Advertisement

The Proteins Elafin, Filaggrin, Keratin Intermediate Filaments, Loricrin, and Small Proline-rich Proteins 1 and 2 Are Isodipeptide Cross-linked Components of the Human Epidermal Cornified Cell Envelope (∗)

Open AccessPublished:July 28, 1995DOI:https://doi.org/10.1074/jbc.270.30.17702
      The cornified cell envelope (CE) is a 15-nm thick layer of insoluble protein deposited on the intracellular side of the cell membrane of terminally differentiated stratified squamous epithelia. The CE is thought to consist of a complex amalgam of proteins cross-linked by isodipeptide bonds formed by the action of transglutaminases, but little is known about how or in which order the several putative proteins are cross-linked together. In this paper, CEs purified from human foreskin epidermis were digested in two steps by proteinase K, which released as soluble peptides about 30% and then another 35% of CE protein mass, corresponding to approximately the outer third (cytoplasmic surface) and middle third, respectively. Following fractionation, 145 unique peptides containing two or more sequences cross-linked by isodipeptide bond(s) were sequenced. Based on these data, most (94% molar mass) of the outer third of CE structure consists of intra- and interchain cross-linked loricrin, admixed with SPR1 and SPR2 proteins as bridging cross-links between loricrin. Likewise, the middle third of CE structure consists largely of cross-linked loricrin and SPR proteins, but is mixed with the novel protein elafin which also forms cross-bridges between loricrin. In addition, cross-links involving loricrin and keratins 1, 2e, and 10 or filaggrin were recovered in both levels. The data establish for the first time that these several proteins are indeed cross-linked protein components of the CE structure. In addition, the data support a model for the intermediate to final stages of CE assembly: the proteins elafin, SPR1 and SPR2, and loricrin begin to be deposited on a preformed scaffold; later, elafin deposition decreases as loricrin and SPR accumulation continues to effect final assembly. The recovery of cross-links involving keratins further suggests that the subjacent cytoplasmic keratin intermediate filament-filaggrin network is anchored to the developing CE during these events.

