Suppression of interneuron programs and maintenance of selected spinal motor neuron fates by the transcription factor AML1/Runx1

Runx1 Is Expressed in Specific Postmitotic Motor Neurons in the Mouse Cervical Spinal Cord.

To determine the role of Runx1 in spinal motor neuron development, we first characterized its expression pattern by examining a previously characterized Runx1^lacZ/+ knockin mouse line in which β-galactosidase (β-gal) expression recapitulates the expression of endogenous Runx1 (13, 18, 19). In embryo day (E)9.5 Runx1^lacZ/+ embryos, β-gal expression coincided with Runx1 protein, detected by using a previously described anti-Runx1 antibody (17), in the second branchial arch, where Runx1 mRNA expression has been demonstrated (18) [supporting information (SI) Fig. S1 A–C]. The specificity of the anti-Runx1 antibody was demonstrated by the absence of immunoreactivity in Runx1-deficient embryos (Fig. S1 E and F).

In the spinal cord of Runx1^lacZ/+ embryos, β-gal expression began at ≈E9.5 in a small number of ventrolateral cells at levels C1–C4 (Fig. 1A, arrowheads). β-gal⁺ cells did not express neuronal progenitor markers such as Pax6 or Nkx2.2, nor the general cell proliferation marker Ki67 (Fig. 1 A–C). Moreover, we observed no detectable overlap between the expression of β-gal and the interneuron markers Evx1 (V0 interneurons), En1 (V1), and Chx10 (V2) (Fig. 1 D–F), although some occasional overlap with Chx10 was observed at later stages (see Fig. 3 below). Instead, β-gal⁺ cells expressed the postmitotic motor neuron markers Isl1 and choline acetyltransferase (ChAT) (3, 4, 20) (Fig. 1 G and H, arrows). Only a subset of the Isl1⁺ cells expressed β-gal, suggesting that Runx1 is expressed in a restricted number of spinal motor neurons. The expression of β-gal faithfully recapitulated the Runx1 expression pattern in the spinal cord (Fig. S1 H–J). These results indicate that Runx1 expression in the spinal cord is first activated in a subpopulation of postmitotic motor neurons, but not their progenitors, at cervical levels C1–C4.

Runx1 Expression Is Activated in Ventrally Exiting Motor Neurons After the Onset of Differentiation.

To determine the identity of the cervical Isl1⁺ neurons in which β-gal is first expressed in Runx1^lacZ/+ embryos, we compared the expression of β-gal with that of the Hb9 protein, which is expressed in virtually all motor neurons whose axons exit via the ventral root and innervate skeletal muscles (3, 4). No visible overlap of β-gal and Hb9 expression was observed at E9.5 (data not shown) and E10.5, the peak of motor neuron generation (Fig. 1 J–L). At E9.5, we also failed to detect an overlap between the expression of β-gal and that of Lhx3, a protein whose expression at this stage transiently marks the majority of ventrally exiting motor neurons (vMNs) as well as their precursors (20) (data not shown). Coexpression of β-gal and Hb9 was first observed in the ventromedial spinal cord at ≈E11 (Fig. 1 M–O, arrows, and data not shown). At this stage, Runx1 expression expands ventrally and can be detected at more caudal levels of the spinal cord (C5–T1) (13). This analysis of Runx1^lacZ/+ embryos suggests that the onset of Runx1 expression in vMNs follows the initial activation of Hb9 expression.

In E9.5-E10.5 Runx1^lacZ/+ embryos, most, if not all, of the β-gal⁺ cells coexpressed the homeodomain protein Phox2b (Fig. 1 P–R), which is selectively expressed in spinal accessory (SAC) motor (nXI) neurons (21). SAC neurons are present at spinal levels C1–C4, express Isl1 but not Hb9, and send their axons out of the spinal cord through lateral exit points located midway along the dorsoventral axis of the spinal cord. The exiting axons assemble into the spinal accessory nerve, which innervates branchial arch-derived muscles in the neck (22). Consistent with these observations, at E15.5 a number of β-gal⁺ cells at levels C1–C4 were retrogradely labeled from the anterior trapezius muscle, a lateral cervical muscle of the neck innervated by the spinal accessory nerve (Fig. 1S). Together, these results show that Runx1 expression is activated in two distinct spatiotemporal patterns. It is first expressed in dorsally exiting SAC motor neurons starting at ≈E9.5 followed by a later expression in selected populations of vMNs after the peak of vMN generation and Hb9 expression (E9.5–E10.5). These results suggest that Runx1 becomes activated in specific vMNs after their initial differentiation and during their developmental maturation.

Runx1 Is Expressed in Selected Types of Ventrally Exiting Motor Neurons and Is Not Required for Their Generation or Survival.

