Three‐reaction high‐resolution melting assay for rapid differentiation of Mycobacterium tuberculosis complex members

Abstract The possibility of introducing a reliable assay for a quick identification and differentiation of the main species of Mycobacterium tuberculosis complex (MTBC) supports the improvement of efficient tuberculosis combating strategies worldwide. Commercially available assays are often based on cultured samples; however, due to the long cultivation time of mycobacteria, results are delayed. Developed PCR approaches have been published previously, though, when testing intricate veterinary samples, the complex composition of multiplex qPCRs frequently leads to assay failure. In order to overcome those limits, a paradigm of a three‐reaction high‐resolution melting (HRM) assay for the simultaneous identification and differentiation of the main members of MTBC was established. The assay is based on single nucleotide polymorphisms within gyrB and gyrA, which have been used as target for the establishment of two highly specific HRM assays (HRM assays 1 and 2) discriminating M. tuberculosis/ Mycobacterium canetti, Mycobacterium bovis/M. bovis BCG, Mycobacterium caprae/rare M. caprae/M. bovis ecotypes, Mycobacterium africanum/Mycobacterium orygis/ Mycobacterium pinnipedii/Clade A1, Mycobacterium microti, and a rare subtype of M. canettii followed by a third HRM assay (HRM assay 3) allowing a further differentiation of M. bovis, M. bovis BCG, and a rare subtype of M. caprae/M. bovis, which is considered to be a novel ecotype. High‐resolution melting assay 1 is described in a previously published report. High‐resolution melting assay 2 showed 100% correlation of all 39 examined isolates with the results of a commercial identification kit. 96% of the clinical samples tested demonstrated concordant results. High‐resolution melting assay 3 showed an accordance of 100% with the results of the commercially available identification kit of all 22 samples analyzed. The proposed strategy of the three‐reaction HRM assay can be used for an accurate differentiation of up to seven groups of MTBC and potentially to identify a rare subtype of M. canettii either on isolates or on clinical samples.


Abstract
The possibility of introducing a reliable assay for a quick identification and differentiation of the main species of Mycobacterium tuberculosis complex (MTBC) supports the improvement of efficient tuberculosis combating strategies worldwide. Commercially available assays are often based on cultured samples; however, due to the long cultivation time of mycobacteria, results are delayed. Developed PCR approaches have been published previously, though, when testing intricate veterinary samples, the complex composition of multiplex qPCRs frequently leads to assay failure. In order to overcome those limits, a paradigm of a three-reaction high-resolution melting (HRM) assay for the simultaneous identification and differentiation of the main members of MTBC was established. The assay is based on single nucleotide polymorphisms within gyrB and gyrA, which have been used as target for the establishment of two highly specific High-resolution melting assay 1 is described in a previously published report. Highresolution melting assay 2 showed 100% correlation of all 39 examined isolates with the results of a commercial identification kit. 96% of the clinical samples tested demonstrated concordant results. High-resolution melting assay 3 showed an accordance of 100% with the results of the commercially available identification kit of all 22 samples analyzed. The proposed strategy of the three-reaction HRM assay can be used for an accurate differentiation of up to seven groups of MTBC and potentially to identify a rare subtype of M. canettii either on isolates or on clinical samples.

