Carbonic anhydrase IX proteoglycan-like and intracellular domains mediate pulmonary microvascular endothelial cell repair and angiogenesis

Abstract

The lungs of patients with acute respiratory distress syndrome (ARDS) have hyperpermeable capillaries that must undergo repair in an acidic microenvironment. Pulmonary microvascular endothelial cells (PMVECs) have an acid-resistant phenotype, in part due to carbonic anhydrase IX (CA IX). CA IX also facilitates PMVEC repair by promoting aerobic glycolysis, migration, and network formation. Molecular mechanisms of how CA IX performs such a wide range of functions are unknown. CA IX is composed of four domains known as the proteoglycan-like (PG), catalytic (CA), transmembrane (TM), and intracellular (IC) domains. We hypothesized that the PG and CA domains mediate PMVEC pH homeostasis and repair, and the IC domain regulates aerobic glycolysis and PI3k/Akt signaling. The functions of each CA IX domain were investigated using PMVEC cell lines that express either a full-length CA IX protein or a CA IX protein harboring a domain deletion. We found that the PG domain promotes intracellular pH homeostasis, migration, and network formation. The CA and IC domains mediate Akt activation but negatively regulate aerobic glycolysis. The IC domain also supports migration while inhibiting network formation. Finally, we show that exposure to acidosis suppresses aerobic glycolysis and migration, even though intracellular pH is maintained in PMVECs. Thus, we report that 1) the PG and IC domains mediate PMVEC migration and network formation, 2) the CA and IC domains support PI3K/Akt signaling, and 3) acidosis impairs PMVEC metabolism and migration independent of intracellular pH homeostasis.

INTRODUCTION

The acute respiratory distress syndrome (ARDS) is caused by a diverse range of clinical problems that manifest as noncardiogenic pulmonary edema and hypoxemia (1). Lung protective ventilatory strategies represent the only mortality-improving ARDS management thus far. These approaches reduce mechanical injury to the lungs (2), but they do not directly address ARDS mechanisms. ARDS still causes high mortality, and survivors experience high rates of morbidity and mortality in the aftermath of their initial illness (3). Efforts have been made to decode the clinical-biological axis of ARDS by subphenotyping inflammatory profiles (4), hoping to identify disease-specific therapeutic targets. Recent reports show how different injuries of ARDS form distinct lung microenvironments that vary in inflammation, antioxidant capacity, and acidosis (5, 6). Furthermore, these microenvironments dictate the effectiveness of therapies (5, 6). In an ex vivo human lung study, the acidic microenvironment of lungs with an infection eliminated the therapeutic benefit of mesenchymal stem cell treatment (6). This finding demonstrates that the acidity of local lung tissue can compromise cell-based therapies (6) and underscores the importance of better understanding cellular function within the heterogeneous microenvironment of ARDS lungs.

ARDS lungs have topographically variable glucose uptake, reflecting locally enhanced metabolism (7–10). High glucose uptake directly correlates with low extracellular pH (11), illustrating that acidosis is heterogeneously distributed throughout the lung. The local tissue pH of these injured regions can range from 6.80 to 6.15 (6). To repair leaky capillaries in this highly acidic microenvironment, pulmonary microvascular endothelial cells (PMVECs) require mechanisms to optimize metabolism and intracellular pH to facilitate wound healing. We have previously demonstrated that carbonic anhydrase IX (CA IX) is critical for the acid-resistant phenotype of PMVECs, supporting intracellular pH homeostasis under acidic conditions (12), and its lung tissue expression is significantly increased during infection (13). We also showed that the roles of CA IX extend beyond pH regulation, to mediating aerobic glycolysis, migration, and network formation that are important for PMVEC barrier integrity (14). However, molecular mechanisms of how CA IX influences such a wide range of PMVEC functional characteristics is unclear.

CA IX is composed of four functional domains known as the proteoglycan-like domain (PG), the catalytic domain (CA), the transmembrane domain (TM), and the intracellular domain (IC) (15). The PG-CA domains constitute the ectodomain and are known to promote pH homeostasis and metastasis in cancer cells (15–19). The PG domain is intrinsically disorganized and enriched with negatively charged amino acids (18, 20), whereas the CA domain is home to the catalytic pocket (21). The IC domain, on the other hand, extends into the cytoplasm and acts as an intracellular tail for CA IX (15). Little is known about the IC domain although it has emerging functions in cell signaling (22), protein-protein interaction (17), and is essential for trafficking CA IX to the membrane (13, 23). Here, we investigated the role of the PG, CA, and IC domains in aerobic glycolysis, intracellular pH homeostasis, migration, network formation, and phosphoinositide-3-kinase (PI3K)/Akt signaling using PMVECs that conditionally express CA IX harboring a specific domain deletion (13).

