Haptoglobin Preferentially Binds β but Not α Subunits Cross-Linked Hemoglobin Tetramers with Minimal Effects on Ligand and Redox Reactions

Human hemoglobin (Hb) and haptoglobin (Hp) exhibit an extremely high affinity for each other, and the dissociation of Hb tetramers into dimers is generally believed to be a prerequisite for complex formation. We have investigated Hp interactions with native Hb, αα, and ββ cross-linked Hb (ααXLHb and ββXLHb, respectively), and rapid kinetics of Hb ligand binding as well as the redox reactivity in the presence of and absence of Hp. The quaternary conformation of ββ subunit cross-linking results in a higher binding affinity than that of αα subunit cross-linked Hb. However, ββ cross-linked Hb exhibits a four fold slower association rate constant than the reaction rate of unmodified Hb with Hp. The Hp contact regions in the Hb dimer interfaces appear to be more readily exposed in ββXLHb than ααXLHb. In addition, apart from the functional changes caused by chemical modifications, Hp binding does not induce appreciable effects on the ligand binding and redox reactions of ββXLHb. Our findings may therefore be relevant to the design of safer Hb-based oxygen therapeutics by utilizing this preferential binding of ββXLHb to Hp. This may ultimately provide a safe oxidative inactivation and clearance pathway for chemically modified Hbs in circulation.


Introduction
Haptoglobin (Hp) has been extensively studied as an hemoglobin (Hb) scavenging protein due to its naturally high binding affinity towards extracellular Hb in plasma [1]. The Hb-Hp protein complex is removed from circulation through the CD163 scavenger receptor on the surface of peripheral blood and tissue monocytes and macrophages [2,3,4]. Free Hb is cleared to prevent cellular and tissue damage caused by oxidative reactions mediated by its heme iron. There has been renewed interest in Hp as a potential therapeutic for indications involving acute and chronic hemolysis, and as an adjuvant for acellular Hb-based oxygen therapeutics [5].
Hp is an acute phase plasma protein that exists in three primary phenotypes: Hp 1-1, Hp 2-1, and Hp 2-2 [6]. A Hp monomer consists of one a subunit and one b subunit linked via disulfide linkage, and one Hp monomer binds with one Hb ab dimer. The Hp molecule consists of two types of a subunits, a1 (approximately 9 kDa) and a2 (approximately 16 kDa), and a single type of b subunit of Hp which weights approximately 45 kDa. The a1 subunit contains a single cysteine residue, while the a2 subunit contains an extra cysteine residue capable of forming multiple disulfide bonds. Therefore, Hp 1-1 is a dimer that consists of two a1 subunits and two b subunits. Hp 2-1 and Hp 2-2 contain a2 subunits to form polymers of different sizes and shapes. The Hb binding site on Hp has previously been mapped to the large b subunit [7].
It has been shown that all types of Hp bind oxygenated Hb almost irreversibly with the equilibrium dissociation constant (K d ) reportedly as low as 10 212 M or 10 215 M [8,9]. Purified Hb subunits can form much weaker complex with Hp, whereas the deoxyHb remains mostly as tetramers and appears to display no binding to Hp. The Hb dimer associates rapidly with Hp under oxygenated conditions, and Hb dimer formation appears to be essential to Hb-Hp binding [10].
In the absence of antioxidant enzymatic system within red blood cells, acellular Hb undergoes a series of oxidative reactions generating potentially toxic reactive oxygen species (ROS) such as superoxide (O 2 N-), hydrogen peroxide (H 2 O 2 ). H 2 O 2 is one of the well-examined Hb pro-oxidants that react with ferrous Hb (Fe 2+ ) in a two-electron transfer process to produce the oxo-ferryl (Fe 4+ = O 22 ) Hb species. When it reacts with met Hb (Fe 3+ ), a protein radical ( N HbFe 4+ = O 22 ) is formed. Both of these higher oxidation states of Hb can cause redox side reactions, damaging Hb itself as well as other nearby proteins and lipids. These reactions have yet to be fully understood, but they are believed to originate from the highly reactive heme group. Radicals are thought to migrate via tyrosine residues to other amino acids [11]. As a consequence, irreversible oxidative modifications of specific amino acids, e.g., Trp15, Met55, Cys93, Cys112 and Tyr145 in the b subunits have been consistently observed in the presence of oxidants such as H 2 O 2 . Cross-linkages between Hb globin chains and heme which can trigger further oxidative reactivity have also been reported to occur [12].
