Human Molecular Genetics, 2002, Vol. 11, No. 1 1-12
© 2002 Oxford University Press
Disease-associated mutations in L1 CAM interfere with ligand interactions and cell-surface expression
Cambridge Institute for Medical Research and Cambridge University Department of Medicine, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 2XY, UK and 1Max Delbrück Center for Molecular Medicine, Robert-Rössle-Strasse 10, 13092 Berlin, Germany
Received July 16, 2001; Revised and Accepted November 5, 2001.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mutations in the L1CAM gene cause a highly variable neurological disease described as X-linked hydrocephalus, MASA syndrome or spastic paraplegia type I. Over one-third of the mutations identified in affected boys are missense, unique to individual families and distributed primarily across the large extracellular domain of the L1 protein. We have examined the effects of 25 missense mutations on binding to homophilic (L1) and heterophilic (TAX-1) ligands as well as on intracellular trafficking. All but three of these result in reduced ligand binding or impaired movement to the surface of COS and CHO cells. Therefore, we demonstrate for the first time that most missense mutations found in affected families have functional consequences. Furthermore, mutations that are predicted to affect the structure of individual extracellular domains are more likely to affect intracellular processing and/or ligand binding than those mutations affecting surface properties of the molecule.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mutations in the gene that encodes neural cell adhesion molecule (CAM) L1 (L1CAM; OMIM 308840) are responsible for the X-linked recessive disorder described as X-linked hydrocephalus, MASA syndrome or SPG-1 (1–3). A high incidence of neonatal lethality is characteristic of this condition; however, survivors exhibit a wide range of clinical signs including varying degrees of hydrocephalus, mental retardation, lower limb spasticity and flexion–adduction deformities of the thumbs (4,5). In addition to enlarged cerebral ventricles, other brain malformations associated with X-linked hydrocephalus include aberrant development of the corpus callosum and the cortico-spinal-tract (6). Many of these features are also seen in L1 knock-out mouse lines and are clearly consistent with a role for L1 in brain histogenesis (7–10).
L1 is a member of a superfamily of immunoglobulin (Ig)-related CAMs that play pivotal roles in the growth and guidance of particular axon tracts in the developing nervous system (reviewed in 11). L1 is composed of six Ig-like domains and five fibronectin type-III (Fn) repeats in the extracellular region, a single pass transmembrane domain and a short cytoplasmic tail. The tertiary structure of L1 is unresolved, although, it is likely that the first four Ig domains form a horseshoe conglomerate similar to that of the related proteins hemolin (12) and axonin-1 (13). This relies on a seven amino acid hinge between Ig domains 2 and 3 (Ig2 and Ig3) that provides enough flexibility for Ig domain 1 (Ig1) to interact with domain 4 (Ig4) and for Ig2 to interact with Ig3.
L1 influences not only axon growth and guidance but also cell migration (14,15), axon fasciculation (16,17) and myelination (18,19). It may act as a ligand or receptor in the developing nervous system and interacts through both its extracellular and intracellular domains with a variety of different molecules (20–22). The complexity of L1 interactions and their relationship to neuronal biology is only partly understood.
The importance of homophilic binding (L1:L1) during axonal development is indicated by the ability of L1 to act as ligand and receptor to promote neurite outgrowth in vitro (23). The in vivo relevance of heterophilic interactions is largely unknown; however, several recent studies have suggested interactions of L1 with integrins (24,25) and Ig superfamily members such as TAX-1/TAG-1/axonin-1 (26), have roles in neurite outgrowth. L1 also interacts with neurocan, a cell-surface proteoglycan that has an inhibitory effect on neurite outgrowth (27) and LI is part of a receptor complex required for repulsion of cortical neurons in response to the semaphorin, Sema3A (28).
Over 100 different L1 mutations have been reported many of which result in frameshift or nonsense changes that effectively eliminate cell-surface L1 (T.L.M. database: http://dnalab-www.uia.ac.be/dnalab/l1/index.html). Patients with these mutations have been described as having a more severe phenotype than those that truncate the cytoplasmic domain or affect single amino acids (29). Missense mutations provide valuable tools for dissecting L1’s different ligand interactions (reviewed in 11). They account for over one-third of the pathological mutations described and are distributed across 10/11 extracellular domains. It is particularly important to identify a functional effect of these missense mutations to confirm that they are indeed disease causing and not rare non-pathogenic variants.
Missense mutations may influence the function of a protein in several ways. If they induce protein misfolding this may cause defects in intracellular processing, affect presentation of a specific ligand-binding site or disrupt the tertiary structure of the protein. Missense mutations that do not affect secondary protein structure may still influence tertiary inter-domain interactions or disrupt protein–protein interaction sites.
Models of L1 domains have been proposed based on resolved structures of related sequences (30). These highlight a number of ‘key’ residues that are considered necessary for correct domain assembly; mutation of these residues would be expected to alter the fold of a domain. Mutations affecting residues with a surface side chain will have variable results. Some of these may also affect the fold or stability of a domain but others may only affect surface properties.
Examining the effect of missense mutations on L1 function allows us to do three things. First, to confirm these missense mutations are indeed pathogenic. Secondly, to determine if any of the mutations affect specific L1 functions and thereby define functional regions of the protein. Thirdly, to better understand whether a reduction in levels of cell-surface L1, disruption of individual ligand interactions or both, are likely to contribute to patient pathology. Using 12 missense mutations primarily in the Ig domains, our previous study showed that several ectodomains are required for wild-type levels of L1 homophilic binding and for heterophilic binding to axonin-1 and to the related protein F3/F11/contactin (31). This allowed us to suggest a preliminary model for interaction with these ligands. Furthermore, these data suggested that some mutations could specifically affect binding to distinct classes of ligand. In the present study, we have extended our assessment to examine the effects of 25 pathological missense mutations, distributed throughout the L1 extracellular domains, on intracellular trafficking as well as binding to L1 and the human homologue of chick axonin-1, TAX-1.
