| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Molecular Medicine and Gene Therapy, Institute of Laboratory Medicine, Lund, Sweden
2 The Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund University Hospital, Lund, Sweden
3 Complex Systems Division, Department of Theoretical Physics, Lund University, Lund, Sweden
4 Department of Cardiology, Heart Lung Center, Utrecht, the Netherlands
5 Laboratory of Experimental Carcinogenesis, National Cancer Institute, Bethesda, Maryland
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
signal transduction; Smad; microarrays
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Because of its importance in several different contexts, thorough investigations have led to considerable knowledge about the mechanisms of TGF-ß signal transduction (13). TGF-ß belongs to a superfamily of cytokines that also includes activins/inhibins, bone morphogenetic proteins (BMPs), growth differentiation factors, and a few other related proteins. The TGF-ß family receptors consist of type I and type II transmembrane serine/threonine kinase receptors. On the cell surface, TGF-ß binds to the constitutively active TßRII. Upon binding of the ligand, two TßRIIs and two TGF-ß type I receptors (Alk5s) are brought together, forming a heterotetrameric complex. The kinase domain of Alk5 is phosphorylated by TßRII, and subsequently Alk5 phosphorylates and activates the intracellular receptor-activated (R)-Smads (Smad2 and -3), which heteroligomerize with the co-mediator, Smad4. This complex translocates into the nucleus, where it recruits transcriptional co-activators and co-repressors to control gene expression. In addition, it has been reported that phosphorylation of Alk5 initiates other pathways of signaling, including the mitogen-activated protein kinase (MAPK) family pathways, comprised of the extracellular signal-regulated kinase (ERK) pathway, c-Jun NH2-terminal kinase (JNK) pathway, and the p38 pathway. The only known function of TßRII is the activation of Alk5, and signaling from solitary TßRIIs has not been reported.
It was earlier thought that Alk5 was the only type I receptor conducting TGF-ß signaling. However, there is evidence that TGF-ß can bind to Alk1 in endothelial cells and that the subsequent signal transduction through Smad1/5/8 inhibits Alk5 signaling (5, 16). Additionally, in epithelial cells Alk2 is involved in TGF-ß-induced epithelial-to-mesenchymal transdifferentiation (9). In this study, we have investigated the possibility of alternative receptors for TGF-ß signaling in fibroblasts, by stimulating Alk5-deficient murine embryonic fibroblasts (MEFs) with TGF-ß. We conclusively demonstrate that Alk5 is necessary for TGF-ß signal transduction in these cells.
Additionally, we present transcriptional profiles of MEF cell lines that are either deficient in Alk5 signaling, have normal signaling, or are stimulated with TGF-ß and identify genes that are differentially expressed as a response to TGF-ß signaling. Our results create a data set of 465 targets of TGF-ß/Alk5 signaling, in which earlier unknown categories of gene targets are revealed.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
In the cell expansion experiment, 2 x 105 cells from the different cell lines were plated in 60-mm plates and cultured with or without the presence of TGF-ß1 (10 ng/ml). Cells were counted and medium was changed at 24, 48, and 72 h postseeding. Three independent experiments were performed.
Western blot analysis.
Western blot analysis was performed as described (10).
Transcriptional reporter assays.
Assays were performed as described (10).
RNA isolation and microarray hybridization.
RNA from harvested MEFs, stored as a dry pellet at –80°C, was isolated using TRIzol reagent (Invitrogen, Stockholm, Sweden) according to manufacturer’s protocol, and reverse transcription and hybridization to microarray slides were performed as described (14). Microarray glass slides containing >37,000 mouse cDNA clones (mouse, 36K; National Institute of Mental Health, Bethesda, MD) were used. Slides were scanned using an Axon 4000 scanner (Axon Instruments, Leusden, the Netherlands) and analyzed with GenePix pro v.4.0 software (Axon Instruments).
