Scalable desolvation synthesis, storage stability, and comprehensive biophysical characterization of human hemoglobin nanoparticles as a potential hemoglobin‐based oxygen carrier

Abstract

Hemoglobin (Hb)‐based oxygen carriers (HBOCs) are materials that exploit Hb's native oxygen transport while avoiding the adverse side‐effects associated with the exposure of the cell‐free, unmodified Hb tetramer (i.e. vasoconstriction, systemic hypertension, and oxidative tissue injury) in circulation. There are many synthetic routes to generate HBOCs including chemical polymerization, polymer surface conjugation, and protein encapsulation that meet the material and O₂ demands of different biomedical applications (e.g. emergency transfusion medicine, ex vivo organ perfusion, tumor oxygenation, etc.) especially when red blood cells may not be immediately available or desirable. There is currently no FDA approved HBOC in the US, therefore leaving room in the field for further HBOC design and optimization. An ideal HBOC should effectively bind and release O₂, have a scalable synthesis to meet material demand, and limit batch‐to‐batch variability to yield consistent, uniform material. This work demonstrates the successful synthesis of human Hb nanoparticles using a desolvation technique (hHb‐dNPs) as a new HBOC candidate. hHb‐dNPs demonstrated a monodisperse size distribution between 130 and 150 nm with spherical morphology, a negative surface charge, and hemocompatibility. hHb‐dNPs exhibited high O₂ affinity with decreased cooperativity similar to that of relaxed‐state polymerized Hb. The methodology demonstrated scalability from 5 to 500 mL starting volume, producing products with similar biophysical properties regardless of the scale. hHb‐dNPs demonstrated structural and biophysical stability for a month at 4°C and −80°C with various cryoprotectants, with the optimal combination being hHb‐dNPs co‐stored at −80°C with HSA. Considering these results, hHb‐dNPs are a promising potential next generation HBOC.

This work describes the development of human hemoglobin nanoparticles using an anti‐solvent precipitation synthesis method as a potential next generation hemoglobin‐based oxygen carrier. The hemoglobin nanoparticles are uniform in size and shape, hemocompatible, stable at various storage conditions, and demonstrate a scalable synthesis technique.

1. INTRODUCTION

Hemoglobin (Hb)‐based oxygen carriers (HBOCs) are a promising temporary red blood cell (RBC) substitute when blood is not immediately available. HBOCs utilize the unique O₂ binding properties of Hb while modifying the molecule in a manner that prevents the deleterious side‐effects of acellular Hb seen in circulation. These modifications can include Hb polymerization, surface conjugation of poly(ethylene glycol) [PEG], Hb encapsulation inside of particles, or formation of solid Hb nanoparticles. Polymerized Hb (PolyHb) is the most well‐established HBOC, having been commercially manufactured, but ultimately not receiving FDA approval due to adverse side‐effects related to the presence of cell‐free Hb and low molecular weight (MW) polymers (<500 kDa) in the preparations.^{1, 2, 3} Next‐generation PolyHbs have demonstrated effective and tunable oxygen delivery, scalability, and safety in vivo because of synthesis and process modifications to remove small MW species (<500 kDa) in the preparations.^{1, 2, 4, 5} One disadvantage of PolyHb is its molecular polydispersity and batch‐to‐batch variability. ⁶

Hb surface PEGylation is the process of conjugating PEG to the surface of Hb to increase its molecular weight and size, therefore increasing its circulatory half‐life and avoiding blood vessel extravasation.^{7, 8, 9} Unfortunately, PEG‐Hb exhibited increased auto‐oxidation rates, Hb tetramer (α₂β₂) dimerization, and heme loss. ¹⁰ Encapsulation of Hb inside vesicles, also known as Hb vesicles (HbV), creates tunable particles that preserve native Hb function.^{11, 12} The disadvantage to HbVs is the disparity between the mass of encapsulated Hb and the mass of lipid in the particle, which is much lower compared to a single RBC, thus requiring transfusion of a higher lipid load to transport the same amount of oxygen compared to RBCs. ¹³ To address this drawback, an alternative HBOC consists of Hb nanoparticles (Hb‐NP), which remove the lipid layer associated with HbVs to yield a particle composed solely of Hb.

The application of NPs in the pharmaceutical industry has been the focus of extensive research recently. NPs are appealing because of their low cost of synthesis, consistent physical properties, controlled release, and improved bioavailability ¹⁴ compared to the unmodified molecule/drug of interest. NPs can also be synthesized from a range of proteins for use in various applications, such as albumin NPs for pharmaceuticals^{15, 16, 17, 18, 19} and curcumin NPs in the food industry.^{20, 21} The versatility of protein‐based NPs can be tailored to specific application requirements. By applying the protein NP synthesis to Hb, a more uniform and consistent HBOC composed entirely of stabilized Hb can be produced with controllable size, morphology, and oxygen affinity. In addition, these NP fabrication methods may allow for the incorporation of other pharmacological agents to improve the outcomes of HBOCs beyond O₂ delivery, thus expanding the application of HBOCs into drug delivery.

Protein NPs can be synthesized via emulsification, ²² desolvation,^{15, 16, 17, 18, 19, 23} nanospray drying,^{24, 25, 26} electrospraying,^{27, 28, 29, 30} and co‐precipitation.^{31, 32, 33} This study focuses on protein NP synthesis via desolvation, or anti‐solvent precipitation, as it has been demonstrated to be a scalable and affordable synthesis route with a straightforward procedure that does not require specialized equipment. Desolvation, for protein nanoparticle synthesis (dNP), is a process in which protein NPs are formed through forced protein precipitation via an antisolvent. Increasing the concentration of antisolvent in the solution increases the supersaturation of the protein of interest and decreases the Gibb's free energy change (ΔG) of the solution. Once the protein has reached a critical supersaturation, the free energy barrier will be overcome and drive protein nucleation.^{34, 35} This point of initial nucleation is known as the critical point, and theoretically, the protein particles are at their smallest and most uniform size. Once protein particles begin nucleating, the concentration/supersaturation of the free protein of interest will decrease in solution. Protein nuclei will continue to form until this concentration falls below the critical saturation. ³⁶ Once no more protein nuclei form, the protein particles will continue to grow through condensation or coagulation. ²⁰

The size and uniformity of the protein NPs synthesized via desolvation are influenced by the selection of solvent and antisolvent, starting concentration of protein, addition rate of antisolvent, stirring rate, mixing profile, volume ratio of solvent to antisolvent, and temperature.^{15, 20} The combination of these parameters is unique to the specific protein of interest. This work determined the appropriate parameters needed for a proof‐of‐concept synthesis of NPs composed of human Hb (hHb), the first application of hHb in protein NP synthesis to the best of the author's knowledge. hHb‐dNPs were fully characterized and further studied for synthesis scalability and particle stability over time. This work aims to demonstrate a promising new potential HBOC.

