Animals models

The use of animal models in this study conforms to ARVO’s Statement for the Use of Animals in Ophthalmic and Vision Research. This study was approved by the West Virginia University Institutional Animal Care and Use Committee (protocol 1603001702). Euthanasia was performed by carbon dioxide inhalation followed by cervical dislocation in all experiments involving retina or eye collection, except for metabolomic analyses, in which mice were euthanized by cervical dislocation only. Intraocular AAV injections were performed under general anesthesia induced by ketamine 80 mg/kg IP/xylazine 10 mg/kg IP, while electroretinography recordings were conducted under inhalational 2% isoflurane anesthesia. Mice undergoing AAV injections or ERG analyses were monitored for signs of distress or complications, and all efforts were made to minimize suffering throughout the study.

TRiC^et mice. The coding sequence of Mus musculus Tcp-1α subunit of chaperonin was generated by Genscript. A FLAG-HA epitope tag flanked by GGG linker on both sides and an internal VKL linker, GGGDYKDDDDKVKLYPYDVPDYAGGG was inserted between the amino acid residues R⁴⁸⁰ and K⁴⁸¹. This construct was used for transgene which contained a Kozak sequence at the 5’ end, a 4.4 kb mouse rhodopsin promoter [16], and a mouse protamine I polyadenylation sequence. The transgene was verified by sequencing and purified and injected into the pronuclei of zygotes from superovulated FVB females at West Virginia University. Transgene integration was confirmed by PCR genotyping of tail DNA using forward primer 5’- ACGACGATGACAAGGTCAAGCTCT and reverse primer 5’- ATCAATCCGAAGGATGGTGATT. Colonies were established by crossing transgenic heterozygotes with wild type partners of 129-E background (Charles River Laboratories).

PhLPs transgenic mice, rhodopsin P23H mice and Gγ₁ knockout mice were previously described [17–20]. Each strain was back-crossed with 129-E mice for at least ten generations. The resulting isogenic strains were crossed with each other, and their offspring was ear-tagged and genotyped using PCR amplification of genomic DNA as described in the original papers. Mice were housed in adjacent cages, under standard diurnal cycle. Equal numbers of males and females were used for the analyses.

TRiC^et pulldown and native PAGE

Frozen retinas (typically 20) or HEK293 cells pellet were homogenized in 0.8mL of 20 mM Hepes-HCl, pH 7.4, 100 mM KCl, 0.5 mM EDTA, 1% Triton X-100, and 10% glycerol by short ultrasonic pulses. The homogenate was cleared by centrifugation and the supernatant was incubated two times with 20 µL of anti-FLAG agarose beads (A2220-1 ML; Sigma-Aldrich) for 30 minutes at room temperature. The beads were combined and washed one time with 0.5mL of the above buffer and two times with 0.5mL of 10 mM Hepes-HCl, pH 7.4. The captured proteins were eluted two times with 15 µL of 10 mM Hepes-HCl, pH 7.4, 0.2 mg/mL 3x FLAG peptide (F4799-4MG; Sigma) for 15 minutes. The eluates were combined and purified by Blue Native PAGE on a Novex 3–12% tris-glycine gel (BN2011BX10, Fisher) using a XCell SureLock Mini-Cell Electrophoresis System (EI0001; Fisher), at room temperature, according to manufacturer’s protocol. Gels were stained with Coomassie and a 900 kDa band was excised from the gel. The protein composition of the band was analyzed by LC MS/MS. In some cases, the band was incubated in 5% β-mercaptoethanol and 0. 5mL of 125 mM Tris/HCl, pH 6.8, 4% SDS, and 6M urea for 20 minutes at 95°C, placed in the well of a polyacrylamide gel, and analyzed by SDS-PAGE and Western blotting.

Electroretinography (ERG)

Scotopic ERG responses were recorded and analyzed using either a Celeris (Diagnosis LLC) or an UTAS BigShot (LKC Technologies) rodent ERG system as previously described [21].

Microscopy

Immunofluorescence microscopy. Confocal images of frozen retinal cross-sections were acquired as previously described [22].

Light microscopy and automated photoreceptor nuclei counting. Eyes were collected and sent to Excalibur Pathology, Inc (Norman, OK, USA) for histology processing. Total nuclei count within the outer nuclear layer (ONL) was determined in continuous ocular cross-sections stained with hematoxylin and eosin. The specimens were digitized using an Olympus VS120 slide scanner, and the resulting TIFF images were analyzed using a custom Cell Profiler pipeline, as previously described [21].

Negative staining transmission electron microscopy (TEM). TRiC^et purified from HEK 293 cells was negatively stained with 1% uranyl acetate on carbon film coated copper grids. TEM images were obtained using a JEOL 1010 Transmission Electron Microscope equipped with an AMT Hamamatsu ORCA-HR Digital Camera. Particles were aligned and classified using SPIDER software image processing. Eyes were enucleated and immersed in fixative containing 2% paraformaldehyde, 2.5% glutaraldehyde, 100mM cacodylate buffer. After fixation, the eyes were transferred to a petri dish containing a drop of 7% sucrose buffer, and the cornea and lens were removed. The eyecups were then returned to the glass vial with the fixative for 48 hours with gentle rolling and then cut into smaller pieces. The pieces were washed five times for five minutes each with 0.1M cacodylate buffer at RT and then incubated with 2% osmium tetroxide in 0.1M cacodylate buffer for 1 hour on ice. After osmication, samples were washed with water and incubated with 1% aqueous uranyl acetate at 4°C overnight. The next day, after washes, samples were dehydrated in an ethanol series on ice. After dehydration, samples were washed with Propylene oxide 2 times for 15 minutes on a rotator at RT. Samples were then infiltrated with a 50:50 mixture of propylene oxide and resin for two hours at RT. Then, the solution was changed to 25:75 mixtures of propylene oxide and resin for overnight incubation at RT. The next day, the samples were transferred to fresh 100% resin and rotated for 2 hours at RT. Samples were then embedded in resin and cured in an oven at 60° C overnight. Ultrathin sections (70 nm) were cut using a LEICA EM UC7 ultramicrotome, post-stained with Reynold’s lead citrate, and imaged the following day on a JEOL JEM-1400 transmission electron microscope.

