Toll-like receptor 5 (TLR5), IL-1β secretion, and
asparagine endopeptidase are critical factors for
alveolar macrophage phagocytosis and bacterial killing
Delphyne Descampsa,b, Mathieu Le Garsa,b,c, Viviane Balloya,b, Diane Barbiera,b, Sophia Maschalidid,e,f, Mira Tohmed,e,f,g,
Michel Chignarda,b, Reuben Ramphala,b,h, Bénédicte Manouryd,f,1, and Jean-Michel Sallenavea,b,c,1
aUnité de Défense Innée et Inflammation, Institut Pasteur, 75724 Paris, France;bInstitut National de la Santé et de la Recherche Médicale (INSERM) U874,
75724 Paris, France;cUniversité Paris Diderot, Sorbonne Paris Cité (Cellule Pasteur), 75013 Paris, France;dINSERM U1013, Hôpital Necker-Enfants Malades,
75015 Paris, France;eEcole Doctorale Gc2iD, Université Paris Descartes, 75006 Paris 5, France;fUniversité Paris Descartes, Sorbonne Paris Cité, Faculté de
Médecine, 75015 Paris, France;gInstitut Curie, INSERM U932, 75005 Paris, France; andhDepartment of Medicine, University of Florida, Gainesville, FL 32610
Edited* by Ruslan Medzhitov, Yale University School of Medicine, New Haven, CT, and approved December 16, 2011 (received for review May 27, 2011)
A deficit in early clearance of Pseudomonas aeruginosa (P. aerugi-
nosa) is crucial in nosocomial pneumonia and in chronic lung infec-
tions. Few studies have addressed the role of Toll-like receptors
(TLRs), which are early pathogen associated molecular pattern
receptors, in pathogen uptake and clearance by alveolar macro-
phages (AMs). Here, we report that TLR5 engagement is crucial for
bacterial clearance by AMs in vitro and in vivo because unflagel-
lated P. aeruginosa or different mutants defective in TLR5 activa-
tion were resistant to AM phagocytosis and killing. In addition, the
clearance of PAK (a wild-type P. aeruginosa strain) by primary AMs
was causally associated with increased IL-1β release, which was
dramatically reduced with PAK mutants or in WT PAK-infected pri-
mary TLR5−/−AMs, demonstrating the dependence of IL-1β produc-
tion on TLR5. We showed that this IL-1β production was important
in endosomal pH acidification and in inducing the killing of bacteria
by AMs through asparagine endopeptidase (AEP), a key endosomal
cysteine protease. In agreement, AMs from IL-1R1−/−and AEP−/−
mice were unable to kill P. aeruginosa. Altogether, these findings
demonstrate that TLR5 engagement plays a major role in P. aeru-
ginosa internalization and in triggering IL-1β formation.
uginosa, is particularly important in nosocomial pneumonia
and in chronic lung diseases such as cystic fibrosis (1). Alveolar
macrophages (AMs) lie at the forefront of lung defense against
pathogens such as P. aeruginosa. The main function of AMs is
to clear pathogens (2), and a deficiency in early recognition of
P. aeruginosa by AMs has been suspected in these pathologies
(3, 4). Research has shown that pathogen-associated molecular
patterns (PAMPs) are recognized by specific Toll-like receptors
(TLRs) at the surface of phagocytes and mucosal epithelial cells.
Surprisingly, although numerous studies have associated the li-
gand-induced TLR engagement to cytokine and chemokine
production from phagocytes (5, 6), comparatively fewer studies
have investigated the importance of TLRs in pathogen phago-
cytosis and killing. Furthermore, these studies have mostly used
macrophages from bone marrow-differentiated cells (BMDMs),
few have used live bacteria, and even fewer have used flagellated
bacteria such as P. aeruginosa. Moreover most studies have used
primed phagocytes (with LPS, zymosan) to boost pathogen up-
take. Despite these caveats, the recruitment of membrane TLRs
to phagosomes upon phagocytosis has been demonstrated (7–
10), except for TLR5. TLR2, TLR4, and the adaptor molecule
MyD88 have been shown to be important molecules in pro-
cessing of heat-killed Escherichia coli and Staphylococcus aureus
by BMDMs in late endosomes and lysosomes (9–11), suggesting
that a blockade in phagosome maturation was occurring in
phagocytes deleted for these molecules.
he opportunist Gram-negative bacterium, Pseudomonas aer-
TLR5 is thought to be one of the key receptors implicated in
the recognition of P. aeruginosa (5, 8), but no information is
available about whether this molecule is important for bacterial
phagocytosis. Here, we sought to investigate specifically the in-
volvement of TLR5 in P. aeruginosa phagocytosis and killing by
AMs. We describe a pathway linking P. aeruginosa engagement
to TLR5 through MyD88, followed by IL-1β release and the
involvement of a lysosomal cysteine protease (asparagine endo-
peptidase, AEP), all events concurring to AM-mediated P. aer-
uginosa phagocytosis and bacterial killing.