MATERIALS AND METHODS

Isolation of Rat Lung Endothelial Cells

The University of South Alabama Institutional Animal Care and Use Committee approved rat pulmonary endothelial cell isolation protocol. As described previously (24), PMVECs were obtained from Sprague–Dawley rats (CD strain, 350–400 g; Charles River). Cells had positive staining for endothelial markers DiI-LDL and factor VIII. PMVECs selectively bind to lectin Griffonia simplicifolia but not Helix pomatia (25, 26). Isolated cells agglutinated in the presence of G. simplicifolia but failed to form cell clusters when treated with H. pomatia, confirming successful isolation of PMVECs (25, 26).

Generation of CA IX Mutant PMVEC Cell Lines

As previously described, CA IX was knocked out (K/O) from PMVECs with CRISPR-Cas9 (12–14). A retro-lentivirus approach was used to rescue expression of full-length CA IX or CA IX harboring a PG, CA, or IC domain deletion (13) (Fig. 1). These cell lines were also engineered to have a Tet-Off system, allowing conditional knockdown of CA IX or CA IX mutant expression in the presence of doxycycline (13, 27). Domain deletion and the Tet-Off system were confirmed through Western blotting and immunocytochemistry (13). ΔIC cells failed to localize the CA IX mutant protein to the cell membrane, trapping it in the perinuclear region of the cell (13, 23).

Figure 1.

Pulmonary microvascular endothelial cell (PMVEC) mutant cell lines. CRISPR-Cas9 and a retro-lentivirus method was used to generate four PMVEC cell lines expressing either a full-length carbonic anhydrase IX (CA IX) (rWT) or a CA IX protein harboring a specific domain deletion [Δproteoglycan-like (ΔPG), Δcatalytic (ΔCA), and Δintracellular (ΔIC)] (12, 13). In addition, these cell lines were engineered to have a Tet-Off system that allows conditional knockdown of the CA IX protein in the presence of doxycycline (13, 27). Domain deletions and the doxycycline Tet-Off system were verified through Western blotting and immunocytochemistry (13). ΔIC cells did not localize the CA IX mutant protein to the cell membrane (13, 23).

In Vitro Acidosis pH Models

HEPES-buffered media were made from bicarbonate-free DMEM (Thermo Fisher, Grand Island, NY), 10% FCS, and 1% penicillin-streptomycin with 30 mM of HEPES. Media were then titrated with 1 N of HCl to achieve a pH of 7.4 or 6.4. A pH meter (Denver Instrument, Bohemia, NY) was used to measure media pH.

Lactate and Glucose Measurements

Cells were seeded at 4.0 × 10⁵ cells per well in six-well plates and incubated in DMEM, 10% FCS, and 1% penicillin-streptomycin at 37°C ambient air with 5% carbon dioxide for 48 h. Media were then replaced with 7.4 or 6.4 pH bicarbonate-free HEPES-buffered media and placed in 37°C room air with 0% carbon dioxide for 24 h. After incubation, media were collected and glucose and lactate measurements were performed with a YS1 2300 STAT Plus Glucose & Lactate Analyzer (YSI, Yellow Springs, OH).

Intracellular pH Measurements

Cells were seeded at 4.0 × 10⁵ cells per well in six-well plates and incubated in DMEM, 10% FCS, and 1% penicillin-streptomycin at 37°C ambient air with 5% carbon dioxide for 48 h. Media were then replaced with 7.4 or 6.4 pH bicarbonate-free HEPES-buffered media and placed in 37°C room air with 0% carbon dioxide for 24 h. Next, cells were rinsed with HBSS before incubating them in 1 µM BCECF-AM (Thermo Fisher Scientific, Waltham, MA) for 15 min. HBSS rinsing was repeated and a SpectraMax iD5 Multi-Mode Microplate Reader (Molecular Devices, San Jose, CA) with dual excitation (440 and 490 nm) and single emission (535 nm) was used to make ratiometric fluorescent measurements. A calibration curve was made by placing cells in 10 µM nigericin sodium (Adipogen, San Diego, CA) that was dissolved in 20 mM of HEPES-buffered media containing 120 mM KCl, 2 mM CaCl₂·H₂O, 1 mM MgCl₂, 10 mM glucose, and a pH ranging from 6.0 to 8.0.

