Introduction
In a 1981 review article, Herzig [
1
] concluded that there was “…no reliable means of learning whether the hemopoiesis reconstituting ability of a marrow with normal cellularity has been diminished by treatment received prior to obtaining marrow samples for storage other than the demonstration of the ability of transplanted marrow to restore hemopoiesis after treatment that produces marrow aplasia.” Although significant advances in experimental hematology have occurred in the 20 years since this review, Herzig's statement largely remains true. Advances in our understanding of the organization of the hematopoietic system and the development of powerful cell separation techniques and reagents to subset and hierarchically order the stem cell compartment have failed to produce consensus about the best way to quantitate stem cells. Thus, a definitive assay to rapidly and reliably predict the long-term reconstitutive ability of stem cells in a clinical setting remains elusive. (Table 1)Table 1Glossary
| Cell culture assays | |
|---|---|
| CAFC | cobblestone area forming cell (stroma-supported progenitor cell that displays a stem cell phenotype over 2–12 weeks in culture) |
| HPP-CFC | high proliferative potential colony-forming cell |
| LTC-IC | long-term culture-initiating cell |
| Cell lines | |
| 32D | a murine, IL-3 dependent cell line |
| KG1a | a human leukemia cell line that does not spontaneously differentiate to granulocytes or macrophages, or respond to colony stimulating factors (CSFs) |
| U-937 | a histiocytic cell line derived from a lymphoma patient; [some stocks were contaminated with KG1 cells] |
| Cell surface molecules and markers | |
| CAS receptor | crk-associated substrate (a focal adhesion protein) receptor [crk is a protooncogene in the tyrosine kinase pathway] |
| CXCR4 | a chemokine receptor for a stroma-derived factor (SDF-1) |
| Ho | Hoechst 33342 dye [low retention of this dye by stem cells] |
| ICAM | intercellular adhesion molecule |
| LFA-1 | lymphocyte function-associated molecule-1 |
| lin | lineage-associated antigens, e.g., CD2, CD3, CD10, CD11b, CD14, CD19, CD20, CD33, CD36, 7B9, and glycophorin-A |
| PECAM | platelet-endothelial cell adhesion molecule |
| Rho | rhodamine 123 dye [low retention of this dye by stem cells] |
| Sca | stem cell antigen (also found on B-cells, stroma, embryonic stem cells, etc.) |
| Thy-1 | a stem cell antigen (also found on thymocytes, brain, and stroma cells) |
| VLA-4 | very late antigen-4 (an integrin) |
| Cytokines and protooncogenes | |
| EPO | erythropoietin |
| flk-2 | fetal liver kinase-2 |
| flt-3 | fms-like tyrosine kinase-3 (a.k.a. fetal liver tyrosine kinase) |
| FL | flt-3 ligand |
| fms | homolog to feline McDonough strain sarcoma virus (encodes receptor for M-CSF) |
| G-CSF | granulocyte colony-stimulating factor |
| GM-CSF | granulocyte-macrophage-colony-stimulating factor |
| IL- | interleukin-[followed by number] |
| c-kit | homolog to the Hardy-Zuckerman 4 feline sarcoma retrovirus (HZ4-FeSV); product of mouse W gene locus, binds SCF |
| M-CSF | macrophage-CSF |
| MIP-1α | macrophage-inhibitory protein-1 alpha |
| SCF (KL) | stem cell factor, c-kit ligand, or Steel factor |
| Steel factor | same as SCF |
| TPO | thrombopoietin |
| Developmental genes | |
| Δ | a ligand that inhibits the notch receptor-bearing cell, but allows migration of the delta-containing cell |
| dishevelled | a signal transduction gene involved in cell polarity |
| jagged | a ligand for the notch receptor that helps to maintain stem cells |
| lunatic fringe | a modulator of the notch signal pathway in segmentation |
| notch | a surface receptor that can signal the nucleus to repress differentiation |
| numb | a transcription factor that is asymmetrically distributed and may help to determine cell fate by interference with notch signals |
| scfr | stem cell frequency regulator genes that control stem cell pool size |
| wnt | a family of genes involved in cell fate determination in tissues |
| Mouse strains | |
| NOD/SCID | non-obese diabetic/severe combined immunodeficiency |
| W, Wv | white dominant spotting (v = viable >1 week to maturity). This mutation on chromosome 5 causes reduced pigmentation, sterility, and macrocytic anemia. The phenotype is similar to the steel (SL) mouse, but the W mutant has a defect in the c-kit protooncogene, a tyrosine kinase receptor for stem cell growth factor (which is defective in SL mice). W mice have severe anemia from day 12 of gestation and die within the first week of birth. Viable dominant spotting (Wv/Wv) mice can survive to maturity. Heterozygous (W/+ or Wv/+) mice have white spotting and are fully viable and fertile. |
To address these issues and identify obstacles that limit the correlation among in vitro assays, in vivo assays, and transplant outcomes, a stem cell assay workshop was held at the National Institutes of Health on September 8–9, 1998, to review the state of the art in surrogate stem cell assays. Participants at this meeting discussed the immunophenotypic, molecular, and functional characteristics of hematopoietic stem cells, with a view to the systematic establishment of rational and reliable benchmarks for the rigorous comparison and evaluation of surrogate assays. This initial workshop was followed by a smaller working group meeting on July 30, 1999, to develop a strategy for standardization of surrogate stem cell assays relevant to clinical transplantation.
In this summary of the workshop and working group proceedings, the presentations and discussions of the participants have been edited and reorganized for the sake of brevity and clarity. The participants are listed in the Appendix, and only key references to published work are given in the bibliography.
Description of the issues in stem cell assay development
To open the meeting, Peter Quesenberry summarized some major obstacles to the development of predictive stem cell assays. They may be characterized at the levels of:
- 1.Intrinsic properties of the stem cell. Stem cell properties may be selected for or altered merely by assays. Stem cells may vary in their ability to enter cell cycle, to engage in asymmetric divisions, or to home to and proliferate in the bone marrow.
