A novel mouse model of hemoglobin SC disease reveals mechanisms underlying beneficial effects of hydroxyurea Open Access

Sickle cell disease (SCD), a significant global health problem, which affects millions worldwide,¹ results from a mutant HBB gene where valine is encoded instead of glutamic acid at the 6th codon (HBB^E6V), and produces “sickle” hemoglobin (HbS) instead of normal adult Hb (HbA).² HbS polymerizes upon deoxygenation, resulting in sickle-shaped red blood cells (RBCs). The homozygous form (HbSS) is the most severe and common form, comprising 60% of SCD. Sickle cell HbC (HbSC) disease, compound heterozygosity for HbS, and HbC (where lysine is encoded in place of glutamic acid at the 6th codon of the HBB gene, HBB^E6K), is the second most common SCD genotype, comprising 30% of SCD (40%-50% of SCD in the Caribbean and Ghana).^1,3-7 HbC crystallizes,⁸ and promotes KCl-cotransporter activity, causing RBC dehydration/xerocytosis.^9,10 Xerocytosis increases the intracellular HbS concentration, promoting HbS polymerization and RBC sickling, despite HbS being in “trait-state” (50% of Hb).⁸ Hence, HbSC-SCD has the same symptoms as HbSS-SCD.

Recent studies have challenged the notion that HbSC-SCD is milder than HbSS-SCD,^5,11 because it leads to significant morbidity and mortality.¹² Besides, HbSC-SCD has some distinctive features, attributed to increased blood viscosity¹¹: 33% to 43% of patients with HbSC develop proliferative retinopathy earlier (aged 15-24 years) compared with 3% to 14% of patients with HbSS (aged 25 to 39 years).^8,13-15 Patients with HbSC-SCD also have a similar degree of avascular necrosis as HbSS-SCD patients and a higher frequency of splenic sequestration than those with HbSS-SCD.^11,16 Understanding the unique properties of HbC and mechanism(s) of xerocytosis is critical to developing effective therapies for HbSC-SCD.¹⁷ 

Although extensive research/treatment strategies have focused on HbSS-SCD, patients with HbSC-SCD have mostly been excluded,^16,18 and therefore deprived of specific therapies. For instance, they are excluded from transformative genetic therapies, because of the lack of preclinical models to assess their efficacy in the presence of HbC. A prospective randomized trial of hydroxyurea (HU) for HbSC-SCD was reported in 2024,¹⁹ 37 years after being approved for HbSS-SCD.^19-21 Perplexingly, both in this trial,¹⁹ and several previous retrospective/nonrandomized studies,^22-27 HU resulted in negligible/small increases in fetal Hb (HbF) in patients with HbSC-SCD, but showed highly significant clinical benefit in SCD symptoms^19,22-27 through an unknown mechanism. In contrast, in HbSS-SCD, HU not only upregulates HbF at significantly higher levels, but its levels determine the degree of correction of SCD symptoms.²⁸ Generation of humanized SCD mice (Townes²⁹ and Berkeley³⁰ mice), which exclusively express human Hb, phenocopying human HbSS-SCD, revolutionized research and translation in HbSS-SCD. No animal model exists for HbSC-SCD, creating a large knowledge gap in identifying mechanisms or developing targeted/effective therapies for HbSC-SCD.

Herein, we generated HbSC mice on the same genetic background as HbSS mice, except changing the HbS to HbC mutation on 1 allele. This novel HbSC model phenocopies human HbSC-SCD, allowing direct comparisons of therapies currently effective in HbSS-SCD to HbSC-SCD. We found unique insights into the HbF-dependent and HbF-independent mechanisms by which HU improves HbSC-SCD.

The Townes HbSS (hα/hα::hβ^S/hβ^S) mice were obtained from Tim Townes^29,31 and HbAA (hα/hα::hβ^A/hβ^A) mice, derived by Tim Townes, were kindly provided by Cheryl Hillary (Madison, WI). HbAS females were mated with HbSS/HbAA males to derive HbSS/HbAA mice. Mice were housed in the Cincinnati Children’s Hospital Medical Center vivarium and studied using institutional animal care and use committee–approved protocols.³² 

Adult HbSS/HbSC mice were given HU 50 mg/kg body weight (BW) intraperitoneally, 3 days per week for 4 weeks. Pregnant HbSC females (mated with HbSS males) were treated with HU (25 mg/kg BW per day) on days 15 through 21 of gestation by gavage. Pups (HbSC and HbSS) born from HU-treated females were given HU 50 mg/kg BW intraperitoneally, 3 days per week. Controls (pregnant dams, adult mice, and pups) received phosphate-buffered saline (PBS).

