The macrophage genetic cassette inr/dtor/pvf2 is a nutritional status checkpoint for developmental timing

Starvation and macrophage deficiency delay ecdysone biosynthesis and metamorphosis in Drosophila larvae

Drosophila larvae lacking nutrients at the pre–critical weight (pre-CW) phase (fig. S1A) and hemolectin (hml)–expressing macrophages both display a developmental delay in the onset of metamorphosis also known as the larva-to-pupa transition or pupariation (31, 42, 43). Pupariation of larvae starved at the pre-CW phase, 88 hours after egg laying (AEL), for a period of 8 hours (see Materials and Methods and fig. S1A) was delayed by ~8 hours in comparison to controls fed on standard molasses medium (Fig. 1A). Delayed pupariation of larvae starved at the pre-CW phase is due to delayed ecdysone biosynthesis by the prothoracic gland (fig. S1B) (42, 43). Accordingly, at the time of the normal onset of pupariation under our laboratory conditions (104 hours AEL), transcription of the ecdysone target eip75b (e75b) (Fig. 1B) (44) and ecdysone levels (Fig. 1C) were reduced by ~90% in comparison to control. Macrophage-less larvae (Fig. 1C) (27, 45, 46) also presented with an ~8-hour delayed pupariation, but irrespectively of feeding conditions (fig. S1A and Fig. 1, D and E). A time course analysis indicated that transcription of the ecdysone target eip75b in these conditions is delayed by ~8 hours in comparison to controls (Fig. 1F), and ecdysone levels are reduced by ~90% at 104 hours AEL (Fig. 1G). Transcription of the ecdysone biosynthetic enzymes (47)—neverland (nvd), phantom (phm), and shadow (sad)—was also delayed by ~8 hours (Fig. 1H) as observed in starved larvae (fig. S1B). These data suggested that both pre-CW starvation and macrophage deficiency increase the length of the L3 larval (feeding) stage before pupariation by delaying the biosynthesis of ecdysone. Macrophages can act as nutrient sensors (22), and we therefore considered the hypothesis that they may regulate ecdysone biosynthesis to optimize the timing of pupariation according to the nutritional status of the larvae (Fig. 1I). To test this hypothesis, we performed genetic screens to identify the putative macrophage secreted growth factor(s) or cytokines that could directly or indirectly control developmental timing via ecdysone synthesis, along with the putative sensor(s) for larval nutritional status that could regulate the production of the secreted factor(s) by macrophages.

Production of the PDGF family member pvf2 by macrophages controls the timing of ecdysone production and metamorphosis

Drosophila macrophages express nutritional sensors, such as nuclear receptors and the insulin receptor (InR), and are a source of secreted factors including cytokines and growth factors that can mediate adaptive responses of local or distant tissues in response to external or internal cues (27, 28, 48). We thus designed a targeted knockdown screen for genes encoding secreted factors and nutrient-sensing receptors that are expressed in larval macrophages before the onset of pupariation (96 hours AEL) (figs. S2 and S3) (49), testing their role in the timing of pupariation and ecdysone production. We found that selective knockdown of the platelet-derived growth factor (PDGF)/vascular endothelial growth factor (VEGF) conserved secreted factor pvf2 in macrophages, using three independent RNA interference (RNAi) transgenes and three hemocyte drivers recapitulated the ~8-hour delay in pupariation (Fig. 2, A to C, and fig. S4, A to C). Pupariation was also delayed in the pvf2^c06947 loss-of-function mutant (pvf2^−/−) (Fig. 2B and fig. S4D) (50). Nevertheless, knockdown of pvf2 in other tissues such as the fat body, prothoracic gland, trachea system, and glia cells did not affect pupariation (Fig. 2B and fig. S4E). In addition, we found that pvf2 transcript levels were selectively low in larvae deficient in macrophages (Fig. 2D) and in macrophages from starved larvae (Fig. 2E and fig. S4F). Furthermore, knockdown of pvf2 in macrophages delayed transcription of the ecdysone target eip75b (Fig. 2F) and reduced ecdysone production at 104 hours AEL (Fig. 2G), as observed in macrophage-less and starved larvae. Conversely, supplementation with the active form of ecdysone hormone, 20-hydroxyecdysone (20-HE; 0.5 mg/ml in food), rescued the developmental delay of larvae lacking pvf2 in macrophages and macrophages-less larvae (fig. S4G). Last, expression of a pvf2 transgene in macrophages also fully rescued the developmental delay and ecdysone production in starved larvae (Fig. 2, H and I) as well as in the pvf2^−/− mutant (Fig. 2, J and K), although it did not trigger early pupariation before 104 hours AEL (Fig. 2, H and J). Together, these data indicate that transcriptionally regulated production of Pvf2 by macrophages controls an 8-hour window in the timing of ecdysone production and metamorphosis, based on whether the larvae have eaten. This suggests that the production of Pvf2 by macrophages is a nutritional status checkpoint for the developmental timing of metamorphosis.

