IAI Accepts, published online ahead of print on 28 October 2013 Infect. Immun. doi:10.1128/IAI.00703-13 Copyright © 2013, American Society for Microbiology. All Rights Reserved.
1
Human lung tissue explants reveal novel interactions during
2
Legionella pneumophila infections
3
Running title: Legionella pneumophila in human lung tissue
4
Jens Jäger1*, Sebastian Marwitz2*, Jana Tiefenau1, Janine Rasch1, Olga Shevchuk1,3,
5
Christian Kugler4, Torsten Goldmann2#, Michael Steinert1#
6
1: Institut für Mikrobiologie, Technische Universität Braunschweig, Germany
7
2: Clinical and Experimental Pathology, Research Center Borstel, Airway Research
8
Center North (ARCN), Member of the German Center for Lung Research (DZL)
9
3: Department of Molecular Structural Biology, Cellular Proteomics, Helmholtz Centre for
10
Infection Research, Braunschweig, Germany
11
4: LungenClinic Großhansdorf, Airway Research Center North, Member of the German
12
Center for Lung Research (DZL)
13
* Equal contributions
14
#
Corresponding authors, e-mail:
[email protected],
[email protected]
15
1
16
Abstract
17
Histological and clinical investigations describe late stages of Legionnaires’ disease, but
18
cannot characterise early events of human infection. Cellular or rodent infection models
19
lack the complexity of tissue or have non-human backgrounds. Therefore we developed
20
and applied a novel model for Legionella pneumophila infection comprising living human
21
lung tissue. We stimulated lung explants with L. pneumophila strains and outer
22
membrane vesicles (OMVs) to analyse tissue damage, bacterial replication and
23
localization as well as the transcriptional response of infected tissue. Interestingly, we
24
found that extracellular adhesion of L. pneumophila to the entire alveolar lining precedes
25
bacterial invasion and replication in recruited macrophages. In contrast, OMVs
26
predominantly bound to alveolar macrophages. Specific damage to septa and epithelia
27
increased over 48 h and was stronger in wildtype-infected and OMV-treated samples
28
compared to infections with the replication-deficient type IVB secretion-deficient strain
29
DotA-. Transcriptome analysis from lung tissue explants revealed a differential regulation
30
of 2499 genes after infection. The transcriptional response included the upregulation of
31
uteroglobin and the downregulation of the macrophage receptor with collagenous
32
structure (MARCO). Immunohistochemistry confirmed the downregulation of MARCO at
33
sites of pathogen-induced tissue destruction. Both host factors have never been
34
described in the context of L. pneumophila infections. This work demonstrates that the
35
tissue explant model reproduces realistic features of Legionnaires’ disease and reveals
36
new functions for bacterial OMVs during infection. Our model allows us to characterise
37
early steps of human infection, which otherwise are not feasible for investigations.
2
38
Introduction
39
Histopathologically, Legionnaires’ disease, caused by the Gram-negative bacterium
40
Legionella pneumophila, is an acute fibrinopurulent pneumonia. Since the first
41
documented outbreak of Legionnaires’ disease in 1976, several autopsy series have
42
been published [1]. Samples from patients who died from L. pneumophila pneumonia
43
exhibit a massive infiltration of neutrophils and macrophages into the alveoli and
44
destruction of alveolar septa. Moreover, the alveolar epithelium shows sloughs and
45
inflammatory cells exhibit intense necrosis. L. pneumophila is mainly present in alveoli
46
and tends to cluster inside macrophages. In late infection stages, bacteria disseminate
47
to the patient’s spleen, kidneys, bone marrow and lymph nodes [1-4].
48
Different models have been established to analyse specific aspects of infection. Besides
49
human monocellular systems like macrophages and epithelial cells, protozoa such as
50
Acanthamoeba castellanii, Hartmannella vermiformis and Dictyostelium discoideum
51
were used to study the cellular and molecular pathogenicity of L. pneumophila [5-9].
52
These studies revealed that L. pneumophila primarily enters phagocytes and resides
53
within a unique membrane-bound compartment termed Legionella-containing vacuole
54
(LCV). The establishment of this replication niche requires the translocation of about 300
55
effector proteins into the host cell via a functional Dot/Icm type IV secretion [10-12].
56
Studying transcriptional responses of L. pneumophila-infected macrophages and
57
D. discoideum vegetative cells also shed light on the cellular mechanisms of
58
Legionnaires’ disease [13-16]. Moreover, proteomic approaches were shown to be
59
powerful tools to characterise both sides of the host-pathogen interaction [17-19].
60
Mammalian models such as guinea pigs, mice, rhesus monkeys and marmosets were 3
61
used to address immunological, pathological and pharmacological questions [20-22].
62
Despite providing enormous progress in the knowledge about mechanisms of
63
L. pneumophila infections, each of the current infection models has intrinsic limitations.
64
Cell culture assays lack the complex interaction networks between the specialised cell
65
types and extracellular components in the human lung. Guinea pig infections require
66
intraperitoneal or intratracheal inoculation techniques; and, owing to a different genetic
67
and immunologic background, the adequacy and transferability to humans can be
68
questioned.
69
Given the different model-immanent limitations, numerous intra- and extracellular
70
interactions of L. pneumophila factors with human lung tissue structures remain
71
unknown.
72
histopathology studies were performed post mortem. Even conspicuous subcellular
73
structures, like the abundant outer membrane vesicles (OMVs) shed by L. pneumophila,
74
have not yet been investigated in human lung tissue. OMVs contain high amounts of
75
degradative enzymes and other virulence-related proteins, which could execute
76
destructive and inflammatory activities [23, 24]. The spectrum of underexplored
77
interactions also includes pathogen-induced effects on the extracellular matrix (ECM),
78
transcriptomic and proteomic responses of the pathogen and the infected tissue, as well
79
as other yet unrecognised molecular processes, which only occur in the context of
80
human tissues.
81
In this study, we thoroughly analysed L. pneumophila-infected human lung tissue
82
explants (HLTEs) at multiple levels. We characterised the pathologic features of infected
83
tissues and determined the localization and growth kinetics of L. pneumophila wildtype
For
example,
early
infection
4
events
appear
underexplored,
since
84
and mutant strains in time course experiments with HLTEs. Moreover, we analysed the
85
contribution of OMVs to tissue destruction and demonstrated that the transcriptional
86
response of L. pneumophila-infected HLTEs differs from previous results in monocellular
87
models.
5
88
89
Experimental procedures
90
Ethics statement
91
For human lung tissue explants, written informed consent was obtained and all
92
procedures were performed according to German national guidelines and approved by
93
the Ethik-Kommission der Medizinischen Fakultät der Universität Lübeck (03/153).
94
Bacterial strains and culture
95
L. pneumophila Corby and a DotA-negative strain [25, 26], kindly provided by Antje
96
Flieger, Robert-Koch-Institut, Wernigerode, Germany) were cultivated in YEB (with 20
97
µg/mL kanamycin for the mutant) to the early stationary phase. For infection, the
98
bacterial suspension was diluted to 107 bacteria/mL in RPMI 1640 (Gibco, Darmstadt,
99
Germany) with 10 % FCS, 20 mM HEPES and 1 mM sodium pyruvate. OMVs were
100
isolated from early stationary L. pneumophila cultures as described [27] and diluted to
101
100 µg/mL (total protein) in RPMI with supplements.
