Engineering Synthetic Erythrocytes as Next-Generation Blood Substitutes

Abstract

Blood scarcity is one of the main causes of healthcare disruptions worldwide, with blood shortages occurring at an alarming rate. Over the last decades, blood substitutes has aimed at reinforcing the supply of blood, with several products (e.g., hemoglobin-based oxygen carriers, perfluorocarbons) achieving a limited degree of success. Regardless, there is still no widespread solution to this problem due to persistent challenges in product safety and scalability. In this Review, we describe different advances in the field of blood substitution, particularly in the development of artificial red blood cells, otherwise known as engineered erythrocytes. We categorize the different strategies into natural, synthetic, or hybrid approaches, and discuss their potential in terms of safety and scalability. We identify synthetic engineered erythrocytes as the most powerful approach, and describe erythrocytes from a materials engineering perspective. We review their biological structure and function, as well as explore different methods of assembling a material-based cell. Specifically, we discuss how to recreate size, shape, and deformability through particle fabrication, and how to recreate the functional machinery through synthetic biology and nanotechnology. We conclude by describing the versatile nature of synthetic erythrocytes in medicine and pharmaceuticals and propose specific directions for the field of erythrocyte engineering.

Graphical Abstract

Engineered red blood cells (erythrocytes) are extensively researched as blood substitutes to minimize the negative effects of blood shortages in healthcare. However, no widespread solution has yet been found. In this Review, we describe the most recent and important advances in erythrocyte engineering, and discuss ways of fabricating the next generation of blood substitutes based on functional materials.

1. Introduction

Blood supply and demand is often at a thin equilibrium as hospitals, blood banks, and research institutes are completely reliant on blood donations to maintain supply. From 2008 to 2018, whole blood collections increased in all World Health Organization (WHO) regions^[1], but multiple factors are expected to hinder this trend. An aging population, lack of effective screening methods for certain infectious diseases (e.g., malaria)^[2], wars, natural disasters, and pandemics^[3] are expected to reduce donor availability and increase blood demand, further straining this fragile equilibrium.

Blood scarcity is also attributed to biological factors. Preserved blood elements have short shelf-lives, ranging from 5 days for platelets^[4] to 42 days for red cell concentrates^[5], after which the blood products are no longer suitable for use. Moreover, not all blood is universally safe to transfuse. Red blood cells, or erythrocytes, may or may not present surface antigens which, when transfused to an incompatible recipient, can trigger a life-threatening reaction. These antigens famously originated the categorization of blood into different types, allowing the safe matching of donors and patients. Blood whose erythrocytes lack the two most common antigens, O-negative blood, is considered the only universal blood type, and is therefore highly needed. Strikingly, only 7% of the population have it.^{[6, 7]} Antigen-low and antigen-free erythrocytes seem to be, therefore, in high demand but in low supply.

To meet the current need for transfusion supplies, erythrocyte engineering is one of the most promising approaches. Using functional materials, it is possible to create tailored artificial erythrocytes without any natural adverse elements (e.g., antigens) and with synthetically added benefits (e.g., longer shelf-life, higher scalability, faster manufacturing). Moreover, erythrocytes are inherently minimalistic, as they neither contain a nucleus nor mitochondria, which renders them comparably easier to engineer than other cells.

Erythrocytes play a key role in blood function. Most blood substitutes have therefore aimed at replicating erythrocyte function. The history of blood substitution started with saline, serum, and gelatin solutions in the first half of the 20^th century^{[8, 9]}, and evolved into hemoglobin products, perfluorocarbons, synthetic O₂ carriers, and stem cell-derived erythroids throughout the second half. Over the years, numerous hemoglobin-based products and perfluorocarbons reached clinical trial stages, although with limited success due to cardiovascular side effects or complement activation symptoms. Currently, only one product (Hemopure^®, hemoglobin-based) is approved for human transfusions, and only marketed in South Africa and Russia.^[10] Nowadays, most advances in blood substitution are focused on improving product safety and production scalability, with several comprehensive reviews recently published on the topic.^[11–14]

In this review, we summarize all the current advances in erythrocyte engineering and present them from a new perspective of natural, synthetic, or hybrid approaches. Moreover, we provide new considerations on the potential of fully synthetic approaches in blood substitution, focusing on their safety and scalability. We conduct a novel point-by-point analysis on how to engineer an erythrocyte using functional materials, covering both common (e.g., O₂ transport) and underexplored (e.g. size, biomechanics) aspects. We conclude the review by addressing current challenges, limitations, and future directions in the field of erythrocyte engineering.

2. Structure and Function of Erythrocytes

Erythrocytes are the most abundant cellular components of blood, accounting for approximately 35–45% of its volume, a value known as hematocrit.^[15] Fulfilling their multiple roles in blood, these cells circulate for approximately 120 days before natural removal due to biological aging.^[16] Erythrocytes are enucleated, biconcave-shaped cells, with a diameter range of 7.5 to 8.7 μm and thickness range of 1.7 to 2.2 μm (Figure 1A).^[17] Their shape is determined by a lipid-rich cell membrane containing phospholipids and cholesterol, alternated with membrane proteins such as glycophorins, Na²⁺ and K⁺-dependent adenosine triphosphatases, band 3 anion (Cl⁻ and HCO₃⁻) transporter proteins, ankyrin, and others.^[17] Through band 3 and ankyrin, the cell membrane is tethered to an intracellular 2D spectrin network, forming a structural arrangement responsible for upholding the shape and mechanical integrity of the cell (Figure 1B).^[17–19]

Figure 1.

A. Size and shape of a natural erythrocyte. B. Cell membrane of an erythrocyte with asymmetric lipid distribution between leaflets. Integral proteins anchor the membrane to the inner spectrin network. Marker of “self’, CD47, and surface antigens from Rh and ABO group systems also integrate the membrane. Based on ref.^[35] C. Lipid composition of the outer leaflet of the erythrocyte membrane. SM: sphingomyelins; PC: phosphatidylcholines; PS: phosphatidylserines; PI: phosphatidylinositol, a glycolipid. Based on ref.^[20] D. Mechanical properties of erythrocytes and respective value ranges. E. Fahräeus-Lindqvist effect, whereby blood flow in small vessels (<300 μm) is divided into an axial flow, rich in erythrocytes and other cells, and a cell-free marginal flow which facilitates gas and nutrient exchange in tissues.

Erythrocyte membranes play a key role in cell signaling and function. The phospholipid-rich bilayer of erythrocytes is asymmetric, with distinct lipid compositions in its inner and outer leaflets.^[20] In a healthy erythrocyte, the outer leaflet is rich in sphingomyelins (SM), phosphatidylcholines (PC), and phosphatidylinositol (PI) (Figure 1C), while the inner leaflet is mostly composed of PC, phosphatidylethanolamine (PE), and phosphatidylserine (PS).^[20] PS lipids are recruited to the outermost leaflet of the membrane to signal cell senescence or the need for macrophage sequestration and removal from circulation.^[21] This class of lipids also showed higher pro-coagulating activity.^{[22, 23]} Cholesterol is also present in the membrane, although its exact distribution between leaflets remains controversial.^{[20, 24]} Cholesterols interact strongly with sphingomyelins through hydrophobic interactions to form liquid-ordered phases, or regions with tight lipid packing, which often include membrane proteins involved in signal transduction^[25] and erythrocyte reshaping.^[26] Erythrocyte membranes also incorporate markers which regulate cell circulation and immune response: the “marker of self”, CD47, is an integrin-associated protein that marks the non-foreign identity of the cells. It interacts with the surface of macrophages as a recognition marker that allows the free circulation of cells in the bloodstream.^[27] Conversely, surface antigens are responsible for a severely adverse response upon contact with foreign, incompatible blood antibodies.^[28]

The mechanical properties of erythrocytes, although frequently studied, present considerable variability (Figure 1D). These cells demonstrate both solid-like elastic and fluid-like viscous properties and typically exhibit viscoelastic behavior under external stress. Young’s modulus (E) of 1–26 kPa^[29–32], shear modulus (μ) of 5–10 μN m⁻¹,^[19] and elastic constant (k_e) of 1.9–19 μN m⁻¹ have been reported.^[19] Membrane viscosity (η_m), while subject to conflictive studies, is most accurately estimated at 10⁻⁷-10⁻⁶ mPa·s.^[33] Creep, or the loading response at a fixed force, generally reaches a plateau after 0.1–0.6 s, with higher η_m resulting in slower dynamics.^{[33, 34]} Similarly, stress relaxation is also slower with increased η_m, reaching a plateau after 0.2–0.6 s.^[33] This considerable variability is attributed not only to the cell’s intrinsic spectrin network fluctuations, but also to the use of different characterization techniques, with multiple reports addressing this issue.^{[19, 35, 36]}

In terms of rheology, erythrocytes are the primary contributors to the non-Newtonian behavior of blood, in particular its unique flow profile under microcirculation. Known as the Fåhraeus-Lindqvist effect, this phenomenon consists in the shear-thinning of blood observed in small blood vessels (diameters < 300 μm)^[37], caused by a preferential flow of erythrocytes in the axis and a slower flow of cell-free plasma in the margins (Figure 1E).^[38] This flow enables a more efficient exchange of gas, nutrients, and metabolites between the plasma and the surrounding tissues, further facilitated by the thin endothelium lining of the vessels.

Erythrocytes are largely responsible for blood function, particularly oxygen (O₂) and carbon dioxide (CO₂) transport, nitric oxide (NO) metabolism, redox regulation, and blood rheology.^{[15, 39]} Hemoglobin (Hb) is an intracellular protein responsible for the transport of O₂ and CO₂, as well as for the metabolism of NO. Each Hb molecule is composed of four units, two α- and two β-globins, each with a ferrous (Fe²⁺) heme group responsible for binding O₂. The O₂-carrying capacity of Hb, known as the Hüfner constant, is well-established at 1.39 ml of O₂ per gram of Hb, even though blood can also carry additional small amounts of O₂ through plasma or methemoglobin (i.e. Hb with oxidized heme groups).^[40] In contrast, the CO₂-carrying capacity of Hb varies from arterial to venous blood.^[40] It occurs through the reaction of CO₂ with the available amino groups of each globin to form carbamates; however, it only accounts for 6% of the total CO₂ transport.^[40] Instead, the majority of CO₂ is transported under the form of bicarbonate ions (HCO₃⁻) converted by the catalytic action of carbonic anhydrases. Upon arrival to environments with low partial pressure of CO₂, such as the lungs, the chemical reaction is reverted and CO₂ is released, not only from its HCO₃⁻ form, but also from the carbamate form. This effect facilitates the binding of O₂ to Hb, in the well-known process of pulmonary gas exchange.^{[39, 41]} Hb also mediates the metabolism of NO in a complex, multi-step interaction. In short, it is believed that deoxygenated Hb functions as an allosterically controlled nitrite reductase, converting nitrite into NO, which will serve as a vasodilation factor.^[42–44] Interestingly, NO is also thought to play a role in erythrocyte deformability and cell response to mechanical stress.^{[15, 45–47]} Finally, the ferrous heme needs to be maintained in its reduced state to guarantee an efficient O₂ transport capacity.^[39] With the high O₂ levels, the autoxidation of Hb, and the reactive oxygen species (ROS) resulting from cellular activity, erythrocytes need potent antioxidant machinery to achieve this. This antioxidative toolbox includes catalase, glutathione peroxidase (GPx), superoxide dismutase (SOD), peroxiredoxin 2 (Prx2), or cytochrome b5 reductase, responsible for neutralizing ROS or reverting ferric heme to its ferrous state.^{[39, 48, 49]}

Undoubtedly, the primary objective of next-generation erythrocyte engineering is the reproduction of cell form and function in a mass production-oriented manner. The aim is to efficiently bridge the gap between blood supply and demand regardless of urgency, worldwide location, and amount of blood needed. To achieve this, engineered erythrocytes (EEs) should be designed in line with manufacturing-related aspects while complying with the technical requirements which ensure their functionality and safety (Figure 2). Manufacturing-oriented aspects include the potential for production scalability, batch reproducibility, long shelf-life, and independence from blood, as it limits scalability and is currently one of the main challenges in the field. Technical requirements, on the other hand, are elaborated as follows:

Figure 2.

