From April 2022 to March 2023, 105 MG-positive samples were used as validation samples for nested real-time PCR experiments to evaluate laboratory performance, such as concordance rate with Sanger sequencing method, limit of detection (LoD), and so on. Of these samples, 17 were male urogenital swabs and 88 were female urogenital swabs; 87 cases were from pre-treatment samples and 18 cases were from test-of-cure (TOC) samples. Each three specimens with different common urogenital pathogens and Mycoplasma pneumoniae have been used for cross-reactivity validation, respectively. The study was approved by the Medical Ethics Committee of the Third Affiliated Hospital of Guangzhou Medical University (No.2022044).

MG was detected by real-time simultaneous amplification and testing (SAT) (Rendu Biotechnology Co., Shanghai, China) with a LoD of 1000 copies/reaction. Nucleic acid was extracted from 300μL eluent of positive samples using the Smart Lab Automatic Nucleic Acid Extractor (Advanced Nanotech Co., Taiwan, China) and DNA/RNA Extraction Kit (Health Gene Technologies Co., Ningbo, China). 50μL of nucleic acid samples are stored at -80°C until use.

Step 1: The first round of PCR experiments was performed according to the instructions of the Platinum SuperFi II PCR Master Mix Kit (CAT# 12368010, Thermo Fisher, USA) with a total reaction volume of 50 μL: 25 μL of 2 × Platinum SuperFi II PCR Master Mix, 2.5 μL primers (10 μM), 5 μL nucleic acid sample and 15 μL water. The primers used as follows: 23S rRNA 5’- CAACTCTATGCCAAACCGTAAG-3’ and 5’-GTC AAACTGCCCACCTAACAC-3’, parC 5’-AAACCAGTACAAAGACGGATCT-3’ and 5’-GAGGTTAGGCAGTAAGGTTGG-3’.

The reaction was performed on a Veriti PCR instrument (Applied Biosystems, USA) under the following PCR conditions: 98°C, 30 seconds, 1 cycle; 98°C, 10 seconds, 60°C, 10 seconds, 72°C, 30 seconds, 35 cycles; 72°C, 5 minutes and 4°C hold.

Step 2: The PCR product of Step1 was diluted 1:1000, and real-time PCR was performed according to the manual of the FastFire qPCR PreMix (Probe) kit (CAT# FP208, Tiangen Biotechnology Co., Beijing, China), with a total reaction volume of 20 μL, including 10 μL of 2 × qPCR Mix, 0.4 μL primers (10 μM), 0.2 μL probes (10 μM), 2 μL diluted Step1 PCR product, and 7 μL water. A Cobas z480 fluorescence PCR instrument (Roche Diagnostics, Germany) was used for detection, with the following reaction conditions: 95°C, 5 minutes, 1 cycle; 95°C, 10 seconds, 70°C, 30 seconds (FAM channel acquisition), 25 cycles; 37°C, 30 seconds, 1 cycle. All primers and probes above were designed by Primer Premier software (Version 6, Premier Biosoft Inc., CA), then were synthesized by Guangzhou IGE Biotechnology Co. Primer and probe sequences used in fluorescent PCR experiments are shown in Table 1 . Seven probes were used to detect mutations at A2058 and A2059 positions of 23S rRNA, one probe to detect A2060 of 23S rRNA was used as a negative control, four probes were used to detect mutations at S83 amino acid site of parC, and four probes were used to detect mutations at D87 amino acid site of parC.

In order to check the amplification efficiency of the fluorescent PCR phase, wild-type plasmids were diluted in a ten-fold gradient, followed by detecting with S1 and P1 primer/probe set. The primer/probe efficiency was calculated automatically by Roche z480 instruments based on the standard curves from three independent experiments.

Twelve pGSI plasmids containing different mutant fragments (23S rRNA A2058C, A2058G, A2058T, A2059C, A2059G, A2059T; parC A247C (S83R), G248T (S83I), G248A (S83N), G259A (D87N), G259T (D87Y), A260G (D87G) and the pGSI plasmid containing the wild-type fragment were used to validate the LoD of the overall experimental protocols. Different plasmids were diluted to 100000, 10000, 1000, 500, 200, 100 and 50 copies/reaction and the experiments were repeated at least ten times. The lowest concentration with 100% amplification is the LoD of each genotype and the LoD of the whole experiment is the highest LoD of all genotypes.

Twenty-four urogenital swab specimens and three pharyngeal swab specimens of Mycoplasma pneumoniae were confirmed for the presence of different pathogens using National Medical Products Administration (NMPA)-approved reagents from Liferiver BioTech Co., including Ureaplasma urealyticum, Mycoplasma hominis, Chlamydia trachomatis, Neisseria gonorrhoeae, Human papillomavirus, Herpes simplex virus, Candida albicans, Ureaplasma parvum and Mycoplasma pneumoniae. All these samples were MG-negative and were used to verify the cross reaction of the nested real-time PCR assay with other pathogens.

