AAC Accepts, published online ahead of print on 22 December 2014 Antimicrob. Agents Chemother. doi:10.1128/AAC.04058-14 Copyright © 2014, American Society for Microbiology. All Rights Reserved.
1
Molecular diagnosis of fluoroquinolone resistance in Mycobacterium tuberculosis: opening
2
Pandora’s Box
3
Christine Bernard,a,b,c Nicolas Veziris,a,b,c Florence Brossier,a,b,c Wladimir Sougakoff,a,b,c
4
Vincent Jarlier,a,b,c Jérôme Robert,a,b,c Alexandra Aubrya,b,c#
5 6
Running head: Fluoroquinolone resistance in M. tuberculosis
7 8
Sorbonne Universités, UPMC Univ Paris 06, CR7, Centre d’Immunologie et des Maladies
9
Infectieuses, CIMI, team E13 (Bacteriology), Paris, Francea ; INSERM, U1135, Centre
10
d’Immunologie et des Maladies Infectieuses, CIMI, team E13 (Bacteriology), Paris, Franceb ;
11
AP-HP, Hôpital Pitié-Salpêtrière, Centre National de Référence des Mycobactéries et de la
12
Résistance des Mycobactéries aux Antituberculeux, Bactériologie-Hygiène, Paris, Francec
13 14
#Address correspondence to: Alexandra Aubry,
[email protected] 15 16 17 18 19 20 21 22 23 24
Keywords: DNA gyrase, extensively drug-resistant tuberculosis, drug susceptibility testing
25
Abstract
26
Objectives: As a consequence of the use of fluoroquinolones (FQ), resistance to FQ has
27
emerged, leading to nearly untreatable, extensively drug-resistant tuberculosis. Mutations in
28
DNA gyrase represent the main mechanism of FQ resistance. A full understanding of the
29
pattern of mutations found in FQ-resistant (FQ-R) clinical isolates, and of their proportions, is
30
crucial for improving molecular methods for the detection of FQ resistance in Mycobacterium
31
tuberculosis. Wwe reviewed the detection of FQ resistance in isolates addressed to the French
32
National Reference Center for Mycobacteria from 2007 to 2012 with the aim to evaluate the
33
performance of PCR-sequencing in a real life context.
34
Methods: gyrA and gyrB sequencing performed prospectively on M. tuberculosis clinical
35
isolates was compared to FQ susceptibility to ofloxacin 2mg/L by the reference proportion
36
method.
37
Results: A total of 605 isolates, of which 50% were multidrug-resistant, were analysed. The
38
increase in FQ-R strains among MDR strains during the time study was alarming (8% to 30%).
39
The majority (78%) of isolates with gyrA mutations were FQ-R whereas only 36% of those
40
with gyrB mutations were FQ-R. Only 12% of the FQ-R isolates had a single mutation in gyrB.
41
Combined gyrA and gyrB sequencing led to greater than 93% sensitivity in resistance detection.
42
The analysis of the four false positive and the five false negative results of the gyrA and gyrB
43
sequencing illustrated the actual limitations of the reference proportion method.
44
Conclusions: Our data emphasize the need for combined gyrA and gyrB sequencing in the
45
investigation of FQ susceptibility in M. tuberculosis and challenge the validity of the current
46
phenotype-based approach as diagnostic gold-standard for FQ resistance.
47 48 49
50
Introduction
51
Tuberculosis (TB) remains one of the most serious infectious diseases worldwide causing 1.8
52
million deaths each year (1). The emergence and transmission of drug-resistant Mycobacterium
53
tuberculosis strains further threatens TB control efforts. Treatment of infection due to
54
multidrug-resistant (MDR) M. tuberculosis (i.e. resistant to isoniazid and rifampin, MDR-TB)
55
requires the use of fluoroquinolones (FQ) since it has been correlated with good prognosis (2).
