Optimization of globomycin analogs as novel gram-negative antibiotics

Article abstract of DOI:10.1016/j.bmcl.2020.127419

Discovery of novel classes of Gram-negative antibiotics with activity against multi-drug resistant infections is a critical unmet need. As an essential member of the lipoprotein biosynthetic pathway, lipoprotein signal peptidase II (LspA) is an attractive target for antibacterial drug discovery, with the natural product inhibitor globomycin offering a modestly-active starting point. Informed by structure-based design, the globomycin depsipeptide was optimized to improve activity against E. coli. Backbone modifications, together with adjustment of physicochemical properties, afforded potent compounds with good in vivo pharmacokinetic profiles. Optimized compounds such as 51 (E. coli MIC 3.1 μM) and 61 (E. coli MIC 0.78 μM) demonstrate broad spectrum activity against gram-negative pathogens and may provide opportunities for future antibiotic discovery.

Full text of DOI:10.1016/j.bmcl.2020.127419

Journal Pre-proofs
Optimization of globomycin analogs as novel gram-negative antibiotics
Keira Garland, Homer Pantua, Marie-Gabrielle Braun, Daniel J. Burdick,
Georgette M. Castanedo, Yi-Chen Chen, Yun-Xing Cheng, Jonathan Cheong,
Blake Daniels, Gauri Deshmukh, Yuhong Fu, Paul Gibbons, Susan L. Gloor,
Rongbao Hua, Sharada Labadie, Xiongcai Liu, Richard Pastor, Craig Stivala,
Min Xu, Yiming Xu, Hao Zheng, Sharookh B. Kapadia, Emily J. Hanan
PII:
S0960-894X(20)30530-8
DOI:
Reference:
https://doi.org/10.1016/j.bmcl.2020.127419
BMCL 127419
To appear in:
Bioorganic & Medicinal Chemistry Letters
Received Date:
Revised Date:
Accepted Date:
8 June 2020
13 July 2020
14 July 2020
Please cite this article as: Garland, K., Pantua, H., Braun, M-G., Burdick, D.J., Castanedo, G.M., Chen, Y-C.,
Cheng, Y-X., Cheong, J., Daniels, B., Deshmukh, G., Fu, Y., Gibbons, P., Gloor, S.L., Hua, R., Labadie, S., Liu,
X., Pastor, R., Stivala, C., Xu, M., Xu, Y., Zheng, H., Kapadia, S.B., Hanan, E.J., Optimization of globomycin
analogs as novel gram-negative antibiotics, Bioorganic & Medicinal Chemistry Letters (2020), doi: https://
doi.org/10.1016/j.bmcl.2020.127419
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. 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.
© 2020 Published by Elsevier Ltd.

Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com
Optimization of globomycin analogs as novel gram-negative antibiotics
Keira Garland,a,† Homer Pantua,a Marie-Gabrielle Braun,a Daniel J. Burdick,a Georgette M. Castanedo,a
Yi-Chen Chen, a Yun-Xing Cheng,b Jonathan Cheong, a Blake Daniels,a,‡ Gauri Deshmukh,a Yuhong Fu, b
Paul Gibbons,a,§ Susan L. Gloor,a, Rongbao Hua,b Sharada Labadie,a Xiongcai Liu,b Richard Pastor,a Craig
Stivala,a Min Xu,a Yiming Xu,a Hao Zheng,a Sharookh B. Kapadia,a and Emily J. Hanan,a*
aGenentech, Inc., 1 DNA Way, South San Francisco, CA, 94080, USA
bPharmaron Beijing Co., Ltd, No. 6 Taihe Road, BDA, Beijing, 100176, P.R. China
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
Discovery of novel classes of Gram-negative antibiotics with activity against multi-drug resistant
infections is a critical unmet need. As an essential member of the lipoprotein biosynthetic
pathway, lipoprotein signal peptidase II (LspA) is an attractive target for antibacterial drug
discovery, with the natural product inhibitor globomycin offering a modestly-active starting point.
Informed by structure-based design, the globomycin depsipeptide was optimized to improve
activity against E. coli. Backbone modifications, together with adjustment of physicochemical
properties, afforded potent compounds with good in vivo pharmacokinetic profiles. Optimized
compounds such as 51 (E. coli MIC 3.1 M) and 61 (E. coli MIC 0.78 M) demonstrate broad
spectrum activity against gram-negative pathogens and may provide opportunities for future
antibiotic discovery.
Keywords:
Signal peptidase
LspA
Globomycin
Gram-negative
———
* Corresponding author. Tel.: +1-650-225-4577; e-mail: hanan.emily@gene.com

Present Address: †Proneurotech, ‡Wilson Sonsini Goodrich & Rosati, §ORIC Pharmaceuticals, EpiCypher, Inc.

