In Vitro Amyloidogenic Peptides of Galectin-7

Background: Characterization of amyloid precursor protein and the mechanism of amyloidogenesis in primary localized cutaneous amyloidosis have not been elucidated previously. Results: Galectin-7 fragments containing β-strand peptides are highly amyloidogenic in vitro. Conclusion: Galectin-7 is an amyloid precursor protein in primary localized cutaneous amyloidosis. Significance: We have proposed the possible mechanism of amyloid deposition in primary localized cutaneous amyloidosis. Pathogenesis of primary localized cutaneous amyloidosis (PLCA) is unclear, but pathogenic relationship to keratinocyte apoptosis has been implicated. We have previously identified galectin-7, actin, and cytokeratins as the major constituents of PLCA. Determination of the amyloidogenetic potential of these proteins by thioflavin T (ThT) method demonstrated that galectin-7 molecule incubated at pH 2.0 was capable of binding to the dye, but failed to form amyloid fibrils. When a series of galectin-7 fragments containing β-strand peptides were prepared to compare their amyloidogenesis, Ser31-Gln67 and Arg120-Phe136 were aggregated to form amyloid fibrils at pH 2.0. The rates of aggregation of Ser31-Gln67 and Arg120-Phe136 were dose-dependent with maximal ThT levels after 3 and 48 h, respectively. Their synthetic analogs, Phe33-Lys65 and Leu121-Arg134, which are both putative tryptic peptides, showed comparable amyloidogenesis. The addition of sonicated fibrous form of Ser31-Gln67 or Phe33-Lys65 to monomeric Ser31-Gln67 or Phe33-Lys65 solution, respectively, resulted in an increased rate of aggregation and extension of amyloid fibrils. Amyloidogenic potentials of Ser31-Gln67 and Phe33-Lys65 were inhibited by actin and cytokeratin fragments, whereas those of Arg120-Phe136 and Leu121-Arg134 were enhanced in the presence of Gly84-Arg113, a putative tryptic peptide of galectin-7. Degraded fragments of the galectin-7 molecule produced by limited trypsin digestion, formed amyloid fibrils after incubation at pH 2.0. These results suggest that the tryptic peptides of galectin-7 released at neutral pH, may lead to amyloid fibril formation of PLCA in the intracellular acidified conditions during keratinocyte apoptosis via regulation by the galectin-7 peptide as well as actin and cytokeratins.

Primary localized cutaneous amyloidosis (PLCA) 2 is a skinlimited amyloidosis that is clinically characterized by chronic pruritic papules (lichen amyloidosus) or macules (macular amyloidosis) and histologically characterized by the deposition of amyloid in the superficial dermis (1). PLCA is relatively common in South America and Southeast Asia, and most cases are the acquired type with unknown pathogenesis. The amyloid precursor protein of PLCA has been hypothesized to be cytokeratins based on the immunohistochemical reactivity of amyloid deposit with cytokeratin antibodies (2,3). The immunohistochemical positivity with cytokeratins in the lesional skin, however, does not always indicate that cytokeratins are amyloidogenic precursor proteins. Because cytokeratins are major intracellular proteins of keratinocytes that are present as a stable form of keratin intermediate filaments (KIF), immunoreactive epitopes of cytokeratins may remain intact even after completion of apoptotic degradation and amyloid accumulation. In addition, it has been demonstrated that KIF and its proteolytic fragments do not react with thioflavin T (ThT) (4). Furthermore, the secondary structure of cytokeratin consisits of ␣-helical, coiled-coil rod domains (5), whereas more than 20 amyloidogenic proteins that can bind Congo red or ThT share common cross-␤ sheet structures, which is defined as ␤-fibrillosis (6). Recently, we identified galectin-7 (Gal-7), actin, and cytokeratins as well as serum amyloid P component (SAP) and apolipoprotein E (Apo E) as the major components of amyloid deposits of PLCA in the water soluble fraction recovered from lesional skin, and we found that all of these proteins were present in the amyloid deposits of PLCA according to immunohistochemical studies (2,(7)(8)(9). SAP and Apo E are universal nonfibrillar constituents of amyloid deposits and will therefore be excluded from the candidates for amyloid precursor protein. The next questions are which protein (Gal-7, actin, and cytokeratins) is directly related to amyloidogenesis and how the protein aggregates to form amyloid fibrils.
