Division of Translational Medicine, Wadsworth Center, New York State Department of Health, Empire State Plaza, Albany, NY 12201, USA
March 2009
List of Abbreviations: AA, amino acid; AFP, alpha-fetoprotein; APC, antigen presenting cells; AR, androgen receptor; CF, cystic fibrosis; Coup, chicken nuclear receptor; CTL, cytotoxic T-lymphocyte; DC, dendritic cells; E, estrogen; E1, estrone; E2, estradiol; ear, human v-erbA-related receptors; EGF, epidermal growth factor; erbA, putative thyroid hormone receptor proto-oncogene products; ERR, central nervous system receptor; FSH, follicle stimulating hormone; GR, glucocorticoid receptor; HAFP, human alpha-fetoprotein; HCC, hepatocellular carcinoma; HER, human estrogen receptor; HLA, human lymphocyte antigen; HRE, hormone response element on DNA; IFNgamma, interferon gamma; IGF, insulin growth factor; IL, interleukin (IL-2, IL-13, etc.); LPS, lipopolysaccharide TCR, T cell receptor; MER, mouse estrogen receptor; MHC, major histocompatibility complex; MR, mineralocorticoid receptor; NKC, natural killer cells; nur-77, mouse nerve growth factor-like receptor; PDGF, platelet-derived growth factor; POLY I: C, polyinosinic:polycytidylic acid; PR, progesterone receptor; RAP, receptor auxiliary (accessory) proteins; RAR, retinoic acid receptor; RXR, retinoic-X-receptor; T3R, thyroid hormone (triiodothyronine) receptor; TCGF, T cell growth factor; TNF, tumor necrosis factor; VDR, vitamin D receptor.
Abstract
The structure of alpha-fetoprotein (AFP) is presented in light of AFP membership and position in the albuminoid gene family in comparison to other gene family members. Ontogenetic AFP gene expression is reviewed considering AFP mRNA presence in various tissues at different times during development. The multiple molecular variant forms of AFP are discussed in relation to published reports of AFP binding proteins and cell surface receptors. The atlas further shows AFP as a protein consisting of multiple peptide-cassettes consisting of amino acid (AA) sequence stretches matched to peptide segments on prohormones and biological response modifier proteins. Such AFP peptide segments could potentially serve as delivery agents which could target vascular, neuroendocrine, or gastrointestinal cells. A discussion follows in which peptide epitopes, extracellular matrix proteins, serine proteases, extracellular matrix, and cellular adhesion AA identity sites on AFP are considered. The AFP molecule is also viewed as a carrier/transport protein based on AA sequence comparison of various proteins that bind hydrophobic ligands and heavy metals similar to AFP binding of such components. An analysis of transcription factors, tumor suppressors, and AA-rich motifs follows, interfaced with dimerization and nuclear localization sequence matches identified on the AFP molecule. Emphasis is further placed upon homeodomain and apoptosis AA sequence identities given that AFP serves as a fetal, phase-specific protein throughout embryogenesis, histogenesis, and organogenesis. AFP AA sequences are further presented as peptide identification sites for growth factors, receptors, cytoskeletal proteins, and chemokines. A closing discussion summarizes the multiple and varied motifs of peptide sequences matched to AAs on each of AFP's three domains. This study indicates that short peptide segments, in addition to full-length AFP, and domain and subdomain fragments of AFP, could be employed as functional agents to help unravel the complexity of biological roles ascribed to human AFP.
I. Introduction
Human alpha-fetoprotein (HAFP) is a single-chain glycoprotein with a molecular mass of 69 kDa or greater depending on carbohydrate (glycan) content (Crandall, 1981; Mizejewski, 1985; Deutsch, 1991). HAFP, a tumor-associated fetal protein, is classified as a member of a three-domain albuminoid gene family that consists of four members: albumin (ALB), vitamin-D binding protein (DBP), AFP, and alpha-albumin (α-ALB) (McLeod and Cooke, 1989; Lichenstein et al., 1994) (Figure 1). Similarly to ALB, HAFP 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 (Mizejewski, 1995a, 1997). Unlike ALB, high concentrations of hydrophobic ligands (i.e., fatty acids, estrogens) have been reported to induce multiple conformational transition forms, which are reversible, in the tertiary structure of HAFP (see ref. (Mizejewski, 2001a) for review). Discordant HAFP levels have long been associated with aberrant growth conditions, and it was assumed that these AFP levels were coincident events, rather than the cause of such malformations. Although AFP may not be the direct cause of the altered growth structures observed in birth defects, the belief is held that some shock/stress-induced conformational (variant) forms of this fetal protein may influence, modulate, or contribute to such events (Mizejewski, 2001a). Over the last two decades, reports have emerged that some of these HAFP forms can serve as dual regulators of growth, capable of either enhancement or inhibition of growth, in cancer as well as fetal cells (Mizejewski, 2002; Semenkova et al., 2003). Furthermore, AFP has been reported to promote growth via the protein kinase A pathway (Li et al., 2002b). The growth regulatory properties of HAFP are one of the more notable features that distinguish this fetal protein from adult or fetal serum ALB.
Objectives:
At present, there are only a few published reports that have attempted to map the many structure/function sequences identified on the entire HAFP amino-acid sequence (Mizejewski, 2001a; Terentiev and Moldogazieva, 2006, 2007). This atlas will survey the structure/function peptide sites on HAFP that have been identified over the last two decades in addition to the most recently discovered protein peptide sites. First, the structure of HAFP will be presented as a member of the albuminoid gene family. Second, the relationship of AFP's biological activity to the naturally occurring genetic and structural variant forms of AFP will be presented. Third, the multiple names and sites of the functional motifs discovered on HAFP will be presented in illustrated forms together with a narrative that attempts to link AA structural findings with physiological activities. The present atlas is intended to largely serve as a compendium source that maps most of the functionally-known and proposed active sites on the entire AA sequence of HAFP, an established biologic response modifier (Mizejewski, 1997). Since the present review may not be inclusive for all physiological activities of AFP, the reader is directed to earlier papers that cover these multiple biological functions (Uriel, 1989; Deutsch, 1991; Nunez, 1994; Mizejewski, 2001a, 2002).
II. The Albuminoid Gene Family
AFP, as a member of an albuminoid gene family, is structurally characterized by 32 cysteine residues that result in folding of the polypeptide chain into layers that form loops dictated by disulfide bridging; this results in a triplet domain, V-shaped molecular structure (McLeod and Cooke, 1989; Lichenstein et al., 1994) (Figures 1, 2). The three domains of AFP and the ALB gene family members have been confirmed by X-ray crystallography, electron dot imaging, and helix-ribbon modeling (Luft and Lorscheider, 1983; Carter and He, 1990) (Figures 1, 2). The albumin gene family members share structural similarities, homologous AA sequence stretches, and reflect similar cysteine disulfide bridge cluster patterns (Figure 1 and Table 1). In humans, the four albuminoid genes lie in tandem on chromosome 4 within the 4q11-q22 region, encompassing 15 exons and 14 introns (Yang et al., 1985). The Gc-protein alone is truncated in the third domain (see dotted square in Figure 1B) and contains only 13 exons, which results in a protein with a lower molecular mass (Ray et al., 1991). The newest member of this gene family, α-ALB, was discovered in both rodents and humans and has been cloned in the last decade (Belanger et al., 1994; Allard et al., 1995). 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 adhesion, copper-stimulated lipid peroxidation, and fatty acid, heavy metal, and actin binding, among others (Gutteridge, 1986; Constans, 1992; Nathan et al., 1993). Although the function of the recently discovered αALB remains obscure, it may play a role (vitamin-E binding and immunoregulation) similar to its orthologs in lower vertebrates, a 74 kD albumin-like molecule found in fishes, sharks, amphibians, and reptiles (Ohkawa, 1987; Butterstein and Mizejewski, 1999).
III. Genetic Variants
The genetic variants of mammalian AFP have been studied primarily in the rodent and to a lesser extent in humans (Table 2). The major fetal and tumor-derived mammalian AFP mRNA consists of a 2.2 kilobase (kb) transcript that translates to 68-72 kD molecular forms in humans and rodents, depending on its carbohydrate content (Petropoulos et al., 1983; Petropoulos et al., 1985). 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 (Morinaga et al., 1982). In both species, a 1.35 kb form is retained intracellularly as a non-secreted form previously described for uterine and cancer cells (Smalley and Sarcione, 1980; Sarcione and Hart, 1985). 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 certain orphan steroid receptors, which dimerize with other nuclear receptors, have similar truncated structures (Mizejewski, 1993; Seol, 1998; Resnick et al., 2000). In this form, truncated AFP variants could potentially dimerize with steroid receptors as previously proposed in computer models (Mizejewski, 1993). The adult form of human AFP, like the rat forms, might also be derived from multiple RNA transcripts (i.e. 2.2, 1.7, 1.6 kb). It is well established that the classical form of human AFP detected in most RIAs and EIAs represents the 69,000 kD (2.2 kb) polypeptide. It is highly probable that other AFP mRNA transcripts exist and are translated in cells and/or tissues but are undetectable by present-day immunologic assays (see below). Although isolation and characterization of these predicted forms are difficult to measure due to their vanishingly small concentrations, i.e. nanogram to picogram levels, recombinant PCR technologies could someday aid in resolving this issue.
