Experimental Biology and Medicine 226:377-408 (2001)
© 2001 Society for Experimental Biology and Medicine
MINIREVIEW
Alpha-fetoprotein Structure and Function: Relevance to Isoforms, Epitopes, and Conformational Variants
Gerald J. Mizejewski1,
Division of Molecular Medicine, Wadsworth Center, New York State Department of Health, Empire State Plaza, Albany, NewYork 12201
 |
Abstract
|
|---|
Mammalian alpha-fetoprotein (AFP) is classified as a member of the albuminoid gene superfamily consisting of albumin, AFP, vitamin D (Gc) protein, and alpha-albumin. Molecular variants of AFP have long been reported in the biomedical literature. Early studies identified isoelectric pH isoforms and lectin-binding variants of AFP, which differed in their physicochemical properties, but not in amino acid composition. Genetic variants of AFP, differing in mRNA kilobase length, were later extensively described in rodent models during fetal/perinatal stages, carcinogenesis, and organ regeneration. With the advent of monoclonal antibodies in the early 1980s, multiple antigenic epitopes on native AFP were detected and categorized, culminating in the identification of six to seven major epitopes. During this period, various AFP-binding proteins and receptors were reported to inhibit certain AFP immunoreactions. Concomittantly, human and rodent AFP were cloned and the amino acid sequences of the translated proteins were divulged. Once the amino acid composition of the AFP molecule was known, enzymatic fragments could be identified and synthetic peptide segments synthesized. Following discovery of the molten globule form in 1981, the existence of transitory, intermediate forms of AFP were acknowledged and their physiological significance was realized. In the present review, the various isoforms and variants of AFP are discussed in light of their potential biological relevance.
Key Words: alpha-fetoprotein • isoforms • epitopes • variants • conformation • structure • function
|
Introduction
|
|---|
Alpha-fetoprotein (AFP), a tumor-associated fetal protein, has long been employed as a serum fetal defect/tumor marker to monitor distress/disease progression (1–3). Similar to albumin, serum AFP is known to bind and transport a multitude of ligands such as bilirubin, fatty acids, retinoids, steroids, heavy metals, dyes, flavonoids, phytoestrogens, dioxin, and various drugs (4, 5). High concentrations of some of the hydrophobic ligands (i.e., fatty acids and estrogens) have been reported to induce a conformational change in the tertiary structure of AFP (see Ref. 6 for review). Altered serum AFP levels have been observed concurrent with aberrant growth manifestations, but it was usually assumed that these levels were a coincident effect rather than the cause of such changes. Although AFP may not be the direct cause of the altered growth manifestations observed in birth defects and cancer, it is conceivable that some shock/stress-induced conformational (variant) forms of this fetal protein may influence or contribute to such events. In the last decade, reports have emerged that some of these AFP forms may serve as dual regulators of growth, capable of both enhancement and inhibition (7, 8).
Molecular variants of mammalian AFP have been reported in the scientific literature since the 1970s. Initially, some of the variant forms were attributed to carbohydrate microheterogeneity and alterations in isoelectric points (9, 10). Additional AFP isoforms were genetic variants and lectin glycoforms that were detected and isolated by isoelectric focusing, electrophoresis, and chromatography (11, 12). Further isoforms were detected following high-pressure liquid chromatography (HPLC), and from lectin, heavy metal, and hydrophobic solid phase separation methods. The advent of monoclonal antibodies further permitted the detection and analysis of the epitopic domains and subdomains that comprise the total antigenic sites of this fetal protein (13, 14). Finally, the recent discovery and characterization of the molten globule form (MGF) have provided a new level of understanding regarding the various intermediate transition forms of proteins such as AFP (15).
Presently, there is no catalogue of the multiple isoforms and epitopic and conformational variants of AFP in the literature. Hence, the aims of the present review are 3-fold. First, the myriad of molecular forms of AFP found in human biological fluids and tissues, and in animal models, will be presented. Second, the various AFP isolates and isoforms that can be induced and purified in the laboratory from in vivo sources or from in vitro cell culture models will be described. Third, the relationship of AFP's biological activity to both naturally occurring and experimentally induced variant forms of AFP will be presented. The multiple names and classifications of the molecular variants of AFP in this rapidly advancing field justify a review that attempts to link structural findings with physiological effects. Finally, this review is intended to serve as an extension and update of a review on AFP as a biologic response modifier previously published in this journal (7). The present review is not intended to be all-inclusive for AFP, therefore the reader is directed to consult earlier reviews. (16, 17).
|
The Albuminoid Gene Family
|
|---|
AFP is classified as a member of an albuminoid gene family, which consists of four members to date: albumin (ALB), vitamin D-binding (Gc) protein (DBP), AFP, and alpha-ALB (
ALB), termed afamin in humans (18, 19). This family is structurally characterized by cysteine residues that are folded into layers that form loops dictated by disulfide bridging, resulting in a triplet domain, U-shaped molecular structure (see Fig. 1
). The three domains of these gene family members have been confirmed by X-ray crystallography (20, 21) (Figs. 1 and 2). The ALB gene family members display structural similarities, homologous amino acid sequence stretches, and similar cysteine disulfide bridge clusters (Fig. 1 and Table I). In humans, the four albuminoid genes lie in tandem on chromosome 4 within the 4q11-q22 region, encompassing 15 exons and 14 introns (22). DBP alone is truncated in the third domain (see Fig. 1) and contains only 13 exons, which results in a protein with a smaller molecular mass (23). The newest member of this gene family, -ALB, was discovered in both rodent and human, and was cloned in the last 6 years (24, 25). All gene members are capable of ligand/carrier transport function, but display a vast array of other functions, including chemotaxis, oxygen free radical scavenging, esterase activity, leukocyte adherence, copper stimulated lipid peroxidation, and fatty acid, heavy metal, and actin binding, among others (26–28). Although the function of the recently discovered ALB remains obscure, it may play a role (ligand binding and immunoregulation) similar to its counterpart in lower vertebrates, a 74-kD ALB-like molecule found in fishes and amphibians (29, 30).
View larger version (32K):
[in this window]
[in a new window]
|
Figure 1. Molecular configurations of HAFP (left), human ALB (center), and human Gc DBP protein (right) based on the predicted secondary structures (See Ref. 117). Stars indicate predicted beta-turns and branched dark squares signify the sole carbohydrate side chain of HAFP. The dashed arrow indicates the proposed ``hinge'' region (lack of disulfide bridging) that is present in AFP, but absent in the other two gene family members. Note the truncated third domain of the Gc protein, representing a lack (dashed line) of a carboxy terminal amino acid chain compared with AFP and ALB. This figure was derived from composite diagrams redrawn from Refs. 117, 22, and 6.
|
|
View larger version (35K):
[in this window]
[in a new window]
|
Figure 2. A three-panel diagrammatic mapping of the tri-domain structure of HAFP. (A) represents the cysteine-loop configuration, while (B) shows the electron dot contour mass map of HAFP. In comparison, (C) predicts an epitopic map of HAFP derived from 30 or more MoAbs (Refs. 14, 161). The precise epitope domain locations have been detected, but have yet to be localized. The three-panel diagram represents a composite redrawn from Refs. 6, 117, 21, 14, and 161.
|
|
View this table:
[in this window]
[in a new window]
|
Table I. Biochemical and Structural Properties of the Four Known Protein Members of the Human Albuminoid Gene Family
|
|
Genetic Variants.
