Reduced PU.1 expression underlies aberrant neutrophil maturation and function in β-thalassemia mice and patients

β-Thalassemia is the result of mutations in the β-globin gene, causing reduced or absent β-globin synthesis, ultimately leading to the precipitation of α-globin proteins and formation of deleterious reactive oxygen species (ROS) in erythroid progenitors.¹  Consequently, erythropoiesis is impaired triggering erythroid apoptosis, hemolysis, and severe anemia. In the most severe form, β-thalassemia major, patients require ongoing blood transfusions for survival. Repeated blood transfusion to counteract anemia leads to iron overload, necessitating iron chelation therapy. Chelating agents allow excretion of iron through urine and feces, although chelation therapy is often insufficient to prevent the onset of iron-related pathologies, including cardiac, hepatic, and endocrine dysfunction.^2-4 

In addition to aforementioned morbidities, multiple immune abnormalities affecting both innate and adaptive immunity are a major concern in β-thalassemia patients.^5,6  These immune abnormalities include decreased responses and functionality from innate immune cells, including phagocytic cells and natural killer (NK) cells, as well as increased numbers of T and B lymphocytes and increased levels of serum immunoglobulins.^5,6  Iron-mediated chronic inflammation and aberrant neutrophil effector function have been identified as key determinants of susceptibility to infection.^6-8  Additional evidence linking iron overload to aberrant neutrophil effector function comes from studies of host iron status, and iron supplementation in both animal models and human patients.^9-11 

Neutrophils form an essential part of the innate immune system and function as a first line of defense against infections. In addition, neutrophils are thought to play an important role in wound healing through their ability to clear infectious organisms and cellular debris from the tissues before the influx of macrophages and fibroblasts, which aid the healing process through stimulation of angiogenesis and granulation.¹²  In humans, neutrophils are produced at a rate of 1-2 × 10¹¹ per day and account for 50% to 70% of circulating leukocytes.^13,14  Upon differentiation, mature neutrophils are released from the bone marrow (BM) and enter the bloodstream. Neutrophils are recruited to the site of infection or inflammation in response to microbial molecules, cytokines such as interleukin-1β (IL-1β), IL-6, and tumor necrosis factor α, or chemokines. Production of chemokine ligands CXCL1, 2, and 5 by sentinel and helper macrophages stimulates neutrophil chemotaxis via activation of the chemokine receptors CXCR1 and 2.¹³  Neutrophils also act as instructors of other immune cell types such as macrophages, NK cells, dendritic cells (DCs), T cells, and B cells.^13,14  The mechanisms by which neutrophils protect against infection are diverse and include opsonophagocytosis, reduced NAD phosphate (NADPH) oxidase-dependent ROS production, and generation of neutrophil extracellular traps (NETs). In addition, neutrophils use several antimicrobial molecules to combat infection, including myeloperoxidase, neutrophil elastase, cathepsin G, and defensins.^13,14  β-Hemoglobinopathy patients have impaired neutrophil effector functions affecting chemotaxis, phagocytosis, ROS production, and NET formation, which collectively contribute to disease progression.^5,15-18  Given the clinical significance of infection-related morbidity and mortality in β-thalassemia and related conditions, we explored aspects of neutrophil biology that may contribute to impaired immune function using a clinically relevant mouse model of β-thalassemia (Hbb^th3/+).¹⁹  We noted a striking expansion of immature band neutrophils as well as the presence of myelocytes and metamyelocytes in Hbb^th3/+ mice.^20,21  Importantly, expression of the ets-family transcription factor PU.1, required for terminal myeloid differentiation and effector function,^22-24  was significantly lower in Hbb^th3/+ mice compared with normal controls. Mirroring those of Hbb^th3/+ mice, neutrophils isolated from hemoglobin E (HbE)/β-thalassemia patients were hyposegmented and displayed reduced PU.1 expression, indicating an aberrant or arrested state of neutrophil maturation.^22,25  Here, we show that Hbb^th3/+ mice closely mimic the immune dysfunctions observed in β-thalassemia patients. Our findings provide mechanistic insights into the compromised neutrophil function in the β-hemoglobinopathies.

