The Mevalonate Pathway Controls Heart Formation in Drosophila by Isoprenylation of Gγ1

See allHide authors and affiliations

Science  01 Sep 2006:
Vol. 313, Issue 5791, pp. 1301-1303
DOI: 10.1126/science.1127704


The early morphogenetic mechanisms involved in heart formation are evolutionarily conserved. A screen for genes that control Drosophila heart development revealed a cardiac defect in which pericardial and cardial cells dissociate, which causes loss of cardiac function and embryonic lethality. This phenotype resulted from mutations in the genes encoding HMG-CoA reductase, downstream enzymes in the mevalonate pathway, and G protein Gγ1, which is geranylgeranylated, thus representing an end point of isoprenoid biosynthesis. Our findings reveal a cardial cell–autonomous requirement of Gγ1 geranylgeranylation for heart formation and suggest the involvement of the mevalonate pathway in congenital heart disease.

Mutations in genes controlling heart development frequently cause fatal cardiac malformations, the most common type of birth defect in humans. Because many of the mechanisms involved in heart development are evolutionarily conserved, the fruit fly Drosophila melanogaster represents a powerful model for genetically dissecting this complex developmental process. The Drosophila heart, or dorsal vessel, which pumps bloodlike cells through an open circulatory system, is composed of parallel rows of contractile cardial cells (cardioblasts) tightly attached to pericardial cells; the latter perform supportive and secretory functions (Fig. 1A) (1).

Fig. 1.

Mutants in different genetic loci gave rise to a common broken hearted (bro) cardiac defect. (A) Schematic drawing of a late stage 17 embryonic heart (dorsal view, anterior to the left). (B to G) Stage 17 embryonic heart labeled by Hand-GFP (3) in wild-type embryo (B) or five bro homozygous mutants [(C) to (G)] (pericardial cells are indicated by arrows) (C) HMGCR, bro1, [l(3)01152]; (D) GGPPS/qmL14.4, bro2; (E) βGGT-IS-2554, bro3; (F) Gγ1, bro4, [l(2)k08017]. (G) Sar1, bro5, [l(2)k07408].

We performed a P-element genetic screen (2) for Drosophila mutants with heart defects using transgenic flies harboring a green fluorescent protein (GFP) transgene under control of the Hand enhancer (3), which is specific for cardial cells, pericardial cells, and the lymph gland—a hematopoietic organ in fruit flies (Fig. 1B). The Hand-GFP transgene allows visualization of the developing heart at single-cell resolution. Among a collection of mutants with cardiac abnormalities, we observed a heart defect in which pericardial cells dissociated from cardioblasts in the dorsal vessel at the end of embryogenesis. We termed this phenotype “broken hearted” (bro). Here, we describe five such mutants of different genetic loci (Fig. 1, C to G). In contrast to the wild-type dorsal vessel in which the pericardial cells are intimately associated with cardioblasts, in each of these mutants, the relative positions of pericardial cells and cardioblasts changed with each heartbeat.

The P element in the bro1 locus [l(3)01152] is located in the first exon of the hydroxymethyl-glutaryl (HMG)coenzyme A (CoA) reductase gene (HMGCR) (Fig. 1C and fig. S1, A, C, and D), which is expressed in the dorsal vessel and the gonadal mesoderm, where it is required for migration of primordial germ cells (4). Mutants trans-heterozygous for HMGCR01152 and a deficiency line Df(3R)Exel9013, in which the HMGCR gene is deleted, or two EMS mutants (2), HMGCRclb26.31 and HMGCRclb11.54 (4), showed similar, but more severe, cardiac defects than homozygous HMGCR01152 mutants (Fig. 2, A and B, and fig. S1B). Expression of HMGCR in the heart, with the use of a Hand-GAL4 driver and a UAS-HMGCR transgene, rescued the cardiac defects in the HMGCR01152 mutant (Fig. 2C).

Fig. 2.