The genetic variants of mammalian AFP have been studied primarily in the rodent and to a lesser extent in humans (Table II). However, it is in mouse and rat that multiple gene expression has been described in greatest detail, including phase-specific expression of AFP mRNAs and their differential regulation during fetal and adult life, liver regeneration, and tumorogenesis (31–34). Although all of the AFP mRNA variants have been translated into proteins in vitro, not all have been detected in vivo. The major fetal- and tumor-derived AFP mRNA consists of a 2.2-kb transcript that translates to a 69,000 to 73,000 molecular form in human and rodents, depending on its carbohydrate content (35, 36). The fetal rat liver primarily expresses the 2.2-kb transcript, but produces a 1.7-kb moiety that translates into a 50- to 65-kD protein that appears in the last quarter of the rat gestation period (37, 39) (Table II). In the neonatal rat liver, a 1.5-kb message (in addition to the 2.2 and 1.7 kb) is detected, which gives rise to a 48-kD protein present for only 4 to 8 weeks following birth. In contrast, the adult liver AFP mRNA transcripts contain only traces of the 2.2-kb moiety with low amounts of the 1.7- and 1.5-kb transcripts and a more recently detected 1.35-kb transcript (37 kD)(39, 40). This latter 1.35-kb form lacks one-third of the amino terminal first-domain of rat AFP (41, 42). Finally, the hepatoma and liver regeneration forms of rat AFP display largely the 2.2-kb form and lesser amounts of the 1.7-kb variant (35, 36).
View this table:
[in this window]
[in a new window]
|
Table II. Expression Patternsa of Encoded Rat Alpha-Fetoprotein (AFP) mRNAs and Their Translated Proteins During Developmental and Adult Stages
|
|
The 1.35-kb AFP mRNA found in adult rat liver and kidney deserves special mention in that a similar transcript has been detected in human embryonal carcinomas transplanted into nude mice (43). In both species, a 1.35-kb form is retained intracellularly as a non-secreted form first reported in uterine and cancer cells (44, 45). This variant represented only 40% of the normal human AFP molecule translated from the 2.2-kb mRNA; it lacked the entire first domain and one-third of the second domain. It may be of interest that orphan steroid receptors, which dimerize with other nuclear receptors, have a similar truncated structure (46–48). Thus, truncated AFP forms could potentially bind to steroid receptors as previously proposed (49).
The adult form of human AFP, like the rat form, might also be derived from multiple RNA transcripts (i.e., 2.2, 1.7, and 1.6 kb). It is well established that the human form of AFP detected in most radioimmunoassays (RIAs) and enzyme-immunoassays (EIAs) represents the 69,000- to 70,000-kD (2.2 kb) polypeptide. It is highly probable that other AFP mRNA transcripts are present and are translated in serum and/or tissues, but are undetectable by present immunologic assays. Isolation and characterization of these predicted forms would be even more difficult since their concentrations would be vanishingly small (i.e., nanogram to picogram levels).
An interesting composite can be drawn when the genetic variants of the rat, and presumably human, are considered. As discussed above, the genetic variants of rat AFP mRNA consist of sizes ranging from 2.2 to 1.35 kb, representing translated proteins ranging from 72 kD down to 37 kD, respectively (41, 42). The smaller protein forms are found to be truncated from the amino-terminal end. This indicates that truncation eliminates domain 1 (amino-terminus) and a portion (up to one-third) of domain 2 of the rat AFP molecule as in the extreme case of the 1.35-kb form. As demonstrated in a previous phylogenetic review of AFP (6), it is the amino-terminal portion (domain 1) that distinguishes rat AFP from other mammalian forms (i.e., human AFP). In that review it was shown that domain 3 exhibits greater amino acid sequence identity among the various mammals than do domains 1 and 2. Domain 3 is also known to contain a major hydrophobic-binding site on human (H) AFP and a proposed dimerization motif (7, 49), now supported by experimental evidence (50). Domains 1 and 2 contain binding sites for both fatty acids and bilirubin; however, domain 2 also displays amino acid sequences related to cellular and extracellular matrix (ECM) adherence regions and bears sites for the carbohydrate attachment via asparagine. It may be more than coincidence that the truncated 1.35 kb (third domain) is retained within the cell as a nonsecreting form, since it bears no carbohydrate side chains (for blood circulation) and might be capable of heterodimerization to other intracellular proteins as do the truncated steroid nuclear receptors (47, 48). It is germane to this discussion that most commercial companies utilize as their capture antibody, monoclonal antibodies (MoAbs) directed to the first domain of AFP, and use for detection labeled secondary antibodies to the less specific second and/or third domain of AFP. If this is the case, most assay kits would contain antibodies directed largely against the 2.2-kb translated protein (72 kD); hence, the other truncated forms would never be detected.
Regulation and Expression.
The genetic regulation and expression of mammalian AFP is a highly complex area of study, and space does not permit adequate coverage in the present review. The reader is referred to recent reviews by Chiu and by Lazarevich (51, 52) for excellent detailed descriptions and overviews of this exciting field. Other papers using derivatives of DNA or mRNA for AFP as a genetic vector, as antisense, and as mRNA and cDNA isolates will only be briefly mentioned. For example, the presence or absence of HAFP mRNA in peripheral blood has been employed as a predictor of outcome in patients with hepatocellular carcinoma (53). Another study employed the AFP gene promoter for gene therapy against high AFP-secreting hepatoma cells using a retrovirus vector carrying a herpes simplex thymidine kinase gene (54). These same investigators then used a variant of the AFP gene promoter for gene therapy against a low AFP-producing hepatoma (55). In other studies, an AFP antisense strategy was employed to inhibit the growth of human liver cells in culture (56). Two further genetic immunization schemes were developed using AFP as a target for T-cell immune responses. In one study, dendritic cells were engineered to express AFP-produced potent AFP-specific cytotoxic T-lymphocytes directed against hepatoma cells (57, 58). Studies were also reported in which HAFP was employed as a recombinant reporter to target proteins to the plasma membrane of cells (59). Potential uses of this technique may include vaccine development, tissue engineering, genetic research, bioseparations, and disease treatment. This latter study served to show that high levels of monomeric and dimeric proteins could be targeted to the cell membrane by the proper selection of a transmembrane domain attached to a reporter protein (AFP). Also, reports of additional AFP gene promoters and repression of the AFP gene continue to emerge (60–62). Finally, an AFP transcription factor has also been described that is a nuclear receptor related to the Drosophila FTZ-F1 gene family (326).
|
Free and Bound Molecular Forms
|
|---|
As stated above, many genetic variants of AFP mRNA have been reported in both human and rodent. Some of these translated AFP products are present in both bound and free forms. The reader is directed to previous reviews by the author, which detail the various bound forms of mammalian AFP in both serum and tissues (7, 63). Although elusive and difficult to detect, AFP complexed to binding proteins has a documented history dating back to the original observation by Norgaard-Petersen in 1976 (64). Since that time, sufficient articles describing AFP-binding proteins and receptors have been reported to validate their existence and justify their place in the scientific literature (65–69).
