The Drosophila Forkhead/Fox transcription factor Jumeau mediates specific cardiac progenitor cell divisions by regulating expression of the kinesin Nebbish

Abstract

Forkhead (Fkh/Fox) domain transcription factors (TFs) mediate multiple cardiogenic processes in both mammals and Drosophila. We showed previously that the Drosophila Fox gene jumeau (jumu) controls three categories of cardiac progenitor cell division—asymmetric, symmetric, and cell division at an earlier stage—by regulating Polo kinase activity, and mediates the latter two categories in concert with the TF Myb. Those observations raised the question of whether other jumu-regulated genes also mediate all three categories of cardiac progenitor cell division or a subset thereof. By comparing microarray-based expression profiles of wild-type and jumu loss-of-function mesodermal cells, we identified nebbish (neb), a kinesin-encoding gene activated by jumu. Phenotypic analysis shows that neb is required for only two categories of jumu-regulated cardiac progenitor cell division: symmetric and cell division at an earlier stage. Synergistic genetic interactions between neb, jumu, Myb, and polo and the rescue of jumu mutations by ectopic cardiac mesoderm-specific expression of neb demonstrate that neb is an integral component of a jumu-regulated subnetwork mediating cardiac progenitor cell divisions. Our results emphasize the central role of Fox TFs in cardiogenesis and illustrate how a single TF can utilize different combinations of other regulators and downstream effectors to control distinct developmental processes.

Introduction

While at least eight Forkhead (Fkh/Fox) domain transcription factors (TFs) are required for proper cardiac development in mammals^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16}, and mutations in four Fox genes have been linked to human congenital heart defects^{17,18,19,20,21,22,23,24}, relatively little is known about the molecular mechanisms or the downstream target genes by which these Fox TF-mediated developmental functions are brought about²⁵. Our previous work had identified the conserved cardiogenic roles of two Fox genes, jumeau (jumu) and Checkpoint suppressor homologue (CHES-1-like) in both the specification and subsequent division of progenitor cells in the Drosophila heart^26,27,28.

We showed previously that both jumu and CHES-1-like function in a mutually redundant manner to transcriptionally activate the expression of heartless (htl) and frizzled (fz), which respectively encode the receptors in a FGF-signaling pathway and a Wnt-signaling pathway required for cardiac progenitor specification; consequently, lacking both Fox genes results in a failure of much of the heart to be specified²⁶. We also demonstrated that jumu and CHES-1-like mediate three distinct categories of cell division that generate the correct subtypes and numbers of cells constituting the heart—asymmetric cardiac progenitor cell divisions, symmetric cardiac progenitor cell divisions, and cell divisions at an earlier stage that produced the cardiac precursor cells—by regulating the activity of the conserved kinase Polo²⁸. In addition, our investigations revealed that the TF Myb acts synergistically with Jumu, CHES-1-like, and Polo to mediate only two of these three categories of cardiac progenitor cell division: symmetric and cell division at an earlier stage²⁷. In contrast, asymmetric cardiac progenitor cell division involved Jumu and CHES-1-like utilizing Polo to phosphorylate Partner of numb (Pon) and thereby bring about the localization of membrane-associated Numb protein to one pole of the dividing cell^28,29.

Additionally, we found that Fox TF binding sites were significantly enriched in the enhancers of genes expressed in the heart^27,30, and that loss-of-function mutations in jumu or CHES-1-like significantly altered the expression levels of many known cardiac genes²⁶. Taken together, these data indicate that the Fox TFs regulate a large number of genes in addition to polo, htl, and fz in order to mediate cardiogenesis.

Collectively, these observations raise three intriguing developmental questions. First, are additional Fox TF-regulated genes besides polo, htl, and fz involved in the two distinct cardiogenic processes—cardiac progenitor specification and cardiac progenitor cell division—that we have discovered as being mediated by the Fox TFs to date?

Second, jumu and CHES-1-like were both seen to be simultaneously regulating the activity of each relevant downstream target gene (htl, fz, or polo) whenever any one of these target genes was characterized as mediating a cardiogenic process. Thus, a particularly germane question is whether there are any downstream cardiogenic targets and, by extension, associated pathways or subnetworks, that are exclusively regulated and mediated by one, but not the other, of these two Fox TFs.

Third, while Myb itself acts in concert with both Fox genes at the same hierarchical level to mediate only two of the three categories of cardiac progenitor cell divisions, polo, the downstream target of both jumu and CHES-1-like, is utilized for all three. Thus, another pertinent question is whether every Fox TF-regulated gene utilized in cardiac progenitor cell division also plays critical roles in all three categories of cell division or is necessary for only certain specific subsets.

Here we show that the kinesin-encoding gene nebbish (neb) is activated by only one Fox gene, jumu; that neb is required in conjunction with polo and Myb to mediate cardiac progenitor cell divisions; and that neb is necessary for only two of the three categories of cardiac progenitor cell division—symmetric and cell division at an earlier stage—thereby extending our understanding of the Fox-regulated subnetworks mediating cardiogenesis.

Results

jumu, but not CHES-1-like, activates mesodermal expression of neb

One of our goals was to discover downstream genes mediating cardiogenic processes, if present, that were regulated by one, but not the other, of the twin Fox TFs Jumu and CHES-1-like. We had previously used flow cytometry and Affymetrix microarrays to obtain genome-wide transcription expression profiles of purified mesodermal cells from wild-type embryos, embryos homozygous for a null mutation in jumu, and embryos homozygous for a null mutation in CHES-1-like at embryonic stages 11 and early 12, i.e. when cardiac progenitor specification and cell division occur in Drosophila²⁶. Reasoning that cardiogenic genes regulated by a Fox TF are likely to exhibit differential expression between the heart-producing mesoderm in wild-type and Fox loss-of-function mutant embryos, we compared the transcription expression profiles between mesodermal cells from wild-type embryos and the Fox gene mutants. We found that neb had its expression levels significantly downregulated in jumu null mutants [log₂(fold change) =  − 0.69058; adjusted P-value = 1.65 × 10⁻⁵ after correction for multiple hypothesis testing] but remained essentially unaffected in CHES-1-like null mutants [log₂(fold change) = 0.14551; adjusted P-value = 0.195].