Manufacture-oriented and technical requirements for erythrocyte engineering.

1. The diameter of an EE should not overall exceed 8 μm, and it specifically should not exceed the diameter of the smallest capillary in the human body (5 μm) if its stiffness is higher than 26 kPa;

2. The shape of an EE should preferably be biconcave to facilitate blood flow and O₂ release; however, spherical and blunt-edged constructs may also be employed, as long as the former parameters are not affected;

3. The deformability of an EE should allow the undisturbed flow through the smallest capillaries and ensure a sufficiently long circulation time for application; these conditions can usually be met with constructs of E between 1 and 26 kPa;

4. O₂ transport should match that of the natural O₂-carrying capacity of 20 ml of O₂ per 100 ml of blood^[50], or remains below a hemotoxic level;

5. CO₂ transport should be such that it prevents adverse effects of CO₂ accumulation;

6. NO regulation, most relevant for application in ischemic treatments, should be such that there is a minimum beneficial effect under hypoxia (e.g., control of vasodilation);

7. Antioxidative machinery should be employed with elements that require redox regulation, either to maintain function (e.g., Hb) or blood and tissue viability (e.g., peroxides);

8. The rheology of engineered blood fluids, linked to the mechanical properties of EEs, should allow the efficient exchange of gas and nutrients in microcirculation; this condition is typically met with fluid viscosity similar to blood (3.5–5.5 mPa s).^[51]

In the following Sections, we summarize some of the most recent approaches in the field of erythrocyte engineering, with a review of fabrication methods, technology, and significance in light of technical and manufacture-oriented requirements.

3. Engineered Erythrocytes

3.1. Bio-engineered and stem cell-derived erythrocytes

Stem cell differentiation is the main process of expansion of natural erythroid cells in vitro. Human erythropoiesis, the natural production process of erythrocytes, is a well-understood cellular lineage commitment process that firstly occurs in the fetal liver and is maintained through life with the differentiation of hematopoietic stem cells (HSCs) into an erythroid lineage.^[52–54] Found in the bone marrow, umbilical cord blood or, more rarely, in adult peripheral blood, HSCs can differentiate into megakaryocyte-erythrocyte progenitors, which in turn can differentiate into burst-forming unit erythroids and later colony-forming unit erythroids. These cells terminally differentiate into erythroblasts as a response to erythropoietin, a cytokine released in hypoxic environments, which in turn differentiate into reticulocytes by loss of cellular nucleus. Lastly, these cells mature into the final erythrocytes, in a process that lasts a minimum of 18 days in vitro since the first HSC culture (Figure 3A).^[52–56]

Figure 3.

A. Schematics of stem cell differentiation into erythrocytes. Simplified based on refs.^{[52, 56]} B. Differentiation (day 4–18) and filtration of CD34⁺ cell-derived reticulocytes (pre-stage erythrocytes) from an adult cell line and an engineered cell line (BEL-A). Cells were stained with May–Grunwald Giemsa reagent and analyzed by light microscopy. Scale bars 10 μm. Reproduced under terms of the CC-BY license.^[61] Copyright © 2017, Trakarnsanga et al., published by Springer Nature Ltd.

The rare occurrence of HSCs in adult blood, coupled with the long differentiation process, limits the production of erythrocytes for blood substitution. The isolation of erythroid progenitors from pathological cell lines, such as the ones from myelogenous leukemia, may result in a higher yield, although with phenotypes unfit for a safe transfusion. Embryonic stem cells (ESCs) have also been successfully differentiated into erythrocytes, with enucleation rates of approximately 50%^[57], although with embryonic hemoglobin phenotypes instead of adult.^[58] Similarly, induced pluripotent stem cells (iPSCs) were successfully differentiated into erythroid cell lines, but several challenges in terms of full maturation of iPSC-derived cells into enucleated erythrocytes.^{[59, 60]} Recently, adult erythroid lines were immortalized from adult bone marrow CD34⁺ cells, with the ability to enucleate into reticulocytes resembling the ones isolated from regular adult cell-lines (Figure 3B).^{[61, 62]} A landmark clinical trial – the first involving lab-grown erythrocytes – is currently ongoing and aims to evaluate the survival of these cells in humans.^[63] While this constitutes a breakthrough in the field of erythrocyte engineering and blood disease treatment, challenges persist regarding the scalability and batch-to-batch reproducibility of cell-derived blood substitutes.

3.2. Hybrid engineered erythrocytes

To overcome some of the manufacturing issues of natural EEs, while retaining natural function, hybrid approaches for erythrocyte engineering emerged. Hybrid EEs can be defined as semi-synthetic constructs with at least one blood cell-derived element, having provided some of the first blood substitutes in history. Starting in the early 20^th century with the use of serum and gelatin solutions,^{[8, 9]} the field rapidly evolved into hemoglobin-based oxygen carriers (HBOCs), erythrocyte “ghost” membrane-coated particles, or a combination of both (Figure 4A), along with the less common, but equally promising serum-derived albumin constructs.

Figure 4.

A. Summary of hybrid engineered erythrocyte (EE) approaches. B. Combination of polymerized human cord hemoglobin (polyCHb) with ascorbic acid (AA). C. Lipid and DNA peroxidation in the kidneys of guinea pigs during pre-treatment (sham) and post-exchange transfusion with polyCHb (ET) and polyCHB+AA (ET+AA). Malondialdehyde (MDA) indicates lipid peroxidation; 8-hydroxy 2 deoxyguanosine (8-OHdG) indicates DNA peroxidation. Reproduced with permission from ref.^[82]. Copyright © 2023, Kong, Zhou, He, Zhang, Li, Zhong and Liu. D. Hemoglobin encapsulated in a lipid vesicle (HbV). E. Oxygenation in a pre-eclampsia Rosa26:Luc rat model through bioluminescence in vivo. Effect of saline (left) or HbV (right) injections in fetal tissue oxygenation. Reprinted (adapted) with permission from ref.^[87]. Copyright © 2015, Springer Nature Ltd. F. Schematic of the production of a synthetic Ca(OH)₂ biconcave template, later coated in Hb and an erythrocyte “ghost” membrane and treated to remove the inorganic base. Based on ref.^[97] G., H. Cell viability and hemolysis results of the different intermediate products of red blood cell (RBC) micromotors. Reprinted (adapted) with permission from ref.^[97]. Copyright © 2019, American Chemical Society. I, J. Schematic of the production of rebuilt red blood cells (RRBCs) based on silicified erythrocytes and layer-by-layer deposition of polymers and an erythrocyte “ghost”. Circulation of RRBCs through 5 μm microchannels as human blood capillary mimics. Scale bar equals 5 μm. Reprinted (adapted) with permission from ref.^[98]. Copyright © 2020, American Chemical Society.

3.2.1. Hemoglobin-based approaches

Hemoglobin solutions were initially developed as volume-restoring, oxygen-delivering alternatives to blood. Patients transfused with these solutions, however, displayed severe side effects such as low blood pressure, bradycardia, and renal function damage.^{[64, 65]} Further investigation pointed to stromal proteins as a potential cause of such adverse effects, prompting the addition of a purification step in hemoglobin extraction. Despite the use of stroma-free hemoglobin, patients still displayed symptoms of nephrotoxicity^[66], an effect later attributed to the dissociation of the hemoglobin tetramer into two hemoglobin dimers.^[67] To maintain the integrity and function of hemoglobin, polymerization and intramolecular crosslinking strategies were investigated, followed by strategies of functionalization of hemoglobin with polyethylene glycol (PEG).^[68] Several HBOC products along these lines of design reached clinical trial stages (e.g. HemAssist^®, PolyHeme, Hemospan^®), but all investigations were interrupted or refused for approval from the United States Food and Drug Administration (FDA) due to an increased risk of mortality and cardiovascular complications.^[69] To date, only two chemically modified HBOCs have been approved: one for human use in South Africa and Russia (Hemopure^®), and another one for veterinary use in the European Union and the United States (Oxyglobin^®).^[10] Both Hemopure^® and Oxyglobin^® consist of polymerized bovine Hb with a size of approximately 250 kDa, or four Hb molecules.^[10] HEMO₂life^®, a solution of purified and gamma-irradiated extracellular Hb isolated from marine lugworm Arenicola marina, is currently also approved for organ preservation in the European Union.^{[70, 71]} Other HBOCs at clinical trial stage are described in more detail in other reviews.^{[10, 72]}

Concurrently, other solutions for Hb protection were investigated. In 1957, Thomas Chang incorporated erythrocyte lysates in polymeric microparticles^[73], in a first approach towards the development of increasingly complex HBOCs.^[74–77] In 1980, Djordjevich and Miller prepared one of the first vesicle-encapsulated Hb structures using phospholipids and cholesterol, thereby introducing the first synthetic erythrocyte membrane.^[78] Vesicle compositions were then fine-tuned over the years, in parallel with advances in pharmaceutical research. The use of PEG-functionalized lipids, for example, contributed to a tremendous improvement of nanomedicine bioavailability, as PEGylated liposomes could more easily evade the reticuloendothelial system and circulate for longer in the bloodstream.^{[79, 80]} Over the last two decades, however, concerns have emerged regarding the safety and functionality of PEGylated liposome injections^[81] – an issue that is addressed in greater detail later in this review.

