|Home | About | Journals | Submit | Contact Us | Français|
Eight different nonsense mutations in the human rhodopsin gene cause retinitis pigmentosa (RP), an inherited degenerative disease of the retina that can lead to complete blindness. Although all these nonsense mutations lead to premature termination codons (PTCs) in rhodopsin mRNA, some display dominant inheritance, while others are recessive. Because nonsense-mediated decay (NMD) can degrade mRNAs containing PTCs and modulate the inheritance patterns of genetic diseases, we asked whether any of the nonsense mutations in the rhodopsin gene generated mRNAs that were susceptible to degradation by NMD. We hypothesized that nonsense mutations that caused mild RP phenotypes would trigger NMD, whereas those that did not engage NMD would cause more severe RP phenotypes—presumably due to the toxicity of the truncated protein. To test our hypothesis, we transfected human rhodopsin nonsense mutants into HEK-293T and HT-1080 human cells and measured transcript levels by qRT-PCR. In both cell lines, rhodopsin mutations Q64X and Q344X, which cause severe phenotypes that are dominantly inherited, yielded the same levels of rhodopsin mRNA as wild type. By contrast, rhodopsin mutations W161X and E249X, which cause recessive RP, showed decreased rhodopsin mRNA levels, consistent with NMD. Rhodopsin mutant Y136X, a dominant mutation that causes late-onset RP with a very mild pathology, also gave lower mRNA levels. Treatment of cells with Wortmannin, an inhibitor of NMD, eliminated the degradation of Y136X, W161X, and E249X rhodopsin mRNAs. These results suggest that NMD modulates the severity of RP in patients with nonsense mutations in the rhodopsin gene.
Retinitis pigmentosa (RP) is an inherited degenerative disease of the retina. Patients with RP initially start to lose peripheral and night vision, as rod photoreceptors degenerate. This early stage is followed by cone photoreceptor degeneration, leading to a loss of color vision and in some instances leaving the patients completely blind (Hartong et al., 2006). RP affects 1 in every 4000 individuals worldwide, with mutations in more than 40 different genes (Berger et al., 2010). Mutations in the rhodopsin gene account for ~30% of all RP cases. In fact, more than 150 mutations in rhodopsin cause RP (http://www.sph.uth.tmc.edu/retnet/; http://www.hgmd.org), making it the single most affected gene in this disease and a prime candidate for gene therapy (Chan et al., 2011; Wilson and Wensel, 2003).
Rhodopsin is a light-detecting G-protein coupled receptor expressed specifically in the rod photoreceptor cells of the retina. Its protein component, opsin, in combination with the chromophore, 11-cis-retinal, is responsible for efficient photon capture under dim light conditions. Upon capture of a photon, the chromophore is converted to all-trans-retinal, causing a conformational change in rhodopsin that activates the associated G-protein transducin, initiating the phototransduction cascade and ultimately triggering the sensation of vision. Rhodopsin is densely packed in the disks of rod outer segment, providing a critical structural component of rod photoreceptors, as well (Lem et al., 1999; Price et al., 2012). The intertwined roles of rhodopsin in rod cell structure and function may account for the large number of human rhodopsin mutations that have been found cause RP.
RP mutations in the rhodopsin gene include missense and nonsense mutations, insertions and deletions, as well as splice site mutations. These mutations can affect many aspects of rhodopsin function, including protein folding, post-translational modification, transport to the outer segment, and signaling capabilities. Nearly all mutations in rhodopsin cause dominant RP, with just five of more than 150 giving rise to recessive RP. Two of these recessive phenotypes are due to nonsense mutations W161X and E249X, which generate premature termination codons (PTCs) in the mRNA (Kartasasmita et al., 2011; Rosenfeld et al., 1992). But not all rhodopsin nonsense mutations cause recessive RP. Five rhodopsin nonsense mutations—Y60X, Q64X, Q312X, Q341X and Q344X—give rise to dominant RP (Cideciyan et al., 1998; Eisenberger et al., 2013; Macke et al., 1993; Sung et al., 1991; Zhao et al., 2001). In addition, nonsense mutation Y136X, although dominant, causes a very mild form of late onset RP (Sanchez et al., 1996). It is unclear whether the different phenotypes of these eight nonsense mutations result from varying toxicities of the truncated proteins or from the differential effects of nonsense-mediated decay (NMD) on the stability of the PTCcontaining rhodopsin mRNAs.
