Domain Specific Partitioning of Uterine Artery Endothelial Connexin43 and Caveolin-1

Abstract

Uterine vascular adaptations facilitate rises in uterine blood flow (UBF) during pregnancy, which are associated with gap junction connexin (Cx) proteins and eNOS. In uterine artery endothelial cells (UAECs) ATP activates eNOS in a pregnancy (P) specific manner that is dependent on Cx43 function. Caveolar subcellular domain partitioning plays key roles in ATP-induced eNOS activation and NO production. Little is known regarding the partitioning of Cx proteins to caveolar domains or their dynamics upon ATP treatment. We observed that Cx43-mediated gap junction function upon ATP stimulation is associated with Cx43 re-partitioning between the non-caveolar and caveolar domains. Compared to UAECs from nonpregnant (NP) ewes, levels of ATP, PGI₂, cAMP, NOx, and cGMP were 2-fold higher (P<0.05) in P-UAECs. In P-UAECs ATP increased lucifer yellow dye transfer, a response abrogated by Gap27, but not Gap 26 indicating involvement of Cx43 but not Cx37. Confocal microscopy revealed domain partitioning of Cx43 and Cav-1. In P-UAECs LC/MS/MS analysis revealed only Cx43 in the caveolar domain. In contrast, Cx37 was located only in the non-caveolar pool. Western analysis revealed that ATP increased Cx43 distribution (1.7-fold;P=0.013) to the caveolar domain, but had no effect on Cx37. These data demonstrate rapid ATP-stimulated repartitioning of Cx43 to the caveolae, where eNOS resides and plays an important role in NO-mediated increasing UBF during pregnancy.

Introduction

Numerous maternal cardiovascular adaptations occur during pregnancy, including decreases in uterine vascular resistance, which accommodate dramatic increases in uterine blood flow (UBF) that facilitates fetal growth¹. Endothelial nitric oxide synthase (eNOS) is partly responsible for the observed increases in UBF via production of NO². Pregnancy is characterized as a physiologic state of vasodilation where the endothelial-derived vasodilators PGI₂^3–5 and NO ^2,6,7 and their respective second messengers cAMP ^8–10 and cGMP ^6,8–13 are elevated. We reported in the ex vivo uterine artery endothelium (UAendo) model elevated expressions of connexin (Cx) 37, 43, and eNOS during pregnancy¹⁴. We demonstrated requirement of the gap junction (GJ) protein Cx43 for ATP-induced Ca²⁺ bursts associated with eNOS activation both ex vivo¹⁴ and in UAECs in vitro¹⁵. Moreover, the normal pregnancy (P) associated Ca²⁺ bursts upon ATP stimulation were higher than in the nonpregnant (NP) UAECs¹⁵, demonstrating enhanced GJ communication between endothelial cells is crucial to for eNOS/NO-mediated changes during gestation.

Caveolae are specialized protein-rich plasmalemmal invagination sites of signal transduction regulation and are “hubs” for NO regulation¹⁶. Within caveolae, eNOS is bound to the scaffolding protein caveolin-1 (Cav-1), maintaining eNOS in an inactive state^17,18. Cav-1 and eNOS interactions occur via the caveolin scaffolding domain on Cav-1, which is the same binding region as the eNOS calmodulin binding domain and is necessary for Ca²⁺/calmodulin-dependent eNOS activation. Some proteins which selectively localize to caveolar microdomains regulate eNOS activity¹⁶, which is substantially activated with ATP stimulation and require Cx43, but not Cx37, for NO production^14,15. Little is known regarding trafficking and partitioning of connexin proteins to caveolar domains, the home for eNOS, and their dynamics with ATP treatment.

We hypothesized that Cx43-mediated GJ function upon ATP stimulation is associated with Cx43 re-partitioning from the non-caveolar to the caveolar domains. We investigated: 1) endogenous levels of ATP, vasodilators (PGI₂ and NO), and their respective cyclic nucleotides (cAMP and cGMP) in NP-UAECs vs. P-UAECs; 2)if the actions of ATP on Lucifer Yellow dye transfer between P-UAECs specifically involves Cx43 and re-partitioning of Cx43 and Cx37 proteins to the caveolar fraction; and 3)the presence of Cx43 and Cx37 proteins relative to other junctional proteins in the caveolar compartment.

Methods

Detailed methods see online supplement http://hyper.ahajournals.org.

