IAI Accepts, published online ahead of print on 6 October 2014 Infect. Immun. doi:10.1128/IAI.01340-13 Copyright © 2014, American Society for Microbiology. All Rights Reserved.
1
Mycobacterium tuberculosis strains lacking the surface lipid phthiocerol
2
dimycocerosate are susceptible to killing by an early innate host response.
3
Tracey A. Daya*, John E. Mittlerb, Molly R. Nixona, Cullen Thompsona, Maurine D.
4
Minera, Mark J. Hickeya, Reiling P. Liaoa, Jennifer Panga, Dmitry Shayakhmetovc,
5
David R. Shermana,#
6 7
Seattle Biomedical Research Institute, Seattle, Washington, USAa; Department of
8
Microbiology, University of Washington, Seattle, Washington, USAb; Department
9
of Pediatrics and Medicine, Emory University, Atlanta, GA, USAc
10 11
Running Head: MTB resistance to innate immunity requires PDIM
12 13
#
14
[email protected].
15
*Present address: HIV Vaccine Trials Network, Fred Hutchinson Cancer
16
Research Center, Seattle, Washington, USA
17 18
Address correspondence to David R. Sherman,
19
ABSTRACT
20
The innate immune response plays an important but unknown role in host
21
defense against Mycobacterium tuberculosis (MTB). To define the function of
22
innate immunity during tuberculosis, we evaluated MTB replication dynamics
23
during murine infection. Our data show that the early pulmonary innate immune
24
response limits MTB replication in a MyD88-dependent manner. Strikingly, we
25
found that little MTB death occurs during the first two weeks of infection. In
26
contrast, MTB deficient in the surface lipid phthiocerol dimycocerosate (PDIM)
27
exhibited significant death rates and consequently total bacterial numbers were
28
reduced. Host restriction of PDIM-deficient MTB was not alleviated by the
29
absence of interferon-γ, iNOS, or the phagocyte oxidase subunit p47. Taken
30
together, these data indicate that PDIM protects MTB from an early innate host
31
response that is independent of IFN-γ, reactive nitrogen intermediates, and
32
reactive oxygen species. By employing a pathogen replication tracking tool to
33
evaluate MTB replication and death during infection, we identify both host and
34
pathogen factors affecting the outcome of infection.
35 36
Keywords:
Mycobacterium
tuberculosis,
37
dimycocerosate, PDIM, MyD88, neutrophil
innate
immunity,
phthiocerol
38 39
INTRODUCTION
40
Phagocyte innate antimicrobial mechanisms are crucial in host defense
41
against intracellular pathogens. For Mycobacterium tuberculosis (MTB), and
42
many other pathogens, interferon gamma (IFN-γ) mediated activation of infected
43
macrophages is central in controlling bacterial growth. Indeed, IFN-γ is required
44
to induce the majority of known macrophage effector functions, including
45
generation of reactive nitrogen and oxygen intermediates, phagosome-lysosome
46
fusion, and autophagy (1). In the case of MTB, antigen specific CD4+ T cells are
47
the primary in vivo source of IFN- γ, but this adaptive immune response does not
48
begin to develop until after two weeks post infection (p.i.) (2). Whether innate
49
immunity contributes to MTB control prior to the arrival of antigen specific T cells
50
in the lung remains an outstanding question in the field.
51
Recent work bearing on this question revealed that innate responses
52
dependent on the central signaling molecule MyD88 are critical in defense
53
against MTB, as MyD88-/- mice were highly susceptible despite normal adaptive
54
immune responses (3). Moreover, human studies indicate that genetic variation
55
in the innate immune response influences human susceptibility to TB (4). These
56
observations suggest that innate immunity may play an important role in host
57
defense against mycobacteria; however the precise mechanisms remain
58
unknown.
59
To define the function of innate immunity in controlling MTB growth we
60
monitored MTB replication dynamics during murine infection. This previously
61
validated technique (5) exploits an unstable plasmid as a replication clock,
62
allowing bacterial replication and death rates to be determined from a simple
63
differential plating technique. In this study, we apply this approach to determine
64
how innate immunity impacts MTB replication dynamics in vivo and how MTB
65
resists these early host responses. We show that innate immunity serves to limit
66
MTB replication rate but does not effectively kill MTB during the first two weeks of
67
infection.
68
The unique lipid-rich mycobacterial cell envelope is thought to contribute
69
to MTB’s success as a pathogen and evidence points to both a passive role in
70
resisting harsh environments as well as an active role in modulating host immune
71
response (6). One critical cell wall component is the surface-associated lipid,
72
phthiocerol dimycocerosate (PDIM). PDIM has long been recognized as an
73
important virulence factor, as PDIM-deficient MTB exhibit a growth defect in
74
multiple animal models (7) (8). However, precisely how PDIM is involved in
75
virulence remains unknown. Here we demonstrate that PDIM-deficient MTB
76
strains are attenuated due to a susceptibility to host killing during the first two
77
weeks of infection.
78 79
METHODS
80
Bacterial strains and growth. M. tuberculosis H37Rv was obtained from
81
American Type Culture Collection (ATCC #25618). Following passage and single
82
colony selection, we noted that this strain, H37Rv:att::pBP10 (H37Rv:att),
83
exhibited decreased growth in mice. We then received a virulent stock of M.
84
tuberculosis H37Rv from Chris Sassetti, (University of Massachusetts Medical
85
School), referred to here as wild-type (WT) H37Rv. Eric Rubin (Harvard
86
University)
87
H37Rv:drrA.Tn1, described previously (9). The H37Rv.Tn.ppsE strain was
kindly
provided
the
defined
PDIM-deficient
MTB
strain,
88
generated in a transposon mutagenesis screen and the transposon position
89
within ppsE (Rv2935) was determined by sequencing (10).
