Impact of Urea on Monoclonal Antibodies: Multiple Destabilization and Aggregation Effects for Therapeutic Immunoglobulin G Proteins

Abstract

We performed nano differential scanning fluorimetry (nanoDSF) measurements of immunoglobulin G (IgG) in urea gradient solutions under thermal unfolding. Our results show that the denaturing effect of urea on individual IgG domains can be monitored via a linear mapping of thermal shift curves to the corresponding urea concentrations. Assignment of IgG domains to each thermal shift curve allows for a reliable differentiation of the underlying mechanisms. Further results show a decisive influence of salt-induced electrostatic screening effects. We are able to explain all findings by preferential binding mechanisms in combination with electrostatic effects. The results of our study shed more light on the complex interaction mechanisms between buffer solutions and complex proteins, which are important for improving the shelf life of protein therapeutic formulation.

Introduction

Cosolute-induced stabilization and destabilization effects on biological macromolecules like proteins or DNA are the focus of recent experimental and computational research.¹⁻⁴ In recent years, advanced thermodynamic arguments and descriptions were developed in order to rationalize the reversible conformational changes of the macromolecules in the presence of cosolutes.^2,3,5−10 Despite controversial discussions, it is nowadays a consensus that preferential binding and exclusion mechanisms as derived from molecular theories of solutions provide a straightforward explanation of the observed effects.^2,4,11 In more detail, protein destabilizers like urea or guanidinium hydrochloride are known to accumulate in close vicinity around macromolecular surfaces in terms of a preferential binding behavior, whereas protein stabilizers like trimethylamine-N-oxide or ectoine are preferentially excluded to higher-order solvation shells and the bulk solution. A possible reason for this behavior was recently provided in agreement with the underlying combination of electronic properties related to the macromolecule, the solvent, and the cosolute.^4,12 As can be shown in terms of extended Kirkwood–Buff theories of solutions,^{2,9,11,13−16} the binding behavior of the cosolute affects the chemical potential of the macromolecule, which rationalizes the changes in the free energy difference between the unfolded and the native state.^11,17

Hence, destabilizers shift the chemical equilibrium to the denatured state, while stabilizers increase the free energy difference such that the native state is stabilized.^2,11 One can thus conclude that stabilization and destabilization effects are closely related to the binding or exclusion behavior of the cosolute in front of macromolecular surfaces.^2,11,17 Whereas these effects were often studied for simple proteins and peptides, less is known about the influence of cosolutes on complex proteins like monoclonal antibodies (mAbs) with a multitude of different folded and unfolded states. Interestingly, understanding these effects is of far greater relevance for such molecules, since mAbs are an important class of modern drugs that are often affected by stability problems in terms of reduced shelf lives.¹⁸⁻²⁰ Thus, it is well-known that mAbs in buffer solution are prone to aggregation effects, which lead to insoluble protein complexes. The occurrence of these protein complexes in human bodies is responsible for the occurrence of adverse events like immunogenicity.²¹ Therefore, a more detailed understanding of the stabilities of mAbs such as immunoglobulin G (IgG) in buffer solutions containing multiple cosolutes is of paramount importance to reduce such undesirable effects.

Noteworthily, the direct monitoring of structural changes and aggregation effects of multidomain IgGs is challenging. Spectral techniques often face issues due to the convolution of individually folded multiple domains during the measurements. In contrast, indirect measurements like differential scanning calorimetry with the determination of melting points T_m are widely used methods but do not reveal that high level of resolution.^22,23 As was shown in previous studies, the IgGs undergo multiple transition steps, followed by an irreversible aggregation step. Such transition steps include multiple regions in the monoclonal antibody,^24,25 which reveals some difficulties to individually detect and quantify the stability change for convoluted multistate domains.²⁶⁻²⁸ In addition, protein conformational changes can also be monitored by thermal unfolding curves in the presence of denaturing cosolutes in terms of “thermal shift” conditions.²⁹⁻³² A promising alternative that combines indirect and direct measurements is intrinsic fluorescence (IF) differential scanning fluorimetry (DSF). IFDSF, also called nanoDSF, provides insight into structural changes by measuring the protein’s emission intensity due to the presence of the apolar amino acids tryptophan, tyrosine and phenylalanine, which are typically more localized in the hydrophobic core.³³ In this article, we study the orthogonal thermal and urea-induced denaturation of a multidomain IgG protein to discriminate the thermal unfolding profiles for the different domains in the presence of urea. The detailed evaluation of different urea concentrations and their individual impact on domain stabilities reveal a complex binding and destabilization behavior. Further analysis also points to close connections between unfolding mechanisms and aggregation tendencies. Our results are discussed in the context of thermodynamic and electrostatic binding and repulsion effects.

