Proliferation kinetics of subpopulations of human marrow cells determined by quantifying in vivo incorporation of [²H₂]-glucose into DNA of S-phase cells

Blood cell and platelet production in healthy adult mammals is provided by a heterogeneous population of committed and primitive progenitors normally found in the bone marrow. Subpopulations of hematopoietic progenitors not only differ in their self-renewal, proliferation, and differentiation capacities,^1-5  but they also have different cell cycle and turnover kinetics.^1-3,6-8  For example, in vivo bromodeoxyuridine (BrdU)–labeling studies demonstrated that murine marrow primitive progenitors capable of long-term reconstitution of lethally irradiated mice have turnover times of 30 days or longer.^6,7  In contrast, granulocyte/macrophage committed progenitors have rapid turnover times of only 1 to 2 days.^3,8  Results from murine in vivo BrdU studies also demonstrated that hematologic disruptions observed in response to some diseases, chemicals, and endotoxins are associated with alterations in cell cycle and turnover kinetics of some subpopulations of hematopoietic progenitors.^8,9  In vitro and in vivo studies demonstrated that HIV infection, chemotherapy, radiation, and hematopoietic stem cell transplantation have short-term and long-term suppressive effects on human marrow progenitor/stem cell numbers and function.^10-14  Indirect measurements indicated that some of the observed effects were due to alterations in cell cycle and turnover kinetics of subpopulations of progenitor/stem cells.^12,13  Direct measurements of in vivo turnover kinetics of cell populations in human subjects have been limited in part because the myelosuppressive effects of BrdU have precluded in vivo analyses in healthy individuals.⁵ 

Results from recent reports indicate that the frequency and the cell cycle and turnover kinetics of primitive progenitors in the marrow of larger animals are significantly different from those parameters in mice.^15-18  For example, in studies by Albkowitz et al,^16,17  the estimated number of long-term repopulating cells per 10 million nucleated cells was 800 in marrow from mice, but only 6 in marrow from Safari cats, and the replication rates of marrow long-term repopulating cells were significantly slower in Safari cats than in mice. In addition, results from indirect measurements suggest that human primitive progenitor/stem cells have even slower replication rates.¹³  The effects of disease, chemotherapy, transplantation, and cell manipulations on turnover kinetics of subpopulations of hematopoietic progenitors observed in murine studies also may be different in human subjects.

Recently, a stable isotope–mass spectrometric technique (SIMST) was adapted for measurements of in vivo cell proliferation kinetics.¹⁹  SIMST was previously developed as a nonradioactive and nontoxic method to measure in vivo biosynthesis and turnover kinetics of cholesterol and other polymers.²⁰  For cell proliferation studies a nontoxic stable isotope-labeled form of glucose, 6,6-[²H₂]-glucose, is administered to animals or human subjects. The infused [²H₂]-glucose becomes incorporated into replicating DNA of all S-phase cells through the de novo nucleotide synthesis pathway. Various times after infusion of [²H₂]-glucose, cells are removed and the percent [²H₂]-glucose incorporated into DNA in the form of [²H₂]-deoxyadenosine (that is, %[²H₂]-dA enrichment) is determined by gas chromatographic–mass spectrometric analysis. The rate constant of replacement of unlabeled by labeled DNA strands is then used in calculating the turnover kinetics of the cell population. Results from studies using in vivo [²H₂]-glucose-SIMST to investigate T-cell turnover kinetics in human subjects demonstrated that T-cell turnover kinetics determined from the in vivo [²H₂]-glucose-SIMST studies were consistent with results of studies using other approaches. In addition, SIMST was sensitive enough to detect differences in T-cell turnover kinetics in healthy and HIV-infected individuals and differences in turnover kinetics of memory/effector T cells and naive T cells.^21,22  In those studies in vivo [²H₂]-glucose-SIMST provided a nontoxic and sensitive assay to compare the effects of disease and treatments on the turnover kinetics of multiple subpopulations of cells from the same tissue sample.

