Hydroxyurea reduces hyperadhesion measured by a VLA-4–targeting PET tracer in sickle transgenic Townes mice

Visual Abstract

TO THE EDITOR:

In sickle cell disease (SCD), sickle hemoglobin (HbS) polymerizes within red blood cells (RBCs), deforming them into sickle cells with altered membrane integrity and high propensity to hemolyze.1, 2, 3 Sickling and hemolysis cause sterile inflammation that enhances the expression of adhesion molecules on blood and endothelial cells and collectively predisposes patients to painful vaso-occlusive episodes (VOE).⁴ The integrin very late antigen-4 (VLA-4, α₄β₁) is upregulated on RBC precursors and mediates heightened RBC-endothelial adhesion during VOE.^5,6 We demonstrated that the positron emission tomography (PET) tracer ⁶⁴Cu-LLP2A that binds with high affinity to activated VLA-47, 8, 9, 10 detects VLA-4–mediated hyperadhesion in the transgenic homozygous (SS) Townes mouse model, offering a novel approach for the noninvasive imaging of VOE-related vascular pathology.¹¹

Hydroxyurea (HU), the most important disease-modifying drug for SCD, induces fetal hemoglobin (HbF), thereby reducing HbS polymerization.¹¹ Experimental and clinical evidence suggests it also exerts important HbF-independent effects, including the attenuation of inflammation, leukocyte and platelet activation, endothelial dysfunction, and cellular adhesion. However, there is currently no direct visualization or quantification of the HbF-independent effects of HU in patients, which constrains our ability to optimize dosing strategies and to rationally combine HU with emerging therapies. Direct, quantitative assessment of the HbF-independent effects of HU is therefore critical for its precision use in heterogeneous SCD populations.12, 13, 14 In this study, we investigated whether HU could reduce VOE in Townes mice —in which HU benefits are uncoupled from HbF induction¹⁵— by measuring its effect on VLA-4–mediated hyperadhesion via noninvasive PET imaging.

One week before the baseline scan, blood samples were obtained from the animals (n = 5 per group; female Townes mice, 7-month old, The Jackson Laboratory, Sacramento, CA) via the facial vein for complete blood count. ⁶⁴Cu-LLP2A was prepared as previously described,⁷ injected IV (∼7.4 MBq; ∼200 μCi) into anesthetized mice, and baseline PET/computed tomography (CT) images were acquired at 4 hours postinjection. One day later, all animals were injected with either HU (n = 5; 100 mg/kg of body weight in 50 μL) or saline vehicle (n = 5; 50 μL) intraperitoneally for 5 days per week over 5 weeks. After treatment, all mice received 1 injection of lipopolysaccharide (LPS, 0.1 μg/kg of body weight) to cause vaso-occlusion, and ⁶⁴Cu-LLP2A PET/CT images were acquired at 4 hours postinjection followed by blood collection for complete blood count (VetScan HM5 analyzer). Standard uptake values (SUVs) for the humerus and femur, typical sites of VOE pain in humans, were measured using VivoQuant software (Invicro). To quantify SUVs, the regions of interest (ROIs) were delineated on coronal CT images along the full proximal-to-distal span of the humeri and femurs. For background normalization, muscle ROIs were defined on axial CT slices in either the forelimb or hind limb. Mean SUVs from the humeral and femoral ROIs were normalized to muscle uptake to obtain SUV ratios (SUVr), correcting for nonspecific background signal and intersubject variability.

We observed focal uptake of ⁶⁴Cu-LLP2A in PET/CT images, indicating its ability to visualize the humeri and femurs (Figure 1B-C) and thereby reflecting regions of VLA-4–mediated hyperadhesion. In the control group, the mean humeri SUVr increased from 6.5 ± 0.92 at baseline to 8.5 ± 1.7 post-LPS (P = .01; Figure 2A), reflecting elevated hyperadhesion and enhanced vaso-occlusion in the absence of treatment. Similarly, the mean femur SUVr rose from 8.4 ± 2.2 at baseline to 11 ± 1.6 post-LPS (P = .01; Figure 2E). Conversely, in the treatment group, the mean humeri SUVr at baseline was 6.2 ± 1.9 and decreased to 4.1 ± 1.8 post-LPS (P = .003; Figure 2B), indicating a reduction in hyperadhesion following HU treatment. The mean femur SUVr showed no increase post-LPS in the treatment group (ie, from 5.8 ± 3.8 at baseline to 6.0 ± 2.3 post-LPS; P = .93; Figure 2F), but we did not observe a reduction with HU, possibly due to the spillover of bladder radioactivity causing high background signal.

Figure 1.

