We first validated that OCIAD1 interacts with the mitochondrial AAA⁺ proteases, by co-immunoprecipitation experiments on the human immortalized cell line HEK293T (Supplementary Figure S1A). To investigate the relevance of AAA⁺ proteases in hematopoiesis, we surveyed the Leukemia MILE (Microarray Innovations in Leukemia) datasets available in the BloodSpot database (https://www.fobinf.com/?gene=AFG3L2&dataset=all_mile;https://www.fobinf.com/?gene=YME1L1&dataset=all_mile) and found that both AFG3L2 and YME1L have elevated expressions across various subtypes of leukemia including AML (acute myeloid leukemia), ALL (acute lymphoblastic leukemia) and CML (chronic myelogenous leukemia), when compared to healthy bone marrow (Bagger et al., 2019; Haferlach et al., 2008). Further, HemaExplorer datasets (https://www.fobinf.com/?gene=AFG3L2&dataset=nl_human_data_HemaExp_v_1; https://www.fobinf.com/?gene=YME1L12&dataset=nl_human_data_HemaExp_v_1) showed that AAA⁺ protease expression was highest in the earliest blood stem and progenitor populations of the bone marrow compared to differentiated blood cells such as B cells and T cells (Bagger et al., 2019; Bagger et al., 2013) (Supplementary Figure S1B,C), suggesting AAA⁺ regulation may control hematopoietic homeostasis. Further, upon searching the Fly Hemocyte Atlas (http://big2.hanyang.ac.kr/flyscrna/), a single-cell RNA-sequencing database for Drosophila hemocytes (Cho et al., 2020) reveal that the expression of AFG3L2 (CG6512) and YME1L are higher in differentiated hemocytes like plasmatocytes and adipohemocytes compared to progenitors or pre-progenitors (Supplementary Figure S1D). Hence, we used the powerful and relevant model of Drosophila larval hematopoiesis, to investigate the possible roles of the proteases in hematopoiesis.

Knockdown of either afg3l2 or yme1l from progenitors [using dome-Gal4>UAS-GFP or domeMeso-Gal4>UAS-GFP or tep4-Gal4>UAS-GFP drivers], or from the niche [using pCol85-Gal4>UAS-GFP or antp-Gal4 drivers] resulted in over 95% lethality on YME1L depletion and about 50% lethality in AFG3L2 depleted larvae by the third instar stage (Supplementary Figures S2A–E). This suggested that AFG3L2 and YME1L function in blood cells is essential for larval health and survival.

Blood progenitors are also influenced by the status of differentiated hemocytes and systemic signals. Hemocyte-specific knockdown (KD), using hmlΔ-Gal4>UAS-GFP (AFG3L2 KD–hmlΔ-Gal4>UAS-GFP/+;UAS-AFG3L2 RNAi/+ or YME1L KD - hmlΔ-Gal4>UAS-GFP/+;UAS-YME1L RNAi/+), showed a ∼50% decrease in the pupariation rate of YME1L KD flies, suggesting impaired hemocyte function in metamorphosis (Supplementary Figure S2F). While there was no significant change in the eclosion rate of survivors (Supplementary Figure S2G), there was a mild yet significant increase in the life span of YME1L KD flies (median survival: control – 54 days, AFG3L2 KD – 52.5 days, YME1L KD – 59 days; Supplementary Figure S2H). To further validate the functional efficacy of the knockdown of the proteases in the hemocytes using hmlΔ-Gal4>UAS-GFP, aspects of mitochondrial morphology like mean and median branch lengths were probed in control and AAA + KD backgrounds, which revealed severely impaired mitochondrial morphology in the KD backgrounds (Supplementary Figure S3A–E) as previously described in other model organisms (Anand et al., 2014; Consolato et al., 2018).

To probe for effects of hemocyte-specific depletion of AAA⁺ proteases on larval hematopoiesis, we first examined the gross LG morphology in third instar larvae, by neutral red staining, as previously described (Rodrigues et al., 2021b). Though the hmlΔ-Gal4 driver is active only in differentiated hemocytes of the CZ and in circulation, we observed effects on all LG lobes, including posterior lobes that harbour only progenitors. Primary lobes were partially or completely absent, while posterior lobes exhibited severe hyperplasia with multiple tumorous structures (Figures 1A–D). We also found that the size of the posterior lobes was significantly higher in the AAA + KD LGs as quantified in Figure 1F. Total cell counts (DAPI⁺ nuclei) were significantly high in AFG3L2 KD and YME1L KD posterior lobes (Figures 1G–K. Although the number of H3P+ (mitotically active) proliferating cells were higher in the KD backgrounds, their increase was proportionate with the increase in total cell numbers (Figures 1G–J). Stage-specific analysis during larval development [ 60 h After Egg Laying (AEL) (second instar), 72 h AEL (early third instar), 96 h AEL (mid third instar), and 120 h AEL (late third instar)] showed relatively larger primary lobes in AAA ⁺ KD second instar and early third instar LGs with a concomitant increase in numbers of DAPI⁺ cells and Hml⁺ (Hemolectin) hemocytes. Subsequently, primary lobes histolysed and posterior lobes exhibited over-proliferation by mid third instar stage, accompanied with emergence of Hml⁺ hemocytes in the posterior lobes (Supplementary Figure S4A–D).

