Macrophage modulation by polymerized hemoglobins: Potential as a wound-healing therapy

Abstract

Chronic skin wounds are hypoxic and are stalled in a pro-inflammatory state. Hemoglobin (Hb)-based oxygen carriers have shown potential in increasing oxygen delivery to aid wound healing. Macrophages also take up Hb, thus altering their phenotype and the regulation of inflammation. Herein, we compared the effect of Hb and polymerized Hbs (PolyHbs) on the phenotype of human macrophages. Macrophages were incubated with Hb or different forms of PolyHbs, and the inflammatory secretion profile was analyzed. PolyHbs were produced by polymerizing Hb in the relaxed (R) or tense (T) quaternary state and by varying the molar ratio of the glutaraldehyde crosslinking agent to Hb. Hb decreased the secretion of most measured factors. PolyHb treatment led to generally similar secretion profiles; however, Hb had more similar trends to R-state PolyHb. Ingenuity pathway analysis predicted positive outcomes in wound healing and angiogenesis for T-state PolyHb prepared with a 30:1 (glutaraldehyde:Hb) polymerization ratio. When tested in diabetic mouse wounds, T-state PolyHb resulted in the greatest epidermal thickness and vascular endothelial CD31 staining. Thus, the effects of PolyHb on macrophages are affected by the polymerization ratio and the quaternary state, and T-state PolyHb yields secretion profiles that are most beneficial in wound healing.

INNOVATION

Topical application of hemoglobin (Hb) is an experimental wound-healing therapy that has shown some potential. It is generally thought that topical Hb may mediate its beneficial effects by increasing oxygen tension in the wound. In general, extracellular Hb is taken up and degraded by macrophages—a process that results in profound alterations in the phenotype of macrophages. Furthermore, macrophage phenotype—and, in particular, switching from pro-inflammatory to anti-inflammatory phenotypes—is central in the progression of wound healing. Herein, we reported that extracellular Hb and less toxic polymerized versions of Hb (PolyHbs) at levels that are well below those needed to significantly influence oxygen transport profoundly affect macrophage secretory functions. PolyHbs, when topically applied at low levels on mouse wounds, had a significant stimulatory effect on angiogenesis and wound healing.

INTRODUCTION

Chronic wounds are a major healthcare problem in the United States, affecting 6.5 million people¹. A wound is chronic, if it remains open for greater than 1 month and does not show signs of healing²; in many cases, such wounds remain open for longer than 12 months³. These wounds are usually located in areas with impaired blood flow, such as extremities (legs/feet) in the case of diabetic, venous, and arterial ulcers, or underneath bony surfaces for pressure ulcers, which are common in patients with spinal cord injury. Although the underlying etiology of chronic wounds may be different, they exhibit several characteristics such as being stuck in a pro-inflammatory state, having poor vascularization and low oxygen levels, and resisting regeneration of dermal and epidermal skin layers, which also makes them prone to infection³.

Typical treatment approaches include wound debridement, delivery of antibiotics, and periodic wound dressing changes. If wounds do not respond to these treatments, then more advanced therapies are used. One example is oxygen delivery-based approaches such as hyperbaric oxygen therapy (HBOT) or topical oxygen therapy (TOT). HBOT involves a pressurized chamber at two times atmospheric pressure, containing 100% oxygen⁴. As patients lay within the chamber, increased levels of oxygen enter their lungs and blood plasma, thus increasing pO₂ within the wound. In addition to increased oxygen transport, other reported benefits of HBOT include reduced wound edema, stimulation of progenitor stem cells, and angiogenesis as well as improved fibroblast function⁵. TOT involves direct delivery of oxygen gas to the wound surface via a pump, either at normobaric or pressurized conditions⁶. Both HBOT and TOT have been reported to accelerate the healing of chronic ulcers in specific instances^4–6. The disadvantages of these approaches are limited patient mobility, lengthy treatment protocols, and requirement of additional, specialized machinery that is costly and not readily available. An alternative approach would be to use an oxygen carrier to enhance oxygen delivery locally even under normal atmospheric conditions.

Hb is the protein in our red blood cells responsible for binding and delivering oxygen throughout the body⁷. It achieves this process by changing conformation in response to the partial pressure of oxygen (pO₂)—a relaxed state (R-state) when pO₂ is high and Hb is saturated with four oxygen molecules (O₂) bound, and a tense state (T-state) when pO₂ is low and no oxygen is bound. Hb can readily change from the R- to T-state depending on the surrounding pO₂, and as a result, oxygen is delivered throughout the body. To control oxygen delivery, Hb can be chemically crosslinked into the R- or T-states⁸. This has been achieved through the development of hemoglobin-based oxygen carriers (HBOCs), particularly polymerized hemoglobins (PolyHbs). PolyHbs can also have a range of molecular weights, depending on the extent of cross-linking, which are much greater than those of native Hb⁹. HBOCs have been traditionally studied as alternatives to blood transfusions, as they have extended shelf life and there is no need to match red blood cell type^7,10. More recently, their oxygenation potential has been studied in other applications such as islet transplantation¹¹ and treatment of tumors¹².

