Gγ1, a Downstream Target for the hmgcr-Isoprenoid Biosynthetic Pathway, Is Required for Releasing the Hedgehog Ligand and Directing Germ Cell Migration

The isoprenoid biosynthetic pathway leading from the production of mevalonate by HMGCoA reductase (Hmgcr) to the geranylation of the G protein subunit, Gγ1, plays an important role in cardiac development in the fly. Hmgcr has also been implicated in the release of the signaling molecule Hedgehog (Hh) from hh expressing cells and in the production of an attractant that directs primordial germ cells to migrate to the somatic gonadal precursor cells (SGPs). The studies reported here indicate that this same hmgcr→Gγ1 pathway provides a novel post-translational mechanism for modulating the range and activity of the Hh signal produced by hh expressing cells. We show that, like hmgcr, gγ1 and quemao (which encodes the enzyme, geranylgeranyl diphosphate synthetase, that produces the substrate for geranylation of Gγ1) are components of the hh signaling pathway and are required for the efficient release of the Hh ligand from hh expressing cells. We also show that the hmgcr→Gγ1 pathway is linked to production of the germ cell attractant by the SGPs through its ability to enhance the potency of the Hh signal. We show that germ cell migration is disrupted by the loss or gain of gγ1 activity, by trans-heterozygous combinations between gγ1 and either hmgcr or hh mutations, and by ectopic expression of dominant negative Gγ1 proteins that cannot be geranylated.


Introduction
Two distinct cell types, the primordial germ cells and the somatic gonadal precursor cells (SGPs), coalesce to form the Drosophila embryonic gonad (for review, see [1,2]). These cells arise in different regions of the embryo and are specified by completely different mechanisms. The SGPs are derived from the lateral mesoderm in parasegments 10-13 during mid-embryogenesis and are specified by the input from a combination of cell-cell signaling pathways and zygotic patterning genes [3,4]. By contrast, the primordial germ cells, or pole cells, are formed on the outside surface of the embryo at the posterior end during the syncitial blastoderm stage and are specified by determinants localized in the posterior pole plasm during oogenesis [5,6]. In order for pole cells to assemble into a gonad with the SGPs, they must traverse from the posterior end into the middle of the embryo and then subsequently move to the lateral mesodermal cell layer, which contains the newly formed SGPs. This is a multistep process that begins at gastrulation when the pole cells are carried into the interior of the embryo by the midgut invagination [1,2]. They then pass through the midgut epithelium, and move along the surface of the midgut until they split into two groups. The germ cells in each group migrate laterally and this brings them into contact with the gonadal mesoderm on either side of the embryo. The germ cells align themselves in a row with the SGPs in parasegments 10-13 and these juxtaposed cells coalesce into the embryonic gonad.
Analysis of the different migration steps has suggested that a combination of repulsive and attractive cues guide germ cell migration through the midgut and toward the somatic gonadal mesoderm. Repulsive clues, whose production depends upon Wunen and Wunen2, are thought to hasten the movement of the germ cells away from the midgut epithelium [7,8]. Once the germ cells exit the midgut and migrate along its surface, attractive cues produced by the SGPs are thought to entice the germ cells towards that lateral mesoderm and promote their subsequent association with the SGPs.
One of the first genes implicated in the production of the germ cell attractant by the SGPs was hmgcr [9]. hmgcr is initially expressed broadly in the embryonic mesoderm; however, by the time germ cells commence their migration into the mesoderm, hmgcr expression is largely restricted to the SGPs [9]. In hmgcr mutants germ cells fail to migrate towards the SGPs and instead either remain associated with the midgut or scatter through the mesoderm. Conversely, ectoptically expressed hmgcr can induce germ cells to migrate towards tissues expressing the Hmgcr protein. Another gene that functions to induce migration towards the SGPs encodes the signaling molecule hedgehog (hh) [10]. Both ectopic expression of Hh and mutations that compromise the production or transmission of the Hh ligand by the SGPs induce mismigration. Since Hh functions as a morphogen in other contexts, one explanation for its effects on germ cell migration is that it acts indirectly by inducing cells to assume a SGP identity so that they can produce the actual attractant. However, a number of findings argue that Hh acts directly as an attractant. For one, the two known receptors of the Hh signal, Patched (Ptc) and Smoothened (Smo) are required in the germ cells for their proper migration. In the absence of the Hh ligand, the transmembrane receptor Ptc inhibits the 7-pass transmembrane protein Smo from mediating signal transduction [11][12][13]. When Hh binds to the Ptc receptor, the physical association between these two proteins is thought to relieve the negative influence of Ptc resulting in the relocalization of Smo to the cell membranes, and this in turn activates the signal transduction cascade downstream of the Hh signal. Consistent with their reciprocal functions in hh signaling, germ cells compromised for ptc or smo activity behave differently. For ptc the germ cells clump prematurely near the midgut as if they had already received the full Hh signal. For smo the germ cells behave as if they are 'signal-blind' and scatter randomly in the posterior of the embryo [10]. A second line of evidence supporting a direct role for Hh as the germ cell attractant comes from the discovery that hgmcr is required for the release of the Hh ligand by hh expressing cells. In embryos compromised for hmgcr activity [14], Hh is inappropriately retained in the hh expressing cells. Conversely, the range and strength of the Hh signal can be substantially enhanced by ectopic expression of hmgcr. Critically, ectopic hmgcr only had an effect on hh signaling when it was expressed in cells that normally produce the Hh ligand, while there was no effect when hmgcr is ectopically expressed in cells that normally receive the Hh ligand.
One important issue left unresolved by these studies is how hmgcr promotes the release of the Hh ligand by hh expressing cells. The hmgcr gene encodes HMGCoA reductase which is responsible for the conversion of 3-hydroxy-3-methylglutaryl coenzyme A to mevalonate. In mammals, mevalonate is a precursor for cholesterol which is used in the modification of the Hh protein.
However, providing precursors for cholesterol biosynthesis is not a likely function for hmgcr in the hh signaling pathway as the genes encoding the enzymes required to synthesize cholesterol from appear to be absent in flies [15]. Mevalonate is also a precursor for many different compounds including carotenoids, isoprenoids, ubiquitones and vitamins A and E [16]. Recent studies by Santos and Lehmann [15] on the role of hmgcr in germ cell migration have implicated the isoprenoid branch of the mevalonate precursor pathway. The isoprenoids farnsyl-pyrophosphate (FPP) and geranylgeranyl-pyrophosphate (GGPP) are used in the posttranslational modification of proteins and are covalently attached to the C terminus of target proteins by farnysyl transferase and type I or type II geranylgeranyl transferases respectively. Santos and Lehmann showed that mutations in farnesyl-diphosphate synthetase (fpps) (which synthesizes FPP), geranylgeranyl diphosphate synthetase (qm) (which in turn converts FPP to GGPP), and geranylgeranyl transferase type I (b-ggt1) disrupt germ cell migration. They also found that germ cell migration is perturbed when fpps and qm are ectopically expressed. Though the effects were much less dramatic than observed for ectopic hmgcr, this is not altogether unexpected since these two genes differ from hmgcr in that they are widely expressed in mid-to-late embryogenesis.
While these findings indicate that the pathway leading from hmgcr to GGPP is important in germ cell migration because some critical target protein requires geranylation, the identity of this protein and the nature of its function in the production of the germ cell attractant remain to be established. Additionally, Santos and Lehmann [15] did not test whether hh signaling also depends upon this same isoprenoid biosynthetic pathway. Thus, the possibility remains open, especially if there is another germ cell attractant besides Hh, that hmgcr has some other function in hh signaling beside the production of isoprenoids. A possible answer to these questions comes from recent studies on cardiac development in flies. Hmgcr and downstream enzymes in the mevalonate pathway are required in cardioblasts to ensure their proper adhesion to the neighboring pericardial cells. Yi et al. [17] found that the endpoint for the isoprenoid branch of the hmgcr mevalonate pathway in heart development is the geranylgeranylation of the heterotrimeric G protein c subunit 1 (Gc1) [17,18]. The C-terminus of the Drosophila Gc1 protein has the isoprenylation CAAX motif sequence, Cys-Thr-Val-Leu. The leucine residue at the terminal position (X) specifies lipid modification by geranylgeranylation. Gc1 requires this modification for membrane association and is inactive when geranylation is blocked. Significantly, gc1 would be a quite plausible downstream target for hmgcr activity in the hh signaling pathway (and thus in the production of the germ cell attractant). Though heterotrimeric G proteins are normally thought to mediate the transduction of extracellular signals by G-protein coupled receptors, recent studies indicate that these G protein complexes have other intercellular functions. In particular, the Gc1:Gb heterodimer together with Ga have been implicated in the transport of cargo from the trans-Golgi network (TGN) to the basolateral plasma membrane [19][20][21]. The involvement of machinery targeting proteins to the basolateral membrane from the TGN would make sense in the context of hh signaling as autoprocessed and fully modified Hh protein is found to preferentially accumulate in a punctate pattern along the basolateral membranes of Hh expressing cells in the embryonic ectoderm [22][23][24]. This protein is then released from the cell, through a Dispatched (Disp) dependent mechanism that is thought to involve translocation of the Hh puncta from their docking sites along the basolateral membranes to the apical membrane [25,26]. In the studies reported here we have asked whether the hmgcrRGc1 pathway is important for hh signaling and whether gc1 is required for proper germ cell migration as is the case for hmgcr.