Blood samples from patients with HbSS/HbSC and controls were obtained through an institutional review board-approved protocol.

RBC lysates were fractionated through ion-exchange high-performance liquid chromatography (HPLC; Alliance-2690), using a PolyCAT-A column (3.54CT0510; Poly-LC Inc).³³ 

Complete blood counts were analyzed using Element HT5 (Heska) or ADVIA-2120 hematology analyzers. Reticulocytes were assessed either by staining with 0.01 mg/mL thiazole orange (BD Biosciences) on the fluorescence-activated cell sorter (FACS) BD FACSCanto flow cytometer (BD Biosciences), or via the Advia-2120 (Siemens Healthineers).

Mice were given 3 mg N-hydroxysulfosuccinimidobiotin (Sigma) IV, followed by serial microbleeds for 2 weeks. RBCs labeled with peridinin-chlorophyll-protein–conjugated streptavidin were analyzed on the BD FACSCanto flow cytometer.^33,34 

RBC deformability was measured on the Lorrca Maxsis, oxygen-gradient or osmoscan ektacytometer (RR Mechatronics, Zwaag, The Netherlands).^33,34 

Whole blood viscosity was measured at different shear rates on the DV3T rheometer (Ametek Brookfield Engineering).

Plasma vascular cell adhesion molecule 1 (VCAM-1) was measured using an enzyme-linked immunosorbent assay (Abcam), and tumor necrosis factor α using the mouse Luminex assay (R&D Systems).

Anesthetized mice received an intracardiac fluorescein isothiocyanate–dextran (10 kDa; Sigma) and were euthanized after 2 minutes; the eyes were removed, fixed in 4% paraformaldehyde (11 minutes, 4°C), and washed with PBS. Retinas were dissected, fixed in 4% paraformaldehyde (1.5 hours, room temperature), washed, and cut into leaflets for flat-mounting. Fluorescent images were captured using a Nikon C2 confocal microscope.

RBCs were fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer, followed by 1% osmium tetroxide,³⁵ then dehydrated through a graded ethanol series, mounted on aluminum specimen stubs, and sputter-coated with 10 nm of carbon (Safematic CCU-010; Rave Scientific). The specimens were imaged using a Zeiss FE-SEM (Supra 35 VP; Zeiss).

Organs were fixed in 10% neutral-buffered formalin for 72 hours, then ethanol for 72 hours, and embedded and sectioned (4 μm).^36,37 Hematoxylin and eosin, Masson trichrome, and Prussian-Blue staining was performed by the Cincinnati Children’s Hospital Medical Center Pathology Core Services.

Mice were challenged with 8% hypoxia for 10 hours followed by 2 hours of reoxygenation in ambient air to mimic acute vaso-occlusive crisis, as previously described.³⁸ 

Whole blood samples were stained with anti-Ter119, anti-CD41, and anti-CD45 antibodies (BD Biosciences) followed by washes and resuspension in 0.05 mM CM-H2-DCFDA (5′,6′-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate; Invitrogen) at 37°C, 30 minutes; resuspended in 100 to 150 μL FACS buffer; and acquired on the flow cytometer.

Whole blood (3 μL) was washed with Hanks balanced salt solution +0.1% bovine serum albumin, resuspended to 5% hematocrit, and deposited onto polylysine-coated coverslips. After 20 minutes, adherent cells were fixed (1% paraformaldehyde, 0.0075% glutaraldehyde), washed, permeabilized, and blocked (permeabilization/wash buffer, 1% bovine serum albumin). Cells were stained with anti–ferryl Hb antibody (1:100) for 1 hour,³⁹ followed by secondary antibodies (anti-mouse immunoglobulin G–AlexaFluor488, 1:300; Ter119-BV421, 2:100) for 30 minutes. Coverslips were mounted with Fluoromount-G, dried, and imaged on an EVOS 7000 microscope (60×). Images were analyzed with EVOS software.