AAV vectors design, production, and subretinal injections

The coding sequences of mouse phosducin-like protein (Pdcl) or Gβ₁ (Gnb1) were synthesized by Genscript and inserted into pcDNA3.1 + . PhLPs-6x-His consisted of the cDNA sequence encoding Pdcl with DNA bases corresponding to amino acids 1–83 removed and a 6x-His tag added to the C-terminus. ^∆130-136 PhLPs-6x-His contains additional deletion of amino acids 130–136. Myc-Gβ₁ consisted of the cDNA sequence encoding Gnb1 with a myc-tag added to the N-terminus. Myc-Gβ1AA contained two alanine insertions after Leu7 and Ala11. Each construct was placed under the mouse opsin promoter (mOP), which limits the expression to rod photoreceptors, and packaged into AAV-PHP.eB capsid, as previously described [21]. The AAV production was conducted at the Viral Core of West Virginia University. The subretinal AAV injection was performed using 1–2-month-old 129-E mice, as previously described [21].

Tissue culture

Plasmids for the expression of human tcp-1α^et, PhLPs-6x-His, ^∆130−136 PhLPs-6x-His, myc-Gβ₁, myc-Gβ₁^AA and HA-Gγ₁ in mammalian cell culture were made by GenScript. HEK 293 cells were maintained in DMEM/F-12 medium supplemented with 10% FBS and 1% penicillin-streptomycin in an incubator at 37°C and 5% CO₂. For transient protein expression, cells were plated in six-well plates and grown to 80% confluency. The following day, cells were transfected with 2.5 μg of plasmid DNA using polyethylenimine (PEI) at a 1 (plasmid DNA): 2.5 (PEI) ratio. Empty pcDNA3.1 + vector was used as a control. Cells were collected 48h after transfection, frozen on dry ice, and stored at −80°C.

Antibodies

Proteins were detected using the following antibodies: rat anti-HA (Roche, 11867423001), rabbit anti-FLAG (Proteintech, 20543–1-AP), mouse anti-tcp-1β (Santa Cruz Biotechnology, sc-374152), rat anti-tcp-1α (Santa Cruz Biotechnology, sc-53454) mouse anti-6x-His tag (MA1–21315, Invitrogen).

Mass spectrometry

Epitope-tagging of TRiC in rod photoreceptors

To enable affinity purification of the chaperonin complex, an epitope tag was inserted into the coding sequence of the tcp-1α subunit at a protruding equatorial loop (Fig 1A). The epitope-tagged tcp-1α transgene, driven by a rod photoreceptor-specific promoter, was expressed in mice. Transgene expression was confirmed by Western blotting, either through anti-FLAG pulldown or direct probing of retinal extracts (Fig 1BC). The combined fluorescence of transgenic and endogenous tcp-1α bands in transgene-positive retinas matched the fluorescence of the tcp-1α band in wild-type retinas, indicating no significant overexpression in transgenic mice (Fig 1C).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Epitope-tagging of chaperonin in rod photoreceptors.

A. DYKDDDK (FLAG) and LYPYDVPDYA (HA) sequences inserted into tcp-1α (magenta) transgene expressed in rods of TRiC^et mice. Resulting epitope tags (et), illustrated on a TRiC^et cartoon, protruded from the loop indicated with arrow on side- and top TRiC view with color-coded subunits (PDB ID 6QB8). B. TRiC^et detected by anti-FLAG pulldown and Western blotting in the retinas of wild-type (WT) and TRiC^et mice. C. The level of the endogenous and epitope-tagged tcp-1α quantified by Western blotting (SEM, n = 11). Significance was determined by Mann-White rank sum test. D. TRiC^et visualized in retinal cross-section by immunofluorescence confocal microscopy with rat anti-HA (orange); PR-photoreceptors, OS-outer segments, IS-inner segments, ONL -outer nuclear layer, OPL-outer plexiform layer. Staining with mouse anti-rhodopsin (green) and DAPI (blue) are included on 60x panel as markers of rod outer segments and nuclei. No anti-HA staining was detected in wild-type retinas (not shown). E. Visual responses of dark-adapted rods were recorded using a Celeris (Diagnosis LLC) rodent ERG system. The amplitude of elicited ERG a-wave is shown as a function of flash intensity (SEM, n = 8). Responses of wild-type (black symbols) and TRiC^et (white symbols) mice were compared at two months of age. Significance was determined using t-test. Each data set was fitted with a simple rectangular hyperbola with two parameters using SigmaPlot 13 software.

https://doi.org/10.1371/journal.pone.0345843.g001

The transgene was uniformly expressed in rod photoreceptors and localized to all subcellular compartments except the outer segment (Fig 1D). Importantly, its expression had no adverse effect on rod visual function (Fig 1E). To confirm that the epitope-tagged tcp-1α assembled into TRiC, it was pulled down using anti-FLAG agarose, eluted under non-denaturing conditions, and analyzed by native gel electrophoresis (Fig 2A). Additionally, the tcp-1α construct used for transgene production was transiently expressed in HEK293 cells. Pulldowns from both the cell culture and retinal extracts ran as a ~ 900 kDa band, closely resembling TRiC isolated from the retinas of PhLPs mice. Negative-stain TEM further confirmed that the captured complex had the characteristic ring-shaped structure, similar to the endogenous TRiC captured by its co-factor PhLP (Fig 2BC). Based on these results, the generated transgenic strain was named TRiC^et.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Chaperonin complex runs on a Blue Native gel as the 900kDa band.

A. Anti-FLAG pulldown of TRiC^et under non-denaturing conditions separated on Blue Native Gel and stained with Coomassie. HEK293 cells were transfected with empty vector or tcp-1α^et vector; retinas were collected from wild type 129E mice, TRiC^et and PhLPs mice at 21 days of age. B. Negative stain TEM of anti-FLAG pulldown from HEK293 cells transfected with PhLP-FLAG showing ring-shaped particles that were absent in the pulldown from HEK293 cells transfected with empty vector (not shown). C. Global average image of 513 particle images aligned showing the most common top view TRiC orientation with eight subunits visible (top), and the same data classified into 30 classes (bottom).

https://doi.org/10.1371/journal.pone.0345843.g002

TRiC loss-of-function model

Phosducin-like protein (PhLP, gene: Pdcl1) is a co-chaperone that facilitates the folding of Gβ and binds chaperonin with high affinity [33]. Previously, we found that the short splice isoform of PhLP, PhLPs, which lacks the Gβ-binding domain, induces rod outer segment breakdown and cell death [17,18,34]. We proposed that PhLPs causes this phenotype by inhibiting the folding of nascent proteins by chaperonin.

