Ultrahigh-concentration biologic therapeutics enabled by spray drying with a glassy surfactant excipient

Abstract

Biopharmaceuticals such as peptides and antibodies have become critical to health care. Despite their exceptional potency and specificity, biopharmaceuticals are prone to aggregation, which can limit efficacy. These therapies therefore often require low-concentration formulations as well as cold storage to maintain stability; however, high doses are required to treat many diseases. Most approved protein drug products are administered intravenously, imposing excessive burdens on patients. New approaches are needed to formulate proteins at high concentrations to enable less burdensome subcutaneous injection, preferably with an autoinjector that can be used directly by patients. To address this challenge, we report a subcutaneously injectable protein delivery platform composed of spray-dried protein microparticles suspended in a nonsolvent liquid carrier. These microparticles contain only biopharmaceuticals and a high–glass transition temperature polyacrylamide-derived copolymer excipient that affords key benefits over traditional excipients. First, the excipient improved stabilization of biopharmaceuticals through the spray-drying process, and second, it improved morphology and properties of the spray-dried particles, enhancing suspension injectability. We demonstrated with albumin, human immunoglobulin G, and an anti-COVID monoclonal antibody (IDBiologics) that this technology enables ultrahigh-concentration protein formulations (exceeding 500 milligrams per milliliter) that are injectable through standard needles with clinically relevant injection forces. In addition, experiments with two clinically relevant antibody drugs show that these ultrahigh-concentration formulations reduce required injection volumes without altering pharmacokinetics or efficacy in mice. This approach could nearly triple the number of commercial protein drugs amenable to subcutaneous administration, improving access to these critical biopharmaceuticals.

Editor’s summary

High-dose protein-based therapeutics are frequently delivered intravenously as dilute solutions that avoid protein aggregation. This route of delivery requires trained personnel and clinical infrastructure, which limits the potential use of such drugs. Here, Jons et al. developed a spray-drying manufacturing process for protein therapeutics that uses a small amount of a polymer excipient to stabilize proteins, reduce the potential for aggregation, and facilitate ultrahigh-concentration formulation. They tested this system with model proteins and antibody-based therapeutics, demonstrating that the spray-dried particles can be delivered at a high dose with standard injection forces compatible with standard subcutaneous autoinjector pens. The technology could open the door for home injection of many protein-based drugs that are only available in the clinic. —Molly Ogle

INTRODUCTION

Biopharmaceuticals are an important class of therapeutics used to treat a wide range of diseases including cancer, autoimmune disorders, and viral infections (1, 2). Monoclonal antibodies (mAbs) dominate the biopharmaceutical market, comprising more than 50% of the almost 200 biologic product approvals in the past 4 years and 80% of the global annual sales of protein-based drugs (3). The success of mAbs is attributable to their complex tertiary structure that enables highly selective targeting of key molecules, improving both their efficacy and safety profiles (4). Moreover, the long serum half-life of mAbs (typically around 21 days but can reach to 70 days for some variants) prolongs the time frame over which the drug remains in the therapeutic window (5). Unfortunately, the high molecular weight and delicate tertiary structure of mAbs require them to be delivered parenterally, typically by intravenous or subcutaneous injection (3, 6). To avoid the burden and costs associated with intravenous infusion, many new mAb drug products have been developed, which are administered by subcutaneous injection (2, 7). Although subcutaneous injections are preferable on account of their simplicity and reduced burden, the volume of administration is limited (0.5 to 1.5 ml), requiring that the formulation have a high protein concentration (>100 mg/ml) (5, 8-10) or that the product is coformulated with enzymes (such as hyaluronidase) that degrade the tissue sufficiently to enable higher dosing volumes (11). Furthermore, mAbs are amphiphilic and have hydrophobic regions that are highly susceptible to irreversible denaturation and aggregation in solution (5, 8, 12-14), issues that are exacerbated at elevated concentrations even under benign storage conditions (15, 16). These challenges hinder the adoption and implementation of mAb therapeutics for treatment of many diseases, particularly in under-resourced locations around the world (17-19).

Biologic drug products rely on excipients for adequate formulation stability. Surfactant excipients such as Tween 20 or 80 (also called polysorbates) are present in 94% of US Food and Drug Administration (FDA)–approved high-concentration mAb drug products because they can screen interactions with hydrophobic interfaces (such as air-water or water-plastic), thereby preventing protein denaturation and aggregation. We previously reported the development of a surfactant excipient comprising a polyacrylamide-derived copolymer poly(acryloylmorpholine-co-N-isopropylacrylamide) (MoNi) that exhibits improved stabilization of proteins, such as insulin and mAbs, during stressed aging conditions (including elevated temperature, shaking, and freeze-thawing) (20-24). In addition to these surfactants, sugars, such as trehalose and sucrose, can act as hydrophilic cosolutes to crowd proteins and prevent aggregation (25). Last, buffering additives such as amino acids (histidine and arginine) have been shown to preserve protein structure and reduce formulation viscosity by minimizing electrostatic repulsion between mAbs (26-28). Unfortunately, the optimization and combination of these various classes of excipients have only enabled the commercialization of a few high-concentration mAb drug products (100 to 200 mg/ml) for subcutaneous administration (29), demonstrating a critical need for alternative formulation approaches, which simultaneously achieve sufficiently high protein concentrations (>350 mg/ml) and formulation stability to enable most of the mAb drug products to be dosed subcutaneously.

To circumvent the challenges associated with standard aqueous formulations, an alternative approach using dried proteins suspended in a carrier fluid that does not solubilize the protein, known as a nonsolvent, has recently been explored (30). Drying antibodies, commonly with lyophilization in the presence of various excipients (such as trehalose), can stabilize the protein for years by preventing physical and chemical degradation (31). There are several clinically approved mAb products consisting of lyophilized protein and excipients that are reconstituted at high concentrations just before use (2). Unfortunately, reconstitution for these products can require upward of 30 min, exposing the mAb to nonequilibrium conditions that induce aggregation (5, 8, 9, 32, 33). To avoid reconstitution, it is possible to either mill lyophilizate into injectable microparticles or directly dry protein and excipients into microparticles with spray-drying processes (30). These protein microparticles are extremely stable and can form injectable suspensions when dispersed in a nonsolvent liquid (34-36). Yet, even the spray-dried suspension formulations reported to date have required at least 33 wt % of a high–glass transition temperature (T_g; the temperature at which materials exhibit a transition in their mechanical properties from exhibiting soft and rubbery to hard and glassy behaviors) additive, such as trehalose. These high-T_g additives stabilize the mAbs during the spray-drying process and enable formation of robust, injectable suspensions. As a result, these microparticle suspensions have typically required a high total solid content to match liquid counterparts (300 mg/ml solids for 200 mg/ml mAb). Although dry protein suspensions hold promise for improving the accessibility and delivery of biopharmaceuticals, current formulations have failed to make a clinical impact because of the modest increases in protein concentration.

We hypothesized that our surfactant excipient, MoNi, which has an inherently high T_g on account of its polyacrylamide-derived chemical structure, could provide superior protein stability compared with standard sugar excipients at substantially lower concentrations in microparticle suspensions. Several studies have determined that surfactants are crucial additives for protein stability during spray drying because these molecules preferentially occupy air-water interfaces after aerosolization and exclude proteins from these hydrophobic interfaces more effectively than sugars and amino acids (37-40). Typical surfactants (such as Tween or Pluronics) are polyethylene glycol (PEG)–based and exhibit low T_g values (T_g of −64°C) that negatively affect the mechanical properties of the resulting microparticles (41). Use of these PEG-based excipients generally requires the additional inclusion of high-T_g sugars such as trehalose (T_g of 106°C) to stabilize the particles at the expense of protein loading (41). In contrast, we hypothesized that a high-T_g surfactant such as MoNi may encompass both the crucial functions of protecting proteins from unfavorable interface interactions during aerosolization and enhancing the mechanical properties of the microparticles to simultaneously improve suspension injectability and stability. In this work, we leveraged the excipient MoNi to generate high–protein-content (95 wt %) spray-dried microparticles with both albumin and human immunoglobulin G (hIgG). We demonstrate that this approach enables ultrahigh-concentration (UHC) suspension formulations exceeding 500 mg/ml that are injectable through standard needles with clinically relevant injection forces. The MoNi excipient further stabilizes protein cargo during spray drying and storage as a liquid suspension. In addition, experiments in mice show that these UHC formulations reduce required injection volumes without altering pharmacokinetics. Overall, we demonstrate that the high-T_g excipient MoNi enables readily injectable protein formulations at the highest reported protein concentrations to date, nearly tripling the number of commercial protein drugs amenable to subcutaneous administration. These advances could enable the development of drug products that reduce the need for infusion of certain therapeutics and improve access to critical biopharmaceuticals.

