Endothelial-adipocyte Cx43 Mediated Gap Junctions Can Regulate Adiposity

Abstract Obesity is a multifactorial metabolic disorder associated with endothelial dysfunction and increased risk of cardiovascular disease. Adipose capillary adipose endothelial cells (CaECs) plays a crucial role in lipid transport and storage. Here, we investigated the mechanisms underlying CaEC-adipocyte interaction and its impact on metabolic function. Single-cell RNA sequencing (scRNAseq) revealed an enrichment of fatty acid handling machinery in CaECs from high fat diet (HFD) mice, suggesting their specialized role in lipid metabolism. Transmission electron microscopy (TEM) confirmed direct heterocellular contact between CaECs and adipocytes. To model this, we created an in vitro co-culture transwell system to model the heterocellular contact observed with TEM. Contact between ECs and adipocytes in vitro led to upregulation of fatty acid binding protein 4 in response to lipid stimulation, hinting intercellular signaling may be important between ECs and adipocytes. We mined our and others scRNAseq datasets to examine which connexins may be present in adipose capillaries and adipocytes and consistently identified connexin 43 (Cx43) in mouse and humans. Genetic deletion of endothelial Cx43 resulted in increased epididymal fat pad (eWAT) adiposity and dyslipidemia in HFD mice. Consistent with this observation, phosphorylation of Cx43 at serine 368, which closes gap junctions, was increased in HFD mice and lipid-treated ECs. Mice resistant to this post-translational modification, Cx43S368A, were placed on an HFD and were found to have reduced eWAT adiposity and improved lipid profiles. These findings suggest Cx43-mediated heterocellular communication as a possible regulatory mechanism of adipose tissue function.


Introduction
Adipose tissue is comprised of several cell types other than its tissue resident, adipocyte. 1Specifically, adipose tissue is innervated with a robust vascular network primarily made up of capillaries. 2 , 3Capillaries consist mostly of endothelial cells (ECs) and pericytes.Here, we focus on capillary adipose endothelial cells (CaECs) and the role these particular cells play in the development and progression of endothelial dysfunction in obesity.
Capillaries differ from arteries and veins in a few ways. 4 , 5ne is the speed of which blood flows through the vessel.Capillaries are a low flow, low pr essur e envir onment allowing for proper gas and nutrient tr ansfer betw een tissues and the b lood. 6Specificall y, capillaries in adipose tissue must take up and transport serum lipids to adipocytes for proper stor age . 7mpr oper stora ge of lipids can r esult in lipoto xicity and o xidati v e str ess both at the cellular and tissue lev el. 8 Due to their location, CaECs face the majority of the lipid uptake and storage burden.To perform this specialized task, CaECs must have a unique gene expression profile for specific proteins, which will take up and transport lipids to neighboring adipocytes.The mechanism by with CaECs build this specialized store of proteins remains unknown.Does their location leave them primed to quickly respond lipid stimulation?We aim to answer these questions by looking at another way in which capillaries differ from arteries and veins, which is in their mural cell cover age .Arteries and veins are almost completely encased in a coordinated sheath of vascular smooth muscle cells (SMCs). 5apillaries, howev er, ar e partiall y cov er ed b y pericytes, 9 lea ving ECs exposed to the interstitial space and tissue specific cells.Lack of complete cov era ge allows for an interesting opportunity for CaEC cell membranes to physically interact with the cell membranes of neighboring adipocytes.This is a concei v a b le heterocellular interaction that has not yet been extensi v el y explor ed.Her e , w e hypothesized that CaECs ar e a b le to uniquel y respond to lipid stimulation in their environment due to their heterocellular contact with adipocytes.
Heterocellular contact is often facilitated by gap junctions comprised of connexin proteins.2][13][14][15] To form a functional gap junction, each cell type must contribute a connexon that come together to form a gap junction channel connecting the cytoplasm of two cells. 14Importantly for their function, connexin pr oteins ar e highl y modified by post translational modifications. 16 , 17Phosphorylation of specific amino acids on the carbo xyl-terminus of conne xin 43 (Cx43) proteins can cause opening or closing of the gap junction.The opening and closing of gap junctions in response to certain stimuli can trigger various cellular signaling cascades by allowing for or blocking the passage of second messengers from one cell cytoplasm to the next.
In the v asculatur e specificall y, connexins hav e been demonstrated to connect ECs with SMCs in the arterial wall of small r esistance v essels. 16 , 17This heterocellular contact dictates the v asodilator y pr operties of the arter y itself. 18Her e , w e present data demonstrating CaEC and adipocyte heterocellular contact facilitated by Cx43 mediated gap junctions can regulate epididymal white adipose tissue (eWAT) adiposity and metabolic function.