To further support this mechanism, we deleted amino acids 130–136 of PhLPs, a region implicated in TRiC binding (Fig 3A) [35]. This mutant, termed ^∆130-136PhLPs-6x-His, and the control PhLPs-6x-His were co-expressed with epitope-tagged Tcp-1α in HEK293 cells, followed by TRiC^et pulldown assays. Western blot analysis of the pulldowns, both directly and after isolating the 900 kDa TRiC^et band by native gel electrophoresis (Fig 3B), showed that PhLPs-6x-His co-precipitated with TRiC^et more efficiently than ^∆130-136PhLPs-6x-His. Importantly, the expression levels of the transfected proteins and endogenous chaperonin were comparable. These results confirm that the ∆130–136 deletion reduces PhLPs binding to chaperonin.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. PhLPs binding to TRiC leads to rod photoreceptors death.

A. Phosducin-like protein (gene: Pdcl; protein: PhLP) and its epitope-tagged short splice isoform PhLPs-6x-His and ^∆130-136PhLPs-6x-His mutant. B. Western blot analysis of HEK293 cells co-transfected with tcp-1α^et and PhLPs-6x-His or ^∆130-136PhLPs-6x-His: anti-FLAG pulldown (left), whole cell extract (center), the 900kDa band from Blue Native Gel of anti-FLAG pulldown (right). C. Comparison of visual responses of mice that received subretinal injection of AAV-PhLPs-6x-His (white circles) and AAV -^∆130-136PhLPs-6x-His (black circles) 2-months post-injection. Electroretinographic responses were recorded using a UTAS BigShot (LKC Technologies) rodent ERG system. The amplitude of elicited ERG a-wave is plotted as a function of flash intensity (SEM, n = 4). Each dataset was fitted with a simple rectangular hyperbola with two parameters. The significance was determined by t-test and Mann-White rank sum test, with P value <0.05 (*) and <0.01 (**). D. Ultrastructure of rod photoreceptors of mice in C analyzed by TEM: outer segments (OS), inner segments (IS), outer nuclear layer (ONL). E. Ocular cross-section stained with hematoxylin and eosin of wild type 129E mice at 14 days of age, and PhLPs mice at 14 and 18 days of age. ONL: outer nuclear layer containing stacked photoreceptors’ nuclei. F. ONL nuclei counts in the entire retinal cross-sections measured by ONLyzer: bars are mean value with SD, P-value determined by t-test, n = 31(WT), 22 (PhLPs, 14 days-old), 28 (PhLPs, 18 days-old). G. Ultrastructure of rod photoreceptors at 14 days of age analyzed by TEM: outer segments (OS), inner segments (IS).

https://doi.org/10.1371/journal.pone.0345843.g003

Next, we compared the in vivo effects of PhLPs-6x-His and ^∆130-136PhLPs-6x-His using an AAV delivery system. The amplitude of the a-wave recorded in dark-adapted mice via electroretinography served as a measure of total number of rods in the retina and phototransduction efficiency in their outer segments (Fig 3C). Rod ultrastructure was analyzed by TEM (Fig. 3D). Mice treated with AAV: ^∆130-136PhLPs-6x-His exhibited significantly larger a-wave amplitudes, retained more rod nuclei in the outer nuclear layer (ONL), and had longer rod outer segments than those treated with AAV: PhLPs-6x-His. These findings suggest that the toxicity of PhLPs is attenuated when TRiC binding is disrupted, supporting the hypothesis that TRiC is a primary target of PhLPs in rods.

To determine the onset of rod photoreceptor degeneration, we analyzed transgenic PhLPs mice, which begin expressing PhLPs-FLAG in rods at postnatal day 8. The total number of photoreceptor nuclei within the ONL was quantified using hematoxylin and eosin-stained paraffin-embedded retinal cross-sections (Fig 3EF). PhLPs-expressing mice and wild-type controls of the same genetic background were compared at postnatal days 14 and 18. At day 14, 95% of PhLPs-expressing photoreceptors were still present, but this percentage declined to 74% by day 18. Additionally, PhLPs mice displayed shortened rod outer and inner segments as early as day 14 (Fig 3G).

These results further support the role of PhLPs in disrupting chaperonin-mediated protein folding, leading to rod photoreceptor degeneration. Importantly, our findings suggest that postnatal day 14 and possibly 18 represent key time points when TRiC function is already fully suppressed by PhLPs, yet rods have not undergone widespread cell death. This provides an opportunity to analyze proteome changes associated with TRiC loss-of-function before major degeneration occurs in conjunction with the analysis of the retina specific TRiC^et interactome.

Identification of proteins interacting with TRiC^et in mouse retina

To identify proteins interacting with TRiC^et in rod photoreceptors, this protein complex was captured using an anti-FLAG pulldown, separated as a 900 kDa band on native gel electrophoresis, and analyzed by LC-MS/MS. Pulldowns from age-matched wild-type retinas processed under identical conditions served as the baseline. The abundance of identified proteins was expressed as the fold change in TRiC^et versus wild-type samples. To account for potential instrumentation bias, two independent datasets from 2 orthogonal approaches, DDA and DIA were generated at the proteomic facilities of Duke University and the IDeA Proteomic Facility at the University of Arkansas respectively.

In both datasets, the tcp-1 subunits of chaperonin were identified with high confidence, reinforcing that the 900 kDa band represented an intact chaperonin complex. Additionally, the Duke and IDeA datasets contained 131 and 109 proteins, respectively, that were enriched by more than twofold (S1 Table). The overlap between these datasets was limited, with only 14 shared proteins, six of which were known TRiC substrates (Gnb1, Gnb2, Gnb5, Hdac1) or TRiC co-chaperones (Pdcl, Pdcl3) (Fig 4AB). However, each dataset included additional known TRiC interactors that were not detected by any specific data filter. Consequently, proteins meeting the method specific enrichment criteria in either DDA or DIA were merged into a single dataframe containing 226 protein candidate TRiC associated interactome for subsequent pathway and hypothesis generating analyses (S2 Table). For both acquisition strategies, candidate TRiC interactors were defined using a twofold enrichment threshold, log2FC(TRiCet/WT) ≥1, together with the method specific multiple testing criterion described in the Methods. Because DDA and DIA have complementary detection properties, and because TRiC–client interactions are expected to be transient and often substoichiometric, we considered proteins meeting significance criteria in either acquisition strategy as candidate TRiC-associated proteins. The merged dataset was therefore used as a discovery level interactome rather than as a list of validated direct substrates. Direct substrate assignment was made only when supported by additional evidence, such as known TRiC biology, loss of function sensitivity, or altered association in the PhLPs-TRiC or Gγ1-deficient models

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Analysis of TRiC^et interactome in mouse retina.