Cell Preparation/Culture

Protocols were approved by the University of Wisconsin-Madison Animal Care Committee¹⁹. NP-UAECs and P-UAECs were isolated and cultured from nonpregnant (n=6–10) and late pregnant (n=6–10) ewes¹⁹. Passage 4 UAECs (~90% confluence) were: a)transferred to 96-well plates for experimental treatments and supernatant medium was used for enzyme immunoassays, b)transferred to chamber slides for experimental treatments, immunofluorescence, and Lucifer yellow dye transfer studies, or c)lysed for Western Analyses.

Measurements of Basal ATP, PGI₂, cAMP, NOx, and cGMP levels

We used ViaLight Plus High Sensitivity Cell Proliferation and Cytotoxicity Assays to measure ATP²⁰. PGI₂ (measuring the stable non-enzymatic hydrolysis product, 6-keto-PGF1α)³, cAMP, and cGMP production levels were determined using EIA kits. Total NO was measured using HPLC, NOx Analyzer (ENO-30).

Lucifer Yellow Dye Transfer

P-UAECs were grown on Lab-Tek II chamber slides with glass coverslips and treated (30min) with ATP (100 μmol/L) in the absence or presence of Cx peptide inhibitors (300 μmol/L) for Cx43 (43, 37)Gap27, Cx37 (40, 37)Gap26, or scramble control peptide (43,37 scramble)GAP27. UAECs were scraped at several sites with a razor and incubated with 0.05% Lucifer Yellow and tetramethylrhodamine (TMR) dextran ¹⁵, and viewed by fluorescent microscopy (triplicates). Areas of dye spread (fluorescent areas) were quantified using ImageJ software.

Immunofluorescence Confocal Microscopy for Cav-1 and Cx43

Performed in the absence and presence of ATP (100 μmol/L; 30 minutes) as described previously²¹.

Caveolar Isolation

Performed as described previously²¹.

Proteomic Analysis

Peptides were analyzed by nano-Liquid chromatography-tandem mass spectrometry (LC-MS/MS) as described previously¹⁶.

Western analyses

Western Analyses performed on whole cell lysates¹⁹ used antibodies for Cav-1 (1:10000), Cx43 (1:3000), and Cx37 (1:5000), secondary antibodies for anti-rabbit or anti-mouse, and detection using enhanced chemiluminescence (ECL) and HyperFilm ECL (GE Healthcare Bio-Sciences Corp., Piscataway, NJ) with β-actin as a loading control.

Statistical Analysis

Data are presented as Means±SEM fold of Control. Group differences were analyzed using one-way or two-way ANOVA (SigmaPlot 11) followed by post hoc multiple pairwise comparisons Student-Newman-Keuls/Bonferroni tests. Significance was established a priori at P<0.05.

Results

Basal ATP, PGI₂, cAMP, NOx, and cGMP levels

Compared to NP-UAECs, P-UAECs had more (P<0.05) endogenous ATP(1.9-fold) and the endothelial-derived vasodilators PGI₂(2.5-fold; P<0.05) and NOx(1.4-fold; P<0.05) were elevated along with proportional increases of their respective second messengers cAMP(2.6-fold; P<0.001) and cGMP(1.8-fold; P<0.05)(Table 1).

Table 1.

Basal ATP, PGI₂, cAMP, NOx, and cGMP levels in cultured NP-UAECs versus P-UAECs.

Compound	NP-UAECs	P-UAECs
ATP (RLU)	223 ± 7 (n=6)	434 ± 19 (n=6) *
PGI2 (pg/ml × mg protein)	97 ± 7 (n=10)	228 ± 27 (n=10) *
cAMP (pg/ml × mg protein)	677 ± 65 (n=10)	1732 ± 18 (n=10) †
NOx (nmol)	1101 ± 70 (n=6)	1590 ± 190 (n=6)*
cGMP (pg/ml × mg protein)	359 ± 26 (n=6)	721 ± 60 (n=10) *

Values are Means ± SEM, n=number of cell lines (duplicates).

^*Increase (P<0.05) in compound in P-UAECs compared to NP-UAECs.

^†Increase (P<0.001) in compound in P-UAECs compared to NP-UAECs.

ATP measured in Relative Light Units (RLU). PGI₂ measured as, 6-keto-PGF1α. NOx measured by NO metabolites, nitrates and nitrites.

