Spinal cord has served as an excellent model for studying molecular and genetic control of cell fate specification and differentiation. At an early stage of spinal cord development, the homogeneous neuroepithelial cells undergo dorsal–ventral patterning by signaling molecules produced from both the dorsal and ventral midline structures. It has been proposed that Sonic hedgehog produced from the notochord and floor plate functions as a morphogen to selectively induce or repress expression of homeodomain transcription factors (TFs) in a concentration-dependent manner, creating a nested pattern of TF expression in the ventral neural progenitor cells (Roelink et al., 1995; Ericson et al., 1997). The combinatory expression of these TFs constitutes a homeodomain code that divides the ventral neuroepithelium into five distinct domains (pMN, p0–p3), with each domain expressing a unique combination of TFs and generating a specific neuronal subtype (motor neurons or MNs, V0–V3 ventral interneurons; Briscoe et al., 2000). Different subtypes of ventral neurons can be readily identified by their expression of specific postmitotic homeodomain TFs, such as HB9 (MNs), Evx1 (V0), En1 (V1), and Chx10 (V2). Recent studies have suggested that these homeodomain proteins function primarily as postmitotic determinants of neuronal subtype identities. For instance, HB9 and Evx1 are known to be required for the consolidation of motor neuron fate and V0 interneuron fate, respectively (Arber et al., 1999; Thaler et al., 1999; Moran-Rivard et al., 2001).
In the dorsal spinal cord, several members of the bone morphogenetic protein (BMP) family are produced in the dorsal midline structures (ectoderm and roof plate; Lee and Jessell, 1999). There is both in vitro and in vivo evidence for a gradient of BMP-dependent activity in controlling neuronal subtype identities (Barth et al., 1999; Timmer et al., 2002). The gradient BMP signaling, together with other signal molecules (e.g., Wnt), could be responsible for the differential expression of many TFs in the dorsal neural progenitor cells. Intriguingly, many of the TFs that are involved in the dorsal neural patterning are members of the basic helix-loop-helix (bHLH) family, known as proneural bHLH factors, which include Math1, Ngn1, Ngn2, and Mash1 (Gowan et al., 2001). Based on the combinatorial expression of certain proneural bHLH factors and other homeodomain factors, the dorsal neuroepithelium is divided into six progenitor domains (dp1–6), from which arise six early-born (E9.5–E12.0) dorsal interneuron populations called dI1–6 neurons (Gross et al., 2002; Muller et al., 2002). Like their ventral counterparts, different subclasses of dorsal interneurons express distinct combinations of postmitotic homeodomain TFs, which are similarly involved in cell fate consolidation (for reviews, see Matise, 2002; Caspary and Anderson, 2003; Helms and Johnson, 2003). For instance, Lbx1 is specifically expressed in dI4–6 neurons and is required for their ultimate fate determination (Gross et al., 2002; Muller et al., 2002).
Interestingly, two later-born (E12–E14) neuronal populations, termed dILA and dILB, are generated from the Mash1+/Gsh1/2+ dp3–5 domains of dorsal neuroepithelium in a salt and pepper manner (Mizuguchi et al., 2006; Wildner et al., 2006). While early neurons are born at embryonic day (E) 9.5–E12, dILA and dILB neurons are mostly generated between E12 and E14. Soon after their birth, both dILA and dILB neurons migrate into the dorsal superficial layers where they intermingle and differentiate into association interneurons. Although the dILA and dILB neurons appear to be generated from the same neural progenitor cells and both express Lbx1, they can be readily distinguished by their expression of distinct combination of other homeodomain TFs. While dILA neurons express Pax2 and Lim1/2, the dILB neurons express Brn3a, Lmx1b, and Tlx3 (Gross et al., 2002; Muller et al., 2002). However, at later stages, the postmitotic TFs seem to be differentially down-regulated in dILB neurons so that Brn3a+ and Lmx1b+/Tlx3+ neurons are found in distinct laminae of the dorsal horn (Gross et al., 2002; Muller et al., 2002).