90
Wild-type M. tuberculosis H37Rv and mutant strains were grown at 37oC
91
in Middlebrook 7H9 medium (Becton Dickinson) with 0.05% Tween-80 and
92
albumin, dextrose, catalase (Middlebrook ADC Enrichment, BBL Microbiology) or
93
on Middlebrook 7H10 (Becton Dickinson) medium with oleic acid plus albumin,
94
dextrose,
95
Hygromycin (50 μg/mL) and kanamycin (25 μg/mL) were added when
96
appropriate (i.e. hygromycin for strains harboring pBP8 and kanamycin for strains
97
harboring pBP10).
catalase
(Middlebrook
OADC
Enrichment,
BBL
Microbiology).
98
M. tuberculosis H37Rv and H37Rv:drrA.Tn1 were transformed with the
99
plasmid pBP8 (K.G. Papavinasasundaram, University of Massachusetts Medical
100
School) as previously described (5) and transformant colonies were selected on
101
7H10 medium containing 50 μg/ml hygromycin.
102 103
Aerosol infection of mice. C57BL/6, IFN-γ−/−, NOS2-/-, and CCR2 -/- mice were
104
obtained from the Jackson Laboratory (Bar Harbor, ME). MyD88-/- mice were
105
provided by Dr. Shizuo Akira, University of Osaka, Japan. p47phox-/- (Phox-/-) mice
106
were provided by Dr. Edward Miao, University of North Carolina at Chapel Hill.
107
We maintained 6–8-week-old mice in a biosafety level 3 animal facility in
108
accordance with Seattle Biomedical Research Institute Institutional Animal Care
109
and Use Committee protocols. We performed infections in an aerosol infection
110
chamber (Glas-Col) as previously described (11). At selected time points, mice
111
were sacrificed and the lungs aseptically removed and homogenized in PBS with
112
0.05% Tween-80. Lung homogenates were serially diluted and plated on 7H10 ±
113
30 μg/ml kanamycin or ± 50 μg/ml hygromycin. Differences in the means of
114
experimental groups were analyzed with the two tailed Student’s t-test using
115
GraphPad Prism software.
116 117
Lipid analysis. Purified PDIM-A was synthesized by Adriaan Minnaard as
118
described previously (12). For lipid extractions, late-log cultures were pelleted,
119
washed twice with ethanol, and dried overnight. Pellets were resuspended in
120
hexane, vortexed for 40 minutes, dried under nitrogen and resuspended in 500μl
121
dichloromethane.
122
chromatography using 200 micron aluminum-backed silica plates and petroleum
123
ether:ethyl acetate (v/v 50:1). Spots were visualized after staining with 8%
124
phosphoric acid, 3% cupric acetate solution and charring.
PDIM
was
resolved
by
one
dimensional
thin-layer
125 126
Neutrophil depletion. C57BL/6 mice were infected with 100 CFU of
127
H37Rv:drrA.Tn1 and were administered 300μg of purified Ly6G-specific (clone
128
1A8) antibody (BD Biosciences) or an isotype control (purified 2A3 (Rat IgG2a))
129
i.p. on day 9 p.i. as described previously (13, 14). Lungs were harvested on day
130
14 p.i. Half portions of lungs were plated to enumerate bacteria. The remaining
131
half lungs were analyzed to verify neutrophil depletion. Half lungs were
132
processed for single cell suspensions using Liberase Blendzyme 3 (Roche)
133
digestion of perfused lungs and percoll gradient centrifugation as described
134
previously (15).
Fc receptors were blocked with anti-CD16/32 (2.4G2). Cells
135
were suspended in PBS containing 0.1% NaN3 and 2.5% fetal bovine serum and
136
stained using antibodies specific for Ly6G (clone 1A8), Ly6C (clone RB6), and
137
F4/80 obtained from BD Biosciences. Samples were fixed in a 1%
138
paraformaldehyde solution in PBS overnight and analyzed using flow cytometry
139
(LSR II by BD) and FlowJo software (Tree Star, Inc.).
140 141
RESULTS
142
Innate immunity limits MTB replication rate. The early innate inflammatory
143
response could function to control MTB by either limiting growth rate or by
144
causing death. To evaluate the role of innate immunity in controlling MTB within
145
lungs early during infection, we employed the unstable plasmid pBP10 to monitor
146
MTB replication dynamics. This plasmid lacks several maintenance genes and as
147
a result it is steadily lost from a proportion of daughter cells as a consequence of
148
replication (16). Thus, the percentage of bacteria containing the plasmid serves
149
as a marker for the number of replications performed by the population. pBP10
150
has been used to track MTB replication in vivo in several recent publications (5,
151
17, 18).
152
Mice were infected with a low dose (100 CFU) of MTB strain H37Rv
153
harboring pBP10 (H37Rv::pBP10) via aerosol and at two weeks p.i. the total and
154
plasmid-containing bacteria in lungs were enumerated. Only innate immunity was
155
expected to operate during this time as MTB-specific T cells are not detected in
156
lungs until day 15 p.i. (2, 19). To verify that this represented the innate phase we
157
simultaneously analyzed MTB replication dynamics in Rag1-/- mice that lack
158
adaptive immunity. Plasmid frequencies averaged 70% on day 1 p.i. and dropped
159
to 12% by day 15 p.i. in both wild-type (C57BL/6, WT) and Rag1-/- mice,
160
indicating that the MTB population had undergone a similar number of
161
replications in each mouse strain (Fig. 1A). Total bacterial numbers reached
162
4.8x105 CFU/lung by day 15 in WT and Rag1-/- mice (Fig. 1B). These data
163
indicate that MTB behavior is not affected by the absence of adaptive immunity
164
during the first two weeks of infection; therefore, this period represents an
165
exclusively innate phase of the immune response.