Materials and Methods

For our nano differential scanning fluorimetry (nanoDSF) measurements, we prepared 24 samples of an aqueous IgG solution with different concentrations of urea ranging from 0 to 8.8 mol/L. The experimentally measured isoelectric point of the protein is located at pI = 7.6–7.8 as measured by isoelectric capillary electrophoresis. For the chosen buffer solution with a pH value of 5.5, one can thus assume a slightly positively charged IgG. A stock solution of 27 mg/mL IgG4 (148 kDa) was stored at −70 °C. The stock solution was thawed at 4 °C overnight and diluted into 1 mg/mL IgG4 with the following 24 gradient urea concentrations: 0.0, 0.49, 0.98, 1.47, 1.96, 2.45, 2.94, 3.43, 3.92, 4.41, 4.9, 5.39, 5.88, 6.13, 6.37, 6.62, 6.86, 7.11, 7.35, 7.6, 7.84, 8.09, 8.33, 8.58 M urea. Another set of 24 samples was produced with identical IgG4 and urea concentrations with 0.3 M NaCl.

In total, 200 μL each of 48 samples were generated in a 96-well plate and incubated at room temperature for 72 h sealed with a plastic cover to avoid evaporation. Around 10 μL was taken using Prometheus NT4.8 capillaries (NanoTemper Technologies GmbH, München, Germany). Thermal unfolding of these 48 samples in the capillaries were measured using the nanoDSF device Prometheus Panta (NanoTemper Technologies GmbH, München, Germany). Emission spectra from 310 to 390 nm and excitation at 295 nm were obtained at temperatures ranging from 20 to 95 °C with a heating rate of 1 °C/min.

The thermal denaturation curves were achieved from the ratio of the fluorescence intensity at 350 and 330 nm (F₃₅₀/F₃₃₀) recorded against temperatures. Since tyrosine has a lower absorption coefficient than tryptophan, excitation wavelengths around 295 nm were selected to reduce the influence of tyrosine.³⁴ The fluorescence emission peak shift of the protein folded state was located around 330 nm, while the unfolded state was associated with wavelengths of 350 nm.³⁵ Thermal parameters from the denaturation curves could be achieved by fitting the curves into a two-state or a three-state sigmoidal model to determine the onset temperature, T_on, and the midpoint temperature, T_m. However, we simply used an inflection point (IP) automatically calculated on the software PR control (NanoTemper Technologies GmbH, München, Germany) from the average of the triplicate measurements. In the identical instrument, the onset temperature of aggregation, T_agg was calculated based on the turbidity determined by the light extinction of samples. Further analysis such as curve fitting and data visualization was done on Origin(Pro), Version 2019 (OriginLab Corporation, Northampton, MA, USA.) and Python Matplotlib.

Results

The corresponding curves for the different melting temperatures in combination with the urea concentration and the corresponding ratio of fluorescence at 350 and 330 nm, r = F₃₅₀/F₃₃₀, are presented in Figure 1. Most of the thermal unfolding curves show various IPs, which can be identified with the help of the first derivatives of the ratio r (data shown in the Supporting Information). During the thermal unfolding, IgG domains are known to unfold independently, as highlighted by multiple transition and melting temperatures from previous studies.^27,36 Therefore, the observed corresponding IPs can be assigned to different melting temperatures for single or multiple domains of the IgG. For increasing urea concentrations, one can observe that the IPs are shifted to lower temperatures. Such thermal shifts induced by increasing urea concentrations highlight different destabilizing effects of urea on various domains, as reflected by lower melting temperatures. The corresponding calculated melting temperatures associated with the different IPs in combination with the urea concentration are shown in Figure 2. As shown in the Supporting Information, one can identify one onset temperature T_on, three different melting temperatures T_(m,1), T_(m,2), and T_(m,3) associated with IP1, IP2, and IP3, respectively, and certain conditions at low urea concentrations with the corresponding aggregation temperatures T_agg. The aggregation temperatures were measured according to the occurrence of turbidity effects. The associated thermal shift curves for T_on and T_(m,1) reveal a nonlinear decay with increasing urea concentration, while the aggregation temperatures T_agg and the melting temperatures T_(m,2) and T_(m,3) show a rather linear decay. The presence of different IPs highlights the individual stability conditions for a multitude of local domains in combination with the global native IgG structure. For reasons of clarity, a simple two-state protein with one domain would show only one single melting temperature associated with one IP.

Figure 1.

3D plot of thermal unfolding curves for IgG samples with a concentration of 1 mg/mL in aqueous urea solutions, including urea concentrations from 0 (purple) to 8.8 mol/L (yellow).

