|Home | About | Journals | Submit | Contact Us | Français|
Roundabout (Robo) family proteins are immunoglobulin-type cell surface receptors that are expressed predominantly in the nervous system. The fourth member of this family, Robo4, is distinct from the other family members in that it is expressed specifically in endothelial cells. In this study, we examined the expression of Robo4 in hematopoietic stem cells (HSCs) and its possible role in HSC regulation. Robo4 mRNA was specifically expressed in murine HSCs and the immature progenitor cell fraction but not in lineage-positive cells or differentiated progenitors. Moreover, flow cytometry showed a correlation between higher expression of Robo4 and immature phenotypes of hematopoietic cells. Robo4high hematopoietic stem/progenitor cells presented higher clonogenic activity or long-term repopulating activity by colony assays or transplantation assays, respectively. A ligand for Robo4, Slit2, is specifically expressed in bone marrow stromal cells, and its expression was induced in osteoblasts in response to myelosuppressive stress. Interestingly, overexpression of Robo4 or Slit2 in HSCs resulted in their decreased residence in the c-Kit+Sca-1+Lineage−-side population fraction. These results indicate that Robo4 is expressed in HSCs, and Robo4/Slit2 signaling may play a role in HSC homeostasis in the bone marrow niche.
Roundabout (Robo) family proteins are immunoglobulin-type cell surface receptors that are expressed predominantly in the nervous system . Slit, a ligand for Robo, is a large secreted protein that is also expressed in brain . The Robo family comprises four family members, Robo1–Robo4, and the Slit family consists of three family members, Slit1–Slit3. Robo and Slit were first described in Drosophila as critical molecules in axon path finding and migration of neuronal cells . In mammals, Slit/Robo signaling acts as a repulsive axon guidance cue for developing neurons and inhibits neuronal migration, thus playing a critical role for correct wiring of the neuronal network . In the hematopoietic system, Slit2 has been shown to inhibit chemotactic migration of lymphocytes induced by SDF-1 . In addition, Slit2/Robo1 signaling plays a critical role in tumor angiogenesis . Robo4 was first identified by in silico database mining as a homolog of Robo1 [6, 7]. Robo4 is unique in its expression pattern that is specific to endothelial cells, whereas it shows functional similarity to other Robo family members, such as a binding with Slit2 or an inhibitory effect for cellular migration . In hematopoiesis, Forsberg et al. have reported an extensive transcriptome analysis of long-term (LT)-hematopoietic stem cells (HSCs), short-term (ST)-HSCs, and multipotent progenitors (MPPs) and showed that Robo4 is one of the differentially expressed genes among these three primitive hematopoietic cell populations . However, differential expression of Robo4 in whole hematopoietic system has not been examined.
HSCs are a rare population of cells that can support life-long hematopoiesis. They are characterized by their unique capacity to self-renew and differentiate into all blood cell lineages. HSCs reside in the specific microenvironment known as the niche in the adult bone marrow (BM). The niche is thought to be located on the surface of trabecular bones, and osteoblasts lining the surface of these bones are reported to be one of the critical niche components [9, 10]. Side population (SP) phenotype, identified by as the activity of Hoechst 33342 dye efflux, is one of the hallmarks of quiescent HSCs in the BM niche [11, 12]. It has been shown that many of the quiescent HSCs reside in the c-Kit+Sca-1+Lineage− (KSL)-SP, and angiopoietin-1/Tie-2 signaling induces HSC quiescence and increases cells in the KSL-SP . Moreover, quiescent HSCs move from the SP to a main population (MP) of non-SP cells when they are recruited into the cell cycle upon myelosuppressive stimuli, such as 5-fluorouracil (5-FU) treatment. However, the molecular mechanism regulating this transition remains unclear.
In a search for novel surface molecules expressed on HSCs, we found that Robo4 was highly expressed in the HSCs and the immature progenitor cell fraction. Moreover, Slit2 was specifically expressed in bone marrow stromal cells, and interestingly, the expression was induced in osteoblasts in response to myelosuppressive stimuli, such as 5-FU treatment. Surprisingly, enhanced Slit2/Robo4 signaling in HSCs resulted in the shift of their residence from SP to non-SP. These results suggest a possible involvement of Slit2/Robo4 signaling in the regulation of HSC homeostasis by the BM niche.