      INTRODUCTION

      During the process of terminal differentiation in stratified squamous epithelia such as the epidermis, a 15-nm thick layer of protein is deposited on the intra-cellular surface of the cell periphery. This cornified cell envelope (CE)1
      The abbreviations used are: CE
      cornified cell envelope
      KIF
      keratin intermediate filaments
      PTH
      phenylthiohydantoin
      SPR
      small proline-rich (class of proteins)
      TGase
      transglutaminase
      TGase 1 (etc.)
      suggested new nomenclature for transglutaminases
      HPLC
      high performance liquid chromatography.
      plays a critical role in barrier function of the tissue and for the organism(
      • Hohl D.
      ,
      • Greenberg C.S.
      • Birckbichler P.J.
      • Rice R.H.
      ,
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ). In the case of the epidermal CE, several distinct proteins have now been identified and characterized as potential CE components, including: involucrin(
      • Rice R.H.
      • Green H.
      ,
      • Eckert R.L.
      • Yaffe M.B.
      • Crish J.F.
      • Murthy S.
      • Rorke E.A.
      • Welter J.F.
      ), cystatin α(
      • Takahashi M.
      • Tezuka T.
      • Katunuma N.
      ,
      • Kartasova T.
      • Cornelissen B.J.C.
      • van der Putte P.
      ), several small proline-rich proteins (SPR1, SPR2, in epidermis as well as SPR3 in cultured keratinocytes)(
      • Kartasova T.
      • Cornelissen B.J.C.
      • van der Putte P.
      ,
      • Kartasova T.
      • van de Putte P.
      ,
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Gibbs S.
      • Fijneman R.
      • Wiegart J.
      • Geurts van Kessel A.
      • van de Putte P.
      • Backendorf C.
      ), loricrin(
      • Mehrel T.
      • Hohl D.
      • Rothnagel J.A.
      • Longley M.A.
      • Bundman D.
      • Cheng C.
      • Lichti U.
      • Bisher M.E.
      • Steven A.C.
      • Steinert P.M.
      • Yuspa S.H.
      • Roop D.R.
      ,
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Yoneda K.
      • Hohl H.
      • McBride O.W.
      • Wang M.
      • Cehrs K.U.
      • Idler W.W.
      • Steinert P.M.
      ,
      • Yoneda K.
      • Steinert P.M.
      ), and possibly trichohyalin(
      • Lee S.-C.
      • Kim I.-G.
      • Marekov L.N.
      • O'Keefe E.J.
      • Parry D.A.D.
      • Steinert P.M.
      ), filaggrin(
      • Richards S.
      • Scott I.R.
      • Harding C.R.
      • Liddell J.E.
      • Powell G.M.
      • Curtis C.G.
      ,
      • Steven A.C.
      • Steinert P.M.
      ), keratin intermediate filaments (KIF)(
      • Steven A.C.
      • Steinert P.M.
      ,
      • Abernethy J.L.
      • Hill R.L.
      • Goldsmith L.A.
      ,
      • Ming M.E.
      • Daryanami H.A.
      • Roberts J.E.
      • Haimowitz J.E.
      • Kvedar J.C.
      ), and a putative cysteine-rich protein (
      • Tezuka T.
      • Takahashi M.
      ) that may be elafin(
      • Tezuka T.
      • Takahashi M.
      ,
      • Wiedow O.
      • Schroder J.-M.
      • Gregory H.
      • Young J.A.
      • Christophers E.
      ,
      • Schalkwijk J.
      • de Roo C.
      • de Jongh G.J.
      ,
      • Nonomura K.
      • Yaminishi K.
      • Yasuno H.
      • Nara K.
      • Hirose S.
      ,
      • Saheki T.
      • Ito F.
      • Hagiwara H.
      • Saito Y.
      • Kuroki J.
      • Tachibana S.
      • Hirose S.
      ,
      • Molhuizen H.O.F.
      • Alkemade H.A.C.
      • Zeeuwen P.L.J.M.
      • de Jongh G.J.
      • Wieringa B.
      • Schalkwijk J.
      ). These proteins are thought to be cross-linked together to assemble the CE by way of disulfide bonds(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ), as well as the Nϵ-(γ-glutamy)lysine isodipeptide cross-link, formed by the action of one or more of the three known epidermal transglutaminases(
      • Hohl D.
      ,
      • Greenberg C.S.
      • Birckbichler P.J.
      • Rice R.H.
      ,
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ,
      • Rice R.H.
      • Green H.
      ,
      • Eckert R.L.
      • Yaffe M.B.
      • Crish J.F.
      • Murthy S.
      • Rorke E.A.
      • Welter J.F.
      ,
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Aeschilimann D.
      • Paulsson M.
      ).
      However, the design of experiments to prove that these are indeed CE structural proteins has proven difficult, because the isodipeptide cross-link itself cannot be hydrolyzed to release the intact proteins without use of reagents that also cleave peptide bonds. Likewise, few data are available on the order of assembly of these proteins into the CE, how they are cross-linked together, and which residues in the proteins are utilized in cross-linking,
      Nevertheless, several approaches have been explored. Indirect mathematical modelling has provided important clues on the abundance of the proteins assembled in epidermal CEs(
      • Steven A.C.
      • Steinert P.M.
      ). This was based on least-squares fitting methods using the known amino acid compositions (from deduced cDNA sequences) of most of the identified proteins listed above, and the amino acid compositions of purified CEs. In this way, it was estimated that loricrin is the major component (66%), together with smaller amounts of cysteine-rich protein (14%), filaggrin (10%), SPRs (5%), involucrin, and cystatin α (2-5% each), but no detectable KIF. At the time of these calculations, the exact amino acid composition of cysteine-rich protein was not known. However, assuming that the well-characterized protein elafin of known sequence(
      • Saheki T.
      • Ito F.
      • Hagiwara H.
      • Saito Y.
      • Kuroki J.
      • Tachibana S.
      • Hirose S.
      ,
      • Molhuizen H.O.F.
      • Alkemade H.A.C.
      • Zeeuwen P.L.J.M.
      • de Jongh G.J.
      • Wieringa B.
      • Schalkwijk J.
      ) is in fact cysteine-rich protein, then the data (of (
      • Steven A.C.
      • Steinert P.M.
      )) can be recalculated to: loricrin 70%, filaggrin 8%, elafin 6%, SPRs and cystatin α 5% each, and involucrin and KIF about 2% each. This large amount of loricrin is consistent with the abundance of its mRNA in the epidermis(
      • Mehrel T.
      • Hohl D.
      • Rothnagel J.A.
      • Longley M.A.
      • Bundman D.
      • Cheng C.
      • Lichti U.
      • Bisher M.E.
      • Steven A.C.
      • Steinert P.M.
      • Yuspa S.H.
      • Roop D.R.
      ,
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ). Furthermore, transgenic experiments (
      • Yoneda K.
      • Steinert P.M.
      ) have suggested that loricrin is the last or one of the last components added during CE assembly by addition to a pre-existing scaffold of such proteins as involucrin and cystatin α. Another approach to address these questions has been to attack isolated CEs with chemicals such as CNBr or solvents to release protein species for characterization by Western blotting or other biochemical techniques. This has been used to show that involucrin is a component of the CE of cultured epidermal keratinocytes(
      • Yaffe M.B.
      • Murthy S.
      • Eckert R.L.
      ). This method also released proteins termed pancornulins that are probably the same as the SPRs(
      • Baden H.P.
      • Kubilus J.
      • Phillips S.B.
      • Kvedar J.C.
      • Tahan S.R.
      ,
      • Phillips S.B.
      • Kubilus J.
      • Grassi A.M.
      • Goldaber M.
      • Baden H.P.
      ,
      • Greco M.A.
      • Lorand L.
      • Lane W.S.
      • Baden H.P.
      • Parameswaran K.N.P.
      • Kvedar J.V.
      ). Likewise, immunological data with existing protein-specific antibodies have identified fragments of keratins and filaggrin in isolated CE preparations(
      • Baden H.P.
      • Kubilus J.
      • Phillips S.B.
      • Kvedar J.C.
      • Tahan S.R.
      ,
      • Haftek M.
      • Serre G.
      • Mils V.
      • Thivolet J.
      ). Clearly, each of these studies suggest association with but do not necessarily prove attachment of these proteins to purified CEs. This is an important problem because of the difficulty in preparing purified CEs from either cultured keratinocytes or intact epidermis free of soluble cytoplasmic proteins(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Yaffe M.B.
      • Murthy S.
      • Eckert R.L.
      ), or proteins solubilized during the isolation procedures.
      A third approach was predicated on data which showed that low specificity proteases can digest isolated CEs to their constituent amino acids and the isodipeptide cross-link within 2-4 days(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steinert P.M.
      • Idler W.W.
      ,
      • Yuspa S.H.
      • Ben T.
      • Steinert P.
      ,
      • Tarcsa E.
      • Kedei N.
      • Thomazy V.
      • Fesus L.
      ). In a recent study(
      • Steinert P.M.
      ), the time course of release of protein from epidermal CEs during the first 36 h of trypsin and proteinase K digestion was followed by amino acid analyses, and these data were then subjected to mathematical modelling to estimate the protein contents remaining in the insoluble CE remnants(
      • Steven A.C.
      • Steinert P.M.
      ). These experiments were also followed by immunogold electron microscopy of the remnants using a series of monospecific antibodies(
      • Steinert P.M.
      ). An initial digestion with trypsin removed primarily keratin and filaggrin epitopes from one side of the CEs. During 24 h of digestion with proteinase K, primarily only loricrin, SPR, and the novel protein elafin were removed, based on both amino acid composition and immunogold criteria. The remnant was enriched in elafin, cystatin α, and involucrin, of which involucrin epitopes were still abundantly available on one side(
      • Steinert P.M.
      ). The appearance and disappearance of gold particles on only one side of the CE remnants provides strong support for the idea that the proteases were removing CE structural proteins from the cytoplasmic surface(
      • Steven A.C.
      • Bisher M.E.
      • Roop D.R.
      • Steinert P.M.
      ). Finally, after 36 h of proteinase K digestion, cystatin α and involucrin comprised most of the remnant, but in this case, involucrin epitopes were visible on both sides(
      • Steinert P.M.
      ), consistent with the idea that it may be associated directly to the lipid envelope(
      • Downing D.T.
      ).
      In this way, a three-stage model for CE structure was suggested which is an elaboration of the existing multi-stage hypotheses(
      • Hohl D.
      ,
      • Greenberg C.S.
      • Birckbichler P.J.
      • Rice R.H.
      ,
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ,
      • Rice R.H.
      • Green H.
      ,
      • Eckert R.L.
      • Yaffe M.B.
      • Crish J.F.
      • Murthy S.
      • Rorke E.A.
      • Welter J.F.
      ,
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ), in which: the outer (cytoplasmic surface) third of the CE consists mostly of loricrin/SPRs/filaggrin; the middle third consists of elafin/loricrin/SPRs; and the innermost third adjacent or attached to the lipid envelope consists of involucrin and cystatin α and perhaps other as yet unknown proteins. Also KIF may be buried throughout the CE (
      • Steinert P.M.
      ).
      However, the most rigorous evidence for the involvement of a protein in CE structure, that circumvents the concerns of degraded, loosely-associated, or contaminating solubilized proteins, would be to demonstrate identifiable protein sequences directly adjoined by isodipeptide cross-links in isolated CEs, as has been done in preliminary experiments for loricrin(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ). In this study, the limited proteolytic digestion paradigm has been explored further. A large number of peptides was harvested from CEs during the first 9 h of digestion with proteinase K, roughly corresponding to the outer two-thirds of CE structure, which were then subjected to amino acid microsequencing. In this way, we show for the first time that several CE proteins are indeed cross-linked components of the human epidermal CE. The data also provide robust support for the proposed modified model for CE structure.

      MATERIALS AND METHODS

      Preparation and Purification of Human Epidermal CEs

      CEs were prepared from the stratum corneum of human foreskins. The epidermis was separated by heat treatment by standard procedures and then extracted in a buffer of 8 M urea containing 50 mM Tris-HCl (pH 7.6) and 1 mM EDTA. In the absence of reducing agent, this buffer dissolves only the inner living layers and perhaps the transition layer of the epidermis, leaving the stratum corneum intact (
      • Steinert P.M.
      • Idler W.W.
      ,
      • Dale B.A.
      • Holbrook K.A.
      • Kimball J.R.
      • Hoff M.
      • Sun T.-T.
      ). The extract was then filtered through nylon gauze (
      • Yuspa S.H.
      • Harris C.C.
      ) (mesh size 0.1 mm) to collect the stratum corneum sheets. CEs were then prepared from this material by exhaustive boiling and sonication in 2% SDS, 20 mM dithiothreitol, 0.1 M Tris-HCl (pH 8.0), and 5 mM EDTA as described previously(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ). Following purification by pelleting at 5,000 × g through a 20% Ficoll solution in the same buffer, the CE fragments were washed three times in phosphate-buffered saline to remove the bulk of the SDS and reducing reagent. Amino acid analysis of acid-hydrolyzed samples was used to confirm the purity of the CEs(
      • Richards S.
      • Scott I.R.
      • Harding C.R.
      • Liddell J.E.
      • Powell G.M.
      • Curtis C.G.
      ).

