Introduction 1 2 3 4 5 3 5 Materials and methods Cell culture and treatment NMuMG cells were cultured in Dulbecco's modified Eagle's medium containing 10% serum and 10 μg/ml insulin, until reaching 70–80% confluency. The cells were treated with 4 ng/ml TGF-β1 for 1, 6 or 24 hours. Untreated cells were used as reference samples at each time point, and were referred to as controls in the analyses. Four independent replicates of each experiment were performed. Cell morphology was examined under phase contrast using an Olympus CK40 microscope (Melville, NY, USA). Growth inhibition assay 3 3 3 3 RNA sample preparation NMuMG cells were lysed and homogenized in Trizol reagent (Invitrogen, Carlsbad, CA, USA), and total RNA was prepared according to the manufacturer's instructions. Two rounds of extraction were performed to isolate pure, high-quality total RNA. Thirty micrograms of total RNA from each sample were used for reverse transcription and cyanine (Cy)-dye incorporation. cDNA microarray hybridization  The reference RNA samples from nontreated NMuMG cells were labeled with Cy5, and the test RNA samples from NMuMG cells treated with 4 ng/ml TGF-β for 1, 6, and 24 hours were labeled with Cy3. The Cy3-labeled and Cy5-labeled samples were hybridized simultaneously to the same array. Four independent replicates of TGF-β treatment, RNA isolation and labeling, and microarray hybridization were performed. To estimate 'system noise' that may be due to differences in Cy-dye labeling between samples, we performed a 'self-to-self' hybridization, where the same control sample was labeled separately with Cy3 and Cy5, and then hybridized. Oligo dT for reverse transcription was synthesized by Invitrogen. SuperScript II reverse transcriptase (Invitrogen) was used. A Qiagen PCR purification kit was used for purification of probes (Qiagen, Valencia, CA, USA). Slides were prehybridized to eliminate nonspecific interactions in prehybridization solution: 1% BSA, 5 × sodium chloride/sodium citrate buffer (SSC), 0.1% SDS at 65°C for 45 min. Hybridizations were performed at 65°C for 14–16 hours in a humidified hybridization chamber (Corning, Acton, MA, USA). After hybridization, the slides were washed once each in solutions 1–3 (wash solution 1, 2 × SSC, 0.1% SDS; wash solution 2, 1 × SSC; wash solution 3, 0.1 × SSC) for 5 min at 55°C with gentle stirring/agitation. Washed slides were centrifuged in a conical tube for 5 min at 1600 rpm to dry them. Microarray slides were scanned using a Genepix 4000 B scanner (Axon Instruments, Union City, CA, USA) at a resolution of 10 μm, and the original data files were generated by GenePixPro software (version 4.0; Axon Instruments). Data analysis A merged 15,000 data file was constructed by combining the three original 5000 data files generated from the three contiguous slides on which the total 15,000 gene clones were printed. The GenBank accession number and the full annotation gene name for each gene were incorporated. Data analyses were performed using the GeneSpring 4.1.5 software package (Silicon Genetics, Redwood City, CA, USA). The normalization for the two-color cDNA microarray data was designed as follows. The net signal intensity in each channel (Cy3 or Cy5) was determined by subtracting the local background from signal intensity values. Each gene's measured intensity was divided by its control channel value (the reference RNA sample channel of Cy5). A Cy3/Cy5 ratio represents the relative abundance of a target transcript in TGF-β-treated and nontreated samples respectively. When the control channel value was below 1000.0, the data point was considered too weak and was discarded. Each sample was normalized to the 50th percentile of all measurements. The bottom 10th percentile was considered background, and was subtracted from all the other values. t  P  t  P  Northern hybridization Not Sal Escherichia coli  32 For northern blot analysis, 10 μg total RNA was separated by electrophoresis on a 1% agarose, 0.66 M formaldehyde gel and transferred to Hybrid Nylon transfer membranes (Amersham Biosciences, Piscataway, NJ, USA). After transfer, the membrane was UV cross-linked (Stratalinker; Stratagene, La Jolla, CA, USA). All cDNA probes were hybridized overnight at 42°C in ULTRAhyb solution (Ambion, Austin, TX, USA). Each filter was washed twice for 5 min in 2 × SSC and 0.1% SDS at 42°C, followed