Rickettsia prowazekii methionine aminopeptidase as a promising target for the development of antibacterial agents

Article abstract of DOI:10.1016/j.bmc.2016.11.013

Methionine aminopeptidase (MetAP) is a class of ubiquitous enzymes essential for the survival of numerous bacterial species. These enzymes are responsible for the cleavage of N-terminal formyl-methionine initiators from nascent proteins to initiate post-translational modifications that are often essential to proper protein function. Thus, inhibition of MetAP activity has been implicated as a novel antibacterial target. We tested this idea in the present study by targeting the MetAP enzyme in the obligate intracellular pathogen Rickettsia prowazekii. We first identified potent RpMetAP inhibitory species by employing an in vitro enzymatic activity assay. The molecular docking program AutoDock was then utilized to compare published crystal structures of inhibited MetAP species to docked poses of RpMetAP. Based on these in silico and in vitro screens, a subset of 17 compounds was tested for inhibition of R. prowazekii growth in a pulmonary vascular endothelial cell (EC) culture infection model system. All compounds were tested over concentration ranges that were determined to be non-toxic to the ECs and 8 of the 17 compounds displayed substantial inhibition of R. prowazekii growth. These data highlight the therapeutic potential for inhibiting RpMetAP as a novel antimicrobial strategy and set the stage for future studies in pre-clinical animal models of infection.

Full text of DOI:10.1016/j.bmc.2016.11.013

Accepted Manuscript
Rickettsia prowazekii Methionine Aminopeptidase as a Promising Target for the
Development of Antibacterial Agents
Travis R. Helgren, Congling Chen, Phumvadee Wangtrakuldee, Thomas E.
Edwards, Bart L. Staker, Jan Abendroth, Banumathi Sankaran, Nicole A.
Housley, Peter J. Myler, Jonathon P. Audia, James R. Horn, Timothy J. Hagen
PII:
S0968-0896(16)31158-0
DOI:
http://dx.doi.org/10.1016/j.bmc.2016.11.013
BMC 13380
Reference:
Bioorganic & Medicinal Chemistry
To appear in:
Received Date:
Revised Date:
Accepted Date:
9 September 2016
6 November 2016
8 November 2016
Please cite this article as: Helgren, T.R., Chen, C., Wangtrakuldee, P., Edwards, T.E., Staker, B.L., Abendroth, J.,
Sankaran, B., Housley, N.A., Myler, P.J., Audia, J.P., Horn, J.R., Hagen, T.J., Rickettsia prowazekii Methionine
Aminopeptidase as a Promising Target for the Development of Antibacterial Agents, Bioorganic & Medicinal
Chemistry (2016), doi: http://dx.doi.org/10.1016/j.bmc.2016.11.013
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Rickettsia prowazekii Methionine Aminopeptidase as a
Promising Target for the Development of Antibacterial
Agents
Travis R. Helgren†, Congling Chen†, Phumvadee Wangtrakuldee†, Thomas E. Edwards‡, Bart L. Staker₴,Δ,
Jan Abendroth‡, Banumathi SankaranⱵ, Nicole A. Housleyǁ, Peter J. Myler₴,Δ,₸, Jonathon P. Audiaǁ, James
R. Horn†, Timothy J. Hagen†*
† Department of Chemistry and Biochemistry, Northern Illinois University, 1425 W. Lincoln Hwy, DeKalb,
IL 60115, USA
‡ Beryllium LLC, 7869 NE Day Road West Bainbridge Island, WA 98110, USA
₴ Center for Infectious Disease Research, formerly Seattle Biomedical Research Institute, 307 Westlake
Avenue N., Seattle, WA 98109, USA
Δ Seattle Structural Genomics Center for Infectious Disease (SSGCID), Seattle, WA, USA
₸ Department of Global Health and Department of Biomedical Informatics and Medical Education,
University of Washington, Seattle, WA 98195, USA
Ⱶ Berkeley Center for Structural Biology, Ernest Orlando Lawrence Berkeley National Laboratory,
Berkeley, CA 94720, USA
ǁ Department of Microbiology and Immunology and The Center for Lung Biology, University of South
Alabama College of Medicine, Laboratory of Infectious Diseases, 307 North University Blvd, Mobile, AL
36688, USA
KEYWORDS: MetAP, methionine aminopeptidase, inhibition, metalloenzyme, epidemic typhus, Rickettsia
prowazekii, lung endothelial cells
ABSTRACT: Methionine aminopeptidase (MetAP) is a class of ubiquitous enzymes essential for the
survival of numerous bacterial species. These enzymes are responsible for the cleavage of N-terminal
formyl-methionine initiators from nascent proteins to initiate post-translational modifications that are
often essential to proper protein function. Thus, inhibition of MetAP activity has been implicated as a
novel antibacterial target. We tested this idea in the present study by targeting the MetAP enzyme in
the obligate intracellular pathogen Rickettsia prowazekii. We first identified potent RpMetAP inhibitory
species by employing an in vitro enzymatic activity assay. The molecular docking program AutoDock was
then utilized to compare published crystal structures of inhibited MetAP species to docked poses of
RpMetAP. Based on these in silico and in vitro screens, a subset of 23 compounds was tested for
inhibition of R. prowazekii growth in a pulmonary vascular endothelial cell (EC) culture infection model
system. All compounds were tested over concentration ranges that were determined to be non-toxic to
the ECs and 10 of the 23 compounds displayed substantial inhibition of R. prowazekii growth. These data
highlight the therapeutic potential for inhibiting RpMetAP as a novel antimicrobial strategy and set the
stage for future studies in pre-clinical animal models of infection.