      Isolation and Quantitation of the Isodipeptide Cross-link

      Aliquots of CE preparations were subjected to total enzymic digestion over a period of 4 days exactly as described previously(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steinert P.M.
      • Idler W.W.
      ,
      • Yuspa S.H.
      • Ben T.
      • Steinert P.
      ,
      • Tarcsa E.
      • Kedei N.
      • Thomazy V.
      • Fesus L.
      ). Control reactions contained only enzymes. The products were resolved by amino acid analysis to measure the amount of the Nϵ-(γ-glutamyl)lysine cross-link (Accurate Biochemical Corp.), which elutes at 29.0 min on a Beckman 6300 analyzer, immediately after methionine. The molar content of the cross-link was then calculated from the total amount of CE protein as determined by acid hydrolysis (in vacuo at 106°C for 22-48 h). In the combined batches of foreskin epidermal CEs used in this work this was 89 nmol/mg CE protein.

      Proteolytic Digestion of CEs and Fractionation of Peptides

      CEs were resuspended (1 mg/ml) in a buffer of 50 mM Tris-HCl (pH 8.0) and 5 mM CaCl2 and first digested with stirring with trypsin (Sigma, sequencing grade, 1% by weight, at 37°C) for 6 h(
      • Steinert P.M.
      ). Following removal of the solubilized material by centrifugation at 14,000 × g for 10 min, and washing in buffer, the CE remnants were resuspended in the same buffer and digested with proteinase K (Promega, 3% by weight, at 37°C) for 3 h. The CE remnant was pelleted, washed, and redigested for a 6-h time interval. This stepwise digestion procedure was repeated for a total of five 6-h intervals (33 h digestion in total). The amount of soluble peptide material released into the supernatant of each time interval was quantitated by amino acid analysis. Aliquots were resolved by HPLC using a reverse-phase ultrapshere ODS column (4.5 × 250 mm) with a gradient of 0-100% acetonitrile containing 0.08% trifluoroacetic acid. Conditions of digestion with proteinase K were optimized so as to release the maximal amount of protein from the CEs, in order to yield cross-linked peptides that were still sufficiently long to obtain unambiguous sequence information about the cross-linked protein(s). The size of peptides was determined empirically from their times of elution from the HPLC column. In general, peptides which eluted with <25% acetonitrile (<50 min) were <8 residues long and few contained cross-links, while those which eluted >30% acetonitrile contained cross-links and were longer than 10 residues. In this way, it was found that digestion with 3% proteinase K during the first two time intervals (3 h followed by 6 h) generated peptides that provided useful sequence information. Each collected peptide was neutralized with trimethylamine and concentrated to about 10 pmol/μl.
      During the course of the work, many minor peptide peaks were identified which potentially contained useful sequence information. To recover these in yields sufficient for sequencing (but not amino acid analysis), the fractions between major collected peaks from several HPLC runs were pooled and rerun.

      Amino Acid Sequencing

      Amino acid compositions of aliquots of resolved peaks were measured when quantities permitted to identify those peptide species that contained at least 1 Glx and 1 Lys residue, which were thus candidates for containing an isodipeptide cross-link. Candidate peptides (0.5-200 pmol) were then sequenced for 5-15 Edman degradation cycles in a LF3000 gas-phase sequencer and following the manufacturer's specifications (Porton). In most cases, the peptides were first covalently attached to a polyvinylidine difluoride solid support (Sequelon-AA, Millipore). The released PTH residues of each cycle were resolved and quantitated by on-line HPLC to elucidate the sequences (Beckman Instruments, using System Gold software). The diPTH-isodipeptide species eluted near PTH-Leu. The PTH-derivative of cysteine is unstable and could not be ascertained directly, but was inferred from the appearance of PTH-dehydroalanine.
      Assignment of protein sequences to the released PTH-derivatives from each cycle were based on the known protein sequences of several human CE proteins, and are: elafin(
      • Saheki T.
      • Ito F.
      • Hagiwara H.
      • Saito Y.
      • Kuroki J.
      • Tachibana S.
      • Hirose S.
      ,
      • Molhuizen H.O.F.
      • Alkemade H.A.C.
      • Zeeuwen P.L.J.M.
      • de Jongh G.J.
      • Wieringa B.
      • Schalkwijk J.
      ), filaggrin(
      • McKinley-Grant L.J.
      • Idler W.W.
      • Bernstein I.A.
      • Parry D.A.D.
      • Cannizzaro L.
      • Croce C.M.
      • Huebner K.
      • Lessin S.R.
      • Steinert P.M.
      ), keratins 1(
      • Steinert P.M.
      • Parry D.A.D.
      • Idler W.W.
      • Johnson L.D.
      • Steven A.C.
      • Roop D.R.
      ), 2e(
      • Collin C.
      • Moll R.
      • Kubicka S.
      • Ouhayoun J.-P.
      • Franke W.W.
      ), and 10(
      • Zhou X.-M.
      • Idler W.W.
      • Steven A.C.
      • Roop D.R.
      • Steinert P.M.
      ), loricrin(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ), and SPRs(
      • Kartasova T.
      • van de Putte P.
      ,
      • Gibbs S.
      • Fijneman R.
      • Wiegart J.
      • Geurts van Kessel A.
      • van de Putte P.
      • Backendorf C.
      ). In other cases, sequences were identified by computer searches of the Swiss Protein data base.

      Computer Analyses

      Secondary protein structure predictions were done using a suite of algorithms compiled by the University of Wisconsin Genetics Computer Group (
      • Kyte J.
      • Doolittle R.F.
      ,
      • Devereux J.
      • Haeberli P.
      • Smithies O.
      ) and the IBI Pustell sequence software version 4.0 (International Biotechnologies Inc).

      RESULTS

      Release and Fractionation of Peptide Material from CEs

      The purpose of the present study was to more rigorously test a recent three-stage elaboration of existing models for the assembly and structure of the human epidermal CE(
      • Steinert P.M.
      ). Proteinase K digestion procedures were modified so as to recover peptides suitable for microsequencing, which thus could yield unambiguous information about cross-linked protein constituents and CE structure.
      CEs were initially digested with trypsin which released about 6% of CE protein mass. Such peptides were too short for useful sequencing (<8 residues), but based on amino acid analyses, most were likely to be filaggrin and perhaps some keratin. The virtual absence of Lys suggested that they contained little or no cross-linked protein material. Indeed, direct measurement of the total amount of isodipeptide cross-link in the starting CE preparation revealed that this fraction contained 0.5 nmol of cross-link/mg of protein, corresponding to ~0.6% of the total. Because keratin and filaggrin are the major proteins of the epidermal tissue, it seems possible that this trypsin digestion is simply removing contaminating solubilized epidermal proteins(
      • Steinert P.M.
      ). Subsequent release of useful cross-linked peptide material was done by titration of the time of proteinase K digestion, followed by generation of an HPLC profile, and measurement of the amino acid compositions of the peptide products recovered. In this way, it was found that digestion for a 3-h time interval released 30% of CE protein mass as peptides that were up to 50 residues long (Fig. 1A). A subsequent 6-h digestion released another 35% of CE protein mass and these peptides were up to 40 residues long (Fig. 1B). Likewise, measurement of the total amounts of isodipeptide cross-link revealed that the 3- and then 6-h digests released 35 nmol/mg total CE protein (39%) and 36.4 nmol/mg (41%), respectively. Further 6-h digestion intervals for a total of 33 h released an additional 24% of CE protein mass, and 14 nmol of cross-links/mg of CEs, but these peptides were <10 residues long and were too short for unambiguous sequencing (data not shown). Together with previous immunogold analyses of the fragments recovered in these times(
      • Steinert P.M.
      ), these data mean that the peptides released in the first 3 h arose from roughly the outer (cytoplasmic) third of the CE, and the second 6-h digestion released material from the middle third.
      Figure thumbnail gr1
      Figure 1:Fractionation of proteinase K peptides by HPLC. A, 3 h; B, next 6-h digestion time. All major and minor peaks were collected for analysis and microsequencing. The acetonitrile gradient is shown.
      Of the 197 peptides containing cross-links, the positions of the Gln and Lys residues involved in the one or more isodipeptide cross-links could be solved in 187, giving a total of 145 unique peptides. In 10 other multi-branched peptides involving only loricrin sequences, there was no unique solution to the assignment of which Gln and Lys residues were cross-linked to each other in the several likely cross-links. Of the 145 unique peptides containing one or more cross-links, 144 contained at least one recognizable loricrin sequence, of which 98 involved loricrin-loricrin cross-links only, and 46 involved loricrin cross-linked with another protein(s). One peptide contained cross-links adjoining three non-loricrin sequences (see below). The sequences of many peptides are presented in Tables I-IV.

