Expansion of Drosophila haemocytes using a conditional GeneSwitch driver affects larval haemocyte function, but does not modulate adult lifespan or survival after severe infection Open Access

Vertebrate macrophages enact diverse functions, including clearance of infections, removal of apoptotic corpses and tissue remodelling as part of development (Wynn et al., 2013). Thus, macrophages are by necessity highly heterogeneous populations of cells, showing high levels of variation at both functional and transcriptomic levels (Murray et al., 2014; Colin et al., 2014; Gordon and Taylor, 2005; Gordon et al., 2014). As a result of this heterogeneity, the study of macrophages in vertebrate systems is challenging, especially considering the necessity to also consider contributions from the adaptive immune system. Therefore, the use of invertebrate models which represent vertebrate macrophages well is of great value to the field of immunology. The fruit fly Drosophila melanogaster is such a model system. Flies lack an adaptive immune system and solely harbour an innate immune system, which consists of circulatory blood cells known as haemocytes (Buchon et al., 2014; Holz et al., 2003). Broadly, haemocytes consist of three distinct cell types: crystal cells, lamellocytes and plasmatocytes. The Hml promoter is active in intermediate prohaemocytes, which are precursor cells within the cortical zone of the lymph gland (Sinenko, 2024), as well as most plasmatocytes and crystal cells, but not lamellocytes (Goto et al., 2003). This last group of cells is responsible for the encapsulation of large foreign objects. Functionally, crystal cells are thought to mediate melanisation, whereas plasmatocytes, which dominate the haemocyte population in quantity, are considered highly homologous in function to vertebrate macrophages (Yoon et al., 2023; Buchon et al., 2014; Wood and Jacinto, 2007; Evans et al., 2003). These cells are able to migrate to wound sites, engulf pathogens and deposit extracellular matrix, all of which are essential characteristics of macrophages, as well as displaying transcriptomic and functional heterogeneity (Coates et al., 2021; Ratheesh et al., 2015). Considering these features of this class of Drosophila haemocyte, the fruit fly has been utilised frequently as a model of macrophages and their associated behaviours (Ratheesh et al., 2015; Wood and Jacinto, 2007).

Considering the wide array of roles for which Drosophila haemocytes are responsible, it would seem logical for flies genetically engineered to lack haemocytes to be non-viable. Interestingly however, the developmental lethality of flies in which haemocytes are ablated appears to vary depending on the specific genetic driver used. Different studies have used different parts of the promoter region of the Hemolectin (Hml) gene to genetically ablate haemocytes. Flies in which haemocytes are ablated using the HmlΔ driver complete development, albeit with an approximately halved ratio of successful eclosion from pupae relative to control flies, but this phenotype is rescued by germ-free or antibiotic-supplemented conditions (Shia et al., 2009; Charroux and Royet, 2009; Arefin et al., 2015; Defaye et al., 2009; Ayyaz et al., 2015). In some studies, melanotic spots were observed in flies in which haemocytes had been ablated, proposed to be the result of a lamellocyte expansion in HmlΔ-positive cell-ablated conditions (Defaye et al., 2009; Arefin et al., 2015). Comparatively, flies in which haemocytes are ablated using a different region of the Hml promoter, named Hml^P2A, are severely pupal lethal, and are unable to be rescued (Stephenson et al., 2022). However, haemocyte ablation has not been investigated intensively with condition-dependent drivers. The benefit of conditional drivers is that all genetics are controlled for, with a genetic system conditionally activated by specific drugs (Clausen et al., 1999; Bockamp et al., 2008; Osterwalder et al., 2001) or environmental variables, such as temperature (McGuire et al., 2003; del Valle Rodríguez et al., 2011) and light (de Mena and Rincon-Limas, 2020). A small number of experiments have used a temperature-sensitive Gal80 to ablate haemocytes conditionally (Defaye et al., 2009; Monticelli et al., 2024; Ayyaz et al., 2015), one of which revealed a reduced post-infection survival phenotype upon conditional haemocyte ablation (Defaye et al., 2009). Another showed that ablation of haemocytes impairs intestinal regeneration following oxidative stress-mediated damage or infection with the pathogens Erwinia carotovora carotovora strain 15 or Pseudomonas entomophila (Ayyaz et al., 2015). However, never before has a non-temperature-dependent conditional approach been used to ablate haemocytes. In addition, it has not been established whether haemocyte function is altered by increasing the number of cells; in other words, the literature suggests that a lack of haemocytes negatively affects haemocyte-driven functions, but does an excess of haemocytes produce the inverse of this?

Here, we show the importance of the use of conditional drivers, especially for lifespan as a phenotype. Lifespan is highly polygenic making it critical to homogenise genetic backgrounds or use conditional systems where possible (Ford and Tower, 2005). Our results differed between experiments using constitutive and conditional drivers, despite both systems using the same promoter region, HmlΔ. The constitutive HmlΔ driver (HmlΔ-Gal4) successfully expanded and ablated Drosophila Hml-positive haemocytes through UAS-regulated ras85D^V12 and bax transgenes, respectively, and this correlated with an apparent modulation of lifespan. However, with our newly generated conditional driver, HmlΔ-GeneSwitch, no lifespan phenotype was observed, in spite of a modulation of Hml-positive cell numbers comparable to the constitutive driver experiments. Surprisingly, altering Hml-positive cell numbers conditionally did not change post-infection survival, raising the possibility that somehow the conditional modulation of Hml-positive haemocyte numbers did not change overall function. However, expansion of the Hml-positive cell population did modulate haemocyte function in larvae, which we demonstrated using a genetic model of self-encapsulation (Mortimer et al., 2021).

