An autophagic deficit in the uterine vessel microenvironment provokes hyperpermeability through deregulated VEGFA, NOS1, and CTNNB1

Abstract

The uterus undergoes vascular changes during the reproductive cycle and pregnancy. Steroid hormone deprivation induces macroautophagy/autophagy in major uterine cell types. Herein, we explored the functions of uterine autophagy using the Amhr2-Cre-driven atg7 deletion model. Deletion of Atg7 was confirmed by functional deficit of autophagy in uterine stromal, myometrial, and vascular smooth muscle cells, but not in endothelial cells. atg7^d/d uteri exhibited enhanced stromal edema accompanied by dilation of blood vessels. Ovariectomized atg7^d/d uteri showed decreased expression of endothelial junction-related proteins, such as CTNNB1/beta-catenin, with increased vascular permeability, and increased expression of VEGFA and NOS1. Nitric oxide (NO) was shown to mediate VEGFA-induced vascular permeability by targeting CTNNB1. NO involvement in maintaining endothelial junctional stability in atg7^d/d uteri was confirmed by the reduction in extravasation following treatment with a NOS inhibitor. We also showed that atg7^d/d uterine phenotype improved the fetal weight:placental weight ratio, which is one of the indicators of assessing the status of preeclampsia. We showed that autophagic deficit in the uterine vessel microenvironment provokes hyperpermeability through the deregulation of VEGFA, NOS1, and CTNNB1.Abbreviations: ACTA2: actin, alpha 2, smooth muscle, aortic; Amhr2: anti-Mullerian hormone type 2 receptor; ANGPT1: angiopoietin 1; ATG: autophagy-related; CDH5: cadherin 5; CLDN5: claudin 5; COL1A1: collagen, type I, alpha 1; CSPG4/NG2: chondroitin sulfate proteoglycan 4; CTNNB1: catenin (cadherin associated protein), beta 1; DES: desmin; EDN1: endothelin 1; EDNRB: endothelin receptor type B; F3: coagulation factor III; KDR/FLK1/VEGFR2: kinase insert domain protein receptor; LYVE1: lymphatic vessel endothelial hyaluronan receptor 1; MAP1LC3B: microtubule-associated protein 1 light chain 3 beta; MCAM/CD146: melanoma cell adhesion molecule; MYL2: myosin, light polypeptide 2, regulatory, cardiac, slow; MYLK: myosin, light polypeptide kinase; NOS1/nNOS: nitric oxide synthase 1, neuronal; NOS2/iNOS: nitric oxide synthase 2, inducible; NOS3/eNOS: nitric oxide synthase 3, endothelial cell; OVX: ovariectomy; PECAM1/CD31: platelet/endothelial cell adhesion molecule 1; POSTN: periostin, osteoblast specific factor; SQSTM1: sequestosome 1; TEK/Tie2: TEK receptor tyrosine kinase; TJP1/ZO-1: tight junction protein 1; TUBB1, tubulin, beta 1 class VI; USC: uterine stromal cell; VEGFA: vascular endothelial growth factor A; VSMC: vascular smooth muscle cell.

Keywords: Atg7; autophagy; permeability; uterus; vascular factor.

Figure 1.

Autophagic deficit in atg7^d/d uteri. (A) Uterine cell type-specific deletion of Atg7 was achieved by crossing Atg7 ^f/f mice with Amhr2-Cre transgene-expressing mice. This cross produced atg7^d/d mice with gene deletion in the uterine stroma and myometrium, while Atg7 was retained in the epithelium and subsets of stromal and myometrial cells. In atg7^d/d uterus, the gene-deleted region in the mesenchyme is shown in gray and the gene-retained region is shown in pink. The bottom figure shows an enlarged image of the spiral artery. atg7^d/d uterine vessels presumably lose gene expression in mesenchyme-derived VSMCs and pericytes (P), but not in the endothelial cells (EC). ge, glandular epithelium; myo; myometrium; s, stroma; le, luminal epithelium; M, mesometrial side; AM, anti-mesometrial side. (B) Ovariectomy-induced accumulation of SQSTM1 and LC3B-I in atg7^d/d uteri. OVX-induced SQSTM1 accumulation in atg7^d/d uteri. Each sample was prepared from a single mouse. (C) Age-dependent accumulation of SQSTM1 in atg7^d/d uteri. Uteri were collected from random cycling mice of different ages. 4 w, 4-week-old; 8 w, 8-week-old; 1 y, > 1-year-old; sh, 2-week rest after OVX; lo, 6-month rest after OVX; (+), OVX atg7^d/d uterus. Each sample was prepared from a single mouse. (D) Immunofluorescence staining of SQSTM1 in the tissue sections of OVX Atg7 ^f/f and atg7^d/d uteri treated with vehicle (oil) or E₂ (100 ng/0.1 ml oil). AM, antimesometrial side; M, mesometrial side; le, luminal epithelium; s, stroma. Scale bar: 100 μm. (E) Immunofluorescence staining of SQSTM1 (green) and PECAM1 (red) in the uterine tissue sections of mice used in (C). lm, longitudinal muscle, cm, circular muscle; s, stroma; le, luminal epithelium; ec, endothelial cells; vm, vascular smooth muscle cells. Scale bar: 100 μm

