Immune Dysregulation and Tumor-Associated Gene Changes in Recurrent Respiratory Papillomatosis: A Paired Microarray Analysis

Abstract

Recurrent respiratory papillomas are benign airway tumors, caused primarily by human papillomaviruses types 6 and 11. The disease is characterized by multiple recurrences after surgical removal, with limited effective therapy. In order to identify novel targets for future therapy, we established transcriptional profiles for actively growing papillomas compared to autologous, clinically normal, laryngeal epithelia (adjacent tissue). Total RNA from 12 papillomas and 12 adjacent tissues were analyzed by microarray, and the matched sets of tissues compared by paired t-test, to identify differentially expressed genes in papilloma tissues while minimizing variations intrinsic to individual patients. Quantitative PCR was used to confirm the relative expression levels for a subset of genes. Within the 109 differentially expressed transcripts were two large groups of genes with related functions whose expression varied by at least 3-fold. The first group consisted of eighteen genes related to host defense, including both innate and adaptive immunity. The second group contained 37 genes that likely contribute to growth of papillomas as benign tumors, since the altered pattern of expression had also previously been reported in many cancers. Our results support our previous studies that document a systemic T_H2-like adaptive immune response in RRP, and suggest that there is a role for altered innate immunity in RRP as well. We propose that HPV 6 and 11 infection establishes a tumorigenic microenvironment characterized by alteration of both innate inflammatory signals and adaptive immune responses that prevent effective T_H1-like responses, in conjunction with altered expression of numerous genes that regulate cellular growth and differentiation.

INTRODUCTION

Recurrent respiratory papillomatosis (RRP)3 is primarily caused by human papillomavirus (HPV) types 6 and 11, with all other HPV types causing less than 2% of disease (1, 2). These viruses induce the growth of benign tumors in the larynx, and less frequently, in the lower respiratory tract. Standard treatment is repeated surgery to remove papillomas that, because of their location in the airway, cause significant morbidity, and on occasion mortality (3, 4). The interval between surgical intervention varies between patients, ranging from 3 weeks to several years (3). Differences in host immune responses to HPV infection may explain this variability.

An effective immune response to viral infection involves activation of both innate and adaptive immunity, with a balance between T_H1-, T_H2-like, and T_H17-derived chemokines and cytokines, and appropriate signaling through receptors they bind (5)We previously reported differences in HPV-specific immune responses by RRP patients and controls that predict disease susceptibility and severity (6–9). Peripheral blood mononuclear cells (PBMC) from these patients respond to HPV 6/11 E6 protein by expressing T_H2-like cytokines and IL-10 (8, 9). Recently we also identified increased levels of the T_H2-like chemokine CCL18 (10) in patient serum. Select class I MHC and class II MHC genes are enriched in RRP (6, 7, 11, 12), associate with disease predisposition, and/or disease severity (7, 12), and correlate with PBMC expression of IFN-γ when challenged with E6 protein (6). Thus, we proposed that the inability of RRP patients to eliminate HPV-infection is likely due to an HPV-specific, T_H2-like/IL-10-biased microenvironment within papillomas that suppress effective T_H1-like responses, and thereby favors recurrent disease.

The immune system also plays a complex role in regulating the growth and metastasis of malignancies (13, 14), however, its role in the development of benign lesions is less well understood. To better understand the expression of specific immune response genes within papilloma tissues, and to identify host genes that are important in the pathophysiology of RRP, we compared the gene expression profiles of paired laryngeal papilloma tissues and autologous adjacent epithelia. We found differences in expression of both innate and adaptive immune response genes, and in many genes associated with a variety of malignancies.

MATERIALS AND METHODS

Patients

Biopsies of papilloma and adjacent epithelia were obtained from patients with RRP undergoing surgery at Long Island Jewish Medical Center following informed consent as approved by the North-Shore Long Island Jewish Health System Institutional Review Board. None of the patients included in our study had high grade dysplasia in their papillomas. Surgical pathologic studies of these papillomas were performed by the Pathology Department at the Long Island Jewish Medical Center and were negative for high grade dysplasia.

RNA isolation and cRNA synthesis

Total RNA was extracted immediately (RNeasy spin columns, Qiagen,Valencia, CA), and stored at −70°C. Matched sets yielding 2 μg total RNA from both tissues (n=12 pairs) were studied. Double stranded cDNA was synthesized from 2 μg of total RNA. (Superscript Double-Stranded cDNA Synthesis Kit, Invitrogen, Carlsbad, CA). Total cDNA was used generate biotinylated cRNA. (BioArray High Yield RNA Transcript Labeling Kit, Enzo Life Sciences Inc., Farmingdale, NY).

Microarray and data analysis

20μg of fragmented cRNA was hybridized to Human U133A (n=2) or U133A2.0 (n=10) microarray chips (Affymetrix, Santa Clara, CA.). Samples were processed on a Gene Chip 450 fluidics station (Affymetrix), scanned (Gene Chip 3000 scanner), and analyzed (Affymetrix MAS 5.0 software) according to the manufacturer. Data mining was performed using Genesifter software (VizX Labs, Seattle, WA). Log transformed data sets were normalized using GC-RMA. Both “pair wise” analysis between autologous specimens, and group analysis (adjacent tissue vs. papilloma) using the false discovery rate algorithm, the Benjamini and Hochberg correction, a fold change of 3, and a p<0.05 were employed. A subsequent filtering step excluded candidate genes that either failed to show a change for half of the matched data sets, or if both normal and papilloma groups were classified as “absent call”. Two subsets of biologically relevant genes were further analyzed namely, immune response genes, and genes associated with malignant transformation. Hierarchical clustering was performed on only the 10 paired data sets obtained using the U133A2.0 arrays on all 22,000 probe sets with GeneSpring GX 7.3 software (Silicon Genetics). Briefly, .cel files were transformed using RMA, normalized by setting values below 0.01 to 0.01, normalized to 50^th percentile per chip and to median by gene. Genes with significant differences (p<0.04) were used to create a condition tree and a relevant gene tree.

