Evolutionary rescue of phosphomannomutase deficiency in yeast models of human disease

  1. Ryan C Vignogna
  2. Mariateresa Allocca
  3. Maria Monticelli
  4. Joy W Norris
  5. Richard Steet
  6. Ethan O Perlstein
  7. Giuseppina Andreotti
  8. Gregory I Lang

  • Curated by eLife

    eLife logo

    Evaluation Summary:

    Vignogna et al. used yeast genetics, experimental evolution and biochemistry to investigate human congenital disorders of glycosylation, often caused by mutations in PMM2. They took advantage of the observation that the budding yeast gene SEC53 is almost identical to human PMM2, and used experimental evolution to find interactors of SEC53/PMM2. Mutations in genes corresponding to other human CDG genes, including PGM1, were overrepresented. The mechanisms of how reduced pgm1 activity could compensate for defects of sec53 are not yet clear.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The most common cause of human congenital disorders of glycosylation (CDG) are mutations in the phosphomannomutase gene PMM2, which affect protein N -linked glycosylation. The yeast gene SEC53 encodes a homolog of human PMM2 . We evolved 384 populations of yeast harboring one of two human-disease-associated alleles, sec53- V238M and sec53 -F126L, or wild-type SEC53 . We find that after 1000 generations, most populations compensate for the slow-growth phenotype associated with the sec53 human-disease-associated alleles. Through whole-genome sequencing we identify compensatory mutations, including known SEC53 genetic interactors. We observe an enrichment of compensatory mutations in other genes whose human homologs are associated with Type 1 CDG, including PGM1 , which encodes the minor isoform of phosphoglucomutase in yeast. By genetic reconstruction, we show that evolved pgm1 mutations are dominant and allele-specific genetic interactors that restore both protein glycosylation and growth of yeast harboring the sec53 -V238M allele. Finally, we characterize the enzymatic activity of purified Pgm1 mutant proteins. We find that reduction, but not elimination, of Pgm1 activity best compensates for the deleterious phenotypes associated with the sec53 -V238M allele. Broadly, our results demonstrate the power of experimental evolution as a tool for identifying genes and pathways that compensate for human-disease-associated alleles.

Article activity feed

  1. Version published to 10.7554/elife.79346 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    1. While the authors identify the suppressors in known genetic interactors (GIs) of the yeast SEC53, it is worth testing if the compensatory mutations are rewiring the GIs, thereby explaining the lack of comparable compensations observed in reconstituted strains. If altered GIs explain the suppression, then while yeast serves as an excellent tool to perform these assays, the human context of the disease may require a different set of genetic suppressors and, therefore, a different target than the yeast PGM1 ortholog.

    Our data show that pgm1 mutations alone greatly improve growth of sec53-V238M strains. Our data also indicate other pathways of compensation. Whether each of these compensatory mechanisms translate to humans is unknown. However, the observed enrichment of compensatory mutations in genes whose human homologs are associated with Type 1 CDG, suggests that many of these genetic interactions are likely to be conserved.

    Also, are Sec53 and Pgm1 proteins directly interacting in yeast and whether these mutations are on the interaction interface?

    As we mention above, there is no support for a direct physical interaction between Sec53 and Pgm1.

    1. Based on the data obtained between pACT1 and pSEC53-driven expression of the SEC53 mutant alleles, the pattern of suppressors appears to be different. Authors report that the variants expressed from strong pACT1 promoters show more suppressors than those driven by native promoters. Is this a general trend in experimental evolution that slower-growing strains tend to show lesser suppressors? For example, on Page 6, line 154, "compensating for Sec53-F126L dimerization defects are rare or not easily accessible". The statement suggests that the authors did obtain suppressors that compensate for the dimerization defect. At the same time, while rare (also, are authors suggesting suppression of dimerization defect as in better dimerization?), the rate of obtaining suppressors seems to be linked to the severity of the fitness defects of the strains. The lack of suppressors may be a limitation of the evolution experiments. Indeed later in the manuscript, the authors noticed that while PGM1 suppressors obtained in V238M can also suppress F126L alleles, the suppression was not as efficient. Could it be that evolution experiments in slower-growing strains predominantly enrich suppressors in other pathways (i.e., not in the CDG orthologs) that restore the growth better and compete out the relatively weaker suppressors in PGM1? In fact, the authors report similar effects on Page 7, lines 204-210. These two paragraphs are contradictory and should be explained further.

    All of our sequencing was performed on strains with sec53 under the control of the pACT1 promoter. While we did not identify unique sec53-F126L suppressors, we cannot exclude that sec53-F126L suppressors exist, so we describe them as “rare or not easily accessible”. While it is possible that the slower growth rate of the sec53-F126L allele could impact the likelihood of observing suppressors, we think it is more likely due to the nature of the variant (dimerization defect versus stability defect) rather than growth rate. In other laboratory evolution experiments the same beneficial mutation typically has a greater effect in slower-growing backgrounds (for example: doi.org/10.1126/science.1250939).

