5 open reading frame could be maintained episomally in keratinocytes (E5 Stop cells) (34). These cells maintain the HPV16 genome at levels equivalent to the wild type and have normal levels of E6 and E7 expression (34). They are therefore a valuable tool to understand the role of E5 in the context of the episomal genome. Therefore, to determine whether there is a role of HPV16 E5 on gene expression in stroma, rafts were created using E5 Stop keratinocytes. Note in Fig. 1c the abundant koilocytes in the wild-type HPV16+ raft that were lacking in the E5 Stop raft, consistent with the report that E5 is required for koilocytosis (71). Array analysis was performed using stromal RNAs as before. When we analyzed the microarray data for stromal gene expression from rafts grown with E5 Stop cells compared to stroma from rafts containing wild-type HPV16-containing cells, 381 genes were altered 1.5-fold or more (see Table S2). Of these, 215 were found to be upregulated, while 166 were downregulated. An analysis of pathways upregulated in the stromal microenvironment due to HPV infection in rafts grown with E5 Stop cells compared to those in wild-type HPV16 cells revealed nucleosome assembly, extracellular matrix organization, locomotion (regulation of cell migration, membrane-to-membrane docking, and membrane raft organization) and cell cycle-related (cell-division and mitotic cell cycle process) pathways (Table 4). Remarkably, defense response pathways (immune system process, regulation of type III interferon production, cytokine-mediated signaling pathway, and inflammatory response) were also upregulated compared to that in stroma from rafts containing wild-type HPV16 (Table 4), suggesting that E5 may contribute to the ability of HPV16 to suppress immune responses in stromal fibroblasts. Pathways affected among downregulated genes belonged to heterogeneous GO terms such as cornification, cell death, cell junction organization, water homeostasis, response to wounding, and cell matrix adhesion (Table 5). Table 6 shows the top 25 upregulated individual genes; only 18 genes were downregulated in rafts containing E5 Stop cells versus that in wild-type HPV16. Among upregulated genes, 11 genes were recognized to be part of the immune response system (RSAD2, IFIT1, MUC4, HERC5, OASL, CRISPLD2, DDX60, IFI44L, CCL7, DDX58, and LCN2) or extracellular matrix organization (CRISPLD2, SERPINE1, SULF1, and ELN). The downregulated genes included many noncoding RNAs (Table 6). The significance of this finding is not clear, although one of the ribosomal 5S pseudogenes was recently found to regulate IFN responses through the RIG-I pathway (72), and the most downregulated gene, H19, is thought to be involved in insulin-like growth factor signaling and epithelial to mesenchymal transition (EMT) (73).
Pathway analysis of genes upregulated in E5 Stop versus HPV16 rafts
Pathway analysis of genes downregulated in E5 Stop versus HPV16 rafts
Genes altered in rafts containing E5 Stop versus wild-type HPV16
Differing effects of E5 on stromal gene expression.Among the top 25 pathways affected by epithelial HPV16 in stromal fibroblasts, two processes stand out clearly—extracellular matrix organization and immune system function. It appeared that several genes and pathways related to these processes were downregulated in the stromal microenvironment in response to HPV infection in the epithelium (Tables 1 and 3) but were upregulated in rafts grown with HPV16 E5 Stop cells (Tables 4 and 6). This effect would suggest a role of HPV16 E5 in the suppression of the stromal immune response and ECM remodeling. Sixty-three and 69 genes were found to be associated with extracellular matrix (ECM) organization (GO:0030198) and immune response (GO:0006955), respectively, in our microarray data set. In order to better visualize the effect of E5 in the suppression of stromal gene expression, heat maps were created focusing on these two sets of downregulated genes using ClustVis (Fig. 2). Unsupervised hierarchical clustering organized genes into groups based on expression patterns. In each heat map, column 1 represents the expression levels of genes in stroma of rafts grown with HFK only, the middle column represents the expression profile in stroma from rafts with keratinocytes harboring wild-type HPV16, and column 3 represents gene expression in stroma of rafts grown with HPV16 E5 Stop cells. In both the ECM and immune response heat maps, gene expression was suppressed in the presence of wild-type HPV, as expected from the overall microarray data. Heat map analysis revealed that the effect of E5 on stromal gene expression could be divided into four general patterns. First, group 1 toward the tops of the heat maps includes genes whose expression levels were further suppressed in stroma from rafts grown with HPV16 E5 Stop cells compared to that in HPV16 cells. Next, group 2 includes genes that were suppressed approximately equally in the presence or absence of E5. These genes were presumably suppressed by the actions of E6 and/or E7 (which are expressed at normal levels in E5 Stop cells [34]) with no contribution from E5. Group 3 includes genes whose expression levels were partially restored, indicating that E5 contributed to suppression but is not responsible for the whole effect. Finally, group 4 includes genes whose levels in E5 Stop-containing rafts were equal or exceed the levels in HFK-containing rafts. These are genes whose suppression in the stroma depended entirely on the presence of E5 in the epithelium. In both the ECM and immune-related gene sets, expression levels were at least partially recovered in the absence of E5 for the majority of the genes (groups 3 and 4), with E5 having the more marked effect in the immune-related set compared to that in the ECM-related set. Pathway analysis of the individual gene groups did not reveal any common molecular features that could account for their differential regulation by E5 (not shown). These findings demonstrate a novel activity of E5 in influencing the expression of ECM and immune-related genes in the stromal fibroblasts of raft cultures in a paracrine manner.
Heat map analysis of ECM maintenance and immune response genes regulated in the stroma by HPV16. Genes identified as belonging to the ECM and immune response gene ontogenies (GO:0030198 [ECM] and GO:0006955 [immune response]) were used to create heat maps using ClustVis. Red indicates higher levels of gene expression, while blue indicates lower levels. Unsupervised hierarchical clustering was used to organize the genes into categories based on expression patterns. Groups of genes were identified based on the pattern of expression in E5 Stop cells, thus representing the impact of E5 on expression of these genes in the stroma.
