Mechanism of Site-Directed Protein Cross-Linking. Protein-Directed Selectivity in Reactions of Hemoglobin with Aryl Trimesates

Article abstract of DOI:10.1021/jo991514n

Site-directed cross-linking of hemoglobin has become an efficient way to produce a structurally defined altered protein with desirable functional properties. The reagent trimesoyl tris(3,5-dibromosalicylate) (1) introduces a bis amide cross-link derived from the ∈-amino groups of the side chains of the two β-Lys-82 residues in human hemoglobin. The basis of its specificity was investigated using a set of analogues of 1 (2-12). There are marked differences in the reaction patterns of these compounds with amino groups in hemoglobin compared to reactions with n-propylamine. The compounds that effectively modify the protein contain a carboxyl group ortho to the phenolic oxygen of the ester, while materials with meta or para carboxyl groups give little or no reaction. In contrast, the reactions with n-propylamine are slowest with the ortho carboxyl materials. Addition of the unreactive compound 5 to a solution containing hemoglobin reduces the ability of 1 to modify the protein, showing that the unreactive compound binds but does not react. On the basis of these observations and the known reaction patterns of salicylates, it is clear that the environment in the protein controls the reaction, regardless of the inherent reactivity of the reagent. We propose that the carboxyl group positions the reagent critically within the protein. Only the ortho arrangement permits transfer of the acyl function to the nucleophile.

Full text of DOI:10.1021/jo991514n

214
J . Org. Chem. 2000, 65, 214-219
Mech a n ism of Site-Dir ected P r otein Cr oss-Lin k in g.
P r otein -Dir ected Selectivity in Rea ction s of Hem oglobin w ith Ar yl
Tr im esa tes
Ronald Kluger* and Vittorio De Stefano
Lash Miller Laboratories, Department of Chemistry, University of Toronto,
Toronto, Ontario, Canada M5S 3H6
Received September 27, 1999
Site-directed cross-linking of hemoglobin has become an efficient way to produce a structurally
defined altered protein with desirable functional properties. The reagent trimesoyl tris(3,5-
dibromosalicylate) (1) introduces a bis amide cross-link derived from the ꢀ-amino groups of the
side chains of the two â-Lys-82 residues in human hemoglobin. The basis of its specificity was
investigated using a set of analogues of 1 (2-12). There are marked differences in the reaction
patterns of these compounds with amino groups in hemoglobin compared to reactions with
n-propylamine. The compounds that effectively modify the protein contain a carboxyl group ortho
to the phenolic oxygen of the ester, while materials with meta or para carboxyl groups give little
or no reaction. In contrast, the reactions with n-propylamine are slowest with the ortho carboxyl
materials. Addition of the unreactive compound 5 to a solution containing hemoglobin reduces the
ability of 1 to modify the protein, showing that the unreactive compound binds but does not react.
On the basis of these observations and the known reaction patterns of salicylates, it is clear that
the environment in the protein controls the reaction, regardless of the inherent reactivity of the
reagent. We propose that the carboxyl group positions the reagent critically within the protein.
Only the ortho arrangement permits transfer of the acyl function to the nucleophile.
Reagents that introduce cross-links into hemoglobin by
hemoglobin.5 Klotz suggests that the bromine substitu-
ents specifically increase electrostatic interactions be-
tween the reagent and protein groups.5
highly selective reaction processes create structurally
defined altered proteins with properties that can be
related to structure.1 Trimesoyl tris(3,5-dibromosalicy-
late) (1) is a remarkably specific and efficient reagent,
reacting with hemoglobin to produce a cross-linked
product (linked between ꢀ-amino groups of lysine-82 side
chains in the â subunits) with the third ester site either
intact or hydrolyzed.2 The leaving group in 1 is a salicylic
acid derivative, containing a carboxyl group adjacent to
the phenolic ester oxygen. The adjacent carboxyl is
expected to facilitate the hydrolysis of the ester but not
the addition of amines, such as the side chain amino
group in a protein.3 Since the residual ester is useful for
making interesting derivatives of cross-linked hemoglo-
bin,4 hydrolysis is an undesired reaction. Isomers of
salicylates, in which the carboxyl is para or meta to the
phenolic oxygen are less susceptible to hydrolysis than
are salicylates,3 but we do not know their reactivity with
the protein. On the basis of this model, a para-substituted
analogue of 1 should be more resistant to hydrolysis and
equally reactive toward amines.
We have prepared a set of isomers and analogues of
(1) to probe the origin of its specificity and reactivity with
hemoglobin. How do its selectivity and reactivity relate
to the substitution pattern on the salicylate leaving
group? Does the ortho carboxyl have a specific effect in
the reaction of the ester with amino groups of the protein?
Do the halogen substituents in triesters have especially
enhanced reactivity? The structures of the compounds we
have prepared are summarized below.
The contrasting reaction patterns of the amino groups
in the protein and in n-propylamine with 1-12 reveal
fundamental information about the effects of functional
group arrays on the leaving group in the reaction of the
amino groups in the protein with esters.
Exp er im en ta l Section
Solvents used in syntheses were dried prior to use. Reagents
for buffers and for chromatography were of analytical grade
or better. Solutions of human hemoglobin A were obtained
from Hemosol, Inc. The structure of newly synthesized materi-
als was assessed by a combination of NMR spectroscopy, mass
spectrometry, infrared spectroscopy, and ion spray mass
spectroscopy. Proton and carbon NMR spectra were recorded
at 200 and 100 MHz, respectively.
