Ex vivo treatment of proliferating human cord blood stem cells with stroma-derived factor–1 enhances their ability to engraft NOD/SCID mice

Much interest is currently focused on improving in vitro procedures for expanding hematopoietic stem cells (HSCs) that do not compromise either their original developmental or engrafting potential. Growth factor (GF) combinations have now been identified that promote self-renewal divisions in vitro of transplantable human HSCs identified as competitive repopulating units (CRUs) based on their ability to repopulate the marrow of nonobese diabetic-scid/scid (NOD/SCID) mice with lymphoid and myeloid progeny.1-7 We have exploited the precision and selectivity afforded by this assay to show that cord blood (CB) CRUs proliferating in vitro lose their ability to engraft as they progress through the S/G₂/M phases of the cycle,8 as has also been shown for adult human9 and murine HSCs10 detected by similar transplantation assays. Previous approaches to increase the transplantability of proliferating murine and human HSCs have exposed the cells to transforming growth factor-β₁ (TGF-β₁) for 24 hours.8,11 Although this arrested many of the cells in G₀/G_1, no increases in transplantable HSC activity were seen. We subsequently found that a more prolonged exposure to either TGF-β₁ or the chemokine stroma-derived factor–1 (SDF-1) can block the cycling of long-term culture-initiating cells (LTC-ICs) as well as primitive human colony-forming cell (CFC) progenitors both in vitro and in vivo.12 We therefore reasoned that such treatments might similarly affect proliferating HSCs and thereby increase their ability to be detected as CRUs. The experiments described here were designed to test this hypothesis. Accordingly, CD34⁺ cell-enriched suspensions of CB were first cultured in the presence of a GF combination that we have previously shown stimulates all of the CRU to proliferate within 4 days.13 The cells were then switched to a different GF cocktail to test the effect of added TGF-β₁ and/or SDF-1 on HSC detection. The GF cocktail selected for use in these secondary cultures consisted of Steel factor (SF) plus thrombopoietin (TPO) based on previous studies indicating that these GFs are less mitogenic but still able to maintain the viability and stem cell properties of HSCs.14-17 

CB was obtained with informed consent, lineage marker–positive (lin⁺) cells removed using StemSep columns (StemCell Technologies, Vancouver, BC) from previously cryopreserved low-density (< 1.077g/mL) cells pooled from at least 5 donors, and the lin⁻ cells then cultured at a density of 1 × 10⁵/mL to 2 × 10⁵/mL for 4 days in serum-free medium (SFM) containing a serum substitute (BIT9500, StemCell), 10⁻⁴ M 2-mercaptoethanol (Sigma Chemicals, St Louis, MO), 40 μg/mL low-density lipoproteins (Sigma), and the following 5 recombinant human GFs: 20 ng/mL interleukin-3 (Novartis, Basel, Switzerland), 20 ng/mL interleukin-6 (Cangene, Mississauga, ON), 20 ng/mL granulocyte colony-stimulating factor (StemCell), 100 ng/mL SF (expressed and purified in the Terry Fox Laboratory), and 100 ng/mL Flt3 ligand (Immunex, Seattle, WA) as previously described.13 The cells were then harvested, washed twice in SFM, aliquots removed for phenotype and functional measurements (see below), and the remainder resuspended in new SFM and cultured further as indicated (either with the same 5 GFs or with 50 ng/mL human SF and 50 ng/mL human TPO (Genentech, Palo Alto, CA) with or without either 5 ng/mL TGF-β₁ (R&D Systems, Minneapolis, MN) or 100 ng/mL SDF-1 (synthesized and purified as previously reported18) or both.

Expression of CD34 was analyzed after staining aliquots of cells with fluorescein isothiocyanate–conjugated 8G12 (anti-CD34) antibody (kindly provided by P. M. Lansdorp, Terry Fox Laboratory) and of CXCR4 using a phycoerythrin-conjugated anti-CXCR4 antibody (Pharmingen, Baie d'Urfe, Quebec). The cell cycle distribution of CD34⁺ cells was determined after propidium iodide (PI) (Sigma) staining as previously described.8 

CFC and LTC-IC assays were performed as previously described.19 

CRU frequencies were determined as described in detail previously.2 Briefly, 8- to 12-week-old NOD/SCID mice were given a sublethal dose of 350 cGy of total body ¹³⁷Cs γ-irradiation and were then injected intravenously with test cells plus 10⁶ irradiated (15 Gy) normal human bone marrow cells as carriers. After 6 to 8 weeks, the presence of viable (PI⁻) human B-lineage (CD34⁻CD19/20⁺) and human myeloid (CD45/71⁺CD15/66b⁺) cells in the marrow of the mice was determined by fluorescence-activated cell sorter (FACS) analysis. A detection limit of 5 human lymphoid (CD34⁻CD19/20⁺) cells and/or 5 human myeloid (CD45/71⁺CD15/66b⁺) cells per 2 × 10⁴ PI⁻ cells analyzed was used to identify positively engrafted mice. Gates were set to exclude more than 99.99% of nonspecifically stained PI⁻ cells incubated with irrelevant isotype-matched control antibodies labeled with the corresponding fluorochromes. CRU frequencies (Table1) were calculated from the data pooled from all experiments using Poisson statistics and the method of maximum likelihood (with L-calc software, StemCell). Engraftment levels varied from 0.2% to 71% human cells according to the number of CRU transplanted, consistent with previous observations.2 

Mean values (± SEM) were calculated from the data pooled from replicate experiments. Differences between groups were assessed using the Student t test.

