Inflammasomes are cytosolic multi-protein complexes assembled by intracellular nucleotide-binding
oligomerization domain (NOD)-like receptors (NLRs) and initiate innate immune responses
to invading pathogens and danger signals by activating caspase-1
1
. Caspase-1 activation leads to maturation and release of the proinflammatory cytokines
interleukin (IL)-1β and IL-18 as well as a lytic inflammatory cell death termed pyroptosis
2
. Recently, a novel non-canonical inflammasome was described that activates caspase-11,
a proinflammatory caspase required for LPS-induced lethality
3
. This study also highlighted that previously generated caspase-1 knockout mice lack
a functional allele of Casp-11, making them functionally Casp-1/Casp-11 double-knockouts
3–6
. Previous studies have shown that these mice are more susceptible to infections with
microbial pathogens
1
, including the bacterial pathogen Salmonella enterica serovar Typhimurium (S. Typhimurium)
7,8
, however the individual contributions of caspase-1 and caspase-11 to this phenotype
are not known. Here, we show that non-canonical caspase-11 activation contributes
to macrophage death during S. Typhimurium infection. TLR4/Trif-dependent IFN-β production
is crucial for caspase-11 activation in macrophages, but is only partially required
for pro-caspase-11 expression, consistent with the existence of an interferon-inducible
activator of caspase-11. Finally, Casp-1−/−
mice were significantly more susceptible to infection with S. Typhimurium than mice
lacking both proinflammatory caspases (Casp-1−/−/Casp-11−/−). This phenotype was accompanied
by higher bacterial counts, the formation of extracellular, bacterial micro-colonies
in the infected tissue and a defect in neutrophil-mediated clearance. These surprising
results indicate that caspase-11-dependent cell death is detrimental to the host in
the absence of caspase-1-mediated innate immunity, resulting in extracellular replication
of a facultative intracellular bacterial pathogen.
Previous studies have shown that logarithmic phase S. Typhimurium induce a rapid NLRC4-dependent
cell death in cultured macrophages that requires the SPI-1 Type 3 Secretion System
(T3SS)
9
. We previously reported that Salmonella grown to stationary phase (decreased SPI-1
expression) do not induce rapid activation of NLRC4, but establish themselves in an
intracellular niche
10
. Intracellular Salmonella are detected by inflammasome receptors NLRP3 and NLRC4
and mature IL-1β/IL-18 are released 12–17 hours post-infection (Supplementary Fig.
1 and 2a, b)
10
. In addition, intracellular Salmonella induce an uncharacterized form of lytic cell
death that is independent of the SPI-1 T3SS
11
. To investigate host factors involved in this type of Salmonella-induced macrophage
death, we infected macrophages deficient for specific inflammasome components with
stationary phase wild-type Salmonella. Macrophage death, which required infection
with live bacteria, did not require NLRP3 or the adaptor protein ASC, but was partially
dependent on NLRC4 (Supplementary Fig. 2a and 3). Since macrophages from Casp-1−/−/Casp-11−/−
mice did not release LDH, we investigated if caspase-1 and caspase-11 were activated
and processed in response to intracellular Salmonella (Supplementary Fig. 2a). Processed
caspase-1 p20 and caspase-11 p30 subunits were detected during Salmonella infection,
indicating that caspase-11 activation correlated with cell death (Supplementary Fig.
2a). Consistently, Casp-11−/−
macrophages were significantly more resistant to death than WT macrophages (Fig. 1a),
demonstrating an important role for caspase-11 in cell death caused by intracellular
Salmonella. In contrast to L. pneumophila infectons
12
, intracellular growth of Salmonella in WT and Casp-11−/−
macrophages was not significantly different (data not shown).
