Differentially Expressed and Novel Transcripts in Highly Purified Chronic Phase CML Stem Cells

The chronic phase of CML is sustained by rare BCR-ABL+ stem cells. These cells share many properties with normal pluripotent hematopoietic stem cells, but also differ in critical ways that alter their growth, drug responsiveness and genome stability. Understanding the molecular mechanisms underlying the biological differences between normal and CML stem cells is key to the development of more effective CML therapies. To obtain new insights into these mechanisms, we generated Long Serial Analysis of Gene Expression (SAGE) libraries from paired isolates of highly purified lin-CD34+CD45RA-CD36- CD71-CD7-CD38+ and lin-CD34+CD45RA-CD36-CD71-CD7-CD38- cells from 3 chronic phase CML patients (all with predominantly Ph+/BCR-ABL+ cells in both subsets) and from 3 control samples: a pool of 10 normal bone marrows (BMs), a single normal BM and a pool of G-CSF-mobilized blood cells from 9 donors. In vitro bioassays showed the CD34+CD38+ cells were enriched in CFCs (CML: 3–20% pure; normal: 4–19% pure) and the CD34+CD38- cells were enriched in LTC-ICs (CML: 0.2–26% pure; normal: 12–52% pure). Each of the 12 libraries was then sequenced to a depth of ~200,000 tags and tags from libraries prepared from like phenotypes were compared between genotypes using DiscoverySpace software and hierarchical clustering. 1687 (355 with clustering) and 1258 (316 with clustering) transcripts were thus identified as differentially expressed in the CML vs control CD34+CD38− and CD34+CD38+ subsets, respectively. 266 of these transcripts (11 with clustering) were differentially expressed in both subsets. The differential expression of 5 genes (GAS2, IGF2BP2, IL1R1, DUSP1 & SELL) was confirmed by real-time PCR analysis of lin-CD34+ cells isolated from an additional 5 normal BMs and 11 CMLs, and lin-CD34+CD38− cells from an additional 2 normal BMs and 2 CMLs (with dominant Ph+ cells). GAS2 and IL1R1 transcript levels were correlated with BCR-ABL transcript levels in both primitive subsets, and predicted differences in expression of IL1R1 and SELL were apparent within 3 days in CD34+ cord blood cells transduced with a lenti-BCR-ABL-IRES-GFP vs a control lenti-GFP vector (n=3). These findings support a direct role of BCR-ABL in perturbing the expression of these 3 genes. Further comparison of the meta CD34+CD38− and CD34+CD38+ CML cell libraries with most publicly accessible SAGE data revealed 69 novel tags in the CD34+ CML cells that correspond to unique but conserved genomic sequences. Nine of these were recovered by 5′- and 3′- RACE applied to cDNAs pooled from several human leukemic cell lines. These results illustrate the power of SAGE to reveal key components of the transcriptomes of rare human CML stem cell populations including transcripts of genes not previously known to exist. Continuing investigation of their biological roles in primary CML cells and primitive BCR-ABL-transduced human cells offer important strategies for delineating their potential as therapeutic targets.