trithorax is an essential regulator of cardiac Hox gene expression and anterior-posterior patterning of the Drosophila embryonic heart tube

Embryonic development requires precise transcriptional regulation which is partially controlled by the actions of the Trithorax group (TrxG) and Polycomb group (PcG) proteins. The founding member of the Drosophila TrxG genes, trithorax (trx), encodes a SET-domain-containing histone methyltransferase (HMT) within the COMPASS-like family of histone 3 lysine 4 (H3K4) methylation complexes (Kingston and Tamkun, 2014). The trx homologs, trithorax-related (trr) and SET domain containing 1 (Set1), also encode HMTs and form their own unique COMPASS-like complexes. The post-translational H3K4 methylation marks deposited by these complexes serve as docking sites to recruit chromatin remodeling complexes and other regulatory machinery necessary for transcriptional activation and maintenance during development (Cheng, 2014). Several studies have identified the trx and trr mammalian orthologs, lysine (K)-specific methyltransferase A-D (Kmt2a-d, also known as Mll1-4), as important epigenetic regulators of mammalian embryonic development (Shilatifard, 2012). Although the COMPASS-like HMTs activate and maintain the transcription of diverse gene classes, trx has maintained its evolutionarily conserved role as a positive regulator of Hox expression. Hox activity is a major determinant of the anterior/posterior (A/P) patterning in both insects and mammals and Hox genes have essential roles in orchestrating organ development (Roux and Zaffran, 2016; Soshnikova and Duboule, 2009). Multiple Hox genes cooperatively regulate the contributions of cardiac progenitor cell populations to embryonic heart patterning in both invertebrate and vertebrate organisms (Deschamps and van Nes, 2005; Lescroart and Zaffran, 2018). In mammals, the inactivation of the trx ortholog Kmt2b results in severe cardiac abnormalities and early embryonic lethality, indicating that these genes are necessary for proper cardiac gene expression (Glaser et al., 2006; Goldsworthy et al., 2013). Therefore, the function of COMPASS-like HMTs in the regulation of Hox and other cardiac genes remains an important area of investigation in developmental biology (Wang, 2012).

In this report, we investigated the role of the COMPASS-like HMTs trx, trr, and Set1 in cardiac development using the embryonic dorsal vessel (linear heart tube) of the fruit fly Drosophila melanogaster, which offered several advantages as an experimental system. The initial stages of invertebrate and vertebrate embryonic heart development share morphological, anatomical, and genetic similarities including the evolutionary conservation of a cardiac development regulatory network (Ahmad, 2017; Bodmer and Frasch, 2010; Bodmer and Venkatesh, 1998; Cripps and Olson, 2002; Meganathan et al., 2015; Olson, 2006; Tao and Schulz, 2007; Vogler and Bodmer, 2015). However, the Drosophila dorsal vessel develops as a single linear epithelial tube consisting of exactly 104 contractile cardial cells surrounded by a sheath of supportive and nephrocytic pericardial cells, thereby allowing the investigation of a wide-range of developmental and cell division defects at a single cell resolution. Additionally, while genetic duplication of the COMPASS-like HMTs has occurred within the mammalian genome, the Drosophila orthologs exist as single genes thereby reducing potentially confounding effects due to functional redundancy and compensation (Chintapalli et al., 2007; El-Brolosy and Stainier, 2017; Ocorr et al., 2007; Shilatifard, 2012).

Numerous cell division defects were identified throughout the dorsal vessel in the individual lethal mutants of trx, trr, and Set1 indicating an independent contribution of each gene to cardiac cell division control. Furthermore, we describe a critical role for trx in controlling the collinear expression of Hox genes, which pattern the A/P axis of the dorsal vessel. trx inactivation induces a remarkable homeotic transformation of the posterior heart-proper segment into an aorta-like fate due to the loss of abdominal A (abd-A) expression within the posterior dorsal vessel. Additionally, trx is necessary for the proper cardiac expression of Antennapedia (Antp), Ultrabithorax (Ubx), and Abdominal B (Abd-B). Together, these data indicate that the COMPASS-like HMTs are essential developmental regulators of cardiogenesis necessary for both cardiac cell division and heart patterning.

