AAC Accepts, published online ahead of print on 4 November 2013 Antimicrob. Agents Chemother. doi:10.1128/AAC.01400-13 Copyright © 2013, American Society for Microbiology. All Rights Reserved.
1 2
Application of DNA chip scanning technology for the automatic detection of Chlamydia trachomatis and Chlamydia pneumoniae inclusions
3 4
Running title: Chlamydia inclusion detection system ChlamyCount
5 6 7
Anita Bogdanov1*, Valeria Endrész2*, Szabolcs Urbán3, Ildikó Lantos2, Judit Deák1, Katalin Burián2, Kamil Önder4, Ferhan Ayaydin5, Péter Balázs3, Dezs P. Virok1#
8 9
1, Institute of Clinical Microbiology, University of Szeged, Szeged, Hungary Address:
10
H-6725 Szeged Semmelweis str. 6.
11
2, Institute of Medical Microbiology and Immunobiology, University of Szeged,
12
Szeged, Hungary Address: H-6720 Szeged, Dóm sqr. 10.
13
3, Department of Image Processing and Computer Graphics, Institute of Informatics,
14
University of Szeged, Szeged, Hungary H-6720, Szeged, Árpád sqr. 2.
15
4, Department of Dermatology, Paracelsus Medical University, Salzburg, Austria
16
Address: 5020 Salzburg, Strubergasse 21.
17
5, Cellular Imaging Laboratory, Biological Research Center, Hungarian Academy of
18
Sciences, Hungary Address: H-6726 Szeged, Temesvári krt. 62.
19 20
*: these authors contributed equally
21
#: corresponding author
22
email of the corresponding author:
[email protected] 23 24
1
25
Abstract
26 27
Chlamydiae are obligate intracellular bacteria that propagate in the inclusion, a
28
specific niche inside the host cell. The standard method for counting chlamydiae is
29
the immunofluorescent staining and manual counting of chlamydial inclusions. High
30
or medium throughput estimation of the reduction in chlamydia inclusions should be
31
the basis of testing antichlamydial compounds and other drugs that positively or
32
negatively influence chlamydial growth, yet low-throughput manual counting is the
33
common approach. To overcome the time-consuming and subjective manual
34
counting we developed an automatic inclusion counting system based on a
35
commercially available DNA chip scanner. Fluorescently labeled inclusions are
36
detected by the scanner, and the image is processed by ChlamyCount, a custom
37
plugin of the ImageJ software environment. ChlamyCount was able to measure the
38
inclusion counts over a one log dynamic range with high correlation to the theoretical
39
counts. ChlamyCount was capable of accurately determining the minimum inhibitory
40
concentration of the novel antimicrobial compound PCC00213 and the already
41
known antichlamydial antibiotics moxifloxacin and tetracycline. ChlamyCount was
42
also able to measure the chlamydial growth altering effect of drugs that influence
43
host-bacterium
44
cycloheximide. ChlamyCount is an easily adaptable system for testing antichlamydial
45
antimicrobials and other compounds that influence Chlamydia-host interactions.
interaction
such
as
interferon-gamma,
DEAE-dextran
and
46
2
47
Introduction
48 49
Chlamydia pneumoniae (C. pneumoniae) and various C. trachomatis serovars
50
are involved in medically important human diseases. C. pneumoniae is a frequent
51
source of community acquired pneumonia and is suspected of participating in the
52
pathogenesis of chronic diseases such as asthma and atherosclerosis (1, 2). C.
53
trachomatis serovars A-C are involved in trachoma pathogenesis; D-K serovars
54
induce pelvic inflammatory diseases (PID), infertility and reactive arthritis, while LGV
55
serovars are the pathogens that cause lymphogranuloma venereum, an STD with
56
systemic manifestations. The antibiotic therapy of chlamydia infections includes
57
tetracyclines and macrolides (3-5). While the antibiotic therapy is effective in the
58
majority of the cases, tetracycline and azithromycin resistance was reported (6, 7).