We found that elemental analysis did not give reproducible
results with the final products (triacids). Thus, in all cases,
purity was assessed by HPLC analysis and the structure was
analyzed spectroscopically.
The reactivity of 1 with hemoglobin should be signifi-
cantly affected by its bromine substituents. While bro-
mine atoms as substituents on the aromatic ring of the
salicylate leaving group increase the reactivity of acetyl
salicylates toward amine nucleophiles in general, the
substituent effect is much greater with amino groups of
(1) Kluger, R. Chemical cross-linking and protein function; Creigh-
ton, T. E., Ed.; IRL Press: Oxford, U.K., 1997; pp 185-214.
(2) Kluger, R.; Song, Y.; Wodzinska, J .; Head, C.; Fujita, T. S.; J ones,
R. T. J . Am. Chem. Soc. 1992, 114, 9275-9279.
(3) St. Pierre, T.; J encks, W. P. J . Am. Chem. Soc. 1968, 90, 3817-
3287.
Tris(3,5-dibromosalicyl) trimesate (1) was prepared by the
published procedure.2 All other tris trimesate esters were
(5) Walder, J . A.; Zaugg, R. H.; Iwaoka, R. S.; Watkin, W. G.; Klotz,
(4) Kluger, R.; Song, Y. J . Org. Chem. 1994, 59, 733-736.
I. M. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 5499-503.
10.1021/jo991514n CCC: $19.00 © 2000 American Chemical Society
Published on Web 12/16/1999

Hemoglobin Reactions with Aryl Trimesates
J . Org. Chem., Vol. 65, No. 1, 2000 215
prepared by variations on this route, which involves coupling
the tert-butyl salicylate ester with an activated form of trimesic
acid, followed by removal of the tert-butyl group.
(s), 164.09 (s), 151.60 (s), 136.14 (s), 134.95 (s), 132.72 (s),
132.40 (s), 127.32 (s), 124.77 (s), 124.39 (s).
Trimesoyl tris(5-bromosalicylate) (3) was prepared from tert-
butyl 5-bromosalicylate, which was prepared from the acid as
tert-Butyl salicylate was prepared as a precursor for tris
(salicyl) trimesate (2). DCC (8.0 g, 0.039 mol) dissolved in dry
THF (50 mL) was added dropwise over 30 min to a stirred
suspension of salicylic acid (5.00 g, 36.1 mmol) and N,N-
dimethylaminopyridine (DMAP) (0.17 g, 0.0014 mol) in dry
tert-butyl alcohol (125 mL). The mixture was stirred at room
temperature overnight and then concentrated. The residue was
stirred in diethyl ether (50 mL), and oxalic acid (5.3 g, 0.059
mol) was introduced in portions to decompose excess DCC and
precipitate DMAP. The mixture was filtered, and the filtrate
was washed with an aqueous solution with three 40 mL
portions of 0.3 M sodium bicarbonate solution. The organic
solution was concentrated to dryness, leaving an oil, which
was purified using flash chromatography (silica gel in dichlo-
romethane). Yield: 6.0 g (86%) of tert-butyl salicylate. 1H NMR
(CDCl3): δ 1.62 (s, 9H, C(CH3)3), 6.82 (dt, 1H, ArH), 6.95 (dd,
1H, ArH), 7.42 (dt, 1H, ArH), 7.77 (dd, 1H, ArH), 11.05 (s, 1H,
ArOH).
1
above in 91% yield. H NMR (CDCl3): δ 1.6 (s, 9H, C (CH3)3),
6.84 (d, 1H, ArH), 7.48 (dd, 1H, ArH), 7.84 (d, 1H, ArH), 11 (s,
1H, ArOH). The protected tris ester, trimesoyl tris(1-(tert-
butoxycarbonyl)-5-bromosalicylate) was cleaved with TFA,
giving 3 (24% yield from monoester, mp 214 °C). 1H NMR
(DMSO-d6): δ 7.48 (dd, 3H, ArH), 7.95 (dd, 3H, ArH), 8.12 (d,
3H, ArH), 9.02 (s, 3H, ArH). IR (KBr): 1750 (s), 1708 (s), 1476
(m), 1202 (s) cm-1. MS (negative FAB): 803 (M - 1). 13C NMR
(DMSO-d6): δ 164.09 (s), 162.83 (s), 149.08 (s), 136.72 (s),
135.09 (s), 133.83 (s), 130.89 (s), 126.30 (s), 125.90 (s), 118.83
(s). Anal. Calcd for C30H15O12Br: C, 46.36; H, 1.56. Found: C,
46.09; H, 1.85.
Trimesoyl tris(3,5-dichlorosalicylate) (4) was prepared by a
similar sequence. tert-Butyl(3,5-dichlorosalicylate) was pre-
pared by DCC-DMAP coupling of tert-butyl alcohol and 3,5-
dichlorosalicylic acid in 64% yield. 1H NMR (CDCl3): δ 1.60
(s, 9H, C(CH3)3), 7.54 (dd, 1H, ArH), 7.66 (dd, 1H, ArH), 11.48
(s, 1H, ArOH). The material was neutralized with potassium
tert-butoxide and combined with trimesoyl trichloride in THF
(15 mL). Extraction, drying, and concentration were followed
by removal of the tert-butyl groups with trifluoracetic acid. The
product was collected by filtration as a white powder (34%
yield; mp 175 °C). 1H NMR (acetone-d6): δ 8.02 (dd, 3H, ArH),
8.07 (dd, 3H, ArH), 9.40 (s, 3H, ArH). IR (KBr): 1757 (m), 1701
(s), 1455 (m), 1209 (s) cm-1. MS (negative FAB): 773 (M - 1).
