Microinjection of monoclonal antibodies specific for one intermediate filament protein in cells containing multiple keratins allow insight into the composition of particular 10 nm filaments

Monoclonal antibodies specific for vimentin (V9), keratin 7 (CK 7) and keratin 18 (CK5) have been microinjected into three human epithelial cell lines: HeLa, MCF-7 and RT-4. The effect of the injection on other keratin polypeptides and vimentin filaments has been observed by double label immunofluorescence and in some instances by immunoelectron microscopy using gold labels of different sizes. Microinjection of V9 into HeLa cells causes the vimentin to collapse into a perinuclear cap leaving the keratin filaments unaffected. Injection of CK5 does not affect the vimentin filaments but disrupts the keratin filaments revealing keratin aggregates similar to those seen in some epithelial cell lines during mitosis. The keratin aggregates obtained after microinjection in HeLa contain the keratins 8 and 18 and probably also other keratins, as no residual keratin filaments are observed with a keratin polyclonal antibody of broad specificity. Aggregates in mitotic HeLa cells contain at least the keratins 7, 8, and 18. In MCF-7 cells keratins 8, 18, and 19 are observed in the aggregates seen 3 h after microinjection which, however, show a different morphology from those seen in HeLa cells. In MCF-7 cells a new keratin filament is built within 6 h after the injection which is composed mainly of keratin 8 and 19. The antibody-complexed keratin 18 remains in spherical aggregates of different size. The results suggest that in HeLa cells vimentin and keratin form independent networks, and that individual 10 nm filaments in epithelial cell lines can contain more than two keratins.     

Tissue specificity of epithelial keratins: differential expression of mRNAs from two multigene families.

Human epithelial cells cultured from stratified and simple squamous tissues all produce keratins of 40,000 to 58,000 daltons, but within this range the number and sizes vary with different epithelial cells. We have shown that this tissue-specific variation in the keratins is not due to posttranslational modification or processing, but rather to the differential expression of a family of heterogeneous but closely related mRNAs. All of these epithelial keratin mRNAs can be further grouped into two distinct subfamilies by their ability to hybridize with either of two cloned epidermal keratin cDNAs. All of the keratin mRNAs hybridize to one or the other, but not both, of the two cloned cDNAs. However, the mRNAs within each group hybridize with varying degrees of stringency, indicating that they are of similar but not identical sequence. Both types of keratin mRNAs are always expressed in every epithelial cell line studied, suggesting that filament assembly is dependent on the presence of both types of keratins. Within each of these two groups, the slight sequence differences in each class may reflect subtle tissue-specific variations in the structural and functional requirements of the epithelial cytoskeleton.     

Cell type-specific expression of bovine keratin genes as demonstrated by the use of complementary DNA clones

Cytokeratins are a family of polypeptides that form the intermediate-sized filament characteristic of epithelial cells. The cytoskeletons of different types of epithelial cells have been reported to possess specific combinations of the members of this protein family. Therefore, we have sought to examine the correspondence between such differential protein expression and the expression of cytokeratin genes at the nucleic acid level. A library of recombinant plasmids carrying cDNA sequences synthesized from bovine epidermal mRNAs was constructed. Clones of about 10(3) base-pairs coding for all the major epidermal keratins of molecular weights of 50,000, 54,000, 59,000, 60,000 and 68,000 were identified by means of hybridization-selection, followed by one and two-dimensional gel electrophoresis of products of translation in vitro. Under stringent conditions, each of these clones hybridizes specifically with its corresponding mRNA and does not show significant cross-hybridization with mRNAs coding for the other keratins, including those belonging to the same subfamily. Using these clones in RNA blot hybridization analysis, we have studied the expression of keratin genes in diverse bovine epithelial tissues (muzzle epidermis, cornea, esophagus, bladder urothelium, liver) and cultured cell lines from kidney (MDBK) and mammary gland (BMGE + H, BMGE -H). In each case we have found a correlation between the respective keratin polypeptides and the corresponding mRNAs. Whereas mRNA coding for keratins Ia and VIb have been found only in epidermis, genes coding for other epidermal keratins are expressed also in certain non-epidermal epithelia and in cells of the BMGE + H line. In contrast, epidermal keratin mRNA sequences have not been detected in liver or bladder tissue, nor in cultured kidney cells (MDBK) or mammary gland cells of the BMGE - H line, which all express a set of cytokeratin polypeptides entirely different from those of epidermis. In all cases, only one mRNA size species has been found, suggesting that in different cell types the same mRNA species is synthesized from the same keratin gene. We conclude that the mechanisms controlling the cell type-specific synthesis of the diverse keratin genes act at a pre-translational level.     