1. Introduction
Rickettsia prowazekii is an obligate intracytoplasmic pathogen and the causative agent of
epidemic typhus fever in humans.1,2 Transmission of R. prowazekii is mediated by the human body louse
and typically associates with deteriorated social conditions resulting in population crowding coupled
with compromised sanitation and hygiene.3 These Gram-negative bacteria are vectored via growth in
louse gut epithelium and R. prowazekii is highly stabile when shed in feces, which facilitates the
potential for aerosol transmission. R. prowazekii is highly virulent,4,5 and populations with a low level of
immunity are particularly susceptible to this disease.6 Patients with typhus exhibit signs and symptoms
such as fever, malaise, myalgia, rash, and confusion that are common to numerous diseases, resulting in
difficulty in establishing a timely diagnosis.7 R. prowazekii is designated a select agent pathogen and
classified by the Center for Disease Control and Prevention (CDC) as a category B pathogen and potential
bioterrorism agent.8
Upon feeding on a human host, the R. prowazekii-infected louse defecates at the feeding site
and host scratching results in sub-dermal inoculation. Post-inoculation, the rickettsiae ultimately enter
the circulation and primarily infect vascular endothelial and immune cells. The rickettsiae attach to a
host cell, induce phagocytosis, and rapidly escape the phagolysosome in a phospholipase A2-dependent
manner to grow free in the host cell cytoplasm. R. prowazekii typically grows to high numbers in the
host cell cytoplasm and are released via host cell lysis to repeat the infection cycle2. The obligate
intracellular nature of R. prowazekii significantly limits the repertoire of clinically effective antibiotic
treatments to doxycycline, tetracycline and chloramphenicol. Considering the limited number of
antibiotics available to treat rickettsioses and reports of R. prowazekii strains resistant to both
tetracycline and chloramphenicol,9 the identification of novel targets for the development of anti-
rickettsial therapeutics is needed.

Nascent bacterial protein synthesis is initiated with an N-terminal formyl-methionine residue
that may be subsequently removed by the enzymatic activity of methionine aminopeptidases (MetAP)
to facilitate proper protein function making this ubiquitous enzyme essential to many bacteria.10
Previous studies have implicated these enzymes as potentially useful targets for the discovery of novel
antibiotics. For example, a nanomolar (IC50) inhibitor of Burkholderia pseudomallei MetAP1 was found to
exhibit growth inhibition of B. thailandensis.11 In the present study, the inhibition activities of three
distinct chemical classes of compounds (furoic acids, 1,2,4-triazoles and quinolinols) were evaluated
against R. prowazekii MetAP1. Structure-activity relationship (SAR) analyses, along with molecular
docking, were used to explore potential interactions of these inhibitors within the RpMetAP active site.
Finally a subset of 23 inhibitors identified in these in silico and in vitro screens were tested for inhibition
of R. prowazekii growth in a cell culture model infection system.
2. Results and Discussion
2.1 Crystal Structures of RpMetAp
The crystal structure of R. prowazekii MetAP was solved in two different crystal forms, to 1.7 Å
and 2.0 Å resolution. Seven different constructs were designed using structure-based alignments to
Escherichia coli MetAP (EcMetAP). Constructs varied by small truncations at both the N and C terminus
of the RpMetAP sequence of 259 residues (see Supplemental Information, Table S2) as well as 2 full-
length constructs with different N-terminal His-tagged extensions, where A and B refer to cleavable and
non-cleavable His-tag constructs, respectively. These seven constructs were processed through the
Seattle Structural Genomics Center for Infectious Disease (SSGCID) high throughput protein production
pipeline concurrently. Constructs A1(1-259), A2(1-247), and A3(1-249) expressed insoluble protein.
Constructs B1(1-259), A4(3-259), and A6(3-249) expressed soluble protein, were purified and subjected

to crystallization trials. Construct A5(3-247) failed at initial cloning, likely due to technical reasons, and
was not pursued further. Given that an N-terminal methionine is added to each sequence during
cloning, the only difference between the amino termini of the sequences A1(1-259) and A4(3-259) as
well as A3(1-249) and A6(3-249) is a single threonine residue at position 2 in the wild type sequence.
This deletion of one residue led to the production of soluble protein in the A4 and A6 constructs, as
compared to the insoluble A1 and A3 constructs.
Constructs A4(residues 3-259) and A6(residues 3-249) both crystallized and diffracted X-rays to
high resolution. The structure of construct A4(3-259) bound to methionine and the apo structure of
construct A6(3-249) were refined to 1.7 Å and 2.0 Å, respectively, and deposited into the Protein Data
Bank (PDB) as 3MX6 and 3MR1. The A4 structure contained two copies of the protein in the asymmetric
unit and the A6 construct contained four copies of the protein in the asymmetric unit. The two crystal
structures are highly similar with an average root mean squared difference (r.m.s.d.) of only 0.34 Å for
C atoms. The C-terminal portions of both the A4 and A6 proteins are present in both crystal structures
α
with A4 being ordered through residue Tyr259 and the A6 construct being ordered through Leu249; we
note that the numbering scheme used for the crystal structures does not include the N-terminal
methionine and thus the above residues are labeled Tyr258 and Leu248.
The overall structural fold of RpMetAP was highly similar to MetAP enzymes from other species
ranging from bacteria to human. The root mean squared difference (r.m.s.d.) of the core 250 residue
'pita bread' fold between the R. prowazekii MetAP as compared to Pseudomonas aeruginosa (PDB:
4FO7), E. coli (PDB: 1XNZ12) and Homo sapiens (PDB: 4U1B13) was 0.985 Å, 0.711 Å, and 0.865 Å,
respectively. For both the A4 and A6 RpMetAP structures, no divalent metal ion was included in the
crystallization conditions. Two metal ions were clearly present in the electron density. These features
were modeled as Mn(II) ions based off similar structures in the literature as well as proper refinement
with B-factors similar to those of surrounding protein atoms, octahedral geometry and mean bond

distances of 2.1-2.2 Å which are appropriate for Mn(II).14 The CheckMyMetal server15 was largely in
agreement on metal ion selection, although it suggested copper or cobalt as possible alternatives;
however, neither of these metal ions refined with B-factors equivalent to those of the surrounding
residues. For the A4 structure, an additional electron density feature was present, which modeled well
as methionine, a byproduct of the enzymatic activity (PDB: 3MX6, Figure 1). In contrast, the A6 construct
crystallized in the presence of phosphate buffer, and a phosphate appears bound to both metal ions; the
phosphate ion overlays well with the carboxylate of the bound methionine residue. This phosphate
orientation is likely biologically relevant (i.e. reflects the binding mode observed for methionine, Figure
1B) since the enzymatic reaction would proceed through a sp3 hybridized transition state and similar
interactions with His78, His176, and Glu202 (His77, His175 and Glu201, respectively in Figure 1) would
be expected.
Figure 1: RpMetAP1 (PDB: 3MX6) with bound methionine. A: RpMetAP represented as a cartoon. B:
Zoom-in of the RpMetAP active site. Purple spheres designate Mn(II) cofactors and the green residue is
the bound methionine.

