Experimental Biology and Medicine 229:439-463 (2004)
© 2004 Society for Experimental Biology and Medicine
MINIREVIEW
Biological Roles of Alpha-Fetoprotein During Pregnancy and Perinatal Development
Gerald J. Mizejewski1,
Division of Molecular Medicine, Wadsworth Center, New York State Department of Health, Empire State Plaza, Albany, New York 12201
1To whom requests for reprints should be addressed at Division of Molecular Medicine, Wadsworth Center, New York State Department of Health, Empire State Plaza, Albany, NY 12201. E-mail: Mizejew{at}Wadsworth.org
 |
Abstract
|
|---|
The use of alpha-fetoprotein (AFP) as a serum marker in cancer actually predates its employment in the detection of congenital defects; however, the latter use of AFP as a fetal defect marker has propelled its clinical utilization. Although the serum-marker capacity of AFP has long been exploited, less is known of the biological activities of this oncofetal protein during fetal and perinatal development. In the present review, the biological activities of AFP are discussed in light of this glycoprotein’s presence in various biological fluid compartments: embryonic and fetal tissues, serum, urine, and reproductive fluids. After a review of the histochemical detection of AFP in various cells and tissues during development, AFP concentrations within various biological fluids were discussed in the context of gestational age and anatomic location. Discussion follows concerning the relationships and roles of AFP in developmental events such as erthyropoiesis, histogenesis/organogenesis, and ligand binding and in developmental disorders such as hypothyroidism, folate deficiencies, and acquired immunodeficiency disorder (AIDS). Based on its association with so many types of birth defects, malformations, and congenital anomalies, AFP can be viewed as a molecular "troubleshooter" until signal transduction pathways are established during pregnancy and prenatal development. The review concludes with a discussion of the place of AFP in the rapidly expanding field of proteomics.
Key Words: alpha-fetoprotein • differentiation • pregnancy • growth • fetus • perinatal • development • proliferation • infancy
|
Introduction
|
|---|
Historical.
A fetal component not commonly found in adults was first detected as a postalbumin migrating protein in fetal serum by Bergstrand and Czar in 1956, using paper electrophoretic techniques (1); subsequently, Masopust and Kotal assigned to the unknown developmental protein of Bergstrand and Czar the name "fetoprotein" (2). Gitlin and co-workers (3) then devised the name "alpha-fetoprotein" (AFP) for the electrophoretic 
1-migrating human fetal protein. In 1963, Abelev and co-workers (4) reported, in hepatoma-bearing mice, a protein that migrated in the 1 region of an electrophoretogram, and in 1965 Tatarinov described a similar protein in the sera of humans bearing hepatomas (5). In the early 1970s, Brock and coauthors/co-workers reported elevated AFP levels in human amniotic fluid (6) and in maternal serum (SAFP; Refs. 7, 8) that correlated with the presence of neural tube defects in the fetus. Thus, studies involving the 1-migrating fetal protein as a gestational age–dependent fetal defect marker actually postdated its recognition as a tumor marker, and the name "oncofetal protein" was assigned (9).
Mammalian AFPs are single-chain glycoproteins with molecular masses ranging from 66 to 72 kDa and a 3%–5% carbohydrate (glycan) content (10–12). Alpha-fetoprotein is a tumor-associated fetal protein classified as a member of a three-domain albuminoid gene family that currently consists of four members: albumin (ALB), vitamin-D binding protein (DBP), AFP, and alpha-albumin (-ALB; Refs. 13, 14). Similarly to ALB, SAFP is known to bind and transport a multitude of ligands, including bilirubin, fatty acids, retinoids, steroids, heavy metals, dyes, flavonoids, phytoestrogens, dioxins, and various organic drugs (15, 16). Unlike ALB, high concentrations of hydrophobic ligands (i.e., fatty acids, estrogens) have been reported to induce an irreversible conformational change in the tertiary structure of AFP (see Ref. 17 for review). Altered SAFP levels have been observed concurrent with aberrant growth manifestations, but it was usually assumed that these AFP levels were coincident events rather than the cause of such changes. Although AFP may not be the direct cause of the altered growth manifestations observed in birth defects, it is conceivable that some shock/stress-induced conformational (variant) forms of this fetal protein influence, modify, or contribute to such events. Over the past decade, reports have emerged that some of these AFP forms can serve as dual regulators of growth, capable of both enhancement and inhibition of growth, in cancer as well as fetal cells (18, 19). The growth regulatory property of AFP is one of the most prominent characteristics that distinguishes this fetal protein from serum ALB.
|
Objectives
|
|---|
To date, there exists a paucity of published reports in the biomedical literature addressing the biological activities of AFP during fetal development. Moreover, the physiological roles of this oncofetal protein in the regulation of growth and differentiation during mammalian development have not been recently reviewed or updated. The objectives of the present review are 4-fold. First, the various uses of AFP, as an investigational probe and/or marker during development, especially in experimental settings, will be described. Second, the relationship of AFP’s biological (physiological) activities in experimentally induced mammalian (including human) growth/differentiation models will be presented. Third, the multitude of congenital malformations, disease states, and biological activities ascribed to AFP in recent years justify a review that links the presence of developmental anomalies of AFP with physiological status. Finally, the biology of AFP during growth and development will be emphasized; many prior reviews have focused on AFP only as a diagnostic fetal defect marker. For a more extensive exposure to the physical chemistry and genetics of AFP, the reader should consult previous reviews (20–24). The present review is also intended to serve as an update and extension of previous reviews on AFP published in this journal (16, 17).
