Loss of NRF2 function exacerbates the pathophysiology of sickle cell disease in a transgenic mouse model

Sickle cell disease (SCD) is a group of inherited blood disorders with a wide range of clinical symptoms caused by polymerization of hemoglobin S (HbS) in red blood cells (RBCs) and chronic hemolysis leading to anemia, inflammation, and elevated reactive oxygen species (ROS).^1,2  Previously, nuclear factor (erythroid-derived 2)-like 2 (NRF2), the master regulator of the cellular oxidative stress response,³  was shown to mediate γ-globin gene activation^4-6  through direct promoter interactions.⁶  More recently, we demonstrated the ability of NRF2 to bind the γ-globin promoter and locus control region to mediate chromatin looping during normal erythropoiesis.⁷  These findings were complemented by in vivo genetic⁸  and pharmacologic^9-12  activation of NRF2, which reduces intravascular hemolysis to improve the phenotype of SCD mice.

To expand on these findings, we investigated the effects of NRF2 loss on fetal hemoglobin (HbF) expression during gestational erythropoiesis and severity of the SCD phenotype using a transgenic mouse model. Our data demonstrated that genetic NRF2 knockout silenced γ-globin gene transcription and HbF expression during development while exacerbating oxidative stress and proinflammatory phenotypes in adult SCD mice. These findings confirm the critical role of NRF2 in hemoglobin regulation and warrant the development of therapeutic agents that activate NRF2 expression as a strategy for SCD treatment.

The humanized Townes SCD mouse model (B6;129-Hba^tm1(HBA)TowHbb^{tm2(HBG1,HBB*)Tow}/Hbb^{tm3(HBG1,HBB)Tow}/J) was characterized previously.¹³  The C57BL/6J NRF2 knockout (NRF2^−/−) mouse¹⁴  was provided by Dr. Yamamoto (Tohoku University, Sendai, Japan). The SCD and NRF2^−/− mice were crossbred to generate the SCD/NRF2^−/− mouse (supplemental Methods; supplemental Figure 1A; available on the Blood Web site) after approval by the Augusta University Institutional Animal Care and Use Committee.

See supplemental Methods for additional details.

To assess the effect of NRF2 function on the pathophysiology of SCD, we established the SCD/NRF2^−/− mouse model (supplemental Figure 1). Although the NRF2 gene is transmitted through Mendelian genetics, NRF2^−/− produced variable penetrance demonstrated by 2% SCD/NRF2^−/− pups (supplemental Figure 1B), supporting embryonic lethality on a SCD background. In addition, the fertility of female SCD/NRF2^−/− mice was decreased, with smaller litter sizes and high mortality in the perinatal period (supplemental Figure 1C-D). These data support a role for NRF2 in fetal and perinatal survival for SCD mice.

To evaluate the hematopoietic effect, we performed complete blood cell counts with differentials, which confirmed a significant decrease in RBCs and an increase in total white blood cells along with lymphocytes and monocytes in SCD/NRF2^−/− mice compared with SCD/NRF2^+/+ mice (Figure 1A); other hematologic parameters were not affected. We next examined globin gene expression during development. The γ-globin messenger RNA (mRNA) levels were significantly lower in comparison with β^S-globin mRNA (Figure 1B), and the γ:(γ+β^S) globin ratios in SCD/NRF2^−/− mice was decreased in embryonic day 13.5 and embryonic day 18.5 fetal liver cells from 0.44% to 0.37% and from 0.21% to 0.14% (P < .05), respectively. In 2- to 3-month-old SCD/NRF2^−/− adult spleen and bone marrow cells, the γ:(γ+β^S) ratio was decreased by 29% to 33% (P < .05). As further evidence of NRF2 involvement in globin gene regulation, SCD/NRF2^−/− bone marrow hematopoietic progenitors showed a decrease in HbF-expressing cells (F-cells) from 3.2% to 1.2% (P < .05), however, expression of the differentiation markers CD71, Sca1, and c-Kit were not changed significantly (Figure 1C).

Because the spleen is a primary site for hematopoiesis in SCD,¹⁵  we performed globin gene analysis in adult SCD/NRF2^−/− spleens. Reflective of γ-globin silencing during gestation, we observed a 73% (P < .05) decrease in HbF expression (Figure 1D). In parallel, F-cell levels were decreased from 2.5% to 1.4% (P < .05) (Figure 1E). Interestingly, HbS levels were unchanged, suggesting that NRF2 does not regulate the adult β^S-globin gene and confirming our recent findings that NRF2 preferentially regulated the γ-globin gene in human erythroid progenitors.⁷  To support a functional role of NRF2, we next determined if the decrease in HbF affected the propensity of RBCs to sickle. Under normal oxygen conditions, NRF2^−/− had no effect on RBC sickling (Figure 1F), however, under 1% O₂ conditions, we observed 54% and 39% sickling in SCD/NRF2^−/− and SCD/NRF2^+/+ mice, respectively (P < .05). The regulation of γ-globin expression by NRF2 in SCD mice supports our findings that NRF2 modulates chromatin structure in the β-locus.⁷ 

In SCD mice, the spleen is enlarged due to compensatory hematopoiesis, chronic hemolysis, and oxidative stress.¹⁶  The loss of NRF2 function in SCD/NRF2^−/− mice exaggerated splenomegaly (Figure 2A), supporting the profound effect of oxidative damage with NRF2 loss. To explore the role of NRF2 in the pathophysiology of SCD, tissue morphology was assessed, which revealed expansion of the splenic red pulp and obscuring of the architecture in SCD/NRF2^−/− mice. Although the spleens of SCD/NRF2^+/+ mice showed expansion of red pulp, the architecture remained intact (Figure 2B).