A. Volcano plots displaying differential proteins of the 900 kDa band in preparations from TRiC^et and wild type retinas. The x-axis represents the log₂ fold change (log₂FC) and the y-axis shows (–log₁₀(p-value), with adjusted p < 0.05 and |log₂FC| > 1 corresponding to ≥2-fold linear enrichment, inclusion criteria for all identified proteins. The chaperonin subunits were removed from the plot. Shown datasets were independently obtained at the proteomic facilities of Duke University and the IDeA Proteomic Facility at the University of Arkansas. B. Venn diagrams illustrating the overlap between the Duke and IDeA datasets (top) and previously published TRiC interactors (bottom) with gene names tabulated. C. Distribution of molecular weights in across both mass spec datasets of TRiC^et interactors. The 60 kDa line indicates the theoretical size limit of the TRiC folding chamber. D. Domain architecture frequency analysis in the combined dataset of TRiC^et interactors using InterPro/Pfam searches. E. Heatmap illustrating the enrichment comparison of both KEGG and Reactome pathways in the Duke and IDeA datasets as determined by Metascape enrichment meta-analysis. The pathway significance is encoded using a gradient based on –log₁₀(p-value), highlighting statistically significant pathways associated with TRiC interactors gray rows indicate no enrichment found. F. Protein–protein interaction networks identified in the combined dataset of TRiC^et interactors and the related biological process.

https://doi.org/10.1371/journal.pone.0345843.g004

The structural and physicochemical properties of these 226 TRiC^et-interacting proteins were examined, revealing that approximately 60% (130 out of 226) had molecular weights below 60 kDa (Fig. 4C), aligning with the size constraints of TRiC’s folding chamber. The most common structural features included RNA recognition motifs and beta-propeller folds (Fig 4D). Many identified proteins were found to participate in mRNA processing and regulation of cell cycle (Fig 4EF). We concluded that, in addition to bona fide TRiC substrates encapsulated in the folding cavity, our dataset includes proteins that regulate TRiC activity. To determine which proteins are directly folded by TRiC, we analyzed changes in steady-state protein levels in the retina in response to PhLPs transgene expression, which suppresses TRiC-mediated protein folding.

Changes in protein levels in the retina of PhLPs mice

Retinas of PhLPs mice were collected at postnatal days P14 and P18, and steady-state protein levels were compared to those in age-matched wild-type mice using isobaric labeling with tandem mass tags TMT-MS (S3 Table). At P14, 24 proteins were significantly reduced in PhLPs retinas, including 11 TRiC^et interactors (Fig 5AB). The largest group consisted of tubulin family proteins (Tubb5, Tubb4a, Tubb2a, Tubb2b, Tubb4b, and Tuba4a). Given that alpha- and beta-tubulins are well-established TRiC substrates, this finding supports the notion that PhLPs inhibit TRiC-mediated tubulin folding, leading to its accelerated proteolytic degradation. This observation provides a compelling explanation for the severe underdevelopment of the rod outer segment in PhLPs mice. The rod outer segment, a microtubule-based compartment, undergoes rapid growth at P14, and its development is arrested when TRiC-mediated tubulin folding is disrupted.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Changes of retina proteome caused by the PhLPs transgene.

A. Volcano plots showing differential protein levels in the retina of PhLPs mice vs. wild type controls 14- and 18-days after birth (P14 and P18, respectively). The x-axis represents the log₂ fold change (log₂FC) and the y-axis shows (–log₁₀(p-value), with adjusted p < 0.05 and absolute linear fold change > 1.5 corresponding to log₂FC ≥ 0.59 inclusion criteria. The data were collected at the IDeA Proteomic Facility at the University of Arkansas. B. Proteins decreased in the retina of PhLPs mice grouped and color-coded based on their biological function. The identified TRiC^et interactors (Sup. Table II) are in bold font. C. Heatmap of differentially expressed proteins (adjusted p < 0.05) across the entire PhLPs dataset (left) and the identified TRiC^et interactors (right), with z-scores indicating the relative expression (red: upregulated; blue: downregulated). D. Proteins increased in the retina of PhLPs mice grouped and color-coded based on their biological function. The identified TRiC^et interactors are in bold font. E. Enrichment of GO biological processes in the PhLPs dataset assessed by GSEA using ClusterProfiler. Each dot represents a GO term, with its position on the x-axis reflecting the normalized enrichment score (NES), positive NES values denote processes up-regulated/activated in the differential protein set, while negative NES values denote down-regulated/suppressed processes. Dot size corresponds to the number of proteins contributing to that term, and color shading indicates statistical significance (adjusted p < 0.05).

https://doi.org/10.1371/journal.pone.0345843.g005

Another confirmed TRiC^et interactor that was reduced in PhLPs retinas was thioredoxin domain-containing protein 9 (Txndc9), a homolog of phosducin-like protein. Txndc9 has been proposed to negatively regulate TRiC ATPase activity and the folding of nascent tubulin and actin [36], which may explain its reduced levels in rods experiencing tubulin deficiency. Three additional TRiC^et-interacting proteins, WD and tetratricopeptide repeat-containing protein 1 (Wdtc1), uridine-cytidine kinase-like 1 (Uckl1), and triosephosphate isomerase (Tpi1, were cytosolic proteins under 80 kDa, suggesting they may also serve as TRiC substrates. Additionally, Wdtc1, which has been proposed to act as a substrate receptor for CRL4 E3 ligase complex mediating ubiquitin-dependent protein degradation [37], may play a role in regulating TRiC turnover. A novel TRiC partner identified in this study was ATRX, a 279 kDa transcriptional regulator involved in chromatin remodeling and transcriptional control.

Interestingly, several non-TRiC interactors that were notably reduced at P14 clustered around Ca² ⁺ signaling and homeostasis. These included neuroplastin (Nptn), a chaperone of plasma membrane Ca² ⁺ ATPase; Ca² ⁺ -dependent protein phosphatase (Ppp3cc); the metallopeptidase Lap3; and annexin A1 (Anxa1), a regulator of phagosome-actin cytoskeleton interactions. Another downregulated group consisted of proteins involved in intracellular membrane trafficking (e.g., Rab10, Eipr1, Snap25, Rab15) and actin cytoskeleton regulation (Tln1, Cdc42, Flna). Additionally, we observed a reduction in Hacd3, an enzyme catalyzing long-chain fatty acid elongation in the endoplasmic reticulum, and Atp1a3, a Na ⁺ /K ⁺ -ATPase essential for maintaining membrane potential in visual signaling.

By P18, the number of proteins reduced in PhLPs retinas increased to 53 (Fig 5AB). However, only eight of these proteins were confirmed TRiC^et interactors, including tubulin (Tuba4a) and triosephosphate isomerase (Tpi), whose reductions were also observed at P14. The observed reduction of rod-specific transducin beta subunit (Gnb1) supports previous findings that Gnb1 is a major TRiC substrate affected by PhLPs[17]. Additionally, heterogeneous nuclear ribonucleoprotein A1 (Hnrnpa1) displayed characteristics of a potential TRiC substrate. However, this was deemed unlikely for two membrane proteins (Vdac1, Vdac2), as TRiC primarily folds soluble substrates.

