Bis[2-(4-carboxyphenoxy)carbonylethyl]phosphinic acid (BCCEP): A novel affinity reagent for the β-cleft modification of human hemoglobin

Article abstract of DOI:10.1016/S0968-0896(98)00032-7

The design, synthesis, and hemoglobin cross-linking studies of a novel organic reagent, bis[2-(4 carboxyphenoxy)carbonylethyl]phosphinic acid (BCCEP, 1) have been reported. The reagent was designed with the aid of molecular modeling, employing crystal coordinates of human hemoglobin A₀. It was synthesized in three steps commencing from 4-t-butoxycarbonylphenol. The tri-sodium salt of 1 was employed to cross-link human oxyHb. While SDS-PAGE analyses of the modified hemoglobin product pointed to the molecular mass range of 32kDa, the HPLC analyse suggested that the cross-link had formed between the β₁-β₂ subunits. The oxygen equilibrium measurements of the modified hemoglobin at 37°C showed significantly reduced oxygen affinity (P₅₀=31.3Torr) as compared with that of cell-free hemoglobin (P₅₀=6.6Torr). The sigmoidal shape of O₂ curves of the modified Hb pointed to reasonable retainment of oxygen-binding cooperativity after the cross-link formation. Molecular dynamics simulation studies on the reagent-HbA₀ complex suggested that the most likely amino acid residues involved in the cross-linking are N-terminus Val-1 or Lys-82 on one of the-chains, and Lys-144 on the other. These predictions were consistent with the results of MALDI-MS analyses of the peptide fragments obtained from tryptic digestion of the cross-linked product. Copyright (C) 1998 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(98)00032-7

BIOORGANIC &
MEDICINAL
CHEMISTRY
Bioorganic & Medicinal Chemistry 6 (1998) 767±783
Bis[2-(4-carboxyphenoxy)carbonylethyl]phosphinic Acid
(BCCEP): A Novel Anity Reagent for the ꢀ-Cleft
Modi®cation of Human Hemoglobin
Ramachandra S. Hosmane,a,* S. Prasad Peri,a Vishweshwar S. Bhadtia
and Victor W. Macdonaldb
aLaboratory for Drug Design and Synthesis, Department of Chemistry and Biochemistry, University of Maryland, Baltimore County,
Baltimore, MD 21250, USA
bBlood Research Detachment, Walter Reed Army Institute of Research, Washington D.C., 20307, USA
Received 22 September 1997; accepted 8 December 1997
AbstractÐThe design, synthesis, and hemoglobin cross-linking studies of a novel organic reagent, bis[2-(4 carboxy-
phenoxy)carbonylethyl]phosphinic acid (BCCEP, 1) have been reported. The reagent was designed with the aid of
molecular modeling, employing crystal coordinates of human hemoglobin A0. It was synthesized in three steps com-
mencing from 4-t-butoxycarbonylphenol. The tri-sodium salt of 1 was employed to cross-link human oxyHb. While
SDS±PAGE analyses of the modi®ed hemoglobin product pointed to the molecular mass range of 32 kDa, the HPLC
analyse suggested that the cross-link had formed between the b1-b2 subunits. The oxygen equilibrium measurements of
the modi®ed hemoglobin at 37 ꢀC showed signi®cantly reduced oxygen anity (P50=31.3 Torr) as compared with that
of cell-free hemoglobin (P50=6.6 Torr). The sigmoidal shape of O2 curves of the modi®ed Hb pointed to reasonable
retainment of oxygen-binding cooperativity after the cross-link formation. Molecular dynamics simulation studies on
the reagent-HbA0 complex suggested that the most likely amino acid residues involved in the cross-linking are N-ter-
minus Val-1 or Lys-82 on one of the-chains, and Lys-144 on the other. These predictions were consistent with the
results of MALDI-MS analyses of the peptide fragments obtained from tryptic digestion of the cross-linked product.
# 1998 Elsevier Science Ltd. All rights reserved.
Introduction
inadequate oxygen anity and the absence of a meta-
bolic mechanism other than the expired air for excre-
tion, coupled with the associated immunotoxicity, the
use of ¯uorocarbons as resuscitation ¯uids may be
impractical.2 This leaves hemoglobin as the best candi-
date for a blood substitute. Being a natural oxygen car-
rier, hemoglobin o€ers a number of obvious advantages
including, but not limited to, its abundance, facile
metabolism, and high tolerance by the body.4
The current health scare of blood-borne diseases such as
AIDS and hepatitis has heightened the interest in
developing an alternative to blood for emergency trans-
fusion. The shortage, brief shelf-life, and storage di-
culties of whole blood, coupled with the necessity for
typing, cross-matching, and thorough screening for
bacteria and viruses, have prompted continuous e€orts
to develop an acellular oxygen carrier for well over half
a century.1 In this regard, the two intensely studied
blood substitutes include ¯uorocarbon emulsions1±3 and
cell-free hemoglobin solutions.1±4 However, owing to
There are, however, two major problems inherent in the
use of cell-free hemoglobin that must be overcome
before it can be employed in place of whole blood. First,
separated from red blood cells, hemoglobin, which nor-
mally exists as a tetramer, easily dissociates into two
a,b-dimers, and thus does not stay in circulation very
long. It undergoes renal excretion within 1±4 h after
Key words: Hemoglobin (Hb); molecular modeling; cross-link-
ing; blood substitutes; P50; Hill coecient; MALDI MS analyses.
*Corresponding author.
0968-0896/98/$19.00 # 1998 Elsevier Science Ltd. All rights reserved
PII: S0968-0896(98)00032-7

