The Flexible Genome of Acidophilic

Over the last couple of decades there has been considerable progress in the identification and understanding of the mobile genetic elements that are exchanged between microbes in extremely acidic environments, and of the genes piggybacking on them. Numerous plasmid families, unique viruses of bizarre morphologies and lyfe cycles, as well as plasmid-virus chimeras, have been isolated from acidophiles and characterized to varying degrees. Growing evidence provided by omic-studies have shown that the mobile elements repertoire is not restricted to plasmids and viruses, but that a plethora of integrative elements ranging from miniature inverted repeat transposable elements to large integrative conjugative elements populate the genomes of acidophilic bacteria and archaea. This article reviews the diversity of elements that have been found to constitute the flexible genome of acidophiles. Special emphasis is put on the knowledge generated for Sulfolobus (archaea) and species of the bacterial genera Acidithiobacillus and Leptospirillum . Also, recent knowledge on the strategies used by acidophiles to contain deletereous exchanges while allowing innovation, and the emerging details of the molecular biology of these systems, are discussed. Major lacunae in our understanding of the mobilome of acidophilic prokaryotes and topics for further investigations are identified.


Introduction
Bacteria and archaea have evolved efficient genome modification mechanisms that allow them to adapt rapidly and effectively to ever-changing environmental conditions and to colonize a plethora of ecological niches. Some of these mechanisms are universal (e.g. transposition), while others are significantly more frequent in the prokaryotic world (e.g. transduction, conjugation). Prokaryotic genomes are modified by intrareplicon, interreplicon and interorganismal gene movement of pre-evolved traits, occasionally circumventing the species barrier (Treangen and Rocha, 2011). These mechanisms are collectively known as horizontal gene transfer (HGT).
Scientific literature on the adaptive role of HGT in prokaryotes is extremely rich (Soucy et al., 2015); for instance, the acquisition of resistance to harmful compounds and elements or the acquisition of metabolic capabilities allowing the use of new resources are frequent events in all well-studied ecosystems. Extensive HGT between lineages that occupy similar ecological niches, which may drive radical changes of lifestyle, have also been documented. HGT seems to have contributed to the adaptation of Thermoplasmatales archaea to extremely acidic and hot environments. Thermoplasma acidophilum and Picrophilus torridus share approximately the same fraction of genes (65% of their proteomes) between them, as they do with Sulfolobus solfataricus, a very distant relative that often inhabits the same environment, whereas only 35% of their genes are present in Pyrococcus furiosus, a closer relative (Fütterer et al., 2004). Beyond its role in prokaryotic genome innovation and evolution (Koonin, 2016), HGT also influences dramatically the function of complex prokaryotic communities (Rodriguez-Valera et al., 2016).
Gene flow between and within microbes is driven by a vast repertoire of Mobile Genetic Elements (MGEs). MGEs are mostly DNA-based platforms that encode enzymes and other proteins that mediate or facilitate self movement and that support different genetic cargos. The full set of MGEs of a microorganism and the genes they carry constitute the mobilome (Frost et al., 2005) or the flexible genomic compartment (Mira et al., 2014). Research in microbial MGEs has progressed rapidly over the last couple of decades. Some MGEs are mainly intrachromosomal (e.g. transposons) and others extrachromosomal or episomal (e.g. plasmids), and yet others exist in a fluctuating equilibrium between an integrated and an episomal state (e.g. temperate viruses). MGEs are classified depending on the molecular mechanism underlying their physical movement. Translocative elements, including transposons and insertion sequences, depend on transposases which catalyse `cut and paste´ or `copy paste´ movements of DNA. Integrative elements relying on site-specific recombination of DNA mediated by a diverse family of enzymes called integrases (or recombinases), include integrons, temperate viruses and a wide diversity of chimeric elements. Other replicative and non-replicative elements which can move between microorganisms, such as plasmids and genomic islands, are reckoned as dispersive or transmissible elements and are further classified as conjugative (those that code all components needed for transfer) or mobilisable according to their transfer ability. Thus, the mobilome is an assortment of selfish MGEs (Figure 1) composed of many different types of elements, and accretion of individual MGEs (e.g. transposon) on higher-order mobile elements (e.g. a conjugative plasmid). Several chimeric elements such as transposable proviruses, integrative plasmids and mobile integrons muddle the picture considerably and favour the view of MGEs as mosaics of functional gene modules of diverse evolutionary origins (Flores-Ríos et al., 2019a).
Lack of informative classification systems and adequate tools for MGE prediction in sequenced genomes has hampered analysis of the prokaryotic mobilome. Both impediments have began to be addressed in recent years. Improved classification criteria have began to emerge for certain types of MGEs. Such is the case of proteobacterial plamids, for which a relatively comprehensive picture of the diversity, types and mechanisms have started to emerge (Garcillán- Barcia et al., 2013;Fernandez-Lopez et al., 2017). Also, several MGE prediciton strategies (e.g. Gugliemini et al., 2011), tools (e.g. Guliaev and Semyenova, 2019) and pipelines (summarized in Languille et al., 2010) have been developed and made publically available in recent years. Although comprehensive genomics of the prokaryotic 'mobilome' still lags behind most other genomics initiatives, sequencing efforts and comparative genomics and metagenomics studies have revealed that the different types of MGEs are found in all branches of the prokaryotic tree of life, and that they occur at varying frequences, even in the most extreme environments. From these studies new relevant definitions have emerged, like: (i) core genome, which defines the genes present in all strains of a prokaryotic species; (ii) flexible (or dispensable) genome, which concerns genes that are present in some, but not all, strains of a species; (iii) pan genome, which entails the sum of the former two, and (iv) supergenome (or communal gene pool), which consists of all the genes encoded on MGEs and that are readily available to all permissive prokaryotes within a particular setting (Medini et al., 2005;Norman et al., 2009). This has signified a paradigm shift in that the individual gene pool is now thought to represent a vast resource for MGEs to exploit, enabling the sharing of beneficial traits among prokaryotic populations restricted only by the host ranges of the resident MGEs and the selection pressures exerted by the surrounding environment. Stable appropriation of foreign genetic material by different hosts (or host ranges) is determined by a number of factors and is less likely as the phylogenetic distances increase. In the case of plasmids, the host-range is influenced by factors affecting their capacity of transference and replication within their new hosts (Jain et al., 2013). In the case of viruses, it is mainly determined by specific host attachment sites and cellular factors necessary for viral multiplication (Hyman and Abedon, 2010). In addition, a number of molecular mechanisms encoded by prokaryotes that counteract uptake and stabilization of foreign DNA molecules serve to further limit MGE host ranges : these include CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats)-Cas and RM (Restriction-Modification) Systems.
Given that the mobilome and its genetic cargo seem, in many cases, tightly linked to the environment of the host population, it is thought to vary considerably from ! 234 one environment to another. In this review, current knowledge gathered on MGEs that favour HGT in extreme acidic econiches and the strategies used by acidophiles to contain their dispersal, are explored. Knowledge on the mobilome of acidophiles has been organized on the basis of the type of element and covers both acidophilic bacteria and archaea. Long standing lacunae in our understanding of the of the mobilome of acidophilic prokaryots are highlighted.

