Methods 54 (2011) 175–180

Contents lists available at ScienceDirect

Methods

journal homepage: www.elsevier .com/locate /ymeth

Dimer–tetramer association equilibria of human adult hemoglobinand its mutants as observed by analytical ultracentrifugation

Fumio Arisaka a,⇑, Yukifumi Nagai b, Masako Nagai b

a Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, 4259 B-9 Nagatsuta, Yokohama 226-8510, Japanb Research Center for Micro-Nano Technology, Hosei University, Tokyo 184-0003, Japan

a r t i c l e i n f o

Article history:Accepted 13 January 2011Available online 20 January 2011

Keywords:Sedimentation velocitySedimentation equilibriumHemoglobinSubunit interfaceMutant hemoglobinsDimer–tetramer association equilibrium

1046-2023/$ - see front matter � 2011 Elsevier Inc. Adoi:10.1016/j.ymeth.2011.01.003

⇑ Corresponding author. Fax: +81 45 924 5713.E-mail address: farisaka@bio.titech.ac.jp (F. Arisak

1 Abbreviations used: Hb, hemoglobin; Hb A, humanphoglycerate; IHP, inositol hexaphosphate; T, tensefocusing; r, recombinant; SV, sedimentation velocity; Sc(s), sedimentation coefficient distribution function.

a b s t r a c t

Dimer–tetramer equilibrium of human adult hemoglobin in CO form (COHb A) and its mutants weremeasured by sedimentation velocity and sedimentation equilibrium. In sedimentation velocity, the asso-ciation constants were estimated by measuring the concentration dependence of the weight average sed-imentation coefficients at pH 6 and 7 and fitting the data to the theoretical binding isotherms withassociation constants as a parameter. Association constants of wild type Hb A and three mutant Hbs,Hb Hirose(bW37S), recombinant (r)Hb(bW37H) and rHb(aY42S), in which an amino acid was replacedat the a1b2 interface, were measured in the presence and absence of inositol hexaphosphate (IHP). Allthe three mutations lowered the value of association constants, but the presence of IHP shifted the equi-librium toward tetramer. Although the association constant between dimer and tetramer of rHb(bW37H)and rHb(aY42S) were similar, sedimentation coefficient distribution function, c(s), analysis indicated thatthe association and dissociation rate constants of the former is higher than the latter.

� 2011 Elsevier Inc. All rights reserved.

1. Introduction

Hemoglobin (Hb)1 has been extensively studied in decades dueto its essential biological function as the oxygen-carrying protein.Hb is an allosteric protein showing cooperativity in oxygen binding,and the elucidation of a structural mechanism of allostery has been amajor subject of Hb studies. Hb A consists of two stable ab dimers.The interaction at the a1b2 interface between the two hetero-dimersis relatively weak which makes it possible to rotate each other at theinterface and materialize the tense (T) to relaxed (R) conformationalchange [1]. Upon switching from T to R conformation, the side chainof a197His residue transfer the position between b2Pro44 andb2Thr41 to between b2Thr41 and b2Thr38 [2]. Concomitantly, thespace constituted by the four subunits becomes smaller, so that itcan no longer bind 2,3-diphosphoglycerate (DPG). It is known thatinositol hexaphosphate (IHP) has the same effect as DPG and bindsto hemoglobin even more tightly. On the basis of X-ray crystallo-graphic analysis, Baldwin and Chothia [3] have suggested that liga-tion-induced rearrangement of the a1b2 interface is an essentialcomponent of the cooperative mechanism. There are two importantcontacts, that is, the contact between a194Asp and b237Trp and the

ll rights reserved.

a).adult Hb; DPG, 2,3-diphos-

; R, relaxed; IEF, isoelectricE, sedimentation equilibrium,

contact between a142Tyr and b299Asp. The former is called ‘‘flexiblejoint’’ because this region allows the quaternary structural change tooccur. The latter contact is designated as a ‘‘switch region’’ becausethe side chains of residues in this region take also alternative posi-tions depending upon quaternary structure. The hydrogen bonds be-tween a142Tyr and b299Asp in the switch region and a194Asp andb237Trp in the flexible region have been thought to be importantfor stabilizing the T structure of deoxyHbA (Fig. 1) [3]. These hydro-gen bonds are broken upon oxygen binding.

