Myogenic Differentiation (MyoD) Gene Expression in Cornea and Role in Corneal Myofibroblast Dedifferentiation
Ratnakar Tripathi, Nishant R. Sinha, Swati Sood, Suneel Gupta, Rajnish Kumar, Prashant R. Sinha, Praveen K. Balne, Shyam S. Chaurasia, Arkasubhra Ghosh, Rajiv R. Mohan
TL;DR
This study explores the role of the MyoD gene in corneal fibrosis and shows that silencing MyoD reduces fibrosis markers in corneal cells.
Contribution
The study is the first to demonstrate MyoD's role in corneal myofibroblast dedifferentiation and fibrosis regulation.
Findings
MyoD expression is significantly higher in fibrotic corneas compared to non-fibrotic ones.
Silencing MyoD in corneal myofibroblasts reduces fibrotic gene expression and promotes a fibroblast phenotype.
In vivo experiments show reduced corneal scarring with MyoD gene silencing.
Abstract
Myogenic differentiation (MyoD), a class II basic helix–loop–helix transcription factor, regulates multiple cell functions, including fibrosis in many organs, but remains unknown in the cornea. This study characterized the expression of MyoD in non-fibrotic and fibrotic rabbit and human donor corneas and investigated the effects of MyoD gene silencing on corneal myofibroblast dedifferentiation in vitro and fibrosis in vivo. New Zealand White rabbits, human donor corneas, human corneal stromal fibroblasts (CSFs), and human corneal myofibroblasts (CMFs) were used. MyoD shRNA or scrambled shRNA was delivered into CMFs via Lipofectamine 3000 and rabbit cornea via 2-kDa polyethylenimine conjugated to gold nanoparticles (PEI2-GNPs). Quantitative real-time polymerase chain reaction (qRT-PCR), immunofluorescence, and western blotting quantified the expression of profibrotic genes, intermediate…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8| Gene | Primers (5′–3′) |
|
|---|---|---|
|
| ||
| Forward | CAC AAC GGA CGA CTT CTA TG | 60 |
| Reverse | AGT GCT CTT CGG GTT TCA | 60 |
|
| ||
| Forward | TGGGTGACGAAGCACAGAGCT | 60 |
| Reverse | CTTCAGGGGCAACACGAAGC | 60 |
|
| ||
| Forward | CGCAGCTTCGAGATCAGTGC | 60 |
| Reverse | TCGACGGGATCACACTTCCA | 60 |
|
| ||
| Forward | TGTGGCCCAGAAGAACTGGTACAT | 60 |
| Reverse | ACTGGAATCCATCGGTCATGCTCT | 60 |
|
| ||
| Forward | AGGTGTTGACGGCTTACCTG | 60 |
| Reverse | TTGAGTCCCGGTAGACCAAC | 60 |
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCorneal Surgery and Treatments · Corneal surgery and disorders · Proteoglycans and glycosaminoglycans research
The myogenic differentiation (MyoD) gene is a class II basic helix–loop–helix (bHLH) transcription factor that regulates myogenesis in myoblasts by activating genes involved in the cell cycle, adhesion, cytoskeletal organization, apoptosis, and fibrogenesis.1^–^5 The bHLH proteins were initially categorized based on tissue distributions, DNA-binding specificities, and dimerization but were subsequently classified into six groups based on sequence comparisons, E-box binding, presence of residues in other parts of the motif, and availability of additional domains.6 MyoD plays a crucial role in cellular differentiation, specifically during development and myogenesis by activating muscle genes, forming heterodimers with E proteins, and binding to DNA. Recent studies suggest that MyoD functions also contribute to cell dedifferentiation in addition to myogenesis. The expression of MyoD in cornea and its functional roles in corneal wound healing, particularly in CMF dedifferentiation and fibrosis modulation, remain unknown.
Previously, we identified the expression and functional role of another member of bHLH transcription factor, inhibitor of differentiation (Id) proteins, in the cornea.9^,^10 MyoD and Id are both bHLH proteins, but they belong to different classes and perform distinct functions. MyoD promotes gene expression and differentiation by binding to DNA; in contrast, Id proteins do not bind to DNA and function by forming protein–protein dimers in proliferating cells in a tissue-specific manner.11^–^14 Additionally, MyoD has been shown to regulate myofibroblast pliability, particularly during cellular differentiation, dedifferentiation, and fibrosis in heart and lung.15^–^22
Myofibroblasts play a pivotal role in stromal repair and the development of fibrosis in the cornea.23 Following ocular trauma, myofibroblasts are generated in stroma from resident quiescent keratocytes via activation of fibroblasts to facilitate wound repair under the influence of transforming growth factor β1 (TGFβ1) and other cytokines.23 Its formation and timely disappearance after wound closure are essential steps for the restoration of corneal function and vision.23 Corneal myofibroblasts (CMFs) are attributed to alpha smooth muscle actin (αSMA) stress fibers, and they secrete extracellular matrix (ECM) proteins to facilitate wound healing.23 However, excessive CMF production and their persistence in stroma after wound closure lead to corneal fibrosis and vision impairment.23^–^26
Cellular dedifferentiation was first reported in 1959 in amphibian limbs, showing that muscle cells could transform into progenitor cells.27 Myofibroblasts have been shown to proliferate, dedifferentiate, and/or transform into precursor cells in various organs, as well as in lung and heart fibrosis models.4^,^17^–^21 Literature suggests that inhibition of TGFβ1 signaling and abnormal CMF formation reduces corneal fibrosis in vitro and in vivo.28^–^35 Maltseva et al.36 demonstrated that CMFs are not terminally differentiated cells and could be induced to a fibroblast phenotype in a culture consisting of fibroblast growth factor and heparin. A recent study demonstrated that CMFs could be dedifferentiated into precursor corneal stromal fibroblasts.37 Mounting literature shows that MyoD is also involved in the regulation of myofibroblast dedifferentiation and fibrosis in lungs, kidneys, and the heart.4^,^17^,^38^–^42^.^ Whether MyoD regulates CMF dedifferentiation and fibrosis in the cornea is still unknown.
The present study examined the expression of MyoD gene in non-fibrotic and fibrotic rabbit and patient-derived human corneas and determined the functional effects of MyoD gene silencing on CMF dedifferentiation and fibrotic response using established human in vitro and rabbit in vivo models.