      Peptides from the Outer (Cytoplasmic) Third of CE Structure Are Mostly Loricrin and SPRs

      3-h digests, 133 peptides were identified, recovered, and sequenced from two similar proteinase K digestion experiments, of which 91 contained one or more cross-links. These cross-linked peptides varied in amounts from ~0.5 to 850 pmol, for a total of 27.5 nmol, and contained 31.5 nmol of cross-links or an average of 1.15 cross-links/mol. They accounted for a total of 90% of isodipeptide cross-links in this fraction (31.5 of 35 nmol cross-link/mg of CEs). This means that ~10% of the cross-links in the 3-h fraction were not found in the harvested and sequenced peptides. Presumably this small amount was lost as short peptides that eluted before 50 min on the HPLC column. However, analyses of the 90% recovered are likely to provide relevant and useful data on the composition and structure of this level of the CE. When expressed as a percentage of the 27.5 nmol of total peptides, it was found that 94% of the molar mass involved loricrin sequences (all 91 peptides), 3% were SPR1 (7 peptides), 2% were SPR2 (4 peptides), 1% was elafin (1 peptide), and trace amounts were due to keratin 1 (0.05%) (4 peptides) and filaggrin (0.08%) (1 peptide) sequences.

      Peptides from the Central Third of CE nvolve Elafin as Well as Loricrin and SPRs

      An additional 143 peptides (118 with cross-links) were characterized from the subsequent 6-h digestion interval. Of these 118, 64 were identical to those obtained in the 3-h digestion and 54 were unique. Most of the 64 common peptides involved loricrin-loricrin cross-linked species. In this case, 33.2 nmol of isodipeptide cross-links (91% of the total amount) could be accounted for in the harvested and sequenced peptides, and the total molar mass of characterized peptides was 34.8 nmol (average of 1.05 mol of cross-links/mol). Of the molar total of peptides, 81% involved loricrin sequences (117 of 118, 24 new peptides), 11% were elafin or preproelafin (17 new peptides), 3% were SPR1 (4 new peptides), 2% were SPR2 (3 new peptides), 0.1% were total keratins (4 new peptides), 0.1% (2 new peptides) were filaggrin, and a single peptide involved desmoplakin I or II (0.1%). Thus, while most of the peptides were still composed of loricrin at this digestion time interval, a notable difference was the appearance of numerous cross-linked peptides involving elafin.

      Intra- and/or Intermolecular Loricrin Cross-links

      The molar amounts of each of the 14 Gln and 7 Lys residues of loricrin that were used in cross-links was calculated from the yields of the 286 separate loricrin sequences identified in all of the peptides solved here (Fig. 2). Of seven possible Lys residues, the terminal Lys315 accounted for 57% of the molar total, while two positions (residues 17 and 296) were not used at all. In the case of Gln residues, 70% of the molar total was attributable to residues 215 and 216, and residue position 158 was not used. These data are consistent with the predicted secondary structure of loricrin (Fig. 2): the most used residues were located in sequences likely to be exposed on the surface, while unused residues (residues 17, 158, and 296), and infrequently used residues are predicted to be buried in or near the glycine loop motifs.
      Figure thumbnail gr2
      Figure 2:Molar utilization of Lys (upper panel) and Gln (middle panel) residues in in vivo cross-links involving loricrin. The data were compiled from all 286 separate loricrin sequences in the peptides sequenced in this study. The arrows designate those Lys and Gln residues that were not utilized in the present body of data. The lower panel displays a predicted surface probability structure of loricrin.
      The cumulative data of Fig. 2 show that a total of 75 nmol of lysines and 78 nmol of glutamines in loricrin are used per mg of the outer (cytoplasmic) two-thirds of CE protein, of which 81-94% (weighted mass average 86%) or 33 nmol is loricrin (molecular mass 26 kDa(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      )). This means that from all of the loricrin-containing peptides sequenced here, there were an average of ~2.3 mol of isodipeptide cross-links/mol of loricrin in the outer portions of the CE. However, the bulk of the sequence data cannot specify whether the cross-links are intrachain, interchain, or both, although 5 solved peptides (Table 1) and 10 unsolved peptides clearly involved interchain cross-links since the same terminal residues were found multiple times.

      Cross-links Involving SPR1 and SPR2

      A total of 19 peptides cross-linked with loricrin involved interchain cross-links with the SPR1 and SPR2 proteins (molar amount about 5%) (Table 2), and were approximately equally distributed between the two digestion time intervals. Since their initial discovery, these proteins have been predicted to be CE precursors, on the basis of sequence and immunological data(
      • Kartasova T.
      • Cornelissen B.J.C.
      • van der Putte P.
      ,
      • Kartasova T.
      • van de Putte P.
      ,
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Gibbs S.
      • Fijneman R.
      • Wiegart J.
      • Geurts van Kessel A.
      • van de Putte P.
      • Backendorf C.
      ). This idea is also supported by a recent immunogold decoration study(
      • Steinert P.M.
      ). The present data confirm this hypothesis for the first time. In addition to Gln- and Lys-rich terminal sequences that are homologous to loricrin(
      • Gibbs S.
      • Fijneman R.
      • Wiegart J.
      • Geurts van Kessel A.
      • van de Putte P.
      • Backendorf C.
      ,
      • Backendorf C.
      • Hohl D.
      ), the notable feature of their sequences is the presence of a conserved octapeptide motif repeated 6 times in SPR1 and a nonapeptide motif repeated 3 times in SPR2. These repeats also contain Gln and Lys residues (
      • Kartasova T.
      • van de Putte P.
      ,
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Gibbs S.
      • Fijneman R.
      • Wiegart J.
      • Geurts van Kessel A.
      • van de Putte P.
      • Backendorf C.
      ) which have been postulated to provide additional cross-linking sites(
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Gibbs S.
      • Fijneman R.
      • Wiegart J.
      • Geurts van Kessel A.
      • van de Putte P.
      • Backendorf C.
      ,
      • Backendorf C.
      • Hohl D.
      ). In the present data, however, no peptides involved cross-links with the peptide repeat sequences: 18 of 19 peptides used either the carboxyl-terminal or penultimate carboxyl-terminal residue only; in the other peptide, SPR1 Lys6 was used. As for loricrin, both these sequences are predicted to be highly exposed for reaction (data not shown). These data strongly suggest that in fact the SPRs are serving as cross-bridging proteins among and between the more numerous loricrin molecules, in confirmation of an earlier prediction(
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ).

