1286 Articles | JNCI Vol. 102, Issue 16 | August 18, 2010 available online). The pLXSN SATB1 vector and pLXSN empty vector control were transfected into the PT67 packaging cell line by use of FuGene (Roche). Stable cell lines expressing pLXSN constructs were selected by incubation with G418 at 1 mg/mL. Virus-containing supernatants from PT67 cells were collected after 48 ��� 96 hours of incubation, and then, 70% confluent cell cultures were infected with virus by adding the supernatant in the presence of polybrene (Sigma) at 8 �� g/mL, and the cultures were incubated overnight. Forty-eight hours later, SKBR3 cells con- taining integrated pLXSN constructs were selected for 5 days in G418 at 1.5 mg/mL, NIH3T3 cells containing integrated pLXSN constructs were selected in G418 at 0.7 mg/mL, and MCF7 and HME3 cells containing integrated pLXSN constructs were se- lected in G418 at 0.5 mg/mL. Pooled populations of transduced SKBR3, NIH3T3, MCF7, and HME3 cells that were obtained after 5 days of G418 selection without subcloning were used for experiments in cell culture and in mouse models. Analysis of SATB1 mRNA and Protein Expression Protein and RNA samples were extracted from subconfluent cul- tures of MDA-MB-231 and BT549 parental cells and correspond- ing cells expressing vector control or SATB1 shRNA, as well as SKBR3, MCF7, NIH3T3, and HME3 parental cells and corre- sponding cells expressing vector control or pLXSN SATB1 in the exponential phase of growth. Analysis of mRNA Expression by Quantitative Real-Time Reverse Transcription ��� Polymerase Chain Reaction (qRT- PCR). Total RNA was purified by use of the TRIzol reagent (Invitrogen) according to the manufacturer���s instructions. A 1- �� g RNA sample was reverse transcribed by use of the High Capacity RNA to cDNA Kit (Applied Biosystems, Foster City, CA) fol- lowing the manufacturer���s directions. qRT-PCR was performed with 100 ng of cDNA from each sample. SYBR green 2X master mixture (Roche) was used in a total volume of 20 �� L. Primers for target genes were as follows: SATB1 sense 5 ��� -TGCAAAGG- TTGCAGCAACCAAAAGC-3 ��� and SATB1 antisense 5 ��� - AACATGGATAATGTGGGGCGGCCT-3 ��� and SATB2 sense 5 ��� -ATGTGAGCATGGTCTCCTCG-3 ��� and SATB2 antisense 5 ��� -GCGCCGTCCACCTTAATAG-3 ��� . GAPDH was used as the endogenous control, with GAPDH primers as follows: sense 5 ��� -TGTTGCCATCAATGACCCCTT-3 ��� and antisense 5 ��� - CTCCACGACGTACTCAGCG-3 ��� . Reaction conditions were 5 minutes at 95��C, followed by 45 cycles of 95��C for 15 seconds, 60��C for 15 seconds, and 72��C for 15 seconds. Each sample was analyzed in triplicate by use of a LightCycler 640 Real-Time PCR System (Roche). Normalized target gene expression was calculated relative to expression of GAPDH endogenous control and adjusted relative to expression in parental cells. Immunoblot Analysis. SATB1 protein expression was assessed by immunoblot analysis in cell lysates (40 ��� 60 �� g of protein in lysis buffer [20 mM HEPES at pH 7.9, 25% glycerol, 0.5 N NaCl, 1 mM EDTA, 1% Nonidet P-40, 0.5 mM dithiothreitol, and 0.1% deoxycholate]) containing protease inhibitors (Roche). Proteins were separated by sodium dodecyl sulfate ��� polyacrylamide gel elec- trophoresis and transferred onto nitrocellulose membranes (BioRad, Hercules, CA). The membranes were incubated for 1 hour in blocking buffer (Tris-buffered saline with 0.1% Tween [TBS-T] and 5% nonfat dry milk) and incubated overnight at 4��C with anti- SATB1 mouse monoclonal antibody (BD Biosciences, San Jose, CA) at a dilution of 1:1000 in blocking buffer. After washing in TBS-T, the blots were incubated with horseradish peroxidase ��� conjugated secondary antibody against mouse IgG, and the signals were visualized by the enhanced chemiluminescence system as described by the manufacturer (Amersham Pharmacia, GE Healthcare Biosciences, Piscataway, NJ ). The blots were reprobed with anti-actin monoclonal antibody (Abcam, Cambridge, MA) to confirm equal loading of the different samples. Microarray Analysis of SATB1 mRNA Expression MCF7 cells expressing pLXSN SATB1 or pLXSN control vector were used for this experiment. Total RNA samples were prepared according to the manufacturer���s instructions (Ilumina, Inc, San Diego, CA) and analyzed on the HumanWG-6 Illumina platform. Raw expression data were quantile normalized (which forces the arrays to have absolutely identical distributions on the basis of the assumption that the RNA populations that hybridized