Methods
In Vivo Rabbit Studies
All animal procedures were performed in accordance with the approved Animal Care and Use Committee protocol of the University of Missouri, Columbia. The animal study complied with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. New Zealand White rabbits (2–3 kg) were purchased from Charles River Laboratories (Wilmington, MA, USA). Rabbits were anesthetized VIA an intramuscular injection of 50 mg/kg ketamine hydrochloride (JHP Pharmaceuticals, Parsippany, NJ, USA) and 10 mg/kg xylazine hydrochloride (Bimeda, Dublin, Ireland). Under general anesthesia, the central cornea was wounded by a topical exposure of 0.5-N NaOH by placing an 8-mm filter-paper disc wetted with alkali for 30 seconds after instilling two drops of topical anesthetic, proparacaine (0.5% proparacaine hydrochloride). The contralateral unwounded eye served as non-fibrotic control. The progression of corneal fibrotic events was monitored clinically with a slit-lamp microscope periodically. All rabbits underwent humane euthanasia following reported methods.32 Non-fibrotic (unwounded) and fibrotic (alkali-wounded) corneas were used in the investigations. Three observers (SG, SS, RT, or RK) in a masked manner used the Fantes scale to estimate corneal haze levels and the amount and intensity of fibrotic scar in the rabbits.32
Patient-Donated Non-Fibrotic and Fibrotic Human Cornea Experimental Design
All human tissue samples were collected at Narayana Nethralaya, Bangalore, India, with prior approval of the Institutional Ethics Committee (ECR/187/Inst/KAR/2013) as per the guidance of the Indian Council for Medical Research, India, and tenets of the Declaration of Helsinki. Written, informed consent was obtained from all patients or their legal guardians undergoing penetrating keratoplasty or corneal transplantation for vision-compromising fibrotic scar, allowing both the surgical procedure and the use of excised tissue for research purposes. The study also followed the ARVO ethical human tissue use guidelines. Donated human corneas used in the study were processed and analyzed at the GROW Research Lab, Narayana Nethralaya Foundation, Narayana Nethralaya, Bangalore, India.
Primary Human Corneal Cell Cultures and In Vitro Corneal Fibrosis Model
Healthy human donor corneas obtained from an eye bank (Saving Sight, Kansas City, MO, USA) were used to isolate human corneal epithelial (HCE) cells, corneal stromal fibroblasts (CSFs), and human corneal endothelial (HCN) cells and to establish primary CSF cultures. A total of 30 human donor corneas were used for RNA isolation and cDNA preparation. Briefly, corneas were washed twice with sterile 1× phosphate-buffered saline (PBS), and epithelial cells were collected with a #64 surgical blade by gentle scraping; they were transferred into 1.5-mL Eppendorf tubes, lysed in 350-µL Buffer RLT (QIAGEN, Hilden, Germany) containing β-mercaptoethanol (BME), and processed for RNA isolation. For endothelial cell isolation, the corneal endothelium was stained with Trypan blue (0.25%, 5–10 µL covering tissue from inside for 1 minute). The stained endothelial cells were carefully peeled off using a surgical needle and tweezers and transferred into a 1.5-mL Eppendorf tube containing 350 µL RLT+BME for RNA isolation. The primary fibroblast cultures were established from the corneal stroma obtained after removal of the epithelium and endothelium as mentioned above, following an earlier published method.43 In brief, the stromal tissue was cut into eight to 10 small pieces, placed onto a sterile T-25 culture flask, and incubated at 37°C in a humidified 5% CO_2_ incubator in Minimum Essential Medium (MEM) containing 10% fetal bovine serum (FBS) for 3 to 4 weeks. The fibroblast cells (CSFs) obtained from the stromal explants were maintained in culture for two to six passages (logarithmic growth phase), and used for RNA isolation, myofibroblast generation, and in vitro wound healing experiments.
A standard human in vitro corneal fibrosis model involving transdifferentiation of CSFs (non-fibrotic) to CMFs (fibrotic) by TGFβ1 was used.44 In brief, CSF cultures at ∼40% confluence were serum starved for 8 hours and thereafter were grown in the presence or absence of TGFβ1 (5 ng/mL; PeproTech, Cranbury, NJ, USA) for up to 72 hours under serum-free conditions and fed with fresh serum-free MEM every 24 hours until the termination point. A phase-contrast microscope was used to monitor changes in cellular morphology. All of the experiments involving donor corneas were performed following the guidelines of the Institutional Review Board, tenets of the Declaration of Helsinki, and ARVO guidelines.
RNA Isolation and Reverse Transcriptase PCR
Total mRNA was isolated from HCE, CSF, and HCN cells using the RNeasy Mini Kit (74126; QIAGEN). The purity of RNA was assessed using a NanoDrop microvolume spectrophotometer with a 260/280 ratio of 2, considered as pure RNA. Briefly, 2 µg of total mRNA isolated from all corneal cells was reverse transcribed into cDNA with the Avian Myeloblastosis Virus (AMV) Reverse Transcriptase enzyme provided in a commercial kit (Promega, Madison, WI, USA) following the manufacturer's instructions.
Conventional reverse-transcription polymerase chain reaction (RT-PCR) was performed to detect the expression of the MyoD gene in HCE, CSF, and HCN cells using primers specific for MyoD (NM_002478): forward 5′-GCC AGC ACT TTG CTA TCT-3′ and reverse 5′-TGG TTT GGA TTG CTC GAC-3′, as described previously.45 A 50-µL PCR reaction mix contained 10 µL of 5× colorless GoTaq Flexi Buffer, 4-mM magnesium chloride (8 µL), 0.2-mM deoxynucleotide triphosphates (dNTPs; 1 µL), 0.4-mM forward (1 µL) and reverse (1 µL) primers, 1.25 U (0.25 µL) of GoTaq Hot Start Polymerase (Promega), 100 ng (2 µL) cDNA, and 0.1% diethylpyrocarbonate (DEPC)-treated sterile water (26.75 µL). The PCR cycling conditions consisted of an initial denaturation at 95°C for 3 minutes, followed by 40 cycles of denaturation at 95°C for 30 seconds, annealing at 55°C for 30 seconds, and extension at 72°C for 60 seconds. A forward primer, 5′-AGG CCA ACC GCG AGA AGA TGA CC-3′, and a reverse primer, 5′-GAA GTC CAG GGC GAC GTA GCA C-3′, were used to amplify the β-actin housekeeping gene. All PCR reactions were conducted in triplicates in a MyCycler thermal cycler (Bio-Rad Laboratories, Hercules, CA, USA). The resulting amplicons were resolved on 1% agarose gel and subjected to electrophoresis using a 100-bp DNA ladder.