|
Ontogeny of AFP
|
|---|
The original observation that AFP was synthesized by a variety of mammalian tissues prior to and following parturition stemmed from the pioneering studies of David Gitlin (25). In 1972, Gitlin and co-workers demonstrated that AFP was synthesized by fetal liver and yolk sac; subsequent cell culture studies by his group indeed demonstrated that AFP was synthesized by a multitude of tissues, especially those of gastrointestinal origin (26). The detectability of AFP during ontogeny precedes the detectability of ALB by a considerable period of time. During the rodent 21-day gestation period, ALB is not synthesized until Days 12–13, while AFP is detected by Day 6 following fertilization and implantation (27). Moreover, AFP synthesis in rodents is initiated by events prior to and during implantation. In the mouse, AFP (MAFP) has been histochemically detected in the inner cell mass of the blastocyst, in both the outer and the inner layers of the primitive endoderm (28, 29). The outer endodermal layer gives rise to the parietal endoderm, while the visceral endoderm emerges from the inner endoderm layer. Bovine AFP (BAFP) has also been detected in the 14-day trophoblast, and by Day 16, BAFP is secreted into the amniotic fluid (30). Similar to the AFPs of marmosets and rodents, BAFP is detected in both the preimplantation and the postimplantation conceptus (31). In summary, all mammalian species studied thus far show histochemical evidence of AFP in the pre- and postimplantation embryos, yolk sac, amnion, embryonic disc, and early primitive streak stages.
The synthesis of AFP in the liver anlage (primordium) and other embryonic tissues has been extensively studied through analysis of expression patterns of AFP mRNA in both human and mouse embryos (36). Human AFP (HAFP) is expressed in the yolk sac, hindgut/midgut endoderm, and the foregut hepatic diverticulum at 26 days postovulation. At 32–52 days postovulation, HAFP was found to be strongly expressed in the mesonephric duct and tubules; however, HAFP was only transiently expressed in the pancreas at 40–50 days. Although HAFP was not expressed in the metanephric kidney, expression was apparent in the bile duct and gallbladder endoderm.
In comparison, MAFP mRNA is expressed in the primitive hepatocytes of the hepatic buds of 9.5-day embryos (27). At this stage, MAFP expression was also observed in midgut and hindgut endoderm up to Day 13.5. As seen in the human embryo at a corresponding stage, pancreatic expression was at best weak and transient only, during the Day 11.5–Day 13.5 period. In contrast to the human pattern, MAFP expression was not detected in either the mesonephric or the metanephric kidneys.
|
AFP in Developing Brain
|
|---|
The presence of AFP in tissues of fetal brain was first described by Gillin and co-workers (5) in culture studies employing tissues of the human conceptus (26). Later, in murine brain cytosols, it was reported that the major estradiol binding protein in the fetal brain was actually AFP, a protein synthesized in liver and yolk sac (32). Human fetal brain tissues obtained at autopsy and stained by immunohisto-chemical procedures revealed positive AFP staining in nerve cells of the cerebral wall, brain stem nuclei, and the epithelial layers of the choroid plexus (CP; Ref. 33). The presence of AFP within cells of the CP suggested that the fetal protein was transudated from the blood to the cerebrospinal fluid (CSF) via a cellular route across CP epithelial layers. These observations prompted the proposal that presence of AFP in the brain plays a role in neuronal differentiation and/or development. Indeed, the localization of AFP and other serum proteins in various neurons of the developing mouse brain was confirmed by means of immunohistochemical methods throughout the 1980s (34–45).
Immunohistochemical procedures were further used to document that AFP was present in developing rat and mouse brains throughout fetal and postnatal development and for up to 20–25 days following birth (34, 36, 39, 40). Large tissue areas and groups of cells in many regions of the developing brain, from the olfactory bulb to the medulla oblongata, stained positively for AFP at various time intervals during development. Intracellular labeling localized AFP to the cytoplasm of the neuronal cells and extending into these cells’ axonic and dentritic extensions. In human brain tissues, peak levels of brain AFP were detected in fetuses up to and including the 20th week of gestation (41, 42). However, no AFP could be detected in fetal brains obtained from the third trimester of human pregnancies. In sheep fetal brain homogenates, histochemical detection of AFP was reported prior to 60 days’ gestation in many immature neurons in the neuroependymal layers and in several layers of the developing cortical plate (35). In like fashion, AFP was localized in fetal pig brain tissues representing the ventricular ependyma, meningeal envelopes, CP, and blood vessel walls of the brain (37). As seen in humans, porcine AFP was found only in the cytoplasm of differentiating neurons at the axonal pole of the cell and in the dendritic processes of pyramidal cells (43). Studies in the 9-week-old fetal baboon brain further demonstrated the presence of AFP in neural tube and neural crest derivatives and in the ventricles, which displayed an intracytoplasmic staining pattern (38). The staining localization of baboon AFP appeared to decline as myelination and glial-cell development progressed. Uriel and co-workers proposed that the binding and transport of polyunsaturated fatty acids by AFP (see Ref. 44) could explain the presence of this fetal protein in the developing nervous system (see the section "AFP and Proteomics"). Studies in human embryos further showed immunohistochemical localization of AFP to in the lateral ventricular zones of the fore-, mid-, and hindbrain (41, 42).