It was Jeffrey Wyman who first explicitly formulated the effectof pH upon oxygen binding and the association of Hb ab dimer intotetramer by the concept of linked functions [4]. John Schellman la-ter formulated the relation between macromolecular associationand Wyman’s binding potential [5]. Although hemoglobin is oneof the most extensively and intensively studied proteins and thedetailed atomic structure of both oxy and deoxy states are known,not much quantitative measurements of association equilibriahave been reported for Hb A and mutant hemoglobins.

Previously, association–dissociation equilibria of Hb has beenstudied mainly by analytical molecular sieve chromatography [6–10], but also by osmotic pressure [11], sedimentation velocity[12–14] and sedimentation equilibrium [13]. Recently, static lightscattering was also applied to measure the association equilibria[15]. In old literatures of analytical ultracentrifugation (AUC) upto ’80 s’ Spinco or Beckman Model E was used. The only advantageof Model E was that it can accommodate 30 mm path length cellsso that solutions of lower concentrations can be measured and

Fig. 1. Inter-subunit interactions of deoxyHb A at the interface between the a1 and b2 subunits as revealed by X-ray crystallography, PDB: 2HHB [34]. Tyra42 is hydrogenbonding with Aspb99, and Trpb37 with Aspa94 of the G-helix. These hydrogen bonding are absent in the liganded form [3]. The F-helix contains the proximal His (F8).

176 F. Arisaka et al. / Methods 54 (2011) 175–180

measurements at higher temperatures, up to 60 �C was possible(Beckman-Coulter XL-A/I can be used up to 40 �C.). Tremendousdevelopments in the analysis of the AUC data was made since1991 when new analytical ultracentrifuge, Beckman XL-A/XL-I(presently Beckman-Coulter XL-A/I) were introduced. Differentfrom Model E, the raw data were directly incorporated into PCwhich allowed one to apply much more sophisticated mathemati-cal treatments. Since that time, a number of softwares have alsobeen developed by a number of authors, and it was 1998 that SED-FIT was reported by Peter Schuck at NIH, which eliminated the ef-fect of diffusion to obtain c(s) [16]. Based on the remarkabledevelopments in analytical ultracentrifugation, we have decidedto revisit and measure association–dissociation equilibria of Hband analyze the data by SEDFIT.

In the present study, the wild type hemoglobin, Hb A, and threemutant hemoglobins were studied in order to elucidate the func-tional roles of amino acid residues at the a1b2 interface. For thatpurpose, association–dissociation properties as well as oxygenbinding property of these hemoglobins were studied.

2. Materials and methods

2.1. Hemoglobins

Hb A was prepared from fresh human blood by a preparativeisoelectric focusing electrophoresis (IEF) using 5% Ampholine (pH6–9) [17]. Hb Hirose was purified from patient’ hemolysate by pre-parative IEF (pH 7–8) [17].

For the synthesis of recombinant (r) Hb, the Hb A expressionplasmid pHE7 [18] containing human a- and b-globin genes andthe Escherichia coli methionine aminopeptidase gene was kindlyprovided by Professor Chien Ho of Carnegie Mellon University.Plasmids for rHb(aY42S) and rHb(bW37H) were produced bysite-directed mutagenesis using an amplification procedure ofclosed circular DNA in vitro [19]. These plasmids were transformed

into E. coli JM109. E. coli cells harboring the plasmid were grown at30 �C in TB medium [18,20]. Expression of recombinant hemoglo-bin was induced by adding isopropyl b-thiogalactopyranoside tobe 0.2 mM. The culture was then supplemented with hemin(30 lg/ml) and glucose (15 g/l), and the growth was continuedfor another 5 h at 32 �C. The cells were harvested by centrifugationand stored frozen at �80 �C until needed for purification.

