Department of Molecular Microbiology and Immunology, Oregon Health and Sciences University, Portland, Oregon, USA
Department of Molecular Microbiology and Immunology, Oregon Health and Sciences University, 3181 SW Sam Jackson Park Road, L220, Portland, Oregon 97201-3098, USA. Telephone: 503-494-5312; Fax: 503-494-6862
There is limited understanding of CD34− hematopoietic cells and the linkage between CD34 antigen expression and cell proliferation. In this study, early CD34− CD38− LIN− (CD34−) cells were purified from mobilized adult peripheral blood and carefully analyzed in vitro for growth and modulation of CD34. Mobilized CD34+CD38− LIN− (CD34+) cells were used for comparison. Expression of CD34, CD38, and LIN antigens was determined, and proliferative responses were assessed with PKH tracking dye, expression of Ki67 antigen, and uptake of pyronin Y. Suspension cultures of adult CD34− cells generated CD34+ cells and progenitors for >8 weeks. Stromal cultures demonstrated the presence of long-term culture-initiating cells within the CD34− fraction. While CD34− cells were slower to initiate growth than the CD34+cells were, no significant difference in hematopoietic cell output was found. Upon cultivation of CD34− cells, CD34 antigen appeared within 48 hours but was restricted to those cells that had initiated growth. Surprisingly, CD34+ precursors lost CD34 expression in culture if they remained in G0 for more than 2 days. Those cells later regained expression of CD34 antigen upon initiation of growth. Comparison of cells that did or did not rapidly modulate CD34 antigen revealed no differences in long-term growth potential. In conclusion, in vitro expression of CD34 by CD34− and CD34+ populations is tightly linked to cellular proliferation. In this culture system, expression of CD34 antigen by LIN− cells constitutes an early hallmark of growth. Measurement of CD34 expression by LIN− cells in expansion culture underestimates the total content of hematopoietic cells.
CD34 antigen, a transmembrane glycoprotein, is a member of the sialomucin family . The antigen was initially discovered on a myeloblastic leukemia cell line . While the precise function of CD34 is unknown, involvement in either cytoadhesion or differentiation (or both) has been suggested [1,3]. Lacking its own kinase activity, transduction of signals from CD34 may involve the intracellular adapter protein CrkL . Due to its expression in early hematopoietic development, the antigen has become a valuable tool for identification and purification of human hematopoietic cells [5–8]. Thus, it was surprising to learn in 1996 that CD34 expression in mice was not mandatory for survival  and that murine hematopoiesis could be reconstituted by CD34− LIN− cells [10–14]. CD34− hematopoietic cells were subsequently identified in other species [15–17]. Engraftment activities were demonstrated in CD34− fractions of human bone marrow, mobilized peripheral blood progenitor cell (PBPC) products, and umbilical cord blood (UCB) [18–21; reviewed in 22–24].
In vitro culture systems have been used to investigate human CD34− cells and their modulation of CD34 antigen. Those experiments proved difficult because of the limited growth potential of CD34− UCB and marrow cells in stroma-free culture [21, 23, 25–29]. Nonetheless, data suggested that colony-forming activity and CD34 antigen expression were present after 7–10 days'cultivation [17, 26, 29, 30]. Experiments in murine systems showed that CD34− stem cells expressed CD34 after mobilization or treatment with 5-fluorouracil [11, 31–33]. Taken together, those results suggested an association between cellular activation (cytokine responsiveness) and upregulation of CD34. However, the exact nature of “cellular activation” has been questioned , and a direct link between proliferation and CD34 expression has not been shown. Other questions remain. The kinetics of upregulation are not well defined, and it is uncertain whether CD34− cells can divide prior to upregulating CD34 . Furthermore, it is not clear if all cells or only growing cells in a CD34− culture upregulate CD34. Last, relatively little information is available about the properties of CD34− cells from PBPC products . Since PBPC represents an important source of stem cells for transplantation, characterization of circulating adult CD34− cells is important.