IV. Structural Variants
Molecular variants of HAFP have been known in the scientific literature since the 1970-1980 era (Sarcione et al., 1983; Sarcione and Biddle, 1987). Initially, some of the different forms have been attributed to carbohydrate microheterogeneity and variations in isoelectric points (Smith and Kelleher, 1980; Crandall, 1981). Some AFP isoforms were lectin glycoforms that were detected and isolated by isoelectric focusing, electrophoresis, and chromatographic methods (Breborowicz, 1988; Taketa et al., 1998). Other isoforms were detected following high-pressure liquid chromatography (HPLC), and 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 (Karamova et al., 1998; Yakimenko et al., 1998). Finally, the discovery and characterization of the molten globule forms of HAFP have provided a new level of understanding regarding the various intermediate transition forms of the fetal protein (Bychkova, 1993). A variety of molecular variants of HAFP have further been reported as a result of clinical assays which detected aberrant molecular forms of the fetal protein. Several such reports of aberrant AFP molecules first appeared in the clinical cystic fibrosis (CF) literature resulting in confusion of the clinical usefulness of AFP for this genetic disorder (Table 3). In the 1970s and 1980s prior to the development of monoclonal antibodies, polyclonal antibody assays for AFP were not as precise and sensitive as today's immunoassays. Such factors resulted in disparate baseline levels of HAFP in the sera of non-CF and normal adult patients which ranged in concentrations from 5-30 ng/ml. To add to the complexity, a previously reported cationic migrating form of HAFP was determined to be HAFP complexed with IgM molecules; this cationic form has now been described in several reports (Mizejewski, 1997, 2001a, 2002; Beneduce et al., 2004). Additional aberrant forms of HAFP have also been detected in fluids found in the reproductive/and urinary tract of various clinical patients (Lippes et al., 1983), and in the tissues and sera of patients bearing breast cancer (Sarcione et al., 1983; Sarcione and Hart, 1985; Sarcione and Biddle, 1987). A non-secreted form of HAFP, lacking the N-terminal signal sequence segment, was also reported in recombinant AFP studies employing yolk sac tumors (Fukasawa et al., 2005). Furthermore, truncated forms of HAFP (~50 kDa) have been detected in cell cultures comprised of hepatomas, testicular embryonal carcinomas, and breast tumors (Mizejewski, 2002). Variant forms of HAFP transcripts from non-translated regions of the AFP mRNA have recently been reported in CD34+ hematopoietic progenitor cells derived from mesodermal germ cells (Kubota et al., 2002). These latter investigators described two variant forms of HAFP mRNA that were not expressed in mature adult cells. The variant AFP mRNAs differed from the authentic AFP mRNA transcripts by incorporating exons from the 5'-untranslated region of the HAFP gene. The abnormal AFP transcript was found only in bone marrow, thymus, and brain tissue. The various folding intermediate forms of HAFP have further been investigated using bacterial and yeast recombinant methodology (Yazova et al., 2003; Leong and Middelberg, 2006). The folding of both glycosylated (yeast) and non-glycosylated forms of recombinant HAFP was studied following protein purification from aggregation-prone E. coli inclusion bodies. After AFP was denatured, it readily refolded under dilution, redox reactions, and ELISA assay conditions in both recombinant produced AFP forms. In summary, the denaturation of recombinant-derived HAFP was found to be a reversible process independent of its source of origin, fatty acid relationship, and glycosylated state.
V. Biological Activities of AFP: Mapping of Structure/Function Peptide Sites
The many and varied biological activities of HAFP reported in the biomedical literature have been studied for many years (see (Crandall, 1981; Mizejewski, 1985; Deutsch, 1991; Mizejewski, 1997, 2001a, 2002)). Such functional activity sites of HAFP have included: a) immunoregulation and peptide epitopes; b) extracellular matrix binding and cell adhesion; c) enzyme-related modulation; d) transcription factor/homeodomains; e) apoptosis-associated activities; f) hydrophobic, ion, and metal ligand binding; g) hormone and growth factors; h) cell surface receptors; i) dimerization and nuclear localization signals; j) AA rich repeat sequences; k) cytoskeletal proteins, filaments, and microtubules, and l) multiple chemokine sites. In the remainder of this atlas, the discussion and mapping of the above protein sequences specific peptide sites on the AFP molecule will be addressed and illustrated (see Figures 5A-L below).Amino Acid Sequence Matching Analysis: Multiple AFP AA sequence matching searches were performed using the GenBank database derived from GCG (Wisconsin Program) software. The identity and similarity of the various protein sequences matched in the GCG-GenBank database were detected using the protein FASTA program. To determine the significance of the match, FASTA employed a Z-score algorithm in order to evaluate identity/similarity relationships between protein and/or peptide AA sequences. For comparison between short peptide sequences (less than 40 AA), a wordscore of 1 to 2 is employed with a default set at 2.0 and above. Having set the wordscore, an E-value between 1 and 10 is considered statistically significant. Significant search results were then recorded and tabulated. Only identity/similarity matches totaling 30% and above are presented.
A) Immunoregulatory Sites and AFP Peptide Epitopes The major histocompatibility proteins: The major histocompatibility complex (MHC) genes encode three major sets of molecules; the class I, II, and III proteins. The Class-I proteins address the adaptive and innate immune systems employing CD8 cytotoxic T-cells (CTL), while Class II molecules are involved in immunological recognition as the major histocompatibility complex (MHC)-restricting CD4 helper (Th) cells; Class III molecules are concerned with the complement cleavage (lysis) cascade in the inflammatory response. While the antigen receptors expressed by B lymphocytes can recognize all types of antigens, most T-cells recognize peptides that are bound to class I or class II cell surface proteins are encoded by genes in the MHC protein family (Yamada and Hayami, 1983). Thus, T-cell class I recognition is largely confined to antigenic peptide fragments that are bound to the antigen presenting cells (APC) surface MHC proteins (Figure 3). A well-characterized activation signal for regulated secretion by T-cells (cytokines, granzymes, etc.) is engagement of the T-cell receptor (TCR) by peptides bound to MHCs on the surface of APCs during T-cells activation (Johnson et al., 2000) (Figures 3, 4). In a similar fashion, the MHC class II proteins encompass a great variety of proteins that serve to present antigenic peptides to CD4(+) helper T-cells (Kawai, 2007). In a classic immune response to class-II antigens, naive effector T-cells differentiate in two types of CD4 helper cells that can respond to foreign antigens. The first type, the T-helper type-1 (Th1) cells act to clear intracellular pathogens, while the second cell type, T-helper type-2 cells (Th2), control parasitic infections (Hanke et al., 2002). More recently, a third class of T-helper cell is induced by interleukin-17 and responds to autoimmune tissue injury and is referred to as Th17 cells (Ayaru et al., 2007).
Major Histocompatibility Complex (MHC) Class Proteins: The MHC class I molecules are essential for the functions of both the adoptive and innate immune response systems (Sherman, 2001). The MHC class I molecules are sub-divided into two groups, the class Ia and the class Ib types. The highly polymorphic MHC class Ia molecules have a seminal role in adaptive immunity, while MHC class Ib molecules display more limited forms and regulate innate immunity encompassing pathogens and tumors. The MHC class Ia molecules, termed human leukocyte antigen (HLA)-A, HLA-B, HLA-C, and HLA-DR are glycoproteins that present bound antigenic peptides to receptors expressed by T-cells. The class Ia molecules are multi-chain complexes which consist of a B2-microglobulin (B2-M) that is non-covalently associated with a heavy chain (Figure 2). The Ia heavy chain contains the antigen-binding clefts that serve as anchor points to typically bind short peptides of 8-9 AAs (Table 3) in length (Terabe and Berzofsky, 2007). The multivariate docking sites of the MHC class Ia molecules are localized within peptide antigen-binding clefts of the APCs which accommodate the presentation of a large number of antigenic peptides which can bind to clusters of co-segregating class Ia molecules (Figures 3, 4). The selection of antigenic peptides (8-10 AAs) that bind within the class Ia cleft is usually determined by two binding pockets that specify the primary anchor AA (AA) sites, with one or more secondary anchor sites that fine-tune the peptide binding motifs (Figure 4, Table 3). In contrast, the MHC class Ib molecules (including NKC receptors) often possess a larger number of primary anchor sites that are common to some class Ia molecules (Sherman, 2001). The differences in the peptide-binding repertoires between class Ia and class Ib appears to reflect the different roles of the two classes in the immune response.
HAFP Homologies: Identities and Similarities: HAFP displays a MHC class I AA identity peptide site on its second domain (Figure 4A) which overlapped an antigenic epitope. This AA sequence stretch was 23 AA long (AA313-336) with an AA identity (ID) of 39% and a similarity (Sim) of 56%. Class-II proteins consist of two noncovalently clustered peptides that traverse the plasma membrane at their COOH terminus. HAFP also displays a MHC Class-II identity site (AA #25-50) on domain I (Figure 4A). This AFP site demonstrated a 33% identity stretch (Sim = 52%) over 25 AA. Finally, the class II molecules (complement-associated) displayed AA identity sites on HAFP within all three AFP domains (see light gray-shaded areas, Figure 5A). The Class-III AA identity sequences ranged from 50-60% and were 62% identical over AA lengths from 10-12 AA residues. Based on these sequence homologies, it is tempting to speculate that HAFP may be involved (as a mimic or soluble receptor) in the immunoregulation of natural killer, MHC, and complement interactions at the onset of pregnancy, throughout gestation, and possibly in the neonatal/infant period (see below).The T-cell receptors: The immune system is able to distinguish non-self pathogens from self-proteins/tissues and this discrimination is mediated by the T-cell receptor (TCRs) family of proteins (De Mees et al., 2006). The TCRs recognize antigenic peptide fragments (epitopes) that are attached, bound, and displayed by the MHC molecules on the surface of the APCs (see Figure 3A bottom). The peptide-bound MHC cluster binds to a groove localized on the TCR molecular surface (Figures 3, 4). These peptide epitopes are generated by the proteolytic degradation of self or foreign proteins within cells expressing MHC class I or II molecules in a process known as antigen presentation (Dauphinee and Mizejewski, 2002). The T-cell recognition of foreign peptides is vital for defense against invading micro-organisms; however, recognition of self-peptides could lead to autoimmune disorders resulting from mutation accumulation, aging, disease, and viral/bacterial infections. Thus, the boundaries separating foreign, self, and altered self-peptides can be overlapping and it has been reported that viral/bacterial peptides are able to mimic self-peptides (Vakharia and Mizejewski, 2000; Harrington et al., 2005). The MHC proteins participate in antigen presentation via a three-way interaction involving the T-cell receptor (TCR), MHC molecules, and APC-processed (peptide) antigens (Figure 3). The T-cell receptors include multiple types composed of heterodimer combinations of α-, β-, γ-, or δ-chains, consisting of constant and/or variable chain regions. Both rodent and human AFP have been previously implicated in the regulation of immune responses of both humoral and cell-mediated types (Monaco et al., 1990; Butterfield et al., 1999; Cavin et al., 2004). Thus, it was not surprising that HAFP AA identity segments involving TCRs were detected on all three domains (Figure 5A). HAFP domain I bears a β-chain TCR site that overlapped an AFP epitope site and exhibited a 19-AA stretch bearing 39% identity, and Sim = 33% (AA #156-175); while the domain II site possessed a 19-AA sequence with a 47% identity, and Sim = 21%. HAFP domain II revealed a double site consisting of a 27-AA sequence length with a 41% identity (Sim = 25%) (AA #260-287). Finally, the third domain of HAFP demonstrated the presence of two identity sites composed of shorter AA sequence lengths representing the δ-chains of the TCR (AA #472-484; and AA #497-521), which displayed 40% identities over a 12-24 AA lengths (Figure 5A).