In summary, we show that AAA⁺ proteases in Drosophila blood progenitors and PSC are essential for survival. In parallel, these proteases regulate hematopoietic homeostasis through their functions in the differentiated hemocytes even though not negatively impacting the life span of AAA ⁺ KD flies. Higher number of H3P⁺ proliferating cells and Hml⁺ cells emerging in the posterior lobes of AAA ⁺ KD third instar LGs indicate proliferation and premature differentiation of the posterior lobe progenitors, which we have previously shown to be more resilient to differentiation than the primary lobe progenitors (Mandal et al., 2007).

During normal larval development under homeostatic conditions, the LG primary lobe progenitors differentiate to plasmatocytes and crystal cells, while the posterior lobe progenitors remain quiescent. Upon receiving specific cues including signaling pathway modulations and immune challenges, anterior progenitors differentiate into additional plasmatocytes and crystal cells, while lamellocytes are induced only in response to wasp infestation (Banerjee et al., 2019; Rodrigues et al., 2021a).

Remnants of the histolysed primary lobes of AAA ⁺ KD larval LGs showed no significant change in proportion of progenitors or hemocytes from the control samples (Figures 2A–H). However, there was a marked decrease in DomeMesoBFP⁺/PI (progenitor/total nuclei) proportions in the secondary lobes with an increase in the tertiary lobes in AAA + KD backgrounds. (Figures 2A–D; control–domeMesoBFP/+;hmlΔ-Gal4>UAS-GFP/+; AFG3L2 KD - domeMesoBFP/+;hmlΔ-Gal4>UAS-GFP/+;UAS-AFG3L2 RNAi/+, YME1L KD - domeMesoBFP/+;hmlΔ-Gal4>UAS-GFP/+;UAS-YME1L RNAi/+). Further, there was a dramatic increase in Hml⁺ hemocyte numbers in AAA ⁺ KD LG posterior lobes (Figures 2E–H), with more drastic effects on the secondary lobes compared to tertiary lobes. There was a minor increase in the Antp⁺ PSC cell numbers as well in the AAA + KD primary lobes (Figures 2I–L). We next examined the differentiation status of the expanded posterior lobes.

Expectedly, AAA ⁺ KD lymph glands had significantly elevated numbers of P1⁺ plasmatocytes, which are typically Hml⁺, and PPO⁺ (Prophenoloxidase) crystal cells in the posterior lobes, compared to controls (Figures 3A–H). Surprisingly, L1⁺phalloidin^high lamellocytes which emerge only in situations such as wasp infestations, were also present in the AAA ⁺ KD secondary lobes (Figures 3I–L, marked by white arrowheads). As observed previously, the differentiation phenotypes were more drastic in the secondary lobes compared to the tertiary lobes (Figures 3D,H,L). In circulation, the proportion of Hml⁺ cells among total DAPI⁺ cells was lower in AAA ⁺ KD samples, likely due to the increase in Hml⁻ or Hml^low hemocytes (Supplementary Figure S5A–D). P1+ plasmatocyte numbers were significantly lower in AFG3L2 KD samples, consistent with lower Hml + hemocyte numbers while that of YME1L KD samples were comparable to that of the control samples (Supplementary Figure S5E–H). Further, PPO⁺ crystal cells were increased in circulation in AAA⁺ KD larvae as observed in the AAA ⁺ KD LGs (Supplementary Figure S5I–L). Larger L1⁺phalloidin^high lamellocytes were also present among AAA ⁺ KD circulating hemocytes in significant numbers (Supplementary Figure S5M–P). The total number of circulating hemocytes however were unchanged in AFG3L2 KD larvae, and slightly decreased in YME1L KD larvae, indicating that the histolysed primary lobes in the AAA + KD backgrounds do not significantly contribute to the circulating hemocyte pool (Supplementary Figure S5Q). Taken together, our data demonstrate that depleting AAA⁺ proteases in differentiated hemocytes results in unbiased differentiation of the progenitor population in AAA ⁺ KD LGs, leading to increased numbers of all three types of hemocytes.

