The misuse of antibiotics has resulted in a rapid increase in antimicrobial resistance among clinically relevant microorganisms, particularly gram-negative bacteria (Al-Tawfiq et al., 2020; Koulenti et al., 2019; Maraki et al., 2020; Wang et al., 2017; Yusef et al., 2018). Resistance to beta-lactam antibiotic is an emerging problem in health care due to the extremely limited therapeutic options. Additionally, resistance to this group of antimicrobials is often associated with resistance to other drugs (Koulenti et al., 2019; Tooke et al., 2019). The emergence of ESBL enzymes highlights the critical importance of understanding how associated genes could change abundances in gastrointestinal microbiome under prolonged antibiotic pressure. Since it is difficult to study this phenomenon using culturomics based approaches and subsequent detection of phenotypes, because the cultivation requirements for majority of gastrointestinal microbiome representatives are not known, the only alternative is to employ molecular biology approach. Therefore, we developed a panel for detection and long-term abundance and prevalence assessment of BLs in H. pylori-infected patients before and after a single eradication event.
Within the samples of experimental sample group, we were able to identify majority of the targeted ESBL clusters. However, there were also some BL gene clusters that were not detected by our panel. Although the most obvious explanation to this would be the absence of target within the pool of extracted bacterial DNA, it is also possible that in some cases the primer was primarily binding to off-target region due to high sequence homology or failed to bind entirely to intended target due to such wet-lab related aspects as incompatibility with employed annealing temperature. However, to evaluate the validity of these speculations it would be necessary to perform detailed examinations involving qPCR or even digital droplet PCR.
Acquired results revealed that most of BLs originated from gram-negative bacteria from genus Enterobacteriaceae. This finding is in agreement with results of several studies, which demonstrated that the prevalence of ESBL-producing Enterobacteriaceae is increasing, even in healthy asymptomatic individuals (Higa et al., 2019; Karanika et al., 2016), where it is believed to serve as a reservoir for the spread of ESBLs. The most abundant and prevalent ESBLs within the experimental group were blaEC from carbapenem-targeting class C ESBLs (Ur Rahman et al., 2018), the cephalosporinase gene cblA from class A BL targeting ESBLs (Smith et al., 1994) and blaACI from class A ESBLs. (Ur Rahman et al., 2018)As infections caused by ESBL producers are associated with increased mortality, length of hospital stay, and increased treatment cost (Ahn et al., 2017; Coque et al., 2008; Sianipar et al., 2019), the widespread identification of ESBLs in our patient samples indicates that this is indeed a growing problem in modern society. To date, a large proportion of ESBL types have been poorly characterized and the focus is mainly on the clinically relevant ones. Therefore, decoding the effect of ESBLs in the host or environment in the context of our study is challenging. However, large-scale ESBL gene identification in patient samples by the most sensitive molecular methods can still provide valuable information on their epidemiology.
Considering the abundance of ESBL clusters at the level of each sample, we found that most were absent in the individual samples of all three datasets, with the lowest number of clusters found in samples from the metagenomic dataset. While this might seem a striking result, we also observed that there are several overlapping clusters between the datasets, possibly reflecting the main population of ESBLs typical for our patient group. Four of these clusters originated from class A, three from class C, one from class D, and one from an unknown BL class. Furthermore, the number of ESBL clusters differed between the treatment states. In all three groups (experimental, validation-targeted, and validation-metagenomic)–the number of ESBL gene clusters was higher in the pre-eradication state than in the post-eradication state, suggesting that during eradication therapy a proportion of ESBL producers might be eliminated, resulting in an overall decrease in BL producers. Generally, monotherapy is used to treat common infections such as pneumonia (Rossio et al., 2015) and urinary tract infections (Dowson et al., 2020), while dual antimicrobial treatment is used for H. pylori. Therefore, it is plausible that the use of two types of antimicrobial agents may have a synergistic effect that favors the reduction of BL producers (Schmid et al., 2019). If true, it is possible that one of the ways to limit the spread of antimicrobial resistance would be the simultaneous use of several antimicrobial agents, not only in the case of H. pylori infection, but also in the case of other infectious diseases.