In order to independently verify the effects of these two Fox genes on neb expression, we used the pan-mesodermal twi-GAL4 driver to knock down either jumu or CHES-1-like via RNA interference (RNAi) exclusively in the mesoderm. We then assessed and compared the resulting neb transcript levels at stages 11–12 with that in control embryos by reverse transcription quantitative real-time PCR (RT-qPCR). Since jumu and CHES-1-like are also expressed in tissues other than the mesoderm³¹, null mutations in the Fox genes could potentially also affect neb expression levels in these non-mesodermal tissues. Utilizing exclusively mesoderm-targeted RNAi knockdowns of these Fox genes instead of null mutations in jumu or CHES-1-like thus ensured that any detected alterations in neb transcript levels in total RNA were due solely to the effects of the Fox genes in the mesoderm. Our RT-qPCR assays revealed a decrease in neb expression levels in mesoderm-targeted jumu knockdowns compared to that in control embryos containing only the twi-GAL4 driver, but detected no such reduction in mesoderm-targeted CHES-1-like knockdowns (Fig. 1).

Figure 1

jumu, but not CHES-1-like, activates mesodermal expression of neb. Relative jumu, CHES-1-like, and neb gene expression levels in jumu or CHES-1-like RNAi knockdown embryos assessed and compared to control embryos via RT-qPCR and the 2^−ΔΔC_T method. Error bars indicate standard deviations. Mesoderm-targeted jumu and CHES-1-like RNAi do indeed reduce jumu and CHES-1-like gene expression levels compared to controls, respectively, indicating that these RNAi constructs result in the relevant Fox gene knockdowns. However, while mesoderm-targeted jumu RNAi knockdown also decreases neb transcript expression levels relative to the control, no such reduction is detected in mesoderm-targeted CHES-1-like knockdowns.

Collectively, both our microarray-based expression profiling data and RT-qPCR-mediated assessment of Fox gene knockdowns indicated that jumu, but not CHES-1-like, regulates neb in the mesoderm by activating its expression.

neb is expressed in the cardiac mesoderm during cardiac progenitor specification and cell division

neb had previously been described as being expressed in the nervous tissue during mid-embryogenesis³¹. Using in situ hybridization, we found that the neb transcript is also expressed, albeit at lower levels, in the cardiac mesoderm and in subsets of the somatic mesoderm at embryonic stages 11 and 12, i.e. when the cardiac progenitor cells are specified and undergo cell division (Fig. 2a,b). However, the expression pattern of the transcript made it difficult to determine unequivocally whether the expression level of neb is lower specifically in the cardiac mesoderm of jumu loss-of-function mutants by in situ hybridization assays. Regardless, the observations that neb is both expressed in the cardiac mesoderm and is transcriptionally activated mesodermally by jumu, collectively raise the question of whether neb has a critical a role in cardiogenesis.

Figure 2

neb mRNA expression in the cardiac mesoderm. (a) Antisense riboprobe hybridizing to the mRNA shows that the neb transcript is expressed in the cardiac mesoderm in stage 11 wild-type embryos. (b) Control in situ hybridization with a sense riboprobe complementary to that used in (a) developed under identical conditions for the same period. The absence of similar staining in (b) indicates that the staining in (a) is neb-specific and not an artifact. (c) In situ hybridization of stage 11 embryos of the genotype TinD-GAL4/+; jumu^{Df(3R)Exel6157} svp-lacZ/jumu^{Df(3R)Exel6157} UAS-neb with the same antisense neb riboprobe as in (a) developed using identical conditions but for a much shorter period. The specific TinD-GAL4-targeted expression of neb to the cardiac mesoderm is clearly observed while the staining for endogenous neb expression is considerably fainter due to the shorter period used. The arrows indicate cells of the cardiac mesoderm in all three panels (a–c).

neb is required in the cardiac mesoderm for two distinct categories of cardiac progenitor cell divisions

The metamerically repeated and stereotyped pattern of distinct cardiac cell types in the embryonic Drosophila heart makes it particularly amenable to discovering whether neb has one or more crucial cardiogenic roles, and what those roles might be, by examining the phenotypes of neb loss-of-function mutations. The heart itself is comprised of two major cell types: an inner tube of Myocyte enhancer factor 2 (Mef-2)-expressing contractile cardial cells (CCs) and an outer sheath of Zn finger homeodomain 1 (zfh1)-expressing nephrocytic pericardial cells. From anterior to posterior, the cardial cells consist of two Seven up-expressing CCs (Svp-CCs) followed by four Tinman-expressing CCs (Tin-CCs) in every repeating hemisegment. These cardial cells are surrounded by a larger number of pericardial cells, which includes two Seven up-expressing PCs (Svp-PCs) that are located laterally to the CCs in each and every hemisegment. The different cellular subtypes comprising these cardial and pericardial cells arise through a stereotyped series of cardiac progenitor cell divisions^32,33,34,35. Specifically, cell division at an earlier stage ensures that there are two Svp precursor cells per hemisegment. Subsequently, the two Svp-CCs in wild-type cardiac hemisegments arise by two asymmetric progenitor cell divisions from these two Svp precursor cells, with each cell division generating one Svp-CC and one Svp-PC (Fig. 3a, yellow and red cells, respectively). In contrast, two symmetric cell divisions give rise to the four Tin-CCs (Fig. 3a, green cells) per hemisegment from two other precursor cells.

