Microarray analysis of differentially expressed genes in vaginal tissues from women with stress urinary incontinence compared with asymptomatic women

Abstract

BACKGROUND: The pathophysiology of pelvic floor dysfunction resulting in stress urinary incontinence (SUI) in women is complex. Evidence suggests that there is also a genetic predisposition towards SUI. We sought to identify differentially expressed genes involved in extracellular matrix (ECM) metabolism in vaginal tissues from women with SUI in the secretory phase of menses compared with asymptomatic women. METHODS: Tissue samples were taken from the periurethral vaginal wall of five pairs of premenopausal, age-matched SUI and continent women and subjected to microarray analysis using the GeneChip Human Genome U133 oligonucleotide chip set. RESULTS: Extensive statistical analyses generated a list of 79 differentially expressed genes. Elafin, keratin 16, collagen type XVII and plakophilin 1 were consistently identified as up-regulated ECM genes. Elafin, a serine protease inhibitor involved in the elastin degradation pathway and wound healing, was expressed in pelvic fibroblasts and confirmed by Western blot, quantitative competitive PCR and immunofluorescence cell staining. CONCLUSIONS: Genes involved in elastin metabolism were differentially expressed in vaginal tissue from women with SUI, suggesting that elastin remodelling may be important in the molecular aetiology of SUI.

Introduction

Pelvic floor dysfunction resulting in stress urinary incontinence (SUI) is a major health and quality-of-life issue for women in their reproductive and post-menopausal years. Epidemiological studies report that up to 50% of women over the age of 60 have symptoms of urinary incontinence (Diokno et al., 1986; Thom, 1998; Brown et al., 1999). In premenopausal women, there is a disproportionately higher incidence of urinary dysfunction compared with men, with a ratio of 4:1 (Feneley et al., 1979; Thomas et al., 1980). While there are treatments for affected women, there is currently no method to identify women at risk of developing SUI or to prevent this condition. However, the percentage of elective Caesarean sections among women with no indicated risk factors has more than tripled since 1996, despite insufficient data supporting this as a preventive treatment modality (Declercq et al., 2005). Few studies, however, have focused on the possible genetic predisposition to SUI, and there has been no way of examining genes that might predispose women to SUI until now.

The pathophysiology of SUI is complex and not well understood. Damage to connective tissues, muscles and nerves is thought to result in pelvic floor dysfunction. Yet it is puzzling why some women develop SUI in the absence of pregnancy or childbirth, while others with multiple vaginal deliveries suffer no symptoms. Epidemiological data suggest that there may be genetic factors that increase a woman’s likelihood of developing SUI or pelvic floor dysfunction.

In the mechanically active environment of the pelvic floor, cells respond to mechanical stimuli by regulation of the extracellular matrix (ECM) structure (Al-Jamal and Ludwig, 2001; Lee et al., 2001). Several investigators have documented differences in the connective tissues of women with pelvic floor dysfunction compared with continent women. In tissue from affected women, both collagen content in pelvic ligamentous tissues and its degradative enzymes are altered (Falconer et al., 1996; Keane et al., 1997; Rechberger et al., 1998; Chen et al., 2002).

These data give us some insight into the molecular pathophysiology, but progress in this area has been slow due to the complex ECM interactions that occur simultaneously. Additionally, ECM metabolism is modulated by reproductive hormones (Jackson et al., 2002), thus making the phase of the menstrual cycle (menses) an important factor in study design. We hypothesized that ECM metabolism was altered in women with SUI when compared with continent women, and used microarray-based methodology to screen for genetic differences in anterior vaginal wall tissues from continent and incontinent women in the estrogen-only (proliferative) phase of the menstrual cycle (Chen et al., 2004). The chip-based technology uses microarrays of oligonucleotides as gene-specific probes to measure tens of thousands of human genes simultaneously (Lockhart et al., 1996). We identified 62 up-regulated and 28 down-regulated genes in SUI women compared with controls in the proliferative phase.

To further understand the effects of in vivo gonadal steroids on gene expression in SUI, we examined the differential expression of ECM genes in SUI patients compared with continent women in the secretory phase of menses. Screening all genes simultaneously in the two phases of the menstrual cycle is a necessary first step to identify which genes are implicated in SUI. We believe this study, in conjunction with our previous study (Chen et al., 2003), will help researchers better understand the molecular pathophysiology of SUI and develop an effective genetic screening tool for women at risk.

