Novel antibody-antibiotic conjugate eliminates intracellular S. aureus

Article abstract of DOI:10.1038/nature16057

Staphylococcus aureus is considered to be an extracellular pathogen. However, survival of S. aureus within host cells may provide a reservoir relatively protected from antibiotics, thus enabling long-term colonization of the host and explaining clinical f

Full text of DOI:10.1038/nature16057

ARticLE
doi:10.1038/nature16057
Novel antibody–antibiotic conjugate
eliminates intracellular S. aureus
Sophie M. Lehar1, thomas Pillow2, Min Xu3, Leanna Staben2, Kimberly K. Kajihara1, Richard Vandlen4, Laura DePalatis4,
helga Raab4, Wouter L. hazenbos1, J. hiroshi Morisaki1, Janice Kim3, Summer Park3, Martine Darwish4, Byoung-chul Lee4,
hilda hernandez5, Kelly M. Loyet5, Patrick Lupardus6, Rina Fong6, Donghong Yan3, cecile chalouni7, Elizabeth Luis4,
Yana Khalfin5, Emile Plise8, Jonathan cheong8, Joseph P. Lyssikatos2, Magnus Strandh9, Klaus Koefoed9, Peter S. Andersen9,
John A. Flygare2, Man Wah tan1, Eric J. Brown1 & Sanjeev Mariathasan1
Staphylococcus aureus is considered to be an extracellular pathogen. However, survival of S. aureus within host cells
may provide a reservoir relatively protected from antibiotics, thus enabling long-term colonization of the host and
explaining clinical failures and relapses after antibiotic therapy. Here we confirm that intracellular reservoirs of S. aureus
in mice comprise a virulent subset of bacteria that can establish infection even in the presence of vancomycin, and we
introduce a novel therapeutic that effectively kills intracellular S. aureus. This antibody–antibiotic conjugate consists
of an anti-S. aureus antibody conjugated to a highly efficacious antibiotic that is activated only after it is released in
the proteolytic environment of the phagolysosome. The antibody–antibiotic conjugate is superior to vancomycin for
treatment of bacteraemia and provides direct evidence that intracellular S. aureus represents an important component
of invasive infections.
S. aureus is the leading cause of bacterial infections in humans world- Intracellular MRSA are protected from antibiotics
wide and represents a major health problem in both hospital and To confirm the hypothesis that mammalian cells provide a protective
community settings1. However, S. aureus is not exclusively a patho- niche for S. aureus in the presence of antibiotic therapy, we compared
gen and commonly colonizes the anterior nares and skin of healthy the efficacy of three major antibiotics that are currently used as the
individuals. When infection does occur, the most serious infections standard of care for invasive MRSA infections (vancomycin, dapto-
such as endocarditis, osteomyelitis, necrotizing pneumonia and sepsis mycin and linezolid) against extracellular planktonic bacteria versus
take hold after dissemination of the bacteria into the bloodstream2. bacteria sequestered inside murine macrophages (Table 1). All three
Over the last several decades, infection with S. aureus has become antibiotics failed to kill a highly virulent community-acquired MRSA
increasingly difficult to treat due to the emergence and rapid spread of strain, USA300, sequestered inside macrophages exposed to clinically
methicillin-resistant S. aureus (MRSA), which is resistant to all known achievable concentrations of the antibiotics, consistent with other stud-
β-lactam antibiotics3. Alarmingly, reduced susceptibility to vancomycin ies showing that the majority of existing antibiotics are inefficient at
and resistance to linezolid and daptomycin have already been reported killing intracellular S. aureus both in vitro21 and in vivo22.
in MRSA clinical strains4.
Investigations going back at least 50 years have revealed that Intracellular S. aureus spreads infection
S. aureus is able to invade and survive inside mammalian cells, includ- To compare directly the virulence of intracellular bacteria versus
ing the phagocytic cells that are responsible for bacterial clearance5–11
.
free-living planktonic bacteria and to determine whether the intracel-
S. aureus is taken up by host phagocytic cells, primarily neutrophils and lular bacteria are able to establish infection in the presence of vancomy-
macrophages, within minutes after intravenous infection12. While the cin in vivo, mice were infected with equivalent doses of S. aureus taken
majority of the bacteria are effectively killed by these cells, incomplete directly from broth culture or bacteria sequestered inside host peritoneal
clearance of S. aureus inside blood-borne phagocytes can allow these
infected cells to act as ‘Trojan horses’ for dissemination of the bacteria
Table 1 | Antibiotic minimum inhibitory concentrations for MRSA.
away from the initial site of infection. Indeed, patients with normal neu-
Antibiotics
Extracellular MRSA Intracellular MRSA
Serum Cmax
trophil counts may be more prone to disseminated disease than those
with reduced neutrophil counts5,13,14. Once delivered to the tissues,
S. aureus can invade various non-phagocytic cell types, and intracellular
S. aureus in tissues is associated with chronic or recurrent infections
including osteomyelitis15, recurrent rhinosinusitis16, pulmonary infec-
tions17 and endocarditis18. In addition, intracellular S. aureus has been
shown to undermine innate immune responses and induce subsequent
destruction of neutrophils19,20. Together, these data suggest that ablat-
ing intracellular S. aureus is key to clinical success, but until now there
has been no way to test this hypothesis directly.
(MIC (μg ml−1))
(MIC (μg ml−1))
>100
(μg ml−1
50
)
Vancomycin
Daptomycin
Linezolid
1
4
>100
60
0.3
>20
20
Rifampicin
0.004
50
20
Extracellular MIC is the minimum antibiotic dose that prevented growth of MRSA seeded in
trypticase soy broth. Intracellular MIC is the minimum dose that prevented survival of MRSA
sequestered inside murine macrophages (> indicates continued growth at the highest antibiotic
dose tested). Data are summarized from three different experiments. Serum Cmax, the expected
serum concentrations for clinically relevant antibiotics37
.
1Infectious Diseases Department, Genentech Inc., South San Francisco, California 94080, USA. 2Medicinal Chemistry Department, Genentech Inc., South San Francisco, California 94080, USA.
3Translational Immunology Department, Genentech Inc., South San Francisco, California 94080, USA. 4Protein Chemistry Department, Genentech Inc., South San Francisco, California 94080,
USA. 5Biochemical and Cellular Pharmacology Department, Genentech Inc., South San Francisco, California 94080, USA. 6Structural Biology Department, Genentech Inc., South San Francisco,
California 94080, USA. 7Pathology Department, Genentech Inc., South San Francisco, California 94080, USA. 8Drug metabolism and Pharmacokinetics Department, Genentech Inc., South San
Francisco, California 94080, USA. 9Symphogen A/S, Pederstrupvej 93, DK-2750 Ballerup, Denmark.
0 0 M o n t h 2 0 1 5
| V o L 0 0 0 | n A t U R E | 1
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
Infection with:
a
b
we developed a novel therapeutic that is activated specifically inside
mammalian cells that have taken up S. aureus. The antibody–antibiotic
conjugate (AAC) consists of an anti-S. aureus antibody (THIOMAB,
engineered to contain unpaired cysteines) covalently linked via the
introduced cysteines to a highly efficacious antibiotic using a cath-
epsin-cleavable linker containing a novel quaternary ammonium salt
(Fig. 2a). The resulting AAC has no direct antibacterial activity when
bound to planktonic S. aureus and does not diffuse into mamma-
lian cells, owing to the size of the antibody. However, when AAC-
opsonized bacteria are taken up by host cells, intracellular proteases
cleave the linker and readily release the antibiotic in its active form23
(Fig. 2b). Depending on the antibody specificity, we found that thou-
sands of AACs can bind to a single bacterium and deliver a sufficient
concentration of free antibiotic within the phagosome to result in
bacterial killing.
The antibody and antibiotic components were carefully chosen
and optimized for maximal efficacy. A panel of more than 40 anti-
S. aureus antibodies were cloned and purified from B cells derived
from the peripheral blood of patients recovering from various S. aureus
infections and screened for binding to a panel of clinically relevant
S. aureus strains and to USA300 isolated directly ex vivo from the kid-
neys of infected mice. We found the greatest extent of binding with
antibodies directed against wall-teichoic acids (WTAs), pathogen-
specific polyanionic glycopolymers that are connected to the thick pep-
tidoglycan layers of Gram-positive bacteria (Fig. 2c). S. aureus produces
WTAs composed of phospho-ribitol repeating units that are further
modified by either α- or β-O-linked N-acetylglucosamine (GlcNAc)
sugars mediated by TarM or TarS glycosyltransferases, respectively24.
A human immunoglobulin G1 (IgG1) that recognizes β-O-linked
GlcNAc sugar modifications on WTA bound to all S. aureus strains
tested. Monoclonal antibodies recognizing the α-O-linked GlcNAc
bound well to S. aureus strains cultured in vitro; however, expression
of the α-O-linked GlcNAc was absent on some S. aureus isolates.
On those strains that co-expressed the α- and β-O-linked GlcNac,
β-specific antibodies yielded consistently higher binding to in vivo-
derived MRSA (Fig. 2c). Antigen specificity of the antibodies was con-
firmed by genetic means, such that antibodies against α- or β-GlcNAc
sugar modifications on WTAs failed to bind to S. aureus strains lacking
the TarM or TarS glycosyltranferase, respectively (Fig. 2d). We esti-
mate that ~15,000 β-WTA-specific antibodies can bind to protein-A-
deficient S. aureus USA300 by measuring binding of radiolabelled
antibodies (data not shown).
To characterize antibody binding to WTA at the molecular level
further, we co-crystallized the Fab fragment from an anti-β-WTA
antibody in complex with a synthetic form of the minimal repeating
β-WTA unit (C2 carbon of ribitol-1-phosphate linked to the C1 posi-
tion in GlcNAc via a β-glycosidic bond) and determined a crystal struc-
ture of the complex at 1.7 Å resolution (Extended Data Table 1). The
structure reveals how the minimal β-WTA epitope binds to the com-
plementarity determining region (CDR) of the Fab antibody fragment
(Fig. 2e). The heavy chain H1 and H3 CDR loops form a binding site for
the GlcNAc, with the GlcNAc hexose ring stacking against the Trp 33
indole side chain. Importantly, the antibody has an extended L1 loop
containing two arginine residues (Arg 27d and 28) that interact with
the phosphate group on the ribitol backbone. This arginine ‘tweezers’
motif triangulates the ribitol phosphodiester backbone in relation to the
GlcNAc moiety, and is probably a key reason for the β-anomer-specific
recognition of GlcNAc-WTA by the antibody.
