Tagged for Cleansing

A career switch brought me to the most satisfying science I had ever done. I was lucky enough to come across one of the most clean and straightforward of cellular processes: the ubiquitin-proteasome system (UPS). Here, a series of enzymes attach a chain of ubiquitin molecules to other proteins, marking them for degradation by the proteasome. The beauty of studying degradation is that you are examining a process in which the protein is first present but then disappears. It is almost like the binary system, in which only two states are possible: 0 or 1. It is a fast process, eliminating a large number of proteins in a matter of minutes, and it can be location specific—inducing degradation of proteins within certain subcellular compartments while leaving others untouched. Finally, with thousands of enzymes and different components, the UPS is highly specific when selecting targets—the family of ubiquitin ligases is the largest family of enzymes in eukaryotes. It is equal to phosphorylation in complexity and importance but lacks the background: no noise, and no shades of gray.

I imagine the cell as a network of molecular machines that are perfectly synchronized with each other. Different gears of these machines (e.g., cellular enzymes) need to appear (be synthesized) at the right time and then disappear (be degraded) once their function is executed. In a sense, regulation by ubiquitin-mediated proteolysis is the most definite and elegant of regulatory systems—no reversible modifications, just simple elimination.

I grew up in Napoli, Italy. Every day, an hour before starting my regular high school classes, I would knock on the door of Professore Moncharmont and begin the next part of the experiment he had prepared for me. I grew bacteria using different media or followed the movement of paramecia in a Petri dish when a slight electrical charge was applied to the water, recording my observations and results in my laboratory notebook. I would sit in Professore Moncharmont's laboratory and he would tell me about the life of a scientist—the excitement of discovery, the international travel and community. I was hooked!

After high school, I relented to my father's wishes and studied medicine. He was a cardiologist at a local hospital in Napoli and expected that I follow in his footsteps. Bitten by the research bug, I could not stay away from the experimental aspects of the field. In 1989, I earned my MD, but also snuck in a specialty diploma in molecular endocrinology (a sort of Italian equivalent of a PhD). My experimental thesis focused on the effects of phosphorylation on the hormone-binding ability of the estrogen receptor.

P27 binds to cyclin dependent kinases (CDK), preventing them from pushing the cell through division. When we looked at the levels of the p27 protein in cells, we saw them drop at the threshold of G1 and S-phase. The levels of p27 gene transcription and translation, however, remained constant throughout the cell cycle. It was our first evidence to suggest that degradation played a major role in cell cycle progression.

In 1990, I moved to Germany for postdoctoral training with Giulio Draetta at the European Molecular Biology Laboratory (EMBL), where I worked on cyclin-dependent kinases (CDKs), which drive the cell division cycle by phosphorylating a number of important regulatory proteins. The build-up of CDKs in the cell was thought to control the cell's entry into mitosis (the final part of cell replication just before the cell physically divides into two daughter cells). But Giulio had just discovered that one CDK, activated by cyclin A, was active many hours earlier. With my experience in phosphorylation, I started working on this problem and proved that cyclin A regulates the S-phase, in which the cell replicates its DNA in preparation for cell division.

Only two and a half years into my postdoctoral training at the EMBL, Giulio moved to Cambridge, Massachusetts, launching a new biotech company called Mitotix. I joined him and became a scientific co-founder of the company, investigating CDK inhibitors. As key cell cycle regulators, CDKs were, and still are, of keen interest to biotech companies as potential targets for cancer treatment. I decided to work on the regulation of CDKs by the inhibitor p27, which had only recently been discovered and for which I felt there was a bit less competition since, at the time, most cell cycle labs were focusing on the CDK inhibitors p16 and p21.

My first step was to look at the levels of p27 across the stages of the cell cycle. I observed that p27 levels change depending on the phase; it was high during the cell's quiescent phase, but it started to drop late in G1 when, in response to nutrients and growth factors, the cell commits to a round of cell division and readies the machinery for replicating its genome (see graphic above). However, when I looked at the expression of the gene encoding p27, I noticed that the mRNA levels, as well as its translation into protein, were constant throughout the cell cycle. If the cell was churning out p27 at a constant rate, why was I getting variable protein levels in the cell across the cell cycle? It had to be degradation. Measuring the half-life of p27 protein revealed that p27 was indeed being degraded at that critical moment in the cell cycle when CDK activation is required for the consequent push into S-phase.

At the time, researchers didn't know that ubiquitin played such a central role in degrading important cellular regulators. For years, scientists had focused on how proteins are made, not on how they are eliminated. It was assumed that, with time, proteins grew old and eventually became oxidized, misfolded, damaged, and, finally, degraded by cellular organelles, such as lysosomes. When the ubiquitin system was discovered by Avram Hershko and colleagues at the Technion-Israel Institute of Technology in Haifa, people realized that proteolysis was not only an entropic process but also energy-dependent and regulated. However, for years, many scientists regarded the ubiquitin system as a pathway specialized for the proteolysis of particularly short-lived proteins.

