sources (fig. S1, D and E). Uptake assays for
(^14) C-glycerol and (^14) C-glucose showed that the
mutant was deficient because it could not
transport glycerol and glucose, whereas both
the parent and the complemented strain could
(fig. S1, F and G).
To confirm that loss of PPE51 function was
responsible for these phenotypes, we created a
site-specific deletion of theppe51gene (fig. S2A).
This knockout did not recapitulate the pheno-
type of the drug-selected mutant—it showed
less resistance to 3bMP1 in the medium with-
out Tween 80 and was able to use glycerol and
glucose as its sole carbon sources, unlike the
drug-selected mutant (fig. S2, B and C, and
Fig. 2, A and B). After further analysis, we
discovered that the mutant had lost the ability
to synthesize an abundant cell wall–associated
nonpolar wax phthiocerol dimycocerosate
(PDIM), containing a single nucleotide dele-
tion (618delA) in thefadD26gene, which en-
codes a fatty acid–coenzyme A ligase in the
PDIM biosynthesis pathway (Fig. 2C and fig.
S1H). PDIM loss is a common occurrence
during in vitro culture ofM. tuberculosis( 20 ).
PDIM makes up 46% of the total lipids of
M. tuberculosis, amounting to 11% of the total
weight of dried bacteria ( 21 ). When we reselected
a PDIM-positive deletion mutant ofppe51,the
deletion mutant displayed a level of resistance
to 3bMP1 similar to that of the drug-selected
mutant, as well as an inability to grow on
glycerol and glucose (fig. S2, B and C, and Fig.
2, A and B). No obvious growth difference
was observed in standard rich medium or in
media containing oleic acid or hexanamide
as sole carbon sources, indicating that the
PDIM-positiveppe51knockout mutant did
not have a general growth defect (fig. S2, D
to F). The PDIM-positiveppe51deletion
mutant also showed a defect in uptake of
(^14) C-labeled glycerol and glucose (Fig. 2, D
and E). Restoration of PDIM production in
the originalppe51deletion mutant was
achieved by providing a wild-type copy of
fadD26, which resulted in an inability to
grow on glycerol and glucose and showed
defective uptake of^14 C-labeled glycerol and
glucose (Fig. 2, A to E). Notably, both growth
and uptake defects in the PDIM-positiveppe51
knockout mutant were complemented by ex-
pression of wild-typeppe51or partially com-
plemented by expression of the porin gene
mspAofM. smegmatis(Fig.2,AtoE).Together
with the observation that loss of PDIM re-
scued growth and uptake defects, this result
indicates that PPE51 was localized to the outer
membrane ofM. tuberculosisand had a simi-
lar channel function to MspA.
Subcellular fractionation ofM. tuberculosis
cells expressing His-tagged PPE51 showed
that this protein was membrane-associated
(Fig. 2F). To distinguish between the inner and
outer membrane proteins, we performed a
cell-surface protein biotinylation assay—which is
based on surface detection of proteins in whole
cells using antibodies—using a membrane-
impermeable biotinylation reagent ( 22 ) and
flow cytometry analyses. As shown in Fig. 2,
G and H, His-tagged PPE51 was biotinylated
with amine-PEG11-biotin and was detected
with a monoclonal anti-His antibody in flow
cytometry experiments, whereas an inner mem-
brane protein (PrrB), an inner membrane–
associated protein (MbtG), and a cytosolic
protein (GroEL) were not detected. The func-
tionality of His-tagged PPE51 was confirmed
by showing that it restored sensitivity to 3bMP1
(fig. S2G). These experiments indicate that
PPE51 is localized to the outer membrane of
M. tuberculosis.
The PE/PPE proteins are thought to assem-
ble as heterodimers. The first reported crys-
tallographic structure of a PE and PPE protein
heterodimer (PE25 and PPE41) revealed that
the proteins are tightly associated via hydro-
phobic regions along the long axis of a four-helix
bundle ( 23 ). The similarity of this heterodimer
to the structure of an ESAT-6 secretion system 1
(ESX-1)–secreted protein inM. tuberculosis,
EspB, has been invoked to explain the substrate
specificity of the type VII secretion system,
which also transports PE/PPE proteins across
the cytoplasmic membrane ( 24 ). EspB oligo-
merizes into heptamers that form pores in the
phagosomal membrane of the host cell during
infection byM. tuberculosis. Our data were con-
sistent with a similar porin-like role for PPE51,
in combination with a PE protein, embedded
in the PDIM (fig. S4).
To identify the partner PE protein for
PPE51, we performed saturating transposon
mutagenesis on wild-typeM. tuberculosis,se-
lecting for resistance to 3bMP1. In addition
to hits inppe51, we identified transposon hits
inpe19that conferred resistance (table S2).
We subsequently constructed a series of mu-
tants lacking each of the individual genes in
Wanget al.,Science 367 , 1147–1151 (2020) 6 March 2020 3of5
Fig. 3. PPE51 interacts with PE19 inM. tuberculosis.
(A)Lossofpe19inM. tuberculosisleads to resistance
to 3bMP1. (B)Thepe19gene is negatively regulated by
the antisense expression ofpe18. Antisense expression of
pe18was induced using 100 ng/ml anhydrotetracycline
(aTC). In (A) and (B), experiments were performed in
7H9/OAD medium containing 0.02% tyloxapol. Data are
generated from at least two independent experiments,
both done as technical duplicates, and error bars
represent SD [in (A):P< 0.0001, wt versusDpe19;in(B):
P= 0.0008, wt versus wt+pe18anti (+aTC);P=0.0082,
wt+pe18anti (−aTC) versus wt+pe18anti (+aTC); two-
way ANOVA of matched values]. (C)Invivointeraction
between PPE51 and PE19.M. tuberculosiscoexpressing
His-tagged PPE51 and HA-tagged PE19 (or PE25) were
cross-linked by DSP. PPE51 was precipitated using
nickel–nitrilotriacetic acid resin, and coprecipitated PE
proteins were visualized by Western blot (WB) using
anti-HA antibodies. DTT, dithiothreitol; IP, immuno-
precipitation. (D) Export of PPE51 inDppe51and
Dppe25-pe19mutants. TheDppe51andDppe25-pe19
cells expressing His-tagged PPE51 were grown in
detergent-free Sauton’s medium. Surface proteins
were extracted by 1%n-octyl-b-D-glucopyranoside
(OBG). The presence of PPE51 in WCL, OBG extracts,
and culture filtrate (CF) was examined by anti-His antibodies.
C
WB IP
37KD
15KD
PPE51-His
PE19-HA
PPE51-His
PE25-HA
Anti-His
Anti-HA
50KD
15KD
DTT
PPE51-HisPE19-HAPPE51-His
PE25-HA
Anti-HA
WCLOBGCFWCLOBGCF
Δppe51+
ppe51his
Δppe25-pe19+
ppe51his
PPE51His
GroEL
Esat-6
D
AB
0
50
100
150
10 -2 10 -1 100 101 102 103
Drug concentration (μg/ml)
Bacteria Growth (%)
wt
Δppe25-pe19
Δppe25
Δppe26
Δppe27
Δpe18
Δpe19
10 -2 10 -1 100 101 102 103
Drug concentration (μg/ml)
0
50
100
150
Bacteria Growth (%)
wt
Δpe19
wt+pe18anti (-aTC)
wt+pe18anti (+aTC)
RESEARCH | REPORT