Here we introduce the principle, structure, and instruction of the intelligent high-throughput antimicrobial sensitivity testing/phage screening system. Its application is illustrated by using Salmonella isolated from poultry in Shandong, China, as an example. The Lar index is calculated, and its significance in evaluating antimicrobial resistance is discussed comprehensively.
To improve the efficiency of antimicrobial susceptibility testing (AST) and phage high-throughput screening for resistant bacteria and to reduce the detection cost, an intelligent high-throughput AST/phage screening system, including a 96-dot matrix inoculator, image acquisition converter, and corresponding software, was developed according to AST criteria and the breakpoints of resistance (R) formulated by the Clinical & Laboratory Standards Institute (CLSI). AST and statistics of minimum inhibitory concentration (MIC) distributions (from R/8 to 8R) of 1,500 Salmonella strains isolated from poultry in Shandong, China, against 10 antimicrobial agents were carried out by the intelligent high-throughput AST/phage screening system. The Lar index, meaning “less antibiosis, less resistance and residual until little antibiosis”, was obtained by calculating the weighted average of each MIC and dividing by R. This approach improves accuracy in comparison with using the prevalence of resistance to characterize the antimicrobial resistance (AMR) degree of highly resistant strains. For the strains of Salmonella with high AMR, lytic phages were efficiently screened from the phage library by this system, and the lysis spectrum was computed and analyzed. The results showed that the intelligent high-throughput AST/phage screening system was operable, accurate, highly efficient, inexpensive, and easy to maintain. Combined with the Shandong veterinary antimicrobial resistance monitoring system, the system was suitable for scientific research and clinical detection related to AMR.
As antimicrobial agents have been widely used to prevent bacterial infectious diseases, antimicrobial resistance (AMR) has become a global public health problem1. Combating AMR is the current main mission of monitoring AMR of epidemiological pathogens and synergistic therapy of sensitive antimicrobial agents and lytic bacteriophages2.
In vitro antimicrobial sensitivity testing (AST) is the mainstay for monitoring therapy and detecting the level of AMR. It is an important part of antimicrobial pharmacology and the critical basis for clinical medication. The Clinical and Laboratory Standards Institute (CLSI) of the United States and the European Committee on Antimicrobial Susceptibility Testing (EUCAST) have formulated and revised international criteria of AST and continuously modified and supplemented AST methods and the breakpoints to determine the MIC of one certain "organism-antimicrobial agent" combination as sensitive (S), resistant (R) or intermediate (I)3,4.
From the 1980s to the 1990s, automatic micro broth dilution instruments were rapidly developed and applied to clinical practice, with examples including Alfred 60AST, VITEK System, PHOENIXTM, and Cobasbact5,6,7. However, these instruments were expensive, required high-cost consumables, and their detection ranges were designed for clinical patient medication5,6,7. For these reasons, they are not suitable for veterinary clinical examination and detection of large quantities of highly resistant strains. In this study, an intelligent high-throughput AST/phage screening system, including a 96-dot matrix inoculator (Figure 1), image acquisition converter (Figure 2), and corresponding software8, was developed to conduct AST for a batch of bacteria strains against multiple antimicrobial agents at one time by the agar dilution method. Moreover, the system was also used to detect and analyze the lysis patterns of phages against antimicrobial-resistant bacteria9, and lytic phages were selected efficiently from the phage library. This system was found to be efficient, affordable, and easy to operate.
Figure 1: Structural diagram of the 96-dot matrix inoculator. 1: Inoculation pin plate; 2: Mobile carrier; 3: Seed block; 4: Incubated plate; 5: Base; 6: Operating handle; 7: Limit pin. Please click here to view a larger version of this figure.
Figure 2: Structural diagram of the image acquisition converter. 1: Shell; 2: Display screen; 3: Image acquisition room; 4: Detection board base; 5: Detection board in and out of warehouse; 6: Control board; 7: Image acquisition conversion device; 8: Light source; 9: Image scanner. Please click here to view a larger version of this figure.
The Salmonella strains used in this study were collected from poultry in Shandong, China, after obtaining approval from the Biosafety Committee of the Institute of Animal Sciences and Veterinary Medicine, Shandong Academy of Agricultural Sciences, China.
1. Application of the intelligent high-throughput AST system8
2. Intelligent high-throughput phage screening system9
Following the protocol of the intelligent high-throughput AST system, its application was illustrated by Salmonella from poultry in Shandong, China, as an example.