The initial event of amyloid formation in PLCA is thought to be related to keratinocyte apoptosis because 1) TUNEL-posi-tive keratinocytes are commonly detected in the epidermis overlying the amyloid-laden dermis (10), and 2) amyloid deposits are restricted to the papillary dermis just beneath the thinned epidermis, suggesting that a hyperproliferative thickened epidermis undergoes extensive apoptotic cell death (filamentous degeneration) (11) and is then substituted by amyloid materials via an unknown mechanism. Apoptosis of keratinocytes, which is induced by UVB irradiation, reactive oxygen species (ROS), drugs or chemical reagents accompanies the activation of certain proteolytic machinery (12). Three major apoptosis-related enzymes of keratinocytes have been reported, including caspases, cathepsin D (cath D), and trypsin-like enzymes, although their final target molecules and their exact roles in apoptosis are not fully known. Nine caspases are expressed in normal human keratinocytes; of these, caspases 3, 8, and 9 are activated in UVB-induced apoptosis (13); in particular, caspase 3 appears to be important because its activity is enhanced by apoptosis-induced Gal-7 (14). Cath D, a major lysosomal, acidic protease of the epidermis, induces apoptosis via directly priming caspase-8 activation (15,16). Keratinocytes potentially synthesize multiple forms of trypsin-like serine proteases including trypsinogen 4, 5, and kallikrein-related peptidases (KLKs) (17,18). A tryptic serine protease secreted by cultured keratinocytes potentially induces keratinocyte apoptosis, which is blocked by soybean trypsin inhibitor, but the precise mechanism is currently unknown (19).
The general mechanism of amyloid fibril formation is not fully understood. A nucleation-dependent polymerization model is proposed for the general mechanism of amyloid fibril formation in several types of amyloidosis, such as ␤2-microglobulin in dialysis amyloidosis and amyloid-␤ (A␤) in Alzheimer disease. Their amyloidogenic models consist of two fundamental phases, nucleation (seed formation), and extension (deposition). Seed formation requires the assembly of monomeric protein into a template, representing the rate-limiting step in amyloid fibril formation. Once the seed has been formed, growth of the template by the deposition of additional monomeric protein becomes thermodynamically favorable, resulting in rapid extension of amyloid fibrils (20,21).
Galectins constitute a large family of ␤-galactoside-binding lectins. At least 14 mammalian galectins that share structural similarities in their carbohydrate recognition domains have been reported. Despite the lack of signal peptide, galectins can be secreted by an endoplasmic reticulum (ER)-Golgi-independent pathway. They can be found both intracellularly (cytoplasmic and/or nuclear) and extracellularly as well as, sometimes, in association with the plasma membrane, although they do not contain a transmembrane domain. From these variable subcellular locations, galectins have been implicated in a wide range of cellular processes, including cell-cell and cell-matrix interactions, extracellular matrix remodelling, cell cycle, cancer biology, intracellular trafficking, and apoptosis. Some galectins are pro-apoptotic, whereas others are anti-apoptotic; some galectins induce apoptosis by binding to cell surface glycoproteins, whereas others regulate apoptosis through interacting with intracellular proteins (22,23). Gal-7 is a prototype galectin, with a molecular weight of 14 kDa (136 amino acid residues), that is expressed only in stratified epithelia (24). Its structure consists of ␤-sandwich with the packing of two ␤-sheets (25). Gal-7 is associated with multiple keratinocyte functions including cell migration and apoptosis. Gal-7-deficient mice display reduced reepithelialization potential compared with wild-type littermates (26). Gal-7 expression is markedly induced during UVB-induced apoptotic processes of epidermal keratinocytes, paralleling p53 stabilization (27); the keratinocytes overexpressing Gal-7 undergo apoptosis, indicating that Gal-7 is a pro-apoptotic protein (14,28).
