Implications
Lactation has an evolutionary origin and biological context within which its secretory processes and milk constituents evolved. This review provides an evolutionary scenario within which the specific functions of milk constituents can be investigated. It is hoped that this review will stimulate increased attention to the evolution of lactation and milk constituents in a large variety of disciplines, including molecular biology, evo-devo, comparative genetics, protein structural studies, dairy chemistry, paleobiology, phylogenetic systematics, mammary gland biology, reproductive evolution, and comparative nutrition. The goal is an in-depth understanding of how the remarkable process of lactation came into being.
Introduction: the importance of milk to mammals
Milk secretion is a characteristic feature of all mammalian species, large or small, social or solitary, arctic or tropical. The famous taxonomist Carolus Linnaeus selected the name Mammalia in 1758 in uniting terrestrial ‘quadrupeds’ with dolphins and whales to reflect the fact that females of both groups bear mammae, or mammary glands (Gregory, Reference Gregory1910). Lactation is unique to mammals. No other extant organism produces copious glandular skin secretions to feed its young, although the larvae of some fish ingest modest amounts of nutritive mucus from surface mucous glands (Buckley et al., Reference Buckley, Maunder, Foey, Pearce, Val and Sloman2010); terrestrial-breeding frogs provide water, antimicrobial compounds, and perhaps some nutrients to their eggs via granular and/or mucous gland secretions (Taigen et al., Reference Taigen, Pough and Stewart1984; Oftedal, Reference Oftedal2002a); and the young of some live-bearing caecilians feed on sloughed skin, skin secretions, or perhaps both (Kupfer et al., Reference Kupfer, Müller, Antoniazzi, Jared, Greven, Nussbaum and Wilkinson2006). The ‘crop milk’ of pigeons, flamingoes and emperor penguins is a lipid-rich material produced by holocrine secretion by epithelial cells of the esophagus or crop (Horseman and Buntin, Reference Horseman and Buntin1995), but it does not match milk in complexity, magnitude or duration of secretion. Lactation can persist for many years in humans, great apes, toothed whales, and elephants (West et al., Reference West, Oftedal, Carpenter, Krames, Campbell and Sweeney2007), but may be as little as 3–4 days in the ice-breeding hooded seal (Oftedal et al., Reference Oftedal, Bowen and Boness1993a).
Lactation is highly complex and apparently of ancient evolutionary origin (Oftedal, Reference Oftedal2002b; Lefevre et al., Reference Lefevre, Sharp and Nicholas2010). The detailed signaling crosstalk among epithelial and underlying mesenchyme cells that is required for the differentiation, ductal branching, and proliferation of mammary tissue is only now being unraveled (Watson and Khaled, Reference Watson and Khaled2008), and the functional significance and patterns of expression of thousands of mammary genes are under investigation (Lemay et al., Reference Lemay, Lynn, Martin, Neville, Casey, Rincon, Kriventseva, Barris, Hinrichs, Molenaar, Pollard, Maqbool, Singh, Murney, Zdobnov, Tellam, Medrano, German and Rijnkels2009). Secreted milk is extremely varied in composition – from trace levels of fat in rhinos to more than 60% fat in some ice-breeding seals (Oftedal and Iverson, Reference Oftedal and Iverson1995) – and contains unique proteins (α-, β-, and κ-caseins, β-lactoglobulin, α-lactalbumin, whey acidic protein (WAP)), membrane-enclosed lipid droplets, and sugars (lactose, milk oligosaccharides) that are not found elsewhere in nature.
A dependency on milk is key to the life history strategy of every mammal. Many mammals specialize on diets that would be too difficult for small offspring to capture or to digest, or that might fail to cover the nutrient needs of rapidly growing young if milk was not available (Pond, Reference Pond1977). The ability to provide offspring with a highly digestible, nutritionally balanced, and variably concentrated food has enabled mammals to evolve a wide range of developmental and reproductive strategies. Lactation provides the essential nutrients required by the smallest and most altricial of neonates, such as monotreme hatchlings and marsupial neonates (Griffiths, Reference Griffiths1978; Tyndale-Biscoe and Renfree, Reference Tyndale-Biscoe and Renfree1987), as well as the largest and most precocial of offspring, such as hooded seals that only nurse for 4 days, but in this time double their mass (Oftedal et al., Reference Oftedal, Bowen and Boness1993a). Milk production can be miniscule, as it must have been to feed newborn Tasmanian tigers (or thylacines) which, by extrapolation from living dasyuromorph marsupials, weighed less than 40 mg at birth, equivalent to 0.00015% of the mass of the 25 kg mother, surely the smallest birth mass : maternal mass ratio of any mammal (O. Oftedal, unpublished results). However, milk production can also be immense, as in the 100 000 kg blue whale. The pregnant female feeds in polar and cold temperate waters where food is abundant and where she lays down massive stores of blubber. She then travels part way around the globe to warm or tropical areas, gives birth to a calf, and while fasting or feeding a little, produces an estimated 220 kg milk per day, containing approximately 4000 MJ energy (Oftedal, Reference Oftedal1997). Over the course of the 6-month lactation, the blue whale cow transfers an estimated 700 000 MJ to her calf, enough energy to feed approximately 200 people for a year.
The ability to lactate allows mothers to collect nutrients and deposit them in their tissues at one time and location, but then provide these nutrients to their young as milk at a later time and/or another location, such as in a nest or burrow (rodents), cave (bats, bears) or snow den (polar bear) or on an isolated island (sea lion, fur seal) or on floating rafts of pack ice (various seals, walrus; Oftedal, Reference Oftedal2000). Some bears lactate in the winter den for 2 months before emerging to feed and drink, a feat made possible by mobilization of extensive body reserves, production of milk low in sugar (minimizing the glucose demand of the mammary glands), and recycling of water and perhaps nitrogen from cubs back to the mother when she ingests their excreta (Oftedal et al., Reference Oftedal, Alt, Widdowson and Jakubasz1993b). In bats, lactation must continue until pups have deposited enough protein and mineral that their muscles and bones have sufficient strength to withstand the pressure and torque associated with flight, or the pups cannot feed themselves; bat pups may individually reach 70% of maternal mass before they are weaned (Hood et al., Reference Hood, Oftedal and Kunz2011). In social carnivores, an entire pack may hunt prey that is brought back to the lactating female, who converts food constituents into mammary secretions. The lactating female is in essence a ‘milk factory’, rearing very large litters of rapidly growing young (Oftedal and Gittleman, Reference Oftedal and Gittleman1989).
How did such a remarkable fluid as milk and such remarkable biochemical factories as mammary glands evolve? The mammary gland and its secretion represent major evolutionary novelties, without any known intermediates. In the mid-19th century the complexity and interdependence among mammary glands, milk and dependent suckling young posed a challenge to Charles Darwin's theory of evolution by natural selection. Darwin rose to the challenge, devoting most of a chapter of the 1872 edition of On the Origin of Species to a discussion of the problems of evolutionary novelty, such as the origin of the eye and of the mammary gland. Believing that seahorse eggs are hatched and reared in a brood pouch, Darwin (Reference Darwin1872: pp. 295–296) suggested by analogy that if mammalian young were protected in a pouch containing cutaneous glands, a gradual increase in the nutritive quality of the fluid and of the secretory complexity of the glands would be favored by natural selection, leading ultimately to breasts producing milk. Over the years since, a variety of authors have speculated on the origin and evolution of lactation, trying to envision a process in which something so complex could have evolved step by step and be favored by natural selection (reviewed by Oftedal, Reference Oftedal2002b). Unfortunately, most of these scenarios (including a recent suggestion that lactation originated from the innate immune system (Vorbach et al., Reference Vorbach, Capecchi and Penninger2006; McClellan et al., Reference McClellan, Miller and Hartmann2008)) have been put forth without reference to an evolutionary timescale and do not acknowledge the repeated radiations of increasingly mammal-like taxa since the first appearance of synapsids approximately 310 million years ago (mya).
