Dietary components modulate yolk protein gene transcription in Drosophila melanogaster

The three yolk proteins (YP) of Drosophila melanogaster (Bownes & Hames, 1977) are encoded by single copy genes (ypl, yp2, yp3) on the X chromosome (Barnett et al. 1980). The genes are expressed in the fat body and ovarian follicle cells of the adult female (Bownes & Hames, 1978; Brennan et al. 1982). They thus provide an interesting system for studying sex-limited, tissue-specific and developmentally regulated gene expression.

Two regulatory systems have been described. The two insect hormones, 20-hydroxyecdysone and juvenile hormone, can affect the production of YP. 20-hydroxyecdysone affects YP synthesis in the fat body of males and females (Postlethwait & Shirk, 1981; Postlethwait et al. 1980a,b), and juvenile hormone affects YP synthesis by ovarian and fat body cells in females (Postlethwait & Shirk, 1981; Bownes et al. 1983a). These effects are mediated at the level of transcription (Bownes & Blair, 1986), but there is no evidence for a direct interaction of the hormone-receptor complexes with cú-acting DNA sequences flanking the YP genes (Bownes et al. 1987).

A second system is that of the sex-determination hierarchy of regulatory genes such as transformer, transformer-2 and doublesex which act in trans to permit transcription of the yp genes only in females (Postlethwait et al. 1980a; Bownes & Nöthiger, 1981; Belote et al. 1985). It is not clear how these two systems are superimposed.

DNA including ypl and yp2 and the intergenic region give the correct sex specificity, tissue specificity and developmental profile when introduced into flies by P-element transformation (Tamura et al. 1986). The sequence requirements for expression of the yp genes in the fat body and follicle cells are separable (Garabedian et al. 1985, 1986) and located between ypl and yp2. The fat body specificity and sex specificity have been located to a 125-base-pair fragment from the intergenic region (Garabedian et al. 1986). As yet nothing is known about the sequence requirements for the correct expression of yp3 nor about those required for 20-hydroxyecdysone and juvenile hormone to exert their effects on the genes.

If we are to understand the molecular basis of how transcription of this gene family is regulated we must understand how the sexual and hormonal regulatory systems interact and identify any other factors important in controlling their expression. It is well known that Drosophila females lay eggs in relation to their environment, and starvation of flies or removal of specific dietary components leads to a cessation of egg production (Sang & King, 1961; Bownes & Blair, 1986).

In this paper we show that, without feeding, the yp genes of newly eclosed females are transcribed at a basal level over several days. In contrast, if food is available, yp transcripts rapidly accumulate. This effect is specific to the yp genes and is not due to a failure of fat body development. It is not the result of any nutrient (e.g. a specific amino acid) becoming limiting and the effects are partially reversed in a dose-dependent fashion by 20-hydroxyecdysone and juvenile hormone. We propose that the presence of various dietary components in the gut activates, via an intermediate molecule or molecules, an enhancer of yp transcription in the fat body. The mechanism may be related to the induction of vitellogenesis in response to feeding as observed, for example, after a blood meal in the mosquito, Aedes aegypti (Van Handel & Lea, 1984).

Flies were maintained on normal food (standard yeast, maize meal, sugar and agar) or on 3 % agar at 25 °C. The wild-type strain, Oregon R, was used throughout, except in the experiment using ypl-Adh transformants, which have the genotype ypl-Adh Adh^nLA248cn bw.

Flies sexed as pupae were allowed to eclose into agar vials and either treated topically with ZR515 or injected with 20-hydroxyecdysone as in Bownes & Blair (1985).

Flies were maintained from eclosion or transferred from 3% agar onto either normal food, 3% agar, 3% agar + 80% sucrose, 3% agar + casein gel, 3% agar + amino acid (aa) mix (contains all 20 amino acids at 0 ·5 M). Further sucrose, aa mix and casein gel were added via soaked strips of filter paper. 2-day-old flies maintained on 3% agar were either injected with 0 ·5 μM-aa mix or had 80 % sucrose applied to their proboscis while etherized, for the other trials.