- 2.External influences. These factors may be soluble (e.g., cytokines) or cell-bound. Assay results also can be affected by small perturbations in the cell culture environment.
- 3.Host factors. There may be immunologic disparities between the donor and the host and the transplant conditioning regimen may affect engraftment. Other physiologic influences include circadian rhythms and estrous cycles.
- 4.Definitions of engraftment. Short-term and long-term endpoints for hematopoietic reconstitution in vivo are defined arbitrarily.
Ivan Bertoncello emphasized the impact of these factors and the limitations of stem cell assays. Hematopoietic stem cells are operationally defined by their ability to sustain life-long, multi-lineage production of mature blood cells in a steady state or after perturbation and by their capacity to regenerate the entire hematopoietic system long term after transplantation. Whereas a single cell or a few extremely rare cells are able to fulfill these criteria, larger sets of primitive cells with varied proliferative histories, maturational ages, and differentiative potentials have been used to restore life-long hematopoiesis when transplanted in large numbers in hosts treated with different preparative regimens. Since the engraftment potential of different subpopulations can be altered by their intrinsic state as well as external factors, it may be impossible to determine whether a surrogate assay monitors or predicts the behavior of a compartment or individual cells within that compartment. Furthermore, surrogate in vitro assays that measure the function of a small sample often monitor the function of only a few highly selected cells from the total stem cell pool. Other phenotypic and functional characteristics, such as homing, engraftment, and renewal, may not be examined, so the ability to predict the fate of transplanted cells and transplant outcomes is inherently limited [
2
].In addition to these issues, Ruud Hulspas described other problems with the standardization of assay techniques. There have been a wide variety of markers, protocols, and surrogate assays, which should be correlated with the ability of stem cells to proliferate and differentiate in vivo [
3
]. Cell preparations may differ among laboratories because stem cell purification methods cannot be easily or precisely replicated. Failure to control variables such as mouse strains, antibody specificity or cocktails, buffer composition, preparation time, and flow cytometry settings have made the comparison of data impossible. Dr. Hulspas cited an example where two technicians processed and sorted aliquots of the same cell preparation following identical protocol but obtained different results. Slight discrepancies in the pH of media and buffers or the overnight refrigeration of cells, for instance, can alter cellular fluorescence or stem cell function. For CD34+ cell analysis, disparities of about 1 log in a single sample are not uncommon when measured by different laboratories, yet variations of 10 to 20% or less are required for this assay to be useful. Major obstacles to standardization of stem cell assays are the preference of investigators for their own assays and the difficulty of sharing techniques and limited cell samples for cross comparison. A potential solution to some of these problems is to include a standard or reference assay in published data.In vivo assays
Measurement of long-term, multi-lineage hematopoietic reconstitution in vivo is used to validate other surrogate assays of stem cell potential. While impractical for routine evaluation in clinical transplantation, competitive repopulation assays in mice have provided a rationale for the study of the potential of human hematopoietic stem cells in NOD/SCID mice and fetal sheep.
David E. Harrison described a competitive repopulation assay in mice. Whereas long-term engraftment following human transplants may take years to assess, this measurement requires 6 to 8 months in mice. In this assay, stem cells from a test donor mouse are mixed with a standard, congenic, “competitor” marrow, distinguished by genetic differences in hemoglobin and glucosephosphate isomerase (GPI). The mixture is infused into an irradiated, congenic host to determine the relative growth of the transplanted cells, whose counts stabilize after 21 days. After 3 to 4 months, repopulation has been achieved by functional, primitive hematopoietic stem cells (PHSCs), proportional in number to the bone marrow cells infused. This repopulating ability is expressed relative to the standard competitor as repopulating units (RUs), where each unit is equivalent to that of 105 adult marrow cells. Since this method compares the repopulating ability of a group of cells, limiting dilution can be used to estimate small numbers of functional stem cells in a set of 20 or more mice with a Poisson model. Although the total number of PHSCs stays constant with age, this assay has revealed differences between different mouse strains [
4
]. Marrow cells from young BALB/cByJ and DBA/2J mice, for instance, repopulate better than cells from older mice, with relative values of 1.67 in fetal liver, to a standard of 1.0 in marrow from young animals, to .63 with marrow from older animals. For C57BL/6J mice, the older cells repopulate as well as younger ones, and when the recipient marrow is used for secondary transplants, there is less loss of repopulating ability than with marrow from the former strains. Both the competitive repopulation and dilution assays can measure long-term repopulating ability of genetically marked PHSCs in vivo without lengthy cultures, and the relative growth potential of these populations is compared to the same standard. The disadvantages for these assays are that they require 3 to 6 months for completion and they cannot be done in humans.John Dick discussed the immunodeficient mouse assay, which is based on the reconstitution of nonobese diabetic mice with severe combined immunodeficiency disease (NOD/SCID mice) [
5