Milk secretion in the context of synapsid evolution
The phylogenetic branch that would ultimately lead to mammals (Synapsida) first diverged from the branch leading to ‘reptiles’ and birds (Sauropsida) in the mid-Pennsylvanian period, approximately 310 mya. After this separation, there were a series of sequential extinctions, with only a limited number of taxa surviving into the succeeding geologic periods as the basis for future radiations (Sidor and Hopson, Reference Sidor and Hopson1998; Oftedal, Reference Oftedal2002b; Kemp, Reference Kemp2005). Thus, the basal synapsids radiated in the Pennsylvanian period and the early Permian period; however, most lineages went extinct by the mid–late Permian period, when they were succeeded by a radiation of therapsids. Most therapsid taxa disappeared during a massive extinction event at the end of the Permian period, but cynodonts survived to radiate in the Triassic period. Most of these lineages in turn disappeared in the late Triassic period, but a subset – the mammaliaforms – radiated in the late Triassic and Jurassic periods. It was from within the mammaliaforms that true mammals evolved, perhaps in the late Jurassic period, approximately 160 mya.
These sequential radiations incorporated an increasing number of anatomic traits that now characterize mammals, that is, they became progressively more mammal-like (Sidor and Hopson, Reference Sidor and Hopson1998; Kemp, Reference Kemp2005). Some of these traits involved changes in locomotion (from a sprawled lizard-like gait to an upright stance with improved running ability), in growth pattern (from a periodic pattern of bone mineralization to more continuous growth), in respiratory ability (development of diaphragmatic breathing), in heat exchange (using respiratory surfaces in the nasal cavity for cooling and moisture retention) and in food processing ability (rearrangement of skull and jaw bones for increased jaw musculature, diversification and specialization of teeth and tooth cusps). The picture is one of increasing metabolic expenditure, increased growth rates, increased activity, and a presumed upregulation of basal metabolism and increased refinement of temperature regulation associated with endothermy.
Reproductive patterns also changed. The early synapsids were egg-laying, like sauropsids (the lineage leading to ‘reptiles’ and birds) (Oftedal, Reference Oftedal2002a). It was the development of a complex of extraembryonic membranes to facilitate respiration, use of stored yolk nutrients and segregation of waste products within the egg that allowed the amniotes (including sauropsids and synapsids) to venture onto land and to gain a measure of independence from free water (Packard and Seymour, Reference Packard and Seymour1997). However, unlike sauropsids, the synapsids never evolved a calcified eggshell, or at least no fossil record of such has ever been discovered (in contrast to the plethora of fossilized eggshells from dinosaurs, crocodilians, turtles, birds and other sauropsids; Oftedal, Reference Oftedal2002a). An egg with a parchment-like shell is the ancestral condition of both groups; however, most sauropsids went down a different evolutionary path, developing elaborate eggshells of diverse crystalline form. These calcified eggshells are very resistant to vapor loss, having only small pores for gas exchange. By contrast, the eggs with parchment-like shells both gain and lose water rapidly. Eggs with parchment-like shells are ectohydric, relying on environmental water for completion of egg development, whereas eggs with calcified shells are by necessity endohydric, relying on water invested in the egg before calcification and laying. Eggs with parchment-like shells dry out rapidly when exposed to a vapor pressure gradient, as in air that is not fully saturated with moisture, or when exposed to a positive temperature gradient from egg to environment (Oftedal, Reference Oftedal2002a).
Early synapsids must either have buried their eggs in moisture-laden soil, as do living lizards and snakes that produce eggs with parchment-like shells, kept them hydrated by contact with moist skin (as terrestrial-breeding frogs and salamanders often do) and/or retained them within a moist pouch, such as the egg-laying echidnas in Australasia. Yet during the long period from the Pennsylvanian period to the Jurassic period, the therapsids, cynodonts and mammaliaforms are believed to have increased metabolic rates and elevated body temperatures, and eggs were presumably incubated to increase thermal stability (Oftedal, Reference Oftedal2002a). A buried egg has access to soil moisture, but is at soil temperature, whereas an incubated egg in a nest is both isolated from soil moisture and – due to thermal gradients that create vapor pressure gradients – subject to rapid desiccation. Pythons that incubate eggs with parchment-like shells above ambient temperatures wrap tight body coils about the eggs to restrict moisture loss (Lourdais et al., Reference Lourdais, Hoffman and Denardo2007). Some lizards and snakes overcome this constraint by internal egg-retention, including development of placental structures to provide nutrients to the developing embryos (Thompson et al., Reference Thompson, Biazik, Lui, Adams and Murphy2006). However, early mammals were still egg-laying (as indicated by egg incubation in extant monotremes and the persistence of egg teeth in some marsupial neonates; Oftedal, Reference Oftedal2002a and Reference Oftedal2002b).
It is not clear why the synapsids lineage did not evolve egg-retention as a reproductive strategy, as this occurred independently many times among sauropsids (Blackburn, Reference Blackburn2006). Perhaps some radiations did, but went extinct without leaving descendants. The synapsids that ultimately survived to become mammaliaforms and mammals opted for a different course: reduction in egg size and a switch to lactation as the primary nutrient source for offspring.
Yet it seems impossible that the increasingly endothermic synapsids could have incubated their eggs without facing lethal egg dehydration unless there was an exogenous source of moisture. I have argued that this required the evolution of lactation, initially as a source of moisture for eggs (Oftedal, Reference Oftedal2002a). Early synapsids apparently had glandular skin, and like extant frogs and salamanders that rear terrestrial eggs, they may have kept eggs moist via skin secretions (see ‘The evolution of mammary secretion’ section). As eggs and hatchlings were in contact with skin secretions rich in antimicrobial and other protective compounds (see Whey Proteins section, below), the evolution of nutrient-rich constituents in these secretions could occur gradually and over a long time period. Thus, milk was born.
It is not known whether eggs and hatchlings were kept in a pouch, as Darwin had hypothesized, but this would have been an effective means of reducing moisture loss by eggs and hatchlings. Many paleontologists have considered epipubic bones as evidence that mammaliaforms and early mammals (including early eutherians) carried eggs or young in a pouch or suspended from the abdomen; however, others argue that epipubic bones rather had a function in locomotion as the structure of the pelvis evolved (Novacek et al., Reference Novacek, Rougier, Wible, McKenna, Dashzeveg and Horovitz1997; Reilly and White, Reference Reilly and White2003). Pouches have evolved in both monotremes (e.g. the echidnas) and marsupials (e.g. kangaroos and wallabies), but these may not be of ancestral origin; pouches appear to have been gained and lost multiple times in marsupials (Tyndale-Biscoe and Renfree, Reference Tyndale-Biscoe and Renfree1987).
During the Triassic and the early Jurassic periods, the cynodonts and mammaliaforms were not only developing elevated metabolic rates but also were becoming progressively smaller in size (Kemp, Reference Kemp2005). This miniaturization of body size would have required miniaturization of eggs as well. Even if eggs were kept in a pouch to prevent desiccation, once they hatched the young would be too small to be effective homeotherms (Hopson, Reference Hopson1973). In birds, reduction of egg size in small species is accompanied by the reduction of incubation time and hatching of altricial (incompletely developed) young (Starck and Ricklefs, Reference Starck and Ricklefs1998). Hopson (Reference Hopson1973) argued that the diminutive mammaliaforms (some no more than a few grams as adults; Luo et al., Reference Luo, Crompton and Sun2001) must have been producing altricial young, as small eggs could not hold enough yolk to allow development of precocial (well developed) hatchlings. However, altricial young would require feeding, indicating that lactation had already evolved.
This conclusion is bolstered by the fact that late cynodonts and the mammaliaforms developed a reduction in tooth replacement (Kemp, Reference Kemp2005). Early synapsids, like most sauropsids, had teeth that were replaced continuously as the animal grew, allowing smaller teeth to be replaced by larger teeth as the jaw lengthened with age. Most mammals, by contrast, have only two sets of teeth, an initial set of deciduous ‘milk teeth’ and the adult dentition. This developmental strategy, termed diphyodonty, is possible because tooth eruption is delayed as the jaw develops in utero and during the lactation period. There is no need for a robust, adult-type teeth in dependent offspring that do not need to capture or consume an adult-type diet. The fact that late cynodonts and early mammaliaforms already had evolved diphyodonty indicates that they were already reliant on milk (Oftedal, Reference Oftedal2002b).
I conclude that the sequential radiations of basal synapsids, therapsids, cynodonts and mammaliaforms from the Pennsylvanian period through the Jurassic period (i.e. from approximately 310 to 160 mya) was a time of increasing reliance on glandular skin secretions, to keep skin moist and pliable, to provide water to eggs, and finally as a nutrient source for offspring.