Ten microarrays were performed to identify downstream targets of TGF-ß signaling. These were made using one RNA extraction from the KO1 MEF line and two extractions from the KO2 line (KO2 and KO2_2), with one of the Alk5+/+ cell lines (WT1) as reference, or the two Alk5+/+ cell lines (WT1 and WT2) stimulated with TGF-ß for 1, 5, and 16 h preharvest, again with the WT1 cell line as reference. Alk5–/– and TGF-ß-stimulated samples were labeled with Cy5 and reference sample with Cy3 dye in all experiments, except for one dye-swap experiment (KO1_DS) performed to exclude technical errors. The KO2_2 extraction functioned as a technical repeat and was performed 6 mo after the KO2 experiment. The subsequent microarray analysis was carried out using a completely new batch of reagents.
In addition, three arrays were performed to investigate the possibility of alternative receptors for TGF-ß signaling. They include experiments done using cDNA from the two Alk5–/– cell lines stimulated with TGF-ß for 8 h, with cDNA from unstimulated cells of the same Alk5–/– cell line as reference, and one experiment using cDNA from RNA extractions of two independently harvested WT1 cultures.
Microarray data, including information about the mouse cDNA clones, are available at the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) website (http://www.ncbi.nlm.nih.gov/geo/) with GEO accession no. GSE1742.
Analyses of microarray data.
The data extracted with the use of GenePix were stored in BioArray Software Environment (BASE; Ref. 19), and BASE was used for quality control and normalization. Briefly, on the basis of the parameters extracted by the GenePix software, our quality filter required each spot to have a median-based signal-to-noise ratio of at least 3 for each channel and at least 50 foreground pixels. Spots that did not fulfill these requirements were treated as missing values in the subsequent analysis. In addition, for the 10 arrays used to identify targets of TGF-ß signaling, spots with more than two missing values across arrays were excluded from further analysis. Each experiment was subsequently normalized using LOWESS regression as implemented in BASE.
Next, we looked for clones showing a large variation in expression across arrays, and selected spots for which the standard deviation of log (base 2) expression ratios was >0.5. To identify clones differentially expressed between TGF-ß-stimulated Alk5+/+ and Alk5–/– cells, we employed the rank-based Wilcoxon test as implemented in the MCD software (12). To investigate whether differentially expressed clones were significantly associated with specific Gene Ontology (GO) terms (6), we used the GoMiner software (24). GoMiner accepts gene symbols as input, and we used ACID (17) to map our clones to gene symbols (based on UniGene mouse build 141). In the GoMiner analysis, differentially expressed clones were compared with all clones that passed the quality filter. GO terms with P-under 
0.05 and at least five downregulated clones or P-over 0.05 and at least five upregulated clones were selected for discussion. To investigate genes in common between our data and previous studies, public data sets were mapped to gene symbols according to UniGene mouse build 141, using ACID.
Quantitative RT-PCR.
Quantitative (q)RT-PCR (TaqMan; Applied Biosystems, Stockholm, Sweden) was performed in an ABI Prism 7700 (Applied Biosystems), according to the manufacturer’s protocol, on cDNA from Alk5–/–, Alk5+/+, 5-h TGF-ß-stimulated Alk5+/+, and TGF-ß-stimulated Alk5–/– MEFs. RNA was reverse transcribed (SuperScript II; Invitrogen, Stockholm, Sweden), according to manufacturer’s protocol, in the presence of random hexamers, and gene-specific primers (Applied Biosystems) were used to analyze the expression of nine differentially expressed genes from the microarrays [inhibitor of DNA binding (Id)-1, Id2, Id3, growth arrest and DNA damage-inducible-45
(Gadd45g), eukaryotic translation initiation factor-5A (Eif5
), myelocytomatosis oncogene (c-myc), growth arrest specific (Gas5), activated leukocyte cell adhesion molecule (Alcam), and decorin (Dcn)] together with the housekeeping gene hypoxanthine guanine phosphoribosyl transferase (Hprt). Each assay was performed in triplicate, the results were normalized to Hprt levels, and relative gene expressions between Alk5–/– or TGF-ß1-stimulated Alk5+/+ and unstimulated Alk5+/+ cells were examined. Additionally, qRT-PCR analysis was performed on all MEF lines, using primers for Alk1–6 (Applied Biosystems), and on murine embryonic endothelial cells, using primers for Alk1. Experiments were performed in triplicate and normalized to Hprt levels.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
It has been shown previously that TGF-ß has an inhibitory effect on cell proliferation through blockage in the late phase of G1 (8). To examine whether TGF-ß-induced proliferation inhibition was altered in our MEF cell lines, we performed a cell expansion experiment. Each cell line was cultured in cell culture medium and serum with or without the addition of TGF-ß. The Alk5–/– cells proliferated equally well, irrespective of the presence of TGF-ß, whereas the cells from the Alk5+/+ animals showed an 
2.5-fold decrease in expansion after 72 h of stimulation (Fig. 1D). These results demonstrate the potent inhibitory effect of TGF-ß on cell proliferation and how this effect is dependent on Alk5.