Soluble Forms.
Some of the bound forms of AFP in solution can be detected directly by immunological methods, while others can only be observed following use of physical separation solutions such as 0.4 M KCl or even more harsh organic chemical exposure (urea, guanidine-HCl, etc.). Some of the reported candidates for AFP-binding proteins include IgG, IgM, actin, TGF-ß, osteonectin, and protease substrates and inhibitors (reviewed in Ref. 7). It has also been well documented that AFP undergoes self-aggregation to form dimers, trimers, and oligomers (70, 71) in vitro, and at least dimers have been detected in vivo (72). The latter investigators (in vivo study) also reported that the dimeric form of AFP may contribute to its function regarding E2-stimulated growth.
Membrane Receptors.
Four cell surface receptors for AFP have been described that might bind various forms of AFP in addition to the native protein. These receptors have been described as both endothelial components and epithelial cell surface membrane receptors (73–76). Eighteen-, 31-, and 60-kD cell surface receptors (73) have been found in the vascular endothelium of many tissues (heart, lung, epididymus, etc.). The 18- and 31-kD binding proteins represent scavenger receptors that bind chemically modified or denatured albumin and AFP, while the 60-kD form is an endothelial cell surface sialoglycoprotein found only on continuously lined (not sinusoidal) endothelium. However, a canonical AFP cell surface receptor has been localized on monocytic, reproductive, immunologic, and tumor cells, notably, hepatomas and MCF-7 breast cancer cells (74–76, 338). This AFP receptor is a 62- to 67-kD protein first detected on human breast cancer cells (77) and later purified from monocyte cell membrane preparations (74). In summary, the field of AFP-binding proteins and receptors is in the early stages of discovery and these receptors have yet to be scrutinized and subjected to cloning, chemical characterization, and physiological study.
|
AFP Fragments
|
|---|
Plasma proteins such as ALB and AFP could possibly serve as circulating protein reservoirs of biological response-modifying peptide fragments. Many proteins are known to serve as precursor molecules and to contain multiple modular sequences or cassette segments generated by proteolytic processing to produce smaller biologically active peptides (78–81). Limited proteolysis of larger proteins appears to be a general mechanism for in situ generation of a variety of regulatory peptides in the circulation (clotting, bleeding, and complement fixation) (69). In many cases, a simple, large polypeptide serves as the precursor pro-protein (pre-proprotein) for a host of biologically active peptides that function in the gastrointestinal, endocrine, cardiovascular, and nervous systems. The peptidic and short amino acid segments (fragments) cleaved from the substrate often display biological effects that differ from the ``parent'' protein. One such example would be the angiogenic peptides developed by Judah Folkman et al. (82) such as angiostatin and endostatin, which are fragments derived from collagen and plasminogen, respectively. Endostatin, for example, is cleaved from the collagen protein by elastase (83). Such peptidic fragments may participate in a host of biologic roles, including biochemical antagonism and agonism, hemostasis, feedback control, and hormonal modulation/regulation. It has been previously reported that short amino acid sequences of ALB and AFP demonstrate sequence similarity with cleaved fragments from neurotensin and neuromedin (84). These peptides were cleaved from a kinetensin ``parent'' protein, which generated peptides capable of modulating biological responses in the endocrine, cardiovascular, digestive, reticuloendothelial, and central nervous systems (see below and Table III). It is of interest in this discussion that excessive placental secretions of neurokinin have recently been associated with pre-eclampsia in the third trimester of pregnancy (333). It has long been known that pre-eclampsia is associated with elevated AFP levels (334, 335).
View this table:
[in this window]
[in a new window]
|
Table III. Human Alpha-Fetoprotein (HAFP)-Derived Peptides Listed According to Domain, Amino Acid Location, and Proposed/Observed Functional Activity
|
|
Enzymatic Fragments.
The ALB molecule has been repeatedly subjected to enzymatic digestion using trypsin, pepsin, chymotrypsin, and others to determine ligand-binding sites (85). In contrast, few studies employing enzymatic digestion of AFP have been reported, other than those using mass spectrometric analysis of the fetal protein (86). However, a recent study by Dudich et al. (50) subjected HAFP to time-limited (mild) peptic hydrolysis using charcoal stripped ligand-free preparations. These investigators obtained and characterized two major AFP peptide fragments, 38 and 32 kD, which were further hydrolyzed to two proteolytic resistant moieties of 23 and 26 kD, respectively. These latter two fragments retained reactive secondary, tertiary, and antigenic structure and were representative of the compact, rigid forms of domains 1 and 3. In contrast, the chemical behavior of the interconnecting domain 2 demonstrated a secondary structure, but lacked a rigid tertiary structure, a form consistent with the ``molten globule'' state. When subjected to tests of biological activity employing apoptosis, it was determined that both P23 and P26 (domains 1 and 3) in a dimer state, together with participation of domain 2 were required for apoptotic signaling in a Raji cell line. Both full-length AFP and its fragments have been implicated in both apoptosis and tumor cytotoxity.
Synthetic Peptide Fragments.
The first biologically active synthetic peptide derived exclusively from HAFP was reported by the author's laboratory for a segment that regulated estrogen-induced growth (87). In that report, an amino acid segment from HAFP 445 to 480 was synthesized by F-Moc chemistry, purified by reverse phase (C-18) HPLC, and structure-confirmed by electrospray mass spectroscopy and amino acid sequence analysis. Subjection to circular dichroic spectroscopy and Fourier infra-red spectroscopy revealed a secondary structure consisting of 10% alpha-helix, 50% beta sheets and turns, and 40% random coil. The peptide, numerically designated synthesis No. P149, was capable of inhibiting both steroid-dependent and -independent growth events, including thyroid-induced frog metamorphosis (30). Computer modeling of the P149 peptide revealed a largely hydrophilic linear peptide with a right angle beta-turn in the last third of the carboxy-terminus. Subsequent studies have shown that the peptide was capable of growth suppression in both estrogen-dependent and -independent tumors (173).