Bleeding larvae to obtain haemocyte populations revealed a large increase in overall haemocyte cell numbers when HmlΔ-GeneSwitch was used to drive ras85D^V12. However, in adult bleeds this was not the case, even though the Hml-positive population did grow in number in vivo. Conversely, our ablation conditions in adults did result in low numbers of Hml-positive cell numbers, but no change in overall bled cell numbers was observed. This suggests mechanisms that regulate total blood cell numbers could compensate for the ablation of Hml-positive haemocytes, despite strong initial manipulations in the larva. Our findings show that haemocyte numbers in larvae have functional relevance; more careful study of these functions in vivo will be possible using well-controlled conditional genetic systems such as presented herein.

All experimental flies were kept at 25°C. Flies were kept for mating for 2 days following eclosion and were subsequently sorted into sexes and the desired genotypes for the experiment. Fly food consisted of the following components, as previously described (McCracken et al., 2020a,b): 8% yeast (Kerry), 13% table sugar (Tate and Lyle), 6% cornmeal (Triple Lion Medium Cornmeal), 1% agar (United States Biological) and 0.225% nipagin (Sigma-Aldrich) (all w/v). In the case of growing bottles, 0.4% (v/v) propanoic acid (Sigma-Aldrich) was also added. Where the GeneSwitch construct was utilised, food was supplemented with RU486 (Fisher; 200 μmol l⁻¹ final media concentration; dissolved in 8.6 ml absolute ethanol (Fisher) per 1 litre of fly media and mixed into the media) or control food lacking RU486 (but still containing 8.6 ml absolute ethanol per 1 litre of fly media). The food used for control and RU486 food was split from a single media preparation, meaning the exact same food was used for both conditions at the same time. For all experiments using HmlΔ-GeneSwitch, RU486 was supplied to flies throughout development.

All fly genotypes utilised in this study are listed in Table 1, including source, where applicable. HmlΔ-GeneSwitch transgenic flies available on request.

The ELAV-GeneSwitch plasmid (Osterwalder et al., 2001), deposited by Haig Keshishian (Addgene plasmid #83957; RRID:Addgene_83957), was digested at 37°C for 1 h with BglII and PspXI (New England BioLabs), following the manufacturer's standard protocol. Genomic DNA was isolated from 20 Drosophila melanogaster w¹¹¹⁸ strain flies, utilising the DNeasy Blood & Tissue Kit (QIAGEN), following the manufacturer's standard protocol. The HmlΔ promoter was subsequently amplified from 80 ng genomic DNA using Platinum SuperFi II Polymerase (Invitrogen), with the following primers (BglII and PspXI restriction sites shown in lower case; additional bases added to maintain the GeneSwitch frame underlined), as previously used in the literature (Makhijani et al., 2011): F primer 5′-CGGTCACTcctcgaggCAAAAGTTATTTCTGTAGGC-3′; R primer 5′-CGGTAACTagatctGGCTGCTGGGAGTCCATTTTGTTAGGCTAATCGGAAATTG-3′. The amplified HmlΔ promoter and digested ELAV-GeneSwitch plasmid were each run on a 1% (w/v) agarose TAE gel at 80 V for 1 h and gel purified using the QIAquick Gel Extraction Kit (QIAGEN). Ligation of the insert and backbone was undertaken at 4°C for 16 h followed by room temperature (22°C) for 2 h, at a molar ratio of 10:1 (insert:backbone, using 50 ng insert DNA), performed using 1 U T4 DNA ligase (Invitrogen), in a 20 µl reaction. TOP10 Competent Cells (ThermoFisher Scientific) were transformed with the ligation mix using the heat-shock method, recovered in a 37°C shaking incubator at 225 rpm for 1 h, and plated on agar plates supplemented with carbenicillin (100 µg ml⁻¹), which were incubated at 37°C overnight. Discrete colonies were picked and following growth overnight in carbenicillin-supplemented LB broth at 37°C and 225 rpm, plasmid DNA was isolated using the QIAquick Spin Miniprep Kit (QIAGEN). The entire plasmid was sequenced using Oxford Nanopore long reads (Plasmidsaurus) and transformed into w¹¹¹⁸ Drosophila melanogaster flies using P-element transposition (GenetiVision). Standard fly husbandry methodology was used to generate homozygous flies containing the HmlΔ-GeneSwitch construct, which were subsequently tested for their ability to drive the UAS-Stinger transgene in a haemocyte-specific manner. Plasmids used to generate transgenic flies available on request.

Prior to imaging, adult flies were immobilised by incubation at −20°C for 10 min. An MZ205 FA fluorescent dissection microscope with a 1× PLANAPO objective lens (Leica) was subsequently used to image individual flies on the GFP channel (ET-GFP filter), using LasX image acquisition software (Leica). The following settings were used: exposure=800 ms, gain=1.0, zoom=30×. Haemocyte numbers were quantified using the Find Maxima tool in Fiji (Schindelin et al., 2012), with Prominence set to 20. Differences between conditions were assessed using a two-way ANOVA. Tukey HSD post hoc tests were used to specifically assess the statistical significance of differences relative to control flies. Flies containing the HmlΔ-Gal4 driver were imaged 1 day post-eclosion, whereas HmlΔ-GeneSwitch flies were imaged at 1 day post-eclosion to confirm the same effects were present as found using the HmlΔ-Gal4 driver, but were also imaged in a separate experiment at 6 days post-eclosion. This latter time point both allowed flies to feed on RU486-supplemented food after pupation and allowed assessment of whether modulation of haemocyte numbers lasted for a prolonged period into adulthood. Of note here is that, with the HmlΔ-Gal4 driver, cell numbers were not quantified further into adulthood, since this driver is too weak to accurately visualise Stinger with the imaging methods used.