AFP Peptide Epitopes and the MHC Proteins: As shown above, AFP-derived peptide epitopes can be recognized by heterologous human T cells in the context of MHC class I antigens (Mizejewski, 2009). Researchers have now determined the identity of additional AFP-derived peptides, 8-9 amino-acids in length, presented as HLA-A*0201 restricted antigens, that could be recognized by the human T cell repertoire (Butterfield et al., 1999) (Table 3). After screening 74 peptides, one research group identified 3 new AFP epitopes, HAFP AA #137-145, HAFP AA #158-166, and HAFP AA #325-334, in addition to the previously reported HAFP AA #542-550 epitope (Figure 5A). Each peptide possessed an anchor hydrophobic AA residue at each end of the 8-9 mer peptide, and each stabilized HLA-A*0201 antigens on T-cells in a concentration-dependent MHC class I binding assay (Table 3). The AFP peptides showed stability for 2-4 h in dissociation-kinetic assays. The peptides also induced epitope-specific primed T-cells in vitro from several normal HLA-A*0201 donors. Importantly, the HAFP peptide-specific T-cells also were capable of recognizing HLA-A*0201(+)/AFP(+) tumor cells in both cytotoxicity and IFN-gamma induction assays. The immunogenicity of each peptide was assayed in vivo with HLA-A*0201 in Kb-transgenic mice following immunization with each peptide emulsified in adjuvant; also, draining mouse lymph node cells were shown to produce IFN-gamma cytokines on recognition of cells stably transfected with HAFP. AFP peptide-specific T-cells were also identified in the spleens of mice immunized with dendritic cells transduced with an AFP-expressing adenovirus (AdVhAFP). Three of the four AFP peptides were identified by mass spectrometric analysis of cell surface peptides derived from an HLA-A*0201 human HCC cell line. In summary, the authors presented the first and compelling immunological and physiochemical evidence which showed that four or more HAFP-derived epitopes were immunogenic, naturally processed, and presented in the context of MHC class I antigens. These investigators confirmed that such AFP epitopic sites could serve as prime targets for hepatoma immunotherapy (Butterfield et al., 2001). The immune response of lymphocytes was further studied after activation of dendritic cells (DCs) sensitized by a cytotoxic T lymphocyte (CTL)-based peptide from the second domain of human alpha-fetoprotein (HAFP AA #218-226 LLNQHACAV) (Figure 5A). High purity DCs were obtained from plastic-adherent monocytes cultured from healthy human donors of HLA-A2(+) peripheral blood which had been co-incubated for seven days with granulocyte-monocyte colony-stimulating factor (GM-CSF) and interleukin-4 (IL-4). Self-lymphocytes were also stimulated with DCs treated with the HAFP AA #218-226 peptide in a culture medium containing interkeukin-2 (IL-2) for 7 days. IL-12 and TNF levels were analyzed in culture medium and specific lysis activity of lymphocytes was assayed against four strains of primary hepatocellular carcinoma cells. After stimulation by DCs that were activated with HAFP #218-226 peptide, lymphocytes appeared unchanged and the culture medium of the activated lymphocytes contained high levels of Th1-type cytokines namely, IL-12 and TNF. Activated lymphocytes not only specifically lysed HLA-A2(+) HepG2 tumor cell line, but also showed cytotoxicity against three other primary hepatocellular carcinoma cell lines and against T2 hepatoma target cells activated with the HAFP peptide. The results of these experiments provided the underpinnings for developing a DC-based vaccine with a HLA-A2 restricted peptide epitope derived from an HAFP epitope directed against AFP positive primary HCC cells (Deng and Mariuzza, 2007).
B) Extracellular Matrix, Integrins, and Cell Adhesion Molecules The major classes of molecules that regulate cellular development and physiology include growth and differentiation factors, cell adhesion molecules, and components of the extracellular matrix (ECM). Components of the ECM and their receptors interact with virtually all cell types (including the nervous system) in embryonic and adult organisms (Reichardt, 1993) (Figure 5B). Thus, the ECM presently is defined to include essentially all secreted molecules that are immobilized in the matrix outside of cells. The functions of these diverse molecules include inhibition of neurite outgrowth, cell migration during embryogenesis, modulation of synaptic transmission, anti-adhesive activities, growth factor interaction, ECM anchoring, binding of proteases and/or protease inhibitors, initiation of intracellular signaling cascades, cytoskeletal signal transduction and pH modulation of Na+/H+ antiporter activity (Reichardt, 1993). The cell adhesion molecules (CAMs) mediate cell-cell and ECM interactions during embryonic development, maintenance of adult tissue architecture, the inflammatory response, wound healing, and tumor metastasis. The adhesion molecules and/or receptors thus influence many biological activities and processes including cell proliferation, differentiation, cell junction formation, and polarity by targeting cell-surface adhesion to specific ECM proteins/ligands on adjacent cells. Multiple families of adhesion molecule/receptors have been identified as the following: a) the heterodimeric integrin receptors; b) the immunoglobulin-like adhesion molecules; c) the homophilic cadherin calcium-dependent proteins; d) the LEC-CAMs with lectin-like domains; and e) homing receptors that target lymphocytes to specific lymphoid and nonlymphoid tissues (i.e., CD44) (Albelda and Buck, 1990). The finding that cells can respond to signal transduction from the ECM via integrin receptors is highly significant (Legate et al., 2009). The signal apparently is transduced through the integrins linked to intracellular talin, vinculin, and actinin which transmit to the actin cytoskeleton. Extensive signaling "cross-talk" can occur between the receptors and other CAMS involving both tyrosine kinases and polyphosphoinositide hydrolysis. It is thought that induction messages of early immediate genes can then be transmitted to the nucleus via actin-associated cytoskeletal elements (Legate et al., 2009). HAFP appears to display multiple adhesive sites of the short variety (3-10 AA) in addition to longer sequence homologies to the adhesive molecules themselves (Figure 5B). Examination of the primary AA sequence structure of HAFP reveals the presence of both RGD (AFP AA #271-273) (arginine-glycine-aspartate) and LRE (leucine-arginine-glutamate) CAM recognition sites (Pierschbacher and Ruoslahti, 1984; Yamada and Kennedy, 1985). The RGD functions as a cell-attachment site within several different extracellular matrix glycoproteins including fibronectin, thrombospondin, von Willebrand's factor, and vitronectin. The LRE triamino sequence (AFP AAs 213-215) is a crucial determinant for the binding of various motoneurons to s-laminin and also serves to inhibit neurite axonal outgrowth at synaptic sites (Hunter et al., 1991). HAFP, as shown in Figure 5B, displays some sequence homologies with a multitude of ECM and adhesion proteins as detected in the GenBank. Examination of the ECM sequence sites identified in the figure revealed different matrix-related molecules including plasma-secreted, cell surface, basement membrane, enzyme, and fibril forms. It can be observed that many AA sequence identities were localized on HAFP domain II with lesser sequences on domain I, and fewer still on domain I. In fact, the majority of ECM-protein homologies resided on domain II. Overall, the AA sequence identities ranged from 30% to 47% over stretches of 17 to 28 AAs in length, respectively. The proteins included collagens, fibrinogen, and thrombotic molecules related to clot formation/dissolution, platelet attachment, and serine protease activities. It is evident that HAFP domain II and to a lesser extent, domain I, are associated with ECM activities, which encompass hemostasis, basement membrane architecture, fibril scaffolding and crossbanding, cytoskeletal attachment, and signal transduction. The second domain of AFP has short peptide sequences common to ECM proteins bearing short sequence cellular adhesion motifs (CAMs). These findings distinguish AFP from other albuminoid family members whose proteins lack the short peptide sequence similarities to the ECM protein family (laminin, fibronectin, collagen, vitronectin, thrombospondin, etc.) of macromolecules (Yamada and Kleinman, 1992). Adhesive ECM macromolecules have potential utility in the study of developmental stages 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 the ECM molecules (Tashiro et al., 1989; Nomizu et al., 1993; Tuszynski et al., 1993). Some of the CAM-derived synthetic peptides have been found to block cell differentiation, tumor growth, and angiogenesis (angiostatin). Several ECM proteins (laminin, collagen, fibronectin) contain a whole host of biologically-active peptide sequences with differing activities specific to cell type (Grant et al., 1989). Furthermore, 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. Human AFP (especially the second domain) contains many short peptide sequences that are common to the ECM-derived CAMs. Thus, HAFP 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 extracellular matrix proteins (see Refs 72-89, listed in Figure 5B were compared to human AFP, ALB, αALB, and DBP numbered peptide sequences (8). Upon inspection of the matched peptide sequences, it was observed that AA matches occurred on AFP with CAM-like sequences in the second domain of AFP (AA #212 to 388). Exact AA matches appeared less frequently on human ALB and DBP molecules. It becomes apparent that: 1) a large number of diverse cellular adhesion activities might be involved and; 2) 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 integrin molecules as previously suggested (Mizejewski, 1997). However, such capabilities of AFP might ultimately depend on its ligand-bound state and extracellular localization during growth. It is of special interest that a periodic spacing of the signal recognition ECM 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 (AA #213-215). There follows a gap of approximately 40 AAs before a second signal peptide (LDV) occurs at HAFP AA #260 to 262. It is at this point that a pattern of regularity arises. The next four signals (including LDV) occur approximately every 10 AA sequences from AA #259 to 289, 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 ((Carter and He, 1990); see Figures 2B and 5). The next proposed signal peptide at AFP AA #327 (PNLDR) is similar to laminin B1, followed by a scrambled RGD at AA #334, and a downstream sequence at AA #370-375 (ILRVXK) comparable to laminin-A. Scrambled RGD is known to be a growth inhibitor sequence, AA #335-337 (Kijimoto-Ochiai and Noguchi, 2000). Teleologically, it would be important if an abundant fetal protein were to possess adhesion recognition sequences in its primary structure. Such adhesion capabilities would provide a protein with the modulatory properties of cell growth and differentiation, cellular attachment and migration, cytoarchitectural structuring, and morphogenetic cell movements and migrations during both embryogenesis and histogenesis. Examples of such regulation have been reported with the integrin family receptors, beginning at gastrulation and continuing throughout development, involving processes such as neural and myoblast movements, sex cell migrations, and neurite extension on Schwann's cells (Albelda and Buck, 1990). In addition, some adhesion molecules (i.e., tenascin) are known to consist of both adhesive and anti-adhesive intrinsic domains or signal-recognition sites (Spring, 1989). Hence, it would not be surprising to identify ECM, adhesion, and anti-adhesion signal-like AA sequences on different portions of the AFP molecule. Such potential AFP matches have been detected from computer-derived sequence matching data (see Figure 5B).