Figure 2.

Dilation of uterine vessels and enhanced stromal extravasation in OVX atg7^d/d uteri. (A) Paraffin sections of OVX Atg7 ^f/f and atg7^d/d uteri were stained with hematoxylin and eosin. OVX mice (8-week-old) were subcutaneously injected with sesame oil (vehicle, 0.1 ml/mouse) or E₂ (100 ng/0.1 ml/mouse) after 2 weeks of rest. The edematous stroma of atg7^d/d uteri (*s) in comparison with that of Atg7 ^f/f mice. cm, circular muscle; s, stroma; le, luminal epithelium; ge, glandular epithelium. Scale bar: 50 μm. (B) Uterine Evans blue content measured by Miles assay. Significant increase in uterine tissue extravasation in atg7^d/d uteri. Scale bar: 1 cm. ¹*p = 0.0024, t = 3.313, df = 15; ²ns, p = 0.0829, t = 1.463, df = 14. (C) Thin sections of E₂-injected OVX Atg7 ^f/f and atg7^d/d uteri. Mice (8-week-old) were ovariectomized and administered sesame oil (vehicle, 0.1 ml/mouse) or E₂ (100 ng/0.1 ml/mouse) after 2 weeks of rest. A representative set of images is shown at 900X (yellow arrows indicate enlarged lumen of vessels). Two sets of thin sections from different mice were chosen and blood vessels in the subepithelial stroma and their luminal areas (15 and 17 areas of Atg7 ^f/f and atg7^d/d uteri, respectively) were quantified using NIH ImageJ software (Fig. S2B). Bars represent means ± SEM. ¹ns, p = 0.1655, t = 0.9883, df = 30; ²***p = 0.0003, t = 3.706, df = 48. Scale bar: 7.5 μm. (D) LYVE1-positive lymphatic vessels in OVX Atg7 ^f/f and atg7^d/d uteri. Mice (8-week-old) were ovariectomized and administered sesame oil (vehicle, 0.1 ml/mouse) or E₂ (100 ng/0.1 ml/mouse) after 2 weeks of rest. Several large open lymphatic vessels with stretched morphology were observed in atg7^d/d uteri (white arrows). Scale bar: 100 μm

Figure 3.

Characterization of cellular abnormalities in OVX atg7^d/d uteri. (A) Immunofluorescence staining of SQSTM1 and PECAM1 in OVX atg7^d/d uteri. Localization of SQSTM1 (green signal) in vascular mural cell layers surrounding PECAM1⁺ (red signal) endothelial cells. The area marked with a yellow rectangle is enlarged in the right panel. Myo, myometrium; le, luminal epithelium; s, stroma; VSMC, vascular smooth muscle cell; EC, endothelial cells. Scale bar: 100 μm. (B) The average number of USCs per mouse harvested by enzymatic digestion method. Usually 3–5 mice were used for USC preparation for cell sorting, culture, RNA preparation, or proteome array experiments. ¹*p = 0.029, t = 1.950, df = 41; ²*p = 0.0271, t = 2.181, df = 10. (C) Co-immunofluorescence staining of cell markers and SQSTM1 in USCs isolated from E₂-injected OVX Atg7 ^f/f and atg7^d/d mice. Red puncta in subsets of atg7^d/d USCs indicating the sites of SQSTM1 accumulation. Scale bar: 20 μm. (D) USCs were isolated from E₂-injected OVX Atg7 ^f/f and atg7^d/d mice and subjected to cell sorting. Five sets of independent analyses were performed using pooled samples from 3 or 4 mice and a representative set is shown here. (E) Co-immunofluorescence staining of MCAM (green signal) and other markers in USCs isolated from E₂-injected OVX atg7^d/d mice. Subsets of MCAM⁺ cells also expressed DES, CSPG4, or ACTA2. White scale bar: 20 μm.(F) Immunofluorescence staining of COL1A1 and PECAM1 in OVX Atg7 ^f/f and atg7^d/d uteri. Enlarged images of subepithelial stromal regions are shown. Scale bar: 100 (left panel) or 10 μm (middle and right panels)

Figure 4.