To image haemocytes in larvae, wandering L3 larvae of the appropriate genotypes were selected and washed in distilled water and blotted dry. They were then imaged in distilled water on ice using the same microscope as above (2× PLANAPO objective lens, ET-GFP filter, exposure=300 ms, zoom=1.34, gain=3.0).

Pseudomonas entomophila (a gift from Julia Cordero, University of Glasgow, UK) and Staphylococcus aureus (a gift from Pedro Vale, University of Edinburgh, UK) bacteria (Vodovar et al., 2005; Perochon et al., 2021) were individually grown overnight at 29°C in LB broth, from stocks frozen at −80°C, to an OD_600nm of 0.5. Flies were individually immobilised using a CO₂ pad and a 0.15 mm diameter stainless steel insect minutien pin was washed in the bacterial stock, before being intrathoracically inserted into the immobilised fly for 2 s. Each fly was subsequently housed alone in a vial containing standard fly media [containing 8% (w/v) yeast] food, labelled with the time of infection and incubated in a temperature and humidity controlled room. The following day, flies were checked for death every 30 min from 1 h after the light turned on. Cox proportional hazard models were used to test the effect of RU486 in the context of each genotype. The ‘coxph’ function in the survival package in R (https://CRAN.R-project.org/package=survival) was used to assess statistical significance.

Flies were transferred to a custom cage to minimise disturbances, as previously described (McCracken et al., 2020a; Good and Tatar, 2001). No more than 100 flies were housed in each cage, with sexes separated. Following caging, every other day, the number of deaths were recorded (and dead flies subsequently removed) and a fresh vial of food was provided, until no flies remained. In order to conditionally drive the HmlΔ-GeneSwitch construct throughout development, RU486-containing or control food was utilised from early larval stages. Proportional Cox hazard models were used between conditions in each genotype and each sex where applicable, including the ‘cage’ as a random effect to correct for pseudoreplication within cages (McCracken et al., 2020b; Therneau et al., 2003; Ripatti and Palmgren, 2000). The coxme package in R (https://CRAN.R-project.org/package=coxme) was used to assess statistical significance to correct for the shared environment of cage. This was used in place of the ‘coxph’ function as it allows for inclusion of between-cage effects, which are not relevant for post-infection assessment of survival where only one fly is kept in each vial. Flies that escaped or that were squashed were censored; no other exclusion criteria were applied here or elsewhere so long as animals were the correct genotype as per discrimination via balancer chromosomes and markers. Flies were randomly assigned to different population cages for ageing with different drug regimens.

To visualise an active cell-based immunity response, we used the tuSz¹ mutation, which results in haemocytes incorrectly recognising fat body cells as non-self and leading to self-encapsulation and associated melanisation (Mortimer et al., 2021). Virgin females containing the tuSz¹ mutation (w^tuSz1) were crossed with w¹¹¹⁸;If/UAS-ras85D^V12;HmlΔ-GeneSwitch males at 25°C on food containing either RU486 or control food. After 24 h at 25°C, parents were removed and the vial containing eggs switched to either 29°C to enable the temperature-sensitive tuSz¹ melanotic phenotype to manifest, or kept at 25°C as negative controls. To assess if expansion of haemocytes caused lethality, offspring on both diets at both temperatures were sexed and counted, as well as being genotyped using the visible marker If to discriminate the presence of UAS-ras85D^V12. At the L3 stage of development, 27 male larvae on a control diet and 25 male larvae on an RU486-containing diet at 29°C were randomly picked and their melanisation level scored from 1 to 5, based on a novel, semi-quantitative scoring system (shown in Fig. 6; flies were scored blinded). Each larva was individually housed in a fresh vial and eventually checked for genotype upon eclosion. Flies which did not eclose were checked for their genotype, via If, by removing the puparium. Larvae were imaged on a MZ205 FA fluorescent dissection microscope with a 1× PLANAPO objective lens (Leica) on the brightfield channel, using LasX image acquisition software (Leica), with the following settings: exposure=700 ms, gain=1.0 and zoom=40×. To image in detail the fat body of developmentally arrested flies following puparium removal, the same microscope was utilised, but with zoom set to 120×.

Wandering L3 larvae of the appropriate genotype were selected and washed in distilled water, blotted dry and then transferred to 75 μm of S2 media [Schneiders medium (Merck) supplemented with 1× Pen/Strep (Gibco) and 10% heat-inactivated FBS (Gibco)] on ice in a Petri dish. Larvae were opened using size 5 forceps from their posterior spiracles towards the anterior end to release haemocytes into S2 media and then agitated for 10 s to release more tightly adhered cells. S2 media (75 μl) containing haemocytes was then transferred to a 96-well plate (Greiner) and a further 75 μl of S2 media added. For adult bleeds, flies were placed at −20°C for 3 min ahead of dissection to stop movement. Two newly-eclosed adult flies (day 1) of the same sex were then dissected in a single 75 μl droplet of S2 media on ice. Flies were decapitated and then cut longitudinally along their head, thorax and abdomen within that droplet and then an additional 75 µl S2 media was added. Carcasses were agitated ten times by pipetting before cell suspensions were transferred to a well in a 96-well plate. For both larval and adult bleeds, cells were allowed to adhere at room temperature and in the dark for 45–60 min before fixation.

