New sickle cell disease therapies modulate red blood cell deformability via Ca²⁺, ATP, and 2,3-DPG in normoxia and hypoxia Open Access

Sickle cell disease (SCD) is a genetic disorder characterized by the presence of an abnormal hemoglobin (Hb); hemoglobin S (HbSS). HbSS polymerizes under deoxygenated conditions and deforms red blood cells (RBCs) into rigid and sickled shapes.^1,2 Even at normal oxygen (O₂) tension (normoxia), residual HbS polymers impose modest membrane tension that activates the mechanosensitive Piezo1 channel, permitting subclinical calcium (Ca²⁺) influx and subtly reducing deformability.^3-5 Under hypoxia, polymerization intensifies, sharply increasing membrane stress, Piezo1 activation, and Ca²⁺ influx. Elevated intracellular Ca²⁺ triggers the Gardos channel, driving K⁺ efflux and water loss.^6,7 Consequently, triggering calpain-mediated cytoskeletal breakdown, Band 3 phosphorylation, and thus reducing RBC deformability.⁸ Impaired RBC deformability leads to severe microvascular occlusion, hemolytic anemia, and painful vaso-occlusive crises characteristic of SCD.^2,9 Beyond HbSS polymerization, oxidative stress and chronic inflammation further exacerbate membrane lipid peroxidation and protein oxidation, compounding RBC deformability and microvascular flow.^9-11 Understanding the intricacies of RBC deformability has implications for disease management and therapeutic intervention.

Recent therapeutic advances have focused on increasing hemoglobin oxygen (Hb-O₂) affinity to prevent HbSS polymerization.^12,13 Voxelotor binds the α-chain of HbSS to stabilize its oxygenated (R-state) conformation, reducing HbSS polymerization.¹⁴ Similarly, GBT021601, a second-generation HbS polymerization inhibitor, functions through a comparable mechanism to voxelotor.¹⁵ Complementing these direct polymer inhibitors, pyruvate kinase (PKR) activators such as mitapivat and etavopivat, which are currently being tested in SCD clinical trials.^16,17 PKR activators have multifaceted effects on RBC glycolytic pathways, including increasing the activity of PKR, thereby reducing 2,3-diphosphoglycerate (2,3-DPG) and elevating adenosine triphosphate (ATP) levels.^13,18 The therapeutic effect of decreasing 2,3-DPG is to allosterically increase the Hb-O₂ affinity, thus delaying the formation of HbS polymers, reducing RBC sickling, and hemolysis, thereby mitigating the anemia and potentially other clinical manifestations of SCD.^17,19,20 

Despite the clinical benefits, the direct impacts of voxelotor, GBT021601, and PKR activators on RBCs' total intracellular calcium content ([Ca²⁺]_total) homeostasis, cytoskeletal stability, and Band 3 tyrosine phosphorylation, a key regulator of membrane-cytoskeleton interactions,²¹ remain unstudied. Herein, we systematically evaluated levels of 2,3-DPG, ATP, [Ca²⁺]_total, and Band 3 phosphorylation in SCD (HbSS) and healthy (HbAA) RBCs under normoxic and chemically induced hypoxic (hypoxia) conditions. Furthermore, by using selective inhibitors of the mechanosensitive Piezo1 channel and plasma membrane Ca²⁺-ATPase (PMCA), we investigated the contribution of Piezo1 activation vs PMCA activity to [Ca²⁺]_total regulation and the resulting impact on microcapillary RBC deformability with treatment.

Using the OcclusionChip, a microcapillary occlusion assay,^22-24 we evaluated RBC deformability in HbSS and HbAA samples under normoxic and hypoxic conditions. Unlike ektacytometry, which measures elongation of bulk RBCs under shear stress, the OcclusionChip quantifies the percentage of microchannels blocked by individual cells, providing a single-cell, physiologically relevant assessment that can detect subtle differences in RBC deformability.²⁴ We hypothesized that voxelotor, GBT021601, and PKR activators enhance RBC deformability by differentially regulating intracellular 2,3-DPG, ATP, and [Ca²⁺]_total levels. We observed that all 3 agents significantly reduced [Ca²⁺]_total, potentially via Piezo1 inhibition and modification of Ca²⁺ clearance pathways, which potentially could involve preserving residual PMCA activity or, alternatively, efflux mechanisms. GBT021601 but not voxelotor lowered 2,3-DPG, whereas PKR activators uniquely increased ATP under hypoxia. These changes in intracellular biomarkers correlated with decreased Band 3 tyrosine phosphorylation, suggesting treatment-specific restoration of sickle RBC integrity via modulation of membrane-protein interactions.