Quantitative PCR

To validate microarray results, 16 representative genes from both the immune and tumor-associated groups were examined by quantitative reverse transcriptase PCR (Q-RT-PCR) with gene specific primers (Table III) and a probe from the Universal Probe Library Set (Roche, Mannheim, Germany). IL-1F9 was measured using a Taqman probe. Samples were amplified with an Applied Biosystems 7900 HT thermocycler and results analyzed using the delta-delta C_t method.

Table III.

Confirmation of gene expression by Q-PCR

Gene Descriptions	Gene Name	ARRAY	Q-PCR	Oligonucleotide Primer Sequencesa	Probe Sequence
C-C Ligand 5	CCL5	−3.7	−4.7	TTGTCAAAAGGAAGTCTCTAG GTTC CTTGTCACAGAGCCCTTGC	AGCCAGAG
C-C Ligand 14	CCL14	−6.4	−99.7b	GCTTCCCACAGCATGAAGA CCCTAGGGCGATGGTGAT	CTTCCTCC
C-C Ligand 19	CCL19	−3.8	−28.9	AGTGGCACCAATGATGCTG GTACCCAGGGATGGGTTTCT	CTGCTGCC
C-C Ligand 20	CCL20	5.1	11.1	GTGGCTTTTCTGGAATGGAA CAACCCCAGCAAGGTTCTT	AGCCCAAG
C-C Ligand 21	CCL21	−4.3	−4.4	AGAAAGGAAAGGGCTCCAAA AGGCTTCAAGCGTTGGTG	CCTGGAGC
C-X-C Ligand 1	CXCL1	3.7	4.7	AAGCAAATGGCCAATGAGAT ATCTAAACAGTTACAAAACAG ATGTGC	GAAGGCAG
C-X-C Ligand 6	CXCL6	3.3	10.9	TGACACTTGTGAAAAGGCTTGT A AGCAAAAATAGAAATTCACAA CCA	CTCCTCCC
Interleukin 1 family member 9	IL1F9	6.9	13.1	TTCAGAGCTCATGCGCGTTA GGAATAAAGCAAAACAGAAAC AGAGA	CCACGATGGCA TGACTAGCACA GAGC
Plasminogen Activator, tissue	PLAT	3.6	7.8	TCCTCAAAAGCACCCTTGAC CCTTCTGAGAGCCAGGGAGT	CTCCTTCC
Parathyroid Hormone-like-Hormone	PTHLH	9.3	15.2	TCCAAGGACATATTGCAGGA CAATGTGCAGTTTCATAGAGC AA	GGAGACAG
Inhibitor of DNA binding 1	ID1	5.4	2.7	CCAGAACCGCAAGGTGAG GGTCCCTGATGTAGTCGATGA	AGGTGGAG
Inhibitor of DNA binding 2	ID2	3.1	2.1	AGGTCTTTTCAGAGCGTGGA GCCTTGGCATAGTTTGGAGA	GGAAGGAG
Vascular Endothelial Growth Factor A	VEGFA	4.8	5.3	TTTTGCTAACACTCAGCTCTGC CCCTCTTTCAAAGGAATGTGTG	CTGGCTCC
S100 calcium binding protein A7	S100A7	8.6	11.8	CACCAGACGTGATGACAAGAT T GTTGGGGAAGTTCTCCTTCA	GCCTGCTG
S100 calcium binding protein A12	S100A12	3.3	11.2	TCATATCCCTGGTAGCCATTG ACCTACTCTTTGTGGGTGTGGT	GCTGCCCA
Tenascin XB	TNXB	−5.8	−22.3	GGCAGGTGACTACTCCATCC GTCGTACTGGGCGAACACA	GGGCTGGG

^aTop sequence is forward primer, bottom is reverse. All primers are displayed 5’ → 3’

^bfold change may be greater than calculated (40 cycle cut-off)

Association between gene expression fold change and disease severity

Disease severity criteria (3, 8, 9) were used to classify the RRP patients into 2 separate disease categories, namely either severe, or mild/moderate. These disease severity groups were treated as ordinal variables when calculating disease associations. A quantitative measure of disease severity can be calculated by determining the extent, location, number, and size of the lesions, and dividing that value by the time between surgical interventions measured in days. Individual severity scores range from 0.001 to greater than 0.8. We have empirically established a cut-off at 0.6, with scores exceeding that value being classified as severe, and those below that value as mild/moderate. Our study contained equal numbers of severe and mild/moderate patients. The differences between groups for select genes (upregulated immune and angiogenic genes with fold change ≥3.7) was compared using SAS software V9.1.3 (SAS Institute, Cary, NC) via a two-tailed, unpaired t-test.

Ingenuity Analysis

Data were analyzed by the Ingenuity Pathways Analysis (Ingenuity Systems®, www.ingenuity.com). A dataset containing gene identifiers was mapped to its corresponding gene object in the Ingenuity knowledge base using a fold change of > 3.0. These genes (Appendix Table I) were overlaid onto a global molecular network developed by Ingenuity. Networks were then algorithmically generated based on their connectivity. Pathways were constructed using both direct and indirect relationships. Gene products are represented by shapes, with the biological relationship between two genes represented as a line which is supported by at least one reference from the literature.

