Regioselective covalent modification of hemoglobin in search of antisickling agents

936

J . Med. Chem. 2003, 46, 936-953

Regioselective Cova len t Mod ifica tion of Hem oglobin in Sea r ch of An tisick lin g

Agen ts

Soobong Park,^†,§Brittany L. Hayes,^†,Fatima Marankan,^†,#Debbie C. Mulhearn,^‡Linda Wanna,^‡

Andrew D. Mesecar,^‡Bernard D. Santarsiero,^‡Michael E. J ohnson,*^,‡and Duane L. Venton*^,†

Department of Medicinal Chemistry and Pharmacognosy and Center for Pharmaceutical Biotechnology, College of Pharmacy,

University of Illinois at Chicago, 833 S. Wood Street, Chicago, Illinois 60612

Received August 20, 2002

Although the molecular defect in sickle hemoglobin that produces sickle cell disease has been

known for decades, there is still no effective drug treatment that acts on hemoglobin itself. In

this work, a series of diversely substituted isothiocyanates (R-NCS) were examined for their

regioselective reaction with hemoglobin in an attempt to alter the solubility properties of sickle

hemoglobin. Electrospray mass spectrometry, molecular modeling, X-ray crystallography, and

conventional protein chemistry were used to study this regioselectivity and the resulting

increase in solubility of the modified hemoglobin. Depending on the attached R-group, the

isothiocyanates were found to react either with the Cysâ93 or the N-terminal amine of the

R-chain. One of the most effective compounds in the series, 2-(N,N-dimethylamino)ethyl

isothiocyanate, selectively reacts with the thiol of Cysâ93 which, in conjunction with the cationic

group, was seen to perturb the local hemoglobin structure. This modified HbS shows an

approximately 30% increase in solubility for the fully deoxygenated state, along with a

significant increase in oxygen affinity. This compound and a related analogue appear to readily

traverse the erythrocyte membrane. A discussion of the relation of these structural changes to

inhibition of gelation is presented. The dual activities of increasing HbS oxygen affinity and

directly inhibiting deoxy HbS polymerization, in conjunction with facile membrane traversal,

suggest that these cationic isothiocyanates show substantial promise as lead compounds for

development of therapeutic agents for sickle cell disease.

In tr od u ction

acceptable approach in their search for antisickling

agents.¹This rationale centers on the very large amount

of HbS that must be targeted. For noncovalent drugs

with dissociation constants lower than 10^-6molar,

nearly all of the drug would be bound to HbS because

of the unusually high concentration of the target Hb

protein,⁶i.e., approximately 2 × 10^-3molar Hb concen-

tration in the blood at normal hematocrit. As it is

generally accepted that at least 20% of the total

hemoglobin must be altered to produce a clinically

significant outcome,^4,7this would require a minimum

drug concentration of approximately 0.4 mM for the

noncovalent casesa blood level that would be considered

unacceptable. Furthermore, for the noncovalent case,

the putative drug would be continuously excreted or

metabolized, requiring chronic administration at high

doses.

On the other hand, a covalent drug administered

repeatedly and in low doses, could progressively modify

the target macromolecules due to the typically longer

half-life of the covalent complex. In the case of Hb, this

half-life is on the order of 120 days,⁸and could translate

into a blood-level advantage of greater than 2 orders of

magnitude for the covalent approach. Although there

can be toxicity problems with this approach to drug

design, it should be noted that important drugs, such

as aspirin and phenoxybenzamine, which act through

covalent modification of the cyclooxygenase,⁹and the

â-adrenergic receptor,¹⁰respectively, clearly demon-

strate that such an approach can exhibit an acceptable

therapeutic window.

Sickle cell anemia results from the substitution of Val

for Glu at position 6 in the hemoglobin â-chain. This

substitution results in abnormal rigidity of erythrocytes

at low oxygen tension, due to intracellular polymeriza-

tion of the sickle hemoglobin (HbS). In turn, this leads

to occlusions in the microcirculation.¹The polymeriza-

tion of HbS involves both axial and lateral contacts,²

with one mutant Valâ6 residue participating in a lateral

intermolecular hydrophobic contact through the interac-

tion between the donor Valâ6 and the acceptor Pheâ85

and Leuâ88.^2-4

Hydroxyurea is currently the most useful therapy for

reducing the symptoms of sickle cell disease, but

patients exhibit variable responses, and it does not

eliminate symptoms.⁵Thus, there is still interest in

finding therapeutic agents that directly block HbS

polymerization. Unlike the development of most drugs,

researchers have adopted covalent modification as an

* Corresponding authors. D.L.V.: Tel (312) 996-5233. FAX (312)

996-7107. e-mail venton@uic.edu. M.E.J .: Tel (312) 996-9114. FAX

(312) 413-9303. e-mail mjohnson@uic.edu.

^†Department of Medicinal Chemistry and Pharmacognosy.

^‡Center for Pharmaceutical Biotechnology.

^§Current address: Soobong Park, University of Minnesota Cancer

Center, 790E CCRB, 425 E. River Road, Minneapolis, MN 55455. Tel.

(612) 624-0638. e-mail parkx108@umn.edu.

Current address: Brittany Hayes, CEM Corporation, Life Sciences

Division, 3100 Smith Farm Road, Matthews, NC 28106. e-mail

brittany.hayes@cem.com.

#

Current address: Fatima Marankan, Institute of Molecular Biol-

ogy and Biochemistry, Simon Fraser University, Burnaby, British

Columbia V5A 1S6, Canada.

Published on Web 02/07/2003

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 937

One approach to reduce the HbS polymerization has

been to disrupt or to destabilize the intermolecular

interactions of the deoxy (T) state conformation by direct

covalent modification of the HbS protein. In essence, the

effect of the pathologic point mutation is reversed by

modification of an amino acid at another site in the

protein. Several structurally unrelated compounds have

been partially successful in this regard.¹¹Any specificity

of covalent reaction of agents on the Hb surface must

ultimately result from differences in the reactant’s

structure interacting with the diverse topological fea-

tures on the protein’s surface. On the reactant side,

these structural features can be arbitrarily divided into

two categories: (1) the nature of the reactive functional-

ity, and (2) the structure of the group attached to the

reactive function. In the latter case, regioselectivity for

a particular site on the molecule is expected to result

from favorable or unfavorable interactions with the

protein’s surface, e.g., binding complementarity of the

R group with the protein prior to reaction, or perhaps

steric hindrance leading to exclusion from reaction at a

particular site.

In the present work, we examine a series of isothio-

cyanates, in which only the R-group is changed, for their

regioslective reactions on the Hb surface. The initial

measure of selectivity involved determining whether the

agents reacted covalently with the R- or the â-chain of

the Hb tetramer. Selected isothiocyanates were tested

for antisickling activity, and for those showing the

greatest antisickling activity, the amino acid side chain

responsible for the covalent modification was also

determined. The ability to inhibit HbS polymerization

was shown to correlate with both the site of modification

and the structure of the R-group of the reacting isothio-

cyanate. In addition, the biological activity was cor-

related with the extent of HbS modification for one of

the more active agents, a dimethylaminoethyl isothio-

cyanate. Molecular modeling also suggests that the

R-group may be directing the isothiocyanate moiety in

its reaction with the Hb surface. Finally, molecular

modeling also suggests that the activity of the dimethyl-

aminoethyl isothiocyanate in inhibiting deoxy HbS

polymerization may result from perturbation of inter-

molecular electrostatic interactions within the HbS

polymer fiber.

F igu r e 1. Isothiocyanates prepared for reaction with HbA.

Sch em e 1

adapted from Seligman utilizing thiophosgene.¹²The

syntheses of isothiocyanates 1-4, 6, 8-18 was ac-

complished starting from their respective primary

amines, following a previously reported procedure¹³

using di-2-pyridyl thionocarbonate. This procedure gave

isothiocyanates 1-4, 6, 8-18 in fairly high yields, after

either column chromatography or vacuum distillation.

Dialkylaminoalkyl isothiocyanates 7 and 19-21 were

prepared by a previously reported procedure from

McElhinney,¹⁴using carbon disulfide and HgCl₂in

acetone. Removal of the mercury salts by filtration,

followed by vacuum distillation, provided the isothio-

cyanates (7 and 19-21).

The synthesis of hydrochloride salts 22-25 is outlined

in Scheme 1. Following a modified preparation,^15,16the

dialkylamine (26 or 27), chloronitrile (28 or 29), and

sodium iodide, in water, were stirred for 5 days. Vacuum

distillation yielded dialkylaminonitriles 30-33. Cata-

lytic hydrogenation over a rhodium-alumina catalyst

afforded mixtures of primary (34-37) and the corre-

sponding secondary amines. Unable to easily separate

these amines, the mixture was used to prepare isothio-

cyanates 38-41, as only the primary amine will form

the isothiocyanate, which were then isolated by vacuum

distillation. The white hydrochloride salts (22-25) were

Ch em istr y

Isoth iocya n a te P r ep a r a tion s. Structurally diverse

isothiocyanates (Figure 1) were required to evaluate

their chemical reaction with the Hb surface. The start-

ing amines for preparation of the isothiocyanates were

chosen from a set of chemically diverse amines that were

commercially available. The derivatives included both

aryl and alkyl derivatives containing neutral, positive,

and negatively charged functionality, as well as different

structural variations such as cyclic and acyclic struc-

tures. For example, the aryl isothiocyanates 1-6 are

composed of two acids (5, 6), two bases (3, 4), and two

neutrals (1, 2). Similarly, the alkyl isothiocyanates,

7-21, are composed of an acid (9), seven neutrals (8,

10-12, 15-17), and seven bases (7, 13, 14, 18-21).

The isothiocyanates shown in Figure 1 were synthe-

sized by one of three different methods. Isothiocyanate

5 was formed from p-aminosalicylic acid via a procedure

938 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

Ta ble 1. Competitive Reaction of Isothiocyanates for the R-

and â-Subunits of Normal Hemoglobin^a

% of subunit modification (100 µM HbA) at 3 h

compound

(200 µM)

R-subunit

â-subunit

1

72

66

72

51

40

36

42

48

35

42

0

3

2

0

7

6

11

10

7

0

8

9

2

3

8

17

16

15

10

3

4

5

6

7

13

34

40

41

31

36

35

25

28

30

18

24

58

33

21

13

38

37

8

9

10

11

12

13

16

17

18

20

21

F igu r e 2. Mass shift analysis of covalently modified HbA

processed by maximum entropy deconvolution. (a) HbA modi-

fied with isothiocyanate 1 (MW 179), (b) HbA modified with

isothiocyanate 7 (MW 130).

formed by bubbling HCl gas through an ether solution

of the crude isothiocyanate.

a

Products were identified by the shift in mass of the corre-

sponding covalently modified Hb subunits. Multiply charged ion

spectra were processed using a maximum entropy deconvolution,

and the yield was obtained from the equation: yield ) height of

modified/(height of unmodified + height of modified).

Selective Hb r- a n d â-Ch a in Mod ifica tion by

Isot h iocya n a t es. Reactions of the various isothio-

cyanates with the R- and â-chains of HbA were analyzed

by electrospray mass spectrometry (ESMS) as previ-

ously described.¹⁷Isothiocyanates 1, 4, and 7-9 were

tested in triplicate. The remaining analyses were sin-

gular. The variability in triplicate experiments was

about 10%, suggesting the general uncertainty expected

for this type of analysis. As can be seen, there are clear

differences between the reactivity of these derivatives

and the R- and â-subunits of the HbA molecule, sug-

gesting that the R-group of the isothiocyanate does, in

fact, influence the course of the reaction.

For example, the ESMS of HbA modified with aryl

isothiocyanate 1 and alkyl isothiocyanate 7 are shown

in Figure 2. In electrophilic addition of the isothio-

cyanate with a nucleophile, there is no other side

product, and the mass of the product is equal to the sum

of the mass of the isothiocyanate and the nucleophile,

here the Hb subunit. As may be seen in Figure 2a, there

is a 15306 m/z ion corresponding to the modified

R-subunit, and one at 16046 m/z for the â-subunit,

indicating that both subunits react, at least to some

extent, with the aryl isothiocyanate 1 (MW 179). There

was no indication that either subunit was modified by

more than one molecule of isothiocyanate. The latter

would, of course, be indicated by a family of peaks

separated by 179 amu. All of the reactions studied

between the isothiocyanate and the two subunits showed

only a one-to-one reaction under the conditions used.

The selectivity of isothiocyanates 1-13, 16-18, 20, 21

for the R- and â-subunit are listed in Table 1.

Thus, ESMS analysis of the reaction of alkyl isothio-

cyanate 7 (MW 130) with HbA indicated no detectable

reaction with the R-subunit. Instead, the ions repre-

sented by the peak at 15997 m/z in Figure 2b indicate

almost exclusive reaction with the â-subunit. Indeed,

examination of the isothiocyanates tested revealed a

general trend (Table 1). Isothiocyanates 1-6, where the

isothiocyanate is directly attached to an aromatic ring,

showed selectivity for the R-subunit. On the other hand,

alkyl isothiocyanates 7-13, 16-18, 20, 21, where the

reactive moiety is attached to an sp³hybridized carbon,

were found to preferentially modify the â-subunit.

However, this selectivity is not absolute. For example,

aryl isothiocyanates 3 and 4 showed a 2-fold selectivity

for the R-subunit relative to the â-subunit. Furthermore,

we could not achieve complete, selective modification

of the R-subunit over the â-subunit, using various

conditions of reaction time and concentration, or reac-

tant ratios. Alkyl isothiocyanates also showed varying

degrees of selectivity depending on the structure of the

substituent. For example, amine isothiocyanate 7 and

acid isothiocyanate 9 showed a clear preference for the

â-subunit. But heterocyclic isothiocyanates 12 and 13

had about a 10-20% selectivity for the R-subunit

relative to the â-subunit. Taken together, these results

strongly suggest that different isothiocyanates are

reacting at different sites on the Hb molecule and with

different degrees of regioselectivity as a function of the

attached R-group. To better understand this selectivity,

we determined the specific amino acid residues being

covalently modified with selected isothiocyanates and

The clear preference for the aromatic alcohol 1 to react

with the R- versus the â-subunit suggests selectivity of

this reagent for the R-subunit (>7×). However, this

property was not shared by all of the isothiocyanates.

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 939

examined the reaction specificity using molecular mod-

eling and X-ray crystallography.

generated from modification by the alkyl isothiocyanates

is unstable to the experimental conditions. Further, it

was shown by ESMS that, although the covalent bond

with the alkyl isothiocyanate was stable to the acidic

acetone used to remove the heme, incubation in the

slightly basic ammonium bicarbonate buffer (pH 8.5)

that was used for the tryptic digestion, resulted in

cleavage of the isothiocyanate bond with the protein for

the adducts of 7-9. Since the anticipated thiourea that

would be formed by addition of the isothicyanate to the

N-terminal amine should be stable under these condi-

tions, this result suggested that modification of some

other residue was occurring.