      Cross-links Involving Keratin and Filaggrin

      A quantitatively very minor portion of the total peptides recovered here involved eight cross-links with keratin 1, 2e, or 10 chains (totalling about 60 pmol or about 0.15% on a molar basis) (Table 3). The lysine residue used in the keratin 1 and 2e chains is situated in the V1 amino-terminal end domain sequences in a 20-residue window highly conserved among the Type II keratins expressed in stratified squamous epithelia(
      • Gan S.-Q.
      • Idler W.W.
      • McBride O.W.
      • Markova N.G.
      • Steinert P.M.
      ). The cross-link involving keratin 10 was located close to its amino terminus. Three cross-links between loricrin and filaggrin (0.1% molar yield) (Table 3) utilized a Gln-Gln dipeptide sequence in filaggrin in a narrow conserved sequence window in a protein of otherwise highly variable sequence(
      • McKinley-Grant L.J.
      • Idler W.W.
      • Bernstein I.A.
      • Parry D.A.D.
      • Cannizzaro L.
      • Croce C.M.
      • Huebner K.
      • Lessin S.R.
      • Steinert P.M.
      ,
      • Gan S.-Q.
      • Idler W.W.
      • McBride O.W.
      • Markova N.G.
      • Steinert P.M.
      ). In the case of both the keratin chains and filaggrin, these data confirm for the first time their direct involvement in CE structure.
      The total molar amount of keratins and filaggrin cross-linked to the isolated CEs corresponds to ~0.1% each, which means that on a weight basis, <0.5% of the total CE protein consists of these proteins directly cross-linked by isodipeptide bonds. Based on mathematical modelling of amino acid compositions, it was estimated (
      • Steven A.C.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ) that keratins and filaggrin constitute about 3 and 7%, respectively, of the total “purified” CE protein mass. These data mean that most of the keratin and filaggrin present in the initial CE preparations was not cross-linked, but presumably was retained as contaminating adherent protein.

      Cross-links Involving Elafin

      An additional 18 peptides contained cross-links involving the novel protein elafin (Table 4). They were almost entirely recovered from the second digestion time interval, corresponding roughly to the middle third of CE structure (Table 4). Elafin is an abundant differentiation protein product of the epidermis and skin(
      • Tezuka T.
      • Takahashi M.
      ,
      • Wiedow O.
      • Schroder J.-M.
      • Gregory H.
      • Young J.A.
      • Christophers E.
      ,
      • Schalkwijk J.
      • de Roo C.
      • de Jongh G.J.
      ,
      • Nonomura K.
      • Yaminishi K.
      • Yasuno H.
      • Nara K.
      • Hirose S.
      ,
      • Saheki T.
      • Ito F.
      • Hagiwara H.
      • Saito Y.
      • Kuroki J.
      • Tachibana S.
      • Hirose S.
      ,
      • Molhuizen H.O.F.
      • Alkemade H.A.C.
      • Zeeuwen P.L.J.M.
      • de Jongh G.J.
      • Wieringa B.
      • Schalkwijk J.
      ,
      • Nara K.
      • Ito S.
      • Suzuki Y.
      • Ghoneim M.A.
      • Tachibana S.
      • Hirose S.
      ), and the present data document for the first time its role in the epidermis as an important CE component. The precursor of elafin, preproelafin of 11 kDa, contains four conserved 12-residue peptide repeats, each of which contains Lys and Gln residues. This motif is similar to the cross-linking region of the seminal vesicle protein substrate of the prostate TGase 4 enzyme (
      • Moore J.T.
      • Hagstrom J.
      • McCormick D.J.
      • Harvey S.
      • Madden B.
      • Holicky E.
      • Stanford D.R.
      • Wieben E.D.
      ). Also, model synthetic peptides of this sequence motif can participate in transglutaminase cross-linking reactions in vitro(
      • Molhuizen H.O.F.
      • Alkemade H.A.C.
      • Zeeuwen P.L.J.M.
      • de Jongh G.J.
      • Wieringa B.
      • Schalkwijk J.
      ). Mature elafin (6 kDa) constitutes the Cys-rich carboxyl-terminal portion of the precursor, and is produced by cleavage in the fourth repeat. In most of the peptides discovered here, the elafin appears to serve as an intermediary or cross-bridging protein between two other protein sequences, which in fact, is consistent with its suggested role in cross-linking reactions(
      • Nara K.
      • Ito S.
      • Suzuki Y.
      • Ghoneim M.A.
      • Tachibana S.
      • Hirose S.
      ). Most of the peptides utilized Gln and Lys residue sites located near the predicted amino terminus of mature elafin, or at its carboxyl-terminal end. Interestingly, three peptides were found to involve the precursor form, using Lys or Gln residue sites in the last 12-residue repeat just prior to the point of cleavage to generate mature elafin.
      The idea that elafin functions as a cross-bridging protein is further supported by the recovery of two other peptides (Table 4) (yields 0.1 and 0.08%, respectively) in which elafin spans between keratin 1 (at Lys73) and loricrin molecules, and between keratin 1 and desmoplakin I/II (at Gln1646). In the case of desmoplakin, this is the last Gln residue, located near its carboxyl terminus(
      • Green K.J.
      • Parry D.A.D.
      • Steinert P.M.
      • Virata M.L.A.
      • Wagner R.M.
      • Angst B.D.
      • Nilles L.A.
      ). This is the first report of the linkage of desmoplakin to the CE. Desmoplakin is a major structural protein component of desmosomes of many types of cells, including epithelia. It is a flexible rod >100 nm long which projects into the cytoplasm such that its carboxyl-terminal domain may be available for direct (
      • Green K.J.
      • Parry D.A.D.
      • Steinert P.M.
      • Virata M.L.A.
      • Wagner R.M.
      • Angst B.D.
      • Nilles L.A.
      ,
      • Kouklis P.D.
      • Hutton E.
      • Fuchs E.
      ) or indirect (
      • Green K.J.
      • Parry D.A.D.
      • Steinert P.M.
      • Virata M.L.A.
      • Wagner R.M.
      • Angst B.D.
      • Nilles L.A.
      ) association with the KIF cytoskeleton. The present finding would seem to support the idea of an indirect association through elafin.

      DISCUSSION

      Definition of a CE Constituent Protein

      The functional definition we have used throughout this and earlier studies (
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ) for the involvement of proteins in the CE is the documentation of an identifiable protein sequence cross-linked by the transglutaminase-catalyzed isodipeptide bond. The use of this rigorous definition is important for two reasons. First, it has proven very difficult to recover CEs free from soluble cytoplasmic proteins, or abundant epidermal proteins such as keratins and filaggrin that are solubilized by the obligatory exhaustive extraction procedures(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Yaffe M.B.
      • Murthy S.
      • Eckert R.L.
      ). Indeed, we have found that a simple trypsin digestion can remove 6-8% of purified CE protein mass as short peptides. Based on direct cross-link data adduced here (Table 3), most of this probably originates from contaminating filaggrin and keratin. The second reason is an attempt to confirm rigorously a number of reports in the literature concerning likely or putative CE protein constituents. Many studies to date have employed methods such as indirect immunofluorescence at the light microscope level using antibodies to suspected CE proteins. A cell peripheral staining pattern was taken as evidence for involvement of the protein in the CE(
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ). However, the resolution of this method is limited to the wavelength of light, that is, roughly 500 nm. Because this is many times the width of the CE structure itself, there remains some doubt as to whether the putative protein is in fact involved in the CE structure, or located nearby on some other structure instead. Labeling with immunogold-tagged antibodies has far greater resolution of course, but so far, rigorously controlled studies have been performed for only involucrin(
      • Eckert R.L.
      • Yaffe M.B.
      • Crish J.F.
      • Murthy S.
      • Rorke E.A.
      • Welter J.F.
      ,
      • Steinert P.M.
      ), loricrin(
      • Mehrel T.
      • Hohl D.
      • Rothnagel J.A.
      • Longley M.A.
      • Bundman D.
      • Cheng C.
      • Lichti U.
      • Bisher M.E.
      • Steven A.C.
      • Steinert P.M.
      • Yuspa S.H.
      • Roop D.R.
      ,
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ,
      • Steven A.C.
      • Bisher M.E.
      • Roop D.R.
      • Steinert P.M.
      ), and SPR2(
      • Steinert P.M.
      ).

