Extended role for insertion sequence elements in the antibiotic resistance of Bacteroides

Abstract

The Bacteroides species are important micro-organisms, both in the normal physiology of the intestines and as frequent opportunistic anaerobic pathogens, with a deeply-rooted phylogenetic origin endowing them with some interesting biological features. Their prevalence in anaerobic clinical specimens is around 60%-80%, and they display the most numerous and highest rates of antibiotic resistance among all pathogenic anaerobes. In these antibiotic resistance mechanisms there is a noteworthy role for the insertion sequence (IS) elements, which are usually regarded as representatives of ‘selfish’ genes; the IS elements of Bacteroides are usually capable of up-regulating the antibiotic resistance genes. These include the cepA (penicillin and cephalosporin), cfxA (cephamycin), cfiA (carbapenem), nim (metronidazole) and ermF (clindamycin) resistance genes. This is achieved by outward-oriented promoter sequences on the ISs. Although some representatives are well characterized, e.g., the resistance gene-IS element pairs in certain resistant strains, open questions remain in this field concerning a better understanding of the molecular biology of the antibiotic resistance mechanisms of Bacteroides, which will have clinical implications.

Key Words: Antibiotic resistance, Antibiotic resistance genes, Bacteroides fragilis, Insertion sequence elements

SIGNIFICANCE OF BACTEROIDES

The species of the genus Bacteroides are the most prominent human pathogenic anaerobic bacteria. Additionally, they have other important specialities: they are one of the most important members of the mammalian normal intestinal microbiota and they are the best-studied organisms of a separate and early diverged phylum, Bacteroidetes, of Bacteria. As regards their pathogenic nature, they account for 60%-70% of the total anaerobic pathogens cultivated from clinical samples, and despite the relatively low number of such materials, they often cause high mortality in various infectious processes, such as abscesses and other soft tissue infections, and often cause anaerobic sepsis[1].

Their phylum is a phylogenetic relative of the group of green-sulfur photosynthetic bacteria, the Chlorobiales, and the best-known and most frequently isolated species, Bacteroides fragilis (B. fragilis), as a type species for anaerobic bacteria, is often referred to as the anaerobic Escherichia coli[2]. B. fragilis was first isolated as ‘B. fragilis’ and later renamed as B. fragilis. Until the late 1970s, almost all Gram-negative anaerobic bacilli were classified in this genus, and only the more recent molecular techniques, such as DNA-DNA homology measurements and 16S rRNA sequence comparisons, allowed a more accurate classification. Thus, the genera of Bacteroides, Prevotella and Porphyromonas were formed from the earlier Bacteroides genus during the late 1980s. 16S rRNA phylogeny and other molecular classification methods were then applied, making the picture more diverse[3]. The parent genus Bacteroides now contains 41 described and well-characterized species. Some other former Bacteroides species were reclassified into the newly formed Parabacteroides genus[3], which belongs in the Porphyromonadaceae family (P. distasonis and P. merdae) (Table 1). Bacteroidaceae, Marinilabiaceae, Porphyromonadaceae, Prevotellaceae and Rikenellacaea are the families of the Bacteroidales order. Together with other important but aerobic taxa (the Cytophagales, Flavobacteriales and Sphingobacteriales orders), they form the Bacteroidetes phylum. The current situation regarding the species of the Bacteroides and Parabacteroides genera, the subjects of the current review, is summarized in Table 1 with some implications in respect of their pathogenic potential. In a recent study, the phylogenetic relations between Bacteroides species were analysed by multilocus sequence analysis, and thus these species could be ranked into 10 subgroups also showing some common characteristics regarding their pathogenic nature and sites of isolation[4].

Table 1 A list of the related species comprising the ‘Bacteroides fragilis group’ at present.