C) Enzymes, Fibrin Binding, and Serine Proteases Human tissue-type plasminogen activator (TPA) is a serine protease enzyme responsible for dissolving fibrinogen in blood clots (Banyai et al., 1983). Interestingly, an AA sequence site detected on HAFP domain I (Figure 5C), located by computer GenBank analysis, showed a 35% sequence identity over 20 residues to γ-fibrinogen (Figure 5B, AA #83-102) and antithrombin-III (AA #29-54). Both plasminogen and TPA are bound by the fibrin polymer in clots to facilitate enzyme: substrate binding alignment. It has further been reported that certain human TPA residues between AA #20-148 display significant sequence peptide corresponding to the last 39 residues at the carboxyl terminus (domain III) of HAFP (AA #571-609) (Baker, 1985). A portion of this segment on TPA (AA #82-128) overlaps a region that was previously shown on AFP to be related to the epidermal growth factor motif module (Gray et al., 1983). TPA identity sites have also been detected on HAFP at the border of domains II and I. The domain II-I overlap site showed 32% extending over 19 residues. An NH2-terminal portion of TPA sequence (AA #26-62) contains the fibrin-binding finger-domains for the serine protease/substrate contact and this region of TPA resembles HAFP residues 433-461 in the third domain of the AFP molecule (Figure 5C). Interestingly, the region (AA #423-444) is known to be a major hydrophobic binding pocket of HAFP, which corresponds to the estrogen-binding site of rodent AFP (Nishi et al., 1991). Residues crucial for estrogen binding on rat AFP are AA #s 428, 430, 433, 434, and 435 (Nishi et al., 1993). Thus, the estrogen binding pocket of RAFP may reside along side or overlap a portion of a site that could function as a serine protease site for binding substrate. If this were the case, AFP might bind serine protease substrate and/or inhibitors, but not cleave them as was previously proposed (98). Indeed, a search of the literature reveals a series of reports that are consistent with this hypothesis (refs. (Mizejewski et al., 1975; Baker et al., 1978; Scott et al., 1989; Luscher and Eisenman, 1990a; Mahoney et al., 1991; Kato et al., 1992; Gehring et al., 1994; Manak, 1994; Olson and Rosenthal, 1994)). Other enzyme-related homologies include superoxide dismutase, kinesins, tyrosine kinases, cyclin-dependent-kinases, ATP binding protein, protein kinases, and many others (Figure 5E).
D) Oncogene Transcription and ExpressionTranscription factors, tumor suppressors, and homeodomain proteins. The oncoproteins are nuclear phosphoprotein products of the proto-oncogenes, which now include ras, rel, myc, myb, fos, jun, ski, ets, cbl, erb A, and many others (Luscher and Eisenman, 1990a; Secombe et al., 2004). Most oncoproteins were determined to be transcription factors, which function as molecular switches that sense incoming signals and modulate the transcription of specific genes (Lewin, 1996; Hess et al., 2004). Although predominantly localized in the nucleus, they are capable of mediating and shuttling specific transcriptional responses to signals originally generated in the plasma membrane or cytoplasm (Martin, 1991; Bernards and Settleman, 2009; Montagut and Settleman, 2009; Plentz et al., 2009; Quinlan and Settleman, 2009; Sharma and Settleman, 2009). Some, like myc, are present in nearly all cell types, while others like myb, are restricted largely to hematopoietic cells (Luscher and Eisenman, 1990b; Secombe et al., 2004). Many oncoproteins have functional partners with which they heterodimerize to bind DNA, such as fos and jun, myc and max, etc. Like other proteins, the oncoprotein transcription factors are synthesized in the cytoplasm and transported into the nucleus via a nuclear localization signal within their primary structure (Hess et al., 2004; Quinlan and Settleman, 2009). Protein-protein and protein-DNA interactions of the nuclear oncoproteins are often mediated through helix-loop-helix and leucine zipper motifs as discussed below (Mizejewski, 1993; Babu et al., 2004). Among the transcription factors, the discovery of the homeobox gene superfamily that translates the homeodomain proteins represents a hallmark discovery in embryonic development (Gehring et al., 1994; Manak, 1994). These embryonic induction factors have long been sought since Hans Speman first speculated their existence (induction factors) in the early 1900s. First discovered in insects, homeobox proteins and their homologs are now known to exist in nematodes, rodents, humans, plants, and yeast (Scott et al., 1989). Homeoproteins serve to direct and control pattern/positional body development in embryos concerning anterior/posterior, trunk-thorax segmentation, dorsal/ventral axis, body polarity, neural-tube formation, and caudal/gut formation. In fact, birth defects are often homeotic transformations (or mutations) resulting in developmental anomalies in which one part of the body develops in the likeness or dissimilarity of another in contrast to bilateral symmetry. Pattern formation in the embryonic germ layers usually involve a network of feedbacks between intrinsic programs of gene expression in developing precursor cells and extrinsic signals exerted by the surrounding embryonic environment (Olson and Rosenthal, 1994). The homeotic proteins frequently direct 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 of that particular germ-layer derivative. Examples of the homeodomain proteins would include Pou, antennepedia, crumbs, Wnt, Sonic Hedgehog, forkhead, and Pax (Manak, 1994) (Figure 5D). Since the homeodomain proteins are present during early embryogenesis, it would seem reasonable that AFP displays short homeodomain sequences in molecular mimicry of these pattern-recognition 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 (Mizejewski, 1997). In mammals, Pou domains are expressed during early embryogenesis in many regions of the developing brain including forebrain and nerve cord (Figure 5D). 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 (Wingless) 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 AA sequence homologies with the homeodomain proteins, which are endowed with embryonic body positional information. Some of the identified transcription factors, tumor suppressors, and homeodomain protein sequence homologies to HAFP are cataloged in Figure 5D. As shown in the figure, AA matching identities for the transcription factors (ranging from 29 to 54% over lengths from 13 to 24 AA) are found largely in HAFP domains I and II. In comparison, the tumor-suppressor (retinoblastoma) protein (141) identities appear to reside largely on domain I of HAFP. However, a fat-protein (cadherin-related) site (Mahoney et al., 1991) was detected on domain III, demonstrating a 100% AA identity over only a 5-AA length. Thirdly, the homeodomain proteins are found to exhibit AA identities ranging from 41% to 67% over stretches of 10-27 AA in AFP domains I and II. However, shorter stretches of AAs were also detected on domain III (67%-100%) including homeodomain signature motif that is characterized by only four conserved AAs and residues in the C-terminal third domain of all mammalian homeotic proteins including AFP (Scott et al., 1989). Another 8-AA segment of homeotic proteins is highly conserved, though not invariant, in stretches of AA that were either amino terminal or carboxyl terminal to the homeodomain signature sequences (HS in Figure 5D). Many oncoproteins have been implicated in the control of normal cell growth and proliferation, in cell-cycle control, and in apoptosis. In cancer, they may be involved with aberrant growth in mutations that activate the oncogenic potential of cellular proto-oncogenes and cause loss of control of normal cellular proliferation and differentiation. Such activating mutations, as seen in growth-suppressing proteins (p53 and retinoblastoma protein), which bind transcription factors, constitute fundamental steps in the development of human cancers. Although the precise steps in molecular oncogenesis are unknown, mutations can inactivate certain oncoprotein properties such as nuclear localization, transcriptional activation, dimerization, and DNA binding (Kato et al., 1992). In the case of c-erb A (Figures 5H-I), dominant-negative mutations can inhibit the transcriptional regulatory function of the thyroid hormone receptors, resulting in a loss of hormone responsiveness and hormone-induced differentiation (Damm, 1993). This is comparable to the tumor-suppressor genes, where a loss of function can induce the transformation process. In this way, normal hormone-activated nuclear receptors, such as the thyroid and retinoic acid receptors, can function as growth suppressors since their differentiated cellular targets lose their proliferative potential (Liu et al., 1995).
E) Apoptosis (Programmed Cell Death) Mammalian development is accomplished by a combination of cell proliferation, differentiation, and apoptosis or programmed cell death (Figure 5E). Apoptosis routinely occurs during embryogenesis, histogenesis, metamorphosis, endocrine-dependent tissue atrophy, and normal adult tissue turnover (Gerschenson and Rotello, 1992; Cohen, 1993). Apoptosis can be readily distinguished from necrosis, a condition that results from pathologic injury, complement-based cellular attack, severe hypoxia, hyperthermia, lytic viral infection, and toxin exposure. Apoptosis is characterized by nuclei condensation and segmentation, DNA degradation, membrane microvilli loss, and cell surface blebbing. Apoptosis occurs naturally in the immune system (lymphocytes, thymocytes) induced by antigen-receptor complexing or by glucocorticoid induction (Caron-Leslie, 1994; Hockenbery, 1994). NKCs or cytotoxic T cells also exhibit apoptotic induction in target cells (see Figure 5A) and an NKC site overlap the FAS region on HAFP (Figure 5E). Tumor regression is often mediated through apoptosis as a result of UV or, X-irradiation, and chemotherapeutic exposure in cancer cells (Nagata, 1994). Throughout mammalian development, and especially in the early embryo, a competition exists between cell death signals and survival (rescue) signals. For example, gastrulation in the mouse embryo is accomplished by two signals; an apoptotic induction that forms the inner cell mass and a rescue (reversal) signal that allows the survival of columnar cavity-lining cells (Lewin, 1996). Although the programmed cell death signal occurs in the embryonic inner mass cells, the rescue command is thought to be contained within the extracellular matrix cells in juxtaposition to the ectoderm. It is in this fashion that the embryo forms tubes with a single layer of cells lining the lumens. Cell death helps to create lumens (cavities) in a variety of embryonic structures, with the dead cells engulfed by phagocytosis. It has further been shown that the integrin receptors of the ECM play a role in regulating apoptosis during mammary differentiation in the mouse (Lewin, 1996). One previous study demonstrated that AFP and AFP-receptor antibody blocked the induction of apoptosis in HL-60 leukemia cells in culture (Laderoute et al., 1994). The induced cell death by AFP was associated with cellular adherence in microtiter plates. Cell death was assessed by morphology, shrinkage of dying cells, inducibility and reversibility kinetics, and DNA fragmentation patterns. Concomitant studies by others had already shown that cells cultivated on substrates that prevent cell adhesion rapidly progressed into apoptosis (Collins, 1987). It has now been ascertained that the signal for programmed cell death is mediated by a cell-surface transmembrane protein termed the Fas antigen belonging to the tumor-necrosis factor/nerve-growth factor receptor family (Caron-Leslie, 1994). Likewise, it has determined that the rescue signal from apoptosis is conferred by the cytosplasmic Bcl-2 protein derived from β-cell leukemias and from lymphomas (Hockenbery, 1994). As stated above, a delicate balance exists during mammalian development regarding cell death and rescue (survival) signals for cell growth, proliferation, and differentiation. Hence, it should be feasible to detect remnants of both signal types on individual growth-regulating proteins such as AFP. The elegant studies by Dudich et al. have revealed that AFP blocks apoptosis inhibition while promoting cell death effectors (caspases) in tumor and cell-free systems (Dudich et al., 2000; Semenkova et al., 2003; Dudich et al., 2006). On the basis of GenBank identification, an AA sequence resembling a Fas-like peptide stretch (AA #59-86) is readily detectable on domain I (Figure 5E) of HAFP (39% identity; 23 AA in length). In a similar fashion, slightly downstream on domain I of HAFP, a sequence stretch identifying with the tumor necrosis factor receptor type 1 and TNF (AA #96-120) itself can also be discerned (50% identity, 12 AA in length). Both stretches of AAs lie in juxtaposition to each other on the first domain of HAFP. One could speculate that dimerization or binding of AFP to these signal proteins could blunt the cell-death signal resulting in enhancement or continuation of cell growth. In contrast, a rescue/survival signal represented by a Bcl-2-like AA stretch (AA #476-484) was detected at the amino terminal end of HAFP domain III (41%, 19 AA in length). Interestingly, this Bcl-2 site is localized adjacent to APO-6 (Figure 5E) and near a proposed hinge region, which allows rotational flexibility. A conformational change in the tertiary structure of AFP, possibly induced by ligand binding or shock signal, could expose a rescue site (i.e., Bcl-2), which was normally hidden in a molecular crevice (Mizejewski et al., 1996). As above, one might speculate that dimerization or binding of AFP to a Bcl-2 protein could block the rescue/survival signal, resulting in the induction of cell death. A fetal protein such as AFP might be expected to display a FAS-like growth enhancement (domain I) segment in its normal molecular configuration, while a concealed Bcl 2 interface might surface following a stress-induced conformational change. In this fashion, AFP could function to enhance growth while conditions conducive to stress or shock (heat or glucose shock, hypoxia, hyperthermia, excessive ligand exposure) might expose a signal site resulting in cessation of growth or cell death. In this manner AFP could function in both the up- and down-regulation of growth by employing a binding or dimerizing mechanism to apoptotic mediators. Several other proposed apoptosis sites labeled APO-1 to 6 are shown in Figure 5E in addition to sites of peptide to tumor necrosis factor (TNF) and its receptor (TNFR).
