WO2024258896A1 - Methods of determining viscosity - Google Patents

Abstract

The present disclosure is in the field of determining viscosity of pharmaceutical compositions comprising antigen-binding protein, wherein the methods comprise measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein under a condition that reduces the surface charge of the antigen-binding protein, and calculating the k_D of the formulation based on the measured plurality of D.

Description

METHODS OF DETERMINING VISCOSITY

[0001] This application claims priority benefit of United States Provisional Application No. 63/507,969, filed on June 13, 2024, and which is incorporated herein by reference in its entirety.

FIELD OF THE INVENTION

[0002] The present disclosure is in the field of determining viscosity of pharmaceutical compositions comprising antigen-binding protein.

BACKGROUND

[0003] Monoclonal antibodies (mAbs) have emerged as a class of fast-growing therapeutics over the last three decades. As of Aug. 2022, the FDA has approved over 110 therapeutic antibodies and Fc fusion proteins across diverse therapeutic areas covering immunology, oncology, infectious diseases, genetic diseases, etc. There are even many more mAb candidates currently being tested at various stages of clinical trials. Extensive screening and engineering efforts go into the candidate selection process to identify the lead mAb candidate for clinical testing. In addition to assessing the therapeutic potential of mAb candidates, significant efforts need to be put into the evaluation of their developability profiles, which encompasses expression titer, stability, manufacturability, delivery options, pharmacokinetic properties, and others. Progression of a lead candidate with a desirable developability profile to clinical trial can reduce the risks, timelines, and resources associated with product development and increase the odds of regulatory approval and market competitiveness.

[0004] The majority of commercially approved mAbs are administered via intravenous (IV) infusion or subcutaneous (SC) injection (Ryamn et al). SC administration of mAbs with autoinjectors or pre-filled syringes offers several advantages over the common IV route as can potentially allow patients to self-administer at home and reduces hospital visits. The SC route is particularly beneficial for patients with chronic diseases as frequent dosing is often required. On the other hand, SC administration is usually limited to an injection volume less than 2 mL (Mathaes et al.) and this in turn requires a formulation concentration as high as 100-200 mg/mL to deliver enough active ingredients in a single dose. mAbs at high concentrations can show elevated tendency to self-associate and manifest high viscosity, posing challenges during product manufacturing, storage, and SC injection. Direct measurement of viscosity at 100 mg/mL and above is often not feasible due to limited materials at the early stage of candidate screening. Predictive assays with low material consumption would serve as a powerful tool to ensure successful identification of lead candidates with low viscosity.

[0005] High viscosity of concentrated mAb solutions has long been hypothesized to originate from formation of a transient protein network driven by weak protein-protein interactions (PPIs) (Buck et al. 2015. 12(1 ): p. 127-139 and Tomar et al. MAbs. 2016. Taylor & Francis.).

Predictive assays that measure PPIs in dilute solutions may inform on viscosity behavior at high concentrations. Diffusion interaction parameter (k_D) is such a label-free biophysical measure among many other techniques, and it has been shown to correlate with mAb viscosities at high concentrations (Connolly et al., 2012. 103(1): p. 69-78; Kingsbury et al. 2020. 6(32): p. eabb0372; Yadav et al. 2012. 101 (3): p. 998-1011 ; Zhou et al. U.S. 7,235,641 ). Common techniques to measure k_D include dynamic light scattering (DLS) and Taylor dispersion (Lavoisier et al., Latunde-Dada et al. 2016. 8(2): p. 386-392), with the former being more commonly used for screening purposes because it can be adapted to high-throughput analyses in microplate-based format. Connolly et al. measured the k_D of 29 mAbs in different buffers and found a strong linear dependence between k_D and the exponential coefficient of concentration dependent viscosity profiles (Connolly et al., 2012. 103(1 ): p. 69-78). A recent paper reported that mAb k_D measured in 10 mM histidine buffer pH 6.0 well predicts whether their viscosity (at 150 mg/mL in 10 mM Histidine 8% Sucrose pH 6.0) falls into a high (> 30 cP) or low (< 30 cP) viscosity category (Kingsbury et al. 2020. 6(32): p. eabb0372;). However, a scatter plot showing correlation between k_D and viscosity is not reported in the paper. Despite some successful case studies of using k_D to predict viscosity, significant challenges still exist for several reasons (Woldeyes et al., 2019. 108(1 ): p. 142-154; Dear et al, 2017. 34(1): p. 193-207). Firstly, PPIs measured in dilute solutions is not necessarily representative of molecular interactions seen at high concentrations. The strength of various molecular interactions is strongly dependent on intermolecular distances; therefore, their relative contributions can differ remarkably at low versus at high concentrations (Neergaard et al. 2013. 49(3): p. 400-410; Chari et al., 2009. 26(12): p. 2607-2618). In dilute solutions, the average surface-to-surface intermolecular distance of charged mAb molecules is far larger than their Debye screening length, and mAb molecules can freely rotate and see each other as point charges (Yadav et al., 2011 . 28(7): p. 1750-1764). As a result, long-range global electrostatic interactions dominate PPIs in dilute solution. This is supported by several findings in the literature (Chari et aL, 2009. 26(12): p. 2607-2618, Lehermayr et al., 2011. 100(7): p. 2551-2562, Yadav et al. 2010. 99(3): p. 1 152- 1168). In contrast, at high concentrations, protein molecules are in closer proximity with a concomitant higher tendency to experience short-range interactions (e.g. van der Waals, electrostatic, hydrophobic interactions, hydrogen bonding, hard sphere repulsion), which have been found as key factors driving viscosity (Chari et al., 2009. 26(12): p. 2607-2618). Secondly, the 1 st order PPIs manifested in the form of k_D may not suffice to predict the viscosity of a highly concentrated mAb solution which is essentially a summed result of 1 st and higher order interactions. These gaps can lead to poor prediction of viscosity using ko. Indeed, there is a large set of data indicating a weak correlation between k_D and the natural logarithm of mAb viscosity at high protein concentrations (e.g., 140 mg/mL) when both were measured under a certain set of conditions (e.g., in a formulation at pH 5.2). There is a need to improve using k_D as a predictive tool for mAb candidate selection.

SUMMARY

[0006] During developability assessment of therapeutic monoclonal antibody (mAb) candidates, utilization of robust high-throughput predictive assays enables rapid selection of top candidates with low risks for late-stage development. Predicting viscosities of highly concentrated mAbs using limited materials is an important aspect of developability assessment because high viscosity can complicate manufacturability, stability, and administration.

Disclosed herein is a high-throughput assay measuring protein-protein interactions to predict/determine mAb viscosity.

[0007] The present disclosure provides methods of determining the viscosity of a pharmaceutical formulation comprising an antigen-binding protein, which requires a small amount of antigen-binding product and may be carried out at an early stage in the development process of the antigen-binding product. The diffusion interaction parameter (k_D) of a formulation comprising an antigen binding protein at high concentrations (e.g., 100 mg/ml or higher) is indicative of the viscosity of that formulation. The disclosed methods comprise the steps of measuring the diffusion coefficient (D) of a plurality of compositions under a condition that reduces the surface charge of the antigen-binding protein in the formulation, and calculating the k_D of the formulation based on the measured diffusion coefficient.

[0008] The monoclonal antibody (mAb) candidates often have pl values well above the formulation pH of 5.2 and therefore carry strong net positive charges in a low ionic buffer. As a result, ko reflects primarily long-range electrostatic repulsions because it is measured in dilute solutions where short-range interactions are insignificant. This is further supported by the observation that k_D values show a strong correlation with pl values of mAbs. Short-range interactions that play a key role in driving viscosity in concentrated mAb solutions are not being sufficiently captured by the k_D measurements in common buffers, such as low ionic buffers. To circumvent this issue, the disclosure provides method of determining the viscosity of a pharmaceutical composition in which salts are added to purposely reduce antigen-binding protein surface charge. By tuning down long-range charge repulsions, the disclosed methods of measuring a k_D will capture short-range interactions, if any. To develop this modified method of determining viscosity, the salt type and concentration was optimized to maximize the predictability of k_D. In addition, the disclosed methods were validated using one engineering panel of 10 mAbs and another mAb panel with 17 mABs with minimal sequence similarity in the complementarity-determining region (CDR). A consolidated data set consisting of these data points suggests a notable improvement in the prediction power of ko for viscosity using the optimized ko method described herein.

[0009] In one embodiment, the disclosure provides for method of determining viscosity of a pharmaceutical formulation comprising an antigen-binding protein, the method comprising a) measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein under a condition that reduces the surface charge of the antigen-binding protein, and b) calculate k_D of the formulation based on the plurality of Ds from step a), wherein ko of the formulation is indicative of the viscosity of the formulation. Methods of calculating kD using the measurement of Ds are well known in the art, for example, k_D can be calculated using the formula D = D_o (1 + k_D * c), wherein D is diffusion coefficient of samples, D_o is the diffusion coefficient of mAb at infinite dilution, and c is protein concentration. Dividing the slope (D_o* k_D) by the intercept (D_o) returns k_D..The plurality of compositions comprises the pharmaceutical formulation but have different concentrations of the antigen-binding protein in order to calculate kD, which is indicative of viscosity of the formulation which comprises a high concentration of an antigen-binding protein. The disclosed method uses lower concentrations of antigen-binding protein to predict the viscosity of the formulation comprising high concentration of said antigen-binding protein.

[0010] Diffusion interaction parameter (k_D) measures colloidal self-association in dilute solutions and has been reported to be predictive of mAb viscosity at high concentrations. However, k_D of IgG 1 mAb candidates measured in acetate buffer having a pH of 5.2 shows only weak correlation to their viscosities at 140 mg/mL. Measuring ko in low strength acetate buffers reflects primarily long-range electrostatic repulsions because most internal lgG1 mAb candidates carry strong net positive charges in this low ionic strength formulation with pH (5.2) well below pl values of mAb candidates. However, the viscosity of high concentration mAbs depends heavily on short-range molecular interactions. The improved methods described herein, comprise measuring D under a condition that reduces the surface charge of the antigenbinding protein, such as adding salt to the formulation to reduce the surface charge by suppressing charge repulsions and allowing for detection of key short-range interactions in dilute solutions. Overall, the method improved the Pearson R2 between kD and viscosity (6-230 cP) from 0.42 to 0.80 for a data set consisting of 31 mAbs.

[0011] In the methods described herein, a low ko, is indicative of a high viscosity formulation while a high k_D is indicative of a low viscosity formulation. For example, a low viscosity formulation has a k_D ranging from 10 cP to 30 cP. In some embodiments, a low viscosity formulation has a k_D of about 10 cP, or about 1 1 cP, or about 12 cP, or about 13 cP, or about 14 cP, or about 15 cP, or about 16 cP, or about 17 cP, or about 18 cP, or about 19 cP, or about 20 cP, or about 21 cP, or about 22 cP, or about 23 cP, or about 24 cP, or about 25 cP, or about 26 cP, or about 27 cP or about 28 cP or about 29 cP or about 30 cP. In any of the methods disclosed herein, a k_D ranging from 10 cP to about 30 cP may be used a cut-off for determining if the formulation has a low viscosity or high viscosity.