59
Discovery of novel antimicrobial compounds for treatment and possible
60
chemoprevention is a medically important research task. High-throughput testing of
61
potentially antichlamydial compounds is hampered by the small size and obligate
62
intracellular propagation of the bacterium. After infecting the host cell, the chlamydia
63
propagates in a distinct cellular space called an inclusion. Since, at a low multiplicity
64
of infection, one chlamydia can form one inclusion, the original chlamydia count is
65
indirectly measured by labeling and manual microscopy counting of inclusions. To
66
circumvent the labor-intensive and subjective manual counting, we designed a
67
relatively low-cost, easy-to-use system that automatically counts chlamydial
68
inclusions. The system consists of a commercial DNA chip scanner and custom-
69
made image analysis software. DNA chip and microarray scanners are widely used
70
devices for the high resolution scanning of glass slides that contain a large number of
71
DNA probes that bind fluorescently labeled cDNAs or DNAs. We applied this system
72
to detect fluorescently labeled chlamydial inclusions in host cells propagated in a 16-
73
well chamber slide. We designed ChlamyCount, a custom ImageJ plugin for the
74
completely automatic detection of fluorescently labeled chlamydia inclusions on the
75
scanned image. ImageJ is a Java-based, platform-independent, freely available
76
image analysis software package (8). ImageJ can be specifically tailored to a given
77
application using user-made plugins. The image processing with ChlamyCount is
78
almost fully automatic, including the extraction of the areas of the 16 wells, the
79
automatic detection of inclusions in each well, dissecting and individually counting
80
inclusions and automatic reporting. 3
81
ChlamyCount was used to determine the minimum inhibitory concentration
82
(MIC) of the known antichlamydial antibiotics tetracycline and moxifloxacin and the
83
novel antimicrobial compound PCC00213. ChlamyCount was also used to measure
84
the effect of compounds that indirectly influence the chlamydial growth cycle such as
85
interferon-gamma (IFN-γ), DEAE-dextran and cycloheximide. The ChlamyCount
86
ImageJ
87
szeged.hu/ipcg/projects/medical.html.
plugin
is
available
for
non-commercial
use
at
http://www.inf.u-
88 89
Materials and Methods
90 91
Chlamydial strains. Two Chlamydia species were used in this study: C. trachomatis
92
(serovar D, UW-3/CX reference strain and serovar L2, strain VR-577, ATCC) and C.
93
pneumoniae (CWL029, ATCC). Chlamydia strains were propagated and partially
94
purified according to methods described previously with modifications (9, 10). Briefly,
95
DEAE-dextran (45 mg/ml in Hank’s balanced salt solution, HBSS) treated McCoy
96
cells were infected with C. trachomatis serovar D, and after a 2-day incubation in a
97
culture medium (MEM with Earle salts supplemented with 10% heat-inactivated fetal
98
bovine serum (FBS), 0,5% glucose, 2 mmol/l L-glutamine, 1x non-essential amino
99
acids, 8 mmol/l HEPES, 25 µg/ml gentamicin) in the presence of 1 µg/ml
100
cycloheximide, the infected cell layers were washed with PBS, frozen and covered
101
with 50 µl/cm2 sucrose-phosphate-glutamic acid buffer (SPG). After 2 freeze-thaw
102
cycles the cells were collected and separated from the cell debris by a 10 min, 800 g
103
centrifugation. C. trachomatis serovar L2 and C. pneumoniae were cultured in HEp2
104
cells and partially purified and concentrated by 30000 g centrifugation as described
105
previously (11).The pelleted EBs were re-suspended in SPG. Stocks of chlamydial
106
EBs were aliquoted and stored in SPG until use at -80 °C. The titer of the infectious
107
EBs was determined by inoculation of serial dilutions of the EB preparation onto
108
McCoy (C. trachomatis) or HEp-2 (C. pneumoniae) monolayers grown on 13mm
109
diameter cover slips. Inoculated cells were centrifuged for 1 h at 800 g, and after 24 h
110
(C. trachomatis) or 48 h (C. pneumoniae) culture in cycloheximide containing growth
111
medium, the cells were fixed with acetone and stained with monoclonal anti-
112
Chlamydia LPS antibody (AbD Serotec, Oxford, UK) and FITC-labeled anti-mouse
113
IgG (Sigma-Aldrich, St. Louis, MO, USA). The number of cells containing chlamydial
4
114
inclusions was counted under a UV microscope, and the titer was expressed as
115
IFU/ml.
116 117
Culture of chlamydiae on a chamber slide. Chamber slides with 16 wells (Lab-Tek
118
Chamber Slide System) consisting of a removable, plastic chamber attached to a
119
specially treated standard glass slide were used to culture host cells for infection with
120
the Chlamydia strains. The slides were treated with 0.01% poly-L-lysine at room
121
temperature (RT) for 15 minutes in order to optimize cell attachment. McCoy cells
122
were transferred into the wells of the chamber slides at a density of 3x104 cells/well in
123
100 µl of culture medium (see above). The slides were incubated for 1 hour at room
124
temperature in order to reduce the edge effect (12, 13), then overnight at 37 °C under
125
5% CO2 atmosphere to obtain a 90% confluent cell layer. For C. trachomatis serovar
126
D infection, the wells were washed with 200 µl/well of Hank’s balanced salt solution
127
(HBSS), then a 1% DEAE-dextran solution (80 µl/well) was added to all wells and
128
were incubated for 15 minutes, RT. After removal of the DEAE-dextran solution, the
129
cells were infected at 1 multiplicity of infection (MOI, 1 IFU/cell) in each well or with
130
serial 2-fold dilutions of stock in SPG starting with 1 IFU/cell. SPG was measured into
131
similarly treated control wells. The chamber slides were incubated at room
132
temperature in a vertical shaker (180 rpm) for 2 hours. The slides infected with C.