TLR5 Engagement Is Required for P. aeruginosa Clearance. To de-
termine whether the TLR5 ligand flagellin/flagella were involved
in P. aeruginosa clearance by AMs, we infected murine alveolar
MH–S cells with either WT P. aeruginosa strain PAK or PAK
mutants deficient in the expression of flagellin, the primary fla-
gellar subunit (the unflagellated PAKΔfliC), or expressing a fla-
gellin monomer mutated in the TLR5-recognition site (PAKL88
and PAKL94; Table S1). FACS analysis indicated that MH–S
cells expressed TLR5 (Fig. S1A). In the first part of our study,
both AM supernatants and cell lysates were pooled for P. aeru-
ginosa numeration (killing assays). In that context, MH–S cells
significantly killed PAK, whereas no clearance of PAKΔfliC or
flagellin-mutated PAKL88 or PAKL94 was observed, regardless
of the duration of infection or the multiplicity of infection (MOI)
used (Fig. 1A and Fig. S1B). These results were confirmed in
primary WT AMs (Fig. 1B). In all of these experiments, the
bacterial counts of all mutants continued to increase, in contrast
to WT PAK. Moreover, AMs from TLR5−/−mice were unable to
kill PAK, compared with WT AMs (Fig. 1C), whereas the bac-
terial clearance ability of AMs from TLR4−/−mice was not im-
paired. Importantly, MyD88−/−mice, the TLR5 mRNA levels of
which were comparable to those of WT mice (Fig. S1C), were
also unable to eliminate bacteria, underscoring the role of
MyD88 in TLR5 signaling.
The resistance of unflagellated PAKΔfliC to AM killing clearly
pointed to the P. aeruginosa flagellum as the organelle mainly
implicated in AM stimulation (12). However, our data did not
rule out that the flagellin monomer might be responsible for AM
Author contributions: D.D., M.C., R.R., B.M., and J.-M.S. designed research; D.D., M.L.G.,
V.B., D.B., S.M., M.T., and B.M. performed research; S.M., M.T., R.R., B.M., and J.-M.S.
contributed new reagents/analytic tools; D.D., M.L.G., S.M., M.T., M.C., B.M., and J.-M.S.
analyzed data; and D.D., B.M., and J.-M.S. wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
1To whom correspondence may be addressed. E-mail: firstname.lastname@example.org or
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| January 31, 2012
| vol. 109
| no. 5
induction of bacterial killing. We, therefore, used an unflagel-
lated PAK mutant, PAKD (Table S1), which, unlike PAKΔfliC,
is still able to release high amounts of flagellin in the supernatant
(Fig. S2A). Fig. 1D shows that PAKD, despite its ability to signal
through TLR5 (TNFα output, 1201 ± 454 pg/mL vs. 2473 ± 487
pg/mL for WT PAK), was completely resistant to primary AM
killing. Bacteria centrifuged onto cells to force bacteria–cell
contact, did not increase PAKD killing by primary AMs, rein-
forcing the importance of specific functional flagellum–TLR5
interaction for P. aeruginosa clearance.
The implication of TLR5 recognition in P. aeruginosa clear-
ance was further confirmed in vivo. Following intratracheal
P. aeruginosa administration, we showed that 67% of the WT PAK
inoculum was recovered in WT mice bronchoalveolar lavage
(BAL) 2 h postinfection, whereas 101% of the same inoculum
remained after mouse TLR5−/−infection (Fig. 1E). Moreover,
all of the original PAKL94 inoculum remained and continued to
proliferate after infection of WT mice (124%), suggesting a delay
in bacterial clearance, compared with WT PAK. Importantly,
these differences cannot be attributed to variations in the cellular
composition of the airspaces (97−100% of AMs and 0−3% of
neutrophils), and the total cell counts were not significantly
different regardless of the bacteria strains used, thereby impli-
cating solely the AMs in the events described.
TLR5/MyD88 Signaling Is Important for P. aeruginosa Phagocytosis.
Because bacterial clearance could not be explained either by
bactericidal activity of AM-infected supernatants (Table S2) or
by AM cytotoxicity, we assessed whether the differential killing
between WT and PAK mutants was attributable to different AM
uptake. MH–S cells were incubated with bacteria for 1 h, fol-
lowed by antibiotic treatment, and then the amount of phago-
cytosed bacteria was determined in cell lysates only. Fig. 2A
shows a deficit in phagocytosis for PAKL88 and PAKL94 com-
pared with WT PAK, whereas mutant CFU counts increased in
supernatants. Using FITC-labeled bacteria (Fig. 2B), FACS
analysis showed a near complete absence of labeling in MH–S
cells with PAKΔfliC-FITC, i.e., an almost complete deficit of
phagocytosis, whereas PAKL88-FITC and PAKL94-FITC pre-
sented an intermediate phenotype of AM uptake, with PAKL88-
FITC being less phagocytosed than PAKL94-FITC. To confirm
that the P. aeruginosa uptake is dependent upon flagellum–TLR5
(and not flagellin–TLR5) interaction, we performed phagocyto-
sis assays with PAKD. In Fig. S2B, PAKD (despite high levels
of secreted flagellin; Fig. S2A) was less phagocytosed than WT
P. aeruginosa, even when bacteria–cell contact was increased
by centrifugation, demonstrating that bacterial internalization is
not attributable to flagellin/TLR5 engagement. Immunoblotting
of intracellular flagellin also reflected differences in bacterial
after AM infection (%)
MH S cells
inoculum in BAL (%)
after AM infection (%)
after AM infection (%)
PAK, PAKΔfliC, PAKL88, or PAKL94 mutants [MOI: 0.1 (B) or 10 (A)]. (C) Primary WT, TLR5−/−, TLR4−/−, or MyD88−/−AMs were infected for 4 h with WT PAK
(MOI: 0.1). (D) Primary AMs were infected for 4 h with either WT PAK or PAKD (MOI: 0.1) with or without centrifugation. (A–D) CFU were quantified in AM
supernatants pooled with cell lysates. Results are means ± SD of three experiments (**P < 0.01; ***P < 0.001) and are expressed as percentages as follows:
(CFU counts recovered without AMs − CFU counts recovered after AM infection) × 100. (E) WT or TLR5−/−mice were infected intratracheally with 105CFU
of PAK or PAKL94. BALs were performed 2 h later and plated on LB agar plates to quantify total CFU counts. Results are means ± SD of three experiments
(*P < 0.05; ***P < 0.001) and are expressed as percentage of surviving bacteria in BAL = (total CFU counts in BAL/ CFU counts of inoculum) × 100.
