Severe congenital neutropenia (SCN) comprises a heterogeneous group of disorders characterised by early childhood onset of profound neutropenia and recurrent life threatening infections.1 2 The bone marrow usually shows a reduced number of mature myeloid cells with an arrest in myelopoiesis at the promyelocyte or myelocyte stage. Genetic analyses in patients with SCN indicated that 35–69% of cases were attributable to heterozygous mutations in the gene encoding neutrophil elastase (NE), ELA2.1^–7 Recently, homozygous germline mutations in the HAX1 gene were identified in a subset of patients with SCN (SCN3, OMIM 610738).8 Kostmann syndrome is a subtype of SCN, originally described in a Swedish population, with an autosomal recessive mode of inheritance.9 10 Klein et al8 reported 23 patients of European and Middle Eastern ancestry with HAX1 deficient SCN, in addition to the original Kostmann family in Sweden.

HAX1 is a ubiquitously expressed human gene.11 12 Although a number of cellular and viral proteins are known to interact with HAX1, its function is still not completely understood. With this limited knowledge, HAX1 has been assigned functions in signal transduction and cytoskeletal control. A recent study showed a significant inverse correlation between levels of HAX1 protein and the degree of cell death, suggesting a role for HAX1 in mammalian cell death.13^–15 Klein et al8 reported that HAX1 was important in maintaining the inner mitochondrial membrane potential and protecting against apoptosis in myeloid cells. Thus, HAX1 may be a regulator of myeloid homeostasis, underlining the significance of the genetic control of apoptosis in neutrophil development.

The first Japanese case of a HAX1 deficient SCN patient with neurodevelopmental deficit was recently reported.16 However, the prevalence and clinical features of patients with HAX1 mutations remain unclear. In the present study, we extended the analysis of genetic abnormalities, including those of the HAX1 gene, in Japanese SCN patients and identified another four patients with HAX1 deficiency who carried homozygous R86X and compound heterozygous R86X combined with R126fsX128 mutations. All of the patients presented with neurodevelopmental deficits and severe neutropenia.

METHODS

Patients

The diagnosis of SCN was made according to accepted criteria, including chronic neutropenia (below 500/μl in peripheral blood), maturation arrest at the promyelocyte or myelocyte level in the bone marrow, absence of circulating anti-neutrophil antibodies as determined by a granulocyte indirect immunofluorescence test, and the onset of severe infections at the age of less than 12 months. Eighteen Japanese patients with SCN (11 males, seven females) were enrolled in the study. The patients’ characteristics are summarised in table 1. Neurodevelopmental features were obtained from medical records. The haematological findings presented were from the time of diagnosis, before treatment with granulocyte colony stimulating factor (G-CSF). No significant differences in differential counts of blood cells of either peripheral blood or bone marrow were noted among the 18 SCN patients (data not shown). Seven patients received stem cell transplants (SCT) from HLA matched donors, and the others were administered G-CSF. None of the patients developed myelodysplastic syndrome and/or acute myelogenous leukaemia.

Sequencing of the HAX1 gene

This study was approved by the Institutional Review Board for Human Genome Research (No. 14, and 123, Hiroshima University, Japan). Peripheral blood samples were obtained, with informed consent from patients and/or their guardians. Genomic DNA was extracted from peripheral blood leucocytes. In patients who underwent SCT, the samples used were collected before SCT. Polymerase chain reaction (PCR) was performed using primers that spanned all HAX1 exons (5′-CGTTTACGACAGTGTCAGGATCG-3′ and 5′-TGACAAA CTGACATGG CCCCAG-3′) and an Expand High Fidelity PCR system (Roche Diagnostics GmbH, Mannheim, Germany), according to the manufacturer’s instructions. The PCR products were sequenced using a BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, California, USA), an ABI PRISM 310 genetic analyzer (Applied Biosystems), and the following primers: 5′-CGTTTACGACAGTGTCAGGATCG-3′, 5′-TGACAAACTGACATGGCCCCAG-3′, 5′-TTTTGATG TGGCAGCCAGTC-3′, 5′-CCCTAAGGGGCAACCTGTG-3′, 5′-GAAAGGTGGCAGGTGTTTGC-3′, 5′-GGCTGGTCTC CAACTCCTG-3′, 5′-GAAGGAGTGTGTAAATAAGGC-3′, 5′-GGTAAAGGAAAGTATGGCCAG-3′, and 5′-CTACAA AGAGAAGTCCCCAC-3′.

