Previous Article | Next Article ![]()
Eukaryotic Cell, February 2004, p. 232-240, Vol. 3, No. 1
1535-9778/04/$08.00+0 DOI: 10.1128/EC.3.1.232-240.2004
Copyright © 2004, American Society for Microbiology. All Rights Reserved.
Lehrstuhl für Allgemeine und Molekulare Botanik, Ruhr-Universität Bochum, 44780 Bochum, Germany
Received 29 August 2003/ Accepted 24 October 2003
|
|
|---|
|
|
|---|
The sexual life cycle of ascomycetes can be either heterothallic (self-sterile) or homothallic (self-fertile). Heterothallic fungi, such as N. crassa, exist in two mating types, A and a, and mating only occurs between sexual structures of opposite mating type. However, in S. macrospora, a homothallic fungus, every strain is able to complete the sexual cycle without a mating partner. Fruiting body development starts with the formation of female gametangia, which are enveloped by sterile hyphae to form closed spherical prefruiting bodies (protoperithecia), which can be considered primordial stages of mature fruiting bodies. Subsequent cell differentiation gives rise to an outer-pigmented peridial tissue and inner ascus initials embedded in sterile paraphyses, which leads to the formation of mature flask-like fruiting bodies, called perithecia.
It has been demonstrated for a number of ascomycetes that several genes control not only sexual development but also asexual sporulation (1). In contrast to other fungal model organisms, S. macrospora produces only meiotically derived ascospores, while asexual spores, such as conidia, are absent. Since there is no interference between sexual and asexual developmental programs, S. macrospora is an ideal model to identify genes involved in the sexual differentiation process. Molecular genetic procedures have been applied to characterize the genes that regulate this developmental process in S. macrospora (24, 31, 46). To isolate additional regulatory genes from S. macrospora, we used ethyl methanesulfonate mutagenesis to generate mutants with defects in fruiting body formation. Within our collection of mutants, we identified a class of strains which are arrested in the transition from the protoperithecium to the perithecium. Sterile mutants of this type were all designated with the prefix pro- to indicate that they form protoperithecia only. Using the pro1 mutant, we previously identified a Zn2C6 transcription factor which is essential for the morphological transition from protoperithecia to perithecia (24).
In this study, the molecular and functional analysis of mutant pro11 is illustrated, which serves the identification of the pro11 gene controlling essential steps in fruiting body development. The predicted gene product is highly homologous to mammalian brain proteins belonging to the striatin family. Members of this family, including striatin, zinedin, and SG2NA, are found in the cytosol and are also associated with membranes. In addition, they are endowed with diverse protein-protein association modules, for instance, a caveolin-binding motif, a coiled-coil structure, a Ca2+-calmodulin-binding domain, and multiple WD40 repeats homologous to WD40 repeats in the beta subunit of trimeric G-proteins. Due to these domains, proteins of the striatin family are engaged in multiple protein associations and are thought to act as scaffolds as well as signaling proteins (5, 10, 11, 16).
Here, we demonstrate that a mouse striatin gene is able to rescue the sterile phenotype of the S. macrospora pro11 mutant, strongly suggesting functional conservation of the fungal pro11 gene and the mammalian striatin gene. Complementation studies with C-terminally truncated versions of the PRO11 protein revealed different functional roles of this multidomain protein, and experimental evidence is provided for a membrane association of the PRO11 polypeptide, which shares this feature with animal homologues.
|
|
|---|
Plasmids. Cloning of S. macrospora DNA fragments with vectors pBCKS+, pQE31 (Qiagen), pDrive (Qiagen), pANsCos1 (32), and pEHN2, a derivative of plasmid pIG1783 without the egfp gene (33), was performed by standard techniques (39). Vector pEHN2 is a fungal expression vector with the strong gpd promoter of Aspergillus nidulans upstream of a single NotI cloning site. The cosmids and plasmids used in this investigation are listed in Table 1.
|
View this table: [in a new window] |
TABLE 1. Cosmids and plasmids used for transformation experiments
|
Antibody production. His tag fusion derivatives of a pro11 gene fragment encoding the N terminus of the PRO11 polypeptide were synthesized in Escherichia coli strain M15. For this purpose, an 857-bp SalI/PstI fragment was amplified with primers p11-31 (5'-GTCGACATGGGCACCAACGGCGTTCATGTC; SalI cloning site italic) and p11-32 (5'-CTGCAGTCTAGACGAACGCAACGCTCCATGGG; PstI cloning site italic), with S. macrospora cDNA as a template.
