Antibody modeling assessment II. Structures and models.

To assess the state-of-the-art in antibody structure modeling, a blinded study was conducted. Eleven unpublished Fab crystal structures were used as a benchmark to compare Fv models generated by seven structure prediction methodologies. In the first round, each participant submitted three non-ranked complete Fv models for each target. In the second round, CDR-H3 modeling was performed in the context of the correct environment provided by the crystal structures with CDR-H3 removed. In this report we describe the reference structures and present our assessment of the models. Some of the essential sources of errors in the predictions were traced to the selection of the structure template, both in terms of the CDR canonical structures and VL/VH packing. On top of this, the errors present in the Protein Data Bank structures were sometimes propagated in the current models, which emphasized the need for the curated structural database devoid of errors. Modeling non-canonical structures, including CDR-H3, remains the biggest challenge for antibody structure prediction.

Second antibody modeling assessment (AMA-II).

To assess the state of the art in antibody 3D modeling, 11 unpublished high-resolution x-ray Fab crystal structures from diverse species and covering a wide range of antigen-binding site conformations were used as a benchmark to compare Fv models generated by seven structure prediction methodologies. The participants included: Accerlys Inc, Chemical Computer Group (CCG), Schrodinger, Jeff Gray's lab at John Hopkins University, Macromoltek, Astellas Pharma/Osaka University and Prediction of ImmunoGlobulin Structure (PIGS). The sequences of benchmark structures were submitted to the modelers and PIGS, and a set of models were generated for each structure. We provide here an overview of the organization, participants and main results of this second antibody modeling assessment (AMA-II). Also, we compare the results with the first antibody assessment published in this journal (Almagro et al., 2011;79:3050).

Anti-CAIX BBζ CAR4/8 T cells exhibit superior efficacy in a ccRCC mouse model.

Improving CAR-T cell therapy for solid tumors requires a better understanding of CAR design and cellular composition. Here, we compared second-generation (BBζ and 28ζ) with third-generation (28BBζ) carbonic anhydrase IX (CAIX)-targeted CAR constructs and investigated the antitumor effect of CAR-T cells with different CD4/CD8 proportions in vitro and in vivo. The results demonstrated that BBζ exhibited superior efficacy compared with 28ζ and 28BBζ CAR-T cells in a clear-cell renal cell carcinoma (ccRCC) skrc-59 cell bearing NSG-SGM3 mouse model. The mice treated with a single dose of BBζ CD4/CD8 mixture (CAR4/8) showed complete tumor remission and remained tumor-free 72 days after CAR-T cells infusion. In the other CAR-T and control groups, tumor-infiltrating T cells were recovered and profiled. We found that BBζ CAR8 cells upregulated expression of major histocompatibility complex (MHC) class II and cytotoxicity-associated genes, while downregulating inhibitory immune checkpoint receptor genes and diminishing differentiation of regulatory T cells (Treg cells), leading to excellent therapeutic efficacy in vivo. Increased memory phenotype, elevated tumor infiltration, and decreased exhaustion genes were observed in the CD4/8 untransduced T (UNT) cells compared with CD8 alone, indicating that CD4/8 would be the favored cellular composition for CAR-T cell therapy with long-term persistence. In summary, these findings support that BBζ CAR4/8 cells are a highly potent, clinically translatable cell therapy for ccRCC.

Identification of critical amino acid residues for human iNOS functional activity.

Nitric oxide (NO) is a short-lived signaling molecule that mediates a variety of biological functions, including vascular homeostasis, neurotransmission, antimicrobial defense and antitumor activities. Three known NOS isoforms (eNOS, nNOS and iNOS) have been cloned and sequenced. Here, we show that upon expression in Escherichia coli using a novel expression vector, an iNOS sequence containing three mutations (A805D, F831S and L832P) within the iNOS reductase domain produced very little functionally active iNOS protein compared to the wild type (wt) iNOS. Each of these point mutations also was individually constructed into the wt iNOS sequence. The activity of the iNOS protein containing the A805D mutation was comparable to wt, while a drastic reduction in iNOS activity was observed for the F831S and L832P mutants. A comparison of the molecular models of the reductase domain of the wt and mutant iNOS revealed a reduced core packing density for the F831S and L832P mutations compared to wt. In addition, the modeling also suggests altered hydrogen bonding, van der Waals and hydrophobic interactions of these mutants.

Characterization of Gpr101 expression and G-protein coupling selectivity.

This report describes the identification and characterization of the murine orphan GPCR, Gpr101. Both human and murine genes were localized to chromosome X. Similar to its human ortholog, murine Gpr101 mRNA was detected predominantly in the brain within discrete nuclei. A knowledge-restricted hidden Markov model-based algorithm, capable of accurately predicting G-protein coupling selectivity, indicated that both human and murine GPR101 were likely coupled to Gs. This prediction was supported by the elevation of cyclic AMP levels and the activation of a cyclic AMP response element-luciferase reporter gene in HEK293 cells over-expressing human GPR101. Consistent with this, over-expression of human GPR101 in a yeast-based system yielded an elevated, agonist-independent reporter gene response in the presence of a yeast chimeric Galphas protein. These results indicate that GPR101 participates in a potentially wide range of activities in the CNS via modulation of cAMP levels.

The functional repertoire of rabbit antibodies and antibody discovery via next-generation sequencing.

