Characterization of the Drosophila Adult Hematopoietic System Reveals a Rare Cell Population With Differentiation and Proliferation Potential

While many studies have described Drosophila embryonic and larval blood cells, the hematopoietic system of the imago remains poorly characterized and conflicting data have been published concerning adult hematopoiesis. Using a combination of blood cell markers, we show that the adult hematopoietic system is essentially composed of a few distinct mature blood cell types. In addition, our transcriptomics results indicate that adult and larval blood cells have both common and specific features and it appears that adult hemocytes reactivate many genes expressed in embryonic blood cells. Interestingly, we identify a small set of blood cells that does not express differentiation markers but rather maintains the expression of the progenitor marker domeMeso. Yet, we show that these cells are derived from the posterior signaling center, a specialized population of cells present in the larval lymph gland, rather than from larval blood cell progenitors, and that their maintenance depends on the EBF transcription factor Collier. Furthermore, while these cells are normally quiescent, we find that some of them can differentiate and proliferate in response to bacterial infection. In sum, our results indicate that adult flies harbor a small population of specialized cells with limited hematopoietic potential and further support the idea that no substantial hematopoiesis takes place during adulthood.


INTRODUCTION
Several aspects of blood cell development and functions are conserved between mammals and Drosophila (Hartenstein, 2006;Banerjee et al., 2019). Hence, this insect has been extensively used as a simple genetic model organism to study the fundamental bases underlying hematopoiesis and blood cell functions (Letourneau et al., 2016;Boulet et al., 2018). Yet, while much effort has been devoted to the characterization of Drosophila blood cells at the embryonic and larval stages, the adult stage remains less well characterized and conflicting data have been published (see below). Thus, further work is essential to gain a better understanding of the composition and dynamics of the adult hematopoietic system.
As in vertebrates, Drosophila hematopoiesis occurs in successive waves and stems from mesoderm-derived blood cell progenitors (called prohemocytes) (Banerjee et al., 2019). These progenitors can give rise to three main differentiated cell types, collectively called hemocytes and related to vertebrate myeloid cells: plasmatocytes, crystal cells, and lamellocytes (Gold and Bruckner, 2015;Parsons and Foley, 2016). Plasmatocytes are macrophages that form the bulk of the population; they are essentially implicated in tissue remodeling and in the cellular immune response, while crystal cells are implicated in clotting and melanization (an insect-specific immune response). Lamellocytes are normally barely present but their differentiation is massively induced in the larvae in response to pathological situations, such as infestation by parasitoid wasp eggs.
Plasmatocytes and crystal cells are first produced in the embryo after a small group of prohemocytes emerges from the early head mesoderm and differentiates (Tepass et al., 1994;Lebestky et al., 2000;Bataille et al., 2005). These embryo-derived hemocytes compose the larval peripheral blood cells, which are observed either as free circulatory cells in the hemolymph or attached under the cuticle (Holz et al., 2003;Makhijani et al., 2011). Interestingly, differentiated larval plasmatocytes can not only proliferate but also transdifferentiate into crystal cells (Makhijani et al., 2011;Leitao and Sucena, 2015), and it was proposed that the increase in peripheral blood cells during larval development relies on self-renewing plasmatocytes (Gold and Bruckner, 2015). Yet, undifferentiated blood cells were also described among peripheral hemocytes (Sinenko et al., 2010) and recent single-cell sequencing experiments suggest that proliferative plasmatocytes retain a progenitor signature (Cattenoz et al., 2020(Cattenoz et al., , 2021Tattikota et al., 2020). In parallel, a second wave of hematopoiesis takes place during the larval stages in a specialized organ called the lymph gland (Lanot et al., 2001). The lymph gland precursors derive from the embryonic lateral mesoderm and develop in close association with the anterior part of the dorsal vessel (Crozatier et al., 2004;Mandal et al., 2004). In third instar larvae, the lymph gland is composed of three to four pairs of lobes separated by pericardial cells (Lanot et al., 2001;Rodrigues et al., 2021). The posterior lobes are essentially composed of prohemocytes (Rodrigues et al., 2021), while the anterior lobes contain blood cells progenitors, their differentiated progenies (plasmatocytes and crystal cells), and a small cluster of cells that form a niche, called the posterior signaling center (PSC) (Jung et al., 2005;Letourneau et al., 2016). In normal situations, lymph gland blood cells are released into circulation during pupation when the lymph gland disperses (Honti et al., 2010;Grigorian et al., 2011).
By comparison, the adult hematopoietic system of the fly remains poorly characterized (Banerjee et al., 2019). Initial examinations failed to reveal the presence of an hematopoietic organ but showed that hemocytes are scattered throughout the imago, mostly as sessile populations, and accumulate along the heart in the abdomen (Elrod-Erickson et al., 2000;Lanot et al., 2001), a general feature among insects (Yan and Hillyer, 2020). In addition, it seemed that all the hemocytes were non-dividing phagocytic cells whose number and phagocytic activity decrease with age (Lanot et al., 2001;Mackenzie et al., 2011;Woodcock et al., 2015). Besides, transplantation experiments showed that both embryonic and larval hematopoietic anlagen contribute to the adult blood cells (Holz et al., 2003). Hence, the prevailing view was that the adult hematopoietic system was solely composed of mature plasmatocytes derived from the embryonic and larval stages, and that no hematopoiesis took place during adulthood. Yet, some data suggested a more contrasted situation. First, the analysis of different markers revealed that crystal cells and distinct subpopulations of plasmatocytes are present in the adult (Kurucz et al., 2007;Clark et al., 2011). Second, a seminal study by Ghosh et al. (2015) suggested that hematopoiesis also occurs in the imago. Indeed, the authors proposed that blood cell progenitors capable of differentiation persist in the adult and they observed plasmatocyte proliferation in response to Escherichia coli infection (Ghosh et al., 2015). However, these findings were refuted by a subsequent publication (Sanchez Bosch et al., 2019), in which the authors disproved the claim that the GATA factor Serpent (Srp) is a marker of adult blood cell progenitors and found no evidence for blood cell proliferation or differentiation.
Here, we set out to better characterize the Drosophila adult hematopoietic system. First, we established the gene expression profile of adult hemocytes and compared it to their larval ascendant (larval peripheral hemocytes and lymph gland) to define the common and specific features of these immune cells. Next, we used a panel of hematopoietic markers to assess their expression in the imago and to gain a better appreciation of the adult blood cell landscape diversity and evolution with age. Finally, we focused our analysis on a small population of adult blood cells that does not express hemocyte differentiation markers. We show that they originate from the PSC rather than from larval blood cell progenitors and that they can proliferate and differentiate in response to infection albeit unfrequently. These PSC-derived cells most probably account for the previously reported population of adult blood cell progenitor. Our findings are discussed in view of the current controversies in the field.

