The two-chain coiled-coil molecule of native epidermal keratin intermediate filaments is a type I-type II heterodimer.

The composition of the two-chain coiled-coil molecule of murine epidermal keratin intermediate filaments (KIF) containing keratins 1 (type II) and 10 (type I) has been explored using native-type KIF as well as KIF reassembled in vitro from protein dissolved in urea solutions or from mixtures of 3H-labeled and unlabeled purified chains. By use of cross-linking, high resolution polyacrylamide gel electrophoresis and blotting for 3H-labeled keratins or with an anti-mouse keratin 10 antibody, it was found that individual keratin chains form type I or type II homodimers and homotetramers in solution that do not assemble into KIF in vitro. When mixed in urea solutions of 5 M or greater, such homo-oligomers rapidly rearrange into mostly heterodimers and heterotetramers that support filament assembly. On the other hand, prekeratin, isolated from newborn mouse epidermis with 0.1 M sodium citrate buffer, pH 2.6, under conditions that do not dissociate the native coiled-coil molecule, consists exclusively of type I-II heterodimers and heterotetramers. It is necessary to dissolve prekeratin in 8-9.5 M urea for several hours in order to dissociate the native heterodimer molecule and incorporate tracer amounts of a single 3H-labeled keratin chain. These data establish that native KIF consist of heterodimer coiled-coil molecules. Furthermore, heterodimers are much more stable than homodimers and are the favored form of association in solution. However, some homodimers (10-30% of total) always form after dissolution in concentrated urea and can be assimilated into KIF during reassembly in vitro. The isolation of alpha-helix-enriched dimer particles from the 2B region of the rod domains upon limited proteolysis confirmed the presence of mostly heterodimer and some homodimer molecules in reassembled KIF. These properties of keratin chains in urea solutions hereby clarify a number of conflicting reports in the literature concerning the composition of the coiled-coil molecule. The presence of some homo-oligomeric species in reassembled KIF correlates with earlier observations of polymorphism as well as stoichiometry.     

The in vitro assembly of hair follicle keratins: comparison of cortex and companion layer keratins

The hair follicle consists of a complex system of multiple tissue compartments that are clearly distinguishable by their morphology and type of differentiation. We have synthesized hair follicle-specific keratins from the companion layer (K6hf, K17) and the hair cortex (Ha1, Hb3, Hb6) in Escherichia coli. The assembly of purified keratins in mixtures of K6hf/K17 and in mixtures of hair cortex keratins was compared in urea solutions, low ionic strength and physiological strength buffers, by urea melting gels, electron microscopy and analytical ultracentrifugation. Both types of keratin mixtures, keratins from the companion layer and keratins from the hair cortex, formed heterotypic complexes at 5 M urea. In low ionic strength buffers, the keratins from the companion layer were assembled to bona fide intermediate filaments. In contrast, mixtures of hair cortex keratins stayed in an oligomeric state with a mean s value of 9 as determined in sedimentation velocity experiments. Hair cortex keratins were, however, assembled into intermediate filaments at physiological salt conditions. A point mutated hair cortex keratin [Hb6(Glu402Lys)] formed no long filaments when mixed with Ha1; instead, the assembled structures showed a length distribution of 50.8 +/- 13.4 nm, comparable to the size distribution of assembly intermediates called 'unit-length' filaments.     

Differential extraction of keratin subunits and filaments from normal human epidermis

We have investigated keratin interactions in vivo by sequentially extracting water-insoluble proteins from normal human epidermis with increasing concentrations of urea (2, 4, 6, and 9.5 M) and examining each extract by one- and two-dimensional gel electrophoresis, immunoblot analysis using monoclonal anti-keratin antibodies, and EM. The viable layers of normal human epidermis contain keratins K1, K2, K5, K10/11, K14, and K15, which are sequentially expressed during the course of epidermal differentiation. Only keratins K5, K14, and K15, which are synthesized by epidermal basal cells, were solubilized in 2 M urea. Extraction of keratins K1, K2, and K10/11, which are expressed only in differentiating suprabasal cells, required 4-6 M urea. Negative staining of the 2-M urea extract revealed predominantly keratin filament subunits, whereas abundant intermediate-sized filaments were observed in the 4-urea and 6-M urea extracts. These results indicate that in normal human epidermis, keratins K5, K14, and K15 are more soluble than the differentiation-specific keratins K1, K2, and K10/11. This finding suggests that native keratin filaments of different polypeptide composition have differing properties, despite their similar morphology. Furthermore, the observation of stable filaments in 4 and 6 M urea suggests that epidermal keratins K1, K2, and K10/11, which ultimately form the bulk of the protective, nonviable stratum corneum, may comprise filaments that are unusually resistant to denaturation.     

