A Haplotype of Two Novel Polymorphisms in δ-Sarcoglycan Gene Increases Risk of Dilated Cardiomyopathy in Mongoloid Population

The role of genetic abnormality of δ-sarcoglycan (δ-SG) gene in dilated (DCM) and hypertrophied (HCM) cardiomyopathy patients is still unfolding. In this study we first defined the promoter region and then searched for polymorphisms/mutations among the promoter, 5'-untranslated region, and the encoding exons in δ-SG gene in 104 Chinese patients with DCM, 145 with HCM, and 790 normal controls. Two novel polymorphisms were found, an 11 base-pair (bp) deletion (c.-100~-110; -) in the promoter region and a missense polymorphism of A848G resulting in p.Q283R in the highly conserved C-terminus. The prevalence of homozygous genotype -/- of c.-100~-110 was slightly higher in DCM (14.42%) and HCM patients (14.48%), as compared with normal controls (11.01%). The prevalence of genotype of 848A/G was significantly higher in DCM (6.73%; OR = 9.43; p = 0.0002), but not in HCM patients (1.38%; OR = 1.37; p = 0.62), as compared with controls (0.76%). Haplotype -_G consisting c.-100~-110 and A848G was associated with increased risk of DCM (OR = 17.27; 95%CI = 3.19–93.56; p = 0.001) but not associated with HCM (OR = 1.90; 95%CI = 0.38–9.55; p = 0.44). Co-occurrence of the genotypes -/- of c.-100~-110 and 848A/G was found in 5 patients with DCM (4.81%; OR = 39.85; p = 0.0001), none of HCM patients, and only 1 of the controls (0.13%). Both polymorphisms were also found in the Japanese population, but not in the Africans and Caucasians. C.-100~-110 resulted in a decrease of δ-SG promoter activity to 64±3% of the control level (p<0.01). Both co-immunoprecipitation and in vitro protein pull-down assays demonstrated that δ-SG-283R interacts normally to β- and γ-SG, but significantly decreased localization of β/δ/γ-SG on the plasma membrane. In conclusion, haplotype -_G composed of c.-100~-110 and A848G confers higher susceptibility to DCM in the Mongoloid population.