gc1 Mutations Suppress the Gain-of-Function Wing Phenotypes of hh Mrt
To test whether gc1 is a component of the hh signaling pathway, we took advantage of the hh Moonrat (hh Mrt ) mutation [27]. hh Mrt is a

Author Summary
Previous studies have shown that HMGCoA reductase (Hmgcr) is required for the production of a germ cell attractant by the somatic gonadal precursor cells (SGPs) and for the release of the Hedgehog (Hh) ligand by hh expressing cells. However, it was not clear what role mevalonate, the biosynthetic product of Hmgcr, played in either of these processes or whether the hmgcr-dependent germ cell attractant corresponds to the Hh ligand (which is known to be expressed by the SGPs). We show here that the downstream target for Hmgcr both in generating the germ cell attractant and in releasing the Hh ligand is the G protein, Gc1. Gc1 must be geranylated in order to function, and the substrate for this posttranslational modification, geranylgeranyl-pyrophosphate, is one of the biosynthetic products of mevalonate. In addition to demonstrating a critical role for Gc1 (as well as the hmgcr isoprenoid biosynthetic pathway) in releasing Hh from hh expressing cells, our findings provide additional evidence that Hh protein produced by the SGPs is an hmgcrdependent germ cell attractant.
dominant gain-of-function hh allele that disrupts patterning of the wing as a heterozygote and is lethal as a homozygote. In wild type wing discs, hh expression is confined to the posterior compartment and it orchestrates wing development by signaling to cells in the anterior compartment along the compartment boundary to upregulate the expression target genes such as ptc and decapentaplegic (dpp). In hh Mrt /+ animals, in addition to being expressed normally in the posterior compartment, hh is ectopically activated in the anterior compartment of the wing disc. As a result dpp is expressed in a pattern that leads to overgrowth of the anterior tissues and the partial duplication of distal wing structures. The anterior-toposterior transformations induced by the hh Mrt allele can be dominantly suppressed by mutations in hh signaling pathway genes like disp and hmgcr that are required to promote hh signaling in the sending cells. The gain-of-function wing phenotype can also be suppressed by mutations in genes like toutvelu (ttv) that are required to promote hh signaling in the receiving cell (unpublished data).
If gc1 functions as the downstream target for hmgcr in the hh signaling pathway, then mutations in gc1 would be expected to dominantly suppress the hh Mrt wing defects. To test for suppression we used two different gc1 mutants. The first, gc1 N159 , is an EMS induced mutation [28]. The gc1 open reading frame encodes a protein of 70 amino acids and this mutation inserts a stop codon at amino acid 59. The second, gc1 k0817 , has a P-element insertion in the splice donor of the first gc1 exon and produces aberrant transcripts. To assess the effects of these gc1 mutations, the Mrt wing blades were assigned to 5 different classes based on the severity of the wing defects, with I being wild type (not shown), and V being the most severely deformed wing (not shown, for details see 27). Under the conditions of this experiment about 70% of the hh Mrt /+ flies were abnormal (see the class III wing in panel A of Figure 1). By contrast when the hh Mrt /+ flies were heterozygous for gc1 N159 quite strong suppression was observed and more than 80% of the wings belonged to class I (panel B). The suppression of the Mrt gain-of-function phenotype does not appear to be due to some non-specific background effect as the wing defects could also be dominantly suppressed by gc1 k0817 (data not shown). Thus like hmgcr and other factors that function to promote hh signaling, gc1 shows genetic interactions with the hh Mrt . Moreover, the extent of suppression is similar to that observed previously with hmgcr [14].
gc1 Is Required To Maintain wg and en Expression The dominant suppression of the Mrt wing phenotypes suggests that like hmgcr, gc1 functions in hh signaling. To test this possibility further we examined wingless (wg) expression during embryogenesis. In wild type embryos, wg stripes are activated by the pair-rule genes at the onset of gastrulation. Once the pair-rule gene products decay later in embryogenesis the maintenance of the wg Figure 1. gc1 can dominantly suppress the wing abnormalities including pattern duplication induced by hh MRT . Panel A shows an example of Class III type of wing defects induced by ectopic expression of Hh in the anterior compartment in hh MRT animals. Panel B shows a wing from an animal of the genotype hh MRT /gc1. Almost complete suppression (Classified as Class I or II) of the wing phenotype can be seen. Panel C shows a graphic representation of the suppression of the wing defects. Roughly 250 single wing blades of the indicated genotypes were analyzed and classified into 5 different categories depending on the severity of the phenotype as previously described in Felsenfeld and Kennison, [27]. All the experimental as well as control crosses testing hh MRT suppression were carried out at 18uC. doi:10.1371/journal.pgen.1000333.g001 expression depends upon hh signaling by the cells immediately posterior to the wg stripe and in hh mutants wg expression begins disappearing by stage 10/11 of embryogenesis. Maintenance of the wg stripes also requires hmgcr activity and in hmgcr mutant embryos the stripes begin to fade around stage 11. However, unlike hh mutants, residual wg expression can still be detected in older hmgcr mutant embryos. Since maternal and zygotic hmgcr activity cannot be completely eliminated, this difference likely reflects (at least in part) the presence of residual Hmgcr in the hmgcr mutant embryos.
If gc1 functions downstream of hmgcr in the hh signaling pathway, then defects in wg expression should also be evident in gc1 mutant embryos. To determine if this is the case we compared wg expression in gc1 2 embryos with their heterozygous gc1 2 /+ sibs. We found that wg expression in the homozygous mutant embryos is initially like wild type (or gc1 2 /+); however as shown in Figure 2B for gc1 N159 , the accumulation of Wg protein begins to decrease around stage 11-12 (compare the gc1 N159 homozygote in panel B with the gc1 N159 /+ sib control in panel A). Similar results were obtained for the gc1 k0817 (compare panel C and D in Figure 1). The extent of reduction in Wg protein in the two gc1 mutants is not as severe as that seen in embryos compromised hh; however, as noted above this was also observed for hmgcr and likely reflects the perdurance of the maternally derived Gc1.
Another gene whose expression in mid-embryogenesis depends upon hh signaling is engrailed (en). en is part of an autoregulatory loop that is established between the neighboring hh and wg expressing cells. en is transcribed in the hh expressing cells in response to the Wg ligand. When wg signaling is disrupted because of a reduction in hh signaling, en transcription is in turn downregulated. As would be expected from the effects of gc1 mutations on wg expression, we find that the accumulation of En protein is reduced in embryos homozygous mutant for both of the gc1 alleles compared to their wild type (gc1 2 /+) sibs (see Figure S1).