By the early 1980s, it had become evident from published reports that innate intracellular synthesis of AFP in brain cells was unlikely. Although AFP and other serum proteins were found intraneuronally in developing mammalian brains, observations showed that local (brain) production, of either AFP proteins or the mRNAs coding for them, could not account for this localization (45). These data supported the concept of a nonbrain origin of AFP. Indeed, later reports confirmed that radiolabeled AFP was actually taken up and incorporated into developing brain cells undergoing neuronal differentiation (44). The evidence against the synthesis of AFP in rat brain cells was further strengthened by reports that no mRNA transcripts for AFP could be found in fetal brain and heart tissues, in contrast to the situation in fetal intestine, lung, liver, and kidney tissues (46). However, both AFP and ALB gene transcripts have since been reported by in situ hybridization, in various rat tissue sections, using S35-labeled AFP and ALB cDNA probes (47). In that report, AFP and ALB mRNAs were found to be present in distinct cell populations of developing rat brain and kidney tissues. Cellular transcripts were localized to the cytoplasm during fetal/postnatal life and were found in the nucleus at 3–5 weeks, suggesting that post-translational mechanisms are involved in the control of stage-specific AFP/ALB gene expression.
|
AFP and Homeodomain Proteins
|
|---|
The discovery of the homeobox transcription factor gene family remains a hallmark discovery in embryonic development and differentiation (for review, see Ref. 16). These embryonic inducers, first discovered in insects, are now known to exist as homeotic proteins in nematodes, plants, yeast, rodents, and human beings. Homeoproteins serve to direct and control pattern/positioning body development in embryos regarding anterior/posterior, trunk-thorax segmentation, dorsal/ventral axis, body/cell polarity, neural tube formation, and caudal/gut formation. For example, birth defects are often homeotic transformations resulting in developmental abnormalities in which one part of the body develops in the likeness or dissimilarity of another. Pattern formation in the embryonic germ layers usually involve a network of feedback systems between intrinsic factors of gene expression in developing precursor cells and extrinsic signals exerted from the surrounding embryonic matrix environment (16, 17). The homeotic proteins frequently modulate or mediate inductive pathways that partition early axial germ layers into structures or segments with distinct regional identities. These morphogenetic processes are then linked to the terminal differentiation stages of that particular germ-layer derivative. Examples of the homeodomain transcription proteins include Pou, Crumbs, Hox, Antennepedia, Wnt, Sonic Hedgehog, Notch, and Pax (16). Human AFP itself appears to contain amino acid sequence identity/similarity stretches to the homeodomain proteins on all three of its domains (Table 1
).
View this table:
[in this window]
[in a new window]
|
Table 1. Genbank Amino Acid Sequence Matching of Human Alpha-Fetoprotein (HAFP) Domains 1, 2, and 3 with Conserved Sequences from Various Homeodomain Protein Segments.a,b
|
|
Since the homeodomain proteins are present during early embryogenesis, it would seem reasonable that AFP might display short homeodomain sequences in molecular mimicry of these pattern-regulating proteins. For example, mutations in the Pax-3 domain result in central nervous disorders relevant to AFP such as anencephalies and spina bifida, in addition to abnormalities associated with neural crest structures (16). In mammals, Pou domains are expressed during early embryogenesis in many regions of the developing brain including forebrain and nerve cord (Table 1
). Aside from binding the major groove of DNA, the Pou domain is required for homo-and heterodimerization of the Pou domain proteins. The Pit-1 gene of the Pou domain controls development of the anterior pituitary, and mutations of this gene display failure of adenohypophysis development. The Wnt gene codes for proteins that are expressed in the midbrain-hindbrain border, and mutations in this gene results in the absence of these brain regions. Finally, Crumbs protein mutations have led to severe disorganization and degeneration of ectodermally derived embryonic epithelia. Thus, it may be more than a coincidence that HAFP segments share short amino acid sequence homologies with the homeodomain proteins, which are endowed with embryonic body positional information.
|
AFP Levels in Urine
|
|---|
The detection of AFP in human pregnancy urine was first described by Smith and co-workers in 1971 (48); AFP was detected in the urine of hepatoma patients in that same year (49). Due to the lack of sensitive immunoassays, the samples were scored as positive only for highly concentrated urine specimens; however, AFP mass values were still considerably lower than in serum. Later in 1973, AFP was detected in the urine of pregnant and hepatoma-bearing rats, but not in normal adults, by means of Ouchterlony immunodiffusion (50). In the gestating rat, AFP is produced in the fetal liver and yolk sac, secreted into the fetal circulation and amniotic fluid, passaged into the maternal circulation via the placenta/allantois, and excreted in maternal urine. In the rat, AFP is reportedly excreted in the urine for at least 6 days postpartum (51). After intrauterine death and resorption of the rat conceptus, urinary excretion of AFP persists for several days, while remaining signs of pregnancy regress to nonrecognizable levels; nevertheless, AFP is still detected in the urine. Excretion of AFP into urine can also be detected in very early stages of chemically induced carcinogenesis, in rats fed carcinogenic diets containing carbon tetrachloride (52). In rats fed 3-methyl-4-dimethylaminoagobenzine, AFP in urine was detected in 90% of the experimental animals. Readers are further referred to a recent publication (53) by the author for an updated review of AFP in pregnancy biological fluids including urine.