For this report, we chose to study the CD34− CD38− LIN− (CD34− ) subset, since this is a particularly primitive hematopoietic fraction . Mobilized CD34− and CD34+CD38− LIN− (CD34+) cells were purified, and their proliferative potentials were characterized. CD34 antigen modulation was examined, and the relationship of CD34 expression to cell proliferation and growth potential were directly determined. Here we report that both CD34− and CD34+ cells modulate CD34 antigen expression shortly after initiation of culture. In both cases, modulation of CD34 reflects the proliferative status of LIN− cells in culture.
Materials and Methods
PBPCs were collected following treatment with granulocyte colony-stimulating factor with or without chemotherapy. Samples were held in liquid nitrogen until selected for analysis. Samples were obtained with approval of the American Red Cross Committee for the Protection of Human Subjects.
Flow Cytometric Analysis and Isolation of Primitive Hematopoietic Progenitors
This laboratory's method for lineage depletion and flow cytometric sorting has been described in detail [36–39]. In brief, thawed PBPCs were enriched for LIN− cells using immunomagnetic separation (Stem Cell Technologies Inc., Vancouver, Canada) targeted against glycophorin A, CD2, CD3, CD14, CD16, CD19, CD24, CD56, CD66b plus added CD41. The LIN− cell-enriched fraction was labeled with CD34 PE-Cy5 (clone 581, class III, Coulter-Immunotech Co., Miami, FL), fluorescein-6-isothiocyanate (FITC)–conjugated antibodies against LIN antigens (CD3, CD14, CD16, CD19, CD20, and CD56; Becton Dickinson Immunocytometry Systems, San Jose, CA), as well as CD42a-FITC and CD38 APC (Becton Dickinson) or isotypic control antibodies (from the corresponding vendor). Cells were sorted using a Coulter ELITE ESP unit, using propidium iodide (PI) as a viability marker (if PKH26 were not present). For data analysis, at least 10,000 events were collected. Reanalysis of sorted cells indicated >97% purity.
Sorting and Analysis of Cells Stained with Tracking Dyes
Suspensions enriched for LIN− cells were stained with PKH26 or PKH67 (Sigma Immunochemicals, St. Louis, MO) . PKHBRIGHT cells were sorted into CD34− and CD34+ fractions. An aliquot of unsorted PKH-stained cells was treated with 1.0% paraformaldehyde and then stored at 4°C for subsequent reference of initial PKH fluorescence. Sorted cells were placed in suspension culture for the indicated period of time, and samples were collected for flow cytometric determination of PKH fluorescence and immunophenotype after restaining. In some cases, cultured cells were resorted and returned to culture. Where indicated, cellular RNA was stained with pyronin Y, as described by Ladd et al.  including the preincubation step with Hoechst 33342, which blocks pyronin Y staining of DNA. This stain distinguishes quiescent G0 cells (PYLOW) from actively cycling cells (PYHIGH) [40,41]. Alternatively, cells were permeabilized and stained with Ki67 antibody following manufacturer's instructions (Becton Dickinson).
Suspension and Stromal Cultures
For stroma-free, serum-free suspension cultures, sorted cells were incubated in 24 well trays or T-25 flasks (Corning Inc., Corning, NY) containing 0.5 to 1.0 mL/well or 10 mL/flask serum-free expansion medium (SFEM; Stem Cell Technologies, Inc.) containing penicillin or streptomycin, low-density lipoproteins (Sigma-Aldrich, St. Louis, MO), recombinant human (rh) stem cell factor (SCF; rhSCF 50 ng/mL), Flt3 ligand (rhFL; 50 ng/mL), interleukin-3 (rhIL-3; 12.5 ng/mL), interleukin-6 (rhIL-6; 10 ng/mL), and thrombopoietin (rhTPO; 100 ng/mL). SCF was a generous gift from Amgen (Thousand Oaks, CA); FL, IL-3, and IL-6 were generously provided by Immunex Corporation (Seattle, WA). TPO was purchased from Genzyme Corporation (Cambridge, MA). CD34− cultures were initiated with an average of 82,000 cells (range: 1,800–183,000, depending on sort yield) and CD34+ cultures, with an average of 53,000 cells (range: 900–154,000). In stromal cultures, the murine cell line S-17 (known to support myeloid and lymphoid differentiation) was used. In this case, the medium consisted of Iscove's modified Dulbecco's medium (IMDM), 30% fetal bovine serum (FBS), 10% bovine serum albumin (BSA), 10− 4 M mercaptoethanol, 1% penicillin or streptomycin, 2 mM l-glutamine, 10− 6 M hydrocortisone hemisuccinate, and the same cytokines used in suspension culture. Cytokines were replenished once per week during demi-depopulation of stromal cultures and two to three times per week in liquid cultures. In rapidly expanding suspension cultures, cells were removed weekly to return the cell concentration to ∼105 cells/mL. Cultures were maintained at 37°C in a humidified atmosphere with 5% CO2.