The use of other AFP-derived peptides has increased in the last 5 years (Table III
and Fig. 3). These studies have involved peptides located near the carboxy-terminus of domain 3 and some located near the amino-terminal side of domain 1. A synthetic peptide from domain 1, LDSYQC, was studied to determine its influence on glucose uptake by human erythrocytes (RBCs). The LDSYQC peptide was found to stimulate the entry of glucose into RBCs from insulin-dependent diabetic children after a 1-hr treatment in 10-8 M to 10-6 M peptide in vitro (88). It was of interest that the peptide effect mimicked that of insulin at 10-9 M to 10-7 M concentrations. An amino acid sequence in insulin (amino acids 17–21) was found to be homologous to the AFP peptide. It was this same sequence on AFP (LDSYQCT) that also increased the antiproliferative potency of the drug ``Cytozar'' on cultured lymphocytes from chronic lymphoid leukemia patients (89). The LDSYQCT peptide at 10-8 M to 10-7 M significantly increased the suppression of lymphocyte proliferation in cells from patients experiencing drug resistance. Further, sequence matching of the AFP amino acids 13 to 19 by these investigators showed amino acid identity/similarity not only to the insulin alpha-chain (see above), but also to epidermal growth factor (amino acids 26–32) and glycodelin (amino acids 67–67 and 114–120) (See Ref. 90).
View larger version (94K):
[in this window]
[in a new window]
|
Figure 3. Amino acid peptide sequences of HAFP with known or proposed biological (physiological) activities. Sequences include both enzyme-derived fragments and synthetic segments. ABS, (proposed) actin-binding site; ARP, apoptosis-related peptide; BBS 1,2, (predicted) bilirubin-binding sites; EBS, (documented) estrogen-binding site (in rodents); EGFL, epidermal growth factor-like segment; FBS, (confirmed) fatty acid-binding site; GIP, (documented) growth inhibitory peptide; HCS, histocompatability Class II (confirmed) segment; HMS 1,2, (predicted) heavy metal-binding sites; ILS, (GenBank derived) insulin-like segment; KLS, kinesin-like segments; LPH, leucine predicted heptads; L-A, (GenBank derived) laminin-A segment; L-B1, laminin-B1 segment; LRE, LDV, and RGD, (documented) cell adhesion sequences (one-letter amino acid code); MPS, (documented) milk casein peptide segment; NAI, nonALB identity site (confirmed); PAS, (GenBank derived) plasminogen activator segment; PRS, proline-rich sequence; RBS, (proposed) HAFP receptor-binding site; SRGD, segment reversed RGD site; ZBS, (proposed) zinc-binding site. Data obtained from Refs. 7, 49, 63, 87 through 94, 96, and 97.
|
|
An AFP-derived peptide has also been identified as a potential HLA-A2.1-restricted peptide epitope obtained from a computer analysis of the HAFP complete sequence. Amino acid sequence 542 to 550 from the third domain of HAFP (GVALQTMKQ) was found to bind with low affinity to CD8 dendritic T-cells bearing HLA-A2.1 class I alleles, and displayed slow dissociation kinetics (58). The No. 542 to 550 HAFP peptide generated the induction of human lymphocyte T-cell clones in culture; this peptide in transgenic mice recognized AFP-transfected targets in both cytotoxicity and cytokine release assays. These studies demonstrated that AFP-reactive clones have not been deleted from the human T-cell repertoire, and that AFP is a potential target for T-cell-based immunotherapy using peptide-based specific strategies.
|
Cellular Adhesion Sequences
|
|---|
The present discussion concerns features unique to the second domain of AFP, which has short peptide sequences common to ECM proteins bearing cellular adhesion motifs (CAMs). This finding distinguishes AFP from ALB, ALB, and DBP, which do not include short peptide sequence similarities to the ECM protein family (laminin, fibronectin, collagen, vitronectin, thrombospondin, etc.) of adhesion macromolecules (91). Adhesive macromolecules have potential utility in developmental and disease states involving growth, differentiation, cell migration, and tumor metastasis. Using synthetic peptides derived from the ECMs, the functional significance of such short signal peptide sequences has been identified from one or more domains of these molecules (92–94). Some of the CAM-derived synthetic peptides have been found to block cell differentiation, tumor growth, and angiogenesis. Some ECM proteins (laminin, collagen, and fibronectin) contain multiple biologically active peptide sequences with differing activities specific to cell type (95). Also, the various cellular receptors for a particular active site sequence may differ slightly among specific cells, permitting a large diversity of biological functions and cell regulatory roles. Mammalian AFP, especially the second domain, contains short peptide sequences that are common to the ECM proteins. Further, AFP appears to possess a multiplicity of such peptide sequences, which suggests involvement in a diverse range of biological activities.
A variety of cell attachment peptide sequence sites from ECM proteins (see Refs. 91–108) have been listed in Table IV (left column) together with mouse AFP, and HAFP, ALB, ALB, and DBP numbered peptide sequences. On inspection of the matched peptide sequences, it is observed that frequent amino acid matches of AFP with CAM-like sequences occur in the second domain of AFP (amino acids 190–394). Exact amino acid matches appear less frequently on human ALB and DBP molecules. The physiological functions of the adhesion sequences of the CAMs and the cell types involved are listed in the right hand column of Table IV. It becomes apparent that a large number of diverse cellular adhesion activities might be involved and that AFP shares short sequence similarities with a variety of the different CAM segments identified to date. Such a diverse array of peptide recognition signals suggests that AFP might share functional properties with adhesion molecules, as previously suggested (7). However, such capabilities of AFP might ultimately depend on its ligand-bound state and extracellular localization during growth.
View this table:
[in this window]
[in a new window]
|
Table IV. Cell Adhesion Sequences in Extracellular Matrix Proteins Compared With Sequences on Alpha-Fetoprotein, Albumin, Vitamin-D-Binding Protein, and Alpha-Albumin
|
|
It is of special interest that a periodic spacing of the signal recognition peptides is readily observed in the second domain of AFP. The first of the known peptide signals, LRE, is positioned at the amino-terminal side of domain two on the HAFP molecule (amino acids 194–196). There follows a gap of approximately 40 amino acids before a second signal peptide (LDV) occurs at HAFP amino acids 241 to 243. It is at this point that a pattern of regularity arises. The next four signals (including LDV) occur approximately every 10 amino acid sequences from amino acids 240 to 270, including LDV (fibronectin), RGD (fibronectin), DGEK (collagen I, IV), and YICSQ (laminin B1). The RGD sequence on AFP may be positioned on a tight turn of an exposed loop, as seen on the crystal structure of ALB, which is similar to AFP (20) (Fig. 2). The next proposed signal peptide at AFP amino acid 309 (PNLDR) is similar to laminin B1, followed by a scrambled RGD at amino acid 316, and a sequence at amino acid 351 (ILRVK) comparable with laminin-A. Scrambled RGD is known to be a growth inhibitor sequence (97).
Functional Significance of CAMs.
Functional subdomains of the CAMs have been described and their biological significance has been realized and delineated (9, 13, 28, 91). Multiple studies of CAM recognition signals have shown that they modulate an array of critical biological activities (See Refs. 98–105). Some of these activities include cell adhesion, migration, differentiation, growth, neurite outgrowth, tumor spread, enzyme activity, angiogenesis, and heparin, fibrin, collagen interaction (Table IV). The CAMs located in the ECM interact with a vast array of cell/tissue types, which include epithelial, muscle, fat, peripheral neural, fibrous connective, platelet, endothelial, and tumor cells (melanoma and adenocarcinoma cells) (91). This interaction involves binding of the CAM to cell surface integrins.
