|Home | About | Journals | Submit | Contact Us | Français|
Desmoplakin is a ubiquitous component of desmosomes and desmosome-like structures, such as the cardiomyocyte area composita. Two major isoforms, desmoplakin I (DSPI) and desmoplakin II (DSPII) are encoded by alternative mRNA transcripts differentially spliced from the same gene. The resulting proteins are identical in amino acid sequence with the exception that DSPII contains only one third of the central alpha-helical rod domain present in DSPI. Here we describe a novel minor isoform of desmoplakin that is also produced by alternative splicing of the desmoplakin gene and that we name desmoplakin Ia (DSPIa). DSPIa is an alternatively spliced DSPI mRNA with a unique splice donor site that is 90% homologous to and downstream of the DSPII specific donor. The resulting DSPIa mRNA is in-frame and encodes a protein that has a central alpha-helical rod domain of intermediate size and that is 156 amino acids larger than DSPII and 443 amino acids smaller than DSPI. We demonstrate, through recombinant expression and short interfering RNA knockdown, that the DSPIa protein is readily detectable, albeit at substantially lower levels than the dominant isoforms, DSPI and DSPII. DSPIa mRNA has a similar tissue distribution to that of DSPI and of DSPII.
The online version of this article (doi:10.1007/s00441-010-0989-1) contains supplementary material, which is available to authorized users.
Desmoplakin (DSP) is a key junctional protein necessary for the morphogenesis and integrity of epithelial and vascular tissues, being a linker protein providing attachment for cytoskeletal elements such as intermediate filaments (IFs; Getsios et al. 2004). DSP ablation in mice results in an embryonic lethal phenotype (Gallicano et al. 1998), whereas conditional targeting to the epidermis leads to severe trauma-induced intercellular separation (Vasioukhin et al. 2001). Human mutations in the DSP gene can result in a range of clinical phenotypes (Cabral and South 2009); from striate palmoplantar keratoderma (Armstrong et al. 1999; Whittock et al. 1999), a mild skin condition, to sudden death from heart defects (Norgett et al. 2000, 2006; Uzumcu et al. 2006), through to early lethality as a result of devastating skin blistering and subsequent water loss (Jonkman et al. 2005). DSP is a ubiquitous component of desmosomes and desmosome-like structures, such as the cardiomyocyte area composita (Angst et al. 1990; Franke et al. 1982, 2006). Two major isoforms, desmoplakin I (DSPI) and desmoplakin II (DSPII) are encoded by alternative mRNA transcripts differentially spliced from the same gene (Green et al. 1990). Although usually expressed together, DSPI is the predominant isoform in heart (Angst et al. 1990; Uzumcu et al. 2006). DSPI and DSPII differ only in the size of the central alpha-helical rod domain, DSPII having two thirds fewer amino acids within this domain (Green et al. 1990). Human mutation leading to DSPI ablation while leaving DSPII intact results in severe arrhythmogenic right ventricular cardiomyopathy (ARVC) and early heart failure coupled with a mild skin phenotype of woolly hair and palmoplantar keratoderma in a single patient aged 4 years. These data reveal the importance of DSPI in heart tissue but highlight the ability of DSPII to support early heart development and to maintain skin homeostasis in inter-follicular epidermis (Uzumcu et al. 2006).
Here, we described the identification and characterisation of a novel human DSP isoform, DSPIa. DSPIa is expressed as a result of an in-frame alternative splice donor site within the DSPI gene. DSPIa mRNA encodes a protein identical to DSPI and DSPII with the exception of the central alpha-helical rod domain, which is of intermediate size, being 156 amino acids larger than DSPII and 443 amino acids smaller than DSPI. This new isoform has been previously overlooked presumably because of similarities in molecular weight and a much reduced expression level compared with DSPI and DSPII. We describe the tissue distribution of DSPIa, showing it to be similar to that of DSPI and of DSPII, and demonstrate that DSPIa is detected in protein lysates from epidermal cells and heart tissue.