The evolution of mammary secretion
The mammary gland is a complex alveolar–ductal organ involving a layer of secretory epithelial cells (lactocytes) that secrete milk into the alveolus by a variety of mechanisms (including exocytosis of secretory vesicles and budding out and pinching off of milk fat globules, MFGs), a surrounding layer of myoepithelial cells that expel milk from the alveolus via contraction, and a basement membrane that surrounds the epithelial compartment. The many alveoli are connected to the skin surface by a much-branched ductal network that terminates in canals, or galactophores, that penetrate either thickened epithelial structures (the nipples) in eutherians and marsupials or a specialized mammary patch or areola in monotremes (Oftedal, Reference Oftedal2002b).
The entire structure initiates development early in embryonic life (before hatching from the egg in monotremes) via coordinated reciprocal signaling between the epithelial cells of ectodermal origin and the underlying mesenchyme of mesodermal origin. This signaling directs gene expression, morphogenesis of tissue architecture, and differentiation of tissue-specific functions (Robinson, Reference Robinson2004; Watson and Khaled, Reference Watson and Khaled2008). For example, signaling compounds generated by the mesenchyme/stroma that have receptors in the epithelial cells and that help determine ductal and alveolar development include hepatocyte growth factor (HGF), IGF-1, activin/inhibin B, epimorphin, neuroregulin and keratinocyte growth factor (also called fibroblast growth factor-7; Nelson and Bissell, Reference Nelson and Bissell2006). This type of branching morphogenesis driven by epithelial–mesenchymal interactions and involving coordinated development with stimulatory signaling in part from HGF and epidermal growth factor, balanced by inhibitory signaling from members of the transforming growth factor-β family, is also found in tissues of even more ancient evolutionary origin, such as the pancreas, lung, kidney, prostate and salivary glands (Nelson and Bissell, Reference Nelson and Bissell2006). Thus, the developmental pathways of the mammary gland must derive from some pre-existing tissue, presumably glandular tissue associated with the skin, which was co-opted for a new function, the secretion of a nutritive fluid for feeding of the young. Although the basic pattern of mammary development, and its regulation, may derive from a more ancient model, the extent of glandular proliferation and output – the remarkable repeated cycles of proliferation and secretion followed by cellular apoptosis and gland involution – and the types of secretory products formed represent evolutionary novelties.
The synapsid lineage that led to mammals derived from pre-amniotic tetrapods, a group now represented by living amphibians (termed lissamphibia by paleontologists). Amphibian skin is characterized by a relatively thin epidermis containing a few cell layers of stratum corneum and dense coverage of small multicellular secretory glands that have evolved to secrete primarily mucus (mucous glands) or bioactive constituents (granular glands) onto the skin surface (Clarke, Reference Clarke1997). Both gland types apparently secrete via exocytosis of secretory vesicles; however, secretory cells in granular glands also swell with contents and release bulk material into the lumina of the glands via a holocrine process. This material may then be discharged onto the skin surface by contraction of myoepithelial cells when the animal is agitated or stressed. Synapsids apparently inherited a glandular skin from the tetrapods. A remarkable early Permian fossil of the integument of the therapsid Estemmosuchus (Dinocephalidae) includes a dense pattern of concave lens-like structures; Chudinov (Reference Chudinov1968) interpreted these as multicellular, flask-shaped alveolar glands – similar to the glands of amphibian skin – and argued that a glandular skin is a primitive synapsid feature still evident in mammals. It should also be recognized that the complex combinations of α- and β-keratins, multi-layered scales and paucity of skin glands in living reptiles evolved after the separation of sauropsids and synapsids and is not ancestral to the mammalian integument (Dhouailly, Reference Dhouailly2009).
However, could such glands evolve into mammary glands? On the basis of evidence that elements of the innate immune system are incorporated into milk constituents and that certain signaling pathways of the innate immune system also have a role in regulating mammary development, Vorbach et al. (Reference Vorbach, Capecchi and Penninger2006) speculated that mammary secretion first developed as part of an inflammatory response by mucous secreting cells, and McClellan et al. (Reference McClellan, Miller and Hartmann2008) supported this view. The innate immune system certainly has an ancient origin, including the antimicrobial compounds associated with epithelial structures and secretions (Beutler, Reference Beutler2004). In frogs and other amphibians, mucous glands are important to keep the skin surface moist, facilitating exchange of respiratory gases across the very thin epidermis (Lillywhite, Reference Lillywhite2006), and amphibian cutaneous glands are indeed veritable factories of antimicrobial constituents. At least 500 antimicrobial peptides have been found in amphibian skin glands to date (Jenssen et al., Reference Jenssen, Hamill and Hancock2006). However, mucous and granular glands are very different in structure and secretion from mammary glands. In mammals, mucous glands are restricted to oral and internal epithelial surfaces. Although Vorbach et al. (Reference Vorbach, Capecchi and Penninger2006) are certainly correct that mammary glands must derive, ultimately, from the simple glandular skin structures found in pre-amniotic tetrapods (Quagliata et al., Reference Quagliata, Malentacchi, Delfino, Brunasso and Delfino2006), there must have been intermediate stages that more closely resemble mammary glands.
If one compares mammary glands to other mammalian skin glands, they bear close resemblance to only one type of gland, the apocrine glands on the general skin surface of mammals (Oftedal, Reference Oftedal2002b). Apocrine glands and mammary glands secrete constituents both by exocytosis of secretory vesicles and by a budding out and pinching off of cellular contents with loss of cytoplasm. In apocrine glands, the latter process is considered an apocrine mode of secretion, in contrast to merocrine secretion employing exocytosis of vesicles or holocrine secretion in which cells swell with the secretory product that is released via apoptotic disruption of cellular integrity. In mammary glands, budding out and pinching off occur during secretion of MFG; cytoplasmic crescents may be present but are minimal (Mather and Keenan, Reference Mather and Keenan1998; Mather, Reference Mather2011). It is likely that MFG secretion is a highly derived form of apocrine secretion in which upregulation of milk fat secretion has required incorporation of novel membrane constituents (see Milk Fat Globule section, below). Unfortunately, little is known about the details of secretion in generalized apocrine glands or about the genes expressed and proteins synthesized (Oftedal, Reference Oftedal2002b), although at least some types do not produce milk-specific proteins, such as β-casein (Gritli-Linde et al., Reference Gritli-Linde, Hallberg, Harfe, Reyahi, Kannius-Janson, Nilsson, Cobourne, Sharpe, McMahon and Linde2007). From an evolutionary perspective, it would be very interesting to compare the array of genes expressed by developing and secreting apocrine glands with those expressed during mammary gland development, milk secretion, and mammary involution. For example, nearly 200 milk protein genes and more than 6000 other genes have been identified as expressed in the mammary glands in virgin, pregnant, lactating, involuting, and mastitic cows (Lemay et al., Reference Lemay, Lynn, Martin, Neville, Casey, Rincon, Kriventseva, Barris, Hinrichs, Molenaar, Pollard, Maqbool, Singh, Murney, Zdobnov, Tellam, Medrano, German and Rijnkels2009), but how many of these genes are expressed in apocrine glands is unknown. It would also be instructive to compare gene expression of both apocrine and mammary glands with that occurring in amphibian skin glands to document similarities and differences that may indicate ancestral and derived conditions.
In most mammals, an apocrine gland on the general skin surface is typically associated with both a hair follicle and a sebaceous gland in a triad termed an apo-pilo-sebaceous unit (APSU). The development of the APSU occurs in coordinated fashion, no doubt because of crosstalk between the differentiating epithelial cells and underlying mesenchyme, as well as differences in signaling pathways and receptors of the hair follicle, apocrine gland, and sebaceous gland (Hatsell and Cowin, Reference Hatsell and Cowin2006; Andrechek et al., Reference Andrechek, Mori, Rempel, Chang and Nevins2008; Mayer et al., Reference Mayer, Foley, De La Cruz, Chuong and Widelitz2008). The apocrine gland duct typically opens into the infundibulum of the hair follicle, such that secretion contacts the hair shaft. Remarkably, in monotremes, a similar relationship exists between mammary glands, hair follicles, and sebaceous glands; the three form what can be termed a mammo-pilo-sebaceous unit (MPSU; Oftedal, Reference Oftedal2002b). The galactophores (lactiferous ducts) also open up into the infundibula of enlarged, specialized mammary hairs (Griffiths, Reference Griffiths1978). The mammary glands in monotremes are organized into a small oval mammary patch or areola consisting of 100 to 200 MPSUs (Griffiths, Reference Griffiths1978; Oftedal, Reference Oftedal2002b); there is no nipple. In the area surrounding the mammary patch, APSUs develop. The mature, lobular mammary gland in mid-to-late lactation is very much larger, more branched, and contains many more secretory epithelial cells than an apocrine gland, but in earliest lactation — when monotreme eggs are incubated and hatched – the mammary gland is still relatively small and tubular (Griffiths, Reference Griffiths1978), and thus has a superficial resemblance to an apocrine gland.