Smo Protein Is Mislocalized in gc1 Embryos
While the effects of gc1 mutations on wg and en expression would be consistent with a role in hh signaling, it is also possible that gc1 activity is required at some other point in the hh-wg autoregulatory loop, for example, in the expression of the Wg or En proteins. For this reason we next examined the effects of gc1 on the Smo receptor which is a more direct target for the Hh ligand in the receiving cells. Upon reception of the Hh signal the Smo receptor is relocalized from intracellular membrane vesicles to the cell surface [29,30]. When hh signaling is compromised, this relocalization does not occur, and the Smo protein remains predominantly cytoplasmic in the receiving cells. Since Smo is not properly relocalized in hmgcr mutant embryos, a similar defect would be expected in gc1 mutants if gc1 functions downstream of hmgcr in the hh signaling pathway.  . Reduced wingless expression in embryos compromised for gc1. Panels A and B: Embryos from gc1 N159 / Cy0, en:LacZ stock were collected and fixed using standard procedures. Embryos were genotyped by probing with b-galactosidase antibodies (imaged in red: not shown), while Wg accumulation was visualized by probing with Wg (imaged in green) antibodies. Embryos carrying en:LacZ express bgalactosidase whereas homozygous mutant embryos do not. The embryo in panel A was positive for b-galactosidase (not shown) and has at least one wild type copy of gc1. Note the high level of Wg accumulation in the stripes. The embryo in panel B b-galactosidase negative, and is homozygous for the gc1 N159 mutation. Note the lower level of Wg expression. Panels C and D: Embryos from a gc1 k0817 /Cyo, en:LacZ stock were collected and fixed using standard procedures. As in panels A and B embryos were genotyped by probing with bgalactosidase antibodies (imaged in red: not shown), while Wg accumulation was visualized by probing with Wg (imaged in green) antibodies. The embryo in C is positive for b-galactosidase, while the embryo in D is not. Note the difference in Wg accumulation in the blowup of three Wg stripes. doi:10.1371/journal.pgen.1000333.g002 embryos (see panel C) it is much less distinct than in the heteterozygous sibs (panel A). Panel D shows that Smo remains largely cytoplasmic in most of the cells in each segment and is not tightly localized at the plasma membrane as it is in wild type or gc1 2 /+ heterozygous embryos.

Embryos Compromised for gc1 Display an Altered Hh Protein Distribution
The finding that Smo protein does not properly relocalize in gc1 mutant embryos would be consistent with the idea that Gc1 acts downstream of hmgcr to promote the efficient release and/or transport of Hh protein. To test this hypothesis further, we compared the pattern of Hh accumulation in gc1 mutant embryos with their heterozygous sibs. The distribution of Hh protein in gc1 2 /+ embryos resembles that seen in wild type [22][23][24][25][26]. Hh is expressed in each parasegment in a two cell wide stripe, and most of the protein in these Hh expressing cells is distributed in the cell membrane in a fine grain or punctate pattern (see arrowheads in panel A of Figure 4). Emanating in both directions from the two cell wide stripe is an Hh protein gradient that appears to extend through much of the parasegment. In this gradient the highest levels of Hh protein are observed associated with cells adjacent to the two Hh expressing cells, while lower levels of protein are found in more distant cells. The distribution of Hh protein in gc1 mutant embryos (panels B and C) resembles that seen in hmgcr mutant embryos [14]. First, in spite of the fact that the overall level of Hh expression is expected to be reduced in these embryos because of the disruption in the wg-hh positive autoregulatory loop (see above), the relative amount of Hh in cells in the hh stripes appears higher than in wild type embryos, while there is a concomitant reduction in the amount of Hh in the gradient that extends through the interstripe region (compare panels A with B & C). Second, the normal grainy or punctate pattern of Hh protein localized around the basolateral membrane of the hh expressing cells that is seen in wild type embryos (see arrows in panel A and in the enlargement in panel D) is largely lost. Instead, Hh accumulates in larger ''clumps'' or aggregates (see arrows in panels B and C and in the enlargement in panels E and F) that in many instances seem to be displaced from the cell membranes (see top arrows in panel E and F).
qm Is Required To Promote the Release/Transmission of the Hh Ligand The results described in the previous sections demonstrate that like hmgcr, gc1 is required for the efficient release/transmission of the Hh ligand by hh producing cells. Since the role of the isoprenoid branch of the hmgcr mevalonate pathway in heart development is the geranylgeranylation of Gc1, a plausible idea is that the function of hmgcr in hh signaling is to provide substrates for the modification of the Gc1 protein. If this model is correct, then gene products that are downstream of hmgcr in the geranylgeranylation pathway should also be required for the release/ transmission of the Hh ligand. To test this prediction we examined the distribution of Hh protein in qm mutant embryos. As described above, qm encodes geranylgeranyl diphosphate synthetase and this enzyme produces the substrate, GGPP, that is used by the geranyl transferase to modify Gc1. Figure 5 shows the distribution of Hh in a homozygous qm mutant embryo (panel B) and its heterozygous qm 2 /+sibs (panel A). As observed for both hmgcr [14] and gc1 (see above), the Hh ligand is inappropriately retained in the hh producing cells in the qm mutant embryos (compare Hh distribution in panels A and B). Like gc1 the characteristic punctate distribution of Hh protein around the membranes of hh expressing cells (arrowheads in pane A) is reduced or lost and instead Hh accumulates in clumps or large aggregates (arrows in panel B). It should also be noted that this particular qm mutation appears to cause a more pronounced defect in the release/ transmission of the Hh ligand than is observed for gc1 (compare Figures 4 and 5), while the defects in Hh distribution evident in hmgcr mutant embryos [14] are roughly intermediate between that in qm and gc1.