RNA was isolated from frozen skin sections by using the RNAeasy mini kit (Qiagen, Calif., USA) and from cultured cells at 70%–90% confluency by using the RNA-Bee reagent (AMS Biotechnology, Spain) according to the manufacturers’ specifications. cDNA was generated by using M-MLV reverse transcription (RT; Promega, Calif., USA). RT-polymerase chain reaction (RT-PCR) for expression profiling of DSP isoforms in a range of tissues was carried out on normal skin cDNA isolated as described and on Human Multiple Tissue cDNA panel 1 and Human Cardiovascular Multiple Tissues cDNA panel (BD Biosciences, N.J., USA). The G3PDH (glyceraldehyde-3-phosphate dehydrogenase) gene was amplified as a control of cDNA abundance according to the manufacturer’s specifications. Primers, which amplified cDNA corresponding to DSPI, DSPIa and DSPII isoforms, were as follows: DSPIF, 5′ CTTGGAACTAAGGAGCCAG; DSPIaF, 5′ GCTTGATAGACTTTCAAGGG; DSPIIF, GATCGAAGTTTTGGAAGAGG; DSPR, CCACCTGAGTACACTGATTC.
Human liver and aorta total RNA was purchased from Clontech (BD Biosciences) and human ventricle and atrium total RNA was purchased from AMS Biotechnology Europe (Abingdon, UK). cDNA was constructed as described. Real-time quantitative PCR was performed by using the MiniOpticon detection system (Bio-Rad, UK) and the DyNAmo SYBR Green qPCR kit (Finnzymes, Finland). DSP isoform-specific primers were as described above. EF1alpha or B2M were used as reference genes. Cycling conditions were according to the manufacturer’s instructions. Relative quantification was established by defining the difference between the reference and target cycle threshold values for each sample.
Small fragments of redundant atrium tissue were obtained during cardiac surgery and transported to the laboratory in dry ice. These small fragments were solubilised into an appropriate volume of 0.1 M TRIS-HCl pH 6.8, 0.2 M dithiothreitol, 4% (w/v) SDS, 0.2% (w/v) bromophenol blue and 20% (v/v) glycerol and subsequently boiled for 5 min before resolution on NuPAGE Novex 3%–8% TRIS-acetate Mini gels according to the manufacturer’s specifications (Invitrogen, The Netherlands). Whole-cell protein extracts were prepared from cells lysed in 0.125 M TRIS-HCl pH 6.8, 4% (w/v) SDS, 20% (v/v) glycerol, 0.001% (w/v) bromophenol and 1.44 M β-mercaptoethanol, boiled for 5 min and resolved on SDS-PAGE as described. The primary antibody used was 11-5F (mouse monoclonal anti-DSPI and DSPII), a generous gift from David Garrod (Parrish et al. 1987).
DSPIa was cloned from primary human keratinocyte RNA by using standard molecular biology techniques. The use of the pBabe-puro retroviral vector (Morgenstern and Land 1990) and phoenix packaging system (Kinsella and Nolan 1996) to introduce full length DSPIa was as described elsewhere (South et al. 2003).
Isoform-specific short interfering RNAs (siRNAs) were designed by using the custom siRNA design tool from Thermo Fisher Scientific (Waltham, Mass., USA) according to the manufacturer’s specifications. siRNA sequences targeting DSPI/DSPIa were as follows: si2, ATAAGGAGATCGAGAGACT; si3, GGCCTGTGGCTCTGAGATA; si4, AGATAGAACTGAAGCAGGT. siRNA sequences targeting DSPI were as follows: siDSPI 1, GAGCTTATCTGAAGAAATA; siDSPI 2, CGAACAGAGGAGAGCGTAA. The siRNA designated as siI/II, which targets all three isoforms, was published previously (Wan et al. 2007). Transient transfections of HaCaT cells were performed according to the DharmaFECT general transfection protocol (Thermo Fisher Scientific, UK) for a 6-well plate format. Briefly, 2×105 cells per well of a 6-well dish were seeded and incubated in antibiotic-free medium containing fetal bovine serum, at 37°C, for 24 h prior to siRNA transfection. In separate polystyrene tubes, 100 nM siRNA (final concentration) and 6 μl DharmaFECT 1 were mixed in serum- and antibiotic-free media and incubated at room temperature for 5 min. The siRNA-containing medium was added to the tube containing the DharmaFECT 1 reagent and these contents were mixed and incubated for 20 min at room temperature. Serum-containing media was added to the mix and the cells were incubated in this siRNA-containing media for 4 days and subsequently prepared for Western blot. Cells transfected with a pool of four non-targeting siRNAs (on target plus siControl non-targeting pool) and cells incubated with DharmaFECT 1 transfection reagent only (mock) were used as negative controls.