Insulin/TOR signaling is required for of Pvf2 production by macrophages

While testing the effect of known molecular sensors of nutritional status, we found that knockdown in larval macrophages of the insulin-like receptor (inr) (51) and target of rapamycin (dtor) (fig. S5A) (52), but not of the ecdysone receptor itself, also recapitulated the developmental delay observed in starved, macrophage-less, and macrophage-pvf2–deficient larvae (Fig. 2, L and M, and fig. S3). The insulin/target of rapamycin (TOR) pathway is a conserved sensor of nutritional stress in metazoans, controlling cell and organism growth (53–55) and particularly the timing of differentiation (14). The array of nutrition-related signaling inputs to this pathway has been defined (56), but the growth-regulatory outputs are less clear. Overexpression of a dominant-negative inr (uas-inr^DN) in macrophages or a second independent dtor RNAi line recapitulated the pupariation delay (fig. S5, B and C). In addition, macrophage inr and dtor knockdown both specifically decreased pvf2 expression (Fig. 2N and fig. S5D), ecdysone production (Fig. 2O), and transcription of eip75b (Fig. 2P). Moreover, we found that the epistatic rescue of dtor knockdown larvae by expression of a pvf2 transgene in macrophages restored developmental timing (Fig. 2Q) and ecdysone production (Fig. 2R). Of note, InR can regulate metabolism and growth via two main downstream pathways: activation of Tor and inhibition of FoxO (51), but FoxO overexpression did not affect developmental timing (fig. S5E). These results indicate that the InR/TOR pathway in macrophages regulates ecdysone production and the developmental timing of pupariation via the production of Pvf2.

Macrophage-derived Pvf2 regulates ecdysone biosynthesis via its receptor Pvr in the prothoracic gland

Together, the above data identify an inr/dtor/pvf2 macrophage genetic cassette that acts as a nutritional status checkpoint that can delay ecdysone production. Ecdysone is mainly produced by the prothoracic gland, an endocrine gland in the Drosophila larva (12). As noted above, ecdysone production in the prothoracic gland is under the control of the Drosophila GnRH superfamily hormone Crz-producing neurons, via the PTTH produced by the PTTH neurons, and of Drosophila insulin-like peptides (DILPs) produced by insulin-producing neurons (4, 5, 11–13). However, tyrosine kinase receptors including Pvr can also regulate developmental timing (57). Pvr is the only known Pvf receptor in Drosophila (58) and is expressed ubiquitously by macrophages, neurons, and fat body cells (59, 60). Thus, we next investigated whether macrophage Pvf2 may control ecdysone synthesis in the prothoracic gland directly or indirectly through its effect on macrophages themselves, neurons, or fat body cells.

A time course analysis of Pvr expression in the prothoracic gland using an anti-Pvr antibody indicated that Pvr is first detected at 88 hours AEL and stays detectable until the onset of maturation at 104 hours AEL (Fig. 3A). Consistently, we found that transcription of pvr in dissected brain/ring gland tissue was up-regulated between 80 and 88 hours AEL to reach its highest level at 104 hours AEL (Fig. 3B). Furthermore knockdown of pvr in the prothoracic gland recapitulated the 8-hour developmental delay observed in starved; macrophage-less; and pvf2-, inr-, and dtor-knockdown models above (Fig. 3C and fig. S6A). The hypomorphic pvr⁰¹ mutant (58) recapitulated the developmental delay (fig. S6B). Knockdown of pvr in fat body, neurons, or macrophages did not affect the developmental timing (Fig. 3D and fig. S6C). In contrast, knockdown of pvr in the prothoracic gland did decrease ecdysone synthesis at 104 hours AEL (Fig. 3E) and delayed transcription of the target gene eip75b by 8 hours (Fig. 3F). As observed in the starved and macrophage-less inr and dtor models (see Fig. 1G and figs. S1B and S6D), we found that knockdown of pvf2 in macrophages and knockdown of pvr in the prothoracic gland delayed transcription of the ecdysone biosynthetic pathway genes nvd, phm, and sad by 8 hours (Fig. 3G). In contrast, starvation or inr and dtor knockdown did not decrease pvr levels in the brain–ring gland complex (Fig. 3H and fig. S6E). Last, pvr overexpression in the prothoracic gland (57) fully rescued and even accelerated the developmental timing and ecdysone production in the pvf2^−/− mutant (Fig. 3, I and J, and fig. S6, F and G). Of note, the latter contrasts with pvf2 overexpression that did not accelerate the developmental timing (Fig. 2, H and J, and fig. S6H). We also examined a possible redundancy between the role of Pvf2 and its paralogs Pvf1 and Pvf3, as they share the same receptor Pvr (58). However, starvation and inr and dtor knockdown did not alter pvf1 and pvf3 transcript levels (figs. S4F and S5C). In addition, knockdown of pvf1 and pvf3 did not affect developmental timing or ecdysone synthesis (Fig. 2A and figs. S2D and S7). Together, these data indicate that InR/dTor signaling in macrophages is required for Pvf2 production, which is, in turn, required to trigger pupariation at 104 hours AEL, by acting on its receptor Pvr, expressed in the prothoracic gland at the onset of pupariation to initiate ecdysone synthesis. However, these data also reveal that, in the absence of the Pvr-dependent signaling axis, which we describe here, ecdysone biosynthesis and pupariation are initiated 8 hours later in a Pvr-independent manner.