Scratch Wound Migration Assay

Cells were seeded at 4.0 × 10⁵ cells per well in six-well plates and incubated in DMEM, 10% FCS, and 1% penicillin-streptomycin at 37°C ambient air with 5% carbon dioxide for 48 h. A 200 µL pipette tip was used to create a scratch wound across the cell monolayer. Media were then replaced with 7.4 or 6.4 pH bicarbonate-free HEPES-buffered media and incubated at 37°C room air with 0% carbon dioxide for 24 h. Images were taken of the scratch wound and quantified with ImageJ’s macro language “MRI_Wound_Healing_Tool.”

Matrigel Network Formation Assay

Overnight, Matrigel (Corning Inc., Corning, NY) was placed in 4°C to thaw. During the experiment, 96-well plates were placed on ice and cooled before coating the wells with 30 µL of Matrigel. The Matrigel-coated plate was then incubated at 37°C in room air with 0% carbon dioxide for 1 h. Cells close to 70% confluency in 100-mm dishes were trypsinized and transferred to a centrifuge tube in DMEM, 10% FCS, and 1% penicillin-streptomycin. After centrifugation at 1,500 g for 10 min, media were removed, and 7.4 pH bicarbonate-free HEPES-buffered media were added. Next, 5.0 × 10⁴ cells were added to each Matrigel-coated well and incubated for 24 h at 37°C with 0% carbon dioxide ambient air. Images were taken and analyzed with ImageJ’s macro language “Angiogenesis Analyzer.”

Akt Stimulation and Cell Lysate Collection

Cells were seeded at 1.5 × 10⁵ cells per well in a six-well plate and incubated in DMEM, 10% FCS, and 1% penicillin-streptomycin at 37°C ambient air with 5% carbon dioxide for 5 h. Cells were then rinsed with HBSS and incubated in bicarbonate DMEM with low serum (0.1% FCS) at 37°C ambient air with 5% carbon dioxide overnight. The following morning, media were removed, and cells were further conditioned in FCS-free bicarbonate DMEM for 2 h at 37°C ambient air with 5% carbon dioxide. Conditioned cells were then rinsed with HBSS, and Akt phosphorylation was stimulated by adding 7.4 pH bicarbonate DMEM with 10% FCS at 37°C ambient air with 5% carbon dioxide. After 30 min, cells were immediately placed on ice and cell lysate was collected with radioimmunoprecipitation assay (RIPA) lysis buffer (Boston BioProducts, Ashland, MA) containing protease (Sigma-Aldrich, St. Louis, MO) and phosphatase (Boston BioProducts, Ashland, MA) cocktail inhibitors. Cell lysates were centrifuged at 14,000 rpm for 10 min at 4°C, and protein extracts were collected. The protein concentration of extracts was quantified with a Bradford protein assay (Thermo Fisher Scientific, Waltham, MA).

Western Blots for Akt Phosphorylation Measurements

Protein extracts were reduced with 1.7% SDS (Boston BioProducts, Ashland, MA) and boiled for 10 min at 70°C. An equal amount of total protein was loaded into a precast SDS-PAGE 4%–12% gradient Bis-Tris gel (Thermo Fisher Scientific, Waltham, MA). Gels were run at 130 V on ice. Proteins were transferred to a nitrocellulose membrane (Bio-Rad, Hercules, CA) at 30 V for 2 h on ice. Nitrocellulose membranes were blocked for 1 h at room temperature with 5% nonfat dry milk in TBST (Tris-buffered saline, 0.1% Tween) (Bio-Rad, Hercules, CA). Primary antibodies phospho-Akt (Ser473), Pan-Akt, and GAPDH antibodies (Cell Signaling Technology, Danvers, MA) were used at a 1:1,000 dilution overnight at 4°C. The following morning, nitrocellulose membranes were rinsed with TBST and incubated with a goat anti-rabbit antibody (Abcam, Cambridge, UK) at a 1:4,000 dilution for 1 h at 4°C. Finally, blots were visualized with a SuperSignal West Femto Maximum Sensitivity Substrate kit (Thermo Fisher Scientific, Waltham, MA). Band intensities were measured using ImageJ software.

Statistics

One-way ANOVA and a Tukey post hoc test were used to assess statistical differences between groups. Significance was set at P < 0.05.