]. The SCID mouse-repopulating cells (SRCs) are more primitive than progenitors detected by in vitro assays for colony-forming cells (CFCs) and long-term culture-initiating cells (LTC-ICs). SRCs are rarely transduced with retroviruses, which distinguishes them from CFCs and LTC-ICs. By limiting dilution, the frequency of SRCs is about one in 617 human cord blood CD34+CD38− cells, and 1 SRC transplanted into a mouse produces about 400,000 progeny after 6 weeks. There is approximately one SRC in 3.0 × 106 adult bone marrow cells, one in 6.0 × 106 mobilized peripheral blood cells from normal donors, and 1 in 9.3 × 105 cord blood cells. When CD34+CD38− cord blood cells are cultured in serum-free media for 4 days, the total number of CD34+CD38− cells increases fourfold, with a 10-fold rise in CFCs and a twofold to fourfold increase in SRCs. However, longer expansion (for 9 days) results in the loss of all SRCs, despite further increases in total cell number, CFC content, and CD34+ cell numbers. The CD34− population, which does not proliferate in long-term cultures, also has been assayed in NOD/SCID mice for SRCs. Although such mice often develop thymomas after 6 months (which makes lengthy studies difficult), they can be repopulated long-term with CD34− CD38− lin− cells. However, the frequency of SRCs in these cells is low (about one in 1.3 × 105). To exclude the possibility of contamination by CD34+ cells, CD34− cells were expanded in serum-free or human umbilical vein endothelial cell (HUVEC)–conditioned media for 4 days. With the latter media, fewer cells became CD34+, yet the efficiency of engraftment improved to 1 to 2–3 × 104 cells. It is not known if selective death of differentiated cells or increased expression of homing molecules during culture led to these results.Esmail Zanjani has developed a xenotransplant model in which the engraftment potential of human stem cells is assessed in fetal sheep. Recently, he examined the repopulating potential of human CD34− cells [
6
] by transplanting these cells into primary and secondary recipients and employing a stem cell exhaustion strategy. In primary recipients, human CD34+ and CD34− cells persisted for about 20 months posttransplant. As few as 75 CD34+ CD38− cells were required for engraftment, but upon expansion in vivo, the relative percentage of CD34+CD45+ cells has been consistently higher for the group receiving CD34− grafts than for the animals receiving the CD34+ grafts (9.6 ± 2.8% vs 2.4 ± 0.7%, n = 8). While human cell activity in the group receiving CD34+ grafts remained relatively unchanged, a significant increase in donor cells in the group receiving CD34− grafts started to occur 14 months post-transplant. The long-term potential of CD34− cells also was demonstrated by re-transplantation of human CD34− cells from the bone marrow of chimeric primary animals into secondary recipients, which resulted in multilineage engraftment that persisted for at least one year. The human cells in animals that became chimeric after receiving 4 × 104 CD34+ cells were depleted following four cycles of treatment with human IL-3 + GM-CSF, but no significant effect was seen in animals that were chimeric after receiving 6 × 104 CD34− cells. These observations in primary and secondary hosts suggested that the latter cells can undergo self-renewal and may be more primitive than the CD34+ cell population.In vivo determinants of stem cell number and function
Christa Muller-Sieburg discussed the effect of genetic diversity on the size of the stem cell compartment [
7
]. In the outbred human population, there is more than 100-fold variation of LTC-ICs that cannot be explained by donor, age, or sex. For inbred mice, LTC-IC levels vary little within a strain, but inter-strain differences are dramatic. DBA/2 mice have the highest levels of LTC-ICs, and C57BL/6 mice have a far lower number. The strains 129 and FvB also have fairly high levels of LTC-IC. To examine the genetic basis for variation in stem cell pool size, Dr. Muller-Sieburg studied inbred mice and showed multigenic control of the phenotype. Two candidate loci were identified, whose allele patterns were significantly associated with the quantitative variation in stem cell frequency. These loci have been named stem cell frequency regulator (Scfr) genes. Both are located on chromosome 1; each comprises about 10 centimorgans, and both are syntenic to parts of chromosome 1 in humans. Confirmation of the location of the Scfr-1 locus was obtained through congenic mouse experiments. The congenic strain B6.C-H25 has a C57BL/6 background and carries a segment of chromosome 1 derived from BALB/c. Current efforts are directed toward further mapping the loci by testing reduced congenics. In other experiments with chimeric mice, Dr. Muller-Sieburg showed that intrinsic properties of stem cells contributed 80% toward LTC-IC frequency, and the environment accounted for the rest.Peter Quesenberry discussed the variations in stem cell phenotype and engraftment associated with cell cycle [
8
]. Most marrow hematopoietic stem cells are quiescent, as shown by studies in which mice were fed oral bromodeoxyuridine (BrdU) for a month. About 70% of the cells were labeled, but about 30% of them did not cycle. Less than 2% of lin− Hoechst(low)/Rhodamine(low) [lin−Holow Rholow] and 10 to 15% of lin−Sca+ murine cells were in S-phase; however, more than 50% of the former were high-proliferative potential colony-forming cells (HPP-CFCs). Exposure of HPP-CFCs to IL-3, IL-6, and IL-11, ± Steel factor drove them into S-phase as shown by incorporation of 3H-thymidine. The first cell division took 36 to 40 hours with more than 60% of cells in S-phase, but each of five subsequent divisions occurred every 12 hours. If male mouse marrow was stimulated with cytokines for various periods and injected with fresh female marrow into an irradiated BALB/c mouse, a competition assay revealed cell cycle variation in engraftment potential. Engraftment was better when male cells were cultured for 40 hours (presumably G1), but poorer at 32 hours (presumably late S/early G2). A defect in stem cell homing may explain these observations, but the results are supported by studies where 5-fluorouracil (5-FU) was used to treat murine marrow donors before transplantation into nonmyeloablated hosts. After 6 days of 5-FU, the marrow engrafted poorly, but the marrow repopulating potential returned to normal by 12 to 35 days after 5-FU treatment. In other experiments, male mouse marrow was transplanted into female nonablated hosts, followed by a short dose of hydroxyurea (900 mg/kg) at 0, 3, 6, 12, and 15 hours. Only at the 12-hour point were more than half of the cells that contributed to long-term hematopoiesis killed. Presumably they were in S-phase and sensitive to hydroxyurea. Thus, variations in stem cell cycle can affect PHSC assay, cell harvesting, and engraftment outcomes.David Bodine described the use of competitive repopulation assays to define stem cell populations for gene therapies. Long-term repopulating cells were characterized as lineage negative (lin−), c-kithi cells. Approximately 30 to 50 of these cells were able to reconstitute a w/wv congenic mouse, and about 500 of these cells were required to compete almost equally with 2 × 106 unfractionated bone marrow cells. Large stem cells contained more amphotropic retroviral receptor mRNA than small cells, which made them more amenable to nonspecies-specific transduction [