Materials and methods

Patient selection

The Institutional Review Board of Stanford University Medical School approved this study. Premenopausal women undergoing benign gynaecological surgery were screened. Women with a history of endometriosis, gynaecological malignancies, pelvic inflammatory conditions, connective tissue disorders, emphysema and prior pelvic surgery were excluded. Women undergoing surgery for urinary incontinence with pelvic organ prolapse no greater than stage I by POP-Q (Bump et al., 1996) were identified as cases, while continent women no greater than stage I were controls. We chose participants with minimal pelvic organ prolapse to investigate the molecular changes in SUI only, without confounding the results by possible changes in advanced prolapse. We selected only tissue samples from participants in the secretory phase of menses for this study. Menstrual phase was confirmed by endometrial histology. All participants with urinary incontinence symptoms underwent urodynamic studies before surgery to confirm their diagnosis.

Tissue collection and RNA isolation

After informed consent had been obtained, approximately 1 cm² of full-thickness, periurethral vaginal wall was excised 1 cm lateral to the urethrovesical junction from the incontinent women undergoing surgery for urinary incontinence. Smaller, 0.5 cm² biopsies of vaginal wall from a similar area were excised from the control group undergoing benign gynaecological surgery. Pelvic ligamentous tissues are often inaccessible or difficult to obtain in control subjects (asymptomatic women). Because previous work from Keane and colleagues (Keane et al., 1997) showed that collagen from the periurethral vaginal wall closely resembles that from endopelvic fascia, we used tissue from this source for our study. The urethrovesical junction was identified by insertion of a Foley balloon in all participants. The area lateral to this junction was infiltrated with saline, and a vaginal wall incision was made with a scalpel. A full-thickness vaginal wall biopsy was taken after sharp dissection down to the avascular space of loose areolar tissue. Larger specimens were available in participants undergoing anterior colporrhaphy or pubovaginal slings. Immediately after excision, the tissues were frozen in liquid nitrogen and stored at –80°C.

Total RNA was extracted with the Trizol reagent according to the manufacturer’s suggested protocol (Gibco BRL Life Technologies, P/N 15596-018, Grand Island, NY, USA). At least 30 µg total RNA was extracted from the tissue. A portion of this RNA was subjected to gel analysis to verify the integrity of the RNA.

Extraction, amplification and labelling of mRNA

Extraction of total RNA and amplification and labelling of mRNA were carried out as previously described (Mahadevappa and Warrington, 1999) and as published in the GeneChip® Expression Technical Manual (www.affymetrix.com).

Fragmentation, array hybridization and scanning

The labelled target was fragmented by incubation at 94°C for 35 min in the presence of 40 mM Tris-acetate, pH 8.1, 100 mM potassium acetate and 30 mM magnesium acetate (Warrington et al., 2000). The hybridization solution consisted of 20 µg fragmented cRNA and 0.1 mg/ml sonicated herring sperm DNA in 1× MES buffer (containing 100 mM 2-[N-Morpholine]ethanesultonic acid (MES), 1 M Na⁺, 20 mM EDTA and 0.01% Tween 20). Hybridization, subsequent washing, and staining of the arrays were carried out as outlined in the GeneChip^® Expression Technical Manual (Affymetrix, Santa Clara, CA) on HuGeneFl arrays (Affymetrix). The arrays were synthesized using light-directed combinatorial chemistry as described previously (Fodor et al., 1993). We used the GeneChip© Human Genome U133A oligonucleotide chip set from Affymetrix. In each chip set there are two microarrays containing over 1 000 000 unique features and covering more than 39 000 transcript variants. This results in a representation of more than 33 000 well-substantiated human genes (Affymetrix). Following washing and staining, probe arrays were scanned twice (multiple image scan) at 3 µm resolution using the GeneChip^® scanner 3000.