In the present study, we 1) determine the amyloidogenic potentials of the candidate proteins, Gal-7, actin, and cytokeratins; 2) determine which part of Gal-7 is involved in amyloidogenesis; and 3) show whether degraded fragments of Gal-7 produced by keratinocyte apoptosis-related proteases can form amyloid fibrils.
Insoluble cytokeratins were extracted from cultured human keratinocytes (HaCaT cells) in phosphate-buffered saline (PBS) containing 0.6 M KCl, 1% Triton X-100, and 0.2 mg/ml DNase I (31). The precipitate was dissolved in an 8 M urea/0.1 M ␤-ME at room temperature for 4 h and then centrifuged at 10,000 ϫ g for 30 min. The supernatant was dialyzed four times against PBS; then, the resulting aggregates, which largely comprise cytokeratins (referred as insoluble cytokeratins), were redissolved in 8 M urea/0.1 M ␤-ME solution and reaggregated by dialysis against PBS. The procedure was repeated four times (4). Salt soluble cytokeratins were extracted from the insoluble cytokeratin aggregates using the buffer (10 mM Tris-HCl, pH 7.5 containing protease inhibitors; 1 mM EDTA, 1 mM NEM, and 1 mM PMSF) (32), which was followed by purification with molecular sieve chromatography. The fractionated proteins with molecular weights of 45-70 kDa were subjected to SDS-PAGE and then confirmed as cytokeratins by immunoblot analysis using anti-pankeratin (polyclonal, DAKO, Glostrup, Denmark) or cytokeratin-5 antibody (monoclonal, Novus Biologicals Inc., Littleton, CO) (not shown). The average molecular weights of cytokeratin monomers were estimated as 65 kDa.
Amyloid fibrils were sequentially extracted from the lesional skin of localized cutaneous amyloidosis with distilled water as described previously (7). The yield of the protein in the watersoluble fraction was ϳ90 g.
In Vitro Formation of Amyloid Fibril Aggregates-Prior to starting the thioflavin T (ThT) assay, we found that some peptides were easily dissolved in water, but some were difficult. In some experiments, we prepared a stock solution of the peptides with 100% Me 2 SO (33), and we adjusted the reaction mixtures to give the same final concentration of Me 2 SO. Proteins or peptides were placed into oil-free PCR tubes and, then transferred into a DNA thermal cycler (TP600, Takara Bio Inc., Otsu, Japan). The plate temperature was elevated from 4°C to 37°C at a maximal speed. The reaction was stopped by placing the tubes on ice. The buffers we used were 50 mM citrate-HCl buffer (pH 1.0, 2.0, and 4.0), 50 mM sodium phosphate buffer (pH 6.0), 50 mM Tris-HCl buffer (pH 8.0), and 50 mM glycine-NaOH buffer (pH 10.0). All buffers contained 100 mM NaCl because well-organized, mature amyloid fibrils were readily formed in the presence of a high dose of NaCl (34). Amyloid fibril formation was measured with the ThT method (35,36). Proteins or peptides were dissolved in each buffer at various doses and were then incubated at various pH ranges at 37°C for 3, 6, 12, 18, and 24 h, or 2-7 days. An aliquot (10 l) was mixed with 1 ml of 5 M ThT (Wako Pure Chemical Industries Ltd., Osaka, Japan) solution in 50 mM glycine-NaOH buffer, pH 8.5. After 1 min, the fluorescence of thioflavin T was determined on a Hitachi F-2000 fluorescence spectrophotometer at an excitation wavelength of 435 nm and emission wavelength at 485 nm. Assays were performed in triplicate, and the values were averaged to provide final data. The error in each data set was below 18%.