by two 15 min washes each in 0.5 × SSC and 0.1% SDS at 42°C, followed by two 15 min washes in 0.1 × SSC, 0.1% SDS at 42°C. Equal loading of gel lanes was confirmed by hybridization with the house keeping gene 1B15. Results were visualized and quantitated using a FUJIFILM-FLA-5000 Phosphoimager (Fuji Photo Film Co. Ltd, Stamford, CT, USA). Results Biological and gene expression responses of NMuMG cells to TGF-β 1 2 3 1  Functional class distribution of the novel and known TGF-β1-regulated genes  1 2 Northern analysis verification of microarray data 4 6 4 7 8  1 5 Examination of cell cycle control and cell adhesion-related genes by microarray 2 myc  2 myc 2 5 6 Our results are consistent with previous studies, but we identified several novel TGF-β-regulated genes that could play important roles in both EMT and cell cycle regulation, including RhoB, mCalpain, actinin 3, IQGAP1, Ikki and protein phosphatase 2A (PP2A)-PR53. Discussion 5 4 myc  9 10 5 1 11 8 12 13 14 15 myc  1  1  myc  myc  1 Myc  9 2 myc 2 1  16 17 2+ 18 Our results also demonstrate that TGF-β regulates multiple genes involved in regulation of the actin cytoskeleton. Some of these components include actinin α3, the Arp2–Arp3 complex, and myosin light chain 2. Four different isoforms of tropomyosin were upregulated. Actinin α3 is a calcium-dependent cytoskeletal protein with an actin-binding domain. Actinin α3 is associated with adherens junctions and desmosomes, together with E-cadherin, α-catenin, β-catenin and γ-catenin, vinculin, α-actinin, and polymerized actin. The genes α-actinin, myosin light chain 1/myosin light chain 2, and tropomyosin are components of myofilaments involved in cell contractility and motility. 19 20 21 Altogether, our results are consistent with the important role of TGF-β in regulating focal adhesions, integrin-based adhesion, actin cytoskeletal architecture, and cell motility. TGF-β plays a profound role in the dedifferentiation of epithelial cells, causing depolarization, disruption of epithelial interactions, altered expression of extracellular matrix proteins, rearrangement of the cytoskeleton, and formation of actin stress fibers. These TGF-β responses provide cells with increased metastatic potential and with increased motility. Over the past few years, tremendous progress has been made in identifying signal transduction pathways activated by TGF-β family members. In the present study, we have combined microarray analysis with the examination of specific signaling pathways. Because of the important role that TGF-β plays in controlling cell proliferation and cell adhesion, we used the signaling elements involved in these processes as a template for the analysis of our microarray results. Many of the genes that we identified as being regulated by TGF-β were previously reported to be downstream targets of TGF-β. We also discovered several novel TGF-β target genes known to play roles in regulating cell adhesion, and thus EMT, and also cell cycle regulation. Examining the biological functions of these novel TGF-β target genes will increase our knowledge of the mechanisms by which TGF-β mediates its cellular effects. Furthermore, our results indicate several possible points of convergence between TGF-β signaling and other major intracellular and intercellular signaling systems. Conclusion A 15,247 cDNA microarray was used to examine TGF-β-regulated gene expression in mouse mammary NMuMG cells. Expression of 10% of the genes examined (939) was altered by TGF-β treatment. We have reported a comprehensive analysis of the coordinated regulation of genetic programs induced by TGF-β in mammary epithelial cells during the processes of cell cycle arrest and EMT. In addition, several genes previously not known to be regulated by TGF-β at the transcriptional level were identified, including Akt, RhoB, IQGAP1, mCalpain, actinin α3, Ikki and PP2A-PR53. Competing interests None declared. Abbreviations BSA = bovine serum albumin; Cdk = cyclin-dependent kinase; Cy = cyanine; EMT = epithelial to mesenchymal transition; IKKi = inducible IkB kinase; IQGAP1 = IQ motif-containing the GTPase-activating protein 1; NF = nuclear factor; NIA = National Institutes of Aging; PCR = polymerase chain reaction; PP2A = protein phosphatase 2A; RT = reverse transcriptase; SSC = sodium chloride/sodium citrate (buffer); TGF-β = transforming growth factor beta.