14-3-3 proteins associate with phosphorylated simple epithelial keratins during cell cycle progression and act as a solubility cofactor

14-3-3 is a ubiquitous protein family that interacts with several signal transduction kinases. We show that 14-3-3 proteins associate with keratin intermediate filament polypeptides 8 and 18 (K8/18) that are expressed in simple-type epithelia. The association is stoichiometrically significant (> or = one 14-3-3 molecule/keratin tetramer), occurs preferentially with K18, and is phosphorylation- and cell cycle-dependent in that it occurs during S/G2/M phases of the cell cycle when keratins become hyperphosphorylated. Binding of phospho- K8/18 to 14-3-3 can be reconstituted in vitro using recombinant 14-3-3 or using total cellular cytosol. Phosphatase treatment results in dissociation of 14-3-3, and dephosphorylation of phospho-K8/18 prevents reconstitution of the binding. Three cellular keratin subpopulations were analyzed that showed parallel gradients of keratin phosphorylation and 14-3-3 binding. Incubation of 14-3-3 with keratins during or after in vitro filament assembly results in sequestering of additional soluble keratin, only in cases when the keratins were hyperphosphorylated. Our results demonstrate a stoichiometrically significant cell cycle- and phosphorylation-regulated binding of 14-3-3 proteins to K18 and in vitro evidence of a simple epithelial keratin sequestering role for 14-3-3 proteins.     

Complex formation and kinetics of filament assembly exhibited by the simple epithelial keratins K8 and K18

We have generated human recombinant keratins K8 and K18 and describe conditions to quantitatively follow their assembly into filaments. When renatured individually from 8M urea into a low ionic strength/high pH-buffer, K8 was present in a dimeric to tetrameric form as revealed by analytical ultracentrifugation. In contrast, K18 sedimented as a monomer. When mixed in 8 M urea and renatured together, K8 and K18 exhibited s-value profiles compatible with homogeneous tetrameric complexes. This finding was confirmed by sedimentation equilibrium centrifugation. Subsequently, these tetrameric starter units were subjected to assembly experiments at various protein concentrations. At low values such as 0.0025 g/l, unit-length filaments were abundantly present after 2s of assembly. During the following 5 min, filaments grew rapidly and by measuring the length of individual filaments we were able to generate time-dependent length profiles. These data revealed that keratins K8/K18 assemble several times faster than vimentin and desmin. In addition, we determined the persistence length l(p) of K8/K18 filaments to be in the range of 300 nm. Addition of 1 mM MgCl(2) increases l(p) to 480 nm indicating that magnesium ions affect the interaction of keratin subunits within the filament during assembly to some extent.     

The role of keratin subfamilies and keratin pairs in the formation of human epidermal intermediate filaments

The four major keratins of normal human epidermis (molecular mass 50, 56.5, 58, and 65-67 kD) can be subdivided on the basis of charge into two subfamilies (acidic 50-kD and 56.5-kD keratins vs. relatively basic 58-kD and 65-67-kD keratins) or subdivided on the basis of co-expression into two "pairs" (50-kD/58-kD keratin pair synthesized by basal cells vs. 56.5-kD/65-67-kD keratin pair expressed in suprabasal cells). Acidic and basic subfamilies were separated by ion exchange chromatography in 8.5 M urea and tested for their ability to reassemble into 10-nm filaments in vitro. The two keratins in either subfamily did not reassemble into 10-nm filaments unless combined with members of the other subfamily. While electron microscopy of acidic and basic keratins equilibrated in 4.5 M urea showed that keratins within each subfamily formed distinct oligomeric structures, possibly representing precursors in filament assembly, chemical cross-linking followed by gel analysis revealed dimers and larger oligomers only when subfamilies were combined. In addition, among the four major keratins, the acidic 50-kD and basic 58-kD keratins showed preferential association even in 8.5 M urea, enabling us to isolate a 50-kD/58-kD keratin complex by gel filtration. This isolated 50-kD/58-kD keratin pair readily formed 10-nm filaments in vitro. These results demonstrate that in tissues containing multiple keratins, two keratins are sufficient for filament assembly, but one keratin from each subfamily is required. More importantly, these data provide the first evidence for the structural significance of specific co-expressed acidic/basic keratin pairs in the formation of epithelial 10-nm filaments.     

