Neuroglobin (Ngb, LocusID: 58157; GenBank: DQ008010) is a newly discovered globular heme protein in the vertebrate brain that displays a high affinity for oxygen like myoglobin (Mb) and hemoglobin (Hb) 1. Ngb is preferentially expressed in the neurons, as well as some endocrine tissues 12. The highest Ngb concentration has been found in the retina, which is also the highest O₂-consuming organ of the body 3.

Ngb is hypoxia-inducible in cultures of cerebral cortical neurons 4 and some neuronal cell lines, such as PC12, immortalized rat hippocampal neuron (HN33) 56 and hybrid dorsal root ganglia neuroblastoma cell (ND15) 7. Prolonged or sustained hypoxia increased Ngb expression in vitro 5 or in vivo 8. The increasing magnitudes of Ngb mRNA levels, ranging from 150% to 500%, can be explained by various experimental conditions or cell lines 689. For example, Ngb mRNA was time-dependently upregulated during the first 8 h in a constant hypoxic environment of 1% O₂, while no significant upregulation occured before 48 h in 3% O₂ 6. However, there were contradictory data as to whether Ngb was upregulated under hypoxic conditions in brain 1011. Several factors, namely species, brain region, severity and duration of hypoxia, and the pattern of hypoxic exposures, could have accounted for such disparate findings 9. Cerebral ischemia is another stimulus to Ngb expression, although findings vary depending on the model used and the experiment designed 12. Focal cerebral ischemia is clearly associated with Ngb induction, especially in the penumbra 4. Transient global forebrain ischemia increases Ngb expression in the gerbil cerebral cortex and serum, but not in the hippocampus 13.

Ngb appears to show the neuroprotective potential following hypoxic and ischemic insults 131415. In cultured cortical neurons, antisense-mediated down-regulation of Ngb decreases cell viability under hypoxia, whereas additional Ngb improves HN33 cells survival 4. After transfected with Ngb-expressing plasmid, human neuroblastoma cells (SH-SY5Y) are resistant to oxidative injury induced by H₂O₂ 16. Intracellular delivery of Ngb by human immunodeficiency virus-1 transactivator of transcription (TAT) protein transduction domain fails to rescue rat retinal ganglion (RGC-5) or SH-SY5Y cells from combined oxygen and glucose deprivation 17, whereas in another case, Ngb-TAT-treated cultured rat cortical neurons shows reduced sensitivity to hypoxia 18. Administration of anti-sense oligodeoxynucleotides directed against Ngb exacerbates experimental stroke in vivo, while intracerebral administration of Ngb with an adeno-associated virus vector reduces the focal cerebral infarct size indeed 15. Furthermore, overexpression of Ngb in a transgenic mouse model reduces cerebral infarct size following middle cerebral artery occlusion (MCAO) 1419.

However, the mechanisms underlying Ngb-mediated neuronal protection during hypoxic/ischemic stress remain largely unknown. It seems to us unlikely that Ngb is primarily an O₂ transporter, because of its fairly low average tissue concentration (approximately micromolar) compared with Mb (0.2 mM or so) in muscles 20. Several proposed mechanisms include Ngb acting as an oxygen sensor and storage molecule 2122, operating as a guanine nucleotide dissociation inhibitor and then increasing levels of free Gβγ 23, interacting with neuronal membrane proteins including Na-K-ATPase 24 and flotillin-1 25, and even acting as an intracellular ROS (reactive oxygen species)/RNS (reactive nitrogen species) scavenger 2026 via reduction reactions leading to ferric-Ngb conversion to ferrous-Ngb by endogenous reducing enzyme systems 27. Ngb not only decreases oxidative stress induced ROS/RNS overproduction and lipid peroxidation 41428, but also attenuates subsequent mitochondrial dysfunction, apoptosis, and cell death 3029. Its antioxidant properties were consistent with the protective role against oxidative stress-induced injury 31. Recently met NGB (ferric-NGB) was found acting not only as scavenger of reactive species, but also as a target of the self-generated reactive species 32.

Whatever its function and mechanism is, it can be summarized that the majority of studies agrees that Ngb is beneficial for neuronal survival during hypoxic/ischemic stress. Based on aforementioned considerations, we hypothesize that Ngb may be one of the candidate genes associated with ischemic stroke. Human Ngb, located on chromosome 14q24, has a 3 intron/4 exon structure 133. The aim of this study is to investigate the association between Ngb polymorphisms or genotypes and ischemic stroke in the Southern Chinese Han population, which is the first allele association study on human Ngb and stroke.

