Genomic Comparison Reveals Natural Occurrence of Clinically Relevant Multidrug-Resistant Extended-Spectrum-β-Lactamase-Producing Escherichia coli Strains

A total of 2,769 cefotaxime-resistant bacteria (CRB) were isolated from 1,535 cattle raised on pasture without antibiotics during their entire life span at two different research facilities (Fig. 1A). The prevalence of CRB in beef cattle was 42.6% on both farms. Of the CRB, 293 isolates from 200 cattle (prevalence = 13.0%) carried either CTX-M or CMY-2 genes confirmed by PCR typing, and 176 isolates were identified as E. coli by using selective media, ChromAgar E. coli, and 16S rRNA genotyping (Fig. 1B). Most of the E. coli isolates carried either a CTX-M (33.5%) or CMY-2 (64.2%) gene, while 4 isolates carried both genes. Of the E. coli isolates, we selected 36 strains (9 CMY-2 positive and 27 CTX-M positive), based on farm location and animal sources, to conduct an antibiotic susceptibility test (AST) to evaluate whether these CRB were multidrug resistant (MDR). All isolates were resistant to cefotaxime (>4 μg/ml), and 83.3% of them were resistant to a clinically important level (>64 μg/ml), as shown by the MICs (Fig. 1C) (18). Thirteen antibiotics belonging to 8 classes, including sulfonamides, aminoglycosides, tetracyclines, fluoroquinolones, chloramphenicol, penicillins, cephalosporins, and polymyxins, were tested. All isolates were resistant to ampicillin, ceftiofur, and cephalothin, but relatively low or no resistance was observed against gentamicin, amikacin, nalidixic acid, or colistin. None of the isolates, including those with intermediate colistin resistance, carried the MCR-1 gene that confers resistance to colistin by modifying lipopolysaccharide (LPS) (19). All isolates were MDR, being resistant against three or more different antibiotic classes, with 10 (27.8%), 6 (16.7%), 7 (19.4%), and 13 (36.1%) isolates resistant to 3, 4, 5, or 6 different antibiotic classes, respectively. In particular, these isolates showed either resistance (10 isolates) or intermediate susceptibility (20 isolates) against amoxicillin-clavulanic acid, a combination of a penicillin class antibiotic and an ESBL inhibitor, indicating that these strains also produce traditional β-lactamase.

To investigate the genetic relationship of these ESBL- and CMY-2-producing MDR E. coli isolates in food-producing animals raised in two different research facilities about 350 kilometers away from each other, whole genomes of the MDR E. coli isolates were sequenced and subjected to phylogenetic analysis. Of these 36 isolates, 3 (JEONG5446, JEONG5776, and JEONG9567) belonged to a single isolate clade according to core-genome-based single-nucleotide polymorphism (SNP) analysis, while the other 33 isolates clustered into 7 multi-isolate clades. Strains clustered into the same clades had limited numbers of SNPs (0 to 30 SNPs), indicating they are clonal variants. Strains were not interspersed between the two farms but clustered together based on farm location, suggesting no transmission occurred between the farms (Fig. 2). Mauve alignment of whole genomes within the same clade were conducted in order to seek genome rearrangement and genome architecture. The alignments revealed that isolates within the same clade had extensive similarity in the chromosomal contents, with minimal genome rearrangement (Fig. S1 in the supplemental material). Based on these data, we concluded that strains in the same clade are clonal variants and selected 11 genetically distinctive ESBL- or CMY-2-producing MDR E. coli strains, belonging to 10 sequence types (STs), from each clade for comparative and functional genomic analyses to understand MDR occurrence.

The 11 representative ESBL- and CMY-2-producing MDR E. coli were identified as 10 different serotypes, with 3 O-antigen nontypeable and 10 unique multilocus sequence types (MLSTs). Except for one strain, all isolates carried Inc group plasmids (Table 1). All genome-sequencing contigs were annotated using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) and submitted to the Comprehensive Antibiotic Resistance Database (CARD) to further identify antibiotic resistance genes (ARGs), resulting in a total of 74 antibiotic resistance-related genes belonging to 19 categories and 9 classes (Fig. 3). All isolates encoded efflux pumps that confer multidrug resistance potential. In addition, we found genes conferring resistance to aminoglycoside, β-lactam, fluoroquinolone, sulfonamide, polymyxin, and peptide antibiotics. This ARG profile supports the results of AST analysis, which described all of the isolates as MDR (Fig. 1C).