Vascular abnormalities in OVX atg7^d/d uteri. (A) Western blot analyses of PECAM1 and several junctional proteins. Whole uterine samples were prepared from individual mice and quantitation of chemiluminescence signals from four samples was performed. TJP1, tight junction protein 1; CLDN5, claudin 5; pCDH5, phosphorylated VE-cadherin; CDH, VE-cadherin; CTNNB1, beta-catenin. Values of TUBB1 chemiluminescence were used to normalize the data. ¹ns, p = 0.0816, t = 1.839, df = 3; ²*p = 0.021, t = 3.384, df = 3; ³**p = 0.0081, t = 4.906; df = 3; ⁴**p = 0.0076, t = 5.017, df = 3; ⁵***p = 0.0001, t = 46.88, df = 3. (B) Western blot analyses of smooth muscle markers in OVX atg7^d/d uteri. Whole uterine samples were prepared from individual mice and quantitation of chemiluminescence signals from four samples was performed. pMYL2, phosphorylated myosin light chain 2; MYL2, myosin light chain 2; MYLK, MLC kinase; ACTA2, actin, alpha 2, smooth muscle, aorta. Decreased ratio of pMYL2/MYL2 indicates a phenotype of muscle relaxation. ¹**p = 0.041, t = 6.284, df = 3; ²***p = 0.0002, t = 17.50, df = 3; ³*p = 0.0491, t = 2.373, df = 3. (C) Western blot analyses of EDN1, a vasoconstrictor, and its receptor in OVX atg7^d/d uteri. Whole uterine samples were prepared from individual mice and quantitation of chemiluminescence signal from four samples was performed. EDN1 and EDNRB levels significantly increased in OVX atg7^d/d uteri. ¹***p = 0.0007, t = 11.40, df = 3; ²*p = 0.0477, t = 2.406, df = 3. (D) Western blot analyses of NOS in OVX atg7^d/d uteri. Whole uterine samples were prepared from individual mice and quantitation of chemiluminescence signals from four samples was performed. NOS1, neuronal NOS; NOS3, endothelial NOS. NOS2, the inducible form, was undetectable. ¹**p = 0.0037, t = 6.500, df = 3; ²*p = 0.0194, t = 3.526, df = 3. (E) Co-immunofluorescence staining of NOS1 (green) and PECAM1 (red) in OVX atg7^d/d uteri. Scale bar: 50 μm. White arrows indicate overlapping signals (yellow) between NOS1 and PECAM1

Figure 5.