2. MATERIALS AND METHODS

2.1. Materials

Dextran from Leuconostoc spp. (MW ~40 kDa), glycine(C₂H₅NO₂), 2,4‐dinitrophenylhydrazine (DNPH, C₆H₃(NO₂)₂NHNH₂), and sodium cyanoborohydride (NaCNBH₃) were purchased from Sigma Aldrich (St. Louis, MO). Sodium chloride (NaCl), potassium chloride (KCl), sodium phosphate monobasic (NaH₂PO₄), sodium phosphate dibasic (Na₂HPO₄), sulfuric acid (H₂SO₄), and sodium periodate (NaIO₄) were purchased from Fisher Scientific (Pittsburgh, PA). 8 kDa cellulose dialysis tubing was purchased from Fisher Scientific (Pittsburgh, PA). Hollow fiber tangential flow filtration (TFF) modules S02‐S05U‐05‐N (polysulfone [PS], 0.05 μm pore size), D02‐S05U‐05‐N (PS, 0.05 μm pore size), and D02‐S050‐05‐N (PS, 50 kDa pore size) were purchased from Repligen (Rancho Dominguez, CA). Amicon Ultra‐15 (50 kDa) and Amicon Ultra‐4 (50 kDa) centrifugal filters were purchased from Millipore Sigma (Burlington, MA). Ethanol was purchased through The Ohio State University Stores (product number 90721). Folded capillary zeta cells were purchased from Melvern Panalytical (Worcestershire, UK).

2.2. Oxidized dextran synthesis

Oxidized dextran (odex) was synthesized via a modified method to that previously described in Hickey and Palmer. ²³ In summary, 40 kDa dextran from Leuconcostoc spp. (Sigma Aldrich, St. Louis, MO) was prepared as a 10 wt% solution in deionized (DI) water. A separate solution of sodium meta‐periodate in equal volume of DI water was prepared at a 1:1 molar ratio of periodate to glucose subunits in 40 kDa dextran (~222 glucose subunits). In a light‐protected container, the two solutions were mixed and allowed to react for an hour at room temperature with constant mixing. Following the oxidation reaction, the odex solution was dialyzed for 48 h at 4°C using an 8 kDa cellulose dialysis membrane (Fisher Scientific, Pittsburgh, PA). The product was frozen at −80°C and lyophilized for 72 h. The final material was stored at −20°C in foil covered falcon tubes until use.

2.3. Aldehyde activity assay

The aldehyde activity assay was performed in accordance with the method outlined in Hickey and Palmer. ²³ In summary, a 2,4‐dinitrophenylhydrazine (DNPH) solution was prepared by mixing 35 mL of 95 v% ethanol, 10 mL of DI water, and 1.5 g of DNPH over ice. 7.5 mL sulfuric acid (H₂SO₄) was added slowly, and the solution temperature was controlled from exceeding 20°C. Once all the H₂SO₄ was added, the vessel was heated until all the DNPH dissolved or the temperature reached 60°C. The solution was allowed to cool overnight, where excess DNPH would precipitate. The extinction coefficient of the DNPH solution at 264 and 360 nm was determined by constructing a calibration curve of DNPH concentration of 0.1 (Stock), 0.075, 0.05, 0.025, and 0.01 mM versus UV–visible spectrometry absorbance. An example of the DNPH calibration curve can be found in Figure S1.

The purpose of the DNPH aldehyde activity assay is to determine the number of active aldehyde groups in solution. This assay was used to calculate how many aldehyde groups are generated on the oxidized dextran chain to determine the appropriate amount of odex solution needed for effective Hb crosslinking. For this assay, the odex sample was diluted 3× with DNPH solution and pipette‐mixed until a yellow precipitate formed (positive for aldehydes). A control sample was made with DI water, where no precipitation should occur. The samples were centrifuged at 5000g for 3 min to separate the precipitates. The supernatant was measured with UV–visible spectrometry using the extinction coefficients determined with the DNPH calibration curve at 264 and 360 nm.^{37, 38} Therefore, the difference between absorbances of the control and odex sample represents the molarity of DNPH consumed, based on the equimolar reaction of DNPH to individual aldehyde groups; this was directly taken as the concentration of aldehydes in the sample.

2.4. Small scale desolvation nanoparticle synthesis

Human Hb (hHb) was purified via the procedure outlined in Palmer et al. ³⁹ from expired human red blood cells (RBCs) acquired from Versiti (Milwaukee, WI) with approval from the Institutional Review Board (IRB, protocol #2022R00000046 under A.F.P.) at The Ohio State University. Purified hHb was buffer exchanged from phosphate buffer into DI water using a 50 kDa Amicon Ultra‐4 centricon filter (3 × 2900 g for 20 min). Over ice, the hHb solution was diluted in DI water to 7.86 mg/mL with a final volume of 5 mL. While mixing the solution at 500 rpm, ethanol (1.5× starting volume) was added at 3 mL/min using a syringe pump. Once all the ethanol was added, odex (20 mg/mL) was added at a 2:1 molar ratio of odex aldehydes to hHb primary amines while mixing rapidly. The mixture was allowed to react at 4°C for 2 h with intermittent manual swirling.