Whole blood samples from deidentified healthy controls (HbAA) and individuals with SCD (HbSS) were collected in EDTA tubes at University Hospitals Cleveland Medical Center. Ethical approval (05-14-07C) was granted by University Hospitals Cleveland Medical Center institutional review board, following the Declaration of Helsinki. Samples were stored at 4°C and processed within 10 hours. HbSS was confirmed via high-performance liquid chromatography. Samples from those on voxelotor therapy were excluded. Clinical data for HbSS donor, including samples on hydroxyurea and transfusion medications, are summarized in supplemental Table 1. Additional details on materials and methods are included in the supplemental Information.

Whole blood samples were processed by centrifugation to isolate RBCs, which were then washed with PBS containing 5 mM glucose and resuspended in a solution with dimethyl sulfoxide to serve as a vehicle control. For drug treatments, RBCs from both control HbAA and SCD patients HbSS were incubated with specific concentrations of PKR activator (HbSS + PKR), voxelotor (HbSS + Vox), or GBT021601 (HbSS + GBT021601) at 37°C for 6 hours on an orbital shaker. Additionally, some samples were pretreated with Grammostola spatulata mechanotoxin 4 (GsMTx4), a Piezo1 inhibitor,²⁵ or sodium orthovanadate, a PMCA inhibitor,^26,27 to study their effects in combination with these drugs, creating various experimental groups for subsequent assays (supplemental Notes 1).

As previously described, the OcclusionChip is a functional microcapillary occlusion assay in which RBCs flow through a microfluidic channel packed with a gradient of 20 μm to 4 μm micropillar arrays.^23,28 Less deformable RBCs are trapped earlier in the channel, whereas more deformable cells travel further, with side channels mimicking arteriovenous shunts to maintain flow despite upstream occlusion (Figure 1A). The OcclusionChip is cast in polydimethylsiloxane (PDMS) via soft lithography,^22,28 then plasma bonded to glass. Before experiments, the chips are functionalized; details are described in supplemental Notes 2.

RBC samples were introduced into the OcclusionChip under normal oxygen conditions (normoxia; pO₂ ≈ 100 mm Hg) for 15 minutes. The chip was imaged to assess the occlusion index (OI), which measures RBC deformability through the percentage of blockages in microchannels, using manual counting and software analysis (Figure 1A).

The occlusion assay was similar to the normoxia assay but conducted with chemically deoxygenated RBCs to simulate low oxygen mixed venous (pO₂ ≈ 44 mm Hg) conditions. Samples were perfused for 12 minutes, and the setup was otherwise identical, focusing on how hypoxia affects RBC occlusion (Figure 1A). Details in supplemental Notes 2.

[Ca²⁺]_total levels in packed RBCs were quantified using atomic absorption spectroscopy²⁹ after treating the cells with hydrochloric acid buffer, with samples processed through multiple centrifugation steps to isolate [Ca²⁺]_total for measurement described in supplemental Notes 3.

RBC membranes were prepared as “ghosts” for western blotting to analyze Band 3 protein phosphorylation, a critical determinant of RBC membrane integrity that regulates membrane-cytoskeleton interactions.²¹ This involved lysing cells, washing the membranes, and then performing protein quantification and blotting. Details are described in supplemental Notes 4 and 5.

RBCs were prepared for analysis by adding stable isotope-labeled standards and using methanol for protein precipitation.³⁰ The levels of ATP and 2,3-DPG were then measured using liquid chromatography–mass spectrometry, with detailed standard solutions and equipment calibration described in supplemental Notes 6 and 7.

Data from these experiments were analyzed using Minitab software (release 2023, version 21.4), employing various statistical tests like analysis of variance, paired t tests, Wilcoxon or Mann-Whitney U tests, and Shapiro-Wilk test, with significance set at P value <.05.