The Drosophila COMPASS-like HMTs trx, trr, and Set1 catalyze H3K4 methylation marks on chromatin associated with poised or active transcription of important developmental regulatory genes (Shilatifard, 2012). To explore the function of COMPASS-like HMTs in cardiogenesis, we utilized the Drosophila dorsal vessel. The Drosophila dorsal vessel at embryonic stage 16 consists of a myoepithelial tube that exhibits axial symmetry, A/P patterning, and a repeated cellular pattern (Figs 1A and 6A). This structure can simply be described as a linear heart tube composed of myoepithelial cardial cells (CCs) that generate the apical lumen and create the contractile force to propel hemolymph throughout the vessel. The CCs of the abdominal A2 to A8 hemisegments are organized into repeated metameric cellular structures which are primarily derived from the seven-up (svp) and tinman (tin) cardiac progenitor lineages (Alvarez et al., 2003; Han and Bodmer, 2003; Ward and Skeath, 2000). Each cardiac hemisegment consists of a repeated pattern of two Svp CCs followed by four Tin CCs, except for A8, which is abbreviated to two Svp CCs followed by only two Tin CCs (Figs 1A and 6A). In each cardiac hemisegment, a cell division event at an earlier stage generates two Svp cardiac progenitor cells, with each Svp progenitor cell subsequently undergoing asymmetric cell division to produce a Svp CC and an associated odd skipped-expressing pericardial cell (Alvarez et al., 2003; Han and Bodmer, 2003; Ward and Skeath, 2000). Thus, any deviation from the expected number of two Svp CCs in a hemisegment indicates a defect in either asymmetric or earlier cell division. In contrast, in the tin cardiac lineage, an initial symmetric cell division produces two Tin cardiac progenitor cells, each of which subsequently divides symmetrically to produce two myocardial Tin CCs (Alvarez et al., 2003; Han and Bodmer, 2003; Ward and Skeath, 2000). Hence a total of four Tin CCs are generated in each hemisegment and any alteration in the expected number of Tin CCs reflects defective symmetric cardiac progenitor cell division (Ahmad et al., 2014, 2012; Kump et al., 2021).

Drosophila embryos hemizygous or homozygous for lethal H3K4 HMT alleles of trr^B, Set1^G12, and trx^E2 were evaluated for cardiac cell division defects using this strategy. Svp and Tin CC numbers in individual hemisegments in these HMT mutants and wild-type embryos were assessed and compared by immunostaining with either anti-H15 (Neuromancer1, Nmr1) and anti-Seven up (Svp) antibodies (which labeled all CCs and exclusively Svp CCs, respectively) or anti-Myocyte enhancer factor 2 (Mef2) and anti-Svp antibodies (which labeled all CCs and exclusively Svp CCs, respectively) (Fig. 1A-D). To quantify cardiac cell division defects, the A2 to A8 hemisegments of homozygous trr^B or Set1^G12 mutant embryos were examined and scored for an increase or decrease of Tin CCs, corresponding to symmetric cell division defects, or Svp CCs, indicating earlier symmetric and/or later asymmetric cell division defects, compared to wild type. Our investigation found that embryos homozygous for either trr^B or Set1^G12 mutations exhibited a significant increase over wild type in the fraction of hemisegments exhibiting cell division defects for both the Tin and Svp lineages (Fig. 1A-C,E, Table S1). To confirm that these cardiac phenotypes were indeed a consequence of loss or reduction of trr or Set1 functions, embryos hemizygous and homozygous for the trr^C2375X and Set1^G5 alleles, respectively, were also evaluated for changes in CC numbers. Similarly significant increases over wild type in cell division defects were also obtained for these two alleles, demonstrating that trr and Set1 were essential for mediating proper cardiac cell divisons (Fig. 1E, Figs S1G-J and S2D-K, Table S1).

The trx^E2 homozygous mutant embryos, however, exhibited an intriguing, completely penetrant A2 hemisegment-specific phenotype. The expected two Svp CCs in each A2 hemisegment in the wild-type dorsal vessel were reduced to a single Svp CC in each and every one of the 30 trx mutant A2 hemisegments we examined (Fig. 1A,D, arrowheads, Fig. 2A, Fig. S2A). While we did observe changes in Svp CC number in A2 hemisegments in the other HMT mutants, they were not anywhere near as penetrant (4/30 for trr^B mutants and 7/30 for Set1^G12 mutants compared to 0/30 for wild-type embryos). Unlike the trx mutants, where the Svp CC number in the A2 hemisegments was consistently reduced to one, trr and Set1 mutants also exhibited A2 hemisegments with more than two Svp CCs compared to wild type. To verify that this cardiac phenotype was due to disruption of trx function, and not a consequence of a mutation at a second site, transheterozygous embryos of the genotypes trx^E2/trx^B11 (where trx^B11 is an amorphic trx allele) and trx^E2/Df(3R)BSC470 (where the Df(3R)BSC470 deficiency, henceforth referred to as Df(trx), completely deletes the trx gene) were evaluated in comparison with the individual trx^E2/+ and trx^B11/+ heterozygotes. Homozygous trx^B11 stage 16 embryos were not recoverable due to likely early embryonic lethality. The reduction of two Svp CCs to one in the A2 hemisegments in the trx^E2 mutants (Fig. 2) was also observed in the trx^E2/trx^B11 and trx^E2/Df(trx) transheterozygous embryos (Figs S1D-F and S3A).