In E13.5 Runx1^lacZ/+ embryos, when distinct motor columns are discernable, two groups of β-gal⁺ motor neurons were observed at the forelimb level (C5–C8) (Fig. 2 A–L). One group was composed of motor neurons of the medial component of the axial muscle-innervating median motor column (MMCm) (20), based on their ventromedial location (Fig. 2A, vertical arrows) and the expression of ChAT, Lhx3 (Fig. 2 A–F), and Isl1 (Fig. 2 J–L). The second group consisted of motor neurons of the lateral motor column (LMC), based on their ventrolateral location and expression of retinaldehyde dehydrogenase 2 (RALDH2) (Fig. 2 G–I, horizontal arrows). β-gal expression was found in subpopulations of both medial LMC (LMCm) motor neurons, which express Isl1 but not Lim1, and project to ventral limb muscles, and lateral LMC (LMCl) motor neurons, which express Lim1 and innervate dorsal limb muscles (17, 23) (Fig. 2 G–O). Consistent with these results, a group of β-gal⁺ neurons was retrogradely labeled from the deltoideus muscle (Fig. S2 A–C). This result was specific because β-gal⁺ neurons were not retrogradely labeled from the pectoralis muscle (Fig. S2 D–F). Together, these results argue that Runx1 is expressed in selected populations of ventrally exiting MMC and LMC motor neurons.

To characterize the function of Runx1 in motor neuron development, we examined two separate lines of Runx1-deficient mice, termed Runx1^lacZ/rd and Tie2-Cre;Runx1^{Flox:lacZ/Flox:lacZ}, respectively (19, 24). Transheterozygous Runx1^lacZ/rd embryos are Runx1-null and die at ≈E12.5 because of a lack of fetal liver-derived hematopoiesis (19). This embryonic lethality can be circumvented by examining Tie2-Cre;Runx1^{Flox:lacZ/Flox:lacZ} embryos, in which the expression of Cre recombinase is under the control of the endothelial/hematopoietic-specific Tie2 promoter (25, 26), resulting in a selective reactivation of Runx1 in hematopoietic cells but not in the nervous system. Concomitantly, this conditional reactivation abolishes β-gal expression in hematopoietic but not neuronal cells (Fig. S3). The total number of β-gal⁺ cells in the spinal cord of both Runx1^lacZ/rd and Tie2-Cre;Runx1^{Flox:lacZ/Flox:lacZ} embryos was the same as in their control littermates at all stages examined (Fig. 3 A–C). This result shows that Runx1 is not essential for β-gal expression and maintenance nor for the generation and/or survival of the selected spinal SAC, MMC, and LMC neurons in which it is normally expressed.

Runx1 Is Important for Persistent Suppression of Interneuron Differentiation Programs and Sustained Expression of Motor Neuron Genes.

We traced the fate of motor neurons that normally express Runx1 by following β-gal expression in Runx1 mutant embryos. The numbers of β-gal⁺ cells that coexpressed general vMN markers such as Isl1 and ChAT were decreased in mutant embryos compared with control littermates (Fig. 3 D and F). The expression of specific markers of MMC (i.e., Lhx3) or LMC (i.e., RALDH2) motor neurons was also decreased in Runx1-deficient spinal neurons (Fig. 3 D and F). In Runx1 mutant embryos of both genotypes, we also noted the presence of increased numbers of cells coexpressing β-gal and the exclusive spinal interneuron markers Pax2 and Chx10 (3, 27) (Fig. 3 E and G). Moreover, we found that Pax2, which is not normally expressed in Isl1⁺ spinal motor neurons (ref. 27; see also Fig. 3 H–J), was coexpressed with Isl1 and β-gal in Runx1 mutant motor neurons at E11.5 (Fig. 3 K–M). This situation was observed at several rostrocaudal positions, including level C2–C3 (Fig. S4 A–H) where Runx1-expressing cells correspond to SAC motor neurons (Fig. 1 P–R). We did not detect any significant change in the expression of Phox2b in Runx1 mutant SAC motor neurons at E11.5 (Fig. S4 I–O), suggesting that at least certain Phox2b⁺ SAC neurons might coexpress Pax2 in Runx1-deficient embryos at this stage. Together, these results strongly suggest that Runx1 inactivation results in a derepression of interneuron-specific genes in postmitotic motor neurons.