Figure 2.

Thermal shift curves as a function of urea concentration for pure water–urea (circle) and aqueous 0.3 M NaCl–urea solutions (crosses) with the onset temperature T_on (black), the melting temperatures T_m,1 (blue), T_m,2 (orange), and T_m,3 (green), and the aggregation temperature T_agg (red). The curve for T_m,3 was fitted by the logistic decay curves. The aggregation areas based on the identification of the T_agg conditions are marked in red.

The thermal denaturation profile at low urea concentrations includes two melting temperatures, as represented by T_(m,1) and T_(m,2). For low urea concentrations below 2 mol/L, one can identify a linear decay for all melting temperatures. In addition, one observes aggregation conditions for temperatures higher than 70 °C and urea concentrations lower than 4 mol/L. For urea concentrations higher than 6 mol/L, the aggregation tendencies vanish and one observes the occurrence of a new melting temperature T_(m,3) as obtained by a logistic fit. The details of the curve fitting result can be found in the Supporting Information. Since the thermal shift curve of T_(m,2) merges into the T_(m,3) curve at urea concentrations above 6 M, it can be assumed that the corresponding IgG domain has reached the completely unfolded state. Comparable conclusions about fully unfolded IgG domains can also be assumed for T_(m,1), which disappears at high urea concentrations. Thus, the unfolding process of the two domains is directly coupled to an irreversible aggregation step at high temperatures and low urea concentrations. Noteworthy, specifically the curve for T_(m,2) is closely located to T_agg, which indicates coupled transition steps including aggregation tendencies as initiated from one domain. For urea concentrations higher than 4 mol/L, all aggregation tendencies vanish, which highlights the crucial influence of urea. Around this condition, the thermal shift curve of T_(m,2) changed from a linear-like decrease to a logistic decay curve.

Furthermore, we studied the effects of NaCl–urea solutions with a salt concentration of 0.3 mol/L on IgGs. It was often discussed that electrostatic interactions either stabilize or destabilize protein structures.¹¹ As can be seen by the modified melting temperatures in the presence of salt, the conformational stabilities of IgG domains related to T_(m,1) and T_(m,2) are significantly decreased. The onset temperature T_on and T_(m,1) reveal a similar decay but with a constant offset value in the presence of the salt. Interestingly, at high urea concentrations, one can observe, in terms of T_(m,2) that the salt even stabilizes the protein and thus counteracts the denaturing effects of urea. The corresponding results reveal rather complex unfolding mechanisms and aggregation tendencies of the IgG antibody. In accordance with the observation of different melting temperatures, one can also assume an independent destabilization of different domains in the presence of urea. At high urea concentrations above 6 mol/L, any aggregation tendency has disappeared and only a single melting temperature is observed. Moreover, a significant shift of the onset temperature to higher values becomes evident. Notably, the presence of salt leads to higher melting temperatures T_(m,3) when compared with simple aqueous urea solutions. The corresponding results can be interpreted as follows. We associate this behavior with an unfolding of the protein to the completely unfolded state. The high urea concentrations form saturated cosolute shells around the protein such that global unfolding mechanisms are triggered. This unfolding mechanism becomes evident in terms of the significantly changed onset temperatures, which highlight a completely different unfolding mechanism and structure for the protein when compared to lower urea concentrations. Corresponding conclusions can also be drawn regarding the influence of salt ions, which stabilize this unfolded structure due to higher melting temperatures. It was already discussed in previous publications that salt ions in complex mixtures can either induce stabilization or destabilization effects depending on their local arrangement around the protein.^2,37,38 In addition, one can observe indistinguishable onset temperatures in the presence and absence of salts. Such findings reveal that electrostatic attraction or repulsion interactions due to different electrostatic screening lengths in the presence or absence of salt¹² are not important for this observation. In addition, completely unfolded states usually do not aggregate since their apolar side chains are randomly distributed,³⁹ which rationalizes the absence of aggregates. As a further rationale, one can assume a crucially affected chemical potential of the protein in the presence of high urea concentrations, which also does not favor aggregation effects.