Isot h iocya n a t e R ea ct ion Sit e on t h e r-Ch a in .

Similar to the methods of Desiderio,¹⁸we used trypsin

digestion with reversed phase HPLC/ESMS analysis of

the resultant peptide fragments for identification of the

isothiocyanate reaction site on the Hb R-chain. We

selected two aryl isothiocyanates, 1 and 3, as represen-

tative of this group for defining the site of reaction on

the R-subunit. These two isothiocyanates provide a high

degree of R-subunit modification with different struc-

tural characteristics (1, alcohol; 3, amine). Samples of

HbA‚CO were first individually modified with aryl

isothiocyanates 1 and 3. These modified HbA molecules

were stirred in acidic acetone to remove heme, and then

the precipitated globins were digested with trypsin. The

resultant peptide mixtures were analyzed by reversed-

phase LC/MS, and the resulting chromatograms were

compared to that of unmodified HbA‚CO (data not

shown).

Chiancone has reported that the Cysâ93 sulfhydryl

group is solvent accessible and available for nucleophilic

reactions.¹⁹To determine if this residue was the site of

reaction, the ionic character of the HbA R-chain modified

with alkyl isothiocyanates 7 (positive), 8 (neutral), and

9 (negative) was tested using electrophoresis. The

sulfhydryl group is neutral under physiological condi-

tions, and the new functional group formed from the

reaction between the sulfhydryl group and the isothio-

cyanate is a dithiocarbamate that is also neutral under

physiological conditions. On the other hand, if the

reaction site is positively charged, for example an

ꢀ-amino group of Lys, this charge would be removed

upon reaction with the isothiocyanate via the formation

of a thiourea. Thus, the charge-state of the modified

protein would be reduced by one. The electrophoretic

pattern of HbA modified with the three different alkyl

isothiocyanates, 7, 8, and 9, matched that expected

when a sulfhydryl group was the site of reaction (data

not shown).

Extracted ion chromatograms of the R-subunit were

used to identify the residue responsible for the modifica-

tion. Trypsin digestion of the R-subunit of hemoglobin

should theoretically generate fourteen peptide frag-

ments, since trypsin is specific for Lys and Arg residues.

Thus, eleven cuts at Lys and three cuts at Arg residues,

where one of the Arg residues is C-terminal, should

occur. If the isothiocyanate modified a Lys, trypsin

hydrolysis would be unlikely at that particular Lys

residue. The molecular weight of such an unhydrolyzed

fragment would be the sum of the molecular weight of

the two theoretical fragments and that of the isothio-

cyanate minus water. The ESMS total ion chromato-

grams were scanned for these possible molecular weights.

However, peaks corresponding to such fragments were

not detected.

Additional evidence that the aliphatic isothiocyanates

react with the â93Cys residue was obtained by titration

of free sulfhydryl groups in the modified Hb using the

p-chloromercuribenzoic acid method of Boyer.²⁰HbA has

6 Cys residues, two from the two R-subunits, and four

from the two â-subunits. Among these, only the two

Cysâ93s are solvent accessible and reactive.¹⁹Given the

above work, we expected two free sulfhydryl groups from

HbA modified with the aryl isothiocyanates that react

with the N-terminus of the R-subunit and no free

sulfhydryl group from HbA modified with the alkyl

derivatives.

If the isothiocyanate modified an amino acid within

a fragment, the molecular weight of the modified

fragment would be the sum of the molecular weight of

the unmodified fragment and the given isothiocyanate.

The ESMS total ion chromatogram scanned for these

masses allowed us to identify one new peak correspond-

ing to the addition of the isothiocyanate to the fragment

from the hydrolysis of the first Lys (R7); i.e., m/z ) 908

from the alcohol-substituted isothiocyanate 1, and m/2z

) 504 and m/z ) 1006 from the amine-substituted

isothiocyanate 3. For the R1-R7 fragment (sequence

VLSPADK) there are two possible nucleophiles for

reaction with an isothiocyanate: the N-terminal amine

and the ꢀ-amino group of LysR7. However as mentioned,

hydrolysis of a modified Lys residue is unlikely. There-

fore, it was concluded that the reaction site on the

R-subunit for the aryl isothiocyanates was the N-

terminal residue.

HbA was modified with aryl isothiocyanate 5 and

alkyl isothiocyanates 7, 8, and 9, and then treated with

2 equiv of p-chloromercuribenzoic acid. The UV absor-

bance for these reaction products was then compared

with a standard curve that was prepared with unmodi-

fied HbA treated with p-chloromercuribenzoic acid. The

results of this titration indicate that HbA modified by

the aryl isothiocyanate 5 has two free sulfhydryl groups,

but HbA modified by alkyl isothiocyanates 7, 8, and 9

did not have any free sulfhydryl groups, as expected

(data not shown). On the basis of this evidence and the

electrophoresis data, it was concluded that Cysâ93 was

the reaction site for the alkyl isothiocyanates. This

conclusion was verified by X-ray crystallographic analy-

sis of deoxy Hb modified with 2-(N,N-dimethylamino)-

ethyl isothiocyanate 7, as discussed below.

Isoth iocya n a te Rea ction Site on th e â-Ch a in . In

an attempt to determine the reaction site on the

â-subunit for the alkyl isothiocyanates, trypsin digestion

was again applied as in the aryl isothiocyanate case,

expecting that the â-chain N-terminal amine would be

similarly modified. The aliphatic isothiocyanates 7-9

were used for these analyses. Unexpectedly, the HPLC

chromatograms of the digested peptide were the same

as that for the unmodified HbA controls. One possible

explanation for this could be that the functional group

E r yt h r ocyt e P er m ea b ilit y t o Isot h iocya n a t es.

For these compounds to be therapeutically useful, they

must be taken up by erythrocytes, and covalently react

940 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

with the intracellular Hb. To evaluate their potential

for uptake, we have incubated 7 and 24 with washed

erythrocytes, washed the cells to remove excess com-

pound, and then measured Hb modification by mass

spectrometry. We find that both compounds produce

significant Hb modification, with 24 exhibiting about

2-fold higher modification than 7, as might be expected

from their relative cLogP values (1.0 and 2.5 for 7 and

24, respectively). Thus, uptake and modification of

intracellular Hb is clearly feasible, with optimization

of cLogP probably the most important parameter for

maximizing uptake.

the attraction of the isothiocyanate to this region.

Inspection of the rest of the HbS molecular surface

suggests that there are other regions of high electrone-

gativity, e.g., around heme groups. However, there is

no other region of high electronegativity which contains

a pocket into which small ligands can “fall”.

A similar Autodock analysis with aryl isothiocyanate

4 indicates that it also prefers Cysâ93 as its lowest

energy binding site. Further, since it is known that aryl

isothiocyanates are more reactive than alkyl isothio-

cyanates toward nucleophilic attack,²²one would an-

ticipate that the aryl isothiocyanates would also react

with the thiol of Cysâ93. However as we have shown,

the preferred reaction site is the N-terminus of the

R-chain. A visual comparison of these two sites provides

a probable explanation. The R-N-terminal amines are

located near the R-entrance to the center cavity of the

hemoglobin, where there is sufficient space for numer-

ous (at least two) isothiocyanates, thus resulting in an

unhindered reaction path for the isothiocyanates to the

N-terminal amine. In contrast, the Cysâ93 site is a

relatively small pocket (8 Å wide and deep). The

nucleophilic thiol group of interest is located approxi-

mately midway down into the pocket, tucked under the

carbonyl of the surface Aspâ94 (i.e., the thiol is not on

the surface of the protein and is only accessible from

one side). Molecular docking of the aryl and alkyl

isothiocyanates into the Cysâ93 pocket shows the as-

sociation of the thiol and isothiocyanate groups prior

to the nucleophilic attack, indicating steric and electro-

static effects. Docking 7 into this site positions the

isothiocyanate approximately 3.5 Å away from the thiol,

with the alkyl chain bending away from the thiol group.

Docking of the aryl isothiocyanate 4 positions the

isothiocyanate group approximately 4 Å from the thiol

and has the phenyl ring positioned between the two

surface histidines, which are on either side of the

Aspâ94. For the nucleophilic attack to actually occur,

these isothiocyanate groups must move closer to the

thiol. In the alkyl case, this is easily accomplished due

to the negligible steric interactions with the neighboring

residues and the alkyl chain flexibility. In the aryl case,

the phenyl ring is favorably sandwiched between two

histidines and is already in close proximity (less than 3

Å) to the carbonyl of the Aspâ94, prior to the reaction

occurring. Due to the phenyl group’s bulk and rigidity,

it is sterically prevented from moving closer to the thiol

by this Aspâ94, and instead, rotates itself away, so that

the isothiocyanate directly attached to the ring is unable

to get close enough to the thiol for the nucleophilic

attack to take place. Therefore, we postulate that the

aryl isothiocyanates are sterically prevented from react-

ing at the Cysâ93 site and are driven to react with the

less reactive N-terminal amine group, while the smaller

N,N-alkyl derivative reacts at the â93 residue.

Au tod ock An a lysis of th e Rea ction of 2-(N,N-

Dim eth yla m in o)eth yl Isoth iocya n a te (7) w ith th e

Cysâ93 Site of Deoxy-HbS. The original premise of

this work was that if one screened a series of reactive

agents (e.g., substituted isothiocyanates), one might find

agents which would covalently react with specific sites

on a macromolecule. Further, specific alteration in the

chemical structure of the protein could lead to specific

changes in the biological activity of the protein, in this

case Hb. The above experimental results suggest that

such site-specific reactions on a protein surface are

possible. Thus, whereas the alkyl isothiocyanates tend

to react with the Cysâ93 residue, the aryl isothiocyan-

ates tend to react preferentially with the N-terminus

of the R-chain and, to a lesser extent, with the N-

terminus of the â-chain.

To better understand the topological interactions

underlying this regioselectivity, we have used Au-

todock²¹to analyze the binding of the 2-(N,N-dimethyl-

amino)ethyl isothiocyanate ligand 7 to deoxy HbS (pdb

code: 2HBS),²based on the assumption that initial

noncovalent interaction with the protein influences the

final covalent reaction. The results indicate that the

isothiocyanate ligand binds preferentially to the â93

site. The lowest energy result indicates that the isothio-

cyanate unit is approximately 3.5 Å away from the

sulfur of the cysteine. The orientation of the molecule

is such that the isothiocyanate moiety is pushed into

the pocket first, with the N,N-dimethylamino group

sticking slightly out, near the surface of the HbS

molecule. Visual analysis of the electrostatic maps show

the Cysâ93 site to be a rather deep pocket that is

dominated by high electronegativity in the area in and

around the Cysâ93. This electronegativity could be

expected to stabilize the positively charged amino-

isothiocyanate ligand in this region. Further, the posi-

tively charged N,N-dimethylamino group shows an

attraction for the nearby Aspâ94 carboxyl group on the

perimeter surface of the pocket. The isothiocyanate

ligand is positioned between two surface histidines,

Hisâ97 and Hisâ146, which are approximately 4-6 Å

away. These two histidines are separated on one side

by the Aspâ94, and on the other side of the pocket by

ProR37, LysR40, and ThrR41. The Cysâ93 residue is

actually underneath the Aspâ94, but has the reacting

sulfur protruding slightly into the pocket so it is easily

accessible to the incoming alkyl isothiocyanate unit.

Biologica l Activity a n d Cova len t Mod ifica tion .

Ariens²³and Stephenson²⁴introduced the concept of

affinity and efficacy to account for the alteration of

protein function by noncovalent ligand-receptor inter-

actions. It may be of value to examine similar concepts

in discussing biological activity in the context of covalent

modification. For example, in the present work, aryl

isothiocyanates tend to modify the N-terminus of the

R-subunit, while alkyl isothiocyanates preferentially

Since this particular isothiocyanate is not bulky

compared to the size of the pocket, it easily orients so

the isothiocyanate can react with the sulfur of Cysâ93,

while positioning the positively charged amino end

toward the Aspâ94. This provides one explanation for

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 941

react with the Cysâ93 residue. These specificities of the

covalent reactant can be equated to differences in

“affinity” of the reactant for these two modification sites.

Covalent modification of a protein receptor by a ligand

can be described by the following relationship:

similar to different drugs reacting noncovalently at a

single receptor site, as has been suggested above.

The alkyl isothiocyanates used for the evaluation of

antisickling activity are all simple molecules, yet they

have differing charge states: 7, 14, 18-25 are positively

charged amines; 8, 15-17 are neutral compounds; 9 is

negatively charged under the physiological conditions

tested. Among them, amines 7, 14, 19-21, 22-25 had

the most potent antisickling activity. This suggests that

the positive charge (-N⁺at physiological pH) now near

the Cysâ93 residue destabilizes the deoxy HbS polymer

fiber structure.

Rela tion sh ip betw een An tisick lin g Activity a n d

th e Am ou n t of Cova len tly Mod ified HbS. The

anticipated linear relationship between the amount of

protein modification and biological activity was tested

using a reconstitution experiment with HbS modified

with N,N-dimethyl isothiocycante 7. HbS was com-

pletely modified with isothiocyanate 7 and then purified

by gel filtration to remove any unreacted isothiocyanate

and its byproducts. The resulting modified HbS was

then mixed with unmodified HbS to prepare 10%, 50%,

80%, and 90% modified HbS of known concentration.

All samples were screened for antisickling activity

following the standard C_satassay and the data graphi-

cally presented in a plot of solubility increase of HbS

over control against percentage of modification. This plot

is shown in Figure 3 and indicates that an approxi-

mately linear relationship exists between the percentage

of covalent modification and antisickling activity. Drugs

acting through covalent modification should show such

a relationship as predicted above. More specifically, the

observed behavior suggests that 7-modified HbS is not

excluded from the polymer, but is incorporated into the

polymer in a manner that partially destabilizes fiber

formation.

Mech a n ism for t h e An t isick lin g Act ivit y of

Isoth iocya n a te-Mod ified HbS. It is not completely

clear how the alkyl isothiocyanates reacting at the

Cysâ93 residue inhibit polymerization of HbS, although

it appears to occur via more than one mechanism. Both

the X-ray structure of the covalently modified HbS and

molecular modeling studies were used to clarify the

mechanism(s) of antisickling.

P + L T P‚L f P-L f functional change

where P is the protein, L is the covalent modifier, P‚L

is a noncovalent complex, and P-L is the covalently

modified protein leading to a given functional change.

