Introduction

Hyperthermophilic (superheat-loving) organisms are extremophiles and thrive at temperatures around 80 °C or higher. The first hyperthermophile discovered in Yellowstone National Park by Thomas Brock was Sulfolobus acidocaldarius (Brock et al. 1972). Since then, over 90 hyperthermophilic species have been discovered (Stetter 2006a, b, 2013), most of them belonging to the domain of archaea, but some hyperthermophilic bacteria have also been characterized. Throughout evolution, hyperthermophilic organisms were able to adapt to changing environments such as up-shifts in temperature. Horizontal gene transfer (HGT) probably played an important role in the adaptation to more extreme environments. Except from a role in evolution, exchange of DNA might have also functioned in other mechanisms like DNA repair. DNA exchange is a widespread phenomenon that occurs in all domains of life. This emphasizes its significance for life on earth.

Three main mechanisms for transferring DNA have been described for archaea and bacteria: natural transformation, conjugation and transduction. Natural transformation is the uptake of DNA from the external environment, mostly emerging from lysed cells. Incoming DNA can be degraded and/or can be incorporated into the chromosomal DNA. The recipient cell is usually in charge of what DNA is and what DNA is not taken up (Chen et al. 2005; Lorenz and Wackernagel 1994; Thomas and Nielsen 2005). Conjugation on the other hand is a more invasive mechanism in which the donor has control over the transferred DNA. It requires direct contact between two cells that are not necessarily closely related. Mostly, small plasmids are transferred. Conjugation is suggested to be the main mechanism responsible for HGT (Halary et al. 2010; Norman et al. 2009; Wozniak and Waldor 2010). Transduction involves viruses that function as vehicles enabling DNA exchange between closely related species (Lang et al. 2012; Thomas and Nielsen 2005). Other mechanisms not belonging to these three include: the transfer of DNA via gene transfer agents (GTAs) which are virus-like elements encoded by the host genome (Lang et al. 2012), membrane vesicles (Gaudin et al. 2012) and nanotubes which are cellular protrusions that can bridge neighboring cells (Dubey and Ben-Yehuda 2011). Finally, DNA exchange has been shown to occur between hyperthermophilic Sulfolobus species (Grogan 1996). This DNA exchange is mediated by UV-inducible pili from S ulfolobus (Ups) (Fröls et al. 2008). The different DNA transfer methods among hyperthermophiles are listed in Table 1. This review will describe the fate and function of the DNA transferred between hyperthermophilic prokaryotes; furthermore, it will explain in what way the above-mentioned mechanisms contributed to DNA transfer between prokaryotes in hot environments.

Table 1 DNA transfer methods among (hyper)thermophiles
Full size table

Horizontal gene transfer and evolution

Bacteria and archaea are not able to reproduce sexually; instead they use binary fission for reproduction where DNA is transferred vertically from the mother to the daughter cells. In theory, this does not lead to genetic diversity. Yet, gene-mutation, -loss and -duplication can occur, introducing genetic variation. Moreover, archaea and bacteria can obtain DNA from their environment or nearby organisms and incorporate it into their own genomic DNA (Dickerson 1980). Early studies showing the spread of antibiotic resistances (Datta and Kontomichalou 1965) already made scientists believe that transfer of DNA occurred among bacteria. However, despite these observations, the importance of HGT in genome evolution was underappreciated for decades (Cohan 1994a, b). Therefore, early models of evolution were only based on vertical transfer of DNA (Levin 1981). Much later studies involving comparative genomics showed the extent to which HGT actually played a role in genome evolution (Jain et al. 2003; Koonin and Wolf 2008; Ochman et al. 2000). Nowadays it has become clear that throughout evolution, archaea and bacteria exchanged genes which allowed them to adapt to the changes in the dynamic environments they live in (Guglielmini et al. 2013).

Horizontally transferred DNA can be incorporated into the chromosome using different methods (Thomas and Nielsen 2005; Zaneveld et al. 2008). Besides homologous recombination (HR) between similar DNA molecules, which is executed by DNA repair proteins (described in the following section), other mechanisms exist. Integration can for instance be performed by site-specific integrases such as those found in the Sulfolobus spindle-shaped virus 1 (SSV1) or conjugative Sulfolobus plasmid pNOB8 (described in the “Transduction and HGT among hyperthermophilic viruses” and “Conjugation” sections, respectively) (She et al. 2004, 2006). Moreover, transposition might facilitate HGT (Zaneveld et al. 2008), transposons or insertion sequences (ISs) carrying genes from a previous host might integrate into the genome of another organism that contains the same transposable elements. HR between the ISs or transposons then leads to incorporation of the DNA. An example of the latter mechanism is a 16-kb DNA fragment flanked by ISs that has been transferred between Pyrococcus furiosus and Thermococcus litoralis (Diruggiero et al. 2000).

Bioinformatic methods designed to detect HGT, identify unusual base composition and codon usage within genomes (Garcia-Vallvé et al. 2000; Lawrence and Ochman 1998; Nakamura et al. 2004), or perform phylogenetic analyses on individual genes (Beiko et al. 2005; Puigbò et al. 2010). Studies using these methods estimated that at least 20 % of the bacterial genes and 40 % of the archaeal genes were transferred during evolution (Popa and Dagan 2011; Thomas and Nielsen 2005).

HGT occurs across species boundaries, even between organisms from different domains of life. The latter contributed tremendously to the adaptation of bacteria to hot environments, which is illustrated by the fact that hyperthermophilic bacteria obtained more genes originating from archaea than generally observed with mesophilic bacteria. For instance, the hyperthermophilic Aquifex aeolicus and Thermotoga maritima acquired 16.2 and 24 % of their genes, respectively, from archaea. In comparison, mesophilic Bacillus subtilis, Sunechocystis species, Borrelia burgdorferi and Escherichia coli obtained only 5 % or less archaeal genes (Aravind et al. 1998; Nesbø et al. 2001). Though it must be noted that HGT might, next to the cause of adaptation, also be the consequence of being hyperthermophily as was hypothesized by Nesbø et al. (2001).