Western blot analysis for HAX1

Cell extracts from peripheral blood leucocytes were subjected to SDS-PAGE, and the proteins were transferred to polyvinyldifluoride membranes (Immobilon-P, Millipore, Bradford, Massachusetts, USA), using a semi-dry electroblotter (Atto Corp, Tokyo, Japan). The membranes were incubated with a monoclonal antibody against HAX1 (BD Biosciences Pharmingen, Franklin Lakes, New Jersey, USA), washed, and incubated with secondary anti-mouse antibody. Immunoreactions were visualised by enhanced chemiluminescence (Fujifilm Corp, Tokyo, Japan) using SuperSignal (Pierce Biotechnology Inc, Rockford, Illinois, USA).

HAX1 cloning

Total RNA was extracted from peripheral blood mononuclear cells (PBMCs) using Isogen (Nippon Gene Co, Tokyo, Japan), and cDNA was synthesised from 5 μg of total RNA by using a SuperScript first strand synthesis system for reverse transcriptase (RT)-PCR (Invitrogen, Carlsbad, California, USA). PCR of the full length HAX1 was performed with primers that span the entire coding region (5′–AATTGAATTCCGCCTCGCT CAATTTCTCACAGG-3′ and 5′–TATATCTAGATG AGAGGTGGAAAGGACTTGAAGG-3′). The PCR products were cloned into pGEM-T Easy vector (Promega, Madison, Wisconsin, USA). We sequenced several clones and obtained two types of HAX1 isoforms (isoforms a and b).

HAX1 expression in brain tissue

Total RNA extracted from cerebellum, cerebral cortex, and putamen were purchased from Clontech (Mountain View, California, USA). The cDNA from each was synthesised from 5 μg of total RNA. RT-PCR analysis of HAX1 expression was performed using the primers 5′–GTCTGCGAATGGAC CACTG-3′ and 5′-AAACCTATGAAATGGCCTCTG-3′.

RESULTS

Recent studies have identified several genes involved in SCN; these include ELA2, HAX1, Gfi-1, WAS, and P14.2^–7 17^–19 We analysed the sequences of these genes in 18 Japanese patients with SCN and found that five patients had homozygous or compound heterozygous mutations in the gene encoding HAX1, and 11 patients had heterozygous mutations in the ELA2 gene encoding NE (table 1). We did not identify any presently known gene defects in two of the SCN patients. None of the patients showed mutations in both ELA2 and HAX1. Thus, approximately 90% (16/18) of the Japanese SCN patients examined showed mutations in the gene encoding NE or HAX1. We found two novel mutations in the ELA2 gene, G203D and 380–382del (supplemental fig 1). The other eight mutations in the ELA2 gene have been reported previously.2^–7 These results suggest that a majority of Japanese SCN patients can be classified into two groups, ELA2 mutation and HAX1 deficiency, based on the analysis of these genes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Mutation and Western blot analyses of patient 1. (A) Sequences of the HAX1 gene in patient 1 and his parents. (B) Western blot analysis of HAX1 in patient 1 and his parents.

Sequence and Western blot analysis of HAX1

Three of the five patients (1, 2 and 3) with HAX1 gene mutations showed a homozygous single base pair substitution at 256 (256C>T), leading to a nonsense change, R86X (fig 1A). These three patients were unrelated. Two sibling cases (4 and 5) showed a compound heterozygous mutation, consisting of 256C>T, derived from their mother and a 59 bp deletion at nucleotides 376–434 derived from their father (fig 2A). This 59 bp deletion is, to our knowledge, novel and leads to a frameshift and a premature stop codon at nucleotides 441–443 (R126fsX128). Heterozygous carriers had no detectable phenotype, but a deficiency in HAX1 protein was detected by Western blot analysis of leucocytes from both homozygous and compound heterozygous patients (fig 1B, 2B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Mutation and Western blot analyses of patients 4 and 5. (A) Sequences of the HAX1 gene in patients 4 and 5 and their parents. (B) Western blot analysis of HAX1 in patient 5 and his parents.

Neurological findings in patients with homozygous R86X mutations and compound heterozygous R86X combined with R126fsX128 mutations in the HAX1 gene

The SCN patients with homozygous R86X mutations and compound heterozygous R86X combined with R126fsX128 mutations in the HAX1 gene specifically displayed mild to severe developmental delays, with or without epileptic seizures (table 1). The neurological findings in patients with homozygous R86X mutations and compound heterozygous R86X combined with R126fsX128 mutations are summarised in table 2.