Nitrite, another powerful oxidant, reacts with oxyHb in a complex process of multiple reaction steps influenced by Hb quaternary state intrinsic oxygen and ligand reactivity, and heme accessibility [13]. Nitrite also reacts with deoxyHb to generate metHb and NO. The latter enzyme-like function may have some physiological implications in the regulation of blood vessel dilation and blood pressure control. Hp has been demonstrated to enhance nitrite reductase activity of deoxyHb by approximately ten-fold, much like that of Hb dimers [14,15]. Moreover recent studies have shown that Hp impedes radical formation, or stabilizes the damaging radicals formed in the presence of H 2 O 2 , which led us and others to explore the protective mechanisms of Hp more fully for possible therapeutic applications [5].
Hp interactions with several chemically modified Hbs have been previously evaluated using fluorometric methods and by surface plasmon resonance (SPR) analysis [16]. This analysis consistently revealed a strong affinity of Hp for Hbs, but a much lower affinity for internally cross-linked Hbs. Furthermore, these studies suggest that the aa subunits cross-linked Hbs may have more reduced affinity to Hp.
The Hp binding with modified and unmodified Hbs was examined closely in the present study using SEC-HPLC, native gel electrophoresis, and stopped-flow kinetic techniques. The functional effects of Hp binding on oxygen dissociation and ligand binding to internally cross-linked, non-dissociable tetrameric Hb were studied in comparison with that of unmodified Hb. The redox reactions of Hb with typical oxidants such as H 2 O 2 and nitrite were assessed in order to determine the impact of specific chemical cross-linking and Hp binding of these reactions. Our results suggest that bb subunit cross-linked tetrameric Hb retains substantial Hp binding capacity which does not alter ligand binding or redox reactions beyond that observed with chemical modification alone. This can have an important application in the design of safe cross-linked and/or polymerized hemoglobin-based blood substitutes. These results are also discussed in the light of a recently published crystal structure of the porcine Hp-Hb complex [17]. A. Representative chromatograms of Hb, aaXLHb, and bbXLHb in the presence or absence of Hp. The absorbance was monitored at 405 nm for heme, and the peaks appearing between the elution time 13 and 19 minutes are Hb and Hp complexes. B. The percentage of Hp bound with Hb was determined and compared among aaXLHb, bbXLHb, and Hb based on baseline normalized area under curve for peaks within the boxed region of the eluting profiles (from Panel A). Statistical analysis was performed using a One-way analysis of variance with a Bonferroni's Multiple comparison test to evaluate between group differences. Data are represented as mean areas +/2 sem, significance was set at p,0.05. All analysis were performed using GraphPad Prism software. doi:10.1371/journal.pone.0059841.g001

Materials
Human HbA was prepared as previously described via ammonium sulfate precipitation and anion-exchange FPLC, and stripped of endogenous organic phosphate cofactors [18]. Both aa subunits cross-linked Hb (aaXLHb) and bb subunits cross-linked Hb (bbXLHb) by bis(3,5-dibromosalicyl)fumarate were prepared as previously described [19], and provided by Research Institutes of the United States Army (Washington, DC). The Hp sample containing primarily dimers (Hp 1-1) and to a lesser extent polymers (Hp 2-1, and Hp 2-2) was purified [20], and kindly provided by Bio Products Laboratory (BPL, Hertfordshire, UK). All chemicals and reagents were purchased from Sigma Aldrich (Saint Louis, Missouri) or Fisher Scientific (Pittsburgh, Pennsylvania) unless indicated otherwise, and gases were purchased from Roberts Oxygen Company, Inc. (Rockville, Maryland) or Matheson Tri Gas (Basking Ridge, New Jersey).

HPLC and Gel Electrophoresis Analyses
Hp was mixed in excess with Hb (Hp:Hb, in a 2:1 ratio) to allow for the complete binding of Hb to Hp. The extent of Hb binding to Hp was measured by size exclusion chromatography on a BioSep-SEC-S3000 column ((600 mm67.5 mm), Phenomenex, Torrence CA) attached to a Waters 2535 quaternary gradient module and 2948 photodiode array detector (Waters Corporation, Milford, MA). Each run was normalized using a 50 ml injection loop. The mobile phase consisted of 50 mM potassium phosphate buffer, pH 7.4, at 22uC, pumped at a flow rate of 1.0 mL/minute over a 30 min run time. An injection loop flush step was performed prior to each run using 500 mL of mobile phase. The percentage of Hb bound to Hp was determined by dividing the area of Hb-Hp peaks (15-17 min elution time) at 405 nm by the total area of Hb containing peaks (15-21 min elution time) multiplied by 100.