Cx43, but not Cx37, GJs mediate lucifer yellow dye transfer

Lucifer yellow dye transfer in P-UAECs was elevated by ATP treatment, a response abrogated by GAP27 peptide, but not GAP26(Figure 1). Control-scrambled peptide for GAP27 had no effect on dye transfer with ATP treatment, indicting functional importance of GJ intercellular communication (GJIC) via Cx43, not Cx37.

Figure 1. ATP-induced Lucifer Yellow Dye Transfer is Cx43 not Cx37 mediated.

Representative fluorescence dye microscopy images following treatment with 100uM ATP and/or inhibitory and control GAP peptides in P-UAECs. (A) Left: Tetramethylrhodamine (TMR) dextran staining at the cell membrane. TMR dextran is too large to pass through GJs and serves as a measure of cell damage at the “wounding site” and a marker for Lucifer yellow entry. Middle: Lucifer yellow dye that passes through GJs starting at the “wound site” into adjacent cells. Right: Merged photo of TMR dextran and Lucifer yellow dye images. P-UAECs treated with ATP displayed increased Lucifer yellow dye transfer NP-UAECs. GAP27, but not GAP26 or scrambled control peptide abrogated ATP-mediated increases in lucifer yellow dye transfer. Dotted lines (---) denote “wound site” and denote distance the Lucifer Yellow dye traveled from the “wound site”. (B) Fold of control quantification of the fluorescence area in dye-coupled cells. * Increase (P<0.05) P-UAEC treatment groups compared to Control (n=3).

Cx43, but not Cx37, co-localizes with Cav-1

In P-UAECs, confocal microscopy revealed subcellular locations of Cx43 (red) with the marker for the caveolae Cav-1 (green). We observed distinct ATP-mediated repartitioning and colocalization between Cx43 and Cav-1 (merger yellow) in both the caveolar and noncaveolar domains (Figure 2).

Figure 2. ATP repartitions Cx43 to co-localize with Cav-1 to the plasma membrane.

Representative confocal images demonstrate that in response to ATP (100 μmol/L;30 min) Cx43 (red) is co-localizes with Cav-1 (green) in the cytosol of control cells, whereas Cx43 and Cav-1 are predominately co-localized (merge yellow) to the plasma membrane.

Proteomic analysis of caveolar Cx43 and Cx37

We observed the presence of Cx43, but not Cx37 (not shown) in the P-UAEC caveolar fraction. Representative MS/MS spectra of Cx43 illustrate Y ion series assignment showing peptide fragment ions from the C-terminus (Figure 3A). B ion series from the N-terminus confirm Cx43 peptide assignment. We further observed an excellent sequest cross-correlation score for Cx43 peptide sequences and low parent mass error of −0.79. Proteomic analysis did not identify Cx37 in P-UAEC caveolar fractions. The identified Cx43 peptide sequence had 17% coverage with four unique peptides, four unique spectra, and five total spectra (Figure 3B).

Figure 3. Representative MS/MS spectra of Cx43.

(A) Good Sequest cross correlation score and low parent ion tolerance were noted. (B) Identified Cx43 protein sequence with four unique peptides, four unique spectra and 5 total spectra were identified with 17% sequence (peptide) coverage. The yellow highlighted areas show the actual peptides detected by the mass spectrometry, and the green highlighted areas show oxidized or methylated amino acids.

Caveolar domain specific localization of Cx43, but not Cx37

Using western analysis we determined the presence of Cx43 in both P-UAEC caveolar and non-caveolar domains. ATP significantly increased Cx43 levels 1.7 fold(P<0.05) in the caveolar, but not in the non-caveolar domain(Figure 4A). Low levels of Cx37 were observed in the caveolar domain. Although Cx37 was predominately located in the non-caveolar domain treatment with ATP did not significantly alter Cx37 distribution in either the caveolar or non-caveolar domains (Figure 4B). Thus, western analysis confirmed in a semi-quantitative fashion the proteomic analysis showing the presence mainly of Cx43 (not Cx37) in the caveolar domain even after ATP stimulation in P-UAECs.

Figure 4. Effects of ATP treatment on P-UAEC Cx43 and Cx37 levels in caveolar and non-caveolar domains.