[0012] For example, the condition to reduce the surface charge of the antigen-binding protein is the presence of salt in the composition, such as adding salt to the plurality of compositions comprising different concentrations of the antigen-binding protein and measuring the Ds of these plurality of compositions. The salt may be any known salt, such as sodium chloride, potassium chloride, lithium chloride, magnesium chloride, calcium chloride, ammonium chloride, sodium sulfate, ammonium sulfate, potassium sulfate, lithium sulfate, magnesium sulfate, calcium sulfate, sodium phosphate, ammonium phosphate, potassium phosphate, lithium phosphate, magnesium phosphate, calcium phosphate, sodium bromide, ammonium bromide, potassium bromide, lithium bromide, magnesium bromide, calcium bromide, sodium nitrate, ammonium nitrate, potassium nitrate, lithium nitrate, magnesium nitrate, calcium nitrate, sodium chlorate, ammonium chlorate, potassium chlorate, lithium chlorate, magnesium chlorate, calcium chlorate, sodium iodide, ammonium iodide, potassium iodide, lithium iodide, magnesium iodide, calcium iodide, sodium perchlorate, ammonium perchlorate, potassium perchlorate, lithium perchlorate, magnesium perchlorate, calcium perchlorate, sodium thiocyanate, ammonium thiocyanate, potassium thiocyanate, lithium thiocyanate, magnesium thiocyanate, or calcium thiocyanate, or a combination thereof. [0013] In some embodiments, the salt is sodium chloride (NaCI), for example the NaCI is at a concentration of about 10 mM to about 200 mM, about 10 mM to about 190 mM, about 10 mM to about 180 mM, about 10 mM to about 170 mM, about 10 mM to about 160 mM, about 10 mM to 150 mM, about 20 mM to about 180 mM, about 20 mM to about 160 mM, about 20 mM to about 150 mM, about 50 to about 150 mM, about 50 mM to about 160 mM, about 50 mM to about 150 mM, about 100 mM to about 150 mM, about 100 mM to 160 mM, about 100 mM to about 170 mM, about 100 mM to about 180 mM, about 100 to about 190 mM, about 100 mM to about 200 mM, about 100 mM to about 250 mM, about 100 mM to about 300 mM, about 100 mM to about 400 mM, or about 100 mM to about 500 mM. In some embodiments, the NaCI is at a concentration of about 10 mM, about 20 mM, about 30 mM, about 40 mM, about 50 mM, about 60 mM, about 70 mM, about 80 mM, about 90 mM, about 100 mM, about 1 10 mM, about 120 mM, about 130 mM, about 140 mM, about 150 mM, about 160 mM, about 170 mM, about 180 mM, about 190 mM, about 200 mM, about 250 mM, about 300 mM, about 400 mM or about 500 mM.

[0014] In another embodiment, the salt is ammonium sulfate. For example, the ammonium sulfate is at a concentration of about 5 mM to about 100 mM, about 5 mM to about 90 mM, about 5 mM to about 80 mM, about 5 mM to about 70 mM, about 5 mM to about 60 mM, about 5 mM to 50 mM, about 8 mM to about 80 mM, about 8 mM to about 60 mM, about 8 mM to about 50 mM, about 10 to about 50 mM, about 10 mM to about 60 mM. In some embodiments, the ammonium sulfate is at a concentration of about 5 mM, 8.33 mM, about 10 mM, about 20 mM, about 30 mM, about 40 mM, about 50 mM, about 60 mM, about 70 mM, about 80 mM, about 90 mM, or about 100 mM.

[0015] The disclosure also provides for a method of determining viscosity of a pharmaceutical formulation comprising an antigen-binding protein, the method comprising a) measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein under a condition wherein the pH of the composition is near to the pl of the antigen-binding protein, and b) calculate k_D of the formulation based on the plurality of Ds from step a) wherein k_D of the formulation is indicative of the viscosity of the formulation. Methods of calculating k_D using the measurement of Ds are well known in the art, for example, ko can be calculated using the formula D = D_o (1 + ko * c), wherein D is diffusion coefficient of samples, D_o is the diffusion coefficient of mAb at infinite dilution, and c is protein concentration. Dividing the slope (D_o* k_D) by the intercept (Do) returns k_D The plurality of compositions comprises the pharmaceutical formulation but have different concentrations of the antigen-binding protein in order to calculate k_D, the measurement of the which is indicative of viscosity of the formulation which comprises a high concentration of an antigen-binding protein. The disclosed method uses lower concentrations of antigen-binding protein to predict the viscosity of the formulation comprising high concentration of said antigenbinding protein. For example, the pH of the formulation ranges from about 5.5 to about 9.5. For example, the pH of the formulation ranges from about 5.5 to about 9.0, or from about 5.5 to about 8.0, or from about 5.5 to about 7.7, or from about 5.5 to about 7.5, or from 5.5 to about 7.0, or about 6.0 to about 9.5, or from 6.0 to about 9.0, or from 6.0 to about 8.6, or from 6.0 to about 8.0, or from about 6.0 to about 7.8, or from about 6.0 to about 7.5, or from about 6.5 to about 9.5, or from about 6.5 to about 9.0, or from about 6.5 to about 8.6, or from 6.5 to about 8.0, or from 6.5 to about 7.8, or from about 6.5 to about 7.5, or from 7.5 to about 9.5, or from7.5 to about 9.0, or from 7.5 to about 8.6, or from 7.5 to about 8.0, or from 7.7 to about 9.5, or from 7.7 to about 9.0, or from 7.7 to about 8.6, or from 7.7 to about 8.7, or from 7.9 to 9.0, or from 8.3 to about 9.5, or from 8.4 to about 9.5, or from 8.5 to about 9.5, or from 7.7 to about 9.2, or from about 7.7 to about 9.3, or from 7.7 to about 9.5.

[0016] For example, the pH of the formulation is about 5.5, or about 5.8, or about 6.0 or about 6.5 or about 6.8, or about 7.0, or about 7.1 or about 7.2, or about 7.3, or about 7.4, or about 7.5, or about 7.6, or about 7.7, or about 7.8, or about 7.9, or about 8.0, or about 8.1 or about 8.2 or about 8.3 or about 8.4, or about 8.5 or about 8.6, or about 8.7, or about 8.8, or about 8.9, or about 9.0, or about 9.1 or about 9.2 or about 9.3 or about 9.4, or about 9.5.

[0017] In any of the disclosed methods, the plurality of compositions comprises different concentrations of the antigen-binding protein. For example, the concentrations of the plurality of antigen-binding protein is less than or equal to (<) 20 mg/ml, such as the concentration of antigen-binding protein ranging from about 2 mg/ml to about 20 mg/ml, about 2 to about 19 mg/ml, about 2 mg/ml to about 18 mg/ml, about 2 mg/ml to about 17 mg/ml, 2 mg/ml to about 16 mg/ml, about 2 mg/ml to about 15 mg/ml, about 5 mg/ml to about 20 mg/ml, about 5 to about 19 mg/ml, about 5 mg/ml to about 18 mg/ml, about 5 mg/ml to about 17 mg/ml, about 5 mg/ml to about 16 mg/ml, about 5 mg/ml to about 15 mg/ml, about 10 mg/ml to about 20 mg/ml, about 10 to about 19 mg/ml, about 10 mg/ml to about 18 mg/ml, about 10 mg/ml to about 17 mg/ml, about 10 mg/ml to about 16 mg/ml, about 10 mg/ml to about 15 mg/ml, about 3 mg/ml to about 10 mg/ml, about 3 mg/ml to about 12 mg/ml, about 3 mg/ml to about 15 mg/ml, 3 mg/ml to about 18 mg/ml or 3 mg/ml to about 20 mg/ml. [0018] Any of the disclosed methods may be used to determine viscosity of the pharmaceutical formulations comprising antigen-binding proteins at different temperatures, such as temperatures ranging from 5-37 °C.

[0019] In another embodiment, the disclosure provides for methods of screening an excipient for preparation of a pharmaceutical formulation comprising an antigen-binding protein, wherein the method comprises determining the viscosity of the pharmaceutical formulation comprising the excipient using any of the methods disclosed herein, and comparing the viscosity of the pharmaceutical composition comprising the excipient with the viscosity of the pharmaceutical composition without the excipient. Some of the disclosed methods, the method further comprises the step of determining the viscosity of the pharmaceutical formulation without the excipient using any of the methods disclosed herein. Any of the disclosed methods may be used to screen a panel of excipients, and the screening may be carried out with high-throughput screening.

[0020] For example, the disclosure provides a method of screening an excipient for preparation of a pharmaceutical formulation comprising an antigen-binding protein, the method comprising i) determining the viscosity of the pharmaceutical formulation comprising the excipient by: a) measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein and different concentrations of the excipient under a condition that reduces the surface charge of the antigen-binding protein or under a condition wherein the pH of the formulation is near to the pl of the antigen-binding protein, and b) calculating k_D of the formulation based on the plurality of Ds from step a), wherein k_D of the formulation is indicative of the viscosity of the formulation, and ii) comparing the k_D of the formulation comprising the excipient with the k_D of the formulation without the excipient. In any of the disclosed methods, the term “plurality” refers a number of compositions in which the D is measured. For example, the plurality is least 3 compositions, or at least 4 compositions or at least 5 compositions.

[0021] In any of the disclosed methods, the diffusion coefficient (D) can be measured by any method known in the art. For example, in certain embodiments, D is measured by dynamic light scattering, including capillary based DLS. In addition, the D is measured by Fluorescence Correlation Spectroscopy, nuclear magnetic resonance spectroscopy, macroscopic mass transfer (Taylor dispersion) or by measuring the concentration spatial profile using optical techniques within a diffusion chamber. [0022] In any of the disclosed methods, the pharmaceutical formulation comprises a saccharide and a surfactant. In some of the disclosed methods, , the pharmaceutical formulation further comprises a buffer. For example, the buffer is an isotonic butter, an acetate buffer, a glutamate buffer, a citrate buffer, a lactic buffer, a succinate buffer, a tartrate buffer, a fumarate buffer, a maleate buffer, a histidine buffer, or a phosphate buffe, r2-(N- morpholino)ethanesulfonate buffer or a combination thereof. In some embodiments, the buffer is present in the formulation at a concentration ranging from about 5 mM to about 200 mM (or about 10 mM to about 50 mM

[0023] In any of the disclosed methods, the pharmaceutical formulation comprises a saccharide. In some embodiments, the saccharide is monosaccharide or a disaccharide. In some embodiments, the saccharide is a sugar alcohol (e.g., sorbitol). In some embodiments, the saccharide is sucrose, galactose, fructose, xylose, xylitol, maltose, trehalose, mannitol, sorbitol or a combination thereof. In some embodiments, the saccharide is present in the formulation at a concentration ranging from about 1 to about 15% (w/V) (or about 9 to about 12% (w/V) or about 5% to about 12% (w/V) or about 7% to about 12% (w/V)).