768
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
transfusion, often causing massive hemoglobinurea.4
Second, cell-free Hb has too high an oxygen anity,
which prevents it from optimal delivery of oxygen,
acquired from lungs, to tissues. A signi®cant part of these
problems has now been traced to the hemoglobin's nat-
ural allosteric regulator 2,3-diphosphoglycerate (DPG),
a highly anionic small molecule that is present in intact
erythrocytes, but de®cient in Hb isolated from red blood
cells. DPG aligns itself between the two b subunits
(molecular mass=16 kDa each) of the hemoglobin tetra-
mer, and is surrounded by several positively charged
amino acid residues residing on the b-subunits, thus
forming an anionic sink that is commonly referred to as
the b-cleft or the DPG pocket. Inside the red blood cells,
the predominant function of DPG is to lower, and thus
tune, the oxygen anity of Hb by shifting the deoxy±oxy
equilibrium toward the deoxy form. DPG favors the
deoxy form over the oxy because of the relatively larger
pocket size of the former. The covalent cross-linking of
hemoglobin subunits with a DPG mimic, preferably in
the b-cleft site, is anticipated to considerably alleviate
these intrinsic problems of cell-free hemoglobin.1
In an attempt to address these issues, we report herein
our work on the design and synthesis of a novel bifunc-
tional organic reagent, bis[2-(4-carboxyphenoxy)carbonyl-
ethyl]phosphinic acid (BCCEP, 1), employing
a
combination of modern molecular modeling and syn-
thetic organic techniques. Also reported are the hemo-
globin cross-linking studies using BCCEP, along with
physicochemical and functional data on the modi®ed
hemoglobin, as well as the elucidation of sites of cross-
link employing MALDI MS analyses of tryptic digests
of the cross-linked protein.
Materials and Methods
Molecular modeling studies
E€orts have since been focused on intramolecular
(interdimer)/intermolecular (inter-tetramer) cross-link-
ing with multifunctional organic reagents such as di-
aldehydes,5 bis-imidates,6 diaspirins,7,12c triaspirins,8a,8b
tetraaspirins,8c bis-isothiocyanates,9 bis-enol-ethers,10
bis-pyridoxal phosphates,11 and bis/tris(methyl phos-
phates).12 With only a few notable exceptions that are
currently undergoing clinical trials,1 the majority of
modi®ed hemoglobins still su€er from too high oxygen
anity, too low intravascular retention time or too
facile autooxidation to MetHb that lacks the capacity to
carry oxygen.1 Although some reagents are known to be
speci®c for the b-cleft, most of them are non-speci®c or
random cross-linkers that yield heterogeneous mixtures
of species, ranging in molecular mass from 64 kDa to
600 kDa.1±4 The most desirable reagent would be one
that not only possesses adequate physicochemical and
stereoelectronic characteristics so as to be speci®cally
directed to the DPG pocket (b-cleft) of hemoglobin like
an anity label, but also invokes optimally stabilizing
Hb-reagent interactions even after the cross-link forma-
tion with little, if any perturbation of the native struc-
ture of hemoglobin. In this regard, even those Hb
cross-linking reagents that are known to be speci®c for
the b-cleft might be less than ideal. The reagent should,
in addition, enable cross-linking under convenient,
ambient, oxygenated reaction conditions noting that a
number of earlier Hb cross-linkings have been performed
on deoxyHb under stringent deoxygenated media.
Finally, the reagent should be easy and inexpensive to
synthesize on a large scale, considering the fact that sev-
eral otherwise promising reagents were not pursued by
industry because of perceived high costs of production.1
Molecular modeling was performed on a Silicon Gra-
phics workstation, using the software Insight/Discover
(Molecular Simulations, Inc., San Diego, CA, USA).
The X-ray coordinates of human HbA0,13 imported
from the Brookhaven National Laboratory, Upton,
NY, USA, were employed for molecular modeling stu-
dies. Reagent 1 was designed as described under Results
and Discussion. It was energy-minimized and docked
into the b-cleft (DPG pocket) of hemoglobin, followed
by energy minimization of the reagent-Hb complex. All
atoms that were 12 A or farther from the ligand were
®xed with a temperature constant of 300 K. No con-
straints were applied to the remaining residues in and
around the ligand site. The complex was minimized to
convergence using consecutive Steepest Descent and
Conjugate Gradient (VAO9A and Newton) minimiza-
tion protocols (see Figure 1). In order to simulate the
natural environment even further, a femtosecond mole-
cular dynamics simulation (5000 iterations) at 300 K
was performed on the above energy-minimized protein-
ligand complex by soaking the latter in an aqueous layer
of 5 A thickness all around the complex. There were no
morse or cross terms. The distance ranges of the ester
carbonyl carbon atoms of the reagent from the appro-
priate "-amino nitrogen atoms of lysine residues or the
a-amino nitrogen atom of the N-terminus valine were
computed from graphs derived from the dynamics tra-
jectories.
Synthesis of reagent BCCEP (1). Commercial reagents
were employed without further puri®cation. Organic
reagents and solvents were purchased from Fischer

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
769
Figure 1. (a) A stereoview of the energy-minimized BCCEP-HbA0 complex. Only those amino acid residues of SFHb that lie within a
12 A radius from the reagent in the b-cleft are shown. (b) A closeup view of the above in monoview with the reagent and the three
concerned amino acid residues shown in solid render.
Scienti®c Co., Aldrich Chemical Co., or Lancaster
Synthesis, Inc., and the solvents were dried before use.
The purity of samples of newly synthesized materials
was assessed by a combination of NMR spectroscopy,
mass spectrometry, analytical thin layer chromatography,
and elemental analysis. The latter was performed by
Atlantic Microlab, Inc., Norcross, Georgia. The syn-
thesis of Reagent 1 is outlined in Scheme 1.
The organic layer was dried over MgSO4, ®ltered, and
the ®ltrate evaporated in a rotavapor to give the crude
product which was puri®ed over silica gel to give a col-
1
orless oil (4.1 g, 80%); H NMR (CDCl3) (8.03 (d, 2 H,
Ar-H), 7.18 (d, 2 H, Ar-H), 6.63 (d, 1 H, CH=CH2),
6.32 (dd, 1 H, CH=CH2), 6.03 (d, 1 H, CH=CH2), 1.59
(s, 9H, CH3); 13C NMR (CDCl3) d 164.53, 163.51,
153.67, 132.60, 130.65, 129.38, 127.41, 121.06, 80.74,
27.88.
4-t-Butoxycarbonylphenyl acrylate (3). 4-Hydroxy-t-butyl
benzoate (2)14 (4 g, 20.6 mmol) was dissolved in anhy-
drous THF (250 mL) under nitrogen, and the solution
was cooled in an ice bath. To this was added NaH
(0.5 g, 21 mmol, 96% dispersion in mineral oil). The
suspension was allowed to stir under nitrogen for
15 min, after which acryloyl chloride (1.9 g, 21 mmol)
was added slowly under nitrogen. The reaction mixture
was allowed to stir for an additional 4 h and allowed to
come to room temperature. It was then diluted with
200 mL of CH2Cl2 and washed with water (3Â200 mL).
Bis[2-(4-t-butoxycarbonyl)phenoxycarbonylethyl]phosph-
inic acid (4). To an ice cold stirred solution of hypo-
phosphorus acid (0.17 g, 2.6 mmol) in dry CH2Cl2 was
added N,O-bistrimethylsilylacetamide (1.85 g, 9.1 mmol)
and the above acryloyl ester (1.3 g, 5.2 mmol), and the
reaction mixture was allowed to stir for 1 h. It was par-
titioned between CH2Cl2 and H2O, and the organic
layer was washed with aq. HCl (0.1N, 3Â100 mL). The
organic layer was then dried with MgSO4, ®ltered, and
the ®ltrate evaporated to give an oil. Upon trituration