RESULTS

Spray drying of protein microparticles with a high-T_g excipient

In this work, we report the development of a biotherapeutic formulation comprising microparticles of protein and the MoNi excipient suspended in a nonsolvent liquid carrier (for example, triacetin) that is readily administered by injection through standard syringes and needles. We synthesized MoNi as previously reported and characterized the copolymer excipient with size exclusion chromatography (SEC) and ¹H nuclear magnetic resonance (NMR) (figs. S1, A to D, and S2 and table S1) (23). Differential scanning calorimetry (DSC) measurements demonstrated that MoNi exhibits a T_g of 140°C (fig. S3). For biotherapeutic formulation, protein and the excipient can be directly dried from an aqueous solution into microparticles through spray drying (Fig. 1A) (42). In these processes, an aqueous feed solution is atomized to form small droplets, which are then exposed to hot air in the spray-drying chamber, leading to rapid liquid evaporation, and the resultant dried microparticles are separated and collected using a cyclone separator (43, 44). Because only volatile compounds such as the aqueous medium are driven off during the drying process, the bulk composition of the final microparticles is determined by the ratio of protein and excipients added to the initial aqueous feed solution, and the microparticle surface composition is determined by the relative surface activity of protein and excipients (37, 38, 45). With excipients that are more surface active than their protein counterparts, the surface composition can be enriched up to 10-fold (37, 38, 45).

Fig. 1. Spray drying of protein microparticles with the MoNi excipient.

(A) Schematic diagram of protein and MoNi spray-drying process in aqueous solution yielding protein microparticles that can be made into UHC protein suspensions by mixing in nonaqueous liquid. Schematics created with BioRender in Inkscape. (B) Schematic diagram of MoNi surfactant surface enrichment during the drying of atomized droplets. Soluble protein diffuses toward the core, and surface-active surfactants remain at the air-water interface. (C) Glass transition temperature for commercial surfactant Tween 20 compared with the polyacrylamide-derived copolymer surfactant MoNi. (D) Calculation of the T_g of protein microparticles as a function of composition at the surface and core of microparticles using the Fox equation. (E) Estimation of the T_g of the protein surfactant mixture in the microparticles with the Fox equation.

Here, we spray-dried aqueous solutions of either model protein, bovine serum albumin (BSA), or therapeutic protein, hIgG, with the MoNi excipient at a feed ratio of 20:1 (w/w), resulting in microparticles that were ~95 wt % protein and 5 wt % MoNi. Although the resulting microparticles were only 5 wt % MoNi by bulk, the surface composition of the microparticles was expected to be heavily enriched with MoNi on account of its high surface activity (Fig. 1B and fig. S4). When considering a 10-μm protein microparticle and defining a surface depth of 5 nm, only 0.3 wt % loading of a highly surface-active additive would be needed to completely occupy the surface layer. Thus, the MoNi surfactant loading in our system is in roughly 15-fold excess of the amount necessary to dominate the surface composition. At a concentration of 5 wt %, the properties of the microparticle surface will be dominated by the T_g of the surfactant (Fig. 1C). A conservative calculation of the T_g of the surface layer of microparticles prepared with either protein alone, protein with MoNi, or protein with the standard PEG-based surfactant Tween 20 indicated that the MoNi is expected to yield robust, glassy particles at room temperature (25°C), whereas particles formed from either protein alone or with Tween 20 would be soft and tacky (Fig. 1, C to E) (37, 38, 46).

Rheological and stability characterization of a spray-dried model protein

Triacetin was chosen as a nonsolvent liquid carrier in these initial experiments on account of its high density (1.16 g/ml), low viscosity (23 centipoise), low vapor pressure (0.0024 mmHg), and inclusion in FDA-approved parenteral drug products (47, 48). Spray-dried particles prepared with BSA and MoNi and dispersed in triacetin at a concentration of BSA (520 mg/ml) by vortexing formed a stable fluid suspension. In contrast, BSA particles prepared without MoNi and dispersed in triacetin at the same BSA concentration formed a thick and solid paste (Fig. 2, A to D). Rheological characterization demonstrated that the inclusion of MoNi in the spray-dried particles substantially improved the flow properties and injectability of the formulation. The inclusion of MoNi in the BSA particles reduced the storage modulus values of the resulting suspensions by four orders of magnitude as measured by angular frequency sweeps (Fig. 2, A and B). We examined the viscoelastic behavior of the two suspensions by the tan δ (G″/G′) values. The suspension formulation comprising MoNi exhibited a tan δ value of 2.4, indicating a liquid-like behavior, whereas the suspension formulation without MoNi exhibited a value of 0.08, indicating a solid-like behavior similar to a thick paste at 10 rad/s (Fig. 2, A and B). Steady shear flow sweeps, measuring viscosity relative to the shear rate, further demonstrated that the BSA suspensions were shear thinning and that formulations comprising MoNi exhibited both lower viscosities and smoother flow than comparable formulations without MoNi (Fig. 2C). Scanning electron microscopy (SEM) images of spray-dried microparticles illustrated that BSA microparticles spray-dried both with and without MoNi were spherical and had an average diameter of 5 to 10 μm, although inclusion of MoNi produced a smoother surface morphology (Fig. 2D). We assessed the stability of the BSA after spray drying by SEC characterization and compared the signal of monomer and aggregate peaks between fresh BSA and the two spray-dried formulations. The inclusion of MoNi resulted in both a higher monomer peak fraction and negligible formation of high–molecular weight aggregates compared with the BSA spray-dried without MoNi (Fig. 2E). The improvement in BSA stability and suspension flow behaviors with inclusion of MoNi was not unique to spray-dried microparticles but was also observed when comparing protein microparticles formed by lyophilization with and without MoNi followed by ball milling (fig. S5, A to E).

Fig. 2. Rheological and stability characterization of spray-dried microparticles.

(A) Storage modulus and loss modulus and (B) tan δ from the angular frequency sweep of BSA microparticles without MoNi (520 mg/ml) resuspended in triacetin and BSA microparticles with MoNi (520 mg/ml, with 5 wt % MoNi) resuspended in triacetin. (C) Flow sweep of BSA microparticles (520 mg/ml) resuspended in triacetin and BSA microparticles (520 mg/ml) containing 5 wt % MoNi resuspended in triacetin. (D) SEM image of microparticles and image of the resultant suspension in a vial and on a glass slide in triacetin at 520 mg/ml for microparticles formulated without and with 5 wt % MoNi. Scale bars, 5 μm. (E) Representative SEC trace of fresh BSA control and dissolved antibodies from spray-dried BSA without MoNi and spray-dried BSA with 5 wt % MoNi. PBS with sodium azide was used as an eluent. Approximate elution times for BSA were 32 to 35 min for monomers, 30 to 32 min for dimers, and 14 to 17 min for high–molecular weight aggregates. Data in (A) to (C) and (E) represent n = 1.

Injectability and stability of UHC protein formulations in triacetin

For therapeutic applications in subcutaneous injection, biopharmaceutical drug products need to be injectable through clinically relevant needles (25 to 27 gauge) with clinically relevant injection forces (less than 30 N) (11, 34, 36, 49). To assess the injectability of the fluid-like suspensions, we quantified the injection force required to inject a suspension of BSA-MoNi microparticles in triacetin (BSA, 520 mg/ml) through a 27-gauge, ½-inch needle at varied flow rates using a force sensor attached to a syringe pump (fig. S6A). The required injection force increased linearly with the injection flow rate and decreased with the use of either smaller gauge needles or thin-walled needles, both of which have larger inner needle diameters (fig. S6, B to D). Using an ultrathin-walled (UTW) 27-gauge needle and a flow rate of 1 ml/min, the BSA suspensions (520 mg/ml) comprising MoNi exhibited a required injection force of only 14 N (fig. S6D), which was within the injection force range achievable in standard pen autoinjectors (25- to 30-N maximum force) (49). Last, long-term storage experiments to assess the physical stability of the suspensions demonstrated that they exhibit minimal particle settling in the triacetin nonsolvent (fig. S7A) and comparable injection forces (21.4 N versus 22.4 N) over the course of 35 days of storage at room temperature (25°C) in a BD (Becton Dickinson) 1-ml syringe (fig. S7B).