While cloning DSPI, we amplified a shorter than expected cDNA product by using DSPI-specific PCR primers. Direct sequencing revealed this PCR product was 1329 bp shorter than DSPI but corresponded to the DSPI sequence. Alignment analysis showed that this shorter cDNA had been generated by splicing involving the exon 24 splice acceptor (common to DSPI and DSPII) and a previously unidentified splice donor within exon 23 of DSPI. The spliced DSPI sequence was in-frame and potentially coded for a desmoplakin protein intermediate in size between DSPI and DSPII. We named this cDNA sequence DSPIa. Figure 1a shows the organisation of the DSP gene at exons 23-24 for DSPI, DSPII and the novel DSPIa mRNA.
Sequence analysis demonstrated that the DSPIa donor splice site is 90% homologous to the exon 23 donor site of DSPII (Whittock et al. 1999; Fig. 1a). Splice analysis of the 3000-bp region spanning the end of exon 23 and the beginning of exon 24 (see Fig. 1a) predicted DSPII and DSPIa to harbour the two most efficient splice donors with equal or highly similar prediction scores (Figs. 1b, S1a). Interestingly, neither of the splice prediction software packages that we used identified the DSPI donor site by using default settings (Figs. 1b, S1a). Reducing the cut-off of detection did eventually identify the DSPI donor site but the sequence gave a considerably lower value than DSPII, DSPIa and other potential splice donor sites within this region (Fig. S1b, data not shown).
In order to establish whether DSPIa was present in the EST database (dbEST; Boguski et al. 1993), we searched for DSP sequences by using the BLAST 2.2.22 megablast program (accessed from http://blast.ncbi.nlm.nih.gov/Blast.cgi on 17-02-2010; Zhang et al. 2000). Using 50 bp spanning each of the three isoform unique exon-exon junctions, we identified 36 separate human EST entries for DSPI, five separate human entries for DSPII and one human entry for DSPIa. The prostate cancer cell line EST homologous to DSPIa (accession AI525952) was of poor sequence quality but displayed 86% homology over 342 bp spanning the DSPIa exon-exon boundary by using the BLAST 2.2.22 blastn program (accessed on 17-02-2010; Altschul et al. 1997). Continuous homology between this EST and DSPI or DSPII did not extend across exon-exon boundaries demonstrating that this EST represented DSPIa and that DSPIa is expressed in the prostate cancer cell line DU-145. Figure 1c depicts the number of human EST identified by searching dbEST with the corresponding sequences.
To identify the extent of DSPIa expression in various human tissues, we designed a DSP isoform-specific RT-PCR (Fig. 2a). Using this RT-PCR assay, we examined the expression of DSPIa by using multiple tissue cDNA panels and demonstrated that DSPIa expression was similar to that of DSPI and of DSPII but, surprisingly, was the only isoform detected in the aorta (Fig. 2b, c).
In order to verify our findings in the aorta and to examine the relative level of DSPIa in cultured keratinocytes, epithelial and heart tissues, we used quantitative PCR (Q-PCR) and total RNA-derived cDNA to show that DSPIa relative to DSPI levels are similar when comparing epidermal keratinocyte cells and cell lines (ratio: 3–7 per relative 100 DSPI) but that the relative level of DSPIa in simple epithelial tissue (liver) and heart was lower (ratio: 0.1–2 per relative 100 DSPI; Fig. 2d). Using this approach, we were able to detect DSPI and DSPIa in the aorta and show that the expression of both these DSP isoforms was significantly lower compared with other heart compartments (Fig. 2e).
