Charting ESCRT function reveals distinct and non-compensatory roles in blood progenitor 1 maintenance and lineage choice in Drosophila 2 3

8 Tissue heterogeneity permits diverse biological outputs in response to systemic signals but requires 9 context-dependent spatiotemporal regulation of a limited number of signaling circuits. In addition to 10 their stereotypical roles of transport and cargo sorting, endocytic networks provide rapid, adaptable, 11 and often reversible means of signaling. Aberrant function of the Endosomal Sorting Complex 12 Required for Transport (ESCRT) components results in ubiquitinated cargo accumulation, uncontrolled 13 signaling and neoplastic transformation. However, context-specific effects of ESCRT on developmental 14 decisions are not resolved. By a comprehensive spatiotemporal profiling of ESCRT in Drosophila 15 hematopoiesis in vivo, here we show that pleiotropic ESCRT components have distinct effects on blood 16 progenitor maintenance, lineage choice and response to immune challenge. Of all 13 core ESCRT 17 components tested, only Vps28 and Vp36 were required in all progenitors, whereas others maintained 18 spatiotemporally defined progenitor subsets. ESCRT depletion also sensitized posterior progenitors 19 that normally resist differentiation, to respond to immunogenic cues. Depletion of the critical Notch 20 signaling regulator Vps25 did not promote progenitor differentiation at steady state but made 21 younger progenitors highly sensitive to wasp infestation, resulting in robust lamellocyte 22 differentiation. We identify key heterotypic roles for ESCRT in controlling Notch activation and thereby 23 progenitor proliferation and differentiation. Further, we show that ESCRT ability to regulate Notch 24 activation depends on progenitor age and position along the anterior-posterior axis. The phenotypic 25 range and disparity in signaling upon depletion of components provides insight into how ESCRT may 26 tailor developmental diversity. These mechanisms for subtle control of cell phenotype may be 27 applicable in multiple contexts. 28


Introduction
Blood progenitors are exposed to a plethora of signals and need to respond to a rapidly changing 66 environment for stem-and progenitor cell maintenance and controlled differentiation. Previous 67 genetic screens and knockout-based functional analyses in both Drosophila and mouse models 68 showed a role of ESCRT in maturation of specific blood cell types in erythroid and lymphoid lineages 69 and a possible functional link of ESCRT to blood cell homeostasis [6][7][8]. Several key endosomal proteins 70 such as Rabex5, Atg6, Rab5 and Rab11 actively control endocytic trafficking and are implicated in 71 developmental signaling and blood cell homeostasis [9][10][11][12][13]. The blood cell-enriched conserved 72 endosomal regulator of hematopoiesis, Asrij interacts with ADP Ribosylation Factor 1 (ARF1-GTP), 73 regulates the endocytic proteome and maintains stemness of blood progenitors in the Drosophila 74 lymph gland [14]. Loss of Asrij promotes activation of Notch signaling in hematopoiesis, thereby 75 leading to precocious differentiation to crystal cells [14]. Also, asrij mutant hemocytes accumulate 76 Notch intracellular domain (NICD) in the sorting endosomes, mimicking an endosomal sorting defect. 77 Mouse Asrij also maintains hematopoietic stem cell quiescence [15]. This suggests conserved 78 mechanisms of endosomal protein sorting in blood cell homeostasis that merit further investigation. 79 The Drosophila larval lymph gland serves as a simple yet powerful model to study conserved 80 mechanisms of hematopoiesis in situ. Using this in vivo model, we investigated the role of ESCRT 81 components in spatiotemporal control of blood progenitor homeostasis and myelopoiesis. Blood 82 progenitors reside in the multi-lobed lymph gland that flanks the cardiac tube in segments T3 and A1. 83 The primary lobe of the lymph gland comprises of three distinct zones enriched in progenitors, 84 differentiated blood cells (plasmatocytes, crystal cells and lamellocytes) and the hematopoietic niche. 