Transmissible elements
Plasmids are self-replicating, extrachromosomal, transmissible replicons that act as key agents of change in microbial populations. They range from a few kilobase (Kb) to few hundred Kb replicons, termed megaplasmids. A plethora of plasmids from all three domains of life and almost every conceivable environmental niche have been described, yet only 70 plasmids from nearly 160 acidophiles are known by means of their sequence or description (Table 1). Although the genome sequences of an increasing number of plasmids from microbial acidophiles have been determined, knowledge on their biology, biochemistry and molecular mechanisms lags considerably far behind. Insights into the replication, mobilisation/conjugation and regulation of some of these plasmids has been gained, yet in most cases they remain cryptic from a functional perspective, and their impact on genome plasticity remains largely unexplored. Knowledge generated for well-characterised plasmid families is covered herein, including archaeal conjugative plasmids, plasmid-virus hybrids and bacterial mobilisable and cryptic plasmids. Replicons of acidophiles, which based on size criteria fall within the category of megaplasmids (larger than 100 Kb) have been omitted from this revision, since no thorough study of these has been reported to date.

Small cryptic plasmids
Some acidophilic archaea and bacteria possess small plasmids of less than 10 Kb, considered to be cryptic in nature. Examples include the high copy number crenarchaeal pRN and "pRN-like" family plasmids, the medium copy number Acidiphilum plasmids pAM/pACRY/pACM and the broad host range, low copy number Acidithiobacillus pTF/pAca plasmids, among others recently added from sequencing projects (Table 1).
The research on cryptic plasmids of acidophiles has focussed mostly on the pRN plasmid family from Sulfolobus spp. and Acidianus spp. Members of this family seem to be exclusive to the crenarchaeal phylum and appear unrelated to any other known eubacterial or archaeal plasmids. Several of these plasmids have been used for the construction of E. coli -Sulfolobus shuttle vectors andhostvector systems (Berkner and Lipps, 2008;Hwang et al. 2015;Suzuki and Kurosawa, 2016). The pRN1 plasmid was the first crenarchaeal plasmid isolated and sequenced, and represents the prototype of this rather remarkable family (Lipps, 2009). All family members share a conserved genomic region and three highly conserved gene products which have been characterised biochemically (Peng et al., 2000). Recent studies have revealed that a single multifunctional replication protein (ORF904) with helicase, primase and DNA polymerase activity, reminiscent of the type of replication performed by some linear bacteriophages, is required to initiate plasmid replication (Lipps, 2011;Berkner et al., 2014). In ! 235 ! 236 Table   1. marked contrast to all DNA polymerases, ORF904 is able to polymerise DNA de novo, i.e. without a primer. The enzyme is not a processive DNA polymerase, it does not have proofreading activity and its specific activity is rather low. Gene orthologs of orf56 encode a fairly variable sequence-specific DNA-binding protein, designated CopG, that has a role in plasmid copy control through down-regulation of plasmid replication initiation. The orf80 orthologues are partly annotated as plrA (plasmid regulatory gene A) but their function is currently unclear.
The pRN family also includes pSSVx from Sulfolobus islandicus REY 15/4, which is a hybrid between a pRN plasmid and a fusellovirus. The genome of pSSVx contains the three highly conserved ORFs of the plasmid family and two conserved ORFs originated from SSVs fuselloviruses, probably involved in packaging (Lipps, 2006). It is believed that pSSVx replicates in the host cell as a plasmid and that upon superinfection with the helper SSV viruses it spreads as a virus satellite (Stedman et al., 2003). pSSVi resembles members of the pRN plasmid family in genome organization but encodes an SSV-type integrase. Indeed, a few other pRN-type elements (pXQ1, pST1 and pST3) have been discovered in an integrated form in the chromosomes of Sulfolobus species , all of which encode an SSV-type integrase, likely responsible for plasmid integration into the crenarchaeal genome (e.g. She et al., 2002).
Additional small high-copy-number cryptic plasmids isolated from other Sulfolobus species and strains, grouped in the "pRN-like" family, have been reported and some of them characterised. The plasmids pIT3, pTAU4, pORA1 and pTIK4 differ from the pRN family plasmids in that they encode a much less conserved or a completely different replication protein (Zillig et al., 1998;Zillig et al., 1996). This provides evidence that the replication apparatus of Sulfolobus plasmids can be quite diverse. Unlike the situation in the crenarchaea, where a large number of genetic elements have been characterised, little is known of extrachromosomal elements in the Thermoplasmatales. To date, the only sequenced and characterized plasmids from this phylogenetic order are the low copy number pTA1 isolated from Tp. acidophilum and pPO1 isolated from Picrophilus oshimae. Small cryptic plasmids are missing in closely related Thermoplasma volcanicum and P. torridus.
The search for native plasmids in the acidithiobacilli has yielded a large number of cryptic plasmids ranging in size from 4 to nearly 10 Kb (Rawlings, 2001;Rawlings and Kusano, 1994). Comparative studies of these plasmids using restriction endonuclease digestion and southern hybridisation have profiled plasmid occurence in diverse strains from different geographical origins but have yielded little insight into the nature of the sequence similarity of this group of small independent replicons.
In 1995 Chakravarty and colleagues presented evidence for a common plasmid replicon present in several strains of At. ferrooxidans (Chakravarty et al., 1995). They found that four strains (TFI91, TFI92, TFI85 and TFI29) contained a related 9.8 Kb plasmid sharing similar restriction endonuclease sites to the prototype plasmid pTFI91. Similarity of pTFI91-family plasmids was also extended to ! 239 plasmid pTfA-4 and related plasmids present in Chilean isolates of At. ferrooxidans obtained from copper mines several hundred kilometres apart, as well as to the larger pTF5 plasmid isolated from the type strain of Acidithiobacillus ferridurans (ATCC 33020) (Dominy et al., 1997). This indicated that these plasmids have a very wide geographic distribution and are either positively selected for or difficult to cure. Later findings confirmed that the pTFl9I-family replicon is a unique entity not related at the nucleotide level to other known broad host range plasmids. Acidithiobacillus caldus sequenced strains ATCC 51756 and SM-1 also contained 9.8 Kbp small cryptic plasmids. Plasmids pAcaTY.2 and pLAtc1 have been characterised bioinformatically and shown to be different from all currently sequenced acidithiobacilli plasmids and also from each other (Acuña et al., 2013;You et al., 2011a). Plasmid pLAtc1 encodes a mobilisation system similar to that described for pRSB105, but is otherwise very different from it. In turn, plasmid pAcaTY.2 from the type strain encodes a disrupted ortholog of the replication protein RepA of other acidithiobacilli and a gene whose product is similar to the plasmid exclusion protein Exc1, known to prevent the entry of IncItype plasmids (Garcillán-Barcia and de la Cruz, 2008). Restriction profile analysis have recently revelated that plasmid pAcaTY.2 corresponds to plasmid pTK1 reported in the original description of the type strain (Shelly Deane, Stellenbosch University,personal communication). Representatives of this set of broad host range plasmids or parts of them (e.g. the replication origin) have been recently used for the development of expression vectors (Zhang et al., 2014) and shuttle vectors between E. coli and the Acidithiobacillus spp. (US Patent 8.163.558 B2; 2012).

