Dopa-Responsive Dystonia: Functional Analysis of Single Nucleotide Substitutions within the 5’ Untranslated GCH1 Region

Background Mutations in the GCH1 gene are associated with childhood onset, dopa-responsive dystonia (DRD). Correct diagnosis of DRD is crucial, given the potential for complete recovery once treated with L-dopa. The majority of DRD associated mutations lie within the coding region of the GCH1 gene, but three additional single nucleotide sequence substitutions have been reported within the 5’ untranslated (5’UTR) region of the mRNA. The biologic significance of these 5’UTR GCH1 sequence substitutions has not been analyzed. Methodology/Principal Findings Luciferase reporter assays, quantitative real time PCR and RNA decay assays, combined with bioinformatics, revealed a pathogenic 5’UTR GCH1 substitution. The +142C>T single nucleotide 5’UTR substitution that segregates with affected status in DRD patients, substantially attenuates translation without altering RNA expression levels or stability. The +142C>T substitution disrupts translation most likely by creating an upstream initiation start codon (uAUG) and an upstream open reading frame (uORF). Conclusions/Significance This is the first GCH1 regulatory substitution reported to act at a post-transcriptional level, increasing the list of genetic diseases caused by abnormal translation and reaffirming the importance of investigating potential regulatory substitutions in genetic diseases.


Introduction
Dopa-responsive dystonia (DRD) was first described by Segawa et al. [1]. It is a movement disorder of childhood onset, typically beginning in a foot followed by spread, over several years, to involve other limbs, rendering walking difficult. DRD displays a diurnal fluctuation with increasing severity over the course of the day and a marked, sustained response to low therapeutic doses of levodopa. However, diagnosis remains a challenge with many cases being misdiagnosed as cerebral palsy or spastic diplegia. DRD is frequently caused by heterozygous mutations in the guanosine triphosphate (GTP) cyclohydrolase 1 gene (GCH1) [2,3], with this form of dystonia now referred to as DYT5 [4]. GTP cyclohydrolase 1 catalyzes the first step in the biosynthesis of (BH4). BH4 is a cofactor for three aromatic amino acid hydroxylases, including tyrosine hydroxylase (TH), which is the rate-limiting enzyme in dopamine biosynthesis [5]. Thus, low levels of BH4, due to mutations in the GCH1 gene, can lead to a reduction in dopamine synthesis and dysfunction of dopaminergic nigrostriatal neurochemical transmission [6].
Noncoding GCH1 mutations have also been reported spanning its 5' upstream region [16][17][18] and intron-exon splice sites [19][20][21]. Within the 5' upstream region of the human GCH1 gene, three different single base pair (bp) substitutions have been identified at positions -22C>T [16,18], -39C>T and -132T>C [17], relative to the GCH1 translation start codon (ATG), in DRD individuals who lack mutations in the coding region ( Figure 1). As the GCH1 transcription start site has been identified, we have now demarcated the substitutions relative to the GCH1 transcription start site (+1) to be in accordance with revised genetic nomenclature. Thus, the -22C>T substitution is referred to as +142C>T, the -39C>T as +125C>T and the -132T>C as +32T>C. The +125C>T and +32T>C substitutions were identified as co-existing in cis in a single subject with DRD [17], making it impossible to independently determine hereditary disease association for each substitution. The +142C>T substitution has been identified in two different reports, one of a single individual with DRD [16] and another in multiple generations of a family with DRD, in whom the +142C>T substitution segregated with affected status [18]. Whether any of these single nucleotide substitutions have an impact on GCH1 expression levels has not been investigated.
In the present paper, we demonstrated that among the three aforementioned 5'UTR GCH1 substitutions, only the +142C>T is a functional substitution that decreases protein expression. However, neither mRNA level nor stability were affected by this substitution. The 5'UTR mediated regulation of translation may reflect alterations in secondary RNA structure, changes in RNA binding motifs (RBPs) and associated proteins, or the creation of upstream open reading frames (uORFs) [22][23][24][25]. The +142C>T substitution within the GCH1 5'UTR introduces an upstream AUG (uAUG) that potentially creates an out-of-frame uORF. This uORF partially overlaps with the physiological ORF (pORF) and likely acts in a competitive manner to down regulate translation. To our knowledge, this is the first report of a translational disruption causing DRD or any other type of dystonia.

Cell cultures
The human neuroblastoma cell line SK-N-BE(2)-M17 was purchased from ATCC (Manassas, VA) and maintained in MEM/F-HAM, supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin, at 37°C and 5% CO 2 . Lymphocytes isolated from DRD patients and healthy individuals were Epstein Barr Virus (EBV) -transformed to stable cell lines, as previously described [26] and maintained in RPMI 1640 with 5% fetal calf serum, 1% penicillin/streptomycin and 2 mM glutamine at 37°C and 5% CO 2 .

