CHAPTER 1: MELANOTRANSFERIN:

A 25–YEAR HALLMARK

“The journey of a thousand miles begins with one step.” Lao Tzu, 600 BC – 531 BC

This Chapter is adapted from:

1. Suryo Rahmanto, Y., Dunn, L.L., and Richardson, D.R. (2007) The melanoma tumor antigen, (p97): a 25-year hallmark – from iron to tumorigenesis. Oncogene. 26(42):6113-24. IF 2006: 6.5.

2. Dunn, L.L., Suryo Rahmanto, Y., and Richardson, D.R. (2007) Iron in the new millenium. Trends Cell Biol. 17(2):93-100. IF 2006: 12.4.

3. Suryo Rahmanto, Y., Sekyere, E.O., Dunn, L.L., and Richardson, D.R. (2007) The function of the membrane-bound homologue, melanotransferrin (melanoma tumour antigen p97). In: Iron Metabolism and Disease, Chapter 9 (Fuchs, H. ed.). Transworld Research Network. [Invited Book Chapter].

1

1.1. General introduction

Melanotransferrin (MTf), or melanoma-associated tumour antigen p97, is a membrane- bound glycoprotein that is a homologue of the transferrin (Tf) family of non-haem iron

(Fe)-binding [Brown et al. 1982]. Melanotransferrin was one of the first cell surface markers associated with melanoma [Brown et al. 1981b]. In fact, it was previously described as an oncofoetal antigen, being expressed in only small quantities in normal tissues, but in much larger amounts in neoplastic cells (especially malignant melanoma cells) and foetal tissues [Brown et al. 1981a; Woodbury et al. 1980].

More recently, there have been reports of MTf being identified in normal tissues, including sweat gland ducts [Alemany et al. 1993; Natali et al. 1987], avian eosinophils [McNagny et al. 1996], porcine foetal intestinal epithelial cells [Danielsen

& van Deurs 1995], liver endothelial cells [Alemany et al. 1993; Sciot et al. 1989], rabbit and mouse cartilage [Kawamoto et al. 1998; Nakamasu et al. 1999], avian proximal kidney tubules [McNagny et al. 1996] and the normal endothelium and reactive microglia of of Alzheimer's disease (AD) patients [Jefferies et al. 1996;

Kennard et al. 1996; Rothenberger et al. 1996]. Normal sera contains very low levels of soluble MTf (sMTf) protein [Brown et al. 1981a] and increased sMTf protein has been described in patients with AD [Kennard et al. 1996] and arthritis [Kato et al. 2001]. In addition, MTf has been shown to be expressed at mRNA levels in a wide variety of human tissues [Richardson 2000]. From a range of 50 normal human tissues, the highest

MTf mRNA levels were found in the salivary gland [Richardson 2000]. Later studies in mice also demonstrated that MTf mRNA levels were high in this tissue as well as the pancreas and epididymis [Sekyere et al. 2005]. Hence, the differential expression of

2

MTf suggests a specific function that may have physiological and pathological indications.

Melanotransferrin shares many important characteristics with serum Tf, including: (i) a of 37-39% with human serum Tf, human (Lf) and chicken Tf; (ii) the MTf is located on 3, as are those for Tf and the 1 (TfR1); (iii) the presence of many disulphide bonds in MTf that are also present in Tf and Lf; (iv) MTf has an N-terminal Fe-binding site that is identical to the N-lobe Fe-binding site found in serum Tf; and (v) isolated and purified MTf can bind Fe from Fe(III) citrate complexes [Baker et al. 1992; Brown et al. 1981b; Brown et al. 1982; Food et al. 2002; Richardson & Baker 1991a; Rose et al. 1986]. These observations suggested that MTf played a role in Fe transport.

The most obvious difference between MTf and serum Tf is that MTf is mainly bound to the cell membrane by a glycosyl phosphatidylinositol (GPI) anchor, while Tf is found free in the plasma [Alemany et al. 1993; Food et al. 1994; Kennard et al. 1995;

Nakamasu et al. 1999; Yamada et al. 1999]. The sMTf may also be secreted from cells, although at very low levels [Food et al. 1994]. Recent investigations have shown that deletion of the GPI-anchor via genetic engineering resulted in a recombinant sMTf that was far less effective than Tf at donating Fe to cells [Food et al. 2002]. In addition, sMTf cannot be bound by the TfR1 nor (TfR2), despite its high structural homology to Tf [Food et al. 2002; Kawabata et al. 2004].

3

Due to its high homology with Tf, its ability to bind Fe and its high expression in melanoma cells, MTf was hypothesised to play a role in Fe uptake by these tumour cells

[Brown et al. 1982; Richardson & Baker 1990; Richardson 2000; Rose et al. 1986].

Despite the very high levels of expression, MTf has been demonstrated not to play a marked role in Fe uptake by melanoma cells, when either bound to the cell membrane

[Richardson 2000] or as a soluble molecule [Food et al. 2002]. Indeed, while MTf binds

Fe, it does not rapidly internalise it into the cell [Richardson & Baker 1991a;

Richardson & Baker 1990; Richardson & Baker 1991b; Richardson 2000]. Nonetheless, the ability of MTf to bind Fe remains the only well characterised function of the protein.

Hence, this was an important idea to investigate, since Fe is critical for DNA synthesis and cancer cell proliferation [Kalinowski & Richardson 2005; Kwok & Richardson

2002].

Other reports have suggested that MTf is found in the and serum of Alzheimer’s patients [Jefferies et al. 1996; Kennard et al. 1996; Rothenberger et al. 1996; Yamada et al. 1999] and in the serum of patients suffering arthritis [Kato et al. 2001]. Recent investigations have also suggested that MTf could play a role in endothelial migration and angiogenesis [Sala et al. 2002], plasminogen activation [Demeule et al. 2003;

Michaud-Levesque et al. 2007], differentiation [McNagny et al. 1996; Suardita et al.

2002] and MTf may transcytose across the blood-brain barrier [Demeule et al. 2002].

However, its exact biological function remains unclear and difficult to determine from in vitro studies alone. This Chapter will discuss the molecular character and expression of MTf and its potential biological roles, particularly in relation to melanoma tumourigenesis.