Bacteroides			Parabacteroides
B. acidifaciens	B. fluxus	B. propionifaciens	P. distasonis1,2
B. barnesiae	B. fragilisa,b	B. pyogenes	P. goldsteinii1
B. caccae1,2	B. galacturonicus	B. rodentium	P. gordonii
B. cellulosilyticus	B. gallinarium	B. salanitronis	P. johnsonii
B. chinchillae	B. graminisolvens	B. salyersiae1	P. merdae1,2
B. clarus	B. helcogenes	B. sartorii
B. coagulans	B. heparinolyticus	B. stercoris1,2
B. coprocola	B. intestinalis1	B. thetaiotaomicron1,2
B. coprophilus	B. massiliensis1	B. uniformis1,2
B. coprosuis	B. nordiia	B. vulgatus1,2
B. dorei	B. oleiciplenus	B. xylanisolvens
B. eggerthii1,2	B. ovatus1,2	B. xylanolyticus
B. faecis	B. paurosaccharolyticus	B. zoogleoformans
B. finegoldii	B. plebeius

¹The main pathogenic species of Bacteroides that were included in an antibiotic susceptibility study and are most frequently isolated from clinical specimens; ²The 10 Bacteroides species earlier comprising the B. fragilis group are marked with a superscript b.

Genomic studies have revealed the important genetic characteristics of this group of anaerobic bacteria and contributed to extensive metagenomic analyses of their habitat, the intestines and the participating microbiota[5]. These studies have reconfirmed that Bacteroides species are important symbionts there and opened up new ways for the investigation of this firmly interacting ecosystem. Besides the earlier cultivation and microscopic methods, metagenomic analyses have also proved that the two most abundant taxa there are Bacteroidales and Firmicutes (low G + C Gram-positives)[5,6]. The composition of the mammalian intestinal microbiota depends on the type of food intake (herbivorous, carnivorous or omnivorous)[7], but in the case of human beings three enterotypes can be distinguished as regards the prevalence of the main abundant constituents (Bacteroides, Prevotella and Ruminococcus); it is suspected that this is determined by the host and does not depend on the geographic origin[8]. The Bacteroides as one of the groups of predominant constituents of the human intestines exert beneficial effects for the host. However, experiments involving the application of metagenomics suggest that the intestinal microbiota, including Bacteroides, can affect not only the food intake, but also the development and physiology of the intestines and the immune system, and such distant organs as the liver, muscles, circulation and central nervous system. Thus, their roles regarding participation in illness states such as obesity and inflammatory bowel diseases have been the subjects of previous and ongoing investigations[9-11].

Virulence mechanisms of Bacteroides spp.

Though Bacteroides can be regarded as only opportunistic pathogens since they reside in the intestines in high cell numbers and cause diseases with underlying predisposition circumstances such as trauma, circulation defects and immunosuppression, they usually possess a pathogenic repertoire with which they participate in infections. B. fragilis, the earliest identified and thus the type species, is isolated most frequently from anaerobic infections with a prevalence of 60%-70%. As it is estimated to have a prevalence in the intestines of only 0.5%-5.0% and to be localized to the epithelium rather than to the lumen, it can be regarded as the most pathogenic species among the Bacteroides, and this is supported by the experimental data[1]. The most frequent infections that it causes are intra-abdominal and intra-pelvic, lung and brain abscesses, appendicitis, diarrhea, inflammatory bowel disease, lower respiratory and soft-tissue infections, and sepsis. The main predisposing factors are usually surgery, mixed aerobe-anaerobe infections, immunosuppression, diabetes and circulatory defects. However, besides the prominent pathogenic role of B. fragilis, most Bacteroides are capable of adhering, evading and destroying the tissues with their direct and indirect virulence mechanisms, which are production of capsules, fimbriae and adhesins, tissue destruction enzymes (fibrinogenases, haemolysins, neuraminidase and enterotoxin) and properties of aerotolerance, evasion of the host immune system, and antibiotic resistance mechanisms[1].