In this scenario, the covalent modifier interacts with the

protein to form a reversible complex, which then co-

valently modifies the protein to alter its biological

function. The on/off rate of the reversible P‚L complex

is likely to be much faster than the covalent bond

forming reaction. In this case, the noncovalent interac-

tion with the protein, as a function of its K_d, could

effectively increase the concentration of L at a specific

site on the protein relative to other solvent accessible

sitessone source of regioselectivity. Under proper condi-

tions, the R-group of the incoming ligand may provide

proper orbital alignment for enhanced reaction at a

given site on the protein enhancing reaction at that

given site.

It should also be recognized that reactants that react

at the same site on a protein can change the structure

of the protein in different ways depending on the

attached R-group. In turn, this can lead to possible

structure/function changes in the protein. The latter

would be comparable to different drugs having different

“efficacy” in the case of noncovalent receptor binding.

After sufficient time for the chemical reaction to reach

completion, one would expect that any biological activity

caused by alteration in the protein’s structure would

show a linear relationship with respect to the concen-

tration of the P-L covalent complex so formed, and that

the intrinsic efficacy of the reacting ligand would be

represented by the slope of this line.

With these thoughts in mind, we tested the antisick-

ling activity of several isothiocyanates to identify which

covalent site gave the greater antisickling activity. In

addition, we examined the relationship of activity with

respect to the amount of covalently modified protein and

evaluated the concept of “efficacy” of structurally dif-

ferent isothiocyanates reacting at the Cysâ93 site.

An tisick lin g Activity of Sick le Hem oglobin Co-

va len tly Mod ified w ith Va r iou s Isoth iocya n a tes.

HbS modified with two aryl (1, 5) and fifteen alkyl

isothiocyanates (7-9, 14-25) were screened for the

inhibition of HbS aggregation using the standard satu-

ration solubility, or C_sat, assay.²⁵As shown in Table 2,

all compounds showed varying degrees of antisickling

activity regardless of the reaction site, i.e., Cysâ93 or

the amino terminus of the R-subunit. However, in

general, alkyl isothiocyanates tended to have better

antisickling activity than aryl isothiocyanates. This

suggests that reaction at the Cysâ93 residue provides

greater antisickling activity than reaction at the N-

terminus of the R-subunit for the compounds studied.

HbS modified with alkyl isothiocyanates 7-9, 14-25

which all reacted at the Cysâ93 residue, also showed

varying degrees of antisickling activity. This result

implies that the substituent on the reactive functional

group plays a role in the inhibition of polymerization

In Table 2, hemoglobin modified with the three alkyl

isothiocyanates 7, 8, and 9 showed increased oxygen

affinity compared with unmodified hemoglobin. This

conclusion was based on their P₅₀values, defined as the

oxygen pressure at which hemoglobin is 50% oxygen-

ated. It has been reported that Cysâ93 is connected with

Hisâ146 through an ionic interaction that is part of a

salt bridge that stabilizes the deoxy conformation of

hemoglobin, and most covalent modifications at Cysâ93

have shown increased oxygen affinity.²⁶The â93Cys

residue is also located adjacent to the proximal Hisâ92

which interacts directly with the heme iron. Thus, it is

reasonable that modification of Cysâ93 could influence

the oxygen affinity of hemoglobin.³The X-ray analysis

of HbS covalently modified with the N,N-dimethylamino

isothiocyanate 7 supports this contention (see below).

On the other hand, in the C_satassay, HbS is first

completely deoxygenated through the chemical reaction

with sodium dithionite and then in the deoxy state,

incubated to generate the polymer. Under these condi-

942 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

Ta ble 2. Inhibition of HbS Polymerization, Oxygen Affinity, and Hill Coefficients for Selected Isothiocyanate-Modified Hemoglobins

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 943

Ta ble 2 (Continued)

a

b

∆log P₅₀of of modified vs unmodified HbA; negative sign indicates that P₅₀of modified Hb is lower than that of unmodified Hb. Hill

coefficient of unmodified HbA: 3.16 lit. from Bonaventura.²⁷

F igu r e 4. Hill plot of HbA modified with various isothio-

cyanates. [ Unmodified HbA, b HbA modified with 8, 1 HbA

F igu r e 3. Inhibition of HbS polymerization vs percent

modified with 9, 9 HbA modified with 7.

modification of HbS with N,N-dimethyl isothiocyanate 7.

dimer interaction and therefore decrease the cooperat-

ivity.

tions, the C_satassay detects only the direct inhibition

X-r ay an d Molecu lar Modelin g An alysis of Deoxy-

H b S Mod ified b y 2-(N,N-Dim et h yla m in o)et h yl

Isoth iocya n a te (7) a t Cysâ93. The X-ray crystal

structure of native deoxyHbA, covalently modified with

the 2-(N,N-dimethylamino)ethyl isothiocyanate 7, was

determined to 1.9 Å resolution and to R-factors of 17.9%

(R_cryst) and 21.3% (R_free), respectively. A summary of the

X-ray data collection and refinement statistics is pre-

sented in Table 3. The final model consists of one (Râ)₂

tetramer per asymmetric unit, with 574 amino acid

residues in four chains, four Fe-heme groups, and 239

water molecules. Analysis of F_o- F_cdifference electron

density maps in the vicinity of Cysâ93, in both â chains

of the tetramer, revealed interpretable density for

2-(N,N-dimethylamino)ethyl isothiocyanate modified

cysteine residues (Lcy93 in Figure 5). The modified

cysteines were built into electron density and then

refined to their final positions as shown in Figure 5.

of HbS polymer formation, and the effect of the in-

creased oxygen affinity or destabilization of the T state

in the inhibition of HbS polymerization is not involved.

Accordingly, the C_satresults listed above indicate that

alkyl isothiocyanates covalently bound to HbS inhibit

the polymerization of HbS not only by increasing the

oxygen affinity, but also by direct inhibition of inter-

molecular contacts in the polymer. It is thus likely that

the shift to the R state will be the initial effect of these

inhibitors, with direct inhibition of polymerization com-

ing into play only under conditions of extensive deoxy-

genation.

HbA modified with 7, 8, and 9 showed smaller Hill

coefficients relative to unmodified HbA (see Figure 4

and Table 2), and ∆log P₅₀decreases of ∼0.17 (7) and

∼0.27 (8 and 9). The Hill coefficients are equivalent to

those observed for Cysâ93 S-nitrosated Hb, and the ∆log

P

₅₀s bracket the ∆log P₅₀of 0.211 observed for Cysâ93

S-nitrosated Hb.²⁷The functional changes for HbA

modified at Cysâ93 by 7, 8, or 9 are very similar to the

effect of NO adduct formation at Cysâ93, suggesting a

common mechanism for functional changes. The Cysâ93

is located close to R₁â₁and R₂â₂contacts within the

tetramer.²⁸Thus, covalent modification of Cysâ93 with

the alkyl isothiocyanate could also disturb the dimer-

Covalent modification of Cysâ93 by the N,N-dimethyl-

amino isothiocyanate 7 wedges the modified side-chain

between residues Aspâ94 and Hisâ146 and consequently

breaks the salt bridge between these two residues. As

previously discussed, this salt bridge is important for

stabilization of the T-state conformation of hemoglobin.

In the T-state, the side-chains of Aspâ94 and Hisâ146

944 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

F igu r e 5. Stereoviews of omit (F_o- F_c) difference Fourier electron density maps, contoured at 2.5σ, for deoxy HbA modified with

2-(N,N-dimethylamino)ethyl isothiocyanate (7). Residues Cysâ93, Aspâ94, and Hisâ146 were omited from Fourier calculations.

The â-subunit B-chain is shown in panel A, and the â-subunit D-chain is shown in panel B in approximately the same orientation.

Differences between the subunits near the modified Cysâ93 (Lcy93) are clearly observable.

Ta ble 3. X-ray Data Collection and Statistics for HbA

Modified with Isothiocyanate 7

In addition, the polar group of the isothiocyanate

ligand’s (N⁺(CH₃)₂) group comes within 4.0 Å of the side

Data Collection

chain of Aspâ94 in the B-chain â-subunit, suggesting

the potential formation of a salt bridge (Figure 5a).

unit cell dimensions

a ) 63.618, b ) 83.792,

c ) 112.259 Å

P2₁2₁2₁(#19)

50.0-1.90 (1.97-1.90)

275074 (1133 rejected)

44466

6.2 (=6)

92.5 (92.4) %

16.0 (2.1)

0.090 (0.527)

The average side-chain B-factor values for Aspâ94 (36

Å²) and Hisâ146 (36 Å²) in the covalently modified

version of HbA lie between the values observed for the

R-state (Aspâ94 ) 50 Å²and Hisâ146 ) 79 Å²) and

T-state (Aspâ94 ) 27 Å²and Hisâ146 ) 29 Å²). The

increased B-factor values suggest that an increase in

the motion of these residues occurs upon modification

of Cysâ93. The electron density surrounding residue

Hisâ146 in both the B- and D-chains of the â subunit is

partially disordered, with slightly more disorder ob-

served in the D-chain, suggesting an increase in the

mobility of the Hisâ146 residue compared to the T-state.

Increased motion of these residues is also consistent

with a previous study in which modification of Cysâ93

by NO produced complete disappearance of the electron

density for the Hisâ146 residue.²⁹Together, these

observations strongly suggest that the covalent modi-

fication disrupts the Hisâ146-Aspâ94 salt bridge nor-

mally observed in the T-state, producing increased

oxygen affinity, reduced cooperativity, and a shift

toward the R-state.

space group

resolution

observations

unique, averaged reflections

redundancy

completeness

I/σ(I)

R(merge)

ø²

1.290 (0.863)

Refinement

resolution

reflections

included (88.0%)

test set (4.7%)

R(cryst)

R(free)

FOM

RMSD Lengths

RMSD Angles

water molecules

20.0-1.90 (1.91-1.90)

42192 (900)

2231 (60)

0.179 (0.293)

0.213 (0.328)

0.862

0.009

1.252

239

Average B-Values (Å²)

all chain atoms

all Fe(porphyrin) atoms

A

B

C

D

29.6

26.8

23.6

30.1

28.3

23.0

22.0

28.6

The covalently modified, deoxyHbA structure was

superimposed with the native deoxyHbA structure using

the program LSQMAN.³⁰Analysis of the overall struc-

ture of covalently modified deoxyHbA indicates that the

main conformational changes that occur in the tertiary

and quaternary structures are localized near the Hisâ146

and Aspâ94 side chains. Therefore, upon covalent

modification, the R-state character of the protein in-

creases, which may explain the observed increase in

oxygen affinity. It is worth noting that the R-state of

hemoglobin does not form a gel, and for this reason, a

HbA conformation with increased R-state character

would be less prone to gelation.²⁹The salt bridge

between Hisâ146-Aspâ94 has been suggested to account

for about half of the overall Bohr effect.^31,32Thus,

disruption of this interaction may explain the observed

increased oxygen affinity of the modified hemoglobin.

Chain RMSD Comparisons

C(R) only

all atoms

A/C

B/D

0.186 (141)

0.288 (146)

0.521 (1069)

0.815 (1132)

are separated by approximately 2.8 Å. This distance is

increased to approximately 7.0 Å in the covalently

modified HbA, and to 9.1 Å in R-state hemoglobin

structures. Similar shifts in the positions of Aspâ94 and

Hisâ146 have been observed in molecular modeling

studies using glutathione as a covalent modifier of

Cysâ93.²⁸The electron density surrounding Aspâ94

extends toward the nitrogen of the (N⁺(CH₃)₂) group in

in both the B- and D-chains of the â-subunit (Figure 5).

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 945

F igu r e 6. (top) Stereoview of the X-ray crystal structure of deoxy hemoglobin HbA covalently modified with 2-(N,N-dimethylamino)-

ethyl isothiocyanate (7), superimposed into the double strand structure of deoxy HbS². Lysine (dark blue), glutamate, and aspartate

(both red) residues are within 15 Å from the covalently modified Cys â 93 (light blue) site. (bottom) Stereo close up of the substitution

region.

The covalently modified deoxy HbA was also super-

imposed onto the deoxy HbS structure with the Gluâ6Val

substitution. The resulting superposition is shown in

Figure 6a in an orientation depicting the double strand

of hemoglobin found in the deoxyhemoglobin S crystal.²

A close up of the region of interest is shown in Figure

6b, where the insertion of a cationic group at Cysâ93

occurs in a “gap” region where the overall electrostatic

potential would be dominated by four Lys residues

on neighboring molecules (Lysâ8, Lysâ120, and two

946 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

Lysâ17’s), as well as two other Lys residues slightly

further away from the Cysâ93 modification (Lysâ65 and

Lysâ66). Besides these Lys residues on the neighboring

molecules of the adjacent strand(s), there are three Lys

residues near the covalently modified Cysâ93 (LysR40,

Lysâ95, and Lysâ144 which are not shown due to their

close proximity to Cysâ93) for the native HbS polymer

fiber. Some of these Lys residues are neutralized locally

by nearby Glu or Asp residues. This is the case for the

Gluâ121 next to Lysâ120, as well as the Gluâ90 (also

not shown in Figure 6) near Lysâ144, and Lysâ95 on

the other side of Aspâ94. There are a few other Asp and

Glu residues nearby that are unlikely to contribute

significantly toward the effect of the Lys residues’

positive potential (Aspâ73, Aspâ79, Gluâ7, and AspR47).

This is due to the participation of these Asp and Glu

residues in the intermolecular lateral contact site, and

hence influence the Lys residues in this “gap” region

very little.

dimethylamine could be designed to see if inhibitory

effects are less, due to the protonated amine being too

far from the COO^-of the Aspâ94, but yet effective due

to the influence on the electric field.

Su m m a r y

In this work, we have shown that design of agents

for regioselective modification of Hb is possible through

selection of an appropriate group for molecular recogni-

tion, coupled with a group that permits covalent reaction

at the recognition site. Although it is well-known that

isothiocyanates can react with both free amines and

thiol groups, we have shown that for the Hb system,

the isothiocyanate reactive functionality, coupled to

aromatic groups, reacts selectively with the N-terminal

amino groups, with a preference for the R-chain. In

contrast, the isothiocyanate functionality, coupled to

aliphatic groups, reacts exclusively with the Cysâ93

thiol when added at stoichiometric or sub-stoichiometric

levels.

Although this represents only a low resolution case

for reaction specificity, it implies that design of a

molecule with a precise fit on the protein surface might

provide a molecular orientation which could greatly

enhance covalent reactivity at one site over another on

a protein surface. The ability to produce such a specific

covalent modification could represent a unique approach

to altering protein structure/function in pharmaceutical

design.

Thus in the present case, compounds 7, 8, and 9, with

cationic, neutral hydroxyl, and anionic R-groups, re-

spectively, provide intriguing contrasts in their func-

tional effects. All three show nearly identical effects on

the oxygen transport function of Hb, with similar

reductions in both the Hill coefficients and the P₅₀

values, indicating similar shifts from T- toward R-state

for all three modified Hbs. In this respect, the effects

are similar to a variety of compounds including glu-

tathione,²⁸N-ethylmaleimide,^28,33iodoacetamide,³³and

NO.²⁷There is a striking difference in their effect on

HbS polymerization, however, with the cationic 7 (and

its analogues 23-25) approximately three times more

effective at inhibition of HbS polymerization than the

anionic 9 or the neutral hydroxylic 8, which both exhibit

inhibitory activities similar to glutathione,²⁸NO,²⁷and

other SH-modifying reagents.