(A) Cx43 was located in both the caveolar and non-caveolar domains and upon ATP treatment treated (100 μmol/L;30 min) was significantly increased (* P=0.013) only in the caveolar domain. (B) Cx37 was predominately localized in the non-caveolar domain and was not significantly altered by ATP treatment.

Cav-1 is expressed in vivo in UAendo and displays linear correlations with phosphorylated and total eNOS

We previously reported in ex vivo samples of UAendo, substantial local unilateral pregnancy associated local increases in Cx37, Cx43, and phosphorylation of eNOS associated with in vivo rises in UBF and shear stress¹⁴. Herein we reexamined those UAendo blots for the presence of Cav-1 and observed substantial increases in Cav-1 during pregnancy that was limited only to the UAendo ipsilateral to the gravid uterine horn (Figure S-1). We observed significant correlations between UAendo Cav-1 and P⁶³⁵ eNOS (Figure S-2;r=0.57) and total eNOS (Figure S-3;r=0.62), providing additional physiologic in vivo¹⁴ evidence supporting our current in vitro data showing that Cx43 association with Cav-1 facilities GJ function (Fig 1) and eNOS activation^15,21.

Discussion

We report herein that compared with NP-UAECs, P-UAECs exhibit higher endogenous ATP levels and elevations in the potent endothelial vasodilators PGI₂&NOx and proportional elevations in their second messengers cAMP&cGMP. P-UAEC elevations of ATP and these cyclic nucleotides are novel observations and consistent with other studies in vivo^9,11–13,19. These observations demonstrate a vasodilatory phenotype of UAECs for pregnancy specific programming with elevations in basal PGI₂^3,19 and NO^2,19 production as well as key UA endothelial-derived proteins, enzymes, and receptors^3,8,19. We previously showed ex vivo in UA explants that PGI₂ was responsible for the cAMP production and NO for cGMP production¹⁴. Elevations in these endothelial vasodilators are responsible for proportional inductions of the respective elevated levels of cAMP^8,13,19 and cGMP ^8,13,19. Signaling of cyclic nucleotides are further implicated in the formation and assembly of GJs through posttranslational modifications of Cx43²². Thus this signaling process involves a feedforward loop between ATP, cyclic nucleotides, and Cx43-mediated GJIC to establish this pregnancy specific vasodilatory phenotype.

GJ proteins Cx43 and Cx37 are the most prevalent connexin subtypes observed in the UAendo and are elevated during pregnancy¹⁴. We directly showed that ATP increases GJIC functionality, which is solely Cx43 mediated (GAP27 inhibited) and that Cx37 is not involved since GAP26 peptide did not alter lucifer yellow dye transfer. These data are consistent with cultured P-UAECs¹⁵ and ex vivo¹⁴ studies using these same peptide inhibitors where Cx43, but not Cx37, was required for Ca²⁺-associated ATP-mediated rises in intracellular NO. This supports the notion that eNOS activation and deactivation in caveolae is intimately coupled to changes in intracellular Ca²⁺. These studies highlight that Cx43-mediated GJIC and NO biosynthesis are interdependently linked for vasodilatory function and that the relationship between Cx43 and NO is surmised by studies showing that both connexins and eNOS influence the expression and activity of each other in association with their interaction with Cav-1^23–25 (Figs S-2&S-3).

We present, confocal microscopy data showing an increase in the rapid ATP-induced co-localization of Cx43 and Cav-1 at the plasma membrane. This is consistent with studies showing that Cav-1 and Cx43 both co-localize^26,27 and translocate from the golgi to the plasma membrane. Reduction of Cav-1 in endothelial cells directly affected connexin location in the plasma membrane and formation and function of GJs²⁷. Caveolae localize to myoendothelial GJs²⁸ and ATP-stimulated translocation of Cx43 may increase transmission of signaling molecules across myoendothelial GJs connecting endothelial and vascular smooth muscle cells. It is unclear how ATP functions to target the Cx43 specifically to the caveolae. Schubert et al²⁶ examined in an unstimulated state how connexins are specifically targeted to lipid raft domains in order to interact with Cav-1 in human embryonic kidney cells. Our work using caveolar enriched fractions directly corroborates and visualizes their suppositions as they demonstrated that Cx43 co-localizes, co-fractionates, and co-immunoprecipitates with Cav-1²⁶. However, we further show that Cx43 partitioning into caveolar fractions change with ATP treatment demonstrating the rapid and dynamic nature of these events as they relate to NO production. We also show specific Cx43 responses relative to Cx37 especially for ATP-induced Ca²⁺-mediated activation of eNOS^14,15. Upon stimulation with the Ca²⁺ mobilizing physiologic agonist ATP, uterine endothelial Cx43 translocates to the caveolar compartment of cell membranes where they interact with and activate eNOS in order to facilitate NO rises.