A first series of experiments were undertaken to determine the rate at which proliferating CD34⁺ CB cells would be arrested by TGF-β₁ and/or SDF-1. Proliferating CD34⁺ cells were isolated by FACS from 4-day primary (“proliferation”) cultures originally initiated with lin⁻ CB cells. These cells were then transferred into new SFM containing SF and TPO with TGF-β₁ and/or SDF-1. As shown in Figure 1A, exposure to TGF-β₁ in the secondary cultures caused a rapid increase (within the first 24 hours) in the proportion of CD34⁺cells in G₀/G₁ (from an initial value of 54% ± 2% to > 75%). Maximum levels of G₀/G₁ cells (82% ± 3%) were reached another 24 hours later. In the same cultures, the proportion of CD34⁺ cells undergoing apoptosis (identified by a DNA content of < 2n) also increased significantly but somewhat more slowly (from 3.6% ± 0.1% to 47% ± 4%; P < .01, Figure 1B). In contrast, SDF-1 did not affect either of these parameters (P > .05).

Based on these findings, a 48-hour period of exposure to TGF-β₁ (and/or SDF-1) was chosen to evaluate the effects of these agents on the functional activities of various progenitors identified by quantitative in vitro and in vivo assays. Accordingly, a second series of similar experiments were set up with larger numbers of lin⁻ CB cells. In these, all of the cells (not just the CD34⁺ subset) harvested from the 4-day primary proliferation cultures were transferred into secondary cultures. Forty-eight hours later, aliquots were removed for FACS analyses of their cell cycle parameters and expression of CD34 and CXCR4 and also for assessment of the number of CFCs and LTC-ICs present. The remainder (> 90% of all the cells harvested from the secondary cultures) were injected into irradiated NOD/SCID mice. Cell cycle analyses revealed similar distributions of the total cell population as shown in Figure 1A for the CD34⁺ compartment (data not shown). Despite the inhibition of proliferation seen in all cultures exposed to TGF-β₁, the total number of CFCs present increased over the 2 days of incubation in all of the secondary cultures (Figure 2). However, this expansion was generally less in the TGF-β₁–containing cultures and, as expected,20 the presence of TGF-β₁selectively suppressed the output of erythroid CFCs (P < .01). Yields of LTC-ICs from all cultures were more variable, but neither TGF-β₁ nor SDF-1 had a significant effect on their numbers (P > .1, data not shown). Also as expected,21 exposure to SDF-1 (but not TGF-β₁) down-regulated expression of the SDF-1 receptor (CXCR4), resulting in a decrease of CXCR4⁺CD34⁺ cells from 72% ± 3% to 45% ± 6% (P = .012). The results of the NOD/SCID repopulation assays are shown in Table 1. Significantly more CRUs (2.3-fold,P < .02) were detectable in the cells harvested from the secondary cultures containing SDF-1 than in those containing SF plus TPO alone or in those maintained in the original 5-GF cocktail used in the primary cultures. In contrast, TGF-β₁ did not affect the number of CRUs detected and even antagonized the enhancing effect of SDF-1.

There are 3 ways that SDF-1 could have elicited the increases in CRU numbers seen. One mechanism could be the stimulation of a faster rate of CRU proliferation (faster cell cycle transit time). However, given the evidence that SDF-1, like TGF-β₁, inhibits the proliferation of primitive hematopoietic cells, including LTC-ICs,12,22,23 such a possibility seems unlikely. A second mechanism would be the induced dedifferentiation of cells that had already progressed to a later progenitor stage prior to their exposure to SDF-1. Although such a possibility cannot be formally ruled out, the increased numbers of CFCs seen do not favor this explanation; nor was there any evidence of an increase in LTC-ICs, which are generally viewed as closely related to CRUs. A third way that increased numbers of CRUs could have been produced is by recruitment from a previously undetectable, but developmentally equivalent, cell population. Such a situation is known to exist because proliferating HSCs transit S/G₂/M,8,10 and the recent evidence of a cytostatic effect of SDF-1 on closely related progenitor types12 is consistent with a proliferation blockade of HSCs.12,22,23 Moreover, the 2-fold higher number of CRUs detected when the cells were exposed to SDF-1 is exactly what would be anticipated from an arrest in G₁ of those that would otherwise have entered S/G₂/M. Alternatively, SDF-1 may exert its effects on the transplantability of HSCs independent of effects on HSC cycling. However, incubation of freshly isolated quiescent CB HSCs with SDF-1 has been found to decrease the ability of these HSCs to engraft NOD/SCID mice.21 An SDF-1–mediated arrest of cycling HSCs thus seems a more plausible explanation for the enhancing effect observed here. The effectiveness of SDF-1 in the absence of stroma also suggests that a direct action on HSCs was obtained. Similar results have been recently reported for preactivated primate HSCs maintained for 4 days in secondary cultures containing SF and retronectin (but no other GFs or chemokines).24 

At first glance, our finding that TGF-β₁ did not have a similar effect on the number of CRUs detected seems contradictory. However, given the high frequency of apoptotic cells seen in the CD34⁺ compartment when these cells were exposed to TGF-β₁, such an effect may have offset any increase in CRU numbers. In line with this interpretation was the finding that a net increase in CRUs detected was also not obtained when the cells were cultured in TGF-β₁ as well as SDF-1. However, regardless of the mechanism(s) ultimately found to explain the findings described here, they suggest an interesting new approach to enhancing HSC yields in a variety of cellular and gene therapy strategies that involve stimulating HSCs to proliferate ex vivo prior to their transplantation in vivo.

The authors thank Gayle Thornbury, Giovanna Cameron, and Rick Zapf for assistance in cell sorting; the staff of Stem Cell Assay Service for initial hematopoietic cell processing; and Amy Ahamed for manuscript preparation. The authors also thank Amgen, Cangene, Immunex, Novartis, StemCell, and Peter Lansdorp for generous gifts of reagents.