As discussed above, we speculate that introduction

of a cationic group at Cysâ93 and the consequent

displacement of the Hisâ146 side chain imidazole com-

bine to produce a subtle modification of the electrostatic

field that opposes the field from residues in neighboring

molecules within the fiber. In addition, increasing the

oxygen affinity of HbS is known to decrease HbS

polymerization within the red cell under conditions of

partial deoxygenation, and to reduce the red cell mor-

phological changes associated with sickle cell disease.³⁴

Regardless of the mechanism of action, in the present

work the substantial increase in inhibitory activity

against HbS polymerization, combined with the effect

of increasing Hb oxygen affinity and relatively facile

traversal of the erythrocyte membrane suggest that the

cationic isothiocyanates 7 and 23-25 show substantial

promise as lead compounds for the development of

therapeutic agents for sickle cell disease.

In the native structure Aspâ94 forms a salt bridge

with Hisâ146, effectively neutralizing the charge for

both residues. Upon adduct formation with 7 at Cysâ93,

the imidazole ring of Hisâ146 is displaced, inserting a

weakly basic (cationic) group approximately into the

center of the positive electrostatic potential generated

by these Lys residues. Local charge neutralization of

Aspâ94 appears to be maintained by the cationic group

of 7 (Figure 5), as well as a likely hydrogen bond

formation between the protonated amine of 7 and the

carboxylate group of Aspâ94 (7 to Aspâ94 N- - -O

distances are ∼2.5 Å and 3.3 Å). The change in electric

field vectors can be seen in Figure 7, which shows that,

in the presence of the positively charged 7, the largest

change in electric field occurs adjacent to the side chain

imidazole of Hisâ146 and the Aspâ94 COO^-group. The

change of the field surrounding the Hisâ146 is more

likely due to the movement of the imidazole ring being

pushed down from the Aspâ94 and minimally due to

the nearby positively charged ligand. The change of the

field surrounding the Aspâ94, however, from one which

is pointed away from the Asp’s COO^-group in the

unbound Hb to one where a portion of the field has been

eliminated in the liganded Hb is likely due to the

additional positive charge of the ligand influencing the

electric field vectors in and around this COO^-.

For the anionic 9 (COO^-group) and neutral 8 (OH),

inhibitory activities are likely because of the disruption

of the Aspâ94-Hisâ146 salt bridge. However, they would

not produce charge neutralization of the Aspâ94 side

chain, which accounts for their lower inhibitory activi-

ties. The fact that they show essentially equivalent

activities may be accountable by two effects. The result-

ing anionic group of 9 (which is farther into this “gap”

region due to the longer chains of 9 compared to 7) may

interact favorably with the Lys-generated electrostatic

potentialsin effect, partially compensating the attrac-

tion-repulsion effects between the Aspâ94 carboxylate/

Cysâ93-adduct and the displaced Hisâ146 imidazole.

Although 8 lacks a charge, it does have a hydrogen

available for hydrogen bonding interactions with the

COO^-of Aspâ94. This would imply that both charge

effects and hydrogen bonding effects at the Aspâ94 may

influence the degree of inhibitory activity of these

molecules. To test this hypothesis, a longer alkyl chain

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 947

F igu r e 7. Electric field vectors adjacent to residues Lcyâ93, Aspâ94, and Hisâ146. (Top) Field for native Hb. (Bottom) Field for

modified Hb.

the Hb surface. Mass spectra were recorded with a MAT 90

for CI, Finnigan LCQ for APCI and routine ESMS, and a

Micromass Quattro II for Hb analyses and HRMS. UV absor-

bances were recorded on a J asco V550 spectrophotometer.

Melting points were determined on a Thomas-Hoover melting

point apparatus and are uncorrected. High-performance liquid

chromatography (HPLC) was performed on a Waters 510 liquid

chromatograph. Thin-layer chromatography (TLC) was carried

out on 0.25 mm precoated Whatman SILG/UV silica plates.

Routine flash column chromatography was effected on silica

gel 60 purchased from Aldrich. All solvents used were either

Exp er im en ta l Section

Gen er a l. All nuclear magnetic resonance (NMR) were

obtained using Bruker Avance 300 and 500 MHz FT NMR

spectrometers. All absorptions are expressed in parts per

million relative to tetramethylsilane as the internal standard.

All infrared (IR) spectra were obtained using a J asco V550

FT/IR-410 spectrophotometer. All elemental analyses were

performed by Midwest Microlab of Indianapolis, IN, and were

within (0.4% of the theoretical values. Elemental analyses

were obtained for all isothiocyanates tested biologically, but

not for those used only to examine regiospecific reactions on

948 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

HPLC grade (Fisher Scientific) or anhydrous (Aldrich). All

other chemicals were purchased from either Aldrich or Fisher

Scientific.

mixture was diluted with CH₂Cl₂(50 mL), washed with 3 N

HCl (50 mL), and then extracted twice with 10% aqueous Na₂-

CO₃(50 mL). The combined aqueous solutions were washed

with CH₂Cl₂(50 mL), acidified to pH 1.0 with concentrated

HCl, and extracted twice with CH₂Cl₂(50 mL). The combined

organic layers were dried over MgSO₄. Column chromatogra-

phy (1:1 hexane/EtOAc) yielded a white solid (6) (0.10 g, 0.52

mmol, 83% yield); mp 142-144 °C. IR (KBr): 2194, 2138, 1695

Gen er a l P r oced u r e for th e Syn th esis of Isoth iocya n -

a tes 1-4, 6, 8-18. These compounds were prepared by a

previously reported procedure.¹³To a stirred solution of the

primary amine in CH₂Cl₂was added di-2-pyridyl thionocar-

bonate (1.0 equiv) at 25 °C. The reaction mixture was stirred

for 24 h, unless otherwise noted. The solvent was removed by

rotary evaporation. The crude product was purified by either

column chromatography on silica gel or vacuum distillation.

4-(2-Hyd r oxyeth yl)p h en ylisoth iocya n a te (1). 4-(2-Hy-

droxyethyl)aniline (0.14 g, 1.02 mmol) was dissolved in CH₂-

Cl₂(20 mL). To this mixture was added di-2-pyridyl thiono-

carbonate (0.23 g, 1.02 mmol). Column chromatography (1:1

hexane/EtOAc) yielded a yellow oil (1) (0.16 g, 0.89 mmol, 87%

yield), which crystallized in the freezer upon solvent removal;

cm^-1. ¹H NMR (CDCl₃): δ 7.30-7.15 (m, 4H), 3.66 (s, 2H). ¹³

C

NMR (CDCl₃): δ 177.31, 132.87, 131.05, 130.92, 126.33,

126.30, 40.87. APCIMS m/z 192([M - H]^-, 56), 148([M -

CO₂]^-, 100).

4-Isoth iocya n a tobu ta n ol (8). 4-Aminobutanol (0.09 g, 1.0

mmol) was dissolved in CH₂Cl₂(20 mL). To this mixture was

added di-2-pyridyl thionocarbonate (0.23 g, 1.0 mmol). Column

chromatography (1:2 hexane/EtOAc) yielded a pale yellow oil

(8) (0.12 g, 0.92 mmol, 91% yield). IR (neat): 3347, 2946, 2186,

1

2108 cm^-1. H NMR (CDCl₃): δ 3.71 (t, J ) 6.6 Hz, 2H), 3.60

mp 42-44 °C. IR (neat): 3313, 2945, 2869, 2172, 2092 cm^-1

.

¹H NMR (CDCl₃): δ 7.35-7.11 (m, 4H), 3.83 (t, J ) 6.6 Hz,

2H), 2.85 (t, J ) 6.6 Hz, 2H). ¹³C NMR (CDCl₃): δ 138.68,

135.26, 130.57, 129.78, 126.24, 63.66, 39.09. CIMS m/z

180([M + H]⁺, 100), 162([M - OH]⁺, 43). Anal. (C₉H₉NOS) C,

H, N.

(t, J ) 6.6 Hz, 2H), 1.99 (bs, 1H), 1.88-1.76 (m, 2H), 1.76-

1.64 (m, 2H). ¹³C NMR (CDCl₃): δ 62.21, 45.36, 29.88, 26.99.

CIMS m/z 132([M + H]⁺, 95), 114([M - OH]⁺, 100). Anal.

(C₅H₉NOS) C, H, N.

4-Isoth iocya n a tobu tyr ic Acid (9). 4-Aminobutyric acid

(0.20 g, 1.94 mmol) was dissolved in CH₂Cl₂(20 mL). To this

mixture was added di-2-pyridyl thionocarbonate (0.45 g, 1.94

mmol). Column chromatography (1:1 hexane/EtOAc) yielded

a clear oil (9) (0.21 g, 1.45 mmol, 75% yield). IR (neat): 2191,

4-Isoth iocya n a top h en ol (2). 4-Aminophenol (0.11 g, 1.01

mmol) was dissolved in CH₂Cl₂(20 mL). To this mixture was

added di-2-pyridyl thionocarbonate (0.23 g, 1.01 mmol). Col-

umn chromatography (1:2 hexane/EtOAc) yielded a pale

yellow solid (2) (0.11 g, 0.073 mmol, 73% yield); mp 42-44 °C.

1

2111, 1709 cm^-1. H NMR (CDCl₃): δ 10.75 (bs, 1H), 3.65 (t,

1

IR (KBr): 3397, 2173, 2119 cm^-1. H NMR (CDCl₃): δ 7.15-

J ) 7.2 Hz, 2H), 2.55 (t, J ) 7.2 Hz, 2H), 2.03 (q, J ) 7.2 Hz,

2H). ¹³C NMR (CDCl₃): δ 179.05, 44.59, 31.06, 25.29. CIMS

m/z 146([M + H]⁺, 62), 128([M - OH]⁺, 100), 87([M - NCS]⁺,

50). Anal. (C₅H₇NO₂S) C, H, N.

7.09 (m, 2H), 6.82-6.76 (m, 2H), 5.75 (s, 1H). ¹³C NMR

(CDCl₃): δ 155.15, 134.17, 127.62, 124.10, 116.78. APCIMS

m/z 150([M - H]^-, 100).

4-Isoth iocya n a tocycloh exa n ol (10). To 4-aminocyclohex-

anol HCl (0.10 g, 0.66 mmol) in CH₂Cl₂(20 mL) and 10%

aqueous Na₂CO₃(5 mL) was added di-2-pyridyl- thionocar-

bonate (150 mg, 0.66 mmol) at room temperature. After 5 h,

the solution was diluted with CH₂Cl₂(50 mL) and washed with

water (30 mL) and 3 N HCl (30 mL). The organic layer was

dried over MgSO₄and evaporated under reduced pressure.

Column chromatography (1:2 hexane/EtOAc) afforded a color-

less oil (10) (0.060 g, 0.39 mmol, 47% yield). IR (neat): 3347,

2940, 2150, 2097 cm^-1. ¹H NMR (CDCl₃): δ 3.86-3.72 (m, 1H),

3.72-3.55 (m, 1H), 2.22-2.05 (m, 2H), 2.10-1.75 (m, 3H),

1.71-1.55 (m, 2H), 1.52-1.20 (m, 2H). ¹³C NMR (CDCl₃): δ

130.76, 68.05, 55.01, 31.87, 30.28. CIMS m/z 158 ([M + H]⁺,

52), 99([M - NCS]⁺, 69), 81([M - NSC - H₂O]⁺, 100).

4-(2-(N,N-Diet h yla m in o)et h yla m in oca r b on yl)p h en yl

Isot h iocya n a t e Mon oh yd r och lor id e Sa lt (3). Procain-

amide HCl (0.2 g, 0.74 mmol) in CH₂Cl₂(20 mL) was cooled to

0 °C in an ice bath. To this were added 1 mL of H₂O and

NaHCO₃(500 mg). This was then followed by the addition of

di-2-pyridyl thionocarbonate (0.17 g, 0.74 mmol). After 30 min,

the solution was diluted with CH₂Cl₂(50 mL) and washed with

water. The organic layer was dried over MgSO₄and evapo-

rated. The residue was diluted with ethyl acetate (30 mL), and

then HCl was bubbled through the solution for an additional

5 min at 0 °C. The mixture was warmed to room temperature

and stirred for 30 min. The precipitate that formed was filtered

and washed with ethyl acetate. Recrystallization from acetone

afforded a yellow solid (3) (0.072 g, 0.023 mmol, 31% yield);

mp 178-180 °C. IR (KBr): 3238, 2603, 2188, 2133, 1655 cm^-1

.

N-(3-Isoth iocya n a top r op yl)-γ-bu tyr ola cta m (11). N-(3-

Aminopropyl)-γ-butyrolactam (0.20 g, 1.41 mmol) was dis-

solved in CH₂Cl₂(20 mL). To this mixture was added di-2-

pyridyl thionocarbonate (0.33 g, 1.41 mmol). Column chromatog-

raphy (EtOAc) yielded a yellow oil (11) (0.08 g, 0.43 mmol,

¹H NMR (CDCl₃): δ 11.65 (bs, 1H), 9.11 (bs, 1H), 8.17-8.11

(m, 2H), 7.32-7.26 (m, 2H), 3.96-3.87 (m, 2H), 3.29-3.23 (m,

2H), 3.23-3.12 (m, 4H), 1.44 (t, J ) 7.5 Hz, 6H). ¹³C NMR

(CDCl₃): δ 166.78, 134.91, 132.03, 129.63, 126.16, 53.47, 48.82,

35.76, 9.04. APCIMS m/z 278([M + H]⁺, 100).

43% yield). IR (neat): 2187, 2111, 1682 cm^-^{1 1}H NMR

.

(CDCl₃): δ 3.59 (t, J ) 6.6 Hz, 2H), 3.47-3.32 (m, 4H), 2.43

4-Isot h iocya n a t o-N,N-d im et h yla n ilin e Mon oh yd r o-

ch lor id e Sa lt (4). 4-Amino-N,N-dimethylaniline HCl (0.10 g,

0.58 mmol) in CH₂Cl₂(20 mL) was cooled to 0 °C in an ice

bath. To this were added 1 mL of H₂O and NaHCO₃(500 mg).

