Biography of Arthur L. Horwich

Inside cells, crucial helpers latch onto nonnative proteins, preventing them from misfolding. These chaperonins trap unfolded proteins inside molecular cages and facilitate folding to the final native form. Arthur L. Horwich, professor at Yale University School of Medicine (New Haven, CT), uncovered chaperonin action in 1989 (1). More than 15 years later, chaperonins remain the major focus of Horwich's research. In his Inaugural Article in this issue of PNAS (2), Horwich, who was elected to the National Academy of Sciences in 2003, probes the behavior of the bacterial chaperonin GroEL, showing that the unfolded substrate polypeptide is not a passive bystander but has an effect on the activation mechanism.

Figure 1.

Some of the current members of the Horwich laboratory: (left to right) Eunice Park, Wayne Fenton, Krystyna Furtak, Jörg Hinnerwisch, Horwich, Fernando Agarraberes, and George Farr. Photograph courtesy of Yale Biomed Communications.

Hooked from the Start

Horwich was born in 1951 and grew up in Oak Park, IL, west of Chicago. “I always loved science as a kid,” he says. Horwich remembers sitting next to a distant relative at one holiday dinner. He did not know the relative all that well but became enthralled by his stories of being a ham radio operator during World War II. “Within a year, I was a ham operator myself,” he says. Such a hobby, however, required federal certification, and Horwich “got a kick out of taking the federal exam as a 10-year-old.” He passed the exam, but when he started high school he became interested in other science fields. During that time, he thought it would be interesting to see medical ward rounds and accompanied some of his father's friends who were doctors. Horwich remembers, “They were pretty sure that one look at patients would kill any possibility of my being interested—they were wrong about that—it was just the opposite reaction.” He liked that the doctors had “incredible human contact but were still thinking on a basic science level about the clinical problem.” When the time came to choose a college, Horwich felt ready to delve into both science and clinical practice. In 1969, Horwich entered Brown University (Providence, RI) as part of a new program that combined the undergraduate degree with medical school. During medical school, Horwich studied fat cell metabolism in the laboratory of Dr. John Fain. At the end of 6 years, in 1975, Horwich graduated as valedictorian of the first class to complete the combined program.

Horwich went on to do an internship and residency in pediatrics at Yale–New Haven Hospital (New Haven, CT). Midway through, Horwich was not sure about an entirely clinical future: “I got fascinated by the literature on the malignant transforming viruses” like simian virus 40 (3). “The genetic action [of the viruses], with a single gene able to transform a cell, really blew me away,” he says. Thus, a week after completing his residency, Horwich joined the Salk Institute (La Jolla, CA) for a postdoctoral position in molecular biology and virology. At Salk, Horwich worked in Walter Eckhart's laboratory alongside Tony Hunter and witnessed Hunter's discovery of tyrosine phosphorylation (4). Horwich credits this time with sharpening his skills as a scientist: “Tony taught me the nuts and bolts of thinking about a problem.”

Back to New Haven

In 1981, after 3 years at Salk, Horwich contemplated having either a purely research-based career or a hybrid between the clinic and the bench. He found the perfect harmony in the field of medical genetics. At the time, scientists in the field began to clone genes related to mammal disease processes (5). Horwich moved back to New Haven for a postdoctoral fellowship at Yale University Medical School. In the laboratory of Leon Rosenberg, Horwich found “an incredibly fruitful interaction and a wonderful scientific mentor.” Together they cloned the urea cycle gene ornithine transcarbamalase (OTC) (6), an event Horwich cites as the beginning of his current research pathway. OTC had both clinical and cellular attractions as a target of study. OTC deficiency is an X-linked trait that results in toxic, and often lethal, ammonia build-up in affected newborn male infants. Despite being encoded on the X chromosome and translated in the cytosol, the OTC enzyme is used inside the mitochondria, presenting a problem of protein transport that interested Horwich. The cloning work illuminated an N-terminal mitochondrial targeting sequence sufficient to ticket other attached proteins to mitochondria (6). With the cDNA isolated, they also performed DNA diagnosis of the disease (7).