Each fly was divided into ovary, fat body and haemolymph, and the proteins were solubilized in 50 μl Laemmli buffer (Laemmli, 1970). Polyacrylamide gel electrophoresis was as in Bownes et al. (1983). To radioactively label proteins, flies were injected with 0 ·1 MBq [³⁵S]methionine (29 ·6 TBq mmol ⁻¹, Amersham). After 2h, flies were prepared for electrophoresis as above. Incorporation of label into protein was estimated by precipitation with 5 % trichloroacetic acid (TCA). After electrophoresis, gels were prepared for fluorography using ‘Enlightening’ (New England Nuclear).

RNA was isolated from ovaries, carcasses and whole flies as in Bownes et al. (1983b). Northern blots were made onto Hybond membrane (Amersham) as in Thomas (1980), as modified in Bownes et al. (1983a).

25 ng of pGem YP1 (the whole yp1 gene was cloned into a pGem vector from Promega Biotech (M. Bownes & M. Blair, unpublished data)) or α-tubulin DNA was randomly primed as in Feinberg & Vogelstein (1983) using Amer-sham’s multiprime DNA-labelling system, to a specific activity of approx. 10⁹ctsmin ⁻¹μg⁻¹.

Spectrophotometric assays and tissue staining for ADH were carried out as described by Bonner et al. (1984).

Females that hatched either into agar vials or vials of normal food were analysed after 0–1, 1–2 and 2–3 days. Some samples were transferred between agar and normal vials. As can be seen in Table 1, starving dramatically reduced egg production. After 2–3 days, no new oocytes enter vitellogenesis. Gel electrophoresis of the total proteins present in the haemolymph and ovaries showed little difference in the proportion of YP present compared to total protein that had accumulated over the 3-day period (Table 2). Fat bodies also looked similar showing only trace amounts of these rapidly secreted proteins. The results suggest that as egg production and vitellogenesis ceases, so does YP synthesis. Thus the relative amount of YP in the haemolymph remains static. Similarly, in the ovary, as yolk uptake is reduced so is the production of all the other cell types. As oogenesis is arrested, again the relative proportion of YP to other proteins remains constant. The similarity of the starved population to the fed population in the fat body YP profile indicates that YP are not synthesized and stored under conditions of starvation.

When injected with [³⁵S]methionine, between 86 and 96 % of the label was taken into the ovaries and carcasses from the haemolymph, where it was injected, and there was no difference between starved and fed flies (data not shown). Similarly, when the amount of label incorporated into TCA-precipitable material was measured, there was no significant difference between the starved and fed flies. Throughout the 3 days, 6–10 % of the label taken up into carcasses (which include the fat body) was incorporated in macromolecules; in the ovary, the figure was 40–80 % (data not shown) indicating that under all circumstances labelled amino acids are accumulated in tissues and incorporated into newly synthesized proteins. Thus, the effects of starvation are not to inhibit protein synthesis completely. It should be noted that it is likely that the amino acid pools in the haemolymph will gradually become depleted in starved flies. This will result in more of the labelled methionine becoming incorporated into protein in starved flies than in fed ones. Thus the results may be biased to give an apparently higher overall rate of protein synthesis in the starved flies. We can only say, therefore, that the rate of synthesis of proteins is not dramatically reduced by starvation.

Analysis of these newly synthesized proteins by gel electrophoresis and autoradiography showed that there was only a low level of YP present in the fat body, as expected, since they are secreted proteins. In the ovaries, newly synthesized YP were detected in all fed flies. The amount was much greater in 2- and 3- day-old flies than in 1-day-old flies. In the starved flies, YP were synthesized on days 0–1, but were not visible in the ovaries on later days.

In the haemolymph samples, which represent the newly synthesized proteins secreted from the fat body, the results were quite different in the two populations as can be seen in Fig. 1. The proportion of YP synthesized in relation to total protein synthesis remained fairly constant in the starved flies, but in the fed flies there was a dramatic increase on days 1–2, followed by a small decrease on days 2–3, presumably due to the rapid transfer of some of the YP from the haemolymph into oocytes. These results indicate that, in starved flies, the pattern of protein synthesis is different to fed flies and the normal increase in YP synthesis occurring during days 1–2 does not occur.