Transmission Electron Microscopy-An aliquot was taken from the reaction mixtures, which had been prepared to determine the ThT fluorescence, applied to carbon-coated copper grids, and then subjected to electron microscopic observation by negative staining with 2% phosphotungstic acid, pH 7.0 (37) using a JEM-1010 electron microscope (JEOL Ltd., Tokyo, Japan).
Deposition of Gal-7 Peptides onto Preformed Fibrils-To determine the deposition ability of the Gal-7 peptides, we prepared seeds composed of Ser 31 -Gln 67 or Phe 33 -Lys 65 mono-mers as previously described (38). Briefly, the solutions of Ser 31 -Gln 67 and Phe 33 -Lys 65 (500 M each) were centrifuged at 4°C for 1 h at 15,000 ϫ g to remove preformed aggregates. Both peptides were incubated at 37°C in 50 mM citrate buffer, pH 2.0, containing 100 mM NaCl for 3 h, which was the time required to reach the maximal ThT level. The fibrils formed, designated as f(Ser 31 -Gln 67 ) or f(Phe 33 -Lys 65 ) were pelleted by centrifugation at 15,000 ϫ g for 1 h and suspended in distilled water at the concentration of 25 nmol/100 l. The recovery of the aggregates was more than 86%, as judged by measurement of the ThT fluorescence before and after centrifugation. The aggregates were sonicated twice (pulse: 1.2 s, output level: 2) for 5 min with an ultrasonic disruptor (UC100-D, Olympus, Tokyo, Japan) at 4°C. An aliquot of this fibril suspension was added to Ser 31 -Gln 67 , Phe 33

Fluorescence Spectra of ThT in the Presence of Amyloid Fibrils
Extracted from PLCA-Water-soluble amyloid fibrils (fPLCA) (30 g/ml) were extracted from lesional skin of PLCA as previously described (7) and showed a novel fluorescence at 485 nm with an excitation maximum at 435 nm in the presence of 5 M of ThT in 50 mM glycine-NaOH buffer, pH 8.5 (Fig. 1A). The excitation and emission wavelengths in the following ThT assays were fixed at 435 and 485 nm, respectively. -7), Actin, and Cytokeratin Molecules-The emission spectra of the Gal-7, actin and cytokeratin molecules at the dose of 50 M each were determined with an excitation spectrum at 435 nm. Gal-7 showed a novel peak at 485 nm at pH 2.0 after 3 h, but there was no fluorescence peak at pH 7.0 (Fig. 1B). Actin and cytokeratins showed no fluorescence at both pH 2.0 and 7.0 (not shown).

Fluorescence of ThT in the Presence of Galectin-7 (Gal
The effects of pH on ThT fluorescence in the presence of Gal-7, actin and soluble cytokeratins (50 M each) were examined. ThT fluorescence of Gal-7 was pH-dependent with a maximal fluorescence after 3 h at pH 2.0, while actin and soluble cytokeratins showed no increase in ThT fluorescence in any pH ranges (Fig. 1C).
The gradual decrease of fluorescence for 24 h, but actin and soluble cytokeratins showed no significant change for 24 h (Fig. 1D).
Electron microscopic observation of Gal-7 after 3 h of incubation at pH 2.0 showed short, curled fibrillar structures that were not consistent with the morphology of amyloid fibrils (Fig. 1E).
