Chapter 2

The Agouti and Extension series of Alleles, Umbrous and Sable

Although there are a great number of loci which influence the synthesis of melanin in mice, ¹ there are only two loci which control the nature of the pigment formed. Thus the agouti and extension series of alleles determine the relative amount and distribution of yellow pigment (phaeomelanin) and black or brown pigment (eumelanin) in the hairs of the coat.

I. The Agouti Locus

The agouti locus (chromosome 2) is particularly interesting because it is a coat-color determinant which acts via the hair follicle. It also appears to be a complex locus. The number of alleles which have been described as being part of this complex—17 in all—and the number of reported mutations (see Chapter 12, Section IV), also exceeds that recorded for any other locus concerned with the synthesis of melanin. Some of these mutations are listed by Grüneberg ( 1952, 1966a), Wallace ( 1954, 1965), L. Russell and Maddux ( 1964), and Dickie ( 1969a).

A. Alleles

1. Lethal Yellow (A^y)

Lethal yellow ( A^y) represents the top dominant of the agouti series in the sense that it does not matter with which of the series it is heterozygous, the phenotype is either a rich yellow or orange color ( Plate 1-A) or, on some backgrounds, a "sooty yellow" or "sable" color (see Section III). This allele, which represents an old mutant of the mouse fancy, was described first by Cúenot ( 1905) who recognized the fact that it was lethal when homozygous (see also Castle and Little, 1910; Ibsen and Steigleder, 1917). However, while Cúenot believed that an A^y-bearing egg could not be fertilized by an A^y-bearing sperm, this is not the case. A^y/A^y embryos are formed but display characteristic abnormalities at the morula or blastocyst stage ( Kirkham, 1919; Robertson, 1942a) and die early on the sixth day of gestation after the trophectoderm of the blastocyst has come into contact with the uterine epithelium, i.e., before implantation is complete (see Eaton and M.M. Green, 1962, 1963; Eaton, 1968; Pederson, 1974; Calarco and Pederson, 1976). Although according to Eaton and M.M. Green ( 1963) death is due to a lag in the differentiation of the trophoblast giant cells, preventing the normal interaction between embryo and endometrium necessary for successful implantation, the primary lesion responsible for the failure of giant cells to differentiate from the trophectoderm is not known. ² Robertson ( 1942b) demonstrated that A^y/A^y embryos which develop in a nonyellow mother, a feat accomplished by mating a/a females bearing A^y/a ovarian isografts with A^y/a males, proceed to a somewhat more advanced stage of implantation, attaining about twice the number of cells, than A^y homozygotes which develop in yellow mothers. Indeed, there is some evidence that this maternal influence on the development of A^y homozygotes holds for A^y heterozygotes as well ( G. Wolff and Bartke, 1966).

The A^y allele is also of interest because it often is associated with stimulation of normal body growth ( Heston and Vlahakis, 1961a), obesity (Dickerson and Gowen, 1946, 1947; Grüneberg, 1952), sterility ( M.C. Green, 1966a), and a diabetes-like syndrome ( Hummel et al., 1972). Moreover, this allele has been associated with an increased susceptibility to both spontaneous ( Heston and Deringer, 1947) and induced ( Heston, 1942) pulmonary tumors, to spontaneous hepatomas in males and spontaneous mammary tumors in females ( Heston and Vlahakis, 1961a), to induced skin tumors ( Vlahakis and Heston, 1963), and to spontaneous reticular neoplasms ( Deringer, 1970). ³ The fact that all of these pleiotropic effects are unique to A^y, or to animals of yellow phenotype, i.e., these effects are not associated with any of the nonyellow agouti series alleles, is undoubtedly significant both in terms of the structure of the locus and its mode of action. Indeed, the observation that A^y differs in so many ways from the other members of the series led Sir Ronald Fisher to suggest that it may represent a deletion covering more than one locus ( Wallace, 1954).

2. Viable Yellow (A^vy)

Viable yellow ( A^vy) arose spontaneously in the C3H/HeJ strain ( Dickie, 1962a) and although it may produce a phenotype similar to A^y, it is not lethal. The most characteristic feature of the A^vy allele is the variation in phenotypes associated with it ( G. Wolff, 1971). Some homozygous viable yellow animals when weaned are a reasonably clear yellow, indistinguishable from A^y/—, but many become sooty with successive molts ( Dickie, 1962a); some (homozygotes and heterozygotes) have a peculiar mottled pattern ( Plate 1-B; see also Figure 7-6b ), first observed when they are 4 or 5 days old, which may vary from a small black patch on an otherwise yellow coat to a complete intermixture of small black and yellow patches ( Dickie, 1962a); still others display a marked visual resemblance to agouti mice even though, on close examination, disparities are apparent ( Galbraith and G. Wolff, 1974). ⁴ The frequency of this "pseudoagouti" phenotype seems to be strain dependent ( G. Wolff, 1971). Inasmuch as genetically identical littermates often display these variations in pigmentation, and since A^vy animals of agouti-like phenotype often produce yellow offspring with variable degrees of mottling ( Dickie, 1962a; G. Wolff and Pitot, 1973), it is apparent that a considerable amount of the variation in pigmentation displayed by these mice must be a consequence of nonhereditary factors ( G. Wolff and Pitot, 1973) (see note 9 and Chapter 7, Section VI). Nevertheless, taken together, the expression of the A^vy gene seems to strike a precarious balance between the A^y and A alleles ( Galbraith and G. Wolff, 1974).