| INTRODUC TI ON
Tuberculosis is a major cause of human death induced by only one infectious agent resulting in approximately 10 million new infections per year along with about 1.6 million deaths in 2017 (WHO, 2018). Tuberculosis persists as a major health concern not only in humans but also in veterinary medicine.  (Brites et al., 2018). Furthermore, Mycobacterium canettii is a genetically more diverse and recombinogenic organism as observed earlier (Fabre et al., 2010), only leading to opportunistic human infections from time to time (Boritsch et al., 2016;Supply et al., 2013). Although its similarity of the nucleotide codes to the species of MBTC, it is not considered to be part of MTBC (Brites et al., 2018). Mycobacterium canettii is mainly limited to the horn of Africa and most of the known strains were isolated in the Republic of Djibouti (Blouin et al., 2014). Mycobacterium tuberculosis is known to be the major source of human tuberculosis; however, numerous cases of infection with other members of the complex are known. Mycobacterium bovis and more rarely M. caprae are the causative agents for bovine tuberculosis, which is recognized to be an important zoonosis responsible for significant economic loss (Rodriguez-Campos, Smith, Boniotti, & Aranaz, 2014). A recent study (Loiseau et al., 2019)  In regions where tuberculosis is endemic, neonates were vaccinated with the attenuated M. bovis strain BCG. In immunocompromised children, this procedure can cause a disease pattern similar to the one of tuberculosis (Hesseling et al., 2006). In order to evaluate a zoonotic risk of MTBC, it is important to rely on a fast and accurate method capable of identification and differentiation of the species of MTBC leading to improved programs in public health surveillance and enhanced food safety.
High-resolution melting (HRM) approaches are cheap and rapid assays, which are able to detect single nucleotide polymorphism (SNP) according to altered melting temperatures (T m ) of dissociating PCR amplicons (Vossen, Aten, Roos, & Dunnen, 2009). A fluorescent nucleic acid dye is intercalating with the resulting PCR amplicons, which are dissociating upon increase in temperature and thus resulting in a decrease in fluorescence intensity. The determinate of T m is based on its nucleotide sequence, length, and level of GC. The userfriendly single-plex HRM assay can be completed within roughly 2 hr. Moreover, as a major advantage, HRM assays can be performed using samples directly extracted from clinical tissue. High-resolution melting assays have been used to discriminate various bacteria species (Esteves et al., 2018;Jeffery, Gasser, Steer, & Noormohammadi, 2007;Robertson et al., 2009;Stephens, Inman-Bamber, Giffard, & Huygens, 2008;Winchell, Wolff, Tiller, Bowen, & Hoffmaster, 2010) or to analyze antibiotic resistance in M. tuberculosis (Anthwal et al., 2017;Chen et al., 2011;Yadav et al., 2012). Moreover, HRM assays have been established for differentiation of nontuberculous mycobacteria (NTM) and confining them from MTBC (Issa et al., 2014;Khosravi, Hashemzadeh, Hashemi Shahraki, & Teimoori, 2017;Perng et al., 2012). Some studies combined HRM with multiplex qPCR assays targeting at the region of difference (RD) (Pinsky & Banaei, 2008;Pounder et al., 2010).
We have previously reported the design and evaluation of a HRM assay (HRM assay 1) for the identification and differentiation of MTBC into three groups most relevant for veterinarians  another HRM assay (HRM assay 3) based on two SNPs within gyrA and thus completing the three-step paradigm. For most diagnostic applications, however, a combination of HRM assays 1 and 2 will be sufficient.

| Samples and reference strains
Sixty-one samples positive for MTBC were collected from 39 different animals (

| Culture and DNA extraction
Sample preparation, culture, and DNA extraction were conducted as described in a former study (Ghielmetti et al., 2017). GenoType MTBC test (Hain Lifescience), spoligotyping (Ruettger et al., 2012), and multilocus variable number tandem repeat analysis using an internationally established 24-loci panel (Supply et al., 2006) were used for species identification of cultured isolates. Standard biosecurity procedures were followed for handling of samples.

| HRM development
High-resolution melting development of assay 1 was described in a former study . The differentiation of the main members of MTBC is performed applying a paradigm of a three-reaction HRM assay based on four SNPs located on gyrB (base pair positions 675, 756, 1,410, and 1,450 [Niemann et al., 2000] and one SNP on gyrA [base pair position 1,323]). In order to extend the obtained SNP differentiation scheme with the recently described two rare ecotypes of M. caprae/M. bovis (ecotypes I and II), a rare M. canettii subtype (Loiseau et al., 2019) and the animal-adapted MTBC clades (Brites et al., 2018), two additional SNPs on gyrB (base pair positions 1,437 and 1,439), and one SNP on gyrA (base pair position 1,359) were included in the assay design ( Figure 1). were designed for the three-reaction HRM paradigm (Table 2).

| qPCR and melting conditions
Each HRM assay was processed separately on a Rotor-Gene Q system (Qiagen) with the Type-it HRM PCR Kit (Qiagen). The qPCRs were performed as described previously .
High-resolution melting ramping from 76°C to 93°C was applied. Additionally, for HRM assay 3, M. bovis BCG Tice ATCC 27289 was used. As a negative control, ultrapure water was included in each experiment. Rotor-Gene Q Software 2.3.1 (Qiagen) was used for data analysis to generate normalized and difference plots as described in a previous study . To prevent false-negative results possibly deriving from inhibition, clinical samples were analyzed in duplicate undiluted and in the form of a 1:5 dilution. The cultured isolates were analyzed at concentrations between 100 pg and 10 ng. TA B L E 1 MTBC-positive samples used for the development of the HRM assays 2 and 3 To investigate the intra-and interassay variability of the T m , illustrating the repeatability of the developed HRM assays, a randomly chosen subset of 22 cultured isolates and 18 clinical specimens for HRM assays 2 and 9 cultured isolates and 7 clinical specimens for HRM assay 3 were tested, respectively. The variability assays were conducted in triplicates in three single runs at three different days.