An extensive set of criteria was used to choose the optimal antibiotic
for the AAC platform. We chose the rifamycin class of antibiotics for
their high potency, unaltered bactericidal activity in low phagolysoso-
mal pH, and their ability to withstand intracellular insults. A rifamycin
derivative (a rifalogue) that allowed connection to the linker antibody
through a tertiary amine (T. Pillow, manuscript in preparation) had
optimum physicochemical properties to allow for efficient bacte-
rial killing. The chemical groups known to make direct contact with
Planktonic
bacteria
Intracellular
bacteria
108
107
106
105
104
103
102
+ PBS
Planktonic
bacteria
+ Vanco
+ PBS
Intracellular
bacteria in
host cells
Planktonic
bacteria
+ Vanco
PBS
Vanco
PBS
Vanco
+ Vanco
c
d
Planktonic bacteria
Planktonic bacteria
Intracellular bacteria
1011
1010
109
108
107
106
105
104
103
102
101
+ MG63 cells
+ MG63 cells
64,815
30,093
105
104
103
102
101
100
Intracellular bacteria
+ vanco
259
ND
Planktonic bacteria
s
s
l
+ vanco
e
n
cel
No
None
MG63
+
+
MG63 cell
0
5
10
15
20
25
Planktonic bacteria
Intracellular bacteria
Time (h)
Figure 1 | Intracellular MRSA are protected from vancomycin.
a, Experimental design for generating planktonic versus intracellular
bacteria for infection and treatment with vancomycin (vanco).
b, Bacterial loads in kidney, 4 days after infection. c.f.u., colony-forming
units. Each point represents data from a single mouse (n =5 mice per
group). c, Planktonic or intracellular bacteria were suspended in media
containing vancomycin and cultured alone or on a monolayer of MG63,
and surviving bacteria were enumerated 1 day later. Error bars represent
means standard deviation (s.d.). d, Survival of MRSA cultured with
vancomycin with or without MG63 cells. Error bars show s.d. ND, none
detected. Dashed lines indicate the limit of detection. Representative of
three independent experiments.
macrophages and neutrophils (Fig. 1a). Mice infected with intracellu-
lar bacteria had equivalent or higher bacterial burdens in the kidneys
4 days after infection compared with those infected with planktonic
bacteria (Fig. 1b). Infection with intracellular bacteria also resulted in
more consistent colonization of the brain (Extended Data Fig. 1). To
characterize this observation further, we infected MG63 osteoblasts
with either planktonic MRSA or an equivalent number of intracellular
MRSA, in the presence of vancomycin. Planktonic bacteria exposed
to vancomycin alone were efficiently killed, (Fig. 1c). However, we
recovered a small number of surviving bacteria (approximately 0.06%
of input) associated with the MG63 cells 1 day after infection, which
had been protected from vancomycin by invasion into osteoblasts.
MRSA that were sequestered inside peritoneal cells showed a sig-
nificant increase in both survival and efficiency of MG63 infection
in the presence of vancomycin. About 15% of intracellular MRSA in
the leukocytes survived under vancomycin exposure identical to that
which sterilized cultures of planktonic bacteria. Intracellular S. aureus
that were sequestered in MG63 cells (Fig. 1d), primary human brain
endothelial cells and A549 bronchial epithelial cells (Extended Data
Fig. 2) were also able to increase by almost tenfold over a 24h period
under constant exposure to a concentration of vancomycin that killed
free-living bacteria. Together, these data suggest that intracellular reser-
voirs of MRSA in myeloid cells can promote dissemination of infection
to new sites in the presence of active antibiotic treatment, and that
intracellular growth can occur in a variety of cell types despite constant
antibiotic therapy.
Designing the antibody–antibiotic conjugate
These findings suggested that therapies aimed at eliminating intracel-
lular bacteria may improve clinical success. To test this conclusively,
2
| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
a
O
S. aureus
opsonization
b
Phagolysosomal
protease cleavage
OMe
O
O
N
OAc
OH
OH
MRSA
OH
HO
O
O
N
O
HN
O
O
O
N
S
H
N
N
N
H
N
H
O
O
Antibiotic
(rifalogue)
NH
O
NH2
Anti-S. aureus
Fc-mediated uptake Release of antibiotic
Intracellular killing
Intracellular protease-sensitive
linker
mAb
S. aureus escape
of bacteria
inside phagolysosome of bacteria
Linker drug =
L1
K30
100
80
60
40
20
0
100
d
e
Ribitol
c
80
60
40
20
0
W50
10,000
1,000
100
WT
2-N-acetyl
glucosamine
Anti-β-WTA mAb
Anti-α-WTA mAb
R28
L32
1
1
1
10
10
10
100 1,000 10,000
1
1
1
10
10
10
100 1,000 10,000
100 1,000 10,000
100 1,000 10,000
G97
100
80
60
40
20
0
100
80
60
40
20
0
D96
R27d
F32
ΔtarM
ΔtarS
Anti-CMV-gD
(Isotype control)
H3
1-Phosphate
Y91
Y96
100 1,000 10,000
G95
100
80
60
40
20
0
100
80
60
40
20
0
10
H1
W33
L3
1
0.01 0.10
1
10
100
H35
100 1,000 10,000
Antibody concentration (μg ml–1
)
APC-A
APC-A
Anti-β-WTA mAb Anti-α-WTA mAb
f
g
i
107
h
1.5
1.0
0.5
0
AAC
AAC + cathepsin
1010
0.5
Rifampicin
Rifalogue
106
105
104
103
Second
antibiotic
added
None
Cipro
Cipro, then rifampicin
Cipro, then rifalogue
108
106
104
102
100
0.4
0.3
0.2
0.1
0
PBS
Rifampicin
Rifalogue
10–10 10–9 10–8 10–7 10–6
Antibiotic concentration (M)
0
4
8
12 16 20 24
0
2 4 6 8 10 20 40 60 80 100
Time of antibiotic exposure (h)
Time of antibiotic exposure (h)
AAC (μg ml–1
)
Figure 2 | AAC design. a, Model of AAC (not drawn to scale).
b, Mechanism of AAC action. c, Binding of Alexa-488 anti-β-GlcNAC
WTA monoclonal antibody (mAb) or anti-α-GlcNAC WTA
monoclonal antibody, or isotype control antibody, anti-cytomegalovirus
glycoprotein-D (gD) to USA300 isolated from infected kidneys (n =3).
MFI, mean fluorescence intensity. d, Binding of anti-GlcNAC WTA
antibodies (red) or isotype control (grey) to protein-A-deficient USA300
lacking tarM or tarS (n =3). WT, wild type. e, Crystal structure of
anti-β-GlcNAc WTA Fab bound to a synthetic minimal β-WTA unit.
Antibody light chain (pink) and heavy chain (blue) are shown. f, MIC
determination for rifampicin and rifalogue on USA300 (n =5). g, Survival
of stationary phase USA300 incubated with 1 ×10−6 M rifampicin or
rifalogue (n =4). h, USA300 bacteria were incubated without antibiotic
(black) or with 3 μg ml−1 ciprofloxacin (Cipro; green, red and grey).
1 μg ml−1 of rifalogue (red) or rifampicin (grey) was added as indicated
(n =3). i, Intact AAC does not kill planktonic bacteria but does after pre-
treatment with cathepsin-B (n =3). g–i, Error bars show s.d. for triplicate
samples (n =biological repeats).
bacterial RNA polymerase are conserved between rifampicin and the fluorescent molecules are covalently associated, fluorescence at 488 nm
rifalogue25. Consistent with this, the minimum inhibitory concentra
-
is quenched; upon cleavage of the linker, quenching is lost. Video
tion (MIC) for the rifalogue is identical to rifampicin (4×10−9 M), as microscopy demonstrated that the linker was cleaved within minutes
measured in a conventional growth inhibition assay using planktonic of uptake of the MRSA opsonized with the FRET conjugate (Fig. 3a).
bacteria (Fig. 2f). The rifalogue, but not rifampicin, resulted in a more We confirmed by mass spectrometry analysis that free rifalogue was
than 1,000-fold decrease in the number of viable, but non-replicating, released inside macrophages after uptake of MRSA opsonized with
bacteria after overnight incubation in minimal PBS buffer (Fig. 2g). The active AACs, but not when opsonized with a non-cleavable version
rifalogue was also able to kill classically defined persister cells, bacteria of the AAC, which was prepared by replacement of the natural amino
that presumably enter a dormant state to survive antibiotic treatment acid l-citrulline with d-citrulline at the P1 position, making the linker
(for example, ciprofloxacin) of growing cultures26 (Fig. 2h), while uncleavable by cathepsins (Fig. 3b).
treatment with rifampicin had no effect on recovery of persister cells,
in agreement with previous observations26. Additionally, rifalogue, AACs eradicate intracellular S. aureus infections
but not rifampicin, was able to kill non-dividing MRSA sequestered When the AAC was used to opsonize MRSA, it readily killed bacteria
inside murine macrophages, in part due to its capacity to accumulate inside the macrophages, whereas the same bacteria opsonized with
rapidly in mammalian cells (Extended Data Fig. 3). As predicted, the non-cleavable β-WTA AAC or monoclonal antibody alone survived
rifalogue had no ability to kill extracellular bacteria when linked to the (Fig. 3c). AAC-opsonized MRSA were killed inside every cell type
anti-β-WTA monoclonal antibody in the AAC format, but was fully tested, including human macrophages, endothelial and epithelial cell
bactericidal after the AAC was treated with cathepsin B to release the lines (Fig. 3d). These data demonstrate that the AAC releases active
active antibiotic (Fig. 2i).
antibiotic and kills S. aureus only after internalization inside host
To confirm that the linker used in the AAC is cleaved after internali- cells, providing a unique tool to assess the importance of intracellular
zation of the AAC inside cells, we examined cleavage of the linker with S. aureus in various settings. To determine whether specifically tar-
a fluorescence resonance energy transfer (FRET)-based probe consist- geting intracellular S. aureus could prevent cell-to-cell transfer of bac-
ing of the same anti-MRSA antibody conjugated to two dye molecules teria, infected peritoneal cells were incubated with MG63 osteoblast
that were separated by the same linker used in the AAC27. When the cells in the presence of vancomycin (Fig. 3e). Addition of AAC to the
0 0 M o n t h 2 0 1 5
| V o L 0 0 0 | n A t U R E | 3
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
a AAC-coated S. aureus
S. aureus
opsonization
Cleaved by intracellular
protease
MRSA
MRSA
FRET
b
TAMRA
12 s
Alexa-488
c
Overlay
Macrophages
Murine macrophages
Murine macrophages
15
10
5
106
105
104
103
102
101
100
1 min
25 μm
25 μm
0
μg ml–1
2
5 min
20 min
50 50 10
mAb
2
50 10
AAC
1
3
1
3
1
3 hrs
only
AAC
(cleavable) (non-cleavable)
e
25 μm
25 μm
AAC
or
rifampicin
Pre-infected
myeloid cells
Human
105
104
103
102
101
100
macrophages
HL-60
MG63
A549
HBMECs
MG63 cells + vanco
d
105
104
103
102
101
100
104
103
102
101
100
104
104
103
102
101
100
105
104
103
102
101
103
102
101
100
μg ml–1
50
50
AAC
50
mAb
mAb
only
mAb AAC
mAb AAC
mAb AAC
mAb AAC
mAb AAC
+ rifampicin (5 μg ml–1
)
Figure 3 | AAC linker is cleaved after internalization of bacteria.
a, Live cell imaging monitoring cleavage of AAC linker in macrophages
with FRET-based antibody conjugate (representative of three fields).