Colony Formation Assay
Committed hematopoietic progenitors from stromal cultures were quantitated in standard semisolid cultures in triplicate, using 1 mL of Methocult GF+ (#H4435, Stem Cell Technologies, Inc.), which consists of 1% methylcellulose in IMDM, 30% FBS, 1% BSA, 2 mM l-glutamine, 10− 4 M 2-mercaptoethanol, 50 ng/mL SCF, 20 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF), 20 ng/mL granulocyte colony-stimulating factor (G-CSF), 20 ng/mL IL-3, 20 ng/mL rhIL-6, and 3 U/mL erythropoietin. Nonadherent cells from stromal cultures were washed twice with IMDM/2% FBS before plating. Progenitors from serum-free suspension cultures were quantitated in triplicate using 1 mL of serum-free Methocult SFbit (#H4436, Stem Cell Technologies, Inc.) containing 1% methylcellulose in IMDM, 1% BSA, 2 mM l-glutamine, 10− 4 M 2-mercap-toethanol, 10 μg/mL bovine pancreatic insulin, 200 μg/mL human transferriniron saturated plus the same six cytokines contained in methylcell H4435. Cells from liquid culture were washed twice with SFEM before assay. Plates were scored for erythroid, granulocyte-macrophage, and mixed-lineage colonies after culturing for 14 days at 37°C, 5% CO2.
RT/PCR Analysis of CD34 Gene Expression
Cells were lysed in dimethyl pyrocarbonate–treated water containing 0.8% NP-40, 5 mM dithiothreitol (DTT), 1 unit ribonuclease inhibitor (RNasin), and 1 μg transfer RNA (Sigma). Supernatants from the lysates were digested with 1 U Deoxyribonuclease. Seven-microliter aliquots were reverse transcribed with GeneAmp RNA PCR kits (Perkin Elmer, Rockford, IL) and random hexamers (2.5 μM) as primers, RNase inhibitor (20 units) and murine leukemia virus reverse transcriptase (50 units). Amplification was conducted with the PCR core kit (Perkin Elmer) using 35 cycles of 1 minute denaturation at 94°C, 1 minute annealing at 65°C, and 2 minutes synthesis at 72°C. Oligonucleotides for CD34 gene expression analysis were sense primer 5′-AGGTATGCTCCCTGCTCCTGGCCC-3′ and antisense primer 5′-AAGAACAGCCTCTGAGGTGTGTGC-3′. Sense primer 5′-GGTCGGAGTCAACGGATTTGG-3′ and antisense primer 5′-TGTGGGCCAT GAGGTCCACCAC-3′ were used as a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) reference. As a negative control, reverse transcriptase was omitted. Twenty microliters of the amplified products were run in 2% agarose gels and stained with ethidiumbromide dye.
Analysis of Data
In stroma-free and stroma-based cultures, the starting cell number was normalized to 10,000 cells to allow pooling of data between experiments. In stroma-free cultures (but not in stroma-based cultures) the weekly total viable cell count was corrected for periodic removal of cells at the time of medium change. To determine the proportion of cells having undergone 0, 1…. n cell divisions, PKH26 profiles were analyzed with ModFit LT software (Verity Software House, Topsham, ME). Significance of differences was determined with paired t-tests or Wilcoxon Signed Rank test, as indicated (Sigma Stat [SMSS Inc., Chicago, IL] or GB Stat [Dynamic Microsystems, Inc., Silver Spring, MD]).