The most potent virulence mechanism of B. fragilis has been demonstrated to involve certain capsular polysaccharide (CPS) species[12]. In the mid-1980s, the use of CPS material of B. fragilis was shown to evoke abdominal abscesses experimentally in a rat model, and the nature of the immuno-modulation involved the induction of a humoral response[13]. The chemical structures of two CPS species, CPS-A and B, of B. fragilis NCTC 9343 were later determined and their abscess-inducing properties were proved to be due to a zwitterionic structure[14,15]. The capsules participate in immuno-modulation by other usual modes of interactions, the inhibition of phagocytosis and complement action. Electron microscopically, B. fragilis may be seen to have small or large capsules or only an electron-dense layer which is implicated in complement resistance[16-18]. Studies have led to the cloning of these CPS operons[19,20], and subsequently altogether 8 operons with similar structures have been found in the genomic sequence of B. fragilis NCTC9343 that have a common regulatory property, the possession of invertible promoters[21]. This special regulatory feature may result in numerous variable surface compositions through activation of on-off switches (about 2⁸) in the case of a single strain. The examination of B. fragilis YCH46 and 638R genomes demonstrated that at least 10 CPS operons can be located on these genomes, which may have different alleles (n = 28), allowing a much higher number of variations in possible surface compositions[22]. The large and small capsule phenotypes are suspected of being regulated by the expression of the gene BF2782 (or BF2790 in B. fragilis 638R), which is a putative sugar transferase participating in the synthesis of the CPS species and is also the subject of invertible promoter structure[22]. Similar CPS operons are suspected of functioning under the regulation of invertible promoters in other Bacteroides species (B. caccae, B. ovatus, B. thetaiotaomicron, B. uniformis, B. vulgatus, P. merdae and P. distasonis), but not in the oral Bacteroides relatives (Prevotella and Porphyromonas)[23]. The CPS-A of B. fragilis is capable of regulating the maturation of the immune system which, in turn, is an important contribution to the overall symbiotic interactions between Bacteroides and the host[24,25].

Another important virulence factor is the enterotoxin of B. fragilis, which may cause diarrhea especially in young mammalians. This enterotoxin is a metallo-protease capable of the specific cleavage of the E-cadherin protein in the zonula adhaerens portion of the intestinal epithelium. This causes specific processes leading to the symptoms of diarrhea; disorganization of the actin cytoskeleton, epithelial fluid loss, inflammation, and possible penetration of the enterotoxinogenic B. fragilis cells into the nearby and distant tissues[26]. The inflammatory action of the B. fragilis enterotoxin may be so pronounced that the malignant transformation it causes can be detected both clinically and experimentally[27,28]. The genes of the enterotoxin, consisting of three main types (bft1-3), lie on a specific portion of a ‘pathogenicity island’, which is a conjugative transposon resembling other B. fragilis genome-borne conjugative transposons[29-31]. The similar CTn86 and CTn9343 elements have molecular variants resulting from (1) insertion of a ca. 6 kb region containing the bft genes into CTn86 (accordingly, bft-positive CTn86s are enterotoxinogenic, whereas bft-negative CTn86s are not); (2) replacement of the 3’ regions of both CTns; and (3) insertion of a novel ca. 7 kb region into some CTn9343s[32,33]. A more detailed summary of the pathogenicity and virulence factors of Bacteroides is to be found in an excellent recent review[1].

Antimicrobial resistance of Bacteroides spp. and its genetic background

As the Bacteroides are the most significant human anaerobic pathogens, detection of their antimicrobial susceptibilities has a significant history, and trends have been observed in the most frequent resistance rates and the most numerous resistance mechanisms among their clinical isolates. As time has passed, these latter resistance trends have become more pronounced. In the 1960s and 1970s, the strains were much less resistant to all groups of antibiotics than more recently[34]. In the meantime, the recommended susceptibility measurement methods have changed. Since the 1980s, the recommended method for the detection of their antibiotic susceptibilities has been agar dilution[35]. Regular studies have been carried out, especially in the United States and in Europe[36,37], and the breakpoint recommendations of the NCCLS (National Committee for Clinical Laboratory Standards, currently the Clinical Laboratory Standards Institute-CLSI, http://www.clsi.org) in the United States have been widely used for resistance categorization; additionally, we now have the recommendations of another influential body, EUCAST (http://www.eucast.org). Since the Bacteroides in the intestines are readily exposed to antibiotics administered orally and excreted into the bile, a continuous increase in resistance rates has been observed for all major antibiotics. The resistance to tetracycline has changed most profoundly, which may be explained by the intensive use of tetracycline and the fact that the spread of tetracycline resistance elements is highly enhanced by tetracycline (see the explanation below)[34].