770
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
Scheme 1.
with 2-propanol, a white solid precipitated out (1.4 g,
96%) which was directly used for the next reaction
0.15 mmol), dissolved in 50 mL of EtOH, was loaded
and the column was then eluted with an additional
200 mL of water. Evaporation of solvent gave an oil
which, upon addition of CH3CN, precipitated out as a
powder (0.07 g, 90%). The solid was dried under
vacuum and used for Hb modi®cation without further
ꢀ
1
without further puri®cation, mp 147±149 C; H NMR
(CDCl3) d 7.99 (d, 2 H, Ar-H), 7.18 (d, 2 H, Ar-H), 4.90
(br s, 1 H, P-OH), 2.88 (m, 4 H, CH2), 2.18 (m, 4 H, CH2),
1.50 (s, 18 H, two C4H9). Anal. calcd for C28H35PO10:
C, 59.78; H, 6.27. Found: C, 59.98; H, 6.30.
1
puri®cation; H NMR (D2O) d 7.92±7.84 (m, 4 H, Ar-
H), 7.160 (t, 2 H, Ar-H), 6.89 (br, t, 2 H, Ar-H), 2.84 (br
m, 2 H, CH2), 2.49±2.47 (br m, 2 H, CH2), 1.95±1.77 (m,
4 H, CH2).
Bis[2-(4-carboxyphenoxy)carbonylethyl]phosphinic acid
(1). A solution of 4 (0.1 g, 0.17 mmol) was re¯uxed with
1 mL of CF3CO2H in anhy. CH2Cl2 (25 mL) for 3 h.
After evaporation of the solvent on a rotavapor, a white
solid (1) was obtained which was recrystallized from 2-
Hemoglobin cross-linking studies. Hemoglobin was iso-
lated from red blood cells according to literature proce-
dure.15 The stroma-free hemoglobin was a gift from the
Walter Reed Army Institute of Research, Washington,
D. C. The Hb cross-linking was carried out at ambient
temperature (ꢁ25 ꢀC) under oxygenated conditions in a
PBS bu€er (phosphate-bu€ered saline solution made by
dissolving 8 g of NaCl, 1.15 g of Na2HPO4, 0.2 g of
KH2PO4, 0.1 g of MgCl2, 0.1 g of CaCl2, and 0.1 g of
KCl in a liter of H2O) (pH 7.4) using a fourfold molar
excess of reagent 1 (Na+ salt) as compared to Hb
(0.5 mM). After the reaction time of 1 h, the crude mix-
ture was dialyzed against 0.2 M glycine bu€er (pH 8.0).
The sample was then applied to a DEAE-cellulose col-
umn equilibrated with the same bu€er. The unreacted
propanol, yield (0.7 g, 92%), mp 230±232 ꢀC; H NMR
1
(DMSO-d6) d 8.00 (d, 4 H, Ar-H) 7.26 (d, 4 H, Ar-H),
4.90 (br s, 1 H, P-OH) 2.85 (m, 4 H, CH2), 2.05 (m, 4 H,
CH2); 13C NMR: d 170.83 (d), 166.58 (s), 154.00 (s),
130.82 (s), 128.35 (s), 121.98 (s), 26.99 (s), 23.97 (d); MS
(FAB) m/z 451 (MH+). Anal. calcd. for C20H19
.
PO10 H2O: C, 51.29; H, 4.52. Found: C, 51.26; H, 4.48.
Trisodium salt of bis[2-(4-carboxyphenoxy)carbonylethyl]
phosphinic acid. An ion exchange column (2.5Â25 cm),
packed with the resin AG 50W-X8 (H+ form), was
converted into the Na+ form by equilibrating with a
solution of 1 N NaOH (200 mL). Reagent 1 (0.07 g,

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
771
Hb and the cross-linked products were eluted using a
30±150 mM NaCl gradient. The spectrophotometric
readings of the eluted fractions at lmax 540 nm gave two
distinct peaks, corresponding to the unmodi®ed Hb and
the cross-linked product (see Figure 2). The fractions
were concentrated individually in an Ultra®ltration
CellR (Amicon). Both peaks were analyzed by HPLC
(Figure 3), SDS±PAGE (Figure 4), and IEF (Figure 5).
SDS±PAGE and IEF were run on Bio-RadR Mini Pro-
tean II and Mini IEF Cell, respectively. HPLC analyses
were performed on a Perkin-Elmer Series 400 Liquid
Chromatograph system (Vydac C4 protein column, pore
size 300 m, particle size 10 m, employing a gradient of 35±
50% CH3CN-H2O, containing 0.1% TFA, as an eluting
solvent with a ¯ow rate of 1 mL/min, over a period of
75 min. The MALDI MS analysis of the cross-linked
product (Figure 6) was performed as described under
Mass Specral Analyses below.
implemented in MATLAB (version 4.0 for Windows;
The Math Works, Inc., Natick, MA, USA). The partial
pressure of oxygen where hemoglobin was half-satu-
rated (P50) was calculated from the ®tted dissociation
curves by non-linear minimization, using the Gauss±
Newton method19 in MATLAB.
Tryptic digestion analyses. Tryptic digestion was per-
formed20 on an isolated cross-linked b-chain (0.2 mg) of
Figure 2, at a trypsin:protein concentration of 1:100 (w/
w), for 4 h in 0.1 M NH4HCO3 bu€er, pH 8.5. For
comparison, the tryptic digestion of the uncross-linked,
pure b-chain of Figure 3a was also carried out under the
same reaction conditions. The peptides were separated
by reverse phase HPLC on a Vydac C4 large-pore col-
umn, using a linear gradient (130 mins) between 5%
CH3CN in H2O containing 0.1% TFA (Solution A) and
50% CH3CN in H2O, containing 0.1% TFA (Solution
B), which ran at a ¯ow rate of 1 mL/min. The fragments
obtained were collected manually from the HPLC
column as they eluted out and lyophilized individually.
The HPLC elution pro®les of the uncross-linked and
cross-linked b-chains are shown in Figure 8a and b,
respectively.
Oxygen equilibrium (P50) measurements. The oxygen
equilibrium curves (Figure 7) were recorded by reoxy-
genation of nitrogen-equilibrated hemoglobin in 0.1 M
Bis-Tris bu€er (pH 7.4, 37 ꢀC) in the spectral cuvette of
a Hemox Analyzer (TCS Medical Products Co., Hun-
tington Valley, PA, USA). During each run, data was
downloaded to a computer for subsequent analysis.
Oxygen dissociation curves were ®tted to the Adair
equation16 by unconstrained non-linear optimization,
using the Nelder-Mead simplex search algorithm17,18
Mass spectral analyses. Mass spectral analyses of tryptic
fragments were carried out21 on a Kratos/Shimadzu
Kompact MALDI III Instrument (Manchester, UK),
equipped with a Nitrogen Laser (337 nm) and a desktop
Figure 2. The elution pro®le, monitored at lmax 540 nm, of the product of reaction of oxyhemoglobin with reagent BCCEP, separated
by chromatography on a DEAE-cellulose column. The ®rst peak at fraction numbers 14±20 is due to native Hb. The peak at fractions
52±60 is due to modi®ed Hb, cross-linked between b1±b2 subunits.

772
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
computer (Sun Computer Inc.). All experiments were
performed using a-cyano-4-hydroxycinnamic acid
(Aldrich) as the matrix. The matrix solution consisted of
a mixture of 7:3 aqueous 0.1% tri¯uoroacetic acid:-
acetonitrile. Each lyophilized tryptic fragment was dis-
solved in H2O (10 mL), and 0.5 mL of the solution was
applied to a sample well, followed by an equal volume of
the matrix solution. The samples were allowed to dry at
room temperature before placing into the spectrometer.
Results and Discussion
The reagent BCCEP (1) was designed through the use of
molecular modeling (Insight/Discover from MSI),
Figure 4. SDS±PAGE analyses of the modi®ed product
obtained by DEAE-cellulose chromatography. Lane 1: Mole-
cular weight markers, Lane 2: SFHb, Lane 3: cross-linked Hb
(retention time 50.53 min on HPLC).
Figure 3. HPLC chromatogram of (a) stroma free Hb (SFHb),
(b) cross-linked product, and (c) a mixture of SFHb and cross-
linked product. Peaks in (c) corresponding to retention times
48.48 min (b-chain), 50.53 min (b-b) and 54.33 (a-chain) were
characterized by SDS±PAGE, which showed molecular mass
ranges of 16 kDa, 32 kDa, and 16 kDa, respectively. The HPLC
was performed on a Vydac C4 protein column, pore size 300 m,
particle size 10 m, employing a 75 min linear gradient of 35±50%
CH3CN/H2O containing 0.1% tri¯uoroacetic acid, as an elut-
ing solvent, with a ¯ow rate of 1 mL/min.
Figure 5. Isoelectric focussing analysis (IEF) of stroma-free
hemoglobin (SFHb) and BCCEP-modi®ed hemoglobin.