In marsupials, such as opossums and kangaroos, there is also a developmental association of mammary glands with hair follicles and sebaceous glands. According to an early work by Bresslau (Reference Bresslau1912 and Reference Bresslau1920), an oval primary-primordium separates into nipple primordia, which deepen into knobs and bud out into hair follicles (primary sprout), mammary glands (secondary sprout) and sebaceous glands (tertiary sprouts). In the opossum, for example, eight hair follicle sprouts are associated with eight mammary sprouts and eight sebaceous sprouts; that is, the nipple primordium develops into eight MPSUs. The hair follicles penetrate the nipple epithelium during development, but are subsequently shed, each leaving a duct (galactophore) by which the mammary gland communicates to the surface of the nipple. As opossums have a dozen or more nipples, approximately 100 MPSUs are involved. In the adult marsupial, the ‘mammary hairs’ are no longer evident, but the galactophores bear testimony to their previous existence. Among different species, the number of ducts penetrating the nipple reflects the numbers of primary hair follicle buds, and can vary from 3 to 33 (Tyndale-Biscoe and Renfree, Reference Tyndale-Biscoe and Renfree1987); however, in species with large numbers of presumptive MPSUs, it is not certain that all secondary sprouts develop into functional mammary lobules.
In eutherian mammals, the association of apocrine glands with hair follicles (APSUs) is maintained, but the association of mammary glands with hair follicles – the presumed ancestral condition – appears to have been lost. In 2002, I hypothesized that this must be due to the inhibition of hair follicle development in the vicinity of mammary glands, and suggested that if the presumptive inhibiting compound (s) could be blocked at the earliest stages of mammary development, hair follicles may develop in association with mammary buds (Oftedal, Reference Oftedal2002b). Although the actual signaling pathways are undoubtedly complex, with both shared and differing sensitivities to signaling compounds among different epithelial cell types, it is intriguing that bone morphogenetic proteins (BMPs) inhibit hair follicle formation, and that when Mayer et al. (Reference Mayer, Foley, De La Cruz, Chuong and Widelitz2008) reduced BMP signaling in the mouse by transgenic overexpression of a BMP antagonist, nipple epithelium was converted into pilo-sebaceous units. Mayer et al. (Reference Mayer, Foley, De La Cruz, Chuong and Widelitz2008) hypothesized that the BMP pathway had been co-opted during evolution of the nipple to suppress hair follicle formation. If so, this occurred following the separation of marsupials and eutherians, given that hair follicle development is not suppressed during nipple development in marsupials; rather, hair follicles subsequently regress and the hairs are shed. Bresslau (Reference Bresslau1920) even described a developmental stage in the koala, after the nipple has everted (extended above the skin surface), in which a tuft of mammary hairs protrude through galactophores at the apex of the nipple.
Thus, available evidence is consistent with the hypothesis that mammary glands developed from an apocrine-like gland that had an association with hair follicles and sebaceous glands. This is not in conflict with the Vorbach et al. (Reference Vorbach, Capecchi and Penninger2006) hypothesis that mammary gland secretion had its origin in mucous glands, as the apocrine-like glands themselves must have originated from earlier tetrapod cutaneous glands. The innate immune system itself is of even more ancient origin, with components shared among invertebrates and vertebrates (Beck and Habicht, Reference Beck and Habicht1996; Hoffmann et al., Reference Hoffmann, Kafatos, Janeway and Ezekowitz1999; Fujita, Reference Fujita2002), and thus the co-option of innate immune system components into the regulatory elements of epithelial–mesenchyme signaling (McClellan et al., Reference McClellan, Miller and Hartmann2008) probably occurred long before the origin of mammary glands.
The evolution of milk constituents
The secretory products of the mammary gland represent the expression of a large number of genes that are upregulated during lactation, but many of these products remain unknown (except as genes) or their functionalities are poorly understood (Smolenski et al., Reference Smolenski, Haines, Fiona, Bond, Farr, Davis, Stelwagen and Wheeler2007; Lemay et al., Reference Lemay, Lynn, Martin, Neville, Casey, Rincon, Kriventseva, Barris, Hinrichs, Molenaar, Pollard, Maqbool, Singh, Murney, Zdobnov, Tellam, Medrano, German and Rijnkels2009). Comparisons of monotreme, marsupial and eutherian genomes suggest that milk and mammary genes tend to be conserved, that is, have not diverged as much as other genes (Lemay et al., Reference Lemay, Lynn, Martin, Neville, Casey, Rincon, Kriventseva, Barris, Hinrichs, Molenaar, Pollard, Maqbool, Singh, Murney, Zdobnov, Tellam, Medrano, German and Rijnkels2009). Nonetheless, there are major constituents of milk in these three lineages for which comparative data provide evidence of an evolutionary origin, especially when milk proteins are compared with closely related gene products.
Caseins
All mammalian milks that have been studied contain multiple casein proteins, characterized as α-, β- and κ-caseins. The caseins are phosphorylated during synthesis, and aggregate into large micelles containing calcium bound to phosphorus in calcium phosphate nanoclusters (Smyth et al., Reference Smyth, Clegg and Holt2004). Multiple caseins participate in these micelles; however, κ-casein plays a particularly important role in stabilizing the micelle in secreted milk. The caseins provide a large proportion of the amino acids, calcium, and phosphorus transferred to the young, and via curd formation in the neonatal stomach they play a part in fat and protein digestion. The caseins had already diverged into the three primary types, α-, β-, and κ-caseins, before the separation of monotremes, marsupials and eutherians, as each casein type is found in all three taxa (Rijnkels, Reference Rijnkels2002 and Reference Rijnkels2003; Lefevre et al., Reference Lefevre, Sharp and Nicholas2009 and Reference Lefevre, Sharp and Nicholas2010). Thus, caseins have a pre-mammalian origin.
The caseins are members of a much larger family of proteins of unfolded nature that are secreted from cells, usually in association with tissue mineralization or regulation of calcium at target tissues. These proteins, termed secretory calcium-binding phosphoproteins (SCPPs), are secreted by secretory epithelial cells or cells derived from underlying ectomesenchymal cells, and have an ancient history in the evolution of mineralized vertebrate tissues (Kawasaki and Weiss, Reference Kawasaki and Weiss2003; Kawasaki, Reference Kawasaki2009). As unfolded proteins, all SCPPs are low in cysteine and therefore cystine disulfide bridges, and a subclass of the proteins (P/Q-rich SCPPs), including the caseins, are particularly rich in proline and glutamine (Kawasaki and Weiss, Reference Kawasaki and Weiss2003; Kawasaki et al., Reference Kawasaki, Lafont and Sire2011). On the basis of the relative locations and structures of exons of these P/Q-rich SCPPs, as well as their phylogenetic distribution, Kawasaki et al. (Reference Kawasaki, Lafont and Sire2011) proposed that the α- and β-caseins derived via gene duplication and exon changes from an ancestral gene (CSN1/2) that derives from another SCPP gene, either ODAM or SCPPPQ1 (which itself is derived from ODAM), whereas κ-casein derives from the SCPP gene FDCSP (which is also derived from ODAM).
Many P/Q-rich SCPPs, including the ODAM and SCPPPQ1 derived proteins, are expressed in mammalian ameloblasts and are involved in mineralization of tooth enamel; follicular dendritic cell secreted peptide (from FDCSP) is found in soft connective tissue (periodontal ligament), where it is thought to prevent spontaneous precipitation of calcium phosphate; it is also expressed in the mammary gland (Kawasaki, Reference Kawasaki2009; Kawasaki et al., Reference Kawasaki, Lafont and Sire2011). Kawasaki et al. (Reference Kawasaki, Lafont and Sire2011) suggest that the initial function of an ancestral SCPP (probably a κ-casein precursor) in protolacteal secretion may have been to regulate calcium delivery to the surface of an egg and to prevent precipitation of calcium phosphate on the parchment-like eggshell. Kawasaki et al. (Reference Kawasaki, Lafont and Sire2011) hypothesize that this may have occurred before the divergence of sauropsids and synapsids, although an ancestral CSN1/2 has yet to be found in a sauropsid genome.