During hematopoiesis, cellular ROS levels regulate critical cell fate decisions, including differentiation, maintenance of quiescence and lineage bias, through regulation of signaling pathways (Filippi and Ghaffari, 2019; Anderson et al., 2017). As mentioned in the introduction, as both AFG3L2 and YME1L have been shown to regulate ROS levels, we hypothesized that AAA⁺ protease depletion from hemocytes may lead to increased ROS levels and thereby affect pathways involved in blood progenitor maintenance.

Probing for ROS levels in the circulating hemocytes using the cellular ROS sensitive dye, DHE (dihydroethidium), revealed higher DHE intensity in AAA ⁺ KD hemocytes (Supplementary Figure S6A–D). As previously reported, cells with lower Hml intensity had higher DHE intensity (Myers et al., 2018). Since ROS levels in AAA ⁺ KD hemocytes were higher, we explored the possibility of rescuing the AAA ⁺ KD LG phenotypes by reducing ROS levels. AAA ⁺ KD larvae grown in food containing 10 mg/mL NAC (N-acetyl cysteine), (Chen et al., 2023), a free radical scavenger that controls ROS levels by promoting glutathione production (Kelly, 1998), resulted in intact primary lobes in the AAA ⁺ KD larvae (AFG3L2 KD + NAC and YME1L KD + NAC) (Figures 4D,E’), although the DAPI⁺ cell numbers were elevated (Figure 4I). Further, there were higher number of Hml⁺ cells in the NAC-treated AAA ⁺ KD LG primary lobes compared to that of untreated control (Figures 4A–E,L,M’). This suggests that NAC treatment could rescue AAA ⁺ KD phenotypes of primary lobe disintegration, but not that of differentiation of the resident progenitors. Interestingly, there was complete rescue of the AAA ⁺ KD phenotype in the posterior lobes upon NAC-treatment, with DAPI⁺ cell numbers restored to control levels and absence of Hml⁺ hemocytes (Figures 4A,A’,D–E’).

To validate the specific role of ROS levels on AAA ⁺ KD phenotypes, we overexpressed the ROS scavenger catalase (Cat) in AAA ⁺ KD hemocytes. Compared to the AAA ⁺ KD lymph glands (Figures 4B,C'), AFG3L2 KD Cat OV (;hmlΔ-Gal4>UAS-GFP/UAS-Cat;UAS-AFG3L2 RNAi/+) and YME1L KD Cat OV (;hmlΔ-Gal4>UAS-GFP/UAS-Cat;UAS-YME1L RNAi/+) LGs (Figures 4G,H’) did not have primary lobe histolysis and were comparable to wild type (Figures 4A–C’,G,H’). Posterior lobes had significantly lower number of DAPI⁺ cells and no Hml⁺ cells. Further, PPO + crystal cell numbers were also rescued in AAA ⁺ KD LGs upon NAC treatment and catalase overexpression (Supplementary Figure S7A–K) Hence, pharmacological or genetic reduction of ROS levels in AAA ⁺ KD flies, indicate that ROS levels act as an important systemic effector downstream of AAA⁺ proteases in the regulation of Drosophila hematopoiesis.

Among other signaling pathways, ROS levels were shown to activate Drosophila Yorkie or its mammalian ortholog YAP1 (Yes-associated protein 1) in Drosophila enterocytes and human glioma cell lines (Dixit et al., 2014; Nagai et al., 2021), a molecule which has been shown to be directly upregulating tissue proliferation (Milton et al., 2014; Snigdha et al., 2019). Since, ROS levels emerged to be critical in the AAA ⁺ KD LG phenotypes, we hypothesized that the ROS-Yorkie axis may be a key mechanism by which AAA⁺ proteases mediate LG homeostasis.

Hemocyte-specific depletion of Yorkie (Yorkie KD - hmlΔ-Gal4>UAS-GFP/+;UAS-Yorkie RNAi/+) had no effect on LG progenitor differentiation compared to control (Figures 5A,A’,D,D’). Depleting Yorkie in AFG3L2 KD hemocytes (hmlΔ-Gal4>UAS-GFP/+;UAS-AFG3L2 RNAi/UAS-Yorkie RNAi) restored both DAPI⁺ cell number and hemocyte differentiation to control levels as indicated in Figures 5A–F and area analyses of the LG lobes of mentioned samples (Figures 5G–I). However, YME1L KD Yorkie KD (hmlΔ-Gal4>UAS-GFP/+;UAS-YME1L RNAi/UAS-Yorkie RNAi) was comparable to YME1L KD in DAPI + cell numbers and Hml + hemocytes (Figure 5). Taken together, our data show that Hippo signaling mediated by Yorkie is involved in AAA⁺ protease mediated regulation of LG maintenance.