HMGCR is required for Drosophila heart formation. (A) Trans-heterozygous mutant HMGCR01152/Df(3R)Exel9013 showed a severe bro defect. (B) HMGCR EMS mutant HMGCRclb11.54 showed a more severe cardiac defect in which the posterior region of the heart tube was abnormally dilated (arrowhead). (C) Expression of HMGCR specifically in the heart using Hand-GAL4 and UAS-HMGCR is sufficient to rescue the bro defect in HMGCR01152 embryos. (D) Embryos injected with 0.1 μM mevinolin showed cardiac defects similar to those of HMGCR null mutants at stage 17. The dissociated pericardial cells in (A), (B), and (D) are indicated by arrows. (E) HMGCR regulated isoprenoid synthesis pathway in Drosophila.

HMGCR controls a rate-limiting step in the conversion of HMG-CoA into mevalonate, a precursor for the synthesis of cholesterol and isoprene derivatives that modify the C termini of proteins containing a CAAX motif (C, cysteine; A, aliphatic amino acid; X, any amino acid) (5) (Fig. 2E). In contrast to mammalian cells, Drosophila does not use the mevalonate pathway to synthesize cholesterol. Injection of embryos at the syncytial blastoderm stage with 0.1 μM mevinolin, a statin drug that lowers cholesterol level by inhibiting HMGCR activity, caused cardiac defects at stage 17 similar to those of the HMGCR mutants (Fig. 2D).

To investigate whether either of the two major isoprenoids (Fig. 2E), farnesyl pyrophosphate (farnesyl-PP) and geranylgeranyl pyrophosphate (geranylgeranyl-PP), might be required for heart formation, we examined mutants in the genes encoding geranylgeranyl pyrophosphate synthase (GGPPS) and geranylgeranyl transferase type I β subunit (βGGT-I), which act downstream of HMGCR and are required for the biosynthesis of geranylgeranyl-PP or transfer of geranylgeranyl-PP to protein, to find out whether they also cause cardiac defects. Indeed, GGPPS (also called qm) mutant embryos showed 100% penetrance for the bro phenotype, just as HMGCR mutants did (Fig. 1D), and at least 30% of the βGGT-I mutants displayed the same phenotype (Fig. 1E). In contrast, two deficiency lines [Df(2L)Exel6010 or Df(3R)Exel6269] deleting either the farnesyl transferase α(CG2976) or β(CG17565) subunit did not display similar cardiac defects (fig. S2). These findings suggested that the cardiac defects of HMGCR mutant embryos resulted from a failure of geranylgeranylation of a target substrate protein required for the adhesion between cardioblasts and pericardial cells.

Analysis of another bro mutant (bro4) (Fig. 1F) suggested that the G protein γ subunit 1 (Gγ1), which contains a C-terminal CAAX motif, is the substrate of this geranylgeranylation modification required for heart formation. The P element in the bro4 locus l(2)k08017 is inserted into the splice donor site after the first exon of the Gγ1 gene (fig. S3A). Gγ1 expression level was reduced by more than 50% in homozygous l(2)k08017 embryos (fig. S3, B and C), which suggested that l(2)k08017 is a hypomorphic mutant allele of the Gγ1 gene. Mutants trans-heterozygous for the l(2)k08017 insertion and a deficiency that deletes the Gγ1 gene [Df(2R)H3E1] or for Df(2R)H3E1 and a Gγ1 null allele (6) showed the same cardiac defects as the homozygous l(2)k08017 embryos (Fig. 3A and fig. S3D). Double mutants of the hypomorphic HMGCR and Gγ1 alleles showed a more severe cardiac defect than either single mutant (Fig. 3B). A fifth bro mutation was mapped to the Sar1 gene (Fig. 1G and fig. S4), which encodes a guanosine triphosphatase that controls budding of vesicles overlaid with coat protein complex II (COPII) from the endoplasmic reticulum (ER) to the Golgi network (7).