Human ECs
A pr eviousl y pub lished scRNA-seq dataset fr om human adipose tissue was used to create these data. 19ECs were subsetted, using subset() function, from previously published data using clusters defined by the original authors and confirmed by our arterial, v enous, capillar y, and l ymphatic endothelial markers.ECs fr om visceral adipose tissue were used for analysis for more appropriate comparison with epididymal adipose ECs.

Murine Adipocytes
A pr eviousl y pub lished snRNA-seq dataset fr om mouse e pididymal adipose tissue was used to create these data. 20eWAT data were used to isolate out the LFD or lean adipocyte populations.

Murine ECs
ScRNAseq from mouse adipose endothelium has been previously described in. 3Briefly, the generation of single-cell indexed libraries was performed by the School of Medicine Genome Analysis and Technology Core, RRID:SCR 018883, using the 10X Genomics chr omium contr oller platform and the Chr omium Single Cell 3 Library and Gel Bead Kit v3.1 reagent.Around 5000 cells were targeted per sample and loaded onto each well of a Chromium Single Cell G Chip to generate single cell emulsions primed for r ev erse transcription.After br eaking the em ulsion, the single-cell specific barcoded DNAs were subjected to cDNA amplification and QC on the Agilent 4200 TapeStation Instrument, using the Agilent D5000 kit.A QC run was performed on the Illumina Miseq using the nano 300Cycle kit (1.4 Million reads/run), to estimate the number of targeted cells per sample using the Cellranger 3.0.2function.After run completion, the Binary base call (bcl) files were converted to fastq format using the Illumina bcl2fastq2 softw ar e raw reads in fastq files were mapped to the mm10 r efer ence m urine genome and assigned to individual cells by CellRanger 5.0.Data from tw o separ ate experiments r e pr esenting cells fr om 12 mice wer e anal yzed in RStudio (2022.07.1) with the Seur at pac kage (4.3.0).Sequencing yielded 27 944 cells with 54 100 features.To ensure high quality data, cells were excluded if they contained less than 200 genes, more than 5000 genes, if their transcriptome was more than 5% mitochondrial encoded, and more than 5% hemoglobin beta.This resulted in a final dataset of 17 164 cells.Data were combined using SCTransform, normalized, and 3000 v aria b le featur es wer e chosen.UMAPs were generated using 20 principal components.Clusters were generated using a resolution of 1. Non-ECs were excluded based on low expression of Pecam1 and Cdh5 as well as high expression of non-endothelial markers (Col1a1, Acta2, Cd3g, Ptprc, Ccr5, Adipoq).Dotplots were made with assay = "RNA" to capture the most appropriate comparisons between genes.

Electr on Micr oscopy and Ima ge Analysis
Six male C57BL6/N mice, purchased from Taconic were fed an NC diet (5% Kcal Fat) were used between the ages of 15-20 wk.Mice were first perfused with PBS with a subsequent second perfusion with a solution consisting of 4% PFA and 0.5% glutaraldehyde in PBS.Epididymal adipose tissue was removed and placed in a 4% PFA/0.5% glutaraldehyde solution to post fix for 48 h at 4 • C. Samples were then processed for electron microscopy studies by the UVA adv anced micr oscopy facility.Grids were imaged on a Jeol 1230 for transmission electron microscopy (TEM) analysis.For TEM anal ysis, m ultiple fields of view were used for each mouse , eac h mouse r e pr esents an N v alue .Membr anous contact was defined as any two membranes residing within 10 nm of each other.Data r e pr esented as percentage of total contact sites observed within adipose tissue.

