HomeResearchProtein Folding in the Cell

Our Scientists

Protein Folding in the Cell

Research Summary

Arthur Horwich has been interested in understanding how chaperonin ring machines assist protein folding in the cell and, more recently, in discovering how a misfolded protein causes motor neuron degeneration resembling ALS (Lou Gehrig's disease).

The final step of information transfer from DNA to effector protein involves the proper folding of a newly translated polypeptide into its biologically active, native form. Although all of the information necessary to direct this process is contained in the sequence of amino acid building blocks that are strung together to make a polypeptide chain, there is a class of specialized proteins known as molecular chaperones that provide assistance to the folding process in the intact cell, in particular preventing polypeptide chains from aggregating with each other during the folding process in the cellular environment, where the temperature is relatively high and where macromolecules are present at high concentration.

The common feature recognized by molecular chaperones is exposed hydrophobic surface, which is ultimately buried to the interior of a folded, native protein. Exposed hydrophobic surfaces allow nonnative proteins to stick to each other, causing multimolecular aggregation, a state that carries no biological activity and that also can be damaging to the cell, as in a number of neurodegenerative diseases. Molecular chaperones prevent such aggregation from occurring by proffering their own hydrophobic binding sites, capturing nonnative proteins. Following release, a chaperoned protein has a fresh chance to fold properly. We focused for many years on one type of chaperone, called a chaperonin, a double-ring "machine" that has the remarkable ability to promote proper folding of an initially captured protein to its native state inside an encapsulated chamber.

GroEL, the bacterial chaperonin that we studied, assists proper folding through two principal actions. One action involves binding a nonnative protein in an open ring, via a hydrophobic lining of the ring that recognizes exposed hydrophobic surfaces in the substrate protein. Such binding can serve to "mask" hydrophobic surfaces, preventing them from aggregating, and can even reverse misfolding by effectively pulling the protein conformation apart, giving it a fresh chance for correct folding. The second assisting action involves releasing the protein, in the presence of ATP and a small "lid" structure, GroES, into a now GroES-encapsulated chamber where folding can proceed in solitary confinement in a chamber, without the chance of aggregation. The walls of the GroES-bound chamber are electrostatic in character, and our studies of the past few years suggest that this is a "no-stick" surface that does not interfere with the folding process, allowing it to occur as if at infinite dilution in a sea of water.

A Misfolding-Induced Neurodegenerative Disease
Mutations in the abundant cytosolic free radical–scavenging enzyme superoxide dismutase (SOD1) have been associated with ~2% of cases of amyotrophic lateral sclerosis (ALS; Lou Gehrig's disease). Overexpression of such mutant forms of SOD1 in mice likewise leads to motor neuron dysfunction and death, associated with muscle weakness and ultimately paralysis. Studies have shown that, as with other neurodegenerative diseases, disease is not due to a loss of function (because, for example, many patients retain normal enzyme activity), but rather to a toxic gain of function related to misfolding and aggregation.

We have engineered a transgenic mouse strain that overproduces a fusion protein of a completely misfolded mutant version of SOD1, G85R, which is unable to fold properly, with the yellow fluorescent protein (YFP). Unlike a wild-type SOD1-YFP transgenic mouse expressing the same amount of protein, these animals develop paralysis by 5–6 months of age. This paralysis is associated with the presence of YFP-fluorescent aggregates in their motor neuron cell bodies (appearing as early as 1–2 months of age). Such aggregation occurs despite observed association of a substantial amount of the misfolded protein with the molecular chaperone Hsc70.

At early times before clinical disease, we also observe a number of abnormalities in neurons, the nature and order of which we seek to understand. For example, we observe many enlarged mitochondria in cell bodies of the motor neurons; we observe shriveled myelin sheaths of axons innervating the lower extremities; and neuromuscular junctions in the lower extremities are abnormal, lacking synaptic vesicles. We are seeking to understand how the mutant protein interacts with components of these systems to produce toxic injury and are carrying out a variety of experiments aimed at protecting the animals from development of disease.

Recently, we have been able to model toxicity to axonal transport of vesicles using purified mutant SOD1-YFP and axoplasm extruded from squid giant axon, in a collaboration carried out at the Marine Biology Lab in Woods Hole, MA, with Drs. Yuyu Song (Harvard Medical School) and Scott Brady (University of Illinois, Chicago). The added mutant protein blocks forward transport of vesicles, which is associated with activation of a kinase cascade (including the kinases ASK1 and p38). Remarkably, addition of molecular chaperones corrected both of these abnormalities, the chaperone Hsc70 partially correcting them, and the nucleotide exchange factor Hsp110, which interacts with Hsc70, completely correcting the transport defect and reversing kinase activation. Interestingly, we and others have previously found that Hsc70 is associated with misfolded SOD1 in the spinal cords of mutant SOD1-containing transgenic mice. We have also observed that Hsp110 is associated with mutant SOD1 as well. Further, we have observed Hsp110 expression is increased in motor neurons from the spinal cords of our mutant animals. These observations raise the question of whether supplying additional molecular chaperone function can have effects on SOD1-linked ALS.

In a further study of our mice, we have characterized the intrinsic firing properties of spinal cord motor neurons, using spinal cord slices, prepared in a novel way that preserves motor neuron viability and allows for patch-clamp recording. In normal animals, recordings identified four distinct firing types of motor neuron, most easily classifiable by firing frequency.  We observed, using a fluorescent probe injected into muscles and retrograde-transported into spinal cord motor neuron cell bodies, that fast firing motor neurons innervate fast twitch muscle, while slow firing neurons innervate slow twitch muscle. In the ALS animals, all four motor neuron firing types were present at 2 months of age. Strikingly, by 4 months of age, a time where minor ALS symptomatology has commenced, the fastest firing type of motor neuron was absent. This correlated with loss of ~50 percent of motor neurons by this age, associated with aggregation of mutant SOD1 protein preferentially in fast firing motor neurons. The results offer a possibility of further understanding motor neuron vulnerability in ALS.     

Grants from the National Institutes of Health supported a portion of this work.

As of December 22, 2014

Scientist Profile

Yale University
Cell Biology, Neuroscience