Synthetic peptides containing the ␤-strand domain (Leu 8 -Pro 38 (Fig. 4, A and C, left panel). Phe 33 -Lys 65 , Ser 31 -Lys 65 , Phe 33 -Ser 64 , and Arg 32 -Glu 66 (200 M each) showed considerable but relatively lower amyloidogenic potentials than those from the Ser 31 -Gln 67 fragment for the initial 18 h maintaining the high ThT values for 7 days (not shown), whereas Phe 33 -Asn 63 , Val 35 -Lys 65 and His 34 -Lys 65 exhibited an increased ThT level for 3, 5, and 7 days after a lag time of 1 day. Phe 33 -Gln 67 had the lowest amyloidogenic potential for the 7-day incubation (Fig. 4A, right panel). The fragments with a high ThT fluorescence were found to have varied appearances that depended on the peptide size and amino acid residues at both the N-and C-terminal ends by electron microscopy (Fig.  4B). The Leu 121 -Phe 136 and Leu 121 -Arg 134 fragments showed considerable amyloidogenic potentials with lower ThT levels than Arg 120 -Phe 136 , while Arg 120 -Ile 135 and Leu 121 -Ile 135 had the lowest amyloidogenic activity (Fig. 4C, right panel). Amyloid fibrils with various morphologies that were formed from Arg 120 -Phe 136 , Leu 121 -Arg 134 , and Leu 121 -Phe 136 were observed on an electron microscope (Fig. 4D).
Extension Reaction of Ser 31 5A, middle panel). The combination of soluble Phe 33 -Lys 65 (10 or 50 M) and f(Phe 33 -Lys 65 ) (2.5 or 5 nmol/100 l) at pH 2.0 also gave higher ThT values than the individual soluble or fibril forms (Fig. 5A, right panel). Electron microscopic studies demonstrated that the amyloid fibrils that formed after incubation with the soluble form (Ser 31 -Gln 67 or Phe 33 -Lys 65 peptide) at pH 2.0 appeared to be extended (Fig. 5B) compared with the fibrils that were composed of their respective monomers (See Fig. 4A).
In Vitro Amyloid Fibril Formation of Protease-digested Fragments of Gal-7-Because proteolytic fragments of amyloid precursor proteins such as A␤ and ␤2-microglobulin have been shown to be more amyloidogenic than intact proteins (39,40), we re-examined the amyloidogenesis of the degradation products of three candidate proteins (Gal-7, actin, and insoluble cytokeratins) that are produced by the digestion with major apoptosis-related proteases of keratinocytes, including trypsin, cath D, and caspase 3. The degradation patterns of these proteins by trypsin, cath D, and caspase 3 are shown in Fig. 7A. Incubation of the degradation products of Gal-7 at pH 2.0 obtained after 3-h trypsin treatment showed increased ThT fluorescence during the 72-h incubation, but the degraded fragments after 1-and 8-h trypsin treatments did not (Fig. 7B). Gal-7 fragments, obtained by cath D and caspase 3, as well as actin and cytokeratin fragments, obtained by the digestions with three proteases, demonstrated no significant increase in ThT fluorescence (not shown). Electron microscopy of samples taken from 3-h trypsin digests after 72-h incubation at pH 2.0 showed straight non-branching amyloid-like structures (Fig.  7C), which was not the case for other samples from 1-and 8-h trypsin digests (not shown).

Amyloidogenesis of Gal-7, Actin, and Cytokeratin Molecules-
Determination of the amyloidogenic potentials of Gal-7, actin, and cytokeratin molecules in vitro revealed that Gal-7 alone showed an increase in fluorescence at an acidic pH (pH 2.0), and the increase was not enough to form typical amyloid fibrils, which is possibly because of the incomplete steric interaction between the ␤-strand peptides of the Gal-7 molecule (41,42).