Nutritional status–dependent Pvf2 biological activity is regulated by transcription and macrophage numbers

We have shown above that macrophage depletion (hmlΔ>rpr, Fig. 2D) decreases pvf2 levels, ecdysone synthesis, and delay pupariation. Down-regulation of ras85D in macrophages, which also reduces macrophage numbers by >50% (61), has the same effects pvf2 levels, ecdysone synthesis, and pupariation (fig. S8, A to D). Pvf2 promotes proliferation and survival of macrophages via Pvr in a cell autonomous manner (50, 58, 62); therefore, regulation of macrophage numbers represents a possible confounding factor for the analysis of the effects of pvf2 down-regulation. Macrophages are reduced by ~30% in pvf2 macrophage knockdown larvae (Fig. 4A). Thus, manipulating this axis could alter signal production by macrophages by altering their quantity. However, pvr knockdown in hemocytes, which also reduces macrophage numbers by ~30% (Fig. 4A), does not decrease pvf2 production by macrophages and does not delay pupariation in comparison to controls (Fig. 4B). Macrophage numbers are also reduced by ~30% in starved larvae as well as macrophage inr- and dtor-knockdown larvae (Fig. 4A). However, reverse transcription quantitative polymerase chain reaction (RT-qPCR) quantification showed a 60 to 80% decrease, of pvf2 transcripts on a per cell basis, in fluorescence-activated cell sorting (FACS)–sorted macrophages from starved and inr-, dtor-, and pvf2-knockdown larvae (Fig. 4B). Therefore, starvation and inr, dtor, and pvf2 knockdown moderately decrease macrophage numbers, which may contribute, in part, to the overall decrease in pvf2 production, in addition to the transcriptional control of pvf2 in macrophages.

In addition, we considered the possibility that the production of Pvf2 could be restricted to a subset of macrophages whose survival or proliferation is selectively affected by starvation, deficient InR/dTor, and Pvr signaling. Analysis of inr, dtor, pvr, and pvf2 expression in a recent Drosophila larva single-cell RNA sequencing (RNA-seq) dataset (63) showed that inr, dtor, and pvr expression is widely distributed and is not restricted to any potential subset (Fig. 4C). pvf2 expression itself, although distributed throughout many hemocyte clusters, appears more abundant on a per cell basis in clusters of activated hemocytes (LM1; as well as PM5, PM3, and PM8) in the dataset analyzed in (63), consistent with the idea that a threshold of hemocyte activation is associated with pvf2 transcription. To further investigate whether pvf2 expression by hemocytes is reduced at the single-hemocyte level under starvation conditions, we used an enhancer-trap transgenic fly that expresses a Gal4 transcription factor under the control of the pvf2 locus (64), to drive expression of green fluorescent protein (GFP) (mcd8::gfp) in Pvf2-expressing cells. Flow cytometry analysis of hemocytes from starved and control hmlΔ-dsred;pvf2^tra-gal4>mcd8::gfp larvae indicated that ~6% hemocytes express mcd8::gfp in control conditions versus ~2% in starved conditions (Fig. 4D). These results confirm that Pvf2 biological activity is regulated at the transcriptional level under starvation conditions. RNA-seq analysis also indicated that pvf2-expressing hemocytes express hml, srp, and he (Fig. 4, D and E), consistent with the biological effects of pvf2 depletion with the corresponding hemocyte drivers (fig. S4, B and C).