The growth of Salmonella strains on agar plates with ampicillin (R of 32 µg/mL) at concentrations from 2 to 256 µg/mL determined by the image acquisition converter is shown in Figure 3. The horizontal 1st well A1 was the negative control and showed no colony growth; A2 and A3 were the quality control strains with MIC 4 µg/mL (forming a colony on the agar plate with 2 µg/mL ampicillin, but not on that of 4 µg/mL), within the quality control range standardized by CLSI (2-8 µg/mL). The MIC of the Salmonella strain in A4 was 64 µg/mL, while that of A5 was 16 µg/mL. MIC distributions of 93 Salmonella strains against ampicillin were calculated automatically by the software.
Figure 3: Morphology of Salmonella on a series of culture plates with ampicillin. 8 horizontal: A-H, 12 vertical: 1-12. Please click here to view a larger version of this figure.
The high-throughput AST system was applied to determine the AMR of Salmonella strains from animals in Shandong Province. The MIC data was uploaded to the database of Varms (http://www.varms.cn/)12. The statistical results are shown in Table 1. A total of 10 antimicrobial agents were tested against 300-1,500 Salmonella strains, showing the high-throughput advantages of this system.
According to the resistant breakpoints of the CLSI standard, the resistance rate (R%) was calculated as a percentage of strains with MIC ≥ R among the tested strains (Table 1). The R% values of ampicillin, ciprofloxacin, and amoxicillin-clavulanic acid were higher than 50%, the R% values of doxycycline, florfenicol, cefotaxime, and enrofloxacin were 30%-50%, and the R% values of gentamicin, amikacin, and meropenem were less than 30%. Meropenem was not used in industrially raised animals and showed an R% of 7%.
The R% indicated the proportion of the bacterial strains with MICs higher than R, while MIC distributions showed the number of strains with each MIC to describe the overall AMR of Salmonella more accurately. For example, the R% of ampicillin was 73%, and the maximum number of samples (916 strains) was concentrated with MIC ≥ 256 µg/mL, indicating that the resistance of Salmonella to ampicillin was quite serious.
Table 1: MIC distributions, R%, and Lar index values of Salmonella from animals in Shandong Province. MICs corresponding to bold font are the R values of antimicrobial agents. Please click here to download this Table.
For uniformity and comparability, 3 gradients were extended forward and backward, centered on the R of each drug. According to the MIC distributions of seven gradients, the Lar index was calculated by the formula:
.
As an example for ampicillin, the R-value was 32 µg/mL, the number of samples was 1,414, and Lar = (4/32) x (245/1414) + (8/32) × (16/1414) + (16/32) × (117/1414) + (32/32) × (27/1414) + (64/32) × (36/1414) + (128/32) × (57/1414) + (256/32) × (916/1414) = 2-3 × (245/1414) + 2-2 × (16/1414) + 2-1 × (117/1414) + 20 × (27/1414) + 21 × (36/1414) + 22 × (57/1414) + 23 × (916/1414) = 5.48. By this formula, Lar indices of other antimicrobial agents were calculated and are shown in Table 1.
The significance of the Lar index was to indicate the severity degree of AMR accurately. Taking ciprofloxacin and amoxicillin-clavulanic acid as examples, their R% values were similar, at 68% and 65%, respectively, but their Lar indices differed significantly, at 4.57 and 1.76, respectively. The reason for this was clearly illustrated by the distribution of the high MIC values in Table 1, in which 71.3% of ciprofloxacin-resistant strains were distributed in 8R (32 µg/mL), while the MICs of amoxicillin-clavulanic acid-resistant strains were mostly concentrated on R and 2R, and the proportion of 8R was low (8.69%). Therefore, the Lar index of ciprofloxacin was higher than that of amoxicillin-clavulanic acid, indicating that the proportion of highly ciprofloxacin-resistant strains was higher than that of amoxicillin-clavulanic acid-resistant strains. The Lar index was a more accurate indicator of the degree of AMR than the resistance rate.
According to the formula of the Lar index, if the MICs of all strains with respect to a certain drug were R, the Lar index would be 1; if the MICs of all strains were 2R, the Lar index would be 2. Therefore, the Lar index showed multiple relationships between the comprehensive MICs and the corresponding R value, except for the edge concentrations. The Lar index was used to assess the degree of AMR, and the advantage was more pronounced if AMR was higher. For uniformity and comparability, the calculation range of the Lar index was seven MICs with the front and rear 3 gradients centered on the R value, and the weighted average was obtained by integrating the different antimicrobial agents, the number of strains, and MIC distributions. Therefore, the value range of the Lar index was 0.125-8. The closer the Lar was to 0.125, the lower the AMR, and the closer it was to 8, the higher the AMR. However, there was no proportional relationship between Lar and R at the edge concentration. When the antimicrobial agents and the calculation range of the Lar index were definite, the general Lar was normalized to an intuitive comprehensive value used to directly compare and evaluate the degree and change trend of AMR under the different conditions of different bacteria, users, years, regions, etc.