13C NMR (acetone-d6): δ 163.62 (s), 162.60 (s), 146.52 (s),
136.82 (s), 134.58 (s), 132.77 (s), 131.68 (s), 131.15 (s), 130.61
(s), 127.59 (s).
Potassium tert-butoxide (0.50 g, 0.0025 mol) was added to
a solution of tert-butyl salicylate (0.28 g, 0.0025 mol) in
anhydrous THF (30 mL). The mixture was stirred at room
temperature for 15 min. A solution of trimesoyl trichloride
(0.22 g, 0.000 85 mol) in anhydrous THF (15 mL) was added
dropwise over 15 min. The mixture was stirred overnight at
room temperature and then concentrated. The solid was
dissolved in diethyl ether (35 mL), and the mixture was
washed with water (2 × 20 mL). The organic phase was dried
and concentrated to give trimesoyl tris(1-(tert-butoxycarboxy-
carbonyl)salicylate). The white solid was dissolved in anhy-
drous trifluoroacetic acid (20 mL) and stirred at room tem-
perature for 1 h. Ether (20 mL) was added, and the solution
was kept overnight at 4 °C. The product (tris(salicyl) trimesate
(2)) was collected by filtration as white crystals (0.36 g, 75%;
mp >240 °C). 1H NMR (acetone-d6): δ 7.5 (dt and dd, 6H, ArH),
7.76 (dt, 3H, ArH), 8.15 (dd, 3H, ArH), 9.18 (s, 3H, ArH). IR
(KBr): 1757 (m), 1708 (s), 1609 (m), 1209 (s) cm-1. MS
(negative FAB): 569 (M - 1). 13C NMR (acetone-d6): δ 165.50
Tris(4-carboxylphenyl) trimesate (5) was prepared using
tert-butyl-4-hydroxybenzoate, which was prepared by DCC
coupling of tert-butyl alcohol and p-hydroxybenzoic in 45%
yield: 1H NMR (CDCl3): δ 1.58 (s, 9H, C(CH3)3), 6.40 (broad,
s, 1H, ArOH), 6.83 (dd, 2H, ArH), and 7.90 (dd, 2H, ArH).
Using the general procedure described above, the product was
1
obtained in 25% yield (mp >240 °C). H NMR (DMSO-d6): δ
7.52 (d, 6H, ArH), 8.08 (d, 6H, ArH), 9.05 (s, 3H, ArH). IR

216 J . Org. Chem., Vol. 65, No. 1, 2000
Kluger and Stefano
(KBr): 1750 (s), 1694 (s), 1602 (m), 1209 (s) cm-1. MS (negative
FAB): 569 (M - 1). 13C NMR (DMSO-d6): δ 166.43 (s), 162.41
(s), 153.59 (s), 135.19 (s), 130.91 (s), 130.54 (s), 128.82 (s),
122.02 (s).
Tris(4-carboxyl-2-chlorophenyl) trimesate (6) was prepared
by the same procedures.
3086, 1711, 1213 cm-1
129.28, 130.90, 132.64, 134.70, 135.14, 153.06, 162.75, 164.68,
165.83.
.
13C NMR (DMSO-d6): δ 124.02, 124.57,
Tris(2,6-dicarboxylphenyl) trimesate (10) was prepared as
above from di-tert-butyl-2-hydroxyisophthalate. 2-Hydroxy-
isophthalic acid6 was converted to the tert-butyl ester7 in 56%
1
yield. H NMR (CDCl3): δ 1.80 (s, 18H, C(CH3)3), 6.82 (t, 1H,
tert-Butyl 3-chloro-4-hydroxybenzoate was prepared using
DCC coupling as a white powder in 75% yield. 1H NMR
(CDCl3): δ 1.60 (s, 9H, C(CH3)3), 5.95 (broad, s, 1H, ArOH),
7.20 (dd, 1H, ArH), 7.83 (dd, 1H, ArH), 7.97 (d, 1H, ArH).
Reaction with trimesoyl trichloride and removal of the tert-
butyl groups gave the product as a white powder in 43% yield
(mp 138 °C). 1H NMR (acetone-d6 with sonication): δ 7.76 (dd,
3H, ArH), 8.15 (dd, 3H, ArH), 8.23 (d, 3H, ArH), 9.29 (s, 3H,
ArH), 7.85 (d, 2H, ArH), 11.90 (s, 1H, ArOH). The ester was
combined with trimesoyl trichloride and the product converted
to the triacid with trifluoroacetic acid. The residue (10) is a
pale brown powder (mp 225 °C, dec) in 7% yield. 1H NMR
(DMSO-d6): δ 6.65 (t, 3H, ArH), 7.87 (d, 6H, ArH), 8.64 (s,
3H, ArH). IR (KBr, cm-1): 3450, 1704, 1252. 13C NMR (DMSO-
d6): δ 165.89, 168.41, 169.26, 113.92, 117.43, 131.94, 133.61,
135.50.
ArH). IR (KBr): 1757 (m), 1708 (s), 1420 (m), 1202 (s) cm-1
.
MS (negative FAB): 671 (M - 1). 13C NMR (DMSO-d6): δ
165.37 (s), 161.41 (s), 149.49 (s), 135.63 (s), 130.85 (s), 130.81
(s), 130.70 (s), 130.16 (s), 129.62 (s), 126.06 (s), 124.59 (s).