2.2 Compound Syntheses
Various classes of inhibitors have been determined to inhibit MetAP activity, including furoic
acids, 1,2,4-traizoles, and quinolinols; compounds of these scaffolds were therefore utilized for our
study.16 It is noteworthy that some compounds based upon the quinolinol scaffold have been identified
as pan-assay interference compounds (PAINS)17 due to both their promiscuous inhibition of
metalloenzymes and their ability to form reactive methides. As this is early stage, proof of concept
research, this information can useful in the design of next generation compounds that are devoid of
such characteristics. Importantly, this subset of compounds (24-26) do exhibit selective toxicity towards
R. prowazekii growing within mammalian host cells (see below) and we have therefore included a
discussion of their observed activity and SAR.
Regarding our routes to obtain a screening test set, compounds (1 – 11) bearing the furoic acid
scaffolding were purchased from a commercial source (Sigma Aldrich, St. Louis, MO). The synthesis of
the triazole species was previously reported 11 (Scheme 1). Briefly, the compounds were synthesized by
base promoted addition of substituted benzyl bromides to the 5-amino-4H-1,2,4-triazole-3-thiol starting
material. The reactions were carried out under aqueous conditions and afforded product in fair yield (47
– 92%). Regarding the synthesis of oxine Mannich derivatives, the starting materials were suspended in
EtOH and heated overnight to afford the desired products in good yield (43 – 95%) (Scheme 2).11

Scheme 1
Scheme 2
2.3 Enzymatic Activity Assay
2.3.1 Assay Procedure
Following the synthesis of compounds designed to evaluate the SAR of published bacterial
MetAP inhibitors, an enzymatic activity assay was used to evaluate inhibition potencies. The assay
employed was adopted from published reports,11,18 where MetAP activity is monitored as a function of
product formation. Utilizing methionyl aminomethylcoumarin (Met-AMC) as the substrate, enzymatic
turnover will cleave the peptide bond producing free AMC, a fluorescent compound with emission at
460 nm.

2.3.2 Activity of Furoic Acid Derivatives
R2
R3
R4
R1
O
X
OH
Table 1: Activity of furoic acid species against RpMetAP1
Compound
(1)
(2)
X
S
R1
H
H
H
H
H
CF3
H
Me
H
Cl
Cl
R2
H
NO2
CF3
H
H
H
H
H
R3
H
H
H
H
Me
H
Br
H
H
H
R4
H
H
H
H
H
H
H
H
H
H
Cl
IC50 (µM)
74 ± 7
77 ± 4
46 ± 2
37 ± 3
57 ± 2
0.5 ± 0.1
60 ± 5
0.9 ± 0.1
26 ± 1
Hill Slope
1.2 ± 0.2
1.0 ± 0.1
1.0 ± 0.1
1.0 ± 0.1
1.0 ± 0.1
2.1 ± 0.1
1.0 ± 0.1
1.3 ± 0.1
1.0 ± 0.1
1.2 ± 0.1
0.6 ± 0.1
O
O
O
O
O
O
O
O
O
O
(3)
(4)
(5)
(6)
(7)
(8)
(9)
Me
H
H
(10)
(11)
0.6 ± 0.1
0.6 ± 0.1
H
Furoic acid based inhibitors of bacterial MetAPs are among the most potent reported in the
current literature.16 Species of this general structure only exhibit potent inhibition of MetAPs containing
Mn(II) cofactors, with essentially no activity observed for MetAPs possessing Fe(II), Co(II) or Ni(II)
metals.16 Although this class of inhibitors have demonstrated activity against a number of bacterial
MetAPs, including Mycobacterium tuberculosis,18-19 B. pseudomallei,11 E. coli.12, 20 and Acinetobacter
baumannii,21 there are currently no reports detailing the selectivity of bacterial versus human MetAPs.
Published reports detailing the inhibition of bacterial MetAPs by furoic acids and thiophenic
acids11-12, 18-21 have suggested the most potent compounds contain small substituents at the ortho (R1)
position. Generally, the compounds bind via bidentate coordination through the carboxylate to one of
the active site metals, with one of the oxygen atoms also exhibiting coordination to the other cofactor

(PDB: 3MAT).12 This binding mode results in a competitive mechanism of inhibition where the catalytic
site is blocked by inhibitor coordination, effectively eliminating access by substrate.
In the present study, compound (6) utilizes a trifluoromethyl substituent at the ortho (R1)
position and was observed to be the most active compound of this chemical series (IC50 = 0.5 μM) (Table
1). Such activity was essentially mirrored by other compounds utilizing substitution at R1, namely (8),
(10) and (11), with inhibitory values of 0.9, 0.6, and 0.6 μM, respectively. Only one species containing
meta (R2) substitution, 3-methyl derivative (9), exhibited greater activity to that of unsubstituted 5-
phenyl-2-furoic acid (4), with inhibitory values of 26 and 37 μM, respectively. All other 3 and 4
substituted aryl furoic acids were found to only weakly inhibit enzymatic activity. As with other
reports,20d 5-phenyl-2-thiophenic acid (1) was found to exhibit weaker activity than that of the
corresponding furan (IC50 = 74 and 37 μM, respectively). It is noteworthy that compound (6) exhibited
the only Hill slope outlier (2.1 ± 0.1) and also exhibited the most potent inhibitory activity. As the
enzyme concentration was fixed at 1 µM, this may reflect the minimum inhibitory concentration
detectable by this assay.