Hbb^th3/+ mice were obtained from the Jackson Laboratory (Bar Harbor, ME) and maintained on a C57BL/6 background.¹⁹  Wild-type (WT) C57BL/6 mice were used as control mice. Six- to 7-month-old, age- and sex-matched Hbb^th3/+ mice were used as they exhibit a higher level of iron accumulation relative to WT mice.^26,27  All animal experiments were approved by the Murdoch Childrens Research Institute Animal Ethics Committee.

Murine neutrophils were identified as CD49b^negCD11b⁺Ly6G^highLy6C^low cells; immature myeloid cells were identified as CD49b^negCD11b⁺Ly6G^lowLy6C^low, and monocytes were identified as CD49b^negCD11b⁺Ly6G⁻Ly6C^high (supplemental Figure 1, available on the Blood Web site).²⁰  Antibodies used in this study are listed in supplemental Table 1A. Cells were acquired on an LSR II cytometer (BD, San Jose, CA) using FACS Diva software. The data were analyzed using FlowJo software (Tree Star, OR). Measurement of neutrophil chemotaxis,²⁸  phagocytosis,²⁹  and ROS²⁹  was performed as previously described (supplemental Methods).

Stocks of S pneumoniae (strain WU2) were prepared as described previously.³⁰  Mice were injected IV with 2000 colony-forming units (CFU). The animals were observed for signs of morbidity at regular intervals, and the experiment was terminated prior to mice reaching a moribund state. Organs were collected following infection, and bacteria were enumerated using serial dilutions and plating on horse blood agar plates.²⁸ 

Total RNA was extracted using the RNeasy mini kit (Qiagen). A total of 50 ng RNA was reverse transcribed to complementary DNA using SuperScript III First-Strand Synthesis Kit (Invitrogen). When flow cytometry–sorted cells were used, gene-specific primers (supplemental Table 2) were used instead of oligo(dT)₂₀ to generate complementary DNA. Gene expression was analyzed by reverse transcription quantitative polymerase chain reaction (RT-qPCR), performed with SYBRGreen PCR Master Mix (Life Technologies). Housekeeping genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and U6 small nuclear RNA were included for normalization, and relative expression was calculated by “comparative Ct method.”

Seventeen β-thalassemia/HbE patients (12 nonsplenectomized and 5 splenectomized) and 8 healthy controls with an age range from 22 to 53 years were recruited. The Mahidol University Central Institutional Review Board approved the study protocol. Characterization of peripheral blood neutrophils was performed as previously described³¹  (supplemental Methods). The human-specific antibodies that were used are listed in supplemental Table 1B. Hematological parameters of HbE/β-thalassemia patients at the time of assay are listed in supplemental Table 3.

Survival analysis was analyzed using the Kaplan-Meier method. Differences in bacterial load in WT and Hbb^th3/+ mice were assessed using the Mann-Whitney U test. All other data were analyzed using Student t test applied using GraphPad PRISM 6.0 (GraphPad Software, Inc). Data are presented as mean ± standard error of mean (SEM). Results were considered statistically significantly different if P ≤ .05.

It has been well established that iron overload conditions and β-thalassemia have been associated with defects in neutrophil functional abilities, including chemotaxis, phagocytosis, and ROS production.^9-11,32  To determine whether neutrophil chemotaxis is altered in Hbb^th3/+ mice, neutrophil migration in response to CXCL2 was investigated in a transwell migration assay. Interestingly, neutrophils from the blood and spleen of Hbb^th3/+ mice showed a significant decrease in CXCL2-induced chemotaxis compared with WT neutrophils (Figure 1A-C), whereas BM neutrophils showed no significant difference (Figure 1A).

Neutrophil trafficking is antagonistically regulated by CXCR2/CXCR4 signaling.³³  Therefore, we examined expression levels of CXCR2 and CXCR4 in CD11b⁺Ly6G^high populations from BM, blood, and spleen. Interestingly, surface expression of CXCR2 was significantly decreased in neutrophils from Hbb^th3/+ blood and spleen compared with WT mice, reflecting the impaired chemotactic response of these cells to CXCL2 (Figure 1D-F,J). No difference in CXCR2 expression levels was found on BM neutrophils. In addition, no difference in CXCR4 expression levels on Hbb^th3/+ and WT neutrophils in the blood, spleen, or BM was detected (Figure 1G-I,K).