105 MG-positive samples were detected by the evaluated method and Sanger sequencing respectively, and the consistency of the results was assessed.

All the Step1 PCR products of MG-positive samples were subjected to 1% agarose gel electrophoresis before sequencing to analyze their specificity, and then sequencing experiments were performed by Guangzhou IGE Biotechnology Co. Step1 PCR products were sequenced in both directions, then were analyzed by Chromas software and compared with Mycoplasma genitalium G37 strain genome sequence (GenBank NC _ 000908). The 23S rRNA mutation sites are represented by the E. coli numbers. The parC mutations and amino acid sites are represented by the Mycoplasma genitalium numbers. Data were analyzed using SPSS software, version 19.0 (IBM Co., Armonk, NY, USA). All data were expressed and analyzed as crosstabs, the Wilson test with a correction for continuity was used to calculated the 95% confidence interval (CI) of a proportion.

For each sample, 16 fluorescent PCR amplification reactions were performed simultaneously (including a negative control). The 23S rRNA and parC genotypes of each sample were assessed comprehensively by combining the detection curves of seven 23S rRNA probes and eight parC probes. The fluorescent PCR detection results of the three independent samples are shown in Figure 1 . If the sample contain a single MG population, three amplification curves will appear, respectively for 23S rRNA, parC S83 and parC D87 regions, as shown in Figures 1A , B . The reactions with the S1, P3 and P5 probes showed an S-shaped amplification curve in the sample shown in Figure 1A , and the corresponding genotypes were 23S rRNA WT and parC G248T (S83I). The reactions with the S3, P1 and P7 probes showed an S-shaped amplification curve respectively in another sample shown in Figure 1B , and the corresponding genotypes were 23S rRNA A2059G and parC G259T (D87Y).

In fact, mixed MG populations may be present in individual samples. Figures 1C, D showed the nested real-time PCR and sequencing results of one sample containing mixed mutants. The reactions of nested real-time PCR with probes S3, P1, P3, P7 and P5 exhibited an S-shaped amplification curve, and the corresponding genotypes were 23S rRNA A2059G, parC G248T (S83I)/G259T (D87Y) or G248T (S83I)/G259T (D87Y)/WT. These mixed MG populations could also be distinguished by the sequencing results.

To demonstrate the specificity of this method, 1% agarose gel electrophoresis was performed with the real-time PCR products (Step 2) from a wild-type sample. The results showed that there were bands in the S1, P1 and P5 lanes ( Figure 1E ), with the S1 fragment 147 bp in length and the P1 and P5 fragment 83 bp in length, as expected.

In the real-time fluorescence PCR stage (Step2), the S1 primer/probe set efficiency was 1.900 ± 0.021, and the P1 primer/probe set efficiency was 1.838 ± 0.044.

Plasmids with different concentrations were detected ten times with the corresponding probes respectively. At a concentration of 200 copies/reaction, all 23S rRNA reaction systems can be amplified, whereas at a concentration of 500 copies/reaction, all parC reaction systems can be amplified. In summary, the LoD of the nested real-time PCR was 500 copies/reaction.

The assay didn’t cross-react with common urogenital pathogens such as Ureaplasma urealyticum, Mycoplasma hominis, Chlamydia trachomatis, Neisseria gonorrhoeae, Human papillomavirus, Herpes simplex virus, Candida albicans and Ureaplasma parvum ( Figures S1A–H ). There were no S-shaped amplification curves for any of the assay systems. However, the 23S rRNA assay cross-reacted with Mycoplasma pneumoniae ( Figure S1I ).

The detection rate of the nested real-time PCR method was 100% (105/105). 23S rRNA sequencing failed in 2 samples and parC sequencing failed in 7 samples. A comparison of the nested real-time PCR and Sanger sequencing methods is shown in Table 2 . Most of the results were consistent. However, in 5 samples, the results of the two methods were not consistent, including 2 23S rRNA WT by sequencing but WT/A2059G by the nested real-time PCR, 1 parC WT by sequencing but WT/G259A (D87N) by the nested real-time PCR, and 2 parC WT by sequencing but WT/G248T (S83I) by the nested real-time PCR. The evaluated method showed that there were mixed MG populations in the sample, but the sequencing results showed that it was unable to distinguish the low peak at the mutation site from background noise on the sequencing map.

The sensitivity and specificity of 23S rRNA were 100% (95% CI; 93.3 - 100) and 94.3% (95% CI; 79.4 - 99.0), and the consistency was 98% (95% CI; 92.5 - 99.7) with a kappa value of 0.96 (P < 0.001). The sensitivity for parC was 100% (95% CI; 93.4 - 100), the specificity was 89.7% (95% CI; 71.5 - 97.3) and the consistency was 96.9% (95% CI; 90.7 - 99.2). The kappa value was 0.92 (P < 0.001).