Endothelin-1 receptor antagonists regulate cell surface-associated protein disulfide isomerase in sickle cell disease

Abstract

Increased endothelin-1 (ET-1) levels, disordered thiol protein status, and erythrocyte hydration status play important roles in sickle cell disease (SCD) through unresolved mechanisms. Protein disulfide isomerase (PDI) is an oxidoreductase that mediates thiol/disulfide interchange reactions. We provide evidence that PDI is present in human and mouse erythrocyte membranes and that selective blockade with monoclonal antibodies against PDI leads to reduced Gardos channel activity (1.6±0.03 to 0.56±0.02 mmol·10¹³ cell⁻¹·min⁻¹, P<0.001) and density of sickle erythrocytes (D₅₀: 1.115±0.001 to 1.104±0.001 g/ml, P=0.012) with an IC₅₀ of 4 ng/ml. We observed that erythrocyte associated-PDI activity was increased in the presence of ET-1 (3.1±0.2 to 5.6±0.4%, P<0.0001) through a mechanism that includes casein kinase II. Consistent with these results, in vivo treatment of BERK sickle transgenic mice with ET-1 receptor antagonists lowered circulating and erythrocyte associated-PDI activity (7.1±0.3 to 5.2±0.2%, P<0.0001) while improving hematological parameters and Gardos channel activity. Thus, our results suggest that PDI is a novel target in SCD that regulates erythrocyte volume and oxidative stress and may contribute to cellular adhesion and endothelial activation leading to vasoocclusion as observed in SCD.—Prado, G. N., Romero, J. R., Rivera, A. Endothelin-1 receptor antagonists regulate cell surface-associated protein disulfide isomerase in sickle cell disease.

Clinical manifestations of sickle cell disease (SCD) are based in part on intravascular erythrocyte sickling within capillaries and small vessels leading to vasoocclusion and impaired blood flow. In SCD, erythrocyte hydration status and density are associated with erythrocyte sickling, organ damage, hemolysis (1) and increased viscosity (2). Increases in intravascular hemolysis, ischemia-reperfusion damage, and chronic proinflammatory state are associated with an imbalance between oxidant and antioxidant availability that contributes to increased oxidative stress and tissue damage (3, 4). There is evidence that the increased oxidative state observed in SCD is further exacerbated by polymerization/depolymerization of hemoglobin S (5) and is associated with increases in cytokine expression such as endothelin-1 (ET-1; refs. 6, 7) and numerous adhesion molecules (8, 9) contributing to a vicious cycle of increased vasoocclusive episodes and a proinflammatory state in SCD. Indeed, we have proposed an important role for ET-1 and potassium intermediate/small conductance calcium-activated channel, subfamily N, member 4 (KCNN4; Gardos channel) activity, an important contributor to dense cell formation, in SCD (10).

ET-1 is a potent vasoconstrictor that plays an important proinflammatory role in SCD (6, 7) and contributes to oxidative stress (11). We reported that ET-1 receptor antagonists (ETRAs) in vivo blocked dense cell formation and Gardos channel activity while improving hematological parameters in SAD sickle cell transgenic mice (10); results that were confirmed more recently by others (12) through mechanisms that are still not entirely clear. Elevated ET-1 levels contribute to increased oxidation of sickle red blood cells (sRBCs) that, in turn, may be responsible for altered cytoskeletal protein function contributing to membrane rigidity and altered cation transport in these cells (13, 14). However, few studies have evaluated the role of oxidation on Gardos channel regulation.

Glutathione (GSH) is a major intracellular antioxidant that protects cells against oxidative stress. Gardos channel activity increases following GSH depletion in intact sickle and normal erythrocytes (15). In addition, lowering GSH (thereby increasing cellular oxidation state) is followed by induction of cellular dehydration (16). However, the mechanism by which the redox state affects plasma membrane protein activities in sRBC remains unclear, in part due to the scarcity of information on the redox enzymes that participate in the pathophysiology of SCD and the very limited studies available in erythrocytes. In nucleated cells, addition of dithiobis-2-nitrobenzoic acid (DTNB), an impermeant oxidizing agent, reduced by 97% the voltage-independent intermediate K⁺ channel activity in inside-out membrane preparations from bovine aortic endothelial cells that was reversibly activated by GSH replenishment (17). Recently, the impermeant oxidizing agent, pCMBS, was shown to bind to the KCNN4 pore region, leading to increased open state probability of inside-out patch-clamp preparations of KCNN4 transfected into human kidney cells (18). Furthermore, inside-out patch-clamp experiments with DTNB show reduced KCNN4 activity that was partially restored by addition of dithiothreitol (DTT) or GSH (17). These data suggest that thiol/disulfide interactions may regulate Gardos channel activity in sRBCs.