Subsequently, the types of caseins, and also the numbers of genes involved in producing each type, increased via gene duplication and exon changes (Rijnkels, Reference Rijnkels2002; Lefevre et al., Reference Lefevre, Sharp and Nicholas2009 and Reference Lefevre, Sharp and Nicholas2010). By an unknown evolutionary process, the different caseins became involved in the formation of complex micelles stabilized by calcium and phosphate bonds (Smyth et al., Reference Smyth, Clegg and Holt2004). This transformation of ancestral SCPPs into a complex of micelle-forming proteins was essential in converting milk from an egg supplement to a major source of nutrients for the suckling young. Given the small size of mammaliaforms in the late Triassic and Jurassic periods, and hence the small size of their eggs (Hopson, Reference Hopson1973; Oftedal, Reference Oftedal2002b), the novel nutritive function of these SCPPs must have developed before this time, for example, during the Permian and Triassic periods. This is consistent with the estimated loss of multiple vitellogenin genes beginning approximately 170 mya in the Jurassic period (Brawand et al., Reference Brawand, Wahli and Kaessmann2008). Vitellogenin genes could only be inactivated once the nutrient transport function of the caseins had made egg yolk proteins dispensible.
The Milk fat globule (MFG)
Mammals vary tremendously in the fat content of their milk (from less than 1% in rhinos and some lemurs to 60% in some seals; Oftedal and Iverson, Reference Oftedal and Iverson1995); however, in all species studied, milk lipids are secreted as specialized structures known as MFGs. MFGs are lipid spheres bounded sequentially by a phospholipid monolayer, an inner protein coat, a bilayered phospholipid membrane, and a glycosylated surface (Mather and Keenan, Reference Mather and Keenan1998). The collective term for the multi-layered structure or envelope that encloses the lipid sphere is the milk fat globule membrane (MFGM), which is a structure found only in milk. The method of milk lipid secretion appears to be unique, as it has not been found in other organs (McManaman et al., Reference McManaman, Reyland and Thrower2006); if this generalization withstands further investigation of lipid secretion in other gland types, the MGM must be considered a key evolutionary novelty of lactation.
What is of particular interest is that the MFGM contains proteins that appear essential to the synthesis and secretion of MFGs. In particular, two proteins, butyrophilin and xanthine oxidoreductase (XOR), play an obligatory structural role in MFGM synthesis, and if they are reduced or eliminated from mouse mammary cells via knockout of the genes that code them, mice fail to produce normal milk; the triacylglycerols within the secretory cells fail to be secreted into milk fat droplets, but rather accumulate in the cytoplasm or leak into the alveolar lumen as unstructured, amorphic lipid masses (Vorbach et al., Reference Vorbach, Scriven and Capecchi2002; Ogg et al., Reference Ogg, Weldon, Dobbie, Smith and Mather2004). Although the details of protein–protein interactions during formation of the MFGM are not fully understood (Mather, Reference Mather2011), these two proteins have apparently been co-opted from other cellular functions during the evolution of the mammary gland.
The butyrophilin in milk is now correctly specified as butyrophilin1A1, as it is the gene product of only one of the genes (BTN1A) that code for the family of proteins known as butyrophilins (Rhodes et al., Reference Rhodes, Stammers, Malcherek, Beck and Trowsdale2001). The butyrophilins are part of the immunoglobulin superfamily and contain two folded immunoglobulin domains, a transmembrane domain and a C-terminal end that may include a large B30.2 domain. In addition to its role in the MFGM, butyrophilin1A1 has been found to be expressed within the thymus; other butyrophilins are more widely expressed among tissues (Smith et al., Reference Smith, Knezevic, Ammann, Rhodes, Aw, Palmer, Mather and Trowsdale2010a). Butyrophilins and related proteins of the immunoglobulin superfamily appear to play a role in the regulation of proliferation, cytokine secretion and activity of T-cells, and butyrophilin1A1 retains this function, at least in vitro (Smith et al., Reference Smith, Knezevic, Ammann, Rhodes, Aw, Palmer, Mather and Trowsdale2010a). It is interesting that butyrophilin1A1 is the only butyrophilin that appears to bind XOR via its B30.2 domain, and it is this binding that is believed to be critical to MFG secretion from lactocytes (Jeong et al., Reference Jeong, Rao, Xu, Ogg, Hathout, Fenselau and Mather2009). It appears that the ancestral butyrophilin protein was a transmembrane protein in secretory cells that had functions in local immune response, and subsequently evolved a role in synthesis and/or stabilization of the MFGM.
Xanthine oxidoreductase, or XOR, is best known for its role in catalysis of the last two steps in the formation of uric acid, an important nitrogenous waste product, but it has multiple enzymatic functions, and is a member of the molybdo-flavoenzyme (MFE) protein family (Garattini et al., Reference Garattini, Mendel, Romão, Wright and Terao2003). The MFEs are believed to have evolved as an ancestral XOR in prokaryotes (Garattini et al., Reference Garattini, Mendel, Romão, Wright and Terao2003). Although the XOR gene is sometimes considered to code for a housekeeping protein (Vorbach et al., Reference Vorbach, Scriven and Capecchi2002), this is debatable, as XOR is unequally expressed in cells (Garattini et al., Reference Garattini, Mendel, Romão, Wright and Terao2003). In mammals, XOR is initially synthesized as the enzyme xanthine dehydrogenase, but it is readily converted to xanthine oxidase, which is the form typically recovered from milk (Enroth et al., Reference Enroth, Eger, Okamoto, Nishino and Pai2000; Nishino et al., Reference Nishino, Okamoto, Eger, Pai and Nishino2008). Xanthine oxidase generates free radical and reactive nitrogen species, and is upregulated and appears to be during inflammation, leading to the hypotheses that XOR has important antimicrobial activities, perhaps even in milk (Martin et al., Reference Martin, Hancock, Salisbury and Harrison2004), and that XOR may have had an important role in the evolution of innate immunity (Vorbach, Reference Vorbach2003). Certainly, XOR and innate immunity are both of pre-eukaryote origin, and were important long before mammary glands evolved. Yet the upregulation and apical membrane localization of XOR in mammary epithelial cells during mammary gland development (McManaman et al., Reference McManaman, Palmer, Wright and Neville2002); the binding of XOR to the B30.2 domain of butyrophilin (Jeong et al., Reference Jeong, Rao, Xu, Ogg, Hathout, Fenselau and Mather2009); and the failure of MFG formation in heterozygous XOR knockout mice (Vorbach et al., Reference Vorbach, Scriven and Capecchi2002) all indicate a novel function for XOR in the MFGM. Other MFGM components, such as adipophilin, also have important, if incompletely understood, functions (Mather, Reference Mather2011), and each of these has an evolutionary history.
MFG secretion presumably evolved from some previous form of fat secretion, perhaps by tetrapod or synapsid skin glands. Certainly, some extant frogs secrete lipids as a means of reducing water loss across the skin (Lillywhite et al., Reference Lillywhite, Mittal, Garg and Agrawal1997; Lillywhite, Reference Lillywhite2006), and secreted lipids applied to eggs could have had an impact on egg moisture loss (Oftedal, Reference Oftedal2002a). However, much more research is required to understand the differences and similarities of secretory mechanisms. Mammary glands bear developmental and structural resemblance to apocrine glands, which led to the hypothesis that mammary glands are derived from ancient apocrine-like glands (Oftedal, Reference Oftedal2002b). One can imagine a scenario in which an ancestral apocrine secretion entailed the secretion of apical blebs containing cytoplasm, secretory vesicles and perhaps cytoplasmic lipid droplets, similar to the process described for some specialized apocrine glands, such as human axillary apocrine glands, glands of Moll, ceruminous glands in the outer ear canal, and rodent Harderian glands (Gesase and Satoh, Reference Gesase and Satoh2003; Stoeckelhuber et al., Reference Stoeckelhuber, Stoeckelhuber and Welsch2003; Stoeckelhuber et al., Reference Stoeckelhuber, Matthias, Andratschke, Stoeckelhuber, Koehler, Herzmann, Sulz and Welsch2006; Stoeckelhuber et al., Reference Stoeckelhuber, Schubert, Kesting, Loeffelbein, Nieberler, Koehler and Welsch2011). With an upregulation of mammary fat synthesis, mechanisms may have evolved to minimize cytoplasmic loss to a few cytoplasmic crescents, as in milk secretion. The increase in lipid secretion must have occurred before the miniaturization of the mammaliaforms in the Triassic period and gradually replaced the nutritional role of lipids provided by yolk, allowing the inactivation of vitellogenins involved in transport and storage of lipids in the egg yolk (Brawand et al., Reference Brawand, Wahli and Kaessmann2008).