Interestingly, the destabilization of two potential TRiC substrates, Uckl1 (involved in the pyrimidine salvage pathway) and Tpi1 (a key enzyme in glycolysis), triggered widespread effects on protein networks related to glucose metabolism and energy production. As a result, several non-TRiC-interacting proteins (e.g., Ldhb, Pfkl, Gapdh, Pkm, Pgk1, Hk1, Eno3, Tkt, Aldoc, Cox4i1, Cyc1, Mdh2, Slc25a12, Uqcrq, and Immt) were decreased at P18 (Fig 5B). Similarly, disruption of tubulin folding led to secondary effects on cytoskeletal proteins (e.g., Cap2, Krt73, Epb41l2, Flna), membrane trafficking proteins (e.g., Rab10, Rab11b, Rab11a, Rab15, Snap25, Anxa1), and phototransduction proteins involved in visual signaling in the outer segment (e.g., Gnat1, Gucy2e, Gucy2f, Pde6a, Rho, Sag, Atp1a3). However, none of these proteins appeared to directly interact with TRiC^et. These findings suggest that the observed changes in the retinal proteome at P18 increasingly reflect secondary effects stemming from the destabilization of TRiC substrates at P14.

In addition, we observed adaptive responses resulting from the suppression of protein folding by TRiC. This category included proteins whose levels in the retina increased in response to the PhLPs transgene (Fig 5CD). For example, at P14, the levels of 19 proteins were significantly elevated in the retinas of PhLPs mice. Among these, four were confirmed TRiC interactors, including protein involved in cytoskeletal maintenance (Map6), a DNA-binding regulator of telomerase (Ylpm1), the rod-specific transducin beta subunit (Gnb1), and phosducin-like protein (Pdcl).

When categorizing all upregulated proteins based on biological function, they clustered into two major groups: proteins involved in maintaining proteostasis (e.g., Psmc5, Psmd6, Psmd8, Psmd11, Psmd12, Psmd14, Hspa1a, and Hspa1b) and regulators of cytoskeletal dynamics (e.g., Coro2b, Lima1, Tmod2, Ina, Map6, Dock7, and Ank3). This pattern suggests that the disruption of proteostasis and cytoskeletal dynamics may be the primary triggers of the observed responses at P14. The increased Pdcl levels were attributed to the expression of a truncated Pdcl protein by the PhLPs transgene. Additionally, the upregulation of Gnb1 was consistent with PhLPs forming a stable tertiary complex with TRiC preloaded with nascent Gβ₁ [34].

By P18, the levels of 34 proteins, including five confirmed TRiC^et interactors, were significantly elevated (Fig 5D). The most upregulated proteins were the G-protein beta subunits Gnb2 and Gnb3, neither of which is typically expressed in rods at significant levels. These subunits closely resemble Gnb1, a component of rod-specific transducin, a heterotrimeric G protein involved in visual signaling. Given that PhLPs inhibits the TRiC-mediated folding of nascent Gnb1, it is plausible that the upregulation of Gnb2 and Gnb3 represents a compensatory response to transducin deficiency.

Additional upregulated TRiC^et interactors included histone-binding and chromatin remodeling proteins (Hdgfl2 and Hdac1) and the DNA-binding telomerase regulator Ylpm1. When analyzed as a whole, the upregulated proteins at P18 fell into four major functional groups. The largest group comprised transcriptional regulators and histones (e.g., Macroh2a2, Hdgfl2, H2ac4, H2ac6, Eif2b3, H2ac7, H2ac8, H2ac11, H2ac13, Hist1h2an, Hist1h2ao, Hist1h2ap, H2ac25, Hist1h2af, H2ac12, H2ac15, H2aj, H2ac1, Ddx3y, Ylpm1, Hnrnpul1, Eif2b3, and Hdac1). Three smaller groups included proteins involved in cytoskeletal regulation (e.g., Radil, Macf1, Dock7, and Camk2b), proteostasis maintenance (Psmd6 and Psmd13), and mitochondrial function (Pgam5 and Atp5f1e). Additional upregulated proteins included Prpsap1, which is involved in nucleotide biosynthesis, and Osbpl5, which participates in lipid transport. These findings suggest that by P18, the primary adaptive response of rod cells to the PhLPs transgene involves transcriptional regulation and chromatin remodeling.

Gene set enrichment analysis (GSEA) of the entire dataset revealed distinct effects of the PhLPs transgene at different time points after its expression. By P14, PhLPs predominantly suppressed recycling endosome, a subcompartment of the Golgi apparatus, and mitochondrial membrane while activating the proteasome and peptidase complexes (Fig 5E). At P18, the most strongly suppressed components included mitochondria, photoreceptor outer segment, and the cilium, whereas activation extended beyond the proteasome to include the translation initiation complex, nucleosome, DNA packaging complex and chromatin.

Analysis of proteins within the PhLPs-TRiC complex

We next examined how PhLPs influence the protein interactions of TRiC. To do this, we isolated the PhLPs-TRiC complex via FLAG-tagged PhLPs and analyzed its protein composition using LC-MS/MS (S4 Table). The same protocol was applied as for TRiC^et, with the analysis conducted at the proteomics facility of Duke University. Data from eight pulldowns performed on retinas from PhLPs and wild-type controls were averaged, and significantly enriched proteins that were also present in TRiC^et pulldowns were identified. To compare protein abundance, the mean intensity of the eight TRiC subunits in each dataset was calculated, and the intensity of each target protein was normalized to this value. The resulting normalized mean intensities were used to identify proteins that were significantly different between the PhLPs-TRiC and TRiC^et complexes (Fig 6).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. PhLPs transgene suppresses protein interactions of TRiC^et.