Flow cytometry was used to measure the percentage of neutrophils (CD11b⁺Ly6G^highLy6C^low) capable of phagocytosing opsonized Staphylococcus aureus labeled with AF488 (supplemental Figure 1). The Hbb^th3/+ neutrophils (CD11b⁺Ly6G^highLy6C^low) from BM, blood, and spleen displayed impaired phagocytic activity when compared with WT neutrophils (CD11b⁺Ly6G^highLy6C^low) (Figure 2A-C). To further understand the impaired process responsible for reduced Hbb^th3/+ neutrophil phagocytosis, cell surface expression of CD11b, a component of complement receptor 3 (CR3), was measured in BM, blood, and spleen of naive mice. Interestingly, CD11b expression on Ly6G^highLy6C^low neutrophils was significantly lower in the spleen of Hbb^th3/+ mice compared with WT mice, whereas BM and blood Ly6G^highLy6C^low neutrophils expressed similar levels of CD11b (Figure 2D-G).

As well as playing a role in phagocytosis and migration, surface expression of CD11b is upregulated on neutrophils following activation, and this can be used as a marker to identify recently activated neutrophils (CD11b^high).³⁴  In this study, we were interested in measuring the kinetics of neutrophil activation following incubation with S aureus.¹³  Following S aureus stimulation, the majority of WT splenic (Ly6G^highLy6C^low) neutrophils rapidly increased CD11b expression within 30 minutes, whereas the majority of Hbb^th3/+ splenic (Ly6G^highLy6C^low) neutrophils failed to upregulate CD11b expression, instead remaining in the CD11b^low population (Figure 2H-I). We next compared the phagocytic activity of CD11b^low and CD11b^high neutrophils. We found that the majority of WT phagocytic neutrophils (Ly6G^highLy6C^low) resided in CD11b^high population, whereas a significant proportion of Hbb^th3/+ CD11b^high splenic neutrophils (Ly6G^highLy6C^low) had very low phagocytic activity (Figure 2J). The failure in Hbb^th3/+ splenic neutrophils (Ly6G^highLy6C^low) to upregulate CD11b expression upon S aureus stimulation is consistent with a suppressed state of activation and phagocytosis.

To test neutrophil ROS activity, neutrophils from Hbb^th3/+ and WT mice were stimulated in vitro with killed S aureus. The cells were subsequently stained with 2′,7′-dichlorofluorescin diacetate (DCFH-DA), which is converted to fluorescent DCF in the presence of ROS. Flow cytometry was used to measure neutrophil (CD11b⁺Ly6G^high) ROS activity over 60-minute periods (Figure 3; supplemental Figure 3). Following DCFH-DA treatment, Hbb^th3/+ blood and splenic neutrophils (CD11b⁺Ly6G^high) displayed a reduced ROS production in comparison with WT mice (Figure 3B-C) (P < .046 and P < .0054, respectively). No significant difference in ROS production was observed in BM neutrophils (CD11b⁺Ly6G^high) (Figure 3A).

The altered functional properties of Hbb^th3/+ neutrophils led us to evaluate their morphology. Microscopic analysis of sorted neutrophils from the spleen of age- and sex-matched Hbb^th3/+ mice revealed features characteristic of immature neutrophils. Interestingly, ∼40% of Hbb^th3/+ Ly6G^highLy6C^low neutrophils displayed band morphology, indicating a failure to achieve terminal maturation (Figure 4A-D; supplemental Figure 4A). In addition, a striking expansion of a subset of neutrophils (CD11b⁺Ly6G^lowLy6C^low) was detected in Hbb^th3/+ mice (Figure 4A). This population contained immature granulocytes, including myelocytes and metamyelocytes (Figure 4E). Intriguingly, this heterogeneous group of cells morphologically and phenotypically resemble myeloid-derived suppressor cells (MDSCs).^35,36 