The Bacteroides have displayed a significant rate of resistance to ‘normal β-lactams’ (penicillins and 1^st and 2^nd generation cephalosporins) throughout the studied periods, but some increases have also been observed. One important issue relating to the ‘normal β-lactam’ resistance is the breakpoint categorization, since the MIC values for all such drugs are scattered widely, ranging from the low 0.25 μg/mL to the very high 256 μg/mL. Thus in a 1990 European study, only 12% of B. fragilis strains were found to be resistant to ampicillin at a breakpoint of 32 μg/mL[38], whereas in a study in 2000 with 2/64 μg/mL as breakpoints, 99.3/27% were categorized as resistant[39]. It was additionally observed that the distribution of ‘normal β-lactam’, especially ampicillin, resistance distribution is bimodal, with modes at about 32 and ≥ 256 μg/mL[38]. Since the majority of Bacteroides isolates exhibit β-lactamase activities, this was proposed to be the main resistance mechanism[34]. The gene cepA of an Ambler Class A β-lactamase is very prevalent[40] (about 70%, according to our own unpublished observations) among B. fragilis and other Bacteroides strains. Little is known concerning the mechanisms of resistance to β-lactam/β-lactamase inhibitor combinations; however, the rates in the United States (< 1%) and in Europe (10.3%) have been increasing continuously in recent years, probably because of the extensive use of such drugs[36,37].

Cefoxitin was earlier a very effective antibiotic for the treatment of Bacteroides infections, but the levels of resistance to this drug rose by 6% and 10.3% in Europe and the United States, respectively[39,41], during the 1990s, though this has been followed by a decrease (12% vs 9%) in the past decade in the United States[36]. The main resistance mechanisms involve the decreased affinity of the penicillin-binding proteins to cefoxitin and the production of another Ambler Class A β-lactamase capable of hydrolyzing cephamycins[42]. The gene for this latter β-lactamase, cfxA, has been located on a mobilizable transposon MTn4555[43], which has been proved to have several variants at its 3’ end[44,45].

Some B. fragilis isolates are also resistant to the wide-spectrum carbapenems, due to a metallo-β-lactamase coded by the cfiA (ccrA) gene[46,47]. Despite the low prevalence of carbapenem-resistant B. fragilis isolates (about 1%), this has displayed a continuous rise since the introduction of these drugs[37,41]. It is very interesting that the cepA and cfiA genes are found mutually exclusively among B. fragilis isolates and define two genetic groups (Division I -cepA-positive and Division II-cfiA-positive)[48,49] that can also be differentiated by the levels of DNA-DNA homologies[50], PCR typing methods[48], ribotyping[51], multi-locus enzyme electrophoresis[49] and MALDI-TOF mass spectrometry[52,53].

The rate of resistance to the macrolide-lincosamide-streptogramin antibiotics of 32.4% is not an exception; the rates of resistance presumed to be caused by ermF genes has a steep rise[37]. Our recent investigations based on susceptibility measurements and resistance gene detection (unpublished) of clinical Bacteroides isolates revealed that other resistance genes (ermB, ermG, mefA and msrSA)[54,55] may participate significantly in the development of clindamycin resistance.

Resistance to 5-nitroimidazoles is caused either by alterations in the cellular redox system that can diminish the lethal action of these drugs or by 5-nitroimidazole reductases that reduce the nitro group of 5-nitroimidazoles to an amino group without the formation of toxic intermediates[56-61]. The 5-nitrimidazole reductases are coded by nim genes that bear about 60%-70% mutual homologies and have 9 representatives (nimA-I)[60,62-64]. The form nimI has been found only among Prevotella baroniae isolates, while some Bacteroides-specific nim genes are present in other source organisms too[65-69]. The rates of resistance to metronidazole are fortunately very low among Bacteroides strains in most places (< 1%).

Tetracycline resistance has been estimated to be approaching 100%; the resistance gene is tetQ, coding a ribosomal protection protein. The tetQ genes are found on conjugative transposons[70]. Interestingly, Bacteroides carry another tetracycline resistance gene tetX (or its amino-terminally truncated, 60% homologous variant tetX1)[71], which is capable of oxidizing the tetracycline molecule[72], but since it requires oxygen for this process, its role in the tetracycline resistance of Bacteroides is very limited. Bacteroides is resistant at a low level (1.7%) to tigecycline, a synthetic minocycline, glycylcycline derivative. In such tigecycline-resistant cases, a direct role of the tetX and tetX1 genes has not been confirmed for the Bacteroides[73].