Together, our data indicate that InR and dTor signaling in macrophage licenses ecdysone synthesis via the transcriptional regulation of pvf2 and, to a lesser extent, by the regulation of hemocyte numbers themselves. These data are consistent with the hypothesis that the expression of pvf2 by macrophages acts as a developmental checkpoint that licenses Pvr-dependent pupariation if the larval nutritional status, as monitored by InR and dTor, is adequate.

The macrophage inr/dtor/pvf2 cassette controls the size of adult flies

Delays in critical transitions such as pupariation have systemic effects on organismal and tissue growth (4, 17, 43, 57). In particular, delayed pupariation in response to suboptimal food supplies is believed to minimize the effect of nutrient deficiency on the size and fitness of the adult by extending the growth period (55). We thus investigated the consequences of the Pvf2/Pvr-dependent developmental delay for the size of pupae and adult Drosophila. We found that abrogating the pupariation delay in starved larvae by a pvf2 transgene further reduced the size of the pupae (Fig. 5A). Conversely, enforcing a Pvr-dependent developmental delay by knockdown or expression of dominant negative forms of inr, dtor, and pvf2 in macrophages using several macrophage specific drivers and pvf2 RNAi lines and in pvf2^−/− mutants always resulted in larger pupae under ad libitum feeding condition (Fig. 5, B to D). In addition, the genetic rescue of pvf2^−/− mutants and the epistatic rescue of dtor knockdown larvae by overexpression of a pvf2 transgene both restored wild-type size (Fig. 5D). In contrast, knockdown of pvf2 in glia, prothoracic gland, trachea, and fat body had no effect on pupal size (Fig. 5E). Knockdown of pvf1 and pvf3 in macrophages had no effect on pupal size either (fig. S7F). Last, the knockdown of inr, dtor, and pvf2 in macrophages also generated larger adult flies, both female and male, as measured by wing size (Fig. 5F). To further confirm that the effect of the inr/dtor/pvf2 cassette on pupal and adult size is mediated by Pvr signaling in the prothoracic gland, we examined the role of pvr knockdown, loss of function, or gain of function. Knockdown of pvr in the prothoracic gland, but not in macrophages, increased the size of pupae (Fig. 5G). In addition, the pvr hypomorph mutant phenocopied the knockdown of pvr in the prothoracic gland (Fig. 5G). Knockdown of pvr in the prothoracic gland also increased the sizes of adult female and male flies (Fig. 5H). In contrast, overexpression of pvr in the prothoracic gland, which shortens the pupariation time (see Fig. 3G and fig. S6E), decreased the size of pupae (Fig. 5I) and adult flies (Fig. 5H). Last, the size increase observed in null pvf2^−/− mutants was epistatically rescued by pvr overexpression in the prothoracic gland (Fig. 5I). Together, these data demonstrate that control of the Pvr-dependent ecdysone synthesis and pupariation delays by the macrophage inr/dtor/pvf2 cassette regulates the size of pupae and adult flies from both sexes (Fig. 5J). Of note, pupal and adult size of macrophage-depleted flies was unchanged in comparison from control (fig. S8, E and F), suggesting that macrophage depletion affects other processes that control growth in addition to effects mediated by Pvf2 as also suggested by other studies where macrophage depletion caused reduced (31, 65) or unchanged size (27, 45, 66).

The macrophage inr/dtor/pvf2 cassette is a developmental timing checkpoint that increases fitness

We report here that validation of the nutritional status of the larvae by the macrophage inr/dtor/pvf2 genetic cassette is a checkpoint that allows Pvr-dependent ecdysone synthesis in the prothoracic gland and the onset of metamorphosis to proceed. If this nutritional status is not met, here in the case of a starvation event, low Pvf2 production delays ecdysone synthesis and opens a time window of an extra ~8 hours for preparing pupariation, after which Pvr-independent ecdysone synthesis and pupariation occur (Fig. 5J). The physiological significance of this time window controlled by macrophages and Pvr expression in the prothoracic gland for the fitness of the host is best illustrated by its consequence on the fitness, here the size, of pupae and adult flies, as opening the window allows the formation of larger flies and can rescue the effect of a starvation episode at the larval stage (42, 55).