To gain insight into the functional antibody repertoire of rabbits, the VH and VL repertoires of bone marrow (BM) and spleen (SP) of a naïve New Zealand White rabbit (NZW; Oryctolagus cuniculus) and that of lymphocytes collected from a NZW rabbit immunized (IM) with a 16-mer peptide were deep-sequenced. Two closely related genes, IGHV1S40 (VH1a3) and IGHV1S45 (VH4), were found to dominate (~90%) the VH repertoire of BM and SP, whereas, IGHV1S69 (VH1a1) contributed significantly (~40%) to IM. BM and SP antibodies recombined predominantly with IGHJ4. A significant proportion (~30%) of IM sequences recombined with IGHJ2. The VK repertoire was encoded by nine IGKV genes recombined with one IGKJ gene, IGKJ1. No significant bias in the VK repertoire of the BM, SP and IM samples was observed. The complementarity-determining region (CDR)-H3 and -L3 length distributions were similar in the three samples following a Gaussian curve with average length of 12.2 ± 2.4 and 11.1 ± 1.1 amino acids, respectively. The amino acid composition of the predominant CDR-H3 and -L3 loop lengths was similar to that of humans and mice, rich in Tyr, Gly, Ser and, in some specific positions, Asp. The average number of mutations along the IGHV/KV genes was similar in BM, SP and IM; close to 12 and 15 mutations for VH and VL, respectively. A monoclonal antibody specific for the peptide used as immunogen was obtained from the IM rabbit. The CDR-H3 sequence was found in 1,559 of 61,728 (2.5%) sequences, at position 10, in the rank order of the CDR-H3 frequencies. The CDR-L3 was found in 24 of 11,215 (0.2%) sequences, ranking 102. No match was found in the BM and SP samples, indicating positive selection for the hybridoma sequence. Altogether, these findings lay foundations for engineering of rabbit V regions to enhance their potential as therapeutics, i.e., design of strategies for selection of specific rabbit V regions from NGS data mining, humanization and design of libraries for affinity maturation campaigns.

Affinity maturation of a humanized rat antibody for anti-RAGE therapy: comprehensive mutagenesis reveals a high level of mutational plasticity both inside and outside the complementarity-determining regions.

Antibodies that neutralize RAGE (receptor for advanced glycation end products)-ligand interactions have potential therapeutic applications in both acute and chronic diseases. We generated XT-M4, a rat anti-RAGE monoclonal antibody that has in vivo efficacy in an acute sepsis model. This antibody was subsequently humanized. To improve the affinity of this antibody for the treatment of chronic indications, we used random and targeted mutagenesis strategies in combination with ribosome and phage-display technologies, respectively, to generate libraries of XT-M4 variants. We identified a panel of single-chain Fv antibody fragments (scFv's) that was improved up to 110-fold in a homogeneous time-resolved fluorescence competition assay against parental XT-M4 immunoglobulin G (IgG). After reformatting to bivalent scFv-Fc fusions and IgGs, we observed similar gains in potency in the same assay. Further analysis of binding kinetics as IgG revealed multiple variants with subnanomolar apparent affinity that was dictated primarily by improvements in the off-rate. All variants also had improved binding to cell surface-expressed human RAGE, and all retained, or had improved, apparent affinity for mouse RAGE. F100bL in V(H) (variable region of the heavy chain) complementarity-determining region 3 (CDR3) was one of a number of key mutations that correlated with affinity improvements and was independently identified by both mutagenesis strategies. Random mutagenesis coupled with ribosome display and high-throughput screening revealed an unexpectedly high level of mutational plasticity across the whole length of the humanized scFv, suggesting greater scope for structural optimization outside of the primary antigen-combining site defined by V(H) CDR3 and V(kappa) CDR3. In summary, our comprehensive mutagenesis approach not only achieved the desired affinity maturation of XT-M4 but also defined multiple mutational hotspots across the antibody sequence, provided an insight into the specificity-determining residues of the antibody paratope, and identified additional sites within the CDR loops where human germ-line amino acids may be introduced without affecting function.

IL-22R, IL-10R2, and IL-22BP binding sites are topologically juxtaposed on adjacent and overlapping surfaces of IL-22.

Interleukin (IL) 22 is a type II cytokine that is produced by immune cells and acts on nonimmune cells to regulate local tissue inflammation. As a product of the recently identified T helper 17 lineage of CD4(+) effector lymphocytes, IL-22 plays a critical role in mucosal immunity as well as in dysregulated inflammation observed in autoimmune diseases. We used comprehensive mutagenesis combined with mammalian cell expression, ELISA cell-based, and structural methods to evaluate how IL-22 interacts with its cell surface receptor, IL-22R/IL-10R2, and with secreted IL-22 binding protein. This study identifies those amino acid side chains of IL-22 that are individually important for optimal binding to IL-22R, considerably expands the definition of IL-22 surface required for binding to IL-10R2, and demonstrates how IL-22 binding protein prevents IL-22R from binding to IL-22. The IL-22R and IL-10R2 binding sites are juxtaposed on adjacent IL-22 surfaces contributed mostly by helices A, D, and F and loop AB. Our results also provide a model for how IL-19, IL-20, IL-24, and IL-26 which are other IL-10-like cytokines, interact with their respective cell surface receptors.