Cells adhered to wells were fixed in 4% formaldehyde (diluted from 16% EM grade, Thermo Fisher) solution in PBS for 15 min, before permeabilisation using 0.1% Triton X-100 (Sigma) in PBS (Oxoid) for 5 min. Cells were stained using either TRITC–phalloidin (1:1000, Invitrogen; all dissections with the exception of HmlΔ-Gal4, constitutive L3 larvae) or Texas Red–Phalloidin (1:400, Invitrogen; HmlΔ-Gal4, constitutive L3 larvae) along with nucBlue nuclear stain (2 drops per ml, Invitrogen). PBS washes were performed after each step and cells were imaged in a final volume of 200 μl PBS using an MDX ImageExpress Hi-Content microscope (10× lens). DAPI, Cy3, Texas Red and GFP filter sets were used to image nucBlue, TRITC-phalloidin, Texas Red-phalloidin and Stinger fluorescence, respectively. Nine sites were imaged per well.

Images from at least six fields of view (FOV) per well were exported as Tiff files using ImageExpress software. For larval bleeds, cell counts per FOV were made from the DAPI/nucBlue channel images. These images were segmented (thresholded, fill holes, watershed) to create binary images, which were then counted automatically using the ‘analyse particles’ tool in Fiji. All contrast adjustment and thresholding were applied equally to each Tiff file within each coherent data set.

For adult dissections, contrast adjustments were applied to Tiff images for each channel to create an RGB merged image; as previously each channel was treated equally across experimental groups. Using the point selection tool and channels tool in Fiji the number of Hml-positive cells with nucBlue-stained nuclei were counted in the green/EGFP and DAPI channels; HmlΔ-Gal4 or HmlΔ-GeneSwitch were used to drive expression from UAS-Stinger in these samples. Non Hml-positive haemocytes were then scored on the basis of cell morphology (F-actin staining via Phalloidin/red channel fluorescence) and the presence of nucBlue staining. Cells without nucBlue staining were excluded.

Statistical significance was calculated using ANOVAs for the conditional driver, with pairwise comparisons assessed using Tukey's HSD post hoc testing, or Student's two-tailed t-tests for constitutive data. Normality of the residuals from the ANOVA models was investigated visually using histograms as well as heteroscedasticity by using plotting of predicted values against residuals.

We used HmlΔ-Gal4 to constitutively drive either the pro-apoptotic gene bax, a potent activator of apoptotic pathways in the fly (Gaumer et al., 2000), in order to ablate HmlΔ-positive cells, or ras85D^V12, a constitutively active mutant variant of ras used to drive expansion of the HmlΔ-positive cell population (Hobbs et al., 2016; Ashburner et al., 1999; Simcox et al., 2008). Hml was used to drive these in vivo manipulations as it is expressed in the majority of differentiated Drosophila haemocytes as well as in prohaemocyte precursor cells (Goto et al., 2003). Expression from UAS-Stinger, which encodes a nuclear-localised GFP (Barolo et al., 2000) enabled labelling of Hml-positive cells. Virtually no Hml-positive cells were visible in flies where HmlΔ-Gal4 was used to drive UAS-bax, as found previously (Defaye et al., 2009), and increased numbers of Hml-positive cells were observed in flies where HmlΔ-Gal4 was used to drive UAS-ras85D^V12 (Fig. 1A,C). We observed a slight increase in the developmental lethality of flies carrying HmlΔ-Gal4 and either UAS-bax or UAS-ras85D^V12, relative to control flies only carrying HmlΔ-Gal4 (Fig. 1B). The increase in mortality in flies carrying UAS-bax flies achieved statistical significance (⁠statistic=6.43, P-value=0.011), whereas the apparent increase in UAS-ras85D^V12 flies was not significant (⁠ statistic=1.40, P-value=0.236).

HmlΔ-Gal4 is a constitutive driver and leads to the expression in differentiated blood cells from larval stages onwards. This approach, however, requires the crossing of different genetic lines that are unlikely to be genetically identical. Conditional drivers have the benefit of controlling fully for genotype, therefore we generated a HmlΔ-GeneSwitch transgenic line. The HmlΔ-GeneSwitch construct we made resulted in strong expression of a UAS-Stinger transgene by 1 day post-eclosion, when activated by RU486 throughout development (Fig. S1A). Under both ablated and expanded conditions (HmlΔ-GeneSwitch driving UAS-bax and UAS-ras85D^V12, respectively, with RU486 provided throughout development and continuing post-eclosion), a modulation of cell number was detected with both transgenes 6 days post-eclosion (Fig. 2; ablation data for 1 day post-eclosion shown in Fig. S1B), demonstrating that HmlΔ-GeneSwitch can fully reproduce results obtained via HmlΔ-Gal4 (Fig. 1). Note that our ability to visualise cells using UAS-Stinger requires induction of this system, thus we only visualise cells that have an active HmlΔ promoter.