Conjugative plasmids
The thermo-acidophilic crenarchaeote Sulfolobus hosts a number of medium sized, moderate to high copy number, conjugative plasmids (Table 1). These appear to be present in about 3% of all isolated Sulfolobus strains . Sulfolobus conjugative plasmids have been grouped on the basis of their genomic characteristics as pKEF-like plasmids and pARN-like plasmids. Full genomic sequence comparison and phylogenetic analysis of representative genes has revealed that pKEF9, pING1, pHVE14, pSOG2 and pNOB8 are very similar to each other in gene content and organization, whereas pARN3, pARN4, pSOG1 and pTC form a related, but separate, group (Wang et al., 2015;Greve et al., 2004). Natural deletion variants fitting this classification have also been described (e.g. pING2 del. var, of pING4; pARN4 del. Var. of pARN3). They all encode a cluster of five to six core proteins that have been implicated in the conjugative process and, with few exceptions, also a non-partitioning type integrase (She et al., 2004). This integrase is known to be responsible for reversible chromosomal integration of pNOB8 plasmid in S. tokodaii, S. acidocaldarius and Acidianus hospitalis chromosomes. The presence of a defective conjugative plasmid-like element in the chromosome of several Sulfolobus species and Metalosphaera sedula DSM 5348, and the discovery of pAH1 in Ad. hospitalis W1, extends the host range of this group of conjugative plasmids to genera other than Sulfolobus. Being also integrative elements, these plasmids have been implicated in facilitating chromosomal DNA conjugation for some Sulfolobus species (e.g. Reilly and Grogan, 2001).

! 240
The best-characterized plasmids are the pKEF family plasmids. They range in size between 20 and 40 Kb and are all medium copy number plasmids in S. solfataricus test strain PH1 (20 to 40 copies/chromosome); although slightly lower copy numbers occur in their natural hosts . Conserved genes within this family are organized in three functionally distinct genomic sections grouping the conjugation genes, the origin of replication and the proposed replication genes which resemble pRN family plasmids repA, copG and plrA genes (Greve et al., 2004). The genomic regions between these sections are more variable in size, gene content and sequence, and harbour putative recombination motifs which contribute to plasmid variability. Indeed, mixtures of variant plasmids have been detected upon propagation of these plasmids in laboratory S. solfataricus strains. Even so, pKEF plasmids are stably maintained in the original plasmid-carrying strains, which is consistent with them carring orthologs of bacterial proteins involved in partitioning (ParA, ParB).
Despite the fact that conjugational exchange of plasmids DNA has been observed between Sulfolobus spp. , it is still unclear what the exact mechanism is, what proteins are involved, and whether there is an origin of transfer. However, one of the well-conserved regions of pKEF plasmids has provided hints in this respect. This region contains six ORFs predicted to encode the core conjugative apparatus in Sulfolobus, and two of these genes products (pKEF9: p12 and p01) exhibit sequence motifs and domain structures characteristic of the bacterial Type IV secretion system coupling protein (VirD4/ T4CP) and relaxase (VirB4) proteins required for DNA transfer. The Sulfolobus T4CPs are considerably larger than bacterial VirD4, while the VirB4-like proteins are slightly smaller than their bacterial counterparts, and both proteins have Ctermini and/or N-termini that are unrelated to other family members (Lipps, 2006). The other four ORFs encode predicted membrane proteins with up to 10 transmembrane helix motifs, and are believed to be involved in transmembrane pore formation required for conjugational exchange of plasmid DNA, mimicking the single membrane spanning protein that facilitates self-transfer in some small plasmids of Gram-positive bacteria. Although VirB2 orthologs are absent from pKEF family plasmids, genes encoding small proteins with pilin-like hydropathy profiles (pKEF9: p05) are colinear with genes encoding VirB4-like subunits and might assemble to form the pili detected on Sulfolobus pKEF9 donor cells (Schleper et al., 1995).
To date, limited progress has been made in characterising the mechanisms of replication and copy number control of the conjugative plasmids of archaeal acidophiles, and very little is known about the interplay between the different elements within a host. Superconjugation immunity studies, in which a plasmid containing host is challenged with a second plasmid after sucessive conjugative transfers, have been performed for some of these plasmids . These experiments have revealed incompatibility of closely related plasmids and different quantitative responses on the copy number of certain conjugative plasmids. While the incompatibility of bacterial plasmids is mainly a consequence of competition between two plasmids for the same replication or ! 241 partition apparatus, the mechanisms behind incompatbilities between archaeal plasmids still needs to be determined.
In the case of acidophilic bacteria, only a few of the known plasmids encode the T4SS required for conjugative self-transfer. These include the MOBP-type 59 Kb pY0007 from Sulfobacillus thermotolerans (Deane and Rawlings, 2011) and its related episomal pL15 in Sb. thermotolerans L15 and integratedplasmid-like region in Sb. thermotolerans Kr1 (Panyushkina et al., 2019). Also, in this category are two MOBF-type large plasmids, pACVM2 and pACRY03, from Acidiphilium multivorum and Acidiphilium cryptum. A deletion variant of pACRY03 lacking the coupling protein and the mating bridge genes is also found in A. multivorum under the pACMV3 designation. The Sb. thermotolerans large, high G+C content plasmids pY0017, pL15 and the plasmid-like region in Sb. Thermotolerans Kr1 encode a potential non-pheromone conjugation 14-gene operon containing homologs of both type IV and II secretion systems (T4SS and T2SS), typical of Gram-positive bacteria. Although relatively little is known about this conjugation system, it has been suggested that it might be involved in establishing physical contact between donor and recipient cells in order to facilitate conjugal transfer of the plasmids (Deane and Rawlings, 2011). None of the Acidiphilium conjugative plasmids have been analyzed to any extent to date and conjugative transfer capacity still awaits experimental evaluation for most acidophilic bacterial conjugative plasmids.