Verification of gene expression in stromal fibroblasts in response to HPV16 in the epithelium.To confirm our microarray results, verification studies were carried out using reverse transcriptase quantitative PCR (RT-qPCR), including both the same samples used for microarray analysis and newly prepared samples with different keratinocyte and fibroblast donors. Because of the critical role of immune evasion in the persistence of HPV infections (8, 9), we chose to focus subsequent analyses on immune-regulated genes in the stroma. Representative interferon-stimulated genes (ISGs) were selected from the microarray results. For all selected genes, downregulation was detected in stromal fibroblasts in response to the presence of HPV16 in the epithelium (Fig. 3). In some cases, this downregulation was dramatic, as in the 115-fold downregulation of IFIT1, while in other cases, the downregulation was less so, as with 40% downregulation of SMAD3 (Fig. 3a to c); but, in all cases, patterns consistent with the microarray were observed. Consistent with the patterns shown in Fig. 2, expression levels of some but not all of these genes were increased in the absence of E5. For some genes, including CCL7, IFI44L, IFIT1, Mx1, and OAS2, suppression by HPV16 was less efficient by at least 2-fold when E5 was absent, suggesting that E5 plays an important role in their suppression (Fig. 3a). In other cases, such as CXCL12, IFI44, PARP9, STAT1, and XAF1, suppression by HPV16 was fully retained in the absence of E5 (Fig. 3b). In some cases, transcripts were recovered to similar or higher levels in E5 Stop-containing rafts compared to that in HFK-containing rafts (Fig. 3c). The cases in which loss of E5 did not fully restore expression levels to that seen in HFK rafts indicate that other viral factors contributed to the suppressive effect of wild-type HPV16. A few genes were upregulated in stromal fibroblasts in response to HPV16. RT-qPCR confirmed that levels of two genes, CXCL8 (IL-8) and long noncoding RNA H19, were upregulated due to HPV16, and this upregulation was either lost or reduced in rafts containing E5 Stop cells (Fig. 3d). These results confirmed the microarray results indicating that ISG levels in stromal fibroblasts are reduced in the presence of HPV16 and that at least part of this effect is dependent on E5.
RT-qPCR was used to confirm stromal gene expression patterns identified in the microarray experiment. (a, b, and c) Interferon response genes whose expression levels were suppressed in stroma from rafts containing wild-type HPV16+ cells and which were recovered to various degrees in rafts containing E5 Stop cells. (d) H19 and CXCL8 mRNA levels were increased in the stroma in the presence of wild-type HPV16+ keratinocytes but to a lesser degree in the presence of E5 Stop keratinocytes. All significance relative to the HPV16 values. *, P < 0.05; **, P < 0.01; ***, P < 0.001; NS, not significant.
Immunofluorescence analysis of raft cultures.Further validation of microarray and RT-qPCR results was carried out by immunofluorescence (IF) staining of raft tissue sections to determine whether expression changes were seen at the protein levels. IF staining also allowed us to confirm that the changes in ISG levels were occurring in the stromal compartment and were not due to residual epithelial tissue adhering to the dermal equivalent at the time of raft harvest. As shown in Fig. 4a, staining intensity for IFIT1 was significantly decreased in stroma of rafts grown with HPV16 in epithelium compared to those with no virus in the epithelium. Furthermore, staining intensity was partially regained in the stroma when E5 Stop cells were present in the epithelium (Fig. 4a). Quantification of these results revealed that although the number of cells positive for IFIT1 did not differ depending on the presence of the virus, the intensity of staining was reduced in an E5-dependent manner (Fig. 4b and c). A similar pattern was observed for OAS2 and CCL7 (Fig. 5 and 6). Intensity of staining for IFI44L was also reduced by HPV16, but this reduction did not depend on E5 (Fig. 7). On the other hand, CXCL8 was minimally expressed in the stroma in the absence of HPV16 in the epithelium, but both the number of positive cells and the intensity of staining increased when HPV16 was present (Fig. 8). CXCL8 levels in the stroma returned to levels similar to that in HFKs when HPV16 E5 Stop cells were present in the epithelium. These findings are consistent with the RT-qPCR results described above that HPV16 can regulate stromal innate immune responses in a paracrine manner dependent in part on E5.
Stromal IFIT1 levels. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for IFIT1. (Top) IFIT1 (green); (middle) magnified image of the inset from the top row; (bottom) overlay of IFIT1 signal with DAPI (blue). Quantification of IFIT1 signals from three sections from three independent cultures indicating the percentage of IFIT-positive cells (b) and intensity of IFIT1 staining (c). ***, P < 0.001; NS, not significant.
Stromal OAS2 levels. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for OAS2. (Top) OAS2 (green); (middle) magnified image of the inset from the top row; (bottom) overlay of OAS2 signal with DAPI (blue). Quantification of OAS2 signals from three sections from two independent cultures indicating the percentage of OAS2-positive cells (b) and the intensity of OAS2 staining (c). **, P < 0.01; ***, P < 0.001; NS, not significant.
Stromal CCL7 levels. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for CCL7. (Top) CCL7 (green); (middle) magnified image of the inset from the top row; (bottom) overlay of CCL7 signal with DAPI (blue). Quantification of CCL7 signals from three sections from two independent cultures indicating the percentage of CCL7-positive cells (b) and the intensity of CCL7 staining (c). ***, P < 0.001; NS, not significant.
Stromal IFI44L levels. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for IFI44L. (Top) IFI44L (green); (middle) magnified image of the inset from the top row; (bottom) overlay of IFI44L signal with DAPI (blue). Quantification of IFI44L signals from three sections from three independent cultures indicating the percentage of IFI44L-positive cells (b) and the intensity of IFI44L staining (c). ***, P < 0.001; NS, not significant.