For Ngb is hypoxia/ischemia-upregulated and shows the neuroprotective potential during hypoxic/ischemic stress, the Ngb polymorphisms may be one of the candidates for being a genetic factor associated with ischemic stoke. In the present study, we have investigated the association of Ngb polymorphisms with ischemic stoke in the Southern Chinese Han population.

It has been suggested that the pathogenesis in ischemic stroke may be heterogeneous, differing between patients with LVD stroke, SVD stroke or CE stroke. Therefore, the studies of genetic factors for ischemic stroke should subdivide according to the different clinical or etiological subtypes to avoid false positive or negative results 3839. TOAST criteria were used to determine stroke etiology, because they are thought to be the best available clinical criteria to separate different stroke etiologies, although they present an imperfect relationship to the underlying inherited disease mechanism.

The cDNA sequence of Ngb is highly conservative. No mutations or polymorphisms within cDNA sequence were found in subjects analyzed. However, two polymorphisms 89+104c>t and 322-110(6a)>5a were detected in exon flanking sequence (95–303 bp). Other polymorphisms like rs28988618, rs101400032, rs7149300 and rs10133981 were not identified in the present study.

According to the SNP on NCBI Reference Assembly, the t carrier frequency of 89+104c>t was 50%, 21.4%, 66.7% in European, African and Asian populations respectively 40. All the above data came from small sample ranged from 14 to 46 subjects. The t carrier frequency of the control group in the present study was 58.5%. The 6a allele frequency of 322-110 (6a)>5a was lower in the control gruop of our study (52.5% vs 75%, 87.5% and 63% in European, African and Asian populations) 41.

The significant differences in 89+104c>t genotype distribution and allele frequency were found between LVD patients or SVD patients and controls, which indicated this polymorphism might be correlated with LVD and SVD. T allele frequencies of LVD patients (36.4%) and SVD patients (40.3%) were significantly lower than that of the controls (58.5%). Conditional logistic regression analysis showed that the presence of at least one t allele in 89+104c>t were probably potential independent protective factor, while hypertension was independent risk factors for both LVD and SVD stroke. DM2, TG and LDL-C were independent risk factors for LVD stroke but not for SVD stroke, while HDL-C was protective factor for LVD stroke.

As we all known, confounding bias should conceal or exaggerate the association between exposure factor and the disease, thus resulted in the false negative or false positive. Age-bias and gender-bias are important considerations in case-control study for cerebrovascular disease. Data was analyzed by cluster stratification analysis to exclude the interference of these confusion factors in this study. T allele of 89+104c>t might be one of the independent protect factors for LVD in females and people over 60 years. The risk of LVD was further decreased in female t allele carriers but not in males. Age stratified analysis suggested that t allele could reduce the LVD risk on people over 60 years, but its protective effect on people under 60 years was not significant. We also confirmed that t allele might be a new protective factor for female SVD. Since t allele was much more frequent in females than in males, and t allele was significant protective especially for females, we presumed that there should be some correlations between Ngb protective effect and female hormones, which remains to be confirmed by further studies.

The genotype distributions and allele frequencies of Ngb 322-110 (6a)>(5a) were similar between cases and controls, which suggesting this site had no obvious correlation with LVD or SVD. However, the haplotype 89+104c/322-110(5a) showed a higher distribution frequency in cases than in controls (OR = 1.838 for LVD, 1.997 for SVD separately), indicating significant association with LVD and SVD. Gradual increase of statistical power with the inclusion of two polymorphisms supports the validity of our conclusion that the Ngb is a susceptibility locus for ischemic stroke.

Our data also showed that Ngb 89+104 t allele carriers have a lower LDL-C level (P = 0.003), less frequent hypercholesterolemia (P = 0.008) and hypertension (P < 0.0001) compared with c/c homozygotes. It is possible that Ngb plays its role against ischemic injury related to blood lipid metabolism or blood pressure regulation, which probably find new way to investigation of Ngb protective mechanism. When the sex was analyzed separately, we found that the carrier of t allele have low LDL-C (P = 0.002), slightly high HDL-C level (P = 0.012) and low frequent hypercholesterolemia (P = 0.003) and hypertension (P < 0.0001) in male subgroup. We concluded that the 89+104c>t polymorphism show significant association with plasma lipids level (LDL-C and HDL-C), hypercholesterolemia and hypertension especially in Chinese male.