NMD is the cell’s quality control mechanism for degrading PTC-containing mRNA transcripts, which can produce potentially deleterious truncated proteins (Popp and Maquat, 2013). In several cases, NMD is known to modulate the severity of the disease phenotype and the inheritance pattern (Holbrook et al., 2004; Khajavi et al., 2006). In the classic example, nonsense mutations in the β-globin gene that engage NMD give rise to a recessive form of β-thalassemia, whereas mutations that escape NMD can cause a dominant version of the disease (Chang and Kan, 1979; Hall and Thein, 1994). Although it is not possible to predict with certainty whether a particular PTC-mRNA will engage NMD, PTCs that are beyond a critical boundary (50–55 nucleotides in front of the last splice junction) typically escape NMD, whereas PTCs upstream of that boundary usually engage NMD (Nicholson et al., 2010).
For our studies, we chose to test Q344X, one of the three dominant mutations downstream of the critical boundary, and Q64X, Y136X, W161X and E249X, four of the five mutations upstream of the boundary. The phenotypes of these mutations—Q64X and Q344X are severe, whereas Y136X, W161X, and E249X are mild—do not match the simplest expectations for NMD based on their positions relative to the critical boundary. As we show, however, their phenotypes are congruent with actual measurements of NMD: Q64X and Q344X escape NMD, whereas Y136X, W161X and E249X engage NMD.
Because NMD occurs in the context of mRNA splicing, we used the full-length rhodopsin gene, including all exons and introns, to build test plasmids. We showed previously that a full-length rhodopsin gene in a similar construct was expressed and spliced normally in HT1080 cells (Intody et al., 2000). The full-length rhodopsin gene was synthesized by GeneScript in two pieces: segment a extends from the beginning of exon 1 to the midpoint of intron 2; segment b extends from the midpoint of intron 2 to the end of exon 5. Each segment was cloned into the HindIII site of the pUC57 plasmid. Four constructs were generated by GeneScript: pRa, pRa-W161X, pRb and pRb-E249X. We introduced the Y136X point mutation separately in the pRa plasmid using the QuickChangeII site-directed mutagenesis kit (Agillent #200521) to generate plasmid pRa-Y136X. Appropriate segments were assembled into full-length rhodopsin genes by recombineering, using the In-FusionHD kit (Clontech #638909), following the manufacturer’s protocol. We used the high fidelity KOD Hot Start DNA polymerase (EMD Millipore, 71086) for all In-Fusion PCR reactions, following the manufacturer’s instructions. In-Fusion primers were designed using Clontech’s Primer Design Tool for In-Fusion and synthesized by Sigma-Aldrich. We assembled full length sequences by In-Fusion cloning: pRb was cloned into pRa to generate pRab-WT, pRb was cloned into pRa-Y136X to generate pRab-Y136X, pRb was cloned to pRa-W161X to generate pRab-W161X, and pRb-E249X was cloned into pRa to generate pRab-E249X. Full length rhodopsin genes were then moved by In-Fusion cloning to HindIII-linearized pcDNA3.1(+) (Invitrogen #V790-20), leading to plasmids: phRhoWT, phRhoY136X, phRhoW161X and phRhoE249X, each of which is driven by the cytomegalovirus (CMV) promoter and carries a downstream Neomycin gene flanked by an SV40 promoter and an SV40 poly A site (Figure 1A). Rhodopsin mutations Q64X and Q344X were generated on phRhoWT by ProNovus Bioscience, who used site-directed PCR mutagenesis to produce plasmids phRhoQ64X and phRhoQ344X. All constructs were sequenced by Baylor College of Medicine’s DNA Sequencing Core and Lone Star Labs to confirm the presence of the target mutations.
Plasmids containing the normal β-globin gene and the 39Ter (Q39X) mutation were provided by Dr. Lynn Maquat (Zhang et al., 1998). The normal and mutant β-globin genes were cloned into pcDNA3.1 so that they were flanked by a CMV promoter and a bovine growth hormone (BGH) poly A site (Figure 1B).
