We studied the genetic and engraftment phenotype of highly purified murine hematopoietic stem cells (lineage negative, rhodamine-low, Hoechst-low) through cytokine-stimulated cell cycle. Cells were cultured in interleukin (IL)-3, IL-6, IL-11, and steel factor for 0 to 48 h and tested for engraftment capacity in a lethally irradiated murine competitive transplant model. Engraftment showed major fluctuations with nadirs at 36 and 48 h of culture and recovery during the next G1. Gene expression of quiescent (0 h) or cycling (48 h) stem cells was compared with lineage positive cells by 3′ end PCR differential display analysis. Individual PCR bands were quantified using a 0 to 9 scale and results were visually compared using color-coded matrices. We defined a set of 637 transcripts expressed in stem cells and not expressed in lineage positive cells. Gene expression analyzed at 0 and 48 h showed a major shift from “stem cell genes” being highly expressed at 0 h and turned off at 48 h, while “cell division” genes were turned on at 48 h. These observations suggest stem cell gene expression shifts through cell cycle in relation to cell cycle related alterations of stem cell phenotype. The engraftment defect is related to a major phenotypic change of the stem cell.
Hematopoietic stem cells, despite being undifferentiated, can express specific stem cell functions, such as rapid homing to bone marrow (1, 2), cell surface molecule expression (3, 4), rapid motility (5, 6), and settlement in their final endosteal bone marrow niche (7). These cells are capable of quasi-infinite self-renewal and have a tremendous differentiation potential, a single to a few cells being capable of repopulating the entire hematopoietic system of a lethally irradiated recipient (8). Murine stem cells have been defined based on their undifferentiated characteristics (i.e., lineage negativity), as well as specific positive markers, such as Sca-1 and c-kit. Another approach was based on the relative quiescence of stem cells and used the DNA binding dye Hoechst 33342 and the mitochondrial binding dye rhodamine 123. Isolation was based on a negative lineage selection using magnetic beads, followed by low Hoechst and low rhodamine expression (LRH; references 9 and 10).
Previous work with unseparated marrow has shown a loss of engraftment potential when whole marrow was cultured in cytokine cocktails such as IL-3, IL-6, IL-11, and Steel factor (11, 12) or Flt3, TPO, and stem cell factor (SCF)* (13). This defect was reversible and cell cycle mapping defined late S-early G2 as the lowest engraftment period. In contrast to general dogma, primitive long-term renewing stem cells are not quiescent. BRDU incorporation experiments showed that more than 60% of stem cells had cycled at least once within 4 wk (10, 14, 15). They probably enter and exit cycle frequently, alternatively they may exist in a prolonged G1 period. Furthermore, primitive stem cells are easily induced into active cell cycle after either in vivo transplantation or in vitro cytokine exposure (16, 17). Whole bone marrow stem cells probably cycle in an asynchronous fashion. Hematopoietic stem cells, purified by the Hoechst/rhodamine approach, are relatively quiescent, and when exposed to cytokines show a highly synchronous progression through cell cycle (17). We have now studied the engraftment function of highly purified LRH cells as they transit through cell cycle under cytokine stimulation.
To identify specific cell cycle related changes that could explain the engraftment defect, one needs to analyze in detail the pattern of stem cell genes expressed in noncycling and cycling cells. Using different sources of stem cells, several groups described stem cell gene expression analysis using subtracted libraries and array technology. Phillips, et al. (18) have reported on over 2,000 nonredundant gene products from fetal liver hematopoietic stem cells using a subtracted cDNA library to generate a micro array chip. They have identified several genes specific to fetal liver hematopoietic cells. In addition, when comparing fetal hematopoietic cells with adult hematopoietic cells (Rhodamine-123 low c-Kit+ Sca-1+) they found several genes that were coexpressed in fetal and adult stem cells, as well as genes specific for either fetal or adult stem cells. More recently, gene expression profiling of human and murine bone marrow, as well as other types of stem cells, has been reported (19, 20).