Fresh EDTA-anticoagulated blood was mixed with 0.5% methyl violet (2:3 ratio), incubated for 60 minutes at room temperature, and smeared on glass slides for examination under a 100× oil-immersion objective (Nikon microscope). Positive controls used human blood treated with 1 mg/mL acetyl-2-phenylhydrazine (1:20 blood-to-reagent ratio) at 37°C for 2 hours.

RBC ghosts were processed for mass spectrometry for β-globin^93Cys–containing peptides, quantified using Proteome Discoverer 3.0⁴⁰ (details provided in the supplemental Methods; supplemental Figure 6, available on the Blood website).

Groups were compared using analysis of variance, and pairs using Mann-Whitney U test using the GraphPad Prism 10.0 (San Diego, CA). P < .05 was considered significant.

Fertilized eggs from the Townes HbSS×HbAS mouse timed-crosses were retrieved, electroporated with single guide RNA–CRISPR–associated protein 9 ribonucleoprotein and single-stranded HBB donor DNA carrying the HBB^E6K mutation and implanted into pseudopregnant females. Pups were screened by polymerase chain reaction amplification of the targeted locus and confirmed by Sanger sequencing. Founder HbSC mice exhibited both the HBB^E6K (aAG) and the HBB^E6V (GtG) alleles (supplemental Figure 1). HPLC on blood from founder mice confirmed HbC and HbS protein expression (Figure 1A-C). Germ line transmission was ascertained before the expansion of the HbSC colony. HbSC (expressing HBB^E6V/E6K) mice were compared with HbSS (expressing HBB^E6V/E6V) and HbAA (expressing HBB^E6/E6) controls.

HbSC mice were anemic compared with HbAA mice. Although their Hb and RBC counts were higher than HbSS mice, their hematocrits were lower (Figure 1D-F), suggesting that HbSC RBCs were more dehydrated than HbSS. Indeed, their mean corpuscular Hb concentration (MCHC) was higher than HbSS/HbAA RBCs (Figure 1G). HbSC mice showed less pronounced reticulocytosis than HbSS (Figure 1J), suggesting lesser hemolysis. Because mean corpuscular volume (MCV) is higher in reticulocytes than RBCs, MCV was highest in HbSS mice (supplemental Figure 2A-B). Despite much higher reticulocytosis, HbSC erythrocytes had a similar MCV as HbAA RBCs because of xerocytosis. HbSC mice also had leukocytosis, albeit milder than HbSS mice, consistent with chronic inflammation in SCD (Figure 1H).

Similar RBC and white blood cell features were present in patients with HbSC-SCD when compared with normal (HbAA) controls and patients with HbSS-SCD, except for the MCV, which was higher in HbSC and HbSS mice from the high reticulocytosis compared with patients with SCD (supplemental Figure 3A-J); and platelets, which were normal in human patients with HbSS/SC but lower in HbSS and HbSC mice (Figure 1I).

HbSC reticulocyte parameters, analyzed using the ADVIA-2120 analyzer showed similar reticulocyte counts to those obtained by flowcytometry (Figure 1J-K). The corpuscular Hb concentration mean of reticulocytes (CHCM_Retic) was highest in HbSC reticulocytes, significantly exceeding HbAA and HbSS, indicating xerocytosis is also present in HbSC reticulocytes (Figure 1L). Furthermore, dense RBC and reticulocytes were higher in HbSC vs HbSS mice (supplemental Figure 2C-D).

HbSC RBCs had a longer half-life (9.2 days) than HbSS RBCs (2.6 days; Figure 1M). Overall, the RBC parameters of the HbSC mice were similar to those in patients with HbSC (supplemental Figure 3A-I). However, it is important to note that many patients with HbSS-SCD were taking HU and, therefore, may have a less severe hematological profile.

SEM showed severely dehydrated HbSC erythrocytes folded into “pita shapes,” “pretzel shapes,” or with a double dimple from membrane folding around scant cytoplasm (Figure 2B,D-F). HbSS controls showed irreversibly sickled RBCs, larger-sized reticulocytes, and some folded dehydrated RBCs; HbAA controls had classic doughnut-shaped RBC (Figure 2A,C).