RESULTS

The CA and IC Domains Negatively Regulate Aerobic Glycolysis in PMVECs

Endothelial cells utilize aerobic glycolysis as their primary bioenergetic pathway (28–30). We have shown CA IX promotes aerobic glycolysis in PMVECs, partially by increasing the expression of hexokinase I (HXK 1) and lactate dehydrogenase A (LDHA) (14, 31). To identify the CA IX domains that regulate aerobic glycolysis, we measured glucose and lactate in rescued wild-type (rWT), ΔPG, ΔCA, and ΔIC cell lines after incubation in 7.4 and 6.4 pH media for 24 h (Fig. 2, A and B). Under 7.4 pH media, ΔPG cells did not have higher media lactate but did show decreased media glucose (Fig. 2B; rWT 7.4 pH vs. ΔPG 7.4 pH, P < 0.05). ΔCA cells had increased media lactate production and a trend toward increasing glucose consumption compared with the rWT (Fig. 2, A and B; rWT 7.4 pH vs. ΔCA 7.4 pH, P < 0.05 and P < 0.18, respectively). Similarly, the ΔIC cells showed increased media lactate production and glucose consumption (Fig. 2, A and B; rWT 7.4 pH vs. ΔIC 7.4 pH, P < 0.05). Higher lactate production in ΔCA and ΔIC cells was unexpected since full CA IX K/O cells have reduced aerobic glycolysis (14). We assessed whether this discordance was consistent when directly compared with CA IX K/O cells. ΔCA and ΔIC cells had significantly higher lactate production in contrast to CA IX K/O cells (Supplemental Fig. S1; see https://doi.org/10.6084/m9.figshare.19847512), showing that single domain deficiencies have an opposing effect on aerobic glycolysis than full CA IX K/O. In line with our previous report (14), acidosis suppressed aerobic glycolysis (Fig. 2, A and B; rWT 7.4 pH vs. rWT, ΔPG, ΔCA, and ΔIC 6.4 pH, P < 0.05), but no significant difference in lactate production and glucose consumption was detected between cell lines in 6.4 pH conditions (Fig. 2, A and B; rWT 6.4 pH vs. ΔPG, ΔCA, and ΔIC 6.4 pH, P = ns). Thus, these findings suggest that the CA and IC domains negatively regulate aerobic glycolysis in PMVECs under physiological pH conditions.

Figure 2.

The catalytic (CA) and intracellular (IC) domains negatively regulate aerobic glycolysis. rWT, Δproteoglycan-like (ΔPG), ΔCA, and ΔIC cell lines were seeded at 4.0 × 10⁵ cells per well in six-well plates and grown for 2 days under bicarbonate-buffered media. Media were replaced with 7.4 or 6.4 pH bicarbonate-free HEPES-buffered media and glucose-lactate measurements were made 24 h later. A: media lactate measurements. B: media glucose measurements. ΔPG cells had reduced media glucose but no difference in media lactate. ΔCA cells showed increased lactate production and a trend toward higher glucose consumption in 7.4 pH media. ΔIC cells also had elevated lactate production and glucose consumption in 7.4 pH media. Acidosis decreased media lactate and increased media glucose. No difference was seen between cell lines under 6.4 pH media. Data represent means ± SD. One-way ANOVA and a Tukey’s post hoc test were used to compare groups. *Significant difference (P < 0.05) from 7.4 pH rWT.

The PG Domain Supports Intracellular pH Homeostasis in PMVECs

Highly glycolytic cells require strong pH regulation to handle proton byproducts (28). PMVECs exhibit high acid resistance, in part, from expressing CA IX (12). Although CA IX is important for pH homeostasis (12), the domains that regulate intracellular pH in PMVECs are not known. To investigate this, intracellular pH was measured using the pH sensitive fluorescent dye BCECF-AM in rWT, ΔPG, ΔCA, and ΔIC PMVECs after 24 h of incubation in 7.4 or 6.4 pH media. ΔPG cells had lower intracellular pH than the rWT cells (Fig. 3; rWT 7.4 pH vs. ΔPG 7.4 pH, P < 0.05), but no other cell line showed a significant change in their intracellular pH in 7.4 pH media (Fig. 3; rWT 7.4 pH vs. ΔCA and ΔIC 7.4 pH, P < 0.05). Acidosis had no effect on intracellular pH (Fig. 3; rWT 7.4 pH vs. rWT, ΔPG, ΔCA, and ΔIC 6.4 pH, ΔPG 7.4 pH vs. ΔPG 6.4 pH, P = ns). These results indicate the PG domain supports intracellular pH homeostasis in physiological pH conditions and confirms the remarkable capacity of PMVECs to sustain intracellular pH against an acidic environment.

Figure 3.

The proteoglycan-like (PG) domain supports intracellular pH homeostasis. rWT, ΔPG, Δcatalytic (ΔCA), and Δintracellular (ΔIC) cell lines were seeded at 4.0 × 10⁵ cells per well in six-well plates and grown for 2 days under bicarbonate-buffered media. Media were replaced with 7.4 or 6.4 pH bicarbonate-free HEPES-buffered media for 24 h before measuring intracellular pH with BCECF-AM. ΔPG cells had lower intracellular pH under 7.4 pH media, but no drop in intracellular pH was seen in rWT, ΔCA, and ΔIC cells. No difference in intracellular pH was seen between cell lines during acidosis conditions. Data represent means ± SD. One-way ANOVA and a Tukey’s post hoc test were used to compare groups. *Significant difference (P < 0.05) from 7.4 pH rWT.

The PG and IC Domains Mediate PMVEC Migration

We previously showed CA IX as a critical protein in PMVEC migration and wound healing (14). To identify specific domains responsible for PMVEC migration capacity, a scratch wound assay was performed on rWT, ΔPG, ΔCA, and ΔIC cells under 7.4 and 6.4 pH media (Fig. 4, A–D). Twenty-four hours after injury, ΔPG and ΔIC cells had larger wound areas (Fig. 4, C and D; rWT 7.4 pH vs. ΔPG and ΔIC 7.4 pH, P < 0.05), but there was no difference in wound size in ΔCA cells (Fig. 4, C and D; rWT 7.4 pH vs. ΔCA 7.4 pH, P = ns) compared with rWT cells under 7.4 pH media. Acidosis suppressed wound healing (Fig. 4, C and D; rWT 7.4 pH vs. rWT, ΔPG, ΔCA, and ΔIC 6.4 pH, P < 0.05), but no difference was seen between cell lines in 6.4 pH media (Fig. 4, C and D; rWT 6.4 pH vs. ΔPG, ΔCA, and ΔIC 6.4 pH, P = ns). The study suggests that the PG and IC domains mediate PMVEC migration and acidosis suppresses PMVEC repair after injury.

Figure 4.

The proteoglycan-like (PG) and intracellular (IC) domains mediate pulmonary microvascular endothelial cell (PMVEC) migration. rWT, ΔPG, ΔCA, and ΔIC cell lines were seeded at 4.0 × 10⁵ cells per well in six-well plates and grown for 2 days under bicarbonate-buffered media. A pipette tip was used to induce a scratch wound across the cell monolayer. Media were replaced with 7.4 or 6.4 pH bicarbonate-free HEPES-buffered media, and 24 h later, wound size was measured using ImageJ software. A: images of the baseline scratch wound. B: quantification of the baseline scratch wound size. Cell lines had similar baseline scratch wound sizes. C: images of the scratch wound 24 h after the injury in 7.4 or 6.4 pH media. D: quantification of the wound size 24 h after the initial scratch in 7.4 or 6.4 pH. ΔPG and ΔIC cells had larger wound sizes than the rWT cells under 7.4 pH media. Acidosis suppressed PMVEC wound healing although no difference was detected between cell lines under 6.4 pH media. Data represent means ± SD. One-way ANOVA and a Tukey’s post hoc test were used to compare groups. *Significant difference (P < 0.05) from 7.4 pH rWT.

The PG Domain Promotes While the IC Domain Inhibits Network Formation in PMVECs

Angiogenesis is vital to pulmonary vascular repair (32, 33). To determine whether any CA IX domains have a functional role in PMVEC network formation, Matrigel assays were conducted in 7.4 pH media for 24 h. ΔPG cells failed to form networks, showing a profound decrease in the number of nodes, junctions, total segment length, master junctions, and branches compared with the rWT cells (Fig. 5, A–F; rWT vs. ΔPG, P < 0.05). Interestingly, a time-lapse video revealed ΔPG cells aggregate into cell clusters but do not stretch into angiogenic branches (Supplemental Video S3; see all Supplemental Videos S1–S6 at https://doi.org/10.6084/m9.figshare.15138030), indicating that failure to form networks is not due to the migration defect in ΔPG cells. Prior studies show network formation in CA IX K/O cells that also lack the PG domain (12, 14). We tested whether this phenotypic discrepancy was consistent when ΔPG cells were directly compared with CA IX K/O cells. As previously reported, CA IX K/O cells formed networks but ΔPG cells were unable to branch into networks (Supplemental Fig. S2; see https://doi.org/10.6084/m9.figshare.19857634). No difference in network formation was found between the rWT and ΔCA cells (Fig. 5, A–F; rWT vs. ΔCA, P = ns). In contrast with ΔPG cells, ΔIC cells had enhanced network formation, displaying increased nodes, junctions, total segment lengths, and master junctions (Fig. 5, A–E; rWT vs. ΔIC, P < 0.05). Together, these results indicate that the PG domain is a critical mediator of network formation whereas the IC domain inhibits network formation.

