Research article

Acellular haemoglobin attenuates hypoxia-inducible factor-1α (HIF-1α) and its target genes in haemodiluted rats

Dominador J. Manalo, Paul W. Buehler, Jin Hyen Baek, Omer Butt, Felice D'agnillo, Abdu I. Alayash


Hb (haemoglobin)-based blood substitutes represent a class of therapeutics designed to correct oxygen deficit under conditions of anaemia and traumatic blood loss. The influences of these agents on HIF-1α (hypoxia-inducible factor-1α) target genes involved in adaptation to hypoxia have so far not been studied. In the study presented here, rats underwent 80% ET (exchange transfusion) with either HS (hetastarch) or a polymerized Hb OG (Oxyglobin®). HS induced dramatic EPO (erythropoietin) gene transcription, reaching a maximum at 4 h post-ET. In contrast, OG suppressed EPO transcription until approx. 24 h post-ET. Large plasma EPO levels that were observed post-ET with HS were significantly blunted in animals transfused with OG. OG, unlike HS, induced a sharp increase in HO-1 (haem oxygenase-1) transcription at 4 h, which declined rapidly within 24 h, whereas modest increases in iNOS [inducible (nitric oxide synthase)] and constitutive NOS [eNOS (endothelial NOS)] were detected over the control. Our results demonstrate for the first time that severe haemodilution-induced erythropoietic responses in kidneys were attenuated by a low-oxygen-affinity cell-free Hb and suggest that tissue-specific oxygen-sensing pathways can be influenced by allosterically modified Hbs.

  • erythropoiesis
  • haematocrit
  • haemodilution
  • haemoglobin (Hb)
  • kidney
  • hypoxia-inducible factor-1α (HIF-1α)


Hundreds of mammalian genes are known to be regulated by dynamic changes in oxygen tension [1,2]. As the master regulator of oxygen homoeostasis, HIFs (hypoxia-inducible factors) have been characterized for their important functional role for mammals in the transcriptional regulation of major pathway genes that mediate the adaptive responses to hypoxia and oxidative stress, such as anaemia and high altitude [3,4]. During hypoxia, HIF forms α–β heterodimers that bind DNA consensus sequences called HREs (hypoxia-response elements) at its target genes [5,6]. The HIF-1β subunit, also known as ARNT (aryl-hydrocarbon nuclear transporter), is a constitutively expressed nuclear protein in almost all cells. However, there are three known isoforms of HIF-α subunits: HIF-1α, HIF-2α and HIF-3α; the stability and activity during hypoxia of HIF-1α and HIF-2α are regulated by the dynamic ‘oxygen-sensing’ activity of PHDs (prolyl hydroylases) [79]. PHDs function to hydoxylate proline residues within the ODD (oxygen degradation domain) of HIF-α proteins, which targets each towards pVHL (von Hippel–Lindau protein) ubiquitination and proteasomal degradation during normoxia [1012]. Physiologically, HIF-αs function to mediate the adaptive responses to tissue hypoxia and appear to serve non-redundant, cell- and tissue-specific roles in essential processes that include erythropoiesis, aerobic respiration, energy metabolism and glycolysis, as well as in haem homoeostasis, antioxidant defence, vascular tone and angiogenesis [3,4].

A dynamic balance between oxygen-carrying capacity of RBCs (red blood cells) and tissue oxygen-sensing mechanisms of the PHD/HIF system exists in mammalian tissues, which appears to modulate energy metabolism with aerobic respiration [13]. There has been remarkable progress in delineating the molecular basis of tissue and cellular oxygenation, particularly in oxygen-mediated regulation of activated HIF-1α gene transcription [14,9]. Acute anaemic haemodilution created by exchanging 50% of rat blood with non-oxygen-carrying pentastarch was recently shown to sufficiently induce hypoxia and subsequently lead to increased expression of HIF-1α, eNOS [endothelial NOS (nitric oxide synthase)] and VEGF (vascular endothelial growth factor) in the cerebral cortex of these mildly anaemic rats [14]. Oxygen release by control fresh RBCs and RBCs treated with an allosteric modifier of Hb (haemoglobin) oxygen affinity, ITPP (myo-inositol trispyrophosphate), were recently compared regarding their ability to deliver oxygen to cultured human MECs (microvascular endothelial cells) under normoxic and hypoxic conditions. RBCs loaded with ITPP were found to release more oxygen than their normal counterparts, and that the levels of VEGF and HIF-1α, elevated in the human MECs under hypoxia, were dramatically reduced or even suppressed in the presence of ITTP-loaded RBCs [15].