Recent efforts on HBOC development have focused on further improving the stability of Hb by optimizing polymerization and encapsulation methods. One strategy involved isolating Hb from human cord blood, removing stromal proteins through hypotonic hemolysis, cross-linking it with glutaraldehyde to a molecular weight below 100 kDa, membrane-filtering, and redispersing it in phosphate buffered saline (PBS) at 8.0 g L⁻¹.^[82] A further addition of ascorbic acid produced an injectable oxygenating antioxidant solution (Figure 4B), capable of alleviating oxidative stress in the kidney of guinea pigs and reducing the levels of lipid and DNA peroxidation (Figure 4C).^[82] Another crosslinking strategy consisted in using α-succinimidyl-ω-maleimide and dithiothreitol to form a Hb nanoparticle, later coated with albumin and loaded with catalase and SOD.^[83] Interestingly, catalase demonstrated good stability in the matrix over time, unlike SOD, a smaller enzyme, which was released prematurely from the construct. Nevertheless, the complex nanoparticles showed prolonged antioxidant activity from catalase, good O₂ binding affinity, and a satisfactory hemocompatibility profile in vitro.^[83] Lastly, it was proposed that the adverse cardiovascular side effects of crosslinked Hb transfusions were due to its low molecular weight^{[69, 84]}, so human Hb was further crosslinked to reach molecular weights of over 500 kDa.^[85] In comparison with stored blood, highly crosslinked Hb administered in the resuscitation of guinea pigs showed a lower risk of liver and cardiac damage, and lower activation of the sympathetic nervous system.^[85] Higher molecular weight Hb proved, therefore, a promising strategy to enhance the safety of hemoglobin products.

Hemoglobin vesicles (Hb-V), or liposome-encapsulated Hb (Figure 4D), represent another simple and elegant approach to achieve stable Hb transport. One strategy for the production of Hb-Vs consists in first isolating and purifying human Hb, stabilizing it with pyridoxal 5-phosphate, and encapsulating it in 0.25 μm liposomes through lipid film rehydration.^[86–89] In all studies, liposomes were composed of 1,2-dipalmitoyl-sn-glycero-3-phosphatidylcholine (DPPC), cholesterol, different types of succinyl-L-glutamate lipids, and a PEGylated lipid (1,2-distearoyl-sn-glycero-3-phosphatidylethanolamine-N-PEG5000, DSPE-PEG₅₀₀₀) at approximate molar ratios of 5:4:1:0.3.^{[86, 87, 89]} All Hb-Vs were resuspended in saline solution to a concentration of 10 g dL⁻¹,^[87] demonstrating a marked effect in the oxygenation and resuscitation of pre-eclampsia rat models (Figure 4E) and in massive obstetric hemorrhage in rabbit models.^{[87, 90]} Phase I clinical trials of this product showed that Hb-V administration can cause light adverse effects (e.g. fever), but biomarker analysis suggested little to no risk of myocardial infarction.^[89] Additionally, the circulation half-time of Hb-Vs was determined at 8 hours, a sufficiently long period for resuscitation in emergency situations.^[89]

Hemoglobin matrices made of polymeric and inorganic materials have also been used as oxygen carriers. In one study, pectin-hemoglobin microcapsules were produced through an electrospraying method optimized for particle sizes of 5–10 μm and biconcave shape, approaching the natural form of erythrocytes.^[91] In another study, anisotropic mesoporous silica nanoparticles of average length and width 90 nm and 43 nm, respectively, were loaded with bovine Hb and lipid-coated through vesicle fusion.^[92] Lipid vesicles were made from 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), and a PEGylated lipid (DOPE-PEG₂₀₀₀) in a molar ratio of approximately 10:5:1.^[93] These particles showed efficient Hb encapsulation, maintained Hb activity after encapsulation, and little endothelium adhesion after injection in a zebra fish model.^[93] Moreover, the lipid coating improved colloidal stability and limited the release of hemoglobin from the construct.^[93] Another study focused on the development of a metal-organic framework (zeolitic imidazole framework-8, or ZIF-8) loaded with bovine Hb for the treatment of hemorrhagic shock in a mouse model.^[94] Animals showed satisfactory survival rates two hours after injection, proving the utility of metal-organic frameworks (MOFs) as another type of hemoglobin carrier.^[94]

3.2.2. Erythrocyte lipid membrane-based approaches

Lysed erythrocyte lipid membranes, or “ghosts” membranes, are also used in the development of EEs. The stealth of the membrane, attributed to the presence of CD47, is believed to provide an excellent option as a coating for injectable systems for long-term circulation. To prove this, perfluorocarbons (PFCs) were encapsulated into poly(d,l-lactide-co-glycolide) (PLGA) nanoparticles of sizes 0.27–0.29 μm through a double emulsion solvent evaporation method, and subsequently coated with an erythrocyte lipid membrane through incubation and stirring at 4°C.^[95] Of note, the “ghost” coating was expected to present the correct inside-out orientation after surface deposition, presumably due to the higher electrostatic affinity of PLGA towards the lipids of the inner leaflet compared to the ones in the outer leaflet of the bilayer membrane.^[95] The final constructs exhibited a prolonged circulation time, hypoxia alleviation eight hours post-injection, and enhanced tumor reduction after radiotherapeutic treatment.^[95] Similarly, another strategy involved the application of an erythrocyte lipid membrane to PFC nanoemulsions and further using them as coatings for Au nanowires, with the purpose of creating complex nanomotors for intracellular O₂ delivery.^[96] The constructs showed significant hypoxia alleviation in mouse macrophages after ultrasound-induced cell uptake and a maintained cell viability, attributed to the use of the natural cell coating.^[96] More recently, researchers produced erythrocyte mimics at distinct Young’s moduli (E = <5 MPa, 5–10 MPa, and 25–30 MPa) by either creating i) erythrocyte “ghost” vesicles as soft mimics; ii) “ghost”-coated PEGDA nanogels of <150 nm with 10% crosslinking as elastic mimics; or iii) the same “ghost”-coated nanogels at 40% crosslinking as hard mimics.^[32] Elastic mimics showed the most prolonged circulation half-time, with overall lower protein adsorption, lower uptake at the reticuloendothelial system, and better tumor penetration than softer or harder particles, proving its potential as tumor targeting hybrid vehicles.^[32]

3.2.3. Combined hemoglobin-erythrocyte lipid membrane approaches

Combining HBOCs with erythrocyte “ghosts” is another relatively common approach in the production of blood substitutes. As an example, erythrocyte-mimicking magnetic micromotors were produced through a templating technique and used in active photodynamic therapy.^[97] A biconcave-shaped Ca(OH)₂ template was produced and loaded with Hb, Fe₃O₄ nanoparticles, and a photosensitizer. The template was subsequently coated with more Hb to form the micromotor, coated with an erythrocyte “ghost” membrane, and etched to remove the Ca(OH)₂ template (Figure 4F).^[97] The final particles showed low hemolysis rates, maintained cell viability (Figure 4G,H), and an effective photodynamic therapy effect.^[97] Another interesting strategy consisted in coating an inorganic erythrocyte template with natural polymers, Hb, and a “ghost” membrane to produce rebuilt red blood cells (RRBCs).^[98] Firstly, blood was silicified through the protein-catalyzed condensation of silicic acid at the surface of all cellular features, resulting in silica templates of all blood cells. Alginate and chitosan were then deposited on silicified erythrocytes using a layer-by-layer technique and the template was subsequently dissolved through etching. The hollow structures were then loaded with different cargo (e.g., Hb) within the polymer layers and finally coated with an erythrocyte “ghost” (Figure 4I).^[98] These complex RRBCs could successfully flow through 5 μm microchannels and retain biconcave and discoidal shape (Figure 4J), displaying a similar circulatory trend in blood capillaries of chicken embryos.^[98] The RRBCs demonstrated low hemolysis, excellent in vitro and in vivo compatibility, good Hb loading compared to natural cells (8.4 vs. 22–37 μg Hb per million cells), and a relatively long circulation half-time of 41.8 hours – a result mostly attributed to the stealthy natural cell coating.^[98] Finally, the enrichment of Hb-“ghost” complexes with a synthetic antioxidant was proposed. Bovine Hb-laden PLGA nanoparticles were produced and subsequently coated with layers of poly-L-lysine (PLL), CeO₂ nanoparticles, and a “ghost” membrane.^[99] The CeO₂ nanoparticles acted as a synthetic alternative to catalase and SOD as the antioxidant agent, showing satisfactory scavenging of superoxide anion radicals.^[99] Moreover, the cell membrane coating prevented particle uptake by endothelial cells, endowing the construct with erythrocyte-like stealth properties.^[99]

Despite the promising technologies used in the fabrication of Hb-“ghost” complexes, some aspects should be taken into consideration. The effect of the multiple construct layers (e.g., polymers, coatings, “ghosts”) on O₂ release remains largely unaddressed, even though it could limit the effectiveness of Hb in tissue oxygenation; and the inclusion of functional nanoparticles in the constructs warrants additional tests on hemocompatibility and cytotoxicity, as particle release could trigger multiple adverse responses.

3.2.4. Albumin-based approaches

Albumin, the most abundant protein in blood, is typically isolated from serum and has been employed as a stabilizer for a number of synthetic oxygen carriers. As a recent example, albumin-coated perfluorodecalin microspheres of sizes 0.1–3 μm were fabricated through an ultrasonic emulsification method.^[100] Particles showed improved O₂ affinity when compared to commercial PFC products, and displayed a reduction of tissue damage caused by hypoxia upon injection in rats with massive hemodilution.^{[100, 101]} Additionally, albumin was investigated as a perfluorocarbon stabilizer in nanoemulsions produced via high-pressure homogenization. After an iterative adaptation of the production method, both bovine and human serum albumin-stabilized PFC emulsions showed oxygenation rates similar to oxygen carriers that had previously reached clinical trial stages.^[102] These emulsions demonstrated satisfactory stability up to 4 months – more than twice longer than donated red cell concentrates.^[102]

To date, hybrid EEs constitute one of the most successful approaches to blood substitution; however, there are several adverse aspects worth considering. Firstly, the use of human Hb and erythrocyte “ghost” membranes contradicts the final objective of blood substitution, as it uses a scarce raw material to build its own substitute. Secondly, the use of animal Hb would raise ethical concerns in large-scale manufacture, as large amounts of animal blood would need to be withdrawn. Finally, while protein sterilization methods are quite effective and the risk of infection is low^[103], the possibility of yet unknown transfusion-transmitted diseases is still a concern. This issue is rooted in the public perception of the HIV/AIDS epidemic and the blood transmissibility of the neurodegenerative Creutzfeld-Jakob disease.^{[10, 104, 105]} With this in mind, the exploration of blood-independent approaches has become progressively more interesting in the field of blood substitution.

3.3. Synthetically engineered erythrocytes and synthetic blood

Synthetic EEs have become the most promising approach for on-demand manufacture of blood substitutes. These fully synthetic, material-based cells do not depend on blood products, are easily scalable, and can provide longer shelf lives than natural blood. Although scarce, distinct approaches have been adopted to establish fully synthetic erythrocyte substitutes, typically involving perfluorocarbons or recombinant hemoglobin. To complement these works, erythrocyte-resembling microgels and oxygen-releasing microparticles have been extensively fabricated and, although not originally designed for blood substitution, could be easily adapted into synthetic EEs (Figure 5A).