Milk sugar synthesis
All mammalian milks contain at least traces of sugar (Oftedal and Iverson, Reference Oftedal and Iverson1995); in most eutherians the predominant sugar is lactose, whereas in monotremes, marsupials and some eutherian carnivores oligosaccharides predominate (Urashima et al., Reference Urashima, Saito, Nakamura and Messer2001; Messer and Urashima, Reference Messer and Urashima2002; Uemura et al., Reference Uemura, Takahashi, Senda, Fukuda, Saito, Oftedal and Urashima2009; Senda et al., Reference Senda, Hatakeyama, Kobayashi, Fukuda, Uemura, Saito, Packer, Oftedal and Urashima2010). Both lactose and oligosaccharides with lactose at the reducing end are unique to milk (Toba et al., Reference Toba, Nagashima and Adachi1991) and require a novel synthetic pathway.
In the mammary secretory cell, the synthesis of lactose begins with the synthesis of a unique milk protein, α-lactalbumin, in the rough endoplasmic reticulum (Brew, Reference Brew2003). α-lactalbumin is then transported to the Golgi apparatus. A transmembrane protein in the trans Golgi, β-1,4-galactosyltransferase1 (β4gal-T1), binds UDP-galactose, producing a conformational change that allows α-lactalbumin to be bound (Ramakrishnan and Qasba, Reference Ramakrishnan and Qasba2001). When α-lactalbumin binds to β4gal-T1, it alters the specificity of β4gal-T1, allowing glucose to become the acceptor sugar for galactose transfer, resulting in the synthesis of lactose. Thus, α-lactalbumin acts as a regulator of β4gal-T1, and without α-lactalbumin β4gal-T1 does not synthesize lactose under physiological conditions (Brew, Reference Brew2003).
It has been apparent for many years – from amino acid sequence similarity, three-dimensional structure and the structure of the exons that code for α-lactalbumin – that α-lactalbumin is most closely related to c-type lysozyme and is derived from it via gene duplication and base pair substitution (Prager and Wilson, Reference Prager and Wilson1988; Qasba and Kumar, Reference Qasba and Kumar1997; Brew, Reference Brew2003). The estimated date of origin of α-lactalbumin from c-lysozyme is ancient, before the time of the split of synapsids from sauropsids approximately 310 mya (Prager and Wilson, Reference Prager and Wilson1988). Many authors have been puzzled by this date, as it was assumed that mammary glands did not arise until the appearance of ‘early mammals’ (i.e. mammaliaforms) 100 million or more years later (Hayssen and Blackburn, Reference Hayssen and Blackburn1985; Prager and Wilson, Reference Prager and Wilson1988; Qasba and Kumar, Reference Qasba and Kumar1997; Messer and Urashima, Reference Messer and Urashima2002), but it is consistent with an ancient origin of lactation.
It is probable that c-lysozyme, as a normal antimicrobial constituent of epithelial secretions and egg white (Callewaert and Michiels, Reference Callewaert and Michiels2010), would have been present in tetrapod and early synapsid skin secretions, including secretions delivered to eggs; a c-type lysozyme is present in amphibian skin secretions (Zhao et al., Reference Zhao, Jin, Lee and Zhang2006). Although the antimicrobial function of c-lysozyme would presumably help protect eggs (as it does in egg white), what immediate advantage would accrue to eggs or hatchlings from the conversion of c-lysozyme function to that of α-lactalbumin, with the resultant synthesis of lactose? α-lactalbumin does not have lysozyme activity, whereas lysozymes do not bind to β4Gal-T1, because of differences in amino acid composition at key positions involved in binding of substrate (in lysozyme) or binding of β4gal-T1 (in α-lactalbumin; Ramakrishnan and Qasba, Reference Ramakrishnan and Qasba2001; Messer and Urashima, Reference Messer and Urashima2002; Brew, Reference Brew2003; Callewaert and Michiels, Reference Callewaert and Michiels2010), and thus an ‘intermediate’ with both functions may not have been possible. In addition, one must assume that any lactose thus synthesized would have been indigestible to embryos or hatchlings, given that the intestinal brush-border enzyme lactase could not have evolved without a substrate to digest, and lactose does not occur elsewhere.
Messer and Urashima (Reference Messer and Urashima2002) argue that the ancestral function of α-lactalbumin as a β4Gal-T1 regulator may have been the production of lactose-containing free oligosaccharides, rather than free lactose per se. The amount of α-lactalbumin synthesized in the monotreme mammary gland is minor, and Messer and Urashima (Reference Messer and Urashima2002) assume this to be the ancestral condition. A wide range of glycosyltransferases would have been present in the trans Golgi of tetrapods and early synapsids, as these are part of the normal synthetic machinery for glycosylation of glycoproteins, glycolipids, and proteoglycans in vertebrates, and they are of ancient origin (Varki, Reference Varki1998; Lowe and Varki, Reference Lowe and Varki1999). A low rate of lactose synthesis, coupled with high activity of other glycosyltransferases that could glycosylate lactose, may have produced free oligosaccharides rather than free lactose, similar to what is observed in extant monotremes and marsupials. Milk oligosaccharides have antimicrobial or probiotic effects, for example by leading pathogens to ‘mistake’ free oligosaccharides for the oligosaccharide chains of the glycocaylx on apical cell membranes (Newburg, Reference Newburg1996), and thus to fail to bind to these surfaces. Such an effect might benefit the mammary gland, an egg surface, or the digestive tract of a hatchling even before the evolution of the lactase enzymatic mechanism. It is intriguing that marsupial young that consume milk containing oligosaccharides but not lactose do not have intestinal lactase (Crisp et al., Reference Crisp, Messer and Cowan1989), but whether this is the ancestral mammalian condition is not known.
In eutherians, lactose accumulating in the Golgi apparatus creates an osmotic gradient, which draws water into the Golgi; this aqueous phase (including lactose, α-lactalbumin, other whey proteins, caseins, electrolytes, etc.) is subsequently packaged into secretory vesicles for transport to the apical plasma membrane of the secretory cell (Shennan and Peaker, Reference Shennan and Peaker2000). This model of milk secretion entails substantial upregulation of α-lactalbumin and lactose synthesis, and the transcription of β4Gal-T1 is also upregulated above constitutively expressed levels via the use of a second transcriptional start site, regulated by a stronger promoter and by more efficient translation of the truncated transcript (Shaper et al., Reference Shaper, Charron, Lo and Shaper1998). Although long considered the ‘standard’ model of milk secretion, this may represent a derived feature of eutherian lactation that only developed after the young evolved the ability to digest lactose. One group of eutherian mammals, the fur seals and sea lions (Pinnipedia: Otariidae), have secondarily lost the ability to synthesize α-lactalbumin because of gene mutations and changes in transcription rates (Sharp et al., Reference Sharp, Cane, Lefevre, Arnould and Nicholas2005; Reich and Arnould, Reference Reich and Arnould2007; Sharp et al., Reference Sharp, Lefevre and Nicholas2008); therefore the milk is devoid of lactose or lactose-based oligosaccharides (Oftedal et al., Reference Oftedal, Boness and Tedman1987a and Reference Oftedal2011). These taxa manage to produce large volumes of high-fat milk (Oftedal et al., Reference Oftedal, Iverson and Boness1987b; Arnould and Boyd, Reference Arnould and Boyd1995; Arnould et al., Reference Arnould, Boyd and Socha1996), but the secretory processes by which the aqueous phase is secreted have not been studied. In mice, knockout of the gene for α-lactalbumin results in a very low level of secretion of high-fat milk, and the offspring do not survive (Stinnakre et al., Reference Stinnakre, Vilotte, Soulier and Mercier1994; Stacey et al., Reference Stacey, Schnieke, Kerr, Scott, McKee, Cottingham, Binas, Wilde and Colman1995).