The etiology genes of inherited DCM are still far from clarified. The thirty-three causative genes reported could explain only 30~35% of the total disease [1,3,7]. Titlin and Lamin A/C gene mutations have been shown to be the most common genetic abnormality [7][8][9]. About half of the genes encode sarcomeric proteins including α-cardiac actin, myosin heavy chain, troponin, tropomyosin, metavinculin, α-actinin and therefore overlap with those in HCM [1,3,7,10]. DCM specific genes are those encoding cytoskeletal proteins, including dystrophin, sarcoglycan, metavinculin, desmin, Cypher/ZASP, α-Bcrystalin, and LIM domain protein-3 [7,10]. Autosomal recessive, mithochondrial, and X-linked DCM have been described, but familial DCM is mainly transmitted as an autosomal dominant disease. However, clinically non-isolated forms of DCM represent only 10% of all familial DCM and majority of DCM patients carry sporadic and autosomal dominant mutations [10]. Given the fact that morbid mutations identified represent only a small percentage of familial DCM, it is reasonable to speculate that a large number of morbid genes remain to be discovered.
The role of δ-SG gene mutation in the pathogenesis of CM was first discovered in a Syrian hamster model of inherited CM and very mild muscular dystrophy [17]. In 1997, Sakamoto et al reported that a 30 kb deletion located in 5' upstream of the exon-2 of the gene was the direct etiology for the model [17]. Later in humans, Tsubata et al screened 50 young DCM patients (age 3 days-18 years) and one DCM family and detected two mutations in the δ-SG gene [18]. The p.S151A mutation was detected in three family members of one family and a 3bp deletion at position 238 (c.K238) in two sporadic cases. The p.S151A and c.K238 mutations caused a relatively severe form of DCM characterized by sudden cardiac death and heart failure at young age, but with no signs of LGMD-2F. Both mutations were autosomal dominant and the p.S151A was later proven to cause DCM in transgenic mice [19] and Drosophila [20], but did not cause DCM when it co-occurred with A131P, a LGMD-2F-causing mutation reported in German population [21]. In 2003, Karkkainen, et al reported the third causative p.R71T that also autosomal dominantly resulted in mild form of DCM without affecting the skeletal muscle in Finland population [22]. However, Sylvius, et al reported that δ-SG gene was only marginally implicated in French patients with DCM [23] and so did Date et al report in Japanese patients with HCM [24]. Since 2007, the mutation-searching studies have shifted to the 5'untranslated region (5'UTR) of the gene. A G-94C polymorphism in the 5'UTR has been reported to be associated with the HCM in Japanese population [25]. The -94C allele was recently identified to be strongly associated with genetic susceptibility for HCM in Amerindians in Mexico as well [26,27].
The inference that δ-SG could be an isolated etiology for CM is based only on the scarcity of above evidence and has been controversial and inconclusive [21]. On the other hand, there is no data on δ-SG mutation/polymorphism in Chinese CM patients. It is still unclear whether there is any CM-causing abnormality in the promoter region, as those reported in Syrian hamster model. Till now, no defined δ-SG promoter has been reported.
In this study, we first determined the promoter region of δ-SG gene and subsequently screened for single nucleotide polymorphism (SNP)/mutations in its promoter, the 5'UTR, and the coding regions in Chinese patients with DCM and HCM. Here we report two novel polymorphisms, one with an 11bp deletion (c. -100~-110 ) in the promoter region and the other with a heterozygous missense polymorphism of A848G resulting in p.Q283R in exon-8. The prevalence of 848A/G genotype was 8.86 times higher in DCM and 1.81 times higher in HCM patients. Five of six cases with co-occurrence of the homozygous c. -100~-110 and p.Q283R are found in DCM patients. Both polymorphisms are found only in the Mongoloid population. δ-SG-283R mildly but significantly decreased localization of β/γ/δ-SG on the plasma membrane. Our results suggested that the co-occurrence of homozygous c. -100~-110 and p.Q283R contribute to the pathogenesis of DCM in eastern Asia countries.

Methods
Additional details are available in S1 File: Expanded Methods.

Subjects and DNA Samples
The study was approved by the institutional review board of Xinhua Hospital Affiliated to Shanghai Jiaotong University. Written informed consents for participation were obtained from all patients and healthy control subjects.
The patient pool consisted of 104 patients with DCM and 145 with HCM. None of the patients had symptoms of LGMD. The control group included 790 unrelated healthy Chinese. The clinical and echocardiography data are summarized in Table 1. Genomic DNA was extracted from whole peripheral blood. 459 cases of normal Japanese DNA samples were also

Amplification of Genomic Fragments, Allele and Genotype Analysis
The genomic structure of the human δ-SG gene is established in Fig 1A. Primers for amplification of the genomic fragments (F) and exons are listed in S2 and S3 Files. PCR products were subjected for sequencing or endonuclease digestion.

Transient Expression and Firefly Luciferase Reporter Assay
Luciferase activities were measured in L6 cells with a dual-luciferase reporter assay system (Promega).

Nuclear Protein Extraction (NPE)
Nuclear protein was extracted from the left ventricular myocardial tissue of golden hamster (S6 File).

Cloning of DGC Components' cDNA
The cDNAs representing either full length or extracellular domain of genes in DGC complex were amplified from a human heart cDNA library (Clontech, CA), and cloned into pcDNA3.1-myc/his(+)A or pGEX-4T-1 (Amersham Biosciences). The mutant of δ-SG-283R mutant cDNA was obtained by site-directed mutagenesis. Human Hand1, Hand2, and MEF2C's cDNA were cloned into pQE-30.

Immnuoprecipitation
Each protein in DGC complex was over-expressed in HEK-273T cells, harvested, and immunoprecipitated with their cognant antibodies.

In vitro Transcription-translation
In vitro transcription and translation was performed using the TNT kit (Promega, WI) as directed by the manufacturer.

GST-and 6His-fusion Protein Expression and Purification
All GST-and 6His-fusion proteins were expressed in E. Coli and purified.

