A total of 64 bacterial strains were used in this work to develop the LAMP assay (Supplemental Materials). K. pneumoniae ATCC BAA-2146 carrying bla_NDM−1 and K. pneumoniae ATCC BAA-1705 carrying bla_KPC−2 were used as the positive control. Twenty-one non- K. pneumoniae bacterial species maintained in our microorganism center, including common clinical infectious and opportunistic pathogens, were included to evaluate the specificity of the LAMP assay. One hundred and ten clinical sputum samples containing suspected K. pneumoniae strains and multi-resistant infections were collected from ICU hospitalized patients with cough or pneumonia in the Wujing hospital, 307 hospital and 301 hospital in Beijing, China. The species identification was carried out using an automated system (Phoenix and BD systems) and matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS). The sequences of 16S ribosomal DNA (rDNA) and rcsA were validated by PCR-based sequencing and showed 100% identity with the sequences of previously reported genes.

Seven housekeeper genes including gapA, infB, mdh, pgi, phoE, ropB, and tonB were detected by PCR. The allele number for each gene was assigned to the MLST database (http://bigsdb.web.pasteur.fr/klebsiella/klebsiella.html). A combination of the allelic sequences of the 7 genes yielded the allelic profile. Antimicrobial susceptibility testing was performed by microbroth dilution VITECaccording to the Clinical and Laboratory Standards Institute (CLSI, Performance standards for antimicrobial susceptibility testing; Twenty-third informational supplement CLSI Document M100-S23. Wayne, PA, USA 2013) and Etest strips (bioMérieux) for carbapenems.

The strains were screened for the presence of known MBL and other β-lactamase genes (bla_NDM−1, bla_KPC−2, bla_TEM, bla_VIM, bla_IMP, bla_CTX, bla_SIM−1, bla_AIM−1, and bla_OXA−48) by PCR with primers as reported previously (Poirel et al., 2007; Patzer et al., 2009).

The bacterial strains and clinical samples were cultured in brain heart infusion (BHI) broth at 37°C according to a standard protocol. Genomic DNA was extracted using the Wizard Genomic DNA Purification Kit (Promega Co. USA).

The sequence of the rcsA gene (http://www.ncbi.nlm.nih.gov/gene/7946097) was downloaded from NCBI GenBank database and further analyzed by Primer Explorer (Version 4, http://primerexplorer.jp/elamp4.0.0/index.html). Five primer sets were designed (Table 1 and Table S2). To compare the sensitivity of the LAMP and conventional PCR assay, PCR was conducted with the KP-27F3 and KP-27B3 primer pair (Table 1), which amplifies a 176-bp fragment. All primers were synthesized commercially (Sangon Biotech Co., Ltd, Shanghai, China).

The LAMP reactions were performed in a final volume of 25 μl containing 12.5 μl reaction mixture, 1 μl Bst DNA polymerase, 2 μl template by using the Loopamp DNA Amplification kit (Loopamp DNA Amplification Kit; Eiken Chemical Co., Ltd, Tochigi, Japan) for real-time turbidimeter and another 1 μl calcein/Mn²⁺ solution (Eiken Chemical Co., Ltd) for visual detection. Primers were used at a concentration of 40 pmol for FIP and BIP, 20 pmol for LB and LF, and 5 pmol for F3 and B3. Finally, the reaction mix was overlaid with the protectant to prevent cross-contamination of samples by aerosol and the reactions were performed in the reaction tubes (Eiken Chemical Co. Ltd.) for 60 min at 61°C. During the amplification, the protectant could melted to liquid without disturbing the reaction and solidified as the temperature in the tubes decreased to room time (Patent: ZL201210371448.5 in china).

Two different methods, based on sample turbidity or fluorescence, were used to detect the LAMP products. Real-time changes in turbidity were monitored with spectrophotometric analysis by recording the optical density (650 nm) every 6 s with a Loopamp Realtime Turbidimeter (LA-320c; Eiken Chemical Co., Ltd). For direct visual detection, 1 μl of calcein/Mn²⁺ fluorescent detection reagent was added to the reaction. LAMP amplification results in a green fluorescent emission as a result of magnesium ions forming a complex with calcein. The color change from orange to green when samples are positive is visible to the naked eye under natural light or with the aid of UV light (Tomita et al., 2008). Each experiment was performed at least three times.