Park et al. (21) have also reported on murine hematopoietic stem cell gene profiling. They used a 5,000 cDNA array obtained by subtraction of cDNA from lineage positive cell populations and studied both hematopoietic adult stem cells and multipotent progenitors (with minimal self renewal capacity). Genes primarily expressed in stem cells were transcription factors, RNA binding proteins, chromatin modifiers, and protein kinases.
We have used differential display, which was developed for comparative gene expression studies, and allows for a systematic and nonbiased screening for molecular differences at the level of mRNA expression, between or among different cells or tissues. The comparisons use a gel-based method that employs the display of 3′ end fragments of cDNA generated by cutting with specific restriction enzymes. The amplified cDNAs labeled with radioisotope are then distributed on a denaturing polyacrylamide gel and visualized by autoradiography (22). Side-by-side comparison of mRNA species from two or more related samples allows identification of both up- and down-regulated genes.
In this paper, we describe the effects of cell cycling on engraftment and transcript expression using highly purified hematopoietic stem cells cultured in IL-3, IL-6, IL-11, and steel factor.
Materials And Methods
6 to 8 wk-old male or female BALB/c mice were purchased from Taconic Farms and housed in a conventional clean facility for at least 1 wk before experimental use. All experiments were approved by the University of Massachusetts and Roger Williams Medical Center Institutional Animal Care and Use Committee.
All mice received mouse chow and acidified water ad libitum.
Hematopoietic Stem Cell Purification.
Bone marrow was isolated from iliac bones, femur, and tibiae of BALB/c mice 6 to 8 wk of age. A low-density fraction (<1.077 g/cm2) was isolated on Nycoprep 1.077A (Accurate Chemical and Scientific Corporation). These cells were lineage depleted by a magnetic bead separation with the following antibodies: Ter119, B220, Mac-1, GR-1, Lyt-2, L3t4, and YW25.12.7 and Dyna beads MW450 anti–rat IgG (Dynal). The lineage depleted cell were labeled with Rhodamine 123 at a concentration of 0.1 mg/ml and Hoechst 33342 at 10 mM. Cells were incubated in the dark for 30 min at 37°C, washed, and followed by an additional warm buffer (37°C) incubation for 20 min at 37°C to efflux the Rhodamine. This last incubation was performed twice in some experiments before sorting using FACS®. The 1st through the 13th percentiles of Rhodamine fluorescence and 1st through the 3rd percentiles of Hoechst fluorescence were isolated. This Lineage negative, Rhodamine low and Hoechst low fraction is abbreviated LRH.
The LRH stem cells were cultured at an initial density of 5,000 cells/ml in Dulbecco's Modified Essential Medium low glucose (Life Technologies/GIBCO BRL) containing 15% heat-inactivated fetal calf serum (HyClone Laboratories), 1% penicillin (100 U/ml)/1% streptomycin (100 mg/ml), and 1% l-glutamine (100 mg/ml). The cytokine cocktail used was rmIL-3, 50 U/ml (collaborative), rmIL-6, 50 U/ml (R&D Systems), rhIL-11, 50 ng/ml (Genetics Institute), and rm steel factor, 50 ng/ml (SCF; R&D Systems). The LRH cells were cultured with the above cytokines for 24 to 48 h in nonadherent Teflon bottles in a humidified 5% CO2, 37°C water-jacketed incubator.
Autoradiographic Detection of 3H-thymidine–labeled Nuclei.
The number of cells in S phase was determined by autoradiography of the cells after pulse labeling with 3H-thymidine (17, 23). Briefly, duplicate aliquots of cells (500 to 1,000), at 6-h intervals, as they are progressing through cell cycle, were incubated with 2.5 μCi/ml of 3H-thymidine at 37°C in 5% CO2 containing humidified incubator for 30 min. The incorporation of 3H-thymidine was terminated by cytospin centrifugation and immediately fixing with methanol/acetic acid (2:1 vol/vol), followed by three washes in methanol. Slides were allowed to air dry overnight and a thin film of Kodak nuclear track NTB3 emulsion (Eastman Kodak Co.) was applied. Slides were then incubated in the dark for 4 d, developed, and fixed with Kodak Decktol Developer and Kodak Fixer, respectively. Slides were washed extensively and stained with Giemsa. We determined the percentage of cells with labeled nuclei (representing cells in S phase) by counting 200 cells per slide. In addition, in order to estimate population doubling, cell counts were determined using a hemocytometer.