56
Unfortunately, the extensive use of FQ has led to the emergence of extensively drug-resistant
57
(XDR) isolates, defined as MDR with resistance to any FQ and at least one of the three
58
injectable second-line drugs (amikacin, kanamycin or capreomycin). Disease caused by XDR
59
M. tuberculosis is associated with very poor treatment outcomes close to the historical
60
outcomes when no chemotherapy for TB was available (3).
61
FQ are antibiotics with broad-spectrum antimicrobial activity and are therefore widely used for
62
the treatment of bacterial infections of the respiratory, gastrointestinal and urinary tracts, as
63
well as of sexually transmitted diseases and chronic osteomyelitis (4). FQ resistance rates in TB
64
patients have been estimated to lie between 0.15 and 30% depending on the country (5-8).
65
Rates are higher among patients exposed to FQ prior to the diagnosis of TB, i.e. in cases of
66
“acquired” resistance (6). However, in some countries, especially where FQ consumption is
67
very high, FQ resistance is often due to transmission of already FQ-resistant (FQ-R) strains
68
(7,8).
69
FQ resistance is defined by the WHO as resistance to 2mg/l of ofloxacin (OFX) (9).
70
Conventional methods for drug susceptibility testing (DST) involve primary culture of
71
specimens and isolation of M. tuberculosis. This process has a long turnaround time of several
72
weeks. Over the past several years, molecular techniques have been developed for rapid
73
detection of resistance to antituberculous agents (10). DNA gyrase, the only target of FQ in
74
M. tuberculosis, is a GyrA2GyrB2 tetrameric enzyme containing two A and two B subunits
75
encoded by the gyrA and gyrB genes, respectively. Mutations within the highly conserved
76
region, the so-called quinolone resistance-determining region (QRDR) of gyrA and/or gyrB
77
have been reported to be responsible for at least 70% of FQ resistance in M. tuberculosis,
78
affecting most commonly the GyrA subunit and less frequently also the GyrB subunit (11). The
79
vast majority of published studies are retrospective and thus the molecular diagnosis was made
80
for strains already known to be fluoroquinolone-resistant or -susceptible. To test the potential
81
of molecular diagnosis in a real life context, we conducted a 6-year prospective study using
82
combined gyrA and gyrB sequencing for initial evaluation of fluoroquinolone susceptibility.
83 84
Materials and methods
85
Bacterial strains and selection of isolates
86
This prospective study included all isolates for which a genotypic and phenotypic diagnosis of
87
FQ resistance was established between 2007 and 2012 at the French National Reference Center
88
for Mycobacteria (NRC). These isolates comprised (i) all MDR strains isolated in France and
89
(ii) some isolates that had been sent to the NRC because of high suspicion of being MDR (and
90
that had been tested for second-line drug susceptibility immediately after isolation but
91
eventually turned out to be resistant to isoniazid or rifampin only, or even susceptible to both
92
drugs) and (iii) some non-MDR isolates from patients for whom treatment with FQ was
93
considered because of intolerance or resistance to antituberculous drugs.
94
Drug susceptibility testing
95
In-vitro DST was performed on Löwenstein-Jensen (LJ) medium following the proportion
96
method (12), with rifampin (40mg/l), isoniazid (0.2 and 1mg/l), OFX (2mg/l), amikacin and
97
capreomycin (40 mg/l) and kanamycin (30 mg/l), and the critical proportion of 1% (9).
98
DNA sequencing of quinolone resistance-associated genes
99
DNA sequencing of gyrA and gyrB was performed using primary culture of all isolates.
100
Genomic DNA was isolated from bacteria grown on LJ medium. A loopful of culture was
101
suspended in water (500 µl) and heated at 95°C for 15 min. The DNA used for PCR
102
amplification was obtained by heat shock extraction (1 min at 95°C and 1 min on ice, repeated
103
five times). A volume of 5 µl was used in PCRs with the oligonucleotide primers described
104
below. For FQ resistance, the QRDRs of gyrA and gyrB were amplified and sequenced using
105
primers PRI8 (5’-YGGTGGRTCRTTRCCYGGCGA-3’) and PRI9 (5’-
106
CGCCGCGTGCTSTATGCRATG-3’) for gyrA and primers gyrBa (5’-
107
GAGTTGGTGCGGCGTAAGAGC-3’) and gyrBe (5’-CGGCCATCAAGCACGATCTTG-3’)
108
for gyrB (13).