133
trachomatis serovar L2 with MOI ranging from 1 IFU/cell to 1:64 IFU/cell in SPG were
134
incubated at 35 °C under 5% CO2 for 2 hours without DEAE-dextran pretreatment.
135
For C. pneumoniae infection the DEAE-dextran treatment was used, or the slides
136
were centrifuged with C. pneumoniae stock with MOI ranging from 1:8 to 1:512, 400
137
g, 60 minutes at RT. After infection, the EB containing inoculums in the wells were
138
replaced with a culture medium containing 1 µg/ml cycloheximide. The slides were
139
incubated at 37 °C under 5% CO2 for 24 or 48 hours after infection with C.
140
trachomatis D and L2 or C. pneumoniae, respectively, and the cells were fixed for
141
immunofluorescence staining.
142
Moxifloxacin (Avelox,
143
Inhibition of chlamydial growth with antibiotics and IFN-
144
Bayer Pharma AG) diluted in a culture medium and tetracycline hydrochloride powder
145
(Sigma-Aldrich) dissolved in distilled water were used. The concentration range of
146
0.25-0.004 µg/ml for moxifloxacin and of 0.04-0.0006 µg/ml for tetracycline with 2-fold
5
147
dilutions were tested. The stock solution of antibacterial drug candidate PCC00213
148
(10 mg/ml) was prepared in DMSO, and was diluted 2-fold in DMSO until 0.156
149
mg/ml. Cycloheximide containing culture medium was prepared by a 100-fold dilution
150
of the DMSO-diluted compound, resulting in a series of concentration ranging from
151
100 to 1.56 µg/ml with 1% DMSO content. After infection of McCoy cells with C.
152
trachomatis D (MOI 1), the culture medium with cycloheximide was supplemented
153
with the serial 2-fold dilutions of the respective antibiotics and was added to duplicate
154
wells. Control infected wells were cultured without adding any antibiotics; for
155
PCC00213, 1% DMSO containing medium was added to the control wells. Murine
156
IFN-γ (Peprotech, Rocky Hill, NJ, USA) was reconstituted according to the
157
manufacturers’ instructions and the 20 000 U/ml stock solution was stored at -80 °C
158
until use. On the day before infection, the established cell layers were treated with
159
serial twofold dilutions of murine IFN-γ over the concentration range of 100-0.046
160
IU/ml. As a control, human IFN-γ (Peprotech, Rocky Hill, NJ, USA) was also added
161
over a concentration range of 100 IU/ml-1.5 IU/ml. After the infection procedure, IFN-
162
γ diluted in a cycloheximide-free medium was added at the same concentration as
163
that used for the pretreatment of cells before infection.
164 165
Immunofluorescent labeling and scanning. Cells in chamber slides infected with
166
C. pneumoniae and C. trachomatis were evaluated by immunofluorescent staining.
167
After removing the culture medium from the slides, the cells were washed twice with
168
PBS (200 µl/well). Then the chamber structure was detached from the slides and the
169
cells were fixed with pre-cooled 100% acetone for 10 min at -20 °C. Alexa-647-
170
labeled anti-chlamydia LPS antibody (AbD Serotec, Oxford, UK) diluted 1:200 was
171
used for the detection of chlamydial inclusions. After incubation for 1 h at 37 °C, the
172
cells were washed three times with PBS for 7 minutes each and finally with distillated
173
water. Fluorescence signals were analyzed with an Axon GenePix Personal 4100A
174
DNA-chip scanner and GenePix Pro 6.1 software (Molecular Devices, Sunnyvale,
175
CA, USA) using the Cy5 channel and a 5 µm resolution.
176 177 178
6
179
Image Processing. The scanned images were about 50-60 MB single-image 16-bit
180
.tiff files with an intensity dynamic range of four orders of magnitude. The images
181
were processed by ChlamyCount in two phases: preprocessing and analysis. In the
182
preprocessing phase, after loading the image, the contrast was enhanced by
183
performing the “Histogram normalization” method of ImageJ. After, the regions of
184
interests (ROI) containing the 16-well areas of the scanned image were cropped. The
185
cropping scheme was specifically designed for the 16-well Lab-Tek Chamber Slide
186
System. In the subsequent steps, the image of each well was processed
187
independently. To reduce noise effects, pixels of the ROI below a predefined
188
threshold were eliminated using the ImageJ threshold operation. The default intensity
189
threshold was set to 15.000, which value had been empirically determined.