Stromal CXCL8 levels. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for CXCL8. (Top) CXCL8 (green); (middle) magnified image of the inset from the top row; (bottom) overlay of CXCL8 signal with DAPI (blue). Quantification of CXCL8 signals from three sections from two independent cultures indicating the percentage of CXCL8-positive cells (b) and the intensity of CXCL8 staining (c). **, P < 0.01; ***, P < 0.001; NS, not significant.
Regulation of stromal STAT1 signaling.To gain insight into the mechanism of this effect, we used IF to examine levels of STAT1, a critical transcription factor driving expression of ISGs in response to IFN signaling and which is itself an ISG (17). STAT1 staining was decreased in stromal fibroblasts in rafts containing HPV16 in the epithelium compared to that in the absence of HPV16 infection. Furthermore, the levels of STAT1 increased when HPV16 E5 Stop cells were present in the epithelium (Fig. 9). This pattern was seen at the protein level through IF even though increased levels of STAT1 mRNA were not observed in E5 Stop stroma by RT-qPCR (Fig. 3b). Careful examination of the images indicated that STAT1 staining, even the residual staining seen in stroma from HPV16-containing rafts, overlapped the nuclei of fibroblasts, suggesting that it was at least partly activated. We thus stained for STAT1 phosphorylated on tyrosine 701 (p-STAT1), which is indicative of STAT1 activation by the IFNAR/JAK pathway (17). Similarly to total STAT1 levels, we observed that p-STAT1 was detectable in rafts containing uninfected HFKs but that the number of cells showing p-STAT1 signal was reduced in rafts containing wild-type HPV16 and again increased in rafts containing E5 Stop cells (Fig. 10). Both wild-type and E5 Stop HPV16 reduced the per-cell intensity of p-STAT1 staining, even though E5 was required to reduce the number of cells positive for p-STAT1 (Fig. 10b versus c). p-STAT1 staining was observed primarily in the nucleus with some cytoplasmic signal. These findings are consistent with the hypothesis that the IFNAR/JAK/STAT1 pathway is constitutively active in the stroma of rafts containing uninfected HFKs but that the pathway is suppressed by HPV16 in a manner at least partially dependent on E5.
Stromal STAT1 levels. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for total STAT1. (Top) Total STAT1 (green); (middle) magnified image of the inset from the top row; (bottom) overlay of total STAT1 signal with DAPI (blue). Quantification of total STAT1 signals from three sections from three independent cultures indicating the percentage of total STAT1-positive cells (b) and the intensity of total STAT1 staining (c). ***, P < 0.001; NS, not significant.
Regulation of STAT1 phosphorylation in the stroma. (a) Paraffin-imbedded sections of rafts containing the indicated cell types in the epithelium were stained with antibody specific for STAT1 phosphorylated at tyrosine 701 (p-STAT). (Top) p-STAT (green); (middle) magnified image of the inset from the top row; (bottom) overlay of p-STAT signal with DAPI (blue). Quantification of p-STAT signals from three sections from two independent cultures indicating the percentage of p-STAT-positive cells (b) and the intensity of p-STAT staining (c). **, P < 0.01; ***, P < 0.001; NS, not significant.
Low levels of STAT1 activation in the stroma of rafts containing HPV16 suggested that the IFNAR/JAK/STAT pathway was activated at lower levels in the stroma of these rafts than in HFK rafts. HFKs are known to produce constitutive levels of various cytokines, including IFN-κ, which HPV is known to suppress (19–21, 74, 75). We therefore examined the levels of type I IFNs in the epithelial layer of raft cultures. Because these cultures were not stimulated with IFN or with PRR ligands, levels of IFN-α and IFN-β were low and somewhat variable in all samples (cycle threshold [CT] values of 32 to 34), but they were lower in HPV16+ rafts than in HFK or E5 Stop rafts (Fig. 11a). IFN-κ transcripts were strongly expressed in HFK rafts (CT, 22 to 24), but were much lower in HPV16+ rafts; E5 Stop rafts epithelia also had low levels of IFN-κ transcripts, which did not correspond to the increased levels of ISGs in the stroma of these rafts.
(a) Raft cultures were created with the indicated keratinocyte cell types. Following 14 days of incubation, RNA was harvested from the epithelial layer and subjected to RT-qPCR analysis for the indicated mRNAs. Significance was calculated in comparison to HPV16+ cells. (b) Conditioned supernatant from the indicated cell type (or unconditioned control medium) was added to monolayers of HFFs and incubated for 24 h. RNA was isolated and subjected to RT-qPCR analysis for the indicated mRNAs. *, P < 0.05; ***, P < 0.001; NS, not significant.
To clarify the effect of HPV on ISG induction in fibroblasts, supernatants were conditioned by incubation with monolayer cultures of HFKs, HPV16+ cells, and E5 Stop cells. Control medium was incubated without keratinocytes. Supernatants were added to monolayer cultures of HFFs and incubated for 24 h, and RNA was subjected to RT-qPCR analysis for several ISGs. As seen in Fig. 11b, levels of ISG transcripts were generally increased in HFFs treated with HFK supernatant compared to that in the control. This increase was either not observed or was reduced when HFFs were treated with supernatant from HPV16+ cells, but supernatants from E5 Stop cells increased ISG mRNA levels similar to that in HFKs for two-thirds of the ISGs. These results indicate that keratinocytes secrete a soluble factor that can induce ISGs in fibroblasts but that HPV16 can inhibit this factor in a manner partially dependent on E5.
