Search

Background The usefulness of antibodies and antibody derived artificial constructs in various medical and biochemical applications has made them a prime target for protein engineering, modelling, and structure analysis. The huge number of known antibody sequences, that far outpaces the number of solved structures, raises the need for reliable automatic methods of antibody structure prediction. Antibodies have a very characteristic molecular structure that is reflected in their modelling technique. Currently, the most accurate models are produced using a quite peculiar modelling strategy, developed among others by our group: the framework regions are modelled with a standard comparative modelling approach, whereas the hypervariable loops are predicted using the ad-hoc “canonical structure method”, historically based on expert analysis of the available antibody solved structures. More than thirty years passed since this modelling method was initially developed, nonetheless there is still a huge effort in the academic and pharmaceutical communities to improve its accuracy. The reason for this lies in several error sources in the current modelling process. First of all, given the large amount of available structures, it was impossible to manually update “canonical structure” classes and rules. Moreover, the lack of specific studies on the packing between the VL and the VH domains and on possible conformational changes occurring upon antigen binding was impairing the integration in the modelling techniques of such factors. Aim The general aim of this study is to carry out an extensive characterization and annotation of immunoglobulin molecules i.e. to deepen our understanding of the molecular basis of their specificity using a combination of bioinformatics sequence- and structure-based analysis. I carried out improvements to the antibody modelling protocols by revising the canonical structure definitions and by minimizing the errors arising from VL and the VH domain packing at the same time by taking care of the conformational changes occurring upon antigen binding. Results During the past years, we successfully improved the description of the structural repertoire of immunoglobulins with lambda light chains, which has both practical (design, engineering and humanization) and theoretical applications (improvement of the antibody modelling)[1]. Our large-scale analysis of the association of heavy and light chain variable domains in antibodies showed that there are essentially two different modes of interaction that can be identified by the presence of key amino acids in specific positions of the antibody sequences [2]. Interestingly, we also found that the different packing modes are related to the volume and type of recognized antigen. These findings are clearly relevant for the design of antibodies and of antibody libraries. The investigation of the antibody conformational changes upon antigen binding allowed us to identify sections on variable and constant regions that show significant flexibility when comparing the antigen bound/unbound forms of immunoglobulins. The results of all the above-mentioned analyses have been implemented in our in-house immunoglobulin structure prediction server (PIGS, automatic Prediction of ImmunoGlobulin Structure), thus helping to minimize the sources of errors in the current modelling process. Consequent to our results, we were asked to write a chapter in Encyclopaedia of Biophysics on antibody modelling [3]. A further step in the direction of improving the understanding of antibody recognition mechanisms was to put together all the annotations of immunoglobulins in a publicly available database. To this aim, we constructed a database of immunoglobulin sequences and integrated tools (DIGIT) [4], which is becoming an extensively used resource by the community. DIGIT stores sequences of annotated immunoglobulin variable domains and offers to the user several tools for searching and analysing them. Our experience in antibody modelling allowed us to approach two biomedical problems in collaboration with Prof. Arcaini (University of Pavia) and Prof. Fabio Ghiotto (University of Genova). More specifically, by applying the tools we developed and all our theoretical knowledge we successfully analysed the immunoglobulin repertoires of SMZL (splenic marginal zone lymphoma) and CLL (chronic lymphocytic leukaemia) patient data. Both the CLL and SMZL patients are known to have a biased usage of immunoglobulin (IG) heavy variable (IGHV) genes and stereotyped B-cell receptors (BCRs), used as a marker in disease prognosis. We extended these analyses by taking into account VL germlines, VL-VH pairing and structural information, thus giving a more detailed view of the immunoglobulin repertoire in terms of sequence, structure and function. Analysing the immunoglobulins of patients with CLL, we discovered statistically significant differences among immunoglobulins in patients with favourable and unfavourable prognosis. A paper describing this work has been submitted [5]. The poster describing the results of SMZL repertoire analysis was accepted at the 2012 American Society of Haematology (ASH) meeting and published as an abstract [6]. Reference: 1. Chailyan, A., P. Marcatili, et al. (2011). "Structural repertoire of immunoglobulin lambda light chains." Proteins 79(5): 1513-1524. 2. Chailyan, A., P. Marcatili, et al. (2011). "The association of heavy and light chain variable domains in antibodies: implications for antigen specificity." FEBS J 278(16): 2858-2866. 3. Marcatili P., A. Chailyan, D. Cirillo and A. Tramontano. Modelling of antibody structures. Encyclopaedia of Biophysics. Springer (2012). 4. Chailyan, A., A. Tramontano, et al. (2012). "A database of immunoglobulins with integrated tools: DIGIT." Nucleic Acids Res. doi:10.1093/nar/gkr806. 5. Marcatili P., F. Ghiotto, C. Tenca, A. Chailyan, A. N. Mazzarello, X. Yan, M. Colombo, E. Albesiano, D. Bagnara, G. Cutrona, F. Morabito, S. Bruno, M. Ferrarini, N. Chiorazzi, A. Tramontano, F. Fais. "Immunoglobulins produced by chronic lymphocytic leukaemia B cells show limited binding site structure variability." submitted 6. Marcatili P., S. Zibellini, S. Rattotti, A. Chailyan, M. Varettoni, L. Morello, E. Boveri, M. Lucioni, M. Bonfichi, M. Gotti, V. Fiaccadori, M. Paulli, A. Tramontano, L. Arcaini. "Hierarchical Clustering of B-Cell Receptor Structures in Splenic Marginal Zone Lymphoma", abstract, American Society of Haematology (ASH) meeting.

Background Although the etiology of chronic lymphocytic leukemia (CLL), the most common type of adult leukemia, is still unclear, strong evidence implicates antigen involvement in disease ontogeny and evolution. Primary and 3D structure analysis has been utilised in order to discover indications of antigenic pressure. The latter has been mostly based on the 3D models of the clonotypic B cell receptor immunoglobulin (BcR IG) amino acid sequences. Therefore, their accuracy is directly dependent on the quality of the model construction algorithms and the specific methods used to compare the ensuing models. Thus far, reliable and robust methods that can group the IG 3D models based on their structural characteristics are missing. Results Here we propose a novel method for clustering a set of proteins based on their 3D structure focusing on 3D structures of BcR IG from a large series of patients with CLL. The method combines techniques from the areas of bioinformatics, 3D object recognition and machine learning. The clustering procedure is based on the extraction of 3D descriptors, encoding various properties of the local and global geometrical structure of the proteins. The descriptors are extracted from aligned pairs of proteins. A combination of individual 3D descriptors is also used as an additional method. The comparison of the automatically generated clusters to manual annotation by experts shows an increased accuracy when using the 3D descriptors compared to plain bioinformatics-based comparison. The accuracy is increased even more when using the combination of 3D descriptors. Conclusions The experimental results verify that the use of 3D descriptors commonly used for 3D object recognition can be effectively applied to distinguishing structural differences of proteins. The proposed approach can be applied to provide hints for the existence of structural groups in a large set of unannotated BcR IG protein files in both CLL and, by logical extension, other contexts where it is relevant to characterize BcR IG structural similarity. The method does not present any limitations in application and can be extended to other types of proteins.