Ultrastructural colocalization of nidogen-1 and nidogen-2 with laminin-1 in murine kidney basement membranes.

Nidogen-1, a key component of basement membranes, is considered to function as a link between laminin and collagen type IV networks. Recently a new member of the nidogen family, nidogen-2, has been characterized. Preliminary immunohistochemical data indicated that nidogen-1 and nidogen-2 show a similar tissue distribution at the light microscopic level. We have now localized nidogen-1 and nidogen-2, as well as their corresponding mRNAs, at the light and electron microscopic levels in adult mouse kidney, by in situ hybridization and immunogold histochemistry, as well as carrying out double labeling with laminin-1. Both nidogen-1 and nidogen-2 mRNAs are found not only in mesenchymal cells of embryonic tissues, but also in all epithelial and endothelial cells in adult mouse kidney. Both nidogens are ubiquitous basement membrane components in the mouse kidney, being found in glomerular, tubular, and capillary compartments and Bowman's capsule. Furthermore, a substantial fraction of nidogen-1 and nidogen-2 colocalizes with laminin-1. The results indicate that nidogen-1 and nidogen-2 could well substitute for one another in some of their biological activities in kidney, for example, stabilizing basement membrane networks in vivo.

Recombinant domain IV of perlecan binds to nidogens, laminin-nidogen complex, fibronectin, fibulin-2 and heparin.

Domain IV of mouse perlecan, which consists of 14 immunoglobulin superfamily (IG) modules, was prepared from recombinant human cell culture medium in the form of two fragments, IV-1 (IG2-9, 100 kDa) and IV-2 (IG10-15, 66 kDa). Both fragments bound to a heparin column, being eluted at ionic strengths either below (IV-2) or above (IV-1) physiological level, and could thus be readily purified. Electron microscopy demonstrated an elongated shape (20-25 nm), and folding into a native structure was indicated by immunological assay and CD spectroscopy. Solid-phase and surface plasmon resonance assays demonstrated strong binding of fragment IV-1 to fibronectin, nidogen-1, nidogen-2 and the laminin-1-nidogen-1 complex, with Kd values in the range 4-17 nM. The latter binding apparently occurs through nidogen-1, as shown by the formation of ternary complexes. Only moderate binding was observed for fibulin-2 and collagen IV and none for fibulin-1 and BM-40. Fragment IV-2 showed a more restricted pattern of binding, with only weaker binding to fibronectin and fibulin-2. None of these activities could be demonstrated for recombinant fragments corresponding to the N-terminal perlecan domains I to III. This indicates a special role for domain IV in the integration of perlecan into basement membranes and other extracellular structures via protein-protein interactions.

Nidogen-2: a new basement membrane protein with diverse binding properties.

Human nidogen-2 was cloned and sequenced (1375 residues) and found to share 46% sequence identity and a similar domain arrangement with the previously characterized basement membrane protein nidogen-1. Recombinant nidogen-2 was purified as a 200 kDa protein from transfected mammalian cell medium, showed a high level of N and O-glycosylation, and could be clearly distinguished from nidogen-1 (150 kDa) by specific antibodies. Electron microscopy demonstrated that the two isoforms have a similar shape, consisting of three globular domains connected by two threads, but differ somewhat in length. Northern blots and immunological assays demonstrated co-expression of the nidogens in various tissues and cultured cells. Immunofluoresence revealed colocalization in vessel walls and other basement membrane zones but some differences in heart and skeletal muscle. Nidogen-2 interacted with collagens I and IV, and perlecan at a comparable level to nidogen-1 but failed to bind to fibulins. Nidogen-2 bound to laminin-1, but only moderately to the epitope on the laminin gamma1 chain, which promotes high-affinity binding of nidogen-1. Both nidogens were cell-adhesive for a restricted number of cell lines, with nidogen-2 having a higher activity. Together, these data suggest that nidogen-2 can compensate for some but not all functional activities ascribed to nidogen-1.

Structural and genetic analysis of laminin-nidogen interaction.

High-affinity binding of nidogen to laminins involves a single binding site on the laminin gamma 1 chain and is thus a property shared by almost all laminin isoforms. This binding mediates the connection of laminins to the collagen IV network, perlecan and other proteins and is considered to be an essential step in the stabilization of basement membranes. Nidogen binding has been located to a single LE module (gamma 1III4) by recombinant analysis. Site-directed mutagenesis and X-ray crystallography demonstrated that three amino acids (Asp, Asn, Val) in loop a of gamma 1III4 are crucial for binding and are supported by some other residues. A restricted complementary binding region seems to exist on nidogen domain G3. A mutant laminin gamma 1 chain gene that lacks the region encoding gamma 1III4 was prepared in mouse embryonic stem (ES) cells by homologous recombination. ES cells homozygous for this defect were shown to assemble laminin-1 into a cruciform structure and to secrete it properly. Yet the mutant laminin failed to associate with nidogen. The mutant ES cells were still able to form embryoid bodies with a similar differentiated histology as the wild type. Immunofluorescence, however, indicated an impaired deposition of nidogen into basement membrane-like structures.