Isolation of CD34− Hematopoietic Cells
The frequencies of CD34− CD38− LIN− (CD34− ) and CD34+CD38− LIN− (CD34+) cells in PBPC were too low to be measured accurately. However, they were readily detected in the LIN-depleted fraction where CD34− cells were significantly more numerous than CD34+ cells (4.6% versus 1.5%, p < .009) (Table 1). These data compare favorably with values of 6% versus 3% published recently .
Table Table 1.. Frequency (%) of primitive hematopoietic cells in PBPC after depletion of LIN+ cells
Samples were analyzed after immunomagnetic depletion of LIN+ cells.
Frequencies of each population are expressed as a percentage of cells in the lineage-depleted samples.
aSignificantly greater than the corresponding CD34+ fraction, p < .009.
Figure 1 illustrates the purification of CD34− and CD34+ fractions by flow cytometric sorting. To achieve high purity, narrow gates were used to capture cells with very low expression of CD34 and LIN. Cells of interest were low in side scatter, low to moderate in forward scatter, LIN−, in the lowest 10% in CD38 fluorescence, and either CD34BRIGHT or CD34− (gates D and F, Fig. 1, Panel II). Cells lacking CD34 expression were defined as those with less than the modal brightness of the PE-Cy5 isotypic control. The higher frequency of CD34− than CD34+ cells is apparent in Figure 1. Reanalysis of the CD34− fraction showed that more than 97% of the cells were CD34− . CD34− cells from PBPC were noticeably smaller than corresponding CD34+ cells . Data presented below (Figures 3, 5, 6, Tables 2, 3) and RT-PCR analyses (not shown) confirmed that CD34+ and CD34− fractions were well separated.
Table Table 2.. Phenotypic analysis of PKHBRIGHT and PKHINTERMED fractions (% of cells in each column with indicated phenotype) in CD34− and CD34+ cultures after short-term incubation
CD34− cells and CD34+ cells were cultured for 2–3 days, after which PKHBRIGHT and PKHINTERMED populations were analyzed for immunophenotype. PKHBRIGHT identifies cells that have undergone little or no growth (cols. 1 and 3). PKHINTERMED cells have undergone several divisions (cols. 2 and 4). Lineage commitment was higher in PKHINTERMED than PKHBRIGHT cells (p < .02 in both cases). Data averaged from seven experiments. Significantly greater than (*) or less than (**) column 2 (p < .02). Significantly greater than (#) or less than (##) column 4 (p < .02). Significantly greater than (&) column 3 (p < .02). Not significantly different (∥) from column 3 (p = .15). Data analyzed with Wilcoxon signed rank test. Only 32% of PKHINTERMED cells in the CD34− culture expressed CD34 (col. 2). This was most likely due to rapid differentiation of growing cells and incomplete resolution of overlapping PKHBRIGHT and PKHINTERMED populations after short incubation (Fig. 3).
92 ± 8* ∥
65 ± 18
81 ± 14
53 ± 41
87 ± 7*
20 ± 17
60 ± 11#
3 ± 2
CD34− CD38− LIN−
70 ± 11* &
11 ± 10
47 ± 8#
2 ± 2
7 ± 7**
32 ± 16
20 ± 15
47 ± 42
4 ± 4**
55 ± 32
11 ± 12##
59 ± 29
Table Table 3.. CD34− cells initiate growth more slowly than do CD34+ cells
PKH-labeled CD34− cells and CD34+ cells were incubated in serum-free medium containing five cytokines. After 2 or 3 days incubation, PKH fluorescence profiles were obtained. Data were analyzed with ModFit software to determine the proportion of cells that had undergone less than one division or greater than three divisions. Table displays mean values ± SD obtained in five trials. p values (Wilcoxon signed rank) compare the significance of differences between the CD34− and CD34+ fractions after 2 or 3 days.
Abbreviation: SD, standard deviation.
