A multi-species comparative structural bioinformatics analysis of inherited mutations in α-D-Mannosidase reveals strong genotype-phenotype correlation
© Khan and Ranganathan. 2009
Published: 3 December 2009
Skip to main content
© Khan and Ranganathan. 2009
Published: 3 December 2009
Lysosomal α-mannosidase is an enzyme that acts to degrade N-linked oligosaccharides and hence plays an important role in mannose metabolism in humans and other mammalian species, especially livestock. Mutations in the gene (MAN2B1) encoding lysosomal α-D-mannosidase cause improper coding, resulting in dysfunctional or non-functional protein, causing the disease α-mannosidosis. Mapping disease mutations to the structure of the protein can help in understanding the functional consequences of these mutations and thus indirectly, the finer aspects of the pathology and clinical manifestations of the disease, including phenotypic severity as a function of the genotype.
A comprehensive homology modeling study of all the wild-type and inherited mutations of lysosomal α-mannosidase in four different species, human, cow, cat and guinea pig, reveals a significant correlation between the severity of the genotype and the phenotype in α-mannosidosis. We used the X-ray crystallographic structure of bovine lysosomal α-mannosidase as template, containing only two disulphide bonds and some ligands, to build structural models of wild-type structures with four disulfide linkages and all bound ligands. These wild-type models were then used as templates for disease mutations. All the truncations and substitutions involving the residues in and around the active site and those that destabilize the fold led to severe genotypes resulting in lethal phenotypes, whereas the mutations lying away from the active site were milder in both their genotypic and phenotypic expression.
Based on the co-location of mutations from different organisms and their proximity to the enzyme active site, we have extrapolated observed mutations from one species to homologous positions in other organisms, as a predictive approach for detecting likely α-mannosidosis. Besides predicting new disease mutations, this approach also provides a way for detecting mutation hotspots in the gene, where novel mutations could be implicated in disease. The current study has identified five mutational hot-spot regions along the MAN2B1 gene. Structural mapping can thus provide a rational approach for predicting the phenotype of a disease, based on observed genotypic variations.
α-D-mannosidase is a lysosomal enzyme which is involved in the catabolism of N-linked glycoproteins through the sequential degradation of high-mannose, hybrid and complex oligosaccharides . The deficiency of this enzyme results in a recessively inherited lysosomal storage disease, called α-mannosidosis, which has been observed in different species in the animals, including domestic cows (Bos taurus), cats (Felis catus), guinea pigs (Cavia porcellus), sheep (Ovis aries) and in humans (Homo sapiens). It was first characterized in humans by Oeckermann in 1967 . Mutations in the MAN2B1 gene, located on chromosome 19 (19 p13.2-q12), encoding lysosomal α-D-mannosidase cause improper coding resulting in dysfunctional or non-functional protein and hence causing the disease. Characterized by immune deficiency, facial and skeletal abnormalities, hearing impairment, and intellectual disability, α-mannosidosis occurs in 1 of 500,000 live births . However, clinicians, geneticists and molecular biologists have not been able to correlate the genotypic mutations with the observed phenotype .
Mapping disease mutations to the structure of a protein can help in understanding the finer aspects of the pathology and clinical manifestations of a disease. Although restricted to diseases where the protein concerned has a known 3-D structure, such an approach is adequately detailed at the molecular level to provide rational explanation for the pathological role of mutations, using protein 3D structure (SOX9 ; human factor H [4, 5]). Therefore, we have attempted a structural bioinformatics approach to understand the role of the different mutations causing α-mannosidosis with differing phenotypes.
From OMIM (Online Mendelian Inheritance in Man) , OMIA (Online Mendelian Inheritance in Animals)  and published literature , a list of inherited mutations for α-mannosidosis has been identified. Various mutations like missense, nonsense, insertions, deletions and also some splicing mutations have been described in the four species to date. Of these only the missense mutations result in a substitution in the protein sequence and were modeled to study their effect on the phenotype. All the other mutations result in the truncation of the protein and its improper function.