Knockdown of IFN pathway in stromal fibroblasts.The finding that IFN levels in epithelia of rafts containing HPV16 were lower than in HFK rafts suggests that the changes in stromal ISGs were simply a reflection of IFN production in the epithelium. Because HPV is well known to inhibit IFN responses (11, 15, 16), this may be to some extent a trivial observation. We therefore chose to investigate whether the stromal IFN response has consequences for HPV replication and gene expression in the epithelium. We investigated four factors in the IFN response pathway (Fig. 12a): IRF3 and IRF7, which are important for the production of IFN, and IFNAR1 and STAT1, which are important for sensing and signaling in response to IFN. Although the other factors are present in unstimulated cells, IRF7 is not normally present in the cell but is induced following IFN treatment to amplify the response (18). When we examined IRF7 levels in the fibroblasts of organotypic cultures, we found that “rafts” consisting of fibroblasts alone grown without an epithelium had low IRF7 levels, but IRF7 was induced in the presence of HFKs, consistent with the constitutive secretion of IFN-κ and other cytokines from normal keratinocytes (Fig. 12b). IRF7 was induced less efficiently in the presence of HPV, however, indicating that the IRF7-inducing signal was inhibited by HPV16. E5 Stop rafts showed induction of IRF7 comparable to that for HFKs, suggesting that E5 is necessary to reduce the induction of IRF7 in the stroma (Fig. 12b).
HPV16 suppresses constitutive stromal IRF7 induction by keratinocytes. (a) Diagram of the IFN production and signaling pathways. Shaded components were knocked down in this study. (b) Rafts were grown with fibroblasts in the absence of keratinocytes (none) or with the indicated keratinocyte types. Following 14 days of growth, RNA was isolated from the stromal fibroblasts and subjected to RT-qPCR for to measure IRF7 mRNA levels. Statistics are relative to rafts lacking keratinocytes (none). **, P < 0.01.
IRF3, IRF7, IFNAR1, and STAT1 were stably knocked down in human foreskin fibroblasts (Fig. 13a to d). We found that constitutive IRF7 protein levels were low in HFFs grown in monolayer culture, and so RT-qPCR was used to confirm knockdown. Raft cultures were created using these knockdown HFF strains in the stroma and wild-type HPV16+ cells in the epithelium. Stratification and differentiation of these cultures were grossly normal as observed by hematoxylin and eosin (H&E) staining (Fig. 13e and f). Rafts were cut in half for DNA and RNA isolation from the same cultures. Viral DNA levels in the epithelium were higher in rafts containing STAT1, IRF3, and IRF7 knockdown fibroblasts but were unchanged in rafts containing IFNAR1 knockdown fibroblasts (Fig. 14a). This suggests that stromal IFN signaling generally acts to suppress HPV replication but that different components of the pathway may make different contributions. Levels of early (E6/E7), intermediate (E1̂E4), and late (L1) viral transcripts were measured by RT-qPCR and found overall to be elevated in the knockdown cultures (Fig. 14b to d). Again, although knockdown of the different IFN signaling components had different effects on viral transcripts, the overall effect was positive, indicating that stromal IFN signaling inhibits viral gene expression. The effect of knocking down the stromal IFN response was most obvious on L1 transcript levels compared to that for other viral transcripts. This finding suggests that the stromal IFN response has a particularly important effect on the late stage of the HPV life cycle. Comparing the effect of stromal IFN signaling on viral DNA versus RNA levels, we observed that although viral replication and gene expression were both suppressed by stromal IFN signaling, the levels of viral transcripts were not proportional to the levels of viral DNA in the same culture (for example, compare the effects of IFNAR1 knockdown [KD] and STAT1 KD). This indicates that the increase in viral transcripts was not simply due to an increase in viral templates available for transcription.
Western blots showing stable knockdown of IRF3 (a), STAT1 (b), or IFNAR1 (c) in human fibroblasts along with nontarget controls (NTC). (d) RT-qPCR of IRF7 mRNA levels in NTC or IRF7 KD human fibroblasts. H&E sections of rafts grown with wild-type HPV16+ keratinocytes and fibroblasts with NTC (e) or the indicated factor knocked down (f). ***, P < 0.001.
Stromal IFN signaling suppressed HPV16 late gene expression. Rafts were grown with wild-type HPV16+ keratinocytes in the epithelium and fibroblasts in the stroma with IFNAR1, STAT1, IRF3, or IRF7 knocked down. (a) Following 14 days of growth, total DNA was harvested from the epithelial layer and subjected to qPCR analysis. (b to e) Rafts were created with the indicated genes knocked down in stromal fibroblasts. RNA was harvested from the epithelial layer and subjected to RT-qPCR analysis to measure levels of the indicated viral transcripts. NTC, nontarget control; *, P < 0.05; **, P < 0.01; ***, P < 0.001.
To better understand the molecular basis of suppression of HPV late gene expression by stromal IFN, we hypothesized that loss of the stromal IFN response may result in lower levels of IFN signaling in the epithelium. When we measured levels of IFNA, IFNB, and IFNK transcripts in epithelia from our raft cultures, we found that the levels of these IFN genes were either unchanged or increased rather than decreased in the presence of reduced IFN signaling in the stroma (Fig. 15a and b). SP100, an ISG previously shown to be important in suppressing HPV gene expression (76), was reduced in the IFNAR1 KD rafts but not changed in the others (Fig. 15b). These findings are not compatible with the idea that reduced stromal IFN signaling results in a corresponding reduction in epithelial IFN and suggest that the derepression effect on viral gene expression of knocking down the IFN response in the stroma was not due to reduced IFN responses in the epithelium.