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
773
Figure 6. The MALDI MS analyses of the b chains of Figure 3c.
mutants,26 self-association,27 and packing28 of hemo-
globins, the documented use of molecular modeling for
the de novo design of hemoglobin cross-linking reagents
is somewhat rare.
The initial structural framework for the cross-linking
reagent was laid down by assessment of the nature and
dimension of the DPG pocket, the identi®cation of the
target amino acids on the diagonally opposed b1 and b2
subunits for cross-linking, as well as the tether length,
geometry, and the electrophilic functional groups
required for ecient cross-linking. To this end, the
modeling studies were ®rst carried out using DPG itself
as the ligand. The extent and scope of potentially stabi-
lizing hydrophobic and hydrophilic Hb±reagent inter-
actions elicited by a variety of functional groups when
present on the reagent were then explored by perform-
ing energy minimizations and dynamics simulations on
each new version of the reagent±Hb complex so as to
arrive at the most suitable functional groups for intro-
duction into the reagent's skeletal framework. In this
context, it must be pointed out that several other
potentially promising reagents besides BCCEP were also
suggested by molecular modeling. The choice of 1 over
others for the present study, principally aimed at
exploring the feasibility of using molecular modeling
techniques for the design of hemoglobin cross-linking
reagents, was largely based on the anticipated facile
synthesis of the reagent.
Figure 7. The oxygen equilibrium curves of unmodi®ed
stroma-free Hb (SFHb) (1) as contrasted with those of a-a
cross-linked Hb (2), and BCCEP-cross-linked Hb (3) in 0.1 M
Bis-Tris, pH 7.4, at 37 ꢀC.
employing crystal coordinates of human HbA0,13
imported from the Brookhaven National Laboratory.
While molecular modeling techniques employing crystal
coordinates of Hb have been used for extensive investi-
gations of allosteric modi®cations of hemoglobin,22,23
drug delivery of hemoglobin bioconjugates,24 and bio-
physical characterizations of genetic variants,25

774
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
Figure 8. The HPLC pro®les of the tryptic digests of (a) pure b-chain, and (b) the cross-linked b1-XL-b2 chain. The peptides were
separated on a reverse phase Vydac C4 column, using a linear gradient of CH3CN-H2O, containing 0.1% CF3CO2H.
Reagent BCCEP (1) is equipped with two activated ester
functionalities to serve as the two cross-linking sites.
The two end carboxylate groups and the central phos-
phinic hydroxyl are anticipated to provide the necessary
anionic charges at biological pH for the molecule to be
speci®cally drawn to the b-cleft of hemoglobin. Initial
docking and energy-minimization studies revealed that
four di€erent possibilities existed for the reagent to align
itself within the b-cleft: (a) between b1-Val-1 and b2-Lys-
144, (b) between b1-Lys-82 and b2-Lys-144, (c) between
b1-Lys-132 and b2-Lys-144, and (d) between b1-Lys-82
and b2-Lys-82. The energy-minimization and molecular
dynamics simulation studies performed individually on
these four di€erent arrangements of the BCCEP-Hb
complex, suggested that the best ®t with least energy
(total Insight energy=737.097 kcals, average absolute
derivative=0.00012, standard deviation of absolute
derivative=0.00012, and average RMS derivative
=0.00017) and maximum reagent±Hb interactions (see
below) occurs when BCCEP is aligned between Lys-144
on the b2 subunit and midway to N-terminus Val-1 and
Lys-82 on the b1 subunit (intermediate between
arrangements a and b described above), as shown in
Figure 1. The energy-minimized BCCEP±Hb complex
with this arrangement revealed the existence of several
hydrogen-bonding interactions between the protein and

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
775
the ligand: (a) Reagent-p-Ph-CO2 . . .HO-Thr-87-b2
(1.45 A), (b) Reagent-P(O2)CH2CH2C(O)O. . .HN-His-
143-b2 (2.41 A), (c) Reagent-PO2 . . .H2N(")-Lys 144-b2
(1.92 A), (d) Reagent-p-Ph-CO2 . . .HN-(CO)-Asp-79-
b1, (2.22 A), (e) Reagent-p-Ph-CO2 (1). . .HN-(CO)-Leu-
81-b1 (1.93 A), (f) Reagent-p-Ph-CO2 (2). . .HN-(CO)-
Leu-81-b1 (1.97 A), (g) Reagent-p-Ph-CO2 (2). . .H2N(")-
Lys-82-b1 (1.86 A), and (h) Reagent-p-Ph-CO2 (2). . .
HN-(a)Lys-82-b1 (2.20 A). The two end phenyl groups
of the reagent ®t snugly into the two hydrophobic
pockets created by the following amino acids, 14 on
each side of the two b-subunits: Val-1, Leu-78, Asp-79,
Asn-80, Leu-81, Lys-82, Thr-84, Phe-85, Ser-89, Lys-
132, Ala-135, Val-133, Leu-141, and His-143 on the b1-
subunit, whereas Leu-78, Leu-81, Lys-82, Thr-84, Phe-
85, Ala-86, Thr-87, Ser-89, Ala-135, Gly-136, Val-137,
Ala-140, His-143, and Lys-144 on the b2-subunit.
The reagent was converted to its tri-sodium salt by ion-
exchange chromatography, and was used for cross-link-
ing human HbA0. Cross-linking was carried out under
oxygenated conditions at ambient temperature (ꢁ25 ꢀC)
in a PBS bu€er (pH 7.4), using a fourfold molar excess
of reagent 1 (Na+ salt) over Hb. The same reaction,
however, failed to proceed in deoxygenated media
inspite of otherwise identical experimental conditions.
After extensive dialysis of the reaction mixture, the
product was separated from the unmodi®ed Hb by
DEAE±cellulose chromatography (see Figure 2), con-
centrated, and analyzed by high performance liquid
chromatography (HPLC) (Figure 3), sodium dodecyl
sulfate-polyacrylamide gel electrophoresis (SDS±PAGE)
(Figure 4), and isoelectric focusing (IEF) (Figure 5). The
HPLC pro®le of the cross-linked product (Figure 3b), as
contrasted with that of the unmodi®ed Hb (Figure 3a),
suggested that the modi®cation was on the b-chain of
Hb. The SDS±PAGE analyses of the modi®ed product
(see Figure 4) clearly pointed to the presence of cross-
linked product with a molecular mass of 32 kDa, cor-
responding to the dimers of Hb subunits. The SDS±PAGE
analyses of the individual a and b chains of the modi®ed
product of Figure 3b revealed molecular masses of &16
and 32 kDa, respectively, con®rming that the modi®ca-
tion was on the b chain. The most convincing evidence
for the cross-link came from the MALDI MS analysis
of the b chains of Figure 3c which revealed (see Figure 6)
molecular masses corresponding to the unmodi®ed chain
(m/z 15874.89 Da) and the BCCEP-modi®ed b chain
[obsd. m/z 31923.93; calcd. m/z 31923.78=2Â15874.89
(two b chains)+176 (the cross-linking fragment -C(O)
(CH2)2-P(O)(OH)-(CH2)2-C(O)- of BCCEP)-2 (two "-
or a-amino H's removed upon cross-linking)].
A femtosecond molecular dynamics simulation (300 K,
5000 iterations) of the above energy-minimized BCCEP±
Hb complex in the preferred arrangement, soaked with
an aqueous layer of 5 A thickness all around, showed
the concerned lysine residues on each b-subunit, lying in
close proximity to the cross-linking sites of BCCEP. The
range of distance between the appropriate ester carbo-
nyl of BCCEP and the "-amino group of lysine residues
or the a-amino group of N-terminus valine, as com-
puted from the graphs derived from dynamics trajec-
tories, indicated that the closest contacts of b1-Val-1
(5.61 A), b1-Lys-82 (5.83 A), and b2-Lys-144 (3.62 A) are
all well within the combined Van der Waals' radii of the
two bonding atoms of the concerned amino and carbo-
nyl functions of Hb and BCCEP, respectively. Thus, the
most preferred amino acids to participate in cross-link-
ing by BCCEP appeared to be Val-1 and Lys-82 on the
b1-subunit and Lys-144 on the b2-subunit, which corre-
spond to arrangements (a) and (b) described above.
While the latter two are the suggested preferred
arrangements, our simulation studies did not totally rule
out the arrangements (c) and (d) described earlier, which
would require the involvement of an additional amino
acid on each of the b chains: Lys-132 on b1 and Lys-82
on b2. However, we did not ®nd any experimental evi-
dence, based on MALDI MS analyses of tryptic digests
of the cross-linked product (see below), supporting these
latter two arrangements.
The modi®ed hemoglobin was further analyzed for
oxygen anity (P50)29 characteristics at 37 ꢀC. For the
sake of comparison, P50 values were also measured
under the same conditions for the uncross-linked,
stroma-free Hb (SFHb) as well as for the a-a cross-
linked Hb.7 The graphs of log PO2 vs % Hb saturation
for (1) SFHb, (2) a-a cross-linked Hb7, and (3) the
BCCEP-modi®ed Hb are shown in Figure 7. The com-
puted P50 value for the modi®ed Hb was 31.3 Torr, as
contrasted with that of cell-free Hb (6.6 Torr) and the
a-a cross-linked Hb (27.9 Torr). Thus, the BCCEP-
modi®ed as well as the a-a cross-linked Hb's showed
signi®cantly lowered oxygen anity as compared with
cell-free hemoglobin. This was further attested by the
observed right-shifting of O2 equilibrium curves of both
cross-linked products relative to that of cell-free Hb.
Furthermore, the sigmoidal shapes of O2 curves of the
BCCEP-modi®ed Hb, comparable to those of the a-a
cross-linked Hb,7 pointed to reasonable retainment of
oxygen binding cooperativity after the cross-link for-
mation. Whereas the Hill Coecient n,30 generated
The reagent BCCEP (1) was synthesized (Scheme 1) by
condensation of p-t-butoxycarbonylphenyl acrylate (3),
prepared from p-t-butoxycarbonylphenol (2) by reaction
with acryloyl chloride, with hypophosphorous acid in
the presence of a silylating reagent, bis(trimethylsilyl)
acetamide. The product, upon removal of the t-butyl
protecting groups with tri¯uoroacetic acid in re¯uxing
methylene chloride, gave reagent BCCEP as a crystal-
line solid in 72% overall yield.