It is known that oxidative stress contributes to the clinical severity of SCD.²  Although NRF2 is the master regulator of the oxidative stress response,³  we observed low erythrocyte ROS levels in C57BL6/NRF2^−/− mice on a nonhemolytic background (Figure 2C), suggesting that other proteins control oxidative stress. By contrast, significantly higher ROS levels were observed in SCD/NRF2^+/+ mice (P < .01), which increased further with NRF2 loss (P < .05), supporting a more prominent role of NRF2 in a chronic hemolysis phenotype. To maintain the oxidative stress response, NRF2 regulates several antioxidative proteins.^9-11,17,18  In particular, the levels of HMOX1, NQO1, GCLC, and catalase were decreased by 31%, 30%, 60%, and 48%, respectively (P < .05) in SCD/NRF2^−/− mice (Figure 2D). High ROS level combined with free heme produce a detrimental effect, promoting RBC adhesion and vaso-occlusion in SCD.²  Therefore, we determined ICAM-1, VCAM-1, and VEGF expression. Whereas VEGF protein was not changed, ICAM-1 and VCAM-1 levels were increased 1.7-fold and 2.0-fold (P < .05), respectively, in SCD/NRF2^−/− adult spleen cells (Figure 2E), suggesting that NRF2 loss would promote adhesion. Additional studies will determine whether ICAM-1 and VCAM-1 are direct targets of NRF2.

Chronic inflammation contributes to the clinical severity of SCD^2,19,20 ; therefore, we investigated the proinflammatory molecules interleukin-1β (IL-1β), IL-6, and tumor necrosis factor-α (TNF-α) in adult spleen cells. The mRNA levels of these genes were higher in SCD mice compared with C57BL/6 mice; strikingly, IL-1β, IL-6, and TNF-α levels increased further in SCD/NRF2^−/− mice (Figure 2F). By contrast, expression of the anti-inflammatory cytokine IL-10 was not changed significantly (Figure 2F).

Previous studies have demonstrated that genetic and pharmacologic NRF2 activation mediates antiadhesion and anti-inflammatory effects in SCD mice.^9-11  We recently confirmed similar outcomes after chronic treatment with the NRF2 activator dimethyl fumarate.¹²  Another attractive function of NRF2 is directing l-glutamine into glutathione synthesis.^21,22  Because l-glutamine levels are reduced in SCD,²³  oral therapy improves the redox potential of sickle RBCs²⁴  and reduces adhesion to human umbilical vein endothelial cells.²⁵  In a recently completed clinical trial, Endari (l-glutamine) was shown to reduce oxidative stress and pain episodes in SCD patients, leading to Food and Drug Administration approval.

Although pharmacologic NRF2 activation provides insights into its protective functions, genetic knockout is required to confirm a role in globin gene regulation during development. Our SCD/NRF2^−/− mouse model established the requirement of NRF2 for γ-globin gene transcription during the embryonic and adult gestational stages of erythropoiesis and confirmed γ-globin silencing in the absence of NRF2 expression. Furthermore, NRF2 knockout on a hemolytic background discovered a protection role of NRF2 in the survival of SCD mice during gestation and the perinatal period. Additionally, we observed severe spleen damage, RBC sickling, and increased expression of proinflammatory and adhesion molecules in SCD/NRF2^−/− mice. Questions beyond those addressed in this study, such as the role of NRF2 in protecting against severe complications of SCD, including acute chest syndrome, stroke, and kidney disease, among others, can be answered using the SCD/NRF2^−/− model, thus expanding its usefulness in the field. Our findings herein combined with published data^8-12,22-25  demonstrate the protective role of NRF2 in SCD and support the development of small-molecule activators of NRF2 for the treatment of β-hemoglobinopathies.

The authors thank Masayuki Yamamoto for contributing the NRF2^−/− mice. The authors also thank Dorothy Tuan, Hongyan Xu, and Huidong Shi for their helpful insights and discussions of data interpretation.

This work was supported by training grant HL117684 from the National Institutes of Health (NIH), National Heart, Lung, and Blood Institute (X.Z.); R01 grant HL069234 (B.S.P.) from the NIH, National Heart, Lung, and Blood Institute; and R01 grant NS101967 (B.T.) from the NIH, National Institute of Neurological Disorders and Stroke.

Contribution: X.Z. designed and performed the research studies, analyzed the data, and wrote the manuscript; C.X. performed the research studies and analyzed the data; B.T. contributed to the research design, data analysis, and manuscript writing; and B.S.P. contributed to the research design, data analysis, manuscript writing, and supervision of the study.

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

Correspondence: Xingguo Zhu, Department of Pediatrics, Augusta University, 1120 15th St, CN-4116, Augusta, GA 30912; e-mail: xzhu@augusta.edu.