The importance of HGT for adaptation to changing environments is illustrated by a genome comparison between Deinococcus radiodurans and Thermus thermophilus (Omelchenko et al. 2005). The two related species diverged from a common, probably mesophilic ancestor (Weisburg et al. 1989), but have surprisingly different phenotypes that correspond to their distinct natural habitats: D. radiodurans is a mesophile that is extremely resistant to ionizing radiation (Anderson et al. 1956), whereas T. thermophilus is a thermophile that shows a rather average sensitivity to ionizing radiation (Oshima and Imahori 1974). It was shown that in both species, adaptation to the highly dissimilar environments was mediated by the loss, and more importantly gain, of several genes during evolution. D. radiodurans obtained bacterial genes that enhanced its ability to survive different kinds of stresses. T. thermophilus on the other hand, acquired numerous archaeal genes that contributed to its thermophilic adaptation. Of these, about 50 % are encoded on a mega plasmid which has an increased plasticity. This is illustrated by the fact that between related T. thermophilus HB8 and T. thermophilus HB27, gene content and order of the plasmid differ substantially. However, both plasmids are implicated in growth at higher temperatures. They both encode several proteins involved in the putative mobile thermophilic-specific DNA repair system. In addition, a gene encoding a reverse gyrase was found on the mega plasmid of T. thermophilus HB8. On the other hand, in the megaplasmid of T. thermophilus HB27, a pseudogene of the reverse gyrase was found suggesting that this gene was once acquired from a hyperthermophilic organism by a common ancestor of the two strains, but was decayed from T. thermophilus HB27 (Brüggemann and Chen 2006; Omelchenko et al. 2005).

The reverse gyrase is found in many thermophiles and in all hyperthermophiles and is probably the best-documented example of HGT being involved in adaptation to hot environments (Déclais et al. 2000; Forterre 2002). Phylogenetic analyses suggest that an early archaeal reverse gyrase gene was acquired by an ancient bacterium. This was then followed by distribution of the gene within the Thermotogales and Aquificales via HGT (Brochier-Armanet and Forterre 2007; Gribaldo and Brochier-Armanet 2006). Its contribution to life at higher temperatures is illustrated by the fact that a reverse gyrase deletion strain of Thermococcus kodakaraensis KOD1 shows growth defects especially at higher temperatures (Atomi et al. 2004). It might therefore be possible that it is only essential for life above 90 °C. Moreover, moderately thermophilic Nautilia profundicola Am-H lives in an environment near hydrothermal vents with fluctuating temperatures; the reverse gyrase of this strain showed a significant upregulation upon up-shifts in temperature (Campbell et al. 2009). The reverse gyrase was previously thought to increase the DNA stability at high temperature by introducing positive coils into the DNA (Kikuchi and Asai 1984). However, not all hyperthermophiles have positively coiled DNA. More recent studies therefore suggest that the reverse gyrase is rather involved in the protection of DNA against degradation which occurs at high temperatures (Kampmann and Stock 2004; Napoli et al. 2004). Although the exact function of the reverse gyrase in this process is not yet fully resolved, its acquisition was almost certainly an important step in the evolution of many thermophilic bacteria (Heine and Chandra 2009).

Adaptations to environmental changes can also be rather subtle. On Vulcano Island (Italy) for instance, there are several shallow vents which have dissimilar and fluctuating physiological conditions (Capasso et al. 1999; Sedwick and Stuben 1996). The genome sequences of several Pyrococcus isolates from these vents revealed extensive genome rearrangements in specific genomic “hot spots” containing mobile genetic elements. It is therefore thought that a number of HGT events took place, possibly contributing to the adaptation to the rapidly changing environments on Vulcano Island (White et al. 2008).

Rachel Whitaker et al. used Sulfolobus islandicus as a model organism to study evolutionary biology in hyperthermophilic archaea (Zhang et al. 2013). By comparing the genomes of 12 different S. islandicus strains from a single hotspring in Kamchatka (Russia), they investigated the rate and mode of recent evolutionary events. Two coexisting groups of S. islandicus species were found that seem to have exchanged DNA mainly within the groups (Cadillo-Quiroz et al. 2012). This is a clear indication of speciation (Dykhuizen and Green 1991). Given the fact that Sulfolobus species exchange DNA in species-specific aggregates (Ajon et al. 2011), the barrier between the two S. islandicus groups could be the inability to successfully create mating pairs (Cadillo-Quiroz et al. 2012). This species-specific recognition is probably driven by Ups-pili and glycosylation patterns (van Wolferen and Albers, unpublished), as will be described later. Strains from the two different S. islandicus groups might therefore show variations in these two structures. Future studies need to confirm whether or not strains from the two different groups are indeed unable to form mixed aggregates. Moreover, glycosylation patterns and Ups-pili between strains from the two different groups need to be compared.

Taken together, genome evolution has been essential for the adaptation of hyperthermophiles to changing and extreme environments in which HGT seems to have been the main player. Without the ability to exchange DNA between species, hyperthermophilic bacteria probably would not exist.