For sequence verification, the amplified fragment was subcloned into the vector pDrive (Qiagen). After sequence verification, the amplified fragment was subsequently cloned into vector pQE31 (Qiagen, Germany) to generate plasmid pAK2 (Table 1). This plasmid was used to express the 5' end of the cDNA of the pro11 gene in E. coli. Proteins containing an N-terminal His tag were purified on Ni-nitrilotriacetic acid-agarose columns (Qiagen) according to the supplier's instructions. For antiserum production, purified eluates of E. coli strains expressing the 5' end of the pro11 open reading frame were dialyzed against 50 mM ammonium carbonate and evaporated. The production of polyclonal antiserum against PRO11 in rabbits was performed by Eurogentec (Belgium).
Protein fractionation and Western analysis. Cell fractionation procedures for S. macrospora were performed by a series of centrifugation steps as described by Bowman et al. (8). The 10,000 x g, 100,000 x g (100K), and soluble protein fractions were separated in sodium dodecyl sulfate-polyacrylamide (SDS-PAGE) gels. The proteins were transferred to polyvinylidene difluoride Western blotting membranes (Roche, Germany) with a semidry blotting system (Biometra, Germany). Detection was carried out with a polyclonal anti-PRO11 antibody, a polyclonal antibody against actin of Aspergillus niger (Acris Antibodies, Bad Nauheim, Germany), and an antibody against mitochondrial porin from Saccharomyces cerevisiae (REF; generous gift of W. Kunau, Bochum University) with the chemiluminescence Western blotting kit (Roche, Germany) following the manufacturer's instructions.
Light microscopy. Light microscopic observations were conducted with a Zeiss Axiophot microscope with appropriate filter sets. For light microscopy, nuclei were stained with DAPI (4',6'-diamidino-2-phenylindole, 0.5 µg/µl).
DNA and protein sequence analysis. Sordaria macrospora protein sequences were used as query sequences in Blast searches (2, 17). Preliminary A. fumigatus genome sequences were obtained from the the Institute for Genomic Research (TIGR) (http://www.tigr.org). Sequences of Magnaporthe grisea were obtained from the Magnaporthe Sequencing Project (Release I, Ralph Dean, Fungal Genomics Laboratory at North Carolina State University, http://www.fungalgenomics.ncsu.edu) and the Whitehead Institute/MIT Center for Genome Research (www-genome.wi.mit.edu). Sequences of the basidiomycete Phanerochaete chrysosporium were obtained from the DOE Joint Genome Institute (http://www.jgi.doe.gov/programs/whiterot.htm).
Amino acid sequences and sequence alignments were done with the Clustal W program (44). Prediction of coiled-coil domains of the proteins was performed with the programs Coils (23), Paircoil (7), and Multicoil (48) at the ExPASY server (http://ca.expasy.org/); prediction of WD40 repeats was carried out at the Biomolecular Engineering Research Center (BMERC) (http://bmerc-www.bu.edu/bioinformatics/WD40repeat.html). and prediction of potential calmodulin binding sites was done at the Calmodulin Target Database (http//calcium.oci.utoronto.ca).
Nucleotide sequence accession number. Sequence data for the S. macrospora pro11 gene have been submitted to the EMBL database under accession number AJ564211.
|
|
|---|
![]() View larger version (126K): [in a new window] |
FIG. 1. Sexual developmental stages of the wild-type (wt) and mutant (pro11) strains. Differential interference microscopy identified ascogonia (wild type, pro11), protoperithecia (wild type, pro11), and young perithecia (only wild-type strain). Strains were grown on fructification medium and examined after growth at 25°C for the number of days indicated.
|
Complementation analysis identifies a gene for a WD40 repeat protein. The pro11 gene was cloned by a recently described genomic complementation strategy (24) with an indexed cosmid library of S. macrospora (34). A single clone with an insert size of 34 kb, designated pANsC7, yielded fertile transformants that showed perithecium and ascospore formation. In subsequent complementation experiments, we succeeded in isolating a 3.6-kb ApaI/EcoRI subfragment of cosmid pANsC7 (Fig. 2A) that was able to complement pro11.