Embryos Compromised for gc1 Activity Display Germ Cell Migration Defects
The findings describe above indicate that gc1 represents an endpoint for the isoprenoid branch of the hmgcr mevalonate pathway in the hh signaling pathway, and that like hmgcr, gc1 is required for the efficient release/transmission of the Hh ligand. Since several components of this hmgcr mevalonateRisoprenoid pathway have also been implicated in the production or transmission of the germ cell attractant by the SGPs [15], we tested whether mutations in gc1 have any effects on germ cell migration. Embryos collected from gc1 k0817 and gc1 N159 stocks Embryos from gc1 N159 / Cy0; en:LacZ stock were collected, fixed using standard protocol and were subsequently identified by staining simultaneously with the b-galactosidase and anti-Hh antibodies. Hh specific signal was imaged with the secondary antibodies coupled to Alexa 546 (Red). The figure shows stage 10-11 wild type (panel A) and gc1 N159 (panels B and C) embryos. In wild type embryos, two rows of cells per segment express Hh protein. In these cells Hh protein is distributed around the membrane in a grainy or punctate pattern (see arrowheads). The Hh ligand is released and it spreads in a graded fashion through the parasegment. In the gc1 N159 embryos (panels B and C) Hh is not properly released from hh expressing cells and high levels of Hh protein accumulate in these cells. The grainy pattern of Hh protein around the cell membrane is less evident and instead Hh protein accumulates in larger clumps or aggregates that appear to be distributed in the cytoplasm rather than just at the membrane (see arrows). Panels D, E and F are enlargements of the Hh stripe in the embryos shown in panels A, B and C respectively. Arrows in panel D show the grainy punctate pattern of Hh protein in wild type embryos (the LPS), while the arrows in panels E and F show the larger clumps or aggregates of Hh protein that accumulate in cells of mutant embryos. doi:10.1371/journal.pgen.1000333.g004 carrying an en:LacZ marked 2 nd chromosome balancer were stained with b-galactosidase antibodies to identify the homozygous mutant embryos and Vasa antibodies to visualize the germ cells.
In wild type embryos (or in gc1 2 /Cyo en:LacZ embryos) germ cells associate with the SGPs in parasegments 10-13 at stages 12-13 (see stage 13 WT embryo in Figure 6A) and the two cell types coalesce into the embryonic gonad at stages 14-15 (see stage 15 WT embryo Figure 6B). Although germ cells that fail to coalesce into the embryonic gonad are sometimes seen in wild type embryos, the number of lost germ cells is generally rather low. The germ cells in gc1 2 embryos appear to have no difficulty in exiting the midgut pocket at stage 9-10, while movement along the surface of the midgut also appears to be comparatively normal. However, as illustrated in panel C of Figure 6, defects in migration are clearly evident by stage 13. In this embryo, several of the germ cells are not properly aligned with the SGPs in PS10-13 (compare with wild type in panel A). This problem persists and in stage 15 gc1 2 embryos germ cells that haven't coalesced into the embryonic gonad can be seen scattered in the posterior (see panel D). Quantitation indicates that in wild type the vast majority (90%) of the stage 15 embryos (n = 20) have few if any (0 to 2) scattered germ cells. In contrast, about 36% of the gc1 1N159 embryos (n = 20) have 3 to 4 scattered germ cells while nearly 40% have 5 or more scattered germ cells. Similar results were obtained for the second gc1 allele, gc1 k0817 (see Figure S2). Though the severity of the germ cell migration defects in the two gc1 2 mutants is similar to that reported for embryos zygotically compromised for either fpps or qm, it is not as strong as that observed for hmgcr mutant embryos [9], or for embryos that lack both zygotic and maternal (m 2 z 2 ) fpps [15]. While compromising both maternal and zygotic gc1 would likely increase the severity of the germ cell migration defects as seen for fpps, the very severe patterning abnormalities observed in m 2 z 2 embryos [28] would make effects on germ cell migration impossible to interpret. These findings indicate that gc1 is involved in germ cell migration just like the three enzymes, fpps, qm and b-GGT1 that are downstream of hmgcr in the geranylgeranylation branch of the mevalonate pathway.

Ectopic Expression of Gc1 Induces Germ Cell Migration Defects
Ectopic expression of hmgcr, qm or fpps can induce the production of the germ cell attractant in inappropriate tissues. This ectopic source of attractant competes with the attractant produced by the SGPs and confuses the germ cells, disrupting their migration towards the SGPs [9,15]. If gc1 is functioning in the same pathway as these three enzymes, then it should also be possible to confuse germ cells by ectopically expressing gc1. To test this hypothesis, females carrying the CNS driver elav-GAL4 were mated to males carrying a UAS transgene that drives a flag-tagged Gc1 protein and the resulting elav-GAL4/UAS-flag gc1 embryos were stained with Vasa antibodies to mark the germ cells. Figure 6E and 6F show that misexpression of Gc1 in the central nervous system leads to a weak but reproducible germ cell migration defect. In wild type 90% of the stage 14-15 embryos have 0-2 scattered germ cells, while about 10% have 3 or more scattered or lost germ cells. In contrast in elav-GAL4/UAS-flag gc1 embryos (n = 51), more than 40% of the embryos have 3 or more lost germ cells. The effects of elav driven Gc1 expression are less than that reported for elav driven Hgmcr expresson (100% have 3 or more scattered germ cells) but equivalent to that observed for elav dependent misexpression of either Fpps or Qm (approximately 40% with 3 or more scattered germ cells: see 15).

Ectopic Expression of gc1 in hh Producing Cells Induces Aberrant Germ Cell Migration
In previous studies we found that expression of Hmgcr protein in hh producing cells was much more effective in inducing aberrant germ cell migration than when it was expressed in hh receiving cells. If gc1 functions downstream of hmgcr in the production of the germ cell attractant, then ectopic Gc1 should also have a more pronounced effect on germ cell migration when it is expressed in hh producing cells then when it is expressed in hh receiving cells.
The experiment in Figure 7A shows that this prediction holds. There is little or no effect on germ cell migration when Gc1 expression is induced in hh receiving cells by a ptc-GAL4 driver. In this case, less than 10% of the embryos have 3 or more lost germ cells, which is comparable to that seen in wild type embryos (see bar graph in Figure 7A). In contrast, nearly one half of the embryos have 3 or more lost germ cells when Gc1 is expressed in hh producing cells under the control of hh-GAL4 driver ( Figure 7A). While this result indicates that like Hmgcr, Gc1 must be misexpressed in hh producing cells in order to induce aberrant germ cell migration, it is important to note that the effects of ectopic Gc1 are less severe than that produced when Hmgcr expression is driven by the same hh-GAL4 driver [10]. Consistent with this difference, we do not observe any obvious alteration in the parasegmental distribution of Hh protein in UAS-gc1/hh-GAL4 embryos (not shown). By contrast, substantially more Hh protein is found in the interstripe regions when Hmgcr expression is driven by hh-GAL4 in hh producing cells [10].