102
Human lung tissue explants and assessment of bacterial replication
103
Tumour-free pulmonary tissue samples of approximately 1 cm³ were obtained from
104
surgery patients as described [28]. Samples were infected with the respective
105
L. pneumophila strain and incubated at 37 °C and 5 % CO2 for up to 48 h. Microscopic
106
inspection of untreated samples at different time points assured tissue vitality (see
107
below). 6
108
For CFU determination, triplicate samples from eight donors were infected. At indicated
109
time points, samples were weighed and homogenised in PBS. Dilutions were plated on
110
BCYE and incubated at 37 °C and 5 % CO2 for four days. Extracellular replication of L.
111
pneumophila in HLTEs was excluded by control experiments which showed that
112
acellular tissue homogenate or tissue supernatant do not support bacterial growth. The
113
CFU/g of tissue were determined; means and standard deviations of samples were
114
compared by the student’s t-test.
115
Tissue processing and histology analysis
116
Tissue samples were fixed with the HOPE technique (HEPES-glutamic acid Organic
117
solvent Protection Effect, DCS Diagnostics, Hamburg, Germany [29]). Briefly, samples
118
were incubated in HOPE solution I at 4 °C for 18 h and dehydrated in acetone at 4 °C for
119
6 h. After overnight incubation in paraffin at 54 °C, the samples were embedded in
120
paraffin and stored at 4 °C. Tissue blocks were cut on a microtome and deparaffinised
121
as described [30]. L. pneumophila was visualised by immunostaining with the
122
antibody 2D8 [31] (diluted 1:50), or an -MOMP antibody (diluted 1:100; kindly provided
123
by Joe Vogel, Washington University, St. Louis, USA). Human alveolar epithelial cells
124
were visualised by immunostaining of Aquaporin 5 (diluted 1:100, clone EPR3747,
125
Abcam, Cambridge, UK) and human alveolar macrophages by anti-CD68 (diluted 1:200,
126
DAKO Cytomation, Glostrup, DK). For detection, a HRP polymer system (Zytomed
127
Systems, Berlin, Germany) was applied according to the manufacturer’s instructions with
128
aminoethylcarbazole as the chromogen. Slides were counterstained with Mayer’s
129
haemalaun, dehydrated in a graded ethanol series and mounted with Pertex (Medite,
7
-Mip
130
Burgdorf, Germany). Images were taken with a Lumenera Infinity 4 digital ccd camera
131
on a Leica DMLB microscope.
132
Characterization of histological damage and statistical analysis
133
Using haematoxylin-eosin-stained slides, a qualitative tissue damage score was set up
134
based on three criteria: protein exudate in the alveoli, epithelial delamination and
135
alveolar septa destruction. Damage severity was graded as 0 (no damage), 1 (little
136
damage, distributed infrequently), 2 (damage, distributed frequently) to 3 (severe
137
damage as the dominating pattern), and added up. Statistical analyses were performed
138
with GraphPad Prism 5 (GraphPad Software, La Jolla, CA, USA). Medians and
139
interquartile ranges of damage scores were compared by the Mann-Whitney test with a
140
Bonferroni-corrected confidence interval of 98.3 % [32]. Affected cell types and
141
compartments were identified and validated by trained pathologists.
142
Transcriptome analysis
143
RNA was isolated from HOPE-fixed tissue samples as described [30]. RNA quality and
144
integrity were analysed with the Agilent RNA 6000 Nano Assay on a Bioanalyzer
145
(Agilent, Böblingen, Germany). Transcriptome analysis was conducted according to the
146
manufacturer’s instructions (Agilent One-Color Microarray-Based Gene Expression
147
Analysis, Low Input QuickAmp Labeling Kit, Version 6.6). 1650 ng Cy3-labeled DNA of
148
each sample were hybridised on one Agilent Human Gene Expression 4x44K
149
microarray. Tiff images of hybridised samples were obtained by scanning with an Agilent
150
SureScan microarray scanner and raw gene expression data were extracted using
8
151
Agilent Feature Extraction Software (v11.0.1.1). For hierarchical clustering, fold change
152
analysis and Gene Ontology term analysis Agilent GeneSpring software (v12.1) was
153
used. Quantile-normalised gene expression data were computed from raw data with
154
DirectArray software (OakLabs, Hennigsdorf, Germany) as described [33].
155
Quantitative Real-Time PCR (qRT-PCR)
156
To validate the transcriptional analysis, 450-650 ng of total RNA from L. pneumophila-
157
infected lungs and matched medium controls were isolated and reverse-transcribed into
158
cDNA (Maxima First Strand cDNA Synthesis Kit for qRT-PCR, Thermo Scientific,
159
Schwerte, Germany). DNAse I digest to remove genomic DNA was included during
160
synthesis; intron-spanning primers were designed with the Universal Probe Library
161
(Roche Applied Science, Mannheim, Germany) assay design center to target mRNA of
162
human uteroglobin (forward: ctcaccctggtcacactgg; reverse: ctgaaagctcgggcagat; Probe
163
no. 84) and RPL32 as the reference gene (forward: ccaccgtcccttctctctt; reverse:
164
gggcttcacaaggggtct; Probe no. 10) with NM_000994.3 and BC004481.2 as the input
165
sequences, respectively. Samples were initially denaturated at 95 °C for 5 min, followed
166
by 45 cycles of amplification (10 s 95 °C, 30 s 60 °C) on a LightCycler 480II (Roche
167
Applied Science) using the Light Cycler 2x Probes Master Mix (Roche Applied Science),
168
0.4 µM oligonucleotides and 0.2 µM FAM-labeled hydrolysis probes (Universal Probe
169
Library, Roche Applied Science) in a final volume of 10 µl. Negative controls without
170
cDNA templates were included. To omit differences in the amplification reaction, pooled
171
cDNA from the investigated samples was used to produce standard dilutions (5-fold)
172
and included in every run individually to determine the reaction efficiency. Data were
9
173
processed for advanced relative quantification within the LightCycler 480 software
174
version 1.5. and are shown as the normalized ratio adjusted for reaction efficiency.
175
Detection of uteroglobin in human HLTE supernatants
176
The supernatant of human HLTEs infected with L. pneumophila Corby or the DotA-
177
negative strain as well as matched medium controls were used for the ELISA
178
determination of secreted uteroglobin (Human Uteroglobin ELISA DuoSet, R&D
179
Systems, Wiesbaden, Germany).
180
Results
181
L. pneumophila causes tissue damage in infected human lung tissue explants
182
To characterise the histopathologic effects of L. pneumophila on HLTEs, infected and
183
non-infected tissue sections were fixed with the HOPE technique and stained with
184
haematoxylin-eosin (Fig. 1A-G). Generally, the predominant observed damage
185
phenotypes were protein exudate in the alveolar lumen (Fig. 1A *), delamination of
186
alveolar epithelial cells (Fig. 1G, arrow), and disintegrating connective tissue (Fig. 1A).
187
Moreover, dead macrophages, identified by nuclear breakup, can be observed in
188
infected tissue samples (Fig. 4E). The damage increased over the course of infection
189
(Fig. 1E-G).