RESULTS

Differential Gene Expression Arrays

The use of “paired, autologous tissue sets” helped identify genes that might be obscured in an analysis using non-autologous, control tissue, because of intrinsic patient-to-patient variability. 364 candidate genes were identified in the initial pair wise comparison of papillomas and adjacent tissue. A filtering step eliminated genes with marginal or absent expression, resulting in 134 unique identifiers comprised of 109 individual candidate genes. 73 genes were up-regulated, and 36 others down-regulated in papillomas (Table V).

Hierarchical clustering revealed a clear discrimination between papillomas and adjacent tissues (Figure 1). There was no correlation between the overall transcriptional profile of papillomas from patients with severe disease, in comparison to those from patients with mild/moderate disease, with respect to age, gender, or disease severity. Differentially expressed genes could be divided into 3 groups: 1) immune response genes 2) genes that likely play a role in papilloma formation since they are also associated with malignant tumors and 3) genes whose role in RRP is not yet apparent. There were significant differences between papilloma and adjacent tissue in expression of multiple immune response genes as seen in Figure 2, and the associated Table I. The expression pattern of specific genes suggested a local tissue bias away from a robust T_H1 response. The T_H1-like chemokines CCL19 and CCL21, and the chemo attractants CCL5, CCL14, and CXCL12, which would promote T_H1-like immune infiltrates, were all decreased in papillomas. In contrast, the T_H2-like chemokine CCL20, and the interleukin IL-23, which maintains T_H17 cells, were both increased.

FIGURE 1. Unsupervised hierarchical clustering.

The comparison of overall gene expression patterns of 20 tissue samples, 10 papilloma and 10 adjacent clinically normal tissues from patients with RRP. The dendrogram was obtained by unsupervised hierarchical cluster analysis using Gene Spring software. The analysis included all genes contained on the Human Affymetrix GeneChip U133A 2.0. A primary branching pattern reveals two distinct expression patterns showing segregation of all papilloma samples (right) from adjacent tissue (left). The color codes are shown in the color bar where blue represents transcripts below the median, and red represents transcripts above the median.

FIGURE 2. Genes associated with immune response.

Immune response genes identified in Table 1, show a bias of chemokine and interleukin gene expression in all samples (n=12). The expression of 12 genes (bottom), were increased in papillomas, and 8 genes (top) were decreased in papillomas relative to the corresponding autologous adjacent tissue. Highly expressed genes are shown as red boxes with low expressed genes shown as blue boxes and with intermediately expressed genes shown as yellow boxes.

Table I.

Immune Response Genes

Gene Name	Gene	NCBI Accession	Fold Change
Decreased Expression
Chemokine (C-C motif) ligand 5	CCL5	NM_002985	−3.6
Chemokine (C-C motif) ligand 14	CCL14	NM_004166	−6.4
Chemokine (C-C motif) ligand 19	CCL19	U88321	−3.8
Chemokine (C-C motif) ligand 21	CCL21	NM_002989	−4.3
Chemokine (C-X-C motif) ligand 12	CXCL12	U19495	−6.1
ATP-binding cassette, sub-family A, member 8	ABCA8	NM_007168	−14.1
Duffy blood group, chemokine receptor	DARC	NM_002036	−3.1
Interleukin 1 receptor, type 2	IL1R2	NM_004633	−5.1
Increased Expression
Chemokine (C-C motif) ligand 20	CCL20	NM_004591	5.1
Chemokine (C-X-C motif) ligand 1	CXCL1	NM_001511	3.7
Chemokine (C-X-C motif) ligand 6	CXCL6	NM_002993	3.3
Chemokine (C-X-C motif) ligand 8	CXCL8	NM_000584	4.8
Chemokine (C-X-C motif) receptor 7	CXCR7	AI817041	3.7
Defensin β 4	DEFβ4	NM_004942	16.1
Heat shock protein A8	HSPA8	AB034951	5.6
Interleukin 1 F9	IL1F9	NM_019618	6.9
Interleukin 23 A	IL23A	NM_016584	3.5
S100 calcium binding protein A2	S100A2	NM_005978	3.4
S100 calcium binding protein A7	S100A7	NM_002963	8.6
S100 calcium binding protein A12	S100A12	NM_005621	3.3

Papillomas also showed altered expression of many innate immune genes that could affect the balance between alternative adaptive immune responses. Of these, hBD4, S100A2, S100A7 and S100A12 were all elevated, while ABCA8, required for release of IL-1β, was decreased. Intriguingly, IL-1F9 was markedly elevated in papillomas. This interleukin is an agonist for IL-1Rrp2, a newly described member of the IL-1 receptor family that likely induces an alternative to classical IL-1β signaling. This suggests that the innate immune system is also altered in RRP.

Several other differentially expressed immune response genes appear inconsistent with suppression of a T_H1-like response. While CXCR7, the receptor for CXCL12 was elevated, CXCL12 itself was down-regulated. IL-1R2, a decoy receptor that suppresses IL-1β signaling was decreased, but the simultaneous reduction of ABCA8 would limit IL-1β signaling. The pro-inflammatory CXCL1, CXCL6, and CXCL8 (IL-8) were all elevated in papillomas, however these chemokines also have angiogenic functions, that apparently supersede their immunoregulatory function in the pathogenesis of HPV-induced respiratory papillomas. In addition, a number of the innate immune response genes altered in papillomas are also associated with malignancy. These include HSPA8, DARC, S100A2 and S100A7. In every case, the direction of gene expression in papillomas was the same as that reported in malignant tumors.

