Tuft protein: protein cross-linking in enamel development

During development, dental enamel, unlike most miner-alized tissues, loses almost its entire extracellular proteinmatrix (1). This is degraded, removed, and replaced, firstby fluid, and subsequently almost entirely by hydroxy-apatite crystals (2). The loss is selective, however, suchthat the very small amount of residual material does notresemble the original bulk matrix. What remains of theoriginal matrix in mature enamel comprises fragments,often small peptides and amino acids (3–5).However, one protein material found in matureenamel does not fall into this category. Complete disso-lution of enamel, particularly human molars, revealed aconsiderable amount of highly insoluble material on thedentine surface (6). This material was not soluble indilute mineral acids with or without chaotropic agents(8 M urea or guanidine) or in EDTA, with or withoutchaotropic agents. Treatment with SDS at room tem-4 perature and after boiling and phenol also failed to dis-5 solve tuft protein. Most of these reagents were used in6 combination at both low and high ionic strengths, also7 without success. Analysis showed this material to beproteinaceous in nature, being composed of more than80% amino acids (7, 8). Microscopic examination of thismaterial showed it to be associated with enamel tufts,histological features that appear as blades of organicmaterial sinuously between, and contiguous with, enamelprism boundaries (9, 10). For this reason the materialwas assigned the name tuft protein. This is not to beconfused with the similarly named tuftelin reported byDeutsch et al. (11).Microscopic examination revealed that the insolubletuft protein extended beyond the histological tuftsthemselves as thick ribbons of material containing apattern that appeared to outline enamel prisms (12). Tuftprotein has been located in all tooth types but can beseen most obviously in human molars, particularly wherethe prism structure is very complex. Similar structuresfor tuft protein were also reported in considerable detailby Amizuka et al. (13).The specific composition of tuft protein remainsenigmatic. The amino acid composition is fairly consis-tent (8) but is not unequivocally related to any othercomponents of the enamel matrix. Tuft protein wouldappear to be a mixture, as the presence of keratin andtuftelin has been indicated by immunochemistry (14). Afragment of ameloblastin sequence has been detected(15), which was subsequently supported by positiveimmunoreactivity to a 13–17 kDa so-called sheathlin 8protein corresponding to ameloblastin or amelin (13).This antibody did not react with all histological com-ponents of tuft, reinforcing the view that tuft is a mixtureof protein components. 9In an effort to clarify the origin of tuft protein, anti-bodies were raised to a suspension of solid tuft protein.When applied to sections of developing rat incisor,antibody reactivity appeared to follow a secretorypathway from the ameloblasts (12). On present evidencethis is likely to be ameloblastin (13, 15). As ameloblastin-degradation products have been immunologically relatedto prism boundaries (16) these data are consistent per-haps with a role relating to prism formation.The presence of other components, such as keratins, isintriguing and may relate to fragments of ameloblastsretained by the tissue (17).The reason for the location of tuft protein primarily atthe dentino–enamel junction is not clear. This region is,