Can Kayatekin Ph.D.

POSTDOC PROJEC‍‍‍TS

Protein perturbation sensors reveal unique signatures of human pro‍‍‍teotoxicities.

The Lindquist and Khalil labs recently published a method, yTRAP, which uses protein-transcription factor fusions to track the solubility of proteins in the cell (Newby, Kiriakov, and Hallaçlı, et al., Cell 2017). Working in close collaboration with another postdoc in the lab, Erinç Hallaçlı, we are using these yTRAP sensors as readouts for any changes in the protein state. We have generated libraries of yTRAP fusions and are using these libraries as prec‍‍‍ise tools to track the interference caused by toxic protein expression in cells. We have discovered that each toxic protein has a unique yTRAP perturbational signature and we are using changes in yTRAP signal to uncover the mechanisms of proteotoxicity.

I created a new yeast model to tackle one of the most prevalent protein misfolding diseases, diabetes, which afflicts nearly 10% of the American population. Deposits of aggregated IAPP that are frequently present in the pancreatic islets of patients with type-2 diabetes, which contribute to β-cell death and dysfunction. I modeled IAPP toxicity by expressing IAPP oligomers in the secretory pathway of yeast, which produced a profound growth defect. Using genome-wide overexpression and deletion screening in this yeast IAPP model, we found that the strongest suppressor of toxicity was the overexpression of Ste24, a protease responsible for digesting peptides that clog the cytoplasm-to-ER protein-conducting channel. We validated the protective role of Ste24 in a rat insulinoma cell line, demonstrating that the mammalian Ste24 homolog, ZMPSTE24, also serves this protective function. Leveraging the strong evolutionary conservation of this protease we examined the functional consequences of clinically relevant amino acid changes in ZMPSTE24 in our yeast IAPP model. (Look for the paper in the March 22nd issue of Cell)

The translocon declogger Ste24 protects cells against IAPP oligomer toxicity.

Prion-like domains interact with polyglutamine and direct its aggregation into a non-toxic form.

Using genetic screens in yeast, I discovered that proteins with prion-like domains could modify polyglutamine (polyQ) toxicity.  Although prions are typically associated with disease, in our yeast model of polyQ toxicity and in complementary experiments with human cell lines, proteins with glutamine-rich domains resembling those of yeast prions suppressed polyQ toxicity. Intriguingly, this protective effect was recapitulated by simply expressing the prion-like domain of the protein. These domains co-aggregated with an otherwise highly toxic 103-glutamine expanded huntingtin exon 1 protein (Htt103Q), forming non-toxic ag‍‍‍gregates and eliminating populations of diffusible oligomeric species. Thus, although polyQ was aggregated in either case, when it co-aggregated with a prion-like domain, it formed a benign aggregate. This work showed how a single protein can produce multiple aggregated forms, with different effects on the cell. (Kayatekin and Matlack et al., PNAS 2014)‍‍‍

Prions as‍‍‍ heritable, reversible regulators of protein function.‍‍‍

I have been deeply interested in the biology of prions since starting in the Lindquist lab. In yeast many proteins have been discovered with the capacity to take on a self-perpetuating aggregated from, i.e. a prion. In many cases these prions confer benign phenotypes under optimal growth conditions and can be beneficial or harmful under non-optimal conditions. Thus, we hypothesize that they allow the yeast to sample greater phenotypic diversity without permanent changes to its genetic code.  I have been involved with several prion-related projects in the lab and plan to pursue prion research in my independent work. Image: Genetically identical yeast carrying different strains of the [PSI+] prion.‍‍‍

GRADUATE PROJECTS

KINETICS AND THERMODNAMICS OF SUPEROXIDE DISMUTASE FOLDING‍‍‍

The misfolding of Cu, Zn Superoxide Dismutase is one of the leading causes of familial amyotrophic lateral sclerosis. Mutations are spread throughout the amino acid sequence (above) and cover nearly every secondary structure element of the protein. ‍‍‍My graduate work was dedicated to understanding the folding properties of this protein and how ALS-causing variants predispose what is normally an incredibly stable protein to unfold and aggregate. ‍‍‍

Enthalpic barriers limit SOD1 monomer folding, making it a very slow folde‍‍‍r

ALS variants have the largest effect on the metal-free, disulfide-reduced SOD1 monomer.

The rate-limiting step in the formation of the native dimeric state SOD1 is an excruciatingly slow monomer folding reaction that governs the lifetime of its unfolded state.  To determine the thermodynamic properties of the transition state ensemble limiting the folding of this high‐contact‐order β-sandwich motif, we performed a combined thermal and urea denaturation thermodynamic and kinetic analysis and discovered that the barriers to folding and unfolding are dominated by the activation enthalpy. In contrast, the activation entropy is favorable and reduces the barrier height for both reactions. The absence of secondary structure formation or large-scale chain collapse prior to crossing the barrier for folding led us to conclude that the dehydration of non-polar surfaces in the transition state ensemble are responsible for the large and positive activation enthalpy. These results indicated a crucial role for water in dictating the rate-limiting step of SOD1 folding, thereby enhancing its aggregation propensity. (Kayatekin C, Cohen NR, and Matthews CR., JMB 2012)‍‍‍

Through quantitative measurements of protein folding and stability, we discovered that newly synthesized SOD1, lacking the intramolecular disulfide bond and metal ions, were the most destabilized by ALS-variants. For instance, in this form, the common A4V variant was half unfolded under physiological conditions. Curiously, loss of the disulfide bond also reduced the apparent Zn affinity of the SOD1 monomer by 750-fold, raising it into the nanomolar range, where it may be unable to compete for free Zn in the cell. These results indicated that disulfide-‍‍‍bond formation may be the critical step in SOD1 maturation, after which misfolding becomes much less likely. Thus, we hypothesized that the newly synthesized SOD1 molecules lacking the disulfide bond were a likely candidate for aggregation. (Kayatekin C, Zitzewitz JA, and Matthews CR, JMB2010)‍‍‍

Zinc binding stabilizes the entire folding free energy surface of SOD1.

Supposing that aggregates of partially-folded states are primarily responsible for toxicity, we examined the role of the structurally-important zinc ion in stabilizing the folding free energy surface of dimeric SOD1 using thermodynamic and kinetic analyses. The presence of zinc was found to decrease the free energies not only the fully folded SOD1 dimer and folded monomer, but also a peptide model of the unfolded state of SOD1. The unfolded state bound zinc weakly with a micromolar dissociation constant, while the folded monomeric intermediate and the native dimeric form both bind zinc tightly, with sub-nanomolar dissociation constants. This progressively tighter binding of zinc across the folding free energy surface shifted the populations of SOD1 towards more well-folded states. This decrease in the populations of unfolded and less-well folded species would be expected to diminish the potential for aggregation by itself, but zinc binding had yet another beneficial effect. It also increased the folding rate of SOD1 by nearly 100 fold, thereby dramatically shortening the lifetime of the unfolded state. (Kayatekin C, Zitzewitz JA, and Matthews CR, JMB 2008)‍‍‍

USING YEAST TO STUDY PROTEIN AGGREGATION

Many fundamental cellular pathways are conserved from yeast to humans, including those that are often thought to be impacted in protein aggregation diseases. By using yeast models of proteotoxicity, we can evaluate the impact of toxic human proteins in one of the most tractable model organisms. In the Lindquist lab, I developed and studied various proteotoxic models in yeast. Below are brief descriptions of some of this work.