      Characterization of Peptide Material from CEs

      A previous study from this laboratory which met the above rigorous criterion was the first to document a protein cross-linked to the human epidermal CE in vivo(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ). Four peptides were recovered by limited proteolysis that contained two loricrin sequences adjoined by the isodipeptide cross-link. More recently, the efficacy of this method to ascertain CE protein composition and structure has been extended(
      • Steinert P.M.
      ). Limited progressive digestion with proteinase K for 36 h could release up to 90% of the CE protein mass as soluble peptides and thereby provided important clues about the possible order of assembly of the CE constituent proteins. Based on amino acid compositions of the released material, together with immunogold labeling of CE fragments and remnants recovered during the digestions(
      • Steinert P.M.
      ), the protein was apparently removed from the cytoplasmic side rather than the lipid envelope side. That is, the progressive digestion protocol excavated “deeper” into the CE structure. The purpose of the present study was to adapt the digestion procedures so as to release peptides that were suitable for amino acid microsequencing in order to (i) confirm for the first time that certain putative proteins are indeed cross-linked components of the CE; and (ii) provide rigorous information about CE structure and assembly. Optimized digestion conditions involved three steps: an initial trypsin digestion removed apparently contaminating non-cross-linked proteins; a 3-h proteinase K digestion released the outer (cytoplasmic) 30% of CE protein mass; and a second 6-h digestion interval released another 35%, corresponding to a middle third of CE mass.
      By direct amino acid sequencing of the many peptides recovered in this way, we present for the first time direct evidence for the eight proteins elafin, filaggrin, keratins 1, 2e, and 10 of KIF, SPR1, SPR2, and desmoplakin, in addition to loricrin, as isodipeptide cross-linked components of the human epidermal CE.
      Furthermore, notable differences in the sequences of the peptides released in the two proteinase K digestion intervals provide important information about likely CE composition and structure.

      A Model for the Structure of the Cytoplasmic (Outer) Two-thirds of the CE

      Peptides released in the first 3-h proteinase K digestion interval, corresponding to the outermost 30% of CE structure, consisted almost exclusively of loricrin (94%) and SPRs (5%). These data provide the strongest evidence to date that loricrin is the major component of the human epidermal CE. While almost every lysine and glutamine residue of loricrin was employed in the identified cross-links, the majority of the cross-links recovered utilized Gln215, Gln216, and Lys315 (Fig. 2, Table 1) in inter- and/or intrachain bonds. The intrachain cross-links would necessarily fold the molecule into a compact form. The predicted secondary surface structural features of loricrin indicate there are four glycine loop domains, flanked or interspersed by five Lys- or Gln-Lys-rich segments each of which harbors multiple utilized cross-linking sites (
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ) (Fig. 2). Because the glycine loop motifs are likely to be highly flexible in structure(
      • Hohl D.
      • Mehrel T.
      • Lichti U.
      • Turner M.L.
      • Roop D.R.
      • Steinert P.M.
      ,
      • Steinert P.M.
      • Mack J.W.
      • Korge B.P.
      • Gan S.-Q.
      • Haynes S.
      • Steven A.C.
      ), the present data indicate that the bulk of the CE consists of a flexible three-dimensional mesh-like array of compact cross-linked loricrin molecules. In addition, this array is also interspersed by smaller amounts of SPR1 and −2 (2-3% each). Interestingly, all of the 19 peptides involving SPRs utilized Lys or Gln residues located on or near their termini, rather than the multiple internal residues of each protein (Table 2). This data base supports the notion (
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ) that these proteins function as intermediary cross-bridges between and among the much larger amounts of loricrin.
      The second 6-h proteinase K digestion interval released peptides that probably originate from the central third of CE structure. While loricrin was again the quantitatively major component (about 81%), and SPRs represented 5%, this time interval was notable for the appearance of many peptides involving the novel protein elafin or its precursor (about 11%). As for the SPRs, it appears that elafin functions as a cross-bridging protein among loricrin. Thus the present data support an earlier hypothesis for the role of elafin in skin cross-linking reactions(
      • Nara K.
      • Ito S.
      • Suzuki Y.
      • Ghoneim M.A.
      • Tachibana S.
      • Hirose S.
      ).
      These data provide robust support for our new model on especially the latter stages of CE assembly (Fig. 3). Unknown early initiation stages of CE assembly presumably involve deposition of involucrin, cystatin α, and possibly other proteins, corresponding to the innermost third of the CE(
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ), and these are (or become) attached to the lipid envelope(
      • Downing D.T.
      ). Our new data indicate that a second major step involves addition to this initial scaffold of elafin, SPRs, and loricrin, corresponding approximately to the middle third of CE structure. Later, it appears that elafin deposition is reduced, while that of the SPRs and loricrin continues in a final stabilization event, corresponding to the outer third (cytoplasmic surface) of the CE.
      Figure thumbnail gr3
      Figure 3:Model of structure for the outer two-thirds of the human foreskin epidermal CE. The outermost (cytoplasmic surface) consists almost entirely of loricrin (large circles) enmeshed by SPRs (horizontal ovoids). Below the surface, significant amounts of elafin (vertical ovoids) is present. Keratin intermediate filaments bound together by filaggrin (small circles) are associated at these levels. These proteins are deposited over a scaffold of unknown structure likely to consist of involucrin, cystatin α, and perhaps other as yet unknown proteins. Together, this isodipeptide cross-linked proteinaceous component of the CE is attached to the lipid envelope(
      • Downing D.T.
      ).
      One implication of this model for the latter stages of CE assembly and structure is that the admixture of SPRs and elafin with loricrin would likely alter the flexibility characteristics of the loricrin array. This might constitute a novel method for altering the physical properties of the cornified layers of the epidermis. Indeed, elafin, SPR1, SPR2, and especially SPR3 are induced in response to epidermal injury (such as by UV light or drugs) or in hyperproliferative disorders (such as psoriasis)(
      • Kartasova T.
      • Cornelissen B.J.C.
      • van der Putte P.
      ,
      • Kartasova T.
      • van de Putte P.
      ,
      • Wiedow O.
      • Schroder J.-M.
      • Gregory H.
      • Young J.A.
      • Christophers E.
      ,
      • Schalkwijk J.
      • de Roo C.
      • de Jongh G.J.
      ,
      • Nonomura K.
      • Yaminishi K.
      • Yasuno H.
      • Nara K.
      • Hirose S.
      ), but it remains to be seen whether the CEs formed in these cases contain increased levels of the proteins. Further studies on the structure and expression of each of these proteins will be necessary to test this hypothesis.
      The cross-links involving keratin and filaggrin are also of interest. The total molar amounts recovered are <0.5% of CE protein mass, which corresponds to <0.05% of the estimated total amount of keratin/filaggrin in the cornified cell. It is possible that these have arisen as a result of random cross-linking by transglutaminases, and do not represent a physiologically important event in CE assembly; that is, they may have arisen as predicted by the “dustbin” hypothesis (
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ,
      • Michel S.
      • Schmidt R.
      • Robinson S.M.
      • Shroot B.
      • Reichert U.
      ). However, we consider this unlikely for two reasons. First, there was remarkable specificity in the lysines used in the keratin chains: of the 26 or 27 lysines in the keratin 1 or 2e chains, only the Lys73 or Lys70, respectively, were utilized (Table 3), which are located in a highly conserved window of sequences near the amino termini of the chains. This specificity would not be the anticipated result of the dustbin hypothesis. Moreover, it has been shown recently that a single-point mutation affecting this lysine residue in keratin 1 (Lys73→ Ile) is the cause of one case of the skin disease non-epidermolytic palmaplantar keratoderma (
      • Kimonis V.
      • DiGiovanna J.J.
      • Yang J.-M.
      • Doyle S.
      • Bale S.J.
      • Compton J.G.
      ). Indeed, these disease data together with the present finding support the alternative idea that this Lys residue has been precisely conserved so that it may be available for transglutaminase cross-linking reactions. Our data suggest that one such function may be linkage of the KIF cytoskeleton to the CE by the isodipeptide bond. This may provide a means for anchorage of the two structures and coordination of cornified epidermal cell structure. Further experiments are in progress to test this concept. Likewise, the three cross-links involving filaggrin utilized Gln residues in a highly conserved window of filaggrin sequences, suggesting specificity of transglutaminase reaction. This conclusion is not inconsistent with data suggesting that the bulk of filaggrin is loosely associated with and may be a contaminant of the isolated CEs: some filaggrin in the form of the filament-matrix component of the bulk of the epidermal cell mass is, like the keratins, also attached to the CE.
      The implications of the present identification of a single peptide involving desmoplakin sequences as a potential CE protein constituent are not yet clear. Because the sequence to which desmoplakin was cross-linked involved the commonly used elafin Gln2/Lys6 residues and keratin 1 Lys73 residue (Table 4), it seems unlikely that this is a dustbin peptide. However, it must be pointed out that two other important CE proteins, involucrin and cystatin α, were not seen in the present data set, despite the fact that they are thought to be quantitatively major CE protein components(
      • Hohl D.
      ,
      • Greenberg C.S.
      • Birckbichler P.J.
      • Rice R.H.
      ,
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ,
      • Rice R.H.
      • Green H.
      ,
      • Eckert R.L.
      • Yaffe M.B.
      • Crish J.F.
      • Murthy S.
      • Rorke E.A.
      • Welter J.F.
      ,
      • Steven A.C.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ). The expectation is that these two proteins occupy the inner portion of the CE, perhaps attached directly to the lipid envelope(
      • Steven A.C.
      • Steinert P.M.
      ,
      • Steinert P.M.
      ,
      • Downing D.T.
      ). It is possible that desmoplakin (and other proteins) may also constitute an inner component of CE structure. Further work is in progress to explore these issues.

