Coupling adipose tissue architecture and metabolism via cytoophidia

Tissue architecture determines its unique physiology and function. How these properties are intertwined has remained unclear. Here, we show that the metabolic enzyme CTP synthase (CTPS) form filamentous structures termed cytoophidia along the adipocyte cortex in Drosophila adipose tissue. Interestingly, loss of cytoophidia, whether due to reduced CTPS expression or a point mutation that specifically abrogates its polymerization ability, leads to downregulated Collagen-Integrin signaling, weakened adipocyte adhesion, and defective adipose architecture. Strikingly, CTPS specifically binds with Integrin subunit α2, which influences Integrin function and Collagen IV deposition. cytoophidia promote Collagen IV mRNA expression and thus its extracellular deposition to strengthen adipocyte adhesion. Remarkably, Collagen IV-Integrin signaling reciprocally regulates cytoophidium formation at a post-translational level. Together, we demonstrate that a positive feedback signaling loop containing both cytoophidia and Integrin adhesion complex couples tissue architecture and metabolism in the fly adipose.


INTRODUCTION
Tissue structure or architecture refers to the organization of individual cells and their extracellular matrix (ECM) in such a way that it is best suited to perform the unique physiological function of each tissue. It is mainly accomplished, irrespective of whether the tissue of concern is an epithelium or a connective tissue such as the adipose, by tissue-specific regulation of cell-cell and cell-ECM adhesion (Mian et al. 2003;Pozzi et al. 2017). Despite its importance, how tissue architecture determines or is determined by its physiological functions, for example, its metabolism, has remained largely unclear.
As a metabolic organ and an equivalent of the mammalian adipose tissue, the Drosophila fat body has in recent years emerged as an excellent model to understand fat metabolism, due in part to its relatively simple and welldefined tissue organization and architecture (Baker and Thummel 2007; Arrese and Soulages 2010; Li et al. 2019). Briefly, the fly fat body is a singlelayered tissue of adipocytes encapsulated by a dense basement membrane (BM) (Baker and Thummel 2007;Pope et al. 2016). Similar to the BM anchoring the basal side of an epithelium, the BM in the fat is also composed of multiple ECM components, including the Laminin and the Collagen IV network of polymers, which are interconnected by numerous "adaptor/linker" proteins such as Nidogen and Perlecan (Cummings et al. 2016;Jayadev and Sherwood 2017). Drosophila Collagen IV polymers are formed by an α1 chain and an α2 chain (Natzle et al. 1982;Fessler and Fessler 1989). In addition to contributing to forming the BM, Collagen IV and other ECM 4 / 60 components are also deposited between adipocytes. Indeed, intercellular deposition of Collagen IV, via binding to its membrane-receptor Integrin, is essential for cell-matrix adhesion and the fly adipose tissue organization (Dai et al. 2017).
As the primary ECM receptor, Integrins are heterodimeric membrane proteins composed of an α and β subunit (Shattil et al. 2010). In Drosophila, the Integrin family consists of five αPS subunits (αPS1 to 5) and two β subunits (βPS and βν) (Brown 1993;Brower et al. 1995). Together with its co-receptor Syndecan (Sdc) and cytoplasmic downstream components, including Integrin-linked kinase (ILK), Particularly interesting cysteine-histidine-rich protein (PINCH), and Talin, Integrins form an "adhesion complex," which binds to the actin cytoskeleton and transduces environmental signals to the cytoplasm and the nucleus (Hannigan et al. 2005;Morgan et al. 2007;Dai et al. 2017). This integrin adhesion complex thus serves as a gateway linking both the microenvironment and the inside of a cell and plays an essential role in mechanosensing, cell migration, differentiation, etc. (Diamond and Springer 1994;Giancotti and Ruoslahti 1999;Hynes 2002;Campbell and Humphries 2011;Kechagia et al. 2019;Moreno-Layseca et al. 2019).
The metabolic enzyme Cytidine 5'-triphosphate synthase (CTPS) catalyzes the conversion of UTP to CTP, a rate-limiting step in the de novo synthesis of CTP (Lieberman 1956;Levitzki and Koshland 1969). The active enzyme is a homo-tetramer with each monomer consisting of a glutamine aminotransferase domain, a kinase-like ammonia ligase domain, and an αhelical linker (Levitzki and Koshland 1976;Robertson 1995). Remarkably, it has been widely observed in both prokaryotes and eukaryotes that CTPS 5 / 60 often forms filamentous structures termed cytoophidia although the functional significance of filament formation has remained unknown (Ingerson-Mahar et al. 2010;Liu 2010;Noree et al. 2010;Carcamo et al. 2011;Liu 2011;Liu 2016;Daumann et al. 2018;Zhou et al. 2020).
We recently found that cytoophidia exist in the fly fat body (Zhang et al. 2020).
Here, we hypothesized that cytoophidia are essential for the fly adipocytes.
Surprisingly, we found that cytoophidia regulate Integrin and Collagen IV function and play an important role in fat body adipocyte adhesion and tissue architecture.