Primers for RT and PCR

To confirm our microarray data, three genes—elafin, IL-1 receptor antagonist (IL-1RA), and receptor (calcitonin) activity-modifying protein 1 (RAMP-1)—were chosen for quantitative competitive RT-PCR and Western blot confirmation. These genes were selected from the raw image data because they were differentially expressed (greater than 2-fold change) in at least four out of five pairs. Specific sequences of oligonucleotide primers for elafin, IL-1RA and RAMP-1 were obtained from the GenBank Database of the National Center for Biotechnology Information (NCBI) of the National Institutes of Health (NIH, Internet address: http://www.ncbi.nih.gov/Genbank). The corresponding set of primers for elafin, IL-1RA and RAMP-1 was found with the help of the program OLIGO 5.0 Primer Analysis Software (National Bioscience, Plymouth, MN, USA) and synthesized by the Protein, Amino Acid and Nucleic Acid (PAN) facility at the Beckman Center of Stanford University. The human β-actin primers that were used to amplify an external standard were obtained from Clontech Laboratories (Palo Alto, CA, USA). β‐actin mRNA expression, employed as an external positive control, was detected in all the samples studied, thus confirming the integrity of the RNA and the RT‐PCR process.

RT

For RT-PCR, the Gen Amp RNA PCR kit was used (Perkin-Elmer, Foster City, CA, USA). Nineteen millilitres of RT-Mastermix for each sample were prepared containing 5 mmol/l MgCl₂; 1× PCR buffer II; 1 mmol/l of each deoxy-NTP; 2.5 µl/l oligo-deoxythimidine; 20 IU ribonuclease inhibitor (all from Perkin-Elmer); 100 IU Moloney murine leukaemia virus reverse transcriptase (Gibco BRL); and 1 µg total RNA diluted in 1 µl DEPC-treated H₂O. They were then filled into 0.5 ml thin-wall PCR tubes (Applied Scientific, South San Francisco, CA, USA). The RT-Mastermix in PCR tubes was covered with 50 µl of light white mineral oil (Sigma, St Louis, MO, USA) and kept on ice until the RT. RT was carried out in the DNA Thermal Cycler 480 (Perkin-Elmer) using a programme with the following parameters: 42°C, 15 min; 99°C, 5 min; then quenching at 4°C. After the reaction was completed, samples were stored at –20°C until the PCR. As the negative control, a sample of 1 µl DEPC-treated H₂O without RNA was subjected to the same RT reaction.

Construction of the competitive and target cDNA fragment

A 455 base pair (bp) fragment of elafin, a 422 bp fragment of IL-1RA cDNA, and a 446 bp fragment of RAMP-1 (the targets) were obtained by PCR amplification of reverse-transcribed total RNA from vaginal mucosa with the regular 3′ and 5′ primers. The PCR product was visualized by agarose gel electrophoresis and stained with ethidium bromide. The cDNA was extracted from the gel, purified with an agarose gel extraction kit (Amersham Pharmacia Biotech, Cambridge, UK), and quantified by spectrophotometry (Pharmacia Biotech, Cambridge, UK). To construct a competitive cDNA fragment, a floating primer with a sequence complementary to the cDNA between the 3′ and 5′ primer binding sites was designed by attaching the complementary sequence of the binding site of the original 3′-end elafin, IL-1RA or RAMP-1. After PCR with the regular 5′-end primer and the 3′-end floating primer, the PCR product was visualized by agarose gel electrophoresis stained with ethidium bromide. Then cDNA extraction and purification and determination of the concentration were performed as described above. This step resulted in cDNA fragments of elafin (278 bp), IL-1RA (155 bp) and RAMP-1 (240 bp).

Quantitative competitive PCR

The standard curve for elafin, IL-1RA and RAMP-1 was constructed by co-amplification of a constant amount of competitive cDNA (1 attomol each) with declining amounts of target cDNA (0.3–0.625 fmol) obtained by serial dilution. Sixty microlitres of the cDNA mix was added to a 40-µl PCR-Mastermix containing 1.9 mmol/l MgCl₂ solution; 10× PCR buffer II; 0.2 mmol/l of each deoxy-NTP; 2.5 U Taq polymerase (Perkin-Elmer); corresponding paired primers in a concentration of 0.2 µmol/l of each primer to a total volume of 100 µl; and 14.5 µl DEPC-treated H₂O. The reaction was covered with 50 µl light white mineral oil and put in the Perkin-Elmer DNA Thermal Cycler 480. PCR cycles were started at 95°C for 5 min to denature all proteins: 30 cycles for 45 s at 94°C; 45 s at 55°C; and 60 s at 72°C. The reaction was terminated at 72°C for 5 min and was quenched at 4°C.