Following the crosslinking reaction, the ethanol in solution was washed out via buffer exchange with four diacycles of DI water over a 0.05 μm TFF module (surface area [SA] = 190 cm², 80 mL/min). A bolus injection of NaCNBH₃ (at a 1:1 molar ratio of aldehydes in solution to NaCNBH₃) was added slowly to the reaction vessel while it continued circulating over the 0.05 μm TFF module with the permeate line closed. The quenching reaction proceeded in the closed TFF circuit for 15 min. Following the quenching reaction with NaCNBH₃, the hHb‐dNPs were washed with 60 mL of 1 M glycine (pH 7.4) over the TFF module (permeate line opened again) to provide additional quenching of the crosslinking reaction. Next, the hHb‐dNPs were buffer exchanged (8 diacycles) into PBS over the TFF module and concentrated. The final particle solution was spun on a centrifuge at 5000 g for 10 min to separate any large protein aggregates that may have formed throughout the synthesis; however, hHb‐dNPs remained suspended in solution. Additional concentration of the hHb‐dNPs was performed with 50 kDa Amicon Ultra‐15 centricon filters for all subsequent data analysis. The benchtop hHb‐dNP synthesis and purification process is outlined in Figure 1.

FIGURE 1.

hHb‐dNP benchtop synthesis and purification process. Figure made with Biorender.com

2.5. Scaled‐up desolvation nanoparticle synthesis

The demonstration of hHb‐dNP scalability was performed with 100× the starting volume (500 mL starting volume) of the small‐scale benchtop system (5 mL starting volume). hHb was buffer exchanged (3 diacycles) from PB into DI water using a 50 kDa PS TFF module (SA = 190 cm²). Over ice, the hHb was diluted in DI water to 7.86 mg/mL and a final volume of 500 mL. While rapidly mixing the solution, ethanol (1.5× starting volume) was added at 70 mL/min to limit the addition time to 10 min. Once all the ethanol was added, odex (20 mg/mL) was added at a 2:1 molar ratio of odex aldehyde groups to hHb primary amines while mixing rapidly. The solution was allowed to react at 4°C for 2 h with rapid mixing.

Following the reaction, the ethanol in the solution was washed out via buffer exchange with four diacycles of DI water over a 0.05 μm TFF module (SA = 1000 cm², 80 mL/min). A bolus injection of NaCNBH₃ (a 1:1 molar ratio of aldehydes in solution to NaCNBH₃) was added slowly to the reaction vessel while it continued running over the 0.05 μm TFF module with the permeate line closed. The quenching reaction in the closed TFF was allowed to proceed for 30 min. Following quenching with NaCNBH₃, the hHb‐dNPs were washed with 1 M glycine (pH 7.4) over the TFF module (2.5 diacycles, permeate line opened again) to provide additional quenching of the crosslinking reaction. Next, the hHb‐dNPs were buffer exchanged (8 diacycles) into PBS over the TFF module and concentrated. The final hHb‐dNP solution was centrifuged at 5000 g for 30 min to separate any large protein aggregate species that may have formed throughout the synthesis, while hHb‐dNPs remain suspended in solution. The scaled‐up hHb‐dNP synthesis and purification process is outlined in Figure 2.

FIGURE 2.

Large‐scale hHb‐dNP synthesis and purification process. Figure made with Biorender.com

2.6. Particle diameter and distribution span

Nanoparticle diameter and size distribution span were measured using tunable resistive pulse sensing (TRPS) on the qNano Gold instrument (Izon Science, Medford, MA) per the manufacturer's instructions. ⁴⁰ A NP200 (measurement range: 80–500 nm) pore (Izon Science) was used with CPC200 200 nm control particles (Izon Science). An additional pore coating step with the proprietary Izon Coating Solution was performed following the manufacturer's instructions to avoid protein charged interactions from occurring with the nanopore and skewing particle measurements. Particle size distribution span was calculated in the qNano system using the following equation:

$$\text{Span}=\frac{d90-d10}{d50}$$

where d90, d10, and d50 are the particle diameters below which 90%, 10%, and 50% of the sample's volume are contained, respectively.

2.7. Zeta potential

The zeta potential of the hHb‐dNPs was measured using a Malvern Panalytical Zetasizer Helix instrument (Worcestershire, UK) according to the manufacturer's recommendations. ⁴¹ For sample preparation, 3 μL of hHb‐dNPs was suspended in 1 mL of DI water and placed in a folded capillary zeta cell.

2.8. Hemoglobin and methemoglobin concentration

Hb and metHb concentrations of hHb‐dNPs were measured using the cyanomethemoglobin assay described in Drabkin et al. ⁴² using UV–visible spectrometry (Olis, Inc., Bogart, GA). Synthesis yield was determined from the initial mass of hHb used in the desolvation reaction and the total mass of Hb in the hHb‐dNPs.

2.9. Scanning electron microscopy

Samples were prepared for scanning electron microscopy (SEM) by buffer exchanging the final hHb‐dNPs from PBS into DI using a 50 kDa Amicon Ultra‐15 centricon filter. One microliter of desalted hHb‐dNPs was pipetted onto a hydrophilic carbon‐backed copper mesh grid (Electron Microscopy Sciences, Hatfield, PA) and allowed to dry completely. Images were obtained using an FEI Apreo microscope (Thermo Fisher Scientific) with beam settings of 25 pA and 5 kV.

2.10. Oxygen equilibrium curve

The Hemox Analyzer (TCS Scientific Corp, New Hope, PA) was used to measure the O₂ equilibrium curve (OEC) for the hHb‐dNPs following the procedure outlined in Palmer et al. ⁴³ The data were fit to the Hill equation to regress the cooperativity coefficient (n_H) and the oxygen affinity (P₅₀, the pO₂ in which 50% of the hHb is saturated with O₂). ⁴⁴

2.11. Hemocompatibility

The hemocompatibility of hHb‐dNPs was evaluated using a method outlined in Gu et al. ⁴⁵ In summary, 1 mL of fresh whole human blood (WB) with approval from the Institutional Review Board (IRB, protocol #2022R00000046 under A.F.P.) at The Ohio State University, was mixed with 1 mL of hHb‐dNPs (15 mg/mL in PBS). The solution was incubated at 37°C for 30 min. The sample was spun down and the supernatant collected. The concentration of cell‐free hHb in the supernatant was measured using the cyanomethemoglobin method ⁴² to detect hemolysis. The negative control consisted of 1 mL of whole human blood mixed with 1 mL of PBS (C1). The positive control consisted of 1 mL of whole human blood mixed with 1 mL of DI water to induce complete hemolysis (C2). The concentration of hHb‐dNPs alone was C3 and NP/WB mixture was C4. The equation to determine percent hemolysis is below. Hemolysis less than 5% indicates hemocompatibility.^{45, 46}

$$\text{Hemolysis}\left(\%\right)=100\times \frac{C4-C1-C3}{C2-C1}$$

2.12. Storage studies

Nanoparticles were stored in one of three cryoprotectants: (1) none; (2) 5 wt% human serum albumin (HSA); or (3) 20 mM trehalose. Samples of each hHb‐dNP/cryoprotectant combination were aliquoted and stored at either 4°C, −20°C, or − 80°C for 1 month. Samples were thawed on days 4, 7, 14, 21, and 28, where they underwent a full biophysical characterization using the techniques listed above.