Synergistic Genetic Interactions between gc1 and Either hmgcr or hh Disrupt Germ Cell Migration
While there are few if any defects in germ cell migration in hmgcr 2 /+ embryos, synergistic interactions are observed when hmgcr 2 is combined with mutations in two components of the hh signaling pathway hh and disp [10]. The perturbations in germ cell migration observed in the trans-heterozygotes taken together with Embryos from qm 2 / Cy0, ftz-LacZ stock were collected, fixed using standard protocol and were subsequently identified by staining simultaneously with the b-galactosidase (not shown) and Hh antibodies (imaged in red). Panel A: Wild type control (qm 2 /Cy0, ftz-LacZ). Panel B: qm 2 embryo. In the control embryos, Hh protein synthesized in two rows of cells per parasegment is released and spreads through the segment. Within the Hh expressing cells, Hh protein is localized around the membrane in a grainy or punctate pattern (see arrowheads in Panel A). In qm the release/transmission of the Hh protein is abnormal. The level of Hh in the interstripe region is considerably diminished suggesting that like gc1, the qm gene is required for the efficient release and/or transport of the Hh protein. Consistent with this suggestion, Hh protein appears to accumulate in the Hh expressing cells. Like gc1, the distribution of Hh in the expressing cells is abnormal. Instead of the characteristic grainy or punctate pattern of Hh protein localized around the cell membranes, Hh accumulates in clumps or aggregates (see arrows). doi:10.1371/journal.pgen.1000333.g005 the effects of hmgcr on the release/transmission of the Hh ligand from hh producing cells lent support to the hypothesis that the primary function of hmgcr in the production of the germ cell attractant by the SGPs is to potentiate the Hh signal emanating from these cells. Since the results presented above suggest that gc1 also functions in the release/transmission of the Hh ligand, we wondered whether equivalent synergistic genetic interactions would also be observed for gc1.
We first tested for interactions between gc1 and hmgcr. Like hmgcr, there are at most only very modest defects in germ cell migration in gc1 N159 /+ embryos. However, more than 60% of the trans-heterozygous embryos have 7 or more lost germ cells (see Figure 8A). Next we tested for genetic interactions between gc1 N159 /+ and hh. As shown in Figure 8B, the minor germ cell migration defects observed in hh 2 /+ embryos are greatly enhanced when the hh mutation is combined with gc1 N159 . These results support the idea that gc1 could function in the germ cell migration pathway by facilitating the release/transmission of the Hh ligand.

Misexpression of the gc1 CAAX Deletion Mutant Induces Germ Cell Migration Defects
The results described in the previous sections suggest that the hmgcr mevalonate pathway is required in germ cell migration because gc1 must be geranylgeranylated in order for it to potentiate Hh signaling by the SGPs. To test this idea further, we examined the effects of misexpressing either the wild type Gc1 or Gc1 proteins that have mutations in the C-terminal CTVL geranylgeranylation motif in the mesoderm using a twist-GAL4 driver. We anticipated that misexpressing wild type Gc1 using the twist driver would induce aberrant germ cell migration because it would inappropriately potentiate signaling by hh expressing cells elsewhere in the mesoderm such as the fat body precursor cells (FBP). The hh signal emanating from these cells would compete with the signal from the SGPs, and this would confuse the migrating germ cells. The results shown in Figure 7B indicate that this expectation is met. While there are only a few wild type embryos in this experiment which have more than 2 lost germ cells, more than half of the twist-GAL4:UAS-gc1 embyros have at least 3 lost or mismigrated germ cells.
We also anticipated that misexpressing Gc1 proteins that have mutations in the C-terminal geranylgeranylation motif would induce germ cell migration defects as well. Gc1 forms a heterodimer with a second G protein Gb and together these two proteins interact with a third G protein, Ga to form a heterotrimeric complex. In order to form a functional complex with Ga and also interact with other factors and effectors, the Gb:Gc1 heterodimer must be anchored to the membrane and this is thought to be dependent upon geranylgeranylation of the Gc1 protein [31,32]. We reasoned that Gc1 mutant proteins that cannot be geranylated would likely behave as dominant negatives because they would compete with the endogenous Gc1 protein for complex formation with Gb, and thus reduce the effective concentration of functional membrane bound Gc1:Gb heterodimers. This idea is support by studies on the Drosophila eye specific Gc protein Gce. The C-terminal sequence Gce is C-V-I-M which corresponds to the signal for farnesylation rather than geranylgeranylation. Mutations in Gce that eliminate farnesylation have no effect on the formation of Gce:Gbe heterodimers; however, these heterodimers do not interact with the membrane and are nonfunctional [32]. When the mutant Gce protein is overexpressed it competes with the endogenous Gce protein for heterodimer formation with Gbe, reducing the amount of functional membrane associated Gce:Gbe heterodimers and disrupting signal transduction. If gernaylation defective Gc1 proteins also behave like dominant negatives, they would be expected to interfere with the efficient release of the germ cell attractant by the SGPs when they are ectopically expressed in mesodermal cells and this should perturb germ cell migration. We tested two different Gc1 mutant proteins, one in which the C-terminal CTVL motif is deleted (Gc1-DCAAX) and the other in which the geranylated Cys residue is replaced by Ser (Gc1-C67S) [16]. As shown in Figure 7B and Figure S3, ectopic expression of the Gc1-DCAAX protein disrupts  Embryos between stages 12-15 of the indicated genotype were stained with anti-Vasa and b-galactosidase antibody and staining was visualized with standard immunohistochemcial techniques. Total number of germ cells that failed to associate with SGPs and remained scattered were counted per embryo. 25 embryos of each genotype were analyzed. Top panel shows that when embryos are heterozygous either for gc1 or hmgcr, more than 80% of the embryos display 0-2 lost germ cells (blue and red bars respectively). But when embryos are simultaneously compromised for both gc1 and hmgcr, more than 60% of the embryos have 7 or more lost germ cells (yellow bars). Bottom panel shows similar synergistic interaction between gc1 or hmgcr. Although the enhancement in germ cell migration defects is less severe compared to that seen with hmgcr (30% of the total number of embryos of the genotype gc1/+; hh/+ show more than 7 lost germ cells), the germ cell migration defects in embryos simultaneously compromised for gc1 and hh are clearly more severe as opposed to either gc1/+ or hh/+ embryos. doi:10.1371/journal.pgen.1000333.g008 Females carrying two copies of UAS-gc1 were mated independently either with the ptc-GAL4/ptc-GAL4 males (panels on left) or hh-GAL4/TM6 Ubx-LacZ males (panels on right). Embryos (10-14 hr old) were collected, fixed and then probed with b-galactosidase and Vasa antibodies. In the case of hh-GAL4 driver, embryos of the correct genotype were identified by the absence of b-galactosidase. The staining was visualized using standard immunohistochemical techniques. As can be seen by the comparison of the panels, gc1 is able to induce germ cell migration defects only when it is overexpressed using hh-GAL4 whereas gc1 overexpression using ptc-GAL4 leads to essentially wild type germ cell migration. As shown in the bardiagram in the lower half of A, more than 90% of the ptc-GAL4/UAS-gc1 embryos (represented with red colored bars) display less than 2 lost germ cells whereas close to 30% of the hh-GAL4/UAS-gc1 embryos (represented with blue colored bars) have more than 5 lost germ cells. hh-GAL4/UAS-gc1 (n = 74), ptc-GAL4/UAS-gc1 (n = 53). (B) Ectopic expression of gc1 and gc1-DCAAX in the mesoderm induces germ cell migration defects. In this experiment embryos produced by females carrying the UAS:gc1 or UAS gc1-DCAAX mated to twist GAL4 males were stained with Vasa antibody and the number of lost or scattered germ cells in each embryo was counted. As shown in the bar graph, ectopic expression of either Gc1 or Gc1-DCAAX using the twist driver induced germ cell migration defects. As explained in the text, these defects likely arise for different reasons. Ore R (n = 20), UAS-gc1 (N = 69), UAS-gc1-DCAAX (n = 53). doi:10.1371/journal.pgen.1000333.g007 germ cell migration and about 75% of the twist-GAL4/UAS-gc1-DCAAX transgene embryos have 3 or more lost germ cells. (Note: Figure S3 shows that ectopic expression of Gc1-DCAAX in germ cells also disrupts their migration.) With the caveat that there may be differences in expression levels of the UAS transgene, it would appear that the germ cell migration defects induced by the geranylation defective Gc1-DCAAX protein are somewhat more pronounced than those observed with wild type Gc1. Consistent with this possibility, the Gc1-C67S mutant protein also induces more extensive germ cell migration defects than wild type (not shown).

