TGF-β signaling–deficient hematopoietic stem cells have normal self-renewal and regenerative ability in vivo despite increased proliferative capacity in vitro

Transforming growth factor-β (TGF-β) is a multifunctional cytokine with critical functions in a wide range of physiologic and pathologic processes such as immune response, angiogenesis, tumor development, and wound healing.¹  A large number of in vitro studies have shown that TGF-β also plays a critical role in the regulation of hematopoiesis, in particular as a growth inhibitor of progenitors and stem cells.²  However, the actions of TGF-β are strongly context dependent and may, therefore, vary, depending on the cell type, its differentiation stage, and the environment.³ 

There are 3 mammalian isoforms, TGF-β1, 2, and 3, that signal through 2 transmembrane serine-threonine kinase receptors known as the type I and type II TGF-β receptors (TβRI and TβRII).⁴  Signaling can only occur in the presence of both type I and type II receptors. TGF-β ligand binds type I and II receptors in a heterotetrameric complex, whereby TβRII phosphorylates the kinase domain of TβRI. TβRI propagates the signal by phosphorylating the intracellular mediators Smad2 and Smad3 that subsequently associate with Smad4 to enter the nucleus where transcription of TGF-β target genes is regulated.⁴  Several studies on different cell lines have shown a direct relation between TGF-β signaling and expression of cell cycle regulators, leading to growth arrest.^5-7 

Growth inhibitory effects of TGF-β on hematopoietic progenitors and stem cells have been widely reported. Addition of TGF-β to primary hematopoietic cells in vitro leads to potent growth inhibition of multilineage progenitors, whereas more committed progenitors are less affected or even in some cases stimulated to proliferate.^8,9  Direct growth inhibitory effects of TGF-β on murine and human primitive hematopoietic cells have been demonstrated through in vitro culture experiments in which exogenous TGF-β causes a reversible inhibition of cell division in purified primitive progenitor cells.^10,11  Blocking endogenous TGF-β signaling has been tested in vitro. Treatment of early human hematopoietic progenitors with anti–TGF-β antibodies or antisense oligonucleotides releases these cells from quiescence and enhances the frequency of colony formation.^12,13  By using a competitive repopulation assay in mice, it was shown that culture of bone marrow cells in the presence of an anti–TGF-β antibody before transplantation, enhanced both their short- and long-term repopulation ability.¹⁴  Together, these studies have implied an important role for TGF-β in maintaining quiescence of hematopoietic stem cells (HSCs). Furthermore, they have highlighted the possibility that interfering with TGF-β signaling could be a possible means of activating and expanding HSCs in clinical cell and gene therapy protocols.

Little is known about the role of TGF-β in hematopoiesis in vivo. Indirect evidence for an important role in growth regulation of hematopoietic cells in vivo comes from studies on human malignancies in which loss of expression of TβRI and TβRII has been observed in both myeloid and lymphocytic leukemia.^15,16  Furthermore, administration of TGF-β to mice has been shown to inhibit proliferation of early progenitor cells in vivo.¹⁷  TGF-β1 knock-out mice develop a multifocal inflammatory disorder and die shortly after weaning.^18,19  These mice exhibit increased numbers of proliferating cells in spleen and lymph nodes,²⁰  as well as elevated levels of lymphocytes and neutrophils.¹⁸  However, this hyperproliferation has been shown to be secondary to the inflammatory phenotype.^21,22  Mice lacking the TGF-β receptors TβRI and TβRII die early during embryogenesis from a yolk sac circulation defect.^23,24  We recently reported that the circulation defect in TβRI null embryos is primarily caused by defective angiogenesis, while the hematopoietic potential of yolk sac progenitors is intact.²⁴  However, because of the early lethality of these knock-out models, they have failed to provide detailed information about the role of TGF-β in hematopoiesis and HSC function in vivo.

We, therefore, created a conditional knock-out model in which disruption of the TβRI gene could be induced in adult mice. We used this model to analyze the consequences of deficient TGF-β signaling for hematopoiesis and HSC function in vivo. Surprisingly, HSCs from these mice exhibit normal regenerative ability following bone marrow transplantation despite increased proliferative capacity in vitro. Thus, despite pronounced effects in vitro, TGF-β signaling is dispensable for normal regulation of hematopoiesis and HSCs in vivo.