P. aeruginosa clearance by AMs is dependent on efficient TLR5-Flagella interaction. MH–S cells (A) or primary AMs (B) were infected for 4 h with WT
signaling. (A and C) MH–S cells were infected for 1 h with WT PAK, PAKL88,
or PAKL94 mutants (A), or primary WT, TLR5−/−, MyD88−/−, or TLR4−/−AMs
were infected with WT PAK (MOI: 10) for 1 h (C). After stimulation, cells
were washed and treated with tobramycin and the number of ingested
bacteria was determined by counting CFU from AM lysates on LB agar plates.
Results are means ± SD of three experiments (***P < 0.001) and are
expressed as percentages of relative phagocytosis index = (CFU counts in
mutant PAK-treated cells/CFU counts in WT PAK-treated cells) × 100. (B)
FACS analysis of MH–S cells following 1 h infection with FITC-labeled WT or
FITC-mutant P. aeruginosa strains. Results are representative of three in-
Phagocytosis of P. aeruginosa by AMs is dependent on TLR5-MyD88
| www.pnas.org/cgi/doi/10.1073/pnas.1108464109Descamps et al.
uptake (Fig. S2A). Phagocytosis was shown to be TLR5 receptor-
specific, because centrifugation did not significantly abolish dif-
ferences in AM uptake between WT PAK and mutants (Fig. S2
B–D). Moreover, competition experiments with P. aeruginosa
flagellin significantly reduced PAK binding (Fig. S2E). Finally,
such a deficit in P. aeruginosa phagocytosis was shown in TLR5−/−
and MyD88−/−primary AMs (Fig. 2C). In contrast, TLR4 was
shown to play no role in P. aeruginosa internalization by primary
AMs. Taken together, these observations established the impor-
tance of flagellum/TLR5–MyD88 engagement for P. aeruginosa
phagocytosis of primary AMs.
IL-1β Production Is Dependent on TLR5 Signaling. Although PAKL94
was not killed, it was shown to be better phagocytosed than
PAKL88, suggesting that other intracellular PAMP receptors
might be important in the detection and the killing of P. aeru-
ginosa by AMs. Because the intracellular PAMP receptor ICE
protease-activating factor (IPAF) participates in the processing
of pro-IL-1β to IL-1β through caspase-1/inflammasome-de-
pendent activation in P. aeruginosa-infected macrophages (13,
14), we measured mature secreted IL-1β levels in primary AM
supernatants as a potential index of inflammasome involvement.
We observed that high levels of IL-1β were associated with AM-
mediated PAK killing and low IL-1β levels were detected with
PAKL94 infection (Fig. 3A). The reduction of IL-1β by PAKL94-
treated AMs was associated with a delay and a reduction in pro-
IL-1β (31 kDa) production, leading to an important inhibition in
mature IL-1β conversion (17 kDa) 4 h postinfection (Fig. 3B).
We have demonstrated that purified flagellin was able to induce
the pro-IL-1β synthesis but not the caspase-1 p10 and, therefore,
was not involved in the mature IL-1β generation (Fig. S3 A–C).
The dependence of IL-1β production on TLR5 was further
demonstrated by using PAK-infected primary TLR5−/−AMs.
Indeed, for these cells, a significant reduction of IL-1β secretion
was observed, compared with WT AMs, caused by a delay in pro-
IL-1β production 2 h postinfection and a decrease in mature IL-
1β conversion (Fig. 3 C and D).
Importantly, the deficiency in mature IL-1β conversion ob-
served in PAKL94-infected WT AMs was correlated with a re-
duction in mature caspase-1 p10 production (Fig. S4A), whereas
this was not the case with PAK-infected TLR5−/−AMs, com-
pared with WT AMs (Fig. S4B).
IL-1β Is Required for P. aeruginosa Bacterial Killing. To examine the
connection between IL-1β production and bacterial killing, we
studied AM infection with PAKΔpscF, a PAK mutant deficient
in the type III secretion system (T3SS) (14), which has been
described to block the caspase-1-dependent processing of IL-1β
(15). Whereas TNFα secretion was similar between WT PAK-
and PAKΔpscF-infected AMs, released IL-1β levels were un-
detectable in supernatants after infection of primary AMs with
this mutant (Fig. 4A). Western blot analysis showed that, al-
though PAKΔpscF induced high levels of pro-IL-1β, mature IL-
1β was not produced (Fig. 4B). This is likely attributable to its
inability to activate caspase-1, as evidenced by the absence of the
caspase-1 p10 subunit in PAKΔpscF-infected AMs 4 h post-
infection (Fig. S4C). Importantly, although PAKΔpscF was as
efficiently phagocytosed by AMs (125% ± 28%) as WT PAK
(100% ± 25%), this nonmature IL-1β producer was not killed
by primary AMs (Fig. 4C), validating a link between defective
killing and a major reduction of IL-1β secretion.