Mobilisable plasmids
A considerably larger number of mobilisable plasmids have been identified in acidophilic bacteria, mostly in the Proteobacteria. Most of the reported or predicted mobilisable plasmids are fairly large elements (ranging from ~ 5 to 90 Kb), encoding at least a relaxase and/or a coupling protein. Species-specific plasmid backbones accompanied by different accessory genes, appears to be a recurring theme in acidophiles such as At. ferrooxidans (e.g. Dominy et al., 1998) and At. caldus (e.g. van Zyl et al., 2008a).
A family of mobilisable plasmids (pTcM1 family) ranging in size from approximately 30 to 70 kb has been found in a variety of At. caldus strains from different continents (Table 1). A number of these plasmids have been characterised to different degrees, including pTcM1, pTcF1 and pCSH12, pLAtc3 and pAcaTY.1. These plasmids share a common region that spans 11 to 26 Kb and includes the plasmid replicon (RepA-type) and other functions associated with plasmid stability and maintenance (ParA-ParB), as well as an IncQ-like mobilisation relaxase of the MobA/RepB-type and a PasAB plasmid addiction system (van Zyl et al., 2008a(van Zyl et al., , 2003. No oriT site was identified within the common 26-kb region, yet pTcM1 is mobilisable. The plasmid backbone appears to be conserved with respect to replicons from microorganisms thriving in different environments. Also, a gene encoding a predicted invertase-like protein with motifs in common with the PinR family of site-specific recombinases is present in pTcM1-family plasmids and other plasmids isolated from acidophilic bacteria (pTCF14 and pTF5), the role of which still remains elusive.

! 242
Two closely related broad host-range, mobilisable, medium copy number plasmids (10-15 plasmids per chromosome), were discovered in the acidithiobacilli (Rawlings, 2005). Plasmids pTF-FC2 (IncQ-2a) and pTC-F14 (IncQ-2b) share with other IncQ-family replicons the presence of repB, repA and repC genes encoding the primase, helicase and iteron-binding protein, respectively (Dorrington and Rawlings, 1990;Rohrer and Rawlings, 1992), as well as a 22-bp iteron based oriV (Rawlings and Tietze, 2001). Despite their similarity, their origins of replication are compatible (Gardner et al., 2004). Both plasmids lack an active partitioning system and their stability depends on the toxin-antitoxin system PasAB, which also seem to play a role in copy number control (Matcher and Rawlings, 2009;Deane and Rawlings, 2004;Gardner et al., 2001), resembling CopG-mediated control of crenarchaeal pRN1-family plasmids. Both plasmids have a 5-gene mobilisation module encoding a MobA relaxase of the MOBP-type (Garcillán- Barcia et al., 2009). While MobA and MobC appear to be essential for plasmid mobilization, the non-essential MobD and MobE only seem to affect the frequency of mobilisation (van Zyl et al., 2003). Even though no conjugative partner plasmids were identified at the time of isolation for pTF-FC2 and pTC-F14, both can be efficiently mobilised by RP4, an IncP conjugative plasmid (van Zyl et al., 2003). The pTC-F14 IncQ replicon has been cloned into plasmid R388, which could be transfered from E. coli HB101 to At. caldus, and used to allow the replication of this shuttle vector in At. caldus for the construction of allelic exchange mutants (van Zyl et al., 2008b). Additional mobilisable plasmids have been isolated from At. ferrooxidans (pTF4.1), At. ferridurans ATCC 33020 T (previously referred to as At. ferrooxidans pTF1) and At. ferrivorans CF27 (AFERRIp) encoding either MobA/MobL or MobL/ MobS relaxases that enable the plasmids to be mobilised at high frequency by IncP group plasmids (Drolet and Lau, 1992;Tran et al., 2017).
Significanlty less is known about the plasmids of the leptospirilli. No plasmid from this group of acidophiles was known until 2005 when a survey in 10 strains of L. ferrooxidans and 6 strains of L. ferriphilum, identified the first two plasmids for a member of this unique group of bacteria (Table 1). Plasmids p49879.1 and p49879.2 were isolated from L. ferrooxidans ATCC 49879 (previously referred to as R3), and thereafter sequenced and experimentally characterised (Coram et al., 2005). Plasmids p49879.1 and p49879.2 were found to be of similar sizes, i.e., ~30 Kb, but with slightly different overall G+C mol% ratios. Approximately half of the ORFs present on the two plasmids had products related to known proteins in databases. Although no clear candidates for replication proteins could be identified and neither plasmid was capable of replication in E. coli or P. putida, both plasmids contained four 20-bp iterons that may serve as replication protein binding sites. Other plasmid-associated functions such as a mobilisation and stability systems have also been characterised: two divergently transcribed ORFs encoding MobS-like and MobL-like proteins, a MOBQ relaxase and a region of DNA with features typical of the oriT regions of pTF1 and the IncQ plasmids. In spite of this, when cloned in a test plasmid, neither of the candidate oriT could be mobilised by IncP plasmids such as RP4 in E. coli S17.1 (van Zyl et al., 2003). Both plasmids encode proteins with high amino acid sequence identity to ParAlike ATPases responsible for active partitioning of plasmids into progeny cells ! 243 upon cell division, albeit from different ParA families. This divergence may partly account for the observed plasmid compatibility of p49879.1 and p49879.2. Stability experiments showed the par system of p49879.2, but not that of p49879.1, to be functional in E. coli.
A number of accessory genes carried by the mobilisable plasmids found in acidophiles encode an eclectic collection of functions. These include energy metabolism related functions (e.g. a terminal quinol oxidase complex), cofactor biogenesis functions (e.g. heme maturation proteins), electron carriers (e.g. 4Fe-4S ferredoxin), transporters, detoxification functions and regulators (Coram et al., 2005;Gardner et al., 2001;Tran et al., 2017;Zhang et al., 2020). However, their functionalty and whether they contribute to host cell fitness remains unclear.

Integrative Genetic Elements
Many mobile elements that physically integrate in the chromosome (or accesory replicons) populate most prokaryotes (Johnson and Grossman, 2015;Wozniak et al., 2010). Although different bacteria and archaea seem to be differently prone to acquire Integrative Elements (IEs), few of them have been characterised in any detail, particularly in acidophiles. This section focuses on genetic elements of acidophiles that integrate in a site-specific fashion in the host chromosome. This group of elements include proviruses, conjugative self-transmissible elements (known as ICEs; Integrative Conjugative Elements), mobilisable elements (known as IMEs; Integrative Mobilizable Elements) and other elements that are no longer or were never self-transmissible but can be mobilised under certain circumstances with the help of other MGEs, generaly referred to as 'Genomic Islands' (GIs).
These general definitions group together an overarching diversity of elements ranging in size from less than 10 Kb to nearly 500 Kb, sharing one or more diagnostic features. Hallmarks of IEs are functional or cryptic genes encoding integrases and small (16-50 bp) perfect or almost perfect direct repeats flanking the element. The integrases are necessary and sufficient to mediate integration and are also required for excision, although this proces requires additional factors in most cases (Wozniak et al., 2010). Most of the bacterial integrases that mediate site-specific recombination between extra-chromosomal DNA elements and chromosomes are members of the tyrosine recombinase family, and most integration target sites are tRNAs or other structural RNAs (e.g. tmRNAs). Chromosomal integration in archaea differs from bacteria in that the integrase genes become split into two segments which border the integrated element, and that integration can be reversed only if the intact integrase is produced (Wang et al., 2015 and references therein). These mechanistic differences may have implications for productive HGT across the boundaries of the three domains of life.
In recent years, many IEs have been identified on the basis of sequence data and comparative studies of closely related strains or species representing most of the prokaryotic taxonomic spectra. Although several of these elements may actually ! 244 be degenerate, experimental evidence has began to build up that supports different aspects of their functionality.