F) Ligand Binding and Carrier/Transport Peptide Sites There exists overwhelming evidence that AFP is an embryonic and fetal carrier/transport molecule for a multitude of ligands including fatty acids, bilirubin, heavy metals, steroids, retinoids, drugs, dyes, and antibiotics (Hirano et al., 1984a; Hirano et al., 1984b; Hirano et al., 1985). However, the precise binding location on the AFP molecule is known for only some of those ligands. For example, a major fatty acid-binding site for long-chain fatty acids has been documented to lie between AA #227 and 246 on HAFP domain II (Nishihira et al., 1993). Lysine AA #242 appears to be one of the most essential AAs for this fatty acid binding site. Studies employing Scatchard binding/saturation analysis had previously demonstrated that at least three possible ALB binding sites (Ka = 10-7 M, n = 3) exist for the polyunsaturated fatty acids (i.e., arachidonic and decosohexanoic acids) (Parmelee et al., 1978; Aussel and Masseyeff, 1983). If AFP is similar to ALB in this regard, then it is probable that one fatty acid-binding site exists on each of the three domains, and such may indeed be the case (Reed, 1986). The remaining two fatty acid-binding locations can be proposed on the basis of GenBank-derived AA comparisons to fatty acid-related proteins. One such example of a proposed HAFP domain I site (AA #60-79) could reside adjacent to AA-residues 55-88 that shows an AA peptide to fatty acid synthetase (28% identity, 32 AA). A proposed folic acid binding site is also proposed to occur at this position (AA #64-75). A second possible site on domain I could reside at AA #189-222, wherein lies another fatty acid protein-like stretch (32% identity, 19 AA in length) together with a Vitamin-D binding site. On domain II, a fatty acid denaturase homolog site was identified on AA #242-256 overlapping the confirmed fatty acid-binding site at AA #227-246. The third AFP binding site resides on domain III (AA #437-460) and appears to overlap the estrogen-binding site homologous to rat AFP according to previous competitive-binding reports (Benassayag et al., 1979; Mizejewski, 1995b). The estrogen-binding interface on rodent AFP has been determined to occupy a region between AA #438 to 452 (human AFP AA #445-459) forming an α-helix segment lying adjacent to a β-sheet/turn structure extending from residues 464 and 498 (Nishi et al., 1991; Nishi et al., 1993). The former site represents a major hydrophobic binding pocket on HAFP that binds little or few estrogenic steroids, in contrast to the rodents (Swartz and Soloff, 1974). In humans, a high affinity E2-binding region (AAs 445-459) displays overlapping binding sites for fatty acids, diethylstibesterol (DES), protease inhibitors/substrates, retinoids, warfarin, coumarin, phenylbutazone, pyrazolic drugs, and anthranilic acid (Hirano et al., 1984a, 1985). A low affinity E2 binding site on HAFP extends from AA #464 to 498 (Vakharia and Mizejewski, 2000). Competitive-binding studies employing rat AFP determined that estrogen was indeed bound on RAFP domain III together with retinoids and fatty acids (Aoyagi, 1979). Employing competitive inhibition-binding analysis, it was determined that bilirubin did not compete with fatty acids for the same binding sites on HAFP, and both sites were totally distinct from the retinoid-binding region (Aoyagi, 1979; Hirano et al., 1984b; Bansal, 1990). Earlier studies had already ascertained that HAFP possessed two separate binding sites for bilirubin while ALB has three sites for this heme pigment (Berde et al., 1979). The former study concluded that HAFP bound both fatty acid and bilirubin noncompetitively on domain I and domain II, while steroids and retinoids (lesser than fatty acids) were bound on domain III (Berde et al., 1979). Regarding bilirubin-binding on HAFP, the two sites appear to reside on either side of the single tryptophan residue on HAFP as determined by spectral and competitive-binding analyses. With such data, it can be speculated that the location of the two bilirubin binding regions in light of GenBank AA identities. The domain I site could be positioned in the amino-terminal direction to the tryptophan (present at AA #181) between AA #176 to 186. At this position, HAFP AA identities were found with pigment-associated proteins (β-globin, 40%, 20 AA in length; chloroplast protein, 35%, 29 AA in length). In comparison, the domain II site might be located in the carboxy-terminal direction to the tryptophan residue and to the proposed fatty acid-binding site. The second bilirubin site is proposed to lie between AA #282 and 290, where Vitamin-D and pigment-related protein AA identities were also detected (heme oxygenase, 33%, 12-AA length; cytochrome-c oxidase, 82%, 11-AA length; a chloroplast protein, 46%, 13-AA length; and phycoerythrin, 58% identity, 19-AA length). HAFP has been shown to bind to heme (Zizkovsky et al., 1983) and that binding could occur at this latter site. It may be of interest that a benzodiazepine receptor peptide site (Figure 5H) has also been identified at this same region (AA identity 38%, 24-AA in length) that also includes an adjacent serotonin receptor region at AAs 277-289 (AA identity 42%, 12-AA length). Both of the latter compounds bind nitrogen-ring aromatic hydrocarbons that mimic those of bilirubin. In contrast, ALB has these latter two binding sites on domain III instead of domain II (Kragh-Hansen, 1991). The binding sites of heavy metals (Cu2+, Zn2+, Pb2+, etc.) on HAFP have only been approximated; however, multiple binding sites have been reported (Aoyagi et al., 1978; Wu et al., 1987), in accordance with AFP purification methods using metal-chelated column chromatography (Andersson et al., 1987). Early studies of ALB binding to heavy metals frequently have implicated the first domain (Aoyagi et al., 1978) and AFP was found to be similar. Published reports have supported the involvement of histidine residues in the binding of Cu2+ and Zn2+ ions, probably at the imidazole group (Aoyagi et al., 1978; Wu et al., 1987) since AFP has been reported to bind tryptophan (Ingvarsson and Carlsson, 1978; Baker et al., 1980). Cysteine residues have also been implicated with Zn2+ in binding to DNA (Wahli and Martinez, 1991); however, HAFP has only two free cysteines, located at AA #37 and 86. These residues of free cysteines occur in the first domain of HAFP in keeping with the ALB metal-binding sites. It may be no mere coincidence that HAFP sequence identities with Cu/Zn-superoxide dismutase occur at this region (Figure 5C) from AA #29 to 54 (31%, 25-AA length) (Figure 5J). A Cu2+-transporting ATPase identity stretch (26%, 19-AA length) was also found between AAs 156 to 175, wherein lies an Arg-His-Pro sequence in domain I. On the second domain of HAFP, a histidine-rich region (Figure 5J) was also found at AA residues #262-270 that contains three histidines. This site would qualify as a Cu2+- and/or Zn2+-binding region since it overlaps a Cu2+-transporting ATP identity site (33%; 21-AA length) located there (Figure 5F). From these data, one could propose that heavy-metal binding can largely be assigned to HAFP domains I and II.