Tris(3,5-dicarboxylphenyl) trimesate (11) was prepared from
di-tert-butyl-5-hydroxyisophthalate, which was prepared from
5-hydroxyisophthalic acid as above (white solid, 50% yield).
1H NMR (CDCl3): δ 1.60 (s, 18H, C(CH3)3), 7.11 (s, 1H, ArOH),
7.77 (d, 2H, ArH), 8.11 (t, 1H, ArH). This was combined with
trimesoyl trichloride and then treated with trifluoroacetic acid
to give 11 as white crystals, mp >250 °C (dec). Yield: 30%.
1H NMR (DMSO-d6): δ 8.21 (s, 6H, ArH), 8.44 (s, 3H, ArH),
Tris(4-carboxyl-3-chlorophenyl) trimesate (7) was prepared
from tert-butyl 2-chloro-4-hydroxybenzoate. The latter was
prepared in 15% yield. 1H NMR (DMSO-d6): δ 1.60 (s, 9H,
C(CH3)3), 6.78 (dd, 1H, ArH), 6.88 (dd, 1H, ArH), 7.75 (d, 1H,
ArH), 10.14 (s, 1H, ArOH). Combination with trimesoyl
trichloride and cleavage with trifluoroacetic acid gave the
9.80 (s, 3H, ArH). IR (KBr): 3088, 1718, 1229 cm-1 13C NMR
.
(DMSO-d6): δ 127.07, 127.80, 130.70, 132.98, 135.59, 150.67,
163.06, 165.88.
1
product as white crystals. Yield: 17%, mp >240 °C. H NMR
(DMSO-d6): δ 7.51 (dd, 3H, ArH), 7.75 (d, 3H, ArH), 7.95 (dd,
Tris(phenyl-2-acetic acid) trimesate (12) was prepared from
tert-butyl-2-hydroxyphenyl acetate. The tert-butyl ester was
prepared as follows. A mixture of 2-hydroxyphenylacetic acid
(5.0 g), liquid isobutylene (100 mL), and concentrated sulfuric
acid (2 mL) was shaken in a pressure vessel (a European beer
bottle with a glass-rubber seal and wire bail) for 7 h.
(Isobutylene was condensed at -78 °C into a graduated
cylinder under nitrogen. The condensed liquid was transferred
into the reaction bottle, and the stopper was inserted. After 7
h, the mixture was removed, dissolved in ethyl acetate, and
washed with saturated sodium bicarbonate. The organic layer
was concentrated, and the resulting yellow oil was purified
using flash chromatography (neutral alumina, eluted with
chloroform): 1.2 g (23%). 1H NMR (CDCl3): δ 1.45 (s, 9H,
C(CH3)3), 3.50 (s, 2H, CH2), 6.88 (m, 2H, ArH), 7.12 (m, 2H,
ArH), 8.02 (s, H, ArOH).
3H, ArH), 9.02 (s, 3H, ArH). IR (KBr, cm-1): 1750 (s), 1708
(s), 1595 (m), 1195 (s). MS (negative FAB): 671 (M - 1).13
C
NMR (acetone-d6): δ 165.83 (s), 162.24 (s), 152.27 (s), 135.32
(s), 132.62 (s), 132.11 (s), 130.37 (s), 129.19 (s), 124.29 (s),
120.95 (s).
Tris(3-carboxylphenyl) trimesate (8) was prepared from tert-
butyl-3-hydroxybenzoate. 1H NMR (CDCl3): δ 1.60 (s, 9H,
C(CH3)3), 7.09 (m, 1H, ArH), 7.27 (t, 1H, ArH), 7.55 (m, 1H,
ArH), 7.64 (m, 1H, ArH), 7.74 (s, 1H, ArOH)) and trimesoyl
trichloride followed by reaction with trifluoroacetic acid. The
product is a white powder, mp 188-191 °C. Yield: 43%. 1H
NMR (DMSO-d6): δ 7.66 (m, 6H, ArH), 7.94 (m, 6H, ArH),
9.06 (s, 3H, ArH). HRMS (FAB+): 571.0851 (MH)+ (theoretical
571.0876). IR (KBr): 3083, 1699, 1222 cm-1 13C NMR (DMSO-
.
d6): δ 122.88, 126.57, 127.36, 130.17, 130.89, 132.60, 135.39,
150.52, 163.11, 166.56.
Reaction of the ester with trimesoyl trichloride and removal
of the tert-butyl groups gave 12 as a yellow powder, mp 167
°C (dec). Yield: 22%. 1H NMR (DMSO-d6): δ 3.61 (s, 1H, CH2),
6.90 (m, 1H, ArH), 7.45 (m, 3H, ArH), 8.94 (s, 3H, ArH). LRMS
(ESMS): 611 (M - H)- (theoretical 611). IR (KBr, cm-1): 3446,
1744, 1218. 13C NMR (DMSO-d6): δ 27.83, 122.28, 128.94,
129.58, 130.88, 131.32, 131.52, 153.72, 162.01, 164.29, 170.52.
Cr oss-Lin k in g of Deoxyh em oglobin . The method de-
scribed elsewhere for related reagents was used to prepare
hemoglobin in the deoxy form.8 The cross-linking reagent (2
mol/mol of hemoglobin) was dissolved in 1 mL of dioxane and
4 mL of 0.1 M sodium borate (pH 9.0). The flask was flushed
with nitrogen three times. The reagent solution was added to
the hemoglobin solution and the mixture was left for 24 h.