Adult Blood Cells Preparations
To bleed adult flies, aged-matched mated individuals were anesthetized, washed in 70% ethanol and air dried before cutting the last abdominal segment with a clean scalpel. Then a fine glass needle was inserted in the anterior part of the thorax and PBS was perfused under air pressure. Flushed hemocytes were collected in a 12-well plate (Nunc) containing 600 µl PBS and a round glass coverslip. In general, six drops of PBS (∼60 µl) were collected per fly and five flies were bled in a single well. The plates were centrifuged at 1000 rpm for 2 min before adding 300 µl of 16% formaldehyde for 30 min. For in situ observation of blood cells in adult abdomen, flies were dissected essentially as described in Ghosh et al. (2015). Briefly: flies were anesthetized, stuck on their dorsal side in a paraffin-coated plastic dish and dissected in PBS. The wings and most of the thorax were discarded and the ventral abdomen was incised on both sides with thin scissors before delicately removing the gonads and the gut. The dissected abdomens were fixed in 4% formaldehyde for 30 min. Samples were then processed as described below for immunofluorescence and/or confocal imaging.

Bacterial and Wasp Infections
Escherichia coli (DH5α strain) were grown overnight in LB broth and pelleted at 4000 rpm for 15 min. A fine tungsten needle was dipped in the bacterial pellet and inserted into the dorsal abdomen or under the wing hinge of anesthetized 1week-old females. A heat-sterilized tungsten needle was used to prick the flies that were used as non-infected control. Three groups of 10 flies were used in each experiment. The flies were then reared for 16 h (FUCCI experiments) to 96 h (G-TRACE experiments) and dissected or bled as described above. For wasp infestations, late second instar larvae were subjected to parasitism by ∼10 Leptopilina boulardi (strain G486) females for 2 h, as described previously (Benmimoun et al., 2015). Successful infestation was assessed by looking for melanotic nodule 48 h later. The corresponding larvae were transferred to fresh vial and the surviving adult flies were bled 5 days after emergence.

EdU Incorporation Assays
For EdU incorporation assays, 2-day-old domeMeso-GAL4, UAS-EYFP females were grown on 2 ml of fly food medium supplemented with 0.4 mM EdU (5-ethynyl-2 -deoxyuridine) for 5 days before infection with E. coli as described above. After 24 h on medium containing fresh EdU, infected and control flies were bled and EdU incorporation in hemocytes was assessed using Click-IT Plus EdU kit with AlexaFluor 555 dye according to the manufacturer's instructions (Molecular Probes #C10638).

RNA-Sequencing
For RNA-seq experiments, wandering third instar larvae or 4-to 5-day-old adult w 1118 females were used to prepare independent biological triplicates. Adult blood cells were retrieved by perfusing the flies with PBS as described above and directly collected in a 1.5 ml Eppendorf tube on ice. Peripheral larval blood cells were collected by delicately peeling the dorsal cuticle with forceps and dripping the larva in a 5 µl drop of PBS on parafilm. Upon microscopic examination, we found that >90% of the cells retrieved with this protocol were hemocytes (Pxn-RedStinger + ), with no visible contamination by fat body cells. The anterior lobes of the larval lymph glands were dissected in PBS. Blood cells and lymph glands were transferred to 1.5 ml Eppendorf tubes, pelleted by centrifugation at 1000 rpm for 2 min and processed for RNA extraction using Arcturus PicoPure RNA kit (ThermoFisher #KIT0204). For each replicate, ∼100 adult flies and 20 larvae were used. RNA samples were run on Agilent Bioanalyzer to assess sample quality and concentration. Samples were converted to cDNA using Nugen's Ovation RNA-Seq System (Catalog # 7102-A01). Libraries were generated using Kapa Biosystems library preparation kit (#KK8201) and multiplexed libraries were sequenced on a 1x75 flow cell on the HiSeq2000 device (Illumina). Reads were filtered and trimmed to remove adapter-derived or low-quality bases using Trimmomatic and checked again with FASTQC. Illumina reads were aligned to Drosophila reference genome (dm6 Ensembl release 70) with Hisat2. Read counts were generated for each annotated gene using HTSeq-Count. RPKM (Reads Per Kilobase of exon per Megabase of library size) values were calculated using Cufflinks. Reads normalization, variance estimation and pair-wise differential expression analysis with multiple testing correction was conducted using the R Bioconductor DESeq2 package. Heatmaps and hierarchical clustering were generated with "pheatmap" R package. Gene ontology enrichment analyses were performed using Genomatix. The RNA-seq data were deposited on GEO under the accession number GSE174107.