166
To determine whether early innate responses resulted in any MTB death,
167
we used a mathematical model described previously (5) to derive MTB
168
replication and death rates from plasmid frequency and CFU data. Between day
169
1 and 15 in WT mice, the MTB lung population replication rate per day averaged
170
0.68, corresponding to a 27.5 hour generation time (Fig. 1C). Death rates
171
averaged 0.08, signifying very little MTB death during this time interval. Similar
172
results were obtained in Rag1-/- mice (Fig. 1C). Thus, minimal MTB death occurs
173
during the early innate phase of infection.
174
MyD88 is a central signaling molecule in multiple innate pathways, and
175
mice lacking MyD88 have severe defects in innate responses to numerous
176
pathogens including MTB (3, 20-22). To examine innate immune function during
177
MTB infection, we analyzed MTB replication dynamics in MyD88-/- mice. MTB
178
exhibited increased replication in the absence of MyD88 as indicated by
179
significantly reduced plasmid frequencies on day 15 p.i. (Fig. 1A). This increased
180
MTB replication was likely responsible for the 1 log higher bacterial loads in
181
these mice (Fig. 1B). To confirm this, we determined MTB replication and death
182
rates in MyD88-/- mice. Indeed MTB replication rates were increased to 0.88 in
183
MyD88-/- mice, corresponding to a 24.5 hour generation time. Death rates did not
184
increase significantly (p value >0.05; Fig. 1C). Thus, higher bacterial numbers in
185
MyD88-/- mice was primarily due to increased MTB replication rates. Taken
186
together, these results demonstrate that innate immunity serves to limit MTB
187
replication via a MyD88-dependent mechanism but does not effectively kill the
188
pathogen.
189 190
PDIM-deficient MTB are susceptible to innate killing. Our data indicate that
191
MTB replicates during the first two weeks of infection with few casualties. These
192
results differ from our earlier published studies, in which measurable bacterial
193
killing was detected (5). We noted however that the strain used in our earlier
194
studies was somewhat attenuated in mice, yielding approximately 1 log fewer
195
CFU in lungs on day 15 p.i. than expected for fully virulent H37Rv (Fig 2B,
196
H37Rv:att). Day 15 plasmid frequencies and replication rates for H37Rv:att in
197
WT mice were similar to fully virulent H37Rv (WT); however, death rates were
198
significantly increased for the attenuated strain (Fig. 2A, C). Therefore, the lower
199
bacterial levels achieved by H37Rv:att was due to increased death.
200
Loss of the virulence associated surface lipid PDIM has been shown to
201
occur spontaneously during in vitro manipulations of H37Rv strains resulting in
202
attenuation in vivo (23-25). We therefore tested whether a defect in PDIM
203
synthesis might explain attenuation of H37Rv:att. Lipid extracts were prepared
204
from this strain, from a fully virulent H37Rv strain, and from a strain disrupted in
205
the PDIM biosynthesis gene, ppsE (H37Rv:Tn:ppsE). Thin-layer chromatography
206
of the lipid extracts revealed that H37Rv:att lacked PDIM (Fig. 3). We therefore
207
hypothesized that increased death of H37Rv:att was due to loss of PDIM from
208
the bacterial cell surface. To test this hypothesis we obtained a mutant H37Rv
209
strain in which drrA, a gene involved in PDIM transport, was disrupted by
210
transposon insertion (9). As previously described, this strain synthesizes PDIM
211
but is unable to transport the lipid to the cell surface. We reasoned that use of a
212
transport mutant would allow us to examine the role of surface PDIM in MTB
213
virulence while minimizing potential effects on bacterial metabolism (23, 26).
214
Murrry et al. previously showed that drrA disruption results in attenuation
215
in mice infected intravenously (9). We infected mice with a low dose by aerosol
216
and enumerated bacterial loads in lungs over 90 days (Fig. 4A). The drrA mutant
217
strain was attenuated for growth early after infection. CFU were approximately 2
218
logs lower than WT MTB by day 14 p.i and remained at that level for 90 days. To
219
determine whether the drrA mutant had a growth defect immediately upon
220
infection we examined its levels prior to day 14 and found that bacterial burdens
221
were similar to WT MTB at 7 days p.i. but were significantly reduced by 14 days
222
p.i (Fig. 4B). We found that plasmid frequency data at day 7 p.i. for the drrA
223
strain varied from 17.8% to 53.2% and thus, rate data for this interval could not
224
be calculated reliably. Despite this variability plasmid frequencies for the drrA
225
strain were significantly lower than WT MTB at day 7. Thus the drrA mutant had
226
impaired growth over the first 2 weeks of infection. Nonetheless, the drrA mutant
227
strain persisted in mice at a reduced but stable level for 90 days.
228
Lower bacterial burdens for the PDIM-deficient drrA mutant could be due
229
to slower replication or increased death. To address this question, we again
230
employed the unstable plasmid replication tracking tool. Since the drrA mutant
231
strain contained a kanamycin resistance marker in the transposon, we used an
232
alternate version of the unstable plasmid, pBP8, in which the kanamycin
233
resistance selection marker is replaced by a hygromycin resistance marker. We
234
validated the use of pBP8 as a replication clock in the same way as pBP10 (5).
235
Briefly, MTB harboring pBP8 (H37Rv::pBP8) were maintained in log phase in
236
vitro culture for 21 generations by sub-culturing every 3-4 days without antibiotic.
237
Like pBP10, pBP8 plasmid loss was found to occur at a linear rate that was
238
proportional to growth rate (Fig. 5). The segregation constant (plasmid loss per
239
generation) for pBP8 was calculated as 0.11, which differs somewhat from that
240
determined for pBP10 (pBP10 s=0.18). C57BL/6 mice were infected with a low
241
dose of WT H37Rv::pBP8 and total and plasmid containing bacteria were
242
enumerated in lungs. Plasmid frequencies on day 14 p.i. were higher for
243
H37Rv::pBP8 than for H37Rv::pBP10, consistent with the lower segregation
244
constant determined for pBP8 (Fig. 1A and Fig. 4C). MTB replication and death
245
rates were calculated using the previously developed mathematical model (5)
246
and the segregation constant calculated for pBP8. Replication and death rates
247
for H37Rv::pBP8 during the first 2 weeks of infection were similar to those found
248
for H37Rv::pBP10 (Fig. 1C and Fig. 4D).