A large number of non-immune response genes whose expression was altered in the papillomas have also been associated with malignant tumors. These tumor-associated genes included angiogenesis and growth factors, matrix-associated proteins, cell cycle regulators, and tumor suppressors, all of which affect cell growth, differentiation, or survival. Changes in expression of other genes, e.g. keratins, most likely reflect abnormalities in keratinocyte growth and differentiation, but are not important in pathogenesis. In all but three genes in Table II, the direction of change in the papillomas was the same as reported by others for malignancies. Thus, a subset of these genes, are likely required for growth of both benign and malignant tumors.

Table II.

Non-Immune Genes Associated with Malignancy

Gene Name	Gene	NCBI Accession	Fold Change in RRP	Direction of Change in Malignancya
Increased Expression
Angiogenesis
Fibroblast growth factor bp 1	FGFBP1	NM_005130	5.3	↑
Phosphoglycerate kinase 1	PGK1	S81916	4.5	↑
Placental growth factor	PGF	BC001422	3.1	↑/↓
Vascular endothelial growth factor	VEGFA	AF022375	4.8	↑
Hypoxia-induced
Carbonic anhydrase II	CAII	M36532	8.3	↑
Carbonic anhydrase XII	CAXII	BC000278	4.0	↑
Hypoxia-inducible factor 1α	HIF1A	NM_001530	3.1	↑
Hypoxia-inducible protein 2	HIG2	NM_013332	5.5	↑/↓
Growth, Differentiation and Apoptosis
Cyclin-dependent kinase inhibitor 1A	CDKN1A	NM_000389	3.7	↓
Inhibitor of DNA binding 1	ID1	D13889	5.4	↑
Inhibitor of DNA binding 2	ID2	NM_002166	3.1	↑
Insulin-like growth factor bp 3	IGFBP3	BF340228	3.4	↑/↓
Parathyroid hormone-like hormone	PTHLH	BC005961	9.3	↑
TP53 apoptosis effector	PERP	NM_022121	5.8	↑
Membrane, Adhesion and Extracellular
Matrix-Associated
Calcium chloride channel activated 2	CLCA2	NM_006536	4.4	↑/↓
Fascin 1	FSCN1	NM_003088	4.2	↑
Kallikrein-related peptidase 12	KLK12	NM_019598	4.5	↑
Lectin, galactoside-binding, soluble, 7	LGALS7	NM_002307	7.7	↑
Plakophilin 1	PKP1	AI378979	5.4	↓
Enzymes and Enzyme Inhibitors
Aldo-keto reductase family 1 B10	AKR1B10	NM_020299	3.6	↑
Cathepsin L2	CTSL2	AF070448	3.7	↑
Serine Protease Inhibitor B3	SERPINB3	AB046400	4.8	↑
Serine Protease Inhibitor B4	SERPINB4	U19557	3.3	↑
Serine Protease Inhibitor B13	SERPINB13	AF169949	4.1	↑
Decreased Expression
Extracellular Matrix Associated
Dermatopontin	DPT	AI146848	−6.8	↑
Mucin 5AC	MUC5AC	AW192795	−11.7	↓
Tenascin XB	TNXB	M25813	−5.8	↓
Tumor Suppressors
Apolipoprotein D	APOD	NM_001647	−4.1	↓
Four and a half LIM domains 1	FHL1	NM_001449	−3.7	↓
Insulin-like growth factor bp 5	IGFBP5	L27560	−4.0	↓
SPARC-like 1 (mast9, hevin)	SPARCL1	NM_004684	−3.7	↓
Other
Apolipoprotein J	APOJ	AI982754	−7.0	↓
Glutathione S-transferase A2	GSTA2	NM_000846	−5.6	↓

^athe direction of change in malignancy is based on an extensive literature review of the specific gene alternations in multiple tumors including breast, colon, lung, cervical, and brain tumors.

The direction and magnitude of the gene fold changes identified by microarray were confirmed by Q-PCR for a subgroup of immune response and tumor-associated genes (Table III). The direction of change in papillomas was the same as identified by microarray, however, the magnitude of differential expression was in general, greater.

Association of gene expression and disease severity

Since the expression of many immune response and angiogenesis-related genes were altered in papillomas, we asked whether the expression of these genes varied between patients with severe disease as compared to those with mild/moderate disease. Four genes, IL-1F9 which may determine the type of innate inflammatory response initiated by the host, chemokines CXCL1 and CXCL8, and the growth factor VEGFA all showed more significant elevations in patients with severe disease (Table IV). Taken together, elevation in expression of these particular genes in patients with severe RRP, suggests that angiogenesis and the regulation of innate inflammatory responses are key factors in RRP pathogenesis. In contrast, the only transcripts which were differentially regulated to a greater degree in patients with mild/moderate disease severity were those for hemoglobin alpha, and hemoglobin beta, which were decreased in the papillomas from patients with less severe disease. The disparity in hemoglobin transcripts might reflect larger numbers of erythrocytes resulting from increased vascularity in papillomas from patients with severe disease.

Table IV.

Genes that show significant correlation between differential expression and disease severity

	Fold Change
Gene Name	Mild/Moderate	Severe	p
IL-1F9	4.6 ± 4.3	26.0 ± 2.6	0.04
CXCL1	1.9 ± 3.3	11.3 ± 2.8	0.02
IL-8	1.1 ± 2.6	11.3 ± 1.5	0.0003
VEGFA	2.46 ± 2.0	7.0 ± 1.7	0.02

Mean±SD for mild/moderate (n=6) and severe (n=6) patient groups. Two-tailed T-test.

