Qualitative and quantitative detection of surgical pathogenic microorganisms Escherichia coli and Staphylococcus aureus based on ddPCR system

Bacterial culture and drug susceptibility testing are used to identify pathogen infections. Nevertheless, the process requires several days from collection to the identification of bacterial species and drug-resistance patterns. The digital PCR system is a rapidly developing quantitative detection technology widely applied to molecular diagnosis, including copy number variations, single nucleotide variant analysis, cancer biomarker discovery, and pathogen identification. This study aimed to use a droplet digital PCR system to identify bacteria in blood samples and explore its ability to identify pathogen in bacteremia. Then, we designed primers and probes of SWG-9 and COA gene for E. coli and S. aureus to identify in blood samples with the ddPCR system. The system had demonstrated extremely high detection accuracy in blood samples, and the detection rate of E. coli was 13.1–21.4%, and that of S. aureus was 50–88.3%. Finally, blood samples containing both E. coli and S. aureus were tested to evaluate further the accuracy and applicability of this method, indicating the detection rates range from 18.1% to 97%. The ddPCR system is highly promising as a qualitatively and quantitatively screening method for rapidly detecting pathogen.

www.nature.com/scientificreports/ bacterial culture growth characteristics, non-pathogenic bacteria contamination, and improper material selection, the test results in actual work may generate false negatives or false positives. Incorrect identification results not only in delayed treatment, wasting medical resources, but in mistaken diagnosis and treatment. For these reasons, rapid, accurate identification of pathogenic microorganisms has become an intense focus of research. In recent years, molecular biology techniques have been applied to detect pathogenic microorganisms. They have attracted attention due to their high specificity, lower time-requirement, and reduced incidence of crossinfection. These techniques include Enzyme-Linked Immunosorbent Assay 2 , Polymerase Chain Reaction 5 and Gene microarray chips 6 . To date, Polymerase Chain Reaction technology has been widely used in many research fields because of its simple, intuitive, economical, and rapid detection characteristics 7 . Digital polymerase chain reaction (dPCR) was first proposed by Kenneth Kinzler and Bert Vogelstein in 1999. This is a PCR technology that truly achieves absolute quantification after qPCR technology 8 . Droplet digital PCR (ddPCR) is a new technology that enables absolute quantification of nucleic acids with high analytical sensitivity and precision. ddPCR splits PCR reagents into tens of thousands of nanoliter or picoliter partitions by a microfluidic chip so that each droplet contains 0 or 1 DNA template. After PCR amplification and fluorescence detection, the target nucleic acids are calculated from the number of positive and negative droplets by Poisson statistics 9 . Compared with traditional qPCR, digital PCR has higher sensitivity, specificity, and accuracy. It plays an essential role in many fields, including early diagnosis of tumor markers 10 , gene expression product analysis 11 , food safety testing 12 , pathogenic microorganism testing 13 , common genetic disease testing, and non-invasive prenatal diagnosis 14 .
In the present study, we designed primers and probes of SWG-9 and COA gene for E. coli and S. aureus. Then, ddPCR was applied to test the advantage and reliability for identifying E. coli and S. aureus in simulated bacteremia blood samples to provide the theoretical basis for subsequent clinical sample testing.

Results
Assay design and identification of bacterial strains using the ddPCR system. To examine the method's dynamic range and detection limit, serial dilutions of E. coli and S. aureus DNAs were subjected to ddPCR. Bacterial DNA was quantified using Qubit 3.0, and the estimated copy number was calculated. Then, we diluted the bacterial DNA templates in different concentrations.
The ddPCR-based bacterial strains detection workflow mainly consisted of five steps: sample preparation, preparation of reaction mixture, droplet generation, PCR amplification, and fluorescence detection and data analysis. For strains of bacteria, and the number of bacterial nucleic acids can be detected simultaneously. For NC, all experiments performed in this study showed no positive signal in either the FAM or the VIC channel, indicating no contamination in the ddPCR system. To examine the dynamic range of the method, serial dilutions of E. coli or S. aureus were quantified by ddPCR, and the results of three replicates were analyzed (Fig. 1,  S3; Tables 1, 2, S11, S12).
The ddPCR system identifies various E. coli ATCC25922 nucleic acid (Fig. 1A, corresponding to A05 to H05). The ddPCR signals were linear over the range from 0 to 10 5 copies/μL with R 2 = 0.9995 for all sites. The actual copy number of bacteria from A05 to E05 measured using the ddPCR system ( Fig. 1A Table 1). The ddPCR system identified various S. aureus ATCC25923 nucleic acid (Fig. 1B, corresponding to A01-H01), and the measured number of bacterial nucleic acid templates were very close to the theoretical value. The ddPCR signals were linear over the range from 0 to 10 4 copies/μL with R 2 = 1 for all sites. The actual copy number of bacteria nucleic acid templates from A01 to C01 measured using the ddPCR system ( Fig. 1 B, Table 2). We compared the lowest detection limit of ddPCR and real-time quantitative PCR in different numbers of bacterial nucleic acid templates. When the number of nucleic acid templates is less than 100, the real-time quantitative PCR system cannot produce a useful amplification curve (Figure S1, S2; Table S8-S10).
The specificity of ddPCR for simultaneously identified two bacterial strains. To validate the specificity of this method, two groups of ddPCR amplification experiments, including E. coli nucleic acid and S. aureus nucleic acid, were performed in triplicates for each detection site. In the two sets of experiments, the actual DNA copy numbers of ATCC25922 and ATCC25923 were very close to the theoretical values, and there was almost no interference between them. The detection rate of bacteria was as high as 80% or even 90% (Fig. 2, Tables 3 and 4).