We employed the OcclusionChip to quantify microcapillary occlusion reported as OI, a measure of RBC deformability^22-24 in samples HbAA and those with HbSS under both normoxic and hypoxic conditions (Figure 1A). In HbAA RBCs, OI was low and unchanged between normoxia (1.16% ± 0.67%) and hypoxia (2.34% ± 0.10%; P > .05; Figure 1B). By contrast, untreated HbSS RBCs displayed elevated OI under normoxia (8.19% ± 1.43%; ∗∗∗P < .001 vs HbAA; Figure 1B) that further increased under hypoxia to 38.72% ± 3.46% (∗∗∗P < .001 vs normoxia and ∗∗∗P = .001 vs HbAA; Figure 1B,E). Under normoxia, HbSS + Vox, HbSS + PKR, and HbSS + GBT021601 each markedly reduced OI in HbSS RBCs (from 8.19% ± 1.43% to 8.41% ± 1.67%, 7.69% ± 2.17%, and 9.11% ± 1.88%, respectively; ∗P < .05 vs untreated; Figure 1C,E). Under hypoxia, these agents similarly lowered OI from 38.72% ± 3.46% to 18.49% ± 2.78% (HbSS + Vox), 23.94% ± 3.28% (HbSS + PKR), and 19.93% ± 3.58% (HbSS + GBT021601; ∗∗P < .01 and ∗∗∗P < .001; Figure 1D-E). Voxelotor reduced OI by approximately 52.23%, PKR activators by 38.96%, and GBT021601 by 48.50%. These findings suggest that (1) HbSS RBCs are far less deformable than HbAA RBCs and (2) treatment with voxelotor, PKR activators, or GBT021601 restores deformability, lowering occlusion to HbAA RBC levels in both normoxia and hypoxia.

We quantified ATP and 2,3-DPG in RBCs, since high 2,3-DPG and low ATP are correlated with enhanced membrane dynamics and reduced deformability.^31-33 Here, we found that under normoxia, HbAA RBCs exhibited higher ATP (190.5 ± 18.6 μg/mL) than HbSS RBCs (138.6 ± 13.9 μg/mL; ∗P < .05; Figure 2A) HbSS + Vox and HbSS + PKR did not alter ATP levels in HbSS RBCs (137.6 ± 14.3 and 129.1 ± 9.6 μg/mL, respectively; P > .05), whereas HbSS + GBT021601 modestly decreased ATP (101.0 ± 12.6 μg/mL; P = .058; Figure 2A). Normoxic 2,3-DPG was lower in HbAA (111.6 ± 11.6 μg/mL) compared to HbSS RBCs (157.6 ± 35.4 μg/mL; ∗P < .05; Figure 2B). PKR activators (86.0 ± 18.9 μg/mL) and GBT021601 (64.3 ± 10.9 μg/mL) significantly reduced 2,3-DPG compared to untreated HbSS (157.6 ± 35.4 μg/mL; ∗∗P < .01; Figure 2B), whereas voxelotor (139.5 ± 32.2 μg/mL) had no effect (P > .05; Figure 2B).

Under hypoxia, HbAA ATP (130.5 ± 9.6 μg/mL) remained significantly higher than HbSS ATP (78.6 ± 9.9 μg/mL; ∗P < .05; Figure 2C). ATP levels of untreated HbSS RBCs (78.6 ± 9.9 μg/mL) did not differ from either HbSS + Vox (70.8 ± 8.26 μg/mL) or HbSS + GBT021601 (55.8 ± 3.31 μg/mL; ∗P > .05 for both comparisons; Figure 2C). HbSS + Vox and HbSS + GBT021601 did not significantly alter hypoxic ATP levels (70.8 ± 8.3 and 55.8 ± 3.3 μg/mL, respectively; P > .05), whereas HbSS + PKR activators increased ATP levels to 97.0 ± 16.3 μg/mL (∗∗P < .01; Figure 2C). Hypoxic 2,3-DPG was lower in HbAA (147.0 ± 8.6 μg/mL) than in HbSS RBCs (210.6 ± 19.2 μg/mL; ∗P < .05; Figure 2D). HbSS + PKR (120.1 ± 10.6 μg/mL) and HbSS + GBT021601 (115.0 ± 18.9 μg/mL) significantly reduced hypoxic 2,3-DPG vs untreated (210.6 ± 19.2 μg/mL; ∗P < .05), whereas voxelotor (190.5 ± 17.2 μg/mL) remained the same (P > .05; Figure 2D). These findings indicate that voxelotor and GBT021601 do not affect ATP, but PKR activators increased ATP only under hypoxia. Both PKR activators and GBT021601 consistently lower 2,3-DPG in HbSS RBCs, whereas voxelotor has no impact on 2,3-DPG in either normoxia or hypoxia.

Given the pivotal role of Ca²⁺ in RBC deformability, we hypothesized that voxelotor, PKR activators, and GBT021601 would modulate intracellular [Ca²⁺]_total levels. Elevated Ca²⁺ activates calpain and calmodulin, which remodel cytoskeletal proteins to influence the membrane.^5,7 Pathological Ca²⁺ overload contributes to RBC dehydration and hypoxia-induced sickling. Thus, we assessed how these drugs, by altering [Ca²⁺]_total dynamics, impact RBC occlusion under normoxia and hypoxia.