In a recent study, we have reported for the first time a correlation between the oxygenation/oxidation states of cell-free Hb and renal HIF-1α-binding activity in a rat and guinea-pig model of 50% blood replacement [16]. Using a more severe model of haemodilution designed specifically to induce hypoxia, the present study tests the hypothesis that an 80% ET (exchange transfusion) in rats with HS (hetastarch), a non-oxygen-carrying volume expander, will activate HIF-regulated/hypoxia-induced gene expression, in an adaptive response to severe haemodilution in kidney tissues. Alternatively, a cell-free Hb stabilized in a low-oxygen-affinity state will reverse these responses in proportion to its oxygen-carrying properties and maintained ferrous iron state in vivo. For this study, we utilized a well-investigated glutaraldehyde-polymerized bovine Hb, OG (Oxyglobin®), an FDA (Food and Drug Administration)-approved oxygen therapeutic for use in dogs with anaemia that is chemically similar to the human analogue, Hemopure™, currently under clinical evaluation as a blood substitute [16,17]. Selected HIF-regulated target pathway genes were analysed as potential biomarkers for efficacy and safety, and they included EPO (erythropoietin) for erythropoiesis and HO-1 (haem oxygenase-1) for both haem degradation and antioxidant/anti-inflammatory defences. We also measured eNOS and iNOS (inducible NOS) respectively for both vascular tone and inflammatory responses. Renal transcriptional responses were profiled by real-time PCR over the course of 72 h post-ET with either HS or OG. We report that a low-oxygen-affinity cell-free Hb, unlike HS, attenuates HIF target gene activity by either suppressing specific genes (e.g. EPO, iNOS and eNOS) or enhancing others (e.g. HO-1) and that this activity is dependent on the ability of the Hb to retain its redox-active iron in the ferrous form to bind and deliver oxygen to hypoxic tissue.


Animals and surgical preparation

Male Sprague–Dawley rats and Hartley guinea-pigs were purchased from Charles River Laboratories (Wilmington, MA, U.S.A.) and acclimated for 1 week upon arrival at the FDA/CBER animal care facility. All animals were fed with normal diets throughout the acclimation period, and they weighed 350–400 g at the time of study. Animal protocols were approved by the FDA/CBER Institutional Animal Care and Use Committee with all experimental procedures performed in adherence to the NIH guidelines on the use of experimental animals. On days of surgery, rats were anaesthetized via the intraperitoneal route with a cocktail of ketamine HCl (100 mg/kg) and xylazine HCl (5 mg/kg) (Phoenix Scientific, St. Joseph, MO, U.S.A.). Under aseptic conditions, a midline incision was made around the neck region allowing for blunt dissection and exposure of the right common carotid artery and the left external jugular vein. Saline filled catheters containing 50 IU (international units) of heparin/ml prepared from sterile PE50 tubing (Clay Adams; Becton Dickinson, Sparks, MD, U.S.A.) were placed in each vessel and tunnelled under the skin to the back of the neck. Immediately after surgeries, animals were administered a subcutaneous dose of buprenorphine (0.1 mg/kg; Reckitt and Coleman, Kingston-Upon-Hull, U.K.) and allowed 24 h of recovery prior to experimentation. Fully conscious and freely moving rats (n=40) underwent an 80% ET, replacing blood with either OG or HS. Arterial and venous catheters were extended, tethered and connected to separate syringe pumps (Model 11 Harvard Apparatus, Holliston, MA, U.S.A.) set on withdrawal (1 ml/min) and infuse (1 ml/min) respectively. The 80% ET volume in the rat was calculated using the equation 80% ET (ml)=[0.06 (ml/g)×body weight (g)+0.77]/2 (see [18]). Plasma from heparinized blood that was withdrawn from each transfused animal was obtained to determine the total Hb content. A separate group of sham control animals (n=5) underwent surgical catheter implantation and 24 h recovery. Therefore these animals were treated as all other animals in the study, but were absent during the process of ET. At the end of recovery, this group of animals was killed and tissues were collected (as described in the following section). Our sham control group thus represents the initial time point (zero time or baseline) to which group and time statistical comparisons are made. A separate group of guinea-pigs was exchange transfused and plasma samples were collected as described in the subsection ‘Blood/tissue collection and preparation’ below. Ascorbate is a primary antioxidant in circulation that maintains extracellular Hb in its reduced oxygen-carrying form. The guinea-pig, unlike the rat, represents a small animal non-ascorbate-producing species and thus serves as a positive control in this study to demonstrate OG oxidative instability.

To validate our model of haemodilution-induced hypoxia, rats (n=3 per group) were exchange transfused with HS at 70, 80 and 90% of their total BV (blood volume). For establishing the maximum level of ET achievable with a non-oxygen carrier, 72 h survival was the determining factor. This study revealed that a maximum 80% controlled ET with a non-oxygen carrier allowed for a model with 100% survival up to 72 h after haemodilution (results not shown).