Hb-based therapies have also been investigated in wound healing applications. For example, Plock et al. intravenously delivered Hb vesicles to mice with ischemic skin flaps¹³. They observed that oxygenation, tissue survival, and healing of the skin flap edges were improved. Furthermore, Hb vesicle injection resulted in higher capillary counts and endothelial nitric oxide synthase (eNOS) expression. Another Hb-based therapy, called Granulox (Sastomed GmbH, Georgsmarienhütte, Germany), is approved in Europe for the treatment of surgical wounds and diabetic, venous, and arterial ulcers¹⁴. In human chronic wounds, Granulox increased oxygen levels¹⁵ and reduced wound exudate, wound size, and pain levels¹⁶.

In these examples, the regenerative effects of Hb-based therapies have been attributed to oxygen delivery, but we aimed to investigate their effects on macrophages, which play important roles in wound healing. Chronic wounds are characterized by a highly inflammatory environment that is dominated by pro-inflammatory M1 macrophages¹⁷. These cells have high levels of damaging reactive oxygen species (ROS) and secrete pro-inflammatory cytokines and chemokines such as tumor necrosis factor-α (TNF-α), interleukin 6 (IL-6), and interferon-γ (IFN-γ), to name a few¹⁸. In healing wounds, the macrophage population transitions to an anti-inflammatory M2 phenotype, which secretes factors such as IL-10, vascular endothelial growth factor (VEGF), and others that reduce inflammation and promote vascularization and regeneration¹⁹. In chronic wounds, the M1–M2 macrophage phenotype transition does not occur, and thus, the healing is stalled. It is plausible that interventions that attenuate the M1 phenotype while promoting the M2 phenotype may help in the healing of chronic wounds.

It is known that Hb can interact with macrophages to elicit an M2-like phenotype through the heme oxygenase 1 (HO-1) pathway^20–22. For example, in atherosclerosis, in areas of intraplaque hemorrhage, red blood cells are ruptured and iron is released from Hb^23,24. In these high-iron areas, a unique macrophage phenotype was identified, with higher expression of the M2 markers CD206 and CD163 and lower expression of the M1 marker TNF-α.²⁵ Hb also forms tight complexes with the plasma protein haptoglobin. The complexes are then internalized by the monocyte/macrophage-specific receptor CD163. Intracellular breakdown of heme activates the HO-1 pathway, resulting in downstream upregulation of the anti-inflammatory cytokine IL-10^25–27. As Hb can interact with macrophages and alter their secretion profile, further investigation on Hb/macrophage and PolyHb/macrophage effects is warranted when studying their potential as chronic wound healing therapies. The current study aimed to characterize the inflammatory secretion profile of macrophages treated with Hb and PolyHbs to determine which type may be the most beneficial for wound healing.

Here, we studied the effect of Hb and PolyHbs on macrophage phenotype in a pro-inflammatory, in vitro environment that mimics chronic wounds. For PolyHbs, both R- and T-state forms were tested, including two polymerization molar ratios. We hypothesized that PolyHbs would be less toxic to macrophages, as the chemical crosslinks hinder the release of the toxic heme group, which leads to oxidative damage. This was verified experimentally, and several key proteins secreted by macrophages were identified, which exhibited significant differences between Hb and PolyHb treatments. Furthermore, T-state PolyHb exhibited the most potential in stimulating wound healing and angiogenesis.

MATERIALS AND METHODS

PolyHb synthesis

PolyHb was synthesized as described in Zhang et al.⁸. Briefly, human red blood cells were lysed by exposure to hypotonic conditions^28,29. To remove cell debris, the lysate was passed through a glass wool column and then further purified using a three-step tangential flow filtration process. Purified Hb was diluted in phosphate buffer solution. R-state PolyHb was synthesized by reacting glutaraldehyde with completely oxygenated Hb for 2 h at 37°C. T-state PolyHb was synthesized by reacting glutaraldehyde with completely deoxygenated Hb under the same condition as that for T-state PolyHb. Two different polymerization molar ratios were used to synthesize R- and T-state PolyHbs—30:1 and 35:1 (glutaraldehyde:Hb molar ratio), denoted as R:30, R:35, T:30, and T:35. After 2 h, NaBH₄ was added to quench the polymerization reaction. Following polymerization, PolyHb was subjected to diafiltration to remove unpolymerized Hb and other small molecules from the solution. PolyHb was concentrated to 100 mg/mL in a modified Ringer’s lactate buffer, sterile-filtered, and stored at −80°C until needed.