Specific short peptides of amyloidogenic proteins, rather than full-length molecules, easily aggregate to form amyloid fibrils. For instance, in A␤ of Alzheimer disease highly amyloidogenic peptides 1-40 and 1-42 are processed from non-pathological amyloid precursor protein (APP) (43), and, in ␤2-microglobulin of dialysis amyloidosis, specific peptides (Ser 20 -Lys 41 or Asp 59 -Thr 71 ) form amyloid fibrils more readily than intact ␤2-microglobulin molecule (33,40,44). This suggests that the fragments of Gal-7, actin, or cytokeratins, even though their full-length molecules have little to no amyloidogenic properties, may exhibit strong amyloidogenesis. To examine this possibility, the amyloidogenic potentials of 1) synthetic peptides with ␤-strand fragments of Gal-7 and actin, and 2) proteolytic fragments of candidate proteins (Gal-7, actin, and cytokeratins) produced by apoptosis-related proteases (trypsin, cath D, and caspase-3) were determined.
Amyloidogenesis of Synthetic Peptides-Of the synthetic peptides of Gal-7 molecule, Gal S3-S5 (Ser 31 -Gln 67 ) and Gal S2-F1 (Arg 120 -Phe 136 ) showed strong amyloidogeneses at pH 2.0 with the lag time of 3 h and 48 h, respectively (see Fig. 3C). Kinetic studies on aggregation using various Ser 31 -Gln 67 analogs showed that the Phe 33 -Lys 65 and Ser 31 -Lys 65 variants that lack two amino acid residues at both ends and at the C-terminal end, respectively, were capable of maintaining their amyloidogenic potential, whereas Phe 33 -Gln 67 lacking N-terminal two amino acid residues did not express any amyloidogenesis at least within 7 days (see Fig. 4A). This indicates that 1) Ser 31 -Gln 67 has the highest aggregation activity, 2) Phe 33 -Lys 65 seems to be the minimum sequence required for maintaining a high aggregation potential, because variants of Phe 33 -Lys 65 lacking one or two amino acid residues at N-terminal (His 34 -Lys 65 and Val 35 -Lys 65 ) or C-terminal end (Phe 33 -Asn 63 and Phe 33 -Ser 64 ) all show lower aggregation potentials, and 3) the N-terminal two amino acid residues of Ser 31 -Gln 67 (Ser 31 -Arg 32 ) seem to have a promoting potential whereas C-terminal two residues (Glu 66 -Gln 67 ) have an inhibiting activity, when variants of Ser 31 -Gln 67 lacking two amino acid residues at N-or C-terminal end (Phe 33 -Gln 67 , Ser 31 -Gln 67 , and Ser 31 -Lys 65 ) are compared.
Among the analogs of the Gal S2-F1 fragment (Arg 120 -Phe 136 ), Leu 121 -Phe 136 , and Leu 121 -Arg 134 were amyloidogenic, compared with Arg 120 -Ile 135 and Leu 121 -Ile 135 , for the 7-day incubation, indicating that the C-terminal amino acid residue of Arg 120 -Phe 136 is important for amyloidogenesis. It is of particular interest that amyloidogenic peptides, Phe 33 -Lys 65 , Leu 121 -Phe 136 , and Leu 121 -Arg 134 , are all putative tryptic peptides (See Fig. 2B); therefore, several amyloidogenic fragments may be released after the digestion of Gal-7 with trypsin-like enzymes, although the tryptic cleavage sites Arg 54 and Arg 134 are located in the Phe 33 -Lys 65 and Leu 121 -Phe 136 peptides, respectively, and their susceptibilities to the trypsin-like enzymes of keratinocytes are unknown. These in vitro studies using synthetic peptides suggest that tryptic degradation of Gal-7 will play a major role in amyloidogenesis of PLCA.