In order to further characterize the CD11b⁺ population, 3 distinct myeloid populations were measured in the BM, blood, and spleen of Hbb^th3/+ and WT mice: neutrophils (CD11b⁺Ly6G^highLy6C^low), immature myeloid cells (CD11b⁺Ly6G^lowLy6C^low), and monocytes (CD11b⁺Ly6G⁻Ly6C^high).²⁰  Among the CD11b⁺CD49^neg cells, the proportions of neutrophils (Ly6G^highLy6C^low) in BM, blood, and spleen of Hbb^th3/+ and WT mice were comparable (Figure 4F). This observation suggests that the block in neutrophil maturation (ie, from band to segmented) lies after the acquisition of the CD11b⁺Ly6G^highLy6C^low phenotype. Interestingly, there was an approximately fourfold increase in the proportion of immature myeloid cells (CD11b⁺Ly6G^lowLy6C^low) in the spleen of Hbb^th3/+ mice (Figure 4G). The proportion of immature myeloid cells in spleens of Hbb^th3/+ mice was comparable to the proportion of CD11b⁺Ly6G^lowLy6C^low cells in the BM of Hbb^th3/+ and WT mice. The percentage of monocytes (CD11b⁺Ly6G^negLy6C^high) in BM, blood, and spleen of mice was also measured, demonstrating a significant reduction in this population in the spleens of Hbb^th3/+ mice compared with WT mice (Figure 4H). Together, these results suggest that in Hbb^th3/+ mice, terminal neutrophil maturation in the spleen may be compromised.

PU.1 is 1 of the key myeloid regulators that directs terminal maturation of both neutrophils and macrophages.^22,24  Disruption of normal PU.1 expression is associated with a block or attenuation in terminal maturation of myeloid lineages, particularly neutrophils and macrophages, as demonstrated in PU.1-deficient mice^37,38  and humans with myeloid disorders.^25,39 

We measured PU.1 messenger RNA (mRNA) and protein levels in all 3 myeloid populations sorted by flow cytometry. The BM-derived myeloid populations from WT and Hbb^th3/+ mice displayed low levels of PU.1 expression, reflecting their immature state (Figure 5A-F). However, when PU.1 protein levels were measured in splenic-derived myeloid populations from Hbb^th3/+ mice, we observed a significant reduction in PU.1 protein (Figure 5J-L) and mRNA (Figure 5G-I) levels in neutrophils (Ly6G^highLy6C^low), immature myeloid (Ly6G^lowLy6C^low), and monocytes (Ly6G^negLy6C^high), relative to WT mice. Expression of C/EBPα (an additional transcription factor involved in myeloid differentiation) was also measured in sorted neutrophils (Ly6G^highLy6C^low), and no differences in mRNA expression levels were detected (supplemental Figure 4E). Finally, mRNA levels for granulocyte-colony stimulating factor receptor (G-CSFR) (regulated by PU.1)⁴⁰  were reduced in splenic neutrophils (Ly6G^highLy6C^low) (supplemental Figure 4F).

We next investigated splenic gene expression levels in response to in vivo infection. IV infection with S pneumoniae was used as an experimental infection model representative of an infectious agent frequently causing infection in β-thalassemia patients. Figure 6A-B shows that in contrast to WT mice, Hbb^th3/+ mice do not upregulate expression levels of chemoattractant CXCL2 and its receptor CXCR2 in the spleen following infection with S pneumoniae.

NADPH oxidase is a multisubunit enzyme complex comprising membrane (gp91^phox and p22^phox) and cytosolic (p47^phox, p67^phox, p40^phox, and G-proteins Rac2 and Cdc42) components. Following activation, the cytoplasmic components translocate to the cell membrane to form the active oxidase complex generating superoxide anions.¹³  To understand the molecular mechanism responsible for suppressed ROS production, splenocytes from naive mice were analyzed for expression of NADPH oxidase components. No significant difference in expression of p22^phox, p67^phox, and gp91^phox was observed between naive WT and Hbb^th3/+ mice, whereas p40^phox was significantly reduced in naive Hbb^th3/+ spleens (Figure 6C-F). However, following in vivo infection with S pneumoniae, WT splenocytes increased p22^phox, p91^phox, p40^phox, and p67^phox expression, whereas Hbb^th3/+ splenocytes failed to induce p22^phox, p91^phox, p40^phox, and p67^phox (Figure 6C-F). These data demonstrate that defective ROS production in Hbb^th3/+ mice is associated with the failure to induce expression of critical NADPH components.