Although not lethal A^vy is associated with many of the other effects of A^y. Moreover, these pleiotropic effects, which include an influence on normal and neoplastic growth, fat deposition, and hormonal and enzymatic levels ( Heston and Vlahakis, 1968; G. Wolff, 1965, 1970a, 1970b, 1970c; G. Wolff and Flack, 1971; G. Wolff and Richard, 1970; G. Wolff and Pitot, 1972a, 1972b, 1973), are particularly interesting because they appear to be correlated with the amount of yellow in the coat (G. Wolff, 1965, 1971). For example, homozygous and heterozygous A^vy mice which are predominantly yellow are more likely to become obese than agouti viable yellow animals ( G. Wolff, 1965; Dickie, 1969a), and A^vy mice which are of agouti phenotype are less likely than mottled yellow mice to develop hepatomas ( G. Wolff and Pitot, 1972b). Clearly, for some reason, A^vy mice of agouti phenotype are metabolically more similar to a/a mice than they are to mottled A^vy animals ( G. Wolff, 1965; G. Wolff and Pitot, 1973) (see note 23).

3. Intermediate Yellow ( A^iy)

Intermediate yellow ( A^iy) originated in the C3H/HeJ strain. The original mutant ( A^iy/A) was a very sooty yellow with a slightly lighter belly, which is the characteristic phenotype of most animals bearing this mutation ( Dickie, 1969a). Intermediate yellow mice are not as mottled with dark areas as some A^vy mice, but like A^vy they occasionally resemble the wild (agouti) phenotype. The ears of A^iy mice appear darker than those of lethal or viable yellow animals. Both homozygotes and heterozygotes, which look alike, become obese as adults. However, the maximum weight of these animals is lower than A^vy/— or A^y/— genotypes. As in the case for A^vy, A^iy animals of agouti phenotype do not become as obese as, or display the very mild hyperglycemia of, phenotypically yellow A^iy mice ( Dickie, 1969a). ⁵

4. Sienna Yellow (A^sy)

Sienna yellow (A^sy) occurred in the C57BL/6J strain. Although little information is available on this allele, A^sy heterozygotes appear to be a very dark or sooty yellow (displaying yellow hairs with black tips) whereas homozygous A^sy mice are phenotypically a clearer yellow. A^y/A^sy heterozygotes are viable and fertile ( Dickie, 1969b).

5. White (or Yellow) Bellied Agouti (A^w)

It should be noted that this allele has also been referred to as A^L (light-bellied agouti). Mice displaying this phenotype have a typical agouti dorsum, i.e., a unique pattern of pigmentation characterized by a subapical yellow band on an otherwise black (or brown) hair. This black-yellow-black pattern results from a rapid shift from deposition of eumelanin to deposition of phaeomelanin and back again to eumelanin in the hair shaft. This pattern has been described in detail by Werneke ( 1916), Dry ( 1928), Dunn ( 1936), Kaliss ( 1942), E. Russell ( 1949b), and Galbraith ( 1964). The belly of A^w mice is white, cream, or tan (yellow) as a consequence of the fact that, depending on the genetic background, the hairs originating on the ventrum are either frequently nonpigmented, possess yellow pigment, or are predominantly yellow with black bases ( Silvers, 1958b). A^w is dominant to A and all lower alleles.

6. Agouti (A)

This is the so-called wild type coloration ⁶ characterized by an agouti dorsum, ⁷ identical in all respects to the A^w dorsum described above, but a darker than A^w ventrum ( Plate 1-C). Although the hairs of the ventrum are predominantly banded, some have yellow tips and black bases.

7. Intermediate Agouti (Aⁱ)

Intermediate agouti ( Aⁱ) arose spontaneously in the C57BL/6J strain. The belly of the original deviant (proven to be Aⁱ/a) was dull yellow. There were some agouti hairs on the sides of the body but the subterminal yellow band was almost absent from the dorsal hairs, so that the back had a very dark appearance. As Aⁱ/a heterozygotes age, the dorsum lightens so that the animals resemble A^w/— mice ( Dickie, 1969a). Aⁱ/Aⁱ, Aⁱ/A, and Aⁱ/a^td genotypes are indistinguishable from light-bellied agouti ( Dickie, 1962b; M.C. Green, 1966a). Young Aⁱ/a^t animals, like Aⁱ/a, have a dark back and light belly with agouti hairs along the sides ( Dickie, 1969a).

8. Tanoid (a^td)

Tanoid ( a^td) also represents a mutation in the C57BL/6J strain. The deviant animal had a tan (yellow) belly, typical of that found on black-and-tan ( a^t) mice (see 9) and banded "agouti" hairs among the predominantly black hairs on the sides of the body. The dorsum was very dark but with the first molt the number of "agouti" hairs increased, spreading over the animal's back so that the animal resembled a dark modification of "agouti"( A or A^w) ( Loosi, 1963). Although precise information is lacking, tanoid animals must, at least at some stage of life, bear a striking similarity to young Aⁱ/a^t and Aⁱ/a mice. The tanoid allele is dominant to nonagouti (a) and black-and-tan ( a^t) and A^w/a^td and A/a^td heterozygotes are phenotypically indistinguishable from A^w/— animals.