Cx37 function in non-caveolar domains remains unclear. Proteomic analysis revealed Cx43, but not Cx37, in the caveolar proteome. Yet, Cx37 was predominately located in the non-caveolar domain and was unaltered by agonistic ATP treatment. We do know that Cx37 is not intimately involved in ATP-stimulated NO production^17,18. We interpret this lack of Cx37 levels in the caveolar domain to more fully explain why the GAP26 in P-UAECs ¹⁸ or UAendo¹⁷, which blocks Cx37 had no effect on ATP-mediated Ca²⁺ burst associated rises in NO production. Moreover, others reported in microvascular endothelial cells that Cav-1 did not co-localize with Cx37²⁹, which is fully consistent with our study.

In conclusion, UAECs exhibit pregnancy-specific endogenous increases in ATP levels and a vasodilatory phenotype associated with elevations in PGI₂-cAMP and NO-cGMP production. Consistent with the model described in Figure 5, ATP-stimulated increases in lucifer yellow dye transfer occurs primarily via Cx43 by inducing increased Cx43 levels and cAMP-mediated GJ co-localization with Cav-1 in caveolar domains where eNOS resides. This rapid temporal and spatial co-localization of Cx43 to the caveolae demonstrates the need for GJIC and quick cell signaling in gestation to maintain rises in intracellular Ca2+ for increased NO production^17,18. Cx37 was predominately located in the non-caveolar domain and unaltered by ATP stimulation, but may have a role in myoendothelial GJs with underlying vascular smooth muscle. Our data are suggestive that the pregnancy specific programming of UAECs at the level of signaling involves enhanced Cx43 function and localization within the caveolae to rapidly increase the endothelial synthesis of NO to maintain UBF required for a healthy fetal growth and survival.

Figure 5. Cx43 and Cav-1 localization.

Basal State: Cav-1 maintains eNOS in its inactivated state in caveolar domains and thereby limits NO production. Cx43 GJs are present near the caveolae. ATP-Stimulated: ATP stimulates cAMP-mediated Cx43 trafficking and localization to caveolae increasing Cx43 GJ plaques and GJIC. ATP stimulation increases intracellular calcium within and between adjoining cells, eNOS disassociates from cav-1, and is phosphorylated to increase NO production and vasodilation.

Perspectives

The specific mechanisms by which GJs regulate vascular tone through the modulation of vasodilatory pathways during pregnancy are unclear. However, in vivo and ex vivo studies have shown that these vasodilatory pathways involve the requirement of a specific GJ protein, Cx43, to regulate ATP-induced eNOS activation and NO production that is Ca²⁺-mediated^14,15. These endothelial actions of ATP are mediated through caveolar microdomains that possess protein machinery for eNOS-NO regulation²¹. We noted that ATP-stimulates rapid Cx43 repartitioning from the noncaveolar to the caveolar domain where eNOS resides and is activated for NO production. Delineating mechanisms establishing the vasodilatory phenotype of pregnancy provides greater understanding of specific mechanisms functioning abnormally in the pathophysiology of vascular diseases such as hypertension and preeclampsia.

Novelty and Significance.

1) What Is New?

• The uterine artery endothelium exhibit a pregnancy-specific vasodilatory phenotype associated with elevations in ATP, PGI₂-cAMP and NOx-cGMP levels.

• ATP-stimulates rapid Cx43 repartitioning from the noncaveolar to the caveolar domain where eNOS resides and is activated for NO production.

2) What Is Relevant?

Pregnancy-specific UAEC elevations of ATP increase GJIC and rapid localization of Cx43 to the caveolae to maintain NO-mediated elevated UBF. Understanding GJ regulation gives more insight of the mechanisms controlling normal UBF during gestation, which may be dysfunctional in preeclampsia.

Summary

These findings demonstrate that the ATP physiological effects in the uterine vasculature during pregnancy involves Cx43 translocation to the caveolar compartment of the endothelial plasma membrane to regulate eNOS activation and NO production to maintain UBF.