The development of N-α-(2-carboxyl)benzoyl- N 5-(2-fluoro- 1-iminoethyl)- l-ornithine amide (o-F-amidine) and N-α-(2-carboxyl) benzoyl-N5-(2-chloro-1-iminoethyl)-l-ornithine amide (o -Cl-amidine) as second generation protein arginine deiminase (PAD) inhibitors

ARTICLE

pubs.acs.org/jmc

The Development of N-α-(2-Carboxyl)benzoyl-N⁵-

(2-fluoro-1-iminoethyl)-^L-ornithine Amide (o-F-amidine) and

N-α-(2-Carboxyl)benzoyl-N⁵-(2-chloro-1-iminoethyl)-L-ornithine

Amide (o-Cl-amidine) As Second Generation Protein Arginine

Deiminase (PAD) Inhibitors^†

Corey P. Causey,^§,XJustin E. Jones,^‡,XJessica L. Slack,^‡Daisuke Kamei,^{^}Larry E. Jones, Jr.,^§

Venkataraman Subramanian,^‡Bryan Knuckley,^‡Pedram Ebrahimi,^||Alexander A. Chumanevich,^||Yuan Luo,^§

Hiroshi Hashimoto,^{^}Mamoru Sato,^{^}Lorne J. Hofseth,^||and Paul R. Thompson*^,‡

^‡Department of Chemistry, The Scripps Research Institute, 130 Scripps Way, Jupiter, Florida 33458, United States

^§Department of Chemistry & Biochemistry, University of South Carolina, 631 Sumter Street, Columbia, South Carolina 29208,

United States

Department of Pharmaceutical and Biomedical Sciences, South Carolina College of Pharmacy, University of South Carolina,

770 Sumter Street, Columbia, South Carolina 29208, United States

^{^}Graduate School of Integrated Science, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan

S Supporting Information

b

ABSTRACT: Protein arginine deiminase (PAD) activity is upregulated in a number of

human diseases, including rheumatoid arthritis, ulcerative colitis, and cancer. These enzymes,

there are ﬁve in humans (PADs 1ꢀ4 and 6), regulate gene transcription, cellular dif-

ferentiation, and the innate immune response. Building on our successful generation of F-

and Cl-amidine, which irreversibly inhibit all of the PADs, a structureꢀactivity relationship

was performed to develop second generation compounds with improved potency and

selectivity. Incorporation of a carboxylate ortho to the backbone amide resulted in the

identiﬁcation of N-α-(2-carboxyl)benzoyl-N⁵-(2-ﬂuoro-1-iminoethyl)-L-ornithine amide

(o-F-amidine) and N-α-(2-carboxyl)benzoyl-N⁵-(2-chloro-1-iminoethyl)-L-ornithine amide

(o-Cl-amidine), as PAD inactivators with improved potency (up to 65-fold) and selectivity

(up to 25-fold). Relative to F- and Cl-amidine, the compounds also show enhanced potency

in cellulo. As such, these compounds will be versatile chemical probes of PAD function.

’ INTRODUCTION

in all tissue and cell types, the expression patterns of the remaining

PADs are much more restricted, with PADs 1, 3, 4, and 6 being

most predominantly expressed in the skin, hair follicles, immune

cells, and oocytes, respectively.³

It has long been recognized that posttranslational modiﬁca-

tions (PTMs) play critical roles in maintaining the homeostasis

of biological systems. While many PTMs, and their putative roles

in disease etiology, are well documented (e.g., protein phos-

phorylation), most others are less well understood. One example

of the latter is protein deimination (Figure 1A).¹This PTM is

generated by the hydrolytic removal of ammonia from the gua-

nidinium of a substrate arginine residue to generate a ureido

group.^1,2Because this modiﬁcation results in the formation of

citrulline, this process is also termed citrullination. The enzymes

responsible for generating this PTM are known as the protein

arginine deiminases (PADs), and in humans and other mammals,

there are ﬁve isozymes (i.e., PADs 1, 2, 3, 4, and 6) that possess a

high degree of interisozyme sequence homology (∼50%).^1,3Given

that these isozymes possess similar, but not identical, substrate

speciﬁcities,^4,5it is perhaps not surprising that the major distinguish-

ing feature of individual family members is their tissue distribution

patterns.³For example, whereas PAD2 is near universally expressed

Although one or more of these enzymes are expressed in

virtually all cell types, our understanding of their physiological

functions, especially with regard to PADs, 1, 2, 3, and 6, is limited.

However, it is known that these enzymes play incompletely deﬁned

roles in a diverse number of processes, including the corniﬁcation

of the skin (PADs 1, 2, and 3), hair follicle formation (PAD3),

nerve myelination (PAD2), and fertility (PAD6).^3,6ꢀ10Given the

early links suggesting that dysregulated PAD4 activity contri-

butes to the onset and progression of rheumatoid arthritis (RA;

see below and ref 1 for a review), our understanding of the

physiological roles of PAD4, while still limited, is more advanced.

For example, a fraction of PAD4 is localized to the nucleus where

Received: July 8, 2011

Published: September 01, 2011

r

2011 American Chemical Society

6919

dx.doi.org/10.1021/jm2008985 J. Med. Chem. 2011, 54, 6919–6935

|

Journal of Medicinal Chemistry

ARTICLE

patients with RA produce autoantiboides targeting both citrulli-

nated proteins and PAD4 itself, and these autoantibodies ﬁrst

appear in the preclinical phase of the disease.^30ꢀ33Additionally,

mutations in the gene encoding PAD4 confer an increased risk of

developing RA, at least in Asian populations.^34,35Finally, PAD4

and citrullinated proteins are present in the inﬂamed joints of RA

patients.^28,36However, the fact that PAD2 is also overexpressed

in these tissues²⁸complicates the seemingly simple picture that

dysregulated PAD4 activity is the only PAD that contributes to

the etiology of RA. Thus, it is unclear whether the inhibition of

single or multiple PADs is required to treat these diseases. One

way to address this issue is the development of isozyme-speciﬁc

PAD inhibitors that can be used to discern the relative contribu-

tions of individual PADs to human disease. To begin to develop

such compounds and also identify compounds with enhanced

potency and bioavailability, we performed StructureꢀActivity

Relationships (SARs) on our previously described PAD inhibi-

tors F- and Cl-amidine (Figure 1B). Herein we describe the

design, synthesis, and testing of a series of new PAD inhibitors

with enhanced potency, selectivity, and bioavailability. In parti-

cular, we describe the development of o-F-amidine and o-Cl-

amidine as two of the most potent and partially selective PAD

inhibitors described to date.

Figure 1. Protein arginine deiminase reaction and inhibitors. (A) The

guanidinium group of an arginine residue is converted into a ureido

group to form citrulline. (B) Structures of Cl- and F-amidine.

it has been shown to modify the unstructured tails of histones

H3 and H4, and this modiﬁcation has been associated with the

decreased transcription of genes under the control of a number of

transcription factors, including the estrogen receptor, thyroid

receptor, and p53.^11ꢀ14More recently, PAD4 was shown to

deiminate ELK1, and this PTM was associated with increased

transcription of the proto-oncogene c-Fos.¹⁵

In addition to modulating gene transcription, PAD4 activity

has been shown to aﬀect a variety of processes, including apoptosis

(inhibition and overexpression of PAD4 induces apoptosis)^13,14,16

and diﬀerentiation (PAD4 inhibition triggers the diﬀerentiation

of HL60 cells into HL60 granulocytes and monocytes).¹⁷Emer-

ging evidence also suggests that PAD4 may play a key role in

regulating at least a subset of the innate immune response. For

example, PAD4 has been shown to deiminate a number of che-

mokines (e.g., CXCL 5, 8, 10, 11, and 12), and this modiﬁcation

reduces their ability to trigger chemotaxis and cell signaling in

vitro.^18ꢀ20Additionally, PAD4 activity appears to be critical for

the formation of Neutrophil Extracellular Traps (NETs), as Cl-

amidine (Figure 1B), a pan-PAD inhibitor developed by our

lab,²¹prevents NET formation;^22,23neutophils extravate their

chromatin to form net-like structures in response to a number of

signaling molecules of both bacterial (e.g., LPS and fMLP)^22ꢀ24

and human origin (e.g., CXCL2 and CXCL8).²⁵

Dysregulated PAD expression and activity has been associated

with a number of human diseases, including, rheumatoid arthritis

(RA), ulcerative colitis, Crohn’s disease, glaucoma, multiple sclero-

sis, Alzheimer’s disease, and even cancer (for a review, see ref 1).