9
]. However, receptor mRNA in the small cells could be upregulated sixfold to 10-fold by incubation with IL-3 (10 ng/mL), IL-6, and SCF for up to 144 hours. Transfected human cells also can be assayed in NOD/SCID mice, but extinction of expression of transduced genes remains a problem. This effect may be lessened by the use of internal promoters, rather than reliance on the retroviral LTR. Human progenitor cells are hard to transduce, but Dr. Bodine has found that frozen and thawed cord blood has a 10-fold higher level of amphotropic retroviral receptor mRNA than fresh samples.Stroma-based in vitro assays
Rob Ploemacher discussed in vitro stroma-based assays for stem cells, which may be more reliable than phenotypic analyses [
10
]. These assays were developed because stem cells proliferate better with stromal contact or in stroma-conditioned medium (SCM), irrespective of the addition of cytokines. He tried to correlate cobblestone-area-forming cell (CAFC) numbers with in vivo stem cell assays that monitor spleen colony formation (CFU-S), marrow-repopulating ability (MRA), and long-term repopulating ability (LTRA) using sex-mismatched hemopoietic chimerism. He showed that cells selected by wheat germ agglutinin (WGA) lectin were not able to initiate long-term in vitro cultures on a stromal layer in vitro, but could produce CFU-S at day 12. The WGA-dim cells contained the repopulating ability, with a CAFC frequency at 4 to 5 weeks of culture that correlated with LTRA at 6 to 12 months post-transplant. These isolation conditions allowed for a 590- to 850-fold enrichment of LTRA over normal bone marrow. It has not been possible to take any single cobblestone area and reconstitute an irradiated animal, probably because most of the primitive stem cells that produce the cobblestone area lose their repopulating ability. These functional assays, both in vitro (CAFC, LTC-IC) and in vivo (repopulation of NOD/SCID mice), have not been validated in the human transplant setting. Dr. Ploemacher also used these methods to study human cord blood. He pooled cord blood samples and cultured them for 2 to 12 weeks in Flt-3L, SCF, TPO, and IL-6. The CD34+CD38− cells (which also contained CD34+ CD19+ pre-B-cells) underwent 103- to 106-fold expansion, but the number of CAFCs increased only 10- to 100-fold after 6 weeks. When 2-week cultures were injected into NOD/SCID mice, the number of cells required to repopulate them after 6 weeks fell from 100,000 to less than 10,000. Thus, this assay may be helpful in validating the functional capacity of stem cell expansion products.Catherine Verfaillie described a method to identify single cells that have the capacity for self-renewal and multilineage differentiation. Called the myeloid-lymphoid initiating cell assay, it adapts the LTC-IC protocol with a lymphoid differentiation step [
11
]. Single human bone marrow CD34+lin− HLA-DR− or CD38− cells were obtained by FACS and cultured on stroma from AFT024 murine fetal liver cells in 96-well plates. The cytokines Flt-3L, IL-7, and SCF were added to the medium for 4 to 6 weeks. Then the contents of each well were trypsinized and split into eight parts, with each part placed into corresponding wells of eight secondary plates. Four plates were given myeloid stimuli, and four were given lymphoid stimuli. After another 5 to 7 weeks, the secondary colonies were tested for myeloid or lymphoid markers, such as CD15 (myeloid), CD56 (NK cells), CD19 (B cells), CD1a or CD11b (dendritic cells), and T-cell epitopes. This assay demonstrated the existence of multipotent progenitor cells, which are about 10 times less frequent than the usual LTC-ICs. The experiments take about 15 weeks and are difficult to perform because of the risk of contamination.Beverly Torok-Storb described an approach to dissect the role of stroma in stem cell assays [
12
]. Because the marrow microenvironment is complex, her overall strategy has been to clone functionally distinct stromal cell lines and to examine their gene products. In addition to ELISA assays, she has collaborated with a biopharmaceutical company (Genetics Institute, Cambridge, MA, USA) to screen for differential expression of about 250 human genes with oligonucleotide arrays. Two human cell lines were described. The first one, HS-5, was immortalized with a replication-defective recombinant retrovirus (LXSN-HPV16 E6E7), which contained human papillomavirus E6/E7 genes. These genes interfere with the tumor-suppressor proteins p53 and retinoblastoma (Rb) and prevent cell cycle arrest without significant transformation. HS-5 cells resemble fibroblasts, lack contact inhibition, and produce extracellular matrix. They secrete G-CSF, GM-CSF, M-CSF, MIP-1, IL-1, IL-6, IL-8, and IL-11. Media conditioned with these cells support the proliferation of hematopoietic progenitors, but the greatest increase in CFU from CD34+ cells is seen when kit-ligand (KL) or flt3-ligand (FL) is added. When human CD34+CD38− cells are cultured in HS-5–conditioned medium plus 100 ng/mL FL for 15 days and injected into fetal sheep, they engraft and can be transplanted serially. A clinical protocol to expand CD34+ cells for autotransplantion is in progress. Another cell line, HS-27a, does not support the growth and differentiation of CD34+38lo cells in serum-free media and does not secrete detectable levels of cytokines. However, it does promote the formation of “cobblestone” areas. HS-27a expresses the human gene hJagged1, whose product can bind to the notch1 receptor to repress differentiation. Contact between HS-27a and the hematopoietic precursor 32D cell line also activates notch1 and prevents G-CSF–induced differentiation. These cells may be used to study notch/jagged signaling and possibly to identify peptides or antibodies that can function as agonists or antagonists of this pathway.Clonogenic in vitro assays
Ivan Bertoncello described the advantages and limitations of the HPP-CFC assay as a surrogate measure of stem cell potential [
13
]. HPP-CFC are defined operationally by their relative resistance to near-lethal doses of 5-FU, their obligatory requirement for multiple cytokines, and their formation of macroscopic colonies (>.5 mm diameter with ⩾50,000 cells). In mice, HPP-CFCs co-fractionate with PHSCs and regenerate after 5-FU treatment, similar to PHSCs with long-term reconstitution potential. Although experiments show that HPP-CFCs are among the most primitive hematopoietic cells detected in clonal agar culture and are a reliable surrogate assay of stem cell activity, investigators need to be aware of their limitations. HPP-CFCs are heterogeneous and comprise a hierarchical order of at least four sub-populations of primitive hematopoietic progenitors characterized by their growth factor preferences [Bertoncello I (1992) Status of high proliferative potential colony-forming cells in the hematopoietic stem cell hierarchy. In: Müller-Sieburg C, Torok-Storb B, Visser J, Storb R. (eds.) Hematopoietic Stem Cells: Animal Models and Human Transplantation Berlin, Heidelberg: Springer-Verlag. Curr Topics Microbiol Immunol 177:83–94
