An ABC Transporter Controls Export of a Drosophila Germ Cell Attractant

See allHide authors and affiliations

Science  13 Feb 2009:
Vol. 323, Issue 5916, pp. 943-946
DOI: 10.1126/science.1166239


Directed cell migration, which is critical for embryonic development, leukocyte trafficking, and cell metastasis, depends on chemoattraction. 3-hydroxy-3-methylglutaryl coenzyme A reductase regulates the production of an attractant for Drosophila germ cells that may itself be geranylated. Chemoattractants are commonly secreted through a classical, signal peptide–dependent pathway, but a geranyl-modified attractant would require an alternative pathway. In budding yeast, pheromones produced by a-cells are farnesylated and secreted in a signal peptide–independent manner, requiring the adenosine triphosphate–binding cassette (ABC) transporter Ste6p. Here we show that Drosophila germ cell migration uses a similar pathway, demonstrating that invertebrate germ cells, like yeast cells, are attracted to lipid-modified peptides. Components of this unconventional export pathway are highly conserved, suggesting that this pathway may control the production of similarly modified chemoattractants in organisms ranging from yeast to humans.

In most organisms, primordial germ cells migrate from their site of origin to the somatic part of the gonad, where they develop into mature eggs and sperm. In Drosophila, germ cells migrate as single cells in a stereotyped manner and are guided by repellent and attractive cues toward the somatic gonad in the mesoderm (1). 3-hydroxyl-3-methyl-glutaryl-CoA reductase (HMG-CoAr or HMGCR) activity controls germ cell attraction to the mesoderm and the recruitment of germ cells to the somatic gonad. In hmgcr mutant embryos, germ cells fail to reach the somatic gonad; moreover, ectopic HMGCR expression is sufficient to attract germ cells to a new location (2). Embryos mutant for several enzymes in the hmgcr pathway, including the β subunit of type I geranygeranyl transferase (βGGTI), are similarly defective in germ cell migration, suggesting that geranylation is critical in attracting germ cells to the mesoderm (3).

Because hmgcr or βggtI mutant embryos show a rather specific germ cell migration defect, we favored the idea that the HMGCR pathway was required to geranylate a critical germ cell attractant rather than regulating a pathway that controlled synthesis or secretion of the attractant (3). This idea posits secretion of a geranyl-modified germ cell attractant, which would preclude secretion by a classic, signal peptide–dependent secretory pathway and require an alternative export mechanism. Such a mechanism has been described in yeast, where adenosine triphosphate (ATP)–binding cassette (ABC) transporters export farnesylated pheromones required for cell mating (4, 5). We therefore asked whether a similar export mechanism exists in Drosophila and is required for germ cell attraction.

ABC transporters are conserved in organisms ranging from bacteria to humans and shuttle hydrophobic lipophilic compounds in an ATP-dependent manner (6, 7). In Saccharomyces cerevisiae and Schizosaccharomyces pombe, the ABCB family members Ste6p and Mam1 play essential roles in the export of farnesyl-modified a-type and M-type mating factors, respectively. The human ABCB family member mdr1 (multidrug resistance) gene is amplified in multidrug-resistant cells, and its homolog mdr3 is a functional homolog of STE6 (8, 9), suggesting a close relationship between the ability of this class of transporters to export drugs and lipid-modified signaling molecules. To determine whether ABCB transporters have a role in exporting the putative Drosophila germ cell attractant, we analyzed expression patterns and germ cell migration in embryos mutant for ABCB transporters (7) (table S1). Among these, only mdr49 showed an expression pattern and mutant phenotype consistent with a role in germ cell migration (Fig. 1A, fig. S1, and table S2). We generated a strong loss-of-function allele, mdr49Δ3.16 (Fig. 1B, fig. S1A, and table S2), and mdr49Δ3.16 mutant embryos showed defects in germ cell migration, like embryos in which the HMGCR pathway was mutant: Germ cells migrated through the posterior midgut but then failed to associate with the somatic gonad (Fig. 1C), which was properly specified (fig. S1E). This migration phenotype was observed only in mdr49 mutants and not in mutants for other ABCB transporters, such as Mdr50, Mdr65, and CG7955 (table S1). To determine whether mdr49 function is required in the mesoderm, we restored mdr49 expression selectively in the mesoderm (fig. S1F) and found that it fully rescued the mdr49 mutant phenotype (Fig. 1D). To test whether Mdr49 acts as an ABCB transporter, we asked whether Ste6p rescued the germ cell migration phenotype observed in mdr49 embryos. Expressing STE6 in the mesoderm rescued the migration defect (Fig. 1D and fig. S1F), suggesting that Mdr49 is functionally equivalent to Ste6p and acts in mesodermal cells to attract germ cells.

Fig. 1.