VEGFA- and NO-regulated hyperpermeability in atg7^d/d uteri. (A) USC lysates were prepared from 5 or 6 OVX Atg7 ^f/f and atg7^d/d mice and assayed with the Mouse XL Cytokine Array containing 111 secreted factors. Three independent experiments (17 mice per group) were performed and factors showing more than 1.2-fold increase in atg7^d/d USC lysates in all three sets are shown in Table 1. (B) USC lysates were prepared from 5 or 6 OVX Atg7 ^f/f and atg7^d/d mice and assayed with the Proteome Profile^TM Mouse Phospho-Receptor Tyrosine Kinase (RTK) Array. Three independent experiments (17 mice per group) were performed and factors showing more than 1.2-fold increase in atg7^d/d USC lysates in all three sets are shown in Table 2. (C) qPCR analysis of factors with significant changes in both 3ʹ mRNA sequencing and cytokine array. Relative gene expression was normalized to ribosomal protein L7 (Rpl7). Values represent the mean ± S.E.M. ¹*p = 0.0141, t = 2.564, df = 10; ²***p < 0.0001, t = 7.918, df = 10; ³***p < 0.0001, t = 6.207, df = 10; ⁴***p < 0.0001, t = 17.43, df = 10; ⁵*p = 0.0436, t = 1.895, df = 10; ⁶***p < 0.0001, t = 13.26, df = 10. (D) Age-dependent increase of VEGFA and NOS1 levels in atg7^d/d uteri. Uteri were collected from random cycling mice. 4 w, 4-week-old; 8 w, 8-week-old; 1 y, > 1-year-old. Each sample was prepared from a single mouse. EDN1 levels did not show an age-dependent increase like VEGFA and NOS1. (E) Experimental scheme for examining the effect of L-NAME on plasma extravasation in OVX atg7^d/d uteri. Mice received daily intraperitoneal injections of L-NAME (0.625 mg/0.1 ml PBS) for 10 (Miles assay) or 14 d (western blotting). Uteri were collected 24 h after the last injection and subjected to further analyses. (F) Uterine Evans blue content after L-NAME injection measured by Miles assay. ¹**p = 0.0049, t = 2.934, df = 16; ²ns, p = 0.0506, t = 1.739, df = 16; ³*p = 0.047, t = 1.780, df = 15. (G) Western blot analyses of CTNNB1 in L-NAME-injected atg7^d/d uteri. ¹ns, p = 0.1510, t = 1.129, df = 6; ²*p = 0.0176, t = 2.708, df = 6. (H) Co-immunofluorescence staining of CTNNB1 (green) and PECAM1 (red) in OVX atg7^d/d uteri. Scale bar: 50 μm. White arrows indicate CTNNB1 signal in PECAM1-positive endothelial cells

Figure 6.

Atg7^d/d mice exhibit partial protection against L-NAME-induced PE. (A) Experimental scheme for producing PE-like effects by injecting L-NAME in pregnant atg7^d/d mice. Mice were bred naturally with stud ICR male mice. From day 8 to 17 of pregnancy, mice were subcutaneously injected with L-NAME (1.25 mg/0.1 ml PBS) daily. Mice were sacrificed on day 18 of pregnancy, and data were obtained as shown in (B) to (F). (B) The number of live conceptuses was counted at the time sacrifice on day 18 of pregnancy. (C) Fetal resorption and death rates for each mouse at the time of sacrifice were plotted. Although not statistically significant, the resorption rate in L-NAME-injected Atg7 ^f/f mice was higher than that in L-NAME-injected atg7^d/d mice (bracket with an arrow). (D) Individual fetal weight plotted for each group. Treatment of both Atg7 ^f/f and atg7^d/d mice with L-NAME induced growth retardation. In the L-NAME-injected Atg7 ^f/f group, a number of fetuses showed severe retardation (bracket with an arrow), while this effect was not observed in the L-NAME-injected atg7^d/d group. ¹***p < 0.0001, t = 3.954, df = 153; ²***p < 0.0001, t = 4.333, df = 132; ³*p = 0.0485, t = 1.670, df = 153. (E) Individual placental weight plotted for each group. ¹ns, p = 0.4972, t = 0.00711, df = 153; ²**p = 0.0083, t = 2.341, df = 132; ³*p = 0.0104, t = 2.341, df = 132. (F) The FW/PW ratio was individually plotted. Treatment with L-NAME had adverse effects on Atg7 ^f/f mice, reducing the ratio of fetal/placental weight. In contrast, L-NAME treatment did not significantly influence the FW/PW ratio in atg7^d/d mice. ¹**p = 0.0028, t = 2/816, df = 153; ²ns, p = 0.4276, t = 0.1829, df = 132; ³**p = 0.0016, t = 3.001, df = 132

Figure 7.

A schematic diagram depicting the role of autophagy in uterine vasculature in mice. The uterus exhibits dynamic changes in vascularity during the estrous cycle and pregnancy. Impaired autophagic flux in the uterine vascular mural cells and fibroblasts in atg7^d/d mice may lead to accumulation of vascular factors, including VEGFA and NOS1. Enhanced signaling by VEGFA (vascular mural cells) via KDR (endothelial cells) (Figure 5A and 5B) may in turn induce NO accumulation in the endothelial compartment (Figure 4D and 4E) via endothelial NOS1. NO promptly causes nitrosylation of CTNNB1, a seminal component of endothelial junctions. Nitrosylated CTNNB1 dissociates from the CDH5 complex in adherens junctions, leading to degradation of junctional proteins (Figure 4A) and subsequent junctional destabilization and hyperpermeability. L-NAME, an inhibitor of NOS, can reduce the exaggerated vascular permeability in atg7^d/d uteri