We further examined expression of PU.1 and critical downstream targets associated with neutrophil maturation in splenocytes in response to S pneumoniae infection. Notably, bacterial sepsis was associated with upregulated PU.1 and G-CSFR expression levels in WT splenocytes, whereas PU.1 and G-CSFR expression in Hbb^th3/+ splenocytes remained unresponsive following infection (Figure 6G-H).

Moreover, PU.1 has also been reported to bind to the miR-223 promoter,^41,42  increasing expression of the microRNA and promoting myeloid progenitor differentiation and maturation.^43,44  Notably, in WT and Hbb^th3/+ splenocytes, miR-223 expression levels were comparable in naive animals. However, in response to S pneumoniae infection, WT mice significantly upregulated miR-223 expression, whereas miR-223 expression in Hbb^th3/+ splenocytes remained unchanged (Figure 6I). Collectively, our findings indicate that in response to bacterial infection, Hbb^th3/+ mice fail to upregulate key neutrophil regulators required for terminal maturation and effector functions, and this failure is associated with inadequate PU.1 expression.

To explore if β-thalassemic mice have impaired host resistance to acute bacterial infections, 6-month-old Hbb^th3/+ and age-matched WT mice were injected IV with 2000 CFU of S pneumoniae. The animals were examined for signs of illness at regular intervals, and the experiment was terminated after 30 hours or before mice reached a moribund state. All Hbb^th3/+ mice developed a severe infection or became moribund and were euthanized, whereas all WT mice survived (Figure 6J). The spleen, liver, and blood were harvested at 24 to 30 hours postinfection from WT and Hbb^th3/+ mice to measure the bacterial load in each organ. The average bacterial load in the blood of Hbb^th3/+ mice was 2.2-fold higher than in WT mice (Figure 6K). The spleen and liver of Hbb^th3/+ mice showed a significantly increased bacterial load compared with WT mice (Figure 6L-M). In a separate set of experiments, we demonstrated that in vitro growth rate of S pneumoniae in fresh blood from Hbb^th3/+ or WT mice was comparable (data not shown). Taken together, these data suggested that Hbb^th3/+ mice are more susceptible to bacterial infection than WT mice.

The integrity and proper function of the spleen are considered important for the protection against infectious diseases. Upon gross examination of the spleen from 6-month-old Hbb^th3/+ mice, we observed several structural abnormalities, including aberrant red pulp sinusoidal architecture and irregular distribution of white pulp nodules relative to WT mice (supplemental Figure 5). Moreover, 6-month-old Hbb^th3/+ mice displayed extensive iron overload in spleen sections, relative to Hbb^th3/+ 6-week-old and WT mice (supplemental Figure 5). These observations are consistent with previous studies that demonstrate Hbb^th3/+ mice progressively iron overload with age.^26,27  Interestingly, in 6-week-old Hbb^th3/+ mice, no difference was detected in the percentage of mature segmented cells among the CD11b⁺Ly6G^highLy6C^low neutrophils compared with WT mice (supplemental Figure 6E). We also found no difference in the proportion of immature myeloid cells (CD11b⁺Ly6G^lowLy6C^low) when compared with 6-week-old WT mice, whereas 6-month-old Hbb^th3/+ mice displayed an increased proportion of immature myeloid cells (supplemental Figure 6F). Taken together, these data suggest that the aberrant maturation of myeloid cells in Hbb^th3/+ mice is age related, and the maldistribution of iron in the spleen of Hbb^th3/+ mice overtime may contribute to impaired maturation and function of neutrophils.