| Sensitivity
The analytical sensitivity of HRM assay 2 was measured by tripli-

| HRM of cultured isolates
All cultured isolates tested were amplified successfully yielding a melting curve. The resulting species-specific T m (Tables 3 and 4,   Tables A3 and A4)
In HRM assay 3, efficiencies were determined to be 89% for M. bovis BCG and 103% for M. bovis ( Figure A2).
High-resolution melting assay 2 showed a LOD for the lowest dilution of which the acceptance criteria (standard deviation < 0.5 and

| D ISCUSS I ON
In the present study, the establishment of a three-reaction HRM paradigm in the form of three HRM assays is described, which can which is a benefit comparing to methods, which are based on timeconsuming procedures (Kamerbeek et al., 1997) or require cultured samples (Kasai et al., 2000;Niemann et al., 2000).  Figure 5).
The three HRM assays can be performed either consecutively or in parallel since the qPCR conditions are equal. The interpretation of the three-reaction paradigm is straightforward and simple to achieve. For routine laboratories, a simple combination of HRM assays 1 and 2 will lead to a rapid detection and differentiation of the most significant agents of tuberculosis appearing worldwide.
T m ranges deriving from HRM assays are partially overlapping (Tables 3-6). Therefore, based solely on T m it is not possible to clearly differentiate the members of MTBC. However, by transforming melting curves into normalized and difference plots using algorithms of the Rotor-Gene Q Software 2.3.1 (Qiagen), the members of MTBC can be separated into distinct groups (Figures 2-4). Thereby, the species-specific melting profiles showed an explicit behavior. The  CVs. Since HRM assays 1 and 2 are proven to be 100% specific, they can be used unambiguously for the identification and differentiation of MTBC. Moreover, the assays yielded PCR efficiencies of more than 91% and 89%, respectively. The sensitivity of HRM assay 2 showing a LOD of 10 GE is adequate. The LOD of HRM assay 3 is slightly higher with a measured LOD of 100 GE. However, since the tested collective of samples did not cover species of all eight genotype groups, it is suggested to further evaluate the assay by testing a more extensive collection of isolates. isolates. Several months of cultivation time may be saved by using these potent HRM assays. Since most species within MTBC are implicated in human infections (Huard et al., 2006), it is of advantage to have early knowledge of transmission of tuberculosis for consequently choosing an appropriate drug therapy for humans or a proper eradication strategy when dealing with veterinary samples.

| CON CLUS ION
Tuberculosis surveillance policies and public health management depend on powerful and affordable diagnostic tools such as this paradigm of a three-reaction HRM assay, which could be easily implemented in laboratories worldwide.

ACK N OWLED G M ENTS
We are grateful to the kind provision of a M. microti wild boar isolate from Lucía de Juan Ferré and Beatriz Romero Martínez, European Union Reference Laboratory for Bovine Tuberculosis, Spain.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
PL and SS conceptualized, drafted, and investigated the data; involved in formal analysis; and wrote the original manuscript. MJAS edited and provided sequencing data. RS and SS wrote, reviewed, and edited the manuscript.

E TH I C S S TATEM ENT
The recommendations of Swiss federal regulations (TSV 916.401 and VSFK 817.190) were followed. The animal samples were analyzed in the context of a monitoring program of lymph nodes aiming at an early recognition of bovine tuberculosis and NTM infections.
No animals were killed for the purposes of this research project, and no ethical approval was required.

DATA AVA I L A B I L I T Y S TAT E M E N T
Raw data sets from intra-and interassay variability runs are comprehended in the appendix. On request, additional raw data can be obtained from the corresponding author.

O RCI D
Simone Scherrer https://orcid.org/0000-0001-9548-8798 TA B L E A 3 Raw data set and statistical parameters generated from the intra-and interassay variability of high-resolution melting (HRM) assay 2 using a randomly chosen subset of 22 cultured samples TA B L E A 5 Raw data set and statistical parameters generated from the intra-and interassay variability of high-resolution melting (HRM) assay 2 using a randomly chosen subset of 18 clinical specimens