These findings support and extend the general concept that macrophages can act as metabolic sensors that integrate environmental signals to contribute to organismal homeostasis (22, 26). More precisely, the results from this study shows that macrophages represent a step in a complex and highly regulated interorgan communication process that allows for fine-tuning of developmental timing. The role of macrophages appears to be to delay the timing of the developmental program when the nutritional status of the larvae is poor, as assessed by InR/dTor signaling, but early/increased pvf2 expression does not initiate premature Pvr-mediated pupariation. In contrast, the timing of Pvr-mediated pupariation is determined by expression of Pvr in the prothoracic gland, which is up-regulated just before the onset of normal pupariation, while pvr early/overexpression induces early pupariation and strongly decreases the size of the adult flies. Moreover, in the absence of Pvf2 or Pvr, pupariation will eventually proceed, in a Pvr-independent manner, at the cost of a reduced size of the flies.

Macrophages assess nutritional status via InR/Tor signaling

Our results suggest that macrophages sense the nutritional status of the larvae via insulin/TOR signaling. This is consistent with seminal studies that have shown that the insulin/TOR signaling network transduces nutritional signals that regulate cell growth and developmental timing at the organismal level (14, 17, 67, 68). DILP synthesis and secretion by insulin-producing cells, like insulin in mammals (51, 69), are stimulated directly and indirectly by fat body–secreted factors such as Unpaired 2, Eiger, and Stunted that are decreased in starvation conditions (Fig. 5J) (70–73). DILPs circulate in the hemolymph and are thus available to macrophages. In addition, dTor, as well as the mammalian TOR complex, is activated by InR signaling and also directly by nutrients, specifically amino acids (52). The InR also signals independently of Tor by inhibiting Foxo (51), but our results indicate that FOXO overexpression does not affect the timing of pupariation (see fig. S7D), while dTOR is both necessary and sufficient for Pvf2 production by macrophages and for pupariation. We did not observe an effect of the knockdown of individual nuclear receptors in macrophages (see fig. S3), suggesting that they do not regulate Pvf2 production. Therefore, it is likely that insulin/TOR signaling in hemocytes is activated, at least, in part, by the availability of DILPs and amino acids; nevertheless, it is very possible that macrophages integrate additional yet unidentified signals to control pvf2 transcription and steroid hormone biosynthesis. On the other hand, InR/TOR signaling in the prothoracic gland itself also controls the timing of metamorphosis (17, 42, 68), indicating that this critical pathway collects and integrates information from different tissues to regulate developmental timing at the organismal level.

Bioavailability of Pvfs in the prothoracic gland

Drosophila Pvr has three ligands coded by separate genes—pvf1, pvf2 and pvf3—with partially redundant functions in cell survival, migration, and proliferation during embryonical stages (50, 58, 74, 75). Our results suggest that Pvf2, but not Pvf1 and Pvf3, is required to activate Pvr in the prothoracic gland to regulate ecdysone synthesis and developmental timing under physiological conditions. Among possible explanations, we show here that pvf2 expression is 10- to 100-fold higher in larval macrophages as compared to pvf1 and pvf3 (fig. S2C). In addition, there are 13 isoforms of Pvr (https://flybase.org/); therefore, it is possible that the prothoracic gland expresses Pvr(s) isoform(s) with extracellular ligand binding domain specific for Pvf2. It is also possible that larval macrophages that produce Pvf2 are in close contact with the prothoracic gland or that Pvf2 produced by hemocytes circulates in hemolymph and thus accesses the prothoracic gland more readily than Pvfs produced by other tissues. In addition, we cannot eliminate the possibility of unknown biochemical differences between Pvfs that could affect, for example, their solubility in the hemolymph, of the existence of putative coreceptor(s). Last, although Pvf2 is required, the contribution of the other Pvfs cannot be ruled out. It was shown that Pvr can regulate ecdysone synthesis in the prothoracic gland in an autocrine manner via Pvf2 and Pvf3, which are both expressed in the prothoracic gland at late stages during development (57), and a local source of Pvf3 may thus contribute to early pupariation of pvf2 mutant overexpressing pvr.