Cytoophidia localize at the cortex of fly adipocytes
We first determined CTPS expression in the fat body during various stages of fly development. To visualize the subcellular localization and dynamics of endogenous CTPS in vivo, we created a "knock-in" fly line in which the coding sequences of the fluorescent protein mCherry and the V5 tag were inserted in-frame at the C-terminus of CTPS, referred hereafter to the CTPS-mCh line. Using Western blotting analysis, we confirmed that the V5-tagged CTPS protein was expressed at the first-instar larval (L1), the second-instar larval (L2), and the early, middle, and late third-instar larval (L3) stages ( Fig.   1A, B). Using fluorescent confocal microscopy, we found that cytoophidia were present at all of the stages examined (Fig. 1C). Interestingly, however, both the lengths and numbers of cytoophidia progressively reduced from the early L3 stage to the late L3 stage (Fig. 1D, E).
Next, we used Stimulated Emission Depletion (STED) microscopy to examine the subcellular localization of cytoophidia at a high resolution. We found that almost all cytoophidia were located at the cell cortex, as marked by intense phalloidin staining of the actin filaments ( Fig. 1Fi-ii). This was in contrast to the nuclei, which were centrally located in the adipocytes ( Fig.   1Fi-ii). To determine whether this striking feature of cytoophidium localization was an artifact introduced by protein fusion between CTPS and mCherry or the V5 tag, we performed immunofluorescence microscopy and directly detected CTPS protein expression in the fat body of the wild type w 1118 fly line at the L2 stage. Again, we found that cytoophidia, as detected by the CTPS antibody, were present at the adipocyte cortex (Supplemental Fig. 1A), 7 / 60 thus confirming the results from the CTPS-mCh knock-in line. Together, the results show that cytoophidia are present in the fly fat-body adipocytes and that they align along the cortex of adipocytes.

CTPS reduction causes defective adipocyte adhesion
The cortical localization of cytoophidia correlated with adipocyte adhesion sites. Therefore, we asked whether cytoophidia play a role in adipocyte adhesion. To answer this question, we used the RNAi technology and created a fly line in which CTPS expression levels were specifically reduced in the fat body under the control of the Cg GAL4 driver (Cg GAL4>CTPS-RNAi), referred to as the CTPS-Ri hereafter. Using qPCR, we found that CTPS mRNA expression in the fat body was reduced by ~73% in the CTPS-Ri larvae when compared with the wild-type larvae (Cg GAL4>w 1118 ) at the L3 stage ( Fig. 2A). Unlike the control fat body at this stage, which was a thin, tight, ribbon-like structure, we found that the CTPS-Ri fat body was loosely connected, often with adipocytes breaking off of the adipose tissue (Fig. 2B).
We speculated that the mutant fat-body suffered from an adhesion defect where adipocytes were not as tightly connected as they should. Therefore, we next examined the tissue integrity of the CTPS-Ri fat body at the L2 stage, before the tissue dissociation phenotype became apparent at the L3 stage. Using fluorescent confocal microscopy to examine the actin cytoskeleton and the nuclei, we observed the typical polygonal adipocytes with tight cell-cell contacts in the wild-type w 1118 fat body (Fig. 2C).
Remarkably, the mutant fat body contained many gaps at the bi-cellular or tri-cellular contact sites (Fig. 2D). By performing Z-stack image acquisition and reconstruction, we confirmed that intercellular gaps were present 8 / 60 throughout the Z-axis, including in the middle and at the top of the singlecell-layered adipose tissue (Fig. 2D, E). Together, our results show that CTPS reduction leads to weakened adipocyte adhesion and dissociation of the adipose tissue in the fat body.