Figure 5.

The proteoglycan-like (PG) domain promotes whereas the intracellular (IC) domain inhibits network formation. Ninety-six-well plates were coated with 30 µL of Matrigel per well and incubated in 37°C ambient air for 1 h. rWT, ΔPG, Δcatalytic (CA), and ΔIC cell lines were then seeded at 5.0 × 10⁴ cells per well with 7.4 pH bicarbonate-free HEPES-buffered media. Angiogenic parameters were measured 24 h later using ImageJ software. A: Matrigel network images of rWT, ΔPG, ΔCA, and ΔIC cell lines. B–F: number of nodes (B), junctions (C), total segment length (D), master junctions (E), and branches (F) formed by each cell line. ΔPG cells were unable to form networks, showing fewer nodes, junctions, total segment length, master junctions, and branches. No difference in Matrigel networks was detected between ΔCA and rWT cells. ΔIC cells had increased network formation, showing more nodes, junctions, total segment length, and master junctions than rWT cells. Data represent means ± SD. One-way ANOVA and a Tukey’s post hoc test were used to compare groups. *Significant difference (P < 0.05) from rWT.

The CA and IC Domains Mediate Akt Phosphorylation in PMVECs

CA IX was previously shown to enhance activation of the PI3K/Akt signaling pathway (22), a critical regulator of aerobic glycolysis, cell migration, and angiogenesis (21). Therefore, we investigated which domains of CA IX mediate the phosphorylation and activation of Akt. To stimulate Akt phosphorylation, subconfluent cells were serum-starved overnight and then exposed to serum media. Immunoblot analysis of cell lysate revealed ΔCA and ΔIC cells have reduced phospho-Akt (Ser473) compared with rWT cells (Fig. 6, A and B; rWT vs. ΔCA and ΔIC, P < 0.05). No difference in Pan-Akt and GAPDH expression was seen between cell lines (Fig. 6A). Collectively, these results suggest that the CA and IC domains mediate the phosphorylation and activation of the PI3K/Akt signaling pathway.

Figure 6.

The catalytic (CA) and intracellular (IC) domains mediate Akt phosphorylation in pulmonary microvascular endothelial cells (PMVECs). Cells were seeded at 1.5 × 10⁵ cells per well in a six-well plate and incubated in bicarbonate DMEM with low serum (0.1% FCS) overnight. The next morning, cells were further incubated in FCS-free bicarbonate DMEM for 2 h. After conditioning, cells were rinsed with HBSS and 7.4 pH bicarbonate DMEM with 10% FCS was added for 30 min. Whole cell lysate was then collected and subjected to immunoblotting for Phospho-Akt (Ser473), Pan-Akt, and GAPDH. Western blots were analyzed using ImageJ software. A: Western blot bands for phospho-Akt (Ser473), Pan-Akt, and GAPDH from rWT, Δproteoglycan-like (ΔPG), ΔCA, and ΔIC cell lines. B: the ratiometric measurements of phospho-Akt (Ser473) to Pan-Akt band intensities. ΔCA and ΔIC cells showed lower phospho-Akt (Ser473) compared with rWT during serum stimulation. No difference in Pan-Akt and GAPDH expression was seen between cell lines. Data represent means ± SD. One-way ANOVA and a Tukey’s post hoc test were used to compare groups. *Significant difference (P < 0.05) from rWT.

DISCUSSION

Acute noncardiogenic pulmonary edema is a key clinical feature of ARDS (32, 34, 35). Surviving patients with ARDS exhibit a long-term reduction in gas exchange and exercise capacity, and they have a worse quality of life (32, 36–40). The mechanisms involved in acute alveolar-capillary barrier dysfunction and its progression to chronically impaired healing are not known. A better understanding of how PMVECs repair wounds in the highly glycolytic and acidic microenvironment of ARDS lung tissue may offer insight into the impact that lung-protective strategies have on ARDS progression.

We have previously shown that CA IX is critical to PMVEC acid resistance and repair (12, 14). In this study, we identified the specific CA IX domains regulating PMVEC aerobic glycolysis, intracellular pH homeostasis, cell migration, network formation, and Akt signaling: 1) the PG domain mediates intracellular pH homeostasis, cell migration, and network formation; 2) the CA domain inhibits aerobic glycolysis but promotes Akt signaling; 3) the IC domain inhibits aerobic glycolysis and network formation while supporting cell migration and Akt signaling; and 4) acidosis suppresses PMVEC aerobic glycolysis and repair independent of intracellular pH homeostasis. These results are summarized in Fig. 7.

