Clarifying results and prevailing models for the evolution of Puf proteins and their RNA targets

The PNAS paper “Recurrent rewiring and emergence of RNA regulatory networks” by Wilinski et al. (1) has a number of flaws. The paper incorrectly interprets the data underlying its major, novel claim about target rewiring. The paper also presents models that conflict with previous results and does not consider existing models that do explain all available results.

First, the major novel result from Wilinski et al. (1) is that Neurospora crassa Puf4/5 targets are enriched for cytosolic ribosomal protein mRNAs and that these interactions have been conserved since the common ancestor of Saccharomycotina and Pezizomycotina. This result conflicts with published results (2) and is not supported by Wilinski et al.’s (1) data. Reanalysis of the ribosomal protein mRNAs listed in the paper’s (1) dataset S2 (tab Nc_Puf45, line 7) shows that the overwhelming majority are mitochondrial ribosomal proteins (30 of 36), leaving only six that could be cytoplasmic, which is not statistically significant. Thus, the N. crassa Puf4/5 targets are enriched with mitochondrial ribosomal protein mRNAs and not cytosolic ones, confirming the previous result (2).

Second, the majority of results presented by Wilinski et al. (1) have been previously published (2–5). The authors (1) take claim for previously published results in several cases. In one series as an example, Wilinski et al. (1) take claim for previous results (2) that identified Puf4 and Puf5 as paralogs, that Puf4 was duplicated in Saccharomycotina, and that Puf4 and Puf5 binding targets diverged in specificity after duplication. Previous work provided evidence that the ancestral Puf4 had broad binding specificity and that Saccharomycotina Puf4 and Puf5 both diverged and specialized in binding specificity after the duplication (2); this model not only conflicts with the model presented by Wilinksi et al. (1), but the results published in Hogan et al. (2) do not support their model.

As a final example, Wilinski et al. (1) propose that both Puf3 and Puf4 were bound to mitochondrial targets in the ancestor of Pezizomycotina and Saccharomycotina fungi, then Puf3 binding sites were lost in Pezizomycotina, and Puf4 binding sites were lost in Saccharomycotina. This model also conflicts with published results (2). By studying early-diverging lineages of Pezizomycotina fungi, previous work (2) identified species in which both Puf3 and Puf4 bind mRNAs encoding mitochondrial proteins (hereafter “mitochondrial targets”). This result led to a model of the evolutionary history in which Puf3 is the ancestral regulator of mitochondrial targets; within the Pezizomycotina lineage, not before it, regulation transitioned to Puf4 through an intermediate in which both Puf3 and Puf4 bound and each had a selective advantage. Thus, binding of mitochondrial targets by both Puf3 and Puf4 in early-diverging Pezizomycotina lineages provides evidence against the Wilinski et al. (1) model.

Accurately accounting for the history of Puf protein and target evolution is an important step toward understanding how and why changes have occurred. Maintaining scholarly rigor when considering all available data and models is important for progress in the field.