OG was purchased from Biopure Corp. (Cambridge, MA, U.S.A.). This solution consists of a heterogeneous mixture of glutaraldehyde-polymerized bovine Hb prepared at a concentration of 13 g/dl in modified Lactated Ringer's (Oxyglobin™ package insert, Biopure). A detailed description of the physicochemical properties of the mixture as a whole and each individual fraction has been described elsewhere [19,20]. Functionally, OG exhibits a less sigmoidal OEC (oxygen equilibrium curve) that saturates at a much higher oxygen tension than normal 2,3-DPG (2,3-diphosphoglycerate)-stripped Hb or whole blood. The P50 (oxygen affinity at which Hb is half-saturated) of this Hb has been reported by us to be ∼46.0 mmHg in phosphate buffer, and when measured in whole animal (rat) blood immediately after infusion, it was reported to be approx. 36.4 mmHg (see [16]). HS (6%; HESpan®) was purchased from B. Braun Medical (Irvine, CA, U.S.A.). HS is a high-molecular-mass starch (number average=600 kDa) that maintains volume expansion greater than whole blood but equal to OG in duration. Additionally, both solutions exert similar colloid oncotic pressures at their respective concentrations [21].

Blood/tissue collection and preparation

A 350 μl portion of blood was sampled before ET, immediately after ET, and then at the time of killing (4, 24, 48 and 72 h post-ET) to assess Hct (haematocrit) levels and to isolate plasma for the determination of EPO, total Hb (oxy/deoxy), ferrous (oxy) and ferric (met) Hb. At the end of each time point (0, 4, 24, 48 and 72 h), animals were anaesthetized via intravenous administration of Euthasol (Delmarva Laboratories, Des Moines, IA, U.S.A.) and immediately perfused with 0.9% normal saline solution. Kidneys were harvested at the same time points, sliced medially and placed in either formaldehyde solution for histopathology or RNALater Solution (Invitrogen) to optimize for RNA stability, or snap-frozen in liquid N2 for both cytosolic and nuclear extraction protocols. Each time point represents n=5 animals.

MS of plasma samples

Plasma samples (10 μl) were desalted using C18 ZipTips (Millipore, Bedford, MA, U.S.A.) according to the manufacturer's instructions. A 1 μl aliquot was pipetted on to a stainless steel MALDI-MS (matrix-assisted laser-desorption ionization MS) sample plate and mixed with 1 μl of sinapinic acid (3,5-dimethoxy-4-hydroxycinnamic acid; Sigma Chemical Co., St. Louis, MO, U.S.A.) saturated in 50% acetonitrile/0.1% TFA (trifluoroacetic acid). The sample/matrix was air-dried and analysed on a PerSeptive Biosystems DERP MALDI—TOF (time-of-flight) mass spectrometer calibrated manually with purified human serum albumin (Sigma Chemical Co.) and using Voyager 5.1 software with Data Explorer (Applied Biosystems, Framingham, MA, U.S.A.) operated in linear mode.

Plasma EPO detection with ELISA

Plasma EPO levels of each rat in the present study were detected by ELISA with an EPO kit (R&D Systems, Minneapolis, MN, U.S.A.) and conducted according to the manufacturer's protocol with the following exception. A 1:4 dilution of rat plasma was necessary to maintain accurate measurement of EPO within the standard curve range for human EPO protein after HS treatment. Absorbance (A) was measured using a microplate reader at 405 nm wavelength (Tecan GENios™, Research Triangle Park, NC, U.S.A.).


HIF-1α accumulation in rat kidney sections was assessed by a polyclonal HIF-1α antibody (1:100 dilution; Santa Cruz Biotechnology) incubated overnight at 4 °C. Detection was performed using the Vectastain Universal Elite ABC kit according to the manufacturer's instructions (Vector Laboratories, Burlingame, CA, U.S.A.). Sections were washed, counterstained with haematoxylin, washed and dehydrated in graded ethanol and xylene. Photomicrographs were obtained using an Olympus IX71 inverted microscope (Olympus America, Melville, NY, U.S.A.) equipped with an Olympus DP70 digital camera. Microscope settings and software were standardized and applied during image capture and processing.

Nuclear/cytosolic extraction

Nuclear and cytosolic extracts were prepared from frozen kidney by Dounce homogenization using Buffer A [10 mM Hepes, pH 7.9, 10 mM KCl, 0.4% Igepal-690, 0.1 mM EDTA, 1 mM DTT (dithiothreitol), 1 mM PMSF, 1.2 mM Na3VO4 (sodium orthovanadate) and protease inhibitor cocktail solution (Roche Diagnostics)] on ice. Samples were incubated on ice for 30 min, and then centrifuged at 7826 g for 10 min at 4 °C. Supernatants were collected as cytosolic extracts. Nuclear pellets were resuspended in Buffer C [20 mM Hepes, pH 7.9, 420 mM NaCl, 1 mM EDTA, 25% (v/v) glycerol, 1 mM DTT, 1 mM PMSF, 1.2 mM Na3VO4 and protease inhibitor cocktail solution]. Samples were then incubated on ice for 30 min, snap-frozen on dry ice and thawed on ice. Samples were centrifuged at 7826 g for 15 min at 4 °C. All supernatants were collected into aliquots and stored at −80 °C prior to analysis.