4

1.2. Genomic and molecular organisation of melanotransferrin

The human MTf gene and its equivalent have been cloned and characterised in several organisms (Table 1.1). To date, this gene has been identified in ten organisms, namely human, chimpanzee, rabbit, rhesus monkey, cow, mouse, rat, dog, chicken and zebrafish, where the molecules shares 46-88% protein homology to human MTf. Most of the for MTf family members are predicted to encode molecules of 721-1540 amino acids in length, while the chimp gene is predicted to encode a much larger protein of 2031 amino acids (Table 1.1). However, BLAT analysis (database version

March 2006) demonstrated the presence of extra sequence within chimp MTf that corresponded to the human discs large homologue 1 (Drosophila, Dlg1) gene. Hence, the large size of chimp MTf as reported in the NCBI database (XM_526438; accessed:

July, 2006) may not be representative of the actual gene. Only human MTf protein has been purified and characterised in detail [Brown et al. 1981b; Brown et al. 1982] and the software predictions of the size of other MTf family members (Table 1.1) must be confirmed experimentally. A MTf-like gene has been identified in the fruit fly

(Drosophila melanogaster) and also in the mosquito (Anopheles gambiae) [Adams et al.

2000]. However, unlike other members of the MTf family, these insect molecules show far lower homology to other MTf family members and the single Fe-binding site is not conserved [Sekyere & Richardson 2000]. Thus, the role of these MTf-like proteins in insect Fe metabolism is unclear.

5

Table 1.1. Comparison of known MTf gene and protein sequences between species.

% Chromosome Genbank mRNA Protein % Protein Name Nucleotide ^ Number Accession No. (bp) ^ (aa) Similarity Similarity MTf Tf TfR

NM_005929 2377 100 738 100 Human 3 3 3 #NM_033316 1651 78 302 78

Chimp §XM_526438 6096 64 2031 84 3 3 3

Rabbit AB010995 2388 86 736 86 ? ? ?

Rhesus §XM_001096034 4623 89 1540 86 2 2 2

§ Cow XM_589607 2346 88 739 88 1 1 ?

Mouse NM_013900 4158 82 738 83 16 916

Rat §XM_237839 2462 82 738 84 11 8 11

§ Dog XM_545158 3582 79 1193 78 33 23 33

Chicken NM_205207 3547 63 738 60 9 9 9 § Zebrafish XM_689207 3095 54 721 46 6 ? 2

^ Coding region (CDS) used for nucleotide homology and score taken based upon ClustalW software for both nucleotide and protein homology (http://clustalw.genome.jp/). The % similarity is relative to human MTf. § Predicted sequences by automated computational analysis ? Data has not been completed nor published. # Denotes the short MTf transcript i.e., hΔMTf

6

The MTf amino acid sequence data from all ten animal species known have been aligned and a phylogenetic tree constructed (Figure 1.1) by the neighbour-joining method (http://align.genome.jp/; [Nakamasu et al. 1999]). Analysis of this phylogenetic tree shows the relationship of the ten MTf members compared with human Tf, Lf and chicken ovoTf. Melanotransferrin diverged from the other Tf’s at position “X”, while fish and avian MTf diverged from mammals at position “Y” (Figure 1.1). Additionally, it has been previously shown that MTf diverged from Tf after the divergence of vertebrates and invertebrates [Nakamasu et al. 1999]. Hence, there is evidence that in evolutionary terms, MTf is more distantly related to serum Tf than Lf [Lambert et al.

2005a; Lambert et al. 2005b]. Considering this, it is possible that the molecular architecture of the Tf molecule could be a useful scaffold for “building” other molecules with different functions and this is explored further in Section 1.5.

7

Figure 1.1. The phylogenetic tree of MTfs and the other Tf family members. Whole and partial protein sequences of Tf super-family members available in the databases were aligned to construct a phylogenetic tree using global alignment by the neighbour-joining method (http://clustalw.genome.jp/). Adapted from [Suryo Rahmanto et al. 2007a; Suryo Rahmanto et al. 2007c].

8

1.2.1. Chromosomal localisation of melanotransferrin

The human MTf gene (also denoted as MFI2) is localised to region q28-29 on the reverse strand of and resides in the same region as the Tf and TfR1 genes

[Plowman et al. 1983; Seligman et al. 1986; Yang et al. 1984]. The MTf, Tf and TfR1 genes co-localise to the same chromosome in a number of different organisms. In fact, in human, chimpanzee, rhesus monkey, mouse, rat, dog and chicken, TfR1 co-localises with MTf on chromosome 3, 3, 2, 16, 11, 33 and 9, respectively (Table 1.1).

Interestingly, Tf which has homology to MTf co-localises to the same chromosome in only humans, chimpanzees, rhesus monkey, cow and chicken (Table 1.1). However, the significance of this observation has not been elucidated.

1.2.2. Melanotransferrin gene structure

The gene structure of melanotransferrin is best characterised in human and mouse and this is described in the Sections below. Clearly, the structure of the mouse gene is important to discuss considering the development of MTf knockout and transgenic models to investigate the function of the molecule in vivo.

1.2.2.1. Human melanotransferrin gene

The human MTf gene (hMTf) spans over 26 kb and is comprised of 16 exons (Figure

1.2; NCBI GeneID: NM_005929; [Le Beau et al. 1986; Plowman et al. 1983]). The hMTf promoter region (601 bp, http://www.genomatix.de/) is mapped from +32 to -569, taking the translation initiation codon (AUG) as +1. This promoter contains three Sp1 transcription factor binding sites and a high GC content (71%). However, the minimal region required for promoter activity is defined from positions -204 to -1, which has two

9

potential binding sites for Sp1 [Duchange et al. 1992]. The Sp1 transcription factor is essential for regulating genes with GC boxes, stimulating or repressing transcription at promoters and enhancers and has a role in development and differentiation [Courey et al. 1989; Saffer et al. 1991]. Furthermore, an enhancer element has also been detected 2 kb upstream from the hMTf promoter region and its deletion results in impairment of hMTf in melanoma cell lines [Duchange et al. 1992]. In addition to

Sp1, other studies also indicate the presence of three potential AP1 binding sites in the enhancer element of hMTf from -2312 to -2268 bp [Roze-Heusse et al. 1996].

Figure 1.2. BLAT analysis arising from the human MTf (hMTf) and mouse MTf (mMTf) genes. The long hMTf and mMTf transcripts consist of 16 exons, while human short MTf

(hΔMTf) consists of 7 exons. The hΔMTf transcript is composed of exons 1-6 and a final

3' exon consisting of part of intron-6 from hMTf which was denoted previously as exon-

6b [Sekyere et al. 2002]. Adapted from [Suryo Rahmanto et al. 2007a; Suryo Rahmanto et al. 2007c].