Furthermore, we studied how the absence of Alk5 affected the transcriptional response of the fibroblasts. We examined the expression of the inhibitor of the DNA binding/differentiation family of genes, known to be transcriptionally repressed by TGF-ß signaling (11, 20). qRT-PCR reveals that these genes were not affected by TGF-ß in Alk5–/– MEFs, whereas in Alk5+/+ cells Id expression was strongly downregulated after TGF-ß stimulation (Fig. 2A).
|
Microarray experiments performed on TGF-ß-stimulated Alk5+/+ cell lines resulted in differential expression of between 2.6 and 3.9% of the clones printed, demonstrating the vast effect of TGF-ß on transcriptional regulation in the genome.
Global gene expression analysis reveals 465 targets of Alk5 signaling.
The substantial effects of TGF-ß made us conduct more extensive global gene expression studies to get a database of gene targets of Alk5 signaling. Total RNA was extracted from unstimulated Alk5–/– and Alk5+/+ MEFs stimulated with TGF-ß for three different time periods (1, 5, and 16 h). cDNA from each sample was hybridized on the microarray slides together with cDNA from an unstimulated Alk5+/+ common reference sample. We speculated that this design would result in four different data sets: early, intermediate, and late differentially expressed clones plus all the clones differentially expressed in the Alk5-deficient fibroblasts. A dye swap control experiment using one of the Alk5–/– lines (KO1) showed that a replicate of one experiment using the same RNA but different dyes for the sample and reference produced highly similar results (R2 = 0.96 and 0.97 for the KO1 and the reference, respectively; data not shown), thus demonstrating the reliability of our protocols. For the 10 arrays used in this investigation, 15,202 clones passed the quality filtering. Next, we performed hierarchical cluster analysis on all the clones (2,223) that had a standard deviation in expression values over the data set >0.5. This unsupervised analysis divided the microarray experiments in two clusters: one that included all the Alk5–/– hybridizations and another containing all the stimulated Alk5+/+ experiments (Fig. 3A). This was good evidence that the differences in gene expression among the samples were primarily due to the presence or absence of Alk5 and not because of other biological or technical variances. Because the 1-h TGF-ß-stimulated Alk5+/+ cell lines seemed to be highly related, whereas the similarities in gene expression patterns in the 5- and 16-h stimulations were more dependent on the cell line, we considered our data to contain differentially expressed clones representing early responsive genes (Alk5+/+ MEFs stimulated for 1 h), late responsive genes (Alk5+/+ MEFs stimulated for 5 and 16 h), and genes differentially expressed in fibroblasts deficient in TGF-ß signaling. Within this set of 2,223 clones with variation among the samples, we isolated the ones that were differentially expressed due to TGF-ß stimulation or Alk5 signaling deficiency and hence likely to be the targets of Alk5 signaling. By dividing the samples into the two groups (ALK5–/– and TGF-ß-stimulated Alk5+/+ samples) defined by the unsupervised hierarchical clustering, and performing a ranked-based Wilcoxon test on the varying clones, we were able to identify 465 clones (of which 445 could be mapped to a UniGene cluster, representing 369 unique genes) that were significantly (P 0.01) differentially expressed as a response to TGF-ß stimulation or Alk5 signaling deficiency (Supplemental Fig. S1; available at the Physiological Genomics web site).1
We expected 22 clones by chance to be differentially expressed using this P value for the 2,223 clones. Hierarchical clustering of these clones divided them in two groups, either upregulated as a response to TGF-ß stimulation and downregulated due to Alk5 deficiency, or the opposite (Fig. 3B). To get an overview of the 465 clones, they were categorized according to the biological or molecular function assigned to them in the GO database (Fig. 3C). An interesting observation was the finding of a large number of clones in the "protein folding/chaperone activity" category.