Recombinant Hbs were isolated and purified according to themethod of Looker et al. [21] with some modifications. For the puri-fication of rHbs, three chromatographic steps were employed [21].Before loading onto the column, the samples were always satu-rated with CO gas. The first column, Q-Sepharose fast flow column(2.5 � 20 cm), was equilibrated with 20 mM Tris–HCl/0.1 mMTETA, pH 7.4, at 4 �C. This step captured a large amount of bacterialprotein and the remaining nucleic acid, while rHb passed through.The rHb fraction was collected, concentrated and dialyzed against20 mM Tris–HCl, pH 8.3, at 4 �C (Q1 fraction). The second column,Q-Sepharose fast flow column (1.5 � 17 cm), was equilibrated with20 mM Tris–HCl, pH 8.3, at 4 �C. After loading the Q1 fraction andthen washing the column with equilibration buffer, the boundrHb was eluted with a linear gradient from 0 to 160 mM NaCl inequilibration buffer. The third column, SP-Sepharose fast flow col-umn (1.5 � 45 cm), was equilibrated with 10 mM sodium phos-phate buffer, pH 6.8, at 4 �C. A linear gradient of equilibrationbuffer versus 10 mM sodium phosphate buffer, pH 8.3, at 4 �Cwas used to elute the rHb. Fractions were pooled as purified rHbon the basis of high ratio of ellipticity of CD at 260 nm and absor-bance at 572 nm (ratio >45) to remove the rHb with reversed hemeorientation [22].

2.2. Analytical ultracentrifugation

Sedimentation velocity experiments were performed by an Op-tima XL-I analytical ultracentrifuge (Beckman-Coulter) using afour-hole An60Ti or an eight-hole An50Ti rotor at 20 �C. Prior to

F. Arisaka et al. / Methods 54 (2011) 175–180 177

centrifugation, samples were dialyzed against 50 mM phosphatebuffer, pH 7.0 and the dialysate was used as reference solutionsin all experiments. After CO-gas was refilled into the sample solu-tion, 400 lL of sample solution and 430 lL of the dialysate wereapplied in a standard double sector cells. The rotor speed was50,000 rpm. The hemoglobin concentrations were monitored at568 nm for most of the experiments, but when measurementswere made for protein concentration dependence, the wavelengthof 419 and 510 nm were also used to keep the absorbance lowerthan 1.5. In all experiments, the material of the centerpiece wasEpon charcoal-filled. When the absorbance was lower than 1.5,normal centerpiece with the optical path length of 12 mm wasused, but when the absorbance exceeded 1.5, centerpiece withthe path length of 3 mm was used. When the protein concentrationexceeded about 3 mg/mL, Wiener-skewing effect was observed, inwhich situation, several of the initial scans were discarded to avoidthat effect. All data were acquired without time intervals betweensuccessive scans. When deoxyHb A was measured, the sector wasonce rinsed by buffer containing Na-dithionite and then the sam-ple containing Na-dithionite (1 mg/ml) was carefully applied inthe sector cell and sealed. In order to confirm that the Hb was indeoxy state, wavelength scan was performed in the VIS regionand the presence of the absorption at 560 nm characteristic fordeoxyHb was verified.

Sedimentation coefficient distribution function, c(s), was ob-tained by using SEDFIT program [16,23,24]. Molecular mass distri-bution, c(M), was obtained by converting c(s) to c(M) on theassumption that the frictional ratio f/f0 was common to all themolecular species (as implemented in the SEDFIT program). Itwas noted that higher concentration of Hb results in higher rmsdthan 0.01 which is very likely due to non-ideality with somewhatlarger frictional ratio and, therefore, c(M), was not reliable. Evenso, the c(s) profile and the weight average molecular weight ofthe s-values are not significantly affected. We, therefore, stickedto use c(s) and did not utilize c(M). For plotting and curve-fittingof weight average sedimentation coefficients vs. protein concentra-tions, Kaleidagraph was used. Stable dimer of Hb A was regarded asa monomer and dimer–tetramer equilibrium was fitted to mono-mer–dimer equilibrium scheme. The weight average sedimenta-tion coefficients can be described as below:

sw ¼s1K½A� þ 2s2K½A2�

KA0

¼ s1ðffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1þ 8KA0p

� 1Þ þ s2ð1þ 4Ka0 �ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1þ 8KA0p

Þ4KA0

;

where s1 and s2 are sedimentation coefficients of monomer and di-mer, respectively, K is the association constant and A0(= [A] + 2[A2])is the total molar concentration of Hb in a monomer base. s1 and s2,in the present study correspond to the sedimentation coefficientsfor stable dimer and tetramer, respectively. For the fitting proce-dure, the value of s1 was set to 2.7S which was taken from the s-va-lue of the first peak of Hb Hirose and that of s2 was set to 4.3 Swhich was taken from that of deoxyHb A.