Pathway analysis

We utilized the Ingenuity Pathways Analysis to establish relationships between the full set of differentially expressed genes (Appendix Table I). The top scoring network (z=51) was cellular movement/immunological disease/cellular growth and proliferation with a p value of 10⁻¹⁵ (Figure 4a). This network included 25 of the genes we identified, with a central role for VEGFA, linked to several growth factors, and NF-κB, linked to many cytokines. Moreover, there were associations between VEGFA and NF-κB. The pathway with the second highest significance (Z-score) was cancer/cellular movement/reproductive system disease (Figure 4b), which revealed prominent involvement of HIF1α, IL-8, and CXCL12, with central importance of p38 MAPK, PI3K, AP1 and Akt, consistent with our previous reports of PI3K and p38 pathway activation in papillomas (15, 16). The identification of these pathways underscores the relationship between gene expression in papillomas induced by HPV’s with low oncogenic potential, and polarization of cellular pathways reminiscent of malignancy.

FIGURE 4. Ingenuity pathway analysis.

Multiple pathway interactions showing both direct (solid line) and indirect (dashed line) associations between multiple dysregulated genes, having a fold change greater than 3.0, including: VEGF, NFKB, PLAT and various chemokines (4a), having a significance score of 51, and MAPK, PI3K, AkT, Ap1, Pkc, and HIF1α and IL8 (4b) having a significance score of 28. This analysis was performed on the 109 genes listed in Appendix Table V. Square: cytokine/growth factor; vertical diamond: enzyme; horizontal diamond: peptidase; circle: other; parallelogram: transporter; circle-in-circle: complex; oval: trans-membrane receptor; shaded circle-in-circle: group.

DISCUSSION

We have found altered expression of many immune response genes in papillomas compared to autologous laryngeal epithelium. These differences could contribute to the persistence of infection and recurrence of disease by biasing the papilloma microenvironment away from effective T_H1-like responses. Previously, we reported that PBMC’s from RRP patients respond to HPV proteins with increased expression of T_H2-like and regulatory cytokines without adequate expression of IFN-γ (8). We now have evidence that there is a T_H2-like chemokine bias in papillomas (increased expression of CCL20), and concomitant down-regulation of T_H1-like chemokines (CCL19 and CCL21). CCL19 and CCL21 are ligands for CCR7, are required for recruitment of naïve CCR7⁺ T-cells that become T_H1-like memory cells, (17) and direct activated CCR7⁺ antigen presenting cells into inflamed tissues (18). Mice with a genomic deletion of both CCL19 and CCL21 have increased numbers of T_H2-inducer-type myeloid dendritic cells (CD8α⁻ CD11b⁺) (19), and defective CD8⁺ T-cell responses to influenza virus (17). Additionally, CCL19 and CCL21 have been used as adjuvants to enhance CTL responses to tumors (20). Consistent with this, papillomas do not contain significant numbers of CTLs (9). In addition, both CCL5 and CCL14 were down-regulated in papillomas. CCL14 is chemo-attractant for both T-cells and monocytes (21), while CCL5 attracts monocytes, memory T-helper cells, and eosinophils (22). Taken together, changes in these chemokines would result in a relative absence of effector T-cells, especially CTL and T_H1-like T-cells, in papillomas.

T_H17-like T-cells are a new addition to the classical T_H1/T_H2 paradigm (23). T_H17-like T-cells selectively inhibit T_H1-like cells by their expression of IL-17 and IL-23 (24). Conversely, T_H1-like T-cells inhibit T_H17-like T-cells cells by their expression of IFN-γ and IL-12, neither of which were expressed in papillomas. CXCL1, CXCL6, hBD4, and CCL20 were all upregulated in papillomas and interestingly are all expressed by human bronchial epithelial cells when treated with IL-17A (25). IL-23 was elevated in papillomas and is required for maintenance of T_H17-like T-cells in vivo (26). Thus, polarization of T_H0-like T-cells towards the T_H17-like lineage may occur in RRP. However, expression of IL-6 and TGFβ₁, in the absence of IL-21 and IL-22, by most papillomas, suggests that T-regulatory cells (T-regs), but not TH17-cells, are preferentially induced in papillomas (27). We have detected increased numbers of CD4⁺ CD25⁺ FoxP3⁺ CD127low T-regs in papilloma tissue, compared to autologous blood, (28) supporting the possibility that functional T-regs in RRP may be responsible for the absence of inflammation caused by T_H1-, and TH17-like cells in this RRP.

Expression of multiple innate immune response genes were also altered in papillomas. Of particular interest was the elevated expression of IL-1F9 mRNA levels that were significantly higher in patients with severe disease (p=0.03), suggesting a central role for IL-1F9 in predisposing to severe disease. IL-1F9 binds to IL-1Rrp2 (29), and is thought to alter innate immune response signaling. Subsequent polarization of the adaptive immune response remains unknown (30), however, several lines of evidence (31, 32) suggest that IL-1F9 likely induces an alternative to IL-1β signaling. IL-1F9 has recently been implicated in allergen-induced T_H2-like bronchial hyper-responsiveness (Abhr1) in mice (33). Furthermore, stimulation of human bronchial epithelial cells with Pseudomonas aeuruginosa induced the expression of IL-1F9, suggesting that IL-1F9 regulates T_H2-like innate responses that normally occur following bacterial exposure. This suggests that there may be a distinct, non IL-1β-inducible innate pathway stimulated by this interleukin in humans. We speculate that IL-1F9 expression induces a yet to be characterized innate response in papillomas that polarizes adaptive T-cells away from a T_H1-like response.