14
]. Consequently, whereas all PHSCs are HPP-CFCs, not all HPP-CFCs are stem cells. The most primitive HPP-CFCs, which are closely related if not identical to PHSCs, require three to seven cytokines to express their full potential. On the other hand, HPP-CFCs stimulated by mixtures of two cytokines are relatively mature and are the immediate precursors of committed progenitor cells. Some investigators fail to appreciate that the presence of HPP-CFCs, grown in the presence of multiple cytokines, does not necessarily indicate an obligatory requirement for each of these cytokines. In order to determine precisely the nature of HPP-CFCs, one must analyze their growth in the presence of each permutation and combination of cytokines. Sometimes colony size can be an unreliable index of HPP-CFC content and stem cell potential. During bone marrow regeneration in vivo or ex-vivo expansion of stem cell progeny, the kinetic status of cells in the assay may lead to the generation of large colonies, simply because colony-forming cells are more likely to grow more rapidly. Growth factor preferences of these cells need to be determined carefully. Likewise, culture conditions and overplating of cells can lead to underestimation of HPP-CFC content. Despite these caveats, the HPP-CFC assay seems to be one of the most informative, reliable, and versatile short-term in vitro assays of stem cell potential.Stem cell characteristics
A major problem in the identification of stem cells is that they are considered to be quiescent until triggered to proliferate. Jeffrey Moore discussed this stem cell phenotype as defined by a new legume lectin called FRIL [
15
]. FRIL (Flt-3 receptor-interacting lectin) was identified by its ability to stimulate proliferation of 3T3 fibroblasts transfected with the Flt-3 receptor, but not those with the related fms receptor or untransfected cells. When growth conditions of these 3T3 cells were designed to permit only Flt-3 receptor ligands to rescue cells from death, fractionation and purification of phytohemagglutinin-leukocyte–conditioned medium (PHA-LCM) showed that the new agent came from red kidney bean extracts (which also contain the mitogen PHA). However, in contrast to PHA and other mitogenic lectins, FRIL did not stimulate the secretion of IL-6 or other cytokines or exhaust the culture medium by inducing cell proliferation and differentiation. Instead, a small population of cells in a dormant state usually persisted at the end of the culture period. FRIL preserved human cord blood progenitor cells in suspension cultures for up to a month without exogenous cytokines or stroma and maintained the number of cord blood SCID repopulating cells (SRC) for 3 to 13 days in suspension culture. If the latter cells were transplanted into NOD/SCID mice, they gave rise to lymphoid, myeloid, and erythroid progeny. Subsequent exposure of FRIL-cultured cells to combinations of early-acting cytokines (without FRIL) expanded the number of total mononuclear cells, progenitors, and possible SRC. The eventual clinical applications may include: (1) synchronization of the stem cell cycle to improve gene transfer and later engraftment, (2) expansion of the number of stem cells and improvement of their quality for autologous and allogeneic transplantation, and (3) protection of stem and progenitor cells from the toxicity of S-phase specific drugs for cancer, to allow for higher doses or longer duration of chemotherapy.Anthony Ho examined asymmetric cell divisions of hematopoietic progenitors, in which one daughter cell remains a stem cell, but the other becomes committed to differentiate [
16
]. He used index sorting of various CD34+ subsets and time-lapse photography with the cell membrane labeling dye, PKH26, to monitor divisional history and correlated it with growth pattern and cloning efficiency. The fluorescence intensity of this dye is reduced by half following each cell division and can be used to gauge replication history. The first mitosis of fetal liver CD34+CD38− cells occurred after 36 to 38 hours, but then, in about 20–30% of cells, asymmetric divisions took place every 12 hours. The cells may have moved apart slightly. With progressive ontogenic sources of CD34+CD38− cells, the percentage of cells that underwent asymmetric division dropped. When CD34+ CD38− fetal liver cells were exposed to single cytokines such as FL, TPO, and SCF, and cell divisions were monitored every 3 to 12 hours for up to 8 days, cloning efficiency decreased (∼9–30%). A cocktail of SCF, IL-3, IL-6, GM-CSF, and EPO was better (∼68% cloning efficiency), but the ratio of asymmetric divisions versus the total number of cells still fell. However, the fraction of asymmetric divisions compared to the total number of divisions (asymmetric division index = ADI) remained constant at about 40%. The results indicated that mitotic rate and cloning efficiency could be altered with various combinations of growth factors, but that the ADI is not. This observation suggest that, although the pattern of commitment can be skewed by extrinsic signals, the proportion of cells undergoing asymmetric divisions is determined by intrinsic factors.Henry Chang suggested that an assay for stem cells could be based on the hypothesis that stem cells have a full complement of DNA, as manifested by long telomeres. It may be difficult to detect this property and then select viable cells to perform functional stem cell assays. Current methods allow the reverse, flow cytometric isolation of stem cells based upon surface markers, followed by an assay for telomere length of single non-viable cells. Since telomeres tend to shorten with each mitosis, an alternate strategy would be to use stem cell quiescence as a surrogate stem cell characteristic. Donor cells could be stained with stably bound vital dyes, such as PKH26 or PKH2, which do not require mitosis for incorporation and which become diluted with cell division. The cells would be cultured and colonies harvested to identify quiescent cells that have not divided. After sorting the cells the most highly fluorescent would be candidate stem cells since they would have divided the least. A similar approach has been reported where hematopoietic stem cells were pre-selected by incubation of human marrow with IL-3 and kit ligand, and the stimulated cells were killed by 5-FU [
