Mol. Hum. Reprod. Advance Access originally published online on March 8, 2006
Molecular Human Reproduction 2006 12(3):187-207; doi:10.1093/molehr/gal018
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
In vitro culture significantly alters gene expression profiles and reduces differences between myometrial and fibroid smooth muscle cells
Centre for Women’s Health Research, Monash University Department of Obstetrics and Gynaecology, Monash Institute of Medical Research, Clayton, Victoria, Australia
1 To whom correspondence should be addressed at: Centre for Women’s Health Research, Monash University Department of Obstetrics and Gynaecology, Monash Institute of Medical Research, 246 Clayton Road, Clayton, Victoria 3168, Australia. E-mail: marina.zaitseva{at}med.monash.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Cultured myometrial (M) and fibroid (F) smooth muscle cells (SMCs) have been widely used as a model for the study of F growth. The aim of this study was to compare gene expression profiles using microarrays between six paired M and F tissues from hysterectomy specimens, as well as cells isolated from the same tissues and cultured for up to three passages. A total of 2055 genes were differentially expressed by ANOVA between all experimental groups. Among them, 128 genes were found to be statistically different between M and F tissues. More than 1100 genes were significantly changed between tissues and cultured cells, with 648 genes common between both M and F cells at P0 and P3. Expression profiles of six genes including estrogen receptor-
(ER) and progesterone receptor (PR) were also validated using real-time PCR. These data demonstrate that large changes occur in SMC gene expression in culture, reducing differences between M and F cells. They also show that ER and PR levels are reduced in cells compared with whole tissue. These results indicate that although M and F cell cultures provide an important tool to study these tumours, in vitro studies must be carefully planned and evaluated to provide meaningful results.
Key words:
fibroids/gene expression/in vitro culture/ER/PR
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Uterine leiomyomas (fibroids) are benign neoplasms of the smooth muscle cells (SMCs) of the uterus and are the most common tumours in humans, affecting up to 77% of women of reproductive age (Cramer and Patel, 1990
). Although these tumours are not malignant, for one-third of these women, fibroids are a significant cause of reproductive and gynaecological problems, such as menorrhagia, dysmenorrhoea, chronic pelvic pain, infertility, recurrent miscarriage, preterm delivery and postpartum haemorrhage (Stewart, 2001; Flake et al., 2003). Uterine fibroids are a leading cause of hysterectomy in both Australia and USA, and they account for over 200 000 hysterectomies in US women each year (Farquhar and Steiner, 2002).
Despite the fact that fibroids adversely affect many women, their aetiology is poorly understood. These tumours grow during the reproductive years and regress after menopause, implicating the sex hormones estrogen (E) and progesterone (P) as primary factors driving fibroid growth (Flake et al., 2003). Obesity and early-age menarche, which increase the body’s lifetime exposure to ovarian steroids, increase the risk of fibroids, whereas parity, use of progestogen-containing oral contraceptives, menopause, exercise and smoking are thought to decrease the risk (Cook and Walker, 2004).
There is little understanding of the processes of fibroid growth and regression at the cellular and molecular levels. In recent years, multiple-published studies have used microarray platforms to try to elucidate genes important in fibroid growth (Ahn et al., 2003; Weston et al., 2003; Hoffman et al., 2004; Arslan et al., 2005; Lee et al., 2005; Vanharanta et al., 2005; and others). Several genes and pathways appear consistently altered in different microarray studies, including genes involved in the retinoic acid pathway, IGF2 metabolism, TGF-ß signalling and extracellular matrix formation. Some of these genes have been previously reported to be important in fibroid growth, and some genes need further investigations to confirm whether changes in gene expression translate into protein concentrations and also the relevance of these genes in fibroid pathophysiology.
The majority of published fibroid studies have used observational or correlational approaches to investigate the differences between fibroids and adjacent myometrium. Mechanistic studies have been limited by lack of suitable in vitro and in vivo models. Fibroids are rare or do not occur at all in nonhuman species, making development of animal models difficult. The best-characterized animal model is the Eker rat model, in which approximately 65% of female Tsc2EK/+ carriers develop fibroids (Walker and Stewart, 2005).
Most studies published to date have used in vitro culture models of isolated fibroid and myometrial SMCs or whole tissue explants to look at factors that may affect fibroid growth. Although it has been shown that fibroid SMCs can behave differently in culture to myometrial SMCs, it is not clear how relevant the cultured cells and their responses are to the in vivo situation. For example, E and P have been implicated as major regulators of fibroid growth, and the presence of E and P receptors (ER and PR) in fibroid and matching myometrium is well documented (Viville et al., 1997; Benassayag et al., 1999). At the same time, there is conflicting evidence about the presence of ER and PR in cultured myometrial and fibroid SMCs. Some studies report rapid loss of these receptors in culture (Sadovsky et al., 1992; Severino et al., 1996; Sampath et al., 2001), whereas others show effects of E and P on cultured cells (Matsuo et al., 1999; Swartz et al., 2005; Xu et al., 2005). A recent study (Mangioni et al., 2005) demonstrated the presence of ER protein in cultured SMCs by western blot, whereas in studies undertaken in our laboratory we were able to demonstrate the presence of ER in SMCs and ER and ERß in endothelial cells (ECs) isolated from fibroids and matching myometrium at passages 1–3 by both mRNA and protein analyses (Gargett et al., 2002a). However, this study did not compare levels of receptors in cultured cells with levels found in the tissue that these cells had been isolated from. No study to date has compared gene expression profiles of cultured cells versus tissue.