DNA transfer for repair

Next to the crucial role DNA transfer played in genome evolution, other destinations of transferred DNA have become evident. It has been suggested that prokaryotes use imported similar DNA to repair their own DNA with HR (Bernstein et al. 1981). HR is the genetic recombination between two similar or identical molecules of DNA, and is the only efficient mechanism for accurately repairing double stranded DNA breaks (DSBs). It is dependent on another copy of the damaged DNA. This other copy can be the second chromosome that is naturally present during the G2 phase. However, taking up other DNA evidently increases the chances of having a non-damaged homologous template for DNA repair (Bernstein et al. 2012). Bernstein et al. (2012) even propose that DNA repair is the primary function of DNA transfer. For recent reviews describing the DSB repair via HR in bacteria and archaea, see Ayora et al. (2011) and White (2011).

Many bacteria induce their competence genes upon DNA damage which underlines the role that transferred DNA might play in DNA repair. In Streptococcus pneumoniae, competence is induced upon treatment with DNA damaging agents mitomycin C and fluoroquinolone (Claverys et al. 2006). Competence of Helicobacter pylori is induced by ciprofloxacin which causes DSBs (Dorer et al. 2010). B. subtilis shows DNA uptake upon UV stress (Michod et al. 1988), and Legionella pneumophila strongly induces competence upon the addition of several different DNA damaging agents (Charpentier et al. 2011). The only competence system studied in detail in thermophilic prokaryotes is that of Thermus species; this system, however, was shown to be constitutively active (Hidaka et al. 1994). Hyperthermophilic Sulfolobus species on the other hand, exchange DNA upon DSB induction with UV-light or chemical compounds, like bleomycin (Fröls et al. 2008). The mechanism behind this exchange is still unclear, but possibly it is related to the competence system as will be described later. Charpentier et al. (2011) speculate that DNA transfer might have evolved as a DNA damage response in SOS-deficient bacteria. Among the above-mentioned bacteria, S. pneumoniae, H. pylori and L. pneumophila indeed do not show an SOS response. This hypothesis might also be true for Sulfolobus species, as archaea in general have so far not been shown to induce SOS responses. It remains to be shown if other hyperthermophilic bacteria and archaea also display DNA damage-inducible DNA transfer mechanisms that may or may not be similar to that found in Sulfolobus species.

Further evidence supporting the idea that transferred DNA functions in DNA repair is the fact that DNA uptake is often strongly biased towards DNA from the same or closely related species (Seitz and Blokesch 2012). In the Pasteurellaceae and Neisseriae it was shown that this bias is provoked by recognizing so-called DUS sequences that are present multiple times in their own DNA (Danner et al. 1980; Elkins et al. 1991; Fitzmaurice et al. 1984; Mell et al. 2012; van Passel 2008). In Vibrio cholerae on the other hand, the competence system is strongly induced by the presence of the species-specific auto inducer CAI-1, which also leads to a higher uptake of DNA from the same species (Suckow et al. 2011). Other bacterial species make use of restriction modification systems that degrade foreign DNA (Murray 2002) and increase thereby the chances of incorporating self-DNA. Likewise, CRISPR-Cas systems can target foreign DNA, and thereby hamper HGT, as was shown for Staphylococcus epidermidis (Marraffini and Sontheimer 2008). A recent study even showed that the presence of CRISPR loci seems to be linked to the presence of an active competence system (Jorth and Whiteley 2012). Uptake of self-DNA can additionally be increased by the induction of cell death of neighboring cells from the same species (fratricide), as has been shown for S. pneumoniae (Guiral et al. 2005; Håvarstein et al. 2006) and probably also H. pylori (Dorer et al. 2010). Lysed cells provide free DNA that can be taken up for possible DNA repair. Lastly, in the archaeal Sulfolobus species, DNA exchange occurs within species-specific cell aggregates (Ajon et al. 2011). In that way, only species-specific DNA is exchanged.

A bias towards species-specific DNA also occurs at the level of HR. In general, this process evidently gets less efficient for more dissimilar sequences to integrate into the DNA (Majewski and Cohan 1998, 1999; Wolf et al. 2001; Zawadzki et al. 1995). Therefore, in all prokaryotic species there is a natural bias towards integration of DNA from the same or closely related species.

In (hyper)thermophilic organisms one could hypothesize that DNA repair mechanisms are of particular importance as the rates of spontaneous DNA mutations are elevated at high temperatures (Lindahl 1993). However, the genomic mutation rate in the hyperthermophilic archaeon S. acidocaldarius was shown to be equal to mesophilic organisms (Grogan et al. 2001). Given the fact that DNA stability is more important at higher temperatures, mutations probably have a more drastic effect in hot environments. In (hyper)thermophiles, mutation rates should therefore theoretically be even lower than in mesophiles to give the same effect. By extrapolating the mutation rate of used markers to the whole genome, recent studies indeed showed that the rate of base substitutions in S. acidocaldarius and T. thermophilus is lower compared to mesophilic organisms (Drake 2009). Spontaneous mutations in the DNA of (hyper)thermophiles must consequently be repaired so rapidly that they cannot be measured. Thus, efficient DNA repair systems seem to be present; this may involve efficient HR mechanisms that use transferred DNA as a template. In general, the role of DNA transfer in DNA repair of prokaryotes might be far more important than previously thought.

DNA as a nutrient source

An alternative fate for imported DNA might be the use of nucleotides as a nutrient source (Palchevskiy and Finkel 2009; Redfield 1993a). This idea is supported by the fact that B. subtilis and Haemophilus influenzae induce competence upon nutrient starvation and not upon, for instance DNA damage (Redfield 1993a, b). Moreover, non-homologous or partially degraded DNA molecules might not be suitable for HR and therefore better used as nutrient source (MacFadyen et al. 2001; Redfield et al. 1997; Redfield 1988). Proof for this theory is amongst others the observation that E. coli is able to grow with DNA as sole carbon and energy source, which is dependent on the presence of homologs of competence genes in H. influenzae and Neisseria gonorrhoeae (Finkel and Kolter 2001; Palchevskiy and Finkel 2006).