Although the association between dysregulated PAD activity and

human disease seems clear, especially with our recent demon-

stration that Cl-amidine decreases disease severity in the collagen

induced arthritis (CIA) model of RA and the dextran sodium

sulfate (DSS) model of ulcerative colitis,^26,27what is not known

are the speciﬁc identities of the isozymes involved in the onset

and progression of these pathologies. From the available evi-

dence, the most likely candidates appear to be PADs 2 and 4, as

these enzymes are expressed by immune cells (the diseases listed

above all have an autoimmune component), and both are over-

expressed in these diseases.^1,3,28,29With respect to RA, the

evidence linking dysregulated PAD4 activity to the onset and

progression of this disease are particularly strong. For example,

’ RESULTS AND DISCUSSION

Design, Synthesis, and Characterization of X-Amidine

Side Chain Analogs. The structure of F-amidine, which is used

as a baseline for comparison, can be divided into three main

portions: the warhead, the side chain, and the backbone (Figure 1B).

To better understand the relative contribution of the side chain

to inhibitor potency, we initially synthesized 2-fluoro-N-propy-

lacetamidine (4) (Figure 2), which is essentially a propylamine

modified warhead, to mimic the three-carbon side chain portion

of F-amidine. The compound was synthesized by reaction of

propylamine with 2-fluorothioacetimidate and purified, and IC₅₀

values were obtained for PAD4. The results indicate that 4 is a

relatively weak PAD4 inhibitor (IC₅₀> 1 mM; Table 2) and that

the IC₅₀value is comparable to that obtained for the warhead

alone, i.e., 3. The fact that F-amidine is approximately 2 orders of

magnitude more potent than 4 indicates that the side chain

component of F-amidine is not responsible for the relatively high

potency of F-amidine.

Given that a tryptophan residue (i.e., W347) lines the active

site of PAD4 and interacts with the side chain portion of F-amidine

via hydrophobic and van der Waals interactions (Figure 3), we

considered the possibility that the incorporation of an aromatic

moiety, in place of the aliphatic side chain, would interact

favorably with the W347 indole group. To test this hypothesis,

the ﬂuoroacetamidine warhead was installed on two structurally

similar aromatic amines, i.e., aniline and benzylamine. However,

neither the benzyl (5: IC₅₀> 1 mM) nor the aniline (6: IC₅₀

>

500 μM) derivatives were potent PAD4 inhibitors. Nevertheless,

the fact that the aniline derivative was more potent than the

baseline compound, i.e., the propyl derivative, suggested that its

incorporation into the full inhibitor might enhance inhibitor

potency. To examine this possibility, we synthesized and com-

pared the relative potencies of phenylalanine and phenylglycine

derivatives in which the warhead is placed meta to the α-carbon.

Note that both the phenylalanine and phenylglycine derivatives

were synthesized to explore how diﬀerences in the connectivity

to the backbone aﬀect inhibitor potency. The results of these

6920

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

Figure 2. Structures of side chain analogues (A) and backbone analogues (B).

SAR studies indicated that the phenylalanine derivative, i.e., 7

(IC₅₀= ∼300 μM), was more potent than the phenylglycine

derivative (8) (>2 mM) suggesting that the extra methylene unit

between the backbone and warhead is advantageous. Given that

these values are signiﬁcantly less than those obtained with

F-amidine and represent only a slight improvement over the

aniline derivative on its own, we considered the possibility that

the lack of potency was due to a less than ideal positioning of the

warhead about the phenyl ring. To partially test this hypothesis,

two para-substituted phenylalanine derivatives, i.e., compounds 9

and 10, were synthesized. The ﬁrst compound attaches the

warhead directly to the phenyl ring, mimicking the structure of

the aniline, and the second compound attaches the warhead via a

one carbon methylene linker that was expected to project the

warhead further into the active site where it might react more

favorably with the enzyme. Unfortunately, the IC₅₀values obtained

for compounds 9 and 10 (IC₅₀∼300 μM and >1 mM, respectively)

are similar to those determined for the ortho-substituted derivatives,

6921

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

Table 1. Crystallographic Data and Reﬁnement Statistics^a

inhibitor

o-F-AM

o-Cl-AM

soaking

5 mM, 6 h

crystallographic data

space group

cell dimensions

C2

a = 146.4, b = 60.4, a = 145.7, b = 60.8,

c = 115.5 Å

β = 124.4ꢀ

50.0ꢀ2.1

(2.18ꢀ2.10)

155,187

c = 114.3 Å

β = 123.9ꢀ

50.0ꢀ2.5

(2.59ꢀ2.50)

98,188

resolution range (Å)

total observations

unique observations

completeness (%)

R_merge(%)

44,771

27,393

91.8 (66.7)

4.5 (28.8)

14.9

94.8

6.2 (40.1)

14.8

Figure 3. Crystal structure of PAD4 F-amidine complex. This ﬁgure

3

ÆIæ/σÆIæ

was prepared with UCSF Chimera using the coordinates for the

reﬁnement statistics

resolution (Å)

PAD4 F-amidine complex (PDB ID: 2DW5).

3

20.0ꢀ2.1

22.0/25.0

20.0ꢀ2.5

20.3/24.4

R/R_free(%)

the case, we synthesized compound 11 (Figure 2B), which lacks

the C-terminal carboxamide moiety. The IC₅₀value obtained for

this compound (IC₅₀g 500 μM) confirms that both amide

groups are critical for the overall potency of F-amidine. We next

considered the effect of substituting the benzoyl-carboxamide

for a sulfonamide (12). While this substitution retains the

H-bond acceptor and in fact introduces a second one, the

spatial orientation is somewhat different; the sulfonamide has a

tetrahedral geometry, while the carboxamide has a trigonal

planar geometry. Although it was impossible to predict exactly

how this substitution would affect the potency, the experi-

mental evidence (IC₅₀> 1 mM) clearly indicates that the effect

was deleterious. To determine whether the lack of potency

was solely due to the loss of the C-terminal carboxamide,

the introduction of the sulphonamide, or a combination of

both changes, we next synthesized compound 13, which

retains the C-terminal carboxamide but replaces the ben-

zoyl-carboxamide with a sulfonamide. Interestingly, this com-

pound was also a poor PAD4 inhibitor, which, combined with

the results described above for compounds 11 and 12,

suggests that hydrogen bonding between R374 and the two

backbone carbonyls contributes to the potency of F-amidine

in a synergistic fashion.

Given that substitutions altering the two backbone carbonyls,

as well as the side chain, yielded inhibitors with decreased

potency, we considered that substitutions of the benzoyl group

might yield inhibitors with increased potency. In particular, we

considered substitutions that would mimic an extension of the

peptide chain. Our ﬁrst attempt replaced the phenyl ring with an

imidazole ring (14). This substitution maintains the aromatic

character of the phenyl ring while incorporating an NH moiety

that might mimic a peptide NH, thus becoming a hydrogen bond

donor; an additional hydrogen bond could signiﬁcantly increase

the binding aﬃnity of the inhibitor. The results of inhibition

studies with 14 showed a 10-fold decrease in potency relative to

F-amidine (IC₅₀= 230 ( 110 μM). This result was somewhat

unexpected given that even in the absence of an additional

H-bond, the imidazole moiety was still aromatic and thus should

be, at worst, comparable to the phenyl moiety. Upon further

consideration, however, we realized that the imidazole was likely

protonated under the reaction conditions resulting in a positively

charged species. Therefore, we concluded that the decrease in

potency may actually arise from a chargeꢀcharge repulsion between

root mean square deviation

bond lengths (Å)

bond angles (deg)

0.005

0.887

0.009

1.241

^aValues in parentheses are for the highest resolution shell.

Table 2. IC₅₀Values of PAD4 Inhibitors^a

compd

IC₅₀

1

5.9 ( 0.3 μM

22 ( 2.1 μM

>10 mM

2

3

4

>1 mM

5

>1 mM

6

>0.5 mM

7

310 ( 37 μM

2.1 ( 0.3 mM

310 ( 110 μM

>1 mM

8

9

10

11

12

13

14

15

16

17

18

>1 mM

500 ( 60 μM

230 ( 110 μM

19 ( 3.0 μM

12 ( 6.4 μM

1.9 ( 0.2 μM

2.2 ( 0.3 μM

^aGiven that these compounds are time dependent irreversible inactiva-

tors, these IC₅₀values should be considered apparent measures of aﬃnity.

thereby indicating that the replacement of the aliphatic side chain

of F-amidine with a phenyl moiety does not lead to inhibitors

with enhanced potency.

Design, Synthesis, and Characterization of X-Amidine

Backbone Analogues. Since none of the side chain analogues

weremorepotent than F-amidine, weconsidered whether changes

to the backbone might yield inhibitors with enhanced potency.