TAMRA, tetramethylrhodamine. b, Mass spectrometric quantification
of released antibiotic inside macrophages from AAC made with standard
(cleavable) or non-cleavable linker. mAb, monoclonal antibody.
c, d, USA300, opsonized with antibody, AAC, or non-cleavable AAC,
enumerated 48 h after incubation with indicated cells lines. HBMECs,
human brain microvascular endothelial cells. Horizontal line indicates
the limit of detection. e, Intracellular USA300 as in Fig. 1a were added to
a monolayer of MG63 with antibody, AAC, or a mixture of antibody plus
rifampicin in media containing vancomycin (vanco). Surviving bacteria
were enumerated 24 h later. b–e, Error bars represent means s.d. from
triplicate wells. Data are representative of 2–4 biological replicates.
culture media resulted in a significant reduction in the number of viable due to antibiotic escape through survival inside host cells. To test this
intracellular bacteria recovered 1 day after infection. Similar treatment directly, mice were treated with a single dose of AAC 24h after infec-
with the unconjugated anti-MRSA antibody plus a tenfold higher molar tion. A single dose of AAC was efficacious and proved superior to twice
concentration of rifampicin, purported to be one of the best-known daily vancomycin treatment initiated at the same time (Fig. 4c and
antibiotics for treatment of intracellular bacteria28, had very marginal Extended Data Fig. 6a).
efficacy. To confirm that targeting intracellular S. aureus could also
Mouse serum has no appreciable anti-S. aureus antibody. To deter-
prevent transfer of infection from intracellular reservoirs in vivo, mice mine whether endogenous anti-WTA antibodies found in normal
were infected by intravenous injection with equivalent doses of plank- human serum might compete for binding with the AAC, SCID mice
tonic or intracellular MRSA and then treated with PBS or vancomycin, were reconstituted with physiological levels of human IgG by daily
with or without a single dose of AAC. In the presence of vancomycin treatment with IGIV. The resulting SCID-huIgG mice had sustained
only intracellular bacteria were able to efficiently colonize the brain levels of at least 10 mg ml−1 of human IgG in the serum and were
(Extended Data Fig. 1); however, treatment with a single dose of AAC equally susceptible to infection with MRSA compared to untreated
effectively eliminated the bacteria that had escaped vancomycin treat- controls (data not shown). Despite the presence of potentially compet-
ment (Extended Data Fig. 4)
ing antibodies, a single dose of anti-MRSA AAC administered at 24h
after infection with MRSA was still more efficacious than vancomycin
(Fig. 4d). An AAC made with an irrelevant antibody (anti-cytomegalo-
AAC is superior to vancomycin in vivo
To determine the role of intracellular infection in bacteraemia, AACs virus glycoprotein D antibody)—which could still bind to protein A
were tested in an intravenous infection model. First, we tested the on MRSA—had much less efficacy. AACs made with anti-β-GlcNAc
efficacy of unconjugated anti-MRSA antibodies by treating mice WTA antibodies appeared to be more efficacious than those made with
with either anti-β-WTA antibodies, anti-α-WTA antibodies or intra- anti-α-GlcNAc WTA antibodies (Extended Data Fig. 6b), suggesting
venous immunoglobulin (IGIV), a pooled immunoglobulin prepa- that the extent of antibody binding determines antibiotic delivery.
ration from ~10,000 humans, 1h before intravenous infection with Treatment with the AAC was more effective than treatment with an
wild-type USA300 MRSA. S. aureus is a common colonizer of human equivalent dose or repeated dosing of unconjugated rifalogue, and
skin and mucosal surfaces and preliminary analysis of multiple sources release of the antibiotic from the AAC was essential for efficacy, as
of human serum, including IGIV, demonstrated that human serum an AAC generated with the non-cleavable linker was not efficacious
contains approximately 300μg ml−1 of anti-S. aureus antibodies, of in vivo (Fig. 4e).
which ~70% are directed towards the GlcNAc modifications of WTA
(Extended Data Fig. 5). Even in this prophylactic treatment setting, Discussion
unconjugated antibodies did not have any efficacy in preventing infec- Treatment of patients with invasive MRSA infections with conven-
tion, whereas treatment with vancomycin eliminated detectable bac- tional antibiotics results in failure rates of up to 50% (refs 29–32),
teria (Fig. 4a). However, the efficacy of vancomycin was limited when in most cases without measurable outgrowth of antibiotic-resistant
treatment was initiated several hours after infection was established strains. Survival of antibiotic-susceptible subpopulations of S. aureus
(Fig. 4b). In at least one model of bacteraemia, 95% of the bacteria in the presence of antibiotic treatment is well documented and is
in blood are found to be associated with neutrophils within 15 min probably due to multiple factors, including survival of persister bac-
(ref. 6), suggesting that failure of vancomycin in this model could be teria that are not actively dividing and poor antibiotic exposure of
4
| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
a
b
c
109
108
107
106
105
104
103
102
+ 1 dead
+ 1 moribund
109
108
107
106
105
104
103
102
+ 3 moribund
+ 1 dead
109
108
107
106
105
104
103
102
Saline
Anti-β-WTA Anti-α-WTA
mAb
IGIV
Vanco
Saline Anti-β-WTA
Vanco
AAC
1
7
24
h
Saline
mAb
mAb
only
Initiation of vanco treatment
after infection
*
d
e
**
***
109
108
107
106
105
104
103
102
*
109
*
108
107
106
105
104
103
102
Saline
AAC
AAC
(non-
cleavable)
Rifalogue
(once)
Rifalogue
(daily)
Anti-β-WTA Vanco
mAb
only
Anti-
MRSA
AAC
Anti-CMV
AAC
Figure 4 | AAC is a more effective treatment than vancomycin after
intravenous infection. a, Wild-type (WT) mice (n =8 per group) were
treated with 50 mg kg−1 of the indicated anti-MRSA antibodies 1 h before
MRSA infection or twice daily with 110 mg kg−1 vancomycin (Vanco).
b, Treatment of wild-type mice (n =5 per group) with 110 mg kg−1
vancomycin (twice daily) was initiated either at 1 h, 7 h or 24 h after
infection. c, Wild-type mice (n =5 per group) were treated with
saline, anti-β-WTA antibody used in the AAC (monoclonal antibody
(mAb)), vancomycin (twice daily), or anti-MRSA AAC (a single dose of
50 mg kg−1) starting 24 h after infection. d, e, SCID mice (d, 5 mice per
group; e, 10 mice per group) were injected with human IgG to achieve
a concentration of 10 mg ml−1, then infected with MRSA. Treatment as
indicated was begun 24 h after infection. Vancomycin, 110 mg kg−1 twice
daily; AAC, 50 mg kg−1 a single dose on day 1; rifalogue, 0.5 mg kg−1
either once on day 1, or once daily (days 1–3). a–e, Bacteria in kidneys
were determined 4 days after infection. Schematic of each experimental
design is shown in Extended Data Fig. 7. Each point shows data from a
single animal. Bars show geometric mean. Dashed lines indicate limit of
detection. Mann–Whitney U-test. *P <0.05; **P <0.01; ***P <0.005.
bacteria in protected niches33–35. However, it has been difficult to extent to which other common human pathogens beyond S. aureus
determine conclusively what role intracellular S. aureus plays in dis- depend on this intracellular protected niche merits investigation; the
ease pathogenesis and antibiotic failure. To address directly the extent AAC platform promises to enhance antibiotic efficacy against these
to which this ability to survive and replicate inside cells contributes infectious diseases.
to the failure of conventional antibiotics, we developed a novel ther-
Online Content Methods, along with any additional Extended Data display items and
Source Data, are available in the online version of the paper; references unique to
these sections appear only in the online paper.
apeutic in the form of an AAC that is specifically designed to be
activated only within host cells. We found that a single dose of AAC
treatment is effective in a murine model of bacteraemia, in which
Received 15 December 2014; accepted 6 October 2015.
vancomycin, the current standard of care for MRSA infection, failed
Published online 4 November 2015.
(Fig. 4). This suggests that ablation of the intracellular pool of path-
1. Diekema, D. J. et al. Survey of infections due to Staphylococcus species:
ogens is key for clinical success in the treatment of potential fatal
frequency of occurrence and antimicrobial susceptibility of isolates collected
in the United States, Canada, Latin America, Europe, and the Western Pacific
region for the SENTRY Antimicrobial Surveillance Program, 1997–1999. Clin.
Infect. Dis. 32 (suppl. 2), S114–S132 (2001).
MRSA infection.
In designing the AAC to test this hypothesis, we used an antibiotic
that kills intracellular bacteria. This rifalogue antibiotic also ablates
2. Lowy, F. D. Staphylococcus aureus infections. N. Engl. J. Med. 339, 520–532
non-replicating bacteria, and antibiotic-resistant persister cells. We
(1998).
posit that this characteristic, along with its ability to accumulate
3. Boucher, H. W. et al. Bad bugs, no drugs: no ESKAPE! An update from the
within the intracellular milieu, is essential for AAC success. When
the closely related rifampicin, which does not have these character-
istics, was conjugated to the same anti-WTA antibody it did not kill
intracellular bacteria and exhibited poor efficacy in vivo (Extended
Data Fig. 8). These studies show that arming antibodies with unique
bactericidal antibiotics can result in promising new therapies. Many
potent antibiotic-like compounds with the desired characteristics to
kill difficult-to-treat intracellular pathogens fail in clinical practice
owing to poor pharmacokinetic properties or undesired host toxic-
ity. The use of an AAC might be able to overcome these problems,
making it a novel platform for delivering potent antibacterial com-
pounds that may not have a suitable profile as unconjugated drugs.
Antibiotic failure has been associated with the ability of a variety of
bacterial pathogens to survive within host phagocytic cells36. The
Infectious Diseases Society of America. Clin. Infect. Dis. 48, 1–12 (2009).
4. Nannini, E., Murray, B. E. & Arias, C. A. Resistance or decreased susceptibility
to glycopeptides, daptomycin, and linezolid in methicillin-resistant
Staphylococcus aureus. Curr. Opin. Pharmacol. 10, 516–521 (2010).
5. Thwaites, G. E. & Gant, V. Are bloodstream leukocytes Trojan Horses for
the metastasis of Staphylococcus aureus? Nature Rev. Microbiol. 9,
215–222 (2011).
6. Rogers, D. E. & Tompsett, R. The survival of staphylococci within human
leukocytes. J. Exp. Med. 95, 209–230 (1952).
7. Gresham, H. D. et al. Survival of Staphylococcus aureus inside neutrophils
contributes to infection. J. Immunol. 164, 3713–3722 (2000).
8. Kapral, F. A. & Shayegani, M. G. Intracellular survival of staphylococci. J. Exp.
Med. 110, 123–138 (1959).
9. Anwar, S., Prince, L. R., Foster, S. J., Whyte, M. K. & Sabroe, I. The rise and rise of
Staphylococcus aureus: laughing in the face of granulocytes. Clin. Exp. Immunol.