2.13. Statistical analysis

Data where applicable are reported as the average ± standard deviation. Statistical analysis was performed using JMP software (JMP Pro, Version 18. SAS Institute Inc., Cary, NC).

For comparisons between benchtop and scaled‐up hHb‐dNP particles, a two‐sample pooled t‐test was performed. Normal quantile plots of studentized residuals were used to confirm the data were normally distributed. The Leverne equal variance test was used to confirm the equal variance assumption (p > 0.05). Should both the normal and equal variance assumptions be met, a pooled t‐test with a p < 0.05 indicated significance.

For the storage stability study, a fit model analysis was required to analyze significance. The model for this experiment is listed below:

$$y_{\mathit{\text{ijkl}}}=\mu +{\propto }_i+{\beta }_j+{\gamma }_k+{\left(\mathit{\alpha \beta }\right)}_{\mathit{ij}}+{\left(\mathit{\alpha \gamma }\right)}_{\mathit{ik}}+{\left(\mathit{\beta \gamma }\right)}_{\mathit{jk}}+{\left(\mathit{\alpha \beta \gamma }\right)}_{\mathit{ijk}}+b_l+{\epsilon }_{\mathit{\text{ijkl}}}$$

where y _ijkl is the response variable of interest at a given storage condition, μ is the mean, α corresponds to the effect of storage time (day, i = 1, 4, 7, 14, 21, 28), β is the effect of temperature (j = 4°C, −20°C, −80°C), γ is the effect of the cryoprotectant (k = none, HSA, trehalose), b _l is the random block intercept for batch variability, and ε _ijkl is the residual error. The variables in the middle of the equation are the cross effects or interaction effects between day, temperature, and cryoprotectant. The normal distribution and equal variance assumptions were validated with studentized residuals. Following their validation, the ANOVA effects test was run to determine which of the effects were significant (p < 0.05). For any significant effects, subsequent Tukey Tests were used to compare the groups within those effects for individual significance (p < 0.05). Only effects from a significant cross effect between day * temperature * cryoprotectant were shown on figures. Each parameter ANOVA effects tests, post‐hoc Tukey tests, and blocking variable analyses are in the supplemental section (Tables S1 through S15).

3. RESULTS AND DISCUSSION

In this study, hHb‐dNPs were synthesized using desolvation precipitation. The purpose of this work was to design a novel HBOC that is composed entirely of hHb and is consistent across several production batches to improve upon existing HBOC technologies as outlined in the Introduction section. In the design of hHb‐dNPs, there were several ideal parameters that were considered for an effective HBOC. First, hHb‐dNPs, as an HBOC, must effectively bind and release O₂. Additionally, these materials need to be synthesized on a large enough scale to be able to meet volume demands for blood and blood products.

Next, the hHb‐dNPs should have a monodisperse size distribution between 100 and 300 nm. Particles with a diameter greater than 100 nm prevent blood vessel extravasation into the tissue space, a common occurrence for unmodified, cell‐free Hb and low MW HBOCs (<500 kDa). Extravasated Hb scavenges NO in muscle tissues, causing vasoconstriction and systemic hypertension, avoidable adverse side effects.^{47, 48, 49, 50, 51} Additionally, by ensuring the particle size remains below 300 nm, premature clearance from the bloodstream via phagocytosis can be ameliorated.^{52, 53} Maintaining a monodisperse particle size within this range should potentially increase the material's circulatory half‐life, ensuring adequate O₂ delivery for longer periods of time.

Finally, hHb‐dNPs should have a negative surface charge, which helps ensure that particles experience repulsive interactions with each other to avoid particle aggregation or condensation. In addition, most cells and proteins in the bloodstream also have a negative surface charge, thus having a similar negative charge will avoid interactions with bloodstream systems that may trigger an immunogenic effect.^{53, 54, 55}

3.1. Benchtop hHb‐dNP synthesis

3.1.1. Synthesis design

The synthesis of hHb‐dNPs was based on the desolvation mechanism previously used to synthesize albumin (both HSA and BSA)^{15, 16, 17, 18, 19} and bovine Hb (bHb) NPs. ²³ bHb‐dNPs were synthesized in prior work as a potential next generation HBOC, but human Hb dNPs have yet to be explored. Since both bovine and human Hb have been used as the precursors in the synthesis of HBOCs, the synthesis of hHb‐dNPs was a logical step in HBOC exploration.

For the majority of albumin and bHb dNP syntheses, ethanol was selected as the antisolvent and DI water or aqueous buffer are typically chosen as the protein solvent.^{15, 16, 17, 19, 23, 56} Although other organic solvents such as acetone and methanol have been successfully tested in albumin NP systems,^{57, 58} the well‐established use of ethanol, in addition to it being generally recognized as safe in the pharmaceutical industry, ⁵⁹ was why it was selected as the desolvating agent in this current study.