The results demonstrate that L1 binding to itself or to TAX-1 requires many but not all L1 domains. All missense mutations predicted to distort the structure of individual domains affect the intracellular processing of the L1 protein and reduce cell-surface expression. Mutations that only affect ligand binding exclusively affect residues with surface side chains and highlight the importance of L1 ligand interactions during development. Our study shows that most L1 mutations associated with disease do have a functional consequence. Furthermore, mortality is increased for patients that have structural versus surface mutations reflecting their multiple effects on L1 trafficking and ligand binding.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
In an earlier study we examined the effects on ligand binding of 12 pathological missense mutations of L1 located primarily in the Ig domains (31). We have now extended this analysis to include 13 additional mutations, the majority of which are located in the Fn domains. We analysed the effect of these new mutations on L1 homophilic binding and studied the effects of all 25 mutations on heterophilic binding to human TAX-1 and on intracellular protein trafficking (Table 1).
|
Missense mutations from several domains disrupt homophilic binding
Our previous study indicated that L1 homophilic binding involves at least the first five Ig domains of L1 (31). Reduced binding seen for surface mutation H210Q also suggested a specific role for Ig2 in L1:L1 contact. However, the analysis of a single mutation in Fn domain 2 (Fn2) (V768F) also implicated membrane-proximal domains in L1:L1 binding. Thirteen additional pathological missense mutations, primarily affecting Fn domains, have now been assessed for their effect on homophilic binding. Fluorescent beads were coated with Fc-chimeric protein containing the extracellular region of L1, with or without amino acid substitutions. Beads were allowed to adhere through homophilic binding between L1 molecules and the proportion of clustered versus single beads was measured by fluorescence-activated cell sorter (FACS) analysis. The results are shown alongside those obtained for the previous 12 (asterisks in Fig. 1A and Table 1). The 25 mutations have been grouped in Table 1 according to whether they are predicted to affect the structure of the protein (Group A) or surface properties (Group B) based on the modelling studies of Bateman et al. (30).
|
Five additional Ig domain mutations (I219T, G370R, L391P, D544N and S542P) reduce homophilic binding. In Ig2, I219T is a highly conserved residue in rodent (PIR-ASDB accession nos X12875 and X59149), chick (Z75013), fugu (Z71926), goldfish (U55211) and zebrafish (X89204) at the beginning of a loop in the same ß-sheet as the surface mutation H210Q, previously found to reduce homophilic binding. This supports the contention that this region of domain 2 houses a contact site between L1 molecules. In Ig4, G370R and L391P affect key residues and dramatically reduce binding. Gly370 lies in a putative intra-molecular contact site between Ig1 and Ig4 in the horseshoe structure predicted from comparison with hemolin (12). A similar reduction was previously found for Ig4, key residue mutation P333R. More modest effects on binding were observed for two Ig domain 6 (Ig6) mutations S542P and D544N. Introduction of a proline at S542 would disrupt the hydrogen bonding of a ß-sheet and destabilize this domain. D544 is a surface site that is highly conserved between L1 orthologues in a loop region of uncertain structure. These are the first examples of Ig6 mutations that significantly reduce L1:L1 binding and these data suggest a role for this domain in homophilic interaction.
Fn domain mutations had more variable effects on binding. In Fn domain 1 (Fn1), surface mutation K655E, does not affect homophilic binding. In contrast, G698R, a key residue mutation, reduces homophilic binding. In the second Fn domain (Fn2), V752M and Y784C disrupt homophilic binding. Combined with result from V768F, this finding supports the hypothesis that the integrity of Fn2 is required for homophilic binding. All three of these are key residue mutations. The only surface mutation in Fn2, M741T, does not reduce binding and therefore this residue is unlikely to be part of an L1-binding site. In Fn domain 4 (Fn4), mutations L935P and P941L affect surface residues. However, mutating L935 to a proline would interrupt the hydrogen bonding of the ß-sheet, destabilizing this domain. Residue P941 is absolutely conserved in L1 homologues and at the beginning of an inter-strand loop of a ß-sheet. Changing the size and shape of this surface residue may disrupt local loop structure. Neither mutation affects homophilic binding. Previously, we have shown that Y1070C, a mutation affecting a surface residue in Fn domain 5 (Fn5), does not affect binding to L1. Here we show that even a key residue mutation in this domain, W1036L, has no effect. Therefore, mutations in Fn4 and Fn5, regardless of their affect on the structural integrity of the protein, do not disrupt homophilic binding.
Heterophilic binding is also affected by mutations in many L1 domains
To examine the interaction of L1 mutant proteins with TAX-1, the human orthologue of axonin-1, a two-channel FACS detection system was used to quantify mixed aggregates of L1–Fc (red) and TAX-1–Fc (green) coated beads (Materials and Methods). Twelve of the mutations have previously been examined for their effects on interaction with axonin-1 using a cell-based assay system (31). These gave identical binding profiles with TAX-1 with the exception of D598N, which showed reduced binding to axonin-1 but not TAX-1. Hence, although binding requirements are quite conserved across evolution there may be some differences between chick and human ligands. Therefore, human TAX-1 protein was used for examining all 25 mutations in this study (Fig. 1B; Table 1).
Overall the results are remarkably similar to those obtained for homophilic binding with mutations in Ig domains 1–6 and Fn1 and Fn2 affecting the L1:TAX-1 interaction. As observed for L1, more mutations in the membrane-distal domains are deleterious and, for the reasons outlined above, Fn4 and Fn5 are not required for L1/TAX-1 binding. Indeed destabilizing Fn5 may even improve binding. However, there are some notable differences in the effects of mutations on heterophilic interaction and these involve surface changes. For example, E309K, a mutation of Ig3 that would alter a surface charge on a ß-sheet, reduces binding to TAX-1 although it does not affect homophilic binding. This was also noted using chick axonin-1 as the L1 ligand in a cell-based assay system (31). This strongly suggests that this region of the domain may be part of a specific binding site. In contrast, mutation H210Q, affecting the surface of a ß-sheet in Ig2, increases binding to TAX-1. Again, this effect had been observed for interaction with the chick protein using the cell-based assay system. Thus, the increase in binding does not depend on the assay used or whether the ligand is from chick or man. Interestingly, a mutation only nine residues away, I219T, dramatically reduces both TAX-1 and L1 binding.