Western blots for HIF-1α

Western blots for HIF-1α accumulation were performed using kidney nuclear extracts (120 μg) on PVDF membranes using a goat polyclonal HIF-1α antibody by R&D Systems (1:100 dilution). HRP (horseradish peroxidase)-based protein levels were carried using the ECL®-Plus™ Western Blotting Detection System (Amersham, GE Healthcare).

HIF-1α DNA binding assay

HIF-1α DNA binding activity was measured using an ELISA-based method (Transbinding HIF-1α Assay kit) from Panomics (Redwood City, CA, U.S.A.) according to the manufacturer's specifications. Briefly, half kidney sections (∼100–200 mg) were homogenized (Tekmar, Cincinnati, OH, U.S.A.) using the Nuclear Extraction kit (P/N 12494; Panomics, Freemont, CA, U.S.A.). Protein concentrations for extracts were measured using the Quant-iT™ Protein Assay kit (Q33210; Molecular Probes, Carlsbad, CA, U.S.A.).

RNA isolation/cDNA synthesis

Total RNA was isolated from rat kidney sections (∼200–250 mg each, cut medially) in 5 ml of TRIzol® Extraction (Ambion) after tissue homogenization (Tekmar). Total RNA was washed and further purified using RNeasy® mini spin columns (Qiagen Sciences, MD, U.S.A.). Purified total RNA (100 μg) from rat kidney cDNA (2 μg) was then synthesized with the Applied Biosystems High Capacity cDNA Reverse Transcription kit. Nucleic acid samples for RNA and DNA purity were based on A260/A300 spectral reading ratios of 1.80 and 2.00 respectively and were obtained from an ND-1000 spectrophotometer (NanoDrop Technologies).

Real-time PCR

Real-time PCR reactions containing 100 ng of cDNA each were performed in triplicate using the ABI (Applied Biosystems) TaqMan® Gene Expression Assay system and its TaqMan® Fast Universal PCR Master Mix, according to the manufacturer's protocol (ABI, Foster City, CA, U.S.A.). Fluorescence detection was achieved using an ABI Model 7900HT, and measurements were analysed with the ABI version 2.3 SDS (Sequence Detection Systems) software. The following ABI ‘inventoried’ rat-specific TaqMan® gene probes for HIF target genes were used for real-time PCR gene expression assays: Rn99999916_s1 (GAPDH), Rn01481376_m1 (EPO), Rn00561387_m1 (HO-1), Rn00561646_m1 (iNOS) and Rn02132634_s1 (eNOS).

Pharmacokinetic analysis

Plasma from heparinized blood was obtained to determine the total Hb (OG) content during the ET period. Blood samples (350 μl) were obtained from the arterial catheter prior to infusion (baseline), at the end of ET and prior to killing (0, 4, 24, 48 and 72 h). Plasma concentrations of ferrous OG and ferric OG were determined using a photodiode-array spectrophotometer (model 8453; Hewlett–Packard, Palo Alto, CA, U.S.A.). Plasma from the baseline (pre-ET) sample for each animal was used to correct for background interference and turbidity. Concentrations of ferrous OG (oxy/deoxy), ferric OG and hemichromes were determined using a multicomponent analysis based on the molar absorption coefficients for each Hb species [22].

Data presentation and statistics

Real-time PCR experiments for each gene were performed on each rat in triplicate and are normalized to rat GAPDH (glyceraldehyde-3-phosphate dehydrogenase) gene expression values to determine mean Ct (threshold cycle) values. Gene expression measurements were used to determine actual Ct values, while final fold change values were calculated based on the log(2) difference in mean ▵Ct values between treated and non-treated groups [–log2(▵▵Ct) values]. Results are presented as fold change and are expressed as means±S.E.M. (n=4–5 rats per group). Statistical analysis real-time PCR was performed on raw ▵Ct data. Within-group differences for all comparisons were determined by ANOVA with a post hoc analysis for determination of differences between groups. Significance was set at P≤0.05 using JMP™ Statistical Discovery software, version 5.1 (SAS Institute, Cary, NC, U.S.A.).


Figure 1(A) demonstrates an Hct reduction of approx. 80% after ET with both OG and HS. From 4 h onwards, Hct levels increased steadily to a maximum approx. 2-fold increase at 72 h by the end of ET. These results demonstrate an augmented erythropoietic effect that was further supported by an increase in spleen-to-body weight ratios of 0.416±0.187 and 0.477±0.028, in OG- and HS-transfused animals respectively (results not shown), a nearly 2-fold increase in mass compared with sham control rats (0.232±0.008).