The Bacteroides are now becoming resistant to the once fully effective fluoroquinolones such as trovafloxacin and moxifloxacin, reaching resistance rates of > 40% and 13.6% in the United States and Europe, respectively. Additional data on the antibiotic resistance rates of Bacteroides are to be found in a recent exhaustive review[74].

INSERTION SEQUENCE ELEMENTS

There are a huge variety of transposable and conjugally mobile genetic elements, in particular among prokaryotes. IS elements are short (from 600 to 2000 bp long), double-stranded integrative DNA sequences that code for only a transposase gene, bordered by inverted repeat sequences; during their integration, they usually cause target site duplications of a small number of nucleotides. A general scheme relating to their organization is presented in Figure 1. They are to be found in all three domains of life (Archaea, Bacteria and Eukarya). Their classification is based on the ends of their inverted repeats and the conserved amino acid residues of the transposase genes[75]. In this way, about 25 families are distinguished among prokaryotes and are usually named after their earliest and best-examined members. Some families fulfil the above-mentioned description criteria, but molecularly represent a more divergent type of elements, e.g., the application of different transposition mechanisms to the main groups of IS elements which harbor transposases with an indispensable aspartate-aspartate-glutamate (DDE) motif forming the active catalytic center. Similar motifs can be found in the integrase proteins of retroviruses and, among others, in RNase H ribonuclease, in the DNA polymerase I 3’-5’ proofreading activity domain and in the RuvC recombinase proteins of bacteria forming the RNase H enzyme superfamily. The reader can find further data on the classification and transposition mechanisms in some older and more recent reviews[75,76] and in the IS Finder database (www-is.biotoul.fr)[77].

Figure 1 Schematic structure of an insertion sequence element. DR: Direct repeats; IR: Inverted repeats; Po: Outward-oriented promoter; Ptnp: Promoter of the transposase. The transposase gene is denoted as tnp. The borders of the insertion sequence (IS) are indicated by the closed arrowheads below.

The simple genetic organization in sensu stricto is sufficient for IS element dispersal, and they can therefore be regarded as appropriate examples of selfish genetic elements. However, thorough examinations of their prevalence, genetic structure and transposition suggest that they not only parasitize their hosts, but sometimes participate in firm interaction with them. Such interaction with the host may be accomplished via (1) a promoter supply for the host genes; (2) increased evolution rates; and (3) a metabolic load. The activation of the expression of nearby genes by promoter supply is mediated by outward-oriented promoters and is specific for a small subset of ISs. This way, various bacterial genes can be activated, resulting in antibiotic resistance most notably, and the ISs act among others on various antibiotic resistance genes, e.g., bla_TEM(pBR322) of Escherichia coli[78], bla_OXA-51Acinetobacter baumannii[79] and oprD of Pseudomonas aeruginosa[80]. They can influence the evolution potential of their host by their mutagenization of the host genomes by hopping activity[81]. It is also known that the introduction of a copy of an additional accessory genetic element, e.g., a plasmid, and the amplification of their physiological copy number means a fitness decrease first and then an adaptation[82], which is also valid for IS elements. IS elements participate in activation of Bacteroides antibiotic resistance genes.

The discovery that erythromycin and clindamycin resistance is due to an MLS_B resistance mechanism (capable of causing cross-resistance to the chemically different macrolid, lincosamide and streptogramin B antibiotics), mediated by ermF genes, and the subsequent linking of these genes to similar compound transposons, was the first indication of IS element involvement in antimicrobial resistance among Bacteroides. Clindamycin resistance plasmids were first detected in clindamycin-resistant isolates[83]. Such plasmids as pBF4 (pIP410), pBFTM10 (pCP1) and pBI136 were characterized very well molecularly in the 1980s[84]. pBF4 (41 kb) harbored Tn4351 bordered by inverted copies of IS4351 and in between ermF and an aerobic-type tetracycline resistance gene, tetX[85]. Tn4551 accounts for a large portion (about 30%) of pBFTM10 (15 kb) and contains ermF in direct repeats of IS4351[84]. pIB136 (80 kb) also contains Tn4551, but with a high preponderance to lose this structure[86].