776
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
from the Hill plots of P50 values, would be a quantita-
tive indicator of oxygen-binding cooperativity, the
computed value of n=2.0 from such plots for the
BCCEP-modi®ed Hb is, nevertheless, less meaningful in
view of the anticipated and observed (see below) multi-
ple cross-linking sites, and the consequent heterogeneity
of the BCCEP-modi®ed Hb.
were directly observed, the fragments T10 (1421) and T11
(1126) could only be seen as a single peptide (T10+T11-
H2O) at m/z 2529. Likewise, fragment T12 (1720) was
observed either as a cysteine dimer (cystine) with a di-
sul®de bridge at m/z 3438 or as part of a three-peptide
fragment (T12+T13+T14 2H2O), resulting from
incomplete tryptic digestion, at m/z 4234. In view of the
absence of disul®de bridges in normal human hemoglo-
bin in the native state, the formation of cystine from
Cys-112 of T12 fragment may be due to air oxidation
during digestion, sample preparation, or analyses.
Another important observation in the MALDI mass
spectra was the presence of Na and K ions in a number
of observed fragment peaks. Since neither of these ions
was present in the matrix of mass spectral analysis or
during tryptic digestion, they must have come from the
phosphate bu€er solutions employed during the isola-
tion and puri®cation of hemoglobin from red cells. One
other important point to be noted is that the typical
mass accuracy available in conventional MALDI MS
analyses is one mass per thousand.32
In order to determine the location of the cross-links, the
tryptic digestion analyses of both the pure uncross-
linked b-chain and the cross-linked b1-XL-b2 chain were
carried out, employing standard procedures and reac-
tion conditions.20 The peptide fragments were separated
by reverse-phase HPLC on a Vydac C4 protein column.
A comparison of the HPLC pro®le of the tryptic frag-
ments of the pure b-chain (Figure 8a) with that of the
cross-linked product (Figure 8b) revealed that the new
peptide fragments were mainly clustered into two broad
humps, each consisting of 5±6 individual peaks, approxi-
mately centered at retention times of 89.71 and 99.44
minutes. In addition, the broad peak at the highest
region of the HPLC pro®les of Figure 8a and b, starting
at approximate retention time of 106 minutes, showed
some conspicuous di€erences.
The MALDI MS analyses of tryptic digest fragments of
the cross-linked b1-XL-b2 chain (see Figure 8b and
Table 2) provided evidence for the sites of cross-links in
the modi®ed hemoglobin. The most important clues
came from the major peaks of retention times 98.85 and
99.44 minutes. The observed MH+ ions from the two
peaks at m/z 6255 and m/z 6207, respectively (see Figure
9a,b and Table 2), corresponded to fragments (T1±T2)-
XL-(T12±T14) and (T8±T10)-XL-(T13±T15). These frag-
ments, along with other shorter, tell-tale fragments (see
below), undeniably suggested that the cross-links
involved an amino acid residue taken from each of the
three di€erent groups of tryptic peptides containing
residues 1±17 (T1±T2) (group I), 60±95 (T6±T10) (group
II), and 105±146 (T12±T15) (group III). Since the ®rst
two groups are each connected to the common third
group, it appeared that the cross-links involved one of
the amino acids from residues 1±17 or 60±95 on one of
the b chains, and an amino acid from residues 105±146
on the other b chain. This is indeed consistent with the
results of molecular modeling described above, which
suggested the possibility of cross-link formation
between Val-1 or Lys-82 on one of the b-chains, and
Lys-144 on the other (refer to arrangements a and b
above). While there are several other amino acid resi-
dues that are capable of cross-linking, besides Val-1
(group I), Lys-82 (group II), and Lys-144 (group III),
especially the lysines that are present in each of the three
groups, for example, Lys-8 and Lys-17 (group I), Lys-
61, Lys-65, Lys-66, and Lys-95 (group II), and Lys-120,
Lys-132 (group III), their considerably distant locations
(515 A) from the cross-linking sites of BCCEP in the
b-cleft render them the unlikely candidates for cross-
linking. Although cross-links are theoretically possible
The matrix-assisted laser desorption/ionization mass
spectrometry (MALDI MS)21 of each of the tryptic
fragments of Figure 8a and b was performed to deter-
mine their compositions. Where it was not possible to
isolate a totally pure fragment peak (e.g., the clusters of
retention times 89.71 and 99.44 mentioned above) the
MALDI MS were run on the peak of highest possible
purity, which normally consisted of two or three com-
ponents. However, the presence of multiple components
in a peak did not hinder determination of composition
of individual peptide fragments or cross-linking sites
using MALDI MS analyses. In this regard, a recent
report of successful MALDI MS analyses, carried out
directly on the crude tryptic digests of hemoglobin, is
noteworthy.21 Listed in Tables 1 and 2 are the retention
times and observed average masses (MH+ ions) by
MALDI MS analyses of predominant peptide fragments
of Figure 8a and b, respectively, along with the calcu-
lated average masses and tentative structural analyses of
each of those fragments.
The MALDI MS analyses of the tryptic digests of native
b-chain (see Figure 8a and Table 1) showed the expected
(MH+) ions for those tryptic peptides whose masses
exceeded 900 Da. The MALDI mass assignments are
normally dicult for peptides of masses considerably
lower than 1 kDa because of interfering low mass matrix
ions.31 Thus, tryptic fragments T6 (mass=245), T7
(411), T8 (146), and T15 (318) were not analyzed. While
ions corresponding to T1 (952), T2 (932), T3 (1314), T4
(1274), T5 (2059), T9 (1670), T13 (1378), and T14 (1149)