Cre/loxP gene targeting of exon 3 of the TβRI gene was performed in embryonic stem (ES) cells as described.²⁴  Following transient transfection with a Cre-expressing plasmid, ES cell clones were selected that had excised the neomycin resistance gene but still carried the loxP flanked (“floxed”) exon 3. These cells were injected into C57BL/6 blastocysts, and chimeric mice were generated. Following germ line transmission, floxed mice were mated to homozygosity (f l/f l) mice and were subsequently crossed with MxCre transgenic mice to generate f l/f l Cre mice. Cre expression was induced with 3 intraperitoneal injections (at 2-day intervals) of 250 μg polyinositolic polycytidylic acid (polyIC). Genotyping for detection of wild-type, floxed, and excised alleles was done using a 3-primer polymerase chain reaction (PCR) described previously.²⁴  Detection of the Cre transgene was done by PCR as described.²⁵  Mice were housed and bred in ventilated racks. All animal experiments were approved by Lund University's Animal Ethical Committee.

Blood samples were collected from the retroorbital plexus of anesthetized (isoflurane) mice and analyzed either manually (Bürker chamber) or on a Boule Medonice CA 530-16 blood cell analyzer to determine cell counts. Single-cell suspensions from bone marrow (BM) and spleen were obtained by passage through a 70-μm cell strainer (Becton Dickinson [BD], San Jose, CA). Cells were kept in phosphate-buffered saline (PBS; Gibco-BRL, Paisley, United Kingdom) containing 2% fetal calf serum (FCS; Gibco-BRL). When necessary, red blood cells were lysed with ammonium chloride (NH₄Cl; Stem Cell Technologies, Vancouver, BC).

For lineage depletion, cells were incubated with unconjugated rat antibodies against murine CD4, CD8, CD5, Gr1, Mac1, B220, and TER119 (Pharmingen, San Diego, CA). Following incubation with sheep antirat immunoglobulin G (IgG) crystallizable fragment (Fc)–conjugated immunomagnetic beads (Dynal, Oslo, Norway), lineage-positive cells were removed with a magnetic particle concentrator (MPC-6; Dynal).

Lineage-depleted BM cells were incubated for 2 hours in the presence or absence of 10 ng/mL TGF-β1. Western blot analysis for detection of phosphorylated Smad2 was performed as previously described.²⁶ 

Rat antibodies against murine Mac1, Gr1, B220, CD4, CD3, CD8, Ter119, Sca1, c-kit, CD34, CD45.1 (Ly5.1), CD45.2 (Ly5.2), and mouse anti-Ki67, were obtained from Pharmingen. Antibodies were conjugated with fluorescein isothiocyanate (FITC), phycoerythrin (PE), allophycocyanin (APC), or biotin. Unconjugated antibodies were detected with tricolor conjugated goat F(ab′)²  antirat IgG(H+L) (Caltag Lab, Burlingame, CA). Biotinylated antibodies were detected with streptavidin APC (Pharmingen). Dead cells were excluded through staining with 7-aminoacid (7-AAD; Sigma-Aldrich, St Louis, MO). For cell cycle analysis, cells were fixed and permeabilized by incubation in 0.4% formaldehyde and 0.2% Triton-X overnight before staining with anti-Ki67 and 7-AAD. Cells were sorted on a FACSVantage Cell Sorter (BD) or analyzed on a FACSCalibur (BD), and results were analyzed with CellQuest software (BD).

For granulocyte-macrophage colony-forming units (CFU-GMs), 30 000 fresh BM cells were cultured in 35-mm Petri dishes in 1 mL methylcellulose medium (M3534; Stem Cell Technologies) containing stem cell factor (SCF; 50 ng/mL), interleukin 3 (IL-3; 10 ng/mL), and IL-6 (10 ng/mL). For erythroid burst-forming units (BFU-Es), 150 000 BM cells were plated per milliliter serum-free methylcellulose (M3236; Stem Cell Technologies) supplemented with 50 ng/mL SCF (Amgen, Thousand Oaks, CA), 50 ng/mL thrombopoietin (TPO; Kirin Brewery, Japan), and 5 U/mL human erythropoietin (hEpo; Janssen-Cilag AB, Sollentuna, Sweden). Colonies were scored on day 7.