De novo synthesis and specific assembly of keratin filaments in nonepithelial cells after microinjection of mRNA for epidermal keratin

Poly(A)+ RNA isolated from bovine muzzle epidermis was microinjected into nonepithelial cells containing only intermediate-sized filaments of the vimentin type. In recipient cells keratin polypeptides are synthesized and assemble into intermediate-sized filaments at multiple dispersed sites. We describe the time course and the pattern of de novo assembly of keratin filaments within living cells. These filaments were indistinguishable, by immunofluorescence and immunoelectron microscopic criteria, from keratin filament arrays present in true epithelial cells. The presence of extended keratin fibril meshworks in these injected cells is compatible with cell growth and mitosis. Double immunolabeling revealed that newly assembled keratin was not codistributed with microfilament bundles, microtubules or vimentin filaments. We suggest that assembly mechanisms exist which in vivo sort out newly synthesized cytokeratin polypeptides from vimentin.     

Posttranslational regulation of keratins: degradation of mouse and human keratins 18 and 8.

Human keratin 18 (K18) and keratin 8 (K8) and their mouse homologs, Endo B and Endo A, respectively, are expressed in adult mice primarily in a variety of simple epithelial cell types in which they are normally found in equal amounts within the intermediate filament cytoskeleton. Expression of K18 alone in mouse L cells or NIH 3T3 fibroblasts from either the gene or a cDNA expression vector results in K18 protein which is degraded relatively rapidly without the formation of filaments. A K8 cDNA containing all coding sequences was isolated and expressed in mouse fibroblasts either singly or in combination with K18. Immunoprecipitation of stably transfected L cells revealed that when K8 was expressed alone, it was degraded in a fashion similar to that seen previously for K18. However, expression of K8 in fibroblasts that also expressed K18 resulted in stabilization of both K18 and K8. Immunofluorescent staining revealed typical keratin filament organization in such cells. Thus, expression of a type I and a type II keratin was found to be both necessary and sufficient for formation of keratin filaments within fibroblasts. To determine whether a similar proteolytic system responsible for the degradation of K18 in fibroblasts also exists in simple epithelial cells which normally express a type I and a type II keratin, a mutant, truncated K18 protein missing the carboxy-terminal tail domain and a conserved region of the central, alpha-helical rod domain was expressed in mouse parietal endodermal cells. This resulted in destabilization of endogenous Endo A and Endo B and inhibition of the formation of typical keratin filament structures. Therefore, cells that normally express keratins contain a proteolytic system similar to that found in experimentally manipulated fibroblasts which degrades keratin proteins not found in their normal polymerized state.     

Expression of keratin K14 in the epidermis and hair follicle: insights into complex programs of differentiation

Keratins K14 and K5 have long been considered to be biochemical markers of the stratified squamous epithelia, including epidermis (Moll, R., W. Franke, D. Schiller, B. Geiger, and R. Krepler. 1982. Cell. 31:11-24; Nelson, W., and T.-T. Sun. 1983. J. Cell Biol. 97:244-251). When cells of most stratified squamous epithelia differentiate, they downregulate expression of mRNAs encoding these two keratins and induce expression of new sets of keratins specific for individual programs of epithelial differentiation. Frequently, as in the case of epidermis, the expression of differentiation-specific keratins also leads to a reorganization of the keratin filament network, including denser bundling of the keratin fibers. We report here the use of monospecific antisera and cRNA probes to examine the differential expression of keratin K14 in the complex tissue of human skin. Using in situ hybridizations and immunoelectron microscopy, we find that the patterns of K14 expression and filament organization in the hair follicle are strikingly different from epidermis. Some of the mitotically active outer root sheath (ORS) cells, which give rise to ORS under normal circumstances and to epidermis during wound healing, produce only low levels of K14. These cells have fewer keratin filaments than basal epidermal cells, and the filaments are organized into looser, more delicate bundles than is typical for epidermis. As these cells differentiate, they elevate their expression of K14 and produce denser bundles of keratin filaments more typical of epidermis. In contrast to basal cells of epidermis and ORS, matrix cells, which are relatively undifferentiated and which can give rise to inner root sheath, cuticle and hair shaft, show no evidence of K14, K14 mRNA expression, or keratin filament formation. As matrix cells differentiate, they produce hair-specific keratins and dense bundles of keratin filaments but they do not induce K14 expression. Collectively, the patterns of K14 and K14 mRNA expression and filament organization in mitotically active epithelial cells of the skin correlate with their relative degree of pluripotency, and this suggests a possible basis for the deviation of hair follicle programs of differentiation from those of other stratified squamous epithelia.     