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
777
Table 1. MALDI MS analyses of the peptide fragments obtained from the tryptic digestion of the uncross-linked, pure b-chain of
human hemoglobin
Retention time (min) Calcd av. massa Obsd av. massb Fragment residues
Tentative structural analysis of tryptic fragment
17.59
35.28
35.96
36.48
41.20
43.46
952.1
974.1
990.2
951.9
974.3
990.2
1±8
1±8
1±8
T1
(T1-1)+Na
(T1-1)+K
1149.4
1171.4
1187.5
1149.8
1171.5
1187.1
133±144
133±144
133±144
T14
(T14-1)+Na
(T14-1)+K
1378.6
1400.5
1416.7
1378.4
1400.4
1416.4
121±132
121±132
121±132
T13
(T13-1)+Na
(T13-1)+K
1126.2
1314.4
1400.5
1126.7
1314.3
1400.4
96±104
18±30
121±132
T11
T3
(T13-1)+Na
1449.7
1471.7
1487.8
1450.0
1471.7
1488.1
133±146
133±146
133±146
(T14+T15) H2O
(T14+T15-H2O) 1+Na
(T14+T15-H2O) 1+K
932.1
954.1
970.2
932.6
954.5
970.0
9±17
9±17
9±17
T2
(T2-1)+Na
(T2-1)+K
58.93
59.73
61.16
1798.1
1274.5
1799.1
1274.4
66±82
31±40
(T8+T9) H2O
T4
2059.3
2081.3
2097.4
2059.7
2081.1
2097.1
41±59
41±59
41±59
T5
(T5-1)+Na
(T5-1)+K
61.71
66.29
87.86
98.45
1669.9
2529.8
4234.0
3438.2
1669.8
2529.2
4234.9
3439.6
67±82
83±104
T9
(T10+T11
)
H2O
105±144
(T12+T13+T14
)
2H2O+Na
2Â(105±120)
(T12)2±2 (dimer of T2 with a disul®de bridge)
aThe average mass of peptide fragment was calculated using the computer program, GPMAW (General Protein Mass Analyzer for
Windows), Ver. 2.0, available from Lighthouse Data, Engvej 35, DK-5230, Odense M, Denmark. The following are the calculated
masses for the peptide fragments resulting from tryptic digestion of the b-chain of hemoglobin, and were employed in all subsequent
computations involving these fragments: T1 (1±8), 952.08; T2 (9±17), 932.09; T3 (18±30). 1314.42; T4 (31±40), 1274.53; T5 (41±59),
2059.28; T6 (60±61), 245.33; T7 (62±65), 411.46; T8 (66±66), 146.19; T9 (67±82), 1669.90; T10 (83±95), 1421.59; T11 (96±104), 1126.24;
T12 (105±120), 1720.11; T13 (121±132), 1378.55; T14 (133±144), 1149.36; T15 (145±146), 318.34.
bThe observed mass was calibrated against protein nerve growth factor (PNGF), MH+=2004.3.
between b1-Lys-132 and b2-Lys-144 (arrangement c) and
between b1-Lys-82 and b2-Lys-82 (arrangement d), we
found no evidence for such cross-links in the MALDI
MS analyses of any of the tryptic digest fragments of
modi®ed hemoglobin (see Table 2). Therefore, we
believe that the cross-links lie between b1-Val-1 and b2-
Lys-144 (arrangement a) and between b1-Lys-82 and b2-
Lys-144 (arrangement b). Furthermore, in view of the
relatively large number of tryptic fragments that corre-
sponded to arrangement b (see Table 2), the cross-link
appears to be predominantly between b1-Lys-82 and b2-
Lys-144 (arrangement b).
[(66±82)-XL-(133±146)], 3856 [(62±82)-XL-(133±146)],
4068 [(60±82)-XL-(133±146)], 4505 [(66±82)-XL-(121±
144)], 5103 [(60±82)-XL-(121±144)], and 5888 [(66±95)-
XL-(121±144)] are consistent with the notion that the
cross-link is between one of the amino acids from resi-
dues 60±95 and one from residues 121±146. Also, the
fragment ions at m/z 5583 and m/z 3198 of respective
retention times 100.09 and 115.43 min corresponded to
the cross-linked fragments (67±104)-XL-(133±144) and
(66±82)-XL-(133±144), respectively. Lys-82 and Lys-144
were also part of the cross-linked tryptic fragments, (66±
82)-XL-(105±144) and (66±95)-XL-(121±146), which
matched the observed MH+ ions at m/z 6267 and m/z
6286, respectively, both of which came from the tryptic
digest peak of retention time 97.43 min.
The other peptide fragments of Figure 9a corroborated
the above notion. The observed MH+ ions at m/z 3426

778
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
779

780
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
Figure 9. The MALDI MS analyses of tryptic fragments of retention times (a) 98.85 min, and (b) 99.44 min.
Several other fragment ions listed in Table 2 gave fur-
ther insight into not only the process of cross-linking
but also the subsequent hydrolysis that seemed to occur
in a number of instances. Many of the above-mentioned
tryptic peptides involved in cross-linking also exhibited
ions that indicated that the peptides contained the
reagent BCCEP either as a covalent mono adduct or as
the product of hydrolysis of the latter. For example, the
observed fragment ion at m/z 1287 from the HPLC peak
of retention time 60.32 min corresponded to the cova-
lent mono adduct of BCCEP with T1. Similarly, the
fragment ions at m/z 2002 and m/z 2026 of respective