249
To determine whether PDIM-deficient MTB reached lower bacterial loads
250
because they were killed at an increased rate, mice were infected with a low
251
dose of the PDIM-deficient mutant MTB strain (H37Rv:drrA::pBP8) and results
252
were compared to those obtained with WT H37Rv::pBP8. Plasmid frequencies
253
on day 14 p.i. were similar for WT and drrA mutant MTB (Fig. 4C). In agreement,
254
replication rates calculated for drrA and WT MTB were also similar (Fig. 4D). In
255
contrast, death rates, which averaged 0.08 for WT MTB, averaged 0.60 for the
256
drrA mutant, a 6-fold increase. Thus, reduced bacterial numbers on day 15 p.i for
257
the drrA mutant strain were due to increased death. These results indicate that in
258
the absence of surface PDIM, MTB are killed more during the innate phase of
259
infection.
260 261
Absence of IFN-γ, RNI, or ROS does not alleviate attenuation of PDIM-
262
deficient MTB. Interferon gamma (IFN-γ) regulates numerous antimicrobial
263
mechanisms that are key in host defense against MTB. Among these, reactive
264
nitrogen intermediates (RNI) and reactive oxygen species (ROS) are potent
265
phagosomal effectors responsible for killing numerous pathogens. Accordingly,
266
we examined whether these effectors were involved in the innate phase killing
267
we observed for PDIM-deficient MTB. C57BL/6, IFN-γ−/−, and NOS2-/- mice were
268
infected via aerosol with WT or drrA mutant MTB harboring the unstable plasmid
269
pBP8 and on 14 days p.i. total and plasmid containing bacteria were enumerated
270
in lungs. WT bacterial loads were slightly increased in IFN-γ−/−, and NOS2-/- mice
271
compared to C57BL/6 mice, although this was not significant (Fig. 6A). When
272
infected with the drrA mutant, bacterial loads were identical in C57BL/6, IFN-γ−/−
273
and NOS2-/- mice. Plasmid frequencies were also similar for the drrA mutant in all
274
mice indicating similar bacterial replication dynamics in all groups (Fig. 6B).
275
Thus, absence of IFN-γ-/- and NOS2-/- dependent effector mechanisms did not
276
alleviate the susceptibility of PDIM-deficient MTB. These data support and
277
extend the work of Murry et al., who reported that drrA attenuation was not
278
relieved in the absence of IFN-γ or iNOS (9).
279
The phagocyte oxidase subunit, p47phox, is required for generation of ROS
280
including
superoxide
anion,
hydrogen
peroxide,
hydroxyl
radical,
and
281
hypochlorous acid, which participate in microbial killing within phagosomes (27).
282
To test whether ROS contribute to killing of PDIM-deficient MTB, we compared
283
replication dynamics of the drrA mutant in C57BL/6 and p47phox-/-mice (Phox-/-)
284
during the first two weeks of infection. We found that total bacterial loads and
285
plasmid frequencies were similar in both groups (Fig. 6 C and D). Taken
286
together, these results show that in vivo attenuation of PDIM-deficient MTB
287
occurs independently from IFN-γ, RNI, and ROS mediated effector mechanisms.
288 289
Neutrophil depletion does not affect growth of PDIM-deficient MTB.
290
Neutrophils are among the first cells detected in lungs following MTB infection,
291
arriving between day 7 and 12 p.i. and increasing in numbers through day 21 (13,
292
28). This early arrival coincides with the kinetics of H37Rv:drrA death and thus,
293
neutrophils could mediate killing of PDIM-deficient MTB. The monoclonal
294
antibody, Ly6G, clone 1A8 (1A8) has been shown to deplete neutrophils from
295
MTB infected mice through day 14 p.i. when given as a single dose of 300 μg on
296
day 9 (13, 14). We employed this same strategy to deplete neutrophils from mice
297
infected with H37Rv:drrA::pBP8 to determine whether this would alleviate
298
attenuation of the drrA mutant (Fig. 7A). Specific neutrophil depletion was verified
299
by flow cytometric analysis of lung cells on day 14 p.i. (Fig. 7B, C). An alternative
300
monoclonal antibody that binds to both monocytes and neutrophils, anti-Ly6C
301
clone RB6, was used in flow cytometric evaluations to verify that 1A8 treatment
302
did not reduce numbers of pulmonary monocytes (Fig. 7B, lower panel). Despite
303
having obtained a significant and lasting depletion of neutrophils, the relative
304
bacterial burdens in lungs for 1A8 and isotype control treated mice were similar
305
(Fig. 7C). Thus, neutrophil depletion did not impact susceptibility of PDIM-
306
deficient MTB.
307 308
Recruited monocytes are not required to mediate H37Rv:drrA attenuation.
309
Inflammatory monocytes are important in the early defense against some
310
pathogens, such as T. gondii (29, 30). In a recent publication, these cells were
311
shown to infiltrate the lung transiently during the first two weeks of MTB infection
312
(28). In addition, we observed that pulmonary monocytes were present both in
313
control and neutrophil depleted mice infected with H37Rv:drrA on day 14 p.i.
314
(Fig. 7B, C). We therefore hypothesized that killing of PDIM-deficient MTB was
315
mediated by these cells. Recruitment of inflammatory monocytes requires the
316
chemokine receptor, CCR2, and thus in mice lacking this gene (CCR2-/- mice)
317
inflammatory monocytes trafficking to the lung in response to MTB infection is
318
reduced (31). To test whether CCR2-dependent inflammatory monocyte
319
recruitment was required for H37Rv:drrA restriction, we compared the early
320
growth of H37Rv:drrA in lungs of WT and CCR2-/- mice. Our data show no
321
significant difference in bacterial numbers in lungs or in plasmid frequencies on
322
day 14 p.i. for WT and CCR2-/- mice (Fig. 8A, B). These results suggest that
323
recruitment of inflammatory monocytes into the lung is not required for host
324
restriction of H37Rv:drrA.