Gc1 Is Required for Hh Signaling
Hh functions as an instructive cue in many different biological contexts. The signaling molecule is secreted from hh expressing cells and it induces morphogenesis in a concentration dependent fashion in neighboring cells by regulating the transcription of downstream target genes. Several mechanisms control the range and inductive activity of the Hh protein. These include the autoprocessing and lipidation [33][34][35]. Hh has two different lipid modifications that are important for the proper functioning of the Hh ligand. One is the palmitoylation of the N terminus which seems to be critical for signaling activity, while the other is the addition of cholesterol to the C terminus. The C-terminal cholesterol moiety is thought to be important for the dimerization of the Hh protein and for its assembly into LPSs (Large Punctate Particles) prior to secretion [36][37][38][39]. The LPSs appear to be lipid vesicles or micelles and they are thought to provide a hydrophobic environment for the lipid modified Hh which facilitates its movement through the extracellular matrix after it is secreted. The release and subsequent transport of the Hh ligand also requires specialized proteins that function in either Hh producing cells or in cells/compartments that are destined to receive the Hh ligand. The transporter class protein, Disp, and a secreted protein Shifted (Shf) are required in hh expressing cells for the efficient release and transmission of the Hh ligand [25,[40][41][42]. In shf mutants, the basolateral accumulation of Hh protein in the wing disc is disrupted, while apical accumulation appears to be normal. The subsequent transport of the Hh ligand to the receiving cells depends upon the glypicans Dally-like (Dlp) and Dally, which are components of the extracellular matrix, and enzymes that are needed for glycosaminoglycan biosynthesis namely Sulfateless and Tout-velu [43][44][45]. The glycosaminoglycan is thought to promote long range signaling by Hh and other signaling molecules such as Wg by passing the ligand from one cell to its neighbor instead of presenting the ligand to the receptor [46]. It is thought to do so by directing the ligands to the lateral membranes where endocytosis is less efficient [47].
There are likely to be an extensive array of accessory factors like disp and dally that are required for the efficient release of the Hh ligand from hh expressing cells and its subsequent transport or transmission from one neighboring receiving cell to the next. We have previously shown that one such gene encodes the mevalonate biosynthetic enzyme Hmgcr [14]. We found that hmgcr is required in Hh expressing cells to facilitate the release or transmission of the Hh ligand; however, it was not clear from our studies why the biosynthesis of mevalonate would be important for the release/ transmission of the Hh ligand in flies. The obvious explanation, that it is required for the synthesis of the cholesterol that is used to modify Hh, was not likely to be correct as flies do not have the downstream enzymes for cholesterol biosynthesis [15]. In the work reported here we have resolved this question. We show that the downstream target for hmgcr in the hh signaling pathway is the heterotrimeric G protein, Gc1, which must be geranylated in order to be active [17,31,48]. Like hmgcr and other genes that are required to promote hh signaling, mutations in gc1 dominantly suppress the gain-of-function wing phenotypes of hh Mrt in adult flies. In the embryo, the expression of wg which is activated by hh in the receiving cells is downregulated in both hmgcr and gc1 mutants. This is also true for the en gene which is normally activated by wg signaling in hh expressing cells as part of the autoregulatory circuit that sustains hh and wg expression as the embryo develops. These transcriptional defects arise because the Hh signal is not properly conveyed to hh receiving cells. In wild type embryos Smo protein is redistributed to the membranes of the receiving cells when they receive the Hh signal transmitted from the neighboring Hh producing cells. As observed for hmgcr, Smo protein is not correctly relocalized in gc1 mutant embryos, and instead it remains largely cytoplasmic. Finally, in the ectoderm of wild type embryos there is a gradient of Hh protein extending into the parasegment from the two cell wide stripe of hh expressing cells. Like hmgcr, this gradient is not properly formed in gc1 mutant embryos, and instead Hh protein is inappropriately retained in the Hh producing cells.
Since isoprenoid modifications, either farnesylation or geranylation, are known to be critical for the functioning of the Gc family of proteins, these observations would argue that hmgcr is required for the release of the Hh ligand because it provides a precursor that is needed for the geranylgeranylation of Gc1. This conclusion is supported by the finding that qm, which synthesizes the activated substrate, GGPP, that is used to geranylate Gc1, is also required for the release of the Hh ligand from hh expressing cells. While these results implicate the biosynthetic pathway leading from mevalonate to the geranylation of Gc1 in the proper release of the Hh ligand, we cannot exclude the possibility that there are important targets for geranylation in addition to Gc1 or that other products of mevalonate might play some role in the hh signaling pathway. These possibilities remain open for a number of reasons.
First, the defects in the release of Hh observed in antibody staining experiments seem to be more severe in the qm mutant (and to a lesser extent in hmgcr: see 14) than in the gc1 mutants we examined. One explanation for this difference is that the Qm enzymatic product, GGPP, is used for the geranylation of other proteins that are important for the release of the Hh ligand. However, this could also be due to, for example, differences in the perdurance of the maternal Qm and Gc1 proteins.
Second, ectopic expression of Hmgcr in hh expressing cells causes a readily discernible change in Hh protein distribution across the parasegment and relatively high levels of Hh are found even near the middle of the interstripe region. By contrast, we could not detect an equivalent alteration in Hh distribution when Gc1 was ectopically expressed in hh producing (or receiving) cells. This difference could mean that the mevalonate produced by Hmgcr has uses in hh signaling besides the synthesis of GGPP and the geranylation of Gc1.
An alternative and perhaps more interesting possibility is that the differences in the effects of misexpression on the release/ transmission of Hh protein reflect the fact that Hmgcr is limiting whereas Gc1 is not. Consistent with this idea, the distribution of hmgcr mRNAs becomes progressively restricted as development proceeds and by mid-embryogenesis (stages 10-15) hmgcr mRNAs are only detected in the SGPs [9]. By contrast, mRNAs encoding Gc1, as well as several of the enzymes that are downstream of Hmgcr in the biosynthesis of GGPP, are much more widely expressed in the embryo at this stage [15,28]. A possible consequence of this difference in mRNA distribution is that the levels of Hmgcr protein would increase dramatically when it is ectopically expressed in the ectoderm during mid-embryogenesis while this would not be true for Gc1 or, for that matter, the other GGPP biosynthetic enzymes. The idea that Hmgcr is a limiting component of signaling pathways that depend upon geranylation Gc1 (or other targets) is also supported by the defects in germ cell migration that are induced by ectopic expression of these proteins. For example, expression of hmgcr in the CNS cause much more severe abnormalities in germ cell migration than those observed when fpps, qm, or gc1 are misexpressed [15]. If these ideas were correct, than inducing or repressing the expression of the hmgcr gene would provide a novel posttranslational mechanism for regulating the potency of signaling molecules like Hh.
The effects of gc1, qm and hmgcr on the distribution of Hh in the ectoderm indicates these genes are required for the release of the Hh ligand from hh expressing cells. For Gc1, a role in releasing the Hh ligand from the sending cells would dovetail nicely with a recently discovered function of this G protein and its partners, Gb and Ga in the transport of cargo from the trans-Golgi network (TGN) to the basolateral plasma membrane [19][20][21]. Since Hh protein appears to be specifically targeted to the basolateral membrane in punctate structures (LPSs) prior to secretion [22][23][24][25]38,39], it is not altogether surprising that components of the machinery needed for the transport of cargo from the TGN to the basolateral membrane would play a key a role in transmitting the Hh signal. Moreover, since Gc1 requires geranylation for membrane association and activity, the retention of Hh in qm and hmgcr mutants would also be explained by a disruption in Gc1-dependent TGN-plasma membrane transport. In this context it is interesting to note that while the levels of Wg are reduced in hmgcr [14] and gc1 mutants, there is no obvious over accumulation of the Wg protein inside wg expressing cells like that observed for Hh. That the hmgcrRqmRgc1 pathway would have no apparent effect on the release of Wg would make sense since this is thought to occur preferentially through the targeting of mRNAs to the apical surface of the cell [49].
Though the precise mechanisms for TGN-plasma membrane transport have yet to be elucidated, it is thought that the heterotrimeric G protein complexes mediate the release of cargo from the TGN by promoting membrane fission [25]. In one scenario, interaction of the cargo with an unidentified receptor in the TGN leads to the activation of the trimeric Gc1:Gb: Ga and the release of Ga The Gc1:Gb heterodimer in turn activates several targets including phosphokinase C and a phosphoinostidespecific phospholipase C (PI-PLC) that generates diacylglycerol. PKC participates in cargo release from the TGN by activating Protein Kinase D (PKD) while locally high concentrations of diacylglycerol produced by PI-PLC are thought to change the properties of the TGN membranes and promote membrane fission. After membrane fission, the vesicle containing the cargo is then targeted to the basolateral plasma membrane [24,25]. A requirement in the formation of cargo containing vesicles would fit well with the effects of hmgcr, qm and gc1 on the formation of the puncate Hh particles, or LPSs, that normally assemble along the basolateral membranes of Hh expressing cells. These LPSs are largely absent in hmgcr, qm and gc1 mutant embryos and instead Hh accumulates in much larger aggregates or clumps. While the precise origin of the LPSs is not known, they are thought to be lipid containing vesicles (or micelles) and it would be reasonable to think that they could be generated by the budding of Hh containing vesicles from the TGN. In this case, the large aggregates or clumps of Hh protein seen in hmgcr, qm and gc1 mutants would likely represent Hh trapped either in the TGN or in aberrant vesicles/structures that accumulate in the mutant cells when efficient cargo release from the TGN is disrupted.
While the idea that Gc1 promotes the transport of Hh from the TGN to plasma membrane would seem to fit best with the known functions of Gc1 and its collaborating G proteins, it is also possible that Gc1 (plus Gb and Ga) functions at earlier steps in the secretion pathway, for example. in the transport of Hh from Endoplasmic Reticulum to the Golgi [50]. Alternatively, it is possible that some novel activity of Gc1 at the plasma membrane rather than in the TGN is needed. For example, it could function to prevent the newly formed LPSs from clumping together into larger aggregates. Further studies will be required to distinguish between these and other possible mechanisms.