Whey proteins as amino acid sources
Caseins have a loosely folded structure with few cystine disulfide bonds, and as a consequence contain a relative deficit of sulfur-containing amino acids (SAA, i.e. methionine and cysteine) relative to the requirements of offspring. In cow's milk, αs-, β-, and κ-caseins contain approximately 2.9% to 3.7% SAA, by mass, whereas α-lactalbumin and β-lactoglobulin contain approximately 7% to 8% SAA (calculated from data presented in Fox Reference Fox2003). Suckling mammals appear to require that SAA be present as 4% to 6% of total amino acids in order to attain maximal growth (Foldager et al., Reference Foldager, Huber and Bergen1977; Burns and Milner, Reference Burns and Milner1981; Fuller et al., Reference Fuller, McWilliam, Wang and Giles1989; National Research Council, 1995). Methionine can substitute for cysteine in most cases (except, perhaps, in premature human infants; Fomon et al., Reference Fomon, Ziegler, Nelson and Edwards1986; Thomas et al., Reference Thomas, Gruca, Bennett and Parimi2008), but cysteine can only replace approximately half of the methionine requirement in growing animals (Fuller et al., Reference Fuller, McWilliam, Wang and Giles1989). In formulating casein-based diets, supplemental cysteine or methionine are required to compensate for the SAA deficit in caseins (e.g. Reeves et al., Reference Reeves, Nielsen and Fahey1993; National Research Council, 1995). This suggests that other proteins had to coevolve with caseins if milk was to be a balanced source of amino acids, rather than just a supplement. The major milk-specific whey proteins, depending on species, are α-lactalbumin (e.g. in human milk), β-lactoglobulin (e.g. in cow's milk), and WAP (e.g. in rat milk), but additional whey proteins have been identified in marsupials, such as early lactation protein, late lactation protein, and trichosurin (Nicholas et al., Reference Nicholas, Messer, Elliott, Maher and Shaw1987; Piotte and Grigor, Reference Piotte and Grigor1996; Demmer et al., Reference Demmer, Ross, Ginger, Piotte and Grigor1998; Piotte et al., Reference Piotte, Hunter, Marshall and Grigor1998).
β-lactoglobulin is the major whey protein in most ruminant milks (including dairy animals such as dairy cattle, goats, sheep and water buffalo), but does not have any indisputable biological role beyond supplying amino acids to the offspring (Sawyer, Reference Sawyer2003). As β-lactoglobulin occurs in the milks of monotremes (platypus), several marsupials (brushtail possum, wallabies and kangaroos) and at least 35 species of eutherians, it must have evolved before the divergence of these groups in the Jurassic or Cretaceous period. The discovery that β-lactoglobulin was similar in structure to retinol-binding protein (RBP) led to the hypothesis that β-lactoglobulin might have a role in the transport to the young of vitamin A, vitamin D, fatty acids, or some other essential lipophilic compounds, or may play a role in intestinal uptake of these constituents (Pervaiz and Brew, Reference Pervaiz and Brew1985; Perez and Calvo, Reference Perez and Calvo1995; Yang et al., Reference Yang, Chen, Chen, Wu and Mao2009). However, in ruminants, vitamin A is associated with the fat globule and not β-lactoglobulin; in pigs and horses, β-lactoglobulin does not bind either retinol or fatty acids; and, in suckling pups of mice that have not been genetically modified as per Yang et al. (Reference Yang, Chen, Chen, Wu and Mao2009), vitamin D is obviously absorbed from milk despite the absence of β-lactoglobulin or the pups would develop vitamin D deficiencies. Thus, if β-lactoglobulin has any role in transport and/or intestinal uptake of these lipophilic constituents, it is neither essential nor universal (Perez and Calvo, Reference Perez and Calvo1995). A major problem in ascribing a functional role is that β-lactoglobulin is absent from the milks of so many mammals, including laboratory mice and rats, guinea pigs, domestic rabbits, dromedary camels, llamas and humans (Sawyer, Reference Sawyer2003).
Both β-lactoglobulin and RBP are members of a large family of small extracellular proteins, termed lipocalins, that have similar tertiary structure, specific amino acid sequence motifs and exon–intron structure of coding genes (Flower, Reference Flower1996; Akerstrom et al., Reference Akerstrom, Borregaard, Flower and Salier2006). This ancient protein family (or superfamily) apparently derives from a bacterial protein and is characterized by a barrel-shaped lipophilic cavern surrounded by a series of eight β-strands, and that is open on one end (Ganfornina et al., Reference Ganfornina, Sanchez, Greene and Flower2006). Many lipocalins are known to function via transport and/or sequestration of hydrophobic compounds in this ‘barrel’ and occasionally at secondary binding sites. Analysis of the molecular evolution of the lipocalins suggests that RBP diverged from the other lipocalins first, followed by β-lactoglobulin, suggesting that β-lactoglobulin is of more ancient origin than other vertebrate lipocalins (other than RBP), including lipocalins that are found in fish and amphibia (Ganfornina et al., Reference Ganfornina, Gutierrez, Bastiani and Sanchez2000; Sanchez et al., Reference Sanchez, Ganfornina, Gutierrez and Marin2003 and Reference Sanchez, Ganfornina, Gutierrez, Gauthier-Jauneau, Risler and Salier2006). It is likely that the ancestral β-lactoglobulin had similar function to that of an ancestral RBP-like protein, that is, transporting hydrophobic compounds in extracellular and/or secreted fluids, long before the appearance of milk as we know it. Although β-lactoglobulin retains a generalized ability to bind a variety of hydrophobic ligands, because of the elasticity of the outer parts of the barrel (Konuma et al., Reference Konuma, Sakurai and Goto2007), its current role in milk appears to be primarily a nutritional one, and in species in which other whey proteins predominate, β-lactoglobulin has become superfluous and has been lost. Two β-lactoglobulin genes have been observed in ruminants, but one is non-coding and is thus a pseudogene (Sawyer, Reference Sawyer2003). A β-lactoglobulin pseudogene is also suspected in the human genome, but there may be confusion with the glycodelin gene (Kontopidis et al., Reference Kontopidis, Holt and Sawyer2004). Other lipocalins (trichosurin, late lactation protein) are expressed in marsupial milk (Demmer et al., Reference Demmer, Ross, Ginger, Piotte and Grigor1998; Piotte et al., Reference Piotte, Hunter, Marshall and Grigor1998); however, these are only distantly related to β-lactoglobulin and are apparently of more recent origin (Ganfornina et al., Reference Ganfornina, Gutierrez, Bastiani and Sanchez2000).
Among whey proteins, WAP has the highest sulfur amino acid content, approximately 17% to 20% by mass. The key feature of WAP is the presence of two or three domains of approximately 40 to 50 amino acids, each of which contains eight cysteine residues involved in four disulfide bonds; as the domain was first recognized in WAP, it is termed the Whey Acidic Protein Four-Disulphide Core (WFDC) domain. There are at least 33 distinct (non-homologous) proteins among vertebrates and invertebrates that include one to four WFDC domains (see PROSITE, www.expasy.org/cgi-bin/prosite), including proteins with antibacterial, antiviral and anti-inflammatory functions, as well as several proteinase inhibitors. All are secreted proteins, including proteins in respiratory, reproductive and other epithelial secretions (Hagiwara et al., Reference Hagiwara, Kikuchi, Endo, Huqun, Usui, Takahashi, Shibata, Kusakabe, Xin, Hoshi, Miki, Inooka, Tokue and Nukiwa2003; Bingle et al., Reference Bingle, Cross, High, Wallace, Rassl, Yuan, Hellstrom, Campos and Bingle2006). The structural similarity of WAP to other WFDC-containing proteins has led to speculation that WAP may also have antibacterial or proteinase inhibition functions, but attempts to demonstrate this have failed (Hajjoubi et al., Reference Hajjoubi, Rival-Gervier, Hayes, Floriot, Eggen, Piumi, Chardon, Houdebine and Thepot2006; Sharp et al., Reference Sharp, Lefevre and Nicholas2007). There is evidence that WFDC domains influence cell proliferation and growth in vitro and in transgenic mice (reviewed by Topcic et al., Reference Topcic, Auguste, De Leo, Lefevre, Digby and Nicholas2009), but when the WAP gene is deleted in knockout mice, the mice continue to develop normal mammary glands indicating that WAP is not essential for mammary cell differentiation or proliferation (Triplett et al., Reference Triplett, Sakamoto, Matulka, Shen, Smith and Wagner2005). The primary effect of WAP deletion in mice appears to be growth retardation of the young during the second half of lactation. Thus, the functional role of WAP in milk, other than as a rich source of sulfur amino acids for the young, remains unclear.