Despite that many bHLH factors are involved in the fate specification of neural progenitor cells, there is growing evidence that bHLH factors also participate in the differentiation of neuronal subtypes. For instance, Bhlh4 is transcribed in bipolar neurons in the retina (Bramblett et al., 2002), and its expression is required for rod bipolar cell maturation (Bramblett et al., 2004). Recently, Bhlhb5, a closely related member of Bhlhb4 in the Olig family (Xu et al., 2002; Bertrand et al., 2002), has also been reported to be expressed in embryonic central nervous system (CNS) and in other tissues as well (Brunelli et al., 2003). However, its expression in the developing nervous system has not been characterized at the cellular level, and its function in neural development remains unknown. As a first step to elucidate its role in neuronal differentiation, we developed a specific polyclonal antibody against the Bhlhb5 protein and performed detailed studies on its expression in the developing spinal cord. Here we report that Bhlhb5 is selectively expressed in subsets of spinal interneurons, specifically the early-born dI6, V1 and V2 neurons, and a subpopulation of late-born (E12.5 and later) dILA and dILB dorsal interneurons that settle down in the superficial laminae relatively late.
It was previously reported that Bhlhb5 mRNA is transcribed in several embryonic tissues, including the neural tube (Brunelli et al., 2003). However, its spatial and temporal patterns of expression in the CNS have not been analyzed in detail. To fully characterize the expression and regulation of Bhlhb5 in the developing spinal cord, we set out to develop polyclonal antibodies against the N′-terminal sequence of the Bhlhb5 protein. Western immunoblotting with the affinity-purified antibody showed that it recognized a single protein band of the expected size (∼39 kDa) in HEK293 cells transfected with pcDNA4-BHLHB5 expression vector, but not with the control vector (Fig. 1A). The antibody specificity was further substantiated by the virtually identical staining patterns of Bhlhb5 in situ RNA hybridization and the anti-Bhlhb5 immunofluorescence on adjacent sections prepared from E10.5 and E17.5 mouse spinal cord tissues (Fig. 1B–E). At E10.5, Bhlhb5+ cells were located in the gray matter adjacent to the ventricular zone mostly in the ventral spinal cord (Fig. 1B,C). At E17.5, Bhlhb5+ cells remained confined to the gray matter but were found in both the ventral and dorsal regions (Fig. 1D,E). The expression of Bhlhb5 in the gray matter persisted in young postnatal spinal cord (Fig. 1F–H) but was eventually down-regulated in the adult (data not shown). The restriction of Bhlhb5 expression to the gray matter strongly suggested that Bhlhb5 is likely to be expressed in neurons. Consistent with this idea, many Bhlhb5+ neurons gradually gained the expression of the neuronal marker NeuN (Mullen et al., 1992) as they migrated away from the ventricular zone (Fig. 2), indicating the Bhlhb5 is initially expressed in young immature neurons.
To define the specific neuronal subtypes that express Bhlhb5 at the early stage of spinal development, we first investigated whether Bhlhb5+ neurons can be generated from dorsal neural progenitor cells by examining the expression of Bhlhb5 in relation to that of Pax7 and Mash1 in E10.5 spinal cord. At this stage, Pax7 is expressed in the entire dorsal neuroepithelium (dp1–6), whereas Mash1 specifically marks the dp3–5 domains of dorsal neuroepithelial cells (Fig. 3B; Matise, 2002; Helms and Johnson, 2003; Caspary and Anderson, 2003). Double-labeling experiments revealed that, in the dorsal region, Bhlhb5 was expressed in a small number of neurons flanking the lowest region of the Pax7+ dorsal neuroepithelial cells (Fig. 3A) but below the Mash1+ neural progenitor cells (Fig. 3B), suggesting that Bhlhb5 is selectively expressed in dI6 interneurons derived from the Pax7+/Mash1− neural progenitor cells. Consistently, this group of Bhlhb5+ neurons also coexpressed the dI4–dI6 marker Lbx1 (Fig. 3C), but not Brn3a (Fig. 3D), which marks dI1–3 and dI5 neurons or the dI5-specific marker Lmx1b (Fig. 3E). The dI6 identity of Bhlhb5+ neurons in the dorsal spinal cord was further substantiated by their coexpression of Pax2 and Lim1/2 (Fig. 3F,G), two homeodomain TFs that are expressed in dI6 and other interneurons as well (Fig. 3H).