Fig. 3.

Geranylgeranylation of Gγ1 is required for heart formation. (A) Transheterozygous mutant Gγ1k08017/Df(2R)H3E1 shows the same bro defect as the Gγ1k08017 mutant. (B) Gγ1k08017; HMGCR01152 double mutants show a much more severe bro defect than either single mutant. (C) Gγ1 can be efficiently geranylgeranylated, as labeled by [3H]GGPP, but cannot be farnesylated by [3H]FPP. Recombinant GST (27 kD) and GST-Gγ1 (36 kD) proteins are shown by Coomassie blue staining. (D) Expression of Gγ1 in the heart using Hand-GAL4 driving UAS-Gγ1 is sufficient to rescue the Gγ1k08017 mutant cardiac defects. (E) Targeted expression of Gγ1 (C67S) in the heart failed to rescue the Gγ1k08017 cardiac defects. The dissociated pericardial cells [(A), (B), and (E)] are indicated by arrows. (F) Subcellular localization of wild-type or mutant Gγ1 following exposure to mevinolin or HMCGR RNAi in S2R+ cells. Gγ1 is localized to the cytosolic compartment (a and a′), whereas the two mutant forms show nuclear and cytoplasmic localization (b, b′, c, and c′). In the presence of mevinolin (d and d′) or HMGCR double-stranded RNA treatment (e and e′), Gγ1 shows a ubiquitous localization as seen with the two mutant forms.

The developmental onset of cardiac defects was identical in the HMGCR, Gγ1, GGPPS/qm, βGGT-I, and Sar1 mutants. Cardioblasts and pericardial cells were properly specified and aligned until stage 16 (fig. S5, A to D). However, at stage 17, pericardial cells began to dissociate from the dorsal vessel. These observations suggest that these genes are required to maintain cardiac integrity. The phenotypes of the different mutants were also comparable, except for the two HMGCR EMS mutants or the HMGCR01152/Df(3R)Exel9013 mutant, which was more severe and showed distortion of the shape of the dorsal vessel.

The final C-terminal residues of all G protein γ subunits contain a CAAX motif (5) in which the variable amino acid X determines the type of lipid modification: If X is serine, methionine, alanine, or glutamine, the cysteine is modified by farnesylation, whereas if X is leucine or valine, it is modified by geranylgeranylation (8, 9). Using an in vitro prenylation assay, we found that Drosophila Gγ1 protein, which contains a CAAX motif of Cys-Thr-Val-Leu (CTVL), was modified by geranylgeranylation, but not by farnesylation (Fig. 3C), in agreement with the requirement of GGPPS/qm and βGGT-I for cardiac development.

To determine directly if geranylgeranylation of Gγ1 is essential for heart development, we tested whether wild-type and mutant forms of Gγ1 protein could rescue the cardiac defect of the Gγ1 mutant. Targeted expression of wild-type Gγ1 in the heart was sufficient to rescue the cardiac defects of Gγ1 mutants (Fig. 3D), whereas mutant forms of Gγ1, in which geranylgeranylation was abolished by either a substitution of Ser for Cys67 (Gγ1-C67S) in the CAAX box or a deletion of the CAAX box (Gγ1-ΔCAAX), failed to rescue the cardiac defects in Gγ1 mutants (Fig. 3E and fig. S3E). We conclude that geranylgeranylation of the CAAX motif of Gγ1 is required for its normal activity during Drosophila heart formation.

Lipid modification of the CAAX motif facilitates the association of proteins with membranes (5, 10). To further explore how geranylgeranylation of Gγ1 affects its biological function, we examined the subcellular localization of the Gγ1 protein in Drosophila S2R+ cells. Wild-type Gγ1 protein was always excluded from the nucleus in S2R+ cells, whereas the two mutant forms of Gγ1, which were not geranylgeranylated, were located throughout the cytoplasm and nucleus (Fig. 3F, a to c and a′ to c′). Because Gγ1 is a small protein and can enter the nucleus freely, the specific localization of wild-type Gγ1 protein to the cytoplasm likely reflects its interaction with membranous structures, which requires modification by geranylgeranylation.