Sedimentation equilibrium experiments were carried out in anOptima XL-I with an 8-hole An50Ti rotor. One hundred and twentymicroliters of sample solution was loaded and 130 lL of the corre-sponding dialysate was applied as a sample and reference solution,respectively, in standard six channel centerpieces or double sectorcells. Concentration profiles were monitored by absorbance at568 nm. The sample concentrations were at A568 of 0.2, 0.3 and0.5 and the measurements were made at 20 �C. Rotor speeds were8100, 13,700 and 24,000 rpm, which were estimated by SEDFITaccessory function based on the molecular weight of dimer andtetramer of hemoglobin. Scans were made every 2 h and the equi-librium of the system was judged by superposition of the last four

scans. Extraction of the data points for analysis and global fittingwas carried out by ‘‘non-lin’’ software implemented in the Beck-man-Coulter software package. Before global fitting, each scanwas used to calculate weight average molecular weight. Some datapoints of each scan which were away from the calculated theoret-ical curve were deleted and in some cases the scans as a whole oflow quality were eliminated. The selected datasets out of totallynine datasets were globally fitted to a single species model firstto determine the weight average molecular weight. The partialspecific volume of Hb A was set to 0.749 [25] and used for mutantrHb as well. Solvent density was calculated by Sednterp program[26]. Afterwards, dimer–tetramer association equilibrium was as-sumed and the association constants were obtained by fixing thedimer molecular weight of 32,270. The association constants KA

thus obtained had the unit of absorption which was converted tomolar-base association constants Km by the equation:

Km ¼ aAMA

2KA;

where aA, MA denote molar extinction coefficient and molecularweight, respectively.

3. Results and discussion

In the present study, we have utilized AUC to study the dimer–tetramer equilibrium of Hb A and three mutant Hbs. The advantageof AUC over other methods such as molecular sieve chromatogra-phy is that it does not require carrier or supports which may havesome non-specific interactions to the solute. Besides, rigorous the-oretical basis of the methodology has been established since thedevelopment of AUC since 1920s. We have used both sedimenta-tion velocity (SV) and sedimentation equilibrium (SE), but SVplayed the major role in this study. The advantage of SV over SEis that the former has much higher sensitivity with respect to het-erogeneity than the latter and that the determination of Kd doesnot depend on vbar. Recent development of c(s) analysis by PeterSchuck as implemented in SEDFIT software has proven much ver-satile and applicable in various systems [27].

3.1. Oxygen equilibrium properties of mutant hemoglobins

Hb Hirose (bTrp37 ? Ser, bW37S) shows high oxygen affinityand low cooperativity (P50 = 2.1 mmHg, Hill’s n = 1.5) [28,29]. Com-pared with Hb A at pH 7.0 (P50 = 11.3 mmHg, Hill’s n = 2.8),rHb(aY42S) gives about four times higher oxygen affinity thanHb A and significant cooperativity (P50 = 2.7 mmHg, n = 2.1) [30].On the other hand, rHb(bW37H) exhibits 7 times higher oxygenaffinity than Hb A and low cooperativity (P50 = 1.7 mmHg,n = 1.8) [30]. However, the addition of IHP lowered oxygen affinityand significant increase in cooperativity of rHb(bW37H)(P50 = 6.9 mmHg, n = 2.1). The mutant hemoglobins at the b37Trpare known to dissociate into dimer upon oxygenation [29,31].