Bleeding of larvae and adults allows measurement of total numbers of blood cells (including cells not labelled by UAS-Stinger), with the caveat that it does not capture cells tightly integrated into tissue or strongly adhered, which may therefore be harder to release during dissections. Despite this limitation, imaging of L3 larvae ahead of dissection demonstrated a strong enhancement of numbers of Hml-labelled cells compared with levels in controls upon constitutive expression of ras85D^V12 via HmlΔ-Gal4, whereas almost no cells were visible on expression of bax (Fig. S2A–C). However, larval bleeds from the same genotypes revealed significant increases in total haemocyte numbers via both strategies (Fig. S2D–G). Very few Stinger-positive cells survived upon expression of bax and many of those cells exhibited a lamellocyte-like morphology, whereas almost all cells present upon ras85D^V12 expression were Stinger-positive (Fig. S2E,F).

Bleeding of adult flies revealed that a significant difference was only achieved for flies in which bax was driven using constitutive HmlΔ-Gal4, with only a mild reduction in haemocytes observed (and then only in females). Furthermore, driving ras85D^V12 did not significantly expand numbers of bled cells (Fig. S2H,I). As per larvae, many cells present in adult bleeds upon expression of bax exhibited lamellocyte-like morphologies (data not shown).

Conducting the equivalent experiments using HmlΔ-GeneSwitch, for which the food was supplemented with RU486 throughout development, revealed very similar results to the constitutive approaches (Fig. 3): in HmlΔ-GeneSwitch larvae driving ras85D^V12, the number of cells was increased more than 30-fold, whereas larvae in which bax was driven showed a trend towards an increase in cell number, although this did not achieve statistical significance (P=0.06; Fig. 3A–E).

In adults where transgenes were driven by HmlΔ-GeneSwitch, no statistically significant changes in total cell numbers were observed (Fig. 3F), although the number of Hml-positive cells (interpreted from GFP labelling) was significantly increased in flies where ras85D^V12 was driven by HmlΔ-GeneSwitch, relative to those in which bax was driven (Fig. 3G).

These data suggest the existence of limitations or feedback mechanisms that control numbers of haemocytes that can be present within adults, given that, despite dramatic increases in cell numbers in larvae, the effect on numbers of haemocytes in adults is less pronounced upon expression of ras85D^V12. Potentially, rather than modulating the number of total haemocytes, our genetic tools may in fact specifically modulate numbers of Hml-positive cells.

Considering that immune function is central to ageing (Stanley et al., 2023), we wanted to investigate whether manipulating the haemocyte population would impact longevity. Upon testing this with large sample sizes under highly controlled conditions, we observed alterations to lifespan upon both Hml-positive cell expansion and ablation using constitutive expression of ras85D^V12 or bax via HmlΔ-Gal4 (Fig. S3A). However, this effect was not replicated when repeating this using the conditional HmlΔ-GeneSwitch system and continuous application of RU486, despite ensuring that all other aspects of the experiment were identical (Fig. 4; statistics shown in Table S1A). This strongly suggests that small genetic differences between the UAS lines explained the difference in longevity observed via the constitutive HmlΔ-Gal4 approach.

Indeed, similarly, bax, ras85D^V12 and control crosses with the conditional driver showed the same lifespan differences; the bax crosses lived for the shortest times, whereas the ras85D^V12 crosses lived longest, but with no additional modulation by driving the ablation or expansion using RU486 (Fig. 4). Additionally, to test this hypothesis further, we crossed the same bax, ras85D^V12 and control lines to a yw background and found similar lifespan differences to those observed when using the constitutive HmlΔ-Gal4 driver, particularly for the reduced lifespan observed with bax (Fig. S3B). These observations fit with the idea that lifespan is a highly polygenic trait and that the accumulation of deleterious mutations through drift can strongly determine lifespan (Cui et al., 2019), underscoring the importance of conditional driver systems, especially when studying ageing (Ford and Tower, 2005). It remains possible that stronger enhancement of haemocyte numbers could impact lifespan, whereas the non-Hml positive haemocytes that appear expanded upon expression of bax could mask effects on lifespan linked to compromised immunity.

We investigated whether Hml-positive cell number affects survival from a pathogen infection, as previously reported using constitutive drivers (Defaye et al., 2009; Charroux and Royet, 2009) and including a small number of experiments using a temperature-sensitive conditional system (Defaye et al., 2009). Immune challenges with two commonly used pathogens (Pseudomonas entomophila or Staphylococcus aureus) to study immunity in the fly were not affected by conditional Hml-positive cell ablation or expansion (Fig. 5; statistics shown in Table S1B and dose optimisation shown in Fig. S4). These results therefore imply that neither lifespan nor post-infection survival are severely impacted by modulating the number of Hml-positive cells as performed here. Again, more significant expansion of Hml-positive cells or the presence of increased numbers of Hml-negative haemocytes could obscure immune phenotypes in these assays.