Under normoxia, [Ca²⁺]_total levels in untreated HbAA RBCs (80.6 ± 21.9 μmol/L) were comparable to that in RBCs treated with voxelotor (87.2 ± 14.9 μmol/L) or PKR activators (99.8 ± 23.5 7 μmol/L; P > .05 for all; Figure 3A). In contrast, HbSS RBCs showed elevated [Ca²⁺]_total (145.20 ± 10.85 μmol/L), which was significantly reduced by voxelotor (102.68 ± 7.69 μmol/L), PKR activators (109.23 ± 9.01 μmol/L), or GBT021601 (99.92 ± 13.58 μmol/L; ∗∗∗P < .001 for all; Figure 3B).

Under hypoxia, similar drug-induced reductions occurred: voxelotor (102.2 ± 13.6 μmol/L), PKR activators (137.8 ± 15.7 μmol/L), and GBT021601 (125.0 ± 13.8 μmol/L) all lowered [Ca²⁺]_total relative to untreated HbSS RBCs (222.6 ± 23.6 μmol/L; ∗∗∗P < .001; for all; Figure 3B). To probe the role of Piezo1, we treated HbSS RBCs with the Piezo1 inhibitor GsMTx4. Under normoxia, GsMTx4 alone (105.0 ± 8.2 μmol/L) yielded [Ca²⁺]_total levels comparable to cotreatments with voxelotor (86.7 ± 10.2 μmol/L), PKR activators (91.8 ± 8.9 μmol/L), or GBT021601 (85.9 ± 9.2 μmol/L; P > .05; Figure 3C) Similarly, under hypoxia, GsMTx4 (116.0 ± 18.9 μmol/L) likewise matched cotreatments (voxelotor, 97.0 ± 17.9 μmol/L; PKR activators, 94.8 ± 18.3 μmol/L; GBT021601, 80.0 ± 14.7 μmol/L; P > .05 for all; Figure 3C). We next inhibited PMCA with sodium orthovanadate to isolate its contribution to [Ca²⁺]_total with drug treatment. Under normoxia, vanadate increased [Ca²⁺]_total (186.5 ± 14.9 μmol/L), which was then reduced by cotreatment with voxelotor (121.8 ± 17.7 μmol/L), PKR activators (125.2 ± 12.5 μmol/L), or GBT021601 (123.8 ± 17.9 μmol/L; ∗∗∗P < .001 for all; Figure 3D). Under hypoxia, vanadate further elevated [Ca²⁺]total (238.1 ± 29.7 μmol/L), but each drug reduced this (voxelotor, 149.0 ± 18.6 μmol/L; PKR activators, 122.0 ± 15.0 μmol/L; GBT021601, 148.4 ± 18.5 μmol/L; ∗∗∗P < .001 for all; Figure 3D). Finally, we measured the OI after GsMTx4 or vanadate treatments (Figure 3E-F). GsMTx4 under normoxia reduced OI (18.0% ± 1.7% vs 25.3% ± 3.3%; ∗P < .05), but cotreatments did not further improve it (voxelotor [15% ± 0.94%], PKR activators [16.53% ± 0.89%], or GBT021601 [16.307% ± 1.5%]; P > .05 for all; Figure 3E). Under hypoxia, GsMTx4 also decreased occlusion (55.3% ± 6.3% vs 63.5% ± 7.7%; ∗∗P < .01) with no added benefit from cotreatments and voxelotor (46.9% ± 7.6%), PKR activators (46.2% ± 5.68%), or GBT021601 (38.54% ± 6.37; P > .05 for all; Figure 3E). Vanadate increased occlusion under normoxia (18.7% ± 1.4% vs 14.3% ± 1.4%; ∗P < .05) but was reduced by cotreatments voxelotor (11.6% ± 1.9%), PKR activators (13.16% ± 1.81%), or GBT021601 (11.34% ± 1.4%; ∗P < .05 for all; Figure 3F). Under hypoxia, vanadate also increased occlusion (62.4% ± 1.3% vs 32.0% ± 8.3%; ∗∗∗P < .001). Cotreatments significantly reduced occlusion in both conditions: voxelotor (48.01% ± 6.3%), PKR activators (37.31% ± 4.33%), or GBT021601 (47.96% ± 6.3%; ∗P < .05, ∗∗P < .01, ∗P < .05, respectively; Figure 3F). These results indicate that (1) these agents selectively lower elevated [Ca²⁺]_total in HbSS RBCs without affecting HbAA RBCs, (2) Piezo1 inhibition induces further [Ca²⁺]_total reduction, suggesting a shared mechanism via Piezo1, (3) PMCA inhibition elevates [Ca²⁺]_total and occlusion, and (4) Cotreatment with voxelotor, PKR activators, or GBT021601 attenuates PMCA inhibitor induced [Ca²⁺]_total overload and occlusion, suggesting modification of alternative Ca²⁺ clearance pathways.