Importantly, Fig. 4D showed that exogenously added IL-1β
restored PAKL94 killing to levels equivalent to that of WT PAK.
We also demonstrated that IL-1 receptor antagonist (IL-1RA),
an inhibitor of IL-1β signaling, was able to inhibit MH–S cell
killing of WT PAK (Fig. 4E), without affecting IL-1β release
(3867 ± 649 pg/mL with IL-1RA vs. 3393 ± 309 pg/mL without
IL1-RA; 5 × 105AMs). To reinforce that point, we performed
additional experiments on primary AMs from IL-1R1−/−mice.
Fig. 4F clearly demonstrated that primary IL-1R1−/−AMs were
unable to kill WT P. aeruginosa, compared with WT AMs. Fur-
thermore, we measured the same amount of IL-1β in AMs from
IL-1R1−/−(333 ± 144 pg/mL) and WT (271 ± 34 pg/mL) mice,
demonstrating the autocrine feedback of IL-1β on P. aeruginosa
clearance by primary AMs.
IL-1β-Induced Bacterial Killing Is Dependent on AEP. Because bac-
terial fate is ultimately determined in acidic phagolysosome
compartments, we then set out to determine the mechanisms
responsible for PAK killing in AMs. We showed that blocking
phagolysosome acidification by bafilomycin A (BafA) treatment
(Fig. S5A), an inhibitor of vacuolar type H+-ATPase, inhibited
significantly PAK killing in primary AMs (Fig. S5B), whereas IL-
1β release was unaffected (Fig. S5C). Using specific probes to
measure the pH in endolysosomes, we showed that exogenous
administration of IL-1β significantly reduced the pH in MH–S
cell endosomes (Fig. 5A), suggesting that IL-1β-mediated acidi-
fication of AM phagolysosomes may be a key element in the
control of bacterial killing.
Because most of the lysosomal cysteine proteases are de-
pendent on a low pH for their activities, we chose to study the
role of the protease AEP in PAK clearance. Using confocal
microscopy, AEP staining was detected in endosomal EEA1+
vesicles in noninfected primary AMs. In contrast, AEP labeling
was significantly decreased upon PAK challenge, whereas a re-
duction of staining was much less pronounced in PAKL94-
infected AMs (Fig. 5B). Immunoblotting (Fig. 5C) and enzy-
matic activity assays using AEP-specific fluorogenic substrate
(Fig. 5D) confirmed a significant AEP consumption in WT PAK-
infected AMs 4 h postinfection, whereas this was not the case
with PAKL94-infected primary AMs. However, exogenously
added IL-1β restored AEP consumption in PAKL94-infected
AMs 3 h postinfection (Fig. 5 C and D). AEP consumption
was associated with mature IL-1β secretion but was independent
of bacterial uptake, because PAKΔpscF, which is efficiently
WT Primary AMTLR5/
2 h 4 h
2 h 4 h
[IL1- ] (ng/mL)
[IL1- ] (ng/mL)
2 h 4 h-
2 h 4 h
primary AMs from WT (A and B) or TLR5−/−(C and D) mice were infected with
WT PAK, PAKL88, or PAKL94 (MOI: 1). IL-1β secretion was measured in
supernatants 4 h postinfection (A and C). Results are means ± SD of three
experiments (***P < 0.001). NI, noninfected cells; nd, not detected. Lysates
obtained from primary WT (B) or TLR5−/−(D) AMs infected with bacteria
were analyzed by immunoblotting for pro-IL-1β processing. Equal loading
was controlled for by β-actin detection. Results are representative of three
IL-1β production in primary AMs is dependent on TLR5. A total of 105
Descamps et al.PNAS
| January 31, 2012
| vol. 109
| no. 5
phagocytosed but unable to produce IL-1β, did not affect AEP
expression during infection (Fig. S6A).
Furthermore we showed that recombinant AEP can exert
a dose-dependent direct lytic activity on P. aeruginosa at pH 6
(Fig. S6B). Finally, the importance of AEP activity in AM-
mediated bacterial killing of P. aeruginosa was confirmed in
primary AEP−/−AMs. Although PAK was phagocytosed
at comparable levels in primary AEP−/−AMs and WT AMs
(Fig. S6C), we observed that primary AEP−/−AMs were un-
able to kill PAK, compared with WT AMs (Fig. 5E). The
absence of killing by primary AEP−/−AMs was not caused by
a deficiency in pro-IL-1β processing, because these cells did
not secrete less mature IL-1β compared with WT AMs (Fig. S6
D and E).