The backbone component of F-amidine contains two amide

bonds that are presumed, although not yet proven, to be important

for binding to the enzyme. To determine whether this was indeed

6922

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

Table 3. Inhibition of PAD Isozymes by Haloacetamidine-Based Inhibitors

compd

isozyme

IC₅₀^a(μM)

k_inact(min^ꢀ1

)

K_I(μM)

k_inact/K_I(M^ꢀ1min^ꢀ1

)

F-amidine

PAD1

PAD2

PAD3

PAD4

PAD1

PAD2

PAD3

PAD4

PAD1

PAD2

PAD3

PAD4

PAD1

PAD2

PAD3

PAD4

29 ( 1.32

51 ( 8.96

367 ( 189

22 ( 2.10

0.8 ( 0.3

17 ( 3.1

6 ( 1.0

0.30 ( 0.03

ND^b

110 ( 40

ND^b

2800

380

0.05 ( 0.01

1.0 ( 0.1

2.3 ( 0.1

ND^b

293 ( 193

330 ( 90

62 ( 11

ND^b

170

3000

Cl-amidine

o-F-amidine

o-Cl-amidine

37000

1200

0.056 ( 0.005

2.4 ( 0.2

1.7 ( 0.1

ND^b

28 ( 7.3

180 ( 33

9.4 ( 1.8

ND^b

2000

5.9 ( 0.3

1.4 ( 0.41

g50

13000

180900

7500

34 ( 31.9

1.9 ( 0.21

0.84 ( 0.12

6.2 ( 0.7

0.69 ( 0.34

2.2 ( 0.31

ND^b

6700

0.5 ( 0.17

ND^a

16 ( 9

ND^a

32500

106400

14100

10345

38000

0.3 ( 0.06

0.5 ( 0.11

29 ( 12

13 ( 5.2

^aGiven that these compounds are time dependent irreversible inactivators, these IC₅₀values should be considered apparent measures of aﬃnity. ^bND =

Not determinable. Values for k_inactand K_Icould not be determined because saturation was not observed in the plots of k_obsversus inactivator

concentration. k_inact/K_Ivalues were determined from linear ﬁts of the data.

the protonated imidazole and a positively charged residue on

the enzyme.

chloride derivative was synthesized and assayed, we found that

the inhibitory potency of o-Cl-amidine (18) (IC₅₀= 2.2 ( 0.31

μM) was nearly identical to that of o-F-amidine.

In order to circumvent this potential chargeꢀcharge repul-

sion, we chose to substitute the phenyl ring with a pyrollidine

ring (15). This substitution maintains both the aromaticity and

the NH bond as a potential H-bond donor, but given the much

higher pK_a, the pyrollidine remains neutral under the assay

conditions. Synthesis and evaluation of this derivative was carried

out, and the results indicate a potency that is comparable to that

of F-amidine. Although this inhibitor is not more potent than the

parent compound, the fact that the IC₅₀is the same as the benz-

oylated version suggests two things: (i) the NH may contribute

little to overall binding; and (ii) the loss of potency seen with the

imidazole derivative is likely due to a chargeꢀcharge repulsion.

Because the results from the imidazole and pyrollidine derivatives

suggest the presence of a positively charged residue in close proximity

to the active site, we postulated that the incorporation of a negatively

charged moiety on the benzoyl group would increase inhibitor

potency. Toward this end, we chose to substitute the phenyl ring

of F-amidine with a para-nitro-phenol (16). We postulated that

the positioning of the nitro group para to the hydroxyl would

suﬃciently depressthe pK_aof the hydroxyl to generate an inhibitor

that is at least partially deprotonated. This negative charge should

then be positioned to interact with the positively charged residue

on the enzyme without being sterically cumbersome. Once this

derivative was synthesized and tested, the results showed a small

decrease (∼2-fold) in potency (IC₅₀= 12.1 ( 6.4 μM).

Given these promising results, we next designed and synthe-

sized an F-amidine derivative that incorporated a carboxylate

ortho to the backbone amide (17). Synthesis of this derivative,

which is denoted o-F-amidine, was accomplished by using

phthalic anhydride in the place of benzoyl chloride. As predicted,

the IC₅₀of this compound (1.9 ( 0.21 μM) is 10-fold lower than

that obtained with the parent compound, F-amidine. Given that

the substitution of chloride for ﬂuoride typically increases the

potency of X-amidine inhibitors by ∼5-fold, we anticipated a

similar trend in these derivatives as well. However, once the

Mechanism of Inhibition. Since the parent compounds, i.e.,

F- and Cl-amidine, are mechanism based inactivators, we next

determined whether o-F- and o-Cl-amidine also irreversibly

inactivate PADs 1, 2, 3, and 4. For these studies, a preformed

complex of inactivator and enzyme was generated, then dialyzed

for 20 h, at which point the residual activity was measured. As

shown in Figure S1 of the Supporting Information, little to no

recovery of activity was observed for all of the PADs tested,

thereby confirming that, like F- and Cl-amidine, o-F- and o-Cl-

amidine are irreversible PAD inactivators. In order to help

confirm that PAD inactivation is due to the modification of an

active site residue, substrate protection experiments were per-

formed. For these studies, product formation was measured as a

function of time at 2 different concentrations of substrate (10 and

2 mM benzoyl-^L-arginine ethyl ester (BAEE)) with and without

inactivator. The results (Figure S2 in the Supporting Information)

indicate that for PADs 1, 2, 3, and 4 the rates of inactivation are

slower at higher substrate concentrations. Such a result is most

consistent with enzyme inactivation being due to the modifica-

tion of an active site residue, which, based on precedence, is likely

the active site cysteine.

Selectivity Studies. To evaluate the selectivity of o-F- and o-

Cl-amidine, k_inact, K_I, and k_inact/K_Ivalues were determined for all

of the PADs (Table 3). Note that for irreversible inhibitors these

rate constants provide a more accurate assessment of selectivity

than IC₅₀values. For these experiments, the residual activity was

measured as a function of time with increasing concentrations of

inactivator (see Figure 4A,C for representative data). From these

curves, the pseudo-first order rate constants of inactivation (i.e.,

k_obs) were determined and plotted against inactivator concentra-

tion and fit to eq 4 to provide values for k_inact, K_I, and k_inact/K_I

(see Figure 4B,D for representative data). Interestingly, based

on the k_inact/K_Ivalues, o-Cl-amidine preferentially inactivates

PAD1, as the k_inact/K_Ivalue obtained for this isozyme is 8-, 10-,

6923

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

Figure 4. Inactivation of PAD4 with o-F- and o-Cl-amidine. The residual activity of PAD4 was measured versus time with increasing concentrations of

(A) o-F-amidine and (C) o-Cl-amidine. The rates (k_obs) were plotted versus the concentration of (B) o-F-amidine and (D) o-Cl-amidine to obtain the

inactivation parameters.

Figure 5. Overlays of the crystal structure of PAD4 with (A) F-amidine and o-F-amidine bound and (B) Cl-amidine and o-Cl-amidine bound.

(C) Stereoview of the crystal structure of PAD4 bound to o-F-amidine.

6924

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

3-fold higher than that obtained for PADs 2, 3, and 4, respec-

tively. Similar trends were observed for o-F-amidine, where the

k_inact/K_Ivalues are 24-, 27-, and 6-fold higher for PAD1 than for

PADs 2, 3, and 4, respectively. Given that the relative selectivity

of o-Cl-amidine for PADs 1 and 4 is relatively modest (∼3-fold),

this compound is a PAD1 and 4 selective inhibitor, whereas o-F-

amidine is primarily a PAD1 selective inhibitor. It is also

important to note that o-F-amidine is 60-fold more potent than

the parent compound F-amidine.

complex where the effects of the ortho-carboxylate on enzyme

inactivation are expected to be most profound, we used site

directed mutagenesis to identify the residue(s) that is/are most

critical for inhibitor recognition. Specifically, Q346, W347, and

R374 were individually converted to alanine, and IC₅₀values

were determined for the mutant enzymes. To aid our analysis, we

also generated the Q346E, W347F, and R374K mutants. The

results of these studies (Table 4) indicate that mutation of Q346

to either an alanine or a glutamate had essentially no effect on the

IC₅₀values determined for o-F- and o-Cl-amidine, suggesting

that the enhanced potency of these compounds is not due to an

interaction between the carboxylate and carboxamide moieties

on the inhibitor and Q346, respectively. In contrast, the IC₅₀

values determined for the R374K and R374A mutants are ∼2.4-

fold and 28-fold higher for o-F-amidine when compared to the

values obtained for the wild-type enzyme. Since similar effects on

potency were observed for F-amidine, these results suggest that

the interaction between the ortho-carboxylate and the R374

guanidinium is not crucial for the enhanced potency of these

PAD inactivators. Nevertheless, this arginine residue is conserved

in both PADs 1 and 4, and the presence of R374 may explain why

o-F- and o-Cl-amidine preferentially inactivate these enzymes.

Unfortunately, neither of the tryptophan mutants were active

(k_cat/K_me 0.2 M^ꢀ1s^ꢀ1); thus, the importance of the indole NH

group to inhibitor potency could not be evaluated directly.