Monocyte isolation and macrophage differentiation

Human blood/buff y coat donations were purchased from the New York Blood Center (New York City, NY). Primary monocytes were isolated using Ficoll-Paque density gradient centrifugation and CD14⁺ magnetic bead separation (Miltenyi, Bergisch Gladbach, Germany) in a process similar to that reported in Faulknor et al.³⁰. CD14⁺ cells were cultured at 5 × 10⁵ cells/mL with 5 ng/mL granulocyte-macrophage colony-stimulating factor (GM-CSF; R&D Systems, Inc., Minneapolis, MN) for 7 days at 37°C and 5% CO₂ to differentiate them into M1 macrophages. Complete media for all monocyte/macrophage cell cultures were Advanced RPMI 1640 (Life Technologies, Carlsbad, CA) containing 10% fetal bovine serum, 1% penicillin-streptomycin, and 4 mM L-glutamine.

Cell culture

Macrophages were cultured in complete media in a 37°C, 5% CO₂ incubator at 5 × 10⁴ cells/well in plastic 24-well plates and allowed to attach for 24 h. They were then activated with 1 μg/mL lipopolysaccharide (LPS) from Escherichia coli (Sigma-Aldrich, Saint Louis, MO) to induce inflammation. Simultaneously, macrophages were treated with 0.2, 2.0, or 20.0 mg/mL of Hb (H) or PolyHb (T:30, T:35, R:30 or R:35) for 48 h, following which supernatants were collected and stored at −80°C until use in the multiplex immunoassay.

Metabolic activity measurement

Following supernatant collection, Alamar Blue Cell Viability Reagent (Life Technologies Corporation, Carlsbad, CA) was mixed in a 1:10 ratio with media containing 1 μg/mL LPS. Five hundred microliters of the prepared mixture were added to each well to measure net cellular metabolic activity. The assay is a fluorescent-based detection method, wherein resazurin is reduced and converted to a fluorescent compound, resorufin, by living cells. After 6 h, fluorescence measurements (excitation 535 nm; emission 595 nm) were performed on a DTX 880 Multimode Detector plate reader with Multimode Detection Software (Beckman Coulter, Brea, CA).

Multiplex immunoassay

A panel of 27 cytokines, chemokines, and growth factors related to inflammation was measured using a Bio-Plex Pro Human Cytokine 27-plex Assay (BIO-RAD, Hercules, CA). All measured factors are listed on the heatmap in Fig. 2a. The assay was carried out according to the manufacturer’s instructions. Measurements were performed on media controls and 2.0 mg/mL Hb/PolyHb samples on a Bio-Plex 200 System (Bio-Rad, Hercules, CA). Results are represented as fold changes (FCs). Raw secretion concentrations were normalized to media baseline values. Then, FC was determined by taking the log₂ of the media-normalized value.

Principal component analysis

Matlab (Mathworks, Natick, MA) was used to run principal component analysis (PCA) on the secretion profile data sets. The code produced a clustergram for the data set and determined the covariance of each treatment pair. Principal component scores based on the weights of 27 inflammatory factors were generated. Principal component values (PC1, PC2, and PC3) for each treatment were plotted against each other.

Ingenuity pathway analysis

Ingenuity pathway analysis (IPA) version 01–13 (Qiagen, Venlo, Netherlands) was used to make predictions of biological/wound healing outcomes of Hb/PolyHb treatments from the secretion data sets. Th e average FCs from three experiments for each condition were included in an expression core analysis with no mutations, including direct and indirect relationships; interaction and causal networks; all node types and data sources; experimentally observed and high (predicted) confidence; and restricted to fibroblast and macrophage cell lines and primary human cells including endothelial cells, keratinocytes, fibroblasts, and immune cells (dendritic cells, granulocytes, mononuclear leukocytes, and peripheral blood leukocytes). z-Scores were used to assign a value to the predicted up-/downregulation of vascular and wound healing events from the treatment compared to media baseline conditions. The z-scores represent the predicted activity of biological events using the expression patterns of the downstream factors, based on relationships published in the literature.