Macrophage death was not totally abrogated in Casp-11−/−
macrophages infected with wild-type Salmonella, indicative of an additional cell death
pathway (Fig. 1a and Supplementary Fig. 2c). Since NLRC4 contributed to macrophage
death (Fig. 1a and Supplementary Fig. 2a), we determined if NLRC4 activation accounted
for the remaining cell death seen in Casp-11−/−
macrophages. The second T3SS (SPI-2), which is used by intracellular Salmonella to
inject effector proteins into the host cell, can also inject flagellin
10
, a ligand for NLRC4
13–16
. Comparable levels of cell death were observed in Casp-1−/−/Casp-11−/−
and Casp-11−/−
macrophages infected with a ΔSPI-2 or a Δflagellin strain of Salmonella (Fig. 1a and
Supplementary Fig. 2c), suggesting that wild-type Salmonella induce cell death through
two separate pathways, one controlled by NLRC4 and the other requiring caspase-11.
Finally we compared the levels of cell death in macrophages derived from WT, Casp-1−
/
−
Casp-11
−
/
−
, Casp-11
−
/
−
or Casp-1
−
/
−
(Casp-1
−
/
−
/Casp-11tg) mice
3
infected with Salmonella. Cell death in Casp-1−/−
macrophages infected with a SPI-2-deficient strain was similar to levels seen in WT
macrophages (Supplementary Fig. 2d), indicating that Salmonella strains that cannot
inject flagellin exclusively engage caspase-11-dependent cell death. Consistently,
Casp-11−/−
macrophages infected with the ΔSPI-2 strain did not die (Supplementary Fig. 2d). In
contrast, wild-type Salmonella induced cell death via canonical (NLRC4/caspase-1)
and non-canonical signaling pathways (caspase-11) (Supplementary Fig. 1 and 2d).
NLRP3-mediated cytokine production triggered by non-canonical inflammasome stimuli
depends on both caspase-1 and caspase-11
3
. We therefore investigated if the NLRP3 pathway induced by intracellular Salmonella
required both caspases. Since IL-1β and IL-18 release in response to intracellular
Salmonella is exclusively mediated by NLRC4 and NLRP3 (Supplementary Fig. 2a, b),
we studied the response to SPI-2- or flagellin-deficient Salmonella
10
. Cytokine maturation in response to these strains required both caspase-1 and caspase-11
(Supplementary Fig. 4). In contrast, cytokine maturation induced by wild-type Salmonella,
which activates both NLRP3 and NLRC4, was only partially dependent on caspase-11,
but absolutely required caspase-1 (Supplementary Fig. 4). These results indicate that
intracellular Salmonella activate a non-canonical inflammasome similar to LPS/CTB
and the enteric bacteria C. rodentium, V. cholerae and E. coli
3
. Formation of ASC foci (specks), a measure of NLRP3/ASC complex formation, required
caspase-11 but not caspase-1 (Supplementary Fig. 5a, b), indicating that caspase-11
acts upstream of the NLRP3/ASC complex.
Stimulation of resting macrophages with LPS or IFN-g induces pro-caspase-11 expression
3,4,17
. To determine whether Salmonella-dependent activation of the non-canonical inflammasome
was dependent on TLR-mediated recognition, we infected Tlr4−/−
macrophages with flagellin-deficient Salmonella (which activate the caspase-11-dependent,
non-canonical inflammasome pathway exclusively; Supplementary Fig. 2 and 4). Caspase-11
activation, IL-1β secretion, and cell death depended on TLR4 (Fig. 1b). Caspase-11
processing and cell death required the TLR4-dependent signaling adaptor Trif, but
not the TLR4-dependent signaling adaptor MyD88 (Fig. 1c). In contrast, IL-1β maturation
was reduced in both MyD88−/−
and Trif−/−
, suggesting that cytokine maturation requires both adaptors. Expression of pro-IL-1β
required MyD88-signaling (Supplementary Fig 6a, b), explaining the lack of mature
IL-1β release in MyD88−/−
macrophages (Fig. 1c). Since cytokine production and cell death required Trif, we
measured pro-caspase-11 expression in MyD88−/−
, Trif−/−
, and MyD88−/−
/Trif−/−
macrophages infected with Salmonella. Although induction of pro-caspase-11 expression
was delayed in MyD88−/−
and Trif−/−
macrophages, the levels of pro-caspase-11 protein in MyD88−/−
/Trif−/−
macrophages were significant (Fig. 1d and Supplementary Fig. 6a). Thus, pro-caspase-11
protein expression is partially dependent on TLR-signaling. However, other pathways
likely contribute. Intriguingly, non-canonical inflammasomes were not activated in
Trif−/−
macrophages (Fig. 1c) even though significant amounts of pro-caspase-11 were present.