The aims of the present study were (i) to analyse gene expression profiles of paired myometrial and fibroid samples using gene microarray, (ii) to compare these to the gene expression profiles of cultured cells isolated from the same tissues and cultured up to three passages and (iii) to compare gene expression of ER and PR between tissue and cultured cells.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Tissue collection
Human myometrial and fibroid tissue was obtained from six ovulating women (mean age 47 years, range 45–53) undergoing hysterectomy for fibroids. Informed consent was obtained from each patient, and ethical approval was obtained from Southern Health Human Research Ethics Committee B. Part of the tissue was snap frozen on dry ice immediately after excision, the rest was collected in HEPES-buffered M199 culture medium (Gibco BRL, Gaithersburg, MD, USA) containing 10% fetal calf serum (FCS) (CSL, Melbourne, Australia) and 10x antibiotic–antimycotic solution (Gibco BRL)—with final concentrations of penicillin 1000 U/ml, streptomycin 1000 U/ml and fungizone 2.5 µg/ml—stored overnight at 4°C and then processed for cell culture. Snap-frozen tissue was stored at –80°C until RNA extraction. Fibroid samples were stored and processed separately from adjacent, normal (host) myometrium.
Isolation and culture of myometrial and fibroid SMCs
Myometrial and fibroid tissue was finely chopped and dissociated with collagenase type-2 and deoxyribonuclease type-I, followed by a short trypsin digestion to produce single-cell suspensions, as described previously (Gargett et al., 2000, 2002a). ECs were removed using UEA-1-coated Dynabeads (Dynal, Oslo, Norway). Myometrial and fibroid SMCs (MSMC and FSMC) were grown on uncoated plastic flasks in M199 medium containing 10% FCS, antibiotic–antimycotic solution, 10 nM 17ß estradiol and 100 nM progesterone (both from Sigma, St. Louis, MO, USA). Cells were routinely cultured and collected for RNA extraction before first passage (P0) (2 weeks in culture) and after passage 3 (P3) (approximately 5 weeks in culture). At the time of collection, SMC cultures were checked for contaminating EC using CD31 immunostaining and fluorescence-activated cell sorter (Gargett et al., 2002b).
RNA extraction
Six RNA samples were collected from each patient: myometrial and fibroid tissue, MSMC and FSMC at P0 and MSMC and FSMC at P3. Total RNA was extracted from cells and tissue samples using TrizolTM reagent (Invitrogen, Carlsbad, CA, USA) as previously described (Weston et al., 2003). RNA was further purified using Rneasy columns with on-column DNase-I treatment (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Resulting RNA was ethanol-precipitated, resuspended in RNase-free water and stored at –80°C. The final concentration of total RNA was approximately 5 µg/µl. Reference RNA for the microarray experiment was extracted from a mixture of unrelated myometrial and fibroid samples using the same protocol but without DNase-I treatment.
Microarray hybridization
Glass microarray slides spotted with approximately 8000 human cDNA sequences were purchased from the Australian Genome Research Facility (AGRF), Melbourne, Australia. An indirect labelling procedure was used to create fluorescent-labelled cDNA. First strand cDNA synthesis was performed using the SuperScriptTM Indirect cDNA Labelling System (Invitrogen). Total RNA at 50 µg per slide with an equal amount of reference RNA was used for each reaction. RNA samples were thawed on ice, briefly spun and made up to 15.5 µl volume with RNase-free water. Two microlitres of anchored Oligo(dT)20 primer (2.5 µg/µl) was added to each RNA sample, and the mixture was incubated at 70°C for 10 min. After this time, tubes were placed on ice for 5 min. Next, the following were added: 6 µl of 5x First-Strand buffer, 3 µl of 0.1 M dithiothreitol (DTT), 1.5 µl of dNTP mix (including amino-modified nucleotides) and 2 µl of SuperScriptTM III RT (400 U/µl). Reverse transcription was performed at 46°C for 2.5 h. After the reaction was complete, original RNA was degraded by addition of 15 µl of 0.25 M NaOH with incubation at 70°C for 10 min. The reaction was neutralized by addition of 25 µl of 0.2 M acetic acid, and 20 µl of 3 M sodium acetate was added to each reaction mixture. Some samples were stored overnight at –20°C before proceeding to cDNA purification and labelling. cDNA was mixed with five volumes of PB-binding buffer, loaded onto Qiaquick PCR purification columns (Qiagen), washed twice with 700 µl of PE buffer, and the column was spun for 1 min to dry it. Cy3 and Cy5 dyes from Cyscribe post-labelling reactive dye kit (Amersham, Buckinghamshire, UK) were resuspended in 15 µl of 0.1 M NaHCO3 (pH = 9). Resuspended dye was loaded directly onto the Qiaquick membrane with bound cDNA. Cy5 was used to label experimental samples; Cy3 was used to label reference samples. Columns were incubated for 1 h in the dark, and labelled cDNA was eluted with 80 µl of water. A second purification step was performed to remove any unbound dye. All samples were mixed with PB-binding buffer, and labelled experimental and reference cDNA was combined by loading them onto a single fresh Qiaquick column. cDNA was washed twice with PE buffer and eluted in 2 x 50 µl volumes of water. cDNA was dried in the vacuum drier, resuspended in 16.2 µl water, and a combination of blockers (5 µl Cot1 DNA [5 µg/µl], 5 µl Salmon sperm DNA [10 mg/ml] [both from Invitrogen] and 3.8 µl PolyA [10 mg/ml] [Sigma]) was added to reduce nonspecific binding to the slides. Samples were mixed with 14.7 µl formamide, 14.7 µl 10x standard saline citrate (SSC), 0.6 µl 10% sodium dodecyl sulphate (SDS) and heated for 2 min at 100°C to denature cDNA, allowed to cool to room temperature and applied to microarray slides under a cover slip for hybridization. Hybridization was performed at 42°C for 18 h in a humidified chamber. Slides were washed at room temperature with 1x SSC/0.2% SDS for 5 min, 0.1 SSC/0.2% SDS for 5 min and twice with 0.1 SSC for 2 min and dried by centrifuging at 1000 g for 5 min.
Scanning and quantification
Microarray slides were scanned using a GenePix 4000B scanner, and the data were quantified and extracted using GenePix Pro 5.0 software (Axon Instruments, Union City, CA, USA).