Nevertheless, this interaction is likely important when one

considers the fact that the introduction of the ortho-carboxylate

led to enhanced potency among all the isozymes and the fact that

this residue is universally conserved among the PAD isozymes.

Bioavailability of o-F- and o-Cl-Amidine. To illustrate the

utility of o-F- and o-Cl-amidine as chemical probes of in vivo

PAD4 function, we set out to evaluate their activity in cellulo.

Initially, the bioavailability of the compounds was determined by

monitoring their effects on histone H3 citrullination in HL60

granulocytes; PAD4 is known to deiminate histone H3 in this cell

line in response to the calcium ionophore A23187.^37,38Western

blots, using an anticitrullinated histone H3 antibody, demon-

strated that the levels of deiminated histone H3 are decreased

upon a 30 min incubation with the inhibitors. Significantly, the

improved in vitro potency of o-F- and o-Cl-amidine is paralleled

in cellulo, as exemplified by the fact that as little as 1 μM of

Structural Studies. To help determine the molecular basis for

the enhanced potency of both o-F- and o-Cl-amidine, the structures

of these compounds bound to the PAD4 calcium complex were

3

determined and compared to the structure of PAD4 calcium F-

3

amidine (Figure 5A,B). The structures of these complexes are

virtually superimposable (rmsd = 0.222 Å for comparable 573

Cα) and revealed the presence of three potential residues, i.e.,

Q346, W347, and R374, that could provide favorable hydrogen

bonding and/or electrostatic interactions with the ortho-carbox-

ylate that would explain the enhanced potency of o-F- and o-Cl-

amidine (Figure 5C). While the distance between the indole NH

of W347 and the ortho-carboxylate is characteristic of a hydrogen

bond (i.e., 2.9 Å), the distances between the ortho-carboxylate

and the carboxamide of Q346 (i.e., 3.9 Å) and the guanidinium

group of R374 (i.e., 4.6ꢀ5.6 Å) are greater than that typically

associated with such an interaction. Note that in the o-F-amidine

complex, water-mediated hydrogen bonds between Q347 and

the carboxylate of o-F-amidine are observed because of higher-

resolution data. Given that these structures represent the dead-

end complex and as such do not represent the initial encounter

Table 4. PAD4 IC₅₀Values

IC₅₀

o-F-amidine o-Cl-amidine IC₅₀F-amidine IC₅₀Cl-amidine

enzyme

(μM)

wild-type 1.9 ( 0.21

2.2 ( 0.31

2.4 ( 0.17

1.1 ( 0.19

5.3 ( 0.54

22 ( 0.75

22 ( 2.1

21 ( 2.2

51 ( 2.6

15 ( 5.5

530 ( 42

5.9 ( 0.3

4.0 ( 1.6

3.4 ( 0.8

9.0 ( 1.6

15 ( 4.0

Q346E

Q346A

R374K

R374A

1.1 ( 0.31

0.4 ( 0.02

4.6 ( 2.79

32 ( 11

Figure 6. (A) Levels of citrullinated histone H3 in HL-60 granulocytes after treatment with o-F- and o-Cl-amidine. (B) o-F- and o-Cl-amidine induce the

diﬀerentiation of HL-60 cells. HL-60 cells were exposed to either o-F- or o-Cl-amidine over the course of 48 h with time points taken after 6, 12, 24,

and 48 h.

6925

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

Figure 7. Eﬀects of o-F-amidine (A) and o-Cl-amidine (B) in combination with doxorubicin on cell viability of HL-60 cells. Cell viability was measured,

using a standard MTT assay, after a 24 h incubation with the inhibitors. Each data point represents an average of three trials (p < 0.05). The combination

of o-F-amidine (C) and o-Cl-amidine (D) with doxorubicin resulted in synergistic killing of HL-60 cells. Synergistic, additive, and subadditive eﬀects of

the combination therapy were determined by a comparison of the O/E ratios, where O/E < 0.8 is considered synergistic, O/E = 0.8ꢀ1.2 is additive, and

an O/E > 1.2 is subadditive.

Table 5. Cytotoxicity of o-F- and o-Cl-amidine in

o-F-amidine decreased the levels of deiminated histone H3 to the

same extent as 100 μM of Cl-amidine (Figure 6A).

Combination with Doxorubicin in HL-60 Cells^a

Effect of o-F- and o-Cl-Amidine on Cell Viability. Having

demonstrated that the additional carboxylate does not affect the

bioavailability of these inhibitors, we next investigated whether

these compounds potentiate the cell killing effects of doxorubi-

cin, similarly to the parent compounds F- and Cl-amidine.¹⁷For

these experiments, the effect of doxorubicin on HL-60 cell survival

was determined in the absence and presence of increasing

concentrations of o-F- and o-Cl-amidine using the standard

MTT assay. As depicted in Figure 7, addition of o-F- and o-Cl-

amidine caused the doseꢀresponse curves to shift to the left,

resulting in an approximate 6-fold decrease in the EC₅₀for

doxorubicin (Table 5). The fact that the magnitude of the effect

is similar for both o-F- and o-Cl-amidine is consistent with the

fact that these compounds are equipotent PAD4 inhibitors.

Relative to 1 μM Cl-amidine, which reduced the EC₅₀by ∼2-

fold, both o-F- and o-Cl-amidine decreased the EC₅₀by ∼6-fold

at concentrations as low a 100 nM. In addition to decreasing the

EC₅₀of doxorubicin, both compounds also decrease cell survival

to background levels, i.e., no viable cells were detectable. To

determine whether the effects imparted by o-F- and o-Cl-amidine

on doxorubicin cell cytotoxicity were subadditive, additive, or

synergistic, the additive model was used.^39,40This model predicts

that two drugs are subadditive when the observed/expected

conditions

EC₅₀(μM)

% survival

doxorubicin

2.50 ( 1.16

0.39 ( 0.02

0.34 ( 0.03

0.34 ( 0.12

0.45 ( 0.12

0.44 ( 0.03

0.42 ( 0.01

1.42 ( 0.02

1.29 ( 0.04

0.14 ( 0.09

2.33 ( 1.74

1.30 ( 0.66

0.49 ( 0.25

12.0 ( 4.4

ND^b

doxorubicin + 100 nM o-Cl-amidine

doxorubicin + 1 μM o-Cl-amidine

doxorubicin + 10 μM o-Cl-amidine

doxorubicin + 100 nM o-F-amidine

doxorubicin + 1 μM o-F-amidine

doxorubicin + 10 μM o-F-amidine

doxorubicin + 1 μM Cl-amidine

doxorubicin + 10 μM Cl-amidine

doxorubicin + 100 μM Cl-amidine

doxorubicin + 1 μM F-amidine

doxorubicin + 10 μM F-amidine

doxorubicin + 100 μM F-amidine

^aEC₅₀is deﬁned as the concentration of agent that reduces cell viability

to 50% of maximum. EC₅₀values were determined by ﬁtting the

doseꢀresponse data to eq 1. % survival is based on the cell viability at

the maximum agent concentration tested. ^bND: no viable cells were

detected as deﬁned by the lack of absorbance above the 100% killing

control.

ND^b

8.14 ( 5.10

ND^b

6926

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

(O/E) ratio is 1.2, additive when the O/E ratio is between 0.8

and 1.2, and synergistic when the O/E ratio is <0.8. As depicted

in Figure 7C,D, o-F- and o-Cl-amidine potentiate the cell killing

of doxorubicin in HL-60 cells in a synergistic and dose dependent

fashion.

colitis by removing the cells that drive colitis. With the increased

potency of o-F-amidine, we thought it would be prudent to test

the hypothesis that o-F-amidine has a more potent impact on

inflammatory cell apoptosis than Cl-amidine. Results shown in

Figure 8 are consistent with this hypothesis. For example,

representative scatterplots of Annexin V/propidium iodide stain-

ing following the exposure of TK6 lymphoblastoid cells to o-F-

amidine (50 μM) or Cl-amidine (50 μM) for up to 8 h demon-

strate that o-F-amidine increases the number of apoptotic cells to

a greater extent than an equivalent concentration of Cl-amidine

(Figure 8A). Figure 8B depicts the quantification of apoptosis

(4th quadrant). Additionally, o-F-amidine induces PARP clea-

vage earlier and to a greater extent than the same concentration

of Cl-amidine (Figure 8C).

o-F- and o-Cl-Amidine Mediated Differentiation of HL60

Cells. Given that F- and Cl-amidine can trigger the differentiation

of multiple cancer cell lines,¹⁷including the differentiation of

HL60 cells into HL60 granulocytes, we evaluated whether o-F-

and o-Cl-amidine can also act as differentiating agents. For these

experiments, PAD4 expression levels were monitored after the

treatment of HL60 cells with o-F- and o-Cl-amidine; PAD4

expression levels are increased upon the differentiation of

HL60 cells. Western blots, using an anti-PAD4 antibody, indicate

that o-F- and o-Cl-amidine treatment causes a dose and time

dependent increase in PAD4 protein expression. For example, at

the highest doses of o-F- and o-Cl-amidine tested (i.e., 10 μM),

PAD4 expression was apparent after only 6 h of treatment;

however, with lower doses (e.g., 1 μM), increased levels of this

enzyme were observed only after 12 and 24 h. In total, the results

of these experiments demonstrate that these inhibitors differentiate

HL-60 cells into HL-60 granulocytes in a manner similar to that

of the parent compounds.