The 1990s revealed other important links between IS elements and the antibiotic resistance of Bacteroides. After cloning of the determinant for the carbapenem resistance, cfiA[46,47], PCR detection and parallel molecular methods have demonstrated that carbapenem-resistant mutants can arise in single-step mutations from cfiA-positive but carbapenem-susceptible B. fragilis isolates[87], which proved to be insertions of IS1186[88] and IS942[47]. Later studies confirmed these findings and the roles of a series of other IS elements were identified in carbapenem-resistant strains from such different geographic regions and countries as Europe (France, the United Kingdom, Hungary, Sweden, Switzerland, Norway and Italy), the United States, Japan, Korea and Kuwait[48,89-95]. The 5-nitroimidazole resistance genes, nims, also carry various IS elements in their upstream regions. In these cases, the nim gene types, the carrying genetic elements and the activating ISs were linked (Table 2). The presence of IS elements has been demonstrated in the upstream regions of other β-lactamase genes, cepA and cfxA. In a high cephalosporinase-producer strain, B. fragilis CS30, this feature was caused by a specific DNA sequence that contained an IS21-like region (ISBf1) at its 3’ end adjacent to the cepA gene[96]. In the case of a representative strain (B. vulgatus CLA341) for cfxA-mediated cephamycin resistance, the upstream region of the cfxA gene also contained an IS element (ISBf8) that was identified later by bioinformatics analysis[97,98]. The majority of the cfxA genes of Bacteroides do not normally contain this (ISBf8) and another mobile element (MITEBf2) in their upstream region[45]. Interestingly, again in high β-lactamase-producer cfxA-positive strains, the presence of IS614-like elements has been revealed in the upstream region of the resistance gene by inverse PCR[44].

Table 2 The 5-nitroimidazole resistance nim genes of interest for Bacteroides.

Nim gene type	Carrying genetic element	Activating IS	No. of isolates1
nimA
	pIP417 (7.7 kb)	IS1168	10[102,114,115]
	10 kb uncharacterized plasmid	IS1168	2[102]
	8.2 kb uncharacterized plasmid	IS614	1[102]
	Chromosomal	IS1168 or Unknown	3[102]
	Unknown	IS1168	12[115]
	Unknown	IS1169	1[116]
nimB
	Chromosomal	IS1168 or IS612 or IS614	8[102,114]
	Unknown	IS1168	3[116]
nimC
	pIP419 (10 kb)	IS1170	4[115]
	Chromosomal	IS1170	2[102]
	Unknown	IS1170	2[63,116]
nimD
	pIP421 (7.3 kb)	IS1169	1[102,117]
	Chromosomal	Unknown	1[102]
	Unknown	IS1169	6[116]
nimE
	pBF388c (pWAL610, 8.3 kb)	ISBf6	5[102,118]
nimF
	Chromosomal	Unknown	1[116]
nimG
	Unknown	Unknown	1[63]
nimH
	Unknown	Unknown	-2

¹The number if isolates with the given genotypes are indicated with references;

²Taken from GenBank (http://www.ncbi.nl.nih.gov, acc. no. FJ969397).

For cfiA and cfxA, a heterogeneous resistance phenotype has been detected by diffusion methods (especially the Etest) in strains that have elevated agar dilution MICs and do not have IS elements in the upstream region of the resistance genes[45,99].

Though specific for some representative strains and resistance genes, the Bacteroides IS elements have been shown to be capable of activating all IS-requiring resistance genes. In this way, IS4351 can activate cfiA[48], the IS elements of the nim and cfiA genes are interchangeable, and an IS element discovered for cfiA, IS614 (or its variant), has been found to activate the cfxA gene also[44]. However, little is known about the prevalence and epidemiology of the resistance gene-activating ISs apart from their being found in resistant isolates. The best-studied examples are the nim and cfiA-carrying strains, but these differ considerably with respect to the prevalence of ‘silent’ and activated genes. Thus, all well-characterized nim genes are associated with an IS element (Table 2), but the majority of the cfiA genes are ‘silent’, and not associated with ISs[52,87,89,100,101]. Examinations of the insertion sites of ISs among nim and cfiA genes revealed that for nims the insertion sites are well defined and conservative for a particular nim gene type[102], whereas for the cfiA genes they vary[91,92]. This means a well-known mechanism for the emergence of nitroimidazole resistance by the nim and IS combination, which was investigated and discussed recently, especially for the β-lactam resistance mechanisms of Enterobacteriaceae[103,104], involves the consecutive steps of emergence, adaptation and spreading. For the Bacteroides the nim gene IS combinations first emerged, which were then inserted into specific replicons (plasmids and chromosomes) and subsequently were spread in the Bacteroides population.