In hyperthermophilic environments, unprotected DNA is degraded faster than in mesophilic surroundings (Lindahl 1993). One could therefore speculate that bad-quality DNA that is commonly present is taken up by competent hyperthermophiles and used as a nutrient source. However, no evidences in agreement with this hypothesis have been shown so far. Moreover, it seems unlikely that the main function of DNA uptake is to serve as a food source when far less energy consuming processes are capable of taking up nutrients just as well. Hence, it seems more plausible that only DNA that cannot be incorporated into the genome is used as a source of nutrients.

Conjugation

Conjugation is the unidirectional transfer of DNA between cells by a process requiring cell–cell contact (de la Cruz et al. 2010; Gomis-Rüth and Coll 2006). This process has a host range larger than observed for transformation or transduction (Guglielmini et al. 2013) and is thought to be the main mechanism involved in HGT (Halary et al. 2010; Norman et al. 2009; Wozniak and Waldor 2010). During conjugation, DNA transfer is carried out by the donor cell by means of a bridge-like connection between two cells; the recipient cell seems to have little control in this process (Pérez-Mendoza and de la Cruz 2009). Mostly plasmids and ICEs (integrating conjugative elements) are transferred. Conjugation has thereby contributed significantly to the rapid spread of antibiotic resistance, virulence, and social traits among prokaryotes (Gomis-Rüth and Coll 2006; Guglielmini et al. 2013; Schröder and Lanka 2005).

Conjugation systems form the largest subfamily of the Type IV secretion systems (T4SS); the latter also include DNA uptake and release systems and protein translocation systems (Schröder and Lanka 2005). The T4SS is a large protein complex spanning the complete bacterial cell envelope. Proteins essential for T4SSs are: a TraU/VirB4 ATPase energizing both the assembly of the system as well as the substrate transfer, and 12–20 mating-pair formation proteins (MPFs) promoting physical contact with the recipient cell (Schröder and Lanka 2005). Besides the general T4SS proteins, conjugation-specific proteins include a relaxase and a type IV coupling protein (T4CP) (de la Cruz et al. 2010). The relaxase initiates conjugative transfer by binding and nicking the DNA at the origin of transfer (oriT). The T4CP then couples the DNA to the channel-forming T4SS that subsequently transfers the nucleoprotein complex through the membrane of the donor cell and delivers it into the recipient cell (de la Cruz et al. 2010; Vogelmann et al. 2011). A gene cluster encoding a VirB4 ATPase, a T4CP and a relaxase is considered to encode a putative conjugative system; without the relaxase, the system is thought to be a protein-exporting T4SS (Guglielmini et al. 2011).

In all different conjugation systems, assembly of the pilus is essential to draw cells close to each other and allow exchange of macromolecules from cell to cell. The best-characterized T4SS surface structures are the conjugative pili from Gram-negative bacteria. There is a huge diversity in composition and structure of these filaments. Pili encoded by the F plasmid from E. coli are long (2–20 μm) and flexible, with a diameter of 8 nm, while P-pili from IncP plasmid RP4 are very short, rigid and have diameters from 8 to 12 nm (Alvarez-Martinez and Christie 2009; Lawley et al. 2003; Silverman 1997).

Conjugation in hyperthermophilic Bacteria

Thermophilic Thermus species are equipped with a highly efficient natural competence system (described in the following section). Recently, it was speculated that also conjugation occurs among Thermus species (César et al. 2011). Given the low solubility of oxygen in geothermal environments, certain T. thermophilus strains express a nitrate reductase, which makes anaerobic growth possible. This nitrate reductase is encoded in the nir-nar-nor cluster, present as a variable region on mega plasmid pTT27 (Alvarez et al. 2011; Bricio et al. 2011). It was shown that T. thermophilus strains NAR1, HB27 and PRQ25 were able to exchange pTT27 among each other, thereby enabling anaerobic growth (Alvarez et al. 2011; Ramírez-Arcos et al. 1998). In addition, when a hygromycin resistance marker (hyg) was placed on pTT27, this gene could be transferred to non-competent cells with frequencies of up to 10−2 (César et al. 2011). These findings support the presence of an active conjugation system. However, no conjugation-like homologous have been found in the above-mentioned T. thermophilus strains. Therefore, another unrelated conjugation system must be present in these species (César et al. 2011). A recently sequenced megaplasmid pTHTHE1601 from T. thermophilus SG0.5JP17-16 revealed the presence of a putative VirB operon, possibly involved in conjugation. Proteins encoded by this operon include TrbC/VirB2 (Ctr1), VirB4 (Ctr3) and VirD4 (Ctr7) ATPases. Future experiments need to confirm the participation of this T4SS in conjugation (César et al. 2011).

Conjugation in hyperthermophilic Archaea

Based on bioinformatics methods (hidden Markov models) scoring for co-localization of genes encoding a virB4/traU ATPase, a T4CP, a relaxase and MPF-specific proteins, only two archaeal conjugative elements could be found: one ICE encoded in the chromosome of the thermoacidophilic Aciduliprofundum boonei, and one conjugative plasmid in Haloarcula marismortui (pNG500). However, many VirB4 homologues were found encoded on archaeal chromosomes or plasmids, often associated with T4CP-like proteins. Because conjugative plasmids were found in Sulfolobus species (described below) it might be possible that unknown relaxases exist in archaea (Guglielmini et al. 2011).

