Introduction

Parkinson’s disease (PD) is a common movement disorder, characterized by resting tremor, rigidity, and bradykinesia [1]. There are a growing number of PD patients in China [2]. Alpha synuclein deposition is the hallmark of PD [3].

To discover genes associated with PD will help us better understand the pathogenesis of PD. Iron metabolism is associated with neurodegeneration [4]. Recently, HMOX-1, a gene related with iron metabolism, has been paid much attention to several disease with neurodegeneration, such as restless legs syndrome (dopamine A11 deficiency) and multiple sclerosis (disrupted myelin synthesis) [5, 6]. Two single nucleotide polymorphisms (SNPs), HMOX1 rs2071746 (35,380,679 A > T) and HMOX1 rs2071747 (35,381,192 G > C), were studied in these studies. HMOX1 rs2071746 was proven to be associated with restless legs syndrome and multiple sclerosis. In the study conducted by Ayuso et al., HMOX1 rs2071746 was also associated with PD [7]. Since PD is also a neurodegenerative disorder with dopaminergic deficiency, there might be a similar cause with restless legs syndrome and multiple sclerosis [8]. In this study, we tried to discover the association between HMOX1 and sporadic PD in southern Chinese.

Methods

Study population

PD was diagnosed by movement disorders specialists with diagnostic criteria brought up by movement disorders society (MDS) [9]. We excluded secondary causes, such as toxic, inflammatory, vascular, and drug-induced parkinsonism. Parkinsonism with neurodegenerative disorders, such as multiple system atrophy and progressive supranuclear palsy, were also excluded. PD patients with positive family history were also excluded. Control subjects were recruited through local healthy communities and evaluated by movement disorders specialists. All PD patients and healthy control subjects were all in our local communities. They were all Han ethnicity. We also recorded the history of hypertension and diabetes mellitus. Patients whose age was below than 45 years old were regarded as early onset PD (EOPD). As for PD patients whose age was above than 45 years old, late onset PD (LOPD) was regarded [10].

All participants and their caregivers were fully informed. They signed written consent forms after informed. This study was approved by the ethic committee of East Hospital, Tongji University School of Medicine. There were no participants whose age was under 18 years old when we assessed.

DNA preparation and genotyping

We collected 2 ml blood samples from PD patients and healthy controls with phenol–chloroform-isopropyl alcohol method for extracting DNA. Primer BLAST of National Center for Biotechnology Information (NCBI) was used to design primers for polymerase chain reaction (PCR). The primers for PCR were as follows: Forward primer of HMOX1 rs2071746: TGG CAT CCT GCT TTA TGT GTG; reverse primer of HMOX1 rs2071746: TCT GGG TGT GAT TTT GCT CC; forward primer of HMOX1 rs2071747: CGC CGA GCA TAA ATG TGA CC; reverse primer of HMOX1 rs2071747: AGG GAC ACT CCC ATC TTG GT. PrimeSTAR HS DNA Polymerase (Takara, Code No. R010Q) was used into PCR. The PCR system was as follows: 5 × PrimeSTAR buffer 10 μl, dNTP mixture 4 μl, forward primer 1 μl, reverse primer 1 μl, DNA template 1 μl, PrimeSTAR HS DNA polymerase 0.5 μl, and ddH2O 32.5 μl. The PCR condition was as follows: 95 °C 2 min—95 °C 30 s—54 °C 30 s—72 °C 30 s – back to step 2 for 40 cycles—72 °C 5 min—16 °C forever. SeqMan software was used to identify SNPs.

Statistical analysis

R (version 3.5.0) and CATT package (version 2.0) and epitools package (version 0.5–10) were used to perform statistical analysis. Student t-test was used to compare differences of age between PD patients and controls. Chi-square analysis was used in comparing differences of gender, assessing Hardy–Weinberg equilibrium (HWE) and allele distribution between PD and controls. We also analyzed genetic models. As for rs2071746, the additive model is AA vs AT vs TT. The dominant model is AA vs (AT + TT). The recessive model is (AA + AT) vs TT. The overdominant model is AT vs (AA + TT). As for rs2071747, the additive model is GG vs GC vs CC. The dominant model is GG vs (GC + CC). The recessive model is (GG + GC) vs CC. The overdominant model is GC vs (GG + CC). Logistic regression model with odds ratio (OR) with its 95% confidence interval (CI) for each SNPs of HMOX1 according to dominant, recessive, and overdominant models was calculated. We also adjusted age, gender, and education. Cochran-Armitage trend test was used to calculate additive models. Genetic powers were also calculated [11, 12].