Real-Time Quantitative Polymerase Chain Reaction
For RT-quantitative polymerase chain reaction (qPCR), RNA samples were prepared from adult or larval bleeds and larval lymph glands (dissected as described above) using RNeasy kit (Qiagen) with an additional on-column DNAse treatment with RNase-Free DNase Set (Qiagen). Reverse transcription was performed with SuperScript IV Reverse Transcriptase (ThermoFisher) according to manufacturer's instruction on 100 ng of RNA and using a mix (1:1) of random primers (Invitrogen) and oligo dT (Promega). qPCRs were performed with SsoFast EvaGreen (Biorad) on a LightCycler 480 Instrument II (Roche Life Science). The sequences of the primers used to assess the expression of the different genes are provided in Supplementary Table 1. qPCR data were analyzed with Ct method and gene expressions were normalized to rp49 (synonym: RpL32). All experiments were performed using biological triplicates.

Adult Blood Cells Exhibit a Distinct Gene Expression Profile as Compared to Larval Blood Cells
As a first step to characterize Drosophila adult hematopoietic system, we used a perfusion protocol similar to the one used to bleed mosquitoes (Muller et al., 1999) in order to retrieve adult blood cells and define their gene expression program (see the section "Materials and Methods"). We found that ∼99% of the cells collected with this technique in 5-day-old flies expressed the pan-hemocyte marker srp as revealed by using the srpD-GAL4 driver (Waltzer et al., 2002) or the srpHemo-H2A::3mCherry reporter (Gyoergy et al., 2018;Figures 1A,C,E), indicating that this approach allows to obtain clean blood cell preparations. Of note, when using young adults (24-48 h after eclosion), the bleeds were contaminated with remnants of the larval fat body ( Figure 1B, arrow), which is known to dissociate into large individual adipose cells during metamorphosis (Nelliot et al., 2006) and gets cleared in the first 2 days of adulthood (Aguila et al., 2007). Using aged flies (35-day-old), we found that ∼4.6% of the cells did not express srp and we recovered fewer hemocytes (Figures 1D-F), which is consistent with previous reports showing that blood cell number decreases with aging (Mackenzie et al., 2011;Woodcock et al., 2015;Sanchez Bosch et al., 2019).
Next, we established the gene expression profile of adult hemocytes collected from 5-day-old flies by RNA sequencing. For comparison, we also established the transcriptome of third instar larval peripheral hemocytes (i.e., embryo-derived hemocytes) and lymph gland hemocytes. All experiments were performed in biological triplicates and at least 14 million mapped reads per sample. We found that 6802 genes were expressed with a RPKM > 1 in all three samples of adult hemocytes (corresponding to 39% of the genes in Drosophila reference genome dm6) (Supplementary Table 2). Similarly, larval peripheral blood cells and lymph gland hemocytes expressed 6282 and 6220 genes (Supplementary Tables 3, 4), respectively. As shown in Figure 2A, adult hemocytes (AD), lymph gland (LG) and larval peripheral hemocytes (PH) shared the expression of 5351 genes, i.e., between 79 and 86% of their respective transcriptome. Among the 268 genes annotated in Flybase (FB_2021_02) as being expressed in blood cells (i.e., hemocytes, plasmatocytes, crystal cells, lamellocytes or prohemocytes, Supplementary Table 5), 216 were retrieved in our adult hemocyte data set (2.1-fold over-enrichment, p-value < 5.5E-45), 197 in the lymph gland (2.1-fold overenrichment, p-value < 3.5E-37), and 198 in peripheral hemocytes (2.0-fold over-enrichment, p-value < 3.4E-37), with 188 being common to all three samples (Supplementary Figure 1). Thus, most hemocyte markers expressed in larval blood cells are also expressed in the adult. Notably, 11 of these hemocyte markers (Col4a1/Cg25C, Ppn, Nplp-2, Pxn, Fer2LHC, Fer1HCH, PPO2, PPO1, crq, Sr-Cl and Glt) were among the 60-most strongly expressed genes in adult blood cells (Supplementary Table 2). In contrast, gcm and gcm2, which were shown to be expressed in embryonic but not in larval blood cells (Avet-Rochex et al., 2010;Bazzi et al., 2018;Ramond et al., 2020), were not detected in adult (or larval) hemocyte RNA-seq samples (Supplementary Tables 2-4).
Despite expressing many genes in common, adult and larval hemocytes also exhibit distinct features. Indeed, principal component analysis showed that 77% of gene expression variance among samples was accounted by stage differences (adult versus larva) ( Figure 2B). Furthermore, differential gene expression analysis with DESeq2 revealed that 1074 genes were differentially expressed between adult blood cells and both larval peripheral hemocytes and lymph glands (adjusted p-value < 0.01 and fold change > 2), with 864 genes up-regulated and 210 downregulated in adult hemocytes (Figures 2C,D and Supplementary  Table 6). For comparison, 330 genes were differentially expressed between larval peripheral hemocytes and both adult blood cells and lymph glands (151 up-regulated, 179 down-regulated), and 218 genes were differentially expressed between lymph glands and both adult and larval peripheral hemocytes (99 up-regulated, 119 down-regulated). For instance, consistent with previous immunostaining results (Kurucz et al., 2007), we observed that Hemese (He) expression was down-regulated in adult hemocytes as compared to larval ones (Supplementary Table 6). In addition, to validate these results, we performed RT-qPCR on several genes that were differentially expressed between adult and larval blood cells using independent RNA samples. Thereby, we confirmed that hsp23, Tig, mthl6 and CG15115 were downregulated in adult hemocytes while ninaD, Fhos, Tps1, Cyp12a4, Ald and Men were up-regulated ( Figure 2E). Furthermore, reporter lines for NimC4 and Mp also showed that these two genes were expressed in adult but not in larval hemocytes (Figures 2F,G).
Gene Ontology enrichment analyses underlined the upregulation of genes implicated in small molecule metabolism processes, including ATP metabolism (as seen with the overexpression of several enzymes involved in glycolysis, TCA cycle, or oxidative phosphorylation) and a down-regulation of genes regulating chromatin organization in adult hemocytes ( Table 1). Along the same line, Gene Ontology analysis considering the 1000 most strongly expressed genes in larval, lymph gland or adult blood cells revealed shared features between them, with for instance a strong over-representation for genes implicated in translation, vesicle-mediated transport or immune response, but also confirmed the singular enrichment for ATP metabolic processes among highly expressed genes in adult hemocytes ( Table 2).
Surprisingly, one of the top-enriched genes in adult hemocytes is NimC4, which was recently shown to be highly enriched in stage 16 embryonic hemocytes as compared to larval descendants (i.e., larval peripheral hemocytes) in a comparative transcriptomic analysis (Cattenoz et al., 2020). Moreover, reminiscent of the situation that we observed, embryonic and their larval progenies were found to have distinct metabolic gene signature. We thus reanalyzed the RNA-seq data from Cattenoz et al. (2020) using the same settings as for our transcriptomes. Thereby, we established a list of 4892 differentially expressed genes between embryonic and peripheral hemocytes (adjusted p-value < 0.01 and fold change > 2; Supplementary Table 7) that we compared with our results. Strikingly, 49% (422/864) of the genes up-regulated in adult versus larval hemocytes are also over-expressed in embryonic blood cells as compared to larval blood cells (2.9-fold over-enrichment, p < 4.8E-113) (Figure 2H). Similarly, 46% (97/210) of the genes downregulated in adult versus larval hemocytes are also repressed in embryonic blood cells (4.2-fold over-enrichment, p < 1.3E-38) ( Figure 2H). The common over-expressed genes in embryonic and adult hemocytes versus larval ones were implicated in small molecule/ATP metabolic processes, while down-regulated genes were involved in cell cycle (Table 3). Hence, many genes and processes differentially regulated between adult and larval hemocytes are similarly regulated between embryonic and larval hemocytes.
In sum, these data show that adult hemocytes share a significant portion of their transcriptome with larval blood cells but also exhibit distinct features, some of which appear to be shared with embryonic hemocytes.