A. Heatmap of relative protein abundance in the 900 kDa band in isolations from PhLPs and TRiC^et retinas, with z-scores indicating the relative normalized amount of protein within the PhLPs-TRiC complex compared to the TRiC^et complex (red: increased; blue: decreased). B. Identified proteins (genes) ranked based on the numerical values of the log₂ fold change (log2FC) color-coded as red: increased, blue: decreased with adjusted p < 0.05. The eight chaperonin subunits are in green.

https://doi.org/10.1371/journal.pone.0345843.g006

In addition to the eight core subunits of the TRiC complex, 60 proteins were significantly enriched in both the PhLPs-TRiC and TRiC^et datasets. However, only two proteins—Pdcl and Plekhb1—were more abundant in the PhLPs-TRiC complex, while all other TRiC interactors were markedly reduced. The increase in Pdcl was expected, given that PhLPs is a splice isoform of Pdcl. The second enriched protein, pleckstrin homology domain-containing family B member 1 (gene: Plekhb1, protein: PKHB1), was increased by ninefold. Plekhb1 is known to be expressed in ciliated sensory neurons that utilize heterotrimeric G protein signaling, such as rod photoreceptors, though its specific function remains unclear [38–40]. Plekhb1 is an integral membrane protein that is not folded by TRiC. Interestingly, it has been shown to interact with the transducin Gβ₁γ₁ dimer [39], which is folded and assembled by TRiC and Pdcl [41,42]. This suggests that Plekhb1 may play a role in regulating Gβγ biosynthesis.

All other identified proteins, including known and potential TRiC substrates such as Actb, Hdac1, Psmd14, Tubb5, Gnb1, and Gnb5, were significantly reduced in the PhLPs-TRiC complex. This finding provides further evidence that PhLPs suppresses TRiC-mediated protein folding in rod photoreceptors.

Changes of metabolic profiles in the retinas of PhLPs mice

Many TRiC interactors identified in our proteomic screen were enzymes, including two enzymes involved in glucose and nucleotide metabolism − Tpi1 and Uckl1 − whose levels were significantly reduced in the retinas of PhLPs mice at postnatal day 14 (Fig 5B). Given the central role of TRiC in proteostasis, we hypothesized that its deficiency might extend beyond protein folding to broader metabolic disruptions in rod photoreceptors. To test this, we performed targeted metabolomics covering major metabolic pathways including glucose, nucleotide, amino acids and fatty acid metabolism [30]. In addition to PhLPs mice (treated group) and wild type 129E mice (control group), we included another group as disease control, composed of homozygous knock-in mice carrying the P23H mutation in the rhodopsin gene (Rho^P23H/P23H). Both PhLPs and Rho P23H mice undergo progressive rod degeneration with similar onset and severity, but the underlying pathogenic mechanisms differ. Thus, the Rho P23H group served as a disease control to distinguish metabolic changes specifically attributable to TRiC dysfunction from those more generally associated with retinal degeneration.

Multivariate analysis of metabolomics data with PLSDA clearly separated treated from control retinas, demonstrating the overall difference in retinal metabolome between the control and treated groups (Fig 7A). Across 155 metabolites reliably detected in the retina, 37 and 51 showed statistically significant differences in the PhLPs and Rho P23H groups, respectively, compared to wild type controls (Fig 7BC, S5 Table). In PhLPs retinas, metabolic disruption was particularly evident in energy homeostasis and mitochondrial pathways (Fig 7D). We detected reductions in ATP, NAD, and NADH, accompanied by decreases in glycolytic intermediates (glucose, glucose 6-phosphate, 2-phosphoglyceric acid), TCA cycle metabolites (fumarate, α-ketoglutarate), and multiple acylcarnitines (octanoyl-, propionyl-, oleoyl-, isobutyryl-, and butyrylcarnitine). Together, these patterns suggest impaired glycolytic flux, reduced oxidative phosphorylation, and defective mitochondrial fatty acid oxidation. The loss of acylcarnitines, in particular, points to an underutilization of fatty acids as energy substrates, which may compromise the high bioenergetic demands of photoreceptors.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Metabolic changes in the retina of PhLPs and Rho P23H mice 14-days after birth.

A. Partial Least Squares Discriminant Analysis (PLS-DA) scores plot illustrating group separation among WT, PhLPs, and Rho P23H retinas. Axes correspond to the first two latent variables (LV1 and LV2; % variance in Y explained shown on each axis). Ellipses indicate the 95% confidence regions for each group. B. Volcano plots of detected metabolites comparing either PhLPs or RhoP23H retinas to wild type 129E controls (n = 4 mice per group, 8 retinas analyzed independently). The x-axis shows log2 fold change (log2FC); the y-axis shows –log10(p- value). The horizontal dashed line marks p = 0.05 (–log10p = 1.3). Significance was assessed by nominal p-value from two-tailed, two-sample t-tests assuming equal variance (p < 0.05). C. Heatmaps of the significantly altered metabolites for each comparison (left: PhLPs vs WT; right: RhoP23H vs WT). Rows represent metabolites, columns represent individual retinas. Values are row-scaled Z-scores, and both rows and columns were hierarchically clustered to emphasize tissue-specific patterns. D. Ranked plot of nominally altered metabolites selected using p < 0.05 and |log2FC| > 0.4. The plot was generated in GraphPad Prism from exported fold-change and nominal p-value tables. Color-coded as blue: decreased, red: increased.

https://doi.org/10.1371/journal.pone.0345843.g007

Beyond energy metabolism, TRiC deficiency also impacted NAD⁺ signaling pathways. Reduced cyclic ADP-ribose implies dysregulation of NAD⁺ metabolism, possibly reflecting overactivation of PARPs in response to cellular stress, further depleting NAD⁺ pools. This could exacerbate the energy crisis by limiting redox balance and sirtuin-mediated metabolic regulation. The decline in acetylglutamic acid, a critical urea cycle cofactor, suggests additional disturbances in nitrogen disposal and amino acid catabolism. Interestingly, these TRiC-dependent metabolic signatures were not mirrored in the Rho P23H model, despite comparable photoreceptor degeneration. Instead, Rho P23H retinas showed elevated levels of amino acids and methylation-related intermediates (Fig. 7D), pointing toward compensatory anabolic or stress-response pathways rather than energy collapse. In addition, deregulation of retinal cGMP level commonly observed in retinal degeneration models was readily detectable in Rho P23H group, but not in PhLPs group (Fig 7D).

Taken together, these results suggest that TRiC loss leads to a metabolic “energy crisis” in photoreceptors, characterized by collapse of glycolysis, TCA cycle activity, and fatty acid oxidation. This contrasts with the metabolic remodeling seen in Rho P23H retinas, underscoring the unique role of TRiC in safeguarding energy metabolism. We speculate that impaired proteostasis in PhLPs retinas may directly interfere with the folding and stability of key metabolic enzymes, thereby coupling chaperone failure to energetic collapse. Such TRiC-linked metabolic vulnerability could represent a previously unrecognized driver of photoreceptor degeneration.