Following the protocol of the intelligent phage screening system, the application was demonstrated by taking 96 phages of Salmonella to lyse AMR Salmonella strains as an example, and the phage lytic pattern was analyzed.
Ninety-six phages were transferred to double-layered plates containing Salmonella by a 96-dot matrix inoculator. The morphology of the formed spots is shown in Figure 4. There were four main types (although not limited to four): clear round spot (●), collection of plaques (), turbid lytic spot (
), and no lytic spot (○).
Figure 4: Morphology of Salmonella spots on double-layer agar plates. 1 and 2: the patterns produced by 96 phages on different Salmonella strains. "●" round clear spot, "" collection of plaques, "
" turbid lytic spot, "○" no lytic spot) Please click here to view a larger version of this figure.
On the double-layer plates, the lytic spots of different phages on different host bacteria were diverse in morphology10. The "round clear lytic spot" resulted from a phage that could reliably kill the host on the plate but which may or may not successfully replicate at that host's expense. In this case, further dilution was necessary to conclusively determine the type. The "collection of plaques" was formed by true plaques. Each individual zone of lysis was clearly produced from a single infectious center, which demonstrated that the phage had replicated at the expense of the bacteria on the plate. The "turbid lytic spot" resulted from a phage that did not reliably kill the host on the plate and which may or may not successfully replicate at that host's expense. In this case, there was more than one possibility, thus requiring confirmation based on further research interests. The "no lytic spot" indicated nonlytic nature.
Moreover, using this system for preliminary testing could essentially determine whether the strains and bacteriophages were duplicated. If different Salmonella species were selected and the patterns of two bacteriophages were identical, it indicated that the bacteriophages might be duplicated. If nonrepetitive bacteriophages were selected to infect unknown Salmonella species from clinical sources and the phage pattern was the same, it indicated that Salmonella strains might be the same strain. Furthermore, the number and proportion of duplicates had reference values for epidemiological investigation.
The agar dilution method has been well-established and used widely. The principle of the high-throughput AST system was that of the agar dilution method. One of the critical steps within the protocol was the accurate high throughput transfer of 96 inocula at one time, which was performed multiple times in a row. To complete this critical step, the pins of the 96-dot matrix inoculator were uniform and very smooth. The natural deposition of each pin was a volume of approximately 2 µL, which aggregated into small droplets on the surface of the agar media to be quickly absorbed into the agar, not flowing or splashing to cause cross-contamination. The second critical step was the statistical processing of large AST data. With high-throughput inoculation, test data needed to be determined, and the workload was huge. The intelligent analysis and statistical results produced by the intelligent image acquisition converter and supporting software were a perfect solution to this problem. The AST results were automatically uploaded directly to the Varms database. The efficiency and accuracy of result interpretation were improved, and the errors caused by human factors were reduced13.
The third critical step was to propose the Lar index of AMR. At present, there are few comprehensive evaluation indices of AMR in the world. According to the literature, a drug resistance index (DRI) was described by Laxminarayan and Klugman to measure changes in antibiotic effectiveness14,15. It combined the prevalence of resistance and relative frequencies of prescription but could not characterize the degree of AMR and evaluate the changes in high drug resistance levels. The drug effectiveness index (DEI)16, another index derived from the DRI, has the same disadvantage as the DRI. Thus, the Lar index was proposed, which consisted of three steps: (1) The MICs of bacterial strains were normalized based on the respective R-value, eliminating the differences in antimicrobial agents caused by different R values of the AST standard; (2) According to the MIC distributions, the weighted average values were calculated to reflect the degree of AMR more accurately than the resistance rate; (3) The arithmetic mean of Lar indices of multiple antimicrobial agents, that is, the general Lar, could reflect the comprehensive situation of AMR, offering convenience for the assessment and comparison of AMR at different levels.
The hardware design of these apparatuses was reasonable and simple to operate, and all parts ran smoothly. There was no jamming problem or fault. The design of the supporting software matched the personalized requirements of AST and phage screening and was easy to operate and use. One instrument was equipped with 4-5 boxes of inoculation pins for multiple applicable scenarios of batch bacterial transfer. The core components of this instrument were the inoculation pin plate and pins made of stainless steel, which were adaptable to a variety of environments and were able to be autoclaved, disassembled and replaced. The inoculation pins were designed to be combined in any way.