To determine whether ESBL (CTX-M-1 and CTX-M-32) and CMY-2 genes were borne by plasmids or chromosomes, we performed PLACNET analysis to assign contigs to either a chromosomal or plasmid network. Contigs carrying the CTX-M or CMY-2 genes were identified (Fig. 4A to C). Contigs with plasmid-encoded replication initiator protein (RIP) and relaxase protein are shown with green and red dots, respectively (Fig. 4A to C). Both CTX-M and CMY-2 genes were found on the plasmid or chromosomal network highlighted with purple dots (Fig. 4A to C). Three strains, JEONG9566, JEONG9567, and JEONG9592, carried CTX-M-32 gene in the chromosomal DNA, while the other 6 isolates carried CTX-M-1/CTX-M-32 or CMY-2 genes on their plasmids. Interestingly, CMY-2 genes were located only within plasmids, whereas CTX-M genes were found within plasmids and chromosomes. Additionally, plasmid replication types were identified using PlasmidFinder. Various types of Inc plasmids, including IncA/C₂, IncFIA, IncFIB, IncFII, IncN, IncR, and IncY, were identified (Table 1). The genetic environments surrounding the CTX-M and CMY-2 genes were found to be associated with mobile elements, as insertion sequence (IS) elements were identified near the CTX-M and CMY-2 gene sequences (Fig. 4D). In JEONG5446 and JEONG5776, insertion sites of the plasmid-carried CMY-2 genes were distinct, mediated by different IS elements, indicating that the transposition events of the CMY-2 genes originated independently. Although insertion sites of CTX-M-32 genes in JEONG9566 and JEONG9567 were in the chromosomes at different loci, transposition events were associated with the same mobile elements, IS1380 and IS5.

To investigate potential pathogenicity of these 11 representative MDR E. coli isolates, we analyzed the virulence gene profiles of the isolates by comparing them to clinically relevant pathogenic E. coli strains, one isolated from pig, ExPEC PCN033 (20), and one from a human, KCJ1409 (10). The virulence factors were identified by using the Virulence Factor Database (VFDB). Numerous virulence factors necessary for adherence, chemotaxis, invasion, and iron uptake were identified (Fig. 5). Genes encoding curli (csgDFG), enterobactin (entABCEFS), and the iron uptake system (fepABCDG) were conserved in all isolates, whereas fimACDEFG (type 1 fimbriae), yagZYXWV (ecpABCDE; E. coli common pilus), and gspDEFGHIJK (general secretion pathway) were missing in two isolates, JEONG9566 and JEONG9592. Genes (iucABCD) encoding proteins for aerobactin synthesis, which is related to iron uptake, were only present in 3 isolates, JEONG5446, JEONG5776, and JEONG9499. The afa-VIII gene cluster encoding the AfaE-VIII adhesin was found in JEONG5776 and KCJ9499, and yersiniabactin biosynthetic proteins (ybtAEPQSTUX, fyuA, irp1, and irp2) were found in JEONG5253 and KCJ9499. K88 fimbriae (faeCDEFHIJ) and P fimbriae (papBCDFG) genes were found in JEONG5776 and KCJ9499, respectively. Among 11 representative MDR E. coli isolates, KCJ9499 showed a virulence gene profile that was more similar to that of the human clinical isolate KCJ1409 than the other 10 isolates. These two strains shared unique adhesins and iron uptake proteins, including AfaE-VIII adhesin, P-fimbriae, and yersiniabactin biosynthetic proteins, consistent with KCJ9499 having the capacity to colonize in the gastrointestinal (GI) tract of humans. Additionally, JEONG5446, identified as serotype O84:H2, carried the type 3 secretion system (T3SS) and its secreted effector proteins, including translocated intimin receptor (Tir), several E. coli secreted proteins (Esp), and non-LEE effectors (Nle) (Fig. 5). Furthermore, JEONG5446 carried a Shiga toxin-producing gene, stx₁AB, the product of which causes hemolytic uremic syndrome (HUS) in humans (21). Taken together, ESBL- or CMY-2-producing MDR E. coli strains carried virulence factor profiles similar to those of the clinically relevant isolates from human and swine origins.