Patients and controls
Lymphoblastoid cultures were established from 10 family members, spanning three generations, in whom the +142C>T substitution segregated with affected status [18]. Six subjects Figure 1. Alignment of the human, mouse and rat 5′ upstream GCH1 regions. Nucleotides are numbered from the transcription start site (+1) as previously identified for the human (top), rat (middle) and mouse (bottom) GCH1 genes. Position of each GCH1 5'UTR substitution identified to date is shaded and the nucleotide substitution is shown beneath the sequence. An asterisk (*) indicates sequence identity across the three species. were heterozygous for the +142 SNP (C/T) and met the clinical criteria for DRD [7], while the remaining four family members were homozygous for the +142C nucleotide (C/C) and had no evidence of dystonia or other basal ganglia dysfunction on examination. Two additional control samples (+142C/C) were utilized from the discarded sample collection of the Massachusetts General Hospital Neurogenetics DNA Diagnostic Laboratory. Full GCH1 gene sequencing (exons 1-6), including intron/exon junctions was performed for all samples and no other nucleotide changes were found [18]. The local institutional review board approved the study. All participants gave informed consent.

Reporter constructs and mutagenesis
Human GCH1 luciferase reporter constructs have been previously described [9]. We utilized a 615 bp and a 5399 bp 5' upstream GCH1 fragment linked to the firefly luciferase (Fluc) cDNA in the promoter-less basic Picca Gene Basic Vector 2 (PGV-B2) (Toyo Ink Co., Ltd, Tokyo) [27] referred to as 615GCH1/WT (wild type) and 5399GCH1/WT, respectively ( Figure 2A & 2C). Mutant reporter constructs were generated from the parental 615GCH1/WT by replacing the +142C with a T (615GCH1/+142T), the +125C with a T (615GCH1/+125T), the +32T with a C (615GCH1/+32C), the +125C with a T and the +32T with a C in the same construct (615GCH1/+125T/ +32C) (Figure 2A). A mutant reporter construct was generated from the parental 5399GCH1/WT by replacing the +142C with a T (5399GCH1/+142T) ( Figure 2C). All mutant products were generated using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) and appropriate sets of primers (sequences available upon request). All constructs were confirmed by DNA sequencing.

Transfections and luciferase assays
Transient transfections were performed using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) in 1.0 to 1.5 x 10 6 cells/well plated onto 24-well tissue culture plates. For each Fluc reporter construct used in transfections, two independent plasmid Maxipreps (Qiagen, Valencia, CA) were used. Each transfection consisted of 0.8 μg DNA of Fluc reporter construct and 0.4 μg CMV-Renilla luciferase (Rluc) vector bearing Rluc under the cytomegalovirus (CMV) promoter, as an internal control of transfection efficiency (Promega, Madison, WI). Luciferase activity was analyzed 40 hrs post-transfection, using a dual luciferase reporter kit (Promega) and a luminometer (Dynex, Richfield, NM). Luciferase assays were repeated at least four times, and within each experiment each construct was transfected in quadruplicate. Results were expressed as a mean of quadruplicate measures (+ S.E.M.) of Fluc normalized to Rluc activity. Statistical significance was determined by Student's t-test, comparing the activity of each mutant construct with the wild type parental construct.

Quantitative real-time PCR
RNA was isolated from SK-N-BE(2)-M17 cells transfected with the 615GCH1/WT or the 615GCH1/+142T construct or from 5.0 x 10 6 human lymphoblastoid cells using the miRNA purification kit (Qiagen) and then subjected to DNase treatment (Qiagen). Two hundred ng of RNA was then reverse transcribed using Omniscript (Qiagen) and qPCR reactions were performed in a 25 µl reaction mixture using Power SYBR® Green PCR Master Mix (Applied Biosystems, Foster City, CA) and 160 nM of each primer. Amplification conditions consisted of: (1) 1 cycle of 50°C, 2 min; (2) 1 cycle of 95°C, 10 min; (3)  post-transfection, SK-N-BE(2)-M17 cells were supplemented with ActinomycinD (5µM/ml, Sigma-Aldrich, St. Louis, MO) and RNA was extracted at 0, 2, 4 and 6hrs while lymphoblastoid cultures were treated with ActinomycinD (5µM/ml) for 0, 6, 12 and 24hrs. Normalized Ct values were converted to relative mRNA levels, set to 100% for t=0. Decay kinetics were obtained using linear regression analysis and the half life for each examined mRNA was calculated based on the slope and the intercept. The mRNA decay assays were repeated 3 times for each case, results were expressed as the mean ± S.E.M. and statistical significance was determined using Student's t-test.