10

1.2.2.2. Mouse melanotransferrin gene

Analysis of the mouse MTf gene (mMTf, MFi2), demonstrated that it is similar to hMTf, spanning approximately 26 kb and consisting of 16 exons (Figure 1.2; GeneID:

NM_013900; [Nakamasu et al. 2001]). The mMTf gene resides on chromosome 16 and shares a homologous region with human chromosome 3q [DeBry & Seldin 1996; Le

Beau et al. 1986; Nakamasu et al. 2001]. The mMTf promoter region (666 bp, http://www.genomatix.de/) is mapped from +49 to -617, taking the translation initiation codon (AUG) as +1. As seen in hMTf, the mouse promoter also contains three Sp1 transcription factor binding sites with a slightly lower GC content of 53%.

1.2.3. Melanotransferrin gene transcription

The hMTf gene encodes two transcripts:- (i) a 16 exon long isoform, hMTf (2377 bp,

Genbank Accession: NM_005929) and (ii) a shorter seven exon isoform, hΔMTf (1651 bp, Genbank Accession: NM_033316; [Sekyere et al. 2003]). The most recent Genbank information describes the full-length hMTf transcript as 2377 bp (coding region 70-

2283). However, the initial studies by Rose and colleagues identified a polyadenylated transcript of over 4 kb, the additional sequence being 3’-untranslated region (UTR;

[Rose et al. 1986]). In the hΔMTf transcript, the first 6 exons are identical to those of hMTf [Sekyere et al. 2005]. However, the seventh and last exon, termed 6b, is composed of intron 6 sequence (Figure 1.2). The sequence of exon 6b has an in-frame termination signal resulting in the truncated hΔMTf transcript [Sekyere et al. 2002].

Recent studies have aimed to characterise hMTf transcripts and their expression patterns in normal human tissues and a variety of tumour cell lines [Sekyere et al. 2005].

11

Northern blot analysis of 12 normal tissues demonstrated two transcripts across all human tissues assessed, at approximately 3 kb and 4 kb [Sekyere et al. 2005].

Additionally, a third transcript was detected in the heart and skeletal muscle at approximately 2 kb [Sekyere et al. 2005]. This smaller mRNA may equate to the hΔMTf transcript and was not detected in the other tissues examined [Sekyere et al. 2005].

Alternative splicing of the pre-mRNA accounts for the two known hMTf transcripts that were observed at 4 kb [Rose et al. 1986] and 2 kb [Sekyere et al. 2005], while the presence of the 3 kb transcript could only be explained by the use of alternative polyadenylation sites [Sekyere et al. 2005]. Further, in nine tumour cell lines assessed, the hΔMTf transcript was detected by RT-PCR at high levels in the melanoma cell lines,

SK-Mel-2 and SK-Mel-28, but was absent or only very slightly expressed in the other cell types [Sekyere et al. 2005].

The physiological significance of the MTf splice variants in both normal tissues and cell lines remains unknown and it is unclear whether the hΔMTf transcript leads to a functional gene product. However, examining the resultant protein sequence, the truncated product arising from the hΔMTf transcript has lost one Fe-binding residue

(His-279) and has no GPI-anchor motif and thus, may not be membrane-bound [Sekyere et al. 2002]. In fact, Kyte-Doolittle analysis indicated that if a protein resulted from translation of hΔMTf, it would have increased hydrophilicity compared with the full length protein encoded by hMTf and may be a soluble molecule [Sekyere et al. 2005].

However, the sMTf identified from spent culture supernatants of melanoma cells is much larger 80-97 kDa [Food et al. 1994] than that predicted for the protein product of hΔMTf (302 amino acids; approximately 30 kDa; Table 1.1).

12

1.3. Protein structure of melanotransferrin

Like other members of the Tf super-family [Morgan 1981; Richardson & Ponka 1997], the hMTf protein is composed of two major lobes [Baker & Lindley 1992; Rose et al.

1986]. The molecule consists of 738 amino acids, with a predicted molecular weight of

80.2 kDa. The protein can divided into four key regions:- (i) a 19 residue hydrophobic signal peptide; (ii) an extracellular N-terminal consisting of 342 amino acids that contains the functional Fe-binding site; (iii) an extracellular C-terminal of 352 amino acids; and (iv) a 25 residue hydrophobic GPI-anchor domain present on the membrane- anchored form of the protein [Rose et al. 1986]. The N- and C-terminals share high internal homology that includes 14 cysteine residues, most of which are also conserved across the Tf family of proteins [Bailey et al. 1988; Baker et al. 1992; Brown et al.

1982; Rose et al. 1986].

In addition to the membrane-bound form of hMTf, sMTf is probably a secreted form of the protein that apparently is not passively shed from the cell [Food et al. 1994]. The sMTf has been detected in spent culture medium from melanoma cells, as well as in urine, saliva, serum and cerebrospinal fluid [Desrosiers et al. 2003; Food et al. 1994;

Kim et al. 2001; Yamada et al. 1999]. At present, it is unclear whether the soluble form seen in cell culture supernatants is the same molecular species that is seen in vivo e.g., in serum [Kennard et al. 1996].

The first observation of hMTf using reducing SDS/PAGE analysis showed the protein migrated at approximately 97 kDa [Woodbury et al. 1980]. In later studies using non-

13

reducing conditions, Desrosiers and colleagues demonstrated two MTf isoforms that migrate at 79 kDa and 82 kDa and are recognised by the specific anti-MTf monoclonal L235 [Desrosiers et al. 2003]. These MTf isoforms are likely the result of post- translational modifications such as differential glycosylation [Desrosiers et al. 2003]. In fact, glycosylation sites have been identified at residues 38 and 135 of the N-terminal and residue 515 in the C-terminal [Brown et al. 1981b; Rose et al. 1986].

A major difference between MTf and other members of the Tf family is the ability of

MTf to bind only one ferric atom in its N-terminal region [Baker et al. 1992]. As found in Tf, the critical Fe-binding ligands in the N-terminal of MTf consists of the DYRYH motif (Table 1.2), that is composed of Asp78, Tyr107, Arg136, Tyr210 and His279

[Baker et al. 1992]. In contrast to Tf, the C-terminal of MTf consists of SYSYH, in which Asp-421 and Arg-482 are replaced by serine resulting in a general disruption of

H-bonding and loss of the Fe-binding motif [Bailey et al. 1988; Baker & Lindley 1992;

Nakamasu et al. 1999; Sekyere & Richardson 2000]. Analysis of the protein sequence encoded by the hΔMTf transcript suggests that the potential protein product does not bind Fe, as the His279 residue has been replaced by valine [Sekyere et al. 2005].