Two human cell lines were used. HEK293 cells were maintained in plates containing DMEM with 4.5 g/L d-glucose (GIBCO, 11965) supplemented with l-glutamine (GIBCO, 25030). HT1080 cells were maintained in plates containing DMEM/F-12 (GIBCO, 11320), supplemented with 1% MEM nonessential amino acids (GIBCO, 10370). All cells were maintained in fresh growth medium supplemented with 10% HyClone Fetal Bovine Serum (Thermo Scientific, SH30071.03HI) at 37°C in a humidified incubator with 5% CO2 for the duration of the experiments.
Cells were transfected with 1 µg of plasmid per well of a six-well plate, using Lipofectamine 2000 (Invitrogen, 11668019) and following the manufacturer’s protocol. Cells were trypsinized 48 hours after transfection and total RNA was extracted using a NucleoSpin RNA extraction kit (Macherey-Nagel, 740955), following the manufacturer’s protocol, and supplemented with β-mercaptoethanol (Sigma-Aldrich).
Because NMD, once engaged, is known to be a rapid and quantitative process (Lejeune et al., 2003), it is standard practice to monitor it by measuring steady-state mRNA levels, and that is the approach we have taken, using quantitative RT-PCR. RNA quality and concentration were assessed using the Experion automated electrophoresis system with RNA StdSens chips (BioRad Laboratories, 700-7103), following the manufacturer’s protocol. First strand synthesis was performed on 2 µg of total RNA per sample in a final volume of 20 µL, using the Omniscript RT kit (QIAGEN, 205111), following the manufacturer's instructions. We prepared the quantitative PCR reaction using the QuantiFast SYBR Green PCR Kit (QIAGEN, 204054) and 1 µL cDNA reaction per sample in a final volume of 25 µL, according to the manufacturer's protocol. PCR reactions contained 0.5 µM of each primer. Primer set 1 (5′-GTGAACGCTCCCGGCTTG and 5′-CCAGGTAGTACTGTGGGTAC) specifically amplified the human rhodopsin gene; primer set 2 (5′-GTGCCTTTAGTGATGGCCTG and 5′-GCCACCACCTTCTGGAAGG) specifically amplified the β-globin gene; and primer set 3 (5′-CTTTTCTGGATTCATCGACTGTG and 5′-CAAGAAGGCGATAGAAGGCG) specifically amplified the neomycin gene. As designed, each primer set amplified its target with 95–100% efficiency. The locations of these primer pairs are shown in Figure 1. Amplifications were performed on a CFX96 Real-Time PCR Detection System (Bio-Rad Laboratories). The PCR cycling conditions were 5 minutes at 95°C, followed by 40 cycles of 95°C fo r 10 seconds and 60°C for 30 seconds. Following amplification, the temperature was ramped from 60°C to 95°C at a rate of 0.5°C every 5 seconds for a melt curve anal ysis. Each melt-curve analysis gave a single peak consistent with the melting of the expected PCR fragment.
Measurements of the levels of rhodopsin and β-globin mRNAs were obtained from two independent transfection experiments, each with three replicates. For each replicate, rhodopsin and β-globin mRNAs were measured relative to neomycin mRNA in the same sample. The relative levels of mRNA were calculated by comparing the number of cycles at which the PCR products became detectable above the basal threshold, which is defined as the crossover point (Livak and Schmittgen, 2001). The ratio of rhodopsin or globin mRNA to neomycin mRNA was determined three times for each sample and averaged to obtain the mean for each replicate. The six means for each ratio determination were averaged and the standard deviation was calculated. In Figure 2, the data were plotted after normalizing to the mean ratio for wild-type rhodopsin/neomycin or wild-type β-globin/neomycin, which were defined as the controls. Statistical analyses of significance were conducted using Student’s t-test to compare the mean and standard deviation of each experimental sample with the mean and standard deviation of the corresponding control, using GraphPad software.