Stromal IFN signaling regulates terminal keratinocyte differentiation. Rafts were grown with wild-type HPV16+ keratinocytes in the epithelium and fibroblasts in the stroma with STAT1, IFNAR1, IRF3, or IRF7 knocked down, or NTC fibroblasts. Following 14 days of growth, RNA was harvested from the epithelial layer and subjected to RT-qPCR analysis to measure levels of IFN-α, IFN-β (a), IFN-κ, Sp100 (b), or the keratinocyte differentiation markers keratin 10 (K10), transglutaminase 1 (TGM1), filaggrin (FLN), or loricrin (LOR) (c). *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Because the effect of knocking down stromal IFN signaling was most obvious for the L1 transcript, we considered the possibility that keratinocyte differentiation may be affected by stromal IFN. L1 expression is strictly dependent on keratinocyte differentiation and occurs at approximately the granular layer of the epithelium (77, 78). We measured the levels of four keratinocyte differentiation markers—keratin 10 (K10), transglutaminase 1 (TGM1), filaggrin (FLN), and loricrin (LOR)—to determine whether the keratinocyte differentiation program is affected in rafts with disabled stromal IFN signaling. K10 and TGM1 transcript levels were not consistently affected in the knockdown rafts (Fig. 15c). However, FLN and LOR were markedly increased in the knockdown rafts. Similarly to L1, FLN and LOR are expressed in the granular layer of the differentiating epithelium (79). The increase in FLN transcripts was reflected in increased filaggrin protein levels in the epithelia of knockdown rafts (Fig. 16). Thus, at least in an organotypic context containing HPV16, constitutive stromal IFN signaling exerts an inhibitory effect on late keratinocyte differentiation.
Knockdown of the IFN-related factors in stromal fibroblasts enhances filaggrin expression in HPV16+ epithelia. Rafts were grown with HPV16+ keratinocytes in the epithelium and fibroblasts in the stroma with NTC (a), STAT1 (b), IFNAR1 (c), IRF3 (d), or IRF7 (e) knockdown shRNAs. Paraffin-embedded sections were stained with antibody specific for filaggrin (FLN, green) and counterstained with DAPI (blue). Images are representative of at least three individual rafts for each knockdown.
DISCUSSION
In this study, we have used organotypic raft cultures as a model to study how HPV infection in the epithelium impacts gene expression in stromal fibroblasts. Our results indicate that many genes and pathways are regulated by HPV in the stroma, consistent with the wide array of soluble factors dysregulated in infected keratinocytes by the viral oncogenes (48–50, 54, 80, 81). Most previous studies related to global gene expression patterns during HPV infection have focused on the effect of HPV infection on the epithelium, and mostly as relating to carcinogenesis (49, 54, 82–88). Some gene expression profiling studies have considered stromal changes in the context of HPV infection and cancer development (53, 54, 88, 89), but a clear picture of how the stroma impacts HPV (and vice versa) during benign productive infection is lacking. Organotypic raft cultures represent a stripped-down model of epithelial-stromal interactions in which only two cell types are present, keratinocytes and fibroblasts, and both are amenable to experimental manipulation. This model has potential to define clear molecular mechanisms that mediate communication between stroma and the epithelium and the impact on HPV pathogenesis. As a first effort in this direction, we have shown that (i) HPV16 can regulate stromal gene expression in a paracrine manner, (ii) the most significant pathways regulated are innate immunity and ECM maintenance pathways, (iii) stromal IFN responses act to dampen the late stage of the HPV16 life cycle, probably by regulating the late stages of keratinocyte differentiation, and (iv) the poorly understood HPV oncogene E5 can act as a regulator of IFN responses in the tissue microenvironment.
Some previous studies have examined the role of the stroma in HPV-associated cancers (53, 69). Of these, the most similar to the present study is that by Spurgeon et al., in which E6/E7 transgenes were expressed in the cervical epithelia of mice and the resulting epithelial and stromal gene expression changes were determined using laser-capture microdissection and array analysis (69). Similarly to the present study, those authors found that the viral oncogenes had a major effect on stromal gene expression, with more stromal genes suppressed in the presence of the oncogenes than activated. In particular, ECM-related factors were significantly suppressed. Because our studies were not performed in an animal, we did not observe the contribution of inflammatory cells, with the resulting gene expression changes, seen in their results (69). On the other hand, our studies were performed using the complete viral genome in human cells, which allowed us to uncover a role for E5, which was not a factor in their work. The contribution of other differing factors—species, tissue of origin, hypoxia, and the presence of other cell types—will no doubt be worked out through additional experiments using different but complementary experimental approaches. It is noteworthy that Spurgeon et al. (69) observed the upregulation of typically epithelial-specific genes and pathways in their stromal samples, which we also observed (Table 2). The significance of that finding remains to be determined.
Among the pathways regulated in the stroma by HPV16 were immune response and ECM maintenance. A critical physiological function of fibroblasts is the maintenance of the ECM, and this function is frequently dysregulated during cancer development (35). Many ECM-related genes are upregulated in both the stromal and the epithelial cells in cervical cancers (88). Low-grade productive HPV infections are not cancers, but they have some features that resemble cancers, such as dysregulated proliferation, resistance to apoptosis, and immune evasion (90). Although we did not observe that HPV infection in the epithelium drives tumor-promoting CAF differentiation, our array results suggest that HPV16 may be able to impact the ECM maintenance function of stromal fibroblasts. What effect this regulation may have on the viral life cycle is not yet known.
HPV carries many tools to suppress IFN responses (9, 39). This study reveals two more: that the ability of HPV to suppress IFN responses is not just in the infected cells but also in the stromal microenvironment, and that E5 can serve as an innate immune suppressor. Consistent with our studies, suppression of immune genes has been observed in very early (CIN1) HPV-containing lesions, and major changes in stromal gene expression were observed starting around CIN2 (54). Gene expression changes were also observed in adjacent normal epithelia as well as stroma (54), confirming that HPV-infected epithelia have a significant paracrine effect on the tissue microenvironment. It should be noted that the IFN pathway was not artificially stimulated in any of our studies. All of the changes we have observed represent “baseline” activities of the cells under long-term culture conditions. IFNs are constitutively expressed in epithelial keratinocytes, resulting in constitutive low levels of innate immune activation in these tissues (10, 20). Three pieces of data from our studies reflect this observation. First, p-STAT1 staining was observed in fibroblasts of rafts grown with uninfected HFKs, suggesting that IFN produced by the HFKs was stimulating the fibroblasts constitutively (Fig. 10). Second, we observed that levels of IRF7 transcripts, which require IFN stimulation in order to be expressed (17), were induced by the addition of HFKs to the raft cultures compared to “rafts” made with HFFs alone (Fig. 12). Third, conditioned supernatants from HFKs were able to induce ISG expression when added to HFFs (Fig. 11b). Thus, the raft culture system reflects an important aspect of constitutive innate immune activity of the epithelia in vivo. It is likely that the constitutive activation of IFNs in epithelial keratinocytes represents a barrier for the establishment and maintenance of HPV infections (21, 91). Importantly, both of these markers of chronic IFN signaling were reduced by HPV16 in our experiments.