Gc1 Is Required for Germ Cell Migration
Studies by Santos and Lehmann [15] provided convincing evidence that hmgcr is required in the soma for germ cell migration because its biosynthetic product, mevalonate, is the precursor for the synthesis GGPP by Qm. They also found that GGPP is used in turn by geranylgeranyl transferse type 1 (b-GGT1) for the geranylation of some unknown target(s). The experiments presented here indicate that one (if not the only) somatic target in the germ cell migration pathway is Gc1. Thus, the effects of both gain and loss of gc1 function on germ cell migration closely resemble those reported for hmgcr, fpps, qm, and b-GGT1. Also supporting the idea that Gc1 must be a relevant target for the hmgcr-isoprenoid biosynthetic pathway, we find that Gc1 proteins that cannot be geranylated behave as dominant negatives when ectopically expressed in the mesoderm and disrupt germ cell migration. In addition, there are other significant similarities between the two genes that have been studied in most detail, gc1 and hmgcr. First, both genes show synergistic genetic interactions with components of the hh signaling pathway that perturb the process of germ cell migration. Second, germ cell migration can be disrupted when gc1 or hmgcr are ectopically expressed in hh producing cells; however, there are no apparent effects when the genes are ectopically expressed in hh receiving cells.
Taken together with the fact that the SGPs are known to be a source of Hh these findings would argue that a critical function of the biosynthetic pathway leading from hmgcr to the geranylation of Gc1 is to upregulate Hh signaling in the SGPs, and it is the Hh ligand produced by these cells that serves to attract the migrating germ cells. Importantly, this model accounts for a number of different observations. Since Hmgcr is expressed at high levels in the SGPs, but is not expressed elsewhere in the mesoderm, it would explain how Hh signaling could be specifically potentiated in a special sub-population of cells. It would also explain why the effects of hmgcr misexpression are much greater than misexpression of the other genes in the hmgcrRgc1 pathway that are more broadly transcribed in the embryo. Finally, it would explain why germ cells can be misdirected by ectopic expression of hh, hmgcr, the downstream genes in the geranylation biosynthetic pathway, and gc1 in a variety of different tissues. By contrast, if the SGPs were to induce germ cell migration by expressing some unique and dedicated hmgcrRgc1 dependent attractant, it is hard to understand how misexpression of these different upstream genes would be able orchestrate the production of this special molecule in a variety of cells and tissues that have little resemblance to the SGPs. It should be noted, however, that our results would also be compatible with more complicated models. For example, it is possible that the potentiation of Hh signaling by the hmgcrRgc1 pathway induces the production of a specialized and as yet unknown germ cell attractant. Likewise, we also can not exclude the possibility that there is some other target for geranylation besides Gc1 which is important for the production or activity of a second germ cell attractant and that this unknown molecule functions in concert with Hh to direct germ cell migration towards the SGPs. However, in either of these more complicated scenarios, the unknown germ cell attractant would have to be a molecule that can be induced in many different cell types in the embryo, but apparently only if these cells also express the Hh protein.

