Process development of a FGF21 protein-antibody conjugate.

A scalable, viable process was developed for the Fibroblast Growth Factor 21 (FGF21) protein-antibody conjugate, CVX-343, an extended half-life therapeutic for the treatment of metabolic disease. CVX-343 utilizes the CovX antibody scaffold technology platform that was specifically developed for peptide and protein half-life extension. CVX-343 is representative of a growing number of complex novel peptide- and protein-based bioconjugate molecules currently being explored as therapeutic candidates. The complexity of these bioconjugates, assembled using well-established chemistries, can lead to very difficult production schemes requiring multiple starting materials and a combination of diverse technologies. Key improvements had to be made to the original CVX-343 Phase 1 manufacturing process in preparation for Phase 3 and commercial manufacturing. A strategy of minimizing FGF21A129C dimerization and stabilizing the FGF21A129C Drug Substance Intermediate (DSI), linker, and activated FGF21 intermediate was pursued. The use of tris(2-carboxyethyl)phosphine (TCEP) to prevent FGF21A129C dimerization through disulfide formation was eliminated. FGF21A129C dimerization and linker hydrolysis were minimized by formulating and activating FGF21A129C at acidic instead of neutral pH. An activation use test was utilized to guide FGF21A129C pooling in order to minimize misfolds, dimers, and misfolded dimers in the FGF21A129C DSI. After final optimization of reaction conditions, a process was established that reduced the consumption of FGF21A129C by 36% (from 4.7 to 3.0 equivalents) and the consumption of linker by 55% (from 1.4 to 0.95 equivalents for a smaller required amount of FGF21A129C ). The overall process time was reduced from ∼5 to ∼3 days. The product distribution improved from containing ∼60% to ∼75% desired bifunctionalized (+2 FGF21) FGF21-antibody conjugate in the crude conjugation mixture and from ∼80% to ∼85% in the final CVX-343 Drug Substance (DS), while maintaining the same overall process yield based on antibody scaffold input.

Systematic Investigation of EDC/sNHS-Mediated Bioconjugation Reactions for Carboxylated Peptide Substrates.

1-Ethyl-3-(3-(dimethylamino)propyl)carbodiimide (EDC) bioconjugations have been utilized in preparing variants for medical research. While there have been advances in optimizing the reaction for aqueous applications, there has been limited focus toward identifying conditions and side reactions that interfere with product formation. We present a systematic investigation of EDC/N-hydroxysulfosuccinimide (sNHS)-mediated bioconjugations on carboxylated peptides and small proteins. We identified yet-to-be-reported side products arising from both the reagents and substrates. Model peptides used in this study illustrate particular substrates are more susceptible to side reactions than others. From our studies, we found that bioconjugations are more efficient with high concentrations of amine nucleophile but not sNHS. Performing bioconjugations on a model affibody protein show that the trends established with model peptides hold for more complex systems.

Novel approach for optimization of a 'difficult' peptide synthesis by utilizing quantitative reaction monitoring assays.

Three different synthetic strategies were tested to synthesize EGFRvIII peptide (1), which is a functional variant of the epidermal growth factor receptor, a protein that has been well validated as a target for cancer therapy. The initial synthesis was performed with Applied Biosystem (Carlsbad, CA, USA) 433A peptide synthesizer and it indicated that the last three amino acids coupling and Fmoc removal rates were lower than the rest of the sequence. Purity of the crude peptide was 54.5%. The second synthesis was performed manually utilizing C S Bio (Menlo Park, CA, USA) synthesizer and the Kaiser test for reaction monitoring. Because of non-optimized reaction conditions and an unexpected by-product, lower purity crude peptide (40.5%) was obtained. Quantitative assays to monitor reactions were developed and demonstrated in gram scale synthesis with C S Bio synthesizer. The optimized synthetic conditions improved the peptide purity to 68.1%.

Anti-nicotine vaccines: Comparison of adjuvanted CRM197 and Qb-VLP conjugate formulations for immunogenicity and function in non-human primates.

Anti-nicotine vaccines comprise nicotine-like haptens conjugated to a carrier protein plus adjuvant(s). Unfortunately, those tested clinically have failed to improve overall long term quit rates. We had shown in mice that carrier, hapten, linker, hapten load (number of haptens per carrier molecule), aggregation and adducts, as well as adjuvants influence the function of antibodies (Ab) induced. Herein, we tested an optimized antigen, NIC7-CRM, comprised of 5-aminoethoxy-nicotine (NIC7) conjugated to genetically detoxified diphtheria toxin (CRM197), with hapten load of ~16, no aggregation (~100% monomer) and minimal adducts. NIC7-CRM was tested in non-human primates (NHP) and compared to NIC-VLP, which has the same hapten and carrier as the clinical-stage CYT002-NicQb but a slightly different linker and lower hapten load. With alum as sole adjuvant, NIC7-CRM was superior to NIC-VLP for Ab titer, avidity and ex vivo function (83% and 27% nicotine binding at 40ng/mL respectively), but equivalent for in vivo function after intravenous [IV] nicotine challenge (brain levels reduced ~10%). CpG adjuvant added to NIC7-CRM/alum further enhanced the Ab responses and both ex vivo function (100% bound) and in vivo function (~80% reduction in brain). Thus, both optimal antigen design and CpG adjuvant were required to achieve a highly functional vaccine. The compelling NHP data with NIC7-CRM with alum/CpG supported human testing, currently underway.

Molecular attributes of conjugate antigen influence function of antibodies induced by anti-nicotine vaccine in mice and non-human primates.

Anti-nicotine vaccines aim to prevent nicotine entering the brain, and thus reduce or eliminate the reward that drives nicotine addiction. Those tested in humans to date have failed to improve quit rates over placebo, possibly because antibody (Ab) responses were insufficient to sequester enough nicotine in the blood in the majority of subjects. We have previously shown in mice that the carrier, hapten and linker used in the nicotine conjugate antigen each influence the function (nicotine-binding capacity) of the Ab induced. Herein we have evaluated immunogenicity in mice of 27 lots of NIC7-CRM, a conjugate of 5-aminoethoxy-nicotine (Hapten 7) and a mutant nontoxic form of diphtheria toxin (CRM197), that differed in three antigen attributes, namely hapten load (number of haptens conjugated to each molecule of CRM197), degree of conjugate aggregation and presence of adducts (small molecules attached to CRM197 via a covalent bond during the conjugation process). A range of functional responses (reduced nicotine in the brain of immunized animals relative to non-immunized controls) were obtained with the different conjugates, which were adjuvanted with aluminum hydroxide and CpG TLR9 agonist. Trends for better functional responses in mice were obtained with conjugates having a hapten load of 11 to 18, a low level of high molecular mass species (HMMS) (i.e., not aggregated) and a low level of adducts and a more limited testing in cynomolgus monkeys confirmed these results. Thus hapten load, conjugate aggregation and presence of adducts are key antigen attributes that can influence Ab function induced by NIC7-CRM.