Around 3 % of all isolated Sulfolobus strains contain self-transmissible conjugative plasmids (Prangishvili et al. 1998). The best-characterized plasmids are pNOB8 from Japanese Sulfolobus strain NOB8H2 (Schleper et al. 1995; She et al. 1998) and pING1 isolated from S. islandicus strain HEN2P2 (Prangishvili et al. 1998; Stedman et al. 2000). The first archaeal plasmid shown to be horizontally transferred is a relatively large plasmid (45 kb) pNOB8. It can propagate easily in liquid cultures of its original host in mating mixtures with Sulfolobus solfataricus and S. islandicus via conjugative-like transfer mechanism. pNOB8 has very high copy numbers of between 20 and 40 plasmids per chromosome. When pNOB8 is transformed to Sulfolobus cells; the cells become a donor and are able to transfer the plasmid to other cells. The same was observed for recipient cells containing the transmitted plasmid (Schleper et al. 1995). Interestingly, upon mixing a donor and recipient strain, cellular aggregates could be observed. Moreover, electron microscopy revealed intercellular cytoplasmic bridges that connect two or more cells (Fig. 1a) (Schleper et al. 1995), which resemble those found in halophilic Haloferax volcanii. The cytoplasmic bridges of H. volcanii mediate the bidirectional transfer of chromosomal DNA representing the first known mating system in archaea (Rosenshine et al. 1989). pNOB8 was also shown to be able to integrate site-specifically into the host genome within a tRNA gene (in the so-called attP site) using a self-encoded integrase (She et al. 2004).

Fig. 1
figure 1

a Scanning electron micrographs of donor strain S. solfataricus NOB8H2 (upper left) and 1:1 mating mixtures with S. solfataricus NOB8H23, showing cell aggregates and intercellular bridges. Reprinted from Schleper et al. (1995). b Transmission electron micrographs from cross sections of nanotubes from Thermococcus sp., scale bar 200 nm (a) and 100 nm (b) (a and b). Nanotube associated with a MV, scale bar, 100 nm (c). Scanning electron micrograph of a network of cells linked by nanotubes, scale bar 500 nm. Inset partially disrupted nanotube showing an artificially twisted internal core extruding from an outer envelope; scale bar, 100 nm) (d). Reprinted from Marguet et al. (2013)

Full size image

Several other archaeal conjugation plasmids have been characterized during the last two decades. All were found in Sulfolobus species and they are grouped into pKEF or pARN plasmids (Alvarez-Martinez and Christie 2009; Greve et al. 2004). In all analyzed conjugative plasmids only a few genes encode proteins homologous to bacterial conjugation proteins, including TraG/VirD4, TrbE/VirB4 and partitioning proteins ParA and ParB. Therefore, despite the fact that conjugational exchange of plasmids has been observed between Sulfolobus species, the exact mechanism is still unclear.

Importantly, Sulfolobus species were also shown to bidirectionally exchange chromosomal DNA, possibly via a conjugation-like mechanism (Aagaard et al. 1995; Ajon et al. 2011; Ghane and Grogan 1998; Schmidt et al. 1999). This mechanism is mediated by UV-inducible pili and will be discussed separately in one of the following sections.

Overall, conjugational exchange of plasmid DNA has been observed both in hyperthermophilic bacteria and archaea; however, only very few of the classical conjugation proteins are found in these species. It seems therefore that conjugation at high temperatures involves different specialized machineries that sofar have not been identified and characterized.

Natural transformation

Natural transformation refers to the uptake of exogenous DNA by naturally competent species (Dubnau 1999). It is a widespread phenomenon in bacteria that has been described for at least 70 bacterial species from all major taxonomic groups. In contrast to conjugation, competence does not require mobile genetic elements (Johnsborg et al. 2007). As previously discussed, imported DNA can be used for different cellular processes, such as DNA repair, nutrition and the introduction of genetic diversity (Seitz and Blokesch 2012). The process of natural transformation starts with the induction of competence, followed by the binding of environmental DNA, which is subsequently fragmented whereupon it enters the cells. If the entered DNA shows homology to certain genomic DNA regions it can be integrated into the chromosome (Averhoff 2009).

Despite the fact that competence systems are common in bacteria, initiation and regulation of the systems differ between species. Some organisms have highly regulated competence that is triggered by certain signals like pheromones, nutrient limitation or high cell density. Competence in Gram-positive B. subtilis and S. pneumonia has been studied in detail and seems to be a general response to stress (Johnsborg et al. 2007; Lorenz and Wackernagel 1994). In Gram-negative bacteria, regulation of competence is less well understood. However, it is known that H. influenza induces competence upon starvation (Herriott et al. 1970) and Vibrio cholera upon growth on chitin (Meibom et al. 2005). Moreover, as described previously, many bacteria show induction of competence upon DNA damage (Charpentier et al. 2011; Dorer et al. 2010; Michod et al. 1988), and thus competence may serve for DNA repair via HR. The only well-studied thermophilic bacteria harboring natural competence systems belong to the genus Thermus. These competence systems seem to be constitutively active (Hidaka et al. 1994), as was also shown for mesophilic N. gonorrhoeae (Dubnau 1999). Bacterial competence is in general subject to regulation by complex signal transduction pathways [for reviews see Claverys et al. (2006), Seitz and Blokesch (2012)].

Due to their vastly different cell envelopes, DNA uptake mechanisms in Gram-positive and -negative bacteria structural differ, in particular, the presence of a secretin ring is unique for the outer membrane of Gram-negative bacteria. However, most of the other proteins are shared (reviewed in Krüger and Stingl 2011) indicating a large similarity in DNA uptake mechanisms. Most bacterial competence systems are composed of a DNA-translocation complex that is coupled to a type IV pilus (T4P) or an evolutionary related type II secretion system (T2SS), proteins in both systems include a pre-pilin peptidase, a secretion ATPase, a polytopic transmembrane protein and pilin- or pseudopilin-subunits. The DNA uptake systems of H. pylori and Campylobacter jejuni form an exception as they are related to the dissimilar T4SSs (described in the “Conjugation” section) (Hofreuter et al. 2000; Stingl et al. 2009).