Viruses and proviruses
Evidence collected over the past 30 years points to a rich diversity of viruses infecting acidophiles from both high and low temperature environments ( Table 2). Most of these viruses have been associated with archaeal acidophiles including Sulfolobus spp., Acidianus spp., the ARMAN nanoarchaea, Nanoarchaeota spp., Metallosphaera spp. and Ferroplasma spp. (Munson-McGee et al., 2020;Wagner et al., 2017;Wang et al., 2018;Wang et al., 2015;Dellas et al., 2014;Andersson and Banfield, 2008;Allen et al., 2007;Prangishvili et al., 2006). These viruses have proved to be unique in terms of their morphology, which includes tailedspindle, filamentous, polihedral, bottle-, lemon-and droplet-shapes, and also in terms of their life cycles, which span integrated and silent proviruses that parasitize their host, infective and lytic forms that kill the cells they infect, and chronic infecting virions that keep their hosts viable while releasing viral progeny Liu et al., 2018;Liu et al. 2017;Hochstein et al., 2016;Wang et al., 2015;Dellas et al., 2014). Tailed viruses infecting bacterial extreme acidophiles of the genus Acidiphilium (Ward et al., 1993) and Acidithiobacillus have also been reported (Covarrubias et al., 2018;Tapia et al., 2012).
Viruses from acidophiles seem to be capable of integrating their genome into that of their host, establishing lysogeny and moving genes around in acidic econiches. Support for the occurrence of proviruses in acidophiles comes from genomic, metagenomic and experimental evidence. Blocks of unique genes with anomalous G+C contents, of putative viral origin, have been found in the "F. acidarmanus" fer1 genome, and in the "alphabet-plasma" Thermoplasmatales and the leptospirilli datasets from the Iron Mountain acid mine drainage (AMD) system (Andersson and Banfield, 2008;Allen et al., 2007). Also fuselloviruses, fusellocaudaviruses, bicaudaviruses and integrative plasmid-virus hybrids inserted at tRNA genes in the genomes of Sulfolobus spp. and Acidianus spp. (e.g. Guo et al., 2011;Hochstein et al., 2016) have been described in the literature. In the case of bacteria, a 59-kb inducible temperate Myoviridae-like prophage located in the srrA tmRNA of the At. caldus type strain genome has been identified (Covarrubias et al., 2018;Tapia et al., 2012). Recently, Hochstein and colleagues, using viral metagenomics and bioinformatic approaches identified NAV1, the first viral genome predicted to infect a member of the Nanoarchaeota from acidic hot springs of Yellowstone National Park, USA, unraveling a novel obligate three-way interaction between a virus, its Nanoarchaeota viral host, and the Sulfolobales cellular symbiont of the Nanoarchaeota (Munson-McGee et al., 2020). Further research on pro/viruses of acidophiles is certain to provide greater insights into the diversity and the unique biology of viruses from acidophiles.
Two large unique genome segments have been identified in At. ferrooxidans strains ATCC 23270 T and ATCC 53993 which account for about 16% difference in gene content between the two strains ). The first is a ~150 Kb long genomic island found exclusively in At. ferrooxidans ATCC 53993 that contains genes encoding mercury, arsenic and copper metal resistance determinants (Valdés et al., 2010). Experimental data have shown that At. ferrooxidans ATCC 53993 has a much higher resistance to CuSO4 (>100 mM) than the type strain (<25 mM), which could be possibly explained by the presence of the additional copper resistance genes in this GI (Orellana et al., 2011;Martínez-Bussenius et al., 2016). Another genomic island, twice as long, is present in At. ferrooxidans ATCC 23270 T Levicán et al., 2009). Its genetic organization resembles that of other well-characterised integrative conjugative elements (ICEs), and includes integration/excision, conjugative DNA processing, transfer, maintenance and regulation proteins organized in a number of discrete gene modules. Excision of the element, expression of a complete set of genes encoding self-transfer functions under normal and DNA-damaging growth conditions and detection of ICEAfe1 T4SS TraA in a pilin pili-like structure of extracellular and intercellular localization, suggests this element is indeed functional (Bustamante et al., 2012;Flores-Ríos et al., 2019b). As in most mobile genetic elements, approximately 60% of its ORFs encode proteins with unknown or hypothetical functions, but additional genes encoding features that may confer selective advantages to the host strain have been identified, including a exopolysacharide biosinthesis cluster a CRISPR-Cas system and a cluster of 36 tRNA genes covering all 20 amino acids. All these accessory functions have been found to ocurr in the mobilome of other sequenced acidithiobacilli (Talla et al., 2014;Travisany et al., 2014;Acuña et al., 2013;Tran et al., 2017).
Comparative genomic studies of At. caldus ATCC 51756 and SM-1 have revealed the presence and activity of >10 integrative mobile elements in At. caldus ranging in size from ~10 to 200 Kb (Acuña et al., 2013). Five of them encode components of type IV secretion systems (T4SS) producing P-type pili for conjugative plasmid transfer that define them as ICE, but only two of these (ICE-2 and ICE-3 type elements) have all the essential components of a functional T4SS, and one encodes at least the relaxase and the coupling protein typically present in non conjugative mobilisable plasmids (IME-1). Experimental analyses revealed the coexistence of the integrated and excised forms of all elements in a collection of At. caldus strains, indicating that the elements are capable of forming circular intermediates suitable for further transfer to recipient hosts. Most ICEs and the IME are widespread geographicaly, indicating that conserved or variant versions occur in the At. caldus population as active excising elements that may be transferred horizontally when needed.
High levels of genomic divergence due to differences in the mobile gene complement have also been identified in other acidophiles. Two dominant and ! 247 recombining strains of Leptospirillum group II from the Iron Mountain AMD showing a very high similarity at the rRNA level (0.3% difference), have differences in gene content approaching 20% (Goltsman et al., 2009;Simmons et al., 2008;Lo et al., 2007). A similar scenario has been pinpointed for Hydrogenobaculum sp. strains (Romano et al., 2013). Integrated plasmid-like elements, virus-like elements and genomic islands encoding conjugal transfer proteins are present in representative strains of all these acidophilic genera. Several of these IEs encode copper-, silver-, arsenic-, and mercury-resistance genes, among other diverse functions, suggesting that they are relevant contributors to fitness and niche adaptation aiding in the dispersal of these and other functions.
Nevertheless, little evidence on the activity and impact on fitness of these IEs and their genetic cargo has been obtained to date. Proteomic studies performed on Iron Mountain AMD leptospirilli found that few proteins encoded by genes in the integrated mobile regions were identifiable, while many of the conjugative transfer proteins encoded by genes in a similar extrachromosomal plasmid were indeed expressed (Goltsman et al., 2009). Further studies have shown that proteins associated with MGEs in the leptospirilli were significantly elevated in low developmental stage biofilms from this environment (Mueller et al., 2010). In turn, studies on Leptospirillum group IV grown in bioreactors have shown that genes encoded in mobile elements are overrepresented in transcriptomic data sets (Goltsman et al., 2013). Further work is required to understand the role and relevance of these large MGEs in adptation and evolution of acidophiles.