The molecular weight compositions of the Hb, Hp, and Hb-Hp samples were assessed by native PAGE using an Invitrogen NovexH Minigel System and NativeMark TM Unstained Protein Standard (Carlsbad, CA). A 4-16% precast NativePAGE TM Novex Bis-Tris gel was employed, and each lane was loaded with 3-5 mL of sample solutions (Invitrogen NativePAGE TM sample prep kit) containing about 100 mM Hb and Hp in excess. The gel was run at 120 V for 90 minutes on a PowerEase 500 Power Supply. After electrophoresis, the gel was stained overnight with Coomassie blue G250 stain buffer, and then destained in water.

Stopped-flow Fluorescence Measurement
The rapid reaction of Hb and Hp was monitored in an Applied Photophysics microvolume stopped-flow spectrophotometer

Rapid Kinetics of Ligand Reactions
The kinetics of oxygen dissociation from oxy Hb or the Hb-Hp complex, and binding of carbon monoxide (CO) to deoxy Hb or the Hb-Hp complex were measured in an Applied Photophysics microvolume stopped-flow instrument as previously described [21,22]. Hb solutions (30 mM in heme) were rapidly mixed with an equal volume of 1.5 mg/mL sodium dithionite, and the absorbance changes of the oxygen dissociation process were monitored at 437.5 nm in 50 mM Bis-Tris buffer at pH 7.4 at room temperature. The CO binding kinetics were measured at 437.5 nm in the same instrument and buffer containing freshly made 1.5 mg/mL sodium dithionite. The CO solution was prepared by saturating the degassed buffer with a flow of prewashed CO gas. For each reaction, at least three kinetic traces were averaged and fit to exponential equations using the Marquardt2Levenberg fitting routines included in the Applied Photophysics software.

Nitrite-and H 2 O 2 -induced Hb Oxidation
The spectral changes of the Hb reaction with nitrite were measured in an Agilent 8453 diode array spectrophotometer in the presence or absence of Hp. Rapid mixing methods using Applied Photophysics microvolume stopped-flow instrument were used to measure the kinetics of nitrite oxidation reaction of Hbs with and without Hp. Hb solutions (30 mM) were mixed with equal volumes of freshly prepared nitrite solution (6 mM) in potassium phosphate buffer, pH7.4, to initiate oxidation, and the absorbance changes were monitored as a function of time. The reaction time courses were recorded at a single wavelength, 577 nm, and repeated at least three times for each reaction condition.
The oxidation of met Hbs by excessive amount of H 2 O 2 was performed under pseudo-first-order conditions using Applied Photophysics microvolume stopped-flow spectrophotometer with a diode array detector at 25uC, as described previously [23]. Ferric (met) Hb solutions (20 mM) with and without excess Hp in 0.5 mM Tris buffer, pH 7.4, were rapidly mixed with H 2 O 2   solutions of increasing concentrations up to 100 mM. The absorbance spectral changes (at least 200 spectra) were recorded as a function of time. The whole set of spectral data were subjected to global curve fitting analysis (Applied Photophysics software) to derive the reaction rate constants of the oxidation reaction of met Hb to ferryl Hb. The second-order rate constants were obtained from the dependence of the apparent rate constants on H 2 O 2 concentrations.

The Measurement of Ferryl Hb Intermediate Formation and Decay
The intermediate formation of ferryl Hb in the reaction of Hb or Hb-Hp with H 2 O 2 was detected by its reaction with sodium sulfide (Na 2 S), generating sulfonated Hb (sulfHb) that is spectrally detectable [24]. In a typical experiment, metHb (50 mM) was mixed with 1:5 molar ratio of H 2 O 2 in 50 mM potassium phosphate buffer, pH 7.4, to initiate the oxidative reaction in a cuvette monitored in an Agilent 8453 diode array spectrophotometer. After a 2 minute incubation time, Na 2 S (2 mM) was added to the reaction mixture and spectral changes between 450 nm and 700 nm were recorded for the conversion of ferryl Hb to sulfHb. The sulfHb concentrations were calculated using the extinction coefficient of 20.8 mM 21 cm 21 at 620 nm [24].