We next tested whether Runx1 was sufficient to cause a suppression of interneuron-specific genes. To this end, we electroporated a GFP expression plasmid alone or together with a Runx1 expression plasmid into Hamburger and Hamilton (HH) (28) stage 14–16 chicken embryo neural tubes. After incubation, the majority of GFP⁺ cells coexpressed Runx1 (Fig. S5). We determined the proportion of GFP⁺ cells expressing motor neuron and interneuron markers at HH stage 27–28 (Fig. 4 A–E). Compared with control embryos expressing only GFP, Runx1 caused a significant decrease in the expression of the interneuron-specific proteins Pax2 and Chx10, but it did not affect the expression of Evx1, a V0 interneuron marker (Fig. 4F). In addition, ectopic Runx1 expression led to an increase in the expression of general motor neuron markers such as Hb9 and Isl1, as well as the motor- and interneuron markers Lim3 and Lim1/2 (Fig. 4G). These effects were observed at both brachial level, where Runx1 is expressed, and lumbar level, where Runx1 is not expressed (17), but occurred only in the ventral, and not dorsal, half of the spinal cord. Importantly, these effects were phenocopied by the oncogenic human fusion protein AML1/ETO (hereafter referred to as Runx1/ETO for consistency) (Fig. S6). Runx1/ETO harbors the DNA-binding domain of Runx1 fused to the protein eight-twenty one, a potent transcriptional repressor that replaces the normal C-terminal transcription activation and repression domains of Runx1 (5, 29). The ensuing fusion protein retains the DNA-binding specificity of Runx1 and the dedicated transcription repression activity of ETO. Thus, our results suggest that transcriptional repression is important for the effect of exogenous Runx1 on motor neuron and interneuron development in the ventral spinal cord. Together, the findings of this work strongly suggest that Runx1 suppresses interneuron differentiation programs within developing motor neurons.

Mouse Lines.

Runx1^lacZ/+ and Runx1^rd/+ mice were generated and genotyped as described (19, 30). Runx1^lacZ/+ males were crossed to Runx1^rd/+ females to generate double-heterozygous Runx1^lacZ/rd embryos. Runx1^lacZ/rd embryos lack Runx1 activity and die by E12.5 because of failed fetal liver-derived hematopoiesis, similar to Runx1^rd/rd embryos (19). Runx1^Flox:lacZ/+ mice were generated and genotyped as described in ref. 24. Runx1^Flox:lacZ/+ females were crossed to Tie2-Cre;Runx1^Flox:lacZ/+ males, in which Cre recombinase expression is driven by the endothelial/hematopoietic-specific Tie2 promoter (25). For embryonic staging, the day of appearance of the vaginal plug was considered as E0.5. All animal procedures were conducted in accordance with the guidelines of the Canadian Council for Animal Care.

Immunohistochemistry.

Mouse embryos were collected and fixed, and cryostat sections were prepared as described (13, 29). Sections were subjected to immunohistochemistry as described in ref. 13, with the following antibodies: mouse monoclonals against En1 (clone 4G11; 1:10), Evx1 (clone 99.1–3A2; 1:100), Hb9 (clone 81.5C10; 1:25), Isl1 (clone 39.4D5; 1:50), Lim1/2 (clone 4F2; 1:5), Lim3 (clone 67.4E12; 1:25), Nkx2.2 (clone 74.5A5; 1;100), Pax6 (clone Pax6; 1:10), and β-gal (clone 40–1A) (obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA); rabbit polyclonals against β-gal (1:5,000; Cappel), Chx10 (1:4,000) and Runx1 (1:2,000) (gifts from T. Jessell, J. Dasen, and S. Brenner-Morton, Columbia University, New York), En1 (1:1,000; a gift from A. Joyner, Skirball Institute, New York), Lhx3/4 (1:3,000; a gift from S. Pfaff, The Salk Institute for Biological Studies, La Jolla, CA), Pax2 (1:100; Zymed Laboratories), RALDH2 (1:1,500; a gift from P. McCaffery, University of Massachusetts Medical School, Waltham, MA), and Phox2b (1:1,000; a gift from J. F. Brunet, Ecole Normale Superieure, Paris, France); goat polyclonals against β-gal (1:5,000; Biogenesis) and ChAT (1:100; Chemicon); sheep polyclonal against Chx10 (1:500; Chemicon); and guinea pig polyclonal against Isl1/2 (1:5,000; a gift from S. Pfaff). All images were captured by using a DVC black and white camera mounted on an Axioskop fluorescence microscope (Zeiss).

In Ovo Electroporations.

For details, see SI Materials and Methods.

Retrograde Axonal Labeling.

E15.5 Runx1^lacZ/+ embryos were collected into ice-cold PBS, and several muscles, including the anterior trapezius, deltoideus, and pectoralis, were injected with a solution of rhodamine-conjugated dextran (molecular weight, 3,000; Molecular Probes), as described (3, 23). Embryos were incubated for 5 h at 30°C in oxygenated PBS, followed by fixation, embedding in OCT compound, cryosectioning, and immunohistochemistry.