’ CONCLUSIONS

In conclusion, SAR led to the development of two new PAD

inhibitors, o-F- and o-Cl-amidine, that are signiﬁcantly (up to 65-

fold) more potent than F- and Cl-amidine. The increased potency is

due to the carboxylate functionality in the ortho position of the

benzoyl ring. Although the inclusion of this group enhanced

potency for all of the PADs, the eﬀects are isozyme and warhead

dependent. For example, with o-Cl-amidine, the increase in

k_inact/K_Iranged from 3-fold for PADs 1 and 4 to 12-fold for PAD2,

whereas for o-F-amidine, the k_inact/K_Ivalues were increased by

Inflammatory Cell Apoptosis. We have recently shown that

Cl-amidine induces inflammatory cell apoptosis in vitro and

in vivo²⁷and have suggested that this inhibitor may suppress

Figure 8. o-F-amidine is more potent at driving apoptosis in inﬂammatory cells than Cl-amidine. TK6 cells, a human lymphoblastoid cell line, were

exposed to either Cl-amidine (50 μg/mL) or o-F-amidine (50 μg/mL) for indicated time periods. (A) Representative scatterplots from one of three

separate experiments. (B) Average percentage apoptosis (from percentages in the 4th quadrant) for Cl-amidine and o-F-amidine as indicated from 3

experiments. (C) PARP cleavage by Cl-amidine and o-F-amidine.

6927

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

11-, 20-, 39-, and 65-fold for PADs 4, 2, 3, and 1, respectively. The

observed warhead dependence on k_inact/K_Ilikely reﬂects diﬀer-

ences in the ability to position the warhead for nucleophilic attack

by the active site cysteine. This result is signiﬁcant because it

suggests that the warhead itself can act as a selectivity determi-

nant. Consistent with this notion are the interisozyme selectivity

data. For example, relative to Cl-amidine, the inclusion of the

ortho-carboxylate in o-Cl-amidine actually decreases the selec-

tivity for PADs 2 and 3 by 2- to 3-fold. In contrast, with o-F-

amidine the selectivity trends are the reverse, as the k_inact/K_I

values are increased by 2- to 3-fold for PADs 2 and 3 (to 24- and

27-fold, respectively) and 6-fold for PAD4. Given these results, o-

Cl-amidine is a PAD1 and 4 selective inhibitor, whereas o-F-

amidine is primarily a PAD1 selective inhibitor. This preference

for PAD1 suggests that o-F-amidine will serve as a versatile tool

that can be used to assign physiological functions to PAD1. On

the basis of the crystallographic data, as well as mutagenesis

studies, the enhanced potency and selectivity is likely due to

interactions between the ortho-carboxylate and the indole nitro-

gen of W347 of PAD4. In cellulo studies indicate that these two

compounds are better than F- and Cl-amidine in their ability to

inhibit histone deimination, trigger HL-60 cell diﬀerentiation,

and potentiate the cell killing eﬀects of doxorubicin in HL-60

cells in a synergist and dose dependent fashion. Finally, o-F-

amidine induces apoptosis of inﬂammatory cells to a greater

extent than Cl-amidine. Given these ﬁndings, the incorporation

of a carboxylate at this position will likely prove to be an

important feature of third and fourth generation PAD inhibitors.

Future studies will focus on evaluating the eﬃcacy of o-F- and o-

Cl-amidine in mouse models of RA, colitis, and cancer.

2-Fluoro-N-phenylacetimidamide HCl (6). To a solution of

3

ethyl 2-fluorothioacetimidate (174 mg, 1.0 mmol) in dry ethanol (3 mL)

at 0 ꢀC was added aniline (93 mg, 1.0 mmol). After 10 min, the reaction

was allowed to warm to rt, and stirring was continued for an

additional 50 min. The mixture was partitioned between ether

(3 mL) and H₂O (5 mL); the aqueous layer was separated and

lyophilized to yield the title compound as an off-white solid (150 mg,

80%). ¹H NMR (300 MHz, D₂O) δ (ppm): 7.50ꢀ7.35 (m, 3H), 7.25

(m, 2H), 5.36 (d, J_HF= 44.7 Hz, 2H). ¹³C NMR (75 MHz, D₂O) δ

(ppm): 163.52 (d, J_CF= 19.8 Hz), 133.28, 130.94, 127.56, 125.31,

77.86 (d, J_CF= 179.1 Hz). HRMS (ESI) m/z calculated for

[C₈H₁₀FN₂⁺], 153.0820; observed, 153.0832.

N¹-Benzoyl-(3-(2-fluoroacetimidamido))-L-phenylalanine

Amide TFA (7). Rink AM Amide resin (300 mg, 0.2 mmol) was

3

treated twice with 5 mL of 20% piperidine (in DMF) for 20 min and

subsequently washed with dry DMF (3 ꢁ 5 mL). Fmoc-(3-nitro)Phe-

OH (100 mg, 0.23 mmol), HOTT (S-(1-oxido-2-pyridyl)-thio-N,N,N⁰,

N⁰-tetramethyluronium hexafluorophosphate; 303 mg, 0.35 mmol), and

triethylamine (0.064 mL, 0.46 mmol) were dissolved in dry DMF

(3 mL) and allowed to react for 10 min before being added to the resin.

The reaction mixture was rocked at rt for 3 h before the resin was filtered

and washed with DMF (3 ꢁ 5 mL). The resin was treated with acetic

anhydride (0.19 mL, 2.0 mmol) and triethylamine (0.56 mL, 4.0 mmol)

in DMF (5 mL) for 2 h to cap any remaining amino groups on the resin.

The resin was collected by filtration, washed with DMF (2 ꢁ 5 mL), and

treated twice for 20 min with 5 mL of 20% piperidine in DMF. The resin

was then washed with DMF (3 ꢁ 5 mL), resuspended in DMF (5 mL),

and treated with benzoyl chloride (0.093 mL, 0.8 mmol) and triethy-

lamine (0.22 mL, 1.6 mmol) and rocked at rt overnight. The resin was

washed with DMF (3 ꢁ 5 mL), then treated overnight at rt with 2 M

tin(II)chloride (6 mL), collected by filtration, and washed with DMF

(3 ꢁ 5 mL). The resin was then treated overnight with a solution of ethyl

2-fluorothioacetimidate (128 mg, 0.8 mmol) in DMF (5 mL), then

collected by filtration, washed with DMF (3 ꢁ 5 mL), ethanol (3 ꢁ

5 mL), and DCM (3 ꢁ 5 mL), and dried for 1 h under reduced

pressure. The dry resin was treated three times (1 h each) with 5 mL

of TFA/TIS/H₂O (95/2.5/2.5). The filtrates were collected, com-

bined, and concentrated. Cold ether (25 mL) was added to the

remaining residue, and the resulting precipitate was collected by

centrifugation and purified by RP-HPLC to yield the title compound

’ EXPERIMENTAL SECTION

Chemicals. Cl-amidine (1), F-amidine (2), and 2-fluoroacetami-

dine (3) were synthesized as previously described.^41,42PADs 1ꢀ4 were

purified as previously described.^4,43Cell culture media, RPMI 1640, and

fetal bovine serum were obtained from ThermoScientific (Pittsburgh,

PA). All compounds were purified to g95% purity. Purity was assessed

by HPLC.