Figure 7.

The domain specific functions of carbonic anhydrase IX (CA IX) in pulmonary microvascular endothelial cell (PMVEC) integrity and repair. CA IX is critical to PMVEC repair, but the specific function of each CA IX domain is unknown. CA IX is composed of four domains known as the proteoglycan-like domain (PG), catalytic domain (CA), transmembrane domain (TM), and the intracellular domain (IC). The PG domain mediates intracellular pH homeostasis, cell migration, and network formation. The CA and IC domains promote the PI3k/Akt signaling pathway, but negatively regulate aerobic glycolysis. In addition, the IC domain inhibits network formation while supporting cell migration. Acidosis suppresses aerobic glycolysis and migration independent of intracellular pH homeostasis.

In line with several other studies (18, 19, 41), our results found that the PG domain preserves intracellular pH in PMVECs. At the cell membrane, CA IX complexes with multiple acid-base transporters, including bicarbonate (16, 17, 42–44) and monocarboxylate transporters (MCTs) (18, 19, 41). The MCT is a symporter that removes lactate and protons from highly glycolytic cells, conserving intracellular pH (18, 19, 41). The PG domain contains 26 negatively charged amino acids that are suggested to attract protons as they move through the MCT, acting as a “proton antenna” that increases proton flux out of the cell (18). Decreased intracellular pH and migration in ΔPG cells may, in part, be due to the absence of PG domain’s function to facilitate proton flux out of the cell to optimize intracellular pH, which may be important for proper actin trailing.

ΔPG cells migrate in Matrigel to form cell clusters, but they do not stretch out to generate thin branches to make typical capillary-like networks as wild-type and other mutant PMVECs do. Such a cell clustering Matrigel pattern has been similarly shown in our previous study when we treated PMVECs with 2-deoxy-glucose and observed a concomitant increase in N-cadherin expression (14). Multiple reports demonstrated the interactions between CA IX and other junctional proteins such as β-catenin, α2/β1 integrin, and CD98c (17, 45), and Rho-GTPase signaling (46), suggesting that the PG domain may mediate network formation via its interactions with junctional and cytoskeletal proteins. However, it is puzzling that the total CA IX K/O PMVECs that also lack the PG domain can still form normal networks in physiological pH (12, 14). Since the PG domain uniquely mediates intracellular pH homeostasis distinctive from all other domains and the total CA IX K/O PMVEC are able to form capillary-like network under physiological pH, intracellular pH may be a critical determinant of Matrigel network forming pattern. These findings indicate complex interplay between individual CA IX functional domains and their microenvironment.

Although traditionally known as a pH regulator, CA IX has an emerging role in its direct involvement in metabolic pathway regulation. Previous reports have shown that CA IX promotes aerobic glycolysis and expression of key glycolytic enzymes, including HXK1 and LDHA (14, 31). However, in the current study, we demonstrate that, counterintuitively, the CA and IC domains inhibit aerobic glycolysis. Similar opposing effects of full-length CA IX compared with the individual domain on metabolism has been observed in our previous study where we found the CA IX inhibitor SLC-0111 enhances aerobic glycolysis in PMVECs (14, 47). These findings may suggest that each CA IX functional domain possess distinctive roles and their collective balance determines full-length CA IX function. However, potential confounders coming from limitations of genetic modification techniques such as variable mutant protein expression and inadvertent selection of certain phenotype during colony isolation need to be considered. Further rigorous replicating studies using multiple clones may help answer these questions.

The IC domain promotes migration but inhibits network formation. These uncoupled effects of the IC domain on migration and network formation are notable particularly compared with the PG domain, which promotes both functions. We considered whether the effects of the IC domain were due to downstream changes in the PI3K/Akt signaling pathway, a major modulator of aerobic glycolysis, cell migration, and angiogenesis (21). To our surprise, we found both the CA and IC domains mediate the phosphorylation and activation of Akt. The impairment of migration and Akt activation yet enhanced network formation in the ΔIC cells implies that the IC domain may regulate other angiogenic signaling pathways and that these pathways compensate for loss of IC domain signaling in response to angiogenic cues. Characterizing the role of the IC domain in other signaling pathways may provide mechanistic insight into how it modulates network formation.