85 Blood progenitors of Drosophila are linearly arranged in anterior (primary) and posterior lobes 86 (secondary, tertiary and often quaternary) of the lymph gland and are characterized by the expression 87 of several markers such as Domeless, TepIV and DE-Cadherin. The lymph gland develops in an anterior 88 to posterior sequence, with younger progenitors in the posterior lobes [16]. Previous studies showed 89 that the progenitor population is heterogeneous in gene expression, mitochondrial morphology and 90 dynamics, signaling, differentiation potential and immune function [16,17]. Posterior progenitors are 91 refractile to immune challenge due to immature mitochondrial morphology and differential activation 92 of JAK/STAT and Notch signaling [16,17]. The lymph gland harbors the entire blood progenitor 93 population of Drosophila, and hence allows complete sampling and a comprehensive study of 94 progenitor homeostasis. 95 The Drosophila ESCRT is comprised of 13 core components [2,18]. Stam) binds to the 96 ubiquitinated cargoes through a ubiquitin-interacting motif. It then recruits ESCRT-I (Vps28,Tsg101,97 Vps37A, Vps37B) and ESCRT-II (Vps25,Vps22 and Vps36), which act as a bridging complex to assemble 98 ESCRT-III (Vps32,Vps24,Vps20,Vps2). ESCRT-I-dependent membrane inward budding (negative 99 curvature) and ESCRT-III-dependent membrane scission lie at the heart of endosomal protein sorting, 100 resulting in the formation of intraluminal vesicles (ILV) containing the sequestered cargo [19]. Vps32 101 is the principal filament-forming component that undergoes activation and polymerization upon 102 binding with various nucleating factors and integrates previous steps of endosomal sorting [5]. The 103 final step involves disassembly of ESCRT subunits and scission of the membrane neck of the 104 intraluminal vesicles, which is mediated by the Vps4-Vta1 mechanoenzyme complex. 105 Here, we provide a functional map of the role of all 13 Drosophila core ESCRT components in 106 ubiquitinated cargo sorting and blood cell lineage choice across distinct progenitor subsets. We show 107 that though ESCRT components are expressed in all cells of the lymph gland (LG), their roles in 108 controlling lineage-specific differentiation and immune response of blood progenitors are distinct and 109 position-dependent. We also find that ESCRT dysfunction primarily affects Notch activation-110 dependent crystal cell differentiation. Our study supports heterogeneity of blood progenitors and 111 highlights the role of ESCRT in spatiotemporal segregation of signaling. 112 113 114

Fly stocks and genetics 116
Drosophila melanogaster stocks were maintained at 25 o C as described previously [14]. The details of 117 fly stocks, genetics and control genotypes used are in supplementary methods. 118 119

Immunostaining analysis 120
Drosophila third instar larval lymph glands were dissected in PBS as described before and 121 immunostained for microscopic analysis [20]. The detailed protocol and reagents used are in 122 supplementary methods. 123 124 Wasp parasitism assay 125 Wasp infestation was performed following standardised protocol as described in Rodrigues et al., 126 2021a [16]. Details are mentioned in supplementary methods. 127 128

Quantification and statistical analysis 129
Blood cell differentiation was quantified as described in Ray et al., 2021 [17]. To explore whether ESCRT function may determine blood progenitor identity or potential, we 139 depleted each of the 13 core ESCRT components individually in the lymph gland by RNAi mediated 140 knockdown (KD) in domeless (dome) expressing blood progenitors (domeGal4>UAS ESCRT RNAi). As 141 ESCRT plays an active role in ubiquitinated cargo sorting, the accumulation of ubiquitinated cargoes 142 serves as a hallmark of dysfunctional ESCRT machinery and impaired endosomal protein sorting. A 143 comprehensive analysis of conjugated ubiquitination (Ub) status (see methods) in the primary, 144 secondary and tertiary lobes by immunostaining the complete lymph gland showed a range of effects 145 with the phenotype varying among ESCRT components within a given ESCRT complex and between 146 complexes. 