Dual role for Insulin/TOR signaling in the control of hematopoietic progenitor maintenance in Drosophila

Thanks to the phylogenetic conservation of the genetic pathways and developmental strategies at stake during blood cell development, Drosophila has emerged as a valuable model organism to study the fundamental mechanisms underlying hematopoiesis (Hartenstein, 2006). Notably, the Drosophila larval hematopoietic organ, the lymph gland, provides a paradigm to reveal the relationships between the hematopoietic niche, the blood cell progenitors and their differentiated progenies (for a review, see Krzemien et al., 2010a). The lymph gland produces three mature blood cell types functionally related to mammalian myeloid cells: the plasmatocytes, the crystal cells and, upon some immune challenges, the lamellocytes (Lanot et al., 2001; Meister and Lagueux, 2003). During the two first instar larval stages, the lymph gland grows and is primarily composed of stem cell-like progenitor blood cells called prohemocytes (Crozatier et al., 2004; Jung et al., 2005; Krzemien et al., 2010b; Minakhina and Steward, 2010). In third instar larvae, the mature lymph gland consists of a pair of primary lobes and several smaller secondary lobes (Fig. 1A). The primary lobes are organized in three zones: the cortical zone (CZ), the medullary zone (MZ) and the posterior signaling center (PSC) (Jung et al., 2005). The MZ as well as the secondary lobes comprise quiescent prohemocytes whereas the CZ contains differentiated hemocytes that emerge from the MZ. Reminiscent of the mammalian hematopoietic system, the balance between stem cell-like and differentiated blood cells is controlled by the PSC, which acts as a niche and secretes diffusible molecules promoting prohemocyte maintenance (Krzemien et al., 2007; Mandal et al., 2007; Sinenko et al., 2009). However, only a handful of signaling pathways controlling niche development or hematopoietic progenitor maintenance are known.

An important challenge for every metazoan is to adjust its development to food availability, and growing evidence indicates that nutrition can influence stem/progenitor cell fate (Amcheslavsky et al., 2009; Chell and Brand, 2010; McLeod et al., 2010; Sousa-Nunes et al., 2011). The interconnected Insulin/IGF signaling (IIS) and Target of Rapamycin (TOR) pathways (Fig. 1B) are the integrators between tissue growth and dietary conditions (Tennessen and Thummel, 2011). Here, we show that these pathways play a crucial role during hematopoiesis in the Drosophila lymph gland. First, they regulate the size and activity of the hematopoietic niche. Second, they cell-autonomously control blood cell progenitor maintenance. Finally, they couple blood cell development to the larval nutritional status.

The following fly stocks were used: Pcol85-Gal4, UAS::mCD8-GFP (Krzemien et al., 2007) referred here as col-Gal4, ppl-Gal4/CyOGal80; slif^Anti, ppl-Gal4/CyO (Colombani et al., 2003), PTEN¹⁰⁰/CyO Act-GFP, PTEN¹¹⁷/CyO Act-GFP (Oldham et al., 2002), UAS::TSC1, UAS::TSC2 (Tapon et al., 2001), UAS::PTEN (Huang et al., 1999), UAS::PI3K^caax/Cyo (Leevers et al., 1996), UAS::dFoxo (Puig et al., 2003), tepIV-Gal4 (Avet-Rochex et al., 2010), Cg25C-GFP and Bc-GFP (Sorrentino et al., 2007), Hh-GFP (Tokusumi et al., 2011). The RNAi stocks were obtained from Vienna Drosophila RNAi Center (VDRC).

Embryos were collected for 6 hours and equivalent numbers of hatching first instar larvae were transferred to vials with regular food medium at 25°C until wandering third instar larval stage. Starvation experiments were performed as established by Ikeya et al. (Ikeya et al., 2002). Larvae from crosses of UAS or Gal4 lines to w¹¹¹⁸ were used as controls.