Real-time quantitative PCR
RNA for all 36 samples used in the microarray experiment was used for subsequent real-time quantitative PCR (RT–QPCR). One microgram of DNase-treated RNA was mixed with 1 µl random primers (3 µg/µl; Invitrogen), 2 µl 10 mM dNTPs (Roche, Basel, Switzerland), 4 µl 5x RT buffer (Roche), 0.5 µl RNAsin (Promega, Madison, WI, USA), 2 µl 0.1 M DTT (Invitrogen) and 0.2 µl AMV-RT (Roche) and incubated at 42°C for 1 h.
A Roche Light Cycler and LC fast-start DNA master SYBR green kit (Roche) was used to perform RT–QPCR, according to the manufacturer’s instructions. The primer sequences used are summarized in Table I. Relative mRNA levels for each of the genes were determined using specific cDNA standards. All results were normalized using 18S RNA as a housekeeping gene to correct for differences in concentration of the starting template.
|
Statistical analysis
Data files generated by GenePix Pro software were imported into Gene Spring (Silicon Genetics, Agilent, Palo Alto, CA, USA) version 7.0, and analyses were performed using default settings of the software. Imported data were log transformed and normalized using per chip intensity-dependent (LOWESS) normalization, which accounts for differential labelling efficiency of the two fluorescent dyes. This adjusts the control channel for each gene based on an algorithm that takes into account the ratio of the signal versus control and the magnitudes of the signal and control.
Parametric analysis of variance (ANOVA) with Benjamini–Hochberg false discovery rate correction at P = 0.05, followed by Tukey post hoc tests were performed on log-transformed ratios of normalized data to identify genes that were significantly altered by culture conditions. The generated gene lists were further analysed according to GO (gene ontology) and other gene-related information collected using Gene Spring, Gene Tools (web-based software and databases created by NTNU; http://www.genetools.no) and NCBI databases.
To validate results of microarray experiments, microarray results were compared with results of RT–QPCR using Graph Pad PrizmTM software. Pearson correlation and linear regression were used to correlate RT–QPCR and microarray results. To confirm results of Gene Spring analyses, microarray data were also log transformed and analysed by parametric ANOVA, followed by Tukey post hoc testing.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
cDNA microarray analyses
Microarray slide data were analysed in six groups: myometrial tissue (Myo), fibroid tissue (Fib), myometrial SMC at passage 0 (MSMC P0), fibroid SMC at passage 0 (FSMC P0), myometrial SMC at passage 3 (MSMC P3) and fibroid SMC at passage 3 (FSMC P3). Parametric ANOVA with Benjamini–Hochberg false discovery rate correction at P = 0.05, followed by Tukey post hoc test identified 2055 genes that were differentially expressed between all groups. Table II summarizes results of the ANOVA and the distribution of differentially expressed genes between different groups. The smallest differences were found between Myo and Fib and between cultured cells, particularly cells at P3, whereas the largest differences were found between Myo and Fib tissues and cultured cells. A gene tree was created using Gene Spring from 2055 genes identified by ANOVA to visualize differences in gene expression (Figure 1).
|
|
Genes significantly different between myometrium and fibroid
A total of 128 genes showed significantly different expression between myometrial and fibroid tissue by ANOVA. This list was further filtered to identify genes that were up- or down-regulated at least 1.5 times (36 and 78, respectively) or two times (22 and 37, respectively) in fibroids compared with myometrium.
Genes up-regulated in fibroids
The most prevalent groups of up-regulated genes belonged to the biological process groups of cell growth and cell maintenance (nine genes), development (five genes) and to the molecular function groups of binding (seven genes) and catalytic activity (nine genes). Genes up-regulated in fibroids included IGF2 (two probes), PTK7, KLF10 and three different collagens (COL1A2, COL5A2 and COL3A1) (Table III).
|
Genes down-regulated in fibroids
The most prevalent groups of down-regulated genes belonged to the biological process groups of cell growth and cell maintenance (24 genes), cell communication (five genes) and development (six genes) and to the molecular function groups of binding (10 genes), catalytic activity (20 genes) and signal transducer activity (eight genes). Genes down-regulated in fibroids included angiogenic factor CYR61, prostaglandin D2 synthase PTGDS, dual specificity phosphatases DUSP1, DUSP3 and DUSP8, ANXA1, CCL21, transcription factors KLF 4, KLF6, KLF7, NR2F2, NR3C1 and proteins responsible for cell–matrix adhesion ADAM15, FBLN5, SGCE, ITGA5, ITGA7 and ITGB5 (Table IV).
|
Genes significantly altered by culture conditions
More than a thousand genes were altered for each cultured cell type when compared with the tissue from which it was derived, with a slight increase in the number of altered genes at P3 compared with P0. Genes that were found to be altered by culture conditions were filtered to obtain genes with at least 1.5 and two-fold up- or down-change. Between 300 and 500 genes were either up- or down-regulated for each cell type at both passages when compared with the tissue. A total of 635 genes were significantly altered between Myo and MSMC at P0 and P3, with 308 genes up-regulated and 327 down-regulated at least two-fold. A total of 594 genes were significantly altered between Fib and FSMC at P0 and P3, with 292 genes up-regulated and 302 down-regulated at least two-fold (Table V). These lists were compared between each other, revealing a large degree of similarity between lists. Out of these genes, 635 genes were commonly altered in all cells compared with tissues at least 1.5-fold, with 329 being up-regulated and 306 being down-regulated, and 507 genes were commonly altered in all cells compared with tissues at least two-fold, with 242 being up-regulated and 265 being down-regulated (Table V). Genes altered by culture conditions at least two-fold in each cell type were also identified among the 128 genes that were found to be statistically differentially expressed between Myo and Fib. Between 17 and 24 genes were up-regulated, between 6 and 9 genes were down-regulated, with 12 genes out of 128 up-regulated and two genes down-regulated in all cell types (Table VI). Some genes were represented by two probes and showed similar fold change. EIF2S2 probes were up-regulated 2.3- and 2.9-fold, ENO1 probes 6.5- and 4.8-fold. EPHB4 probes were down-regulated 0.33- and 0.36-fold, RPL4 0.22- and 0.29-fold and SMTH0.17- and 0.17-fold, respectively. Genes with at least two-fold differences that were common between both Myo and Fib tissue and all four cell types are summarized in Tables VII and VIII.