Experimental design and representative PET/CT scans of mice at baseline and post-LPS challenge from the control and treatment groups. (A) Experimental design. (B) Representative PET/CT scan at baseline (left) and post-LPS challenge (right) from the control (saline) group. The blue arrows in panels B-C represent differences in tracer uptake in the humeri and femurs (vaso-occlusive sites). (C) Representative PET/CT scan at baseline (left) and post-HU and post-LPS challenge (right) from the treatment group. BW, body weight; M-F, Monday through Friday.

Figure 2.

Treatment with hydroxyurea mitigates LPS-induced hyperadhesion and inflammation measured by a VLA-4-targeting PET tracer and complete blood count analysis in sickle mice. (A-B,E-F) Mean SUV for the humeri and femurs divided by mean SUV for muscle for an SUVr normalized to background at baseline and post-LPS challenge for treatment and control group. Baseline and post-LPS challenge means are shown as a black circle with standard deviation (SD) error bars. (C-D,G-H) Complete blood count analysis at baseline and post-LPS challenge for both the treatment and control group. The bars represent mean ± SD for hemoglobin, WBC, neutrophil, and platelet counts. Statistical analysis between baseline and post-LPS used a paired 2-tailed Student t test. HGB, hemoglobin; NEU, neutrophil; PLT, platelet; WBC, white blood cell.

All mice exhibited splenomegaly, consistent with the role of the spleen as a hematopoietic organ in the adult mouse, and high splenic ⁶⁴Cu-LLP2A uptake at baseline and post-LPS challenge, reflecting the organ’s robust VLA-4 expression. In addition, mild thymic uptake was observed in some animals, consistent with the known VLA-4 expression in the thymus and a characteristic finding in ⁶⁴Cu-LLP2A PET imaging, whereas cardiac uptake was minimal.⁷

Although LPS has a short circulating half-life in mice (2-4 minutes),¹⁶ it induces a sustained inflammatory and endothelial activation response that persists beyond its clearance and drives downstream hematologic effects. As anticipated, the control group exhibited a significant reduction in hemoglobin levels following LPS challenge, indicating enhanced hemolysis during VOE (Figure 2C). In contrast, the treatment group demonstrated no changes in hemoglobin levels post-LPS challenge compared with those at baseline, suggesting that HU effectively mitigated hemolysis and inflammation. Following LPS challenge, control mice exhibited a marked reduction in circulating total white blood cell and neutrophil counts, consistent with acute neutrophil margination and sequestration. In contrast, HU-treated mice demonstrated a significantly attenuated decline in both after LPS exposure (Figure 2D,G). These findings are consistent with LPS-induced sequestration of neutrophils by activated endothelium, mediated in part by VLA-4–dependent adhesion, a process that is substantially blunted by HU treatment. Collectively, these data suggest that HU dampens the inflammatory response to LPS by limiting neutrophil margination and endothelial adhesion in vivo.^17,18 Post-LPS challenge, the control group, but not the HU-treated group, exhibited a reduction in platelet counts relative to those at baseline. This may have been due to reduced splenic platelet sequestration and margination in the HU-treated group. HU may have also attenuated the decline in neutrophil and platelet counts through additional mechanisms, including direct myelosuppressive effects, anti-inflammatory dampening of the cytokine cascade that drives cellular activation and consumption, prevention of neutrophil stiffening that would otherwise trap cells in the microvasculature, and inhibition of tissue factor expression that reduces coagulation-mediated platelet consumption. Finally, HU may have also reduced platelet activation by acting as a nitric oxide donor.

Our results align with and extend prior preclinical and clinical evidence demonstrating that HU confers clinically meaningful benefits independent of HbF induction. For example, individuals with genetically determined high baseline HbF levels still derive clinical benefit from HU, indicating that HbF induction is not a prerequisite for therapeutic response.¹⁹ Similarly, in the REACH clinical trial, HU led to marked reductions in clinical complications well before maximal HbF elevations were achieved, strongly implicating HbF-independent mechanisms. In addition, HU administered to adult HbSC and HbSS mice after γ-globin silencing²⁰ did not induce HbF yet reduced reactive oxygen species in RBCs, ferryl hemoglobin, and Heinz body formation, thereby reducing membrane damage and improving hematologic parameters.¹⁴

A limitation of this study is that our experimental design focused on chronic HU administration and therefore cannot disentangle effects attributable to long-term exposure from those arising after acute dosing. Prior studies in sickle cell mouse models have shown that a single administration of HU can rapidly elicit biological effects, including nitric oxide–mediated signaling and changes in vascular function, which may contribute to early therapeutic benefit.^21,22 Consequently, some of the effects observed with chronic treatment in our model may reflect the cumulative or sustained impact of mechanisms that are already initiated acutely.

In summary, our findings show the potential utility of HU in patients with SCD and inflammatory or adhesive phenotypes regardless of baseline HbF levels and provide a rationale for combination therapeutic strategies.

Conflict-of-interest disclosure: E.M.N. is an advisory board member/consultant for Novo Nordisk, Novartis, and Chiesi. The remaining authors declare no competing financial interests.