Figure 3

Schematic showing cell lineage relationships in a wild-type Drosophila embryonic heart and the expected phenotypes due to defects in cardiac progenitor specification or cell division. (a) Lines connect daughter cells arising from the division of each precursor cell. (b) Defects in cardiac progenitor cell specification would be expected to result in the absence of all the cardial cells in one or more hemisegments. Defects in asymmetric cell division would result in an increase or reduction in the number of Svp-CCs accompanied by a corresponding decrease or increase in the number of Svp-PCs, or larger Svp-CC nuclei with missing corresponding Svp-PCs due to errors in karyokinesis. Defective symmetric cell division would result in an alteration in the expected number of four Tin-CCs per hemisegment, while errors at the earlier stage of cell division producing the two Svp precursors would result in hemisegments with either one or three pairs of Svp-CCs and Svp-PCs instead of the customary two pairs.

These lineage relationships allow us to use the numbers of the Tin-CCs, Svp-CCs, and Svp-PCs in individual hemisegments in neb loss-of-function mutants to determine whether neb is mediating one or more of these three known categories of cardiac progenitor cell divisions (Fig. 3b). If neb mutations cause asymmetric cell division defects, we would expect to see an increase or reduction in the number of Svp-CCs accompanied by a corresponding decrease or increase in the number of Svp-PCs, or larger Svp-CC nuclei with missing corresponding Svp-PCs due to errors in karyokinesis. Defective symmetric cell division, on the other hand, would be identified as any alteration in the expected number of four Tin-CCs per hemisegment, while errors at the earlier stage of cell division producing the two Svp precursors would result in hemisegments with either one or three pairs of Svp-CCs and Svp-PCs instead of the customary two pairs²⁸. Our examination found that embryos homozygous for the neb^k05702 loss-of-function allele exhibited a significant increase over wild-type both in the fraction of hemisegments with excess or fewer Tin-CCs (P = 2.00 × 10⁻⁶), corresponding to symmetric cell division defects, and in the fraction of hemisegments with either one or three pairs of Svp-CCs and Svp-PCs (P = 0.016), corresponding to earlier cell division defects affecting the number of Svp precursors (Fig. 4a–c, Supplementary Table S1). The fraction of hemisegments exhibiting asymmetric cell division defects, in contrast, was miniscule and not significantly different (P = 0.481) from that in wild-type embryos (Fig. 4a–c, Supplementary Table S1). Collectively, our results demonstrate that neb is necessary to mediate both symmetric cell divisions and the earlier round of cell divisions that give rise to the Svp precursors, but is not required for asymmetric cell divisions. Of note, we did not detect any instance where all the cardial cells in one or more entire hemisegments were missing, as would be expected in the case of cardiac progenitor specification defects, and were seen when htl or fz functions were disrupted^{26, 36,37,38}.

Figure 4

Cardiac progenitor cell division defects associated with neb loss of function. (a) A heart from an otherwise wild-type embryo bearing one copy of the svp-lacZ enhancer trap showing hemisegments consisting of four Tin-CCs (green), two Svp-CCs (yellow), and two Svp-PCs (red). (b) A heart from an embryo homozygous for the neb^k05702 null mutation (and also carrying one copy of the svp-lacZ enhancer trap) exhibiting both cardiac progenitor symmetric cell division defects (arrowheads) and defects at an earlier round of cell division specifying the number of Svp precursors (arrows). In these images (derived by flattening z-stacks), a few Svp-PCs are hidden underneath the CCs or appear quite faint in certain hemisegments, but all were clearly discernible in the individual planes of the z-stacks from which the images were obtained. (c) Percentage of hemisegments exhibiting each type of cardiac progenitor cell division defect in embryos that are wild-type (n = 196 hemisegments) or homozygous for the neb^k05702 mutation (n = 186 hemisegments). The significance of each type of cell division defect in the neb mutants compared with wild-type is shown. (d) A wild-type heart stained to show only Tin-CCs (green) and Svp-CCs (yellow). Each hemisegment contains four Tin-CCs and two Svp-CCs. (e) Similar staining in embryos where neb has been knocked down specifically in the cardiac mesoderm exhibits changes in Tin-CC numbers and Svp-CC numbers consistent with symmetric (arrowheads) and earlier (arrows) cell division defects, respectively.

We next attempted to ascertain whether this neb function, necessary for mediating both symmetric cell divisions and divisions at the earlier stage affecting the number of Svp precursor cells, is required specifically in the cardiac progenitors. We thus knocked down neb activity specifically in the cardiac progenitors by cardiac mesoderm and heart-targeted RNA interference (RNAi) directed by the TinD-GAL4³⁹ and Hand-GAL4⁴⁰ drivers and examined the resulting embryos for alterations in expected cardial cell numbers consistent with errors in cardiac progenitor cell divisions. Staining with appropriate antibodies showed hemisegments with localized changes in the number of Tin-CCs corresponding to symmetric cell division defects and Svp-CCs corresponding to defective cell division at an earlier stage (Fig. 4d,e, Supplementary Fig. S1, Supplementary Table S2).

Taken together, these results demonstrate that the requirement of neb for mediating two distinct categories of cardiac progenitor cell division is autonomous to the cardiac mesoderm and raise the question of which cell division pathway or subnetwork neb is involved in, one that we address in the following experiments.