Two distinct hypotheses could explain this result. The cell destined to be the anteriormost Svp CC in the A2 hemisegment in the wild-type embryo could simply fail to express detectable Svp protein in the trx mutant. An alternative explanation is a missing Svp CC in the A2 hemisegment in trx mutant embryos as a consequence of either an asymmetric or earlier cell division defect. The Svp CCs in each A2 hemisegment in the wild-type embryo are immediately preceded anteriorly by the 12 CCs of the anterior aorta that do not express Svp. Hence, if the first hypothesis, that of the anteriormost Svp CC of the A2 hemisegment failing to express Svp in trx mutants is correct, then the single Svp CC in the A2 hemisegment would be immediately preceded by 13 CCs that do not express Svp. In contrast, if the second hypothesis is correct, that a cell division defect prevents the development of more than one Svp CC in the A2 hemisegments of trx mutants, then the sole Svp CC in the mutant hemisegment and the two Svp CCs in wild-type A2 hemisegment would both be immediately preceded anteriorly by the same number of CCs, the 12 cells of the anterior aorta lacking Svp expression. Our quantification of the CCs immediately preceding the Svp CC(s) in A2 hemisegments revealed no statistically significant difference in cell number between wild-type embryos and trx^E2 mutants in either the left or the right row of the embryonic heart (the left row displayed an average of 12.1 CCs in wild-type embryos compared to the mutant's average of 12.6 CCs, P=0.592; while the wild type right row exhibited an average of 12.4 CCs compared to the trx mutant average of 12.6 CCs, P=0.879) (Fig. 2B). These results indicate that the precursors of the Svp lineage in A2 hemisegments of embryos lacking trx function did not reproducibly complete the stereotypical cell divisions to generate two Svp CCs (Alvarez et al., 2003; Gajewski et al., 2000; Han and Bodmer, 2003; Ward and Skeath, 2000).

Because trx mutants exhibited this additional completely penetrant A2 hemisegment-specific alteration in Svp CC number, we were concerned that incorporating data for the A2 hemisegments could potentially contaminate and skew our statistical assessment of cell division defects in the more posterior cardiac hemisegments. In order to eliminate any such influence or bias, we compared hemisegments A3 to A8 exclusively between embryos lacking trx function, embryos heterozygous for trx mutants, and wild-type embryos for changes in Tin CC and Svp CC numbers. Our results showed that trx mutants also exhibited significantly more symmetric cell division defects (indicated by more or fewer Tin CCs) compared to wild-type embryos in hemisegments A3 to A8 (Fig. 1F, Figs S1 and S2, Table S1). However, in contrast to our observations with trr and Set1 mutants, we found that the fraction of A3 to A8 hemisegments exhibiting changes in the expected Svp CC number in trx^E2 mutants was not significantly different from that in wild type embryos (Fig. 1F, Table S1). Similarly, the fraction of A3 to A8 hemisegments exhibiting changes in the Svp CC number in the trans-heterozygous trx^E2/Df(trx) embryos was not significantly different from that in heterozygous trx^E2/+ embryos, and the fraction in trx^E2/trx^B11 was not significantly different from that in either trx^E2/+ or trx^B11/+ embryos (Fig. 1F, Fig. S1, Table S1). Thus, in the posterior A3 to A8 hemisegments, trx is essential for mediating symmetric cardiac cell divisions, but is not required for either asymmetric cardiac progenitor cell divisions, or for the cell divisions at an earlier stage that give rise to a pair of Svp progenitors.

The Drosophila dorsal vessel is patterned into two anatomical areas along the A/P axis: the posterior heart-proper region (hemisegments A6 to A8), which absorbs fluid from the posterior embryo and propels hemolymph into the aorta, and the aorta (hemisegments T2 to A5), which transports hemolymph to the anterior embryo (Fig. 6A). The heart-proper contains columnar CCs with elliptical nuclei, which surround an enlarged lumen, while the long aorta consists of cuboidal CCs, which generate a narrow lumen. The aorta itself can be further subdivided into two regions based upon the presence of Svp CCs. The anterior aorta (hemisegments T2 to A1) is devoid of Svp CCs, while the posterior aorta (hemisegments A2 to A5) and heart-proper are both organized into hemisegments containing both Svp CCs and Tin CCs. However, the Svp CCs in the posterior aorta can be distinguished from those in the heart-proper since only the latter differentiate into valvular inlet ostia cells as a result of both svp and abd-A activity (Curtis et al., 1999; Lo et al., 2002; Lovato et al., 2002; Molina and Cripps, 2001; Ponzielli et al., 2002).