To determine whether nonsense mutations in the rhodopsin gene can activate NMD, we compared rhodopsin transcript levels in human cells transfected with various forms of the human rhodopsin gene (Figure 1). To control for transfection efficiency, we measured rhodopsin mRNA relative to mRNA expressed from the neomycin gene on the same plasmid (Figure 1). We chose to test two different human cell lines—HEK293 and HT1080—both of which are competent to carry out NMD (Huang et al., 2012; Micale et al., 2009; Stoecklin et al., 2001). These cells were transfected with plasmids carrying the neomycin gene and a rhodopsin gene, and RNA was harvested 48 hours later. After conversion of mRNA to cDNA, we measured rhodopsin and neomycin transcript levels by quantitative reverse transcription PCR (qRT-PCR), using the primers indicated in Figure 1. In both cell lines, the Q64X and Q344X gave levels of rhodopsin mRNA that were indistinguishable from those obtained with wild type rhodopsin (Figure 2A). By contrast, Y136X, W161X, and E249X yielded significantly reduced levels of rhodopsin mRNA, approximately 40% lower than for wild type rhodopsin (Figure 2A). These observations suggest that the nonsense mutations Y136X, W161X, and E249X trigger mRNA degradation.
To confirm that HEK293 and HT1080 cells can indeed carry out NMD, we compared transcript levels for wild-type β-globin with those for the well-characterized NMD-inducing β-globin 39Ter mutation (Lejeune et al., 2003). As described above for rhodopsin, we used qRT-PCR to quantify globin mRNA relative to mRNA expressed from a neomycin gene on the same plasmid (Figure 1). As shown in Figure 2A, β-globin 39Ter yielded about 75–85% less globin mRNA than wild type. These results indicate that both cell lines are capable of robust degradation of PTC-containing mRNAs, consistent with active NMD. To verify that β-globin 39Ter mRNA degradation was due to NMD, we chemically inhibited a critical component of the NMD pathway, SMG1 kinase, using Wortmannin (Yamashita et al., 2001). In response to a ribosome stalled at a PTC, SMG1 kinase phosphorylates UPF1, which leads to recruitment of enzymes that degrade the mRNA (Lejeune et al., 2003; Popp and Maquat, 2014). We treated cells with 10 µM Wortmannin (Sigma) 42 hours after transfection for 6 hours and then immediately isolated total RNA for analysis. As shown in Figure 2B, Wortmannin treatment substantially increased expression of globin mRNA by the 39Ter nonsense mutant, indicating, as expected, that the degradation observed in the absence of Wortmannin was due to NMD.
We also used Wortmannin treatment to test whether the reduction in mRNA levels observed for rhodopsin mutants Y136X, W161X, and E249X was also due to degradation by NMD. In both HEK293 and HT1080 cells, Wortmannin eliminated the difference between wild type rhodopsin mRNA levels and the mRNA levels in the nonsense mutants (Figure 2B). These results indicate that the NMD pathway is responsible for the degradation of rhodopsin mRNA in the Y136X, W161X, and E249X nonsense mutants.
With the exception of Q64X, these results are in accord with the empirical rule that PTCs more than 50–55 nucleotides upstream of the last splice junction induce NMD, whereas PTCs downstream of that boundary do not. In the case of Q344X, we have data in a Q344X mouse knockin that shows that the level of Q344X mRNA is the same as that in a comparable wild type knockin (Sandoval et al., 2014). How Q64X manages to avoid NMD is unclear. In the β-globin and triosephosphate isomerase (TPI) genes, where NMD has been extensively studied, PTCs near the AUG start codon—within 23 codons for β-globin and within 10 codons for TPI—typically evade NMD (Peixeiro et al., 2012; Zhang and Maquat, 1997). In both cases, avoidance of NMD is associated with reinitiation of translation downstream of the PTC, at Met14 in TPI and Met55 in β-globin. Efficient reinitiation and avoidance of NMD depend on the length of the upstream open reading frame (ORF) created by the PTC and the presence of a downstream AUG. The efficiency of reinitiation decreases with increasing length of the upstream ORF (Kozak, 2001), and in one study it was predicted to decrease to zero when the upstream ORF reached 55 codons (Luukkonen et al., 1995). Thus, Q64X would seem to create an ORF that is too long for efficient reinitiation. Further experiments will be required to resolve the mechanism by which Q64X evades NMD.
