Injectable oxygenation therapeutics: evaluating the oxygen delivery efficacy of artificial oxygen carriers and kosmotropes in vitro

Abstract The aim of this paper was to utilise an existing in vitro setup to quantify the oxygen offloading capabilities of two different subsets of injectable oxygenation therapeutics: (1) artificial oxygen carriers (AOCs), which bind or dissolve oxygen and act as transport vectors, and (2) kosmotropes, which increase water hydrogen bonding and thereby decrease the resistance to oxygen movement caused by the blood plasma. Dodecafluoropentane emulsion (DDFPe) was chosen to represent the AOC subset while trans sodium crocetinate (TSC) was selected to represent the kosmotrope subset. PEG-Telomer-B (PTB), the surfactant utilised to encapsulate DDFP in emulsion form, was also tested to determine whether it affected the oxygen transport ability of DDFPe. The in vitro set-up was used to simulate a semi closed-loop circulatory system, in which oxygen could be delivered from the lungs to hypoxic tissues. Results of this study showed that (1) 0.5 ml of a PFC outperformed 6.25 ml of a kosmotrope in a controlled, in vitro setting and (2) that PTB and sucrose do not contribute to the overall oxygen transportation efficacy of DDFPe. These results could be therapeutically beneficial to ongoing and future pre-clinical and clinical studies involving various oxygenation agents.


Introduction
With respect to a variety of acute medical conditions including traumatic brain injury [1][2][3], haemorrhagic shock [2,4], stroke [5,6], and acute respiratory distress syndrome, intravenous therapeutics that are able to facilitate respiratory gas exchange have become a promising avenue for treatment as post-onset prophylactics. Though these injectable therapeutics serve to restore adequate oxygen delivery after acute hypoxic injury, the mechanism through which they facilitate tissue oxygenation can greatly vary. One subset of these oxygenation therapeutics is known as artificial oxygen carriers (AOCs), which physically carry and transport oxygen (O 2 ) by either binding or dissolving diatomic O 2 molecules. AOCs found their origin as haemoglobin based oxygen carrier solutions, due to the stability and efficacy of haemoglobin as an O 2 carrier, and the field grew from there [7,8]. These O 2 carriers are efficient drug delivery systems in that they are cyclic transport vectors and can repeatedly load and unload O 2 for as long as they remain in circulation in vivo. Moreover, in lower O 2 partial pressure environments, AOCs with high O 2 binding affinities may deliver even more O 2 than red blood cells (RBCs) [7].
Within this field of AOCs, perfluorocarbons (PFCs) have a long-standing history of use and exploration as RBC substitutes. PFCs are highly fluorinated carbon chains with an ideal gas like chemical inertness, which allows for the dissolution of similarly inert gasses, such as O 2 , CO 2 , N 2 and NO, into stable CF 3 pockets formed by adjacent PFC molecules [9]. The efficacy of PFCs results from their increased O 2 delivery capability, which takes priority over O 2 absorption ability in vivo [10]. Consequently, although PFCs, which dissolve O 2 , possess a lower O 2 affinity than RBCs, which physically bind O 2 , the combination of this lower O 2 affinity in conjunction with higher diffusion rates, creates a reservoir of O 2 that is more readily available for extraction by the tissues [11]. Additionally, PFCs preferentially dissolve CO 2 over O 2 , which could aid in the targeted oxygenation of hypoxic tissues primarily undergoing anaerobic respiration and consequently, producing large quantities of CO 2 [11,12]. In order to be administered intravenously as an effective therapeutic, PFCs must be encapsulated as a microbubble or nanodroplet in a lipid monolayer and stabilised in an emulsion-a process that has continuously been refined over decades. When improving the efficacy of PFCs, two main variables must be considered: (1) the PFC itself and (2) the PFC encapsulation surfactant. For this reason, third generation emulsions have focussed on optimising the O 2 carrying capacity of PFCs, by increasing the CF 3 group density [13] while decreasing the molecular weight and boiling point [14], and utilising surfactants such as PEG-Telomer-B (PTB), which decrease the reactivity of the PFC with surrounding material, allowing for increased shelf life and stability during in vivo circulation.
Another subset of oxygenation therapeutics is known as kosmotropes. These are unique order-inducing chemical compounds that increase protein stability, reduce hydrophobic molecule stability, and form strong hydrogen bonds with water [15]. The mechanism of action of kosmotropes, with respect to O 2 delivery, is based on studies that identify blood plasma as accounting for nearly 70% of the resistance to O 2 movement experienced as O 2 diffuses from RBCs to tissue [16,17]. Due to their hydrophobic intermolecular forces, kosmotropes can form strong hydrogen bonds with surrounding water molecules, which make the water more structured on a microscopic scale, thereby decreasing its density. This physical change "opens up" the water phase to facilitate rapid O 2 diffusion and decrease plasma resistivity [18].
The aim of this experiment was to utilise an existing in vitro setup [12], capable of evaluating the O 2 offloading ability of a drug product in a hypoxic environment, to assess the relative efficacies of these two subsets of injectable oxygenation therapeutics in vitro. For the purposes of this experiment, dodecafluoropentane emulsion (DDFPe) was chosen to represent the AOC subset while trans sodium crocetinate (TSC) was selected to represent the kosmotrope subset. PTB, the surfactant used in DDFPe, was also tested to determine whether it affected the O 2 transport ability of DDFPe. The utilised setup allows for the simulation of O 2 transfer from the lungs to hypoxic tissues through a semiclosed circulatory system, with 0.9% saline serving as a blood plasma substitute. Testing drug products through such a setup allows for the evaluation of the relative efficacy of O 2 delivery while accounting for the effects of cyclic oxygenation and de-oxygenation of "blood" as seen in biological circulatory systems, as well as the additional pressures exerted by a semi-closed circulatory system [12].