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
781
retention times 60.32 and 54.25 min corresponded to the
covalent monoadduct of mono- and disodium salt of
BCCEP with T9, respectively. The partially hydrolyzed
T1-BCCEP and T2-BCCEP adducts were seen at m/z
1145 and m/z 1907 at retention time 40.02 and 42.66
minutes, respectively. Likewise, the HPLC peaks of
retention time 44.93 min revealed several ions that mat-
ched the tryptic fragment of residues 133±146 (T14±T15)
with a partially hydrolyzed BCCEP as a covalent mono
adduct. The latter was seen in several di€erent forms
(see Table 2): as the parent ion at m/z 1641, a mono
sodium salt at 1663, a mono potassium salt at 1679, and
a dipotassium salt at 1721. The same adduct was also
seen as a monosodium±monopotassium salt at m/z 1701
at retention time 42.66 min.
hydrolysis of the intact mono adducts would produce
the partially hydrolyzed mono adducts of T1 and T9,
along with the respective cross-linker-free fragments.
Finally, in view of formation of a number of partially
digested peptide fragments from apparently incomplete
tryptic digestion of the cross-linked product, it was
decided to repeat the experiment by extending the time
of digestion to 24 h from the 4 h duration that was
employed earlier, although incomplete tryptic digestion
of modi®ed hemoglobin has been documented.21 Since
the digestion conditions employed for both native and
cross-linked b chains have been nearly identical, it seems
that the modi®cation of an amino acid residue by
BCCEP causes diculty for trypsin to access the vicinity
of the modi®ed residue, resulting in partially digested
fragments. If this is the case, the products obtained from
the prolonged digestion should not be much di€erent
from the ones obtained from the 4 h digestion. If, on the
other hand, the complete hydrolysis of the reagent
occurred during the prolonged period of digestion,
resulting in liberation of cross-linker-free peptide frag-
ments as discussed above, then trypsin should be able to
further degrade these peptides into smaller fragments in
the usual manner. Indeed, the MALDI MS of the tryp-
tic fragments obtained from the 24 h tryptic digestion
turned out to be a lot less informative than the ones
from the 4 h digestion. Most of the cross-linked tryptic
fragments obtained earlier from the shorter digestion
were now absent, and many of the observed ions were
found to be due to self-digestion of trypsin. It appears,
therefore, that the cross-link formed by BCCEP is prone
to hydrolysis under conditions employed (pH 8.5) for
tryptic digestion. Nevertheless, the cross-linked product
was found to be stable for inde®nite periods of time
when suspended in a PBS (pH 7.4), as evidenced by
HPLC pro®les taken at intervals of several days. As
there are no prior reports of hydrolysis, to the best our
knowledge, of such cross-linkers that form bis-amide
linkages with hemoglobin subunits, we speculate
that the central phosphinic acid might be playing a
crucial role in facilitating the process of hydrolysis
by providing an anchimeric assistance via formation of
a ®ve-membered ring intermediate. Nonetheless, this
can only be a tentative explanation in the absence of
careful and thorough scrutiny of the mechanism of
hydrolysis.
The above tryptic fragments that correspond to covalent
mono adducts of BCCEP with T1 (m/z 1287) and T9 (m/z
2002 and 2026) lend further support to the notion
based on molecular modeling described earlier (see Fig-
ure 1) that one end of the reagent BCCEP aligns itself
midway between the N-terminus Val-1 (T1) and Lys-82
(T9) on one of the b chains, while the other end lies close
to Lys-144 (T14) on the other b chain. In this orienta-
tion, there is an equal probability that at any one
moment, both Val-1 and Lys-82 belonging to the same
subunit will be modi®ed by BCCEP forming a mono
adduct each. However, only one of the mono adducts
can form a cross-link with Lys-144 of the other b-sub-
unit at any instant, leaving the other monodduct intact.
This is further corroborated by the fact that no such
intact covalent mono adducts of BCCEP were observed
with Lys-144 (T14).
While there were only three intact covalent mono
adducts of BCCEP with T1 and T9, as described above,
there was an abundance of partially hydrolyzed BCCEP
adducts with T1 and T9, as well as with T14. These spe-
cies could arise from hydrolysis of the above intact
mono adducts as well as from that of the cross-linked
products, presumably during tryptic digestion, HPLC
separation or sample preparation for MALDI MS ana-
lysis. While the latter two processes employed acidic
conditions (0.1% TFA), the ®rst one was performed
under a basic medium (pH 8.5). Although it is not clear,
and was not explored further, as to at what stage it
might have taken place, the hydrolysis could, never-
theless, satisfactorily explain why the cross-linker-free
tryptic fragments of T1, T9, and T14 were detected in
spite of their implicated involvement in cross-linking.
Apparently, upon hydrolysis, the cross-linked product
would initially produce a partially hydrolyzed mono
adduct, containing a tryptic fragment from either the
b1 or b2 side plus the one-end-hydrolyzed reagent part.
A second hydrolysis of such a mono adduct would pro-
duce the reagent-free peptide, as observed. Similar
Conclusion
We have designed and synthesized a novel bifunctional
organic reagent aided by modern molecular modeling
and synthetic organic methods. The reagent BCCEP has
a number of desired characteristics for a hemoglobin
cross-linker. It is, like DPG, speci®c for the b-cleft