To evaluate whether our findings in Hbb^th3/+ mice are reflected in β-thalassemia patients, we analyzed neutrophils isolated from the peripheral blood of HbE/β-thalassemia patients (Figure 7A-D). The number of nuclear lobes in human neutrophils was used to distinguish between immature and mature states of development. Typically, mature human neutrophils contain 3 to 5 nuclear lobes. Interestingly, morphological analysis of neutrophils from HbE/β-thalassemia subjects revealed a significant increase in hyposegmented nuclei. In nonsplenectomized HbE/β-thalassemia patients, the proportion of neutrophils displaying 2 to 3 lobes was significantly higher, with a reduction in the proportion of morphologically mature cells possessing 4 to 5 lobes, relative to healthy control individuals (Figure 7C; supplemental Table 4). Similarly, splenectomized HbE/β-thalassemia patients displayed a greater proportion of neutrophils with hyposegmented nuclei (≤2 lobes), and a corresponding decrease in the number of cells with 4 lobes relative to healthy control individuals (Figure 7D).

We next analyzed whether PU.1 expression was also impaired in HbE/β-thalassemia neutrophils. Similarly to our data from the Hbb^th3/+ mice, PU.1 expression in neutrophils from nonsplenectomized and splenectomized HbE/β-thalassemia patients was significantly lower compared with normal individuals. This observation supports our hypothesis that neutrophils in β-thalassemia fail to fully mature due to reduced expression of the PU.1 transcription factor (Figure 7E).

Several studies in β-thalassemia patients have clearly pointed to disruptions in the innate immune system as factors contributing to poor outcomes following infection.^5,6  Underlying the dysfunctional innate immune system are defects in neutrophil migration, phagocytosis, and activation of pathogen-specific immune responses. The molecular mechanisms by which neutrophils fail to mount appropriate responses are unknown. To further investigate innate immune dysregulation, we explored neutrophil effector functions in β-thalassemia patients and in Hbb^th3/+ mice, focusing on neutrophils that displayed a mature phenotype, characterized in mice by the expression of high levels of surface molecule Ly6G and intermediate levels of Ly6C.

Migration of neutrophils to the site of infection is orchestrated by chemokines, including CXCL1 and CXCL2, which ligate receptors CXCR1 and CXCR2, respectively. Because CXCR2 is an important component of the neutrophil chemotactic response,^13,14  we reasoned that the reduced migration by Hbb^th3/+ neutrophils might be due to aberrant CXCR2 expression. Indeed, a significant reduction in CXCR2 expression was observed in Hbb^th3/+ neutrophils, and these cells failed to induce expression of CXCR2 and its ligand CXCL2 upon exposure to S pneumoniae, suggesting a serious impairment in the CXCR2-CXCL2 axis. Such inability to respond to chemokines may result in impaired neutrophil migration and reduced resistance to infection, as has been shown in infection studies in CXCR2-deficient mice.⁴⁵ 

Impaired neutrophil phagocytosis has been reported in β-thalassemia patients,^5,16,46  and we detected an impaired phagocytic response in Hbb^th3/+ neutrophils, in particular, in the spleen. Recognition and binding of pathogens involve a wide variety of neutrophil surface receptors, including pattern-recognition receptors, such as Toll-like receptors, Fcγ receptors, and the CRs.^13,14,47  Interestingly, when we measured the expression of CD11b, 1 of the 2 CR3 subunits (CD11b/CD18), we found a significant reduction in CD11b expression in Hbb^th3/+ splenic neutrophils, which was associated with impaired phagocytic activity of the neutrophils following S aureus stimulation. CR3 is expressed at high levels on neutrophils, monocytes, macrophages, and NK cells and plays a key role in leukocyte adhesion, migration, and phagocytosis.^13,14,47  Notably, CR3 deficiency in mice⁴⁸  and humans⁴⁹  leads to life-threatening infections, and the reduced expression levels of CD11b may contribute to reduced resistance to infection in the Hbb^th3/+ mice.