Amyloidogenesis of Proteolytic Fragments-We next examined whether the mixture of degraded peptides of Gal-7 obtained after trypsin digestion (1-8 h) might be amyloidogenic. Gal-7 fragments produced by trypsin under the restricted condition of 3 h digestion were amyloidogenic. The digestion time-dependent amyloidogenesis suggests that the susceptibilities of the cleavage sites (a total of 16 sites) of Gal-7 to trypsin are not the same. The K and R residues (Lys 7 , Arg 21 , Arg 23 , Arg 54 , Lys 65 , Arg 75 , Lys 99 , Arg 111 , and Arg 134 ) located inside the ␤-sheet domains will be relatively resistant to trypsin because the tertiary structure of ␤-sheet may inhibit the access of the enzyme compared with Arg 83 , Arg 113 , Arg 118 , and Arg 120 , which are located outside the ␤-sheet domains (See Fig.  2B), except Arg 15 and Arg 72 , which are difficult to cleave because their C-terminal amino acid residues are acidic (P and E, respectively) (45). During the short incubation time (3 h), the peptides, Gly 84 -Arg 113 and Leu 121 -Phe 136 , of which tryptic cleavage sites are outside the ␤-sheet domains will be released; the latter peptide is shown to be amyloidogenic and the former peptide, although not amyloidogenic by itself, potentially stimulates Leu 121 -Phe 136 to form amyloid fibrils (see Fig. 6B). As digestion progresses (8 h), the tryptic sites inside the ␤-sheet domains begin to be cleaved, resulting in the degradation of the Gly 84 -Arg 113 peptide by the tryptic sites (Lys 99 and Arg 111 ) located inside the peptide and the production of Leu 121 -Arg 134 peptide with a relatively low amyloidogenic potential (see Fig.  4C). These hypotheses seem to explain well why limited trypsin treatment for 3 h produces amyloid fibrils. Keratinocytes do not express pancreatic type trypsin (trypsinogens 1 and 2), but they do produce trypsinogens 4 and 5 and KLKs. Their functions and substrate specificities have not been fully elucidated, but among them, KLK 4, 5, 6, and 8 are known to exhibit trypsin-like specificity (17,18). Although the identification of the tryptic enzymes involved in Gal-7 degradation in vivo and the possibility of the involvement of other species of caspases and cathepsins still remain to be determined, the present studies suggest that at least limited degradation of Gal-7 by tryptic enzymes is involved in the production of amyloidogenic fragments in PLCA.
Amyloid Fibril Extension-We used Ser 31 -Gln 67 and Phe 33 -Lys 65 fragments in the extension studies. Amyloidogenic peptides, Ser 31 -Gln 67 and Phe 33 -Lys 65 monomers were both capable of depositing on their respective preformed aggregates and of elongating the amyloid fibrils, suggesting that amyloid fibrils consisting of Ser 31 -Gln 67 and Phe 33 -Lys 65 peptides are formed via two steps, seed formation (the assembly of monomeric amyloid protein into a template) and extension (the growth of the template by depositions of additional monomeric amyloid protein) (20). However, incubation of the intact Gal-7 molecule in the presence of f(Ser 31 -Gln 67 ) or f(Phe 33 -Lys 65 ) failed to increase the rate of elevation of ThT fluorescence. This is in contrast with dialysis amyloidosis in which the combination of the K3 (Ser 20 -Lys 41 peptide) seed and intact ␤ 2 M monomer resulted in the extension of amyloid fibrils (33,38), suggesting that intact Gal-7 cannot be a promoter for amyloid fibril extension and that degradation of Gal-7 is essential for the extension of amyloid fibrils.
Acidified Condition of Apoptotic Cell-The optimal pH for amyloid fibril formation by galectin-7 segments (Ser 31 -Gln 67 , Phe 33 -Lys 65 , Arg 120 -Phe 136 , Leu 121 -Arg 134 , etc.) was acidic (pH 2.0), which is not physiological. However, cytoplasmic acidification is now recognized as a general feature of apoptosis in a variety of cell lines, including neutrophils, T-lymphocytes and leukemia cells. Because intracellular acidification, resulting from pH dysregulation, which is in turn due to dephosphorylation of protein exchangers, occurs at early stages of apoptosis (46 -48), amyloidogenic galectin-7 segments should be pro-duced by tryptic enzymes at neutral pH prior to amyloid fibril formation in the intracellular acidified conditions. Previous report that intracellular acidification of apoptosis is preceded by protease (including caspases) activation and that inhibition of the protease activity prevents cytoplasmic acidification in Jurkat T-lymphoblasts (49), strongly suggests a close relationship between apoptosis-induced proteolysis and subsequent cytoplasmic acidification. This supports our hypothesis on the mechanism of amyloid fibril formation in PLCA, which will be discussed for Fig. 8.