The native adhesion proteins (i.e., laminin, collagen, fibronectin, etc.) are found primarily in basement membranes, often tethered to collagen and fibrillar networks. Since each of these proteins contains multiple biologically active adhesion signal peptides, they collectively exert a growth regulatory function on cell migration, spreading, and differentiation. Modulation at these control levels is also a crucial factor in tumor growth and metastasis. Some adhesion proteins are known to contain two signals of opposing activities (YIGSR and SIKVAV of laminin), demonstrating that additional fine-tuning controls are in place on the same molecule to even further regulate adhesion activities (106). Two such sites may not be simultaneously available on a protein due to ligand and/or conformational masking and could be progressively unveiled during a cascade such as the clotting/anticlotting network. Although the integrin cell receptors for a particular adhesion peptide are distinct for specific cell types, it is known that some integrins can recognize multiple sites on CAMs such as the YIGSR, RGD, and SIKVAV sequences for laminin (96).
Cell Adhesion During Development.
These adhesion peptides have also been studied during various developmental stages and have been shown to influence cell differentiation, tissue elongation, cell migration of mesenchyme and neural crest cells, and specific stages of embryonic development (100, 107). Specific peptide sequences (YIGSR) can determine basement membrane cellular attachment, Sertoli cell cord formation, endothelial and salivary cell alignment on basement membranes (SIKVAV), cell adhesion and spreading, and angiogenesis on chorio-allantoic membranes (100, 108). In neural growth, active synthetic peptides can regulate neurite outgrowth, promote the adhesion of ciliary ganglia, recognize various neuronal cells, and influence tyrosine hydroxylase activity in neural transmitter precursors (96). In cancer growth, active site peptides can block adenocarcinoma cell adhesion to collagen and/or laminin or enhance melanoma cell adhesion to such proteins (100, 101). Finally, in hemostasis, adhesion peptides can regulate angiogenesis, plasminogen activation, collagenase IV activity, platelet adhesion, and sites for heparin binding. Most cells use the various integrin-receptor systems as a basis for such activities (109, 110). These data suggest that AFP might be capable of binding to the integrin-receptor system and should prompt further binding studies (111).
Proposed Significance for AFP.
From the teleologic standpoint, it would be reasonable that a fetal protein with reported growth regulatory activities would display an array of adhesion peptides sites in its primary structure (See review in Ref. 111). It would be logical that a serum fetal protein, which ontogentically precedes ALB formation and synthesis, could possess an extensive armamentarium of adhesion site signals that might influence growth and cell differentiation. However, the similarity on AFP of such a diverse multitude of sequence signals common to a variety of adhesion macromolecules may be more than coincidence. This suggests that AFP may play a role in fine-tuning the architectural interstitial growth patterns in developing organisms.
Following the aforementioned logic, it appears that AFP could be armed with pairs of adhesion peptide signal sequences, some of which are known to function with opposing activities. Thus, AFP might possess an array of peptide sequences that differ slightly from the CAM sequence and thus serve as inhibitor sequences (see Table I, GDR versus RGD). As alluded to above, all adhesion sites may not be concurrently accessible since AFP is thought to express different functions at various developmental stages (4). Some sites may be unmasked or made conformationally available during progressive stages of embryonic and fetal development. As in the case of RGD, it appears that the peptide sequences and tertiary structure surrounding that particular peptide site could be important for its eventual activity or its competition in the native AFP molecule, as was shown for the cyclic peptide CRGDCL (112). It is also known that several upstream (amino terminal) sequences in fibronectin function as synergistic cohorts regarding RGD activities (113, 114). It is of interest that AFP binding of fatty acids, bilirubin, and various drugs occurs in the second domain of the molecule (7). In fact, fatty acids are known to be bound at amino acids 209 to 227, in a region lacking CAM sites (115, 116).
The proposed activity of these adhesion peptides in AFP must of course await experimental verification. The peptides are easily synthesized; however, suitable experimental models to assay their biological activity are not as readily available. Such cell adhesion models do exist and efforts to document their activities might prove rewarding for therapeutic and diagnostic purposes. Undoubtedly, there are additional active peptide sequences in AFP, especially in the non-CAM peptide region (amino acids 197–240) described above or in other domains. The use of synthetic AFP-associated CAM peptides could more clearly delineate and possibly unravel developmental mechanisms concerning growth, differentiation, and tissue architecture. Such AFP peptides may find utility as therapeutic, diagnostic, or screening (marker) agents in diseases such as cancer and hematological disorders. Others might find uses in biotechnology procedures such as cell cloning and hybridization, matrigel cell culture assays, coating of artificial transplants/implants, and adsorption to prosthetic devices (91). Finally, it is conceivable that cells or proteins armed with adhesion peptides could be targeted to areas of chronic or acute inflammation to possibly accelerate wound healing, inhibit metastases, and minimize the formation of scar tissue at lesion sites.
|
Molecular Microheterogeneity
|
|---|
Carbohydrate Isoforms.
In contrast to ALB, AFPs are glycoproteins containing carbohydrate moieties in one to three N-linked glycans. While HAFP contains only one N-linked glycan at Asn 232 on domain 2, mouse AFP displays three sites at Asn 232, Asn-310 on domain 2, and Thr 483 on domain 3, and rat AFP possesses three sites at Asn 232, Ser 96 (domain 1), and Asn 310 on domain 2 (117, 6, 7). In human, the structure of the glycan on AFP produced by a hepatoma differed from that produced by a yolk sac tumor (11). The AFPs produced in cells from different organs and under various pathological conditions have the same sequence and are immunologically the same (2.2-kb form); thus microheterogeneity is displayed by AFP's reaction to different lectins and to pH environments (see below).
Different isoforms can be demonstrated by electrophoretic, chromatographic, and isoelectric techniques in combination with methods employing lectins (See Refs. 118–126). Therefore, most researchers denote AFP variants as binding (reactive) or not binding (nonreactive) with their respective lectins (Table V). The three lectins that have been most useful for studies of AFP are concanavalin-A (Con A), Lens culinaris agglutinin (LCA), and Vicia faba agglutinin (VFA) with a d-mannose specificity (11, 118). Lectin specificity for d-galactose is also demonstrable with Ricinus communis agglutinin-1 (RCA), and Viscum album agglutinin-1 (VAA), both of which react and bind to the carbohydrates of HAFP. The galactose-specific lectins are thought to react with the penultimate galactose of the biantennary AFP glycan chain. The lack of reactivity with other lectins permits one to determine the structure of the glycan of all AFP variants (126, 127). However, the mannose-specific lectins require the presence of one fucose bound to the glycosyl-N-acetyl (GlcNAc) next to the Asn residue. Studies have shown that most molecules of HAFP possess a biantennary glycan substituted with two sialic acid residues (11). Some lectins such as phytohemagglutinin can be induced to bind to HAFP following removal of sialic acid residues (123). Wheat germ agglutinin, which does not bind to HAFP, requires an additional fucose adjacent to the GlcNAc next to the Asn residue (125).