Figure 1 Hct levels (A) and redox status of acellular Hb in plasma (B, C and D) at time points before and after ET

(A) Percentage Hct: the closed bars represent OG and the open bars represent HS-transfused animals. (B) Plasma OG concentrations: ●, total OG; ■, ferrous (oxy/deoxy) OG; ▲, ferric (met) OG. ‘*’ indicates the significance (*P<0.05) from the end of ET [time point 0.5 in (A) and zero time in (B, C)]. All concentration values are expressed as haem concentration (μM×103). (C) MALDI-MS spectra of plasma samples collected at baseline and after ET until 72 h. Each spectrum is normalized to 100% intensity to enhance changes that may be masked by lower abundance due to elimination. Albumin is indicated in the baseline (BL) sample at m/z 66313 [M+H]+1. ‘–α’ indicates the absence of the α-globin chain that should occur at m/z 15074 [M+H]+1, but is involved in cross-linking (α–α) at m/z 30990 [M+H]+1. ‘β’ indicates the β-globin chain observed at m/z 16092 [M+H]+1 that is not involved in covalent intra- or inter-molecular cross-linking. No changes in the spectra can be assigned over the 72 h period post-ET, indicating stability of the protein in vivo over the 72 h time course. (D) The positive control MALDI-MS spectra of plasma samples collected from guinea-pigs at the end of ET and at 48 h post-ET. The time point of the maximum percentage of ferric (Fe3+) OG collected is represented by the 48 h trace and demonstrates extensive accumulation of modified α-globin chains (m/z 15250 [M+H]+1) and modified β-globin chains (m/z 16125 [M+H]+1) not observed in the initial plasma sample, which, similar to rat plasma, demonstrated no α-globin chains and a β-globin chain (mass ions at m/z 16093 [M+H]+1).

The iron oxidation states of OG from rat plasma samples were measured post-ET, since only ferrous (Fe2+) but not ferric (Fe3+) forms of Hb can bind and deliver oxygen. Figure 1(B) shows that the maximum concentrations (Cmax) of plasma total iron (oxy+deoxy+ferric) OG (3663.7±162 μM haem) and ferrous iron OG (3000.0±173 μM haem) decline similarly over time with a half-life (t1/2=ln2k) of 18.5 h. The Cmax of ferric OG (130.61±57.0 μM haem) remains consistently low (5–10%) in circulating plasma, as previously demonstrated in ascorbate-producing rodents [16]. RBC Hb had significantly increased from baseline (HS: 2132.8±135 μM haem; OG: 2303.4±265 μM haem) by 72 h post-ET to more than 2-fold for both HS and OG groups: 4478.9±310 μM and 4137.6±284 μM haem respectively (results not shown). To further illustrate the redox stability of OG in the rat, plasma samples were collected over 72 h. Since oxidative destabilization of OG αβ-globin subunits and haem was previously reported to influence HIF binding activity [23], we evaluated plasma samples for changes in circulating OG that could be attributed to (i) protein degradation/oxidation and (ii) haem iron oxidation. Plasma samples for the globin chain region and the cross-linked globin chain region were evaluated in both rat and guinea-pig and are shown in Figures 1(C) and (D). The spectra are normalized to 100% intensity to magnify changes in the protein as OG is eliminated over time. In Figure 1(C), OG tetramers and polymers are cross-linked between α-globin chains. Therefore the β-globin chains are visible in the mass spectra (m/z 16093 [M+H]+1) along with the cross-linked α–α-globin chains that ionize poorly at m/z 30990 [M+H]+1. Accumulation of bovine Hb α-globin chain at m/z 15074 [M+H]+1 is an indication of cross-linked protein breakdown that is associated with haem loss, as we have reported previously [20]. However, this was not observed, indicating that OG remains stable in rat plasma during the course of the experiment. Conversely, in Figure 1(D), the initial guinea-pig plasma sample post-exchange indicates the absence of α-globin chains similar to rat plasma at the same time point. Gradually, OG oxidized in the guinea-pig's circulation (results not shown) until 48 h post-transfusion when circulating plasma OG reached a maximum of 40% oxidized haem iron (Fe3+). MALDI-MS analysis of the 48 h plasma sample demonstrates extensive accumulation of modified α-globin chains (m/z 15250 [M+H]+1) and modified β-globin chains (m/z 16125 [M+H]+1). In addition, haem release was monitored and compared with the mass ion of haemin-spiked plasma m/z 616.7 [M+H]+1. No corresponding mass ions were observed in plasma samples and all plasma sample mass spectra post-ET were found to be identical with baseline plasma mass spectra (results not shown). Therefore OG appears to have undergone minimal oxidative damage at both the protein and haem levels in the rat.

Immunohistochemical analyses for HIF-1α and HIF-2α in rat kidney were then performed 4 h post-ET. Sparse levels of HIF-1α (Figures 2A and 2D) and HIF-2α (results not shown) were detectable in cortical regions of sham control rats. In contrast, HS treatment produced a more pronounced accumulation of both HIF-1α (Figures 2B and 2E) and HIF-2α (results not shown) relative to sham control rats. Interestingly, OG treatment appeared to blunt HIF-1α (Figures 2C and 2F) and HIF-2α (results not shown) accumulation relative to HS-treated rats. Although only cortical regions of rat kidney are shown, detection for HIF-1α and HIF-2α revealed ubiquitous staining with HS in medullary regions as well (results not shown).