Synergistic genetic interactions between neb, jumu, polo, and Myb

We had previously shown that jumu and CHES-1-like regulate polo to mediate all three categories of cardiac progenitor cell divisions²⁸; and that Myb functions synergistically with jumu, CHES-1-like, and polo to control both symmetric and earlier cardiac progenitor cell divisions²⁷, the same two cardiogenic processes mediated by neb. These observations raise the questions of whether neb functions through the same genetic pathways as Myb, jumu, and polo, and, if it does, for which specific categories of cardiac progenitor cell divisions. Operating on the premise that we may expect to see synergistic, i.e. more than merely additive, genetic interactions between mutant alleles of neb and another gene if they both function together in the same genetic pathway, we quantified and compared the phenotypes of single heterozygotes of mutations in neb and single heterozygotes of mutations in jumu, polo, or Myb, with those of embryos that were doubly heterozygous for mutations in both neb and jumu, neb and polo, and neb and Myb, respectively.

Double heterozygotes for both neb and jumu mutations exhibit symmetric cell division defects and earlier Svp precursor-determining cell division defects at frequencies that are significantly more severe (P = 1.80 × 10⁻⁵ and P = 8.23 × 10⁻⁴, respectively) than the additive effects of both the neb single heterozygotes and the jumu single heterozygotes (Figs. 5a–c, 6a, Supplementary Table S1). However, the increase in the frequency of asymmetric cell division defects in neb^MI0225/+; jumu^{Df(3R)Exel6157}/+double heterozygotes over the additive sum of those in the component neb^MI0225/+and jumu^{Df(3R)Exel6157}/+single heterozygotes is barely significant (P = 0.0429) (Figs. 5a–c, 6a, Supplementary Table S1). Our results thus indicate clearly that jumu and neb do indeed mediate both symmetric and earlier Svp precursor-determining cardiac progenitor cell divisions through the same genetic pathways, but the evidence for neb functioning in concert with jumu to bring about asymmetric cardiac progenitor cell divisions as well is not as strong. Note that these genetic interaction results are also consistent with our initial finding of neb being required for symmetric and earlier, but not asymmetric, cardiac progenitor cell divisions.

Figure 5

neb exhibits synergistic genetic interactions with jumu, polo, and Myb, but not with CHES-1-like. Representative hearts from embryos (a) heterozygous for the neb^MI02225 allele, (b) heterozygous for the jumu^{Df(3R)Exel6157} allele, (c) doubly heterozygous for both the neb^MI02225 and jumu^{Df(3R)Exel6157} alleles, (d) heterozygous for the polo¹⁰ allele, (e) doubly heterozygous for both the neb^MI02225 and polo¹⁰ alleles, (F) heterozygous for the Myb^MH107 allele, (g) doubly heterozygous for both the neb^MI02225 and Myb^MH107 alleles, (h) heterozygous for the CHES-1-like¹ allele, and (i) doubly heterozygous for both the neb^MI02225 and CHES-1-like¹ alleles. All of these embryos carry one copy of the svp-lacZ enhancer trap, thereby allowing the identification of Tin-CCs (green), Svp-CCs (yellow), and Svp-PCs (red). In these images (derived by flattening z-stacks), a few Svp-PCs are hidden underneath the CCs or appear quite faint in certain hemisegments, but all were clearly discernible in the individual planes of the z-stacks from which the images were obtained. Asymmetric cell division defects are denoted by asterisks, symmetric cell division defects by arrowheads, and earlier Svp precursor-determining cell division defects by arrows.

Figure 6

Quantification and significance of cardiac progenitor cell division defects in genetic interaction assays. (a) Percentage of hemisegments exhibiting each type of cardiac progenitor cell division defect in embryos that are heterozygous for the neb^MI02225 allele (n = 188 hemisegments), heterozygous for the jumu^{Df(3R)Exel6157} allele (n = 210 hemisegments), or doubly heterozygous for both the neb^MI02225 and jumu^{Df(3R)Exel6157} alleles (n = 238 hemisegments). (b) Percentage of hemisegments exhibiting each type of cardiac progenitor cell division defect in embryos that are heterozygous for the neb^MI02225 allele, heterozygous for the polo¹⁰ allele (n = 196 hemisegments), or doubly heterozygous for both the neb^MI02225 and polo¹⁰ alleles (n = 210 hemisegments). (c) Percentage of hemisegments exhibiting each type of cardiac progenitor cell division defect in embryos that are heterozygous for the neb^MI02225 allele, heterozygous for the Myb^MH107 allele (n = 193 hemisegments), or doubly heterozygous for both the neb^MI02225 and Myb^MH107 alleles (n = 179 hemisegments). (d) Percentage of hemisegments exhibiting each type of cardiac progenitor cell division defect in embryos that are heterozygous for the neb^MI02225 allele, heterozygous for the CHES-1-like¹ allele (n = 210 hemisegments), or doubly heterozygous for both the neb^MI02225 and CHES-1-like¹ alleles (n = 192 hemisegments).

Since jumu brings about all three categories of cardiac progenitor cell divisions by regulating the activity of Polo kinase²⁸, we used similar genetic interaction assays next to assess whether neb and polo also function through the same genetic pathways. Our results revealed highly significant synergistic genetic interactions between neb and polo for both symmetric cell divisions and earlier cell divisions (P = 3.89 × 10⁻⁴ and P = 1.00 × 10⁻⁶, respectively), but no significant interactions in the case of asymmetric cardiac progenitor cell divisions (P = 0.117) (Figs. 5a,d,e, 6b, Supplementary Table S1).

Since Myb has been shown to regulate polo expression⁴¹, and act synergistically with jumu to mediate only symmetric and earlier Svp precursor-determining cardiac progenitor cell divisions²⁷, we next addressed whether neb works in concert with Myb via genetic interaction assays, and in which classes of cardiac progenitor cell divisions. Significant synergistic genetic interactions between neb and Myb were also detected in the case of both symmetric cell divisions and earlier cell divisions (P = 7.60 × 10⁻⁵ and P = 7.43 × 10⁻⁴, respectively), but not for asymmetric cell divisions (P = 0.340) (Figs. 5a,f,g, 6c, Supplementary Table S1).