The finding that protein synthesis continued in starved flies and that the proportion of YP synthesized was selectively reduced suggested to us, since YP are not abundant in methionine, that perhaps yp transcription was altered during starvation. We measured transcript levels, as shown in Fig. 2A, and found that yp transcripts increase over the first 3 days from eclosion in fed flies. In starved flies, there is an increase over 0–1 days, they then remain constant. The fluctuation on days 2–3 is due to loading of the gels as can be seen by probing with rDNA (Fig. 2B). Thus transcription is likely to be maintained at a low level in starved flies. These results mirror those observed at the protein level. To be sure that transcription in general is not affected, the filter was reprobed with a a-tubulin DNA (Fig. 2C) and shows a pattern similar to that obtained with the ribosomal probe, suggesting a similar abundance of this transcript in starved and fed flies.

Flies were allowed to hatch into agar-containing vials and maintained for days. They were then treated by injection with 20-hydroxyecdysone or topical application of a juvenile hormone analogue (ZR515) at doses ranging from 10⁻²-10⁻¹⁰M. They were returned to agar vials for a further day before their total proteins in the haemolymph, ovary and fat body, their rates of protein synthesis and the yp transcripts present were analysed. The 12 h time point was chosen after pilot experiments treating starved flies with hormones had shown that the optimal effect on YP synthesis occurred at this time. Controls from the same population of flies were starved and treated with appropriate solvents as above, or fed for the whole period (Fig. 3).

Vitellogenesis in the flies treated with 20-hydroxyecdysone never resumed, even at 10⁻²M. This has been observed many times in various experiments and reflects the fact that 20-hydroxyecdysone acts specifically on the fat body (Jowett & Postlethwait, 1980; Bownes et al. 1983a; Bownes & Blair, 1986). However, those treated with ZR515 had early vitellogenic stages present (Table 1). The restoration of vitellogenesis was observed in a few flies using 10⁻⁸ and 10⁻⁶M-ZR515 and in all flies using 10⁻⁴ and 10⁻²M-ZR515. Note that some mature, retained eggs are present in all cases; this occurs in starved controls and is not the result of the hormone. Injection tends to cause the oocyte in the oviduct to be laid and hence is counted as a laid egg rather than a mature oocyte (Table 1).

The total proteins were analysed by gel electrophoresis. As can be seen in Table 3, the treatments with ZR515 and 20-hydroxyecdysone made little difference to the total amount of YP compared to the starved or fed controls. In all cases, the fat bodies contained less than 1 % YP due to the rapid secretion of these proteins. This shows that there is no accumulation of YP in the fat body. To look at the dynamics of YP synthesis, flies were injected with [^3:>S]meth-ionine. The uptake of injected radioactive methionine into organs remained above 90 % in all cases. The incorporation into proteins was also similar to the controls ranging from 30–48 % in the ovaries, 4–9% in the carcasses of the ZR515 experiment, 24–32 % in the ovaries and 4–8 % in the carcasses of the 20-hydroxyecdysone experiment. Thus, treatment with these hormones neither dramatically stimulated nor inhibited protein synthesis in general. Since all these experiments are done with the same population of flies, methionine pools are likely to be similar between groups. After gel electrophoresis and autoradiography there were some interesting results. In the fed controls, the YPs were very obvious in the haemolymph and ovaries and were detectable in the fat body. In starved controls, they were absent from the fat body in all cases. After treatment with 20-hydroxyecdysone, they were observed in the haemolymph at 10⁻²M and 10⁻⁴M (Fig. 3A). They could be readily detected in fat bodies at 10⁻²M and were just detectable at 10⁻⁴ and 10⁻⁶M. There was no labelling of ovaries, indicating that there was no uptake of YP or ovarian synthesis stimulated at any dose of hormone.

After treatment with ZR515 at 10⁻², 10⁻⁴ and 10⁻⁶M, YP could be seen in the fat body and were above control levels in the haemolymph at 10⁻²M (Fig. 3B). All three yps are translated in response to the hormones though YP3 is often less abundant than YP1 and YP2.

These results were also observed at the level of yp- transcript accumulation (e.g. Fig. 4), thus indicating that this regulation occurs at the level of transcription or transcript stability.