Quantitative Real-Time PCR
Quantitative real-time PCR (qRT-PCR) reactions were performed using the StepOnePlus Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) to measure the relative levels of various genes under different conditions as reported previously.32^,^44^,^46 The nucleotide sequences of the forward and reverse primers of the MyoD, αSMA, fibronectin (FN), collagen type I (Col-I), Col-IV, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) are provided in the Table. A 20-µL reaction mixture contained 100 ng of cDNA (2 µL), 200 nM of forward primer (2 µL) and reverse primer (2 µL), and 10 µL of 2× SYBR Green qPCR SuperMix (Applied Biosystems), and the remaining volume was adjusted with DEPC water (4 µL). The reaction mixture was processed at the universal cycle conditions involving an initial denaturation cycle at 95°C for 10 minutes followed by 40 cycles of denaturation at 95°C for 15 seconds and annealing at 60°C for 60 seconds. The GAPDH gene was used for normalization of the data. The 2^–^^ΔΔCt^ method was used to calculate relative fold changes over the respective control values. For each sample, qRT-PCR was performed in triplicate, and the average fold changes in mRNA levels were reported.
shRNA Gene Silencing in Human CMFs In Vitro and Rabbit Cornea In Vivo
The functional role of MyoD in the modulation of TGFβ1-mediated corneal fibrosis was evaluated using shRNA gene silencing in vitro*.* MyoD shRNA and control scrambled shRNA plasmids were obtained from Victor Thannickal, MD, University of Alabama, Birmingham.4 The DNA sequences of each plasmid were verified through the DNA core facility of the University of Missouri, Columbia. Gene transfer experiments were performed following published protocols.45 In brief, MyoD shRNA or scrambled plasmid was delivered into CMF using 150 µL of transfection mixture prepared by combining 2 µg of plasmid with 10 µL of the Invitrogen Lipofectamine 3000 transfection reagent (L3000001; Thermo Fisher Scientific, Waltham, MA, USA). The mixture was applied dropwise in each well of a six-well plate containing CMFs at 70% confluence. The cultures were incubated for 20 minutes at room temperature (RT) and subsequently maintained in a standard humidified tissue culture incubator for 72 hours, with fresh serum-free MEM provided every 24 hours. The cellular viability and morphology were assessed with a phase-contrast microscope equipped with a camera and fluorescence capabilities.
The polyethylenimine conjugated to gold nanoparticles (PEI2-GNPs) were used to deliver MyoD shRNA or scrambled plasmid in rabbit cornea as reported previously.32 In brief, the PEI2-GNPs were mixed with MyoD shRNA or scrambled shRNA at a nitrogen-to-phosphate (N/P) ratio of 180 by stirring 37.5 µL of 150-mM PEI2-GNPs with 10 µg of DNA, 10% glucose (w/v), and balanced salt solution (BSS) to bring the volume to 100 µL. This transfection solution was incubated at 37°C for 30 minutes and topically applied onto the cornea 1 day after alkali injury as described earlier.32 Each rabbit received alkali injury (Group 1; 8-mm filter disc with 0.5-N sodium hydroxide; n = 6), alkali injury and PEI2-GNP–scrambled shRNA (Group 2; n = 6), or alkali injury and PEI2-GNP–MyoD shRNA (Group-3; n = 6). The contralateral untreated eyes served as naïve controls.
Immunofluorescence
Immunofluorescence was used to detect the level, localization, and distribution of MyoD and αSMA proteins in non-fibrotic and fibrotic rabbit and human corneas, as well as in vitro CSFs and CMFs, using antibodies specific to MyoD (sc-377460; Santa Cruz Biotechnology, Dallas, TX, USA) and αSMA (M0851; Dako, Glostrup, Denmark). For dedifferentiation studies, immunostaining was performed with the Ki67 (9129T; Cell Signaling Technology, Danvers, MA, USA), FSP1 (ab-41532; Abcam, Cambridge, UK), and intermediate filaments, such as Invitrogen Vimentin Polyclonal Antibody (PA1-16759; Thermo Fisher Scientific) and Invitrogen Desmin Polyclonal Antibody (PA5-19063; Thermo Fisher Scientific), in CMFs with or without MyoD shRNA or scrambled shRNA. Samples were incubated in 1× PBS for 10 minutes followed by blocking with 5% horse serum in 1× PBS + 0.05% Tween 20 (PBST). They were then incubated in primary antibody at 1:500 dilution overnight at 4°C and washed with PBST three times (10 minutes each). This was followed by incubation with suitable secondary antibodies at 1:1000 dilution for 1 hour at RT: Invitrogen Goat anti-Chicken IgY (H+L) Secondary Antibody, Alexa Fluor 594 (A-11042; Thermo Fisher Scientific); Invitrogen Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 594 (A-21125; Thermo Fisher Scientific); Invitrogen Donkey anti-Goat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 (A-11055; Thermo Fisher Scientific); Invitrogen Donkey anti-Goat IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 594 (A-11058; Thermo Fisher Scientific); or Invitrogen Goat anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 (A-11001; Thermo Fisher Scientific). Thereafter, samples were washed three times with 1× PBST (10 minutes each) and mounted in VECTASHIELD containing 4′-6-diamidino-2-phenylindole (DAPI; H1200; Vector Laboratories, Newark, CA, USA). Images were captured with a Leica DM4000B fluorescent microscope (Leica Microsystems, Wetzlar, Germany) equipped with a digital camera (SPOT RT KE; SPOT Imaging, Sterling Heights, MI, USA).