These results indicate that caspase-11 activity requires a Trif-dependent signal.
The adaptor Trif induces NFkB activation and signals through IRF3 to induce the expression
of type-I-interferons (type-I-IFNs)
18
. To investigate if type-I-IFNs could be the Trif-dependent signal required for caspase-11
activation, we compared the levels of IL-1β release, cell death and pro-caspase-11
expression in WT, Casp-1−/−/Casp-11−/−
and IRF3−/−
macrophages (Fig. 2a,b and Supplementary Fig. 6c). IRF3−
/
−
macrophages were significantly impaired in their ability to initiate caspase-11-dependent
IL-1β release and cell death even though significant levels of pro-caspase-11 were
present, albeit at slightly reduced levels when compared to WT macrophages. To confirm
the requirement of type-I-IFN signaling for caspase-11 activation, we measured non-canonical
inflammasome activation in macrophages lacking components of type-I and type-II interferon
signaling cascade. IFNαR−/−
or STAT-1−/−
macrophages infected with Salmonella did not process caspase-11 or induce non-canonical
cell death (Fig. 2c), and this was not due to a lack of pro-caspase-11 expression
(Fig. 2d). Macrophages lacking IFNγR were indistinguishable from WT thereby excluding
an involvement of IFN-γ. To confirm a dependency on type-I-IFN signaling for caspase-11
activation, macrophages infected with Δflag Salmonella were treated with recombinant
murine IFN-β (Fig. 2e). Consistent with an important role for type-I-IFN in caspase-11
activation, exogenous IFN-β restored cell death and caspase-11 processing in infected
MyD88−/−/Trif−/−
but not Casp-1−/−/Casp-11−/−
macrophages (Fig. 2e and Supplementary Fig. 7). Importantly, uninfected macrophages
treated with IFN-β did not induce LDH release, confirming that IFN-β alone cannot
induce non-canonical cell death in the absence of an infection. Our results reveal
a previously unreported requirement for type-I-IFN signaling in caspase-11 activation
that is consistent with a model in which an interferon inducible activator mediates
caspase-11 activation in response to intracellular Salmonella (Supplementary Fig.
1).
Finally, to extend our findings to an in vivo setting, we infected WT, Casp-1−/−/Casp−11−/−
, Casp-1−/− (Casp-1−/−/Casp-11tg) and Casp-11−/−
mice orally with wild-type Salmonella. As reported previously, Casp-1−/−/Casp-11−/−
mice had significantly higher bacterial loads in tissues compared to WT mice (Fig.
3a)
7,8
. Surprisingly, the levels of bacteria in Casp-1−/−
mice were significantly higher compared to Casp-1−/−/Casp-11−/−
mice (Fig. 3a). Interestingly, bacterial loads in Casp-11−/−
mice were comparable to WT mice in all organs examined. Unexpectedly, these results
imply that activation of proinflammatory caspase-11 is detrimental to the host in
the absence of caspase-1. Consistent with these findings, the bacterial loads in Nlrp3−/−/Nlrc4−/−
animals, which activate caspase-11, but not caspase-1 in response to Salmonella (Supplementary
Fig. 2), were significantly higher than in Casp-1−/−/Casp-11−/−
mice (Supplementary Fig. 8), essentially phenocopying Casp-1−/−
.