14

Table 1.2. Comparison of the Fe- and anion-coordinating amino acids present in MTfs and other Tf homologues.

Protein Source N-terminal repeat C-terminal repeat (position of each amino acid) (position of each amino acid)

Melanotransferrin Human DYRYH (78,107,136,210,279)a SYSYH (421,452,482,556,625) Chimp DYRYH (149,178,207,281,641) SYSYH (1116,1137,1177,1251,1367) Mouse DYRYH (78,107,136,210,279) RYSYH (421,451,482,556,625) Rat DYRYH (78,107,136,210,279) RYSYH (421,442,482,556,625) Rabbit DYRYH (78,107,136,210,279) NYSDH (421,451,482,556,625) Rhesus EYRYH (228,260,289,363,432) SYSNH (574,595,635,709,778) Dog DYRYH (824,959,988,1080,1184) RYSSH (1474,1495,1718,1835,1954) Chicken DYRYR (77,106,135,209,278) GYRYQ (420,450,481,555,624) Cow DYRYH (78,107,136,210,279) KYSYH (421,442,482,556,625) Zebrafish SYRYR (78,107,137,209,269) DYRYH (405,426,472,545,614)

Transferrin Humanb DYRYH (82,114,143,207,268) DYRYH (411,431,475,536,604) Cockroach DYRYH (78,111,141,225,295) DYRYH (429,457,487,573,642)

Lactoferrin Humanc DYRYH (80,112,140,212,273) DYRYH (415,455,485,548,617)

Ovotransferrin Chicken DYRYH (79,111,140,210,269) DYRYH (414,434,479,543,611)

GenBank accession numbers (protein ID): human Tf – NP_001054; cockroach Tf – A47275; human Lf – NP_002334; chicken ovoTf – NP_990635; other proteins (see Table 1.1).

a In parentheses are positions of each Fe-coordinating amino acid in the protein sequence. b Invariant in human, rabbit, bovine, porcine, equine and chicken serum Tf. c Invariant in human, bovine and murine Lf.

15

The Fe-binding site of MTf, like other members of the Tf family [Morgan 1981], has been suggested to be capable of binding other metal cations [Richardson 2000]. In addition to this, a putative (II)-binding motif (HExxH) is present in the hMTf protein (344LGHEYLHAMK353) and is also predicted in chimpanzee MTf

(1386LGHEYLHAMK1395, GeneID: XM_526438). It has been previously postulated that hMTf may have a Zn(II)-binding function and possible thermolysin/metalloprotease activity (Figure 1.3; [Garratt & Jhoti 1992; Nappi & Vass 2000]). If metalloprotease activity is an important function of MTf, one may expect the motif to be conserved across species. However, this motif is absent in the mouse, chicken and rabbit, suggesting that at least in these organisms, MTf probably has no significant metalloprotease activity [Sekyere & Richardson 2000].

In the mouse, the mMTf protein is a similar length to that found in the human viz., 738 amino acids long. It shares an amino acid identity of 83% with hMTf and a homology of

41% with human serum Tf [Rose et al. 1986; Yang et al. 1984]. Again, the cysteine residues are conserved, as is the Fe-binding motif in the N-terminal [Baker & Lindley

1992]. The same substitutions occur in the C-terminal of mMTf as those in hMTf

[Baker et al. 1992; Nakamasu et al. 1999]. Mouse MTf also possesses a GPI-anchor sequence and is a membrane-bound protein [Nakamasu et al. 1999]. The presence of a soluble form of the mMTf protein in the serum or other fluids of the mouse remains to be determined.

16

Figure 1.3. Amino acid sequence and other motifs in human MTf. The two MTf terminals (N- and C-) are separated by the black line; the Fe-binding residues and substitutions are boxed in red; the Zn-binding motif in green. The signal peptide is boxed in grey and the GPI-anchor signal in blue. Adapted from [Suryo Rahmanto et al. 2007a].

17

1.4. Expression and localisation of melanotransferrin

Melanotransferrin was originally found to be expressed at high levels in melanomas, other tumours and foetal tissues, but at low levels in normal adult tissues [Brown et al.

1981a; Woodbury et al. 1980]. To date, the highest levels of MTf are detected in melanomas, in particular the melanoma cell line SK-Mel-28, that has up to 3.8 x 105

MTf binding sites per cell [Brown et al. 1981b]. For this reason, most in vitro studies examining MTf function have utilised this cell line. Until recently, there were only intermittent reports of MTf expression in normal adult tissues such as sweat gland ducts, liver sinusoids, endothelial cells and brain endothelium [Natali et al. 1987;

Rothenberger et al. 1996; Sciot et al. 1989]. The claim that MTf expression may be elevated in non-malignant conditions such as AD [Jefferies et al. 1996] underlined the necessity to characterise the expression of MTf in normal human tissues.

Analysis of hMTf mRNA expression has been determined across a broad range of normal human tissues and may be more sensitive than the earlier employed immunological assays [Richardson 2000]. In these studies, hMTf mRNA expression was detected across all 42 adult and foetal tissues assayed, with the highest levels in the salivary gland, kidney and pancreas (Figure 1.4; [Richardson 2000]). The ubiquitous but differential expression pattern of hMTf was entirely different to that of both Tf and the

TfR1, suggesting that hMTf may have roles outside of classical Fe metabolism.

Immunohistochemical staining of normal tissues using the monoclonal antibody, L235, that is specific for MTf, confirmed that hMTf expression was highest in the epithelial structures, ducts and tubules of the salivary gland, pancreas, kidney, epididymis and sweat gland ducts of the skin (Figure 1.5; [Sekyere et al. 2005]). Other data from our

18

laboratory indicates a somewhat similar MTf tissue expression profile in the mouse

[Sekyere et al. 2005]. The protein has also been detected in tissues from other species such as porcine intestinal epithelial cells, chicken eosinophils and chondrocytes from the rabbit and mouse [Alemany et al. 1993; Danielsen & van Deurs 1995; Kawamoto et al. 1998; McNagny et al. 1996].

19

Figure 1.4. Mouse mRNA (poly A+) Master Blot™ sequentially hybridised to mouse melanotransferrin (mMTf) cDNA, mouse (TfR1) cDNA and hypoxanthine guanine phosphoribosyl transferase (HGPT) cDNA (loading control). Probes were labelled and hybridised to the blot using established procedures [Sekyere et al. 2005]. The autoradiogram was developed after exposure of the membrane for 2 days (mMTf and TfR1) or 2 weeks (HGPT). Adapted from [Sekyere et al. 2005].

20

Figure 1.5. Immunohistochemistry using anti-hMTf MoAb L235 compared with a relevant control without the antibody demonstrating the distribution of hMTf in: (A) skin, (B) liver, (C) kidney and (D) salivary gland. (A) The epidermis of the skin shows positive staining in the epidermis (see star) while the dermis is largely negative. Sweat gland ducts also stain positive (see arrow). Magnification: 100x. (B) Hepatocytes of the liver stain positive (see star), while connective tissue of interlobular septa is negative (see arrow). The sinusoids are largely negative (see triangle). Magnification: 200x. (C) Epithelium lining the tubules of the kidney stain positive (see star) while the glomeruli (see arrow) and connective tissue are negative (see triangle). Magnification: 100x. (D) Epithelium lining the ducts of the salivary gland stain positive (see arrow), while glandular serous alveoli (see star) and connective tissue (see triangle) are negative. Magnification: 200x. Adapted from [Sekyere et al. 2005].

21

1.4.1. GPI-anchored melanotransferrin

Melanotransferrin was initially thought to be bound to the plasma membrane through its hydrophobic protein tail [Brown et al. 1982]. Evidence for membrane anchoring through a GPI moiety was published over 10 years later [Alemany et al. 1993; Food et al. 1994]. Melanotransferrin was demonstrated to attach to the cell surface via GPI- anchor through studies examining: (i) the sensitivity of the protein to bacterial phosphatidylinositol-specific phospholipase C (PI-PLC), which is a common method to cleave GPI-anchored proteins from membranes; (ii) biosynthetic labelling with

[3H]ethanolamine; and (iii) partitioning of MTf in Triton X-114 [Alemany et al. 1993;

Food et al. 1994; Low 1989].

It is of interest to note that a variety of cell surface proteins are attached to the plasma membrane via a GPI-anchor [Lisanti et al. 1990]. These proteins lack a transmembrane domain, have no cytoplasmic tail and are located exclusively on the extracellular side of the plasma membrane [Lisanti et al. 1990]. The GPI-anchor is a glycolipid that is used to attach proteins to the membrane bilayer via phosphoethanolamine, glycans and phosphatidylinositol lipid [Low 1989]. The GPI-anchored proteins are synthesised and segregated into the endoplasmic reticulum (ER) with an extra hydrophobic GPI attachment signal at their C-terminus [Brown & Waneck 1992; Doering et al. 1990].

This signal is recognised and cleaved by an enzyme that attaches the preformed GPI by a transamidation reaction [Brown & Waneck 1992; Doering et al. 1990].

The GPI-anchor may be important in MTf function, as GPI-anchors can regulate cell surface localisation, transmembrane signalling, surface molecule turnover and

22

biological function of membrane-bound proteins. The family of GPI-anchored proteins form a diverse array of molecules that includes membrane-associated enzymes, adhesion molecules, activation antigens, differentiation markers, protozoan coat components and other miscellaneous glycoproteins [Brown & Waneck 1992]. The GPI- anchor imparts some functional characteristics to proteins, such as: (i) a strong apical targeting signal in polarised epithelial cells [Lisanti et al. 1989; Lisanti et al. 1990]; (ii) the ability to concentrate into specialised lipid domains in the membrane, including the so-called “smooth pinocytotic vesicles” (caveolae; [Sargiacomo et al. 1993]); (iii) GPI- anchored proteins may act as activation antigens in the immune system and may modulate antigen presentation by major histocompatibility complex molecules [Brown

& Waneck 1992; Brown 1993; Lisanti et al. 1989; Loertscher & Lavery 2002]; and (iv) cleavage of GPI-anchor by PI-PLC or PI-phospholipase D may generate second messengers for signal transduction [Low & Huang 1991; Sharom & Lehto 2002]. To date, the biological roles of MTf in relation to the presence of its GPI-anchor remain unresolved. Although, it is of interest that MTf is expressed on many epithelial surfaces

(e.g., gland ducts; [Alemany et al. 1993; Sekyere et al. 2005]) where apical targeting may be important in its function.

1.4.2. Soluble melanotransferrin

As discussed previously, sMTf has been identified in cell culture supernatants [Food et al. 1994; Liao 1996] and in the serum of patients with arthritis [Kato et al. 2001] and

AD [Kennard et al. 1996; Kim et al. 2001]. A recent study examining the role of genetically engineered sMTf in Fe uptake by cells showed that despite its homology to

Tf, sMTf does not bind to the TfR1 nor any other high affinity receptor [Food et al.

23

2002]. In fact, compared with Tf, sMTf was inefficient at donating Fe to cells and was non-specifically internalised by melanoma cells [Food et al. 2002] and reticulocytes

[Richardson & Morgan 2004]. Moreover, sMTf was degraded by the [Food et al. 2002]. This is in contrast to Tf, which is internalised and largely re-cycled in many cell types [Morgan 1981], including melanoma cells [Richardson & Baker 1991b].

It has been previously shown that sMTf protein does not contain a GPI-anchor signal and is not likely to originate from shedding of the membrane-bound form [Alemany et al. 1993; Food et al. 1994]. However, it is still unclear how sMTf may arise and its precise function and physiological significance is unknown. Considering the genesis of sMTf, defects in either the biosynthesis of the GPI-anchor, or the addition of the anchor to the polypeptide chain are known to occur [Miyata et al. 1994; Ware et al. 1998] and could potentially result in release of the soluble form of the molecule. Secondly, the presence of GPI-specific phospholipase D (GPI-PLD) in mammalian plasma which recognises a conserved portion of the anchor [Metz et al. 1991] could, at least in part, account for sMTf.

1.5. Functional role of melanotransferrin: iron transport and metabolism

One of the well-characterised properties of hMTf is the sequence similarity with members of the Tf super-family and its ability to bind one Fe-atom [Brown et al. 1982].

A general hypothesis initially existed that MTf may be over-expressed in melanoma cells to assist with their increased Fe requirements for proliferation [Brown et al. 1982;

Richardson & Baker 1990; Sekyere & Richardson 2000].

24

A variety of in vitro studies have been performed to ascertain the relationship between

MTf and Fe metabolism in melanoma cells [Brown et al. 1982; Richardson & Baker

1991a; Richardson & Baker 1990; Richardson & Baker 1991b]. Iron uptake was examined in the SK-Mel-28 cell line that express the highest levels of naturally produced MTf known to date (3.8 x 105 binding sites per cell; [Brown et al. 1982]). Iron uptake in this cell line was found to occur by two processes consistent with:- (i) receptor-mediated of the Tf-TfR1 complex and (ii) adsorptive pinocytosis of Tf [Richardson & Baker 1991a; Richardson & Baker 1992a; Richardson & Baker

1990; Richardson & Baker 1991b; Richardson & Baker 1994]. A pronase-sensitive, temperature-dependent, membrane-bound Fe-binding component that was consistent with MTf was also identified, although it could not efficiently donate Fe to the cell from

Tf or low molecular weight Fe complexes [Richardson & Baker 1990; Richardson &

Baker 1991b; Richardson & Baker 1992b; Richardson & Baker 1994]. Indeed, removal of MTf from the cell with PI-PLC had no significant effect on Fe uptake from 59Fe- citrate [Richardson 2000]. Collectively, these studies demonstrated that MTf did not play a significant role in Fe uptake by melanoma cells.

It is of interest to note that Kennard et al., showed that MTf transfected Chinese hamster ovary (CHO) cells could internalise Fe from Fe(III) citrate complexes [Kennard et al.

1995]. These CHO cells expressed 1.2 x 106 MTf molecules per cell, a level three- to four-fold higher than that of the most highly expressing, naturally-derived melanoma cell line, SK-Mel-28 (300-380,000 sites/cell; [Brown et al. 1981b; Sekyere &

Richardson 2000]). However, after an incubation of 240 min, the Fe uptake from 59Fe- citrate was only 2.4-times greater than control CHO cells not expressing MTf [Kennard

25

et al. 1995]. Furthermore, hyper-expression of MTf did not result in any increase in Fe uptake from Tf when compared with control cells [Kennard et al. 1995]. These studies again indicated that the role of MTf in Fe uptake was not marked.

Additional investigations were conducted to examine whether MTf expression in SK-

Mel-28 melanoma cells was regulated in a similar way to molecules involved in Fe uptake e.g., TfR1 [Richardson 2000; Seligman et al. 1986]. Irrespective of intracellular

Fe or cellular proliferation status, MTf mRNA or protein expression was not regulated like TfR1, which is increased by Fe-deficiency or during proliferation [Richardson &

Ponka 1997; Richardson 2000; Seligman et al. 1986]. In addition, the expression of MTf mRNA amongst tissues was shown to be very different to TfR1 mRNA, with the molecule not being expressed at marked levels in tissues with high Fe requirements

(e.g., placenta, liver or bone marrow; [Richardson 2000]). Other studies also suggest that MTf does not play a key role in Fe metabolism. For instance, unlike TfR1 which is up-regulated to supply Fe for DNA synthesis in the S phase [Neckers & Cossman

1983], MTf expression remained constant throughout the cell cycle [Kameyama et al.

1986]. All these data demonstrate that MTf expression was not regulated in a way that would be expected if it was playing a vital role in Fe uptake.

Although in vitro studies have clearly demonstrated that MTf has only a very minor role in Fe uptake [Richardson 2000], it is not possible to completely rule out a role in vivo.

The future development and characterisation of MTf knockout and transgenic mouse models would address this question.

26

1.6. Other possible roles of melanotransferrin

Increased MTf expression in neoplastic cells, in particular melanoma cells, has been well documented [Brown et al. 1980; Brown et al. 1981a; Garrigues et al. 1982;

Woodbury et al. 1980]. Recent data suggests MTf is involved in: (i) angiogenesis, cell migration and plasminogen activation [Demeule et al. 2003; Michaud-Levesque et al.

2005b; Sala et al. 2002]; (ii) the pathogenesis of AD and Fe transport across the blood brain barrier (BBB; [Demeule et al. 2002; Jefferies et al. 1996; Moroo et al. 2003;

Rothenberger et al. 1996]); (iii) arthritis and chondrogenesis [Kato et al. 2001;

Kawamoto et al. 1998; Nakamasu et al. 1999; Suardita et al. 2002]; and (iv) eosinophil differentiation [McNagny et al. 1996]. The role of MTf in each of the described physiological and pathophysiological events is discussed in the sub-Sections below.

1.6.1. Cell migration and proliferation, plasminogen activation and angiogenesis

Melanoma tumours are highly vascularised and there is a correlation between angiogenesis and metastatic potential [Denijn & Ruiter 1993; Neitzel et al. 1999; Ribatti et al. 1992]. Investigations on the effect of recombinant sMTf on angiogenesis in chick chorioallantoic membrane and chemotactic cell migration using the Boyden chamber assay suggested that recombinant sMTf in vitro exerts: (i) an angiogenic response quantitatively similar to that elicited by fibroblast growth factor-2 (FGF-2) and (ii) induces chemotactic migration of vascular cells [Sala et al. 2002]. These data also suggested that MTf and vascular endothelial growth factor-1 (VEGF1) may play a role in various stages of melanoma progression. In fact, VEGF1 is a known stimulator of migration and angiogenesis in melanoma cells [Birck et al. 1999; Graeven et al. 2001].

27

Further, it was concluded by Sala and co-workers that recombinant sMTf may participate in the vascularisation of solid tumours [Sala et al. 2002].

A role for sMTf in angiogenesis and cell migration has been suggested to occur through its interaction with the low-density lipoprotein receptor-related protein (LRP) and the urokinase:plasminogen activator: complex [Demeule et al. 2003].

During the past decade, there has been increasing evidence for the involvement of the urokinase-type plasminogen activator (uPA) system in cancer metastasis [Andreasen et al. 1997]. The urokinase:plasminogen activator:urokinase receptor system can enhance cell migration and may have an important role in angiogenesis [Andreasen et al. 2000;

Choong & Nadesapillai 2003; Stahl & Mueller 1994]. In fact, the uPA, via the generation of plasmin, has been suggested to play a critical role in the extracellular matrix dissolution mediated by malignant melanoma cells [Montgomery et al. 1993].

The uPA molecule is a 50 kDa serine proteinase that is one of two types of plasminogen activator (the other being tissue plasminogen activator; tPA) which is able to convert the ubiquitous pro-enzyme, plasminogen, into plasmin which can degrade a range of proteins [Andreasen et al. 1997]. Plasminogen activation is thought to occur on the cell surface, where a specific urokinase receptor localises and enhances uPA activity.

It has been reported that membrane-bound MTf acts as a binding site for plasminogen on the cell surface and enhances its activation to plasmin (Figure 1.6; [Demeule et al.

2003; Michaud-Levesque et al. 2005b]). The process involved a specific interaction between sMTf with both pro-uPA and plasminogen that stimulates plasminogen activation by decreasing the Km of pro-uPA for plasminogen and by increasing the Vmax

28

of the reaction [Demeule et al. 2003]. Hence, MTf may target the formation of plasmin to the cell surface and protect it from inactivation by natural inhibitors [Michaud-

Levesque et al. 2005b]. Interestingly, sMTf was shown to antagonise this process, demonstrating that the balance between endogenous MTf and sMTf could modulate cell migration in melanoma and endothelial cells (Figure 1.6; [Demeule et al. 2003;

Michaud-Levesque et al. 2005b]). However, the physiological significance remains unclear, as the concentration of sMTf in the serum is exceedingly low [Kennard et al.

1996]. Critically, in all these studies examining angiogenesis and plasminogen activation and the effect of sMTf on these processes, it is notable that the Fe status of sMTf was unspecified. It is known that Fe-binding to the Tf super-family of proteins results in marked conformational changes that affect biological activity [Baker &

Lindley 1992]. Therefore, controlled studies examining the Fe-status of sMTf need to be performed to clearly delineate the role of this protein in angiogenesis, cell migration and plasminogen activation.

29

Figure 1.6. Overview of the postulated role of MTf in plasminogen activation and angiogenesis. It has been proposed that membrane-bound MTf binds plasminogen and stimulates activation through tissue plasminogen activator (tPA) and/or urokinase-type plasminogen activator (uPA), thereby increasing levels of plasmin [Michaud-Levesque et al. 2005b; Michaud-Levesque et al. 2007]. This could result in fibrinolysis, matrix degradation and endothelial cell detachment [Michaud-Levesque et al. 2007]. These processes may facilitate tumour cell invasion, angiogenesis and metastasis. Studies with recombinant sMTf may inhibit the above-mentioned processes by possibly competing with membrane-bound MTf for plasminogen [Michaud-Levesque et al. 2007]. This could provide a potential therapeutic strategy to prevent angiogenesis, metastasis and tumour growth. Adapted from [Suryo Rahmanto et al. 2007c].

30

A recent investigation examining the role of sMTf in tubulogenesis in vitro and angiogenesis in vivo have demonstrated that it can antagonise the pro-angiogenic effects of membrane-bound MTf expressing cells [Michaud-Levesque et al. 2007]. The mechanism of this process entails disruption of the balance between membrane-bound

MTf and sMTf, resulting in a reduction of plasminogen activation by tPA at the cell surface. This results in the limitation of the pro-angiogenic activity of MTf by sMTf and thus may provide therapeutic avenue for limiting tumour progression or growth (Figure

1.6; [Michaud-Levesque et al. 2007]).

These authors have also shown that sMTf could inhibit in vitro tubulogenesis of human umbilical vessel endothelial cells [Michaud-Levesque et al. 2007]. Furthermore, sMTf was shown to inhibit neovascularisation induced by SK-Mel-28 melanoma, basic fibroblast growth factor or VEGF1 [Michaud-Levesque et al. 2007]. However, these results are in contrast to previous studies by others described above which showed that sMTf itself was pro-angiogenic and involved VEGF1 [Sala et al. 2002]. In addition, there appears to be some paradoxical results from the same laboratory, with activation of plasminogen by pro-uPA being increased by sMTf [Demeule et al. 2003], while a later study indicated that sMTf decreased plasminogen activation [Demeule et al. 2003].

Clearly, further work is required to clarify the role of sMTf and membrane-bound MTf in plasminogen activation and angiogenesis. It is also important to note that recombinant sMTf may not represent the physiological form of MTf that is present on the membrane and at very low levels in the serum. Hence, the significance of these

31

findings remain unknown. At high doses, recombinant sMTf may be a therapeutic strategy to inhibit angiogenesis and tumour growth [Michaud-Levesque et al. 2007].

1.6.2. Blood brain barrier transport

Recent studies reported that human recombinant sMTf could transcytose and transport

Fe across the BBB and accumulate in mouse brain [Demeule et al. 2002; Moroo et al.

2003]. The trans-endothelial transport was temperature-sensitive and occurred through a saturable, active mechanism that was shown not to be via the TfR1 [Demeule et al.

2002]. It was suggested that recombinant sMTf was possibly interacting with the LRP.

[Demeule et al. 2002]. Interestingly, LRP has also been suggested to be the receptor for another Tf homologue, lactoferrin, in the brain [Fillebeen et al. 1999].

Both Demeule et al. and Moroo et al., reported that recombinant sMTf transported Fe into the brain more efficiently than Tf in both in vitro BBB models and in situ mouse brain perfusion [Demeule et al. 2002; Moroo et al. 2003]. However, two independent studies by others did not agree with these later findings. When sMTf protein uptake by in situ brain perfusion was compared with that of albumin, the results were comparable and sMTf uptake was not significant [Pan et al. 2004]. Additionally, Richardson and

Morgan demonstrated that when rats were injected intravenously with recombinant sMTf, 59Fe uptake in the brain was only 10% of that measured for Tf [Richardson &

Morgan 2004]. Although a small fraction of sMTf was taken up by the brain from the plasma, the uptake by the liver was much higher, a factor not assessed in the previous

BBB and brain perfusion studies [Demeule et al. 2002; Moroo et al. 2003]. Considering these latter data in conjunction with the low concentration of sMTf in the serum

32

(7 ng/mL in normal patients compared with 44 ng/mL in AD; [Kennard et al. 1996]) compared with Tf (2-3 mg/mL; [Richardson & Ponka 1997; Richardson & Morgan

2004]), it is doubtful that sMTf plays a significant role in brain Fe transport [Richardson

& Baker 1992b].

1.6.3. Alzheimer’s disease

Over the past eight years, it has been argued that MTf may play a role in AD. As discussed previously, Kennard et al., showed that Fe uptake was directly mediated by

MTf in CHO cells transfected with very high levels of hMTf [Kennard et al. 1995]. In other studies it was shown that: (i) MTf and TfR1 co-localised at the capillary endothelium in sections of human brains [Rothenberger et al. 1996] and (ii) that MTf was increased in the reactive microglia of amyloid plaques in brains from deceased AD sufferers [Jefferies et al. 1996]. Investigations have since examined the relationship between Fe, MTf, BBB and AD. Interestingly, AD patients appeared to have increased sMTf in their serum and cerebrospinal fluid and this led to the suggestion that sMTf was a “cellular hallmark” of the disease [Kim et al. 2001; Ujiie et al. 2002; Yamada et al.

1999].