325 326
Restriction of PDIM-deficient MTB occurs independently of MyD88. Our
327
previous experiments showed that the early innate immune response to MTB
328
infection serves to limit MTB replication rate via a MyD88-dependent mechanism
329
(Fig. 1C). Since MyD88 has a critical role early during MTB infection, we
330
reasoned that it may also be involved in killing of PDIM-deficient MTB. Therefore
331
we compared the growth of H37Rv:drrA in WT and MyD88-/- mice during the first
332
two weeks of infection. Surprisingly, we observed similar bacterial numbers and
333
plasmid frequencies in WT and MyD88-/- mice (Fig. 8A, B). Thus, despite its key
334
role in the early innate response against MTB, susceptibility of H37Rv:drrA
335
occurs independently of MyD88.
336 337
DISCUSSION
338
MTB pulmonary infection induces a substantial influx of innate immune
339
cells including monocytes, neutrophils, dendritic cells, NK, and NKT cells (32-34).
340
However, it is unknown how these cells promote early host defense against MTB.
341
The early innate inflammatory response could function to control MTB by either
342
limiting growth rate and/or by causing death. Conventional plating for colony
343
forming units indicates increasing bacterial numbers during the first few weeks of
344
infection; however this method does not discern the relative contributions of
345
replication and death to total bacterial numbers at a given time. To determine the
346
rate of MTB killing early during murine infection we employed a replication
347
tracking tool, the unstable plasmid pBP10 that is steadily lost from a replicating
348
MTB population. Our results show that little MTB death occurs during the first two
349
weeks of infection. Thus, despite the documented presence of innate cells shown
350
capable of IFN-γ production and limiting MTB growth in vitro, WT MTB resists
351
killing by this early innate response in mice. In addition, we found that MTB
352
deficient in the complex surface lipid PDIM are susceptible to killing by an
353
unknown early host innate response.
354
During the first 4 weeks of infection, the observed bacterial death rates for
355
WT MTB were low suggesting that early host defense may play a limited role in
356
containment of WT MTB. However, the extreme susceptibility of MyD88-/- mice to
357
MTB infection despite normal adaptive immune responses revealed an important
358
role for innate immunity. Of the three known signaling pathways in which MyD88
359
is involved, disruption of the IL-1R pathway most closely recapitulates the
360
phenotype of MyD88-/- mice, whereas deficiencies in the toll-like receptor and
361
IFN-γR pathways do not (35-37) (unpublished data). To gain insight into the
362
critical role of innate immunity during MTB infection, we analyzed MTB replication
363
dynamics in MyD88-/- mice. We found that MTB numbers were 1 log higher in
364
MyD88-/- as early as 15 days p.i. By measuring MTB replication and death rates,
365
we found that higher CFU in MyD88-/- mice were due to increased replication
366
rates. Absence of MyD88 did not significantly affect death rates. Thus, MyD88-
367
dependent mechanisms are insufficient to kill WT MTB during the first two weeks
368
of infection; however they do slow MTB replication. Alternatively, MTB
369
pathogenesis may differ in MyD88-/- mice, resulting in conditions that enhance
370
MTB division in vivo. For example, more bacteria may be extracellular, as has
371
been shown for later time points (3).
372
In contrast to our current findings, we previously reported that MTB death
373
rates in C57BL/6 mice were highest during the first two weeks of infection (5).
374
Here we show that the strain used in those studies had lost production of PDIM
375
(H37Rv:att) and was attenuated for growth in vivo. Loss of virulence in H37Rv
376
strains is a commonly reported experience in research labs (23-25). Researchers
377
should be aware that during single colony selection there is a potential for
378
spontaneous PDIM-deficiency in the cloned strain, as we believe occurred during
379
plasmid transformant selection of our H37Rv strain.
380
PDIM is a well-known virulence factor in MTB but how it functions to
381
promote MTB growth in hosts is not clear. Previous studies showed that MTB
382
disrupted in PDIM transport to the cell surface were attenuated for growth in mice
383
(9, 38). Our data extend these studies showing in a low dose aerosol infection
384
model, that the drrA mutant exhibits an early pulmonary growth defect with 2 log
385
lower CFU by 14 days p.i. We show that PDIM-deficient MTB replicates at a
386
similar rate to WT MTB and that lower bacterial loads achieved by PDIM-deficient
387
MTB are due to increased death occurring between day 1 and 14 p.i. After 4
388
weeks of infection, we found that replication and death rates for WT MTB and
389
PDIM-deficient MTB were similar. This is in agreement with our previous study
390
reporting that stable CFUs during chronic murine infection are a result of slow
391
replication and death in balance (5). Thus, even though we later discovered that
392
the strain used in that study lacked PDIM, we can now confirm this holds true for
393
PDIM-containing WT MTB strains as well. Together, these data suggest that
394
surface PDIM protects MTB from an early innate host response.
395
The mechanism by which PDIM prevents MTB death remains unclear.
396
PDIM has been shown to be involved in macrophage receptor-mediated
397
phagocytosis of MTB and acidification of phagosomes (39), but MTB numbers
398
were not affected (40). Other studies have identified roles for PDIM in cell
399
permeability and resistance to RNI (38, 41), however growth in unactivated and
400
activated macrophages was similar for WT and PDIM-deficient MTB (9, 38). A
401
recent study concluded that PDIM on the bacterial surface masks underlying
402
pathogen-associated molecular patterns (PAMPs) that otherwise would signal
403
recruitment of microbicidal macrophages (42). In the mouse this recruitment
404
could occur during the first 2 weeks of infection, the time we find that PDIM-
405
deficient MTB are killed whereas WT MTB are not (Fig. 3).