Protein disulfide isomerase (PDI) is a ubiquitously expressed oxidoreductase present in the plasma membrane and endoplasmic reticulum that mediates thiol/disulfide interchange reactions at the cell surface of many cell types (19). PDI, a member of the thioredoxin superfamily, is a multifunctional 57-kDa enzyme that provides essential chaperone activities and can function as an isomerase or reductase depending on the ambient reducing potential (20). The enzyme contains two active sites with two vicinal cysteines (CXXC). In the oxidized state, PDI interacts with two cysteinyl thiol groups of an adjoining protein to form a disulfide bond. In the reduced state, it cleaves neighboring protein disulfide bonds. There is evidence that PDI is located at the surface of the plasma membranes of lymphocytes (21), pancreatic cells (22), and hepatocytes (23), where it has been shown to be involved in the assembly and function of pore-forming proteins and receptors linked to protein transport processes. Other surface associated functions of PDI include its interaction with the diphtheria toxin and the thyrotropin receptors (24, 25), binding to extracellular thrombospondin (26), as well as its capacity to bind nitric oxide for transport into megakaryocytes and maturation of platelets (27). More recently, PDI has been demonstrated to play an important role in thrombus formation (28) and cellular adhesion to the endothelium (29). However, its potential role in SCD has not been described.

In this study, we investigated the role of the erythrocyte cell surface-associated PDI (sa-PDI) on Gardos channel function and sRBC dehydration in SCD. We hypothesized that PDI mediates the effects of ET-1-induced Gardos channel activity and cellular dehydration in SCD. We provide evidence that PDI is present in the plasma membrane of erythrocytes and that its blockade modifies the cellular hydration status of sRBCs. Furthermore, we show that ETRAs in vivo can regulate PDI activity and improve hematological parameters in a sickle cell transgenic mouse model of severe pathology.

MATERIALS AND METHODS

Drugs and chemicals

The A23187 was purchased from Calbiochem (La Jolla, CA, USA). ⁸⁶Rb and ¹²⁵I were purchased from PerkinElmer Life Sciences (Boston, MA, USA). PDI antibody (monoclonal RL90) was from AbCam (Cambridge, MA, USA) and β-actin antibody from Cell Signaling (Danvers, MA, USA). All other reagents were from Sigma-Aldrich (St. Louis, MO, USA).

Blood samples

Human blood samples were collected after signed informed consent, following approval by Boston Children's Hospital Institutional Review Board, and compliance with U.S. Health Insurance Portability and Accountability Act (HIPAA) regulations.

Animals

We used Berkeley (BERK) sickle cell transgenic mice on a mixed genetic background (The Jackson Laboratories, Bar Harbor, ME, USA). BERK mice have a transgene containing normal human α-, γ-, and δ-globins and sickle β-globin and targeted deletions of murine α- and β-globins (α^−/−, β^−/−, Tg). Our mouse colony was generated by breeding α^−/−, β^−/−, Tg males with α^−/−, β^+/−, Tg females. Three- to 6-mo-old male and/or female BERK and BERK-trait mice (homozygous for the α knockout, hemizygous for the β knockout and BERK transgene) were used. BERK mice have severe disease that simulates human sickle cell anemia (hemolysis, reticulocytosis, anemia, extensive organ damage, and shortened life span) and have high levels of oxidative stress (30). Transgenic mice expressing exclusively human hemoglobin A and knockout mouse globins (HbAKO: Hba⁰//Hba⁰: Hbb⁰//Hbb⁰) were used (31). The β_S-Antilles (Hba⁺/Hba⁺//Hbb⁰/Hbb⁰) transgenic mice were kindly provided by Dr. Mary Fabry (Albert Einstein College of Medicine, Bronx, NY, USA). We also used 4- to 6-mo-old SAD transgenic male mice on C57BL/6J background (32) that were kindly provided by Dr. Seth Alper (Beth Israel Deaconess Medical Center, Boston, MA, USA). SAD mice carried the human β_S (β6Val), β_S-Antilles (β23Ile), and D-Punjab (β121Glu) globin β-chain transgene. All procedures for study, animal care, and euthanasia followed approval by the Boston Children's Hospital Animal Care and Use Committees.

In vivo studies

We studied BERK mice that were placed on an ETRA regime for 14 d essentially as previously reported by us in SAD mice (10). Briefly, mice were intraperitoneally injected for 14 consecutive days and received either sterile mouse saline (0.1 ml) or 0.1 ml of an ETRA mixture that consisted of selective ET-1 antagonist subtype A (BQ123; 0.2 mg/ml) and selective ET-1 antagonist subtype B (BQ788; 0.2 mg/ml) dissolved into 1 ml mouse saline. Animals were fed standard mouse chow and given water ad libitum during treatment. Mice were then euthanized, and whole blood was immediately collected into heparinized or EDTA tubes for further experimentation. Plasma concentration of ETRA in mice was determined as a function of time after administration and is shown in Supplemental Figs. S1 and S2.

Hematological parameters

Blood cell counts were determined using the ADVIA analyzer (Bayer Diagnostics, Tarrytown, NY, USA) and a software program specific for either mouse or human EDTA-collected blood as described previously (33). Mouse plasma samples were isolated for cytokine determinations by ELISA (R&D Systems, Minneapolis, MN, USA).