To further understand the stability of BSA protein in the triacetin-based suspensions, we tested monomer loss and aggregate formation in the suspensions under stressed aging conditions (elevated temperature). BSA microparticle suspensions were prepared with protein (520 mg/ml) in triacetin using spray-dried BSA particles either with or without MoNi. An aqueous control solution was prepared with BSA at 20 mg/ml in phosphate-buffered saline (PBS). The three formulations were heated at 60°C for 30 min, and the BSA stability was then assessed by SEC (fig. S8A). Standard aqueous solution formulations of BSA in PBS showed aggregation and loss of monomer peak fraction with 76% of the monomer protein remaining. Both BSA microparticle suspensions, with and without MoNi, remained more stable with monomer peak fractions exceeding 89% (compared with fresh BSA with 90% monomer fraction) (fig. S8B). To further evaluate the stability, the suspensions were stored at 4°, 25°, and 37°C, and BSA stability was assessed by SEC over time. Suspensions comprising MoNi had lower light-scattering intensity of the high–molecular weight aggregate peaks than suspensions of BSA alone across all temperatures, indicating that the MoNi excipient improves the protein’s stability even in the dry state within the microparticles (fig. S9, A to F).

Characterization of low-viscosity nonsolvent additives for UHC protein formulations

To further reduce the injection force of protein suspensions and enable UHC protein formulations, we evaluated nonsolvent mixtures comprising triacetin and lower-viscosity additives including dimethylacetamide (DMAc) and benzyl alcohol (BA) (50, 51). The notation “X triacetin:Y DMAc:Z BA” indicates the volume fraction of triacetin, DMAc, and BA in the nonsolvent mixtures evaluated. Triacetin alone exhibited the highest injection force, which decreased with the addition of DMAc and BA (Fig. 3, A and B). For BSA-MoNi suspensions prepared at an equivalent protein content, the injection force at a given flow rate was reduced from 17.5 ± 2.4 to 2.5 ± 0.1 N (sevenfold reduction in injection force) with the use of viscosity-reducing, nonsolvent additives (Fig. 3B). The addition of DMAc at more than 50% did not further decrease the injection force (Fig. 3C). We selected 70 triacetin:30 DMAc (v/v) as our preferred nonsolvent mixture because it afforded a greater than twofold reduction in the injection force while demonstrating minimal subcutaneous injection site irritation in mice (fig. S10, A and B).

Fig. 3. Low-viscosity nonsolvent additives allow for ultrahigh-concentration protein formulations.

(A) Injection force curves for injection of BSA (506 mg/ml), 5 wt % MoNi suspensions at 1 ml/min through 26-gauge, ½-inch needles with DMAc and BA nonsolvent additives and (B) corresponding plateau injection forces (n = 2). Bar graphs show means ± SD. (C) Plateau injection force of BSA (506 mg/ml), 5 wt % MoNi suspensions at 1 ml/min through a 26-gauge, ½-inch needle as a function of DMAc content in triacetin (n = 2). Graph shows means ± SD. (D) Comparative protein concentration as predicted from density measurements and measured by absorbance with NanoDrop (n = 4) from a known volume of suspension. Bar graph shows means ± SD. (E) Injection force as a function of concentration for BSA microparticles (with mol %–matched MoNi or Tween 80) in 70 triacetin:30 DMAc and predicted fit to a particle jamming model. (F) SEM of particle morphology for BSA MoNi and BSA Tween microparticles. Scale bars, 10 μm. (G) Comparative injection force curves and (H) plateau injection forces for BSA microparticle suspensions (506 mg/ml; with mol %–matched MoNi or Tween 80) in triacetin at 1 ml/min through 26-gauge, ½-inch needles (n = 3). Bar graphs show means ± SD. Statistical comparison between two groups was made using a two-tailed Student’s t test in GraphPad Prism with statistical significance as P < 0.05.

Using the 70 triacetin:30 DMAc nonsolvent mixture, the injection force was measured for a series of BSA-MoNi microparticle suspensions as a function of BSA concentration. To ensure accuracy of reported BSA concentrations, the density of the BSA-MoNi microparticles was measured using a pycnometer and found to be 1.31 g/cm³. The expected BSA concentration calculated using this density value was corroborated by evaluating the BSA concentration by absorbance (NanoDrop) after dissolving a known volume of suspension formulation in aqueous media (Fig. 3D). Injection force measurements with BSA-MoNi microparticle suspensions in 70 triacetin:30 DMAc demonstrated injection of BSA formulations (600 mg/ml) with a clinically relevant injection force of 17 N through a 26-gauge, ½-inch needle at 1 ml/min (Fig. 3E). Concentration versus injection force data were fit to a modified version of the Krieger & Dougherty particle jamming model (52, 53). Using a spherical shape factor, we found that BSA-MoNi microparticles achieved the theoretical maximum packing density for spheres (volume fraction > 0.74) before jamming in these suspensions (Fig. 3E and fig. S11, A to C), indicating that these microparticles are behaving as smooth, hard spheres likely because of the enrichment of the high-T_g MoNi excipient in the surface layer of the microparticles. By contrast, BSA-Tween microparticles spray-dried with an equal molar loading of Tween 80 exhibited similar size and surface morphology (Fig. 3F), but suspensions comprising these particles exhibited higher injection forces at equal BSA concentrations (P = 0.005), onset of particle jamming at much lower BSA concentrations, and lower maximum concentrations that were injectable under relevant conditions (Fig. 3, G and H).

UHC protein suspensions enable in vivo administration and subcutaneous absorption

To demonstrate the translational potential of these UHC protein microparticle suspensions in vivo, we sought to quantify the absorption of protein after subcutaneous administration. In these studies, SKH1e mice were injected subcutaneously over the rear flank with fluorescently tagged BSA [BSA (506 mg/ml) with MoNi in triacetin suspension or BSA (20 mg/ml) without MoNi in PBS solution]. Fluorescent images collected from an in vivo imaging system (IVIS) were then used to study BSA absorption from the site of injection (Fig. 4A). Fluorescent signal over time characterized for the suspension and solution formulations was normalized and fit to a single-phase exponential decay curve to provide the half-life of BSA absorption from the subcutaneous space (Fig. 4B). The half-life of BSA absorption was about 4 hours and was not significantly different between the two formulations evaluated (P = 0.97) (Fig. 4C), despite a greater than 20-fold difference in required injection volume to achieve similar doses between the UHC suspension and standard solution formulations (Fig. 4D). This result was repeated with a DMAc and triacetin mixed suspension [BSA (506 mg/ml) with MoNi in 70 triacetin:30 DMAc] that demonstrated equivalent BSA half-lives to bolus solution and UHC suspensions in triacetin (fig. S12, A and B).

Fig. 4. UHC BSA suspensions allow for in vivo administration and subcutaneous absorption.

(A) Representative IVIS images demonstrating subcutaneous absorption of fluorescently tagged BSA administered by a PBS bolus injection or a UHC protein suspension. hr, hours. (B) Representative fluorescent signal in the subcutaneous space is fit to a single-phase exponential decay mode to identify half-life of subcutaneous absorption. Dots represent data points. Lines represent single-phase exponential decay curves. Representative of n = 3 to 5. (C) Comparative half-life of subcutaneous absorption for BSA administered by PBS bolus injection or UHC protein suspensions formulated with triacetin (n = 3 to 5). Bar graph shows means ± SEM. (D) Comparative volume of administration for bolus injection and UHC protein suspension. Bar graph shows means ± SEM. Statistical comparison between two groups was made using a two-tailed Student’s t test in GraphPad Prism with statistical significance as P < 0.05.

UHC suspension technology can be effectively used to administer hIgG

After these successes with formulation of BSA into UHC suspensions using the MoNi excipient, we applied this technology to a clinically relevant protein, hIgG. hIgG is the active pharmaceutical ingredient in several dose-limited, commercial drug products (such as Cuvitru and Gammagard from Takeda and Xembify from Grifols) (2, 54) and is comparable to mAb therapeutics in chemical structure, solubility, and size. Mirroring our previous BSA formulation studies, the hIgG-to-MoNi ratio in the final spray-dried product was fixed at 20:1 (w/w), and the initial feedstock solid concentrations and spray-drying parameters were selected to produce particles with a desired size and morphology. Spray-dried hIgG particles comprising MoNi (5 wt %) were determined to be 5 to 20 μm in diameter and exhibited a smooth morphology with slightly lower sphericity than the comparable BSA-based particles (Fig. 5A). As previously observed with BSA, the inclusion of the MoNi excipient stabilized hIgG during the spray-drying process. SEC characterization of fresh hIgG and hIgG spray-dried with and without MoNi demonstrated that spray drying hIgG with MoNi resulted in a higher monomer peak fraction and mitigated the formation of high–molecular weight aggregates (Fig. 5B). A control hIgG formulation stabilized with 30 wt % trehalose had a similar stabilizing effect as 5 wt % MoNi, and the addition of 25 wt % trehalose to the 5 wt % MoNi formulation afforded no additional stability benefit to the spray-dried hIgG product compared to MoNi alone (fig. S13, A and C). In summary, we illustrated that inclusion of only 5 wt % MoNi was sufficient to stabilize hIgG during the spray-drying process in a manner similar to state-of-the-art particle formulations comprising a minimum of 30 wt % trehalose (55).