Synthesis and fate of keratins 8 and 18 in nonepithelial cells transfected with cDNA

To study the assembly of intermediate filaments in vivo we have transfected fibroblast cell lines with the cDNAs coding for keratins 8 and 18 under the control of the promoter of the SV40 early region and followed keratin expression by RNA hybridization, two-dimensional gel electrophoresis, and immunofluorescence analysis. When expressed individually, keratins 8 and 18 failed to polymerize into intermediate filaments but formed granular aggregates of variable size distributed throughout the cytoplasm as seen by staining with specific antibodies. The expression of one of these two keratins did not induce the synthesis of its partner or of any other keratin. Coexpression of the two keratins produced filamentous structures, frequently perinuclear, indicating that the two types of polypeptides were able to assemble into intermediate filaments but could not form the cytoskeleton characteristic of epithelial cells. These results demonstrate that assembly in heterocomplexes stabilizes keratins against cellular degradation, helping to explain why excess pools of simple keratins have never been detected.     

Characterization of early assembly intermediates of recombinant human keratins

The intermediate filaments (IFs) form major structural elements of the cytoskeleton. In vitro analyses of these fibrous proteins reveal very different assembly properties for the nuclear and cytoplasmic IF proteins. However, keratins in particular, the largest and most heterogenous group of cytoplasmic IF proteins, have been difficult to analyze due to their rapid assembly dynamics under the near-physiological conditions used for other IF proteins. We show here that keratins, like other cytoplasmic IF proteins, go through a stage of assembling into full-width soluble complexes, i.e., "unit-length filaments" (ULFs). In contrast to other IF proteins, however, longitudinal annealing of keratin ULFs into long filaments quasi-coincides with their formation. In vitro assembly of IF proteins into filaments can be initiated by an increase of the ionic strength and/or lowering of the pH of the assembly buffer. We now document that 23-mer peptides from the head domains of various IF proteins can induce filament formation even under conditions of low salt and high pH. This suggests that the "heads" are involved in the formation and longitudinal association of the ULFs. Using a Tris-buffering protocol that causes formation of soluble oligomers at pH 9, the epidermal keratins K5/14 form less regular filaments and less efficiently than the simple epithelial keratins K8/18. In sodium phosphate buffers (pH 7.5), however, K5/14 were able to form long partially unraveled filaments which compacted into extended, regular filaments upon addition of 20 mM KCl. Applying the same assembly regimen to mutant K14 R125H demonstrated that mutations causing a severe disease phenotype and morphological filament abnormalities can form long, regular filaments with surprising efficiency in vitro.     

The human keratins: biology and pathology

The keratins are the typical intermediate filament proteins of epithelia, showing an outstanding degree of molecular diversity. Heteropolymeric filaments are formed by pairing of type I and type II molecules. In humans 54 functional keratin genes exist. They are expressed in highly specific patterns related to the epithelial type and stage of cellular differentiation. About half of all keratins--including numerous keratins characterized only recently--are restricted to the various compartments of hair follicles. As part of the epithelial cytoskeleton, keratins are important for the mechanical stability and integrity of epithelial cells and tissues. Moreover, some keratins also have regulatory functions and are involved in intracellular signaling pathways, e.g. protection from stress, wound healing, and apoptosis. Applying the new consensus nomenclature, this article summarizes, for all human keratins, their cell type and tissue distribution and their functional significance in relation to transgenic mouse models and human hereditary keratin diseases. Furthermore, since keratins also exhibit characteristic expression patterns in human tumors, several of them (notably K5, K7, K8/K18, K19, and K20) have great importance in immunohistochemical tumor diagnosis of carcinomas, in particular of unclear metastases and in precise classification and subtyping. Future research might open further fields of clinical application for this remarkable protein family.     