![]() View larger version (31K): [in a new window] |
FIG. 2. Complementation analysis to restore fertility in pro11 and structural comparison of eukaryotic WD40 repeat proteins with PRO11 (AJ564211
[GenBank]
). (A) Molecular organization of the pro11 gene. Restriction map of the sequenced 4.0-kb fragment carrying the pro11 gene. The location of the open reading frame is indicated by an arrow, and the three introns are marked by grey boxes. Fragments generated by reverse transcription-PCR are shown below. (B) Recombinant plasmids carrying inserts of the pro11 gene were used for complementation experiments. The ability of transformants to restore fertility (wild-type phenotype) is indicated by a +, while no complementation is shown by ; (+) denotes transformants which produced few fertile perithecia but showed a hyphal morphology similar to that of the mutant strain. Abbreviations: A, ApaI; B, BamHI; Bg, BglII; E, EcoRI; H, HindIII; P, PstI; X, XhoI; Xb, XbaI. (C) Schematic representation of PRO11 and some closely related proteins. White boxes represent coiled-coil domains, and numbered black boxes show the position of WD40 repeats. Black boxes indicate the position of calmodulin binding sites and caveolin binding domains in striatin, SG2NA, and zinedin. Accession numbers: Mm, Mus musculus striatin (055106); Hs, Homo sapiens SG2NA (Q13033) and H. sapiens zinedin (NP_037535
[GenBank]
); Dm, Drosophila melanogaster CKA (Q9VLT9); Ce, Caenorhabditis elegans protein encoded by the K07C5.8 gene (Q17406); Sp, Schizosaccharomyces pombe protein encoded by the SPBC1773.01 gene (O94560). (D) Alignment of the seven predicted WD40 repeats in PRO11. Putative ß strands are shaded and their positions are indicated with arrows (A to D). The WD40 repeat core consensus is given at the bottom. The coordinates of the amino acid positions are marked on the left.
|
To verify that we did not identify a suppressor gene of the pro11 mutation, the mutant allele was sequenced. It differs from the wild-type pro11 gene by a single point mutation within the coding region, resulting in an in-frame stop codon (TGG Trp to TGA stop) at amino acid position 546 of PRO11 (Fig. 2C). In addition, transformation experiments were carried out with clones containing truncated versions of the pro11 gene (Fig. 2B). Only the full-length version of the pro11 gene was able to fully complement the sterile phenotype of strain pro11. Truncated versions containing the coding region for the N-terminal part of PRO11 or a deletion in the C-terminal coding part of the pro11 open reading frame (pSB10 and pSB22, respectively) resulted in only partial complementation (Fig. 2B). Partially complemented transformants were able to form fertile fruiting bodies with asci and ascospores. However, the number of fruiting bodies and ascospores was greatly reduced, and the transformants still showed an increased density of aerial hyphae. However, no complementation was observed when the N-terminal part of the pro11 open reading frame contained a frameshift mutation (pSB10H [+4]).
Characterization of the WD40 repeat protein PRO11. Database searches revealed that the PRO11 protein showed significant sequence similarity to three mammalian brain proteins, all WD40 proteins of the striatin family, which includes striatin (055106), SG2NA (Q13033), and zinedin (NP_037535 [GenBank] ). These multimodular WD40 repeat and calmodulin binding proteins are thought to act as scaffolds, as signaling proteins, or as catalytic subunits of protein phosphatase 2A (5, 10, 11, 29). In addition, the PRO11 protein shows distinct homology with the recently identified connector of kinase to AP-1 (CKA) protein of Drosophila melanogaster (Q9VLT9), a multidomain WD40 protein, which regulates the Jun N-terminal kinase signal transduction pathway and, thereby, epithelial sheet movement that occurs during dorsal closure (12). In addition to proteins of the striatin family, PRO11 was found to be homologous to WD40 proteins of unknown function (Fig. 2C).
Within filamentous ascomycetes, the pro11 gene seems to be very conserved. Blast searches through the genomic sequences of N. crassa, Magnaporthe grisea, Aspergillus fumigatus, and the basidiomycete Phanerochaete chrysosporium indicated that they contain a single gene encoding a protein which displays 94.6%, 62.8%, 48.4%, and 28.4% amino acid identity, respectively, to PRO11. In contrast, no pro11 homologue could be identified in the genome of the yeast Saccharomyces cerevisiae.