The exact role of the pili or pseudopili in competence is still not well understood, but their presence is essential for successful DNA uptake (reviewed in Krüger and Stingl 2011). It has been suggested that DNA is brought close to the cell surface by binding to the (pseudo)pili that subsequently retract. Until recently, this hypothesis was questioned as no binding of DNA to (pseudo)pili had ever been observed (Assalkhou et al. 2007; Provvedi and Dubnau 1999). However, a very recent paper described the binding of minor pilin ComP of competent Neisseria meningitidis to DUS sequences suggesting that pili are indeed bringing DNA to the cell surface (Cehovin et al. 2013). For detailed reviews about bacterial competence systems, see Averhoff and Friedrich (2003), Chen and Dubnau (2004), Claverys et al. (2009), Johnsborg et al. (2007) and Krüger and Stingl (2011).

Naturally competent hyperthermophilic Bacteria

Of the at least 70 naturally competent bacteria that are found so far (Johnsborg et al. 2007), six are (hyper)thermophilic. One of these is the thermophilic cyanobacterium Thermosynecoccus elongates BP-1 (Onai et al. 2004), all others belong to the Thermus genus, these are: T. aquaticus, T. caldophilus, T. flavus, T. thermophilus (Koyama et al. 1986) and T. scotoductus (Gounder et al. 2011). With transformation frequencies of up to 10−2, hyperthermophilic T. thermophilus was shown to have the most efficient DNA uptake system of all studied naturally competent bacteria (Koyama et al. 1986). This was confirmed by the fact that DNA uptake had a speed of around 40 kb/s per cell, which is extremely fast (Schwarzenlander and Averhoff 2006). In comparison, other bacteria like B. subtilis and H. influenza show rates of 4 and 16 kb/s per cell, respectively (Deich and Smith 1980; Dubnau 1991). T. thermophilus has for that reason become a model organism for studying natural transformation in bacteria.

Additional properties that make the T. thermophilus DNA uptake system of great interest is that it is constitutively active (Hidaka et al. 1994) and that it is equipped with a broad substrate specificity. T. thermophilus HB27 does not show any bias towards certain types of DNA; genetic material from all domains of life can be taken up with equal efficiencies (Schwarzenlander et al. 2009; Schwarzenlander and Averhoff 2006). From this point of view, one could speculate that the Thermus competence system is involved in HGT rather than in DNA repair. This hypothesis is strengthened by the fact that throughout evolution Thermus species have indeed undergone several HGT events; these events are suggested to be involved in the gain of thermophilicity (Averhoff and Müller 2010; Omelchenko et al. 2005). It is therefore hypothesized that Thermus species developed a highly efficient competence system to acquire all sorts of DNA, which allowed them to adapt through the incorporation of foreign, probably primarily archaeal, genes.

Through whole-genome comparisons and mutational analyses, 16 competence genes were identified in T. thermophilus (Friedrich et al. 2001, 2002, 2003). Genome comparisons further revealed that these genes are conserved among all sequenced Thermus species (César et al. 2011). The proteins encoded by those genes are divided into three groups: type IV pili-related proteins (PilA1–4, PilD, F, C, Q), DNA translocator proteins (ComEA, ComEC, DprA) and novel Thermus-specific proteins (ComZ, PilM, N, O, W) (Averhoff 2009) (Fig. 2a). Among the type IV pili-related proteins there is a conserved pre-pilin signal peptidase PilD which is, as was shown in many other bacteria and archaea, responsible for the maturation of the pilin subunits (Nunn and Lory 1991). Four predicted pilin subunits are present: PilA1, PilA2, PilA3 and PilA4, which are essential for pili formation (Friedrich et al. 2002). So far, no ATPase energizing the assembly of the type IV pili has been identified in T. thermophilus. However, a deletion mutant of a second traffic-ATPase, PilF, resulted in piliated but non-competent cells (Friedrich et al. 2001), PilF is therefore thought to be a retraction ATPase functioning similar to PilT in N. gonorrhoeae (Maier et al. 2002; Wolfgang et al. 1998). Other type IV pili-related proteins are: the inner membrane protein PilC, anchoring the pili into the inner membrane, and the outer membrane protein PilQ whose structure was recently elucidated (Burkhardt et al. 2011). The latter is similar to secretins in other Gram-negative bacteria and was shown to bind DNA. PilQ is thought to form a membrane channel that guides the DNA through the outer membrane to the DNA translocator which is present in the inner membrane (Burkhardt et al. 2012; Rumszauer et al. 2006; Schwarzenlander et al. 2009).

Fig. 2
figure 2

Proposed DNA transfer systems in (hyper)thermophilic Thermus and Sulfolobus species. a The natural competence system of T. thermophilus: Pilin subunits PilA1–4 are processed by signal peptidase PilD and subsequently build into a pilus. PilC anchors the structure into the inner membrane. DNA enters the cell envelope via outer membrane protein PilQ. Upon pili retraction energized by the ATPase PilF, DNA enters the periplasmic space and it is handed over by ComEA to channel-forming ComEC transporting the DNA further into the cytoplasm. Before entering the cell, one strand of the DNA is degraded by a yet unknown nuclease while the other strand of incoming DNA is guided by DprA to RecA. Accessory proteins PilM, PilN, PilO, PilW and ComZ, are either involved in the assembly of the transporter and/or assist transport of DNA across the inner membrane. b The UV-inducible pili system in of Sulfolobales (Ups-system): Pilin subunits UpsA and UpsB are processed by signal peptidase PibD and subsequently build into a pilus, energized by ATPase UpsE. UpsF anchors the structure into the membrane. OM outer membrane, PG peptidoglycan, IM inner membrane, M membrane

Full size image

The three DNA translocator proteins found in T. thermophilus are all homologous to conserved competence proteins of other bacteria: ComEA is a protein anchored to the inner membrane that binds DNA and subsequently guides it to ComEC. ComEC forms a channel in the inner membrane and transports the DNA into the cytosol. During this process, incoming DNA is probably linearized by an unknown nuclease that degrades one of the two strands. Single stranded DNA is subsequently thought to be handed over to RecA by DprA as shown for S. pneumoniae (Mortier-Barrière et al. 2007). As RecA is essential for transformation (Martin et al. 1995), transferred DNA might then be integrated into the genomic DNA via a RecA-dependent process.