In vivo wound healing studies

All animal protocols were approved by the Rutgers University Institutional Animal Care and Use Committee. Ten-week old, male, genetically diabetic mice—BKS.Cg-Dock7m +/+ Leprdb/J (The Jackson Laboratory, Bar Harbor, ME)—were used. Animal handling, wounding procedure, and histological analyses were performed as previously described^31–33. On the day prior to surgery (Day –1), mice were anesthetized using inhaled isoflurane (Henry Schein Animal Health, Melville, NY) delivered through a nose cone. Th e backs were shaved using clippers to remove the majority of the hair where the wounds would be created. Residual hair was chemically removed by the topical application of Nair™ (Church & Dwight Co. Inc., NJ) for 60 s and then cleaned with a paper towel and warm water. Mice were caged individually from this point onward. On the day of wounding (Day 0), mice were again anesthetized. The wound area was cleaned three times alternately with betadine antiseptic surgical scrub (Avrio Health L.P., Stamford, CT) and 70% ethanol. A 1 × 1 cm template was traced on the mouse skin to demarcate the edges of the wound to be excised. By using autoclaved tweezers in one hand, the center of the traced region was lifted and then cut through the center with surgical scissors. Then, the remainder of the skin was removed by cutting along the edge of the traced region. An image was taken of the wound with a ruler to serve as a reference so that the initial wound size could be determined.

Mouse wounds were topically treated with either human Hb, human T-state PolyHb (30:1), human R-state PolyHb (30:1), or Ringer’s Lactate (vehicle control). Twenty milligrams of each treatment (200 μL of 100 mg/mL solution) sample were applied on Day 0. Tegaderm™ (3M, Saint Paul, MN) dressing was sutured over the wound to hold the treatment in place. Following surgery and treatment on Day 0, mice were injected subcutaneously with analgesic (buprenorphine, 0.05 mg/kg) and returned to their cages. Mice received an additional 10 mg (100 μL of 100 mg/mL) of treatment sample delivered every 7 days for 4 weeks.

Images of the wound were taken with a ruler for reference on Days 0, 3, and 7, and once weekly until all wounds were closed. Then, the images were analyzed using Image J (NIH, Bethesda, MD). Th e wound area was measured by tracing the wound edges with the polygon tool and converting the area from pixels² to cm². “100% closed wounds” were defined as those that had no visible open skin or scab. During the course of healing, percent wound closure (W_p) was normalized to wound size on Day 0 of each individual wound and defined by:

$$W_p=\left(1-\frac{W_x}{W_0}\right)\times 100.$$

where W_x is the wound area on Day x and W₀ is the initial area on Day 0.

Histological staining

On Day 35, mice were sacrificed and wound/scar tissues were harvested. Tissues were fixed in 10% formalin for 24 h and then stored in ethanol at 4°C. Tissues were then paraffin-embedded, sectioned (5 μm), and stained with hematoxylin and eosin (H&E). The immunohistochemical staining of CD31 was performed as described in Kumar et al.³⁴. A primary rabbit polyclonal anti-CD31 antibody (Abcam, Cambridge, MA; 1:200) was used, followed by secondary antibody (Biotinylated Goat Anti-Rabbit, 1:200; Vector Laboratories, Burlingame, CA).

Next, 4× images of histological sections were captured. ImageJ software was used to analyze the images obtained. Epidermal thickness was measured in three locations per slice (center, left, and right edges). Th e density of CD31 was determined by measuring the dermal area in the image view and then counting positively stained areas.

Statistics

GraphPad Prism 8.1.1 (330) (GraphPad Software, San Diego, CA) was used for all statistical analyses as well as to generate all plots other than those shown in Fig. 2. One-way ANOVA with Tukey’s post-hoc analysis was used to identify significant trends in all analyses, except for metabolic activity (Fig. 1) and inflammatory secretion (Fig. 3) analyses, in which two-way ANOVA was used. *, +, and # are used to identify significance between groups, as specified in the figure legends. Increasing *, +, or # indicates the increasing level of significance. For example, * denotes p < 0.05, ** denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001. Th e same holds true for increasing numbers of + and # symbols.

RESULTS AND DISCUSSION

Effect of Hb/PolyHb on the cellular metabolic activity of macrophages

First, we tested a range of concentrations of Hb/PolyHb on macrophages to determine doses at which net cellular metabolic activity/viable cell number is not negatively affected. M1 macrophages seeded 24 h prior were stimulated with 1 μg/mL of LPS. At the same time, 0.2, 2.0, or 20.0 mg/mL of Hb/PolyHb was added. PolyHb treatments included both T-state and R-state polymers, at the polymerization molar ratios of 30:1 and 35:1. After 48 h of incubation, supernatants were collected and net metabolic activity per well was measured. Results are shown in Fig. 1 and are normalized to media baseline measurements.

Figure 1. Cellular metabolic activity versus Hb/PolyHb concentration.

Metabolic activity of macrophages cultured in media with or without 0.2, 2, or 20 mg/mL Hb T:30 PolyHb, R:30 PolyHb, T:35 PolyHb, or R:35 PolyHb. Results were obtained using the Alamar blue assay, and relative fluorescence intensity was measured on a plate reader. Results are grouped by treatment type, across all three concentrations. Black bars represent values for the media group (controls). * indicates significance compared to the media baseline. # indicates significance between bracketed groups.