157, 216–224 (2009).
10. Fraunholz, M. & Sinha, B. Intracellular Staphylococcus aureus: live-in and let die.
Front. Cell. Infect. Microbiol. 2, 43 (2012).
0 0 M o n t h 2 0 1 5
| V o L 0 0 0 | n A t U R E | 5
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
29. Kullar, R., Davis, S. L., Levine, D. P. & Rybak, M. J. Impact of vancomycin
exposure on outcomes in patients with methicillin-resistant Staphylococcus
aureus bacteremia: support for consensus guidelines suggested targets.
Clin. Infect. Dis. 52, 975–981 (2011).
30. Fowler, V. G. Jr et al. Daptomycin versus standard therapy for bacteremia and
endocarditis caused by Staphylococcus aureus. N. Engl. J. Med. 355, 653–665
(2006).
31. Yoon, Y. K., Kim, J. Y., Park, D. W., Sohn, J. W. & Kim, M. J. Predictors of
persistent methicillin-resistant Staphylococcus aureus bacteraemia in patients
treated with vancomycin. J. Antimicrob. Chemother. 65, 1015–1018 (2010).
32. Johnson, L. B., Almoujahed, M. O., Ilg, K., Maolood, L. & Khatib, R.
Staphylococcus aureus bacteremia: compliance with standard treatment,
long-term outcome and predictors of relapse. Scand. J. Infect. Dis. 35, 782–789
(2003).
33. Levin, B. R. Noninherited resistance to antibiotics. Science 305, 1578–1579
(2004).
34. Grant, S. S., Kaufmann, B. B., Chand, N. S., Haseley, N. & Hung, D. T. Eradication
of bacterial persisters with antibiotic-generated hydroxyl radicals. Proc. Natl
Acad. Sci. USA 109, 12147–12152 (2012).
35. Lewis, K. Persister cells. Annu. Rev. Microbiol. 64, 357–372 (2010).
36. Kaiser, P. et al. Cecum lymph node dendritic cells harbor slow-growing bacteria
phenotypically tolerant to antibiotic treatment. PLoS Biol. 12, e1001793
(2014).
37. Bryskier, A. Anti-MRSA agents: under investigation, in the exploratory phase
and clinically available. Expert Rev. Anti Infect. Ther. 3, 505–553 (2005).
11. Garzoni, C. & Kelley, W. L. Return of the Trojan horse: intracellular phenotype
switching and immune evasion by Staphylococcus aureus. EMBO Mol. Med. 3,
115–117 (2011).
12. Rogers, D. E. Studies on bacteriemia. I. Mechanisms relating to the persistence
of bacteriemia in rabbits following the intravenous injection of staphylococci.
J. Exp. Med. 103, 713–742 (1956).
13. Velasco, E. et al. Comparative study of clinical characteristics of neutropenic
and non-neutropenic adult cancer patients with bloodstream infections. Eur. J.
Clin. Microbiol. Infect. Dis. 25, 1–7 (2006).
14. Venditti, M. et al. Staphylococcus aureus bacteremia in patients with
hematologic malignancies: a retrospective case-control study. Haematologica
88, 923–930 (2003).
15. Bosse, M. J., Gruber, H. E. & Ramp, W. K. Internalization of bacteria by
osteoblasts in a patient with recurrent, long-term osteomyelitis. A case report.
J. Bone Joint Surg. Am. 87, 1343–1347 (2005).
16. Clement, S. et al. Evidence of an intracellular reservoir in the nasal mucosa of
patients with recurrent Staphylococcus aureus rhinosinusitis. J. Infect. Dis. 192,
1023–1028 (2005).
17. Jarry, T. M., Memmi, G. & Cheung, A. L. The expression of α-haemolysin is
required for Staphylococcus aureus phagosomal escape after internalization in
CFT-1 cells. Cell. Microbiol. 10, 1801–1814 (2008).
18. Que, Y. A. et al. Fibrinogen and fibronectin binding cooperate for valve infection
and invasion in Staphylococcus aureus experimental endocarditis. J. Exp. Med.
201, 1627–1635 (2005).
19. Greenlee-Wacker, M. C. et al. Phagocytosis of Staphylococcus aureus by human
neutrophils prevents macrophage efferocytosis and induces programmed
necrosis. J. Immunol. 192, 4709–4717 (2014).
20. Kobayashi, S. D. et al. Rapid neutrophil destruction following phagocytosis of
Staphylococcus aureus. J. Innate Immun. 2, 560–575 (2010).
21. Barcia-Macay, M., Seral, C., Mingeot-Leclercq, M. P., Tulkens, P. M. & Van
Bambeke, F. Pharmacodynamic evaluation of the intracellular activities of
antibiotics against Staphylococcus aureus in a model of THP-1 macrophages.
Antimicrob. Agents Chemother. 50, 841–851 (2006).
22. Sandberg, A., Hessler, J. H., Skov, R. L., Blom, J. & Frimodt-Møller, N.
Intracellular activity of antibiotics against Staphylococcus aureus in a mouse
peritonitis model. Antimicrob. Agents Chemother. 53, 1874–1883 (2009).
23. Dubowchik, G. M. et al. Cathepsin B-labile dipeptide linkers for lysosomal
release of doxorubicin from internalizing immunoconjugates: model studies of
enzymatic drug release and antigen-specific in vitro anticancer activity.
Bioconjug. Chem. 13, 855–869 (2002).
24. Winstel, V., Xia, G. & Peschel, A. Pathways and roles of wall teichoic acid
glycosylation in Staphylococcus aureus. Int. J. Med. Microbiol. 304, 215–221
(2014).
Acknowledgements This research used resources of the Advanced Photon
Source, a US Department of Energy (DOE) Office of Science User Facility
operated for the DOE Office of Science by Argonne National Laboratory under
contract no. DE-AC02-06CH11357.
Author Contributions S.M.L. designed and executed the in vitro and in vivo
analysis of the AAC mechanism of action. T.P., L.S. and J.A.F. designed and
synthesized antibiotics and linker drugs. M.X., J.K., S.P. and D.Y. designed
and analysed in vivo models for intravenous infection. H.R., L.D., M.D. and
R.V. designed and conjugated linker antibiotic to antibodies. K.K.K., W.L.H.,
J.H.M. and S.M. characterized the anti-MRSA antibodies. Y.K., H.H., K.M.L.,
E.P. and J.C. did mass spectrometry analysis of the rifalogues during in vitro
efficacy studies. P.L. and R.F. performed X-ray crystallography of anti-β-WTA
monoclonal antibody. J.P.L. designed the synthesis of β-phospho-ribitol. B.-C.L.
and C.C. characterized FRET constructs and helped with video microscopy. E.L.
determined the number of antibody-binding sites on MRSA. M.S., K.K. and P.S.A.
isolated anti-MRSA antibodies from patients. M.W.T. contributed to bacterial
genetics and data analysis. E.J.B. and S.M. initiated the project and S.M. led the
project. S.M.L., E.J.B. and S.M. composed the paper with input from all authors.
25. Campbell, E. A. et al. Structural mechanism for rifampicin inhibition of
bacterial RNA polymerase. Cell 104, 901–912 (2001).
26. Conlon, B. P. et al. Activated ClpP kills persisters and eradicates a chronic
biofilm infection. Nature 503, 365–370 (2013).
27. Fischer, R., Hufnagel, H. & Brock, R. A doubly labeled penetratin analogue as a
ratiometric sensor for intracellular proteolytic stability. Bioconjug. Chem. 21,
64–73 (2010).
28. Nielsen, S. L. & Black, F. T. Extracellular and intracellular killing in neutrophil
granulocytes of Staphylococcus aureus with rifampicin in combination
with dicloxacillin or fusidic acid. J. Antimicrob. Chemother. 43, 407–410
(1999).
Additional Information The structure of the anti-β-WTA Fab bound to the
synthetic WTA fragment (β-GlcNAc anomer) has been deposited in the
Protein Data Bank under accession number 5D6C. Reprints and permissions
information is available at www.nature.com/reprints. The authors declare
competing financial interests: details are available in the online version of
the paper. Readers are welcome to comment on the online version of the
paper. Correspondence and requests for materials should be addressed to
E.J.B. (brown.eric@gene.com) or S.M. (sanj@gene.com).
6
| n A t U R E | V o L 0 0 0 | 0 0 M o n t h 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
MeThODS
solution as a control for extracellular bacteria that were not subject to neutrophil
Ethics statement. All animal procedures were conducted under a protocol killing.
(#08–1990) approved by the Genentech Institutional Animal Care and Use Infection of osteoblasts, HBMEC and A549 cells. MG63 cell line (CRL-1427) and
Committee in an Association for Assessment and Accreditation of Laboratory A549 cells (CCL185) were obtained from ATCC and maintained in RPMI 1640
Animal Care (AAALAC)-accredited facility in accordance with the Guide for the tissue culture media supplemented with 10mM HEPES and 10% fetal calf serum
Care and Use of Laboratory Animals and applicable laws and regulations. For (RPMI-10). HBMEC cells (catalogue #1000) and ECM media (catalogue #1001)
cloning of antibodies from human B cells, informed written consent was obtained were obtained from ScienceCell Research Labs. The cells were used without fur-
from all donors and was provided in accordance with the Declaration of Helsinki. ther authentication or testing for mycoplasma contamination. Cells were plated in
Approval was obtained from the health research ethics committee of Denmark 24-well tissue culture plates and cultured to obtain a confluent layer. On the day
through the regional committee for the Capital Region of Denmark.
of the experiment, the cells were washed once in RPMI (without supplements).