Materials and methods
The in vitro set-up illustrated in Figure 1, established by Jayaraman et al. [12], was designed to simulate gas exchange and O 2 delivery from the lungs to hypoxic tissues. The O 2 offloading capability of various stabilised solutions including DDFPe, TSC, and PTB was measured through a series of assays (Table 1) utilising the set-up depicted in Figure 1.

Preparation of product
Sodium chloride was purchased from Sigma Aldrich. Sodium sulphite was purchased from J.T. Baker. Purified water (18MXÁcm) from an in-house purification system was used as the diluent. Dissolved O 2 measurements were obtained using an Oakton DO110 metre. Silastic tubing (1.47 mm I.D. Â 1.96 mm O.D.) was purchased from VWR. The finished medicinal product, 2% w/v dodecafluoropentane emulsion (DDFPe), manufactured by NuvOx Pharma (Tucson, AZ) was used [19]. Specifically, a 30% sucrose solution was homogenised along with PTB and DDFP. The emulsion was processed using a semi-sealed, stainless steel containment system attached to an Avestin Emulsiflex-C50 homogeniser keeping the temperature below 8 C. The homogenates were then subject to terminal sterile filtration immediately prior to filling into 10 ml vials. Particle sizing by Nycomp showed a mean particle size of approximately 250 nm. The PTB þ Sucrose solution was prepared by adding 0.779 g of purified PTB into 150 ml of a buffered sugar solution composed of sodium phosphate (Sigma Aldrich) and sucrose (Emprove Low Endotoxin Sucrose). Lastly, the stabilised TSC solution was prepared in accordance with the referenced United States Patent 6,060,511 by combining TSC with a mixture of 8% cyclodextrin (Sigma Aldrich) and 2.3% mannitol (Sigma Aldrich) [20].