190
Nevertheless, depending on the used fluorescent antibody and scanner type, the
191
image intensity may change; hence the user can manually adjust the inclusion
192
intensity and area thresholds. In a further noise-reduction step, a greyscale
193
morphological opening (a morphological erosion followed by a dilation) was applied,
194
using the minimum and maximum filters of ImageJ on a 3x3 window.
195
In the analysis phase, the images of each well were processed independently. First,
196
a binarized copy of the result of the preprocessing phase was produced. Then, the
197
ImageJ Analyze Particles function was called to encounter and outline the particles
198
(potential inclusion or inclusions) in the well. Since confluent inclusions of various
199
shapes could occur, no size or circularity constraints were set during the process. In
200
the next step, the median area of the particles was computed, and particles having
201
an area greater than the calculated median area were further processed. If the
202
perimeter of the particle was less than 500 pixels, then the particle was decomposed
203
by Voronoi-tesselation using the local greyscale minima as seeds, and the Euclidean-
204
distance. However, no splitting was performed if the perimeter of the particle was
205
greater than 500 pixels, because those regions were likely to represent areas of
206
potentially aspecific labeling of the background and/or host cells. Setting a higher
207
intensity threshold in the preprocessing step could reduce the occurrence of such
208
areas. After the splitting step, the ImageJ Analyze Particles function was again called
209
to identify the final number of particles and to find their boundaries, which were
210
highlighted in red on the original greyscale image. Finally, the result of the analysis
211
was reported in .txt, .xls, and .pdf files. The .pdf file contains the numerical results
212
and the processed images of the 16 areas. It should be noted, that for easier visibility 7
213
of the inclusions on the small scale Figures, the images of the 16 areas were further
214
contrast enhanced using the duotone feature of the PhotoFiltre image processing
215
software.
216 217
Confocal Microscopy and Imaging. Confocal laser scanning microscopy was
218
performed using Olympus FV1000 confocal laser scanning microscope (Olympus Life
219
Science Europe GmbH, Hamburg, Germany). Microscope configuration was the
220
following: objective lenses: UPLSAPO 10x (NA:0.4) UPLSAPO 20x (NA:0.75),
221
LUMPLFL 40x (NA:0.8); sampling speed: 4 or 8 µs/pixel; line averaging: 2x; confocal
222
aperture: 200 µm; image dimension: 512x512 pixels; scanning mode: sequential
223
unidirectional; excitation: 633 nm HeNe laser; laser transmissivity: 45%, AlexaFluor
224
647 was detected between 645-745 nm. Transmitted light images were also captured
225
and paired with each fluorescence image using 633 nm laser. Using a 10x objective,
226
16 successive images were captured to encompass the full diameter of each well of
227
the 16-well chamber slide. Images were stitched together to make an image strip of
228
512x8192 pixels using "import image sequence" and "make montage" functions of
229
ImageJ software. Infected cells were identified and manually scored on these
230
composite images. Since the area of the image strip is 12.2 times smaller than the
231
area of the whole well, to estimate the approximate total number of inclusions/well
232
the counted inclusions were multiplied by 12.2. For close-up analyses, microscopy
233
slides
234
(SouthernBiotech, Birmingham, AL, USA) and 40x immersion objective was used to
235
capture images.
were
mounted
with
Fluoromount-G
antifade
mounting
solution
236 237 238
Results
239 240
ChlamyCount software. McCoy epithelial cells grown on a 16-well chamber slide
241
were used for chlamydial infection. Depending on the chlamydial strain, we removed
242
the chamber 24 or 48 hours post infection (p.i.), fixed the host cells and stained the
243
chlamydial inclusions with the Cy5 analogue Alexa-647-labeled anti-chlamydial LPS
244
antibody. The stained inclusions were scanned with a commercially available Axon
245
GenePix 4100 DNA chip scanner. The scanner is capable of scanning Cy3 or Cy5 8
246
labeled spots on a regular microscope glass slide with a maximum resolution of 5
247
µm, which is comparable to the size of a chlamydial inclusion (Fig 1A-D). The
248
scanned image is processed by the ChlamyCount software. The thresholds for the
249
minimum intensity and size of the inclusion are adjustable by the user, otherwise
250
preset threshold values are applied (Fig. 2A). The two threshold values are the only
251
parameters that can be adjusted by the user; all the subsequent steps are automatic.