Not much is known about the Thermus-specific competence proteins; PilM, N, O and ComZ are all present in the cytoplasmic membrane (Friedrich et al. 2003; Rumszauer et al. 2006) and are suggested to be involved in the assembly of the DNA-translocation system (Averhoff and Müller 2010). PilW is thought to be responsible for proper localization of PilQ in the outer membrane (Rumszauer et al. 2006). As T. thermophilus lives at extremely high temperatures, the unique competence proteins might be an evolutionary adaptation to these special environmental conditions (Averhoff and Müller 2010). For detailed reviews about the competence system in T. thermophilus see, Averhoff (2009), Averhoff and Müller (2010) and César et al. (2011).

Naturally competent hyperthermophilic Archaea

Only four archaeal species have so far been found to be naturally competent (Bertani and Baresi 1987; Sato et al. 2003; Waege et al. 2010; Worrell et al. 1988), including two hyperthermophilic organisms. The first naturally transformable hyperthermophilic archaeon was Thermococcus kodakarensis (Sato et al. 2003), a sulphur reducing euryarchaeon that lives in marine hydrothermal vents and has an optimal growth temperature of 85 °C. It can be transformed naturally with linear and circular DNA, which is subsequently integrated into the genomic DNA. The observed transformation frequencies of 10−7 are rather low (Sato et al. 2003, 2005). Another naturally competent member of the Thermococcales is P. furiosus which grows at temperatures around 100 °C (Waege et al. 2010). Also this species can be transformed with linear and circular DNA and shows transformation frequencies of up to 10−3 approaching those of highly competent bacteria (Lipscomb et al. 2011). Among the crenarchaeota, so far no naturally competent organisms have been found. Nevertheless, Grogan and Stengel (2008) were able to show that the hyperthermophilic S. acidocaldarius is able to constitutively take up short single stranded oligonucleotides. Though, as earlier experiments revealed that no DNA uptake could be observed when cells were mixed with purified marker DNA, culture supernatant or lysed cells (Grogan 1996), one could question if we can speak here about true natural competence for S. acidocaldarius.

The mechanisms of natural transformation in (hyperthermophilic) archaea have not been studied, and the systems involved have not been identified as homologs of bacterial competence systems have sofar not been found (Averhoff 2009; Claverys et al. 2009). However, the complete DNA sequence of hyperthermophilic crenarchaeon Thermoproteus tenax, revealed the presence of a few homologs of bacterial genes involved in competence (dprA and comF). Since these have not been detected in any other archaea, they were probably acquired from bacteria via HGT (Siebers et al. 2011). However, the existence of a functional bacterial-like competence system in T. tenax needs to be proven. It is likely that naturally transformable archaea that did not acquire competence genes from bacteria have other unrelated DNA uptake mechanisms, which will be a challenge for future studies.

Transduction and HGT among hyperthermophilic viruses

Transduction refers to the transfer of DNA from one cell to another by means of viruses and was first described for Salmonella typhimurium (Zinder and Lederberg 1952). In this process, viruses accidentally package host DNA along with their own DNA and subsequently infect other prokaryotes. In that way, surviving host cells might acquire genetic material from previous hosts (Canchaya et al. 2003). Transduction therefore probably contributed significantly to HGT. Whole-genome analyses indeed suggest that phages played an important role in the acquisition of new genes thereby promoted genetic diversity (Bordenstein and Reznikoff 2005; Brüssow et al. 2004; Ochman et al. 2000; Pallen and Wren 2007). Moreover, when two different viruses infect the same host simultaneously, HGT might not only occur between previous and current hosts, but also among the viruses (Krupovic et al. 2011), leading to viral genetic diversity.

Similar to viruses, gene transfer agents (GTAs), which are virus-like elements encoded by the host-genome, contribute to HGT. It has been demonstrated that GTAs can horizontally transfer a kanamycin resistance gene very efficiently between bacterial species (McDaniel et al. 2010), which implies a general importance of GTAs in HGT. However, so far no GTAs have been found in hyperthermophilic organisms.

Since 1959, about 6,300 prokaryotic viruses have been described morphologically (Ackermann 1996, 2001, 2007; Ackermann and Prangishvili 2012; Eisenstark 1967). Most of these are bacterial viruses (98.5 %), however, over the past few years increasing numbers of archaeal viruses have been described, especially those infecting hyperthermophilic crenarchaea. All studied hyperthermophilic viruses were shown to contain DNA and not RNA, but a recent study revealed the presence of viral RNA in acidic hot springs in Yellowstone National Park, USA, probably originating from viruses infecting hyperthermophilic archaea. These unidentified RNA viruses might form a novel group of phages (Bolduc et al. 2012). For a detailed review about bacterial and archaeal viruses, see Krupovic et al. (2011).