Cell Surface Biotinylation
HEK-293T cells were washed with PBS and treated with 1 mM membrane-impermeable NHS-LC-biotin in 50 mM HEPES (pH 7.4). After incubation at room temperature for 30 min, the reaction was quenched in 100 mM Tris-HCl (pH 7.4). The biotin-labeled cells were washed with PBS and total proteins were extracted with TLB buffer. The biotin-labeled proteins were isolated by incubating cell lysate with avidin-agarose. The flow-through fraction was collected and the avidin-agarose was then washed with PBS with 0.1% Tween-20 and the biotin-labeled proteins were eluted from the avidin-agarose with SDS sample buffer.

Western Blotting
The proteins were separated by SDS-PAGE and electrophoretically transferred to Immobilon-P (Millipore, MA). Immunodetection was accomplished using the ECL Western blotting (WB) kit (Thermo-Pierce).

Statistic Analysis
Genotype frequencies in cases and controls were compared using a χ 2 test with Yates' correction. Odds ratio (OR) and 95% confidence interval (CI) were calculated by unconditional logistic regression analysis. Linkage disequilibrium (LD) between c. -100~-110 and A848G was estimated in the control group using Lewontin's standardized coefficient D' and LD coefficient r 2 . The haplotypes were inferred from all genotype data, and haplotype analyses were conducted using SNPStats (http://bioinfo.iconcologia.net/snpstats/start.htm) [28]. All the statistical analyses were performed with the SPSS15.0 software, with two-sided tests and a significant level of p value <0.05.

Localization of the Promoter Region of the Human δ-SG Gene and Identification of the Active E-boxes
In order to define the promoter region, we first conducted a luciferase assay with luciferase reporter constructs carrying different deletional fragments of 5'UTR of d-SG gene (F-1~10) in L6 cells. As shown in Fig 1B, the luciferase activities demonstrated a 37.2~38.2-fold increase for F-1, -2 and -3, but only a 5.8~8.8-fold increase for F-4 and -5. These results suggest that the promoter is likely located in the region between the F-4 and -1, an 814bp fragment including 476bp 5'UTR1 and its upstream 338bp. To further confirm the promoter region and rule out any possible effects of 2301bp interon-1 and 46bp 5'UTR-2 (exon-1) on the activity, we performed luciferase assay with F-6~10. As shown in Fig 1C, the reporter activities were increased 61.6~64.5-fold for F-6, -7, and -8, indicating that the interon-1 and 5'UTR-2 inhibit the reporter activity. We observed a 48-fold increase in the activity for deletional mutant F-9. However, further deletion (F-10) showed only an 11.1-fold increase, indicating that the high promoter activity came from the region between F-9 and -10. These results are consistent with those in Fig 1B. Taken together, we identified that F-8 (-338 to +476bp) represents the whole promoter region of human δ-SG gene. Sequence analysis revealed that the above identified δ-SG promoter contains 3 putative MEF2C motifs ( Fig 1D) with MEF2C-1, -2 highly conserved among species. However, as shown in Fig 1C, MEF2C-1 and -2 would only account for about 20% promoter activity by calculating the difference between F-8 and -9.
Further analysis identified 7 putative E-box ( Fig 1D), with E-2, E-3, and E-6 are highly conserved. Based on the results shown in Fig 1, F-9 gave the highest luciferase activity and contains all 7 putative E-boxes without MEF2C-1 and -2 sites. We therefore chose F-9 construct for further mutagenesis analysis. To determine the key important binding motif, we performed serial mutations to delete those motifs. As shown in Fig 1E, we found that the E-2, E-3, and E-6 played a functional role in regulating δ-SG gene expression.
C. -100~-110 Located at Immediate Downstream of a Putative E-box Next, we searched for SNP/mutations in the 814bp promoter region in CM patients by direct sequencing. In some samples, compared to the WT (Fig 2A), overlapped spikes always started from a certain area in both forward ( Fig 2B) and reverse (Fig 2C) sequencing directions. These overlapping spikes disappeared when we sequenced a sample from a patient carrying an 11bpdeletion in both alleles. This indicated that the overlapped sequencing spikes was due to a het-  (Fig 2E), which was used as an easy identification in mass genotyping ( Fig  2F).
As shown in Table 2 Our results from both 790 normal Chinese and 469 normal Japanese are consistent with the data of the Asian population (286 including 97 Han Chinese in Beijing, 100 Southern Han Chinese, and 89 Japanese in Tokyo) in the "1000 Genomes Project" (Table 3). However, our data using both 110 Caucasians and 110 West Africans normal controls are slightly different from those in the "1000 Genomes Project", which illustrates 2.0% heterozygous c. -100~-110 in 246 Africans and 1.7% heterozygous c. -100~-110 in 379 Europeans.
Within the promoter region of δ-SG, several other loci of SNPs (A108T; C148G; A217G; G426C) were found in both CM patients and normal Chinese controls. However, those SNPs frequency showed no significant difference among HCM, DCM, and controls (data not shown).
A Novel Missense Polymorphism of A848G in the Highly Conserved Exon-8 Next, we searched for SNP/mutations in all 8 exons in CM patients by direct sequencing. No previously-reported mutations or SNPs were found in exon-1~7 [18,[21][22][23][24][25]. In exon-8, a novel nucleotide substitution of A848G (CAG!CGG), resulting in conversion of p.Q283R (Fig 3A), was discovered in the universally conserved extreme C-terminus ( Fig 3B). This polymorphism creates a new restriction enzyme site for Bsl I (Fig 3C), which we used for genotyping ( Fig 3D).     This polymorphism is not on the list of the "1000 Genome Project", indicating it is a novel mutation or a rare SNP. The genotype for both patients and controls were heterozygous. The genotype frequency of 848A/G was significant higher in DCM (7/104, 6.73%; OR = 9.43; p = 0.0002), as compared with HCM patients (2/145, 1.38%; OR = 1.37; p = 0.62) and controls (6/790; 0.76%; Table 4). This 848G allele was also found in Koreans, Japanese, and Thai population, but not in the Africans and Caucasians (Tables 2 and 3).

