Association between the neuron-specific RNA-binding protein ELAVL4 and Parkinson disease
- First Online:
- Cite this article as:
- Noureddine, M.A., Qin, XJ., Oliveira, S.A. et al. Hum Genet (2005) 117: 27. doi:10.1007/s00439-005-1259-2
- 244 Downloads
Inflammatory processes have been implicated in the cascade of events that lead to nerve cell death. In the nervous system, a number of genes involved in inflammation pathways are regulated post-transcriptionally via the interaction of their mRNAs with specific RNA-binding Hu proteins, the vertebrate homologues of the Drosophila ELAV (for embryonic lethal abnormal vision). The gene encoding ELAVL4, a member of the Hu family of proteins, is located 2 Mb from the chromosome 1p linkage region peak for age-at-onset (AAO) of Parkinson disease (PD) (LOD=3.41). Nine single-nucleotide polymorphisms (SNPs) in ELAVL4 were genotyped for 266 multiplex families (1,223 samples). Additional genotyping in 377 singleton families was performed for a subset of five SNPs (SNPs 1–5) that were not in linkage disequilibrium. SNP 2 (located in the first intron of ELAVL4) showed a strong significant association with AAO of PD (P=0.006), and SNP 5 (a coding SNP in ELAVL4) showed a moderately significant association (P=0.035). Haplotype analysis revealed that the A-C haplotype at SNPs 2 and 3 has the strongest significant association with AAO (P=0.0001) among all combinations of two or three loci. The A-C haplotype remained significant for AAO after the inclusion of the C allele at SNP 5 to this haplotype (A-C-C haplotype, P=0.00018). Although SNP 5 was found to associate with PD risk in the early-onset subset of PD families (at least one affected with AAO <40 years, 60 families), we believe that it is a by-product of its association with AAO. Taken together, these results suggest a potential role for ELAVL4 as a modifier gene for AAO of PD.
Parkinson disease [PD (MIM168600)] is a progressive neurologic disorder characterized by the loss of dopaminergic neurons within the substantia nigra (SN) of the midbrain. The cardinal symptoms of PD are bradykinesia, resting tremor, and loss of postural flexibility. The etiology of PD is challenging, as evidence suggests that both genetic and environmental factors could influence disease risk. While a small proportion (5%) of familial PD cases can be attributed to mutations in certain genes, such as α-synuclein (Polymeropoulos et al. 1997; Kruger et al. 1998) and parkin (Kitada et al. 1998), other genetic factors influencing idiopathic PD await identification.
Our recent work has shown that genetics also plays a role in controlling age-at-onset (AAO) of neurodegenerative disorders (Li et al. 2002). In that study, several linkage peaks for AAO of PD and Alzheimer disease (AD) were identified, two of which map to chromosomes 1 and 10. We speculated that these linkage regions might harbor candidate genes that could impact AAO of these disorders. In a follow-up study, we reported significant association between glutathione s-transferase omega-1 (GSTO1) and AAO for AD (P=0.007), and to a lesser extent, PD (P=0.026). GSTO1 is located under the linkage peak of chromosome 10 (Li et al. 2004). This finding was provocative for two reasons. First, common clinical and pathological features are found in AD and PD brains, most notably, inflammation. Second, GSTO1 is thought to be involved in the post-translational modification of interleukin-1, a major player in inflammatory pathways (Laliberte et al. 2003; Griffin and Mrak 2002; Grimaldi et al. 2000). Recent reports have demonstrated the therapeutic effectiveness of targeting immune and inflammation-associated components in neurodegenerative diseases (Weiner and Selkoe 2002; Ringheim and Conant 2004), including PD (Gao et al. 2003; Benner et al. 2004). This tremendous momentum will certainly be maintained by identifying additional biologic targets for such therapies.