Lymph glands were processed and stained as described (Avet-Rochex et al., 2010). The following primary antibodies were used: anti-Antp, anti-Hnt (Developmental Studies Hybridoma Bank), anti-Col (Crozatier et al., 2004), anti-P1/NimrodC1 (Vilmos et al., 2004), anti-H3P (Upstate), anti-β-Gal (Cappel), anti-GFP (Torrey Pines). For in situ hybridization, DIG-labeled antisense RNA probes against TepIV, Ance and α-PS4 were used. Samples were visualized with a laser scanning confocal microscope (Leica). Confocal images are displayed as maximum intensity projections of the lymph glands.

For quantification, lymph glands were scanned at an optimized number of slices using a Leica SP5 microscope. Crystal cell and plasmatocyte differentiation indexes were defined as the number of Hnt⁺ cells (quantified using Volocity software) or the level of Cg25C-GFP expression (measured with Image J software), respectively, in each primary lobe reported to the estimated lobe’s volume. PSC cell number and MZ or lobe size were quantified using Image J software. A minimum of 12 lobes was scored per genotype.

To test whether the IIS pathway controls lymph gland homeostasis, we used a combination of Pten mutant alleles for which third instar larvae can be obtained (Oldham et al., 2002). Pten codes for a protein phosphatase that antagonizes IIS-induced activation of the Phosphoinositol-3-Kinase (PI3K) (Fig. 1B). Hence, Pten larvae exhibit increased IIS activity. In contrast to wild-type lymph glands, which exhibited strong TepIV and Ance expression in the MZ of their primary lobes and in their secondary lobes, Pten lymph glands displayed markedly reduced expression of these two prohemocyte markers (Fig. 1C-F; supplementary material Fig. S1). Conversely, immunostaining against the crystal cell marker Hindsight (Hnt; Peb – FlyBase) (supplementary material Fig. S1) or the plasmatocyte marker P1, showed that whereas differentiated hemocytes are normally restricted to the CZ, they filled Pten lymph glands primary lobes and were also present in the secondary lobes (Fig. 1E-H), which is a characteristic sign of precocious differentiation (Owusu-Ansah and Banerjee, 2009). In addition, in situ hybridization against α-PS4 revealed the presence of lamellocytes, which are seldom observed in the normal situation (supplementary material Fig. S2). Furthermore, Pten lymph glands frequently exhibited premature primary lobe dispersal and displayed overgrown primary and secondary lobes (supplementary material Fig. S5). Consistent with this observation, anti-phospho-H3 labeling showed that Pten lymph glands contained numerous proliferating cells (Fig. 1G,H). Thus, as in mammals, in which Pten loss induces hematopoietic stem cell exhaustion and myeloproliferative disease (Yilmaz et al., 2006; Zhang et al., 2006), Pten is required to prevent the aberrant differentiation of the blood cell progenitors into the three mature blood cell types found in Drosophila.

To ensure that the massive differentiation observed in Pten mutants was not due to a defect in PSC cell specification, we monitored the expression of the two key regulators of PSC fate: Antennapedia (Antp) and Collier (Col; Kn – FlyBase) (Krzemien et al., 2007; Mandal et al., 2007). Although immunostaining against these two proteins demonstrated that PSC cells were correctly specified, we found that the size of the niche was strongly increased (Fig. 1I,J; data not shown), suggesting that IIS normally controls PSC cell number. As the presence of a larger niche was nonetheless associated with increased prohemocyte differentiation, we surmised that these contradictory phenotypes might arise from a dual activity of the IIS in the PSC versus the MZ. We thus assessed its role specifically in these two domains.