Using our novel HmlΔ-GeneSwitch driver, we were able to change the number of Hml-labelled cells, but this was not associated with infection or lifespan phenotypes. Therefore, we wanted to test another known haemocyte-related function upon manipulation of their numbers, as a positive control. To do this, we used the tuSz¹ mutant, which has a mutation on the X chromosome, 35 bp upstream of the GCS1 transcription start site. This mutation results in a loss of self-tolerance in the posterior fat body and a haemocyte-driven self-encapsulation phenotype that is temperature dependent (Mortimer et al., 2021). We developed a novel scoring system in order to quantify this self-encapsulation phenotype (Fig. 6). We hypothesised that expansion of the haemocyte cell population would lead to increased targeting of misrecognised ‘self’ tissue in the presence of the tuSz¹ mutation (Mortimer et al., 2021). Indeed, we observed developmental lethality when ras85D^V12 was overexpressed using HmlΔ-GeneSwitch, with lethality stronger at 29°C (the temperature at which the melanisation phenotype presents itself in control flies). In males at the higher temperature, where the tuSz¹ mutation cannot be rescued by a second copy of the X chromosome, near-complete developmental lethality was observed (Fig. 7A and Table 2); only a single fly eclosed at 29°C where UAS-ras85D^V12 had been activated by HmlΔ-GeneSwitch and this fly died within 3 days of eclosion. Flies that overexpressed ras85D^V12 alone at 29°C, without the tuSz¹ mutation, showed no developmental lethality (Fig. 7B and Table 2). Flies overexpressing ras85D^V12 in the same genetic background but at 25°C also showed no developmental lethality (Fig. S5A and Table 2), similarly, the 25°C RU486-driven flies showed no significant difference to flies grown at 25°C where ras85D^V12 was driven by the constitutive HmlΔ-Gal4 driver (Table 2, counts from Fig. 1B; these flies were female only so as to match the constitutive counts). This therefore suggested that there is no difference in effect on lethality of UAS-ras85D^V12 between the constitutive and conditional HmlΔ drivers. We also repeated these experiments using flies carrying the UAS-bax transgene to test whether reduction of the Hml-positive cell population using UAS-bax driven by HmlΔ-GeneSwitch affected development and ability to eclose in the presence of the tuSz¹ mutation or on a w¹¹¹⁸ background. In this experiment, we observed no successful eclosers for either sex in 29°C RU486 conditions in the tuSz¹ background, although flies did successfully eclose in the w¹¹¹⁸ background. As with the equivalent UAS-ras85D^V12 experiments, activation of the HmlΔ-GeneSwitch driver through supplementation of RU486 reduced eclosion rates at both temperatures in the context of the tuSz¹ background, although unlike the ras85D^V12-carrying flies, RU486 supplementation to UAS-bax flies also reduced eclosion rates in the context of the w¹¹¹⁸ genetic background. Furthermore, again as found with the ras85D^V12 experiment, the strength of effect of HmlΔ-GeneSwitch driving UAS-bax did not significantly differ from that of HmlΔ-Gal4 driving the same transgene (Fig. S5B,C and Table 2).

Melanisation is most easily detected at larval stages (Fig. 7E), but it was clear that the ras85D^V12 flies that failed to eclose showed excessive melanisation and this could be observed even without removal of the pupal case (Fig. 7F,G). Removal of the puparium revealed that these pupal-lethal flies consistently died at approximately the P12 pupal stage (Bainbridge and Bownes, 1981) and showed significant levels of excessively melanised tissue in the fat body, which was likely responsible for the observed developmental lethality (Fig. 7C).

Conditional expansion of the Hml-positive cell population also increased melanisation during larval stages. RU486-fed larvae showed significantly higher melanisation scores relative to the control diet-fed larvae (Fig. 5D,E). For this purpose, we single-housed larvae to determine their genotype retrospectively (using If as a marker to genotype flies; see Materials and Methods). As in our earlier experiment, no male larvae containing both HmlΔ-GeneSwitch and UAS-ras85D^V12 that had been fed on RU486-containing media successfully developed and eclosed at 29°C. Furthermore, the genotypic proportions of pupae which eclosed were approximately the same as observed when larvae were not kept in individual vials, ruling out the possibility that the lethality phenotype was a result of competition between larvae of different genotypes (Fig. 7A compared with Fig. 7D). Overall, these results demonstrate that although no lifespan or infection survival phenotypes are observed upon conditional Hml-positive cell number modulation, manipulation of Hml-positive cell number using HmlΔ-GeneSwitch can produce strong functional haemocyte-specific effects.

Our work demonstrates that a constitutive haemocyte-specific driver (HmlΔ-Gal4) is able to not only reduce Hml-positive cell numbers, but also expand the population, via expression of bax and ras85D^V12 transgenes, respectively (Fig. 1A,B). Additionally, a conditional HmlΔ-GeneSwitch driver produced comparable effects (Fig. 2). However, bleeding haemocytes from adults and larvae suggested that we were not affecting all immune cells and that compensatory mechanisms can impact the overall population in vivo (Fig. 3 and Fig. S2). We confirmed that Hml-positive cell number expansion modulated a known haemocyte-dependent phenotype (self-encapsulation; Fig. 7). However, manipulating Hml-positive cell populations had no effect on lifespan (Fig. 4) or survival post-immune challenge with two different pathogens (Fig. 5), in spite of a lifespan modulation being initially observed via a constitutive HmlΔ-Gal4-mediated approach (Fig. S3A). No difference was observed between the strength of the constitutive and conditional HmlΔ drivers in terms of lethality as a result of expanding the haemocyte population (comparison between counts in Fig. 1C and Fig. S5A), thus reducing the likelihood that no effect was observed using our conditional HmlΔ as a result of it being weaker than HmlΔ-Gal4. These negative findings on longevity and especially post-infection survival contrast to a number of studies in the literature. There are several reasons for these differences that we discuss below, with the foremost reason being that other Hml-negative immune cells likely replace Hml-positive haemocytes on ablation with bax.