147 Control LG showed low or no Ub in primary, secondary, and tertiary lobes (Fig 1A, B). A similar trend 148 was seen on depletion of ESCRT-0 components Hrs or Stam, with an occasional increase in Ub in 149 primary lobes, which was not significant ( Figure 1A, B). In contrast ESCRT-I, -II and -III depletion had 150 effects on all lobes, though not all components affected the Ub status. Among ESCRT-I components 151 (Vps28,Tsg101,Vps37A,Vps37B), depletion of Vps28 or Tsg101 very significantly increased Ub in the 152 primary lobe, Vps28 and Vps37A affected the secondary lobe and Vps37A showed an increase in Ub 153 in the tertiary lobe. Vps37B depletion had a mild non-significant effect on the Ub status of the LG 154 ( Figure 1A, B). ESCRT-II components Vps25, Vps22 and Vps36 had no effect on the primary lobe. 155 However, Vps22 depletion caused a dramatic increase in Ub in the secondary and tertiary lobes, where 156 Vps25 and Vps36 depletion had no effect. Finally, depletion of ESCRT-III components (Vps32,Vps20 157 and Vps2) caused a significant increase in Ub in the primary lobes whereas secondary and tertiary 158 lobes were sensitive only to Vps20 depletion. 159 In summary, our data indicates non-uniform response of progenitors to perturbation of the cargo 160 sorting machinery, depending on the ESCRT component that is depleted as well as the target 161 progenitor population ( Figure 1B schematic). Of the 13 core ESCRT components, 7 caused increased 162 Ub in the LG when depleted. Interestingly, the effects were not uniform amongst progenitor subsets 163 -5 affected Ub status in the primary lobes, 4 in the secondary lobes and 3 in the tertiary lobes. This is 164 in agreement with the anterior-posterior developmental and functional heterogeneity of progenitors 165 reported earlier [16] and suggests that younger progenitors have a reduced requirement for ESCRT 166 function. Thus, our analysis provides a spatiotemporal correlation of Ub status to ESCRT depletion in 167 LG progenitor subsets. We next tested whether this correlation reflects the response of progenitors 168 to maintenance and differentiation cues. 169 170 ESCRT components play distinct roles in lymph gland progenitor maintenance. 171 As ESCRT components regulate ubiquitinated cargo sorting in the blood progenitors, they might 172 potentially regulate progenitor homeostasis. Hence, we checked whether depleting the 13 ESCRT 173 components individually from dome + progenitor subsets (marked by GFP expression) 174 knockdown did not show any significant change in progenitor status indicating non-essential roles for 183 these in progenitor maintenance (Fig 2A, B). Depletion of ESCRT-I components Vps28 and Tsg101 184 caused reduction in progenitor fraction in the primary lobes whereas secondary lobe progenitors were 185 reduced by depletion of Vps28, Vps37A or Vps37B but not of Tsg101. Interestingly, ESCRT-I 186 components had no effect on tertiary lobe progenitors. Among ESCRT-II components, Vps25 had an 187 effect on proliferation causing an absolute increase in primary lobe cell numbers with a concomitant 188 decrease in progenitor fraction (Fig 2A, B; S1A, C; S6; Supplementary results). Vps22 also did not affect 189 LG progenitor fraction. In contrast, Vps36 drastically reduced progenitor fraction in all LG lobes, with 190 phenotype severity increasing from anterior to posterior. ESCRT-III had very restricted effects on 191 progenitors with Vps32 KD causing a reduction only in anterior progenitors, Vps2 KD reduced both 192 anterior and posterior progenitors and Vps24 and Vps20 had no effect ( Fig Vps32, that had no effect on progenitors, caused an increase in plasmatocyte numbers only in the 222 secondary lobes. This indicates that though there is no effect as assessed by progenitor marker 223 analysis, Hrs or Vps32 depletion has sensitized the tissue to respond to proliferation and 224 differentiation cues. Along similar lines, Vps36 and Vps2 depletion drastically reduced the secondary 225 progenitor fraction and increased plasmatocyte differentiation. Except for Vps28, ESCRT KD did not 226 induce plasmatocytes in the tertiary lobes, even when progenitors were lost (e.g. Vps36 KD and Vps2 227 KD). Notch signaling, which may contribute to crystal cell differentiation and also differential response of 278 the progenitor subsets. 