109
After amplification, unincorporated nucleotides and primers were removed by filtration with
110
Microcon® 100 microconcentrators (Amicon Inc., Beverly, MA), and the amplicons were
111
sequenced using the Big Dye® Terminator v3.1 Cycle Sequencing Kit (Life Technologies)
112
following the manufacturer’s instructions.
113
The amino acid substitution S95T in GyrA was omitted from the summary tables because it
114
corresponds to a well-known polymorphism that does not confer drug resistance (11).
115
When more than one substitution was observed in one strain (double or triple substitution), we
116
noted two scenarios: (i) either substitution was observed as a single substitution elsewhere, or
117
(ii) the substitutions were never observed independent of one another. In both scenarios, the
118
substitutions were listed as single substitutions, and as multiple substitutions. Mutations that
119
were never observed independent of one another are noted in the tables. All substitutions in
120
GyrA and GyrB were annotated according to the numbering system of M. tuberculosis where
121
the QRDR of M. tuberculosis GyrA ranges from codon 74 to 113 and the QRDR of GyrB from
122
codon 500 to 540 (the most frequently GyrB numbering system used in the literature
123
(CAB02426.1)) (11, 14). Correspondences with the proposed numbering system are given as
124
footnotes in Tables 3 and 4 (11). Implication of DNA gyrase substitutions in FQ-resistance
125
noted in tables correspond to results of biochemical studies found in the literature whose
126
references are noted in the tables.
127
Statistics tests.
128
Chi2 tests were performed using BiostaTGV website (http://marne.u707.jussieu.fr/biostatgv/).
129
Results were considered significant for p-value ˂ 0.05.
130 131
Results
132
Drug susceptibility pattern of the isolates included in the study (Table 1)
133
A total of 605 M. tuberculosis isolates obtained from clinical specimens were sent to the NRC
134
during 2007-2012 for diagnosis of FQ resistance. The number of isolates studied rose every
135
year, from 71 in 2007 to 137 in 2012, but the proportion of MDR isolates remained the same
136
(ca. 50%) (Table 1). Most of the isolates (78%) were resistant to isoniazid or rifampin or both:
137
297 (49%) were resistant to both isoniazid and rifampin (i.e. MDR or XDR), and 176 (29%)
138
were resistant either to rifampin (n=25) or isoniazid (n=151). Phenotypic OFX resistance was
139
found in 70 (12%) of the 605 isolates. Among the OFX-resistant (OFX-R) isolates, most (66
140
[94%] of 70) were MDR, whereas one was resistant to isoniazid but susceptible to rifampin,
141
two were resistant to rifampin but susceptible to isoniazid, and one was susceptible to both
142
isoniazid and rifampin.
143
The percentage of OFX-R isolates included in the study rose significantly, from 4% in 2007 to
144
18% in 2012 (p=0.003), mainly due to the increase in OFX-R isolates among MDR isolates
145
(from 8% in 2007 to 30% in 2012).
146
Amino acid substitutions in GyrA
147
Eighty-one substitutions in the QRDR of GyrA were observed in 74 (12%) of the isolates
148
(Table 2). Out of these, 63 isolates harboring 69 substitutions were MDR (including 24 XDR),
149
seven isolates harboring seven substitutions were resistant to either rifampin or isoniazid, and
150
four isolates harboring five substitutions were susceptible to both isoniazid and rifampin. The
151
amino acid substitutions were located at six different positions: T80, G88, A90, S91, D94, and
152
Q101 (Table 2). The most common substitution occurred at position D94 (n=37, i.e. 46%) and
153
resulted in 13 D94G, 13 D94A, four D94N, four D94H, and three D94Y substitutions. The
154
second most common substitution occurred at position A90 (n=21) and resulted in 18 A90V
155
substitutions, two A90G substitutions, and one A90E substitution. The latter two were
156
systematically associated with other substitutions in GyrA (Table 3). The third most common
157
substitution was T80A (n=16). The remaining substitutions were S91P (n=5), G88C (n=1) and
158
Q101E (n=1).