Bacteria count.
We added E. coli and S. aureus to 50 ml of Luria-Bertani bacterial culture medium and placed it in an orbital shaker at 37 °C overnight. We then diluted the bacterial solutions 10 6 times and performed a rolling ball scratching board. We calculated the number of colonies in the culture medium and found that the number of colonies of E. coli ATCC25922 was 107, and the number of colonies of S. aureus ATCC25923 was 70 (Fig. 3). Conversion is made according to the bacterial solution's dilution factor. The concentration of the E. coli ATCC25922 bacterial solution was 1.25 × 10 9 cells/ml, and the concentration of S. aureus ATCC25923 bacterial solution was 6.7 × 10 8 cells/ml. Detection of simulated bacteria using the ddPCR system. We used the ddPCR system to identify bacterial nucleic acid extracted from the simulated bacteremia blood samples. We counted the number of bacteria by counting the number of nucleic acid templates. The number of bacteria detected using the ddPCR system and the theoretical number of bacteria are listed in Tables 5 and 6. We took the average number of bacteria detected in the three groups and calculated the logarithm of the average number of bacteria to perform fitting analysis (Fig. 4, ATCC25922: R 2 = 0.9998, ATCC25923: R 2 = 0.9996). The E. coli ATCC25922 nucleic acid detection rate in simulated bacteremia blood samples was 13.1-21.4%, and the S. aureus ATCC25923 nucleic acid detection rate in simulated bacteremia blood samples was 50-88.3%. www.nature.com/scientificreports/ We used the ddPCR system to measure bacterial nucleic acid extracted from the simulated bacteremia blood samples contaminated by E. coli and S. aureus simultaneously (Table 7 and 8). We took the logarithm of the measured number of bacteria to make a fitting curve (Fig. 5, ATCC25922: R 2 = 0.9532, ATCC25923: R 2 = 0.9958). The bacterial nucleic acid detection rate of E. coli ATCC25922 was 18.1-77%, and that of S. aureus ATCC25923 was 44.9-97%. Figure 1. The dynamic range of the ddPCR-based assay. Detecting the bacterial nucleic acid by ddPCR system, (A) The 1D droplet spots of FAM fluorescence amplitude for eight detection sites for E. coli ATCC25922 (from A06 to H06: the estimated copy numbers of nucleic acid template were 10 5 , 10 4 , 10 3 , 10 2 , 25, 6.25, 1.56, and NC) were performed for each detection site. Linear fitting lines are shown on the right (B) The 1D droplet spots of VIC fluorescence amplitude for four detection sites for Staphylococcus aureus ATCC25923 (from A01 to H01: the estimated copy numbers of nucleic acid template were 10 4 , 10 3 , 10 2 , and NC) were performed for each detection site. Linear fitting lines are shown on the right. www.nature.com/scientificreports/     www.nature.com/scientificreports/