MS10, PA6, and OP9 cells [14, 15] were cultured as previously described . Primary BM stromal cells were obtained by culturing the adherent fraction of whole BM cells from B6 mice in α-minimal essential medium/10% fetal bovine serum on the culture dish and expanding them for 2 weeks.
Messenger RNA (mRNA) was isolated using a Micro-FastTrack 2.0 mRNA Isolation Kit (Invitrogen, Carlsbad, CA, http://www.invitrogen.com) according to the manufacturer's protocol. First-strand cDNA was synthesized with SuperScript II reverse transcriptase (Invitrogen). The quantity of cDNA was normalized according to the expression of glyceraldehyde-3-phosphate dehydrogenase measured by real-time reverse transcription (RT)-polymerase chain reaction (PCR) using a Light Cycler Fast Start DNA SYBR Green I kit (Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com). Semiquantitative RT-PCR was performed using Ex Taq-HS polymerase (Takara Bio, Shiga, Japan, http://www.takara-bio.com). The sequences of primers used for RT-PCR are shown in supporting information Table 1.
Anti-mouse c-Kit-allophycocyanin (APC) (2B8), Sca-1-fluorescein isothiocyanate (FITC) (D7), Sca-1-APC (D7), and CD34-FITC (RAM34) antibodies were from BD Pharmingen (San Diego, http://www.bdbiosciences.com/index_us.shtml). Anti-mouse c-Kit-phycoerythrin (PE)-Cy7 (2B8) was from BioLegend (San Diego, http://www.biolegend.com). Anti-mouse Robo4 monoclonal antibodies were raised against the extracellular domain of Robo4 by immunizing rats with Y3Ag1.2.3 cells expressing mouse Robo4. The purified antibodies were labeled with PE using PhycoLink R-Phycoerythrin Conjugation Kit (PROzyme, San Leandro, CA, http://www.prozyme.com).
BM cells were obtained by flushing out femurs and tibias from 8–12-week-old mice with phosphate-buffered saline (PBS). Depletion of lineage-positive cells and staining and fluorescence-activated cell sorting (FACS) of KSL and CD34−KSL cells were described previously . Stained cells were analyzed by FACSAria or FACSVantage (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com). Hoechst staining of lineage depleted cells was performed with 5 μg/ml Hoechst 33342 (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) at 37°C for 90 minutes as previously described by Goodell et al. .
Retrovirus was produced as previously described . Briefly, retrovirus plasmids  were transiently transfected into PLAT-E  cells using Fugene (Roche Diagnostics), and retrovirus super-natant was collected after 48 hours of transfection. Infection of BM cells with retrovirus was carried out using Retronectin (Takara Bio) according to the manufacturer's protocol. Bone marrow transplantation of infected cells was performed as previously described  using B6-Ly5.1 mice (Sankyo Lab Service Co., Tokyo, http://www.sankyolabo.co.jp) as donors and B6-Ly5.2 mice (CLEA Japan, Tokyo, http://www.clea-japan.com) as recipients and competitors. For competitive repopulation assay of Robo4+KSL or Robo4−KSL cells, 100 cells for each group were sorted from BM mononuclear cells of B6-Ly5.1 mice and transplanted with 2 × 105 competitor cells (whole BM cells from Ly5.2 mice) into lethally irradiated (950R) recipient mice (B6-Ly5.2). Eight- to 12-week-old mice were used in all experiments. All animal experiments were reviewed and approved by the institutional review board of the Institute of Medical Science, University of Tokyo.
For the colony-forming cell assay, cells were sorted by FACS and deposited into MethoCult GF M3434 (StemCell Technologies, Vancouver, BC, Canada, http://www.stemcell.com). The number and type of colonies were assessed at day 7 of culture.