In vitro model
The in vitro oxygenation experimental setup has previously been described in detail by Jayaraman et al. [12]. The setup was designed to simulate the O 2 uptake by the blood in the lungs, the transport of O 2 by the blood to the tissues, and finally, the uptake of O 2 by the tissues from the blood. The fluid reservoir (80 ml of 0.9% saline solution) represented O 2 uptake while the gas exchange vessel (900 ml of 0.9% saline solution) was purposed for O 2 offloading (see Figure 1). Silastic TM tubing placed in the gas exchange vessel simulated the circulatory system by permitting gas exchange and offloading O 2 to areas of low O 2 concentration. Sparging with nitrogen gas was used to lower the dissolved O 2 concentration of saline in the gas exchange vessel to approximately 1.75 mg/l. A continuous supply of nitrogen (2.0 standard ft 3 / minute) was also pumped into the headspace of the gas exchange vessel throughout the experiment to displace air since the vessel was not completely sealed. Once the sparging process was complete, the solution in the fluid reservoir received a constant influx of O 2 (2.0 l/min). Fluid was pumped out of the reservoir through the silastic tubing with a peristaltic pump at a rate of 30 ml/min. A galvanic dissolved oxygen (DO) sensing probe (DO 110 m, Oakton Instruments), which utilises a Clark electrode to assess the dissolved O 2 concentration, was used to monitor the O 2 concentration within the gas exchange vessel, which was recalibrated each day. Each assay was conducted over a 45-min duration at $21 C.
Mechanism of action study. About 0.5 ml of DDFPe was tested against 0.5 ml of 0.9% Saline and 6.25 ml of TSC (Table 1) to determine what effect the mechanism of action of the injected material has on its respective O 2 transportation abilities. The doses of DDFPe and TSC tested were determined from clinical literature [18,21] for an assumed 80 kg individual with 5 l of blood, which translated to 80 ml of saline in the fluid reservoir ( Figure 1). These assays were conducted under the same experimental conditions as mentioned above. It is important to note, however, that the TSC solution experimentation was performed in a dark environment to minimise potential photodegradation.
Artificial oxygen carrier (DDFPe) vs kosmotrope (TSC) equivalent dosage calculations: Assumptions: Average human blood volume ¼ 5 l Weight of patient ¼ 80 kg   To account for most effective usage of material, a 0.5 ml dose of DDFPe was selected for experimentation. The scaling factor of the clinical dilution to the in vitro dilution of DDFPe was calculated to be 3.906. This scaling factor of 3.906 was then used to determine the equivalent in vitro dilution of TSC based on the current clinical dilution of TSC (20 ml TSC/ 1 l blood). This equivalent in vitro dilution of TSC was determined to be 6.25 ml TSC for experimentation. The clinical and in vitro dosages of DDFPe and TSC listed under the heading "Clinical and In Vitro DDFPe and TSC Dose Comparison" are meant to depict the differences in the magnitude of active ingredient administered with respect to each AOC.
PTB þ sucrose control study. About 0.5 ml of a PTB þ Sucrose solution was tested against 0.5 ml of DDFPe and 0.5 ml of 0.9% Saline (Table 1) to observe whether the O 2 offloading behaviour of DDPFe was affected by the presence of PTB and Sucrose in the emulsion during testing. These assays were conducted under the same experimental conditions as mentioned above.

Statistical analysis
A statistical analysis to determine the significance of experimental results was conducted on Excel via a two-tailed, two-sample unequal variance (heteroscedastic) t-test. A heteroscedastic t-test was chosen since the compared samples had different variances and sample sizes, and were observed and analysed from independent experimental runs. Data are expressed as means ± standard error of mean (SEM). Significance was defined as p .05. Figure 2 displays the O 2 offloading data for assays 1-4 (Table  1). Assay 1, the saline control run, exhibited the lowest net O 2 increase, while assay 2, the DDFPe experimental run, resulted in the greatest net O 2 increase within the gas exchange vessel. The "net" increase or decrease refers to the total change in the O 2 concentration as observed from the beginning to the end of the experimental run. In comparing the magnitudes of O 2 transfer of DDFPe and TSC over the course of a 45-min run, an injection of 0.5 ml of DDFPe resulted in a net O 2 concentration increase of 5.95 mg/ l ± 0.283 mg/l, while an injection of 6.25 ml of TSC solution resulted in a net increase of 4.59 mg/l ± 0.187 mg/l. A standard t-test was used to verify the statistical significance of these results (p-value ¼ .03798).

Mechanism of action study
When assessing the mechanism of oxygenation observed in assays 1-3, Figure 3 displays the average oxygenation curves for the tested compounds over the course of 45-min. Based on Figure 3, it appears that both DDFPe and TSC display a logarithmic oxygenation behaviour, with neither compound having reached its maxima within the 45-min period. However, the relative slopes of the curves imply that, given sufficient time, the maximal oxygenation potential of TSC is most likely lower than that of DDFPe.  (Table 1). Assay 2, the DDFPe experimental run, resulted in the greatest net O 2 increase within the gas exchange vessel, while assay 1, the saline control run, exhibited the lowest net O 2 increase. The PTB þ Sucrose solution (Assay 4) exhibited no significant change in oxygenation when compared to saline over the course of 45 min. The net increase in O 2 concentration resulting from administration of DDFPe, PTB þ Sucrose, and Saline were 5.95 ± 0.283 mg/l mg/ l, 3.67 mg/l ± 0.103 mg/l, and 3.59 mg/l ± 0.088 mg/l, respectively.