Characterization of the Adult Hematopoietic System Landscape
To better define the cellular composition of the adult hematopoietic system, we then analyzed the expression of a series of well-characterized reporters classically used to study embryonic and larval blood cell types (see the section "Materials and Methods").
First, we assessed the expression of plasmatocyte and crystal cell differentiation markers. In 5-day-old flies, the plasmatocyte reporters for Peroxidasin (Pxn), Collagen 25C (Cg25C, also known as Collagen type IV alpha1), Hemolectin (Hml), and croquemort (crq) were expressed in ∼95%, 90%, 65%, and 25% of the blood cells, respectively, while those for the crystal cell markers Black cells (Bc, also known as Prophenoloxidase 1, PPO1) and lozenge (lz) were expressed in 8 to 12% of the cells (Figures 3A-E Supplementary Figures 2A-F). The proportion of hemocytes expressing these reporters was similar in 35-day-old flies, except for a slight but significant decline in the proportion of Pxn + cells ( Figure 3G). The analysis of double transgenic lines revealed that (almost) all the cells expressing Hml co-expressed srp and Pxn (Figures 3H-H",K-K",Q). Besides, ∼90 of the crq + cells were Hml + (Figures 3O-O",Q). Among the Bc + cells, 100% expressed srp and Pxn, ∼90% Cg25C and 60% Hml (Figures 3I-I",L-L",N-N",P-P",Q). Of note, it was previously observed that the "plasmatocyte" markers Cg25C, Pxn and Hml are also expressed in larval crystal cells (Jung et al., 2005). We did not observe significant shifts in the repartition of these populations between 5-and 35-day-old flies, indicating that the overall composition of the adult hematopoietic system remains stable during aging ( Figure 3Q).

,G and
Next, we tested whether lamellocytes could be retrieved in the adult. Intriguingly, we found that msnF9 reporter, whose activity is induced in lamellocytes in the larvae (Tokusumi et al., 2009b), was expressed in ∼92% of the adult hemocytes (Supplementary Figure 2G). However, we did not observe any cell with high actin content and/or expressing the β-integrin Myospheroid  (Mys), two characteristic features of lamellocytes (Irving et al., 2005; Supplementary Figures 3C,I), suggesting that lamellocytes are not present and that the msnF9 reporter is constitutively active in adult blood cells. Since lamellocytes are absent in healthy larvae but massively induced in response to infection by eggs of the parasitoid wasp Leptopilina boulardi (Lanot et al., 2001), we assessed their presence in adult flies that survived wasp infestation. Still, β-integrin and phalloidin staining did not reveal the presence of lamellocytes in infected adults, while they were readily recovered in bleeds from infected larvae (Supplementary Figures 3B,D,H,J). Furthermore, when using flies carrying the gain-of-function HopT Tum−l , which exhibit constitutive activation of the JAK kinase Hopscotch leading to lamellocyte production in the larvae (Luo et al., 1995), we did not observe lamellocytes in the adult either ( Supplementary  Figures 3E,F,K,L). Thus, it appears that lamellocytes do not persist during adulthood. Finally, to reveal the presence of putative blood cell progenitors, we made use of tepIV-GAL4, Ance-GFP, dome PG125 -GAL4 and domeMeso-GAL4, all of which have been shown to label larval prohemocytes in the lymph gland (Krzemien et al., 2007;Avet-Rochex et al., 2010;Louradour et al., 2017;  et al., 2020; Rodrigues et al., 2021) or in peripheral blood cells (Fu et al., 2020). While we did not detect any expression of tepIV-GAL4, Ance-GFP or dome PG125 -GAL4 in adult hemocytes (Supplementary Figures 2H-J), we found that domeMeso-GAL4 was expressed in ∼4% of cells in both 5-and 35-day-old flies ( Figures 3F,G). domeMeso was expressed in cells exhibiting a characteristic fusiform shape, with one or two filopodial extension, but not in round hemocytes (Figures 3F,J-J"' ,M-M"'). Importantly, none of the fusiform cells expressed the blood cell differentiation markers tested above (Pxn, Cg25C, Hml, crq, Bc, lz and msn), but they expressed srp, albeit at lower level than the other blood cells (Figures 3H"'-P"' and Supplementary Figures 4A-H). Also, immunostaining against the plasmatocyte marker P1/NimC1 as well as in situ hybridization against Pxn or against the crystal cell marker Bc/PPO1 confirmed that domeMeso + /fusiform cells did not express these blood cell differentiation markers (Supplementary Figures 4I-K), strongly suggesting that they could be blood cell progenitors.
In sum, our data indicate that the adult hematopoietic system is essentially composed of plasmatocytes (∼85%) and crystal cells (8 to 10%) and could harbor a small population of undifferentiated cells (∼4%). Furthermore, as schematized in Figure 3R, adult blood cells can be classified into different subcategories (depending on the expression of Pxn, Hml, crq, Bc and domeMeso) whose proportions are relatively stable between young and old flies.