Figure 5.

A. Summary of synthetic engineered erythrocyte (EE) approaches. B. Method of production of perfluorocarbon-poly(lactide-co-caprolactone) (PFC-PLC) biconcave particle through a Shirasu porous glass (SPG) membrane emulsification method, using dichloromethane (DCM) as a solvent and isopropanol (IPA) as a shape changer. Reprinted (adapted) with permission from ref^[115]. Copyright © 2021 Wiley-VCH GmbH. C. Hypoxia relief on EGFP-modified HeLa cells upon addition of 45 vol% concave shape-deformable perfluorocarbons (cDFCs). EGFP fluorescence indicates a hypoxic state. Scale bars equal 50 μm. Reprinted (adapted) with permission from ref^[115]. Copyright © 2021 Wiley-VCH GmbH D. Recombinant hemoglobin rHbA(X) combined with three molecules of recombinant human serum albumin (rHSA) to form a rHbA(X)-rHSA₃ tetramer. E. Oxygen saturation of red blood cells (RBCs) and different modifications of rHbA(X)-rHSA_3. Reprinted (adapted) with permission from ref.^[120] Copyright © 2020, Royal Society of Chemistry. F. Schematics of production of hyaluronic acid (HA) spheres and capsules from the deposition of HA on solid and mesoporous silica templates. After depositions, the erythrocyte-sized templates (7.0 μm) were removed. Based on ref.^[123] G. Flow of HA capsules (left) and spheres (bottom right) in a microchannel environment. Scale bars equal 10 μm. Reprinted (adapted) with permission from ref.^[123] Copyright © 2015, American Chemical Society. H. Schematics of assembly of hydrophobic oxygen-generating microparticles (HOGs) based on solid peroxides. I. Hydrogen peroxide release from hydrogel-laden HOGs over 12 days. Reprinted (adapted) with permission from ref.^[132] Copyright © 2021 The Authors. Advanced Functional Materials published by Wiley-VCH GmbH. J. Hydrogen peroxide release over 7 days from HOGs coated with catalase (catalase outside the particle, COTP) and embedded within a hydrogel. Dashed line indicates cytotoxic threshold. Reprinted (adapted) with permission from ref.^[137] Copyright © 2022 Wiley-VCH GmbH.

3.3.1. Perfluorocarbons

Perfluorocarbons (PFCs) emerged as the first entirely synthetic blood substitutes. These liquid fluorinated hydrocarbons are chemically stable, metabolically inert, and more effective than water or blood plasma in dissolving and absorbing oxygen in the lungs. PFCs presented, therefore, a remarkable potential as an alternative to hemoglobin.^[12] In 1966, Leland and Clark proved that mice could survive for hours submerged in oxygen-enriched PFCs^[106], opening a new avenue in the development of oxygenation products. Fluosol-DA^®, a mixture of perfluorodecalin and perfluorotripropylamine, was one of the first PFCs employed in a clinical setting, having obtained FDA approval for use in high-risk angioplasty.^[107] In 1980, facing a delay in blood supply, a patient at high risk of hemorrhagic shock was administered Fluosol-DA^®, promptly showing signs of improvement in blood pressure, pulse rate, and blood-gas levels.^[108] Over time, however, other solutions for angioplasty rendered Fluosol-DA^® obsolete.^[107] Others studies within the so-called first-generation of PFCs followed, although with limited success. Synthetic stabilizers (e.g. Pluronic^® F-68) were thought to trigger flu-like symptoms and complement activation, and PFCs required freezing storage conditions, which proved impractical for long-distance shipment.^{[109, 110]} Second-generation PFCs emerged, such as Oxygent^®, OxyFluor^®, and Oxycyte^®, using natural stabilizers (e.g. egg-derived phospholipids) and presenting longer shelf-lives at room temperature.^[109] However, most second-generation PFC trials were still suspended due to the unchanged reports on side effects (e.g. fever, nausea) and heightened risk of stroke.^{[12, 107, 109, 111–113]}

Following Fluosol-DA^®, only two other PFCs obtained approval for clinical use. Perftoran^®, a mixture of perfluorodecaline and perfluoro-N-4-(methylcyclohexyl)-piperidine, was approved for clinical use in Russia and also temporarily approved in other countries (Ukraine, Kazakhstan, and Mexico).^[113] Perflubron^®, a perfluorooctyl bromide, was approved by the FDA for use in magnetic resonance imaging and computed tomography as a contrast agent.^[107] More notably, it was recently granted orphan designation for use in respiratory stress disease by the European Medicines Agency (EMA).^[114]

Nowadays, advances in PFCs are dedicated to minimizing side effects through the improvement of fabrication techniques, stabilization methods, and other innovative strategies. To improve the biological response of PFCs, perfluorooctyl bromide particles were fabricated and encapsulated in a poly(lactide-co-caprolactone) shell through a Shirasu porous glass (SPG) membrane emulsification process.^[115] Particles were then treated with isopropanol to transform into a biconcave shape (Figure 5B).^[115] This material was shown to alleviate hypoxia in a tumor cell line (Figure 5C), and could successfully deform through 4.5 μm gaps despite having a higher Young’s modulus than natural cells (93 kPa vs 1–26 kPa).^[115] Despite the great interest in new-generation PFCs, concerns remain over the environmental costs of fluorinated compounds and their large-scale production, prompting the discovery and exploration of more environmentally-friendly blood substitutes and methods of production.

3.3.2. Recombinant hemoglobin

Recombinant DNA technology has provided a substantial contribution to the scalable production of blood proteins, especially Hb and serum albumin, for use in blood substitutes. The first reports on recombinant Hb (rHb) date back to 1984 and 1990, with the expression of human β-globin and complete human Hb, respectively, in Escherichia coli.^{[116, 117]} Recombinant DNA technology offered a promising way to circumvent issues related to natural Hb, such as tetramer dissociation, in a single step through the direct modification of encoded gene sequences to express Hb mutants. Multiple strategies have since been described on how to engineer rHb for enhanced O₂ affinity and release, reduced NO scavenging as a competitor of O₂, reduced autoxidation, and longer shelf-life.^[118] Enhanced variants of rHb have been produced under good manufacturing practice (GMP) conditions at large industrial scale, and some entered Phase II clinical trials, although all investigations were eventually ceased due to unwanted NO scavenging effects or complement activation issues.^[119]

Recently, different variants of rHb produced in yeast species Pichia pastoris were covalently wrapped with three recombinant human serum albumins (rHSA) (Figure 5D).^[120] One of the modified variants of rHb combined with rHSA showed similar O₂ saturation properties as native erythrocytes, showing good potential as a synthetic substitute (Figure 5E).^[120] Other variants of rHb-albumin are being investigated by the group, with the aim of improving O₂ transport efficiency between the lungs and peripheral tissues. However, due to the use of natural human serum albumin instead of rHSA, these second constructs do not classify as fully synthetic EEs.^[120] Further studies should focus on the safety profile of these particles in vitro and in vivo for intravenous injection.

The use of rHb as a blood substitute is not free from drawbacks. The participation of Hb in numerous cellular processes, such as NO scavenging, has affected the maximum O₂ transport levels of rHb products, thereby limiting their effectiveness. This aspect has warranted considerable modification efforts to reach a satisfactory O₂ response.^[119–121] Moreover, the cost of production and purification of recombinant proteins remains high, with particular concerns for the elimination of lipopolysaccharides, a highly inflammatory component of the membrane of E.coli and other host systems.^[118]

3.3.3. Erythrocyte-resembling microgels

Considerable efforts have been dedicated to closely mimicking the size, shape, and deformability of erythrocytes, as these properties are linked to safer injectability, better evasion of phagocytosis, and longer circulation time. Although not originally intended for blood substitution, their adaptation to this purpose can be achieved.

To study the biodistribution of erythrocyte-resembling microgel (ERM) biodistribution, erythrocyte mimics with different deformability profiles were first investigated. Specifically, microgels with 6 μm in diameter were produced with varying degrees of stiffness by cross-linking 2-hydroxyethyl acrylate (HEA) with PEG-diacrylate (PEGDA).^[122] It was seen that softer microgels (E = 7.8 kPa) circulated longer than stiffer microgels (E = 63.9 kPa) after intravenous injection in mice, with a >30-fold increase in circulation half-time.^[122] Notably, stiffer microgels were entrapped in the thin pulmonary capillary beds, causing significant distress to the animals and calling forth their premature sacrifice.^[122] It was therefore concluded that microgel deformability dictated the safety of intravenous injection, and that soft microgels (E < 8 kPa) were considered safe-to-inject.

To assess the effect of microgel structure on particle behavior under flow, hyaluronic acid (HA) structures were fabricated through the combination of templating and continuous assembly of polymer techniques.^[123] In short, methacrylate-modified HA was deposited on silica or mesoporous silica particles of 7.0 μm in diameter and later polymerized through atom transfer radical polymerization (ATRP).^[123] The silica templates were then removed to reveal HA spheres or HA capsules, respectively (Figure 5F). Despite having identical stiffness at the nanoscale, HA capsules flowed more easily through blood capillary-mimicking microchannels than HA spheres, indicating that the structure of a particle influences its deformability (Figure 5G).^[123]

In conclusion, the few existing studies on the injectability and flow safety of ERMs showed that some micromaterials, unlike previously believed, can be injected intravenously without blocking capillary blood flow. This observation opens new doors for the fields of blood substitution and drug delivery systems alike.

3.3.4. Oxygen-releasing microparticles

Oxygen-releasing microparticles (ORMs) are generally composed of two distinct categories: oxygen carriers and oxygen generators. Oxygen carriers of fully synthetic origin consist of PFCs, O₂ host molecules (e.g., cyclodextrin, rHb), or lipid-stabilized O₂ microbubbles, and represent the most common types of synthetic ORMs. Having reviewed PFCs and rHb, cyclodextrins and microbubbles are discussed in this section.

