roller phenotype from RNAi?

Hi All,

Does anyone know whether there are any genes that produce a good roller phenotype with feeding RNAi? This is for a class demonstration, so I am not picky about which gene it is. From Wormbase, it seems that RNAi on many rol genes produces either no phenotype or lethality. RNAi for rol-4 says you see rollers but also Dpys and some lethality. I would like to get the most robust and clean roller phenotype possible so the students don’t get confused. I have the Ahringer RNAi library, so I should have what I need in the freezer.




I believe that the roller phenotypes arise from dominant missense mutations in some of the cuticle collagens. These mutations lead to a strangely kinked collagen subunit the is twisted and leads to a twisted cuticle. As far as I know, decreasing any of the cuticle collagens with RNAi will not lead to a roller phenotype. This is true of rol-4 (sqt-3) and rol-6.

On the other hand, a strong DPY phenotype can be quite distinctive. I am not sure which cuticle protein would give you the strongest phenotype.


rol-5 RNAi works well, while rol-6 RNAi does not (su1006 is a gain of function allele and the sequence is used as an co-injection marker).

I agree that rol-6 RNAi is ineffective at producting a Rol phenotype, but I should point out that the canonical allele of rol-6, e187 (, is a recessive loss-of-function mutation. It is true that we all use the stronger, neomorphic, gain-of-function allele for co-injection markers. But logically, one would expect a Rol phenotype from rol-6 RNAi, but I am guessing the efficiency of knocking down this gene is poor.


Using an RNAi hypersensitive strain might be good to get loss of function rollers out. As Steve points out, that is the rol-6 lf phenotype.

Another possibility is the Sqt genes, where heterozygotes are Rol but homozygotes are Sqt? Presumably sqt RNAi could induce a Rol phenotype.

My recommendation for a class would be unc-22: RNAi causes the coolest Unc phenotype there is.

e187 is recessive for Rol, but is missense, so it could easily be recessive gain-of-function. A better case can be made by reference to n1268 and n1270, which are identical W36ochre stop mutations and so are likely null mutations (n1178 is a late amber, and so less obviously a null).