Mice treated with 5-FU (150 mg/kg, i.p.) for 3 days were fixed by perfusing them with Zamboni's fixative. Femurs were then dissected and decalcified in 10% EDTA at 4°C. The bones were embedded in Tissue-Tek OCT Compound (Sakura Finetek, Torrance, CA, http://www.sakura.com), frozen, and sectioned in 5-μm-thick slices using a cryostat (Leica, Heerbrugg, Switzerland, http://www.leica.com). Sections were blocked with biotin and normal donkey serum and stained with anti-Slit2 (G-19; Santa Cruz Bio-technology Inc., Santa Cruz, CA, http://www.scbt.com) and anti-osteopontin (Takara Bio) antibodies, which were then subjected to secondary staining with anti-goat IgG-biotin and anti-mouse IgGCy3 and tertiary staining with streptoavidin-Cy2. Sections were washed three times for 5 minutes in PBS between steps and finally mounted in Vectashield anti-fading medium (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com) containing 4,6-diamidino-2-phenylindole (Sigma-Aldrich) for nuclear labeling. Fluorescent images were examined and captured using a laser confocal microscope (Olympus, Tokyo, http://www.olympus-global.com).
All statistical analyses were performed by unpaired Student's t test using GraphPad Prism software (GraphPad Software, La Jolla, CA, http://www.graphpad.com).
During a search for novel surface molecules expressed on HSCs, we noticed that cells in the hematopoietic and the nervous systems share a number of expressed genes [20-22]. These observations prompted us to investigate the expression of Robo family genes in the hematopoietic system. We first checked the expression of Robo family genes by RT-PCR. Interestingly, Robo4 was specifically expressed in murine KSL cells, which contain HSCs and immature hematopoietic progenitors. However, it was not expressed in lineage marker-positive (Lin+) cells or various early hematopoietic progenitors, such as common myeloid progenitor, megakaryocyte/erythroid progenitor, granulocyte/monocyte progenitor, and common lymphoid progenitor (Fig. 1A, 1B). These data suggest that Robo4 is specifically expressed in murine HSCs and immature hematopoietic progenitors. In fact, Robo4+CD34−KSL cells induced to differentiate in vitro no longer expressed Robo4 (supporting information Fig. 1), indicating that Robo4 is downregulated along with differentiation. Collectively, these results suggest that Robo4 is expressed in murine HSCs and immature hematopoietic progenitors, and its expression declines as the cells differentiate.
We then generated a monoclonal antibody specific for Robo4 and examined its expression by flow cytometry (FACS). The antibody was specific to Robo4 (supporting information Fig. 2) and was not cross-reactive with other Robo family members (data not shown). Compatible with RT-PCR data, Robo4 was not expressed in lineage marker-positive cells in the murine BM (data not shown). As shown in Figure 2A, the highest expression of Robo4 was detected in the tip population of KSL cells (fraction a), which is enriched in HSCs. Furthermore, CD34−KSL cells, a highly enriched population of long-term HSCs, expressed higher Robo4 compared with CD34+KSL cells, an enriched fraction of short-term HSCs (Fig. 2B). Conversely, Robo4highKSL cells contained more CD34−KSL cells compared with Robo4low/−KSL cells (13.6% vs. 4.7%, respectively) (Fig. 2C). Taken together, these results confirm high expression of Robo4 in murine HSCs and immature hematopoietic progenitors.
We next examined the physiological relevance of Robo4 expression and the HSC activities. As shown in Figure 3A and 3B, Robo4highCD34−KSL cells presented higher multilineage differentiation (shown as nmEM in Fig. 3A) and proliferative potential (shown as high proliferating potential-colony-forming cells in Fig. 3B) compared with other fractions. Interestingly, transplantation assays revealed that long-term repopulating activity of KSL cells was detected only in the Robo4high fraction and not in the Robo4low/− fraction (Fig. 4). These results suggest that higher expression of Robo4 defines a population of cells with higher capacities for multilineage differentiation, proliferation, and long-term repopulation.
To obtain clues for the role of Robo4 in HSC regulation, we examined the expression of Slit2, a ligand for Robo4, by RT-PCR. As shown in Figure 5A, Slit family genes were not expressed in hematopoietic cells, such as KSL or Lin+ cells. Interestingly, however, Slit2 was expressed in murine BM stromal cell lines (Fig. 5B) and the primary BM stromal cells (Fig. 5C). Of note, primary BM stromal cells also showed slight or moderate expression of Slit 1, Slit3, and Robo1. These results suggest that Slit2 is expressed in the BM microenvironment.
