What is SRD5A3?
SRD5A3 is the name of an evolutionarily conserved gene that encodes an enzyme called Steroid 5alpha Type 3. SRD5A3 belongs to a group of enzymes called polyprenol reductases but its functions cannot be compensated for by the related polyprenol reductases. In the absence of SRD5A3, polyprenol is no longer converted (reduced) to dolichol. Dolichol is the lipid carrier that shuttles oligosaccharides to their site of transfer onto asparagine residues in freshly synthesized proteins. The featured image above is sufficient to illustrate the role of SRD5A3 in N-linked glycosylation.
The early days of diagnostic whole exome sequencing (WES) sequencing between 2010 and 2012 were a boon to researchers and clinicians studying inborn errors of metabolism and seeing “CDG-1x” patients. That is, patients who were diagnosed with a congenital disorder of glycosylation based on blood tests and clinical presentation but whose causal gene and pathogenic mutation(s) had not yet been identified by conclusive genetic testing.
In the case of SRD5A3-CDG, the first clinical reports appeared around the same time from several groups, including two seminal papers from an international team that set the stage for SRD5A3 research: Cantagrel et al., 2010 and Morava et al., 2010. Since 2012, nearly all SRD5A3-CDG papers have been n-of-1 clinical reports of additional SRD5A3-CDG pediatric and even adult cases from around the world. These publications are cited below and some are open access. In 2018, the first viable SRD5A3 mouse model was published by a French group: Medina-Cano et al., 2018.
I will review these papers here. I will conclude with a discussion of how new collaborative research can fill in discovery and preclinical gaps, and help the SRD5A3-CDG community seize clinical opportunities.
First clinical description of SRD5A3-CDG
Morava et al is a clinical report of the first diagnostic cohort comprising 12 SRD5A3 children from 9 families from Turkey, Poland and Iran. Not many more SRD5A3-CDG patients have been identified since and the expectation is that this will always be an ultra-rare disease. SRD5A3-CDG shares the core diagnostic criteria of all Type 1 CDGs with endoplasmic reticulum-based defects, namely developmental delay, cerebellar malformations and nystagmus (involuntary repetitive eye movements). Targeted DNA sequencing of the SRD5A3 gene and measurement of elevated plasma polyprenol levels confirm the diagnosis.
To quote the relevant passage in Morava et al: “The diagnostic criteria of this novel inborn error of glycosylation, SRD5A3-CDG, are psychomotor retardation, nystagmus, visual impairment due to variable eye malformations, vermis anomalies and abnormal coagulation. Ichthyosiform skin lesions may support the clinical suspicion.”
Early-onset ocular presentation and congenital eye defects called colobomas and glaucoma appear unique to SRD5A3-CDG. SRD5A3 is expressed at high levels in human brain tissue, especially in the cerebellum, at moderate levels in the eye and heart, and at low levels in other organs. The cerebellar atrophy observed in SRD5A3-CDG kids results from increased sensitivity of developing cerebellar tissue to loss of SRD5A3 enzyme function. The developing eye is another tissue that is particularly sensitive to loss of SRD5A3. As we imagine therapeutic interventions, it’s critical to keep in mind that some damage or defects may be irreversible depending on the time since the insult occurred in development and depending on the tissue/cell type.
Like CDGs as a category, the clinical presentation of SRD5A3-CDG varies widely from patient to patient, with some severely challenged by a multisystem disorder and with some displaying mild neurological defects. This range exists despite the fact that in SRD5A3-CDG the vast majority of documented mutations lead to premature stop codons and a truncated, functionless form of the SRD5A3 protein, e.g., W19X. That means SRD5A3 patients are effectively null mutants. This fact deprioritizes a therapeutic strategy like pharmacological chaperones which can only act upon full-length if unstable or misfolded proteins, not the stumps of proteins. A better therapeutic thesis to pursue are nonsense suppressors that enable the cell to read through a premature stop codon and create functional protein.