17
]. The cells that retained PKH26 fluorescence were CD34+ and c-Kit+ and required a stromal feeder layer to proliferate in vitro.Cell homing and migration
A difference between the data generated using in vitro assays and that from in vivo assays may be due to the fact that the latter assays also depend on homing of the PHSC to the bone marrow. Homing may be difficult to distinguish from cell trapping. Visser and colleagues isolated progenitor cells with wheat germ agglutinin and labeled them with PKH26 before injection into mice [
18
]. The cells were found in the spleen and bone marrow, but surprisingly, about 75% of them were not accounted for. This loss occurred within 1 day, so it probably was not due to stem cell differentiation into nonhematopoietic tissues. Further research is needed to understand this phenomenon. Dye-stained, sca-1+, c-kit+, lin− cells that homed to the spleen could be transplanted and could repopulate secondary recipients faster than those that have homed to the marrow [19
], but the latter cells (prepared by elutriation) appeared to possess long-term reconstituting ability [20
]. Also, one expansion protocol diminished the homing of dye-stained cells 10-fold. In the future, it would be of interest to find nascent colonies in tissue sections, isolate them with techniques such as laser-capture microdissection, and characterize the most highly stained cells molecularly.Bernhard Palsson has begun to look at the effect of stem cell migration with a custom-built automated time lapse microscope system (ATLMS). It includes a fluorescence microscope with infinity-correlated objectives and a motorized stage with micromanipulators [
21
]. The system is enclosed in a sealed plastic chamber with proportional/integral/derivative (PID) temperature control and a CO2 sensor. A cooled charge coupled device (CCD) camera captures the images to be processed on a computer workstation. Dr. Palsson showed pictures of CD34+ cells extending processes up to hundreds of microns long that could be visualized with PKH26 or PKH2 fluorescent dyes. At any time, about 5% of primary CD34+ cells, but up to 60 to 70% of the human leukemia cell line, KG1a cells, extend these long podia. This phenomenon likely depends on microenvironmental variables, such as distance to neighboring cells and local cytokine or substrate concentrations. The podia display integrins, including CD11a, CD18, CD29, CD49d, and CD49e, plus other adhesion molecules, such as CD44, CD54, and CD62L. They formed on surfaces coated with fibronectin, laminin, and collagen IV, but not as well on plastic. Even though their function is not understood, these cell podia could be important for sensing gradients. A migration assay was developed by tilting the microscope 7° to watch KG1a cells migrate uphill. After a fixed time, the cells that had migrated past a defined “finish line” were counted. This assay was used to compare the effects of various cytokines and substrates and showed that the cells actively migrate unless they are preparing to divide. The setup may be used to study asymmetric cell divisions in addition to interactions with stromal cells.Malcolm Moore spoke about stem cell motility, which helps infused cells find their way to the bone marrow. Dr. Moore has used transwells with 3 μm pores to develop an assay, where a human marrow-derived endothelial cell line (BMEC-1) is grown to confluence in the top chamber, and a mouse stromal cell line (MS-5) is layered in the bottom chamber as a source of the chemokine, stroma-derived factor-1 (SDF-1) [
22
]. About 5 × 104 CD34+ cells isolated from peripheral or cord blood are added to the top chamber for 3, 12, 24, or 48 hours, and the cell number, the progenitor cells, and the CAFC content of all the cells that pass through the chamber are enumerated. The migratory cells comprised about 50 to 60% of the input and were mostly BFU-E's. Dr. Moore found that the rate of transit correlated with the expression of CXCR4 receptors on CD34+ CD38− cells. Recombinant SDF-1 (100 ng/mL) also can be placed in the lower chamber to produce chemotaxis, and an antibody to the CXCR4 receptor (12G5) will block this effect. A comparison of cord blood cells and G-CSF–mobilized peripheral blood cells showed that the former were more efficient in migration, 75% vs 40% over 48 hours, respectively. However, the expression of CXCR4 was only slightly higher in cord blood than in adult blood, so it is likely other factors are involved in stem cell migration. The expression of SDF-1 or unique sinusoid adhesion molecules has not been determined in bone marrow. Knockout mice for SDF-1 and CXCR4 are characterized by hematopoiesis in the liver, but not in bone marrow, which supports the role of these molecules in stem cell migration. The effect of cytokines on stem cell motility also was examined. When progenitors were expanded in vitro for 7 to 14 days with FL, KL, and TPO, migration increased twofold. However, if IL-3 (50 ng/mL) was added, committed progenitors were increased, but there was a 30% drop in chemotaxis as compared to controls. The clinical outcome of engraftment may therefore be based on a complex relationship between homing and proliferation of stem cells.Stem cell gene expression
Recent advances in genomics, bioinformatics, and cDNA microarray technologies have provided an opportunity to molecularly define stem cells on the basis of specific genetic markers or unique gene expression patterns. Ihor Lemischka described the analysis of gene expression in hematopoietic stem cells. His laboratory has built a comprehensive database of molecular phenotypes of stem cell and its microenvironment, for both humans and mice. It uses conserved stem cell properties as an approach to identify relevant regulatory molecules. Differential gene expression technologies, such as subtraction hybridization, differential display, and high-density microarrays, were coupled with sophisticated bioinformatics tools to study highly purified stem cells [
23