[0024] In some embodiments, the formulation comprises a surfactant. In some embodiments, the surfactant is polysorbate 20, polysorbate 40, polysorbate 60, polysorbate 80, poloxamer 188, poloxamer 407, triton X-100, polyoxyethylene, PEG 3350, PEG 4000, or a combination thereof. In some embodiments, the surfactant is present in the formulation at a concentration ranging from 0.004 Any to about 0.5% (w/V) (or about 0.001 to about 0.01% (w/V), or about 0.001 to about 0.5% (w/V) or about 0.004 to about 0.01 % (w/V)).

[0025] In an exemplary embodiment, the method is carried out on a pharmaceutical formulation having a pH of 5.2. For example, the pharmaceutical formulation comprises a 10 mM acetate buffer, sucrose and polysorbate, at pH 5.2. In another embodiment, the method is carried out on a pharmaceutical formulation comprising 10 mM glutamate, 9% (w/V) sucrose and 0.01% (w/V) polysorbate 80, and wherein the pH of the pharmaceutical formulation is 4.2. In addition, the method is carried out on a pharmaceutical formulation comprising histidine, at pH 6.0, which may also comprise sucrose and a polysorbate.

[0026] In any of disclosed methods, the antigen-binding protein. The term “antigen-binding protein” is a protein comprising a domain that binds a specified target antigen. For example, the antigen-binding protein is a polypeptide, large peptide, peptide conjugate, antibody, antibody fragment, antibody fusion peptide or antigen-binding fragment thereof. In some embodiments, the antigen-binding protein is a polyclonal antibody, monoclonal antibody or bi-specific antibody construct. For example, the antibody is an IgG monoclonal antibody, such as IgG 1 or lgG2.

[0027] In some embodiments, the pharmaceutical formulation of the disclosure is a liquid formulation.

[0028] It should be understood that while various embodiments in the specification are presented using “comprising” language, under various circumstances, a related embodiment may also be described using “consisting of” or “consisting essentially of” language. The disclosure contemplates embodiments described as “comprising” a feature to include embodiments which “consist of” the feature. It is to be noted that the term “a” or “an” refers to one or more, for example, “an immunoglobulin molecule,” is understood to represent one or more immunoglobulin molecules. As such, the terms “a” (or “an”), “one or more,” and “at least one” can be used interchangeably herein.

[0029] It should also be understood that when describing a range of values, the characteristic being described could be an individual value found within the range. For example, “a pH from about pH 4 to about pH 6,” could be, but is not limited to, pH 4, 4.2, 4.6, 5.1 , 5.5 etc. and any value in between such values. Additionally, “a pH from about pH 4 to about pH 6,” should not be construed to mean that the pH of a formulation in question varies 2 pH units in the range from pH 4 to pH 6 during storage, but rather a value may be picked in that range for the pH of the solution, and the pH remains buffered at about that pH.

[0030] When the term “about” is used, it means the recited number plus or minus 5%, 10%, 15% or more of that recited number. The actual variation intended is determinable from the context.

[0031] In any of the ranges described herein, the endpoints of the range are included in the range. However, the description also contemplates the same ranges in which the lower and/or the higher endpoint is excluded. Additional features and variations of the invention will be apparent to those skilled in the art from the entirety of this application, including the drawing and detailed description, and all such features are intended as aspects of the invention. Likewise, features of the invention described herein can be re-combined into additional embodiments that also are intended as aspects of the invention, irrespective of whether the combination of features is specifically mentioned above as an aspect or embodiment of the invention. Also, only such limitations which are described herein as critical to the invention should be viewed as such; variations of the invention lacking limitations which have not been described herein as critical are intended as aspects of the invention.

[0032] All references cited herein are hereby incorporated by reference in their entireties.

BRIEF DESCRIPTION OF THE FIGURES

[0033] Figures 1 A-1 H provide DLS D plot of the 10 mAb in Buffer A with different levels of NaCI (A-F: 0-150 mM) or ammonium sulfate (G-H: 8.33 and 50 mM).

[0034] Figures 2A-2B demonstrate that addition of NaCI (A) or (NH₄)2SO4 (B) gradually decreased DLS k_D values of the 10 mAbs in Buffer A.

[0035] Figures 3A-3I demonstrates the progressive increase in the predictive power of k_D measured under charge screening conditions with NaCI (A-F: 0-150 mM) or ((NH₄)2SO4 (G-H: 8.33 and 50 mM) for viscosity at 140 mg/mL in Buffer A (N=3) measured as cP. The Pearson R² gradually improves as a function of ionic strength (I). The best linear correlation is observed for ko measured in Buffer A with 150mM NaCI (F).

[0036] Figures 4A-4D provide DLS k_D plots of an engineering panel of 10 IgG 1 (Test set A) measured in Buffer A with and without 150 mM NaCI (A and B). The scattering plots show correlation between k_D (C) or k_D,i50Naci (D) (N=2) measured as cP. Candidate mAb A5 shows an apparent D_o substantially lower than expected values of around 4.0x10^-7 cm²/s (B) and is therefore considered as an outlier (D).

[0037] Figures 5A-5D provides DLS k_D plots of a test panel of 11 IgG 1 mAbs (Test set B) measured in Buffer A with and without 150 mM NaCI (A and B). The scattering plots show correlation between k_D, Buffer A (C) or k_D, isomM Naci (D) (N=3) measured as cP.

[0038] Figures 6A-6B provide correlation between ko (A) or kD.isoNaci (B) and viscosity at 140 mg/mL in Buffer A at 25 ^eC measured as cP. All samples (31 mAbs) tested in the study are consolidated in this figure. Compared to ko, ko.isoNaci shows improved correlation with viscosity.

[0039] Figures 7A-7C provide the averaged surface-to-surface intermolecular distance of mAb molecules as a function of protein concentration (A1 , B1 , C1 ). The calculated Debye length is plotted as a function of solution ionic strength (A2, B2, C2). When k_D is measured in Buffer A, mAb molecules experience predominantly strong charge repulsions and the thick Debye layer prevents formation of possible short-range interactions (A3). When ko is measured in Buffer A in the presence of salt, the Debye layer becomes thinner due to charge screening and allows possible short-range interactions to form (B3). At high concentrations, mAb molecules are in proximity in the “crowded” solution environment despite the presence of thick Debye layer (C3).

[0040] Figures 8A-8B provides a comparison of mAb k_D measured in Buffer A with salts at 25 mM (A) or 150 mM (B) ionic strength.

[0041] Figures 9 provides a scatter plot of k_D, isoNaci versus viscosity (at 140 mg/mL in Buffer A) on logarithmic scale. k_D, isoNaci was calculated from the diffusion coefficient at a single concentration of around 12 mg/mL.

[0042] Figure 10 provides the viscosity profile of several IgG 1 mAbs (denoted as mAb C1 , mAb C2, mAb 03 and mA 04) as a function of concentration at different temperatures 5-37 °C. mAb 01 and mAb 02 show significant increase in the viscosity as the temperature cools down to 5°C.

[0043] Figures 1 1 A-D provide DLS ko plots of several IgG 1 mAbs (denoted as mAb 01 , mAb C2, mAb C3 and mA C4) measured in Buffer A with and without 150 mM NaCI at different temperatures 5-37 °C (N=3).

[0044] Figures 12A-B provide scatter plots showing correlation between k_D and temperature in Buffer without (A) and with 150 NaCI (B).

DETAILED DESCRIPTION

[0045] The use of k_D to predict the viscosity of concentrated mAb solutions has been reported in the literature with some success. It has been demonstrated that k_D exhibits weak correlation to mAb viscosity when both parameters are measured in a buffer having a pH of 5.2. It was hypothesized that k_D measured in common antigen-binding protein buffer reflects primarily global long-range charge repulsions. On the other hand, viscosity at high concentration is presumably linked to the formation of protein “clusters” driven by attractive short-range interactions (e.g. van der Waals attractions, electrostatic, hydrophobic interactions, hard sphere repulsion). Therefore, there is a clear disconnect between what k_D measures and the key interactions driving viscosity. To fill this gap, disclosed herein are methods of measuring k_D under charge conditions where long-range interactions are intentionally suppressed to allow for capturing key short-range interactions highly relevant to viscosity at high concentrations.

[0046] As described in the Examples herein, salt type and concentrations were screened to optimize the prediction power of k_D for viscosity. Titrating either NaCI or (NH^SC gradually improved the predictive performance of k_D. When directly comparing k_D values in the presence of NaCI or (NH^SC , the majority of mAbs show a more negative k_D value in the presence of (NH₄)₂SO₄ compared to NaCI at 50 mM ionic strength (Figure 8A). This can probably be explained by the ability of ammonium to promote hydrophobic interactions as a kosmotrope in addition to a charge screening effect. At 150 mM ionic strength, k_D measured in these two salts are more comparable across these 10 mAbs (Figure 8B). Overall, ko measured in antigenbinding protein buffer in the presence of 150 mM NaCI (k_D, isoNaci) was found to be an optimal condition. This optimized approach was further validated with two panels of mAb molecules, and in both test cases k_D, isoNaci show improved predictability compared to k_D measured in the antigen-binding buffer.

[0047] In the methods described herein, salt was used to suppress long-range electrostatic interactions of proteins so that short-range self-associations can be captured in dilute conditions to predict viscosity at high concentrations. As a result, the contribution of overall short-range interactions is being intentionally magnified, whereas electrostatic interactions (both long- and short-range) are being heavily suppressed. This is not an exact representation of complex molecular interactions at high concentrations where the interplay of electrostatic interactions, short-range interactions and others together govern viscosity. Despite these caveats, k_D, isoNaci well predicts viscosity based on the data. While not wish to be bound by any specific theory, this result implies that 1 ) short-range interactions are key to the viscosity of concentrated mAb solutions; 2) electrostatic interactions are probably comparable across the mAbs tested in this study described herein because of their high pl and strong positive charges, therefore they are probably not the determining factor in differentiating viscosities.