2.3.3 Activity of Triazole Derivatives
Table 2: Activity of triazole species against RpMetAP1
Compound
(12)
(13)
R1
H
H
H
Cl
H
Cl
H
H
R2
H
Me
F
H
Cl
Cl
iPr
tBu
IC50 (µM)
6.6 ± 1.0
23 ± 3
28 ± 6
15 ± 2
30 ± 2
7.7 ± 0.9
14 ± 1
14 ± 1
Hill Slope
1.1 ± 0.2
0.9 ± 0.1
1.0 ± 0.2
1.2 ± 0.1
1.6 ± 0.6
1.3 ± 0.3
0.6 ± 0.2
1.5 ± 0.4
(14)
(15)
(16)
(17)
(18)
(19)
Species bearing the 1,2,4-triazole scaffold have been shown to inhibit MetAPs from both
bacterial11, 18-19, 21-22 and human23 sources. These compounds have been shown to most effectively inhibit
MetAPs utilizing Co(II) or Ni(II) cofactors in vitro, with weaker activity observed for Mn(II) cofactors.16
Interestingly, MetAP species bearing Fe(II) cofactors exhibit essentially no reduction in enzymatic
turnover in the presence of triazole inhibitors.
Crystal structures detailing the binding of 1,2,4-triazoles to bacterial MetAP targets exist (PDB:
3IU8, 3IU9),19 where compounds bind via coordination of the 1N and 2N atoms of the triazole ring to the
divalent metal cofactors embedded within the active site of MetAPs. Additional binding contacts
generally include π-π stacking with aromatic residues (phenylalanine, histidine, and tyrosine) occupying
adjacent space to the substrate-binding pocket. A similar binding mechanism is observed for HsMetAP2,
with coordination to the metal cofactors by the triazole ring and additional π-π stacking to aromatic
residues (PDB: 2OAZ).23

Concerning the inhibition of RpMetAP by 1,2,4-triazoles, we determined the most potent
compounds utilize electron withdrawing groups (Cl or F) at the R1 and R2 position; other reports detailing
the inhibition of bacterial MetAPs by this chemical class have confirmed this observation (Table 2).18, 21
However, the most potent compound containing benzyl thioethers employed in our study was found to
be unsubstituted derivative (12) with an IC50 value of 6.6 µM, although 2,4-dichloro derivative (15)
exhibited comparable activity (IC50 = 7.7 µM).
2.3.4 Activity of Quinolinol Derivatives
Table 3: Activity of quinolinol species against RpMetAP1
Compound
(20)
(21)
R1
CH3
H
H
H
R2
H
H
Cl
NO2
R3
H
H
H
H
IC50 (µM) Hill Slope
1.4 ± 0.1 1.1 ± 0.1
1.0 ± 0.3 1.2 ± 0.1
0.9 ± 0.1 1.9 ± 0.2
1.3 ± 0.7 1.7 ± 0.5
(22)
(23)
(24)
(25)
(26)
H
H
H
Cl
2.5 ± 0.1 1.5 ± 0.1
NO2
NO2
73 ± 38
16 ± 4
1.2 ± 0.4
1.0 ± 0.1
Compounds composed of the 8-quinolinol scaffold have been screened against MetAPs from B.
pseudomallei,11 M. tuberculosis,24 E. coli,25 Staphylococcus aureus,25c and against those of human
origin.25c The chelating ability of 8-quinolinol derivatives affords an interesting mechanism of binding for
metalloenzymes as demonstrated by the crystal structure of an 8-quinolinol derivative bound to E. coli

MetAP (PDB: 2BB7),25b although the inhibitor is the related compound, 8-quinolinyl-
methanesulfonamide. The crystal structure shows the inhibitor binding to a tertiary Mn(II) cofactor
found within the active site, and the inhibitor potency was found to be dependent upon the
concentration of MnCl2 added to the assay buffer.25b Given the structure-activity relationship previously
discovered for the inhibition of B. pseudomallei,11 we felt it was appropriate to screen a class of
aminoalkylated quinolinol analogues against R. prowazekii MetAP.
In our previous study investigating the inhibition of BpMetAP1,11 it was found that quinolinol
derivatives utilizing NO2 groups para (R2) to the hydroxyl group exhibited the most potent activity.
Species bearing Cl groups at the same position were essentially inactive. In this study, the opposite
appears to be true, with all Cl derivatives being more potent inhibitors than the corresponding NO2
derivatives. However, the metals employed as cofactors for this study were Mn(II), while Co(II) was used
previously in the study of B. pseudomallei MetAP inhibitors. This may explain the observed difference in
activity regarding Cl or NO2 substitution, as observed activities have been documented to be highly
dependent upon the identity of the metallic cofactors.25b
When compared to aminoalkylated derivatives, unsubstituted quinolinols exhibited superior
activity. For example, the most active species was found to be 5-chloro-8-quinolinol (22), with an IC50
value of 0.9 µM. All other derivatives (2-methyl-8-quinolinol (20), 8-quinolinol (21), and nitroxoline (23))
exhibited comparable activity to that of (22). Concerning aminoalkylated derivatives, all were inferior
inhibitors than the corresponding quinolinol fragments.
2.4 Molecular Docking
To discern potential binding interactions and to aid in the development of new MetAP
inhibitors, test compounds from each class were docked into the crystal structure of R. prowazekii

MetAP1 (PDB: 3MX6) and the docking poses and scores were compared to the observed inhibitory data
and currently available crystal structures. The docking studies were performed using the open-source
program AutoDock.26 The numerous crystal structures of E. coli MetAP including substituted 5-aryl-2-
furoic acids (PDB: 1XNZ,12 2EVM,20e 2Q92,27 2Q93,27 2Q94,27 2Q95,27 2Q96,27 and 3IU719) served as
models to validate the docking output from AutoDock (see Supplemental Information). A sequence
alignment was performed to discern the conservation between EcMetAP1 and RpMetAP1 (PDB: 1XNZ12
and 3MX6), which are 53% identical (Figure 2). Based upon the three-dimensional structure, the active
site, including the five residues responsible for metal cofactor binding and residues composing the active
site surface, is highly conserved.
Figure 2: Sequence alignment of EcMetAP1 (PDB: 1XNZ)12 and RpMetAP1 (PDB: 3MX6) shaded according
to alignment (dark = identity; light = similarity; no shading = non-conserved). Alignment reveals 53%
identity between the proteins. Residues involved in cofactor binding are marked as X
The crystal structure of the EcMetAP/furoic acid (10) complex reveals contacts responsible for
potent inhibition (Figure 3B). The most important interactions between the ligand and receptor involve
coordination to both the metal cofactors via the carboxylate, π- π interactions with Tyr62 and
hydrophobic interactions with His63, Tyr65, His79, Phe177, His178 and Trp221 (Table 4). Additionally,