Figure 2 Immunohistochemistry analysis of HIF-1α accumulation in the rat kidney cortical region at 4 h post-ET in sham control (A, D), HS- (B, E) and OG- (C, F) treated rats

Low-magnification images (left panels, ×10 objective) show an overall increase in HIF-1α staining in HS-treated rats compared with sham and OG-treated rats. Higher magnification images (right panels, ×40 objective) reveal increased nuclear localization of HIF-1α in HS-treated rats. Haematoxylin was used as the nuclear counterstain (blue).

Next, we analysed HIF-1α by Western blotting and an ELISA-based assay for HIF-1α HRE transbinding activity using rat kidney nuclear extracts (Figures 3A and 3B). Western blots for representative kidney tissue (Figure 3A) at time points 0, 4 and 24 h provide the most reliable comparison between HS and OG, since circulating OG is maximal over the initial 24 h. HIF-1α protein expression compared with control animals was increased 2.4-fold (4 h) and 3.7-fold (24 h) after HS transfusion and 0.3-fold (4 h) and 0.5-fold (24 h) after OG transfusion. HIF-1α transbinding activity, when normalized to the sham control group, reached a peak at 24 h post-ET for both HS- (179.5%±58.6 S.D.) and OG- (98.3%±29.7 S.D.) treated rats. Notably, HS provoked greater HIF-1α transbinding activity compared with OG at each corresponding time point up to 72 h.

Figure 3 HIF-1α protein (A) and HIF-1α/HRE transbinding activity (B) from rat kidney nuclear extracts

(A) Western blot of HIF-1α control (zero time), 4 h and 24 h after ET. HIF-1α/actin fold change values, relative to control, were obtained by using densitometry and are shown below individual lanes. The vertical dotted line indicates Western-blot image cropped from the same gel. (B) ELISA-based HIF-1α transbinding activity (Panomics; see the Materials and methods section) was performed from rat kidney nuclear extracts (20 μg) and are presented as percentage change differences between HS- and OG-treated rats over control rats (baseline, sham). Values for HIF-1α activity are normalized for protein. HS (open bars) and OG rats (closed bars) are compared for HIF-1α transbinding activity. ‘*’ indicates significance from baseline or sham control animals (*P<0.05) and ‘†’ indicates significance between each treatment group HS compared with OG (*P<0.05).

As the principal hormone that mediates haematopoietic responses to hypoxia, EPO is the best characterized mammalian gene that is regulated by HIF activity. As illustrated in Figure 4(A), HS treatment resulted in a dramatic induction of EPO gene expression that reached a peak at 4 h (27.5-fold over baseline control). Although EPO gene expression began to decrease after 24 h, expression remained high up to 72 h (6.8-fold, over baseline control). OG transfusion, however, completely inhibited this response during the first 4 h. However, after 24 h, EPO gene expression significantly increased at 24 h (9.5-fold over baseline control) and reached a peak at 48 h (17.8-fold over baseline control). As shown in Figure 4(B), circulating plasma EPO levels mimicked the behaviour of mRNA EPO, increasing steadily and reaching maximum fold increase in circulating plasma EPO of 230±116-fold over baseline at 48 h post-HS transfusion, then declining towards baseline levels at 72 h post-transfusion. Plasma EPO levels increased in a similar yet diminished pattern after OG transfusion, reaching maximum fold increase in circulating plasma EPO at 48 h of 69±29-fold greater levels over baseline EPO concentrations.

Figure 4 Renal EPO gene expression profile (A) and plasma EPO levels (B) at time points before and after ET

(A) EPO gene expression profiles in rat kidney at 80% ET from 0 to 72 h with either OG (●) or HS (◯) treatments. EPO gene expression profiles were determined through real-time PCR with either OG (●) or HS (◯) treatments. Gene expression levels were determined through real-time PCR and are presented as relative fold change over untreated control rats (zero time, fold change=1.0) (see the Materials and methods section); ‘*’ indicates significance from baseline or sham control animals (*P<0.05) and ‘†’ indicates significance between each treatment group HS compared with OG (*P<0.05). (B) Rat plasma EPO levels as assayed by ELISA (R&D Biosystems) between HS (open bars) and OG (closed bars) groups. EPO levels (pg/ml) are shown as means+S.E.M. Sample size for each group, n=5. ‘*’ indicates significance from baseline or sham control animals (*P<0.05) and ‘†’ indicates significance between each treatment group HS compared with OG (*P<0.05).