|
|
|
|
Genes up-regulated in culture
The largest groups of up-regulated genes belonged to the biological process groups of cell growth and cell maintenance (77 genes), cell communication (19 genes) and development (25 genes) and to the molecular function groups of binding (23 genes), catalytic activity (52 genes), signal transducer activity (17 genes) and defence immunity protein activity (five genes), with some overlap between similar groups. There were genes responsible for cell adhesion (ADAM9, ADAM10, CDH11, CDH2, ITGA5, ITGB5, LOXL2, NID2, THBS1 and THBS2) and cytoskeleton and muscle fibre organization and contractility (AFAP, KBTBD10, DAG1, MRCL3, MYO6 and TUBB). A large group included 67 genes involved intranspor within the cell and across the membranes, including 10 proteins that belong to different families of solute carriers, 16 genes responsible for electron transport, eight genes responsible for metal ion transport, 12 genes involved in protein and three in amino acid transport and two genes for glucose transport. Up-regulated genes also included a large number of genes responsible for cellular metabolism, including fatty acid and lipid metabolism (PNPLA4, PCCA, DHRS3 and ACSL4), carbohydrate metabolism (SMPD1, ACLY, GBE1 and B4GALT4) and protein and amino acid metabolism (GOT1, PTS, PLOD2 and GGH). There were also 20 genes with function in metal ion binding, including nine involved in calcium binding and six different metallothionein genes.
Genes down-regulated in culture
Down-regulated genes belonged to the biological process groups of cell growth and cell maintenance (74 genes), cell communication (28 genes) and development (16 genes) and to the molecular function groups of binding (38 genes), catalytic activity (38 genes), signal transduction activity (20 genes), defence immunity protein activity (11 genes) and transport (10 genes). There was some overlap between similar groups, with several genes belonging to more than one group.
There were several groups of genes that stood out during the analyses. A total of 27 different receptors were down-regulated by culture, including interleukin receptors IL10RB, IL11RA and IL17R, platelet-activating factor receptor, transforming growth factor beta receptor-III, tumour necrosis factor receptor superfamily member 25 and retinoic acid receptor-ß. Large number of genes tresponsible for transcription was also down-regulated. Expression of 10 different zinc finger proteins was reduced in cells compared with tissue. Other genes involved in transcription included MYST 3 and 4, HOX A10 and D4, NR2F6, CNOT8, SCML2, SOX13, TGFB1/4(TSC22) and TTF1. Another prominent group was genes important in cytoskeleton and muscle fibre organization and muscle contractions (MYH11, GSN, SMTH, CNN1, EBP49, DIAPH2, TPM4 and ARHGDIB). There were a number of down-regulated genes associated with EC, including platelet/EC adhesion molecule (CD31 antigen), endothelial differentiation lysophosphatidic acid G-protein-coupled receptor-2 and thrombomodulin, which were expected, given that EC were selectively removed from our SMC cultures.
Validation of microarray results by RT–QPCR
Four genes were chosen to be validated by RT–QPCR, annexin-1 (ANXA1), thrombospondin-1 (THBS1), gelsolin (GSN) and transforming growth factor-ß-stimulated protein TSC22 (TSC22). These four genes were chosen as they were up-regulated (ANXA1, P = 6.97E–10 and THBS1, P = 8.15E–09) or down-regulated (GSN, P = 1.75E–14 and TSC22, P = 4.05E–09) by culture conditions. To confirm results of Gene Spring analyses and to further validate microarray data, mRNA expression data for both techniques were log transformed and analysed by ANOVA, followed by Tukey’s post hoc testing. Results of all analyses were similar for both techniques. The data for mRNA expression profiles from both microarray and RT–QPCR are shown in Figure 2. All genes showed significant correlation between mRNA expression profiles obtained from microarray and RT–QPCR experiments (THBS1, P = 0.0171, r2 = 0.16; TSC22, P < 0.0001, r2 = 0.53; ANXA1, P < 0.0001, r2 = 0.4; GSN, P = 0.0026, r2 = 0.27) (Figure 3).
|
|
ER and PR expressions by microarray and RT–PCR
Expression of ER and PR was further investigated owing to the importance of E and P in fibroid biology and ongoing controversy regarding receptor expression in cultured cells. ERß could not be amplified in cultured cells because MSMC and FSMC do not express this receptor (Gargett et al., 2002a). PR expression was not significantly altered according to microarray data, but the level of expression was very low, or below detection, for all samples. ER expression was significantly down-regulated by microarray analysis in three out of four experimental groups when compared with respective tissue (in MSMC P0 and P3 and FSMC P0). FSMC P3 analysis was not significant due to two aberrant data points where a problem occurred in the reading for the control channel, and these two points were excluded from further analyses. After exclusion of these data points, both microarray and PCR data showed a significant down-regulation (P < 0.001) of ER mRNA in all cells compared with the tissue. PR-PCR results demonstrated a significant reduction in PR mRNA levels between Myo and MSMC P0 (P < 0.01) and Myo and Fib and cells at P3 (P < 0.001). There was no significant difference in expression between Fib and FSMC at P0. Levels of PR mRNA were approximately 10x higher than levels of ER mRNA, and there was an approximate 10-fold decrease in receptor levels for both ER and PR at P0, and 30-fold at P3. There was a significant correlation between microarray and RT–PCR data for ER (P < 0.0001; r2 = 0.76) and no correlation for PR (r2 = 0.01). Microarray and RT–PCR data and correlation graphs are shown in Figure 4.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
This is the first study to report gene expression profiles of myometrial and fibroid cultured cells at two different passages (P0 and P3) and compare them with the gene expression profiles of the tissues from which the cells have been derived. We observed significantly altered gene expression profiles in cultured cells compared with their original tissue, with more than 500 genes altered in all cell–tissue comparisons, regardless of tissue type or time in culture. Under in vitro conditions, the gene expression profiles of myometrial and fibroid SMCs were more similar to each other, reducing differences that exist between myometrium and fibroids in vivo. Studies performed in other tissues, mostly tumours, report similar findings to those reported here.