Irradiation and Transplant.
Female mice were exposed to 10 Gy at a rate of 0.94–0.96 Gy/min in one fraction using a Cesium 137 gamma source (Gamma cell 40; MDS Nordian) at least two hours before transplant. Donor mice were killed by cervical dislocation. Bone marrow was isolated from femurs, tibias and iliac bones with cold PBS. The cells were counted in a hemocytometer, washed and resuspended for injection in PBS. For competitive transplant, 250,000 female marrow cells were mixed with a volume corresponding to 500 Lin− Rhlow Holow cells initially introduced into the culture vial.
Southern Blot Analysis of Engraftment.
DNA was extracted by lysis in 150 mM NaCl, 20 mM Tris (pH 7.5), 20 mM EDTA, 1% SDS, 0.5 mg/ml proteinase K, and 50 mg/ml pancreatic RNase A at 55°C overnight. Purification was completed by organic extraction with phenol-chloroform and ethanol precipitation. 5 μg of each DNA sample was digested with DraI (Boehringer), and separated by gel electrophoresis in 0.8% agarose (GIBCO BRL). DNA fragments were alkaline transferred onto Zetaprobe nylon membranes (Bio-Rad Laboratories). The presence of Y chromosome–specific DNA sequences was assessed using a pY2-cDNA probe (24, 25; donated by Dr. I. Lemischka, Princeton University, Princeton, NJ). Sample loading variability was assessed and adjusted for by reprobing membranes with a cDNA for IL-3 (donated by J. Ihle, St. Jude Children's Research Hospital, Memphis, TN, and DNAX, Palo Alto, CA). Probes were labeled with 32P using a random primed labeling kit (Boehringer), and autoradiography was performed using Kodak XRP x-ray film (Eastern Kodak Co.). Blots were exposed to PhosphorImaging plates (Molecular Dynamics), and the percentage of male and female DNA quantified after scanning the plates with a 400A PhosphorImager (Molecular Dynamics).
cDNA Preparation and PCR Amplification of cDNA 3′ Ends.
Poly A+ RNA was isolated from fresh or cultured cells using Oligotex (QIAGEN). cDNA was generated using the Superscript Choice System (GIBCO BRL; references 26 and 27). Briefly, 10 μg of PolyA+ RNA and 2 pmol of 1 of the 2-base anchored oligo(dT) primers with a heel sequence (20) were mixed with other components for first-strand synthesis reaction. This was followed by reverse transcription using 2 μl of Superscript reverse transcriptase (200 U/ml; Invitrogen/GIBCO BRL). Second-strand synthesis was then performed and the cDNAs were precipitated with ethanol. A Y-adaptor was assembled from 2 oligonucleotides with sequences TAGCGTCCGGCGCAGCGACGGCCAG and GATCCTGGCCGTCGGCTGTCTGTCGGCGC, respectively. The cDNA was digested with BglII and an aliquot of this reaction mixture (∼4 ng) was then used for ligation to 100 ng of the Y-shaped adaptor. After ligation, the reaction mixture was diluted with water, T4 ligase was denaturated at 65° and aliquots were made for PCR amplification (28). The primers used corresponded to the heel sequence for the 3′ end and to one of the Y-adaptors for the 5′ end (TAGCGTCCGGCGCAGCGAC). The PCR products were identified on the display gel by mixing a small proportion of [γ32P]ATP labeled 5′ primer to the cold primer. PCR was performed with manual hot start for 28–30 cycles of 94°C for 30 s, 56°C for 2 min, and 72°C for 30 s. PCR products (2.5 μl) were analyzed on a 6% polyacrylamide sequencing gel. Bands of interest were extracted from the display gels, reamplified using the 5′ and 3′ primers, and directly sequenced or subcloned into pCR-Script using the PCR-Script cloning kit (Stratagene). Sequencing was performed on an ABI automated sequencer. A schematic of the procedure is presented in Fig. 1.