Translocative elements
Translocative or transposable elements (TPEs) are components of nearly all prokaryotic genomes. They are capable of moving from one chromosome location to another, regardless of the existence of sequence homology between the element and the target site, by the action of a transposase (TPase) in a process called transposition. Transposases constitute a highly diverse group of enzymes and they are the single largest component in sequenced genomes (encompassing 1-10% of the genome) and environmental metagenomes (Aziz et al., 2010). By transposing into transmisible and integrative elements, they can transfer between hosts and serve as sites of homologous recombination. This allows plasmids and other elements to become inserted into chromosomes and to loop out again, sometimes carrying segments of adjacent chromosomal DNA. These elements can also serve as sites of chromosomal DNA deletions, inversions and a variety of other rearrangements; therefore, they contribute greatly to genetic flexibility (or plasticity) and the entry of genetic material into the horizontal gene pool. Microbial transposable elements can be classified into two large groups: (i) autonomous elements with open reading frames that encode the products required for transposition (i.e. the TPase) together with Insertion Sequences (ISs) and Transposons; (ii) non-autonomous elements that derive from the former by deletion of the TPase while retaining the sequences necessary for in trans transposition by cognate TPases, which include Miniature Inverted repeat Transposable Elements (MITEs).

Insertion Sequences
Insertion Sequences (ISs) are fairly abundant in acidophilic archaea and bacteria ( Figure 2). More than 200 copies of intact ISs of at least 25 different types were found in the S. solfataricus P2 genome and a comparable number in S. tokodaii (Brügger et al., 2002), being this the largest number of transposable elements found in any sequenced genome to date. In the Sulfolobales, the types of ISs vary substantially between species, and even closely related strains can differ greatly with respect to their complement of total and active ISs (Blount and Grogan, 2005). Only a few of the Sulfolobales ISs have been observed experimentally to transpose, exhibiting differences in their transposition frequencies as well as in the degree of target-site specificity (e.g. Martusewitsch et al., 2000). While S. acidocaldarius maintains a very stable genome organization, evidence of extensive IS-catalysed rearrangements in S. solfataricus P2 and S. tokodaii genomes has been decribed (Redder and Garrett, 2006). In turn, in S. islandicus and Ad. hospitalis, the majority of potentially transposable autonomous and nonautonomous mobile elements, as well as many degenerate copies of the former, fall into a large variable region of their genomes, helping preserve overall genome organization (e.g. Guo et al., 2011;You et al., 2011b). In contrast to the crenarchaeota, acidophilic euryarchaeota have significanlty fewer IS elements.
! 249 Figure 2. Number of insertion sequences in complete genomes of acidophilic prokaryotes. The number of predicted insertion sequences found in sequenced genomes of acidophiles, as organized by phylum, is shown in grey. The number of ISs for the most abundant family in each phylum is shown in white. The number of ISs normalized by the number of genomes in each category is indicated above the dotted lines. Allen et al. (2007) reported less than 100 transposase insertions in the genome of the "F. acidarmanus" fer1 isolate and ~50 in the environmental population fer1(env) from which fer1 was derived, representing at most 10 families of ISs. Even fewer ISs are present in Tp. volcanium (27 ISs) and Tp. acidophilum (4 ISs), while P. torridus seems to be completely devoid of them (Brügger et al., 2002). The small size of P. torridus genome (1.55 Mb) and its extreme gene density have been suggested to explain its lack of ISs, yet the dramatic differences observed beween these acidophilic archaeal phyla still lack an adequate explanation.
Although none of the sequenced acidophilic bacteria contains as many ISs as the Sulofolobales, the number and diversity of IS is also high in this group (López de Saro et al., 2013). The occurrence of many IS elements has been reported in recent genome announcements and comparative genomic analyses, but descriptions on active transposable elements have only been published for a small number of acidophiles (e.g. Holmes et al., 2001). Early southern hybridisation experiments discovered two families of actively transposing ISs of about 1.5 kb in size (originally named IST1 and IST2) in a variety of laboratory strains and natural isolates of At. ferrooxidans obtained from different parts of the world. Among IS-positive strains, the number of IST2 copies varied from 15 to 25, and the number of IST1 copies, from 1 to 10 (Holmes and Haq, 1989;Yates and Holmes, 1987). IST2 occurs in a small region of the genome of At. ferrooxidans ATCC 19859, considered to be a hot spot of transposition, and seems to exhibit two different patterns of insertion depending on the growth conditions (Kondrat'eva et al., 2005;Cádiz et al., 1994;Yates et al., 1988). ISTI (currently known as ISAfe1) is an ISL3 family insertion sequence from At. ferrooxidans ATCC 19859 (Holmes et al., 2001). It is about 1.3 kbp in size, has 26-bp imperfectly paired inverted terminal repeats, 5-bp target duplications, and encodes a transposase (OrfA) and a second ORF of unknown function (OrfB). Specific changes in the position of IST1 have been associated with the formation of a large-colony variant of At. ferrooxidans ATCC 19859 that reversibly lost the capacity to oxidise iron and exhibited a swarming phenotype (Schrader and Holmes, 1988). Phenotypic switching was later shown to be correlated with the high frequency insertion and excision of ISAfe1 into, and out of, the resB gene which encodes a cytochrome c-type maturation protein relevant for the iron oxidation pathway (Cabrejos et al., 1999). The cause of the swarming phenotype remains unexplained.
Recent full and partial sequencing of several strains of Acidithiobacillus have provided further insight into the diversity of ISs and their genomic distribution. For instance, natural isolates tend to have more ISs than long laboratory-kept type strains, and these tend to cluster within a few genomic regions, several of which have been identified as different types of genomic islands and transposition hot spots (e.g. Acuña et al., 2013). The genome of an At. caldus strain (SM-1) isolated from a gold bioreactor in China carried many more copies of ISL3, IS4, IS5 and IS21 elements than all other known Acidithiobacillus genomes (You et al., 2011a), and the genome of At. thiooxidans Licanantay strain, isolated from a copper pile in northern Chile, carried 8 times more transposases that the type strain (Travisany et al., 2014). Comparative genomic analysis has also provided ! 250 evidence for recent proliferation of certain families; IS5 and ISL3 in the At. caldus (27 and 37 copies, respectively) and for the ISAs1 family in At. ferrooxidans ATCC 53993 (24 copies) (López de Saro et al., 2013). Only a fairly small percentage of these elements (<10%) have been found to actually interrupt coding regions or to mediate gene acquisition.
A large number of ISs have also been found in Leptospirillum species (Mueller et al., 2010), and comparable differences in the relative abundance of some IS families between closely related species and strains have also been reported for these acidophiles.