Real-Time Quantitati v e Pol ymer ase Chain Reaction (PCR)
Total RNA was extracted from mouse tissues and adipocyte fractions using the Aurum Total RNA Fatty and Fibrous Tissue Extraction Kit (Biorad: #732-6870).RNA from cells was extracted using Zymo Research R1055 Quick-RNA MiniPr e p Kit, Zymo Resear c h Kit (Genesee: 11-328).RN A concentr ation was measured using the Nanodrop1000 spectrophotometer (Thermo F isher).RN A w as stor ed at −80 • C befor e r ev erse transcription with SuperScript III F irst-Str and Synthesis system (Thermo Fisher: 18080051) using random hexamer primers on 1 μg of template RNA.Real-time quantitati v e PCR w as performed using Taqman Gene Expression Master Mix (Thermo Fisher: 4369016) and Taqman Real-Time PCR assays in MGB-FAM for Cx43 (Mm01179639 s1), Fabp4 (Hs00609791 m1), and were normalized to β-2-microglobin/B2M in VIC-PL (Hs00364808 m1; Mm00437762 m1).Reactions were run in a CFX Real-Time Detection System (Applied BioSystems) and threshold cycle number (CT) was used as part of the 2-DDCT method to calculate fold change from control.

Genomic Excision
DN A was extr acted fr om lung tissue and digested using pr oteinase K (Bioline: BIO-37084) for genomic excision gels.A set of two primers were used, forward: GCTA CTTCTTGCTTTGA CTCT-GA TT A and r ev erse: GCTCACTTGA T AGTCCACTCT AAGC.Nonexcised Cx43 mice have a no band and Cx43 is a positi v e excision band at 686 bp.

Western Blotting
Cells and tissue lysates were generated in RIPA (50 mmol/L Tris-HCL, 150 mmol/L NaCl, 5 mmol/L EDTA,1% deoxyc holate , 1% Triton-X100) in PBS and pH adjusted to 7.4 supplemented with protease inhibitor cocktail (Sigma: P8340).Lysates were rocked at 4 • C for 30-60 min to solubilize proteins, sonicated briefly, and centrifuged for 15 min at 1 300 G to pellet cell debris.Protein concentration was determined using the Pierce BCA method (Thermo Fisher: 23227).Total protein of 20 μg was loaded into each sample well.Samples were subjected to sodium dodecyl-sulate (SDS) gel electr ophor esis using 8% or 4-12% Bis-Tris g els (Invitrog en) and transferr ed nitr ocellulose membranes for imm unob lotting.Membr anes w er e b locked for 1 h at r oom temperature in a solution containing 3% BSA in Tris buffered saline, then incubated overnight at 4 • C with primary antibodies against Cx43 (Sigma: C6219, 1:1000) and Phospho368-Connexin-43 (Sigma: SAB4504371, 1:100).Membranes wer e w ashed and incubated in LiCOR IR Dye secondary antibodies (1:10 000) for 1 h and viewed/quantified using the LiCOR Odyssey CLx with Image Studio softw ar e. Licor Total Pr otein stain w as used for loading normalization.Re pr esentati v e western b lot ima ges hav e been cropped for presentation.

Animals
Only male mice were used, 20-23 wk of age, on a C57Bl/6 genetic background.Cdh5ERT2 + /Cx43fl/fl mice were cared for under the provisions of the University of Virginia Animal Care and Use Committee, while Cx43WT and Cx43S368A animals were cared for under the provisions of Virginia Tech Animal Care and Use Committee.Both facilities follow the National Institutes of Health guidelines for the care and use of la borator y animals.Animals were subject to a 12-h light dark cycle.The inducible, EC-specific Cx43 knockout mice (Cdh5ERT2 + /CX43fl/fl) were generated by crossing Cdh5ERT2 + /CxWT/WT mice (a kind gift from Dr Ralf Adams, Max Plank Institute, Germany) with Cdh5ERT2 −/Cx43fl/fl mice. 23To conditionally induce Cx43 deletion in the vascular endothelium, Cdh5ERT2 −/Cx43fl/fl (EC Cre − Cx43fl/fl) and Cdh5ERT2 + /Cx43fl/fl (EC Cre + Cx43fl/fl) littermates were fed tamoxifen diet (Envigo: TD130856) for two consecuti v e weeks starting at 6 wk old.Littermates were then fed HFD (60% Kcal from fat Bio-Serv-The Foster Corp F3282) starting at 8 wk of age for 12 consecutive weeks.For all assessments of b lood, b lood w as collected via terminal cardiac punctur e using a syringe fitted with 25 G needle, coated with ethylene glycolbis( β-aminoethyl ether)-N ,N ,N ,N -tetraacetic acid (EGTA) to prevent clotting and deposited in gold cap blood collection tubes.Blood lipids (cholesterol and triglycerides) were processed by UVA clinical la borator y.