the furoic acid inhibitor exhibits a dihedral angle of 40.6o between the furan and aryl ring. As previously
stated, this rotated bi-aryl ring system is necessary for potent inhibition, maximizing interactions with
Tyr62. Thus, the incorporation of large substituents at the ortho position affords hindered rotation
about the bi-aryl axis, resulting in an increase in the dihedral angle, which serves as the basis for the
observed increase in activity for this substitution pattern.
Table 4: Comparison of binding interactions for (10) in crystal and docked structures
Distancea,b
EcMetAP1 Actual
RpMetAP1 Docked
Mn(II) Chelation
Mn(II) Coordination
π – π Stacking (Tyr62)
Hydrophobic (His63)
Hydrophobic (Tyr65)
Hydrophobic (His79)
Hydrophobic (Phe177)
Hydrophobic (His178)
Hydrophobic (Trp221)
2.2, 2.3
2.0
4.0
3.5
3.6
3.6
4.0
3.3
3.4
1.7, 2.1
1.8
3.1
6.9c
3.5
3.9
3.3
3.6
3.1
Angle
Mn(II) – OL – Mn(II) Angle
107o
41o
155o
18o
(10) Dihedral
a. Distances correspond to closest observed interaction and are reported in Å
b. Residue numbering corresponds to EcMetAP1 (PDB: 1XNZ)12
c. This is a non-conserved residue and exists as Lys in RpMetAP
The series of furoic acid inhibitors (Table 4) was screened in silico against RpMetAP1 to optimize
the docking method. Because AutoDock does not automatically assign charges to metal ions, the charge
of the Mn(II) cofactors found within the RpMetAP1 structure (3MX6) was manually set at 0.5 elementary
charge units (see Supplemental Information). Additionally, compound (10) was previously crystallized
with EcMetAP1 as a bound structure (1XNZ),12 allowing for the comparison of the predicted docking
pose and crystal structures for two highly conserved MetAPs of bacterial origin.

The highest scored docking pose of (10) reveals a binding mode similar to that observed in the
crystal structure of (10) bound to EcMetAP1 (Figure 3). As demonstrated in Table 4, the differences in
protein-ligand interactions between the crystal and docked structures ranged from 0.1 – 0.9 Å.
However, the most obvious difference between the docked and crystal structures is the dihedral angle
between the ring systems of the furoic acid inhibitor. The dihedral angle for the docked structure was
significantly reduced, at 18o, compared to that of the actual pose (see Supplemental Information).
Examination of the Mn-O-Mn bond angle and Mn-O bond distances demonstrates the closer spatial
orientation of the inhibitor in the docked pose as compared to that of the actual crystal structure. The
large Mn-O-Mn bond angle (155o) and shortening of the chelation distance between the Mn and
carboxylate for the docked pose revealed stronger predicted interactions than actually observed (Table
4). The chloride substituent was also found to point into the middle of the substrate binding pocket,
rather than coordinate to residues lining the active site as seen in the crystal structure.

Figure 3: Comparison of docked and observed binding interactions of (10). A: Docked pose of (10) with
RpMetAP (PDB: 3MX6). B: Crystal structure of (10) bound to EcMetAP (PDB: 1XNZ).12 Note that residue
numbering differs by two between the two MetAP species (Ex: RpMetAP Tyr60 and EcMetAP Tyr62 are
equivalent).
2.5 Comparison of HsMetAPs and RpMetAP1
To develop selective inhibitors that target bacterial MetAPs, a detailed understanding of the
differences between bacterial and human MetAP structures is necessary. The sequence alignment of
RpMetAP1 and HsMetAP1 indicates the enzymes are 43% identical, with all 5 residues responsible for
cofactor binding being conserved (Figure 4, bottom). Active site residues were found to be highly
conserved, with only a few significant non-conserved residues lining the entrance to the binding pocket
(Lys61Tyr, Asp164Ser, Thr166Cys, Ser179Asn, and Leu213Thr, indicating the residue change from Hs to
Rp) as demonstrated by the colored surface model shown in Figure 4. Selective inhibitors will likely
utilize binding to the metal cofactors as an anchor for proper binding orientation while targeting
interactions with the non-conserved residues. The closest non-conserved residue near the active site of
RpMetAP1 is Lys61. Inhibitors have been found to interact with this residue position for EcMetAP112

(histidine for the E. coli isoform (His63), see Figure 3) and selective inhibition of RpMetAP1 may result
from designing inhibitors targeting Lys61. Although docking studies of (10) with RpMetAP1 suggests
Lys61 lies too far from the active site to allow favorable interactions with bound inhibitors (Table 4),
new molecular species could be designed to capitalize upon this lack of conservation and yield selective
inhibition.
Of the species screened for RpMetAP1 inhibitory activity (1 – 26), compounds bearing the
quinolinol scaffold have been evaluated against both HsMetAP1 and HsMetAP2, previously.28 Because
HsMetAP1 was determined to be most active with Co(II) cofactors, this metal was employed in the
screening of quinolinol compounds in the published report.28 Additionally, Mn(II) has been suggested as
the native cofactor for HsMetAp2 and this metal was used in the corresponding assays.28-29 In the
present study, the compounds exhibit selective inhibition of RpMetAP1 over HsMetAp1, with the
compounds generally being inactive against HsMetAp1 (Table 5). This result is interesting, as quinolinol
species have been demonstrated to inhibit MetAP species utilizing Co(II) cofactors over species utilizing
Mn(II) cofactors.16 The quinolinol compounds were not as selective against HsMetAP2, with the
observed activities being comparable to RpMetAP1 (Table 5). However, HsMetAP2 inhibition has been
suggested as a treatment for tumor growth abatement (antiangiogenic properties)30 and a lack of
specificity regarding these two targets may not be of striking clinical significance.

Figure 4: Comparison of HsMetAP1 and RpMetAP1 sequences. A: Active site of RpMetAP1 colored
according to sequence alignment (green = identity; yellow = similarity; magenta = non-conserved). B:
Surface diagram of RpMetAp1 colored according to sequence alignment (green = identity; yellow =
similarity; magenta = non-conserved). The blue ligand is a bound methionine and orange spheres are
manganese cofactors. C: Sequence alignment of RpMetAP1(PDB: 3MX6) and HsMetAP1 (PDB: 2B3K)
shaded according to alignment (dark = identity; light = similarity; no-shading = non-conserved).
Alignment reveals 43% identity between the proteins. Residues responsible for cofactor binding are
marked as X.