Taken together, these results and our initial findings show that neb is necessary for only two of the three categories of cardiac progenitor cell divisions, symmetric and earlier, and that it mediates these two types of cell divisions by working in concert with jumu, Myb, and polo through the same genetic pathways.

Similar synergistic genetic interactions are not detected between neb and CHES-1-like

As described previously, both of the jumu-regulated cardiogenic processes that we had discovered, cardiac progenitor specification and cell division, are also regulated by another Fox TF, CHES-1-like. Furthermore, the downstream genes used for each of these processes, htl and fz in the case of cadiac mesoderm specification, and polo in the case of cardiac progenitor cell divisions, are each also regulated by both jumu and CHES-1-like^27,28. neb, however, provides an exception to this trend because, based on our expression profiling data, it is transcriptionally regulated only by jumu, and not by CHES-1-like. Does the observation that CHES-1-like does not regulate neb imply that neb functions completely independently of CHES-1-like, or are they both components of the same genetic subnetwork?

In an attempt to answer this question, we assessed potential genetic interactions between neb and CHES-1-like in mediating cardiac progenitor cell divisions, based on the premise that such interactions are likely to be synergistic if neb and CHES-1-like function in concert in the same pathways. The frequency of cardiac progenitor cell division defects in CHES-1-like¹/+; neb^MI0225/+ double heterozygotes was not significantly different from the additive sum of those of the CHES-1-like¹/+ and neb^MI0225/+ single heterozygotes during asymmetric (P = 0.266), symmetric (P = 0.721), or earlier Svp precursor-determining cell divisions (P = 0.530) (Figs. 5a,h,i, 6d, Supplementary Table S1). The lack of synergistic genetic interactions between neb and CHES-1-like in any of the three cardiac progenitor cell division categories suggests strongly that these two genes function independently of each other in these cardiogenic processes.

Cardiac mesoderm-targeted ectopic expression of neb partially rescues both symmetric and earlier cardiac progenitor cell division defects in embryos lacking jumu function

Our results so far show that jumu activates neb and that neb is specifically required in the cardiac mesoderm for mediating symmetric and earlier Svp precursor-determining cell divisions, i.e. two of the three categories of jumu-regulated cardiac progenitor cell divisions. The most parsimonious hypothesis that can explain these observations is that jumu transcriptionally upregulates neb in the cardiac mesoderm to bring about symmetric and earlier cardiac progenitor cell divisions, and that in jumu loss-of-function mutations the ensuing reduction in neb expression levels in this tissue results in defects in only these two types of cell divisions. However, the nature of neb expression prevented us from qualitatively assessing via in situ hybridization assays whether the reduction of neb transcript levels in the entire mesoderm in jumu mutants reflected a reduction of neb levels specifically in the cardiac mesoderm.

Therefore, to test this hypothesis, we ectopically expressed full-length neb specifically in the cardiac mesoderm in jumu loss-of-function mutants under the control of the TinD-GAL4 driver (Fig. 2c). If our hypothesis that neb acts downstream of jumu specifically in the cardiac mesoderm is correct, then the cardiac mesoderm-specific ectopic expression of neb should be able to partially restore neb levels and rescue the symmetric and earlier cardiac progenitor cell division defects in jumu mutants. We found that ectopic expression of neb throughout the cardiac mesoderm does indeed significantly reduce the severity of defects of jumu mutants in symmetric cardiac progenitor cell divisions (P = 0.0144). Defects in Svp precursor-determining cell divisions are also rescued, albeit with borderline significance (P = 0.0632). In contrast, no significant difference is detected in asymmetric cell division defects between jumu mutants alone and jumu mutants expressing full-length neb (P = 0.781) (Fig. 7a–c, Supplementary Table S1). Collectively, our results thus indicate a downstream requirement of neb in the cardiac mesoderm for correct symmetric and earlier cell divisions mediated by jumu.

Figure 7

Cardiac mesoderm-targeted ectopic expression of neb partially rescues both symmetric and earlier Svp precursor-determining cardiac progenitor cell division defects in embryos lacking jumu function. (a) Representative heart from an embryo homozygous for the jumu^{Df(3R)Exel6157} null mutation exhibiting asymmetric cell division defects (asterisks), symmetric cell division defects (arrowheads), and earlier Svp precursor-determining cell division defects (arrow). (b) Representative heart from an embryo homozygous for the jumu^{Df(3R)Exel6157} null mutation but ectopically expressing neb in the cardiac mesoderm exhibiting only asymmetric cell division defects. All of these embryos possess one copy of the svp-lacZ enhancer trap, allowing the identification of Tin-CCs (green), Svp-CCs (yellow), and Svp-PCs (red). (c) Quantification and significance of each type of cardiac progenitor cell division defect (n = 193 hemisegments for embryos homozygous for the jumu^{Df(3R)Exel6157} null mutation alone, and n = 205 hemisegments for embryos homozygous for the jumu^{Df(3R)Exel6157} null mutation but ectopically expressing neb in the cardiac mesoderm).

Discussion

Our previous work had identified the roles of htl and fz as Fox TF-regulated genes that mediate cardiac progenitor specification and polo as yet another Fox TF-regulated gene that mediates three distinct categories of cardiac progenitor cell division^26,28. In this study, we identify another Fox TF-regulated downstream gene, neb, which also mediates cardiac progenitor cell divisions. The previously characterized Fox downstream targets htl, fz, and polo are each regulated by both jumu and CHES-1-like simultaneously. Our expression profiling results indicate that neb, in contrast, is transcriptionally regulated only by Jumu, and not by CHES-1-like. This independence from CHES-1-like regulatory control is also supported by neb exhibiting synergistic genetic interactions with jumu in cardiac progenitor cell division subnetworks, but not with CHES-1-like.

