A major component of barrier function in stratified squamous epithelia is the cornified cell envelope (CE). This is a multi-component 15 nm thick layer of highly insoluble protein deposited on the inner surface of the plasma membrane of the cells during terminal differentiation. In the case of the epidermis, a 5 nm thick layer of ceramide lipids (lipid envelope) is attached to the exterior surface. The insolubility of the protein envelope is due in large part to the crosslinking of several structural proteins by transglutaminases. Studies on the biology and assembly of the protein and lipid components are a major effort of this laboratory. Specifically, we are studying: (i) the crosslinking of proteins in CEs isolated from a variety of sources to explore which proteins are crosslinked together through which glutamines and lysines, and to provide information on structure and function; (ii) two key structural proteins and their genes, loricrin and the small proline rich protein (SPR) families; (iii) the ceramide lipids which become covalently attached to the CE; (iv) the earliest stages of CE assembly by use of immunogold electron microscopy on CE fragments produced in cultured keratinocytes; and (v) an attempt to recreate a CE-like structure using an in vitro synthetic lipid vesicle (slv) model system.CE protein envelope structure and assemblyWe are studying the features of CEs isolated from human foreskin epidermal stratum corneum, immature terminally differentiating foreskin epidermis, cultured human epidermal keratinocytes induced to terminally differentiate, mouse forestomach epithelium, and human gingiva epithelium. We have used controlled proteolysis to dissect apart the CEs, separate crosslinked peptides, and perform protein sequencing. In all cases except the gingiva, the bulk of the protein envelope consists of loricrin (70-80%) admixed with smaller amounts (2-20%) of SPRs. The gingiva CEs contain >50% SPRs. In all cases, the SPRs appear to function as promiscuous crossbridging proteins, by linking together various proteins, most often loricrin or themselves, through multiple adjacent glutamine and lysine residues on their head and tail domains only. In addition, we have found that there is a direct correlation between the amount of SPRs present in CEs and the presumed physical characteristics and exposure to physical trauma of the epithelium: human or mouse trunk CEs contain little or no SPRs; human foreskin epidermal CEs 5% SPRs; mouse foot pad and lip epidermal CEs 10% SPRs; mouse forestomach CEs 20% SPRs; and gingiva CEs >50% SPRs. These data suggest that crossbridging SPRs serve to modulate the biomechanical properties of the CEs in which they are expressed. Moreover, we have shown that the CE is crosslinked to the keratin intermediate filament cytoskeleton, which further suggests that the SPRs may contribute in important ways to the biomechanical properties and requirements of an entire epithelium. In addition, we have recovered many crosslinks involving the keratins, involucrin, and other cell junctional proteins including desmoplakin and envoplakin. These findings prompted a more detailed study on the earliest stages of CE assembly. We recovered CEs from 2, 3, 5 and 7 day cultured normal human epidermal keratinocytes and used them for immunogold electron microscopy, as well as sequencing of the 3 and 7 day CEs. Our data are consistent with the possibility that CE assembly is initiated along the plasma membrane at interdesmosomal sites by head-to-head and head-to-tail crosslinking of involucrin to itself, and perhaps to envoplakin and periplakin. Shortly later, involucrin deposition spreads to desmosomal sites so that a continuous layer of involucrin, envoplakin and perhaps periplakin is formed along the cell periphery: this layer perhaps forms a scaffold for the later stages of CE assembly involving substantial deposition of other proteins such as loricrin and SPRs. Loricrin We have expressed human loricrin in bacteria and used it to characterize its structure, biochemical properties, and crosslinking by epidermal transglutaminases (TGase) in vitro. By biophysical measurements it has some structure in solution associated with its multiple tyrosines. It is a complete TGase substrate because it is oligomerized by all three epidermal TGases in in vitro reactions, although with different kinetic efficiencies, and utilization of different glutamines and lysines. From comparisons of the residues used in vitro with those used in vivo from sequencing of CEs, we can conclude that both TGase 1 and TGase 3 are required for the correct crosslinking of loricrin in vivo. Future studies will be aimed at determining the structure of loricrin by use of solution nmr methods on either full length expressed protein, or synthetic peptides from selected portions of it. The proximal promoter of the human loricrin gene resides within the first 160 bp above the transcription initiation start site. Interactions of c-fos/c-jun proteins at an AP1 site are essential for epithelial expression in keratinocytes. Unlike the mouse loricrin gene, a calcium responsive element lies within this region. We are actively characterizing a series of negative elements which lie just upstream of the 160 bp region. Small proline rich proteins SPRs consist of three distinct families consisting of from one to 11 members. We have expressed one member of each of the human SPR1, SPR2 and SPR3 proteins for in vitro studies. By circular dichroism, they have little organized structure in solution. What structure is present can be attributed to the central proline-rich peptide