Discussion
The specific genes of bacterial strains were selected according to the GenBank database. Nandakafle and Brozel first discovered the SWG-9 UIDA gene in 2015. The majority of E. coli strains carry the SWG-9 gene, which contains 566 bases and encodes a protein product containing 188 amino acids. So we designed a primer and MGB probe for this gene to detect E. coli. Also, Most S. aureus carries the Staphylococcus coagulase gene, which can encode protein products of 647 amino acids, and this gene plays a vital role in the pathogenic process of S. aureus 15 . Therefore, we selected the conserved sequence of this gene in the experiment to design PCR primers and probes to identify strains. Furthermore, in this study, the sensitivity and accuracy of the ddPCR system have obvious advantages compared with real-time fluorescent quantitative PCR. The ddPCR system retains high sensitivity and accuracy when simultaneously detecting two pathogenic microorganisms. Also, the ddPCR system performs absolute quantification without the need for a standard curve. The detection rate of simulated bacteremia blood samples using the www.nature.com/scientificreports/ ddPCR system is lower than that of purified bacteria nucleic acid samples, mainly because of the bacterial nucleic acid extraction rate in simulated bacteremia blood samples. When the bacterial nucleic acid is extracted, human blood, EDTA in the anticoagulation tube, whether the cell walls are easily digested and cleaved by enzymes, and the elution and enrichment of nucleic acid, various operating processes and factors have a negative impact during the bacterial nucleic acid extraction. Even so, the rate of extracting bacterial nucleic acid from bacterial blood samples by the ddPCR system was between 15 and 80%. We found that when the concentration of bacteria in the blood was greater than ten cells/ml, the ddPCR system detected bacteria stably. At present, clinical laboratories determine whether bacteremia is present based on the blood culture instruments reporting positive, and the reporting time is about 6-18 h. Generally speaking, it takes about three days from the collection of blood samples to the completion of bacterial culture and antibiotic sensitivity testing. Suppose we design unique probes and primers for specific bacterial and drug-resistance genes. In that case, the ddPCR system can identify bacterial species and drug-resistance genes simultaneously, with detection times less than 4 h. Therefore, our research suggests that the ddPCR technical platform is superior for the identification of pathogenic microorganisms. The ddPCR system accurately and effectively detect pathogenic microorganisms, allowing for more targeted antibiotic therapy and minimizing misuse of antibiotics, drug toxicity, and drug resistance, all of which could reduce economic burdens on patients.
Currently, the coronavirus (COVID-19) epidemic is spreading all over the world. The pathogenic microorganism of this pneumonia has been identified as a new coronavirus (Severe Acute Respiratory Syndrome Coronavirus 2, SARS-CoV-2). In acute respiratory infections, reverse transcription-polymerase chain reaction is usually used to detect pathogenic viruses from respiratory secretions in nucleic acid testing. It is the "gold standard" for the diagnosis of related cases 16 . However, the RT-PCR test result can also produce false negatives 17 . The reasons are the time of specimen collection, the quality or type of specimen, the transportation of the specimen, the ability of the virus to mutate, and the low sensitivity of RT-PCR and polymerase chain reaction inhibition [18][19][20] . Therefore, an efficient, fast, and accurate laboratory diagnosis technology is the key to ensuring early diagnosis, timely treatment, and preventing the development of the epidemic. Based on its advantages, digital PCR provides a new, efficient, and accurate detection method for pathogenic microorganisms such as viruses and bacteria detection.

Methods
Bacterial strains, experimental reagents, and instruments. The bacterial strains used were provided by the Laboratory of Peking University People's Hospital (Table S1), and the reagents and instruments used were provided by the laboratory of Tsinghua University (Tables S2 and S3).

Extraction of bacterial nucleic acid. The bacterial strains were propagated and cultured in a 50 ml
Luria-Bertani bacterial culture medium. A total of four strains of bacteria were used in the study, S. aureus ATCC25923, S. aureus ATCC29213, E. coli ATCC25922, and a strain of E. coli named E. coli-clin extracted from a patient. According to the methods suggested in the manufacturer's instructions, nucleic acids were extracted from the four bacteria using the QIAamp cador Pathogen Mini Kit (QIAGEN, Germany). We used Nanodrop to measure the extracted nucleic acid concentrations and stored samples at -20 °C.
Design of primers and probes. The specific genes of bacterial strains were selected according to the Gen-Bank database. The specific genes of E. coli and S. aureus were SWG-9 and coagulase COA gene, respectively. Primers and probes were designed for conserved regions in SWG-9 and coagulase COA gene sequences (Table S4 and S5). We used Oligo 7.0 and Primer Express 3.0 software to design the primers and probes for the nucleic acid sequence containing the site to be detected. The sequences were delivered to Thermo Fisher Scientific (CN) to synthesize the primers and probes. The 5 ' end of the E. coli probe was labeled with a 6-FAM fluorescent molecule, the 5′ end of the S. aureus probe was labeled with VIC fluorescent molecule.
Workflow of the ddPCR system. According to the manufacturer's instructions, we carried out ddPCR using the QX200 ddPCR system (Bio-Rad, CA). The QX200 ddPCR system (Bio-Rad, CA) consists of four steps: (1) preparation of the reaction mixture using a ddPCR amplification volume 20 µL (Table S6); (2) droplet generation; (3) ddPCR amplification (Table S7); and (4) droplet reading and analysis of results using QuantaSoft 3.0 software. Detection of bacterial strains using a real-time quantitative PCR system. The real-time quantitative PCR system with the MGB probe system was used to quantify the gradient-diluted nucleic acid of E. coli standard strain (ATCC25922) to determine the sensitivity range of the two real-time quantitative PCR. The template dilution gradient (copies/20 µL) of ATCC25922 was 10 6 , 10 5 , 10 4 , 10 3 , 10 2 , 10 1 , 1, and 0. Each group of experiments was repeated three times in parallel.
Detection of bacterial strains by ddPCR system. We used the ddPCR system to quantify the gradientdiluted nucleic acid of E. coli (ATCC25922) to determine the sensitivity range of the ddPCR system. The template dilution gradient (copies/20 μl) of ATCC25922 were 10 5 , 10 4 , 10 3 , 10 2 , 25, 6, and 0. Each group of experiments was repeated three times in parallel.
The ddPCR system simultaneously detects two bacterial strains. Various numbers of DNA templates, specific primers, and probes (ATCC25922: 5′-FAM/3′MGB, ATCC25923: 5′-VIC/3′MGB) of the two bacterial strains were added to a ddPCR reaction system (Group A, bacteria counts of E. coli: bacteria counts of S.