1

as an off-white solid (16.5 mg, 18%). H NMR (400 MHz, D₂O)

2-Fluoro-N-propylacetamidine HCl (4). To a solution of ethyl

3

δ (ppm): 7.52 (m, 3H), 7.36 (m, 4H), 7.14 (m, 2H), 5.28 (d, J_HF

=

2-fluorothioacetimidate (174 mg, 1.0 mmol) in dry ethanol (3 mL) at

0 ꢀC was added propylamine (59 mg, 1.0 mmol). After 10 min, the

reaction was allowed to warm to rt, and stirring was continued for an

additional 50 min. The mixture was partitioned between ether (3 mL)

and H₂O (5 mL); the aqueous layer was separated and lyophilized to

yield the title compound as an off-white solid (134 mg, 87%). ¹H NMR

(400 MHz, D₂O) δ (ppm): 5.07 (d, J_HF= 45.2 Hz, 2H), 3.10 (t, J = 7.2

Hz, 2H), 1.46 (q, J = 7.2 Hz, 2H), 0.73 (t, J = 7.2 Hz, 3H). ¹³C NMR

44.4 Hz), 4.72 (dd, J = 5.6, 9.6 Hz, 1H), 3.26 (dd, J = 5.6, 14.0 Hz,

1H), 2.97 (dd, J = 10.0, 14.0 Hz, 1H). ¹³C NMR (100 MHz, D₂O)

δ (ppm): 175.80, 170.63, 163.26 (d, J_CF= 19.1 Hz), 139.42, 132.61,

132.45, 132.29, 130.50, 130.38, 128.65, 127.03, 126.04, 123.99,

77.60 (d, J_CF= 178.8 Hz), 54.71, 36.65. HRMS (ESI) m/z calculated

for [C₁₈H₂₀FN₄O₂⁺], 343.1570; observed, 343.1565.

N¹-Benzoyl-(3-(2-fluoroacetimidamido))-L-phenylglycine

Amide TFA (8). Rink AM Amide resin (300 mg, 0.2 mmol) was

(100 MHz, D₂O) δ (ppm): 162.21 (d, J_CF= 19.8 Hz), 77.53 (d, J_CF

=

3

treated twice with 5 mL of 20% piperidine (in DMF) for 20 min and

subsequently washed with dry DMF (3 ꢁ 5 mL). Bz-(3-NO₂)Phg-OH

(180 mg, 0.6 mmol), HBTU (2-(1H-benzotriazol-1-yl)-1,1,3,3-tetra-

methyluronoium hexafluorphosphate; 228 mg, 0.6 mmol), HOBt (92

mg, 0.6 mmol), and triethylamine (0.17 mL, 1.2 mmol) were dissolved

in dry DMF (5 mL) and allowed to react for 10 min before being added

to the resin. The reaction mixture was rocked at rt for 3 h before the resin

was collected by filtration and washed with DMF (3 ꢁ 5 mL). The resin

was rocked overnight at rt in 2 M tin(II)chloride (6 mL), collected by

filtration, and washed with DMF (3 ꢁ 5 mL). The resin was then treated

overnight with a solution of ethyl 2-fluorothioacetimidate (128 mg, 0.8

mmol) in DMF (5 mL), then collected by filtration, washed with DMF

(3 ꢁ 5 mL), ethanol (3 ꢁ 5 mL), and DCM (3 ꢁ 5 mL), and dried for

1 h under reduced pressure. The dry resin was treated three times (1 h each)

177.4 Hz), 43.65, 20.15, 10.30. HRMS (ESI) m/z calculated for

+

[C₅H₁₂FN₂], 119.0984; observed, 119.0993.

N-Benzyl-2-fluoroacetimidamide HCl (5). To a solution of

3

ethyl 2-fluorothioacetimidate (174 mg, 1.0 mmol) in dry ethanol (3 mL)

at 0 ꢀC was added benzylamine (107 mg, 1.0 mmol). After 10 min, the

reaction was allowed to warm to rt, and stirring was continued for an

additional 50 min. The mixture was partitioned between ether (3 mL)

and H₂O (5 mL); the aqueous layer was separated and lyophilized to

yield the title compound as a white powder (147 mg, 73%). ¹H NMR

(300 MHz, D₂O) δ (ppm): 7.29 (m, 5H), 5.18 (d, J_HF= 45 Hz, 2H),

4.44 (s, 2H). ¹³C NMR (75 MHz, D₂O) δ (ppm): 162.82 (d, J_CF= 19.8

Hz), 133.84, 129.27, 128.67, 127.84, 77.85 (d, J_CF= 177.45 Hz), 45.74.

+

HRMS (ESI) m/z calculated for [C₉H₁₂FN₂], 167.0984; observed,

167.0979.

6928

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

with 5 mL of TFA/TIS/H₂O (95/2.5/2.5). The filtrates were collected,

combined, and concentrated. Cold ether (25 mL) was added to the

remaining residue, and the resulting precipitate was collected by centri-

fugation and purified by RP-HPLC to yield the title compound as a white

powder (19.5 mg, 22%). ¹H NMR (400 MHz, D₂O) δ (ppm): 7.64 (m,

2H), 7.48 (m, 3H), 7.36 (m, 3H), 7.28 (m, 1H), 5.54 (s, 1H), 5.34 (d,

J_HF= 44.8 Hz, 2H). ¹³C NMR (100 MHz, D₂O) δ (ppm): 175.11,

173.95, 170.42, 163.54 (d, J_CF= 19.8 Hz), 138.40, 132.98, 132.60,

132.55, 131.21, 129.70, 128.76, 127.36, 125.97, 124.87, 120.11, 77.69 (d,

TFA/TIS/H₂O (95/2.5/2.5) at rt for 3 h. The resulting filtrates were

collected, combined, and concentrated under a stream of nitrogen. Cold

ether (20 mL) was added to the remaining residue, and the resulting

precipitate was collected by centrifugation and purified by RP-HPLC

to afford N-α-benzoyl-^L-(4-aminomethyl)phenylalanine amide. Ethyl

fluoroacetimidate hydrochloride (10 mg, 0.073 mmol), dry triethyla-

mine (0.004 mL, 0.029 mmol), and N-α-benzoyl-^L-(4-aminomethyl)-

phenylalanine amide (6 mg, 0.015 mmol) were dissolved in dry DMF

(1 mL), and the mixture was stirred under nitrogen for 16 h. Purification

by RP-HPLC afforded the title compound as a hygroscopic white

powder upon lyophilization. ¹H NMR (400 MHz, CD₃OD) δ (ppm):

7.63 (d, J = 7.7 Hz, 2H), 7.42 (t, J = 7.4 Hz, 1H), 7.33 (t, J = 7.6 Hz, 2H),

7.28 (d, J = 8.0 Hz, 2H), 7.19 (d, J = 8.0 Hz, 2H), 5.13 (d, J = 45.4 Hz,

2H), 4.73 (dd, J = 5.5, 10.0 Hz, 1H, obtained in D₂O, 500 MHz), 4.40 (s,

2H), 3.27 (dd, J = 5.4, 13.9 Hz, 1H, obtained in D₂O, 400 MHz), 2.94

(dd, J = 9.9, 13.9 Hz, 1H). ¹³C NMR (100 MHz, CD₃OD) δ (ppm):

175.87, 170.78, 162.63 (d, J_CF= 18.9 Hz), 137.13, 132.85, 132.42,

132.38, 129.78, 128.68, 127.80, 127.11, 77.62 (d, J_CF= 177.3 Hz), 55.03,

+

J_CF= 178.1 Hz), 57.63. HRMS (ESI) m/z calculated for [C₁₇H₁₈FN₄O₂],

329.1414; observed, 329.1417.

N¹-Benzoyl-(4-(2-fluoroacetimidamido))-L-phenylalanine

Amide TFA (9). Rink AM Amide resin (300 mg, 0.2 mmol) was

3

treated twice with 5 mL of 20% piperidine (in DMF) for 20 min and

subsequently washed with dry DMF (3 ꢁ 5 mL). Fmoc-(4-nitro)Phe-

OH (100 mg, 0.23 mmol), HOTT (303 mg, 0.35 mmol), and triethylamine

(0.064 mL, 0.46 mmol) were dissolved in dry DMF (3 mL) and allowed

to react for 10 min before being added to the resin. The reaction mixture

was rocked at rt for 3 h before the resin was filtered and washed with

DMF (3 ꢁ 5 mL). The resin was treated with acetic anhydride (0.19 mL,

2.0 mmol) and triethylamine (0.56 mL, 4.0 mmol) in DMF (5 mL) for

2 h to cap any remaining amino groups on the resin. The resin was

collected by filtration, washed with DMF (2 ꢁ 5 mL), and treated twice

for 20 min with 5 mL of 20% piperidine (in DMF). The resin was then

washed with DMF (3 ꢁ 5 mL), resuspended in DMF (5 mL), and

treated with benzoyl chloride (0.093 mL, 0.8 mmol) and triethylamine

(0.22 mL, 1.6 mmol), and rocked at rt overnight. The resin was washed

with DMF (3 ꢁ 5 mL) then treated overnight at rt with 2 M

tin(II)chloride (6 mL), collected by filtration, and washed with DMF

(3 ꢁ 5 mL). The resin was then treated overnight with a solution of ethyl

2-fluorothioacetimidate (128 mg, 0.8 mmol) in DMF (5 mL), then

collected by filtration, washed with DMF (3 ꢁ 5 mL), ethanol (3 ꢁ

5 mL), and DCM (3 ꢁ 5 mL), and dried for 1 h under reduced pressure.