A 25 μl reaction volume was used for all PCRs, with mixtures that contained the following components: 12.5 μl of PCR MasterMix reagents (Tiangen Biotech Co., Ltd, Beijing, China), 9.5 μl of double-distilled water, 1 μM KP-27F3 and KP-27B3 primers, and the same amount of DNA template as was used in the LAMP reaction. The PCR cycling parameters were: initial PCR activation, 95°C for 5 min; amplification, 30 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 30 s; final extension, 72°C for 10 min. The products were separated with 1% agarose gel (Amresco) electrophoresis and stained with ethidium bromide. Images were documented with a Gel Doc EQ imaging system (Bio-Rad).

Five sets of primers were initially tested to detect K. pneumoniae. All five sets amplified the target sequence under the same reaction conditions. The KP-27 primer set began to amplify the target gene in the shortest time (Figure 1) and was therefore chosen as the optimal primer set for K. pneumoniae detection with LAMP (Table 1).

Reaction temperatures ranging from 59°C to 69°C at 2°C intervals were compared for optimal amplification. As shown in Figure 2, the most suitable reaction temperature range was 59–63°C. Finally, we chose 61°C as the optimal reaction temperature.

K. pneumoniae ATCC BAA-2146 was used as the positive control and double-distilled water as the negative control when evaluating the specificity of LAMP for the detection of K. pneumoniae. Twenty-one other non- K. pneumoniae bacterial strains (Supplemental Materials) were also tested. As shown in Figure 3, both methods of analysis positively identified the K. pneumoniae. All other strains, including the blank control, tested negative, indicating that the LAMP assay was specific to K. pneumoniae.

To compare the detection limit of LAMP using either real-time turbidity measurements or visual color change with traditional PCR, pure genomic DNA was extracted from K. pneumoniae ATCC BAA-2146 and subjected to serial 10-fold dilution, from 115.0 ng/μl to 0.0115 pg/μl. As shown in Figure 4, the detection limit of the real-time turbidity and visual detection was both 0.115 pg/μl, which was 100-fold more sensitive than traditional PCR assay.

One hundred and ten clinical sputum samples were collected for LAMP-based surveillance of K. pneumoniae from ICU patients with clinically suspected multi-resistant infections in three top hospitals of Beijing. Ten sputum samples from healthy people were collected as controls. Both LAMP and PCR assay were involved to analyze the clinical samples. As shown in Figure 5, of the 110 clinical samples, LAMP detected 32 positive samples while 25 were detected by PCR. Then, 32 K. pneumoniae strains were successfully cultured from these positive samples. The healthy control samples all tested negative in each of the assays. The sequence analysis of the rcsA genes from K. pneumoniae isolates confirmed conservation with the nucleotide sequences of reported gene.

MLST analysis showed that the 32 K. pneumoniae strains belonged to different sequence type (ST) including ST11, ST21, ST30, ST37, ST40, ST84, ST104, ST147 or ST322, respectively (Supplemental Materials). To further characterize the clinical K. pneumoniae isolates, PCR screening of MBL and other β-lactamase genes was performed and antimicrobial susceptibility tested, the results showed a co-occurrence of resistance genes in most of the K. pneumoniae clinical isolates. The positive PCR products were sequenced and the result showed 100% identities with previously reported genes. It is interesting to note that 18 of the 32 K. pneumoniae strains contain the bla_KPC−2 gene, resulting in increased resistance to β-lactams antibiotics (Supplemental Materials). Moreover, the isolate named K. pneumoniae WJ-64 simultaneously contained bla_KPC−2 and bla_NDM−1which has emerged in recent years and attracted wide attention, presented increased resistance to all β-lactams (MIC > 128μg/ml for meropenem and imipenem), cephalosporins and aminoglycosides, and was only susceptible to tigecycline. The isolate was also positive for imipenem-EDTA double-disk synergy test (DDST) and modified Hodge test (MHT).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.