We went on to investigate the role of Slit2-Robo4 signaling in HSC physiology by examining the expression of Slit2 in the BM under a stress condition. We treated mice with a cytotoxic agent, 5-FU, and periodically monitored the expression of Slit2 in the BM by RT-PCR. Interestingly, Slit2 was transiently induced in the bone marrow on days 3–6 after 5-FU treatment, whereas Slit1 and Slit3 were not induced at all (Fig. 6A). Fractionation of BM cells by CD45 expression revealed that the induction occurred in CD45− cells, suggesting that Slit2 was induced in nonhematopoietic cells (Fig. 6B). Of note, a slight induction of Slit2 was also seen in CD45+ cells after 9 days of 5-FU administration. The expression of Robo4 was also upregulated in the BM after 5-FU treatment, probably because of the concentration of immature hematopoietic cells in the BM (supporting information Fig. 3).
Given that Slit2 is induced in CD45− cells, we speculated that Slit2 might be induced in the BM niche in response to myelosuppressive stress. To test this hypothesis, we performed immunostaining of Slit2 in the BM treated with 5-FU for 3 days. As shown in Figure 6C, Slit2 was specifically induced in the osteoblasts at the endosteal surface of the bone marrow, where the HSC niche was reported to be located. Of note, expression of Slit2 was not detected in other places in the BM, such as endothelial cells (data not shown). These results suggest that Slit2-Robo4 signaling may play a role in HSC physiology in the osteoblastic niche upon myelosuppressive stress.
To gain more insight into the physiological role of Slit2/Robo4 signaling in HSCs, we enhanced Slit2/Robo4 signals by overexpressing Robo4 or Slit2 in the hematopoietic stem/progenitor cells and examined its effect on their SP phenotype. We retrovirally transduced Robo4 or Slit2 into the BM cells taken from 5-FU-treated mice and transplanted them into lethally irradiated recipients (Fig. 7A, 7B; supporting information Fig. 4). Robo4-or Slit2-expressing cells homed to and engrafted to the recipient's BM, albeit to a lesser extent compared with the mock control, and their contribution to the peripheral blood progressively decreased over time (supporting information Fig. 5). To our surprise, the percentage of KSL-SP was significantly lower in cells expressing Robo4 or Slit2 (green fluorescent protein [GFP]+) compared with cells expressing mock vector (GFP+) or nontransfected donor-derived cells (GFP−) after 12 weeks of transplantation. In contrast, the percentage of KSL-MP was increased or remained constant in Slit2- or Robo4-transfected populations. On the basis of these observations, we speculated that Slit2/Robo4 signals might be acting to drive cells in KSL-SP into the KSL-MP or KSL-non-SP fraction. It is also possible that KSL-SP cells underwent apoptosis by enhanced Slit2/Robo4 signals. However, we speculate this is not the case, since we did not see any increased apoptosis in Robo4- or Slit2-overexpressing cells. In support of above hypothesis, Robo4 was highly expressed in KSL-SP cells (supporting information Fig. 6), indicating that these cells are sensitive to Slit2 stimuli. Taken together, these findings suggest that Slit2/Robo4 signaling promotes SP to non-SP transition of HSCs.
During the search for novel surface molecules expressed on primitive hematopoietic cells, we noticed that cells in the hematopoietic and nervous systems share a number of expressed genes. For example, we have previously reported that mKirre, which is abundantly expressed in brain and BM stromal cells, plays a critical role in maintaining HSC functions [20, 21]. Ephrins and their receptors, Ephs, are expressed in both the nervous and the hematopoietic systems, and they play critical roles in various aspects of neurogenesis and hematopoiesis [23-25]. Goolsby et al. also reported that hematopoietic progenitors express a set of neural genes . These observations prompted us to focus on Robo family proteins, cell surface receptors that play a critical role in the nervous system. Within this family, Robo4 drew our particular attention, as the expression of Robo4 was specific to endothelial cells (ECs), and many HSC markers, such as CD34 and Flk1, were shared with ECs [26-28]. As expected, this study demonstrated that Robo4 is highly expressed in murine HSCs and immature hematopoietic progenitors. Moreover, transplantation experiments revealed that KSL cells with high Robo4 expression possessed higher long-term repopulating activity. These results demonstrate that the expression of Robo4 correlates with higher HSC capacities, such as long-term repopulation and differentiation to multiple lineages. During the course of our study, Forsberg et al. reported an extensive transcriptome analysis of LT-HSCs, ST-HSCs, and MPPs and showed that Robo4 is one of the differentially expressed genes among these three primitive hematopoietic cell populations . In the current study, however, we demonstrated that the expression of Robo4 is highest in LT-HSCs not only among primitive hematopoietic cells, but also among all hematopoietic cells, including various progenitors and lineage marker-positive cells. In addition, we confirmed this observation on both the mRNA and the protein levels. Furthermore, we demonstrated that Robo4 could be used for isolating HSCs with long-term repopulating potential when combined with KSL phenotype.