Fig. 5. UHC suspension technology can be effectively used to deliver hIgG.

(A) Representative SEM of spray-dried IgG with MoNi particle morphology demonstrating individual particle morphology (top) and distribution of morphologies across the population (bottom). Scale bars, 10 μm. (B) SEC trace of dissolved antibodies from fresh hIgG control, spray-dried hIgG without MoNi, spray-dried hIgG with 5 wt % MoNi, and spray-dried hIgG with 25 wt % trehalose and 5 wt % MoNi. Represents n = 1. (C) Injection force as a function of concentration for hIgG microparticles with MoNi in triacetin (n = 3 to 5). Graph shows means ± SD. Injection force as a function of particle concentration is fit to a particle jamming model. (D) SEC trace of dissolved antibodies from fresh hIgG control and hIgG formulated in histidine buffer (100 mg/ml), glycine buffer (100 mg/ml), and triacetin (500 mg/ml) after 4 days of stressed aging at 50°C. Represents n = 1. (E) Comparative volume of administration for bolus injection and UHC protein suspension. (F) Representative IVIS images demonstrating subcutaneous absorption of fluorescently tagged hIgG administered by a PBS bolus injection or a UHC protein suspension. Representative of n = 5. (G) Fluorescent signal in the subcutaneous space is fit to a single-phase exponential decay mode to identify the half-life of subcutaneous absorption (n = 5). Graph shows means ± SEM. (H) Comparative half-life of subcutaneous absorption for hIgG administered by PBS bolus injection or UHC protein suspensions formulated with triacetin (n = 5). Graph shows means ± SEM. Analysis by an unpaired two-tailed Student’s t test. (I) Comparative volume of administration for bolus injection and UHC protein suspension. (J) hIgG serum concentration in hFcRn transgenic mice after hIgG administered by a PBS bolus injection or a UHC protein suspension (n = 5 or 6). Data in (B) and (D) represent n = 1. Graphs (G), (H), (J), (K), (L), and (M) show means ± SEM. Longitudinal comparison between two groups was conducted with a two-way ANOVA in GraphPad Prism with statistical significance as P < 0.05. Corresponding (K) C_max, (L) serum half-life, and (M) bioavailability as determined by normalizing the area under the curve for serum concentrations throughout the duration of the experiment. Bar graphs in (K) to (L) show means ± SEM (n = 5 or 6). Analysis by a two-tailed Student’s t test.

With these high–hIgG content microparticles (about 5 wt % protein) prepared with MoNi, we formulated triacetin-based suspensions and measured injection forces as a function of hIgG concentration. In these studies, we used a particle density of 1.35 g/cm³, measured by pycnometry, to determine formulation composition. Injection force measurements of hIgG formulations (450 mg/ml) of hIgG-MoNi microparticles in triacetin exhibited an injection force of 17 N through a 26-gauge, ½-inch needle (Fig. 5C). The concentration versus injection force data for these suspensions were again fit to a modified version of the Krieger and Dougherty particle jamming model (52) using a spherical shape factor (Fig. 5C). hIgG-MoNi suspensions exhibited jamming at a volume fraction of 0.67, indicating that microparticles were behaving as smooth, hard spheres likely because of the enrichment of the high-T_g MoNi excipient in the surface layer of the microparticles. Furthermore, similar to our observations with BSA-MoNi microparticle suspensions, the injection force of hIgG-MoNi microparticle suspensions could be further reduced by altering the nonsolvent composition (fig. S14).

We then assessed the stability of the hIgG protein in suspension through stressed aging assays. In these studies, hIgG-MoNi microparticle suspensions were prepared with protein (500 mg/ml) in triacetin. To directly compare with commercial IgG drug products (such as Gammagard Liquid) (2, 54), aqueous control solutions were prepared with hIgG at 100 mg/ml in histidine or glycine buffers. Formulations were exposed to stressed aging conditions of 50°C with shaking for 1 week, and IgG stability was assessed over time by SEC (Fig. 5D). hIgG-MoNi microparticle suspensions in triacetin showed minimal aggregation at all time points and less high–molecular weight aggregate formation than aqueous controls (fig. S15, A to I). In additional stressed aging studies of the suspension through multiple freeze-thaw cycles, we evaluated the stability of the hIgG-MoNi microparticle suspensions in triacetin and aqueous controls. The hIgG-MoNi suspension was unchanged after 10 freeze-thaw cycles, whereas aqueous controls exhibited an increase in high–molecular weight species (fig. S16, A and B). These studies indicated that formulation of the hIgG with MoNi in these microparticle suspensions improved antibody stability.

UHC suspensions demonstrate comparable pharmacokinetic profiles to subcutaneous bolus administration

We next sought to evaluate the subcutaneous absorption kinetics of hIgG from the microparticle suspension and from a clinically relevant formulation. SKH1e mice were injected subcutaneously through 27-gauge insulin needles with equal protein doses of fluorescently tagged hIgG, formulated either in a triacetin suspension formulation (300 mg/ml) comprising MoNi or in a PBS solution (100 mg/ml). This high-concentration IgG solution was chosen to be representative of standard aqueous IgG solution drug products (such as Gammagard Liquid). The hIgG-MoNi particles were delivered in 20 μl, whereas the PBS formulation required an injection volume of 60 μl (Fig. 5E). Fluorescent IVIS images were collected over time after the subcutaneous administration to study the kinetics of hIgG signal at the site of injection (Fig. 5F). Fluorescent signal over time for each formulation was normalized and fit to a single-phase exponential decay curve to provide a half-life of hIgG subcutaneous absorption (Fig. 5G). The half-life of hIgG subcutaneous absorption was about 9 hours (Fig. 5H), in agreement with previous literature values (56). Similar to our previous observations with BSA-based suspensions, the difference in half-life of absorption was not statistically significant (P = 0.165) between the UHC hIgG-MoNi suspension and standard solution formulations despite a threefold difference in the injection volume required for equivalent doses (Fig. 5E). To further investigate the in vivo pharmacokinetics of these UHC hIgG-MoNi suspensions and standard solution formulations, B6.Cg-Fcgrt^tm1DcrTg(CAG-FCGRT)276Dcr/DcrJ (hFcRn) transgenic mice were injected subcutaneously with equal doses of hIgG in either a triacetin-based suspension (350 mg/ml) comprising MoNi or a PBS solution (100 mg/ml) (Fig. 5I). Serum was collected for 9 weeks, and hIgG serum concentrations were quantified by enzyme-linked immunosorbent assay (ELISA) (Fig. 5J). Values for C_max (Fig. 5K), serum half-life (Fig. 5L), and bioavailability (Fig. 5M) were not significantly different (P > 0.05) between the two formulations despite the greater than threefold difference in injection volume (Fig. 5I).

UHC suspension technology can be effectively used to deliver a commercial mAb drug product

After these successes formulating hIgG into UHC suspensions using the MoNi excipient, we applied this technology to an infectious disease mAb hindered by critical dose limitations. mAbs are useful for preexposure or postexposure prophylaxis for infectious diseases such as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection. Regeneron has commercialized a highly potent anti-spike mAb cocktail for neutralization of SARS-CoV-2 variants (57). This mAb cocktail requires ~1200 mg of mAb to be delivered intravenously in a hospital setting at a relatively low solution formulation concentration of 20 mg/ml in histidine buffer (Fig. 6A). Here, we demonstrate reformulation of a human anti-COVID mAb (20 mg/ml; IDBiologics) as a UHC suspension at 400 mg/ml amenable to subcutaneous administration (Fig. 6A), enabling relevant doses to be delivered by an autoinjector in two 1.5-ml subcutaneous injections.

Fig. 6. UHC suspension technology can be effectively used to deliver a commercial mAb drug product.

(A) Schematic illustrating that a commercial intravenous antibody drug product (ADP) requires time-intensive delivery in a hospital setting, whereas ADP reformulated as a UHC protein suspension enables delivery with an autoinjector at home. Schematics created with BioRender and Adobe Illustrator. (B) Representative SEM of spray-dried mAb with MoNi particle morphology demonstrating individual particle morphology (top) and distribution of morphologies across the population (bottom). Scale bars, 10 μm. (C) Representative SEC trace of dissolved antibodies from fresh mAb control and spray-dried mAb with 5 wt % MoNi. Representative of n = 3. (D) Injection force of mAb microparticles (400 mg/ml) with MoNi in triacetin (n = 3). Bar graph shows means ± SD. (E) Comparative volume of administration and mAb dose for bolus injections and UHC protein suspension. (F) mAb serum concentration in humanized BL6 mice after mAb administered by a bolus injection (of 200 μl for a dose of 4 mg) or a UHC protein suspension (of 20 μl for a dose of 8 mg) (n = 5). Graph shows means ± SEM. Longitudinal comparison between two groups was conducted with a two-way ANOVA in GraphPad Prism with statistical significance as P < 0.05. Corresponding (G) C_max, (H) total drug exposure (AUC), and (I) day 1 IC₅₀ fold dilution from neutralization. Bar graphs in (G) to (I) show means ± SEM (n = 5). Analysis by an unpaired two-tailed Student’s t test.