Samples were analyzed after 2 and 12 h. They were saturated
with carbon monoxide, passed through a Sephadex G-25
column (0.1 M MOPS, pH 7.2), collected, and stored in a vial
at 0 °C. Hemes and the globin (R and â) chains were separated
by reversed-phase HPLC using a 330 Å pore C-4 Vydac column
Tris(2,4-dicarboxylphenyl) trimesate (9) was prepared as
follows. Di-tert-butyl-4-hydroxyisophthalate was prepared from
4-hydroxyisophthalic acid (1.0 g) dissolved in a solution of
thionyl chloride (30 mL) and oxalyl chloride (1 mL). The
solution was refluxed overnight. Solvents were evaporated
using a rotary evaporator and the dry product was then
dissolved in anhydrous THF (20 mL). The solution was slowly
transferred to a stirred suspension of potassium tert-butoxide
(1.8) in anhydrous THF (20 mL) and stirred overnight. The
organic solution was concentrated to dryness. The resulting
product was dissolved in ether (30 mL) and washed twice with
saturated sodium bicarbonate. The organic solution was
concentrated to dryness to give a white powder. Yield: 0.7 g
1
(56%). H NMR (CDCl3): δ 1.58 (s, 9H, C(CH3)3), 1.62 (s, 9H,
C(CH3)3), 6.90 (d, 1H, ArH), 8.02 (dd, 1H, ArH), 8.45 (d, 1H,
ArH), 11.50 (s, 1H, ArOH).
Potassium tert-butoxide (0.3 g) was added to a solution of
di-tert-butyl-4-hydroxyisophthalate (0.7 g) in anhydrous THF
(25 mL). The mixture was stirred for 15 min. A solution of
trimesoyl trichloride (0.3 g) in anhydrous THF (25 mL) was
added dropwise, and the solution was stirred overnight. The
mixture was concentrated to dryness, dissolved in ether, and
washed with saturated sodium carbonate (2 × 25 mL). The
organic phase was dried and concentrated to give tris(2,4-di-
tert-butoxycarbonylphenyl) trimesate as a solid. The product
was dissolved in anhydrous trifluoroacetic acid (30 mL) and
stirred for 1 h. Ether was added to induce crystallization and
the solution was kept at 4 °C overnight. The product (9) was
collected by filtration to give yellow crystals, mp 245 °C (dec).
and developers containing 0.1% trifluoroacetic acid in
a
gradient of acetonitrile-water (20-60%). The effluent was
monitored at 220 nm. The procedure was repeated three times
for each reagent.
P ep tid e P a tter n An a lysis. Dilute hemoglobin solutions
were concentrated by ultrafiltration. Hemoglobin was sepa-
rated into its constituent hemes and globin chains (including
cross-linked chains) by reversed-phase HPLC (330 Å C-4 Vydac
(6) Sprengling, G. R.; Freeman, J . H. J . Am. Chem. Soc. 1950, 72,
1982-1985.
(7) Gringauz, J . J . Med. Chem. 1968, 11, 611-612.
(8) Paal, K.; J ones, R. T.; Kluger, R. J . Am. Chem. Soc. 1996, 118,
10380-10383.
1
Yield: 0.5 g (24%). H NMR (DMSO-d6): δ 7.64 (d, 3H, ArH),
8.26 (d, 3H, ArH), 8.56 (s, 3H, ArH), 9.05 (s, 3H, ArH). HRMS
(FAB+): 703.0533 (MH)+ (theoretical 703.0571). IR (KBr):

Hemoglobin Reactions with Aryl Trimesates
J . Org. Chem., Vol. 65, No. 1, 2000 217
Ta ble 1. Mod ifica tion of Deoxyh em oglobin Aa
column) using developers containing 0.1% trifluoroacetic acid
and a gradient of acetonitrile-water (20%-60%). The globin
chains were then dissolved in 8 M urea (to increase suscep-
tibility to hydrolysis) and kept at room temperature for 4 h.
The solution was diluted to 2 M urea with 0.08 M pH 8.5
ammonium bicarbonate. Trypsin (2% solution) was added, and
the mixture was left for 24 h (room temperature). The tryptic
hydrolysate was heated in boiling water for 2 min and diluted
to 1 M urea. Staphylococcus aureus V8 endoproteinase Glu-C
(1% solution) was added to this mixture, and the resulting
mixture was left for an additional 72 h at room temperature.
The hydrolysates were filtered through a 0.45 fm filter prior
to injection onto the reversed-phase HPLC using a C-18 Vydac
column.9
The HPLC cross-linked peaks were identified by peptide
pattern analysis.2 Peptide fragments were separated using
HPLC procedures.10 Separations were made using developers
of 0.1% trifluoroacetic acid and a gradient of acetonitrile-
water (0%-100% acetonitrile generated over 1 h). In all cases,
products were identified by comparison with those obtained
from the reaction of hemoglobin with trimesoyl tris(methyl
phosphate).9 The resulting peptide fragments were also ana-
lyzed by comparison with peptide patterns of unmodified
hemoglobin.10
resulting modified protein
relative amount of each class
of cross-linked protein (%)
fraction â cross-linked
cross-linked
with ester
hydrolyzed
subunits
altered
with ester
intact
three-way
cross-linked
reagent
1
2
1.00
0.55
0.83
0.89
0.00
0.00
0.00
0.00
0.10
0.00
0.00
0.00
77
0
5
62
27
21
0
18
38
34
13
0
3
39
66
0
4
5
6
0
0
0
7
0
0
0
57
0
0
0
0
8
0
0
9
0
43
0
10
11
12
0
0
0
0
0
a
Yields are based on quantitative HPLC analysis of globin
chains. Deoxyhemoglobin A was reacted with the reagents in 0.1
M sodium borate (pH 9.0), 37 °C.
were also used to obtain the first-order rate constant for
hydrolysis (khyd) by extrapolation to the rate in the absence of
amine.
