Polymerized human hemoglobin facilitated modulation of tumor oxygenation is dependent on tumor oxygenation status and oxygen affinity of the hemoglobin-based oxygen carrier

Administration of hemoglobin-based oxygen carriers (HBOCs) into the systemic circulation is a potential strategy to relieve solid tumor hypoxia in order to increase the effectiveness of chemotherapeutics. Previous computational analysis indicated that the oxygen (O2) status of the tumor and HBOC O2 affinity may play a role in increased O2 delivery to the tumor. However, no study has experimentally investigated how low- and high-affinity HBOCs would perform in normoxic and hypoxic tumors. In this study, we examined how the HBOC, polymerized human hemoglobin (PolyhHb), in the relaxed (R) or tense (T) quaternary state modulates O2 delivery to hypoxic (FME) and normoxic (LOX) human melanoma xenografts in a murine window chamber model. We examined microcirculatory fluid flow via video shearing optical microscopy, and O2 distributions via phosphorescence quenching microscopy. Additionally, we examined how weekly infusion of a 20% top-load dose of PolyhHb influences growth rate, vascularization, and regional blood flow in the FME and LOX tumor xenografts. Infusion of low-affinity T-state PolyhHb led to increased tissue oxygenation, decreased blood flow, decreased tumor growth, and decreased vascularization in hypoxic tumors. However, infusion of both T-state and R-state PolyhHbs led to worse outcomes in normoxic tumors. Of particular concern was the high-affinity R-state PolyhHb, which led to no improvement in hypoxic tumors and significantly worsened outcomes in normoxic tumors. Taken together, the results of this study indicate that the tumor O2 status is a primary determinant of the potency and outcomes of infused PolyhHb.

Due to their ability to modulate O 2 delivery from the circulatory system, hemoglobin-based oxygen carriers (HBOCs) are promising O 2 therapeutics that may increase the effectiveness of chemotherapy [6][7][8][9][10][11][12] . However, HBOCs are still not clinically approved despite decades of research 13,14 . The elevated renal toxicity and hypertension associated with previous generations of commercial HBOCs have mainly been attributed to this delay in development. Current improvements in reactor design and product purification are now able to exclude the low molecular weight (MW) HBOC fractions (< 250 kDa ) that contributed to the toxicity of early generation HBOCs [15][16][17][18] .
The chemical modifications that are necessary to make HBOCs safe for infusion typically alter the HBOC O 2 affinity 16,17,[19][20][21] . For example, conjugating polyethylene glycol to the surface of hemoglobin (Hb) results in an increased O 2 affinity (P 50 : 5-6 mm Hg) compared to unmodified Hb 22,23 . Alternatively, commercial polymerized Hbs (PolyHbs) prepared via glutaraldehyde or O-raffinose crosslinking in the tense (T) quaternary state all have significantly lower O 2 affinity (P 50 : 30-40 mm Hg) compared to unmodified Hb [24][25][26] . Recently, we demonstrated that the O 2 affinity of PolyHbs could be controlled by polymerizing the Hb under fully oxygenated or deoxygenated conditions 16,17,21 . By fully oxygenating the Hb during polymerization, the PolyHb is effectively locked into the high O 2 affinity, relaxed quaternary state (R-state). Alternatively, polymerizing the Hb while deoxygenated locks the PolyHb into the low O 2 affinity, T-state.
Previously, we computationally determined that alterations in the O 2 affinity of the HBOC impact how much O 2 an HBOC will supply to the surrounding tumor tissue 20,27 . From the results of these studies, we anticipate that R-state PolyHb and T-state PolyHb should increase O 2 delivery to hypoxic tissue. Whereas in normoxic tissue, T-state PolyHb should increase oxygenation. To date, no study has examined how infusing an HBOC may impact the growth and vascularization of hypoxic and normoxic tumors. Previous studies only examined hypoxic tumors or a single class of HBOC [28][29][30][31][32][33] . Understanding how both normoxic and hypoxic tumors respond to different modes of enhanced O 2 delivery is vital to defining how these materials would be applied clinically.