Previous studies have reported elevated tyrosine phosphorylation of Band 3 in HbSS RBCs.^8,21 To evaluate the impact of voxelotor, PKR activators, and GBT021601 on membrane protein phosphorylation, we quantified Band 3 tyrosine phosphorylation levels under normoxic and hypoxic conditions. Under normoxia, phosphorylation levels in untreated HbAA RBCs (99.9 ± 6.0) were comparable to those in HbAA RBCs treated with voxelotor (100.12 ± 7.44) or PKR activators (96.86 ± 5.78) with no significant differences (P > .05 for both; Figure 4A-B).

In contrast, HbSS RBCs exhibited significantly elevated Band 3 phosphorylation levels (110.28 ± 4.38), which were markedly reduced following treatment with voxelotor (75.39 ± 4.94), PKR activators (69.83 ± 6.23), or GBT021601 (75.98 ± 12.75) with respective statistical significance (∗∗∗P < .001, ∗∗∗P < . 001, ∗P < .05, respectively; Figure 4A,C). Under hypoxia, Band 3 phosphorylation levels in untreated HbSS RBCs (225 ± 7.38) increased further, but voxelotor (162.4 ± 4.56), PKR activators (125.6 ± 7.89), and GBT021601 (145.89 ± 3.45) significantly reduced phosphorylation levels (∗∗P < .01 for all; Figure 4A,D). These findings demonstrate that voxelotor, PKR activators, and GBT021601 selectively reduce Band 3 tyrosine phosphorylation in HbSS RBCs without affecting HbAA RBCs. The pattern mirrors prior observations of drug-induced normalization of intracellular [Ca²⁺]_total levels and is consistent with a potential link between [Ca²⁺]_total dynamics and Band 3 phosphorylation in SCD.⁸ 

Recent developments in disease-modifying therapies for SCD have significantly expanded the therapeutic armamentarium.^34,35 The use of Hb-O₂ affinity–modifying drugs, such as voxelotor, GBT021601, and activation of erythrocyte PKR, presents an intriguing avenue for treatment. These approaches increase Hb-O₂ affinity and limit HbSS polymerization.^31-34 However, the mechanisms by which increased Hb-O₂ affinity improves RBC deformability, particularly via surrogate markers of cytoskeletal stability under normoxic and hypoxic conditions, remain insufficiently explored.

In this study, we leveraged our understanding of HbSS RBC deformability under normoxic (pO₂ ≈ 100 mm Hg) and mixed venous hypoxic (pO₂ ≈ 44 mm Hg) conditions.^24,36 Using the OcclusionChip microcapillary assay to measure microcapillary occlusion OI in RBCs from HbAA and HbSS donors under normoxia and chemically induced hypoxia.^22,23 We showed that untreated HbSS RBCs exhibited dramatically elevated OI under normoxia (∼8% vs ∼1% in HbAA; ∗∗∗P < .001) that worsened under hypoxia (∼39%; ∗∗∗P < .001 vs normoxia and HbAA). These findings align with evidence that a measurable residual deoxygenated-HbS persists and forms small polymer fibers under normoxia, even at arterial oxygen saturations above 90% to 95%,^37-39 generating sufficient membrane tension to activate Piezo1 channel, elicit subclinical Ca²⁺ influx, and subtly impair deformability.^40,41 Under hypoxic stress, HbS polymer formation mechanically deforms the RBC membrane to a far greater extent than under normoxia, an effect reflected in our data by markedly increased OI under hypoxia. This mechanical deformation directly activates the mechanosensitive Piezo1 Ca²⁺ influx channel, leading to accumulation of [Ca²⁺]_total that exceeds the capacity of PMCA Ca²⁺ efflux, and thus triggers the Ca²⁺-sensitive Gardos (K⁺) channel.^5,42 The resulting K⁺ efflux and osmotic water loss dehydrate the cell, further stiffening the membrane and exacerbating sickling, thus tightly linking hypoxia-induced polymerization to disrupted Ca²⁺ homeostasis and the dehydration pathway in SCD. The cascade of intracellular Ca²⁺ overload, calpain-mediated cytoskeletal breakdown, and Band 3 phosphorylation culminates in which turn, reduces RBC deformability.^8,41 