    1. Authors report that the LOF of PGM1 compensates for the SEC53 mutations. However, the evolution experiments did not capture any LOFs in PGM1. The fitness comparisons in evolution experiments are different as many different genotypes compete in a mix. Therefore, the fitness assays in a clonal population may not represent these differences well. To test this argument, authors can try to mimic the evolution experiments by mixing two genotypes to check competitive fitness, like the co-culture of pgm1 suppressor obtained via evolution experiments with pgm1Δ.

    Though we did not perform a direct head-to-head competition between a pgm1 suppressor and a pgm1Δ, our data suggest that the pgm1 delete would outcompete some of the lower-fitness suppressors. In the Discussion we speculate as to why we do not see deletion mutations: “Given that most of the evolved clones containing pgm1 mutations are more fit than the reconstructed strains, it is possible that other evolved mutations interact epistatically only with non-loss-of-function pgm1 mutations.”. Though it is beyond the scope of the present manuscript, it would be possible to rerun the evolution experiment in sec53-V238M strains carrying either a pgm1 missense suppressor or a pgm1Δ. Under the hypothesis of additional interacting loci, only the pgm1 missense suppressors would be more likely to acquire additional compensatory mutations.

    Reviewer #3 (Public Review):

    Vignogna et al. used yeast genetics, experimental evolution and biochemistry to tackle human congenital disorders of glycosylation (CDG), a disease mostly caused by mutations in PMM2. They took advantage of the observation that the budding yeast gene SEC53 is almost identical to human PMM2, and used experimental evolution to find interactors of SEC53/PMM2. They found an overrepresentation of mutations in genes corresponding to other human CDG genes, including PGM1. Genetic and biochemical characterizations of the pgm1 mutations were carried out. This work is solid, although authors did not reveal why reduction of pgm1 activity could compensate for defects of a particular mutant allele of sec53.

    Out of curiosity, if the authors were to simply focus on the preexisting mutations, would they have gotten the materials for most of the experiments in this article? In other words, how important is the experimental evolution?

    The evolution experiment was crucial as the specific pgm1 mutations we identified here have not been reported elsewhere, nor have the orthologous mutations been identified in human PGM1.

    A strain table with full genotypes is needed.

    We added a strain genotype table (Supplemental Dataset 2).

  3. eLife

    Evaluation Summary:

    Vignogna et al. used yeast genetics, experimental evolution and biochemistry to investigate human congenital disorders of glycosylation, often caused by mutations in PMM2. They took advantage of the observation that the budding yeast gene SEC53 is almost identical to human PMM2, and used experimental evolution to find interactors of SEC53/PMM2. Mutations in genes corresponding to other human CDG genes, including PGM1, were overrepresented. The mechanisms of how reduced pgm1 activity could compensate for defects of sec53 are not yet clear.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)

  4. eLife

    Reviewer #1 (Public Review):

    The manuscript presents an interesting study that uses the previously generated yeast strains harboring human disease-specific mutations modelled in the yeast ortholog of SEC53 (V238M & F126L variants). These variants are either controlled by a strong heterologous promoter (pACT1) or a less-efficient native promoter. In either scenario, the strains manifest growth defects. The current study uses an experimental evolution strategy to evolve the strains to identify genetic suppressors of the slow growth phenotype. The authors identify several mutations in evolved strains and find a significant number of the suppressors in phosphoglucomutase 1, PGM1 (congenital disorders of glycosylation, CDG type I human ortholog). The synthetic setup replicates the compensatory mutations, but the growth rescue did not match the primary suppressors with several other mutations suggesting synergistic effects. Furthermore, reconstituted strains harboring LOF of PGM1 also showed the growth rescue, yet none of the evolved strains possessed a LOF of PGM1. The authors identify the PGM1 suppressors to be dominant. Finally, the protein activity assays reveal that the mutations in PGM1 reduce the protein activity rather than eliminate it. Overall, the assays show the power of yeast genetics for discovering the potential therapeutic targets in human diseases such as Congenital Disorders of Glycosylation 1.

    1. While the authors identify the suppressors in known genetic interactors (GIs) of the yeast SEC53, it is worth testing if the compensatory mutations are rewiring the GIs, thereby explaining the lack of comparable compensations observed in reconstituted strains. If altered GIs explain the suppression, then while yeast serves as an excellent tool to perform these assays, the human context of the disease may require a different set of genetic suppressors and, therefore, a different target than the yeast PGM1 ortholog. Also, are Sec53 and Pgm1 proteins directly interacting in yeast and whether these mutations are on the interaction interface?

