A novel homozygous missense mutation p.P388S in TULP1 causes protein instability and retinitis pigmentosa

Study participants

This study was approved by the Institutional Review Board of the Srikiran Institute of Ophthalmology and followed the tenets of the Declaration of Helsinki. The proband and his family members were recruited and examined after informed consent was received. All participants underwent detailed ophthalmologic evaluations including fundus examination by a retina fellowship-trained ophthalmologist.

Exome sequencing

Approximately 4 ml of blood was drawn from each subject by venipuncture and stored in BD Vacutainer blood collection tubes with K2EDTA (Becton Dickinson, Franklin Lakes, NJ) at 4° C. Genomic DNA from peripheral leukocytes was isolated using the QIAsymphony automated DNA extraction system and QIAsymphony DNA Midi Kit (Qiagen, Hilden, Germany) per manufacturer’s protocols.

We performed exome sequencing on genomic DNA of the proband. Library construction and target enrichment were performed using the IDT xGen Exome capture kit (Coralville, IA). The libraries were then sequenced to mean 100X on-target depth on an Illumina sequencing platform (San Diego, CA) with 150 base pairs paired-end reads. Sequences were aligned to the human reference genome b37, and variants were called using the Genome Analysis Toolkit (Cambridge, MA) [14] and annotated using SnpEff [15].

We filtered for rare missense, nonsense, splicing, or frameshift homozygous mutations with a minor allele frequency (MAF) less than 0.01 in the 1000 Genomes Project and genome aggregation (gnomAD) databases. Variants with a Genomic Evolutionary Rate Profiling (GERP²⁺) score greater than 2.0 and a Combined Annotation Dependent Depletion (CADD) score greater than 15 were considered. Known RP susceptibility-conferring genes [16] were screened with priority. Sanger sequencing was used to validate variants of interest in the proband and family members.

Generation of TULP1 constructs

The cDNA encoding for WT human TULP1 was purchased from the DNASU Plasmid Repository (HsCD00820883, Tucson, AZ). To generate the P388S mutation, Q5 site-directed mutagenesis (New England Biolabs, NEB, Ipswich, MA) of full-length human TULP1 was performed using the following primers: 5′-CGG GCA GAA CTC ACA GCG TGG-3′ and 5′-TTG TCA AAG ACC GTG AAG CGG-3′. To generate the C-terminal green fluorescent protein (GFP)-tagged WT and P388S TULP1, Gibson Assembly (HiFi Master Mix, NEB) was used to insert a Kozak sequence (DNA sequence: GCCACC) upstream of the TULP1 start codon, and a flexible linker (amino acids: GGGGS) separating TULP1 and enhanced GFP (eGFP). This TULP1-GGGGS-eGFP DNA was inserted into the peGFP-C1 vector backbone via the SalI and NheI restriction sites. All constructs were verified with Sanger sequencing.

Cell culture

Human embryonic kidney (HEK-293A, Life Technologies, Carlsbad, CA) cells were cultured at 37 °C with 5% CO₂ in Dulbecco’s minimal essential medium (DMEM) supplemented with high glucose (4.5 g/l, Corning, Corning, NY), 10% fetal bovine serum (FBS, Omega Scientific, Tarzana, CA), and 1% penicillin-streptomycin-glutamine (Gibco, Waltham, MA). For a 24-well plate, cells were plated at a density of 100,000 cells/well, and for a 12-well plate, cells were plated at a density of 180,000–200,000 cells/well. Cells were transfected the following day with either 500 ng (24 well) or 1 μg (12 well) of midi-prepped endotoxin-free plasmid DNA (Qiagen, Germantown, MD) using Lipofectamine 3000 (Life Technologies) according to the manufacturer’s protocol. Forty-eight hours after transfection, fresh media was added, and the cells were harvested 24 h later (72 h posttransfection) and processed for western blotting or quantitative PCR (qPCR). As a positive control for some experiments, cells were treated with tunicamycin (an unfolded protein response inducer, 1 μM, 24 h, Sigma cat# T7765, St. Louis, MO) and processed similarly for western blotting or qPCR. Human immortalized RPE (ARPE-19, CRL-2302, American Type Culture Collection, Manassas, VA) cells were cultured in DMEM/F12 media supplemented with 10% FBS (Omega Scientific), HEPES (Corning, Corning, NY), and penicillin/streptomycin and glutamine (PSQ, Gibco, Germantown, MD). For a 24-well plate, ARPE-19 cells were plated at a density of 100,000 cells/well and transfected the following day with 500 ng of midi-prepped endotoxin-free plasmid DNA (Qiagen) using Lipofectamine 3000 (Life Technologies). All cells used were verified for authenticity using short tandem repeat (STR) profiling (Appendix 1, University of Arizona Genomics Core, Tucson, AZ). Note that STR verification cannot distinguish among different variants of the 293-based cell lines (i.e., 293 versus 293A versus 293T).