While the trr mutant and Set1 mutant embryos exhibited a dilated lumen in the heart-proper region similar to that of wild-type embryos (Fig. 1A-C), embryos lacking trx function exhibited heart-proper regions that resembled the narrow-luminal, cuboidal epithelial appearance of the posterior aorta (Fig. 1D). The average lumen width of the posterior aorta and heart-proper region within the wild type is 0.66 µm and 2.18 µm, respectively (Fig. S4). Specifically, our analysis revealed that while the mean width of the lumen of the heart-proper regions was significantly greater than that of the posterior aorta in wild-type, trr mutant and Set1 mutant embryos (P<0.0001 for wild type, P=0.0021 and P<0.0001 for trr^B/Y and trr^C2375X/Y genotypes, respectively, and P<0.0001 for either Set1^G12 or Set1^G5 genotypes), there was no significant difference in the width of the lumen between these two regions in embryos of the genotypes trx^E2 (P=0.1976), trx^E2/Df(trx) (P=0.0601), and trx^E2/trx^B11 (P=0.1453) (Fig. S4). To better visualize the morphology of the dorsal vessel, wild-type and trx mutant embryos were immunostained using antibodies directed against Svp and H15, a cardiac T-box transcription factor with expression restricted to the CC lineage (Qian et al., 2005a). Confirming our results in Fig. 1D, we identified the aorta-like morphology of the heart-proper region within trx mutant embryos (Fig. 3A-B″,AA-BB″). Additionally, closer inspection revealed another significant discrepancy between the CCs in wild-type and trx mutant embryos: although the Svp CCs are readily identifiable within trx mutants, the Svp CCs of the dorsal vessel in embryos lacking trx function express reproducibly lower levels of Svp protein compared to those in wild-type controls (Fig. S3A′,B′,AA′,BB′). Similar reductions in Svp protein levels were also identified within the trx^E2/trx^B11 and trx^E2/Df(trx) mutants (Fig. S1E′,F′).

Within the A6-A8 hemisegments of the heart-proper, both svp and abd-A are necessary for the induction of wingless (wg) expression within the Svp cardiac progenitors and the differentiation of these cells into the valvular ostial cells (Lo et al., 2002; Ponzielli et al., 2002; Trujillo et al., 2016). Consequently, in wild-type embryos, the Svp CCs of the heart-proper (hemisegments A6 to A8) are distinguishable from the Svp CCs in the posterior aorta (hemisegments A2 to A5) since the former express both svp and wg while the latter express svp alone.

The narrow-luminal aorta-like morphology of the heart-proper region in embryos lacking trx function that we observed phenocopies that of abd-A mutant embryos (Lo et al., 2002; Lovato et al., 2002; Perrin et al., 2004; Ryan et al., 2005) and suggests that the posterior dorsal vessel in trx mutant homozygotes may have undergone a homeotic transformation and adopted the anterior fate of the aorta. If this is true, then the Svp CCs in the homeotically transformed heart-proper region of trx mutant embryos would express only svp, not both svp and wg (Lo et al., 2002; Perrin et al., 2004). While Wg immunostaining readily identifies the Svp CC ostial pairs within the wild-type embryo, Wg protein is completely lost within the trx mutant embryo and coincides with the absence of the expected ostial cell morphology (Fig. 3C-D″,CC-DD″, arrows). Therefore, the heart-proper region is indeed homeotically transformed in embryos lacking trx function in a manner which phenocopies the loss of wg-expressing ostial cells within abd-A mutants.

Several extracellular proteins necessary for the development and maintenance of the cardiac lumen exhibit either uniform or differential expression within the trx dorsal vessel. Whereas slit expression is necessary for cardiac luminal development and is localized uniformly across the entire myoepithelia of dorsal vessel (Fig. 3E-E″,EE-EE″), high expression of Multiplexin (Mp) is required for the wider heart-proper luminal dilation and is localized to a more posterior domain (Harpaz et al., 2013; MacMullin and Jacobs, 2006; Qian et al., 2005b; Santiago-Martínez et al., 2006; Volk et al., 2014). Specifically, a gradient of Mp is detected within the wild-type dorsal vessel with levels precipitously declining along the posterior aorta in a posterior-to-anterior manner until the anterior region is devoid of signal (Fig. 3G-G″,GG-GG″). Slit immunostaining reveals similar levels of expression within both wild-type and trx mutant embryos suggesting that aorta lumen formation and epithelialization may not be significantly affected by trx loss of function (Fig. 3E-F″,EE-FF″). However, Mp immunostaining reveals a dramatic absence of this protein within the trx mutant dorsal vessel compared to that in the wild-type control consistent with a complete loss of heart-proper specification (Fig. 3G-H″,GG-HH″). Collectively, these data and the observations described in the previous sections identify several heart-proper and posterior dorsal vessel patterning defects within the trx mutant and confirm a homeotic transformation of the heart-proper into a posterior aorta-like fate.