Although previous studies have shown that Casp-1−/−/Casp-11−/−
mice are more susceptible to many pathogens, the exact mechanism underlying these
findings is not well understood
1,19
. Host defense against systemic Salmonella infection requires neutrophils, since Salmonella
replication in the liver and spleen is exacerbated in neutropenic mice
20,21
. In addition, Salmonella becomes vulnerable to neutrophil-mediated clearance when
it transits between host cells
22
. We therefore investigated if caspase-1 or caspase-11-deficiency resulted in reduced
neutrophil influx into the spleen. Splenic neutrophil counts (Ly6G+ cells) were reduced
in mice lacking either caspase-1 or caspase-11 compared to WT mice, with Casp-1−/−/Casp-11−/−
mice having the most significant reduction of neutrophil counts compared to WT animals
(Fig. 3b). Since the levels of neutrophils in the mice deficient for caspase-1 or
-11 were very similar to each other, and did not correlate with the bacterial loads,
we concluded that differences in neutrophil levels alone could not account for the
higher bacterial levels in Casp-1−/−
animals (Fig. 3a). We next examined if caspase-1 deficiency resulted in any functional
defects in splenic neutrophils. Neutrophils have been previously implicated in phagocytosing
and killing extracellular Salmonella released by pyroptotic macrophages
22
. Interestingly, the percentage of neutrophils (Ly6G+) among all Salmonella-associated
cells (Salmo+) was significantly reduced in Casp-1−/−
and Casp-1−/−/Casp-11−/−
mice compared to WT and Casp-11−/−
mice (Fig. 3b). These data suggested that the lack of caspase-1 resulted in reduced
bacterial uptake by neutrophils or reduced association with Salmonella during infection.
However, since this reduction was observed in both Casp-1−/−
and Casp-1−/−/Casp-11−/−
mice, it did not explain the significantly higher CFUs in Casp-1−/−
mice (Fig. 3a).
NLRC4/caspase-1 induced lysis releases intracellular Salmonella, thus making them
accessible to neutrophil-mediated uptake and killing
22
. Since both caspase-1 and caspase-11 can induce lysis of infected cells, we speculated
that the lack of caspase-11 in Casp-1−/−/Casp-11−/−
mice might delay the egress of Salmonella from infected macrophages. Consistent with
this model, we found that Salmonella were present to a higher degree in macrophages
in Casp-1−/−/Casp-11−/−
mice compared to WT, Casp-11−/−
and Casp-1−/−
mice (Fig. 3b). Gram stain and specific anti-Salmonella antibody stainings of tissues
revealed that WT and Casp-11−/−
liver sections contained low levels of Salmonella, consistent with the CFU data (Supplementary
Fig. 9). Casp-1−/−/Casp-11−/−
liver sections contained high levels of bacteria within cells in the sinusoids (Supplementary
Fig. 9), which is consistent with our FACS data indicating that a larger proportion
of Salmonella are associated with macrophages in these mice (Fig. 3b). Casp-1−/−
liver sections contained mats of extracellular bacteria within typhoid nodules and
expanded sinusoids (Fig. 3c and Supplementary Fig. 9, 10a). Finally, mice infected
with Salmonella were injected with the membrane-impermeable antibiotic gentamicin
to distinguish intracellular bacteria from extracellular bacteria. Gentamicin treatment
significantly reduced bacterial counts in Casp-1−/−
mice, consistent with our histological finding that Salmonella is largely extracellular
in these mice (Supplementary Fig. 10b).
We conclude that caspase-1-deficiency results in reduced neutrophil-mediated clearance
of Salmonella released from infected macrophages, thus supporting extracellular growth
of this facultative intracellular pathogen. In keeping with this observation, it has
been reported that Salmonella rapidly replicates extracellularly in the liver of neutropenic
mice
23
. This phenotype is alleviated in Casp-1−/−/Casp-11−/−
animals, since bacterial egress from the infected macrophages is delayed. Thus, caspase-11-mediated
cell death results in detrimental effects to the host in the absence of caspase-1.
Our results indicate that the ability of neutrophils to phagocytose bacteria is dependent
on a caspase-1-mediated function. Since Casp-1−/−/Casp-11−/−
, IL-1R
−/− or IL-1β
−/−
/IL-18
−/− mice have similar levels of Salmonella
8
, the mechanism is not likely dependent on IL-1β/IL-18 maturation. Future studies
are required to identify and characterize this function. Finally, we provide evidence
that caspase-11-dependent cell death is exploited by Salmonella in the absence of
caspase-1 to cause disease in the host, highlighting the need to determine if caspase-11
activation has similar detrimental effects for the host in other infectious disease
models.