|
AFP in Reproductive Fluids
|
|---|
Alpha-fetoprotein has long been detected in the biological fluids of the reproductive tract. The mammalian oviduct secretes, by transudation and exudation, a variety of fluids that may influence multiple reproductive activities, including sperm function. Such reproductive fluids, which include cervical mucus, oviductal fluid, and follicular fluid, can facilitate sperm capacitation (aging/maturation), fertilization, zygotic cleavage, and early embryonic nourishment (54–56). Electrophoretic studies of oviductal fluids from rabbits, monkeys, and women have demonstrated that the mammalian oviduct secretes a variety of proteins such as ALB, ß-globulins, and AFP (54, 55). Earlier studies provided evidence that the production of proteins found in human oviductal fluid (HOF) is under hormonal control and that levels of such protein secretions correlate with the estrogen peak of the menstrual cycle (56). Since the proteins appear to be estrogen regulated within the oviduct, they may have functional similarities while differing in molecular mass. The proteins derived from HOF have two sources: (i) proteins originating from serum transudation and (ii) those synthesized and secreted by the uterine tubal mucosa (55). One such secreted protein, designated as "human oviductin-1" (HOV-1), displayed a molecular mass of 54 kDa and a PI of 4.5 and contained a carbohydrate moiety (56). Subsequent studies in rabbits detected a similar protein in the post-ALB range of an electrophoregram and showed evidence of induction by 17ß-estradiol while demonstrating Periodic Acid Schiff (PAS)–positive staining (55, 56). The 54-kDa human oviductal protein contained 8% carbohydrate (vs. 3%–5% in SAFP), showed isoform heterogeneity, and did not react with anti-ALB antisera. The human HOV-1 protein was of nonserum origin and did not cross-react with extracts of human liver, kidney, or ovary, but it did react against human oviductal tissue. Studies using tritiated leucine confirmed that the protein was a secretory product of oviductal origin. Furthermore, reducing SDS gels indicated a molecular mass of 54 kDa, while the same protein displayed a molecular mass of 66 kDa in non-reducing gels. Thus, the protein appeared to be a heavily glycosylated, AFP-like molecule. At the time of publication, the authors had not ruled out immunological cross-reactivity with -ALB (human afamin).
In subsequent studies of the HOV-1 molecule, fresh donated human sperm were incubated with (i) a mixture of HOF-specific proteins or (ii) HOV-1. Using indirect immunofluorescence, the investigators studied the ability of the HOF proteins to bind to the human sperm (57). While the mixture of HOF proteins bound diffusely over the entire surface of the sperm, cell HOV-1 binding was restricted to only its head region. The authors stated that the HOV-1 protein acted as an acrosome-stabilizing factor, serving to prevent premature acrosome activation. However, they were unable to differentiate between capacitation and the acrosomal reaction in their sperm populations. In subsequent purifications of HOV-1 protein, determination of the amino acid and carbohydrate compositions of HOV-1, together with isofocusing and Western immunoblot analysis using monoclonal anti-HAFP antibodies, confirmed that HOV-1 was antigenically identical to HAFP (58). However, its molecular mass suggested that HOV-1 is a truncated form of AFP (17). Thus, HOV-1 was a globular, non-collagenous protein with carbohydrate attachment via an N-glycosidic linkage between N-acetyl-glucosamine and asparagines, as is the case for HAFP. The authors proposed that HOV-1, like AFP, was secreted into the luminal spaces of the oviduct as a result of AFP biosynthesis by mucosal cells of the Fallopian tube (58–60). Since total HOF has been shown to prolong sperm survival, AFP as a constituent protein may serve to mediate sperm survival and motility. In the same study, HOV-1 was found in the HOF of a patient during the periovulatory period (59, 60). Interestingly, AFP has been detected in the follicles of rats prior to and during the ovulatory cycle (15). The detection of HAFP in postovulatory stages of the human ovum lends credence to a role for AFP in reproductive-tract physiology (28, 29). Overall, female reproductive tract secretions in vitro have been shown to maintain follicular fluid-induced hyperactive sperm motility while simultaneously decreasing the response of the acrosomal reaction so as to prolong sperm viability/function (58, 60). It is germane to this discussion that AFP binds to the male antifertility drug gossypol, which inhibits estrogen binding to AFP (61). Thus, AFP may serve a role in the fertilization process that is not yet fully understood or appreciated (see below).
In a recent study, a "knockout" of the AFP gene in mice was reported by a group of investigators in Belgium (62). They used gene targeting to show that AFP is not absolutely required during embryonic development, at least in the mouse. The AFP-null embryos developed normally, and transplanted embryos homozygous for the AFP-null trait developed normally, in an AFP-deficient microenvironment. However, while mutant homozygous adult male offspring appeared to be viable and fertile, AFP-null females were infertile. These investigators determined that the female infertility defect was related to a dysfunction in the hypothalamic/pituitary-ovary axis, as previously proposed (63). Although AFP does not seem to be required during development, due to compensatory actions of ALB and -ALB, it plays a critical, nonredundant role in determining the future fertility of female offspring in the mouse. In light of the reported presence of AFP in the developing follicles of the ovary, in HOF, and in the developing and newborn brain, AFP may somehow be involved in the maturation and programming of the positive feedback exerted by ovary-derived estrogen on brain LH and FSH levels. Indeed, earlier studies in rats showed that the postnatal decline in SAFP was strongly linked to the progressive increase in tissue-to-serum ratios of estradiol E2 during the first 5 weeks of life (64). During these postnatal weeks, AFP and FSH levels declined, and an LH surge developed in a progressive, stepwise fashion (65). If AFP is injected during this period, it causes a significant rise in plasma FSH, low levels of free E2, and a delay in the onset of puberty. These symptoms mimic the polycystic ovary syndrome in humans, in which anovulation persists in the presence of reduced FSH levels, concomitant with increased LH levels. The Belgian investigators suggested that the lack of AFP in null mice might mimic the polycystic ovary syndrome in humans and that further study was required to elucidate these issues (see Ref. 76).
|
AFP Antibodies During Development
|
|---|
The effect of heterologous or autologous antibodies to rat AFP (RAFP) in living systems has been extensively investigated. The presence of anti-autologous AFP antibodies in pregnant rodents was effective in interrupting development in a significant proportion of embryos (66–68). Using rabbit antiserum to rodent AFP, it was shown that the number of litters born was smaller, the birth weights were lower, the occurrence of stillbirths was higher, and the postnatal mortality was increased. Earlier studies had also documented a fetotoxic effect of heterologous anti-AFP antibodies in pregnant rats (66), whereas passive immunization of mice to AFP was only partially effective in fetal growth disruption (69, 70).
