(M1230-07-42) Investigating Protein-Excipient Interactions of a Multivalent V_HH Therapeutic Protein using NMR Spectroscopy

Purpose: Multispecific therapeutic proteins come in variety of formats such as bi- and tri-specific antibodies, dual-variable domain antibodies, cross-mAbs and similar constructs. These multivalent proteins are engineered to interact with multiple therapeutic target proteins with high specificity. The complexity of solution interactions may translate into developability and manufacturing challenges. In this work, we use Nuclear Magnetic Resonance (NMR) spectroscopy to study the solution behavior of multivalent V_HH molecule, comprised of three flexibly linked heavy-chain only domains that show dramatic stabilization against thermal degradation in the presence of sucrose. We provide a framework to characterize and begin to understand the role of molecular flexibility in protein stabilization with potential applications in design of novel therapeutic protein scaffolds that include multivalent proteins, fusion proteins, antibody-drug conjugates and proteins modified with flexible lipids.

Methods: The MV-V_HH protein was recombinantly produced in-house using mammalian expression system and standard purification train. MV-V_HH was dialyzed overnight into 10mM acetate buffer at pH between 4.6 and 5.5 with 0, 4, 8, 12, 16, or 20% sucrose. Samples were filtered using 0.22 µm filter (EMD Millipore Burlington, MA) and used for NMR analysis. Stressed stability Study:  A 20 mg/mL sample of the MV-V_HH protein with different levels of sucrose was exposed to stress temperature of 50°C for 10 days and corresponding unstressed controls were stored at 5°C for the duration of the study. Profiling and Structural Fingerprinting: A simple ¹H NMR spectrum of formulated proteins provides a rapid, low-resolution assessment of protein structural changes.  Initial data consisted of a protein profile that uses a diffusion filter to reduce the NMR signal from sucrose and buffer molecules. Protein-enhanced DOSY Pulsed magnetic field gradient NMR experiments were used to measure the translational diffusion of multiple molecules in complicated mixtures. A modified protein-enhanced DOSY experiment was made by starting the initial gradient magnetic field at 30% which efficiently reduced the excipient and water signals leaving mostly the protein DOSY NMR signal.  CPMG Proton R2: Proton relaxation measurements were used to probe molecular motions ranging from nanosecond to millisecond timescales. A conventional CPMG proton-R2 measurement (effective v_CPMG = 500  s^-1) was modified with a diffusion filter to produce protein-enhanced R2 measurement that reduces the 1H NMR signals from all non-protein buffer components.

Results: A stress stability study (50°C) was carried out to evaluate the impact of increase in level of sucrose towards the molecular stability. The results from the stress (50°C) stability were compared to a 5°C control. Increase in sucrose content resulted in decrease in the HMW formation upon stress (Fig. 1). In the absence of sucrose, the ∆HMW was highest at 18%. Concentration dependent decrease in the ∆HMW of 3.8% was observed at 20% sucrose concentration. In Fig 2A, MV-V_HH protein spectra were compared with 0, 4, 8, 12, 16 and 20% sucrose. Experimental data was collected at 27°C, 10 days after the samples were prepared and stored at 5°C (unstressed). Site-specific binding in these experiments would be detected as peak shifts or intensity variations. The identical traces in Fig 3A indicate that there is no detectable protein self-interactions or conformational change due to sucrose addition. Intensity variations were, however, detected in MV-V_HH spectra (Fig. 2B) collected on samples with 0, 4, 8, 12, 16 and 20% sucrose prepared and stored at 50°C (stress) for 10 days prior to data collection at 27°C. R2 was measured for sidechain methyl (˜) and backbone amides (▼) found in the core of the folded aC and aD domains, as well as linker amides (¢) (Fig. 3A). While anticipated trends of increasing aC and aD R2s with increasing sucrose were observed, two unanticipated results were obtained. First, an inflection point is reached at ~10% sucrose where the R2s originating from aC and aD atoms starts increasing more dramatically. At 0% sucrose, the linker R2 in Figure 3A is consistent with greater flexibility than the methyl and aC and aD amide R2s, as expected.

Conclusion: Key insights were revealed in studies of MV-V_HH dynamic behavior. Sucrose-dependent changes in protein dynamics were measured by R2 relaxation studies in which linker and V_HH atoms could easily be distinguished. Protein R2 measurements indicate a reduction of slow V_HH domain motions expected from viscosity alone. L₁-linker dynamics were profoundly different than folded domain motions; samples with higher amounts of sucrose showed faster linker dynamics, clearly trending in an opposite direction expected from measured solution viscosity. Studies like these can be valuable to understand details of protein dynamics and protein-excipient interactions in a wide range of bioconjugated molecules that include antibody-drug-conjugates and an ever-widening array of multivalent, multispecific proteins where linker behavior will affect overall stability.

High molecular weight species generation upon thermal stress of MV-VHH at 50°C for 10 days. Sucrose concentration ranges from 0 to 20%. The difference between stressed and unstressed samples is plotted as the ∆HMW



Effect of sucrose on MV-VHH protein behavior from NMR fingerprinting. A) Overlay of 1H NMR spectra of MV-VHH with 0 (black) , 4 (magenta), 8 (green), 12 (orange), 16 (blue) and 20 % (red) sucrose show that NMR signals from the MV-VHH protein are not affected by sucrose addition with no thermal stress. B) Overlay of 1H NMR spectra of MV-VHH with 0, 4, 8, 12, 16 and 20 % sucrose showing loss of NMR signal following thermal stress, consistent with monomer lost in forming high MW species (from Fig. 1). C) Variation of NMR signal intensities from A (gray) and B (blue). Intensities for unstressed samples vary by 1%, while thermal stress reduces the NMR signal by more than 20% for samples without sucrose. D) Overlay of the MV-VHH 1H, 13C sfHMQC data with 0% (black) and 20% (red) sucrose with no stress. No direct protein-sucrose interaction was detected using 2D NMR fingerprints. No changes in B or C suggest the sucrose-depended MV-VHH stabilization is not a result of site-specific binding and does not result in detectable changes in structure or conformation.



Proton R2 relaxation data reveals changes in motion as a function of sucrose for methyl groups, aC/aD amides, and L1 amides. A) 1H R2 for different regions of the protein show distinct behavior in response to sucrose. B) R1*R2 and R2/R1 reveal anisotropic behavior for different regions of the MV-VHH. Arrows show the direction of increasing sucrose concentration for each region. Dramatically different behavior is observed for the linker amides.