Dome-Gal4,UAS-2xEGFP, Antp-Gal4 were gifts from Utpal Banerjee. tep4-Gal4,UAS-GFP was gifted by Lucas Waltzer. pCol85-Gal4,UASmCD8-GFP was gifted by Michele Crozatier, HmlΔGal4,UAS-2xEGFP, UAS-catalase (w (Mochizuki-Kashio et al., 2021); P{w[+mC] = UAS-Cat.A}2) (BDSC #24621), domeMeso-BFP and domeMeso-Gal4,UAS-GFP were gifts from Tina Mukherjee. UAS-Yorkie-RNAi (y (Mochizuki-Kashio et al., 2021) v (Mochizuki-Kashio et al., 2021); P{y[+t7.7] v[+t1.8] = TRiP.HMS00041}attP2) (BDSC #34067) was gifted by Carmen Coelho. UAS-AFG3L2-RNAi (y (Mochizuki-Kashio et al., 2021) sc[*] v (Mochizuki-Kashio et al., 2021) sev (König et al., 2016); P{y[+t7.7] v[+t1.8] = TRiP.HMS01331}attP2) (BDSC #34343) and UAS-YME1L-RNAi (y (Mochizuki-Kashio et al., 2021) v (Mochizuki-Kashio et al., 2021); P{y[+t7.7] v[+t1.8] = TRiP.HMC03303}attP2) (BDSC #51752) were obtained from Bloomington Drosophila Stock Center (BDSC). Canton-S was used as the wild type strain. The stocks were maintained under standard rearing conditions at 2 °C, unless specified otherwise.

RNAi knockdown was validated by qRT-PCR and mitochondrial morphology phenotype.

For NAC treatment, fly food was mixed with NAC (A9165, Sigma Aldrich) at a concentration of 10 mg/mL and parent crosses were cultured for examining emerging larvae at wandering third instar larval stage.

Lymph glands were dissected in PBS (phosphate buffered saline) and fixed using 4% PFA (– 158,127, Sigma Aldrich, India) (paraformaldehyde), followed by washing with PBS. The LGs were incubated in 0.2% neutral red solution (72,210-25G, Merck, Germany) in PBS for 5 min, after washing with PBS. After washing post-incubation, they were imaged in Olympus SZX12 stereozoom microscope (OLY-SZX12-B, Olympus, Japan) at 4X magnification for blinded quantification of LG morphology by two independent researchers to qualitatively classify the LG lobes of randomly numbered samples into three categories–regular (as observed under normal conditions), disintegrated (histolyzing and histolyzed lobes) and tumorous (abnormal lobes with tumorous growths) Representative images were acquired in Leica DMi8 (Germany) microscope.

For LG analyses, wandering third instar larvae (144 h AEL) were dissected in PBS, fixed in 4% PFA for 20 min, followed by washing in PBS. The LGs were then permeabilized using 0.3% Triton X-100 (X100-5ML, Merck, Germany) for 20 min and washed in PBS. After permeabilization, the LGs preps were incubated in 20% normal goat serum (RM10701,HiMedia, India) for 1 h. Primary antibody dilutions were then added to the samples and incubated overnight at 4 °C. Primary antibody incubation was followed by washes in PBS and incubation with appropriate secondary antibodies for 2 hours. Secondary antibody incubation was followed by extensive washes in PBS. The samples were then mounted in 70% glycerol (Q24505, ThermoFisher Scientific, India) containing 4′6-diamidino-2-phenylindole (DAPI) (D1306, ThermoFisher Scientific, United States) and imaged in Zeiss LSM880 confocal microscope at 20X magnification. To visualize the entire lymph gland, multiple images were captured from the anterior-posterior axis and stitched manually.

For circulating hemocyte immunostaining, individual larvae of each genotype were bled in 50 µL of PBS and allowed to adhere to coverslip bottomed 96 well plates (P96-15H-N, CellVis, United States) for 45 min, following which, the cells were fixed using 4% PFA for 15 min. Fixation was followed by washing and permeabilization with 0.4% NP40 (Nonidet P-40; 492016, Sigma Aldrich, India). The cells were then washed again and incubated with 2 mg/mL BSA (bovine serum albumin; MB083, Himedia, India) for 1 h. Blocking was followed by primary antibody incubation at 4 °C overnight. Primary antibody incubation was followed by washes in PBS and incubation with appropriate secondary antibodies for 2 hours. The cells were then incubated in a DAPI:PBS solution for 10 min, and washed. Imaging was performed in Zeiss LSM880 confocal microscope at 40X magnification. All antibody dilutions were made in PBS.