In 1984, Horwich moved across the hall from Rosenberg's laboratory to start his own laboratory as an assistant professor in the department of genetics. He enjoyed the proximity and still collaborated with members of the Rosenberg laboratory, including Wayne Fenton, a collaborator to this day. As an independent researcher, Horwich asked whether the pathway that imported OTC in mammalian cells also could work in yeast. He found that the human enzyme subunit, when expressed in yeast, could be imported faithfully and processed to produce active OTC enzyme in the yeast mitochondria (8). Demonstrating that the pathway was operative in yeast meant that genetic screens were now possible. With a bank of temperature-sensitive lethal yeast mutants containing an inducible copy of OTC, Horwich operated on the basic principle that if mitochondrial protein biogenesis fails, so does the cell. An important observation concerning the import process had been made several years beforehand and ultimately came to the forefront in the mutant studies: imported proteins had to be unfolded to enter the mitochondria (9). After screening mutants for the ability to transport across the membrane or have signal sequences cleaved, Horwich says, “It dawned on us to ask if there is a mutant that imports protein precursors and processes their signal peptide but then fails to fold the mature protein subunits into active forms.”

His hunch was correct, and he quickly isolated the mutant; OTC seemed to have entered the mitochondria and was proteolytically matured, but it had no activity (1). While further tests of the mutant were being carried out, Ulrich Hartl and Walter Neupert at the University of Munich (Germany), experts in import biochemistry, approached Horwich, interested in collaborative study. “When they heard about the folding mutant, they were initially skeptical that lack of folding was the reason for that particular mutant's failure, so they asked if they could have a look at it.” But it did not take Hartl and Neupert long to be convinced. Horwich recalls, “They called back two weeks later and said, `It's clearly inside. It's clearly not reaching native state. Unbelievable!”' The gene involved in this folding-defective behavior, called Hsp60, turned out to encode a heat-inducible subunit of a large double ring assembly found inside mitochondria a year before by Richard Hallberg (10). It now became increasingly clear that this protein was required under all conditions and that it mediated protein folding.

This requirement for a chaperone seemed somewhat heretical, as early work of Christian Anfinsen had shown that a protein could assume its final fold spontaneously, directed simply by its primary amino acid sequence (11). Chaperones had been implicated only in the final steps of oligomeric protein assembly. But later discoveries proved that the principles of Anfinsen were not challenged after all. Chaperones do not carry information for folding; they only provide assistance to proteins under cellular conditions where conditions like high temperature can lead to misfolding and aggregation.

Machinery on a Molecular Level

Hsp60 is a member of a family of oligomeric complexes composed of two rings, each formed from seven subunits. The result is a cylindrical cage with a hydrophobic lining that snares the hydrophobic residues of unfolded proteins (12). The presence of ATP and a molecular lid formed by the cochaperonin GroES results in a protein-folding machine. Horwich has spent more than 15 years teasing apart the workings of this molecular apparatus. He began early on trying to acquire structural information with crystallography through collaboration with a fellow Yale researcher. “I was lucky to be living in the same institution as Paul Sigler,” says Horwich. “At that time it didn't look like there would be enough computer power to deal with structures of this large size.” Three years later, in 1993, they managed to grow what Horwich refers to as “decent crystals,” and even then it was unexpected: “The molecule that worked the best was just an accident.” A PCR error led to a benign amino acid substitution and a resulting protein that crystallized well. “It was really sort of a religious moment to look at the chaperonin for the first time, even at low resolution before phase extension had been carried out,” says Horwich (13).

“The molecule that worked the best was just an accident.”