Purified hHb was buffer exchanged into DI water prior to dilution as the salts present in the hHb storage buffer can serve as nucleation points, where particle formation occurs rapidly and leads to uncontrolled particle growth and aggregate formation.^{16, 17, 20, 21} The starting solution had a concentration of 7.86 mg/mL hHb. In preliminary studies, hHb concentrations higher than this resulted in immediate aggregation of hHb‐dNPs with the addition of EtOH. Although most albumin NP syntheses were performed at room temperature, the hHb‐dNP synthesis was performed at 4°C to limit Hb oxidation and particle diffusion, ensuring small particle formation and to avoid uncontrolled particle growth.^{20, 21} The diluted solution of hHb in DI water maintained at low temperatures allowed for the formation of small hHb‐dNPs with a limited risk of coagulation from decreased frequency of particle‐particle interactions.^{34, 60}

The desolvation reaction occurred when ethanol was added at 3 mL/min ²³ to the hHb solution mixing at 500 rpm. A fast mixing rate ensures that the miscible solution homogenizes quickly for a more even nucleation and generation of monodisperse particles.^{20, 61} Oxidized dextran (odex) was selected as the crosslinker over glutaraldehyde (GA) as it follows the same crosslinking reaction chemistry as GA but is more biocompatible and low cost.^{62, 63, 64, 65} The large size of odex (40 kDa) also promotes the formation of a coating on the surface of the NPs that helps to decrease the particle surface charge and therefore could potentially increase the particle's circulatory half‐life.^{53, 54, 55} The crosslinking reaction proceeded for 2 h to aid with hHb entrapment and particle stability. Although a 1‐h reaction time produced smaller particles in preliminary experiments, the yield was lower versus a 2‐h reaction.

Following synthesis, the hHb‐dNPs underwent several purification steps. Following the crosslinking reaction, the hHb‐dNPs were buffer exchanged into DI water as the hHb‐dNPs would crash out of solution with the addition of quenching agents to the EtOH‐rich reaction solution. Quenching of the crosslinking reaction was also performed with two reagents: NaCNBH₃ and glycine. NaCNBH₃ was used as a hydrogen donor to reduce the Schiff base formed between the odex aldehydes and the primary amines on the surface of hHb. ⁶⁶ Glycine was then used to cap any unreacted aldehydes to ensure no additional crosslinking occurred with the highly reactive aldehyde groups.^{23, 67} TFF was used versus centrifugation for particle purification as it allows for a higher throughput and is ultimately a more scalable process. Finally, excess washing steps with PBS were performed to remove all unreacted hHb, NaCNBH₃, glycine, and odex from the hHb‐dNPs.

3.1.2. Benchtop hHb‐dNP characterization

The hHb‐dNPs were initially synthesized and optimized on a small scale (5 mL starting volume). The final nanoparticles were composed entirely of hHb and covalently stabilized with odex. The benchtop synthesis exhibited a yield of 56.9 ± 12.0%. This is on par with some albumin and bHb‐dNP syntheses, which typically exceed 60%^{17, 56} but a few studies achieved upwards of 95% yield.^{15, 23} A few studies showed that increasing the ratio of ethanol to DI water increased the particle yield,^{19, 68} although an ethanol ratio of 1.5× the volume of DI water produced a high yield for previously synthesized bHb‐NPs, ²³ it may be worth exploring the effects of higher ratios for hHb‐dNP synthesis. hHb‐dNPs exhibited a negative surface potential at −31.8 ± 2.7 mV. Below −20 mV is ideal for biological based systems to ensure repulsive interactions in vivo. This heavily negative surface charge is consistent with other protein NPs in literature.^{16, 17, 23}

Figure 3A shows representative size distribution and Figure 3B shows the morphology of the hHb‐dNPs. The particle size distribution was measured using Izon's TRPS qNano Gold system. The final materials synthesized on the benchtop scale had an average particle diameter of 138.3 ± 6.5 nm with a distribution span of 0.28 ± 0.04. The diameter of hHb‐dNPs is smaller than the reported sizes from Hickey and Palmer ²³ and Chen et al. ⁵⁶ who had mean particle diameters of 243 and 580 nm for bovine Hb‐dNPs, respectively. The smaller particle diameter corresponds to a larger surface area‐to‐volume ratio, better enabling potential O₂ transport in vivo. A span value close to zero is indicative of a monodisperse particle size distribution. Finally, the hHb‐dNPs were imaged with SEM to determine their morphology and confirm their size. The hHb‐dNPs were spherical in shape, consistent with all previous albumin and bHb dNPs, and the size of the NPs was confirmed to be between 100 and 200 nm.

FIGURE 3.

Benchtop hHb‐dNP size distribution and morphology (A) Representative hHb‐dNP size distribution, and (B) Representative SEM image.

In addition to hHb‐dNP size and morphology, the O₂ equilibrium curve (OEC) of the hHb‐dNPs was measured using a Hemox Analyzer and the P₅₀ (oxygen affinity) and Hill coefficient (n_H) were regressed using the Hill equation ⁴⁴ (Figure 4). The small‐scale hHb‐dNPs had a P₅₀ of 0.88 ± 0.27 mm Hg and a n_H of 0.72 ± 0.05. The benchtop hHb‐dNPs have a left‐shifted OEC compared to hHb with a distinct loss in the sigmoidal shape of hHb, characteristic of hHb's cooperative O₂ binding behavior. Similar to other chemically modified HBOCs, hHb‐dNPs loss of cooperativity (n_H ≤1) is a common result of the chemical crosslinking used to stabilize these materials. Chemical crosslinking will typically lock the once flexible hHb molecule in either its tense (T) or relaxed (R) quaternary state or a mixture of the two quaternary states depending on the reaction conditions. The OEC of benchtop hHb‐dNPs is almost identical to that of R‐quaternary state polymerized Hb (PolyhHb), another HBOC that has demonstrated an ability to increase chemosensitization in hypoxic tumor environments,^{69, 70} demonstrating a potential application for this new HBOC.

FIGURE 4.

Oxygen equilibrium curves for small‐ and large‐scale hHb‐dNPs, relaxed‐quaternary state polymerized human Hb (R‐state PolyhHb), and hHb.

The small, spherical morphology, replicable synthesis, and high O₂ affinity of the benchtop hHb‐dNPs demonstrated their potential value as a next generation HBOC candidate. To be commercially relevant, the hHb‐dNPs must also demonstrate a scalable synthesis route to meet the supply demand for blood and have a stable shelf‐life.