Multidrug-resistant (MDR) infections of Gram-negative
bacteria remain a significant medical challenge and there is an
urgent need for new classes of antibiotics to address the emerging
resistance to the current classes of antibiotics.1 Despite this clinical
need, there have been no new approved classes of antibiotic for
Gram-negative pathogens in over fifty years. The lipoprotein
biosynthetic pathway presents a compelling opportunity for
antibacterial discovery, with multiple enzymes located in the inner
membrane that are required for the post-translational processing of
lipoproteins.2,3 Lipoprotein signal peptidase II (LspA) is
responsible for cleaving the signal peptides from diacylated
prolipoproteins to enable their subsequent modification and
trafficking to the outer membrane (OM), and is essential for
growth in all clinically relevant Gram-negative bacteria.4
Inhibition of LspA in wild-type Escherichia coli results in mis-
localization and accumulation of peptidoglycan-linked Braun’s
lipoprotein (Lpp) in the inner membrane leading to bacterial cell
death.4-6
strain USA300, given that previous reports have demonstrated that
S. aureus strains with defective lipoprotein biosynthesis are viable
in vitro.16 Globomycin itself shows no activity against S. aureus
up to 100 M.
Both solution-phase and solid-phase syntheses of the
globomycin scaffold have been previously reported.17,18 We first
synthesized the lipophilic tail fragments utilizing the Oppolzer’s
sultam chiral auxiliary to generate the chiral aldol product, with
auxiliary cleavage providing the chiral carboxylic acids A1 and A2
(Scheme 1).19 We synthesized the peptide backbone utilizing
standard solid-phase synthesis techniques, and triphosgene to
facilitate the final coupling between A1 or A2 and the hindered N-
methyl peptide.20 After cleavage from resin, macrocyclization of
the linear peptide was effected under Shiina lactonization21
conditions, and subsequent global deprotection of amino acid side
chains afforded the final products (Scheme 2).
Two natural products are known to selectively inhibit LspA:
globomycin, produced by Streptomyces species, and
myxovirescin, produced by myxobacteria (Figure 1).7-10 Although
non-natural product derived inhibitors of LspA have been
reported, they lack measurable growth inhibition against wild-type
Gram-negative bacteria.11 Globomycin, a cyclic depsipeptide, has
previously been optimized with some improvement in activity
against Enterobacteriaceae. However, these analogs still lacked
activity against non-fermenter species such as Acinetobacter and
Pseudomonas species.12,13 Additionally, all analogs retained a
metabolically labile ester moiety, and hence were unsuitable for in
vivo evaluation. The 2016 publication of a co-crystal structure of
globomycin with LspA offered new opportunities for structure-
based drug design and further optimization of the globomycin
scaffold.14
Scheme 1. Reagents and conditions: (a) propionyl chloride,
HO
O
O
O
NH
O
O
HN
HO
(S)
O
O
HN
HN
O
O
(S)
O
O
NH
OH
(S)
(R)
O
N
OH
OH
Figure 1. Natural product inhibitors of LspA
globomycin
myxovirescein A1
As described by Vogeley et al., globomycin occupies the
putative active site of LspA and forms both polar and hydrophobic
interactions with the protein.14 The peptide backbone of
globomycin engages in the majority of the polar interactions in the
center of the binding site, with the lipophilic side chains and n-
hexyl tail nestled against the membrane-facing surface of the
protein. While we sought to optimize these interactions, we also
sought to improve permeation of these molecules through the OM.
The mechanism by which globomycin penetrates the OM is not
entirely clear: its size likely precludes it from transversing through
general porins, and the significant difference in minimum
inhibitory concentration (MIC) against wild-type and a membrane
permeabilized E. coli strain are consistent with the OM
permeability barrier of Gram-negative bacteria limiting
globomycin penetrance.15 In addition, with a 39-fold improvement
in MIC when tested in an E. coli efflux pump (tolC) mutant strain
(Table S1), gobomycin appears to be a substrate of the TolC efflux
pump, which likely plays a role in limiting its intracellular
concentration. However, interpreting the magnitude of efflux
liability using the tolC strain is complicated by the impact on OM
integrity with the loss of the TolC protein.
triethylamine, 4-dimethylaminopyridine, THF; (b) R1-CHO,
CaH2, TBDMSOTf, Et3N, TiCl4, DCM, 0 C-r.t.; (c) LiOH or
KOSi(CH3)3, H2O, THF
The SAR described by Kiho et al. indicated a correlation
between increasing potency and extension of the linear aliphatic
tail and no change in potency when inverting the side chain
stereochemistry of the allo-isoleucine residue to an isoleucine. We
were able to recapitulate these results with globomycin and
compounds 1 and 2 (Table 1). We were particularly interested in
modifying the serine residue based on observations in the co-
crystal structure of globomycin and LspA, in which the sidechain
hydroxyl appears to form a bridging hydrogen bond between the
two putative catalytic aspartate residues in the active site.14
To inform our SAR studies, we measured MICs in both the
wild-type uropathogenic clinical E. coli isolate (CFT073) and an
OM-permeable mutant of CFT073 (CFT073imp4213).
Differences in activity between these two strains can elucidate how
changes to an inhibitor affects its ability to penetrate the outer
membrane of the bacteria, vs. its ability to bind to the target
protein. To avoid optimization of cell killing via off-target activity
or general membrane disruption, we also monitored activity
against the Gram-positive Staphylococcus aureus (S. aureus)