At increasing concentrations of Hb, net metabolic activity decreased. This was significant at 20 mg/mL, compared to media baseline and lower Hb concentrations (0.2 and 2.0 mg/mL). In contrast, net metabolic activity per well remained close to baseline conditions for macrophages treated with any type and concentration of PolyHb (Fig. 1). The decrease in net metabolic activity at 20 mg/mL of Hb may be a reflection of a decrease in the attached cell number (Supporting Information, Fig. 1). Additional morphological characterization and ROS level measurements of macrophages treated with increasing concentrations of Hb were also performed and confirmed that 20 mg/mL Hb damages the macrophages (Supporting Information, Fig. 2). Overall, these results suggest that 0.2–2.0 mg/mL of Hb is a safe concentration range for macrophages. Furthermore, this range extends to 20 mg/mL for PolyHbs, which were less toxic to macrophages at this higher dose than that for unmodified Hb.

One possible reason for less toxicity of PolyHbs versus Hb at high concentrations may be that the damaging heme group is less readily released from PolyHbs due to chemical crosslinking. In unmodified Hb, the heme group is easily released from the protein and causes oxidative damage, which can ultimately lead to apoptosis of cells³⁵. Based on these results, further experiments were carried out at 2.0 mg/mL of Hb/PolyHb or lower, within the safe range for both Hb and PolyHbs, thus allowing for easy comparison between the groups.

Net effect of Hb/PolyHb on the secretion of inflammatory factors from macrophages

To characterize the macrophage phenotype in response to Hb/PolyHb treatment, we measured inflammatory protein secretion using a 27-plex immunoassay. Figure 2a provides a heatmap of, the secretion results for each treatment and protein. Results were normalized to media conditions and represented as FCs above or below (red/green, respectively) media levels. Overall, a majority of factors (21/27) decreased with Hb treatment, whereas trends with PolyHbs were more variable. For about half of the factors (15/27; from IP-10 to IL-4), trends for Hb were generally different than those for PolyHbs. For example, for IL-6, treatment with Hb resulted in an increase in secretion, whereas treatment with all PolyHbs led to a decrease in secretion. Similarly, for MIP-1α, the Hb group exhibited a decrease in secretion, whereas all PolyHbs exhibited an increase in secretion. Several other factors (IL-15, IL-17, IFN-γ, and IL-4) decreased with Hb treatment and remained close to baseline with PolyHbs. In the second half of the clustergram (IL-5—RANTES), Hb treatment had similar results to PolyHbs. For example, IL-5 and IL-9 decreased in all Hb/PolyHb groups. From IL-1RA to RANTES (9/27 factors), Hb and R-state PolyHb groups had similar, decreasing trends, and T-state PolyHb groups showed slightly higher values. These trends generally hold true regardless of the polymerization molar polymerization molar ratio of T-or R-state PolyHbs. Overall Hb treatment decreased a majority of the factors and had similar trends to PolyHbs for about half of the measured factors, but displayed more similarity to R-state PolyHb group trends.

Figure 2. Net analysis of macrophage secretion profiles.

Analysis of secretion profiles measured by multiplex immunoassay of macrophages treated with 2 mg/mL media, Hb, T-state PolyHb (30:1), R-state PolyHb (30:1), T-state PolyHb (35:1), and R-state PolyHb (35:1). (a) Clustergram showing differences in the secretion of cytokines/chemokines/growth factors from macrophages treated with Hb/PolyHbs. All results are presented as average fold changes (FCs) for each respective factor normalized to the media group measurement. Shades of red indicate an increase in the secretion, black indicates a negligible change in the secretion, and shades of green indicate a decrease in the secretion. Color brightness indicates the relative FC, as shown by the scale bar. The orange dashed line between IL-4 and IL-5 indicates two groups—above the line, the H group generally has opposite trends to PolyHbs and below the line, the H and PolyHb groups have similar trends. (b) Table with correlation values between treatment groups. Groups are listed in order of similarity to the H group in the first column. A value of +1 indicates exact correlated, a value of zero indicates no correlation, and a value of −1 indicates reverse correlation. (c) PC1 and PC2 values plotted for each treatment. (d) PC2 and PC3 values plotted for each treatment.