Ion Sp r a y Ma ss Sp ectr oscop y. The exact masses of cross-
linked hemoglobin were determined by electrospray mass
spectroscopy (ESMS).11 ESMS on the cross-linked hemoglobin
products were performed at the Medical Sciences Facility,
University of Toronto.
Since each reagent contains three esters, the overall ami-
nolysis reaction is kinetically complex. Preliminary measure-
ments over the course of the whole reaction (formation of the
tris amide) gave data that do not fit standard rate laws. It is
likely that aminolysis of the first ester group gives an amide-
diester product whose reactivity is different from that of the
triester. Therefore we used initial rates to obtain the rate of
the first reaction separately as the basis for analyzing reactiv-
ity.12 Rates were measured for about 2% of the total reaction.
The observed rate was statistically corrected to take into
account that these reagents have three ester functions.
Ra te of Cr oss-Lin k in g Rea ction s. A series of reactions
in which aliquots of cross-linked hemoglobins were removed
from the reaction vessel at 20 min intervals were performed
to track the progress of the reaction. Six samples were taken
over 2 h. Each aliquot was quenched with n-propylamine,
exposed to carbon monoxide for 10 min and stored at pH 7.2.
The cross-linking reaction with 1 was complete after 40 min.
In h ibition of Cr oss-Lin k in g. Do the reagents that do not
modify hemoglobin fail to do so because they do not associate
with the protein? Reagent 5 is a triester with para carboxyl
substituents and no ortho carboxyl. It does not alter hemo-
globin. This reagent was combined with hemoglobin in the
presence of a reagent that is an effective cross-linker (1, in
pH 9 borate buffer). 5 (1, 2, and 10 equiv) was added to reaction
solutions containing deoxyhemoglobin along with 1 equiv of
1, and the products were measured after 2 h.
Am in olysis of Tr iester s. Kinetic data for the reaction of
1-12 with n-propylamine were acquired by recording the UV/
vis absorbance at the wavelength of greatest change for the
leaving group (267-328 nm). Preparative reactions were run
with the same materials and the products identified spectro-
scopically. Each reagent (1-12) was combined with an excess
of n-propylamine (as pH 10 buffer) in solution. The product
that is in common from all reagents is benzene-1,3,5-tricar-
boxylic acid tris-n-propylamide 1H NMR: δ 0.09 (t, 9H,
C-CH3), 2.6 (q, 6H, C-CH2-C), 3.2 (q, 6H, C-CH2-N), 8.6
(s, 3H, ArH), 9.0 (s, 3H, C-NH-C). In addition, each reagent
releases the corresponding phenol derivative. The second-order
rate constant for reaction of each reagent with n-propylamine
was determined from the dependence of the observed first-
order rate coefficient (average of three measurements of k0)
upon the concentration of free n-propylamine (total n-propy-
lamine concentrations: 0.2, 0.4, 0.6, 0.8, and 1.0 M). Solutions
for the reaction were kept at ionic strength 1.0 (with added
sodium chloride) and maintained at 25.0 °C. The reagents
(∼0.001 g) were dissolved in 2 mL of DMSO, and 0.020 mL of
the solution was then transferred to a cell containing 3 mL of
the n-propylamine buffer. The data for the aminolysis reactions
Resu lts
We prepared substituted phenolic triesters of trimesic
acid (1-12) and have evaluated their reaction patterns
with human deoxyhemoglobin. Trimesyl tris(3,5-dibro-
mosalicylate) (1) had previously been prepared.2 It reacts
very efficiently and selectively with amino groups in
hemoglobin to produce a principal cross-linked product
(linked between ꢀ-amino groups of lysine-82 side chains
in the â-subunits) with the third ester site either intact
or hydrolyzed.2,4 The general synthetic route for the
triester analogues of 1 combines trimesoyl trichloride
with an excess of the tert-butyl ester of a substituted
phenol. The tert-butyl groups of the esters are then
removed with trifluoroacetic acid. (The tert-butyl protect-
ing group is necessary for the synthesis of 1, and we
extended the general route2,13 in all cases.) The tert-butyl
esters of the phenols are prepared by DCC coupling of
the carboxylic acid and tert-butyl alcohol.13 In one case
(8), the synthesis of the tert-butyl ester precursor did not
succeed by this procedure and a number of other stan-
dard methods also gave no product. In that case we were
finally successful with a route in which the carboxylic
acid is converted to the acid chloride, followed by reaction
with potassium tert-butoxide.
The esters in 1 contain a carboxyl group ortho to a
phenolic ester oxygen as well as two bromine atoms as
aromatic substituents. It is an effective cross-linking
reagent. It gives the highest yield of cross-linked hemo-
(9) Kluger, R.; Wodzinska, J .; J ones, R. T.; Head, C.; Fujita, T. S.;
Shih, D. T. Biochemistry 1992, 31, 7551-9.