Bacterial strains. All in vivo experiments were done with MRSA-USA300 NRS384 MRSA or infected peritoneal cells were diluted in complete RPMI-10 and van-
obtained from NARSA (https://www.beiresources.org) unless noted otherwise. comycin was added at 5μg ml−1 immediately before infection. Peritoneal cells
The generation of protein-A-deficient strain Δmcr USA300 NRS384, as well as were added to the osteoblasts at 1×106 peritoneal cells per ml. A sample of the
protein-A-deficient USA300 lacking tarM or tarS has been described pre- cells was lysed with 0.1% Triton-X to determine the actual concentration of live
viously38,39. The protein-A-deficient strains were used only in some in vitro experi- intracellular bacteria at the time of infection. The actual titre for all infections was
ments to determine antibody specificity.
determined by plating serial dilutions of the bacteria on tryptic soy agar with 5%
MIC determinations for extracellular bacteria. The MIC for extracellular defibrinated sheep blood.
bacteria was determined by preparing serial twofold dilutions of the antibiotic Generation of the anti-S. aureus antibodies. The human IgG antibodies against
in tryptic soy broth. Dilutions of the antibiotic were made in quadruplicate in anti-β-GlcNAc WTA monoclonal antibody (mAb) and anti-α-GlcNAc WTA mAb
96-well culture dishes. MRSA (NRS384 strain of USA300) was taken from an were cloned from peripheral B cells from patients after S. aureus infection using
exponentially growing culture and diluted to 1 ×104 c.f.u. ml−1. The bacteria a monoclonal antibody discovery technology that conserves the cognate pairing
were cultured in the presence of antibiotic for 18–24h with shaking at 37°C and of antibody heavy and light chains40. Antibodies were expressed by transfection
bacterial growth was determined by reading the optical density (OD) at 630 nm. of mammalian cells41. Supernatants containing full-length IgG1 antibodies were
The MIC was determined to be the dose of antibiotic that inhibited bacterial harvested after 7 days and used to screen for antigen binding by enzyme-linked
growth by >90%.
immunosorbent assay (ELISA). These antibodies were positive for binding to
MIC determinations for intracellular bacteria. Intracellular MIC was determined cell wall preparations from USA300. Antibodies were subsequently produced in
on bacteria that were sequestered inside mouse peritoneal macrophages (see later 200-ml transient transfections and purified with protein A chromatography
for generation of murine peritoneal macrophages). Macrophages were plated at a (MabSelect SuRe, GE Life Sciences) for further testing.
density of 4×105 cells ml−1 and infected with MRSA at a ratio of 10–20 bacteria Synthesis of the linker drug. Synthesis of the rifalogue linker drug was per-
per macrophage. Macrophage cultures were maintained in growth media sup- formed as follows. Protease cleavable linker MC-VC-PAB-OH23 (1.009 g,
plemented with 50 μg ml−1 of gentamycin to inhibit the growth of extracellular 1.762 mmol, 1.000, 1,009 mg) was taken up in N,N-dimethylformamide
bacteria and test antibiotics were added to the growth media 1 day after infection. (6 ml, 77 mmol, 44, 5,700 mg). To this was added a solution of thionyl chlo-
The survival of intracellular bacteria was assessed 24h after addition of the antibi- ride (1.1 equiv., 1.938mmol, 1.100, 231mg) in dichloromethane (DCM) (1ml,
otics. Macrophages were lysed with Hanks buffered saline solution supplemented 15.44 mmol, 8.765, 1,325 mg) in portions dropwise (half was added over 1 h,
with 0.1% bovine serum albumin (BSA) and 0.1% Triton-X, and serial dilutions stirred for 1h at room temperature, then the other half was added over another
of the lysate were made in PBS solution containing 0.05% Tween-20. The number hour). The solution remained a yellow colour. Another 0.6 equiv. of thionyl
of surviving intracellular bacteria was determined by plating on tryptic soy agar chloride was added as a solution in 0.5ml DCM dropwise, carefully. The reaction
plates with 5% defibrinated sheep blood.
remained yellow and was stirred sealed overnight at room temperature. The
In vivo transfer of infection model. USA300 stocks were prepared for infection reaction was monitored by liquid chromatography mass spectrometry (LC/MS),
from actively growing cultures in tryptic soy broth. Bacteria were washed three indicating 88% conversion to benzyl chloride. Another 0.22 equiv. of thionyl
times in PBS and aliquots were frozen at −80 °C in PBS 25% glycerol.
chloride was added dropwise as a solution in 0.3 ml DCM. When the reaction
Intracellular bacteria infections. Seven-week-old female A/J mice (Stock 000646) approached 92% benzyl chloride, the reaction was bubbled with N2. The con-
were obtained from Jackson Labs and infected by peritoneal injection with centration was increased from 0.3 M to 0.6M.
5×107 c.f.u. of USA300. Mice were killed 1 day after infection and the perito-
MC-VC-PAB-Cl (0.9mmol) was cooled to 0°C and rifalogue (dimethyl piper-
neum was flushed with 5ml of cold PBS. Peritoneal washes were centrifuged for azinebenzoxazinorifamycin42 (0.75g, 0.81mmol, 0.46, 750mg)) was added. The
5 min at 1,000 r.p.m. at 4°C in a table-top centrifuge. The cell pellet containing mixture was diluted with another 1.5ml of DMF to reach 0.3M. Stirred open to
peritoneal cells was collected and cells were treated with 50μg ml−1 of lysostaphin air for 30min. N,N-diisopropylethylamine (3.5mmol, 3.5mmol, 2.0, 460mg) was
(Cell Sciences, CRL 309C) for 20min at 37°C to kill contaminating extracellular added and the reaction stirred overnight open to air. Over the course of 4 days,
bacteria. Peritoneal cells were washed three times in ice-cold PBS to remove the four additions of 0.2equiv. N,N-diisopropylethylamine base were added while the
lysostaphin. Peritoneal cells from donor mice were pooled, and recipient mice reaction stirred open to air, until the reaction appeared to stop progressing. The
were injected with cells derived from five donors per each recipient by intrave- reaction was diluted with DMF and purified on high-performance liquid chroma-
nous injection into the tail vein. To determine the number of live intracellular tography (HPLC; 20–60% ACN/FA·H2O) in several batches to give MC-VC-PAB-
colony-forming units, a sample of the peritoneal cells were lysed in HB (Hanks rifalogue (0.38g, 32% yield) m/z=1,482.8.
balanced salt solution supplemented with 10mM HEPES and 0.1% BSA) with 0.1%
The non-cleavable rifalogue linker drug was synthesized using the exact same
Triton-X, and serial dilutions of the lysate were made in PBS with 0.05% Tween-20. method, but replacing MC-VC-PAB-OH with MC-V-D-Cit-PAB-OH.
Free bacteria infections. A/J mice were infected with various doses of free bacteria Conjugation of the linker drug to antibody. Construction and production of
using a fresh aliquot of the glycerol stocks used for the peritoneal injections. Actual the THIOMAB variant of anti-WTA antibody was done as reported previously43
.
infection doses were confirmed by c.f.u. plating. For the data shown in Fig. 1a the Briefly, a cysteine residue was engineered at the Val 205 position of the anti-WTA
actual infection dose for intracellular bacteria was 1.8×106 c.f.u. per mouse, and light chain to produce its THIOMAB variant. The thio anti-WTA was conju-
the actual infection dose for free bacteria was 2.9×106 c.f.u. per mouse. Selected gated to MC-vc-PAB-rifalogue. The antibody was reduced in the presence of
mice were treated with a single dose of 110mg kg−1 of vancomycin by intravenous 50-fold molar excess dithiothreitol (DTT) overnight. The reducing agent and
injection immediately after infection.
the cysteine and glutathione blocks were purified away using HiTrap SP-HP col-
In vitro transfer of infection to non-phagocytic cells. Generation of MRSA- umn (GE Healthcare). The antibody was re-oxidized in the presence of 15-fold
infected peritoneal cells. Six-to-eight-week-old female A/J mice (see earlier) were molar excess dehydroascorbic acid (MP Biomedical) for 2.5h. The formation of
infected with 1×108 c.f.u. of the NRS384 strain of USA300 by peritoneal injec- interchain disulfide bonds was monitored by LC/MS. A threefold molar excess
tion. The peritoneal wash was harvested 1 day after infection, and the infected of the linker drug (MC-VC-PAB-rifalogue) over protein was incubated with the
peritoneal cells were treated with 50 μg ml−1 of lysostaphin diluted in HEPES THIOMAB for 1h. The AAC was purified by filtration through a 0.2μm SFCA
buffer supplemented with 0.1% BSA (HB buffer) for 20min at 37°C. Peritoneal filter (Millipore). Excess-free linker drug was removed by filtration. The conjugate
cells were then washed twice in ice-cold HB buffer. The peritoneal cells were was buffer exchanged into 20mM histidine acetate pH 5.5/240mM sucrose by
diluted to 1 ×106 cells ml−1 in RPMI 1640 tissue culture media supplemented dialysis. The number of conjugated MC-VC-PAB-rifalogue molecules per mAb
with 10 mM HEPES and 10% fetal calf serum, and 5 μg ml−1 vancomycin. was quantified by LC/MS analysis. Purity was also assessed by size-exclusion
Free MRSA from the primary infection was stored overnight at 4 °C in PBS chromatography.
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
Mass spectrometric analysis. LC/MS analysis was performed on a 6530 Generation of a synthetic WTA fragment. Synthesis of dibenzyl phosphorochlori-
Accurate-Mass Quadrupole Time-of-Flight (Q-TOF) LC/MS (Agilent date. A mixture of NCS (3.5g, 26.6mmol) was suspended in toluene (80ml). Then
Technologies). Samples were chromatographed on a PRLP-S column, 1,000 Å, dibenzyl phosphonate (2.0g, 7.6mmol) was added. The mixture was stirred at
8μm (50mm×2.1mm, Agilent Technologies) heated to 80°C. A linear gradi- room temperature overnight. The white solid was filtered off and the organic phase
ent from 30–60% B in 4.3min (solvent A, 0.05% TFA in water; solvent B, 0.04% was evaporated to give dibenzyl phosphorochloridate (1; 2.1g, 96%) as light yellow
TFA in acetonitrile) was used and the eluent was directly ionized using the elec- oil. 1H NMR (300 MHz, CDCl3, 25°C) δ 7.36 (s, 10H), 5.20 (m, 4H).
trospray source. Data were collected and deconvoluted using the Agilent Mass Synthesis of 4-O-(2-acetamido-3,4,6-tri-O-acetyl-2-deoxy-β-d-glucopyrano-
Hunter qualitative analysis software. Before LC/MS analysis, AAC was treated syl)-1-O-acetyl-d-ribitol-5-dibenzylphosphate. A mixture of 2 (described in ref. 44)
with lysyl endopeptidase (Wako) for 30min at 1:100 w/w enzyme to antibody (500mg, 0.95mmol) dissolved in pyridine (12ml) was cooled to −30 °C and 1
ratio, pH 8.0, and 37°C to produce the Fab and the Fc portion for ease of analysis. (described ref. 44) (595 mg, 2.0 mmol) was added, stirring for 2h at −30 °C and
The drug-to-antibody ratio (DAR) was calculated using the abundance of Fab and warmed to room temperature for 4 h. The mixture was added to H2O, and concen-
Fab+1 calculated by the MassHunter software.
trated in vacuo. The residue was purified by column chromatography (silica gel:
Flow cytometry to compare expression of anti-MRSA antibodies. Analysis of 200 to ~300 mesh; dichloromethane: methanol in a 30:1 as eluent) to give 4-O-(2-
bacteria isolated from infected mice. Balb/c mice were infected with 1×107 c.f.u. acetamido-3,4,6-tri-O-acetyl-2-deoxy-β-d-glucopyranosyl)-1-O-acetyl-d-ribitol-
of MRSA (USA300) by intravenous injection and kidneys were harvested on day 5-dibenzylphosphate (3; 190 mg, 24%) as light yellow solid. 1H NMR (300 MHz,
3 after infection. Kidneys were homogenized using a GentleMACS dissociator Acetone-d6, 25°C) δ 7.29–7.23 (m, 10H), 7.08 (d, 1H), 5.08 (t, 1H), 4.99–4.78
in 5 ml volume per two kidneys using M-Tubes and the program RNA01.01 (m, 6H), 4.31–3.97 (m, 8H), 3.82–3.63 (m, 3H), 1.88 (s, 3H), 1.86 (s, 6H), 1.79 (s, 3H),
(Miltenyi Biotec). Homogenization buffer was: PBS plus 0.1% Triton-X-100, 10μg 1.69 (s, 3H). LC/MS (m/z) ES+ 784 [M+H]+.
ml−1 DNAase (bovine pancreas grade II, Roche) and protease inhibitors (com- Synthesis of 4-O-(2-acetamido-3,4,6-tri-O-acetyl-2-deoxy-β-d-glucopyranosyl)-
plete protease inhibitor cocktail, Roche 11-836-153001). After homogenization, 1-O-acetyl-d-ribitol-5-phosphate. A mixture of 3 (150 mg, 0.19 mmol)
the samples were incubated at room temperature for 10min and then diluted with dissolved in MeOH (6 ml) was hydrogenated over 10% Pd/C (20 mg) for 2h at
ice-cold PBS and filtered through a 40 μm cell strainer. Tissue homogenates were room temperature. Then the mixture was filtered, and the filtrate was evaporated
washed twice in ice-cold PBS and then suspended in a volume of 0.5 ml per two to give 4-O-(2-acetamido-3,4,6-tri-O-acetyl-2-deoxy-β-d-glucopyranosyl)-1-O-
kidneys in HB buffer (Hanks balanced salt solution supplemented with 10mM acetyl-d-ribitol-5-phosphate (4; 100 mg) as light yellow oil. LC/MS (m/z) ES+
HEPES and 0.1% BSA). The cell suspension was filtered again and 25μl of the 604 [M+H]+.
bacterial suspension was taken for each staining reaction (Fig. 2c).