CTPS polymerization promotes its protein levels
Next, we asked whether it is the filaments, rather than the protein expression level per se, of CTPS that are essential for adipocyte adhesion. To answer this question, we needed to specifically disrupt the polymerization of CTPS proteins without affecting their enzymatic functions. Previous studies showed that the Histidine amino acid at the 355th position, or His 355 , of human CTPS protein is essential for its polymerization, but not enzymatic function (Lynch et al. 2017;Sun and Liu 2019). By aligning the amino-acid sequences of both human and fly CTPS proteins and performing structural analysis, we identified that the His 355 amino acid of the fly CTPS in the glutamine amidotransferase domain would also be at the interface between two consecutive tetramers and thus should also be critical for filament formation ( Fig. 3A, B).
We predicted that a replacement of His 355 with Alanine, which is uncharged and smaller than Histidine, would disrupt the polymerization ability of the fly CTPS protein in a similar way to the human CTPS situation. Thus, using the CRISPR/Cas9-mediated knock-in technology, we created a point mutant that would lead to His 355 to Ala 355 conversion in the endogenous CTPS protein.
Moreover, the mutant CTPS was fused with mCherry and three hemolymph agglutinin tags (3x HA) at its C-terminus so that the fusion protein could be easily detected in various assays (Supplemental Fig. 2). Using the genomic 9 / 60 DNA extracted from the knock-in fly line, referred to as the CTPS MU -mCh line hereafter, we confirmed the intended point mutation was successfully introduced as designed (Supplemental Fig. 3). The homozygous CTPS MU -mCh adult flies were viable, with no discernible developmental defects other than a smaller body size.
Next, we examined the adipose tissues in both the control CTPS-mCh and the CTPS MU -mCh flies. As expected, we observed numerous cytoophidia in the control adipocytes, but none in CTPS MU -mCh adipocytes (Fig. 3C, C', D, D'). These results thus confirmed our above prediction that the mutant CTPS proteins would be unable to polymerize, and suggest that, as in the human situation, His 355 is also essential for fly CTPS proteins to form filaments.
Furthermore, phalloidin staining of the actin cytoskeleton showed that numerous gaps emerged between adipocytes in the mutant but not the control adipose tissues, indicating that adipocyte adhesion is defective in the mutant fly line (Fig. 3C'-E). Interestingly, CTPS protein levels, as detected by Western blotting analysis using an antibody against the mCherry fusion protein, were greatly reduced in the CTPS MU -mCh fly larvae when compared with the control (Fig. 3F, G). The data thus show that a failure to form CTPS polymers leads to a reduction in its protein levels.
Thus, while adipocyte adhesion is defective in the fat body of the CTPS MU -mCh fly larvae, we concluded that the defect is most likely a secondary consequence caused by reduced CTPS expression, rather than a direct result from a lack of cytoophidia in the mutant flies. Moreover, the data suggest that filament formation increases CTPS protein levels in fly 10 / 60 adipocytes, presumably by stabilizing the otherwise labile unpolymerized proteins.
Cytoophidia, rather than CTPS protein levels, are essential for adipocyte adhesion To definitively determine whether cytoophidia are essential for adipocyte adhesion, we needed to disrupt filament formation without reducing its protein expression levels. To this end, we created a transgenic fly line, the CgG4/CTPS MU-OE , in which a mutant CTPS protein, tagged with mCherry-HA at its C-terminus, was forcefully expressed in the fat body as controlled by the Cg GAL4 driver. We used Arg 355 , which is negatively charged and is bulkier than Alanine used above, to replace His 355 to ensure that the mutant CTPS protein is incapable of forming filaments. As a control, a transgenic line CgG4/CTPS WT-OE , in which the wild-type CTPS tagged with mCherry-HA was also created. Western blotting analysis confirmed that protein expression of the mutant CTPS was at a similar level as the wild-type CTPS CTPS-mCh adipocytes (Fig. 5K). Similar results were also observed when we examined Cg25C mRNA expression, although it was also significantly increased in the CgG4/CTPS MU-OE adipocytes (Fig. 5K).
Together, the combination of loss-and gain-of-function studies of CTPS conclusively demonstrate that cytoophidia specifically promote the production of Collagen IV, rather than other ECM components such as Nidogen. Further, the data show that cytoophidia regulate Collagen IV expression at the transcriptional level.