For spleen colony-forming units (CFU-Ss), 75 000 fresh BM cells were injected into lethally irradiated mice (950 cGy, 110 cGy/min, ¹³⁷Cs gamma rays). Mice were killed after 12 days, and macroscopic spleen colonies were counted.

Lineage negative (lin^–), Sca1⁺, c-kit⁺ cells from induced mice were purified by fluorescence-activated cell sorting (FACS) and plated in Terasaki multiwell plates (Nunc, Kamstrup, Denmark) at one cell per well. Cells were grown in 20 μL serum-free medium (X-vivo 15; BioWhittaker, Walkersville, MD) supplemented with 1% bovine serum albumin (BSA; Stem Cell Technologies), 100 IU/mL penicillin, 100 μg/mL streptomycin (Gibco-BRL), 2 mM l-glutamin (Gibco-BRL), and 10^–4 M 2-mercaptoethanol (Sigma-Aldrich). The following cytokines were used: SCF (25-50 ng/mL; Amgen), TPO (25-50 ng/mL; Kirin), Flt-3 ligand (FL; 25 ng/mL; Immunex, Seattle, WA), granulocyte colony-stimulating factor (G-CSF; 25 ng/mL; Amgen), granulocyte/macrophage colony-stimulating factor (GM-CSF; 10 ng/mL; gift from Novartis, East Hanover, NJ), Il-3 (10 ng/mL; Novartis), and TGF-β1 (10 ng/mL; R&D Systems, Minneapolis, MN). Proliferating cell clones were scored after 12 days.

BM cells were collected from mice 4 weeks after induction and resuspended in 10 mM HEPES (N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid; Gibco-BRL), 140 mM NaCl, 2.5 mM CaCl₂. Cells were incubated with lineage antibodies (see “Flow cytometry”), PE-conjugated anti–c-kit, and FITC-conjugated anti–Sca-1 to detect the lin^–, Sca1⁺, c-kit⁺ fraction. In addition, the cells were incubated with APC-conjugated Annexin V (Pharmingen) to detect apoptotic cells and 7-AAD to exclude dead cells. Apoptotic stem cells were enumerated by FACS as the fraction of lin^–, Sca1⁺, c-kit⁺, 7-AAD^lo cells that stained positively for Annexin V.

BM cells from induced knock-out and control mice (Ly5.2) were depleted of CD4⁺ and CD8⁺ T cells as described in “Cell preparations” for lineage depletion. Cells (4 × 10⁵) were transplanted together with 8 × 10⁵ equally T-cell–depleted B6SJL (Ly5.1) competitor BM cells to lethally irradiated (950 cGy) B6 nude recipient mice. After 12 weeks, recipient mice were killed, and a half femur–equivalent of BM was transferred to secondary B6 nude recipients. Blood samples were collected at several time points from the tail vein of the mice receiving transplants to determine Ly5.1 and Ly5.2 reconstitution levels by FACS.

The significance of results was analyzed by using the Student t test. P < .05 was considered significant.

To study the effects of TGF-β–signaling deficiency in adult mice we generated conditional knock-out mice for TβRI using the Cre/loxP system. Floxed mice were generated by flanking exon 3 of the TβRI gene with loxP sites (Figure 1A). TβRI-floxed (fl/fl) mice were subsequently crossed with Mx-Cre transgenic mice that can be induced to express Cre through intraperitoneal injections with polyIC.²⁷  Cre/loxP recombination in BM progenitors of polyIC-activated TβRI fl/fl Cre mice was very efficient (> 97%) as measured by PCR on individual colonies grown from BM cells (Figure 1B). This finding is consistent with other studies using polyIC-treated Mx-Cre mice that have shown highly efficient Cre/loxP recombination in the hematopoietic compartment, including progenitors and stem cells.^25,28  Furthermore, lineage-depleted (lin^–) bone marrow cells from polyIC-treated fl/fl Cre mice showed a complete absence of the phosphorylated form of the intracellular TGF-β–signaling mediator Smad2 (Figure 1C). Thus, induced disruption of TβRI blocks TGF-β signaling and occurs with high efficiency in the BM.