Next, we examined the temporal changes in HO-1 mRNA levels in both HS- and OG-treated animals (Figure 5A). The levels of mRNA were markedly increased immediately after infusion with OG. They were maximally increased between the end of transfusion and the 4 h time point (288-fold increase over baseline). HO-1 then decreased rapidly after 4 h to approximately baseline levels by the end of the first 24 h. In the HS group, however, there was a 50-fold increase over sham controls, which were normalized by 24 h post-transfusion.

Figure 5 (A) HO-1, (B) iNOS and (C) eNOS gene expression profiles in rat kidney at 80% ET from 0 to 72 h with either OG (●) or HS (◯) treatments

Gene expression levels were determined through real-time PCR and are presented as relative fold change over untreated control rats (zero time, fold change=1.0) (see the Materials and methods section). ‘*’ indicates significance from baseline or sham control animals (*P<0.05) and ‘†’ indicates significance between each treatment group HS compared with OG (*P<0.05).

Inflammatory responses to HS or OG ET was further inferred by iNOS gene expression up to 48 h (Figure 5B). Interestingly, both HS and OG significantly induced maximal iNOS gene expression after 4 h at 13.4- and 6.5-fold respectively. Furthermore, no significant differences in iNOS expression between HS and OG treatment at 4 h were observed. Vasodilatory responses were inferred by eNOS (Figure 5C) gene expression. In contrast with iNOS, peak eNOS gene expression over baseline occurred at 24 h for both HS (5.4-fold) and OG (3.4-fold); however, only HS-treated rats induced significant levels of eNOS gene expression from 4 to 48 h.


Much of our current understanding of oxygen-sensing mechanisms is derived from biochemical analyses and tissue-culture observations. The existence of several levels of controls over the expression of action of target gene products in the cells of intact organisms are lacking in cell-culture-based models, which may not reflect fully the complex physiology of the whole animal in both health and in disease state [24]. In the present study, we investigated the interplay between blood oxygen-controlling and tissue oxygen-sensing mechanisms and subsequent gene activation under acute severe anaemia conditions in rats, with some emphasis on how changes in allosteric and redox properties of infused chemically stabilized Hb may influence this relationship.

We have previously established the utility of rats over guinea-pigs in monitoring oxygenation, oxidation and toxicokinetics of cell-free Hb in an anaemia model of 50% ET [16]. Rats, unlike guinea-pigs and humans, are able to endogenously maintain sufficient levels of ascorbic acid in circulation and are able to control Hb oxidation and the loss of its function, thus allowing us to monitor these events within a defined timeframe [16]. We have also demonstrated that HIF-1α protein activity in rat kidney correlated inversely with circulating ferrous (functional) Hb over the first 50 h, which confirmed the kidney as a primary sensing organ that responds to hypoxia with HIF activation [25]. In the present study, using the same Hb (OG) in an 80% ET model, we demonstrate that HIF activity can become attenuated by Hb within 24 h relative to its allosteric, in vivo oxidation and pharmacokinetic properties, as reflected by genes whose proteins mediate adaptive responses to RBC loss.

EPO is the principal haematopoietic hormone produced in the adult kidney [26], which not only mediates mammalian erythropoiesis, but also serves diverse roles for non-haematopoietic tissues including cell/tissue survival [27,28], immunity [29] and ventilatory respiration [30]. Further evidence has shown that HIF-1α and HIF-2α subunits regulate hypoxia-inducible genes, as an adaptive response in a selective, non-redundant and cell-specific manner [2,24]. HIF-1α was the first HIF-α subunit to be discovered based on cell-culture studies using a human EPO–HRE enhancer region; however, recent reports suggest that HIF-2α may principally regulate EPO gene expression in a tissue- or cell-type-specific manner [2,24,31]. Immunohistochemistry staining of HIF-1α protein accumulation was observed at 4 h post-ET (Figures 2A–2F). This observation was supported by both Western blotting and HIF-1α transbinding activity following HS and OG transfusion (Figures 3A and 3B). Immunohistochemistry staining for HIF-2α protein accumulation was also confirmed (results not shown). Importantly, the patterns of both proteins qualitatively paralleled both EPO gene expression and EPO plasma levels, as measured by PCR and Western blot/ELISA methods respectively. Rats that are exposed to altitude develop severe anaemia, primarily demonstrating splenic RBC production, which is characterized by increased Hct and body-to-spleen weight ratios [32]. Thus the observation of increased spleen weights in both OG- and HS-transfused animals, coupled with increased Hct and RBC Hb, suggests that early and rapid erythropoiesis is occurring in both groups of animals.