View this table:
[in this window]
[in a new window]
|
Table V. Specificity Reaction or Nonreaction of Selected Lectins With Human Alpha-Fetoprotein (HAFP) IsoForms and Their Clinical Utility as Biomarkers in the Differential Diagnosis of Cancer and Fetal Defects
|
|
The sugar chains are not genetically encoded, but are dependent on the set of glycosylation enzymes present in the endoplasmic reticulum (ER) and the Golgi complex of the host cell. However, these enzymes have different tissue distributions, and the variants of AFP are due to the various tissue specific pathways involved (128–131). Thus, yolk sac-derived AFP differs from AFP of hepatic origin. However, AFP in amniotic fluid consists of a mixture of these two types in different ratios, depending on gestational age. Differences encountered in various tumor types also suggest quantitative as well as qualitative differences in the glycosylation enzymes from each tumor (Table V).
The carbohydrate heterogeneity of HAFP has spurred the use of lectin-affinity separation techniques for use in the diagnosis of specific fetal defects and for distinguishing between different tumor types. It was first shown in 1979 that isoforms in amniotic fluid AFP contained Con-A-binding and Con-A-nonbinding variants as a mixture from both yolk sac- and liver-derived origins (128). The nonbinding variant is decreased in pregnancies with neural tube defects, and can be separated and quantitated by column chromatography. By 1981 it was already obvious that determination of AFP-branched sugar heterogeneity would be a useful tool in the differential diagnosis of cancer (132, 133). Yolk sac tumors and gastrointestinal cancers could be distinguished from hepatomas based on LCA reactivity, and from benign liver disorders (nonreactive LCA) employing imaging procedures (134). The addition of lectin affinity chromatography and affinity electrophoresis in conjunction with autoradiography, indirect enzyme-linked immunoabsorbant assay (ELISA), and blotting techniques has since amplified the detection of a multitude of AFP glycoforms. These techniques have added a new dimension to identification since variants can be detected not only as binding and nonbinding bands, but also as gel retardation moieties.
Using combined lectin affinity-based procedures, i.e., crossed affinity immunoelectrophoresis, at least 10 HAFP glycoforms have now been identified and characterized (126). Lectins have been found to be specific not only for sugar residues, but also for the whole carbohydrate molecule. More recently, inclusion of additional lectins such as Phytohemagglutin-A (E-PHA), allomyrina dichotoma lectin (ALLO-A), and Datura stromonium agglutinin (DSA) has increased sensitivity and specificity of lectin reactivity with AFP (127). With the use of the new lectins and novel methodologies, the glycoforms of maternal serum AFP were found to be similar to those from umbilical cord serum AFP. This finding may be useful for evaluating the developmental state of the fetus by examining the nature of the AFP sugar chain. Initially, the use of MoAbs specific to AFP variants seemed to be unsuccessful. These early negative results may have occurred because the biantennary glycans of HAFP may be a self-antigen for mice, and because of the use of fetal calf serum, which may contain bovine AFP, with similar glycoforms as HAFP. However, recent studies using LCA-reactive and -nonreactive AFP demonstrated that LCA inhibited the binding of two MoAbs to HAFP, showing that a competition existed between the antibodies and LCA for the AFP sugar chain (12).
The present review was not intended to be all-inclusive for the carbohydrate microheterogeneity of AFP glycoforms. Therefore, the reader is referred to the extensive reviews on this topic published by Breborowicz, Lamerz, and Taketa (11, 126, 127, 337). Regarding the physiology of the microheterogenetic carbohydrate forms, rat AFP subjected to lectin columns shared higher estradiol binding affinities with less carbohydrate (KA= 10-8 M with 0.1–0.5 binding sites), while the high carbohydrate forms bound E2 with lower affinity (KA = 10-7 M and 0.5 site) (135). Also, purified mouse AFP devoid of carbohydrate had very high-affinity binding sites for estradiol (KA = 0.7 x 108 M, W = 3) (136). Finally, studies showed that human peripheral lymphocytes were inhibited by AFP in the blast transformation assay in response to PHA and Con-A stimulation, but not pokeweed mitogen (137).
Isoforms of pH Heterogeneity.
Isoforms of AFP have also been identified and characterized following exposure to changes in the pH environment. An early study by Alpert et al. (138) ascertained that HAFP could be separated into two major molecular forms by isoelectric focusing followed by classical chromatographic separation procedures. The two forms, found in both hepatoma and fetal sera, displayed isoelectric points of pH 4.8 and 5.2. The authors reported that a single homogeneous form (pH 5.2) could be produced after treatment with neuraminidase. Two subsequent reports followed in which another group of investigators demonstrated that the two isoelectric isotypes (pH 4.8 and 5.2) could be distinguished by their fatty acid content (139, 140). The pH 4.8 material contained 2.4 mol of fatty acids/mol of protein, while the pH 5.2 form contained no fatty acid. Removal of the fatty acids by treatment with charcoal converted all the AFP into material displaying an isoelectric point of pH 5.2. Addition of fatty acid to this HAFP solution again restored the protein to an isoelectric point of 4.8. More recent studies employing isochromatofocusing of HAFP in cord blood revealed the presence of three isoelectric variants (141). These pH isotypes included a pH 4.5 (52%); a pH 4.7 (43%); and an isovariant (at <pH 4.0) that could be eluted from the column by a 1.0 M NaCl solution. In this report, a pH 5.2 variant was not reported using cord blood as the AFP source. Conformational transition forms can also be induced in AFP by changes in pH as discussed later. AFP molecular transition forms have been reported at both acid and base extremes of the pH range (142). The conformational changes are reversible and gradual, indicating the presence of multiple transition forms of AFP. The authors of this latter paper reported that both rodent and HAFP had a remarkable hydrophilic exposed molecular surface at neutral pH and possess extensive hydrophobic segments hidden in molecular crevices as discussed in both the peptide fragment and epitope sections.
|
Antigenic Variants
|
|---|
AFP, as a member of the albuminoid gene family, displays 38% to 40% amino acid sequence identity to ALB (6, 117). Although unreactive in their native folded states, RIA utilizing reduced and carboxymethylated (CM) proteins as antigens reveal a serological cross-reactivity between AFP and ALB (143). Unfolding of the polypeptide chains by reduction of the disulfide bonds followed by CM produces derivatives that differ from the native state and crossreact in an RIA between AFP and ALB. It was further observed that ALBs from different animal species were equally crossreactive as HALB; overall, the AFP crossreactivity assays were found to lack species specificity (144). Although most serum proteins unrelated to AFP or ALB reacted little if any in the RIAs, transferrin (TRF) unexpectedly demonstrated a crossreactivity. No crossreactivity of TRF with AFP in its native states was observed; however, antisera to CM-AFP reacted with CM-TRF at high titers. Interestingly, the Gc DBP showed only moderate reactivity in this study. All the above RIA associations were further confirmed using Western blotting procedures. These results suggested that both TRF and Gc protein were structurally related to both AFP and ALB. Although Gc protein has been classified as an albuminoid family member, TRF is seemingly unrelated and these findings remain to be explained.