]. Murine fetal liver (AA4.1+ lin−/lo Sca-1+ c-kit+) cells and murine bone marrow (Rholo or Rhohi plus Thylo lin− Sca-1+ c-kit+) cells were used, along with human bone marrow (CD34+ lin− CD38−) cells. Full-length stem cell cDNA libraries were generated and depleted of sequences from more mature hematopoietic cells. These subtracted libraries were enriched up to 200-fold for uniquely expressed gene products and were analyzed by high-throughput DNA sequencing from the 5′ end. These DNA and derived protein sequences constitute a stem cell database, which contains approximately 4,000 entries to date. Bioinformatic analysis has revealed: 1) homologies to regulatory molecules in developmental pathways, including those of invertebrates, 2) members of related protein families, e.g., cell-surface receptors and DNA-binding factors, and 3) potential roles in biological processes, such as apoptosis and cell signaling. It also has been possible to obtain virtual expression profiles with Expressed Sequence Tag (EST) databases, which are now publicly available for about 20 to 40% of the potential coding regions of the mouse and human genomes. Clusters of unique ESTs have been arrayed at high density (>18,000 unique sequences) on commercially available membranes. Hybridization was performed with cDNA probe populations obtained from Rholo, Rhohi, and lin+ cells purified from murine bone marrow, which represent primitive, intermediate, and mature hematopoietic cells, respectively. The procedures were optimized for as few as 10,000 to 20,000 cells. Comparative hybridizations were performed with the Rholo vs the Rhohi and the Rholo vs lin+ subsets, using total complex cDNA as probes. Multiple differences in gene expression were identified. Although the chips are quantitative to at least fivefold, the sensitivity of the approach can be increased by PCR-based subtraction to generate less complex probe populations, enriched for specifically expressed sequences. The substractions were bidirectional, to obtain divergent probe populations for both primitive and more mature cells. Hybridization of these probes to high-density arrays revealed mutually exclusive patterns and uncovered many more differentially expressed gene products. Venn diagrams of gene expression patterns can be constructed for Rholo sequences not expressed in Rhohi or in lin+ subsets. Linkage to cell cycle genes in yeast and developmental molecules such as wnt 5/10, notch, disheveled, and lunatic fringe in Drosophila was noted.Jan Visser compared the gene expression in CD34+ and CD34−lin−kit+sca+ murine cells [
24
]. He used a variety of assays, such as differential display, RT-PCR, and gene expression fingerprint analysis to identify, extract, and clone specifically expressed cDNAs. About 1,000 stem cells gave 2 to 3000 bands, which represented about 10,000 genes expressed. Long-term repopulating cells (mostly CD34−) were identified by limiting dilution, and more than 2,000 of their expressed genes were screened. Of 22 genes that were unique to long-term repopulators, tissue transglutaminase, type II was expressed exclusively in the earliest stem cells. Further progress depends on cell purity, good directional databases, and the ability to associate genes with the 3′ untranslated regions and alternative splice forms of the RNAs. Dr. Visser also tested about 60 other known genes for expression in stem cells and found that CD3-epsilon was turned off in short-term repopulating cells. Further studies with a panel of such markers are needed to establish clinical utility.An approach to standardization of surrogate stem cell assays
The workshop defined the need for researchers to collaborate and compare the results of molecular, cellular, and animal assays with engraftment data. A working group was formed and held a roundtable discussion of July 30, 1999. As a first step, a practical approach was proposed to standardize the measurement of engraftable human hematopoietic cells.
Differences between three major cell sources—bone marrow, mobilized peripheral blood, and umbilical cord blood—were considered with regard to how they could be assayed to predict clinical outcome. Bone marrow may have a higher content of accessory cells that need to be characterized qualitatively and quantitatively to determine their effect on stem cell proliferation and quiescence. When infused, the donor cells could interact with host stroma variably, depending on factors such as histoincompatibility status or cytokine release following a myeloablative conditioning regimen. Since bone marrow aspirates from living donors are limited in quantity and are frequently diluted by peripheral blood, cadaveric marrow samples were proposed for such studies. However, this material is unsuited for human transplants and may limit correlation of in vitro with in vivo data.
The second PHSC source was human peripheral blood, mobilized by treatment with G-CSF or GM-CSF and harvested by apheresis. Even if normal donors are treated uniformly, their mobilized products are not equivalent, which probably reflects genetic or cyclical differences in the donor. A group of donors should be studied to account for individual variation. Mobilized peripheral blood may contain more mature precursor cells and tends to engraft earlier than bone marrow, but this stem cell source should provide sufficient cells for testing.
The third stem cell source considered was umbilical cord blood, which is readily available. Because samples are small, they cannot be distributed widely and specimens may have to be pooled.
At the 1998 meeting, the variation in surrogate human stem cell assay endpoints as compared to animal models was mentioned. Not only are there large variations among individuals, but significant differences are noted when samples from the same person are assayed. Since cell cycle status and other factors are not known at the time of collection, at least 8 to 12 donors in each category may be needed to compare different assay systems. It was felt that five large, individual samples of cadaveric marrow, five samples of mobilized peripheral blood, and possibly five large pools of cord blood could be obtained initially from both clinical and commercial sources. These samples would be frozen in DMSO and stored centrally. This repository later could be expanded to include fetal liver, purified stem, and accessory cell populations. Standardized reagents (e.g., antibodies, GMP-quality biologics) should be utilized for the performance of assays in the participating laboratories. After distribution, each cell aliquot would be checked for viability before use.
Although assays may differ between laboratories, an important goal would be to correlate the results at each site with transplant outcome. Xenogenic transplant models in mice and fetal sheep would be included because competitive repopulation and limiting dilution studies are possible. Statistical analysis of the data would be performed to try to identify the most predictive assays. At a later stage, the best assays could be standardized. Training courses, an interactive Web site, or the exchange of laboratory personnel could be provided to instruct investigators in the performance of the assays of choice.
Acknowledgements
We would like to thank Barbara Alving and Helena Mishoe for review of the manuscript and suggestions.
The opinions expressed in this paper are not those of the N.I.H. or the U.S. government.
Appendix.