9. Black-and-tan (a^t)

This phenotype ( Plate 1-D) was first described by Dunn ( 1928) who obtained a strain of black-and-tan mice from an English fancier, and demonstrated that an allele of yellow, agouti, and nonagouti was responsible for the pigment pattern. Black-and-tan mice have a black dorsum and yellow or cream belly. As in the case of A^w the color of the ventrum varies with the genetic background; in some stocks the small amount of pigment in ventral hairs is phaeomelanin, whereas in other stocks almost all of the ventral hairs are yellow with black bases ( Silvers and E. Russell, 1955). Some yellow pigment is also found in hairs originating on and behind the ears of a^t mice. This allele is particularly interesting because it is recessive to A on the dorsum, but dominant to A on the ventrum. Thus the A/a^t heterozygote is phenotypically indistinguishable from light-bellied agouti ( A^w/—).

10. Nonagouti (a)

Nonagouti ( a) represents another old mutant of the mouse fancy. This allele obtains its name from the fact that a/a mice display nonbanded (eumelanotic) hairs. a/a mice are black ( Plate 1-E) and a/a,b/b animals are brown ( Plate 1-F). Although a/a hairs are almost exclusively pigmented with eumelanin, the hairs originating on and behind the ears as well as the hairs around the genital papilla and mammae, are yellow (at least in part).

11. Extreme nonagouti (a^e)

Extreme nonagouti ( a^e) was found among the descendants of an irradiated mouse ( Hollander and Gowen, 1956). It is recessive to all the other alleles of the agouti series and is characterized as completely eumelanotic— i.e., no yellow hairs occur on or behind the ears or around the nipples and perineum.

12. Mottled Agouti (a^m)

Mottled agouti ( a^m) is an allele at the agouti locus which originated in radiation experiments at Oak Ridge (L. Russell, 1964, 1965). This allele is similar to A^vy in that it induces a mottled coat, many a^m/a animals possessing agouti (actually, as described above for A^vy, "pseudoagouti") ⁸ and nonagouti patches of fur freely intermingled at their edges (L. Russell, 1964, 1965). Moreover, as in the case of A^vy, a^m produces a continuum of phenotypic variability. Unlike A^vy, however, this variability encompasses phenotypes of the lower agouti series alleles. Thus a^m/a^m mice vary in color from agouti to completely black, mimicking in a few cases the phenotypes of a/— and a^e/a^e animals ( G. Wolff, 1971). The fact that A^vy is associated with phenotypes ranging from agouti up to yellow, while a^m produces phenotypes ranging from agouti down to extreme nonagouti, is especially interesting in that these two alleles can produce almost the entire spectrum of agouti locus phenotypes. This has been nicely emphasized by G. Wolff ( 1971) who has established a stock, the so-called VYm stock, in which both of these alleles are maintained (see Figure 2-1 and Chapter 7, Section VI). ⁹

In addition to the above alleles, three others have recently been recovered from radiation experiments at Harwell ( R.J.S. Phillips, 1976); these are agouti umbrous ( a^u), nonagouti with dark agouti belly ( a^da), and nonagouti lethal ( a^l).

13. Agouti Umbrous (a^u)

Agouti umbrous ( a^u) resembles A^s (see 17) in its effect but does not crossover with a. a^u homozygotes have a dark agouti dorsum and ventrum as well as dark agouti pinna hairs. a^u/a animals also display an umbrous back and dark agouti belly but their pinna hairs are yellow. a^u/a^t heterozygotes have an umbrous back, tan (yellow) belly, and yellow ear hairs ( R.J.S. Phillips, 1976) (see note 10).

14. Nonagouti with Dark Agouti Belly (a^da)

This allele when homozygous, or when heterozygous with a, produces either a nonagouti or extreme umbrous back, dark agouti belly, and yellow pinna hairs ( R.J.S. Phillips, 1976).

15. Nonagouti Lethal (a^l)

Nonagouti lethal ( a^l) is presumably a prenatal lethal when homozygous, but it differs from a^x (see 16) in that phenotypically a^l/a mice are indistinguishable from a/a genotypes. a^e/a^l mice resemble a^e/a^e animals, i.e., they have black pinna hairs. The interaction of this new lethal with A^y has not been reported ( R.J.S. Phillips, 1976) (see note 10).

The above 15 alleles have all been ascribed to the agouti locus and although some investigators believe that many of the phenotypes associated with these alleles result from the expression of more than one locus (see section B), there is no concrete evidence to prove this assertion. Nevertheless, there are two mutations which are considered members of the "agouti" series which have been shown to be very closely linked; these are a^x and A^s.

16. Lethal Nonagouti (a^x)

Lethal nonagouti ( a^x) is also a radiation-induced mutation and one which is lethal in utero when homozygous. a^x, which is recessive to A^y, A^w, A, and a^t but dominant to a, is characterized by a slightly paler belly than that found in a/a animals. This mutation shows about 0.5% recombination with A^y ( L. Russell et al., 1963). A^y/a^x mice are perfectly viable and preliminary studies indicate that the lethal effect in a^x homozygotes is different from that seen in A^y homozygotes ( Papaioannou and Mardon, 1978) (see note 10).