Integrin signaling promotes cytoophidium formation
The localization of cytoophidia at the adipocyte cortex, where cell-ECM adhesion occurs, prompted us to ask whether Collagen IV-integrin signaling, which mediates adipocyte adhesion, regulates cytoophidium formation. To Likewise, we also determined whether reduced expression of Cg25C (Collagen at 25C) or Vkg (Viking), which encodes the Collagen IV α1 chain or α2 chain, respectively, affected cytoophidium formation. In both cases, we 14 / 60 observed defective adipocyte adhesion and concurrent absence of cytoophidia in fly lines with reduced expression of Cg25C or Vkg (Fig.6F, G, K, L). By contrast, we found that adipocyte adhesion was normal and cytoophidia were present in fly lines with reduced expression of Laminin, Perlecan, or Nidogen, all of which were also ECM components (Fig. 6H-L).
The data thus suggest that Integrin signaling, and specifically that activated by Collagen IV rather than other ECM components, regulates cytoophidium formation.

CTPS binds to integrin and colocalizes with integrin signaling complex
Next, we sought to determine how Integrin signaling promotes cytoophidium formation. To this end, we first examined whether a reduction in Integrin signaling may affect CTPS mRNA or protein expression level. Using qPCR, we found CTPS mRNA expression was not significantly altered in the fly lines expressing the Mys-Ri, Stck-Ri, or the Ilk-Ri allele (Fig. 7A). Likewise, Western-blotting analysis showed that CTPS protein levels were not significantly changed by reduced Integrin signaling in any of these fly lines ( Fig. 7B, C). Together, these results show that Integrin signaling does not regulate CTPS at either the mRNA or protein expression level.
Alternatively, Integrin signaling may promote cytoophidium formation by binding to its filament precursors and promote one or more steps during its polymerization. Therefore, we next tested whether CTPS could bind to Integrin. We "tagged" CTPS and Integrin with the HA and the V5 peptides, respectively, and expressed them in the S2 cells. These fusion proteins were then subjected to co-immunoprecipitation (Co-IP) assays to test their potential binding. We found that CTPS was able to bind to Integrin (Fig. 7D).

/ 60
Moreover, if CTPS binds to Integrin proteins inside living cells, they should co-localize at the same places. To test this possibility, we crossed the If-GFP and the CTPS-mCh fly lines and examined Integrin and CTPS expression in the adipocytes. We found that Integrin and cytoophidia often colocalized in the fly adipocytes (Fig. 7E, E', H). Likewise, cytoophidia also colocalized with other components of the Integrin signaling, including Syndecan (Sdc), an Integrin co-receptor, and ILK, which were labeled using the Sdc-GFP and the Ilk-GFP trap lines ( Fig. 7F-H).
Taken all together, the above data show that Integrin signaling does not affect CTPS mRNA or protein expression; however, Integrin and most likely other components of the adhesion complex can bind to CTPS protein to promote filament formation.