In vivo [²H₂]-glucose-SIMST to measure turnover kinetics of hematopoietic progenitors in human subjects has not been reported. The purpose of studies in the present report was to determine the feasibility of using in vivo [²H₂]-glucose-SIMST to measure turnover kinetics of subpopulations of hematopoietic progenitor cells. In these studies, marrow was obtained from healthy human subjects various times following a 2-day infusion of [²H₂]-glucose. The %[²H₂]-dA enrichment was measured in subpopulations of cells selected from the marrow samples. The results were sensitive enough to detect the presence of subpopulations of CD34⁺ cells with relatively rapid turnover kinetics and others with slower turnover rates and allowed for the analysis of turnover kinetics of multiple cell populations from the same marrow aspirate. This study presents the first in vivo data using SIMST to measure turnover kinetics of progenitor and precursor cells in the marrow of healthy human subjects and demonstrates the potential in investigating abnormalities of hematopoiesis in a variety of clinical settings.

Study participants were required to give informed consent as participants of National Cancer Institute Institution Review Board–approved protocols. Male and female subjects were between 19 and 84 years old (Table 1) and had no history or current indication of diabetes. The study participants were admitted to the National Institutes of Health Clinical Center and administered 6% to 10% 6,6-²H₂-glucose (Cambridge Isotope Laboratories, Andover, MA) in normal saline by continuous intravenous infusion at 42 cc/h for a total dose of 120 to 200 grams. Throughout the infusion period the subjects were maintained on a low-carbohydrate diet consisting of no more than 50 grams of carbohydrates per day. No adverse side effects were observed in any of the individuals who received the 2-day infusion of 6% to 10% [²H₂]-glucose. Peripheral blood samples for measurements of plasma concentrations of [²H₂]-glucose were obtained during [²H₂]-glucose infusion. Marrow aspirates obtained after the end of the infusion period were used to determine [²H₂]-glucose levels in subpopulations of hematopoietic progenitors and precursors.

Marrow aspirates were obtained from the iliac crest. Low-density (≤ 1.077 g/cm³) cells were collected on a Ficoll-sodium diatrizoate gradient (Lymphocyte Separation Medium; Organon Teknicka, Durham, NC) and enriched for either CD34⁺ or CD133⁺ cells by positive immunomagnetic selection using high-gradient magnetic separation column MiniMacs CD34 Progenitor Cell or CD133 Cell Isolation Kits (Miltenyi Biotec, Auburn, CA). CD34⁺ cells also were selected from CD133^– fractions (CD133^–CD34⁺ cells). The percent CD34⁺ cells and coexpression of CD133 were determined by staining with anti–CD34-fluorescein isothiocyanate (FITC) (anti–HPCA-1 [human progenitor cell antigen 1]; Becton Dickinson, San Jose, CA) and anti–CD133-phycoerythrin (PE) (Miltenyi).

Cells in the CD34^– fractions were labeled with an antibody to glycophorin A and CD41a (GPIIb/IIIa) to select for, respectively, erythroid and megakaryocyte lineage cells. Glycophorin A is first detectable on pronormoblasts, the earliest identifiable progeny of erythroid colony-forming unit (CFU-E) and is expressed throughout all subsequent stages of differentiation.^23-25  The nucleated stages include the dividing precursors (pronormoblasts, basophilic normoblasts, and polychromatophilic normoblasts) and the nondividing orthochromatic normoblasts. CD41a is expressed on the CD34⁺ megakaryocyte progenitor cells and the CD34^– immature and mature megakaryocytic cells found in the marrow.^26,27  Cells in the CD34^– fractions were washed, resuspended in 1% bovine serum albumin in phosphate buffered saline (FACS buffer), and incubated for 30 minutes at 4°C with anti–CD41a-FITC and anti–glycophorin A-PE (PharMingen, San Diego, CA). CD41a⁺/glycophorin A^– and CD41a^–/glycophorin A⁺ cells were sorted with a FACSVantage SE cell sorter (BD Biosciences, San Jose, CA) equipped with a Coherent I-90 laser (Mountain View, CA) emitting at 488 nm with 150 milliwatts of power.

