Adipose triglyceride lipase promotes prostaglandin-dependent actin remodeling by regulating substrate release from lipid droplets

A key factor controlling oocyte quality and fertility is lipids. Even though lipid droplets (LDs) are crucial regulators of lipid metabolism, their roles in fertility are poorly understood. During Drosophila oogenesis, LD accumulation in nurse cells coincides with dynamic actin remodeling necessary for late-stage follicle morphogenesis and fertility. Loss of the LD-associated Adipose Triglyceride Lipase (ATGL) disrupts both actin bundle formation and cortical actin integrity, an unusual phenotype also seen when Pxt, the enzyme responsible for prostaglandin (PG) synthesis, is missing. Dominant genetic interactions and PG treatment of follicles in vitro reveal that ATGL and Pxt act in the same pathway to regulate actin remodeling, with ATGL upstream of Pxt. Further, lipidomic analysis detects arachidonic acid (AA) containing triglycerides in ovaries. Because AA is the substrate for Pxt, we propose that ATGL releases AA from LDs to drive PG synthesis necessary for follicle development. We also find that exogenous AA is toxic to follicles in vitro, and LDs modulate this toxicity. This leads to the model that LDs both sequester AA to limit toxicity, and release AA via ATGL to drive PG production. We speculate that the same pathways are conserved across organisms to regulate oocyte development and promote fertility.

of LD proteins in flies and regulates lipid metabolism by protecting the stored triglycerides from ( Figure 1E), we asked whether these LD regulators have a role in these events. F-actin present 179 at the cortex of the nurse cells throughout early stages of oogenesis ( Figure 1F and G) 180 dramatically thickens during S10B ( Figure 1H); in addition, actin bundles form that hold the 181 nurse cell nuclei in place during nurse cell dumping ( Figure 1H-J). We find that females lacking 182 either PLIN2 or ATGL display two types of actin defects (

187
The combination of actin defects due to loss of PLIN2 or ATGL are rarely seen in other 188 mutants, as most actin regulators impact either actin bundle formation or cortical actin, but not 189 both (Wheatley et al., 1995;Buszczak and Cooley, 2000). However, the same combination of

274
Maximum projection of three confocal slices of representative S10B follicles stained for F-Actin  Defects. Images were brightened by 30% to increase clarity. Green arrowheads indicate

317
We utilized our in vitro egg maturation (IVEM) assay, in which isolated S10B follicles mature in vitro in a simple culture medium (Spracklen and Tootle, 2013). This assay does detect 319 the genetic interaction between Pxt and ATGL, as the majority of S10B follicles from single 320 heterozygotes of pxt and ATGL develop in vitro, while 55% of the follicles from double 321 heterozygotes fail to develop ( Figure 4A). These data recapitulate what we observed when we quantitatively assessed the actin defects in the double heterozygotes ( Figure 3H). Using the 323 same assay, we then tested the role of PGF2a downstream of ATGL. Treatment of S10B follicles 324 with 1.5mM aspirin (an inhibitor of COX enzymes, including Pxt) inhibits ~50% of follicle 325 development, and this is suppressed by addition of PGF2a ( Figure 4B). Of the ATGL mutant 326 follicles, only ~38% develop, but addition of PGF2a results in significant improvements, with