Using nonopsonic conditions, thus mimicking resting and naïve
conditions in the alveolar space, we show that TLR5 and MyD88
(but not TLR4) are of paramount importance for P. aeruginosa
phagocytosis and killing. Indeed, P. aeruginosa mutants for the
TLR5-recognition site of flagellin monomer domain D1 [PAKL88
or PAKL94 (16, 17)] were resistant to AM killing, and primary
TLR5−/−AMs were unable to clear P. aeruginosa. Importantly, in
these unprimed conditions, bacterial flagellum was shown to be
key for AM–P. aeruginosa interaction because unflagellated
mutants, PAKΔfliC or PAKD (the latter secreting flagellin in the
supernatant), were equally resistant to AM killing. Crucially, the
importance of flagellum/TLR5 interaction on bacterial killing was
2 h 4 h 2 h 4 h
after AM infection (%)
after AM infection (%)
Primary AM WT IL-1R1/
after AM infection (%)
after AM infection (%)
aeruginosa. (A) Primary WT AMs (105) were infected
with WT PAK or PAKΔpscF mutant (MOI: 1) for 4 h. TNFα
or IL-1β was measured in supernatants. Results are
means ± SD of three experiments. NI, noninfected cells;
nd, not detected. (B) Lysates from infected AMs were
assessed for pro-IL-1β processing. Equal loading was
controlled by β-actin detection. Results are representa-
tive of three independent experiments. (C) Primary WT
AMs were infected with WT PAK or PAKΔpscF (MOI: 0.1)
for 4 h. (D and E) MH–S cells were infected with WT PAK
or PAKL94 mutant (MOI: 0.1) for 4 h in the presence or
not of IL-1β or IL-1 receptor antagonist (IL-1RA). (F) Pri-
mary WT or IL-1R1−/−AMs were infected for 4 h with WT
PAK (MOI: 0.1). (C–F) CFU were quantified in super-
natants pooled with cell lysates. Results are means ± SD
of three experiments (*P < 0.05; **P < 0.01) and are
expressed as in Fig 1.
IL-1β secretion is required for AMs to kill P.
Results are means ± SD of three experiments. (B) Confocal microscopy of AEP was measured in primary WT AMs infected or not for 3 h with WT PAK or
PAKL94. Quantification was performed using Image J software (n = 5). (C) Immunoblotting AEP expression in lysates from WT PAK- or PAKL94-infected
primary WT AMs (MOI: 1) with or without IL-1β. Equal loading was controlled by β-actin detection. Results are representative of three independent
experiments. (D) AEP activity was measured in lysates from WT PAK- or PAKL94-infected MH–S cells (MOI: 10) with a fluorometer by assessing the hydrolysis of
the specific substrate for AEP. (A–D) Unpaired t test (*P < 0.01; **P < 0.005; ***P < 0.0001). NI, noninfected cells. (E) Primary WT or AEP−/−AMs were infected
with WT PAK (MOI: 0.1) for 4 h. CFUs were quantified in supernatants pooled with cell lysates. Results are means ± SD of three experiments (***P < 0.001) and
are expressed as in Fig 1.
AEP protease is essential for P. aeruginosa killing by primary AMs. (A) Kinetics of endolysosomal pH in MH–S cells ± IL-1β (10 ng/mL) was measured.
| www.pnas.org/cgi/doi/10.1073/pnas.1108464109 Descamps et al.
confirmed by using primary TLR5−/−, TLR4−/−, or MyD88−/−
AMs, as well as in vivo, where we demonstrated that 2 h post-
infection, TLR5 expression on AMs was the most important factor
in clearing bacteria and in containing pulmonary infection.
Flagellum/TLR5-mediated killing was not related to AM-se-
creted antimicrobial activity, but was associated with TLR5-de-
pendent intracellular entry, because PAK mutants for the TLR5-
recognition sites L88 and L94 were less phagocytosed by MH–S
cells than WT PAK. The necessity of TLR5/MyD88 engagement
for P. aeruginosa phagocytosis was also confirmed by using pri-
mary TLR5−/−and MyD88−/−AMs. Importantly, TLR4 signal-
ing did not play any role in P. aeruginosa uptake by primary AMs.
Loss of bacterial motility (18) was not a relevant factor in our
model. Indeed, PAKL88 and PAKL94 are fully motile bacteria
(16); however, none of the mutants was killed by AMs. More-
over, after centrifugation to force bacteria–cell contact, flagellin-
mutated bacteria or unflagellated mutants were still considerably
less phagocytosed than WT bacteria, demonstrating that the
phagocytic phenotype was not attributable to a mobility-deficit or
flagellin/TLR5 engagement but was related to diminished spe-
cific flagellum/TLR5 interaction.
Unexpectedly, the use of PAKL88 and PAKL94 revealed
differences in P. aeruginosa phagocytosis and killing. Indeed,
although both mutants were equally resistant to AM killing,
PAKL88 stayed mostly extracellular, whereas PAKL94 was bet-
ter phagocytosed by AMs, albeit not as efficiently as WT PAK.
Although a short stretch of 10 aa (N-88-LQRIRDLALQ-97) in
the flagellin monomer was known to be crucial for TLR5 rec-
ognition (16–18), our data demonstrate that the L88 residue is
important for binding/recognition by TLR5 and intracellular
entry, whereas the L94 residue may be more important for in-
tracellular downstream events. Indeed, Franchi et al. (14) dem-
onstrated that PAKL94 was much less efficient at inducing
mature IL-1β than WT PAK in BMDMs, presumably because of
a reduced interaction of bacteria with the intracellular receptor
IPAF. These data demonstrated that L94 was a critical residue
for the engagement of the inflammasome via IPAF, in addition
to residues present at the C terminus of the D0 domain of fla-
gellin monomer (14, 17, 19). Alternatively, other Nod-like
receptors such as Naip5 (19, 20) may also be involved in the
differential binding of flagellin L94.