    2. Based on the data obtained between pACT1 and pSEC53-driven expression of the SEC53 mutant alleles, the pattern of suppressors appears to be different. Authors report that the variants expressed from strong pACT1 promoters show more suppressors than those driven by native promoters. Is this a general trend in experimental evolution that slower-growing strains tend to show lesser suppressors? For example, on Page 6, line 154, "compensating for Sec53-F126L dimerization defects are rare or not easily accessible". The statement suggests that the authors did obtain suppressors that compensate for the dimerization defect. At the same time, while rare (also, are authors suggesting suppression of dimerization defect as in better dimerization?), the rate of obtaining suppressors seems to be linked to the severity of the fitness defects of the strains. The lack of suppressors may be a limitation of the evolution experiments. Indeed later in the manuscript, the authors noticed that while PGM1 suppressors obtained in V238M can also suppress F126L alleles, the suppression was not as efficient. Could it be that evolution experiments in slower-growing strains predominantly enrich suppressors in other pathways (i.e., not in the CDG orthologs) that restore the growth better and compete out the relatively weaker suppressors in PGM1? In fact, the authors report similar effects on Page 7, lines 204-210. These two paragraphs are contradictory and should be explained further.

    3. Authors report that the LOF of PGM1 compensates for the SEC53 mutations. However, the evolution experiments did not capture any LOFs in PGM1. The fitness comparisons in evolution experiments are different as many different genotypes compete in a mix. Therefore, the fitness assays in a clonal population may not represent these differences well. To test this argument, authors can try to mimic the evolution experiments by mixing two genotypes to check competitive fitness, like the co-culture of pgm1 suppressor obtained via evolution experiments with pgm1Δ.

  5. eLife

    Reviewer #2 (Public Review):

    Slow growing yeast with a human mutation in SEC53 (Man-6P to Man-1P) that causes a rare human glycosylation disorder (PMM2-CDG) improve growth and glycosylation by natural selection over 1000 generations. Partial deficiencies in PGM1 (Glc-6-P to Glc-1P) do it best. Mechanism and usefulness for other PMM2-CDG mutations is uncertain.

    Strengths
    • This paper provides a nice unbiased way to allow evolution to decide on the best way to compensate for a deleterious, growth-compromising mutation, rather than using preconceptions or testing specific hypotheses. This approach is hypothesis-generating.
    • This approach is possible using a rapidly growing organism, like yeast, that has genes corresponding to the human homologs with sufficiently conserved amino acids or structural features.
    • This is perhaps the best way to investigate potential therapeutic targets for human metabolic disorders, in addition to targeting the specific genetic/enzymatic deficiency.
    • The results described for a single mutation in SEC53 and the best compensatory candidate, PGM1, greatly benefit from having standard, well-established biochemical assays. Perhaps the approach could be applied to other mutations in human PMM2 and even to other genes within glycosylation pathway(s).
    • The best of the 5 PGM1 variants isolated were able to restore normal N-glycosylation.
    • This extends the possibility that metabolic manipulation of overlapping or complementary metabolic pathways is a viable therapeutic approach. This type of approach for treatment of patients with other glycosylation disorder has been demonstrated for supplements of mannose, galactose and fucose.

    Weaknesses
    • The initial experiments examined two different human PMM2 pathological mutations. One, F126L, interferes with obligate dimer formation of the PMM2 enzyme and the other (V231M) forms dimers but the protein is unstable. Only the latter generated important candidates. The former grew too slowly to produce identifiable candidates after 1000 generations. So, while the V231M variant, which retains 39% of control enzymatic activity and 32% of the growth rate, can be improved over the control, the improvement is measured in yeast that express a variant homodimer. But most patients are compound heterozygotes with mutations that compromise enzyme stability, enzymatic activity or dimerization. So, this approach may work for some patient combinations, but those could be limited.
    • The general applicability of these results to other PMM2 mutations in general, even those that only compromise enzyme stability might be quite limited. Each would need to be tested by this method.
    • If the reduction in PGM1 activity is more generally an applicable treatment for PMM2-CDG patients, appropriately reducing that level would be challenging since PGM1 deficiency is also a known glycosylation disorder which is currently treated with galactose supplementation.
    • The authors checked whether purified mutated PGM1 constructs had any effect on the formation or degradation of Glucose-1,6 bisphosphate, but found none. However, this molecule is a cofactor and stabilizer of PMM2. The authors did not report whether the steady state level of this compound increased.

  6. eLife

    Reviewer #3 (Public Review):

    Vignogna et al. used yeast genetics, experimental evolution and biochemistry to tackle human congenital disorders of glycosylation (CDG), a disease mostly caused by mutations in PMM2. They took advantage of the observation that the budding yeast gene SEC53 is almost identical to human PMM2, and used experimental evolution to find interactors of SEC53/PMM2. They found an overrepresentation of mutations in genes corresponding to other human CDG genes, including PGM1. Genetic and biochemical characterizations of the pgm1 mutations were carried out. This work is solid, although authors did not reveal why reduction of pgm1 activity could compensate for defects of a particular mutant allele of sec53.

    Out of curiosity, if the authors were to simply focus on the preexisting mutations, would they have gotten the materials for most of the experiments in this article? In other words, how important is the experimental evolution?
    A strain table with full genotypes is needed.

  7. Version published to 10.1101/2022.04.06.487342v1 on bioRxiv