Hershko's group characterized the general mechanism of ubiquitin attachment to proteins in the early 1980s. It was a three-enzyme process. E1 (enzyme 1) binds and "activates" ubiquitin molecules and transfers them to one of many E2s (or Ubc, for ubiquitin-conjugating enzymes). With the help of one of many E3s (or ubiquitin ligases), each E2 conjugates a ubiquitin chain to the proteins that will be degraded (see graphic below). Ultimately, the E3 confers specificity to the entire system by bringing specific proteins to the E2s.

I suspected that ubiquitin might be the culprit in my system, but there were several other candidates. It all seemed to fall into place when, using the yeast two-hybrid system, our group found that a protein called ubiquitin-conjugating enzyme 9 (Ubc9) was interacting with the C-terminus of the p27. Encouraged by this result, we designed an in vitro system to verify that p27 was being degraded via ubiquitin tagging. Little did we know that Ubc9 was a common false positive in two-hybrid experiments! By a stroke of luck, that artifactual (and unpublished) result had steered us in the right direction. Our in vitro system revealed that p27 was indeed ubiquitylated, not by Ubc9, of course, but by Ubc3.¹

The resulting paper garnered a lot of attention, in part because it demonstrated a major new role for the ubiquitin system in controlling the mammalian cell cycle. Not long after we published our results, I received a call from the father of the field himself, Avram Hershko. Avram went on to win the 2004 Nobel Prize in Chemistry for his work on ubiquitin, but by 1995, he was already well recognized for his contribution. I was taken aback when I realized who was on the line and flattered by his congratulations and interest in our work. In the months that followed, I spoke to Avram several more times. I quickly came to regard him a mentor and turned to him for guidance during my next steps as a researcher. To me, he was, and still is, a model of what a pure scientist should be: someone unconcerned with prestige, politics, and power—someone simply driven to answer a scientific question. In part because Avram instilled in me so much excitement for pure research, I decided to return to academia. I applied for several positions, and in late 1996, I moved to the New York University School of Medicine to start my career as a principal investigator.

One of my first projects there was a collaboration with Massimo Loda, a cancer pathologist at the Dana Farber Cancer Institute, to explore p27's role in human cancer. Everything we knew about p27 indicated that it should be a tumor suppressor, but no one could produce genetic evidence. The problem, however, was that they were looking in the wrong place. Armed with the understanding that degradation, and not gene expression, was the driving force behind the regulation of p27 abundance in the cell, we studied protein levels rather than mutations in stage II colorectal cancers. We chose stage II cancers because we knew that some cases progress slowly and others become very aggressive. By comparing the p27 levels in these samples and then watching the patients' outcomes, we observed that the lower the p27 levels, the more aggressive the tumor.² Since then, many other groups have used p27 as a prognostic marker for other types of cancer. Importantly, in the same study, we showed that low levels of p27 in aggressive tumors are often due to its enhanced degradation rather than its decreased transcription and translation.

In the summer of 1997, I met Avram in person at a Federation of American Societies for Experimental Biology (FASEB) meeting on the UPS in Vermont. We spent many hours sitting on the grass outside of the meeting center discussing science. While I was young and a relative newcomer to the ubiquitin field, he and I saw eye to eye, so I was delighted when he suggested coming to visit my lab during his summer sabbaticals from his institution. By then, I was totally focused on the ubiquitin system. I was captivated by how much influence ubiquitin held in the cell. For example, with every step in the cycle of cell division, the UPS removes all of the enzymes and factors that controlled the previous stage while allowing new enzymes and factors, such as the cyclins, to accumulate, driving the cell cycle forward and making it unidirectional. We knew that p27 was degraded by the UPS, but what gave the UPS system the specificity to degrade p27 at just the right moment in the cell cycle?

Other groups had shown that, in yeast, Ubc3 works with E3 enzyme complexes called SCFs. The SCF complexes are made up of three subunits—Skp1, Cul1, and one of many F-box proteins—the last of which gives the E3 its specificity for degradation (see graphic below). We thought that if we could find all of the human F-box proteins, we'd be one step closer to understanding what controlled p27, and the degradation of many other cellular proteins.

The E1 enzyme transfers ubiquitin (Ub) to an E2, which then binds E3 Ub ligases, such as the SCF protein complex (so called for its Skp1, Cul1 and F-box protein components). The E2 adds its Ub to the growing tail of Ub molecules on a substrate, which is bound to the E3, thus marking the substrate for proteasome-mediated degradation. F-box proteins like Skp2 are specific to a small number of substrates, a specificity that allows the Ub-system to regulate the cell processes.

