by Yuko Tsuchiya Institute for Protein Research, Osaka University
Kenji Mizuguchi National Institutes of Biomedical Innovation, Health and Nutrition
The complementary determining regions (CDRs) of antibodies play a key role in antigen recognition. The contribution of each of the six CDR loops to antigen recognition is different from each other, and even within a single CDR loop, each residue position plays a different role in antigen binding [1, 2]. Therefore, it is necessary to characterize the sequence and structural properties of each position in a CDR loop and understand how different residue positions are utilized for antigen binding.
The H3 loop is the main contributor to antigen recognition among the six CDR loops, because of its sequence diversity and location favorable to antigen binding [3, 4]. The sequence diversity produces diverse conformations, particularly in long H3 loops, and the conformational variety may be required for maintaining antigen specificity and H3’s predominant role in antigen binding.
A better understanding of the effect of diverse H3 loop conformations on antigen binding will be of use to antibody design and affinity maturation. Several studies have revealed sequence-structure relationships, particularly in the stem region of the H3 loops, and classified them into two groups, kinked and extended [5-8]. However, non-stem regions (particularly in long H3 loops) are crucial as main antigen-binding sites and yet, their structures have not been fully characterized because of their diversity . Thus, a novel method will be required for describing non-stem conformations from new perspectives.
Here, we summarize new findings from our recent study  based on a non-redundant set of antibody-antigen complex structures in the following areas: (1) the antigen-binding tendency of each position in the six CDR loops, (2) the effect of H3 loop lengths on the antigen-recognition properties of all the CDR loops, and (3) the relationship between the diverse conformations of long H3 loops and antigen recognition.
1. The antigen-binding tendency of each position in the six CDR loops
To identify the antigen-binding ability of a CDR position, we named the residue positions systematically (e.g., position 1 of H1 and position 3 of H2), based on a structure-based alignment of CDR loops in a non-redundant set of antibody-antigen complexes. Since the CDR loops vary considerably in length and conformation even within a single loop type, we distinguished structurally-definable positions (well-aligned columns in a structure-based alignment, 68 positions) from undefinable positions (27 positions). The undefinable positions are located at the tip of the β-hairpin, around the middle of the loop.
Some of the structurally-definable positions are occupied by only one or a few amino acid types. These sequence-conserved positions are infrequently used in antigen binding (36 positions); the fraction of antibodies in our dataset, in which the residues in these positions are involved in antigen binding, is 0.062 in H3 loops. Most of these positions are located in the stem regions of the loops. On the other hand, structurally-definable and sequence-non-conserved positions (32 positions) tend to be used in antigen binding (the fraction of antigen-binding residues is 50% in H3), which are located at the “very high” (observed mainly in H2) or “high and central” (in H3) positions, relative to the geometric center of the CDR regions, and are mostly in the stem regions.
The H3 loops that consist of more than or equal to 14 residues contain structurally-undefinable positions in the middle of the loops. The locations of these positions are broadly distributed in the CDR space, most of which are located in very high positions compared to the structurally-definable positions. The fraction of antigen-binding residues in the structurally-undefinable positions is 67%, indicating that the structurally-undefinable positions are even more frequently utilized for antigen binding than the structurally-definable positions, particularly in long H3 loops.
We constructed a simple method for distinguishing antigen-binding from non antigen-binding positions, where we assume that 1) structurally-definable and sequence-conserved positions are not involved in antigen binding, 2) structurally-definable and sequence-non-conserved positions are involved in antigen binding, and 3) structurally-undefinable positions in H3 and L1 are antigen-binding sites, while those in L3 are non antigen-binding sites, which are based on our observations of CDR loop conformations.
This method distinguished the residues in all the CDR loops correctly with 72% accuracy, which is better than that in a previous study with a similar rule-based method (Paratome [11, 12]; 41% accuracy). Thus, probable antigen-binding positions can be identified by simple sequence and structural features.
2. The effect of H3 loop lengths on the antigen-recognition properties of all the CDR loops
In H3 loops, the positions utilized in antigen binding vary depending on the loop lengths; short H3 loops utilize structurally-definable and mainly sequence-non-conserved positions, while long H3 loops tend to use structurally-undefinable positions. The H3 loop lengths also affect the contributions of the other CDR loops to antigen binding. In antibodies with short H3 loops, most of the six CDR loops participate in antigen binding, while in those with long H3 loops, H3 loops are the main contributor to antigen binding. This observation is consistent with the shapes of the CDR surfaces; in antibodies with short H3 loops, the CDR surfaces are concave, where all the six CDR loops can contact the antigen. On the other hand, the CDR surfaces in antibodies with long H3 loops are convex, where mainly the H3 loops bind to the antigens and the degree of convexity of the CDR surface depends on the H3 loop conformation. These observations suggest that in antibodies with long H3 loops - these H3 loops, particularly their structurally-definable and sequence-non-conserved or structurally-undefinable positions, are the main contributors to antigen binding among all the CDR loops.