The ABC transporter mdr49 is required for germ cell migration to the mesoderm. (A) In situ hybridization to wild-type embryos shows mdr49 mRNA expression (purple) specifically in mesodermal tissues. (B) Reverse transcription polymerase chain reaction of the mdr49Δ3.16 allele (lane 3), compared to a revertant line (lane 2) or wild-type embryos (lane 1) (with an rp49 loading control). Embryos in (C) and (D) are stage 14, and germ cells are marked by Vasa (brown). (C) mdr49 phenotype. An mdr49Δ3.16 embryo (bottom) shows lost germ cells. (Right) Quantification of different mdr49 mutant lines is as follows: blue, 0 to 3 germ cells; red, 4 to 6 germ cells; yellow, 7 to 9 germ cells; green, 10 to 12 germ cells; black, >12 germ cells. Genotypes are homozygous or as indicated. (D) Rescue of the mdr49Δ3.16 phenotype by expression of the yeast gene STE or Drosophila mdr49 using twi:: GAL4, a GAL4 transgene that drives expression of UAS target genes in the mesoderm. (Right) Bar 1: twi::Gal4, mdr49Δ3.16/CyO (n = 77 embryos); bar 2: twi::Gal4, mdr49Δ3.16 (n = 194); bar 3: twi::Gal4, mdr49Δ3.16; UASmdr49/+ (n = 123); bar 4: twi::Gal4, mdr49Δ3.16; UASSTE6/+ (n = 92) (**P < 0.0001, *P < 0.0005).

The germ cell migration phenotypes caused by mutations in Mdr49 and in components of the HMGCR pathway, such as geranylgeranyl-diphosphate synthase and βGGTI, are strikingly similar (2, 3). Thus, to determine whether Mdr49 acts as a transporter for a germ cell attractant that is geranylgeranyl-modified by the HMGCR pathway, we tested the genetic epistasis between hmgcr and mdr49. Previous experiments showed that overexpression of hmgcr in the central nervous system (CNS) is sufficient to attract germ cells to this tissue (2). We therefore reasoned that Mdr49 function should be necessary for the export of the ectopically produced attractant and that mutations in mdr49 should suppress the hmgcr misexpression phenotype. Indeed, germ cell migration to the CNS was suppressed by reducing mdr49 copy number by means of either the mdr49 deficiency or P-element mutation. (Fig. 2 and table S3). Mutations in other ABCB transporters, such as mdr50, mdr65, and CG7955, did not significantly suppress the hmgcr overexpression phenotype (table S1), which is consistent with a specific role for Mdr49 as an exporter for an HMGCR-dependent geranyl-modified germ cell attractant.

Fig. 2.

HMGCR-dependent germ cell migration to the CNS is suppressed by reduction of mdr49 activity. All embryos are stage 13. (A) In more than 50% of embryos overexpressing hmgcr in the CNS, >5 germ cells (brown) migrate to the CNS (the area between the horizontal lines). Ectopic migration is suppressed when mdr49 dosage is reduced, either by the deficiency (B) or a P-element mutation (C). (Right) Quantification of germ cells in the CNS. Blue, 0 to 3 germ cells; light blue, 4 to 5 germ cells; magenta, >5 germ cells.

Prenylated proteins require additional modifications to be fully functional: cleavage of the AAX residues (A, aliphatic amino acid; X, any amino acid), prenyl proteolysis, and carboxymethylation (10) (Fig. 3A). We therefore asked whether the enzymes that catalyze these reactions are present in Drosophila and are required for germ cell migration. Prenyl proteolysis is catalyzed by Ste24p (prenyl protease type I) and Rce1p (prenyl protease type II), with the former being essential for a-factor modification (11, 12). Carboxymethylation is achieved exclusively through the activity of Ste14p (13). Orthologs of these enzymes are also found in mammals, where they have similar substrates (table S4). A genomic search in Drosophila revealed predicted orthologs for the following: (i) prenyl protease type I as three proteins encoded by a cluster of genes (CG9000/CG9001/CG9002), which we term Dmel\Ste24a, -b, and -c, respectively; (ii) prenyl protease type II, termed Sras; and (iii) the isoprenylcysteine carboxylmethyltransferase as CG11268, termed Dmel\Ste14 (table S4 and fig. S2).

Fig. 3.

Type I prenyl proteases and isoprenylcysteine carboxymethyltransferases are required for germ cell migration to the mesoderm. (A) A schematic of the prenyl processing pathway analyzed. CAAX (C, Cys) represents the prenylation motif to which a farnesyl (F) or geranylgeranyl (GG) group is added. (B) Embryos carrying a deficiency of Dmel\ste24 show germ cells (brown) lost in the posterior mesoderm. Sibling control (top) and mutant (bottom) embryos are shown at left, and quantification is shown at right (blue, 0 to 3 germ cells; red, 4 to 6 germ cells; yellow, 7 to 9 germ cells; green, 10 to 12 germ cells; black, >12 germ cells). (C) Df(3L)Dmel\ste14 suppresses hmgcr-dependent ectopic germ cell migration. Sibling control (top) and suppressed (bottom) embryos are shown at left and quantification at right (germ cells at ectopic sites). Blue, 0 to 5 germ cells; light blue, 6 to 10 germ cells; magenta, >10 germ cells. All embryos are at stage 14/15.