Among all linkage regions identified in our previous study (Li et al. 2002), the region located between D1S2134 and D1S200 on chromosome 1p showed the strongest linkage signal for AAO in PD (LOD=3.41). Here, we search for factors that could be affecting AAO of PD in that chromosome 1p region. The gene ELAVL4, one of the mammalian homologues of the Drosophila ELAV (embryonic lethal abnormal vision, also known as HuD) (Good 1995), maps approximately 2 Mb away from the linkage peak on chromosome 1, and plays an essential role in the development of the nervous system by regulating the temporal and spatial pattern of gene expression, and consequently the proper development and maintenance of neuronal cells (Antic and Keene 1997a; Perrone-Bizzozero and Bolognani 2002; Yao et al. 1993; Akamatsu et al. 1999). In addition, the ELAV/Hu proteins are known to bind to AU-rich element (ARE) sequences in the 3′-untranslated region (3′UTR) of inflammation-associated factors (Nabors et al. 2001). Therefore, both positional and biological reasons led us to further investigate the role of ELAVL4 in PD.
Materials and methods
Individuals with PD and their families were collected by the Duke Center for Human Genetics (DCHG) Morris K. Udall Parkinson Disease Research Center of Excellence (PDRCE) and the 13 centers of the Parkinson Disease Genetics Collaboration (Li et al. 2002). All participating clinicians defined strict consensus clinical criteria prior to the ascertainment of families. Affected individuals possess at least two out of three cardinal signs for PD (bradykinesia, resting tremor, and rigidity), no atypical clinical features, and no other causes of parkinsonism. Unaffected participants demonstrated no signs of the disease. Individuals with history of neuroleptic therapy within 1 year prior to diagnosis, evidence of normal pressure hydrocephalus, or any clinical features suggestive of atypical or secondary parkinsonism, were excluded from this study.
Summary of family data
No. of families
No. of affected individuals
No. of unaffected individuals
In addition to the overall PD data set, we also analyzed two stratified data sets based on AAO. Early-onset families (EOPD, n= 60) are characterized by having at least one affected individual with AAO less than 40 years, while in late-onset families (LOPD, n=583) all affected individuals have an AAO ≥40 years.
DNA extraction and genotyping
Primers and probes for SNP 5 were made as follows: TET-CCTTCTGCTTGTCCCCCCAGGTTCT, and FAM-CCTTCTGCTTGTTCCCCCAGGTTC. Both probes were modified with BHQ-1 at the 3′ end. Primers were 5′-GTGTGTTATCCTTGGTCAGACTGATG-3′ and 5′-CTGTGTGACCAGGGATGTTCATT-3′. PCR amplification was performed in 5-μl reactions (3 ng dried DNA, 1×TaqMan universal PCR master mix from Applied Biosystems, 100 ng of each primer, 200 nM of each probe) using the GeneAmp PCR system 9700 thermocyclers (Applied Biosystems). Amplifying conditions were as follows: [1×(50°C/2 min; 95°C/10 min); 40×(95°C/15 s, 60°C/1 min)]. Taqman primer and probe sets for the remaining SNPs were purchased from Applied Biosystems as Assays on Demand. The fluorescence generated during the PCR amplification was detected using the ABI Prism 7900 HT sequence detection system and analyzed with the SDS software (Applied Biosystems).
All SNPs were tested for Hardy-Weinberg equilibrium (HWE) and LD in the affected group (one affected from each family) and the unaffected group (one unaffected from each family). An exact test implemented in the Genetic Data Analysis (GDA) program (Zaykin et al. 1995) was used to test HWE, in which 3,200 replicate samples were simulated for estimating the empirical P value. We used the GOLD (Graphical Overview of Linkage Disequilibrium) program (Abecasis and Cookson 2000) to estimate the Pearson correlation (r2) of alleles for each pair of SNPs as the measurement of LD. The r2 ranges from 0 (no LD) to 1 (perfect LD). Since there is no clear definition on how to interpret the intermediate r2, we chose an arbitrary cutoff by considering two markers in strong LD if r2>0.60.