Western Blot
Western blot was performed to quantify levels of MyoD and αSMA proteins in CSFs and CMFs following a previously reported method.46 In brief, cells (2 × 10^6^) were lysed with 350 µL radioimmunoprecipitation assay (RIPA) lysis buffer (89900; Thermo Fisher Scientific) containing protease inhibitor cocktail (sc-506202; Santa Cruz Biotechnology) in a 1.5-mL microfuge tube for 60 minutes on ice. After centrifuging for 10 minutes at 10,000g, supernatants were collected and protein concentrations were quantified using the Bradford method. Briefly, 60 µg of total protein from each sample was mixed with NuPAGE LDS Sample Buffer containing a reducing agent (Thermo Fisher Scientific), heated for 5 minutes at 90°C, and loaded onto 10% precast Invitrogen Bis-Tris Mini Protein Gels (Thermo Fisher Scientific). The resolved proteins were transferred onto polyvinylidene difluoride (PVDF) membranes with the help of Trans-Blot Turbo Transfer System (17041150; Bio-Rad Laboratories) at a constant 25 V. The PVDF membranes were blocked with blocking reagent containing 5% fat-free milk in 1× PBST and incubated with primary antibodies against MyoD (sc-377460; Santa Cruz Biotechnology), αSMA (M0851; Dako), and GAPDH (sc-47724; Santa Cruz Biotechnology) with 1:1000 dilution at 4°C overnight, followed by incubation in horseradish peroxidase–conjugated anti-mouse secondary antibody at 1:5000 dilution for 2 hours at RT. The blots were washed three times with 1× Tris-buffered saline with 0.05% Tween 20 (TBST) for 5 minutes each, and the transferred proteins on membranes were detected using enhanced chemiluminescence using a C-DiGit Blot Scanner (LI-COR Biosciences, Lincoln, NE, USA). For each experiment, three separate western blots were performed. ImageJ (National Institutes of Health, Bethesda, MD, USA) was used to normalize quantified protein levels with GAPDH, and the data are expressed as fold changes. HeLa cells were used for authentication of the MyoD primary antibody, and rabbit muscle tissues were used as positive controls to confirm the specificity of the first-time MyoD antibody used for the cornea (data not shown).
Statistical Analysis
Prism 9.2 (GraphPad, Boston, MA, USA) was used for statistical analysis. Student's t-tests, one-way and two-way analyses of variance (ANOVAs), and Bonferroni post hoc tests were used to evaluate statistical significance. P ≤ 0.05 was considered significant. Results are expressed as mean ± SEM.
Results
MyoD Is Constitutively Expressed in Human Corneas and Human Corneal Cells In Vitro
First, we tested the presence and expression of MyoD in the human corneas and in the corneal epithelial (HCE), stromal (CSF), and endothelial (HCN) cells isolated from them. Figure 1 illustrates the MyoD expression across various layers of the donor human cornea. The mRNA analysis of human HCE, CSF, and HCN cells showed substantial gene expression of MyoD in the three layers of the human cornea (Fig. 1A), supported by protein expression via western blot in human-derived corneal cells in vitro (Fig. 1B) and immunofluorescence in healthy human corneal tissue sections (Fig. 1C).
MyoD is expressed in the human cornea. (A–C) RT-PCR (A), western blot (B), and immunofluorescence (C) showing MyoD expression depicted in red staining in three main layers of healthy human donor cornea. β-actin and GAPDH were used as the internal controls for RT-PCR and western blot, respectively. HCE, human corneal epithelium; CSF, human corneal stromal fibroblast; HCN, human corneal endothelium. Scale bar: 100 µM. Blue indicates DAPI-stained nuclei; red, MyoD-expressing cells.
MyoD and αSMA Are Co-Expressed In Vivo in Rabbit and Fibrotic Human Corneas
To test the hypothesis that MyoD expression is co-expressed and regulated by corneal injury in conjunction with fibrotic events, we conducted a detailed examination of αSMA and MyoD expression in a controlled in vivo model of fibrotic rabbit corneas compared to non-fibrotic corneas and fibrotic human corneas using immunofluorescence double staining. Figure 2 shows qualitative and quantitative immunofluorescence data collected from rabbit (Figs. 2A–E) and human (Figs. 2F–J) corneas. The rabbit cornea analysis showed a significant increase in the number of αSMA^+^ cells (Fig. 2B, green) and MyoD^+^ cells (Fig. 2D, red) within the fibrotic cornea at 14 days after alkali injury compared to the non-fibrotic cornea (Figs. 2A, 2C). Further, we investigated the co-localization of αSMA and MyoD in fibrotic conditions. As evident from Figure 2D (inset), αSMA and MyoD showed co-expression in fibrotic rabbit cornea. The quantification of data shown in Figure 2E, exhibited a significantly higher number of αSMA+ (45.08 ± 3.43; P < 0.0001) cells in fibrotic rabbit corneas than the non-fibrotic corneas. Similarly, the fibrotic rabbit corneas had 40.19 ± 2.95 MyoD^+^ cells (P = 0.0003), and non-fibrotic corneas had 16.90 ± 2.18 MyoD^+^ cells. The non-fibrotic and fibrotic human corneas showed a similar pattern of αSMA^+^ and MyoD^+^ cells (Figs. 2F–J). Fibrotic human corneas showed an enhanced number of αSMA^+^ cells (Fig. 2G, green) and MyoD^+^ cells (Fig. 2I, red) compared to non-fibrotic corneas (Figs. 2F, 2H). Quantification shown in Figure 2J found significantly increased αSMA^+^ cells (62.41 ± 5.10; P < 0.0001) in fibrotic human corneas compared to non-fibrotic corneas (3.47 ± 1.37). Likewise, significantly higher MyoD^+^ levels (59.08 ± 5.95; P = 0.0019) in fibrotic human corneas compared to non-fibrotic corneas (34.94 ± 3.91) were noted. These findings illustrate a significant role of MyoD in corneal fibrosis development.
*(A–J) Representative αSMA and MyoD immunofluorescence images of rabbit cornea (A–E) and human cornea (F–J). Rabbit corneal tissue sections showed significantly enhanced αSMA (B, arrow) and MyoD (D, arrowhead) levels in fibrotic corneas compared to non-fibrotic corneas (A, C). An inset in panel D shows co-expression of MyoD and αSMA. Likewise, a significantly high αSMA (G) and MyoD (I) expression is evident in human fibrotic corneas compared to non-fibrotic corneas (F, H). The quantification of αSMA+ and MyoD+ cells in rabbit and human corneas is shown in panels E and J, respectively. **P = 0.0019, ***P = 0.0003, ***P < 0.0001. Error bars represent mean ± SEM. Scale bar: 100 µM. Green indicates αSMA+ cells; red, MyoD+ cells.