Measurement of whole-blood Gardos channel activity

Freshly isolated mouse blood was used to determine Gardos channel activity as described in detail previously (10, 34). Briefly, K⁺ fluxes were followed for 5 min to avoid any potential experimental artifacts, and fluxes were determined from the slope of the K⁺ content (⁸⁶Rb) vs. time and expressed as mmol · 10¹³ cell⁻¹ · min⁻¹.

Measurement of Gardos channel

Freshly isolated erythrocytes were used to determine Gardos channel activity as described in detail previously by us (10, 34, 35) and others (36, 37). Cells were preincubated with or without permeant DTT (10 mM), impermeant DTNB (0.1 mM), phenylarsine oxide (PAO; 0.1 mM), impermeant bacitracin (3 mM) or anti-PDI antibodies (20 ng/ml; RL90, ab2792; Abcam, Cambridge, MA, USA) for 15 min. Preincubations with ET-1 were carried out for 20 min at 37°C in an isotonic sodium medium.

Western blotting and coimmunoprecipitation

Plasma membranes from sickle and normal erythrocytes were prepared and solubilized in buffer containing protease inhibitors as described previously (35). Protein concentrations were determined by Pierce BCA Protein Assay (Pierce; Thermo Fisher Scientific Inc., Rockford, IL, USA; ref. 38). Proteins were resolved on 10% SDS-PAGE (Bio-Rad, Richmond, CA, USA) and detected using primary antibodies: mouse monoclonal antibody against PDI (mAb PDI), rabbit anti-KCNN4, or mouse monoclonal anti-β-actin, and visualized with ECL (GE Healthcare, Piscataway, NJ, USA). For coimmunoprecipitation, 400 μg of precleared, solubilized membranes was incubated with or without the precipitating antibody with 1 μg of rabbit anti-KCCN4 or mouse anti-PDI at 4°C with rotation overnight. Protein G agarose was then added to the antibody-preparation mixture and incubated at 4°C for 3 h, followed by 3 washes in RIPA buffer. Immunoprecipitates were eluted by adding 100 μl of 1× Laemmli SDS sample buffer and loaded onto a 10% SDS-PAGE. Coimmunoprecipitated proteins were detected by anti-PDI or anti-KCNN4. Immunoglobulin G (IgG) was used as control for specificity of the reactions.

Quantitative RT-PCR

Early cultures of mouse aortic endothelial cells from BERK mice were prepared as described previously (38). RNA from sickle cell mouse aortic endothelial cells was prepared using Trizol and reversed transcribed using the Superscript II kit (Invitrogen, Carlsbad, CA, USA), as recently described (38). Real-time PCR was performed using TaqMan Universal PCR mastermix and preformulated primers for PDI and the endogenous controls β-2-microglobulin and GAPDH (Applied Biosystems, Foster City, CA, USA). The cycle threshold (C_t) values for PDI expression were normalized against two housekeeping genes, β-2-microglobulin and GAPDH. Difference in PDI expression between ETRA treated and vehicle treated sickle mice were analyzed according to the 2^−ΔΔCt method.

PDI assays

Radiolabeled Tyr-S-S-PDL assay

sa-PDI activity in intact erythrocytes was measured using ¹²⁵I-iodotyramine linked to poly-d-lysine (¹²⁵I-Tyr-S-S-PDL conjugate) synthesized as described previously by us (19) and others (39). Briefly, radiolabeled erythrocytes were washed and incubated at 37°C in the presence or absence of either bacitracin (1 mM) or mAb PDI (20 ng/ml). The PDI-specific reductive activity was calculated from the difference between the percentage of acid soluble radioactivity in the presence or absence of mAb PDI.

Fluorescence PDI assay

A circulating PDI activity assay was developed following previously described protocols (40). Briefly, [E-GSH] formation is recorded for 30 min (λ_ex: 525 nm; λ_em: 545 nm) at room temperature, and fluorescence intensity is recorded and plotted as a function of time to calculate the conjugate's reduction rate. Measurements were performed in the presence or absence of 20 ng/ml mAb PDI and nonspecific IgG for specificity and quality control. Plasma PDI activity was determined by fluorescence using [abz-GSH] formation at low hemoglobin concentration (λ_ex: 312 nm; λ_em: 415 nm) as described previously, with modifications (41).

Phthalate density profile

Erythrocyte density distribution profiles were obtained using phthalate esters in microhematocrit tubes at 4°C as described by us in detail previously (35) and by others more recently (1). Briefly, phthalate solutions were prepared to give a range of densities between 1.08 and 1.120 g/ml before (baseline) and after 3 h of oxygenation and deoxygenation cycles in a plasma-like buffer. The amount of denser cells was calculated from the total cell content below the oil layer (lower layer) divided by the total amount of cells and expressed as a percentage. The phthalate oil density at 50% (D₅₀) is the phthalate oil density that divides the cell population into two equal parts. This value is used to determine alterations in the cellular density profile of the entire RBC population in a given blood sample (1).