G) AFP Peptide Sites for Hormones and Growth Factor Binding Reports of AFP interactions with a variety of hormones can readily be found in the literature (see below). Such findings include hormones such as follicle-stimulating hormone (FSH), luteinizing hormone (LH), thyroid hormones, and growth factors. Previous studies showed that AFP was indeed subject to hormonal, nutritional, and hematological regulation (Belanger, 1975). These reports indicated that prednisolone, adrenaline, and thyroxine depressed AFP through a selective blockade of its synthesis. In the case of thyroxine, decrease of AFP synthesis has been confirmed both in cell culture studies (Anteby et al., 1993) and in clinical follow-up of hypothyroid newborns (Mizejewski and Pass, 1992). In contrast, estradiol (E2) and estrone (E1) treatment induced the synthesis of AFP in the livers of adult rodents (Aussel, 1976; Kotani, 1987). In vitro studies further have shown that AFP was capable of inhibiting the formation of water-soluble metabolites of E1 and E2 by incubation with microsomes from rat livers in the presence of NADPH (Aussel, 1976). The addition of purified AFP fractions into newborn rats bearing E2-sensitive pituitary tumor cells also prevented oncogenic growth (Sonnenschein and Soto, 1979; Sonnenschein et al., 1980). The authors concluded that AFP was a specific inhibitor of the multiplication of cells that were E2 sensitive for growth. Because of such reports, the author of this atlas chapter undertook a series of studies to investigate and hopefully elucidate the E2-sensitive growth inhibitory properties of AFP in vivo (Mizejewski et al., 1983, 1986; Mizejewski and Warner, 1989; Jacobson et al., 1990a; Mizejewski et al., 1990; Allen et al., 1993). These studies concluded that rodent and human AFP contained an encrypted (occult) E2-sensitive growth regulatory site that is induced to emerge following exposure to high estrogen concentrations (Mizejewski et al., 1990; Mizejewski et al., 1996; Vakharia and Mizejewski, 2000). Such a site could account for the prevention of fetal hyperestrinism in the face of high maternal estrogen levels during pregnancy as first purposed by Vannier and Raynaud (Vannier and Raynaud, 1975) and later addressed by Nunez et al. (Nunez, 1976; Nunez et al., 1989). It was this hidden E2-sensitive growth regulatory site on HAFP that was produced and studied as a purified 34-mer peptide (Mizejewski et al., 1996). Growth enhancement by AFP has long been reported in the scientific literature (Mizejewski, 2001a, 2002). The question of whether AFP contains bona fide growth factor or mitogenic activity was addressed by recent reports implicating protein kinase-A pathways (Li et al., 2002a). However, the complete mechanism of the growth-regulatory properties exhibited by AFP in a decade of reports is yet to be fully elucidated. In this regard, an AA sequence peptide site related to growth hormone (46% over 11 AA) was found on the first domain of HAFP (Figure 5H). A previous report had demonstrated that purified HAFP, together with platelet-derived growth factor, synergistically enhanced the proliferative activity of human medullary breast-carcinoma cell cultures (Leal et al., 1991). It may be germane to this report that HAFP displays an AA stretch on domain I (Figure 5H) with identity to the platelet-derived growth factor-α receptor (29% identity, 21-AA length). The synergistic action of HAFP with various growth factors has also been reported in porcine ovarian granulose cell cultures. It was initially found that physiological levels of HAFP could enhance the mitogenic activity of epidermal growth factor (EGF) and transforming growth factor-α (TGFα) (domains II and III), suggesting that AFP might serve to modulate growth factor-mediated proliferation during development and neoplasia (Leal et al., 1990; Keel et al., 1991). Further studies on these porcine cell cultures revealed that AFP in the presence of EGF plus insulin-like growth factor-α1 (IGF-α1) and platelet-derived growth factor (PDGF) also produced the growth-enhancing effect (Keel et al., 1991). It may be of interest that a PDGF receptor-like sequence has been detected on the first domain of HAFP and IGF sites were found on domain II and III (Figure 5H). Subsequent use of the granulosal cell cultures seemed to indicate that, while HAFP was capable of growth factor-enhanced growth of these cells in vitro, HAFP also inhibited the steroidogenic function of E2 production (Keel et al., 1992). It was later demonstrated that HAFP actually inhibited the FSH-stimulated E2 production in the porcine granulosa cells (Keel, 1993). Those authors suggested that AFP could be inhibiting differentiated functions such as aromatase enzyme activity while enhancing cell proliferation. It could be speculated that the FSH-like sequence shown in Figure 5H may have relevance to this observation. It is readily observed in Figure 5G that a large variety of hormone and growth factor peptide sites are present representing reproductive, growth, and metabolic moieties.
H) G-Protein Coupled Receptor Signaling Unlike cytoplasmic signaling, G-protein signaling cascades begin at a cell surface heptahelical receptor (Figure 3, Panel-1). The G-proteins are heterotrimeric peripheral membrane proteins, anchored to the basolateral portion of the cell membrane by both intrinsic hydrophobicity and by lipid modifications (prenylation) of their subunits, largely the α and γ types (Lefkowitz, 2000). The G-proteins connect to the seven-transmembrane (heptahelical) receptors of the cell membrane that, in turn, affects downstream signaling pathways via protein-to-protein scaffold interaction (see (Mizejewski, 2002)). The G-protein coupled receptor (GPCR) activation (by ligand binding) triggers the exchange of GTP for GDP by the Gα subunit, of the heterotrimeric G-protein complex; this induces a conformational change in the Gα subunit which facilitates its dissociation from the receptor and the Gα subunits that bind and activate downstream effector pathways (Foord and Marshall, 1999; Lefkowitz, 2000). The heptahelical receptors, which become linked to G-proteins, interact with extracellular receptor kinases (ERKs) and adaptor proteins responding to neurotransmitter, hormonal, light, taste, odorant, and growth factor stimulation (Lefkowitz, 2000) (see Figure 5H). Signal transduction of this type can be uncoupled by a physiological process termed "agonist desensitization" (Katragadda et al., 2001). In this process, a blunting of the second messenger responses occurs following prolonged or excessive agonist stimulation (Bockaert and Pin, 1999). Following agonist binding to the cell surface GPCRs, endocytosed clathrin-coated receptosomes are generated (Mizejewski, 2001b). The endocytotic clathrin-induced pathway recruits a G-protein-associated adapter protein termed arrestin and the mechano-enzyme, dynamin (Freedman and Lefkowitz, 1996; Bunemann, 1999). Arrestin binding to the receptor uncouples the G-proteins from the pathway so that no further kinase phosphorylation cascades can occur (Clark et al., 1999; Oakley et al., 1999; McNiven et al., 2000). Clathrin is bound by arrestin and assisted by dynamin in trafficking the receptosome protein/peptide cargo to either an endosome or proteosome destination. The dyamin enzyme mediates the fission (vesicle pinching-off) step subsequent to the initial coated-pit formation in the membrane vesicle recycling process (McNiven et al., 2000). However, rapidly repeated agonist stimulation depletes the cell surface receptor population (without subsequent replacement), which renders the cell insensitive to further stimulation. Receptor blockade, which is related to agonist desensitization, can occur when a decoy agonist ligand binds to the receptor and the signal is insufficient to induce the proper conformational change needed to activate the cell surface receptor to bind the G-proteins; this deprives further ligand binding to the GPCR (Clark et al., 1999; Oakley et al., 1999). AFP and its derived peptides might be able to desensitize or quench the GPCR signal transduction pathways and functionally impair the cellular growth response of multiple cell types, including tumor cells. Examples in Figure 5H include a multitude of receptors in domains II and III. Thus, AFP and other protein-derived peptides could serve as decoy ligands for GPCR, they would bind to the receptors but would not induce subsequent ERK-protein cascade interactions. In turn, this could induce a refractory state in the mitogenic response to MAPK interactions in the G-coupled pathway that regulates downstream mitogenic activities. Such a pathway would represent a non-activating ligand occupancy of the GPCR, thereby depriving the cell of further signal transduction as previously demonstrated (Krupnick et al., 1997; Ghanouni et al., 2001). The various G-protein coupled cell surface receptor peptide sites displayed in Figure 5H include those for serotonin, thyroid, benzodiazepine, platelet-derived growth factor, epidermal growth factor, fibroblast growth factor, somatostatin, and acetylcholine.
I) Heterodimerization and Nuclear Localization Signal Sites Leucine zipper motif. In order for AFP to influence growth positively or negatively at the transcription level, one possible mechanism might involve heterodimerization with nuclear receptors or nuclear accessory proteins via a leucine (repeat) zipper-like motif in the ligand-binding domain. Such motifs have been previously proposed and reported for AFP (Mizejewski, 1993; Dauphinee and Mizejewski, 2002). The carboxy-terminal third domain of murine AFP is known to include an estrogen-binding region as well as retinoic acid-, thyroxine- and fatty acid-binding sites (Hsia et al., 1980; Herve et al., 1982; Aussel and Masseyeff, 1984). The estrogen-binding region in rodent AFP has been identified and mapped to the amino-terminal side of the third domain (AA #420-506, high affinity region 447-480, low affinity) (Nishi et al., 1991) just downstream from a known 'hinge' region. A putative leucine zipper motif appears to be located downstream (toward the carboxy terminus) from the ligand (E2)-binding site. Indeed, such a series of nine sequential heptad repeats have been described (Mizejewski, 1993), beginning with AA #515 in human AFP. All nine heptad repeats extend from AA 515 to 606 in human AFP and are presented in Figure 5I. Many of the postulated requirements for these motifs to qualify for heptad eligibility appear to be satisfied within the third domain of both mammalian species of AFP. These proposed criteria state that positions 1 and 8 of the heptad repeat contain one of the following hydrophobic AAs: leucine (Leu), methionine (Met), isoleucine (Ile), valine (Val) or phenylalanine (Phe). The fifth position in the heptad contains either hydrophobic residues or the charged AAs, arginine (Arg) or glutamic acid (Glu). The AAs listed above at heptad positions 1, 5, and 8 are especially well conserved among the nuclear receptors, particularly the T3R, VDR and RAR receptors (Wahli and Martinez, 1991). As described by Forman et al. (Forman et al., 1989) and Landschulz et al. (Landschulz et al., 1988), the folding of the well-conserved 1, 5, and 8 positions of the heptad into coiled-coil helices (see below) would result in alignment along the same surface of the DNA helix for dimerization. The nine heptads described for human AFP were aligned by computer analysis with the nine heptads proposed for the Group II receptor superfamily members, which include the VDR, RAR, T3R, erb-α2, V-erb and REV-erb receptors (see Figure 5I and list of abbreviations). Although attempts to align the entire third domain of either AFP with any individual receptor proved unsuccessful, notable matching of AAs between AFP and Group II receptor members was observed on an individual heptad-to-heptad comparison, when analyzed by computer AA matching and alignment (Hitachi HIBIO-PROSIS Protein Input and Analysis software). The most conserved AA matching occurred between Group-II nuclear receptors and heptads 1, 2 and 5, (58-61%), while heptads 3, 4, 6, 7, and 9 showed 48-54%, and heptad 8 showed 41%. Computer analysis demonstrated that the most conservative matching (58%) occurred between AFP and the retinoic acid receptors (α, β and γ), followed next by the erb series (51%) and the thyroid (α1, β) (50%) receptors and finally the vitamin D receptors (42%) (Figure 5F). It was next determined whether certain stretches within the nine-heptad AFP dimerization motif could be matched to members of the remaining steroid/thyroid superfamily consensus region in the ligand-binding domain was identified where both a dimerization and a DNA-binding/stabilizing site colocalized (Fawell et al., 1990). Subsequent alignment of this superfamily dimerization region with the nine heptads proposed for AFP revealed a conserved AA region which matched to AFP heptads 3 through 7 (Figure 5I). Thus, these matching data demonstrated AA sequence peptide with five of the nine AFP heptads to both the steroid and the thyroid members of the superfamily. The analysis again confirmed (see above) that the retinoic acid receptors displayed the highest AA peptide (60%) in the dimerization region among the superfamily members. Interestingly, the GR displayed the highest peptide among the steroid receptors. The entire superfamily displayed the