Lessons from disorders of epidermal differentiation-associated keratins

A number of diseases have been associated with mutations in genes encoding keratin intermediate filaments. Several of these disorders have skin manifestations, in which histological changes highlight the role of various different keratins in epidermal differentiation. For example, mutations in either K1 or K10 (the major keratin pair expressed in differentiated keratinocytes) usually lead to clumped keratin filaments and cytolysis. Furthermore, the precise nature of the mutation has direct implications for disease phenotype. Specifically, mutations in the H1 and alpha-helical rod domains of K1/K10 result in bullous congenital ichthyosiform erythroderma, underscoring the critical role for this keratin filament domain in maintaining cellular integrity. However, a lysine to isoleucine substitution in the V1 domain of K1 underlies a form of palmoplantar keratoderma, which has different cell biological implications. Keratins are cross-linked into the cornified cell envelopes through this particular lysine residue and the consequences of the mutation lead to changes in keratin-desmosome association and cornified cell morphology, suggesting a role for this keratin subdomain in cornified cell envelope formation. Recently, to extend genotype-phenotype correlation, a frameshift mutation in the V2 region of the K1 tail domain was identified in ichthyosis hystrix (Curth-Macklin type), in which keratin filaments show a characteristic shell-like structure and fail to form proper bundles. In this case, the association of desmosomes with loricrin was also altered, implicating this keratin domain in organizing the intracellular distribution of loricrin during cornification. Collectively, these mutations in K1/K10 provide a fascinating insight into both normal and abnormal processes of epidermal differentiation.     

Identification of two types of keratin polypeptides within the acidic cytokeratin subfamily I

Cytoskeletal filaments of the α-keratin type (cytokeratins) are a characteristic of epithelial cells. In diverse mammals (man, cow and rodents) these cytokeratins consist of a family of approximately 20 polypeptides, which may be divided into the more acidic (I) and the more basic (II) subfamilies. These two subfamilies show only limited amino acid sequence homology. In contrast, nucleic acid hybridization experiments and peptide maps have been interpreted to show that polypeptides of the same subfamily share extended sequence homology.We compare two polypeptides of the acidic cytokeratin subfamily, VIb (M_r 54,000) and VII (M_r 50,000), which are co-expressed in large amounts in bovine epidermal keratinocytes. These two epidermal keratins can be distinguished by specific antibodies and show different patterns of expression among several bovine tissues and cultured cells. In addition, they differ in the stability of their complexes with basic keratin polypeptides and in their tryptic peptide maps. The amino acid sequences deduced from the nucleotide sequences of complementary DNA clones containing the 3′ ends of the messenger RNAs for these keratins are compared with each other and with available amino acid sequences of human, murine and amphibian epidermal keratins. Bovine keratins VIb and VII share considerable sequence homology in the α-helical portion (68% residues identical) but lack significant homology in the extrahelical portion. Bovine keratin VIb shows, in its α-helical region, a pronounced sequence homology (88% identity) to the murine epidermal keratin of M_r 59,000. In addition, the non-helical carboxy-terminal regions of both proteins are glycinerich and contain a canonic sequence GGG^S_GYGG, which may be repeated several times. Moreover, their mRNAs present a highly conserved stretch of 236 nucleotides containing, in the murine sequence, the end of the coding and all of the non-coding region (81% identical nucleotides). Bovine keratin VII is considerably different from the murine M_r 59,000 keratin but is almost identical to the human cytokeratin number 14 of M_r 50,000, both in the α-helical and in the non-α-helical regions of the proteins, and the mRNAs of the human and the bovine keratins also display a high homology in their 3′ non-coding ends.The results show that in the same species keratins of the same subfamily can differ considerably, whereas equivalent keratin polypeptides of different species are readily identified by characteristic sequence homologies in the α-helical and the non-helical regions as well as in the 3′ non-coding portions of their mRNAs. Among the members of the acidic subfamily I of cytokeratin polypeptides that are co-expressed in bovine epidermis, at least two types can be distinguished by their carboxy-terminal sequences. One type is characterized by its abundance of glycine residues, a consensus GGG^S_GYGG heptapeptide sequence, which may be repeated several times, and an extended stretch of high RNA sequence homology in the 3′ non-coding part. The other type shows a predominance of serine and valine residues, a subterminal GGG^S_GYGG sequence (which has been maintained in Xenopus, cow and man) and also a high level of homology in the 3′ non-coding part of the mRNA. The data indicate that individual keratin type specificity overrides species diversity, both at the protein and the mRNA level. We discuss the evolutionary conservation and the tissue distribution of these two types of acidic keratin polypeptides as well as their possible biological functions.     