This was then followed by the addition of di-2-pyridyl thiono-

carbonate (0.134 g, 0.58 mmol). After 30 min, the solution was

diluted with CH₂Cl₂(50 mL) and washed with water. The

organic layer was dried over MgSO₄and evaporated. The

residue was diluted with ethyl acetate (30 mL), and then HCl

was bubbled through the solution for an additional 5 min at 0

°C. The mixture was warmed to room temperature and stirred

for 30 min. The precipitate that formed was filtered and

washed with diethyl ether. Column chromatography (4:1

hexane/EtOAc) afforded a white solid (4) (0.080 g, 0.037 mmol,

78% yield); mp 144-146 °C. IR (KBr): 2408, 2200, 2143, 1506

(t, J ) 7.8 Hz, 2H), 2.14-2.03 (m, 2H), 2.02-1.85 (m, 2H). ¹³

C

NMR (CDCl₃): δ 176.11, 48.03, 43.39, 40.60, 31.23, 28.19,

18.34. CIMS m/z 185([M + H]⁺, 100), 126([M - NCS]⁺, 79).

2-(In d ol-3-yl)eth yl Isoth iocya n a te (12). 6-(2-Aminoeth-

yl)indole (0.20 g, 1.25 mmol) was dissolved in CH₂Cl₂(20 mL).

To this mixture was added di-2-pyridyl thionocarbonate (0.29

g, 1.25 mmol). Column chromatography (5:1 hexane/EtOAc)

yielded a white solid (12) (0.183 g, 0.91 mmol, 91% yield); mp

46-47 °C. IR (KBr): 3398, 2185, 2097, 1456 cm^-^{1 1}H NMR

.

(CDCl₃): δ 8.04 (bs, 1H), 7.54 (d, J ) 7.8 Hz, 1H), 7.35 (d, J )

7.8 Hz, 1H), 7.26-7.12 (m, 2H), 7.04 (d, J ) 2.4 Hz, 1H), 3.72

(t, J ) 6.9 Hz, 2H), 3.12 (t, J ) 6.9 Hz, 2H). ¹³C NMR (CDCl₃):

δ 136.68, 130.07, 127.22, 123.46, 122.77, 120.12, 118.71,

111.90, 111.59, 46.09, 26.90. APCIMS m/z 203([M + H]⁺, 100),

144([M - HNCS]⁺, 36).

cm^-^{1 1}H NMR (DMSO-d₆): δ 7.38 (d, J ) 8.4 Hz, 2H), 7.06

.

(bs, 2H), 6.20 (bs, 1H), 2.99 (s, 2H). ¹³C NMR (DMSO-d₆): δ

146.96, 127.98, 118.21, 43.35. APCIMS m/z 179([M + H]⁺, 100).

(4-Isoth iocya n a tom eth yl)p yr id in e (13). 4-Aminometh-

ylpyridine (0.15 g, 1.39 mmol) was dissolved in CH₂Cl₂(20 mL).

To this mixture was added di-2-pyridyl thionocarbonate (0.32

g, 1.39 mmol). After following the generalized procedure above,

column chromatography (1:2 hexane/EtOAc) yielded a brown

oil (13) (0.20 g, 1.33 mmol, 96% yield). IR (neat): 2198, 2095

4-Isoth iocya n a top h en yla cetic Acid (6). 4-Aminophenyl-

acetic acid (0.10 g, 0.66 mmol) was dissolved in CH₂Cl₂(20

mL). To this mixture was added di-2-pyridyl thionocarbonate

(0.16 g, 0.66 mmol), which was stirred for 4 h at 25 °C. The

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 949

1

cm^-1. H NMR (CDCl₃): δ 8.65 (d, J ) 4.5 Hz, 2H), 7.28 (d, J

2H). ¹³C NMR (DMSO-d₆): δ 171.79, 162.56, 136.77, 136.53,

132.75, 118.06, 114.81, 113.45. APCIMS m/z 194([M - H]^-,

100), 150([M - CO₂]^-, 12). Anal. (C₈H₅NO₃S) C, H, N.

) 4.5 Hz, 2H), 4.79 (s, 2H). ¹³C NMR (CDCl₃): δ 150.75, 143.59,

134.82, 121.62, 47.97. APCIMS m/z 151([M + H]⁺, 100), 93-

([M - NCS]⁺, 13).

Gen er a l P r oced u r e for th e Syn th esis of Isoth iocya n -

a tes 7, 19-21. These compounds were all prepared by a

previously reported procedure.¹⁴The primary amine in acetone

(5 mL) was cooled to -10 °C in a dry ice/acetone bath. Carbon

disulfide (1.1 equiv) in acetone (5 mL) was added dropwise

via an addition funnel over 15 min. Then, over 45 min, the

temperature was allowed to slowly rise to 10 °C. The reaction

mixture was then cooled to -15 °C. Over 45 min, a solution of

HgCl₂(1.0 equiv) in acetone (15 mL) was added, keeping the

temperature below -10 °C. After warming to 0 °C over 15 min,

Et₃N (2.2 equiv) was added slowly by syringe. Finally, the

reaction mixture was refluxed for 1 h, becoming black. Upon

cooling, the salts were filtered through Celite, and washed

subsequently with acetone (1 × 50 mL) and ether (2 × 50 mL).

The filtrate was treated with MgSO₄. Solvent was removed

by rotary evaporation. To the residue was added ether (75 mL),

and this was filtered. The filtrate was concentrated and

vacuum distillation yielded the isothiocyanates (7, 19-21).

1-(2-Isoth iocya n a toeth yl)p ip er id in e (14). 1-(2-Amino-

ethyl)piperidine (0.13 g, 1.0 mmol) was dissolved in CH₂Cl₂

(10 mL). To this mixture was added di-2-pyridyl thionocar-

bonate (0.23 g, 1.0 mmol). After following the generalized

procedure above, column chromatography (1:2 hexane/EtOAc)

yielded a clear oil (14) (0.082 g, 0.048 mmol, 47% yield). IR-

(neat): 2936, 2190, 2096 cm^-1. H NMR (CDCl₃): δ 3.59 (t, J

1

) 6.6 Hz, 2H), 2.64 (t, J ) 6.6 Hz, 2H), 2.55-2.35 (m, 4H),

1.68-1.48 (m, 4H), 1.49-1.38 (m, 2H). ¹³C NMR (CDCl₃): δ

58.42, 54.81, 43.49, 26.33, 24.53. APCIMS m/z 171([M + H]⁺,

100), 112([M - NCS]⁺, 23). Anal. (C₈H₁₅N₂SCl) C, H, N.

5-Isoth iocya n a top en ta n ol (15). 5-Aminopentanol (0.10 g,

1.0 mmol) was dissolved in CH₂Cl₂(20 mL). To this mixture

was added di-2-pyridyl thionocarbonate (0.23 g, 1.0 mmol).

Column chromatography (1:2 hexane/EtOAc) yielded a clear

oil (15) (0.10 g, 0.69 mmol, 71% yield). IR (neat): 3347, 2940,

2182, 2106 cm^-1. ¹H NMR (CDCl₃): δ 3.68 (t, J ) 6.3 Hz, 2H),

3.54 (t, J ) 6.3 Hz, 2H), 1.82-1.70 (m, 2H), 1.67-1.45 (m, 4H).

¹³C NMR (CDCl₃): δ 62.88, 45.41, 32.20, 30.17, 23.35. APCIMS

2-(N,N-Dim eth yla m in o)eth yl Isoth iocya n a te (7). This

compound has been previously prepared and spectroscopically

characterized.³⁷Anal. (C₅H₁₁N₂SCl) C, H, N.

m/z 146([M + H]⁺, 63), 128([M - OH]⁺, 100). Anal. (C₆H₁₁

NOS) C, H, N.

-

2-(N,N-Dieth yla m in o)eth yl Isoth iocya n a te (19). 2-(Di-

ethylamino)ethylamine (2.0 g, 17.2 mmol) was dissolved in 15

mL of acetone. After following the procedure outlined above,

vacuum distillation (bp 118-122 °C/12 mmHg) yielded a clear

liquid (19) (1.57 g, 9.94 mmol, 58% yield). This compound

matched the previously reported boiling point and spectral

data.^{38 13}C NMR (CDCl₃): δ 131.98, 52.53, 47.44, 44.01, 12.10.

HRMS ESI+ (m/z) calcd. for (M + H)⁺C₇H₁₅N₂S, 159.0956;

found 159.0961. Anal. (C₇H₁₅N₂SCl) C, H, N.

3-(N,N-Dim et h yla m in o)p r op yl Isot h iocya n a t e (20).

3-(Dimethylamino) propylamine (1.0 g, 9.80 mmol) was dis-

solved in 10 mL acetone. After following the procedure outlined

above, vacuum distillation (bp 110-111 °C/10 mmHg) yielded

a clear liquid (20) (0.99 g, 6.88 mmol, 70% yield). This

compound matched the previously reported boiling point and

spectral data.^{38 13}C NMR (CDCl₃): δ 131.08, 55.92, 45.34,

42.91, 27.89. Anal. (C₆H₁₂N₂S) C, H, N.

3-(N,N-Dieth yla m in o)p r op yl Isoth iocya n a te (21). 3-(Di-

ethylamino)propylamine (1.0 g, 7.69 mmol) was dissolved in

10 mL of acetone. After following the procedure outlined above,

vacuum distillation (bp 129-131°C/15 mmHg) yielded a clear

liquid (21) (0.85 g, 4.94 mmol, 65% yield). This compound

matched the previously reported boiling point and spectral

data.^{38 13}C NMR (CDCl₃): δ 131.23, 49.21, 46.86, 43.03, 27.88,

11.74. HRMS ESI+ (m/z) calcd. for (M + H)⁺C₈H₁₇N₂S,

173.1112; found 173.1114. Anal. (C₈H₁₇N₂SCl) C, H, N.

2,2-Dim eth oxyeth yl Isoth iocya n a te (16). Aminoacetal-

dehyde dimethyl acetal (0.20 g, 1.90 mmol) was dissolved in

CH₂Cl₂(20 mL). To this mixture was added di-2-pyridyl

thionocarbonate (0.44 g, 1.90 mmol). The product, obtained by

vacuum distillation (bp 105-106 °C/15 mmHg), was a clear

liquid (16) (0.34 g, 0.231 mmol, 12% yield) after following the

generalized procedure above. This compound matched the

previously reported boiling point,^35,36as well as IR, mass

spectrometry, and proton NMR data.^{35 13}C NMR (CDCl₃): δ

130.98, 101.53, 54.18, 54.17, 46.51. Anal. (C₅H₉NO₂S) C, H,

N.

1-Isoth iocya n a top r op yl Nitr ile (17). 3-Aminopropioni-

trile fumarate salt (0.30 g, 2.34 mmol) in CH₂Cl₂(30 mL) was

cooled to 0 °C in an ice bath. To this were added 1 mL of H₂O

and a small spatula full of NaHCO₃. This was then followed

by the addition of di-2-pyridyl thionocarbonate (0.54 g, 2.34

mmol). The ice bath was removed, and the reaction mixture

was allowed to stir for 24 h at 25 °C. This was followed by an

extraction with saturated NaHCO₃solution (2 × 30 mL) and

saturated NaCl solution (2 × 30 mL) to remove starting

materials and byproducts. The organic layer was dried with

MgSO₄, and the solvent was removed by rotary evaporation.

Column chromatography (1:1 hexane/EtOAc) yielded a clear

oil (17) (0.149 g, 1.33 mmol, 57% yield). IR (NaCl): 2208, 2112,

2086 cm^-1. ¹H NMR (CDCl₃): δ 3.88 (t, J ) 13.1 Hz, 2H), 2.78

(t, J ) 13.1 Hz, 2H). ¹³C NMR (CDCl₃): δ 130.89, 115.98, 41.04,

19.05. Anal. (C₄H₄N₂S) C, H, N.

1-(3-Isoth iocya n a top r op yl)im id a zole (18). 1-(3-Amino-

propyl)imidazole (0.1 g, 0.80 mmol) was dissolved in CH₂Cl₂

(10 mL). To this mixture was added di-2-pyridyl thionocar-

bonate (0.19 g, 0.80 mmol). After following the generalized

procedure above, column chromatography (1:1 hexane/EtOAc)

yielded an orange oil (18) (0.08 g, 0.48 mmol, 60% yield). IR

(NaCl): 2201, 2115, 1653, 1508, 1135 cm^-1. ¹H NMR (CDCl₃):

δ 7.50 (s, 1H), 7.07 (d, J ) 1.1 Hz, 2H), 6.94 (d, J ) 1.3 Hz,

1H), 4.12 (dt, J ) 6.6, 2.0 Hz, 2H), 3.51 (dt, J ) 5.7, 1.7 Hz,

2H), 2.19-2.09 (m, 2H). ¹³C NMR (CDCl₃): δ 137.09, 131.92,

130.01, 118.67, 43.37, 41.76, 30.08. Anal. (C₇H₉N₃S) C, H, N.

4-(N,N-Dim eth yla m in o)bu tyl Isoth iocya n a te Mon oh y-

d r och lor id e Sa lt (22). Following a procedure similar to

Kupchan’s,^15,164-chlorobutyronitrile (28) (10.0 g, 97.1 mmol),

dimethylamine (26) (73.0 mL, 582.5 mmol), and sodium iodide

(1.46 g, 9.71 mmol) were stirred together for 5 d at 25 °C.

Vacuum distillation (78-80 °C/15 mmHg) yielded a clear

liquid, 4-dimethylaminobutyronitrile (30) (3.15 g, 28.1 mmol,

29% yield). This compound matched the previously reported

boiling point.^16,39,40IR (NaCl): 2944, 2248, 1148, 1040 cm^-1

.

¹H NMR (CDCl₃): δ 2.44-2.36 (m, 4H), 2.22 (s, 6H), 1.82-

1.78 (m, 2H). ¹³C NMR (CDCl₃): δ 119.55, 57.34, 45.07, 23.27,

14.52. HRMS ESI+ (m/z) calcd. for (M + H)⁺C₆H₁₃N₂,

113.1079; found 113.1043. Compound 30 (1.7 g, 15.2 mmol)

and the Rh-Al₂O₃catalyst (0.47 g) were dissolved in MeOH

(40 mL). Hydrogenation was carried out in a Parr shaker

under H₂at 37 psi of pressure for 5 h. When hydrogen uptake

was complete, the solution was filtered through Celite to

remove the catalyst. The filtrate was concentrated by rotary

evaporation. The crude product contained a mixture of primary

(N,N-dimethylbutane-1,4-diamine, 34) and secondary amines,

2:1, respectively, which were not separated. IR (NaCl): 3291,

4-Isoth iocya n osa licylic Acid (5). This procedure was

adapted from that of Seligman.¹²p-Aminosalicylic acid (1.53

g, 11.2 mmol) was suspended in water (32 mL) and concen-

trated HCl (3.4 mL). To this suspension, thiophosgene (1.35

g, 11.7 mmol) was added in one portion, and the reaction

mixture was stirred for 3 h at 25 °C. The solid product was

filtered, washed with water (100 mL), and dried under vacuum

over P₂O₅. The resulting solid was recrystallized from benzene

to yield a white solid (5) (1.1 g, 6.1 mmol, 55% yield); mp 185-

1

1466, 1308, 1041 cm^-1. H NMR (CDCl₃) (peaks for 34 only):

187 °C. IR (KBr): 3433, 2077, 1665 cm^-1

d₆): δ 11.45 (bs, 2H), 7.83 (d, J ) 8.4 Hz, 1H), 7.02-6.94 (m,

.