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Global gene expression profiling only detects transcriptional changes, and therefore posttranslational events could be overseen. For example, recently it has been shown that the TGF-ß superfamily type II receptor BMPR-II alone can interact with LIM kinase-1 (LIMK1), and that LIMK1 is released on BMP4 binding to BMPR-II (3). Unbound LIMK1 will then regulate actin dynamics through phosphorylation of cofilin. These kinds of posttranslational events might not be detected by gene expression analysis, although this is unlikely, since such changes in the cell are likely to affect global gene expression profiles to some degree. Thus, taken together, our data strongly suggest that TGF-ß signals exclusively through receptor complexes containing Alk5 in fibroblasts.
One microarray study on murine fibroblasts has previously been published (23) where the authors investigated the transcriptome of MEFs lacking Smad2, Smad3, or ERK signaling while having other TGF-ß signaling pathways intact. Here we have the advantage of Alk5–/– cells that completely lack responses to TGF-ß signaling and thus should reveal the majority of its gene targets. While in our study we identified 465 clones as targets of TGF-ß signaling, Yang et al. presented 360, and among these two data sets, 29 targets were overlapping. The 27 unique genes that these 29 clones represent include both known targets of TGF-ß signaling, like c-myc and Pdgfra, and some novel targets with interesting functions. In our study, Pdgfra is considerably downregulated in all the TGF-ß-stimulated WT MEFs and upregulated in three of four Alk5–/– experiments. Pdgfra is a possible indirect key mediator of the TGF-ß-induced proliferation block, since it stimulates proliferation of fibroblasts and has been shown to be downregulated by overexpression of TßRII in NIH-3T3 fibroblasts (4).
Another gene overlapping with the study by Yang et al. (23) is Bcl-2-associated athanogene-2 (Bag2), previously unknown as a target of TGF-ß signaling. Bag2 is involved in apoptosis and chaperone regulation and was recently reported to be induced by the p38 MAPK pathway (21). In accordance with this, Bag2 is downregulated in three of four of our Alk5–/– experiments while upregulated in all the TGF-ß-stimulated WTs. The finding of Bag2 is part of the interesting observation that two of the most significant differentially expressed categories, according to GoMiner analysis of our data, are the protein folding and chaperone categories. These include ATP-binding cassette, subfamily E (OABP), member 1 (Abce1), Bag2, heat shock protein A (Hspa9a), Hsp105, Hspe1, Hspd1, Hspa5bp1, chaperonin subunit 2 (Cct2), Cct3, Cct8, t-complex protein-1 (Tcp1), peptidylprolyl isomerase D (Ppid), and DNAJ homolog subfamily A, member 1 (Dnaja1). Also interesting is that the vast majority of these genes were upregulated as a response to TGF-ß stimulation or downregulated in the Alk5-deficient cells. Analysis after only 1 h of TGF-ß stimulation revealed a significant overrepresentation of the chaperone activity and protein folding genes. This significance was lost at later time points, suggesting that TGF-ß stimulation rapidly and transiently induces the expression of chaperones. Intriguingly, a large number of the other significantly overrepresented categories of target genes, identified by GoMiner analysis, are also involved in different aspects of protein processing. Among these, the most striking finding is the upregulation of genes involved in translation. Of the 71 genes involved in "translation regulator activity" spotted on the microarray, as many as nine were significantly upregulated in response to TGF-ß stimulation. Additionally, eight of these nine genes were subunits of eukaryotic initiating factors (eIFs), responsible for the first step in protein biosynthesis. These include Eif2, subunit 3, structural gene X-linked (Eif2s3x), Eif3s9, Eif2s1, Eif2s2, Eif5a, and Eif4g1. Furthermore, these categories exhibited kinetics similar to the protein folding and chaperone genes with rapid upregulation as a response to TGF-ß signaling.