Domain deletion studies support missense mutation findings
The histograms in Figure 1 show that mutations of the membrane-distal Ig domains of L1 have profound effects on binding to L1 or TAX-1. Mutations in the more membrane-proximal domains are generally less deleterious, but the dramatic effects of some mutations in Fn1 and Fn2 suggest that the Fn domains may also be involved in binding. To investigate this further, we examined the effects of selectively deleting regions of the L1 protein on ligand binding (Fig. 2). Deletion of Fn1 and Fn2 (
F1,2), Fn3 (F3), Fn4 and Fn5 (F4,5) or indeed 1–5 (6Ig) does not reduce binding to TAX-1 (Fig. 2B). However, deletion of Fn1 alone (F1) has a modest effect on binding and, consistent with data obtained using mutations, deletion of Fn2 alone (F2) significantly reduces binding. Similar effects are seen when these constructs are used for homophilic binding assays (Fig. 2A) although in this case a modest reduction in binding is seen when the more distal Fn domains are missing. However, Fn4 and Fn5 are dispensable. For both ligands, binding of Ig domains 1–4 (4Ig) is at least as efficient as Ig domains 1–6 (6Ig) indicating that the primary-binding site is contained within the Ig domains. That Ig domains 1–3 (3Ig) are insufficient for binding (Fig. 2) demonstrates that at least the first four domains are necessary. This region is capable of adopting a horseshoe structure but whether binding requires this conformation or an extended string of domains is not known. To investigate this we deleted five out of seven amino acids of the hinge region between Ig2 and Ig3 hinge. This should not interfere with folding of neighbouring Ig domains but would reduce flexibility. This protein retains <20% binding to both ligands demonstrating that L1 interactions with both itself and TAX-1 occur with the first four domains folded into a horseshoe conformation.
|
Several L1 missense mutations affect cell-surface expression
Missense mutations are valuable tools for understanding how individual domains contribute to homophilic and heterophilic binding. However, many mutations, especially those affecting structural residues, may cause disease as a result of defects in intracellular processing. To examine this, full-length L1 cDNA clones containing each of the 25 mutations were transiently transfected into CHO cells. As a control we included signal peptide mutation W9S as this would be expected to inhibit trafficking of L1 to the endoplasmic reticulum (ER). Cell-surface expression was assessed using digital image analysis. Many mutations were found to reduce expression of cell-surface L1 (Fig. 3). Those expected to affect domain structure (Group A, Table 1) result in a reduction of cell-surface expression to below 75% of wild-type. However, the amount of protein expressed on the cell surface is variable, from between 12 and 72% of wild-type levels. In contrast, only four out of the 11 mutations that affect surface properties of L1 show a reduction in levels of cell-surface protein to below 75% (Group B, Table 1) and in fact for H210Q an increase in levels of cell-surface expression was detected. Group B mutations that reduce cell-surface expression do so to between 9 and 50% of wild-type levels.
|
We then examined the intracellular fate of mutant proteins by transiently transfecting constructs into COS-7 cells, which are a large, morphologically appropriate cells for examination of intracellular processing defects. Cells were then fixed, permeabilized and double labelled with anti-L1 and anti-KDEL antibodies. KDEL is an ER retention signal only expressed on ER proteins and hence is a marker for this organelle (32). Protein localization was monitored with a confocal microscope by scanning through planar images of individual cells. Several hundred cells were examined in at least three independent experiments (Table 1). Of the 26 mutant proteins investigated (including W9S), eight showed a wild-type (WT) pattern of cell-surface expression. Seven mutant proteins produced the moderate (M) phenotype and 11 mutant proteins were clearly retained within the cell with little cell-surface staining (R).
Mutant proteins with a WT expression pattern primarily expressed cell-surface L1, with marked expression at the periphery of the imaged planes. As expected some L1 protein was also detected inside the cell. In the second category (M), clear cell-surface expression of L1 was evident, but cells also showed increased intracellular accumulation, noticeably in the ER, as detected by co-localization with the KDEL marker. Finally, in the third category of expression (R), mutant proteins were primarily retained within the cell. Frequently, this protein co-localized with the ER marker but in some cells intense regions of juxtanuclear and cytoplasmic L1 staining were observed. Composite images for typical examples of mutated protein displaying grossly R, M or WT trafficking are shown in Figure 4.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Homophilic and heterophilic binding
We have now conducted homophilic (L1) and heterophilic (TAX-1) binding studies for L1 containing 25 pathological mutations and complemented this with specific domain deletions in order to refine models of L1 interaction (31). Mutations distributed across Ig domains 1–6 and Fn1 and Fn2 have now been found to disrupt both L1:L1 and L1:TAX-1 binding suggesting that all of these domains are required. For the Ig domains, mutations that would modify surface properties as well as those that would disrupt domain structure have been found to affect binding indicating that important ligand or intra-molecular contact sites are affected. In particular, the ß-sheet that contains surface residue H210 in Ig2 may have a role in homophilic interaction, whereas, that containing E309 in Ig3 is important for interaction with TAX-1. For the Fn domains, only structural mutations have so far been found to affect binding to either ligand. Domain deletion studies support the contention that wild-type binding requires Ig and Fn domains but indicates that the requirement for Fn domains may not be domain specific. This is highlighted by the result that deletion of Fn2 severely reduces binding to both ligands but deletion of Fn1 and Fn2 together rescues binding. Therefore, Fn2 may provide tertiary structural stability. A similar observation was reported by Kunz et al. (17) for NgCAM, the chick homologue of L1. They suggested that altered domain folding is likely to account for the loss of binding when Fn2 is deleted. Fn4 and Fn5 are clearly not required for either L1 or TAX-1 binding.
It has been suggested that formation of the four Ig domain horseshoe conglomerate of hemolin (12) and axonin-1 (13) is also likely to occur in L1. Su et al. (12) have also suggested that this structure may open to allow the first four domains of L1 to interact in an anti-parallel fashion. Indeed, a recent study visualized L1 in the horseshoe and open conformations using rotary shadowing, sedimentation and negative staining techniques. However, the question still remained as to which conformation L1 adopts for ligand binding (33). Our data indicate that the first four Ig domains interact with ligand in the closed, horseshoe conformation as significant reduction in binding occurs when the hinge region is deleted.