following pattern: RAR>GR>PR>ER, ERR>nur, VDR, MR>T3R>AR>COUP>ear. Thus, human AFP could possess a motif in the ligand-binding domain which might engage in both dimerization activities and stabilization of receptor binding to the DNA helix. Transcription-associated proteins (TAPs), which interact with activated receptors, have been described as transcription co-factors or auxiliary proteins, many of which have yet to be characterized (Bourguet et al., 2000; Love et al., 2000); indeed, AFP might function as such a factor. A highly conserved 30-AA region, bearing a TAP heterodimerization motif in many members of the receptor superfamily, has reportedly been localized to the ligand-binding subdomain of nuclear receptors (Spanjaard et al., 1991). Indeed, such a TAP-like binding region can likewise be visualized in a carboxyterminal region in juxtaposition to the steroid-binding subdomain reported for AFP, namely HAFP (AA #464-496). Alignment of these respective AAs in AFP with a broad spectrum of nuclear receptors (NR) and transcription-associated factors revealed identify/similarity matching of AAs (Figure 5D). On HAFP AA#464-496, the nuclear receptor superfamily displayed the following gradation pattern of AA heptad matching: RAR>RXR>ER>T3R. A second and third TAP-binding region of the T3R, located in the second domain of AFP, consists of 15-24 AA juxtaposed to the ligand binding region (E receptor domain) which is conserved in all members of the c-erbA superfamily (O'Donnell et al., 1991). It is shown in Figure-5D that these potential transcription-associated segments are displayed in the second domain of human AFP at AA #242-256 and AA #358-381. Total AA identify/similarity between AFP and the steroid/thyroid receptors ranged form 40-80%. The steroid receptor peptide in this region could be attributed to the reported observation that this region of the ER has been implicated in the binding of the HSP-90 and HSP-70 to the ER (Pratt et al., 1988).Nuclear localization signals (NLS): Various NLSs are found in various members of the steroid/thyroid/vitamin receptor superfamily (Dang and Lee, 1989; Guiochon-Mantel et al., 1989; Picard et al., 1990); these were then compared to presumptive signal AA cassette sites proposed for the AFP molecule. It became immediately apparent that the AFP molecule itself displayed intrinsic prototypic signal-like sequence homologies among its own three domains (Figure 5I). There further appeared to be multiples of the different signal types on all three domains of AFP. The NLSs have been divided into three groups as follows:First, four compact type of signal-like cassettes present on all three domains of AFP appeared to resemble those of the thyroid/retinoic acid receptors more than the steroid receptor family members, although both share similarities (Picard et al., 1990).Second, two signal sites within the AFP steroid binding domain bore resemblance to comparable bipartite proto-signal sites reported for the steroid nuclear receptor members (LaCasse et al., 1993).Third, two NLS sites appeared to comprise part of a more degenerative signal sequence, one site of which lies on the second domain of AFP (Horowitz et al., 1989). The NLS-type-1 would correspond to the D-domain of most steroid receptors and might be exemplified on AFP as AA #339-354 in which a series of basic compact, concise AAs reside at AA #341-345. The NLS-type-2 region of the glucocorticoid receptor contains two subregions that require hormone-binding for activation, one located toward the amino-end of the ligand binding domain and the other at the carboxy-terminal last third of the domain. These two NLS regions on the glucocorticoid receptor might be analogous to the proposed protosignal AA #416-431 and AA #465-478 on AFP. In the glucocorticoid receptor, these regions of the ligand binding domain required hormone for activation; thus, it may be more than coincidence that AFP AA #416-431 and AA #465-478 lie at the beginning and end of the estrogen binding subdomain reported for rodent AFP (Nishi et al., 1991). The latter proto-signal which appeared to be present at AFP AA #465-478 lies directly in a region crucial to steroid (estrogen) binding regulation in the third domain of AFP (Nishi et al., 1991). These two regions on AFP could serve as potential proto-signals for organelle (nuclear) accumulation as reported for the glucocorticoid, estrogen and progesterone receptors (Picard and Yamamoto, 1987; LaCasse et al., 1993). A compact (concise) signal in the third domain of human AFP might be located within AA #537-562 which lie in a strongly hydrophobic region of leucine zipper-like heptad repeats resembling a stretch of 100 AAs comprising a large putative dimerization subdomain found in the nuclear receptor superfamily (Mizejewski, 1993). AFP is known to undergo a conformational change in the presence of high E2 concentrations, as evidenced by a change in the UV difference spectrum (Jacobson et al., 1990b). The estrogen receptor contains an estrogen-induced proto-NLS-type-2 in addition to three other basic AA-rich proto-signals which cooperate following exposure to estrogen (Ylikomi et al., 1992). Thus, AFP might mimic the estrogen receptor in displaying multiple weaker (degenerative) signal-like sites (AA #416-431, 465-478; Figure 5I). A fourth suspected NLS of the estrogen receptor, although more degenerative, exists toward the amino terminal side of the DNA binding domain. Such discontinuous motifs are known to exist in other superfamily members (such as retinoic acid, thyroid, retinoic-X and mineralocorticoid receptors) (Horowitz et al., 1989; LaCasse et al., 1993), and AFP may indeed have such a degenerative NLS-like site at AA sequences 223-237. Other degenerative (type-3) presumptive sites on the first domain of AFP at AA #80-94, 102-116 and 142-156 could also serve as presumptive signals for other organelle targeting. It may be no coincidence that one of the proposed signal cassette sites on AFP (AAs 532-545; Figure 5I) is associated with a presumptive hetero-dimerization region which was previously proposed for HAFP (Mizejewski, 1993). Since the nuclear targeting escort proteins usually recognize a targeting signal, it would seem reasonable to find (as in AFP) the dimerization sites located in the near vicinity, either flanking the signal cassette or inclusive with it. This observation would suggest a direct association between the presence of the signal cassette and the location of the dimerization interface. It would be conceivable that NLS sites on proteins also function as dimerization recognition and/or docking sites (Smith and Toft, 1993).
J) Amino Acid-Rich Repeat Sites The HAFP molecule contains AA-rich repeat sites which include clustered multiples of glycine, proline, histidine, and alanine in its AA sequence. An interesting AA repeat is found at AA #455-5600 found as AAT and AAT (Figure 5J). The alanine AAs are known to be anti-freeze motifs in fish proteins (Lin et al., 1999). One of the three glycine-rich regions is located on domain I at AA #77-85 while the remaining two on domain III at AA #473-482 and AA #492-498. Glycine is usually found as an AA chain component in beta sheet turns. A histidine-rich region was found on domain II at AA #263-271 suggestive of zinc and cobalt binding sites. Proline-rich regions on AFP were found at five regions namely, AA #131-144 (domain I); AA #172-177 (domain I); AA #323-328 (domain II); AA #492-498 (domain II); and AA #521-524 (domain III). Prolines are known to serve as helix-breakers and are also involved in Src-3(Ser/Thr kinase)-peptide sites (Nguyen et al., 1998). Various regions of basic AA (BAS sites) residues such as lysine, arginine, histidine, can be observed at AA #168-176 (domain I); AA #245-249 (domain II; AA #437-439 (domain III, and AA #283-488 (domain III). In comparison to the basic AAs, the acidic AAs, such as glutamic and aspartic acids (GASR sites), are seen at AA #96-107 (domain I); AA #149-156 (domain I); AA #388-400 (domain II); AA #565-573 (domain III; Figure-5J); and AA #587-595 (domain III). The acidic AA sequences stretches are often affiliated with proteins that bind to nuclear receptor/heat shock protein complexes such as HSP70 and HSP90 (Pratt et al., 1988). Two disease-related sites are located at AA #279 to 293 which are rich in glutamic acid/glutamate; these diseases states are Huntington's and Sjorgren's syndromes.
K) The Cytoskeletal Proteins, Intermediate Filaments, and Microtubule Peptide Sites Actin, an abundant intracellular protein, is known to circulate at micromolar concentrations in peripheral blood (Mc Leod et al., 1989). In mammals experiencing diverse forms of tissue injury, actin is released from dying cells and may be entrapped in fibrin clots adjacent to or directly at tissue damage sites. Actin has been reported to be a noncompetitive inhibitor of the clot-dissolving enzyme, plasmin (Lind and Smith, 1991). Adult plasma reportedly contains two high-affinity actin-binding proteins (i.e., vitamin D-binding protein [Gc globulin] and plasma gelsolin), which serve as actin-sequestering agents to protect against microvascular damage. Thymosin also serves as a major actin-sequestering protein in cells (Nachmias, 1993). Within the cell itself, actin is a dominant, abundant cytoskeletal protein existing in monomeric, filamentous, and protein-complexed forms. Profilin, also a major intracellular protein, binds monomeric G-actin to constitute an intracellular pool of nonfilamentous actin (McWhirter and Wang, 1993; Mizejewski, 1997). The profilin-actin complex is called profilactin and serves as an intracellular repository of available actin. Previous reports have demonstrated that actin can inhibit plasmin's hydrolysis of substrate, suggesting that accessible lysine residues of actin interact with the kringle fingers (lysine-binding regions) of the plasmin molecule (Lind and Smith, 1991). The authors of the latter study suggested that extracellular-released actin may modulate plasmin-dependent biological responses. Since AFP does possess a putative actin-binding site (Mizejewski, 1997), other physiological and pathological spin-off functions might be tenable. For example, an actin-binding function contributes to cell transformation in the Philadelphia-chromosome-positive human leukemias (McWhirter and Wang, 1993). The BCR-Abl oncoprotein complex, if prevented from binding F-actin, showed a reduced ability to transform rat fibroblasts, which mimic leukemia cell transformation. While BCR activates the Abl tyrosine kinase, the presence of both an actin-binding and a DNA-binding domain on the Abl proto-oncogene product suggests that the protein is capable of relaying a signal from the cytoskeleton to the genome in a single step. The actin-binding state of the oncogene transformation induces a redistribution of dispersed intracellular F-actin into punctate aggregates surrounding the cell nucleus (McWhirter and Wang, 1993). This process has been proposed to interfere with the normal signal transduction pathways by disrupting interactions between the actin cytoskeleton and growth factor/cell adhesion receptors at the plasma membrane. It would be logical to implicate AFP in such actin-coupled growth regulatory signal pathways in view of the association of AFP with growth, differentiation, regeneration, cell transformations, and cellular adhesion as discussed below. HAFP appears to further display short AA segment homologies (identities) to the cytoskeletal proteins that could aid in the further elucidation of AFP function based on structural analysis (Figure 5K). These proteins include the actin-associated filaments, the motor and microfilaments, and the intermediate filaments (Pollard, 1993). These stretches of HAFP AAs, extracted from GenBank, are displayed in Figure 5K. After partitioning of the cytoskeletal proteins into their three classes, certain trends and/or patterns emerged. First, the actin-associated proteins' identities appear to be clustered largely on domain II of HAFP. In contrast, the microfibril and motor filament (myosin) AA identities are amassed on HAFP domain I. Finally, the intermediate-filament protein homologies are more widely distributed, being found mainly on domains I and II. However, it is of interest that AA sequence identities of two anchor proteins namely, ankyrin and catenin, were both localized at the carboxy-terminal domain of HAFP. These computer sequence peptide findings would suggest a trend where certain areas on the HAFP molecule might conceivably provide a dimerization interface for interaction with actin, myosin, and the intermediate filaments. Such surface interactions with actin/myosin would implicate muscle contraction, cytokinesis, cytoplasmic streaming, ameboid motion, and cross-linking with the intermediate filaments. The synthesis of AFP has further been reported in neonatal rat skin implants (Mujoo et al., 1983), but has yet to be confirmed. The elevated AFP levels reported in infants with functional and dystrophic epidermolysis bullosa, aplasia cutis congenita, and epidermolysis bullosa letalis (Dolan et al., 1993; Drugan et al., 1995; Gerber et al., 1995) attest to a possible linkage between AFP and skin intermediate filaments. In fact, it has been suggested that AFP screening would obviate the need for fetal skin sampling in the prenatal diagnosis of these disorders (Gerber et al., 1995).