Table 5: Comparison of activity (IC50) against RpMetAP and HsMetAPs
Compound
(19)
RpMetAP1a,b
1.4 ± 0.1
1.0 ± 0.3
0.9 ± 0.1
1.3 ±0.7
HsMetAP1c,d
>15
>15
HsMetAP2c,e
>15
2.03 ± 0.3
1.27 ± 0.6
0.055 ± 0.02
(20)
(21)
12.9 ± 1.0
>15
(22)
a. IC50 values in µM
b. IC50 values and standard error obtained from nonlinear regression analysis of the average observed
activity (in triplicate) versus inhibitor concentration data.
c. In cases where analysis is limited by compound solubility, the minimum estimate of IC50 is
provided, Values from Bhat et al.28
d. Co(II) cofactors employed in assay
e. Mn(II) cofactors employed in assay
2.6 Host-Cell Viability Assay
Based on the in silico and in vitro screens, a subset of 17 RpMetAP-inhibitory compounds was
selected for testing against R. prowazekii growth in a cell culture model infection system (Table 6).
Considering that R. prowazekii is a vasculotropic pathogen, we used primary pulmonary vascular
endothelial cells (ECs) of rat origin for the infection studies. In order to design an assay that was
compatible with working under BSL-3 conditions at moderate throughput, our strategy here was to use
host cell viability as an indirect measure of rickettsial intracellular growth. Under typical infection
conditions, rickettsiae grow inside the host cell to large numbers until the host cell succumbs to
infection and lyses. Thus, any RpMetAP-inhibitory compound that inhibits rickettsial growth will protect
host cell viability over the infection time course. To this end, we used a cell proliferation assay where the
amount of water-soluble tetrazolium salt-1 (WST-1) metabolism directly correlates with the number of
viable cells.
There are several noteworthy considerations pertaining to rickettsial obligate intracellular
growth and our study design. Firstly, rickettsiae are not free-living organisms like E. coli. Rickettsiae
cannot be grown in pure culture and thus, typical antibiotic resistance screening using, for example,
Mueller-Hinton agar and disc diffusion susceptibility assays are not applicable. Secondly, because

rickettsiae must grow inside of a eukaryotic host cell, it is not reasonable to design assays based on
direct exposure of the pathogen to potential antibiotic compounds. A physiologically relevant screening
system must consider the biology of both pathogen and host. This includes diffusion of compounds into
the eukaryotic host cell cytoplasm and metabolism/break-down of the compounds by host cell
metabolism. Lastly, it is important to note that due to their obligatory intracellular growth consideration
of selective toxicity is of paramount importance. Any perturbations that adversely affect host EC health
will indirectly result in R. prowazekii growth inhibition. Thus, great care was taken to determine the
effects of solvents on R. prowazekii and EC viability, and to test for off-target effects of the RpMetAP-
inhibitory compounds on EC viability prior to screening against R. prowazekii-infected ECs. We first
determined the optimal concentration of solvent (DMSO) used to dissolve the RpMetAP-inhibitory
compounds that was non-toxic to both the ECs and R. prowazekii (optimal concentration was
determined to be 0.3% DMSO, data not shown). We next tested the RpMetAP-inhibitory compounds
over a range of concentrations ranges on uninfected ECs and identified concentration ranges that were
non-toxic to the ECs (data not shown). Based on the solubility of the RpMetAP-inhibitory compounds in
100% DMSO and the preliminary EC toxicity screen, we selected the optimal concentration ranges for
testing each of the 23 RpMetAP-inhibitory compounds against R. prowazekii-infected ECs (see Table 6).

Table 6: RpMetAP Inhibitory Compounds Tested in EC Culture Infection Model
Compound
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
(26)
MW a
222.63
257.07
206.27
220.30
224.26
240.72
240.72
275.16
248.35
262.37
195.65
181.62
245.06
190.16
315.20
362.25
325.75
Stock (mM) b
Test Range (µM) c
Growth Inhibition d
100
300, 30, 3, 0.3
No
No
Yes
Yes
Yes
Yes
Yes
No
No
No
No
Yes
Yes
No
No
No
Yes
100
300, 30, 3, 0.3
100
300, 30, 3, 0.3
100
300, 30, 3, 0.3
100
300, 30, 3, 0.3
100
300, 30, 3, 0.3
100
300, 30, 3, 0.3
100
30, 3, 0.3, 0.03
30, 3, 0.3, 0.03
30, 3, 0.3, 0.03
3, 0.3, 0.03, 0.003
30, 3, 0.3, 0.03
30, 3, 0.3, 0.03
3, 0.3, 0.03, 0.003
20, 2, 0.2, 0.02
7.5, 0.75, 0.075, 0.0075
3, 0.3, 0.03, 0.003
100
100
100
100
100
100
40
25
100
a. Compound Molecular Weight in g/mol.
b. Stock inhibitor solutions were initially prepared in 100% DMSO to achieve a 100 mM concentration. Compounds
that were insoluble at 100 mM were adjusted by addition of DMSO until solubility was achieved.
c. All compounds were initially tested for toxicity against uninfected ECs. Working concentration test ranges that
produced no EC toxicity were selected for further testing against R. prowazekii-infected ECs.
d. Inhibition of rickettsial growth was determined using the in vitro host EC viability assay described in the
Experimental Methods section.

