|
|
Blood, Vol. 109, Issue 9, 3972-3981, May 1, 2007
Pre-TCR expression cooperates with TEL-JAK2 to transform immature thymocytes and induce T-cell leukemia
Blood dos Santos et al.
109: 3972
Supplemental materials for: dos Santos et al, Vol 109, Issue 9, 3972-3981
Proliferation assays
For T-cell proliferation assays, thymocytes were seeded in triplicate in 96-well plates precoated with 1 µg/mL or 10 µg/mL anti-CD3 antibody (145-2C11; BD Biosciences, San Jose, CA). Alternatively, the cells were stimulated by 10 ng/mL PMA (Sigma-Aldrich, St Louis, MO), 250 ng/mL ionomycin (Sigma-Aldrich), or both. After a 48-hour incubation at 37°C, proliferation was quantified by 6-3H-thymidine incorporation (for 16 hours; 0.1 µCi (0.0037 MBq)/well GE Healthcare, Chalfont St Giles, United Kingdom). Cell lysates were harvested and spotted on glass fiber filters, washed, and counted in OptiPhase HiSafe 2 (Perkin Elmer, Courtaboeuf, France). Oligonucleotide array analysis
Unsupervised probe set selection and classification
We first filtered out probe sets that had a mean log2 intensity value across all 25 samples less than 5.6 resulting in 7065 probe sets. We used the following steps:
(1) Unsupervised probe set selection based on the two following criteria: (a) a P of a variance test (see below) less than .01, and (b) a “robust” coefficient of variation (rCV) less than 10 and superior to a given rCV percentile. Eight rCV percentile thresholds were used (60%, 70%, 80%, 90%, 95%, 97.5%, 99%, 99.5%), yielding 8 lists of probe sets.
(a) Variance test: For each probe set (P) we tested whether its variance across samples was different from the median of the variances of the 7065 probe sets. The statistic used was ((n−1) × Var(P)/Varmed), where n refers to the number of samples. This statistic was compared to a percentile of the  2 distribution with (n−1) degrees of freedom and yielded a P value for each probe set.
(b) Robust coefficient of variation: rCV for each probe set was calculated by dividing the standard deviation by the mean, eliminating the highest and lowest expression value across the samples for each probe set.
(2) Hierarchical clustering of the 25 samples, using data restricted to each of 8 probe sets lists, for 3 different linkage methods (average, complete and Ward), using 1-Pearson correlation as a distance metric (package cluster V1.9.3) produced 24 dendrograms.
(3) Stability assessment: The intrinsic stability of each of the 24 dendrograms was assessed by comparing each dendrogram to cluster results (dendrograms) obtained after data “perturbation” or “resampling” (100 iterations for each). This comparison yielded a reproducibility score (see below). Here, “perturbation” refers to the addition of random Gaussian noise (µ = 0,  = 1.5 × median variance calculated from the data set) to the data matrix, and “resampling” refers to random substitution of 5% of the samples by virtual samples. Virtual samples were generated by random crossing of existing profiles. To verify that the obtained partition did not depend on the number of probe sets nor of the clustering method, we calculated similar reproducibility scores based on partitions derived from a decreasing number of probe sets and different linkage methods. Reproducibility score: To compare 2 dendrograms, both were cut into k clusters (k = 2.5) yielding a “partition” for each dendrogram. We then calculated a reproducibility score by comparing the 2 partitions for each k. To compare 2 partitions, we calculated the proportion of pairs of samples that are in the same cluster retained in both partitions. The overall stability of the classification was assessed by calculating a mean reproducibility score (for different values of k), using all pairs of dendrograms obtained for a given clustering method with the 8 probe sets lists.
(4) Choice of a representative dendrogram: To choose a representative dendrogram of the series of 24 dendrograms, we identified “consensus” clusters of samples, and kept the dendrogram the most related to these “consensus” clusters. We obtained 100% conservation of samples using k = 3 partition and the Ward linkage across all 8 gene lists (Table S1). Supervised analysis
Univariate t tests (BRB ArrayTools v3.4.beta2, http://linus.nci.nih.gov/BRB-ArrayTools.html) were used to define the differential expressed gene lists with a significance level of each univariate test of P < 0.01 using a random variance model.1 False discovery rates (FDRs) were calculated based on 1000 sample permutations. Fisher exact tests were performed to measure the enrichment of the genes on chromosome 15 in the list of significantly overexpressed 274 genes in TEL-JAK2/Rag2–/– leukemic cells. Mouse array CGH
The genome-wide CIT M3 Mus musculus 1K BAC CGH array was built in the laboratory. This array contains 1081 BACs that were isolated, validated by end-sequencing, and amplified by multiple displacement amplification. The collection was spotted in quadruplicate, in 75% formamide, on Corning Ultragaps slides (Corning, NY). After rehydration and UV fixation, the slides were blocked for 15 minutes with succinic anhydride/1-methyl 2-pyrrolidinone/sodium borate (pH 8.0) solution. The slides were denatured for 2 minutes in boiling water, dehydrated in 95% ethanol, and dried by centrifugation at 100× g. After DpnII digestion (New England Biolabs, Ipswich, MA) and purification with PCR Purification Kit (Qiagen, Courtaboeuf, France), 1.5 µg tumor DNA and 1.5 µg male normal C57BL/6 DNA were, respectively, labeled with dCTP-Cy5 or dCTP-Cy3 (GE Healthcare, Chalfont St Giles, United Kingdom) by random priming (BioPrime DNA Labeling System Kit, Invitrogen, Carlsbad, CA). The reaction was performed at 37°C for 3 hours. After purification on microcon YM30 (Millipore, Billerica, MA) and denaturation at 92°C, these DNAs were prehybridized with 120 µg mouse Cot-1 DNA (Roche, Meylan, France) for 30 minutes at 37°C in 20µL hybridization buffer (50% formamide, 40 mM NaH2PO4, 0.1% SDS, 10% dextran sulfate, 2× SSC, Denhardt solution) and laid on treated microarray slides. After 24 hours of hybridization at 37°C, the slides were washed for 5 minutes in 0.5× SSC, 0.03% SDS at 65°C, and for 5 minutes in 0.5× SSC, 0.03% SDS at 45°C. After drying by centrifugation, slides were detected with GenePix 4000B scanner (Molecular Devices, Union City, CA) and analyzed with GenePix Pro 5.1 software. Normalization was done by the MANOR routine2 and data were visualized on VAMP interface.3 Quantitative RT-PCR primer sequences
Hprt: 5′-GCTGGTGAAAAGGACCTCT-3′ and 5′-CACAGGACTAGAACACCTGC-3′
Ctsw: 5′-ACCGTGACCATCAACATGAA-3′ and 5′-GTCTGCATGCCCTCTTTCTC 3′
Prf1: 5′-CGGTGTCGTGTGGAACAATA-3′ and 5′-GTCATCATCCCAGCCGTAGT 3′
Runx3: 5′-AGCACCACGAGCCACTTC-3′ and 5′-GTAGGGAAGGAGCGGTCAA 3′ REFERENCES 1. Wright GW, Simon RM. A random variance model for detection of differential gene expression in small microarray experiments. Bioinformatics. 2003;19:2448-2455.