Following phagocytosis, engulfed pathogens are killed by ROS generated by the NADPH oxidase complex, and this key process has been reported to be impaired in β-thalassemia patients.^5,50  The NADPH complex is composed of cytosolic and membrane components, which interact following activation. The cytoplasmic components include p47^phox, p67^phox, p40^phox, and the small G-proteins, Rac2 and Cdc42, whereas the transmembrane components include gp91^phox and p22^phox. Upon activation, the cytoplasmic components translocate to the membrane, resulting in an activated complex generating ROS.¹³  Our molecular studies in naive Hbb^th3/+ mice identified a significant decrease in p40^phox and p47^phox components in blood and spleen. Furthermore, following S pneumoniae infection, Hbb^th3/+ splenocytes failed to upregulate expression of both the membrane (p22^phox, p91^phox) and the cytosolic (p40^phox and p67^phox) components of NADPH, explaining the reduced ROS production in Hbb^th3/+ splenocytes.

The altered functional properties of Hbb^th3/+ neutrophils, which showed little change in their phenotype based on Ly6G and Ly6C expression, led us to evaluate their morphology. Unexpectedly, ∼40% of Hbb^th3/+ splenic neutrophils were immature band cells. Such an increase in band neutrophils may explain the differences in phagocytosis and ROS production detected in the Hbb^th3/+ mice, as band cells are less efficient in phagocytosis.^51,52  Another striking observation was the expansion of immature myeloid cells (Ly6G^lowLy6C^low) in the spleen of Hbb^th3/+ mice to levels observed in BM. It is thought that this immature granulocyte population that arises from myeloid progenitors, and under steady state conditions, can quickly differentiate into mature granulocytes, macrophages, and DCs.⁵³  Importantly, several studies have demonstrated that a heterogeneous group of immature myeloid cells and neutrophils termed MDSCs is found to expand, infiltrate, and persist in the spleen, liver, lung, and other lymphoid organs in pathological conditions,^53,54  such as in cancer,^20,55  auto-immunity,^56,57  chronic hepatitis C,³⁵  HIV,³⁵  and cystic fibrosis,⁵⁸  but has not previously been reported in the context of β-thalassemia. Further experiments are required to determine whether the immature myeloid cells detected in the Hbb^th3/+ mice are MDSCs. The presence of an expanded number of MDSCs, which are known to suppress T-cell and NK functions via a suppressive mechanism involving arginase expression and IL-10 secretion,^53,54,59  and are frequently found in inflammatory settings, may contribute to severe organ dysfunction^20,59-61  and could play a role in progressive organ damage in β-thalassemia.

Several genes tightly orchestrate neutrophil maturation, proliferation, and functions. To analyze the molecular mechanism responsible for the maturation arrest and functional defects of Hbb^th3/+ neutrophils, we measured the expression of PU.1, which is essential for terminal neutrophil maturation,^23,25,37  and C/EBPα, which is involved in myeloid lineage commitment.⁴⁰  Interestingly, we observed a significant reduction of PU.1 expression but not in C/EBPα expression in neutrophils isolated from Hbb^th3/+ mice, which correspond to an aberrant and/or arrested state of neutrophil maturation. Further supporting our observations are several key studies reporting failed neutrophil differentiation and function in PU.1-deficient mice.^25,37,38  Importantly, PU.1 dysregulation has been reported in many types of leukemia. In humans, loss or near loss of PU.1 expression has been found in acute myeloid leukemia, which is characterized by maturation arrest of myeloid cells.²⁵  PU.1 plays a central role in neutrophil maturation and function by coordinating the expression of myeloid genes associated with multiple functions, including terminal maturation (G-CSFR⁴⁰ ), phagocytosis (CD11b/CD18^62,63 ), ROS production^64-66  (NADPH oxidase), and miR-223 (lysosomal activities,⁴²  neutrophil elastase,⁶⁷  and myeloperoxidase production⁶⁸ ). Moreover, PU.1 is also involved in promoting apoptosis through direct transactivation of apoptosis-related genes, in particular the tumor necrosis factor–related apoptosis-inducing ligand.³⁹  Collectively, these data suggest that abnormal PU.1 expression in Hbb^th3/+ myeloid cells is a key mechanism underlying aberrant neutrophil functions and potentially extending the survival of dysregulated neutrophils.