Drosophila trx and its vertebrate orthologs partially regulate the collinear Hox selector gene expressions that in turn pattern the A/P axis of the embryo (Breen and Harte, 1991, 1993; Hanson et al., 1999; Ingham, 1998; Yu et al., 1995). Although the knowledge of TrxG regulation of Hox activity in embryo patterning is well understood, trx regulation of cardiac-specific Hox expression in heart development remains an open area of investigation. Fig. 6A summarizes the collinear expression of the Antennapedia (Antp) and Bithorax complex (Bx-C) Hox genes Ubx, abd-A, and Abd-B across A/P axis of the dorsal vessel in wild-type embryos (Lo and Frasch, 2003; Lovato and Cripps, 2016; Monier et al., 2007). Hox inputs play an additional role in Svp CC progenitor specification across this axis (Perrin et al., 2004; Ryan et al., 2005). Together with our observations that the posterior dorsal vessel is homeotically transformed into the aorta in trx mutants in a manner reminiscent of abd-A mutants, these data suggest that trx may control cardiac fate along the A/P axis by regulating collinear Hox expression. If this is true, then the expression of the Hox genes would be expected to be significantly altered in trx mutants. Thus, to investigate Hox expression within the dorsal vessel, stage 16 trx mutant and wild-type embryos were immunostained with a panel of antibodies directed against the known Hox genes expressed within the dorsal vessel (Figs 4 and 5, Figs S5, S6, S7).

Previous studies have identified abd-A as the primary Hox selector gene responsible for specifying the heart-proper region (Lo et al., 2002; Lovato et al., 2002; Ponzielli et al., 2002). In the wild-type control, Abd-A immunostaining identifies high levels of expression within the A5-A8 segments that form the heart-proper (Fig. 4A-A″,AA-AA″, arrows). As predicted from the loss of heart-proper specification, immunostaining in the trx^E2 mutants reveals a complete loss of abd-A expression within the posterior dorsal vessel (Fig. 4B-B″,BB-BB″ compared to Fig. 4A-A″, AA-AA″, arrows). Similar results were obtained for the trx^E2/trx^B11 and trx^E2/Df(trx) transheterozygous mutants compared to trx^E2/+ and trx^B11/+ heterozygotes (Fig. S5C-D″,CC-DD″ compared to Fig. S5A-B″,AA-BB″, arrows). Furthermore, these results replicate the previously reported loss of cardiac abd-A expression of the amorphic trx^B11 strain (Breen and Harte, 1993). Since the loss of trx function abrogates abd-A expression and phenocopies the abd-A mutant cardiac phenotype, the loss of abd-A expression in the trx mutant is likely the cause of the heart-proper to aorta transformation.

Next, we evaluated cardiac Abd-B expression, which is restricted to the posterior-most A8 segment of the wild-type dorsal vessel (Fig. 4C-C″,CC-CC″, arrows). Previous reports indicate that Abd-B activity suppresses the formation of the cardiac mesodermal lineage within the posterior-most parasegments of the wild-type embryo (Lo et al., 2002; Lovato et al., 2002; Ponzielli et al., 2002). Although heart-proper formation is not affected in the Abd-B mutants, the posterior most A8 hemisegment of the dorsal vessel is populated by supernumerary CCs, which may be derived from the loss of Abd-B repression of cardiac mesoderm specification within parasegment 13 (Lo et al., 2002). We found that Abd-B expression is completely lost from the A8 cardiac segment within the trx mutant in contrast to wild-type embryos (Fig. 4D-D″,DD-DD″ compared to Fig. 4C-C″,CC-CC″, arrows). This result was also replicated in the trx^E2/trx^B11 and trx^E2/Df(trx) transheterozygous mutants compared to trx^E2/+ and trx^B11/+ heterozygotes (Fig. S6C-D″,CC-DD″ compared to Fig. S6A-B″,AA-BB″, arrows).

To investigate whether the trx mutants also phenocopy the supernumerary CC defect in Abd-B mutants, the number of CCs within the A8 segment (comprised of both contralateral A8 hemisegments) of the wild-type and trx^E2 mutant dorsal vessels were quantified. This analysis revealed a statistically significant increase within the Tin CC population of the trx^E2 mutant (6.7 CCs) compared to the wild type (4.0 CCs); in contrast, the number of Svp CCs in the A8 hemisegments did not exhibit any significant variation between wild-type and trx mutant embryos (Fig. 2A). Analysis of the A8 Svp and Tin CCs identified similar results within the trx^E2/trx^B11 (6.1 CCs) and trx^E2/Df(trx) (5.2 CCs) transheterozygous mutants compared to trx^E2/+ (4.1 CCs) and trx^B11/+ (3.9 CCs) heterozygotes (Fig. S3B). Thus, trx functions to restrict Tin CC progenitor differentiation in these posteriormost cardiac hemisegments by activating Abd-B expression in the posterior embryonic segments.

Collectively, the findings reported in this, and the previous section identify a critical role for trx in maintaining cardiac abd-A and Abd-B expression within the heart-proper region.