336
In the IVEM assay, ATGL and Pxt genetically interact, as ATGL/pxt follicle development is

366
Our lipidomic analysis did not uncover a significant difference in the content of 367 phospholipids ( Figure 5C) or of total triglycerides ( Figure 5D) between wild-type and ATGL 368 mutant ovaries. The overall FA profile in triglycerides was also similar between the two 369 genotypes ( Figure 5A and B). In addition, there was no significant difference in total triglycerides in S14 oocytes, as detected by thin layer chromatography ( Figure 5E      mechanism to prevent free AA from reaching toxic levels. To test if free AA is also toxic to development using a modified IVEM assay. One of the IVEM medium's key components is fetal 432 bovine serum (FBS), which supplies growth factors and numerous nutrients, including unknown 433 amounts of FAs such as AA. We therefore reduced the amount of FBS from 10% to 2.5%, which 434 still allows the majority of follicles to mature, and then assessed the effects of AA. At ~60µM AA, 435 a slightly higher percentage of follicles developed than in control medium, but that fraction 436 dropped steadily as AA concentrations were raised further ( Figure 6A). Notably, oleic acid (OA) 437 does not impair follicle development at any concentration ( Figure 6A). Thus, it is specifically free 438 AA that is dangerous to the follicles, which provides a rationale for why free AA is rapidly  ( Buszczak et al., 2002). If exogenous AA, as we hypothesize, is incorporated into LDs to protect 447 follicles, then reducing DGAT1 levels should enhance AA toxicity ( Figure 6B). We tested this 448 idea with our modified IVEM assay and treated follicles with 125µM AA. At this concentration,

450
These data support the model that exogenous AA is sequestered into LDs.

451
We next asked whether ATGL is required to release AA from internal LDs stores and 452 thus generate free AA. If this is true, then reducing ATGL levels should partially suppress the 453 toxicity of high levels of exogenous AA ( Figure 6B), as total free AA levels will be lower. Indeed, 454 two-fold more S10B follicles from ATGL heterozygotes develop in the presence of 500µM AA 455 than wild-type follicles ( Figure 6C). These data are consistent with the model that ATGL 456 releases AA from LDs.

457
Together these findings support the model that AA is stored in LDs to prevent toxicity 458 and that ATGL is required to release AA from LDs. This AA can then be used for PG production 459 and thus promotes actin remodeling necessary for follicle development.

543
AA freed from TAG by ATGL (red) might be supplied directly to Pxt or be first incorporated into 544 phospholipids (surrounding the neutral lipid core of LDs or present in the ER or other cellular 545 membranes). In a subsequent step, cPLA2 (light blue) would then release AA from these 546 phospholipids to provide the substrate for PG production. In the second pathway, PLIN2 (blue)

551
Our data reveal that the LDs produced during Drosophila oogenesis are not just lipid and 552 protein stores for the future embryo, but already play a critical role during follicle development.

553
In particular, loss of the triglyceride lipase ATGL results in cortical actin breakdown, cortical 554 contraction failure, and defective actin bundle formation during S10B of oogenesis, an unusual

574
Our data support the model that in Drosophila follicles LD triglycerides are a major 575 source of AA for PG production and that ATGL is responsible for its release. This AA might either directly serve as substrate for Pxt, or it might first be incorporated into phospholipids, to 577 be released by cPLA2 in a subsequent step (Figure 8). Such a model explains the genetic 578 interaction between Pxt and ATGL ( Figure 3B, D, E, and H and Figure 4A), the increase in AA-

579
containing ovary triglycerides in ATGL mutants ( Figure 5G and H), the fact that exogenous suppresses the lipotoxicity of exogenous AA ( Figure 6D).

582
Our studies also provide the first direct evidence that AA has a functional role during

639
It seems likely that LDs also buffer AA released from endogenous sources. Because             Figure 5G and 5H, the signal for two AA-containing triglyceride species was computed, and

951
TKM was removed and the volume of ovaries was estimated visually. Ovaries were kept on ice, 952 to which were added: 2 times the estimated ovary volume of TKM + 1M sucrose, protease 953 inhibitor cocktail to final concentration of 1X protease inhibitor cocktail (Sigma-Aldrich), and calyculin A serine/threonine phosphatase inhibitor (10µL per mL of volume; Cell Signaling).

955
Ovaries were homogenized on ice by grinding with an automated tissue grinder (KONTES pellet 956 pestle) for 1-2 minutes, and then 20µL ovary lysate samples were transferred from the total 957 lysate to 0.65mL Eppendorf tubes and stored at -80°C. The remaining ovary lysates were spun 958 at 13200rpm for 10 minutes in an Eppendorf Microcentrifuge (Model 5145D) at 4°C. The