Altered expression of a small number of genes in papilloma tissues have been reported using qRT-PCR, in situ hybridization, or RNase protection, and many of those changes including hBD4, CXCL8 (34) and VEGFA (35) were also identified in our analysis, further validating our findings. However, we did not detect elevated levels of survivin mRNA (36), or transcripts for p16INK4A and p53 (37). These discrepancies may reflect, in part, our use of matched papilloma and autologous, epithelia pairs, rather than other control tissues.

A number of genes differentially expressed in papillomas have been associated with malignancies, affecting both tumor growth and immune responses. These include cytokines CXCL1, CXCL6, and CXCL8, and VEGFA that can all function as growth and angiogenic factors (38, 39). Furthermore, increased expression of CXCL1, CXCL8 and VEGFA significantly correlated with severe disease (Table IV), suggesting that angiogenesis, a histological hallmark of RRP, is central to the pathology of this disease. Also evident were reductions in expression of three tumor suppressors (IGFbp5, FHL1 and SPARCL1) and elevated expression of several growth factors (PGF, IGFbp3, and PTHLH). Three members of the S100 family of proteins were also altered in papillomas. These proteins are involved in regulation of numerous cellular processes, including cell growth, differentiation, and progression towards cancer (40). They all likely play a role in regulating the innate immune response to pathogens. S100 proteins are damage-associated molecular pattern molecules, which can function as pro-inflammatory factors of innate immunity (41). S100A2 has been reported to both promote tumor growth (42) and function as a tumor suppressor gene (40, 43). S100A7 is over-expressed in breast cancer (44), epithelial skin tumors (45) bladder cancer (46) and is markedly elevated in lesions from psoriasis patients (47), suggesting a role in keratinocyte differentiation, and in regulating the innate immune response associated with epithelial inflammation (46). S100A12, a potent monocyte chemoattractant (48), that mediates allergic inflammation by activating mast cells (49)was also increased in expression in papillomas. These observations suggest that the less oncogenic HPVs can reprogram cellular pathways similar to that described in some malignancies. Studies are underway to compare HPV 6/11 induced changes with those induced by the oncogenic HPV 16, to identify key cellular processes that distinguish their differential expression in benign versus malignant tumorogenesis.

In summary, we have used microarray analysis to identify changes in the transcriptional profiles of papilloma tissue from patients with RRP, as compared to autologous, laryngeal epithelium. We identified several groups of genes that may contribute to the disease process and disease severity. However, genes that are comparably expressed in both tissues would not be detected, even though some may also be important to disease susceptibility and/or severity. Our results support our previous contention that RRP is a disease characterized by a defective T_H1-like response in adaptive immunity. In this communication we now suggest that altered innate responses to HPV are also present. Our findings may be relevant to other HPV-induced diseases, such as cervical cancer, where oncogenesis complicates and overshadows the inherent immunologic responses made to the more oncogenic HPVs. Our results provide new insight into the disease process associated with RRP, identify for the first time that papillomaviruses with low oncogenic potential can induce gene expression changes characteristically found in malignancies, and identify novel targets for future therapeutic interventions in RRP.

FIGURE 3. Genes associated with malignancy.

Genes that are deregulated in various malignancies as identified in Table 2. 22 of 24 genes that are increased in malignancy (tumor promoters/growth factors) are also increased in papillomas (top). 8 of 9 genes which are decreased in malignancy (tumor suppressors) were also decreased in papillomas (bottom). Highly expressed genes are shown as red boxes with low expressed genes shown as blue boxes and with intermediately expressed genes shown as yellow boxes.

Appendix Table V.

Candidate genes that are differentially expressed in papillomas.