A study by Shai (Shai et al., 2005) examined robustness of gene expression profiling in glioma and derived cells using microarray. Similar to our findings, all their cell lines clustered separately from tissue samples during microarray analysis, even though they tested two distinct tumour subtypes in their study, suggesting that gene expression pattern of cell cultures is largely shaped by their in vitro culture environment. In this study, analysis of genes changed by culture conditions revealed enrichment of genes involved in nucleotide and nucleic acid metabolism, cell proliferation and signal transduction. There was also a loss of gene expression involved in development, signal transduction and transporter activity, including brain-specific genes, which may indicate a more undifferentiated phenotype, compared with primary tumour cells.
Studies by Luo and others (Luo et al., 2005a,b) are the only studies published to date that performed microarrays on myometrium, fibroids and cells isolated from these tissues. The authors undertook two comprehensive studies (i) to assess gene expression profiles of untreated and GnRH analogue treated myometrium and fibroids, as well as untreated and treated fibroid and myometrial SMCs and (ii) to investigate changes in gene expression profiles of fibroid and myometrial SMCs in response to TGFß-1. In this study, the authors found a substantial difference in gene expression between tissue and cultured cells. No detailed comparison was provided, but gene ontology assessment indicated that the majority of genes changed in culture when compared with the whole tissue were involved in transcriptional regulation and metabolism, cell cycle regulation, extracellular matrix and adhesion molecules and signal transduction and transcription factors. They also reported that both tissue and cultured cells had altered expression profiles in response to GnRHa and TGFß-1 but could not distinguish fibroids/FSMC and myometrium/MSMC based on single gene markers uniformly expressed only in fibroids and/or myometrium. All these studies are in agreement with the present study which has shown that culture profoundly affects genes involved in cell growth, maintenance, cell-to-cell communication, development and differentiation, signal transduction, metabolism and cellular transport, despite the different tissue origins of the cell types being tested.
Another important finding of our study is that although there are significant differences in gene expression between myometrium and fibroids, similarities between host and diseased tissue are greater than between cells and the tissue from which these cells were isolated. A study comparing gene expression profiles of 60 tumour cell lines, derived from nine different tissues, with the corresponding in vivo tumours and normal tissues, further supports the current findings (Sandberg and Ernberg, 2005). This study demonstrates that gene expression was more similar between normal and tumour tissues, than between tumour tissues and corresponding cell lines. Out of nine tissues tested, four did not show any similarity in gene expression with corresponding cell lines, three showed 1–6% similarity, one of the tissues had 20% and another 55% similarity, with an average of 5% across all comparisons. Similarities between tumour and normal tissue were 36% on average.
Because cellular environment plays a crucial role as a driver of gene expression, altering culture conditions to make them more similar to the in vivo environment, for example, by growing cells in a three-dimensional (3-D) collagen matrix, may reduce differences between in vitro and in vivo gene expression profiles. A study by Oswald (Oswald et al., in press) evaluated the effects of collagen matrix on gene expression profile of CD34+ haematopoietic stem cells (HSC). Using a microarray platform, the authors compared gene expression and clonogenicity of cells grown in collagen matrix with cells grown in suspension cultures. Three-dimensional collagen matrix significantly altered gene expression of HSCs, with increased expression of several growth factors, chemokines and cytokines. The authors speculated that 3-D collagen matrix may more closely mimic the environment of the stem cell niche and thus improve clonogenicity and help maintain cells in an undifferentiated phenotype. Another study (Camphausen et al., 2005) examined the effects of environment on gene expression profiles of two glioma cell lines grown under three different conditions: in vitro as a monolayer and in vivo as s.c. or intracerebral tumours. This study demonstrated that gene expression profiles of the two cell lines differed significantly between the in vitro and in vivo s.c. models. The gene expression profiles were also significantly different between s.c. and intracerebral in vivo models, with the intracerebral model expressing more genes related to CNS function and fewer genes related to cell cycle progression and regulation than the cells grown in vitro. It was also noted that under intracerebral conditions the gene expression profiles of the two cell lines were more similar, whereas under s.c. conditions they were actually closer to cells under in vitro conditions. These data demonstrate that normal environment has profound effects on gene expression. Our study shows that monolayer in vitro myometrial and fibroid cell culture induces drastic changes in gene expression. Growing these cells in collagen 3-D matrix, which more closely mimics these cells in vivo environment, may be a better alternative and could help to reduce differences in gene expression induced by culture conditions.
Another important finding of this study was that ER and PR expression was reduced by culture conditions. This is the first study to undertake a systematic analysis and compare ER and PR expression levels between tissue and cultured cells. We report a significant down-regulation of ER and PR mRNA in cultured myometrial and fibroid SMCs when compared with tissue from which these cells have been isolated. There was around a 10-fold decrease in receptor levels between tissue and cells at passage 0 and a 30-fold decrease between tissue and cells at passage 3.