At E10.5, Bhlhb5+ neurons were also observed in the ventral spinal cord, and an apparent gap was detected between the dorsal and ventral Bhlhb5+ neurons at approximately the position of V0 interneuron. Double-labeling experiments revealed that, in the ventral part, Bhlhb5 neurons were generated immediately adjacent to the Pax6+ neuroepithelial cells (Fig. 4A) above the Olig2+ pMN domain (Fig. 4B). Thus Bhlhb5 is likely to be expressed in the V1 and V2 ventral interneurons interposed between V0 and MNs. Consistent with this idea, Bhlhb5+ neurons coexpressed the V1 marker EN1 (Fig. 4D) and V2 marker Chx10 (Fig. 4E), but not the V0 marker Evx1 (Fig. 4C) nor the MN marker HB9 (Fig. 4F). In further support, Bhlhb5+ neurons in the V1 position also coexpressed Lim1/2 and Pax2, two postmitotic TFs expressed in V0 and V1 interneurons in the ventral spinal cord (Fig. 3F,G).
Of interest, at later stages of embryogenesis (E17.5), Bhlhb5 expression was also detected in the dorsal spinal cord, primarily in the superficial layers (laminae 1–3) of the dorsal horn (Fig. 1D,E). Recent studies demonstrated that neurons that populate this region are predominantly the late-born (after E11) dILA and dILB dorsal interneurons that originate from the Mash1+/Gsh+ dp3–dp5 domains and subsequently migrate dorsally into the superficial laminae (Muller et al., 2002; Gross et al., 2002). To test the possibility that Bhlhb5 is transcribed in dILA and/or dILB neurons, we examined the dorsal expression of Bhlhb5 in relation to the dILA marker Pax2 and the dILB markers Brn3a, Lmx1b, or Tlx3. At E14.5, a small number of Bhlhb5+ cells were detected in the uppermost layers populated by Pax2+ dILA neurons (Fig. 5A), and Lmx1b+/Tlx3+ dILB neurons (Fig. 5B,C). In this region, a few Bhlhb5+ cells coexpressed Pax2, but not Lmx1b or Tlx3. In contrast, the majority of Bhlhb5+ cells in the dorsal horn were situated in the more ventral layers rich in Bran3a+ neurons, with some of Bhlhb5+ cells coexpressing Brn3a (Fig. 5D). Later at E17.5, most Bhlhb5+ neurons were found in the more dorsal layers, coexpressing either the dILA marker Pax2, or the dILB markers Lmx1b, Tlx3, or Brn3a (Fig. 5E–H). However, Bhlhb5 expression was only detected in a fraction of dIL neurons in the dorsal layers at this stage. Together, these results suggest that Bhlhb5+ neurons progressively migrate into the superficial laminae of the dorsal horn between E14.5 and E17.5.
The late arrival of Bhlhb5+ neurons in the superficial layers as compared with other Pax2+ and Lmx1b+ dIL neurons has raised the possibility that they represent a late-born population of dorsal association interneurons. To examine this possibility, we birthdated the dorsal Bhlhb5 neurons by administrating a single injection of bromodeoxyuridine (BrdU) into pregnant mice at E11.5, E12.5, or E13.5. Spinal cord tissues were collected at E18.5 and subject to double immunostaining with anti-BrdU and anti-Bhlhb5. In embryos that received BrdU injection at E11.5, numerous BrdU+ cells were observed in the dorsal horn, but few Bhlhb5+ neurons in this region were BrdU+ (Fig. 6A,D). However, in E12.5-labeled embryos, a large number of Bhlhb5+ neurons in the dorsal horn were strongly immunoreactive to BrdU (Fig. 6B,E). In E13.5-injected embryos, despite a much reduced number of BrdU+ cells in the dorsal horn, Bhlhb5+/BrdU+ double-positive neurons were still detected in the superficial layers (Fig. 6C,F). Double immunostaining further confirmed that a small number of Bhlhb5+ neurons started to be produced at E12.5 from the Mash1+ neuroepithelial cells (Fig. 6G–I), and the number was dramatically increased at E13.5 (Fig. 6J–L). Taken together, these BrdU birthdating and double-immunostaining experiments suggested that Bhlhb5+ dIL neurons were predominantly generated from the dorsal neural progenitor cells at E12.5 and later.