406 407
Our observation that PDIM-deficient MTB are killed during the second
408
week of infection and then persist thereafter at stable levels suggests that the
409
restriction occurs via a specific and possibly transient host response, rather than
410
via a general cell wall defect. However, in agreement with one group (9), our data
411
show that absence of IFN-γ and RNI does not alleviate killing of PDIM-deficient
412
MTB (Fig. 4). In addition we show that drrA mutant killing does not require ROS
413
or MyD88 (Figs. 4 and 6). In this regard our data contrast with Cambier et al.
414
(42), who found that killing of PDIM-deficient bacteria is MyD88-dependent. This
415
difference may be due to differences in infection model and approach, for
416
example Cambier et al. primary examined host cell recruitment and bacterial
417
burdens in M. marinum-infected zebrafish larvae over the first 3 days of infection.
418
In contrast, we focused on bacterial replication dynamics in MTB-infected mice
419
over the first 2 weeks of infection. Alternatively, these differences may indicate
420
multiple roles for PDIM in protection from host-mediated killing. Indeed, as the
421
majority of known phagocyte anti-microbial functions rely on one of these effector
422
molecules, uncovering the role(s) of PDIM in promoting MTB survival early during
423
pulmonary infection may identify a previously unknown innate anti-bacterial
424
mechanism of general importance in host defense.
425
Having eliminated the known macrophage effector functions, we
426
hypothesized that death of PDIM-deficient MTB involved the polymorphonuclear
427
response, which has been shown to be elicited during this time (28, 43).That one
428
third of the pulmonary MTB population was found to reside within neutrophils at
429
day 14 p.i. lent credence to this possibility (44). Depletion of neutrophils,
430
however, did not affect replication dynamics of PDIM-deficient MTB (Fig. 5) . As
431
inflammatory monocytes were observed in the neutrophil depleted mice, we next
432
postulated that they might mediate drrA mutant killing, however this strain was
433
still attenuated in CCR2 deficient mice in which inflammatory monocyte
434
recruitment is blocked (Fig. 6).
435 436
In summary, none of the individual innate defense mechanisms examined
437
was found to be responsible for killing of the PDIM-deficient drrA MTB strain.
438
Further studies are needed to identify the mechanism responsible for the
439
increased death rates observed for PDIM-deficient MTB. A slight increase in CFU
440
was noted for drrA infected mice lacking phagocyte oxidase, neutrophils, and
441
inflammatory monocytes. Although, these were not statistically significant
442
differences, they may indicate that PDIM’s role is complex and could afford some
443
protection against multiple host effectors. Uncovering the means by which WT
444
MTB avoids killing early after infection will lead to a better understanding of MTB
445
persistence mechanisms and may identify targets for novel therapeutics.
446 447 448
Acknowledgements
449
This work has been supported by the National Institutes of Health (grant number
450
NIH HL094915-01) and the Paul G. Allen Family Foundation.
451 452
We thank Eric Rubin and Sarah Fortune of Harvard University for providing the
453
drrA mutant strain and Jeff Murry for helpful discussions. The plasmid pBP8 was
454
a gift from K.G. Papavinasasundaram, University of Massachusetts Medical
455
School. MyD88-/- mice were a gift from Shizuo Akira, Osaka University, Japan.
456
p47phox-/- (Phox-/-) mice were a gift from by Dr. Edward Miao, University of North
457
Carolina at Chapel Hill. We also thank Adriaan Minnaard, University of
458
Groningen, for providing synthetic PDIM-A, Chetan Seshadri, University of
459
Washington, for assistance with lipid analysis and Chetan Seshadri, Kevin Urdahl
460
and Lalita Ramakrishnan for critically reading the manuscript.
461 462
References
463
1. Flannagan, RS, Cosio, G, Grinstein, S. 2009. Antimicrobial mechanisms of
464
phagocytes and bacterial evasion strategies. Nat Rev Microbiol. 7:355-366.
465
2. Wolf, AJ, Desvignes, L, Linas, B, Banaiee, N, Tamura, T, Takatsu, K, Ernst, JD.
466
2008. Initiation of the adaptive immune response to Mycobacterium tuberculosis
467
depends on antigen production in the local lymph node, not the lungs. J Exp Med.
468
205:105-115.
469
3. Fremond, CM, Yeremeev, V, Nicolle, DM, Jacobs, M, Quesniaux, VF, Ryffel, B.
470
2004. Fatal Mycobacterium tuberculosis infection despite adaptive immune response in
471
the absence of MyD88. J Clin Invest. 114:1790-1799.
472
4. Berrington, WR, Hawn, TR. 2007. Mycobacterium tuberculosis, macrophages, and
473
the innate immune response: does common variation matter? Immunol Rev. 219:167-
474
186.
475
5. Gill, W, Harik, N, Whiddon, M, Liao, R, Mittler, J, Sherman, DR. 2009. A
476
Replication Clock for Mycobacterium tuberculosis. Nat Med. 15:211-214.
477
6. Daffe, M, Draper, P. 1998. The envelope layers of mycobacteria with reference to
478
their pathogenicity. Adv Microb Physiol. 39:131-203.
479
7. Goren, MB, Brokl, O, Schaefer, WB. 1974 Jan. Lipids of putative relevance to
480
virulence in Mycobacterium tuberculosis: correlation of virulence with elaboration of
481
sulfatides and strongly acidic lipids. 9:142-149.
482
8. Cox, JS, Chen, B, McNeil, M, Jacobs, WR. 1999. Complex lipid determine tissue
483
specific replication of Mycobacterium tuberculosis in mice. Nature. 402:79-83.