Results

In all, 500 healthy controls and 500 PD patients are included in this study. There was no difference between age, gender, and education. There was no difference of the distribution of positive medical history hypertension and diabetes mellitus in two groups. Twelve (2.40%) patients suffered EOPD, so the analysis between EOPD and control was not available (Table 1). There was imbalance between PD patients and healthy controls of rs2071746. Since there was Hardy–Weinberg disequilibrium in two groups of rs2071746, the analysis of rs2071746 could be ignored (Table 2).

Table 1 Demographic information of cases and controls
Full size table

As for HMOX1 rs2071747, the dominant model was shown statistically significant between PD patients and control after adjustment of age and gender (Dominant model: p value: 0.045, OR: 1.51, 95% CI: 1.01–2.27, adjusted). There is also a trend of association between overdominant model and PD after adjustment of age, gender, and education (overdominant model: p value: 0.064, OR: 0.68, 95% CI: 0.45–1.02, adjusted) (Table 3).

Table 2 Association of models and OR to PD risk of HMOX1 rs2071746
Full size table

As for the comparison between LOPD and control of rs2071747, the associations of additive model and dominant models were found (additive model: p value: 0.047; dominant model: p value: 0.035, OR: 1.55, 95% CI: 1.03–2.34). The association of overdominant model was found (p value: 0.050, OR: 0.66, 95% CI: 0.43–0.99). However, the association was lost after adjustment of age and gender (p value: 0.050, OR: 0.66, 95% CI: 0.43–1.00, adjusted) (Table 3).

Table 3 Association of models and OR to PD risk of HMOX1 rs2071747
Full size table

Discussion

In this study, we found the association of the dominant model of HMOX1 rs2071747 with PD patients. We also found the associations of additive model and dominant model of HMOX1 rs2071747 with LOPD patients. To our knowledge, this is the first study that investigates the association between HMOX1 and PD in southern Chinese population.

HMOX1 encodes an enzyme, heme oxygenase-1 (HO-1), a wide range of pro-oxidant, and inflammatory stimuli [13]. HO-1 is a protective factor of several antioxidant pathways associated with PD, such as Nrf/HO-1 pathway and HO-1/biliverdin reductase pathway [13,14,15,16]. Since it is a key factor as an antioxidant, it could mediate the pathogenesis between several disease or lifestyle and PD, such as high-fat diet induced obesity and diabetes mellitus. Microglial activation and the resulting neuroinflammation are associated with PD [17]. Enhanced HO-1 expression can reduce the neuroinflammation and microglial activation [17, 18].

Besides, HO-1 also participated in iron metabolism. Iron is the most abundant transition metal in human’s brain [19]. It is important to mitochondrial respiration, myelin synthesis, and neurotransmitter synthesis [20]. There is iron deposition in the brain of PD [21]. Nrf2/HO-1 signaling pathway could prevent mitochondria-dependent apoptosis in PD model with treatment [22]. Thus, HO-1 could work as both antioxidant and iron metabolism in PD [23]. Iron could be a treatment target of PD [20, 24].

In our study, we did not find the association between HMOX1 rs2071746 and PD, which is inconsistent with Ayuso et al. [7]. This might be associated with the different genetic background between two different studies.

Our study used widely accepted diagnostic criteria, the MDS diagnostic criteria. Our patients are all sporadic PD. Besides, we observed two new loci (rs2071746, rs2071747) to PD.

This study has some weakness and limitations. First, we did not perform other rating scales to assess PD. Second, we did not test other locations of HMOX1. Third, our study is a single-center study with limited amount, especially EOPD. More multicenter and larger studies are expected.

In conclusion, we found the association of the dominant of HMOX1 rs2071747 with PD. We also found the associations of additive model and dominant model of HMOX1 rs2071747 with LOPD patients. More larger and multicenter studies are warranted.