Peripheral blood samples were obtained at 0, 12, 24, and 48 hours during the infusion period, and determinations of the percent [²H₂]-glucose enrichment in the plasma were measured as previously described.^19,20  The mean plasma [²H₂]-glucose enrichment values (Table 1) were multiplied by a correction factor of 0.65 to account for intracellular dilution of glucose.¹⁹  Measurements of cell kinetics were determined from the percent [²H₂]-glucose incorporated into DNA in the form of [²H₂]-deoxyadenosine (that is, %[²H₂]-dA enrichment). DNA was extracted from 4 × 10⁵ to 1.5 × 10⁶ cells using QIAamp DNA Blood kits (Qiagen, Valencia, CA) following manufacturer's recommended procedures and frozen. All samples from individual donors were processed and analyzed at the same time. DNA was enzymatically hydrolyzed to deoxyribonucleosides. The deoxyadenosine was isolated using an LC18 solid phase extraction (SPE) column (Supelco, Bellefonte, PA) and converted to the triacetylaldonitrile derivative of deoxyribose, for measurement of %[²H₂]-dAenrichment by gas chromatography/mass spectrometry (GC/MS), as previously described.^19,21,22  In an alternative method that gave similar results (data not shown), DNA was precipitated with ethanol and dried and then digested with phosphodiesterase I (Worthington Biochemical, Lakewood, NJ) and DNAse I (Life Technologies, Rockville, MD) followed by bacterial alkaline phosphatase (Life Technologies). Deoxyribonucleosides were separated from salts, enzymes, and proteins on solid phase extraction columns, concentrated by vacuum centrifugation, derivatized with equal parts N, N-dimethylformamide (DMF) and methelute (Pierce Chemical, Rockford, IL), and then reconstituted in DMF and methelute to approximately 50 μmol/L dA. Replicate samples were injected into a Hewlett Packard model 6890 gas chromatograph equipped with a model 5973 mass selective detector operating in electron impact ionization mode. Ions representing permethylated derivatives of dA and [²H₂]-dA were monitored at 292 m/z and 294 m/z, respectively.

The %[²H₂]-dA enrichment was determined by multilinear regression analysis of standard curves generated from 10% to 0.1% of 20 μmol/L and 100 μmol/L [²H₂]-dA samples. Calculations are based on a random replacement model of exponential kinetics.^19,20  The fraction of newly added cells (“f” = %[²H₂]-dA/[plasma [²H₂]-glucose enrichment × 0.65]), fractional replacement rate (k/d = –ln[(1–“f”) × 24 hours]/[hours of infusion]), and replacement half-life (t_1/2 = 0.693/[k/d]) were calculated.^19,20  Turnover rates also were calculated from the decrease in %[²H₂]-dA (delabeling) observed at week 2 and week 4.²⁸  Student 2-tailed t test was used to test for significant differences between different cell populations (P < .05). Paired t test used to test for significant differences between CD133⁺ and CD34⁺CD133^– cells from the same marrow samples.

Cells from marrow aspirates were enriched for CD34⁺, CD133⁺, and CD133^–CD34⁺ cells. More than 95% of the selected cells expressed the CD34 cell surface antigen (Table 2). From 13% to 36% of CD34⁺ selected marrow cells coexpressed the CD133 cell surface antigen, and the percentages were not significantly different from measurements in unseparated marrow aspirates from the same donors. Thus, the proportion of CD34⁺CD133⁺ and CD34⁺CD133^– cells in the marrow aspirates was maintained after selection for CD34⁺ cells. Similar to results by others,^29,30  approximately one third of the CD34⁺ cells in adult marrow coexpressed the CD133 cell surface antigen. CD34⁺CD133⁺ cells were at least 2-fold higher in marrow enriched for CD133⁺ cells without prior selection for CD34⁺ cells. From 9% to 15% of CD34⁺ cells selected from the CD133^– fraction coexpressed CD133, however, intensity of expression was dim (data not shown).

In vivo turnover kinetics of marrow CD34⁺ cells were calculated from the %[²H₂]-dA enrichment in DNA of cells selected from marrows obtained within 2 hours after the end the [²H₂]-glucose infusion period (day 0). Following the 2-day infusion, 28% to 50% of marrow CD34⁺ cells were newly produced cells (one new cell added per cell division) (Table 3). Those measurements represented a fractional replacement rate (k_labeling/d) of 0.28, a t_1/2labeling of 2.5 days (median values) for CD34⁺ cells in healthy donor marrows.

