648 JOURNAL OF COSMETIC SCIENCE hairs. In the straight fiber, the cell types were arranged annularly and evenly within cortex, implying that the averaging of differing structural features would maintain a straight fiber conformation. In the curly fiber, the cell types were bilaterally distributed approximately perpendicular to fiber curvature direction. Previously, this pattern had only been observed in human scalp hairs with extremely high curliness (65). Among low-diameter mammalian hairs, fiber curvature is generally correlated to differences in cortical cell-type distribution across the cortex, with straight hair shafts containing an annular cell-type distribution and curly shafts a more bilateral distribution (46). WITHIN THE FOLLICLE Whatever the cause of curvature is at the level of the hair shaft, that cause originated in the hair follicle and resulted from a program of events that started within the living cells at the base of the anagen follicle bulb (24,66,67). Unlike correlations with curliness in the hair shaft, many of the correlations observed at the level of follicle structure have led directly to hypotheses of causation. It has been proposed that the shape of the hair follicles largely determines the shape of the fibers. Curly hairs from humans (regardless of ethnicity) and sheep typically emerge from curved follicle bulbs and straight fibers from relatively straight follicle bulbs (24). This has led to a common perception that hairs are extruded and take on the form of the follicle that produced them. In the middle of the 20th century, there was an awareness that follicles were in a dynamic state, mobilized by their arrector pili muscle (68). This led Chapman (69) to propose that muscle-induced rotation of the follicle bulb might cause hair shaft curl. However, the extrusion hypothesis is largely discounted based on more recent microscopy studies of follicles. Not only do follicles have a retrograde curvature (the bulb curves the opposite direction to the final fiber), but straight fibers have been observed growing from curved follicles in cultured bovine follicles, indicating that the correlation between follicle morphology and fiber curvature may be noncausative but associative (70). Rather than a cause, follicle bulb curvature may be a consequence of fiber development, such as different rates of protein expression across the cortex (71), cell reshaping and migration (72), or simply redistribution of stresses in a high-pressure biological environment (73). Other features that correlate with, or are associated with, follicles that generate curly hairs are differences in cortical keratin expression patterns (74) and in ultrastructural assembly of macrofibrils (75). Thibaut et al. showed using in vitro growth that curly hair is programmed in the basal area of the follicle, so asymmetry of keratinization was believed to result in a bilaterality of cell-type formation. Also associated with curly-hair–producing follicles is that the outer root sheath and the connective tissue sheath lack symmetry along the follicle (24). This is also known for wool, where the thicker outer root sheath associated with the concave curve of the follicle bulb is called the ental side and the other the ectal side of the follicle (68). Asymmetry of follicle morphology associated with hair curvature may have led Hynd et al. (76) to propose a model in which asymmetrical cell production on either side of the dermal papilla combined with a staggered initiation of keratinization and hardening on different sides of the hair shaft caused curvature. Work by Harland et al. (13) has challenged the role of asymmetric cell division as a cause of curvature in high-curl wool fibers, but the concept of staggered hardening remains intriguing. Further investigation tracking cells during

649 WHAT CAUSES CURLY HAIR? their transit through the follicle will likely be required to identify additional potential contributions of cell division, migration, and reshaping to curvature. PROTEOMIC AND PROTEIN EXPRESSION Our ability to both identify and relatively quantify individual structural proteins and their chemical modifications and associations (e.g., chemical cross-links) between hair samples is rapidly developing. The current state of knowledge from sheep model studies is that the abundance of some structural proteins differs between straight and curly wool fibers from closely related individual animals (fraternal twins). Quantitative mass spectrometric analysis demonstrated that straight wool and curly wool differed in the relative abundance of individual species from high glycine-tyrosine protein families and ultra-high sulfur protein families (46). Similarly, protein abundance differences have been observed between human scalp hair samples chosen for being very straight versus very curly (77). Proteomic analysis was able to pick out individual keratin and keratin-associated proteins that differed between the sample groups (keeping in mind that samples likely differed in other ways than just curliness). Quantitative mass spectrometry is a rapidly advancing technology, and these early results suggest it will be a powerful tool to identify the proteins of interest for understanding phenomena such as hair curliness. The main current limitation of all mass spectrometry techniques is that optimal results require multiple hairs and a lack of data on protein associations or locations, although both of these are areas of current research (78,79). Figure 2. Summary of a generic mammalian anagen phase follicle growing a curly hair, with key structures and features associated with curvature listed.