Immune Tolerance.
AFP, when injected into a foreign species, produces a strong immune response, as does ALB. However, neither ALB nor AFP elicits an immune response in the species in which the proteins originated (145). Moreover, AFPs from different mammalian species are crossreactive and antibodies to HAFP induced in chickens recognized all mammalian species tested (146). Breakage of tolerance to autologous AFP can only be achieved using chemically modified or desialylated AFP injected with complete Freund's adjuvant into animals (147–149). Thus, the tolerance can be terminated by immunizing either with heterologous or hapten-modified homologous AFP, which results in production of antibodies that react with native host AFP (150). These antibodies crossreact with AFP in its native state, and the unfolded (reduced and CM) forms of AFP also show immunologic crossreactivity. The antibodies produced in this fashion were found to diminish the AFP levels normally measurable in the host serum (151, 152). Such antibodies also diminish the elevation of serum AFP seen in hepatoma-bearing rodents, but do not protect against such tumors. In comparison, heterologous immunizations of mouse AFP into rats and vice versa have induced fetal death in pregnant animals (148, 153). The passive immunization of heterologous antibodies into hepatoma-bearing mice was found to induce pathological changes and suppress overall growth of these tumors (154). Furthermore, in vitro studies of such heterologous antibodies were found to be cytotoxic to murine hepatoma cells (155, 156).
Antigenic Determinant Sites.
Serum ALB has been extensively used as an antigen. AFP is also a prime antigen when used as a foreign protein in a host species. The multiple antigenic determinant sites on proteins, lipoglycoproteins, and polysaccharides are termed epitopes, which are specific, limited parts of an antigen molecule that serve as inducers of antibody formation. When an antigen is employed to raise antibodies, the nature of the antigenic specificity should be considered for optimal usage. Induction of antibodies by an antigen molecule depends on rigidity of the molecule, presence of functional groups, size of the determinant sites, and the tertiary structure of the immunogen. Within an epitope, there is one area that is usually more dominant than the others even though the total determinant group may cover a large region. This area is known as the immunodominant site and often protrudes the farthest, but is not necessarily the terminal group (157). The immunodominant site comprises the major region of an epitope that confers individual and species variability and may even vary within an individual. In accordance with reported calculations and observations, the major epitopic site has a molecular size of about 10,000 Daltons; therefore, a molecule such as AFP (65 to 70 kD) should predictively display six to seven major epitopes (158). However, upon denaturation, as many as three internal minor sub-epitopes can be exposed per major epitope. The major epitopes are referred to as the functional valence of the molecule, while the internal sites exposed by denaturation are termed the nonfunctional valence (NFV). Hence, AFP should display at least six major epitopes (valence = 6) and possibly up to 12 additional minor epitopes (NFV = 12) for a possible total of 18 antigenic sites.
Epitope Analysis.
A summary of the epitope analysis of HAFP with an array of MoAbs was reported at an international workshop in 1998 (159). Previous reports using polyclonal antibodies had inferred that HAFP indeed displayed at least six major epitopes (160). A summary of the findings of the 1998 workshop was published in a series of four consecutive reports. Using 30 different antibody clones against HAFP, two reports confirmed the presence of five to six major epitope clusters and identified several additional minor epitopes (161, 14). One of the major epitopes did apparently contain carbohydrate as part of the immunodominant site specificity as reported in these studies (12). Another of the epitopes identified in the HAFP molecule demonstrated both open and cryptic (hidden) forms (13). An overall survey of the literature including both mono- and polyclonal antibodies to AFP was in agreement that four to eight distinct epitopes are present on the AFP molecule (162–171) (Table VI).
View this table:
[in this window]
[in a new window]
|
Table VI. The Major and Minor Epitopes on Human Alpha-Fetoprotein Are Categorized in Accordance With Immunoglobulin-(Ig) Subclass, Number of Cell Clones, and Scientific Utility
|
|
Epitope Mapping.
A study described at the international workshop by Norwegian investigators reported six major epitopes and an additional seven sites of low antibody-binding activity, suggestive of minor epitopic sites (161). The six major epitopes were labeled A, B, C, D, E, and X (a partially hidden site) and were represented by antibodies of high binding affinity (Fig. 2C). A second group led by Russian and Dutch scientists (14) reported at least 11 antigenic clusters, which were categorized into six distinct major epitopes (A–F) derived from 30 HAFP MoAbs. These MoAbs to HAFP exhibited four means by which cross-reactivity occurred: full competition; partial competition; full independence, and enhancement of binding. Additional complexity was encountered when some MoAbs reacted with AFP in solution while others reacted only with AFP adsorbed to a solid surface. Thus, some epitopes were hidden or cryptic in nature and were expressed only on partially denatured AFP adsorbed to plastic or nitrocellulose membranes (NCM). Finally, it was observed that the MoAbs reacted similarly with the major epitopes whether HAFP was derived from umbilical cord serum, placental extracts, or tumor tissue. However, minor epitopic differences between cord sera and placental extracts were found, suggesting conformational alterations in AFP forms from the different tissues (14).
Studies presented by Russian investigators at the workshop (see above) found additional MoAbs reacting against subfractions of native HAFP, suggesting the presence of at least four additional minor epitopic variants (13). These subfractions of AFP displayed similar molecular weights to the native AFP molecule, but showed different serological reactivity when conformationally altered by fixation on NCMs. These epitopes existed in two forms on the native AFP molecule: an open and a cryptic form. Since the hidden sites could be revealed after partial denaturation, these subfractions represented conformational variants that lie partially buried in the compactly folded, intact molecule. Recently, cryptic epitopes on HAFP have been reported to induce spontaneous immune responses in human patients suffering with hepatomas, cirrhosis, and chronic hepatitis (172). The physiological significance of hidden subdomain epitopes on HAFP has emerged as a fertile field for both basic and clinical investigations of AFP-regulated ontogenic and oncogenic growth.
One such hidden epitope on the AFP molecule (P149 peptide) did not react in commercial radioimmunologic, immunoenzymatic, or chemiluminescent assays designed to measure HAFP levels in pregnant and tumor-bearing patients. In contrast, polyclonal antibodies produced in rabbits using keyhole limpet hemocyanin conjugates produced a moderately high-titered P149-peptide antiserum (1:10,000), which was subsequently adapted to direct and indirect ELISA employing an alkaline phosphatase labeled antibody. By means of the peptide-ELISA it was demonstrated that the antibodies could not detect the amino acid peptide sequence on intact, native HAFP that had been purified from term cord serum (173). However, the immunoreactivity of the peptide antiserum was demonstrable following treatment of native HAFP with high concentrations of estrogen or by solid-phase adsorption of HAFP to the microtiter plate surface. The antiserum also was capable of detecting the peptide in amniotic fluid during pregnancy and in the serum of patients with AFP-secreting tumors.