In S2R+ cells treated with three HMGCR inhibitors (atorvastatin, mevinolin, and simvastatin), as well as HMGCR double-stranded RNA, the wild-type Gγ1 protein displayed the same abnormal subcellular distribution as the two mutant forms of Gγ1 (Fig. 3F, d, e, d′, and e′, and fig. S3F). These findings suggest that abnormal subcellular localization of Gγ1 accounts for the cardiac defects in the mevalonate pathway mutants and Gγ1 mutants. Gα has also been shown to be required at an earlier stage of heart development for proper alignment of cardioblasts (11), which is distinct from the function of Gγ1 revealed here.

Cardiac defects of HMGCR or Gγ1 mutants could be completely rescued by targeted expression of UAS-HMGCR and UAS-Gγ1 transgenes, respectively, using a Hand-GAL4 driver (Figs. 2C and 3D), which directs expression in both cardioblasts and pericardial cells, or a Mef2-GAL4 driver, which is expressed in cardioblasts but not in pericardial cells (Fig. 4, A and B). In contrast, targeted expression of HMGCR or Gγ1 using Dot-GAL4, which only drives expression in pericardial cells, failed to rescue the cardiac defects in either mutant (Fig. 4, C and D). These results demonstrate that HMGCR and Gγ1 function specifically in cardioblasts to adhere with pericardial cells and exclude the possibility that the bro cardiac phenotype arises secondarily from general metabolic abnormalities.

Fig. 4.

HMGCR and Gγ1 are specifically required in cardioblasts to maintain cardiac integrity. (A and B) Expression of HMGCR or Gγ1 in cardioblasts is sufficient to rescue the bro defect in HMGCR01152 or Gγ1k08017 mutants, respectively. (C and D) Expressing HMGCR (C) or Gγ1 (D) in pericardial cells cannot rescue the bro defect in HMGCR01152 or Gγ1k08017 mutants. (E) A model summarizing the function of the mevanolate pathway, Gγ1, and Sar1 during Drosophila heart formation.

HMGCR and downstream enzymes in the biochemical pathway leading to the synthesis of geranylgeranyl-PP are specifically required in cardioblasts to modify Gγ1 (Fig. 4E). We propose that geranylgeranylation, which is required for the proper intracellular localization of Gγ1, is in turn required for generating a signal for pericardial cells to adhere to cardioblasts throughout heart formation. Indeed, Gβγ has been shown to control Golgi apparatus organization and vesicle formation during exocytosis in mammalian cells (12, 13). The finding that a mutation in Sar1 causes the same cardiac phenotype as the Gγ1 mutation further supports the possibility that this collection of mutations perturbs the secretion of a factor required for maintenance of cardiac integrity. Inhibition of this pathway with statins results in cardiac defects similar to those resulting from mutations in HMGCR and downstream genes required for isoprenoid biosynthesis, which raises the possibility that congenital heart defects reportedly associated with the use of statins (14), which are contraindicated during pregnancy, may reflect perturbation in a similar developmental pathway.

HMGCR has also been shown to be required for recruitment of primordial germ cells (PGCs) to the gonad in Drosophila (4), but the protein target(s) of the mevalonate pathway that mediate this process have not been identified. Perhaps Gγ1 functions in the gonad mesoderm to guide PGC migration. We speculate that lipid modifications mediated by the mevalonate pathway contribute to directed cell migration and subsequent cell-cell adhesion in diverse cell types. Given the conservation of cardiac developmental control mechanisms, it will be of interest to investigate the potential involvement of the mevalonate pathway in mammalian heart development and congenital heart disease.

Supporting Online Material

Materials and Methods

Figs. S1 to S5


References and Notes

View Abstract

Navigate This Article