The case presentation of patient 1 has already been reported in detail.16 In brief, his development was delayed since infancy, and both his mental and motor status have declined with aging. Neuroradiological studies demonstrated no abnormal findings. Since the age of 4 years, he has suffered from epilepsy.

The development of patient 2 was moderately delayed. She received an allogeneic bone marrow transplantation from an HLA matched sibling at 1 year 9 months of age. The conditioning regimens included busulfan and cyclophosphamide without radiotherapy. Graft versus host disease (GVHD) prophylaxis consisted of cyclosporine and a short course of methotrexate. However, she has been suffering from mild, chronic GVHD. Mild developmental delay was found before bone marrow transplantation. Since the age of 9 years, she has suffered from loss of consciousness for periods of several minutes, sometimes developing secondary generalised tonic–clonic seizures. An interictal electroencephalogram (EEG) showed diffuse irregular spike and wave discharges, predominantly in the right occipital region. Bone marrow transplantation successfully restored severe neutropenia, but did not improve or prevent the progression of neurodevelopmental deficits.

The development of patient 3 has been severely delayed since infancy. A computed tomography (CT) scan of the head performed at the age of 9 years demonstrated mild cerebral atrophy. She experienced two febrile seizures, at the ages of 5 and 6 years. She has presented with loss of consciousness and apnoea since the age of 9 years. An interictal EEG showed multifocal spike discharges, predominantly in the bilateral parieto-occipital region. Her physical growth was stunted. These three patients with epilepsy were successfully treated with antiepileptic drugs. Patient 1 was taking multiple drugs, while the other two were taking single drugs.

Patients 4 and 5 are siblings. The development of the elder sister (patient 4) has been moderately delayed, with the following developmental milestones: head control at 4 months, sitting alone at 10 months, standing without support at 24 months, and speaking single words at 20 months. At 50 months, her total developmental quotient (DQ) was 52. Magnetic resonance imaging (MRI) of the head conducted at the age of 5 years revealed no abnormal findings. She has not experienced seizures. At 5 years of age, she received an allogeneic bone marrow transplantation from an HLA matched unrelated donor, without complications. The conditioning regimens consisted of fludarabine phosphate, melphalan, anti-human T lymphocyte immunoglobulin and total lymphoid irradiation (4 Gy). GVHD prophylaxis consisted of tacrolimus and a short course of methotrexate. At present, there are no obvious improvements in her cognitive function or behaviour after bone marrow transplantation. The development of the younger brother (patient 5) was mildly delayed. His total DQ score was 70 at the age of 4 years. He also did not suffer from seizures.

Thus, all five of the SCN patients with homozygous R86X mutations and compound heterozygous R86X combined with R126fsX128 mutations in the HAX1 gene showed developmental delays. No family members, apart from the affected patients, showed neurological deficits. The three cases with the homozygous R86X mutation demonstrated more severe delays than those with compound heterozygous mutations. Furthermore, the three patients with the homozygous R86X mutation in the HAX1 gene suffered from epileptic seizures, although the pattern of seizures was not consistent among the patients. The neurological findings were not consistent among the HAX1 deficient patients; however, developmental delay and failure to thrive are probably common clinical features in HAX1 deficient SCN patients with the R86X mutation. In contrast, none of the SCN patients with ELA2 mutations presented with neurological abnormalities.

HAX1 expression in brain tissues

The HAX1 gene encodes two isoforms of the HAX1 protein, isoforms a and b, which consist of 279 and 231 amino acids, respectively. The transcription of both isoforms of the HAX1 gene was determined using total RNA extracted from brain tissues. The two transcripts were readily distinguishable in cerebellum, cerebral cortex, and putamen (fig 3). The 559 bp band corresponds to the transcript of isoform a and the 415 bp band is isoform b. Thus, two variants of the HAX1 transcript were identified in human brain tissues.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 HAX1 expression in human brain tissue. Both isoforms a and b were detected in cerebellum, cerebral cortex, and putamen (lanes A, B, and C, respectively). The 559 bp product indicates the transcript of isoform a and the 415 bp product indicates that of isoform b.