Another important finding of our study is that TLR5 is im-
portant for production of IL-1β in AMs. The latter result con-
trasts with that of Franchi and coworkers (14, 21), who studied P.
aeruginosa and Salmonella infection of BMDMs and showed,
without studying the effect on phagocytosis and bacterial killing,
that TLR5 was redundant and that IPAF and ASC were the
two inflammasome constituents regulating IL-1β synthesis. Two
things may explain this discrepancy: first, whereas these authors
used BMDMs, we have used AMs; second, and crucially,
whereas they primed BMDMs with LPS before bacterial in-
fection to increase pro-IL-1β production, our protocol did not
involve priming, and the production of mature IL-1β occurred
following a single event of P. aeruginosa infection (4 h). In our
study, TLR5 is essential for bacterial internalization, production,
and processing of pro-IL-1β (presumably through IPAF). In the
study mentioned above (14, 21), BMDM priming with LPS
probably bypassed the need for TLR5 through other receptors
(e.g., TLR4) for pro-IL-1β generation. The latter events in-
volving inflammasome activation most likely occurred in a simi-
lar fashion in both studies.
The role of inflammasome has been mainly investigated using
intracellular bacteria (Legionella, Shighella) (22, 23), but rela-
tively few studies have addressed this issue in AMs infected with
extracellular bacteria (e.g., P. aeruginosa). How can P. aerugi-
nosa, then, activate the inflammasome if this bacterium fate is
destruction in the phagolysosome? No system to evade the
phagolysosome has been described for P. aeruginosa. However,
a cytosolic leakage/transport of bacterial constituents is a possi-
bility. The main vector for this transfer may be the T3SS, and
the main ligand may be flagellin itself, although most studies
used, for their demonstration, elegant but artificial transfection
methods to introduce flagellin in the cytosol to activate IPAF.
Indeed, we showed that extracellularly added flagellin, although
able to produce pro-IL-1β, is unable to activate procaspase-1
and, consequently, to induce the mature IL-1β secretion.
Moreover, PAKΔpscF, a mutant deficient in T3SS unable to
engage IPAF (13, 14), was not killed by AMs and completely
failed to induce the mature IL-1β production (although pro-IL-
1β was produced), despite the presence of flagellum and its
proper internalization. This demonstrates that IL-1β processing
needs both the flagellum and T3SS to activate the inflamma-
some. Crucially, IL-1β production is not merely a bystander in
the killing process but is causative because WT AMs treated with
IL-1RA or AMs from IL-1R1−/−mice could not kill P. aerugi-
nosa. This clearly demonstrates the existence of an autocrine
loop, involving IL-1β production, IL-1β/IL-1R interaction, and
subsequent P. aeruginosa clearance. It has been shown recently
that the IL-1β processing (procaspase-1/pro-IL-1β) may, in fact,
be operative in lysosomes (24, 25), bringing this machinery in
close contact with incoming bacteria and, presumably, decreasing
the time of reaction for mature IL-1β production upon bacterial
contact. Moreover, this vesicular localization of procaspase-1/
caspase-1 may explain why IL-1β release precedes cell lysis, be-
cause in the time frame of our experiments, no pyroptosis/cell
lysis was observed. However, it is possible that AMs may be
inherently more resistant to pyroptosis because this cell type
performs a crucial sentinel role within tissues. Interestingly,
Carvalho et al. have shown that IL-1RA is produced by AMs in
response to flagellin after 4 h, suggesting that this counter-
regulatory molecule is only secreted at a late time point in the
phagocytic process and may not impair bacterial clearance (26).
We went a step further in the dissection of the mechanism by
showing that IL-1β increased phagolysosomal acidification. Be-
cause the latter is critical for the activation of cysteine proteases
involved in bacterial degradation (27), we studied specifically
one of them, AEP, which controls the activation of other lyso-
somal cysteine proteases (28). By using three independent
techniques, we observed a reduction of AEP levels upon in-
fection with WT PAK, but not with PAKL94 or PAKΔpscF or
purified flagellin (Fig. S3D). AEP was shown to be a key factor
participating in AM-mediated P. aeruginosa killing because
AEP−/−AMs were unable to kill WT PAK.
In summary (Fig. S7), we describe a phagocytic pathway,
demonstrating the role of TLR5 in AM phagocytosis after P.
aeruginosa early infection. This occurs in a T3SS-dependent
fashion, leading to caspase-1 and IL-1β maturation. The latter
induces, after acidification of the phagolysosome, the consump-
tion of AEP, a protease that we demonstrate to be a key factor in
P. aeruginosa killing. Our results give a mechanistic insight into
early events following P. aeruginosa infection of AMs in the lung
and extend our understanding of the crucial role of TLR5 in the
defense of the lung against this important pathogen (29, 30–32).
Materials and Methods
Mouse Primary AMs and MH–S Cells. TLR4−/−, TLR5−/−, MyD88−/−(S. Akira,
Osaka University, Osaka, Japan), and IL-1R1−/−mice (CDTA) were back-
crossed 8 times with C57BL/6J. C57BL/6J mice (Janvier) were used as control
mice. Care and use of the mice was in accordance with Institut Pasteur
guidelines in compliance with the European Animal Welfare regulations.
Mouse primary AMs were isolated after lung washing with PBS (33). AMs
were plated in complete RPMI medium (supplemented with 2 mM L-gluta-
mine, 1% antibiotic, and 5% inactivated FBS). After 2 h, medium was re-
moved and AMs were incubated overnight with fresh medium. Mouse AM
cell line MH–S (CRL-2019; ATCC) was maintained in complete RPMI medium
supplemented with 1% sodium pyruvate. Primary AMs or MH–S cells were
placed into serum- and antibiotic-free RPMI for 5 h before infection.