Mirroring our previous hIgG formulation studies, the mAb-to-MoNi ratio in the final spray-dried product was fixed at 20:1 (w/w), and the initial feedstock solid concentrations and spray-drying parameters matched those used for hIgG. Spray-dried mAb particles comprising MoNi (5 wt %) were determined to have an average volume weighted mean diameter of 14 μm and exhibit a smooth spherical morphology (Fig. 6B and fig. S17, A to D). The inclusion of the MoNi excipient stabilized the mAb during the spray-drying process as evidenced by the preservation of the monomer peak and lack of high–molecular weight aggregates in SEC characterization (Fig. 6C). With these high mAb content microparticles (~95 wt % protein) prepared with MoNi, we formulated triacetin-based suspensions on the basis of a particle density of 1.35 g/cm³ (determined by pycnometry). Injection force measurements of mAb-MoNi microparticle suspensions in triacetin at mAb (400 mg/ml) exhibited an injection force of 6.0 ± 0.1 N through a 26-gauge, ½-inch needle (Fig. 6D), which is below the injection force threshold of 25 N for commercial autoinjectors. Furthermore, these suspensions were stable in triacetin and demonstrated comparable profiles to that of hIgG under stressed aging at 50°C with shaking for 1 week (fig. S18, A to C).

To investigate the in vivo pharmacokinetics of the UHC mAb-MoNi suspension, hFcRn transgenic mice were injected subcutaneously with either 20 μl of a triacetin-based mAb-MoNi suspension comprising protein (400 mg/ml) or 200 μl of a mAb solution (20 mg/ ml), which is the maximum achievable subcutaneous administration volume in a 20-g mouse (Fig. 6E) (58, 59). As such, mice that received UHC mAb suspensions received two times the mAb dose in 1/10 of the injection volume. Serum was collected for 4 weeks, and anti–SARS-CoV-2 mAb serum concentrations were quantified by ELISA with SARS-CoV-2 receptor binding domain (RBD)–coated plates (Fig. 6F). Mice that received UHC mAb-MoNi suspensions had a greater Cmax value (P < 0.001) (Fig. 6G) and greater total drug exposure as determined by the area under the curve (AUC) of serum concentration for the duration of the study (P < 0.001) (Fig. 6H) than mice receiving bolus injections, and these values were commensurate with dosing. In addition, comparative neutralization of serum samples at 24 hours after adjusting for dosing demonstrated that the mAb delivered in UHC mAb suspensions retained neutralization ability (Fig. 6I and fig. S19, A to D). In summary, we demonstrate that human mAb drug products can be successfully reformulated into UHC protein microparticle suspensions, enabling subcutaneous delivery of therapeutic doses that were previously inaccessible.

DISCUSSION

In this work, we leveraged a glassy surfactant called MoNi, which is a polyacrylamide derivative with a T_g exceeding that of high- T_g sugars such as trehalose (140°C compared with 106°C), as an efficient stabilizer for formulating protein therapeutics into readily injectable UHC suspensions. We applied this approach to improve the administration of antibodies, the dominant class of biopharmaceuticals, which require high-concentration formulations to achieve therapeutically relevant doses in a subcutaneous injection (5, 8-10, 54, 60). To date, clinically approved products have relied on unstable and burdensome multi-injection regimens of liquid formulations or the addition of an enzyme (hyaluronidase) as a secondary active biologic to degrade the subcutaneous tissue to enable administration of high volumes (11). Both liquid-based formulation approaches are prone to aggregation of the active biopharmaceutical, limiting shelf-life and global accessibility of these crucial drug products. Recently, formulations of spray-dried protein microparticles suspended in a nonsolvent have been developed to enhance protein stability, but current formulation approaches fail to meaningfully increase injectable doses over liquid formulations (typically <300 mg/ml) because of poor microparticle mechanical properties (for example, low-T_g surface properties) and excessively high (>30 wt %) additive concentrations (34-36). We hypothesized that a glassy surfactant such as MoNi, with a high T_g (140°C) and the ability to stabilize protein drugs against harsh stressed aging conditions, could markedly improve injectability of these protein suspensions by yielding mechanically robust microparticles at minimal additive concentrations and exceptionally high protein loadings (95 wt %).

The addition of surface-active excipients, including surfactants like Tween 20 or Tween 80, to aqueous formulations of proteins before spray drying helps to maintain drug bioactivity and mitigate protein aggregation through the spray-drying process. These surfactants are better stabilizers in this context than sugars because they are amphiphilic and preferentially partition to the interfaces formed as the aqueous feedstocks are aerosolized into droplets during spray drying, preventing protein denaturation and aggregation at these air-water interfaces (37, 39, 40, 61). These surfactants can also improve spray-dried particle morphology. Unfortunately, the amount of these PEG-based surfactants that can be used in these formulations is limited (typically <0.3 wt %) on account of the low T_g of the excipients (~65°C), which negatively affects the surface mechanical properties of the resulting microparticles (62). Given that frictional interactions between particles can increase the viscosity of suspensions and lead to paste-like formulations at high solid content, the poor mechanical properties induced by the surface enrichment of standard PEG-based surfactants yield suspension formulations that are uninjectable at high protein concentrations. For these reasons, these formulations require inclusion of high-T_g (>100°C) additives, which are typically sugars such as trehalose, at quantities exceeding 30 wt % to yield mechanically robust protein microparticles and suspensions that are both stable and injectable under relevant conditions (34-36). Yet, the high loading of these excipient solids within the microparticles severely limits the actual protein content achievable in the resulting suspension formulations.

To address these shortcomings of previous technologies, we replaced these surfactant and sugar excipients with a single glassy surfactant excipient, MoNi, during the spray-drying process. We found that incorporation of MoNi stabilized proteins during the spray-drying process, and the resulting protein microparticles comprising MoNi could be formulated as readily injectable suspensions at exceptionally high protein contents (exceeding 500 mg/ml for BSA and exceeding 400 mg/ml for human antibodies). We demonstrated that these UHC suspensions could be injected through clinically relevant needles (26 and 27 gauge) with low injection forces relevant to standard pen autoinjector devices (<25 N). In addition, experiments in mice showed that UHC suspension formulations of a model protein BSA and two antibody-based drug products (hIgG and an anti–COVID mAb) markedly reduced required injection volumes without altering the pharmacokinetics, protein structure, or efficacy. In summary, we demonstrated that the high- T_g excipient MoNi enables stable and readily injectable protein formulations at the highest reported protein concentrations to date (30, 34, 35).

These MoNi-containing protein microparticles can be readily translated into a clinically relevant product because of the scalable nature of the spray-drying process and promising safety profile of MoNi in preclinical animal models. The MoNi-containing protein microparticles reported in this work were generated using a standard commercial laboratory-scale spray dryer without fully optimized parameters. Given that spray drying is commonly used in the pharmaceutical industry and there are numerous licensed biopharmaceutical drug products that already use particle-based suspension formulations (such as Byetta) (63), we believe that further optimization of the spray-drying process parameters and use of industrial-scale spray dryers are likely to enable the development of suspensions with even higher protein concentrations. To date, the MoNi excipient has been evaluated in several preclinical animal models (20-22, 24) with no observable adverse events. This biocompatibility is likely due to stability (23) and rapid renal clearance (neutrally charged polymers < 25 kDa readily pass through the glomerular filtration barrier). Compared with current commercial excipients for stabilizing biopharmaceuticals (polysorbates and poloxamers), MoNi is orders of magnitude less cytotoxic (22). Therefore, we believe that clinical translation of this protein-MoNi microparticle suspension technology is highly feasible (64).

In addition, new formulation technologies enabling high-concentration, low-volume antibody delivery with less reliance on the cold chain will be essential to facilitating global access to critical biopharmaceutical drugs. Although mAbs are one of the most effective therapeutic modalities for treating many life-threatening conditions, most of the approved mAb drug products require high doses (>100 mg per dose) for therapeutic efficacy (60). Between 1998 and 2021, the FDA approved 27 solution-based high-concentration antibody products (>100 mg/ml) that can be administered by subcutaneous injection (54). Unfortunately, stability and viscosity limitations mean that these drug products are typically limited to mAb concentrations of only 200 mg/ml or lower, and many of these approved mAb drug products still require multiple injections to achieve therapeutically relevant doses (54). Because large subcutaneous injection volumes are often associated with pain, the widely accepted maximum volume for subcutaneous injection is 1.5 ml (65, 66). Many pen autoinjector devices have been engineered to deliver this volume per administration. As formulated, only 8 of the 27 subcutaneous injectable high-concentration mAb drug products (30% of these drugs) can be administered with an injection volume of 1.5 ml or less and achieve therapeutic doses (fig. S20A) (54). In this work, we demonstrate successful reformulation of commercial low-concentration human antibody drugs into UHC protein suspensions at >400 mg/ml. If our protein microparticle suspension technology was applied to reformulation of current mAb drug products, 22 of 27 (>80%) could be delivered as a single 1.5-ml injection (fig. S20, B and C).