Fatty Acid Machinery Is Enriched in CaECs
Single-cell RNA sequencing (scRNAseq) of adipose ECs from normal chow (NC) and high fat diet (HFD)-fed mice ( Figure 1 A) show an enrichment of fatty acid binding and transport machinery (ie, Fabp4 , Fapb5 , CD36 , Gpihbp1 ) in capillary ECs ( Figure 1 B and  C).This increased gene expression is specific to capillaries from HFD-fed mice demonstrating their importance in lipid handling as compared to other EC types.

CaECs Make Contact with Adipocytes
To understand why this gene expression change is specific to capillary endothelium, we examined the ultrastructure of epididymal adipose tissue capillaries ( Figure 2 A).In several images, we found significant instances of cellular contact between CaECs and adipocytes ( Figure 2 B, left).A different form of cellular contact seen in adipose tissue was between pericytes and adipocytes ( Figure 2 B, middle).In all sections of tissue examined, there were no instances of adipocyte to adipocyte cellular contact observed ( Figure 2 B right).Interestingly, only the CaEC and adipocyte contact sites had the ball-and-socket structur e r eminiscent of m y oendothelial junctions (MEJs) betw een endothelium and smooth muscle in arteries (eg, 11 , 24-27 ).These contact sites were quantified as a percentage of total contact between cells observed within adipose tissue ( Figure 2 C).Based on these data, we have found CaEC to adipocyte contact makes up the majority of heterocellular contact observed in adipose tissue.To model this physical interaction in vitro, we developed a co-culture model using tr answ ell inserts with pores to allow for cellular projections to occur ( Figure 3 A).Using our co-culture model, the tr answ ell insert can be sectioned transv ersel y ( Figure 3 B) to examine the morphology (hematoxylin and eosin; H&E) and cellular projections (Phalloidin) of the human adipose micr ov ascular ECs (HAMECs) and differ entiated human adipoc ytes.The trans well inserts can also be immuno-stained en face to look at both the top HAMEC layer and the bottom adipocyte layer.We show HAMECs ( Figure 3

Cx43 May Regulate Capillary Adipose Endothelial and Adipocyte Junctions
Heterocellular contact between CaECs and adipocytes is important for fatty acid gene regulation.Gap junctions are major r egulators of heter ocellular contact, especiall y in the v asculature. 14 , 28Using scRNAseq data from human adipose ECs ( Figure 4 A), the only gap junctional protein detected in capillaries was connexin43 (Cx43-GJA1).To form a functional gap junction between cells via connexin proteins, each cell type contributes half the junction.We also found Cx43 to be one of the main conne xins e xpr essed in m urine adipocytes ( Figur e 4 B).Complimentary to this data, scRNAseq of murine adipose ECs mimics the data shown from human ECs in that Cx43 is the only connexin expressed in the capillary adipose endothelium ( Figure 4 C).The overall conne xin e xpression of murine adipose endothelium closel y r esemb les ov erall conne xin e xpression in the mesenteric endothelium, regardless of vascular bed ( Supplementary Figure S1 ).Gi v en this data, we w anted to utilize our tr answ ell co-culture system to look for Cx43 protein.We observed Cx43 puncta overlap with the cellular projections made between HAMECs and adipocytes ( Figure 4 D).To further understand the functional relevance of these gap junctions, we selecti v el y knocked down Cx43 from HAMECs ( Figure 4 E) and demonstrated a loss of Fabp4 gene expression in endothelium ( Figure 4 F) and adipocytes ( Figure 4 G), reminiscent of a lack of cellular contact in Figure 3 D. Thus, the presence of gap junctions between endothelium and adipocytes may be an important regulator of adipose function.
Connexin proteins are highly regulated by post-translational modifications.These modifications can dictate the channels' open and closed state as well as its localization. 16 , 17 , 29Specificall y, phosphor ylation of Cx43 at serine 368 can incr ease the closing of the Cx43 channel. 30 , 31To further investigate how Cx43 may regulate heterocellular communication between CaECs and adipocytes, we looked for phosphorylated Cx43 specifically at serine 368 (P-Cx43) using transverse sections from our tr answ ell co-culture model.Representative immune-staining ima ges ( Figur e 5 A) show r obust Cx43 staining in HAMECs and adipocytes in both control and lipid-treated conditions.However, upon lipid treatment, phosphorylation of Cx43 increases when quantified from all cells in Figure 5 B. Additionally, we quantified P-Cx43 specifically from the portion of the tr answ ell containing the cellular junctions and saw the same increase in P-Cx43 but not Cx43 itself ( Figure 5 C).To test if P-Cx43 increases via lipid stimulation in vi v o, we looked specifically for in adipose tissue from mice fed a HFD for 12 wk using immunofluorescence on epididymal fat pad sections ( Figure 5 D).P-Cx43 w as incr eased in adipose tissue fr om mice fed a HFD ( Figur e 5 D).Capillary endothelium was marked using Car4. 4 , 32 Additionally, we assessed total Cx43 protein abundance , whic h did not change between NC and HFD-fed mice ( Figure 5 E).However, P-Cx43 increased in HFD-fed animals ( Figure 5 F) mimicking the results we see in vitro.The ratio of phosphorylated Cx43 and total Cx43 was quantified and normalized to total protein ( Figure 5 G).These data show Cx43 is present at the junctions between CaECs and adipocytes and phosphorylation at serine 368 is increased in HFD and may regulate heterocellular communication.