252
If the user adjusts the intensity and/or area thresholds, the effect of the adjustment
253
can immediately be seen by a magnified quarter of the uppermost and lowest left-
254
hand side wells (Fig. 2B). In the next step, ChlamyCount automatically crops the 16
255
areas containing the host cells from the complete image. For each area,
256
ChlamyCount processes the images as follows: i, the pixels with low intensity values
257
are likely noise, thus they are eliminated by thresholding the image with an intensity
258
and area threshold value ii, regions having an area greater than the median of the
259
area of all regions (the suspected size of a single inclusion) are split by finding the
260
local maxima in the regions and then assigning each point to the same closest
261
maximum to form smaller regions and iii, the identified particles are encountered and
262
their boundaries are determined. It is worth to note that the splitting of high intensity
263
areas is not complete, larger high intensity areas likely to contain multiple infected
264
cells without clear boundaries among them are regarded as one particle (inclusion).
265
On the other hand this effect can be observed both at higher and lower MOIs,
266
therefore the correlation of detected particles with the theoretical inclusion count
267
remains. After the image analysis, ChlamyCount provides a detailed .txt, .xls output
268
of the 16 areas with the inclusion counts, total areas above the threshold and if
269
confluent high intensity areas were detected and could not be dissect to individual
270
inclusions by the software the area/median area ratio is also provided. Median area
271
size is suspected to be the area of an individual inclusion when the MOI is not high
272
i.e. equal or below 1. The third output file is a .pdf file, with the above mentioned
273
numerical results and the processed images of the 16 areas (Fig. 2C). A typical
274
scanning time is about 10 minutes and the image analysis time is about 1-5 minutes.
275 276
Measuring the dynamic range of detection of C. trachomatis serovar D, C.
277
trachomatis serovar L2 and C. pneumoniae. The ChlamyCount system was tested
278
on McCoy cells infected with a 1:2 dilution series of C. trachomatis serovar D, C.
279
trachomatis serovar L2 and C. pneumoniae. The C. trachomatis serovar D infection 9
280
was performed using the DEAE-dextran method; the C. trachomatis serovar L2 were
281
performed without DEAE-dextran; and the C. pneumoniae infections were performed
282
by centrifugation (400 g, 60 minutes, RT). Since both C. trachomatis serovars have a
283
faster developmental cycle, the inclusion counting was performed 24 h p.i., while for
284
C. pneumoniae 48 h p.i. Initially the highest MOI was 8, but between 8 and 1 MOI the
285
infected areas were highly confluent and ChlamyCount could not dissect these areas
286
efficiently. For further experiments the starting MOI was either 1 (C. trachomatis
287
serovars D and L2) or 1:8 (C. pneumoniae) and 6 additional 1:2 dilutions were
288
performed in duplicates. The last two wells contained uninfected McCoy cells (Fig.
289
3A-C). With all three Chlamydia species we could measure a high correlation (R2
290
0.94-0.98) between the measured Chlamydia inclusion count (in the report: particle
291
count) and the theoretical inclusion count calculated from the 1:2 dilution curve. The
292
other two calculated values, namely the total area above the intensity threshold and
293
the area/median area ratios, also highly correlated to the calculated theoretical
294
values. When no centrifugation was used for infection (C. trachomatis serovar D and
295
L2), the inclusion counts were closely followed the theoretical inclusion counts
296
between MOI 1 and MOI 1:8-1:16 resulting in an approximately 1 log dynamic range
297
of the ChlamyCount system. The detected inclusion counts did not change
298
substantially after MOI 1:16, marking the lower threshold of detection. The
299
centrifugation was more efficient and permitted us to use lower MOIs for C.
300
pneumoniae infection (Fig. 3C), however we experienced leakage in about 25-30% of
301
the chambers, hence this infection method was not used for other experiments.
302 303
Measuring the MIC of moxifloxacin, tetracycline and the novel antichlamydial
304
compound PCC00213 on C. trachomatis serovar D growth. We tested whether
305
ChlamyCount was capable of determining the MIC of the known antichlamydial
306
antibiotics moxifloxacin and tetracycline. The C. trachomatis serovar D moxifloxacin
307
MIC was characterized before as 0.03-0.05 µg/ml (14-16). We performed C.
308
trachomatis serovar D infections (MOI 1) in the presence of moxifloxacin with a
309
concentration ranging from 0.25 µg/ml to 0.04 µg/ml. Our experiments showed (Fig.