190
No marked damage could be observed after 2 h of stimulation (Fig. 1B and E). After 24
191
h, the most abundant effect was the decreasing integrity of alveolar septa (arrow in Fig.
192
1F). At later time points, control samples were only slightly damaged, while
193
L. pneumophila-infected HLTEs exhibited delamination of epithelial cells from the 10
194
supporting connective tissue and shedding into the alveolar compartment (Fig. 1G,
195
arrow). Furthermore, the normally compact ECM, including the collagen backbone of
196
the remaining septa, appeared loose; even thicker connective tissue close to vessels or
197
bronchioles disaggregated (Fig. 2C). Protein exudate was observed as a slightly reddish
198
signal stained by H&E surrounding the alveolar septa (Fig. 1 *) and inside the alveolar
199
lumen. Protein exudate was markedly more abundant in L. pneumophila-infected
200
specimens than in controls, and was repeatedly found adjacent to damaged tissue
201
components such as affected septa (* in Fig. 1A). Importantly, L. pneumophila was
202
frequently detected close to damaged tissue structures (Fig. 1H).
203
Table 1 shows the distribution of damage scores of the individual phenotypes for each
204
infection condition. In all categories, damage increased in the first 48 h, most strongly for
205
wildtype-infected and OMV-stimulated samples (Fig. 2A-E). Incubation of tissue with
206
L. pneumophila or isolated OMVs led to comparable damage scores, shown by a
207
significant increase between 2 and 24 h compared to the control (Fig. 2A). Infection with
208
a type IV secretion-deficient L. pneumophila mutant (DotA-) did not lead to strong
209
epithelial delamination, and resulted in a total damage score comparable to uninfected
210
controls. After 24 h, the damage score for wildtype-infected samples was significantly
211
higher compared to that caused by DotA- (p = 0.0160; n = 7). Besides this difference
212
and the aforementioned increase in total damage scores compared to uninfected
213
controls, samples infected with wildtype or DotA- L. pneumophila or coincubated with
214
OMVs do not differ significantly from each other at a given time point, assuming a
215
confidence interval of 98.3 %. To compare the tissue damage depending on the
216
bacterial load, we challenged HLTEs of two different patients with 107, 108 and 109 11
217
bacteria/mL for 2h-48h with L. pneumophila wildtype and the DotA-negative mutant.
218
Higher concentrations (108 and 109 bacteria/mL) did not increase the tissue damage
219
compared to 107 bacteria/mL (Fig. 7). The observed tissue damage caused by the DotA-
220
negative strain was still less than that of the wildtype independent of the bacterial load
221
and incubation time.
222
L. pneumophila adheres to the alveolar lining and primarily infects alveolar
223
macrophages
224
In biopsies from Legionnaires’ disease patients, L. pneumophila is found predominantly
225
in alveolar macrophages [1]. In vitro data suggest that the pathogen also replicates
226
within alveolar epithelial cells [6]. In HTLEs, we detected L. pneumophila with anti-Mip
227
and anti-MOMP antibodies. Clear signals were observed in alveolar macrophages (red
228
colour); the staining intensities varied depending on the expected antigen abundance
229
(Fig. 3C, D). Interestingly, infected HLTEs displayed numerous L. pneumophila adhering
230
extracellularly to the alveolar surface (Fig. 4A, B). Where bacteria adhered to epithelial
231
cells (see Fig. 3G for alveolar epithelial cells type I, AECI) at septa and connective
232
tissue, tissue damage and epithelial delamination increased locally (Fig. 1H). We
233
confirmed that macrophages (see Fig. 3H for alveolar macrophages detected by anti-
234
CD68 staining) are the major host cell in the alveolar compartment, since virtually all
235
macrophages in affected alveoli were infected with L. pneumophila (Fig. 3 C, D), while
236
only a fraction of the epithelial cells had been invaded by the pathogen (* in Fig. 4A).
237
Uninfected HLTEs did not yield a signal for any of the two antibodies (Fig. 4C, E). These
238
observations demonstrate that the HLTE infection model produces representative
12
239
results and that it allows us to describe early events of disease progression in a 48 h
240
time frame.
241
242
L. pneumophila replicates within HLTEs
243
To assess the ability of HLTEs to support intracellular L. pneumophila replication, the
244
bacterial load of tissue samples was analysed during 48 h after infection (Fig. 5).
245
Colonies were visible on BCYE agar four days after plating. Wildtype L. pneumophila
246
multiplied by approximately 10-fold in 24 h, similarly to previous studies in human
247
macrophage-like cells [34]. After 24 h, the bacterial load continued to increase at a lower
248
rate. The amount of DotA-negative bacteria, which cannot replicate within host cells, did
249
not increase significantly. These results revealed that HLTEs support L. pneumophila
250
replication and suggest this infection model for the characterization of mutants on the
251
tissue level.
252
253
L. pneumophila OMVs are located in alveolar macrophages
254
L. pneumophila OMVs contain many virulence-related proteins including degradative
255
enzymes and associate with alveolar epithelial cells [23, 35]. Whether OMVs execute
256
destructive activities and how they contribute to the infection on the tissue level is
257
unknown.
13
258
Our results reveal a distinct localization of L. pneumophila OMVs in human lung tissue
259
for the first time. Immunostaining showed that purified OMVs bind predominantly to the
260
surface of alveolar macrophages and can be detected in their cytoplasm (Fig. 3E, F).
261
Stimulating HLTEs with OMVs resulted in distinct tissue damage with epithelial cell
262
delamination in affected alveoli and damage to collagen structures in septa and
263
connective tissue fibres, starting approximately 24 h after infection. This damage is as
264
severe as the effect observed in L. pneumophila-infected samples (Fig. 2A, D; Table 1)
265
and shares a comparable histological damage pattern.
266
267
Transcriptional response of HLTEs to L. pneumophila infection
268
Previous transcriptional analyses identified host responses to L. pneumophila infection
269
[14, 15, 36]. To re-address this question on the tissue level, the transcriptome of
270
L. pneumophila-infected HLTEs was compared to that of non-infected tissue from the
271
same donors. 2499 genes were regulated with a fold change ≥ 2.0 (Table 2) and were
272
clustered hierarchically on entities (similarity of gene expression) and conditions
273
(uninfected vs. infected). The data from two independent experiments showed that
274
distinct response levels could be observed (Fig. 6A). Gene ontology analysis of the
275
highly regulated genes revealed an enrichment of eight different terms among the 2499
276
genes (Table 3).
277
Among the highly upregulated genes, we found uteroglobin, a protein secreted by Clara
278
cells (2.18 log2-fold change; Table 4). Quantitative RT-PCR with material from 8
14
279
experiments confirmed the significant regulation of uteroglobin by L. pneumophila wt
280
(Fig. 6F). Targeting uteroglobin on the protein level via immunostaining did not reveal a
281
marked difference between infected and uninfected samples. Alveolar macrophages
282
were predominantly positive for uteroglobin (Fig. 6B, C). Since it is a secreted molecule,
283
we further quantified by ELISA the amount of uteroglobin in HTLE supernatants.
284
Interestingly, less uteroglobin was found in the supernatants of L. pneumophila wt and
285
DotA- infected tissues, compared to medium controls (Fig.6G).