Studies by Haspel et al. (34) suggest that Ig domains 1–4 are sufficient for homophilic binding but Ig domains 1–6 are required for neurite outgrowth. Our results support the hypothesis that the distal Ig domains including the 4Ig horseshoe contain the major binding sites but indicate some additional roles for other domains.
Trafficking
Eighteen out of 26 missense mutations in L1 affect trafficking of L1 to the cell surface in two different in vitro mammalian expression systems. Furthermore, quantitative reduction of cell-surface expression in CHO cells was reflected by evidence of intracellular accumulation of L1 in COS cells, with only one exception. For mutation E309K, a reduction in cell-surface expression was seen for CHO cells but no intracellular protein accumulation was evident in COS cells. It is possible that this is due to different processing requirements in the two cell types or that the reduction in cell-surface expression is due to lack of mRNA or protein stability. The degree to which cell-surface expression is affected is highly variable and to an extent this is related to the type of mutation. All mutations predicted to affect the structure of a domain (Group A, Table 1) result in reduced expression on the surface of CHO cells and show clear intracellular retention of protein in COS cells. In contrast only three out of 11 of those mutations affecting surface properties of L1 or loop structures (Group B, Table 1) show trafficking defects in both cell types. Although it is not possible to extrapolate from these in vitro data to determine levels of expression on developing neurons in vivo, it is likely that the amount of neuronal surface L1 is also reduced in patients with these mutations. For some mutations, neuronal trafficking machinery may apply more stringent controls, further reducing cell-surface expression of severely misfolded proteins.
Several studies have investigated the cellular mechanisms used to eliminate misfolded proteins (reviewed in 35). During normal protein synthesis, the production of misfolded proteins or incompletely assembled proteins is a common occurrence. Overlapping quality control mechanisms ensure that misfolded proteins are retained in the ER, or retrieved to the ER from other compartments and then dislocated through an ER membrane channel for degradation by cytosolic proteosomes. However, as a result of mutation, gross accumulation in the ER can occur. This may be the result of inappropriate disulfide bond formation due to the exposure of free cysteine residues or due to the exposure of hydrophobic regions that interact to form aggregates. Interestingly, Y1070C is an example of a surface residue mutation that may be retained in the ER due to the presence of a free cysteine residue. In addition, some dislocated intermediates that are highly prone to aggregation when exposed to the cytosolic environment may escape degradation by proteosomes and form cytosolic aggregates such as aggresomes (36). Here we show that although misrouted L1 protein is often co-localized with the ER, accumulation in juxtanuclear regions and the cytosol was also observed in some cells. In comparison with other reports, it is also likely that misfolded mutant proteins are also localized within compartments of the degradation pathway.
We conclude that trafficking defects are likely to contribute to the disease phenotype. This has been suggested previously for a few missense mutations in L1. We showed previously that W9S, a mutation in the signal peptide region of the protein, significantly reduces cell-surface expression in COS cells (31). Moulding et al. (37) reported that two extracellular domain missense mutations, R184Q and D598N, reduced cell-surface expression in three cell types, including COS cells. However, our quantitative analysis in CHO cells as well as immunofluorescent detection in COS cells suggests that D598N does not inhibit L1 surface expression.
Defects in intracellular trafficking of mutant proteins have been shown to be pathological for several genetic disorders, for example cystic fibrosis (38), Charcot–Marie tooth disease (39) and Menkes disease (40). We have now shown that the majority of missense mutations in L1 may cause trafficking problems in vitro adding further weight to the importance of this mechanism in disease pathology.
How much do trafficking problems and ligand-binding effects contribute to patient pathology?
We show that the majority of L1 missense mutations affect binding to at least one of two ligands or disrupt protein processing in vitro. However, which of these effects is more important is unclear. In vivo, defects in intracellular processing of some misfolded proteins may not significantly reduce cell-surface protein below functional levels.
Our work suggests that disruption of L1 ligand interactions may also contribute to the disease pathology as mutations such as H210Q, I219T and A426D have severe effects on ligand binding without affecting intracellular processing. Other studies have also used pathological mutations to identify ligand interactions that may be disrupted during the disease process. A study by Needham et al. (41) showed that certain pathological missense mutations in the cytoplasmic domain of L1 reduce the ability of L1 to recruit the cytoskeletal protein ankyrin. In addition, mutations in the first Ig domain of the extracellular region of the protein have also been shown to disrupt the interaction with the ECM protein neurocan (42). We have shown that for one of these mutations, L120V, neither intracellular trafficking nor ligand binding to TAX-1 (present study), F3/contactin or L1 (31) is affected.
Relating mutations to phenotype
Until an effect on a ligand interaction or on protein processing can be attributed to each missense mutation, the question of whether the mutation is indeed pathological remains. Certainly, the lack of detection of a mutation in a series of healthy subjects cannot be taken as proof of causation. For example, L1 mutation V768I was identified in a family with X-linked hydrocephalus, but did not segregate with disease (43). Our study has revealed two mutations, K655E and M741T, that do not affect any of the functions of L1 investigated to date. Mutation M741T arises in a sporadic case that may not have the X-linked form of disease, whereas, K655E was found to segregate with disease in a large X-linked pedigree (3,43,44). Of course these mutations may primarily disrupt the function of one of the many L1 ligand interactions that have not yet been assessed.