Two per cent agarose gel (Gibco BRL, Gaithersburg, MD, USA) electrophoresis was carried out in an H5 electrophoresis chamber. Gels were stained with ethidium bromide (Sigma). Aliquots (25 µl) of each PCR product and dye buffer were analysed using a 100-bp DNA ladder as a standard (Gibco BRL, Gaithersburg, MD, USA). After completion of electrophoresis, the gel blot was analysed and photocopies of it were printed by UV densitometry (Gel-Doc 1000 system; Bio-Rad Laboratories, Hercules, CA, USA). The logarithmically transformed ratios of target cDNA to competitive cDNA were plotted against the log amount of initially added target cDNA in each PCR to obtain the standard curve. This standard curve was highly reproducible and linear. The values obtained from the regression line of the standard curve (y = b + mx) allowed us to calculate the amount of cDNA transcripts in an unknown sample. One attomol of elafin, IL-1RA or RAMP-1 competitive cDNA was added to each unknown sample before PCR. The ratios of the densities of the sample target elafin, IL-1RA and RAMP-1 cDNA band (455, 422 and 446 bp) to competitive cDNA (278, 155 and 240 bp) were logarithmically transformed and compared with the values on the standard curve.

Western blot analysis

Western blot analysis was used to confirm protein expression of two up-regulated (elafin, IL-1RA) and one down-regulated gene (RAMP-1). Twenty micrograms of total protein from each patient were separated by 10% SDS–PAGE under reducing conditions, and were then blotted onto nitrocellulose membranes (Pierce, Rockford, IL, USA) in an electrophoretic transfer cell (Bio-Rad). Blots were blocked with 5% non-fat milk at 4°C overnight. After blocking, the membrane was washed three times in PBST (phosphate-buffered saline, pH 7.4 and 0.1% Triton). The membrane was incubated for 1 h at room temperature in a 1:100 dilution of polyclonal antibody to human elafin (SKALP) (Cell Science, Norwood, MA, USA), 1 µg/ml monoclonal anti-human IL-1RA antibody (R&D Systems, Minneapolis, MN, USA), or a 1:200 dilution of rabbit polyclonal antibody against RAMP 1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), followed by three washes in PBST. The membrane was then incubated in a 1:5000 dilution of sheep anti-mouse IgG or a 1:5000 dilution of donkey anti-rabbit IgG conjugated horseradish peroxidase (Amersham Pharmacia Biotech) for 1 h at room temperature, followed by three washes in PBST. Blots were developed by chemiluminescence.

Immunofluorescence cell staining

Immunofluorescence cell staining was carried out to verify that pelvic fibroblasts expressed the three proteins identified. Fibroblasts from vaginal wall tissues were cultured in a four-well chamber slide. The cells were fixed with 4% paraform aldehyde (PFA) and treated with 5% Triton. After washing with TBS-T and blocking with 5% normal goat serum and 1% bovine serum albumin, the slides were incubated with 50 µg/ml of rabbit anti-IL-1RA (R&D System, Minneapolis, MN, USA), 1/50 of goat anti-elafin (Cell Sciences, Norwood, MA, USA), or mouse anti-vimentin primary antibody at 4°C overnight (Chemicon, Temecula, CA, USA). After washing, these slides were then incubated with goat anti-mouse IgG-TRITC (vimentin) and goat anti-rabbit-IgG-FITC (IL-1RA), or goat anti-mouse IgG-TRITC (vimentin) and donkey anti-goat-FITC (elafin) at room temperature for 1 h in a dark chamber. After washing and mounting, the slides were examined with a fluorescence microscope.