In this work, we prepared high MW polymerized human Hbs (PolyhHbs) in either the T-or R-state. We then observed how each of these materials modulate O 2 transport within hypoxic (FME) and normoxic (LOX) human melanoma xenografts. By using the LOX melanoma cell line, we are able to generate normoxic tumors to compare with the hypoxic tumors 34,35 . By using phosphorescence quenching microscopy (PQM) on the microcirculatory environment observed within window chamber models, we explored how PolyhHbs modulate blood flow and O 2 transport in vivo. Additionally, we examined how weekly infusions of PolyhHb impact tumor growth and vascularization. With these results, we are better able to define how the oxygenation status of the tumor and the O 2 affinity of an HBOC will impact modulation of O 2 delivery.

Results
Biophysical properties of polyhHbs. The resulting properties of the PolyhHbs used in this study are shown in Fig. 1 tissue o 2 tension. Changes in tissue O 2 tension in both hypoxic (FME) and normoxic (LOX) tumors are depicted in Fig. 2. The average tissue pO 2 of hypoxic (FME) tumors was significantly lower than the average tissue pO 2 in normoxic (LOX) tumors and host tissue. Infusion of 30:1 R-state PolyhHb significantly decreased tissue oxygenation in both normoxic (LOX) and hypoxic (FME) tumors compared to baseline and following infusion of 35:1 T-state PolyhHb. In contrast, infusion 35:1 T-state PolyhHb led to significant increases in average tissue pO 2 in both tumors.

Microhemodynamics.
Changes in the arteriolar and venular blood flow in both hypoxic (FME) and normoxic (LOX) tumors are depicted in Fig. 3A, B. Vessel diameters were separated into two groups for this analysis: one group with diameters less than 30 μm and another group with diameters greater than 30 μm. In hypoxic tumors, infusion of 30:1 R-state PolyhHb led to a significant reduction in the volumetric flow rate in small diameter ( D ves < 30 μm) venules. Infusion of 35     tumor vasculature. We were also interested in experimentally observing how periodic infusions of PolyhHb solutions would alter properties associated with microvascular mass transport. Regional blood flow (RBF) was analyzed with fluorescent microsphere perfusion, and microvascular density (MVD) was estimated via tissue histology. The results of these studies are shown in Fig. 5B, C. In hypoxic FME tumor xenografts, infusion of 35:1 T-state PolyhHb led to a significant decrease in both RBF and MVD. In normoxic LOX tumor xenografts, infusion of T-state PolyhHb led to a significant increase in RBF to the tumor compared to baseline.

Discussion
The principal finding of this study is that the O 2 status of tumors has a strong effect on the effects of PolyhHb coadministration. These results may help explain some of the previous negative results that occurred in rhabdomyosarcomas 28 .
In  www.nature.com/scientificreports/ primarily increased MAP. These increases in MAP is consistent with previous HBOCs containing low MW species 36,37 .
In normoxic tumors, the hemodulution effect of PolyhHb infusion led to significant increases in tumor perfusion. Despite this increase in blood perfusion after the infusion of R-state PolyhHb, the average tissue pO 2 decreased. This decrease is likely a result of the relatively low amount of O 2 extracted from R-state PolyhHb (2.2 ± 0.1% ), which failed to make up for the lack of O 2 carrying capacity from the diluted blood. In comparison, 18.7 ± 4.7% of the total available O 2 from T-state PolyhHb was extracted, which offset the O 2 extraction from both Hb in RBCs and dissolved O 2 in plasma.