<1 division (% total cells)
>3 divisions (% total cells)
<1 division (% total cells)
>3 divisions (% total cells)
74 ± 21
16 ± 18
45 ± 27
39 ± 18
(p < .05)
(p < .05)
(p < .02)
(p < .01)
34 ± 25
36 ± 29
14 ± 17
65 ± 13
In Vitro Proliferative Activity of CD34− Cells
Colony-forming cell (CFC) activity of CD34− and CD34+ fractions was assessed in serum-free methylcellulose containing six cytokines. No significant differences in CFC frequencies could be detected between the CD34− fraction and the CD34+ fraction (7% + 7% and 10% + 11%, respectively; n = 11, paired t-test, p = .29). The distribution of progenitor lineages in the CD34− fraction (55% colony-forming units-granulocyte-macrophage (CFU-GM), 37% BFUE, and 9% colony-forming units-mixed (CFU-Mix) was similar to that observed in the CD34+ population (49%, 36%, and 13%, respectively).
Suspension cultures of PBPC CD34− and CD34+ fractions proved very active, as both expanded approximately 10,000-fold (Fig. 2A). Over 8 weeks, the fractions generated similar numbers of CD34+ cells (Fig. 2D). CD34+ cells appeared in CD34− cultures within just 2 days. After 6 days of cultivation, CD34+ cell frequencies in CD34− and CD34+cul-tures were similar (Fig. 2E). Each CD34− cell produced 630 ± 770 CD34+ cells, while each CD34+ cell produced 1,600 ± 1,800 CD34+ cells, a difference that did not reach significance (n = 5, p = .13, paired t-test).
The prolonged generation of CD34+ cells in CD34− suspension cultures was confirmed in progenitor assays. In CD34− cultures, output of both CFU-GM and CD34+ cells peaked after 7–8 weeks (Fig. 2B, 2D). Similar numbers of progenitors were generated per initial cell (51 ± 63 CFC in CD34− suspensions versus 47 ± 59 CFC in CD34+ suspensions). In stroma-based cultures, both populations generated CFC for more than 80 days (Fig. 2C), demonstrating the presence of long-term culture-initiating cells (LTCIC) and extended LTCIC. Collectively, these data demonstrated that CD34− and CD34+ fractions from adult PBPC collections exhibited similar proliferative capabilities when cultured in vitro.
Modulation of CD34 Antigen Expression
The ready growth of CD34− PBPCs provided an opportunity to examine the relationship between cell proliferation and expression of CD34 antigen. In these studies, attention was focused on the response of the most primitive (LIN− ) cells in the culture. PKH26- or PKH67-loaded CD34− and CD34+ cells were cultured for 48–72 hours, harvested, and restained with antibodies. Cells were analyzed for proliferation and antigen expression. Cells that had undergone little or no growth were PKHBRIGHT, while those that had completed several division cycles were PKHINTERMED.
Figure 3 (top) shows typical PKH profiles of LIN− cells in CD34− and CD34+ cultures. Gate R1 delineates PKHBRIGHT (day zero) fluorescence, while R2 marks the PKHINTERMED region. R2 was placed so that cells with modal brightness in R1 would have to divide twice to reach R2. In the CD34− culture (central panels), only 6% of non-proliferating LIN− PKHBRIGHT cells had upregulated CD34, while 10 times as many (59%) expressed CD34 in the proliferating LIN− PKHINTERMED fraction. For comparison, the bottom panels of Figure 3 present the companion CD34+ culture. In the proliferating LIN− PKHINTERMED fraction, 89% continued to express CD34. Surprisingly, 37% of the undifferentiated LIN− PKHBRIGHT cells in the CD34+ culture lost CD34 expression.
These data suggest a correlation between CD34 expression and cell proliferation. That inference is strengthened by Table 2, which summarizes data from seven similar trials. For example, in CD34− cultures, 32% of growing cells expressed CD34 (column 2), compared with only 7% of PKHBRIGHT cells (column 1), and these values were significantly different (p <.02). Conversely, both the CD34− LIN− fraction and the CD34− CD38− LIN− fraction were enriched in the PKHBRIGHT population relative to the PKHINTERMED population (column 1 versus 2, both fractions p < .02). A similar association between CD34 expression and proliferation was observed in CD34+ cultures. CD34 expression was higher in PKHINTERMED cells than in PKHBRIGHT cells (column 4 versus 3), while CD34− LIN− cells and CD34− cells were more highly enriched in the PKHBRIGHT fraction than in the PKHINTERMED fraction (column 3 versus 4). There was no difference in the proportions of growing cells in CD34− and CD34+ cultures that expressed CD34 (32% versus 47%). In summary, primitive CD34− cells were enriched in the PKHBRIGHT fractions of both CD34− and CD34+cultures; in contrast, CD34+ cells were enriched in the PKHINTERMED fractions.