We have recently reported that mice with an induced disruption of TβRII develop a multifocal inflammatory disease and die 8 to 10 weeks after induction.²⁵  As expected, we observed an identical phenotype in the TβRI (fl/fl Cre) mice following induction with polyIC (J.L., unpublished observations, April 2000). To study hematopoiesis in these mice we decided to perform the analysis at 4 weeks after induction (ie, 2 weeks before the mice show the first clinical signs of disease). As controls to the induced fl/fl Cre knock-out mice, we used equally induced heterozygous fl/+ Cre littermates. When comparing noninduced and induced control mice, no differences in cellularity and distribution of myeloid cells, B cells, and T cells in blood, BM, and spleen could be detected (data not shown). Similarly, there was no difference in numbers of lineage negative (lin^–) Sca1⁺, c-kit⁺ cells in the BM (113 ± 16 versus 121 ± 19 × 10³ cells/femur and tibia, n = 4). Thus, the polyIC treatment per se did not seem to affect the hematopoietic system, which is in agreement with a previous report.²⁹ 

We first asked whether the block in TGF-β signaling would affect the numbers of circulating blood cells. Red and white blood cell counts as well as platelet counts were measured in blood samples taken from the mutant mice. However, no difference in these blood parameters was observed between induced knock-out and control mice (Table 1). Neither was there any significant difference in BM cellularity (knock-out 115 ± 17 versus control 108 ± 20 × 10⁶ cells/2 femurs and tibias, n = 10, P = .4), and cytospin preparations of knock-out BM cells showed a normal distribution of myelopoiesis, erythropoiesis, megakaryopoiesis, and lymphopoiesis (data not shown). To further assess possible effects on commitment and differentiation of hematopoietic cells we performed FACS analysis on cells derived from peripheral blood (PB), BM, and spleen by using markers for myeloid (Mac1, Gr1), B-lymphoid (B220), and T-lymphoid (CD3) lineages. The knock-out mice showed no significant deviation in the distribution of these lineages in any of the hematopoietic organs analyzed, indicating normal homeostasis of mature hematopoietic populations (Table 2). Taken together these results suggest that TGF-β signaling is not required for normal differentiation and maturation of hematopoietic cells in vivo.

The most prominent effect of TGF-β seen on hematopoietic cells in vitro is growth inhibition of primitive progenitors and stem cells. We, therefore, determined the numbers of progenitors and stem cells at different maturation stages in the BM 4 weeks following induction of the mutant mice. Committed progenitors (CFU-GM and BFU-E) were analyzed by colony assays in vitro, whereas multipotent progenitors were analyzed in vivo by spleen colony-forming unit (CFU-S) assays. Surprisingly, the induced block in TGF-β signaling did not affect the numbers of primitive or the more mature populations of hematopoietic progenitor cells (Figure 2A,C). However, progenitors from induced fl/fl Cre mice were completely resistant to the growth inhibitory effects mediated by exogenously added TGF-β (Figure 2B). The same was true also when cells were harvested only 1 day after induction, demonstrating that the induced Cre disruption of TβRI swiftly creates a complete functional block in TGF-β signaling.

With the use of flow cytometry, the numbers of stem cells were estimated by scoring the phenotypic marker profile of lineage negative (lin^–), c-kit⁺, Sca1⁺, and CD34^–, which represents a highly enriched population of quiescent stem cells.³⁰  The size of this cell population was the same in TβRI null and control mice (17.9 ± 7.9 versus 17.5 ± 12.2 × 10³ cells/femur and tibia, n = 9, P = .9), suggesting that physiologic levels of TGF-β do not have any major effects on expansion and proliferation of hematopoietic stem and progenitor cells during endogenous hematopoiesis in vivo.