L) The Chemokine-like Peptide Sites Chemokines are chemical attractant cytokines that mediate the migration of cells, i.e. leukocytes, into and out of tissues and are essential to the immune response and inflammatory reactions (Moser et al., 2004). The chemokine gene family is composed of four member groups based on the amino-terminal positioning of their cysteine (C) AAs (Nakayama et al., 2003). The four member family of chemokine peptides (60-90 AA in length) are divided and subgrouped as follows: Group I has a single cysteine (C) residue in the amino terminus of the molecule; Group II has a CC grouping; Group III has the CXC configuration; and Group IV has the CX3C sequence where X is any AA residue (see top portion of Figure 5L). The chemokines are the ligands (binding agents) that bind to heptahelical receptors; these receptor molecules weave through the cell membrane seven times as transmembrane (TM) domains. Chemokines and their receptors are highly conserved in nature being present in mammals, birds, reptiles, amphibians and fish, and they share AA similarities across species. Chemokines are produced locally in tissues and direct the emigration and immigration of cells from the bloodstream into sites of inflammation, infection, and cell proliferation. Chemokines can direct and influence many physiological processes such as angiogenesis, degranulation, autoimmunity, HIV infection, tumor growth, parasite infection, and leukocyte trafficking (Robledo et al., 2001). Discovered less than 20 years ago, the chemokines are now known to function as regulatory molecules in cell maturation, trafficking, homing, and the development of various tissues (Robledo et al., 2001). The chemokines are now known to comprise a family of nearly 50 ligands and 20 receptors, which despite its size, is remarkably homogeneous with properties similar to IL-8, the first chemokine to be discovered. Many chemokines are secreted as a result of pathological conditions, some fulfill regulatory roles, and others function in cell migration during histo- and organogenesis and in cancer metastasis. Chemokines attracted much attention when it was discovered that some of their receptors function as binding sites for the HIV virus that causes AIDS (Orsini et al., 1999). Although they show receptor binding with HIV, their main function is attracting, homing, regulating, and luring cells into tissue parenchyma (Verani and Lusso, 2002). Full-length AFP, produced during pregnancies of all mammals, has long been associated with hematopoiesis in the fetal liver. Indeed, studies have shown fetal AFP is positively correlated with fetal hemoglobin while negatively correlated with red blood cell levels, hematocrit, and erythropoietin/transferring concentrations (Bartha et al., 1999). Thus, AFP has long been implicated in erythropoiesis myelopoiesis (bone marrow) and lymphopoiesis during pregnancy. These two latter developmental states are now known to be regulated by the chemokines. By means of GenBank AA sequence matching, it was discovered that HAFP resembles a segment of the chemokines, especially the Group-II (CC) members, GROα, MiP-1B, Eotaxin, Rantes, and MCP-1 chemokine (Figure 5L). It is proposed from AA matching evidence that portions of HAFP may possess the properties of chemokine fragments or appear as chemokine-mimicking proteins (peptides); see Figure 5L for the chemokine comparisons and AFP-matched identities. Note in Figure 5L that AFP cysteine bridge clusters resemble those of the CC-chemokine ligands. However, the chemokine-like homologies appear to be restricted to the third domain of AFP beginning at the amino-terminal portion of that domain (Figure 5L). Full-length HAFP has been reported to bind to the CCR5 receptor (Atemezem et al., 2002; Kellner and Mizejewski, 2004). In effect, this would cause suppression in growth factor-stimulated cells such as in prostate and breast cancers as well as HIV cell fusion and infection, suggesting that AFP might mimic the CCR5 ligand. Thus, it might be possible that AFP mimics a cc-chemokine and binds to a cc-chemokine receptor, induces the receptor conformational change and tubulin (or actin) polymerization; in other cases, it might inhibit cell migration (spreading), cell adhesion, and cell proliferation. In other words, AFP may be a partial agonist/antagonist chemokine mimic depending on its intravascular and interstitial concentrations and duration of exposure. Also, AFP would have an advantage over the other chemokine-related inhibitors, in that other chemokines cannot act in a decoy ligand fashion. It could be speculated that AFP could serve as a novel type of chemokine inhibitor, in that, it could still retain the growth regulatory properties, but still be able to induce some of the early (initial) functions of a true chemokine ligand (such as immunostimulation). Overall, AFP might be useful in pathological situations in which chemokines usually are involved. Such conditions could include cancer therapeutics, HIV infection and transmission, inflammatory diseases, angiogenesis, autoimmune diseases, atherosclerosis, multiple sclerosis, diabetes, lung infections (as in CF), malaria, endometriosis, graft rejection, wound healing, and other conditions yet to be discerned. Certainly, neurodegenerative dementias such as Alzheimer's disease might be viable candidate diseases.
VI. Concluding Remarks
At the start of this atlas chapter, it was stated that the many biological roles of AFP have yet to be clarified. In the face of such a challenge, the in vitro carrier/transport ligand-binding function was one of the first functions to be demonstrated. A growth-regulatory role for AFP, although not yet clearly understood, has also been established. The immunoregulatory properties of AFP are still at the threshold of new discoveries regarding the use of peptide epitopes in hepatoma cancer (vaccine) therapy. AFP is not merely a fetal substitute for albumin since ALB is present in nearly all vertebrate classes together with a-ALB, but true AFP is present largely in mammals, and a closely-related molecule has been identified in birds (Mangum et al., 2005). In contrast, true AFP is not present in shark, rays, fish, amphibians, salamanders, and reptiles (McLeod and Cooke, 1989). The 75 kD shark AFP reported by Gitlin (1973) turned out to be alpha-albumin, not AFP (Mizejewski, 1995a). Moreover, only two short stretches of mammalian AFP are nearly identical to various mammalian ALBs studied to date (Moro and Villacampa, 1986). One is located on HAFP domain II (16 AA long) and the other is on HAFP domain III (17 AA long). The domain I site resides in a region of platelet-surface proteins, while the domain III site is localized in an area with high peptide to glutathione peroxidase (58% over 31 AA) (see (Mizejewski, 1997)). Clearly, both regions appear to display functions common to albumin. The present atlas chapter was intended to focus on AFP as a biologic response modifier in order to expand the reader's perspective regarding the functional roles of AFP. By amassing a collection of publications in the biomedical literature, attempts were made to associate computer peptide-derived analyses of AFP domains, subdomains, and motifs/cassettes with previously published reports regarding the physiological roles of AFP. Although the computer findings did not always have a published counterpart, numerous insights were uncovered to provide both a rationale and justification for further studies of AFP physiology based on domain and subdomain peptide structure. More intense future investigations into the functional roles of AFP should prove mutually beneficial to both the biomedical community as well as the biotechnology industry. The separation, synthesis, and recombinant production of AFP subdomains and sequence fragments could provide a novel source of unique or modified pharmaceutical derivatives of AFP fragments for diagnosis and therapy in such diverse fields as endocrinology, hematology, immunology, and neurology. If only a small proportion of the AA sequence identities currently displayed on AFP have real biological relevance, this atlas report should provide a deep insight into the pathophysiology of AFP. For example, if the GenBank-detected sequences were present only in the pre- or pro-form of the protein precursor, then AFP might have the potential of binding and/or dimerizing with that intracellular precursor form of that molecule in question. In theory, this might suppress formation of the mature protein, thereby depleting its intracellular concentration. As presented in previous reviews by the author, AFP probably engages in different functions at various time points during ontogenetic development (Mizejewski, 1985; Mizejewski, 1997, 2001a). Conceptually, the binding of various ligands to AFP during the progressive stages of such development might dictate the specific function of AFP at that particular moment during development. AFP function could also be determined by various environmental factors in the fetal body's internal milieu such as ligand concentration, hyper- or hypoxia, pH state, blood sugar (glucose) concentrations, blood-gas partial pressure, osmolality, and plasma alcohol and lipoprotein/lipid content. Such conditions have been shown to produce conformational transition forms in AFP that alter tertiary form and shape (Mizejewski, 1995a). Many conformational changes are known to be reversible and gradual, after which AFP could revert to its original form (Mizejewski, 1995a). During these transitional changes, sequence sites concealed in molecular crevices on AFP might be exposed during the environmental stress period and then subsequently retract into their concealed state following protein refolding (see (Vakharia and Mizejewski, 2000)). In this manner, different AA sequence stretches normally unavailable might be exposed at various times dependent on the environmental shock or distress during that particular interval. AFP might contain multiple such buried sites in reserve in order to meet a variety of developmental challenges to the fetal environment. Hence, AFP could serve as a "protein for all seasons" and as the "molecular duct tape" of ontogenetic development.
Acknowledgments
The writing, artwork, and library/computer research utilized in this paper were supported by internal funds from the Wadsworth Center of the New York State Department of Health, Albany, NY. The author wishes to express his sincerest thanks to Ms. Tracy L. Godfrey for her commitment and time expenditure in the excellent typing and processing of the manuscript text, references, and tables of this review. The author also extends his deepest gratitude to Ms. Rachel A. Moseley for her provision/production of the excellent and skillful computer graphic artwork and illustrations presented in this report. Finally, the author is grateful to Michael Muehlemann at the Serometrix Biotechnology Company, Syracuse, NY for providing the three-dimensional computer model of the alpha-fetoprotein molecule.
Atlas of Genetics and Cytogenetics in Oncology and Haematology
Mapping of Structure-Function Peptide Sites on the Human Alpha-fetoprotein Amino Acid Sequence
Online version: http://atlasgeneticsoncology.org/deep-insight/20077/mapping-of-structure-function-peptide-sites-on-the-human-alpha-fetoprotein-amino-acid-sequence