It was imperative to establish an AMR monitoring system because of the prevalence of resistant pathogens caused by the misuse of antibiotics. Since 2008, the public health team of the Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences, has carried out AMR monitoring of animals successively in Shandong Province12,13,17. It was necessary to efficiently detect MICs of pathogens to regulate the use of antimicrobial agents because of the high level of veterinary resistance and the large monitoring volume. However, the relevant instruments for AST are expensive, and the cost of the operation and consumables is high and not suitable for a wide range of large-scale farms. For this reason, the development of the intelligent high-throughput AST system and the standardization of its application were conducive to promoting the establishment of a sound system for AMR monitoring technology. According to previous research12,13, the intelligent high-throughput AST system achieved good repeatability and stability to match the standard of CLSI and was applied to AST and analysis of clinical pathogenic bacteria in animals. To date, comprehensive AMR data for more than 20,000 epidemic strains have been accumulated12. For the resistant bacteria found in the monitoring process, this system could also be used for rapid high-throughput screening of lytic phages to cooperate with antimicrobial agents to reduce AMR. The application of the 96-dot matrix inoculator and the image acquisition converter for phage lysis screening was an extended function, and no other instruments had previously been applied in this field.
The intelligent high-throughput AST/phage screening system combined AST with lytic bacteriophages to achieve AMR monitoring, control, and reduction. At the same time, the Lar index was more intuitively and concisely used to evaluate the contributions of various factors and new antibacterial technologies to the reduction of AMR.
The authors have nothing to disclose.
This work was supported by the National Key Research and Development Project (2019YFA0904003); Modern Agricultural Industrial System in Shandong Province (SDAIT-011-09); International Cooperation Platform Optimization Project (CXGC2023G15); Major Innovation tasks of agricultural Science and technology innovation project of Academy of Agricultural Sciences Shandong, China (CXGC2023G03).
96 well culture plate | Beijing lanjieke Technology Co., Ltd | 11510 | |
96-dot matrix AST image acquisition system | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright | |
96-dot matrix inoculator | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | N/A | Patented product |
Agar | Qingdao hi tech Industrial Park Haibo Biotechnology Co., Ltd | HB8274-1 | |
Amikacin | Shanghai McLean Biochemical Technology Co., Ltd | A857053 | |
Amoxicillin | Shanghai McLean Biochemical Technology Co., Ltd | A822839 | |
Ampicillin | Shanghai McLean Biochemical Technology Co., Ltd | A830931 | |
Analytical balance | Sartorius | BSA224S | |
Automated calculation software for Lar index of AMR | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright | |
Bacteria Salmonella strains | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | N/A | Animal origin |
Bacterial resistance Lar index certification management system V1.0 | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright | |
Ceftiofur | Shanghai McLean Biochemical Technology Co., Ltd | C873619 | |
Ciprofloxacin | Shanghai McLean Biochemical Technology Co., Ltd | C824343 | |
Clavulanic acid | Shanghai McLean Biochemical Technology Co., Ltd | C824181 | |
Clean worktable | Suzhou purification equipment Co., Ltd | SW-CJ-2D | |
Colistin sulfate | Shanghai McLean Biochemical Technology Co., Ltd | C805491 | |
Culture plate | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | N/A | Patented product |
Doxycycline | Shanghai McLean Biochemical Technology Co., Ltd | D832390 | |
Enrofloxacin | Shanghai McLean Biochemical Technology Co., Ltd | E809130 | |
Filter 0.22 μm | Millipore | SLGP033RB | |
Florfenicol | Shanghai McLean Biochemical Technology Co., Ltd | F809685 | |
Gentamicin | Shanghai McLean Biochemical Technology Co., Ltd | G810322 | |
Glass bottle 50 mL | Xuzhou Qianxing Glass Technology Co., Ltd | QX-7 | |
High-throughput resistance detection system V1.0 | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright | |
Image acquisition converter | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | N/A | Patented product |
Meropenem | Shanghai McLean Biochemical Technology Co., Ltd | M861173 | |
Mueller-Hinton agar | Qingdao hi tech Industrial Park Haibo Biotechnology Co., Ltd | HB6232 | |
Petri dish 60 mm x 15 mm | Qingdao Jindian biochemical equipment Co., Ltd | 16021-1 | |
Petri dish 90 mm x 15 mm | Qingdao Jindian biochemical equipment Co., Ltd | 16001-1 | |
Salmonella phages | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | N/A | |
Shaker incubator | Shanghai Minquan Instrument Co., Ltd | MQD-S2R | |
Turbidimeter | Shanghai XingBai Biotechnology Co., Ltd | F-TC2015 | |
Varms base type library system V1.0 | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright | |
Vertical high-pressure steam sterilizer | Shanghai Shen'an medical instrument factory | LDZX-75L | |
Veterinary pathogen resistance testing management system | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright | |
Veterinary resistance cloud monitoring and phage control platform V1.0 | Institute of Animal Science and Veterinary Medicine, Shandong Academy of Agricultural Sciences | In-house software copyright |