An important advance provided by our data is that E5 is also able to suppress IFN responses in stromal fibroblasts in a paracrine manner. E5 Stop-containing epithelium did not show higher levels of IFN-κ, and the other type I IFNs were expressed only at very low levels (Fig. 11a). However, conditioned supernatants from E5 Stop cells grown in monolayer were able to stimulate expression of ISGs when incubated with HFFs (Fig. 11b). It is therefore not immediately clear what the signal is that drives suppression of stromal IFN in an E5-dependent manner. Further studies will be required to clarify this issue. Previous studies have shown that mutations in HPV that prevent E5 expression generally do not interfere with the ability of the virus to immortalize or replicate episomally but that the efficiency of the viral life cycle is reduced in the late stages (25, 26). E5 is conserved among high-risk HPV types, and so the modest nature of this effect as well as its molecular basis has been a mystery. If, as we propose, E5 acts as an immune suppressor, then it could explain why acute effects on viral replication are not seen in vitro when E5 is absent. The fact that E5 mutants tend to have late rather than early defects can also be explained by our observation that stromal IFN signaling regulates the late phase of the viral life cycle. A late effect of IFN on HPV is not unprecedented. Sp100, an ISG, was found to inhibit the late rather than early stage of the viral life cycle (76). Furthermore, STAT1, which normally increases upon differentiation in keratinocytes, can impair late events in the HPV life cycle (92). Our observation that stromal IFN responses tend to have a greater effect on L1 mRNA expression than on early transcripts (Fig. 14) and on late rather than early differentiation markers (Fig. 15) would be consistent with the idea that IFN acts late in the viral life cycle.
E6 and E7 are both known to interfere with the IFN response, and both have been shown to regulate the expression of soluble factors that are capable of acting on the stroma (39, 48, 81). Because E6 and E7 are both expressed at normal levels in E5 Stop cells (34), the suppressive effects of these two oncogenes would still be present. This explains the observation that E5 Stop cells retained a portion of wild-type HPV16’s ability to suppress stromal innate immune responses. Further studies will be required to dissect the individual contributions of each viral oncogene. It was also interesting that suppression of some ISGs, which are defined by their common dependence on IFN, depended significantly on E5 expression, while others were suppressed equally regardless of E5. The mechanisms behind these differences remain to be discovered.
We expected that knocking down the type I IFN signaling pathway in the stroma would result in reduced IFN and ISG transcript levels in the epithelium, but unexpectedly, either no effect or a positive effect was observed. Thus, it appears that basal IFN signaling in the stroma is not required for basal IFN signaling in the epithelium in any obvious way. However, we did find that knockdown of stromal IFN signaling resulted in increased HPV late transcript levels. Although both viral replication and gene expression were suppressed by stromal IFN signaling, there was not a clear correlation between viral DNA copy number and transcript levels, as previously reported (93, 94). Rather, increased viral transcripts correlated with increased expression of the late keratinocyte differentiation marker FLN and LOR. Various lines of evidence indicate that IFN-γ (type II IFN) can promote keratinocyte differentiation (95–100), but an ability of type I IFNs to promote differentiation is much weaker or absent (95, 96, 101). Our data suggest that type I IFN signaling in the stromal compartment may exert a negative effect on keratinocyte differentiation. Given the importance of skin integrity for immune function (102, 103), it is not surprising to find that keratinocyte differentiation may respond to immune signals.
MATERIALS AND METHODS
Cell lines and cultivation.Primary human foreskin fibroblasts (HFF) and human foreskin keratinocytes (HFKs) were isolated from discarded neonatal foreskin samples. The foreskin samples were collected from the University Health hospital in Shreveport, LA, according to an institutional review board (IRB)-approved protocol. HFKs were isolated by enzymatic tissue disruption as described previously (104). Human fibroblasts were isolated from discarded foreskin circumcisions by slightly modifying the protocol available at the Thermo Fisher Scientific website (https://www.thermofisher.com/us/en/home/references/protocols/cell-culture/primary-cell-protocols/fibroblast-protocols/isolation-primary-culture-and-cryopreservation-of-human-neonatal-fibroblasts.html). Briefly, neonatal foreskin tissue was washed with phosphate-buffered saline (PBS) containing 1× antibiotic-antimycotic solution (Thermo Fisher) followed by dispase digestion (2.4 U/ml) overnight at 4°C. On the next day, the epidermis was separated from the dermis, and dermal pieces were finely minced with sterile scissors. Tissue pieces were transferred to a conical tube with 5 ml of 1,500 U/ml collagenase and incubated at 37°C for 1 h with vigorous swirling every 10 min. After incubating, the tissue was almost completely digested. The cell suspension was transferred to a new tube, leaving large tissue pieces behind, and centrifuged at 200 × g for 5 min at room temperature. The supernatant was removed, and the pellet was washed with sterile PBS. Finally, the pellet was resuspended and plated in Dulbecco’s modified Eagle’s medium (DMEM) with 10% bovine growth serum with both antifungal and penicillin-streptomycin (pen-strep) solutions. The medium was changed the next day, and colonies of HFFs were usually observed within 48 h of plating.