279 NICD cleavage and transport to the nucleus to activate target gene transcription is key to effecting 280 canonical and non-canonical modes of Notch signaling. Accumulation of NICD may lead to aberrant 281 activation of Notch signaling. Progenitor-specific knockdown of all tested ESCRT components, except 282 Vps25, resulted in increase in the number of cells accumulating NICD in the primary lobe ( Fig 4B). This 283 suggests a role for a majority of the ESCRT components in NICD trafficking, which may affect Notch 284 signaling. The absence of any phenotype due to Vps25 knockdown suggests compensatory 285 mechanisms may regulate cargo trafficking and lineage-specific signaling activation, which is sufficient 286 to maintain progenitors at steady state. 287 Knockdown of 4 components [Hrs, Stam (ESCRT-0), Vps28 (ESCRT-I) and Vps24 (ESCRT-III)] led to an 288 increase in the number of NICD accumulating cells in the secondary lobe (Fig 4B, S4A). However, only 289 Hrs and Stam knockdown resulted in NICD accumulation in the tertiary lobe. This is in concordance 290 with the phenotype of Notch activation upon Hrs, Stam and Vps28 knockdown in the posterior lobes. 291 Our results show that NICD accumulation and Notch pathway activation correlate perfectly with 292 crystal cell differentiation upon ESCRT knockdown. Notch signaling is known to be sensitive to 293 endocytic sorting defects due to ESCRT in other contexts [24]. Similar effects may result in ectopic 294 activation and promoting crystal cell differentiation. Despite a differential effect on NICD trafficking 295 and Notch activation across progenitor subsets, immunolocalization and proximity ligation assay 296 indicated uniform interaction of ESCRT with NICD (Fig S4B, C; see supplementary results). Also, Notch 297 activation triggered by ESCRT depletion in blood progenitors may be independent of the status of 298 Notch ubiquitination (Fig S4D, E;  only Vps36 knockdown triggered lamellocyte differentiation in the tertiary lobe (Fig 5A, B). This 308 indicates that the majority of the ESCRT components are not involved in suppressing lamellocyte 309 differentiation in the refractile posterior progenitors at steady state. However, it is likely that KD 310 progenitors may be more sensitive to immunogenic cues as compared to normal, unperturbed 311 progenitors. 312 Wild type larvae are generally able to mount a sufficiently robust immune response against wasp 313 infestation, that aids their survival and eclosion. Systemic signals are generated upon wasp infestation 314 and are received by the lymph gland progenitors [25], possibly through a complex extracellular matrix 315 [16]. This results in lamellocyte differentiation in the primary lobe followed by disintegration and 316 release of lamellocytes into circulation. Secondary and tertiary lobes are refractile to wasp infestation 317 and do not form lamellocytes even upon immune challenge. Hence, we chose to test the response to 318 wasp infestation in-a] ESCRT KD that had no effect on lamellocyte formation ( Vps25 KD) and b] ESCRT 319 KD that caused lamellocyte differentiation only in the primary lobe ( Vps32 KD). Knockdown of both 320 Vps25 and Vps32 triggered lamellocyte differentiation across all progenitor subsets upon immune 321 challenge with wasp ( Fig 5C). This shows that Vps25 and Vps32 play essential roles in preventing all 322 posterior progenitors from lamellocyte differentiation in response to a natural immune challenge. 323 Further, loss of ESCRT sensitizes progenitors to systemic cues by unlocking differentiation programs. 324 Our detailed analyses of blood progenitor differentiation upon knockdown of the 13 core ESCRT 325 components yielded a functional map that reflects distinct lineage-specific roles of ESCRT in blood cell 326 homeostasis and reduced sensitivity of younger progenitors to endocytic perturbation (Fig 6 A, B). 327

ESCRT regulates cargo sorting in a cell-autonomous manner in blood progenitors. 328