Functional analyses of hydration and sickling were performed on HbSC, HbSS, and HbAA mouse RBCs (mRBCs) and corresponding human RBCs (hRBCs) using osmoscan- and oxygen-gradient ektacytometry. In the osmoscan, RBC deformability (measured as their elongation index [EI]) was determined under fixed shear stress but varying osmolarities. Normal RBCs deform maximally (EI_max) in iso-osmolar conditions and have reduced deformability under hyperosmolar (dehydrated) or hypo-osmolar (swollen) conditions. Xerocytosis causes a left-shifted EI curve, with EI_max occurring at lower osmolarity. Xerocytosis is quantified by O_hyper, the hyperosmolarity at which EI is half of EI_max.

HbSC and HbSS mRBCs and hRBCs had left-shifted osmoscan curves than their normal HbAA counterparts, with the highest shift seen with HbSC RBCs. The corresponding hRBCs had an identical pattern (Figure 2G-H). Quantitatively, the O_hyper of mRBCs and hRBCs was lowest in HbSC, intermediate in HbSS and highest in HbAA RBCs (Figure 2K-M). Hence, the presence of HbC resulted in more severe xerocytosis in HbSC than HbSS RBCs, a hallmark of HbSC-SCD pathobiology.

Sickling kinetics of murine and human HbSC RBCs, along with HbAA and HbSS controls, were determined using oxygen-gradient ektacytometry.⁴¹ Here, EI is measured in RBCs subjected to constant shear stress with increasing hypoxia followed by reoxygenation.⁴² HbSS RBCs retain maximum deformability (EI_max) under normoxia until the partial pressure of oxygen (pO₂) drops between 40 and 60 mm Hg, when HbS polymerization begins. The pO₂ level at which EI drops to 95% of EI_max is termed point-of-sickling (PoS). Hypoxia beyond PoS results in rapid HbS polymerization, causing an exponential drop in EI to a minimum (EI_min). Reoxygenation results in HbS depolymerization, and HbSS RBCs gradually regain deformability.

PoS was lower, and EI_max and EI_min higher, in HbSC RBCs compared with HbSS RBCs in both mice and humans (Figure 2I-J,L,N), indicating milder sickling kinetics in HbSC RBCs because of the presence of 50% nonsickling HbC. As expected, HbAA RBCs had the same EI under all oxygen concentrations. Notably, HbSC RBCs (both murine and human) did not regain their baseline deformability upon reoxygenation, indicating some degree of irreversible damage after hypoxia-reoxygenation, conceivably because of a detrimental effect of HbC.

We simulated sickle acute vaso-occlusive crisis (VOC) in HbSC mice with the hypoxia-reoxygenation (H/R) assay,^43,44 which has been shown to decrease Hb and platelets, and increase soluble VCAM, a marker of endothelial injury and inflammation (tumor necrosis factor α) in HbSS.^45-47 HbSC mice showed a similar drop in RBCs and platelets and a similar increase in soluble VCAM-1 to that in HbSS mice, although their baseline levels were different from HbSS mice (Figure 3A-F). Hence, experimentally induced VOC results in comparable alterations in hematological and endothelial parameters in HbSS and HbSC mice.

H/R caused further xerocytosis of HbSC RBCs, which were the most dehydrated RBCs before H/R (Figure 3G-H). As expected, HbSS RBCs were also significantly dehydrated, after sickling-induced shape change, which promotes cation and water loss.⁴⁸ SEM after H/R confirmed the severe dehydration in HbSC RBCs, evidenced by pronounced membrane folding due to reduced cytoplasmic volume (Figure 3I-J).

Despite milder hemolysis and sickling than HbSS mice, significant organ pathology was observed in 7-month-old HbSC mice, which was either comparable with, or slightly milder than, age-matched HbSS mice (Figure 4; supplemental Tables 1-5). The spleen is an organ of extramedullary stress erythropoiesis in mice, unlike humans. Hence, splenomegaly in HbSC mice was less prominent than in HbSS mice, from lower anemia and hemolysis (Figure 4B). Histologically, the spleens of HbSC mice had remarkably more sinusoidal congestion but had less white-pulp atrophy than in HbSS mice (Figure 4A,C-D; supplemental Table 1). Hepatomegaly was comparable in HbSC and HbSS mice, with foci of infarction and congested veins filled with sickled RBCs, extramedullary hematopoiesis, and iron deposition. (Figure 4E-G). Portal inflammation and fibrosis were higher in HbSS, whereas vascular congestion was higher in HbSC livers (Figure 4H-K; supplemental Table 2). Kidneys showed mild glomerular enlargement with congested capillary tufts in HbSC and HbSS mice. There was vascular congestion and hemosiderin deposition in HbSC kidneys, but we did not observe significant fibrosis (Figure 4L-N; supplemental Table 3). Urine osmolarity was reduced in both HbSC and HbSS mice (Figure 4O). Cardiac sections showed hypertrophied cardiomyocytes in HbSC and HbSS mice, although we did not observe evidence of fibrosis (Figure 4P; supplemental Table 4).