Cyclodextrins are a family of macrocyclic oligosaccharides prepared by the enzymatic treatment of starch, with a conical shape able to host various types of guest molecules in aqueous media.^[124] In this line, cyclodextrins were tested as synthetic O₂ carriers. Recently, a guest-host inclusion complex using tetra-PEGylated tetraphenyl porphyrin and a per-O-methylated cyclodextrin dimer was formed through a molecular threading process.^[125] The resulting complex, named HemoCD, could reversibly bind O₂ in aqueous solution while showing low toxicity and satisfactory clearance in mice six hours after injection, thus presenting a promising alternative to Hb.^{[125, 126]} Current efforts consist in improving the synthetic parameters of different HemoCDs and tuning their affinity towards O₂ versus carbon monoxide (CO).^{[125, 127]} More recently, a similar type of O₂-saturated cyclodextrin polymers were produced and tested for potential use in myocardial infarction treatment.^[128] These constructs demonstrated low hemolysis, sustained O₂ release for 48 hours, and maintained cardiomyocyte viability under hypoxia-reoxygenation conditions, suggesting a therapeutic effect in ischemia/reperfusion settings.^[128]

Lipid-stabilized O₂ microbubbles represent another interesting approach to O₂ transport due to the general safety of lipid-based products as pharmaceuticals. In a study, pH-responsive oxygen nanobubbles were created to selectively relieve tumor hypoxia during radiotherapy and photodynamic therapy.^[129] Lipid-stabilized acetal-dextran carriers were made from the emulsification of acetal-dextran in a lipid solution, forming nanobubbles which were then saturated with O₂.^[129] In vitro and in vivo studies on these nanobubbles demonstrated maintained cell viability, as well as high tumor oxygenation in a nasopharyngeal carcinoma model.^[129] Compared to other synthetic EEs, however, O₂ microbubbles have displayed much shorter circulation times^[130] and an adverse hemodynamic response in vivo.^[131] A more thorough description of these and other erythrocyte-inspired O₂ carriers, as well as their fabrication methods and applications, can be found in a recent review.^[11]

Oxygen generators are alternative solutions based on peroxides^[132], algae^[133], or manganese dioxide (MnO₂) nanomaterials.^[134–136] To date, however, limited efforts have been conducted on O₂ generators which resemble erythrocytes. Photosynthetic algae can only produce O₂ upon contact with light, a fact that limits the use of algae as synthetic EEs due to the vast majority of blood being located relatively deep in the body. Nanomaterials based on MnO₂, or so-called nanozymes, are a catalyst for the conversion of endogenous H₂O₂ into O₂. They have been shown to alleviate tissue hypoxia and have acted as successful phototherapy mediators^[134–136], but concerns remain over their cytotoxicity - an issue probably attributed to their small size and adverse surface chemistry. Hydrogen peroxide (H₂O₂) rapidly decomposes to water and O₂ in biological media, but its burst release process might be cytotoxic at high concentrations. Moreover, controlling the release of O₂ over several weeks has proven difficult to achieve, but not impossible. Solid peroxides (e.g., CaO₂) can be encapsulated in hydrophobic biomaterials and decompose in a more time-controlled manner, providing a good solution for the sustained release of O₂. In two studies, CaO₂ crystals of sub-micron sizes were encapsulated in polycaprolactone (PCL) microparticles through a double emulsion solvent evaporation technique (Figure 5H).^{[132, 137]} The resulting constructs, of 2–8 μm in diameter, were subsequently incorporated in cell-laden hydrogels to build self-oxygenating tissues.^{[132, 137]} It was hypothesized that the slow diffusion of water to the inner hydrophobic matrix of PCL would limit the degradation of CaO₂ into H₂O₂, and sequentially into O₂ by action of catalase (Equation 1, Equation 2), thereby promoting a controlled, non-cytotoxic oxygenation process.

$$Ca{O}_2+2H_{2}O\to Ca{\left(OH\right)}_2+H_2{O}_2$$ (1)

$$2H_2{O}_2\to O_2+2H_{2}O\left(catalase\right)$$ (2)

Indeed, it was observed that PCL-encapsulated CaO₂ exhibited a more prolonged release of O₂ and H₂O₂ (Figure 5I) and a higher cell viability in anoxic condition over a period of 12 days.^[132] More interestingly, the addition of catalase to the outer face of the PCL construct maintained H₂O₂ levels below the cytotoxic threshold^[137] (Figure 5J), suggesting an added benefit in the incorporation of antioxidant elements in synthetic EEs. To the best of our knowledge, these two studies report on the first polymeric O₂ generators with similar features to natural erythrocytes.

Other studies on solid peroxide-loaded microparticles have been published, with promising results on cytocompatibility and hypoxia relief over time.^{[138, 139]} Recently, CaO₂ microparticles were also coated with hydroxyapatite as an alternative way of controlling O₂ release, showing prolonged oxygenation from three to ten days when embedded in a hydrogel.^[140] Particle sizes, however, generally overpass the size range of natural erythrocytes. One of the difficulties of fabricating hydrophobic O₂-generating microparticles is the tunability of particle size to the range of 6–8 μm in a monodisperse suspension. Moreover, hydrophobic particles have been linked to an adverse in vivo response due to their tendency to flocculate in aqueous media and potential immunogenicity. We address these issues in greater detail later in the review.

With the growing concerns over blood shortages and the use of natural blood, synthetic EEs have emerged as one of the most promising blood substitutes. Compared to natural or hybrid alternatives, synthetic erythrocytes appear to be the only blood-independent, mass-producible, and long-lasting blood substitutes for use in times of high demand. However, to date, no artificial blood substitutes have been clinically approved for use in humans at a widespread international level. Substantial challenges persist in finding an ideal solution, but the continued use of advanced materials technologies in the creation of synthetic erythrocytes is expected to bring considerable advances in the field.

4. Addressing Challenges in the Production of Engineered Erythrocytes

4.1. Recreating cell form

Size, shape, and deformability are key factors to consider when producing synthetic erythrocytes. Specifically, particles should be sufficiently small and deformable to flow safely through the thinnest blood capillaries (~ 5 μm diameter in humans), and should preferentially have a biconcave discoidal shape to facilitate deformation and gas exchange (Figure 2).^[17]

4.1.1. Size and shape

Recreating the size and shape of erythrocytes in a single-step process usually requires the use of template-based techniques. These encompass the fabrication of molds with erythrocyte form, after which a precursor material is deposited and cured into a final microparticle structure. Blood silicification, for example, is a high-throughput method of producing erythrocyte-like particles by rendering all blood elements inorganic.^[98] This approach, however, requires the use of donated blood, which severely limits its scalability. Particle replication in nonwetting templates (PRINT^®) is another templating technique based on a patterned elastomeric mold, generally loaded with a pre-polymer precursor which is subsequently photocured or crosslinked and collected.^{[122, 141]} This approach provides great tunability and batch-to-batch reproducibility, constituting a powerful method for EE fabrication. Micromechanical punching consists in creating a material film on a deformable substrate and punching the desired particle shape repeatedly.^[142] This method is scalable, automatable, and high-throughput, even though currently produced structures are two orders of magnitude (>100 μm) larger than native erythrocytes. Calcium hydroxide (Ca(OH)₂) templating consists in dissolving NaOH in a CaCl₂/dextran sulfate solution and subsequently using ultrasonication to obtain biconcave-shaped Ca(OH)₂ particles of different sizes.^{[97, 143]} These templates are then layered with specific materials and etched, forming erythrocyte-like capsules. This scalable method offers the possibility of tuned deformability, even though encapsulation is limited to etch-resistant cargo. Ca(OH)₂) templating, therefore, constitutes another promising way of fabricating synthetic EE templates with life-like size and shape.

Separating fabrication of size and shape in a two-step process is also a common approach in literature. Size recreation can be achieved through, for example, microfluidics or emulsification techniques. Microfluidics enables the generation of droplets with precise size control in a highly monodisperse and reproduceable manner. Within the field of blood substitution, however, the low throughput of many microfluidics systems hinders scalability, and the vast majority of techniques is still oriented towards the fabrication of large microparticles (>10 μm).^[144–148] To counter this trend, one study reported a parallelized microfluidics setup able to produce polycaprolactone (PCL) microparticles of 5–9 μm in a monodisperse way.^[149] This elegant high-throughput system incorporated 20160 flow focusing droplet generators onto a single silicon wafer, enabling the fabrication of up to a trillion erythrocyte-sized microparticles per hour.^[149] However, knowing that each unit of blood (450 mL) contains approximately 2.25 trillion erythrocytes, a minimum of two and a half hours would be needed to produce one single unit of synthetic blood. Moreover, due to the stiffness of these hydrophobic particles, 5 μm structures would still encounter difficulties flowing through small blood capillaries.

Emulsification techniques encompass the generation of droplets by mixing and stabilizing two immiscible phases. One strategy involved a common double emulsification protocol to produce oxygen-generating PCL microparticles: an aqueous solution of calcium peroxide and a dichloromethane solution of PCL were first emulsified using a tip ultrasonicator, after which they were emulsified once more in an aqueous solution of polyvinyl alcohol (PVA) of high hydrolysis rate.^[132] Despite being polydisperse, the resulting particle suspension had a size distribution of 2–8 μm, a range that more closely aligns with the specifications for EEs.

To obtain more monodisperse suspensions, membrane emulsification techniques provide one of the best outputs from a vast range of options. In a previously mentioned study, Shirasu Porous Glass (SPG) membrane emulsification was used to create synthetic EEs.^[115] This emulsification technique consists in pressurizing a dispersant phase onto a continuous immiscible phase, usually an oil or an organic solvent, through a glass membrane with specific pore sizes, creating stable uniform droplets.^[150] Essentially, it offers the possibility of fabricating monodisperse particle suspensions at safe-to-inject sizes using a range of materials (e.g. PFCs, polymers, lipids)^{[151, 152]} in a ultrahigh-throughput manner. Upscaling would be possible, depending on membrane dimensions, maximum pressure range, availability of raw materials (e.g., Shirasu volcanic ash), and other factors. Following the same principle, other membrane and microsieving systems^[153] could also be applied in the fabrication of erythrocyte-like particles in both small- and large-scale settings. Nevertheless, some drawbacks exist: first, the incompatibility with more complex liquids, such as granular or viscous solutions, which can damage the membrane when pressurized; second, the need for mass production of erythrocyte substitutes, at approximately 2.25 trillion particles per unit of blood; and third, the use of oils or organic solvents, which require a thorough removal process prior to in vivo administration.

To further imprint the erythrocyte form, shape changes can be induced in certain materials after particle generation. The most common techniques consist in solvent-induced fluidization, which does not readily solubilize the material but rather alters the conformation of its structure. Specifically, isopropanol was used to imprint a biconcave shape in poly(lactide-co-caprolactone) PFC emulsions^[115] and PLGA microparticles, while tetrahydrofuran was used to partially solubilize and deform polystyrene microbeads.^[154] Despite inhomogeneities, an erythrocyte-resembling form was observed in all structures.^{[115, 154]}

Lastly, a multi-step combination of techniques for size and shape replication has been reported. Particularly, PLGA microparticles of approximately 7 μm in size were first produced through electrohydrodynamic jetting and later used as templates for layer-by-layer deposition of hemoglobin/albumin (Hb/BSA) or polyallylamine hydrochloride/albumin (PAH/BSA).^[154] The polymer was subsequently solubilized to create hollow particles.^[154] Electrohydrodynamic jetting had been previously reported as a promising procedure for the fabrication of simple or complex particles at the nanoscale, sub-micron, and micron-sized (2–3 μm) ranges^[155] and, while monodispersity and scalability are limited, this versatility could prove useful in synthetic EE development.