All 3, voxelotor, GBT021601, or PKR activator treatments, significantly reduced HbSS OI toward HbAA levels in both normoxic and hypoxic conditions but via distinct molecular routes, demonstrating restoration of RBC deformability. Voxelotor and GBT021601 stabilize the oxygenated (R-state) form of HbS via Schiff-base interactions, increasing Hb-O₂ affinity, reducing the deoxy-HbS pool under normoxia, and preventing polymer formation under hypoxia.^43,44 PKR activators mitigate sickling through distinct but complementary mechanisms: diverting from the Rapoport-Leubering shunt and thereby, lowering 2,3-DPG, which indirectly raises Hb-O₂ affinity, and modulates ATP levels.^35,45 

HbSS RBCs have been reported to exhibit elevated 2,3-DPG, reduced ATP, [Ca²⁺]_total, and elevated Band 3 tyrosine phosphorylation levels^21,46,47 (Figures 5A and 6A). Elevated 2,3-DPG reduces Hb-O₂ affinity, thereby promoting HbS polymerization.⁴⁸ Additionally, increased 2,3-DPG interacts with spectrin and other cytoskeletal components, enhancing membrane viscoelasticity, thus contributing to impaired erythrocyte deformability.⁴⁹ Reduced ATP levels adversely affect the activity of the PMCA pump, leading to elevated intracellular calcium, which triggers downstream phosphorylation of membrane proteins such as Band 3, destabilizing membrane-cytoskeletal interactions.^8,21,50 We systematically examined intracellular ATP, 2,3-DPG, [Ca²⁺]_total, and Band 3 tyrosine phosphorylation under both normoxic and hypoxic conditions. Voxelotor did not alter ATP or 2,3-DPG in HbSS RBCs. GBT021601 did not alter ATP but significantly reduced 2,3-DPG under normoxia and hypoxia. Interestingly, despite their shared R-state stabilization of HbS, the divergent effects of GBT021601 and voxelotor on 2,3-DPG underscore distinct interactions with RBC metabolism.^15,44 GBT021601’s potential to reduce 2,3-DPG suggests it also alters the Rapoport-Leubering shunt. Preclinical studies have reported that GBT021601 achieves higher Hb occupancy at lower doses than voxelotor, which can translate into more potent inhibition of polymer formation and a concomitant reduction in 2,3-DPG that may further enhance Hb-O₂ affinity and membrane stability.¹⁵ PKR activators lowered 2,3-DPG in both conditions and increased ATP only under hypoxia. Previous reports show ATP increases with PKR activation under normoxia in vitro¹⁷ and murine SCD models¹⁶ after 10 hours, which was not significant in 6 hours of incubation in our normoxic study. This dichotomy can likely be explained that the ATP increase may manifest after prolonged treatment exposure under normoxia, whereas a decrease in 2,3-DPG reflects the most immediate effect of PKR activators.

All 3 treatments reduced intracellular [Ca²⁺]_total levels in HbSS RBCs via 2 complementary mechanisms without changing [Ca²⁺]_total levels in HbAA RBCs: (1) inhibiting Piezo1 directly or indirectly and (2) modifying [Ca²⁺]_total efflux pathways, either likely by preserving residual PMCA activity via membrane stabilization or by enhancing PMCA activity via increasing ATP or by activation of alternative Ca²⁺ clearance pathways. Piezo1 inhibition with GsMTx4 induced further [Ca²⁺]_total reduction, while PMCA inhibition with vanadate increased [Ca²⁺]_total but was reversed by drug effects, confirming that both reduced influx and preserved efflux are essential. Functionally, these corrected [Ca²⁺]_total balance further decreased OI even in the presence of a channel modulator. In interpreting our GsMTx4 results, it is important to recognize that GsMTx4 broadly alters membrane tension and mechanics, affecting multiple mechanosensitive channels beyond Piezo1 alone.^51,52 Nevertheless, GsMTx4 has been widely adopted as a functional Piezo1 inhibitor in several RBC studies.^25,41 Thus, our observed reductions in [Ca²⁺]_total likely reflect Piezo1 inhibition but may also involve other mechanosensitive pathways. Future studies using more selective Piezo1 inhibitors or direct channel assays are needed to precisely define Piezo1’s role in RBC–calcium homeostasis.

Membrane mechanics play a pivotal role in regulating PMCA function.⁵³ Improving RBC membrane integrity indirectly preserves PMCA activity. Central to the mechano-metabolic link is the Band 3 membrane protein: dephosphorylation of its tyrosine residues enhances ankyrin binding, thereby tightening spectrin-actin linkages and stabilizing the lipid bilayer.^8,53,54 A stabilized membrane-cytoskeleton network preserves an ordered lipid microdomain in which PMCA resides, thereby facilitating Ca²⁺ efflux.⁵³ Band 3 phosphorylation also plays a critical role in RBC homeostasis by modulating membrane-cytoskeleton interactions and cell shape, key determinants of deformability and life span. Phosphorylation of Band 3 disrupts its association with spectrin, actin, and protein 4.1, undermining membrane integrity, impairing ion transporter regulation, and impacting intracellular [Ca²⁺]_total balance and metabolic flux.^21,53,55 

In this study, voxelotor, GBT021601, and the PKR activator each significantly lowered Band 3 phosphorylation in HbSS RBCs under both normoxic and hypoxic conditions, suggesting improved membrane cytoskeletal stability. Importantly, none of the agents increased ATP levels, a primary substrate for PMCA activity, except PKR activators under hypoxia, suggesting modification of alternative Ca²⁺ clearance pathways. By preserving PMCA’s lipid-protein microenvironment, membrane stabilization could likely maintain the residual PMCA pump activity despite vanadate inhibition. A more ordered bilayer may unmask secondary Ca²⁺ efflux pathways such as Na⁺/Ca²⁺ exchange or other low-affinity transporters and also alter the buffering through cytosolic Ca²⁺-binding proteins.^56,57 Additionally, studies where ATP was depleted from RBCs have reported a residual PMCA pump activity fueled by ATP generated from the breakdown of the abundant intracellular 2,3-DPG pool.^27,58 Since we observed a reduction in 2,3-DPG with GBT021601 and PKR activators under normoxia, we speculate that GBT021601 and PKR activators, by increasing Hb-O₂ affinity, promote the catabolism of 2,3-DPG, thereby fueling ATP locally to support residual PMCA efflux activity and contributing to reduced [Ca²⁺]_total and improved deformability.

Elevated [Ca²⁺]_total in HbSS RBCs has been linked to calpain activation,^8,47 which cleaves and inactivates protein tyrosine phosphatase 1B (PTP1B), destabilizing interactions among Band 4.1R, Band 3, actin, and spectrin.^8,33,55,59 Our selective reduction of Band 3 phosphorylation in HbSS but not HbAA RBCs, under both normoxic and hypoxic conditions, suggests a link between restored [Ca²⁺]_total homeostasis to cytoskeletal stabilization, and thus improved deformability in SCD erythrocytes.

Under normoxia, elevated 2,3-DPG drives low-level HbSS polymer formation, and residual polymers generate membrane tension that activate Piezo1 channels and induce mild Ca²⁺ influx; this limits the capability to ATP-driven PMCA to Ca²⁺ efflux and permits calpain activation and Band 3 phosphorylation, thus reducing deformability (Figure 5A); Under hypoxia, accelerated polymerization dramatically heightens membrane stress and Piezo1 activation, driving an increase in [Ca²⁺]_total and, compounded by ATP depletion, further reducing the activity of PMCA-mediated Ca²⁺ efflux and promoting cytoskeletal destabilization (Figure 6A). Voxelotor counters this cycle by stabilizing HbSS in its oxygenated R-state, reducing residual polymer formation without altering 2,3-DPG or ATP, while inhibiting Piezo1 and modifying Ca²⁺ clearance pathways (Figures 5C and 6C). GBT021601 shares R-state stabilization but also robustly lowers 2,3-DPG, suppressing polymers, inhibiting Piezo1, and further preserving the residual PMCA activity or by modifying other Ca²⁺ clearance pathways (Figures 5D and 6D). PKR activators complement these effects by reducing 2,3-DPG, Hb-O₂ affinity, inhibiting Piezo1 (Figure 5B), and uniquely elevating ATP under hypoxia (Figure 6B), which increases PMCA activity, thereby preserving membrane-cytoskeleton integrity. Together, these 3 therapeutic interventions converge to prevent calpain activation, limit Band 3 phosphorylation, and re-establish membrane cytoskeleton stability, restoring HbSS RBC deformability under both normoxic and hypoxic stress.