(10) Shelton, J . B.; Shelton, J . R.; Schroeder, W. A. J . Liq. Chro-
matogr. 1984, 7, 1969-1977.
(11) Adamczyk, M.; Gebler, J . C. Bioconjugate Chem. 1997, 8, 400-
6.
(12) Fleck, G. M. Chemical Reaction Mechanisms; Holt, Rinehart,
and Winston: New York, 1971; pp 47-48.
(13) Delaney, E. J .; Massil, S. E.; Shi, G. Y.; Klotz, I. M. Arch.
Biochem. Biophys. 1984, 228, 627-38.

218 J . Org. Chem., Vol. 65, No. 1, 2000
Kluger and Stefano
F igu r e 1. Chromatogram of the product mixture from the
reaction between 1 and hemoglobin after 2 h in 0.1 M sodium
borate, pH 9.0, 37 °C. Excess n-propylamine was added to the
solution after 2 h to convert residual esters to amides.
F igu r e 2. Chromatogram of the product mixture showing the
reaction between the mixture of 1 and 5 (1:10 molar ratio,
respectively) and hemoglobin after 2 h, in 0.1 M sodium borate,
pH 9.0, 37 °C.
Ta ble 2. Ra te Con sta n ts for th e Am in olysis a n d
globin among the various trimesyl triesters we prepared.
Table 1 summarizes the products obtained from reactions
of the set of reagents with hemoglobin. Only the reagents
with a leaving group containing a carboxyl ortho to the
phenolic oxygen (1-4, 9) produce a cross-link in hemo-
globin. In the one case where there are carboxyls both
ortho and para to the ester, cross-links are produced in
very low yield (10%). The yield of cross-linked hemoglobin
from the ortho-substituted carboxyl reagents is greatest
for the reagent having two bromine substituents (1). The
yield of cross-linked protein is lower with chlorine in
place of bromine and lowest in compounds with no
halogen present.
The reaction with 1 also gives a small amount of the
product in which all three acyl groups react with the
protein.9 The other esters with ortho carboxyl groups give
considerably higher proportions of this product. Reagents
with carboxyl groups in meta and para positions give no
cross-linked hemoglobin products. For these compounds
we detect some acylation of amino groups in the â
subunits only in cases where the leaving group contains
halogens. The product in this instance has the two other
esters hydrolyzed.
Figure 1 shows an HPLC analysis of the globin chains
resulting from the reaction of hemoglobin with the
trifunctional reagent (1). The R-globin chains are not
modified, while all â chains are altered. The altered
chains are identified as peaks: “1” is a combination of
cross-linked â chains with the residual ester hydrolyzed.
Peak “2” is a combination of cross-linked â chains with
the residual ester converted to the n-propylamide. Peak
“3” is the triply linked material with the cross-linked â
chains connected at positions 82, 82, and 1. The nature
of the modification was determined as described previ-
ously.9
Why do some of the reagents not produce a cross-link?
Do they enter the reaction site without producing a cross-
link, or do they fail to associate with the protein? Since
we expect that the positive charge in the effector site of
the protein will attract anions,14 it is reasonable to expect
that these anionic materials should be attracted. We
tested this by adding 1, which is an efficient cross-linking
reagent, to a solution that contained an excess of the para
carboxyl substituted material (5), which does not react
with the protein. In the presence of excess 5 about 80%
of the â subunits react with 1 (Figure 2). In the absence
Hyd r olysis of Tr iester s 1-12a
aminolysis,
hydrolysis,
khyd (s-1
reagents
kN (M-1s-1
)
)
λmax (nm)
1
2
3.0 ( 0.2
0.026 ( 0.005
0.016 ( 0.002
0.038 ( 0.003
0.013 ( 0.004
0.054 ( 0.004
0.008 ( 0.004
0.039 ( 0.003
0.014 ( 0.002
0.038 ( 0.002
not detectedb
0.033 ( 0.004
0.0150 ( 0.0007
315
300
300
300
300
275
267
312
298
307
328
295
0.054 ( 0.006
0.12 ( 0.01
1.35 ( 0.08
0.13 ( 0.01
0.07 ( 0.01
0.09 ( 0.01
0.17 ( 0.01
0.16 ( 0.01
not detectedb
0.16 ( 0.01
0.099 ( 0.002
3
4
5
6
7
8
9
10
11
12
a
b
Errors are two standard deviations. No hydrolysis or ami-
nolysis was observed after 20 h.
of 5 the â subunits would be completely modified. This
suggests that 5 binds but does not react.
To what extent are inherent differences in reactivity
of 1-12 with amines manifested in their reactions with
hemoglobin? As a basis for comparison we measured the
rate of reaction of each with the small, basic amine,
n-propylamine. Due to the kinetic complexity of reactions
of trifunctional reagents, we measured initial rates of
aminolysis so we could obtain the rate constant for amide
formation from one ester group in each compound. The
rate constants for hydrolysis and aminolysis of the
trifunctional reagents, 1-12, are summarized in Table
2. The results were obtained using initial rates to obtain
observed first-order rate coefficients in the range 0.2-
1.0 M n-propylamine (as buffer), pH 10.5, ionic strength
1.0 (with added sodium chloride), 25 °C. (Additional
details are in the Experimental Section.)
The rate constants for aminolysis of reagents that have
one carboxyl per leaving group are largest for those in
which the carboxyl is para to the ester (5-7), followed
by the ortho (1-4) and meta (8) derivatives. This order
is consistent with the order of reactivity for aminolysis
of substituted phenyl acetates15 (relative rate constants
at 25 °C: p-COOH, 1.8; o-COOH, 1.1; m-COOH, 0.6).