Several studies have shown that treatment of primitive hematopoietic progenitors with antisense oligonucleotides or neutralizing antibodies to TGF-β releases these cells from quiescence in vitro.^12,13  We, therefore, asked whether primitive progenitors isolated from the knock-out mice would exhibit a higher potential for recruitment into proliferation when cultured as single cells in vitro. Lin^–, Sca1⁺, c-kit⁺ (LSK) cells from knock-out and control mice were purified by FACS and grown as single cells in serum-free medium, and the number of proliferating clones was scored after 12 days. We first tested the proliferative response under high stimulatory conditions with a rich cocktail of cytokines in the presence or absence of active TGF-β1. There was no significant difference in the number of proliferating clones in the absence of TGF-β. However, addition of TGF-β markedly reduced the number of control clones but had no effect on the knock-out group, confirming a block in TGF-β signaling in the TβRI null LSK cells (Figure 3A). In contrast, there was a 2-fold increase in the number of TGF-β–deficient proliferating clones when cultured in SCF alone (Figure 3B). These findings demonstrate that autocrine TGF-β exerts significant inhibitory effects on HSC proliferation under low stimulatory conditions in vitro.

The inhibitory effects of TGF-β on the cell cycle have been widely reported.^5-7  We, therefore, wanted to investigate whether blocked TGF-β signaling in vivo would cause a shift in the cell cycle distribution of primitive hematopoietic progenitor cells from resting to more activated states. Lin^–, c-kit⁺ cells or LSK cells defined by stringent FACS gating were analyzed for their DNA content by staining with 7-AAD, as well as for their expression of the activation marker Ki67, to determine the fraction of cells in the G₀, G₁, and S/G₂/M phases, respectively (Figure 4A). We observed no difference between knock-outs and controls in the cell cycle distribution within the populations enriched for either progenitors (Figure 4B) or stem cells (Figure 4C). This result is consistent with our finding of normal numbers of progenitors and stem cells in the knock-out mice. Furthermore, the normal numbers of LSK cells residing in the resting (G₀) state suggest that, in contrast to previous in vitro data, TGF-β signaling is not required for maintaining quiescence of HSCs in vivo.

Reports from in vitro experiments have shown that TGF-β can inhibit the viability-promoting actions of hematopoietic growth factors and induce apoptosis of primitive murine hematopoietic progenitor cells.³¹  This finding suggests that regulation of apoptosis might be an additional way, apart from interfering with the cell cycle, through which TGF-β exerts its inhibitory effects on stem/progenitor cells. Using Annexin V staining, we, therefore, determined the fraction of apoptotic cells in the primitive LSK population from the knock-out mice to see whether the lack of TGF-β signaling would affect apoptosis (Figure 5A-B). However, there was no significant difference in the proportion of apoptotic cells between the knock-out and control stem cell populations (Figure 5C), indicating that endogenous TGF-β signaling is not critical for regulation of apoptosis in HSCs.

After determining that steady state hematopoiesis in the knock-out mice is normal, we asked whether the TGF-β signaling–deficient HSCs exhibit enhanced regenerative capacity following BM transplantation to lethally irradiated mice. When healthy immune-competent mice receive transplants with TGF-β receptor–deficient BM, they develop an inflammatory disease and die after 6 to 9 weeks²⁵  (J.L., unpublished observations, June 2000). To bypass this problem we used immune-deficient B6 nude mice as recipients in our transplantation experiments. These mice lack a functional thymus and do not produce T cells.³²  Before transplantation, the donor BM samples were depleted of T cells by using magnetic beads and antibodies against CD4 and CD8. Consequently, B6 nude recipients of knock-out BM did not show any signs of illness when monitored up to 6 months after transplantation.

Knock-out or control BM cells, expressing the Ly5.2 marker, were competed against BM cells from B6SJL mice, expressing Ly5.1, in lethally irradiated B6 nude recipient mice (Figure 6A). The mice receiving transplants were monitored by FACS for expression of the Ly5.1 and Ly5.2 markers. The Ly5.2⁺ knock-out and control cells did not differ by their reconstitution in PB at 2, 6, and 12 weeks after transplantation (Figure 6B). Furthermore, both groups of mice showed a similar contribution of Ly5.2 cells within myeloid, B lymphoid, and the primitive LSK population of the BM when killed after 12 weeks (Figure 6C). These results show that both short- and long-term repopulation ability are unaffected in TGF-β signaling–deficient HSCs and that these cells have a normal differentiation potential.