MyoD Expression Is Increased in Transdifferentiating Corneal Myofibroblasts In Vitro
MyoD is a bHLH transcription factor often regarded as a key regulator of cellular differentiation. Thus, we explored its role in the transdifferentiation of corneal myofibroblasts. Human CSFs treated with TGFβ1 for 72 hours (fibrotic) showed significantly elevated levels of αSMA mRNA (5.56 ± 0.23-fold; P < 0.0001) and sustained upregulation of MyoD mRNA (5.4 ± 0.14-fold; P < 0.0001), as observed by qRT-PCR (Fig. 3A), and protein, as observed by western blot (Fig. 3B), compared to the non-fibrotic group. In both gene expression and protein analyses, GAPDH expression served as an internal control.
*(A, B) The qRT-PCR data (A) and immunoblot data (B) show significantly increased expression of αSMA and MyoD in fibrotic conditions (+TGFβ1) at 72 hours in myofibroblasts in comparison to non-fibrotic conditions (–TGFβ1). ***P < 0.0001. Error bars represent mean ± SEM.
Furthermore, corneal fibroblasts in various transdifferentiation stages (24, 48, and 72 hours of TGFβ1 stimulation) demonstrated progressively increased expression of αSMA and MyoD proteins in a time-dependent manner (Fig. 4). A maximal enhanced expression of αSMA (Fig. 4D) and MyoD (Fig. 4I) proteins was observed at 72 hours post-fibrotic induction by TGFβ1. The quantification data showed that the αSMA^+^ cells were 44.40 ± 1.02 (P < 0.0001) (Fig. 4E) and MyoD^+^ were 41.3 ± 2.01 (P < 0.0001) (Fig. 4J) in the fibrotic condition (+TGFβ1) but not in the non-fibrotic condition (–TGFβ1). As expected, basal MyoD^+^ cells (Fig. 4F) in fibroblasts (–TGFβ1; non-fibrotic) were observed (21.88 ± 1.85) (Fig. 4J).
*Representative immunofluorescence images showing the time-dependent increased numbers of αSMA+ cells (B–D) and MyoD+ cells (G–I) and the significantly enhancement in myofibroblasts under the fibrotic condition (+TGFβ1) in comparison to the non-fibrotic condition (–TGFβ1) (A, F). Some fibroblasts also showed MyoD expression in the non-fibrotic conditions (F). Quantification data exhibited time-dependent enhanced expression of αSMA (E) and MyoD (J) in fibrotic conditions. **P = 0.0080, ***P < 0.0001. Error bars represent mean ± SEM. Scale bar: 100 µM. Blue indicates DAPI, nuclei; green, αSMA; red, MyoD.
MyoD Gene Silencing Reduced αSMA Expression and Myofibroblast Formation
The functional role of MyoD in myofibroblast transdifferentiation in the human cornea remains unclear. Hence, we silenced MyoD using shRNA to assess its role in myofibroblast transdifferentiation (Fig. 5). Silencing of MyoD using shRNA in fibrotic cells resulted in a reduced myofibroblast-like phenotype (Fig. 5C), and, more importantly, dedifferentiation (reversal) of myofibroblast- to fibroblast-like phenotype (arrows) was observed. The cells transfected with scrambled shRNA (Fig. 5B) and cells treated with vehicle (Fig. 5A) retained the myofibroblast-like appearance (arrowheads), indicating the predominant role of MyoD in myofibroblast differentiation. To further validate our findings, CMFs were collected after 72 hours and quantified for αSMA transcript (Fig. 5D) and protein (Fig. 5E) content. The CMFs transfected with MyoD shRNA showed significantly suppressed αSMA mRNA levels (3.26 ± 0.05-fold; P < 0.0001) in comparison to cells transfected with scrambled shRNA (5.52 ± 0.18-fold) in the presence of TGFβ1 (Fig. 5D). Similarly, protein levels in shRNA-mediated MyoD-silenced CMF cells showed inhibited expression of αSMA (1.60 ± 0.19-fold; P < 0.001), but αSMA expression was significantly enhanced in the scrambled shRNA (3.48 ± 0.21-fold) and vehicle-treated (3.78 ± 0.28-fold) CMFs (Fig. 5E). Similarly, MyoD shRNA silencing in CMFs demonstrated significantly decreased αSMA^+^ cells (24.91 ± 03.48; P < 0.001) (Fig. 5H), but CMFs transfected with scrambled shRNA (45.41 ± 02.60) (Fig. 5G) and vehicle-treated shRNA (45.91 ± 03.19) (Fig. 5F) showed robust αSMA^+^ cells.
*(A–C) Representative phase-contrast microscope images showing the effect of shRNA-mediated MyoD silencing on the CMFs. shRNA-mediated MyoD silencing showed fewer numbers of fibroblast phenotypes (arrows) even after being treated with +TGFβ1 (C) versus scrambled shRNA (B); myofibroblast treated with vehicle (A) showed more myofibroblast phenotypes (arrowheads). (D–H) The qRT-PCR (D), western blot (E), and immunofluorescence (F–H) results showing the effects shRNA-mediated MyoD silencing on the expression of αSMA. The shRNA-mediated MyoD silencing indicated suppressed expression of αSMA mRNA (D) and protein (E), respectively, versus scrambled shRNA (and vehicle-treated) CMFs. Representative immunofluorescence images and quantification depict the effects of MyoD silencing on the expression of αSMA. Maximal αSMA+ cells (arrow) were observed in CMF cells transfected with scrambled shRNA (G) and cells that were vehicle treated (F). When MyoD was silenced by shRNA, smaller numbers of αSMA+ cells were observed (H). Quantification data indicate significantly low numbers of αSMA+ cells upon shRNA-mediated MyoD silencing versus scrambled shRNA and vehicle-treated cells (I). ***P < 0.001, ***P < 0.0001; ns, not significant. Error bars represent mean ± SEM. Scale bar: 100 µM.