Taken together, these results show that the effects of starvation on yp transcription in the fat body are overcome, at least in part, by the juvenile hormone analogue and 20-hydroxyecdysone in a dose-dependent fashion. Although these doses are between 100 and 1000 times higher than the physiological level, 50 % of an injection of 1000 times the physiological titre of 20-hydroxyecdysone is catabolized and excreted by the female in 4h and it is virtually undetected after 12h (Smith & Bownes, 1985). Only juvenile hormone restores vitellogenesis (Table 1), which is consistent with earlier results on isolated abdomens (Jowett & Postlethwait, 1980). There is no evidence as to whether these hormones act directly on the fat body or via other interactions in the animal.

It was possible that starvation led to reduction of essential components needed for transcription and translation of yps. Indeed with reduced intake of food the protein reserves of the fly are bound to be insufficient for normal levels of YP production. However, since ZR515 and 20-hydroxyecdysone can restore, at least partially, yp transcription and translation, it is unlikely that any particular amino acid, carbohydrate, salt or vitamin is completely limiting in the starved animals. The flies are able to transcribe and translate these genes, but starvation inhibits the process to a greater degree than would seem to be solely due to lack of individual components of the proteins.

We have further confirmed this by using flies that are first fed normally on complete medium for 3 days and then fed sucrose for 1 day. This is unlikely tc deplete specific nutrients (Sang & King, 1961). Aftei this regime, the flies are very active in protein synthesis, yet after injection of [³⁵S]methionine the synthesis of YP.is seen to be selectively inhibited in both ovaries and fat bodies (Fig. 5). This occurs in flies whether or not they have been mated, and is mediated at the level of transcription or transcript stability (data not shown). These results also show that transcription is not just a function of gene copy number differences in starved and fed flies as both populations were fed normally over the first 3 days when the process of fat body polyploidization occurs.

When these sugar-fed flies are treated with ZR515 and 20-hydroxyecdysone, they respond by reactivating yp transcription and translation and the dose dependency is similar to that described for flies starved from eclosion. This suggests that feeding is continuously needed to maintain high levels of yp transcription. If feeding on a complete diet ceases, the level of yp transcription is reduced or yp transcripts become less stable over a relatively short time span.

To discover if the effects of diet were due to lack of a single compound or more complex, we starved flies for 2 days, then transferred them to various diets or treated them for a further day before injecting [³⁵S]methionine to analyse protein synthesis. The results showed that only adding live yeast gave a dramatic recovery; feeding casein gave a slight response as did feeding an amino acid mix or sucrose. Injecting amino acid mix or distending the stomach with sucrose had no effect. The results are shown in Table 4. Only the fed controls and yeast-fed flies showed vitellogenic stages of oogenesis and laid eggs.

In a further experiment, we allowed flies to hatch onto varying diets and then compared them after 3 days. Again, only the live yeast gave levels of YP synthesis similar to fed controls though amino acid, casein and sucrose were all better than the starved controls (Table 4). Many eggs were laid in fed controls and yeast-fed flies; in those fed with amino acid mix or casein, a small number of vitellogenic stages 8-13, which were absent in sucrose-fed and starved controls, was present.

It seems that the effects of diet are complex. It is likely that the poor recovery with casein may arise because the proboscis is able to take up nutrients from yeast and liquids much more easily than the casein gel.

It should be noted that the effects of ZR515 and 20-hydroxyecdysone are at levels similar to feeding protein, amino acid mix or sucrose and are quantitatively considerably less than the effects of live yeast. If dietary signals operated via ecdysone and juvenile hormone one might expect the hormones to give a much greater stimulation of vitellogenesis, although it may be essential to treat flies with both hormones in correct proportions and in the correct sequence to achieve this. Because of the rapid catabolism of injected hormones and our inadequate knowledge of hormones, it is not possible to design sensible experiments at present to further test this.

Flies transformed with a ypl-Adh fusion gene containing 890bp of 5′ flanking yp DNA in an Adh” background were assayed for ADH activity under fed and starved conditions on the 3 days following eclosion. The results are shown in Fig. 6. ADH activity in both fed and starved flies increases following eclosion.

In fed flies, ADH activity rises steadily until day 3. The starved flies show a much less pronounced rise in activity reaching a maximum approximately one third of that in fed flies at 2–3 days. These results suggest that the DNA sequences fused to Adh are sufficient for the enhancement of yp transcription by feeding.

This was further confirmed by starving ypl-Adh flies for 3 days then providing them with a normal diet. ADH activity rose twofold after 24h (Fig. 7), but did not completely reach normally fed levels.