The IC domain has previously been shown to interact with PI3K through a phosphorylated tyrosine residue, leading to the activation of Akt (22). To our knowledge, however, this is the first time the CA domain has been incriminated in cell signaling. CA IX is structurally homologous to receptor protein tyrosine phosphatases-γ (RPTPγ) (48), a carbon dioxide/bicarbonate sensor in renal proximal tubule epithelial cells (49). Whether CA IX can act as a similar pH sensor is unknown. One possibility is that the catalytic activity of the CA domain “senses” changes in carbon dioxide/bicarbonate and induces a conformational shift in the IC domain, adapting intracellular signals in response to extracellular pH changes. Alternatively, the CA domain could be necessary for the interaction with transmembrane kinases/phosphatases for IC domain phosphorylation, or genetic deletion of the CA domain locks the IC domain in an inactive conformation. A point mutation in the CA domain that eliminates the catalytic activity but preserves CA IX structure and membrane localization may help determine whether CA IX signaling is dependent on its catalytic activity. Considering the vital role of the IC domain in PMVEC survival during serum starvation and infection (13), further investigation into the physiological role of CA-IC domain signaling may help identify potential molecular targets for ARDS therapy.

Previously, we showed that CA IX is a critical mediator of PMVEC wound healing during acidosis (12, 14). No difference in repair and metabolism, however, were found in PMVECs expressing either full-length CA IX or CA IX with a single domain deletion under acidic conditions. These results suggest that deletion of a single domain is not enough to cause a repair deficiency in the setting of acidosis and that the domains of CA IX collectively contribute to PMVEC repair during acid-base disturbances. Studies with CA IX K/O mice will help better determine the importance of CA IX in acid handling and repair in the lung.

Finally, we demonstrated that extrinsic acidosis suppresses PMVEC aerobic glycolysis and migration even though intracellular pH is maintained, showing that individual functional characteristics of PMVECs are not necessarily interdependent. These findings may be explained through three potential mechanisms. First, the remarkable capacity of PMVECs to stabilize global intracellular pH is likely due to the expression of several different types of pH regulatory proteins, including CA IX, Na⁺/H⁺-exchangers, HCO₃⁻ transporters, and V-H⁺-ATPase (12, 14, 50–52). The complementary roles of these pH regulatory mediators cumulatively determine the acid-resistant phenotype of PMVECs, and deletion of a single pH mediator may not be sufficient to compromise global intracellular pH during acidosis. Second, the decoupling of metabolism and migration with intracellular pH indicates PMVECs may express pH receptors that negatively regulate reparative processes within the cell in response to external pH changes. Third, while no change in global intracellular was seen, it is possible that pH shifts occur within compartmentalized regions of the cell (i.e., the cell membrane) over time. Higher-resolution approaches may help connect intracellular pH dynamics to metabolism and cell migration during acidosis.

In summary, we report that the PG domain promotes intracellular pH homeostasis, migration, and network formation. The CA and IC domains mediate PI3K/Akt signaling but negatively regulate aerobic glycolysis. The IC domain supports migration while inhibiting network formation. Acidosis suppresses PMVEC aerobic glycolysis and migration independent of global intracellular pH homeostasis (Fig. 7). The findings suggest CA IX as an important multifunctional mediator of PMVEC integrity and repair.

DATA AVAILABILITY

The data that support the findings of this study will be made available upon reasonable request from the corresponding author.

SUPPLEMENTAL DATA

Supplemental Fig. S1: https://doi.org/10.6084/m9.figshare.19847512.

Supplemental Fig. S2: https://doi.org/10.6084/m9.figshare.19857634.

Supplemental Videos S1–S6: https://doi.org/10.6084/m9.figshare.15138030.

GRANTS

This study was supported by American Heart Association Grant 18CDA34080151 (to J. Y. Lee and T. Stevens) and the National Institutes of Health Grants HL153958 (to J. Y. Lee and T. Stevens), OD010944 (to M. F. Alexeyev), S10OD025089 (to M. F. Alexeyev), HL148069 (to J. Y. Lee and T. Stevens), HL66299 (to T. Stevens), HL60024 (to T. Stevens), and HL160988 (to J. Y. Lee, M. F. Alexeyev, and T. Stevens).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.P.S., T.S., and J.Y.L. conceived and designed research; R.P.S., M.F.A., N.K., V.P., S.S.P., J.B., and D.T.T. performed experiments; R.P.S., M.F.A., and J.Y.L. analyzed data; R.P.S., T.S., and J.Y.L. interpreted results of experiments; R.P.S. prepared figures; R.P.S. drafted manuscript; R.P.S., T.S., and J.Y.L. edited and revised manuscript; R.P.S. and J.Y.L. approved final version of manuscript.