MyoD Silencing Depletes Key ProFibrotic Genes and Intermediate Filaments During Myofibroblast Transdifferentiation
The addition of MyoD shRNA to CMFs in the presence of TGFβ1 for 72 hours caused a significant reduction in profibrotic genes during myofibroblast transdifferentiation. The qRT-PCR data demonstrated that CMFs transfected with MyoD shRNA showed reduced profibrotic gene transcripts: FN (2.46 ± 0.15-fold; P = 0.0015) (Fig. 6A), Col-I (2.57 ± 0.20-fold; P = 0.0011) (Fig. 6B), and Col-IV (8.53 ± 0.20-fold; P = 0.0053) (Fig. 6C) mRNA. In contrast, elevated levels of FN (3.78 ± 0.18-fold; P < 0.0001) (Fig. 6A), Col-I (3.84 ± 0.13-fold; P < 0.0001) (Fig. 6B), and Col-IV (11.33 ± 0.61-fold; P < 0.0001) (Fig. 6C) mRNA was observed in cells transfected with scrambled shRNA compared to vehicle-treated CMFs. During myofibroblast transdifferentiation, fibroblasts attain the myofibroblast precursor phenotype with increased expression of intermediate filaments, vimentin, and desmin.54 To test the role of MyoD in the regulation of intermediate filaments, CMFs treated with MyoD shRNA in the presence of TGFβ1 for 72 hours were compared with the non-fibrotic (–TGFβ1) and fibrotic (+TGFβ1) conditions. Time-dependent changes were observed in the expression of vimentin (arrowheads), with an early peak at 48 hours (49.8 ± 1.43) (Figs. 6N, 6X) and then there was a significant depression at 72 hours (29.3 ± 1.43; P < 0.001) (Figs. 6R, 6X), whereas desmin expression peaked later at 72 hours (46.3 ± 2.3; P < 0.001) (Fig. 6O, double arrows; Fig. 6U) compared to the –TGFβ1 group (Figs. 6F, 6c, respectively). The silencing of MyoD in the presence of TGFβ1 resulted in a significant increase in the expression of vimentin (40.7 ± 2.01; P < 0.001) (Figs. 6V, 6X) and a slight dip in desmin (39.3 ± 2.22; P < 0.001) (Figs. 6S, 6U), compared to the TGFβ1-treated group at 72 hours.
*The qRT-PCR results show the effect of shRNA-mediated MyoD silencing on the expression of profibrotic genes. (A–C) The cells with shRNA-mediated MyoD silencing showed suppressed levels of FN (A), Col-I (B), and Col-IV (C) mRNA compared to the scrambled shRNA and vehicle-treated cells. (F, J, N, R) The representative immunofluorescence images show the time-dependent increased number of vimentin+ cells at 24 hours (J) and 48 hours (N) and decreased levels at 72 hours (R), which were further enhanced after MyoD silencing in the fibrotic condition (+TGFβ1) (V) in comparison to the non-fibrotic condition (–TGFβ1) (F). (g, k, o, s) Minimal desmin expression was observed in the non-fibrotic condition (–TGFβ1) (c) and at 24 hours (g) and 48 hours (k), but significant increased expression was observed at 72 hours (o) in the fibrotic (+TGFβ1) condition which was decreased after MyoD silencing (s). αSMA expression showed time-dependent increases in fibrotic condition (+TGFβ1) at 24 hours (I, f), at 48 hours (M, j), and at 72 hours (Q, n) but it decreased after shRNA-mediated MyoD silencing (U, r). Minimal αSMA+ cells were observed in non-fibrotic (−TGFβ) condition (E, b). D, H, L, P and T similarly a, e, i, m and q panels are DAPI stained cell nuclei. Quantification data show significantly increased numbers of vimentin+ cells (X). In contrast, desmin exhibited decreased positivity upon shRNA-mediated MyoD silencing versus +TGFβ1 at 72 hours (U). G, K, O, S, and W, panels are merged images of Vimentin and αSMA and d, h, i, p and t panels are merged images of Desmin and αSMA. Blue indicates nuclei; green, αSMA; red, vimentin/desmin. ***P < 0.001, ***P < 0.0001, @P = 0.0015, #P = 0.0011, ^P = 0.0053; ns, not significant. Error bars represent mean ± SEM. Scale bar: 100 µM.
MyoD Gene Silencing Stimulates Proliferation and Dedifferentiation of Myofibroblasts to Fibroblast Phenotype In Vitro and Reduces Scarring In Vivo
Myofibroblasts are considered terminally differentiated cells that, by definition, have lost their ability to proliferate. We employed the shRNA approach to inhibit MyoD expression in myofibroblasts to examine their proliferative responses; the results are presented in Figure 7. The shRNA-mediated silencing of MyoD resulted in an increased proliferative response, as evidenced by enhanced Ki67 expression, and reduced αSMA stress fibers (Fig. 7C). The vehicle-treated fibroblast cells (Fig. 7A) maintained their proliferative capacity and showed a greater number of Ki67^+^ cells (25.94 ± 1.9; Fig. 7D), whereas differentiated αSMA-expressing myofibroblasts (Fig. 7B) in CMF cells exhibited less positive Ki67 staining (8.5 ± 1.85) (Fig. 7D) and enhanced αSMA^+^ cells (56.3 ± 2.4) (Fig. 7D). The shRNA-mediated MyoD silencing resulted in increased proliferative responses, as evidenced by enhanced expression of Ki67 (14.94 ± 3.21; P = 0.0349) (Fig. 7D) and reduced stress fibers when stained with αSMA (35.2 ± 1.98; P < 0.001) (Fig. 7D).
*(A–G) Representative immunofluorescence images revealed Ki67+ cells (A, double-headed arrows) and FSP+ cells (E, arrowhead) and minimal αSMA+ cells (arrows) in vehicle-treated CSFs. Also observed was a decrease in Ki67+ cells (B) and FSP+ cells (F) and robust αSMA+ cells in CMFs (B, F) when treated with scrambled shRNA, which further enhanced Ki67+ cells (C) and FSP+ cells (G) but decreased ɑSMA+ cells in MyoD shRNA-treated CMFs (C, G). The bar graph shows enhanced levels of Ki67+ (D) and FSP1 (H) cells in MyoD shRNA-treated cells versus scrambled shRNA-treated cells. Green indicates αSMA; red, Ki67 or FSP1.***P < 0.001, **P < 0.01, P = 0.0349. Error bars represent mean ± SEM. Scale bar: 100 µM.