The speakers for the first meeting on September 8–9, 1998, were: Peter Quesenberry, M.D., from the University of Massachusetts Cancer Center in Worcester, MA, USA; and Ivan Bertoncello, Ph.D., of the Peter MacCallum Cancer Institute, Melbourne, Australia (Co-Chairpersons); David Bodine, Ph.D., from the National Human Genome Research Institute in Bethesda, MD., USA; John Dick, Ph.D., of the Hospital for Sick Children in Toronto, Canada; David E. Harrison, Ph.D., from the Jackson Laboratory in Bar Harbor, ME., USA; Anthony D. Ho, M.D., Ph.D., from the Medizinische Klinik und Poliklinik V in Heidelberg, Germany; Ruud Hulspas, Ph.D., from the University of Massachusetts Medical Center in Worcester, MA, USA; Ihor Lemischka, Ph.D., from Princeton University in Princeton, NJ, USA; Jeffrey Moore, Ph.D., from Phylogix in Scarborough, ME., USA; Malcolm Moore, Ph.D., from Memorial Sloan-Kettering Cancer Center in New York, NY, USA; Christa Muller-Sieburg, Ph.D., from the Sidney Kimmel Cancer Center in San Diego, CA, USA; Bernhard Palsson, Ph.D., from the University of California in La Jolla, CA, USA; Rob Ploemacher, Ph.D., from the Erasmus University in Rotterdam, the Netherlands; Beverly Torok-Storb, Ph.D., of the Fred Hutchinson Cancer Research Center in Seattle, WA, USA; Catherine Verfaillie, M.D., of the University of Minnesota Hospital Center in Minneapolis, MN, USA; Jan Visser, Ph.D., from the New York Blood Center in New York, NY, USA; and Esmail Zanjani, Ph.D., from the Veterans Affairs Medical Center, Reno, NV, USA.
A working group met on July 30, 1999, and included Drs. Stephen Bartelmez, Ph.D., from the Seattle Biomedical Research Institute in Seattle, WA, USA; Ian McNiece, Ph.D., from the University of Colorado Health Sciences Center in Denver, CO, USA; Makio Ogawa, M.D., Ph.D., from the Medial University of South Carolina, Charleston, SC, USA; Robertson Parkman, M.D., from the Children's Hospital of Los Angeles, Los Angeles, CA, USA; and Gerald Spangrude, Ph.D., from the University of Utah Medical Center, Salt Lake City, UT, USA; as well as Drs. Quesenberry, Bertoncello, Bodine, Hulspas, Lemischka, M. Moore, Muller-Sieburg, Torok-Storb, Verfaillie, Visser, and Zanjani from the previous meeting.
References
- Autologous transplantation in cancer therapy.Prog Hematol. 1981; 12: 1-23
- Surrogate assays for hematopoietic stem cell activity.in: Garland J. Quesenberry P. Hilton D. Colony–stimulating factors. 2nd ed. Marcel Dekker Inc, New York1997: 35-47
- Specific flow cytometric phenotypes can accurately quantitate hemopoietic stem cells.Blood. 1999; 94;10 (#2062][abstr): 463a
- Development and aging of primitive hematopoietic stem cells in BALB/cBy mice.Exp Hematol. 1999; 27: 928-935
- Purification of primitive human hematopoietic cells capable of repopulating immune-deficient mice.Proc Natl Acad Sci USA. 1997; 94: 5320-5325
- Human bone marrow CD34− cells engraft in vivo and undergo multilineage expression that includes giving rise to CD34+ cells.Exp Hematol. 1998; 26: 353-360
- Genetic control of the frequency of hematopoietic stem cells in mice.J Exp Med. 1996; 183: 1141-1150
- The fluctuating phenotype of the lymphohematopoietic stem cell with cell cycle transit.J Exp Med. 1998; 188: 393-398
- Improved amphotropic retrovirus-mediated gene transfer into hematopoietic stem cells.Ann N Y Acad Sci. 1998; 850: 139-150
- Stem cells.Baillieres Clin Haematol. 1997; 10: 429-444
- The myeloid-lymphoid initiating cell (ML-IC) assay assesses the fate of multipotent human progenitors in vitro.Blood. 1999; 93: 3750-3756
- Dissecting the marrow microenvironment.Ann N Y Acad Sci. 1999; 872: 164-170
Bertoncello I (1992) Status of high proliferative potential colony-forming cells in the hematopoietic stem cell hierarchy. In: Müller-Sieburg C, Torok-Storb B, Visser J, Storb R. (eds.) Hematopoietic Stem Cells: Animal Models and Human Transplantation Berlin, Heidelberg: Springer-Verlag. Curr Topics Microbiol Immunol 177:83–94
- The relationship between different high proliferative potential colony forming cells (HPP-CFC) in murine bone marrow.Exp Hematol. 1994; 22: 432-440
- cDNA cloning of FRIL, a lectin from Dolichos lablab, that preserves hematopoietic progenitors in suspension culture.Proc Natl Acad Sci USA. 1999; 96: 646-650
- Symmetry of initial cell divisions among primitive hematopoietic progenitor cells is independent of ontogenic age and regulatory molecules.Blood. 1999; 94: 2595-2604
- Functional isolation and characterization of human hematopoietic stem cells.Science. 1995; 267: 104-108
- Homing of fluorescently labeled murine hematopoietic stem cells.Exp Hematol. 1996; 24: 129-140
- Organ-selective homing defines engraftment kinetics of murine hematopoietic stem cells and is compromised by ex vivo expansion.Blood. 1999; 93: 1557-1566
- Hematopoietic stem cell tracking in vivo.Blood. 1999; 93: 1916-1921
- Two new pseudopod morphologies displayed by the human hematopoietic KG1a progenitor cell line and by primary human CD34(+) cells.Blood. 1998; 92: 3616-3623
- Regulation of transendothelial migration of hematopoietic progenitor cells.Ann N Y Acad Sci. 1999; 872 (discussion 185–186): 176-185
- Searching for stem cell regulatory molecules. Some general thoughts and possible approaches.Ann N Y Acad Sci. 1999; 872 (discussion 287–288): 274-287
Hendrikx PJ, Ivanova NB, Debnath G, Radice A, Bartelmez SH, Visser JWM (1999) Comparison of stem cell purification protocols and isolation of a new stem cell marker gene. Blood 94(10)Suppl. 1, p. 682a [abstr. #3021]
Article info
Publication history
Accepted:
March 27,
2000
Received in revised form:
March 15,
2000
Received:
November 8,
1999
Identification
Copyright
© 2000 International Society for Experimental Hematology. Published by Elsevier Inc.