Statistical analysis

Results are expressed as means ± se unless otherwise stated. ANOVA was used to evaluate statistical significance among animal groups. Analysis was done using paired or unpaired t test. Significance was set at P < 0.05.

RESULTS

Gardos channel activity is blocked by PDI inhibitors

Sulfhydryl modifying agents are not selective and affect all thiol groups in the intracellular and/or extracellular compartments. To the best of our knowledge, regulation of sRBC Gardos channel by cell surface thiol-modifying proteins has not been reported. We studied the effects of extracellular thiol group modification on Gardos channel activity using the impermeable agents DTNB and bacitracin and compared them to the permeant thiol blocker, PAO, and the reducing agent DTT (Fig. 1A). We found that the strong oxidant DTNB and PAO blunted Gardos channel activity by 74 and 40%, respectively. On the other hand, the strong reducing agent 10 mM DTT had negligible effects on sRBC Gardos channel maximal activity. These observations suggest that the effects of DTNB or DTT on channel activity are specifically related to the redox state of thiol/disulfide groups. In addition, as PAO and bacitracin are nonselective but well-described PDI inhibitors (39), these results suggest that PDI may regulate the Gardos channel. mAb PDI has been shown to selectively inhibit PDI activity in leukocytes without affecting either thioredoxin or glutaredoxin activity (24). We tested the effect of mAb PDI on channel activity in sRBCs (Fig. 1B). We observed that mAb PDI blocked 66% of the Gardos channel activity, an event not observed when using nonspecific IgG. This inhibitory effect on Gardos channel activity was dose dependent, with an IC₅₀ of 4 ng/ml mAb PDI (n=3) in sRBCs. We then tested the effects of the thyroid hormone and PDI inhibitor, triiodothyronine (T₃) on Gardos channel activity (19). T₃ likewise blocked Gardos channel activity (73%; Fig. 1B). The selectivity of T₃ for PDI has been demonstrated, as T₃ was used to isolate PDI from the surface of mammalian cells (42). Taken together, our results imply that the channel is more active in the reduced form than in the oxidized form, as the above-mentioned oxidizing agents blocked channel activity. These results further implicate a heretofore unknown role for PDI on calcium-activated potassium channel activity and erythrocyte physiology.

Figure 1.

Thiol blockers inhibit Gardos channel activity in ex vivo human erythrocytes. Human sickle erythrocytes were freshly isolated, and channel activity was measured as described in Materials and Methods in the presence or absence of either 0.1 mM DTNB, 0.1 mM PAO, 3 mM bacitracin (Bac), or 10 mM DTT (A) or 0.1 mM triiodothyronine (T3), 20 ng/ml mAb PDI, or 20 ng/ml IgG (B). Channel activity was calculated from the difference of K⁺ flux in the presence or absence of 50 nM charybdotoxin (ChTX). Values represent means ± sd, n = 5 in 3 subjects. *P < 0.05, **P < 0.01; ^βP < 0.001.

sa-PDI activity is increased in SCD

To characterize PDI activity at the erythrocyte cell surface, we measured the release of acid-soluble ¹²⁵I-Tyr-SH from surface-bound ¹²⁵I-Tyr-S-S-poly-d-lysine in intact erythrocytes as described in Materials and Methods. As shown in Fig. 2A, sRBCs (SS patients) showed a significantly greater sa-PDI activity in baseline conditions than in erythrocytes from otherwise healthy subjects (AA). These results demonstrate the existence of a detectable erythrocyte sa-PDI activity.

Figure 2.

RBC sa-PDI activity. Activity of PDI in ex vivo intact mouse and human erythrocytes was measured using the radiolabeled ¹²⁵I-Tyr-S-S-PDL assay as described in Materials and Methods. A) RBC sa-PDI activity in intact erythrocytes from otherwise healthy human blood with hemoglobin A (AA, n=18) and sickle cell blood with hemoglobin SS (SS, n=21), *P < 0.0001. B) RBC sa-PDI activity in intact mouse erythrocytes from BERK sickle cell transgenic, BERK-trait, SAD transgenic, and β^S-Antilles (β^S-Ant) transgenic mice compared to wild-type C57BL/6J and HbAKO mice; n = 6. *P < 0.0003. C) Gardos channel activity correlates with sRBC sa-PDI activity in human sickle erythrocytes. Gardos channel and PDI activities were simultaneously measured in the same blood sample (n=21). Linear correlation with r² = 0.17642, P = 0.029. D) PDI blockade decreases cellular dehydration. Erythrocyte density distribution profile was determined as described in Materials and Methods in sRBCs at baseline before the oxygenation/deoxygenation cycles (solid squares), and in the absence (solid circles) or in presence of mAb PDI (open circles) or nonspecific IgG (open circles). Values represent means ± sd of 3 experiments in 3 patients with SCD.