A nontetrameric species is the major soluble form of keratin in Xenopus oocytes and rabbit reticulocyte lysates

Inside the interphase cell, approximately 5% of the total intermediate filament protein exists in a soluble form. Past studies using velocity gradient sedimentation (VGS) indicate that soluble intermediate filament protein exists as an approximately 7 S tetrameric species. While studying intermediate filament assembly dynamics in the Xenopus oocyte, we used both VGS and size-exclusion chromatography (SEC) to analyze the soluble form of keratin. Previous studies (Coulombe, P. A., and E. Fuchs. 1990. J. Cell Biol. 111:153) report that tetrameric keratins migrate on SEC with an apparent molecular weight of approximately 150,000; the major soluble form of keratin in the oocyte, in contrast, migrates with an apparent molecular weight of approximately 750,000. During oocyte maturation, the keratin system disassembles into a soluble form (Klymkowsky, M. W., L. A. Maynell, and C. Nislow. 1991. J. Cell Biol. 114:787) and the amount of the 750-kD keratin complex increases dramatically. Immunoprecipitation analysis of soluble keratin from matured oocytes revealed the presence of type I and type II keratins, but no other stoichiometrically associated polypeptides, suggesting that the 750-kD keratin complex is composed solely of keratin. To further study the formation of the 750-kD keratin complex, we used rabbit reticulocyte lysates (RRL). The 750-kD keratin complex was formed in RRLs contranslating type I and type II Xenopus keratins, but not when lysates translated type I or type II keratin RNAs alone. The 750-kD keratin complex could be formed posttranslationally in an ATP-independent manner when type I and type II keratin translation reactions were mixed. Under conditions of prolonged incubation, such as occur during VGS analysis, the 750-kD keratin complex disassembled into a 7 S (by VGS), 150-kD (by SEC) form. In urea denaturation studies, the 7 S/150-kD form could be further disassembled into an 80-kD species that consists of cofractionating dimeric and monomeric keratin. Based on these results, the 750-kD species appears to be a supratetrameric complex of keratins and is the major, soluble form of keratin in both prophase and M-phase oocytes, and RRL reactions.     

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.     

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.     

Onset of re-epithelialization after skin injury correlates with a reorganization of keratin filaments in wound edge keratinocytes: defining a potential role for keratin 16

Injury to stratified epithelia causes a strong induction of keratins 6 (K6) and 16 (K16) in post-mitotic keratinocytes located at the wound edge. We show that induction of K6 and K16 occurs within 6 h after injury to human epidermis. Their subsequent accumulation in keratinocytes correlates with the profound reorganization of keratin filaments from a pan-cytoplasmic distribution to one in which filaments are aggregated in a juxtanuclear location, opposite to the direction of cell migration. This filament reorganization coincides with additional cytoarchitectural changes and the onset of re-epithelialization after 18 h post-injury. By following the assembly of K6 and K16 in vitro and in cultured cells, we find that relative to K5 and K14, a well- characterized keratin pair that is constitutively expressed in epidermis, K6 and K16 polymerize into short 10-nm filaments that accumulate near the nucleus, a property arising from K16. Forced expression of human K16 in skin keratinocytes of transgenic mice causes a retraction of keratin filaments from the cell periphery, often in a polarized fashion. These results imply that K16 may not have a primary structural function akin to epidermal keratins. Rather, they suggest that in the context of epidermal wound healing, the function of K16 could be to promote a reorganization of the cytoplasmic array of keratin filaments, an event that precedes the onset of keratinocyte migration into the wound site.     