286
Similarly, we targeted the macrophage receptor with collagenous structure (MARCO), a
287
class A scavenger receptor. MARCO is strongly expressed in uninfected tissue, and is
288
1.96
289
Immunostaining verified this finding, revealing a strong MARCO signal on alveolar
290
macrophages in non-infected HLTEs (Fig. 6D) and a reduced expression on
291
macrophages at sites of tissue destruction (Fig. 6E).
log2-fold
down-regulated
after
infection
292
15
with
L. pneumophila
(Table
4).
293
Discussion
294
The histopathologic descriptions of Legionnaires’ disease are consistent, but restricted
295
to the final stage of disease [1]. Our understanding of L. pneumophila pneumonias is
296
also limited by the poor amenability of infected human tissue. To overcome these
297
restraints, powerful infection models ranging from monocellular host systems to
298
mammals were developed. However, these models lack the communication between
299
different cell types or have a non-human background, respectively. Data from animal
300
models in particular cannot be generalised per se due to important differences in the
301
expression, localization and function of signaling molecules and receptors. Thus, not
302
surprisingly, the function of L. pneumophila virulence factors varies considerably
303
between host systems [37, 38].
304
In the present study, we established a novel L. pneumophila infection model involving
305
human lung tissue explants (HTLEs). Particular aspects of infections with the pathogens
306
Chlamydophila pneumoniae, Streptococcus pneumoniae and Haemophilus influenzae
307
were analysed in similar systems [28, 39, 40]. Although certain characteristics of HLTEs
308
may depend on clinical parameters, the patients’ medical conditions and donor diversity,
309
we obtained statistically robust, reproducible results. Thus, we utilised HTLEs with their
310
multitude of cell types and extracellular components to investigate interactions between
311
L. pneumophila and its human host at a unique level of complexity. The infection route is
312
comparable to that in the human body, with the pathogen entering the alveoli via the
313
bronchioli, albeit in a liquid phase rather than in aerosol droplets. The bacteria can reach
314
the host cells and tissue structures from all sides, similarly as in the setup of a cell
315
culture infection experiment. 16
316
Despite numerous sophisticated infection studies with L. pneumophila, the initial
317
infection processes in the human lung remain unknown. Cell culture models suggest
318
that alveolar macrophages are the most relevant cells for intracellular replication, while
319
epithelial cells are infected to a minor degree. However, it is not clear whether these
320
cells are relevant for the initial contact, or if other cells or extracellular structures are also
321
crucial for infection establishment. In the present study with HLTEs, we found that large
322
numbers of L. pneumophila adhere extracellularly to the entire alveolar surface. With
323
increasing incubation periods the bacteria were detected primarily on and within alveolar
324
macrophages, which are recruited to the alveolar space. Moreover, L. pneumophila
325
could be detected in the connective tissue. Taken together, these results indicate that
326
L. pneumophila initially binds to extracellular, yet unidentified, alveolar tissue
327
components and epithelial cells. The binding and invasion of alveolar macrophages
328
recruited to the alveoli obviously represents a consecutive step of infection.
329
Histopathological analyses revealed that L. pneumophila-infected HLTEs are consistent
330
with well-known features of Legionnaires’ disease [41]. This is not self-evident, since
331
HLTEs do not include all immune system components which normally circulate through
332
blood vessels and enter the tissue at infection sites. Infected HLTEs were characterised
333
by specific damage to the tissue architecture. Alveolar septa were disrupted, alveolar
334
epithelia appeared to be shaved off the underlying basal lamina and protein exudate
335
was detected. L. pneumophila colocalised with the damage to tissue structures at
336
alveolar septa. This indicates that L. pneumophila causes the degradation of tissue
337
barriers, possibly by destructive enzymes present on the surface or in the secretome of
338
the bacterium (in the soluble fraction or in association with outer membrane vesicles) 17
339
[23]. This abolition of alveolar integrity, which could additionally be caused by the
340
induction of tissue-destructive host molecules, likely contributes to the dissemination of
341
L. pneumophila to neighbouring alveoli and other organs [2, 3].
342
Interestingly, the tissue destruction in wildtype-infected HLTEs was markedly stronger
343
compared to HLTEs infected with the DotA-negative strain. This is probably not due to a
344
higher amounts of bacteria and bacterial enzymes at later time points in wildtype-
345
infected samples, as the DotA- L. pneumophila strain fails to cause higher tissue
346
damage than the wildtype strain even if applied at 100-fold higher numbers. However, it
347
is conceivable that pulmonary cells secrete degradative enzymes or activate cell death
348
pathways like pyroptosis in response to wildtype, but not DotA- L. pneumophila [42]. The
349
finding that the transcriptional response to these two strains is different in a human
350
macrophage-like cell line supports this hypothesis [13].
351
To provide evidence that the HLTE model supports L. pneumophila replication and can
352
identify attenuated mutants, we determined the growth kinetics of wildtype and DotA-
353
deficient L. pneumophila in tissue samples. Importantly, in accordance with previous
354
results in cell culture models [13, 43], we observed that wildtype L. pneumophila
355
replicates within tissue samples, while DotA- bacteria, which are unable to multiply
356
intracellularly, did not. Since the DotA-negative strain does not show growth defects in
357
liquid media, the observed bacterial replication takes place within infected host cells,
358
and not extracellularly, where there would be no growth disadvantage. Furthermore, we
359
can conclude that the observed difference between the wildtype strain and the DotA-
360
mutant can be explained by the intracellular growth of the wildtype. Even if applied at
361
100-fold higher numbers, the DotA- bacteria fail to cause a remarkable higher tissue 18
362
damage than the wt strain. These observations show that L. pneumophila-infected
363
HLTEs yield reliable, robust results on bacterial replication which strongly correlate with
364
current cell culture models. Moreover, these results pave the way for the
365
characterization of L. pneumophila mutants under the complex conditions in the human
366
lung.
367
The successful establishment of the HLTE infection model encouraged us to discover
368
novel host-pathogen interactions in complex human tissue. Previously, L. pneumophila
369
OMVs were shown to contain many virulence-related proteins including proteases and
370
lipases. Interestingly, OMVs do not kill host cells, but specifically modulate the cytokine
371
response of alveolar epithelial cells [23]. To study the involvement of OMVs in infections
372
of the human lung, we analysed the localization and putative degradative effects of
373
OMVs in HLTEs. We detected L. pneumophila and its OMVs at similar sites in the
374
tissue. Since OMVs contain proteins involved in adhesion to host cells, such as MOMP
375
and Hsp60 [23], our results indicate that the presence of these proteins is sufficient for
376
the localization of these subcellular bacterial structures. Intriguingly, OMVs also caused
377
histological tissue damage which was qualitatively and quantitatively similar to the
378
destruction caused by L. pneumophila itself. The aforementioned degradative enzymes
379
are likely responsible for this effect in OMV-treated and also in L. pneumophila-infected
380
HLTEs. Following this notion, we propose that OMVs contribute to the extracellular
381
pathogenicity of L. pneumophila, including the dissemination of the infection to other
382
organs.