3. The relationship between the diverse conformations of long H3 loops and antigen recognition
The broad distribution of the spatial locations of structurally-undefinable positions in H3 loops come from the diversity of loop conformations in the non-stem regions. From the observations of the conformations of long H3 loops, we found that a ladder-like “straight” β-sheet conformation, as in 3vg9 in Figure 1, was a minority (only 12 of the 72 long H3 loops in the dataset). Also, many long H3 loops form non-straight conformations, such as the conformations that broaden in the non-stem regions, as in 2j4w (the “broad” conformation, 22 antibodies), or those with no broadening and are bent towards H1 or L3 loops, as in 1osp (the “bent” conformation, 25 antibodies).
The CDR loops start with a β-ladder structure with regular intra-loop hydrogen bonding patterns, but if the hydrogen bond (HB) in a specific position is broken, the ladder cannot extend and it may result in forming various loop conformations, particularly in long loops, such as bent or broad. We found that HB breaks between main-chain atoms in positions 4 and N-3 in H3 loops. This is an important factor to prevent an extension of the β-ladder. The numbers of these HBs highly correlate with the observed long H3 loop conformations; all the long H3 loops with two main-chain HBs between positions 4 and N-3 form a straight conformation (as in 3vg9), while H3 loops with one or no HBs tend to form bent (as in 1osp) or broad (2j4w) conformations, respectively.
Moreover, long H3 loops tend to form non-conserved inter-CDR loop HBs in non-stem regions, which may result in increasing the structural diversity of long H3 loops (Figure 1). These inter-loop HBs are located close to the residues involved in antibody-antigen HBs, and this observation applies to almost all antibodies with non-conserved inter-loop HBs; as shown in Figure 1, the inter-loop HBs exist at the edges of the antibody-antigen HB networks in many cases so as to facilitate the formation of antibody-antigen HBs. Moreover, these inter-loop HBs form a larger HB network including the antibody-antigen HBs and may contribute to increasing the antibody’s affinity for the antigen.
In conclusion, the above described findings suggest that in designing higher-affinity antibodies, we should consider the H3 loop length and (if possible) conformation in determining which positions to introduce mutations. The majority of long H3 loops adopt non-straight conformations that increase the antigen specificities.
About the Authors
Dr. Yuko Tsuchiya is an Assistant Professor at Institute for Protein Research, Osaka University, and is also a visiting researcher at National Institutes of Biomedical Innovation, Health and Nutrition.
Dr. Kenji Mizuguchi is a research group head at National Institutes of Biomedical Innovation, Health and Nutrition and also a guest professor at Graduate School of Frontier Biosciences and Graduate School of Pharmaceutical Sciences, Osaka University. His research focuses on computational drug discovery and development.
- Padlan EA, Abergel C, Tipper JP. Identification of specificity-determining residues in antibodies. FASEB J 1995; 9:133-9.
- Kunik V, Ofran Y. The indistinguishability of epitopes from protein surface is explained by the distinct binding preferences of each of the six antigen-binding loops. Protein Eng Des Sel 2013; 26:599-609.
- Chothia C, Lesk AM, Tramontano A, et al. Conformations of immunoglobulin hypervariable regions. Nature 1989; 342:877-83.
- Al-Lazikani B, Lesk AM, Chothia C. Standard conformations for the canonical structures of immunoglobulins. J Mol Biol 1997; 273:927-48.
- Shirai H, Kidera A, Nakamura H. Structural classification of CDR-H3 in antibodies. FEBS Lett 1996; 399:1-8.
- Morea V, Tramontano A, Rustici M, et al. Conformations of the third hypervariable region in the VH domain of immunoglobulins. J Mol Biol 1998; 275:269-94.
- Shirai H, Kidera A, Nakamura H. H3-rules: identification of CDR-H3 structures in antibodies. FEBS Lett 1999; 455:188-97.
- Kuroda D, Shirai H, Kobori M, et al. Structural classification of CDR-H3 revisited: a lesson in antibody modeling. Proteins 2008; 73:608-20.
- North B, Lehmann A, Dunbrack RL Jr. A new clustering of antibody CDR loop conformations. J Mol Biol 2011; 406:228-56.
- Tsuchiya Y, Mizuguchi K. The diversity of H3 loops determines the antigen-binding tendencies of antibody CDR loops. Protein Sci 2016; 25:815-25.
- Kunik V, Ashkenazi S, Ofran Y. Paratome: an online tool for systematic identification of antigen-binding regions in antibodies based on sequence or structure. Nucleic Acids Res 2012; 40:W521-524.
- Kunik V, Peters B, Ofran Y. Structural consensus among antibodies defines the antigen binding site. PLoS Comput Biol 2012; 8:e1002388.