Modification to an alanine (compound 3) and the resulting
complete loss of activity confirmed the importance of this residue.
We hypothesized that changing this hydroxyl to a positively
charged moiety would enable a stronger salt-bridge interaction. A
variety of positively charged side chains (compounds 4-7) were
evaluated and the 2,3-diaminopropionic (Dap) residue (4) was
found to provide the best activity. Interestingly, in the context of
the n-decyl tail, the improvement in activity was observed only in
the WT strain (compound 4 vs 2), while in the context of the n-
hexyl tail both OM-permeabilized and WT MICs were improved
by 5-fold (8 vs globomycin).
recapitulated in Table 1, we confirmed that increasing the length
of this tail from C6 to C10 enhances potency. However, as shown
in Supplemental Table S2, many compounds with a C10 tail begin
to show apparent off-target activity against the gram-positive
pathogen S. aureus USA300. Activity against S. aureus USA300
roughly correlates with lipophilicity, particularly for compounds
with LogD > 4.0, suggesting the influence of a non-specific
mechanism of action which prompted us to evaluate alternatives
to a linear aliphatic tail (Table 2). The corresponding chiral
carboxylic acids were prepared following methods analogous to
those described in Scheme 1 (see Supplementary Material for
details). Both terminal and interior branching was tolerated with
20 and 21 providing similar potency to 19 and with reduced S.
aureus USA300 activity. A terminal CF3 group (22) as well as
mono-fluorinated compound 23 showed significantly reduced
activity in both OM-permeabilized and wild-type E. coli relative
to 12. The introduction of polar heteroatoms, such as ethers,
sulfones, or amines, typically ablated all activity (Supplemental
Table S3). Compounds 24 and 25 with a tertiary alcohol or ether
(Table 2) were exceptions that retained some activity; the latter
ether 25 was equipotent to compound 19 but with a reduced LogD.
Extension of the methyl group adjacent to the lipophilic tail was
tolerated as long as a combined minimum number of carbons was
maintained, but no significant improvements in potency were
found and lipophilicity was generally increased (Supplemental
Table S4). An attempt to reduce the conformational flexibility of
the hydrophobic tail led to compounds 26-29 shown in Table 3. As
compared to compound 13 (LogD 3.4, S. aureus USA300 MIC 50
M), these compounds afforded good activity, reaching sub-M
MICs with reduced lipophilicity. Norbornyl analogs 26 (exo) and
27 (endo) also demonstrated reduced S. aureus USA300 activity
relative to 13. The tolerance for lipophilic groups with a variety of
constrained 3D character is consistent with non-specific
hydrophobic interactions at the interface of the protein and the
interior of the inner membrane.
Scheme 2. Reagents and conditions: (a) N-Fmoc protected amino
acid, HOBt, DIPEA, DMF; (b) 20% piperidine in DMF; (c) N-
methyl-N-Fmoc protected amino acid, HOBt, DIPEA, DMF; (d)
triphosgene, DIPEA, collidene, THF; (e) 7:2:1
DCM:AcOH:TFE; (f) 2-methyl-6-nitrobenzoic anhydride, 4-
dimethylaminopyridine, THF/DCM; (g) global deprotection:
TFA or Pd-C/H2/EtOAc.
Adjacent to the serine, the allo-isoleucine residue makes
hydrophobic contacts with the protein. As illustrated in Table 1,
we elucidated several trends at this position: preference for
branching location in alkyl chains (allo-isoleucine 8 vs leucine 9),
increased activity for aliphatic over aromatic side chains (10 and
11), and enhanced potency for cyclic aliphatic groups such as
cyclohexyl and cycloheptyl glycine (12 and 13). While this side
chain does reach the edge of the active site, it is expected that this
region of the protein is not solvent facing, but buried within the
inner membrane. Consistent with the highly lipophilic
environment, polar side chains at this position, either with or
without hydrogen bond donors, completely ablated activity (e.g.,
compounds 14 - 16). The next residue in the globomycin scaffold,
N-methyl-leucine, is in a similarly lipophilic environment, and we
again found that polar side chains were not tolerated
(Supplemental Table S2). Branching was also not required and
small aliphatic side chains such as norvaline retained activity
(compound 17), although further reduction in side chain length
(e.g. ethylglycine 18) reduced activity (Table 1).
In order to progress LspA inhibitors towards in vivo studies, we
profiled the pharmacokinetics of globomycin itself both in vitro
and in vivo. It was found to be labile in mouse plasma (49%
remaining after 6 hour incubation), presumed to be the result of
esterase activity. Existing SAR indicated that the corresponding
amide to the globomycin ester has no measurable activity.13
The lipid tail of globomycin appears to make favorable
hydrophobic interactions along a shallow groove in the surface of
the protein adjacent to the active site. As was shown by Kiho and
Table 1. Globomycin analogsa
Compound
A
B
C
n
CFT073imp4213b CFT073b MIC
USA300c MIC
MIC (M)
(M)
(M)
globomycin
Ser
Ser
Ser
Ala
Dap
alloIle
alloIle
Ile
Leu
Leu
Leu
Leu
Leu
1
5
5
5
5
0.25
44
>100
25
1
2
3
4
0.016
0.016
25
5.4
9.4
25
Ile
100
2.2
25
Ile
0.03
13