Immunohistochemistry
The embryo stainings were performed essentially as described in 51. Vasa (from Paul Lasko) and Hh (from Tom Kornberg) antibodies are rabbit polyclonal antibodies. Both were used at a 1:500 dilution. Engrailed and Wingless antibodies are mouse monoclonal antibodies and were used at 1;10 dilution. b-Galactosidase antibody was either a rabbit polyclonal purchased from Kappel (used at 1:1000 dilution) or a mouse monoclonal antibody from Developmental Hybridoma Bank (used at 1:10 dilution). Smoothened antibody (anti-rat) was a kind gift from Steve Cohen and was used at 1:500 dilution. For confocal analysis a magnification of 406 was used in almost all the instances and images were collected using identical settings for the control and experimental samples. Multiple pairs of wild type (sibs) and mutant embryos were imaged in each case and representative examples are presented.

Mutant and Misexpression Analysis
gc1 mutant stocks, gc1 N159 and gc1 k0817 , were obtained from Fumio Matsuzaki while the various 1UAS-gc1 stocks (gc1, gc1 DCAAX and gc1 C67S) were kindly provided by the Olson lab. The other UAS and GAL4 stocks used for the misexpression studies: UAS-hmgcr, hairy-GAL4, elav-GAL4, nanos-GAL4, patched-GAL4, UAS-b-galactosidase, hh-GAL4/TM6 Ubx-LacZ. In most experiments, males carrying two of the copies UAS transgene were mated with virgin females carrying two copies of the GAL4 transgene. The resulting progeny embryos were fixed and stained for subsequent analysis [51]. Figure S1 Engrailed expression is not properly maintained in gc1 mutant embryos. Embryos from the gc1 N159 / Cy0, en:LacZ stock were collected and fixed using standard procedure. Embryos were genotyped by simultaneously staining them with b-galactosidase (imaged in green: not shown) and En (imaged in red) antibodies. Balancer embryos (Panel A) show strong En specific expression in 14 stripes. By contrast, En specific signal starts to decline by Stage 11 in the homozygous gc1 N159 embryos (Panels B and C). As illustrated in these two panels there is some variation in the extent of the reduction in En expression. In some embryos, moderate levels of En protein are detected (B) while in others only low levels are observed (C). Because of the variability in En accumulation in gc1 N159 homozygous embryos, we classified the En staining pattern. For the heterozygous gc1 N159 /+ control, 6/7 embryos had high levels of En accumulation, while 1 embryo had a medium level of accumulation. For the homozygous gc1 N159 embryos 4/11 (37%) had little En protein (like the example shown in the figure) while 4/11 had a medium level of En protein (like the example shown in the figure). The 3 remaining embryos (27%) resembled wild type. We also examined En expression in homozygous gc1 k0817 embryos. In this experiment all of the heterozygous gc1 k0817 /+ control embryos had a high level of En protein (9 embryos Figure S2 Germ cell migration defects are also observed in gc1 k0817 mutant embryos. Embryos from a gc1 k0817 /Cy0 en:LacZ stock were probed with Vasa to mark the germ cells and bgalactosidase antibodies to identify heterozygous and homozygous mutant embryos. Panels A and B are gc1 k0817 /Cy0 en:LacZ embryos (note b-galactosidase expression) while panel C and D are mutant. Panel C shows embryo with 3 scattered cells whereas the embryo in panel D has more than 6 scattered germ cell cells. About 15% (4/22) of the mutant embryos had 3-4 scattered germ cells (example in panel C), while about 40% (8/22) of the mutant embryos had 5 or more scattered germ cells (example in panel D). The remaining embryos (10/22 or 45%) had 2 or fewer scattered germ cells. Recent studies by Kunwar et al. [52] on germ cell migration have suggested that gc1 has a cell autonomous requirement in germ cells. To test the cell autonomous function of gc1 in germ cell migration, these authors rescued the gastrulation defects of progeny from gc1 germline clone mothers using a nullo-GAL4 to drive expression of a UAS-gc1 transgene. They reported that the pole cells in these embryos failed to migrate properly out of the midgut and exhibited phenotypes similar to those found for mutations in the G protein-coupled receptor (GPCR) gene trapped in endoderm 1 (tre1). We wondered whether the geranylated form of Gc1 is also required in the germline. To explore this possibility we ectopically expressed the dominant negative gc1 deletion mutant Gc1-DCAAX in the germline. As shown in Figure S3E and S3F, ectopic expression of the Gc1-DCAAX protein in germ cells using a nos-GAL4 driver disrupts germ cell migration. We found that nearly 60% of the stage 13-15 nos-GAL4/UAS-Gc1-DCAAX embryos had 3 or more lost germ cells, while 33% had 5 or more lost germ cells (n = 100 embryos). This is roughly equivalent to the germ cell migration defects evident when Gc1-DCAAX is expressed in the mesoderm using a twist-GAL4 driver. The effects of the dominant negative protein in germ cells would support the findings of Kunwar et al., and argue that gc1 (specifically geranylated Gc1) has a cell autonomous function in these cells during their migration towards the SGPs. On the other hand, our results differ somewhat from those reported by Kunwar et al. in that we did not observe any obvious defects in the ability of the germ cells to exit the midgut when the Gc1-DCAAX protein was expressed using the nos-GAL4 driver. While there are a number of plausible reasons why a tre1-like phenotype wasn't observed, the most likely explanation is that not enough of the dominant negative Gc1-DCAAX is generated to disrupt the tre-1 dependent migration through the midgut epithelia. In particular, germ cells are known to be transcriptionally quiescent until just before they exit the midgut, and there might not be sufficient time to generate high enough levels of Gc1-DCAAX to effectively inhibit the maternally derived product. If this explanation is correct, it would suggest that gc1 may also function at a later, tre-1 independent step in the germ cell migration pathway since many of the germ cells in nos-GAL4/UAS-Gc1-DCAAX embryos fail to coalesce with the SGPs. Found at: doi:10.1371/journal.pgen.1000333.s003 (4.2 MB TIF)