3.2. Dynamic dimer–tetramer association equilibrium of Hb A

The effect of amino acid replacements at the interface was stud-ied by analytical ultracentrifugation. First, the weight average sed-imentation coefficients of Hb A were measured as a function of theprotein concentration at pH 6 and 7. An example of raw data and ac(s) profile is shown in Fig. 2. Comparison of the sedimentationbehavior at two pH values indicated that lower pH indeed dramat-ically facilitated dissociation of the tetramer into two dimers(Fig. 3, Fig 4). The association constants at pH 6.0 and 7.0 were cal-culated to be 8.1 � 103 M�1 and 3.2 � 105 M�1, respectively, basedon curve-fitting by least square method to fit the experimentallydetermined s-values as a function of the hemoglobin concentration

Fig. 2. c(s) Analysis of sedimentation velocity data by SEDFIT. Top: moving boundaries for COHb A, 0.8 mg/mL in 50 mM phosphate buffer, pH 7.0. A wavelength of 568 nmwas used for scanning. Middle: residuals between raw and theoretically fitted data points. Bottom: obtained distribution function of sedimentation coefficients, c(s).

Fig. 3. c(s) analysis of COHb A in 50 mM phosphate buffer at pH 7 and 6 with different protein concentrations. Measurements were made at wavelength of 510 nm for 10 and3 mg/mL, 568 nm for 1 and 0.3 mg/mL and 419 nm for 0.1 mg/mL. 1.2 cm pathlength centerpiece was used for all the measurements except for the solution of 10 mg/mL,where 3 mm pathlength centerpiece was used. Weight average sedimentation coefficients were calculated by integrating over the whole area of existing peak(s).

178 F. Arisaka et al. / Methods 54 (2011) 175–180

to the theoretically predicted isotherm. It has been reported that atlow concentrations of Hb, the P50 value tends to increase and theHill coefficient n decrease [32]. These properties of Hb are consis-tent with the present observation that Hb A is in dynamic dimer–

tetramer equilibrium and that the lower concentration of Hb facil-itated dissociation into dimers. Although sedimentation velocity(SV) measurements indicated that COHb A is mostly tetramer atpH 7, slight dissociation may affect the c(s) of Hb A. We, therefore,

Fig. 4. Protein concentration dependence of sedimentation coefficients of COHb Aat pH 7 (h) and pH 6 (s) based on the c(s) analysis in Fig. 3. A0 is the loadingconcentration of Hb A.

Fig. 6. An example of the concentration profile of sedimentation equilibrium;rHb(aY42S) at 24,000 rpm with the loading concentration 0.5 at A568 (seesedimentation equilibrium section under Section 2.2).

F. Arisaka et al. / Methods 54 (2011) 175–180 179

measured sedimentation velocity of deoxyHb A, as tetramer in thedeoxy state is known to be much more stable. In order to ensurethat the Hb in the AUC cell retained the deoxy state, UV absorptionspectrum at 6.5 cm from the center of rotation was measured inthe AUC cell and confirmed that the Hb was in the deoxy statebased on the absorption maximum at 560 nm. The s-value of thedeoxyHb A was 4.33S which was indeed slightly higher than thatof COHb A, 4.27 S.

As Hb A and Hb Hirose was isolated from blood, whereas twomutant proteins, rHb(aY42S) and rHb(bW37H) are recombinant,we measured the sedimentation velocity of rHb A (no mutation)to confirm that the rHb A behaves exactly the same as Hb A fromthe blood. The s-value of the rHb A was 4.27S which is identicalto that of Hb A from blood. As expected, SV measurement didnot detect any difference between Hb A from blood and rHb A.

Fig. 5. c(s) Profiles of wild type and three mutant hemoglobins in the presence and absenwas 50,000 rpm and moving boundaries were monitored at 568 nm, 20 �C.

3.3. Association equilibrium of mutant hemoglobins

SV measurements of Hb A and three mutant Hbs were carriedout at a protein concentration of 1.2 mg/mL in the presence andabsence of IHP (Fig. 5). The data were analyzed by SEDFIT andthe weight average sedimentation coefficients of these Hbs werecalculated by integrating the c(s) peaks. Compared with the c(s)peak for Hb A, it is readily seen that all three mutants tend to dis-sociate into two dimers in some different way. In Hb Hirose andrHb(aY42S), two peaks are apparent, but in rHb(bW37H), twopeaks are not separated and the larger peak is only seen as a shoul-der, indicating that the latter Hb is under faster association equilib-rium than the former two mutant Hbs. Note that in these dynamicassociation equilibrium, peaks do not precisely correspond to theexisting molecular species. Indeed, if the dynamic equilibrium isfast enough, only one peak will be seen of which the s-value is

ce of 2 mM IHP at pH 7. The protein concentration was 1.2 mg/mL. The rotor speed

Table 1Weight average molecular weights and association constants of Hb A and threemutant hemoglobins.