3.2. Scaled‐up hHb‐dNP synthesis

3.2.1. Scaled‐up synthesis design

To demonstrate the scalability of this synthesis route, the initial volume was scaled 100× to 500 mL. Successful demonstrations at this scale should be sufficient to assess the ability to scale this system to other larger volumes to meet potential blood demand. For the large‐scale system, operating parameters that were kept constant between scales were the starting hHb concentration, the volume ratio of EtOH to the initial hHb solution, and the crosslinking reaction time. The volumes of each reagent were scaled linearly with the system volume.

For the large‐scale system, the mixing profile of the reaction vessel had to ensure that the antisolvent was quickly homogenized. This necessitated the use of an impellor‐based reaction vessel with ports that fed the EtOH directly into the main impellor path. With high speeds, this mixing system helped to ensure fast homogenization of the antisolvent to promote rapid particle nucleation, resulting in smaller and more uniform particles.^{20, 61} In addition, the rate of EtOH addition was increased to limit the addition time to 10 min. If the same addition rate of 3 mL/min was used at this scale, particle nucleation would have occurred over an extended time period, creating a wider size distribution as particles that nucleated early would have entered the growth phase before the crosslinker was added.^{20, 61, 71} By increasing the addition rate, the monodisperse distribution of the particles was maintained at a larger scale. Table 1 shows the reaction parameter differences between the benchtop and scaled‐up hHb‐dNP systems.

TABLE 1.

Synthesis parameters of benchtop and scaled‐up hHb‐dNP synthesis.

Property	Benchtop	Scaled‐up
Starting volume (mL)	5	500
Starting hHb concentration (g/L)	7.86	7.86
Ethanol‐to‐starting solution volume ratio	1.5:1	1.5:1
Ethanol addition rate (mL/min)	3	70
Reaction time (h)	2	2
Total synthesis time + preparation (h)	6–8	28–32

Note: Yellow highlight indicates synthesis parameter changes.

3.2.2. Scaled‐up hHb‐dNP characterization

Table 2 lists the biophysical properties of the small‐ and large‐scale hHb‐dNPs, R‐state PolyhHb, and hHb. All values for R‐state PolyhHb are from Khan et al. ⁵ except zeta potential, which was measured directly. The characterization of the large‐scale hHb‐dNPs followed the same protocols as the benchtop materials. The scaled‐up hHb‐dNP yield of 56.4 ± 16.5% was insignificantly different compared to the benchtop synthesis (p > 0.05). This similarity in yield confirms that the mixing profiles between the two scales was conducive to rapid, uniform particle formation. The final protein mixture had 1.9 ± 1.8% metHb, an improvement to the 9.52% metHb reported by Hickey and Palmer ²³ for bHb‐dNPs. Despite having a multiday synthesis time, this low amount of hHb oxidation ensures that most of the hHb in the hHb‐dNPs can transport oxygen. ⁷²

TABLE 2.

Biophysical characteristics for small‐ and large‐scale hHb‐dNPs, R‐state PolyhHb, and hHb.

Property	hHb	R‐State PolyhHb 5	Benchtop hHb‐dNP	Scaled‐up hHb‐dNP
Yield (%)	N/A	N/A	56.9 ± 12.0	56.4 ± 16.5
Methemoglobin (%)	0.9 ± 0.7	2.3 ± 0.9	N/A	1.9 ± 1.8
Average diameter (nm)	5.4 ± 0.1 a , 5	24.4 ± 1.9 a	138.3 ± 6.5	142.8 ± 7.9
Span	N/A	N/A	0.28 ± 0.04	0.35 ± 0.05
Zeta Potential (mV)	−3.4 ± 0.5	7.6 ± 0.3	−31.8 ± 2.7	−29.0 ± 2.6
P50 (mm Hg)	11.13 ± 0.25	0.9 ± 0.3	0.88 ± 0.27	2.16 ± 0.50*
n H	2.65 ± 0.19	1.0 ± 0.1	0.72 ± 0.05	0.92 ± 0.04*

^a Values measured using dynamic light scattering, not TRPS qNano.

^*Statistical significance in accordance with a two‐sided pairwise t‐test comparison between the small‐ and large‐scale synthesis methods.

Similar to the benchtop synthesis, the scaled‐up system produced monodisperse, spherical nanoparticles with a negative surface potential. The large‐scale hHb‐dNPs had an average diameter of 142.8 ± 7.9 nm with a distribution span of 0.35 ± 0.05, an insignificant difference compared to the benchtop hHb‐dNPs (p < 0.05 for both). hHb‐dNPs of this size are large enough to prevent blood vessel extravasation, and small enough to have a large surface area‐to‐volume ratio to improve O₂ transport. Figure 5A is a representative size distribution and Figure 5B shows the SEM for the large‐scale hHb‐dNPs. The zeta potential for the hHb‐dNPs is −29.0 ± 2.6 mV, which is below the recommended surface charge of at least ±20 mV for the stabilization of nanoparticles both electrostatically and sterically. ¹⁶ The large‐scale surface charge is insignificantly different from hHb‐dNPs synthesized at the benchtop scale (p > 0.05).

FIGURE 5.

Large‐scale hHb‐dNP size distribution and morphology (A) Representative hHb‐dNP size distribution and (B) Representative SEM image.

Figure 4 also shows the OEC for the large‐scale hHb‐dNPs. The P₅₀ for these particles are 2.16 ± 0.50 mmHg with a Hill coefficient of 0.92 ± 0.04. Although both the P₅₀ and Hill coefficient are significantly different from the benchtop synthesis (p < 0.05 for both), the OEC still falls in the range for a R‐quaternary state HBOC.^{5, 23}

3.3. Hemocompatibility

In addition to the biophysical properties of hHb‐dNP, preliminary analysis of the safety profile of hHb‐dNPs was performed. Samples of hHb‐dNPs were mixed with fresh whole human blood at a 1:1 ratio and incubated for 30 min at 37°C. The concentration of hHb in the supernatant was measured to monitor for hemolysis. The hHb‐dNPs induced 3.3 ± 1.4% hemolysis. As this is less than 5% and there were no observations of coagulation, it is a hemocompatible product according to ISO 10993‐4:2017.^{45, 46}

3.4. Scale‐up storage studies

Finally, the stability of hHb‐dNPs was monitored over a month at various storage temperatures and with different cryoprotectants. The goal of this study was to gain an understanding of the stability of hHb‐dNPs and discern the optimal storage conditions to extend their shelf life. Currently, resuscitation solutions such as whole blood and packed RBCs have limited ex vivo shelf lives of 21 (if stored with citrate–phosphate‐dextrose [CPD]) and 42 days, respectively. In this storage study, each scaled‐up batch of hHb‐dNPs (n = 4) was aliquoted and co‐stored with either 5 wt% HSA, 20 mM trehalose, or no cryoprotectant. Each aliquot was placed directly in their designated storage locations of 4°C, −20°C, or −80°C. Full biophysical characterization was performed on storage days 1, 4, 7, 14, 21, and 28 to monitor changes to key properties.