5
Dab
His
Ile
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Leu
Nva
Abu
5
5
5
1
1
1
1
1
1
1
1
1
1
1
0.39
6.3
25
13
6
Ile
100
50
25
7
Agp
Dap
Dap
Dap
Dap
Dap
Dap
Dap
Dap
Dap
Dap
Dap
Ile
1.6
13
8
alloIle
Leu
Phe
0.053
0.4
8.6
25
>100
>100
>100
>100
>100
50
9
10
11
12
13
14
15
16
17
18
0.8
50
Phg
Chg
C7g
Thr
1.6
71
0.059
0.014
50
3.1
1.1
>100
>100
>100
3.5
8.8
>100
>100
>100
>100
>100
Thr(Me)
Gln
7.7
>100
0.049
0.14
Chg
Chg
aAgp = (S)-2-amino-3-guanidinopropanoic acid; Dap = (S)-2,3-diaminopropionic acid; Dab = (S)-2,3-diaminobutyric acid; Phg = phenylglycine; Chg =
cyclohexylglycine, C7g = cycloheptylglycine, Thr(Me) = O-Methyl-L-threonine, Nva = norvaline; Abu = aminobutyric acid; b E. coli; c S. aureus
Table 2. Effects of hydrophobic tail modification
Compound
R4
R1
CFT073imp4213a MIC CFT073a MIC (M)
(M)
USA300b
MIC (M) (pH 7.4)
LogD
19
-(CH2)9CH3
0.011
0.035
0.034
2.1
3.1
3.1
15
4.6
3.0
3.2
20
21
22
100
100
0.39
25
>100
2.0
CF3
F
23
24
0.39
0.55
25
15
>100
>100
1.6
2.2
F
OH
25
0.048
2.2
>100
2.8
O
a E. coli; b S. aureus
However, simply replacing the ester with the corresponding
ether (30) reduced potency by only 3- fold and the resulting
compound was also stable in plasma (Table 4). Ring enlargement
(31) or substitution at the adjacent carbon (32 and 33) were not
beneficial. Substitution at the -carbon did provide a potency
advantage, with the optimal stereochemistry providing compounds

35 and 36 with comparable activity to compound 12 while also
affording excellent plasma stability.
We also attempted to replace the ester group with an amino
linkage, and found that substantial activity was lost with
compound 38 (Table 5). Interestingly, removing one carbon from
the macrocyclic backbone, to generate an 18-membered ring
instead of the 19-membered ring of globomycin, significantly
improved activity, with 39 providing a MIC of 25 M, and with
no plasma stability issues. Within this context, the aliphatic tail
can be placed at either position (39 vs. 40) or with two shorter
aliphatic groups split between the two positions (41) all with
similar activity. Similar to what was observed for the ether
analogs, the introduction of a methyl group onto the backbone
provided a few fold improvement in both OM-permeabilized and
wild-type E. coli MICs as exemplified by compound 42.
Figure 2: Distance from the allo-threonine globomycin side
chain to the predicted membrane-periplasm interface (PDB 5DIR)
Table 3. Rigidified hydrophobic tails
Compound
R1
CFT073imp4213a MIC CFT073a MIC (M) USA300b MIC
LogD
(pH 7.4)
(M)
(M)
26c
0.041
0.97
0.78
0.4
100
2.8
2.9
3.1
27d
28
0.009
0.009
100
25
29
0.009
0.78
50
2.9
a E. coli; b S. aureus; cMixture of exo diastereomers; dmixture of endo diastereomers

Table 4. Macrocyclic ethers
Compound
X
Y
CFT073imp4213 CFT073a MIC
USA300b MIC
(M)
Plasma stability
(h/m)
a MIC (M)
(M)
8.7
12
-CH2-
0.14
100
100
100
100
71
99% / 96%c
96% / 101%d
91% / 101%d
100% / 92%d
88% / 93%d
93% / 94%d
100% / 91%d
101% / 87%d
30
31
32
33
34
35
36 e
37
a
-CH2-
-CH2-
-CH2CH2-
0.39
(S) -CH(CH3)-
(R) -CH(CH3)-
-CH2-
-CH2-
0.1
12
-CH2-
0.4
35
(S) -CH(CH3)-
(R) -CH(CH3)-
(R) -CH(CH3)-
(R) -CH(CH2CH3)-
1.0
84
-CH2-
0.034
0.048
0.024
2.9
4.4
2.2
71
-CH2-
100
50
-CH2-
b
c
d
e
E. coli; S. aureus; % parent remaining after 6 hour or 24 hour incubation in either human (h) or mouse (m) plasma; N-Me-
norvaline instead of N-Me-leucine