Comparing our differentially expressed genes with identified gene targets from other studies performed on human lung fibroblasts (1) and human dermal fibroblasts (22), we obtained an overlap of 10%. Because of the differences in experimental design and species, extensive comparisons between our study and these two were not performed. However, Myc and connective tissue growth factor (Ctgf) were identified in both our study and the one performed by Yang et al. (23) as well as in the study performed on human lung fibroblasts, indicating these genes to be somewhat species-independent targets of TGF-ß signaling. Furthermore, it is interesting to note the immense downregulation of the Id genes by TGF-ß in our study. This is in contrast to findings in the study of human lung fibroblasts, where the ID genes were induced by TGF-ß (1). On the contrary, a microarray study of epithelial cells reported a repression of ID1, ID2, and ID3 expression as a response to TGF-ß, in agreement with our results (7). This downregulation of ID genes is implicated in the mechanism of proliferation arrest exercised by TGF-ß in these cells. Thus our findings showing an Id expression in MEFs similar to the expression in adult epithelial cells indicate that the downregulation of these genes is part of the proliferation inhibition by TGF-ß in MEFs.
To conclude, the use of Alk5–/– MEFs has given us the unique opportunity to investigate and exclude the use of alternative receptors for TGF-ß in fibroblasts. Additionally, the absence of alternative receptors makes this model ideal for the identification of target genes of TGF-ß signaling, which will be important for understanding the mechanisms behind the diverse effects of this multifunctional cytokine.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: M. Ringnér, Complex Systems Division, Dept. of Theoretical Physics, Lund Univ., Sölvegatan 14A, 223 62 Lund, Sweden (e-mail: markus{at}thep.lu.se).
doi:10.1152/physiolgenomics.00303.2004.
1 The Supplemental Material for this article (Supplemental Fig. S1) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00303.2004/DC1. 
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. S. Shao, L. Lin, Y. Yao, and K. I. Bostrom Expression of vascular endothelial growth factor is coordinately regulated by the activin-like kinase receptors 1 and 5 in endothelial cells Blood, September 3, 2009; 114(10): 2197 - 2206. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. A. Tone, H. Begley, M. Sharma, J. Murphy, B. Rosen, T. J. Brown, and P. A. Shaw Gene Expression Profiles of Luteal Phase Fallopian Tube Epithelium from BRCA Mutation Carriers Resemble High-Grade Serous Carcinoma Clin. Cancer Res., July 1, 2008; 14(13): 4067 - 4078. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Andratsch, E. Feifel, L. Taylor, M. O'Hayre, H. Schramek, N. P. Curthoys, and G. Gstraunthaler TGF-beta signaling and its effect on glutaminase expression in LLC-PK1-FBPase+ cells Am J Physiol Renal Physiol, September 1, 2007; 293(3): F846 - F853. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. V. Cohen, O. Kosti, and C. L. Stewart The nuclear envelope protein MAN1 regulates TGF{beta} signaling and vasculogenesis in the embryonic yolk sac Development, April 1, 2007; 134(7): 1385 - 1395. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Gauldie, M. Kolb, K. Ask, G. Martin, P. Bonniaud, and D. Warburton Smad3 Signaling Involved in Pulmonary Fibrosis and Emphysema Proceedings of the ATS, November 1, 2006; 3(8): 696 - 702. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