Because of the size and similarity of DSP proteins, we speculated that DSPIa might have previously been overlooked in Western blotting because of co-migration and relative abundance. We therefore resolved total protein lysates from cultured human epidermal keratinocytes and human atrium tissue on 3%–8% gradient TRIS-Acetate gels for a considerably longer time than we would normally. Immunoblotting with a specific DSP antibody identified a faint immunoreactive band of the expected size for DSPIa (Fig. 3a). In order to confirm that this protein was indeed DSPIa, we expressed DSPIa in A293 phoenix cells and the epidermal keratinocyte cell line, HaCaT. Recombinant DSPIa migrated at a distance corresponding to the faint immunoreactive band in HaCaT cells (Fig. 3b).
To confirm further the existence of detectable levels of DSPIa, we used siRNA specifically to inhibit the mRNA of DSPI, DSPI and DSPIa and all three isoforms. siRNAs si2, si3 and si4 targeting DSPI and DSPIa (Fig. S2) significantly reduced the level of both corresponding proteins (Fig. 3c), whereas siRNA targeting DSPI only (Fig. S2) did not effect the levels of DSPIa (Fig. 3d). All three isoforms were downregulated with siRNA siI/II, which targets the 5′ end of DSP (Fig. 3d). These data demonstrate that all immunoreactive bands are DSP-specific and that only the knockdown of DSPI does not effect DSPIa expression.
Pre-mRNA splicing is a mechanism of post-transcriptional regulation of gene expression by which non-coding introns are excised and exons are joined to form a mature mRNA transcript. Alternative pre-mRNA splicing is thought to occur in approximately 75% of all human genes (Johnson et al. 2003) and, by this mechanism, human cells are able to synthesise over 90,000 proteins from approximately 26,000 coding genes (Ast 2004). Many of these alternative splice variants are differentially expressed according to tissue type and developmental stage. Alternative splicing involves sequential phosphodiester transfer reactions that are catalysed by a large ribonucleoprotein complex. Three consensus sequences within each intron are required for splicing to occur; 5′ and 3′ splice sites (donor and acceptor sites, respectively) that flank each intron, and a branch point that is located more than 18 nucleotides upstream of the acceptor site and more than 48 nucleotides downstream of the donor site (Matlin et al. 2005). Here, we identify a previously undescribed DSP isoform, DSPIa, which is produced by alternative splicing of the DSP gene. Given that DSPIa shares its acceptor site with all isoforms of DSP and that its donor splice site shares 90% homology with DSPII, differing only in the 10th 3′ base, we speculate that the branch points differ between DSPII and DSPIa. The possibility nevertheless exists that the branch point is shared and that DSPII is the major splice variant because of efficiencies of the spliceosome at the most 5′ sequence but our finding of DSPIa in heart tissue in the absence of detectable DSPII (Fig. 3a) suggests a unique branch point. Following on from this logic, it is also likely that separate branch points direct the variable tissue expression of DSPI and DSPII.
Our Q-PCR data show that we are able to detect all DSP isoforms in all tissues tested with the exception of the aorta in which DSPII is below the threshold of detection, data in agreement with our original RT-PCR panel screen (Fig. 2c, d). This is surprising given that, so far, no DSP protein has been detected in junctions of the vascular endothelium of human vessels, with the exception of the complexus adhaerentes identified in the lymph node sinus (Moll et al. 2009). Such junctions have also been identified in the cultured human umbilical vein endothelial cell line (Valiron et al. 1996). Our finding of extremely low levels of DSP in aorta RNA suggests that the sensitive Q-PCR methodology is able to detect low levels of mRNA transcripts either regulated in a post-transcriptional manner or contaminating RNA from a separate compartment. The ability to detect DSPI by Q-PCR but not by RT-PCR probably reflects template complexity; the RT-PCR cDNA panels used in Fig. 2b, c are produced from polyadenylated mRNA and are at low concentrations (total input approximately 5 ng), whereas the Q-PCR assay cDNA are constructed from total RNA at a concentration of around 400 ng. The lack of detectable DSPII is curious but, notably, PCR is dependent on the efficiency of primer binding coupled with template complexity and we cannot compare absolute levels between the different DSP isoform Q-PCR assays relative to control (Bustin 2000). A definitive demonstration of expression relies on protein detection, which in itself is dependent on antibody specificity. Here, we demonstrate the detection of DSPIa by using an antibody raised against a common DSP epitope (Parrish et al. 1987) and, although a different protein conformation may influence recognition, we feel that this is unlikely. By this method, we have detected DSPIa protein in cultured keratinocytes and heart tissue and no DSPII in heart tissue (Fig. 3a).