4.1.2. Size range

Native erythrocytes have a diameter range of 6–8 μm, and the production of erythrocyte substitutes within that size ranges appears logical. However, erythrocyte function is not inherently connected to its size, and a considerable number of blood substitutes have been based on nano- and sub-micron-sized constructs (<200 nm). Nano-EE approaches are typically chosen instead of micro-EEs, as the intravascular injection of micromaterials still raises serious safety concerns, unlike nanomaterials. Specifically, nanomaterials are less likely to interrupt blood flow in case of aggregation or flocculation^[157]; can be more rapidly cleared from circulation without majorly affecting organs or tissues^{[157, 158]}; and, most importantly, the body of literature on nanoparticle safety is far more extensive than the one on microparticle safety.^[159–161] Moreover, for large-scale manufacturing, monodispersity and batch-to-batch reproducibility are important, and sub-micron particles (0.1–1 μm) are substantially easier to produce than erythrocyte-like microparticles (6–8 μm) in a scalable, monodisperse way. Specifically, sub-micron and nanoparticle synthesis is often performed in bulk, using reagents and stabilizers which form highly precise supramolecular structures as nano-reactors. This process generally results in scalable and monodisperse particle suspensions. This supramolecular precision, however, is lost when scaling the process to micrometer sized reactors. Non-bulk methods have aimed at improving the monodispersity of microparticle production (e.g., microfluidics-based droplet generation or template-based methods), but their scalability and throughput are usually low for the erythrocyte-like size range.

Despite this tendency, nanoparticles are not always safe to administer intravenously. Their high surface area-to-volume ratio amplifies any response derived from cell-material interaction, including adverse ones, such as cytotoxicity (e.g., Au, MnO₂)^{[162, 163]} or hemolysis (e.g. SiO₂).^[164] More interestingly, some studies show that microgels can flow safely through capillary-mimicking microchannels, as long as they fit within certain physical requirements of size and deformability.^[122] With the growing investment in in vitro and in silico technologies such as blood vessel-on-chip^[165–168] and microcirculation simulations^[169–171], microparticle safety profiles are expected to be traced at a faster pace. These advances will hopefully draw a clearer picture on microparticle safety in intravascular injection.

4.1.3. Deformability

Deformability has been identified as one of the key factors in the safe injection of synthetic EEs. Erythrocyte-like deformability can be primarily achieved through microgel volume, microgel structure, and choice of material, and is, by and large, described by the value of elastic modulus E. Firstly, a study using poly(N-isopropylacrylamide) microgels (pNIPAM) proposed that the E of a microgel was inversely proportional to its volume and swelling.^[172] In other words, larger and swollen microgels would be softer than smaller, collapsed ones. Secondly, microgel structure also influences deformability, as hollow capsules tend to flow more easily through narrow microchannels than solid spheres.^[123] Thirdly, the choice of materials is also paramount in the fabrication of ERMs. Depending on their chemical properties, soft micromaterials can be used as-is or crosslinked to different degrees to further tune their deformability, with the final goal for safe capillary flow of E < 100 kPa,^[115] and safe capillary flow and biodistribution at E < 26 kPa, the maximum value determined for natural erythrocytes (Figure 1).^[31] This limit was determined from the observation that microgels with E ≈ 67 kPa, although deformable, tended to accumulate in the lungs of mice after injection and cause pulmonary embolism.^[122]

Some examples of soft microgel materials include calcium-crosslinked alginate (E = 47 kPa, variable),^[145] polymer-PFC emulsions (E = 93 kPa)^[115], or swollen p(NIPAM) microgels (E = 13 kPa)^[172]. It is worth to note that hydrophobic polymers, despite well-known for their FDA approval for biomedical purposes, have a stiffness in the MPa to GPa range^[173], which renders them not suited for use as synthetic EEs. Nevertheless, the structural engineering of these microparticles (e.g., hollowing, higher porosity, thin layer deposition) offers a simple and effective solution to improve particle deformability and injectability safety. Furthermore, while E is a good representative of particle deformability, other parameters such as creep and stress relaxation (Figure 1) could provide additional insight for the validation of synthetic EEs.

4.2. Recreating blood rheology

Blood rheology, a key regulator of physiological function, is primarily attributed to the form and abundance of erythrocytes. Higher cell stiffness (10 kPa < E < 100 kPa) is thought to induce a more Newtonian-like blood flow,^{[174, 175]} consequently limiting the formation of a cell-free plasma layer and potentially affecting blood-tissue exchange. It is important to note, however, that erythrocytes distribute unevenly through capillaries after vessel bifurcation, leading to different hemodynamic effects, which may have implications in organ function.^{[176, 177]} For blood substitution purposes, maintaining a synthetic hematocrit closer to the normal range is, therefore, ideal. This implies the use of ultra-concentrated synthetic EE suspensions (2.25 × 10¹² particles per 450 mL), which in turn requires the use of highly hemocompatible soft microgels (E < 26 kPa) without any dose-dependent toxicity. Due to the large amounts of microgels needed, the possibility of large-scale manufacturing is also highly important. More research on soft microgels is needed to build synthetic “transfusion liquids” with similar rheological behavior as blood – especially at a microvascular scale.

4.3. Recreating the cell membrane

Cell membrane composition also plays a significant role in the stealth and function of erythrocytes. As the component with direct blood contact, the erythrocyte membrane is responsible for managing cell-blood interaction, including the indication of a healthy condition (i.e., continued circulation time) or a senescent state (i.e., warranting removal from bloodstream). This is achieved primarily through the “marker of self”, CD47, and the lipid composition of the outer leaflet. A number of advances have been made towards understanding the specific roles of each membrane lipid and choosing the best synthetic analogs for cell engineering. Erythrocyte membrane “ghosts”, successfully used in multiple nanomaterials, can trigger immunogenic responses due to the presence of natural surface antigens and can only be isolated from scarce blood supplies, as described earlier in the review. Moreover, “ghost” coatings have shown little surface stability over time due to its fragmented nature.^[178] To circumvent the use of erythrocyte “ghost” membranes, most reports use synthetic lipid carriers composed of 2–3 lipids, generally combining PC with cholesterol, PE, and/or PEGylated lipids. Additionally, there is broad consensus that erythrocyte stealth arises from the presence of integrin-associated protein CD47 and the near-absence of PS in the outer leaflet of the membrane.^{[20, 27, 179]}

Phospholipids are the primary constituents of native erythrocyte membranes (EMs) and are, naturally, the most commonly used building blocks in synthetic EMs. Other molecules such as albumin, heparin, and PEG have also served as coatings for blood-contacting biomaterials, but the well-established success of lipid nanocarriers in pharmaceutical sciences has given preference to their use in cell engineering alike. Within natural EMs, sphingomyelins (SMs) and PCs are the two most abundant lipid components of the blood-contacting leaflet, with 55 mol% and 42 mol% of all phospholipids, respectively.^[20] It is interesting to note that SMs are rarely included in synthetic membrane formulations, often being replaced by PCs. This fact is due to i) the similar chemical structure of SMs and PCs; ii) the low commercial availability of synthetic SMs; and iii) the distinctively higher price of both natural and synthetic SMs when compared to PCs. It has been suggested that saturated PCs (e.g. DSPC) could replace SMs in cholesterol-rich liquid-ordered domains to maintain lipid packing.^[180] While promising for the field of cell membrane engineering, more studies are needed to confirm that this substitution does not affect erythrocyte reshaping^[26] or cell signaling.^[181]

Cholesterol is another essential component of the lipid membrane. It regulates its fluidity and compressibility, and is often used in lipid-based pharmaceutical formulations such as Doxil^® and the COVID-19 vaccines.^{[182, 183]} While most cholesterol is derived from animal sources, pharmaceutical-grade synthetic and plant-derived cholesterol are also commercially available.

The use of PEGylated lipids has also been subject to controversy in recent years. Polyethylene glycol (PEG) is one of the most important non-toxic stealth agents in pharmaceutics and biomaterials.^{[79, 80]} However, presumably due to its widespread use in cosmetics, a number of researchers started reporting adverse responses to multiple injections of PEGylated liposomes in healthy animals.^{[184, 185]} Multiple studies have since been conducted on the issue, indeed identifying considerable levels of IgG and IgM after multiple injections of PEGylated nanocarriers.^{[32, 81, 186, 187]} These observations brought into question whether PEG is suitable as a component of small injectable constructs, since these immunoglobulins are linked to an accelerated blood clearance and hypersensitivity reactions.^[186] Similarly, other studies demonstrated that low molecular weight PEG (i.e. 1500 and below) is more cytotoxic than higher molecular weight.^[188] Overall, it is crucial to balance both advantages and disadvantages of using PEGylated molecules in the design of an engineered construct. If its use is favorable, immunogenicity screenings should be performed thoroughly and iteratively during early development stages; however, if the use is unfavorable, a number of PEG substitutes could be tested. These include glycosylated lipids, poly(sialic acid), polysaccharides, zwitterionic polymers, polyglycerols, polyoxazolines, and many others described elsewhere.^[189]

Lastly, to further improve stealth, CD47 coatings can be used in place of synthetic lipid bilayers. In an elegant study, streptavidin-modified particles were combined with recombinant biotinylated human CD47 to form a strongly bound, non-covalent coating based on a synthetic protein.^[190] Within the same study, minimal “self”-peptides were also synthesized based on the sequence of CD47 that binds directly to the SIRP-α receptors of macrophages, showing similar evasive behavior as recombinant CD47.^[190] As an alternative approach, carboxylated particles were coated with mouse recombinant CD47 through carbodiimide chemistry.^[191] The recombinant protein showed remarkably lower uptake in M1 macrophages than in M2, demonstrating an enhanced stealth in pro-inflammatory environments.^[191]

Advances in large-scale production of recombinant proteins could facilitate the use of recombinant CD47 as a synthetic EE coating. However, to ensure long-term stability, it is important to immobilize CD47 on the surface by means of covalent (e.g., plasma treatment) or strong non-covalent (e.g., biotin-avidin) interactions. Similarly, with the reports on the success of minimal “self” peptides in evading phagocytosis, it would be equally possible to conjugate these peptides to synthetic lipid membranes. In this way, both lipids and CD47 could be combined in a scalable way to form exceptionally stealthy, functional cell membranes.

4.4. Recreating CO₂ and NO transport functions

The transport of CO₂ constitutes another key function of erythrocytes. It is mostly attributed to carbonic anhydrases (CA) and Hb, which respectively catalyze the conversion of CO₂ into soluble bicarbonate ions (Equation 3)^[194] or bind CO₂ through its amino groups.^[192] Accumulation of CO₂ in blood can lead to lower rates of oxygen transport and intracellular acidosis, which translate in respiratory failure symptoms.^[193]

$${CO}_2+H_{2}O\rightleftharpoons {HCO}_3^{-}+H^{+}$$ (3)

To replicate CO₂ scavenging, one study reported the use of bovine CA coupled to polymerized Hb, catalase, and SOD to form a complex enzymatic structure.^[77] It was shown, however, that high amounts of CA were needed to recreate the native CO₂-binding capacity.^[77] Commercially available recombinant CA and Hb could be used as enzymatic substitutes, although at a considerably high cost. As an option, these recombinant proteins could be further engineered to display higher number of amino groups with CO₂ affinity, thereby reducing the amount of enzyme needed to achieve native function. Finally, PFCs constitute the least expensive options as O₂ and CO₂ synthetic carriers, although efforts are still in course to improve their biological response.^[12]

The regulation and transport of NO is also a known function of erythrocytes, with an intricate role in vasodilation and endothelial metabolism. In a similar manner as O₂, NO is transported in the heme pockets of Hb, and its conversion from nitrites is catalyzed by nitrate reductase.^{[45, 46]} To recreate this phenomenon synthetically, recombinant Hb or PFCs could be combined with recombinant nitrate reductase. Furthermore, to tune and manage O₂/NO transport, recombinant Hb could be further engineered for more accessible disposition of heme pockets.