From the preceding data, it was not possible to state unequivocally that PKHBRIGHT cells were quiescent. To resolve that question, cells were stained with pyronin Y (PY). Quiescent cells are PYLOW, whereas cycling cells are PYHIGH [40,41]. This was confirmed with fresh (inherently quiescent) CD34+ CD38− PBPCs, 75% + 11% of which were PYLOW (n = 5). In the following experiments, PKH-stained, CD34+CD38− PBPC cells were cultured for 4–6 days. Upon flow analysis, four populations were identified: PKHBRIGHTCD34+ (Fraction A), PKHBRIGHTCD34− (Fraction B), PKHINTERMEDCD34+ (Fraction C), and PKHINTERMEDCD34− (Fraction D). The analysis of those populations for PY content and CD38 expression is presented in Figure 4. In the PKHBRIGHT fractions (Fig. 4A, B), 80%–90% of cells were PYLOW. In contrast, 70%–80% of PKHINTERMED cells were PYHIGH (Fig. 4C, D). This confirmed that PKHBRIGHT cells were quiescent, while PKHINTERMED cells were actively cycling. It is notable that PKHBRIGHT cells that expressed CD34 were predominately CD38+, while PKHBRIGHTCD34− cells were primarily CD38− (see Discussion).
The proliferative status of cultured cells was also determined by assessing their expression of Ki67 antigen. Ki67 is absent from G0 cells but present in proliferating cells in all phases of the cell cycle [41,42]. Preliminary studies showed that staining with Ki67 and PY yielded comparable results (data not shown). Starting with freshly sorted CD34− and CD34+ fractions, cells were examined for expression of Ki67 and CD34. Analysis confirmed clean separation of CD34− and CD34+ populations. Both populations were Ki67LOW (Fig. 5A). CD34− and CD34+ fractions were cultured for 42 hours and analyzed for antigen expression and proliferative status. In the CD34− culture (Fig. 5B), most LIN− cells were CD34LOW (R2) or CD34− (R1). In the CD34+ culture (Fig. 5C), the bulk of LIN− cells were still CD34BRIGHT (R3); however, early dimming of CD34 in LIN− cells was apparent. In both CD34− and CD34+ cultures, Ki67 and PY analysis shows that CD34− cells (R1) were quiescent, while CD34BRIGHT cells (R3) were proliferating. Interestingly, in the CD34LOW fractions (R2), low levels of PY staining and Ki67 expression were seen. These data confirmed the correlation between CD34 antigen expression and proliferation in both CD34+ and CD34− cultures.
Initial CD34 Phenotype and the Onset of Cell Proliferation
Figures 3, 5, and 6 suggest that, on average, CD34− cells initiated growth more slowly than did CD34+ cells. To test the generality of that observation, PKH26 profiles from CD34− and CD34+ cultures were analyzed for the proportion of cells that were PKHBRIGHT and those that had undergone greater than three division cycles. After 2 days of cultivation, the proportion of cells that had undergone at least three divisions was twice as great in CD34+ cultures as in CD34− cultures (p < .05; Table 3). By day 3, the PKHBRIGHT fraction constituted only 14% of the CD34+ culture, compared with 45% of the CD34− suspension (p < .02). More rapid outgrowth of the CD34+ population is not attributable to binding of antibody to CD34 antigen during isolation, since cell cycle parameters are unchanged by such treatment . It was not determined whether the kinetic differences could be eliminated by other cytokine combinations. It was previously shown that CD34− UCB cells were slower than CD34+ cells to proliferate in a stromal system .