Association between C. -100~-110 and A848G
Obtained from LD analysis, D' is 0.1955 and r 2 is 0.00027 between c. -100~-110 and A848G in the control group, indicating low LD between these two loci.
In DCM group, of 7 cases carrying 848A/G, 5 cases were concomitant with homozygous -/of c. -100~-110 and 1 case each with heterozygous -/TGTAATAAGCT and WT, respectively ( Table 2). In HCM group, of 2 cases carrying 848A/G, 1 case each was in coincidence with heterozygous -/TGTAATAAGCT and WT, respectively. In the normal group, of 6 cases carrying 848A/G, 1 case was in coincidence with homozygous -/-of c. -100~-110 and 3 and 2 cases were respectively with heterozygous -/TGTAATAAGCT and WT. Similarly, of 4 cases carrying 848A/G in healthy Japanese, none was concomitant with homozygous -/-of c. -100~-110 and 2 cases each were with heterozygous -/TGTAATAAGCT and WT, respectively. Compared to the prevalence in normal Chinese group (0.13%), cases with co-occurred -/-of c. -100~-110 and 848A/G have a significantly increased prevalence (4.81%) and a remarkably high specificity (OR = 39.85; p = 0.0001) in DCM patients.

Decrease in the Promoter Activity Caused by C. -100~-110
We then studied the impact of c. -100~-110 on the promoter activity. The luciferase activities were compared between WT, F-1, -6, -8, and -9 and matching mutants containing homozygous c. -100~-110 . As shown in Fig 3A, the c. -100~-110 resulted in a significant decrease in the promoter activity to 64±3% of the control levels (p<0.01).
Next, we explored the impact of c. -100~-110 on any potential binding of transcriptional factors to the region. Since c. -100~-110 abolished the putative 3 rd E-box motif, we tested the impact on HAND1 and HAND2 binding. As shown in Fig 4, neither HAND1 nor HAND2 bound to the WT sequence or to the probe containing the c. -100~-110 . Furthermore, since c. -100~-110 is an