DSPIa has not previously been described, presumably because of multiple factors chiefly resulting from the low level of expression relative to DSPI and DSPII (which would lead to a reduced number of EST spanning intron 23 of the DSPIa gene, for example) but also from the difficulty in resolving DSPII protein from DSPIa by using standard Western blotting techniques. Indeed, our ability to resolve DSPIa is clearly dependent upon a number of as yet undetermined parameters; occasionally, DSPIa is particularly prominent (for example, Fig. 3c) and, at other times, it is less pronounced (Fig. 3a). However, DSPIa is clearly a minor isoform of DSP that is expressed in a wide range of human tissues, with DSPIa being translated into a protein that is identical to DSPI and DSPII in both the N and C terminal portions but contains only half of the DSPI alpha helical rod domain. Lack of evidence for DSPIa protein in the extensive documentation of DSP can be attributed to the widespread use of non-human tissue (Mueller and Franke 1983) and cell lines (Green et al. 1991) and insufficient protein resolution. We are, however, tempted to speculate that DSPIa has been previously identified by using Western blotting of protein lysates from FaDu cells (human upper respiratory tract tumour line) by Angst and colleagues (1990) but, again, the resolution and quantity of protein was not sufficient for conclusive evidence.
The function of the various domains of DSP has been extensively investigated (Bornslaeger et al. 1996; Fontao et al. 2003; Green et al. 1990, 1992; Kouklis et al. 1994; Kowalczyk et al. 1994; Meng et al. 1997; Smith and Fuchs 1998; Stappenbeck et al. 1993, 1994; Stappenbeck and Green 1992; Vasioukhin et al. 2001). The C-terminal domain of DSP is responsible for mediating interactions with various types of IFs: vimentin in meninges and dendritic cells of lymph nodes, desmin in the myocardium and a variety of keratins in epithelial tissues (Fontao et al. 2003; Kouklis et al. 1994; Meng et al. 1997; Stappenbeck et al. 1993; Stappenbeck and Green 1992). The N-terminal plakin domain of DSP is important for targeting DSP to the junctional plaque through interactions with the armadillo proteins plakoglobin and plakophilin (Bornslaeger et al. 2001; Kowalczyk et al. 1997, 1999) and DSP has also been shown to be able to bind itself through this domain (Smith and Fuchs 1998). However, whereas interactions with binding partners are known to be facilitated by the N- and C-terminal domains of DSP, the rod domain is thought only to be important for self-association (Green et al. 1990; O’Keefe et al. 1989). DSP isolated from pig tongue has provided evidence that DSPI can homo-dimerise in solution, whereas DSPII remains monomeric (O’Keefe et al. 1989). However, the rod domains of both DSPI and DSPII are predicted to form homophilic coiled-coil dimers (Green et al. 1990; Virata et al. 1992). Such higher order aggregates of DSP are thought to form a filamentous network that interacts with other desmosomal plaque proteins between the membrane and IFs (Getsios et al. 2004). North et al. (1999) have proposed, based on the immunogold labelling of the N- and C- termini revealing consistent spatial localisation in tissues expressing both DSPI and DSPII, that DSPII governs the thickness of the inner dense plaque and that DSPI is folded within the plaque. This logic would translate to DSPIa but the functional consequences of shortening the rod domain remain unknown. Interestingly, the detection of DSPIa in heart tissue raises the possibility that DSP protein with a shorter rod domain, regardless of overall amounts, might be necessary for desmosomal integrity. Notably, however, the assembly of desmosomal components in vitro is possible with DSPI and, therefore, DSPII and DSPIa might simply be “structural accessories” (Bornslaeger et al. 2001).