Presence of ER and PR in myometrium and fibroids in vivo is well documented, but there are confusing data regarding the presence of ER and PR and responsiveness of myometrial and fibroid SMCs to E and P in vitro. Various studies have used primary cultures of myometrial and fibroid SMCs as their experimental model to study responses to E and P with variable results. One study (Severino et al., 1996) reported the rapid loss of ER and PR in human fibroid and myometrial explant cultures (approximately 75% decrease after 8 h), whereas another study (Sadovsky et al., 1992) demonstrated that concentration of ER decreases dramatically (63%) in rabbit uterine myocytes after 9–14 days in culture. More recent work (Sampath et al., 2001) also claimed that they were unable to maintain steroid receptor-positive primary cultures of uterine SMCs under a variety of conditions and chose to transfect cells for their experiments. In contrast, multiple studies have reported proliferative and transcriptional effects of E and P on cultured myometrial and fibroid SMCs. A study by Andersen et al. (1995) has shown that both fibroid and myometrial primary cells have a transcriptional response to estrogen stimulation, with fibroid SMCs showing greater response compared with autologous myometrial cultures. Another study (Matsuo et al., 1999) showed that both E and P increase numbers of proliferating cell nuclear antigen (PCNA)-positive primary fibroid SMCs, whereas only E has this effect on myometrial SMCs. This study also demonstrated that E and P modulate epidermal growth factor (EGF) and EGF-R expression in fibroid but not myometrial cells. Recent work (Xu et al., 2005) reported that the PR modulator CDB-2914 inhibits FSMC proliferation via a decrease in PCNA and an increase in apoptosis by up-regulating cleaved caspase-3 and PARP and down-regulating Bcl-2 but did not report on PR expression. A study by Swartz (Swartz et al., 2005) performed a microarray study on E stimulated FSMC and MSMC. This study used SMCs purchased from two different sources, and no details were given with regard to passage number of the cells used. Results of this study showed 12 genes that were differentially regulated in MSMC and 19 genes in FSMC, with seven of the down-regulated genes common to both lists. This study demonstrated transcriptional effects of estrogen on cultured cells, but it did not examine the presence of ER in these cells. Studies in our laboratory were able to demonstrate ER mRNA and protein in passaged SMCs (Gargett et al., 2002a), and a recent study (Mangioni et al., 2005) demonstrated the presence of ER in cultured myometrial and fibroid SMCs by western blot.
Our study together with other studies demonstrates that cultured myometrial and fibroid SMCs express ER and PR and respond to E and P in culture, but levels of both receptors are reduced in cultured cells compared with the whole tissue, which may indicate reduced responsiveness of myometrial and fibroid cells to steroid hormones in vitro. Another interesting finding was that the levels of PR were around 10 times higher than levels of ER both in tissue and in cultured cells. Despite a significant reduction in expression levels in culture, both receptors were easily detectable by PCR in all experimental groups, which could explain why multiple studies have reported that myometrial and fibroid SMCs respond to E and P treatment in vitro.
In the present study, we have also identified 35 genes that were up-regulated and 78 that were down-regulated in fibroids compared with myometrium. There have been multiple microarray studies to date that have investigated gene expression of fibroids and myometrium using various platforms. We compared our lists with other published studies and found many similarities. Out of 35 genes up-regulated in fibroids, at least nine have been reported previously. These genes include IGF2, which is one of the most common observations in multiple studies (Weston et al., 2003; Arslan et al., 2005; Lee et al., 2005; Vanharanta et al., 2005; and others), collagens COL5A2 and COL3A1 (Tsibris et al., 2002; Weston et al., 2003; Lee et al., 2005; Vanharanta et al., 2005), calpain 6 (Weston et al., 2003), integral membrane protein 2C (Vanharanta et al., 2005), neuronal pentraxin 2 (Lee et al., 2005), protein phosphatase 2 (Weston et al., 2003; Lee et al., 2005) and protein tyrosine kinase 7 (Arslan et al., 2005). High-mobility group AT-hook 2 (HMGA2) gene, which is associated with rearrangement of chromosome 12q15, one of the most common chromosomal abnormalities found in fibroids (Gross et al., 2003), was also up-regulated in fibroid tissue compared with myometrium. Out of 78 genes that were down-regulated in fibroids, 23 have been previously reported by other studies and included genes such as CYR61, DUSP1, ANXA1, ABLIM1, CAV2, CCL21, CITED2, PTGDS and RNASE4 (Arslan et al., 2005). Other genes that were previously reported by at least one other study included component 1 r and s subcomponents (Lee et al., 2005; Vanharanta et al., 2005), Kruppel-like factor 4 (Arslan et al., 2005; Vanharanta et al., 2005), retinoic acid-receptor responder 2 (Lee et al., 2005; Vanharanta et al., 2005) and retinol-binding protein 1 (Vanharanta et al., 2005). Possible mechanisms of involvement of these genes in fibroid development have been discussed in reviews published recently and include pathways associated with growth and proliferation, differentiation, extracellular matrix formation and retinoid metabolism (Catherino et al., 2004; Arslan et al., 2005).
In addition, we also identified groups of genes that belonged to distinct gene families. For example, we identified three dual-activity phosphatases that were down-regulated in fibroids. DUSP-1 has been reported multiple times by previous studies, but DUSP3 and DUSP8 have not been reported before, which may be due to the fact that they only showed a modest decrease in expression levels compared with DUSP1 (–1.8 and –1.7 versus –3.7-fold). In addition to down-regulation of DUSP1 (–7.41-fold), Hoffman (Hoffman et al., 2004) also reported down-regulation of DUSP6 (–2.11-fold). Dual-specificity phosphatases have a central role in the complex regulation of signalling pathways that are involved in cell-stress responses, proliferation and death via their ability to control activity of MAP kinases (Camps et al., 2000; Theodosiou and Ashworth, 2002), and their aberrant expression may contribute to fibroid growth. It has also been reported that retinoic acid inhibits Jun N-terminal kinase via increasing DUSP activity (Lee et al., 1999), which may represent one of the mechanisms by which aberrations in the retinoic acid pathway may contribute to fibroid growth.