383
During acute L. pneumophila infections, dynamic and interrelated responses are
384
triggered in the human lung at multiple levels. To describe such processes at a 19
385
complexity above monocellular cultures, we recorded the transcriptional response of
386
L. pneumophila-infected HLTEs. This approach revealed several new host factors which
387
may be involved in the pathogenesis of Legionnaires’ disease. We found 2499 genes to
388
be differentially expressed after infection, many of which have never been described to
389
be related to L. pneumophila infections. A group of eight Gene Ontology terms is
390
significantly regulated (p < 0.05) after infection, including a large number of extracellular
391
proteins, components of the immune response and, interestingly, lipoprotein transport
392
proteins. We will characterise these data in the future to understand better how the
393
pathogen establishes disease and how it modulates the host response.
394
Our analyses by microarray and validation by quantitative PCR revealed that uteroglobin
395
mRNA was significantly upregulated 24 h after infection. Uteroglobin, also termed Clara
396
cell secretory protein or blastokinin, is a major constituent of the airway extracellular fluid
397
[44]. It inhibits immune cell recruitment both in vitro and after infection of mice with
398
Pseudomonas aeruginosa [45, 46]. Importantly, uteroglobin mRNA and protein levels
399
decrease in lung epithelial cells after stimulation with P. aeruginosa or TNF- [46, 47]. It
400
is conceivable that L. pneumophila follows a strategy similar to that of P. aeruginosa in
401
regard to immune cell recruitment. Interestingly, we observed a reduced concentration
402
of uteroglobin in the supernatants of HLTEs challenged with wildtype and DotA- L.
403
pneumophila. This might be explained by an increased uptake of uteroglobin by alveolar
404
macrophages upon stimulation, since these are shown as the main positive cells by
405
immunostaining.
406
The downregulation of MARCO after an L. pneumophila infection is striking. MARCO is
407
class A scavenger receptor involved in the uptake of the bacterial pathogens Neisseria 20
408
meningitidis, Clostridium sordellii and Streptococcus mutans by macrophages and
409
modulates cytokine responses [48-53]. Intriguingly, S. mutans phagocytosis is partially
410
mediated by MARCO, and the pathogen suppresses this function with a peptidyl-prolyl
411
cis-trans isomerase [50]. L. pneumophila features Mip, a virulence factor with the same
412
enzymatic activity, on its surface and in association with OMVs [23, 54]. Since the
413
collagen-binding protein Mip is also involved in the extracellular pathogenicity of
414
L. pneumophila, it will be interesting to investigate if Mip modulates MARCO activity and
415
thereby contributes to bacterial dissemination within the lung.
416
In summary, the HLTE infection model narrows the gap between current infection
417
models and actual human infections. It allows us to characterise tissue damage,
418
bacterial dissemination and the host’s molecular response after an infection with
419
L. pneumophila in great detail and will contribute to our understanding of the pathogen-
420
host crosstalk and to the development of interventional treatment strategies.
421
21
422 423
Acknowledgments
424
This study was supported in part by a grant from the Deutsche Forschungsgemeinschaft
425
(DFG) STE 838/6-1, by a grant from the German Bundesministerium für Bildung und
426
Forschung (BMBF) LegioProTect 0315831, and by the Helmholtz International Graduate
427
School for Infection Research. The authors would like to thank Antje Flieger (Robert
428
Koch Institute, Wernigerode, Germany) for providing a mutant strain, Joe Vogel
429
(Washington University, St. Loius, USA) for providing an antibody, Andra Schromm and
430
Ulrich Schaible for providing machinery and lab space, as well as Steffi Fox, Maria
431
Lammers and Jasmin Tiebach for their excellent technical assistance.
432
Author contributions
433
JJ, SM, TG and MS conceived the experiments. JJ, SM, JT, JR and CK carried out the
434
experiments. JJ, SM, OS, TG and MS analysed data. JJ, SM, TG and MS wrote the
435
manuscript.
436
The authors declare that no conflict of interest exists.
437
438
22
439
References
440
441
1.
pneumonias. A review of 74 cases and the literature. Hum. Pathol. 12:401-422.
442 443
Winn WC, Jr., Myerowitz RL. 1981. The pathology of the Legionella
2.
Hambleton P, Baskerville A, Fitzgeorge RB, Bailey NE. 1982. Pathological and
444
biochemical features of Legionella pneumophila infection in guinea-pigs. J. Med.
445
Microbiol. 15:317-326.
446
3.
Watts JC, Hicklin MD, Thomason BM, Callaway CS, Levine AJ. 1980. Fatal
447
pneumonia caused by Legionella pneumophila, serogroup 3: demonstration of
448
the bacilli in extrathoracic organs. Ann. Intern. Med. 92:186-188.
449
4.
Theaker JM, Tobin JO, Jones SE, Kirkpatrick P, Vina MI, Fleming KA. 1987.
450
Immunohistological detection of Legionella pneumophila in lung sections. J. Clin.
451
Pathol. 40:143-146.
452
5.
Pearlman E, Jiwa AH, Engleberg NC, Eisenstein BI. 1988. Growth of
453
Legionella pneumophila in a human macrophage-like (U937) cell line. Microb.
454
Pathog. 5:87-95.
455
6.
Mody CH, Paine R 3rd, Shahrabadi MS, Simon RH, Pearlman E, Eisenstein
456
BI, Toews GB. 1993. Legionella pneumophila replicates within rat alveolar
457
epithelial cells. The J. Infect. Dis. 167:1138-1145.
458
7.
Hagele S, Kohler R, Merkert H, Schleicher M, Hacker J, Steinert M. 2000.
459
Dictyostelium discoideum: a new host model system for intracellular pathogens of
460
the genus Legionella. Cell. Microbiol. 2:165-171.
23
461
8.
pneumophila for freshwater and soil amoebae. J. Clin. Pathol. 33:1179-1183.
462 463
Rowbotham TJ. 1980. Preliminary report on the pathogenicity of Legionella
9.
Fields BS, Sanden GN, Barbaree JM, Morrill WE, Wadowsky RM, White E,
464
Feeley JC. 1989. Intracellular multiplication of Legionella pneumophila in
465
amoebae isolated from hospital hot water tanks. Curr. Microbiol. 18:131-137.
466
10.
Bartfeld S, Engels C, Bauer B, Aurass P, Flieger A, Bruggemann H, Meyer 2009. Temporal resolution of two-tracked NF-kappaB activation by
467
TF.
468
Legionella pneumophila. Cell. Microbiol. 11:1638-1651.
469
11.
pneumophila type IV effectors. Annu. Rev. Cell. Dev. Biol. 26:261-283.
470 471
Hubber A, Roy CR. 2010. Modulation of host cell function by Legionella
12.
Gomez-Valero L, Rusniok C, Jarraud S, Vacherie B, Rouy Z, Barbe V,
472
Medigue C, Etienne J, Buchrieser C. 2011. Extensive recombination events
473
and horizontal gene transfer shaped the Legionella pneumophila genomes. BMC
474
Genomics 12:536.
475
13.
Losick VP, Isberg RR. 2006. NF-kappaB translocation prevents host cell death
476
after low-dose challenge by Legionella pneumophila. J. Exp. Med. 203:2177-
477
2189.
478
14.
Fortier A, Faucher SP, Diallo K, Gros P. 2011. Global cellular changes induced
479
by Legionella pneumophila infection of bone marrow-derived macrophages.
480
Immunobiology. 216:1274-1285.