The general syntheses of ether- and amino-linked analogs in
Tables 4 and 5 are described in Schemes 3 and 4. Cyclic sulfamates
could be used to alkylate a hydroxy acid to generate the
corresponding alkoxy ester. Ester hydrolysis and an exchange of
Boc for Fmoc protecting groups afforded the N-Fmoc-protected
amino acids for peptide synthesis. Coupling to resin-bound tetra-
peptides, Fmoc deprotection, and cleavage from the resin afforded
linear amino acids, which were cyclized using HATU coupling
followed by global side chain deprotection to afford the ether-
linked macrocycles. Amino-linked macrocycles were prepared
starting with benzyl or methyl ester protected amino acids, with
two sequential reductive amination reactions providing the tertiary
amines. Deprotection of the carboxylic acid and an exchange of
Boc for Fmoc protecting groups afforded N-Fmoc-protected
amino acids for peptide synthesis. Coupling to resin-bound tetra-
peptides, Fmoc deprotection, and cleavage from the resin afforded
linear amino acids, which were cyclized using HATU coupling
followed by global side chain deprotection to afford the amino-
linked macrocycles. With plasma-stable compounds in hand, we
evaluated the in vivo profile of 36 and 42. Both compounds have
low to moderate CL in rat liver microsomes and hepatocytes, but
high CL was still observed when dosed IV in rats (Table 6). The
increased CL observed in vivo relative to in vitro systems may be
due to a combination of active hepatic uptake and passive renal
filtration. The intrinsic permeability of these compounds is quite
low (36 MDCK Papp = 0.02 x10-6 cm/sec), and thus they may not
be able to overcome any active transport mechanisms with passive
Scheme 3. Reagents and conditions: (a) NaH, TBAF, THF; (b) LiOH,pMeremOeHa,bHili2tyO.; (Wc)hHilCel, daicotixvaene;tr(adn)sFpmorotc-Cmle,cLhiaOnHis,mTsHFc;a(ne) Nbe-
Fmoc amino acid, DIPEA, DMF; (f) piperidine, DMF; (g) N-Fmoc amchinaolleancgidin, gDIPEtoA, HchOaBrat,ctDerMizFe,; (h)siNg-nmifeicthanytll-yN-Fmalotecripnrgotecthede
physicochemical properties of a scaffold can affect transporter
recognition. Globomycin scaffold SAR to this point indicated a
fairly restricted chemical space seemingly without possibilities for
adding additional polar or charged functionality. However, when
the co-crystal structure of globomycin and LspA was analyzed to
predict the boundaries of the membrane in which LspA would be
embedded, the hydroxyl of the allo-threonine side chain is within
4 Å of the calculated IM-periplasm interface (Figure 2).22 We
hypothesized that a positively charged group may provide a
favorable interaction with the negatively charged phosphate head
groups at the periplasmic border of the lipid bilayer. Multiple
positive charges are a common feature among antibiotics with
gram negative activity, and may also enhance the ability of
compounds to penetrate the OM.23

amino acid, HOBt, DIPEA, DMF; (i) HATU or PyAOP, DIPEA, DMF; (j) HFIP, DCM; (k) HATU, DIPEA, THF; (l) global deprotection:
TFA or Pd-C/H2/EtOAc
Scheme 4. PG = Bn or CH3. Reagents and conditions: (a) R8-CHO, NaBH(OAc)3, DCM or EtOAc; (b) NaBH(OAc)3, DCM or EtOAc;
(c) HCl, dioxane; (d) Fmoc-Cl, NaHCO3; (e) LiOH, H2O, THF; (f) Pd-C, H2; (g) resin-bound tetrapeptide, DIPEA, HATU or PyAOP,
DMF; (h) piperidine, DMF; (i) HFIP/DCM; (j) HATU, DIPEA, THF; (k) Pd-C/H2
Consequently, we prepared a series of analogs with positively
charged moieties at this position, following the general synthetic
method described in Scheme 3. Analogs illustrated in Table 7
demonstrate the feasibility of this approach, with multiple
compounds providing <10 M MICs against wild type E. coli.
There is flexibility in the structure of the linker, with both 3-carbon
and 4-carbon linked amino groups providing similar activity
(compounds 43 and 44), and the corresponding tertiary amine 45
only marginally less active. Interestingly, the guanidine 46 had
activity similar to the primary amine, but undesired S. aureus
USA300 activity was increased. Aromatic rings could be included
in the linker, as either a benzyl amine (47) or phenethylamine (48)
led to 2- and 7-fold improvement in wild-type MIC vs 43,
respectively. Moving the ether oxygen farther from the core
macrocycle provided a similar profile (49 vs 50). Adding small
aliphatic rings (51) or a second charged amino group (52 and 53)
to the linker provided some benefit in wild type E. coli compared
to compound 43. A carboxylic acid resulted in loss of activity
against wild type E. coli, even with a positively charged spacer
between the anion and the core macrocycle (compound 54). The
side-chain ether could be replaced with an amino, urea, or amide
spacer with similar activity (compounds 56 – 58 compared to 55).
Table 5. Macrocyclic amines
Compound
R4
n
2
R7
R8
R9
H
CFT073imp4213 CFT073a
USA300b
MIC (M)
Plasma stability
(h/m)c
a MIC (M)
MIC (M)
38
CH3
-(CH2)4CH3
12.5
>100
>100
101% / 100%
39
40
1
1
CH3
-(CH2)4CH3
H
H
0.5
25
12
>100
100
95% / 112%
92% / 81%
-
H
0.39
(CH2)5CH
3
41
42
1
1
-
-(CH2)2CH3
-(CH2)2CH3
H
0.1
6.3
1.6
100
71
NDd
(CH2)3CH
3
-
CH3
0.017
97% / 89%
(CH2)3CH
3
a E. coli; b S. aureus; c% parent remaining after 24 hour incubation in either human (h) or mouse (m) plasma; d ND = not determined
Table 6. Pharmacokinetic parameters for selected LspA inhibitors
Compound
LogD
Plasma stability
rat/mouse
PPB (%)
RLM/MLMb
(mL/min/kg)
Rat CL
(mL/min/kg)
Rat T1/2
(h)
Mouse CL
(mL/min/kg)
Mouse
T1/2 (h)
pH 7.4
(h/r/m)a
globomycin
3.1
2.8
3.4
1.6
2.2
0.7
1.7
0.2
1.5
100% / 1% / 2%
100% / 101% / 91%
97% / 102% / 89%
98% / 96% / 101%
103% / 90% / 84%
103% / 118% / 106%
99% / 94% / 95%
NDc
NS / NSc
37 / 81
20 / 69
27 / 77
20 / 43
28 / 48
ND / 53
21 / 41
ND / ND
<6.1 / 32
286
0.3
36
42
51
59
60
61
62
63
97.1 / 98.5
99.5 / 99.6
96.7 / 98.9
97.8 / 99.3
97.5 / 99.9
97.4 / 99.7
97.4 / 99.6
95.3 / 97.4
148
53
0.3
2.6
2.9
11
12
17
68
4.0
4.0
4.6
2.1
2.5
40
15
24
3.2
5.8
6.8
105% / 58% / 55%