Antibodies to MAFP have likewise been employed as investigational probes and tools in the study of possible physiological roles for AFP. As previously observed in the chicken and in the rat, heterologous antibodies to AFP administered to pregnant mice resulted in developmental arrest, congenital abnormalities, hemorrhagic placental lesions, and fetal wastage (71–74). The cause of fetal death was linked to an antibody-induced anaphylactoid (Schultz-Dale) reaction of the uterus and antibody-mediated inflammatory lesions at the placental interface (73). It was readily apparent from these studies that breakdown of immune tolerance to AFP during pregnancy was hazardous to the completion of a full-term pregnancy. In a similar fashion, immune tolerance can be broken, as reported in pregnant rats, mice, and rabbits (75, 76). Finally, the administration of heterologous antibodies to MAFP in neonates in their first week of life resulted in characteristics resembling ovarian androgenization (i.e., polycystic ovaries), as discussed previously (76).
|
Uptake of AFP by Developing Cells
|
|---|
The uptake of AFP (endocytosis) was first observed in fetal cells using both in vitro and in vivo systems (77–82). The uptake or endocytosis of AFP was reported in cells derived from ecto-, meso-, and endodermally derived tissues of the embryo and fetuses of both birds and mammals (83). Subsequent studies showed the ability of cells of muscle origin to internalize exogenous AFP during fetal, neonatal, and tumor development (84, 85). The AFP is taken up by cells undergoing differentiation, whereas neither undifferentiated nor fully differentiated cells readily incorporate AFP. Furthermore, experimental evidence indicated that the entry of AFP into cells occurs by a process of receptor-mediated endocytosis (85–88). Following uptake, the AFP was first detected in clathrin-coated pits of the cell membrane; it thereafter progressed to vesicles, multi-vesicular bodies, and the trans-Golgi network surrounding the nucleus (89, 90). A series of subsequent studies further showed that AFP uptake also appeared in fetal cells of various species and origins, such as human rat, mouse, chicken, and baboon (38, 85–87). These studies and others strongly suggested the presence of an AFP cell surface receptor that constitutes an integral part of the cell membrane (88–93). Overall, the expression of AFP receptors reflected the acquisition and/or presence of phenotypic properties specific to immature, incompletely differentiated cells.
|
The AFP Cell-Surface Receptor
|
|---|
A study demonstrating the incorporation of AFP by MCF-7 human breast cancer cells was reported in 1984 (90); later that year, the same research group presented evidence for the existence of a specific membrane receptor for AFP on the surface of MCF-7 cells (91). Scatchard binding analysis of the breast cancer cell receptor revealed the presence of at least two high-affinity binding sites, with KD of 10–8 and 10–9 M, and N = 2,000 and 135,000 sites per cell, respectively. A further study demonstrating the incorporation of AFP by human monocytes was reported in 1992 (92). Subsequent studies in using human monocytes, ß-lymphoma cells, and T-leukemic cells confirmed the findings of the earlier binding studies and revealed the presence of two to three specific binding sites, with KD values ranging from 10–6 to 10–10 M, and numerous binding sites per cells (88, 89, 91, 92). Furthermore, the receptors detected on fetal, lymphoid, and tumor cells could be distinguished from the AFP and ALB asialo-receptors, previously described on the surface of vascular endothelial cells, that are involved primarily with blood clearance activities (93).
The AFP receptor was first isolated and characterized as a specific cell-surface receptor on human monocytes in 1992 (92). After that, a human breast tumor membrane receptor that bound AFP was reported (94). This latter study described monoclonal antibodies to a human mammary adenocarcinoma membrane receptor that inhibited AFP binding to cell membranes; together these studies aided in the isolation of 62- and 67-kDa PAS-reactive receptor components linked with a higher-molecular-mass (~200 kDa) molecular entity (94, 95). In 1997, a study confirmed the presence of a heterotrimeric protein complex composed of 65-, 130-, and 185-kDa molecules, yielding a multimer complex with a total molecular mass of 250–300 kDa (96). The trimeric protein complex was glycosylated, exhibited a high carbohydrate-to-protein ratio, and was susceptible to disulfide-bond cleavage. The trimeric complex, comprised of three noncovalently bound subunits, bound AFP (and to lesser extent, ALB) with a KD of 2–4 x 10–10 M. The complex could be histochemically localized on the surfaces of cells from both embryonal and tumor tissues (breast, hepatoma), as had been reported in previous studies (79, 86, 89, 97). Thus, uptake and binding analyses of the AFP receptor provided the basis for a concept of an autocrine AFP/AFP-receptor system described in human monocytes, quiescent T-lymphocytes, activated T-lymphocytes, and activated T-lymphocyte systems (84, 92, 97). This autocrine loop of AFP-growth stimulation was further confirmed in tumor cell lines and was found to be intimately linked to the cell endocytotic delivery of ligands, notably fatty acids (98).
|
AFP: Fatty Acid Binding During Development
|
|---|
The presence of fatty acids on HAFP was first reported in 1978, enumerating six fatty acids: palmitic, stearic, oleic, linoleic, arachidonic (AA), and docosahexaenoic (DHA) acids (99, 100). At that same time, the binding to fatty acids and tryptophan to AFP from fetal pigs was reported in a study by Scandinavian investigators (101). Similar results were found with BAFP, which also bound fatty acids (102). During the study of binding to estrogens, it was revealed that fatty acids in serum were strong competitors of estradiol binding to RAFP (103). The fatty acid levels of RAFP derived from fetuses and pregnancy and hepatoma sera showed that AA and DHA constituted 33% of the total bound fatty acids; the highest levels were found in fetal sera (104). These polyunsaturated fatty acids (PUFAs) displayed a higher association constant (KA = 10–6 M) than did the saturated (KA = 10–4) compounds (105); RAFP was shown to possess one high-affinity binding site and multiple (12, 13) low-affinity sites for PUFA, while HAFP presented only three binding sites, all of equivalent affinity (106–109). Similar patterns of fatty acid binding were found in AFPs of pig, bovine, rat, and human origin (101, 106, 110–113).