This study is not without limitations. Although suspensions were stored in sterile plastic syringes for the purposes of this work, clinical products intended for subcutaneous administration are typically delivered with prefilled syringes or autoinjectors (67, 68). Further studies are warranted to evaluate the injection force and stability of UHC suspensions in these final delivery formats (69). Second, in vivo experiments were conducted in mouse models. Although mice are commonly used in preclinical studies, further evaluation in larger animals, such as pigs or nonhuman primates, is recommended because their subcutaneous tissue architecture more closely resembles that of humans (70). In addition, given that many high-concentration mAb therapies require repeated dosing, future work should assess local histological and immunological responses after repeated administration of UHC suspensions. Last, additional research is needed to define the range of biologic modalities, such as mAbs, antibody-drug conjugates (ADCs), Fc-fusion proteins, and enzyme replacement therapies, which can be effectively formulated and delivered using this platform.

Overall, the protein formulation platform we describe here both improves the shelf stability and markedly increases achievable formulation concentrations of clinically relevant biopharmaceuticals. These advancements have the potential to substantially improve accessibility and greatly reduce patient burden associated with these critical therapeutics.

MATERIALS AND METHODS

Study design

The objective of this study was to demonstrate that a high- T_g surfactant (MoNi) can stabilize protein therapeutics during the harsh conditions of spray drying, enable formulations at UHC that can be injected with clinically relevant forces through clinically relevant needles, and maintain similar pharmacokinetics and pharmacodynamics after subcutaneous administration. In vitro sample sizes were determined on an experiment-by-experiment basis depending on the inherent error of instruments (technical replicates) or material limitations. To ensure reproducibility of clinical-grade formulations, all antibody (IgG and mAb) injection data were collected in n ≥ 3. The initial sample sizes for in vivo experiments were determined with Mead’s resource equation. The number of experimental repeats/ replicates is listed in the figure legends. Animals were randomized to groups by cage, investigators performing injections were not blinded to the group, but investigators conducting analyses (such as ELISAs) were blinded to the group. Exclusion criteria included experimental errors such as incorrect dosing at the start of the experiment. Animal studies were performed with the approval of the Stanford Administrative Panel on Laboratory Animal Care (APLAC-32109).

Materials

All reagent-grade materials and solvents were purchased from Sigma-Aldrich or Thermo Fisher Scientific and used as received. Alexa-647-NHS was purchased from Lumiprobe. Slide-A-Lyzer dialysis cassettes [2-kDa molecular weight cutoff (MWCO)] from Thermo Fisher Scientific were used for polymer purification. BSA (A2153-50G, CAS no. 9048-46-8) was purchased as a lyophilized powder from Sigma-Aldrich. hIgG (catalog no. 340-21, lot: 07 J4627) was purchased as a lyophilized powder from Medix Biochemica. HyPure Cell Culture Grade Water was purchased from Cytiva. PBS (10010-023) was purchased from Gibco. Syringes used for injection force measurements were Fisherbrand 1-ml plastic Luer lock syringes (catalog no. 14955464). Syringes used for formulating protein suspensions with nonsolvent were Thermo Fisher Scientific 5-ml Luer slip plastic syringes (catalog no. S7510-5). Needles used for injection force measurements were BD PrecisionGlide needles (26 gauge, ½ inch, ref: 305111). In vivo protein suspension delivery was performed using BD Insulin Syringes with BD Micro-Fine IV Needles (27 gauge, 12.7 mm).

MoNi synthesis and characterization

Polymers were synthesized and characterized according to previously published methods (20). Briefly, to a 20-ml scintillation vial, 2-cyano-2-propyl dodecyl trithiocarbonate (CDPT; 259 mg, 0.75 mmol), 2,2′-azobis(2-methyl-propionitrile) (AIBN; 24.6 mg, 0.15 mmol), 4-acryloylmorpholine (Morph, Mo; 3.85 g, 27.3 mmol), N- isopropylacrylamide (Nipam, Ni; 1.15 g, 10.2 mmol), and N,N- dimethylformamide (DMF; 9.0 mL) were added. The vial was capped with a PTFE septum and sparged with nitrogen gas for 20 min. The polymerization was conducted at 65°C for 18 hours to a conversion of 99% as determined by ¹H NMR (Bruker Neo, 500 MHz). Polymers were purified by precipitating three times in a 75:25 ether:hexane mixture. Then, the end group was removed by addition of the polymer to a 20-fold excess AIBN (4.1 g, 25 mmol), lauroyl peroxide (LPO; 996 mg, 2.5 mmol), and DMF (45 ml) solution in a 200-ml round-bottom flask, and the reaction was heated to 90°C for 24 hours. Afterward, excess AIBN was removed by precipitation, and the polymer was dialyzed against deionized water for 48 hours to prepare the final product with a composition of 77 wt % Morph and 23 wt % Nipam by ¹H NMR (Morph:Nipam ratio of 2.77), a molecular mass by SEC of M_n 6024 kDa, M_w of 6992 kDa, and dispersity of 1.16. Because MoNi has an extremely high solubility in aqueous solutions, before formulation, the polymer was dissolved at 100 mg/ml in nanopure H₂O.

To characterize the final MoNi polymer, we determined the molecular weight (M_{n SEC}) and dispersity by SEC using a refractive index (RI) detector and polymethylmethacrylate standards. The running solvent was DMF with LiBr (1 g/liter; flow rate: 1 ml/min) heated to 50°C, and samples were prepared at 5 mg/ml. Separation was done through two Jordi Labs Resolve Mixed Bed Low Divinylbenzene (DVB) columns in series, and data were collected by a Dionex Ultimate 3000 Variable Wavelength detector and RefractoMax521 RI detector. The RI traces were normalized, and AUCs for the 310-nm absorbance signals were calculated with Prism 10.

Spray drying BSA microparticles

BSA feed solutions were prepared by dissolving lyophilized BSA in cell-grade water at 2 wt % (20 mg/ml). BSA solutions were dissolved at room temperature for 1 hour before sterile filtering using a 0.2-μm sterile filter. After sterile filtering, 7-kDa MoNi was added to the feed solution at a concentration of 0.1 wt % (1 mg/ml). Feed solutions were stored on ice before spray drying. Samples were spray-dried using a Buchi B-290 Mini Spray Dryer equipped with a high-performance cyclone. Samples were spray-dried using an inlet temperature of 150°C (outlet 67°C), an aspirator pressure of 40 mm, and a pump rate of 20% (6 ml/min). Collected powder was transferred to a 50-ml Falcon tube and stored at 4°C with desiccant.

Spray drying BSA microparticles with Tween 80

BSA feed solutions were prepared by dissolving lyophilized BSA in cell-grade water at 2 wt % (20 mg/ml). BSA solutions were dissolved at room temperature for 1 hour before sterile filtering using a 0.2-μm sterile filter. After sterile filtering, Tween 80 was added to the feed solution at a concentration of 0.0176 wt % (0.176 mg/ml). The Tween 80 concentration was chosen to add equal moles of MoNi and Tween 80 to the spray-drying feed solution. Feed solutions were stored on ice before spray drying. Samples were spray-dried using an inlet temperature of 150°C (outlet 67°C), an aspirator pressure of 40 mm, and a pump rate of 20% (6 ml/min). Collected powder was transferred to a 50-ml Falcon tube and stored at 4°C with desiccant.

Spray drying hIgG microparticles

hIgG feed solutions were prepared by dissolving lyophilized hIgG in cell-grade water at 10 wt % (100 mg/ml). hIgG solutions were dissolved at 4°C for 4 hours before sterile filtering using a 0.2-μm sterile filter. The sterile-filtered product was then dialyzed against cell-grade water overnight using a 30K MWCO Slide-A-Lyzer dialysis cassette to remove residual salts in the lyophilized product. After dialysis, the hIgG concentration of the solution was quantified by NanoDrop, and the volume of the solution was measured using a graduated cylinder. The hIgG solution was then diluted with hIgG (50 mg/ml) using cell-grade water. MoNi (7 kDa) was added to the feed solution at a concentration of 0.25 wt % (2.5 mg/ ml). Feed solutions were stored on ice before spray drying. Samples were spray-dried using an inlet temperature of 80°C (outlet 53°C), an aspirator pressure of 40 mm, and a pump rate of 1 ml/ min. Collected powder was transferred to a 50-ml Falcon tube and stored at 4°C with desiccant.