Robo/Slit signaling acts as a repulsive axon guidance cue and inhibits migration of neuronal cells. In addition, Slit2 inhibits migration of endothelial cells through Robo4 . On the basis of these published observations, it is reasonably speculated that Robo4 signaling inhibits HSC migration. It was also reported that Drosophila Slit/Robo signaling inhibits N-cadherin function in mammalian cells . Since N-cadherin has been shown to play critical roles in the adhesion of HSCs and osteoblasts in the BM niche [9, 13], we also speculated that Robo4/Slit2 signaling might regulate HSC-osteoblast interaction upon myelosuppression. However, our preliminary analysis indicated that Slit2 did not inhibit migration of KSL cells toward SDF-1, and Robo4/Slit2 signaling did not inhibit N-cadherin-mediated adhesion in mammalian cells (unpublished observation). These data do not necessarily exclude the possibility that Robo4/Slit2 system is involved in migration or adhesion of HSCs in the physiological settings, and additional studies using gene-deficient animals will be required to unravel the precise roles of Robo4/Slit2 signaling in the HSC dynamics.
In a similar vein as the regulation of HSCs by the BM niche, it is intriguing that enhanced Slit2/Robo4 signaling resulted in the decreased residence of HSCs in the SP fraction. The SP phenotype is defined as a cell population with a high capacity of Hoechst 33342 dye efflux, and the efflux of Hoechst dye is accomplished by one of the ATP-binding cassette transporters, ABCG2/Bcrp-1 . A gene disruption study in mice revealed that Bcrp-1 is absolutely required for the SP phenotype ; however, the expression of Bcrp-1 is equally observed in SP and non-SPs of CD34−KSL cells, suggesting that the function of Bcrp-1 is regulated by other factors . Our finding that enhanced Slit2/Robo4 signaling led to decreased HSC residence in the SP fraction indicates that Slit2 could be a candidate factor regulating the Bcrp-1 function. In this regard, the induction of Slit2 in osteoblasts upon myelosuppression is of particular interest. It has been reported that HSCs are recruited from the SP to the non-SP fraction on days 4–6 after 5-FU treatment . The induction of Slit2 occurs on days 3–6 after 5-FU administration, which precedes or coincides with the SP to non-SP transition of HSCs. These data suggest that Slit2 induced in the BM niche upon myelosuppression acts on HSCs through Robo4 and may play a critical role in their transition from the SP to the non-SP fraction. Preliminary analysis of Robo4-deficient mice showed that there is only a slight difference, if any, in the proportion of KSL-SP cells between knockout and wild-type animals (data not shown). This suggests that other redundant pathways, in addition to Robo4, could regulate the SP to non-SP transition of HSCs, and further detailed analysis of Robo4-deficient mice would be required to unravel precise roles of Robo4 in vivo. Considering a well-recognized role of Robo/Slit signaling in cellular migration , it is also tempting to speculate that Slit2 might be acting to induce HSC migration out of the niche in response to myelosuppression.
We identified Robo4 as a receptor expressed on hematopoietic stem cells that is potentially involved in the regulation of the SP phenotype. Revealing the physiological role of Robo4/Slit2 signaling would lead to a better understanding of HSC homeostasis in the BM niche.
We thank N. Watanabe, T. Shibata, and S. Saito (FACS Core Laboratory, Institute of Medical Science, University of Tokyo) for FACS sorting, and Dr. Dovie R. Wylie for language assistance. This work was supported in part by a grant from the Ministry of Education, Culture, Sports, Science and Technology of Japan. H.N. is currently affiliated with the Division of Hematology, Department of Internal Medicine, Keio University School of Medicine, Tokyo, Japan.
F.S. and Y.G.-K. contributed equally to this work.
Disclosure of Potential Conflicts of Interest
J.P.H. was employed by R&D Systems.