We performed association analysis on the markers that are not in strong LD using both multiplex and singleton data sets. AAO was treated as a quantitative trait. We used both the orthogonal model (OM) (Abecasis et al. 2000) and Monks-Kaplan (MK) method (Monks and Kaplan 2000) implemented in the QTDT program to test the association between markers and AAO. In addition to providing an association signal, the MK method also detects the direction of association; i.e., negative association for a certain allele is declared when the majority of carriers of this allele have AAO lower than the average AAO, and vice versa for positive association. Strong significant results were declared for markers with P values that meet Bonferroni correction (0.05/the number of markers). We also applied less stringent criteria by setting the significance level at 0.05 to declare moderate significance for markers that did not meet Bonferroni correction.
To perform the haplotype association analysis for AAO, we first used the FBAT—o option (Laird et al. 2000) to estimate the optimal offset of the AAO for each marker. We then performed the HBAT—e option (Horvath et al. 2004) on the adjusted AAO data (subtracting AAO with the average optimal offset estimate) for testing the association between haplotypes and AAO. We tested the association between haplotypes and AAO for all possible combinations of two or three loci. Since a maximum of four haplotypes is tested for each pair of two loci, we set the significance level at 0.0125 (i.e., 0.05/4). Similarly, we used the significance level of 0.00625 for three loci cases.
The pedigree disequilibrium test (PDT) (Martin et al. 2000, 2003) was used to determine association between markers in ELAVL4 and PD risk. Two PDT statistics were used: the PDT-sum statistic, which examines allelic effects, and the genotype-PDT, which examines genotypic effects. Association with PD risk was analyzed in total PD, EOPD, and LOPD data sets. We used HBAT—e option to perform haplotype analysis for the disease trait. We tested the association between haplotypes and PD for all possible combinations of two or three loci.
Pairwise Pearson correlation (r2) for all SNPs genotyped in ELAVL4. The lower triangle is for the unaffected group and upper triangle is for the affected group. Bold numbers represent strong LD (r2>0.6) in both the affected and unaffected groups
SNP 5 is a coding polymorphism (results in a non-synonymous amino acid change), lies within the 3′-most exon, and has a minor allele frequency of 9.7%. The remaining four SNPs typed in ELAVL4 region are intronic, and have a minor allele frequency of 20% or higher. All markers were in HWE.
Allelic association results between markers and AAO obtained by using the orthogonal model and the Monks-Kaplan method
Minor allele frequency
Allelic association between markers and PD risk by using PDT
Overall PD (644a)
In this study we find that a member of the RNA binding proteins, ELAVL4 (HuD), is significantly associated with AAO trait in PD. Our results show that one polymorphism in ELAVL4 (SNP 2) is significantly associated with AAO of PD, while another polymorphism (SNP 5) shows a moderately significant effect. By conducting haplotype analysis for all combinations of two or three loci, we found that the combinations of SNPs 2–3 and 2-3-5 revealed the most significant haplotype association. Overall, the combination of SNP 2-A, SNP 3-C, and SNP 5-C alleles (haplotype A-C-C) associated with earlier AAO, which is consistent with individual allelic association.
Clearly, the haplotype analysis provides more information than single locus association analysis. Although SNP 3 alone does not show significant association with AAO, the SNP 3-C allele clearly acts together with SNP 2-A and has the strongest effects on AAO among other loci combinations. The region surrounding SNPs 2 and 3 may harbor additional important variants for AAO in PD.