The first generation of SRD5A3 disease models
Cantagrel et al is a thorough dissection of the pathophysiology of SRD5A3 using yeast, mouse and patient fibroblast models, with confirmation studies that measured patient plasma polyprenols. They began with a CDG-1x family with four affected children. Genetic mapping experiments homed in on SRD5A3 as the causal gene. Examination of a larger set of nearly 40 CDG-1x families led to diagnosis of a half a dozen SRD5A3-CDG patients.
They then took advantage of the fact that the yeast genome has an evolutionarily related version of SRD5A3 called DFG10. In a set of elegant experiments, human SRD5A3 expressed in yeast cells lacking DFG10 rescue the glycosylation of the yeast analog of transferrin called CPY. However SRD5A3’s more distantly related polyprenol reductase cousins SRD5A1, SRD5A2, SRD5A2L2 and GPSN2 do not rescue the glycosylation of CPY.
Being able to work with simple yet powerful yeast models turned out to be good fortune because a SRD5A3 knockout (null) mouse embryo does not survive past day 12.5 of development. But Calangrel et al were able to measure the accumulation of polyprenols in mutant whole mouse embryos. Accumulation of polyprenols is a universal phenotype across yeast, mice and human cells SRD5A3-deficient model systems. Notice that even though polyprenol levels increase, dolichol levels remain flat. The inescapable conclusion is that cells have another way to synthesize dolichol than just from polyprenol. But how exactly? This remains an open question for new collaborative research project to answer.
A viable SRD5A3 mouse model
Lack of a mouse or any rodent model is a gap that must be filled by every rare disease community. Last year a paper came out describing a viable SRD5A3 mouse model. In order to create a viable SRD5A3 mouse model, Medina-Cano et al knocked out SRD5A3 only in the cerebellum. This mouse model presents a chance to ask questions that were raised by Cantagrel et al. What are the defining features of glycoproteins that are sensitive to loss of SRD5A3 given that dolichol levels remained largely changed? Which cerebellar glycoproteins are particularly sensitive to loss of SRD5A3 and how does their absence or reduced function explain cerebellar atrophy?
First, they confirmed that cerebellar-specific SRD5A3 knockout mice actually have cerebellar defects by documenting a size difference between the cerebella of mutant mice versus control mice and a corresponding loss of motor function in a behavior test called the rotarod. Then they determined that glycoproteins involved in cell-to-cell adhesion in the brain are among the most affected by loss of SRD5A3, including suspected secreted glycoproteins like antibodies. The explanation is that these glycoproteins have multiple (4 or more) asparagine sites that must all be glycosylated for the protein to function.
How to get from here to the clinic?
The shortest and cheapest path is drug repurposing, or finding a new use for an old drug, followed by drug repositioning, or giving new life to a shelved or dormant asset that failed somewhere after a positive Phase 1 safety trial.
As a prerequisite for drug screens, the SRD5A3-CDG community needs more disease models. Yeast models have been validated since Cantagrel et al. Worm, fly and zebrafish models have not yet been generated. Patient fibroblasts exhibit the relevant disease phenotypes from a defect in secreting glycoproteins like DNase I to the accumulation of polyprenols, which means they can be used to validate hits from model organism drug screens. iPSCs can also be generated from patient cells and complement fibroblast studies.
As mentioned above, the preponderance of SRD5A3 nonsense mutations means screening targeting libraries of nonsense suppressors (the classic if questionable example is ataluren) is a worthwhile and ultimately fruitful exercise if done properly.
Delivery of dolichol phosphate, generally referred to a substrate replacement therapy, is another avenue that can be explored once animal and patient derived cellular models have been characterized.
The unsparing ocular presentation suggests a SRD5A3 gene therapy targeted to the eye might be worth pursuing since SRD5A3 is a reasonably sized payload at 318 amino acids. The eye could also be a proving ground for an experimental gene editing therapy that corrects one of the nonsense alleles, restoring the patient to disease-free heterozygous/carrier status.
The big opportunity for drug discovery is to identify modifiers, specifically activators, of the alternate SRD5A3-independent pathway for dolichol production. Model organisms and patient cell models can be exploited to identify novel compounds or genetic suppressors that safely boost this pathway.