Integrated Proteomic–Metabolomic analysis identifies TRiC associated molecular signatures of metabolic dysregulation in PhLPs retinas

While our proteomic and metabolomic analyses independently revealed disruptions in protein homeostasis and energy metabolism, these layers of data only provide partial perspectives. To capture a more complete picture, we next performed an integrated analysis, combining proteomic and metabolomic datasets to identify protein candidates that covary with the observed metabolite alterations using DIABLO. Such integrative approaches are particularly powerful because proteomic changes do not always translate linearly into metabolic outcomes, and metabolites can be influenced by both direct enzymatic regulation and broader cellular stress responses. By linking these data types, we aimed to identify the molecular nodes where TRiC dysfunction intersects with metabolic collapse in rod photoreceptors.

For this analysis, proteomic features were selected using the same global TMT-MS differential abundance threshold used, adjusted p-value <0.05 and absolute linear fold change ≥1.5, equivalent to |log2FC| ≥ 0.59. Metabolomic features were prioritized when they were altered in PhLPs retinas and showed opposite directional behavior compared with the Rho P23H degeneration-control group across five randomly selected replicates, thereby enriching for metabolic changes associated with TRiC dysfunction rather than nonspecific retinal degeneration. This strategy enabled us to disentangle TRiC-specific metabolic alterations from generic degeneration-associated changes. The integration revealed a highly specific set of ten proteins (Rab10, Anxa1, Tuba1a, Txndc9, Uckl1, Wdtc1, Tubb4a, Hspa1a, Hspa1b) that co-varied with ten metabolites in the PhPLs group (Fig 8, S6 Table). Among these protein–metabolite pairs, two metabolites glucose and aminoadipic acid emerged as strongly driven by TRiC deficiency. Notably, glucose levels correlated with Rab10 abundance. Because Rab10 has been implicated in GLUT4 trafficking in other cell types [42], this association raises the possibility that altered Rab10-dependent trafficking could contribute to impaired glucose handling in rods [43]. However, this model remains speculative and will require direct experimental validation, our analysis was used as an exploratory multi-block integration approach to identify coordinated protein–metabolite signatures associated with TRiC dysfunction. In contrast, the relationship between Anxa1 and aminoadipic acid was less clear, lacking a known biochemical or signaling connection, but may suggest unexplored roles for annexins in amino acid homeostasis.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. DIABLO-based integration uncovers TRiC-dependent molecular signatures in PhLPs retinas 14-days after birth.

Input proteomic features were prefiltered using adjusted p-value <0.05 and absolute linear fold change ≥1.5, equivalent to |log2FC| ≥ 0.59. DIABLO was then used to identify correlated multi-omic signatures between the proteomic and metabolomic blocks. A. Multiblock sparse PLS-DA (sPLS-DA) on proteomics and metabolomics datasets. Each sample is plotted on its first latent component for each block; points are colored by condition (wild type 129E control vs PhLPs). Ellipses denote 95% confidence regions. The correlation between the first components of the two blocks is shown (bottom left; r = 0.97). B. Loadings plot highlighting the top proteins and metabolites (highest absolute loading weights) that discriminate wild type 129E control and PhLPs groups in the DIABLO model. Loadings represent model-selected discriminatory features and should not be interpreted as independent differential-abundance statistics. C. Clustered Image Map (CIM) of the selected features across both omic blocks, proteomics and metabolomics. Cell colors represent pairwise correlations between features from DIABLO analysis (red = positive, blue = negative|). Variables with mean loading weight ≈ 0 were omitted.

https://doi.org/10.1371/journal.pone.0345843.g008

This integrated omics analysis highlights how TRiC dysfunction triggers a cascade of metabolic stress centered on glucose utilization and mitochondrial bioenergetics. Importantly, this signature was not recapitulated in Rho P23H retinas, where metabolic changes likely reflect downstream degenerative processes rather than primary energetic defects. Thus, integration of proteomic and metabolomic data helped identify candidate molecular links, including Rab10, that may connect impaired TRiC-mediated proteostasis with metabolic dysfunction. This analysis does not establish causality between individual proteins and metabolites. Rather, it prioritizes candidate molecular nodes whose abundance covaries with the TRiC-dependent metabolic phenotype and might guide future mechanistic studies. Accordingly, we interpret Rab10, Anxa1, Uckl1, and related features as candidate contributors to, or markers of, the metabolic remodeling observed in PhLPs retinas. Collectively, these findings support a model in which TRiC dysfunction may compromise photoreceptor metabolism by affecting trafficking and enzymatic proteins, while the specific role of Rab10-dependent GLUT4 trafficking remains to be tested experimentally.

Effect of substrate dominance on TRiC-mediated protein folding

To evaluate the in vivo effect of a dominant TRiC substrate, we generated a perpetually unfolded mutant of Gβ₁. The final step in the folding of nascent Gβ₁ involves dimerization with the γ subunit [44]. To disrupt this process, we inserted alanine residues at positions 8 and 12 of myc-tagged Gβ₁, disrupting the coiled-coil interaction with Gγ [45] (Fig 9A). This construct, termed myc-Gβ₁^AA, was co-expressed with HA-tagged Gγ₁ in HEK293 cells, with wild-type myc-Gβ₁ serving as a control. Both proteins were readily detectable in whole cell extracts; however, only myc-Gβ₁ was present in HA pulldown assays, indicating that myc-Gβ₁^AA failed to bind HA-Gγ₁ (Fig 9B). We next compared the ability of myc-Gβ₁ and myc-Gβ₁^AA to interact with TRiC in HEK293 cells co-transfected with tcp-1α^et. The 900 kDa TRiC complex was isolated via FLAG pulldown followed by blue native gel electrophoresis, excised, denatured with SDS-containing buffer, and analyzed by Western blotting using anti-myc antibodies (Fig 9C). We found that myc-Gβ₁^AA was enriched ~1.8-fold compared to wild-type myc-Gβ₁ in TRiC^et complexes (Fig 9D), suggesting that the mutant persists in an unfolded state and preferentially binds TRiC (Fig 9E).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 9. Perpetually unfolded Gβ₁ mutant increasingly binds TRiC and is cytotoxic for rods.