The key finding of this study is the strong correlation between induction of NMD and disease phenotype. Nonsense mutations that induce NMD give rise to recessive RP (in the case of W161X and E249X) or to a dominant form with a mild phenotype (in the case Y136X). For Y136X and E249X, it has previously been speculated that NMD might account for their mild phenotypes (Rosenfeld et al., 1992; Sanchez et al., 1996). By contrast, Q64X and Q344X, which evade NMD, give rise to dominant RP. It is known that the mammalian retina, including photoreceptors, has a functional NMD pathway; in a mouse model of age-related retinal degeneration, a nonsense mutation in the Mdm1 gene caused a 40–80% reduction in Mdm1 mRNA via NMD (Chang et al., 2008). The 40% reduction in PTC-containing rhodopsin mRNA observed in cultured cells, if it also occurs in rod photoreceptor cells, would reduce the levels of the encoded protein fragments, potentially contributing to the relatively healthy retinas observed in patients with the Y136X, W161X, and E249X rhodopsin nonsense mutations.
The high level of expression of rhodopsin, which accounts for 70% of the protein in the rod outer segment, is often cited as the reason why such a large fraction of rhodopsin mutations (>95%) are dominant. Whether a 40% reduction in rhodopsin mRNA is enough to modulate the severity of RP greatly depends on how toxic each of the products encoded by Y136X, W161X, and E249X is to rod photoreceptors. NMDinduced mRNA decreases of 35–95% have been linked to recessive inheritance of the phenotype in genetic diseases such as Robinow syndrome, β-thalassemia, human hereditary complement C3 deficiency, Alpers syndrome, Ehlers-Danlos syndrome, Hallopeau-Siemens type recessive dystrophic epidermolysis bullosa, and pseudoxanthoma elasticum (Baserga and Benz, 1988; Ben-Shachar et al., 2009; Chan et al., 2005; Christiano et al., 1997; Hu et al., 2003; Inacio et al., 2004; Neu-Yilik et al., 2011; Reis et al., 2004; Romao et al., 2000; Schwarze et al., 2004). Without further studies in mice, however, we cannot be sure whether the mild phenotypes of Y136X, W161X, and E249X are due primarily to NMD-induced reductions in rhodopsin mRNA, or to the lower intrinsic toxicity of their encoded protein fragments, or to some combination of the two.
It is possible that our knowledge of NMD of rhodopsin nonsense mutations could be exploited for gene therapy of RP patients. For example, drugs that promote ribosomal read-through of PTCs could in principle, be used to treat the dominant nonsense mutations Q64X, Q312X, Q341X, and Q344X, a strategy that has been studied for more that 40 different disease genes (Lee and Dougherty, 2012; Wang and Gregory-Evans, 2015) (Guerin et al., 2008). The major challenge of this approach is getting sufficient read-through to affect the disease phenotype. Would a 5–10% increase in full-length rhodopsin, and a corresponding decrease in the toxic fragment, be sufficient to ameliorate the phenotype in dominant diseases? Would a 5–10% increase be sufficient to improve the phenotype in recessive diseases? In neither case are the threshold levels of effective read-through known.
Alternatively, it may be possible to exploit NMD in a more general approach that would work for most, if not all, of the dominant rhodopsin mutations in the RP patient population. If a stop codon could be introduced into an NMD-sensitive region of a mutant rhodopsin gene, it may be possible to replace the deleterious mutant protein with low levels of a less toxic fragment. In principle, a stop codon could be introduced by homologous recombination with an appropriate fragment, by a nonhomologous end joining-induced shift in the reading frame after cutting with an engineered nuclease, or by incorporating an exon with a potent NMD-inducing PTC. Because of the difficulty in distinguishing the mutant gene from the normal one, this approach may require using a “kill and replace” strategy, in which both genes are targeted and a resistant rhodopsin gene is supplied at the same time (Hernan et al., 2012; Millington-Ward et al., 2011; Petrs-Silva et al., 2011). The main advantage of this approach is that a single set of proven reagents could be used to treat all dominant rhodopsin mutations.
The plasmids containing the β-globin normal and mutant genes were provided by Dr. Lynn Maquat. This work was supported by National Institute of Health Grants (EY11731 to J.H.W., EY07981 to T.G.W., EY002520 core grant for vision research) and the Welch Foundation (Q0035 to T.G.W).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest