The 67LR (67 kDa laminin receptor) is a cell-surface receptor with high affinity for its primary ligand. Its role as a laminin receptor makes it an important molecule both in cell adhesion to the basement membrane and in signalling transduction following this binding event. The protein also plays critical roles in the metastasis of tumour cells. Isolation of the protein from either normal or cancerous cells results in a product with an approx. molecular mass of 67 kDa. This protein is believed to be derived from a smaller precursor, the 37LRP (37 kDa laminin receptor precursor). However, the precise mechanism by which cytoplasmic 37LRP becomes cell-membrane-embedded 67LR is unclear. The process may involve post-translational fatty acylation of the protein combined with either homo- or hetero-dimerization, possibly with a galectin-3-epitope-containing partner. Furthermore, it has become clear that acting as a receptor for laminin is not the only function of this protein. 67LR also acts as a receptor for viruses, such as Sindbis virus and dengue virus, and is involved with internalization of the prion protein. Interestingly, unmodified 37LRP is a ribosomal component and homologues of this protein are found in all five kingdoms. In addition, it appears to be strongly associated with histones in the eukaryotic cell nucleus, although the precise role of these interactions is not clear. Here we review the current understanding of the structure and function of this molecule, as well as highlighting areas requiring further research.
In 1983, a receptor for the ECM (extracellular matrix) glycoprotein laminin was isolated and identified by three different laboratories from murine melanoma cells , human breast carcinoma cells  and normal muscle cells . The receptor could be extracted from the cell membrane and purified by laminin–sepharose affinity chromatography. The isolated purified receptor was a single proteinaceous entity with an apparent molecular mass (as estimated by SDS/PAGE) of approx. 67 kDa. The receptor bound laminin with high affinity and specificity, with a dissociation constant (Kd) of 2 nM [1,4]. Consequently, this receptor was named 67LR (67 kDa laminin receptor) and, more recently, LAMR1 (laminin receptor-1).
Cleavage of the purified protein with cyanogen bromide followed by microsequence analysis revealed a unique octapeptide with the sequence MLAREVLR . The full-length sequence of the protein has not been determined by direct protein sequencing, presumably due to a blocked N-terminus, but the discovery of this octapeptide enabled the identification and isolation of cDNA clones encoding the protein . Northern blotting showed that the mRNA encoding 67LR is approx. 1700 bp. Of these, 888 bp are the coding sequence, resulting in a protein of 295 amino acids and a calculated molecular mass of approx. 32.8 kDa. This precursor protein can also be detected in Western blots of cell lysates and has an apparent molecular mass of 37kDa – hence its name 37LRP (37 kDa laminin receptor precursor). The considerable discrepancy between the observed mass of the mature receptor form of the protein and the polypeptide encoded by its gene is incompletely understood (see below).
Laminin-1, the primary ligand of 67LR, is one of the components of the ECM and is the major glycoprotein of the basement membrane from all types of tissues. The basement membrane is a specialized ECM, comprising type IV collagen, nidogen (entactin), glycoproteins and proteoglycans . Laminin is a high-molecular-mass basement membrane glycoprotein, with a molecular mass of approx. 1 MDa. Electron microscopy of rotary-shadowed molecules revealed an overall structure in the shape of an asymmetric cross, with one long arm and three short arms (Figure 1) . Laminin is a heterotrimer of three subunits: α (∼400 kDa), β (∼200 kDa) and γ (∼200 kDa). Five different human α chains, three β chains and three γ chains have been identified, and they can assemble in various combinations to form at least 15 laminin isoforms that have different tissue distributions and appear at different developmental stages [7,8]. The three short arms of the cross are formed by the N-terminal regions of the three chains and an α-helical coiled-coiled long arm is formed by the more C-terminal regions. The three chains are also joined by interchain disulfide bonds at three sites. The globular domains of the short arms are separated by consecutive cysteine-rich EGF (epidermal growth factor)-like repeats. A large globular region (the G-region) in the α-domain, which is located at the end of the long arm, is further subdivided into five disulfide-bond-stabilized loops . Laminin is implicated in many biological processes, including cell adhesion, differentiation, proliferation, migration and neurite outgrowth, as well as tumour growth and metastasis . Its involvement in these processes is made possible by a number of specialized domains, identifiable as peptide motifs, which bind discretely to specific cell-surface receptors, such as 67LR, as well as to ECM molecules of the basement membrane .
Architecture of the laminin complex
Homologues to 37LRP/67LR can be found in all five kingdoms (archaebacteria, eubacteria, fungi, plants and animals) and are likely to be present in all organisms. Phylogenetic analysis of the sequence suggested that the acquisition of the laminin-binding capability by the 37LRP gene product is linked to the evolution of the C-terminus, particularly the appearance of the palindromic sequence LMWWML. This sequence appeared during evolution concomitantly with laminin and the basement membrane, the key foundation structure for multicellular tissue formation . Since only multicellular animals have laminin-containing basement membranes, it was clear that N-terminal homologies with proteins from unicellular organisms indicated alternative or additional functions in lower animals and in the other four kingdoms. Interestingly, some pathogenic organisms do have cell-membrane-associated receptors for laminin (presumably to aid in attachment to the basement membrane during invasion of the host). For example, Leishmania donovani (a unicellular haemoflagellate parasite, which is the causative agent of leishmaniasis) expresses a 67LR at its cell membrane. However, on the basis of partial sequence analysis and Western blotting, this protein appears not to be homologous with 67LR, although it does have functions in common [13–15].
Phylogenetic analysis carried out on 37LRP from different species indicated that all of these protein sequences are derived from orthologous genes, and that 67LR was, originally, a ribosomal protein that acquired the additional novel function of a laminin receptor through evolution . This can also be seen from the fact that the protein has been found in association with the ribosomes of species as diverse as the yeasts Saccharomyces cerevisiae  and Candica albicans , Arabidopsis thaliana , the sea urchin Urechis caupo  and the archaebacterium Haloarcula marismortui . It should be noted that multi-functionality is a property of other ribosomal proteins.
The N-terminal region of 37LRP/p40 (an acidic ribosomal protein of molecular mass approx. 40 kDa, identical with 37LRP) is highly homologous in eukaryotes as distantly related as Hydra vulgaris , the yeast Candida albicans  and the tapeworm Echinococcus granulosus . Furthermore, this protein has been conserved since archaebacteria in association with the cellular translational machinery, suggesting a direct descent of the vertebrate gene product from a primordial form of the RS2 (ribosomal protein S2) prokaryotic ribosomal protein [20,24]. Indeed, the RS2 protein from Methanobacterium thermoautotrophicum is 43% identical with a 198-amino-acid N-terminal region of human 37LRP.
In metazoans, particularly the vertebrates, 37LRP proteins exhibit very high identity along their complete length, including the C-terminal region, with at least 98.3% similarity between the mouse, bovine and human sequences, and 99% similarity between rat and human sequences. Since the formation of 67LR from 37LRP/p40 has only been described in vertebrates, it is likely that the conserved C-terminus (which is most divergent in other organisms) is responsible for the laminin-binding capabilities of this protein.
Although 67LR isolated from cells migrates at approx. 67 kDa on denaturing polyacrylamide gels, the corresponding gene only encodes a protein of approx. 33 kDa , which migrates with a molecular mass of approx. 37 kDa on SDS/PAGE. This gene product, 37LRP, is recognized by some (but not all) antibodies raised against 67LR . The cause of this anomaly has not been determined. The most common explanations for such behaviour in proteins are post-translational covalent modification or the formation of very tightly associated homo- or hetero-dimers. Although reducing and denaturing gel systems, such as SDS/PAGE, normally completely disassociate protein complexes into their components, it is not unknown for very tightly bound complexes to resist this, at least partially (for example, see ). Interestingly, there is evidence for both these possibilities, but the exact cause of the molecular mass discrepancy remains a mystery. One hypothesis states that the 33 kDa protein encoded by the cDNA clone is a precursor for 67LR which undergoes some form of modification in order to become the mature protein.
The translated sequence of the 37LRP gene also reveals some interesting and unusual features. There are no N-linked carbohydrate-attachment sites or typical transmembrane domains. There is no recognized signal sequence at the N-terminus for entry and translation in the endoplasmic reticulum, which is the conventional starting point for transmembrane proteins destined for translocation to the cell surface. However, the N-terminal region contains a hydrophobic segment (amino acids 86–101) that has been hypothesized to act as a transmembrane domain. The position of this putative domain is consistent with the immunoreactivity of 67LR in intact cells to a panel of antibodies raised to peptide sequences from different parts of the molecule . Finally, there is an unusual C-terminal 70-amino-acid segment, which is trypsin-resistant and highly negatively charged which contains five repeats of the sequence (D/E)W(S/T).
A precursor–product relationship has been established between 37LRP and 67LR. There are two main bodies of evidence that support this hypothesis. The first evidence was provided by the cross-reactivity of two anti-(37LRP peptide) antibodies with a 67 kDa protein on Western blots. The second arose from pulse–chase experiments performed using 37LRP from [35S]methionine-labelled human melanoma cells. During these experiments, radiolabelled 37LRP alone was initially detected, later followed by the appearance of a similarly labelled 67 kDa protein .
One translational modification has been found that distinguishes 67LR from its precursor, and that is fatty acylation. It has been suggested that this modification is a prerequisite to homodimerization  or heterodimerization . Evidence for this hypothesis came from experiments where 67LR was expressed in mammalian cells and the purified protein treated with methyl trans-esterification reagents. GC-MS (gas chromatography MS) identified the covalently bound fatty acids palmitate, stearate and oleate, showing that acylation of 67LR had occurred; the molecular mass [by MALDI–TOF-MS (matrix-assisted laser-desorption ionization–time-of-flight MS)] was 66.7 kDa and showed no heterogeneities that are expected of glycosylated proteins . Lack of glycosylation was confirmed as glycosidase, glycanase or neuraminidase treatments made no difference to the mass of the protein. In a separate study, inhibition of fatty acid synthesis by cerulenin resulted in little or no 67LR formation; instead, there was an accumulation of 37LRP . Thus the mechanism by which 37LRP is converted into 67LR is unlikely to involve glycosylation, but may include acylation.
The amino-acid compositions of 67LR and 37LRP are identical , suggesting a homodimer. Yet 37LRP does not associate in a Y2H (yeast two-hybrid) assay . However this may be because acylation (or some other post-translational modification) is absolutely required for the formation of dimers and that the required post-translational modification is not carried out in yeast cells. In contrast, Butò et al.  have proposed that the formation of the 67LR is due to a heterodimerization between 37LRP and another protein carrying epitopes from galectin-3. This lectin, which was identified in melanoma cells as a laminin-binding protein and initially named HLBP31 (31 kDa human laminin-binding protein), binds to β-galactosides  and the poly-N-acetyl-lactosamine residues of laminin. The hypothesis of heterodimerization is supported by the fact that lectin epitopes are present in 67LR, but not in 37LRP, that laminin recognition by 67LR is lactose-dependent and that 67LR can be eluted from laminin-affinity chromatography columns by lactose . However, Gauczynski et al.  obtained conflicting results. It was shown that both 37LRP and 67LR are present on the plasma membrane of N2a cells , a cell line which does not express galectin-3. This suggests that galectin-3 is not required for the membrane targeting of 67LR, and that, in some cell types at least, 37LRP does not associate with it.
FUNCTIONS OF 67LR AND 37LRP/p40 IN NORMAL CELLS
Laminin, similar to the other main component of basement membranes, collagen IV, is capable of self-assembly into polymeric sheets. In vivo, laminin-1 oligomerization has been postulated to require receptor binding . The interaction of the collagen IV network with that of laminin is facilitated by the proteoglycan perlecan and the protein entactin (also known as nidogen). The cross-linking of laminin to collagen IV by entactin is greatly enhanced by zinc ions and this is thought to be due to the presence of nested zinc fingers that are present in 12 out of the 42 cysteine-rich EGF-like repeats found in the N-terminal short arms of laminin chains α, β and γ. In the case of entactin, the high-affinity site formed in the presence of calcium and zinc is one of three zinc fingers on the short arm of the γ1 chain of laminin (Figure 1) .
Interactions with laminin
It is well-established that 67LR interacts directly with laminin, and at least three regions of the C-terminal domain of the receptor are thought to be involved: (i) the most C-terminal 53 amino acids containing four acidic TEDWS repeats; (ii) a putative helical stretch between amino acids 205–229; and (iii) a heparan-sulfate-dependent laminin-binding region from 161–180 (Figure 2) .
Main functional regions and binding sites of 67LR
YIGSR: the minimal ligand sequence of laminin for interaction with 67LR205–229?
The minimal sequence of laminin that binds to 67LR was first identified from studies using protease-resistant laminin fragments (see below). The protease-resistant biologically active core (see Figure 1) was confined to the three short arms, without the globular domains. The actual binding domain was eventually determined to be one of the LE (laminin-like EGF) domains on the β1 short arm of laminin-1 [37,38]. However, rotary-shadowing electron microscopy showed that 67LR predominantly bound to the long arm of whole laminin, just below the cross intersection .
The minimum sequence needed to displace 67LR binding to radiolabelled laminin is YIGSR. The binding activity of this synthetic peptide may be increased by the presence of an amide group on the C-terminal arginine residue . The corresponding sequence of the receptor that binds such YIGSR-like laminin-derived ligands has been mapped to residues 205–229 of human 67LR, a peptide sequence that is predicted to be α-helical [41,41a] (Figure 3). Modelling of the peptides derived from phage-display mapping has suggested that the helix has two faces. One is a polar heparan-sulfate-binding face, which presents a series of positively charged residues (Lys212, Lys220 and Lys224, in the human sequence) flanked by acidic residues (Glu208, Glu213, Glu219 and Glu225). The presence of both positive and negative charges in this face was proposed to offer the possibility of conformation-dependent binding. The other face is more hydrophobic and binds the laminin β1 LE domain peptide CDPGYIGSR (sometimes referred to as peptide 11) . The peptide-binding properties are independent of heparan sulfate. Zinc ions may enhance this interaction, possibly through the cysteine-rich zinc-finger-like sequences on the β1 chain on laminin. This is certainly the case in L. donovonia, which expresses a cell-surface laminin receptor (see above). This receptor has some common functions with 67LR (although the two proteins appear to be unrelated), and interacts with laminin in a zinc-dependent manner. Interaction of the L. donovonia receptor with laminin has been localized to a peptide (sequence CDPGYIGSR, i.e. identical with peptide 11), which is located in a cysteine-rich repeat in domain III on the β1 chain of laminin-1 [14,37,41,41a]. A similar sequence is found in the C-loop of murine EGF, which, as a synthetic linear peptide, has 67LR antagonistic properties . This suggests that the L. donovonia laminin receptor may interact with laminin in a similar way to 67LR (at least at one site of interaction). How far these parallels can be drawn is not clear as the parasite protein is currently relatively uncharacterized.
Homology model of human 67LR and structures of the proteins used as templates
Peptide G: 67LR161–180
A second laminin-interaction site has been identified in 67LR: the so-called peptide G, which spans residues 161–180 (IPCNNKGAHSVGLMWWMLAR) [25,43]. A synthetic peptide corresponding to this sequence binds to laminin with a high affinity (Kd=50 nM) . Peptide G also binds directly to the sulfated polysaccharide heparin. Heparin and laminin compete for binding to peptide G and this may represent a further level of regulation of the interaction between 67LR and the basement membrane . NMR studies on peptide G showed that it adopts a stable 310-helix in solution. . (Note, however, that in our model of human 37LRP, on the basis of ribosomal protein structures, see below, this region adopts an α-helical structure.) Both the stability of this helix and the affinity for laminin were reduced by deletion of the palindromic sequence LMWWML. Phage-displayed peptides which showed similarity with peptide G were eluted by the YIGSR-containing peptide 11 sequence (see above), suggesting that both 205–229 and 161–180 of 67LR can bind to the same minimal YIGSR region of the β1 chain of laminin-1 . Finally, phage display also revealed clones with inserts similar to the C-terminal TEDWS repeats of 67LR, suggesting a third region of 67LR that can bind to laminin-1 .
67LR as a co-receptor
In addition to direct interactions with laminin, it has been proposed that 67LR facilitates interactions between laminin and integrins. Integrins are a large family of heterodimeric transmembrane molecules consisting of non-covalently associated α and β subunits . Treatment of human carcinoma cells with laminin increased the membrane localization of the α6 and β4 integrin subunits and of 67LR , suggesting an increased translocation of the two associated receptors from the cytoplasm to the cell surface. Treatment of these cells with cytokines down-regulated the expression of integrin subunit α6 and concomitantly decreased the level of 67LR on the cell membrane. In addition, down-regulation of the expression of the gene encoding the α6 integrin subunit, by an antisense oligonucleotide, was accompanied by a proportionate decrease in 67LR cell-membrane localization. Moreover, the total amount of the mature form of 67LR did not change, nor did the total level of 37LRP, indicating that the antisense treatment does not influence overall 37LRP gene expression. In contrast, up-regulation of α6 was shown to increase the amount of 67LR at the membrane, suggesting co-regulation of their localization and interaction between these two molecules. This interaction between 67LR and the α6 integrin subunit seems to occur in the cytoplasm, and is followed by co-translocation to the cell surface. This hypothesis is supported by co-immunoprecipitation of 67LR and the α6 subunit with an anti-α6 monoclonal antibody. The close association between these two receptors suggests that they are involved in some of the same processes.
In vascular endothelial cells 67LR may have an important role in angiogenesis . It was shown that 67LR is expressed by the proliferating retinal vasculature during neovascularization, whereas the quiescent vascular cells in normal control retinas showed comparatively low expression levels. 67LR expression is highest in rapidly proliferating endothelial cells in culture and declines when cells become contact-inhibited, suggesting that 67LR levels need to be constantly renewed during the cycles of rounding-up and re-attachment which dividing cells go through during rapid proliferation . Recently, 67LR has been shown to be involved in the mobilization of HSCs (haemopoietic stem cells) by granulocyte-colony-stimulating factor, and the expression levels of 67LR positively correlate with mobilization efficiency of HSCs . HSC cells stimulate angiogenesis and are thought to give rise to new endothelial cells.
Electron microscopy studies have suggested export of 67LR toward the ECM, but there was no actual evidence provided for the shedding of 67LR from cells. However, cultured cells do release 67LR into the culture medium . Comparison of the shed and membrane-associated 67LR showed no change in molecular mass. Interestingly, the shed 67LR was found both as monomers, as well as associated in high-molecular-mass complexes. Although the shed 67LR is no longer anchored to the membrane, it retained its ability to bind to the cell surface. The observation that lactose increased the release of 67LR suggests that a lectin-type interaction may be being disrupted .
RIBOSOMAL LOCALIZATION OF 37LRP AND IDENTITY WITH p40
In 1988, an abundant mRNA species that is under translational control was identified in mouse tumour cells . The mRNA codes for a 33 kDa polypeptide called p40 . Mouse p40 shows nearly complete similarity to human 37LRP/67LR, including an octapeptide laminin-binding sequence (see below). Essentially, this p40 protein is the murine equivalent of human 37LRP. Ribosomal localization of 37LRP, and its role in translational initiation was first elucidated in a study by Auth and Brawerman . Sucrose-gradient sedimentation of cytoplasmic fractions from murine sarcoma and erythroleukaemia cells revealed that 37LRP was associated with particles ranging from 40S to 60S. These proved to be free ribosome subunits and polysomes. A smaller pool of 37LRP appeared to be soluble. 40S–60S ribosomal subunits containing 37LRP/p40 were also found in U. caupo oocytes and embryos . The p40-containing fractions were more accurately defined in size and sedimented at a position corresponding to the 40S ribosomal subunit.
In reticulocyte lysates, initiation of protein synthesis caused loss of polysome-associated 37LRP, which could be prevented by using the translation elongation inhibitor cycloheximide. 37LRP that was associated with the free cytosolic ribosome subunits, however, was not affected by cycloheximide treatment and was instead released. Reconstitution of polysomes was accompanied by their uptake of 37LRP. 37LRP may therefore shuttle between free ribosome subunits and polysomes as and when required. In both hydra (H. vulgaris) and mammalian fibroblast cells [21,54], 37LRP/p40 was found in association with the actin cytoskeleton. There appeared to be a cell-cycle-dependent change in localization, with p40 accumulating around the nucleus at prophase during chromosome condensation. As mitosis continued, the staining of p40 became cytoplasmic, and then became associated with the cytoskeleton again as cytokinesis ended. This dynamic re-distribution of p40 mirrors that of ribosomes as they move from soluble to cytoskeletal or endoplasmic reticulum locations.
The human 37LRP gene product (UniGene ID, Hs.181357; ribosomal protein name, RPSA) is found in the ribosomes of all tissues investigated . In human heart tissue, 37LRP (the human homologue of bacterial S2 ribosomal protein) is part of the mitochondrial proteome, where it functions as a mitochondrial ribosomal protein . The S2/RS2/S2P/p40/37LRP protein is not limited to mammals, and homologues have been identified in many species [47,53]. A 37LRP-like chromosome-encoded gene product Mrp4p (systematic gene name, YHL004W) was originally discovered in the mitochondrial ribosomes of the budding yeast, S. cerevisiae. It is 33% similar to human 37LRP and a similarity with Escherichia coli S2 protein was noted . Budding yeast has, in fact, two more genes, RPS0A (YGR214W) and RPS0B (YLR048W), which encode proteins that are similar to 37LRP. The disruption of either of the latter two genes alone resulted in reduced cell proliferation, and deletion of both was lethal .
37LRP as a structural component of the ribosome
A number of medium- and high-resolution structures of ribosomes from a variety of species have been solved [58–64]. Examination of the available high-resolution ribosome structures suggests a structural role for 37LRP/p40 homologues, such as the S. cerevisiae protein Rps0a. It makes a number of contacts with other proteins and limited ones with rRNA molecules. Modelling of human 37LRP with the program SwissModel [65,66] returns a result based on the structure of Rps0a (PDB ID, 1S1H ) and on ribosomal protein S2P from Archaeoglobus fulgidus (PDB ID, 1VI5 and 1VI6, ). Both these molecules and the model derived from them have compact globular structures in which a parallel β-sheet structure is surrounded by four α-helices (Figure 3). Of course, this is an automatically generated model on the basis of proteins from organisms evolutionarily distant from humans and some caution needs to be exercised in interpreting it. Furthermore, the C-terminal 43 residues of 67LR do not align with the ribosomal proteins and are thus not included in the model. However, the results are similar to those obtained by more detailed homology modelling using ribosomal protein S2P from Thermus thermophililus . The purification and crystallization of recombinant human 67LR was reported several years ago [70,71]. However, this has not been followed by a description of the high-resolution structure, and nor has the structure been solved by other groups.
Although 67LR has no clearly recognizable transmembrane region, a stretch of sixteen predominately hydrophobic amino acids (residues 86–101) has been proposed as a putative membrane-spanning region [25,72]. Additional evidence for 67LR as a conventional transmembrane receptor with two distinct domains separated by the plasma membrane comes from the observation that antipeptide antibodies raised against sequences from residue 107 to the C-terminus bind to intact cells, whereas antibodies specific to sequences from the first 103 residues reacted exclusively with permeabilized cells . This provides evidence that, under certain circumstances, a C-terminal domain of the protein may be translocated through the cell membrane to create the extracellular laminin-binding region of functionally active 67LR. In the model structures of human 37LRP (Figure 3), the laminin-interacting sequence, peptide G (see above), forms part of a β-strand, linking a region of random coil and an α-helix. This α-helix includes the palindromic sequence and is only partially surface exposed; the β-strand and much of the linking region is buried. For both membrane insertion and high-affinity laminin recognition by peptide G, considerable conformational change would be required (Figure 4). However, it should be noted that, in contradiction of the antibody-binding results, the authors of the more detailed model  considered that the single-domain fold was too compact and stable to permit radical rearrangements of this nature, and thus concluded that the central region of 67LR is unlikely to undergo membrane insertion.
Considerable conformational changes would be required to permit 67LR to act as a transmembrane protein and for peptide G to interact with ligands such as laminin
THE ROLE OF 37LRP IN THE NUCLEUS
There is evidence to suggest that 67LR is a component of the nuclear machinery [46,74]. 37LRP has been localized to the cytoplasmic, perinuclear and perichromosomal regions of the cell [46,74]. Further biochemical analysis showed that 37LRP was attached to both chromatin and the nuclear envelope within the perinuclear region . 37LRP lacks a clearly identifiable nuclear localization signal, but the heparin-binding multifunctional protein midkine binds to 37LRP and acts as a chaperone in the transport of the complex to the nucleus. 37LRP can be bound to DNA cellulose and eluted by moderate ionic strength solutions (0.3 M NaCl). Mixing the protein with a nuclear extract increased the salt concentration (to 0.5 M NaCl) required for elution, suggesting that 67LR tightly associates with some nuclear structures . Affinity chromatography with immobilized 37LRP demonstrated interactions with histones H2A, H2B and H4 . The consequences of this association with chromatin components are not clear. Furthermore, these interactions may be specific to animals. 37LRP from A. thaliana shows sequence similarity to the human form . However, immunofluorescence microscopy showed that it was distributed evenly through the cytosol of the cell, but was absent from the nuclei .
FUNCTIONS OF 67LR AND 37LRP/p40 IN DISEASE
Roles in tumour invasion and aggressiveness
Tumour invasion and metastasis are complex processes involving multiple interactions of tumour cells with host cellular and extracellular structures. One critical event in cancer spread is the formation of micro-metastases at secondary sites. Cancer cells, detached from the primary tumour site, invade new sites by a three-step process. Firstly, there is attachment of the cancer cells to basement membrane components. This is followed by local degradation of the basement membrane by type IV collagenase and other proteases, and finally movement of the cells into the adjacent tissue. The first step may be mediated in part by specific cell-surface receptors which bind to laminin in the basement membrane . Thus it is hardly surprising that interactions between 67LR and laminin are involved in metastatic forms of many cancers.
67LR, the basement membrane and cancer metastasis
Laminin serves as a major adhesion substrate for invasive cancer cells. The observation that laminin bound to the cancer cell surface in a saturable fashion and with high affinity suggested the presence of a high-affinity laminin-binding protein at the cell surface . Several cellular receptors for laminin molecules have been identified, including 67LR. Consequently, the molecule has attracted considerable attention from researchers studying cancer progression and metastasis. Although considerable progress has been made in understanding the role played by 67LR in these diseases, progress has been hindered by a lack of understanding of the basic biochemistry and cell biology of this molecule.
An increase in the expression of 67LR (compared with the corresponding normal tissue) has been found in a variety of common cancers, including breast , cervical , colon , gastric , hepatocellular , lung , ovary , pancreatic , prostate  and thyroid  carcinomas. In many cases, a positive correlation with aggressiveness or metastatic potential is also found.
Basement membranes separate epithelia from the underlying stroma and constitute the major natural barrier against tumour invasion. Interactions between cancer cells and laminin have been shown to play a critical role during this process. Indeed, the ability of cancer cells to attach to laminin has been correlated with their metastatic potential . Cells can interact with laminin via several cell-surface laminin-binding proteins, including 67LR, dystroglycan and the integrins α3β1, α6β1, α6β4 and α7β1 ; the high number of free laminin receptors on the tumour cell surface may facilitate interaction of these cells, through laminin, with the vascular basement membrane . For example, antiserum (raised against a synthetic peptide corresponding to a region close to the C-terminus of 67LR) had an inhibitory effect on cell adhesion . However, because laminin can polymerize or self-assemble, laminin receptors already occupied by endogenous laminin can still (somewhat counterintuitively) bind to exogenous laminin substrata. In support of this, it has been found that antisense laminin vectors down-regulate the adhesion of a cell to laminin substrates , and anti-laminin antibodies inhibit tumour cell attachment to umbilical endothelium and reduce experimental metastasis in vivo .
Alterations in the synthesis and secretion of laminin are also associated with cancers, in which laminin chains may be either up-regulated or absent compared with normal tissues . For example, laminin is produced by neuroblastoma , mesothelioma , oral squamous carcinoma , cervical , breast , lung , colon , melanoma  and prostate  cancers. In choriocarcinoma cells, approx. 25% of laminin is secreted by the cells and the remaining 75% is retained on the cell surface . However, a correlation between tumour-associated laminin levels and disease stage or outcome is often obscure  or absent .
P1: the protease resistant core of laminin-1
There are two factors that contribute to variation in laminin levels in tumours. Detection of laminin by immunocytochemistry may give false negatives due to misassembly of the laminin secreted by cells into aberrant laminae . A further complication is that laminin produced by invasive tumours may be proteolytically cleaved by tumour-associated proteases, such as MMPs (matrix metalloproteinases; gelatinase)  or cathepsin B [98,104]. Prolonged incubation of laminin-1 with trypsin, chymotrypsin, elastase, subtilisin, Staphylococcus aureus protease  or pepsin  results in a common protease-resistant fragment, P1, of approx. 280 kDa in mass. This soluble laminin fragment (Figure 1) has been found at elevated levels in the sera of advanced carcinoma and leukaemia patients [107,108]. P1 retains many (but not all) of the bioactivities associated with native laminin. P1 is equipotent with laminin in enhancement of epithelial cell proliferation  and displaces 125I-labelled laminin from gingival epithelial cells . Similar to laminin, P1 supports cell adhesion; but unlike laminin does not stimulate cell spreading . The P1 fragment displaces radiolabelled laminin from 67LR, but has lost the ability to bind to collagen IV . Other activities retained by P1 include stimulation of type IV collagenase secretion from cancer cells  and induction of eNOS (endothelial nitric oxide synthase) expression .
Most of the activities of the P1 fragment are due to its interaction with 67LR. Indeed, the exact binding site for 67LR emerged from such studies and was finally located to the short arm of β1, where the minimal 67LR-interacting sequence was found (described above). In particular, the synthetic laminin β1 nonapeptide (CDPGYIGSR) and pentapeptide (YIGSR) are both chemotactic towards cancer cells and endothelial cells [42,113]. The solid-phase YIGSR peptide supports cell adhesion, but not spreading [37,40], and the induction of eNOS by P1 is blocked by YIGSR .
It has been observed that the 67LR-derived peptide G is able to inhibit the binding of tumour cells to endothelial cells, but, quite unexpectedly for an antagonist of attachment, this laminin-binding 67LR mimic was found to increase the metastatic spread of human melanoma in nude mice . This may be explained by the effects that 67LR binding exerts on the conformation of laminin. Treatment of laminin-1 with peptide G increases the degradation of laminin by cathepsin B; significantly, YIGSR blocks this effect . The effect of peptide G is to change the degradation pattern produced by cathepsin B; only when pre-treated with peptide G does laminin change conformation to reveal a cryptic cathepsin B cleavage site. The result of peptide-G-assisted cathepsin B cleavage is the release of a highly chemotactic fragment from the G region at the end of the long arm (Figure 1) . Peptide-G-modified laminin effectively increases the affinity of a cell for laminin by increasing the number and affinity of α6-containing integrins [114,115]. The coincidence of cathepsin B with laminin degradation in cancer , and the shedding of 67LR from cancer cells, when considered with the effects of the 67LR mimic (peptide G) on experimental tumour progression, suggest the effects described above are relevant to remodelling of the tumour micro-environment and invasiveness in vivo. Significantly, peptide-G-remodelled laminin selectively increases expression of proteases known to be involved in invasiveness: MT1 (membrane type 1)-MMP and MMP-2, among others .
Malignant breast carcinoma tissues have more unoccupied laminin receptors on their cell surfaces than do non-metastatic breast tumours. This may be due to the increased biosynthesis of 67LR in the more metastatic breast tissue. The level of 37LRP mRNA was quantified in colon adenocarcinoma cells: a 2- to 23-fold increase in levels of 37LRP mRNA compared with those in the adjacent normal mucosa was observed . The level of laminin receptor mRNA was observed to be significantly correlated with the number of laminin receptors . This suggests that the amount of laminin receptor mRNA available for biosynthesis of receptor may be a rate-limiting control step in the regulation of laminin-mediated cellular attachment to basement membranes and is thus a potential site for therapeutic intervention.
A significant correlation between levels of expression of laminin receptor and progression of the disease (as evaluated by the Dukes' classification, the most common method used to estimate the invasive and metastatic properties of colorectal carcinomas) was observed. It was postulated that cancer cells in the primary tumour with expression of high levels of laminin receptor may have a higher metastatic potential. This is also supported by the observation that the amount 37LRP mRNA in poorly differentiated colon carcinomas is as high as or higher than that in well-differentiated carcinomas .
Laminin treatment of human melanoma cells induced a specific increase in levels of both 37LRP and 67LR . This was shown using a combination of immunogold electron microscopy and specific antibodies that distinguish between 37LRP and 67LR. This is consistent with the hypothesis that the 37 kDa polypeptide is the direct precursor of 67LR. A positive feedback loop has been hypothesized in which binding of laminin to the melanoma cell surface mediated by laminin receptors induces the synthesis of 37LRP. The consequence of this is maturation of 37LRP into 67LR and delivery to the cell surface of more laminin-binding proteins in a feedback loop results in enhanced attachment of the melanoma cells to the basement membrane during invasion and metastasis.
67LR AS A RECEPTOR FOR BACTERIA, VIRUSES AND PRIONS
Some pathogens need to enter their host's cells as part of their life cycles. In order to do this, many of them subvert naturally occurring cell-surface receptors. There is growing evidence that some bacteria and viruses use 67LR in this way. In addition, PrP [the prion protein which is believed to be responsible for spongiform encephalopathies, such as BSE (bovine spongiform encephalopathy), scapie and kuru] may also be internalized by a pathway which involves 67LR.
67LR acts as receptor protein for alphaviruses, such as the Sindbis virus, and appears to be important for entry of the virus into mammalian cells . The 67LR/37LRP protein also interacts specifically with the major dengue virus serotypes . The protein may also be important in bacterial infections, such as bacterial meningitis development and the internalization of the E. coli K1 strain into brain endothelial cells . Using Y2H screening, 37LRP was selected from a prey library using CNF1 (cytotoxic necrotizing factor 1) from E. coli as bait; 37LRP anti-sense treatment prevented CNF1-mediated RhoA activation and bacterial uptake . Furthermore, the mature 67LR promotes the internalization of CNF1-positive E. coli .
Using the Y2H system and direct cell-binding studies on neuronal and non-neuronal cells, two binding domains in the PrP which mediate interaction with 67LR were identified. The first one binds directly to the 67LR, whereas the second one depends on the presence of HSPGs (heparan sulfate proteoglycans) on the cell surface (Figure 5) .
Key functional regions in PrP and its interaction sites with 67LR
Human PrP (a 235-residue glycoprotein) is anchored to the cell membrane by a C-terminal GPI (glycosylphosphatidylinositol) moiety. It contains one disulfide bridge and five N-terminal octapeptide repeats . PrP acts as a copper-binding protein via four conserved histidine residues in the N-terminal region of PrP . PrP is at found at particularly high concentrations at synapse junctions where it is required for synaptic copper ion uptake . This copper may then be made available to cupro-enzymes, such as the copper/zinc superoxide dismutase.
PrP functions as a laminin receptor and binds to the C-terminal domain of the γ-1 chain . This interaction is crucial for the process of neuritogenesis. Inactivation or loss of PrP caused retraction of the neurite extensions of PC12 cells growth on a laminin substrate . An interaction between 37LRP/67LR and PrP was first established in a Y2H screen of a HeLa cDNA expression library for PrP-interacting proteins . This selected a number of clones encoding 37LRP, as well as Hsp60 (heat-shock protein 60) molecular chaperones. The physiological relevance of the 37LRP/67LR and PrP interaction was revealed in tissues and cells of scrapie compared with non-scrapie-infected mice. Tissues with high levels of PrPsc (the modified pathogenic form of PrP) accumulation displayed correspondingly high levels of 37LRP, in particular the brain and pancreatic tissue . Both 37LRP and 67LR are expressed on the surface of mouse cortical cells and both forms may act as PrP receptors . The surface expression of functional 37LRP/67LR appears to be a prerequisite for both binding and internalization of PrP, as 37LRP mutants lacking the putative transmembrane domain (residues 86–101) were not retained at the membrane, but mutated 37LRP was isolated from the extracellular space. Neither binding nor internalization of PrP was found to occur on these cells.
The Y2H system was used to map the sites of interaction between PrP and 37LRP/67LR . There are two 37LRP interaction sites on the PrP molecule (Figure 5). The first spans amino acids 144–179 of PrP. This functions as a direct binding site, making contact with residues 161–179 of 37LRP. It is of interest to note that the same region (161–180) of 37LRP corresponds to one (of three) laminin-binding sites, which is known as the peptide G region [25,32]. The second PrP site is an indirect interaction site and is mediated by the presence of HSPGs. This was detected upon expression of 37LRP lacking the PrP–laminin interaction domain (peptide G). This indirect binding site maps to amino acids 53–93 on PrP, and either 101–160 or 180–295 of 37LRP.
ROLE OF 67LR AND LAMININ-INDUCED SIGNALLING
At first, the interaction of 67LR with laminin was considered to be a matter of simple adhesion. However, it has become increasingly clear that 67LR activation also induces additional dynamic processes (increased filopodia, induction of directional motility and modulation of gene expression) that must be a result of active signal transduction within the cell. At present, 67LR signalling is barely understood, but studies have begun to elucidate potential pathways.
One of the mechanisms by which laminin contributes to metastatic spread is the induction of an increase in MMP-2 activity, an ECM-degrading endopeptidase which plays a key role in invasion and metastasis . The involvement of 67LR in the induction of MMP-2 activity was demonstrated by the fact that cells expressing reduced levels of 67LR have lower MMP-2 mRNA level and activity .
Givant-Horwitz et al.  investigated the role of the 67LR in mediating the effects of laminin and the involvement of the MAPK (mitogen-activated protein kinase) cascades and DUSPs (dual-specificity MAPK phosphatases) in laminin signalling in human melanoma cells. MAPK subgroups include: ERKs (extracellular-signal-regulated kinases), JNK (c-Jun N-terminal protein kinase)/stress-activated protein kinase and p38 MAPK. The ERK pathway plays a role in cell proliferation, survival and differentiation. The JNK and p38 pathways are activated in response to chemical and environmental stress and to inflammatory cytokines. MAPK are inactivated by serine/threonine phosphatases, tyrosine phosphatases and DUSP family members, including MKP-1 (MAPK phosphatase 1), PAC-1 (phosphatase of activated cells 1), MKP-4 and MKP-5.
It was shown that cells expressing a reduced level of 67LR demonstrate a less aggressive phenotype and a diminished attachment to laminin. Moreover, 67LR induces prolonged dephosphorylation of ERK, JNK and p38 MAPK, and additional exogenous soluble laminin-1 induces a further temporary dephosphorylation, independently of the 67LR level. In addition, the increase in MAPK phosphorylation in cells expressing reduced levels of 67LR is accompanied by a significant reduction of the MKP-1 mRNA level . Overexpression of the enzyme MKP-1, which dephosphorylates ERK, JNK and p38 MAPK, has been found in several malignancies. These findings suggest that a laminin-related signal transduction pathway exists and that reduced activity of some MAPKs is associated with increased malignancy of tumour cells .
Association with TIMAP [TGF-β (transforming growth factor-β)-inhibited membrane-associated protein]
In endothelial cells, TGF-β treatment caused the strong repression of a previously unidentified gene which was named TIMAP . Sequence analysis of human TIMAP cDNA predicted a protein consisting of an N-terminal nuclear localization signal, a PP1 (protein phosphatase 1)-binding motif (KVSF), followed by four central ankyrin repeats, and a C-terminal CaaX box motif. The latter motif is isoprenylated to allow anchoring to the inner leaflet of the plasma membrane. Full-length TIMAP, used as bait in a Y2H screen, identified 37LRP as an interaction partner; TIMAP interacts with 37LRP in vitro and with 67LR in vivo .
TIMAP binds to the N-terminal cytoplasmic domain of 67LR and the interaction region on TIMAP is confined to the fourth ankyrin repeat (residues 261–290). Immunoprecipitation experiments demonstrated that mature 67LR co-immunoprecipitated with TIMAP and vice versa. In transfected cells overexpressing TIMAP, both 67LR and PP1 associate in a TIMAP-dependent manner, and this interaction is dependent on TIMAP being isoprenylated and hence localized to the membrane. This finding is confirmed in wild-type cells that have detectable levels of endogenous TIMAP. The TIMAP–PP1–67LR complex is concentrated in endothelial filopodia.
The function of the above complex appears to be to control the levels of 67LR phosphorylation. TIMAP-targeting of PP-1 is essential to dephosphorylate 67LR. If the complex is disrupted, 67LR phosphorylation increases and, with it, filopodia formation. Very recently, the multi-functional GSK-3 (glycogen synthase kinase 3) has been shown to play a role in the control of the TIMAP–PP1–67LR complex . PKA (protein kinase A) phosphorylates TIMAP on Ser337 and this primes it for a second phosphorylation on Ser333 by GSK-3β. Double-phosphorylated TIMAP activates the PP1 member of the complex, and this leads to an increase in dephosphorylation of 67LR and TIMAP itself; if GSK3 is inhibited, 67LR phosphorylation increases, accompanied by an increase in filopodia.
CONCLUSIONS AND PERSPECTIVES
Despite over two decades of work, 67LR remains something of a mystery. Clearly, it functions in many parts of the cell in a variety of different unrelated roles. There have been few other proteins which have been proposed to play a role in extracellular signalling, in ribosome assembly and in nuclear processes. The possible fates of 37LRP are summarized in Figure 6. The mystery is compounded by uncertainty about the molecular structure of the protein called ‘67LR’. The cause of the apparent increase in molecular mass remains uncertain. Although evidence has been collected in support of a number of conflicting hypotheses, no firm conclusions can be reached. Consequently, understanding the biochemistry of this molecule is difficult, as the active species has not been properly defined. Furthermore, there has been no experimentally determined structure of 37LRP. Although models can be derived using homologous ribosomal components, the molecules upon which these models are based are not laminin receptors. Presumably, they do not even bind to laminin. Therefore a proper understanding of 67LR–laminin interactions can only be achieved once we know both the molecular composition of the active species along with the three-dimensional structures of both 37LRP and 67LR. Possession of this information would also permit structure-based design of drugs to inhibit or enhance this interaction (or the interaction between 67LR and pathogens).
Possible conversion of 37LRP/67LR within the cell
cytotoxic necrotizing factor 1
dual-specificity MAPK phosphatase
epidermal growth factor
endothelial nitric oxide synthase
glycogen synthase kinase 3
haemopoietic stem cell
heparan sulfate proteoglycan
c-Jun N-terminal protein kinase
- LE domain
laminin-like EGF domain
37 kDa laminin receptor precursor
67 kDa laminin receptor
mitogen-activated protein kinase
an acidic ribosomal protein of molecular mass approx. 40 kDa, identical with 37LRP
protein phosphatase 1
transforming growth factor-β
TGF-β-inhibited membrane-associated protein
Present address: School of Biomedical Sciences, Queen's University Belfast, Medical Biology Centre, 97 Lisburn Road, Belfast BT9 7BL, U.K.
Present address: Agri-Food and Biosciences Institute, 18a Newforge Lane, Belfast BT9 5PX, U.K.
Present address: Biocolor Limited, Valley Business Centre, 67 Church Road, Newtownabbey BT36 7LS, U.K.