Discussion
The results shown in Figure 3 demonstrate that, within the context of this in vitro simulation of hypoxia, AOCs appear to facilitate greater magnitudes of O 2 transfer at lower doses than kosmotropes. However, both compounds display statistically significant improvements in oxygenation when compared to the 0.9% saline control. Furthermore, the results in Figure 2 suggest that PTB and sucrose do not contribute to the overall O 2 transportation efficacy of DDFPe.
The diffusion of O 2 in vivo follows Fick's Law, which explains that the rate of O 2 diffusion is dictated by three variables: (1) the plasma thickness, which refers to the distance that O 2 must diffuse over in order to reach the target cells, (2) the O 2 concentration gradient, and (3) the diffusion coefficient or diffusivity of O 2 [23]. However, plasma thickness is physiologically determined by the anatomy of the vasculature. Therefore, injectable oxygenation therapeutics seek to either increase the O 2 concentration gradient or increase the diffusion coefficient in vivo to encourage greater magnitudes of O 2 diffusion. PFCs fit into the former mechanism, while kosmotropes fit into the latter [24]. Figure 4 further illustrates this concept of how PFCs and kosmotropes theoretically affect normal O 2 transport in vivo.
Under normal physiological conditions, O 2 transport is dictated by the Bohr Effect, which describes how haemoglobin cooperatively binds O 2 in higher pH environments and releases O 2 in lower pH environments caused by increased CO 2 tension [25]. The O 2 released by RBCs then passively diffuses through the blood plasma, based on oxygen tension (pO 2 ) gradients, in order to reach the cell membrane [26]this is the major rate limiting step. At physiological pO 2 , oxygen possesses an extremely low solubility in plasma. Furthermore, the O 2 extraction ratio (OER), which describes the extent to which RBCs release bound O 2 , is only $25% at rest. This indicates that under normal conditions only 25% of bound O 2 is released into the plasma, and even under conditions of hypoxia and exceptional metabolic stress, this ratio rarely, if ever, exceeds 75% [25]. This is in part due to an unstirred, stagnant boundary layer of plasma that forms as a Figure 2. Net oxygen offloading in mechanism of action comparative study. The net increase in oxygen concentration in the gas exchange vessel for assays 1-4 ( Table 1). The error bars displayed reflect the SEM. The statistical analysis conducted was a two-tailed, two-sample unequal variance (heteroscedastic) t-test. p-Values are 0.01520, 0.03798, 0.03208 for Saline vs. DDFPe, DDFPe vs TSC, and Saline vs TSC, respectively. For the PTB þ Sucrose Control study, p-values are 0.01472 and 0.57763 for DDFPe vs PTB þ Sucrose, and Saline vs PTB þ Sucrose, respectively. result of the turbulent flow through arterioles and capillaries [16,17]. These layers surrounding RBCs create local O 2 concentration gradients of lesser magnitude than that of the overall RBC to tissue O 2 concentration gradient, and therefore (1) limit the extent of O 2 released by RBCs and (2) increase the distance over which O 2 molecules must diffuse in order to reach the cell surface. Figure 4(A) depicts O 2 transport and diffusion under normal physiology, which is limited by low solubility and high resistivity experienced in blood plasma. Consequently, although an OER of 25% is sufficient for normal physiological function, in the event of an acute hypoxic event, even an elevated OER of 75% is often not enough to prevent tissue damage without the aid of molecules to facilitate transport through the plasma. Figure 4(B) depicts O 2 transport and diffusion in the presence of PFC based artificial O 2 carriers. It was previously theorised that PFCs act as RBC substitutes due to their high affinity for O 2 without the trade-off of reduced bioavailability. However, in terms of mathematical modelling, the effects observed after administration of DDFPe far outweigh what is expected given the small dosage [10]. Consequently, it is more probable that PFCs act in conjunction with circulating RBCs to improve the efficiency of O 2 transport, as opposed to acting independently as "superior RBCs". As seen in Figure  4(B), PFCs presumably act as an intermediate transport vessel for released O 2 in the vasculature. It is theorised that they improve the efficiency of O 2 transfer by increasing local O 2 gradients caused by unstirred layers, which in turn encourages RBCs to readily release more O 2 . Circulating PFCs then dissolve released O 2 and facilitate O 2 transfer through the resistive plasma, and once at the tissue surface, the weak intermolecular forces experienced between PFCs and dissolved gases, and the larger surface-to-volume ratio of PFCs, are more favourable for gas exchange [12]. The results observed using this in vitro setup support that a small dose of PFC can demonstrate a significant increase in oxygenation.