DISCUSSION
Tissue architecture is essential for organ physiology and function. How it is regulated by individual cell behavior, including cellular metabolism, has remained largely unclear. Here, we show that the metabolic enzyme CTPS regulates cell adhesion and tissue organization of the Drosophila adipose.
CTPS forms filaments at the adipocyte cortex, corresponding to the Integrin-Collagen IV contact sites essential for adipocyte adhesion. Surprisingly, loss of CTPS expression causes defective cell adhesion and tissue dissociation, suggesting that CTPS promotes adipocyte adhesion. Interestingly, expression of mutant CTPS proteins, which have normal enzymatic activities but are unable to form filaments, also leads to defective adipocyte adhesion, suggesting that the polymer form, rather than the absolute amount, of CTPS proteins, is essential for adhesion and tissue architecture. We show that cytoophidia promote Collagen IV mRNA expression and thus its deposition to strengthen Integrin-Collagen IV adhesion. Moreover, a reduction of Integrin signaling due to loss of Integrin or other components of the signaling pathway leads to failure of cytoophidium formation. We show that Integrin signaling does not affect CTPS mRNA expression, but increases the protein expression by promoting its filament formation, which stabilizes the otherwise labile monomers and tetramers. This is supported by the binding and colocalization of components of the Integrin adhesion complex with cytoophidium.

Coupling Adipose Tissue Architecture and Metabolism via Integrin
Feedback Signaling

/ 60
One of our most striking observations was the cortical location of cytoophidia, which were aligned with adipocyte-matrix contact sites known to require Integrin-Collagen IV interactions (Dai et al. 2017). Indeed, loss of Integrin signaling activities, for example, as a result from reduced expression of Collagen IV, Integrin, Syndecan, or downstream signaling components PINCH, or ILK, all led to a failure of cytoophidium formation. These data thus conclusively demonstrate that Integrin-Collagen IV signaling promotes cytoophidium formation (Fig. 8).
Remarkably, the formation of cytoophidia also promotes Integrin-Collagen IV-mediated adipocyte adhesion. In the absence of cytoophidia, whether due to reduced CTPS expression or forceful expression of a mutant CTPS protein that prevents the polymerization event, in turn, led to reduced Integrin signaling, weakened adipocyte adhesion, and defective adipose tissue architecture (Fig. 8). Together, our data demonstrate that there exists a signaling feedback loop between cytoophidium formation and Collagen IV in the matrix, by which intrinsic cellular metabolism of adipocytes is coupled to extrinsic tissue architecture of the adipose (Fig. 8). We speculate that, by coupling these two essential characteristics, the fly adipose, and possibly other tissue and organ systems can best match their structure and physiological function to operate at an optimal condition.
Our data show that the amount of cytoophidia correlated with Collagen IV protein expression levels, suggesting that cytoophidia promote Collagen IV expression. Interestingly, a similar, though weaker, correlation was also observed between cytoophidium amount and Integrin expression levels (Fig.   5). This difference in the amount of promotion between Collagen IV and 18 / 60 Integrin protein expression could be partly explained by the notion that Collagen IV functions upstream of Integrin and thus its increase due to cytoophidium formation secondarily upregulated Integrin expression levels, although to a lesser extent.
It remains unclear the mechanism by which cytoophidia regulate Integrin signaling in fly adipocytes. Surprisingly, CTPS mRNA levels, and possibly the amount of cytoophidia, appeared to be directly correlated with Collagen IV mRNA expression levels (Fig. 5K). These data suggest that CTPS promote Collagen IV expression at the transcriptional level, presumably indirectly via other protein partners. We await with great anticipation the exciting discoveries of these protein partners that, together with CTPS, may regulate Collagen IV expression and Integrin signaling.