310
4A), that C. trachomatis could grow until 0.015 µg/ml moxifloxacin concentration, but
311
was inhibited at a concentration of 0.031 µg/ml resulting in a MIC value of 0.031
312
µg/ml. Tetracyclines are first-choice antibiotics for chlamydial infections (5). The 10
313
tetracycline and doxycycline MICs for C. trachomatis serovar D was previously
314
characterized as 0.03-0.15 µg/ml (14, 17, 18). We performed C. trachomatis serovar
315
D infections (MOI 1) in the presence of tetracycline with a concentration ranging from
316
0.04 µg/ml to 0.0006 µg/ml. Our experiments revealed (Fig. 4B) that C. trachomatis
317
could grow until 0.01 µg/ml tetracycline concentration, but was inhibited at a
318
concentration of 0.02 µg/ml resulting in a MIC value of 0.02 µg/ml. We also used
319
ChlamyCount to determine the MIC of the novel antichlamydial compound
320
PCC00213. As Figure 4C shows, C. trachomatis could grow until 3.1 µg/ml
321
concentration of PCC00213 but was inhibited at a concentration of 6.2 µg/ml,
322
resulting in a MIC value of 6.2 µg/ml. We have to note that the parallel MTT-based
323
host cell viability assays showed that the antichlamydial effect of PCC00213 was
324
partially due to the inhibition of host cell metabolism (data not shown). Parallel to the
325
ChlamyCount based MIC determination we investigated the same slides with
326
fluorescent microscopy and determined the inclusion counts in each chambers. As
327
Figure 4A-C shows, the absolute inclusion counts were generally higher when we
328
applied ChlamyCount, but there was a high correlation between the two inclusion
329
counts (R2 0.94-0.98) (Fig. 4D). Importantly, the MIC values determined by the
330
ChlamyCount and manual methods were identical for all the three tested compounds.
331 332
Measuring the effect of IFN-γ on C. trachomatis serovar D and DEAE-dextran
333
and cycloheximide on C. trachomatis serovar D and on C. pneumoniae growth.
334
The inhibitory activity of IFN-γ on chlamydial growth in human host cells via the
335
degradation of host tryptophan pool is a well known phenomenon (19). Byrne et al.
336
found that IFN-γ had a concentration dependent inhibitory activity on C. psittaci
337
growth in the human uroepithelial cell line T24, resulting an approximately 10-fold
338
reduction of direct inclusion counts at 20 ng/ml IFN-γ concentration. We performed C.
339
trachomatis infection (MOI 1) of McCoy murine fibroblastoid cells in the presence of
340
murine and human IFN-γ. Despite the fact that the host species, cell line and
341
chlamydial species were different, our experiments showed comparable extent of
342
inhibition albeit at lower murine IFN-γ concentration: inhibition of chlamydial
343
propagation was concentration dependent and the maximum inhibition was
344
approximately 3.8 fold at a concentration of 1.5 IU/ml (approximately 0.07 ng/ml) or
345
higher murine IFN-γ concentration (Fig. 5A-B). The control human IFN-γ did not show 11
346
any inhibitory effect even at a concentration of 100 IU/ml (Fig. 5A). It was an early
347
observation that the pretreatment of host cells with DEAE-dextran or treatment with
348
cycloheximide could increase the number of chlamydial inclusions and the
349
recoverable IFU (9, 20-22). We applied ChlamyCount to detect these effects during
350
C. trachomatis serovar D and C. pneumoniae infection. We pretreated HeLa cells
351
with DEAE-dextran (1% DEAE-dextran, 15 minutes, RT) and/or applied 1 µg/ml
352
cycloheximide during the infection and compared the direct inclusion counts to the
353
untreated cells. Our results showed partially different effects of these drugs on the
354
growth of the two Chlamydia species (Fig. 5C-D). For C. trachomatis serovar D the
355
application of DEAE-dextran showed only marginal effect, but the cycloheximide
356
treatment largely independently of the presence of DEAE-dextran increased the
357
direct inclusion count 1.9-2.2 fold. For C. pneumoniae the application of dextran or
358
cycloheximide alone increased the direct inclusion count 2.4-3.3 fold respectively, but
359
the co-addition of the two drugs did not show a further growth promoting effect.