Results
Alignment of the 5' upstream sequence of the human, mouse and rat GCH1 genes revealed that among the three point substitutions examined only the +142C is evolutionarily conserved between the human, rat and mouse genome ( Figure  1). All of the three 5' upstream GCH1 substitutions fall within its 5'UTR (+142C>T; +125C>T; +32T>C; Figure 1) rendering it more likely that any potential effects would involve posttranscriptional mechanisms.

Effects of the GCH1 5'UTR substitution on the luciferase reporter assay
We utilized a previously characterized GCH1 reporter construct [27] encompassing 452 bps upstream of the transcription start site (+1) and 163 bps corresponding to the distance between the +1 and the translation start site ATG (615GCH1/WT; Figure 2A). Three mutant reporter constructs were generated to mimic the 5'UTR GCH1 reported substitutions; 615GCH1/+142T, 615GCH1/+125T and 615GCH1/+32C (Figure 2A). Transient transfection assays were performed in a human neuroblastoma SK-N-BE(2)-M17 cell line, a monoamine (MA) and BH4 containing cell line in which the GCH1 promoter has been reported to be active [28]. Normalized Fluc activity for each of the mutant constructs compared to the wild type (set at 100%) revealed that the 615GCH1/+142T construct caused a reduction in luciferase activity to 32% of the 615GCH1/WT activity (n=4, P ≤ 0.005;  Figure 2B). Likewise, when the +142T substitution was introduced in the larger 5399 bp reporter construct (5399GCH1/WT and 5399GCH1/+142T, Figure 2C) luciferase activity dropped to 37% of the wild type (n=4, P ≤ 0.005; Figure  2D). In contrast, the 615GCH1/+125T, the 615GCH1/+32C and the double mutant 615GCH1/+125T/+32C constructs did not result in a statistically significant reduction in luciferase activity (94%, 89% and 90% of wild type activity, respectively, n=4; Figure 2B). Finally, after generation of the +142T reporter construct, we re-introduced the +142C in the 5399GCH1/+142T or the 615GCH1/+142T constructs and luciferase activity returned to wild type levels (data not shown).

Effects of the +142C>T GCH1 substitution on luciferase mRNA
We then analyzed the effects of the +142C>T substitution on luciferase mRNA levels expressed in SK-N-BE(2)-M17 cells transiently transfected with luciferase reporter constructs. QRT-PCRs revealed no statistically significant differences between the relative mRNA levels of the 615GCH1/WT (set to 100%) and the 615GCH1/+142T (113%) reporter constructs (n=4; Figure 3A). To determine if the +142C>T substitution affected mRNA stability, inhibition of transcription was achieved with ActinomycinD and luciferase mRNA half-lives were determined to 4.1 hrs for the 615GCH1/WT and 4.3 hrs for the 615GCH1/+142T transcript. This difference in half-lives was not statistically significant (n=3, Figure 3B).

Endogenous GCH1 levels in normal individuals and DRD patients
We further investigated the effects of the +142C>T substitution on in vivo GCH1 mRNA levels by qRT-PCR analysis in lymphoblastoid cultures established from DRD patients and control individuals. While the average relative steady-state levels of GCH1 mRNA were lower in DRD patients (0.842) compared to controls (set at 1.0), this reduction was not statistically significant (n=4; Figure 4A). Furthermore, when transcription was inhibited by actinomycinD, there was no statistically significant difference in the GCH1 transcript half-life between DRD and control subjects (11.4 hrs in controls vs 11.8 hrs in patients), (n=3; Figure 4B).

Bioinformatics analysis
RNA secondary structures analysis, performed for the complete 5'UTR GCH1 region with either a C or a T at +142 predicted no structural differences between them ( Figure 5). Since our experimental data did not reveal evidence for transcriptional effects of the +142C>T substitution, we scanned the wild type GCH1 +142C and the GCH1 +142T 5'UTRs for RBP motifs and found no differences (data not shown). However, we observed that the +142C>T substitution created a novel upstream ATG codon (ACG>ATG) (Figure 6). Indeed, scanning the 5'UTR-coding GCH1 sequence with the NCBI ORF Finder software revealed that in presence of a T at +142, an uATG arises and a potential upstream open reading frame (uORF) (sequence in bold; Figure 6), starting 23 nucleotides upstream of the physiological ATG (underlined; Figure 6) and terminating after 219 bases at a STOP codon within an exon of the wild type GCH1 sequence (TGA circled; Figure 6). Thus, the +142C>T substitution is likely to result in production of a 73 amino acid peptide (MAAAAAGPWRRALCGHRRRSRGAPGAAMGSPSGIRRGPG PAGRRRSPRGPRPRARSPRTAGRASGPAARRITS) which is out of frame with the pATG codon initiating translation of the physiological GCH1 protein ( Figure 6). Note that for the 5'UTR GCH1-luciferase constructs used in this study, sequence scanning by the NCBI ORF Finder software showed that the newly generated uORF would encode an out of frame of 14aa truncated peptide (analysis non shown).