Overall, most O₂ transport molecules used in blood substitutes (Hb, PFCs) also bind CO₂ and NO. On the other hand, O₂ generators need to be supplemented with CO₂ and NO transporters to better mimic erythrocyte function. Recombinant proteins and synthetic biology are, so far, the most powerful solutions to the large-scale production of CO₂ and NO transporters, as they are largely comprised of CA and nitrate reductase.

4.5. Recreating antioxidant activity

The alleviation of oxidative stress in erythrocytes is also primarily attributed to enzymatic action. This machinery includes catalase, SOD, GPx, and Prx2. The conversion of superoxide (O₂^•−) to H₂O₂ is ensured by SOD, while its subsequent degradation is attributed to catalase, GPx, and PRx2, among others. Incorporating these enzymes in engineered blood substitutes remains challenging, as small enzymes cannot be easily retained in certain matrices. To illustrate this, it was shown that SOD (32 kDa) could not be retained in Hb-albumin nanoparticles after polymerization, but catalase (240 kDa) could, retaining function for up to 24 weeks.^[83] Enzymes of intermediate molecular weight, such as GPx (83–95 kDa)^[194] or Prx2 (58 kDa)^[195], would have higher chances of retention than SOD; however, their commercial availability is limited, their retention in engineered constructs remains largely unstudied, and their function could still be performed by catalase to a certain extent.^[196] For these reasons, catalase remains the main antioxidant enzyme used in synthetic blood substitutes. While it can be isolated from tissues other than blood (e.g., liver), the use of animal tissue-derived proteins still requires additional sterilization, purification, and quality assurance steps to eliminate the risk for viral infections.

Synthetic antioxidants have emerged as an alternative to antioxidant enzymes. Ascorbic acid, for example, has been combined with Hb for oxidative stress relief.^{[82, 197]} Nanozymes such as Au and CeO₂ nanoparticles have also been incorporated in different HBOCs, both displaying catalase- and SOD-like activity upon contact with H₂O₂.^{[99, 198]} Albumin-coated MnO₂ nanoparticles have been used as multifunctional nanozymes, with roles in O₂ generation, antioxidation, and ROS scavenging.^{[135, 199]} Iron oxide nanoparticles (Fe₃O₄) have also demonstrated peroxidase-like activity^[200], having been incorporated in recombinant human ferritin shells for tumor detection.^[201] Recently, Mn-MOFs were produced via the coordination of Zn²⁺ and Mn^III meso-tetrakis(4-carboxyphenyl)porphyrin in the presence of a stabilizer (polyvinylpyrrolidone) to obtain 2D nanosheets of <300 nm in length.^[202] These MOFs were used as antioxidants in the treatment of rheumatoid arthritis, displaying catalase- and SOD-like activity in the disproportionation of O₂^•− and H₂O₂.^[202] Moreover, they were degradable under fluid stress and had a visible anti-inflammatory effect^[202], rendering them useful also as antioxidants for synthetic EEs.

More recently, carbon nanodots isolated from graphite and activated charcoal through oxidative treatment demonstrated a striking SOD-like activity of over 10,000 U mg⁻¹, similar to that of the natural enzyme.^[203] Another distinct strategy consisted in using ultra-small nitrogen nanobubbles (<50 nm) as short-term antioxidants and ROS scavengers.^[204] These particles were produced through a compression-decompression method in water, described previously by the group.^[205] Both studies present novel promising antioxidant options, but additional studies are needed to clarify their cytocompatibility and effect in blood rheology. Other multifunctional nanozymes are reviewed in more detail elsewhere.^[206]

On a final note, the choice of animal model is important in the correct assessment of oxidative stress response in vivo. Unlike humans and guinea pigs, rats and mice are able to synthesize ascorbic acid, in a diet-dependent manner.^[208] This results in a better tolerance to oxidative stress-inducing materials in the first group of animals, which does not necessarily translate well to humans.^{[208, 209]} Guinea pigs are, therefore, a more suitable animal model to study oxidative stress response of EEs in vivo. Additionally, for animal testing substitutes such as organ-on-chip and other cell-based assays, it is essential to choose appropriate cell lines and preferentially use human blood or blood-derived elements to ensure a reliable output in an eventual clinical trial initiation.

4.6. Recreating O₂ transport

The fundamental and vital function of the vertebrate circulatory system is to ensure the supply of O₂ to tissues. O₂ bound to Hb is transported through the bloodstream, originating from the lungs and reaching the designated tissues. Subsequently, individual cells receive O₂ through diffusive transport facilitated by micro-vessels. The microcirculation serves as the primary location for gas exchange, where both erythrocytes and vessel diameters are comparable in size. Erythrocytes act as cargo carriers, fulfilling the crucial role of delivering O₂ and transporting CO₂ for approximately 120 days before natural removal due to senescence.^[16] The effective delivery of O₂ heavily relies on the reversible O₂-binding capacity of Hb within erythrocytes. Moreover, the prolonged circulation of erythrocytes is made possible by an array of membrane proteins that prevent immune clearance, enabling them to remain in circulation for the specified duration. Nevertheless, the utilization of blood-derived proteins poses the potential risk of viral contamination if adequate sterilization measures are not implemented. Furthermore, their incorporation into products for large-scale manufacturing is hindered by the limited availability of blood resources. As a result, various approaches have been extensively investigated to replicate O₂ transport, including the development of synthetic O₂-binding molecules and O₂-generating compounds. As adjustment of designing the material for the artificial oxygen carrier, the application of its use varies depending on its O₂-binding ability and O₂ releasing properties.

PFCs were initially considered for their ability to store significant amounts of dissolved O₂ in a linear manner.^[113] However, due to severe side effects, researchers sought alternative solutions. Synthetic microparticle- and nanoparticle-based HBOCs were presented as alternatives, and while their lifespans are shorter than natural erythrocytes, they offer immediate infusion capabilities and short-term improvements in tissue oxygenation. Microparticle-based O₂ carriers consist of a core of O₂ gas enclosed by a 2-nm thick phospholipid monolayer, further stabilized with block copolymers to form microparticle emulsions. These carriers are suitable for emergency O₂ delivery. Nanoparticle-based O₂ carriers incorporate larger amounts of Hb per carrier material, increasing the O₂-carrying capacity per particle while minimizing effects on serum viscosity and oncotic properties upon infusion. They are particularly well-suited for addressing acute hemorrhage or temporarily alleviating decreased O₂-carrying capacity. In a different line, stem cell-derived O₂ carriers produced through induced hematopoietic differentiation, which closely resemble natural erythrocytes, can be artificially generated in vitro. However, there are potential challenges concerning limited product shelf-life, storage costs, shipment, and infusion, as well as issues related to inhomogeneity or low efficiencies in stem cell differentiation.

In view of approaching artificial O₂-generating materials, therapeutic microparticles have been extensively engineered in versatile designs. These constructs possess not only an ability to oxygenate tissues, but also have an indirect beneficial effect on the angiogenesis and vascularization of tissue engineered constructs.^[132] A body of literature has been established utilizing the chemical reaction of peroxides in water, using materials such as CaO₂, sodium percarbonate, and MnO₂ to mediate or contribute to the decomposition of H₂O₂ into O₂ in aqueous media.^{[132, 136, 210]} The O₂ release behaviors can be modulated by temperature, solubility, pH and catalysts, which are adjustable parameters in the physiological conditions. Importantly, peroxide decomposition into singlet oxygen, an unstable and toxic species, can also occur in the presence of certain peroxidases.^[211] Catalase, instead, promotes generation of ground state O₂.^[212] Similarly, catalase-like nanozymes, such as MnO₂ or MOFs, could therefore prove important in the correct generation of O₂ for blood and tissue oxygenation.

In general, the rate of O₂ generation is relatively rapid, which is undesirable for an artificial O₂ carrier. Hydrophobic O₂-generating microparticles (OMPs) have therefore been proposed as a solution, as the hydrophobic core-shell structure limits the contact of water with the encapsulated peroxides, thereby preventing the burst release of O₂. OMPs have become widely explored for the direct injection or implantation of OMP-encapsulated scaffolds for application in regenerative therapeutic tools. These systems enable the passive diffusion of O₂ directly to the desired tissues and cells, eventually promoting the angiogenesis in hypoxic injured area. Depending on the specific application, there are cases where therapeutic effects can be achieved with 3–5 days of O₂ release, while in other situations, sustained release for approximately 28 days is necessary for tissue regeneration and vascular reconstruction. Therefore, in a similar manner as EEs, it is crucial to have designs that allow for continuous O₂ release. These applications can be classified as either short-term or long-term therapy based on the targeted tissue area and O₂ diffusion, and can similarly be applied to erythrocyte engineering for short-term (emergency) or mid-term (urgent) care applications.

4.7. Ensuring biocompatibility

Biomaterials need to undergo extensive in vitro and in vivo biocompatibility tests before clinical trials. Some of the primary endpoints of biological evaluation include cytotoxicity, hemocompatibility, degradation, acute toxicity, and chronic toxicity, and EE biocompatibility testing is no exception. Since the primary biological response of EEs occurs upon contact with blood, hemocompatibility plays a defining role in their safety assessment. Criteria for in vitro hemocompatibility tests are described in ISO 10993–4 and ASTM F756, and typically include complement system, coagulation, platelet, hemolysis, leukocyte, and material surface analysis.^[213] While most materials do not fit the ideal standards of hemocompatibility, numerous strategies are nowadays available to enhance them. Conveniently, these strategies also improve other aspects of biocompatibility, such as degradation and toxicity.