DISCUSSION

SCN constitutes a heterogeneous group of diseases. Heterozygous mutations in the ELA2 gene, which encodes neutrophil elastase, have been detected in 35–69% of SCN cases.1 3^–6 In addition, a number of genetic aberrations have been identified in patients with disorders associated with severe neutropenia.1 2 Klein et al recently defined a homozygous germline mutation in HAX1 in individuals with autosomal recessive SCN, known as Kostmann disease.8 In the present study, we analysed 18 Japanese patients with SCN and found five patients with mutations in the HAX1 gene. Among the remaining cases, 11 patients with SCN had heterozygous mutations in the ELA2 gene. None of the patients examined had mutations in both genes. These results suggest that, in Japan, about 60% of SCN cases are attributable to mutations in the ELA2 gene and about 30% involve the HAX1 gene. The frequency of ELA2 mutations is consistent with the data reported in the Severe Chronic Neutropenia International Registry (SCNIR).1 As suggested by Klein et al, the majority of Japanese patients with SCN can be divided into two groups, ELA2 and HAX1 mutations.

All of the HAX1 deficient SCN patients examined in the current study carried homozygous R86X mutations and compound heterozygous R86X combined with R126fsX128 mutations. Klein et al reported three types of homozygous mutations in the HAX1 gene in 23 affected patients of European and Middle Eastern ancestry.8 Nineteen of these 23 patients had a single nucleotide insertion in exon 2 that introduced a premature stop codon, W44X. The original Kostmann family had a homozygous mutation (Q190X), and one Iranian patient showed the R86X mutation, suggesting that the prevalence of HAX1 mutations differs among patients of European, Middle Eastern, and Japanese ancestry. Further studies are needed to determine accurately the frequencies of HAX1 gene mutations in patients with SCN.

In the current study, all of the Japanese patients with the R86X mutation had common findings of developmental delays, with or without intractable neurological deficits. We recently reported the first case of a HAX1 deficient Japanese patient with SCN associated with severe developmental delay and epilepsy (patient 1 in this report).16 Additionally, the original report on HAX1 deficiency included an Iranian patient with the R86X mutation who showed stunted growth.8 The original Kostmann report noted no neurological symptoms in the patients.9 However, Carlsson and Fasth further studied six of the original “Kostmann family” and reported that three of them had neurological deficits: one had mild mental retardation, one had speech deficits and gross motor retardation, and one had a behavioural disorder plus epilepsy.10 No patient with the W44X mutation has shown neurological deficits.8 Further, Germeshhasusen et al reported six additional patients with HAX1 mutations, suggesting an important role of HAX1 isoforms in the clinical phenotype.22

Transcripts encoding both the a and b isoforms of the HAX1 protein were affected in the current five patients. As shown in fig 4, isoform b lacks some of the amino acids encoded by exon 2. We demonstrated that both isoforms were expressed in human brain tissue, the cerebellum, cerebral cortex, and putamen (fig 3), consistent with the results of Germeshhasusen et al.22 Recently, it has been demonstrated that Hax1 is required to suppress apoptosis in lymphocytes and neurons, particularly in the central nervous system. Hax1 knockout mice have extensive apoptosis of neurons in the striatum and cerebellum.23 Taken together, these findings indicate that the presence of isoform b may be important for neuronal function because only patients with HAX1 mutations affecting isoform b presented neurological symptoms, as suggested by Germeshhasusen et al.22 These observations suggest that the association of neurodevelopmental abnormalities in HAX1 deficient SCN is related to specific mutations in the HAX1 gene. However, the difference in the severity of the neurodevelopmental deficits among patients with similar sites of mutation in the HAX1 gene remains unclear.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Two isoforms of the HAX1 protein. The HAX1 protein consists of isoforms a (A) and b (B), which consist of 279 and 231 amino acids, respectively. Isoform b lacks some of the amino acids encoded in exon 2. The 130–131insA mutation (W44X) affects the composition of isoform a, whereas the mutations 256C>T (R86X), 376-434del (R126fsX128), and Q190X lead to amino acid substitutions in both HAX1 isoforms.

The SCNIR has collected data to monitor the clinical course, treatments, and disease outcomes for patients with SCN.24 The registry received six reports of abnormal development that were not associated with known medical conditions. These included four reports of delayed sexual development in females and two reports of developmental delay. Based on analyses of approximately 300 patients in the SCNIR, neurodevelopmental disorders appear to be less frequent in patients with SCN. Although the exact frequency of HAX1 deficiency is unclear in the SCNIR cases, neurodevelopmental abnormalities in SCN patients with HAX1 deficiency due to mutations affecting both isoforms of HAX1 may be a specific clinical feature. A detailed study of many more cases with HAX1 deficiency is needed to establish the findings in a subtype of SCN, Kostmann disease.