An X-ray crystal structure for bovine lysosomal α-D-mannosidase  (PDB ID: 1O7D), solved at a reasonably good resolution of 2.7 Å, is available, albeit lacking two vital disulfide bonds, that hold the five protein chains of the mature α-D-mannosidase protein together, as well as nine of the 20 ligands and a few structurally and functionally important residues. To overcome the limitations of the available 3D structure, we have used homology modeling approaches to reconstruct the complete lysosomal α-D-mannosidase for human, bovine, cat and guinea pig, to which structures we have then mapped all known mutations. The truncation mutations, with the exception of a single truncation in cat, were not modeled, as most of them produced proteins spanning less than two of the five protein chains, leading to a completely compromised active site.
Our comprehensive analysis, taking into account all non-splicing mutations (see Additional file 1: table S1) known to cause α-mannosidosis, reveals a strict genotype-phenotype correlation, contrary to the reports of Malm and Nilssen  and Lyons et al. . This disease can be comparatively well studied as it occurs in different species, providing us with an evolutionary basis for the conserved regions of the protein sequence, as well as active site conservation, where mutations could result in disastrous consequences. Regions with several mutations represent hot-spots where novel mutations could lead to disease. Based on the location of these hot-spots vis-à-vis the active site of the protein, it is also possible to predict the disease phenotype: mild, moderate or severe, extrapolating from known mutations and disease phenotypes. In this paper, we describe a prototype structural bioinformatics analysis method applied to α-mannosidosis, which can be extended to several other diseases for predicting novel disease phenotypes and for developing therapies as well as designing drug/inhibitor molecules.
Observations that the structure of proteins is better preserved during evolution than their sequence, have lead to homology modeling [10, 11] as an reliable methodology for generating 3D structural models of proteins, when the structure of a homologue is available. The most critical step in modeling a correct structure is the alignment of the target sequence with that of the template structure. 3D models of the complete mature wild-type (WT) bovine α-D-mannosidase protein were developed and put through a series of checks for structural verification and analysis to select the best structural model. From this bovine model structure, homology models were constructed for the WT proteins for the other species.
Four complete WT lysosomal α-mannosidase amino acid sequences were retrieved from the Swiss-Prot database , one each, for bovine (Accession No: Q29451), human (Accession No: O00754), guinea pig (Accession No: Q8VHC8) and cat (Accession No: O46432).
Full-length WT protein sequences, spanning all five protein chains, were aligned to the sequence corresponding to the PDB structure, 1O7D, using ClustalX  with default BLOSUM scoring matrices. As the signal peptide (consisting of about 50 N-terminal amino acids) is not present in the mature protein, these were removed from the WT sequences, prior to alignment. The gaps in the alignment were carefully scrutinized and edited manually to preserve chain boundaries and the conservation of structurally and functionally important residues, especially in view of the large segments of residues missing in the PDB structure, 1O7D.
Since α-D-mannosidase is cleaved into five chains, which assemble into a functional enzyme, MODELLER 7V7  proved to be the only homology modeling program allowing multiple chain modeling with ligand inclusion. The models are constructed by optimally satisfying spatial constraints and dihedral angle restraints derived from the alignment of the template structure with the target sequence  and from the CHARMM-22  force field which together enforce proper stereochemistry. Three structural models were generated for each WT and mutated sequence. The model with the lowest current energy and objective function was selected for analysis after carrying out quality assessment and structural refinement.
The model structures were visualized using ICM . The quality of each structural model was evaluated using the three major structural assessment tools, PROCHECK , WHAT IF  and PROVE , which together perform checks on stereo-chemical quality, residue geometry, bond-angle, bond-length and volumetric analysis. These three programs are available as part of the Biotech Validation Suite for Protein Structures.