Results
Purified human Hb, aaXLHb and bbXLHb, and complexs of each protein with Hp were analyzed using both analytical sizeexclusion chromatography (SEC) and native gel electrophoresis methods. SEC consistently revealed the major Hb peaks at the elution time of 20.9 min (monitored at 405 nm) in all Hb samples ( Figure 1A). Three main peaks were observed between elution times 13 min and 18 min for the HbA sample with Hp, in agreement with the previously reported results [25]. The peak eluting at 17.3 minutes represents HbA bound to Hp 1-1, while Hb binding to polymeric species (Hp 2-2 or Hp 2-1) were observed with predominant peaks at 15.5 and 16.4 minutes, respectively. A slightly different elution profile with additional peaks eluting between 13 min and 16 min was observed for the mixture of bbXLHb and Hp, indicating further molecular association and the formation of larger molecular weight complexes. Only small peaks could be detected in the same elution time frame for the mixture of aaXLHb and Hp. Since Hp was added in excess, 99.6% 60.4% purified human Hb sample was bound with Hp and no free Hb peak was observed. It was determined from peak % area calculations within the boxed region that bbXLHb preserved approximately 52.9% 64.1% of the binding capacity with Hp relative to unmodified Hb, whereas aaXLHb and Hp showed minimal binding with less than 15.8% 63.5% including nonspecific bindings and low levels of contaminants ( Figure 1B).
The SEC results were verified and confirmed on a native PAGE (Figure 2). The unmodified Hb was resolved largely into Hb dimers and some tetramers (Lane 2) as contrasted to the native gel molecular markers (Lane 1). The samples of aaXLHb (Lane 3) and bbXLHb (Lane 4) appeared primarily as tetramers, as a result of the chemical cross-linking and tetramer stabilization processes, with some minor protein impurities that may not be Hb species, and were not observed in the SEC-HPLC results. The Hp sample (Lane 8) was resolved primarily as Hp dimers (Hp 1-1), with some Hp polymer species, i.e., Hp 2-1 and Hp 2-2. In the presence of slight excess Hp (Lane 5), the formation of additional protein bands, and the disappearance of free Hb indicated the binding between Hp and unmodified Hb. The Hp dimers, and Hp polymers composed of Hp 2-1 and Hp 2-2 can form multiple protein complexes with native Hb to give extra bands. Conversely, relatively few new protein bands were observed in the case of the aaXLHb and Hp sample (Lane 6), which could represent a merge of the two samples run separately in Lanes 3 and 8. However, protein complexes of high molecular weight polymers formed between Hp and bbXLHb can be seen (Lane 7) with some bbXLHb tetramer also visible. This observation suggests a moderate binding capability, consistent with the SEC data. In addition, Hb dimers bind to Hp with strong monovalent binding characteristics, and bbXLHb binds to Hp with polyvalent binding characteristics. Our previous data suggest that bbXLHb polyvalent binding with Hp leads to a time dependent gelation phenomenon [16]. These binding characteristics also contribute to the observation of multiple band patterns observed in lanes 5 and 7.
Hb binding to Hp is accompanied by the quenching of Hp intrinsic fluorescence as previously reported [10,26]. We measured the kinetics of Hb and Hp binding by monitoring fluorescence changes in a rapid mixing stopped-flow spectrophotometer equipped with a fluorescence detector. Representative time courses of Hp (1 mM) association with 15 mM Hb, aaXLHb, or bbXLHb under pseudo first order conditions are shown in Figure 3A. The interaction of Hb and Hp was fitted to a single exponential equation. No reaction was observed between aaXLHb and Hp under the same experimental conditions. In contrast, bbXLHb and Hp displayed a relatively slower reaction followed by a much slower second kinetic phase, possibly due to polymer formation. These results are consistent with the observations using SEC and native PAGE studies of unmodified and modified Hbs binding with Hp. The Hb concentration dependence of the apparent association rate constants as shown in Figure 3B revealed that bbXLHb and Hb bind to Hp under our experimental conditions with bimolecular rate constants at about 0.01560.004 mM 21 s 21 and 0.05260.004 mM 21 s 21 , respectively. The latter value is close to that reported previously for human Hb binding to Hp type 1-1 [27].
