Worms reveal essential functions for intermediate filaments.

I filaments (IF) represent one of the three major cytoskeletal systems found in animal cells. Their somewhat uninspired name was originally derived from their 10-nm diameter, which lies between that of smaller actincontaining microfilaments and larger microtubules. This name, however, belies their importance as critical players in the organization of cells and tissues of vertebrate systems. As to function, it has become obvious from studies of numerous human disorders, such as those that cause blistering diseases of the skin, that IF play important roles in establishing and maintaining the mechanical integrity of cells (1). Depending on the cell type, IF proteins comprise anywhere from 1 to 85% of total cell protein and, despite these quantities, they remain the least studied and understood of all cytoskeletal systems. Historically, there are many reasons for this lack of understanding of their structure and function, the most obvious relates to the fact that the structural proteins that assemble into IF are not highly conserved. For example, humans contain IF that are encoded by over 50 different members of a multigene family. This family is subdivided into six types on the basis of similarities in their amino acid sequences (2). This is in stark contrast to the other cytoskeletal components whose core structures are comprised primarily of the highly conserved subunits of microtubules, a and b tubulin, and actin, the major subunit of microfilaments. There have been many attempts to use the gene knockout approach in the mouse to determine the function of specific types of IF. In some cases, this approach has provided insights into possible functions, but in other cases this approach has not revealed an obvious phenotype. In the case of the Type III desmin knockout mice, there are severe structural defects in cardiac muscles, demonstrating that IF play a critical role in establishing and maintaining the structural integrity of these cells (3). This is also the case for the keratin 8 knockout mouse, in which the majority of embryos die of internal bleeding because of the fragility of hepatocytes in day 12 embryos (4). However, in a different strain of mice, the same K8 knockout phenotype did not produce embryonic lethality. Postnatal examination of these mice showed only colorectal hyperplasia and an inf lammatory response in the lamina propria and submucosa of the intestinal tract (5). Finally, in the case of two other Type III IF gene knockouts, no obvious phenotypes were initially discovered, and mice negative for vimentin (6) and glial fibrillary acidic protein (GFAP; ref. 7 and references therein) survived to reproductive age. However, more recent work has shown more subtle phenotypes in these mice, such as defects in wound healing in the vimentin knockouts (8), and in the response of astrocytes to traumatic brain injury in the GFAP null animals (7). From these studies, it is obvious that analyses of IF function in mouse models are extremely complex and difficult to interpret. Determinations of IF functions are further complicated by the fact that the expression of different IF genes is temporally regulated during development. In many cases, progenitor cells express one type of IF and, as development proceeds, it is gradually replaced by another type of IF system. This often results in the coexpression of two types of IF in differentiating cells. For example, in the earliest stages of development, many cells (e.g., neuroblasts) destined to become neurons in the central nervous system express the Type III IF protein vimentin, and later during differentiation, vimentin is downregulated as the Type IV ‘‘neurofilament triplet’’ proteins are expressed (9). An additional factor distinguishing IF from other cytoskeletal components is related to variations in the number of proteins required for their assembly. For example, in vitro studies have shown that the Type III IF proteins such as vimentin and desmin form homopolymer IF, whereas the Type I and II keratins can form IF only from heterodimers comprised of one of each type of protein chain, and Type IV neurofilaments require three protein chains, NFL, NFM, and NFH, to form proper IF (2). However, recent studies have revealed that most IF structures formed in vivo may be heteropolymers. For example, vimentin, which readily forms homopolymers in vitro, is frequently found as a heteropolymer with nestin, a Type VI IF protein in vivo. In vitro, nestin cannot assemble into IF on its own, but when mixed with vimentin, it combines to form typical IF (10, 11). Likewise, desmin is frequently associated with another Type VI protein, paranemin (12). The emerging picture from a large number of studies suggests that every distinct cell type has a different cytoskeletal IF composition. In addition, there is an ever-growing number of IF-associated proteins (IFAPs) such as plectin (13), which adds another level of complexity to this large family of proteins. Finally, recent studies using green fluorescent protein (GFP)-tagged IF have demonstrated that they are very dynamic and motile structures in vivo (14). Many of their motile properties have been linked to the activities of microtubule-associated motors such as kinesin (14). Taken together, all of these factors make studies of IF function even more challenging. In this issue of PNAS, Karabinos et al. (15) take the first crack at testing IF function in a far less complicated and widely studied genetically approachable organism, Caenorhabditis elegans. By probing the complete worm genome, there now appear to be only 11 genes