Keratinocyte-based cell lines were created by transfection of the HPV16 genome as previously described (66, 105). Creation of the E5 Stop cells was described in reference 34. Episomal maintenance was confirmed using Southern blot analysis (104) or by a newly developed assay utilizing ExoV nuclease to distinguish episomal from integrated DNA (J. E. Myers and R. S. Scott, manuscript in revision). Keratinocyte-derived cell lines were cultivated in E medium using mitomycin C-treated NIH 3T3 J2 fibroblast feeders as described previously (104). The HFFs were maintained by growing in DMEM plus 10% bovine growth serum. HFKs were used between passages 3 to 8. No HFKs were used that showed signs of senescence, as determined by growth pattern in culture. HPV16+ and E5 Stop cells were created from HFKs between passages 3 and 5 and were used at less than passage 10 following selection. HFFs were used at passages less than 15 following isolation.
Knockdown cell strains.For knockdown experiments, short hairpin RNA (shRNA) lentiviruses were produced by transfecting human embryonic kidney (HEK293T) cells with pLK0.1 TRC vectors from Dharmacon (nontarget control TRC [RHS6848] or IRF3 [RHS4533-EG3661], IRF7 [RHS4533-EG3665], IFNAR1 [RHS4533-EG3454], or STAT1 [RHS4533-EG6772]) together with psPAX2 packaging plasmid and pMD2.G envelope plasmid (kind gifts from Jeremy P. Kamil) using polyethylenimine (Polysciences). Cells were maintained in DMEM containing 10% bovine growth serum (BGS) and antibiotics. At 48 h posttransfection, virus-containing supernatant medium was centrifuged at 200 × g and filtered through 0.22-μm filters. Virus stocks were snap frozen and stored at −80°C until use. At least two different donor backgrounds of HFFs were infected with either nontarget control shRNA lentiviral particles or particles targeting the gene of interest along with Polybrene (5 μg/ml). Cells were used for experiments following selection and expansion. Knockdown was confirmed by Western blotting or, in the case of IRF7, reverse transcriptase quantitative PCR (RT-qPCR).
To generate conditioned supernatants, 3 million HFK, HPV16+, or E5 Stop cells were seeded into a 10-cm dish overnight. Following attachment, 5 ml fresh E medium plus 5% FBS was added and incubated for 48 h. Control medium consisted of E medium plus 5% FBS placed in a dish without keratinocytes. Following conditioning, medium was filtered and stored at −80°C until use. One million HFFs per well were plated in 6-well plates. Following attachment, 2 ml conditioned medium was added to each well, cells were incubated for 24 h, and total RNA was harvested using RNA-Stat 60 (TelTest, Inc.) according to the manufacturer’s instructions (74).
Organotypic raft culture.Low-passage-number HFKs alone or keratinocytes harboring either wild-type HPV16 or the E5 Stop mutant HPV16 genomes were used for the generation of organotypic rafts as described previously (64), except that low-passage-number HFFs rather than murine J2 cells were used to form a dermal equivalent. Keratinocytes to be used in these experiments were grown for at least 1 to 2 passages using mitomycin C-treated HFFs as feeders rather than the usual murine J2 cells to ensure that the keratinocytes were adapted to the presence of human fibroblasts. We created rafts using combinations of at least three different donor HFF strains along with three donor backgrounds of HFK alone, HPV16+ keratinocytes, and HPV16 E5 Stop cells. In a given experiment, all fibroblast samples that were to be compared with each other were from the same HFF donor background. To grow organotypic raft cultures with knockdown fibroblasts, HFFs with the gene of interest knocked down were included in the dermal equivalent, but the protocol was otherwise identical. The rafts were harvested after 14 days of growth.
For experiments involving gene expression analyses, epithelial and stromal compartments were manually separated. RNA samples were isolated by the Feist-Weiller Cancer Center Tissue and Serum Repository Genomic Isolation core lab on a Qiagen Qiacube. Briefly, 350 μl lysis buffer RT plus 2-mercaptoethanol (10 μl per ml lysis buffer) (Qiagen) and a 5-mm stainless steel bead (Qiagen) were added to the sample. The sample was homogenized using a Qiagen Tissuelyser II at 30 Hz for 20 s. The sample was briefly centrifuged, and 350 μl of the supernatant was transferred to a clean tube. Samples were then placed in a Qiacube instrument and processed using standard protocols, with the addition of a QiaShredder column to filter out insoluble materials and reduce sample viscosity. DNase treatment was performed on the column as part of the isolation protocol. Samples were eluted in 50 μl of RNase-free water.
Microarray analysis.For microarray analysis, stromal samples from three rafts grown with HFFs alone or stromal samples grown in the presence of HFKs, wild-type HPV16+ keratinocytes, or HPV16 E5 Stop cells were used. Following RNA isolation, microarray analysis was performed in the LSU Health Sciences Center Genomics Core facility using an Affymetrix HTA-2.0 array, according to the manufacturer’s directions. The data generated from the microarray were analyzed using Affymetrix Transcriptome Analysis Console with human genome version of hg19 to determine the genes significantly differentially expressed between different sets of conditions. Significantly differentially expressed genes were considered to have a P value of <0.05 and fold change in expression of ≥1.5.
To identify and visualize the enriched gene ontogeny (GO) terms in ranked list of genes, GOrilla, an online tool, was employed (http://cbl-gorilla.cs.technion.ac.il/). Two unranked lists of genes that included targets (that were affected in microarray) and background (all genes in the microarray) were used to determine significantly enriched GO terms (67). Further, we fed the output of GOrilla enriched GO terms in REViGO, a web server, to remove the redundant GO terms (http://revigo.irb.hr/). For our analysis, we used the similarity coefficient of 0.7 (medium) and P values for different GO categories generated through GOrilla (68). Pathways returned by the software were further curated to eliminate redundant pathways (substantially similar to higher ranked pathways) or overgeneralized pathways that yielded limited insight.