[0048] In some embodiments, measurement of the optimized k_D is used to select candidate molecules with high k_D, isoNaci values that would translate to low viscosities at high concentrations. However, because k_D, isoNaci is somewhat biased towards short-range interactions and the electrostatics contribution is suppressed due to charge screening, both false positive (i.e. high ko. isoNaci but high viscosity) and false negative (i.e. low k_D, isoNaci but low viscosity) scenarios may occur. In the case of false positive, it is possible that k_D, isoNaci might fail to flag out candidates that have high viscosity issues primarily driven by electrostatic attractions, since the addition of 150 mM salts would screen out charge attractions and result in a high k_D, isoNaci value if other short-range interactions are minimal. This can be an issue in principle, but practically this will be rare or at least can be mitigated for the following two reasons: 1 ) those candidates with charge-driven viscosity issues are often flagged or deselected during an earlier stage of molecular selection because they would most likely have relatively low pl values; 2) ko measurement in common antigen-binding protein buffer will likely be able to flag out those candidates due to their charge driven self-associations. Potentially using a combination of k_D and k_D, isoNaci will be a more effective strategy to identify these viscous mAbs rather than relying on one individual measurement. On the other hand, low k_D, isoNaci may not always translate to high viscosity resulting in false negative selection. This can happen if mAbs exhibit both strong electrostatic repulsions and strong short-range attractions with the former being stronger. Under charge screening conditions, these molecules will show low k_D. isoNaci driven by short-range interactions. At higher concentrations, although molecules are in closer proximity, strong charge repulsion may still outweigh short-range attractions, resulting a low viscosity. Although these mAbs do not show a red flag in terms of viscosity, this behavior might imply a strong tendency to aggregate under charge screening conditions and therefore they may have undesirable PK profiles upon SC injections when mAbs experience neutral pH and charge screening in the SC space. Therefore, k_D, isoNaci will still be meaningful from a broader developability perspective.

[0049] It is encouraging to see a strong correlation between k_D, isoNaci and viscosity (Figure 6). The experiments described herein provide clear evidence that first-order interactions can well predict viscosity, although viscosity is a result of summation of 1 st order and higher order interactions. This optimized ko approach (i.e. ko, isoNaci ) works well for determining mAb viscosity at high concentrations in an exemplary antigen-binding protein formulation. For example, the optimized buffer condition of adding 150 mM NaCI in formulations having a relatively low pH of 5.2. This approach would be applicable for other common mAb formulations, e.g. histidine pH 6.0, but minor optimization of salt concentration may be required to achieve optimal predictability of k_D. Since net positive charges would be less on mAb surfaces at pH 6.0 than at pH 5.2, less salt might be needed to suppress long-range charge repulsions in higher pH antigen-binding protein formulations.

Methods of Measuring the Diffusion Coefficient

[0050] The disclosed methods of determining viscosity are based on measuring the diffusion coefficient (D) of a plurality of compositions. The “compositions” are samples of the pharmaceutical formulation which comprise different concentrations of the antigen-binding protein in order to predict the k_D wherein the k_D correlates with the predicted viscosity of the pharmaceutical formulation. This method allows for determining the viscosity early in the development process; and allows for determining viscosity when the amount of antigen-binding product is limited, such as early in the development process. [0051] In various embodiments, the diffusion coefficient (D) of the compositions can be measured using methods known in the art. For example, in certain embodiments, D is determined using dynamic light scattering (DLS) which measured the intensity fluctuation of light scatted by particles. The D may also be measured using Fluorescence Correlation Spectroscopy (Yu et aL, Front. Phys. 9:110, 2021), which measures the temporal autocorrection of the detected fluorescence signal emitted by the volume of liquid. In addition, macroscopic mass transfer may be used to measure the D, such as methods based on Taylor dispersion within a Poiseuille flow, or the diaphragm cell may be used to measure the D. Direct detection of D may be carried out by measuring the concentration spatial profile using optical techniques within a diffusion chamber (Hamada & de Anna, T ransport in Porous Media 146: 463-474, 2023). Furthermore, NMR may be used to measure D, e.g. as taught in Patil et aL, AAPS J. 2017 Nov; 19(6): 1760-1766.

[0052] The measurement of D allows for the calculation of D which is indicative of viscosity of a formulation which comprises a high concentration of an antigen-binding protein. For example, k_D is calculated using the formula D = D_o (1 + k_D * c), wherein D is the diffusion coefficient of samples, D_o is the diffusion coefficient of mAb at infinite dilution, and c is protein concentration. Dividing the slope (D_o* ko) by the intercept (D_o) returns ko. The plurality of compositions used in the disclosed methods comprise the pharmaceutical formulation but have different concentrations of the antigen-binding protein in order to calculate k_D.

[0053] As used herein, the term “pharmaceutical formulation” relates to a formulation which is suitable for administration to a subject in need thereof. The terms "subject" or "individual" or "animal" or "patient" are used interchangeably herein to refer to any subject, particularly a mammalian subject, for whom administration of the pharmaceutical formulation of the invention is desired. Mammalian subjects include humans, non-human primates, dogs, cats, guinea pigs, rabbits, rats, mice, horses, cattle, cows, and the like, with humans being preferred. The pharmaceutical formulation of the present disclosure is stable and pharmaceutically acceptable, i.e., capable of eliciting the desired therapeutic effect without causing significant undesirable local or systemic effects in the subject to which the pharmaceutical formulation is administered. Pharmaceutically acceptable formulations of the invention may be sterile. Specifically, the term "pharmaceutically acceptable" can mean approved by a regulatory agency or other generally recognized pharmacopoeia for use in animals, and more particularly in humans, but is not limited to those approved by a regulatory agency. [0054] Various aspects of the pharmaceutical formulations are described below. The use of section headings is merely for the convenience of reading, and not intended to be limiting per se. The entire document is intended to be viewed as a unified disclosure, and it should be understood that all combinations of features described herein are contemplated.

Antigen-Binding Proteins

[0055] An “antigen-binding protein” is a protein comprising a domain that binds a specified target antigen (such as CD3 and/or CDH19, MSLN, DLL3, FLT3, EGFRvlll, BCMA, PSMA, CD33, CD19, CD70, CLDN18.2 or MUC17). An antigen-binding protein comprises a scaffold or framework portion that allows the antigen binding domain to adopt a conformation that promotes binding of the antigen-binding protein to the antigen. In exemplary aspects, the antigen-binding protein is an antibody or immunoglobulin, or an antigen-binding antibody fragment.

[0056] The term "antibody" refers to an intact antigen-binding immunoglobulin. An "antibody" is a type of an antigen-binding protein. The antibody can be an IgA, IgD, IgE, IgG, or IgM antibody, including any one of lgG1 , lgG2, lgG3 or lgG4. In various embodiments, an intact antibody comprises two full-length heavy chains and two full-length light chains. An antibody has a variable region and a constant region. In IgG formats, a variable region is generally about 100-110 or more amino acids, comprises three complementarity determining regions (CDRs), is primarily responsible for antigen recognition, and substantially varies among other antibodies that bind to different antigens. A variable region typically comprises at least three heavy or light chain CDRs (Kabat et al., 1991 , Sequences of Proteins of Immunological Interest, Public Health Service N.I.H., Bethesda, Md.; see also Chothia and Lesk, 1987, J. Mol. Biol. 196:901 -917;

Chothia et al., 1989, Nature 342: 877-883), within a framework region (designated framework regions 1 -4, FR1 , FR2, FR3, and FR4, by Kabat et al., 1991 ; see also Chothia and Lesk, 1987, supra). The constant region allows the antibody to recruit cells and molecules of the immune system.

[0057] In some embodiments, the antibody of the pharmaceutical formulation is a bispecific antibody, i.e. , an antibody that binds two different targets (e.g., CD3 and a second, different target).

[0058] The term “bispecific” as used herein refers to an antibody construct that binds to two different target antigens, i.e., it comprises a first binding domain and a second binding domain, wherein the first binding domain binds to one antigen or target (e.g., the target cell surface antigen), and the second binding domain binds to another antigen or target (e.g. CD3). Accordingly, antibody constructs according to the disclosure comprise specificities for two different antigens or targets. The term “target cell surface antigen” refers to an antigenic structure expressed by a cell and which is present at the cell surface such that it is accessible for an antibody construct as described herein. It may be a protein, preferably the extracellular portion of a protein, or a carbohydrate structure, preferably a carbohydrate structure of a protein, such as a glycoprotein. It is preferably a tumor antigen. The invention also encompasses multispecific antibody constructs such as trispecific antibody constructs, the latter ones including three binding domains, or constructs having more than three (e.g. four, five...) specificities.

[0059] Bispecific antibodies and/or antibody constructs as understood herein include, but are not limited to, traditional bispecific immunoglobulins (e.g., BsIgG), IgG comprising an appended antigen-binding domain (e.g., the amino or carboxy termini of light or heavy chains are connected to additional antigen-binding domains, such as single domain antibodies or paired antibody variable domains (e.g., Fv or scFv)), BsAb fragments (e.g., bispecific single chain antibodies), bispecific fusion proteins (e.g., antigen binding domains fused to an effector moiety), and BsAb conjugates. See, e.g., Spiess et al., Molecular Immunology 67(2) Part A: 97- 106 (2015), which describes various bispecific formats and is hereby incorporated by reference. Examples of bispecific constructs include, but are not limited to, diabodies, single chain diabodies, tandem scFvs, bispecific T cell engager (BiTE®) format (a fusion protein consisting of two single-chain variable fragments (scFvs) joined by a linker), and Fab2 bispecifics, as well as engineered constructs comprising full length antibodies. See, e.g., Chames & Baty, 2009, mAbs 1 [6]:1 -9; and Holliger & Hudson, 2005, Nature Biotechnology 23[9]:1126-1 136; Wu et al., 2007, Nature Biotechnology 25[11 ]:1290-1297;Michaelson et al., 2009, mAbs 1 [2]:128-141 ; International Patent Publication No. 2009032782 and 2006020258; Zuo et al., 2000, Protein Engineering 13[5]:361 -367; U.S. Patent Application Publication No. 20020103345; Shen et aL, 2006, J Biol Chem 281 [16]:10706-10714; Lu et al., 2005, J Biol Chem 280[20]:19665-19672; and Kontermann, 2012 MAbs 4(2):182, all of which are expressly incorporated herein.

[0060] In some embodiments, the pharmaceutical formulations described herein comprise a bispecific antibody construct comprises a first binding domain that binds to a target cell surface antigen, a second binding domain that binds to human CD3 on the surface of a T cell, and optionally a third domain comprising, in an amino to carboxyl order, hinge-CH2 domain-CH3 domain-linker-hinge-CH2 domain-CH3 domain. In some embodiments, each of the first and second binding domains comprise a VH region and a VL region. [0061] The term "binding domain" as used herein refers to a domain which (specifically) binds to I interacts with I recognizes a given target epitope or a given target site on the target molecules (antigens), e.g. CDH19, MSLN, DLL3, FLT3, EGFRvlll, BCMA, PSMA, CD33, CD19, CD70, CLDN18.2 or MUC17 and CD3, respectively. The structure and function of the first binding domain (recognizing e.g. CDH19, MSLN, DLL3, FLT3, EGFRvlll, BCMA, PSMA, CD33, CD19, CD70, CLDN18.2 or MUC17) and preferably also the structure and/or function of the second binding domain (recognizing CD3), is/are based on the structure and/or function of an antibody, e.g. of a full-length or whole immunoglobulin molecule and/or is/are drawn from the variable heavy chain (VH) and/or variable light chain (VL) domains of an antibody or fragment thereof. Preferably, the first binding domain is characterized by the presence of three light chain CDRs (i.e. CDR1 , CDR2 and CDR3 of the VL region) and/or three heavy chain CDRs (i.e. CDR1 , CDR2 and CDR3 of the VH region). The second binding domain preferably also comprises the minimum structural requirements of an antibody which allow for the target binding. More preferably, the second binding domain comprises at least three light chain CDRs (i.e. CDR1 , CDR2 and CDR3 of the VL region) and/or three heavy chain CDRs (i.e. CDR1 , CDR2 and CDR3 of the VH region). It is envisaged that the first and/or second binding domain is produced by or obtainable by phage-display or library screening methods rather than by grafting CDR sequences from a pre-existing (monoclonal) antibody into a scaffold.