Figure 5: Effects of RpMetAP inhibitors on R. prowazekii growth. The loss of infected host ECs viability due to
normal R. prowazekii growth was used to indirectly assess whether RpMetAP inhibitors produced antibacterial
effects. The expectation here is that compounds inhibiting rickettsial growth will restore host EC viability to levels
comparable to the uninfected control ECs. Host EC viability was measured via WST-1 metabolism (calculated as
A450nm – A660nm and normalized to media only as background). Solid black bars denote compounds that restored
host EC viability to levels comparable to uninfected ECs (denoted by the dashed lines). U = uninfected control ECs;
Rp = R. prowazekii-infected host ECs; Rp+DM = R. prowazekii-infected host ECs treated with the vehicle/solvent
control (0.3% DMSO); numbers denote compound identification from Table 6. Infections were performed at a MOI
= 100 rickettsiae: 1 host cell. Results represent the average of triplicate infections ± standard deviation.
Figure 5 shows the effects of each of the compounds at the maximal concentration tested.
Control, uninfected EC monolayers were healthy and displayed robust WST-1 metabolism (solid black
bar labeled U). When EC monolayers were infected with R. prowazekii (at an MOI = 100:1), host EC
metabolism of WST-1 was reduced 2-fold at 72 hours post-infection (solid grey bar labeled Rp),
indicating that R. prowazekii intracellular growth reduced host EC viability. This result was confirmed
visually under the microscope where monolayers looked stressed with the presence of visible cellular
debris (data not shown). The combination of R. prowazekii infection and the addition of the solvent

control (0.3% DMSO) gave results comparable to infected ECs without DMSO (solid grey bar labeled
Rp+DM). The remainder of Figure 5 and Table 6 show 10 promising RpMetAP-inhibitory compounds out
of the 23 tested that restored R. prowazekii-infected EC viability to levels equal to or greater than the
uninfected control ECs. The sensitivity of this viability assay was further assessed using EC monolayers
infected with a lower burden of rickettsiae (MOI = 50:1), which resulted in only a 1.2-fold reduction in
host EC viability, and revealed a similar subset of RpMetAP-inhibitory compounds as restorative of host
EC viability (data not shown).
When tested serially at 10-, 100- and 1000-fold dilutions, the RpMetAP-inhibitory compounds
lost inhibitory efficacy indicating that the maximal concentration tested was the most effective at
inhibiting R. prowazekii growth (data not shown). It is unknown as to whether permeability into the host
ECs or compound metabolism/turnover played any role in the observed loss of efficacy. Again, it is
noteworthy that inhibitory compounds will only exhibit anti-rickettsial activity via uptake by the host EC
cells, followed by uptake by the obligate intracellular parasite cells; this assay therefore mimics practical
application better than cases where suspected therapeutics are introduced to free-living pathogens.
Interestingly, the data in Figure 5 show several of the RpMetAP-inhibitory compounds
stimulated host EC metabolism to levels greater than the control, uninfected and untreated cells. In fact,
the preliminary screen of RpMetAP-inhibitory compounds for off-target toxicity on uninfected ECs
revealed 8 of 23 compounds stimulated host EC metabolism (data not shown). The nature of this
metabolic stimulatory effect is unknown. It is noteworthy that of the compounds that stimulated EC
metabolism, only 4 were found to also inhibit rickettsial growth (12,15,16,22). However, studies with
compound (16) visually demonstrate rickettsial death upon treatment (Figure 6). Together these data
suggest that targeting R. prowazekii MetAP is selectively toxic to the rickettsiae and may represent a
novel class of anti-rickettsial therapeutics.

The furoic acid inhibitors were the most potent of the classes of compounds screened in the
enzymatic assay, but did not restore R. prowazekii-infected host cell viability. The results of the host-cell
viability assay therefore suggest furoic acid inhibitors are not readily permeable to host ECs, are not
permeable to bacteria residing within host ECs, or the native cofactors of RpMetAP are not Mn(II).
Considering that free carboxylic acids are known to exhibit minimal membrane permeability, the
observed lack of anti-rickettsial activity is likely the result of poor cellular uptake.31 As such, compounds
(6) and (8) were not screened for host-cell viability, although they exhibited similar activity as
compounds (10) and (11) when screened against RpMetAP1.
All quinolinol-based RpMetAP1 inhibitory compounds were screened in the host-cell viability
assay, regardless of the activity observed against the enzyme, because all but two compounds, (25 – 26),
exhibited IC50 values less than 10 µM. Interestingly, three compounds from this chemical class were
determined to restore R. prowazekii-infected host cell viability. Two were unsubstituted 8-quinolinol
derivatives (21, 22) while the remaining compound (26) was a Mannich derivative of nitroxoline (23).
The most effective compounds for restoring host-cell viability were from the 1,2,4-triazole
series. Because these compounds were determined to exhibit similar inhibitory activities against the
RpMetAP1 target, all eight analogs from this series were tested and five exhibited anti-rickettsial activity
(12 - 16) while restoring host-cell viability. The three analogs (17 - 19) that did not restore host-cell
viability contained lipophilic substituents that increased their cLogP values and lowered their solubility.
Finally, we assayed compound (16), which both inhibited RpMetAP1 activity and stimulated host
EC metabolism, to investigate whether the compound inhibited rickettsial growth by fluorescence
staining and microscopy to provide a visual representation of the observed effects. For these
experiments, doxycycline was included as a clinically relevant positive control to inhibit rickettsial
intracellular growth. Figure 6A shows a healthy, uninfected and untreated control EC monolayer with

Hoechst-stained nuclei (blue pseudocolor) and phalloidin-stained actin outlining cell boundaries (red
pseudocolor). Figure 6B shows the presence of several host ECs infected with intracellular, rod shaped R.
prowazekii stained with an anti-rickettsial antibody (green pseudocolor). Figure 6C shows treatment of
infected host ECs with compound (16) resulted in identification of intracellular debris that stained with
the anti-rickettsial antibody but did not resemble healthy, rod-shaped bacteria shown in Figure 6B; this
data strongly suggests that intracellular rickettsial growth abatement had occurred. We confirmed
visually that addition of compound (16) alone had no overt effect on the appearance of the monolayers
(data not shown). Moreover, a similar pattern of staining with the anti-rickettsial antibody was also
observed when infected ECs were treated with doxycycline as a positive control known to inhibit
rickettsial growth (Figure 6D). Together, these results confirm the promise of compound (16) as
representative of a novel class of anti-rickettsial therapeutics that do not appear to have overt
deleterious effects on host EC ultrastructure by fluorescence staining and microscopy.

Figure 6: Representative fluorescent micrographs showing effects of RpMetAP inhibitor compound (16) compared
to a doxycycline control. Panel A shows uninfected control ECs. Panel B shows ECs infected with R. prowazekii.
Panel C shows the effects of treating R. prowazekii-infected ECs with compound (16) (300 µM final concentration).
Panel D shows the effects of treating R. prowazekii-infected ECs with doxycycline (25 µg/mL) as a positive control.
Uninfected control cells show very little background antibody staining. The infected control shows infection with
many bacilli (white arrows). Treatment with either inhibitor caused bacterial shape changes consistent with
condensation and death (white asterisk). Nuclei were stained with Hoescht (blue pseudocolor); actin was stained
with Texas Red-labeled Phaloidin (red pseudocolor); and rickettsiae were stained with a specific antibody
conjugated to FITC (green pseudocolor).

