The mutations in 23S rRNA and parC have been identified as the cause of failure of macrolide and quinolone in the treatment of M. genitalium [53,54]. In our study, 66.4% (101/152) and 77.7% (108/139) of samples manifested mutations in 23S rRNA and parC genes, and A2072G (59/101, 58.4%) and S83I (79/108, 73.1%) were highly predominating in 23S rRNA and parC genes, respectively. More worryingly, the proportion of mutations in both 23S rRNA and parC genes was as high as 48.6%. This suggests that nearly half of these samples are resistant to both macrolide and quinolones. We also reported other mutations in parC and gyrA genes. However, the significance of these mutations requires further study.
Within the last 10 years, M. genitalium eradication rate has declined gradually [55, 56]. The resistance rate of M. genitalium has been described as a rising phenomenon in many countries [42, 57]. At the time of this study, there are only three locations actively conducting AMR-related research in clinical settings in China. The earliest published macrolide-associated mutations in M.genitalium in China collected samples from 18 symptomatic NGU patients . In this research, the 23S rRNA mutation rate was 94.4%, with A2072G being the most common (55.6%), A2071G the second most (27.8%), and A2071T as the third most common mutation (11.1%), with no double-mutations detectable . Later, in the same hospital, 358 M. genitalium positive samples were collected. The 23S rRNA mutation rate was 88.9%, with A2072G being the most common (61.9%), A2071G the second most (17.6%), and no double-mutations were detected . The parC mutation rate was 90.4%, S83→I was the most common mutation (83.7%) . The double mutation in parC (G248A + G259T) was detected . The gyrA mutation rate was 13.0%, with M95→I being the most common (5.3%), three double-mutations in G244A + G285A, G285A + A309G, and G285A + A317G were detectable . Another earlier study collected samples among men seeking care at an infertility clinic in Changsha, a city in the interior of China . The macrolide mutations rate was similarly extremely high at 96.7% . The two most common mutations in the Nanjing study are also the most frequent mutations in Changsha, that is, A2072G (60.0%) and A2071G (20.0%) . Unlike in Nanjing, the analysis conducted on specimens from Changsha detected double-mutations, and these mutations are frequent enough to be the third most common set of mutations (A2071T + A2072G at 11.7%). Our location, a STI center based in a hospital in Guangzhou, constitutes the third AMR site. Our facility is a provincial STI center situated in Guangzhou, the capital city of Guangdong Province. Guangzhou is an international hub for travel, trade, and commerce and a major destination for migrants and their concomitant illnesses. Servicing the medical needs of such a diverse population, we focus on macrolide and fluoroquinolone resistance-associated mutations in M. genitalium. We extend current knowledge in two key ways. First, we continue monitoring and reporting efforts on macrolide and fluoroquinolone resistance, expanding on reports from the two prior studies based in central and interior China, by adding a major urban migration destination in south China. Second, we expand on AMR surveillance by being the first to report on macrolide and fluoroquinolone-associated mutations in men and women in China.
The 23S rRNA mutation is associated with macrolide resistance . We found that SNPs in region V of the 23S rRNA gene were observed in 101 (66.4%) samples from male and female patients with M. genitalium-positive infection in 2016-2018. Mutations mainly occurred at positions A2071 and A2072 mainly to G (C or T is relatively less). With the exception of a study from Greenland, the mutation frequency (66.4%)  observed was higher than frequencies reported by Russia and Estonia (0.7～10%) , South Africa (10%) , southern Sweden (13%) , France (17%) , Japan (42.2%) , southern USA (48%) , Norway (56%) , and Denmark (57%) . However, our rate of 66.4% is lower than rates reported from England (82.4%) , the US (Alabama: 74.1% HIV positive MSM) , and Australia (79.4%) .
It is widely reported that M. genitalium expressed a diversity of mutations linked to fluoroquinolone resistance-associated in gyrA and parC gene [49, 53, 68]. Similar to extant studies, mutations in the QRDR of the gyrA gene of our samples were rarely detected [53, 71]. The amino acid changes (M95→I and A96→T) in gyrA were found in our specimens. The M to I transition at position 95 of gyrA (G to C at nucleotide position 285) was first reported in 2013 by Tagg et al , most commonly observed from 2013 to 2017 in Japan, and have been reported in moxifloxacin-resistant strains of M. pneumoniae, M. hominis, and Ureaplasma spp [49, 57, 72, 73]. To our knowledge, a gyrA A96→T mutation in the core of the QRDR has not previously been described in M. genitalium and its association with resistance to fluoroquinolone remains unknown. The amino acid changes at G81, S83 and D87, have been previously reported as being associated with fluoroquinolone resistance in M. genitalium and other closely related organisms [44, 46, 49]. Although the majority of published reports have shown the parC S83N and S83I substitution as the two most prevalent base changes at position 248, we find that the S83I substitution accounted for 71.8% (79/110), significantly higher than reports from Japan (13.0-23.2%) , New Zealand (16.7%) , and southwestern France (9.1%) . Among the 139 samples successfully amplified DNA sequences of parC gene, we observed an exorbitantly high mutation rate of 77.7%.
Additionally, 48.6% (67/138) of samples were multidrug resistant and contained both macrolide and fluoroquinolone resistance related SNPs. If SNP on parC is strictly limited to S83I, the multidrug resistance rate was 36.1% (50/138). In Japan, the prevalence of multidrug resistance with A2071G or A2072G in the 23S rRNA and amino-acid change in S83 or D87 of parC has been reported in up to 21.8% from 2010-2017 . Our data showed very high prevalence of the same mutation. This trend of multidrug resistance presents challenges for clinicians because of a lack of suitable alternative therapy after azithromycin and moxifloxacin failure. Pristinamycin as the only third-line treatment has been reported to be only about 75% effective and is not readily available in China .
The high prevalence of mutations in macrolide and quinolone resistance-associated genes observed in our study might be related to the study population and to antibiotic overuse in China. The clinical samples were collected from the STI clinic of Dermatology Hospital, Southern Medical University. As a provincial level STI center, our doctors are referred patients from all over the region when doctors from feeder hospitals are unable to resolve medical ailments locally. These patients likely experienced several prior courses of antibiotic treatment. In addition, in China, it is incredibly easy for the public to obtain antibiotic prescriptions and purchase antibiotics in pharmacies. Data show that antibiotic use in children and hospitalized patients in China is very high . These factors further exacerbate the problems of antibiotics resistance confronting health facilities today .
An important limitation of the study is the lack of epidemiological and clinical information, as well as information about treatment received and clinical evolution of the patients. The significance of several novel mutations in the parC and gyrA genes remains unknown. Nonetheless, the prevalence of mutations associated to macrolide and fluoroquinolone resistance in our study related to phenotypic testing has been previously reported in several studies. Our prevalence rate is a calculation based on a sample of patients seen by clinicians at our STI clinic. During patient intake and consultation, we did not collect patients' history of previous antibiotic use. We hypothesized that patients at our STI clinic were more likely to have previously used antibiotics than the general population, so there might be a possibility of overestimating the prevalence rate when extended to the general population. Secondarily, we lack data for a large sample epidemiological survey of M. genitalium, since samples studied were collected mainly from a single clinic. Hence, our findings might not be representative or readily generalizable to the larger population living in Guangzhou.