Integrin signaling-independent and dependent stages during cytoophidium formation
We show that the replacement of the wild-type CTPS protein by a mutant form, in which H 355 is converted into A 355 or R 355 , in the CTPS MU -mCh or the CgG4/CTPS MU-OE fly lines caused a failure of filament formation in the fly adipocytes. The data are thus consistent with a previous report showing that the H 355 is a crucial amino-acid residue for tetramer formation in human CTPS proteins (Lynch et al. 2017). Moreover, CTPS protein levels were also reduced in the CTPS MU -mCh adipocytes when compared to normal.
Interestingly, despite a reduction in protein expression, CTPS mRNA expression in the CTPS MU -mCh adipocytes was similar to that in the control (Fig. 5K). These data thus suggest that filament formation promotes CTPS 19 / 60 protein expression at a post-transcriptional level, presumably by stabilizing the highly labile monomers and tetramers (Fig. 8).
Remarkably, however, despite a positive control of Integrin signaling on cytoophidium formation, we observed no reduction in CTPS mRNA or protein expression levels in the fly lines, in which Integrin signaling was downregulated. Together, these data suggest that cytoophidium formation is a multi-step process. Specifically, it includes an early, Integrin-independent step whereby the unstable monomers and tetramers form the stable polymers; and a late, Integrin-dependent step whereby polymers further bundle into a higher-order filament, both of which are stable (Fig. 8).
Moreover, considering that the endogenous, wild-type CTPS proteins would Alternatively, Integrin signaling may regulate cytoophidium formation at a metabolic level. Indeed, Integrin signaling is known to regulate glucose metabolism and glutamine transportation, whose substrates may be used by

Generation of transgenic flies
The CRISPR/Cas9 technology was used to establish both the C-terminal mCherry-4V5 tagged CTPS knock-in fly and C-terminal mCherry-3HA tagged CTPS H355A mutation knock-in fly according to homology-directed repair procedures as previously described (Bassett et al. 2013 Step 2, the cDNAs encoding Drosophila CTPS was produced by RT-PCR using the total RNAs extracted from the w 1118 line.
Then, mCherry-3HA sequence was ligated to the C-terminal of CTPS.
Step 3, the offspring flies from step2 were crossed with cre fly to remove nonessential sequence between two Loxp. Thus, C-terminal mCherry-3HA tagged CTPS H355A mutation knock-in fly were finally generated. Then, site-specific integration of pUASTattB-CTPS or pUASTattB-CTPS H355-R plasmid into fly germ line (attp2) that contain attp lading sits was carried out by co-injection with phiC31integrase RNA as previously described 48 at the Core Facility of Drosophila Resource and Technology, SIBCB, CAS.

Fly husbandry and diet preparation
Fly lines were raised on the standard yeast-cornmeal-agar food. To make sure larvae used in this study at the desired developmental stage, we restricted egg collections by allowing female to lay eggs for less 4 hours. All 24 / 60 flies were cultured at 25°C with 50% humidity under a 12 h/12 h light/dark cycle.

Protein expression constructs
For expression of the wild-type CTPS, and If, cDNA fragments were amplified by RT-PCR using the total RNAs extracted from the w 1118 line. The in PBS twice for 5min each and mounted in Vecta shield with DAPI (Vector Labs).
Cell transfections were carried out by using FuGENE HD Transfection Reagent (E2311, Promega, WI) according to the manufacturer's instructions. containing 1X protease inhibitor cocktail (Bimake) for 2 hours on ice. After centrifuging 15000g at 4oC, the supernatants were collected and incubated with anti HA magnetic beads (Bimake) or IgG bound protein A/G magnetic beads (Bimake) followed by an overnight rotation at 4oC. Then the beads were washed three times with 1 ml PBST (0.5% Tween20 in PBS) for 5 minutes, followed by SDS-PAGE and immunoblotting analysis after elution by boiling in 50 μl 1× SDS loading buffer. Each data point represents one image.

Western blot
Larvae or fat body tissues were extracted in RIPA buffer (  steps that were experimentally manipulated in the current study were marked in red and by an asterisk (*). As a result, Integrin-mediated cell adhesion is reduced, resulting in defective adipose architecture.