Progenitor-specific downregulation of ESCRT expression leads to accumulation of ubiquitinated 329 cargoes. However, a majority of the ubiquitin aggregates were found to accumulate in non-progenitor 330 (domeless -) cells, suggesting a possible cell non-autonomous effect. To test whether this could be due 331 to ubiquitin accumulation in progenitors prior to their differentiation (cells lose domeless marker 332 expression), we generated homozygous mutant mitotic recombinant clones for a representative 333 ESCRT gene Vps32 (shrub), in progenitors. Vps32 is a terminally acting ESCRT that affects ubiquitinated 334 cargo sorting and its depletion affects all blood cell lineages (Fig. 6). Hence it serves as a good model 335 to assess cell autonomous function of ESCRT. Staining for conjugated ubiquitin revealed accumulation 336 of ubiquitin aggregates in the homozygous mutant patch of the tissue (GFP -) indicating Vps32 has a 337 cell autonomous role in cargo sorting in blood progenitors ( Fig 7A). Hence, it is likely that ubiquitin 338 seen in domecells ( Fig. 1) accumulated when the cells were still expressing domeless. This suggests 339 that despite a decrease in dome expression in the knockdown cells, ubiquitin aggregates may persist 340 during differentiation, likely due to dysfunctional cargo sorting in a cell-autonomous manner. 341

ESCRT affects progenitor differentiation in a cell non-autonomous manner. 343
We analysed differentiation in progenitor-specific mitotic clones of ESCRT. Vps32 knockdown results 344 in increased crystal cell differentiation and triggers lamellocyte differentiation, as described earlier 345 (Fig 3, 5-6). ProPO staining showed both wild type and mutant origin of crystal cells as revealed by 346 overlap with GFP expression in the mutant tissue ( Fig 7B). As crystal cells are usually present in the 347 lymph gland in low numbers, it is difficult to interpret cell-autonomous origin of crystal cells from 348 mutant progenitors. However, lamellocytes are completely absent in the control lymph gland at 349 steady state (Fig 7C). Phalloidin staining in the Vps32 mutant clone showed GFP + elongated or 350 coalescing cells, indicating the presence of lamellocytes and possibly their precursors (Fig 7C). This 351 suggests non-autonomous regulation of lamellocyte differentiation by ESCRT. Hence, ESCRT may 352 regulate progenitor differentiation in both a cell-autonomous as well as non-cell-autonomous 353 manner. 354 Vps25 knockdown did not affect the status of ubiquitination, progenitor maintenance or 355 differentiation to any particular blood cell lineage despite its expression in the lymph gland. To further 356 verify this, we generated lymph gland progenitor-specific homozygous mitotic clones of Vps25 loss of 357 function mutation (Vps25 A3 ). There was no accumulation of ubiquitin aggregates (Fig S5A) or any 358 change in the status of the progenitor (Fig S5B), plasmatocyte (Fig S5C), crystal cell ( Fig S5D) and lamellocyte differentiation (Fig S5E). However, the mutant lymph glands showed enlargement of the 360 primary lobe, suggesting possible increase in blood cell proliferation upon loss of Vps25, due to non-361 autonomous effects. Also, phalloidin staining revealed appearance of binucleate, large cells and also 362 very small cells occasionally, along with increase in F-actin content in some patches of the tissue, 363 mostly in a cell autonomous manner (visible in GFP negative area of the tissue) ( Fig S5E). Hence, Vps25 364 possibly inhibits uncontrolled cell proliferation and may contribute to critical steps of cell division that 365 may dictate cell shape, number and polarity. Curiously, we observed drastic functional diversity of ESCRT-II components in progenitor fate 390 specification. While Vps36 depletion affected all lineages across progenitor subsets, Vps25 depletion 391 did not affect differentiation. Loss of Vps25 caused hyperproliferation in blood cells and failed to 392 activate signaling pathways such as Notch, which are necessary for progenitor differentiation. Though 393 Vps25 is a critical player in endosomal protein sorting in epithelial tissues [23,26], its redundancy in 394 lineage-specific differentiation suggests alternate routes for endosomal protein sorting in blood 395 progenitors or a temporally regulated, developmental stage-specific role that has not yet been 396 identified. Notably, though Vps25 is dispensable for steady state hematopoiesis, its depletion 397 sensitizes all progenitors to differentiate upon immune challenge. This supports the possibility that 398 the diverse roles of ESCRT may contribute to differential regulation of steady state and stress 399 hematopoiesis.   in the lymph gland upon progenitor-specific knockdown of 12 core ESCRT components (All except 611 Vps25). Dome>2xEGFP (green) marks the progenitors across different lobes. Arrowheads mark 612