360 361 362
Discussion
363
We
364
designed
a low-cost,
medium-throughput
method
for the
rapid
365
enumeration of chlamydial inclusions. Chlamydial inclusions on a 16-well chamber
366
slide were labeled by a fluorescently labeled genus-specific antibody and scanned by
367
a commercial DNA chip scanner. In this detection system, the DNA chip scanner is
368
the most expensive component, however these scanners are easily available in core
369
facilities, and in many cases the new generation sequencing technology makes these
370
scanners infrequently used or redundant. Our technology reuses these scanners in a
371
novel role, when the high resolution images produced by these scanners are used to
372
visualize chlamydial inclusions. The images are processed either completely
373
automatically or after a short intensity and area threshold adjustments on a single
374
desktop computer with an average or low-average year 2013 hardware configuration
375
(Intel Core2 CPU
[email protected] GHz, 2 GB RAM, ATI Radeon HD 3600 Series video
376
card).
377
The golden standard test of inclusion counting is the infection of host cells with
378
serial dilutions of Chlamydia and the subsequent counting of inclusions. Our method
379
was capable of counting the 1:2 dilutions of inclusions of three different chlamydial 12
380
species, with high correlation with the theoretical estimates. The system was capable
381
of measuring the inclusion counts over a 1 log range which was comparable or better
382
than the previously described other methods (12, 23-25). As with other methods, a
383
limitation of the ChlamyCount method at higher MOIs is the accurate dissection of the
384
high number of confluent or nearly confluent fluorescent areas originated from close
385
inclusions. Also, ChlamyCount detects a higher number of inclusions than the manual
386
counting by microscopy likely because the software is detecting smaller fluorescent
387
areas as inclusions. Therefore the current version of ChlamyCount may not be used
388
for absolute quantitation of inclusions. ChlamyCount was designed to measure the
389
effect of various treatments on chlamydial growth and for this task it is enough to
390
follow the changes in inclusion counts with high accuracy and is not necessary to
391
determine their absolute number.
392
Indeed, we could demonstrate that inclusion counts detected by ChlamyCount
393
and manual microscopy closely correlated and therefore could be applied for tasks
394
where the detection of changes in bacterial (inclusion) counts are important such as
395
MIC determination. We used ChlamyCount to determine the MIC of two well-
396
characterized but different action antichlamydial antibiotics, the ribosome inhibitor
397
tetracycline and the gyrase inhibitor moxifloxacin. In both cases ChlamyCount was
398
able to determine MIC values that were identical to the MIC values determined by
399
manual microscopy and also close to the previously described values. ChlamyCount
400
was also able to reproducibly determine the MIC value of the novel antichlamydial
401
compound PCC00213. Besides antibiotics, various chemicals can effect chlamydial
402
growth in a positive or negative manner. ChlamyCount was also able to determine
403
the previously described inhibitory effect of IFN-γ, and growth promoting effect of
404
DEAE-dextran and cycloheximide. Importantly, these data shows that ChlamyCount
405
can be used to quantitatively measure the fold changes in inclusion count between a
406
treated and control samples. This type of relative quantitation makes our method
407
applicable in chlamydia basic biology experiments, where the effect of a given
408
treatment should be quantitatively measured.
409
Considering the previously described methods, the rapid estimation of
410
chlamydial growth can be achieved via two approaches. The first approach uses
411
either a fluorimeter or spectrophotometer to measure the total intensity of Chlamydia
412
specific fluorescently labeled-antibody (26) or indirectly measure the Chlamydia
413
growth by measuring decreased host cell metabolism after chlamydia induced lysis 13
414
(25). These methods are rapid, inexpensive, but do not rely on counting the individual
415
inclusions, hence the aspecific antibody binding or aspecific change in host cell
416
metabolism cannot be excluded. The second approach mostly (12, 24) but not
417
exclusively (23) uses automatic microscopes to take a certain number of images/well,
418
followed by computer image analysis for the specific detection and enumeration of
419
inclusions. ChlamyCount relates to these methods. Compared to the recently
420
described automatic microscope-based inclusion counting method (12) our method
421
has certain advantages and disadvantages. The automatic microscope-based
422
method uses 96-well plates and obviously produces images with a higher resolution.
423
Since analyzing the images requires a significant computational power, the image
424
analysis is performed by a computer cluster with 16 processors. In contrast our
425
method has a lower throughput, but the analysis time is significantly shorter,
426
therefore the processing time of 96 samples (6x16-well chambers) is comparable for
427
at least the first 96 samples. ChlamyCount system set-up cost is generally lower, the
428
required computational support is significantly simpler. Albeit at lower resolution,
429
ChlamyCount also preserves a major advantage of the automatic microscope-based
430
methods; namely it provides topological information about the inclusions. Since DNA
431
chip scanners can scan at two different wavelengths, our method can potentially be
432
applied to provide colocalization information, e.g. allowing testing the effect of anti- or
433
pro-chlamydial proteins recombinantly expressed in the host cells.