Another group of genes that were altered in fibroids is the group of Kruppel-like factors. KLF4 has previously been reported as down-regulated in fibroids by at least two other studies (Arslan et al., 2005; Vanharanta et al., 2005). A study by Vanharanta reported that KLF2 and KLF5 were also down-regulated in fibroids compared with myometrium. In our study, we found that KLF10 was up-regulated, whereas KLF6 and KLF7 were down-regulated along with KLF4. These genes were possibly not picked by other studies due to the modest change (around two-fold in both studies for all genes). Kruppel-like factors are zinc finger-containing transcription factors with diverse regulatory functions in cell growth, proliferation, differentiation and neoplastic transformation. These proteins can function as activators and/or repressors of transcription depending on the member of the family and cellular context (reviewed by Kaczynski et al., 2003). KLF10, the only member that was up-regulated in fibroids, belongs to a different subclass and acts as a repressor of transcription. All down-regulated KLFs belong to the same subclass and all except KLF8 can act as activators of transcription (Kaczynski et al., 2003). It is also interesting to note that there is a crosstalk between KLF factors and PRs (Simmen and Simmen, 2002), and that one of the members of KLF family, PLZF, is induced by P in endometrial stromal cells and myometrial SMCs (Fahnenstich et al., 2003).
Even though many genes that were significantly altered between fibroids and myometrium have previously been reported by multiple studies, large numbers of genes have not been reported before. Similar or less overlap of genes has been reported in previous microarray studies. Most studies that have examined differences between myometrium and fibroids using microarrays produced lists of approximately 100 genes that were differentially expressed between the two tissue types. An earlier review of five fibroid microarray studies (Catherino et al., 2004) reported only eight genes that were common in at least three out of five studies reviewed, with one study having no overlapping genes. A more recent review (Arslan et al., 2005) compared results of nine studies. The authors reported that there were eight genes identified by at least five different studies, 11 genes reported by four studies, 12 genes identified by three studies and 40 genes reported by two studies. The differences in these reported results could be attributed to several factors. Fibroids are highly heterogeneous tumours, with large variability between patients where differences between individuals can be greater than between normal and diseased tissue. Often, only small numbers of subjects are used, and selection criteria for the subjects may vary, with possible racial and other differences that could all contribute to the variability in results. Different studies have used various microarray platforms, including cDNA and oligo-based chips, with different numbers of genes involved. Different statistical analyses have also been used to examine the data generated. All these factors may increase variability in the resulting gene lists produced.
Validation of microarray data by other techniques such as real-time PCR is essential to provide reassurance that results obtained from microarray data are reproducible in other systems. Studies previously performed in our laboratory utilizing a different microarray platform (Ponnampalam et al., 2004) reported difficulties with validation of microarray data and verification of clone sequences spotted on the array. In our study, five out of six genes tested showed a significant correlation between microarray and PCR data. Microarray results for PR, the only gene that did not show a significant correlation, had low signal intensity for both test and reference channels, with signal in several samples below detection levels, which may be the reason for this aberrant result. Reproducibility of results within one sample is important to consider. Replicates of the same sample were not performed in the present study due to the relatively large number of samples done. During the analysis, we found that many genes present on array by two different probes showed similar fold change for each probe, which gives us some added confidence in the reproducibility of our array results. We have also observed that a large number of genes that were changed between tissue and corresponding cultured cells were common between all four tissue–cell comparisons. Many genes that were significantly altered between myometrium and fibroids have been previously reported by other microarray studies. Validation results and consistency between our study and other published studies and a high similarity in changes observed for all cell types tested allow confidence in the results.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Our study provides important information regarding the gene expression profile of myometrium, fibroids and SMCs that have been isolated from these tissues and cultured for up to three passages. We demonstrate that culture conditions significantly alter the gene expression profile of myometrial and fibroid SMCs, reducing differences between them in vitro when compared with in vivo. We also examined expression of ER and PR in myometrial and fibroid tissue and cultured cells and report a significant reduction in mRNA levels for both receptors in cultured cells but show that both receptors are still present at detectable levels. We confirmed findings of other microarray studies on myometrium and fibroids and added a number of previously unreported genes to list of differences between the two tissues. These results indicate that though myometrial and fibroid cell cultures provide an important tool to study these tumours, in vitro studies must be carefully planned and evaluated to provide meaningful results.
|
Acknowledgements |
|---|
We thank Sisters Nancy Taylor and Nicky Sam for the collection of tissue samples and to various gynaecological surgeons affiliated with Monash Medical Centre who provided subjects for the study. We also thank Anna Ponnampalam for expert technical advice and Jane Girling for critical review of the manuscript. This work was supported by the Royal Australian and New Zealand College of Obstetricians and Gynaecologists Members Research Foundation Scholarship to Dr B.J.V. P.A.W.R. is a Principal Research Fellow of the National Health & Medical Research Council of Australia (NH & MRC Fellowship grant No. 134063). M.Z. is a holder of an Australian Postgraduate Award.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Ahn WS, Kim KW, Bae SM, Yoon JH, Lee JM, Namkoong SE, Kim JH, Kim CK, Lee YJ and Kim YW (2003) Targeted cellular process profiling approach for uterine leiomyoma using cDNA microarray, proteomics and gene ontology analysis. Int J Exp Pathol 84,267–279.[CrossRef][ISI][Medline]
Andersen J, DyReyes VM, Barbieri RL, Coachman DM and Miksicek RJ (1995) Leiomyoma primary cultures have elevated transcriptional response to estrogen compared with autologous myometrial cultures. J Soc Gynecol Investig 2,542–551.[CrossRef][ISI][Medline]
Arslan AA, Gold LI, Mittal K, Suen TC, Belitskaya-Levy I, Tang MS and Toniolo P (2005) Gene expression studies provide clues to the pathogenesis of uterine leiomyoma: new evidence and a systematic review. Hum Reprod 20,852–863.
Benassayag C, Leroy MJ, Rigourd V, Robert B, Honore JC, Mignot TM, Vacher-Lavenu MC, Chapron C and Ferre F (1999) Estrogen receptors (ER/ERß) in normal and pathological growth of the human myometrium: pregnancy and leiomyoma. Am J Physiol 276,E1112–E1118.