Do the ends justify the mean? Proline mutations at the ends of the keratin coiled-coil rod segment are more disruptive than internal mutations

Intermediate filament (IF) assembly is remarkable, in that it appears to be self-driven by the primary sequence of IF proteins, a family (40-220 kd) with diverse sequences, but similar secondary structures. Each IF polypeptide has a central 310 amino acid residue alpha-helical rod domain, involved in coiled-coil dinner formation. Two short (approximately 10 amino acid residue) stretches at the ends of this rod are more highly conserved than the rest, although the molecular basis for this is unknown. In addition, the rod is segmented by three short nonhelical linkers of conserved location, but not sequence. To examine the degree to which different conserved helical and nonhelical rod sequences contribute to dimer, tetramer, and higher ordered interactions, we introduced proline mutations in residues throughout the rod of a type I keratin, and we removed existing proline residues from the linker regions. To further probe the role of the rod ends, we introduced more subtle mutations near the COOH-terminus. We examined the consequences of these mutations on (a) IF network formation in vivo, and (b) 10-nm filament assembly in vitro. Surprisingly, all proline mutations located deep in the coiled-coil rod segment showed rather modest effects on filament network formation and 10-nm filament assembly. In addition, removing the existing proline residues was without apparent effect in vivo, and in vitro, these mutants assembled into 10-nm filaments with a tendency to aggregate, but with otherwise normal appearance. The most striking effects on filament network formation and IF assembly were observed with mutations at the very ends of the rod. These data indicate that sequences throughout the rod are not equal with respect to their role in filament network formation and in 10-nm filament assembly. Specifically, while the internal rod segments seem able to tolerate considerable changes in alpha-helical conformation, the conserved ends seem to be essential for creating a very specific structure, in which even small perturbations can lead to loss of IF stability and disruption of normal cellular interactions. These findings have important implications for the disease Epidermolysis Bullosa Simplex, arising from point mutations in keratins K5 or K14.     

Sorting out IF networks: consequences of domain swapping on IF recognition and assembly

Vimentin and keratin are coexpressed in many cells, but they segregate into two distinct intermediate filament (IF) networks. To understand the molecular basis for the sorting out of these IF subunits, we genetically engineered cDNAs encoding hybrid IF proteins composed of part vimentin and part type I keratin. When these cDNAs were transiently expressed in cells containing vimentin, keratin, or both IFs, the hybrid IF proteins all recognized one or the other or both networks. The ability to distinguish networks was dependent upon which segments of IF proteins were present in each construct. Constructs containing sequences encoding either helix 1B or helix 2B seemed to be the most critical in conferring IF recognition. At least for type I keratins, recognition was exerted at the level of dimer formation with wild-type type II keratin, as demonstrated by anion exchange chromatography. Interestingly, despite the fact that swapping of helical domains was not as deleterious to IF structure/function as deletion of helical domains, keratin/vimentin hybrids still caused structural aberrations in one or more of the cytoplasmic IF network. Thus, sequence diversity among IF proteins seems to influence not only coiled-coil but also higher ordered associations leading to 10-nm filament formation and/or IF interactions with other cellular organelles/proteins.     

Functional Differences between Keratins of Stratified and Simple Epithelia

Dividing populations of stratified and simple epithelial tissues express keratins 5 and 14, and keratins 8 and 18, respectively. It has been suggested that these keratins form a mechanical framework important to cellular integrity, since their absence gives rise to a blistering skin disorder in neonatal epidermis, and hemorrhaging within the embryonic liver. An unresolved fundamental issue is whether different keratins perform unique functions in epithelia. We now address this question using transgenic technology to express a K16-14 hybrid epidermal keratin transgene and a K18 simple epithelial keratin transgene in the epidermis of mice null for K14. Under conditions where the hybrid epidermal keratin restored a wild-type phenotype to newborn epidermis, K18 partially but not fully rescued. The explanation does not appear to reside in an inability of K18 to form 10-nm filaments with K5, which it does in vitro and in vivo. Rather, it appears that the keratin network formed between K5 and K18 is deficient in withstanding mechanical stress, leading to perturbations in the keratin network in regions of the skin that are subjected either to natural or to mechanically induced trauma. Taken together, these findings suggest that the loss of a type I epidermal keratin cannot be fully compensated by its counterpart of simple epithelial cells, and that in vivo, all keratins are not equivalent.     

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.     