17. Agouti-Suppressor (A^s)

This determinant is of special interest not only because it displays 0.6% crossing over with A^w and a^t ( R.J.S. Phillips, 1966a), but also because it recently has been shown to have resulted from and inversion ( Evans and R.J.S. Phillips, 1978) and therefore almost certainly reflects some sort of position effect. ¹⁰ It occurred among the offspring of an X-irradiated (C3H x 101)F₁ male ( R.J.S. Phillips, 1961a) and was originally believed to be a recessive allele of the agouti locus and similar to Aⁱ (R.J.S. Phillips, 1959, 1960). When homozygous, A^s reduces the amount of phaeomelanin normally produced by whatever agouti alleles are present. As described by R.J.S. Phillips ( 1966a): "the yellow pinna hairs found with all alleles except a^e become black (or brown on a homozygous brown background etc.); with a^t/a^t the back and belly are black; with A^w/A^w the size of the normal yellow band and the percentage of banded hairs are reduced giving a very dark agouti effect shading to almost complete nonagouti on the top of the head (with A^w/a^t this "nonagouti" patch is rather larger, but the two genotypes are not positively distinguishable). The belly of the A^sA^w/A^sA^w animal is also dark agouti; the A^sa^t/A^sa^t animal is undistinguishable from a^e/a^e."

The fascinating aspect of the behavior of A^s is that it shows a cis-trans position effect ( E. Lewis, 1961), i.e., when heterozygous it affects only the agouti allele to which it is linked. Thus, for example, A^sa^t/+A^w is phenotypically indistinguishable from light-bellied agouti whereas A^sa^t/+ a^t animals have an umbrous back, characterized by a nonagouti mid-dorsum and dark agouti flanks, and a yellow belly; A^sa^t/+ a^e mice are phenotypically identical with extreme nonagouti, while A^sa^e/+ a^t animals look like normal black-and-tan mice (see Table 2-1). In other respects A^s behaves in a normal Mendelian fashion ( R.J.S. Phillips, 1966a). ¹¹

B. Structure of the Locus (Simple vs Complex)

As the above description of the phenotype(s) associated with each agouti locus allele testifies, the locus is a complicated one (see Table 2-2). In general, as one proceeds from top dominance ( A^y) to bottom recessive ( a^e) there is an increase in the amount of eumelanin, indicating that the synthesis of less black is dominant to the production of more. This tendency to produce yellow pigment appears also to be greater in ventral than in dorsal hair follicles and is even greater in the hairs of the pinna. However, there are some important exceptions to these generalizations and these have convinced some investigators that the agouti locus is composed of several very closely linked "mini-loci", each of which controls the proportion of yellow and black pigment for different parts of the body, e.g., dorsum, ventrum, ears, etc. ( Pincus, 1929; Keeler, 1931; Wallace, 1954, 1962, 1965). Indeed, according to Wallace ( 1965), the situation to be envisaged is similar to the pseudoallelic system proposed by Fisher to account for the Rh blood group antigens in man ( Race and Sanger, 1962).

Let us consider the evidence supporting this complex locus theory. Probably the best evidence is that inasmuch as a^x and A^s have already been demonstrated to be closely linked to A^y, and A^w and a^t, respectively, and there is some evidence that A^y is pseudoallelic to a just as it is to a^x ( L. Russell et al., 1963), there is precedence for an even more extensive series of closely linked determinants. Further support for a complex locus stems from the observation that as one proceeds down the agouti locus series, the change from black to yellow occurs in one place in the series for one part of the pelage and in another for a different area. Thus phaeomelanin disappears from the mid-dorsum at compound Aⁱ/a (but not Aⁱ/Aⁱ), from the lateral dorsum at a^t/a^t and its lower compounds, from the ventrum at a/a and its lower compounds, and from the pinna and genital ridge at a^e/a^e. Although these changes may merely reflect the fact that certain regions of the body, e.g. the ventrum and pinna, present a milieu more favorable for the synthesis of phaeomelanin, they are also compatible with the notion that several very closely linked loci are involved, one controlling the proportion of yellow-black pigment for one part of the body, and one for another region. ¹² For example, as pointed out by Wallace ( 1965), if one "mini-locus," D, controlled dorsal banding, and another, V, controlled ventral banding, with dominance of the yellow component in each case, the genotypes A/a^t and A^w/a, which have the same phenotype, light-bellied agouti, would become Dv/d V and DV/dv, respectively. Such an interpretation would explain the observation that the series A^w:A:a^t represents a grading yellow to black dorsally but not ventrally. ¹³

Proponents of the complex locus hypothesis argue that it is difficult to envisage on the simple locus theory how two alleles, A^vy and a^m (and perhaps A^iy should also be included), which are relatively high and low in the dominance hierarchy, produce mottling, whereas it does not appear in any of the intermediate compounds ( Wallace, 1965). According to the multiple locus theory, mottling is attributed to yet another closely linked locus—or possibly two further loci ( Wallace, 1965).

Mutations to light (yellow) belly are much more frequent than mutations to other agouti locus genotypes both in the laboratory ( Isherwood et al. 1960; Dickie, 1969a) ¹⁴ and in the wild ( Wallace, 1954) and the advocates of the pseudoallelism theory believe that this too is consonant with their position. They argue that it is more likely that a locus closely linked to the a locus has a much higher mutation rate than to suppose that some agouti locus alleles differ so drastically from the others in terms of their stability.