As adhesion to epithelial cells is the first step in bacterial colonization of the intestinal tract, the adhesion capabilities of the 11 ESBL- or CMY-2-producing MDR E. coli isolates were evaluated in Caco-2 human epithelial colorectal adenocarcinoma cells. E. coli O157:H7 EDL933, a strain encoding T3SS and Shiga toxins, and E. coli DH5α were compared as positive and negative controls, respectively. All strains except JEONG5650 showed significantly greater adherence capability (P < 0.05) than the negative-control DH5α (Fig. 6A), consistent with a role as a human pathogen.

The stxAB genes are generally present in lambdoid prophage, and the expression of Shiga toxin is controlled by the phage regulatory system mediated by the bacterial SOS response (22). Due to mutations in the phage transcription system, the expression of Shiga toxin is frequently impaired (23). To determine whether JEONG5446 produces Shiga toxin by phage induction, the expression of Shiga toxin was measured by an immuno-dot blot assay after mitomycin C (MMC) treatment, which causes DNA damage that triggers the SOS response, thereby resulting in prophage induction and cell lysis. Bacterial cell lysis was evaluated by measuring the optical density at 600 nm (OD₆₀₀) after treatment with MMC. JEONG5446 was lysed with MMC, like EDL933, encoding Stx1 and Stx2 in prophage BP-933V and BP-933W, respectively, indicating phage induction with MMC treatment, whereas DH5α (negative control without an stx gene) maintained a normal OD₆₀₀ during the experiment in the presence of MMC (Fig. 6B). Immuno-dot blot analysis showed that the positive-control strain, EDL933, expressed both Shiga toxin 1 and 2 with MMC, whereas the negative-control strain, DH5α, did not express them regardless of the presence of MMC. JEONG5446 produced Stx1 in the manner of MMC dependency (Fig. 6C), which is consistent with the cell lysis data, indicating that JEON5446 carries the stx₁ gene, which is expressed by phage induction triggered by lytic factors, including antibiotic treatments.

Since the beef cattle employed in this study were raised without antibiotics and with limited human activities within the nearby vicinity, we hypothesized that the occurrence of these MDR E. coli strains did not likely originate on the farms, but rather, the origins of these pathogens are plausibly from environmental sources. To identify the potential sources from which these MDR E. coli strains were introduced into the farms, we conducted a phylogenetic tree analysis with other relevant global lineages using genome sequences. Recently, efforts have been made for outbreak investigation of infectious diseases by targeting a variety of pathogens through the use of whole-genome-sequence data (24). First, genomes of MDR E. coli isolates were submitted to the NCBI GenomeTrakr pathogen detection system to generate phylogenetic trees with 55,062 E. coli/Shigella genomes (25). Although none of the MDR E. coli strains were identified as clonal variants of known outbreak-associated pathogens, isolates were closely clustered with clinically relevant E. coli strains (Fig. S2A to J). For example, JEONG5776 was clustered with strain PCN033, an ESBL-producing E. coli strain that caused outbreaks in swine in China (20). Furthermore, as shown by the results in Fig. 7, a phylogenetic analysis of sequence type 306 (ST306), which has been reported to cause hemolytic uremic syndrome in humans, using 88 available genomes in EnteroBase, showed three major clades. In the phylogeny, Shiga toxin-producing isolate JEONG5446 clustered with a variety of isolates from humans, animals, and environmental sources (Fig. 7). The two closest strains, PNUSAE004879 and 2015 C-3863, isolated from humans, had relatively small numbers of SNPs in their core genomes compared to the number in JEONG5446, 1,325 SNPs in PNUSAE004879 and 1,836 SNPs in 2015 C-3863, indicating that these isolates have genetically close relatedness. These data suggest that MDR E. coli strains isolated in the GI tract of animals with no antibiotic treatment history were likely transmitted from the environment through unknown vehicles.

Standard practices of animal husbandry were applied to all animals in this study. Research protocols were approved by the University of Florida Institutional Animal Care and Use Committee (IACUC number 2015-68003-22971).

A total of 1,535 fecal samples were collected from beef calves belonging to two different herds housed at two different farms in North Central Florida. The herds were located at the Beef Research Unit (BRU) in Waldo, FL, and the North Florida Research and Education Center (NFREC) in Marianna, FL. The animals were grazing on pasture. None of the animals in this study had been previously exposed to antibiotics for treatment purpose. Sterile cotton swabs were used to collect fecal samples directly from the recto-anal junction (RAJ) of each animal. Following sample collection, fecal swabs were placed in sterile 15-ml centrifuge tubes, transported on ice to the Emerging Pathogens Institute at the University of Florida, and immediately processed.