We showed further that it is the activation of neb by Jumu specifically in the cardiac mesoderm that is necessary for proper cardiac progenitor cell divisions. It is possible that Jumu may also activate neb expression at other mesodermal domains besides the cardiac mesoderm: our expression profiling was based on the entire mesoderm, and in situ hybridization assays showed neb expression in subsets of the somatic mesoderm in addition to the cardiac mesoderm. However, the twin observations that cardiac mesoderm-specific knockdown of neb partially phenocopies jumu loss-of-function cardiac progenitor cell division defects and that cardiac mesoderm-specific ectopic expression of neb partially rescues jumu loss-of-function cell division defects demonstrate clearly that jumu mediates these cardiac progenitor cell division defects by activating neb expression specifically in the cardiac mesoderm.

Our previous examinations of cardiac progenitor cell divisions had shown that while jumu, CHES-1-like, and polo are required for all three categories of cardiac progenitor cell division—asymmetric cell divisions, symmetric cell divisions, and cell divisions at an earlier stage that produce the Svp precursor cells—Myb is required for only the two latter classes^27,28. In this study, we discovered that neb, like Myb, is also required for only the symmetric and earlier cardiac progenitor cell divisions, not asymmetric cell divisions. Furthermore, while neb exhibits no synergistic genetic interactions with CHES-1-like in any of the three classes of cardiac progenitor cell divisions, it displays synergistic genetic interactions with Myb and polo exclusively in symmetric and earlier Svp precursor-determining cell divisions. And while neb and jumu do show synergistic genetic interactions in all three categories of cardiac progenitor cell divisions, the interaction in the case of asymmetric cell divisions is barely significant. Finally, we find that ectopic cardiac mesoderm-specific expression of neb partially rescues symmetric and earlier cardiac progenitor cell division defects in jumu loss-of-function mutants, but not asymmetric cell division defects. Given also prior studies showing synergistic genetic interactions between Myb, jumu, and polo in these same two classes of cardiac progenitor cell divisions²⁷, and the regulation of polo by both jumu and Myb^28,41, our data thus suggests the presence of an additional CHES-1-like-independent, jumu-regulated pathway involving Myb, polo, and neb that mediates only symmetric and earlier cell divisions. This pathway both expands our prior model of Fox TF-regulated cardiogenic processes (Fig. 8) and emphasizes the central role of jumu in heart development by illustrating how it utilizes different combinations of other regulators and downstream effectors to mediate each of these processes.

Figure 8

Model for jumu and CHES-1-like-regulated cardiogenic processes based on previous work and this current study. jumu and CHES-1-like initially transcriptionally activate htl and fz to mediate the specification of cardiac progenitors from the dorsal mesoderm by the FGF and Wnt-signaling pathways²⁶. Polo, activated by the synergistic functions of jumu and CHES-1-like, then brings about the phosphorylation of Pon necessary for the localization of Numb protein during asymmetric cell division of the Svp precursor cells²⁸. Additionally, jumu, CHES-1-like, and Myb also act synergistically to activate Polo to mediate both symmetric and earlier Svp precursor-determining cell divisions^27,28,41. The results of the current study suggest the presence of an additional jumu-regulated pathway (highlighted in peach) that acts independently of CHES-1-like, involves Myb, polo, and neb, and mediates only symmetric and earlier Svp precursor-determining cardiac progenitor cell divisions. Lines without arrowheads connecting neb with polo and with Myb indicate that while these three genes do indeed function synergistically to mediate symmetric and earlier cardiac progenitor cell divisions, our data at present cannot determine whether Myb or polo are acting upstream or downstream of neb. Since both polo and neb are regulated by jumu, and since it is not yet known whether polo acts upstream or downstream of neb, these two genes have been drawn at the same level.

neb encodes a kinesin, the Drosophila ortholog of the mammalian Kif14, that is required for chromatin condensation, chromosome segregation, and cytokinesis^31,42,43,44; thus errors in one or more of these processes could result in the symmetric or earlier cardiac progenitor cell division defects seen in neb mutants. In particular, Neb forms a conserved physical complex with another kinesin, Pavarotti, and the Citron kinase Sticky at the nexus of the central spindle and the actomyosin contractile ring to bring about the cleavage furrow ingression necessary for cytokinesis⁴⁵. Thus it is particularly intriguing to learn from our expression profiling data that both sticky (sti) and pavarotti (pav) are also activated by Jumu in the mesoderm [log₂(fold change) =  − 0.77505; adjusted P-value = 8.27 × 10⁻⁸ and log₂(fold change) =  − 0.54834; adjusted P-value = 2.51 × 10⁻⁶ respectively in jumu mutants] (Supplementary Table S3). These data raise the possibility that Jumu may be mediating one aspect of cardiac progenitor cell divisions by regulating not just one but multiple components of this cytokinesis subnetwork and will be the focus of future investigations.

A particularly interesting question raised by our observations is why loss-of-function mutations in neb affect symmetric cell divisions of cardiac progenitor cells, but not asymmetric cell divisions. One can easily conceive of the converse: mutations in numb or pon, for instance, that disrupt asymmetric cell division without affecting symmetric cell division by failing to appropriately localize Numb to one pole of the dividing cell, and thus disproportionately to one of the resulting daughter cells^27,28,29. Proper Numb localization is required only for asymmetric cell division. But why would mutations in neb, a gene purportedly involved in more general cell division phenomena such as chromosome condensation, chromosome segregation, and cytokinesis, have no effect on cardiac progenitor asymmetric cell division? One possible explanation could be the presence of other redundant genes and/or pathways that may compensate for the lack of neb function during asymmetric cell division. These potential redundant genes and pathways could also be involved in symmetric cell division and may explain why null mutations in neb did not result in symmetric cell division defects in each and every hemisegment.