Next, we asked whether the self-renewal ability of HSCs would be affected by the lack of TGF-β signaling. Thus, we killed the primary recipient mice after 12 weeks and transplanted their BM to secondary recipients. Again, there was no significant difference in the contribution of Ly5.2 cells between the knock-out and control groups when secondary recipients were analyzed at 12 weeks after transplantation (Figure 6D), suggesting that self-renewal ability is normal in the knock-out stem cells. PCR analysis of hematopoietic colonies from BM of secondary recipients killed after 12 weeks showed a similar contribution of knock-out and control progenitors, supporting the FACS results (Figure 6E). Importantly, all (45 of 45) colonies derived from fl/fl Cre and fl/+ Cre donors carried excised floxed alleles, excluding the possibility that nonexcised stem cells could have contributed to hematopoiesis in the recipient mice.

TGF-β is generally considered to be a key negative regulator of hematopoietic stem and progenitor cells. In vitro studies have shown that both exogenous and autocrine TGF-β inhibit the proliferation of early murine and human hematopoietic stem/progenitor cells.²  With the use of conditional gene targeting to disrupt the TβRI gene in adult mice we have studied how TGF-β signaling regulates hematopoiesis and HSC function in vivo. We show here that induced gene disruption of TβRI is very efficient in BM, including HSCs, generating a complete, cell-autonomous block in TGF-β signaling. Consistent with an inhibitory role of TGF-β, HSCs from TβRI null mice showed increased proliferation recruitment when cultured as single cells in vitro. However, the induced mutant mice exhibited entirely normal steady state hematopoiesis in vivo, including normal numbers and cell cycle status of progenitors and stem cells. Furthermore, when the proliferation kinetics of HSCs were challenged by competitive repopulation in transplantation experiments, we observed no difference compared with controls, neither when analyzed in primary nor in secondary recipients. These findings suggest that TGF-β signaling is dispensable for normal proliferation and differentiation of HSCs and progenitor cells in vivo, despite a pronounced inhibitory role in vitro.

It is beyond doubt that certain concentrations of active TGF-β inhibit proliferation of HSCs and progenitors both in vitro and in vivo. This situation has been shown convincingly in studies in which exogenous sources of active TGF-β has been added to cells in culture or injected into mice.^8-10,17  However, the levels and spatial distribution achieved from exogenous sources of active TGF-β could very well exert effects that are not relevant in a physiologic context. Because practically all cell types, including those of hematopoietic origin, express TGF-β receptors, they are also likely to respond to exogenous TGF-β stimulation. This calls for loss-of-function studies to directly assess the function of endogenous TGF-β signaling. Several studies have addressed this by adding neutralizing antibodies or antisense oligonucleotides to hematopoietic cells in culture. In these experiments, blocking TGF-β has released primitive hematopoietic progenitors and stem cells from quiescence and increased their regeneration.^12-14  However, it has been unclear to what extent TGF-β signaling is blocked and whether components of the culture medium, such as serum, could affect the results. Indeed, levels of active TGF-β present in serum-containing medium have been shown to exhibit significant inhibitory effects on the growth of primitive erythroid progenitors (BFU-E) in vitro.³³  We, therefore, studied the in vitro growth kinetics of TβRI knockout HSCs using serum-free conditions. When cultured in serum-free medium with low stimulation (SCF alone) a 2-fold increase in proliferation recruitment of TβRI null HSCs could be detected. This finding is consistent with results from similar experiments with single-cell cultures of human CD34⁺ CD38^– cells in which we blocked TGF-β signaling with a dominant-negative type II receptor expressed from an adenoviral vector.³⁴  Together, these studies show that autocrine TGF-β exerts significant inhibitory effects on HSC proliferation also under very stringent conditions in vitro.