Transposons
Tn21 transposons are probably the best examples to illustrate how TPEs can be vehicles of transmission of specific phenotypic traits. Highly arsenic-resistant strains of At. caldus and L. ferriphilum isolated from the same bio-oxidation process treating a gold-bearing arsenopyrite concentrate at the Fairview mine in South Africa have been found to contain distantly related Tn21-like transposons Tuffin et al., 2005;Tuffin et al., 2004;de Groot et al., 2003).
Occurrence of Tn21-like in broad host range plasmids support their disemination among acidophiles via horizontal gene transfer (Clennel et al., 1995). Both the L. ferriphilum and At. caldus ars transposons (TnLfArs and TnAtcArs) were capable of transposition in E. coli and thus, believed to be functional in L. ferriphilum and At. caldus, respectively (Kotze et al., 2006;Tuffin et al., 2006). Despite their presumed functionality in their cognate acidophilic hosts, TnLfArs and TnAtcArs were sufficiently different from each other to indicate that they were obtained independently and not to have been transferred from one bacterium to the other during their coexistence in the biooxidation tank. This suggests that either HGT might not be possible between these phylogenically distant bacteria or that biooxidation tanks are not a suitable environment for HGT between them to occur.
In addition to Tn21 transposons, Tn7-like transposons have also been described to mediate translocation of heavy metal resistance genes in the Acidithiobacilli. In At. ferrooxidans E-15 truncated remnants of Tn7-like genes provide evidende of the role of this family of transposons in gene duplication and shuffling of mer genes (Inoue et al., 1991). In turn, in the type strain of At. ferridurans (ATCC 33020) a Tn7-like transposon (Tn5468) was found to be inserted at a phylogenetically conserved target site, immediately downstream to glmS gene (Oppon et al., 1998). Southern-blot hybridization experiments have shown the presence of Tn7-like transposons in other At. ferrooxidans strains (ATCC 23270 T and ATCC 19859), suggesting that this family of TPEs might be active in natural isolates from different geographical origins (Oppon et al., 1998). Recent phylognetic analyses using Tn7 TnsA protein sequences from a number of acidophiles and several neutrophiles have shown that Acidithiobacillus spp. Tn7 transposons pertain to different clades, adding to the newly appreciated diversity of elements of this family, with different target sites, traget sites selector proteins, and targeting pathways (Peters, 2014;Peters et al., 2017).

! 251
Miniature inverted repeat transposable elements Miniature inverted repeat transposable elements (MITEs), common in eukaryote genomes, are also present in the genomes of acidophilic archaea. Some MITEs are present in multiple copies of identical or near-identical sequence (> 140 copies in S. solfataricus P2), consistent with their having recently transposed within the genome (Redder et al., 2001). Although they are more prevalent in S. solfataricus P2 than in any other sequenced archaeal genome, MITEs are common in other Sulfolobus species, as well as in Tp. volcanium and Ad. hospitalis. Since MITEs can constitute homologous recombination sites, they have been implicated as being responsible for many of the genomic rearrangements observed in archaea . Although MITEs in general show a low level of transpositional activity, they can also transpose to different genomic regions and mobilise genomic regions located between two MITEs. Since they arise from internal deletions in IS elements but retain the terminal sequences, the parent IS element can be frequently identified. Archaeal MITEs have been classifed into two main types, type I MITEs, including two representatives in Tp. volcanium (which derive from ISE1247) and nine in S. tokodaii (which derive from the IS605 family), and type II MITEs present in the genomes of S. tokodaii and S. islandicus at low abundance and of S. solfataricus, where they constitute up to 0.6% of the genome with predicted partner IS elements ISC1048, ISC1217, ISC1058 and ISC1173. One challenge for the future will be to determine whether, and how frequently, the sequence diversity created by MITE insertions has resulted in altered gene expression or gene products in acidophiles.

Defence mechanisms
Protection against potentially deleterious effects of entry and propagation of foreign MGEs has warranted the emergence and selection of alternative defence strategies by both bacteria and archaea . Some of these strategies entail distinction of self-from foreign-DNA by means of DNA modification (methylation or phosphorothioation) and degradation (e.g. Liang et al., 2007) or restriction (e.g. Vasu and Nagaraja, 2013) of non-self DNA; very much like a prokaryotic version of innate immunity. Other systems, in turn, memorise the encounters with MGEs and attack them specifically afterwards. This is the case of the CRISPR-Cas systems, which have been compared to prokaryotic adaptive immunity systems (Koonin and Makarova, 2009;van der Oost et al., 2009). Little is known about the RM-systems in acidophiles. On the contrary, substantial evidence on the ocurrence and relevance of CRISPR-Cas systems in acidophiles has accumulated in recent years. Hints for HGT between distant prokaryotes has been obtained for both RM and CRISPR-Cas systems; interestingly, these systems act to limit HGT of the mobilome elements but are themselves part of it.

CRISPR-Cas systems
CRISPR-Cas systems are used to acquire heritable protection against invading MGEs by both bacteria and archaea. Essential for the function of these systems are the Cas proteins and the CRISPR loci. CRISPRs are made up of a variable number of repetitive DNA motifs, averaging 30 bp, separated by spacer ! 252 sequences of similar size. The spacers are derived from MGEs and incorporated into active loci after exposure to, and successful destruction of, the foreign elements. Upon entrance of MGEs into the cell, the CRISPR loci precursor transcripts are processed into a number of short RNAs, called crRNAs, and displayed on Cas protein complexes. Recognition of the invading foreign DNA is achieved by an RNA-guided sequence specific mechanism through crRNAs matching the invader DNA or RNA, and subsequent degradation is exerted by Cas nucleases. Because of their inherent mechanism of action, CRISPR-Cas systems are both adaptive and heritable and thus constitute a memory of previous attacks which the cell has survived or which the population as a whole has been able to cope with.
CRISPR loci are well represented in acidophiles and have been studied quite extensively in a couple of acidophilic archaeal and bacterial models, namely the Sulfolobales (reviewed by Garrett et al.,, 2015) and the leptospirilli (Tyson and Banfield, 2008). Nearly all sequenced acidophilic archaea and approximately half of acidophilic bacteria encode CRISPR loci and/or Cas proteins in their genomes, fitting general statistics derived from other econiches. By combining evidence from phylogenetic, comparative genomic and structural analyses, basic structural and functional blocks have been recognised and accommodated into a two classes, six types and multiple subtypes based classification (Makarova et al., 2020). Diversity of CRISPR-Cas system types in acidophiles is summarised in Table 3. The most frequent types in acidophilic archaea are I-A, III-B and III-A and in bacteria I-B, I-C and III-B, with 60% of the acidophilic archaea and only 30% of acidophilic bacteria having simultaneously a type I and a type III system. Type II interference systems are absent from all currently known acidophiles. The accepted consensus is that most type I systems target dsDNA, while type III systems target RNA. Thus, acidophilic archaea seem to be better prepared to confront DNA and RNA MGEs, (e.g. archaeophages) than bacterial acidophiles. For instance, the Sulfolobales host subtypes I-A, I-D, III-B, III-D and III-VIII-I CRISPR-Cas (Vestergaard et al., 2014). Such diversity of the CRISPR-Cas system types and spacers (>4000 in complete sequence genomes of the genus) correlate with the considerable diversity and abundance of genetic elements and viruses infecting the Sulfolobales and thriving in thermo-acidic econiches (Garrett et al., 2015;Dellas et al., 2014). Indeed, several viral families containing circular or linear dsDNA genomes as well as positive strand RNA viruses have been detected in the econiches dominated by these thermo-acidophiles (Bolduc et al., 2012). Less impressive numbers are found in other acidophiles (Table 3), but still more than 70% of the completely sequenced acidophiles carry CRISPRs in their genomes, suggesting that viruses and other MGEs are as common a threat in acidic econiches as in other types of environments. Archaeal acidophiles tend to have more arrays per genome (up to 9) and more repeats per array (up to 457) than bacteria. These observations have been made also in microbial communities inhabiting other types of environments. Such comparisons suggest that extremophiles frequently have more CRISPR-Cas systems and larger numbers of CRISPRs. Most acidophiles carry repeats that are highly conserved in sequence within each predicted locus, fitting well-accepted criteria for functionality.