Cantagrel V, Lefeber DJ, Ng BG, Guan Z, Silhavy JL, Bielas SL, Lehle L, Hombauer H, Adamowicz M, Swiezewska E, De Brouwer AP, Blümel P, Sykut-Cegielska J, Houliston S, Swistun D, Ali BR, Dobyns WB, Babovic-Vuksanovic D, van Bokhoven H, Wevers RA, Raetz CR, Freeze HH, Morava E, Al-Gazali L, Gleeson JG. (2010). SRD5A3 is required for converting polyprenol to dolichol and is mutated in a congenital glycosylation disorder. Cell. 142: 203-217.
Gründahl JE, Guan Z, Rust S, Reunert J, Müller B, Du Chesne I, Zerres K, Rudnik-Schöneborn S, Ortiz-Brüchle N, Häusler MG, Siedlecka J, Swiezewska E, Raetz CR, Marquardt T. (2012). Life with too much polyprenol: polyprenol reductase deficiency. Molecular Genetics and Metabolism. 105: 642-651.
Kara B, Ayhan Ö, Gökçay G, Başboğaoğlu N, Tolun A. (2014). Adult phenotype and further phenotypic variability in SRD5A3-CDG. BMC Medical Genetics. 15: 10.
Kasapkara CS, Tümer L, Ezgü FS, Hasanoğlu A, Race V, Matthijs G, Jaeken J. (2012). SRD5A3-CDG: a patient with a novel mutation. European Journal of Paediatric Neurology. 16: 554-556.
Medina-Cano D, Ucuncu E, Nguyen LS, Nicouleau M, Lipecka J, Bizot JC, Thiel C, Foulquier F, Lefort N, Faivre-Sarrailh C, Colleaux L, Guerrera IC, Cantagrel V. (2018). High N-glycan multiplicity is critical for neuronal adhesion and sensitizes the developing cerebellum to N-glycosylation defect. eLife. 7: e38309.
Morava E, Wevers RA, Cantagrel V, Hoefsloot LH, Al-Gazali L, Schoots J, van Rooij A, Huijben K, van Ravenswaaij-Arts CM, Jongmans MC, Sykut-Cegielska J, Hoffmann GF, Bluemel P, Adamowicz M, van Reeuwijk J, Ng BG, Bergman JE, van Bokhoven H, Körner C, Babovic-Vuksanovic D, Willemsen MA, Gleeson JG, Lehle L, de Brouwer AP, Lefeber DJ. (2010). A novel cerebello-ocular syndrome with abnormal glycosylation due to abnormalities in dolichol metabolism. Brain. 133: 3210-3220.
Taylor RL, Arno G, Poulter JA, Khan KN, Morarji J, Hull S, Pontikos N, Rueda Martin A, Smith KR, Ali M, Toomes C, McKibbin M, Clayton-Smith J, Grunewald S, Michaelides M, Moore AT, Hardcastle AJ, Inglehearn CF, Webster AR, Black GC; UK Inherited Retinal Disease Consortium and the 100,000 Genomes Project. (2017). Association of Steroid 5α-Reductase Type 3 Congenital Disorder of Glycosylation With Early-Onset Retinal Dystrophy. JAMA Ophthalmology. 135: 339-347.
Tuysuz B, Pehlivan D, Özkök A, Jhangiani S, Yalcinkaya C, Zeybek ÇA, Muzny DM, Lupski JR, Gibbs R, Jaeken J. (2016). Phenotypic Expansion of Congenital Disorder of Glycosylation Due to SRD5A3 Null Mutation. JIMD Reports. 26: 7-12.
Wheeler PG, Ng BG, Sanford L, Sutton VR, Bartholomew DW, Pastore MT, Bamshad MJ, Kircher M, Buckingham KJ, Nickerson DA, Shendure J, Freeze HH. (2016). SRD5A3-CDG: Expanding the phenotype of a congenital disorder of glycosylation with emphasis on adult onset features. American Journal of Medical Genetics. 170: 3165-3171.