Abdominal Hematopoietic Hubs Have a Similar Composition as Adult Bleeds
Since adult bleeds may not accurately reflect the composition of the adult hematopoietic system, we also assessed the expression of some of the above reporters in the dorsal abdomen. As shown in Figures 4A-G,A'-G' srp, Pxn, Cg25C, Hml, crq, Bc, or domeMeso expression was observed as dispersed cells in the abdomen, especially in segments A1 to A3 along the heart, in the socalled "hematopoietic hubs." srp expression was also prominent in pericardial cells (Figure 4A, arrows). In the hematopoietic hubs, Hml + cells represented 65% of the hemocytes as judged by co-labeling with srp or Pxn and no Hml + srp − or cells Hml + Pxn − were observed (Figures 4H-H",K-K",P-P"). Crystal cells (Bc + ) and putative prohemocytes (domeMeso + ) represented ∼10% and ∼3% of the hemocytes, respectively (Figures 4I-I",J-J",P-P"). 90% of the crystal cells expressed Cg25C and 55% of them expressed Hml (Figures 4M-M",O-O",P-P"). All domeMeso + cells expressed srp at low levels and none expressed Pxn (Figures 4J-J",L-L",P-P"). Similar observations were made when analyzing the expression of these markers in 35-dayold flies and we did not observe significant differences in the proportion of each blood cell populations between 5-day-and 35-day-old flies ( Figure 4P). These results are consistent with our observations on adult bleeds and indicate that the adult hematopoietic system is composed of different subpopulations of plasmatocytes and crystal cells as well as a few putative blood cell progenitors.

Hemolectin Is Expressed in a Sub-Population of Adult Hemocytes
The Hml driver has been used in several instances to manipulate gene expression in adult blood cells and/or test the function of these cells, notably by generating "hemoless" adults through the expression of an apoptosis inducer in blood cells (Charroux and Royet, 2009;Defaye et al., 2009;Ayyaz et al., 2015;Ghosh et al., 2015;Woodcock et al., 2015;Sanchez Bosch et al., 2019). Yet, our results suggest that this driver is not active in every adult hemocyte but rather labels a specific subpopulation. To confirm this hypothesis, we expressed the apoptotic inducer Bax under the control of Hml either during the whole development or only from day 1 of adulthood by using a tubGAL-80 ts transgene to restrict its expression temporally, and we assessed the presence of hemocytes in 10-day-old adult flies. Hml-driven expression of Bax specifically during adulthood caused an almost total ablation of Hml + blood cell population as compared to control flies (Figures 5A-A"' ,C-C"'), but Pxn + cells were still present in adult bleeds and abdominal preparations (Figures 5B-B"' ,D-D"'). When Bax was expressed from embryogenesis onward, we observed an efficient ablation of Hml + cells and very few Pxn + blood cells in adult abdominal hubs but, in contrast with the normal situation (Figures 5E-E"' ,G-G"'), we retrieved many Pxn − cells when these individuals were bled (Figures 5F-F"' ,H-H"'). While some of these Pxn − cells were fusiform, as in the control, most had the typical round morphology of differentiated hemocytes, suggesting that the chronic expression of Bax in Hml + cells leads to the repression of Pxn. Consistent with this hypothesis, immunostaining against P1/NimC1 showed  and Supplementary Figure 5). These results show that killing Hml + blood cells is not sufficient to generate "hemoless" flies and might alter the gene expression program of the remaining blood cells. These data also indicate that during adulthood Hml + Pxn + and Hml − Pxn + cells form two distinct populations and that fusiform cells are not derived from Hml + cells.