Endothelial Cx43 Can Regulate Ewat Adiposity
To test our hypothesis that EC Cx43 may regulate intercellular communication between CaECs and adipocytes, we selecti v el y deleted Cx43 from endothelium using an inducible Cdh5-Cre ERT2 + Cx43 fl/ fl mouse.Successful deletion is shown via excision of Cx43 ( Figure 6 A) and reduction in both Cx43 RNA and pr otein ( Figur e 6 B-C).By r emoving Cx43 fr om endothelium, we can model the gap junction being closed in vi v o.As a metabolic c hallenge , mice w ere fed a HFD for 12wk.Deletion of Cx43 from ECs caused an increase in both weight gain ( Figure 6 E) and eWAT fat pad mass ( Figure 6 F).Additionally, both serum cholesterol ( Figure 6 G) and triglycerides ( Figure 6 H) were also increased significantly in mice lacking Cx43 in endothelium.Age matched mice were also fed NC diet, however, no changes in body mass, eWAT mass, or trigl ycerides wer e observ ed ( Supplementar y Figur e S2 ) r egardless of genotype.Inter estingl y, cholester ol lev els in NC EC Cr e + Cx43 fl/ fl wer e slightl y elev ated compar ed to NC fed contr ols.Female EC Cre + Cx43 fl/ fl mice, fed both NC or HFD, generally demonstrated a similar trend to males ( Supplementary Figure S3 ), but with significantly lower weights and lipid parameters due to decreased lean mass (eg, 33 ) Additionally, loss of EC Cx43 had no effect on insulin or glucose sensitivity in both NC and HFD-fed mice ( Supplementary Figure S4 A-D).We next examined both epididymal adipose and liver histology to examine tissue morphology ( Supplementary Figure S5 A-B).As expected, gi v en the lack of glucose and insulin phenotype , there w ere no differences observed in overall morphology of adipose tissue ( Supplementary Figure S5 A).Howev er, li v er sections fr om HFD EC Cr e + Cx43 fl/ fl appear ed to contain more lipids ( Supplementary Figure S5 B).Given these data, we performed Oil Red O staining to assess li v er lipid content, quantification of which shows a slight increase in lipid accumulation within the liver ( Supplementary Figure S5 C).The li v er contributes significantl y to the non-esterified fatty acid  (NEFA) concentrations within the b lood, ther efor e , w e examined serum NEFA levels in both NC and HFD-fed mice.Regardless of diet EC Cre + Cx43 fl/ fl mice had increased NEFA levels ( Supplementary Figure S4 E-F) indicating an imbalance in lipid homeostasis.These data demonstrate an important role for EC Cx43 in the regulation of serum lipids especially in HFD.
Loss of EC Cx43 is a mimetic for the channel in a closed state; thus, to test a scenario where Cx43 gap junctions would be more likely to be open, we used a mutant mouse where serine 368 is mutated to an alanine (Cx43 S368A ).Ther efor e , Cx43 S368A (c hannel open) mice would be expected to have the opposite phenotypic parameters to the EC Cre + Cx43 fl/ fl mice that have no Cx43 in endothelium (channel closed).Indeed, Cx43 S368A mice fed a HFD for 12 wk have decreased body w eight ( F igure 6 I) and decreased eWAT mass ( Figure 6 J) compared to wildtype controls (Cx43 WT ).Furthermore, Cx43 S368A mice have decreased cholester ol ( Figur e 6 K) and trigl ycerides ( Figur e 6 L) in r esponse to HFD c hallenge .

Discussion
Adiposity and serum lipids are an important indicator of metabolic health and risk for car dio vascular disease. 34Here, we present a novel mechanism by which CaECs can regulate eWAT adiposity and serum lipids through direct contact with adipocytes.This contact is facilitated by gap junctional protein Connexin43 (Cx43), and phosphorylation of Cx43 at serine 368 regulates heterocellular communication between ECs and adipocytes.Phosphorylation, and therefore closure of the Cx43 gap junction, of Cx43 specifically at serine 368 is increased with lipid treatment and HFD.This increased channel phosphoryla-tion and decreased EC to adipocyte communication, leading to dyslipidemia could be a potential mechanism behind the onset endothelial dysfunction in obesity.
Our scRNAseq data show an enrichment of fatty acid handling mac hinery (ie , F abp4 , F apb5 , Cx36 , Gpihbp1 ), specifically in capillaries in response to HFD.This upregulation specifically CaECs is logical as serum lipids must pass through capillaries in adipose tissue for proper storage in adipocytes.Therefore, CaECs see a higher concentration of lipids than other endothelial beds and must increase fatty acid transport and storage proteins.Our main question based on this data was how do CaECs know their function?What about their environment within adipose tissue allows for this specific and targeted gene expression c hange .What is unique about capillaries is their lack of complete mural cell coverage compared to arteries and veins that hav e v ascular SMCs. 5 Less cov era ge by peric ytes allo ws for contact between CaECs and adipocytes.Ther efor e, we hypothesized contact with adipocytes and exposure to high lipid levels primes CaECs to take upregulate fatty acid machinery in response to HFD.
Here, for the first time to best of our knowledge , w e describe and c har acterize CaEC and adipocyte heterocellular contact.7][18] In fact, the projections and contact sites we observed in adipose tissue between CaECs and adipocytes r esemb les the contact seen between arterial ECs and SMCs of small resistance arteries.These contact points deemed MEJs 11 are key regulators of vasodilation and therefor e systemic b lood pr essur e. 13 , 18 Ther efor e, it is likely the adipose-endothelial junction could regulate adipose tissue function in a similar manner to the MEJ.We demonstrate the majority of heterocellular contact in adipose tissue is between ECs and adipocytes.This positions CaECs as a window between the adipocyte and the blood, serving as an interface to relay critical information on nutrient status.This form of rapid and specific comm unication likel y becomes crucial in times of metabolic distress, ie, HFD/obesity.
We show evidence for the importance of heterocellular contact with lipid stimulation using our co-culture tr answ ell model.Only when in direct contact with adipocytes did human adipose micr ov ascular endothelial cells (HAMECs) increase Fabp4 expr ession.Additionall y, when in dir ect physical contact, tr eating HAMECs with lipids induced a robust increase in adipocyte Fabp4 .This physical coupling primes the adipocyte for its task of lipid uptake and stor age .Proper storage of lipids is crucial to pr ev ent lipotoxicity and the subsequent oxidati v e str ess that follows.Data presented here show HFD increases Cx43 phosphorylation.This phosphorylation occurs specifically at serine 368 and serves as a signal for channel closure, decreasing the communication between CaECs and adipocytes.We hypothesize, this diminished communication between CaECs and adipocytes could contribute to the endothelial and adipocyte dysfunction seen in obesity.
By using mice which lack EC Cx43 (mimetic of channel closure) or express a version of the channel that is more likely to be open, we can test the metabolic effects of decreased communication between CaECs and adipocytes.Mice lacking Cx43 had worse metabolic outcomes when eating an obesogenic diet compared to mice with increased channel opening.Decreased communication between capillaries and adipocytes regarding lipid uptake and storage could account for the increase in adipose tissue mass seen in mice without EC Cx43.Additionally, metabolic disruption can result in dyslipidemia, which is also seen in mice with an EC Cx43 deletion.Increasing the channels pr oba bility of being open through the mutation of serine 368 to an alanine protected mice from metabolic dysfunction associated with HFD.Taken together, our data points to EC Cx43 as a regulator of eWAT adiposity possibly through heterocellular communication between CaECs and adipocytes.
Our data show a gr eater r eliance on EC Cx43 communication during metabolic stress (ie, HFD feeding).We see an increase in total body mass, eWAT mass, cholesterol, and triglycerides with loss of EC Cx43 in HFD.However, mice lacking EC Cx43, which r ecei v ed NC diet, onl y hav e a slightly elevated serum cholesterol level.It is possible caECs and adipocytes rely on different comm unication pathw ays during homeostatic and high fat conditions.These data suggest that under metabolic stress adipocytes and ECs r el y mor e heavil y on Cx43-mediated comm unication.
In addition to the presence of other caEC-adipocyte signaling axis, it is possible there are additional tissues besides adipose contributing to the phenotypes seen here.For example, li v er, skeletal m uscle, and br own adipose ar e also crucial for lipid homeostasis.Here, we show a slight increase in liver lipid accumulation, it is possible skeletal muscle and brown adipose taken from loss of EC Cx43 mice could display a similar phenotype.Tissue lipid accumulation is likely due to the disruptions in lipid re gulation whic h permeate systemicall y.The li v er, skeletal m uscle, and brown adipose are also highly vascularized specifically with capillaries.Looking for potential sites of EC contact with hepatocytes, m y ocytes, and brown adipocytes could further our understanding of how ECs can regulate metabolism.
Our work is not without alternati v e e xplanation.For e xample , w e do not examine or quantify imm une cell pr esence or contact in adipose tissue.Immune cells have been identified as mediators of adipose tissue function and health. 35However, the experimental method we chose to identify contact sites, TEM, is not adequate to identify and c har acterize contact betw een immune cells and adipocytes.We do not discount that immune cells, especially in HFD, can affect the proper function of adipose tissue .Additionally, w e did not identify a signaling molecule, which could potentially pass between cells through Cx43 gap junctions relaying metabolic information.7][38][39] Similarly, we do not directly measure, via patch clamp or dye transfer, the open and closed states of Cx43 gap junctions when phosphorylated or mutated (S368A).However, these experiments have been previously done. 40 , 41Developing therapeutic targets that could toggle the metabolic state of ECs with Cx43 acti v ators and inhibitors could serve as novel mechanism to treat disease.This is an excellent area for future study, which could lead to more targeted methods for weight management in obese patients.Lastly, we do not discount the contributions of extracellular vesicles (EVs) passed between ECs and adipocytes in the regulation of adipose tissue metabolism. 42ur no contact tr answ ell co-culture model indicates the contribution of EVs in our system is minimal as the presence of adipocytes alone was not enough to elicit Fabp4 expression when ECs were treated with lipids.Instead, we explore an alternati v e mechanism for EC and adipocyte communication that could act in conjunction with these pr eviousl y esta b lished pathways.
Statistical anal ysis w as performed using Prism GraphPad 9. Specific statistical test type and N values are listed in each corresponding figure legend.Data represented are the mean with SEM ( + and −) shown as error bars.All N values represent different experiments (in vitro) and individual mice (in vivo).For image quantification, three fields of view were used to average each N value.
C, top) with robust staining of vascular marker Ve-Cadherin and almost no lipid

Figure 2 .
Figure 2. Capillary adipose endothelial cells are in direct contact with adipocytes.(A) Transmission electr on micr ograph (TEM) of an adipocyte (a) and capillary (cap, * denotes lumen).Scale bar denotes 2 μm.(B) TEM showing different types of cellular contact in adipose tissue.Left demonstrates endothelial (EC) and adipocyte (Adipo) contact, middle showing pericyte (PC) and adipocyte contact and right, showing lack of adipocyte-adipocyte contact.Scale bar denotes 500 nm.(C) Quantification of contact seen in adipose tissue normalized as a percentage of total contact sites seen in tissue.N = 4. Arrows indicate sites of contact.

Figure 3 .
Figure 3. Fatty acid gene expression in adipose endothelial cells is dependent on contact with adipocytes.(A) Schematic showing experimental methods for the development of endothelial-adipocyte co-culture model using tr answ ell inserts.(B) Transverse view of tr answ ell insert sectioned and stained with H&E (top) and Phalloidin (bottom).Human adipose endothelial cells (HAMECs) are on the top side of the tr answ ell with differentiated human adipocytes on the bottom.Arrow is pointing to cellular contents (marked by phalloidin) in the pores of the tr answ ell.Scale bar represents 10 μm.(C) En face view of the tr answ ell insert.HAMECs shown in top image staining with EC marker Ve-Cadherin (magenta: top image) and adipocytes from the underside of the same tr answ ell are shown below with robust lipid droplet staining (Bodipy-yellow: bottom image).Scale bar represents 20 μm.(D) Fatty acid binding protein (Fabp4) expression in HAMECs from monolayer (left) nocontact co-culture (middle) and contact (right) co-culture models after 50 μm lipid treatment of HAMECs 50 μm long chain fatty acids (consisting of 12.5 μm linoleic acid, 25 μm Oleic acid, and 12.5 μM Palmitic acid for 8 h) N = 4-5.(E) Adipocyte Fabp4 expression from contact co-culture after lipid stimulation of HAMEC layer, N = 3. Statistics r e pr esent Student's t -test.

Figure 4 .
Figure 4. Cx43 is expressed in capillary endothelial cells and adipocytes.(A) Dotplot of all connexins expressed in human adipose endothelial cells divided by EC subtype.Connexin43 is shown with gene name as GJA1.(B) Dotplot of all connexins expressed in murine adipocytes.(C) Dotplot of all conne xins e xpressed in murine adipose endothelial cells divided by EC subtype.(D) Transverse section of contact co-culture model with Cx43 staining (top) and phalloidin staining (middle) and the overlap (bottom) in the pores of the tr answ ell.Scale bar denotes 10 μm.(E) Endothelial cell (EC) Cx43 expression with siRN A knoc kdown of Cx43, (F) EC Fabp4 expression with or without siCx43, (G) Adipocyte (Adipo) Fabp4 expression with or without EC siCx43.Statistics represent Student's t -test.Long chain fatty acids of 50 μm were used for lipid treatments (12.5 μm linoleic acid, 25 μm oleic acid, and 12.5 μm palmitic acid for 8 h).

Figure 5 .
Figure 5. Phosphorylation of Connexin43 increases with lipid stimulation and high fat diet.(A) Transverse section of contact co-culture tr answ ell from control or lipidtreated human adipose endothelial cells (HAMECs).The top layer of cells are HAMECs, the middle section is the pores of the tr answ ell, and the bottom contains human adipocytes.Connexin43 (Cx43) staining is shown in cyan and phosphorylated connexin43 (P-Cx43) is shown in magenta.Scale bar r e pr esents 10 μm.Lipid-treated cells were dosed with 50 μm long chain fatty acids consisting of 12.5 μm linoleic acid, 25 μm oleic acid, and 12.5 μm palmitic acid for 8 h.(B) Quantification of P-Cx43 and Cx43 from all cells in the transverse section and (C) showing the quantification of just the junctional portion of the contact co-culture transverse section.(D) Imm unofluor escence staining of epididymal fat pad sections for Car4 (capillary marker) and P368-Cx43 in NC (top) and HFD mice (bottom).Quantification of P368-Cx43 fluor escence ar ea normalized to n umber of n uclei (far right).Scale bar r e pr esents 100 μm western b lot of adipose tissue fr om normal chow (NC) and high fat diet (HFD) fed mice imm unob lotting for Cx43 (E) P-Cx43 (F).Total protein is shown in (G) with ratio of P-Cx43/Cx43 signal quantified below.Right most lanes are molecular weight ladders in E-G.Statistics r e pr esent Student's t -test.Data points r e pr esent indi vidual mice (D-G).