Bone histology of the legs in 12-month-old HbSC and HbSS mice showed no evidence of avascular necrosis, although the number of animals in the HbSC group was limited (n = 3; Figure 4Q-R). Of 6 HbSS mice, 1, however, had evidence of infarction of a portion of femoral diaphysis along with coagulative necrosis of the underlying bone marrow (Figure 4S-U; supplemental Table 5).

Retinal vasculature in 7-month-old HbAA mice was normal. However, 4 of 6 (67%) age-matched HbSC mice showed abnormal vascular growth, vascular leakage, and “comma”-shaped vessels at the distal end of the retinal vessels (Figure 5A-G). Of 6 7-month-old HbSS mice, 1 (17%) showed retinopathy (P < .05).

The higher frequency of retinopathy in HbSC-SCD than HbSS-SCD has been attributed to higher blood viscosity conferred by the dense dehydrated HbSC RBC. Indeed, whole-blood viscosity, measured at different shear rates in HbSC, HbSS, and HbAA, showed that both HbSC mice and humans had the highest blood viscosity, followed by HbSS, and then HbAA controls (Figure 5H-I).

In summary, characterization of the hematological parameters, functional analysis of sickling, RBC hydration and VOC, and organ pathology showed that the HbSC mouse model has similar features of HbSC-SCD as human patients except for minor differences in platelets and RBC-turnover rates. Using this model, we studied the mechanism of action by which HU improves clinical symptoms in HbSC-SCD with small increases in HbF and contrasted these with HbSS-SCD.

The humanized Townes HbAA/HbSS mice carry both the HBG and the HBB/HBB^E6V genes and regulatory elements. They switch from HBG gene transcription to HBB or HBB^E6V gene transcription in midgestation, when hematopoiesis moves from the yolk sac to the fetal liver.²⁹ HbF levels in RBCs start dropping thereafter, with 20% HbF containing RBCs, determined using flow-cytometry (F cells) seen at postnatal day 2 (P4), declining to undetectable levels after P14.⁴⁹ Efforts at reactivating HbF in adult HbSS mice with HU have failed.^28,49 The lack of HbF-induction with HU in adult HbSS (and likely HbSC) mice makes them ideal for studying the non-HbF effects of HU.

We first titrated the dose of HU in adult HbSS and HbSC mice from 25 to 100 mg/kg BW. HU at >50 mg/kg BW every other day showed significant cytopenia, toxicity, and lethality. Notably, we did not observe higher toxicity in HbSC mice. Adult HbSS and HbSC mice were administered HU 50 mg/kg BW thrice weekly for 4 weeks (HU-50), along with PBS controls (HU-0). Leukocytes and reticulocytes decreased in HU-50 HbSS/HbSC mice, as in patients; MCV decreased from the reduced reticulocytosis caused by HU. However, no upregulation of HbF was seen by HPLC/F-cell analysis (Figure 6A-B; supplemental Figure 4A-G). MCHC and CHCM_Retic measurements showed no improvement in HbSC RBC/reticulocyte hydration (Figure 6C). However, shear-stress ektacytometry showed a significant improvement in HbSC RBC deformability (Figure 6D; supplemental Figure 5A). Oxygen-gradient ektacytometry confirmed this improved membrane deformability, reflected in higher EI_max in HbSC and HbSS RBCs. Interestingly, PoS was also lower in HbSC/HbSS RBCs (Figure 6E; supplemental Figure 5B).