domeMeso + "Undifferentiated Hemocytes" Are Not Derived From Larval Blood Cell Progenitors but From the Lymph Gland Posterior Signaling Center
A previous report showed that a small fraction of cells expressing Srp did not express the blood cell differentiation markers P1/NimC1 (for plasmatocytes) or Hindsight (for crystal cells), suggesting that they could be hematopoietic progenitors (Ghosh et al., 2015). In addition, these cells were proposed to derive from the posterior lobes of the larval lymph gland based on lineage tracing experiments using the col-GAL4 driver GMR13A11. While col-GAL4-derived cells from the larval lymph gland were shown to contribute to adult plasmatocytes and crystal cells (Ghosh et al., 2015), it remains to demonstrate that these cells can proliferate and/or differentiate during adulthood and thus constitute a genuine adult hematopoietic progenitor population. Our results above strongly suggested that fusiform/domeMeso + cells could correspond to this putative hematopoietic progenitor population. To test this hypothesis, we first assessed their origin. Using the G-TRACE cell lineage tracing technique  with either GMR13A11 or the classically used pcol85 col-GAL4 line, we found that all fusiform cells belonged to the col lineage-traced cells (Figures 6A,D), indicating that they correspond to the previously described adult blood cell progenitors. Interestingly, we found that these cells maintain col driver expression ( Figures 6A-A",D-D"), while we only observed past/lineage-traced expression in round hemocytes ( Figures 6B-B",E-E"). This result was confirmed by immunostaining, which revealed that Col was expressed in fusiform/domeMeso + cells, but not in other adult blood cells (Figures 6G-G",H-H" and Supplementary Figures 6A,B). Furthermore, in accordance with the above characterization of domeMeso + cells, col + cells were Pxn − both in adult bleeds and in abdominal hematopoietic hubs (Figures 6J-J",K-K").
These results are thus consistent with the hypothesis that col + progenitors could give rise to differentiated hemocytes. Importantly though, the col-GAL4 lines pcol85 and GMR13A11 are not only expressed in blood cell progenitors present in the lymph gland posterior lobes but also in the PSC (Benmimoun et al., 2012;Ghosh et al., 2015;Rodrigues et al., 2021). Moreover, recent single-cell RNA-seq analyses revealed that PSC-like cells expressing col and Antp are present in circulating larval blood cells (Cattenoz et al., 2021). Hence, these putative adult hematopoietic progenitors could have diverse origins. First, to test whether they were derived from the Quantifications of the proportion of domeMeso high /fusiform hemocytes (green) and domeMeso low /round hemocytes in tub-GAL80ts, domeMeso-GAL4, UAS2x-EYFP adult flies expressing (RNAi col) or not (ctr) col RNAi during adulthood. Means and standard deviations of at least three independent experiments are presented. A total of 2892 and 4489 hemocytes were scored in ctr and RNAi col flies, respectively. ***p < 0.001 (Student's t-test). (L-Q) Blood cells from tub-GAL80ts, domeMeso-GAL4, G-TRACE flies raised at permissive temperature (L,M) or switched from restrictive to permissive temperature 24 h after adult emergence (N-Q). domeMeso live expression is revealed by nuclear RFP (red) and its lineage-traced expression by nuclear GFP (green). P1/NimC1 expression (red) (Continued) FIGURE 7 | (Continued) was revealed by immunostaining. Nuclei were stained with DAPI. Scale bar: 20 µm. ctr, control non-infected flies; inf, flies infected with E. coli. (R) Quantifications of the proportion of domeMeso + cells expressing NimC1 in tub-GAL80ts, domeMeso-GAL4, G-TRACE flies raised at permissive temperature only during adulthood and infected (inf) or not (ctr) with E. coli. Means and standard deviations of at least seven independent experiments are presented. A total of 369 (ctr) and 449 (inf) domeMeso + cells were scored. ***p < 0.001 (Student's t-test). (S-Y) Representative images of the cell cycle status observed in fusiform or round hemocytes from srpD-GAL4, UAS-FUCCI control (S-U) or E. coli-infected (V-Y) adult flies. Cells in G1 phase express the nuclear GFP only, cells in G2/M express both nuclear GFP and RFP, cells in S phase express the nuclear RFP only. Cell morphology was visualized by phalloidin staining (green). Nuclei were stained with DAPI (blue). Scale bar 20 µm. (Z) Quantifications of the proportion of fusiform or round hemocytes in the different phase of the cell cycle in control (ctr) or in E. coli-infected (inf) srp>FUCCI adult flies. Mean values from at least six independent experiments are presented. A total of 646 (ctr) and 515 (inf) fusiform hemocytes as well as 842 (ctr) and 1336 (inf) round hemocytes were scored. embryo/peripheral hemocytes or from the lymph gland, we performed lineage-tracing experiments with sn-GAL4, which is specifically expressed in embryo-derived hemocytes (Avet-Rochex et al., 2010), and dot-GAL4, which is specifically expressed in lymph gland-derived cells (Honti et al., 2010). In line with the idea that adult blood cells are derived from both waves of hematopoiesis (Holz et al., 2003;Sanchez Bosch et al., Figures 6C,D), demonstrating that they are solely derived from the lymph gland. Next, we asked whether fusiform cells are derived from the lymph gland posterior lobes or the PSC using, respectively, Ubx(M3)-GAL4, which is expressed only in the posterior lobes (Rodrigues et al., 2021), and Antp-GAL4, which is expressed in the PSC but not in the posterior lobes Rodrigues et al., 2021). Importantly, we found that fusiform cells were lineage-traced with Antp-GAL4 (n = 29/38), while Ubx(M3)-GAL4 activity was traced in round hemocytes but not in fusiform cells (n = 0/83) (Figures 6R-W,R'-W' ,R"-W"). Furthermore, immunostaining revealed that the PSC marker Antp is specifically expressed in domeMeso + blood cells (Supplementary Figures 6E,F). In addition, as shown for PSC cells in the larval lymph gland (Tokusumi et al., 2010), the knock-down of srp in domeMeso + cells abrogated the formation of filopodial extensions typical of these cells (Supplementary Figures 6G,H). In sum, these results show that fusiform/col + /domeMeso + cells are derived from the PSC rather than from undifferentiated blood cells originating from the lymph gland posterior lobes.