Discussion
In eukaryotes, regulation of protein synthesis is a finelytuned process involving multiple steps of transcription, posttranscriptional processing and modification, mRNA stability, translation initiation and posttranslational modification. Human disease can result from single nucleotide polymorphisms that affect any step of this finely tuned regulatory system. In particular for diseases linked to noncoding substitutions, functional assays are required to establish a link between the substitution and the disease phenotype. In DRD, three different sequence substitutions have been identified within the 5' untranslated region of the GCH1 gene [16][17][18]. We demonstrated here that the +125C>T and the +32T>C substitutions reported together in a single DRD patient had negligible effects on luciferase activity when tested either individually or combined into a single reporter construct. The presence of these substitutions has been examined in only 3 neurologically normal subjects, who did not harbor either or both substitutions [17]. In contrast, the +142C>T substitution which segregates with affected status in ten members of a single DRD family, spanning three generations but not found in a total of 543 control alleles [18], had a strong negative effect, leading to a 68% reduction in luciferase activity. The effect persisted even when the +142T substitution was introduced into the larger 5399GCH1 reporter construct having a greater degree of endogenous transcriptional regulation. To ensure that no random mutations within the luciferase vector backbone affected our results, we re-introduced the wild type +142C back onto the mutant +142T construct and luciferase activity returned to wild type levels. To our knowledge, no gene dosages studies of the +142C>T substitution carriers has been conducted.
We next investigated if the +142T-triggered reduction in luciferase activity was mediated by a reduction in luciferase mRNA levels. We found that the +142T had no significant effect on luciferase mRNA abundance. Furthermore, RNA decay kinetics demonstrated a similar mRNA half-life for both the +142C and the +142T 5'UTR luciferase reporter constructs. Similarly, there was no statistically significant difference in endogenous steady-state GCH1 mRNA levels or its stability in lymphoblastoid cells derived from DRD and control subjects. Lymphoblastoid cultures from DRD and control subjects were exposed to ActinomycinD, RNA was collected at 0, 6, 12 and 24 hrs and subjected to qRT-PCR. Normalized GCH1 Ct values were converted to relative mRNA levels and results are shown as means ± S.E.M. plotted to a linear regression over time, with time zero set at 100%. Half-lives were calculated based on the slope and intercept; t ♦ 1/2 =11.4hrs for control and t • 1/2 = 11.8hrs for DRD subjects (n=3). Cerebrospinal fluid from these patients was not available to assess the BH 4 level but previous studies support that altered GCH1 mRNA levels can be detected in mononuclear blood cells from DRD patients with coding GCH1 mutations [29]. In agreement with our experimental data, bioinformatics analysis predicted the same RNA structure in the presence of a +142C or a +142T within the GCH1 5'UTR. Taken together these results support that the +142C>T GCH1 5'UTR substitution alters efficiency of translation without affecting RNA levels or RNA stability/structure. Scanning of the wild type +142C and the +142T GCH1 5Ú TR sequences predicted no differences in RBP sites. Thus, it is unlikely that differences between RBP association with the wild type and the 142T 5'UTR sequences are the cause of the observed changes in translation. An alternate level of translational regulation involves the creation of uAUGs and uORFs [25]. Genes can be devoid of uAUGs or bear a single or several uAUGs which mediate complete or partial inhibition of the physiological AUG (pAUG) [22,25,30]. Our in silico analysis revealed that the wild type 5'UTR GCH1 region lacks of any uAUG. However, the +142T substitution introduces a single uAUG and a potential uORF encoding a truncated, out of frame, peptide. The +142T substitution results in reduced luciferase activity indicating that both the truncated and full length luciferase protein are generated, with the truncated peptide competing for efficiency of translation against the full length peptide.
These results do not provide direct evidence of the effects of the +142T substitution on the GCH1 levels in DRD patients. However, similar to the luciferase constructs, the +142T may generate an uORF that encodes a 73 aa out of frame peptide that competes with the translation of wild type GCH1 initiated by the pAUG. A model in which a mixture of wild type and truncated peptides are generated in a competitive manner due to creation of novel uAUGs has been reported in human diseases ranging from a predisposition to cancer [31] to bronchiectasis [32] and bipolar depression [33]. Our study expands this list to include dystonia by revealing that the +142C>T GCH1 5'UTR substitution detected in DRD patients is a functional mutation that reduces translational efficiency and likely leads to reduced wild type GCH1 protein levels, underlying manifestation of DRD.