3. Conclusions
A series of compounds based upon furoic acid, 1,2,4-triazole, and quinolinol motifs were
synthesized or purchased and screened for enzymatic inhibitory activity against R. prowazekii MetAP.
Suspected RpMetAP1 inhibitory compounds were first screened against the enzyme in a fluorescence-
based activity assay, with Mn(II) utilized as the enzymatic cofactors. It was determined that compounds
based upon all three scaffolds exhibited potent inhibitory activity. The most active compounds were
subjected to a host cell viability assay to determine anti-Rickettsial activity in the presence of host
endothial cells, which revealed only quinolinol and triazole-based inhibitors were active at “reasonable”
concentrations. The lack of observed activity from the furoic acid series may be due to limited
membrane permeability of small molecules containing free acid groups. The most potent compound in
the host-cell viability assay was 2-chlorophenyl 1,2,4-triazole derivative (16) and the activity of this
compound was therefore monitored with fluorescence staining and microscopy to provide a visual
representation of the anti-Rickettsial activity in the presence of mammalian host-cells. These data
suggest that compounds based upon both the 1,2,4-triazole and quinolinol scaffolds exhibit efficacy
against the intracellular pathogen while allowing the restoration of host-cell viability. Thus, compounds
of these classes should be optimized against RpMetAP1 and the resulting species should be re-screened
to determine anti-Rickettsial activity in the presence of mammalian host-cells.
4.
Experimental
4.1 Crystallization and Data Collection
Purified RpMetAPs were screened for crystallization in 96-well sitting-drop plates against the
JCSG+ and PACT crystal screens (Rigaku Reagents). Equal volumes of protein solution (0.4 µl) and
precipitant solution were set up at 293 K against reservoir (80 µl) in sitting-drop vapor-diffusion format.

The A4 construct crystallized in the PACT screen condition G8 which contains 0.2 M sodium sulfate, 0.1
M BisTris propane pH 7.5, and 20% w/v PEG 3350. The A6 construct crystallized in the PACT screen
condition A6 which contains 0.1 M SPG (succinic acid, phosphate, glycine) buffer pH 8.0 and 25% w/v
PEG 1500. The crystals were cryo-protected in crystallant plus 25% v/v 1,2-ethanediol and cryo-cooled
by dipping into liquid nitrogen. Data were collected under the Collaborative Crystallography program of
the Berkeley Center for Structural Biology at the Advanced Light Source, Berkeley National Laboratory.
Data were collected at 100 °C on Advanced Light Source beamline 5.0.1 using an ADSC Quantum 315
CCD detector with 1° oscillations at a wavelength of 0.9774 Å. Data were reduced with HKL200032. Raw
X-ray diffraction images are available at the Integrated Resource for Reproducibility in Macromolecular
Crystallography at www.proteindiffraction.org.
4.2 Structure Solution and Refinement
The structure was solved by molecular replacement with Phaser from the CCP4 suite of
programs using PDB entry 1XNZ12 as a search model.33 The structure was refined using iterative cycles of
Refmac5 followed by manual rebuilding of the structure using Coot.34 The quality of all of the structures
was assessed using MolProbity.35 All data-reduction, refinement, and data collection statistics can be
found in Table S3. Structure figures were analyzed and prepared using PyMOL (v.1.5; Schrodinger) and
PISA. Coordinates and structure factors have been deposited with the Protein Data Bank (PDB)
(http://www.rcsb.org/pdb/home/home.do) with accession number 3MR1 and 3MX6.

4.3 RpMetAP Cloning and Purification
Cloning, expression and purification were conducted as part of the Seattle Structural Genomics
Center for Infectious Disease (SSGCID) following standard protocols described previously.36 All
constructs of RpMetAP (Uniprot: Q9ZCD3) were PCR-amplified from Rickettsia prowazekii str. Madrid E
kindly provided by Dr. David H. Walker. A full length (1-259) and five variant open reading frames (ORF)
were cloned into the ligation independent cloning (LIC) expression vector pAVA0421 encoding a
cleavable 6xHis fusion tag followed by the human rhinovirus 3C protease-cleavage sequence
(MAHHHHHHMGTLEAQTQGPGSM) followed by the ORF.37 The human rhinovirus 3C protease-cleavage
site is between the glutamine and glycine residues that are underlined. An additional full length
construct was cloned into the expression vector pBG1861 which encodes a non-cleavable 6xHis fusion
tag (MAHHHHHH). See Supplemental Information, Table S2 for list of constructs. Complete amino acid
sequences are included in supplementary materials.
Plasmid DNA was transformed into chemically competent E. coli BL21 (DE3) R3 Rosetta cells.
Cells were expression tested in 96 well blocks according to the procedures described in Choi, et al.,
2011.36e Briefly, cells are inoculated into 600 µl ZYP-5052 auto-induction medium [Sterile ZY Broth (10
g/L tryptone, 5 g/L yeast extract), 1 mM MgSO4, 1× metals mix, 1× 5052 (0.5% glycerol, 0.05% glucose,
0.2% α-lactose monohydrate) and 1× NPS] supplemented with the correct antibiotics. The block was
sealed and incubated on a plate shaker inside a refrigerated incubator set at 293 K for roughly 27 h to
allow the cultures to reach saturation or early stationary phase. The cultures were not harvested until
OD600nm readings of at least 0.6 were obtained. Once the induced cells were at the correct density,
they were centrifuged at 4 oC and 4300 rpm for 30 min. After centrifugation, the supernatant was
discarded and the block with the semi-dry cell pellets were stored at 193 K. Cell pellets were analyzed
for expression of insoluble and soluble RpMetAP by re-suspension in 600 µl lysis buffer [20 mM HEPES
pH 7.0, 500 mM NaCl, 5% glycerol, 0.5% CHAPS (A.G. Scientific Inc., San Diego, California, USA), 30 mM