To extend our observations beyond mice, we evaluated neutrophil morphology and PU.1 expression in blood neutrophils isolated from HbE/β-thalassemia patients. Importantly, morphological analysis revealed neutrophils isolated from HbE/β-thalassemia subjects displayed significant nuclear hyposegmentation as measured by the proportion of 2- to 4-lobed nuclei and a significant decrease in PU.1 expression in peripheral blood neutrophils compared with healthy controls. Among β-thalassemia patients, susceptibility to infection involves several complex biological processes and at least in part is influenced by blood transfusion, iron overload, iron chelation therapy, splenectomy, and vaccination status. Current clinical guidelines recommend treatment with antibiotics as soon as clinical symptoms of infection present in the patient.⁶⁹  In most countries where patients readily have access to such treatment, this prophylactic therapy helps reduce the frequency of severe infection and associated morbidity and mortality. In this study, we demonstrated that Hbb^th3/+ mice were more susceptible to S pneumoniae infection. However, other defects in Hbb^th3/+ mice, such as the significant alterations in splenic structure and concomitant loss of normal function, may also contribute to increased susceptibility to infection.

In preliminary experiments, we show that the appearance of the aberrant, immature neutrophil population is age-related, with 6-week-old Hbb^th3/+ animals, and mice up to age of 2 months old (supplemental Figure 6F), not showing any differences in neutrophils compared with age-matched WT mice. Interestingly, as demonstrated previously,^26,27  iron loading in the spleen of younger mice was much lower compared with the 6-month-old mice. It is tempting to speculate that the accumulation of iron in the spleen affects the expression of transcription factors such as PU.1, which in turn affects the terminal maturation of the Ly6G^highLy6C^low neutrophils. Further experiments are needed to address this hypothesis, including a more detailed analysis of PU.1 expression in relation to tissue iron levels, in which iron levels are deliberately increased through the use of iron-fortified diet or reduced through the use of iron chelation therapy.

Our data, together with several studies on dysregulated neutrophils, particularly in chronic inflammatory conditions, indicate that aberrant neutrophil functions represent a significant risk factor for disease-associated morbidity and mortality.^13,51  In addition, aberrant neutrophil functions may also contribute to other pathological processes, including organ damage and wound healing.^13,51,70,71  A better understanding of the mechanism(s) that underlie the immune defects may help the development of therapies for all β-thalassemia patients. Taken together, our data advance the understanding of the innate immune dysfunction in β-thalassemia. Reduced PU.1 expression is associated with aberrant neutrophil maturation and effector functions in β-thalassemia mice and human patients. We suggest that therapeutic strategies focusing on neutrophil replenishment and maturation may reduce comorbidities and improve quality of life for patients with β-thalassemia and other related hematological conditions.

The authors thank David Briles, University of Alabama at Birmingham, Alabama, for providing S pneumoniae strain WU2.

The authors acknowledge the financial support from the National Health and Medical Research Council, the Murdoch Childrens Research Institute, the Victorian Government’s Operational Infrastructure Support Program, Thalassaemia and Sickle Cell Society of Australia, Thalassaemia Society of New South Wales, and The Greek Conference. This work was also supported by Thailand Research Fund (DPG5980001, MRG5980043, and IRG5780009), Office of the Higher Education Commission and Mahidol University under the National Research University Initiative, Research Chair Grant, National Science and Technology Development Agency, Thailand and Faculty of Science, and Faculty of Medicine Ramathibodi Hospital. Supported was provided by the Royal Golden Jubilee PhD Research Scholarship from the Thailand Research Fund (P. Siwaponanan).

Contribution: P. Siwaponanan, J.Y.S., R.G., G.Z.-W.N., N.W., I.O., T.N., C.T., P.C., and O.W. performed experiments; P. Siwaponanan, J.Y.S., R.G., T.N., B.M., P. Sutton, C.T., S.F., P.C., S.S., O.W., and J.V. analyzed data and performed statistical analysis; S.F. and S.S. contributed to the patient material with clinical data; P.S., O.W., and J.V. designed the research; and P.S., J.Y.S., R.G., B.M., S.F., P.C., S.S., O.W., and J.V. wrote the paper.

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

Correspondence: Jim Vadolas, Murdoch Childrens Research Institute, The Royal Children’s Hospital, 50 Flemington Rd, Parkville, VIC 3052, Australia; e-mail: jim.vadolas@mcri.edu.au.