Nine human mutations have so far been reported within the rod domain of DSP. Two of these mutations (p.R1113X and p.R1934X) affect all three isoforms; another three (p.R1255K, p.R1267X and p.N1324fsZ23) affect DSPI and DSPIa and four (p.R1775I, p.K1583R, p.L1654P and p.Q1446X) affect only DSPI while leaving DSPIa and DSPII intact (Bolling and Jonkman 2009). Three of the mutations affecting only DSPI are missense mutations causing dominant ARVC, whereas the single nonsense mutation affecting only DSPI, p.Q1446X, causes recessive cardio-cutaneous disease when inherited with another nonsense mutation, p.Q673X, which affects all isoforms (Asimaki et al. 2009). This combination of mutations is predicted to lead to complete ablation of DSP expression from one allele (p.Q673X) and ablation of DSPI only, allowing expression of DSPII and DSPIa from the other allele (p.Q1446X). This contrasts with the homozygous p.R1267X nonsense mutation, which leads to complete absence of both DSPI and DSPIa, leaving DSPII intact, and resulting in recessive cardio-cutaneous disease (Uzumcu et al. 2006). In this case (p.R1267X/p.R1267X), the cutaneous phenotype is relatively mild (compared with p.Q673X/p.Q1446X) presumably because of normal DSPII expression (which is haplo-insufficient in p.Q673X/p.Q1446X and results in more severe cutaneous disease). Interestingly, the cardiac phenotype might be considered more severe in p.R1267X/p.R1267X with cardiac failure at age 4 years compared with cardiac failure at age 9 years in the p.Q673X/p.Q1446X case and therefore we are tempted to speculate that these data show that expression of DSPIa, although not capable of rescuing the cardiac phenotype, can influence its severity and positively contributes towards cardiac cohesion.
The ratio between DSPI and DSPII varies among the different tissues and different cell types; DSPII, which is abundantly expressed in complex epithelial tissues, is expressed at lower levels in cells derived from some other epithelial tissues such as epidermoid carcinoma (vulva) and human bladder carcinoma (transitional epithelium; Angst et al. 1990) and in the heart (Uzumcu et al. 2006). We show here that the relative levels of DSPI, DSPII and DSPIa are consistent in keratinocyte-derived cells and that DSPIa, like DSPII relative levels, is decreased in the liver and heart compared with keratinocytes.
The majority of desmosomal genes are alternatively spliced and little is known about the role of some of these splice isoforms, such as DSC “b” proteins or indeed DSPII. A large volume of research has been published based on recombinant DSPI and whether DSPII or DSPIa has any differential biological role in cell junction formation and signalling remains to be seen. Further work will be necessary to investigate the importance and function of these isoforms.
Below is the link to the electronic supplementary material.
Splice donor analysis of 3,000bp of sequence spanning intron 23 of the DSP gene. a - Table illustrating the NetGene2 2.42 neural network prediction of splice donor sites accessed at http://www.cbs.dtu.dk/services/NetGene2/ (Hebsgaard et al. 1996). Splice donor sites for DSPII and DSPIa are indicated in grey. b - NNSPLICE version 0.9 accessed via http://www.fruitfly.org/seq_tools/splice.html (Reese et al. 1997) identifies the splice donor sites used by DSPII, DSPIa and DSPI (grey). Note the reduced score given to the DSPI donor in relation to all other predicted sites. (PDF 149 kb)
Nucleotide sequences of six short interfering RNAs (siRNA) directed against DSPI (siDSPI 1/2), DSPI and DSPIa (siI), or all three splice isoforms (siI/II) and their targeting sites. (PDF 54 kb)
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.
This work was supported by Barts’ and The London Charity, Medical Research Council and the Dystrophic Epidermolysis Research Association.