3.4.1. Qualitative observations

When hHb‐dNPs are synthesized they appear as a turbid solution with a vibrant grapefruit pink/red color. Over the course of the storage study, the color of the solution changed depending on the conditions in which it was stored. All samples stored at −80°C maintained close to their native color throughout the study, regardless of cryoprotectant used. On the other hand, the sample co‐stored with HSA at 4°C quickly turned a rusty brown color. Based on qualitative analysis alone, the 4°C HSA sample may be experiencing auto‐oxidation quicker than its other 4°C co‐stored counterparts and those stored at −80°C. Qualitative color observations over time can be seen in Figure S2.

In addition to color, the presence of aggregates was also qualitatively monitored. In this study, all samples stored at −20°C with no cryoprotectant or trehalose experienced gelation by day 4. An image of the material stored at this temperature can be found in Figure S3. The −20°C HSA co‐stored samples experienced increasing degrees of particle aggregation until day 28, where almost all hHb‐dNPs precipitated out of solution. Therefore, all additional characterizations only took place on samples stored at 4°C and − 80°C.

At 4°C, there was no evidence of particle aggregation over 28 days of ex vivo storage. At −80°C, trehalose and the no cryoprotectant co‐stored samples showed evidence of aggregate formation by day 4. There was no evidence of aggregates in HSA co‐stored particles at −80°C. The number of aggregates observed in the −80°C samples did not change much with time. Based on their responses to storage at sub‐zero temperatures, it is believed that the slow cooling of the samples to their storage temperature might have disrupted the stability of the particles. It is hypothesized that snap freezing the hHb‐dNPs prior to placing them at their desired storage temperature (−20°C or − 80°C) might mitigate the adverse effects observed. For all subsequent characterization, the −80°C samples were spun on a hand centrifuge for 30 seconds to pellet any aggregates to avoid skewing of measurements due to noise or light scattering from large particles.

3.4.2. Hemoglobin and methemoglobin concentration

Hb and metHb concentrations were measured using the cyanomethemoglobin assay and plotted on Figures 6 and 7. For hHb concentration, the storage time was identified as having a significant effect on the model (F (5) = 3.65, p = 0.0054). Tukey tests indicated that days 21 and 28 had significantly different hHb concentrations than day 1. At 4°C (Figure 6A), there was no apparent change in concentration with time. At −80°C (Figure 6B), there is a steady decline in hHb concentration for hHb‐dNPs stored with no cryoprotectant. This is most likely a result of the formation and subsequent settling of aggregates following sample thaw. None of the changes in individual storage time points were significant (p > 0.05).

FIGURE 6.

Hemoglobin concentration of hHb‐dNPs stored at (A) 4°C and (B) −80°C with various cryoprotectants for 28 days. Figure made with Biorender.com

FIGURE 7.

Methemoglobin concentration of hHb‐dNPs stored at (A) 4°C and (B) −80°C with various cryoprotectants for 28 days. Figure made with Biorender.com

Figure 7A shows the change in metHb level over time for 4°C and Figure 7B −80°C storage, respectively. The metHb level increased faster and resulted in a higher metHb concentration for samples stored at 4°C compared to −80°C. At −80°C, the metHb level remained relatively consistent over the course of the storage study at less than 5% for each cryoprotectant. At 4°C, there was a steady increase in the metHb level over the month for the hHb‐dNPs co‐stored with trehalose and no cryoprotectant. These samples were only at ~5% metHb by day 28, but the trendline indicates this value will most likely continue to increase with time without intervention. Interestingly, the 4°C HSA co‐stored hHb‐dNPs had a higher oxidation rate, exceeding 10% by day 28. This data is consistent with the qualitative color observation where the solution quickly turned a rusty brown color.

Storage time (F (5) = 7.34, p < 0.0001) and the cross effect of (storage time) × temperature (F (5) = 13.76, p < 0.0001) significantly affected the model for metHb. Subsequent Tukey test for storage time indicated that days 4, 7, 14, and 21 were significantly different than day 1. The Tukey test for cross effect of (storage time)*temperature showed at 4°C, day 21 and 28 were significantly different than day 1, 4, and 14. At −80°C, day 1 was significantly different than day 4, 14, and 21 and day 28 was significantly different than day 7. Finally, day 4 and day 28 were significantly different between the two storage temperatures.

3.4.3. Particle size distribution

The hHb‐dNP size distribution was measured with Izon Qnano TRPS. The goal of this measurement was to see if there was any particle growth or deterioration in storage. As the visible aggregates were spun down to avoid clogging the pore on the Qnano system, particle growth aside from the visible aggregates would be observed with the formation of a shoulder or tail on the right‐hand side of the particle size distribution. Particle deterioration would be observed with a left‐hand tail formation. Both issues would quantitatively be observed with an increase in the span of the distribution. ANOVA effects test for mean particle diameter revealed that storage time (F (5) = 3.47, p = 0.0061) is the only parameter to affect the model. Subsequent Tukey tests indicated that day 21 is significantly different than day 1 and 4. For span, both storage time (F (5) = 2.55, p = 0.0324) and cryoprotectant (F (2) = 4.92, p = 0.0091) had significant effects on the model. Tukey tests for storage time showed that day 28 was significantly different than day 1. Tukey tests for cryoprotectant indicated that HSA had a significantly different span than trehalose or no cryoprotectant.