Now that such a large number of mutations have been assessed for several downstream effects a reasonable assessment of mutation type versus disease severity can be made. In Table 1, missense mutations are classified into those expected to disrupt domain integrity through affecting a key residue or secondary structures such as ß-sheets (Group A) and those expected to affect surface properties or loop structures (Group B). Interestingly, 45% of patients with Group A mutations die within 1 year compared with only 13% of patients with Group B changes. Our study shows that the majority of surface residue mutations affect either ligand binding or protein trafficking but not both. In contrast, Group A mutations are more likely to disrupt protein architecture and as a result affect both ligand binding and intracellular processing. Therefore, the higher morbidity associated with structural changes may, at least in part, be a reflection of their effects on trafficking, ligand binding or, more likely, a combination of these effects.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Mutagenesis
L1 cDNAs containing individual missense mutations were constructed by PCR directed in vitro mutagenesis with a method modified from Ke and Madison (45). The first round reaction used 30 cycles of PCR on a wild-type pBS-L1 cDNA template (a gift from J.Hemperley) with BIO-X-ACTTM (Bioline) proof-reading polymerase, a mutagenic primer and flanking L1 primer to produce a megaprimer. An aliquot of this megaprimer was then used with a third long-flanking primer in the second round of PCR to produce an L1 fragment containing the mutation flanked by two unique restriction enzyme sites. The fragment was then cut with the appropriate restriction enzymes (Table 2) and used to replace the same region in an L1, wild-type clone that had previously been manipulated into appropriate vectors (pIG and pcDNA3) as described by De Angelis et al. (31). Clones were sequenced across the mutated cassettes using the PrismTM dye terminator kit (Perkin Elmer) and an ABI 377 semi-automated DNA sequencer. The same mutagenic strategy was used to create the domain deletion constructs but in this case a first round ‘loop out’ primer was designed and used in conjunction with an L1 flanking primer to create a megaprimer deleted for cDNA encoding one or more domains of L1. As before, an aliquot of the megaprimer was then used in conjunction with a third long-flanking primer to create an L1 fragment lacking the individual domains; this was used to replace the same region in the L1 wild-type clone (Table 3).
|
|
The L1–Fc Ig domain 1–4 clone was constructed using 35 cycles of PCR on the wild-type pBS-L1 cDNA template with proof-reading polymerase and primers designed to introduce HindIII and an EcoRV restriction sites at the 5' and 3' end of the product, respectively (Table 3). The 1292 bp product was cut with the appropriate enzymes, cloned into the pIgplus vector and sequenced. The TAX–Fc clone was constructed by engineering the extracellular portion of human TAX-1 cDNA into pIgplus (Invitrogen). Primers TH3F (5'-CTGGACTTTCTCAAGCTCTAGTC-3') and THXE (5-GCCATGATATCCACCATCATTCTAGAGCCTCCATTCCT-3') were used to amplify a 252 bp PCR product from a human full-length TAX-1 clone (a gift from P.Sonderegger, Switzerland). This introduced an Xba1 restriction enzyme site immediately prior to the GPI linkage sequence. This fragment was cut with BstB1 and XbaI and used in a three-way ligation with a 2.8 kb HindIII–BstBI fragment of TAX-1 cDNA corresponding to the upstream coding sequence and HindIII–XbaI-cut pIgplus. The BstBI–XbaI cassette and cloning sites were sequenced. The L1–Fc construct was described by De Angelis et al. (31). Constructs for 3Ig, 6Ig and
F3 were a gift from Professor P.Doherty (London).
Protein production and purification
Constructs containing sequences encoding the extracellular portions of human L1 or human TAX-1 (the orthologue of chick axonin-1 and mouse TAG-1) in conjunction with sequences encoding human IgG Fc domains were used to produce soluble chimeric proteins as described by De Angelis et al. (31). Briefly, COS-7 cells were transiently transfected with 10 µg DNA per 150 mm culture dish, the soluble Fc chimeric protein was allowed to accumulate in the media for 6 days and was recovered and purified by protein A Sepharose affinity chromatography.
Homophilic adhesion assay
The homophilic-binding assay was conducted according to the method described by De Angelis et al. (31). Briefly, Bioclean fluorescent microspheres (red, 0.6 µm; Duke Scientific Corp.) were pre-coated with anti-human IgG antibody (Fc specific). A 2.5 µg aliquot of L1 wild-type, mutant or domain deleted Fc-protein was conjugated to 10 µl of antibody-coated beads by incubating for 2 h at 37°C. Excess unbound protein was removed by washing with PBS/5% FCS. For the homophilic-binding assay, L1–Fc-coated beads were disaggregated to a single bead suspension, then homophilic aggregation was allowed to occur by incubating the single bead suspension at 37°C with samples removed in duplicate over a 30 min time course. Samples were diluted 1:5000 times in ice-cold PBS and analysed using a Becton Dickinson FACSort. Ten thousand particles were sampled for each time point, the number of particles with single bead fluorescence levels was measured in comparison to the number of particles with multiple bead fluorescence levels (clustered particles). Each Fc-protein was assayed at least three times in parallel with wild-type to control for variation between experiments and standardized to a wild-type-binding curve.
Heterophilic adhesion assay
A two-colour heterophilic adhesion assay was developed using red and green fluorescent microsphere beads (red and green, 0.6 µm). Microspheres were coated with anti-human IgG antibody (Fc specific) and conjugated with L1 or TAX-1 Fc chimeric protein as described above. Briefly, for each L1–Fc mutant or wild-type protein chimera sample, 10 µl of antibody-conjugated red microsphere beads were incubated with 2.5 µg of wild-type or mutant L1–Fc protein for 2 h at 37°C. In parallel, TAX-1–Fc protein was captured onto green microsphere beads. Excess unbound protein was removed by washing twice with PBS/5% FCS and the beads were resuspended in 60 µl PBS/5% FCS. The protein-conjugated microspheres were disaggregated to produce single bead suspensions by trituration. A 1:1 mixture of red (L1) and green (TAX-1) microsphere beads was prepared and further disaggregated by sonicating in an iced bath for 30 min. To allow heterophilic aggregation to occur, the microsphere bead mixture was incubated at 37°C with samples removed in duplicate over a 30 min time course and diluted 1:5000 into ice-cold PBS. A FACS analysis was used to assess the proportion of two-colour aggregates in the samples. Ten thousand particles were counted per sample and categorized as homophilic red only (L1:L1), homophilic green only (TAX-1:TAX-1) or mixed aggregates. Every mutant protein was assayed at least three separate times and standardized to a wild-type control always conducted in parallel. As a control to ensure that equivalent concentrations of the Fc chimeric proteins were captured by the antibody-coated beads, 30 µl samples of coated beads were denatured at 95°C in sample buffer, the released Fc chimeric proteins were subjected to SDS–PAGE and shown to be present in equal proportions. No unbound Fc protein was detected under assay conditions as assessed by SDS–PAGE.