To test whether the IIS pathway is cell-autonomously required in the PSC, we used the col-Gal4 driver (Krzemien et al., 2007), expression of which is strictly confined to the PSC during lymph gland ontogeny, as demonstrated by a lineage tracing experiment (supplementary material Fig. S3). In addition, we took advantage of col-Gal4-driven expression in the wing disc to confirm the specificity of the UAS transgenes used in this study (supplementary material Fig. S4). As observed in Pten larvae, over-activation of IIS in the PSC, induced by expressing either Pten RNAi or an active form of PI3K (PI3K^caax) (Leevers et al., 1996), led to a strong increase in PSC size (Fig. 2B,C). This phenotype correlated with a rise in PSC cell number (Fig. 2G). Conversely, knocking down InR by RNAi or overexpressing Pten caused a reduction in PSC cell number (Fig. 2D,G). As IIS impinges on TOR activity (Fig. 1B), we tested whether this pathway also regulates PSC development. PSC cell number diminished when the TOR pathway was inactivated either by overexpressing both TSC1 and TSC2 (gig – FlyBase) or by downregulating raptor by RNAi (Fig. 2E,G; data not shown). Of note, TSC1/TSC2 overexpression seemed to reduce PSC cell size (Fig. 2E′). Conversely, TSC1 RNAi expression, which resulted in a larger PSC (Fig. 2F), did not significantly affect cell number but increased cell size (Fig. 2G,H). This suggests that TOR signaling not only supports PSC cell proliferation but also their growth. Finally we observed a strong drop in PSC cell number when we overexpressed Foxo (Fig. 2G), which is the main effector of IIS and whose targets are concomitantly regulated by the TOR kinase (Edgar, 2006). Together, these data indicate that IIS and TOR pathways are required in the PSC to promote niche cell proliferation/maintenance and growth.

Previous work showed that the PSC controls prohemocyte maintenance (Krzemien et al., 2007; Mandal et al., 2007). Notably, in contrast to what we observed in Pten larvae, increased PSC size has been associated with decreased differentiation (Mandal et al., 2007; Sinenko et al., 2009; Tokusumi et al., 2011). We thus assessed the effect of increased IIS/TOR signaling in the PSC on hemocyte differentiation. Although PI3K^caax overexpression or TSC1 RNAi increased PSC size by more than twofold, P1 and Hnt immunostaining did not reveal an obvious reduction in either plasmatocyte or crystal cell differentiation (Fig. 3A-F). To ensure that we did not overlook a subtle phenotype, we measured crystal cell or plasmatocyte differentiation index on a minimum of 20 samples (see Materials and methods). We found that over-activation of the IIS or TOR pathway in the PSC significantly inhibited the differentiation of both type of hemocytes (Fig. 3G,H). Moreover, expression of an Hh-GFP reporter line (Tokusumi et al., 2011) was observed in all the PSC cells expressing either PI3K^caax or TSC1 RNAi, suggesting that increased PSC size is associated with increased production of this inhibitor of prohemocyte differentiation (Mandal et al., 2007). Together, our results indicate that increased IIS/TOR signaling not only supports Drosophila hematopoietic niche growth but also promotes its activity.

Next, we modulated the activity of the IIS/TOR pathways specifically in the prohemocytes using the TepIV-Gal4 driver (Avet-Rochex et al., 2010). Reminiscent of the phenotypes observed in Pten larvae, over-activation of IIS by PI3K^caax expression induced a massive differentiation of plasmatocytes (labeled with either Cg25C-GFP or P1) and crystal cells (labeled with either Hnt or Bc-GFP) in the primary and secondary lobes, paralleled by a strong reduction in the pool of prohemocytes (labeled with either Ance or a UAS-mCherry) (Fig. 4B,H; supplementary material Fig. S6B,G), and an overgrowth of the secondary lobes (supplementary material Fig. S5). Unexpectedly, knocking down IIS by expressing InR RNAi also favored prohemocyte differentiation (Fig. 4C,H; supplementary material Fig. S6C,G). However, in contrast to IIS activation, IIS knockdown reduced lymph gland growth (supplementary material Fig. S5) and did not cause lamellocyte differentiation (supplementary material Fig. S2). Finally, lowering TOR activity using raptor RNAi promoted hemocyte differentiation and significantly reduced the proportion of prohemocytes present in the primary lobes (Fig. 4D,H; supplementary material Fig. S6D,G). Similar phenotypes were observed when IIS and TOR signaling were reduced in the MZ using the dome-Gal4 driver (data not shown). Thus, a fine-tuning of IIS/TOR pathways in the blood cell progenitors seems to be crucial to prevent their differentiation and to control lymph gland growth. Furthermore, as increased PSC size in Pten zygotic mutants does not prevent the massive differentiation of the prohemocytes, we propose that over-activation of the IIS/TOR pathways in the MZ bypasses the progenitor-promoting activity of the PSC.