Confocal microscopy

A glass-bottom 24-well plate (MatTek Corporation , Ashland, MA) was coated with 1X poly-D-lysine (Sigma Aldrich), rinsed with water, and allowed to dry at room temperature. HEK-293A or ARPE-19 cells were plated at a density of 100,000 cells/well and transfected the following day with 500 ng of midi-prepped endotoxin-free plasmid DNA (Qiagen). Forty-eight hours after transfection, fresh media was added, and 24 h later (72 h post transfection), the cells were washed twice with 1X PBS (Fisher BioReagents, cat# BP2944100, Waltham, MA) followed by incubation with 4% paraformaldehyde (PFA; Electron Microscopy Sciences, Hatfield, PA) for 20 min. After PFA incubation, cells were washed with 1X PBS. For the ARPE-19 cells, the cell nuclei were stained with 300 nM 4’,6-diamidino-2-phenylindole (DAPI), dilactate solution (Molecular Probes, Eugene, OR). For membrane staining, the HEK-293A cells were washed twice with 1X PBS, fixed in 4% PFA, permeabilized in 0.1% Triton X-100 for 3 min, and washed again in 1X PBS. Cells were incubated in blocking buffer (1% bovine serum albumin [BSA] in PBS) for 10 min followed by Alexa Fluor™ 633 Phalloidin (1:50 dilution in PBS; Molecular Probes) for 20 min and washed twice with 1X PBS before being imaged using a 63X oil objective on a Leica SP8 confocal microscope (Buffalo Grove, IL).

Western blotting

Cells were washed with Hanks buffered salt solution (HBSS, Corning), lysed with radioimmunoprecipitation assay (RIPA) buffer (Santa Cruz, Dallas, TX) supplemented with Halt protease inhibitor (Pierce, Rockford, IL) and benzonase (Millipore Sigma) for 3–5 min, and spun at maximum speed (21,000 ×g) at 4 °C for 10 min. The soluble supernatant was collected, and the protein was quantified via bicinchoninic assay (BCA) assay (Pierce). The insoluble pellet fractions were further washed in HBSS and centrifuged, and the pellet was resuspended in 1X sodium dodecyl sulfate (SDS) buffer containing 0.83% beta-mercaptoethanol (BME) and sonicated (30% amplitude, pulse 10 s on/off). Thirty micrograms of soluble supernatant were run on a 4–20% Tris-Gly SDS– polyacrylamide gel electrophoresis (PAGE) gel (Life Technologies) alongside the equivalent amount of insoluble protein and transferred onto a nitrocellulose membrane using an iBlot2 device (Life Technologies). After probing for total protein transferred using Ponceau S (Sigma), the membranes were blocked overnight in Odyssey PBS Blocking Buffer (LI-COR, Lincoln, NE). Membranes were then probed with mouse anti-GFP (1:1,000; Santa Cruz, cat #sc-9996), mouse anti-glucose-regulated protein 78 (GRP78, 1:1,000; Santa Cruz, cat #sc-376768), or rabbit anti-β-actin (1:1,000; LI-COR, Lincoln, NE, cat# 926–42210). Blots were imaged on an Odyssey CLx and quantified using ImageStudio (both from LI-COR).

Quantitative PCR

Transfected HEK-293A cells were trypsinized (0.25% Trypsin EDTA, Gibco), quenched with DMEM, and centrifuged at maximum speed (21,000 ×g) at 4 °C for 10 min. Cell pellets were washed with HBSS and centrifuged again, and then RNA extraction from the cell pellets was performed using the Aurum Total RNA Mini Kit (Bio-Rad, Hercules, CA). Four hundred nanograms of RNA were reverse transcribed using qScript cDNA SuperMix (Quanta Bioscience, Beverly, MA) according to vendor recommended parameters (5 min at 25° C, 30 min at 42° C, 5 min at 85° C), and the cDNA was diluted 5X in DNase/RNase-free water. cDNA was amplified with TaqMan Fast Advanced Master Mix (Thermo Fisher; Applied Biosystems, Waltham, MA (cat# 4444963), 20 sec at 95° C [initial denaturation], 1 sec at 95° C, 20 sec at 60° C, 40 cycles). TaqMan probes used were hTULP1 (cat# hs00163236_m1), hHSPA5 (cat# hs00607129_gH), hDNAJB9 (cat# hs01052402_m1), hASNS (cat# hs04186194_m1), and hACTB (cat# hs01060665_g1; Thermo Fisher; Applied Biosystems) and quantification was performed using using QuantStudio 6 Real-Time PCR software (Thermo Fisher; Applied Biosystems).