METHODS SUMMARY
Mice
Casp-1−/−/Casp-11−/−
(a.k.a caspase-1 knockout), Nlrp3−/−/Nlrc4−/−
, Casp-11−/−
and Casp-1−/−
(Casp-1−/−
/Casp-11tg
) mice were backcrossed to C57BL/6 for at least 10 generations
3,9,10
. All mouse studies were approved by the institutional animal care and use committees
of Genentech Inc. and Stanford University.
Animal infections
Mice (fasted for 12 h) were inoculated orogastrically with 2.4×107–1×108 WT or GFP+
WT S. Typhimurium SL1344. Tissues were harvested at day 4 post-infection, homogenized,
and dilutions were plated on LB agar containing 100 μg/ml Streptomycin. Bacterial
numbers are expressed as CFU/gram tissue.
Cell culture and infections
BMDMs were differentiated as described previously
13
. S. Typhimurium was grown to stationary phase overnight in LB at 37° C with aeration.
Cells were infected at an MOI of 100:1 and centrifuged for 15 minutes at 500 g to
ensure comparable adhesion of the bacteria to the cells. 100 μg/ml gentamicin was
added at 60 min post-infection. Cells were washed at 120 min post-infection followed
by addition of 10 μg/ml gentamicin for the remainder of the infection. Recombinant
murine IFN-β was added at 2 hours post-infection as indicated.
(Online only) METHODS
Bacterial strains
Bacterial strains include WT S. Typhimurium SL1344, GFP+ WT S. Typhimurium SL1344
(smo22) and the following S. Typhimurium mutants: ΔSPI-2 (ssaV::Kan) and Δflag (fljAB::Kan,
fliC::Cm).
Mice
Casp-1−/−/Casp-11−/−
(a.k.a caspase-1 knockout), Nlrp3−/−/Nlrc4−/−
, Casp-11−/−
and Casp-1−/−
(Casp-1−/−
/Casp-11tg
) mice were backcrossed to C57BL/6 for at least 10 generations
3,9,10
. All mouse studies were approved by the institutional animal care and use committees
of Genentech Inc. and Stanford University.
Animal infections
Mice (fasted for 12 h) were inoculated orogastrically with 2.4×107–1×108 WT or GFP+
WT S. Typhimurium SL1344. Tissues were harvested at day 4 post-infection, homogenized,
and dilutions were plated on LB agar containing 100 μg/ml Streptomycin. Bacterial
numbers are expressed as CFU/gram tissue. For in vivo gentamicin protection experiments,
mice were infected as above and injected intra-peritoneally with 1 mg of gentamicin
in 0.2 ml sterile PBS or PBS alone at 48 h, 24 h and 12 h before being euthanized.
Bacterial counts were analyzed as above.
Cell culture and infections
BMDMs were differentiated as described previously
13
. S. Typhimurium was grown to stationary phase overnight in LB at 37° C with aeration.
Cells were infected at an MOI of 100:1 and centrifuged for 15 minutes at 500 g to
ensure comparable adhesion of the bacteria to the cells. 100 μg/ml gentamicin was
added at 60 min post-infection. Cells were washed at 120 min post-infection followed
by addition of 10 μg/ml gentamicin for the remainder of the infection. Recombinant
murine IFN-β was added at 2 hours post-infection as indicated.
Histological analysis
Livers were harvested from infected mice at day 4 post-infection and immediately placed
in buffered formalin (R&D). Paraffin embedding, H&E staining and Gram-staining was
done by Histo-Tec Lab (Hayward, California). For immunoflurescence paraffin was removed
from paraffin-embedded tissue sections with Xylene and graded Ethanol baths. The tissue
was stained with rabbit anti-Salmonella antibodies and Phalloidin. Tissue was imaged
with a Zeiss LSM700 confocal microscope.