Spray drying COVID mAb microparticles

Human anti–SARS-CoV-2 mAb was purchased from IDBiologics. The mAb drug product was formulated at 20 mg/ml in a nondisclosed buffer. The mAb drug product was first buffer exchanged into cell-grade water by centrifugal filtration using an Amicon Ultra-15 30,000 MWCO device spun at 4000g. Buffer exchange was performed by five repeated cycles of 10-fold concentration and dilution with cell-grade water. After buffer exchange, the concentrated mAb was diluted to 100 mg/ml and sterile filtered using a 0.2-μm sterile filter. After sterile filtering, the concentration of mAb in the solution was quantified by NanoDrop, and the volume of the solution was measured using a graduated cylinder. The mAb solution was then diluted with mAb (50 mg/ml) using cell-grade water. MoNi (7 kDa) was added to the feed solution at a concentration of 0.25 wt % (2.5 mg/ ml). Feed solutions were stored on ice before spray drying. Samples were spray-dried using an inlet temperature of 80°C (outlet 53°C), an aspirator pressure of 40 mm, and a pump rate of 1 ml/min. Collected powder was transferred to a 50-ml Falcon tube and stored at 4°C with desiccant.

Formulating suspensions

Suspensions were formulated by combining a known mass of spray-dried microparticles with a known volume of nonsolvent (triacetin, DMAc, or BA). Protein concentration in milligrams per milliliter was determined by assuming that the total volume encompassed nonsolvent volume as well as spray-dried particle volume. When calculating suspension concentration, the spray-dried particle density was taken from pycnometry experiments.

To minimize nonsolvent evaporation when preparing suspensions, spray-dried microparticles were added to the barrel of a 6-ml Luer slip syringe. The mass of spray-dried particles was measured using an analytical balance. The desired volume of nonsolvent or nonsolvent combination was added to the syringe barrel through the syringe tip using a p200 pipette. After the syringe was capped, the protein suspension was mixed inside the syringe using a vortex for 5 min or until all powder was fully dispersed. Protein suspensions were transferred from 6-ml Luer slip syringes to the alternative desired syringe (1-ml Luer lock syringes or insulin syringes) by backloading for force of injection experiments or animal experiments, respectively.

Injection force measurements

Force of injection was quantified by measuring the force required to inject a protein suspension through a known needle gauge at a known flow rate using a syringe of known barrel dimensions. A force sensor was built that encompassed a load cell [FUTEK LLB300 50 lb Subminiture Load Button (model no. LLB300, item no. FSH03954, serial no. 705242)] attached to a syringe pump [KD Scientific Syringe Pump (model no. LEGATO 100, catalog no. 788100, serial no. D103954)]. An Omega Engineering Platinum Series Meter (model no. DP8PT, serial no. 18110196) was used to translate load cell resistance measurements to force values in kilograms (kg). The load cell was calibrated before measuring injection force (fig. S21, A and B).

Injection force experiments were performed as follows. A 1-ml Thermo Fisher Scientific Luer lock syringe with the desired needle gauge was loaded into the syringe pump. The syringe pump height was adjusted so that the load button of the force sensor was in contact with the end of the syringe plunger. The initial force was at or very close to 0 kg. The appropriate syringe barrel dimensions as well as the desired flow rate and injection volume were then selected. The syringe pump moved at the programmed rate injecting the protein suspension through the attached needle. The force sensor coupled with the Omega unit measured the force required to inject the protein suspension at the desired flow rate. A LabVIEW program recorded the forces measured throughout the duration of an injection experiment and displayed a graph of injection force over time. Force of injection was quantified by subtracting the average initial force (background) from the average plateau injection force. Injection force in kilograms was converted to injection force in newtons by multiplying by 9.81.

SEC characterization of protein stability

Protein stability before and after spray drying was characterized by SEC. Protein suspension stability was characterized by SEC as well as by comparative injection force. Protein stability was quantified by SEC absorbance measurement and SEC–multiangle light scattering (MALS) with the ASTRA software package (Wyatt Technology Corporation). Samples were first dissolved in PBS with protein (5 mg/ml) and passed through an SEC column (Superose 6 Increase 10/300 GL) in a mobile phase of PBS with sodium azide at 25°C and a flow rate of 0.5 ml/min. Dissolved proteins were separated by size, and the AUC of SEC absorbance measurements was analyzed at 250 nm to determine the mass fraction of each species. SEC-MALS detection consisted of an Optilab T-rEX (Wyatt Technology Corporation) RI detector operating at 658 nm and a TREOS II light scattering detector (Wyatt Technology Corporation) operating at 659 nm. A dn/dc value of 0.185 was used for BSA and IgG samples. To directly compare the stability of spray-dried protein samples, SEC traces were normalized to the height of the monomer peak.

Methodology for BSA stressed aging

BSA protein suspensions were prepared at 520 mg/ml in triacetin using spray-dried BSA particles with and without 5 wt % MoNi. A BSA aqueous control was prepared by dissolving fresh, lyophilized BSA at 20 mg/ml in PBS. All samples were stored in parafilmed 8-ml scintillation vials. A 500-ml beaker of water was heated to 60°C using a temperature-controlled hot plate. The sample files were submerged in the 60°C water bath so that the protein sample volume was fully beneath the water line. Samples were heated at 60°C for 30 min to encourage protein degradation. After stressed aging, samples were redissolved in PBS at 5 mg/ml, and protein stability was assessed by SEC.

Methodology for hIgG and human mAb stressed aging

hIgG microparticle suspensions were prepared with protein (500 mg/ ml) in triacetin. Aqueous control solutions were prepared with hIgG at 100 mg/ml in 0.25 M histidine or 0.25 M glycine buffers. Formulations were exposed to stressed aging conditions (50°C incubator and shaking at 160 rpm) for 1 week, and IgG stability at multiple time points was assessed by SEC. Human mAb microparticle suspensions were prepared with protein (400 mg/ml) in triacetin, and stability through stressed aging was assessed in the same manner. After stressed aging, samples were redissolved in PBS at 5 mg/ml, and protein stability was assessed by SEC.

Methodology for hIgG stability through freeze-thaw cycles

hIgG microparticle suspensions were prepared with protein (500 mg/ ml) in triacetin. Aqueous control solutions were prepared with hIgG at 100 mg/ml in 0.25 M histidine or 0.25 M glycine buffers. Formulations were freeze-thawed up to 10 times, where each freeze-thaw cycle consisted of allowing the sample to reach −80°C and subsequently allowing the sample to warm to room temperature (20°C). The stability of IgG across multiple freeze-thaw cycle time points was evaluated by SEC after dissolving the samples in PBS at a concentration of 5 mg/ml.

UHC protein suspension storage and injection force measurements

BSA protein suspensions with 5 wt % MoNi were prepared at 520 mg/ml in triacetin. Protein suspensions were loaded into a 1-ml Thermo Fisher Scientific Luer slip syringe and capped with a BD 26-gauge, ½-inch needle wrapped with parafilm to limit solvent evaporation. Injection force for the protein suspension (520 mg/ml) was measured on day 0 and again on day 35. The syringe was stored horizontally at 23°C.

In vivo delivery and imaging of UHC protein suspensions

Animal studies were performed with the approval of the Stanford Administrative Panel on Laboratory Animal Care (APLAC-32109) and were in accordance with the National Institutes of Health guidelines. Fluorescently tagged BSA microparticles were obtained by spray drying AF647-BSA with untagged BSA at a ratio of 1:500. MoNi was kept at 5 wt % in the final particle formulation. Fluorescently tagged BSA microparticles were then further diluted with untagged BSA microparticles with MoNi at a ratio of 1:5, resulting in a final ratio of 1:2500 AF647-BSA:untagged BSA. Protein suspensions in triacetin and 70 triacetin:30 DMAc were formulated as described above. A bolus control consisted of BSA at a ratio of 1:2500 AF647-BSA:untagged BSA dissolved in PBS at 20 mg/ml.