The dry resin was treated three times (1 h each) with 5 mL of TFA/TIS/

H₂O (95/2.5/2.5). The filtrates were collected, combined, and con-

centrated. Cold ether (25 mL) was added to the remaining residue, and

the resulting precipitate was collected by centrifugation and purified by

RP-HPLC to yield the title compound as an off-white solid. ¹H NMR

(400 MHz, CD₃OD) δ (ppm): 7.76 (d, J = 8.3 Hz, 2H), 7.55ꢀ7.49 (m,

3H), 7.44 (m, 2H), 7.29 (d, J = 8.3 Hz, 2H), 5.43 (d, J_HF= 45.2 Hz),

3.66ꢀ3.48 (m, 1H), 3.36 (dd, J = 4.8, 13.8 Hz, 1H), 3.09 (dd, J = 9.4,

13.5, Hz, 1H). ¹³C NMR (100 MHz, D₂O) δ (ppm): 175.99, 170.18,

164.99 (d, J_CF= 19.4 Hz), 140.78, 135.25, 133.15, 132.56, 129.74,

128.55, 126.66, 80.20, 78.42 (d, J_CF= 179.9 Hz), 56.16, 38.93. HRMS

+

45.16, 36.79. HRMS (ESI) m/z calculated for [C₁₉H₂₂FN₄O₂],

357.1227; observed, 357.1732.

N¹-Benzoyl-N⁴-(2-fluoro-1-iminoethyl)-1,4-diaminobu-

tane TFA (11). To a solution of N¹-benzoyl-1,4-diaminobutane TFA

3

(39 mg, 0.13 mmol) in dry methanol (3 mL) was added ethyl

2-fluorothioacetimidate (36 mg, 0.25 mmol) and Cs₂CO₃(62 mg,

0.19 mmol). The reaction was allowed to stir at rt overnight before

being quenched with TFA. The mixture was subsequently diluted with

H₂O and purified by RP-HPLC to yield the title compound as a clear

colorless residue (46.5 mg, 100%). ¹H NMR (300 MHz, D₂O) δ (ppm):

7.60ꢀ7.30 (m, 5H), 5.07 (d, J_HF= 44.7 Hz, 2H), 3.26 (m, 4H), 1.47 (m,

4H). ¹³C NMR 100 MHz, D₂O) δ (ppm): 170.83, 162.34 (J_CF= 19.8

Hz), 133.51, 131.97, 128.65, 126.82, 77.49 (J_CF= 177.4 Hz), 41.61,

39.06, 25.67, 23.95. HRMS (ESI) m/z calculated for [C₁₃H₁₉FN₃O⁺],

252.1512; observed, 252.1517.

N¹-Benzenesulfonyl-N⁴-(2-fluoro-1-iminoethyl)-1,4-dia-

minobutane TFA (12). To a solution of N¹-benzenesulfonyl-1,4-

3

diaminobutane TFA (43 mg, 0.13 mmol) in dry methanol (3 mL) was

added ethyl 2-fluorothioacetimidate (36 mg, 0.25 mmol) and Cs₂CO₃

(62 mg, 0.19 mmol). The reaction was allowed to stir at rt overnight

before being quenched with TFA. The mixture was subsequently diluted

with H₂O and purified by RP-HPLC to yield the title compound as a

1

clear colorless residue (41.5 mg, 80%). H NMR (300 MHz, D₂O) δ

(ppm): 7.72 (m, 2H), 7.51 (m, 3H), 5.07 (d, J_HF= 44.7 Hz, 2H), 3.08

(t, J = 6.9 Hz, 2H), 2.75 (t, J = 6.8 Hz, 2H) 1.45 (m, 4H). ¹³C NMR

(75 MHz, D₂O) δ (ppm): 162.50 (d, J_CF= 19.3 Hz), 138.36, 133.65,

129.68, 126.76, 77.69 (d, J_CF= 177.45 Hz), 42.23, 41.66, 25.99, 23.94.

HRMS (ESI) m/z calculated for [C₁₂H₁₉FN₃O₂S⁺], 288.1182; ob-

served, 288.1183.

+

(ESI) m/z calculated for [C₁₈H₂₀FN₄O₂], 343.1570; observed,

343.1570.

N¹-Benzoyl-4-((2-fluoro-1-iminoethyl)aminomethyl)-L-

N¹-Benzenesulfonyl-N⁵-(2-fluoro-1-iminoethyl)-L-ornithine

phenylalanine Amide TFA (10). Rink AM Amide resin (300 mg,

3

0.186 mmol) was preswelled in DMF (5 mL) for 1 h, filtered, washed

with DMF (2 ꢁ 5 mL), treated with 20% piperidine (in DMF) (2 ꢁ

5 mL) for 20 min, and subsequently washed with dry DMF (3 ꢁ 5 mL).

Fmoc-^L-(4-amino(Boc)methyl)Phe-OH (384 mg, 0.744 mmol), HBTU

(282 mg, 0.744 mmol), and HOBt (114 mg, 0.744 mmol) were

dissolved in DMF (3 mL) and allowed to stand at rt for 10 min before

N-methylmorpholine (0.4 M in DMF) (3.7 mL, 1.49 mmol) was added.

The reaction mixture was rocked at rt for 3 h before the resin was filtered,

washed with DMF (3 ꢁ 5 mL), and treated with 20% piperidine (in

DMF) (2 ꢁ 5 mL) for 1 h and subsequently filtered and washed with

DMF (3 ꢁ 5 mL). A mixture of benzoyl chloride (105 mg, 0.744 mmol)

and N-methylmorpholine (0.4 M in DMF) (3.7 mL, 1.49 mmol) was

added to the resin, and this suspension was rocked at rt. After 16 h, the

resin was collected by filtration and washed with DMF (3 ꢁ 5 mL),

ethanol (2 ꢁ 5 mL), and DCM (2 ꢁ 5 mL) and dried overnight under

reduced pressure. The dry resin was treated twice with a mixture of

Amide TFA (13). Rink AM Amide resin (500 mg, 0.36 mmol) was

3

preswelled in DMF (15 mL) for 1 h, filtered, washed with DMF (2 ꢁ

15 mL), treated with 20% piperidine (in DMF) (2 ꢁ 15 mL) for 20 min,

and subsequently washed with dry DMF (3 ꢁ 15 mL). Fmoc-Orn-

(Dde)-OH (552 mg, 1.07 mmol), HBTU (404 mg, 1.07 mmol), and

HOBt (163 mg, 1.07 mmol) were dissolved in DMF (5 mL) and allowed

to stand at rt for 10 min before N-methylmorpholine (0.4 M in DMF)