Synthesis of 4-O-(2-acetamido-2-deoxy-β-d-glucopyranosyl)-d-ribitol-5-phosphate
Antibody staining for flow cytometry. Bacteria (1×107 of in vitro grown bacteria (5). A mixture of 4 (80mg, 0.16mmol) dissolved in MeOH (10ml) was cooled
(Fig. 2d), or 25 μl of tissue homogenate described earlier (Fig. 2c) were sus- to 5 °C and K2CO3 (30 mg, 0.21 mmol) was added and stirred at 5°C for 3 h.
pended in HB buffer and blocked by incubation with 400μg ml−1 of mouse IgG The reaction was then quenched with 1 N HCl, and concentrated in vacuo. The
(Sigma, I5381) for 1 h. Fluorescently labelled antibodies were added directly crude product was purified by gel filtration (LH-20, MeOH) to give 4-O-(2-
to the blocking reaction and incubated at room temperature for an additional acetamido-2-deoxy-β-d-glucopyranosyl)-d-ribitol-5-phosphate (5; 13.3mg, 23%)
10–20 min. Bacteria were washed three times in HB buffer and then fixed in as a white solid 1H NMR (300MHz, MeOH-d4, 25°C) δ 4.62 (d, 1H), 4.30–4.02
PBS 2% paraformaldehyde before FACS analysis. Test antibodies (anti-β-WTA, (m, 3H), 3.92–3.32 (m, 9H), 2.03 (s, 3H). LC/MS (m/z) ES+ 436 [M+H]+.
anti-α-WTA or isotype control-anti CMV-gD) were conjugated with Alexa-488 Time of kill for free antibiotics on non-replicating bacteria. S. aureus (USA300)
using amine reactive reagents (Invitrogen, succinimidyl-ester of Alexa Fluor was taken from an overnight stationary phase culture, washed once in PBS and
488, NHS-A488). Antibodies in 50mM sodium phosphate were reacted with a suspended at 1 ×107 c.f.u. ml−1 in PBS with no antibiotic or with 1 ×10−6 M
5–10-fold molar excess of NHS-A488 in the dark for 2–3h at room temperature. antibiotic in a 10ml volume in 50ml polypropylene centrifuge tubes. The bacteria
The labelling mixture was applied to a GE Sepharose S200 column equilibrated were incubated at 37°C overnight with shaking. At each time point, three 1 ml
in PBS to remove excess reactants from the conjugated antibody. The number of samples were removed from each culture and centrifuged to collect the bacteria.
A488 molecules per antibody was determined using the ultraviolet method as Bacteria were washed once with PBS to remove the antibiotic and the total number
described by the manufacturer.
of surviving bacteria was determined by plating serial dilutions of the bacteria
For analysis of bacteria in tissue homogenates a non-competing anti-S. aureus on agar plates.
antibody (rF1 (ref. 38)) was conjugated to Alexa-647 to distinguish S. aureus from Killing of persister cells by free antibiotics. S. aureus (USA300) was taken from
similar sized particles. Test antibodies were examined at a range of doses from an overnight stationary phase culture, washed once in tryptic soy broth (TSB) and
80ng ml−1 to 50 μg ml−1. Flow cytometry was performed using a Beckton Dickson then adjusted to a final concentration of 1×107 c.f.u. ml−1 in a total volume of
FACS ARIA (BD Biosciences) and analysis was performed using FlowJo analysis 10ml of either TSB or TSB with ciprofloxicin (0.05mM). Cultures were incubated
software (Flow Jo LLC).
with shaking at 37°C for 6h and then the second antibiotic, either rifampicin
WTA–antibody complex purification and crystallization. The anti-β-WTA (1μg ml−1) or the rifalogue (1 μg ml−1) was added. At the indicated times, samples
antibody Fab fragment was expressed in Escherichia coli and purified on Protein were removed from each culture, washed once with PBS to remove the antibiotic
G Sepharose followed by SP sepharose cation exchange and size-exclusion chro- and re-suspended in PBS. The total number of surviving bacteria was determined
matography. Antibody was concentrated to 30mg ml−1 in MES buffer (20mM by plating serial dilutions of the bacteria on agar plates. At the final time point the
MES pH 5.5, 150mM NaCl) and mixed with a 2:1mol/mol ratio of the WTA remainder of each culture was collected and plated.
analogue (diluted in water) for crystallization trials. Sparse matrix crystallization Cathepsin release assay for AAC. To quantify the amount of active antibiotic
screening provided initial hits in PEG-8000 based conditions, which were further released from AACs after treatment with cathepsin B, AACs were diluted to
optimized to provide diffraction quality crystals. Ultimately, data were collected on 200 μg ml−1 in cathepsin buffer (20 mM sodium acetate, 1 mM EDTA, 5 mM
a crystal grown by the vapour diffusion method in a sitting drop containing 0.5μl l-cysteine, pH 5). Cathepsin-B (from bovine spleen, Sigma C7800) was added at
protein and 0.5μl 0.08M sodium cacodylate pH 6.5, 0.16M calcium acetate, 14.4% 10μg ml−1 and the samples were incubated for 1h at 37 °C. As a control, AACs
PEG-8000, and 20% glycerol. Crystals were cryo-protected in mother liquor, flash were incubated in buffer alone. The reaction was stopped by addition of 9 volumes
frozen in liquid nitrogen, and stored for data collection at 100K.
of bacterial growth media, TSB pH 7.4. To estimate the total release of active
Data collection and structure determination. Data were collected to 1.7 Å at antibiotic, serial dilutions of the reaction mixture were made in quadruplicate
beamline 22ID at the Advanced Photon Source (APS) under cryo-cooled con- in TSB in 96-well plates and MRSA (USA300) was added to each well at a final
ditions (100 K) at a wavelength of 1.0 Å. Data were reduced using HKL2000 density of 2×103 c.f.u. ml−1. The cultures were incubated overnight at 37°C with
and SCALEPACK in the space group P212121, with unit cell parameters of shaking and bacterial growth was measured by reading absorbance at 630 nM
a = 63.7 Å, b = 111.4Å, c = 158.4 Å (see Extended Data Table 1 for process- using a plate reader.
ing statistics). The structure was solved by sequential molecular replacement Synthesis of anti-β-WTA antibody FRET conjugate. We synthesized and conju-
searches using Fab constant and variable regions (Protein Data Bank accession gated a maleimide FRET peptide to the anti-β-WTA THIOMAB antibody. We used
4I77) as individual search models. Iterative rounds of manual model adjustment a FRET pair of tetramethylrhodamine (TAMRA) and fluorescein. The maleimide
with COOT followed by simulated annealing, coordinate, and b-factor refine- FRET peptide was synthesized by standard Fmoc solid-phase chemistry using a
ment with Phenix and BUSTER (Global Phasing) gave a final model with R/Rfree PS3 peptide synthesizer (Protein Technologies; B.-C.L., M.D. and R.V., manu-
values of 20.6% and 23.7% respectively. Ramachandran statistics calculated by script in preparation)27. Briefly, 0.1mmol of Rink amide resin was used to generate
MolProbity indicate that 97.2% of the model residues lie in favoured regions, with C-terminal carboxamide. We used a Fmoc-Lys(Mtt)-OH at the N- and C-terminal
0.5% outliers.
residues in order to remove the Mtt group on the resin and carry out additional
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
side-chain chemistry to attach TAMRA and fluorescein. The sequence of Val- a concentration of 4% in water, sterilized by autoclaving, and aged for 20 days to
Cit-Leu was added between the FRET pair as a cathepsin-cleavable spacer. The 6 months before use. Peritoneal macrophages were harvested 4 days after treatment
crude maleimide FRET peptide or maleimidocarproyl-K(TAMRA)-G-V-Cit-L-K with thioglycolate by washing the peritoneal cavity with cold PBS. Macrophages
(fluorescein) cleaved off from the resin was subjected to further purification were plated in DMEM supplemented with 10% fetal calf serum, and 10mM
by reverse-phase HPLC with a Jupiter 5u C4 column (5μm, 10mm×250mm; HEPES, without antibiotics, at a density of 4×105 cells well−1 in 24-well culture
Phenomenex). Our FRET probe allows monitoring not only of the intracelluar dishes. Macrophages were cultured overnight to permit adherence to the plate.
trafficking of the antibody conjugate, but also the processing of the linker in the Human M2 macrophages. CD14+ monocytes were purified from normal human
phagolysosome. The intact antibody conjugate fluoresces only in red due to the blood using a Monocyte Isolation Kit II (Miltenyi, catalogue 130-091-153) and
fluorescence resonance energy transfer from the donor. However, upon the sub- plated at 1.5×105 cells cm−2 on tissue culture dishes pre-coated with fetal calf
strate cleavage of the FRET peptide in the phagolysosome, the green fluorescence serum (FCS) and cultured in RPMI 1640 media with 20% FCS plus 100ng ml−1
from the donor is expected to appear.
rhM-CSF. Media was refreshed on day 1 and on day 7, the media was changed to
Video microscopy to detect cleavage of the linker inside macrophages. Murine 5% serum plus 20ng ml−1 IL-4. Macrophages were used 18h later.