domeMeso + Cells Can Differentiate and Proliferate
Since col is required for PSC development (Crozatier et al., 2004), we tested its role for fusiform cell development. In contrast with control flies, no domeMeso + cells (or fusiform cells) were present in col −/− adults (Figures 7A,B,D). In addition, RNAimediated knock-down of col specifically in domeMeso + cells was sufficient to ablate this lineage (Figures 7C,D). To test whether the maintenance of col expression in domeMeso + cells is required in the adult, we used a tub-GAL80 ts transgene to restrict col knock-down to adulthood. Following the inhibition of col expression for 48 h in adult flies almost no domeMeso + /fusiform cells were retrieved, yet we observed a few cells that maintained domeMeso expression at lower levels and adopted a round shape similar to differentiated hemocytes (Figures 7E-K).
Interestingly, in contrast with domMeso high /fusiform cells, these domeMeso low cells expressed the plasmatocyte marker P1/NimC1 (Figures 7H-J). Similarly, in situ hybridization revealed that unlike their normal counterparts, these cells also expressed the differentiation marker Pxn (Supplementary Figure 7). These results show that col controls cell-autonomously domeMeso + cells development and suggest that its expression in these cells is continuously required to prevent their differentiation.
In light of the above results, we used the G-TRACE system to test whether domeMeso + cells naturally differentiate during adulthood. Because domeMeso-GAL4 is expressed in blood cell progenitors in the larval lymph gland (Louradour et al., 2017;Rodrigues et al., 2021), we used a tubGAL80 ts transgene to temporally restrict its activity. When domeMeso-GAL4 activity was left unrestricted (i.e., when flies were raised at 29 • C from embryonic to adult stages), we observed lineage-traced expression of domeMeso in round cells (which expressed the plasmatocyte marker P1/NimC1) as well as live (and traced) expression in fusiform cells (Figures 7L-L",M-M"). When dome-GAL4 activity was restricted to the adult (i.e., when flies were raised at 18 • C until adulthood and then switched at 29 • C for up to 2 weeks), domeMeso>G-TRACE expression was observed in fusiform cells but not in other hemocytes, except for one cell from one out of seven experiments (Figures 7N-N",O-O",R and Supplementary Table 8). Strikingly though, we found that upon infection of adult flies by E. coli, in addition to the fusiform cells, some round/NimC1 + cells were recurrently lineage-traced with domeMeso ( Figures 7P-P",Q-Q",R, and Supplementary Table 8). Thus, while domeMeso + cells do not usually differentiate during adulthood under normal conditions, they can give rise to plasmatocytes in response to an immune challenge.
Finally, it was proposed that infection of adult flies with E. coli can induce Hml + cells proliferation at low frequency (Ghosh et al., 2015), but these results were refuted by Sanchez Bosch et al. (2019), who brought several lines of evidence that Hml + cells do not proliferate even in response to infection (Sanchez Bosch et al., 2019). Yet, Hml is not expressed in all blood cells and notably not in domeMeso + /fusiform cells. We thus reassessed the cell cycle status of adult blood cells using srpD-GAL4, which is expressed in all the hemocytes, together with the two-color FUCCI cell cycle indicator (Zielke et al., 2014). We found that 97% of the round hemocytes were in G1 phase and 3% in G2/M but none in S phase, while 73% of the fusiform cells were in G1, 27% in G2/M (Figures 7S-S",T-T",U-U",Z and Supplementary  Table 9). Of note except for one fusiform cell in one experiment out of six, we did not observe cells in S phase. These results strongly suggest that blood cell do not proliferate in uninfected adults. Following infection with E. coli, the proportion of round or fusiform cells in G1 or G2/M remained similar (Figures 7V-V Table 9), suggesting that they maintain their proliferating potential during aging. In addition, when we used EdU incorporation to assess domeMeso + cell proliferation, a few domeMeso + cells (n = 8/370) incorporated EdU following infection with E. coli, while none were observed in control flies (n = 0/225) (Supplementary Figure 8). These results thus indicate that domeMeso + /fusiform cells can proliferate, albeit at low frequency, in response to an immune challenge.