We analyzed the predicted PRO11 protein and found that a 41-amino-acid region near the N terminus of the protein (positions 56 to 96) has the potential to form a coiled-coil structure (Fig. 2C). This domain is one of the common structural motifs mediating protein-protein interactions (22). MultiCoil and Paircoil scores indicate that this region has a >80% probability of forming a dimer (7, 48). A putative Ca2+-calmodulin binding site was predicted in the N terminus between amino acid 80 and amino acid 91. In addition, computer analysis revealed that an expanded C-terminal domain of PRO11 (419 amino acids, positions 426 to 844) consists of seven copies of the degenerate WD40 repeat (Fig. 2D). WD40 repeats are domains of approximately 40 to 60 amino acids. Each repeat contains a variable domain and a core region, which includes preferentially conserved amino acids typically bracketed by the dipeptides Gly-His and Trp-Asp, and correspond to a permutated structural repeat of four antiparallel ß strands (41). All of these features are present in the predicted WD40 repeats of PRO11 (Fig. 2D).
Subcellular localization of the WD40 repeat protein in fungal cells. It is important to consider the subcellular localization of the PRO11 WD40 protein because mammalian homologues, such as striatin, SG2NA, and zinedin, are neuronal proteins that are strictly expressed in the somatodendritic compartments. These are found in the cytosol and are also associated with membranes (11, 16, 38).
To obtain further information about the subcellular location of the fungal homologue PRO11, we raised an antibody against the N-terminal part of PRO11. In a Western blot analysis, the PRO11 antibody detected a band of about 90 kDa in the wild-type strain as well as in transformants expressing the pro11 open reading frame, but no bands in pro11 (Fig. 3). The absence of the PRO11 protein in the mutant can be explained by the missense mutation within the mutated gene affecting the stability of the truncated protein. Further analysis of the 100K and soluble fractions showed that only the 100K membrane fraction contained detectable amounts of the PRO11 protein in the wild-type strain (Fig. 3). This, however, does not exclude that trace amounts of PRO11 are present in other protein fractions as well. As a control, the soluble protein as well as the 100K membrane fraction were characterized with antibodies against actin and the mitochondrial membrane protein porin (15), respectively (Fig. 3).
![]() View larger version (26K): [in a new window] |
FIG. 3. Western blot analysis of polypeptides from different strains. Aliquots (30 µg) of protein fractions, as indicated, were fractionated on denaturing SDS-8% PAGE. Western blot detection of PRO11 protein was performed with a polyclonal PRO11 antibody. As a control, actin and porin were detected with specific antibodies in different protein fractions. Size markers are indicated. Abbreviations: wt, wild-type strain; pro11, pro11 mutant; Tr38, pro11 transformant carrying plasmid pIG1807-24; SF, soluble fraction; 100k, protein fractions as described in the text.
|
Proteins with very similar surfaces are expected to have related functions, and therefore we raised the question of whether the mammalian striatin gene was able to complement developmental defects of the S. macrospora mutant pro11. For this purpose, we amplified the coding region of the mouse striatin gene from a brain cDNA preparation and cloned the resulting PCR fragment into the fungal expression vector pEHN2. The corresponding construct, pIG1808-23 (Table 1), which contains the mouse striatin gene under the control of the gpd promoter of Aspergillus nidulans, was transformed into pro11. Transformants carrying the construct developed fertile perithecia containing rosettes of asci (Fig. 4). DAPI staining of nuclei proved that several asci contained eight ascospores with two nuclei, showing that karyogamy, meiosis I and II, and two postmeiotic mitoses as essential processes of the sexual cycle were fully restored in the transgenic strains (Fig. 4A). The frequency of unusual asci with less than eight ascospores was, however, increased (Fig. 4B). In some instances, the separation of nuclei after meiosis II or postmeiotic mitosis seemed to be impaired and led to multinuclear ascospores. This complementation analysis strongly supports the assumption that the mouse striatin gene is a functional homologue of the fungal pro11 gene.