In this study, we were able to detect several ESBL clusters whose relative abundance differed significantly between the pre- and post-eradication states in both targeted datasets. The annotated taxonomic source indicated that these ESBLs originated from Klebsiella sp. (Sarojamma and Ramakrishna, 2011), Pseudomonas sp. (Laudy et al., 2017), Acinetobacter sp. (Abdar et al., 2019) Achromobacter xylosoxidans (El Salabi et al., 2012), Stenotrophomonas maltophilia (Adegoke et al., 2017) and others. Most of these are known ESBL producers, suggesting gastrointestinal carriage and asymptomatic colonization by these organisms. While the top abundant BL groups were common between the treatment states, the annotated taxonomic source of BLs differed between the experimental and targeted validation groups, indicating, that resistance genes in the post-eradication group were uptaken from the environment during the reestablishment of gut microbiome. If true, our findings suggest that in the future greater emphasis should be paid to the development of novel products and procedures for controlled gut microbiome reestablishment. These products should ensure that patients’ gut is colonized by non-resistant microorganisms thus mitigating the situation where our microbiota serves as an ESBL reservoir.
Considering the prevalence of ESBL clusters, the highest number of clusters was found in the experimental and targeted validation groups. The metagenomic dataset contained only a few ESBL gene clusters, suggesting that ESBLs abundances are at very low level. A precise analysis of the presence or absence of each cluster is possible by amplifying the perspective BLs or by increasing the target read count in metagenomic analyses. However, the amplification step within ESBL panel-based sequencing library preparation might induce bias in the abundance of certain ESBL gene clusters, especially those that share sequence homology.
In this study, we also evaluated the effect of H. pylori eradication therapy on the taxonomic composition of the gut microbiome using data obtained from the shotgun metagenomic sequencing analysis. The microbiome diversity rate was significantly higher in the pre-eradication state than in the post-eradication state. This observation agrees with literature data, which suggests that the gut microbiota is significantly altered immediately after eradication therapy and gradually restores to baseline parameters over the time; however, certain alterations may persist up to a year after completing an eradication therapy (Gudra et al., 2020; Liou et al., 2019; Martín-Núñez et al., 2019; Yap et al., 2016). While we observed that diversity was lower in the post-eradication state than in the pre-eradication state, we did not observe significant changes in the beta diversity analysis, suggesting that the global composition of microbial communities between the pre- and post-eradication states was highly similar. However, during the association analysis, the abundance of certain microbial species varied between the pre- and post-eradication states. For instance, in the pre-eradication subgroup, we observed increased levels of Acidaminococcus intestinalis, which has been previously shown to be increased in overweight adults (Palmas et al., 2021). Similarly, we detected increased levels of Collinsella aerofaciens and Treponema succinifaciens in the pre-eradication subgroup. The former has been associated with low dietary fiber intake (Gomez-Arango et al., 2018), whereas the later was enriched in traditional rural populations (Angelakis et al., 2019). Altogether, minor differences exist in the composition of the microbiome between the pre- and post-eradication states, although these differences could be more related to the diet and general state of health and not to the eradication therapy itself. In our previous study (Gudra et al., 2020) we evaluated the long-term effects of H. pylori eradication on the gut microbiome using patient samples from the current experimental group and analyzed them by sequencing the V3 region of the 16S rRNA gene. In that study we found that the gut microbiome was stable over two years and was more associated with subject-specific parameters, such as age and medical history than to the eradication therapy itself. These results are comparable to the current study concluding that in the long-term gut microbiome recovers after single antibiotic intervention.
