All subjects signed consent forms approved by the Committee on Human Research at UCSF. Systemic patients met ACR classification criteria for systemic sclerosis (13). They were further characterized as the diffuse subset (14). All localized patients met clinical and histologic criteria for morphea (15). Three 5-mm punch biopsies from the lateral forearm, 8 cm proximal to the ulnar styoid were taken for clinically involved skin and three biopsies were taken from the upper outer quadrant of the buttock or flank for clinically uninvolved skin. Four systemic sclerosis patients (two men and two women) with diffuse scleroderma underwent 2 sets of biopsies (Table 1). Four normal controls (one man and three women) underwent the same number of biopsies in the identical locations (forearms and backs). Two biopsies were immediately frozen and stored in liquid nitrogen until RNA extraction. The third biopsy was bisected; half went into 10% formalin for routine histology processing and the second half for fibroblast cell culture (16).

Each set of frozen biopsies was pulverized using a Bio-Pulverizer (Biospec Products, Bartlesville OK) pre-chilled in liquid nitrogen. Frozen tissue powder was quick-thawed directly into lysis buffer from Promega's (Madison, WI) SV Total RNA Isolation System and total RNA prepared. Typically 10 to 25 mg of total RNA was obtained from two 5 mm punch biopsies. cRNA was synthesized by standard protocols and hybridized to Affymetrix Hu95A microrarrays containing 12,000 different elements as described in the supplemental material.

Cells grown in culture were analyzed on cDNA microarrays containing 40,512 elements representing 28,384 UniGene clusters (UniGene Build No. 155, released 9-28-2002) manufactured in the Stanford Microarray Facility (http://www.microarray.org). Hybridization and analysis of these data are described in the supplemental materials and methods. Detailed protocols for cDNA synthesis, cDNA microarray hybridization and data processing have been previously described (6).

Human cell lines
Aortic smooth muscle cells and microvascular endothelial cells were obtained from Clonetics and cultured in the recommended media; Hs578T myofibroblast-like cells were grown in RPMI-1640 supplemented with phenol red, glutamine (2 mM) and 5% fetal calf serum (7); HeLa S3 epithelial cells were obtained from the American Type Culture Collection (ATCC) and cultured as described previously (6). Normal, morphea and scleroderma fibroblasts were derived from normal, morphea and scleroderma skin biopsies as described previously (16).

cRNA synthesis and Affymetrix Microarray Hybridization.
Double-stranded cDNA was synthesized with superscript reverse transcriptase (Gibco Life Technologies) at 42oC from 10ug total RNA using a T7-(dT)24 primer containing a T7 RNA polymerase promoter. Biotinylated cRNA was synthesized using the Enzo BioArray HighYield RNA Transcript Labeling kit (ENZO Diagnostics, Farmingdale, NY) yielding 20-40 mg of cRNA. Labeled cRNA was fragmented to approximately 50-100 nucleotides and hybridized to Affymetrix Hu95A microarrays for 16 hrs at 45oC in the UCSF GCRC Core Laboratory. Microarrays were washed using an Affymetrix fluidics station and stained initially with streptavidin-phycoerytherin (10 mg/ml: Molecular Probes). For each sample the signal was further enhanced by incubation with biotinylated goat anti-streptavidin (3 mg/mL; Vector Laboratories) followed by a second incubation with streptavidin-phycoerytherin (10 mg/ml) and a second round of intensities were measured.

Image segmentation, background corrections, average difference intensities and absent, present and marginal calls were generated by GeneSpring Software (Silicon Genetics, Redwood City, CA). Only those genes for which at least 25% of samples provided adequate data were used for further analysis; genes expressed at detectable levels in fewer than 25% of the samples were excluded. Arrays were scaled such that each had the same overall intensity. Each absolute expression value in a given sample was converted to a ratio by dividing by its average expression value across all samples as previously described (5).

cDNA synthesis, cDNA microarray hybridization and data processing
Poly(A) RNA from each cell line was reverse-transcribed into Cy5-dUTP (Amersham Pharmacia Biotech, Piscataway NJ) labeled cDNA and reference RNA reverse-transcribed into Cy3-dUTP (Amersham Pharmacia Biotech, Piscataway NJ) labeled cDNA using standard methods (29, 30). Cells grown in culture were analyzed on cDNA microarrays containing 40,512 elements representing 28,384 UniGene clusters (UniGene Build No. 155, released 9-28-2002) manufactured in the Stanford Microarray Facility (http://www.microarray.org). Equal amounts of Cy5- and Cy3-labeled cDNA were hybridized to spotted cDNA microarrays and scanned using a GenePix 4000A Scanner (Axon Instruments, Union City CA). Detailed protocols are available at http://brownlab.stanford.edu/protocols.html.

Data were extracted by superimposing a grid over each array using GenePix 3.0 software (Axon Instruments, Union City CA). Spots of poor quality, determined by visual inspection, were excluded from further analysis. Data collected for each array were stored in the Stanford Microarray Database (SMD) and are available from SMD at http://genome-www.stanford.edu/microarray/ (31). Only array elements with signal intensity at least 2-fold above background in either the Cy-5 and Cy-3 channels and with a regression correlation (R2) greater than 0.4 between the Cy-5 and Cy-3 fluorescent signals of individual pixels from a given DNA spot, were analyzed further. Data points that did not meet these criteria are blank in the primary data tables. Log2 (Cy5/Cy3) was retrieved for each data point and used for all analysis, where (Cy5/Cy3) is the normalized ratio of the background-corrected intensities, as defined in SMD (31).

Data integration In order to compare data generated with Affymetrix microarrays to data generated with Stanford cDNA microarrays, the data for each perfect match-mismatch oligo set (average difference) and the data for each clone on our cDNA microarrays in the same UniGene cluster were averaged to produce a single expression profile for each; since UniGene is dynamic, each gene has been identified by its representative accession number taken from LocusLink at the time the data were merged. 2149 genes (represented by unique UniGene clusters) were selected from the skin biopsy data that varied more than two-fold on at least four arrays. Of the 2149 genes differentially expressed in skin biopsies, 1123 were also measured in our study of cell lines. Expression data for these genes were selected from the cell line experiments and merged with data for the same genes from the skin biopsy samples. Gene expression experiments for the cell lines and skin biopsy samples were centered independently on the corresponding group-wise median values and the expression patterns for each set of samples was hierarchically clustered separately based on expression of these 1123 genes.

Selection of differentially expressed genes To identify genes differentially expressed in 14 systemic sclerosis samples as compared to the 13 normal skin samples, we used the rank sum test (Wilcoxon) with p-value cutoff of 0.1 (with rounding) (20). This allowed us to identify genes more highly expressed in either group without assumptions about the background distribution of expression values. All p-values were corrected for multiple testing with the Bonferroni correction.