Properties of the extracellular calcium binding module of the proteoglycan testican.

The extracellular calcium-binding (EC) module of human testican (115 residues) was obtained in native form by recombinant production in mammalian cell culture and thus shown to represent an independently folding domain. This module showed a large loss in alpha-helix upon calcium depletion. Apparently only one of the two EF hands binds calcium, with a moderate affinity (Kd =68 microM) about 100-fold lower than in the homologous BM-40 protein. No clear evidence was obtained for collagen binding, indicating that EC modules found in different proteins may not share similar functions.

Conversion of the Kunitz-type module of collagen VI into a highly active trypsin inhibitor by site-directed mutagenesis.

The recombinant Kunitz protease inhibitor module (domain C5) of human collagen alpha 3(VI) chain was previously shown to lack inhibitory activity for proteases with trypsin-like specificity and some other proteases. We have now prepared mutants in the binding loop region including the P1' site (D2889-->A), the P2' site (F2890-->R) and the P3 site (T2886-->P) and in a more remote region (W2907-->V) either as individual substitutions or combinations of them. These mutants were analyzed for their kinetics of binding to trypsin by surface plasmon resonance and for their capacity to inhibit various proteases. Single substitutions (D-->A, T-->P, W-->V) showed an effect only for D->A which bound to trypsin with Kd = 0.25 microM. A 25-100-fold increase in affinity was observed for the double mutants T-->P/D-->A and F-->R/D-->A and approached the affinity of aprotinin (Kd approximately 0.01 nM) in two different triple mutants. These affinities correlated well with the inhibitory capacities of the mutants for trypsin in the cleavage of a large protein and a small peptide substrate. A similar but not completely identical improvement in inhibitory capacity was also observed for leucocyte elastase but not for thrombin. These data could be interpreted in terms of steric interferences or lack of hydrogen bonding of a few critical residues based on three-dimensional structures available for the C5 domain.

N-methyl-N-nitrosourea-induced mutations in human cells. Effects of the transcriptional activity of the target gene.

In this study we addressed the question as to whether the mutagenesis by methylating agents is affected by the transcriptional activity of the damaged gene. An Epstein-Barr virus (EBV)-derived shuttle vector system was developed where the genetic target for mutation analysis, the bacterial gpt gene, is under the control of an eukaryotic inducible promoter in plasmid pF1-EBV and lacks the eukaryotic promoter in plasmid pF2-EBV. Two human cell lines that episomically maintain these shuttle vectors were established. In clone 6NT cells, which contain pF1-EBV plasmid, the gpt gene is actively transcribed and the transcription rate is regulated by zinc ions. In clone 3 cells, which harbor pF2-EBV plasmid, the gpt gene is not transcribed. Following treatment of both cell lines with the potent alkylating carcinogen N-methyl-N-nitrosourea (MNU), G.C to A.T transitions were the major mutagenic event, consistent with the miscoding potential of O6-methylguanine. The mutations were predominantly generated in the non-transcribed DNA strand of the active gpt gene. The same strand-bias was observed when the gpt gene was transcriptionally inactive, indicating that MNU-induced strand-specific formation of mutations is not due to transcription. Our data identify as major determinants of this phenomenon the sequence-specificity of MNU mutagenesis and the conformational properties of the target protein. Differences in mutation distribution were observed between the transcriptionally active and inactive gpt gene. This finding suggests that the organization of active genes in chromatin might modulate DNA alkylation and/or DNA repair.

Action of gamma endonuclease on clustered lesions in irradiated DNA.

Irradiation of DNA in situ i.e. in phage particles or in the cell leads to alterations of single DNA nucleotides as well as to clustered lesions such as double strand breaks or unpaired DNA regions the latter being sensitive to digestion by S1 nuclease. A contribution will be made to the configuration of such S1-nuclease-sensitive sites (S1 sites). DNA from irradiated lambda phage containing S1 sites was treated with gamma endonuclease from M. luteus which is known to split the nucleotide strand at the position of oxidized pyrimidine base. It was found that the gamma endonuclease induces double-strand breaks at some of the S1 sites indicating double base damage within this site. However, half of the S1 sites are not converted into a double-strand break by the gamma endonuclease, indicating base damage only on one strand within the unpaired region.