Microarray data analysis

The microarray raw signal intensity files were first analysed using Microarray Suite 5.1 MAS 5.0 (Affymetrix). A common method of analysing the scanned intensities from the microarrays is to take the log 2 transformations so that the data will result in the normal curve required by subsequent analysis methods. Recently, other methods have been developed for analysis of these data (Li and Wong, 2001; Irizarry, 2003). Instead of working with the chip file from MAS 5.0 data directly, these methods start from files upstream in the analysis pathway. These CEL files contain signal intensity information from each probe featured on the microarrays. By working directly with CEL files, different background correction, normalization and summarization methods can be applied to enhance the data quality.

Both programs—MAS 5.0 and Robust Multiarray Average (RMA) (Irizarry, 2003)—were run to take advantage of the differing strengths of each algorithm for background correction as well as to build our confidence in the results of the analysis. We used the quantiles normalization to normalize our data, as described by Bolstad (probe level quantile normalization of high-density oligonucleotide array data; B. Bolstad, unpublished manuscript, December 2001, Berkeley, CA, USA). Both algorithms are available in microarray analysis packages (Bioconductor, open source statistical software of the R project, http://www.r-project.org).

For significant gene detection, we used simple t-tests of both parametric and non-parametric formulations, and Prediction Analysis of Microarrays (PAM). These methods were used to minimize the statistical errors that can occur when conducting analyses of so many probes (33 000 genes). We calculated the test statistics using both the raw data (parametric) and their ranks (non-parametric). The calculated p values were then adjusted according to different multiple testing procedures (Bonferroni, Holm, Hochberg, Benjamini and Hochberg, and Westfall and Young) (Westfall and Young, 1993; Dudoit, 2002). A cut-off p value of 0.05 was used. Although PAM was designed for the purpose of classification, we were able to identify a list of significant genes remaining by setting a shrinking parameter. Using all of the above statistical methods, we then generated a list of common genes.

Results

The average age of the women in the SUI group was 42.9 years (range 24–57) and in the continent group it was 43.2 years (range 36–52). In all participants, pelvic organ prolapse was no greater than stage I by POP-Q (Bump et al., 1996). Vaginal wall tissues taken from an additional 14 women with SUI and 12 continent women were used for the quantitative competitive PCR and Western blot analyses. The incontinent and continent women were similar in age, parity and BMI (Table I).

Using the raw intensity microarray data, we identified 20 genes that are up-regulated in all five pairs of women with SUI compared with continent women, with changes ranging from 1.2-fold to 78.8-fold. When we catalogued changes occurring in at least four out of five pairs, the number of the relevant genes increased to 58 up-regulated and 36 down-regulated. From this list, up-regulated genes that appear to be involved in ECM metabolism include skin-derived protease inhibitor 3 (elafin); IL-1RA; keratin 6, 14 and 16; and psoriasin 1. Down-regulated genes included α2 actin; actin depolymerizing factor; smooth muscle myosin; light polypeptide kinase; RAMP-1; tropomyosin 1; microfibril-associated glycoprotein-2; insulin-like growth factor binding protein 7; and collagen type IV α chain. Several large cDNA genes (named KIAA) that encode large, multidomain proteins were also down-regulated. The function of these proteins is mostly unknown, but they are thought to be the framework of assembly protein complexes. It is possible that these proteins are involved in ECM metabolism.

Elafin, IL-1RA and RAMP were chosen for verification of the microarray raw data because of their relatively large changes (greater than 2-fold). The presence of these genes was confirmed by both quantitative competitive PCR and Western blot (Figures 1 and 2), as well as by immunofluorescent cell staining in fibroblasts cultured from anterior vaginal wall tissues (Figure 3). RAMP-1 was not detected by Western blot due to low intensity.

Finally, we conducted extensive statistical analyses on our data to confirm our experimental results. By using MAS 5.0 and RMA to normalize our raw data and then subjecting it to various statistical analyses for comparisons (as described in Materials and methods), we generated two lists of differentially expressed genes: 387 for MAS 5.0 and 480 for RMA. Tables II and III list the common 79 genes found in both lists, with 39 up-regulated and 40 down-regulated. Elafin, keratin 16, collagen type XVII and plakophilin 1 were consistently identified as up-regulated genes by both MAS 5.0 and RMA when examined by all analytical methods.