Kosmotropes, contrastingly, facilitate O 2 transport by increasing the diffusion constant, also known as the diffusivity, of O 2 in plasma. They do so by decreasing the entropy of water molecules in the plasma, which thereby reduces the plasma density and decreases the resistance faced by O 2 molecules [18]. As previously mentioned, blood plasma is comprised of nearly 92% water. The intrinsic structure of water, and by extension plasma, is due to hydrogen bonds formed between adjacent H 2 O molecules. Theoretically, a single water molecule should be able to form up to four hydrogen bonds simultaneously. However, in reality, this number averages between 2 and 3.6 [24]. Kosmotropes are order inducing molecules. TSC, for example, is a large hydrophobic molecule. Consequently, as illustrated in Figure 4(C), administration of a kosmotrope, such as TSC, decreases plasma entropy by interacting with similarly hydrophobic plasma components and therefore encourages the formation of additional hydrogen bonds between polar water molecules. This brings the average number of hydrogen bonds per water molecule closer to 4 [24], and this physical change in density "opens up" the water phase of the plasma allowing O 2 to diffuse towards the vascular wall more easily [18]. This change in density is illustrated in Figure 4(C) with the lighter plasma regions through which the O 2 is diffusing. This in vitro setup utilised 0.9% saline solution as a plasma substitute in simulating hypoxia. Consequently, the increase in oxygenation observed through these experiments supports that kosmotropes increase the diffusivity of O 2 primarily by affecting interactions between water molecules.
Additionally, these results indicate that the use of PTB as a surfactant to encapsulate PFCs does not beneficially nor detrimentally affect the oxygenation abilities of the PFC. The oxygenation capability of PTB was originally tested due to the similarities between the hydrophilic poly(oxyethylene) (POE) regions of Poloxamer 188 and PTB (see region x, Figure  5(A,B)). Poloxamer 188 is a surfactant that has shown success in mitigating the severity of acute chest syndrome episodes in sickle cell anaemia patients [27,28]. However, the mechanism through which Poloxamer 188 improves micro-vascular blood flow is thought to involve binding between the hydrophobic core (see region y, Figure 5(A)) of Poloxamer 188 and adjacent hydrophobic RBC and neutrophil cell surface regions, which appears to block extraneous hydrophobic adhesive interactions in the bloodstream and thereby reduce the blood viscosity. This allows the hydrophilic POE chains (see region x and z, Figure 5(A)) free to interact with the surrounding media [28]. PTB shares the same hydrophilic POE chain (see region x, Figure 5(B)) as Poloxamer 188, however, it lacks the hydrophobic core. Instead, PTB's chemical composition includes a fluorinated tail (see region y, Figure 5(B)). Consequently, based on these in vitro experiments where the hydrophobic silastic tubing is representative of the vasculature, it appears that the chemical similarities between the poly(oxyethylene) hydrophilic units in Poloxamer 188 and PTB, do not amount to the same capabilities of facilitating O 2 transfer in vitro. Rather these in vitro results support the supposition that the success of Poloxamer 188 in vivo is mechanistically more accredited to the hydrophobic core of Poloxamer 188 than its hydrophilic chains.
The results of this study show that 0.5 ml (10 mg DDFP) of a PFC outperformed 6.25 ml (125 mg TSC) of a kosmotrope in a controlled, in vitro setting. It should be stated that the numerical O 2 offloading concentrations measured in this in vitro model cannot be scaled quantitatively to an in vivo situation due to the physiological limitations of an acellular, in vitro model. However, the trends observed in the in vitro model may still be biologically relevant.
With respect to both PFCs and kosmotropes, it would be pertinent to utilise whole blood in this in vitro model as both classes of molecules rely heavily on blood components mechanistically. In terms of PFCs, in these experiments, it is very likely that DDFPe did not display its full oxygenation potential due to the lack of RBCs present. The use of whole blood, therefore, could account for synergistic oxygenation effects of RBCs in tandem with PFCs. Whereas for kosmotropes, the use of whole blood would create a more representative model of the extent to which these molecules can alter the density of plasma and consequent diffusivity of O 2 in vivo.
Additionally, the use of whole blood would provide a clearer sense, in general, of how resistive blood plasma is to O 2 movement in a simulation of extreme hypoxia. Furthermore, it would be of interest to include a test group in which DDFPe and TSC were administered together. Future in vitro studies will aim to incorporate such cellular components in order to create a more representative in vivo mechanistic model.

Disclosure statement
Meghna Jayaraman is an employee at NuvOx Pharma. Kaitlin Graham is an employee of NuvOx Pharma and owns stock in the company. Dr. Evan Unger is President and CEO of NuvOx Pharma, serves on the Board of Directors, and owns stock in the company. Dr. Unger is also a patent holder of the NuvOx Pharma technology. The authors report no other conflicts of interest in this work.