Regulation of Amyloid Fibril Deposit-Apoptotic cells will be replaced by amyloid fibrils via complex and multi-step processes. Apoptotic cell death in the epidermis is frequently detected in skin disorders that are unrelated to PLCA, including lichen planus, lupus erythematodes, drug reaction, graft versus host disease (GVHD), and sun-burn (50). Ovoid, PAS-positive eosinophilic apoptotic bodies of various shapes and sizes that have generally been referred to cytoid bodies or Civatte bodies are often observed close to the epidermis or sometimes in the epidermis of these disorders (51), suggesting that apoptotic changes of the epidermis do not always lead to amyloid fibril deposition, but the conversion from apoptotic keratinocytes to amyloid deposition occurs only in the restricted condition. The rate of amyloidogenesis of Gal-7 fragments (Ser 31 -Gln 67 , Arg 120 -Phe 136 , etc.) was dependent on the content of each Gal-7 fragment (see Fig. 3C), which is controlled by the degree of Gal-7 induction during epidermal apoptosis. We found that limited trypsin treatment (3 h) of Gal-7 was the best condition for the production of amyloidogenic peptides (See Fig. 7B), indicating that appropriate tryptic enzyme activity is crucial for amyloid fibril formation. Additionally, the amyloidogeneses of Ser 31 -Gln 67 and Phe 33 -Lys 65 fragments were inhibited by other amyloid constituents (actin and cytokeratin fragments) of PLCA, whereas the amyloidogenic potentials of Arg 120 -Phe 136 ,
Familial Primary Localized Cutaneous Amyloidosis-Most cases of PLCA are the non-familial, acquired type, but rare cases of familial PLCA (FPLCA) with autosomal-dominant inheritance have been reported (52). Recent genetic analysis has revealed missense mutations in the oncostatin M receptor-␤ (OSMR) or interleukin-31 receptor A (IL31RA) gene, which can form a heterodimeric receptor together through IL31 signaling. OSMR␤ or IL31RA signaling is related to the reduction of signal for keratinocyte apoptosis (53); therefore, genetic defects in the signaling will induce keratinocyte apoptosis. The mechanism by which these mutations are related to Gal-7 is unknown. The subcellular localization of Gal-7 is variable; it can be located in the intracellular or extracellular space and sometimes on the plasma membrane (22), depending on the cellular conditions. Gal-7 may be directly or indirectly involved in the OSMR␤/IL31RA signaling pathway during keratinocyte apoptosis (28), resulting in the abnormal metabolism of Gal-7 in FPLCA keratinocytes.
Finally, the basic mechanism for promoting amyloid fibril formation in PLCA may be summarized as follows (Fig. 8); 1) induction of keratinocyte apoptosis by UV irradiation and more, 2) induction of Gal-7 expression and activation of apoptosis-related proteases, 3) production of amyloidogenic peptides of Gal-7 by apoptosis-related proteases, including trypsinlike enzymes at neutral pH, 4) aggregation of amyloidogenic peptides in the acidic condition, 5) the aggregation is modulated by Gal-7 peptides, actin and cytokeratins, and 6) growth of amyloid fibrils by deposition of additional Gal-7 fragments.
This study is the first step for understanding the pathogenesis of PLCA. To confirm the amyloidogenic mechanism of Gal-7, further studies will be necessary to identify the species of Gal-7 peptides that accumulate in the lesional skin of PLCA. The roles of actin and cytokeratins in amyloid fibril formation of PLCA also should be clarified.