481
15.
Farbrother P, Wagner C, Na J, Tunggal B, Morio T, Urushihara H, Tanaka Y,
482
Schleicher M, Steinert M, Eichinger L. 2006. Dictyostelium transcriptional host
483
cell response upon infection with Legionella. Cell. Microbiol. 8:438-456.
24
484
16.
Mycobacterium and other pathogens. Semin. Cell. Dev. Biol. 22:70-76.
485 486
Steinert M. 2011. Pathogen-host interactions in Dictyostelium, Legionella,
17.
Shevchuk O, Batzilla C, Hagele S, Kusch H, Engelmann S, Hecker M, Haas
487
A, Heuner K, Glockner G, Steinert M. 2009. Proteomic analysis of Legionella-
488
containing phagosomes isolated from Dictyostelium. Int. J. Med. Microbiol.
489
299:489-508.
490
18.
Urwyler S, Nyfeler Y, Ragaz C, Lee H, Mueller LN, Aebersold R, Hilbi H. 2009.
491
Proteome analysis of Legionella vacuoles purified by magnetic immunoseparation
492
reveals secretory and endosomal GTPases. Traffic. 10:76-87.
493
19.
Hayashi T, Nakamichi M, Naitou H, Ohashi N, Imai Y, Miyake M. 2010.
494
Proteomic analysis of growth phase-dependent expression of Legionella
495
pneumophila proteins which involves regulation of bacterial virulence traits. PLoS
496
One 5:e11718.
497
20.
Baskerville A, Fitzgeorge RB, Broster M, Hambleton P, Dennis PJ. 1981.
498
Experimental transmission of legionnaires' disease by exposure to aerosols of
499
Legionella pneumophila. Lancet. 2:1389-1390.
500
21.
Blanchard DK, Djeu JY, Klein TW, Friedman H, Stewart WE, 2nd. 1987.
501
Induction of tumor necrosis factor by Legionella pneumophila. Infect. Immun.
502
55:433-437.
503
22.
Fitzgeorge RB, Baskerville A, Broster M, Hambleton P, Dennis PJ. 1983.
504
Aerosol infection of animals with strains of Legionella pneumophila of different
505
virulence: comparison with intraperitoneal and intranasal routes of infection. J.
506
Hyg. 90:81-89.
25
507
23.
Galka F, Wai SN, Kusch H, Engelmann S, Hecker M, Schmeck B, Hippenstiel
508
S, Uhlin BE, Steinert M. 2008. Proteomic characterization of the whole
509
secretome of Legionella pneumophila and functional analysis of outer membrane
510
vesicles. Infect. Immun. 76:1825-1836.
511
24.
pneumophila cell envelope. Front. Microbiol. 2:74.
512 513
Shevchuk O, Jager J, Steinert M. 2011. Virulence properties of the Legionella
25.
Jepras RI, Fitzgeorge RB, Baskerville A. 1985. A comparison of virulence of
514
two strains of Legionella pneumophila based on experimental aerosol infection of
515
guinea-pigs. J. Hyg. 95:29-38.
516
26.
Aurass P, Pless B, Rydzewski K, Holland G, Bannert N, Flieger A. 2009.
517
bdhA-patD operon as a virulence determinant, revealed by a novel large-scale
518
approach for identification of Legionella pneumophila mutants defective for
519
amoeba infection. Appl. Environ. Microbiol. 75:4506-4515.
520
27.
Legionella pneumophila. Methods. Mol. Biol. 954:225-230.
521 522
Jager J, Steinert M. 2013. Enrichment of outer membrane vesicles shed by
28.
Dromann D, Rupp J, Rohmann K, Osbahr S, Ulmer AJ, Marwitz S,
523
Roschmann K, Abdullah M, Schultz H, Vollmer E, Zabel P, Dalhoff K,
524
Goldmann T. 2010. The TGF-beta-pseudoreceptor BAMBI is strongly expressed
525
in COPD lungs and regulated by nontypeable Haemophilus influenzae. Respir.
526
Res.11:67.
527
29.
Vollmer E, Galle J, Lang DS, Loeschke S, Schultz H, Goldmann T. 2006. The
528
HOPE technique opens up a multitude of new possibilities in pathology. Rom. J.
529
Morphol. Embryol. 47:15-19.
26
530
30.
Marwitz S, Abdullah M, Vock C, Fine JS, Visvanathan S, Gaede KI, Hauber
531
HP, Zabel P, Goldmann T. 2011. HOPE-BAL: improved molecular diagnostics by
532
application of a novel technique for fixation and paraffin embedding. J.
533
Histochem. Cytochem. 59:601-614.
534
31.
Helbig JH, Ludwig B, Luck PC, Groh A, Witzleb W, Hacker J. 1995.
535
Monoclonal antibodies to Legionella Mip proteins recognize genus- and species-
536
specific epitopes. Clin. Diagn. Lab. Immunol. 2:160-165.
537
32.
Seneta E. Probability inequalities and Dunnett's test. Multiple comparisons,
538
selection, and applications in biometry (Hamilton, ON, 1991). Vol. 134. New York:
539
Dekker, Marcel, 1993:29-45.
540
33.
Bolstad BM, Irizarry RA, Astrand M, Speed TP. 2003. A comparison of
541
normalization methods for high density oligonucleotide array data based on
542
variance and bias. Bioinformatics 19:185-193.
543
34.
Juli C, Sippel M, Jager J, Thiele A, Weiwad M, Schweimer K, Rosch P,
544
Steinert M, Sotriffer CA, Holzgrabe U. 2011. Pipecolic acid derivatives as small-
545
molecule inhibitors of the Legionella MIP protein. J. Med. Chem. 54:277-283.
546
35.
Fernandez-Moreira E, Helbig JH, Swanson MS. 2006. Membrane vesicles shed
547
by Legionella pneumophila inhibit fusion of phagosomes with lysosomes. Infect.
548
Immun. 74:3285-3295.
549
36.
Jules M, Buchrieser C. 2007. Legionella pneumophila adaptation to intracellular
550
life and the host response: clues from genomics and transcriptomics. FEBS. Lett.
551
581:2829-2838.
27
552
37.
Rossier O, Dao J, Cianciotto NP. 2009. A type II secreted RNase of Legionella
553
pneumophila facilitates optimal intracellular infection of Hartmannella vermiformis.
554
Microbiology. 155:882-890.
555
38.
Tyson JY, Pearce MM, Vargas P, Bagchi S, Mulhern BJ, Cianciotto NP.
556
2013. Multiple Legionella pneumophila type II secretion substrates, including a
557
novel protein, contribute to differential infection of amoebae Acanthamoeba
558
castellanii, Hartmannella vermiformis, and Naegleria lovaniensis. Infect. Immun.
559
81:1399-1410.
560
39.
Rupp J, Droemann D, Goldmann T, Zabel P, Solbach W, Vollmer E,
561
Branscheid D, Dalhoff K, Maass M. 2004. Alveolar epithelial cells type II are
562
major target cells for C. pneumoniae in chronic but not in acute respiratory
563
infection. FEMS Immunol. Med. Microbiol. 41:197-203.
564
40.