Macrophages license steroid hormone biosynthesis in the endocrine gland

Steroid hormone production in the endocrine gland of the fly and metamorphosis are under the control of the GnRH superfamily hormone Crz and Asta/AstaR via the PTTH produced by the PTTH neurons (4, 7, 11–13). A notable finding from our study is that the production of Pvf2 by macrophages contributes orthogonally to steroidogenesis by promoting transcription of the Halloween steroid biosynthesis genes when a satisfactory nutritional status is detected (see Fig. 5J). The development and organismal homeostasis of Drosophila and mammals are highly conserved (76), and the production of steroid hormones in mammalian puberty is under the control of the hypothalamus GnRH via the anterior pituitary hormones LH and FSH (2, 3, 6), which are similar to the Crz-Ptth axis (13) although LH and FSH are not Ptth orthologs. Although an orthogonal role of macrophages in the regulation of puberty is yet unknown, it is well documented that macrophage-deficient mice present with steroid hormone deficiency and infertility (77–84), and several mouse Pvfs orthologs (PDGFs) are also important for steroid production and male and female fertility (85).

Drosophila husbandry

Plasmatocyte-specific reporter line and driver hmlΔ-gal4;UAS-2xeGFP stock is a gift from S. Sinenko (86). UAS-pvf2 is a gift from B. Lemaitre (61). srp^hemo-gal4 is a gift from N. Perrimon. ras85D- KK108029-IR is a gift from M. Dominguez. pvf2^tra-gal4 is a gift from T. K. Johnson. hmlΔ-dsred (III). spi-TRiP.HMS01120-RNAi, gbb-TRiP.HMS01243-RNAi, dpp-TRiP.JF01090-RNAi, pvf1-TRiP.HMS01958-RNAi, pvf2-TRiP.HMJ23540-RNAi, pvf3-TRiP.HMS01876-RNAi, dilp6-TRiP.HMS00549-RNAi, spz-TRiP.HM05024-RNAi, upd2-TRiP.HMS00948-RNAi, upd3-TRiP.HM05061-RNAi, akh-TRiP.HMS00477-RNAi, idgf2-TRiP.HMC04223-RNAi, idgf6-TRiP.HMC04127-RNAi, pvr-TRiP.HMS01662-RNAi, hr96-TRiP.JF02350, eip78c-TRiP.JF02258, eip75b-TRiP.HMS01530, hr39-TRiP.JF02432, hr3-TRiP.HMC03156, hnf4-TRiP.JF02539, usp-TRiP.HMS01620, hr78-TRiP.JF03424, ecr-TRiP.HMJ22371, err-TRiP.HMC03087, hr4-TRiP.HM05260, svp-TRiP.JF03105, inr-TRiP.HMS03166, elav-gal4, btl-gal4, phm-gal4, ppl-gal4, repo-gal4, he-gal4, UAS-reaper, UAS-inrDN, dtor-TRiP.HMS00904, dtor-TRiP.HMS01114, UAS-pvr, Pvf2c06947, Pvr1, UAS-pvf1.RB, and pvf2d02444 lines are from the Bloomington Stock Center at Indiana University (Bloomington, IN). pvf1-KK112211-RNAi, pvf1-GD3432-RNAi, pvf2-KK110608-RNAi, pvf2-GD966-RNAi, pvf3-KK112796-RNAi, and pvf3-GD5238-RNAi are from Vienna Drosophila Resource Center. Flies were reared in standard molasses fly food from Archon scientific at 25°C (except when indicated) on a 12-hour:12-hour light:dark cycle.

Targeted RNAi screen for genes encoding secreted factors and nutrient-sensing receptors in larval macrophages

Genes involve in tissue growth and developmental timing were obtained from RNA-seq analysis of embryonic macrophages from (49). The expression of genes in larvae macrophages was determined by RT-qPCR of sorted macrophages at 96 hours AEL, and their role in developmental timing was tested using RNAi lines from Bloomington Stock Center at Indiana University (Bloomington, IN) listed above and the hmlΔ-gal4;UAS-2xeGFP flies.

Egg-laying and pupariation formation measurements

We crossed 20 females and males, and, after 48 hours, the parental flies were transferred to grape agar plates (Nutri-fly grape agar premix from Genesee Scientific) with yeast paste and left 3 to 4 hours for egg deposition at 25°C. The flies were removed, and laid eggs were incubated 48 hours at 25°C. Forty-eight hours later, the second-instar larvae were transferred onto Drosophila molasses food (20 larvae per tube) and reared at 26.5°C. We check the transition between larva-to-pupa at 8-hour intervals, with “time 0” designated as 4 hours after the initiation of egg laying and referred to as “AEL.” Median time of pupariation was used to analyze differences between genotypes (50% of larvae that became pupae).

Critical weight and starvation experiments

To determine the length of the pre-CW phase, we transferred the larva at 48 hours AEL into petri dishes with Drosophila molasses food (20 larvae per dish), and we allowed the larva to develop. At the different time points indicated (72, 76, 80, 84, 88, 92, and 96 hours AEL; fig S1A), we transferred the larva to empty petri dishes with a wet paper in the cover to keep larvae hydrated and avoid desiccation to starve the larvae until they enter in pupariation (87). We counted pupariation formation in 4-hour intervals.