During development in the rat, the major fatty acids bound to AFP were DHA and AA, as assayed either from fetal serum or whole fetuses; palmitic and oleic acids were mainly bound by ALB (108, 114). Amniotic fluid AFP contained fewer fatty acids (0.8 mol/mol protein) than did fetal SAFP (1.4 mol/mol protein); especially noticeable was a reduced amount of DHA (107, 108). Levels of serum DHA bound to AFP decreased quickly after birth to a minimum at 8–10 days, which represents a period of maximal accumulation of DHA by the brain and breast-derived colostrums, accompanied by a decreased uptake by liver. It was further reported that adult rat ovarian extracts inhibited the binding of estrogen to RAFP (115). The active component of ovarian extracts, AA, also served as a strong inhibitor of E2 binding. Since AA is a direct precursor of the prostaglandins, AFP was investigated as a regulator of prostaglandin metabolism and synthesis. Interestingly, the prostaglandins showed no affinity for binding to the AFP molecule; however, it later was demonstrated that both prostaglandins and lipoxygenase product formation were reduced when cells were maintained in the presence of AFP (115). Finally, it was shown that, when AFP was used as a protein carrier, the amount of hexaene fatty acid derivatives of linoleic acid recovered in hepatocytes was reduced by up to 50% (109, 111). This effect was explained by an efflux of hexaene derivatives from cells, with AFP as the causative agent.
Previous studies have shown that AFP also plays a role in the intracellular delivery of PUFAs into developing cells. Labeled AFP was found to enter the cells via coated pits and receptosomes and to move to tubular elements of the transreticular portion of the Golgi apparatus (105, 109). Fatty acids bound to AFP are transferred into cells within 5 mins at 37°C, and, following fatty acid release, AFP can be recycled back across the cell surface. Data revealed that the fatty acids (i.e., AA and DHA) bound to AFP were mainly incorporated into cell phospholipids and that 25%–40% of the incorporated AFP was secreted and released undegraded after 60 mins of incubation. The AFP first binds to an AFP cell-surface receptor, and then the fatty acid is endocytosed and transferred within the cell by a specific fatty acid–binding protein. The AFP-mediated uptake of fatty acids has also been demonstrated in human T-lymphocytes and in cancer cells (112, 113, 116).
Human AFP has been reported to bind long-chain polyunsaturated fatty acids (AA and DHA) with high-affinity KA108 M, n 
1–3 (103, 104, 110, 116); RAFP also binds long-chain polyunsaturated fatty acids, with one high-affinity binding site and 10–12 low-affinity sites per molecule (103, 107, 114, 117). The E2-binding sites on RAFP are subject to competition inhibition by oleic and linoleic acids and particularly by AA and DHA. These latter two fatty acids show higher levels of binding to fetal SAFP than to maternal on hepatoma AFP. Thus, it is conceivable that the binding sites for E2 and fatty acids overlap (107). Ligand binding to MAFP has also been studied in detail by many investigators particularly in regard to estrogenic steroids and fatty acids (118–120); E2, AA and diethyl-stilbestrol (DES) all bind to MAFP with decreasing affinities, respectively, KA 0.8 x 108 M–1 (N = 0.3), KA 0.3 x 107 M–1 (N = 4–5), and KA 0.2 x 107 M–1 (N = 0.7; Ref. 15), and MAFP preferentially binds long-chain fatty acids (C22:4 and C22:6) that serve as efficient inhibitors of both E2 and DES. Murine ALB, in contrast, binds virtually no estrogen and shows a higher affinity for AA.
During pregnancy and early infancy, the biological role of HAFP in binding and trafficking of PUFAs is now well established (117, 121–123). Human AFP is known to both regulate and facilitate the entry of fatty acids (especially AA and DHA) into cells undergoing growth and differentiation (102, 112, 113). Human AFP was found to bind 16%–42% of the DHA in total fatty acid content, whereas fetal ALB bound only 4% (100, 102, 104). Human AFP reversibly binds DHA with high affinity (KA = 2 x 107 M–1) and transports the fatty acid mainly during the fetal, perinatal, and neonatal periods (89, 96, 102, 112). Human AFP itself undergoes transplacental passage to the maternal circulation and tissues (117, 123). Mammalian AFP in vitro in fetal hepatocytes has also been shown to enhance the intracellular conversion of saturated to unsaturated fatty acids (110, 118, 123).