Samples were serially diluted (up to 10⁻⁴) with Luria Bertani (LB) broth and then plated on MacConkey agar (BD, USA) containing cefotaxime (4 μg/ml). Plates were incubated at 37°C and examined after 24 h to enumerate bacterial colonies. Resistance to cefotaxime due to the production of extended-spectrum β-lactamase was identified by streaking cefotaxime-resistant isolates on ChromAgar ESBL (CHROMagar, France) as previously described (41, 42). Up to four colonies from each fecal sample with the presence of cefotaxime-resistant bacteria were purified and stored at −80°C in 15% glycerol for future use. Frozen cefotaxime-resistant isolates (n = 2,769) were revived on MacConkey agar plates containing 4 μg/ml cefotaxime and screened by PCR for the presence of the CTX-M gene using primers KCP 685 (5′-TTTGCGATGTGCAGTACCAGTAA-3′) and KCP 686 (5′-CGATATCGTTGGTGGTGCCATA-3′) (544 bp) as described previously (43). In addition, we confirmed the presence of the CMY-2 and MCR-1 genes using primers KCP556 (5′-ATGATGAAAAAATCGTTATGC-3′) and KCP557 (5′-TTGCAGCTTTTCAAGAATGCGC-3′) (1,200 bp) for the CMY-2 gene (44) and primers KCP830 (5′-CGGTCAGTCCGTTTGTTC-3′) and KCP831 (5′-CTTGGTCGGTCTGTAGGG-3′) (305 bp) for the MCR-1 gene (45). Taxonomic identification was conducted at the species level to identify cefotaxime-resistant E. coli. Genomic DNA was extracted with a Qiagen DNA minikit (Qiagen, Valencia, CA) and used as a template for PCR to amplify the 16S rRNA gene using primers KCP 812 (5′-CAGGCCTAACACATGCAAGTC-3′) and KCP 813 (5′-GGGCGGWGTGTACAAGGC-3′) (∼1,300 bp) (46). The PCR products were purified using the QIAquick PCR purification kit and sequenced by the Sanger sequencing method at the Interdisciplinary Center for Biotechnology Research (ICBR) at the University of Florida. The resulting sequences were analyzed using the NCBI nucleotide BLAST program to compare the homology of the sequences to the 16S rRNA gene sequences of other organisms.

A total of 36 cefotaxime-resistant E. coli isolates, as identified by the 16S rRNA gene sequencing, were further subjected to MIC testing against cefotaxime using the broth microdilution method (47). Isolates were further tested for susceptibility to 13 different antimicrobial compounds according to the Clinical and Laboratory Standards Institute guidelines (18). Briefly, the isolates were tested using the standard Kirby-Bauer disk diffusion method on Mueller-Hinton agar to generate antibiograms of the cefotaxime-resistant isolates. The control strains used for the antibiotic susceptibility testing were Escherichia coli (ATCC 35401), Staphylococcus aureus (ATCC 25923), and Pseudomonas aeruginosa (ATCC 27853). The following antimicrobial-disk concentrations were used: amikacin (30 μg), ampicillin (10 μg), amoxicillin-clavulanic acid (30 μg), ceftiofur (30 μg), cephalothin (30 μg), chloramphenicol (30 μg), colistin (10 μg), gentamicin (10 μg), nalidixic acid (30 μg), streptomycin (10 μg), sulfamethoxazole-trimethoprim (23.75 μg/1.25 μg), sulfisoxazole (250 μg), and tetracycline (30 μg) (BD, USA).