Next, we examined the expression of FSP1, which is primarily produced by fibroblasts during fibrotic events and tissue remodeling.47 Treatment of CMF with MyoD shRNA resulted in increased FSP1 expression (18.9 ± 3.1; P < 0.01 (Figs. 7G, 7H) compared to the scrambled shRNA group (3.20 ± 1.85) (Fig. 7F) or non-differentiated (vehicle) group (45.9 ± 1.9) (Figs. 7E, 7H). In contrast, αSMA expression was significantly high and constitutively expressed in the scrambled shRNA groups (53.3 ± 2.4 (Figs. 7F, 7H) compared to CSF (Fig. 7E) and was reduced in the MyoD shRNA-treated groups (30.2 ± 4.9; P < 0.001) (Figs. 7G, 7H). These data indicate that myofibroblasts are not terminally differentiated cells and show early signs of dedifferentiation.
The effects of MyoD gene silencing on corneal fibrosis amount and area in live rabbits (see black circle in Fig. 8) were studied at 1-, 2-, and 3-week intervals with a clinical slit-lamp microscope (Figs. 8A–I) and appraised on the Fantes scale (Fig. 8J). A clinically relevant change in corneal fibrosis levels in MyoD shRNA versus scrambled shRNA delivery was noted at 2 weeks (P = 0.846) (Figs. 8E, 8H) and at 3 weeks (P = 0.368) (Figs. 8F, 8I). Corneas of the alkali injury alone and alkali injury plus scrambled shRNA had analogous fibrosis levels (data not shown). As expected, naïve corneas remained transparent at all tested time points (Figs. 8A–C). This in vivo analysis indicated that decline in fibrosis is derived from myofibroblast dedifferentiation in cornea, however, a detailed follow-up study is required to confirm or refute this inference.
(A–J) Representative slit-lamp images showing the intensity and area of corneal fibrosis in naïve rabbit cornea (A–C), rabbit cornea receiving alkali injury plus scrambled shRNA (D–F), and rabbit cornea receiving alkali injury plus MyoD shRNA (G–I) in vivo at 1, 2, and 3 weeks post-injury and quantification of the Fantes scale (J). A clinically relevant change in corneal fibrosis after MyoD gene silencing was noted at 2 weeks (E vs. H; P = 0.846) and at 3 weeks (F vs. I; P = 0.368) but not at 1 week (D vs. G; P > 0.999). Corneal fibrosis levels for alkali injury alone (not shown) and alkali injury plus scrambled-shRNA (D–F) were similar. The bar graph shows corneal haze scores at the Fantes scale in the naïve, alkali injury alone, alkali injury plus scrambled shRNA, and alkali injury plus MyoD shRNA rabbit eyes observed at 1, 2, and 3 weeks post-injury (J). Dotted circles show the zone of injury, and arrows show the levels of corneal fibrosis observed in rabbit eyes. Error bars represent mean ± SEM. Scale bar: 2.5 mm.
Discussion
This study analyzed the expression of the MyoD gene in non-fibrotic and fibrotic rabbit and patient-derived human corneas and determined the effects of MyoD gene silencing on corneal myofibroblast dedifferentiation and fibrosis retraction using human in vitro and rabbit in vivo models of corneal fibrosis. Regulating CMF dedifferentiation offers a new method to treat corneal fibrosis and blindness. Following ocular trauma, corneal keratocytes/fibroblasts transdifferentiate into myofibroblasts to facilitate wound repair.23 The CMFs are primarily responsible for de novo expression of αSMA along with synthesis and secretion of ECM and tissue remodeling. The development of antifibrotic modalities has focused mainly on targeting early wound healing events, including arresting fibroblast proliferation, inhibition of myofibroblast formation, preventing abnormal ECM accumulation, and/or induction of apoptotic myofibroblast death.28^–^35 Effective arrest of these processes would limit excessive corneal wound healing and scar development. However, these approaches have shown limited success in reversing corneal scars. The reversal of fibrosis in heart and lung experimental fibrosis models has been reported via myofibroblast dedifferentiation through MyoD gene silencing.4^,^17^,^38^–^42 Consistent with these literature reports, results of the present study demonstrate that corneal myofibroblasts also possess similar unique features and could be dedifferentiated into precursor fibroblasts through MyoD gene regulation.
After an initial identification of MyoD gene expression in human and rabbit corneas (Fig. 1), the study advanced through colocalization of αSMA and MyoD in the stroma of patient-derived fibrotic human and alkali-injured rabbit corneas (Fig. 2), as well as human CMF in vitro cultures (Fig. 3). A time-dependent increase in MyoD expression along with αSMA was noted in vitro with a peak at 72 hours (Fig. 4). This analysis suggests that MyoD is linked to the regulation of αSMA in the cornea during wound healing and fibrosis development.