The effects of Hp binding with unmodified Hb and cross-linked Hb tetramers on the rapid kinetics of ligand association and dissociation were evaluated. Figure 4A shows the time courses of O 2 dissociation from Hb and the Hp-Hb complex. The binding of Hp clearly altered the overall time course from a single exponential process (34 s 21 ) to a biphasic reaction with apparent rate constants of 69 s 21 and 14 s 21 based on the nonlinear least squares curve fitting. In contrast, identical O 2 dissociation time courses were recorded for bbXLHb bound or not bound to Hp ( Figure 4B), with no appreciable difference in derived rate constants (Table 1). Figure 5A and 5B illustrate representative time courses of CO association with Hb in the absence and presence of Hp, respectively. The Hb reaction with CO is typically a single exponential process, whereas Hb-Hp complexes exhibit kinetic heterogeneity, including an additional fast phase with approximately 10 times larger second-order rate constant under the same experimental conditions (Table 1). In contrast, the recorded time courses of CO reaction with bbXLHb and the complex of bbXLHb and Hp were identical as depicted in Figure 6, and no significant differences in the derived second-order rate constants were observed ( Table 1).
The effects of Hp binding on Hb oxidative reactions were examined using nitrite and H 2 O 2 as oxidants. It is known that oxygenated Hb reacts with nitrite following a complex process to form metHb and nitrate as end products [28,29,30]. Figure 7A & 7B present the progressive spectral changes from oxy to met Hb over the reaction time of nitrite and native Hb with and without Hp. Fewer intermediate spectra and faster reaction were observed in the presence of Hp ( Figure 7B). Figure 7C shows the typical auto-accelerating time course by monitoring the absorbance change of the oxy Hb and nitrite reaction to completion. Hp binding shifted the curve to the left, enhancing the autoaccelerating reaction of nitrite with the Hb and the Hb-Hp complex and changing the reaction half-time from 19.0 s to 8.4 s under our experimental conditions. Conversely, nitrite-induced oxidation of bbXLHb was not altered by Hp (Figure 7D), and both reactions exhibited similar auto-accelerating process with a half-time of 22-23 s 21 under the same conditions (Table 2).
Hp binding was previously shown to have no effect on Hb pseudoperoxidase activity, the conversion of met to ferryl Hb with H 2 O 2 consumption [2]. The absorbance changes of the formation of ferryl Hb were measured under pseudo first-order conditions in a rapid mixing stopped-flow instrument using a photodiode array detector. The progressive spectral changes in the Soret and visible regions were subjected to global kinetic analysis to de-convolute and reconstruct the spectra of met and ferryl Hbs and to derive the reaction rate constants. The observed rates were a linear function of the H 2 O 2 concentrations, resulting in the second-order rate constants for this oxidation reaction from met to ferryl Hb ( Table 2). No appreciable differences of the H 2 O 2 oxidation rates were obtained for either Hb or bbXLHb in the absence or presence of Hp.
The formation of ferryl Hb is followed by a process of autoreduction to generate metHb which completes a pseudoperoxidase cycle [31]. The transient formation of ferryl Hb can be captured prior to its decay to metHb by adding sodium sulfide to convert the ferryl species to a spectrally more distinct and stable sulfHb species. It was shown in our previous report [15] that although the ferryl Hb concentration reached similar initial levels, the decay of ferryl Hb was attenuated by Hp binding. Figure 8 depicts the kinetic stabilization of ferryl bbXLHb as result of Hp binding in comparison with that of unmodified Hb. The levels of the initial ferryl bbXLHb formation were comparable to that of Hb, and progressive spectral changes focusing on the sulfHb peak at 620 nm is shown as a function of time in the inset. These Data demonstrate that the ferryl form of bbXLHb decayed similarly with initial rates of 0.32 mM/min and 0.36 mM/min in the absence and presence of Hp (Table 2). Although Hp reduced the initial rates from 1.01 mM/min to 0.26 mM/min for wild type Hb, no kinetic effects on ferryl Hb decay were detected in Hp complexed with bbXLHb.

Discussion
The interaction between Hb and Hp is extremely strong and almost irreversible, equivalent to that of antigen-antibody interactions. Hb dimers have long been regarded as the only Hb molecular form that binds with Hp. However, non-dissociable chemically cross-linked Hbs have been observed to bind with Hp [16,32]. The nature of chemically modified Hb and Hp protein complex formation may provide valuable insights into the molecular pathway of Hb clearance under physiological conditions, and may provide a better understanding of Hp mediated attenuation of Hb oxidative reactions [25].