The cardiac Antp activity facilitates the generation of two Svp CC pairs within the A2 segment of the dorsal vessel (Perrin et al., 2004; Ryan et al., 2005). Interestingly, the individual Antp and trx mutants share a common A2 phenotype: the expected two pairs of Svp CCs in the A2 segments are found to be reduced to a single pair (Fig. 1A compared to Figs 1D and 3A′ compared to Fig. 3B′, arrowheads, Fig. S1D-F″ compared to Fig. S1A-C″). Given this similar phenotype, it is likely that trx may regulate cardiac cell divisions along the Svp lineage in A2 by controlling Antp expression. Therefore, Antp immunostaining was performed for both wild-type and trx mutant embryos to assess potential Antp expression changes (Fig. 5A-B″, Fig. S7A-D″). The Antp expression domain within the wild-type embryo is located at the A1/A2 boundary, which separates the anterior and posterior regions of the aorta (Fig. 5A-A″, arrows). In the trx mutant, Antp is completely lost within the anterior dorsal vessel, thereby indicating that trx activity is indeed required for cardiac Antp expression (Fig. 5B-B″ compared to Fig. 5A-A″, arrows). Evaluation of the trx^E2/trx^B11 and trx^E2/Df(trx) transheterozygous mutants revealed a similar loss of Antp protein at the A1/A2 boundary (Fig. S7C-D″ compared to Fig. S7A-B″).

To investigate whether the reduction of A2 Svp CC pairs was a reproducible cell division defect within the trx mutant, the Svp and Tin CCs of the A2 segments were quantified in both wild-type embryos and trx mutants. This analysis revealed a statistically significant reduction of Svp CCs in the A2 segment from a mean of 4.0 in wild-type embryos to a mean of 1.9 in the trx^E2 mutants. In contrast, the number of A2 Tin CCs were not significantly changed between wild-type and trx^E2 mutant embryos (Fig. 2A). Assessment of the trx^E2/trx^B11 and trx^E2/Df(trx) transheterozygous mutants identified similar results (Fig. S3A). As previously mentioned, the number of CCs immediately anterior to the Svp CCs in the wild-type and trx^E2 mutant A2 segments exhibited no significant difference (Fig. 2B), suggesting that Antp is necessary for proper Svp lineage cell division rather than Svp expression. These data indicate that trx is required for Antp expression and subsequent A2 segment-specific cell division in the Svp lineage in a manner similar to that of the Hox gene Antp. The most parsimonious explanation for these observations is that trx activates Antp expression to mediate proper cardiac cell division in the Svp lineage of the A2 hemisegments.

Posterior to the Antp expression domain, Ubx activity within the posterior aorta specifies the Svp CCs of the A3-A5 segments (Ryan et al., 2005). Interestingly, Ubx is expressed at different levels along the dorsal vessel in wild-type embryos, being highest across the A2-A4 hemisegments of the posterior aorta and declining where it overlaps the Antp and abd-A expression domains (Fig. 5C-C″) (Perrin et al., 2004). In contrast to the other Hox genes we observed whose expression is eliminated in embryos lacking trx function, Ubx expression persists within the trx mutant, but its expression pattern is significantly altered (Fig. 5D-D″ compared to Fig. 5C-C″). The noticeably higher levels of Ubx expression within the A2-A4 hemisegments in wild-type embryos are reduced such that a uniform and consistent level of Ubx is expressed throughout the dorsal vessel in trx mutants (Fig. 5D-D″, arrows). Interestingly, Ubx immunostaining of the trx^E2/trx^B11 and trx^E2/Df(trx) transheterozygous mutants identified similarly lower protein levels within the posterior heart tube (Fig. S7E-H″).

Together, the collinear expression of Antp, Ubx, and abd-A within their corresponding domains is necessary to specify the Svp progenitor lineage and ensure the proper cell division of these Svp progenitors since the loss of all Hox function results in a complete loss of Svp CCs (Perrin et al., 2004). The continued expression of Ubx combined with the absence of Antp and abd-A expression within the trx mutant dorsal vessel suggests that this remaining Ubx activity is sufficient to maintain Svp CC lineage specification and cell divisions throughout the dorsal vessel except for the Svp CCs located within A2 hemisegments (Fig. 1D compared to Figs 1A and 3B-B″ compared to Fig. 3A-A″, Fig. S1D-F″ compared to Fig. S1A-C″). However, the expression of merely Ubx in the posterior dorsal vessel in trx mutants is insufficient to drive heart-proper specification, consistent with prior studies identifying the abd-A function as the primary selector for this process (Lo et al., 2002; Lovato et al., 2002; Perrin et al., 2004; Ponzielli et al., 2002).

In addition to their roles of important transcriptional co-activators, the trx, trr, and Set1 genes encode essential HMTs of unique COMPASS-like complexes that catalyze H3K4 methylation, an important chromatin modification associated with transcription (Cheng, 2014; Kingston and Tamkun, 2014; Shilatifard, 2012). Our study reveals that each of these genes contributes to the regulation of Svp and Tin cardiac cell divisions necessary for the cellular organization of the embryonic dorsal vessel. We show further that one of these three HMTs, trx, patterns the cardiac A/P axis by maintaining the proper collinear expression of Hox genes.