64
110% / 66% / 95%
1.0
95.7 / 97.0
26 / 34
8.7
1.6
a% parent remaining after 24 hour incubation in either human (h), rat (r), or mouse (m) plasma; b RLM = rat liver microsomes, MLM = mouse liver microsomes;
c NS = not stable, ND = not determined
Table 7. Polar side chain SAR
Compound
n
R10
R11
CFT073imp CFT073a
4213a MIC MIC (M)
(M)
USA300b
MIC (M)
LogD
pH 7.4
43
44
45
46
0
0
0
0
CH3
CH3
CH3
CH3
-O-(CH2)3-NH2
-O-(CH2)4-NH2
0.28
0.18
0.39
0.2
11
7
100
56
1.5
1.7
2.1
2.0
-O-(CH2)3-N(CH3)2
25
4.4
100
5
H
O
N
NH
NH2
47
48
CH3
CH3
0
0
0.39
6.3
1.6
100
25
2.5
2.4
O
NH2
0.048
O
NH2
49
50
51
H
H
H
1
1
1
-O-(CH2)4-NH2
0.14
0.2
6.3
8.8
3.1
71
1.8
1.7
1.6
-CH2-O-(CH2)3-NH2
100
50
0.048
O
NH2
52
53
H
H
1
1
0.2
6.3
3.1
100
50
1.6
1.1
NH2
N
O
NH2
0.55
O
NH2
54
H
1
-O-(CH2)3-NH-(CH2)3-
CO2H
0.78
100
>100
1.3
55
56
57
H
H
H
1
1
1
-NH-(CH2)3-NH2
0.28
0.39
0.39
3.1
8.8
12
50
1.1
1.8
1.5
-N(CH3)(CH2)3-NH2
100
71
H
N
H
N
O
NH2
O
58
H
1
0.39
18
100
1.3
N
H
NH2
a E. coli; b S. aureus;
Interestingly, while the compounds in Table 7 generally
maintained the wild-type MIC of parent compound 36, the OM-
permeabilized MIC was elevated, with the exception of
compounds 48 and 51. It is possible that analogs with a charged
amine at this position reduced the intrinsic affinity of the inhibitor
for LspA or hindered ability of the inhibitor to access the
membrane-buried active site, resulting in decreased potency vs.
the OM-permeabilized strain. At the same time, the overall
physicochemical properties may enable better OM penetration and
access to the target, with a resulting decrease in shift between OM-
permeabilized and wild-type MICs. Selected compounds were
tested in a tolC-deficient E.coli strain and showed a small shift in
MIC, suggesting that reduction in efflux may also play a role
(Supplementary Table S1). Optimized inhibitors containing both a
charged sidechain and preferred rigidified lipophilic tails were
prepared and are illustrated in Figure 3 (compounds 59 – 64).
Compared to globomycin itself, these inhibitors not only provide
significant improvements in potency against E. coli, but also other
Enterobacteriaceae and non-fermenter P. aeruginosa and A.
baumanii strains, particularly compounds 59, 60, and 62 (Table 8).
While these compounds do show measurable activity against S.
aureus USA300, they do not inhibit the growth of a human Jurkat
cell line, and so are unlikely to be generally cytotoxic. We also
demonstrated that these compounds were specific to LspA as they
show reduced growth inhibition of CFT073 strains that either were
deficient in lpp or have over-expressed LspA, consistent with the
previously published activity of globomycin (Supplementary
Table S5).
In general, compounds 51 and 59-64 are moderately
metabolized in rat and mouse liver microsomes, and have good