Modulation of CD34 Expression: Relation to Growth Potential
Finally, we asked if cells that rapidly modulate CD34 antigen in culture might differ from those that retain their original phenotype. As one approach, we studied the possibility that modulating and nonmodulating cells might differ in their capacity for long-term proliferation. Since it was important to minimize confounding parameters (differentiation and growth-related loss of proliferative potential), analysis was limited to LIN− PKHBRIGHT cells. In a typical trial, CD34− and CD34+ cells were cultured for 3 days, then harvested and sorted for LIN− CD38− PKHBRIGHT subpopulations differing in CD34 expression. In the CD34− culture, two populations were resolved (Fig. 6, Panel I): B1, which remained CD34−, and B2, which had upregulated CD34 expression. Two fractions were isolated from the CD34+ culture (Panel II): B3, which had downregulated CD34 expression, and B4, which remained CD34+. From preceding considerations (Figs. 3, 5), it was anticipated that fraction B2 cells would be rare. Indeed, B2 represented only 1% of PKHBRIGHT LIN− cells. Nevertheless, sufficient cells were obtained for analysis.
Fractions B1–B4, quiescent and LIN−, were resuspended in serum-free culture to assess in vitro proliferative potential. Average expansion of the fractions ranged from 350,000-fold to 730,000-fold, with no significant differences among them (n = 4). The kinetics and total number of CD34+ cells and CFC generated were similar (Fig. 7). Thus, neither rapid up-modulation nor down-modulation of CD34 antigen distinguished subpopulations that differed in growth capacity measured in vitro.
The nature of CD34− hematopoietic cells is in dispute. Most studies have focused on the engraftment activity of CD34− cells from murine bone marrow and human UCB. Those data led many [13, 18, 21, 23, 26, 29, 43, 44] but not all [17, 45–47] researchers to conclude that CD34− cells are more primitive than CD34+ cells. Others have suggested that CD34 expression reflects either “activation” status or developmental change, or both [5, 11, 31–34, 48–50].
Analysis of human CD34− cells from fetal tissue, UCB, and bone marrow has been impeded by the generally poor growth achieved in stroma-free culture [16, 17, 19–21, 23, 25–29]. The present report describes a stroma-free, serum-free suspension culture in which CD34− PBPC generated hematopoietic cells for 8 weeks (Fig. 2). When CD34− and CD34+ PBPCs were compared, no significant differences in progenitor content or long-term growth in culture were observed. The prolonged and robust growth of CD34− cells observed here differs from previous reports. The difference might be due to the use of adult-mobilized PBPC, use of the CD34− subset, reliance on serum-free medium, or the combination of cytokines (see below). Interestingly, in further experiments not shown here, CD34− UCB cells grew poorly under the same conditions: cloning efficiency was low, and growth in suspension was marginal, confirming earlier work .
The proliferative activity of CD34− cultures cannot be ascribed to contamination by CD34+ cells. Reanalysis demonstrated the CD34− fraction contained <3% CD34+ cells. If contaminating CD34+ cells had been responsible for growth of the CD34− fraction, hematopoietic output per CD34− cell would have been just a small percent of that seen in the CD34+ fraction. That clearly was not the case. In addition, CD34+ cells appeared very rapidly in the CD34− fraction, too fast to be explained by low-level CD34+ cell contamination. Also, CD34− and CD34+ populations were clearly different in cell size and their kinetics of growth initiation (Figs. 3, 6; Table 3). When essentially the same cell-separation methods were applied to UCB, the resulting CD34− and CD34+ fractions differed markedly in growth characteristics (data not shown), and these are precisely the results that were expected from earlier studies. Finally, in three trials, reverse transcription-polymerase chain reaction (RT-PCR) analyses of PBPC CD34− cells did not detect CD34 mRNA.