Although this study describes the zygotic role of these genes in cardiac cell division and patterning, we cannot rule out the possibility that maternal RNA or protein persist within the cardiac lineage of these mutants. Indeed, trx, trr, and Set1 are expressed in early embryonic development prior to gastrulation which may provide residual activity in the mutants (Hallson et al., 2012; Prudencio et al., 2018; Sedkov et al., 1994). However, our study does identify and describe the importance of zygotic gene function of these genes in embryonic cardiac development. Recently, Zhu and colleagues (2023) have described specific roles for trx, trr, and Set1 in the larval and adult Drosophila heart. Not only do the COMPASS-like HMTs regulate larval cardiac cell division by restricting over production of Tin CCs, but these genes also maintain global H3K4 methylation levels and myocardial gene expression necessary for normal cardiac physiology in the adult heart (Zhu et al., 2023).

Intriguingly, trx loss of function results in the complete loss of Antp, abd-A, and Abd-B expression within the dorsal vessel which in turn recapitulates the combined cardiac phenotypes of each individual Hox mutant (Lo et al., 2002; Lovato et al., 2002; Perrin et al., 2004; Ponzielli et al., 2002; Ryan et al., 2005). Previous work had identified abd-A as the primary selector gene for heart-proper development since its expression is both necessary and sufficient to drive heart-proper patterning within the dorsal vessel (Lo et al., 2002; Lovato et al., 2002; Perrin et al., 2004; Ponzielli et al., 2002). Thus, the most profound effect of trx inactivation is the loss of abd-A expression. It is this loss of abd-A expression that is responsible for the homeotic transformation of the heart-proper into a posterior aorta-like fate.

Second, the trx regulation of cardiac Antp and Abd-B expression plays important roles in defining the A/P patterning of the cardial cells along the dorsal vessel. At the A1/A2 segment boundary, Antp activity is required for the proper Svp cardiac cell progenitor divisions that produce a symmetric pair of two Svp CCs within the A2 hemisegment; hence the elimination of Antp expression in the dorsal vessel in trx mutant embryos results in only one Svp CC being produced in each A2 hemisegment. At the posterior end of the heart-proper, Abd-B expression suppresses excessive Tin CCs, thereby ensuring the production of a partial contralateral hemisegment consisting of two Svp CCs followed by two Tin CCs each (Lo et al., 2002; Lovato et al., 2002; Perrin et al., 2004). In trx mutant hearts, in the absence of Abd-B expression, this suppression of excessive Tin CCs in the A8 hemisegments is eliminated, resulting in supernumerary Tin CCs.

In contrast to the complete elimination of Antp, abd-A, and Abd-B expression in the dorsal vessel in trx mutant embryos, the loss of trx function merely alters the Ubx expression to a uniformly consistent level along the dorsal vessel. Our results thus suggest that in trx mutants, Ubx, by itself, may provide sufficient Hox activity to drive Svp CC specification. Within the wild-type embryo, Ubx expression is tightly controlled via multiple mechanisms involving auto-regulation, cross-regulation via other Hox genes, and several cis-regulatory elements (Bienz and Tremml, 1988; Karch et al., 1990; Macias et al., 1990; Maeda and Karch, 2006). Although the persistence of Ubx expression in trx mutants might initially suggest a regulatory mechanism that is not directly trx-dependent, we cannot rule out the possibility that trx may be directly responsible for the observed high levels of Ubx expression within wild-type A2-A4 cardiac hemisegments.

In conclusion, we have characterized the developmental functions of the Drosophila COMPASS-like H3K4 HMTs in heart development. Whereas trx, trr, and Set1 all participate in the regulation of cardiac cell division, trx is also necessary for collinear Hox expression and the A/P patterning of the dorsal vessel. Our data suggest an evolutionarily conserved role for trx regulation of cardiac Hox expression in heart development. Given the importance of Hox activity in heart development and our findings in this report, we propose that the trx ortholog Kmt2b may play critical roles in cardiac Hox gene expression during mammalian heart development.