plasma stability (Table 6). We generated in vivo PK profiles for
optimized inhibitors, and were pleased to see that compared to
compounds 36 and 42, these +2 or +3 charged inhibitors were able
to achieve moderate or low CL in both mouse and rat IV studies
(Table 6). For mouse, CL predicted by liver microsomes still
generally overpredicted the measured in vivo CL, potentially due
to the poor penetration of inhibitors into liver cells.
Figure 3. Optimized LspA inhibitors
Table 8. Optimized compounds demonstrate broad spectrum gram-negative activity
Compound
E. coli
E. cloacae
K.
P. aeruginosa
A. baumanii
S. aureus
USA300
MIC (M)
Cytotoxicity
CFT073
MIC (M)
MIC (M)
pneumoniae MIC (M)
MIC (M)
MIC (M)
EC50 (M)a
globomycin
44
100
6.3
6.3
3.1
3.1
1.6
ND
ND
>100
18
>100
100
71
>100
18
>100
50
>50
51
59
60
61
62
63
64
3.1
2.2
1.6
0.78
0.78
6.3
6.3
>100
ND
8.8
3.1
8.8
1.6
25
18
25
50
50
18
>50
100
12
25
50
>100
>100
ND
25
12
100
>100
25
50
100
>100
100
>100
aantiproliferative EC50 in a Jurkat cell growth assay
Based on observed low CL in mouse, compound 51 was
advanced to a mouse neutropenic thigh model (E. coli 25922).
Over a 20 hour IV infusion, an average blood concentration of 27
M of compound 51 was reached, but no reduction in CFU was
observed relative to vehicle treated animals. This is perhaps
unsurprising since the free fraction of 51 in mouse plasma is only
1.1%, and the unbound concentration of compound (0.3 M) is
below the in vitro MIC for the E. coli 25922 strain (3.1 M).
Unfortunately, a higher dose of 51 was not tolerated and we were
unable to evaluate the efficacy of an inhibitor where unbound
exposures were commensurate with the in vitro MIC. The
mechanism of toxicity observed at higher doses is unknown.
Further improvements in MIC may be required to evaluate the in
vivo efficacy of LspA inhibition.
functionality, and through modification of physicochemical
properties with the inclusion of an additional charged residue, we
were able to obtain potent compounds with good in vivo PK
properties. Optimized compounds demonstrate broad spectrum
activity against Gram-negative bacterial pathogens. Despite the
significant improvements in activity and in vivo stability, we were
unable to achieve free concentrations in mice sufficient to
demonstrate in vivo efficacy of LspA inhibition. We anticipate that
optimization of this scaffold to achieve further improvements in
activity and metabolic stability will enable us to demonstrate the
promise of LspA inhibition as a new mode of action against Gram-
negative bacteria.
Declaration of Competing Interest
In conclusion, the natural product globomycin was optimized
to improve activity against wild-type E. coli, informed by the
published co-crystal structure with LspA. While there are polar
residues in the bottom of the active site that interact with backbone
amides in the globomycin scaffold, we found that the overall
lipophilic environment restricted options for optimization. Aside
from a productive salt-bridge that was created with the catalytic
aspartate residues, improvements in specific interactions with the
protein remained elusive. Through replacement of the ester
Authors H.P., M-G.B., D.J.B., G.M.C., Y-C.C, J.C., G.D, S.L.,
R.P., C.S., M.X., Y.X., H.Z. S.B.K, and E.J.H. are employees of
Genentech, Inc.
Acknowledgments
We gratefully acknowledge the Genentech and Pharmaron
analytical chemistry groups as a whole, and Won Choi and Kewei