[0062] In some embodiments, the first binding domain which binds to the target cell surface antigen and/or the second binding domain which binds to CD3E is/are human binding domains. Antibodies and antibody constructs comprising at least one human binding domain avoid some of the problems associated with antibodies or antibody constructs that possess non-human such as rodent {e.g. murine, rat, hamster or rabbit) variable and/or constant regions. The presence of such rodent derived proteins can lead to the rapid clearance of the antibodies or antibody constructs or can lead to the generation of an immune response against the antibody or antibody construct by a patient. In order to avoid the use of rodent derived antibodies or antibody constructs, human or fully human antibodies I antibody constructs can be generated through the introduction of human antibody function into a rodent so that the rodent produces fully human antibodies.

[0063] In some embodiments, the antigen binding protein comprises a single chain antibody construct. An scFv comprises a variable heavy chain, an scFv linker, and a variable light domain. Optionally, the C-terminus of the variable light chain is attached to the N-terminus of the scFv linker, the C-terminus of which is attached to the N-terminus of a variable heavy chain (N-vh-lin ker-vl-C), although the configuration can be switched (N-vl-linker-vh-C). Alternatively, the C-terminus of the variable heavy chain is attached to the N-terminus of the scFv linker, the C-terminus of which is attached to the N-terminus of a variable light chain (N-vl-linker-vh-C), although the configuration can be switched (N-vh-linker-v-C). Thus, specifically included in the depiction and description of scFvs are the scFvs in either orientation.

[0064] The at least two binding domains and the variable domains (VH/VL) of the antibody construct of the present disclosure may or may not comprise peptide linkers (spacer peptides). The term “peptide linker” comprises in accordance with the present invention an amino acid sequence by which the amino acid sequences of one (variable and/or binding) domain and another (variable and/or binding) domain of the antibody construct of the disclosure are linked with each other. The peptide linkers can also be used to fuse the third domain to the other domains of the antibody construct of the invention. An essential technical feature of such peptide linker is that it does not comprise any polymerization activity. Among the suitable peptide linkers are those described in U.S. Patents 4,751 ,180 and 4,935,233 or WO 88/09344, the disclosure of which are incorporated herein by reference in their entireties. The peptide linkers can also be used to attach other domains or modules or regions (such as half-life extending domains) to the bispecific antibody construct described herein.

[0065] In some embodiments, the third domain comprises a “Fc” or “Fc region” or “Fc domain,” which refers to the polypeptide comprising the constant region of an antibody excluding the first constant region immunoglobulin domain. Thus, “Fc domain” refers to the last two constant region immunoglobulin domains of IgA, IgD, and IgG, the last three constant region immunoglobulin domains of IgE and IgM, and the flexible hinge N-terminal to these domains. For IgA and IgM, Fc may include the J chain. For IgG, the Fc domain comprises immunoglobulin domains Cy2 and Cy3 (Cy2 and Cy3) and the lower hinge region between Cy1 (Cy1 ) and Cy2 (Cy2). The bispecific antibody construct is preferably an IgG antibody (which includes several subclasses, including, but not limited to lgG1 , lgG2, lgG3, and lgG4). Although the boundaries of the Fc region may vary, the human IgG heavy chain Fc region is usually defined to include residues C226 or P230 to its carboxyl-terminus, wherein the numbering is according to the EU index as in Kabat. In some embodiments, amino acid modifications are made to the Fc region, for example, to alter binding to one or more FcyR receptors or to the FcRn receptor.

[0066] In some embodiments, the pharmaceutical formulations described herein comprise a bispecific antibody construct which binds human CD3 and human CDH19, or human CD3 and human MSLN, or human CD3 and human DLL3, or human CD3 and human FLT3, or human CD3 and human EGFRvI 11 , or human CD3 and human BCMA, or human CD3 and PSMA, or human CD3 and human CD33, or human CD3 and human CD19, human CD3 and human CD70, or human CD3 and human MUC17, or human CD3 and human CLDN18.2.

[0067] In some embodiments, the pharmaceutical formulation comprises an antigen-binding protein in a concentration ranging from about 2 mg/ml to about 20 mg/ml, about 2 to about 19 mg/ml, about 2 mg/ml to about 18 mg/ml, about 2 mg/ml to about 17 mg/ml, 2 mg/ml to about 16 mg/ml, about 2 mg/ml to about 15 mg/ml, about 5 mg/ml to about 20 mg/ml, about 5 to about 19 mg/ml, about 5 mg/ml to about 18 mg/ml, about 5 mg/ml to about 17 mg/ml, about 5 mg/ml to about 16 mg/ml, about 5 mg/ml to about 15 mg/ml, about 10 mg/ml to about 20 mg/ml, about 10 to about 19 mg/ml, about 10 mg/ml to about 18 mg/ml, about 10 mg/ml to about 17 mg/ml, about 10 mg/ml to about 16 mg/ml, about 10 mg/ml to about 15 mg/ml, about 3 mg/ml to about 10 mg/ml, about 3 mg/ml to about 12 mg/ml, about 3 mg/ml to about 15 mg/ml, 3 mg/ml to about 18 mg/ml or 3 mg/ml to about 20 mg/ml, 10 mg/mL to about 50 mg/mL (or from about 10 mg/mL to about 20 mg/ml_ or from about 15 mg/mL to about 20 mg/mL).

[0068] In some embodiments, the pharmaceutical formulation comprises an antigen-binding protein at a concentration of about 2 mg/ml, about 3 mg/ml, about 4 mg/mL, about 5 mg/mL, about 6 mg/mL, about 7 mg/mL, about 8 mg/mL, about 9 mg/mL, aboutIO mg/mL, about 1 1 mg/mL, about 12 mg/mL, about 13 mg/mL, about 14 mg/mL, about 15 mg/mL, about 16 mg/mL, about 17 mg/mL, about 18 mg/mL, about 19 mg/mL, about 20 mg/mL, about 25 mg/mL, about 30 mg/mL, about 35 mg/mL, about 40 mg/mL, about 45 mg/mL or about 50 mg/mL. In some embodiments, the pharmaceutical formulation comprises an antigen-binding protein in a concentration of about 20 mg/mL.

Buffers

[0069] The pharmaceutical formulation of the invention comprises a buffer, which optionally may be acetate, lactate, aspartate, glutamate, citrate, succinate, tartrate, fumarate, maleate, histidine, phosphate, 2-(N-morpholino)ethanesulfonate or combinations thereof.

[0070] Buffering agents are often employed to control pH in the pharmaceutical formulation. In some embodiments, the buffer is added in a concentration that maintains pH of the pharmaceutical formulation of about 4.0 to about 6.0, about 4.0 to 5.0, or about 4.2. The effect of pH on pharmaceutical formulations may be characterized using any one or more of several approaches such as accelerated stability studies and calorimetric screening studies (Remmele R.L. Jr., et al., Biochemistry, 38(16): 5241-7 (1999)).

[0071] The buffer system present in the pharmaceutical formulation is selected to be physiologically compatible and to maintain a desired pH. The buffer may be present at a concentration between about 0.1 mM and about 1000 mM (1 M), or between about 5 mM and about 200 mM, or between about 5 mM to about 100 mM, or between about 10 mM and 50 about mM. Suitable buffer concentrations encompass concentrations of about 200 mM or less. In some embodiments, the buffer in the pharmaceutical formulation is present in a concentration of about 190 mM, about 180 mM, about 170 mM, about 160 mM, about 150 mM, about 140 mM, about 130 mM, about 120 mM, about 110 mM, about 100 mM, about 80 mM, about 70 mM, about 60 mM, about 50 mM, about 40 mM, about 30 mM, about 20 mM, about 10 mM or about 5 mM. In some embodiments, the concentration of the buffer is at least 0.1 , 0.5, 0.7, 0.8 0.9, 1.0, 1.2, 1.5, 1.7, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1 1 , 12, 13, 14, 15, 16, 17, 18, 19, 20, 30, 40, 50, 60, 70, 80, 90, 100, 200, 500, 700, or 900 mM. In some embodiments, the concentration of the buffer is between 1 , 1.2, 1.5, 1.7, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, 15, 16, 17, 18, 19, 20, 30, 40, 50, 60, 70, 80, or 90 mM and 100 mM. In some embodiments, the concentration of the buffer is between 5, 6, 7, 8, 9, 10, 1 1 , 12, 13, 14, 15, 16, 17, 18, 19, 20, 30, or 40 mM and 50 mM. In some embodiments, the concentration of the buffer is about 10 mM.

Surfactants

[0072] The pharmaceutical formulations described here comprise a surfactant. Exemplary surfactants include but are not limited to, polysorbate 20, polysorbate 40, polysorbate 60, polysorbate 80, poloxamer 188, poloxamer 407, triton X-100, polyoxyethylene, PEG 3350, PEG 4000, or a combination thereof.

[0073] Pharmaceutical formulations described herein comprise at least one surfactant, either individually or as a mixture in different ratios. In some embodiments, the pharmaceutical formulation comprises a surfactant at a concentration of about 0.001 % to about 5% w/v (or about 0.001% to about 0.5%, or about 0.004 to about 0.5% w/v or about 0.001 to about 0.01 % w/v or about 0.004 to about 0.01 % w/v). In some embodiments, the pharmaceutical formulation comprises a surfactant at a concentration of at least 0.001 , at least 0.002, at least 0.003, at least 0.004, at least 0.005, at least 0.007, at least 0.01 , at least 0.05, at least 0.1 , at least 0.2, at least 0.3, at least 0.4, at least 0.5, at least 0.6, at least 0.7, at least 0.8, at least 0.9, at least 1 .0, at least 1 .5, at least 2.0, at least 2.5, at least 3.0, at least 3.5, at least 4.0, or at least 4.5% w/v. In some embodiments, the pharmaceutical formulation comprises a surfactant at a concentration of about 0.001% to about 0.5% w/v. In some embodiments, the pharmaceutical formulation comprises a surfactant at a concentration of about 0.001 to about 0.01% w/v. In some embodiments, the pharmaceutical formulation comprises a surfactant at a concentration of about 0.001 to about 0.01% w/v. In some embodiments, the pharmaceutical formulation comprises a surfactant at a concentration of about 0.001%, about 0.002%, about 0.003%, about 0.004%, about 0.005%, about 0.006%, about 0.007%, about 0.008%, about 0.009%, about 0.01%, about 0.05%, about 0.1%, about 0.2%, about 0.3%, about 0.4%, to about 0.5% w/v. In some embodiments, the pharmaceutical formulation comprises a surfactant incorporated in a concentration of about 0.001 % to about 0.01% w/v. In some embodiments, the surfactant is polysorbate 80 and the polysorbate 80 is present in a concentration of about 0.01% w/v.