Supporters of the complex locus hypothesis believe also that the occurrence of certain phenotypes from known matings may represent crossovers rather than gene mutations. For example, Wallace ( 1954) reported that a nonagouti ( a/a) mouse was produced from an A/a^t female x a/a male mating. While it is conceivable that a mutation to a occurred from A or a^t, it is also possible that the unexpected phenotype resulted from a crossover. If yellow belly is designated as W, a gene closely linked to a, and nonyellow belly as w, its recessive allele, then Wallace's mating can be designated Aw/aW female x aw/aw male and a simple crossover could yield an aw ovum (see Keeler, 1931; Grüneberg, 1966a).

Finally, the advocates for the complex locus hypothesis believe that evolutionary considerations support their contention. They find it difficult to understand why if one locus is involved, two different genotypes, A/a^t and A^w/—, should have as a result of selection the same phenotype ( Wallace, 1954).

In spite of all the evidence supporting the complex locus theory, there are some observations which argue against it. On numerous occasions a/a matings have produced A^w/— offspring ( Little and Hummel, 1947; Bhat, 1949; Hoecker, 1950; Dickie, 1969a) (see note 13). While according to the simple locus concept this merely requires a single mutation, albeit one which "skips a step," i.e., a —> A^w (skipping a^t), the complex locus interpretation requires two simultaneous mutations, one from a —> A and the other from w —> W. Nevertheless, two such simultaneous mutations in a single animal have been documented ( Dickie, 1969a) ¹⁵ and one could conceive that if two loci are involved, they are so closely associated that a mutating agent affecting one would frequently affect the other ( Wallace, 1954). Still more difficult to reconcile with the two locus hypothesis is R.J.S. Phillips' ( 1966a) observation that A^s affects the belly of A^w mice differently than it does the ventrum of a^t animals; the belly of A^sA^w homozygotes is a dark agouti whereas the ventrum of A^sa^t homozygotes is nonagouti. According to the complex locus interpretation, this would indicate that the yellow belly of A^w and a^t mice are not equivalent and would necessitate a locus for banding and a locus for yellow belly rather than one for dorsal and one for ventral banding ( R.J.S. Phillips, 1966a).

Any hypothesis involving two loci would also be expected to produce some phenotypes with the ventrum darker than the dorsum or with plain hair on the dorsum and banded hair on the belly, and these have never been observed. Indeed, the relationship of back, belly, and ears is such that if the ears are dark, so is the belly and if the ventrum is dark, so is the dorsum ( R.J.S. Phillips, 1966a).

Finally it should be noted that the mutation to extreme nonagouti first occurred in an agouti stock ( Hollander and Gowen, 1956) which, according to the complex locus theory, would require at least three distinct mutations: one affecting the dorsum, one affecting the ventrum, and one affecting the ears! ¹⁶

Clearly, the structure of the agouti locus remains to be resolved but it probably will not be until the specific biochemical processes involved in producing the various agouti locus phenotypes are determined.

C. Eumelanin and Phaeomelanin: Structure, Ultrastructure, and Biosynthesis

Regardless of the chromosomal structure of the agouti locus it is apparent that it determines whether eumelanin, phaeomelanin, or both of these pigments are synthesized in the melanocytes of the hair bulb. Moreover, it should be emphasized that these two kinds of pigments are appreciably different from each other when examined under the electron ( Moyer, 1966; Sakurai et al., 1975) or conventional microscope (E. Russell, 1946, 1948, 1949a, 1949b). When viewed under the latter, yellow granules from either A^y/— or A/— hairs are uniformly round and of approximately the same intensity and size, whereas black granules are much more variable in their characteristics (see Figure 4-6) with three different grades of color, four recognizably different shapes, and a wide variation in size (see Table 2-3) ( E. Russell, 1949a). The total pigment volume in black hairs also is about five times higher than in yellow hairs ( E. Russell, 1948) and whereas eumelanin is insoluble in almost all solvents and resistant to chemical treatment, phaeomelanin is soluble in dilute alkali ( Jimbow et al., 1976; see also Ikejima and Takeuchi, 1978).

When examined under the electron microscope, yellow and black granules again display striking differences which suggest that the ontogeny of yellow granules is quite distinct from granules producing eumelanin ( Moyer, 1966; Sakurai et al., 1975). One characteristic feature of the ultrastructure of yellow granules is that they do not possess any organized matrix ( Figure 2-2) so that although pigment is laid down in discrete areas, it is deposited randomly on a tangled mat of exceedingly fine fibers which are almost translucent to the electron beam ( Moyer, 1966). According to Moyer ( 1966), "there is no ordered aggregation of fibers nor is there any organized cross-linking. As the phaeomelanin accumulates, the areas of deposition spread and fuse until finally the granule assumes a very dense homogeneous appearance. Very occasionally the areas of melanin deposition in the granules are found in linear arrays reminiscent of intermediate stages in the ontogeny of eumelanin granules [see Chapter 3; section I, E and F], but even then no organized matrix is visible."

Of particular relevance is the ultrastructural study by Sakurai and his associates ( 1975) on A/A melanocytes. These investigators observed that hair follicle melanocytes in transition from black to yellow possess both eumelanosomes and phaeomelanosomes ( Figure 2-3). This not only constitutes the best evidence to date that the shift from eumelanin to phaeomelanin synthesis, or vice versa, occurs within a single cell, but it indicates that one population of cells is responsible for all the patterns of pigmentation associated with the a locus (see also Galbraith, 1964; Geschwind et al., 1972).