As we all known, exonic polymorphisms have a potential to change amino acid coding sequence. The current model of pre-mRNA splicing is based on the recognition of four canonical intronic motifs (5' splice site, branchpoint sequence, polypyrimidine (PY) tract and 3' splice site), however, it has become clear that the four canonical splice elements do not contain adequate sequence information to ensure accurate splicing. A few families of motifs that are over-represented upstream of weak PY tracts were identified and suggested as intronic splicing enhancers (ISEs) that appear to compensate for a weakened canonical pre-mRNA splicing motif in both short and long human introns 424344. Some intronic SNP, for example, the intronic prothrombin 19911A>G is itself functional and changes splicing efficiency by altering a known functional pentamer motif CAGGG 45. The intronic 89+104 c>t SNP could probably be assumed to influence the splicing efficiency of Ngb by toggling between CCGGG and CTGGG, which is one of the above motifs. Ngb's structure shows a peculiar internal cavity of very large size. Binding of heme ligands is associated to a conformational change involving the heme that "slides" into the pre-existing cavity and makes the sixth coordination position available 46. So the spatial conformation of Ngb is especial important. We presumed that whether this intronic SNP influenced nucleotide splicing and formed different spliceosomes which might change the classical α helical and three-over-three sandwich structure of Ngb, and then resulted in the interference of protein folding, instead of changing the amino acid coding sequence. Of course, additional experimental evidence will need to be obtained to verify the functional significance of the SNP identified by our study.

Some potential biases may still have influenced our final results. Prevalence of risk factors and stroke subtypes differ between hospitalized and community patients with ischemic stroke. This case-control study was mainly hospital-based, with comprehensive and accurate data, but there would be inevitable selection bias compared with community-based study. The lack of integrity data for carotid ultrasound examination in controls may be another weakness of this study, thus we have to reject carotid internal media thickness data in our study.

In conclusion, t allele of 89+104c>t (rs4903565) might have a protective effect against LVD and SVD in the Southern Chinese Han population, especially for female or people over 60 years. Ngb 322-110(6a)>5a polymorphism showed association with neither LVD nor SVD subtype. However, there was a "hitchhiking" effect for ischemic stroke was observed for the 89+104 t/322-110 (6a) genotype combination especially for LVD. To our knowledge, this is the first study on correlation between Ngb polymorphisms and ischemic stroke has been investigated. These results need to be further confirmed by some multicenter case-control study or prospective study in different ethnic populations.

Ngb: Neuroglobin; Mb: myoglobin; Hb: hemoglobin; LVD: stroke caused by large-artery atherosclerosis; SVD: stroke caused by small-vessel occlusion; BMI: body mass index; DM2: diabetes mellitus type 2; LDL-C: low density lipoprotein cholesterol; HDL-C: high density lipoprotein cholesterol; TC: total cholesterol; TG: triglycerides; FIB: fibrinogen; PCR: polymerase chain reaction; RFLP: Restriction fragment length polymorphism; PY: polypyrimidine; ISEs: intronic splicing enhancers.

Neuroglobin, a newly discovered member of the globin superfamily, is expressed predominantly in brain neurons, and modulate hypoxic-ischemic insults. Although the mechanisms underlying Ngb-mediated neuronal protection are still unclear, Ngb appears to be one of the candidate protective factors for ischemic stroke. This is the first case-control study to clarify the association of Ngb polymorphisms with ischemic stroke in the world. And we found that Ngb polymorphism 89+104 t had protective effects on ischemic stroke, including large-artery atherosclerosis (LVD) and small-vessel occlusion (SVD) strokes in the Southern Chinese Han population. Meanwhile, a "hitchhiking" effect was observed for the 89+104 t/322-110 (6a) genotype combination especially for LVD.

YL carried out the molecular genetic studies, participated in the analysis of the data and drafted the manuscript. LF and XHX collected demographic data and risk factor profiles of subjects and participated in analysis and interpretation of data. SXMR analyzed the clinical data of all subjects. NW participated in the design of the study and the acquisition of data. ZYW conceived of the study, and participated in its design and coordination, and revising the manuscript critically for important intellectual content. All authors have read and approved the final version of the manuscript.