      Utilization of Lysine and Glutamine Residues in Cross-links

      While a large body of in vitro data has been accumulated on the sequence specificity adjacent to lysines or glutamines required for the TGase 2 and factor XIIIa enzymes to form cross-links(
      • Folk J.E.
      ), little in vivo information is available, in part because few natural cross-linked structures have been explored to date, and in part because of the likelihood that the different TGase family members may have different substrate specificities. Only limited studies have been performed so far on the TGase 1 (
      • Kim S.-Y.
      • Kim I.-G.
      • Chung S.-I.
      • Steinert P.M.
      ) and TGase 3 (
      • Kim H.C.
      • Lewis M.S.
      • Gorman J.J.
      • Park S.-C.
      • Girard J.E.
      • Folk J.E.
      • Chung S.-I.
      ) enzymes that operate in the epidermis. The present body of cross-link data for several substrates (loricrin, SPRs, elafin, and keratins) indicate a clear preference for terminal sequence regions, and/or sequences that are predicted to be exposed on the protein surface. Similar observations on the in vitro cross-linking of eye lens β-crystallins have shown a clear preference for exposed terminal sequences(
      • Groenen P.J.
      • Smulders R.H.
      • Peters R.F.
      • Grootjans J.J.
      • van den Issel P.R.
      • Bloemendal H.
      • de Jong W.W.
      ). In vitro data using model peptide substrates have shown that the first Gln residue of adjacent Gln-Gln or Gln-rich sequences flanked by hydrophobic residues offer favorable amine acceptors for the TGase 2 enzyme(
      • Walsh F.
      ,
      • Parameswaran K.N.
      • Velasco P.T.
      • Wilson J.
      • Lorand L.
      ). This is the case in the present in vivo loricrin cross-link data: Gln215 of Gln215-Gln216-Val217 was used more often than either Gln of Gln305-Gln306-Lys307, or Gln-rich regions 3-10, 153-158, and 303-308 (Fig. 2). As more data accumulate on in vivo cross-links of natural substrates, better data on the properties and specificities of the TGases should become possible.

      Concluding Remarks

      The present data have provided evidence for the first time on the direct involvement of several proteins in CE structure. In addition, these data afford rigorous support for our modified model (
      • Steinert P.M.
      ) on the structure of the outer cytoplasmic two-thirds of the human epidermal CE, which correspond to the terminal reinforcement or maturation steps of CE assembly(
      • Hohl D.
      ,
      • Greenberg C.S.
      • Birckbichler P.J.
      • Rice R.H.
      ,
      • Reichert U.
      • Michel S.
      • Schmidt R.
      ,
      • Eckert R.L.
      • Yaffe M.B.
      • Crish J.F.
      • Murthy S.
      • Rorke E.A.
      • Welter J.F.
      ,
      • Marvin K.W.
      • George M.D.
      • Fujimoto W.
      • Saunders N.A.
      • Bernacki S.H.
      • Jetten A.M.
      ,
      • Steven A.C.
      • Steinert P.M.
      ). It remains for the future to design experiments to explore the earlier initial stages of CE assembly, presumably involving deposition of involucrin, cystatin α, and perhaps elafin, desmoplakin, KIF, and other proteins. Further carefully controlled digestion procedures using alternative enzymes may be able to “dig” deeper into the CE and recover suitable peptide material from lower levels. An alternative approach may be to use “immature” envelopes from cultured keratinocytes, in which only the earliest stages of CE formation have occurred, prior to the massive deposition of loricrin(
      • Steven A.C.
      • Steinert P.M.
      ).

      Acknowledgments

      We thank Drs. Eleonora Candi, Soo-Il Chung, Laszlo Fesus, John Folk, Tonja Kartasova, Ulrike Lichti, Edit Tarcsa, and Stuart Yuspa for their helpful comments during this work. William Lanahan (Beckman Instruments) provided valuable technical support and advice.