We next assessed the role of each enzyme in germ cell migration. Because specific mutations were not available for these enzymes, we analyzed deficiencies that deleted each gene [as well as other genes (14)]. Embryos with homozygous mutations in Df(2R)Dmel\ste24, which deletes the three Drosophila ste24 genes, showed a specific germ cell migration phenotype similar to that of mdr49 mutant embryos. On average, six germ cells did not associate with the somatic gonad (Fig. 3B and table S2). Next, we asked whether the single Drosophila isoprenylcysteine carboxylmethyltransferase, Dmel\ste14, is required for germ cell migration. We were unable to analyze the germ cell migration phenotype of embryos homozygous for the deletion Df(3L)ED4486 (Df(3L)Dmel\ste14) because of embryonic patterning defects, which resembled those observed in embryos with mutant Drosophila Rho GTPases and gave additional evidence that Dmel\ste14 encodes a general postprenylation processing enzyme that modifies all prenylated proteins (1517) (fig. S4). Instead of analyzing homozygous mutants, we asked whether reducing Dmel\ste14 gene dosage was able to suppress germ cell mis-migration induced by ectopic expression of hmgcr in the CNS. As observed for the ABC transporter mdr49, reducing Dmel\ste14 gene dosage significantly suppressed HMGCR-dependent ectopic germ cell migration, suggesting a requirement for this enzyme in attractant modification (Fig. 3C and table S5).

Our genetic analyses strongly suggest that a geranylgeranylated germ cell attractant is produced and modified in the somatic gonad and exported via an ABCB transporter. In order to more directly test for the production of a molecule that can act as a diffusible attractant, we devised an in vitro germ cell migration assay. We used germ cells sorted by fluorescence-activated cell sorting (FACS) from embryos 2 to 10 hours old that expressed moesin–green fluorescent protein (GFP) specifically in germ cells (18). Germ cells were placed in the upper well of a transwell chemotaxis chamber and scored for their migration toward medium conditioned with secreted proteins (conditioned medium) from Kc cells, an insect cell line (19), in the lower well (Fig. 4A). Germ cells migrated to the bottom well containing conditioned medium from cells overexpressing hmgcr and mdr49 (Fig. 4B). Fewer cells moved toward unconditioned control medium or medium conditioned from parental Kc cells, which express low levels of hmgcr and mdr49 (19) (fig. S5A). To determine whether migration toward parental Kc cells is dependent on hmgcr and mdr49, we used RNA interference (RNAi) to reduce the expression of HMGCR pathway members in Kc cells. We found that reducing hmgcr, βGGTI, and ABC transporter (mdr49) expression fully blocked germ cell migration toward Kc cell–conditioned medium (Fig. 4C and fig. S5B). These results are consistent with our genetic data and support the notion that the prenylated Drosophila germ cell attractant is active in a secreted form.

Fig. 4.

Germ cells migrate toward a diffusible attractant. (A) Scheme of the transwell migration assay procedure. CM, conditioned medium. (a) The box represents the germ cell population collected. (b and c) The FACS-sorted germ cell population is homogenous and expresses germ cell markers: Germ cell–specific moesin-GFP from the Pnos::egfp-moe::nos 3′ untranslated region transgene is expressed at the membrane [green, (b) and (c)] and the Vasa germ cell marker in the cytoplasm [red, (c)]. (B) Quantification of migration toward serum-free media (control), Kc cells, and Kc cells overexpressing hmgcr and mdr49. *P < 0.001, **P < 0.0001. (C) Quantification of migration toward the attractant in control and RNAi knockdown experiments, ***P < 0.001, **P < 0.005, *P < 0.01; Kc cells plus ds-syntaxin 5 (Kc+ds syx5), P = 0.010. All experiments were repeated at least three times, in triplicate.

To test whether the classic, signal peptide–dependent pathway is also required for secretion of the germ cell attractant, we used RNAi to reduce the levels of syntaxin 5, which encodes a SNARE (SNAP and NSF attachment receptor) protein essential for constitutive secretion (20). Germ cells were similarly attracted to conditioned medium from parental cells and syntaxin 5–deficient Kc cells, indicating that production of the attractant does not depend upon constitutive secretion (Figs. 4C and S5A). Our results demonstrate that the Drosophila germ cell attractant is geranylgeranylated and secreted by mesodermal cells in a signal peptide-independent manner through an ABCB transporter. The modifications, processing, and export pathway used to generate and secrete the germ cell attractant strikingly resemble that of a-factor in mating yeast.

ABC transporters have been studied mostly in the context of cancer drug resistance or their role in toxin protection. Our findings reveal a new function for this conserved, non-conventional secretory pathway in a multicellular organism and suggest that this pathway may be used to export signals required for cell-cell communication in organisms other than Drosophila and yeast (2123). Yeast lacks many of the secreted cell signaling molecules, such as Hedgehog, Wnt, and bone morphogenetic protein, used by multicellular organisms to communicate between cells. The use of a prenylated signal may thus be an ancient mechanism of cell communication. It is striking that this pathway is used in yeast and flies to facilitate the migration and adhesion of germ cells, the essential cells for reproduction.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S6

Tables S1 to S5


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article