Discussion

At low urea concentrations below 4 mol/L, one can observe two different melting temperatures and strong aggregation tendencies. The different melting temperatures can be associated with the partial unfolding of two different and independent domains of the protein, as driven by urea. Notably, the melting temperature T_(m,2) roughly coincides with the aggregation temperature T_agg, which highlights certain structural relationships in terms of the underlying unfolded structures. Moreover, one can observe nearly identical T_agg values in the presence of salt, while slight changes can be observed in the absence of salt for increasing urea concentrations. The presence of salt ions results in electrostatic screening lengths of λ ≈ 0.55 nm,¹² which is clearly smaller than the average distance between proteins at this concentration but larger than the distance between different domains in one protein. Thus, all electrostatic interactions of partially charged proteins become screened such that electrostatic repulsion effects between similarly charged proteins are crucially suppressed. Such a driving mechanism thus rationalizes the nearly constant aggregation temperatures in the presence of salt for screened proteins. In addition, the aggregation mechanism is driven by one partially unfolded domain. This can be concluded based on the identical values of T_(m,2) and T_agg in the absence of salt. Moreover, the significantly higher values of T_agg when compared to T_(m,1) indicate that a partially unfolded domain with a melting temperature T_(m,1) does not trigger aggregation effects. Such observations of partially unfolded domains as initiators for aggregation are in perfect agreement with previous assumptions on the formation of reversible aggregates.^18,19 Our results with T_agg changes with increasing urea concentration agree well with the well-known fact that urea suppresses the protein thermal aggregation.^40,41 Notably, in the presence of 0.3 M NaCl, T_agg remains relatively constant upon an increase in the urea concentration, while protein aggregation is inhibited at lower urea concentrations than the samples in the absence of NaCl. Finally, one can observe significant differences for T_on and T_(m,1) in the presence and absence of salt. The presence of salt further induces a destabilization of the domain structure, as highlighted by significantly lower values of T_(m,1) and T_on. One can relate such findings to preferential binding and electrostatic screening effects between differently charged regions of the protein. Thus, electrostatic attraction, which stabilizes the intramolecular structure between protein regions, is reduced in the presence of salt, which thus lowers the corresponding melting temperature.

As already mentioned, the preferential binding and exclusion of urea can be regarded as key drivers in terms of the unfolding mechanisms. At low urea concentrations, preferential binding effects and their influence on the free energy difference between the folded and unfolded protein states are rather moderate. Such observations change for high urea concentrations, which reveal a strong local attraction behavior of urea to certain regions of the protein. The corresponding preferential binding coefficients crucially affect the chemical potential and thus the free energy difference between the folded and the unfolded states.^2,3 Notably, further evidence concerning the impact of preferential binding mechanisms becomes obvious with regard to the functional decay of the temperatures T_on and T_agg. For low urea concentrations below 2 mol/L, one can observe a nearly linear decay, which transforms into a nonlinear decay for higher urea concentrations. In a recent publication,⁴³ we showed that decreasing melting temperatures can be associated with a preferential binding mechanism. The corresponding melting temperature variation can be written as ΔT ∝ −a₃₃Δν with the derivative a₃₃ of the thermodynamic activity a₃ for urea molecules and the difference of the preferential binding coefficients between the native and the partially unfolded state. For low urea concentrations c₃, it is known that a₃ ≈ c₃, such that any derivative a₃₃ = ∂ln a₃/∂c₃ in terms of series expansion shows linear changes upon variation of the urea concentration.⁴² At higher urea concentrations, nonideal clustering effects crucially affect the thermodynamic activity, which thus gradually changes into nonlinear variation. Under the reasonable assumption of nearly constant, moderate urea concentrations,⁴³ one can thus assume that the nonlinear decay of the melting temperatures is mainly induced by increasing urea concentrations and a significantly changing thermodynamic activity.

Conclusions

The discrimination of IgG domains in stability analysis and their impact on shelf lives are important for the improved design of therapeutic drug formulations. Our studies revealed a direct destabilization effect of urea on different IgG domains by using nanoDSF. The conformational stability of IgG domains depends on the preferential binding of urea, which can be monitored by urea-induced thermal shifts. The occurrence of aggregation inhibiting effects for IgG in the presence of salt was considerably strong, as highlighted by the relatively constant T_agg upon increasing urea concentration. Hence, one can conclude that ions or stabilizing cosolutes are reasonable options to modify the shelf life and the colloidal stability upon further addition. The urea-induced thermal shift curves can clearly be discriminated in the presence and absence of salt concentrations. Notably, different effects of the salt on the individual IgG domains were observed. We discussed these effects in terms of preferential binding and electrostatic screening mechanisms, which are independent of the underlying urea concentration. Therefore, one can discriminate between urea-and ion-induced preferential binding, both of which lead to protein domain destabilization but affect the behavior on large scales in different ways. Our study discusses a reasonable approach to monitor thermal denaturation effects of cosolutes in terms of conformational and colloidal stability for independent IgG domains. We hope that the corresponding insights into the complex unfolding and aggregation effects can provide new ideas to extend the shelf life of therapeutic drug formulations.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.3c07323.

• Thermal unfolding curves, results of curve fitting, and turbidity spectra of all samples (PDF)

The authors declare no competing financial interest.