DISCUSSION
In mammals, the life-long production of the different blood cell types relies on the presence of long-term hematopoietic stem and progenitor cells, which are specified in the embryo and reside in specific niches from where they can be mobilized to respond to the needs of the organism (Cool and Forsberg, 2019). While several aspects of hematopoiesis are conserved from mammals to Drosophila (Banerjee et al., 2019), the presence of a similar long-term blood cell progenitor population in Drosophila remains elusive and the overall composition of the adult hematopoietic system is still a matter of debate. In particular, two successive publications came to different conclusions concerning the presence of hematopoietic progenitors and the mere existence of hematopoiesis in adult flies (Ghosh et al., 2015;Sanchez Bosch et al., 2019). Overall, our data indicate that the Drosophila adult hematopoietic system is essentially composed of different subpopulations of mature blood cells, including PSC-derived cells which seem to have some limited hematopoietic capacity, and concur with the hypothesis that no significant hematopoiesis takes place during adulthood.
Our results show that the bleeding protocol that we used ensures a swift and representative collection of the blood cell types present in the adult with minimal contamination from other tissues. This simple protocol is thus suitable to define the adult hemocyte gene repertoire without applying tissue dissociation and cell-purification protocols, which might alter gene expression. Thereby, we could show that adult hemocytes share an important part of their transcriptome with their larval parents, but also exhibit some clearly distinct features.
Notably several blood cell markers are differentially expressed between adult and larval hemocytes. For instance, NimC4 and drpr, which work together for apoptotic cell clearance in the embryo (Kurant et al., 2008), or Fer1HCH and Fer2LHC, which encode the two components of the iron transporter ferritin and control larval blood cell differentiation (Yoon et al., 2017), are strongly up-regulated in adult hemocytes. Moreover, our findings highlight an unexpected degree of similarities between adult and embryonic blood cells as compared to larval blood cells. Indeed, the comparison of our results with the recently described embryo versus larval blood cell gene repertoire (Cattenoz et al., 2020) revealed that almost half of the genes differentially expressed in adult versus larval hemocytes are similarly regulated in embryonic versus larval blood cells. This is particularly striking for adult up-regulated genes as 422 (out of 864) of them are also more strongly expressed in embryonic hemocytes, indicating that adult blood cells reactivate a large number of embryonic blood cell markers. Among those, many are involved in energy metabolism. Actually embryo-derived hemocytes undergo a shift from glycolysis during embryogenesis to lipid β-oxidation during the larval stages (Cattenoz et al., 2020) and our results strongly argue that a second metabolic shift occurs during adulthood. The important role of energy metabolism regulation in Drosophila blood cells is underscored by recent publications showing that fatty acid β-oxidation is required for blood cell progenitor differentiation in the larval lymph gland (Tiwari et al., 2020) and that infection leads to changes in blood cell metabolism which are important for an effective immune response (Bajgar et al., 2015;Krejcova et al., 2019). Prominent changes in energy metabolism have been observed also in mammalian blood cells in response to development, aging, infection or cancer (Rashkovan and Ferrando, 2019;Faas and de Vos, 2020;Nakamura-Ishizu et al., 2020). Getting further insights into the pathways controlling these shifts will be thus of great interest. More generally, the differences between embryonic, larval and adult blood cell gene expression programs likely reflect distinct biological functions which warrant further investigation.
Our analysis of different blood cell differentiation markers indicate that the adult hematopoietic system comprises different subtypes of plasmatocytes as well as ∼10% of crystal cells, but no lamellocytes. These finding are consistent with previous reports (Lanot et al., 2001;Kurucz et al., 2007;Ghosh et al., 2015;Sanchez Bosch et al., 2019), and underscore the heterogeneity of the plasmatocyte/crystal cell populations, which can be subdivided according to the expression of Hml, crq and, to a lesser extent, Cg25C. It will be interesting to analyze whether these subpopulations have distinct functions for instance in the immune response or the removal of apoptotic cells, as recently shown for embryonic macrophage subpopulations (Coates et al., 2021). The proportion of these subpopulations was similar between young and aged flies, suggesting that they do not represent differentiation intermediates inherited from pupal hemocytes. Along that line, our cell ablation experiments argue against the idea that all adult hemocytes are derived from Hml + cells or that adult Hml + Pxn + cells differentiate into Hml − /Pxn + cells during adulthood. Besides, our observations are in line with previous reports showing that adult blood cells do not proliferate (Lanot et al., 2001;Sanchez Bosch et al., 2019) and that their number decreases with aging (Mackenzie et al., 2011;Woodcock et al., 2015;Sanchez Bosch et al., 2019). Therefore, it seems that the adult hematopoietic system is essentially composed of mature blood cells. Ghosh et al. (2015) observed some Srp + blood cells that did not express the differentiation markers NimC1 or Hnt and were lineage-traced with col-GAL4. Although the differentiation and proliferative potentials of these cells during adulthood were not assessed, the authors concluded that blood cell progenitors derived from the lymph gland posterior lobes persist during adulthood. Our results contradict this model. Indeed, while we also observed the presence of a small proportion of srp + cells that do not express hematopoietic differentiation markers and are lineage-traced as originating from the lymph gland, we demonstrated that these cells actually derive from the PSC rather than from the posterior lobes. Actually, they exhibit a characteristic fusiform shape with filopodial extensions reminiscent of PSC cells (Krzemien et al., 2007;Mandal et al., 2007), and they express Col, which is required for PSC development but also for their maintenance in the adult. Moreover, although these cells express domeMeso, they do not express other prohemocytes markers (tep4, Ance or dome-GAL4), and we did not find evidence that they differentiate during adulthood in normal conditions. Nonetheless these cells hold some hematopoietic potential. Notably, upon E. coli infection, some of them can enter S phase and can differentiate into plasmatocytes as evidenced by NimC1 induction and morphological changes. Along the same line, infection of larvae with E. coli was shown to induce PSC cells proliferation and to affect their morphology (Khadilkar et al., 2017). However, given the low number of PSC-derived cells and the limited response we observed in infected adults, it is unlikely that de novo production of mature blood cells is their main function. In the lymph gland, the PSC forms a highly specialized cluster of cells that does not contribute to the general pool of blood cells but controls their fate, in particular in response to immune challenges (Crozatier et al., 2004;Benmimoun et al., 2015). It is tempting to speculate that these PSC-derived cells may have similar functions in the adult. As the knock-down of col resulted in the ablation of the PSC-derived cells and the resulting adults were viable and did not show any obvious phenotype, it would be interesting to challenge these flies with different pathogens to try to shed light on the physiological role of these cells.
In sum, our results strongly suggest that the Drosophila adult hematopoietic system does not harbor a true blood cell progenitor population with significant proliferation and differentiation potential. Still, different types of mature, potentially specialized, blood cells are present during adulthood. Along that line it will be interesting to decipher the function of the PSC-derived population, as these cells have been shown to control the cellular immune response to some specific immune challenges in the larva and may play a similar role in the adult. Finally, it is expected that applying single cell RNA-sequencing approaches to the adult hematopoietic system will bring a better description of its constituents and could reveal some of its functional features.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm. nih.gov/geo/, GSE174107.

AUTHOR CONTRIBUTIONS
MB, FL, BB, and LW performed the experiments. YR performed bioinformatics analyses. MB, FL, YR, LV, and LW analyzed the data. MB, FL, YR, BB, LV, and LW designed the study. LW wrote the manuscript. All authors read, and approved the submitted version.

FUNDING
This work was supported by grants from the Agence Nationale de la Recherche (ANR 13-BSV3-08 and 17-CE12-0030-03), i-Site CAP20-25 and the Indo-French Centre for the Promotion of Advanced Research (IFCPAR/CEFIPRA 5503-1) to LW. MB was supported by Ph.D. fellowships from the Université Clermont-Auvergne and the Fondation pour la Recherche Médicale (FRM).

ACKNOWLEDGMENTS
We thank the Toulouse RIO and Clermont-Ferrand CLIC platforms for assistance with confocal imaging. We thank all our former colleagues for valuable discussions and the Drosophila community, the Drosophila Bloomington Stock Center and the Developmental Study Hybridoma Bank (DSHB) for fly stocks and antibodies.