Differential Display Analysis.
Each PCR was repeated 2–3 times and bands of interest were given a relative intensity from 0 to 9 by visual evaluation. The radioactive signal being carried by the 5′ primer only, the size of the PCR product did not influence the intensity grading. The information on each band was stored into an Excel 2000 spread sheet (Microsoft). The sequences of the bands were compared with public databases by BLAST® analysis (29) and information on matching gene/sequence was collected from NCBI website (30). The gene category was defined based on these similarities.
Database Comparison and Graphical Representation.
The differential display of LRH, Lineage positive (Lin+), and LRH cultured for 48 h (LRH 48) was compared by subtraction of each band intensity and sorted according to the difference. Based on the intensity of expression, a graphical representation was constructed using color-coded squares, each representing a unique cDNA target. In this array, each gene was ordered according to the intensity of the difference between two groups. Thus, an individual gene can be visually compared between different cell populations, their position on each array being the same. In addition this color analysis allowed direct visual comparison of the entire subset of genes to identify patterns. Specific genes represented in this pattern analysis is available in supplementary data (see reference 30).
Data can be assessed on line as follows: open the following URL (http://info.med.yale.edu/genetics/weiss/Linksfor_Hematon.htm) and then go to the “stem cells” link. The total experimental information on gel analyses of RNA expression is collected in a Microsoft Access database with a separate sheet for each gel band that was analyzed.
Engraftment Defect Through Cell Cycle.
In two separate experiments, LRH stem cells were evaluated for engraftability at 8 wk in a competitive transplant model. The initial engraftment level was 50 ± 1% and 39 ± 4% for fresh LRH cells. Two nadirs of engraftment were observed at 36 and 48 h of culture in IL-3, IL-6, IL-11, and steel factor. The chimerism was 10 ± 2 and 13 ± 1 at 36 h and 4 ± 1 and 12 ± 2 at 48 h. A recovery of engraftment was observed at 40 h (Fig. 2). In a separate experiment, LRH stem cells, cultured in the same cytokines, were mapped for cell cycle by 3H-thymidine pulse labeling and cell count doubling. Cell cycle status is shown in the bottom of Fig. 2. These data indicate that engraftment nadirs occurred during late S/early G2.
LRH Specific Genes.
PCR amplification of the cDNA 3′ ends was performed using a Y-adaptor ligated to restriction fragments of the cDNA mixture (see Fig. 1). Gene expression of LRH stem cells and lineage positive (Lin+) fraction was compared by differential display. A total of 637 genes were expressed only in LRH cells. They were all subcloned and sequenced. Among these genes, 411 were unknown with no homologies to GenBank databases. There were 226 cDNA with homologies to GenBank databases and 126 of these corresponded to genes with known function. The large number of transcripts not associated with known genes is noteworthy. Very recent studies using genomic “tiling” arrays have shown that a much larger portion of the genome is transcriptionally active than would have been predicted from the presence of known or putative genes (31). The known genes belonged to 19 different gene categories (footnote, Table I).
LRH Gene Expression Shifts Through Cell Cycle.
To evaluate gene expression through cytokine-induced cell cycle, we prepared cDNA from LRH cells incubated in Teflon bottles for 0 or 48 h with IL-3, IL-6, IL-11, and steel factor (LRH48). 251 transcripts from fresh LRH cells or Lin+ cells were chosen for analysis based on their differential expression patterns. A major shift in gene expression was found with many genes active in LRH but turned off in LRH48 and other genes inactive in LRH turned on in LRH48 (Fig. 4). Analyzing the difference of gene expression between LRH and LRH48, we made a comparison of expression levels between baseline and 48 h of culture. Genes were sorted according to their difference in level of expression with a minimal difference of 4. A total 89 cDNA's were turned off at 48 h of culture with a residual expression level ≤2. These genes, related to noncycling cells are defined as quiescent. Out of these cDNAs, we found 51 sequences with similarities to GenBank. Table II describes 28 genes whose name and function are known. A total of 39 cDNA were turned on at 48 h from a baseline expression level ≤2. These genes, termed cycling represent 25 sequences with homologies to the GenBank database and 14 known genes (Table III). A few cDNA's (29) had a stable expression through cycle, while not expressed in Lineage+ cells. These represent 18 genes with similarities and 13 known genes (Table III, common genes). The gene shift seen here probably did not represent differentiation, as gene expression was dramatically different between the 48 h group and the differentiated lineage+ cells.