While it has been known that CD34− marrow and UCB cells up-modulate CD34 [10, 11, 17, 18, 21, 26, 27, 30–35], many questions remain (see Introduction). For example, a link between expression of CD34 antigen and cell cycling has been proposed but has not yet been definitively shown. In fact, it was suggested that activation may differ from cell cycling [34,35]. Furthermore, stem cells' down modulation of CD34 antigen has been reported [11, 31–35], but samples were obtained months after transplantation, making the relationship to cell cycling difficult to assess. Our finding that mobilized adult CD34− cells proliferate vigorously allowed us to study CD34− precursors in detail. In the present work, experiments showed that cultured CD34− cells upregulate CD34 antigen expression in as little as 42 hours. The rapidity of this process has not been appreciated. Whether rapid surface expression is due to a pool of cytoplasmic antigen is not known. CD34 expression increased as CD34− cells shifted from quiescence to proliferation, as measured by PKH (Fig. 3), PY, and Ki67 (Fig. 5). Cells expressing CD34 at low levels (Fig. 5B, region R2) showed low proliferative activity. Within the resolution of these methods, upregulation of CD34 occurred near the onset of proliferation. It seems unlikely that CD34− cells can grow while maintaining a CD34− phenotype.
With regard to CD34+ cells, CD34 antigen was lost by LIN− cells when quiescence extended beyond 2–3 days in culture (Figs. 3, 5, 6; Table 2). Loss of CD34 antigen expression was not due to maturation (the cells were uncommitted) or cell deterioration (the cells were functional [Figs. 6, 7]). We believe this is the first demonstration that primitive CD34+ cells downregulate CD34 expression in response to (or because of) extended quiescence in culture. After down-modulation of CD34 by quiescent CD34+ cells, cells retained the capacity to re-express CD34 upon initiation of growth (Figs. 6, 7). CD34+ cells retained CD34 expression in culture if they initiated growth quickly or expressed CD38 (Fig. 4A; see below). It is not clear whether CD34 expression is a secondary result of cells' preparation for cycling, or whether CD34 expression is directly involved in the onset of growth. It was recently reported that quiescent mast cells upregulated CD34 as cells approached S phase . Note that the relationship between CD34 antigen expression and cell proliferation differs for cultured cells and circulating cells. That is, both CD34− and CD34+ cells are quiescent in circulation. After 2 days' cultivation, a positive correlation between CD34 expression and proliferation becomes apparent. The reason for this difference is not clear. It is possible that conditions within the circulation either do not support or actively inhibit cycling.
Further insights into CD34 expression patterns were made in Figure 4. When CD34+CD38− cultures were examined after 4–6 days, some quiescent cells expressed CD34 while others did not (Fig. 4A, B). Interestingly, the quiescent cells that expressed CD34 were CD38+, whereas those that had lost CD34 were CD38− . Although CD38 has been often considered a marker of hematopoietic differentiation, it also has been linked to regulation of cell cycle progression, growth enhancement, and activation of calcium flux [51,53]. Thus, PKHBRIGHT PYLOW cells that expressed CD34 and CD38 may have been at an early stage of activation. In some experiments, CD38 expression preceded expression of CD34 by quiescent LIN− cells (Fig. 6, Panel I). Together, the data suggest that expression of CD34 and CD38 by primitive cells may be early indicators of cellular activation or proliferation (or both). These findings, obtained in vitro, do not pertain to the in vivo model of CD34/CD38 reciprocity .
On average, cells in the CD34− population were slower to initiate growth than were CD34+ cells (Table 3). It is possible that mobilization activated both CD34− and CD34+ cells, but to different extents. For example, mobilization might have enhanced CD34− cells' susceptibility to in vitro cytokine stimulation compared to non-mobilized, poorly growing CD34− cells, i.e., those from UCB. Even so, mobilized CD34− cells were slower to initiate growth than were mobilized CD34+ cells.
This report is based on the response of one specific sub-population of CD34− cells cultured in serum-free medium with a single combination of cytokines. Other media, cytokines, and subsets were examined, but little growth was obtained. For example, serum-based medium supported neither proliferation nor colony formation by CD34− PBPC, nor did other cytokine combinations such as [SCF + FL + TPO] and [SCF + TPO]. We also found that the prominent CD34− CD38+ LIN− fraction grew poorly.
In summary, both CD34− and CD34+ PBPC expressed CD34 during ex vivo proliferation. Conversely, during protracted in vitro quiescence, CD34+ cells lost CD34 antigen, and CD34− cells remained CD34− . Modulation of CD34 expression in the first several days' culture does not affect long-term proliferative potential in vitro. These results show that determinations of CD34+ cell frequency in expansion cultures can underestimate hematopoietic content.
This research was supported by the American Red Cross.