Xu F, Droemann D, Rupp J, Shen H, Wu X, Goldmann T, Hippenstiel S, Zabel
565
P, Dalhoff K. 2008. Modulation of the inflammatory response to Streptococcus
566
pneumoniae in a model of acute lung tissue infection. Am. J. Respir. Cell. Mol.
567
Biol. 39:522-529.
568
41.
Glavin FL, Winn WC, Craighead JE. 1979. Ultrastructure of lung in
569
Legionnaires' disease. Observations of three biopsies done during the Vermont
570
epidemic. Ann. Intern. Med. 90:555-559.
571
42.
Case CL, Kohler LJ, Lima JB, Strowig T, de Zoete MR, Flavell RA, Zamboni
572
DS, Roy CR. 2013. Caspase-11 stimulates rapid flagellin-independent pyroptosis
573
in response to Legionella pneumophila. Proc. Natl. Acad. Sci. 110:1851-1856.
28
574
43.
Roy CR, Berger KH, Isberg RR. 1998. Legionella pneumophila DotA protein is
575
required for early phagosome trafficking decisions that occur within minutes of
576
bacterial uptake. Mol. Microbiol. 28:663-674.
577
44.
Singh G, Katyal SL, Brown WE, Kennedy AL, Singh U, Wong-Chong ML.
578
1990. Clara cell 10 kDa protein (CC10): comparison of structure and function to
579
uteroglobin. Biochim. Biophys. Acta. 1039:348-355.
580
45.
Vasanthakumar G, Manjunath R, Mukherjee AB, Warabi H, Schiffmann E.
581
1988.
582
blastocyst rejection. Biochem. Pharmacol. 37:389-394.
583
46.
Inhibition of phagocyte chemotaxis by uteroglobin, an inhibitor of
Hayashida S, Harrod KS, Whitsett JA. 2000. Regulation and function of CCSP
584
during pulmonary Pseudomonas aeruginosa infection in vivo. Am. J. Physiol.
585
Lung. Cell. Mol. Physiol. 279:L452-459.
586
47.
Harrod KS, Jaramillo RJ. 2002. Pseudomonas aeruginosa and tumor necrosis
587
factor-alpha attenuate Clara cell secretory protein promoter function. Am. J.
588
Respir. Cell. Mol. Biol. 26:216-223.
589
48.
Mukhopadhyay S, Chen Y, Sankala M, Peiser L, Pikkarainen T, Kraal G,
590
Tryggvason K, Gordon S. 2006. MARCO, an innate activation marker of
591
macrophages, is a class A scavenger receptor for Neisseria meningitidis. Eur. J.
592
Immunol. 36:940-949.
593
49.
Thelen T, Hao Y, Medeiros AI, Curtis JL, Serezani CH, Kobzik L, Harris LH,
594
Aronoff DM. 2010. The class A scavenger receptor, macrophage receptor with
595
collagenous structure, is the major phagocytic receptor for Clostridium sordellii
596
expressed by human decidual macrophages. J. Immunol. 185:4328-4335.
29
597
50.
Mukouhara T, Arimoto T, Cho K, Yamamoto M, Igarashi T. 2011. Surface
598
lipoprotein PpiA of Streptococcus mutans suppresses scavenger receptor
599
MARCO-dependent phagocytosis by macrophages. Infect. Immun. 79:4933-4940
600
51.
Elomaa, O., et al., Cloning of a novel bacteria-binding receptor structurally related
601
to scavenger receptors and expressed in a subset of macrophages. Cell, 1995.
602
80(4): p. 603-9.
603
52.
Elomaa O, Kangas M, Sahlberg C, Tuukkanen J, Sormunen R, Liakka A,
604
Thesleff I, Kraal G, Tryggvason K. 1995. Cloning of a novel bacteria-binding
605
receptor structurally related to scavenger receptors and expressed in a subset of
606
macrophages. Cell. 80:603-609.
607
53.
Bowdish DM, Sakamoto K, Kim MJ, Kroos M, Mukhopadhyay S, Leifer CA,
608
Tryggvason K, Gordon S, Russell DG. 2009. MARCO, TLR2, and CD14 are
609
required for macrophage cytokine responses to mycobacterial trehalose
610
dimycolate and Mycobacterium tuberculosis. PLoS. Pathog. 5:e1000474.
611
54.
Wagner C, Khan AS, Kamphausen T, Schmausser B, Unal C, Lorenz U,
612
Fischer G, Hacker J, Steinert M. 2007. Collagen binding protein Mip enables
613
Legionella pneumophila to transmigrate through a barrier of NCI-H292 lung
614
epithelial cells and extracellular matrix. Cell. Microbiol.9:450-462.
615
616
30
617
618
Figure legends
619
Figure 1: Progression of tissue damage caused by L. pneumophila within the
620
human lung over time. (A) Haematoxylin-eosin-stained section of an HTLE challenged
621
with L. pneumophila after 48 h at 20× magnification. Protein exudate into the alveolar
622
lumen can be frequently observed in close proximity to sites of tissue damage (*).
623
Collagen fibres within the alveolar septa appear loose (#), indicating decreasing septal
624
integrity. (B, C, D) Haematoxylin-eosin stains of uninfected control HTLEs at 40×
625
magnification after 2, 24 and 48 h of incubation. (E, F, G) Haematoxylin-eosin stains of
626
HTLEs challenged with L. pneumophila at 2, 24 and 48 h after infection. No damage
627
could be observed 2 h after infection (B, E). After 24 h, alveolar integrity is affected,
628
shown by loosening of the collagen backbone in L. pneumophila-infected tissue (# in F)
629
compared to uninfected controls (C). After 48 h of incubation, the integrity of uninfected
630
tissue is moderately compromised (D), while L. pneumophila-infected tissue exhibits
631
severe damage (G). Almost all septa-lining epithelial cells are delaminated (arrow in G)
632
and the underlying scaffold appears disintegrated. (H) High power magnification (100×)
633
shows L. pneumophila targeted by immunohistochemistry with an anti-Mip antibody (red
634
signal) 48 h after infection. The pathogen adhered to alveolar septa and alveolar
635
epithelial cells. At these sites, severe tissue damage was frequently observed.
636
Figure 2: Observed damage score of stimulated HLTEs compared to uninfected
637
control. (A) For each time point and condition, histological damage was classified in
638
regard to protein exudate, delamination of epithelial cells and integrity of alveolar septa
31
639
on a scale from 0-3 and added up for each donor; n = 6-9 per condition and time point.
640
Total damage score medians and interquartile ranges were calculated; values of
641
infected or OMV-stimulated samples were compared to the uninfected control at the
642
respective time point using the Mann-Whitney test with a Bonferroni-corrected
643
confidence interval of 98.3 %. *: p < 0.05, **: p < 0.01, ***: p < 0.001. (B) Uninfected
644
control, (C) L. pneumophila wildtype-infected HLTE, (D) sample infected with the DotA-
645
deficient mutant, (E) sample incubated with 100 µg/ml OMVs; haematoxylin-eosin-
646
stained sections of HLTEs after 24 h of incubation. Infection with wildtype
647
L. pneumophila leads to loosening of collagen backbones in alveolar septa and
648
delamination of epithelial cells (C). The DotA-deficient mutant does not show a
649
significant difference compared to the uninfected control (D). Incubation with OMVs
650
resulted in severe tissue damage, including a wide-spread delamination of epithelial
651
cells and loss of septal integrity (E).