Saccharides

[0074] The pharmaceutical formulations described herein comprise a saccharide. In some embodiments, the saccharide is a monosaccharide or a disaccharide. In some embodiments, the saccharide is glucose, galactose, fructose, xylose, sucrose, lactose, maltose, trehalose, sorbitol, mannitol or xylitol or a combination thereof.

[0075] In some embodiments, the pharmaceutical formulation comprises a saccharide at a concentration of about 0.01% to about 40% w/v, or about 00.1% to about 20% w/v, or about 1% to about 15%, or about 5% to about 12 %, or about 7% to about 12% w/v. In some embodiments, the pharmaceutical formulation comprises at least one saccharide at a concentration of at least 0.5%, at least 1%, at least 2%, at least 3%, at least 4%, at least 5%, at least 6%, at least 7%, at least 8%, at least 9%, at least 10%, at least 11%, at least 12%, at least 13%, at least 14%, at least 15%, at least 16%, at least 17%, at least 18%, at least 19%, at least 20%, at least 30%, or at least 40% w/v. In some embodiments, the pharmaceutical formulation comprises at least one saccharide at a concentration of about 1%, about 2%, about 3%, about 4%, about 5%, about 6%, about 7%, about 8%, about 9%, about 10%, about 11%, about 12%, about 13%, about 14%, or about 15% w/v. In some embodiments, the pharmaceutical formulation comprises at least one saccharide at a concentration of about 1% to about 15% w/v. In a yet further embodiment, the pharmaceutical formulation comprises at least one saccharide at a concentration of about 7%, about 7.5%, about 8%, about 8.5%, about 9%, about 9.5%, about 10%, about 10.5%, about 1 1%, about 11 .5%, or about 12% w/v. In some embodiments, the pharmaceutical formulation comprises at least one saccharide at a concentration of about 7% to about 12% w/v. In some embodiments, the at least one saccharide is in the pharmaceutical formulation at a concentration of about 9% w/v. In some embodiments, the saccharide is sucrose and is present in the pharmaceutical formulation ranging from about 9% to about 12% w/v.

[0076] In a preferred embodiment, the pharmaceutical formulation comprises 10 mM glutamate, 9% (w/V) sucrose and 0.01% (w/V) polysorbate 80, wherein the pH of the pharmaceutical formulation is 4.2. In some embodiments, the pharmaceutical formulation is lyophilized.

EXAMPLES

Example 1 - Materials and Methods

Materials

[0077] In the experiments described herein, three sets of mAbs were tested. The first set consisted of 10 well-characterized mAbs with large material availability (Table 1), and they were used as a “training” sample set to screen different salt conditions for k_D measurement and identify the optimal salt condition in which D shows the greatest correlation with the viscosity of concentrated mAbs. Due to a low number of IgG 1 molecules (only 6) available at the time of testing, four additional lgG2 molecules were also included in the sample set. The other two mAb “test” sets (A and B) were analyzed to validate the optimized method. Test set A is an engineering panel of 10 lgG1 molecules with a high level of CDR sequence similarity. Test set B has 17 lgG1 mAbs with a broad viscosity range at high concentrations. Among these 17 molecules, 12 molecules are conventional monospecific bivalent lgG1 mAbs, whereas the rest 5 are multi-specific IgG 1 molecules (e.g., 3 bispecific molecules with scFv domains; one IgG peptide conjugate; one antibody cytokine conjugate).

Sample Preparation

[0078] A dilution series from 2 to 12 mg/mL was prepared in a buffer having a pH of 5.2 with and without salts. For the “training” mAb set, two types of salt were screened: NaCI (10, 20, 100, 150 mM) and ammonium sulfate (8.33, 50 mM). The optimal buffer condition was identified and validated using two additional panel sets. Protein concentrations were measured using a Droplet Quant UV-Vis spectrophotometer (Perkin Elmer, Waltham, MA) prior to DLS k_D analysis.

Dynamic light scattering (DLS)

[0079] Samples were loaded into an Aurora 384-well clear bottom plate, and the plate was centrifuged at 3000 rpm for 5 min to remove air bubbles. An adhesive film (Microseal C PGR Plate Sealing Film, BioRad) was used to seal the plate and to prevent sample evaporation during measurement. Each sample was measured with 5 acquisitions and 10 second per acquisition in a DynaPro plate reader (Wyatt Technology, Santa Barbara, CA). Sample plates were pre-equilibrated at 25 ^eC for at least 20 min before data collection and measurement was performed at 25 ^SC. Diffusion coefficient was plotted against protein concentration, and data fit to a simple linear function as below:

D = Do (1 + k_D * c) where D is diffusion coefficient of samples, D_o is diffusion coefficient of mAb at infinite dilution. kD is diffusion interaction parameter, and c is protein concentration. Dividing the slope (D₀*kD) by the intercept (D_o) returns ko.

Viscosity Measurement

[0080] All viscosity measurements were performed at 25 °C using a DHR-2 cone and plate rheometer (TA Instruments, New Castle, Delaware) equipped with a 20 mm anodized aluminum geometry at 1 .988° angle. This configuration required approximately 80 pL for each measurement. Reported values represent the viscosity at a shear rate of 1000 s’¹. Viscosity data was plotted as a function of protein concentration (70- 150 mg/mL or higher) using the exponential fit equation defined by Excel. The exponential fit equation is generated by the following equation: y=a*e^A(b*x) where y is the viscosity, x is the concentration, and a and b are the least-squares fitted parameters. The expected viscosities at 140mg/ml_ in an exemplary antigen-binding formulation (denoted herein as “Buffer A”) were then calculated using the exponential fit equation.

[0081]

Example 2 - Improvement in the Predictive Power of ko

[0082] ko measured in an exemplary pharmaceutical formulations having a pH of 5.2 (Buffer A) by dynamic light scattering shows only weak correlation with viscosity at 140 mg/mL. Therefore, the present experiment was carried out to improve the predictability of k_D by screening the buffer conditions in which k_D was measured. To achieve this goal, 10 well characterized mAbs with a large viscosity range (5.8-82.5 cP) at 140 mg/mL at 25 ^eC (Table 1) were used.

Table 1 : List of the 10 mAbs used to optimize the DLS k_D method for viscosity prediction

[0083] In Buffer A, all 10 molecules showed an increase in diffusion coefficient as protein concentration rises (i.e. positive k_D values), indicating net repulsive interactions. These 10 mAbs have pl values at least 2 pH unit higher than the solution pH of 5.2 and therefore carry strong net positive charges, resulting in electrostatic repulsions and positive ko. All the 10 molecules had Do (diffusion coefficient of the mAb at infinite dilution) around 4*10‘⁷ cm²/s, which agrees with the diffusion coefficient of a typical mAb monomer in Buffer A under ideal condition (Figures 1 ). As salt (NaCI or ammonium sulfate) concentration increased, k_D values gradually decreased and eventually all became negative, suggesting net intermolecular interactions shift from repulsive to attractive upon suppression of charge repulsion with salt (Figure 2). This observation implies that ko can potentially capture short-range interactions under charge screening conditions.

[0084] Next, the measured k_D measured in the presence of increasing salt levels was plotted against viscosity (in Buffer A) on a logarithmic scale. For ko measured in Buffer A, the overall correlation is poor (R² = 0.22) (Figure 3A). Although the two molecules with the highest k_D values (> 40 mL/g) show low viscosities (<10 cP), the differentiation in viscosity was minimal for molecules with ko values between 10-30 mL/g. The 2 most viscous molecules otherwise showed intermediate ko values (22.9 and 28.5 mL/g). Therefore, ko data failed to flag the viscous molecules, and this observation supports that k_D measured in Buffer A cannot capture key short-range interactions that are apparently manifested in concentrated mAb solutions. As expected, this result suggests k_D measured in the relatively low pH and low ionic strength Buffer A buffer poorly predicts viscosity. As the NaCI concentration increased in the buffer used for k_D measurement, the correlation between ko and viscosity gradually improved. This improvement in correlation seemed to plateau around 100-150 mM NaCI, therefore higher NaCI concentrations were not tested. Eventually, a 4-fold increase in the correlation (R² from 0.22 to 0.90) between k_D and viscosity was achieved by adding 150 mM NaCI in the buffer when measuring k_D.

[0085] In addition, k_D in Buffer A was measured in the presence of ammonium sulfate to determine whether a better prediction can be achieved than using NaCI. Two concentrations of ammonium sulfate were tested: 8.33 and 50 mM, which are equivalent to 25 and 150 mM NaCI in terms of ionic strength, respectively. Similar to NaCI, the addition of ammonium sulfate gradually improved the increasing correlation between k_D and viscosity (Figures 3G and 3H). However, the best prediction for viscosity was still observed for k_D measured in Buffer A with 150mM NaCI (Figure 3I). Therefore, Buffer A with 150 mM was selected as the optimized buffer condition for DLS k_D measurement, and the diffusion interaction coefficient measured under this condition was designated as kD,i50Naci-

Example 3 - Method Validation

[0086] The optimized k_D method for viscosity prediction described in Example 1 was validated by testing an engineering panel of 10 lgG1 mAbs because screening of an engineering panel is a routine task during the lead molecule optimization stage. In the engineered panel, the mAbs differ by only a few amino acid residues in the CDR region but show diverse viscosities ranging from 7.5 cP to 36.1 cP at 140 mg/mL in Buffer A. All candidates exhibit relatively large positive k_D values in Buffer A, indicative of net self-repulsion (Figure 4A). Although the most viscous molecule (mAb A5) did have the lowest k_D, the second most viscous candidate (mAb A6) has a kD value comparable to those of low viscosity molecules (Figure 4C). In addition, the correlation for low viscosity molecules (< 10 cP) was rather poor. In contrast, all candidates exhibited negative kD. isoNaci and showed attractive self-interactions (Figure 4B). Interestingly, candidate mAb A5 was found to have an apparent D_o ~ 2.2x10-7 cm²/s in the presence of 150 mM NaCI, which is nearly half of the typical value for mAbs (~4.0x10-7 cm²/s) (Figure 4B). This is apparently not because of pre-existing aggregates in the sample, as evidenced by its D_o close to 4.0x1 O'⁷ cm²/s in the absence of NaCI show in Figure 4A. Instead, this result suggests that candidate mAb A5 had a high propensity to self-associate in the presence of 150 mM NaCI even at concentrations well below the DLS sample concentration range. Therefore, kD,i₅oNaci for this molecule probably reflects higher order interactions, and it is not directly comparable to other candidates and is considered as an outlier (Figure 4D). The overall Pearson R² was between kD,i50Naci and viscosity improved to 0.64 or 0.87 (if mAb A5 is considered as an outlier and excluded).