In addition to the evaluation and validation of the ESBL screening panel, we investigated the functional profile and resistome of the microbiome. Thus, we were able to show that abundancy of several genes was increased in the post-eradication state. These genes have a role in molecule and ion transport; cobalamin and extracellular polysaccharide biosynthetic processes; and methylglyoxal, arabinose, and glutamine metabolic processes. All the above-mentioned processes have a positive and beneficial impact on the human host, most profoundly in the case of cobalamin and extracellular polysaccharide synthesis. Additionally, our data also suggests that abundance of several genes were decreased in the post-eradication state. These genes have a role in the DNA restriction-modification system, DNA-templated transcription and initiation, and in ribosomal small subunit biogenesis. Their increased levels might be associated with active bacterial reproduction and their genomic defense from invading foreign DNA. In addition, since we detected minor differences in the abundance and prevalence of ESBL gene clusters between the treatment states, we also evaluated the entire resistome profile. The number of AMR genes detected in the post-eradication subgroup was higher than that detected in the pre-eradication subgroup. Thus, it is apparent that the diversity of AMR genes has increased under the pressure of antimicrobial therapy. There were three AMR genes that were present in all samples of the pre-eradication subgroup – resistance-nodulation-cell division antibiotic efflux pump gene adeF, tetracycline-resistant ribosomal protection protein gene tetQ, and trimethoprim resistant dihydrofolate reductase dfr gene dfrF, while four were detected in all samples of the post-eradication subgroup: Erm 23S rRNA methyltransferase gene ErmF, major facilitator superfamily antibiotic efflux pump gene tet(40), and tetracycline-resistant ribosomal protection protein genes tetO and tetW. Furthermore, the number of AMR genes that confer resistance to macrolides increased in the post-eradication subgroup. In one study subject we were able to detect Chlamydia trachomatis 23S rRNA with mutations conferring resistance to macrolide antibiotics, such as clarithromycin, which was prescribed to the study participants in the current study. Although C. trachomatis is commonly associated with sexually transmitted diseases (Malhotra et al., 2013), it has also been shown that the human gastrointestinal tract might be a site of persistent infection by this pathogen (Borel et al., 2018; Yeruva et al., 2013). Moreover, in the post-eradication state, we observed an increase in the AMR gene families, macrolide esterase and macrolide phosphotransferase, both of which contribute to macrolide antibiotic inactivation. Alterations in the number of AMR genes after H. pylori eradication therapy have been reported in a few studies (Olekhnovich et al., 2019; Wang et al., 2021). Nonetheless, little is known about the functional mechanisms of gut microbiome dynamics following antibiotic treatment.
Despite all this wealth of knowledge this study had several limitations. First, some of the targeted ESBL-coding gene clusters were absent in all patient samples and the reason behind their absence remained ambiguous. Therefore, the developed primers that were targeting ESBL-coding gene clusters should be further validated using such methods as RT-qPCR or digital droplet PCR. Next, while the validation group was able to mimic the experimental group, greater patient involvement is needed to increase the resolution of the diversity and abundance of ESBL and AMR coding genes, especially within the metagenomic dataset. In this study, we also observed that some individuals remained H. pylori-positive after eradication therapy, but the sample size was too small to confirm this observation. Lastly, this study did not evaluate resistance to the prescribed antibiotics, amoxicillin and clarithromycin, and did not involve the genomic characterization of patients’ H. pylori. However, considering the high prevalence of H. pylori in the Latvian population, we believe that such studies are of paramount importance and should be addressed in the near future.
Our study suggests that NGS based large-scale ESBL coding gene screening panel can be used for accurate population screening and surveillance of ESBL genes in symptomatic and asymptomatic infections. The applicability of the currently developed methodology is not limited to ESBL encoding gene determination in the gut microbiome of H. pylori-infected patients, but could also be potentially applied to different samples, populations, and various infection cases that encounter increased resistance to cephalosporins, amoxicillin, penicillin and other. In addition, these results suggest BL recolonization during restoration of the gut microbiome, implying that greater microbiome control would be necessary after antibiotic treatment. In conclusion, we believe that ESBL screening panel is suitable for screening changes in prevalence of ESBL coding genes and in-depth research of the resistome is required to better understand the reservoir of AMR genes in relation to antibacterial therapy, which in future could aid clinicians when choosing antibacterial therapy.