RBC dehydration and deformability are interconnected in SCD pathophysiology, with dehydration further reducing deformability by elevating intracellular HbS concentration, thereby promoting HbS polymerization and increased HbSS RBCs' rigidity.^1,7 In our study, treatment with voxelotor, GBT021601, and PKR activators led to significantly decreased OI, suggesting improved RBC deformability underpinned by enhanced membrane stability, as demonstrated by reduced Band 3 phosphorylation. This effect is likely mediated through inhibition of Piezo1 and modulation of Ca²⁺ clearance pathways, suggesting that the primary therapeutic benefit is restored deformability. However, given the close coupling of cellular volume and shape changes in HbSS RBCs, we cannot fully exclude a concurrent benefit from improved RBC hydration status.

Interestingly, the combination of voxelotor and a PKR activator did not produce additive or synergistic improvements in membrane integrity or Band 3 tyrosine phosphorylation under either normoxic or hypoxic conditions (supplemental Figure 5A-B), suggesting that these agents share overlapping mechanisms that limit the benefit of their combined use. Our findings have underscored the importance of rational combination strategies that employ therapies with mechanistically distinct modes of action, such as hydroxyurea, which induces fetal hemoglobin, or emerging gene-based interventions, to achieve more robust improvements in RBC deformability and clinical outcomes. Although several animal studies and ongoing clinical trials have already demonstrated the individual clinical benefits of these agents, our study reinforces their mechanistic promise in parallel and supports further aggressive therapeutic development.

Importantly, we selected ex vivo concentrations of 600 μM for voxelotor and GBT021601 and 200 μM for the PKR activator based on prior human RBC assays demonstrating meaningful target engagement.^15,17,44 Our insights robustly capture changes in [Ca²⁺]_total, 2,3-DPG metabolism, and Band 3 phosphorylation; however, future studies may need to provide detailed dose-response experiments, in vivo models, and direct drug-target interaction assays to confirm these pathways at clinically relevant exposure levels and to guide rational dosing strategies. The clinical relevance of our findings is profound, as they pave the way for regimens that could dramatically reduce SCD morbidity by concurrently correcting key pathophysiological derangements and improving critical markers of anemia, hemolysis, and vaso-occlusive complications. Moreover, these results refine our understanding of the biochemical pathways underlying sickle-RBC deformability and inform the rational design of next-generation combination therapies to further mitigate vaso-occlusive complications and improve patient outcomes in SCD.

The authors acknowledge the following funding sources: National Heart, Lung, and Blood Institute (awards R41HL172662 and R44HL140739); National Institute of Biomedical Imaging and Bioengineering (award P41EB021911); National Institute of Allergy and Infectious Diseases (award U01AI176469); and Case-Coulter Translational Research Partnership pilot award. The authors acknowledge institutional support from the Blood Research Center at The University of North Carolina. Z.S. acknowledges the American Society of Hematology Graduate Hematology Award.

This article’s contents are solely the authors' responsibility and do not necessarily represent the official views of the National Institutes of Health.

Contribution: Z.S. and U.A.G. conceptualized the idea; Z.S., S.O., B.Z., T.L., F.W., and A.C. assisted with the planning and execution of clinical sample acquisition and testing, including the development of human participant research protocol, patient recruitment, and blood sample collection; Z.S., P.F., A.C., L.W., Y.D., H.B., R.A., J.L., Y.M., and U.A.G. performed the experiments, investigations, data analysis, and visualization; U.A.G. supervised the work, acquired funding, and administrated the project; Z.S. drafted the original manuscript; and all authors reviewed and edited the manuscript.

Conflict-of-interest disclosure: U.A.G. and Case Western Reserve University have financial interests, including licensed intellectual property, stock ownership, research funding, employment, and consulting, in Hemex Health Inc (which offers point-of-care diagnostics for hemoglobin disorders and anemia), BioChip Labs Inc (which offers commercial clinical microfluidic biomarker assays for inherited or acquired blood disorders), and XaTek Inc (which offers point-of-care global assays to evaluate the hemostatic process). U.A.G. has financial interests, including licensed intellectual property, stock ownership, research funding, employment, and consulting in DxNow Inc (which offers microfluidic and bioimaging technologies for in vitro fertilization, forensics, and diagnostics). The competing interests of Case Western Reserve University employees are overseen and managed by the Conflict of Interests Committee according to a Conflict-of-Interest Management Plan. The remaining authors declare no competing financial interests.

Correspondence: Umut A. Gurkan, Case Western Reserve University, 10900 Euclid Ave, Cleveland, OH 44106; email: umut@case.edu.