(5.3 mL, 2.13 mmol) was added. The reaction mixture was rocked at rt

for 16 h before the resin was filtered, washed with DMF (3 ꢁ 5 mL), and

treated with 20% piperidine (in DMF) (2 ꢁ 15 mL) for 1 h and

subsequently filtered and washed with DMF (3 ꢁ 15 mL). A mixture of

benzenesulfonyl chloride (215 mg, 1.42 mmol) and N-methylmorpho-

line (0.4 M in DMF) (7 mL, 2.84 mmol) was added to the resin, and this

suspension was rocked at rt. After 16 h, the resin was collected

by filtration and washed with DMF (3 ꢁ 15 mL). The resin was treated

with 2% hydrazine (in DMF) (15 mL) for 2 h, washed with DMF

6929

dx.doi.org/10.1021/jm2008985 |J. Med. Chem. 2011, 54, 6919–6935

Journal of Medicinal Chemistry

ARTICLE

(3 ꢁ 15 mL), ethanol (2 ꢁ 15 mL), and DCM (2 ꢁ 15 mL), collected by

filtration, and dried overnight under reduced pressure. Ethyl fluoro-

acetimidate hydrochloride (390 mg, 2.77 mmol) and dry triethylamine

(0.38 mL, 2.77 mmol) were added to the resin suspended in DMF

(15 mL), and the mixture was stirred under nitrogen for 16 h. The resin

was collected by filtration and washed sequentially with DMF (3 ꢁ

15 mL), ethanol (2 ꢁ 15 mL), and DCM (2 ꢁ 15 mL). The resin was

incubated three times with a mixture of TFA (10%) and TIS (1%) in

DCM for 1 h each. The filtrates were collected, combined, and concentrated

under a stream of nitrogen. Cold ether (20 mL) was added to the

remaining residue, and the resulting precipitate was collected by centrifuga-

tion. This precipitate was washed with cold ether (2 ꢁ 15 mL) and

lyophilized. Purification by RP-HPLC afforded the title compound as a

N¹-(2-Hydroxy-5-nitro)benzoyl-N⁵-(2-fluoro-1-iminoethyl)-

L-ornithine Amide TFA (16). To a solution of N¹-(2-hydroxy-5-

3

nitro)benzoyl-^L-ornithine TFA (13.7 mg, 0.033 mmol) in dry methanol

3

(1 mL) was added ethyl 2-fluoroacetimidate (9.4 mg, 0.067 mmol) and

Cs₂CO₃(16.3 mg, 0.049 mmol). The reaction was stirred overnight at rt,

quenched with TFA, diluted with H₂O, and purified by RP-HPLC to

yield the title compound as a white powder (14.7 mg, 95%). ¹H NMR

(400 MHz, D₂O) δ (ppm): 8.45 (d, J = 3.2 Hz, 1H), 8.05 (dd, J = 2.8, 9.2

Hz, 1H), 6.88 (d, J = 9.2 Hz, 1H), 5.11 (d, J_HF= 44.8 Hz, 2H), 4.39 (dd,

J = 5.6, 8.4 Hz, 1H), 3.27 (t, J = 6.8 Hz, 2H), 1.95ꢀ1.61 (m, 4H). ¹³C

NMR (100 MHz, D₂O) δ (ppm): 175.95, 166.95, 162.55 (d, J_CF

20.5 Hz), 139.85, 128.96, 125.98, 117.69, 117.13, 77.53 (d, J_CF

=

177.4 Hz), 53.32, 41.32, 28.31, 23.05. HRMS (ESI) m/z calculated for

+

1

[C₁₄H₁₈FN₅O₅], 356.1370; observed, 356.1359.

hygroscopic white powder upon lyophilization. H NMR (400 MHz,

N-α-(2-Carboxyl)benzoyl-N⁵-(2-fluoro-1-iminoethyl)-L-or-

nithine Amide (o-F-amidine) (17). Rink Amide AM resin (300 mg,

0.186 mmol) was suspended in 20% piperidine (in DMF) (5 mL) and

gently rocked at rt for 20 min. The suspension was filtered and treated

with 20% piperidine once more. After 20 min, the resin was filtered and

washed with DMF (3 ꢁ 5 mL). Fmoc-Orn(Boc)-OH (338 mg, 0.74

mmol), HBTU (282 mg, 0.74 mmol), and HOBt (114 mg, 0.74 mmol)

were dissolved in DMF (5 mL), and N-methylmorpholine (0.164 mL,

1.49 mmol) was added. After 10 min, this solution was added to the resin

followed by gentle rocking. After 3 h, the resin was filtered, washed with

DMF (3 ꢁ 5 mL), and treated with 20% piperidine (2 ꢁ 5 mL). The

resin was then filtered, washed with DMF (3 ꢁ 5 mL), and resuspended

in DMF (5 mL). After the addition of phthalic anhydride (110 mg,

0.74 mmol) and triethylamine (0.207 mL, 1.49 mmol), the mixture was

rocked at rt for 16 h. The resin was collected by filtration, washed with

DMF (3 ꢁ 5 mL), ethanol (2 ꢁ 5 mL), and dichloromethane (2 ꢁ

5 mL), dried under reduced pressure, and treated twice with a mixture of

TFA/TIS/H₂O (95/2.5/2.5) for 1 h. The filtrates were collected, com-

bined, and concentrated under a stream of nitrogen. Cold ether (20 mL)

was added to the remaining residue, and the resulting precipitate was

collected by centrifugation, washed twice with ether (10 mL), and lyophilized.

This precipitate (15 mg, 0.038 mmol), ethyl chloroacetimidate hydro-

chloride (10.7 mg, 0.076 mmol), and cesium carbonate (18.5 mg, 0.057

mmol) were dissolved in dry methanol (1 mL) and stirred at rt for 16 h.

The reaction was quenched with TFA, diluted with H₂O, and purified by

RP-HPLC to afford the title compound as a hygroscopic white powder

CD₃OD) δ (ppm): 7.89 (d, J = 8.4 Hz, 2H), 7.65ꢀ7.55 (m, 3H), 5.27

(d, J = 45.4 Hz, 2H), 3.82 (m, 1H), 3.10 (m, 2H, obtained in D₂O, 400

MHz), 1.81ꢀ1.59 (m, 4H). ¹³C NMR (100 MHz, CD₃OD) δ (ppm):

175.68, 164.28 (d, J_CF= 19.5 Hz), 141.86, 133.93, 130.24, 128.17, 78.96

(d, J_CF= 177.9 Hz), 57.13, 42.78, 31.56, 24.73. HRMS (ESI) m/z

calculated for [C₁₃H₂₀FN₄O₃S⁺], 331.1240; observed, 331.1237.

N¹-(4-Carbonyl)imidazole-N⁵-(2-fluoro-1-iminoethyl)-L-

ornithine Amide TFA (14). Rink AM Amide resin (300 mg, 0.2 mmol)

3

was treated twice with 5 mL of 20% piperidine (in DMF) for 20 min and

subsequently washed with dry DMF (3 ꢁ 5 mL). Fmoc-Orn(Dde)-OH

(414 mg, 0.8 mmol), HBTU (303 mg, 0.8 mmol), HOBt (122 mg,

0.8 mmol), and triethylamine (0.22 mL, 1.6 mmol) were dissolved in dry

DMF (5 mL) and allowed to react for 10 min before being added to the

resin. The reaction mixture was rocked at rt for 3 h before the resin was

filtered, washed with DMF (3 ꢁ 5 mL), treated twice with 5 mL of 20%

piperidine (in DMF) for 20 min, and washed with dry DMF (3 ꢁ 5 mL).

1H-imidazole-4-carbonyl chloride HCl (134 mg, 0.8 mmol) and triethy-

3

lamine (0.22 mL, 1.6 mmol) were dissolved in dry DMF (5 mL) and added

to the resin. After rocking overnight at rt, the resin was collected by filtration,

washed with DMF (3 ꢁ 5 mL), and treated twice with 2% hydrazine (in

DMF) (5 mL) for 1 h to remove the Dde-protecting group. A solution of

ethyl 2-fluoroacetimidate HCl (113 mg, 0.8 mmol) and triethylamine

3

(0.22 mL, 1.6 mmol) in DMF (5 mL) was added to the resin, and the

mixture was rocked at rt. After 16 h, the resin was collected by filtration,

washed with DMF (3 ꢁ 5 mL), ethanol (3 ꢁ 5 mL), and DCM (3 ꢁ

5 mL), then dried under reduced pressure for 1 h. The dry resin was treated

three times (1 h each) with 5 mL of TFA/TIS/H₂O (95/2.5/2.5). The

filtrates were collected, combined, and concentrated. Cold ether (25 mL)

was added to the remaining residue, and the resulting precipitate was

collected by centrifugation and purified by RP-HPLC to yield the title

compound as a white powder (49 mg, 48%). ¹H NMR (400 MHz, D₂O) δ

(ppm): 8.65 (s, 1H), 7.92 (s, 1H), 5.10 (d, J_HF= 44.8 Hz, 2H), 4.33 (dd, J =

5.6, 8.4 Hz, 1H), 3.24 (t, J = 6.8 Hz, 2H), 1.89ꢀ1.58 (m, 4H). ¹³C NMR

(100 MHz, D₂O) δ (ppm): 175.77, 162.54 (d, J_CF= 19.1 Hz), 158.74,

135.42, 126.44, 123.44, 120.75, 117.64, 114.74, 77.51 (d, J_CF= 177.4 Hz),

53.38, 41.32, 27.95, 23.17. HRMS (ESI) m/zcalculated for [C₁₁H₁₈FN₆O₂⁺],

285.1475; observed, 285.1468.

1

(15 mg, 23%). H NMR (400 MHz, D₂O) δ (ppm): 7.83 (m, 1H),

7.56ꢀ7.45 (m, 2H), 7.35 (m, 1H), 4.34 (m, 1H), 3.25 (m, 2H), 1.92ꢀ

1.55 (m, 4H). ¹³C NMR (100 MHz, D₂O) δ (ppm): 176.04, 173.00,

170.28, 162.79, 135.99, 135.10, 132.56, 131.51, 130.41, 128.62, 127.39,

53.32, 41.87, 39.05, 27.89, 23.04. HRMS (ESI) m/z calculated for

+

[C₁₅H₂₀ClN₄O₄], 355.1173; observed, 355.1172.

N-α-(2-Carboxyl)benzoyl-N⁵-(2-chloro-1-iminoethyl)-L-

ornithine Amide (o-Cl-amidine) (18). Rink Amide AM resin (300

mg, 0.186 mmol) was suspended in 20% piperidine (in DMF) (5 mL)

and gently rocked at rt for 20 min. The suspension was filtered and

treated with 20% piperidine once more. After 20 min, the resin was

filtered and washed with DMF (3 ꢁ 5 mL). Fmoc-Orn(Boc)-OH (338 mg,

0.74 mmol), HBTU (282 mg, 0.74 mmol), and HOBt (114 mg, 0.74 mmol)