On the basis of these results, all starvation experiments were performed by transferring 88 hours AEL pre-CW larvae to empty petri dishes with a wet paper in the cover to keep larvae hydrated and avoid desiccation (87). After 8 hours, the larvae were refed by transferring then to Drosophila molasses food tubes with 5 ml, and we counted pupariation formation in 8-hour intervals.

Pupae volume measurements

Twenty females and 20 males were crossed and left 24 hours for egg deposition. Parental flies were transferred every 24 hours to fresh tubes, and laid eggs were reared at 25°C. Pupae were collected and photographed with their dorsal side up. Length and width were measured using ImageJ; volume was calculated according to the following formula: v = 4/3π(L/2)(l/2)2 (L, length; and l, width) (88). Leica M80 stereo microscope was used for pupal imagine.

Adult wing area measurements

Twenty females and 20 males were crossed and left 24 hours for egg deposition. Parental flies were transferred every 24 hours to fresh tubes, and laid eggs were reared at 25°C. Adults were collected, and left wings of each individual were excised and rinsed thoroughly with ethanol and mounted in a glycerol-ethanol solution. Wing areas were measured using ImageJ (89). Pictures were taken using an Axio Lab.A1 Zeiss microscope with N-Achroplan 5×/01.25 objective.

Whole-larva mounting and confocal imaging

Mild-late L3 larvae were rinsed with tap water (no longer than 5 min), dried carefully, and anaesthetized using diethyl ether for less than 3.5 min. Larvae were positioning within one cover slide and a slide for mounting immobilizing the sample and imagine. We set up the tile scan with the micrometer-thick Z stack on LSM880 Zeiss microscope with 5× objective, which allows to see the sessile macrophages in the cuticle. Images were acquired with a resolution of 1024 × 1024.

Immunohistochemistry in ring-gland tissue

Brain–ring gland complex was dissected out in cold phosphate-buffered saline (PBS) buffer and fixed in 4% paraformaldehyde for 20 min (90). Brain–ring gland complex was stained overnight at 4°C with chicken anti-GFP (1/500; Life Technologies), and rat anti-Pvr (91) (1/500; a gift from B. Shilo). Secondary antibodies 488 anti-chicken (1/500) and 647 anti-rat (1/500) were purchased from Invitrogen. For DNA and nuclei staining, we used Hoechst (1/1000) for 10 min at room temperature. We set up the tile scan with the micrometer-thick Z stack on an LSM880 Zeiss microscope with 40× oil objective. Images were acquired with a resolution of 1024 × 1024.

Flow cytometry analysis of macrophages

Pvf2 transcriptional reporter

Hemocytes from 10 to 15 hmlΔ-dsred;pvf2^tra>mcd8::gfp larvae at 96 hours AEL in fed condition and starved condition (8 hours starved from 88 hours AEL) were bled and analyzed as above. Percentage (%) of live DsRed positive and GFP-positive hemocytes (DAPI negative), using in on an ARIA III (BD Biosciences), was determined by flow cytometry using Flow Jo 9.9 as above.

May-Grunwald–Giemsa staining

Analysis of macrophages was done on of sorted cells from GFP-positive hemocytes population, using HmlΔ>eGFP. One thousand cells were sorted directly in fetal bovine serum (Invitrogen) and centrifuged for 10 min at 800g with a low acceleration onto Superfrost slides (Thermo Fisher Scientific). After air drying for at least 30 min, the slides were fixed in methanol, air dried for at least 30 min, stained with May-Grunwald (Sigma-Aldrich) solution for 5 to 15 min and then Giemsa (Sigma-Aldrich) solution 14% for 15 to 30 min, and rinsed with Sorenson buffer (pH 6.8). After air drying, the slides were mounted with Entellan (Merck). Pictures were taken using an Axio Lab.A1 Zeiss microscope with N-Achroplan 100×/01.25 objective.