Characterization of the major physiologic phosphorylation site of human keratin 19 and its role in filament organization

Keratin polypeptide 19 (K19) is a type I intermediate filament protein that is expressed in stratified and simple-type epithelia. Little is known regarding K19 regulation or function, and the only other type I keratin that has been studied in terms of regulation is keratin 18 (K18). We characterized K19 phosphorylation as a handle to study its function. In vivo, serine is the major phosphorylated residue, and phosphopeptide mapping of 32PO4-labeled K19 generates one major phosphopeptide. Edman degradation suggested that the radiolabeled phosphopeptide represents K19 Ser-10 and/or Ser-35 phosphorylation. Mutation of Ser-10 or Ser-35 followed by transfection confirmed that Ser-35 is the major K19 phosphorylation site. Transfection of Ser-35 --> Ala K19 showed a filament assembly defect as compared with normal or with Ser-10 --> Ala K19. Comparison of K18 and K19 phosphorylation features in interphase cells showed that both are phosphorylated primarily at a single site, preferentially in the soluble versus the insoluble keratin fractions. K19 has higher basal phosphorylation, whereas K18 phosphorylation is far more sensitive to phosphatase type I and IIA inhibition. Our results demonstrate that Ser-35 is the major K19 interphase phosphorylation site and that it plays a role in keratin filament assembly. K19 and K18 phosphorylations share some features but also have distinct properties that suggest different regulation of type I keratins within the same cells.     

Retrovirus-mediated transgenic keratin expression in cultured fibroblasts: specific domain functions in keratin stabilization and filament formation.

With retrovirus-mediated gene transfer, we used intact and deleted keratin proteins to investigate the molecular basis of intermediate filament function. Three levels of assembly show a different stringency for the involvement of individual keratin domains: protein accumulation requires the alpha helix domains; stable filament formation additionally requires both N- and C-terminal domains of either one of the two interacting keratins, suggesting that head to tail homotypic interaction is important for effective elongation; and higher order organization of the cytoplasmic network depends on correct type I-type II pairing of keratins. The presence of two distinct interaction sites along potentially different axes may explain the characteristic morphology of keratin intermediate filament networks.     

Genomic organization and amplification of the human epidermal type II keratin genes K1 and K5

Keratins are a family of structurally related proteins that form the intermediate filament cytoskeleton in epithelial cells. Mutations in K1 and K5 result in the autosomal dominant disorders epidermolytic hyperkeratosis/bullous congenital ichthyosiform erythroderma and epidermolysis bullosa simplex, respectively. Most disease-associated mutations are within exons encoding protein domains involved in keratin filament assembly. However, some mutations occur outside the mutation hot-spots and may perturb intermolecular interactions between keratins and other proteins, usually with milder clinical consequences. To screen the entire keratin 1 and keratin 5 genes we have characterized their intron-exon organization. The keratin 1 gene comprises 9 exons spanning approximately 5.6 kb on 12q, and the keratin 5 gene comprises 9 exons spanning approximately 6.1 kb on 12q. We have also developed a comprehensive PCR-based mutation detection strategy using primers placed on flanking introns followed by direct sequencing of the PCR products.     

Expression of mutant keratin cDNAs in epithelial cells reveals possible mechanisms for initiation and assembly of intermediate filaments

We have deleted cDNA sequences encoding portions of the amino- and carboxy-terminal end of a human type I epidermal keratin K14, and examined the molecular consequences of forcing the expression of these mutants in simple epithelial and squamous cell carcinoma lines. To follow the expression of our mutant products in transfected cells, we have tagged the 3' end of the K14 coding sequence with a sequence encoding an antigenic domain of the neuropeptide substance P. Using DNA transfection and immunohistochemistry (with an antibody against substance P), we have defined the limits of K14 sequence necessary to incorporate into a keratin filament network in vivo without disrupting its architecture. We have also uncovered major differences in the behavior of carboxy- and amino-terminal alpha-helical mutants which do perturb the cytoskeletal network of IFs: whereas carboxy terminal mutants give rise to aggregates of keratin in the cytoplasm, amino-terminal mutants tend to produce aggregates of keratins which seem to localize at the nuclear surface. An examination of the phenotypes generated by the carboxy and amino-terminal mutants and the behavior of cells at late times after transfection suggests a model whereby initiation of filament assembly occurs at discrete sites on the nuclear envelope and filaments grow from the nucleus toward the cytoplasm.