(566e) Tuning the Biophysical Properties of an IgG1-Based Antibody-Drug Conjugate By Traversing the Formulation Design Space
AIChE Annual Meeting
Wednesday, November 16, 2022 - 4:12pm to 4:33pm
While ADCs offer a promising approach for treating a wide variety of diseases, they also present challenges related to structural stability. In particular, the attachment of payload to antibodies is generally achieved via modification of chemically reactive residues. One common strategy for this conjugation consists of partially reducing the antibodyâs interchain disulfide bonds to yield reactive free cysteines that can react with the linker/drug moiety. However, interchain disulfide bonds play an important role in forming the quaternary structure of antibodies, and therefore reducing these bonds can negatively impact the overall stability of the structure. Moreover, in the case of small molecule payloads, their attachment may introduce an additional hydrophobic moiety that can promote increased self-interaction and aggregation.
This talk will explore our recent work studying an ADC composed of an IgG1 antibody conjugated to an auristatin-based small molecule. The small molecule was attached at cysteine residues of the antibody to yield conjugates with an average drug loading of four as quantified by hydrophobic interaction chromatography. This talk will focus on and discuss the differential behavior observed of the ADC compared to the naked (unconjugated) mAb towards gaining fundamental understanding of how conjugation of surface payload drives structural changes that manifest as differences in product attributes.
One important property of antibody formulations is their degree of opalescence (turbidity) in solution. High opalescence can signal the presence of reversible or irreversible aggregates that impact therapeutic efficacy or elicit in vivo toxicity. As a preliminary experiment, the naked mAb and ADC were formulated in citrate buffer to assess their turbidity. In contrast to the naked mAb which exhibited opalescence as high as ~20 nephelometric turbidity units (NTU), the ADC exhibited substantially higher opalescence of ~55 NTU when both were formulated at ~55 mg/mL. This observation motivated a systematic study to understand the mechanistic underpinning of this phenomenon. A high-throughput screen across ~100 formulation conditions demonstrated that varying solution properties such as pH, ionic strength, trehalose concentration, and PS80 concentration led to changes in opalescence of the conjugated mAb. Notably, the largest difference in opalescence was observed under conditions with varying pH. In particular, a significant reduction in turbidity of up to 400% was observed as the solution pH was increased from pH 5.2 to pH 7.0. Because the significantly high opalescence was observed at a pH much less than the pI of the ADC (pI ~8.30 - ~9.10 depending on drug loading), this suggests that electrostatic interactions are one of the primary drivers for the turbidity observed. Interestingly, and in contrast to the ADC, the turbidity of the unconjugated mAb remained unchanged across the design space. Moreover, the ADC exhibited higher turbidity than the unconjugated mAb in all conditions analyzed.
It was hypothesized that the observed differences in solution behavior of the ADC versus the unconjugated mAb was a result of their inherent structural differences. Therefore, the extent of changes in the ADC upon small molecule conjugation was next probed by both differential scanning calorimetry (DSC) and intrinsic fluorescence measurements. Melting temperatures generally correlate with protein stability; increasing stability is usually reflected by higher melting temperatures as the amount of energy required to unfold the protein increases. DSC indicated two melting transitions in both the naked mAb and ADC corresponding to (1). the unfolding of the CH2 domain of the Fc fragment and (2). the unfolding of the unconjugated Fab domain and CH3 domain of the Fc fragment. However, in the case of the ADC, the modification of the protein gave rise to a third melting transition corresponding to unfolding of the drug-conjugated Fab domain. In comparison to the unconjugated Fab region, this domain exhibited an approximately 6°C lower melting temperature. Intrinsic fluorescence measurements were next conducted to make high throughput melting temperature readings across formulations that had previously been used for studying turbidity. While the naked mAb had a melting temperature that was largely invariant to the different conditions tested, the ADC melting temperature was heavily dependent on the physical properties of the solution, including a 6°C increase in melting temperature as the formulation pH was varied from 5.2 to 7.0. Importantly, solution conditions that led to decreases in melting temperature were the same as those that led to higher turbidity, suggesting that structural destabilization plays a key role in dictating the opalescence observed in the ADC. The use of static light scattering experiments revealed that the ADC also has a substantially lower (up to ~10°C) aggregation temperature compared to the naked mAb. Similarly to melting temperature results, the aggregation temperature decreased in formulations that promoted increased turbidity. Taken together, these studies substantiate that the conjugation of the small molecule moiety leads to structural changes that ultimately give rise to less stable formulations.
An investigation was then done to understand whether these structural differences lead to increased levels of interaction between proteins in solution. It was initially observed that turbidity exhibits a nonlinear dependence on ADC concentration. This is especially interesting considering that normal Rayleigh scattering in dilute media should lead to a linear dependence between concentration and turbidity, and implies the presence of protein-protein interactions contributing to the behavior of the ADC. To probe this further, measurements to quantify interaction parameters (Kd) were made using dynamic light scattering. These studies showed that the ADC exhibited negative Kd values that were also significantly lower than the naked mAb. This suggests the presence of a greater extent of attractive protein-protein interactions in the case of the ADC compared to the naked mAb. Moreover, varying buffer composition led to significant changes in Kd, wherein those conditions that previously had given rise to more opalescence also exhibited more negative Kd values. These results illustrate that structural differences between the naked mAb and ADC are manifested into increased attractive protein-protein interactions that lead to enhanced opalescence. To gain further insight into these experimental observations, molecular modeling was carried out. These studies suggested specific structural differences between the ADC and the naked mAb towards proposing a hypothesis for increased protein-protein interactions in the ADC. Taken together, it is anticipated that the experimental and simulation results presented here can lead to learnings applicable to other IgG1-based conjugates where drug loading is predominantly at cysteine residues of the Fab region.