The precise binding location on the AFP molecule is known only for a few of its ligands. For example, one major fatty acid–binding site for long-chain fatty acids has been documented to lie between residues 210 and 227 on HAFP Domain 2 (113). Lysine residues, especially Lys-223, appears to be essential for the fatty acid binding at this site. Studies employing Scatchard binding/saturation analysis have previously demonstrated that at least three potential ALB binding sites (KA = 10–7 M–1, N = 3) exist for the polyunsaturated fatty acids (i.e., AA and DHA; Refs. 106, 120). If AFP is similar to ALB in this regard, then most probably one major fatty acid–binding site exists on each of the three domains. The remaining two fatty acid–binding locations are speculated on the basis of Genbank-derived amino acid comparisons to fatty acid–related proteins. One such example of a potential HAFP Domain 1 site, residing at residues 36–69, shows an amino acid homology to fatty acid synthetase (38% identity, 32 amino acids). Indeed, when Domain 1 residues 40–60 on HAFP are compared for sequence/identity matching to the documented fatty acid–binding site on Domain 2, several interesting points emerge (Table 2). First, it can be seen that each of the 20 amino acid sequence stretches contains three or more lysines, which are essential for fatty acid binding. Second, one or more lysines are located at the amino-terminal side of the sequence amino acid stretch. Third, the crucial amino acid for complexing to the carboxy group of the bound fatty acid has been identified as the lysine (see asterisk) positioned 13–14 amino acids from the amino-terminal lysine. These three criteria from Domain 2 fit the amino acid sequences on both Domain 1 (residues 41–62) and Domain 3 (residues 418–437), compared to Domain 2 (residues 208–227). Both amino acid sequences stretches on Domains 1 and 2 demonstrate 60%–65% identity/similarity matching to the documented fatty acid–binding site on residues 208–227 of the second domain. The third AFP fatty acid–binding site, which resides on Domain 3, apparently overlaps or lies directly adjacent to the documented estrogen-binding site on RAFP, according to previous competitive-binding reports (75, 168; Table 2, part B).
View this table:
[in this window]
[in a new window]
|
Table 2. Proposed and Documented Amino Acid Sequences for Fatty Acid– and Estrogen-Binding Sites on the Human Alpha-Fetoprotein (HAFP) Moleculea
|
|
Binding, spectral, and immunological studies have demonstrated that conformational changes in the tertiary structures of rodent and human AFPs can be induced by a high free fatty acid (PUFA) environment (121). In contrast to the PUFAs, saturated fatty acids had no effect on the tertiary properties of all AFP species tested. Human AFP measured by RIA/ELISA in fetal, hepatoma, and cord serum showed reductions in AFP levels of 80%, 50%, and 5%, respectively, in these fluids, following the change in tertiary structure. Furthermore, a transient rise in plasma PUFA levels led to a loss of AFP immunoreactivity in 21- and 28-day-old rats (122). In all cases, the fatty acids induced a rapid and reversible conformational change in the RAFP molecule. At the placental interface, PUFAs were found at low levels in maternal blood but at high levels in intervillous/umbilical blood vessels; the AA and DHA concentrations were also highest here (117, 123). The conformational state of AFP was found to differ in the intervillous spaces, suggesting that AFP was heavily loaded with PUFAs at the feto-maternal interface (123). High concentrations of PUFAs stimulated E2 binding and inhibited progesterone (PG) binding, suggesting that PUFAs modulate the steroid hormone message by amplifying the E2 signal and damping the PG signal (119).
|
Interactions of AFP with Estrogen During Development
|
|---|
Although HAFP was initially reported to bind few, if any, estrogenic steroids (124), additional evidence now indicates that HAFP (obtained by butanol extraction of fetal tissues) can bind to immobilized estrogen columns, suggesting the existence of a partially occluded binding site (125). Previous data had implied that <1.0% of the HAFP population was capable of such binding (126, 127) and that the butanol either induced a conformational change or removed a bound ligand that normally interferes with the estrogen binding site (127). This is because AFP is able to bind alcohols; furthermore, HAFP was found to bind longer-chain alcohols more strongly than it did low-carbon chain alcohols (butanol > propanol > ethanol > methanol; Ref. 128). The reported alcohol- and drug-binding affinities of HAFP displayed a KA of 106 M–1 and binding site numbers 1.2–3.4 (129, 130). Thus, HAFP is capable of binding butanol, which may have outcompeted a bound agent to which HAFP had been complexed. The use of purified RAFP also has provided a means by which to study the in vitro binding and in vivo transport of steroids by this fetal protein (131–135). In a study employing RAFP fractions on lectin columns, low-carbohydrate forms bound E2 with high affinity (KA = 108 M) and 0.5 sites per mole of protein (131). It has been demonstrated that RAFP binds strone (2.74 x 107) with higher affinity than it does E2 (KA = 1.83 x 107 M–1; Refs. 137–140), contrast to the behavior of MAFP (see below); MAFP binds with high affinity to E2 (KA = 0.4 x 108 M–1, N = 0.3; Refs. 131, 134, 135). The carbohydrate (CHO) prosthetic group of MAFP plays a significant role in E2 binding, in that a fraction devoid of CHO groups results in the appearance of very high affinity binding sites (~109 M–1). As expected, mouse embryonic serum has a very high affinity for both (E1) and E2, peaking at 15–18 days of gestation, although little E1 is initially bound at Day 12 (135). In fact, the major estrogen-binding component in mouse amniotic fluid is AFP (136), and it serves as a secondary E2-binding agent in rodent uterine tissues (137). Further serum elevation of MAFP in adult mice can be induced by intraperitoneal injections of either estriol or E2 (100–1,000-µantities; Refs. 138, 139). Because MAFP binds with such avidity to both E2 and E1, estrogen-based affinity chromatography has served as a useful means by which to purify the protein (126, 127).