Normal Interaction of δ-SG-283R to DGC Components
Since the WT residue 283Q shares homologous with βand γ-SG (Fig 5C), and conversion of Q!R results in a dramatic change in electric charge, we thus tested its influence on its protein interactions with other five DCG components. We co-transfected 293T cells with full-length cDNA encoding each of the five DGC components in expression plasmids, and WT γ-SG or γ-SG-848A. Immunoprecipitation assays were performed using transfected whole cell lysate. The results showed that both WT δ-SG and δ-SG-848A strongly bind to β-SG and γ-SG with similar affinity, weakly to α-SG with similar affinity, and neither binds to α-DG nor β-DG (Fig 5A). We further examined the effect of A848G on the direct association with the DGC components by pull-down assay. We expressed and purified both GST-δ-SG-CT-283Q (WT) and GST-δ-SG-CT-283R (SNP) from E. Coli (Fig 5A-a). The extracellular fragments of the indicated five components' were obtained by in vitro translation (Fig 5A-b). In vitro protein pulldown assay demonstrated that both GST-δ-SG-CT-283Q and GST-δ-SG-CT-283R equally bind to α-, β-, γ-SG, but not to αand β-DG (Fig 5A-c and 5A-d), similar to what we observed in immunoprecipitation assays.  Association of δ-SG-283R with Mild Decrease in Plasma Membrane Localization of β/γ/δ-SG It has been reported that β-SG and δ-SG were shown to form a core structure capable of targeting to the plasma membrane [29]. Since δ-SG-283R is located between 282C and 285N, whether it can change proper plasma membrane targeting? Above immunoprecipitation and pull-down results demonstrated that δ-SG interacts strongly with βand γ-SG. Thus, we cotransfected wild-type β-SG and γ-SG with either WT δ-SG-283Q or δ-SG-283R in HEK-293T cells and performed cell surface biotinylation assay. We found that δ-SG-Q283R caused a mild but significant reduction of β/γ/δ-SG on the cell surface, as compared with WT δ-SG-283Q (Fig 6). These results indicate that δ-SG-283R has a mild detrimental effect on membrane localization of DGC.