Expression values of genes from the microarray were fed into in an online program, ClustVis (https://biit.cs.ut.ee/clustvis/), to visualize data groupings through heat maps. Original values were ln(x)-transformed; unit variance scaling was applied to rows, and rows were clustered using correlation distance and complete linkage. All the genes present in the related GO terms were used for heat map generation: for ECM based, GO:0030198 was used, while for immune response, GO:0006955 was used.
Reverse transcriptase quantitative PCR and Western blotting.cDNA synthesis was carried out with 1 μg of total RNA using Quanta qScript cDNA synthesis kit (Quantabio). Fifty nanograms total cellular DNA was used for qPCRs of the viral genome. SYBR green PCR master mix (Applied Biosystems) was used for all the qPCRs, and the reactions were performed in a Step One Plus Real-time PCR system (Applied Biosystems). The genes and primer pairs used are described in Table 7. All primers were synthesized by Integrated DNA Technologies (IDT, USA). Fold changes were calculated using the ΔΔCT method, using cyclophilin A as a housekeeping control.
Oligonucleotides used in this study
To confirm knockdown of cellular factors, total protein was isolated using cell lysis buffer (Cell Signaling), and protein concentration was measured using Bradford’s assay (Bio-Rad). One hundred micrograms of total protein was loaded on sodium dodecyl sulfate-polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane. The membrane was incubated overnight at 4°C with antibodies listed in Table 8. The membrane was then washed and incubated with near-infrared (IR) secondary antibodies (LI-COR). The membranes were scanned with an LI-COR Odyssey system. Densitometry analysis was done using Image Studio Light 5.2 after normalizing protein expression levels to the level of GAPDH (glyceraldehyde-3-phosphate dehydrogenase).
Antibodies used in this study
Immunofluorescence staining.Organotypic raft cultures were fixed in 4% paraformaldehyde on ice for ∼1 h and then washed 3× with 1× PBS, with at least 15 to 30 min per wash. Cassettes were transferred to ice-cold 70% ethanol for at least 30 min before storage in fresh ice-cold 70% ethanol. Samples were paraffin embedded and sectioned (4 μm) by the Feist-Weiller Cancer Center Tissue and Serum Repository. The Repository also prepared hematoxylin and eosin slides. To perform IF, sections were deparaffinized with xylene (Fisher Scientific) and then rehydrated with decreasing concentrations of ethanol in water. Antigen retrieval was achieved by heat-induced epitope retrieval either by incubating the slides for 20 min in 95°C citrate buffer (pH 6.0) in a vegetable steamer (Oyster, USA) or for 10 min in a pressure cooker (Instant Pot). The slides were then washed with water and permeabilized for at least 45 min in 0.5% Triton X-100 in 1× PBS (pH 7.0). The permeabilization step was omitted while performing IF staining for CXCL8. Afterwards, the slides were washed with PBS and then blocked with either 5% goat serum (Gibco) or 9% bovine serum albumin (BSA; Fisher Scientific) in PBS for at least 1 h before incubating with primary antibodies. Each sample was incubated overnight with antibodies as listed in Table 8. The next day, slides were washed with PBS followed by an incubation with secondary antibodies (1:500 dilution, Alexa Fluor 488; Life Technologies) for 1 h at 37°C. All antibodies were diluted either in 2.5% goat serum in PBS or 4.5% BSA in PBS. Finally, the slides were dried, and mounting medium with DAPI (4′,6-diamidino-2-phenylindole; Sigma) was added. The experiments were conducted with three different HFF lines at least in duplicates and in some cases triplicates. Images were acquired with a Leica FW4000 microscope with 20× lens objective. Images for presentation in the figures were modified using Adobe Lightroom 7.6.2 to enhance brightness equally for all images. Pictures were taken to avoid parts of the raft that showed processing artifacts (such as detached epithelium, folding, etc.) and included disparate parts of each raft, but otherwise, pictures were random throughout the slide. Boxed areas in the figures were chosen to include several fibroblasts at once to illustrate the differences between samples. Quantification of raw images was performed using ImageJ software (https://imagej.nih.gov/ij/) on all DAPI+ cells in the stromal portion of the section, regardless of their distribution. In ImageJ, the area, integrated density, and mean gray value were selected while analyzing images to generate the corrected total cell fluorescence (CTCF). Only IF signals that overlapped with DAPI were included in quantifications to ensure that real cells and not debris or background staining was counted. At least 100 total cells, including at least three independent experiments, were included in the quantifications.
Statistics.For qPCR/RT-qPCR, Western quantification, and immunofluorescence quantification, statistical significance was calculated using Welch’s unequal variances t test.
ACKNOWLEDGMENTS
We thank members of the Center for Molecular and Tumor Virology and Craig Meyers for helpful discussions. We thank Jeremy Kamil for the gift of lentivirus packaging constructs and Malgorzata Bienkowska-Haba and Katarzyna Zwolinska in the laboratory of Martin Sapp for advice and technical assistance. We also thank the LSU Health Sciences Center—Shreveport Genomics Core Facility for assistance with microarray experiments and the Feist-Weiller Cancer Center Serum and Tissue Repository for help with tissue processing and RNA isolations.
This work was supported by grants from the National Institute of Allergy and Infectious Diseases (R01AI118904 to J.M.B.), the National Institute of Dental and Craniofacial Research (DE025565 to R.S.S.), the National Institute of General Medical Sciences (P30GM110703 to J.M.B. and R.S.S.), and the Feist-Weiller Cancer Center (to B.L.W., M.L.S., and J.M.B.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Feist-Weiller Cancer Center.
FOOTNOTES
- Received 15 March 2019.
- Accepted 29 June 2019.
- Accepted manuscript posted online 10 July 2019.
Supplemental material for this article may be found at https://doi.org/10.1128/JVI.00458-19.
- Copyright © 2019 American Society for Microbiology.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