![]() View larger version (98K): [in a new window] |
FIG. 4. Ascus development in the wild type and a representative pro11 transformant (Tr) carrying plasmid pIG1808-23 and expressing the mouse striatin gene. (A) DAPI staining identified nuclei during karyogamy (K), meiosis II (MII), and postmeiotic mitosis (PM). The second postmeiotic mitosis (2nd PM) occurs after spore wall formation. (B) Rosettes from a mature perithecium are shown. Note that some of the asci of the transformant carry less than eight ascospores, which are regularly seen in a wild-type ascus.
|
|
|
|---|
Proteins orthologous to striatin, SG2NA, and zinedin are present in metazoa throughout their evolution; they have been detected in nonmammalian vertebrates (Danio rerio and Xenopus laevis) and in invertebrates (Drosophila melanogaster and Caenorhabditis elegans) (27). In D. melanogaster, it was demonstrated that the striatin homologue CKA regulates AP-1 activity by organizing a molecular complex of kinases and transcription factors, thus coordinating the spatial and temporal expression of AP-1-regulated genes (12). A recent search for calmodulin binding proteins in the genome of Arabidopsis thaliana revealed that 29 animal calmodulin binding proteins, including striatin, SG2NA, and zinedin, have no homologues in A. thaliana. However, with the exception of S. cerevisiae, striatin gene homologues have been detected in various fungal genomes.
N-terminal part of PRO11 is sufficient to restore fertility. The mutant pro11 gene contains a single G to A substitution, changing a conserved Trp (TGG) codon of the third WD40 repeat into a stop (TGA) codon. This removes all but the first two of the WD40 domains. Mutant pro11 does not seem to produce any detectable amount of a truncated protein, as was shown by Western blot analysis. Somehow, the mutation either prevents translation of the mRNA or destabilizes the translated protein. Interestingly, a truncated version of the wild-type pro11 gene, encoding only the N terminus of the PRO11 protein, was capable of partially complementing the developmental defects of pro11. Thus, the N-terminal domains of PRO11 retain significant functions on their own, and it appears to be more beneficial for PRO11 to be without a WD40 repeat domain than to have a defective one. A comparable observation was described before for the WD40 repeat protein COP1, a photomorphogenesis repressor of A. thaliana. Similar to PRO11, the COP1 protein contains a C-terminal WD40 repeat domain. In A. thaliana, all characterized lethal alleles of the cop1 gene contain mutations within the WD40 repeat domain, whereas weak mutant alleles lack the entire WD40 domain (18, 25). Similarly, in the human lisencephaly-1 gene encoding a WD40 repeat protein whose haploinsufficiency causes a devastating brain malformation known as lissencephaly, mutations cluster in exons which code for WD40 repeats 4, 5, and 6. Several of these are missense mutations resulting in protein truncations (9, 21, 45).
As discussed for the COP1 protein, there may be two possible explanations for the observation that the N-terminal portion of PRO11 itself retains some function, whereas a defective WD40 domain disrupts the function of the N-terminal domain. One possibility is that the truncation of the WD40 repeat in the protein encoded by the mutant pro11 gene disrupts the function of the N-terminal portion, possibly due to an unfavorable conformation. Alternatively, some other WD40 repeat proteins in the fungal cell may be able to take the place of the C-terminal domain. This substitution cannot take place in mutants with a defective WD40 repeat motif as a result of steric hindrance.
Animal homologues of PRO11 are functionally conserved in eukaryotic cell development. Functional conservation of mammalian and fungal genes at the cellular level has previously been demonstrated for yeasts and filamentous fungi. For example, the human homologue of the Schizosaccharomyces pombe cdc2 gene, encoding a serine-threonine protein kinase, was cloned by expressing a human cDNA in fission yeast (20) and the rat c15 gene is capable of rescuing the nuclear movement defect of a nudC3 mutant from Aspergillus nidulans (30).
Similarly, the mouse striatin cDNA was able to restore the wild-type developmental program in pro11. Beside structural similarities, these data clearly indicate direct functional homology between the cognate proteins, implying that the cellular function of the PRO11 protein may be evolutionarily conserved from fungi to humans.