484
9. Murry, JP, Pandey, AK, Sassetti, CM, Rubin, EJ. 2009. Phthiocerol
485
dimycocerosate transport is required for resisting interferon-gamma-independent
486
immunity. J Infect Dis. 200:774-782.
487
10. Pang, JM, Layre, E, Sweet, L, Sherrid, A, Moody, DB, Ojha, A, Sherman, DR.
488
2012. The polyketide Pks1 contributes to biofilm formation in Mycobacterium
489
tuberculosis. J Bacteriol. 194:715-721.
490
11. Lewis, KN, Liao, R, Guinn, KM, Hickey, MJ, Smith, S, Behr, MA, Sherman, DR.
491
2003. Deletion of RD1 from Mycobacterium tuberculosis mimics bacille Calmette-Guerin
492
attenuation. J. Infect. Dis. 187:117-123.
493
12. Casas-Arce, E, ter Horst, B, Feringa, BL, Minnaard, AJ. 2008. Asymmetric total
494
synthesis of PDIM A: a virulence factor of Mycobacterium tuberculosis. Chemistry.
495
14:4157-4159.
496
13. Blomgran, R, Ernst, JD. 2011. Lung neutrophils facilitate activation of naive
497
antigen-specific CD4+ T cells during Mycobacterium tuberculosis infection. J Immunol.
498
186:7110-7119.
499
14. Daley, JM, Thomay, AA, Connolly, MD, Reichner, JS, Albina, JE. 2008. Use of
500
Ly6G-specific monoclonal antibody to deplete neutrophils in mice. J Leukoc Biol. 83:64-
501
70.
502
15. Urdahl, KB, Liggitt, D, Bevan, MJ. 2003. CD8+ T cells accumulate in the lungs of
503
Mycobacterium tuberculosis-infected Kb-/-Db-/- mice, but provide minimal protection. J
504
Immunol. 170:1987-1994.
505
16. Bachrach, G, Colston, MJ, Bercovier, H, Bar, N-D, Anderson, C,
506
Papavinasasundaram, KG. 2000. A new single-copy mycobacterial plasmid, pMF1,
507
from Mycobacterium fortuitum which is compatible with the pAL5000 replicon.
508
Microbiology. 2:297-303.
509
17. Adams, KN, Takaki, K, Connolly, LE, Wiedenhoft, H, Winglee, K, Humbert, O,
510
Edelstein, PH, Cosma, CL, Ramakrishnan, L. 2011. Drug tolerance in replicating
511
mycobacteria mediated by a macrophage-induced efflux mechanism. Cell. 145:39-53.
512
18. Chang, JC, Miner, MD, Pandey, AK, Gill, WP, Harik, NS, Sassetti, CM,
513
Sherman, DR. 2009. igr Genes and Mycobacterium tuberculosis cholesterol
514
metabolism. J Bacteriol. 191:5232-5239.
515
19. Chackerian, AA, Alt, JM, Perera, TV, Dascher, CC, Behar, SM. 2002.
516
Dissemination of Mycobacterium tuberculosis is influenced by host factors and precedes
517
the initiation of T-cell immunity. Infect Immun. 70:4501-4509.
518
20. Takeuchi, O, Hoshino, K, Akira, S. 2000. Cutting edge: TLR2-deficient and
519
MyD88-deficient mice are highly susceptible to Staphylococcus aureus infection. J
520
Immunol. 165:5392-5396.
521
21. Feng, CG, Scanga, CA, Collazo-Custodio, CM, Cheever, AW, Hieny, S, Caspar,
522
P, Sher, A. 2003. Mice lacking myeloid differentiation factor 88 display profound defects
523
in host resistance and immune responses to Mycobacterium avium infection not
524
exhibited by Toll-like receptor 2 (TLR2)- and TLR4-deficient animals. J Immunol.
525
171:4758-4764.
526
22. Henneke, P, Takeuchi, O, Malley, R, Lien, E, Ingalls, RR, Freeman, MW,
527
Mayadas, T, Nizet, V, Akira, S, Kasper, DL, Golenbock, DT. 2002. Cellular activation,
528
phagocytosis, and bactericidal activity against group B streptococcus involve parallel
529
myeloid differentiation factor 88-dependent and independent signaling pathways. J
530
Immunol. 169:3970-3977.
531
23. Domenech, P, Reed, MB. 2009. Rapid and spontaneous loss of phthiocerol
532
dimycocerosate (PDIM) from Mycobacterium tuberculosis grown in vitro: implications for
533
virulence studies. Microbiology. 155:3532-3543.
534
24. Andreu, N, Gibert, I. 2008. Cell population heterogeneity in Mycobacterium
535
tuberculosis H37Rv. Tuberculosis (Edinb). 88:553-559.
536
25. Ioerger, TR, Feng, Y, Ganesula, K, Chen, X, Dobos, KM, Fortune, S, Jacobs,
537
WR, Jr., Mizrahi, V, Parish, T, Rubin, E, Sassetti, C, Sacchettini, JC. 2010. Variation
538
among genome sequences of H37Rv strains of Mycobacterium tuberculosis from
539
multiple laboratories. J Bacteriol. 192:3645-3653.
540
26. Singh, A, Crossman, DK, Mai, D, Guidry, L, Voskuil, MI, Renfrow, MB, Steyn,
541
AJC. 2009. Mycobacterium tuberculosis WhiB3 maintains redox homeostasis by
542
regulating virulence lipid anabolism to modulate macrophage response. PLoS Pathog.
543
5:e1000545.
544
27. Sumimoto, H, Miyano, K, Takeya, R. 2005. Molecular composition and regulation
545
of the Nox family NAD(P)H oxidases. Biochem Biophys Res Commun. 338:677-686.
546
28. Kang, DD, Lin, Y, Moreno, JR, Randall, TD, Khader, SA. 2011. Profiling early
547
lung immune responses in the mouse model of tuberculosis. PLoS One. 6:e16161.
548
29. Mordue, DG, Sibley, LD. 2003. A novel population of Gr-1+-activated macrophages
549
induced during acute toxoplasmosis. J Leukoc Biol. 74:1015-1025.
550
30. Robben, PM, LaRegina, M, Kuziel, WA, Sibley, LD. 2005. Recruitment of Gr-1+
551
monocytes is essential for control of acute toxoplasmosis. J Exp Med. 201:1761-1769.
552
31. Peters, W, Scott, HM, Chambers, HF, Flynn, JL, Charo, IF, Ernst, JD. 2001.
553
Chemokine receptor 2 serves an early and essential role in resistance to Mycobacterium
554
tuberculosis. Proc Natl Acad Sci U S A. 98:7958-7963.
555
32. Tsai, MC, Chakravarty, S, Zhu, G, Xu, J, Tanaka, K, Koch, C, Tufariello, J,
556
Flynn, J, Chan, J. 2006. Characterization of the tuberculous granuloma in murine and
557
human lungs: cellular composition and relative tissue oxygen tension. Cell Microbiol.
558
8:218-232.
559
33. Chiba, A, Dascher, CC, Besra, GS, Brenner, MB. 2008. Rapid NKT cell responses
560
are self-terminating during the course of microbial infection. J Immunol. 181:2292-2302.
561
34. Feng, CG, Kaviratne, M, Rothfuchs, AG, Cheever, A, Hieny, S, Young, HA,
562
Wynn, TA, Sher, A. 2006. NK cell-derived IFN-gamma differentially regulates innate
563
resistance and neutrophil response in T cell-deficient hosts infected with Mycobacterium
564
tuberculosis. J Immunol. 177:7086-7093.
565
35. Reiling, N, Ehlers, S, Holscher, C. 2008. MyDths and un-TOLLed truths: sensor,
566
instructive and effector immunity to tuberculosis. Immunol Lett. 116:15-23.
567
36. Fremond, CM, Togbe, D, Doz, E, Rose, S, Vasseur, V, Maillet, I, Jacobs, M,
568
Ryffel, B, Quesniaux, VF. 2007. IL-1 receptor-mediated signal is an essential
569
component of MyD88-dependent innate response to Mycobacterium tuberculosis
570
infection. J Immunol. 179:1178-1189.
571
37. Mayer-Barber, KD, Barber, DL, Shenderov, K, White, SD, Wilson, MS, Cheever,
572
A, Kugler, D, Hieny, S, Caspar, P, Nunez, G, Schlueter, D, Flavell, RA, Sutterwala,
573
FS, Sher, A. 2010. Caspase-1 independent IL-1beta production is critical for host
574
resistance to mycobacterium tuberculosis and does not require TLR signaling in vivo. J
575
Immunol. 184:3326-3330.
576
38. Rousseau, C, Winter, N, Pivert, E, Bordat, Y, Neyrolles, O, Ave, P, Huerre, M,
577
Gicquel, B, Jackson, M. 2004. Production of phthiocerol dimycocerosates protects
578
Mycobacterium tuberculosis from the cidal activity of reactive nitrogen intermediates
579
produced by macrophages and modulates the early immune response to infection. Cell
580
Microbiol. 6:277-287.
581
39. Astarie-Dequeker, C, Le Guyader, L, Malaga, W, Seaphanh, FK, Chalut, C,
582
Lopez, A, Guilhot, C. 2009. Phthiocerol dimycocerosates of M. tuberculosis participate
583
in macrophage invasion by inducing changes in the organization of plasma membrane
584
lipids. PLoS Pathog. 5:e1000289.
585
40. Pethe, K, Swenson, DL, Alonso, S, Anderson, J, Wang, C, Russell, DG. 2004.
586
Isolation of Mycobacterium tuberculosis mutants defective in the arrest of phagosome
587
maturation. Proc Natl Acad Sci U S A. 101:13642-13647.
588
41. Camacho, LR, Constant, P, Raynaud, C, Laneelle, MA, Triccas, JA, Gicquel, B,
589
Daffe, M, Guilhot, C. 2001. Analysis of the phthiocerol dimycocerosate locus of
590
Mycobacterium tuberculosis. Evidence that this lipid is involved in the cell wall
591
permeability barrier. J Biol Chem. 276:19845-19854.
592
42. Cambier, CJ, Takaki, KK, Larson, RP, Hernandez, RE, Tobin, DM, Urdahl, KB,
593
Cosma, CL, Ramakrishnan, L. 2014. Mycobacteria manipulate macrophage
594
recruitment through coordinated use of membrane lipids. Nature. 505:218-222.
595
43. Tsai, MC, Chakravarty, S, Zhu, G, Xu, J, Tanaka, K, Koch, C, Tufariello, J,
596
Flynn, J, Chan, J. 2006. Characterization of the tuberculous granuloma in murine and
597
human lungs: cellular composition and relative tissue oxygen tension. Cell Microbiol.
598
8:218-232.
599
44. Wolf, AJ, Linas, B, Trevejo-Nunez, GJ, Kincaid, E, Tamura, T, Takatsu, K,
600
Ernst, JD. 2007. Mycobacterium tuberculosis infects dendritic cells with high frequency
601
and impairs their function in vivo. J Immunol. 179:2509-2519.
602 603
Figure 1: Innate immunity limits MTB replication rate.
604
Mice were infected with 100 CFU of MTB::pBP10 and replication dynamics were
605
determined. A. Percent of bacteria containing the plasmid on days 1 and 15 p.i.
606
B. CFU/lung after 15 days p.i. C. Replication and death rates between day 1 and
607
15 p.i. Means and standard deviations are plotted. Data from 2 -7 independent
608
experiments using 5 mice per group are shown. * indicates a P value < 0.01, and
609
“n.s.” not significant from C57BL/6 according to unpaired t-test with a p