To further compare the effects of Hb/PolyHb treatments on macrophage secretion, correlation values between the groups were calculated. The first column of Fig. 2b lists the correlation of each PolyHb versus Hb group, in the order of highest to lowest correlation. Hb was not closely correlated to any of the PolyHbs, but was better correlated with the R-state PolyHbs (0.50–0.51) than the T-state PolyHbs (0.27–0.34). Th ese correlation values were much lower than those in the remaining columns (0.72–0.84), which compare PolyHbs with one another. This indicates that secretion profiles of PolyHb are generally more similar to one another than to that of Hb. This is not surprising, as PolyHbs are more physically similar to each other; regardless of polymerization in the R or T quaternary state, because PolyHbs have undergone chemical crosslinking procedures, unlike Hb. Consequently, PolyHbs have higher molecular weights, as multiple Hbs are bound together (Table 1). These physical and chemical changes may cause PolyHbs to have more similar interactions with macrophages than unmodified Hb, which is evident through the comparison of the resulting secretion profiles.

Table 1.

Physical properties of human hemoglobin and PolyHbs used in studies.

Property	Human hemoglobin (Hb)	35:1 R-state PolyHb	30:1 R-state PolyHb	35:1 T-state PolyHb	30:1 T-state PolyHb
Diameter (nm)	5.5	56.58	39.61	63.49	34.70
[Hb] (g/dL)	18.85	13.03	12.69	10.21	12.80
MetHb (%)	3.5	4.8	4.3	7.7	6.2
P50 (mmHg)	11.09	1.28	1.35	35.42	35.66
Cooperativity (n)	2.69	0.97	0.99	0.91	0.80

Principal component analysis on treatment secretion profiles

PCA was used to more systematically separate treatment groups based on their secretion trends. This method identifies which factor(s) contribute the most in generating different responses among the experimental groups. With these data, the first four PCs accounted for 97.8% of the variance of the data set (Supporting Information, Fig. 3a). Each PC is based on a linear combination of the inflammatory factor data, with each assigned a different weight/score (Supporting Information, Fig. 3b). In Fig. 2c, PC1 and PC2 values are plotted for each treatment. PC1 accounts for 71.9% of the variance of the data set and PC2 accounts for 16.2% of the same. Along PC1, H (PC1 = 0.30) slightly separates from PolyHbs, all of which have similar values (PC1 = 0.46–0.50). Both MIP-1α and IP-10 are factors that have high scores for PC1 (absolute values above 2), and their trends are opposite for Hb versus PolyHbs, suggesting that they contribute to this observed slight separation along PC1. In contrast, RANTES and IL-13 are the top two scoring factors for PC1, and all treatment groups have similar trends, leading to less separation along PC1. Separation is seen to a greater extent along PC2. H has a high positive value (PC2 = 0.9), both R:30 and R:35 have values close to zero, and T:30 and T:35 fall between −0.2 and −0.4, generally clustering by the state of Hb/PolyHb (unmodified, R-state PolyHb, or T-state PolyHb). IL-2 and GM-CSF are in the top eight cytokines that contribute most to the total PC2 value, and in the heatmap, exhibit similar trends within R-state and T-state PolyHbs, regardless of the polymerization molar ratio. These factors may contribute to the separation seen between the groups along PC2 and may be key macrophage markers that react differently with PolyHb quaternary state, possibly due to differences in oxygenation or binding with macrophage receptors.

Figure 2d plots PC2 against PC3 values (accounts for 5.8% variance). Along PC3, treatments separate by polymerization molar ratio. T:30 and R:30 have PC3 values greater than 0.3. In contrast, T:35 and R:35 have PC3 values less than −0.3. H is in-between, with a PC3 value close to zero. VEGF and MIP-1β are two top-scoring contributors to PC3. As seen in the heatmap, VEGF has similar trends when treated with PolyHbs at constant polymerization molar ratios, regardless of the quaternary state. MIP-1β also has similar values between T:30 and R:30. As they are more consistently regulated between the polymerization molar ratio, rather than PolyHb quaternary state, VEGF and MIP-1β secretion in response to Hb-based therapies may be more affected by molecular weight than oxygenation.

Taken together, PC1 represents the most variability of the data set, but only separates Hb/PolyHbs slightly. PC2 contributes to the second most variability in the data set, and the values cluster according to the quaternary state, whereas PC3, contributing to less variance, separates by the polymerization molar ratio. As discussed in the clustergram results, this supports the observation that the most obvious distinction in the data set is between Hb and PolyHbs as a whole (consistent with PC1). Next, it is more apparent that differences in macrophage secretion result from the PolyHb quaternary state (PC2), rather than from the polymerization molar ratio (PC3). The PC separation trends and the percentage of variance explained by each one provide an overarching commentary for the data set.