In this study, we report the detailed analyses of Bhlhb5 expression in the developing spinal cord by immunostaining with a newly generated anti-Bhlhb5 polyclonal antibody. The affinity-purified antibody appears to be specific to the Bhlhb5 protein based on the detection of a single protein band of the expected size by Western immunoblotting in cells transfected with Bhlhb5 expression vector and the identical patterns of Bhlhb5 in situ RNA hybridization and immunostaining signals (Fig. 1). The Bhlhb5 antibody did not seem to cross-react with other related proteins, as its immunostaining pattern did not overlap with those of other members of the Olig family, which includes Olig1, Olig2, Olig3, Bhlhb4, and Bhlh5 itself (Bertrand et al., 2002). For instance, in E10.5 spinal cord, Olig1 and Olig2 are expressed in the pMN domain of the ventral neuroepithelium (Fig. 3B; Takebayashi et al., 2000; Lu et al., 2000; Zhou et al., 2000), whereas Olig3 is expressed in the dI1–3 domains of the dorsal neuroepithelium (Takebayashi et al., 2002; Muller et al., 2005). Bhlhb4 is specifically expressed in the diencephalon–mesencephalon boundary, but not in the spinal cord (Bramblett et al., 2002).
Comparison of Bhlhb5 expression with other well-characterized markers for neural progenitor cells and neuronal subtypes revealed a dynamic pattern of Bhlhb5 expression in embryonic mouse spinal cord. At early stage of spinal cord development (E10.5), Bhlhb5 is expressed in postmitotic immature neurons derived from the dp6, p1 and p2 domains of neural progenitor cells (Figs. 3, 4). Indeed, Bhlhb5+ neurons did not incorporate BrdU in a 2-hr short pulse chase and did not coexpress the mature neuronal marker NeuN (data not shown). However, at E12.5, coexpression of Bhlhb5 and NeuN can be clearly detected in the ventral spinal cord as Bhlhb5 neurons migrate into the mantle zone (Fig. 2).
At later stages, Bhlhb5 expression is also up-regulated in a small number of late-born dIL dorsal interneurons in the superficial layers, coexpressing dIL markers Pax2, Brn3a, Lmx1b, or Tlx3 (Fig. 5). Pervious studies showed that the majority of dIL neurons were born between E11 and E13 (Gross et al., 2002; Muller et al., 2002). Our BrdU birthdating studies revealed that the Bhlhb5+ dorsal association neurons are mostly generated at and after E12.5, with many being produced as late as E13.5 (Fig. 6). Indeed, at E12.5, a small number of Bhlhb5+ neurons started to emerge from the Mash1+ dorsal neural progenitor cells (arrows in Fig. 6G–I). However, this number was dramatically increased at E13.5 (Fig. 6J–L), suggesting that, like other dIL neurons, Bhlhb5+ dorsal association neurons are also generated from the dp3–5 neural progenitor cells. In keeping with the idea that Bhlhb5 labels a late-born population of dIL neurons, they migrate into the dorsal superficial laminae later than most of other dIL neurons. At E14.5, Bhlhb5+ neurons in the dorsal spinal cord were largely localized in the Brn3a+ layer ventral to the dorsal layers populated by Pax2+ and Lmx1b+ dIL neurons (Fig. 5A–D). By E17.5, a majority of Bhlhb5+ neurons were found in the more superficial layers intermingling with other dILA and dILB neurons (Fig. 5E–H), indicating that Bhlhb5+ interneurons migrate into the uppermost layers between E14.5 and E17.5.
The function of Bhlhb5 in neuronal development in the spinal cord is not known at this stage. Previous work showed that mouse Bhlhb5 protein can repress the human Pax6 promoter (Xu et al., 2002). Consistent with this observation, Pax6 expression is down-regulated in Bhlhb5+ neurons as they leave the ventricular zone. Given that Bhlhb5 is initially expressed in NeuN-negative immature neurons before terminal differentiation, it is quite possible that Bhlhb5 may participate in neuronal fate specification and differentiation. Recent studies have indicated that expression of bHLH TFs during the differentiation process can contribute to the specification of distinct neuronal identities (Lee and Pfaff, 2003; Helms et al., 2005). A more direct support for the possible role of Bhlhb5 in the control of neural specification and differentiation came from the most recent observation that Bhlhb5 function is required for the correct specification of γ-aminobutyric acid (GABA)ergic amacrine and Type 2 OFF-cone bipolar subtypes in the developing retina (Feng et al., 2006).