to the arrays are the same) and log 2 transformed before analysis with the beadar- ray package in the R statistical language ( 15 , 16 ). Only transcripts with detection P values of less than.1 were included in further analyses. Data were analyzed by use of a single factorial design with treatment (ie, pLXSN control vector) as the factor and subse- quent single treatment contrasts (ie, pLXSN SATB1), as imple- mented in the limma package in the R language ( 17 ). An Empirical Bayes analysis of the contrasts (ie, pLXSN control vector vs pLXSN SATB1) provided estimates of fold change and adjusted P values of differential expression between treatments. Soft Agar Assay MDA-MB-231 and BT549 parental cells and corresponding cells expressing vector control or SATB1 shRNA, as well as SKBR3, MCF7, NIH3T3, and HME3 parental cells and corresponding cells expressing vector control or pLXSN SATB1, were used for this experiment. Anchorage-independent growth was assessed with a high-throughput soft agar colony formation assay (CytoSelect 96-Well Cell Transformation Assay Cell Biolabs, Inc, San Diego, CA) according to the manufacturer���s instructions. Briefly, 5 �� 10 3 cells were suspended in DMEM containing 10% fetal bovine serum with 0.4% agarose and layered on top of 0.6% agarose in DMEM in 96-well plates. Cultures were maintained for 8 days. Colony formation was measured by agar solubilization followed by cell lysis and quantification of cell number by use of CyQuant GR Dye in a fluorescence plate reader. Data were reported as fluores- cence intensity, which is directly proportional to cell number. Morphology on Matrigel MDA-MB-231 and BT549 parental cells and corresponding cells expressing vector control or SATB1 shRNA, as well as SKBR3 and MCF7 parental cells and corresponding cells expressing vector control or pLXSN SATB1, were used for this experiment. MCF10A cells were used as a control for acinar formation by nor- mal mammary epithelial cells. Invasive protrusions from acinar demonstrate aggressiveness. Cell morphology on matrigel was at Resources Manager on September 10, 2010 jnci.oxfordjournals.org Downloaded from

jnci.oxfordjournals.org JNCI | Articles 1287 studied by plating 2.5 �� 10 4 cells onto 24-well plates that had been coated with 100 �� L of matrigel (BD Biosciences). Cultures were maintained for 5 days and imaged by use of a phase ��� contrast microscope that was equipped with an Olympus digital camera. Each experiment was performed twice, with triplicate samples. Wound-Healing Assay MDA-MB-231 and BT549 parental cells and corresponding cells expressing vector control or SATB1 shRNA, as well as SKBR3 and MCF7 parental cells and corresponding cells expressing vector control or pLXSN SATB1, were used for this experiment. Cell migration was assessed in a wound-healing (scratch) assay. Cells were plated in 24-well plates and allowed to proliferate to form a confluent monolayer. A 200- �� L pipette tip was used to scratch a single wound through the middle of the cell monolayer. Cells were imaged as indicated until the wound was fully closed by use of a light microscope equipped with an Olympus digital camera. Images that were matched by the number of days after wounding were compared visually to determine differences in the rate of cell migration. Each experiment was performed twice, with triplicate samples. Analysis of Tumor Growth in Mouse Mammary Fat Pads and Intravasation We randomly assigned female NCR athymic mice to one of 11 groups (six mice per group) as follows: group 1 = a nonsubcloned pool of SATB1 shRNA 1 MDA-MB-231 cells group 2 = a non- subcloned pool of SATB1 shRNA 2 MDA-MB-231 cells group 3 = a nonsubcloned pool of SATB1 shRNA 3 MDA-MB-231 cells group 4 = control shRNA MDA-MB-231 cells group 5 = parental MDA-MB-231 cells group 6 = parental SKBR3 cells group 7 = SKBR3 empty vector control cells group 8 = SATB1- overexpressing SKBR3 cells group 9 = parental MCF7 cells group 10 = MCF7 empty vector control cells and group 11 = SATB1-overexpressing MCF7 cells. Before injection, 2 �� 10 5 cells were suspended in 200 �� L of matrigel (5 mg/mL) in phosphate- buffered saline for cell lines except MCF7 cells, which were sus- pended at 3 �� 10 6 cells. Cells were injected into the fourth mammary fat pad from flank in all mice to generate one tumor per mouse. Tumor growth was monitored externally biweekly by use of vernier calipers for 6 ��� 12 weeks. Tumor volume was calculated with the formula L �� W 