Significant effects of Hb/PolyHb on the secretion of key inflammatory factors from macrophages

Although the overall trends for the data set were identified in Fig. 2, there were also several key significant trends for specific inflammatory factors (Fig. 3). These trends are divided into three groups. Figure 3a shows results for factors in which Hb treatment results in lower secretion than PolyHbs. For TNF-α and IL-2, the H group resulted in levels that were significantly lower than treatment with any other PolyHb. For GM-CSF and IL-12, H treatment still resulted in the lowest secretion, but it was only significantly lower than T-state groups, and not the R-state groups. This supports the observation that Hb has more similar macrophage secretion trends to R-state PolyHbs than T-state PolyHbs. In the second group of factors, in Fig. 3b (IL-6 and IL-10), PolyHbs have significantly lower secretion than Hb. These cytokines contribute to the observation that PolyHbs generally separate from Hb. Lastly, in Fig. 3c, Hb and PolyHbs act similarly for RANTES and IL-13, by decreasing secretion compared to media. This suggests that, for these particular factors, chemical crosslinking does not affect secretion levels. Perhaps Hb and PolyHbs are similar enough in chemical and physical properties to result in similar secretion trends for pathways in which IL-13 and RANTES secretion are implicated. Overall, Fig. 3 shows inflammatory factors with significant trends that support conclusions made from the clustergram and PCA analysis presented in Fig. 2.

Figure 3. Significant inflammatory factor trends due to Hb/PolyHb treatment.

Key inflammatory factors revealed several significant trends when macrophages were treated with 2 mg/mL H, T:30, R:30, T:35, or R:35. (a) Four factors—TNF-α, IL-2, GM-CSF, and IL-12—in which several groups (media and PolyHbs) are significantly higher than H. + denotes a significant difference between all other groups. * denotes significance compared to H. # indicates significance between the bracketed groups. (b) Secretion levels for IL-6 and IL-10, in which values for H remain close to media baseline, but the remaining PolyHb groups are significantly lower than H. * denotes significance compared to H. (c) Secretion levels for RANTES and IL-13, in which several Hb/PolyHb groups are significantly lower than media. + denotes significance between all other groups. # denotes significance compared to media.

IPA modeling—biological disease and function predictions

Next, we used IPA, which is based on published trends on protein interactions and pathways, to interpret the secretion results and make predictions on the effects of Hb/PolyHbs within a biological/wound healing context. Interesting trends were identified in vascular and wound healing categories and are shown in Fig. 4. For vascular trends (Fig. 4a), T-state PolyHb is the only treatment that has a predicted increase for the migration of endothelial cells, cell movement of microvascular endothelial cells, and tubulation of vascular endothelial cells. The remaining treatments predict decreases for these processes, indicating that T-state PolyHb may promote vascularization, which is necessary for wound healing. These predictions are specifically based on increases in the secretion of PDGF, VEGF, IL-8, and GM-CSF in the T:30 group, as seen in the heatmap for the factors that are implicated in the migration of endothelial cells.

Figure 4. IPA Modeling of inflammatory factor data to predict biological outcomes.

(a) Predictions related to vascularization and (b) wound healing for each Hb/PolyHb treatment are shown on the orange and blue heatmaps. Shades of orange indicate a predicted increase in the function above media baseline, and shades of blue indicate a predicted decrease. The brightness of the blue/orange shade depends on the relative z-score of the prediction. These predictions (specifically for the migration of endothelial cells and migration of cells) are based on the factors listed in the red and green heatmaps. Shades of red indicate an increase in the secretion of the factor compared to media baseline, and shades of green indicate a predicted decrease.

Wound healing-related predictions for T:30 compared to other experimental groups include the most migration of cells, cell viability, and growth of connective tissue (Fig. 4b). These predictions are based on a majority of the cytokines measured in the data set; for example, for the migration of cells, the prediction was based on 25/27 of the cytokines measured. A decrease in a majority of the inflammatory factors due to Hb treatment led to a strong predicted decrease in the migration of cells. Increases in some of these factors in the T:30 group led to a less extreme prediction for the migration of cells. In addition to favorable IPA predictions for T:30, this group also resulted in the lowest levels of intracellular ROS in macrophages, which would be desirable for lowering inflammation in chronic wounds (Supporting Information, Fig. 4). To follow up on these predictions, the next step was to test the Hb/PolyHb treatments in an in vivo wound healing model, with our hypothesis being that T-state PolyHb would lead to faster wound closure and more angiogenesis than those observed for other experimental groups.

Effect of Hb/PolyHb in in vivo murine wounds

The in vivo wound healing study investigated the effect of different Hb-based formulations (nonpolymerized H, R:30, and T:30) versus vehicle controls (Ringer’s lactate) on diabetic mouse wounds. Excisional wounds were created on the backs of mice; 200 μL of 100 mg/mL of Hb/PolyHb solution was topically applied, and the wound was covered with Tegaderm™ wound dressing. Treatment was reapplied once a week for 4 weeks. Images of the wound area were taken on Days 0, 3, and 7 and then weekly until Day 35 until all wounds were closed.