A. Epitope-tagged rod transducin β subunit (gene: Gnb1; protein: Gβ₁), myc-Gβ₁, and its mutant with Ala⁸ and Ala¹² insertions, myc-Gβ₁^AA. B. Western blot analysis of HEK293 cells co-transfected with epitope-tagged rod transducin γ subunit (gene: Gng1; protein: Gγ₁) designated as HA-Gγ₁ and myc-Gβ₁ or myc-Gβ₁^AA: whole cell extract (left), anti-HA pulldown (right). C. Western blot analysis of the 900 kDa band from Blue Native Gel. Anti-FLAG pulldown was conducted in HEK293 cells co-transfected with tcp-1α^et, HA-Gγ₁ and myc-Gβ₁ or myc-Gβ₁^AA. D. Normalized amounts of myc-Gβ₁ and myc-Gβ₁^AA in the 900 kDa band; bars are mean value with SE, P-value determined by t-test, n = 3. E. Cartoon of Gγ₁ completing the TRiC-mediated folding of Gβ₁, disrupted in Gβ₁^AA mutant. F. Western blot analysis of anti-myc pulldown from the retina. Wild type 129E mice received subretinal injection of AAV- myc-Gβ₁ and AAV – myc-Gβ₁^AA and their retinas collected and analyzed 1-month post-injection. G. Comparison of visual responses of mice that received subretinal injection of AAV – myc-Gβ₁^AA (white circles) and AAV- myc-Gβ₁ (black circles) 4-months post-injection. Electroretinographic responses were recorded using a UTAS BigShot (LKC Technologies) rodent ERG system. The amplitude of elicited ERG a-wave is plotted as a function of flash intensity (SEM, n = 5 (myc-Gβ₁) n = 6 (myc-Gβ₁^AA)). Each dataset was fitted with a simple rectangular hyperbola with two parameters. The significance was determined by t-test and Mann-White rank sum test, with P value <0.05 (*) and <0.01 (**).

https://doi.org/10.1371/journal.pone.0345843.g009

To assess its effect in vivo, we delivered myc-Gβ₁^AA to mouse rod photoreceptors via subretinal injection of AAV vectors, using myc-Gβ₁ as control. One-month post-injection, myc signal levels were compared. Because myc-Gβ₁^AA was barely detectable in whole retina extracts, we used myc pulldown to confirm expression, which was substantially lower than that of wild-type myc-Gβ₁ (Fig 9F). Longitudinal electroretinography showed a significant reduction in a-wave amplitude in mice treated with AAV: myc-Gβ₁^AA at 4 months post-injection, indicating rod dysfunction (Fig 9G). Thus, the myc-Gβ₁^AA mutant, unable to bind Gγ₁, was less stable and more cytotoxic in rod photoreceptors than wild-type Gβ₁. Next, we sought to determine how unfolded Gβ₁ influences TRiC protein interactions. Because AAV vectors result in patchy transduction in mouse rods, they were not suitable for proteomic analysis. Instead, we crossed TRiC^et mice with Gγ₁^-/- mice [19]. Given that Gγ₁ is the predominant γ subunit in mouse rods, its absence is functionally equivalent to the AA mutation in producing unfolded Gβ₁. As reported for AAV: myc-Gβ1AA-treated rods, Gγ₁^-/- photoreceptors undergo progressive degeneration [19,46]. We used this genetic model to analyze TRiC interactions by LC-MS/MS using our standard approach. Data from six TRiC^et pulldowns in Gγ₁^-/- and wild-type retinas were averaged, and significantly enriched proteins identified. The same analysis was performed for TRiC^et vs. wild-type samples. For normalization, we calculated the mean intensity of the eight TRiC subunits in each dataset and used this value to normalize the intensity of each target protein. These normalized average intensities were used to identify significant differences in TRiC^et; Gγ₁^-/- vs. TRiC^et datasets (Fig 10A, S7 Table). Using this approach, we identified 10 proteins with increased association, including four previously validated TRiC^et interactors (Cenpv, Pdcd5, Gnb1, and Pdcl) and six new interactors (Gfap, Ica, Tprl, Hnrnpm, Syne1, and Actr1b (Fig 10C). Notably, Gβ₁ (Gnb1) showed a 2.2-fold increase, consistent with the enrichment observed for myc-Gβ₁^AA (Fig 9D), indicating its unfthey were not suitable folded state. The increased TRiC-Gβ₁ interaction likely stimulated binding of Pdcl, a co-chaperone mediating the folding of Gβ₁ [33]. Among 52 proteins with decreased association were 12 known TRiC interactors and substrates (Atg16l1, Wdr31, Txndc9, Tubb5, Wdtc1, Tuba4a, Tubb2a, Hnrnpll, Iars, Gnb3, Gnb5, and Plekhb1). This group also included three bona fide TRiC substrates from the tubulin family: Tubg1, Tubb3, and Tuba1a (Fig 10D). Together, these data demonstrate that unfolded Gβ₁ increasingly occupies TRiC, displacing other substrates when it fails to properly fold due to the absence of Gγ₁. This competitive binding may underlie Gβ₁ cytotoxicity and could, in theory, amplify the detrimental effects of misfolding mutations in other TRiC substrates. For example, when tubulin folding in rods is impaired, microtubules in cilium and elsewhere may become unstable or dysfunctional. Accordingly, the increased interaction of TRiC with the cytoskeleton- and stress-related proteins, such as Cenpv [47], Gfap [48], Pdcd5 [49], Syne1 [50], and Actr1b may reflect an attempt to stabilize or reconfigure microtubules. Likewise, impaired folding of Atg16l1 may compromise the formation of autophagosome [51], which may be further exacerbated by the increased binding of Tiprl, a positive regulator of mTORC1 inhibiting autophagy [52] essential for rod photoreceptor survival [53]. The suppression of autophagy may prevent degradation of TRiC via lysosome-autophagy pathway observed in the HeLa cells treated with cytochalasin D, resulting in TRiC inactivation by unfolded actin [54].

Download:

• PNG
  larger image
• TIFF
  original image

Fig 10. Perpetually unfolded Gβ₁ affects protein interactions of TRiC^et in rods.

A. Volcano plot of differential protein levels in the 900 kDa band isolated from the retina. TRiC^et mice of Gγ₁^-/- background and TRiC^et mice of wild type 129E backgrounds are compared at one month of age. The x-axis represents the log₂ fold change (log₂FC) and the y-axis shows (–log₁₀(adjusted p-value). Proteins were considered differentially associated using adjusted p-value <0.05 and absolute linear fold change ≥1.5, equivalent to |log2FC| ≥ 0.59. The data were collected at the Proteomic Facility at the Duke University. B. Nransduction in mouse rods, they werormalized levels of the eight subunits of TRiC, the numerical values of the log₂ fold change (log2FC), and p-value in the compared groups. C. Normalized levels of proteins increased in TRiC^et preparations from Gγ₁^-/- retinas (red) ranked based on log₂FC. D. Normalized levels of proteins decreased in TRiC^et preparations from Gγ₁^-/- retinas (blue) ranked based on log₂FC. Independently identified TRiC^et interactors (Sup. Table II) are in bold font.

https://doi.org/10.1371/journal.pone.0345843.g010