PDI activity was also evaluated in 4 different sickle cell transgenic mouse models [BERK-trait, BERK (43), β^S-Antilles, and SAD] under similar conditions and compared to wild-type C57BL/6J and HbAKO mice. As shown in Fig. 2B, the SAD, β^S-Antilles, and BERK mice showed elevated PDI activity when compared to C57BL/6J, HbAKO, and BERK-trait erythrocytes. These results show that sickle cell transgenic mouse models likewise have elevated PDI activity as observed above in patients with SCD.

We then studied the relationship between sickle erythrocyte sa-PDI activity and Gardos channel activity. As shown in Fig. 2C, there was a significant and positive correlation (r=0.419, n=21, P=0.029) between baseline sa-PDI and Gardos channel activity; results that further support the contention that sa-PDI activity modulates Gardos channel activity.

To determine whether sa-PDI is involved in regulating erythrocyte volume, we induced dehydration by deoxygenation/oxygenation cycles of human sRBCs for 3 h (33) in the presence or absence of mAb PDI compared to nonspecific IgG (20 ng/ml). As shown in Fig. 2D, the presence of mAb PDI induced a shift to the left of the erythrocyte density profile (D₅₀ from 1.115±0.001 to 1.104±0.001, n=3, P=0.012), suggesting that PDI inhibition results in a reduction of cellular dehydration.

Immunodetection of PDI in erythrocyte membranes

To identify whether PDI is present in the erythrocyte membrane, we prepared solubilized human erythrocyte membranes from otherwise healthy subjects and subjects with SCD, and separated proteins on SDS-PAGE that were immunoblotted using mAb PDI (Fig. 3A). mAb PDI reacted with a major band in both normal RBCs and sRBCs, which corresponds to the molecular mass of PDI (57 kDa; ref. 19). As our functional assays suggested an association between PDI and Gardos channel (Fig. 2C), we performed a reciprocal coimmunoprecipitation assay. We showed that blotting with anti-PDI, the immunoprecipitate was associated with Gardos channel, and blotting with anti-KCNN4, PDI was pulled down (Fig. 3B, C). Specificity was tested by immunoprecipitation with nonimmune rabbit IgG. Immunoblot analysis with anti-KCNN4 antibodies detected the 47-kDa Gardos channel in erythrocyte membranes (Fig. 3B, C). These results strongly support the contention that PDI is present in the erythrocyte membrane and suggest the existence of a Gardos-PDI complex.

Figure 3.

Immunodetection of PDI and KCNN4 in human erythrocytes. A) Immunodetection of plasma membrane-associated PDI in human sickle (SS) and normal (AA) erythrocytes. B) Immunoprecipitation of KCNN4 followed by Western blot for PDI (left panel) or for KCNN4 (right panel). C) Immunoprecipitation of PDI followed by Western blot for PDI (left panel) or for KCNN4 (right panel), as described in Materials and Methods, in sickle erythrocytes. IgG was used as control.

ET-1 stimulates erythrocyte sa-PDI activity

We tested the direct effects of mAb PDI on ET-1-induced Gardos channel activity in the absence of A23187. In sRBCs, we observed an increase in Gardos channel activity on ET-1 receptor activation (Fig. 4A). However, in the presence of mAb PDI, ET-1-induced Gardos channel activity was significantly decreased, but not in the presence of nonspecific IgG. These results support the contention that PDI can modulate the effects of ET-1 receptor activation on Gardos channel.

Figure 4.

ET-1 stimulates RBC Gardos channel activity and sa-PDI activity. A) mAb PDI (20 ng/ml) blocked ET-1-stimulated Gardos channel activity. Channel activity was measured in the absence of A23187 with 1.5 mM extracellular Ca²⁺ in sickle erythrocytes. Values represent means ± sd, n = 9. *P < 0.05, **P < 0.01. B) ET-1 (500 nM) induced sa-PDI activity in sickle erythrocytes. BQ3020 (1 μM) and IRL1620 (1 μM) are two selective ET-1 receptor subtype B agonists used as positive controls. Values represent means ± sd in triplicate determinations. *P < 0.0001.

We then determined the effect of ET-1 on erythrocyte sa-PDI activity following incubation with ET-1 (Fig. 4B). We showed that sa-PDI activity was significantly increased by ET-1. As we previously reported that erythrocytes express ET-1 type B receptors (33), we then evaluated the effects of two distinct and selective ET-1 type B receptor agonists (IRL1620 and BQ3020) on sa-PDI activity and found similar activation (Fig. 4B). Furthermore, we found that mAb PDI was able to block BQ3020-induced sa-PDI activity (8.8±0.5 to 5.02±1.7%, n=3, P<0.01), further supporting the contention that ET-1 type B receptor activation is associated with stimulation of erythrocyte PDI activity.