HO-1 plays a major role against inflammatory events [33,34], antioxidant defence [35], cell survival [36] and anaemia [37], including in kidney tissue in rat models of hypoxia [38,39]. HO-1 catalyses the rate-limiting step in haem oxidation whose by-products include biliverdin and bilirubin, which release iron and carbon monoxide. HIF-1α has been shown to mediate activation of HO-1 transcription in kidney [40]. HO-1 gene expression is tightly regulated and complex and is shown to be orchestrated by several transcriptional regulators including Nrf2 (nuclear factor erythroid 2 p45 subunit-related factor 2) [35], NF-κB (nuclear factor κB) [41,42] and HSF-1 (heat-shock factor 1) [43]. The robust induction of HO-1 expression by OG, which is 6-fold greater than HS at 4 h (Figure 5), is probably mediated through separate regulatory mechanisms, whereby the iron haem group of OG activates haem metabolism and antioxidant pathways via the Nrf2 pathway and HS activates haem iron recycling pathways against anaemia after global hypoxia via HIF-1α. The dramatic expression of HO-1 at 4 h in rats transfused with OG (Figure 5A) is therefore not surprising considering that both modified and unmodified Hbs in different oxidation states have been shown to elicit high levels of HO-1 both in cell culture and in animals [23,44]. The lower levels of post-injury circulating pro-inflammatory cytokines such as IL-6 (interleukin-6) and IL-8 in human subjects that were infused with a polymerized human Hb have been attributed to the fact that HO-1 may have reduced the systematic inflammatory responses, and interventions to induce its expression may be therapeutic [45].

HIFs regulates a battery of vasoactive protein genes including the iNOS [46] and eNOS [47] respectively. Induction of iNOS (Figure 5B) and the constitutively expressed eNOS (Figure 5C) was also observed in both animal groups. These results suggest that each may be biologically active in the kidney of rats. However, as iNOS gene expression was induced by both HS and OG, this could also indicate consequential effects from 80% ET, rather than indicating an inflammatory response by OG. eNOS activity is intimately involved in the physiological regulation of vascular tone and blood flow distribution into hypoxic regions, whereas the inducible form is strongly up-regulated in response to inflammatory stimuli. Yet in spite of the well-documented NO scavenging of HBOCs (Hb-based oxygen carriers) in animals that result in acute vasoconstriction [48], our results indicate no major impediment to tissue oxygenation by OG, as inferred from the attenuation of HIF-1α activity from immunohistochemical staining, transbinding activity and target gene expression in the first 24 h.

In conclusion, ‘cross-talk’ between an oxygen carrier and the HIF transcriptional machinery has clearly been established in our model of severe haemodilution. Our results demonstrate that erythropoietic (EPO) effects increased in response to acute haemodilution with HS. These responses, however, were suppressed by allosterically stabilized Hb, which correlated with its oxygen-carrying capacity (P50) and iron haem oxidative state. However, OG, unlike HS, triggered a transient but considerable activity in HO-1, which is in large part related to haem metabolism and anti-inflammatory events rather than oxygen-carrying activity. These observations not only provide a better understanding and validation of homoeostatic control by an oxygen carrier under these conditions, but also additional insights into potential therapeutic applications with redox-sensitive, allosterically modified Hbs and/or RBCs. Relative to the rate and extent of HBOC stability, clearance and redox properties, when compared with native acellular Hb, the utility of these products could be explored in the context of characterizing pathway processes in whole animals. HBOCs are a class of potentially useful therapeutic agents that may provide treatment modalities in surgery, trauma and anaemia, such as sickle cell disease, when whole blood is unavailable. For example, HBOCs under clinical development possess unique biochemical properties that include auto-oxidative, nitrosative and peroxidative processes and broad oxygen-carrying capacities (oxygen P50) that range from 5.0 mmHg (high O2 affinity) to 40 mmHg (low O2 affinity) [49]. The present study indicates that HBOCs may be tailored to function in a specific fashion in disease states and conditions where modulation of HIF activity and regulation of key pathway target genes may be needed. This may include disease state management, where use could include ischaemic tissue preconditioning, antitumorigenesis or adaptations to hypoxia such as high-altitude acclimation [50], which we are currently investigating.


This work was supported in part by FDA/CBER critical path funding and by a grant from the Defense Advanced Research Projects Agency. The findings and conclusions in this paper have not been formally disseminated by the FDA and should not be construed to represent any agency determination or policy.

Abbreviations: DTT, dithiothreitol; NOS, nitric oxide synthase; eNOS, endothelial NOS; EPO, erythropoietin; ET, exchange transfusion; Hb, haemoglobin; HBOC, Hb-based oxygen carrier; Hct, haematocrit; HIF, hypoxia-inducible factor; HO-1, haem oxygenase-1; HRE, hypoxia-response element; HS, hetastarch; IL, interleukin; MALDI-MS, matrix-assisted laser-desorption ionization MS; MEC, microvascular endothelial cell; iNOS, inducible NOS; ITPP, myo-inositol trispyrophosphate; Nrf2, nuclear factor erythroid 2 p45 subunit-related factor 2; OG, Oxyglobin®; PHD, prolyl hydroylase; RBC, red blood cell; VEGF, vascular endothelial growth factor


View Abstract