Transfection and immunofluorescence on eight-sample multi-test slides
A 12.5 µl aliquot of a 1:10 dilution of COS-7 cells from a confluent T25 flask was added to each spot on an eight-well multi-test slide (ICN) and incubated overnight at 37°C in a humidified incubator. For each spot, 0.75 µl Fugene (Roche) was added to 10 µl Dulbecco’s modified Eagle’s medium (DMEM; Sigma D-5671) without serum and antibiotics and incubated for 5 mins at room temperature (RT). This was then added to 0.2 µg DNA (prepared using QIAprep spin miniprep kit) and incubated for 15 mins at RT. The cells were washed once with DMEM and the fugene/DNA mix was added to cells and incubated at 37°C for 3 h. The transfection mix was removed from the cells, replaced with 25 µl DMEM supplemented with 10% FCS, 200 mM L-glutamine, 100 U/ml penicillin and 100 mg/ml streptomycin and incubated overnight at 37°C prior to fixation and immunofluorescent labelling. The cells were fixed for 30 min with 20 µl 4% paraformaldehyde, permeabilized with 10 µl 0.1% Triton X-100 for 5 min, then blocked with PBS/3% BSA for 3 h at RT. The cells were double labelled with the primary antibodies rabbit anti-human L1 polyclonal antibody (1:600) and rat anti-KDEL monoclonal antibody (1:200; a gift from Sean Munro, Cambridge), for 1 h followed by anti-rabbit Texas Red (1:200) and anti-rat FITC (1:40) secondary antibodies, respectively, for 1 h. Coverslips were mounted on to the slides with immunomount (ICN) and fluorescence was viewed with a 63x objective on a Leica confocal microscope with TCS NT software.
Quantification of wild-type and mutant L1 cell-surface expression in CHO cells
CHO cells were transiently transfected with wild-type or mutant expression plasmids. Cell-surface expression was monitored by immunofluorescence analysis with L1-specific polyclonal antibodies, followed by quantification with digital image analysis as outlined previously (46). To identify the sub-population of transfected cells, the GFP-expressing plasmid pEGFP (Clontech) was co-transfected with L1 plasmids. Cell-surface expression of L1 mutants was quantified as the percentage of GFP-positive cells that have detectable L1 surface expression and values were normalized with respect to wild-type L1 expression for every slide. Several hundred GFP-positive cells were evaluated for each mutant in each experiment and, depending on the clone, between three and eight independent experiments were performed.
|
ACKNOWLEDGEMENTS |
|---|
We would like to thank Fritz G.Rathjen for helpful discussions and generous support and Dieter Jobsky for excellent technical assistance. This work was funded by the Wellcome Trust to S.K. and partly supported by Deutsche Forschungsgemeinschaft grant Br 1217/4 to T.B.
|
FOOTNOTES |
|---|
+ To whom correspondence should be addressed. Tel: +44 1223 762616; Fax: +44 1223 331206; Email: sjk12@mole.bio.cam.ac.uk
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
1 Rosenthal,A., Jouet,M. and Kenwrick,S. (1992) Aberrant splicing of neural cell adhesion molecule L1 mRNA in a family with X-linked hydrocephalus. Nat. Genet., 2, 107–112.[ISI][Medline]
2 Jouet,M., Rosenthal,A., Armstrong,G., MacFarlane,J., Stevenson,R., Paterson,J., Metzenberg,A., Ionasescu,V., Temple,K. and Kenwrick,S. (1994) X-linked spastic paraplegia (SPG1), MASA syndrome and X-linked hydrocephalus result from mutations in the L1 gene. Nat. Genet., 7, 402–407.[ISI][Medline]
3 Vits,L., Van Camp,G., Coucke,P., Fransen,E., De Boulle,K., Reyniers,E., Korn,B., Poustka,A., Wilson,G., Schrander-Stumpel,C. et al. (1994) MASA syndrome is due to mutations in the neural cell adhesion gene L1CAM. Nat. Genet., 7, 408–413.[ISI][Medline]
4 Willems,P.J., Brouwer,O.F., Dijkstra,I. and Wilmink,J. (1987) X-linked hydrocephalus. Am. J. Med. Genet., 27, 921–928.[ISI][Medline]
5 Kenwrick,S., Jouet,M. and Donnai,D. (1996) X linked hydrocephalus and MASA syndrome. J. Med. Genet., 33, 59–65.[Abstract]
6 Yamasaki,M., Arita,N., Hiraga,S., Izumoto,S., Morimoto,K., Nakatani,S., Fujitani,K., Sato,N. and Hayakawa,T. (1995) A clinical and neuroradiological study of X-linked hydrocephalus in Japan. J. Neurosurg., 83, 50–55.[ISI][Medline]
7 Cohen,N.R., Taylor,J.S., Scott,L.B., Guillery,R.W., Soriano,P. and Furley,A.J. (1998) Errors in corticospinal axon guidance in mice lacking the neural cell adhesion molecule L1. Curr. Biol., 8, 26–33.[ISI][Medline]
8 Dahme,M., Bartsch,U., Martini,R., Anliker,B., Schachner,M. and Mantei,N. (1997) Disruption of the mouse L1 gene leads to malformations of the nervous system. Nat. Genet., 17, 346–349.[ISI][Medline]
9 Demyanenko,G.P., Tsai,A.Y. and Maness,P.F. (1999) Abnormalities in neuronal process extension, hippocampal development, and the ventricular system of L1 knockout mice. J. Neurosci., 19, 4907–4920.
10 Fransen,E., D’Hooge,R., Van Camp,G., Verhoye,M., Sijbers,J., Reyniers,E., Soriano,P., Kamiguchi,H., Willemsen,R., Koekkoek,S.K. et al. (1998) L1 knockout mice show dilated ventricles, vermis hypoplasia and impaired exploration patterns. Hum. Mol. Genet., 7, 999–1009.
11 Kenwrick,S., Watkins,A. and De Angelis,E. (2000) Neural cell recognition molecule L1: relating biological complexity to human disease mutations. Hum. Mol. Genet., 9, 879–886.
12 Su,X.D., Gastinel,L.N., Vaughn,D.E., Faye,I., Poon,P. and Bjorkman,P.J. (1998) Crystal structure of hemolin: a horseshoe shape with implications for homophilic adhesion. Science, 281, 991–995.