In regard to the biosynthesis of these two melanins ( Figure 2-4), the prevailing contention has been that eumelanin is a protein conjugate formed by the coupling of a quinoid polymer, indole-5,6-quinone, with protein to form eumelanin granules. Tyrosine is the natural precursor of eumelanin, and a single, copper-containing, enzyme complex, tyrosinase, is involved in the first two steps of the conversion of this colorless amino acid to melanin: step one, the hydroxylation of tyrosine to 3,4-dihydroxyphenylalanine (dopa) and step two, the oxidation of dopa to dopaquinone. According to this scheme, eumelanin is a homopolymer of indole-5,6-quinone units, formed from intramolecular rearrangements of oxidative products of dopaquinone, linked through a single bond type ( Fitzpatrick and Lerner, 1954; Lerner, 1955; Fitzpatrick et al., 1958; Lerner and Case, 1959; Fitzpatrick and Kukita, 1959; Foster, 1965). However, there is also some evidence that eumelanin may actually be a heteropolymer, or a random polymer derived from the linkage of many different indoles, including 5,6-dihydroxyindole ( Nicolaus and Piatelli, 1965), or that it is a highly irregular, three-dimensional polymer composed of several types of monomers joined by different covalent bonds ( Blois et al., 1964; Jimbow et al., 1976).

Although at one time there was evidence to suggest that phaeomelanin might be derived from tryptophan, i.e., that an ortho-aminophenol derived from tryptophan might be oxidized by dopaquinone to produce a yellow pigment under the influence of a genetically controlled switch mechanism ( Fitzpatrick et al., 1958; Fitzpatrick and Kukita, 1959; Foster, 1965), yellow pigment, too, is almost certainly a product of tyrosine metabolism.

Support for the notion that tryptophan might be involved in phaeomelanin synthesis stemmed primarily from the efforts of Foster ( 1951) who found that oxygen uptake in pulverized yellow skin was stimulated considerably when it was incubated with tryptophan (see also Nachmias, 1959). Moreover, the production of a yellow pigment accompanied this reaction. While it was known that incubating yellow skin in tyrosine produced a black pigment, it was believed that yellow skin possessed an inhibitor of tyrosinase since when a/a skin was mixed with A^y/a skin the latter appeared to inhibit the tyrosinase activity normally present in the a/a tissue ( Foster, 1951). Nevertheless, all attempts to demonstrate directly the formation of yellow pigment from tryptophan have failed. Thus while Markert ( 1955) and Coleman ( 1962) were unable to show that ¹⁴C-labeled tryptophan was incorporated into melanin, these investigators found that ¹⁴C-labeled tyrosine was included. The study by Coleman ( 1962) is particularly pertinent since he employed mice of various genotypes. He observed that if labeled tryptophan was injected into yellow, black, or brown baby mice it was not incorporated into any of these pigments, whereas labeled tyrosine occurred in all of them, albeit at a somewhat reduced rate in yellow, as compared with black and brown skin. Coleman observed also that skin slices prepared from yellow animals likewise failed to incorporate [¹⁴C]tryptophan, but did incorporate [¹⁴C]tyrosine at about one-third the normal rate observed for nonagouti skin (see Table 3-1). ¹⁷ These results strongly imply that tryptophan plays no role in the synthesis of melanin and that tyrosine is the common precursor of both eumelanin and phaeomelanin. The fact that tyrosine is incorporated at a reduced rate into yellow pigment may indicate that the normal sequence of events which lead to the synthesis of eumelanin by hair bulb melanocytes is permanently (e.g., in the case of A^y) or periodically (e.g. in the case of A) interrupted, or somehow diverted, resulting in a smaller polymer with altered physical properties ( Wolfe and Coleman, 1966). Indeed, the little that is known about the synthesis of phaeomelanin suggests that it is formed by a modification of the eumelanin pathway, involving the interaction of cysteine with dopaquinone ( Jimbow et al., 1976). ¹⁸