Hb Single species model KAa Kmb (M�1)

Hb A 63,100 2263 3.1 � 107c

Hb Hirose 32,300 0.051 7.0 � 103

rHb(aY42S) 58,700 93.4 1.3 � 106

rHb(bW37H) 41,600 3.59 4.9 � 105

a KA, association constants based on absorption.b Km, association constants based on molar concentration.c This association may contain significant error due to the fact that hemoglobin

under this condition is mostly tetramer and the dimer concentration is very low.

180 F. Arisaka et al. / Methods 54 (2011) 175–180

the result of the reaction boundary. If higher multimers than dimerare in equilibrium with monomer and dimer, there might be multi-ple peaks even if instantaneous equilibrium is established. Addi-tion of IHP to these mutant Hbs caused to shift toward tetramer,concomitantly some mutant Hbs restored cooperativity [30].

Sedimentation equilibrium (SE) measurements of the samehemoglobins were also carried out and the resultant concentrationgradients at equilibrium were analyzed by single species modelfirst and then for the model of dimer–tetramer equilibrium(Fig. 6). The results are shown in Table 1. The results are in reason-ably good agreement with those from SV measurement.

4. Conclusions

Hb A and three mutant Hbs were studied by analytical ultracen-trifugation and the results were utilized to understand the oxygenbinding behavior of Hbs. It is concluded that the impaired cooper-ativity of mutant Hbs is due to dissociation into dimers. c(s) orSEDFIT analyses turned out to be quite powerful in delineatingthe association–dissociation equilibrium and in visually giving in-sights into kinetics of the association.

In the present study, both sedimentation velocity (SV) and sed-imentation equilibrium (SE) were used. SV has advantage in deter-mining the association constant in that we do not need the vbar ofthe solute. Besides, it has much higher sensitivity in detecting theimpurity or heterogeneity of the sample than SE. Furthermore,unraveling the self-association scheme of other species could bedone predominantly by SV. Both SV and SE would encounter theproblem of non-ideality; non-ideality in SV would give rise toapparent higher molecular weights in c(M), whereas SE would giveapparent lower Mw. The advantage of SE over SV would be that SEcould deal with much higher concentration than SV as demon-strated by Hb work by Allen P. Minton [33].

Acknowledgements

We thank Professor Chien Ho for the gift of the E. coli Hb expres-sion plasmid, Dr. Yayoi Aki for constructing plasmids and express-

ing mutant Hbs, Dr. Yuzo Ohba for the gift of blood containing HbHirose, and Dr. Shigenori Nagatomo for the help of the preparationof Fig. 1.

References

[1] G. Fermi, M.F. Perutz, Atlas of Molecular Structures in Biology, 2. Haemoglobin& Myoglobin, Clarendon Press, Oxford, 1981.

[2] R.E. Dickerson, J. Geis, Hemoglobin: Structure, Function, Evolution, andPathology, Benjamin/Cummings, Menlo Park, California, 1983.

[3] J. Baldwin, C. Chothia, J. Mol. Biol. 129 (1979) 175–220.[4] J. Wyman, Adv. Protein Chem. 19 (1964) 223–286.[5] J.A. Schellman, Biopolymers 14 (1975) 999–1018.[6] D.H. Atha, A. Riggs, J. Biol. Chem. 251 (1976) 5537–5543.[7] S.H.C. IP, G.K. Ackers, J. Biol. Chem. 252 (1977) 82–87.[8] R. Valdes, L.P. Vickers, H.R. Halvorson, G.K. Ackers, Proc. Natl. Acad. Sci. USA 75

(1978) 5493–5496.[9] A. Dumoulin, L.R. Manning, W.T. Jenkins, R.M. Winslow, J. Biol, Chem. 272