13 Freigang,J., Proba,K., Leder,L., Diederichs,K., Sonderegger,P. and Welte,W. (2000) The crystal structure of the ligand binding module of axonin-1/TAG-1 suggests a zipper mechanism for neural cell adhesion. Cell, 101, 425–433.[ISI][Medline]
14 Lindner,J., Rathjen,F.G. and Schachner,M. (1983) L1 mono-and polyclonal antibodies modify cell migration in early postnatal mouse cerebellum. Nature, 305, 427–430.[Medline]
15 Asou,H., Miura,M., Kobayashi,M. and Uyemura,K. (1992) The cell adhesion molecule L1 has a specific role in neural cell migration. Neuroreport, 3, 481–484.[ISI][Medline]
16 Honig,M.G., Petersen,G.G., Rutishauser,U.S. and Camilli,S.J. (1998) In vitro studies of growth cone behavior support a role for fasciculation mediated by cell adhesion molecules in sensory axon guidance during development. Dev. Biol., 204, 317–326.[ISI][Medline]
17 Kunz,S., Spirig,M., Ginsburg,C., Buchstaller,A., Berger,P., Lanz,R., Rader,C., Vogt,L., Kunz,B. and Sonderegger,P. (1998) Neurite fasciculation mediated by complexes of axonin-1 and Ng cell adhesion molecule. J. Cell Biol., 143, 1673–1690.
18 Wood,P.M., Schachner,M. and Bunge,R.P. (1990) Inhibition of Schwann cell myelination in vitro by antibody to the L1 adhesion molecule. J. Neurosci., 10, 3635–3645.[Abstract]
19 Haney,C.A., Sahenk,Z., Li,C., Lemmon,V.P., Roder,J. and Trapp,B.D. (1999) Heterophilic binding of L1 on unmyelinated sensory axons mediates Schwann cell adhesion and is required for axonal survival. J. Cell Biol., 146, 1173–1184.
20 Brümmendorf,T. and Rathjen,F.G. (1996) Structure/function relationships of axon-associated adhesion receptors of the immunoglobulin superfamily. Curr. Opin. Neurobiol., 6, 584–593.[ISI][Medline]
21 Kamiguchi,H. and Lemmon,V. (1997) Neural cell adhesion molecule L1: signaling pathways and growth cone motility. J. Neurosci. Res., 49, 1–8.[ISI][Medline]
22 Kenwrick,S. and Doherty,P. (1998) Neural cell adhesion molecule L1: relating disease to function. Bioessays, 20, 668–675.[ISI][Medline]
23 Lemmon,V., Farr,K.L. and Lagenaur,C. (1989) L1-mediated axon outgrowth occurs via a homophilic binding mechanism. Neuron, 2, 1597–603.[ISI][Medline]
24 Felding-Habermann,B., Silletti,S., Mei,F., Siu,C.H., Yip,P.M., Brooks,P.C., Cheresh,D.A., O’Toole,T.E., Ginsberg,M.H. and Montgomery,A.M. (1997) A single immunoglobulin-like domain of the human neural cell adhesion molecule L1 supports adhesion by multiple vascular and platelet integrins. J. Cell Biol., 139, 1567–1581.
25 Yip,P.M., Zhao,X., Montgomery,A.M. and Siu,C.H. (1998) The Arg–Gly–Asp motif in the cell adhesion molecule L1 promotes neurite outgrowth via interaction with the alphavbeta3 integrin. Mol. Biol. Cell, 9, 277–290.
26 Kuhn,T.B., Stoeckli,E.T., Condrau,M.A., Rathjen,F.G. and Sonderegger,P. (1991) Neurite outgrowth on immobilized axonin-1 is mediated by a heterophilic interaction with L1(G4). J. Cell Biol., 115, 1113–1126.
27 Friedlander,D.R., Milev,P., Karthikeyan,L., Margolis,R.K., Margolis,R.U. and Grumet,M. (1994) The neuronal chondroitin sulfate proteoglycan neurocan binds to the neural cell adhesion molecules Ng-CAM/L1/NILE and N-CAM, and inhibits neuronal adhesion and neurite outgrowth. J. Cell Biol., 125, 669–680.
28 Castellani,V., Chedotal,A., Schachner,M., Faivre-Sarrailh,C. and Rougon,G. (2000) Analysis of the L1-deficient mouse phenotype reveals cross-talk between Sema3A and L1 signaling pathways in axonal guidance. Neuron, 27, 237–249.[ISI][Medline]
29 Yamasaki,M., Thompson,P. and Lemmon,V. (1997) CRASH syndrome: mutations in L1CAM correlate with severity of the disease. Neuropediatrics, 28, 175–178.[ISI][Medline]
30 Bateman,A., Jouet,M., MacFarlane,J., Du,J.S., Kenwrick,S. and Chothia,C. (1996) Outline structure of the human L1 cell adhesion molecule and the sites where mutations cause neurological disorders. EMBO J., 15, 6050–6059.[ISI][Medline]
31 De Angelis,E., MacFarlane,J., Du,J.S., Yeo,G., Hicks,R., Rathjen,F.G., Kenwrick,S. and Brummendorf,T. (1999) Pathological missense mutations of neural cell adhesion molecule L1 affect homophilic and heterophilic binding activities. EMBO J., 18, 4744–4753.[ISI][Medline]
32 Munro,S. and Pelham,H.R. (1987) A C-terminal signal prevents secretion of luminal ER proteins. Cell, 48, 899–907.[ISI][Medline]
33 Schurmann,G., Haspel,J., Grumet,M. and Erickson,H.P. (2001) Cell adhesion molecule l1 in folded (horseshoe) and extended conformations. Mol. Biol. Cell, 12, 1765–1773.
34 Haspel,J., Friedlander,D.R., Ivgy-May,N., Chickramane,S., Roonprapunt,C., Chen,S., Schachner,M. and Grumet,M. (2000) Critical and optimal Ig domains for promotion of neurite outgrowth by L1/Ng-CAM. J. Neurobiol., 42, 287–302.[ISI][Medline]
35 Ellgaard,L., Molinari,M. and Helenius,A. (1999) Setting the standards: quality control in the secretory pathway. Science, 286, 1882–1888.