Immunogold labeling of keratin filaments in normal human epidermal cells with two anti-keratin monoclonal antibodies

We report on application of the highly sensitive and specific immunogold labeling method for ultrastructural investigation of keratin intermediate filament antigens in human epidermal cell suspensions. Triton X-100 pretreated cells proved accessible to the colloidal gold conjugate, thus enabling keratin filament bundles to be labeled. Anti-keratin KL1 and KL2 monoclonal antibodies were raised in mice after immunization with either human stratum corneum-isolated keratins or keratins extracted from human epidermal cells suspensions, respectively. Immunoelectron microscopy confirmed immunofluorescence and immunoperoxidase results of epidermal keratinocyte staining, and revealed two different antibody reactivity patterns: KL2 reacted with keratin filaments in keratinocytes of all epidermal layers, whereas antigen to KL1 was detected only on keratin of the suprabasal layers, not on the basal keratinocyte tonofilaments. The monoclonal antibody-recognized epitopes were specific for the keratin filaments. Vimentin-rich cells (melanocytes) were not stained in the same epidermal cell suspensions. Additionally, two distinct ultrastructural patterns of keratin filament epitope labeling were observed. KL1 and KL2 monoclonal antibodies react with two different antigenic determinants, depending on the stage of keratinocyte differentiation, and may therefore be used for immunohistochemical studies of various keratin-containing cells in normal and pathologic conditions.     

Embryogenesis of stratified squamous epithelium. II. Developmental stage-specific immunoreactivity of keratins

Throughout embryogenesis of the rat palate, from the early fetal to adult stages, a consistent subset of keratin proteins is synthesized in the epithelial lining cells. Although the relative abundance of particular keratins has been shown to vary with ongoing palatogenesis, the expression of finite keratins appears to be genetically predetermined. In order to preliminarily ascertain whether conformational changes accompanied intermediate filament 'maturation' from monomeric to polymeric keratin formation, we screened cytokeratins with polyclonal and monoclonal antibodies generated against adult-type keratins. Until epithelial stratification occurred on the 16th day of gestation, the keratin proteins were weakly immunoreactive. On the other hand, subsequent to epithelial thickening, adult-type immunoreactivity was initiated and progressed concomitantly with ongoing palatal development. These findings suggest that the cytokeratin intermediate filaments may progress through conformational 'maturation' during polymerization, and play a role in the eventual acquisition of the adult-type epithelial structure and function.     

Deletions in epidermal keratins leading to alterations in filament organization in vivo and in intermediate filament assembly in vitro

To investigate the sequences important for assembly of keratins into 10- nm filaments, we used a combined approach of (a) transfection of mutant keratin cDNAs into epithelial cells in vivo, and (b) in vitro assembly of mutant and wild-type keratins. Keratin K14 mutants missing the nonhelical carboxy- and amino-terminal domains not only integrated without perturbation into endogenous keratin filament networks in vivo, but they also formed 10-nm filaments with K5 in vitro. Surprisingly, keratin mutants missing the highly conserved L L E G E sequence, common to all intermediate filament proteins and found at the carboxy end of the alpha-helical rod domain, also assembled into filaments with only a somewhat reduced efficiency. Even a carboxy K14 mutant missing approximately 10% of the rod assembled into filaments, although in this case filaments aggregated significantly. Despite the ability of these mutants to form filaments in vitro, they often perturbed keratin filament organization in vivo. In contrast, small truncations in the amino-terminal end of the rod domain more severely disrupted the filament assembly process in vitro as well as in vivo, and in particular restricted elongation. For both carboxy and amino rod deletions, the more extensive the deletion, the more severe the phenotype. Surprisingly, while elongation could be almost quantitatively blocked with large mutations, tetramer formation and higher ordered lateral interactions still occurred. Collectively, our in vitro data (a) provide a molecular basis for the dominance of our mutants in vivo, (b) offer new insights as to why different mutants may generate different phenotypes in vivo, and (c) delineate the limit sequences necessary for K14 to both incorporate properly into a preexisting keratin filament network in vivo and assemble efficiently into 10-nm keratin filaments in vitro.     