During 1997 and 1998, my lab used the yeast two-hybrid system and human Skp1 as bait to fish for human F-box proteins. Using this technique, as well as computer predictions, we identified 26 F-box proteins.³ In the summer of 1998, Avram arrived for the first of seven summer sabbaticals in my lab. He took a bench next to Andrea Carrano, a very talented graduate student, and started working immediately on the puzzle of p27 degradation. Together, we found that the F-box protein called Skp2 specifically binds p27 and promotes its ubiquitylation. The timing of this event is at least partially dictated by the availability of Skp2, whose levels are low during quiescence and increase in late G1.⁴

Although the Skp2 result was an important step, when we took purified component proteins and tried to replicate the ubiquitylation of p27 in a test tube, the experiment failed unless a crude cell extract was added to the reaction—we were missing something that has to be present in the cell extract. Avram, with his experience in protein purification, fractionated the cell extract to find the missing component. The protein he found was Cks1, which we already knew bound CDKs but whose biochemical functions were unknown. Finally, a collaborative effort with Nikola Pavletich's lab at the Memorial Sloan-Kettering Cancer Center in New York led to a crystal structure of the Skp1-Skp2-Cks1-p27 complex, providing important insights into the mechanism of catalysis by the SCF.

We had all the proteins that regulate the degradation of p27, but what regulated the function and availability of Skp2 and Cks1? We dove into the problem, looking for all of the proteins that influenced the degradation of p27 by regulating Skp2 and Cks1, finding proteins that either acted on Skp2 or on other effector proteins upstream of Skp2 and Cks1. At the same time we started the characterization of a second F-box protein, βTrCP, which we found to act both as a suppressor of Cdk1 activity and as an activator of Cdk1, depending on the cell cycle phase.

After the human genome was published and we could simply scan the genome for homologous sequences, the number of F-box proteins jumped from 26 to 69. Of those 69, we and others had only discovered substrates for approximately nine. By 2005, although we hadn't necessarily finished characterizing every part of the cell cycle regulated by Skp2 and βTrCP, we started to have less fun in only working out the smaller details. We decided, instead, to focus on those orphan F-box proteins whose function was still unknown but that were suggested to play a role in cell proliferation or cell cycle checkpoints. The first part of the plan: Find a method to identify substrates of orphan F-box proteins.

Historically, it has been much easier to identify the F-box protein starting from the substrate (as we used p27 to identify Skp2) than vice versa (i.e., to find substrates of a specific F-box protein). Unpublished negative results from our and other labs indicated that standard approaches such as the yeast two-hybrid system used to identify interacting proteins would not be successful in finding substrates of F-box proteins.

The problem was a tall order by most standards, but it was just the kind of challenge that two postdoctoral fellows in the lab, Angelo Peschiaroli and Valerio Dorrello, were looking for. A small trick led us to the substrates. We developed an immunopurification method that enriched for ubiquitylated substrates, based on the ability of SCF ligases to promote the in vitro ubiquitylation of bound substrates, and on the subsequent, selective purification of the ubiquitylated proteins. To validate our assay, we chose βTrCP for two main reasons. First, several substrates of βTrCP were already known, which would help to assess the feasibility and validity of our method. Second, the existence of a characterized βTrCP binding motif would facilitate the identification of the novel substrates.

Through the process of testing and optimizing our technique, we discovered a number of new βTrCP substrates that were not involved in cell cycle control. Two surprises were that βTrCP controlled the degradation of a protein called PDCD4, which is an inhibitor of protein synthesis, and BimEL, a powerful factor that induces cell death. We found that in response to mitogenic stimuli, both proteins are degraded via βTrCP to allow protein synthesis (and therefore cell growth) and cell survival. We also identified a transcriptional repressor, REST, as a novel substrate of βTrCP that needs to be degraded to allow the transcription of a "genome guardian," Mad2.

Soon, the mantra of the lab became: "one postdoc/graduate student, one F-box protein," although some people in the lab do work on other subjects. When we began looking at the substrates of orphan F-box proteins, we ventured even further into new territories when we saw that some F-box proteins were tightly involved in degrading the master regulator molecules of the circadian clock, while others control epigenetic mechanisms by tagging histone H2A with a single ubiquitin molecule rather than a chain. The emerging picture is that certain F-box proteins coordinate the proteolysis of several dozen substrates, while others appear to have a very specialized function and direct the ubiquitylation of only one or two proteins.

Our unbiased proteomic screens have led me far from my original expertise in the regulation of cell cycle progression. However, at the same time, this journey has increased my enjoyment of studying biology since I have had to learn new systems and familiarize myself with new literatures. I have never understood those labs that focus mainly on one single gene (no matter how important it is) or on a single pathway and, in fact, I no longer want to work on just cell cycle control. The large family of F-box proteins has offered me the wonderful opportunity and luxury to work on multiple, critical, and distinct cellular processes. I hope that, being outsiders, we can bring a fresh and novel prospective to the new fields we enter. As we discover the biological roles of more orphan F-box proteins, the picture of this fascinating system will no doubt become still more interesting.

Have a comment? E-mail us at mail@the-scientist.com

Michele Pagano is a Howard Hughes Medical Institute investigator, the Ellen and Gerald Ritter Professor of Oncology at the Department of Pathology, and the Deputy Director of the New York University Cancer Institute at the NYU School of Medicine.