      REFERENCES

        • Hohl D.
        Dermatologica. 1990; 180: 201-211
        • Greenberg C.S.
        • Birckbichler P.J.
        • Rice R.H.
        FASEB J. 1991; 5: 3071-3077
        • Reichert U.
        • Michel S.
        • Schmidt R.
        Darmon M. Blumenberg M. Molecular Biology of the Skin. Academic Press Inc., New York1993: 107-150 (ed) pp. –
        • Rice R.H.
        • Green H.
        Cell. 1979; 11: 417-422
        • Eckert R.L.
        • Yaffe M.B.
        • Crish J.F.
        • Murthy S.
        • Rorke E.A.
        • Welter J.F.
        J. Invest. Dermatol. 1993; 100: 613-617
        • Takahashi M.
        • Tezuka T.
        • Katunuma N.
        FEBS Lett. 1992; 308: 79-82
        • Kartasova T.
        • Cornelissen B.J.C.
        • van der Putte P.
        Nucleic Acids Res. 1987; 15: 5945-5962
        • Kartasova T.
        • van de Putte P.
        Mol. Cell. Biol. 1988; 8: 2195-2203
        • Marvin K.W.
        • George M.D.
        • Fujimoto W.
        • Saunders N.A.
        • Bernacki S.H.
        • Jetten A.M.
        Proc. Natl. Acad. Sci. U. S. A. 1992; 89: 11026-11030
        • Gibbs S.
        • Fijneman R.
        • Wiegart J.
        • Geurts van Kessel A.
        • van de Putte P.
        • Backendorf C.
        Genomics. 1993; 16: 630-637
        • Mehrel T.
        • Hohl D.
        • Rothnagel J.A.
        • Longley M.A.
        • Bundman D.
        • Cheng C.
        • Lichti U.
        • Bisher M.E.
        • Steven A.C.
        • Steinert P.M.
        • Yuspa S.H.
        • Roop D.R.
        Cell. 1990; 61: 1103-1112
        • Hohl D.
        • Mehrel T.
        • Lichti U.
        • Turner M.L.
        • Roop D.R.
        • Steinert P.M.
        J. Biol. Chem. 1991; 266: 6626-6636
        • Yoneda K.
        • Hohl H.
        • McBride O.W.
        • Wang M.
        • Cehrs K.U.
        • Idler W.W.
        • Steinert P.M.
        J. Biol. Chem. 1992; 267: 18060-18066
        • Yoneda K.
        • Steinert P.M.
        Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 10754-10758
        • Lee S.-C.
        • Kim I.-G.
        • Marekov L.N.
        • O'Keefe E.J.
        • Parry D.A.D.
        • Steinert P.M.
        J. Biol. Chem. 1993; 268: 12164-12176
        • Richards S.
        • Scott I.R.
        • Harding C.R.
        • Liddell J.E.
        • Powell G.M.
        • Curtis C.G.
        Biochem. J. 1988; 253: 153-160
        • Steven A.C.
        • Steinert P.M.
        J. Cell Sci. 1994; 107: 693-700
        • Abernethy J.L.
        • Hill R.L.
        • Goldsmith L.A.
        J. Biol. Chem. 1977; 252: 1837-1839
        • Ming M.E.
        • Daryanami H.A.
        • Roberts J.E.
        • Haimowitz J.E.
        • Kvedar J.C.
        J. Invest. Dermatol. 1993; 100 578
        • Tezuka T.
        • Takahashi M.
        J. Invest. Dermatol. 1987; 88: 47-51
        • Wiedow O.
        • Schroder J.-M.
        • Gregory H.
        • Young J.A.
        • Christophers E.
        J. Biol. Chem. 1990; 265: 14791-14795
        • Schalkwijk J.
        • de Roo C.
        • de Jongh G.J.
        Biochim. Biophys. Acta. 1991; 1096: 148-154
        • Nonomura K.
        • Yaminishi K.
        • Yasuno H.
        • Nara K.
        • Hirose S.
        J. Invest. Dermatol. 1994; 103: 88-91
        • Saheki T.
        • Ito F.
        • Hagiwara H.
        • Saito Y.
        • Kuroki J.
        • Tachibana S.
        • Hirose S.
        Biochem. Biophys. Res. Commun. 1992; 185: 240-245
        • Molhuizen H.O.F.
        • Alkemade H.A.C.
        • Zeeuwen P.L.J.M.
        • de Jongh G.J.
        • Wieringa B.
        • Schalkwijk J.
        J. Biol. Chem. 1993; 268: 12028-12032
        • Aeschilimann D.
        • Paulsson M.
        Thromb. Haemostatis. 1994; 71: 402-415
        • Yaffe M.B.
        • Murthy S.
        • Eckert R.L.
        J. Invest. Dermatol. 1993; 100: 3-9
        • Baden H.P.
        • Kubilus J.
        • Phillips S.B.
        • Kvedar J.C.
        • Tahan S.R.
        Biochim. Biophys Acta. 1987; 925: 63-73
        • Phillips S.B.
        • Kubilus J.
        • Grassi A.M.
        • Goldaber M.
        • Baden H.P.
        Comp. Biochem. Physiol. 1990; 95B: 781-788
        • Greco M.A.
        • Lorand L.
        • Lane W.S.
        • Baden H.P.
        • Parameswaran K.N.P.
        • Kvedar J.V.
        J. Invest. Dermatol. 1995; 104: 204-210
        • Haftek M.
        • Serre G.
        • Mils V.
        • Thivolet J.
        J. Histochem. Cytochem. 1991; 19: 1531-1538
        • Steinert P.M.
        • Idler W.W.
        Biochemistry. 1979; 18: 5664-5669
        • Yuspa S.H.
        • Ben T.
        • Steinert P.
        J. Biol. Chem. 1982; 257: 9906-9908
        • Tarcsa E.
        • Kedei N.
        • Thomazy V.
        • Fesus L.
        J. Biol. Chem. 1992; 267: 25648-25651
        • Steinert P.M.
        Cell Death Differ. 1995; 2: 33-40
        • Steven A.C.
        • Bisher M.E.
        • Roop D.R.
        • Steinert P.M.
        J. Struct. Biol. 1990; 104: 150-162
        • Downing D.T.
        J. Lipid Res. 1992; 33: 301-313
        • Dale B.A.
        • Holbrook K.A.
        • Kimball J.R.
        • Hoff M.
        • Sun T.-T.
        J. Cell Biol. 1983; 101: 1257-1269
        • Yuspa S.H.
        • Harris C.C.
        Exp. Cell Res. 1974; 86: 95-105
        • McKinley-Grant L.J.
        • Idler W.W.
        • Bernstein I.A.
        • Parry D.A.D.
        • Cannizzaro L.
        • Croce C.M.
        • Huebner K.
        • Lessin S.R.
        • Steinert P.M.
        Proc. Natl. Acad. Sci. U. S. A. 1989; 86: 4848-4852
        • Steinert P.M.
        • Parry D.A.D.
        • Idler W.W.
        • Johnson L.D.
        • Steven A.C.
        • Roop D.R.
        J. Biol. Chem. 1985; 260: 7142-7149
        • Collin C.
        • Moll R.
        • Kubicka S.
        • Ouhayoun J.-P.
        • Franke W.W.
        Exp. Cell Res. 1992; 202: 132-141
        • Zhou X.-M.
        • Idler W.W.
        • Steven A.C.
        • Roop D.R.
        • Steinert P.M.
        J. Biol. Chem. 1988; 263: 15584-15589
        • Kyte J.
        • Doolittle R.F.
        J. Mol. Biol. 1982; 157: 105-132
        • Devereux J.
        • Haeberli P.
        • Smithies O.
        Nucleic Acids Res. 1984; 12: 387-3932
        • Backendorf C.
        • Hohl D.
        Nature Genet. 1992; 2 91
        • Gan S.-Q.
        • Idler W.W.
        • McBride O.W.
        • Markova N.G.
        • Steinert P.M.
        Biochemistry. 1990; 29: 9432-9440
        • Nara K.
        • Ito S.
        • Suzuki Y.
        • Ghoneim M.A.
        • Tachibana S.
        • Hirose S.
        J. Biochem. (Tokyo). 1994; 115: 441-448
        • Moore J.T.
        • Hagstrom J.
        • McCormick D.J.
        • Harvey S.
        • Madden B.
        • Holicky E.
        • Stanford D.R.
        • Wieben E.D.
        Proc. Natl. Acad. Sci. U. S. A. 1987; 84: 6712-6714
        • Green K.J.
        • Parry D.A.D.
        • Steinert P.M.
        • Virata M.L.A.
        • Wagner R.M.
        • Angst B.D.
        • Nilles L.A.
        J. Biol. Chem. 1990; 265: 2603-2612
        • Kouklis P.D.
        • Hutton E.
        • Fuchs E.
        J. Cell Biol. 1994; 127: 1049-1060
        • Steinert P.M.
        • Mack J.W.
        • Korge B.P.
        • Gan S.-Q.
        • Haynes S.
        • Steven A.C.
        Int. J. Biol. Macromol. 1991; 13: 130-139
        • Michel S.
        • Schmidt R.
        • Robinson S.M.
        • Shroot B.
        • Reichert U.
        J. Invest. Dermatol. 1987; 88: 301-305
        • Kimonis V.
        • DiGiovanna J.J.
        • Yang J.-M.
        • Doyle S.
        • Bale S.J.
        • Compton J.G.
        J. Invest. Dermatol. 1994; 103: 713-719
        • Folk J.E.
        Adv. Enzymol. 1983; 54: 1-56
        • Kim S.-Y.
        • Kim I.-G.
        • Chung S.-I.
        • Steinert P.M.
        J. Biol. Chem. 1994; 269: 27979-27986
        • Kim H.C.
        • Lewis M.S.
        • Gorman J.J.
        • Park S.-C.
        • Girard J.E.
        • Folk J.E.
        • Chung S.-I.
        J. Biol. Chem. 1990; 265: 21971-21978
        • Groenen P.J.
        • Smulders R.H.
        • Peters R.F.
        • Grootjans J.J.
        • van den Issel P.R.
        • Bloemendal H.
        • de Jong W.W.
        Eur. J. Biochem. 1994; 220: 795-799
        • Walsh F.
        Nature. 1963; 198: 188-189
        • Parameswaran K.N.
        • Velasco P.T.
        • Wilson J.
        • Lorand L.
        Proc. Natl. Acad. Sci. U. S. A. 1990; 87: 8472-8475