Aldehyde Dehydrogenase (ALDH) Activity Resolves Neuroblastoma (NB) Cells and Hematopoietic Stem and Progenitor Cells (HPC) in Graft Material: Potential Method for Clinical Purging.

High dose, myeloablative chemotherapy followed by autologous HPC transplantation is used clinically to treat NB in children. The presence of NB cells in transplant material adversely affects outcome. Grafts have been purged by using monoclonal antibodies to select CD34⁺ HPC for transplantation or to remove tumor cells expressing NB surface antigens from graft material. We are exploring the feasibility of purging NB autografts by flow sorting HPC and NB cells based on differential expression of ALDH. Cell populations with high levels of ALDH activity and low side scatter (ALDH^brSSC^lo cells) can be identified in mobilized peripheral blood (PBSC), bone marrow (BM), and umbilical cord blood using the ALDEFLUOR™ [StemCo Biomedical, Inc.] method for flow cytometric measurement of cellular ALDH activity. ALDH^brSSC^lo cells isolated by flow sorting are highly enriched for HPC activity. We report here that none of three established NB cell lines (SK-N-SH, SK-N-BE(2), and IMR-32) express detectable ALDH activity in the ALDEFLUOR™ assay. Furthermore, when any of these NB lines was labeled with the stable, red fluorescent intracellular marker SNARF™ [Molecular Probes, Inc.] and then mixed with BM or PBSC, NB were readily resolved from the green fluorescent ALDH^brSSC^lo cells. In mixtures containing equal numbers of tumor cells and PBSC or BM cells 0–6 events were detected in the ALDH^brSSC^lo region out of a total of 3-10 x 10⁵ events in the SSC vs. FSC gate. Even when mixtures containing 9 tumor cells to 1 leukocyte were used, fewer than 21 SNARF-labeled NB were detected in the ALDH^brSSC^lo region out of a total of 2-9 x 10⁵events in the SSC vs. FSC gate. Additional studies were done by sorting mixtures of SNARF-labeled NB cells and PBSC or BM and measuring tumor burden post-sort. In a representative study, pre-sort NB content of 30% was reduced to 0.04% when care was taken to eliminate NB-HPC doublets by using pulse width discrimination; without this precaution, NB contamination in the sorted cells was 1.3%. RNA was prepared from pre- and post-sorted cells, and real time PCR analysis was performed using probes to NB gene products tyrosine hydroxylase, GD2 synthase and PRAME. PCR signals were readily detected in the starting sample and were reduced by >1000-fold in the sorted ALDH^brSSC^lo cells. To extend these findings with NB cell lines to clinical material, a bone marrow sample was obtained from a NB patient following appropriate institutional consent procedures and was analyzed by multiparameter flow analysis using ALDEFLUOR™ and a monoclonal antibody to GD2 synthase, an antigen strongly expressed by NB cells. Approximately 60% of the cells in this BM sample expressed GD2 synthase, but only 0.03% of the NB cells were detected among ALDH^brSSC^lo events. RNA was prepared from 100,000 flow-sorted ALDH^brSSC^lo cells, and real-time PCR analysis was performed. NB gene products were readily detected in the donor BMs, but no signal was detected in the sorted cell preparations. The NB burden in the flow-sorted ALDH^brSSC^lo population was reduced >100-fold compared to the unpurged, heavily contaminated BM. We are extending this analysis to maximize purging obtainable with clinical material. As ALDEFLUOR™ does not impair any tested functional characteristics of HPC, flow sorting based ALDH activity may provide a new method to purge autografts of NB cells for clinical transplantation.