Within a year, the group constructed 80 functional mutations to begin looking at the protein in action in vitro. They discovered that the apical domains have a hydrophobic binding site where nonnative proteins bind (14), a characteristic that explains GroEL's lack of affinity for native proteins and ability to prevent aggregation of nonnative proteins. Horwich again attributes the result to luck. “We had looked at the wrong surfaces initially by mutational study and then realized that it must be the apical cavity lining. All six or seven of the mutations we placed there in the next round of study killed polypeptide binding,” he says. “We came in one day and saw the result of an in vivo test, and there it was, we had gotten the answers. It was unbelievable.”

The first crystal structures showed the initial binding state, but what about the folding state? This finding required further crystallographic and functional studies, carried out together by the Horwich and Sigler groups. They crystallized a GroEL–GroES complex (15), and “now we could see how this machine moves into folding mode.” Horwich likens it to a jack-in-the-box-type movement. The apical and intermediate domains make rigid body movements that move the hydrophobic surface away from facing the cavity, replacing it with the hydrophilic surface, causing the bound protein to strip off and release into the GroES-lidded chamber where it commences folding (16, 17). But these structural studies of GroEL–GroES complexes at the end of the 1990s revealed a conundrum that led to Horwich's current area of work detailed in his Inaugural Article on page 15005 of this issue of PNAS (2). To trigger proper folding, GroEL, the cylinder, binds ATP and then binds GroES, the lid. ADP also, however, can promote GroES binding and encapsulation of the bound protein but will not promote productive folding (17).

To explain the functional differences in the complexes with ATP versus ADP, Horwich and Sigler initially had crystallized GroEL–GroES–ADP and now wanted to do so with an ATP complex. They used ADP and aluminum fluoride to simulate the γ-phosphate of ATP. Their preliminary data showed a complex virtually identical to the earlier-studied ADP complex (18). Sadly, Horwich lost a collaborator when Sigler passed away in 2000. Horwich forged on to tease out the answer to the question: How could you get to the same end state with ADP or ATP?

Up to this point, all of the complexes had been structurally studied in the absence of substrate, a point first noted by one of Horwich's students: “Then Charu Chaudhry, a graduate student who had started her thesis work with Sigler, who lived, ate, and breathed the project, reminded the group that all of the complexes had been structurally studied in the absence of substrate protein.” The group speculated that the polypeptide itself could be a load on the system. In other words, could the presence of the polypeptide influence the mechanical action of the chaperonin machine? To study the question, his group used fluorescence energy transfer (FRET). Fluorescent tags attached to the apical and equatorial domains of a subunit of the complex reported the apical opening movement that occurred upon binding nucleotide and GroES. With the technique, they could see the machine open in real time. With no substrate, the complex opened rapidly whether ATP or ADP was supplied along with GroES. With unfolded polypeptides, however, the story was different. With ATP, the movement was only marginally slower, but with ADP, there was “essentially no opening of the apical domains,” he says (2). “The machine is stuck in what may be a collision state.” Horwich is planning to study what goes on in this collision state and how it proceeds to the fully open state.

Life Beyond GroEL–GroES–ADP

Chaperonins are not the only things that hold Horwich's attention. Thanks to his eldest son, Mike, a 26-year-old M.D./Ph.D., Horwich has picked up fly-fishing. He and his family, including David, 11, Annie, 23, and wife Martina (also a physician and scientist), play tennis and hike. But when the Connecticut streams freeze over, the tennis courts are buried under feet of snow, and the hiking trails are closed for the season, Horwich has the option of visiting the laboratory space he recently set up as a visiting professor at The Scripps Research Institute (La Jolla, CA). This satellite laboratory gives him access to powerful machines like NMR magnets and single-molecule fluorescence instrumentation to churn out structural information. In collaboration with Professor Kurt Wüthrich, efforts are underway to “see” nonnative protein bound to GroEL with NMR techniques. But for all the high-tech toys, in the end, Horwich is most proud of what was really an accident. “Our initial stumbling across chaperonin function, which was a turning point in the field, got people to buy into the idea of a folding machine”—Horwich included.