159
Seven isolates had double substitutions in GyrA: three harbored T80A + A90E/G substitutions,
160
and four harbored A90V + D94G/H/N/Y substitutions (Table 3).
161
Amino acid substitutions in GyrB
162
Substitutions in the QRDR of GyrB were observed in 22 (4%) of the 605 isolates (Table 5). Out
163
of these 22 isolates, 14 were MDR (including seven XDR), six were resistant to either rifampin
164
or isoniazid and two were susceptible to both isoniazid and rifampin. The amino acid
165
substitutions were located at 11 positions. No substitution was significantly more frequent than
166
any other: four G559A, four E540V, three S486F/Y, two P478A, two D500A/N, and two
167
A543V substitutions and one S470I, one D473N, one A506G, one A547V and one G551R
168
substitution (Table 4).
169
As opposed to what was observed for GyrA, no isolate harbored more than one substitution in
170
GyrB. However, one isolate showed a substitution in both GyrA (D94A) and GyrB (A543V)
171
(Table 3).
172
Distribution of DNA gyrase mutations regarding the susceptibility to FQ
173
Among the isolates included in this study, 535 were OFX-susceptible (OFX-S) and 70 were
174
OFX-R (Table 1). The distribution of substitutions in GyrA and GyrB according to the FQ
175
sensitivity of the isolates is detailed in Tables 2 and 4. Among the 74 isolates with
176
substitution(s) in GyrA, 16 (harboring 18 substitutions) were OFX-S whereas the majority were
177
OFX-R (n= 58 [78%], harboring 63 substitutions). Among the 22 isolates with substitutions in
178
GyrB, more isolates were OFX-S (n=14 [64%], harboring 14 substitutions) than OFX-R (n=8
179
[36%], harboring eight substitutions). Among the latter, one isolate harbored substitution in
180
both GyrA (D94A) and GyrB (A543V). The proportion of substitutions in GyrB associated
181
with FQ resistance was significantly lower than in GyrA (36% vs 78%, p=0.0002).
182
Surprisingly, no substitution in GyrA or GyrB was found in five of the 70 (7%) OFX-R
183
isolates, mainly MDR isolates (n=4). For these isolates, the proportion of colonies growing on
184
OFX containing LJ medium was close to the breakpoint given by the proportion method for
185
four isolates (1%, 1%, 2%, and 5% of mutant), and higher for one isolate (proportion of
186
colonies growing on OFX containing LJ medium, 25%). Sequencing of gyrA and gyrB was
187
therefore also performed on these colonies growing on OFX-containing LJ medium and
188
substitutions were found in four isolates: GyrA D94G/Y substitutions in three isolates, and
189
GyrB N538K substitution in one isolate (Table 4). No substitutions in GyrA or GyrB were
190
found in the remaining isolate although it harbored the highest proportion of colonies growing
191
on OFX containing LJ medium (25%).
192
On the other hand, 30 OFX-S isolates harbored 18 substitutions in GyrA and/or 14 substitutions
193
in GyrB. Among these 30 strains, two strains harbored substitutions known as implicated in
194
FQ- resistance (GyrB D500A and GyrB E540V) (14,15) and another two strains harbored
195
substitutions whose implication in FQ resistance is not known (GyrA Q101E and GyrB S470I)
196
(Tables 2 and 4).
197
Substitutions implicated in FQ resistance were more frequent in MDR than non-MDR isolates
198
(42 vs. 2%, p