434
In conclusion, we developed an easily useable, accurate system for measuring
435
the antichlamydial effect of known and novel antibiotics and for measuring the effect
436
of various compounds on Chlamydial growth. We think that ChlamyCount has the
437
potential to be further optimized, an extended dynamic range and absolute inclusion
438
number estimation could be achieved by applying new raw image processing
439
methods, and improved confluent area dissection algorithm, the goals we are
440
currently pursuing.
441
14
442
Acknowledgments
443
This work was supported by the ERA-NET ChlamyTrans grant (D.P. Virok), OTKA
444
National Research Fund grant PD 100442 (K. Burián), TÁMOP-4.2.2.A-11/1/KONV-
445
2012-0073 (P. Balázs) and TÁMOP-4.2.2.A-11-1-KONV-2012-0035.
446 447
Figure Legends
448
FIG. 1
449
Comparison of the images produced by a DNA chip scanner and confocal microscopy (A) C.
450
trachomatis serovar D infected McCoy cells grow in a 16-well plastic chamber attached to a
451
glass slide. After fixation chlamydial inclusions are fluorescently labeled and scanned by a
452
DNA chip scanner at 5 µM resolution. Scanned image of a well of the 16-well chamber slide.
453
(B) Magnified portion of the scanned image. (C) The same area visualized by fluorescent
454
confocal laser scanning microscopy. Bar=100µm. (D) Further magnification of the lower right
455
fluorescent structure reveals a single infected cell. Bar=10µm.
456 457
FIG. 2
458
ChlamyCount image adjustments and report. (A) Infected host cells are grown in a 16-well
459
chamber slide. Chlamydial inclusions are stained by direct immunofluorescence and scanned
460
by a DNA chip scanner. Before image analysis, the user can adjust the inclusion intensity
461
and area thresholds for detection. (B) The effect of the applied threshold changes on the
462
number and location of the detected inclusions can be readily checked by a magnified
463
quadrant of the left uppermost and lowest wells of the chamber. Based on the applied
464
thresholds ChlamyCount performs the image analysis. (C) ChlamyCount reports the
465
numerical data as a .txt, .xls and .pdf files. The .pdf file contains the numerical data as well
466
as the images of the 16 scanned areas.
467 468
FIG. 3
469
Detection of chlamydial inclusions. McCoy cells were infected with serially diluted (A) C.
470
trachomatis serovar D, (B) C. trachomatis serovar L2 and (C) C. pneumoniae. C.
471
pneumoniae infections were performed by centrifugation (400 g, 60 min, RT). Each infection
472
with a particular MOI was performed in parallel wells. MOIs are shown as simple fractions
473
instead of decimal numbers in order to follow the serial dilutions easier. The last two wells
474
were uninfected. The images of scanned and ChlamyCount processed wells and the
475
numerical data are shown for each species and serovar. For an easier comparison of
476
theoretical and measured inclusion counts, the first theoretical inclusion count was made the
15
477
same as the first measured inclusion count. Data are means +/- standard deviations of the
478
parallel wells.
479 480
FIG. 4
481
MIC estimation of known and novel antimicrobial compounds. McCoy cells were infected with
482
C. trachomatis serovar D (MOI 1) in the presence of various concentrations of (A)
483
moxifloxacin, (B) tetracycline and (C) PCC00213. Each infection with a particular antibiotics
484
concentration was performed using parallel wells. The inclusion counts were enumerated by
485
ChlamyCount and manual confocal microscopy on the same slide. The scanned and
486
ChlamyCount processed well images and the numerical data are shown. Data are means +/-
487
standard deviations for the parallel wells. (D) Correlation between the inclusion numbers
488
enumerated by ChlamyCount method and manual counting. Each data point represents the
489
inclusion number detected in a single well of the 16-well chamber slide.
490 491
FIG. 5
492
Measuring the effect of IFN- , DEAE-dextran and cycloheximide on the direct inclusion
493
counts of C. trachomatis serovar D and C. pneumoniae. McCoy cells were infected with C.
494
trachomatis serovar D (MOI 1) in the presence of various concentrations of (A) human and
495
(A, B) murine IFN- . HeLa cells were infected with (C) C. trachomatis serovar D (MOI 0.5)
496
and (D) C. pneumoniae (MOI 1) after pretreatment with 1% DEAE-dextran (15 minutes, RT)
497
and/or applied 1 µg/ml cycloheximide during the infection. Each infection with a particular
498
compound concentration was performed in two parallel wells. The scanned and
499
ChlamyCount processed well images and the numerical data are shown. Data are means +/-
500
standard deviations for the parallel wells. Student’s t-test was applied for statistical
501
comparison of the treated samples vs. the non treated. * P