¹H NMR (DMSO-

δ 2.71 (t, J ) 6.7 Hz, 2H), 2.61 (t, J ) 6.7 Hz, 2H), 2.22 (s,

6H), 1.51-1.46 (m, 4H). ¹³C NMR (CDCl₃): δ 49.87, 45.43,

950 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

42.09, 25.50, 25.02. The mixture of primary (34) and secondary

amines (1.4 g, 12.1 mmol/molecular weight of 34) was dissolved

in 20 mL of acetone. After following the procedure used for

compounds 7, 19-21, the resulting crude product was a yellow

liquid, 4-(N,N-dimethylamino)butylisothiocyanate (38). IR

(NaCl): 2942, 2173, 2103, 1264, 1041 cm^-1. ¹H NMR (CDCl₃):

δ 3.55 (t, J ) 6.5 Hz, 2H), 2.29 (t, J ) 7.1 Hz, 2H), 2.22 (s,

6H), 1.75-1.72 (m, 2H), 1.61-1.56 (m, 2H). ¹³C NMR

(CDCl₃): δ 129.80, 58.60, 45.39, 44.33, 27.88, 24.54. Crude

isothiocyanate (38) (1.1 g, 6.96 mmol) was dissolved in diethyl

ether (30 mL), and HCl gas was bubbled through the solution.

The white precipitate (22) so formed was very hygroscopic and

could not be filtered. The product was washed with ether (3 ×

30 mL) and decanted to yield a yellow-orange gum. It was then

dried under vacuum (0.91 g, 4.69 mmol, 67% yield, 9% overall

119.58, 58.41, 45.30, 26.41, 23.17, 16.94. HRMS ESI+ (m/z)

calcd for (M + H)⁺C₇H₁₅N₂, 127.1235; found 127.1235.

Compound 32 (1.0 g, 7.94 mmol) and the Rh-Al₂O₃catalyst

(0.25 g) were dissolved in MeOH (40 mL). Hydrogenation was

carried out in a Parr shaker under H₂at 37 psi of pressure for

5 h. When hydrogen uptake was complete, the solution was

filtered through Celite to remove the catalyst. The filtrate was

concentrated by rotary evaporation. The crude product con-

tained a mixture of primary (N,N-dimethylpentane-1,5-di-

amine, 36) and secondary amines, 1:1, respectively, which were

not separated. IR (NaCl): 3293, 1465, 1307, 1041 cm^-^{1 1}H

.

NMR (CDCl₃) (peaks for 36 only): δ 2.69 (t, J ) 6.9 Hz, 2H),

2.52 (t, J ) 7.2 Hz, 2H), 2.22 (s, 6H), 1.50-1.42 (m, 4H), 1.38-

1.30 (m, 2H). ¹³C NMR (CDCl₃): δ 50.03, 45.49, 42.15, 27.66,

25.26, 24.75. The mixture of primary (36) and secondary

amines (1.0 g, 7.69 mmol/molecular weight of 36) was dissolved

in 15 mL of acetone. After following the procedure used for

compounds 7, 19-21, the resulting crude product was a yellow

liquid, 5-(N,N-dimethylamino)pentyl isothiocyanate (40). IR

(NaCl): 2939, 2180, 2105, 1272, 1041 cm^-1. ¹H NMR (CDCl₃):

δ 3.53 (t, J ) 6.7 Hz, 2H), 2.29 (t, J ) 7.4 Hz, 2H), 2.21 (s,

6H), 1.75-1.70 (m, 2H), 1.50-1.43 (m, 2H). ¹³C NMR

(CDCl₃): δ 129.86, 59.22, 45.28, 44.83, 29.72, 26.76, 24.27.

Crude isothiocyanate (40) (0.60 g, 3.49 mmol) was dissolved

in diethyl ether (30 mL), and HCl gas was bubbled through

the solution. The white precipitate (24) so formed was filtered,

washed with ether (3 × 30 mL), and dried under vacuum (0.53

1

yield). H NMR (D₂O): δ 3.66 (t, J ) 6.1 Hz, 2H), 3.19 (t, J )

6.6 Hz, 2H), 2.89 (s, 6H), 1.90-1.73 (m, 4H). ¹³C NMR (D₂O):

δ 57.24, 44.53, 43.00, 26.19, 21.75. Anal. (C₇H₁₄N₂SCl) C, H,

N.

4-(N,N-Dieth yla m in o)bu tyl isoth iocya n a te Mon oh y-

d r och lor id e Sa lt (23). Following a procedure similar to

Kupchan’s,^15,164-chlorobutyronitrile (28) (10.0 g, 97.1 mmol),

diethylamine (27) (60.1 mL, 582.5 mmol), and sodium iodide

(1.46 g, 9.71 mmol) in H₂O (50 mL) were stirred together for

5 d at 25 °C. Vacuum distillation (98-103 °C/15 mmHg)

yielded a clear liquid, 4-diethylaminobutyronitrile (31) (4.54

g, 32.4 mmol, 33% yield). This compound matched the previ-

ously reported boiling point.^15,40IR (NaCl): 2969, 2247, 1200,

1

g, 2.55 mmol, 73% yield, 16% overall yield); mp 93-94 °C. H

1

1073 cm^-1. H NMR (CDCl₃): δ 2.50 (q, J ) 7.2 Hz, 4H), 2.42

NMR (D₂O): δ 3.61 (t, J ) 6.4 Hz, 2H), 3.15 (t, J ) 8.1 Hz,

2H), 2.87 (s, 6H), 1.82-1.71 (m, 4H), 1.53-1.46 (m, 2H). ¹³C

NMR (D₂O): δ 57.89, 44.76, 42.93, 28.62, 23.60, 23.16. Anal.

(C₈H₁₇N₂SCl) C, H, N.

(t, J ) 7.1 Hz, 4H), 1.81-1.72 (m, 2H), 1.01 (t, J ) 7.2 Hz,

6H). ¹³C NMR (CDCl₃): δ 119.93, 50.94, 46.77, 23.52, 14.72,

11.75. HRMS ESI+ (m/z) calcd. for (M + H)⁺C₈H₁₇N₂,

141.1392; found 141.1390. Compound 31 (0.5 g, 3.57 mmol)

and the Rh-Al₂O₃catalyst (0.11 g) were dissolved in MeOH

(20 mL). Hydrogenation was carried out in a Parr shaker

under H₂at 40 psi of pressure for 2.5 d. When hydrogen uptake

was complete, the solution was filtered through Celite to

remove the catalyst. The filtrate was concentrated by rotary

evaporation. The crude product contained a mixture of primary

(N,N-diethylbutane-1,4-diamine, 35) and secondary amines,

1:2, respectively, which were not separated. IR (NaCl): 3292,

5-(N,N-Dieth yla m in o)p en tyl Isoth iocya n a te Mon oh y-

d r och lor id e Sa lt (25). Following a procedure similar to

Kupchan’s,^15,165-chlorovaleronitrile (29) (10.0 g, 85.5 mmol),

diethylamine (27) (53.0 mL, 512.8 mmol), and sodium iodide

(1.28 g, 8.55 mmol) in H₂O (50 mL) were stirred together for

5 d at 25 °C. Vacuum distillation (110-120 °C/15 mmHg)

yielded a clear liquid, 5-diethylamino valeronitrile (33) (7.15

g, 46.4 mmol, 54% yield). This compound matched the previ-

ously reported boiling point.^41,42IR (NaCl): 2968, 2247, 1204,

1

1466, 1306, 1041 cm^-1. H NMR (CDCl₃) (peaks for 35 only):

1

1071 cm^-1. H NMR (CDCl₃): δ 2.50 (q, J ) 7.2 Hz, 4H), 2.44

δ 2.62 (t, J ) 7.2 Hz, 2H), 2.52 (q, J ) 7.2 Hz, 4H), 2.42 (t, J

(t, J ) 7.3 Hz, 2H), 2.39 (t, J ) 7.0 Hz, 2H), 1.74-1.54 (m,

4H), 1.01 (t, J ) 7.2 Hz, 6H). ¹³C NMR (CDCl₃): δ 119.68,

51.72, 46.71, 26.08, 23.44, 16.99, 11.62. Anal. (C₉H₁₈N₂) C, H,

N. Compound 33 (1.6 g, 10.4 mmol) and the Rh-Al₂O₃catalyst

(0.327 g) were dissolved in MeOH (40 mL). Hydrogenation was

carried out in a Parr shaker under H₂at 40 psi of pressure for

2.5 d. When hydrogen uptake was complete, the solution was

filtered through Celite to remove the catalyst. The filtrate was

concentrated by rotary evaporation. The crude product con-

tained a mixture of primary (N,N-diethylpentane-1,5-diamine,

37) and secondary amines, 1:2, respectively, which were not

) 7.2 Hz, 2H), 1.49-1.42 (m, 4H), 1.02 (t, J ) 7.2 Hz, 6H). ¹³

C

NMR (CDCl₃): δ 52.56, 46.58, 41.96, 24.64, 24.19, 11.41. The

mixture of primary (35) and secondary amines (0.51 g, 3.54

mmol/molecular weight of 35) was dissolved in 15 mL of

acetone. After following the procedure used for compounds 7,

11-21, the resulting crude product was a yellow liquid, 4-(N,N-

diethylamino)butyl isothiocyanate (39). IR (NaCl): 2938, 2179,

2100, 1270, 1042 cm^-^{1 1}H NMR (CDCl₃): δ 3.55 (t, J ) 6.5

.

Hz, 2H), 2.51 (q, J ) 7.1 Hz, 4H), 2.44 (t, J ) 7.2 Hz, 2H),

1.77-1.68 (m, 2H), 1.61-1.51 (m, 2H), 1.01 (t, J ) 7.1 Hz, 6H).

¹³C NMR (CDCl₃): δ 129.89, 51.77, 46.65, 44.92, 28.00, 24.00,

11.55. Crude isothiocyanate (39) (0.10 g, 0.54 mmol) was

dissolved in diethyl ether (30 mL) and HCl gas was bubbled

through the solution. The white precipitate (23) so formed was

filtered, washed with ether (3 × 30 mL), and dried under

vacuum (0.08 g, 0.36 mmol, 67% yield, 3% overall yield); mp

1

separated. IR (NaCl): 3291, 1464, 1308, 1040 cm^-1. H NMR

(CDCl₃) (peaks for 37 only): δ 2.69 (t, J ) 7.8 Hz, 2H), 2.54

(q, J ) 7.0 Hz, 4H), 2.42 (t, J ) 7.8 Hz, 2H), 1.54-1.42 (m,

4H), 1.35-1.27 (m, 2H), 1.03 (t, J ) 7.1 Hz, 6H). ¹³C NMR

(CDCl₃): δ 52.71, 46.69, 41.97, 26.72, 25.34, 24.83, 11.45. The

mixture of primary (37) and secondary amines (1.6 g, 10.1

mmol/molecular weight of 37) was dissolved in 5 mL acetone.

After following the procedure used for compounds 7, 19-21,

the resulting crude product was a yellow liquid, 5-(N,N-

diethylamino)pentyl isothiocyanate (41). IR (NaCl): 2941,

1

71-73 °C. H NMR (D₂O): δ 3.66 (t, J ) 5.9 Hz, 2H), 3.24 (q,

J ) 7.1 Hz, 4H), 3.18 (t, J ) 7.2 Hz, 2H), 1.90-1.72 (m, 4H),

1.29 (t, J ) 7.2 Hz, 6H). ¹³C NMR (D₂O): δ 51.47, 48.00, 44.86,

26.66, 21.44, 8.88. Anal. (C₉H₁₉N₂SCl) C, H, N.

1

2175, 2102, 1270, 1040 cm^-1. H NMR (CDCl₃): δ 3.52 (t, J )

5-(N,N-Dim eth yla m in o)p en tyl Isoth iocya n a te Mon o-

h yd r och lor id e Sa lt (24). Following a procedure similar to

Kupchan’s,^15,165-chlorovaleronitrile (29) (1.0 g, 8.55 mmol),

dimethylamine (26) (6.4 mL, 51.3 mmol), and sodium iodide

(0.128 g, 0.855 mmol) were stirred together for 5 d at 25 °C.

Vacuum distillation (98-100 °C/15 mmHg) yielded a clear

liquid, 5-dimethylaminovaleronitrile (32) (0.54 g, 4.29 mmol,

50% yield). This compound matched the previously reported

6.6 Hz, 2H), 2.53 (q, J ) 7.1 Hz, 4H), 2.43 (t, J ) 7.8 Hz, 2H),

1.77-1.68 (m, 2H), 1.49-1.42 (m, 4H), 1.03 (t, J ) 7.1 Hz, 6H).

¹³C NMR (CDCl₃): δ 128.97, 52.52, 46.76, 44.94, 29.86, 26.29,

24.58, 11.50. Crude isothiocyanate (41) (0.66 g, 3.30 mmol) was

dissolved in diethyl ether (30 mL) and HCl gas was bubbled

through the solution. The white precipitate (25) so formed was

filtered, washed with ether (3 × 30 mL), and dried under

vacuum (0.41 g, 1.74 mmol, 53% yield, 9% overall yield); mp

1

boiling point.^15,39IR (NaCl): 2944, 2247, 1262, 1040 cm^-1. H

1

NMR (CDCl₃): δ 2.39 (t, J ) 6.8 Hz, 2H), 2.29 (t, J ) 6.8 Hz,

90-92 °C. H NMR (D₂O): δ 3.61 (t, J ) 6.2 Hz, 2H), 3.22 (q,

2H), 2.21 (s, 6H), 1.73-1.59 (m, 4H). ¹³C NMR (CDCl₃): δ

J ) 7.2 Hz, 4H), 3.15 (t, J ) 8.4 Hz, 2H), 1.81-1.69 (m, 4H),

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 951

1.56-1.44 (m, 2H), 1.28 (t, J ) 7.3 Hz, 6H). ¹³C NMR (D₂O):

δ 51.98, 47.82, 44.97, 28.86, 23.60, 23.19, 8.71. Anal. (C₁₀H₂₁N₂-

SCl) C, H, N.

(2%) for 1 or in buffer only for 3 to give a concentration of 1

and 2 mg/mL, respectively. The solution for 1 (268 µL) was

diluted with ammonium acetate buffer (116 µL, 50 mM, pH

7.3), and the solution for 3 (235 µL) was diluted with am-

monium acetate buffer (149 µL, 50 mM, pH 7.3). HbA‚CO

solution (116 µL 27.9%) was added to each isothiocyanate

solution, and then the resulting mixtures were incubated at

37 °C for 3 h.