D. Role of Sulfhydryls in the Expression of Agouti Locus Alleles

In 1963 Cleffmann reported that if pieces of skin from late embryos or from young mice of different agouti locus constitutions, including A^y/—, were grown in vitro, hair growth continued and eumelanin was produced. From this Cleffman concluded that all melanocytes synthesize eumelanin under standard in vitro conditions. However, if sulfhydryl (SH) compounds such as glutathione were added to the nutrient medium all melanocytes, regardless of their age or genotype (provided of course that they were not from albino donors), could be induced to produce phaeomelanin. The minimum concentration of SH compounds which were required to induce the in vitro synthesis of phaeomelanin varied with the agouti locus genotype of the explant. a/a and a^t/— pigment cells required a significantly higher concentration of SH compounds throughout their entire period in vitro than A^y/— melanocytes. On the other hand, A^w/— and A/— melanocytes displayed cyclic variations in their requirements of SH compounds. During that stage of the hair growth cycle when eumelanin was normally synthesized, a high level of SH compounds, equivalent to that required by a/a cells, was necessary to induce phaeomelanin synthesis, whereas during the period when the yellow band was being formed significantly fewer SH compounds were needed. Even more intriguing was the fact that Cleffmann claimed that the in vitro synthesis of phaeomelanin was not confined to the melanocytes of the hair bulb. The same in vitro conditions which promoted yellow pigment production in the hair follicle induced dermal melanocytes to synthesize phaeomelanin. This was surprising because this is not the case in vivo. Thus A^y/— mice produce phaeomelanin only in a hair bulb environment; in other regions of the animal where melanin is produced, e.g., the eye, ear, tail, harderian gland etc., only eumelanin occurs ( Markert and Silvers, 1956). Cleffmann also observed that pigment cells which are synthesizing phaeomelanin take up more SH compounds and less melanin precursors ([¹⁴C]tyrosine, [¹⁴C]dopa) than eumelanin-producing cells and that these SH compounds are incorporated to a greater extent into phaeomelanin than into eumelanin ( Cleffmann, 1964). Because these results helped account for how the agouti locus operates, i.e., it somehow controlled the amount of SH compounds in the melanocyte, a number of hypotheses on a-locus gene expression have been based upon these findings ( Foster, 1965; R.J.S. Phillips, 1966a; G. Wolff, 1963, 1971; Geschwind et al., 1972). Unfortunately, however, all of these hypotheses must be considered with caution since the original observations of Cleffmann await confirmation. Thus Knisely and his associates ( 1975) observed that neither A^y/a explants synthesized eumelanin under standard in vitro conditions, nor did A/A explants, in the eumelanin phase of pigment production, synthesize phaeomelanin when cultured in a concentration of glutathione which, according to Cleffmann ( 1963), should have induced them to do so. ¹⁹ Galbraith and Patrignani ( 1976) have likewise failed to corroborate Cleffmann's ( 1964) contention that yellow and black melanocytes differ according to their SH metabolism. They found that "yellow and black melanocytes, regardless of genotype, possess equivalent amounts of histochemically detectable sulfhydryl compounds."

E. Action of Genes at the Agouti Locus

The fact that in A^y/— mice phaeomelanin appears to occur only in hair follicles (see Silvers, 1957), ²⁰ as well as the observation that yellow and black granules occur in the same cell in A/A hair bulbs ( Galbraith, 1964; Geschwind et al., 1972; Sakurai et al., 1975), strongly suggest that the genes at the agouti locus act via the follicular environment. However, these findings do not constitute proof that this is the case. Proof requires evaluating the behavior of melanocytes of one agouti-locus genotype in hair follicles comprised of cells of a different agouti-locus constitution. Fortunately this situation can be achieved experimentally by transplanting histocompatible skin from near term or newborn donors to newborn recipients ( Silvers, 1963). This procedure takes advantage of the fact that when such grafts are made to neonatal recipients, some host melanoblasts migrate across the graft boundary and become established in the developing hair follicles of the graft ( Reed and Sander, 1937; Reed, 1938, Reed and Henderson, 1940). Thus employing this technique it was found that when intensely pigmented black mice of the genotype a/a;C/c^e (produced by crossing mice of an inbred color stock, A^y/a;c^e/c^e with C57BL/6, a/a;C/C animals) were grafted shortly after birth with neonatal skin from black-eyed white mice of the genotype A^y/a;c^e/c^e, some intensely colored yellow hairs appeared among the nonpigmented hairs within the border of the graft. Conversely, when neonatal a/a;c^e/c^e skin was transplanted to yellow-pigmented ( A^y/a;C/c^e) recipients, some intensely colored black hairs were found among the predominantly pale brown-pigmented hairs of the ( a/a;c^e/c^e) graft ( Silvers and E. Russell, 1955). Other experiments, utilizing the same technique, demonstrated that when agouti (but albino) skin was transplanted to either intensely pigmented yellow or black recipients, some hairs pigmented with the typical agouti pattern, i.e., black with a yellow band, arose among the albino hairs of the graft ( Silvers, 1958b).

While all of these studies were consistent with the hypothesis that it was the agouti locus constitution of the receiving hair follicle which determined the kind of melanin synthesized by the invading cell, nevertheless, because all the grafts, including those which were nonpigmented, possessed indigenous populations of melanocytes, one had to rule out the possibility that these indigenous populations were not being "turned on" by some chemical infiltrate of host origin. This possibility was eliminated by repeating some of these experiments with grafts from white spots. Inasmuch as white-spotting results from an absence of melanocytes (see Chapter 3, Section II, A), such grafts do not possess any indigenous population that can be "turned on." the results employing these grafts not only confirmed the previous findings ( Silvers and E. Russell, 1955; Silvers, 1958a, 1958b) but they indicated that the number of pigmented hairs within the border of "white spots" was usually greater than in transplants known to contain an indigenous melanoblast population. Evidently, in the latter case the melanoblasts of host origin have to compete with those of the graft for occupancy in hair bulbs.