Turnover rates also were estimated from the decrease in the %[²H₂]-dA enrichment observed at week 2. The estimated t_{1/2delabeling} determined from the 4 donor marrows obtained at day 0 and week 2 was 5 ± 1.4 days (Table 3) and 4 ± 0.1 days (n = 4) for donor marrows obtained at day 1 and week 2. The estimated t_{1/2delabeling} determinations were longer than predicted from the k_labeling/d calculated from the %[²H₂]-dA enrichment in CD34⁺ cells obtained at day 0 (dashed line in Figure 1) and indicated that CD34⁺ cells comprised subpopulations of cells with different turnover kinetics. At week 4, [²H₂]-dA enrichment was still detectable in CD34⁺ cells from 5 of 7 donor marrows (Figure 1). The estimated mean t_{1/2delabeling} calculated from the decrease in the %[²H₂]-dA enrichment between week 2 and week 4 was 24 days (range, 6-82 days). The relatively short t_1/2labeling and the long t_{1/2delabeling} times determined from the persistence of [²H₂]-dA enrichment beyond 2 weeks detected the presence of subpopulations of CD34⁺ cells with relatively rapid turnover kinetics and others with slower turnover rates.

The %[²H₂]-dA enrichment following the 2-day infusion of [²H₂]-glucose was used to compare turnover kinetics of CD133⁺ cells and CD133^–CD34⁺ cells selected from the marrow of healthy donors. At day 0, the fraction of newly produced CD133⁺ cells (one new cell added per cell division) during the [²H₂]-glucose infusion period was less than observed for CD133^–CD34⁺ cells selected from the same marrow aspirate (Table 4). The k_labeling/d for CD133^–CD34⁺ cells was similar for all 3 donor marrows evaluated, and the mean t_1/2labeling was 2.5 days. The turnover kinetics for CD133^–CD34⁺ cells from donor 22 were similar to results observed for CD34⁺ cells selected from low-density cells from the same marrow aspirate (Table 3). In contrast, CD133⁺ cells had slower turnover rates than CD133^–CD34⁺ cells (P = .04, n = 4).

The %[²H₂]-dA enrichment at day 1 also was less for CD133⁺ cells than for CD133^–CD34⁺ cells from the same marrow sample (Table 4, donor 14). By week 2 following the end of the [²H₂]-glucose infusion period, the %[²H₂]-dA enrichment was less than at day 0 or day 1, however, the %[²H₂]-dA enrichment still detectable in CD133⁺ cells was approximately 1.7-fold higher (P = .01, n = 3) than for CD133^–CD34⁺ cells, and the estimated t_{1/2delabeling} was longer for CD133⁺ cells (Table 4). A similar difference in %[²H₂]-dA enrichment was observed in cells from 3 other donor marrows obtained at week 2. At week 4 [²H₂]-dA enrichment was detectable in cells from all 5 donor marrows evaluated, and the levels were higher for CD133⁺ cells (range, 0.09%-4.64%) than for CD133^–CD34⁺ cells (range, 0.03%-0.58%). Although the enrichment levels were still higher for CD133⁺ cells, the t_{1/2delabeling} estimated from week 2 and week 4 measurements from 4 of the donors either were similar to (n = 2), less than (n = 1), or greater than (n = 1) estimates for CD133^–CD34⁺ cells. It is of interest to note that the %[²H₂]-dA enrichment at week 4 for CD133^–CD34⁺ cells (D18 = 0.58%, D11 = 0.38%) was similar to observations at week 2 and that the percent for CD133⁺ cells (D18 = 3.0% and D11 = 4.6%) was higher than week 2 observations.

In vivo labeling kinetics during ²[H]-glucose infusion were determined for CD41a (megakaryocyte lineage) and for glycophorin A (erythroid lineage)–expressing cells remaining in marrow samples after CD34⁺ cell selection. Glycophorin A⁺ cells had a mean t_1/2labeling of 3.5 days (range, 2-6 days) (Table 5). Similar results were observed when turnover rates were determined from the decrease in the %[²H₂]-dA enrichment from day 0 to week 2. The [²H₂]-dA enrichment levels at week 4 were less than 0.1% or not detectable. CD41a⁺ cells had a replacement half-life of (t_1/2labeling) of approximately 10 days (range, 5-21 days). The fractional replacement rates and replacement times for CD41a⁺ cells were significantly different (P = .02) from those values from glycophorin A⁺ cells. The results demonstrate that in healthy donor marrow, CD41a⁺ cells have slower turnover kinetics than glycophorin A⁺ cells.