T-state trended toward faster closure of wounds than R-state, H, and control groups throughout the entire 35 days of the study (Fig. 5). Th is was significant on Day 21, when T-state treated wounds were significantly smaller than those of all other groups. Throughout the study, H and R-state mice exhibited slower healing, with similar wound closure curves. Figure 6 shows Day 35, H&E stained, histological sections of uninjured skin and wound areas that had received various treatments. In the uninjured skin, nicely formed, mature epidermal, dermal, fat, and muscle layers were identified. In the wounded sections, the layers are more difficult to separate, as they are newly regenerated. Th e epidermal layer was bluish/purple and had less folds and surface area than the uninjured epidermis. The neodermis was compact, with newly formed extracellular matrix (pink) and infiltrated cells (blue dots) that have filled the wound bed. No appendages (hair follicles, sebaceous glands) were present in the neodermis. Qualitatively, the epidermis of R- and T-state PolyHb-treated mice appeared thicker than that of the control and Hb-treated mice. Epidermal thickness was measured and quantified in ImageJ and normalized to that of the uninjured skin. Treatment with R- or T-state PolyHbs resulted in the highest epidermal thicknesses, which was significant compared to that for uninjured skin. The T-state PolyHb group was also significantly higher than the control group. H and C groups had lower, but similar, epidermal thicknesses, between 1.5 and 2 times higher than that of the uninjured skin.

Figure 5. Effect of Hb/PolyHbs (H/R:30/T:30) on wound closure in vivo as compared to vehicle-treated controls (C).

Percent wound closure as a function of time, representing the wound closure rate for each treatment. * denotes significantly higher wound closure of T versus all other groups on Day 21. The images below are representative of wound size for each treatment group on Days 0, 7, 14, and 21.

Figure 6. Histological sections from Day 35 mice stained with H&E.

The epidermal thickness of skin was measured using ImageJ. Results were normalized to the epidermal thickness of the uninjured skin. * denotes significance versus uninjured skin. # denotes significance between the indicated groups.

CD31 staining was also performed on Day 35 histological sections (Fig. 7). CD31 is a marker for endothelial cells, indicating blood vessel formation. Qualitatively, CD31 staining of nonpolymerized Hb and R-state PolyHb treatment groups appeared wider than clusters observed in groups treated with T-state PolyHb. Quantitatively, mice treated with T-state PolyHb had significantly higher CD31 density than control, Hb, and R-state PolyHb groups; on average, T-state groups had approximately two times higher CD31⁺ density than other groups. As predicted in IPA, T-state PolyHb also exhibited the most benefits in vivo, in terms of angiogenesis and wound healing (CD31 density and epidermal thickness). Th is may be attributed to increased levels of PDGF, VEGF, IL-8, and GM-CSF detected from T-state PolyHb-treated macrophages in the in vitro studies.

Figure 7. CD31 staining on Day 35 histological sections to indicate blood vessel formation.

Arrows point to positive staining (red/brown areas). Results are quantified in the graph. * denotes significance versus controls (C). # denotes significance between the indicated groups.

CONCLUSIONS

These studies laid the groundwork for the investigation of the effects of PolyHbs in inflammatory conditions, particularly in relation to macrophages and chronic wound healing. Overall, Hb is toxic to macrophages at concentrations >20 mg/mL, unlike PolyHbs. Hb reduced the secretion of a majority of proteins on an inflammatory panel, whereas the effect of PolyHbs was less drastic and more comparable to one another across quaternary states and polymerization molar ratios. Th e inflammatory secretion trends of Hb exhibited more similar trends to R-state PolyHb rather than to T-state PolyHb. IPA analysis identified T-state PolyHb as having the secretion profile most likely to stimulate angiogenesis and wound healing. This was confirmed in an in vivo study, where CD31 density and epidermal thickness were the highest in mouse wounds topically treated with T-state PolyHb. Future work should include the incorporation of PolyHbs into a wound dressing material such as a hydrogel to test therapeutic effects in a clinically relevant delivery system. Furthermore, additional controlled studies should be performed to better understand the results in this in vitro system, specifically, comparing the extent of oxygen delivery, as well as heme release from Hb versus PolyHbs. It is also worthwhile to elucidate whether the HO-1 pathway is affected by Hb versus PolyHb treatment. These studies would provide further insight on whether the observed differences in macrophage cytokine secretion profile are due to differences in Hb/PolyHb oxygen delivery, physical properties, or HO-1 activation. Overall, this work showed that PolyHbs are less toxic to macrophages than Hb and that chemical modifications of Hb can affect inflammatory macrophage secretion, which can ultimately affect wound healing.