To determine whether JH and 20-hydroxyecdysone act on the same DNA fragment, ypl-Adh flies starved for 2 days were treated with the two hormones. Neither caused a significant increase in ADH activity (Fig. 7), indicating that the dietary stimulation of transcription is unlikely to be mediated by 20-hydroxyecdysone or juvenile hormone.

We have presented evidence that there is a system for modulating YP gene expression and egg production in relation to diet. Such effects have been reported in the mosquito in response to a blood meal (Van Handel & Lea, 1984) and in more physiological experiments diet has been shown to be critical for vitellogenesis in the housefly (Lea, 1972).

This system probably acts to enhance transcription of the YP genes or increase mRNA stability. The results have been shown to be specific to the YP and are unlikely to be due to general effects on protein synthesis or transcription. We have shown that the feeding-induced enhanced transcription operates on an 890 bp fragment of DNA but that the hormones do not exert their effect on this sequence. This is consistent with our failure to induce ADH activity in male flies transformed with the same construct (Shirras & Bownes, 1987) and also suggests that the effects of diet do not operate by modulating hormone titres alone.

It is unlikely that the regulation of yp transcription operates by feedback of YP haemolymph levels alone. If starvation inhibited vitellogenesis, haemolymph YP would stabilize and could feed back to reduce transcription; addition of food or ZR515 would lead to uptake and thus deplete haemolymph YP and lead to increased transcription. However, 20-hydroxyecdysone alone increases transcription of the YP genes and in this case there is no restoration of vitellogenesis and reduction of haemolymph YP. Furthermore various mutants with defective vitellogenesis maintain YP synthesis (Postlethwait et al. 1980b; Bownes, 1983) and ovariectomized flies also maintain YP synthesis (Postlethwait et al. 1980 b). So although YP titres in the haemolymph may affect yp transcription at present, the available evidence is against this.

An alternative would be that, in response to various nutrients, signals are propagated in the gut, perhaps gut hormones, which could directly act upon the fat body ultimately altering yp transcription or they could induce hormones or other signals to be synthesized in the brain or the ovaries. It is clear that during starvation or sugar feeding vitellogenesis rapidly ceases whilst YP haemolymph titres are still high, so there must be some kind of regulation operating whereby the progress of oogenesis is modulated by diet. It is not known how this is achieved. The affects on yp transcription may be part of a cascade of events, or the appropriate hormones or signals may act on the ovary to inhibit vitellogenesis and the fat body to reduce yp transcription independently. We also do not know whether this process is a negative one as suggested above or a positive one whereby appropriate signals stimulate these events. Either seem equally possible at this time.

We present in Fig. 8 our current understanding of the regulatory regions lying between the ypl and yp2 genes. Sequences conferring ovarian expression on yp2 lie between ypl and yp2 (Tamura et al. 1985; Garabedian et al. 1985), and a sequence conferring expression of ypl and yp2 in female fat bodies has been shown to be located within a 125 bp fragment (Garabedian et al. 1986; Tamura et al. 1986). Insert sequences that respond to the dsx gene products have been located to an 890 bp fragment (Shirras & Bownes, 1987) which includes the 125 bp identified by Garabedian et al. (1986). It is likely that the dsx products act on this 125 bp sequence since it confers fat body expression of genes in females only, but this has not yet been directly tested. The sites conferring hormone inducibility have not been identified but they lie outside the 890 bp fragment. Although the mechanisms of action are unknown, the dietary signals also operate within the 890bp fragment. It appears that there are likely to be multiple binding sites responding to enhancers or repressors (positive or negative regulators) of transcription which are expressed in either a tissue-specific or sex-specific fashion (e.g. the products of the dsx gene) and also factors produced in relation to the physiological status of the fly (e.g. the nutritional enhancement). There is no evidence yet whether one or several proteins bind upstream of the YP coding sequences and whether the signals initiated in response to diet modulate the YP directly by acting as DNA-binding factors or if they act by inducing or modifying other binding factors.

We thank the Zoecon Corporation for providing the ZR515. We also thank Graham Brown for printing photographs, Annie Wilson for labelling figures and Betty McCready and Jeanette Ferguson for typing the manuscript. This research was supported by the Science and Engineering Research Council. Alan Shirras is supported by a Medical Research Council training fellowship.