Structural quality assessment for the wild-type models
Z -Score average
Residues in most favoured regions
Correlation of the effect of substitution mutations on folding and observed disease phenotype. Mutations leading to severe phenotypes are highlighted in bold font. AS: Active site
Effect on Structure
Close to AS
Destabilizes the fold
Away from AS
Close to AS
Destabilizes the fold
Close to AS
Destabilizes the fold
On the whole, 13 substitution mutations have been characterized to cause α-mannosidosis in humans with experimentally verified loss in enzyme activity. H200N  and H72L  disrupt the active site, while E402K  and S453Y  lead to disruption of the ionic linkages with the surrounding charged residues that would otherwise stabilize the protein structure. H72L disturbs a metal-coordinating residue (Figure 5). H200N alters substrate binding and other catalytic properties of the enzyme resulting in no residual enzyme activity. T355P and P356R [26, 28] are located in first α-helix of chain B (Figure 2). These mutations presumably affect the initiation of the helix and thus are likely to disturb the folding of active-site domain. W714R, L809P, R750W [26, 28] and G801D  are located in chain D, where they perturb the structure of the enzyme minimally and result in mild/viable phenotypes. L518P and R916S  are considered the only exception to our correlation, as they lie away from the active site yet potentially disturb the interaction of small E domain (Figure 2) with active site domain. R916S also damages the hydrogen bonding between R916 and D170. These mutations result in a moderately harmful phenotype. R227W in guinea pigs is due to T<C change in the nucleotide sequence at position 679 causing significant loss in enzyme activity . R227 is a structurally and functionally important conserved residue in all the species and its substitution could affect ligand binding.
Phenotypically, α-mannosidosis has a range of expression, with the most common manifestations including mental retardation, hearing impairment, skeletal deformities, and recurrent infections. Diagnosis relies on demonstration of deficient α-mannosidase enzyme activity in leukocytes or other nucleated cells such as fibroblasts. From a clinical perspective, variation is considerable, ranging from a severe infantile form that includes profound mental retardation, hepatosplenomegaly, severe dysostosis multiplex and early death to a mild juvenile form that includes moderate mental retardation, hearing impairment, milder dysostosis and survival into adult life. The high phenotypic variability, even between siblings with identical mutations, has so far prevented adoption of a standardized clinical typing, further complicating research into potential treatments.
This analysis establishes a significant correlation between the genotype and the phenotype of the disease. The feline 1748del4 mutation, which causes a severe genotype and an equally fatal phenotype leading to the destruction of the enzyme structure thereby rendering it non-functional, is a good example of our derived relationship. α-mannosidosis caused by this mutation is fatal due to the absence of mannosidase activity in the liver of the Persian cats. There are also missense mutations like the bovine F320L, where the phenotype (with only 0.3% of the normal levels of enzyme activity) was as severe as the effect of the mutation on the genotype of the enzyme. The effect of this mutation cannot be explained by sequence analysis alone. Our work suggests that the mutations in MAN2B1 gene are scattered over the entire gene providing us with five mutational hot-spot regions. This gives us an opportunity to predict the degree of severity for a particular mutation and also to predict the residues that are most likely to undergo mutations based on their genotypic location. Moreover, the high degree of mutational heterogeneity of α-mannosidosis is comparable to that observed in many other lysosomal disorders. Based on the co-location of mutations from different organisms (human, cow, guinea pig and cat) and their proximity to the enzyme active site, we have extrapolated observed mutations from one species to homologous positions in other organisms, as a predictive approach for detecting likely α-mannosidosis, based on orthologous positions in the multiple sequence alignment of the α-mannosidase sequences. Our investigation highlights the effect of disease mutations on protein structure and forms the basis for understanding the molecular determinants for phenotypic variations. This study could play a vital role in developing therapies for inherited diseases. Since lysosomal α-mannosidase is an essential enzyme and all observed mutations affect its functioning, our study suggests that rather than drug/inhibitor design, this disease could be tackled through gene therapy.
Other papers from the meeting have been published as part of BMC Bioinformatics Volume 10 Supplement 15, 2009: Eighth International Conference on Bioinformatics (InCoB2009): Bioinformatics, available online at http://www.biomedcentral.com/1471-2105/10?issue=S15.
JMK is grateful to Macquarie University for the award of MQRES research scholarship. Open access publication charges were borne by Macquarie University.
This article has been published as part of BMC Genomics Volume 10 Supplement 3, 2009: Eighth International Conference on Bioinformatics (InCoB2009): Computational Biology. The full contents of the supplement are available online at http://www.biomedcentral.com/1471-2164/10?issue=S3.
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.