Cellular dedifferentiation is a prominent phenomenon attributed to various physiological processes in multiple organs and tissues, reverting from a mature cellular phenotype to a progenitor or precursor state and thereby allowing the cell to regain its proliferative capacity. For example, Schwann cells, cardiac myocytes, and germ cells can dedifferentiate in response to stress.48^–^50 These cellular events hold considerable promise for the development of new therapeutic approaches aimed at addressing corneal fibrosis and ultimately preventing blindness. We investigated the dedifferentiation process from corneal myofibroblasts to fibroblasts, characterized by changes in morphological alterations in shape, size, and cellular components; gene expression profiles of fibrotic genes; regaining proliferative capacity; and the reversing of fibrotic events. The application of MyoD shRNA resulted in morphological changes in the pre-existing CMFs, as they attained a more elongated fibroblast phenotype (Figs. 5A–C). This inhibition was accompanied by a reduction in expression of the αSMA transcript (Fig. 5D), as well as protein (Fig. 5E), and a significant suppression of αSMA^+^ cells in CMFs compared to the scrambled shRNA group (Figs. 5F–I), indicating dedifferentiation of CMF in response to endogenous MyoD silencing. Additionally, MyoD shRNA exhibited a significant decrease in the expression of fibrotic genes such as FN, Col-I, and Col-IV, as described in Figures 6A to 6C. Our findings are consistent with previous studies showing a reduction in αSMA and profibrotic gene expression following the silencing of MyoD4 or treatment with prostaglandin E2.21 In addition, the intermediate filaments, vimentin and desmin, are associated with αSMA during transdifferentiation51 and could represent the earliest precursors to myofibroblasts during corneal wound healing. CMFs, when exposed to MyoD shRNA, caused a notable enhancement in the expression of intermediate filaments (specifically, vimentin) and a reduction in desmin expression compared to the TGFβ1-treated group after 72 hours. Our study, utilizing shRNA technology, uncovered the critical role of MyoD in TGFβ1-induced myofibroblast differentiation, along with the regulation of intermediate filaments, in the human cornea in vitro. The findings of this study are consistent with the literature reporting an association of MyoD with the differentiation of smooth muscle and skeletal muscle, including myofibroblasts.42
MyoD is known to be a master regulator of differentiation of skeletal muscle cells.52 Furthermore, myofibroblasts were traditionally considered to be an irreversibly differentiated cell type.53 However, recent studies have demonstrated that myofibroblasts are not terminally differentiated cells and can proliferate.36 Additionally, previous studies have emphasized the dedifferentiation of cells to retain their proliferative capacity in other organ systems.4^,^10^,^17^,^20^–^22^,^54^,^55 Thus, we tested whether human corneal myofibroblasts have a capacity to proliferate. The present study found that CMFs maintained their ability to proliferate, albeit at lower rates, as indicated by the reduced number of Ki-67-positive cells compared to CSFs. Double staining of Ki67 and αSMA through immunofluorescence demonstrated that this observation was not attributable to the activity of a small subset of (undifferentiated) fibroblasts within a heterogeneous population (Fig. 7). Interestingly, we found increased expression of FSP1, a specific marker for fibroblasts, in MyoD shRNA-treated CMFs compared to the scrambled shRNA group. Furthermore, a drop in the amount or intensity of scarring observed in subjective clinical eye examinations offered indirect support that a MyoD-driven cellular differentiation process plays a role in corneal fibrosis outcomes in vivo (Fig. 8).
The phenomenon of cellular dedifferentiation27 opens new avenues for managing cell fate and developing novel regenerative medicine approaches for corneal diseases. Dedifferentiation varies among tissues and organs, including Schwann cells, cardiac myocytes, and germ cells, but many aspects remain poorly understood. This study revealed that, despite MyoD silencing, there was only a partial resolution of fibrosis, indicating incomplete reversal. This might be due to the stoichiometric differences in protein levels, which can be linked to their relative stabilities; for example, the half-life of αSMA is approximately 100 times longer than that of MyoD.56^,^57
Although these results provide a proof of principle that MyoD silencing in cornea leads to CMF dedifferentiation and reduced fibrosis, the study has limitations. For example, MyoD expression was detected in epithelial, fibroblast, and endothelial cells of non-fibrotic corneas, which raises the possibility that MyoD might be involved in maintaining the functions of these cells, influencing E proteins, the cell cycle, and/or metabolic reprogramming.3 Also, it is important to confirm the expression of MyoD in all corneal cells through immunofluorescence colocalization studies. A lack of direct mechanistic evidence that CMF dedifferentiation is derived from the unique MyoD-driven signaling pathway and not by attenuation of a TGFβ1-driven program is another limitation of the study. A decline in trauma-induced corneal fibrosis in MyoD-silenced rabbit eyes in vivo was assessed by clinical slit-lamp only. It is essential to verify functional MyoD response with histological and molecular studies and to determine if MyoD silencing or CMF dedifferentiation does not cause adverse effects to the cornea or other ocular tissues. Our ongoing and future studies will fill these critical knowledge gaps.
Conclusions
The present study demonstrated the expression of MyoD in cornea and offers a proof of concept that inducing myofibroblast dedifferentiation via MyoD silencing in the cornea is a viable strategy to rescind corneal fibrosis and restore vision.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Lassar AB, Buskin JN, Lockshon D, et al. Myo D is a sequence-specific DNA binding protein requiring a region of myc homology to bind to the muscle creatine kinase enhancer. Cell. 1989; 58: 823–831.2550138 10.1016/0092-8674(89)90935-5 · doi ↗ · pubmed ↗
- 2Wardle FC. Master control: transcriptional regulation of mammalian Myod. J Muscle Res Cell Motil. 2019; 40: 211–226.31301002 10.1007/s 10974-019-09538-6PMC 6726840 · doi ↗ · pubmed ↗
- 3Olson EN. Regulation of muscle transcription by the Myo D family. The heart of the matter. Circ Res. 1993; 72: 1–6.8380257 10.1161/01.res.72.1.1 · doi ↗ · pubmed ↗
- 4Hecker L, Jagirdar R, Jin T, Thannickal VJ. Reversible differentiation of myofibroblasts by Myo D. Exp Cell Res. 2011; 317: 1914–1921.21440539 10.1016/j.yexcr.2011.03.016PMC 3123424 · doi ↗ · pubmed ↗
- 5Dong Y, Lakhia R, Thomas SS, Wang XH, Silva KA, Zhang L, Interactions between p-Akt and Smad 3 in injured muscles initiate myogenesis or fibrogenesis. Am J Physiol Endocrinol Metab. 2013; 305: E 367–E 375.23736539 10.1152/ajpendo.00644.2012 PMC 3742853 · doi ↗ · pubmed ↗
- 6Murre C, Bain G, Vandijk MA, et al. Structure and function of helix–loop–helix proteins. Biochim Biophys Acta. 1994; 1218: 129–135.8018712 10.1016/0167-4781(94)90001-9 · doi ↗ · pubmed ↗
- 7Atchley WR, Fitch WM. A natural classification of the basic helix–loop–helix class of transcription factors. Proc Natl Acad Sci USA. 1997; 94: 5172–5176.9144210 10.1073/pnas.94.10.5172 PMC 24651 · doi ↗ · pubmed ↗
- 8Jones S. An overview of the basic helix–loop–helix proteins. Genome Biol. 2004; 5: 226.15186484 10.1186/gb-2004-5-6-226PMC 463060 · doi ↗ · pubmed ↗