[0087] The modified method further tested our approach to determine whether this method can be applied to screen mAb molecules with even broader viscosity ranges (6-220 cP). The 17 mAbs (12 mABs and 5 multispecifics) tested have diverse CDR sequences and target different proteins. The overall correlation between k_D and viscosity was weak in the absence of NaCI, whereas k_D,i₅oNaci showed improved correlation. As shown in Figures 5C and 5D, the logarithm of viscosity showed stronger correlation with kD,i50Naci (R² = 0.75) than with ko measured in Buffer A (R² = 0.30). This supports that this approach is capable of screening mAb candidates with a rather broad viscosity range. Candidates mAb B2 and mAb B11 exhibited lower apparent Do under salt conditions. Candidate mAb B2 also showed a low apparent D_o in Buffer A, and this might imply that there are pre-existing aggregates in the sample resulting a low apparent DO in both buffers. In contrast, candidate mAb B11 showed a normal D_o (~ 4.0x10-7 cm²/s) in Buffer A, therefore this candidate may have a high propensity to self-associate in salt solutions, a similar behavior observed for candidate mAb A5 (Figure 5B).

[0088] Finally, the data provided in Figure 6 the data for all 37 mAbs tested in this study was consolidated. The correlation between k_D and the logarithm of viscosity shows a moderate R² of 0.24. In contrast, the R² for k_D,i50Naci versus viscosity is 0.80, which is a drastic improvement. It was found that viscosity and k_D,i50Naci follows a function below: viscosity_140 mg/ml__25 ^eC = 10 ^A (0.2976 - 0.0759 * k_D, 150NaCl) where viscosity_140 mg/mL_25 ^eC (in cP) is the viscosity measured in Buffer A at 140 mg/mL at 25 ^eC; kD, 150NaCI (in mL/g) is diffusion coefficient measured in Buffer A with 150 mM NaCI at 25 ^eC. The corresponding k_D,i50Naci value for a viscosity of 10, 20, 30 cP is -9.25, -13.22, -15.54 mL/g, respectively. These k_D,i50Naci can be potentially used as filters during candidate screening to select non-viscous molecules. Example 4 - Mechanism of Modified Method

[0089] To rationalize our findings, we propose a mechanism where the effects of both intermolecular distance and ionic strength on PPIs are taken into consideration. The averaged surface-to-surface intermolecular distance (Ds-s) is a function of protein concentration and can be calculated using the following formula [23]:

[0090] Where c is the protein concentration in mg/mL, MW is the molecular weight of a mAb, NA is the Avogadro constant of 6.022x10²³ mol^-1, and D_h is the hydrodynamic diameter of a mAb (a D_h of 10 nm for a typical mAb is used). D_s-S drastically decreases as protein concentration arises. D_s.s ranges from 17.5 to 39.9 nm within the ko concentration regime (2~12 mg/mL) (Figures 7A1 and 7B1 ), whereas D_s.s decreases to 1.84 nm at 140 mg/mL (Figure 7C1 ).

[0091] On the other hand, Debye length (1 /K) is strongly dependent on solution ionic strength following the expression (Duroudier et al.) below:

Where E is relative permittivity of water, s₀ is the permittivity of free space, k_B is Boltzmann’s constant, T is temperature, is ion concentration, z_t is charge on electron. I is the ionic strength in mol/L. Debye length is a measure of the electric double layer thickness, which becomes thicker at lower ionic strength. Debye length is approximately 3.26 nm and 0.76 nm in Buffer A (I = 8.67 mM) and Buffer A + 150 mM NaCI (I = 159 mM), respectively (Figures 7A2, 7B2 & 7C2).

[0092] When kD is measured in Buffer A (Figure 7A), D_s.s of mAbs (17.5-39.9 nm) is far longer than Debye length (3.26 nm), so electric double layers of mAb molecules do not overlap. Due to Brownian motion, mAb molecules may experience intermolecular interactions when they transiently come into close proximity. These mAb molecules carry strong net positive charges at pH 5.2, and mAb molecules can start to experience strong charge repulsions from a rather long distance due to the presence of thick Debye length in the low ionic strength buffer of Buffer A. This in turn prevents mAb molecules from getting close enough to form possible short-range interactions. Consequently, k_D measured in Buffer A reflects primarily global long-range charge repulsion of mAb molecules.

[0093] Based on the explanation above, it seems clear that the long-range global charge repulsion reflected by k_D measured in Buffer A is not a strong predictor of high concentration mAb viscosity predominantly driven by key short-range interactions. A simple but effective approach is to measure k_D under a salt screening condition by adding 150 mM NaCI (Figure 6B). By doing so, the Debye length is reduced to as low as 0.76 nm which is short enough to allow possible short-range interactions to form even in dilute solutions. These short-range interactions reflected by ko, isoNaci are indicative of key short-range interactions driving viscosity, making k_D, isoNaci a strong predictor for viscosity.

Example 5 - Method Results in Low Apparent Do

[0094] In addition to obtaining the valuable ko data, the ko linear fit was also observed and in particular the value of apparent D_o. In theory, as protein concentration approaches zero, PPIs start to disappear and eventually the diffusion coefficient should reach a consensus value for a subclass of IgG molecules. The majority of mAb molecules show a D_o around 4.0x1 O'⁷ cm²/s in Buffer A at 25 ^eC, which is in agreement with the hydrodynamic size of a typical mAb. However, several outliers that have low apparent D_o values (e.g. mAb A5, mAb B11 ) were also observed. There are two reasons to observe such a deviation. The first reason is the existence of non- dissociable aggregates in the sample. Aggregates diffuse more slowly and therefore have smaller diffusion coefficient than monomers. Since DLS directly measures the Z-averaged diffusion coefficient, the presence of aggregates will result in a negative offset in k_D linear plot and eventually a low apparent D_o. This can be potentially solved by performing multi-mode analysis to resolve monomer species from aggregates if the aggregates are at least 5X larger in size than the monomer. However, it is challenging to separate the monomer peak from relatively small aggregates. Therefore, the purity of the sample should be carefully examined prior to DLS k_D analysis. Another reason is related to a molecule’s high propensity to self-associate in the presence of salt solutions. In this case, first-order self-interactions can occur at very low concentrations (< 2 mg/mL) and instead capture higher order interactions within the DLS relevant concentration region (2~20 mg/mL). Fitting such a region will result in a low apparent D_o. This is a red flag from a broad developability perspective because this behavior often correlates with high viscosity of concentrated mAb solutions at least from the data collected. It may also inform aggregation risks upon subcutaneous injections where highly concentrated mAbs are exposed to a high salt concentration at neutral pH.

Example 6 - Potential ways to further reduce sample consumption

[0095] A typical k_D measurement in a DLS plate reader consumes 0.5-1 mg of material for each candidate using a 384-well plate (e.g. 20 pL * (12 + 8 + 6 + 4) pg/pL = 600 pg). Although the material consumption is low, it would be preferable to reduce the material requirement even further given the limited material availability during the early candidate screening stage. One idea is to use a single-concentration approach to approximate k_D. Given that the theoretical D_o of an lgG1 in a given buffer is constant (i.e. an averaged value of 3.93x10^-7 cm2/s in Buffer A), a linear plot can be constructed between DO and the diffusion coefficient measured at a relatively high concentration (e.g. ~12 mg/mL), and kD can be calculated. To test whether this is a viable method, the k_D, isoNaci was calculated using the diffusion coefficient at the highest testing concentration only for the monospecific mAbs. As shown in Figure 9, there is some level of differentiation between low and high viscosity molecules. For example, molecules with k_D, isoNaci > -10 mL/g are likely to exhibit low viscosity at 140 mg/mL (< 10 cP). However, k_D, isoNaci calculated based on single concentration data show inferior correlation with viscosity compared ko, isoNaci derived from the linear fit of a concentration gradient (Figure 6B). In addition, this method may fail to identify some D_o outliers. For single point measurements, the purity of the sample needs to be ensured to avoid the interference from the pre-existing aggregates on diffusion coefficient. Nonetheless, this approach can be considered especially when sample is limited since using a single concentration of 12 mg/mL only consumes 0.25 mg of protein. Another method to reduce sample consumption for ko is to use a capillary based DLS unit which requires a sample volume as low as 2 pL.

Example 7 - Temperature-Dependent High Concentration mAb Viscosity

[0096] A study was carried out to investigate the effect of temperature on viscosity behaviors of high concentration mAb formulations. The viscosity of several mAb antibodies (denoted as mAb C1 , mAb C2, mAB C3 and mAb C4) at different temperatures was measured using the standard assay to measure viscosity. It was determined that the viscosity of mAB C1 and mAb C2 increased exponentially as the temperature was lowered to 5°C.

[0097] To determine if the disclosed optimized k_D method for viscosity prediction can be used to predict viscosity at different temperatures, the method described in Example 1 was used to predict the viscosity of mAbs C1 - C4 at different temperatures (5°C, 10°C, 15°C, 25°C and 37°C). The diffusion coefficient (cm²/s) of the mAb formulation at each temperature was measured at various concentrations (4, 5, 6, 7, 8, 9, 10, 1 1 , 12, 13 and 14 mg/mL) in order to calculate the k_D of the formulation based on the measured diffusion coefficient. Figure 11 provides the linear plot of the diffusion coefficient vs. the concentration of each mAb composition at each temperature with or without 150 mM NaCI added. Figure 12 demonstrates that the temperature dependency of viscosity correlates with the k_D. Therefore, this study demonstrates that the optimized method disclosed herein predicted viscosity of the pharmaceutical formulations comprising antigen-binding proteins at different temperatures.

References

1. Miller, K.L. and M.J.N.r.D.d. Lanthier, Regulatory watch: Innovation in biologic new molecular entities: 1986-2014. 2015. 14(2): p. 83-83.

2. Mullard, A.J.N.r.D.D., FDA approves 100th monoclonal antibody product. 2021 .

3. Warne, N.W.J.E.j.o.p. and biopharmaceutics, Development of high concentration protein biopharmaceuticals: the use of platform approaches in formulation development. 2011. 78(2): p. 208-212.

4. Xu, Y., et al. Structure, heterogeneity and developability assessment of therapeutic antibodies, in MAbs. 2019. Taylor & Francis.

5. Bailly, M., et al. Predicting antibody developability profiles through early stage discovery screening, in MAbs. 2020. Taylor & Francis.

6. Jarasch, A., et al., Developability assessment during the selection of novel therapeutic antibodies. 2015. 104(6): p. 1885-1898.

7. Jain, T., et al., Biophysical properties of the clinical-stage antibody landscape. 2017. 114(5): p. 944-949.

8. Yang, X., et al. Developability studies before initiation of process development: improving manufacturability of monoclonal antibodies, in MAbs. 2013. Taylor & Francis.

9. Ryman, J.T., B.J.C.p. Meibohm, and s. pharmacology, Pharmacokinetics of monoclonal antibodies. 2017. 6(9): p. 576-588.