Gene name	Mean Fold Change	NLM Accession	Affymetrix Unique Identifier	Gene Description
A. Immunologically Relevant Genes
A1.Chemokines and chemokine receptors
CCL5	−3.7	M21121	1405_i_at	T cell-specific protein precursor; (RANTES)
CCL5	−3.6	NM_002985	204655_at	Chemokine Ligand 5
CCL14	−6.4	NM_004166	205392_s_at	CCL 14
CCL19	−3.8	U88321	210072_at	Small inducible cytokine subfamily A (Cys-Cys), member 19
CCL20	5.1	NM_004591	205476_at	Small inducible cytokine subfamily A (Cys-Cys), member 20
CCL21	−4.3	NM_002989	204606_at	Small inducible cytokine subfamily A (Cys-Cys), member 21
CXCL1	3.7	NM_001511	204470_at	GRO1 oncogene (melanoma growth stimulating activity, alpha)
CXCL6	3.3	NM_002993	206336_at	Small inducible cytokine subfamily B (Cys-X-Cys), member 6
CXCL8	4.8	NM_000584	202859_x_at	Interleukin 8 (also in cancer)
CXCL12	−6.1	U19495	209687_at	Stromal cell-derived factor 1(also in cancer)
CXCR7	3.7	AI817041	212977_at	Chemokine (C-X-C motif) receptor 7(also in cancer)
DARC	−3.1	NM_002036	208355_s_at	Duffy blood group, chemokine receptor (also in cancer)
A2. Interleukins and interleukin receptors
IL1R2	−5.1	NM_004633	205403_at	Interleukin 1 receptor, type II
IL1R2	−3.1	U64094	211372_s_at	Interleukin 1 receptor, type II
IL1F9	6.9	NM_019618	220322_at	Interleukin-1 homolog 1
IL23A	3.5	NM_016584	220054_at	Interleukin 23, alpha subunit p19 (also in cancer)
A3. Innate immunity
ABCA8	−14.1	NM_007168	204719_at	ATP-binding cassette (ABC1)
DEFβ4	16.1	NM_004942	207356_at	Defensin, beta 4
S100A12	3.3	NM_005621	205863_at	S100 calcium-binding protein A12 (calgranulin C)
A4. Immunoglobulin
IGL	−3.5	AJ408433	216401_x_at	Immunoglobulin kappa chain variable region, clone 38
IGHG1	−3.2	AI858004	213674_x_at	Immunoglobin heavy constant gamma 1
IGHG1	−4.8	BC001872	209374_s_at	Immunoglobulin heavy constant gamma 1
IGL	−3.1	M85256	211645_x_at	Immunoglobulin kappa light chain
B. Genes previously associated with malignancy
AKR1B10	3.6	NM_020299	206561_s_at	Aldo-keto reductase family 1, member B10 (aldose reductase)
APOD	−4.1	NM_001647	201525_at	Apolipoprotein D
CA2	8.3	M36532	209301_at	Carbonic anhydrase II
CA12	4.0	BC000278	210735_s_at	Carbonic anhydrase XII
CLU	−7.0	AI982754	222043_at	Clusterin (complement lysis inhibitor, SP-40, apolipoprotein J)
CLU	−6.9	M25915	208791_at	Clusterin (complement lysis inhibitor, SP-40, apolipoprotein J)
CLU	−5.9	M25915	208792_s_at	Clusterin (complement lysis inhibitor, SP-40, apolipoprotein J)
CTSL2	3.7	AF070448	210074_at	Cathepsin L2
CDKN1A	3.7	NM_000389	202284_s_at	Cyclin-dependent kinase inhibitor 1A (p21,Cip1)
DPT	−6.8	AI146848	213068_at	Dermatopontin
DPT	−4.5	AI146848	213071_at	Dermatopontin
DPT	−3.0	NM_001937	207977_s_at	Dermatopontin
FGFBP1	5.3	NM_005130	205014_at	Fibroblast growth factor binding protein 1
FHL1	−3.7	NM_001449	201540_at	Four and a half LIM domains 1
FSCN1	4.2	NM_003088	201564_s_at	Singed (Drosophila)-like (sea urchin fascin homolog like)
GSTA2	−5.7	NM_000846	203924_at	Glutathione S-transferase A2
HIF1A	3.1	NM_001530	200989_at	Hypoxia-inducible factor 1, alpha subunit
HIG2	5.5	NM_013332	218507_at	Hypoxia-inducible protein 2
HSPA8	5.6	AB034951	210338_s_at	Heat shock 70kDa protein 8 (also in cancer)
ID1	5.4	D13889	208937_s_at	Inhibitor of DNA binding 1
ID2	3.0	NM_002166	201565_s_at	Inhibitor of DNA binding 2
ID2	3.1	NM_002166	201566_x_at	Inhibitor of DNA binding 2
IFI27	3.7	NM_005532	202411_at	Interferon alpha-inducible protein 27 (also in cancer)
IGFBP3	3.4	BF340228	212143_s_at	Insulin-like growth factor binding protein 3
IGFBP5	−4.0	L27560	211959_at	Insulin-like growth factor binding protein 5
KLK12	4.5	NM_019598	220782_x_at	Kallikrein-related peptidase 12
KRT6A	6.1	AL569511	214580_x_at	Keratin 6A
KRT6A	5.6	J00269	209125_at	Keratin 6A
KRT6B	12.1	L42612	209126_x_at	Keratin 6A
KRT6B	9.3	AI831452	213680_at	Keratin 6A
KRT14	8.7	BC002690	209351_at	Keratin 14
KRT16	15.7	AF061812	209800_at	Keratin 16
KRT17	5.9	NM_000422	205157_s_at	Keratin 17
KRT17	4.4	Z19574	212236_x_at	Keratin 17.
LGALS7	7.7	NM_002307	206400_at	Lectin, galactoside-binding, soluble, 7 (galectin 7)
MUC5AC	−11.7	AW192795	214303_x_at	Mucin 5AC
MUC5AC	−10.7	AI521646	214385_s_at	Mucin 5AC
PERP	5.8	NM_022121	217744_s_at	TP53 apoptosis effector
PGF	3.1	BC001422	209652_s_at	Placental growth factor, VEGF-related protein
PGK1	4.5	S81916	217356_s_at	Phosphoglycerate kinase 1
PKP1	5.4	AI378979	221854_at	Plakophilin 1
PTHLH	9.3	BC005961	211756_at	Parathyroid hormone-like hormone