Previous work by our laboratory has demonstrated that cytokine stimulated unseparated whole bone marrow cells developed an engraftment defect occurring during the late S/early G2 of cell cycle. Others have also shown cell cycle associated changes in stem cell phenotype. Studying Lin− Sca-1+ Thy1low through cell cycle, Fleming et al. (32) showed decreased engraftment in S/G2/M as compared with G1. In a similar vein, Lin− Sca-1+ cells, selected based on Hoechst staining, were found to have long term engraftment potential only when in G0/G1 and not when in S/G2/M (33). We have also recently shown that purified stem cells (Lin− Sca-1+) evidenced a homing defect coincident with the timing of an engraftment defect (2). In addition, work in collaboration with Dr. Pamela Becker, showed a major fluctuation of cell surface adhesion molecules on LRH cells during cell cycle passage (4). Here, using highly purified murine marrow hematopoietic stem cells, we have confirmed a reversible fluctuation of engraftment through cell cycle, with a comparable timing as was seen with studies on unseparated marrow cells. The engraftment potential was followed for two cell divisions and fluctuation was reproducibly observed.
Using a nonselective approach with differential display analysis, we were able to identify over 600 stem cell genes, most of these being actively expressed in relatively “quiescent” stem cells selected on the base of their quiescent status. The gene machinery is highly active. Previous studies have also shown that LRH cells can move extremely rapidly, thus further negating the dogma that these cells are metabolically inactive (6).
The 48 h culture time point, showing a major engraftment defect, was chosen for analysis of the pattern of LRH gene expression. A major shift of gene expression was demonstrated. Compared with time zero (i.e., fresh LRH), the cells cultured for 48 h had turned down most of the highly expressed genes and turned on most of the genes initially not or faintly expressed. From these data, we can postulate that many genes necessary for engraftment or initial homing are turned off during cell division, thus possibly explaining the observed engraftment defect. A number of genes were turned on and it is unknown whether these genes participate in altering homing and engraftment. In contrast, many genes were comparably expressed in both LRH and LRH48 cells, defining a pattern of stem cell specific genes whose expression does not seem to be modified by cycling. When comparing the categories of genes expressed in LRH and in LRH48 cells, the “quiescent” genes are mainly transcription factors and protein synthesis genes, while “cycling genes” are related to cell cycle regulation and chromatin remodeling.
When comparing the same panel of genes with Lin+ cells, we noted a completely different picture; the genes turned on during cell cycle were not expressed in Lin+ cells with a few exceptions such as Histone H2A gene (see Table III), active in both cycling LRH48 and Lin+ but not in fresh LRH. This pattern thus suggests cell cycle specificity and is fundamentally different from lineage differentiation. This weighs strongly against the hypothesis that cell cycle engraftment defect could be related to lineage differentiation. However, it must be acknowledged that some of the genes turned on and expressed at 48 h could represent an early onset of differentiation, as the gene expression profile of cells early in differentiation might be different from that seen in fully differentiated cells such as are represented by lineage positive cells. This seems unlikely to us, but will only be resolved with experiments showing reversibility of these gene changes.