Choosing the right surface chemistry is the most predominant strategy to ensure a safe biological profile. Stem cell-derived EEs are inherently cytocompatible and hemocompatible, as long as they do not present immunogenic surface antigens; however, this phenotype is extremely uncommon. To solve this problem, cell surface engineering techniques have aimed at shielding these antigens, for example through surface-anchored frameworks^[214] or antigen-blocking nanogels.^[215] For materials-based EEs, biocompatible materials are often elected as the core construct, and generally include PEG^{[32, 122]} or biomolecules (e.g., lipids, proteins, saccharides, nucleic acids, or a combination thereof).^{[89, 93, 120, 216]} However, when less biocompatible core materials are used, surface modifications (e.g., coatings) are typically added. Surface coatings are aimed at improving cytotoxicity and hemocompatibility, but also have an impact in degradation rate and circulation time, which ultimately determine the material’s acute and chronic toxicity. The most common surface coatings in blood-circulating materials include heparin^[217], albumin^[120], lipids^{[86, 218]}, or biological membranes.^{[32, 95, 216, 219, 220]} Heparin is a clinical anticoagulant and is one of the most common coatings in blood-contacting medical devices^[221] along with albumin.^[222] Lipids and cell membranes form highly bioinert coatings, with erythrocyte membranes showing great potential in prolonging the circulation time of nanoparticles.^[223] These coatings, however, need to be stable under the high shear stress of blood flow, and non-covalent coating methods are often not enough to ensure this stability.^[178] Strategies such as plasma treatment, biotin-avidin interactions, or click chemistry are some of the proposed solutions, with specific adaptations needed depending on the base material.

Choosing the right core properties is another important aspect of biocompatibility, especially when ensuring a safe acute and chronic response. Particle size affects biodistribution, with nanoparticles (~100 nm) tending to have a prolonged blood circulation and microparticles (> 3 μm) tending to accumulate in the site of injection.^{[157, 158, 224]} Smaller particle sizes seem, therefore, to be preferred for long-term applications. However, while a higher circulation half-life is indeed associated with a lower acute toxicity, chronic toxicity can still occur from bioaccumulation. To prevent this, the degradation rate of the materials can be tuned in agreement with its circulation time. Another key factor of biological response is deformability, with stiff microparticles (E > 67 kPa) tending to accumulate in the lungs and increasing the chances of pulmonary embolism in mice.^[122] Maintaining soft biomechanical properties (E < 26 kPa) is, therefore, essential in preventing acute toxicity.

To sum up, the biocompatibility of blood substitutes is predominantly determined by their size, surface chemistry, deformability, and degradability. Tuning these properties is essential to direct the use of EEs for short-term or long-term transfusion effects in a safe manner. Importantly, in vitro and in vivo models should be chosen or developed in a way that represents human physiology as accurately as possible. Human vasculature is a complex and unique system, with very few animal models providing similarly-sized blood vessels. Moreover, as mentioned earlier in this review, the blood of certain animal models (e.g., rats) has a more complete antioxidant machinery than human blood, which may result in a misleading preclinical safety assessment.^[208] Considerable efforts are now being directed towards the development of in vitro human vasculature^[225] and organ substitutes^{[226, 227]}, which hopefully will allow for more reliable biocompatibility testing at preclinical stages.

5. Conclusion and Outlook

Engineered erythrocytes are one of the most common and effective solutions for blood substitution. Following some initial success, which did not ultimately result in a widespread approval, numerous innovative products have emerged. The search for an ideal blood substitute continues, with synthetic alternatives slowly emerging as the most promising solutions in terms of universality and scalability. Surprisingly, to date, there is still a scarcity of studies on fully synthetic EEs, which can be justified by the higher versatility and functionality of HBOCs compared to PFCs. The use of Hb or other blood components, however, severely limit their mass-production, and it is therefore important to shift attention to safer and more scalable substitutes.

Remarkable progress has been made regarding the fabrication of functional materials as erythrocyte components. Some of the most innovative and promising examples include nanozymes as enzymatic substitutes, recombinant hemoglobin as an oxygen carrier, synthetic peptides as stealth agents, synthetic lipids as cell membrane constituents, and synthetic polymers as cell cores. We anticipate that these separate advances can be combined to produce extremely powerful blood substitutes, and we believe that a considerable number of hybrid EEs could be effortlessly adapted into fully synthetic EEs by substituting blood elements with functional materials. Moreover, advances in particle fabrication techniques are increasingly working towards large-scale manufacturing, and the synergy between fundamental biology and microfluidics is rapidly facilitating the development of more reliable pre-clinical testing platforms. Such progress is expected to bring an exponential development of new EEs, which will drive the process of clinical translation of functional blood substitution products. Furthermore, synthetic EEs can be tuned to serve multiple purposes beyond cell function – the combination of these cells with drugs, diagnostic agents, self-oxygenating compounds or other bioactive molecules opens new avenues in tissue engineering and nanomedicine.

To summarize, the synergy between multiple areas of expertise, namely materials science, bioengineering, synthetic biology, or organic chemistry, seems indispensable in the manufacturing of a universal “transfusion liquid” for blood substitution. Future efforts in next-generation erythrocyte substitutes should focus in combining different knowledge to build new improved synthetic constructs and, in parallel, leverage the properties of erythrocyte mimicry to develop new solutions for blood substitution, tissue engineering, and nanomedicine.

Table 1.

Microfabrication methods used in the production of erythrocyte-resembling microparticles, categorized into size and shape mimicry.

Fabrication Technique	Description	Estimated Size	Throughput, scalability	Drawbacks	Ref.
Parallelized microfluidics droplet generation	Single-chip containing 20,160 flow focusing droplet generators	5–9 μm	Very high, limited (1 × 1012 particles per hour)	Limited scalability for blood substitution	[149]
Ultrasonic emulsification	Emulsification of organic and water phases containing polymer and oxygen-generating cargo	0.1–20 μm	High, scalable (4.66 × 108 particles per small batch, assuming PCL average 5 μm)	High polydispersity	[132]
Membrane emulsification	Pressurizing dispersed phase through a Shirasu Porous Glass (SPG) membrane, yielding PFC particles of 7.9 ± 2.5 μm sizes	8–30 μm	Medium, limited	Limited scalability	[115, 152]
Electrohydrodynamic jetting	Liquid jetting induced by electrical field, yielding single or Janus particles in the nano-to-micron size range	0.1–9 μm	Medium, limited	Limited scalability	[154, 155]
Bio-silicification	Silicification of whole blood, separation of erythrocytes as templates	6–8 μm	Medium, limited	Need for blood	[98]
PRINT®	Deposition of polymer precursor on a mold, photocuring or crosslinking	0.01–200 μm	Very high, limited (500 mg per minute, or 1.45 × 1010 particles per hour, assuming polystyrene 5 μm)	Limited scalability for blood substitution;	[122, 156]
Micromechanical punching	Deposition of material film on deformable substrate, micro- punching desired particle shape	>100 μm	Medium, limited	Particle size too large	[142]
Ca(OH)2 templating	Producing biconcave Ca(OH)2 microparticles from CaCl2 and dextran sulfate	5–20 μm	High, scalable	Potential polydispersity of templates; selective cargo loading	[97, 143]
Solvent-induced shape change	Resuspension of microparticles in tetrahydrofuran (THF) or isopropanol (IPA) to induce biconcave shape	N/A	High, scalable	Irregular particle deformation	[115, 154]

N/A: not applicable

Table 2.

Substitutes for the enzymatic machinery of erythrocytes.

Enzyme substitute	Surface modification	Size	Function	Drawback	Ref.
Ascorbic acid	N/A	-	Antioxidant	Potentially fast release from carriers	[82]
Au NPs	MOF encapsulation	6 nm	Antioxidant	Potentially fast release from carriers; bioaccumulation in liver, spleen, kidney	[198]
CeO2 NPs	PLGA-PLL encasulation	2–3 nm	Antioxidant	Potentially fast release from carriers; bioaccumulation in liver, spleen, kidney	[207]
Fe3O4 NPs	Recombinant human ferritin	4.7 nm	Peroxidase	Potentially fast release from carriers; bioaccumulation in liver, spleen, kidney	[201]
MnO2 NPs	Bovine serum albumin	5.7 nm	Antioxidant, ROS scavenger, catalyst O2 generation	Bioaccumulation in liver, spleen, kidney	[163, 199]
MOF nanosheets	Polyvinyl-pyrrolidone	<300 nm	Antioxidant	Bioaccumulation in liver, spleen, kidney	[202]
Carbon nanodots	N/A	2 nm	Superoxide dismutase	Bioaccumulation in liver, spleen, kidney	[203]
Ultra-small N2 nanobubbles	N/A	<50 nm	Antioxidant, ROS scavenger	Potential changes in blood rheology	[204]

N/A: not applicable.

Biographies

Francisca L. Gomes is a PhD candidate at the University of Twente (Netherlands). She was awarded her M.Sc. degree in Bioengineering from the University of Porto (Portugal), and her research interests include the design and engineering of small constructs for nanomedicine. She is currently developing synthetic red blood cells as potential blood substitutes.

Dr. Seol-Ha Jeong is a postdoctoral researcher in Professor Su Ryon Shin’s lab at Brigham and Women’s Hospital, Harvard Medical School. She received her B.S. (2012), M.S. (2014) and Ph.D. (2017) degrees in department of Materials Science and Engineering from Seoul National University. She has numerous experiences in bioengineering, tissue engineering and the development of ion-based biological systems using hydrogels for biomedical applications. She is currently developing a promising 3d-printing technology and designing the bioinks which can deliver oxygen to surrounding tissue or to the encapsulated cells for chronic wound applications.

Su Ryon Shin is an Assistant Professor in the Department of Medicine, Division of Engineering in Medicine at Brigham and Women’s Hospital / Harvard Medical School. Her Ph.D. is in Biomedical Engineering from Hanyang University, Seoul, South Korea. She is an expert in nano-biomaterials, 3D bioprinting, microfluidics, biosensors, organs-on-a-chip, and tissue engineering. Specifically, She is developing an oxygenating microparticle to regulate the oxygen level of the stem cell niches that can potentially regenerate critical bone defects, diabetic or chronic wounds, and ischemic disease. She received the American Heart Association Innovative Award and Collaborative Award and the BWH Stepping Strong Awards.

Jeroen Leijten is Professor in the Department of Developmental BioEngineering at University Twente. He earned his Ph.D. in Tissue Engineering at University Twente, and performed postdoctoral research at KU Leuven and Harvard Medical School. His interdisciplinary research group focuses on development of nano- and microscale tools using enabling microfluidic technologies to drive scalable production of multiscale and hierarchically organized bioengineered constructs for biomedical applications and tissue engineering in particular. He has been awarded with the Jean Leray Award by the European Society for Biomaterials, and the Robert Brown Award by the Tissue Engineering and Regenerative Medicine International Society.

Pascal Jonkheijm earned his PhD in macromolecular chemistry from the University of Eindhoven (Netherlands) with Profs. E.W. Meijer and A.P.H.J. Schenning and stayed as an Alexander-von-Humboldt fellow in the Chemical Biology department of Prof. H. Waldmann (Germany, MPI for Molecular Physiology). He is currently Full Professor at the University of Twente leading the Laboratory of Biointerface Chemistry in the Molecules and Organ-on-Chip Centres. He received the 2018 Gold Medal of the Royal Netherlands Chemical Society. Present research interests include cell-instructive biointerfaces, self-assembly, cell-surface interactions, and biomimetics. He co-founded Lipocoat B.V., a high-tech start-up that commercializes bio-inspired coatings for medical devices.