For whole-genome sequencing of 36 isolates, DNA was extracted from each isolate using the DNeasy blood and tissue kit (Qiagen, Valencia, CA) following the protocol for Gram-negative bacteria. DNA libraries were constructed using the Nextera XT sample preparation kit (Illumina, San Diego, CA) according to the manufacturer's protocol. Sequencing was performed using an Illumina MiSeq with cartridges providing 2 × 250-bp paired-end read coverage. The resulting sequence reads were trimmed for quality and length using Sickle (48) and then assembled using SPAdes (version 3.0) (49). The assembled genome sequences of 36 isolates were deposited in NCBI (Table S1 in the supplemental material). Next, a multiple alignment of the de novo assemblies was performed using progressiveMauve (version 2.4.0) (http://darlinglab.org/mauve/user-guide/progressivemauve.html) to understand genome rearrangement and compare whole-genome architecture (50). The phylogenetic trees of 36 sequenced genomes and 88 ST306 strains were generated using Parsnp (https://harvest.readthedocs.io/en/latest/) (51) based on the core-genome single-nucleotide polymorphisms (SNPs). The assembled genomes were used as input files of Parsnp that used PhiPack (52) to detect recombination and generated reliable core-genome SNPs. The set of core-genome SNPs was used to generate maximum-likelihood phylogenetic trees using 1,000 bootstrap replicates with FastTree2 (53) embedded in the Parsnp program. The number of SNPs in each phylogenetic clade was calculated by using NCBI Pathogen Detection (https://www.ncbi.nlm.nih.gov/pathogens/) (54). The phylogenetic trees of each representative strain were generated using Genome Workbench (https://www.ncbi.nlm.nih.gov/tools/gbench/) after uploading the raw reads to GenomeTrakr (25).

Virulence genes were identified through PATRIC (https://patricbrc.org/) (55) by aligning whole-genome sequences against the Virulence Factor Database (VFDB) (56) using BLAST. The identity of each virulence gene was defined by multiplying gene coverage times query identity with subject identity. The Comprehensive Antibiotic Resistance Database (CARD, version 2.0.1, https://card.mcmaster.ca/analyze) (57) was also employed for discovery of additional antibiotic resistance genes. Briefly, the whole-genome sequence of each isolate was submitted to Resistance Gene Identifier (RGI, version 4.0.3) in CARD to predict resistance genes using homology and SNP models.

The multilocus sequence type (MLST), serotype, plasmid replicons, and plasmid MLST (pMLST) of each isolate were determined using MLST 1.8 (58), SerotypeFinder (59), PlasmidFinder and pMLST 1.4 (60) of the Center for Genomic Epidemiology (CGE) (http://www.genomicepidemiology.org/).

To identify whether the CTX-M and CMY-2 genes were in the chromosome or plasmid, we employed PLACNETw (https://castillo.dicom.unican.es/upload/) as previously described (61). PLACNETw assembled sequence reads to generate contigs, and then the contigs were aligned to reference sequences of complete chromosomes and plasmids from NCBI. An original network was generated to illustrate the relatedness among contigs and reference sequences. Following the manual pruning of the original network, the contigs were assigned to either chromosome or plasmid. To find out the location of a CTX-M or CMY-2 gene, alignment was conducted using the sequence of the CTX-M or CMY-2 gene against the chromosomal and plasmid sequences by BLASTn. The genetic environments of CTX-M and CMY-2 genes, i.e., the genes surrounding a CTX-M or CMY-2 gene, were acquired from GenBank files of sequenced strains. Similarity of genetic environments of CTX-M or CMY-2 genes was determined by BLASTn embedded in EasyFig (http://mjsull.github.io/Easyfig/) (62).

An adherence assay was conducted to evaluate the adherence of the representative isolates to Caco-2 cells (human epithelial colon cancer cell). EDL933 and DH5α were used as a positive and a negative control, respectively. Caco-2 cells were maintained in Dulbecco modified Eagle medium (DMEM) (product number 10-017-CV; Corning, USA) supplemented with 20% (vol/vol) heat-inactivated fetal bovine serum at 37°C and 5% CO₂. A total of 10⁵ Caco-2 cells were seeded into each well of a 24-well polystyrene plate, followed by incubation until 90% confluence. Overnight bacterial cultures in LB broth were seeded into new tubes of LB broth (1:250) to produce the main cultures, followed by incubation at 37°C with shaking for 8 h. A total of 10⁶ bacterial cells were washed with sterile phosphate-buffered saline (PBS) three times, resuspended in 500 μl DMEM, and added to each well with 10⁵ Caco-2 cells (multiplicity of infection [MOI] of 10). After a 3-h incubation at 37°C with 5% CO₂, the medium in each well was replaced by 500 μl of new DMEM, followed by another 3-h incubation. After the medium was removed, each well was washed with sterile PBS three times to remove the unattached bacteria. Then, 1 ml of 1% Triton X-100 was added to each well to lyse the Caco-2 cells. Finally, 100 μl of the diluted suspension was spread on LB agar and incubated at 37°C overnight, followed by the enumeration of colonies on the plates. This experiment was conducted in duplicate three times.