36 Johnston,J.A., Ward,C.L. and Kopito,R.R. (1998) Aggresomes: a cellular response to misfolded proteins. J. Cell Biol., 143, 1883–1898.
37 Moulding,H.D., Martuza,R.L. and Rabkin,S.D. (2000) Clinical mutations in the L1 neural cell adhesion molecule affect cell-surface expression. J. Neurosci., 20, 5696–5702.
38 Cheng,S.H., Gregory,R.J., Marshall,J., Paul,S., Souza,D.W., White,G.A., O’Riordan,C.R. and Smith,A.E. (1990) Defective intracellular transport and processing of CFTR is the molecular basis of most cystic fibrosis. Cell, 63, 827–834.[ISI][Medline]
39 Deschenes,S.M., Walcott,J.L., Wexler,T.L., Scherer,S.S. and Fischbeck,K.H. (1997) Altered trafficking of mutant connexin32. J. Neurosci., 17, 9077–9084.
40 Petris,M.J., Mercer,J.F., Culvenor,J.G., Lockhart,P., Gleeson,P.A. and Camakaris,J. (1996) Ligand-regulated transport of the Menkes copper P-type ATPase efflux pump from the Golgi apparatus to the plasma membrane: a novel mechanism of regulated trafficking. EMBO J., 15, 6084–6095.[ISI][Medline]
41 Needham,L.K., Thelen,K. and Maness,P.F. (2001) Cytoplasmic domain mutations of the L1 cell adhesion molecule reduce L1-ankyrin interactions. J. Neurosci., 21, 1490–1500.
42 Oleszewski,M., Gutwein,P., von Der Lieth,W., Rauch,U. and Altevogt,P. (2000) Characterization of the L1-neurocan-binding site. Implications for L1–L1 homophilic binding. J. Biol. Chem., 275, 34478–34485.
43 Gu,S.M., Orth,U., Zankl,M., Schroder,J. and Gal,A. (1997) Molecular analysis of the L1CAM gene in patients with X-linked hydrocephalus demonstrates eight novel mutations and suggests non-allelic heterogeneity of the trait. Am. J. Med. Genet., 71, 336–340.[ISI][Medline]
44 Izumoto,S., Yamasaki,M., Arita,N., Hiraga,S., Ohnishi,T., Fujitani,K., Sakoda,S. and Hayakawa,T. (1996) A new mutation of the L1CAM gene in an X-linked hydrocephalus family. Childs Nerv. Syst., 12, 742–747.[ISI][Medline]
45 Ke,S.H. and Madison,E.L. (1997) Rapid and efficient site-directed mutagenesis by single-tube ‘megaprimer’ PCR method. Nucleic Acids Res., 25, 3371–3372.
46 Marg,A., Sirim,P., Spaltmann,F., Plagge,A., Kauselmann,G., Buck,F., Rathjen,F.G. and Brummendorf,T. (1999) Neurotractin, a novel neurite outgrowth-promoting Ig-like protein that interacts with CEPU-1 and LAMP. J. Cell Biol., 145, 865–876.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  C. O. Law, R. J. Kirby, S. Aghamohammadzadeh, and A. J. W. Furley The neural adhesion molecule TAG-1 modulates responses of sensory axons to diffusible guidance signals Development, July 15, 2008; 135(14): 2361 - 2371. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Arevalo, S. Shanmugasundararaj, M. F. Wilkemeyer, X. Dou, S. Chen, M. E. Charness, and K. W. Miller An alcohol binding site on the neural cell adhesion molecule L1 PNAS, January 8, 2008; 105(1): 371 - 375. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Dequidt, L. Danglot, P. Alberts, T. Galli, D. Choquet, and O. Thoumine Fast Turnover of L1 Adhesions in Neuronal Growth Cones Involving Both Surface Diffusion and Exo/Endocytosis of L1 Molecules Mol. Biol. Cell, August 1, 2007; 18(8): 3131 - 3143. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. G. Randles, I. Lappalainen, S. B. Fowler, B. Moore, S. J. Hamill, and J. Clarke Using Model Proteins to Quantify the Effects of Pathogenic Mutations in Ig-like Proteins J. Biol. Chem., August 25, 2006; 281(34): 24216 - 24226. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Zhang, X. Bo, R. Schoepfer, A. J. D. G. Holtmaat, J. Verhaagen, P. C. Emson, A. R. Lieberman, and P. N. Anderson Growth-associated protein GAP-43 and L1 act synergistically to promote regenerative growth of Purkinje cell axons in vivo PNAS, October 11, 2005; 102(41): 14883 - 14888. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Gavert, M. Conacci-Sorrell, D. Gast, A. Schneider, P. Altevogt, T. Brabletz, and A. Ben-Ze'ev L1, a novel target of {beta}-catenin signaling, transforms cells and is expressed at the invasive front of colon cancers J. Cell Biol., February 14, 2005; 168(4): 633 - 642. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Itoh, L. Cheng, Y. Kamei, S. Fushiki, H. Kamiguchi, P. Gutwein, A. Stoeck, B. Arnold, P. Altevogt, and V. Lemmon Brain development in mice lacking L1-L1 homophilic adhesion J. Cell Biol., April 12, 2004; 165(1): 145 - 154. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R E Harrison, J A Flanagan, M Sankelo, S A Abdalla, J Rowell, R D Machado, C G Elliott, I M Robbins, H Olschewski, V McLaughlin, et al. Molecular and functional analysis identifies ALK-1 as the predominant cause of pulmonary hypertension related to hereditary haemorrhagic telangiectasia J. Med. Genet., December 1, 2003; 40(12): 865 - 871. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. G. M. Frints, P. Marynen, D. Hartmann, J.-P. Fryns, J. Steyaert, M. Schachner, B. Rolf, K. Craessaerts, A. Snellinx, K. Hollanders, et al. CALL interrupted in a patient with non-specific mental retardation: gene dosage-dependent alteration of murine brain development and behavior Hum. Mol. Genet., July 1, 2003; 12(13): 1463 - 1474. [Abstract] [Full Text] [PDF]
|
|
|
|
 |