10. Mathaes, R., et al., Subcutaneous injection volume of biopharmaceuticals — pushing the boundaries. 2016. 105(8): p. 2255-2259. 1. Buck, P.M., et aL, Highly viscous antibody solutions are a consequence of network formation caused by domain-domain electrostatic complementarities: Insights from coarse-grained simulations. 2015. 12(1 ): p. 127-139. 2. Tomar, D.S., et al. Molecular basis of high viscosity in concentrated antibody solutions: strategies for high concentration drug product development, in MAbs. 2016. Taylor & Francis. 3. Connolly, B.D., et aL, Weak interactions govern the viscosity of concentrated antibody solutions: high-throughput analysis using the diffusion interaction parameter. 2012. 103(1): p. 69-78. . Kingsbury, J.S., et aL, A single molecular descriptor to predict solution behavior of therapeutic antibodies. 2020. 6(32): p. eabb0372. 5. Yadav, S., S.J. Shire, and D.S.J.J.o.p.s. Kalonia, Viscosity behavior of high-concentration monoclonal antibody solutions: correlation with interaction parameter and electroviscous effects. 2012. 101 (3): p. 998-1011. 6. Zhou, C., W. Wang, and D. Liu, Formulation optimization for bispecific antibodies. 2021 , Google Patents. 7. Lavoisier, A. and J.-M. Schlaeppi. Early developability screen of therapeutic antibody candidates using Taylor dispersion analysis and UV area imaging detection, in MAbs. 2015. Taylor & Francis. 8. Latunde-Dada, S., et aL, Methodologies for the rapid determination of the diffusion interaction parameter using Taylor dispersion analysis. 2016. 8(2): p. 386-392. 9. Woldeyes, M.A., et aL, How well do low-and high-concentration protein interactions predict solution viscosities of monoclonal antibodies? 2019. 108(1 ): p. 142-154. 0. Dear, B.J., et aL, Contrasting the influence of cationic amino acids on the viscosity and stability of a highly concentrated monoclonal antibody. 2017. 34(1 ): p. 193-207. 1 . Neergaard, M.S., et aL, Viscosity of high concentration protein formulations of monoclonal antibodies of the lgG1 and lgG4 subclass-Prediction of viscosity through protein-protein interaction measurements. 2013. 49(3): p. 400-410. 22. Chari, R., et al., Long-and short-range electrostatic interactions affect the rheology of highly concentrated antibody solutions. 2009. 26(12): p. 2607-2618.

23. Yadav, S., et al., Establishing a link between amino acid sequences and self-associating and viscoelastic behavior of two closely related monoclonal antibodies. 2011. 28(7): p. 1750-1764.

24. Lehermayr, C., et al., Assessment of net charge and protein-protein interactions of different monoclonal antibodies. 2011. 100(7): p. 2551 -2562.

25. Yadav, S., et al., Specific interactions in high concentration antibody solutions resulting in high viscosity. 2010. 99(3): p. 1152-1 168.

26. Duroudier, J.-P., 3 - Solution Activity Coefficients, in Thermodynamics, J.-P. Duroudier, Editor. 2016, Elsevier, p. 107-155.

Claims

What is claimed is:

1 . A method of determining viscosity of a pharmaceutical formulation comprising an antigen-binding protein, the method comprising a) measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein under a condition that reduces the surface charge of the antigen-binding protein, and b) calculate k_D of the formulation based on the plurality of Ds from step a), wherein k_D of the formulation is indicative of the viscosity of the formulation.

2. The method of claim 1 wherein the condition that reduces the surface charge of the antigen protein is the presence of salt in the composition.

3. The method of claim 2, wherein the salt is sodium chloride, potassium chloride, lithium chloride, magnesium chloride, calcium chloride, ammonium chloride, sodium sulfate, ammonium sulfate, potassium sulfate, lithium sulfate, magnesium sulfate, calcium sulfate, sodium phosphate, ammonium phosphate, potassium phosphate, lithium phosphate, magnesium phosphate, calcium phosphate, sodium bromide, ammonium bromide, potassium bromide, lithium bromide, magnesium bromide, calcium bromide, sodium nitrate, ammonium nitrate, potassium nitrate, lithium nitrate, magnesium nitrate, calcium nitrate, sodium chlorate, ammonium chlorate, potassium chlorate, lithium chlorate, magnesium chlorate, calcium chlorate, sodium iodide, ammonium iodide, potassium iodide, lithium iodide, magnesium iodide, calcium iodide, sodium perchlorate, ammonium perchlorate, potassium perchlorate, lithium perchlorate, magnesium perchlorate, calcium perchlorate, sodium thiocyanate, ammonium thiocyanate, potassium thiocyanate, lithium thiocyanate, magnesium thiocyanate, calcium thiocyanate or combination thereof.

4. The method of claim 3, wherein the sodium chloride is at a concentration of about 100 mM to about 500 mM.

5. The method of claim 4, wherein the sodium chloride is at a concentration of about 150 mM.

6. The method of claim 3, wherein the ammonium sulfate is at a concentration of about 5 mM to about 100 mM.

7. The method of claim 3, wherein the ammonium sulfate is at a concentration of about 50 mM.

8. A method of determining viscosity of a pharmaceutical formulation comprising an antigen-binding protein, the method comprising a) measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein under a condition wherein the pH of the composition is near to the pl of the antigen-binding protein, and b) calculate k_D of the formulation based on the plurality of Ds from step a), wherein k_D of the formulation is indicative of the viscosity of the formulation.

9. The method of any one of claims 1 -8 wherein the concentrations of the antigen-binding protein in the plurality of compositions is <20 mg/ml.

10. The method of any of the preceding claims wherein the diffusion coefficient is measured by dynamic light scattering.

11 . The method of any of the preceding claims wherein the formulation comprises a saccharide and a surfactant.

12. The method of claim 1 1 , wherein the formulation further comprises a buffer.

13. The method of claim 12, wherein the buffer is an acetate buffer, a glutamate buffer, a citrate buffer, a lactic buffer, a succinate buffer, a tartrate buffer, a fumarate buffer, a maleate buffer, a histidine buffer, or a phosphate buffer.

14. The method of any one of claims 11 -13, wherein the saccharide is a monosaccharide or a disaccharide.

15. The method of any one of claims 11 -14, wherein the saccharide is glucose, galactose, fructose, xylose, sucrose, lactose, maltose, trehalose, sorbitol, mannitol or xylitol.

16. The method of any one of claims 11 -15, wherein the surfactant is polysorbate 20, polysorbate 40, polysorbate 60, polysorbate 80, poloxamer 188, poloxamer 407, triton X-100, polyoxyethylene, PEG 3350, PEG 4000, or a combination thereof.

17. The method of any of the preceding claims wherein the pharmaceutical formulation has a pH of 5.2.

18. The method of any of the preceding claims wherein the pharmaceutical formulation comprises a 10 mM acetate buffer, sucrose and polysorbate at pH 5.2.

19. The method of any of the preceding claims, wherein the antigen-binding protein is a polypeptide, large peptide, peptide conjugate, antibody, antibody fragment, antibody fusion peptide or antigen-binding fragment thereof.

20. The method of claim 19, wherein the antigen-binding protein is an antibody and the antibody is a polyclonal antibody or monoclonal antibody

21 . The method of claim 20, wherein the antibody is an IgG monoclonal antibody.

22. The method of claim 19, wherein the antigen-binding protein is a bi-specific antibody construct.

23. A method of screening an excipient for preparation of a pharmaceutical formulation comprising an antigen-binding protein, wherein the method comprises determining the viscosity of the pharmaceutical formulation comprising the excipient using the method of any one of claims 1 -22, and comparing the viscosity of the pharmaceutical composition comprising the excipient with the viscosity of the pharmaceutical composition without the excipient.

24. The method of claim 23, further comprising the step of determining the viscosity of the pharmaceutical formulation without the excipient using the method of any one of claims 1-22.

25. The method of claim 23 or 24 wherein a panel of excipients are screened using the method of any claim 23 or 24.

26. A method of screening an excipient for preparation of a pharmaceutical formulation comprising an antigen-binding protein, the method comprising i) determining the viscosity of the pharmaceutical formulation comprising the excipient by: a) measuring a plurality of diffusion coefficients (D) of a plurality of compositions comprising different concentrations of the antigen-binding protein and different concentrations of the excipient under a condition that reduces the surface charge of the antigen-binding protein or under a condition wherein the pH of the composition is near to the pl of the antigen-binding protein, and b) calculating k_D of the formulation based on the plurality of Ds from step a), wherein k_D of the formulation is indicative of the viscosity of the formulation, and ii) comparing the k_D of the formulation comprising the excipient with the k_D of the formulation without the excipient.

27. The method of claim 26, wherein the plurality of diffusion coefficients (D) of a plurality of compositions are measured under a condition that reduces the surface charge of the antigenbinding protein, and the condition that reduces the surface charge of the antigen protein is the presence of salt in the composition.

28. The method of claim 27, wherein the salt is sodium chloride, potassium chloride, lithium chloride, magnesium chloride, calcium chloride, ammonium chloride, sodium sulfate, ammonium sulfate, potassium sulfate, lithium sulfate, magnesium sulfate, calcium sulfate, sodium phosphate, ammonium phosphate, potassium phosphate, lithium phosphate, magnesium phosphate, calcium phosphate, sodium bromide, ammonium bromide, potassium bromide, lithium bromide, magnesium bromide, calcium bromide, sodium nitrate, ammonium nitrate, potassium nitrate, lithium nitrate, magnesium nitrate, calcium nitrate, sodium chlorate, ammonium chlorate, potassium chlorate, lithium chlorate, magnesium chlorate, calcium chlorate, sodium iodide, ammonium iodide, potassium iodide, lithium iodide, magnesium iodide, calcium iodide, sodium perchlorate, ammonium perchlorate, potassium perchlorate, lithium perchlorate, magnesium perchlorate, calcium perchlorate, sodium thiocyanate, ammonium thiocyanate, potassium thiocyanate, lithium thiocyanate, magnesium thiocyanate, calcium thiocyanate or combination thereof.

29. The method of claim 28, wherein the sodium chloride is at a concentration of about 100 mM to about 500 mM.

30. The method of claim 29, wherein the sodium chloride is at a concentration of about 150 mM.

31 . The method of claim 28, wherein the ammonium sulfate is at a concentration of about 5 mM to about 100 mM.

32. The method of claim 31 , wherein the ammonium sulfate is at a concentration of about 50 mM.

33. The method of any one of claims 26-32, wherein the concentrations of the antigenbinding protein in the plurality of compositions is <20 mg/ml.

34. The method of any one of claims 26-33, wherein the diffusion coefficient is measured by dynamic light scattering.

35. The method of any one of claims 26-33, wherein the antigen-binding protein is a polypeptide, large peptide, peptide conjugate, antibody, antibody fragment, antibody fusion peptide or antigen-binding fragment thereof.

36. The method of claim 35, wherein the antigen-binding protein is an antibody and the antibody is a polyclonal antibody or monoclonal antibody

37. The method of claim 36, wherein the antibody is an IgG monoclonal antibody.

38 The method of claim 35, wherein the antigen-binding protein is a bi-specific antibody construct.