FACS analysis
Spleens from infected animals were aseptically removed and crushed between two glass
slides in RPMI containing 10 % (v/v) heat-inactivated FBS, 25 mM HEPES, 2 mM L-glutamine,
1 mM sodium pyruvate and 55 μM 2-mercaptoethanol. Single cell suspensions of spleens
were obtained by passages through 70 μm filters. Red blood cells were lysed in 175
mM ammonium chloride, 10 mM phosphate buffer, pH 7.0. 2×106 cells were stained per
sample. Dead cells were stained using a Live/Dead Fixable dead cell stain kit (Invitrogen).
Cells were washed in FACS buffer and rat anti-mouse CD16/CD32 (BD Biosciences) was
added to block FcIII/IIR prior to staining with analytical antibodies. Cells were
then stained for 30 min on ice with anti-Ly6-G (clone 1A8, BioLegend), anti-F4/80
(eBioscience) and anti-CD11b (BD Biosciences) antibodies. Stained cells were washed
twice prior to flow cytometric analysis. Data were collected on a LSR II (BD Biosciences)
at the Stanford University shared FACS facility, and the data were analyzed with FlowJo
software (Treestar).
Cell culture and infections
BMDMs were differentiated in DMEM (Invitrogen) with 10% vol/vol FCS (Thermo Fisher
Scientific), 10% MCSF (L929 cell supernatant), 10 mM HEPES (Invitrogen), and nonessential
amino acids (Invitrogen). BMDMs were seeded into 6-, 24-, or 96-well plates at a density
of 1.25 × 106, 2.5 × 105, or 5 × 104 per well. For all infections S. Typhimurium was
grown to stationary phase overnight in LB at 37° C with aeration and the BMDMs were
infected at an MOI of 100:1. The plates were centrifuged for 15 minutes at 500 g to
ensure comparable adhesion of the bacteria to the cells. 100 μg/ml gentamycin (Sigma-Aldrich)
was added at 60 min post-infection to kill extracellular bacteria in the cultures.
At 120 min post-infection, the cells were washed once with DMEM and given fresh macrophage
medium containing 10 μg/ml gentamycin for the remainder of the infection. Recombinant
murine IFN-β Sigma was added at 2 hours post-infection when necessary.
Immunofluorescence
BMDMs were seeded onto glass cover slips and infected as described above. Cover slips
were fixed with 4% Paraformaldehyde and stained with Rat anti-ASC (Genentech), Rabbit
anti-mouse Caspase-1 p10 (Santa Cruz Biotech) and Dapi. Cells were imaged with a Zeiss
LSM700 confocal microscope.
Cytokine and LDH release measurement
IL-1β was measured by ELISA (R&D systems). LDH was measured using CytoTox 96 (Non-Radioactive
Cytotoxicity Assay, Promega). To normalize for spontaneous lysis, the percentage of
LDH release was calculated as follows: ( LDH infected - LDH uninfected / LDH total
lysis - LDH uninfected )*100.
Western blotting
The caspase-1 p10 subunit, caspase-11 p30 and processed IL-1β released into the culture
supernatant were determined by Western blotting. Macrophages were washed with plain
pre-warmed DMEM lacking serum and phenol red at 6 hours post infection. The cells
were then cultured in this DMEM lacking serum and phenol red until 17 hours after
infection. The supernatant was collected and precipitated with 10% TCA (vol/vol) for
1 hour on ice. Precipitated proteins were pelleted at 20,000 g for 30 min at 4° C,
washed with ice-cold acetone, air-dried, resuspended in SDS-PAGE sample buffer, and
heated to 95° C for 10 min. Protein from 2.5×106 macrophages was loaded per well of
a 14% acrylamide gel. Western blots were performed with rat anti-mouse caspase-1 antibody
(4B4; Genentech) diluted 1:1,000, rat anti-mouse caspase-11 (17D9; Sigma) at 1:500,
rabbit anti-IL-18 (Biovision) at 1:500 and goat anti-mouse IL-1β antibody (AF-401-NA;
R&D Systems) diluted 1:500. Cell lysates were probed with anti-b-actin antibody (Sigma)
at 1:2,000.
Statistical analysis
Statistical data analysis was done using Prism 5.0a (GraphPad Software, Inc.). Statistical
significance was determined by the Mann-Whitney U test or Student’s t-test.
Supplementary Material
1
2