This report summarizes results of studies using in vivo [²H₂]-glucose-SIMST (stable isotope–mass spectrometric technique) to measure turnover kinetics of subpopulations of CD34⁺ cells and precursor cells in healthy human donor marrows. Human subjects were administered an infusion of 6,6-[²H₂]-glucose, a nontoxic stable isotope–labeled form of glucose. In SIMST the infused 6,6-[²H₂]-glucose becomes incorporated into the deoxyribose moiety of DNA in all S-phase cells, generating 2 labeled DNA strands (one per daughter cell) for each cell division.^19,21,31  The percent [²H₂]-glucose incorporated into DNA in the form of [²H₂]deoxyadenosine (that is, %[²H₂]-dA enrichment) is determined by gas chromatography–mass spectrometry, and the proportion of labeled DNA strands provides a measure of the number of cell divisions that occurred in the presence of [²H₂]-glucose. The proportion of DNA strands labeled during [²H₂]-glucose and remaining 2 to 4 weeks after discontinuing the infusions (delabeling) was used to calculate the turnover kinetics of CD34⁺ cells and to compare the turnover rates of CD133⁺ (for instance, CD133) and CD133^–CD34⁺ from the same donor marrow sample. The results demonstrate that in vivo [²H₂]-glucose-SIMST is sensitive enough to detect differences in turnover rates of subpopulations of CD34⁺ cells and hematopoietic precursors in the marrow of healthy human subjects.

The %[²H₂]-dA enrichment in DNA at the end of the 2-day [²H₂]-glucose infusion period was used to determine turnover kinetics of CD34⁺ cells in healthy donor marrows. The levels of [²H₂]-dA enrichment in DNA at the end of the 2-day [²H₂]-glucose infusion period indicated that from 28% to 46% of the CD34⁺ cells were new cells (one new cell added per cell division). Those results represented a mean replacement rate of 27% of the CD34⁺ cell population per day with a replacement half-life (that is, t_1/2labeling, time for 50% of the cells to be replaced) of approximately 2.5 days. Cell cycle analyses by others demonstrated that from 8% to 25% marrow CD34⁺ cells are in S/G2/M and 40% to 60% are in G1 phase of the cell cycle.^12,32,33  In addition, results from cytokinestimulated cultures of marrow CD34⁺ cells indicated that approximately one third of G1 phase cells rapidly progress into S phase.^32,33  Thus, the range and relatively rapid turnover rates of marrow CD34⁺ cells observed in the present report using in vivo [²H₂]-glucose-SIMST are consistent with results from cell cycle analyses of marrow CD34⁺ cells.

After [²H₂]-glucose infusion is discontinued, the rate constant of replacement of labeled by unlabeled DNA strands also can be used to calculate turnover kinetics of cell populations.^22,31  In studies by McCune et al²²  results from measurements of [²H₂]-dA enrichment levels over more than a month after [²H₂]-glucose infusion was discontinued indicated that memory/effector T cells consisted of subpopulations of cells that differed in turnover kinetics. In the present report, the %[²H₂]-dA enrichment was determined for CD34⁺ cells in marrow aspirates taken 2 weeks and 4 weeks after [²H₂]-glucose infusion was discontinued. As expected, levels of [²H₂]-dA enrichment decreased after [²H₂]-glucose infusion was discontinued. However, the %[²H₂]-dA enrichment remaining at week 2 and week 4 was higher than would be predicted from the k_labeling/d determinations, and the persistence of labeled DNA indicated the presence of subpopulations of CD34⁺ cells with slower turnover rates.²²  The 3- to 9-day estimated replacement half-life determined from the decrease in [²H₂]-dA enrichment observed at week 2 (that is, t_{1/2delabeling}) was longer than that observed by others for the rapidly turning over committed progenitors (CFU-GM [granulocyte macrophage colony-forming unit], CFU-E, BFU-E [erythroid burst-forming unit], and CFU-Meg [megakaryocyte colony-forming unit] in murine marrow.^3,8  Although estimates from the levels of [²H₂]-dA enrichment still detectable at week 4 indicated the presence of more slowly turning-over subpopulations of CD34⁺ cells, the t_{1/2delabeling} of approximately 6 days was less than the 12 to 19 days observed for murine primitive progenitors^6,17  or estimates of more than 5 weeks for primitive progenitors in Safari cat marrow.¹⁷ 