These results show a major shift in gene expression of purified marrow stem cells at a time when engraftment is markedly and reversibly depressed. It is difficult to assign specific roles to individual genes, given the relatively large number that are turned on or off and given the presence of a number of unknowns. Genes turned on at 48 h of culture include those involved in DNA damage repair (34–36), chromatin modification (37, 38), RNA splicing (39), and intracellular signaling (40, 41). Several genes could be involved in cell–cell and cell–matrix interactions, adhesion and cell migration. These are rat basement membrane-associated chondroitin proteoglycan BAMacan (42), mouse cell division control protein 19 (43), and mouse β-1,4-galactosyltransferase (44, 45). The latter appears to be important in cytoskeletal assembly and lamellipodia stability and “mesenchymal cell migration on basal lamina”. Expression of these particular genes could indicate changes in stem cell migration/homing patterns which could negatively effect marrow stem cell engraftment. In a similar vein, genes which are turned off could adversely effect stem cell engraftment into the marrow cavity. Many maintenance genes are turned off and their effects on proliferation and survival could determine ultimate engraftment outcomes. Human MEMD is turned off, and this gene appears to play a role in cell–cell interactions and in migration of mobile cells through tissues (46). Thus, turning off of this gene could also have effects in influencing marrow homing. Altogether the observed engraftment defect is likely to be due to polygenic changes.
Recent studies in our laboratory have indicated that when marrow homing is depressed, stem cells are diverted to other nonmarrow tissues, such as lung (47). Alterations of expression of different migration/adhesion factors thus may be involved in such stem cell diversions. This is, of course, only one of many possibilities. A fuller understanding will have to await knockout (embryonic or siRNA adult) studies of individual gene function and identification of the unknown genes.
The modulation of gene expression is consistent with previous observations of modulation of both adhesion proteins and cytokine receptor expression (4, 48). The observations that primitive stem cells are continuously, if intermittently, passing through cell cycle suggests that their phenotype is probably continuously changing.
During cell cycle progression, chromatin is remodeled. In previous studies, chromatin modulations at the β-globin and lysozyme gene loci were evaluated (49–52). Myeloid specific cis-regulatory elements showed a specific chromatin pattern at the lysozyme locus in myelomonocytic cells at different differentiation stages. This chromatin pattern was also found in multipotent hematopoietic progenitors, but was no longer apparent with erythroid differentiation.
Studies of multipotent hematopoietic stem cell chromatin structure showed that the lineage associated genes, globin, myeloperoxidase, IgH, and CD3δ have accessible control regions before unilineage commitment (50–54). Other studies suggest that chromatin remodeling factors recruited in one phase of cell cycle may determine ultimate action in a later phase. This in turn would determine changes in transcriptional programs (55). A reasonable sequence of events for stem cell phenotype regulation is chromatin remodeling with cytokine induced cell cycle passage, leading to varying levels of transcription factor access to DNA, followed by alteration in gene expression. Thus, these events would prime a stem cell to respond to a specific signal. If the signal was not delivered, the stem cells would again change phenotype. For example, Notch is probably an important primitive stem cell surface receptor. If Notch expression was augmented or present and the stem cell did not see the appropriate ligand, the moment of opportunity would pass and the cell would then be responsive to different signals (56, 57). If, however the cell saw its ligand, ragged-1, on a stromal cell, a sequence of events would follow with activation of PU.1 and induction of myeloid differentiation (58, 59).
In summary, using highly purified murine HSC, we have shown a major shift of gene expression between two specific functional states. Thus, noncycling HSC selectively express mainly transcription regulators and protein synthesis factors, while they are fully capable of repopulating a myeloablated transplant recipient. In contrast, cells in S/G2 have turned down most of the originally active genes, and now express cell cycle related as well as chromatin remodeling genes. This indicates a stem cell phenotype that is continuously changing its potential over time, while HSC slowly travel through the cell cycle.
We are grateful to N.V. Baskaran and H. Habibian for expert technical help.
This study was supported by grants from the National Heart, Lung, and Blood Institute (P0-1 HL56920) and the National Institutes of Diabetes and Digestive and Kidney Diseases (P0-1 DK50222) and (R0-1 DK27424).
Abbreviation used in this paper: SCF, stem cell factor (or steel factor).