In contrast, all in vivo aspects of HSC and progenitor cell function measured in our TGF-β signaling–deficient mice were normal, including cell numbers, differentiation potential, cell cycle status, apoptosis, and repopulation kinetics. Thus, despite substantial evidence from in vitro systems, defining TGF-β as a key negative regulator of HSCs, our study shows that maintenance and function of the stem cell pool in vivo is not dependent on this factor. Because the actions of TGF-β are known to be strongly contextual, this discrepancy is most likely caused by differences in the in vitro and in vivo environments. However, unlike HSCs, many other cell types in TGF-β signaling–deficient mice, such as endothelial cells, keratinocytes, and cells of the immune system, have severely disturbed functions that correlate well with experiences from in vitro experiments.^24,25,35  It is, therefore, possible that the BM microenvironment provides factors with similar or redundant functions to TGF-β to a larger extent than present in other tissues or an in vitro culture system. It is often argued that the very limited numbers and extreme expansion potential of HSCs require a tight regulatory network with more backup mechanisms than other cell types to minimize the risk of uncontrolled growth, leading to leukemia or premature exhaustion of the stem cell pool. We can, therefore, speculate that quiescence of HSCs and maintenance of the stem cell pool is controlled by a number of negative regulators, including TGF-β, with redundant or overlapping functions. This view is supported by a report from Dao et al³⁶  and colleagues, which shows that blocking both the cyclin-dependent kinase (CDK) inhibitor p27 and TGF-β in primitive human hematopoietic progenitors has a synergistic effect in increasing their susceptibility to retroviral transduction, suggesting that p27 and TGF-β cooperate in maintaining quiescence of these cells.³⁶  Similarly, cross talk and redundant mechanisms from other signaling pathways could also play a role in our knock-out mice. TGF-β is a member of a superfamily of cytokines that includes the activins and the bone morphogenetic proteins (BMPs).⁴  TGF-β family members signal through similar types of receptors and use the Smad proteins as intracellular mediators.⁴  For example, the activin signaling pathway is mediated through the same Smad proteins (Smad2 and Smad3) as TGF-β. Both activins and BMPs have been shown to have regulatory functions in hematopoiesis.^37,38  It is, therefore, conceivable that these or other signaling pathways could compensate for some TGF-β functions in our knock-out mice.

It should also be pointed out that TGF-β signaling may have different functions in the contexts of murine and human hematopoiesis. For example, TGF-β has been shown to induce apoptosis in primitive murine hematopoietic progenitor cells.³¹  In contrast, there are reports from studies on human CD34⁺ cells suggesting that the effects of TGF-β are antiapoptotic.³⁹  We can, therefore, not exclude that blocking TGF-β signaling in HSCs in the human system would have consequences other than those we have described here for mice.

To summarize, we have for the first time evaluated the in vivo role of TGF-β in hematopoiesis and found that TGF-β signaling is dispensable for healthy hematopoiesis and healthy function of HSCs in adult mice. These unexpected findings demonstrate a highly context-dependent role of TGF-β in HSC biology and illustrate the limitations of in vitro assays to study the complex regulation of hematopoietic stem and progenitor cells. However, our data on HSC regulation in vivo does not exclude that interfering with TGF-β signaling will play a role in future protocols for ex vivo expansion of HSCs. Our in vitro culture experiments under serum-free conditions support the notion that blocking TGF-β signaling ex vivo may indeed enhance proliferation of HSCs. However, because the BM microenvironment naturally supports the regeneration of HSCs, a better understanding of how TGF-β and other factors regulate HSCs in this context, along with possible mechanisms of cross talk or redundancy, is certainly required to develop novel and better strategies for HSC expansion.

We thank R. Fässler and C. Brakebusch for providing Mx-Cre transgenic mice and for helpful discussions; W. Badn and M. L. Selenica for valuable experimental contributions; L. Wittman for help with transplantation experiments; S-E. Jacobsen and members of the Department of Stem Cell Biology, Lund University, for helpful advice and discussions; Peter ten Dijke for expert advice on TGF-β signal transduction; and H. Mikkola for critical reading of the manuscript.