Discussion

We hypothesize that, in addition to the mechanical stresses on the pelvic floor, some women have genetic differences that predispose them to stress urinary incontinence. These differences are reflected in altered ECM repair pathways in response to both mechanical stresses and hormonal stimulation. We have previously documented a list of relevant genes in vaginal wall tissues from reproductive-aged women with SUI in the proliferative phase of the menstrual cycle. Our data showed good segregation of incontinent and continent women based on the different response of the genes (Chen et al., 2003). To further clarify the role of hormonal stimulation on reproductive-aged women with SUI, we used microarray technology to screen for the relevant genes in the secretory phase of menses.

In this study, we identified differentially expressed genes in vaginal wall tissues from SUI women compared with controls in the secretory phase of menses, and confirmed the expression of three genes selected from the microarray raw image data. Microarray technology enabled analysis of greater than half a million probes, representing more than 33 000 human genes. The false detection rate can become significant when conducting a comparison analysis with such a large number of probes. Thus, it is crucial to apply careful statistical methods. Comprehensive statistical analysis resulted in a common list of 79 differentially expressed genes (Tables II and III). We found that elafin (a serine protease inhibitor) and keratin 14 and 16 were consistently up-regulated, by 16-, 5- and 6-fold respectively, both in the raw data and with all statistical methods.

Elafin is a host-defence protein that is absent in normal skin but present in the keratinocytes of skin affected by psoriasis, in epidermal skin tumours and after wound healing (Pol et al., 2003). Hyperproliferation is also a functional physiological response to wound healing. Other genes whose expression is highly up-regulated in healing include cytokeratins 6, 16 and 17 (Leigh et al., 1995). Similarly, the presence of elafin is strongly connected to abnormal epithelial differentiation and hyperproliferation (Wiedow et al., 1990; Schalkwijk et al., 1993). Up to this point, elafin gene expression has been studied in keratinocytes as well as in breast and lung epithelial cells (Zhang et al., 1997; Bingle et al., 2001). This study is the first to document its presence in pelvic fibroblasts.

The increased expression of elafin and keratin 16 in the hyperproliferative response suggests altered cellular responses in the pelvic tissues of women with SUI. Elastin metabolism is also implicated, since elafin is a serine protease inhibitor involved in the elastin degradation pathway. Similarly, we have documented a decrease in the mRNA and protein expressions of α1 antitrypsin inhibitor, another serine protease inhibitor, in the vaginal tissues of women with SUI in the proliferative phase of menses (Chen et al., 2004). Recent animal experiments confirm that lysyl oxidase-like protein 1 (LOXL-1) is involved in elastin deposition, and that mice deficient in LOXL-1 develop prolapse post-partum (Liu et al., 2004). These data show a causal link between elastic fibre homeostasis and pelvic floor dysfunction. We believe that both collagen and elastin metabolism may be altered in women with SUI and pelvic floor dysfunction, and that the activation of the genes involved in these pathways is hormone-dependent.

In this study and our previous study, we have now identified the relevant ECM genes that are found in vaginal tissues from women with SUI compared with continent women in the proliferative and secretory stages of menses. Genes that were differentially expressed in the proliferative phase (Chen et al., 2003) include TGFβ-3, laminin, collagen type VI, and myocyte function-related proteins. TGFβ-3 is involved in stimulation of ECM, and increases production of collagen I and III, the principal components that provide tensile strength to ligamentous tissues. It appears that collagen metabolism is activated in the proliferative phase, while elastin metabolism changes are more pronounced in the secretory phase. These differences require further protein confirmation.

During the course of these studies, not only has microarray technology evolved from simultaneous screening of 6800 genes to 33 000 genes, but the statistical methods required for these complex analyses have also changed significantly. Using these comprehensive statistical analyses, we have identified target genes for future in vitro experiments to investigate the potential involvement of reproductive hormones in the pathogenesis of SUI, and to develop a genetic screening tool for women at risk of developing SUI, as well as future treatments for prevention of SUI.

Acknowledgements

This work was supported by National Institutes of Health grant AG 17907 and Affymetrix, Inc.

References

Author notes

¹Stanford University, Stanford, CA, USA, ²Affymetrix, Inc., Santa Clara, CA 95051, USA and ³Department of Statistics, Stanford University, Stanford, CA 94305, USA