More recent work has questioned many of the previous findings regarding the role of

MTf in AD pathogenesis. In contrast to the earlier reports of increased sMTf in the cerebrospinal fluid and serum of Alzheimer’s patients [Kennard et al. 1996], Desrosiers et al. have shown that there is no significant difference between sMTf levels in the serum, saliva and urine of AD patients when compared with healthy, normal controls

[Desrosiers et al. 2003]. Additionally, others have demonstrated that sMTf does not

33

bind the TfR1 or TfR2 or any other high-affinity receptor in human melanoma cells and other cell types in culture [Food et al. 2002; Kawabata et al. 2004]. Furthermore, sMTf inefficiently donates Fe to cells via a low affinity process consistent with surface pinocytosis [Food et al. 2002] and played little role in Fe supply to the rat brain and erythropoietic tissue [Richardson & Morgan 2004]. Collectively, the role of MTf in brain Fe metabolism and the pathogenesis of AD remains controversial, with little definitive data being available to support these hypotheses.

1.6.4. Eosinophil differentiation

Another MTf homologue is the 100 kDa chicken antigen, EOS47 [McNagny et al.

1996]. Originally identified on non-peripheral blood eosinophils, the protein has also been demonstrated in the chicken on the brush borders of kidney proximal tubules and intestinal villi, as well as liver sinusoids [McNagny et al. 1996]. While the protein shares high homology with hMTf (60% identity; 28 conserved cysteine residues) and is membrane-bound by a GPI-anchor, a non-conserved mutation (His253Arg) in the N- terminal indicates that EOS47 does not bind Fe [McNagny et al. 1996]. The lack of expression of EOS47 on peripheral blood eosinophils and other mature haematopoeitic cells has prompted suggestions that this protein is a marker of eosinophil differentiation

[McNagny et al. 1998; Querfurth et al. 2000]. However, the exact role of EOS47 in eosinophil differentiation remains unclear.

1.6.5. Chondrogenic differentiation and arthritis

There is data from two different organisms which show that MTf may play a role in chondrogenesis. In the rabbit, MTf is a 76 kDa protein expressed at high levels on the

34

surface of chondrocytes within cartilage [Kawamoto et al. 1998]. However, the largest body of data has been generated upon examination of chondrogenesis in the mouse

[Nakamasu et al. 1999; Nakamasu et al. 2001; Suardita et al. 2002]. As with the rabbit, expression studies of MTf mRNA expression in adult mouse tissues indicated that the highest levels were found in the cartilage [Nakamasu et al. 1999]. Elevated levels of

MTf were also demonstrated in the testes and lesser levels in other tissues, in agreement with other studies [Richardson 2000; Sekyere et al. 2005]. Upon examining the expression of MTf in vitro in the murine embryonic cell line, ATDC5, it was demonstrated that MTf expression paralleled that of collagen and aggrecan, suggesting that MTf plays some role in chondrogenic differentiation [Nakamasu et al. 1999;

Suardita et al. 2002]. It has also recently been shown that MTf can cause changes in cell surface shape and induce chondrogenic differentiation [Oda et al. 2003]. It is possible that high MTf levels in chondrocytes may modulate growth factor signaling pathways.

Considering these results, a patent for the application of assaying arthritis-associated sMTf in the plasma has been filed, although no data has been published in peer reviewed journals [Kato et al. 2001].

1.7. Melanotransferrin as therapeutic target

It has been suggested that MTf could be used as a delivery system across the BBB or for immunotherapy [Estin et al. 1988; Moroo et al. 2003]. However, both therapeutic applications have mounting evidence against them. While it has been demonstrated that recombinant sMTf can be transcytosed in an in vitro BBB model and accumulate in brain following in situ perfusion [Demeule et al. 2002; Moroo et al. 2003], this process is not more efficent than Tf both in vitro and in vivo [Pan et al. 2004; Richardson &

35

Morgan 2004]. The ability of drug-conjugated sMTf to maintain conformation, bind to a receptor and then be transcytosed into the brain would appear to be inefficient. Indeed, it was shown that sMTf does not bind to any high affinity receptor on a number of cell types and is not effectively transported into the brain [Food & Richardson 2002; Food et al. 2002; Richardson & Morgan 2004].

Immunotherapy has been suggested as an alternative, curative strategy in cancer therapy

[Ballen & Stewart 1997; Komenaka et al. 2004; McGee 1991; Nathanson 1979]. In the past, it appeared that MTf was a likely candidate for the treatment of malignant melanoma as: (i) MTf was originally believed to be mainly expressed in melanoma cells

[14]; (ii) inoculation of mice and monkeys with a recombinant vaccinia virus expressing

MTf induced both cell-mediated and humoral immunity [Estin et al. 1988]; (iii) MTf- specific CD4+ T cell clones could eradicate pulmonary metastases [Kahn et al. 1991]; and (iv) bi-specific could abolish melanoma metastases in SCID mice

[Riedle et al. 1998]. However, MTf has been detected at the protein and mRNA levels in normal tissues indicating that it is not tumour-specific [Kawamoto et al. 1998;

Richardson 2000; Rothenberger et al. 1996; Sciot et al. 1989; Sekyere et al. 2005], which may lead to non-specific toxicity of normal cells. Moreover, MTf expression is highly variable depending upon the individual melanoma tumour assessed [Brown et al.

1981a; Woodbury et al. 1980]. Hence, further studies on the therapeutic potential of immunotherapy needs to be performed to validate its use.

36

1.8. Summary and aims of the study

Over the past twenty-five years, the biological function of the Tf homologue, MTf, has remained unclear. Although the molecule specifically binds Fe(III), studies in cell culture indicate that neither the soluble or membrane-bound forms of MTf are significantly involved in cellular Fe uptake. However, the role of Fe in MTf function cannot be ruled out. Indeed, some progress has been made in elucidating the functional role of hMTf in health and disease. These include its purported roles in angiogenesis, cell migration and differentiation. Far more controversial are the possible roles of hMTf in brain Fe metabolism and AD.

Considering MTf has been implicated in a diverse range of biological processes, the exact mechanisms of MTf function in biological process remain to be elucidated. To achieve this aim, the specific objectives of this project were:

1. Phenotypic characterisation of the MTf knockout (MTf -/-) mouse.

2. Examination of the effect of MTf hyper-expression in mouse fibroblast and

human neuroepithelioma cell lines using mMTf and hMTf cDNA, respectively,

under the control of the cytomegalovirus (CMV) promoter.

3. Identification of novel molecular pathways which MTf may be involved in,

through whole-genome microarray analysis of the MTf -/- mouse, mouse fibroblast

and human neuroepithelioma cell lines.

4. Generation and phenotypic characterisation of the MTf hyper-expression (MTf Tg)

mouse.

37