Increased protein casein kinase 2 (CKII) activity in sRBCs is regulated by ET-1 and leads to modulation of sa-PDI activity and cellular dehydration

ET-1 type B receptors belong to the superfamily of G-protein-coupled receptors that mediate their effects via regulation of CKII (44, 45). Our results indicate that ET-1 can induce significant increases in CKII activity in erythrocyte membranes as well as in the cytosolic fractions of human erythrocytes (Fig. 5A). This stimulation was strongly inhibited by 4,5,6,7-tetrabromobenzotriazole (TBB), a specific CKII inhibitor (46). We found that ET-1-induced sa-PDI activity was strongly inhibited by TBB (Fig. 5B) in intact sRBCs, suggesting that CKII might have an important role in the interaction of PDI and Gardos channel. Furthermore, these results suggest that ET-1 type B receptor activation leads to activation of CKII in sRBCs, as shown in other cell types (44).

Figure 5.

Protein kinase CKII inhibitors block ET-1-induced sa-PDI activity and cellular dehydration in sickle erythrocytes. A) CKII activity in membrane and cytosolic fractions was measured as described in Materials and Methods; 500 nM ET-1 and 2 μM TBB, a specific CKII inhibitor, were used. Statistical comparison of ET-1 vs. ET-1 in the presence of TBB. Values represent means ± sd of triplicate determinations. *P < 0.05,**P < 0.001; ^βP < 0.003; ^αP < 0.002; ^#P < 0.004. B) ET-1 induced sa-PDI activity in intact sickle erythrocytes in the presence or absence of 2 μM TBB. sa-PDI was measured by fluorescence as described in Materials and Methods. Values represent means ± sd of 6 determinations. RFU, relative fluorescence units. *P < 0.05, **P < 0.001. C) Erythrocyte density profile of sickle erythrocytes was measured as described in Materials and Methods before (baseline, solid square) or after 3 h oxygenation/deoxygenation cycles in the absence (control, open triangle) or presence of 20 μM apigenin (open circles) or 2 μM TBB (solid triangles). Results are representative of one experiment in duplicate determinations performed in 3 different subjects.

To further characterize a role for CKII on cellular dehydration, we incubated sRBCs for 3 h in deoxygenation/oxygenation cycles, 10 min each cycle in the presence or absence of TBB or apigenin (47), and measured the changes in erythrocyte density distribution, as described in Materials and Methods and shown previously (33). We observed that CKII inhibition by apigenin or TBB led to inhibition of deoxygenation/oxygenation-stimulated cellular dehydration of sRBCs (Fig. 5C). Together, these results implicate CKII as a novel mediator of cellular volume changes in sRBCs.

In vivo treatment of sickle cell transgenic mice with ETRAs leads to decreased PDI activity

We have previously shown that in vivo treatment with ETRAs (BQ788/BQ123 mixture) for 14 d leads to improved hematological parameters when compared to saline-treated SAD mice (10). However the effects of ETRAs in a severe mouse model of SCD is not, to the best of our knowledge, reported. The BERK sickle cell mouse has been proposed as a model of severe pathology and increased oxidative stress (30). The pharmacokinetic parameters of ETRAs following intraperitoneal injections were determined (Supplemental Data). As shown in Fig. 6A, we observed a significant decrease in intact erythrocyte sa-PDI activity (n=5, P<0.02) on treatment with ETRAs when compared to vehicle-treated BERK mice. Since PDI is present in the circulation, we investigated whether plasma PDI activity was also affected by ETRA treatment. We measured PDI activity in mouse plasma collected after treatment and observed that plasma PDI activity was likewise reduced in treated mice compared to vehicle treatment (Fig. 6B, n=9, P<0.003). As there is evidence that activated endothelial cells secrete PDI (48), we studied early cultures of mouse aortic mouse endothelial cells from BERK mice following treatment with ETRA. We found a reduction in PDI mRNA levels in mice treated with ETRA compared to vehicle treatment (Fig. 6C, n=5, P<0.0001). These results suggest that ET-1 blockade signaling pathway might include decreased PDI secretion in vivo.

Figure 6.

In vivo treatment of BERK sickle transgenic mice with ETRAs leads to reduced PDI activity. A) RBC sa-PDI activity was measured with the radiolabeled ¹²⁵I-Tyr-SS-PDL conjugate in intact cells, as described in Materials and Methods; n = 5. *P < 0.02. B) Plasma levels of circulating PDI activity were measured with the fluorescence-labeled conjugate protocol, as described in Materials and Methods; n = 9. *P < 0.003. C) PDI mRNA levels from early cultures of mouse aortic endothelial cells were determined by TaqMan gene expression assays on a singleplex real-time RT-PCR reaction using commercially available probe and primer sets from Applied Biosystems following manufacturer's protocols, as described in Materials and Methods. β-2-microglobulin was used an endogenous control, n = 5. Control, vehicle (saline)-treated BERK mice. *P < 0.0001.

ETRA treatment ameliorates hematological parameters and Gardos channel activity in BERK mice

Consistent with our previous reports, we observed improvements in hematological parameters after antagonist administration. ETRA treatment of BERK mice led to improved mean cellular volume (MCV) and mean corpuscular hemoglobin concentration (MCHC) (Fig. 7A, n=6, P<0.001). MCV was significantly increased after treatment as previously observed in SAD mice (10). This change was accompanied by a decrease in MCHC. In addition, ETRA treatment was associated with lower Gardos channel activity in ex vivo erythrocytes (Fig. 7B, n=6, P<0.001) further supporting a role for ET-1 receptor activation in sickle cell pathology.