Female 10-week-old SKH1-Elite mice were each given subcutaneous injections of either 8.8 μl of fluorescently tagged BSA protein suspension (506 mg/ml; in triacetin or 70 triacetin:30 DMAc) or 200 μl of fluorescently tagged BSA (20 mg/ml) in PBS. Protein suspension injections were administered with a 50-μl Hamilton syringe with a 26-gauge, ½-inch needle. Bolus injections were administered with a 1-ml Luer lock syringe with a 26-gauge, ½-inch needle. The subcutaneous injection sites of animals were imaged using the IVIS (Lago) over a series of time points spanning 2 days. When imaging, mice were anesthetized with isoflurane gas and imaged with an exposure time of 2 s, excitation wavelength of 600 nm, and emission wavelength of 670 nm (binning, medium; F/stop, 1.2). Total radiant efficiency [(photons/s)/(μW/cm²)] was quantified using an equal-sized region of interest surrounding the injection site. Fluorescence intensity at each time point was normalized to the maximum fluorescent intensity, and normalized fluorescence intensity values for each mouse (n = 3 to 5) were fit to a single exponential decay model; half-lives were acquired and averaged using GraphPad Prism.

IVIS imaging of hIgG delivery

hIgG was fluorescently labeled with Alexa-647-NHS from Lumiprobe at a 5 wt % ratio. In brief, 50 mg of dry hIgG was dissolved in 10 ml of PBS, and 500 μl of a DMSO stock solution (5 mg/ml) of Alexa-647-NHS was added to the solution. The reaction proceeded for 24 hours at room temperature, and the free dye was removed with 10-kDa MWCO Amicon spin filter. Fluorescently tagged hIgG microparticles were obtained by spray drying AF647-tagged hIgG with untagged hIgG at a ratio of 1:20. MoNi was kept at 5 wt % in the final particle formulation. Fluorescently tagged hIgG microparticles were then further diluted with untagged hIgG microparticles with MoNi at a ratio of 1:100, resulting in a final ratio of 1:2000 AF647-tagged hIgG:untagged hIgG. Protein suspensions in triacetin were formulated as described above. A bolus control consisted of hIgG at a ratio of 1:2000 AF647-tagged hIgG:untagged hIgG dissolved in PBS at 100 mg/ml.

Female SKH1-Elite mice were each given subcutaneous injections of either 20 μl of fluorescently tagged hIgG protein suspension (300 mg/ml) in triacetin or 60 μl of fluorescently tagged hIgG (100 mg/ ml) in PBS (n = 5). Protein suspensions and bolus injections were both administered subcutaneously into the flank of the mouse with an insulin syringe with a 27-gauge needle. The subcutaneous injection sites of animals were imaged using the IVIS (Lago) over a series of time points spanning 2 days. IVIS imaging and half-life data analysis were performed as described above. Animals were imaged with an exposure time of 1 s.

In vivo pharmacokinetics of hIgG delivery

Female scid FcRn–/– hFcRn Tg mice were each given subcutaneous injections of either 30 μl of hIgG protein suspension (350 mg/ml) in triacetin or 105 μl of fresh hIgG (100 mg/ml) in PBS to result in a uniform protein dose of 10.5 mg of hIgG per mouse. When formulating protein suspensions in triacetin, it was assumed that spray-dried powder had a density of 1.35 g/cm³ and was 95% hIgG by mass. Protein suspension and bolus injections were both administered subcutaneously into the flank of the mouse. At selected time points, mice were anesthetized with isoflurane gas, and blood samples were drawn by tail vein. Time points included 24 and 48 hours and 4, 6, 11, 16, 21, 28, 35, 49, and 63 days. The concentration of hIgG in serum was quantified using an IgG (Total) Human ELISA Kit (Thermo Fisher Scientific, catalog no. BMS2091). Pharmacokinetic readouts included C_max, serum half-life, and bioavailability. C_max for each animal was the highest hIgG titer quantified by IgG Total Human ELISA. Serum half-life was quantified by fitting hIgG titers (48 hours to day 63) for each treatment group to a single-phase exponential decay model, and half-lives were acquired using GraphPad Prism. Bioavailability was quantified by calculating the AUC (0 to day 63) for each animal in the PBS bolus and triacetin suspension using GraphPad Prism. Bioavailability of hIgG in each animal was quantified by dividing the AUC of each animal by the average AUC of the PBS bolus group.

In vivo pharmacokinetics of human mAb delivery

Female 10-week-old B6.Cg-Fcgrt^tm1DcrTg(CAG-FCGRT)276Dcr/DcrJ (hFcRn) transgenic mice were each given subcutaneous injections of either 20 μl of COVID mAb (400 mg/ml; IDBiologics) protein suspension in triacetin or 200 μl of COVID mAb (20 mg/ml; IDBiologics) drug product. When formulating protein suspensions in triacetin, it was assumed that spray-dried powder had a density of 1.35 g/cm³ and was 95% mAb by mass. Protein suspension and bolus injections were both administered subcutaneously into the flank of the mouse. At selected time points, mice were anesthetized with isoflurane gas, and blood samples were drawn by tail vein. Time points included 24 and 48 hours and 4, 7, 10, 14, 21, and 28 days. The concentration of mAb in serum was quantified by ELISA with SARS-CoV-2 (2019-nCoV) Spike RBD-His (A435S) Recombinant Protein (Sino Biological)–coated plates (at 2 μg/ml). Pharmacokinetic readouts included C_max and bioavailability. C_max for each animal was the highest mAb titer quantified by ELISA. Total drug exposure was quantified by calculating the AUC (0 to 28 days) for each animal using GraphPad Prism.

Pseudovirion neutralization experiments

Neutralization experiments were performed as described in Xu et al. (71). Briefly, SARS-CoV-2 spike-pseudotyped lentiviruses encoding a luciferase-ZsGreen reporter were produced in human embryonic kidney (HEK) 293F cells by cotransfection of five plasmids. The five plasmids included a packaging vector (pHAGE-Luc2-IRES-ZsGreen), a plasmid encoding the SARS-CoV-2 spike (HDM-SARS2-spike-WT), and three helper plasmids (pHDM-Hgpm2, pHDM-Tat1b, and pRC-CMV_Rev1b). Fifty milliliters of cells was diluted to a density of ~3 × 10⁶ to 4 × 10⁶ cells/ml. Transfection mixture was prepared by adding five plasmids (50 μg of packaging vector, 17 μg of SARS-CoV-2–encoding plasmid, and 11 μg of each helper plasmid) to 5 ml of Expi-free medium, followed by the dropwise addition of BioT transfection reagent (150 μl, Bioland Scientific) with vigorous mixing. After a 10-min incubation at room temperature, the transfection mixture was transferred to HEK293F cells. d-Glucose (4 g/liter, Sigma-Aldrich) and valproic acid (3 mM, Acros Organics) were then added to the cells immediately posttransfection to increase recombinant protein production. The cells were harvested 3 to 5 days after transfection by spinning the cultures at 300g for 5 min. The supernatant was filtered through a 0.45-μm filter, and 0.5 ml of 1 mM Hepes was added to neutralize the pH. Viral stocks were aliquoted and flash-frozen in liquid nitrogen. They were stored at −80°C and titrated before further use.

Antisera were heat inactivated (56°C, 30 to 60 min) before neutralization assays. Neutralization against SARS-CoV-2 wild type (WT) was analyzed in HeLa-ACE2/TMPRSS2 cells. One day before infection (day 0), cells were seeded at 8000 cells per well in white-walled, white-bottom, 96-well plates (Thermo Fisher Scientific or Greiner Bio-One). On day 1, antisera were serially diluted in D10 media and mixed 1:1 with pseudoviruses for 1 to 2 hours at 37°C before being transferred to cells. The pseudovirus mixture contained SARS-CoV-2 WT, D10 medium, and polybrene (1:500). Assays were read out with luciferase substrates 2 days after infection by removing the medium from the wells and adding 80 μl of a 1:1 dilution of BriteLite in Dulbecco’s phosphate-buffered saline (DPBS, BriteLite Plus, PerkinElmer). Luminescence values were measured using a microplate reader (BioTek Synergy HT or Tecan M200). Percentage infection was normalized on each plate. Neutralization assays were performed in technical duplicates. Median inhibitory concentration (IC₅₀) dilution values were obtained for each serum sample by fitting an inhibitor versus response three-parameter curve in GraphPad Prism.

Statistical analysis

Injection force data and protein quantification by ultraviolet absorbance are reported as means with SD. All other data are reported as means with SE. Comparison between two groups was conducted with an unpaired, two-tailed Student’s t test performed in GraphPad Prism. Comparison between greater than two groups was conducted in GraphPad Prism with a one-way analysis of variance (ANOVA) with Tukey post hoc correction for multiple comparisons. Longitudinal comparison between two groups was conducted with a two-way ANOVA in GraphPad Prism. Results were considered significant if P < 0.05. Before performing parametric tests, data were tested for normality using the Shapiro-Wilk test. n for each experiment is provided in the figure captions. All individual-level data are available in data file S1.

Supplementary Material

The PDF file includes:

Methods

Figs. S1 to S21

Table S1

Legend for data file S1

Other Supplementary Material for this manuscript includes the following:

Data file S1

MDAR Reproducibility Checklist