2 �� 0.4 (where 1 cm 3 = 1 g). Tumors were removed at necropsy and weighed to determine final tumor weight. To determine intravasation of MDA-MB-231 tumor cells, lung tissue (one half of a lung) and 100 �� L of blood were collected from each mouse injected with cancer cells immediately after they were euthanized by ketamine ��� xylazine injection, followed by cer- vical dislocation. Each lung or blood sample was incubated with 0.2% collagenase type 2 in DMEM for 2 hours at 37��C, and cells were dispersed, washed, and plated in DMEM with 10% fetal bovine serum plus puromycin (0.7 �� g/mL) to select tumor cells containing integrated pGIPZ shRNA constructs. Tumor colonies were counted 4 weeks later. Analysis of Metastasis in a Mouse Model We randomly assigned female NCR athymic mice to one of five groups (six mice per group) as follows: group 1 = a pool of SATB1 shRNA 1 MDA-MB-231 cells group 2 = a pool of SATB1 shRNA 2 MDA-MB-231 cells group 3 = a pool of SATB1 shRNA 3 MDA-MB-231 cells group 4 = control shRNA MDA-MB-231 cells and group 5 = parental MDA-MB-231 cells. We injected 1 �� 10 6 cells intravenously via the lateral tail vein in 100 �� L of phosphate- buffered saline. At 10 weeks after injection, all mice were eutha- nized by ketamine ��� xylazine injection, followed by cervical dislocation, and their lungs were removed and fixed in 10% for- malin. The number of surface metastases per lung was determined under a dissecting microscope. Histopathological analysis was per- formed at the Histology Research Core Facility of the Sylvester Comprehensive Cancer Center (University of Miami). Analysis of Oncomine Data For SATB1 expression in cell lines, the Oncomine Research data- base ( http :// www . oncomine . org / ) was searched for gene expression studies involving breast cancer cell lines. We retained any studies (n = 3) with at least 20 different cell lines for which expression of SATB1 was available for further analysis. The processed gene ex- pression data from these studies was obtained from either the Gene Expression Omnibus database ( www . ncbi . nlm . nih . gov / geo / ) or the ArrayExpress repository ( www . ebi . ac . uk / microarray - as / ae / ). Each dataset was analyzed in the R statistical language ( 16 ) by use of an analysis of variance for SATB1 expression with breast cancer subtype as the only factor (luminal, basal A, or basal B) to deter- mine the statistical significance of the association of subtype with SATB1 expression. For analysis of SATB1 expression in patient data, the Oncomine Research database ( http :// www . oncomine . org / ) was searched for gene expression studies involving breast cancer patients for whom overall survival information was available. Those studies (n = 6) with at least 100 patients for which expression of SATB1 was avail- able were retained for further analysis (for details, see Table 1 ). These six studies included a total of 1170 patients. The processed gene expression data from these studies was obtained from either the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov/ geo/) or the ArrayExpress repository (www.ebi.ac.uk/microar- ray-as/ae/). Each dataset was analyzed in the R statistical language by use of the pamr and survival packages ( 24 , 25 ). This analysis involved an initial grouping of patients according to their level of SATB1 expression (fi rst, second, or third tertile). We visually con- fi rmed proportionality using R���s plot function of survival function vs survival time. A survival analysis with Cox proportional hazards models was then conducted to compare the survival between groups (ie, survival time was examined as a function of level of SATB1). Kaplan ��� Meier plots were made of the observed survival times for each group. A forest plot of the odds ratios of death (in which the middle and top tertiles of SATB1 expression vs the lowest tertile of expression were compared) was constructed with the metaplot function from the rmeta package for R ( 26 ). To determine whether there was cross-hybridization between SATB1 probes with SATB2 mRNA, we checked the probes for SATB1 on the Affymetrix platforms for potential cross-hybridization with SATB2 ( Supplementary Tables 2 and 3 , available online). Using BLAST searches (www.ncbi.nlm.nih.gov/BLAST/), we determined that there was little if any overlap between the SATB1 probe sequences and SATB2 mRNA sequence. at Resources Manager on September 10, 2010 jnci.oxfordjournals.org Downloaded from