One main function of the IIS pathway is to link tissue growth to nutrient sensing. Indeed, food shortage induces a reduction of insulin-like peptide (ILP) secretion by the insulin-producing cells (IPC) located in the brain (Edgar, 2006). The consecutive reduction in IIS signaling diminishes larval growth. The fat body acts as a nutrient sensor that restricts global growth through a humoral mechanism: upon diet restriction, ILP production by the IPC is reduced in response to the inactivation of the TOR pathway in the fat body. Accordingly, depletion of the amino acid transporter Slimfast (Slif) in the fat body disrupts TOR signaling and is sufficient to mimic diet restriction (Colombani et al., 2003). As IIS/TOR signaling is active in the lymph gland and controls prohemocyte maintenance, we investigated whether the larval nutritional status alters lymph gland homeostasis. When the larvae were starved for 24 hours, we observed a strong increase in plasmatocyte differentiation as revealed by Cg25C-GFP expression, and a concomitant decrease in Ance expression, indicative of a switch from prohemocyte maintenance towards differentiation (Fig. 4F,J). Unexpectedly, only a few crystal cells were present in the lymph glands of starved third instar larvae and we observed that they burst (Fig. 4J; data not shown). However, the examination of starved second instar larvae revealed the precocious differentiation of crystal cells associated with a decreased in the MZ (Fig. 4J′). Hence, starvation induces precocious differentiation of prohemocytes into both crystal cells and plasmatocytes. Of note, it is possible that starvation also directly affects differentiated hemocytes, notably crystal cells. We then tested whether food shortage was sensed directly by the prohemocytes or mediated by a systemic humoral response. To do so, we mimicked amino-acid deprivation by inhibiting Slif expression either in the prohemocytes or in the fat body, where its downregulation causes a systemic reduction in ILP levels (Colombani et al., 2003). Whereas expression of the RNA antisense slif^Anti in the prohemocytes did not affect their maintenance (supplementary material Fig. S6E,G), its expression with the fat body-specific pumpless-Gal4 (ppl-Gal4) driver caused massive plasmatocyte differentiation and decreased the prohemocyte pool (Fig. 4G,H,K; supplementary material Fig. S6F). Therefore, the systemic response to nutrient availability controls blood cell progenitor maintenance in the lymph gland.

Our results demonstrate that IIS/TOR signaling plays a dual role in the maintenance of the blood cell progenitors by acting both within the hematopoietic niche to control its size and its activity, and within the prohemocytes to control their fate. To gain a comprehensive view of IIS/TOR function in Drosophila hematopoiesis and in light of the recent report showing that differentiated hemocytes can feedback on prohemocyte maintenance (Mondal et al., 2011), it will be interesting to explore the role of these pathways in the differentiated blood cells. In addition, our data are consistent with a model whereby the IIS/TOR pathways link prohemocyte maintenance to the Drosophila larvae nutritional status. We speculate that food shortage, by sensitizing blood cell progenitors to differentiation, might affect the cellular immune response. Along this line, the rate of encapsulation of parasitoid wasp eggs, which relies primarily on the differentiation of lamellocytes, has been shown to diminish in larvae that were deprived of yeast before infestation (Vass and Nappi, 1998). We anticipate that future studies will allow further understanding of how developmental and environmental cues are integrated by IIS/TOR signaling to control blood cell homeostasis.

We thank H. Hafen, P. Léopold, J. Colombani, L. Pick, N. Tapon, M. Crozatier, I. Ando and Bloomington and Vienna Drosophila stock centers for flies and reagents. We thank B. Ronsin for developing the Volocity macro. We thank Toulouse RIO imaging platform and B. Augé for assistance with confocal, and members of our team for discussions.