Although SNPs 2 and 5 show significant association with AAO, it does not warrant them a role of increasing or reducing disease risk. For AAO, we are interested in whether individuals with certain polymorphism tend to have earlier or later AAO. In other words, only affected individuals provide phenotypic information in the analysis. Therefore, although SNP 2 shows the most significant association with AAO, affected and unaffected groups may not necessarily have a significant difference in the number of SNP 2 alleles. In this study, we did not find any risk effects contributed by SNPs 2 and 5 in the overall PD data set. The only significant risk effect was observed at SNP 5 in the early-onset PD data set, in which the T allele significantly decreases the risk of PD. We believe that the protective effect of SNP 5-T in AAO of PD (associated with later AAO) may have resulted in the tendency of early-onset PD patients to have less SNP 5 (T allele) than the late-onset PD patients. Therefore, it is possible that discordant siblings within the early-onset group sharing less T allele at SNP 5 than expected, leading to the negative association between SNP 5-T and PD (Fig. 3). Based on this pattern, we speculate that SNP 5 plays a role in modifying AAO of PD rather than risk.
This is the first report that demonstrates allelic association between a member of the ELAV/Hu gene family and neurodegenerative disease. The ELAV/Hu genes were first identified in Drosophila as essential in the development and maintenance of the nervous system (Campos et al. 1985; Antic and Keene 1997b). The mammalian homologues of this family are known as the ELAV-like proteins or Hu antigens (ELAVL1/HuR, ELAVL1/Hel-N1, ELAVL3/HuC and ELAVL4/HuD) (Brennan and Steitz 2001), and have been implicated in human disease. Patients who suffer from paraneoplastic neurologic syndrome (Hu syndrome) generate autoantibodies that readily cross the blood-brain barrier and recognize ELAV/Hu proteins as target epitopes (Szabo et al. 1991; King et al. 1999; Benyahia et al. 1999). As a result, patients with this disorder develop neurologic degenerations that affect discrete areas of the nervous system, including the dorsal root ganglia, the limbic system, cerebellum, brainstem, motor, or autonomic nervous system (Dalmau and Posner 1999). These findings paved the way for additional work that established the role of such molecules in neuronal cell function (Perrone-Bizzozero and Bolognani 2002).
At the molecular level, the ELAV/Hu proteins regulate mRNA stability by binding preferentially to adenine and uridine (AU) rich elements (AREs; AU-rich are 20, 51 and 19 nt in length) in the 3′UTR of mRNA (Ma et al. 1997; Levine et al. 1993). Target AREs for ELAV/Hu protein binding have been identified in the 3′UTR region of several genes, including the microtubule-associated protein tau (Aranda-Abreu et al. 1999), and neuroserpin, an axonally secreted serine protease inhibitor that protects neurons from ischemia-induced apoptosis (Cuadrado et al. 2002). ELAV/Hu proteins also bind to the 3′UTRs of others transcripts, particularly those encoded by inflammation-associated genes such as TNFα and COX-2 (Sakai et al. 1999; Cok et al. 2003; Sengupta et al. 2003; Sully et al. 2004; Nabors et al. 2001). Curiously, recent findings suggest that brains from patients with PD have elevated density of glial cells expressing inflammatory-associated factors and proinflammatory cytokines, such as cyclooxygenase-2 (COX-2), TNF-α, IL-lβ, IL-1, IL-6, compared with age-matched controls (Hunot and Hirsch 2003).
It is possible that the variations in ELAVL4 that we have analyzed in this study could be influencing AAO by affecting this gene’s interaction with one or more of its target 3′UTRs. Alternatively, other polymorphisms between SNP 2 and 3 may be the actual loci affecting AAO. More work is needed to address these questions.
We are grateful to all of the families whose participation made this project possible. We thank the members of the PD Genetics collaboration Martha A. Nance, Ray L. Watts, Jean P. Hubble, William C. Koller, Kelly Lyons, Rajesh Pahwa, Matthew B. Stern, Amy Colcher, Bradley C. Hiner, Joseph Jankovic, William G. Ondo, Fred H. Allen Jr., Christopher G. Goetz, Gary W. Small, Donna Masterman, Frank Mastaglia and Jonathan L. Haines who contributed families to the study. We would also like to thank Dr. Eden Martin, Dr. Bill Scott, Dr. Jack Keen, and Dr. Peter King for helpful discussions on this work. This research was supported by the NIH/NINDS R01 NS311530-10 and P50-NS-039764 grants.