Cell, Vol. 92, 291–294, February 6, 1998, Copyright 1998 by Cell PressThe Cell as a Collection Overviewof Protein Machines: Preparing theNext Generation of Molecular Biologistswith the macroscopic world, these protein assembliescontain highly coordinated moving parts. Within eachprotein assembly, intermolecular collisions are not onlyrestricted to a small set of possibilities, but reaction Cdepends on reaction B, which in turn depends on reac-Bruce AlbertsPresident, National Academy of Sciences2101 Constitution Avenue NWWashington, D.C. 20418Professor, Department of Biochemistry and Biophysicstion A—just as it would in a machine of our commonUniversity of California, San Franciscoexperience (Alberts, 1984).San Francisco, California 94143Underlying this highly organized activity are orderedconformational changesin one or more proteins drivenbyIntroductionnucleoside triphosphate hydrolysis (or by other sourcesof energy,such as an iongradient). Because theconfor-We have always underestimated cells. Undoubtedly wemational changes driven in this way dissipate free en-still do today. But at least we are no longer as naive asergy, they generally proceed only in one direction.we were when I was a graduate student in the 1960s.An earlier brief review emphasized how the direction-Then, most of us viewed cells as containing a giantality imparted by nucleoside triphosphate hydrolysesset of second-order reactions: molecules A and B wereallows allosteric proteins to function in three differentthought to diffuse freely, randomly colliding with eachways: asmotorproteinsthatmove ina polarizedfashionother to produce molecule AB—and likewise for thealong afilamentora nucleicacidstrand;as proofreadingmany other molecules that interact with each other in-devices or“clocks”thatincreasethefidelityof biologicalsidea cell. Thisseemed reasonable because, aswehadreactions by screening out poorly matched partners;learned from studying physical chemistry, motions atand as assembly factors that catalyze the formation ofthe scaleof molecules areincredibly rapid. Consider anprotein complexes and are then recycled. (See figure 1enzyme,for example. Ifits substratemoleculeis presentin Alberts and Miake-Lye, 1992.)ataconcentrationof 0.5mM,whichisonly onesubstrateSince the time of that review, the number of proteinmolecule for every 105water molecules, the enzyme’sassemblies that are recognized to employsuch devicesactive sitewill randomly collidewithabout 500,000 mol-has substantially increased. In particular, the nearly ubiq-ecules of substrate per second. And a typical globularuitous use of energy-driven conformational changesprotein will be spinningto andfro, turning aboutvariousto promote the local assembly of protein complexes,axes at rates corresponding to a million rotations perthereby creating a high degree of order in the cell, hassecond.become universally recognized. A simple generic dia-But, as it turns out, we can walk and we can talkgram of such a process is presented in Figure 1.because the chemistry that makes life possible is muchWe have also come to realize that protein assem-more elaborateand sophisticated thananything we stu-bliescanbeenormously complex.Considerfor exampledents had ever considered. Proteins make up most ofthe spliceosome. Composed of 5 small nuclear RNAsthe dry mass of a cell. But instead of a cell dominated(snRNAs) and more than 50 proteins, this machine isby randomly colliding individual protein molecules, wethought to catalyze an ordered sequence of more thannow know that nearly every major process in a cell is10 RNA rearrangements as it removesan intron fromancarried out by assemblies of 10 or more protein mole-RNA transcript. As cogently described in this issue ofcules. And,as itcarries out itsbiological functions, eachCell by Staley and Guthrie (1998), these steps involveof these protein assemblies interacts with several otherat least eight RNA-dependent ATPaseproteins and onelarge complexes of proteins. Indeed, the entire cell canGTPase, each of which is presumed to drive an orderedbe viewed as a factory that contains an elaborate net-conformational change in the spliceosome and/or in itswork of interlocking assembly lines, each of which isbound RNA molecule. As the example of the spliceo-composed of a set of large protein machines.some should make clear, the cartoons thus far used toConsider, as an example, the cell cycle–dependentdepict proteinmachines (e.g., Figure1) vastlyunderesti-degradation of specific proteins that helps to drive amate the sophistication of many of these remarkablecell through mitosis. First a large complex of about 10devices.proteins, the anaphase-promoting complex (APC), se-Given the ubiquity of protein machines in biology, welects out a specific protein for polyubiquitination (Kingshould be seriously attempting a comparative analysiset al., 1996; Zachariae et al., 1996); this protein is thenof all ofthe known machines, with the aimof classifyingtargeted to the proteasome’s 19S cap complex formedthem into types and deriving some general principlesfrom about 20 different subunits; and the cap complexfor future analyses. Some of the methodologies thatthen transfers the targeted protein into the barrel of thehave been derived by the engineers who analyze thelarge 20Sproteasome itself, where it is finally convertedmachines of our common experience are likely to beto small peptides (Baumeister et al., 1998 [this issue]).relevant. For example, modern machines comprised ofsubsystems from different “domains” (i.e., mechanical,Ordered Movements Drive Protein Machineselectrical, fluid, thermal) are often analyzed by an en-Whydowe callthelargeproteinassemblies thatunderlieergy-based approach. Here a mathematical descriptioncell function protein machines? Precisely because, likeof the machineisachievedby considering certain scalarfunctions that represent the system energy (i.e., kineticthe machines invented by humans to deal efficientlyCell292dissipateenergy. Any particular part of a machine mightbe modeledas consisting of one or more of these basicconstituent elements. It seems reasonable to expectthat different, but analogous approaches could profit-ably be applied to the protein machines that underliethe workings of all living things.Should We Expect a Protein Machine to BeWell Engineered?It is not hard to see why protein machines are advanta-geousto cells.A mere glance atthecollectionof articlesin this issue of Cell should