782
R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
modulation of hemoglobin. It can be conveniently syn-
thesized in three easy steps from readily available, in-
expensive starting materials, has an inde®nite shelf life
as a crystalline solid, can be easily converted into its tri-
sodium salt for usage in an aqueous medium, and
employed to cross-link Hb under ambient oxygenated
media, requiring no special precautions or reaction
conditions. The oxygen anity of the cross-linked pro-
duct is considerably lower (P50=31.3 Torr) as compared
with that of cell-free Hb (P50=6.6 Torr), and is some-
what comparable to the oxygen anity of whole blood
(P50=27 Torr). The cross-linked product also retains
some of the oxygen-binding cooperativity characteristics
of the native protein, as revealed by the sigmoidal shape
of O2 curve that is comparable to that of the a-a cross-
linked hemoglobin.7 We have also successfully eluci-
dated the sites of cross-links, employing the MALDI
MS analyses of tryptic digests of the cross-linked
product.
advice and assistance in obtaining the MALDI MS spec-
tra, and Colonel John Hess of the Blood Detachment
Center, Walter Reed Army Institute of Research for the
generous gift of stroma-free hemoglobin employed in our
studies. The reported EI/CI/FAB mass spectra were run
at the Michigan State University Mass Spectral Facility,
supported in part by a grant (no. P41RR00480-0053)
from the National Institutes of Health.
References and Notes
1. (a) Winslow, R. M.; Vandegri€, K. D.; Intaglietta, M. Blood
Substitutes: Physiological Basis of Ecacy; Birkhauser: Boston,
1995. (b) see Abstracts of the VI International Symposia on
Blood Substitutes, Montreal, Quebec, Canada, August 5±7,
1996. (c) Kluger, R. Can. J. Chem. 1994, 72, 2193.
2. Miller, I. F. In CRC Critical Reviews in Bioengineering;
1978; pp 149±177.
3. Bolin, R. B.; Beyer, R. P.; Nemo, G. J. In Advances in Blood
Substitutes Research; Alan R. Liss: New York, 1983.
Further structural modi®cation of BCCEP must take
into consideration that while the reagent is very speci®c
for the b-cleft, and is also speci®c for Lys-144 of one of
the b subunits, it has somewhat less speci®city for the
residues lying on the other b subunit, interacting with
both Val-1 and Lys-82. Consequently, achievement of
absolute speci®city, if necessary, may involve interaction
of the reagent with only one of the latter two amino acid
residues from the same Hb subunit, although it is not
obvious at this point if there is any distinct advantage of
achieving such a goal beyond that of the cleft-speci®city,
provided that the cross-linking reaction is uniformly
reproducible and that the oxygen anity as well as the
oxygen-binding cooperativity of the modi®ed Hb are
adequate. In that regard, while the oxygen anity of
BCCEP is in the desired range for a blood substitute, its
cooperativity characteristics could further bene®t from
structural improvements. In any event, our preliminary
molecular modeling exercises call for interchanging and/
or adding carboxy groups to the ortho and/or meta
positions of the terminal phenyl groups of BCCEP so as
to potentially realize the described total speci®city.
4. Winslow, R. M. Hemoglobin-based Red Cell Substitutes;
Johns Hopkins: Baltimore, 1992.
5. (a) Boyiri, T.; Safo, M. K.; Danso-Danquah, R. E.; Orringer,
E.; Criss, A.; Abraham, D. J. Colloq. INSERM 1995, 234, 553.
(b) DeVenuto, F. and Zegna, A. J. Surg. Res. 1983, 34, 205.
6. Mok, W.; Chen, D.-E.; Mazur, A. Fed. Proc. 1975, 34, 1458.
7. Snyder, S. R.; Welty, E. V.; Walder, R. Y.; Williams, L. A.;
Walder, J. A. Proc. Natl. Acad. Sci. U.S.A. 1987, 84, 7280.
8. (a) Kluger, R.; Song, Y.; Wodzinska, J.; Head, C.; Fujita,
T.; Jones, R. T. J. Am. Chem. Soc. 1992, 114, 9275. (b) Jones,
R. T.; Shih, D. T.; Fujita, T. S.; Song, Y.; Xiao, H.; Head, C.;
Kluger, R. J. Biol. Chem. 1996, 271, 675. (c) Paal, K.; Jones, R.
T.; Kluger, R. J. Am. Chem. Soc. 1996, 118, 10380.
9. Kavanaugh, M. P.; Shih, D. T.-B.; Jones, R. T. Biochemistry
1988, 27, 1804.
10. Hosmane, R. S.; Bertha, C. M. Biochem. Biophys. Res.
Commun. 1990, 166, 567.
11. (a) Benesch, R. E.; Kwong, S. Biochem. Biophys. Res.
Commun. 1988, 156, 9. (b) Benesch, R.;E.; Kwong, S. J. Protein
Chem. 1991, 10, 503. (c) Benesch, R. E.; Kwong, S. J. Biol.
Chem. 1990, 265, 14881.
12. (a) Kluger, R.; Wodzinska, J.; Jones, R. T.; Head, C.;
Fujita, T.; Shih, D. T. Biochemistry 1992, 31, 7551. (b) Jones,
R. T.; Head, C. G.; Fujita, T. S.; Shih, D. T.; Wodzinska, J.;
Kluger, R. Biochemistry 1993, 32, 215. (c) Kluger, R.; Shen, L.;
Xiao, H.; Jones, R. T. J. Am. Chem. Soc. 1996, 118, 8782.
Acknowledgements
This paper is dedicated to Professor Nelson J. Leonard
on the occasion of his eighty-®rst birthday. The research
was supported by a grant (no. R01 HL48632) from the
National Heart, Lung and Blood Institute of the
National Institutes of Health. We thank Dr A. Seethar-
ama Acharya and Ashok Molavalli of Albert Einstein
College of Medicine and Professor Robert F. Steiner of
the Department of Chemistry and Biochemistry, UMBC,
for their assistance in hemoglobin cross-linking studies,
Professor Catherine Fenselau and Dr Yetrib Hathout of
the Structural Biochemistry Center, UMBC, for their
13. Coordinates of carbonmonoxy HbA0 were employed for
the molecular modeling studies; see Silva, M. M.; Rogers, P.
H.; Arnone, A. J. Biol. Chem. 1992, 267, 17248.
14. Delaney, E. J.; Massil, S. E.; Shi, G-Y; Klotz, I. M. Arch.
Biochem. Biophys. 1984, 228, 627.
15. Masala, B.; Manca, L. In Methods in Enzymology, Everse,
J.; Winslow, R. M.; Vandegri€, K. D., Eds; 1994; Vol. 231, pp
21±31.
16. Imai, K. In Methods in Enzymology, Antonini, E.; Rossi-
Bernardi, L.; Chiancone, E., Eds; Academic: New York, 1981;
Vol. 76, p 471.

R. S. Hosmane et al./Bioorg. Med. Chem. 6 (1998) 767±783
783
17. Nelder, J. A.; Mead, R. Computer J. 1964, 7, 308.
24. Trimble, S. P.; Marquardt, D.; Anderson, D. C. Biocon-
jugate Chem. 1997, 8, 416.
18. Dennis, J. E., Jr.; Woods, D. J. In New Computing Environ-
ments: Microcomputers in Large-Scale Computing; Wouk, A.,
Ed.; SIAM, 1987; pp 116±122.
25. Pechik, I.; Ji, X.; Fidelis, K.; Karavitis, M.; Moult, J.; Bri-
nigar, W. S.; Fronticelli, C.; Gilliland, G. L. Biochemistry 1996,
35, 1935.
19. Dennis, J. E., Jr. In Nonlinear Least Squares, State of the
Art in Numerical Analysis; Jacobs, D., Ed.; Academic: 1977; pp
269±312.
26. Wajcman, H.; Kister, J.; Prome, D.; Blouquit, Y.; Prehu,
C.; Poyart, C.; Galacteros, F. C. R. Acad. Sci., Ser. III 1995,
318, 785.
20. Lee, T. D.; Shively, J. E. In Methods in Enzymology; Aca-
demic: New York, 1990; Vol. 193, pp 361±374.
27. Scaad, O.; Vallone, B.; Edelstein, S. J. C. R. Acad. Sci.,
Ser. III 1993, 316, 564.
21. Yang, T.; Horejsh, D. R.; Mahan, K. J.; Zaluzec, E. J.;
Watson, T. J.; Gage, D. A. Anal. Biochem. 1996, 242, 55.
28. Cretegny, I.; Edelstein, S. J. J. Mol. Biol. 1993, 230, 733.
29. P50 is de®ned as the partial pressure of oxygen to saturate
50% of Hb.
22. (a) Boyiri, T.; Safo, M. K.; Danso-Danquah, R. E.; Kister,
J.; Poyart, C.; Abraham, D. J. Biochemistry 1995, 34, 15021. (b)
Abraham, D. J.; Safo, M. K.; Boyiri, T.; Danso-Danquah, R.
E.; Kister, J.; Poyart, C. Biochemistry 1995, 34, 15006.
30. Cooperativity of oxygen binding is measured by the Hill
Coecient, n, which is computed by the Hill Equation,
Y=(PO2)n/[(P50)n+(PO2)n], where Y is the fraction of Hb
saturated. The value of n in whole blood ranges 2.3±2.5.2
23. (a) Randad, R. S.; Mahran, M. A.; Mehanna, A. S.;
Abraham, D. J. J. Med. Chem. 1991, 34, 752. (b) Wireko,
F. C.; Kellogg, G. E.; Abraham, D. J. J. Med. Chem. 1991, 34,
758.
31. Biemann, K. Annu. Rev. Biochem. 1992, 61, 977.
32. Beavis, R. C.; Chait, B. T. Anal. Chem. 1990, 62, 1836.