In summary, knowledge of the antigenic structure of AFP is both important and necessary for several reasons. First, knowledge of the major and minor epitopes on HAFP could lead to more rational designs of monoclonal/polyclonal antibodies for immunoassay kits produced by the various pharmaceutical companies. Such designs might include both the major epitopes on the native AFP molecule and the minor (hidden and exposed) epitopes observed in the conformational variants of AFP. It may also be important to know the ratio of conformationally transformed to native AFP in the serum to determine whether it may relate to disease states and/or cancer stages. Such information might prompt further comparative studies among the various kit manufacturers, as well as analysis of specific functional groups on the AFP molecule itself. Such information might also be employed to identify AFP domains or motifs responsible for some of its functions such as ligand binding/transport, cell receptor interactions, growth regulation, and immunoregulatory activities. Finally, these findings could aid in characterizing and distinguishing the variant states of the AFP molecule such as the compact native form, denatured intermediates, conformational variants, and the MGF (see next section).
|
Denatured Intermediates and the MGF
|
|---|
Proteins that reside in living cells are thought to assume a conformational state unlike that of the native, secreted molecules found circulating and/or stored in biological fluids (15). The highly variable environments of the different intracellular compartments dictate that a protein exist in a mildly denatured (unfolded) state, which has been termed the MGF (15). As discussed below, the MGF has now been implicated in a variety of physiological activities including protein translocation, insertion into membranes, heat shock protein (HSP; chaperone) binding and recognition, and protein degradation by lysosomes and/or the ubiquitin system. First discovered in 1981 (174), the MGF represents an intermediate folded state of a protein induced by mild denaturing conditions in contrast to harsh exposure to strong ionic solutions, detergents, and extreme pH values, which cause aggregation. This slightly unfolded state of a protein has now been demonstrated for many different naturally occurring proteins including HAFP (175–196) (see Table VII). Although the MGF can be induced in vitro by low pH, moderate ionic strength solutions (guanidinium chloride [GnCl], urea, etc.) and high temperatures, the MGF can also be produced in physiological environments by point mutations, removal of ligands, and disruption of disulfide bonds (15, 176, 177). The MGF can readily be studied using low and high pH or moderate organic solution concentrations with techniques such as near and far ultraviolet circular dichroism, infrared spectra, and nuclear magnetic resonance (NMR). Extremes of the conditions stated above produce a completely unfolded state, which is the totally denatured or aggregated form (Fig. 4 and Table VII). However, it must be noted that not all proteins have been shown to form stable molten globule states or to fold through MG intermediates.
View this table:
[in this window]
[in a new window]
|
Table VII. Alpha-Fetoprotein and Other Proteins That Form Molten Globule Intermediate States Are Listed (see Fig.4)
|
|
View larger version (25K):
[in this window]
[in a new window]
|
Figure 4. Flow diagram showing the properties and progression of a native protein (see Table 7) to the MGF derived from in vitro studies (Refs. 178–186). Directional arrows indicate bi- and unidirectional flow reactions. The numbered parentheses signify predicted multiple intermediate steps in the progression toward the MGF.
|
|
Characteristics of the MGF.
The MGF is thought to represent a third thermodynamic state of protein molecules induced by mild denaturation (15, 197) (Fig. 4). Formation of the MGF proteins are proposed to occur in a four step sequence as follows: the native-like state; a transitory unfolded form; the true MGF; and the denatured form (see Fig. 4). The MGF, once considered to be an equilibrium state at mild denaturing conditions, is now viewed as a kinetic intermediate in the protein folding process (197). Thus, the properties of the MGF are its relaxed compactness, stable secondary structure, nonrigid tertiary structure, and ability to bind hydrophobic probes due to its less rigid tertiary structure.
It is surprising that the MGF, rather than the classical native protein, has been proposed as the common form within the cell for various physiological purposes (15, 198). The cytoplasm of the living cell is teeming with high concentrations of proteins, nucleic acids, and organelles at low pH (4.5–5.0), which commonly encounter heat and osmotic shock, variable electrostatic fields, and a myriad of membrane interfaces (199). Although intracellular environments might not represent the extreme denaturation conditions that can be produced in vitro (urea, GnCl), cytoplasmic proteins do endure mild denaturation states, which probably alter their overall molecular structure (198, 199). In this more flexible form, a protein can readily adapt itself to changing cell conditions such as interaction with chaperones, membranes, stress/shock extremes, and encounters with other intercellular components and organelles. As hypothesized in Fig. 4, AFP (as example protein) in its intact, native form would embody a compact, rigid molecule with an inaccessible central core (Fig. 5). When exposed to a stress/shock microenvironment (i.e., high serum estrogen levels from the mother), the fetal AFP molecule might progress through one or more intermediate forms (representing molecules exposed to slightly increased increments of denaturation) to arrive at a transitory MGF with slightly looser packing and unfolding chains (Fig. 4, top right). This transitory (intermediate) MGF of AFP would be further characterized by a slightly less rigid tertiary form with a central core accessible to solvents. This transitory AFP form, upon continued exposure to the stress/shock state, might then pass through one or more stages toward the true MGF of AFP. The specific (true) MGF of AFP represents a totally flexible molecule with solvent accessible nonpolar residues and complete mobility of both aromatic and aliphatic side chains. This flexible form of AFP is more protease sensitive, binds hydrophobic ligands (dyes), and displays affinity for cell membranes (Fig. 6). Up to this point, both the transitory and true MGFs are in dynamic equilibria (two-way arrows, Fig. 4) depending on their molar ratios; however, the true MGF is aggregation sensitive and might eventually be susceptible to irreversible denaturation (Fig. 4, lower left). This hypothetical form (MG) of HAFP might comprise only 1.0% or less of the total circulating fetal protein (247), but might be expected to increase in the presence of fetal defects (173) and be the estrogen-binding form of HAFP previously described (339, 340).
View larger version (35K):
[in this window]
[in a new window]
|
Figure 5. Diagrammatic representation of the proposed tri-domain structure of HAFP transitioning from the native to the MG state. This artist's conception is predicted from data derived from Refs. 15, 189, 175 through 178, 194, 197, and 203.
|
|
View larger version (57K):
[in this window]
[in a new window]
|
Figure 6. Flow diagram depicting the pathway of the molecular chaperoned folding of ribosome-translated nascent polypeptide chains through the normal folding process to the functional protein. As shown, the partially folded intermediates can be passaged through a MG stage or deemed a misfolded protein destined for degradation. The far right side pathway depicts a f | |
Top
Abstract
Introduction