Discussion
DCM is a genetically heterogeneous disease and linkage analysis may be of limited value mainly because of the high mortality associated with the disease and the incomplete penetrance of disease mutations. So far almost all reported mutations for DCM have been found only in single families, and only few DCM-associated mutations have been detected in more than one patient [1,3,[7][8][9][10]. Therefore, one of the key strategies to identify morbid genes is to screen for mutations in candidate genes in cohorts of sporadic DCM patients. Based on previous reports, defects in the δ-SG gene seem to be unique and highly heterogeneous among DGC genes: 1) Six different LGMD-2F-causing mutations (E93X, A131P,  [21,[30][31][32][33]. These mutations can be genetically transmitted in a mode of autosomal recessive. 2) Three different DCM-causing mutations (p.S151A, c.K238, and p.R71T) were not associated with the involvement of the skeletal muscle [18,22]. Those mutations can be transmitted in a mode of autosomal dominant. 3) Deletion or mutation in the gene's promoter can cause both DCM and HCM in both hamster and human being [17,26,34]. Those warrant our screening for genetic abnormalities in δ-SG gene in Chinese CM patients.  In the present study, we identified for the first time c. -100~-110 and A848G polymorphisms in δ-SG gene. The co-occurred -/-of c. -100~-110 and 848A/G is corrected with a significantly increased susceptibility of DCM. This is the fourth DCM-associated polymorphism/mutation reported in the δ-SG gene. Based on the fact that the responsible gene mutations reported so far could only account for about 30~35% of the DCM, our results further indicate that it is less likely that the remaining two-thirds mutations are all from new genes, but it is highly possible that a large part of new mutations will be from novel loci of those already-reported genes.
The genotype frequency of homozygous c. -100~-110 was slightly higher in patients with CM. But due to a high prevalence of the allele in the normal controls, we postulated that it is a benign polymorphism that may have limited genetic significance. We tried two approaches to determine the functional relevance of the c. -100~-110 , luciferase reporter assay and EMSA. As shown by luciferase results, homozygous c. -100~-110 unexpectedly caused a significant decrease in promoter activity. However, our EMSA results failed in demonstrating any positive binding of transcriptional factors to the region. So far, transcriptional regulations of the DGC complex genes remain unclear. HAND-1 and HAND-2 are important transcription factors (TFs) for cardiac muscle [35]. They share the same E-box motif (CANNTG) as MyoD, a key TF for skeletal muscle differentiation [36]. Among the DGC genes, αand γ-SG are activated by MyoD [33,37,38]. The TFs that are responsible for controlling δ-SG gene transcription remain unknown, and possible candidates are HAND-1, HAND-2, MEF2C, Nkx2.5, NF-AT3, and CREB [39]. In our identified promoter region of the human δ-SG gene, we found two highly conserved putative binding motifs for MEF2C, one for Nkx2.5, and seven E-boxes. Based on our EMSA results, there was no evidence supporting that HAND-1 and HAND-2 are the TFs regulating human δ-SG gene transcription, nor the possibility that the sequence of c. -100~-110 itself acting as a MEF2C binding site. Our mutagenesis results suggest that at least the 2nd Ebox motif, which is located adjacent to c. -100~-110 , is one of the functional E-boxes. That c. -100~-110 changes the downstream nucleotides of the 2nd E-box motif may contribute to the decreased promoter activity (Fig 3).
Despite a significant decrease in δ-SG promoter activity in the -/-of c. -100~-110 polymorphism, there is no significant association of this polymorphism in patients with CM. On the contrary, the reported polymorphism -94G>C is associated with increased susceptibility of HCM in both the Japanese and Mexican population [25][26][27], the molecular mechanism remains to be unclear. It is possible that even with a mild to moderate decrease in δ-SG gene transcription, δ-SG protein translation could be compensated. Since c. -100~-110 is not located in the 5'-UTR of the gene (Fig 4), it should not detrimentally affect post-transcriptional regulation of δ-SG mRNA stability and translation of the δ-SG protein [40]. The compensation need to be verified by measuring δ-SG gene transcription and translation in myocardial biopsy samples from subjects with -/-of c. -100~-110 .
On the other hand, the 848G allele seems to be sporadic in the Chinese and eastern Asia population. The genotype frequency of 848A/G was significantly higher in our patients with DCM (6.73%) as compared to HCM (1.38%) and normal controls (0.76%), suggesting the A848G is a rare polymorphism related to significantly increased susceptibility to DCM. Importantly, our result showed that when both homozygous c. -100~-110 and heterozygous p.Q283R occurred coincidently in a single individual, the risk for DCM further increased to a surprisingly high level (OR = 39.85), indicating the pathogenic effects from both polymorphisms could be accumulative and amplified. Our results strongly suggest that the combination is one of the etiologies for human DCM in eastern Asia and the prevalence is 4.81% in our DCM patients, much higher than the previous reports where the mutation rate of δ-SG gene is less than 1.5% in both Europe and Japan [18,[21][22][23][24].