peritoneal macrophages were plated on chamber slides (Ibidi, catalogue 80826) Assay protocol. In all experiments bacteria were cultured in TSB. To assess intracel-
in complete media as described for the macrophage intracellular killing assay. lular killing with AACs, USA300 was taken from an exponentially growing culture
USA300 was labelled with Cell Tracker Violet (Invitrogen C10094) at 100μg ml−1 and washed in HB. AACs or antibodies were diluted in HB (Hanks balanced salt
in PBS 0.1% BSA by incubation for 30min at 37°C. The labelled bacteria were solution supplemented with 10 mM HEPES and 0.1% BSA) and incubated with
opsonized with the anti-β-WTA-FRET probe by incubation for 1h in HB buffer. the bacteria for 1h to permit antibody binding to the bacteria (opsonization),
Macrophages were washed once immediately before addition of the opsonized and the opsonized bacteria were used to infect macrophages at a ratio of 10–20
bacteria, and bacteria were added to cells at 1 ×107 bacteria per ml. For no- bacteria per macrophage (4×106 bacteria in 250 μl of HB per well). Macrophages
phagocytosis controls, the macrophages were pre-treated with 60nM Latrunculin were pre-washed with serum-free DMEM media immediately before infection,
A (Calbiochem) for 30min before and during phagocytosis. The slides were placed and infected by incubation at 37 °C in a humidified tissue culture incubator with
on the microscope immediately after addition of bacteria to the cells and movies 5% CO2 to permit phagocytosis of the bacteria. After 2 h, the infection mix was
were acquired with a Leica SP5 confocal microscope equipped with an environ- removed and replaced with normal growth media (DMEM supplemented with 10%
mental chamber with CO2 and temperature controllers from Ludin. The images FCS, 10 mM HEPES) and gentamycin was added at 50 μg ml−1 to prevent growth
were captured every minute for a total time of 30min using a Plan APO CS ×40, of extracellular bacteria45. At the end of the incubation period, the macrophages
N.A: 1.25, oil immersion lens, and the 488nm and 543nm laser lines to excite were washed with serum-free media, and the cells were lysed in HB supplemented
Alexa-488 and TAMRA, respectively. Phase images were also recorded using the with 0.1% Triton-X (lyses the macrophages without damaging the intracellular
543nm laser line.
bacteria). Serial dilutions of the lysate were made in PBS solution supplemented
Quantification of released antibiotic inside macrophages. Primary murine peri- with 0.05% Tween-20 (to disrupt aggregates of bacteria) and the total number of
toneal macrophages or RAW 264.7 cells (purchased from ATCC) were infected surviving intracellular bacteria was determined by plating on tryptic soy agar with
in 24-well tissue culture dishes as described later for the intracellular killing assay 5% defibrinated sheep blood.
with MRSA opsonized with AAC at 100μg ml−1 in HB. The RAW 264.7 cells were ELISA for quantification of anti-MRSA antibodies in human serum. Cell wall
used without further authentication or testing for mycoplasma contamination. preparations (CWPs) were generated from protein-A-deficient S. aureus by incu-
After phagocytosis was complete, the cells were washed and 250μl of complete bating 40mg of pelleted bacteria per ml of 10mM Tris-HCl (pH 7.4) supplemented
media plus gentamycin was added to wells and the cells were incubated for the with 30% raffinose, 100μg ml−1 of lysostaphin (Cell Sciences), and EDTA-free
indicated time points. At each time point, the supernatant and cellular fractions protease inhibitor cocktail (Roche), for 30 min at 37°C. The lysates were centri-
were collected followed by acetonitrile (ACN) addition to 75% final concentra- fuged at 11,600g for 5min, and the supernatants containing cell wall components
tion and incubated for 30min. Cell and supernatant extracts were lyophilized by were collected.
evaporation under N2 (TurboVap; Biotage) and reconstituted in 100 μl of 50%
ACN, filtered using a 0.45 glass fibre filter plate (Phenomenex) and analysed by were pre-coated with CWP and then incubated with human IgG preparations:
LC/MS/MS as follows. purified human IGIV Immune Globulin (ASD Healthcare), pooled serum from
ELISA experiments were performed using standard protocols. Briefly, plates
The rifalogue was separated on an Acquity UPLC (Waters Corporation) healthy donors or from MRSA patients. The concentrations of anti-staphylococcal
under gradient elution using a Phenomenex Kinetex XB-C18 column (100Å, IgG present in the serum or purified IgG were calculated by using a calibration
50×2.1 mm internal diameter, 2.6μm particle size). The column was maintained curve that was generated with known concentrations of anti-peptidoglycan mAb
at room temperature. The mobile phase was a mixture of 10mM ammonium ace- (4479) against peptidoglycan.
tate in water containing 0.1% formic acid (A) and 90% acetonitrile (B) at a flow Intravenous infection model for testing efficacy of the AAC. Seven-week-
rate of 1 ml min−1. The rifalogue was eluted with a gradient of 3–98% B over 1 min, old female mice, Balb/c, were obtained from Jackson West, or SCID mice were
followed by 0.8 min at 98% B, then 0.7 min of 3% B to re-equilibrate the column. obtained from Charles River Laboratories. Infections were carried out by intra-
The injection volume was 10 μl.
venous injection into the tail vein. SCID-huIgG model: CB17.SCID mice were
The Triple Quad 6500 mass spectrometer (Ab Sciex) was operated in a pos- reconstituted with IGIV Immune Globulin (ASD Healthcare) using a dosing
itive ion multiple reaction-monitoring (MRM) mode. The rifalogue precursor regimen optimized to achieve constant serum levels of >10mg ml−1 of human
(Q1) ion monitored was 927.6 m/z and the product (Q3) ion monitored was IgG. IGIV was administered with an initial intravenous dose of 30mg per mouse
895.2 m/z with collision energy at 27 eV and declustering potential at 191 V. The followed by a second dose of 15mg per mouse by intraperitoneal injection after
MS/MS setting parameters were as follows: ion spray voltage, 5,500 V; curtain 6 h, and subsequent daily dosings of 15mg per mouse by intraperitoneal injection
gas, 40 psi; nebulizer gas (GS1), 35psi, (GS2), 50 psi; temperature, 600°C; and for 3 consecutive days. Mice were infected 4h after the first dose of IGIV with
dwell time, 150 ms.
2×107 c.f.u. of MRSA diluted in PBS by intravenous injection. The wild-type
Linear calibration curves were obtained for 0.41–100 nM concentration range USA300, protein-A-sufficient strain was used for all in vivo experiments. Mice
by spiking rifalogue into cell or supernatant fractions (lacking MRSA or AAC) that that received vancomycin were treated with twice daily intraperitoneal injections
were treated similarly to samples. Concentrations of rifalogue were calculated with of 110 mg kg−1 of vancomycin starting between 6 and 24h after infection for the
MultiQuant software (Ab Sciex).
duration of the study. Experimental therapeutics (AAC, anti-MRSA antibodies
In vitro intracellular killing assay. Non-phagocytic cell types. MG63 (CRL-1427) or free rifalogue antibiotic) were diluted in PBS and administered with a single
and A549 (CCL185) cell lines were obtained from ATCC and maintained in RPMI intravenous injection 30 min to 24 h after infection. All mice were killed on
1640 tissue culture media supplemented with 10mM HEPES and 10% fetal calf day 4 after infection, and kidneys were harvested in 5 ml of PBS. The tissue
serum (RPMI-10). HUVEC cells were obtained from Lonza and maintained in samples were homogenized using a GentleMACS dissociator (Miltenyi Biotec).
EGM endothelial cell complete media (Lonza). HBMEC cells (catalogue #1000) The total number of bacteria recovered per mouse (two kidneys) was determined
and ECM media (catalogue #1001) were obtained from ScienceCell Research Labs. by plating serial dilutions of the tissue homogenate in PBS 0.05% Tween on tryptic
The cells were used without further authentication or testing for mycoplasma soy agar with 5% defibrinated sheep blood.
contamination.
Statistical analysis. All experiments were performed on biological replicates.
Murine macrophages. Peritoneal macrophages were isolated from the peritoneum Sample size for each experimental group per condition is reported in appropriate
of 6–8-week-old Balb/c mice (Charles River Laboratories). To increase the yield figure legends and Methods. For cell culture experiments, sample size was not
of macrophages, mice were pre-treated by intraperitoneal injection with 1 ml of predetermined, and all samples were included in the analysis. In animal experi-
thioglycolate media (Becton Dickinson). The thioglycolate media was prepared at ments no statistical methods were used to predetermine sample size (n=number
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
of mice per group), and all animals were used for analysis unless the mice died
or had to be euthanized when found moribund. These cases are annotated in the
figures. The mice were not randomized after infection, and the investigators were
not blinded to outcome assessment. When appropriate, statistically significant
differences between control and experimental groups were determined using
Mann–Whitney tests.
40. Meijer, P. J. et al. Isolation of human antibody repertoires with preservation of
the natural heavy and light chain pairing. J. Mol. Biol. 358, 764–772
(2006).
41. Meijer, P. J., Nielsen, L. S., Lantto, J. & Jensen, A. Human antibody repertoires.
Methods Mol. Biol. 525, 261–277 (2009).
42. Van Duzer, J. et al. Rifamycin Analogs and Uses Thereof (Activbiotics, 2005).
43. Junutula, J. R. et al. Site-specific conjugation of a cytotoxic drug to an
antibody improves the therapeutic index. Nature Biotechnol. 26, 925–932
(2008).
44. Boullanger, P., Descotes, G., Flandrois, J. P. & Marmet, D. Synthesis of
4-O-(2-acetamido-2-deoxy-β-d-glucopyranosyl)-d-ribitol, antigenic determinant
of Staphylococcus aureus. Carbohydr. Res. 110, 153–158 (1982).
45. Vaudaux, P. & Waldvogel, F. A. Gentamicin antibacterial activity in the presence
of human polymorphonuclear leukocytes. Antimicrob. Agents Chemother. 16,
743–749 (1979).
38. Hazenbos, W. L. et al. Novel staphylococcal glycosyltransferases SdgA and
SdgB mediate immunogenicity and protection of virulence-associated cell wall
proteins. PLoS Pathog. 9, e1003653 (2013).
39. Monk, I. R., Shah, I. M., Xu, M., Tan, M. W. & Foster, T. J. Transforming the
untransformable: application of direct transformation to manipulate
genetically Staphylococcus aureus and Staphylococcus epidermidis. MBio 3,
e00277–11 (2012).
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
Infection with:
Planktonic
bacteria
Intracellular
baceria
107
106
105
104
103
102
101
100
PBS
vanco
PBS
vanco
Extended Data Figure 1 | Intracellular MRSA can infect the brain even
in the presence of vancomycin. In vivo infection of mice shown in
Fig. 1a. Mice (n =5) were infected with equivalent doses of free bacteria or
intracellular bacteria and treated with vancomycin at 110 mg kg−1 10 min
after infection and then once per day. Bacterial burden was monitored in
the brain 4 days after infection. Bars show geometric mean.