652
Figure 3: Detection of L. pneumophila and OMVs in the alveolar compartment.
653
Immunohistochemistry with anti-Mip (A, C, E) and anti-MOMP antibodies (B, D, F)
654
followed by visualization with HRP polymer and permanent AEC (red signals).
655
Uninfected controls remain negative for both antibodies (A, B); L. pneumophila is mainly
656
observed in alveolar macrophages with both antibodies (C, D). In contrast to the
657
detection of the whole pathogen, OMVs were only detected sufficiently with the anti-
658
MOMP antibody (F). Alveolar epithelial cells type I, which line the alveolar compartment
659
(see arrows) and the septa, are targeted by Aquaporin-5 (AQP5) in medium control
660
tissue (G). Alveolar macrophages as the most abundant immune cells in the alveolar
32
661
space (see arrows) are targeted by anti-CD68 staining (H). All images at 40×
662
magnification and all HLTEs after 24 h of incubation.
663
Figure 4: L. pneumophila adheres extracellularly to the alveolar compartment and
664
infects epithelial cells. Immunohistochemistry with an anti-Mip antibody on
665
L. pneumophila-infected (A, B, E) and uninfected HLTEs (C, D), 100× magnification.
666
Colonization of alveolar epithelial cells and connective tissue was observed after 24 h
667
(A) and 48 h (B). L. pneumophila infected alveolar epithelial cells (* in A and B) and
668
tended to form clusters over the course of stimulation (B). Furthermore, positive staining
669
for L. pneumophila was frequently observed in dead alveolar macrophages (E).
670
Figure 5: Replication of the L. pneumophila wildtype (wt) and the DotA-deficient
671
strain in HLTEs. Infected HTLEs were weighed, homogenised and plated on BCYE
672
agar at the indicated time points after infection. The graph visualises means and
673
standard deviations of triplicate experiments with tissue from eight donors. **: p < 0.01;
674
n.s.: not significant (p > 0.05) as determined by the student’s t-test for samples with
675
identical variance.
676
Figure 6: Transcriptional response of L. pneumophila-infected HLTEs and
677
targeting of candidate genes on the protein level. (A) Heat map analysis and
678
hierarchical clustering of 2499 genes with a log2-fold change ≥ 2. RNA was isolated
679
from L. pneumophila wildtype-infected and uninfected HLTEs after 24 h of incubation (n
680
= 2 each) to analyse the acute phase of infection. Distinct clusters were found to be
681
differentially regulated after infection. (B-E) Immunohistochemistry against uteroglobin
682
(B, C) and MARCO (D, E) on HLTEs after 24 h of incubation with medium (B, D) and 33
683
L. pneumophila (C, E). Both proteins were mainly detected on alveolar macrophages.
684
MARCO was observed to be down-regulated over the course of stimulation at sites of
685
tissue damage (* in E). Images were taken at 40× magnification with permanent AEC
686
(red signals) as the chromogen. Microarray data for uteroglobin were validated by qPCR
687
normalized to Ribosomal Like Protein 32 (RPL32) from eight independent experiments
688
(F). Uteroglobin is significantly upregulated after infection with L. pneumophila as
689
determined by the student’s t-test (*: p < 0.05). Secretion of uteroglobin in HLTE
690
supernatants was quantified by ELISA (G) from eight independent experiments; no
691
significant difference was detected as computed with one-way ANOVA.
692
Figure 7: Comparison of tissue damage in HLTEs caused by different bacterial
693
concentrations. HLTEs from two different patients were challenged with increasing
694
concentrations of bacteria. The damage score does not increase markedly with the
695
number of infecting bacteria (A; median +/- IQR). The tissue structure is visualized in B
696
(medium control) and C (increasing concentrations of L. pneumophila wildtype and the
697
DotA-negative strain). Sections of paraffin-embedded HLTEs were stained with
698
hematoxylin and eosin and the damage score was evaluated. All images were taken at
699
40x magnification 48 h after infection.
700
701
34
702
Tables
703
Table 1: Damage score categories of HLTEs shown as median ± interquartile range.
2h
24 h
48 h
Protein
Epithelial
Septal
Total damage
exudate
delamination
damage
score
Control
0±1
0±0
0±0
1±1
wt
1±2
0±1
0±0
2±2
DotA-
1.5 ± 2.75
0±0
0±0
1.5 ± 1.75
OMVs
0±1
0 ± 0.5
0 ± 0.5
0 ± 2.5
Control
1±2
1±1
0±1
2±2
wt
2±2
1 ± 1.25
2±1
5 ± 1.5
DotA-
1±1
1±0
1±0
2.5 ± 1.5
OMVs
1 ± 1.5
2±1
2 ± 1.5
6±2
Control
0.5 ± 1.25
0.5 ± 1
1±0
3 ± 1.25
2±1
2±1
3±2
6±1
DotA-
1 ± 0.75
1±0
1.5 ± 1
4 ± 0.75
OMVs
1±1
2.5 ± 2
2.5 ± 2
5.5 ± 1.75
wt
704
705
HTLEs (human lung tissue explants) were incubated with medium, the indicated
706
L. pneumophila strain or 100 µg/mL OMVs (outer membrane vesicles) for the indicated
707
time. Sample sizes were 6-9, depending on the infection condition. Per phenotype, the
708
severity of damages was scaled from 0 to 3. The total damage score was calculated by
35
709
adding up the specific damage scores of each donor at a given time point. The median ±
710
interquartile range of this analysis is shown in the right column.
711
712
36
713 714
Table 2: Total number of differentially regulated genes in HLTEs 24 h after infection with
715
L. pneumophila.
Log2-fold change
N (regulated genes)
≤2.0
1986
>2.0 - 3.0
156
≥ 3.0
20
≤ -2.0
330
≥ -2.0
7
716
717
37
718 719
Table 3: Gene ontology term analysis of genes differentially regulated after infection of
720
HLTEs with L. pneumophila. GO accession
GO term
Corrected p-value
GO:0005576
extracellular region
0.00003
GO:0044421
extracellular region part
0.00003
GO:0005615
extracellular space
0.00020
GO:0002376
immune system process 0.00854
GO:0005929
cilium
0.01026
GO:0042953
lipoprotein transport
0.01302
GO:0035085
cilium axoneme
0.02505
GO:0005930
axoneme
0.03395
721
722
2499 genes with a fold change ≥ 2 were used as input list for Gene Ontology analysis. A
723
Benjamini-Yekutieli correction was applied and a p-value ≤ 0.05 was set as the cut-off.
724
38
725 726
Table 4: Differential gene expression of uteroglobin and MARCO after infection with
727
wildtype L. pneumophila.
Relative expression level Log2 fold change Gene ID
Name
Donor Control
Wildtype
NM_003357 Uteroglobin 1
7288.43
28741.66
2
1157.49
6897.58
mean
4222.96
17819.62
1
3190.08
958.49
2
9088.06
2412.81
mean
6139.07
1685.65
NM_006770 MARCO
+ 2.18
- 1.96
728
729
Quantile-normalized expression levels of uteroglobin and MARCO in two donors
730
obtained from microarray raw data using Direct Array software.
731
39