2. Neuvial P, Hupe P, Brito I, et al. Spatial normalization of array-CGH data. BMC Bioinformatics. 2006;7:264.
3. La Rosa P, Viara E, Hupe P, et al. VAMP: visualization and analysis of array-CGH, transcriptome and other molecular profiles. Bioinformatics. 2006;22:2066-2073.
Files in this Data Supplement:
- Table S1. Clustering stability scores (XLS, 19.5 KB)
-
The stability score is calculated in the same way for all 3 tables (see “Materials and methods”). Each of the 3 tables refers to a different situation. The 2 first tables give individual scores of stability for each of the 24 cluster dendrograms. The upper-most table shows the intrinsic stability of each of the 24 dendrograms after resampling. Here, 5% of the samples (randomly chosen) are replaced by virtual samples (obtained via random combinations of existing samples), then a cluster dendrogram is built and a stability score is computed by comparing it to the original dendrogram. Averaged scores after 100 iterations are shown. Columns correspond to each of the 24 cluster dendrograms, each row corresponds to a partition in k clusters (k = 2...5) obtained after cutting a dendrogram. For example, cell (O,10) gives the stability score (0.992) relative to the partition in 3 clusters of the dendrogram obtained with Ward linkage using 5% of the most varying genes. The second table shows the intrinsic stability of each of the 24 dendrograms when confronted with noise (see “Materials and methods”). Averaged scores after 100 iterations are shown. The third table shows the stability of the topology using different number of genes for a given linkage. It gives the averaged scores (for each linkage method) across the different gene lists for a given k number of nodes.
- Table S2. Supervised analysis of TEL-JAK2 versus TEL-JAK2/Rag2–/– leukemic cells (P<.01) (XLS, 130 KB)
-
Comparison of the gene expression profile of the TEL-JAK2 (TJ) versus TEL-JAK2/Rag2–/– (TJR) leukemic cells. Provided are the probe set ID, gene symbol, gene title, chromosomal location, and the following information if the given probe set met the P value cut-off (P < .01): non-log geometric mean intensity values for each group (eg, Geom mean TJ, Geom mean TJR, etc); fold change (FC) between the different groups (eg, FC TJ/TJR); parametric P value and the false discovery rate (FDR) calculated based on 1000 permutations.
- Figure S1. TEL-JAK2 leukemic cells are hypersensitive to TCR signaling agonists (JPG, 90.3 KB)
-
(A) Leukemic cells from a representative TEL-JAK2 tumor were left untreated in culture medium or treated with either anti-CD3 antibody or 10 ng/mL PMA for 15 hours at 37°C before CD69 cell surface immunostaining. (B) Leukemic cells from a representative TEL-JAK2 tumor (no. 38) and wild-type (WT) thymocytes were left untreated or treated in vitro with 1 µg/mL concanavalin A (ConA), 1 µg/mL plate-bound anti-CD3 monoclonal antibody, 10 ng/mL PMA, or this dose of PMA together with 250 ng/mL ionomycin (iono) for 2 days at 37°C. Proliferation was quantified by 3H-thymidine incorporation. Similar results were obtained for leukemic cells from more than 5 independent mice. Inset shows results for wild-type thymocytes only.
- Figure S2. A small subset of TEL-JAK2 leukemic cells express low CD69 levels (JPG, 63.3 KB)
-
Leukemic cells from two representative TEL-JAK2 transgenic tumors (nos. 1007 and 1130) were stained with fluorescently labeled CD69 antibody or an isotypic control monoclonal antibody.
- Figure S3. Tcra-deficient TEL-JAK2 transgenic leukemic cells are unresponsive to anti-CD3ε in vitro stimulation (JPG, 67.6 KB)
-
Leukemic cells from representative Tcra+/– and Tcra–/– TEL-JAK2 tumors (nos. 29 and 52, respectively) and Tcra+/+ and Tcra–/– thymocytes were left untreated or treated in vitro with 1 µg/mL plate-bound anti-CD3 monoclonal antibody, 10 ng/mL PMA, or this dose of PMA together with 250 ng/mL ionomycin (iono) for 2 days at 37°C. Proliferation was quantified by 3H-thymidine incorporation.
|
|