Through Design of a Robust in vitro HT Platform. Cell Chem. Biol.
2018, 1–21.
12. Kiho, T.; Nakayama, M.; Yasuda, K.; Miyakoshi, S.; Inukai, M.;
Kogen, H. Synthesis and Antimicrobial Activity of Novel
Globomycin Analogues. Bioorg. Med. Chem. Lett. 2003, 13 (14),
2315–2318.
Xu in particular, for their expert contributions in compound
purification and purity analysis. We thank Zhengyin Yan for
advice and discussion on DMPK methods.
Supplementary Material
13. Kiho, T.; Nakayama, M.; Yasuda, K.; Miyakoshi, S.; Inukai, M.;
Kogen, H. Structure–Activity Relationships of Globomycin
Analogues as Antibiotics. Bioorg. Med. Chem. 2004, 12 (2), 337–
361.
14. Vogeley, L.; Arnaout, El, T.; Bailey, J.; Stansfeld, P. J.; Boland, C.;
Caffrey, M. Structural Basis of Lipoprotein Signal Peptidase II
Action and Inhibition by the Antibiotic Globomycin. Science 2016,
351 (6275), 876–880.
Supplementary data associated with this article can be found
online at XXX and includes: Supplementary Tables S1 through S5;
experimental methods describing determination of MICs, plasma
protein binding, plasma stability, liver microsome and hepatocyte
stability, MDCK permeability, and LogD; analytical data, purity
determination, and synthetic methods for lead compounds.
15. Zgurskaya, H. I.; López, C. A.; Gnanakaran, S. Permeability Barrier
of Gram-Negative Cell Envelopes and Approaches to Bypass It.
ACS Infect. Diseases 2015, 1 (11), 512–522.
References and notes
16. Stoll, H.; Dengjel, J.; Nerz, C.; Götz, F. Staphylococcus Aureus
Deficient in Lipidation of Prelipoproteins Is Attenuated in Growth
and Immune Activation. Infect. Immun. 2005, 73 (4), 2411–2423.
17. Kogen, H.; Kiho, T.; Nakayama, M.; Furukawa, Y.; Kinoshita, T.;
Inukai, M. Crystal Structure and Total Synthesis of
1.
2.
Ventola, C. L. The Antibiotic Resistance Crisis: Part 1: Causes and
Threats. P. T. 2015, 40 (4), 277–283.
Arnaout, El, T.; Soulimane, T. Targeting Lipoprotein Biogenesis:
Considerations Towards Antimicrobials. Trends in Biochemical
Sciences 2019, 1–15.
Xia, J.; Feng, B.; Wen, G.; Xue, W.; Ma, G.; Zhang, H.; Wu, S.
Bacterial Lipoprotein Biosynthetic Pathway as a Potential Target
for Structure-Based Design of Antibacterial Agents. CMC 2019, 26,
1–18.
Tokunaga, M; Tokunaga, H; Wu, H.C.; Post-translational
modification and processing of Escherichia coli prolipoprotein in
vitro. Proc Natl Acad Sci USA 1982, 79, 2255–2259.
Zwiebel, L. J.; Inukai, M.; Nakamura, K.; Inouye, M. Preferential
Selection of Deletion Mutations of the Outer Membrane
Lipoprotein Gene of Escherichia Coli by Globomycin. J. Bacteriol.
1981, 145 (1), 654–656.
Hussain, M.; Ichihara, S.; Mizushima, S. Accumulation of
glyceride-containing precursor of the outer membrane lipoprotein
in the cytoplasmic membrane of Escherichia coli treated with
globomycin. J. Biol. Chem. 1980, 255, 3707-3712.
Inukai, M.; Takeuchi, M.; Shimizu, K.; Arai, M. Mechanism of
Action of Globomycin. J. Antibiotics 1978, 31 (11), 1203–1205.
Xiao, Y.; Gerth, K.; Müller, R.; Wall, D. Myxobacterium-Produced
Antibiotic TA (Myxovirescin) Inhibits Type II Signal Peptidase.
Antimicrob Agents Chemother 2012, 56 (4), 2014–2021.
Inukai, M.; Nakajima, M.; Osawa, M.; Haneishi, T.; Arai, M.
Globomycin, a New Peptide Antibiotic with Spheroplast-Forming
Activity. II. Isolation and Physico-Chemical and Biological
Characterization. J. Antibiotics 1978, 31 (5), 421-425.
Globomycin:
Establishment of Relative and Absolute
Configurations. J. Am. Chem. Soc. 2000, 122 (41), 10214–10215.
18. Sarabia, F.; Chammaa, S.; García-Ruiz, C. Solid Phase Synthesis of
Globomycin and SF-1902 a 5. J. Org. Chem. 2011, 76 (7), 2132–
2144.
19. Oppolzer, W.; Lienard, P. Efficient Asymmetric Synthesis of Anti-
Aldols From Bornanesultam Derived Boryl Enolates. Tett. Lett.
1993, 34, 4321–4324.
20. Thern, B.; Rudolph, J.; Jung, G. Triphosgene as Highly Efficient
Reagent for the Solid-Phase Coupling of N-Alkylated Amino
Acids—Total Synthesis of Cyclosporin O. Tet. Lett. 2002, 43,
5013–5016.
21. Shiina, I.; Fukui, H.; Sasaki, A. Synthesis of Lactones Using
Substituted Benzoic Anhydride as a Coupling Reagent. Nat. Protoc.
2007, 2 (10), 2312–2317.
22. Lomize, M. A.; Pogozheva, I. D.; Joo, H.; Mosberg, H. I.; Lomize,
A. L. OPM Database and PPM Web Server: Resources for
Positioning of Proteins in Membranes. Nucleic Acids Res. 2012, 40
(Database issue), D370–D376.
23. Brown, D. G.; May-Dracka, T. L.; Gagnon, M. M.; Tommasi, R.
Trends and Exceptions of Physical Properties on Antibacterial
Activity for Gram-Positive and Gram-Negative Pathogens. J. Med.
Chem. 2014, 57 (23), 10144–10161.
3.
4.
5.
6.
7.
8.
9.
10. Nakajima, M.; Inukai, M.; Haneishi, T.; Terahara, A.; Arai, M.;
Kinoshita, T.; Tamura, C. Globomycin, a New Peptide Antibiotic
with Spheroplast-Forming Activity. III. Structural Determination of
Globomycin. J. Antibiotics 1978, 31 (5), 426-432.
11. Kitamura, S.; Owensby, A.; Wall, D.; Wolan, D. W. Lipoprotein
Signal Peptidase Inhibitors with Antibiotic Properties Identified