Animal homologues of PRO11 have been identified as components in three different protein complexes (Fig. 5). (i) In mammals, striatin, SG2NA, and zinedin are neuronal proteins strictly expressed in somatodendritic compartments and primarily localized to the cytosol and membranes (10, 11). PRO11 is physically present at several sites of the fungal cell, suggesting that the protein functions in multiple cellular processes. Members of the striatin family are thought to be involved in visual and neuron-specific Ca2+/calmodulin signal transduction cascades and have been shown to form a complex with protein phosphatase 2A. As depicted in Fig. 5, it was hypothesized that striatin and SG2NA may target the catalytic subunit C of protein phosphatase 2A to components of Ca2+-dependent signaling pathways (11, 29). As a result, members of this protein family may be involved in Ca2+-dependent neural functions, such as neurotransmission, maintenance, or dismantling of cytoskeletal elements (38). Moreover, it was demonstrated that the proteins of the striatin family directly bind caveolin-1, a membrane associated protein (16). Caveolins are scaffolding proteins able to collect a large number of signaling proteins bearing a caveolin-binding motif and confer the shape and structural organization of caveolae during caveola-mediated endocytosis (13, 40). Although endocytosis has not conclusively been demonstrated for filamentous fungi, the recently sequenced genome of Neurospora crassa led to the identification of genes encoding putative components of a complex endocytic protein machinery (36).
![]() View larger version (29K): [in a new window] |
FIG. 5. Scheme showing the signal transduction pathways of animal striatin-like proteins. All binding partners of the striatin-like proteins from mammals (Homo sapiens and Rattus norwegicus) and Drosophila melanogaster are shown in grey. With the exception of caveolin, genes encoding homologues for all depicted polypeptides have been identified in the fully sequenced genome of Neurospora crassa.
|
(ii) As shown in Fig. 5, an additional function was assumed from the data of a two-hybrid screen with striatin as a bait. Baillart et al. (4) identified phocein, an intracellular protein which bears a few homologies with the
-subunits of clathrin adaptor proteins. With phocein as a bait, nucleoside diphosphate kinase and Eps15a multidomain protein, involved in clathrin-mediated endocytosis, were identified as interacting partners. Since nucleoside diphosphate kinase and Eps15 have been shown to be functional partners of dynamin, a GTPase that plays a critical role in endocytosis, the hypothesis that phocein and striatin might play a role in membrane and vesicular traffic is strengthened (3, 19, 37, 43).
(iii) D. melanogaster mutants lacking the striatin homologue CKA fail in dorsal closure during embryogenesis, causing the embryos to develop a hole in the dorsal epidermis. The Drosophila CKA protein is an essential component of the Jun N-terminal kinase pathway, which is required for this developmental process (12). Interestingly. CKA homologues from C. elegans to mammals have been reported to replace the D. melanogaster CKA in activating the Jun N-terminal kinase signal transduction pathway (12), indicating that striatin binding partners are not limited to calmodulin, caveolin, protein phosphatase 2A, MOB1, and phocein.
In consideration of our data as a whole, it seems reasonable to assume that the S. macrospora multidomain protein PRO11 regulates signaling by organizing protein complexes of kinases, phosphatases, and transcription factors. In this context, it is noteworthy that we have identified the S. macrospora homologues of phocein and MOB1, showing a high degree of similarity (33% and 49%, respectively) to the corresponding human proteins (S. Pöggeler unpublished data). Furthermore, a search of the genome sequence of N. crassa revealed that, with the exception of caveolin, all putative binding partners of PRO11 are encoded by this closely related species (Fig. 5). Therefore, the pathways in which the PRO11 protein might be involved are control of cytokinesis, vesicular traffic, and endocytosis.
Mutant alleles of the striatin, SG2NA, and zinedin genes are not known. However, downregulation of striatin in vivo with antisense oligonucleotides results in decreased locomotor activity in adult rats. In contrast, in vitro downregulation in embryonic motoneurons leads to a blockade of dendritic but not axonal growth, indicating that striatin might play a role in the establishment of polarity within developing neurons (6).
The S. macrospora sexual pathway provides an optimal experimental system for studying the mechanism and regulation of developmental processes that are usually obscured by complex cell-cell interactions involved in animal and plant development.
This work was funded by a grant from the Deutsche Forschungsgemeinschaft (SFB 480) to S.P and U.K.
Dedicated to K. Esser on the occasion of his 80th birthday. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2010 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»