Additional experiments with A^w/— and a^t/— transplants demonstrated that the expression of a-locus genic activity depended not only upon the genotype of the follicular environment, but also upon the location of this environment on the integument. Thus when potentially intensely pigmented melanocytes of any agouti locus constitution invaded a dorsal, nonpigmented, a^t/a^t graft they produced intensely pigmented black hairs, whereas when those cells were introduced into ventral follicles of the same genotype, some all-yellow hairs as well as some yellow hairs with black bases (characteristic of the ventral hairs of many intensely pigmented black-and-tan genotypes) resulted ( Silvers, 1958b). ²¹ These observations, together with the finding that a^t/— and A^w/— melanocytes were likewise able to respond completely to the agouti locus genotype of the receiving hair follicle, implied that in a^t/— and A^w/— mice ventrality and dorsality of location were not important per se but that, in conjunction with their genotype, they presented different follicular environments which influenced the expression of the melanocyte ( Silvers, 1961). The results of these transplantation experiments are summarized in Table 2-4. ²²

Once it was established that the genes at the agouti locus produce their effect by altering the follicular milieu the stage was set to locate more precisely the activity of these alleles. Inasmuch as hair follicles have both an epidermal and dermal component—the epidermal ectoderm forms the epithelial portion of the follicle and the mesoderm the dermal papilla—attention was focused on which of these components was responsible for the agouti pattern. Mayer and Fishbane ( 1972) and Poole ( 1974) treated small pieces of skin, derived from the side of the trunk of 13- to 17-day-old a/a and A^w/A^w (or A/A) embryos, with trypsin so that the epidermis and dermis could be separated and recombined and allowed to differentiate in the testis of a histocompatible host or in a chick embryo. They found that when a/a dermis was recombined with A^w/A^w (or A/A) epidermis the hairs which differentiated did not display the agouti pattern. On the other hand, the reciprocal combination of agouti dermis and nonagouti epidermis always resulted in typical agouti hairs. These findings indicated that the genotype of the epidermal (ectodermal) component had no influence on the kind of pigment synthesized and that a-locus activity was mediated via the dermis. This conclusion had to be revised, however, when it was observed that recombinations of A^y/a with +/+, a/a, or a^e/a^e dermis produced hairs completely pigmented with phaeomelanin (Poole, 1974, 1975). Indeed, yellow pigmented hairs were produced even in recombinations of "young" (13 day) embryonic yellow epidermis with "old" (17 day) a/a or a^e/a^e dermis: dermis in which dermal papillae were already present.

Further investigations demonstrated that only in A^y/— skin does both the epidermis and dermis induce the synthesis of phaeomelanin. ²³ Thus the regional pigmentation pattern displayed by black-and-tan mice, like the agouti pattern, is determined by the dermis regardless of whether it is of ventral or dorsal origin ( Poole and Silvers, 1976a). Experimental recombinations of 14- to 16-day-old embryonic a^t/a^t ventral dermis and a^t/a^t dorsal epidermis give rise to yellow-pigmented hairs, whereas the reciprocal combination of dorsal dermis and ventral epidermis develop hairs pigmented with eumelanin. Black hairs are likewise produced when a^t/a^t dorsal dermis is combined with a/a ventral dermis or when a/a ventral dermis is associated with a^t/a^t ventral epidermis. This latter finding is especially revealing because it demonstrates that the epidermis of a^t/a^t ventral skin, unlike that of A^y/a skin, does not promote the synthesis of phaeomelanin ( Poole and Silvers, 1976a). It appears, therefore, that although yellow and black-and-tan animals have similar ventral phenotypes, only the A^y allele can act via the dermal and epidermal components of the follicle to induce the formation of yellow hairs. A summary of the pigmentation of hairs obtained by recombining dermis and epidermis between a^t/a^t, a/a dorsal and ventral, and A^y/a ventral embryonic skin is presented in Table 2-5. Obviously the combination of both dermal ventrality and the a^t allele is necessary for the development of the black-and-tan ventral phenotype.

In spite of all the effort which has been directed toward investigating the action of genes at the agouti locus, the specific effect of this locus remains unknown. It was once proposed that the banding of agouti hairs was caused by the rate of hair growth, i.e., an elevated mitotic rate in the follicular matrix somehow promoted phaeomelanin synthesis (Cleffmann, 1953, 1960, 1963). While this possibility was consistent with the observation that the agouti pattern could be altered by the administration of the mitotic inhibitor colchicine during the period of yellow band formation ( Galbraith, 1964), as well as with the fact that there was an association between the initiation of yellow banding in A/A mice with an increased mitotic rate in their hair follicles, this notion has had to be discarded. Thus the very meticulous analysis by Galbraith ( 1971) has revealed clearly that a/a hair bulbs (as well as A^y/a bulbs) display an increased mitotic rate at the same phase of hair growth when the yellow band is forming in A/A animals. Indeed, genetically yellow and genetically black bulbs possess virtually identical mitotic rates at all stages of hair growth. Moreover, if there was a causal relationship between increased mitotic activity and yellow pigment production one would expect the cessation of phaeomelanin production in A/A hair bulbs to be associated with a decreased mitotic rate and this, too, does not occur. In fact, the rate of cell division during the return to eumelanin synthesis is, in most instances, higher than during band formation ( Galbraith, 1971). ²⁴

Galbraith and Arceci ( 1974) have also shown that A^y/a and a/a hair bulbs do not differ with respect to their melanocyte populations and so it is most unlikely that a fluctuation in the number of pigment cells during hair growth is responsible for the agouti pattern. ²⁵ Actually the only factor which has been shown to be related to the synthesis of phaeomelanin in agouti mice is bulb mass. As A/A hair bulbs increase in size there is a loss in the capacity of their melanocyte populations to synthesize yellow pigment; no phaeomelanin is produced by A/A melanocytes in hair bulbs that exceed 110 microns in diameter ( Galbraith, 1969). Thus neither guard hairs nor the vibrissae show the agouti pattern.

While, as the above testifies, an abundant amount of information has been collected on the action of the a-locus genes, the crucial experiments defining the primary action of this locus have yet to be reported.