! 253
Nevertheless, with few exceptions, functionality and activity against invading MGEs of most CRISPR loci of acidophiles remain to be evaluated.
A number of studies of the CRISPR-Cas systems of acidophiles have contributed significantly to the general understanding of CRISPR biology. The Sulfolobales have served as a model of study for several aspects of the basic molecular and structural biology of CRISPR-Cas systems (Manica and Schleper, 2013;Lillestøl et al., 2009), providing seminal insights into both the adaptation and interference processes, and their regulation in type I and type III systems (for a review see Garrett et al., 2015). The CRISPR-Cas systems of the acidithiobacilli, which host the rather unique type U loci (for Unclassified), subtype Aferro (for At. ferrooxidans ATCC 23270 T ) (Makarova et al., 2011), have provided insights into the diversity of systems and the challenges associated with their classification. Now acknowledged as Type IV (Pinilla-Redondo et al., 2020), the unusual CRISPR-Cas loci of the acidithiobacilli are located within ICE elements, lack the ubiquitous core genes involved in adaptation (cas1 and cas2) and encode proteins that are highly divergent from other known and characterised Cas (Acuña et al., 2013;Bustamante et al., 2012). Despite the fact that the number and diversification of spacer sequences suggests these loci to be active, it remains unclear whether these are self-sufficient systems or defective systems that capture and utilise pre-existing CRISPR arrays. Further study of the Aferrosubtype CRISPR-Cas systems is likely to enlarge the list of emerging alternative roles of these systems; hints in this direction have begun to emerge in other model bacteria (Crowley et al., 2019). Also, population-level genomic studies in the leptospirilli have proved CRISPRs to be amazing tools for matching viruses to their hosts by means of spacer sequence analysis (Andersson and Banfield, 2008), and for grasping the interactions and dynamics of viruses and microbial host populations of both acidophilic bacteria and archaea (Tyson and Banfield, 2008;Held et al., 2010).

Restriction-modification systems
Foreign DNA is frequently cut into pieces by endonucleases encoded in the genome of the recipient cell, which recognise and cut at specific DNA sequences. Cognate methylases, in turn, protect the hosts DNA from the action of the cuting enzyme by chemically modifying the target sequences, thus achieving an accurate distinction between native and exogenous DNA. Overall, the RM systems limit the frequency of HGT and increase the resistance of cells to viruses. Four types of restriction systems (I-IV) are currently recognized, on the basis of subunit composition, NTP requirement and cleavage mechanism (Roberts et al., 2003). Information stored in the ReBase database reveals that 95% of sequenced bacteria and nearly 100% of archaea encode RM systems, with type II RM systems being by far the most common. This general tendency is conserved in acidophiles (Table 3). Despite their widespread occurrence and high diversity, few RM systems have been characterized so far in acidophiles. These include the Tp.
acidophilus ThaI system (McConnell et al., 1978) and SuaI and SuiI systems from S. acidocaldarius (Pranguishvili et al., 1985) and S. islandicus (Söllner et al., 2006), respectively. While all three systems are capable of DNA restriction, their cognate methylases await discovery and evaluation. Multiple RM-systems ! 254 ! 255 55 Table   3 -Defence systems present in sequenced acidophiles.  populate the genomes of acidophilic bacteria (e.g. > 50 in sequenced acidithiobacilli) and several of these have been shown to map to MGEs (e.g. Acuña et al., 2013;van Zyl et al., 2008a). Studies of RM-systems of acidophiles in general are required in order to overcome some of the barriers to genetic manipulation that many of these models impose.

Outlook
Although acidophiles can be differentiated from neutrophiles and other extremophiles in many aspects of their biology, their mobilome seems to be as diverse and untapped as that of other prokaryotes. Transmissible, integrative and translocative elements described in this chapter represent evidence of the active exchanges of genetic material that have taken place between acidophiles, and have likely had an impact on their adaptation, differentiation and/or niche expansion. Many of the genes being exchanged along with these MGEs belong to functional categories that have previously been shown to be over-represented in the mobilome of model systems and communities thriving in other econiches. In contrast, several other genes provide a clear ecological advantage under nichespecific (acidic) conditions. Several other genes encode hypothetical proteins whose function awaits further discovery. To date, very little insight into the potential advantages (e.g. gene acquisition, genome plasticity, etc.) and disadvantages (e.g. gene inactivation, genome instability, etc.) for cells carrying MGEs has been obtained for acidophiles. Also, the elucidation of the environmental conditions that enhance the proliferation and propagation or containment of these elements by defence systems, are relevant topics to be addressed to improve the understanding of the physiology and ecology of these environmentally-and industriall-relevant microorganisms. Many mechanistic details are likely to emerge from further studies. Both bioinformatic approaches and genetic experiments are required to uncover the dynamic interactions between the genome, the transcriptome and the mobilome. For this to be possible, larger sets of closely related genomes need to get sequenced and a more comprehensive toolbox for genetic manipulation of acidophiles needs to be developed. Both these challenging milestones have begun to be addresed in recent years and will shortly provide new opportunities to expand our knowledge and understanding of the acidophilic mobilome.

Acknowledgement
This manuscript is dedicated to the memory of Douglas Eric Rawlings, to honor his humanity, generosity of spirit, and his many profound contributions to the fields of molecular biology, acidophile microbiology and mineral biotechnology.