Figure 7.

ETRAs in vivo improve hematological parameters in BERK sickle transgenic mice. A) Effect of ETRAs on mean cellular volume (MCV) and mean corpuscular hemoglobin concentration (MCHC). Hematological parameters were determined as described in Materials and Methods; n = 6. *P < 0.001. B) Gardos channel was measured as described in Materials and Methods and expressed as charybdotoxin-sensitive fraction of K⁺ influx in intact mouse erythrocytes. Values represent means ± sd of 6 experiments in triplicate determinations. *P < 0.001.

ETRA treatment decreases inflammatory markers in BERK mice

Cytokine production in SCD is chronically elevated and plays an important role in the vasoocclusive episodes. To assess whether ETRA treatment affects the inflammatory state of BERK mice, we measured plasma levels of several cytokines altered in SCD following ETRA treatment. We observed that tumor growth factor β1 (TGFβ1) was significantly reduced (Table 1), while circulating soluble vascular cell adhesion molecule (sVCAM) and soluble intercellular adhesion molecule (sICAM) were not significantly affected during our 2-wk ETRA treatment of BERK mice.

Table 1.

Plasma inflammatory marker levels in BERK mice after 14 d of treatment with ETRAs

ID	sICAM	sVCAM	TGFβ1	TNFα
HbAKO	417 ± 19	684.7 ± 32	12.8 ± 1	4.1 ± 0.8
BERK + vehicle	460.5 ± 23	735.9 ± 31	30.7 ± 4**	4.7 ± 0.6
BERK + ETRA	517 ± 1.8	859.3 ± 131	16.4 ± 1.4*	4.5 ± 0.3

Circulating cytokine levels were measured by ELISA as described in Materials and Methods. sICAM, soluble intercellular adhesion molecule; sVCAM, soluble vascular cell adhesion molecule; TGFβ1, tumor growth factor β1; TNFα, tissue necrosis factor α. Vehicle = saline. Values represent means ± sd (n=6).

^*P < 0.002

^**P < 0.001.

DISCUSSION

There is evidence of abnormal redox status in sRBCs (13). Modification of plasma membrane redox status has been shown to regulate erythrocyte K⁺ transport and hydration status through mechanisms not clearly described (49). We now report on the presence of the thiol/disulfide interchange enzyme, PDI, in human and mouse erythrocytes and that blockade of its activity leads to reduced Gardos channel activity and modulation of sickle erythrocyte hydration status. In addition, we observed increased PDI activity in sRBCs when compared to PDI activity in cells from otherwise healthy subjects that can be activated by ET-1. Consistent with these observations, we showed that erythrocytes from sickle cell transgenic mouse models have elevated sa-PDI activity in erythrocytes when compared to either wild-type C57BL/6J mice or HbAKO transgenic mice. Furthermore, PDI activity correlated positively with Gardos channel activity in intact human sRBCs. Thus, these results support our hypothesis that PDI plays a novel role in mediating the effects of ET-1-induced Gardos channel activation and cellular hydration status in SCD. In addition, our results suggest that erythrocyte redox status is an important factor in erythrocyte hydration and that altered PDI redox status might contribute to the increased permeability and altered cation homeostasis observed in sRBCs.

We, and others, have provided preclinical evidence that blockade of ET-1 effects with ETRAs in mice might be of therapeutic importance in patients with SCD (10, 12). PDI activity may represent a common characteristic of SCD that may serve as a potential biomarker for disease severity. To this end, we treated BERK SCD mice, a mouse model of increased disease severity, with ETRAs. We showed that in vivo blockade of ET-1 receptor activation improves erythrocyte hydration, cellular hemoglobin concentration, and levels of the inflammatory marker TGFβ1, events that were associated with lower circulating and surface-associated erythrocyte PDI activity. These results confirm and extend previous reports showing numerous beneficial effects of ETRA treatment in SCD in sickle cell transgenic SAD mice (10, 12). In addition, they support the contention that PDI contributes to abnormal redox status in SCD and that the Gardos channel is functionally coupled to PDI. Consequently in SCD, elevated levels of ET-1 contribute to disordered redox status, leading to increased Gardos channel activity, factors that, in turn, may contribute to formation of dense, dehydrated erythrocytes, vasoocclusion, and organ damage. The mechanisms for ET-1's effects involve regulation of ion transport mechanisms and dehydration in sRBCs (10). Our current results expand on this knowledge, as we demonstrate herein that PDI plays a novel role in ET-1 receptor signaling and cell volume regulation that may be regulated by protein kinase CKII in erythrocytes. The role of protein kinase CKII in ET-1-induced channel activation might, in turn, be mediated by calmodulin (50), a regulator of Gardos channel function.