Descamps et al.PNAS
| January 31, 2012
| vol. 109
| no. 5
MA) is a commonly studied P. aeruginosa strain. All modified mutants used
were derived from PAK (Table S1). Bacteria were grown overnight in LB
broth at 37 °C and then transferred to fresh medium and grown by shaking
at 100 rpm for 4–5 h to mid-log phase. The culture was centrifuged at 3,000
× g and the pellet was washed and resuspended in PBS. The OD600nmwas
adjusted to give the desired inoculum. Inoculum was verified by serial
dilutions plated on LB agar to determine the number of colony-forming unit
(CFU). Bacterial growth of each PAK strain used in this study was shown to
Bacterial Clearance Assay. A total of 105AMs were infected with bacteria
(MOI: 0.1). After 2–4 h, supernatants were collected and cells were lysed with
0.1% Triton X-100 (a concentration that did not affect PAK viability) in H2O
in sequential washes to harvest total bacteria. To quantify total viable
bacteria, pooled cell supernatants and lysates were diluted and plated on LB
agar to determine CFU scores. Results are expressed as percentages as fol-
lows: (CFU counts recovered without AMs − CFU counts recovered after AM
infection) × 100. For specific experiments, centrifugation was performed (80
× g; 4 min) to ensure synchronous contact between WT or nonmotile
P. aeruginosa and AMs. To test bacterial clearance in vivo, mice were anes-
thetized by ketamine-xylazine intramuscular injection, and then infected
intratracheally with 105CFU (50 μL). Mice were euthanized 2 h later. BAL
fluids (2 × 250 μL pooled) were performed, diluted, and plated on LB agar
plates to quantify total CFU counts. The percentage of surviving bacteria in
BAL was assessed as follows: (total CFU counts in BAL/CFU counts of in-
oculum) × 100. Total cell counts were measured with a Coulter Beckman
Counter. Cell differential counts were determined after cytospin centrifu-
gation and staining with Diff-Quik products (Medion Diagnostics).
Phagocytosis Assay. A total of 5 × 105AMs were infected with bacteria (MOI:
10) to 1 h. Free and adherent bacteria were removed by washing cells with
PBS and were killed with tobramycin treatment (40 μg/mL; 30 min). Then, cells
were washed and lysed in H2O containing 0.1% Triton X-100. The number of
bacteria in lysates was determined by counting CFU on LB agar plate. The
percentage of relative phagocytosis index was assessed as follows: (CFU
counts in mutant PAK-treated cells/CFU counts in WT PAK-treated cells) × 100.
In other experiments, bacteria were labeled with 0.1 mg/mL FITC (Sigma) in
Na2CO3buffer, pH 9.5, at 37 °C while shaking at 100 rpm for 1 h. After in-
cubation and treatment as described above, cells were resuspended in PBS-
EDTA 2 mM, and then fixed in 3% PFA. To quantify phagocytosed bacteria,
fixed cells were used for FACS analysis of cell-associated fluorescence.
ELISA. IL-1β and TNFα were assayed after stimulation using DuoSet ELISA
(R&D Systems). Recombinant murine IL-1β and IL-1RA were purchased from
Endosomal pH Measurement. Endosomal pH measurement assays have been
described previously (28). MH–S cells were pulsed with 1 mg/mL FITC- and
Alexa-647-labeled 40-kDa dextrans (Molecular Probes) for 10 min at 37 °C
and washed with PBS with 1% BSA. Cells were chased and analyzed by FACS,
via a FL1/FL4 gate selective for cells that have endocytosed the probes.
AEP Protease Activity. Protease activity assays were performed on a FluoStar
Optima (BMG Labtech) by measuring the release of fluorescent N-acetyl-
methyl-coumarin in citrate buffer (pH 5.5) or PBS (pH 7.4) at 37 °C after
incubation of AEP or total lysates with its specific substrate (Z-Ala-Ala-Asn-
AEP Immunofluorescence. Primary AMs were seeded on poly-L-lysine-coated
glass coverslips. Cells were infected with bacteria (MOI: 10) for 3 h, washed,
fixed with 4% PFA, and quenched by adding 0.1 M glycine. Cells were per-
meabilized in PBS/0.05% saponin/0.2% BSA, washed, and then incubated
with anti-AEP Ab. The coverslips were mounted with Fluoromount G and
were analyzed by confocal microscopy (Zeiss confocal microscope LSM 700).
For Z-stack acquisition, several images were acquired.
Statistics. Data are presented as means ± SD. A one-way ANOVA with
Fischer’s protected least significant difference test was conducted.
ACKNOWLEDGMENTS. We acknowledge B. Solhonne for her assistance,
Prof. Z. Xing (Centre for Gene Therapeutics, McMaster University, Hamilton,
Canada) for useful discussions, and Bernhard Ryffel for providing IL-1R1−/−
mice (Unité Mixte de Recherche 6218, Centre National de la Recherche Sci-
entifique, Orléans, France). We thank Vaincre la Mucoviscidose for funding
1. Mogayzel PJ, Jr., Flume PA (2010) Update in cystic fibrosis 2009. Am J Respir Crit Care
2. Gordon S (2007) The macrophage: Past, present and future. Eur J Immunol 37(Suppl
3. Di A, et al. (2006) CFTR regulates phagosome acidification in macrophages and alters
bactericidal activity. Nat Cell Biol 8:933–944.
4. Deriy LV, et al. (2009) Disease-causing mutations in the cystic fibrosis transmembrane
conductance regulator determine the functional responses of alveolar macrophages.
J Biol Chem 284:35926–35938.
5. Kawai T, Akira S (2010) The role of pattern-recognition receptors in innate immunity:
Update on Toll-like receptors. Nat Immunol 11:373–384.