Cycloheximide-chase assay

Twenty-four hours after transfection with the WT TULP1 eGFP or P388S TULP1 eGFP constructs, the HEK-293A cells were treated in 24-well plates with cycloheximide (25 μM; Alfa Aesar, Haverhill, MA, cat# J66901-03) for 0, 1, 3, 6, and 9 h. Cells were washed with HBSS, harvested at each time point, and then processed for western blotting. Membranes were probed with mouse anti-GFP and rabbit anti-β-actin and imaged/quantified as described above.

Proband from a consanguineous family in Southern India

A 25-year-old man (Study ID: SIO221) born of a consanguineous marriage in Andhra Pradesh, India, an area where we previously identified unique autosomal recessive mutations linked to eye disease [17], presented with a history of night blindness since childhood. Best-corrected visual acuity (BCVA) in both eyes was 20/60. He had no nystagmus. Intraocular pressure and anterior segment examinations were normal. Fundus examination revealed a fairly symmetric generalized and parafoveal RPE atrophy with bone spicule-like pigmentation in the midperiphery and arteriolar attenuation (Figure 1A,B). The optic nerve head was normal in appearance. Fundus autofluorescence revealed a parafoveal ring of hypoautofluorescence corresponding to the area of RPE atrophy and a patchy decrease in autofluorescence throughout the retina in both eyes (Figure 1C,D). Optical coherence tomography scans through the macula of both eyes showed atrophy of the outer retinal layers with loss of the ellipsoid zone (EZ) and a thin epiretinal membrane (Figure 1E,F). For comparison, an age-matched healthy control patient was imaged using the same modalities (Figure 1G–L). There was no evidence of posterior staphyloma in the patient. The patient’s axial lengths were 24.47 mm and 24.25 mm, respectively. Systemic examination was normal. The examined parents and sibling (pedigree shown in Figure 2A) were asymptomatic and had normal funduscopic examinations.

Exome sequencing identifies a novel homozygous mutation in the TULP1 gene

Exome sequencing of the proband, followed by application of filtering criteria (described in Methods and the flowchart provided in Appendix 2), revealed ten possible homozygous mutations (Appendix 3), only one of which was in a gene (TULP1) known to cause RP [16]. This variant, a homozygous missense mutation (NC_000006:g.35471576G>A; NM_003322:c.1162C>T; NP_003313:p.Pro388Ser) in exon 12 of the TULP1 gene, results in a substitution of proline by serine in a conserved amino acid position (Figure 2B). Aside from the potentially pathogenic mutation in TULP1, the only known pathogenic mutation (p.Arg89His) that was identified in the affected individual was in the INS gene (Gene ID: 3630, OMIM: 176730; Appendix 4), which is associated with hyperproinsulinemia, a disease not known to result in the described ocular phenotype [18]. Nonetheless, the P388S TULP1 mutation is a novel variant absent from the 1000 Genomes Project database, the Genome Aggregation Database (v2.1.1), the TOPMed database (freeze 5), and the GenomeAsia 100 K Project database [19]. Segregation of the variant in the consanguineous pedigree was examined with Sanger sequencing to reveal that the parents are heterozygous for the mutation (Figure 2A, Appendix 5). P388 is a highly conserved residue among the species tested, including mammals (Figure 2B) with a GERP²⁺ score [20] of 4.95 (Figure 2C). In silico prediction indicates that the change to proline at this position could possibly perturb protein function or contribute to pathogenicity with a PolyPhen-2 score [6] of 0.997 (probably damaging), a CADD score [21] of 26.9, and a PROVEAN score [22] of −7.9 (deleterious). Analysis of known mutations in TULP1 showed an enrichment of mutations occurring in the C-terminus of TULP1 (> amino acid 300), with P338S falling within this region (Appendix 6).

P388S displays similar subcellular localization to WT TULP1

Previously, WT TULP1 has been shown to localize near the plasma membrane and in the nuclear compartments of COS-7 cells [23]. A separate study suggested that missense mutations in TULP1 shift its sub-cellular trafficking, resulting in ER localization [13]. Therefore, we tested whether the P388S mutant displayed localization differences compared to WT TULP1 in cultured cells. HEK-293A cells (STR verified, Appendix 1) were transiently transfected with eGFP, WT TULP1 eGFP, or P388S TULP1 eGFP constructs and analyzed for green fluorescence and counterstained with phalloidin, which binds to F-actin, using laser-scanning confocal microscopy (Figure 3A–C). As expected, expression of eGFP showed the fluorescent signal distributed evenly across the cytoplasm in cells (Figure 3A). WT TULP1 eGFP was localized near the plasma membrane as well as in the nuclear compartments of cells (Figure 3B) similar to previous reports in COS-7 cells [11,23]. Surprisingly, we found that localization of P388S TULP1 eGFP was similar to that of WT TULP1 eGFP in that it also localized predominantly near the plasma membrane and in the nuclear compartment of cells (Figure 3C), suggesting that there are no differences in cellular distribution between WT and P388S TULP1. To confirm that these observations were not cell type-dependent, we also transfected human immortalized RPE (ARPE-19) cells (also STR verified, Appendix 1) with the constructs indicated above and observed that P388S TULP1 eGFP again localized similarly to WT TULP1 eGFP in the nucleus and near the plasma membrane of the cells (Appendix 7).