Mechanistically, HU lowered RBC but not platelet/white blood cell reactive oxygen species (Figure 6F; supplemental Figures 5A and 7A-B). Furthermore, ferryl Hb, detectable in HU-0 HbSC/HbSS RBCs, was remarkably reduced in the HU-50 group (Figure 6J; supplemental Figure 5G). Consequently, Heinz bodies were also significantly reduced in the HU-50 group, both in HbSC and HbSS RBCs (Figure 6G; supplemental Figures 5D and 8A-F). To assess Hb oxidation, we analyzed the association and the oxidation status of β-globin^93Cys (β^93Cys; a position in β-globin prone to oxidation) to the RBC membrane by mass spectrometry. RBC membrane ghosts exhibited a higher association of the unmodified β^93Cys peptides and trioxidized β^93Cys peptides when compared with HbSS mice (Figure 6H-I; supplemental Figure 5E-F). We also observed a trend toward a lower association of the unmodified β^93Cys peptides and the trioxidized β^93Cys peptides with HU (P = .1). In vitro experiments have shown that HU nitrosylates β^93Cys peptides, protecting it from oxidation⁵⁰; we did not detect β^93Cys peptide nitrosylation with in vivo administration of HU to mice, whereas we confirmed the position and site of all detected β^93Cys peptide modifications by peptide-fragmentation sequencing (supplemental Figure 6A-B).

Taken together, the HbF-independent effects of HU on HbSC and HbSS RBCs include reduced oxidative stress and oxidized Hb, reduced membrane damage that results in improved membrane deformability, and sickling at lower pO₂, but no improvement in HbSC RBC hydration.

Using the human analogy of initiation of HU early in infancy for high level of HbF expression,⁵¹ we exposed pregnant HbSC mice from HbSS×HbSC timed-matings to 25 mg/kg BW per day HU in the last week of gestation to allow HU exposure to unborn mice while HBG gene poised for being switched-off, and continued it postnatally (50 mg/kg, 3 days per week; Figure 7A). We observed heterocellular HbF expression in HbSC and HbSS HU-50 mice at 4 weeks of age, which declined to control levels by age 8 weeks of age. Notably, HbF and F cells in HbSC mice were nearly half compared with those in HbSS, despite the same dose given concurrently to these littermates (Figure 7B). Upregulation of HbF, even at these low HbF levels (∼1% HbF) in the 4-week-old HbSC mice, nevertheless, resulted in significant improvement in HbSC RBC hydration, as evidenced by lower MCHC and higher MCV (Figure 7C-D), which reverted with the decline in HbF at age 8 weeks to levels seen in HU-0 controls. No change in MCHC occurred in 4-week-old HbSS mice. The MCV increased in both HbSC and HbSS RBCs, but the increase was significantly greater in HbSC RBCs than in their respective controls. No significant change in Hb/hematocrit (HCT) in HbSC mice occurred (supplemental Figure 9A-B).

We terminally bled HbSC and HbSS P7 pups, when HbF levels should be higher than at 4 weeks of age, to obtain sufficient blood for determining HbF and sickling kinetics. Similar proportions of F cells were present in HbSC (55%) and HbSS (57%) HU-50 P7 pups (Figure 7E-F). However, the HbF levels were nearly 3 times higher in HbSS (19%) than HbSC (7%) P7 pups, despite both getting the same dose of HU concurrently (Figure 7G-H).

Furthermore, oxygen-gradient ektacytometry demonstrated complete abrogation of sickling in HbSC RBCs at 7% HbF (PoS occurred at 20 mm Hg at 1% HbF [HU-0] but PoS was not detectable at 7% HbF; Figure 7I). In HbSS HU-0 P7 pups, the PoS occurred at ∼29 mm Hg with 6% HbF (HU-0), and decreased to 17 mm Hg in HU-50 P7 pups with 19% HbF (Figure 7F,H,J).

HbSC-SCD has remained a neglected SCD genotype compared with HbSS-SCD, depriving a third of patients with SCD of transformative/disease-modifying therapies because of lack of an animal model. Adults with HbSC have substantial morbidity and mortality from severe disease.^8,11 The HbSC mice developed herein exhibited erythrocyte characteristics phenocopying patients with HbSC-SCD: milder anemia, lower RBC turnover, longer RBC half-life, but higher blood viscosity than HbSS-mice, and prominent RBC xerocytosis. HbSC-RBC sickling kinetics resembled those in RBCs of patients with HbSC-SCD. Notably, humanized SCD mice have a more severe RBC phenotype than human patients (HbSS RBC half-life, 2-3 days; 50% reticulocytes). We observe that HbSC mice also have a more severe phenotype than human patients with HbSC-SCD, with ∼20% reticulocytes. Regardless, the HbSC model generated herein and the HbSS models⁵² are invaluable because they mimic the features of human SCD.

Although HbSC RBCs sickled at lower pO₂ than HbSS RBCs, as is expected from the presence of 50% HbS, unlike 100% HbS in HbSS RBCs, HbSC mice showed a similar degree of increase in inflammatory and endothelial activation markers and drop in Hb/platelets as seen in HbSS mice in experimentally induced VOC. Clinically, VOC in patients with HbSC is no less severe than in patients with HbSS. Additionally, HbSC mice showed: (1) marked splenic sinusoidal congestion compared with HbSS mice, because sinusoids have slow vascular flow and therefore congestion is exacerbated by the high blood viscosity in HbSC mice; and (2) sickle retinopathy occurred more frequently in HbSC mice, as seen in patients with HbSC-SCD.

The identical genetic background of the HbSC and HbSS mice, both being knockins for the human HBA and HBB genes, except for the HBB sixth codon difference on 1 allele, allowed us the unique ability to make direct comparisons of the effects of HU in HbSC vs HbSS mRBCs.

HU was first approved for HbSS-SCD in 1997.^53,54 However, until very recently,¹⁹ only small/retrospective studies of HU in HbSC-SCD had been done,^22-27 which showed perplexingly low/modest increases in HbF, unlike much higher HbF induction in patients with HbSS.^55,56 The largest retrospective study²² and the recently published PIVOT trial¹⁹ (133 and 107 patients with HbSC given HU for 12 months, respectively) also showed modest increases in HbF (2%-4%,²² and 7.6% ± 7.5% with HU at maximum tolerated dose¹⁹). However, despite modest HbF induction, there was significant reduction in vaso-occlusive events in both these studies.^19,22 

Because adult SCD mice do not express HbF with HU,^28,49 we were able to uniquely compare the non-F effects on HbSC and HbSS RBCs in this model. We found that HU decreased oxidative damage to Hb by reducing levels of the toxic oxidative transformation of Hb to the ferryl state (HbFe⁴⁺). Ferryl Hb irreversibly oxidizes β-globin at Cys-93.⁵⁷ Oxidized Hb binds to band 3 protein, clustering band 3,⁵⁰ which then results in Heinz-body formation.^58-60 We observed this phenomenon in both HbSC and HbSS RBCs, although a significantly higher proportion of HbSC RBCs had Heinz bodies than HbSS RBCs. This is consistent with studies that show HbC binds more tightly to erythrocyte membrane and alters its structure in HbCC RBCs⁶¹; its oxidation results in Heinz-body formation.^62,63 Heinz bodies have been similarly shown to form with HbS.⁶³ We show a significant reduction in Heinz bodies in HbSC/HbSS RBCs treated with HU, resulting in improved RBC deformability. Increased Heinz bodies quantitatively correlate with a reduction in RBC deformability.⁶⁴ 

We also observed a reduction in PoS with HU in both HbSC and HbSS mice, without HbF upregulation. In vitro exposure of HbS samples to HU was shown to prevent β^93Cys oxidation and delay HbS polymerization.⁶⁵ We observe an increased abundance of the β-globin peptide containing β^93Cys on HbSC RBC membrane ghosts, that tends to be reduced with HU treatment, and the same phenomenon is seen with oxidized β^93Cys peptides. Finally, our data unequivocally show that HbF is required to improve RBC hydration, because HbSC RBCs show no improvement in hydration with HU when they do not upregulate HbF. Given the known role of band 3 in regulating glycolysis via pyruvate kinase and interactions with Hb, and that pyruvate kinase has 2 cysteines in its catalytic domain, which are prone to oxidation, it is plausible that HU prevents their oxidation, which can be explored in future studies.

It has been known for decades that HU reduces VOC symptoms in patients with HbSS-SCD long before HbF upregulation, which takes 4 to 6 months. The paradigm has been that HU benefits HbSS-SCD because of its HbF-independent effects on RBCs (increased MCV and reducing HbS concentration), leukocytes (reduced leukocytosis and inflammation), and endothelial cells (increasing nitric oxide bioavailability).⁶⁶ Herein, we show the non-F effects of HU directly on HbSC and HbSS RBCs, which also explain the early benefits of HU.