The reagents with an ortho carboxyl and two halogens
on the aromatic leaving groups (1 and 4) are the most
reactive toward n-propylamine, and these also react with
hemoglobin. Two of the reagents that have two carboxyl
(15) Arcell, A.; Concillo, C. J . Chem. Res. Miniprint 1992, 1, 201-
226.
(14) Ueno, H.; Manning, J . M. J . Protein Chem. 1992, 11, 177-85.

Hemoglobin Reactions with Aryl Trimesates
J . Org. Chem., Vol. 65, No. 1, 2000 219
Sch em e 1
F igu r e 3. Rate constants for aminolysis vs the pKa of the
acids of the conjugate bases of the leaving groups of the
trifunctional reagents, 1-10, where b are data for the tri-
functional reagents with carboxylates meta and para to the
ester and 1 represents the trifunctional reagents with car-
boxylates ortho to the ester. Reagent 12 is not included in this
plot because the necessary pKa is not available and could not
be determined. The acidity constants are from published tables
or were estimated from Hammett plots.
catalyst in aminolysis. The small value of â for the
correlated compounds is reasonable since the carbonyl
group is stabilized by its own π system rather than by
the phenyl substituents. We note that the rates for the
compounds with two halogen substituents are above the
correlation, suggesting an alternative mechanism may
be involved.
substituents (9, 11) are more reactive toward n-propyl-
amine than the reagents that have one carboxyl, but
these are unreactive with hemoglobin. The derivative
with two carboxyls ortho to the ester (10) is unreactive
with n-propylamine (and does not react with hemoglobin).
The low reactivity is consistent with steric effects in ester
aminolysis.16
Our results show that the reactions of the triesters
with n-propylamine are not sensitive to positions of the
substituents on the ester, paralleling the reaction pat-
terns of monosalicylates with amines.3 This contrasts
dramatically with the reactions of the triesters with
amino groups in hemoglobin where there is a strong
pattern of positional selectivity, with the ortho carboxyl
being essential. Therefore, the control of reactions with
hemoglobin is not based on the electrophilic reactivity of
the reagent. Rather, it must be a consequence of the
environment in the protein in relation to the geometry
of the reagent. The amino groups of hemoglobin that are
modified are in the site that binds the polyanionic
effector, 2,3-diphosphoglycerate.20 The site, between the
â subunits, contains a number of basic amino groups,
including those that react with the reagents.2,21 The
amino groups will be protonated to a significant extent.
However, the amine is not permanently in the form of
the ammonium ion, since the proton can be readily
transferred to an acceptor.
Discu ssion
The yields of cross-linked proteins from the reactions
of amino groups in hemoglobin with 1-12 indicate the
importance of the carboxyl group adjacent to the reacting
ester in the reagent (present in 1-4, 9, and 10). Acyl
phosphate monoesters, which are anionic, acylate hemo-
globin at the same sites as do the salicylate deriva-
tives.9,17 Since the position of the charged group relative
to the reaction site is critical, it is possible that the
carboxylate participates as a catalyst in the reaction with
the protein. However, the reaction pattern is not consis-
tent with what is known about reactions of small amines
with salicylates. St. Pierre and J encks3 found that the
reactions of salicyl esters with basic amines are not
facilitated by the adjacent carboxyl. (The carboxyl func-
tions as an acid to accelerate aminolysis by weak bases,
such as hydroxylamine.) The conjugate base also serves
as an intramolecular catalyst for the addition of water
to the ester.18
We expect the reactions of n-propylamine with 1-12
to follow the normal patterns of ester aminolysis, subject
to steric and electronic effects.19 Figure 3 is a plot of the
rate constant for aminolysis against the pKa of the
conjugate acid of the leaving group for 10 of the com-
pounds. There is a linear correlation for five compounds
with meta and para carboxyl substituents, giving â )
0.14. Compounds with an ortho carboxyl are less reactive
than expected from the pKa of the phenol, a typical steric
effect.19 It is clear that the ortho carboxyl group is not a
A mechanistic diagram for the modification reaction
is presented in Scheme 1:
A similar mechanism would also apply for modification
reactions that use acyl phosphate monoesters, since in
those compounds the anionic site and reaction site are
adjacent. Since the reaction patterns of these compounds
reflect the structure of the protein, it will be interesting
to examine results of reactions of this set of materials
with other proteins.
Ack n ow led gm en t. We thank NSERC Canada for
support through an operating grant and the Protein
Engineering Network of Centres of Excellence. We
thank Omar Green for his help in conducting the
competition experiments.
(16) Suh, J .; Chun, K. H. J . Am. Chem. Soc. 1986, 108, 3057-3063.
(17) Kluger, R.; Grant, A. S.; Bearne, S. L.; Trachsel, M. R. J . Org.
J O991514N
Chem. 1990, 55, 2864-2868.
(18) Fersht, A. R.; Kirby, A. J . J . Am. Chem. Soc. 1967, 89, 4853-
7.
(20) Benesch, R.; Benesch, R. E. Biochem. Biophys. Res. Commun.
1967, 26, 162-167.
(21) Fermi, G.; Perutz, M. F. J . Mol. Biol. 1984, 175, 159-174.
(19) Cline, G. W.; Hanna, S. B. J . Am. Chem. Soc. 1987, 109, 3087-
91.