At 4°C, the mean particle diameter remained constant over 28 days of storage regardless of cryoprotectant (Figure 8A). In addition, the span also remained constant (Figure 8C), indicating that the particles stored at this temperature were stable against particle growth or deterioration. It is interesting to note that the HSA co‐stored particles consistently had a slightly higher mean particle diameter compared to hHb‐dNP co‐stored with trehalose or no cryoprotectant (Figure 8A,B) This is most likely due to physical adsorption of HSA to the surface of the NP.^{73, 74} The additional coating of HSA on the nanoparticles may not be consistent and therefore may also be contributing to the significantly higher span (Figure 8C,D) This hypothesis could be confirmed with future cryo‐EM or TEM^{73, 75} to see if there is a visible corona around the hHb‐dNP and/or wider variation in particle size compared to the particles with no cryoprotectant, which is outside the scope of this study.

FIGURE 8.

Mean diameter (A and B) of hHb‐dNPs stored at (A) 4°C and (B) −80°C and size distribution span (C and D) of hHb‐dNPs stored at (C) 4°C and (D) −80°C with various cryoprotectants for 28 days. Figure made with Biorender.com

At −80°C, there is a slight increase in mean particle size for hHb‐dNPs stored with no cryoprotectant and trehalose, but this slight increase occurred between day 1 and day 4 and plateaued for both conditions. (Figure 8B) This increase in mean particle size was also reflected in an increase in the span (Figure 8D) to a more visible degree for hHb‐dNPs stored with no cryoprotectant than trehalose. The increase in mean particle diameter and span is consistent with the qualitative observation of aggregates in solution, shifting the mean to the right and coinciding with formation of a tail in the particle size distribution. On the other hand, HSA seems to be a more efficient cryoprotectant for hHb‐dNPs at −80°C. This is evidenced by a constant value in the mean particle diameter and span as well as no visible formation of aggregates.

3.4.4. Oxygen equilibrium

O₂ equilibrium curves were measured using the Hemox Analyzer for all samples on days 1, 14, and 28 of storage. The goal was to monitor changes in the equilibrium O₂ binding of hHb‐dNPs over the course of storage. The P₅₀ for hHb‐dNPs remained at ~2 mm Hg for all cryoprotectants and storage temperatures. (Figure 9A,B) Interestingly, for all samples, there was a drop in P₅₀ to about 1 mm Hg midway (day 14) through the study, which resolved itself back to the original value by the end of the storage study. Fit model ANOVA analysis revealed that storage time (F (2) = 14.32, p < 0.0001) is the only parameter to have a significant effect on the P₅₀ of hHb‐dNPs. Subsequent Tukey test indicates that day 14 is significantly different than days 1 and 28. As the data fluctuation is occurring over small P₅₀ values, technical variability is a likely contributor.

FIGURE 9.

P₅₀ (A and B) of hHb‐dNPs stored at (A) 4°C and (B) −80°C and hill coefficient (C and D) of hHb‐dNPs stored at (C) 4°C and (D) −80°C with various cryoprotectants for 28 days. Figure made with Biorender.com

The Hill coefficient describes the cooperativity of O₂ binding with Hb. For this experiment, the cooperativity of hHb‐dNPs remained at 1 over 28 days of storage with no effects from temperature or cryoprotectant (Figure 9C,D). A Hill coefficient of 1 corresponds to non‐cooperative O₂ binding, consistent with most HBOCs that have chemical crosslinking.^{5, 23} Statistical analysis confirmed that there were no significant effects on the model (p > 0.05) from any variables. Overall, the O₂ binding properties remain consistent with the source material even after 28 days in storage.

3.4.5. Zeta potential

Zeta potential was measured using a Melvern ZetaSizer. The zeta potential for hHb‐dNP remained at about −30 mV throughout the storage study, regardless of the cryoprotectant or storage temperature (Figure 10). HSA‐stored hHb‐dNPs consistently had a slightly higher surface charge, but that is most likely attributed to HSA coating of the hHb‐dNP changing its surface properties. Statistical analysis confirmed that there is no significant effect to the model for zeta potential from any of the experimental parameters (p > 0.05).

FIGURE 10.

Zeta potential of hHb‐dNPs stored at (A) 4°C and (B) −80°C with various cryoprotectants for 28 days. Figure made with Biorender.com

Overall, hHb‐dNPs demonstrated stability at various temperatures over the course of 28 days of storage. Based on all the data above, hHb‐dNPs stored at −80°C with HSA as a cryoprotectant seem to be the most optimal storage condition of the parameters tested thus far. hHb‐dNPs stored under these conditions showed no signs of particle aggregation or deterioration with time or freeze–thaw cycles. The material also maintained a low oxidation rate demonstrated with a metHb value similar to that of its starting material at day 1. hHb‐dNP stored at 4°C with no cryoprotectant or trehalose also shows similar stability in terms of the particle size distribution, but its linear increase in oxidation to metHb may prove to not be a beneficial long‐term storage technique.

Moving forward, it may be useful to analyze the effects of snap‐freezing the samples before placing them in their respective storage temperatures to avoid temperature gradients affecting the stability of the particles during the freezing process. Although outside the scope of this study, an analysis of the stability of hHb‐dNPs for longer periods of time (6 months to 1 year) as well as lyophilization would be important to study.

4. CONCLUSION

This work demonstrates the successful synthesis of hHb‐dNPs using an anti‐solvent precipitation‐based synthesis technique. hHb‐dNPs were consistently synthesized to be between 130 and 150 nm in diameter with a spherical morphology and negative surface charge. The synthesis of hHb‐dNP was scalable, producing similar products regardless of the batch size, and hHb‐dNPs were hemocompatible. hHb‐dNPs were also shown to be stable over a month, maintaining their uniform size distribution and O₂ binding properties when stored at −80°C with HSA as a cryoprotectant. These particles have room for optimization to improve yield and O₂ affinity in order to be effective HBOC candidates for use in transfusion medicine to possibly treat hypoxic conditions. Therefore, hHb‐dNPs are a promising new HBOC that may also allow the co‐delivery of pharmaceuticals and other agents with an oxygen carrier thus potentially expanding the reach of HBOCs to drug delivery fields.

CONFLICT OF INTEREST STATEMENT

We declare a conflict of interest: a provisional patent application was filed based on the materials described in this work (US 2022/0313792 A1).

Supporting information

Data S1.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.