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
a
b
Planktonic MRSA alone
1010
1010
109
108
107
106
105
104
103
102
101
100
Planktonic MRSA alone
109
108
107
106
105
104
103
102
101
Intracellular MRSA ( in A549 cells)
+ vanco
Intracellular MRSA ( in HBMEC cells)
+ vanco
Planktonic MRSA + vanco
Planktonic MRSA + vanco
100
0
5
10
15
20
25
0
5
10
15
20
25
Time (hrs)
Time (hrs)
Extended Data Figure 2 | MRSA is able to grow in the presence of
vancomycin when cultured on a monolayer of infectable cells.
a, b, Similar to the set up in Fig. 1d, planktonic MRSA were either seeded
in media alone, or in the presence of vancomycin. Intracellular bacteria
were generated by infecting a monolayer of either A549 bronchial
epithelial cells (a) or HBMECs (b) in the presence of vancomycin (vanco).
In these experiments plates were centrifuged to promote contact of the
bacteria with the monolayer to enhance intracellular infection. At each
time point, the culture supernatant was collected to recover extracellular
bacteria and adherent cells were lysed to release intracellular bacteria.
Extracellular bacteria (planktonic bacteria) grew well in media alone,
but were killed by vancomycin. In wells containing a monolayer of
mammalian cells (intracellular MRSA + vanco) a fraction of the bacteria
were protected from vancomycin during the first 5 h after infection and
were able to expand within the intracellular compartment over 24 h. Error
bars show s.d. from triplicate wells. Representative of three independent
experiments.
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
a
b
100
80
60
40
20
0
100
80
60
40
20
0
Rifampicin
Rifalogue
Rifampicin
Rifalogue
0
10
60
10-10
10-9
10-8
10-7
10-6
Minutes in culture
[Antibiotic] M
Extended Data Figure 3 | Rifalogue can also kill intracellular bacteria.
a, Determining the intracellular MIC for rifalogue and rifampicin. MRSA
was allowed to infect peritoneal macrophages and macrophages were
cultured overnight in gentamycin to kill extracellular bacteria. Various
doses of rifalogue (red) or rifampicin (grey) were added to the culture
medium 1 day after infection and the number of viable intracellular
bacteria was determined 24 h later by spotting macrophage lysates
onto agar plates. Data shown are representative of more than three
independent experiments. b, Diffusion of rifalogue versus rifampicin into
murine macrophages. Murine peritoneal macrophages were incubated
with rifalogue or rifampicin in the culture media. Wells were harvested
at 10 and 60 min and the total amount of antibiotic associated with
the cells was determined by quantitative mass spectrometry. Results
are shown as percentage of input. Error bars show s.d. from triplicate
wells. Representative of two experiments. Rifalogue is more lipophilic
than rifampicin as a result of its two additional fused aromatic rings,
determined by measuring logDs at pH 7, with rifalogue at 3.4, being
100-fold higher than rifampicin (logD 1.3). Additionally, rifalogue has
a more basic amine (pKa 9.7) compared to that found in rifampicin
(pKa 8.2). This balance of lipophilicity and basicity in rifalogue allows
it to localize preferentially in lysosomes. It is challenging to develop
antibiotics with these properties for systemic administration due to
poor pharmacokinetic (PK) properties and toxicity profiles associated
with indiscriminate accumulation of these molecules in all host cells.
However, appending such antibiotics to an anti-MRSA specific antibody
both extends its half-life in the circulation and converts it into an inactive
pro-drug whose properties are manifest only after it has been released in
phagolysosomes of cells infected with MRSA.
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
Infection with:
Planktonic
bacteria
Intracellular
baceria
+ PBS
Planktonic
bacteria
107
106
105
104
103
102
*
+ vanco
+ PBS
Intracellular
bacteria in
host cells
*
Planktonic
bacteria
+ vanco
+ vanco
+AAC
PBS
vanco
PBS
vanco
vanco
+ AAC
Extended Data Figure 4 | AAC kills MRSA that survive treatment with
vancomycin. In vivo infection of mice as shown in Fig. 1a. Mice were
infected with equivalent doses of free bacteria or intracellular bacteria and
vancomycin as described earlier and also treated with a single dose of AAC
at 50 mg kg−1 10 min after infection. Four days after infection, bacterial
burden was monitored in the brain. Each point represents data
from a single mouse (n =5). Bars show geometric mean. *P <0.05,
Mann–Whitney U-test.
treated with either saline (PBS) or vancomycin (vanco) at 110 mg kg−1
10 min after infection and then once per day. Selected mice were given
,
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
a
b
600
500
400
300
IGIV
Incubate with polyclonal IgG;
purified or sera
WTA
PGN
200
100
WT WTA
CWP ELISA
tarM tarS
DKO CWP ELISA
Purified
IgG
IGIV
Healthy
donor
serum
Patient
donor
serum
Extended Data Figure 5 | Human serum contains high levels of anti-
S. aureus antibodies that can compete with the AAC for binding.
a, To estimate the concentration of antibodies that could potentially
compete for binding with the anti-β-WTA antibody used in the AAC,
human IgG from various sources (normal human serum, serum from
MRSA infected patients, purified human IgG (Sigma) or IGIV derived
from pooled normal donors) was tested for binding to various bacterial
cell wall preparations (CWPs) by enzyme-linked immunosorbent assay
(ELISA). CWPs were made from either USA300 (wild type (WT)) or
ΔtarMΔtarS (DKO) USA300; the latter strain is deficient in the WTA-
GlcNAc antigen recognized by the anti-WTA antibodies. For these
studies protein-A-deficient USA300 background strains were used to
minimize non-specific antibody binding. A standard curve was generated
by titrating known amounts of an anti-MRSA antibody directed against
peptidoglycan on both cell wall extracts. b, Estimated concentration of
anti-S. aureus antibodies in human serum. The amount of anti-MRSA
antibodies in each sample was estimated by comparing the signal obtained
for each sample with the standard curve. In the absence of WTA GlcNAc
antigens, ~60–70% less serum IgG binding was observed (DKO ELISA;
red bars). This indicates the high prevalence of natural antibodies against
WTA in adult human serum. Results are reported as μg ml−1 of antibody
per 10 mg ml−1 of total IgG. Error bars show mean s.d. from triplicate
wells. Data are representative of two independent experiments.
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
a
109
108
107
106
105
104
103
102
MRSA
(i.v)
Day 4
Kidney
c.f.u.
t =0
(t = 24hr)
AACs
mAb
only
AAC
***
109
IGIV
b
108
107
106
105
104
103
102
N.S.
MRSA
(i.v)
Day 4
Kidney
c.f.u.
t =0
t =24 hrs
PBS tx
AAC tx
anti-α-WTA
AAC
PBS
anti-β-WTA
AAC
IgG and infected with MRSA. Mice were treated with 60 mg kg−1 of the
indicated AACs in a single intravenous injection 1 day after infection and
bacterial burden was monitored in kidneys 4 days after infection. Each
point represents data from a single animal. Bars show geometric mean.
Mann–Whitney test: ***P <0.005, not significant (NS) P >0.05.
Extended Data Figure 6 | Optimization of the in vivo model of
bacteraemia. a, Titration of AAC in intravenous infection model shown
in Fig. 4c (n =8 mice per group). b, Efficacy of AACs specific for β-WTA
or α-WTA in SCID-IGIV model. SCID mice (n =8 mice per group)
were reconstituted with IGIV Immune Globulin using a dosing regimen
optimized to achieve constant serum levels of >10 mg ml−1 of human
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
a
b
d
MRSA
(i.v)
MRSA
(i.v)
Day 4
Kidney
c.f.u.
Day 4
Kidney
c.f.u.
t =0
t =0
1
7
24hr
(t = -1hr)
Abs
(t = -1hr)
Initiation of Vanco Tx - twice daily (BID)
post infection
or Vanco Tx (BID)
c
IGIV
MRSA
(i.v)
MRSA
(i.v)
Day 4
Kidney
c.f.u.
Day 4
Kidney
c.f.u.
t =0
t =0
(t = 24hr)
AACs
(t =24hr)
AACs
or Vanco Tx (BID)
or Vanco Tx (BID)
e
IGIV
MRSA
(i.v)
Day 4
Kidney
c.f.u.
t =0
AACs/ Rifalogue (t =2
4hr)
or Rifalogue (daily)
Extended Data Figure 7 | Schematic of in vivo experiments presented in Fig. 4.
© 2015 Macmillan Publishers Limited. All rights reserved

reSeArcH Article
a
b
S. aureus
opsonization
+/- exogenous cathepsin-B
0.5
0.4
0.3
0.2
0.1
Rifalogue-AAC
Rifampicin-AAC
6
10
5
10
Rifalogue-AAC + cathepsin
Rifampicin-AAC + cathepsin
4
10
3
10
2
10
0.0
0.0
0.2
0.4
0.6
0.8
1
10
AAC ( g/mL)
0
10
g/mL
mAb
only
Rifalogue
AAC
Rifampicin
AAC
S. aureus
c
opsonization with
Rifalogue-AAC or
Rifampicin-AAC
Intracellular antibiotics (% retained)
d
Extracellular antibiotics
(% released into supernatant)
IGIV
MRSA
(i.v)
Day 4
Kidney
c.f.u.
t =0
t =24 hrs
mAb only
Rifampicin-AAC
Unconjugated Rifalogue inside cells,
Unconjugated Rifalogue in supernatant
Unconjugated Rifampicin inside cells
Unconjugated Rifampicin in supernatant
N.S.
100
50
0
100
50
0
109
108
107
106
105
104
103
102
101
10
20
60
120
180
10
20
60
120
180
mAb
only
Rifampicin
AAC
Minutes post AAC-coated S.aureus phagocytosis in macrophages
Extended Data Figure 8 | Comparison of anti-β-WTA AACs made
with rifalogue and rifampicin. a, Rifampicin–AAC and rifalogue–
AAC release equivalent amounts of free antibiotic after treatment with
exogenous cathepsin-B. Released antibiotics from AACs made with
rifalogue (red) and rifampicin (grey) are equally active as they can kill
USA300 grown in broth culture. Error bars show s.d. from triplicate wells.
b, Intracellular killing assay in primary mouse macrophages as shown in
Fig. 3c indicates that the rifalogue–AAC, but not rifampicin–AAC is able
to kill intracellular S. aureus. Error bars show s.d. from triplicate wells.
c, Greater intracellular retention of unconjugated rifalogue compared
with rifampicin after release from AAC inside macrophage cells. MRSA
was opsonized with AACs and incubated with macrophages (RAW 264.7
cells) to permit phagocytosis. The macrophages were washed to remove
extracellular bacteria and samples of cell lysates or supernatants were
collected in triplicate at indicated time points and the total amount of
released antibiotic was determined by quantitative mass spectrometry.
Error bars represent means s.d. from triplicate wells. a–c, Representative
of two or more independent experiments. d, Rifampicin–AAC is not
efficacious in the SCID-IGIV intravenous infection model as shown in
Fig. 4d. Each point represents data from a single mouse (n =8 mice per
group). Bars show geometric mean. Mann–Whitney U-test: not significant
(NS), P >0.05.
© 2015 Macmillan Publishers Limited. All rights reserved

Article reSeArcH
extended Data Table 1 | Data collection and refinement statistics for anti-β-WTA–WTA complex
*Values in parentheses are for highest-resolution shell.
© 2015 Macmillan Publishers Limited. All rights reserved