PDI is the only vicinal thiol-containing cell surface-associated enzyme (CXXC) to have been positively identified in mammalian cells, yet little is known about its physiological role at the cell surface (51). The evidence points to PDI as a critical modulator of Gardos channel activity, although the contribution of other surface reductases has not been completely ruled out in this study. In this regard, mAb PDI has been shown to selectively inhibit PDI activity in leukocytes without affecting either thioredoxin or glutaredoxin activity (24). We have also shown the specificity of mAb PDI and T₃ to block PDI activity (19). In addition, PDI has essentially no sequence homology with glutaredoxin (52) or its active site, thus reducing the likelihood that any cross-reactivity with mAb PDI or bacitracin may exist. Furthermore, glutaredoxin has not been described in the plasma membrane of mammalian cells. Of importance for our studies, thioredoxin is stimulated by bacitracin in vitro, while PDI is inhibited by this compound, as we have previously shown (24). Thus, the currently available data allow us to propose that sa-PDI regulates the Gardos channel.

Few studies have evaluated the effects of redox status on Gardos channel activity in SCD. In contrast, several studies have evaluated the effects of oxidative damage on the activity of the K/Cl cotransporter (KCC), another K⁺ transporter that contributes to sickle dehydration (53, 54). Joiner et al. (54) reported that DTT can partially block the maximal velocity of the cotransporter and showed that pH-sensitive KCC activation was likewise inhibited by DTT, suggesting that sulfhydryl oxidation may be responsible for the pH sensitivity of the cotransporter. Thus, these results suggest that both volume regulatory transporters, the Gardos channel and KCC, are highly sensitive to redox agents. However, we do not know the role of PDI on KCC activity. The Gardos channel has 9 endogenous cysteine residues, with 4 of these located around the gating site, making them likely candidates for redox modification and gating site modification (18, 55, 56). We observed coimmunoprecipitation of PDI and Gardos channel that suggest their proximity, potential compartmentalization, and coupling functionality. Consequently, further studies are needed to clarify the existence of a PDI-Gardos complex.

Our results suggest that PDI inhibitors and PDI regulators, like ETRAs, may lead to decreases in sRBC dehydration and Gardos channel activity by redox modifications. This hypothesis is supported by results showing increased reductive capacity in the plasma membrane of both human and mouse sRBCs that was blocked by mAb PDI and other PDI inhibitors. In addition, we showed that ET-1 increases the reduced form of PDI, as cell surface reductive capacity was significantly elevated and sensitive to PDI inhibitors. Furthermore, complete blockade of PDI thiols by either bacitracin or DTNB inhibited Gardos channel activity. These results implicate PDI redox modifications in the pathology of SCD. Of interest, studies performed using thiol/disulfide exchange chromatography of SDS-solubilized sRBCs, a technique that allows separation of membrane proteins on the basis of reactive thiols, revealed increased levels of (oxidized) proteins in sRBCs compared with membranes from otherwise healthy subjects (13). However, and as reported by the researchers, this approach does not allow the detection of partially reduced proteins, as the methods are unable to detect that the concentration of oxidants and reductants is linearly related with respect to the level of oxidation (13, 14).

It is now evident that extracellular redox enzymes catalyze thiol/disulfide exchange reactions of target proteins at the cell surface. In this regard, a role for PDI has been reported for platelet activation leading to increases in cell surface thiol groups (28, 57). In addition, recent evidence shows a role for PDI on integrin receptor, P-selectin exposure, and platelet activation, leading to increased thrombotic and coagulation states (57). Of importance, inhibition of these target proteins has been shown to ameliorate or interfere with sickle cell pathology (58–61). In SCD, intravascular hemolysis may lead to increases in PDI levels in circulation that could further accelerate platelet activation and white blood cell adherence, leading to increased vasoocclusive episodes. Alternatively, there is evidence that PDI is also secreted by activated endothelial cells, as observed during thrombosis (48). Activation of the endothelium, as observed in SCD, leads to increased sickle cell adherence and correlates with disease severity (62). Thus it is tempting to speculate that elevated plasma PDI levels might likewise originate from activated endothelium and contribute to SCD pathology. Consistent with this proposal, we observed decreased PDI mRNA levels in early cultures of aortic endothelial cells from mice treated with ETRAs vs. vehicle-treated mice. This is important, as ET-1 is a well-described and important regulator of endothelial cell activation and, its blockade in vivo may lead to improved vascular parameters in SCD mice. In addition, we now provide evidence that PDI activity in circulation is likewise reduced by treatment with ETRAs in our mouse model. However, whether PDI is secreted and released into the plasma and reassociated with the plasma membrane or is translocated to the plasma membrane from intracellular sites to exert its effects is currently not known.

In summary, our findings show evidence that PDI is a novel target of disordered ET-1 levels in SCD and that studying its function may improve our mechanistic understanding about how sRBCs regulate hydration status and oxidative stress, activate the endothelium, and promote cellular adhesion, leading to vasoocclusion.