The literature suggests that haemocyte ablation should affect survival following pathogen exposure (Shia et al., 2009; Charroux and Royet, 2009; Arefin et al., 2015; Defaye et al., 2009). However, the severe pupal lethality and increased self-encapsulation phenotype observed in flies in which the Hml-positive population was expanded in the context of the tuSz¹ mutation suggests that increasing Hml-positive cell number really does upregulate haemocyte activity. Note, however, that we cannot distinguish cell number from other possible pleiotropic effects of expressing constitutive-active Ras, although Ramond et al. (2020) found limited evidence of activation of stress-associated pathways via overlapping approaches to manipulate haemocytes; similar considerations hold for bax as an experimental strategy. Regardless, it is surprising that we did not observe other adult phenotypes. Although the infected flies in our experiments died within a similar timeframe as prior experiments, it is possible that the dose of the immune challenge masks any modulating effect from haemocytes. Potentially therefore, haemocyte number may be critical in the face of infection only at certain levels of infection. Another explanation is that there is compensation within the population of cells. It is also possible that the Hml-negative cells are sufficient to protect against infection and that the expansion in adults via ras expression does not reach a threshold to exert harmful or beneficial effects. It is also now known that haemocytes communicate with cross-talk between these cells and other tissues (Yang and Hultmark, 2016; Green et al., 2018). As such, it remains possible that both ablation and expansion of the Hml-positive population functionally does little, except if cells are falsely attracted to ‘self’. Perhaps, the expanded cells are differentiated such that functioning is increased in recognition of ‘self’ rather than any other functions. Indeed, a recent study used the pro-apoptotic genes hid and reaper to ablate Hml-positive haemocytes and found an increase in Hml-negative plasmatocytes, in line with our data and interpretation (Shin et al., 2020) and expansion of lamellocyte-like cells on ablation was also seen by Arefin et al. (2015). These Hml-negative haemocytes were shown to express Pxn, a classical haemocyte marker gene (Evans et al., 2014), despite the lack of Hml, and therefore it would be beneficial in future work to test whether our Hml-negative haemocyte population also expresses Pxn, in order to fully confirm that our results here identified the same Hml-negative population. The expansion of lamellocyte-like cells observed by both us and Arefin and colleagues is likely a result of lamellocytes lacking Hml expression (Goto et al., 2003), so enabling this lineage of haemocytes to dominate in Hml-positive cell ablating conditions, in part at least as they are thus able to escape expression of bax. A separate study also concluded that depletion of Hml-positive cells was insufficient to generate flies completely lacking haemocytes, as Hml-negative haemocytes remained unaffected (Boulet et al., 2021). Their study also suggested that depletion of Hml-positive immune cells may affect the gene expression of the remaining immune cells, in line with our own work that has previously shown exposure to apoptosis can impact subpopulations at other stages of development (Coates et al., 2021; Brooks et al., 2024). These observations perhaps fit with our observation of an increase in lamellocyte-like cells. Regardless, it is essential that future studies further investigate methods of haemocyte depletion without limiting themselves exclusively to Hml drivers. Furthermore, the populations of Hml-negative immune cells which remain following Hml-positive cell depletion offer an explanation for the lack of infection phenotype in flies where the haemocyte population has been ‘ablated’ or ‘expanded'. These possibilities will prove a fruitful paradigm to study immunology in the fly; an emerging theme in this area of research are functional differences within the plasmatocyte lineage of haemocytes (Shin et al., 2020; Coates et al., 2021).

It is interesting that ras85D^V12 dramatically increased the total cell population in larvae, but not in adults. This is an intriguing discovery, which suggests that potentially a feedback mechanism exists in adult flies to restrict numbers of Hml-positive cells, which are not present or as active in larvae. This is a discovery that is not only relevant for fly biology, but also for potential use as a cancer model. Feedback mechanisms that regulate the number of immune cells are highly relevant to understanding immunoproliferative disorders such as leukaemia or lymphoma (Connors et al., 2020; Tebbi, 2021). Of further relevance, recent work from the Giangrande lab revealed early wave haemocytes can impact later blood cell development via extracellular matrix deposition, demonstrating potential feedback mechanisms are relevant in the fly (Monticelli et al., 2024).

Blood cell numbers are tightly controlled in higher organisms at the level of generation and release from the bone marrow, with clearance from circulation via entry into tissues (Scheiermann et al., 2015). Most commonly, blood cells are ultimately removed via apoptotic cell clearance in organs such as the liver and spleen. The retraction of blood cell numbers we have observed appears to occur between late larval stages and eclosion as adults. Given that no further proliferation has been identified during pupal stages and that haemocytes are released from the lymph gland after the end of larval stages, it is unlikely that feedback mechanisms would supress proliferation or release, as this would be too late to impact cell numbers. Cell death, most likely apoptosis, is thus the prime candidate here. More typically, Ras gain-of-function mutations are associated with a poor prognosis in cancer and increased rates of proliferation/cell survival, including in Drosophila (e.g. Gafuik and Steller, 2011). However, some studies suggest Ras activation can sensitise some tumour cells to chemotherapy or lead to apoptosis (Overmeyer and Maltese, 2011). Therefore, one possibility might be an intrinsic mechanism whereby excessive levels of Ras activity eventually drive apoptosis.