qRT-PCR on sorted macrophages and whole larvae

For qRT-PCR, 20,000 hemocytes were sorted into 600 μl of TRIzol LS (Life Technologies). RNA extraction was performed following the manufacturer’s instructions (miRNeasy Mini Kit, Qiagen). RNA concentration was measured with NanoDrop 2000. cDNA preparation was performed with the Invitrogen SuperScript IV First-Strand Synthesis System (Thermo Fisher Scientific) as per the manufacturer’s instructions. qRT-PCR was done with 5 to 10 ng of cDNA. For whole larvae, qRT-PCR tissue homogenization was performed with a tissue lyser (Precellys Evolution, Bertin Technologies) in 2 ml of tissue-homogenizing mixed beads (Bertin Technologies, ref. P000918-LYSK0-A) of five larvae per sample, and RNA extraction was performed following the manufacturer’s instructions (RNeasy Mini Kit, Qiagen). For whole larva, the same protocol as for the sorted cells and the qRT-PCR was performed on 500 ng of cDNA. qRT-PCR was performed using QuantStudio 6 Flex Real-Time PCR from Life Technologies. qRT-PCR is performed on a QuantStudio 6 Flex using TaqMan Fast Advance Mastermix and TaqMan probes for rp49 (Dm02151827_g1), dib (Dm01843083_g1), phm (Dm01844265_g1), nvd (Dm03419116_m1), sad (Dm02139323_g1), spo (Dm01840221_s1), eip75b (Dm01793666_m1), idgf2 (Dm01842858_g1), idgf6 (Dm01842553_g1), gbb (Dm01843010_s1), dpp (Dm01842959_m1), akh (Dm01822073_g1), spi (Dm02364865_s1), spz (Dm02151534_g1), spz2 (Dm01832735_g1), spz3 (Dm01809417_m1), spz4 (Dm01810735_g1), spz5 (Dm01835260_m1), spz6 (Dm01834239_g1), dilp6 (Dm01829746_g1), upd2 (Dm01844134_g1), upd3 (Dm01844142_g1), pvf2 (Dm01814370_m1), pvf1 (Dm01814376_m1), pvf3 (Dm01813949_m1), pvr (Dm01803625_m1), hr96 (Dm02151377_m1), eip78c (Dm01798025_g1), hr39 (Dm01811265_g1), hr3 (Dm01846760_m1), hnf4 (Dm01803764_g1), usp (Dm01841709_s1), hr78 (Dm01798077_g1), ecr (Dm01811604_g1), err (Dm01835887_g1), hr4 (Dm01798975_g1), hr38 (Dm01842602_g1), hr83 (Dm02137429_g1), hr51 (Dm01824332_m1), svp (Dm02135706_m1), dsf (Dm01842632_m1), tll (Dm02151861_g1), inr (Dm02136218_g1), dtor (Dm01843300_g1), and 4e-bp (Dm01842928_g1).

Ecdysteroid measurement

For ecdysteroid extraction, 10 whole larvae from each condition (from 96 or 104 hours AEL, as indicate in the figure) were collected in 2 ml of tissue-homogenizing mixed beads (Bertin Technologies, ref. P000918-LYSK0-A) and frozen in liquid nitrogen and transfer to −80°C. Samples were homogenized using a tissue lyser (Precellys Evolution, Bertin Technologies) in 0.3 ml of methanol, centrifuged at maximum speed for 5 min at room temperature and the supernatant was transferred to another eppendorf tube. We repeat this procedure by adding 0.3 ml of methanol to the pellet and mix using vortex and a third round using 0.3 ml of ethanol. The pooled sample after homogenization was 0.9 ml per sample. For quantification, the extracted samples were centrifuged at maximum speed for 5 min (room temperature) to remove any remaining debris and divided into two tubes to generate technical replicates. The cleared samples were evaporated overnight (92). The following steps were performed using the 20-HE enzyme-linked immunosorbent assay kit instructions (Bertin Pharma, no. A05120; 96 wells). Absorbance was measured at 410 nm with a microplate reader (SpectraMax i3 from Molecular devices).

20-HE treatment

Larvae were collected as explain above and transferred onto Drosophila molasses at 48 hours AEL. Control larvae (hmlΔ>gfp) and mutant conditions (hmlΔ>pvf2-IR and hmlΔ>rpr) were transferred at 80 hours AEL to fresh food supplemented with 20-HE (0.5 mg/ml; Sigma-Aldrich, H5142) diluted in ethanol. We used ethanol as a control vehicle.

Single-cell gene expression visualization by dot plots

Dot plots were generated using a user-friendly searchable web tool (www.flyrnai.org/scRNA/blood/) where genes can be queried and visualized in the different clusters made by single-cell RNA-seq (63).

Statistical analyses

All statistical analyses were performed using GraphPad Prism Software 9.0 with a 95% confidence limit (P < 0.05). An unpaired t test was used for comparisons between two genotypes or time points.

This PDF file includes: