MYCT1–IFITM2/3 interaction links endothelial endolysosomal trafficking to white adipose tissue expansion

Wetterwald, Köck et al. report the existence of an endothelial metabolic checkpoint where MYCT1 restricts activity of IFITM2/3 to prevent energy waste at the vascular barrier and enable fat storage. This mechanism links endothelial nutrient handling to systemic energy balance, opening new directions for tackling obesity and metabolic disease.

Abstract

Vertebrates rely on a network of blood vessels to meet organ demands for oxygen and nutrients. While endothelial cells are known to transport excess nutrients to white adipose tissue (WAT) for energy storage, how their metabolic state impacts this process remains unclear. Here, we identify MYCT1 as a conserved, pan-endothelial protein essential for WAT expansion. Endothelial-specific MYCT1 deletion limited WAT expansion independently of angiogenesis, adipogenesis, or systemic metabolic parameters. Mechanistically, MYCT1 interacted with the transmembrane endolysosomal proteins IFITM2/3 to restrict nutrient consumption by the vascular barrier. Loss of MYCT1 caused IFITM2/3 accumulation in early endosomes, promoting excessive endolysosomal degradation and mTORC1 hyperactivation, limiting the WAT energy storage capacity. Notably, endothelial-specific mTORC1 activation through TSC1 deletion phenocopied the fat storage defects of MYCT1 deficiency. Our findings establish the MYCT1–IFITM2/3 complex as endothelial metabolic checkpoint regulating systemic energy storage. Targeting MYCT1–IFITM2/3 may offer new therapeutic options for obesity and metabolic disorders.

Graphical Abstract

Introduction

Vertebrates have evolved a closed and ramified circulatory system for efficient resource distribution. Lining the interior of vascular networks, endothelial cells (ECs) are immersed in a diverse milieu of nutrients, hormones, and metabolites (Hasan and Fischer, 2020). In most adult tissues, ECs are quiescent with long half-lives; their own nutrient consumption is kept to a minimum, which preserves resources and maximizes nourishment for the surrounding tissues. Given their strategic position, ECs can sense various stimuli and rapidly adjust their metabolic state during angiogenesis (De Bock et al., 2013; Ong et al., 2022; Kalucka et al., 2018) or buffer postprandial surges of nutrients (Kuo et al., 2017), thereby contributing to systemic metabolic homeostasis.

The major routes of nutrient transport through the EC barrier include specific transporters, fluid-phase endocytosis, and receptor-mediated transcytosis (Hasan and Fischer, 2020). These mechanisms enable ECs to deliver essential nutrients from the bloodstream, supporting tissue metabolism, and allowing excess energy storage in white adipose tissue (WAT), an important evolutionary insurance policy against food scarcity (Sakers et al., 2022). During periods of nutrient abundance, WAT stores excess energy as triglycerides, either by taking up fatty acids or through de novo lipogenesis, which converts surplus carbohydrates and proteins into lipids (Collins et al., 2011; Sakers et al., 2022). Energy stores can then be mobilized during fasting to support whole-body metabolism. While this energy-buffering role was essential for survival throughout evolution, in today’s context of chronic overnutrition, limited WAT expandability drives metabolic dysfunction, leading to obesity, type 2 diabetes, and cardiovascular diseases (Sakers et al., 2022).

Healthy WAT expansion requires coordinated adipogenesis, angiogenesis, and remodeling of the extracellular matrix to safely store excess nutrients without triggering adipocyte death, inflammation, and fibrosis (Crewe et al., 2017). Although recent studies highlight the importance of cross talk between the endothelium and adipose tissue (Robciuc et al., 2016; Monelli et al., 2022; Crewe et al., 2018), the intracellular mechanisms through which ECs influence systemic metabolism and energy storage remain an area of active investigation.

Here, we identify MYCT1 as a pan-endothelial transmembrane phosphoglycoprotein that controls WAT expansion. Endothelial-specific deletion of MYCT1 limits adiposity without altering WAT angiogenesis, adipogenesis, and whole-body metabolism. Mechanistically, we uncover a previously unrecognized interaction between MYCT1 and the interferon-induced transmembrane proteins 2 and 3 (IFITM2/3), which have been previously implicated in endolysosomal trafficking and membrane fusion (Majdoul and Compton, 2022). Our findings demonstrate that this interaction regulates endothelial nutrient consumption. In the absence of MYCT1, IFITM2/3 accumulate in early endosomes, driving excessive endolysosomal cargo degradation and hyperactivating endothelial mTORC1 signaling, ultimately restricting energy accumulation in WAT. Notably, endothelial-specific mTORC1 activation through tuberous sclerosis complex 1 (TSC1) deletion recapitulates the adipose tissue phenotype of MYCT1-deficient mice. These findings uncover the MYCT1–IFITM2/3 complex as a novel link between EC metabolism and systemic energy storage. Identification of the MYCT1–IFITM2/3 interaction as a regulator of endothelial nutrient handling suggests new therapeutic targets for metabolic disorders associated with dysfunctional WAT expansion, such as obesity, type 2 diabetes, and related cardiovascular disorders.

Results

WAT is a highly vascularized metabolic organ where ECs actively coordinate nutrient flux and tissue expansion (AlZaim et al., 2023). To uncover novel endothelial factors that coordinate these processes, we investigated MYCT1, a protein whose functional role in the endothelium remains to be characterized.

Pan-endothelial MYCT1 is required for WAT expansion

MYCT1 is expressed predominantly in ECs and in a subset of hematopoietic stem cells (HSCs) (Aguadé-Gorgorió et al., 2024; Kabir et al., 2021). Analysis of the human vascular cell atlas (Barnett et al., 2024) confirmed its broad expression in ECs, identified by high levels of PECAM1 and CDH5 (Fig. S1 A), with higher MYCT1 levels in arterial and capillary ECs than venous ECs (Fig. S1 B). Across organs, MYCT1 is restricted to blood ECs (BECs) and lymphatic ECs (LECs), with little or no expression in perivascular mural cells such as pericytes and smooth muscle cells (Fig. 1 A). Staining for MYCT1 on human tissue sections revealed robust expression of MYCT1 protein in diverse vascular beds, including intestine, skin, WAT, and small cell lung cancer (Fig. 1 B), highlighting MYCT1 as a pan-endothelial protein.

Figure S1.

Pan-endothelial MYCT1 is required for WAT expansion, related to Fig. 1. (A) MYCT1 expression is specific to the endothelial lineage. Dot plot representation of MYCT1 expression, selected markers of ECs, and other major human cell types. Dataset by Barnett et al. (2024). (B)MYCT1 expression is pan-endothelial with higher levels in arterial and capillary ECs. Dot plot representation of MYCT1 expression and selected markers of human endothelial and mural cell subsets (Barnett et al., 2024). (C) Generation of a conditional Myct1 deletion mouse model. (D)Myct1^ecKO mouse model. See Materials and methods for details. (E)Myct1 is efficiently deleted in adult ECs. RT-qPCR for the indicated transcripts using lung mRNA from long-term monitored wild-type or Myct1^ecKO mice. n = 3–9 mice per genotype; mean ± SD; multiple Mann–Whitney tests, P = 0.003 (*) for Myct1 and P > 0.05 for Cdh5. (F)Myct1 ablation prevents age-associated weight gain. n = 10 male mice per genotype; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons test, P = 0.0403 (*). (G)Myct1 ablation does not affect organ size. Quantification of organ weight to body weight ratio relative to wild-type mice. n = 5 mice per genotype; mean ± SD; multiple Welch’s t tests, P > 0.05. (H)Myct1 ablation does not affect tibia bone length. Quantification of tibia length relative to wild-type mice. n = 7–9 mice per genotype; mean ± SD; Mann–Whitney test, P > 0.05. (E–H) Data from the long-term Myct1 deletion experiment. (I)Myct1^ecko WAT does not express higher levels of browning markers. Heatmap representation of RT-qPCR for the indicated transcripts. BAT was used as a positive control. n = 5 wild-type WAT, 9 Myct1^ecKO WAT, and 4 wild-type BAT. (J) Quantification of fat pad weight to body weight ratio relative to wild-type mice under HFD conditions. n = 5 mice per genotype; mean ± SD; multiple Welch’s t test, P > 0.05 for IsWAT and IsBAT (interscapular fat), P = 0.009 (*) for retroperitoneal, P = 0.033 (*) for gonadal, P = 0.004 (*) for inguinal, and P = 0.016 (*) for mesenteric. Icons used in D were created with BioRender.com and modified in Affinity. BAT, brown adipose tissue; IsWAT, interscapular WAT; IsBAT, interscapular BAT.

Figure 1.

Pan-endothelial MYCT1 is required for WAT expansion. (A) MYCT1 gene expression is specific to the endothelial lineages. Dot plot representation of MYCT1 expression in blood and lymphatic endothelial and mural cells across various human tissues. Dataset by Barnett et al. (2024). (B) MYCT1 protein expression is endothelial-specific across human tissues and tumors. Chromogenic immunostaining of MYCT1 (brown) in human intestine, skin, white fat, and in small cell lung cancer. A, artery; V, vein; C, capillary; L, lymphatic vessel. Arrows, MYCT1 signal. Scale bar, 50 μm. (C) Experimental workflow for long term monitoring of Myct1^ecKO mice phenotype. (D) Adult Myct1^ecKO mice display reduced body weight. n = 10 female mice per genotype; mean ± SD; multiple unpaired t tests, P = 0.0045 (*). (E)Myct1 ablation reduces size of visceral and s.c. WAT. Whole-body CT-scans of wild-type and Myct1^ecKO mice. Is, interscapular fat; Mes, mesenteric fat; Gon, gonadal fat; Ing, inguinal fat. (F) Quantification of interscapular WAT (IsWAT) and BAT (IsBAT) tissue volume from CT scans. BAT, brown adipose tissue. n = 5 mice per genotype; mean ± SD; two-way ANOVA with Sidak’s multiple comparisons test; P < 0.0001 (*) for IsWAT and P > 0.05 for IsBAT. (G) Macroscopic images of interscapular, retroperitoneal, and gonadal fat pads of wild-type and Myct1^ecKO mice. Scale bar, 5 mm. (H) Quantification of fat pad weight to body weight ratio relative to wild-type mice. n = 6–7 mice per genotype; mean ± SD; multiple Welch’s t tests, P = 0.0072 (*) for IsWAT, P > 0.05 for IsBAT, P = 0.018 (*) for Ing (inguinal), P = 0.032 (*) for RP (retroperitoneal), P = 0.015 (*) for Gon (gonadal), and P = 0.037 (*) for Mes (mesenteric). (I)Myct1 ablation reduces the size of adipocytes. H&E staining of retroperitoneal fat sections of wild-type and Myct1^ecKO mice. Scale bar, 50 μm. (J) Quantification of adipocyte size relative to wild-type. n = 6 mice per genotype; mean ± SD; unpaired t test, P = 0.033 (*). (K) Experimental workflow for HFD experiment. (L) HFD accelerates differences in body weight change between wild-type and Myct1^ecKO mice. n = 5 mice per diet and genotype; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons test, P = 0.023 (*) for HFD and P > 0.05 for standard control diet (SD). (M)Myct1 ablation limits WAT (yellow, adipocytes) expansion. Icons used in A, C, K, and M were created with BioRender.com and modified in Affinity. See also Fig. S1.

To investigate MYCT1’s function, we generated endothelial-specific conditional KO mice (Myct1^lox/lox; Pdgfb-Cre^ERT2) for tamoxifen-inducible deletion (Myct1^ecKO) (Fig. S1, C and D). After induction in 10-wk-old mice (Fig. 1 C), efficient deletion was confirmed by lung mRNA analysis (Fig. S1 E).

Long-term monitoring revealed a significantly lower body weight in Myct1^ecKO mice compared with control littermates in both females and males (Fig. 1 D and Fig. S1 F, respectively). Computed tomography (CT) at 42 wk demonstrated selective WAT reduction in subcutaneous (inguinal and interscapular) and visceral (mesenteric and gonadal) depots, with brown adipose tissue unaffected (Fig. 1, E and F). This reduction was consistent across all fat pads (Fig. 1, G and H), while organ weights (heart, lungs, and kidneys) and tibia length remained unchanged (Fig. S1, G and H). Notably, adipocyte size was significantly smaller within 1 mo after deletion (Fig. 1, I and J). However, we did not observe evidence of WAT browning (Fig. S1 I).

To assess MYCT1’s role in adaptive WAT expansion, we subjected mice to a high-fat diet (HFD) for 10 wk (60 kcal% fat) (Fig. 1 K). Myct1^ecKO mice showed significantly attenuated weight gain and reduced WAT fat pad sizes compared with control mice on HFD, with no differences on a control diet (10 kcal% fat) (Fig. 1 L and Fig. S1 J). Collectively, these results demonstrate that inactivation of pan-endothelial Myct1 selectively impairs WAT expansion, without affecting overall growth, indicating a specific role in WAT homeostasis (Fig. 1 M).

Reduced adiposity in Myct1^ecKO mice is independent of angiogenesis, adipogenesis, and systemic metabolic activity

To investigate the mechanisms underlying reduced adiposity in Myct1^ecKO mice, we monitored animals housed individually in metabolic cages. These experiments were performed after short-term HFD exposure (Fig. 2 A), before obvious differences in body composition were observed (Fig. S2 A). Analyses revealed that the reduced adiposity was not due to alterations in food or water intake (Fig. 2 B and Fig. S2 B), physical activity (Fig. 2 C), respiratory exchange ratio (RER) (Fig. 2 D), or energy expenditure (Fig. 2 E and Fig. S2 C).

Figure 2.