(1997) 31326–31332.[10] T. Yamaguchi, J. Pang, K.S. Reddy, S. Surrey, K. Adachi, J. Biol. Chem. 273 (1998)

14179–14185.[11] G. Guidotti, J. Biol. Chem. 242 (1967) 3685–3693.[12] A.G. Kirshner, C. Tanford, Biochemistry 3 (1964) 291–296.[13] S.J. Edelstein, M.J. Rehmer, J.S. Olson, Q.H. Gibson, J. Biol. Chem. 245 (1970)

4372–4381.[14] S.J. Edelstein, Q.H. Gibson, in: B. Chance, T. Yonetoni, A.S. Mildvan (Eds.),

Probes of structure and function of macromolecules and membranes, Probes ofEnzymes and Hemoproteins, vol. 2, Academic Press, Orlando, FL, 1971, pp.417–429.

[15] T. Yamaguchi, K. Adachi, Biochem. Biophys. Res. Commun. 290 (2002) 1382–1387.

[16] P. Schuck, Biophys. J. 75 (1998) 1503–1512.[17] M. Nagai, S. Kaminaka, Y. Ohba, Y. Nagai, Y. Mizutani, T. Kitagawa, J. Biol.

Chem. 270 (1955) 1636–1642.[18] T.-J. Shen, N.T. Ho, M. Zou, D.P. Sun, P.F. Cottam, V. Siplaceanu, M.F. Tam, D.A.

Bell Jr., C. Ho, Protein Eng. 10 (1977) 1085–1097.[19] Z. Chen, D.E. Ruffner, Nucleic Acids Res. 26 (1998) 1126–1127.[20] T.-J. Shen, N.T. Ho, V. Siplaceanu, M. Zou, B.N. Green, M.F. Tam, C. Ho, Proc.

Natl. Acad. Sci. USA 90 (1993) 8108–8112.[21] D. Looker, A.J. Mathews, J.O. Neway, G.L. Stetler, Methods Enzymol. 231 (1994)

364–374.[22] M. Nagai, Y. Nagai, Y. Aki, K. Imai, Y. Wada, S. Nagatomo, Y. Yamamoto,

Biochemistry 47 (2008) 517–525.[23] P. Schuck, Biophys. J. 78 (2000) 1606–1619.[24] P. Schuck, M.A. Perugini, N.R. Gonzales, G.J. Howlett, D. Schubert, Biophys. J. 82

(2002) 1096–1111.[25] E. DeMoll, D.J. Cox, E. Daniel, A.F. Riggs, Anal. Biochem. 363 (2007) 196–203.[26] T.M. Laue, B.D. Shah, T.M. Ridgeway, S.L. Pelletier, in: S.E. Harding, A.J. Rowe,

J.C. Horton (Eds.), Analytical Ultracentrifugation in Biochemistry andPolymer Science, Royal Society of Chemistry, Cambridge, UK, 1992, pp. 90–125.

[27] Peter Schuck, in: Peter Schuck (Ed.), Protein Interactions in the Series ofProtein Reviews, 2007, pp. 289–316, 469–518.

[28] K. Yamaoka, Blood 38 (1971) 730–738.[29] J. Sasaki, T. Imamura, T. Yanase, D.H. Atha, A. Riggs, J. Bonaventula, C.

Bonaventula, J. Biol. Chem. 253 (1978) 87–94.[30] Y. Aki-Jin, Y. Nagai, K. Imai, M. Nagai, ACS Symp. Ser. 963 (2007) 297–311.[31] L. Kiger, A.L. Klinger, L.D. Kwaitkowski, A.D. Young, M.L. Doyle, J.M. Holt, R.W.

Noble, G.K. Ackers, Biochemistry 37 (1998) 4336–4345.[32] K. Imai, T. Yonetani, Biochim. Biophys. Acta 490 (1977) 164–170.[33] A.P. Minton, J. Mol. Biol. 112 (1977) 437–452.[34] G. Fermi, M.F. Perutz, B. Shaanan, R. Fourme, J. Mol. Biol. 175 (1984) 159–174.