Protein expression and solubility of P388S TULP1

Because we did not detect obvious differences between WT and P388S TULP1 at the sub-cellular level, we investigated other potential biochemical differences that might partially explain the RP phenotype observed in the patient with the presumed pathogenic variant, p.Pro388Sser in TULP1. We employed a biochemical approach to detect the expression and solubility of WT and P388S TULP1. Using HEK-293A cells, we transfected WT TULP1 eGFP and P388S TULP1 eGFP, and isolated the soluble and insoluble protein fractions from the cells 24 h later. WT TULP1 eGFP and P388S TULP1 eGFP in the soluble and insoluble fractions migrated as predicted at a molecular weight of about 100 kDa (Figure 4A, TULP1 is about 70 kDa [12], and eGFP is about 26–28 kDa [24]). WT TULP1 and P388S TULP1 were similarly more abundant in the RIPA-soluble fraction, as expected based on previous findings [12] (Figure 4A). However, we detected a significant 27.7±13.8% and 22.2±12.4% decrease in soluble and insoluble P388S TULP1 protein levels compared to WT TULP1, respectively (Figure 4B). Furthermore, these observed differences were not due to variations at the transcript level, as qPCR revealed no statistically significant difference between WT and P388S TULP1 (Figure 4C).

P388S is degraded more rapidly than WT TULP1

Because we observed a significant reduction in P388S TULP1 protein steady-state levels compared to WT TULP1 (Figure 3A,B), we hypothesized that this may indicate that P388S TULP1 is less stable in vitro. To more definitively address whether there were any differences in stability at the protein level between WT TULP1 and P388S TULP1, transfected HEK-293A cells were treated with cycloheximide (CHX), a translation elongation inhibitor, over the course of 9 h. With western blotting, we observed a gradual decrease in protein levels for WT and P388S TULP1 under CHX treatment over time (Figure 5A,B). Initially, we observed an 18.4±18.2% reduction in P388S levels, compared to a 4.8±5.5% reduction in WT TULP1 after 1 h of treatment with CHX (25 μM, Figure 5A–C, not statistically significant). After 3 h of CHX treatment, we observed a statistically significant 48.7±7.90% reduction in P388S levels, in contrast to the stability of WT TULP1 (5.00±12.3%, Figure 5A–C, p<0.01, t test), indicating that P388S is more rapidly degraded at this time point. Finally, at 9 h, we detected a 74.1±11.6% reduction in P388S, whereas WT TULP1 displayed only a 53.6±3.10% reduction in protein levels (Figure 5A–C, p<0.05, t test). These data suggest that P388S is generally more unstable and has a higher turnover rate compared to WT TULP1.

P388S TULP1 does not induce ER stress

Missense mutations in TULP1 have been shown to induce ER stress in vitro [13]. Similarly, we hypothesized that P388S TULP1 may also induce ER stress in cells. To test this hypothesis, we transfected HEK-293A cells and performed qPCR using TaqMan probes that are representative downstream genes of unfolded protein response (UPR) pathway activation [25]. To measure changes in ER stress, we selected the heat shock protein 70 family protein 5 (HSPA5, ATF6 activation, Gene ID: 3309, OMIM: 138120), DnaJ homolog subfamily B member 9 (DNAJB9, IRE1 activation, Gene ID: 4189, OMIM: 602634), and asparagine synthetase (ASNS, PERK activation, Gene ID: 440, OMIM: 108370) genes. We measured the mRNA expression levels of each gene in HEK-293A cells expressing either WT or P388S TULP1 and detected no statistically significant differences in the HSPA5, DNAJB9, and ASNS transcript levels (Figure 6A), suggesting that the presence of P388S does not induce ER stress within cells. We also confirmed these observations at the protein level by analyzing the GRP78 (HSPA5) levels (Figure 6B,C). We found that P388S did not induce statistically significant cellular stress in cultured cells when compared to WT TULP1. These results suggest that the P388S TULP1 variant likely contributes to RP by an alternate mechanism other than ER stress.