The chemical modifications of Hb used in this study are sitespecific in which the intramolecular reagent, bis(3,5-dibromosalicyl)fumarate was used to either cross-link the two Lys99 residues in a subunits of the deoxy or the two Lys82 residues in b subunits of the oxy forms of human Hb, resulting in aaXLHb and bbXLHb respectively [33,34,35,36]. These structural changes produce stable, low oxygen (T-state) and high oxygen (R-state) affinity Hbs for transfusion purposes in animals and in humans [37,38]. We reasoned that Hb quaternary conformations of aaXLHb and bbXLHb differ sufficiently to restrain aaXLHb and Hp interaction, but allow bbXLHb enough flexibility in its R conformation to serve as points of contact with Hp.
Our protein binding analyses clearly showed that the b subunit cross-linked Hb tetramers bind Hp, whereas tetrameric Hb crosslinked between a subunits has minimal interaction with Hp. SEC-HPLC revealed that approximately 50% of bbXLHb formed complex with Hp compared to 100% complex formation with unmodified Hb and Hp and that bbXLHb appeared to form larger sized complexes. Polymer formation is likely a result of bbXLHb divalent interaction with Hp, in which binding can occur at dimeric sites on the stabilized Hb tetramer [32]. The binding patterns of native Hb and bbXLHb with Hp were resolved by non-denaturing electrophoresis gel analysis which confirmed the SEC-HPLC results. Our rapid mixing analyses indicated that bbXLHb reacts with Hp at a rate that is about 4 fold slower than that of the native Hb. This was followed in the case of bbXLHb by a much slower second kinetic phase, which can be attributed to polymer formation.
We have characterized the impact of Hp interactions on the ligand binding reactions of chemically modified Hbs in order to understand the extent of the structural perturbation introduced by Hp. We also reported that both oxygen equilibrium binding and the kinetics of CO association and oxygen dissociation were altered after complex formation. In addition, we find that chemical modification of aaXLHb significantly decreases oxygen equilibrium binding affinity while increasing the oxygen off rate, whereas bbXLHb exhibits a higher oxygen affinity and CO on rate. Our equilibrium (data not shown) and kinetic parameters determined for the bbXLHb and Hp complex are similar to that of the uncomplexed bbXLHb. Interestingly, it appears that the complex formation between the R-state bbXLHb and Hp did not perturb its heme group reaction environment. Therefore, it is possible to alter Hb structure, promote Hp binding and retain desired ligand reaction properties.
In agreement with previous reports, our results showed that Hp does not change the intrinsic reactivity of Hb with H 2 O 2 , but largely increased ferryl Hb stability. It was previously shown that heme accessibility plays a major role in determining the reaction rate with nitrite [13]. We observed enhanced nitrite oxidation of native Hb in the presence of Hp due to Hb dimerization and potentially heme accessibility. Although chemical modification typically enhances redox reactivity of modified Hb deviations, Hp association with bbXLHb did not cause any additional H 2 O 2 and nitrite reactivity changes. This suggests that modified Hb can bind Hp and retain heme reactivities.
Our results are in agreement with a recent crystal structure analysis of the porcine Hb and Hp complex [17]. Although porcine blood was used as source, porcine and human Hb and Hp exhibit 82% homology by sequence alignment, and the same dumbbell shape with two serine protease domains connected by two complement control protein (CCP) domains. The serine protease domain contains several loops exposed on the surface and the amino-terminal region that were determined to constitute the Hb-binding site. Hp interacts with both Hb a and b subunits, and amino acid residues in helix C, G and FG were determined as the primary sites, but also as Hb dimer interface in the formation of tetramer. These structural data also reveal that the conformation of deoxygenated Hb dimers does not promote Hp binding. Remarkably, these data also show that Hb-Hp interaction originates from an initial complex between C terminus of Hb a subunit and the active serine protease of Hp. This may explain the preferred binding to Hp we observed, as a subunits are more surface exposed due to the cross linkage of the b subunits within bbXLHb.
In summary, although Hb binds to Hp with a high affinity via Hb ab dimerization, non-dissociable Hb tetramers may also form protein complex with Hp. Hb tetramers cross-linked between two b subunits retain an R-state like conformation and display much higher Hp affinity than that of a subunit cross-linked Hb tetramers. Because the binding of bbXLHb to Hp like unmodified Hb retains its ability to decrease the propagation of damaging ferryl radicals, site specific cross-linking of the b subunits may provide a basis for an improved design of Hb-based oxygen therapeutics.