Reduced adiposity in Myct1 ^ecKO mice is independent of angiogenesis, adipogenesis, and systemic metabolic activity. (A) Experimental workflow for monitoring early changes in metabolism of Myct1^ecKO mice. (B) Food intake over 24 h (23°C) is similar between wild-type and Myct1^ecKO mice. n = 10 mice per genotype; mean ± SD; Welch’s t test, P > 0.05. (C) Total horizontal activity over 24 h (23°C) is similar between wild-type and Myct1^ecKO mice. n = 10 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (D) RER over 24 h (23°C) is similar between wild-type and Myct1^ecKO mice. n = 10 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (E) Energy expenditure over 24 h (23°C) is similar between wild-type and Myct1^ecKO mice. n = 10 mice; mean ± SD; unpaired t test, P > 0.05. (A–E) Results are shown for males at 23°C. Similar results were obtained at 30°C and for females at both temperatures (data not shown). (F)Myct1 ablation does not affect adipose tissue fibrosis. Masson’s trichrome staining of retroperitoneal fat sections of wild-type and Myct1^ecKO mice. Scale bar, 100 μm. (G) Quantification of fibrosis as percentage of area positive for collagen based on Masson’s trichrome stain. n = 6 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (H)Myct1 ablation does not affect the vascular density of WAT. Staining of retroperitoneal sections for Pecam1 (green) and Fabp4 (magenta). Scale bar, 50 µm. (I) Quantification of vascular density as percentage of Pecam1⁺ adipose tissue area. n = 4 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (J) Quantification of CD45^negCD31⁺ cells as percentage of total cells in fat pad as measured by flow cytometry. n = 7–8 mice per genotype; mean ± SD; Mann–Whitney test, P > 0.05. (K) Experimental workflow for analysis of postnatal retina angiogenesis. (L) Staining of P5 and P7 retina for Pecam1 (black). Scale bar, 1 mm. Magenta circle: outline of the wild-type retina vasculature at the indicated time point; yellow arrow: vascular outgrowth from optic nerve. (M) Quantification of vascular outgrowth for P5 and P7 wild-type and Myct1^ecKO pups. n = 3–6 mice per genotype; mean ± SD; multiple Mann–Whitney tests, P = 0.009 (*) at P5 and P > 0.05 at P7. (N) Quantification of vascular density for P5 and P7 wild-type and Myct1^ecKO pups. n = 3–4 mice per genotype; mean ± SD; multiple Mann–Whitney tests, P > 0.05. (O) Quantification of CD45⁺ immune cells as percentage of total cells in fat pad as measured by flow cytometry. n = 7–8 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (P) Quantification of CD45^negCD31^negSca1⁺ adipocyte progenitor cells as percentage of total cells in fat pad as measured by flow cytometry. n = 7–8 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (J, O, and P) Data acquired during the same experiments. (Q) Experimental workflow for ex vivo adipogenesis assay. (R)Myct1 ablation does not affect ex vivo adipogenesis. SVF isolated from the inguinal fat pad of wild-type and Myct1^ecKO mice was treated with control or adipogenic cocktail for 4 days. Staining of SVF for perilipin (gray) and DNA (blue). Scale bar, 200 µm. Gonadal SVF provided similar results (data not shown). (S) Quantification of perilipin⁺ area normalized to cell number for SVF. n = 9 mice per condition and genotype; mean ± SD; two-way ANOVA, P > 0.05. (T)Myct1 ablation limits WAT (yellow, adipocytes) expansion independent of angiogenesis (red, vasculature). Icons used in A, K, and T were created with BioRender.com and modified in Affinity. See also Fig. S2.

Figure S2.

Reduced adiposity in Myct1 ^ecKO mice is independent of angiogenesis, adipogenesis, and whole-body metabolic activity, related to Fig. 2. (A) Body composition is similar between wild-type and Myct1^ecKO mice after 4 wk of HFD at the end of the metabolic cage experiment. n = 10 mice per genotype; mean ± SD; multiple Mann–Whitney tests, P > 0.05. (B) Water intake is similar between wild-type and Myct1^ecKO mice. n = 10 mice per genotype; mean ± SD; Mann–Whitney test, P > 0.05. (C) Energy expenditure over lean and body weight is similar between wild-type and Myct1^ecKO mice. n = 10 mice per genotype. (D)Myct1 ablation does not change fecal lipid content. Measurement of fecal lipids (mg) normalized to amount of feces collected (g). n = 4–5 mice per genotype; mean ± SD; Mann–Whitney test, P > 0.05. (E)Myct1 ablation does not alter blood metabolic parameters. Blood plasma was analyzed for concentrations of glucose, triglycerides, free fatty acids, and cholesterol from wild-type and Myct1^ecKO mice of the long-term Myct1 deletion experiment: 46-wk-old females in fed state (left), 50-wk-old females after 24-h fasting (middle), and 50-wk-old males after 6-h fasting (right). n = 4–9 mice per genotype; mean ± SD; unpaired t test, P > 0.05. (F)Myct1 ablation does not alter tolerance to lipid. Oral lipid tolerance test performed after 6 h of fasting. n = 4–5 mice per genotype; mean ± SD; mixed effects model, P > 0.05. (G)Myct1 ablation does not affect vascular permeability. Evans blue tissue content was assessed 30 min after i.v. injection in wild-type and Myct1^ecKO mice. n = 3 mice per genotype, 1 non-injected mouse shown as control; mean ± SD; multiple paired t test with Holm–Sidak’s correction for multiple comparisons, P > 0.05. (H) Experimental workflow for analysis of tumor angiogenesis. (I)Myct1 ablation reduces tumor growth. Quantification of LLC tumor volume from wild-type and Myct1^ecKO mice from 6 to 15 days after implantation. n = 8 mice per genotype; mean ± SD; multiple unpaired t tests, P = 0.032 (*). (J)Myct1 ablation reduces tumor weight. Quantification of tumor weight from wild-type and Myct1^ecKO LLC tumors at end of experiment (day 15). n = 8 mice per genotype; mean ± SD; unpaired t test, P = 0.036 (*). (K)Myct1 ablation does not affect tumor vascular density. Staining of LLC tumor sections for Vegfr2 (black). Scale bar, 1 mm. (L)Myct1 ablation does not affect tumor vascular density. Quantification of Vegfr2⁺ tumor area in wild-type and Myct1^ecKO LLC size-matched tumors shown in K. n = 6 mice per genotype; mean ± SD; unpaired t test, P > 0.05. Icons used in E and H were created with BioRender.com and modified in Affinity.

We further assessed alternative mechanisms for reduced adiposity in Myct1^ecKO mice. Masson’s trichrome staining revealed no evidence of increased fibrosis in WAT of Myct1^ecKO mice compared with wild type (Fig. 2, F and G). Fecal fat content did not differ significantly between genotypes, indicating intact intestinal lipid absorption (Fig. S2 D). Metabolic profiling under fed and fasted conditions, including oral lipid tolerance tests, showed no significant differences in plasma triglycerides, free fatty acids, glucose, or cholesterol levels (Fig. S2, E and F). Evans blue extravasation further confirmed unaltered vascular permeability in WAT or other examined tissues (Fig. S2 G). Collectively, these results exclude altered lipid absorption, vascular leakage, and tissue fibrosis as contributors to impaired WAT expansion in Myct1-deficient mice.

Given the critical role of vascularization in adipose tissue homeostasis (AlZaim et al., 2023), we next assessed whether Myct1 deficiency alters WAT vasculature. Immunostaining of retroperitoneal WAT sections for the pan-endothelial marker Pecam1 showed comparable vessel density between Myct1^ecKO and wild-type mice (Fig. 2, H and I). Consistent with this, flow cytometry analysis of WAT stromal vascular fraction (SVF) revealed no differences in Pecam1/CD31⁺ EC content (Fig. 2 J). These results demonstrate that Myct1 deletion does not impair vascular density or endothelial abundance in adipose tissue and is therefore unlikely to act as a direct regulator of angiogenesis. Accordingly, analysis of retinal angiogenesis in pups showed minimal and transient delay in vascular outgrowth and no difference in vascular density at postnatal day P5 between Myct1^ecKO and wild-type mice (Fig 2, K–N).

Prior work reported that Myct1 inactivation reduced tumor angiogenesis and impaired tumor growth (Kabir et al., 2021). Given the shared reliance of tumors and expanding WAT on vascular support—including rapid expansion, high metabolic demands, and nutrient dependency—we tested whether Myct1 deficiency similarly affects tumor progression. We induced endothelial-specific Myct1 deletion in 12–14-wk-old mice via a 2-wk tamoxifen diet, followed by s.c. implantation of Lewis lung carcinoma (LLC) cells (Fig. S2 H). Consistent with published findings, tumor volume and weight were significantly reduced in Myct1^ecKO versus control mice (Fig. S2, I and J). However, vascular density in size-matched tumors assessed by Vegfr2 staining was unaltered (Fig. S2, K and L), indicating an angiogenesis-independent role for endothelial Myct1 in regulating both WAT expansion and tumor growth.

Given the essential roles of immune cells and adipocyte progenitor cells in WAT metabolic homeostasis (Brestoff and Artis, 2015; Lecoutre et al., 2025), we analyzed retroperitoneal fat by flow cytometry. This revealed no differences in CD45⁺ immune cell infiltration or adipocyte progenitor cell populations (CD31^negCD45^negSca1⁺) between genotypes (Fig. 2, O and P). Furthermore, ex vivo differentiation assays demonstrated comparable adipogenic potential in SVF cells isolated from inguinal fat pads of Myct1^ecKO and wild-type mice (Fig. 2, Q–S). Collectively, these data establish that Myct1 deletion limits WAT expansion through mechanisms independent of angiogenesis, adipogenesis, and systemic metabolic activity (Fig. 2 T).

Ablation of MYCT1 upregulates pathways involved in protein folding, ribonucleoprotein assembly, and mTORC1 signaling

To elucidate the mechanism underlying reduced WAT expansion following endothelial MYCT1 ablation, we focused on adipose ECs, as they constitute the primary interface for nutrient and signal delivery to the adipose tissue (Robciuc et al., 2016; Crewe et al., 2018; Monelli et al., 2022). We isolated CD45^negCD31⁺ cells from mesenteric fat SVFs of wild-type and Myct1^ecKO mice for single-cell RNA sequencing (scRNA-seq) (Fig. 3 A and Fig. S3 A). Unsupervised clustering analysis identified five distinct populations: a predominant BEC cluster, alongside LECs, pericytes/vascular smooth muscle cells, mesothelial/fibroblast cells, and immune cells (Fig. 3 B and Fig. S3 B). Myct1 expression was high in control ECs, but absent in Myct1^ecKO BECs, confirming efficient deletion (Fig. 3 C). Differential gene expression analysis revealed maximal transcriptional alterations in BECs versus minimal changes in other clusters (Fig. S3 C).

Figure 3.

Ablation of Myct1 upregulates pathways involved in protein folding, ribonucleoprotein assembly, and mTORC1 signaling. (A) Experimental workflow of the scRNA-seq experiment. CD45^negCD31⁺ cells were sorted from mesenteric fat of 12-wk-old wild-type and Myct1^ecKO mice. n = 2 per genotype. (B) UMAP representation of cells retained in the scRNA-seq and colored according to cell type. BECs, blood endothelial cells; LECs, lymphatic endothelial cells; vSMCs, vascular smooth muscle cells. (C)Myct1 is efficiently deleted from ECs in Myct1^ecKO mice. Distribution of Myct1, Pecam1, and Cdh5 expression (ln[normalized counts +1]) in the BEC cluster in wild-type and Myct1^ecKO cells. (D) GO terms and Hallmark gene sets overrepresented among upregulated (red bars) or downregulated (gray bars) genes in Myct1^ecKO BECs compared with wild-type BECs. (E)Myct1 deficiency in ECs increases protein levels of mTORC1 target Mat2a (methionine adenosyltransferase 2A). Staining of WAT of Myct1^ecKO or wild-type control mice for Mat2a (green) and pan-endothelial marker Erg (magenta). Scale bar, 50 µm. (F) Quantification of data shown in E. n = 8 mice per genotype from three independent experiments. mean ± SD; paired t test, P = 0.0191 (*). Icons used in A were created with BioRender.com and modified in Affinity. See also Fig. S3, A–D.

Figure S3.

MYCT1 limits endothelial mTORC1 signaling, related to Figs. 3 and 4. (A) Flow cytometry gating strategy (CD45^negCD31⁺) for sorting of ECs from mesenteric fat for scRNA-seq. (B) Dot plot of markers for the indicated clusters. Color code: scaled average expression level in each cluster; the dot size denotes the percent of cells in each cluster expressing the given gene. (C) Number of differentially expressed genes (DEGs) between wild-type and Myct1^ecKO cell clusters. (D) Volcano plot of DEGs between the wild-type and Myct1^ecKO mice in the BEC cluster. Mat2a gene was selected for scRNA-seq validation. Mat2a, methionine adenosyltransferase 2A. (E) MYCT1 protein levels in human primary ECs. Western blot analysis for the indicated proteins. HPMECs, human pulmonary ECs; HUVECs, human umbilical vein ECs; HIECs, human intestinal ECs. (F and G) MYCT1 antibody and siRNAs validation for identification of endogenous human MYCT1 protein. Human primary ECs were transfected with two different MYCT1 targeting siRNAs. (F) Staining of ECs for MYCT1 (black) and DAPI (blue). Arrow, not transfected EC. Scale bar, 50 μm. (G) Western blot analysis showing MYCT1 migration profile and siRNA specificity. (H)MYCT1 knockdown increases phosphorylation of S6 but does not affect AKT and ERK1/2 phosphorylation status. Western blot analysis for the indicated proteins. (I) Quantification of p-S6^Ser240/244 levels normalized to total S6 (tot-S6). P = 0.037 (*). (J) Quantification of p-AKT^Ser473 levels normalized to total AKT (tot-AKT). P > 0.05. (K) Quantification of p-AKT^Thr308 levels normalized to total AKT (tot-AKT). P > 0.05. (L) Quantification of p-ERK1/2^{Thr202/Tyr204} levels normalized to total ERK1/2 (tot-ERK1/2). P > 0.05. (M) Quantification of MYCT1 levels normalized to vinculin. P = 0.001 (*). (I–M)n = 5 independent experiments; paired t tests. (N)MYCT1 knockdown increases phosphorylation of p70/S6 kinase (p70/S6K), a key downstream effector of mTORC1 signaling, in response to amino acids. Western blot for the indicated proteins. (O) Quantification of data shown in N. n = 2 independent experiments; mean ± SD; two-way ANOVA with Tukey’s multiple comparison test, P = 0.0194 (*). (P)MYCT1 knockdown hyperactivates mTORC1 signaling in response to amino acids. 2 days after siRNA transfection, confluent ECs were serum- and growth factor–starved overnight, then starved in PBS for 1 h before 30-min stimulation with amino acids, glucose, growth factors, FBS, their combination, or PBS as control. Staining of ECs for p-S6 (gray), VE-cadherin (magenta), and DAPI (blue). Scale bar, 50 μm. Source data are available for this figure: SourceData FS3.

To identify biological processes associated with Myct1 deletion in BECs, we performed overrepresentation analysis using Gene Ontology (GO) and Hallmark gene sets. Surprisingly, this demonstrated significant enrichment in Myct1^ecKO BECs for pathways involving mTORC1 signaling, protein folding, and ribonucleoprotein complex assembly—processes mechanistically linked to mTORC1 activity (Qian et al., 2010; Mayer and Grummt, 2006) (Fig. 3 D; Fig. S3 D; and Tables S1 and S2). Our scRNA-seq analysis was based on a small sample size of two mice per genotype; however, we validated these findings by immunostaining for methionine adenosyltransferase 2A, a direct target of mTORC1 signaling (Villa et al., 2021), which was markedly increased in Myct1-deficient ECs (Fig. 3, E and F). Collectively, scRNA-seq data indicate that Myct1^ecKO BECs exhibit increased mTORC1 signaling and enhanced biosynthetic activity.

MYCT1 limits endothelial mTORC1 signaling

To investigate MYCT1’s role in mTORC1 signaling, we used confluent and quiescent human primary pulmonary ECs (HPMECs) or intestinal ECs (HIECs), as they express high levels of MYCT1 as compared with human umbilical vein ECs (HUVECs) (Fig. S3 E). Following validation of MYCT1 knockdown efficiency and antibody specificity through immunostaining and western blot (Fig. S3, F and G), we assessed key signaling pathways in control and MYCT1^KD cells. MYCT1-depleted ECs exhibited markedly increased phosphorylation of p70/S6 kinase at Thr389 (p-p70/S6K) and ribosomal protein S6 at Ser240/244 (phospho-S6 [p-S6]), canonical mTORC1 activity readouts, while phosphorylation states of AKT (Ser473/Thr308) and ERK1/2 (Thr202/Tyr204) remained unchanged (Fig. 4, A and B; and Fig. S3, H–O). Rapamycin treatment abolished p-S6 elevation, confirming mTORC1 dependence (Fig. 4, A and B).

Figure 4.

MYCT1 limits endothelial mTORC1 signaling. (A) MYCT1 knockdown in quiescent ECs hyperactivates mTORC1 signaling. Confluent ECs were cultured in complete medium for 1.5 days after siRNA transfection and treated overnight with 10 nM mTORC1 inhibitor rapamycin or DMSO as a control. Staining for p-S6 (gray), VE-cadherin (magenta), and DAPI (blue). Scale bar, 50 μm. (B) Quantification of mTORC1 activation in control and MYCT1^KD cells in the presence or absence of rapamycin. The percentage of p-S6⁺ cells was quantified in the indicated conditions. n = 6 independent experiments (3 in the presence of rapamycin); 500–5,000 cells were analyzed per condition for each experiment; two-way ANOVA with Tukey’s multiple comparisons test, P = 0.011 (*) for MYCT1 knockdown effect in control conditions and P < 0.0001 (*) for rapamycin effect on MYCT1^KD cells. (C) MYCT1 limits protein synthesis. Confluent control and MYCT1^KD ECs were cultured in complete medium for 2 days after siRNA transfection and treated for 50 min with 10 µM OPP. Staining for OPP (gray), VE-cadherin (magenta), and DNA (blue). Arrowhead, OPP puncta. Scale bar, 20 µm. (D) Quantification of OPP puncta per cell in control and MYCT1^KD cells. n = 3 independent experiments; 20–40 cells analyzed per condition for each experiment; mean ± SD; Welch’s t test, P = 0.042 (*). (E)MYCT1 knockdown hyperactivates mTORC1 signaling in response to amino acids. Confluent ECs were starved for 1 h and restimulated with amino acids for 30 min. Western blot analysis for the indicated proteins. (F)MYCT1 knockdown hyperactivates mTORC1 signaling in response to amino acids. Staining for p-S6 (gray), VE-cadherin (magenta), and DAPI (blue). Scale bar, 50 μm. (G) Quantification of mTORC1 activation in response to amino acid supplementation. The percentage of p-S6⁺ cells was quantified in the indicated conditions. n = 4 independent experiments; 150–500 cells were analyzed per condition for each experiment; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons test, P = 0.002 (*) for MYCT1 knockdown effect under amino acid supplementation. (H)Myct1 ablation hyperactivates mTORC1 signaling in vivo in a feeding status–dependent manner. Wild-type and Myct1^ecKO mice were starved overnight or fed ad libitum. En face staining of aorta for p-S6 (gray) and VE-cadherin (magenta). Scale bar, 50 μm. (I) Quantification of mTORC1 activation in the aortic endothelium of wild-type and Myct1^ecKO mice, starved or fed ad libitum. The percentage of p-S6⁺ cells was quantified in n = 4–6 mice per genotype and conditions; 300–500 cells were analyzed per aorta; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons test, P = 0.026 (*) for Myct1 ablation effect in fed mice. (J)Myct1 ablation hyperactivates mTORC1 signaling in ECs of the retroperitoneal fat pad. Mice fed ad libitum. Staining for p-S6 (gray/black), endomucin (magenta), and DNA (blue). Arrowheads, high endothelial p-S6⁺ signal. Scale bar, 20 μm. (K) Quantification of p-S6 intensity in ECs of the retroperitoneal fat pad. p-S6 intensity was quantified in endomucin⁺ (EMCN⁺) area and normalized to wild type. n = 6 mice per condition; mean ± SD; paired t test, P = 0.021 (*). (L)Myct1 ablation hyperactivates mTORC1 signaling in capillary ECs of the mesenteric fat pad. Mice fed ad libitum. Whole-mount staining for p-S6 (gray/black) and Pecam1 (magenta). Scale bar, 20 μm. (M) Schematic view of MYCT1 limitation of mTORC1 signaling in ECs. MYCT1 ablation hyperactivates mTORC1 signaling in response to amino acids and nutrients. Icons used in M were created with BioRender.com and modified in Affinity. See also Fig. S3, E–P. Source data are available for this figure: SourceData F4.

Given that mTORC1 activation drives anabolic programs, including protein synthesis (Valvezan and Manning, 2019), we next assessed protein production in ECs. To quantify global protein production, we used O-propargyl-puromycin (OPP), a puromycin analog that incorporates into nascent polypeptide chains. MYCT1 knockdown significantly increased OPP incorporation (Fig. 4, C and D), indicating enhanced protein synthesis. This result is consistent with our scRNA-seq data, which revealed upregulation of protein synthesis pathways in Myct1-deficient endothelium.

mTORC1 integrates diverse environmental cues, including nutrients and growth factors (Valvezan and Manning, 2019). To identify which signal drives mTORC1 hyperactivation in MYCT1-deficient ECs, we subjected confluent control and MYCT1^KD ECs to nutrient and growth factor deprivation, followed by acute (30-min) stimulation with (1) amino acids, (2) glucose, (3) growth factors, (4) FBS, or (5) all factors combined. mTORC1 activity was assessed via detection of p-S6. In control ECs, amino acid supplementation alone induced modest mTORC1 activation, while glucose, growth factors, or FBS alone failed to elicit significant responses (Fig. S3 P). As expected, the combined treatment (5) enhanced mTORC1 activation beyond amino acids alone (1) (Fig. S3 P). Strikingly, MYCT1-depleted ECs exhibited amplified mTORC1 signaling: amino acids alone triggered hyperactivation, while the combined treatment induced even stronger p-S6 phosphorylation (Fig. 4, E–G; and Fig. S3 P). These results establish MYCT1 as a regulator of endothelial mTORC1 pathway sensitivity to nutrient availability.

To evaluate mTORC1 activation status in vivo at single-cell resolution, we employed en face aorta imaging (Mack et al., 2017). In ad libitum–fed control mice, aortic endothelia exhibited rare p-S6⁺ cells (Fig. 4, H and I). In agreement with in vitro data, Myct1^ecKO mice displayed a significantly higher prevalence of p-S6⁺ ECs. This hyperactivation was nutrient dependent, as overnight starvation abolished mTORC1 activity in both wild-type and Myct1^ecKO mice (Fig. 4, H and I). Consistent with WAT phenotypes, elevated mTORC1 signaling was also observed in Myct1^ecKO vessels of retroperitoneal and mesenteric adipose depots (Fig. 4, J–L).

Collectively, these data establish MYCT1 as a critical negative regulator of nutrient-sensing mTORC1 signaling in vascular endothelium (Fig. 4 M). From an evolutionary perspective, this regulatory mechanism optimizes nutrient partitioning to WAT—ensuring efficient energy storage during abundance and mobilization during scarcity. Such endothelial metabolic gatekeeping would have conferred survival advantages by supporting energetically demanding processes like immune responses during famine cycles (Hotamisligil and Erbay, 2008).

MYCT1 is a transmembrane phosphoglycoprotein that interacts with IFITM2/3

Given the lack of detailed characterization of MYCT1, we focused on examining its endogenous form to better understand its properties and interactions. MYCT1 is evolutionary conserved, with sequence alignment revealing high protein homology among representative vertebrates (Fig. S4 A), including several conserved domains (5–15 aa) in the cytoplasmic tail. This conservation, combined with the absence of other family members and the lack of sequence similarity to other proteins, points to a role for MYCT1 in fundamental vertebrate-specific processes.

Figure S4.

MYCT1 is a transmembrane phosphoglycoprotein that interacts with IFITM2/3, related to Fig. 5. (A) The MYCT1 protein is highly conserved across vertebrates. An alignment of human MYCT1 protein sequence with those of the indicated species, amino acid conservation is color-coded as indicated in the legend below. Percentages next to species indicate amino acid sequence identity (left) and homology (right) in comparison with human sequence. (B) Short MYCT1 isoform is predominant in ECs. Cells were transduced with Ad-GFP (control) or Ad-MYCT1 (187-aa isoform with C-terminal V5 tag) adenoviruses. Western blot analysis for the indicated proteins. (C) Workflow for mass spectrometry experiments. MYCT1-negative SW480 colon cancer cells were used to exclude nonspecific interactors pulled down by MYCT1 IgG. n = 2 independent experiments. (D) Venn diagram showing how the short list of MYCT1 interactors was selected. (E) Interaction between MYCT1 and IFITM2/3 was analyzed by PLA in ECs. siRNA-mediated knockdown of either protein confirmed specificity of PLA signal. Staining of ECs for PLA dots (gray), VE-cadherin (magenta), and DNA (blue). Scale bar, 50 µm. (F) Quantification of the number of PLA dots per cell in control, MYCT1^KD, and IFITM2/3^KD cells. n = 3 independent experiments; 500–1,500 cells were analyzed per condition for each experiment; mean ± SD; one-way ANOVA with Dunnett’s multiple comparisons, P = 0.036 (*) for MYCT1 knockdown effect. (G) Positive control (VE-cadherin::β-catenin) and negative control (MYCT1 antibody alone) for PLA signal in human brain sections. PLA dots (gray) and staining of ECs for VE-cadherin or Pecam1 (magenta) and DNA (blue). Arrowhead, colocalization of PLA dots with vascular marker. Scale bar, 20 µm. Source data are available for this figure: SourceData FS4.

Analysis of endogenous MYCT1 in quiescent and confluent primary ECs revealed its predominant localization at cell–cell junctions and isolated puncta (Fig. 5 A). MYCT1 was initially reported to be a nuclear protein (Wu et al., 2016); however, cell fractionation experiments confirmed MYCT1 as a membrane protein (Fig. 5 B), supporting previous studies that used tagged exogenous constructs (Wu et al., 2016; Kabir et al., 2021; Aguadé-Gorgorió et al., 2024).

Figure 5.

MYCT1 is a transmembrane phosphoglycoprotein that interacts with IFITM2/3. (A) Endogenous MYCT1 is located at cell–cell junctions (arrow) and in puncta (arrowhead). Staining of human primary ECs for MYCT1 (black), VE-cadherin (magenta), and DNA (blue). Scale bar, 10 µm. (B) MYCT1 is a membrane protein. Western blot analysis of various EC fractions for MYCT1, GAPDH, PECAM1, H3K27ac, and vimentin proteins. Cy, cytoplasm; Mb, membrane; Nu, nucleus; Ck, cytoskeleton. (C) MYCT1 is glycosylated. Western blot analysis of MYCT1 protein electrophoretic mobility in control and PNGase-F–treated lysates. (D) Schematic model of MYCT1 structure and domains with phosphorylation sites, identified by mass spectrometry. MYCT1 phosphorylation sites are highly conserved as indicated by the color scale. Asterisks indicate sites also described at https://www.phosphosite.org/. (E) Top five proteins interacting with MYCT1 as identified by mass spectrometry, among which IFITM2 and IFITM3. ECs were transduced with recombinant adenoviruses to transiently overexpress MYCT1 or GFP, as a control. Cell lysates were collected 48 h after transduction, immunoprecipitated using MYCT1 antibody or a control IgG, and analyzed by mass spectrometry. Proteins interacting with both endogenous and overexpressed MYCT1 were selected and ranked by normalized spectral abundance factor (NSAF) from two independent mass spectrometry (MS) experiments are shown (31 proteins); the top five proteins are highlighted in magenta. (F) GO terms of the cellular component and biological process overrepresented in the MYCT1 interactome. Fisher’s exact test with adjustment for false discovery rate (FDR). (G) Validation of IFITM2/3 and MYCT1 interaction by co-IP. EC lysates from confluent ECs were immunoprecipitated (IP) with MYCT1 or control IgG and blotted for IFITM2/3. H, IgG heavy chain; L, IgG light chain. (H) IFITM2/3 are constitutively expressed in ECs in vitro and in vivo. Staining of human primary ECs (upper panels), human brain and WAT sections (lower panels) for MYCT1 (gray), IFITM2/3 (green), VE-cadherin (magenta), and DNA (blue). Scale bar, 20 µm (brain) and 50 µm (adipose tissue). (I) MYCT1 and IFITM2/3 interact in brain ECs. Proximity ligation assay (PLA) in human brain sections. Detection of PLA dots (gray) in ECs and staining for VE-cadherin (magenta) and DNA (blue). Arrowheads, colocalization of MYCT1::IFITM2/3 PLA dots and VE-cadherin staining. Scale bar, 20 µm. See also Fig. S4. Source data are available for this figure: SourceData F5.

The electrophoretic migration profile of endogenous MYCT1 showed a broad and diffuse pattern (Fig. 5 B and Fig. S3 G), suggesting the presence of posttranslational modifications. Treatment with Peptide-N-glycosidase F (PNGase F), an enzyme that removes N-linked glycans, significantly reduced the molecular weight of MYCT1 (Fig. 5 C), indicating extensive N-linked glycosylation. De-glycosylated MYCT1 migrated at ∼20 kDa, lower than the predicted 26.6 kDa for the full-length 235-aa protein (UniProt Q8N699) but matching the 187-aa isoform (UniProt D6Q1S4), initiated at a conserved methionine. The long isoform is predicted to have two transmembrane domains (Aguadé-Gorgorió et al., 2024), while the short would possess only one (Phobius prediction) (Käll et al., 2007). To determine the predominant form, ECs were transduced with Ad-GFP (control) or Ad-MYCT1 adenoviruses, the latter encoding 187-aa V5-tagged isoform of MYCT1. Western blot showed similar migration of endogenous and Ad-MYCT1 (Fig. S4 B), confirming that MYCT1 is a single-transmembrane protein in ECs.

Mass spectrometry of immunoprecipitated MYCT1 from Ad-MYCT1–transduced ECs revealed extensive serine and threonine phosphorylation of the cytoplasmic part, which clustered within two conserved domains, along with a conserved p-Tyr residue (Fig. 5 D, Fig. S4 A, and Table S3). Although MYCT1 shares no sequence similarity with known proteins, with a short extracellular domain and a long, heavily phosphorylated cytoplasmic tail, it resembles structurally transmembrane adaptor proteins, which act as key organizers of membrane-proximal signaling complexes in immune cells, such as LAT and LIME (Hořejší et al., 2004). In all, our biochemical analysis confirms MYCT1 as a vertebrate specific, single-pass transmembrane phospho-glycoprotein.

To gain further insight into the functional role of MYCT1, we conducted an unbiased mass spectrometry analysis to identify associated proteins. Two independent mass spectrometry experiments were conducted using quiescent primary human ECs transduced with either MYCT1-expressing adenovirus or a control GFP-expressing adenovirus. Immunoprecipitation was performed using MYCT1 or a control IgG antibody targeting GFP (Fig. S4 C). In addition, we excluded nonspecific MYCT1 antibody interactors immunoprecipitated from a MYCT1-negative cancer cell line SW480 (Fig. S4 C). For analysis, we focused on 31 high-confidence proteins that were immunoprecipitated with both endogenous and overexpressed MYCT1 across both mass spectrometry experiments (Fig. 5 E, Fig. S4 D, and Table S4).

To prioritize candidates for functional follow-up, we performed GO enrichment analysis. This revealed significant overrepresentation of proteins associated with endosomal and membrane compartments as well as interferon (IFN) signaling (Fig. 5 F and Table S5). Among the top interactors (Fig. 5 E), IFITM2 and IFITM3 (collectively referred to as IFITM2/3) stood out due to their known roles in endosomal trafficking and IFN-mediated antiviral responses (Majdoul and Compton, 2022), which align closely with the enriched GO categories. Of note, germline deletion of the IFITM locus in mice leads to obesity (Wee et al., 2015), a phenotype opposite to that observed in Myct1^ecKO mice.

As IFITM2/3 function in ECs is not well understood, we selected IFITM2/3 for further study. We additionally confirmed the interaction of MYCT1 with IFITM2/3 in ECs by co-immunoprecipitation (Fig. 5 G) and proximity ligation assay (PLA) (Fig. S4, E and F). IFITM2/3 levels are typically low and induced by IFN stimulation in epithelial cells and fibroblasts (Brass et al., 2009; Sun et al., 2016), yet quiescent primary human ECs produce high levels of IFITM2/3 even without IFN stimulation (Fig. 5 H, upper), consistent with previous findings (Sun et al., 2016). Immunostaining of human brain and WAT sections confirmed high levels of IFITM2/3 proteins in the vascular endothelium (Fig. 5 H, lower). To validate our in vitro findings, we performed PLA directly on brain sections and confirmed the interaction of MYCT1 with IFITM2/3 in vivo (Fig. 5 I and Fig. S4 G), supporting its relevance in human physiology.

MYCT1 limits enlargement of IFITM2/3⁺ early endosomes

We next sought to elucidate the interdependence of MYCT1 and IFITM2/3 by examining the effects of individual protein depletion. Western blot analysis revealed reduced MYCT1 protein levels in IFITM2/3-deficient cells (Fig. 6 A), which was confirmed by immunofluorescence (Fig. S5, A and B). In contrast, MYCT1 ablation increased total IFITM2/3 protein levels (Fig. 6 A), while mRNA levels remained unchanged (Fig. S5, C and D). Immunofluorescence analysis showed localization of IFITM2/3 in enlarged intracellular vesicular structures in MYCT1^KD cells. (Fig. 6, B and C). IFITM2/3 proteins are predominantly located to late endosomes and lysosomes in various cell types (Feeley et al., 2011; Jia et al., 2012; Spence et al., 2019). We examined colocalization of endogenous IFITM2/3 in ECs with early endosomal RAB5, late endosomal RAB7, and lysosomal LAMP1 markers. We found that in ECs IFITM2/3 predominantly localized to the RAB5⁺ early endosome (Fig. 6, D and E). In agreement with this observation and enlarged IFITM2/3⁺ vesicles, MYCT1-deficient cells exhibited expanded RAB5⁺ early endosomal compartments (Fig. 6, F and G), while no changes were found in RAB7⁺ late endosomal or LAMP1⁺ lysosomal compartments (Fig. S5, E–G). Crucially, Rab5⁺ early endosome was also enlarged in Myct1^ecKO mice en face aorta (Fig. 6, H and I). Collectively, these data indicate that in the absence of MYCT1, IFITM2/3 proteins accumulate in and enlarge RAB5⁺ early endosomes (Fig. 6 J).

Figure 6.

MYCT1 limits enlargement of IFITM2/3⁺ endosomes. (A) MYCT1 and IFITM2/3 display inverse regulatory dynamics. MYCT1 knockdown increases total IFITM2/3 protein levels, whereas IFITM2/3 knockdown reduces total MYCT1 protein levels. Western blot analysis of confluent ECs 48 h after siRNA transfection for the indicated proteins. Quantification of total MYCT1 and IFITM2/3 protein levels relative to the control is indicated as the average of all experiments under the respective blots. n = 4 independent experiments; one-way ANOVA with Dunnett’s multiple comparisons test; P < 0.0001 (*) for MYCT1 knockdown effect on MYCT1, P = 0.004 (*) for IFITM2/3 knockdown effect on MYCT1, P = 0.015 (*) for MYCT1 knockdown effect on IFITM2/3, and P = 0.0016 (*) for IFITM2/3 knockdown effect on IFITM2/3. (B)MYCT1 knockdown increases IFITM2/3⁺ vesicle volume. Staining of ECs for MYCT1 (gray), IFITM2/3 (green), VE-cadherin (magenta), and DNA (blue). Yellow box: magnification of IFITM2/3 from highlighted areas, shown below. Scale bar, 20 µm. (C) Quantification of the IFITM2/3⁺ vesicle volume in control and MYCT1^KD cells. n = 4 independent experiments; 500–1,500 vesicles from 15 to 20 cells were analyzed per condition for each experiment; mean ± SD; unpaired t test; P = 0.039 (*). (D) IFITM2/3 localize mainly to the RAB5⁺ early endosome in ECs, rather than to RAB7⁺ late endosome and LAMP1⁺ lysosome. Staining for IFITM2/3 (gray), RAB5 or RAB7 or LAMP1 (green), and DNA (blue). Yellow box: magnification of IFITM2/3 and endolysosomal markers from highlighted areas, shown below. Yellow arrows, IFITM2/3⁺RAB5⁺ vesicles; magenta arrows, IFITM2/3⁺RAB7^neg or IFITM2/3⁺LAMP1^neg vesicles. Scale bar, 20 µm. (E) Quantification of colocalization of IFITM2/3 with RAB5, RAB7, and LAMP1 by Manders’ overlap coefficients. n = 2–4 independent experiments; 15–25 cells were analyzed per condition for each experiment; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons test, P < 0.001 (*) for both RAB5 versus RAB7 and RAB5 versus LAMP1. (F)MYCT1 knockdown causes enlargement of RAB5⁺ early endosomes. Staining of ECs for RAB5 (gray/black), VE-cadherin (magenta), and DAPI (blue). Cyan box: magnification of RAB5 from highlighted areas, shown below. Scale bar, 10 μm. (G) Quantification of RAB5⁺ vesicle volume. n = 4 independent experiments; 400–800 vesicles were analyzed per condition for each experiment; mean ± SD; unpaired t test, P = 0.0133 (*). (H)Myct1 ablation enlarges Rab5⁺ early endosomes in aortic ECs. En face staining of aorta from wild-type or Myct1^ecKO mice for Rab5 (gray/black) and VE-cadherin (magenta). Cyan box: magnification of RAB5 from highlighted areas, shown below. Scale bar, 5 μm. (I) Quantification of Rab5⁺ vesicle volume in aortic ECs. n = 5 mice per genotype; 1,000–4,000 vesicles were analyzed per mouse; mean ± SD; unpaired t test, P = 0.0128 (*). (J)MYCT1 depletion leads to IFITM2/3⁺ early endosome enlargement. Icons used in J were created with BioRender.com and modified in Affinity. See also Fig. S5, A–G. Source data are available for this figure: SourceData F6.

Figure S5.

MYCT1 restricts endothelial endocytosis and IFITM2/3-dependent mTORC1 activation, related to Figs. 6 and 7. (A) IFITM2/3 antibody and siRNA validation for identification of endogenous human IFITM2/3 proteins. IFITM2/3 knockdown reduces MYCT1 protein levels. Staining of ECs for MYCT1 (gray), IFITM2/3 (green), and DNA (blue). Scale bar, 20 µm. (B) Quantification of MYCT1 protein levels in control and IFITM2/3^KD cells. n = 4 independent experiments; mean ± SD; Welch’s t test, P = 0.0014 (*). (C and D)MYCT1 knockdown does not affect IFITM2 (C) nor IFITM3 (D) mRNA levels in ECs. n = 3 independent experiments; mean ± SD; Welch’s t test, P > 0.05. (E)MYCT1 knockdown does not impact RAB7⁺ late endosomes nor LAMP1⁺ endolysosomes. Staining of ECs for RAB7 (gray), LAMP1 (green), VE-cadherin (magenta), and DNA (blue). Scale bar, 20 µm. (F and G) Quantification of RAB7⁺ (F) and LAMP1⁺ (G) areas per cell in control and MYCT1^KD cells. n = 3 independent experiments; 20–50 cells were analyzed per condition for each experiment; mean ± SD; Welch’s t test, P > 0.05. (H)MYCT1 knockdown increased FITC-dextran uptake. 2 days after siRNA transfection, cells were starved for 1 h in PBS, followed by a 30-min induction with amino acid solution together with 10-kDa FITC dextran. Detection of 10-kDa FITC-dextran (gray) and staining of ECs for VE-cadherin (magenta) and DAPI (blue). Arrow, dextran⁺ puncta. Scale bar, 10 μm. (I) Quantification of the number of dextran⁺ puncta per cell in control and MYCT1^KD cells. n = 3 independent experiments; 30–50 cells were analyzed per condition for each experiment; mean ± SD; Welch’s t test, P = 0.0016 (*). (J) Example of gating strategy (7-AAD^neg CD45^neg CD31⁺) of ECs from gonadal fat pad by flow cytometry. (K) WAT ECs take up higher amounts of labeled plasma proteins compared with colon ECs. Quantification of labeled plasma protein uptake in ECs from s.c. and visceral WAT and colon normalized to plasma Atto-647 signal. n = 10 mice per organ; Friedman test with Dunn’s multiple comparisons test, P > 0.05 for scFAT versus visFAT, P = 0.0052 for scFAT versus colon, and P = 0.001 (*) for visFAT versus colon. (L) Endocytosis inhibition with dynasore rescues mTORC1 hyperactivation caused by knockdown of MYCT1. Staining for p-S6 (gray), β-catenin (magenta), and DAPI (blue). Scale bar, 50 μm. (M) Quantification of mTORC1 activation by amino acid supplementation in control and MYCT1^KD cells in the absence or presence of dynasore. The percentage of p-S6⁺ cells was quantified in the indicated conditions. n = 3 independent experiments; 1,500–6,000 cells were analyzed per condition for each experiment; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons test, P = 0.004 (*) for MYCT1 knockdown effect in control conditions and P < 0.001 (*) for its rescue by dynasore treatment. (N)RAB5 knockdown rescues mTORC1 hyperactivation in MYCT1^KD cells. Staining of ECs for p-S6 (gray), β-catenin (magenta), and DAPI (blue). Scale bar, 50 μm. (O) Quantification of mTORC1 activation in control, MYCT1^KD, and MYCT1-RAB5^KD cells. The percentage of p-S6⁺ cells was quantified in the indicated conditions. n = 3 independent experiments; 7,000-15,000 cells were analyzed per condition for each experiment; mean ± SD; one-way ANOVA with Tukey’s multiple comparisons test, P = 0.0021 (*) for MYCT1 knockdown effect and P = 0.0292 (*) for its rescue by RAB5 double knockdown. (P)IFITM2/3 knockdown rescues mTORC1 hyperactivation in MYCT1-deficient human adipose ECs. Staining for p-S6 (gray), β-catenin (magenta), and DAPI (blue). Scale bar, 100 μm. (Q) Quantification of mTORC1 activation in control, MYCT1^KD, IFITM2/3^KD, and MYCT1–IFITM2/3^KD cells. The percentage of p-S6⁺ cells was quantified in the indicated conditions. n = 2 independent experiments; 1,500–3,000 cells were analyzed per condition for each experiment; mean ± SD; one-way ANOVA with Tukey’s multiple comparisons test, P = 0.0168 (*) for MYCT1 knockdown effect and P = 0.0123 (*) for rescue effect by IFITM2/3 double knockdown.

MYCT1 restricts endothelial endocytosis and IFITM2/3-dependent mTORC1 activation

Given the essential role of RAB5 in receiving and sorting cargo from incoming endocytic vesicles (Naslavsky and Caplan, 2018) and a report indicating that MYCT1 affects endocytosis rate in HSCs (Aguadé-Gorgorió et al., 2024), we sought to investigate whether MYCT1 regulates endocytic uptake in ECs. Analysis of primary human ECs after 30-min exposure to low–molecular weight fluorescent dextran revealed that MYCT1 knockdown increased endocytic uptake (Fig. S5, H and I).

While dextran is commonly used to study endocytosis, it represents an artificial model that may not fully capture physiological cargo dynamics. To investigate physiologically relevant endocytosis in vivo, we labeled the mouse plasma proteome with Atto 647N NHS-ester fluorophore (Yang et al., 2020) and analyzed plasma protein uptake in ECs of wild-type and Myct1^ecKO mice by flow cytometry (Fig. 7 A and Fig. S5 J). We observed significant organ-specific differences in EC plasma protein uptake: adipose ECs from both visceral and s.c. fat pads internalized substantial amounts of labeled plasma proteins, whereas colonic ECs exhibited minimal uptake (Fig. 7 B and Fig. S5 K). We further analyzed the amounts of labeled proteins internalized by adipose ECs, through the mean fluorescence intensity levels of Atto 647, comparing it with the circulating levels of labeled proteins. This analysis revealed that Myct1^ecKO adipose ECs demonstrate an enhanced uptake of labeled plasma proteins compared with wild-type controls (Fig. 7 C).

Figure 7.

MYCT1 restricts endothelial endocytosis and IFITM2/3-dependent mTORC1 activation. (A) Experimental workflow for in vivo Atto647–labeled plasma protein uptake experiment. (B) WAT ECs take up higher amounts of labeled plasma proteins compared with colon ECs. Representative flow cytometry histograms showing Atto647-MFI in ECs from s.c. (sc. fat), visceral (vis. fat) WAT, and colon of two wild-type mice, comparing an injected one (colored) and a non-injected control (gray). (C)Myct1 ablation increases labeled plasma protein uptake in WAT ECs. Atto647-MFI of ECs from visceral WAT of control mice or Myct1^ecKO mice, plotted as a function of corresponding plasma Atto647-fluorescence levels. Each dot corresponds to an individual mouse. Linear regression showing the 95% confidence bands of the best-fit line and the goodness of fit (R²). a.u., arbitrary units, MFI, mean fluorescence intensity. Similar results were obtained for s.c. fat (data not shown). (D)MYCT1 knockdown promotes lysosomal degradation activity. Control and MYCT1^KD cells were treated for 1 h with DQ-BSA. DQ-BSA fluorescence (gray/black) and staining for β-catenin (magenta) and DNA (blue). Lower panels show DQ-BSA. Scale bar, 20 µm. (E) Quantification of the percentage of DQ-BSA signal per cell. n = 4 independent experiments; 60–100 cells were analyzed per condition for each experiment; mean ± SD; unpaired t test, P = 0.0059 (*). (F)IFITM2/3 knockdown rescues RAB5⁺ endosome enlargement in MYCT1-deficient ECs. Staining of ECs for RAB5 (gray/black), VE-cadherin (magenta), and DAPI (blue). Lower panels show RAB5. Scale bar, 10 μm. (G) Quantification of RAB5⁺ vesicle volume in control, MYCT1^KD, IFITM2/3^KD, and MYCT1–IFITM2/3^KD cells. n = 3 independent experiments; 1,000–3,000 vesicles were analyzed per condition for each experiment; mean ± SD; one-way ANOVA with Tukey’s multiple comparisons test, P = 0.028 (*) for MYCT1 knockdown effect, P = 0.0060 (*) for comparison between MYCT1 and IFITM2/3 knockdowns, and P = 0.0070 (*) for rescue effect by IFITM2/3 simultaneous knockdown. (H)IFITM2/3 knockdown rescues mTORC1 hyperactivation in MYCT1-deficient ECs. Staining for p-S6 (gray), β-catenin (magenta), and DAPI (blue). Scale bar, 20 μm. (I) Quantification of mTORC1 activation in control, MYCT1^KD, IFITM2/3^KD, and MYCT1–IFITM2/3^KD cells. The percentage of p-S6⁺ cells was quantified in the indicated conditions. n = 4 independent experiments; 1,500–8,000 cells were analyzed per condition for each experiment; mean ± SD; one-way ANOVA with Tukey’s multiple comparisons test, P = 0.0195 (*) for MYCT1 knockdown effect and P = 0.0118 (*) for rescue effect by IFITM2/3 simultaneous knockdown. (J) MYCT1 ablation leads to accumulation of IFITM2/3, which drives enlargement of early endosomes, increases cargo uptake, and enhances lysosomal degradation. This process results in greater amino acid delivery, thereby activating mTORC1. These phenotypes are reversed upon simultaneous knockdown of IFITM2/3 and MYCT1. Icons used in A and J were created with BioRender.com and modified in Affinity. See also Fig. S5, H–Q.

IFITM2/3 promote shuttling of endocytic cargo to the lysosome (Spence et al., 2019). To establish the fate of extracellular cargo taken up in MYCT1-deficient cells, we used a DQ-BSA assay (Marwaha and Sharma, 2017), in which hydrolysis of DQ-BSA in lysosomal compartment leads to increased fluorescence signal. DQ-BSA fluorescence was significantly increased upon MYCT1 depletion, indicating that endocytic cargo is targeted toward degradation in lysosomes (Fig. 7, D and E). Collectively, these findings establish a novel role of MYCT1 in limiting endocytic trafficking and delivery to lysosomes through its interaction with IFITM2/3.

Finally, we investigated the role of the MYCT1–IFITM2/3 interactions in endosome organization and in the activation of mTORC1 pathway. The treatment with dynasore, an endocytosis inhibitor (Macia et al., 2006), effectively suppressed the mTORC1 hyperactivation observed in MYCT1-deficient cells (Fig. S5, L and M). Furthermore, simultaneous knockdown of RAB5 and MYCT1 reduced mTORC1 activation to levels comparable with those observed in control cells (Fig. S5, N and O). Crucially, simultaneous knockdown of IFITM2/3 and MYCT1 reversed the enlarged endosome phenotype (Fig. 7, F and G) and counteracted hyperactive mTORC1 signaling caused by MYCT1 depletion (Fig. 7, H and I; and Fig. S5 P and Q). Our findings reveal the existence of a new pathway whereby interaction of MYCT1 and IFITM2/3 limits endocytosis and futile mTORC1 signaling activation in ECs (Fig. 7 J).

Endothelial-specific activation of mTORC1 signaling recapitulates the adipose tissue phenotype

Building on our findings, we next asked whether sustained endothelial mTORC1 activation is sufficient to drive the adipose tissue phenotype observed in Myct1^ecKO mice. To address this, we generated endothelial-specific Tsc1 KO mice (Tsc1^ecKO), as TSC1 is a well-established negative regulator of mTORC1 (Huang and Manning, 2008) (Fig. 8 A). En face staining of the aorta for p-S6 confirmed robust mTORC1 activation in the endothelium of Tsc1^ecKO mice (Fig. 8, B and C).

Figure 8.

Endothelial-specific activation of mTORC1 signaling recapitulates adipose tissue phenotype of Myct1 ^ecKO mice. (A) Tsc1 ^ecKO mouse model. See Materials and methods for details. (B) Endothelial Tsc1 ablation activates sustained mTORC1 signaling in vivo. En face staining of aorta for p-S6 (gray) and VE-cadherin (magenta). Scale bar, 50 μm. (C) Quantification of mTORC1 activation in the aortic endothelium of wild-type and Tsc1^ecKO mice. The percentage of p-S6⁺ cells was quantified in n = 5 mice per genotype and conditions; 300–500 cells were analyzed per aorta; mean ± SD; Welch’s t test, P = 0.0121 (*). (D) Workflow of Tsc1^ecKO mouse analysis. (E) Wild-type and Tsc1^ecKO mice were analyzed before significant difference in body weight. Quantification of body weight at the start and end of experiment. n = 10–12 mice per genotype; mean ± SD; multiple unpaired t tests, P > 0.05 at start, P = 0.116 at end. (F) Quantification of fat pad weight to body weight ratio relative to wild-type mice. n = 9 mice per genotype; mean ± SD; multiple unpaired t tests, P = 0.001 (*) for interscapular WAT (IsWAT), P = 0.027 for interscapular BAT (IsBAT), P < 0.001 (*) for inguinal (Ing), P < 0.001 (*) for retroperitoneal (RP), P < 0.001 (*) for gonadal (Gon), and P < 0.001 (*) for mesenteric (Mes). (G)Tsc1 ablation reduces size of adipocytes. Retroperitoneal thick sections stained for Laminin α4 (gray) and Pecam1 (magenta). Scale bar, 50 μm. (H) Quantification of adipocyte size in wild-type and Tsc1^ecKO fat. n = 5 mice per genotype; mean ± SD; unpaired t test, P = 0.0056 (*). (I) Schematic view of the role of endothelial MYCT1–IFITM2/3 complexes in WAT homeostasis. MYCT1 interacts with IFITM2/3, limiting their function and allowing nutrient transport for energy storage. In the absence of MYCT1, IFITM2/3 accumulate in early endosomes and trigger continuous endolysosomal cargo degradation and hyperactivation of mTORC1 signaling in ECs, restricting energy storage in WAT. Icons used in A, D, and I were created with BioRender.com and modified in Affinity.

We focused our phenotypical analysis on mice 6 wk after endothelial Tsc1 deletion, a time point at which overall body weight was only modestly and nonsignificantly lower compared with wild-type littermates (Fig. 8, D and E). Despite the absence of a significant change in total body weight (Fig. 8 E), Tsc1^ecKO mice exhibited a striking limitation in WAT mass accumulation (Fig. 8 F). Quantitative analysis revealed that all major fat depots—including subcutaneous (interscapular and inguinal) and visceral (retroperitoneal, gonadal, and mesenteric) adipose tissues—were significantly smaller in Tsc1^ecKO mice compared with controls (Fig. 8 F), mirroring the phenotype observed in Myct1^ecKO mice. To further characterize this phenotype at the cellular level, we quantified adipocyte size in the retroperitoneal depot. Consistent with impaired adipose tissue expansion, Tsc1^ecKO mice displayed significantly smaller adipocytes compared with wild-type controls (Fig. 8, G and H). Collectively, these findings demonstrate that persistent endothelial mTORC1 activation, as modeled by Tsc1 deletion, is sufficient to selectively limit adipose tissue expansion and recapitulate key features of the Myct1^ecKO phenotype.

In summary, our data establish that the MYCT1–IFITM2/3 axis regulates endolysosomal trafficking and constrains endothelial mTORC1 activation, thereby preventing futile energy expenditure within the endothelium and ultimately enabling systemic energy storage in WAT (Fig. 8 I).

Discussion

ECs are a vertebrate innovation, having evolved ∼500 million years ago in conjunction with the development of closed circulatory systems (Monahan-Earley et al., 2013). This adaptation facilitated efficient blood flow and regulated nutrient exchanges, supporting the increased metabolic demands of complex vertebrate bodies and functions of specialized organ systems, such as the brain. Recent studies have increasingly positioned ECs as active regulators of organ development, regeneration, and metabolism (Augustin and Koh, 2024). Our findings advance the field by uncovering a new signaling checkpoint in controlling systemic energy storage.

WAT serves as the primary site for long-term energy storage and plays a central role in systemic metabolic regulation. ECs are increasingly recognized as active participants in WAT homeostasis. Previous studies have implicated VEGFA and VEGFB-mediated angiogenesis in improving WAT metabolic health (Sung et al., 2013; Robciuc et al., 2016) while other studies uncovered further modes of adipocyte-endothelial communication (Crewe et al., 2018; Monelli et al., 2022). However, these mechanisms primarily act through changes in vascular density or heterotypic cellular interactions. In contrast, our data position MYCT1 as a non-angiogenic endothelial regulator that influences adipocyte size and WAT expansion through regulation of IFITM2/3-dependent endolysosomal trafficking. We propose that a body-wide increase in endothelial endocytosis and mTORC1 activity elevates futile energy expenditure at the vascular barrier, producing a synergistic effect that leaves less energy available for storage, with WAT showing the most pronounced response under homeostatic conditions (Fig. 8 I).

MYCT1 was first identified as a target of the c-MYC transcription factor in myeloid cells and laryngeal cancer cells (Nesbit et al., 2000; Qiu et al., 2003; Fu et al., 2011). However, subsequent studies, including ours, revealed its specific and constitutive expression in ECs as well as in a subset of HSCs in both mice and humans (Kabir et al., 2021; Aguadé-Gorgorió et al., 2024) (Fig. 1 and Fig. S1). While MYCT1 loss in cultured HSCs led to increased endocytosis and cytokine signaling, compromising self-renewal and engraftment capacity through changes in environmental sensing (Aguadé-Gorgorió et al., 2024), our work focused on elucidating MYCT1 function in the endothelium. We demonstrate that MYCT1 plays a central role in restraining energy expenditure at the vascular barrier. In the absence of MYCT1, IFITM2/3 drive continuous endolysosomal cargo transport and degradation, leading to hyperactivation of mTORC1 signaling in ECs, ultimately limiting storage of the excess energy in WAT. The question remains why futile energy expenditure by ECs is not detected by whole-body metabolic assays. Our metabolic assays ruled out behavioral confounders, however they may lack the sensitivity to detect modest, endothelial-specific changes. We propose that small, sustained increases in energy use at the endothelial barrier sum up to a meaningful systemic effect. To test this directly, future studies should apply more sensitive approaches, such as high-resolution respirometry of isolated organs (Monelli et al., 2022). Further studies will be also required to directly test how increased endothelial endocytic activity contributes to systemic energy balance.

MYCT1 undergoes extensive posttranslational modifications, most notably phosphorylation (Fig. 1 E). The structure of MYCT1, with a short extracellular domain and a long, phosphorylated cytoplasmic tail, resembles that of transmembrane adaptor proteins such as LAT, NTAL, PAG, LIME, TRIM, SIT, and LAX (TRAPs), which organize membrane-proximal signaling in immune cells through tyrosine phosphorylation (Hořejší et al., 2004). Like TRAPs, MYCT1 may act as a signaling adaptor, linking serine/threonine receptor kinases to intracellular pathways. In addition to IFITM2/3, we identified the serine/threonine kinase TGFBR2 and the tyrosine kinase JAK1, a key player in both type I and type II IFN signaling pathways, as interactors of MYCT1 (Table S4). It is therefore tempting to speculate that TGF-β– and IFN-dependent phosphorylation may act as a fast switch for regulation of MYCT1 interaction with target proteins and downstream signaling.

Like MYCT1, IFITMs emerged in vertebrates (Hickford et al., 2012). While IFITMs are not endothelial-specific, ECs are uniquely characterized by the constitutive production of IFITM2/3 proteins (Sun et al., 2016). The mechanisms underlying this constitutive expression and the precise way in which MYCT1 prevents IFITM2/3 accumulation in endosomes remain unclear. In mouse embryonic fibroblasts, IFITM2/3 undergoes ubiquitination-mediated degradation via E3 ubiquitin ligase NEDD4 (Chesarino et al., 2015) and, in our study, mass spectrometry analysis identified members of the NEDD4 family (NEDD4, NEDD4L, and ITCH) as interactors of MYCT1 (Table S4). Thus, it will be interesting to investigate whether MYCT1 binding to IFITM2/3 facilitates its continuous turnover. IFITM2/3 are extensively regulated by posttranslational modifications, which influence their localization and activity (Yount et al., 2010; Chesarino et al., 2014). Our data indicate that, in ECs, IFITM2/3 localize to RAB5⁺ early endosomes. This raises the question of how this localization is mediated in ECs and whether IFITM2/3 posttranslational modifications differ from other cell types, potentially driving cell type–specific functions. Importantly, while IFITM2/3 are primarily known for antiviral defense, they have been recognized for other noncanonical roles, including fibrinogen endocytosis in platelets and placental syncytiotrophoblast formation (Buchrieser et al., 2019; Campbell et al., 2022; Lou et al., 2022). Our findings add to this emerging perspective by implicating IFITM2/3 in nutrient transport pathways, particularly in the regulation of endosomal architecture and mTORC1 activity in ECs.

Our findings in mouse tumor models contrast with the study by Kabir et al. (2021), which reported that Myct1 deficiency reduced tumor angiogenesis, enhanced high endothelial venule formation, and promoted anti-tumor immunity, inhibiting tumor progression. The discrepancies between our results and those by Kabir et al. may be attributed to several factors, including differences in genetic backgrounds, housing conditions, and use of germline versus inducible KO models. Alternatively, the absence of MYCT1 leads to increased endothelial nutrient and energy consumption, depriving tumor cells of essential nutrients required for rapid growth. Rapidly expanding tissues such as WAT and tumors may be particularly sensitive to nutrient availability, whereas under starvation conditions, other organs could become affected if nutrients are preferentially utilized at the endothelial barrier—an important future direction.

In summary, we report a novel function of MYCT1 as a non-angiogenic endothelial regulator of adipocyte size and WAT expansion via IFITM2/3-dependent endolysosomal trafficking. Disruption of MYCT1–IFITM2/3 interaction represents a novel therapeutic opportunity to modulate WAT growth and function, potentially offering new strategies to combat metabolic diseases such as obesity, insulin resistance, and related disorders by targeting adipose ECs rather than approaches focused solely on adipocytes.

Our work opens the possibility of additional roles for MYCT1–IFITM2/3 complexes in the vasculature and HSCs. For instance, IFITM3 restricts viral infections by enveloped viruses in epithelial cells and fibroblasts (Feeley et al., 2011; Jia et al., 2012; Spence et al., 2019; Majdoul and Compton, 2022); however, the role of MYCT1–IFITM2/3 complex in EC infection remains to be investigated. Furthermore, IFITM3 increase by inflammation and aging has been implicated in Alzheimer’s disease progression through an increase in γ-secretase activity, and amyloid-β production (Hur et al., 2020; Feng et al., 2025). Since γ-secretase is critical for Notch signaling and Notch signaling is essential for both endothelial and hematopoietic lineage decisions, it will be fascinating to investigate MYCT1–IFITM2/3 interplay in these processes, as well as its potential dysregulation in aging and disease.

Materials and methods

Key resources table

Key resources are listed in Table 1.

Table 1.

Key resources table

Reagent or resource	Source	Identifier
Antibodies
Rabbit polyclonal anti-MYCT1	Abcam	Cat# ab139945, RRID:AB_2861408
Mouse monoclonal anti-IFITM2	Thermo Fisher Scientific	Cat# CF811628
Rabbit polyclonal anti-IFITM3	Proteintech	Cat# 11714-1-AP, RRID:AB_2295684
Goat polyclonal anti VE-cadherin	R&D Systems	Cat# AF1002, RRID:AB_2077789
Mouse anti-β-catenin	BD Biosciences	Cat# 610154, RRID:AB_397555
Goat polyclonal anti-VEGFR2	R&D Systems	Cat# AF644, RRID:AB_355500
Rabbit polyclonal anti-β-catenin	Sigma-Aldrich	Cat# 06-734, RRID:AB_310231
Rabbit polyclonal anti-GFP	Abcam	Cat# ab290, RRID:AB_303395
Mouse monoclonal anti-Rab5	Cell Signaling Technology	Cat# 46449, RRID:AB_2799303
Rabbit monoclonal anti-Rab5	Cell Signaling Technology	Cat# 3547, RRID:AB_2300649
Rabbit monoclonal anti-Rab7	Abcam	Cat# ab137029, RRID:AB_2629474
Mouse monoclonal anti-Lamp1	Santa Cruz Biotechnology	Cat# sc-20011, RRID:AB_626853
Rabbit monoclonal anti-phospho-S6 ribosomal protein (Ser240/244)	Cell Signaling Technology	Cat# 5364, RRID:AB_10694233
Mouse monoclonal anti-S6 ribosomal protein	Cell Signaling Technology	Cat# 2317, RRID:AB_2238583
Rabbit anti-p70/S6K (Thr389)	Cell Signaling Technology	Cat# 9234, RRID:AB_2269803
Rabbit anti-p70/S6K	Cell Signaling Technology	Cat# 9202, RRID: AB_2943679
Rabbit polyclonal anti-Mat2a	Proteintech	Cat No. 55309-1-AP, RRID:AB_2881303
Rabbit monoclonal anti-phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204)	Cell Signaling Technology	Cat# 4376, RRID:AB_331772
Rabbit monoclonal anti-p44/42 MAPK (Erk1/2)	Cell Signaling Technology	Cat# 4695, RRID:AB_390779
Rabbit monoclonal anti-phospho-Akt (Thr308)	Cell Signaling Technology	Cat# 13038, RRID:AB_2629447
Rabbit monoclonal anti-phospho-Akt (Ser473)	Cell Signaling Technology	Cat# 4060, RRID:AB_2315049
Rabbit polyclonal anti-AKT	Cell Signaling Technology	Cat# 9272, RRID:AB_329827
Rabbit polyclonal anti-GAPDH	Sigma-Aldrich	Cat# G9545, RRID:AB_796208
Mouse monoclonal anti-Vinculin	Sigma-Aldrich	Cat# V9131, RRID:AB_477629
Mouse monoclonal anti-Pecam1	Thermo Fisher Scientific	Cat# MA5-13188, RRID:AB_10982120
Rat anti-Pecam1	BD Biosciences	Cat# 553370, RRID:AB_394816
Rat monoclonal anti-endomucin	Thermo Fisher Scientific	Cat# 14-5851-81, RRID:AB_891529
Rabbit polyclonal anti-Pecam1	Abcam	Cat# ab28364, RRID:AB_726362
Rabbit monoclonal anti-Erg	Abcam	Cat# ab92513, RRID:AB_2630401
Goat polyclonal anti-FABP4	R&D Systems	Cat# AF1443, RRID:AB_2102444
Rabbit monoclonal anti-Perilipin	Cell Signaling Technology	Cat# 9349, RRID:AB_10829911
Rat monoclonal anti-Pecam1 (BV421 conjugated)	BioLegend	Cat# 102423, RRID:AB_2562186
Rabbit anti-H3K27ac	Abcam	Cat# ab4729, RRID:AB_2118291
Rabbit anti-GAPDH	Sigma-Aldrich	Cat# G9545, RRID:AB_796208
Mouse anti-Vimentin	Dako	Cat# M0725, RRID:AB_10013485
CD45-APC/Cy7, clone 30-F11	BioLegend	Cat# 561037, RRID:AB_10563075
CD45-PE, clone 30-F11	BioLegend	Cat# 103105, RRID:AB_312970
CD31-BV421, clone 390	BD Pharmingen	Cat# 102423, RRID:AB_2562186
CD31-PE, clone 390	eBioscience	Cat# 12-0311-81
Sca1-PE/Cy7, clone D7	BioLegend	Cat# 108113, RRID:AB_493597
Bacterial and virus strains
Ad-MYCT1 and Ad-GFP adenoviruses	This study	​
Biological samples
Human brain samples	Hôpitaux universitaires de Genève (HUG) (Geneva, Switzerland)	CCER 2022-02109
Human fat samples	Lausanne University Hospital (CHUV) (Lausanne, Switzerland)	CER-VD 38/15, PB_2016-01185
Human normal and cancer samples	CHUV (Lausanne, Switzerland)	CER-VD 464/11
Chemicals, peptides, and recombinant proteins
Tamoxifen	Sigma-Aldrich	T5648
Cremophor EL	Sigma-Aldrich	C5135
PFA	Sigma-Aldrich	158127
EBM2 medium	Lonza	CC-3156
EGM2 SingleQuots	Lonza	CC-4176
Opti-MEM medium	Gibco	51985-026
PBS containing MgCl2 and CaCl2	Sigma-Aldrich	D8662
Lipofectamine RNAiMAX	Thermo Fisher Scientific	13778-075
OPP	Jena Bioscience	NU-931-05
Vitamin solution	Sigma-Aldrich	R7256
Amino acid solution	Sigma-Aldrich	R7131
GlutaMAX	Thermo Fisher Scientific	35050038
Rapamycin	LC Laboratories	R-5000
PhosSTOP	Roche	04906837001
Complete mini, EDTA-free Protease inhibitor	Roche	11836170001
Histodenz	Sigma-Aldrich	D2158
Liberase TL	Sigma-Aldrich	05401020001
HBSS	Gibco	14175-053
Red blood cell lysis	Thermo Fisher Scientific	00-4333-57
Fibronectin	Roche	10838039001
Dynasore	Sigma-Aldrich	324410
Cycloheximide	Enzo Life Sciences	ALX-380-269
Fluorescein isothiocyanate–dextran (10 kDa)	Sigma-Aldrich	FD10S
DQ Green BSA	Thermo Fisher Scientific	D12050
Evans blue	Sigma-Aldrich	E2129
DMEM	Sigma-Aldrich	D5030
Insulin	Sigma-Aldrich	I9278
Dexamethasone	Sigma-Aldrich	D4902
3-Isobutyl-1-methylxanthine (IBMX)	Enzo Life Sciences	BML-PD140
Indomethacin	Enzo Life Sciences	ALX-270-086
Recombinant PNGase F	Provided by P. Schneider (Department of Immunobiology [DIB], University of Lausanne [Unil], Lausanne, Switzerland)	704S; NEB
Fluoromount-G mounting medium	Thermo Fisher Scientific	00-4959-52
Aquatex	Merck	1085620050
Heparin	Sigma-Aldrich	H3149
Tamoxifen diet	Harlan	TD.55125
HFD	Research Diet	D12492
Control diet	Research Diet	D12450B
Anti-rat IgG magnetic Dynabeads	Thermo Fisher Scientific	11035
7-AAD	eBioscience	00-6993-50
Critical commercial assays
Duolink in situ Detection Reagents Orange	Sigma-Aldrich	DUO92007
Duolink in situ PLA probe anti-mouse PLUS	Sigma-Aldrich	DUO92001
Duolink in situ PLA probe anti-rabbit MINUS	Sigma-Aldrich	DUO92005
Duolink in situ PLA probe anti-goat PLUS	Sigma-Aldrich	DUO92003
RNeasy Plus mini kit	Qiagen	74134
Transcriptor First Strand cDNA Synthesis kit	Roche	04 896 866 001
SensiFast Sybr Lo-Rox Mix	Bioline	BIO-94020
SuperSignal West Femto Maximum Sensitivity Substrate	Thermo Fisher Scientific	34095
Pierce BCA Protein assay kit	Thermo Fisher Scientific	23227
Lipofectamine RNAiMAX	Thermo Fisher Scientific	3778075
Protein G magnetic Dynabeads	Thermo Fisher Scientific	10003D
Tyramide SuperBoost Kits with Alexa Fluor Tyramides	Thermo Fisher Scientific	B40931
Deposited data
GEO accession number GSE294341	This study	https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE294341
scRNA-seq data of human ECs from several organs	(Barnett et al., 2024)	https://www.vascularcellatlas.org/
Experimental models: Cell lines
Primary HIECs	Isolated in the lab	Anonymous donors, CHUV, Lausanne
Primary human pulmonary and adipose tissue microvascular ECs	Innoprot, PromoCell	P10552, P10781 C-12281
Primary HUVECs	Lonza	C2519A
Human embryonic kidney 293T cells	ATCC	CRL-1573
SW480 cells	ATCC	CCL-228
LLC cells	ATCC	CRL-1642
Experimental models: Organisms/strains
Myct1 flox/+ mouse	This study	InGenious Targeting Laboratory Inc.
Pdgfb-Cre ERT2 mouse	(Claxton et al., 2008)	Provided by M. Fruttiger (UCL, London, UK)
Tsc1 flox/+ mouse	(Uhlmann et al., 2002)	Provided by O. Dormond (UNIL, Lausanne, Switzerland)
Oligonucleotides
Primers qPCR
18 s (human/mouse) forward	Eurofins	5′-AGG​AAT​TCC​CAG​TAA​GTG​CG-3′
18 s (human/mouse) reverse	Eurofins	5′-GCC​TCA​CTA​AAC​CAT​CCA​A-3′
MYCT1 (human/mouse) forward	Eurofins	5′-CTT​TCC​ATC​CCT​TTC​TGC​AA-3′
MYCT1 (human/mouse) reverse	Eurofins	5′-GTG​CTG​ATG​GTG​GGA​GAG​AT-3′
IFITM2 (human) forward	Eurofins	5′-GTC​ACC​ATG​AAC​CAC​ATT​GTG​CAA​AC-3′
IFITM2 (human) reverse	Eurofins	5′-CCC​CCA​GCA​TAG​CCA​CTT​CC-3′
IFITM3 (human) forward	Eurofins	5′-ACC​ATG​AAT​CAC​ACT​GTC​CAA​ACC​TT-3′
IFITM3 (human) reverse	Eurofins	5′-CCA​GCA​CAG​CCA​CCT​CG-3′
Pecam1 (mouse) forward	Eurofins	5′-AAC​AGA​AAC​CCG​TGG​AGA​TG-3′
Pecam1 (mouse) reverse	Eurofins	5′-GTC​TCT​GTG​GCT​CTC​GTT​CC-3′
Cdh5 (mouse) forward	Eurofins	5′-CCC​ACG​AAG​TCC​CTG​GAC​TAT​G-3′
Cdh5 (mouse) reverse	Eurofins	5′-GGT​CTG​TGG​CCT​CAA​TGT​AGA​ATG-3′
Ucp1 (mouse) forward	Eurofins	5′-CTG​CCA​GGA​CAG​TAC​CCA​AG-3′
Ucp1 (mouse) reverse	Eurofins	5′-GCC​ACA​AAC​CCT​TTG​AAA​AA-3′
Cpt1b (mouse) forward	Eurofins	5′-CCC​ATG​TGC​TCC​TAC​CAG​AT-3′
Cpt1b (mouse) reverse	Eurofins	5′-CCT​TGA​AGA​AGC​GAC​CTT​TG-3′
Prdm16 (mouse) forward	Eurofins	5′-CAG​CAC​GGT​GAA​GCC​ATT​C-3′
Prdm16 (mouse) reverse	Eurofins	5′-GCG​TGC​ATC​CGC​TTG​TG-3′
Cebpb (mouse) forward	Eurofins	5′-CGC​AAC​CTG​GAG​ACG​CAG​CA-3′
Cebpb (mouse) reverse	Eurofins	5′-GGC​TCG​GGC​AGC​TGC​TTG​AA-3′
Ppara (mouse) forward	Eurofins	5′-GAC​AAG​GCC​TCA​GGG​TAC​CA-3′
Ppara (mouse) reverse	Eurofins	5′-GCC​GAA​TAG​TTC​GCC​GAA​A-3′
siRNA
All Stars negative control	Qiagen	1027281
MYCT1 (human) #6 (A)	Qiagen	SI04203052
MYCT1 (human) #8 (B)	Qiagen	SI04281760
IFITM2/3 (human) #5	Qiagen	SI04179721
RAB5A (human) #10	Qiagen	SI03111115
Software and algorithms
GraphPad Prism 10	GraphPad Software	https://www.graphpad.com/
Fiji version 1.54f	(Schindelin et al., 2012)	https://fiji.sc
CellProfiler version 4.2.4	CellProfiler	https://cellprofiler.org
Affinity Photo 2	Affinity	https://affinity.serif.com
Affinity Publisher 2	Affinity	https://affinity.serif.com
Affinity Designer 2	Affinity	https://affinity.serif.com
BioRender	BioRender	https://biorender.com
FlowJo version 10	BD Biosciences	https://www.flowjo.com/
Cellpose version 2.1.1	(Stringer et al., 2021)	https://github.com/mouseland/cellpose
Napari version 0.4.17	(Sofroniew et al., 2025)	https://napari.org/0.4.17/index.html
Scikit-image version 0.19.3	(van der Walt et al., 2014)	https://scikit-image.org/
Python version 3.9.7	Python Software Foundation	https://www.python.org/
PantherDB version 19.0	PantherDB	https://pantherdb.org/
Cell Ranger version 6.0.0	10x Genomics	https://www.10xgenomics.com/support/software/cell-ranger/downloads
R versions 4.3.1 and 4.1.2	The R Project for Statistical Computing	https://www.r-project.org/
Seurat package versions 5.0.1 and 4.3.0	(Hao et al., 2024)	https://cran.r-project.org/web/packages/Seurat/index.html
clusterProfiler package version 4.2.2	(Xu et al., 2024)	https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html
msigdbr package version 7.5.1	(Dolgalev, 2025)	https://cran.r-project.org/web/packages/msigdbr/index.html

Animal models

All animal experiments were approved by the Animal Ethics Committee of the Canton of Vaud, Switzerland. Mice were provided with water and food ad libitum and were on a 12-h light/dark cycle. Experiments were performed with age-, sex-, and litter-matched cohorts; no significant sex-dependent differences were found in the experiments reported. Myct1^fl/+ mice were generated by InGenious Targeting Laboratory. Myct1^fl/+ mice were backcrossed with C57BL/6J mice over 10 generations and further crossed with Pdgfb-Cre^ERT2 mice (Claxton et al., 2008) to obtain Myct1^fl/fl;Pdgfb-Cre^ERT2 (Myct1^ecKO). Myct1^fl/fl were used as wild-type controls. For early postnatal studies, Myct1 deletion was induced by s.c. injection of 40 µg tamoxifen (T5648-5G; Sigma-Aldrich) in 10% Cremophor EL (C5135; Sigma-Aldrich) at birth (P0) and postnatal day 1 (P1). For studies in adult mice, unless indicated otherwise, mice received three intraperitoneal 50 mg/kg injections of tamoxifen. Alternatively, adult mice were fed with tamoxifen-containing diet (TD.55125; Harlan) for 2 wk prior to the start of the experiment. The absence of adipose phenotype was validated in Pdgfb-Cre^ERT2 mice treated with tamoxifen and in untreated Myct1^fl/fl;Pdgfb-Cre^ERT2 mice.

Tsc1^{fl/ +} mice (Uhlmann et al., 2002) were crossed with Pdgfb-Cre^ERT2 mice (Claxton et al., 2008) to obtain Tsc1^fl/fl;Pdgfb-Cre^ERT2 (Tsc1^ecKO). Tsc1 deletion was induced as described for Myct1.

Cell culture

Cells were cultured in a sterile incubator at 37°C and 5% CO₂. Human lung ECs (HPMECs, PromoCell and Innoprot), human adipose tissue ECs (HAdMECs, Innoprot), human intestinal ECs (HIECs, Innoprot) (Haraldsen et al., 1995), and human umbilical vein ECs (HUVECs; Lonza) were cultured in EBM2 medium (CC-3156; Lonza) supplemented with EGM2 SingleQuots supplements (CC-4176; Lonza) in fibronectin-coated (5 μg/cm²; 10838039001; Roche) flasks. 293T, SW480, and LLC cells (ATCC) were cultured in complete DMEM (D5030; Sigma-Aldrich) supplemented with 10% FBS, and 1% penicillin/streptomycin.

Human samples

Brain samples from patients with epilepsy (n = 2) were collected with patient-written consent and with approval of the Ethics Committee of the Canton of Geneva (CCER 2022-02109). Upon collection, samples were frozen in liquid nitrogen without further processing. Mammary fat samples from patients undergoing reduction mammoplasty were collected with patient-written consent and with approval of the Ethics Committee of the Canton of Vaud (CER-VD 38/15, PB_2016-01185). Upon collection, samples were fixed with 4% paraformaldehyde (PFA) and further embedded in paraffin.

Cell transfection

For knockdown experiments, cells were transfected at 70% confluency with 40 nM siRNA using Lipofectamine RNAiMAX (3778075; Thermo Fisher Scientific). When cells were co-transfected with multiple siRNAs, each siRNA was added at 40 nM, and the amount of control siRNA was adjusted accordingly. siRNA list is provided in the key resources table (Table 1). If not indicated otherwise, ECs were analyzed 48 h after transfection.

Cell treatments

For the analysis of protein synthesis, 48 h after transfection, OPP (10 μM, Jena Bioscience) was added to ECs for 50 min, cells were fixed with 4% PFA containing 1 mM CaCl₂ and 0.5 mM MgCl₂, and OPP was revealed by click-chemistry reaction with Alexa Fluor 555–conjugated azide (C10338; Thermo Fisher Scientific).

For amino acid stimulation, 32 h posttransfection cells were serum- and growth factor–starved overnight. 48 h after transfection, cells were starved for 1 h in PBS supplemented with vitamin solution (R7256; Sigma-Aldrich) followed by 30-min stimulation with 1× amino acid solution (R7131; Sigma-Aldrich) supplemented with 1× GlutaMAX (35050038; Thermo Fisher Scientific) or PBS supplemented with vitamin solution as a control. Cells were fixed with 4% PFA containing 1 mM CaCl₂ and 0.5 mM MgCl₂ and further processed for staining.

When applicable, cells were treated for indicated times with 10 nM rapamycin (LC Laboratories), 25 μM Dynasore (324410-10MG; Merck), or co-incubated with 1 mg/ml 10-kDa FITC-dextran (FD10S; Sigma-Aldrich) or 10 μg/ml DQ Green BSA (D12050; Thermo Fisher Scientific).

Western blot and immunoprecipitation

Cells were lysed on ice in RIPA lysis buffer (150 mM NaCl, 1 mM EDTA, 50 mM Tris-HCl, pH 7, 4, 1% sodium deoxycholate, 1% Triton X-100, 0.1% SDS, PhosSTOP [04906837001; Roche], and complete mini protease inhibitor [11836170001; Roche]). Lysates were cleared by centrifugation, and protein concentration was measured using BCA kit (23227; Thermo). 10-20 μg of lysate was run on Mini-PROTEAN TGX Precast gels (10%, 12% or 4–15%) and transferred on Immobilon-P membrane (IPVH00010; Millipore). Membranes were blocked with 5% BSA for 1 h at room temperature, incubated overnight at 4°C with primary antibodies, and incubated for 1 h at room temperature with HRP-conjugated secondary antibodies (Dako). Blots were developed using Super Signal West Femto Maximum Sensitivity Substrate (34095; Thermo Fisher Scientific) with a CCD camera (Fusion Fx, Vilber Lourmat). Antibodies are listed in the key resources table (Table 1).

Subcellular fractions of HIECs and HPMECs were prepared using the Qproteome Cell Compartment Kit (37502; Qiagen). Confluent cells were washed with ice-cold PBS and sequentially extracted into cytosolic, membrane, nuclear, and cytoskeletal fractions following the manufacturer’s protocol. Fractionation was validated by immunoblotting by detecting compartment-specific proteins, as indicated in the figure.

For protein glycosylation analysis, lysates were incubated with glycoprotein denaturing buffer for 5 min at 92°C, put on ice and incubated with or without 20 U PNGase F (704S; NEB) for 1 h, and analyzed by immunoblotting.

For immunoprecipitation, cells were lysed on ice in buffer containing Tris-HCl, NaCl, Triton X-100, EDTA, and phosphatase and protease inhibitors. 300–500 μg of protein was precleared and incubated overnight at 4°C with anti-MYCT1 or control antibody. The next day, protein G Dynabeads (10003D; Thermo Fisher Scientific) were added for 1 h at 4°C with gentle rocking. After washing, samples in SDS-PAGE loading buffer were boiled at 95°C for 10 min, separated by SDS-PAGE, and processed for western blotting or mass spectrometry.

Mass spectrometry

HPMECs and SW480 were starved for 4 h in basal EBM2 (CC-3156; Lonza) before transduction with adenoviruses encoding MYCT1 (Ad-MYCT1) or GFP (Ad-GFP) for 4 h at 37°C. The infection medium was replaced with complete EBM2, and cells were incubated for additional 48 h. Cells were lysed and used for immunoprecipitation as described above. The immunoprecipitates were resolved by SDS-PAGE, stained overnight with Candiano colloidal Coomassie blue, and analyzed by liquid chromatography–tandem mass spectrometry after trypsin or chymotrypsin digestion. Generated spectra were matched to known protein sequences using MASCOT. Proteins identified in MYCT1 pull-downs from HPMECs across two independent experiments, but absent in MYCT1 pull-down from SW480 cells, were considered significant. Relative protein abundances were determined using normalized spectral abundance factors, and GO term enrichment analysis was performed using PantherDB.

Immunostaining of cells

Cells were seeded at confluency on fibronectin-coated coverslips, washed with PBS containing MgCl₂ and CaCl₂ (D8662; Sigma-Aldrich), and fixed for 10 min in 4% PFA with 1 mM CaCl₂ and 0.5 mM MgCl₂. Cells were washed with PBS containing MgCl₂ and CaCl₂, blocked for 30 min (PBS with 0.3% BSA, 5% donkey serum, 0.3% Triton X-100, and 0.1% NaN₃), then incubated with primary antibodies in blocking buffer for 1 h at room temperature, followed by Alexa Fluor–conjugated secondary antibodies for 1 h at room temperature, and mounted with Fluoromount-G mounting medium (00-4959-52; Thermo Fisher Scientific).

Immunostaining of paraffin sections

5–10-μm sections were deparaffinized and subjected to heat-induced epitope retrieval (high/low pH retrieval solution, Dako). They were blocked for 30 min with blocking buffer (PBS containing 0.3% BSA, 5% donkey serum, 0.3% Triton X-100, and 0.1% NaN₃), then incubated with primary antibodies overnight at 4°C, followed by secondary antibody incubation for 1 h at room temperature. Slides were mounted with Fluoromount-G mounting medium (00-4959-52; Thermo Fisher Scientific).

For chromogenic detection, sections were quenched with 3% H₂O₂, blocked, and incubated overnight with primary antibodies. Signal was developed using biotinylated secondary antibodies, HRP-conjugated streptavidin, and amplified with Tyramide SuperBoost (B40931; Thermo Fisher Scientific), followed by 3,3′-diaminobenzidine staining. H&E or Masson’s trichrome sections were mounted in Aquatex (1085620050; Merck).

Immunostaining of cryosections

10–20-μm thick sections were thawed for 30 min at room temperature, fixed for 5 min (pre-fixed tissue) or 30 min (unfixed tissue, e.g., brain) in 4% PFA at room temperature. The slides were blocked for 30 min with blocking buffer (PBS containing 0.3% BSA, 5% donkey serum, 0.3% Triton X-100, and 0.1% NaN₃) and incubated with primary antibodies overnight at 4°C, followed by secondary antibody incubation for 1 h at room temperature. Slides were mounted with Fluoromount-G mounting medium (00-4959-52; Thermo Fisher Scientific).

Proximity ligation assay (PLA)

Fixed and permeabilized cells or freshly cut cryosections were stained as per the manufacturer’s protocol (DUO92007; Sigma-Aldrich) using Duolink in situ PLA probe anti-mouse PLUS (DUO92001), anti-rabbit MINUS (DUO92005), and anti-goat PLUS (DUO92003). The PLA was combined with immunofluorescent staining to co-stain cell–cell junctions (cells) or vessels (sections).

Quantitative real-time PCR (RT-qPCR)

Cells or tissues were lysed in RLT buffer, and RNA was isolated using the RNeasy Plus Mini Kit (74134; Qiagen). 500 ng of total RNA was reverse transcribed into cDNA for qPCR using the Transcriptor First Strand cDNA Synthesis kit (04 896 866 001; Roche). For RT-qPCR, Quant Studio 3 (Applied Biosystems) machine with SensiFAST SYBR Lo-ROX Kit (BIO-94020; Bioline) was used. Data were normalized to 18S and analyzed using the comparative Ct (ΔΔCt) method. Sequences of PCR primers are provided in the key resources table (Table 1).

Production of viruses

MYCT1 cDNA encoding 187 aa of human MYCT1 (UniProt, D6Q1S4) was subcloned into pAd-CMV to produce a C-terminal V5-tagged construct. Replication-deficient adenoviruses, encoding V5-tagged MYCT1 or GFP, were produced as described previously (Ivanov et al., 2013).

Mouse tissue collection and whole-mount immunostaining

Retina

P5 and P7 pups were sacrificed by decapitation; eyes were dissected and fixed in 4% PFA for 4 h at 4°C. Dissected retinas were blocked (PBS with 0.5% BSA, 5% donkey serum, 0.3% Triton X-100, and 0.1% NaN₃) for 6 h at 4°C, incubated with primary and secondary antibodies (each overnight at 4°C), and mounted with Fluoromount-G (00-4959-52; Thermo Fisher Scientific).

Aorta

Anaesthetized mice were intracardially perfused with cold heparinized PBS (with CaCl₂/MgCl₂, 1 kU/ml heparin, and anti-phosphatases), followed by 4% PFA (1 mM CaCl₂ and 0.5 mM MgCl₂) and PBS. The thoracic aorta was dissected, opened, pinned endothelium side up and fixed on ice for 1 h in 4% PFA. For en face staining aortas were blocked for 1 h at room temperature, incubated with primary antibodies overnight at 4°C, then Alexa Fluor–conjugated secondaries for 2 h at room temperature, and mounted with Fluoromount-G (00-4959-02; Thermo Fisher Scientific).

Fat

Following intracardiac perfusion as described above, mesenteric fat was collected and fixed overnight at 4°C. Samples were blocked (PBS with 0.5% BSA, 5% donkey serum, 0.3% Triton X-100, and 0.1% NaN₃) for 6 h at 4°C, incubated with primary and secondary antibodies (each overnight at 4°C), and cleared in Histodenz (D2158; Sigma-Aldrich) before mounting.

Metabolic studies

HFD challenge

For diet-induced obesity studies, 8-wk-old littermates were injected with tamoxifen and fed a HFD (60 kcal% fat, D12492; Research Diet) or control diet (10 kcal% fat, D12450B; Research Diet) for 10 wk.

Metabolic cages

For metabolic studies, 10-wk-old littermates were injected with tamoxifen and fed a HFD for 2 wk. To measure metabolic parameters, the Comprehensive Lab Animal Monitoring System (Columbus instruments) was used. Each cage was an indirect open circuit calorimeter (Oxymax) that provides measures of oxygen consumption (VO₂) and carbon dioxide production (VCO₂). The RER was calculated as the respiratory quotient VCO₂/VO₂). Heat production was estimated from oxygen consumption and from the RER. The system also concurrently measured food and water consumption and locomotor activity. Ambulatory activity was monitored using an intra-Mot activity sensor system. Mice were acclimated to the metabolic cages for 48 h prior to the experiment. Data were collected in consecutive 24-h periods with a 12:12-light:dark cycle at normal ambient temperature of 23 ± 1°C for 2 days and at thermoneutrality (30°C) for an extra day. Body composition was evaluated at the end of the experiment in live, conscious animals by quantitative nuclear magnetic resonance spectroscopy (Echo Medical Systems). Core body temperature was monitored using a rectal thermometer.