Another epidemiological concern besides the interchangeability of the IS elements is their geographical distribution (Figure 2). A number of studies of IS elements in resistant strains indicated that there is little geographical restriction to their spreading worldwide, e.g., IS614 or IS614-like elements were found ubiquitously, though some local tendencies can also be observed (Figure 2, cf. Japan and Korea). It could also be that these IS elements vary in the nucleotide sequence, giving rise to isoforms (not mentioned in full detail here) and could be mosaics/combinations of other elements. This can be explained by the homologous nature of these elements and the fact that they can be harbored coincidentally in an unknown proportion of the strains.

Figure 2 Insertion sequence elements found worldwide in antibiotic-resistant Bacteroides isolates. For Europe, the following abbreviations are used to identify the countries in which the insertion sequences were isolated: At: Austria; Be: Belgium; Ch: Switzerland; Cz: Czech Republic; De: Germany; Dk: Denmark; Fi: Finland; Fr: France; Hu: Hungary; It: Italy; Nw: Norway; Sp: Spain; Sw: Sweden; UK: United Kingdom.

While the role of IS elements in emerging antibiotic-resistant Bacteroides strains is well documented, the process of the movement/skipping of the IS elements from their proper positions has been investigated only poorly. Podglajen et al[87] studied this process in vitro and reported a rough estimation of the development of imipenem-resistant strains, with 10^-8 to 10^-7/cell frequencies in a given culture. Edwards et al[89] detected this process in vivo when the initially susceptible strain in a patient with a B. fragilis infection treated with imipenem became resistant[105].

Overall, the IS elements found among Bacteroides species belong in 9 IS families, members of 5 families being capable of activating antibiotic resistance genes. An overview of these elements is provided in Table 3.

Table 3 The insertion sequence elements involved in the up-regulation of antibiotic resistance genes in Bacteroides.

IS family1	Group1	IS2	Activated genes
IS4	ISPepr1
		IS943	cfiA
		ISBf8	cfxA
IS5	IS5
		IS1186 (IS1168)	cfxA, cfiA, nimA, nimB
		IS1169	cfiA, nimA, nimD
	IS1031
		ISBf6	nimE
IS21	-
		ISBf1	cepA
IS982	-
		IS1187	cfiA
IS31380	IS942
		IS942	cfiA
		IS1170	nimC
		IS612	cfiA, nimB
		IS613	cfiA
		IS614	cfxA, cfiA, nimB
		IS615	cfiA
	-
		IS1188	cfiA
		IS4351	ermF, cfiA
		IS616	cfiA

¹The IS families and the subgroups within them (taken from IS Finder)[77]; - indicates no further classification;

²The species of IS elements activating the resistance genes of Bacteroides spp.; the mosaics and isoforms are not indicated. IS: Insertion sequence.

THE PROMOTERS CARRIED BY THE BACTEROIDES IS ELEMENTS

Although IS element insertion correlates well with antibiotic resistance gene expression, the main reason for their up-regulation is that the IS elements carry outward-oriented promoters capable of driving the expression of the genes. The initial hypothesis for this up-regulation was the lending of IS activation mechanisms from other antibiotic resistance genes for aerobic species, but only E. coli promoter sequences could be investigated for these Bacteroides IS elements at that time. There was a straightforward result concerning the requirements for transcription in Bacteroides when Bayley et al[106] recognized the nucleotide composition of the promoter sequences for several antibiotic resistance and other Bacteroides genes. The promoter consensus sequence for Bacteroides is depicted in Figure 3A. This highly different requirement in promoter sequence was later explained by the results of Vingadassalom et al[107], who proved that the primary σ subunit of the Bacteroides fragilis RNA polymerase is unusual and different from that of other bacteria; whereas it is able to start transcription from original Bacteroides promoters in reconstitution experiments, it clusters firmly together with the suspected primary σ-factors of other Bacteroidetes species, but only distantly to the primary and stationary σ-factors of other bacteria.