Making a connection: direct binding between keratin intermediate filaments and desmosomal proteins

In epidermal cells, keratin intermediate filaments connect with desmosomes to form extensive cadherin-mediated cytoskeletal architectures. Desmoplakin (DPI), a desmosomal component lacking a transmembrane domain, has been implicated in this interaction, although most studies have been conducted with cells that contain few or no desmosomes, and efforts to demonstrate direct interactions between desmoplakin and intermediate filaments have not been successful. In this report, we explore the biochemical nature of the connections between keratin filaments and desmosomes in epidermal keratinocytes. We show that the carboxy terminal "tail" of DPI associates directly with the amino terminal "head" of type II epidermal keratins, including K1, K2, K5, and K6. We have engineered and purified recombinant K5 head and DPI tail, and we demonstrate direct interaction in vitro by solution- binding assays and by ligand blot assays. This marked association is not seen with simple epithelial type II keratins, vimentin, or with type I keratins, providing a possible explanation for the greater stability of the epidermal keratin filament architecture over that of other cell types. We have identified an 18-amino acid residue stretch in the K5 head that is conserved only among type II epidermal keratins and that appears to play some role in DPI tail binding. This finding might have important implications for understanding a recent point mutation found within this binding site in a family with a blistering skin disorder.     

The intermediate filament system of the keratinizing mouse forestomach epithelium: Coexpression of keratins of internal squamous epithelia and of epidermal keratins in differentiating cells

Summary The internal epithelium of mouse forestomach represents a fully keratinized tissue that has many morphological aspects in common with the integumental epidermis. In the present study we have, therefore, analyzed keratin expression in the total epithelium, in subfractions of basal cells and in living and dead suprabasal cells that were obtained by Percoll density gradient centrifugation of trypsin-dissociated forestomach keratinocytes. The keratin analysis revealed that basal forestomach keratinocytes synthesize the same keratin types as basal epidermal cells (60 000, 52 000 and 47 000 daltons), whereas differentiating cells contain both the epidermal suprabasal keratin pair (67 000 and 59 000 daltons) and the suprabasal keratin pair characteristic for other internal squamous epithelia (57 000 and 47 000 daltons). Indirect immunofluorescence using an antibody recognizing the members of the epidermal-type suprabasal keratin pair and in-situ-hybridization experiments using specific cDNA probes for the members of the internal-type keratin pair showed that the two keratin pairs are uniformly coexpressed in living suprabasal forestomach keratinocytes. Furthermore, it could be shown that distinct cells in the basal cell layer acquire the ability to express both the 67 000/59 000 dalton and the 57 000/47 000 dalton keratin pair and that some basal cells apparently lose the ability to synthesize mRNAs for basal keratins.     

Mouse differentiation-specific keratins 1 and 10 require a preexisting keratin scaffold to form a filament network

Keratins 1 (K1) and 10 (K10) are the predominant cytoskeletal intermediate filaments of epidermal cells during transition from the proliferative to the terminal differentiation stage. In situ, formation of the K1/K10 intermediate filament network occurs in the cytoplasm of cells with a preexisting cytoskeleton composed of keratins 5 and 14. To define cytoskeletal interactions permissive for formation of the K1/K10 filamentous network, active copies of mouse K1 and K10 genes were introduced into fibroblasts (NIH 3T3) which do not normally express these proteins. Transient and stable transfectants, as well as heterokaryons produced by fusions with epithelial cells, were evaluated for expression of K1 and K10 proteins and filament formation using specific antibodies. In contrast to keratin pairs K5/K14 and K8/K18, the K1/K10 pair failed to form an extensive keratin filament network on its own, although small isolated dense K1/K10 filament bundles were observed throughout the cytoplasm by EM. K1 and K10 filaments integrated only into the preexisting K5/K14 network upon fusion of the NIH 3T3 (K1/K10) cells with epithelial cells expressing endogenous K5/K14 or with NIH 3T3 cells which were transfected with active copies of the K5 and K14 genes. When combinations of active recombinant gene constructs for keratins 1, 5, 10, and 14 were tested in transient NIH 3T3 transfections, the most intact cytokeratin network observed by immunofluorescence was formed by the K5/K14 pair. The K1/K14 pair was capable of forming a cytoskeletal network, but the network was poorly developed, and usually perinuclear. Transfection of K10 in combination with K5 or K1 resulted in cytoplasmic agglomerates, but not a cytoskeleton. These results suggest that the formation of the suprabasal cytoskeleton in epidermis is dependent on the preexisting basal cell intermediate filament network. Furthermore, restrictions on filament formation appear to be more stringent for K10 than for K1.