Another intriguing question is whether the activation of neb by Jumu is mediated by direct transcriptional regulation, requiring the functional binding of the Fox TF itself to cis-regulatory regions of the neb gene to stimulate its transcription; or indirect, involving regulation by other, Jumu-activated TFs. While this question cannot be completely answered without the identification and functional analysis of the cis-regulatory regions mediating neb transcription in the cardiac mesoderm, we note that ChIP-seq analyses compiled in the modERN and modENCODE databases identify three in vivo Jumu binding regions in the immediate neighborhood of the neb gene, one overlapping the promoter, the other two in the intron, and all three coinciding with the in vivo binding sites of Myb as well as those of the cardiogenic TFs Tinman, Tailup, and Twist^46,47 (Supplementary Fig. S2). These observations raise the possibility that one or more of these three Jumu binding sites may be the cis-regulatory region(s) mediating neb transcription and suggest that neb might be a direct transcriptional target of Jumu as well as of Myb.

We have previously suggested²⁸ that FoxM1 is the functional ortholog of jumu in mammals because FoxM1 regulates Plk1, the mammalian ortholog of polo⁴⁸, because FoxM1 and Myb function as synergistic master regulators of key genetic programs of human B cell proliferation in the germinal center⁴⁹, and because mutations in FoxM1 exhibit cardiac phenotypes—highly irregular orientation of cardiomyocytes with enlarged polyploid nuclei—that are similar to those in jumu mutants¹⁰. Given our discovery of a jumu-regulated pathway for cell division that involves Myb, polo, and neb in Drosophila, and possible cis-regulatory regions of neb that bind both Jumu and Myb, it was very interesting to find that both mammalian FoxM1 and B-Myb bind to the promoter of the neb ortholog Kif14 and activate its transcription in breast cancer cell lines^50,51. And, not surprisingly, Kif14 is considered to play a role as an oncogene in cancer types as varied as breast, lung, liver, gastric, colorectal, ovarian, cervical, and prostate cancers in addition to glioma, medulloblastoma, and retinoblastoma^{52,53,54,55,56,57,58,59,60,61,62,63,64}.

Taken together, these data suggest that the evolutionary role of neb and its ortholog Kif14 is quite crucial, involving not only a conserved complex mediating cytokinesis in both flies and mammals^45,65, but also its potentially conserved regulation by Fox and Myb TFs. It remains to be seen whether the Fox TF-regulated role of this kinesin is also conserved in cardiac progenitor cell divisions in mammals. The observation that the Meckel–Gruber syndrome linked to some mutations of human Kif14 is occasionally found to be also associated with congenital heart defects⁶⁶ suggests that this may indeed be the case.

Methods

Drosophila strains and genetics

The following mutant alleles, deficiencies, and transgenes were used for counting cell division defects: jumu^{Df(3R)Exel6157} [also known as Df(3R)Exel6157; FlyBase ID: FBab0038212] and CHES-1-like¹ [also known as Df(1)CHES-1-like¹; FlyBase ID: FBab0047384]²⁸; neb^k05702 [FlyBase ID: FBal0043429]⁴²; neb^MI0225 [FlyBase ID: FBal0249290]; UAS-neb^RNAi [neb^GD7072; FlyBase ID: FBal0210181]; UAS-neb [neb^UAS.Tag:HA; FlyBase ID: FBal0197607]; svp-lacZ [svp³; FlyBase ID: FBal0016610]³³; polo¹⁰ [polo^S132408; FlyBase ID: FBal0099891]⁶⁷; Myb^MH107 [FlyBase ID: FBal0141652]⁶⁸; UAS-Dcr-2 [Dcr-2^UAS.cDa; FlyBase ID: FBal0211026]⁶⁹; TinD-GAL4 [GAL4^tin.D; FlyBase ID: FBal0267971]^30,39; and Hand-GAL4 [GAL4^HCH.Hand; FlyBase ID: FBal0190607]⁴⁰. They were maintained over FM7c, ftz-lacZ, CyO, ftz-lacZ, and TM3, ftz-lacZ balancers when they could not be homozygosed before crossing. Relevant embryos resulting from these crosses were distinguished from their siblings by the absence of anti-β-galactosidase staining in the ftz-lacZ pattern of the balancers. CHES-1-like and Myb heterozygotes were distinguished from hemizygotes by both positive staining for anti-Sxl since they were female and the absence of anti-β-galactosidase staining in the ftz-lacZ pattern. The same anti-β-galactosidase staining, along with anti-Mef-2 staining, was also used to distinguish the Svp-CCs and Svp-PCs except in the case of RNA interference assays. The neb^MI0225 allele, a weak hypomorphic mutation that fails to complement the neb^k05702 null allele, was used in genetic interaction assays because it allowed cell division defects to be assessed with increased sensitivity. The genotypes of all the embryos used for counting cell division defects (Supplementary Table S1) and presented in the figures are described below.