Modified and unmodified globins were isolated from the

heme according to the method of Marta.⁴³Briefly, hemoglobin

solution (1 mM, 300 µL) was added dropwise, with stirring, to

a 1.5% v/v solution of concentrated HCl in acetone (15 mL)

cooled below 0 °C. This caused the globin to precipitate leaving

the heme in solution. The precipitate was washed three times

by suspending in acetone, centrifuged, and dried under

vacuum.

Trypsin stock solution was prepared by dissolving l-1-

tosylamido-2-phenylethylchloromethyl-ketone-treated (TPCK)

trypsin (1 mg) in ammonium bicarbonate buffer (1 mL, 50 mM,

pH 8.0). Globin (1 mg) was suspended in ammonium bicarbon-

ate buffer (490 µL, 50 mM, pH 8.0) and trypsin stock solution

(10 µL). The resulting mixture was incubated overnight at 37

°C. The tryptic peptide mixture was ultrafiltered (10000 MW

cut off, Amicon, Bedford, MA). The filtrate (20 µL) was injected

into a Finnigan HPLC system equipped with a Vydac C18

reversed-phase column (2.1 × 250 mm, Chrom Tech, Inc. Apple

Valley, MN) with Finnigan LCQ ESMS detection (molecular

weight scan range: 350-2000 m/z). The mobile phase con-

sisted of solution A (water containing 0.1% formic acid) and

solution B (acetonitrile containing 0.1% formic acid): a linear

gradient from 0 to 60% solution B was run over a period of 45

min at a flow rate of 200 µL/min. Peaks were assigned based

on molecular weights of expected fragments (data not shown).

Id en tifica tion of Rea ction Site on â-Su bu n it Usin g Gel

Electr op h or esis. HbA‚CO (1mM) was modified with com-

pound 7, 8, or 9 (3 equiv) for 3 h as described above. ESMS

confirmed at least 95% of the â-subunit had been modified with

less than 5% of R-subunit modification. Electrophoresis was

done as previously described⁴⁴with minor modification. Briefly,

the three modified and unmodified HbA molecules were

applied to a Titan III cellulose acetate plate (Helena Labora-

tories, Beaumont, TX) which was soaked in buffer (0.1 M Tris,

2 mM EDTA, 50 mM boric acid, pH 8.4) just before the

electrophoresis. The electrophoresis was carried out for 15 min

at 300 V in the same buffer above. The resulting plate was

dried by standing for 1 h and stained in Ponceau S (Helena

Laboratories, Beaumont, TX) in 3% aqueous trichloroacetic

acid for 15 min, followed by 3% aqueous acetic acid twice before

rinsing with distilled water. The resulting gels are not shown.

Biology. P r ep a r a tion of Nor m a l a n d Sick le Hem oglo-

bin a n d C_{sa t}Assa y. Preparation of stripped hemoglobins

(HbA from normal adult human donors and HbS from sickle

homozygous patients) and their purification have previously

been described.⁴³Packed red blood cells (26.5 mL) in distilled

water (53.5 mL), which had been washed three times with

isotonic NaCl solution, were lysed in distilled water by first

stirring for 30 min in a cold room. To this solution was added

saturated ammonium sulfate solution (pH 7.00 adjusted with

NaOH, 20 mL) and then stirred for another 30 min in a cold

room. The Hb solution was separated from the precipitate by

centrifugation at 2500g in a Sorvall RC-5B centrifuge for 30

min. Carbon monoxide was bubbled through the Hb solution

for 30 min while on ice. Salt was removed from the Hb by gel

filtration on a Sephadex G-25 column, preequilibrated with

phosphate buffer (5 mM, pH 7.2) at 2-4 °C. Eluates containing

Hb were washed with phosphate buffer (5 mM, pH 7.0, 1 L),

followed by sodium azide solution (0.02%, 1 L). The resulting

solution was concentrated to 15 g/dL by ultrafiltration at 2500

g in a Sorvall RC-5B centrifuge and the concentrated Hb stored

in liquid nitrogen. When ESMS analyses were required,

ammonium acetate buffer (50 mM, pH 7.0) was used in place

of the phosphate buffer.

The saturation concentration (C_sat) of deoxy HbS was

determined in 50 mM phosphate buffer at pH 7.0 following

our previously published methods.²⁵Briefly, CO was removed

by photodissociation under an oxygen atmosphere, then HbS

or inhibitor-modified HbS was deoxygenated by adding 20 µL

of a 0.9 M sodium dithionite solution to 330 µL of 35 g/dL

HbSO₂in SW55 centrifuge tubes on ice, and the resulting

solution then centrifuged for 1 h at 30 °C. After centrifugation,

the supernatant phase was separated and the concentration

determined spectrophotometrically. Aliquots for mass spec-

trometry were also obtained. Further details are in De Croos

et al.²⁵

Ch em ica l Mod ifica t ion of H b w it h Va r iou s Isot h io-

cya n a tes. Modification of HbA‚CO (all stated Hb concentra-

tions are in terms of the R₂â₂tetramer) with isothiocyanates

1-13, 16-18, 20-21 was carried out at pH 7.3 with each

isothiocyanate in a molar ratio of compound:Hb ) 2:1 at a 100

µM final concentration of HbA‚CO. The isothiocyanate was

dissolved in ammonium acetate buffer (50 mM, pH 7.3)

containing dimethyl sulfoxide (2%) to give a concentration of

1 mg/mL. For 3, 7, and 9 only buffer was used without

dimethyl sulfoxide. HbA‚CO solution was added to isothiocy-

anate, and then the resulting mixture was incubated at 37 °C

for 3 h. An aliquot (10 µL) of the reaction mixture was diluted

with acetonitrile/water (1/1 ) v/v, 1 mL) containing formic acid

(0.2%) and analyzed using the Micromass Quattro II mass

spectrometer; cone voltage ) 30 V, source temperature ) 70

°C, flow rate ) 10 µL/min, capillary potential ) 3.5 kV. The

mass spectrometer was scanned from m/z 600-1300 every 10

s for 3 min. For the study, the mass scale was calibrated using

horse heart myoglobin (Mr ) 16951.5). From the multiply

charged peaks envelopes in the ESMS, the molecular weights

and peak intensities of the R-, â-, and modified subunits were

obtained using maximum entropy deconvolution software

(Masslynx, Micromass Boston MA) based on peak width at

half-height of base peak and m/z of each of the multiply

charged peaks. Representative processed spectrums are shown

in Figure 2. The reaction yields were obtained from peak

height normalized by (height of modified)/(height of modified

+ height of unmodified).

Id en tifica tion of Rea ction Site on Nor m a l Hem oglobin

â-Su bu n it by Titr a tion of Cysâ93 Su lfh yd r yl Gr ou p w ith

p-Ch lor om er cu r iben zoic Acid . The free sulfhydryl groups

in HbA modified with compounds 5, 7, 8, and 9 were titrated

with p-chloromercuribenzoic acid (PMB) in ammonium acetate

buffer (50 mM, pH 7.3), according to the method of Boyer.²⁰

Briefly, PMB (1 mg) was dissolved in aqueous NaOH (0.01 N,

1 mL), and a 100 µL aliquot was diluted to 1 mL with

ammonium acetate buffer (50 mM, pH 7.0). The final concen-

tration of this stock solution was determined from UV absor-

bance at 232 nm using the extinction coefficient ꢀ ) 1.69 ×

10⁴M^-1cm^{-1 20}PMB stock solution (100 µL) was diluted with

.

ammonium acetate buffer (50 mM, pH 7.0, 900 µL). The UV

absorbance of this solution relative to the ammonium acetate

buffer (50 mM, pH 7.0) was measured at 250 nm.

A standard curve for the increase in UV absorbance from

the reaction between PMB and the free sulfhydryl group in

unmodified HbA was prepared. PMB stock solution (100 µL)

was added to 0.125, 0.25, 0.375, 0.5, and 0.625 equiv of

unmodified HbA‚CO, and the final volumes were adjusted to

1 mL with ammonium acetate buffer (50 mM, pH 7.0). The

references, which were used for the double beam UV spectro-

photometer, were prepared the same as above but with

ammonium acetate buffer (50 mM, pH 7.0) used in the place

of the PMB. The UV absorbances of the solutions of PMB and

unmodified HbA‚CO mixtures relative to the reference were

Id en tifica tion of Isoth iocya n a te Rea ction Site on

Nor m a l Hem oglobin r-Su bu n it Usin g Tr yp sin Digestion .

Isothiocyanates 1 and 3 were used as representative aryl

isothiocyanates in the identification of the reaction site on the

R-subunit. Modification of HbA‚CO with isothiocyanates 1 and

3 was carried out at pH 7.3 with each isothiocyanate in a molar

ratio of compound:Hb ) 3:1 at 1 mM final concentration of

HbA‚CO. The isothiocyanate was dissolved in ammonium

acetate buffer (50 mM, pH 7.3) containing dimethyl sulfoxide

952 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6

Park et al.

measured at 250 nm. An increase in UV absorbance due to

mercaptide formation from the reaction between PMB and the

free sulfhydryl group in unmodified HbA relative to PMB was

obtained by subtraction of the UV absorbance of diluted PMB

stock solution (measured above) from that of the solutions of

PMB and HbA‚CO mixtures. The increase in UV absorbance

of samples was plotted against the equivalents of free sulfhy-

dryl in the unmodified HbA‚CO. As HbA‚CO has two â-sub-

units, the 0.125, 0.25, 0.375, 0.5, and 0.625 equiv of unmodified

HbA‚CO have 0.25, 0.5, 0.75, 1.0, and 1.5 equiv of free

sulfhydryl, respectively, for reaction with PMB.

In a separate experiment, HbA‚CO was modified with

isothiocyanates 5, 7, 8, and 9, as previously described. The

modified HbA‚CO (0.5 equiv or 1 equiv of sulfhydryl) was

added to PMB stock solution (100 µL), and the final volume

was adjusted to 1 mL with ammonium acetate buffer (50 mM,

pH 7.0). The UV absorbance of each modified HbA solution

was measured at 250 nm. The increase in UV absorbance due

to mercaptide formation was obtained from subtraction of the

UV absorbance of diluted PMB stock solution from that of the

solutions of PMB and modified HbA‚CO mixtures.

set was collected on a single crystal to 1.9 Å resolution. The

data were processed with DENZO (v1.9.6),⁴⁷assuming a

mosaic spread of 0.3°, which refined to 0.225°. The data were

scaled with SCALEPACK⁴⁷and showed a maximum decay of

approximately 25%. Covalently modified HbA crystallized in

space group P2₁2₁2₁with unit cell dimension of a ) 63.65 Å,

b ) 83.76 Å, c ) 112.23 Å, and with one (Râ)₂tetramer in the

asymmetric unit.

The structure was solved by molecular replacement using

the program EPMR.⁴⁸The complete, deoxyhemoglobin tet-

rameric structure 1A0Z⁴⁹was used as the initial model for the

molecular replacement search. Model refinement was carried

out using the program CNS (v1.0) using a bulk-solvent

correction.⁵⁰A random selection of 5% of the total number of

reflections were excluded from the refinement and used to

calculate the free R-factor values. Manual model building was

carried out with O⁵¹and with CCP4.⁵²

Molecu la r Mod elin g. Autodock calculations were per-

formed using AutoDock 3.0.²¹The deoxy-HbS crystal structure,

2HbS, was used. All water molecules were removed from the

crystal structure, as were chains E-H, since the 2HbS crystal

is a dimer, and only one HbS molecule (chains A-D) is needed

for modeling. QUANTA, a program known to handle heme

groups properly, was used to add polar hydrogens and charges

to the HbS molecule. This molecule was then used for generat-

ing the atomic energy grids for autodock. Due to the size of

the HbS molecule, and the desire for fine grids, the HbS

molecule was divided into a 3 × 3 × 3 cube, resulting in a

total of 27 grids. Each grid was approximately 34 × 30 × 30

Å, with an overlap of ∼10 Å, to ensure that any region of the

molecule would not be overlooked. The number of points per

grid was 90 × 80 × 80, with 0.375 Å between points. The

isothiocyanate ligands were created and minimized within

Sybyl 6.7 (Tripos, Inc., St. Louis, MO) and Gasteiger-Marsili

charges assigned to all atoms. These isothiocyanate ligand

initial structures were then docked using all 27 grids, and not

just those grids containing the â93 or R-N-terminus. Reasons

for this were 2-fold: (1) since it is known that HbS has a

number of possible “binding” sites, which are in part dependent

upon the type of ligand, we wanted to have a complete picture

as to where these ligands could theoretically interact, and (2),

we wanted to verify whether the â93 and/or R-N-terminus were

indeed the lowest energy binding site(s) for these ligands over

the whole HbS molecule. Flexible ligand dockings were

performed using the defaults of the Genetic Algorithm plus

pseudo-Solis & Wets, with the exceptions of using a population

size of 50, and 25000 energy evaluations, along with 300

iterations of S&W local search. A total of 30 GA runs were

done with a cluster analysis.

The number of equivalents of free sulfhydryl in HbA‚CO

modified with isothiocyanates was obtained by comparing the

increase in UV absorbance of modified HbA‚CO with that in

the standard curve (data not shown).

Cell P er m ea bility Mea su r em en ts for Com p ou n d s 7

a n d 24. Human packed red blood cells obtained from Blood

Bank (UIC hospital) were washed three times with 4 volumes

of phosphate buffered saline (pH 8.0). Washed cells were

incubated with equal volumes of compound at a 4:1 molar ratio

of each isothiocyanate to Hb (tetramer) in phosphate-buffered

saline (50 mM phosphate, pH 7.4) on a tube rocker for 6 h at

room temprature. After incubation, the cells were washed

three times with phosphate-buffered saline to remove the

excess label. To release hemoglobin from the erythrocytes, cells

were mixed with 2 volumes of deionized distilled water and

stirred gently for 30 min in the cold. The resulting hemoglobin

solution, which also contained cell membranes, was mixed and

stirred for 30 min with 1/4 volume of saturated ammonium

sulfate solution (pH 7.0) in the cold. The mixture was

centrifuged at 12000 rpm for 30 min (at 4 °C), and the

supernatant was collected and concentrated to 5% using an

Amicon Centriflow Membrane Cone (MWCO 2500) (4 °C, 3500

rpm). The concentrated hemoglobin solution was desalted

using Micro DispoDialyzer (Spectrum Lab) (MWCO 25 kD) and

deionized distilled water(8 °C, 24 h) and concentrated to 1.8%.

An aliquot of modified hemoglobin was submitted for mass

spectrometry analysis as described above.