Additionally, we were able to induce HbF in both HbSC and HbSS mice by starting HU administration around the time the mouse HbF switch occurred. To the best of our knowledge, this is the first demonstration of upregulation of HbF with HU to clinically relevant levels in SCD mice (∼7% in HbSC and ∼19% in HbSS mice). Even a minimal increase in HbF with HU improved HbSC RBC hydration (MCV and MCHC), and a modest increase in HbF by 7% ameliorated sickling of murine HbSC RBCs, explaining the clinical benefits seen with low/modest HbF in patients with HbSC-SCD. In cell-free Hb mixing experiments, HbF reduces HbC crystallization and HbS polymerization.⁶⁷ Furthermore, F cells were absent in denser fractions when HbSC RBCs were fractionated on a density gradient.⁸ 

We further show that, despite getting the same exposure to HU, there is nearly threefold higher induction of HbF in HbSS RBCs than HbSC RBCs, despite similar numbers of F cells. Hence, HbSS mice produce nearly 3 times more HbF/F cells than HbSC mice. These data are very similar to what has been seen clinically in patients with HbSS⁵⁵ vs those with HbSC who have been given HU.^19,22 The fact that this can be modeled in mice makes it possible to study the underlying mechanism in future studies.

Overall, the combined HbF and the non-HbF effects seen in this model explain the remarkable clinical benefit seen with HU in HbSC-SCD.^19,22-27 

In conclusion, the lack of an animal model of HbSC-SCD has hindered research and translation of therapies for HbSC-SCD. We report, to our knowledge, the first animal model of HbSC-SCD that mimics human HbSC-SCD phenotype and shows unique mechanistic insights into the HbF-dependent and HbF-independent effects of HU in HbSC vs HbSS SCD. This mouse model will allow mechanistic studies and rapid development of translatable therapies for HbSC-SCD.

The authors acknowledge the Center for Advanced Microscopy and Imaging at Miami University, Oxford, Ohio, for scanning electron microscopy imaging support; the Research Flow Cytometry Core Facility at Cincinnati Children’s Hospital Medical Center (CCHMC) for providing flow cytometry support; the Erythrocyte Diagnostic Laboratory at CCHMC, specifically Theodosia Kalfa and Jenifer Korpik, for support of hematological tests; the Transgenic Animal and Genome Editing Core Facility, CCHMC, for support with the generation of the sickle cell hemoglobin mice, and Jeffrey Bailey in the Mouse Core for assistance with mouse procedures; and Smith Drew, Lori Miller, and Betsy DiPasquale in the Pathology Core, CCHMC, for organ histopathology staining. They thank the University of Cincinnati, Proteomics Core, including Michael Wydmer and Wendy Haffey, and Rashmi Hegde and her laboratory personnel for allowing the use of, and assistance with, the hypoxia chamber. The authors acknowledge and thank Éva Katona for her contribution to the generation of the ferryl hemoglobin antibody. The visual abstract and Figure 7A were generated with BioRender.com.

This study is partially supported by a grant funded by the Center for Clinical and Translational Science and Training Just-In-Time program and the Research Innovation Pilot award (P.M.). J.B. was supported by HUN-REN-DE (11003) Research Network and Project no. TKP2021-EGA-18; and the National Research, Development and Innovation Fund, Ministry for Innovation and Technology of Hungary, National Institutes of Health Office of the Director S10 Shared Instrumentation Program (grant S10OD026717), which, in part, funded the mass spectrometer used in these studies.

Contribution: T.S., M.C., Z.O., M.S., K.S., Z.Z., A.T., H.K., K.D.G., J.B., M.X., K.V., A.S., and T.M.T. performed data acquisition and analyses; Y.-C.H., Z.O., M.S., K.V., K.D.G., T.S., and P.M. performed data plotting/micrography; T.S., M.C., Y.-C.H., and P.M. conducted the experimental methodologies and interpreted data; P.M. conceived this project; T.S. and P.M. designed the experiments, wrote the manuscript; and all authors read and commented on the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Punam Malik, Cincinnati Children’s Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45229; email: punam.malik@cchmc.org.