svp-lacZ/+	Figure 4a
nebk05702; svp-lacZ/+	Figure 4b
nebMI02225/+; svp-lacZ/+	Figure 5b
jumuDf(3R)Exel6157 svp-lacZ/+	Figure 5a
nebMI02225/+; jumuDf(3R)Exel6157 svp-lacZ/+	Figure 5c
polo10 svp-lacZ/+	Figure 5d
nebMI02225/+; polo10 svp-lacZ/+	Figure 5e
MybMH107/+; svp-lacZ/+	Figure 5f
MybMH107/+; nebMI02225/+; svp-lacZ/+	Figure 5g
CHES-1-like1/+; svp-lacZ/+	Figure 5h
CHES-1-like1/+; nebMI02225/+; svp-lacZ/+	Figure 5i
jumuDf(3R)Exel6157 svp-lacZ/jumuDf(3R)Exel6157	Figure 7a
TinD-GAL4/+; jumuDf(3R)Exel6157 svp-lacZ/jumuDf(3R)Exel6157 UAS-neb	Figure 7b
+/+	Figure 4d
UAS-nebRNAi/+	Suppl. Fig. S1
Hand-GAL4/+; TinD-GAL4 UAS-Dcr-2/+	Suppl. Fig. S1
Hand-GAL4/+; TinD-GAL4 UAS-Dcr-2/UAS-nebRNAi	Figure 3e

The following transgenes were used for RT-qPCR: twi-GAL4 [GAL4^twi.PG; FlyBase ID: FBal0040491]^70,71; UAS-Dcr-2 [described earlier]; UAS-jumu^RNAi [jumu^GD4099; FlyBase ID: FBal0209759]^28,69; and UAS-CHES-1-like^RNAi [CHES-1-like^GD4327; FlyBase ID: FBal0206447]^69,72. The genotypes of all the embryos used for the RT-qPCR assays are described below:

twi-GAL4 UAS-Dcr-2/+

twi-GAL4 UAS-Dcr-2/+; UAS-jumuRNAi/+

twi-GAL4 UAS-Dcr-2/+; UAS-CHES-1-likeRNAi/+

RNA interference assays

RNA interference knockdown of neb was performed using the UAS-neb^RNAi (neb^GD7072) construct targeted to the cardiac mesoderm using both TinD-GAL4 and Hand-GAL4 drivers simultaneously. The efficiency of knockdowns was enhanced with UAS-Dcr-2. Cell division defects were counted in the neb knockdown and controls (wild-type, UAS-neb^RNAi alone, and both GAL4 drivers with UAS-Dcr-2 but no UAS-neb^RNAi) by staining with anti-Mef2 and anti-Svp to identify Tin-CCs and Svp-CCs.

Reverse transcription quantitative real-time PCR (RT-qPCR) assays

Details of the RT-qPCR methodology and the mesoderm-targeted RNA interference knockdown of jumu and CHES-1-like used in these assays are included in Supplementary Method S1.

In situ hybridization, immunohistochemistry, microscopy, and cell counting

Embryo fixation, riboprobe synthesis, in situ hybridization, and fluorescent immunohistochemistry were performed as described previously^26,28. The following primary antibodies were used: rabbit anti-Mef2 (1:1000, gift from B. Paterson; 1:7500, gift from J. Jacobs), mouse anti-β-galactosidase (1:500, Catalog no. Z3783 from Promega), chicken anti-β-galactosidase (1:500, Catalog no. ab9631 from Abcam, Inc), mouse anti-Svp (1:5, monoclonal 5B11 from the Developmental Studies Hybridoma Bank), and mouse anti-Sxl (1:20, monoclonal M18 from the Developmental Studies Hybridoma Bank). Fluorescent microscopy was performed on a Zeiss AxioImager with Apotome. Z-stacks of entire stage 16 embryonic hearts were scanned with a 40X objective and 0.31 µm steps, and all planes for each z-stack were examined to count cells and determine cell division defects. Cell counting and assessment of cell division defects were performed blind (i.e. the individual evaluating cardiac progenitor cell division defects did not know the genotypes of the embryos being assessed) to avoid any potential bias. For all quantitative studies, cells in 179 or more hemisegments were counted for each genotype.

Statistical methods

Comparison of cell division error rates between genotypes was done using regression models with the response variable being the proportion of hemisegment errors for each embryo. Due to violation of regression assumptions, e.g., non-normality and heteroscedasticity, permutation (randomization) tests were used to obtain reliable p-values⁷³.

For comparing rates between two genotypes, for example neb^k05702 and wild-type, the following general linear model was used:

$${Y}_{j}= {\beta }_{0}+ {\beta }_{1} {I}_{j}+ {\varepsilon }_{j},$$

where ${Y}_{j}$ is the proportion of hemisegment errors for embryo j and indicator variable ${I}_{j}$ is 1 if embryo j has phenotype neb^k05702 and 0 otherwise. To obtain a permutation p-value for testing ${H}_{0}: {\beta }_{1}=0$, the estimate of ${\beta }_{1}$ for the actual data is compared with the estimates obtained when the genotypes of the embryos are permuted, i.e., the phenotype labels are randomly shuffled among the embryos in the sample. The permutation test p-value is then $p=(n+1)/(N+1)$ where $n$ is the number of permutation estimates for ${\beta }_{1}$ which exceed the estimate for the actual data and $N$ is the number of permutations⁷⁴. In order to obtain highly reproducible p-values, $N={10}^{6}$ permutations were used for all permutation tests.

For determining if cell division error rates are non-additively related to two gene mutations, for example, to detect synergistic interaction between neb and Myb, a general linear model allowing for interaction was used:

$${Y}_{j}= {\beta }_{1}{I}_{n,j}+{\beta }_{2}{I}_{M,j}+{\beta }_{3}{I}_{n,j}{I}_{M,j}+{\varepsilon }_{j},$$

where ${I}_{n,j}$ is 1 if the jth embryo is heterozygous for the neb^MI02225 mutation and ${I}_{M,j}=1$ if it is heterozygous for the Myb^MH107 mutation. Since synergism is present only if ${\beta }_{3}\ne 0$, to detect it ${H}_{0}: {\beta }_{3}=0$ was tested using permutation. Since this is a multiple regression model, a somewhat more sophisticated permutation procedure, the Smith procedure (orthogonalization) was employed⁷⁵.