Bumke-Vogt C, Bahr V, Diederich S, Herrmann SM, Anagnostopoulos I, Oelkers W and Quinkler M (2002) Expression of the progesterone receptor and progesterone-metabolising enzymes in the female and male human kidney. J Endocrinol 175,349–364.[Abstract]
Camphausen K, Purow B, Sproull M, Scott T, Ozawa T, Deen DF and Tofilon PJ (2005) Influence of in vivo growth on human glioma cell line gene expression: convergent profiles under orthotopic conditions. Proc Natl Acad Sci USA 102,8287–8292.
Camps M, Nichols A and Arkinstall S (2000) Dual specificity phosphatases: a gene family for control of MAP kinase function. FASEB J 14,6–16.
Catherino W, Salama A, Potlog-Nahari C, Leppert P, Tsibris J and Segars J (2004) Gene expression studies in leiomyomata: new directions for research. Semin Reprod Med 22,83–90.[CrossRef][ISI][Medline]
Cook JD and Walker CL (2004) Treatment strategies for uterine leiomyoma: the role of hormonal modulation. Semin Reprod Med 22,105–111.[CrossRef][ISI][Medline]
Cramer SF and Patel A (1990) The frequency of uterine leiomyomas. Am J Clin Pathol 94,435–438.[ISI][Medline]
Fahnenstich J, Nandy A, Milde-Langosch K, Schneider-Merck T, Walther N and Gellersen B (2003) Promyelocytic leukaemia zinc finger protein (PLZF) is a glucocorticoid- and progesterone-induced transcription factor in human endometrial stromal cells and myometrial smooth muscle cells. Mol Hum Reprod 9,611–623.
Farquhar CM and Steiner CA (2002) Hysterectomy rates in the United States 1990–1997. Obstet Gynecol 99,229–234.
Flake GP, Andersen J and Dixon D (2003) Etiology and pathogenesis of uterine leiomyomas: a review. Environ Health Perspect 111,1037–1054.[ISI][Medline]
Gargett CE, Bucak K and Rogers PA (2000) Isolation, characterization and long-term culture of human myometrial microvascular endothelial cells. Hum Reprod 15,293–301.
Gargett CE, Zaitseva M, Bucak K, Chu S, Fuller PJ and Rogers PA (2002a) 17ß-estradiol up-regulates vascular endothelial growth factor receptor-2 expression in human myometrial microvascular endothelial cells: role of estrogen receptor- and -ß. J Clin Endocrinol Metab 87,4341–4349.
Gargett CE, Bucak K, Zaitseva M, Chu S, Taylor N, Fuller PJ and Rogers PA (2002b) Estrogen receptor- and -ß expression in microvascular endothelial cells and smooth muscle cells of myometrium and leiomyoma. Mol Hum Reprod 8,770–775.
Gross KL, Neskey DM, Manchanda N, Weremowicz S, Kleinman MS, Nowak RA, Ligon AH, Rogalla P, Drechsler K, Bullerdiek J et al. (2003) HMGA2 expression in uterine leiomyomata and myometrium: quantitative analysis and tissue culture studies. Genes Chromosomes Cancer 38,68–79.[CrossRef][ISI][Medline]
Hoffman PJ, Milliken DB, Gregg LC, Davis RR and Gregg JP (2004) Molecular characterization of uterine fibroids and its implication for underlying mechanisms of pathogenesis. Fertil Steril 82,639–649.[CrossRef][ISI][Medline]
Huang C, Kim Y, Caramori ML, Fish AJ, Rich SS, Miller ME, Russell GB and Mauer M (2002) Cellular basis of diabetic nephropathy: II. The transforming growth factor-beta system and diabetic nephropathy lesions in type 1 diabetes. Diabetes 51,3577–3581.
Kaczynski J, Cook T and Urrutia R (2003) Sp1- and Kruppel-like transcription factors. Genome Biol 4,206.[CrossRef][Medline]
Latil A, Bieche I, Vidaud D, Lidereau R, Berthon P, Cussenot O and Vidaud M (2001) Evaluation of androgen, estrogen (ER and ERß), and progesterone receptor expression in human prostate cancer by real-time quantitative reverse transcription-polymerase chain reaction assays. Cancer Res 61,1919–1926.
Lee EJ, Kong G, Lee SH, Rho SB, Park CS, Kim BG, Bae DS, Kavanagh JJ and Lee JH (2005) Profiling of differentially expressed genes in human uterine leiomyomas. Int J Gynecol Cancer 15,146–154.[CrossRef][ISI][Medline]
Lee HY, Sueoka N, Hong WK, Mangelsdorf DJ, Claret FX and Kurie JM (1999) All-trans-retinoic acid inhibits Jun N-terminal kinase by increasing dual-specificity phosphatase activity. Mol Cell Biol 19,1973–1980.
Luo X, Ding L, Xu J and Chegini N (2005a) Gene expression profiling of leiomyoma and myometrial smooth muscle cells in response to transforming growth factor-ß. Endocrinology 146,1097–1118.
Luo X, Ding L, Xu J, Williams RS and Chegini N (2005b) Leiomyoma and myometrial gene expression profiles and their responses to gonadotropin-releasing hormone analog therapy. Endocrinology 146,1074–1096.
Mangioni S, Vigano P, Lattuada D, Abbiati A, Vignali M and Di Blasio AM (2005) Overexpression of the Wnt5b gene in leiomyoma cells: implications for a role of the Wnt signaling pathway in the uterine benign tumor. J Clin Endocrinol Metab 90,5349–5355.
Matsuo H, Kurachi O, Shimomura Y, Samoto T and Maruo T (1999) Molecular bases for the actions of ovarian sex steroids in the regulation of proliferation and apoptosis of human uterine leiomyoma. Oncology 57,49–58.
Oswald J, Steudel C, Salchert K, Joergensen B, Thiede C, Ehninger G, Werner C and Bornhauser M (in press) Gene expression profiling of CD34+ hematopoietic cells expanded in a collagen I matrix. Stem Cells.
Ponnampalam AP, Weston GC, Trajstman AC, Susil B and Rogers PA (2004) Molecular classification of human endometrial cycle stages by transcriptional profiling. Mol Hum Reprod 10,879–893.