Mouse anti-Antennapedia (1:20, 4C3, DSHB), anti-P1 antibody (1:20) and anti-L1 abtibody (kind gifts from Istvan Ando) mouse anti-ProPO antibody (1:20, self-generated through Bioneeds India Private Limited), and rabbit anti-H3P antibody (1:100, Abcam ab5176, Boston, United States). Phalloidin was conjugated to Alexa-633 (A-22284, Thermofisher Scientific, United States) and secondary antibodies were Alexa-488 (A-11008 and A-11001, Thermofisher Scientific, United States), Alexa-568 (A-11011 and A-11004, Thermofisher Scientific, United States), or Alexa-633 (A-21050 and A-21070, Thermofisher Scientific, United States) conjugated.

For lethality assays, UAS-AFG3L2-RNAi and UAS-YME1L-RNAi were crossed with progenitor and PSC-specific drivers. GFP⁺ or non-GFP⁺ wandering third instar larvae were counted and the proportions with total larvae collected were quantified for all drivers except antp-Gal4. For lethality using antp-Gal4 driver, the Gal4 line was crossed with Canton-S flies which served as the control, and non-tubby flies were counted in control and AAA⁺ depleted genotypes. The experiments were repeated at least thrice, with at least 10 larvae in each replicate.

For pupariation rate, fly crosses were kept for 6 h for egg laying in glass. After 120 h of AEL, 3rd instar larvae were counted and monitored for pupariation. The experiment was repeated thrice with at least 40 flies in each replicate. Once pupariated, the flies were monitored for eclosion and the rates were compared.

For life span assays, male flies post 3-day eclosion were collected and maintained in sterile fly food bottle at 25 °C in cornmeal agar medium. Flies were transferred in fresh bottles every 3 days to avoid any pathogenic challenge. Five sets, each with minimum 12 flies were monitored daily, and natural deaths were recorded. Median survival rate was compared.

For visualizing and quantifying cellular ROS levels, larvae were bled in 50 µL Schneider’s media (21720-024, ThermoFisher Scientific, Paisley, Scotland, United Kingdom) and allowed to adhere onto 96-well coverslip bottomed dishes for 45 min. The cells were then incubated with 5 µM DHE (D11347, ThermoFisher Scientific, United States) in Schneider’s media for 15 min. Incubation was followed by washing with Schneider’s media. The cells were imaged live in a Zeiss LSM880 confocal microscope attached to a sample heater (25 °C).

For live imaging of mitochondria, 4-5 larvae were bled in 50 µL Schneider’s media, and allowed to adhere onto 96-well coverslip-bottomed dishes for 45 min. After washing the cells once with 1X PBS, MitoTracker Deep Red (M22426, ThermoFisher Scientific, United States) at a concentration of 5 nM made in Schneider’s media was added and incubated for 15 min at 25 °C. After washing with Schneider’s media twice, the cells were imaged live in a Zeiss LSM880 confocal microscope attached to a sample heater (25 °C) to control temperature. Images of different fields were acquired, and cells were analysed individually. Mitochondrial morphology analysis was performed on FIJI (https://imagej.net/software/fiji/downloads) with the MiNA (Mitochondrial Network Analysis) plugin (https://github.com/StuartLab/MiNA) as described elsewhere [50].

Maximum intensity projections were made from complete Z-stacks and manually stitched for representative figures. Images were processed uniformly for brightness and contrast using Adobe Photoshop CS5 (Adobe Inc., United States). White dotted lines indicate lymph gland lobe boundaries. Mean fluorescence intensities (MFI) were measured using ImageJ software. Cell count analyses using various markers were performed using the spots module in IMARIS (Oxford Instruments, United Kingdom). Quantifications were performed for the primary, secondary, and tertiary lobes individually.

For all the experiments, statistical significance was deduced by Students’ t-test with Welch’s correction except for life span assay. For life span assay, Mantel-Cox log-rank test was used. Graphs and statistical analyses were performed in GraphPad Prism 8.0 (Dotmatics, United States). Statistically significant differences were indicated by *p < 0.05, **p < 0.01, and ***p < 0.001 and ns indicates nonsignificant. Error bars represent standard error of mean (SEM).