Figure 3 The nucleotide sequences of the promoters of some important Bacteroides antibiotic resistance genes. The consensus sequence with the conserved regions (A), and the actual promoter sequences (B). The match with the consensus is shown in bold capital letters, proven transcriptional initiation sites are marked in bold with an arrowhead next to them; n denotes any nucleotide, and small letters in the consensus indicate less conserved bases. The own promoters of cepA and cfxA were searched for bioinformatically and are not IS elements next to them in the list. The own promoter sequence of cfiA is from our unpublished preliminary experiments made by ‘rapid amplification of cDNA ends’ capable of amplifying in PCR the 5’ end of the mRNA. Underlined -33 regions are parts of compound promoters and these parts originate only from insertion sequence elements in the cases of these promoters.

Several antibiotic resistance gene promoter sequences were recognized during these investigations. The first was for the cepA gene of B. fragilis CS30 in the original work of Bayley et al[106] Similarly, those authors determined the promoter for the cfxA gene on MTn4555 of B. vulgatus CLA341[96]. Interestingly, this promoter is a compound one: the -7 region originates from a prototype MTn4555 backbone and the -33 region from an IS element (ISBf8). MTn4555 insertion of the IS614 elements is associated with increased resistance to cefoxitin, though the exact transcription initiation site for this IS element and the promoter remain to be elucidated[44]. Among these rare data relating to the promoter structures of Bacteroides, those carried by IS612, IS613, IS614, IS615 and IS616 elements activating the cfiA genes have been recognized, thereby furnishing us with important confirmatory data[90,93]. Podglajen et al[108] also determined the outward-oriented promoters of some important IS elements (IS1186, IS942, IS1187 and IS1188) participating in activation of the cfiA genes of some carbapenem-resistant B. fragilis isolates. Although the recognition of the requirements for the Bacteroides promoter nucleotide sequence facilitated an understanding of their antibiotic resistance mechanisms, there was also research into other aspects of their properties, e.g., the CPS on-off regulation[21]. Figure 3B lists the known and some proposed sequences of promoters of antibiotic resistance genes of Bacteroides.

Other resistance mechanisms

Despite the dominance of IS element-borne activation of the antibiotic resistance genes of Bacteroides, natural resistance (to aminoglycosides, 1^st and 2^nd-generation fluoroquinolones and aztreonam), resistance emerging by point mutations, and the enforcement of internal regulatory mechanisms of the genes should be mentioned. Point mutations in the gyrA gene (coding for a subunit of topoisomerase II) can cause ciprofloxacin, moxifloxacin and trovafloxacin resistance[109]. A special, well-characterized resistance mechanism of Bacteroides is coded by tetracycline resistance conjugative transposons harboring the tetQ genes. The tetQ genes have their own promoters that can be up-regulated by tetracycline, as observed in the 1970s and exhaustively analyzed since the 1990s[70]. This is mediated by an attenuation mechanism where the transcription stalls at a leader upstream of tetQ in the absence of tetracycline, but in the presence of tetracycline the transcription proceeds. The tetQ gene is in an operon with the regulatory proteins of rteABC, which upon tetracycline exposure up-regulate the excision, mobilization and conjugation genes[110]. For this regulation to be effective, other regulatory processes are also involved, whose absence makes the conjugative transposons constitutive with respect to tetracycline[111]. Some other important resistance genes code efflux pumps, e.g., bexA (fluoroquinolones)[112], mefA and msrSA (clindamycin)[55], and an endogeneous efflux mechanism, mediated by the bmeABC genes, can be up-regulated by mutations in the amino acid sequence of the coded effector proteins[113].

CONCLUDING REMARKS

Bacteroides species are noteworthy participants and contributors to human health and disease. They comprise a group of bacteria with additional molecular biological specific features as regards their promoter and RNA polymerase structures and a huge number of surface variations due to the invertible promoters at their CPS operons. The regulation of their antibiotic resistance genes is in most cases also specific; they need up-regulatory IS elements for antibiotic resistances to develop. However, there is a paucity of data about the observed associations in resistant strains: the promoters of less characterized IS elements are still to be determined, the roles of up-regulatory IS elements in other resistance genes could be investigated, and the frequencies with which the IS elements move to the upstream positions of the resistance genes could be examined in greater detail. These latter approaches would promote a better understanding of the whole picture of the rather prevalent antibiotic resistances of the Bacteroides species, which in turn would facilitate the design of better antimicrobial therapies against this important group of bacteria in the future.