Blood metabolic parameters

Blood samples were collected from mice via submandibular vein. Plasma was separated by centrifugation and stored at −80°C until analysis. Quantification of glucose, triglycerides, free fatty acids, and cholesterol was performed using standardized enzymatic assays at the university’s metabolic facility (University of Lausanne).

Oral lipid tolerance test

Mice were fasted for 6 h with free access to water prior to the oral lipid tolerance test. Following the fasting period, each mouse received an oral gavage of olive oil at a dose of 10 μl/g. Blood samples were collected from the tail vein before fasting at baseline (0) and at subsequent time points (1, 2, 3, and 4 h) after gavage. Plasma was processed, and plasma triglyceride concentrations were measured as described above.

Mouse CT scan

Whole-body CT scans were acquired using an X-RAD 225Cx irradiator (Precision X-ray). Mice were anesthetized with 2% isoflurane and positioned transversely in the animal holder. A 2D pre-scan confirmed alignment and defined the region of interest (ROI). Scans were performed at 40 kVp and 3 mA, displaying axial, sagittal, and 23 coronal slices. Images were analyzed in OsiriX (Rosset et al., 2004). Adipose tissue volumes were measured by adjusting levels (WL -65, WW 704), delineating the region across slices based on anatomical landmarks, and reconstructing the 3D volume to determine tissue volume and average density in Hounsfield units.

SVF isolation for in vitro adipogenesis assay

Inguinal (ING) and retroperitoneal (RP) adipose tissue depots were dissected from 4–7-wk-old wild-type or Myct1^ecKO mice (1–4 wk after tamoxifen). Tissue was minced with scissors, transferred into 8% Liberase TL (05401020001; Roche) in HBSS buffer (14175-053; Gibco), and incubated for 30 min (RP) or 50 min (ING) at 37°C under agitation. The digested tissue was passed through a syringe with 18-G needle, followed by a 5-min centrifugation at 1,500 rpm at 4°C, and floating adipocytes were aspirated. The cell suspension was filtered through a 100-μm strainer, and red blood cells were lysed (00-4333-57; Thermo Fisher Scientific) for 2 min, followed by two PBS washes and filtration through a 40-μm strainer. To deplete immune cells, the SVF was incubated with rat anti-CD45 antibody (10 min, 4°C), then with anti-rat IgG Dynabeads (11035; Thermo Fisher Scientific) for 30 min. The CD45^neg supernatant was cultured in DMEM with 10% FBS and 1% penicillin/streptomycin.

For adipogenic differentiation, SVF cells were seeded on glass coverslips at confluency and exposed to adipogenic cocktail consisting of 1 μM dexamethasone (D9184; Sigma-Aldrich), 500 μM IBMX (BML-PD140; ENZO), 125 nM indomethacin (ALX-270-086; Enzo), and 5 μg/ml insulin (I9278; Sigma-Aldrich) in complete DMEM. Medium was changed after 48 h to 1 μg/ml insulin in complete DMEM, and cells were analyzed 48 h later by immunofluorescent staining.

Fecal lipids

For analysis of fecal lipids, mice were housed in individual cages for 3 days, during which a minimum of 1 g of feces were collected per mouse. Feces were powdered and lipids extracted using chloroform:methanol (2:1) extraction. Extracts were left to dry completely, then weighed to determine lipid mass.

Vascular permeability

Mice were injected via tail vein with 100 μl of 1% Evans blue in PBS, sacrificed after 30 min, and perfused with ice-cold PBS, and organs were collected. Tissues were incubated in formamide (5 μl/g) for 72 h, then vortexed, centrifuged, and supernatant absorbance measured at 610–650 nm.

Tumor model

Following 2 wk of tamoxifen diet, 12–14-wk-old Myct1^ecko or control mice were injected in the right flank with 5 × 10⁵ LLC cells. Tumors were measured daily from day 6 after implantation with a caliper (#30087-00; Fine Science), and volume was calculated as length × width². Mice were sacrificed either at day 15 or when tumors reached 1 cm³. Mice were deeply anesthetized and intracardially perfused with PBS and then 4% PFA. Tumors were incubated in 4% PFA overnight at 4°C, then overnight in 30% sucrose and embedded in OCT.

Flow cytometry analysis

Single-cell suspension of SVF from WAT was generated as described above. For surface staining, cells were blocked with anti-CD16/32 antibody and subsequently stained for 40 min at 4°C using CD45-APC/Cy7 (clone 30-F11; BioLegend), CD31-PE (clone 390; eBioscience), and Sca1-PE/Cy7 (clone D7; BioLegend) conjugated antibodies. Measurements were performed on LSRFortessa flow cytometer (BD), and data were analyzed using FlowJo.

For labeled protein uptake in vivo, mouse plasma was labeled with ATTO 647N NHS ester as described (Yang et al., 2020). Labeled proteins (1.5–3.5 mg) were injected retro-orbitally, and mice were sacrificed 30 min after injection. The amount of delivered ATTO-labeled proteins was determined by fluorescence quantification of blood plasma samples collected from the tail vein immediately (<1 min) after the injection, using ATTO-labeled plasma as standards. After PBS perfusion, fat pads (inguinal and gonadal) and colon were enzymatically digested to obtain single-cell suspensions, stained with 7-AAD to exclude nonviable cells and CD45-PE– (clone 30-F11; BioLegend) and CD31-BV421– (clone 390; BioLegend) conjugated antibodies, and analyzed by flow cytometry to assess the level of ATTO-647 uptake in EC populations. Samples were acquired in LSR-IIb cytometer from BD Biosciences, and data were analyzed using FlowJo.

Reanalysis of published scRNA-seq data

Raw counts of human global and vascular cells from several organs were extracted from the h5ad file (Barnett et al., 2024) available at https://www.vascularcellatlas.org and imported into R (version 4.3.1). The Seurat package (version 5.0.1) (Hao et al., 2024) was used to convert raw counts to ln(norm.counts+1) and generate dot plots of scaled average expression per gene per global or vascular population.

scRNA-seq and associated data analysis

ECs from mesenteric fad pads were isolated as described (González-Loyola et al., 2021). Cells were loaded onto a 10x Genomics Chromium Controller for droplet encapsulation with the 10x Single Cell 3′ version 3 chemistry. Two sequencing libraries per genotype were generated and paired-end sequenced at the Lausanne Genomic Technologies Facility on a NovaSeq 6000 device. The count function of the Cell pipeline (version 6.0.0) was used to align reads to the Mus musculus reference genome (refdata-gex-mm10-2020-A), to summarize read counts per gene, and to call cells. Filtered matrices of counts per gene per barcode were imported into R (version 4.1.2) for further downstream analysis with the Seurat package (version 4.3.0). We retained 27,226 cells that expressed between 500 and 5,000 genes, <15% mitochondrial genes expression, <20,000 total UMIs, and <10% dissociation-related genes (van den Brink et al., 2017). We performed batch correction using the canonical correlation analysis integration method (Stuart et al., 2019) implemented in Seurat using 30 components as input. Dimensional reduction was first performed using a principal component using the 2,000 most variable genes, followed by Uniform Manifold Approximation and Projection (UMAP) on the 20 first principal components. Cells were clustered using a shared nearest neighbor graph approach followed by Louvain modularity optimization with a resolution parameter of 0.3. Clusters of cells were annotated to cell types according to their expression of known marker genes. Differential gene expression analysis of Myct1^ecKO cells versus wild-type cells within each population was performed using a Wilcoxon rank-sum test implemented in the FindMarkers function. Genes with a Bonferroni-adjusted P value <0.05 were considered as significantly differentially expressed. Up- and down-regulated genes were separately subjected to over-representation analysis of the GO (Aleksander et al., 2023) and the Hallmark gene set collections (Liberzon et al., 2015) using the enrichGO and enricher functions of the clusterProfiler package (version 4.2.2) (Xu et al., 2024). The Hallmark collection was downloaded using the msigdbr package (version 7.5.1). The scRNA-seq data generated for this publication have been deposited in NCBI’s Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE294341.

Image acquisition and quantifications

Images were acquired using either Zeiss LSM800 or LSM880, Hamamatsu NanoZoomer S60, or Zeiss Axio Imager Z1 microscopes. Image analysis was performed using ImageJ, CellProfiler, or with python packages for image analysis, as detailed below.

Retina analysis was performed using ImageJ. Radial expansion was measured as the distance between the vascular front and the central optic nerve. Vascular density was quantified by applying a threshold to Pecam1⁺ staining.

Tumor analysis was conducted on size-matched tumors using ImageJ. Vascular density was quantified by applying a threshold to Vegfr2⁺ staining. The area of staining was then measured as the percentage of the tumor area that was positive for the endothelial marker.

Adipocyte size quantification on H&E-stained sections was performed using ImageJ. Image processing included Gaussian blur and local contrast enhancement to optimize adipocyte delineation prior to thresholding. Analysis of adipocyte cross-sectional areas was conducted using the “Analyze Particles” function.

Vascular density quantification was performed using ImageJ. The vascular area was delineated by applying a threshold to Pecam1⁺ staining. This area was then normalized to the total fat pad area analyzed to determine relative vascular density.

Ex vivo adipogenesis was quantified using CellProfiler. Nuclei were counted based on DAPI staining, while perilipin⁺ areas were identified by applying a threshold to the perilipin channel. The total perilipin⁺ area was then normalized to the number of nuclei to account for cell density.

In vitro EC p-S6 analysis was performed using CellProfiler for cell segmentation and intensity measurement per cell. Cells were detected based on VE-cadherin or β-catenin staining and were classified as p-S6–positive or –negative based on a threshold intensity value determined by manual inspection of raw data for each experiment.

Images from en face aorta underwent maximum intensity z-projection in ImageJ and analysis using Python with the scikit-image library. Cellpose’s “cyto” model detected cells based on VE-cadherin or Pecam1 staining. p-S6 intensity was measured per cell, and cells were classified as p-S6–positive or –negative based on a threshold intensity value determined by manual inspection of raw data for each experiment.

Analysis of p-S6 in sections was performed in ImageJ. Vascular area was defined by endomucin threshold, and p-S6 intensity was measured within this area.

Protein interaction (PLA), dextran uptake, and OPP protein synthesis were similarly quantified using ImageJ. For all analyses, either the “Find maxima” or Analyze particles function enumerated discrete signal dots (PLA, dextran, or OPP) and DAPI-positive nuclei. The number of PLA/dextran/OPP dots was normalized to cell count in each case.

Colocalization analysis of IFITM2/3 with endosomal or lysosomal markers was conducted using the BIOP JACoP plugin in ImageJ. Analysis was performed on individual z-slices, with single cells defined as ROIs) Colocalization was assessed using Pearson’s correlation coefficient, Mander’s overlap coefficients, and visual inspection of fluorograms. These multiple parameters were collectively considered to determine the extent of colocalization between IFITM2/3 and the endosomal/lysosomal markers.

Analysis of IFITM2/3⁺ and RAB5⁺ vesicles was conducted using Python with the scikit-image library. Full z-stack images were processed. Vesicle detection employed filtering and thresholding of IFITM2/3 and RAB5 channels, respectively. Individual vesicle volumes were quantified from the resulting 3D segmentations.

Analysis of DQ-BSA signal was performed using Python with the scikit-image library. Cellpose’s “cyto2” model detected cells based on VE-cadherin or β-catenin staining. DQ-BSA⁺ areas were identified by thresholding, and the DQ-BSA⁺ area per cell was quantified.

Statistical analysis

Graphs were prepared, and statistical analyses were performed using GraphPad Prism software version 10. Data are shown as mean ± SD. Sample size is indicated as n and is described for each quantification in the corresponding figure legends.

To assess Gaussian distribution of the data, we used Shapiro–Wilk test in GraphPad Prism; normal distribution was assumed if P value was >0.05. Statistical tests were selected based on appropriate assumptions with respect to data distribution and variance characteristics. Details of the test used for each analysis can be found in figure legends. Difference between groups was considered statistically significant when P value was <0.05 and indicated as * on the graph. When nothing is indicated, it is not significant.

Figure preparation

Figures were created in part using icons obtained from BioRender.com under an academic license. Icons and graphs were assembled and modified in Affinity.

Text preparation

Text polishing and language refinement were performed using Perplexity AI (version 2) to improve clarity and readability of the manuscript.

Resource availability

Requests for information and resources should be directed to the lead contact, Tatiana Petrova (tatiana.petrova@unil.ch).

Online supplemental material

Fig. S1 contains MYCT1 gene expression data in ECs, details for the Myct1^ecKO mouse model, and additional phenotypic characterization of Myct1^ecKO mice. Fig. S2 contains metabolic cage data, blood metabolic parameters, and analysis of tumor growth and angiogenesis. Fig. S3 shows scRNA-seq data, including dot plot of clusters and volcano plot of differentially expressed genes, MYCT1 antibody and siRNA validation data, as well as validation of mTORC1 pathway activation. Fig. S4 shows MYCT1 sequence alignment and validation data for MYCT1–IFITM2/3 interaction. Fig. S5 shows endocytosis and endosomal vesicle–related phenotypes, as well as rescue experiments for mTORC1 pathway activation. Table S1 contains differentially expressed genes in BEC cluster. Table S2 contains GO terms and hallmark identified by overrepresentation analysis of differentially expressed genes in BEC cluster. Table S3 shows MYCT1 phosphorylation sites identified by mass spectrometry. Table S4 lists MYCT1-interacting proteins identified by mass spectrometry. Table S5 contains GO terms of MYCT1-interacting proteins identified by mass spectrometry.

Supplementary Material

Table S1

contains differentially expressed genes in BEC cluster.

Table S2

contains GO terms and hallmark identified by overrepresentation analysis of differentially expressed genes in BEC cluster.

Table S3

shows MYCT1 phosphorylation sites identified by mass spectrometry.

Table S4

lists MYCT1 interacting proteins identified by mass spectrometry.

Table S5

contains GO terms of MYCT1 interacting proteins identified by mass spectrometry.

SourceData F4

is the source file for Fig. 4.

SourceData F5

is the source file for Fig. 5.

SourceData F6

is the source file for Fig. 6.

SourceData FS3

is the source file for Fig. S3.

SourceData FS4

is the source file for Fig. S4.

Data availability

Resources generated in this study will be made available on request. The scRNA-seq data generated for this publication have been deposited in NCBI’s GEO and are accessible through GEO Series accession number GSE294341.