In vitro and xenogenic transplant studies have shown that almost all of the long-term repopulating cells and most of the detectable CFU-GM from human hematopoietic tissues are found in subpopulations of CD34⁺ cells that coexpress CD133.^29,30,34  In the present report, in vivo [²H₂]-glucose-SIMST was used to compare the turnover kinetics of CD133⁺ cells and CD133^–CD34⁺ cells. Turnover rates determined from day 0 and week 2 measurements indicated that compared to CD34⁺ cells, CD133⁺ cells and CD133^–CD34⁺ cells were somewhat more homogenous with respect to turnover kinetics. The results also demonstrated that CD133⁺ cells were enriched for more slowly turning-over subpopulations of cells than CD133^–CD34⁺ cells. The persistence of higher levels of [²H₂]-dA enrichment at week 4 also suggests that CD133⁺ cells were enriched for a subpopulation of cells with slower turnover rates. Although [²H₂]-dA enrichment was still detectable at week 4, there was a wide range of values and estimates for t_{1/2delabeling}. Possible reasons for the wide range of values observed at week 4 include individual donor variability and the sensitivity threshold of the assay to detect a subpopulation of cells present at very low frequency. In a long-term BrdU-labeling study in baboons, Mahmud et al¹⁵  observed a wide range in the percentage of BrdU-labeled CD34⁺ cells (56%-84%) and BrdU-labeled CD34⁺Ho^Low/Rho^Low cells (9%-37%) between individual animals. Recent modifications of in vivo [²H₂]-glucose-SIMST have been shown to increase the sensitivity and reproducibility of the technique for measurements of DNA synthesis and cell proliferation.³⁵  Those improvements might allow for measurements in subpopulations of CD34⁺ cells shown to be further enriched for primitive progenitors.

In vivo [²H₂]-glucose-SIMST also was used to measure the turnover kinetics of erythroid and megakaryocyte lineage cells in marrow from healthy donors. Cells in the CD34^– fractions were labeled with an antibody to glycophorin A and CD41a to select for, respectively, erythroid and megakaryocyte lineage cells. Results from in vivo [²H₂]-glucose-SIMST studies demonstrated that a mean of 23% of the erythroid cells, but only 8% of the megakaryocyte lineage cells, were replaced per day. Erythroid precursors are rapidly dividing cells, and Ki-67–labeling studies by others demonstrated that more than 70% of the erythroid cells in healthy donor marrows are in cell cycle.^23,36  In the present report, the turnover rates observed for glycophorin A⁺ cells were similar when determined from either the %[²H₂]-dA enrichment at day 0 or the decrease in enrichment levels by week 2. The results indicate that, unlike CD34⁺ cells, glycophorin A⁺ cells are a relatively homogenous turning-over population of cells. Those results and the replacement half-life of 2 to 6 days determined by in vivo [²H₂]-glucose-SIMST are consistent with the high proliferative activity of the erythroid cell compartment and the 3 to 5 days for pronormoblasts to develop into reticulocytes.^23,27,36  There was more donor-to-donor variation and individual donor variation in labeling and delabeling kinetics of CD41a⁺ cells than observed for glycophorin A⁺ cells from the same group of donor marrows. Possible reasons for t_1/2 times for CD41a⁺ cells that ranged from 5 to 21 days include differences in the number of endomitotic cycles as promegakaryoblasts develop into mature megakaryocytes and the loss of the more fragile larger cell stages during the selection procedures.²⁶  Although there was a broad range of values for CD41a⁺ cells, results from in vivo [²H₂]-glucose-SIMST determinations demonstrated that turnover rates for CD41a⁺ cells were slower than observed for erythroid cells.

The present study provides the first data using in vivo [²H₂]-glucose-SIMST to measure turnover kinetics of hematopoietic progenitors in human subjects. Results demonstrate that in vivo [²H₂]-glucose-SIMST is sensitive enough to detect differences in turnover kinetics between erythroid and megakaryocyte lineage cells and subpopulations of CD34⁺ cells in healthy donor marrows that are consistent with other measurements. In addition the technique allows for analysis of more than one cell type from the same sample of marrow cells. These data provide a preliminary baseline by which turnover kinetics of marrow progenitors can be assessed in vivo in the setting of chemotherapy, immune reconstitution, and a variety of disease states.