The WFDC domain itself is of ancient origin, being a component in secreted proteins involved in the regulation of shell mineralization in mollusks such as abalone (Treccani et al., Reference Treccani, Mann, Heinemann and Fritz2006), and in antimicrobial response as part of the innate immunity of crustaceans and perhaps insects (Zou et al., Reference Zou, Evans, Lu, Zhao, Williams, Sumathipala, Hetru, Hultmark and Jiang2007; Jia et al., Reference Jia, Sun, Wang, Wang, Wang, Zhao and Wang2008; Smith et al., Reference Smith, Desbois and Dyrynda2010b). A number of WFDC domain-containing proteins are also secreted by snake venom glands, where they have antibacterial function (Nair et al., Reference Nair, Fry, Alewood, Kumar and Kini2007; Fry et al., Reference Fry, Scheib, van der Weerd, Young, McNaughtan, Ramjan, Vidal, Poelmann and Norman2008), and by skin glands in frogs, where they serve as antimicrobial defensive compounds (Ali et al., Reference Ali, Lips, Knoop, Fritzsch, Miller and Conlon2002; Zhang et al., Reference Zhang, Zhang, Nie, Liu, Xie and Shang2009). Although much more research is required to determine relationships among invertebrate and vertebrate WFDC-containing proteins, it is likely that an ancestral WAP present in the glandular skin secretion of an egg-tending tetrapod or early synapsid served as a defensive compound against microbes as a component of the innate immune system, similar to existing WFDC proteins in mammalian epididymal, respiratory, and oral mucosal secretions (Hiemstra, Reference Hiemstra2002; Hagiwara et al., Reference Hagiwara, Kikuchi, Endo, Huqun, Usui, Takahashi, Shibata, Kusakabe, Xin, Hoshi, Miki, Inooka, Tokue and Nukiwa2003; Bingle and Vyakarnam, Reference Bingle and Vyakarnam2008), and in frog skin secretions (Ali et al., Reference Ali, Lips, Knoop, Fritzsch, Miller and Conlon2002; Zhang et al., Reference Zhang, Zhang, Nie, Liu, Xie and Shang2009).
As with β-lactoglobulin, it appears that WAP may have lost its purported ancestral function as proteinase inhibitor/antimicrobial protein in skin secretions. Of the two WFDC domains in WAP in eutherian milks, designated as DI and DII, only DII retains the characteristic N-terminal motif found in most WFDC domains (Lys-X-Gly-X-Cys-Pro, where X represents various amino acids); amino acid substitutions in this and other areas of DI may have altered the charge distribution, glycosylation sites, and conformation in such a way that original functions are no longer possible (Ranganathan et al., Reference Ranganathan, Simpson, Shaw and Nicholas1999). Monotreme and marsupial WAPs contain two to three WFDC domains, but of differing sequence and arrangement than eutherian WAPs (Sharp et al., Reference Sharp, Lefevre and Nicholas2007), and it is thought that they may retain functions lost in eutherians, but more evidence is required (Topcic et al., Reference Topcic, Auguste, De Leo, Lefevre, Digby and Nicholas2009). At least some eutherians have lost WAP in entirety. Although the genes for WAP synthesis have been found in sheep, goats, and cattle, they are missing a nucleotide at the end of the first exon, causing a frameshift mutation (Hajjoubi et al., Reference Hajjoubi, Rival-Gervier, Hayes, Floriot, Eggen, Piumi, Chardon, Houdebine and Thepot2006). They are not transcribed and are thus pseudogenes.
Conclusion
The proposal that mammary secretion has an ancient origin and long evolutionary history (Oftedal, Reference Oftedal2002a and Reference Oftedal2002b) is increasingly accepted, especially in the face of supportive molecular evidence (Kawasaki and Weiss, Reference Kawasaki and Weiss2003; Vorbach et al., Reference Vorbach, Capecchi and Penninger2006; Brawand et al., Reference Brawand, Wahli and Kaessmann2008; McClellan et al., Reference McClellan, Miller and Hartmann2008; Capuco and Akers, Reference Capuco and Akers2009; Lemay et al., Reference Lemay, Lynn, Martin, Neville, Casey, Rincon, Kriventseva, Barris, Hinrichs, Molenaar, Pollard, Maqbool, Singh, Murney, Zdobnov, Tellam, Medrano, German and Rijnkels2009; Lefevre et al., Reference Lefevre, Sharp and Nicholas2010; Kawasaki et al., Reference Kawasaki, Lafont and Sire2011). It is now possible to formulate a more detailed scenario by which the secreted fluid came to resemble in form and function what we now know as milk. The evolutionary origins of milk appear to be found in the secretions of primitive apocrine-like glands in the skin of early synapsids or even of taxa living before the split of synapsids from sauropsids approximately 310 mya. These glands incorporated elements of the innate immune system in providing protection to the skin and to eggs that were moistened. Membrane and intracellular proteins in secretory epithelial cells, such as butyrophilin and xanthine oxidase, were incorporated into the MFGM, secretory calcium-binding phosphoproteins involved in extracellular regulation of calcium and phosphorus were transformed into casein micelles, and antimicrobial proteins such as lysozyme and WFDC-containing proteins were incorporated as whey proteins, with the transformed lysozyme (i.e. α-lactalbumin) serving to alter sugar synthesis such that new milk-specific sugars were formed. Over a period of perhaps 150 million years, and during the course of multiple radiations of early synapsids and the descendent lineages leading to mammals, this secretory fluid and the glands that produced it became more complex, the volumes produced became greater, and the extent of dependence by hatchlings and growing young increased. The presence of a glandular skin secretion appears to have been essential to endothermic incubation of eggs with parchment-like shells. The sequential radiations of basal synapsids, therapsids, cynodonts, and mammalialforms became increasingly mammal-like in terms of metabolism, locomotor ability, structural adaptations to diet, and reproduction. On the one hand, the predominance of milk secretion led to reduction in egg yolk mass and smaller and less-developed hatchlings; on the other hand, growth patterns changed as offspring relied for longer periods on milk, with delayed tooth eruption and more specialized mature dentition. It was only because lactation was far advanced that Triassic and Jurassic mammaliaforms could become so tiny, with adult body masses of some taxa weighing only a few grams.
Despite the tremendous diversity in milk composition among extant mammals, the specific constituents of milk are common to the primary mammalian lineages, reflecting that they were inherited from pre-mammalian taxa. The major qualitative differences that have arisen among mammalian taxa (such as lack of α-lactalbumin in fur seals and sea lions, lack of WAP in ruminants, and lack of β-lactoglobulin in humans) are due to losses of functional components, rather than de novo creation of new constituents. The evolution of the placenta in eutherians also removed much of the need for nutrient transfer during early development. Marsupials, which have no or very simple placentas, rely on lactation for a greater span of development than eutherians, and not surprisingly, marsupial milks undergo much greater changes in composition over the course of lactation than do most eutherian milks (Oftedal and Iverson, Reference Oftedal and Iverson1995). It is not known whether some of the unique whey proteins that have been found in marsupials represent milk constituents that were lost during eutherian evolution, or if they have evolved since marsupials and eutherians diverged.
When human civilizations finally developed animal agriculture, they were able to select as dairy animals a suite of species (e.g. cows, sheep, goats, water buffaloes, yaks, camels, horses, and asses) characterized by relatively dilute milks containing modest levels of fat, caseins, and whey proteins, and relatively high levels of lactose. Perhaps such species were selected because their milks bore resemblance to human milk, in being dilute, or perhaps such species were easier to milk because their dilute milks accumulated in substantial storage cisterns between bouts of suckling. However, it is only now that we are beginning to appreciate that the remarkable secretory product that is the foundation of the dairy industry had such a very long evolutionary history before mammals appeared on earth.