Alternatively, a more systemic mechanism might be at play, potentially orthologous to control of vertebrate neutrophil numbers: here clearance of apoptotic neutrophils inhibits IL-23 release; since IL-23 would normally stimulate T cell-mediated IL-17 release and drive subsequent elevation of G-CSF, this limits further blood cell production (Stark et al., 2005). Other hypotheses might include insufficient levels of pro-survival factors like TGF beta or PDGF/VEGF-related ligands to maintain this size of population, given the previous roles for these ligands in mediating haemocyte survival (Bruckner et al., 2004; Makhijani et al., 2017). What is clear is that elevated numbers of haemocytes can strongly perturb fly physiology with large expansions demonstrated to lead to lethality under some conditions (Ramond et al., 2020) or ‘selfishly’ reduce access of non-immune organs to nutrients (Bajgar and Dolezal, 2018). Thus, there is a strong rationale for negative regulation of cell numbers to maintain organismal health. Complex feedback networks regulate haemocyte proliferation, differentiation and release in flies (see Bannerjee et al., 2019 for review; Madhwal et al., 2020), though these pathways are more associated with promotion of cell survival than as drivers of immune cell apoptosis, so this remains an exciting avenue for future experimentation.

Flies in which haemocytes are ablated display melanotic spots, purportedly as a result of the lamellocyte population expanding upon HmlΔ-positive cell ablation (Defaye et al., 2009; Arefin et al., 2015). We observed similar melanotic spots when we ablated the Hml-positive cell population with the constitutive HmlΔ-Gal4 driver (data not shown). Furthermore, we observed an increase in melanotic spots in the context of the tuSz¹ mutation when the haemocyte population was expanded. It is not clear how these two melanotic phenotypes are related. The ablation-related spots identified by previous work were found across the whole larval body (both posterior and anterior), whereas the melanisation generated by the tuSz¹ mutation is found only in the larval posterior (Mortimer et al., 2021). Furthermore, the tuSz¹-mediated melanisation persists into adult flies, whereas the melanotic phenotype previously identified when haemocytes were ablated is only mentioned in relation to larvae (Defaye et al., 2009; Arefin et al., 2015). Although these phenotypes appear superficially similar, they may in fact originate via different mechanisms.

The flies utilised in this body of work were not grown in a sterile environment, and so our expectation was that modulation of haemocyte number would affect lifespan. Prior studies have shown other adult phenotypes for haemocyte ablation, and perhaps if we could ablate more or all Hml-positive cells a phenotype would be observed, although very few Hml-positive cells were left in flies in which we ablated this cell population. Indeed, the Hml^P2A driver is suggested to be stronger and active in a larger fraction of the haemocyte population (Stephenson et al., 2022). However, previous uses of the HmlΔ promoter have shown clear immune phenotypes upon challenge with pathogens (Shia et al., 2009; Charroux and Royet, 2009; Arefin et al., 2015; Defaye et al., 2009; Ayyaz et al., 2015). The only obvious difference in the approach we have used here compared with previous constitutive (or temperature-sensitive conditional) approaches is that our driver only starts to become active upon the first instance of larval feeding, as the chemical required to activate the GeneSwitch construct was in our experiments administered via the media. This should have a limited impact, considering that Hml is considered to be active only from early larval stages (Charroux and Royet, 2009); however, there is the potential for a narrow window during which Hml becomes active, so promoting transcription of the GeneSwitch construct, but before RU486 is ingested via first larval feeding. In essence this means that HmlΔ-positive cells may be present for a short period in our conditional haemocyte manipulation system where they would not be present when using a constitutive driver. Similarly, levels of RU486-dependent activation may drop during pupal stages when active feeding has ceased. Overall, it seems unlikely that all previously documented immune phenotypes of ablating haemocytes are the result of ablation at this very early larval or pre-larval stage, and that any future ablation has no immune effect, but it cannot be ruled out.

Overall, our work demonstrates that a conditional HmlΔ-GeneSwitch driver is able to drive not only ablation of Hml-positive cells, but also expansion of this population in flies. However, modulating haemocyte numbers conditionally did not affect longevity, in contrast to our observations when using a constitutive driver, nor did it affect immunity. This leads to our conclusion that rare alleles that are fixed at random but are determinative of lifespan likely explain these differences. Lifespan is thus a phenotype in which it is especially important to control for genetic background, as has been argued before (Ford and Tower, 2005). Our study therefore illustrates the utility of the HmlΔ-GeneSwitch driver to understand macrophage behaviour in the fly model. The work also suggests powerful mechanisms of regulation for overall numbers of immune cells in the fly that may not always be trivial to overcome experimentally. In this process, we identified potentially important but unknown mechanisms that regulate total immune cell numbers, for which conditional and especially temporal manipulation (Simons et al., 2019 preprint) using GeneSwitch provides a promising experimental paradigm.

We thank Katherine Whitley, Karen Plant and the University of Sheffield Fly Facility and Simons Lab technical team for fly husbandry and logistical support. The MDX ImageExpress Hi-content microscope is housed within the Sheffield RNAi Screening Facility and we thank Steve Brown (University of Sheffield) for assistance with that equipment. This work would not be possible without the Bloomington Drosophila Research Centre (NIH P40OD018537) and Flybase (MRC grant MR/N030117/1). We thank Katie Roome for contributions to initial experiments in this project and acknowledge Martin Zeidler (University of Sheffield) for helpful comments and discussions.