In hypoxic tumors, the percentage of O 2 extracted from 30:1 R-state PolyhHb only slightly increased (12.8 ± 1.8%) compared to the percentage of O 2 extracted from 35:1 T-state PolyhHb (37.5 ± 2.6%). This slight increase in O 2 extraction from R-state PolyhHb does not fully offset the O 2 demand from O 2 dissolved in plasma and Hb in RBCs. When the decrease in O 2 supply from R-state PolyhHb is coupled with reduced blood flow resulting from vasoconstriction, overall O 2 delivery is significantly reduced. This requirement is instead offset by significant increases in the O 2 extracted by the dissolved O 2 in the plasma and Hb in RBCs.
In this study, we found that the infusion of PolyhHb led to relatively minor changes in the growth rate of the hypoxic (FME) tumors. However, it appears that this low dosage was unable to replicate the significant reduction in tumor growth (40%) that was previously observed in a triple-negative breast cancer model 20 . This is likely because the dose volume and dose frequency were too low to result in an appreciable effect on tumor growth. For this study, the PolyhHb was delivered weekly; however, previous studies of the pharmacokinetics of similar PolyHbs indicate that these PolyhHbs have a half-life of only 24 h 15 . Taking this into account, the tumors were only exposed to the O 2 modulating effect of PolyhHb for only 25% of the week. Increasing the dosing frequency to once every 2 to 3 days may increase the relative effect.
Unfortunately, infusion of both the T-state and R-state PolyhHb solutions led to a significant increase in tumor growth for normoxic LOX tumors. Infusion of 30:1 R-state PolyhHb led to a 40% increase in tumor volume after the 14-day treatment regime. This is likely because both 35:1 T-state PolyhHb and 30:1 R-State PolyhHb   www.nature.com/scientificreports/ have higher O 2 affinity compared to mouse Hb in RBCs (P 50 = 42 mm Hg). Because of this relative increase in O 2 affinity, we anticipate that less O 2 may be delivered under normoxic conditions, which could decrease host cell survival in the tumor periphery. We anticipate that when applied to a model that more accurately represents human physiology, T-state PolyhHb might decrease the tumor growth rate due to its lower O 2 affinity compared to human Hb in RBCs (P 50 = 26 mm Hg). However, we may also observe further increases in tumor growth due to the increased supply of O 2 to normoxic tumors. Despite observing a growth delay after infusion of 30:1 R-state PolyhHb, infusion of R-state PolyhHb did not lead to significant decreases in RBF or MVD. This is likely because the low dose frequency and high O 2 affinity of R-state PolyhHb were insufficient to trigger an anti-angiogenic response in FME tumors. Baseline values for RBF in hypoxic FME tumors [0.16 ± 0.02 mL/(min g)] and normoxic LOX tumors [0.15 ± 0.02 mL/(min g)] are at the upper range of the values measured for other tumors experimentally [38][39][40] . In the hypoxic (FME) tumors, we observed decreases in MVD after delivery of 35:1 T-state PolyhHb. This decrease in vessel formation indicates an increase in O 2 delivery. In contrast, we observed a significant increase in MVD after weekly infusions of 30:1 R-state PolyhHb in normoxic (LOX) tumors. This is consistent with a decrease in O 2 delivery, which may lead to more aggressive tumor growth and increased angiogenesis. In fact, the MVD of the normoxic LOX tumor after infusion of 30:1 R-state PolyhHb is remarkably similar to the measured MVD within the baseline hypoxic (FME) tumor. This further supports the notion that within the normoxic tumor, R-state PolyhHb is not adequately delivering O 2 , which is in agreement with microvascular simulations performed previously 20 . Despite this, reduction in tumor growth has been previously observed after infusion of high O 2 affinity HBOCs [41][42][43] . Therefore, this decrease in tumor growth may be due to other factors in the environment including production of reactive O 2 species (ROS) 44 and nitric oxide (NO) scavenging 45 . This faster rate of metHb formation will lead to the increased production of ROS which can induce oxidative injury to the tumor mass. This is especially important to consider given that HBOCs can scavenge NO 46 and can oxidize and produce ROS 47 species in vivo. Future studies should investigate these mechanisms in more detail by directly observing changes in hypoxia inducible factors and downstream proteins when working with high O 2 affinity HBOCs.

conclusions
The results from this study indicate that low-dose, infrequent infusions of R-state PolyhHb is not suitable for oxygenating both hypoxic and normoxic melanomas. In general, treatment of normoxic tumors with either high-or low O 2 affinity PolyhHbs aggravated tumor growth and angiogenesis. In contrast, T-state PolyhHbs significantly increased O 2 supply to hypoxic tumors. These results encourage the use of low O 2 affinity PolyhHbs with reduced cooperativity to hypoxic tumors. Additionally, this further emphasizes the need to fully characterize how different tumor types respond to modulating O 2 delivery with HBOCs.

Methods
polymerized hemoglobin synthesis and analysis. Human Hb (hHb) used in these studies was first purified from human red blood cells (RBCs) as described previously 48 . PolyhHb was produced using methods described previously 17 . In brief, the resulting hHb solution was polymerized with glutaraldehyde while fully oxygenated or deoxygenated to form either R-state or T-state PolyhHb, respectively. The resulting PolyhHbs were first clarified on a 0.2 μm hollow fiber filter. After clarification, the PolyhHb solutions were diafiltered on a 100 kDa hollow fiber filter into a modified Ringer's lactate buffer to remove the low MW PolyhHb/hHb species. The cyanomethemoglobin method was used to measure the Hb concentration and the metHb level of hHb/ PolyhHb solutions 49,50 . The size distribution of PolyhHb, by particle volume, was measured using dynamic light scattering (DLS) (Brookhaven Instrument Inc. BS-200M, Holtsville, NY). The O 2 -hHb/PolyhHb equilibrium binding curves were measured using a Hemox Analyzer (TCS Scientific Corp., New Hope, PA). The hHb/ PolyhHb kinetics of O 2 offloading ( k off ,O 2 ) were measured with an Applied Photophysics SF-17 microvolume stopped-flow spectrophotometer (Applied Photophysics Ltd., Surrey, United Kingdom) using protocols previously described by Rameez and Palmer [51][52][53] . The MW distribution was estimated using an Acclaim SEC-1000 column (Thermo Scientific, Waltham, MA) on a Thermo Scientific Dionex Ultimate UHPLC system using previously described methods 17,54 . Dorsal chamber window model. Adult female 8-to 10-week old female BALB/c-nu/nu mice were used for the xenografted tumors according to protocols approved by the University of California San Diego Animal Care and Use Committee. Mice were instrumented with dorsal chamber windows as described previously 55,56 . This experimental model is an excellent system to observe O 2 delivery as it is highly sensitive to changes in O 2 supply 57 . Additionally, this model is especially useful when examining changes in microvascular pO 2 distributions in the unanesthetized state. Human melanomas (FME and LOX) were initiated by implanting a 200-500 μm xenograft into the fascial side of the intact skin layer of the chamber window model. FME and LOX cells were generously donated by Micro-Target Dynamic Therapy (San Diego, CA). The FME and LOX human melanoma cell lines were originally developed by the Rofstad Group at the Institute for Cancer Research (Oslo University Hospital, Norway) 34,35 . After implantation of the xenografts, tumors were allowed to grow for 7 days before analysis. Mice were divided into two groups based on the implanted human tumor cell lines (FME or LOX). Each of these groups was further subdivided into three cohorts: (1) an unsupplemented baseline, (2) infusion of 30 www.nature.com/scientificreports/ then projected to a 4,815 charge-coupled device camera (Cohu Industries, Poway, CA). A LUMPFL-WIR × 40 numerical 0.8 aperture water immersion objective (Olympus, New Hyde Park, NY) was used to carry out the measurements. Mean arterial pressure (MAP) and heart rate (HR) were recorded using a pressure transducer connected to the femoral artery catheter using an MP-150 system (BIOPAC Systems, Goleta, CA). Arterial blood was collected in heparinized capillary tubes and immediately analyzed for PaO 2 , PaCO 2 , and pH using a 248 Blood Chemistry Analyzer (Bayer, Norwood, MA). Total Hb was measured spectrophotometrically using a B-hemoglobin Hemocue (Stockholm, Sweden). Plasma Hb was measured from plasma collected after the capillary tube was centrifuged.
Phosphorescence quenching microscopy (PQM). Phosphorescence quenching microscopy (PQM) was used to analyze the O 2 distribution in the tissue and vascular space, as described previously 58 . This highresolution method allows us to resolve the pO 2 of arterioles and venules within the growing tumor. To determine the pO 2 in this method, we measure the decay rate of the excited palladium-mesotetra-(4-carboxyphenyl) porphyrin (Frontier Scientific Porphyrin Products, Logan, UT) bound to albumin. We then used the measured fluorescence lifetime ( τ p ), fluorescence lifetime in the absence of O 2 ( τ p,0 ), and fluorescence quenching rate constant ( k q ) to calculate the pO 2 using the Stern-Volmer equation, as shown in Eq. (1) 59 .
The probe was injected intravenously 10 min before pO 2 distributions were measured to allow time for the phosphorescent probe to circulate and diffuse into the chamber window model. The exposed tissue within the chamber window was then excited with 420 nm wavelength light. To acquire τ p , the 680 nm emitted phosphorescence signal was collected. Because this method is relatively independent of the probe concentration, we were also able to measure extravascular tissue pO 2 .

Microvascular hemodynamics.
To observe changes in the arteriole and venule diameter, we used a video image shearing method to determine blood vessel diameter 60 . Center-line velocities of arterioles and venules were measured with a 102B Vista Electronics photo-diode velocity tracker (San Diego, CA) using a crosscorrelation method. Volumetric flow rate ( Q ) through the arterioles and venules was then calculated using the radius of the vessel ( r ves ) and average fluid velocity ( − v f ) , as described in Eq. (2). For these calculations, we assume that fluid velocity profile was parabolic in arterioles and venules.
We are also able to calculate the oxygen extraction fraction (OEF) from the various species ( tumor growth model. Similar to the chamber window model study, adult 8-to 10-week old female BALB/c-nu/nu mice were used for the xenografted tumors according to protocols approved by the University of California San Diego Animal Care and Use Committee. Approximately 4 × 10 5 cells of the human melanoma cell lines FME and LOX were injected into the mouse flank. Mice were divided upon the tumor cell lines (FME or LOX). Each of these groups was further subdivided into three cohorts: (1) an unsupplemented baseline, (2) infusion of 30:1 R-state PolyhHb. and (3) infusion 35:1 T-state PolyhHb. Mice were infused with the 100 mg/mL PolyhHb solutions via tail vein injection of 20% of the mouse's blood volume once each week during the study. During tumor growth, the length of the tumor ( L tumor ) and width of the tumor ( W tumor ) were both measured to estimate tumor volume as shown in Eq. (4).
Tumor blood flow. Fluorescently labeled microspheres were used to estimate tumor blood flow in tumors as described previously 61 . In short, 15 μm diameter fluorescent microspheres (Molecular Probes, Eugene, OR) were suspended in saline. 100 μL of this solution was rapidly injected into the animal via the tail vein. Arterial reference samples were simultaneously withdrawn at a constant rate of 100 μL/min for 1 min through an inserted femoral catheter. At the end of the protocol, the mice were euthanized with a lethal dose of sodium pentobarbital. Tumor tissue was then digested in 1 M KOH solution for 24 h. Fluorescent dye was extracted with Cellosolve (Fisher Scientific, Pittsburgh, PA). The fluorescent signal was then measured using an LS 50B luminescence spectrometer (PerkinElmer Corp., Norwalk, CT). Regional blood flow proportional to the fraction of cardiac output was calculated by measuring the number of fluorescent microspheres in the tumor tissue relative to the total in the arterial reference samples.