repeats, and the signal strength is proportional to the number of repeats. The SPR proteins are also complete substrates in in vitro crosslinking reactions for the three TGases commonly expressed in the epidermis. In all cases of SPR proteins studied, the glutamines and lysines used for crosslinking are located only on the end domains, suggesting they may function as crossbridging proteins. However, the details are different. In the case of SPR2 proteins, the TGase 1 enzyme uses only one glutamine residue on the head domain, and only one lysine on the tail domain for interchain crosslinking, whereas the TGase 3 enzyme uses multiple head and tail residues for interchain crosslinking. In the SPR1 proteins, we found that there are two head domain regions termed head A and head B. The former are used only by the TGase 3 enzyme for interchain crosslinking, whereas the latter are used only by the TGase 1 enzyme primarily for intrachain crosslinking. A similar situation was found for SPR3 proteins. Moreover, we correlated these crosslinking data with those for loricrin, and we found that the TGase 1 and 3 enzymes crosslink at common sites on the respective proteins. These data suggest that TGase 3 initiates crosslinking of loricrin and SPRs into small interchain oligomers which are later crosslinked to the CE by the TGase 1 enzyme. Solution nmr structural studies on the SPR2 and 3 proteins have been performed. Unfortunately, these proteins have little organized structure in solution and only short range (2-4 ) interactions were obtained. Nevertheless, the data suggest the central peptide repeat domains adopt novel omega-loop-like protein folds. We have explored the expression of the SPR1 and SPR2 families in mouse epithelia by use of immunocytochemistry, in situ hybridization and RT-PCR. Both families are differentially expressed in different epithelia. In the case of SPR1, the amount expressed in the epidermis varies widely with the site, from none in interfollicular epidermis, to very abundant in the thickened epidermis of the foot pad and lip, for example. In the case of the SPR2 proteins, the entire family is upregulated in the epidermis after injury or chemical assault. The SPR2 family in the mouse genome consists of 11 genes linked in a cluster of about 150 kbp, and includes one presumptive pseudogene and one prematurely terminated very short protein possessing only a head domain. The nine SPR2 proteins differ from one another primarily in the numbers of central domain peptide repeats which vary from 2.3 to 11, that is, of different span lengths. Further work will be directed to explore the likely role of the SPRs as determinants of the biomechanical properties of CEs.Ceramide lipids By alkaline hydrolysis in methanol it has been possible to remove covalently bound ceramide lipids from foreskin CEs. By mass spectrometry, these consist of a heterogeneous population of molecules of varying size, containing fatty alcohol chains of C28-C34 long, and sphingosine chains of C18 to C22. The amount of ceramides attached corresponds to a monomolecular layer on the protein portion of the CE. It is suggested that this layer of ceramides is essential for the organization of other intercellular lipids to effect proper barrier function. Thus anything that disrupts these ceramides or their orderly attachment to the protein CE could be predicted to result in an ichthyosis-like phenotype. To better understand this structure, we used limited alkaline hydrolysis, and have isolated several peptides with attached ceramides. By sequencing, most of these peptides derived from the ancestral head domain sequences of involucrin, as well as lesser amounts from other junctional proteins including periplakin, envoplakin and desmoplakin.An in vitro synthetic lipid vesicle system to explore CE assemblyWe have constructed synthetic lipid vesicles (slv) of composition similar to eukaryote plasma membranes. We have found that the TGase 1 enzyme binds spontaneously to them. Interestingly, involucrin also binds in a Ca-dependent manner and at <1 microM, suggesting that involucrin can attach to plasma membranes in vivo as soon as it is expressed during initiation of terminal differentiation. However, the TGase 1 enzyme does not begin to crosslink involucrin until the Ca concentration is raised to >100 microM. Most crosslinking occurs through glutamine 496, with minor amounts at four other head domain residues, and only one lysine residue located on the head domain: thus involucrin forms head-to-head and head-to-tail oligomers. These data contrast the utilization of >50 glutamine residues in solution TGase 1 assays. In addition, we have performed experiments incorporating a synthetic ceramide, and found that the TGase 1 enzyme is capable of attaching the ceramide to the same four head domain glutamines identified in crosslinking experiments, three of which are the same as found in in vivo experiments on foreskin CEs. The ceramide attachment occurs by ester formation. Thus the TGase 1 enzyme can perform two essential aspects of CE formation in vivo: (a) the crosslinking of CE structural proteins; and (b) the covalent attachment of the ceramide layer. Indeed, we speculate that the severe phenotype of the disease lamellar ichthyosis caused by inactivated TGase 1 enzyme may be due more to the inability to attach the ceramide than protein crosslinking. Further work is now needed to explore the earliest stages of CE assembly by use of this slv system. Such studies may provide valuable new insights into disease etiology as well as provide clues as to how to effectively treat ichthyosis disorders.