Mitomycin C (MMC) treatment was previously described (63). Briefly, the optical densities of bacteria were measured to determine whether the bacterial cells were lysed by phage induction. Overnight bacterial cultures were seeded to LB medium. When the OD₆₀₀ of the bacterial culture reached between 0.5 and 0.7, MMC was added to a final concentration of 0.5 μg/ml. The negative-control and positive-control strains were DH5α and EDL933, respectively. Phage induction was conducted as previously described with minor changes (23). To collect phage particles in the bacterial cultures, 25-ml amounts of bacterial cell cultures were treated without or with MMC (final concentration of 1 μg/ml) when the OD₆₀₀ reached 0.7. After 24 h of incubation, bacterial cultures were centrifuged at 3,700 × g for 30 min at 4°C, followed by filtration of the supernatants through 0.22-μm-pore-size membrane filters (catalog number 09-719A; Fisher Scientific, USA). To precipitate phage particles, 25% (vol/vol) polyethylene glycol 8000 (PEG 8000) (product number 81268; Sigma-Aldrich, USA)-and-NaCl solution (20% PEG 8000 and 10% NaCl) was added to the supernatant. The mixture was incubated at 4°C overnight, followed by centrifugation at 12,000 × g for 1 h. The pellets acquired were resuspended using STE buffer (1 M Tris [pH 8], 0.5 M EDTA [pH 8], and 5 M NaCl), and used for immuno-dot blotting.

To detect the expression of Shiga toxin type 1 (Stx1) and Shiga toxin type 2 (Stx2), we used an immuno-dot blot assay. Briefly, 3 μl of each phage particle sample (bacterial cell cultures of strains EDL933, JEONG5446, and DH5α treated with or without MMC) was loaded to a nitrocellulose membrane (0.2-μm pores; Bio-Rad, USA). The membranes were blocked with 5% skim milk (BD, USA) at room temperature for 1 h, followed by incubation with Stx1 antibody (sc-52726; Santa Cruz, USA) diluted 1:100 in 5% skim milk or Stx2 antibody (sc-52727; Santa Cruz, USA) diluted 1:100 in 5% skim milk for 1 h at room temperature. After washing with PBS with 0.2% Tween 20 (TPBS), the membranes were incubated at room temperature for 1 h with goat anti-mouse IgG₁ secondary antibody (IRDye 800CW, product number 926-32350; LI-COR, USA) diluted 1:10,000 in 5% skim milk. After washing with TPBS, the blots were imaged with an Odyssey CLs (LI-COR, USA) to identify Stx1 and Stx2.

The data from the adherence assay were analyzed using one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test.

Whole-genome sequences of 36 strains have been deposited in the NCBI database. The accession numbers of these sequenced genomes are listed in Table S1 in the supplemental material, including JEONG5084 (NSED00000000), JEONG5114 (NSEE00000000), JEONG5120 (NSEF00000000), JEONG5232 (NSEG00000000), JEONG5250 (NSEH00000000), JEONG5298 (NSEI00000000), JEONG5413 (NSEJ00000000), JEONG5446 (NSEK00000000), JEONG5453 (NSEL00000000), JEONG5507 (NSEM00000000), JEONG5511 (NSEN00000000), JEONG5617 (NSEO00000000), JEONG5650 (NSEP00000000), JEONG5766 (NSEQ00000000), JEONG5776 (NSER00000000), JEONG9566 (PKKV00000000), JEONG9567 (PKKU00000000), JEONG9592 (PKKW00000000), JEONG9593 (QDJW00000000), JEONG9594 (QDJX00000000), JEONG9596 (QDJY00000000), JEONG9597 (QDKF00000000), JEONG9598 (PKKX00000000), JEONG9599 (QDJZ00000000), JEONG9600 (QDKA00000000), JEONG9601 (QDKB00000000), JEONG9602 (QDKC00000000), JEONG9603 (QDKD00000000), JEONG9615 (QDKE00000000), KCJ9488 (QGNB00000000), KCJ9489 (QGNC00000000), KCJ9491 (QGND00000000), KCJ9492 (QGNE00000000), KCJ9493 (QGNF00000000), KCJ9499 (QGNS00000000), and KCJ9511 (QGNG00000000).