PTHLH	5.5	J03580	210355_at	Parathyroid hormone-like hormone
PTHLH	3.4	NM_002820	206300_s_at	Parathyroid hormone-like hormone
SERPINB3	4.3	BC005224	209719_x_at	Serpin Peptidase Inhibitor, clade B (ovalbumin), member 3
SERPINB4	3.3	U19557	210413_x_at	Serpin Peptidase Inhibitor, clade B (ovalbumin), member 4
SERPINB3	4.8	AB046400	211906_s_at	Serine proteinase inhibitor, clade B (ovalbumin), member 3
SERPINB13	4.1	AF169949	211362_s_at	Serine proteinase inhibitor, clade B (ovalbumin), member 13
SERPINB13	3.9	AJ001698	217272_s_at	Serine proteinase inhibitor, clade B (ovalbumin), member 13
SERPINB 13	3.1	AJ001696	211361_s_at	Serine proteinase inhibitor, clade B (ovalbumin), member 13
SPARCL1	−3.7	NM_004684	200795_at	SPARC-like 1 (mast9, hevin)
S100A2	3.4	NM_005978	204268_at	S100 calcium-binding protein A2
S100A7	8.6	NM_002963	205916_at	S100 calcium-binding protein A7 (psoriasin 1)
TNXB	−3.1	NM_007116	206093_x_at	Tenascin XB
TNXB	−5.8	M25813	216333_x_at	Tenascin XB
VEGFA	4.8	AF022375	210512_s_at	Vascular endothelial growth factor
VEGFA	3.0	H95344	212171_x_at	Vascular endothelial growth factor
C. Other
TMPRSS11D	3.3	NM_004262	207602_at	Airway trypsin-like protease
APOL1	5.9	AF323540	209546_s_at	Apolipoprotein L
DST	3.5	NM_001723	204455_at	Bullous pemphigoid antigen 1 (Dystonin)
CLCA2	4.4	NM_006536	206165_s_at	Chloride channel, calcium activated, family member 2
CLCA2	3.7	AF043977	206166_s_at	Chloride channel, calcium activated, family member 2
CLCA2	3.1	NM_006536	206164_at	Chloride channel, calcium activated, family member 2
COL4A1	6.1	NM_001845	211981_at	Collagen, type IV, alpha 1
COL4A1	4.9	NM_001845	211980_at	Collagen, type IV, alpha 1
DSC2	3.7	NM_004949	204751_x_at	Desmocollin 2
DSC3	3.9	NM_001941	206033_s_at	Desmocollin 3
DSG3	5.4	NM_001944	205595_at	Desmoglein 3
DDIT4	3.5	NM_019058	202887_s_at	DNA-damage inducible transcript 4
ENOL1	7.6	U88968	217294_s_at	Enolase 1, (alpha)
FABP5	3.1	NM_001444	202345_s_at	Fatty acid binding protein 5 (psoriasis-associated)
FER1L3	3.7	AF207990	211864_s_at	Fer-1-like 3, myoferlin
FLG	6.5	AL356504	215704_at	filaggrin
GJA1	3.4	NM_000165	201667_at	Gap junction protein, alpha 1, 43kD (connexin 43)
IGF2BP3	3.2	NM_006547	203820_s_at	IGF-II mRNA-binding protein 3
IVL	3.2	NM_005547	214599_at	Involucrin
KLK10	4.5	BC002710	209972_s_at	Kallikrein-related peptidase 10
LY6D	5.6	NM_003695	206276_at	Lymphocyte antigen 6 complex, locus D
NDUFA4L2	7.1	NM_020142	218484_at	NADH:ubiquinone oxidoreductase MLRQ subunit homolog
PLAT	3.6	NM_000930	201860_s_at	Plasminogen activator, tissue
PPP1R3C	4.3	N26005	204284_at	Protein Phosphatase 1, regulatory (inhibitor) subunit 3C
RHCG	7.0	NM_016321	219554_at	Rh type C glycoprotein
SCEL	3.3	NM_003843	206884_s_at	Sciellin
SPRR1B	5.2	NM_003125	205064_at	Small proline rich protein 1B
SPRR1A	5.3	NM_005987	214549_x_at	Small proline-rich protein 1A
SPRR1A	4.9	AI923984	213796_at	Small proline-rich protein 1A
SPRR2D	6.5	NM_006945	208539_x_at	Small proline-rich protein 2B
SPRR2C	4.5	NM_006518	220664_at	Small proline-rich protein 2C
SQLE	3.5	AF098865	209218_at	Squalene Epoxidase
SCD	4.1	AB032261	200832_s_at	Stearoyl-CoA desaturase (delta-9-desaturase)
TGM1	5.6	NM_000359	206008_at	Transglutaminase 1
TGM3	6.4	NM_003245	206004_at	Transglutaminase 3
TMPRSS 11D	4.6	NM_004262	207602_at	Transmembrane Protease Serine 11D
TMPRSS11E	5.0	NM_014058	220431_at	Transmembrane Protease Serine 11E
ADH1C	−7.8	NM_000669	206262_at	Alcohol dehydrogenase 1C (class I), gamma polypeptide
ALDH3A1	−7.6	NM_000691	205623_at	Aldehyde dehydrogenase 3 family, member A1
CES1	−11.8	S73751	209616_s_at	Carboxylesterase 1
C1S	−3.5	M18767	208747_s_at	Complement component 1, s subcomponent
CKAP2	−5.8	D84143	216984_x_at	Cytoskeleton associated protein 2
CFD	−3.5	NM_001928	205382_s_at	D component of complement (adipsin)
DCN	−3.3	AI281593	209335_at	Decorin
GNG11	−4.4	NM_004126	204115_at	Guanine nucleotide binding protein 11
HSD17B11	−3.1	NM_016245	217989_at	Hydroxysteroid (17 beta) dehydrogenase 11
LEPR	−6.1	U50748	209894_at	Leptin receptor
PTGDS	−5.8	NM_000954	212187_x_at	Prostaglandin D2 synthase (21kD, brain)
PTGDS	−4.6	BC005939	211748_x_at	Prostaglandin D2 synthase (21kD, brain)
SRPX	−3.4	NM_006307	204955_at	Sushi-repeat-containing protein, X chromosome
CLEC3B	−11.9	NM_003278	205200_at	Tetranectin
TGFBR3	−3.9	NM_003243	204731_at	Transforming growth factor, beta receptor III
FAM107A	−9.4	AL050264	209074_s_at	TU3A protein
Unknown	−6.7	AV711904	213975_s_at	Unknown