Furthermore, in this study all cases with the polymorphism are heterozygous. What cardiac phenotype will turn out to be in individuals who carry an 848G/G genotype? That may shed light on the role of the polymorphism in the pathogenesis of DCM. Because our patient pool is small, the prevalence of 848G/G is still unclear in DCM patient, but it could be much higher than it in the normal population. Based on our data, it is about 1.4/100,000 in the normal Chinese population. If individuals with 848G/G manifest a DCM phenotype, this number could account for about 7.4% of the etiology of the DCM, since the prevalence of DCM is 19/100,000 in China [41]. If added to the above 4.81% prevalence of the co-occurrence in DCM patients, these could account for >10% of the etiology of DCM, exceeding previously reported Lamin A/C gene mutations [7]. We are planning to expand our patient pool to sufficiently cover some cases of 848G/G to determine the cardiac phenotype.
Normal protein interactions among each DGC components are prerequisite for maintaining DGC normal architecture and function [29,42,43]. δ-SG is in close proximity to βand γ-SG, especially their extracellular domains which consist of the same number of 231 amino acid (AA) residues [42]. The assembly of the complex follows a specific pathway: βand δ-SG first form a core and then γ-SG acts as a linker to connect α-SG to the β/δ-SG core [29,42,43]. Those DGC proteins are divided into four functional domains, including intracellular, transmembrane, extracellular proximal, and extracellular extreme C-terminus [43]. Their extracellular proximal domains are highly conserved and important for normal protein interactions among each DGC components [29]. Their extreme C-terminus domains constitute a trimeric β/δ/γ-SG complex with a symmetric EGF receptor-like structure, which might respond to an unidentified ligand or interact with a yet unidentified protein in the extracellular matrix [43]. Within the domains, there is a cysteine-rich motif that is supposed to form intramolecular disulfide bonds [29,43]. The function of these disulfide bonds is still unclear and presumed to be important for their membrane localization. In case of δ-SG, two disulfide bonds are predicted to form between 264C and 289C and between 266C and 282C.
Mutations in the DGC genes can cause disruption of the complex and subsequently result in LGMD and/or DCM [12][13][14][15][16]. The mechanisms mainly involve in changes in protein localization, trafficking, protein-protein interaction, intramolecular disulfide bonds formation, and glycosylation. In cases of LGMD-2F-causing mutations reported so far, five (p.E93X, p.A131P, p.R165X, p.E262K, and c.656C) are located in the extracellular domain and only one (p.W30X) is located in the intracellular domain of the δ-SG protein [30][31][32][33]. In cases of DCM-causing mutations in the δ-SG gene, all three (p.S151A, c.K238, and p.R71T) are located at the extracellular domain [18,21,22]. p.S151A disrupted the nuclear localization of lamin A/C and emerin in cardiomyocytes due to nuclear sequestration of mutant δ-SG [44]. C.K238 is reported to disturb glycosylation of the protein. The p.R71T is located in the conserved region of the δ-SG gene and is supposed to alter the secondary structure of the protein by unknown mechanism [22].
The p.Q283R polymorphism is located between the disulfide bond-forming 282C and the glycosylation site of 285N. The WT residue Q is homologous to βand γ-SG in the same location (Fig 4E). The conversion of Q!R results in a relatively dramatic change in the electric charge. As reported by Sakamoto et al [17], δ-SG binds to all other five DCG components with different affinity and we thus test the effect of 283R on its protein interactions with other five DCG components. Our results demonstrated that δ-SG binds to β-SG strongly, supporting that β-SG and δ-SG are fundamental to the assembly of the DGC. Our results also showed that δ-SG directly and strongly binds to γ-SG. However, the GST-δ-SG-CT-283R protein held a normal binding affinity with βand γ-SG as GST-δ-SG-CT-283Q did, suggesting that p.Q283R is not affecting its interaction with βand γ-SG. Instead, p.Q283R is associated with a mild reduction of β-, γ-, and δ-SG on the plasma membrane localization. Thus, it can cause disruption of the DGC and subsequently result in DCM [12][13][14][15][16]. In vivo effect of p.Q283R on the assembly of DGC needs to be evaluated in either DCM patients carrying 848G/G or in 848G/G-knock-in mouse model. In addition, p.Q283R may affect trimeric β/δ/γ-SG's EGF receptor-like stereostructure for some unidentified binding proteins in the extracellular matrix [29,43]. Also, δ-SG is reported to be located in the terminal cistern of the sarcoplasmic reticulum independently of dystrophin and thus δ-SG-283R could be involved in the mechanism of the dysfunctional calcium signaling in DCM [45].
There are three limitations in this study. First, the size of the CM patient groups is small at a single center. The results might be affected due to bias of administered patients. Our findings need to be confirmed in a large cohort of DCM patients. Second, all of our patients samples were randomly collected and have no pedigree data. Therefore, the transmission mode of the co-occurred polymorphism is still unclear. Third, we lacked cardiac biopsy evidence in subjects with -/-of c. -100~-110 to show the actual changes in δ-SG gene transcription and in the DCM patients with the co-occurred -/-of c. -100~-110 and 848A/G to further show the mildly decreased plasma membrane localization of δ-SG-Q283R.
In conclusion, our results change our previous notion that the mutation frequency in the δ-SG gene is quite uncommon in patients with DCM, at least in eastern Asia. The co-occurred -/-of c. -100~-110 and 848A/G could be used as a screening marker for genetic determination of DCM susceptibility in the regional countries.