6. Medzhitov R (2007) Recognition of microorganisms and activation of the immune
response. Nature 449:819–826.
7. Underhill DM, et al. (1999) The Toll-like receptor 2 is recruited to macrophage
phagosomes and discriminates between pathogens. Nature 401:811–815.
8. Ozinsky A, et al. (2000) The repertoire for pattern recognition of pathogens by the
innate immune system is defined by cooperation between toll-like receptors. Proc
Natl Acad Sci USA 97:13766–13771.
9. Blander JM, Medzhitov R (2004) Regulation of phagosome maturation by signals from
toll-like receptors. Science 304:1014–1018.
10. Blander JM, Medzhitov R (2006) On regulation of phagosome maturation and anti-
gen presentation. Nat Immunol 7:1029–1035.
11. Doyle SE, et al. (2004) Toll-like receptors induce a phagocytic gene program through
p38. J Exp Med 199:81–90.
12. Mahenthiralingam E, Speert DP (1995) Nonopsonic phagocytosis of Pseudomonas
aeruginosa by macrophages and polymorphonuclear leukocytes requires the pres-
ence of the bacterial flagellum. Infect Immun 63:4519–4523.
13. Miao EA, Ernst RK, Dors M, Mao DP, Aderem A (2008) Pseudomonas aeruginosa ac-
tivates caspase 1 through Ipaf. Proc Natl Acad Sci USA 105:2562–2567.
14. Franchi L, et al. (2007) Critical role for Ipaf in Pseudomonas aeruginosa-induced cas-
pase-1 activation. Eur J Immunol 37:3030–3039.
15. Jyot J, et al. (2011) Type II secretion system of Pseudomonas aeruginosa: In vivo evi-
dence of a significant role in death due to lung infection. J Infect Dis 203:1369–1377.
16. Verma A, Arora SK, Kuravi SK, Ramphal R (2005) Roles of specific amino acids in the N
terminus of Pseudomonas aeruginosa flagellin and of flagellin glycosylation in the
innate immune response. Infect Immun 73:8237–8246.
17. Smith KD, et al. (2003) Toll-like receptor 5 recognizes a conserved site on flagellin
required for protofilament formation and bacterial motility. Nat Immunol 4:
18. Amiel E, Lovewell RR, O’Toole GA, Hogan DA, Berwin B (2010) Pseudomonas aeru-
ginosa evasion of phagocytosis is mediated by loss of swimming motility and is in-
dependent of flagellum expression. Infect Immun 78:2937–2945.
19. Miao EA, et al. (2010) Innate immune detection of the type III secretion apparatus
through the NLRC4 inflammasome. Proc Natl Acad Sci USA 107:3076–3080.
20. Lightfield KL, et al. (2011) Differential requirements for NAIP5 in activation of the
NLRC4 inflammasome. Infect Immun 79:1606–1614.
21. Franchi L, et al. (2006) Cytosolic flagellin requires Ipaf for activation of caspase-1 and
interleukin 1beta in salmonella-infected macrophages. Nat Immunol 7:576–582.
22. Amer A, et al. (2006) Regulation of Legionella phagosome maturation and infection
through flagellin and host Ipaf. J Biol Chem 281:35217–35223.
23. Suzuki T, et al. (2007) Differential regulation of caspase-1 activation, pyroptosis, and
autophagy via Ipaf and ASC in Shigella-infected macrophages. PLoS Pathog 3:e111.
24. Andrei C, et al. (2004) Phospholipases C and A2 control lysosome-mediated IL-1 beta
secretion: Implications for inflammatory processes. Proc Natl Acad Sci USA 101:
25. Wewers MD (2004) IL-1beta: An endosomal exit. Proc Natl Acad Sci USA 101:10241–
26. Carvalho FA, Aitken JD, Gewirtz AT, Vijay-Kumar M (2011) TLR5 activation induces
secretory interleukin-1 receptor antagonist (sIL-1Ra) and reduces inflammasome-as-
sociated tissue damage. Mucosal Immunol 4:102–111.
27. Bird PI, Trapani JA, Villadangos JA (2009) Endolysosomal proteases and their in-
hibitors in immunity. Nat Rev Immunol 9:871–882.
28. Sepulveda FE, et al. (2009) Critical role for asparagine endopeptidase in endocytic
Toll-like receptor signaling in dendritic cells. Immunity 31:737–748.
29. Balloy V, et al. (2007) The role of flagellin versus motility in acute lung disease caused
by Pseudomonas aeruginosa. J Infect Dis 196:289–296.
30. Feuillet V, et al. (2006) Involvement of Toll-like receptor 5 in the recognition of
flagellated bacteria. Proc Natl Acad Sci USA 103:12487–12492.
31. Ramphal R, Balloy V, Huerre M, Si-Tahar M, Chignard M (2005) TLRs 2 and 4 are not
involved in hypersusceptibility to acute Pseudomonas aeruginosa lung infections. J
32. Sutterwala FS, et al. (2007) Immune recognition of Pseudomonas aeruginosa medi-
ated by the IPAF/NLRC4 inflammasome. J Exp Med 204:3235–3245.
33. Gonçalves de Moraes VL, Singer M, Vargaftig BB, Chignard M (1998) Effects of roli-
pram on cyclic AMP levels in alveolar macrophages and lipopolysaccharide-induced
inflammation in mouse lung. Br J Pharmacol 123:631–636.
| www.pnas.org/cgi/doi/10.1073/pnas.1108464109Descamps et al.