Highly concentrated human recombinant interleukin-1 receptor antagonist (IL-1ra) aggregates at elevated temperature without perturbation in its secondary structure. The protein aggregation can be suppressed depending on the buffer ionic strength and the type of anion present in the sample solution. Phosphate is an approximately 4-fold weaker suppressant than either citrate or pyrophosphate on the basis of the measured protein aggregation rates. This is in agreement with the strength of protein-anion interactions at the IL-1ra single anion-binding site as judged by the estimated dissociation constant values of 2.9 mM, 3.8 mM, and 13.7 mM for pyrophosphate, citrate, and phosphate, respectively. The strength of binding also correlates with the anion size and with the number of ionized groups available per molecule at a given pH. Affinity probing of IL-1ra with methyl acetyl phosphate (MAP) in combination with proteolytic digestion and mass spectral analysis show that an anion-binding site location on the IL-1ra surface is contributed by lysine-93 and lysine-96 of the loop 84-98 as well as by lysine-6 of the unstructured N-terminal region 1-7. The replacement of lysine-93 with alanine by site-directed mutagenesis results in dramatically suppressed IL-1ra aggregation. Furthermore, when the unstructured N-terminal region of IL-1ra is removed by limited proteolysis, a 2-fold increase in the time course of the aggregation lag phase is observed for the truncated protein. An anion-controlled mechanism of IL-1ra aggregation is proposed by which the anion competition for the protein cationic site prevents formation of intermolecular cation-pi interactions and, thus, interferes with the protein asymmetric self-association pathway.

No abstract

We demonstrate the use of a Generative Adversarial Network (GAN), trained from a set of over 400,000 light and heavy chain human antibody sequences, to learn the rules of human antibody formation. The resulting model surpasses common in silico techniques by capturing residue diversity throughout the variable region, and is capable of generating extremely large, diverse libraries of novel antibodies that mimic somatically hypermutated human repertoire response. This method permits us to rationally design de novo humanoid antibody libraries with explicit control over various properties of our discovery library. Through transfer learning, we are able to bias the GAN to generate molecules with key properties of interest such as improved stability and developability, lower predicted MHC Class II binding, and specific complementarity-determining region (CDR) characteristics. These approaches also provide a mechanism to better study the complex relationships between antibody sequence and molecular behavior, both in vitro and in vivo . We validate our method by successfully expressing a proof-of-concept library of nearly 100,000 GAN-generated antibodies via phage display. We present the sequences and homology-model structures of example generated antibodies expressed in stable CHO pools and evaluated across multiple biophysical properties. The creation of discovery libraries using our in silico approach allows for the control of pharmaceutical properties such that these therapeutic antibodies can provide a more rapid and cost-effective response to biological threats.

No abstract

No abstract