Recent studies have shown that many postmitotic transcription factors are involved in the control or regulation of the synthesis of particular neurotransmitters. For instance, Pft1a and Pax2 specify GABAergic neurons (Glasgow et al., 2005), whereas Tlx1/Tlx3 homeodomain transcription factors produce glutamatergic neurotransmitter phenotype (Cheng et al., 2005). All early-born Bhlhb5+ neurons and many late-born Bhlhb5+ dorsal association neurons coexpressed Pax2 and, thus, are likely to be GABAergic neurons (Fig. 5; Cheng et al., 2004). However, a small number of Bhlhb5+ dorsal interneurons appear to belong to the glutamatergic dILB neurons as they coexpressed Lmx1b and Tlx3 (Fig. 5; Cheng et al., 2004). Thus, Bhlhb5 expression is not directly related to the switch of GABAergic versus glutamatergic neurotransmitter phenotypes. However, it cannot be ruled out that Bhlhb5 may regulate other neurotransmitter phenotypes (e.g., serotoninergic, cholinergic, or peptidergic). In addition, it is also conceivable that Bhlhb5 may be involved in other developmental processes such as neuronal migration, axonal pathfinding, or circuitry formation. Further functional studies are required to elucidate the role of Bhlhb5 in neuronal development in the developing spinal cord.
Generation of Anti-Bhlhb5 Antibody
The 5′-terminus sequence (residues 2–213) and 3′-terminus sequence (277–356) of Bhlh5 were subcloned separately into the pET-32a vector (Novagen). The fused Trx-Bhlhb5 proteins were expressed in Escherichia coli and purified by the NTA-Ni2+ column according to the manufacturer's instructions. Fusion proteins (1 mg) was then injected into rabbits or guinea pigs (PRF&L Inc.), and the antisera were then collected 2 months later and purified by antigen-conjugated affinity columns. However, only the antiserum against the 5′-terminus fusion protein produced specific immunostaining signal. Western and immunofluorescence data were generated with antisera from guinea pig.
Embryonic and postnatal spinal cord tissues were dissected out and submerged in 4% paraformaldehyde at 4°C overnight. After fixation, tissues were transferred to 20% sucrose in phosphate buffered saline overnight, embedded in OCT medium, and then sectioned (15 μm thickness) on a cryostat. Immunofluorescent procedures with guinea pig anti-Bhlhb5 were previously described by Xu et al. (2000). Anti-Nkx2.2, anti-Evx1, anti-En1, anti-HB9, anti-Pax7, and anti-BrdU were obtained from the Developmental Studies Hybridoma Bank. Anti-Brn3a (Chemicon, Inc.), anti-Pax2 (Zymed, Inc.), and anti-Lmx1b (Abcam Inc.) were purchased from commercial sources. Anti-Chx10, anti-Tlx3, and anti-Lbx1 were generously provided by Drs. Sam Pfaff and Carmen Birchmeier.
BrdU Birthdating Analysis of Bhlhb5+ Cells
Specifically, BrdU (Sigma, 15 mg/ml in 7 mM NaOH) was injected intraperitoneally into pregnant mice (0.12 mg/g of body weight) at E11.5, E12.5, or E13.5. Embryos were harvested at E18.5 and analyzed for incorporation of BrdU in Bhlhb5+ neurons by double immunofluorescence with anti-BrdU (Developmental Studies Hybridoma Bank, Iowa city, IA) and anti-Bhlhb5 antibody as previously described in our previous studies (Xu et al., 2000). The percentage of BrdU+ cells in Bhlhb5+ cells was calculated from three separate sections.
We thank Dr. Sam Pfaff for generously providing the anti-Chx10 antibody and Dr. Carmen Birchmeier for generously providing the anti-Lbx1 and anti-Tlx3 antibody. M.Q. was funded by the NIH, and X.P. was funded by the National Sciences Foundation of China.