Mutational analysis of the karmellae-inducing signal in Hmg1p, a yeast HMG-CoA reductase isozyme

In response to elevated levels of HMG-CoA reductase, an integral endoplasmic reticulum (ER) membrane protein, cells assemble novel ER arrays. These membranes provide useful models for exploration of ER structure and function, as well as general features of membrane biogenesis and turnover. Yeast express two functional HMG-CoA reductase isozymes, Hmg1p and Hmg2p, each of which induces morphologically different ER arrays. Hmg1p induces stacks of paired nuclear-associated membranes called karmellae. In contrast, Hmg2p induces peripheral ER membrane arrays and short nuclear-associated membrane stacks. In spite of their ability to induce different cellular responses, both Hmg1p and Hmg2p have similar structures, including a polytopic membrane domain containing eight predicted transmembrane helices. By examining a series of recombinant HMG-CoA reductase proteins, our laboratory previously demonstrated that the last ER-lumenal loop (Loop G) of the Hmg1p membrane domain contains a signal needed for proper karmellae assembly. Our goal was to examine the primary sequence requirements within Loop G that were critical for proper function of this signal. To this end, we randomly mutagenized the Loop G sequence, expressed the mutagenized Hmg1p in yeast, and screened for inability to generate karmellae at wild-type levels. Out of approximately 4000 strains with Loop G mutations, we isolated 57 that were unable to induce wild-type levels of karmellae assembly. Twenty-nine of these mutants contained one or more point mutations in the Loop G sequence, including nine single point mutants, four of which had severe defects in karmellae assembly. Comparison of these mutations to single point mutations that did not affect karmellae assembly did not reveal obvious patterns of sequence requirements. For example, both conservative and non-conservative changes were present in both groups and changes that altered the total charge of the Loop G region were observed in both groups. Our hypothesis is that Loop G serves as a karmellae-inducing signal by mediating protein-protein or protein-lipid interactions and that amino acids revealed by this analysis may be important for maintaining the proper secondary structure needed for these interactions.