The following mutant alleles and transgenes were obtained from the Bloomington Drosophila Stock Center: trx^E2 [FlyBase ID: FBal0017174] (Kennison and Tamkun, 1988), trr^B [FlyBase ID: FBal0323346] (Kanda et al., 2013), trr^C2375X [FBal0323346] (Kanda et al., 2013), Set1^G12 [FlyBase ID: FBal0265880] (Hallson et al., 2012), Set1^G5 [Flybase ID: FBal0265881] (Hallson et al., 2012), Df(3R)BSC470, designated Df(trx) in text, [Flybase ID: Fbal0265881] (Cook et al., 2012). The trx^B11 strain [FBal0032908] was provided by J. A. Kennison (Mazo et al., 1990). The trx and Set1 alleles were maintained over TM3, ftz-lacZ balancers and trr alleles were maintained over FM7c, ftz-lacZ X-chromosome balancers before self-crossing. The embryos were genotyped by the absence of anti-β-galactosidase immunostaining of the expected ftz-lacZ balancer expression pattern. For validation of the trx^E2 and Set1^G12 mutant alleles, complementation tests were utilized. The trx^E2 allele failed to complement the Df(3R)BSC470 deficiency, henceforth described as Df(trx), and the Set1^G12 allele failed to complement the Set1^G5 allele. All genotypes used for analysis of cell division defects, wg expression, Mp expression, or Hox gene expression are described below.

Embryo fixation and fluorescent immunohistochemistry were performed as described previously examining protein expression (Ahmad et al., 2016, 2014). However, for Svp and Wg protein immunostaining, heat fixation with embryo wash buffer was utilized. Briefly, embryos were boiled in embryo wash buffer (140 mM NaCl, 0.03% Triton-X 100) for 10 s and immediately cooled with 4°C embryo wash buffer before devitellinization in 50:50 mixture of heptane and methanol followed by storage in methanol before staining. The following primary antibodies were used: guinea pig anti-H15 (NMR1, 1:1000, gift from J. B. Skeath) (Leal et al., 2009), rabbit anti-Mef2 [1:1000, Clone ID: Mef2 from Developmental Studies Hybridoma Bank (DSHB) and gift from J. Jacobs], chicken anti-β-galactosidase [1:500, catalog no. ab9361 (RRID:AB_307210) from Abcam, Inc], mouse anti-Svp (1:5, 5B11 from DSHB), rat anti-Mp (1:500, gift from T. Volk) (Harpaz et al., 2013), mouse anti-Ubx (1:10, Clone ID: FP3.38 from DSHB), mouse anti-Abd-A [1:100, Clone C-11 (sc-390990) from Santa Cruz Biotechnology, Inc.], mouse anti-Antp (1:20, Clone ID: 8C11 from DSHB), mouse anti-Wg (1:10, Clone ID 4D4 from DSHB), mouse anti-Slit (1:10, Clone ID: C555.6D from DSHB), and mouse anti-Abd-B (1:12.5, Clone ID: 1A2E9 from DSHB). Fluorescent microscopy was performed on a Zeiss AxioImager with Apotome. Z-stacks of stage 16 embryonic hearts were scanned at 20X with 0.60 µm intervals for anti-Mef2, anti-H15, anti-Svp, anti-Wg, anti-Slit, and anti-Hox immunostaining. Anti-Mp immunostaining was scanned at 40X with 0.31 µm intervals.

The Zen software suite of the Zeiss AxioImager microscope was used to measure the internal lumen diameter at five specific points along the dorsal vessel for Z-stacks of each embryo. The lumen width of the posterior aorta of a particular embryo was calculated as the mean of the internal lumen diameters measured across the middle of the four pairs of Tin CCs in the A2, A3, and A4 segments. The lumen width of the heart proper region of a specific embryo was the mean of the internal lumen diameters measured across the middle of the four pairs of Tin CCs in the A6 and A7 segments (Fig. S3).

A comparison of cell division errors between genotypes using Fisher exact tests is not appropriate as the hemisegments for a given embryo are correlated in their likelihood of having such errors (Ahmad et al., 2012). Comparison of cell division error rates between genotypes was thus 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-values (Manly, 2007).

For quantitation of Tin CC and Svp CC number changes in A2 and A8 hemisegments of wild type and trx^E2 mutant embryos, two-tailed, two-sample unequal variance t-tests were used with significance level of 0.05 (SPSS Statistics software; IBM; Version 29). For comparisons of Tin CC and Svp CC number changes in A2 and A8 hemisegments among multiple genotypes [wild type, trx^E2/+, trx^B11/+, trx^E2/trx^B11, and trx^E2/Df(trx)], a one-way ANOVA was completed with SPSS BCa 5000 sample bootstrapping to assess between group statistical differences and significance for hemisegment specific differences.

ANOVAs with a significance level of 0.05 (SPSS Statistics software; IBM; Version 29) were used to compare the number of CCs immediately anterior to the Svp CCs of the A2 hemisegments.

Two-tailed, two-sample unequal variance t-tests with significance level of 0.05 (GraphPad; Dotmatics) were used for comparing lumen widths of the posterior aorta and the heart proper regions of appropriate genotypes.

We thank the Bloomington Drosophila Stock Center (National Institutes of Health grant P4OOD018537) for providing fly strains. We thank James A. Kennison for the kind gift of the trx^B11 strain. We thank Developmental Studies Hybridoma Bank created by the NICHD of the NIH and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242, USA, for providing antibodies.