Physiological studies on AFP-estrogen interaction have been undertaken in a variety of animal models; RAFP was shown to inhibit the formation of water-soluble metabolites of E2 and E1, when microsomes from rat liver were incubated in the presence of NADPH, and to regulate the activity of 17ß-hydroxy steroid dehydrogenase in vitro (140). Alpha-fetoprotein has also been associated with the delay in the onset of puberty in postnatal rat pups (141). Finally, the injection of AFP during the prepubertal period in rats has resulted in a decreased number of primordial and primary follicles in the ovary (142). The injected RAFP was localized to the zona pellucida of follicles undergoing atresia.
The estrogen-binding interface on rodent AFP has been reported to occupy a region between residues 423 and 444, forming an -helical segment that lies adjacent to a potential ß-sheet/turn structure extending from residues 445–480 (143, 144; Table 2). The former site (residues 423–444) represents a major hydrophobic binding pocket on HAFP that binds few estrogenic steroids, in contrast to rodents (144, 145). In humans, both regions (residues 423–444 and 445–480) display overlapping binding sites for fatty acids, DES, protease inhibitors/substrates, retinoids, warfarin, coumarin, phenylbutazone, pyrazolic drugs, and anthranilic acid (128, 129, 145). Previous competitive-binding studies employing RAFP had already determined that estrogen was bound on RAFP Domain 3, together with retinoids and fatty acids (146, 147).
Both RAFP and MAFP avidly bind estrogens (E2, E1) with high Kd values (10–8 M–1); however, HAFP binds few of these estrogens, as discussed above. One high-affinity region (Site I) and one low-affinity region (Site II) have been reported for AFP (147). Through the use of recombinant technology involving human/rat hybrids, the primary estrogen binding Site I in RAFP has been determined to lie between residues 419 and 436, corresponding to HAFP residues 424–439 (Table 2). The five amino acids crucial to the binding on RAFP are glycine-425, methionine-427, isoleucine-430, alanine-432, and threonine-432 (144). These amino acids match precisely to the MAFP residues glycine-428, methionine-430, isoleu-cine-433, alanine-434, and threonine-435. In contrast, three of five of these amino acids in HAFP, a protein that binds only scant amounts of estrogen, show substitutions, namely, glycine 
arginine, isoleucine threonine, and serine alanine. Interestingly, 8 of the 15 amino acids in this HAFP sequence stretch display substitutions, when compared to the rodent AFPs. Moreover, when the amino acid sequences of human and rodent AFPs are compared to an estrogen-binding region on the human estrogen receptor (HER; residues 419–431), it can be seen that four of the five amino acids crucial for estrogen binding to rodent AFP are matched and aligned to those of HER. As displayed in Table 2, the remaining hydrophobic amino acids in human and rodent AFPs are also present in HER, namely, leucine and valine. It is obvious that in the HAFP molecule, multiple (five) alanines have replaced many of the crucial amino acids present in both RAFP and HER. However, a single alanine that is retained in all species is probably significant. Thus, this 15-amino acid region on the various AFPs seems to represent the high-affinity binding site for rodent and to a lesser extent human AFP.
For the second AFP estrogen-binding region, Site II, a different scenario emerges (Table 2). This segment has been shown to bind E2, through the use of AFP-derived peptides, and it probably serves as a secondary, lower-affinity binding site in both HAFP and rodent AFPs. As seen above, HAFP appears to largely lack crucial amino acids in the primary binding site, but it has retained most of these residues in the secondary binding site. The latter, secondary site is highly hydrophobic (leucines, isoleucines, alanines) in all three AFP species (HAFP, RAFP, MAFP) and in HER (five leucines, two isoleucines, one alanine). The various AFP molecules display a common cysteine, unlike HER, but otherwise they show similar arginine and histidine positionings. In both the AFPs and HER, the composition and placement (position) of leucines, isoleucines, arginines, and histidine as the dominant amino acids are shared. The HER segment is devoid of glycines and cysteine in its secondary binding site, in contrast to the rodent and human AFPs. Thus, the lower binding affinity of estrogen-binding Site II (10–5 M–1) as compared to Site I (10–8 M–1) may be ascribed to the predominant leucine/isoleucine hydrophobic composition of Site II since Site I contains mostly methionine, threonine/serine, and alanine. Cysteine does not seem to be required for estrogen binding at this secondary site. In the native state of the HAFP molecule, these cysteines are disulfide bonded. While AFP Site I is primarily committed to estrogen binding/ligation, Site II is also thought to serve as a docking site for proteins of the heat-shock protein (HSP) family, such as HSP-70, and HSP-90, as in HER (148). Such docking sites for the HSPs are also known to be involved in protein folding/unfolding activities (149).
|
AFP and Insulin/Carbohydrate Chemistry
|
|---|
Insulin is the most potent anabolic hormone known, promoting synthesis and storage of carbohydrates, lipids, and proteins (150). The interaction of insulin with AFP in developing mammals has previously been described (151); however, effects of mammalian AFP in the chick embryo have not been reported until now. In pregnant mice, a single pulse of insulin at Day 13 of gestation produced elevated maternal SAFP levels and increases fetal mass, while long-term treatment with insulin yielded lower levels of maternal SAFP and lower fetal mass per animal (152). Human AFP has been employed in the clinical laboratory as a gestational age–dependent fetal defect marker and is presently utilized as a screening agent for neural tube defects (153) and aneuploidies (154). The neural tube defect screening procedure identifies a distinct subpopulation of women who exhibit insulin-dependent diabetes during pregnancy. These pregnant women display maternal SAFP concentrations that are 20% lower than those of age-matched controls (153). In avian models of insulin teratogenicity, there is seen a form of caudal vertebrae rumplessness in early development that resembles the mammalian syndrome of neural tube defects (155–158). Closure defects of the neural tube in chick fetuses are accompanied by 
Top
Abstract
Introduction