X-r a y Cr ysta llogr a p h y. Cr ysta lliza tion . Native, carbon-

monoxyhemoglobin was first isolated from packed human RBC

according to the procedure described by Perutz.⁴⁵The protein

was further purified using a DEAE-Sephadex column by

gradient elution from 50 mM Tris buffer (pH 8.3 to pH 7.3).⁴⁶

The purified protein was then covalently modified with 2-(N,N-

dimethylamino)ethyl isothiocyanate as described above and

subsequently desalted on a Sephadex G25 column to remove

free reagents and products. The modification and purity of the

protein were checked by electrophoresis and ESMS. Prior to

the crystallization, the sample was first oxygenated to remove

bound CO and then deoxygenated using sodium dithionite in

an anaerobic environment. Crystals were grown by the batch

crystallization method at room temperature from a solution

containing 0.75% protein, 19% PEGMME 2000, 5 mM sodium

dithionite, 100 mM sodium citrate, 200 mM ammonium acetate

at pH 6.6. Crystals suitable for X-ray data collection grew

within a week to crystal sizes of approximately 0.5 × 0.4 ×

0.2 mm. A single crystal was mounted in a 1 mm quartz glass

capillary (Hampton Research, Laguna Niguel, CA), which was

then sealed with epoxy. All manipulations were performed in

a glovebag filled with nitrogen.

To create the double-strand covalent-Hb structure in Figure

6, we began with the crystal structure containing the covalent

ligand. The covalent-Hb dimer was first created by the

symmetry translation of the monomer. Then, using mole-

man2,⁵²the dimer was translated in the ‘a’ direction by -2,

-1, 1, and 2 units. The final structure shown in Figure 6 was

created using Sybyl 6.8 (Tripos, Inc., St. Louis, MO). Sybyl

6.8 was also used to generate the electric fields of this double-

strand structure.

Ack n ow led gm en t. This work was supported in part

by NIH grant 5R01 HL57604. D.C.M. was supported by

NIH NRSA fellowship 5F32 HL10231. B.L.H. was

supported in part by American Heart Association Mid-

west Affiliate fellowship 0020596Z. We thank Loyda

Vida for access to instrumentation and assistance with

oxygen affinity measurements. We thank the Mass

Spectrometry Laboratory of the Research Resources

Center, University of Illinois at Chicago, for use of the

ThermoFinnigan LCQ and Micromass Quattro II mass

spectrometers used in this investigation. The LCQ mass

spectrometer was purchased in part with a grant from

the National Science Foundation (BIR-9111391); the

Da ta Collection a n d Str u ctu r e Deter m in a tion . X-ray

diffraction data were collected at room temperature (298 K)

on a R-AXIS-IIC imaging plate area detector equipped with

a Rigaku RU-200 rotating anode with a Cu target (KR, λ )

1.5418 Å) X-ray operating at 50 kV and 100 mA. A full data

Antisickling Agents

J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 6 953

(26) Moffat, J . K.; Simon, S. R.; Konigsberg, W. H. Structure and

functional properties of chemically modified horse hemoglobin.

3. Functional consequences of structural alterations and their

implications for the molecular basis of cooperativity. J . Mol. Biol.

1971, 58, 89-101.

purchase of the Quattro II was funded by the National

Center for Research Resources (S10 RR10485).

Note Ad d ed a fter ASAP P ostin g

(27) Bonaventura, C.; Ferruzzi, G.; Tesh, S.; Stevens, R. D. Effects

of S-nitrosation on oxygen binding by normal and sickle cell

hemoglobin. J . Biol. Chem. 1999, 274, 24742-24748.

(28) Garel, M.-C.; Donemget, C.; Caburi-Martin, J .; Prehu, C.;

Galacteros, F. et al. Covalent binding of glutathione to hemo-

globin. J . Biol. Chem. 1986, 261, 14704-14709.

(29) Chan, N. L.; Rogers, P. H.; Arnone, A. Crystal structure of the

S-nitroso form of liganded human hemoglobin. Biochemistry

1998, 37, 16459-16464.

(30) Kleywegt, G. J . Use of noncrystallographic symmetry in protein

structure refinement. Acta Cryst. 1996, D52, 842-857.

(31) Garel, M. C.; Caburi-Martin, J .; Domenget, C.; Kister, J .;

Craescu, C. T. et al. Changes of polymerization and conformation

of hemoglobin S induced by thiol reagents. Biochim. Biophys.

Acta 1990, 1041, 133-140.

The Supporting Information paragraph in the version

of the manuscript posted February 7, 2003, has been

removed. The J ournal of Medicinal Chemistry does not

publish elemental analyses data. The corrected version

of the manuscript was posted February 25, 2003.

Refer en ces

(1) Manning, J . M. Covalent inhibitors of the gelation of sickle cell

hemoglobin and their effects on function. Advances in Enzymol-

ogy and Related Areas of molecular Biology; J ohn Wiley &

Sons: New York, 1991; pp 55-91.

(2) Harrington, D. J .; Adachi, K.; Royer, W. E., J r. The high-

resolution crystal structure of deoxyhemoglobin S. J . Mol. Biol.

1997, 272, 398-407.

(3) Bunn, H. F.; Forget, B. G. Hemoglobin: molecular, genetic and

clinical aspects; W. B. Saunders: Philadelphia, 1986.

(4) Eaton, W. A.; Hofrichter, J . Hemoglobin S gelation and sickle

cell disease. Blood 1987, 70, 1245-1266.

(5) Charache, S. Mechanism of action of hydroxyurea in the

management of sickle cell anemia in adults. Semin. Hematol.

1997, 34, 15-21.

(6) Beddell, C. R.; Goodford, P. J .; Kneen, G.; White, R. D.;

Wilkinson, S. et al. Substituted benzaldehydes designed to

increase the oxygen affinity of human haemoglobin and inhibit

the sickling of sickle erythrocytes. Br. J . Pharmacol. 1984, 82,

397-407.

(7) Sunshine, H. R.; Hofrichter, J .; Eaton, W. A. Requirement for

therapeutic inhibition of sickle haemoglobin gelation. Nature

1978, 275, 238-240.

(8) Guyton, A. C. Medical Physiology; W. B. Saunders: Philadelphia,

1986.

(9) Vain, J . R. Inhibition of prostaglandin synthesis as a mechanism

of action for aspirin-like drugs. Nature-New Biol. 1971, 231,

231-235.

(10) Nickerson, M.; Hollenberg, N. K. Blockade of alpha-adrenergic

receptors; Academic Press: New York, 1967; 243-305.

(11) Garel, M. C.; Domenget, C.; Galacteros, F.; Martin-Caburi, J .;

Beuzard, Y. Inhibition of erythrocyte sickling by thiol reagents.

Mol. Pharm. 1984, 26, 559-565.

(12) Seligman, R. B.; Bost, R. W.; McKee, R. L. Derivatives of

p-aminosalicylic acid. J . Am. Chem. Soc. 1953, 75, 6334-6335.

(13) Kim, S.; Yi, K. Y. Di-2-pyridyl thionocarbonate. A new reagent

for the preparation of isothiocyanates and carbodiimides. Tet-

rahedron Lett. 1985, 26, 1661-1664.

(14) McElhinney, R. S. Derivatives of thiocarbamic acid. Part I.

Preparation of 4-substituted thiosemicarbazides. J . Chem. Soc.

C 1966, 950-955.

(15) Kupchan, S. M.; Bondesson, G.; Davies, A. P. Tumor inhibitors.

70. Structure-cytotoxicity relationships among N-acyltriamines

related to solapalmitine. J . Med. Chem. 1972, 15, 65-68.

(16) Price, C. C.; Kabas, G.; Nakata, I. Some amino and ammonio

nitrogen mustard analogues. J . Med. Chem. 1965, 8, 650-655.

(17) Park, S.; J ohnson, M. E.; Venton, D. L. Regioselective chemical

reaction on a protein surface. Presented ath the 217th National

Meeting of the American Chemical Society; Anaheim, CA, 1999.

(18) Desiderio, D., M. Mass spectrometry; Plenum Press: New York,

1994.

(19) Chiancone, E.; Currell, D. L.; Vecchini, P.; Antonini, E.; Wyman,

J . Kinetics of the reaction of the “Masked” and “Free” sulfhydryl

groups of human hemoglobin with p-Mercuribenzoate. J . Biol.

Chem. 1970, 245, 4105-4111.

(20) Boyer, P. D. Spectrophotometric study of the reaction of protein

sulfhydryl groups with organic mercurials. J . Am. Chem. Soc

1954, 76, 4331-4337.

(32) Perutz, M. F.; Muirhead, H.; Mazzarella, L.; Crowther, R. A.;

Greer, J . et al. Identification of residues responsible for the alka-

line Bohr effect in haemoglobin. Nature 1969, 222, 1240-1243.

(33) Domenget, C.; Garel, M. C.; Rhoda, M. D.; Caburi-Martin, J .;

Galacteros, F. et al. Kinetics of polymerization of hemoglobin S

modified by thiol reagents and by oxidation. Biochim. Biophys.

Acta 1985, 830, 71-79.

(34) Eaton, W. A.; Hofrichter, J . Sickle cell hemoglobin polymeriza-

tion. Adv. Protein Chem. 1990, 40, 63-279.

(35) J ochims, J . C.; Seeliger, A. A new synthesis of aliphatic isothio-

cyanates. Angew. Chem., Int. Ed. Engl. 1967, 6, 174-175.

(36) Yamasaki, T.; Kawaminami, E.; Uchimura, F.; Okamoto, Y.;

Okawara, T. et al. Preparation of novel tricyclic ring systems

containing the pyridazinone ring. J . Heterocycl. Chem. 1992, 29,

825-829.

(37) Park, S.; Wanna, L.; J ohnson, M. E.; Venton, D. L. A mass

spectrometry screening method for antiaggregatory activity of

proteins covalently modified by combinatorial library mem-

bers: application to sickle hemoglobin. J . Comb. Chem. 2000,

2, 314-317.

(38) Siatra-Papastaikoudi, T.; Tsotinis, A.; Raptopoulou, C. P.; Sam-

bani, C.; Thomou, H. Synthesis of new alkylaminoalkyl thio-

semicarbazones of 3-acetylindole and their effect on DNA

synthesis and cell proliferation. Eur. J . Med. Chem. 1995, 30,

107-114.

(39) Short, J . H.; Biermacher, U.; Dunnigan, D. A.; Leth, T. D. Sym-

pathetic nervous system blocking agents. Derivatives of guani-

dine and related compounds. J . Med. Chem. 1963, 6, 275-283.

(40) Huber, W.; Clinton, R. O.; Boehme, W.; J ackman, M. Some dialk-

ylaminoalkylamines. J . Am. Chem. Soc. 1945, 67, 1618-1619.

(41) Stevenson, G. W.; Williamson, D. Base strengths of cyanoamines.

J . Am. Chem. Soc. 1958, 80, 5943-5947.

(42) Kurtz, P. Formation of nitriles. I. J ustus Liebigs Ann. Chem.

1951, 572, 23-82.

(43) Marta, M.; Patamia, M.; Lupi, A.; Antenucci, M.; Io, M. D. Bovine

Hemoglobin Cross-Linked through the beta Chains Functional

and structural aspects. J . Biol. Chem. 1996, 271, 7473-7478.

(44) D’Avino, R.; Di Prisco, G. Hemoglobin from the Antarctic fish

Notothenia coriiceps neglecta. 1. Purification and characterisa-

tion. Eur. J . Biochem. 1989, 179, 699-705.

(45) Bolton, W.; Cox, J . M.; Perutz, M. F. Structure and function of

haemoglobin. IV. A three-dimensional Fourier synthesis of horse

deoxyhaemoglobin at 5.5 A resolution. J . Mol. Biol. 1968, 33,

283-297.

(46) Adachi, K.; Asakura, T. Demonstration of a delay time during

aggregation of diluted solutions of deoxyhemoglobin

S and

hemoglobin CHarlem in concentrated phosphate buffer. J . Biol.

Chem. 1978, 253, 6641-6643.

(47) Otwinowski, Z.; Minor, W. Processing of X-ray Diffraction Data

Collected in Oscillation Mode. Methods in Enzymology; Academic

Press: New York, 1997; Part A, pp 307-326.

(48) Kissinger, C. R.; Gehlhaar, D. K.; Fogel, D. B. Rapid automated

molecular replacement by evolutionary search. Acta Crystallogr.

D Biol. Crystallogr. 1999, 55, 484-491.

(21) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart,

W. E. et al. Automated docking using a Lamarckian genetic

algorithm and an empirical binding free energy function. J .

Comput. Chem. 1998, 19, 1639-1662.

(22) Drobnica, L.; Kristian, P.; Augustin, J . The chemistry of the

isothiocyanato group. The Chemistry of Cyanates and Their Thio

Derivatives; Wiley: Chichester, England, 1977; pp 1003-1221.

(23) van Rossum, J . M.; Ariens, E. J . Receptor-reserve and threshold-

phenomena. II. Theories on drug-action and a quantitative

approach to spare receptors and threshold values. Arch. Int.

Pharmacodyn. Ther. 1962, 136, 385-413.

(24) Stephenson, R. P. A modification of receptor theory. Br. J .

Pharmacol. Chemother. 1956, 11, 379-393.

(25) De Croos, P. Z.; Sangdee, P.; Stockwell, B. L.; Kar, L.; Thompson,

E. B. et al. Hemoglobin S antigelation agents based on 5-bro-

motryptophan with potential for sickle cell anemia. J . Med.

Chem. 1990, 33, 3138-3142.

(49) Kavanaugh, J . S.; Weydert, J . A.; Rogers, P. H.; Arnone, A. High-

resolution crystal structures of human hemoglobin with muta-

tions at tryptophan 37beta: structural basis for a high-affinity

T-state. Biochemistry 1998, 37, 4358-4373.

(50) Rice, L. M.; Earnest, T. N.; Brunger, A. T. Single-wavelength

anomalous diffraction phasing revisited. Acta Crystallogr. D Biol.

Crystallogr. 2000, 56, 1413-1420.

(51) J ones, T. A.; Zou, J . Y.; Cowan, S. W.; Kjeldgaard, M. Improved

methods for building protein models in electron density maps

and the location of errors in these models. Acta Crystallogr. 1991,

A47, 110-119.

(52) Bailey, S. The CCP4 suite: programs for protein crystallography.

Acta Crystallogr., Sect. D: Biol.Crystallogr. 1994, D50, 760-763.

J M020361K

Article Doi

Regioselective covalent modification of hemoglobin in search of antisickling agents

DOI: 10.1021/jm020361k

Source and publish data:

Authors:

Article abstract of DOI:10.1021/jm020361k

Full text of DOI:10.1021/jm020361k

Products guided by the article

R&D Labs maybe for 101098-33-3

Relevant to this article

Hot Product