Viruses infecting hyperthermophilic Bacteria

Within the (hyper)thermophilic bacteria, so far only a relatively small number of phages has been found. In total 27 hyperthermophilic bacteriophages have been morphologically described, all infecting Thermus species (Ackermann and Prangishvili 2012). These include: Inoviridae, Myoviridae, Siphoviridae, and Tectiviridae. The genomes of myovirus φYS40 (Naryshkina et al. 2006; Sakaki and Oshima 1975), siphoviruses P23-45, P74-26 (Minakhin et al. 2008; Yu et al. 2006), IN93 and icosahedral virus P23-77 (Jaatinen et al. 2008; Jalasvuori et al. 2009) were sequenced which contributed greatly to the understanding of regulation of transcription and translation as well as the evolution of Thermus phages. The sequence of P23-77 showed an evolutionary link to another Thermus phage, IN93. Interestingly, it also revealed evolutionary relationships to haloarchaeal plasmid pHH205, an integrated Haloarcula provirus, and the Haloarcula virus SH1. These similarities include homologies between two major capsid proteins and a putative packaging ATPase (Jalasvuori et al. 2009). A possible explanation for their relatedness could be that the haloarchaeal and Thermus viruses evolutionarily diverged from each other along with their hosts. Subsequently, the viruses may have evolved into plasmids or proviruses. Another hypothesis is that viruses were able to cross the domain barrier and could thereby mediate HGT between the two domains (Jalasvuori et al. 2009). In the course of this process, other host genes might also have been transferred.

The DNA sequence from another bacteriophage φIN93 revealed the presence of an IS that was shown to originate from its host T. thermophilus (Matsushita and Yanase 2009). Upon a future infection, the virus might again transfer this IS to another organism, leading to HGT.

Viruses infecting hyperthermophilic Archaea

Among the studied hyperthermophilic viruses, most were found to infect members of the crenarchaeota, including: Acidianus, Aeropyrum, Pyrobaculum, Stygiolobus, Sulfolobus and Thermoproteus species. These viruses were shown to be highly diverse both on morphological and genomic level and have been classified into eight viral families: Ampullaviridae, Bacaudaviridae, Clavaviridae, Fuselloviridae, Globuloviridae, Guttaviridae, Lipothrixviridae and Rudiviridae (Ackermann and Prangishvili 2012).

The spindle-shaped fuselloviruses that infect Sulfolobus and Acidianus species are among the best-studied crenarchaeal viruses (Lipps 2006). The genome of Sulfolobus spindle-shaped virus 1 (SSV1), isolated from Sulfolobus shibatae was the very first archaeal viral genome being sequenced (Palm et al. 1991). Since then, 13 more fusellovirus genome sequences have been published (Held and Whitaker 2009; Redder et al. 2009). As was shown for conjugative plasmid pNOB8, SSVs can stably integrate in a Sulfolobus chromosome via integration within a tRNA gene (in the so-called attP site), thereby forming a provirus (Held and Whitaker 2009; Muskhelishvili et al. 1993). This site-specific integration is thought to be an important mechanism for HGT and genome evolution. The SSV1-type integrase that is essential for this integration, is next to the pNOB8-type integrase (described in the “Conjugation” section) of particular interest in studies on the integration of crenarchaeal mobile genetic elements (She et al. 2006; She et al. 2004). SSVs might in addition be involved in the HGT of plasmids; upon an SSV-infection, non-conjugative pRN-like plasmids could be encapsulated into virus-like particles that are released from the cell. Subsequently, they can spread with the help of the viruses and integrate into a following host genome using a self-encoded SSV1-type integrase (Arnold et al. 1999; Wang et al. 2007). Although they cannot infect S. acidocaldarius, SSVs are not host species dependent, it was shown that they can infect different Sulfolobales (Ceballos et al. 2012): Consequently, virus-mediated HGT might have occurred between the different hosts of SSVs (Ceballos et al. 2012).

Next to the Fuselloviridae, members from the Rudi- and Lipothrixviridae also infect Sulfolobus and Acidianus species; these are helical viruses with a linear dsDNA genome (Prangishvili et al. 2006a). Comparative genomics revealed that members from the two virus families horizontally exchanged genes. Interestingly, it was shown that both viruses also obtained genes from their hosts (Peng et al. 2001). The latter include genes encoding a dUTPase: a thymidylate synthase (ThyX) and a Holliday junction resolvase (Peng et al. 2001; Prangishvili et al. 2006b). This acquisition of host genes could have possibly led to HGT between different hosts. Comparable to the fusseloviruses, a lipothrixvirus also shows a relationship with an integrative plasmid. Acidianus filamentous virus 1 (AFV1) mediates the horizontal spread of pAH1 between different hosts. The mechanism involved is so far not known (Basta et al. 2009).

Other crenarchaeal viruses involved in HGT are the Sulfolobus turreted icosahedral viruses (STIVs). A sequence comparison between STIV and related STIV2, revealed a similar genome organization as well as the loss and gain of several genes (Happonen et al. 2010). STIV2 shares genes with other archaeal viruses, including the Fuselloviridae, and it also contains a conserved crenarchaeal gene encoding a DNA-binding protein. Moreover, Thermococcus kodakarensis virus 4 (TKV4) and Methanococcus voltae virus (MVV) were shown to encode capsid proteins and genome packaging ATPases that were similar to those from STIV-like viruses (Krupovic and Bamford 2008). It is therefore clear that HGT has occurred between these archaeal viruses (Koonin 2009; Koonin and Wolf 2008). This transfer of DNA might have been mediated by viruses structurally related to STIV able to infect both eury- and crenarchaeota (Krupovic et al. 2011).

Only two hyperthermophilic euryarchaeal viruses have been described, both infecting members from the order of thermococcales: Pyrococcus abyssi virus 1 (PAV1) (Geslin et al. 2003, 2007) and Thermococcus prieurii virus 1 (TPV1) (Gorlas et al. 2012). They are morphologically similar to the Fuselloviridae and infect cells without causing lysis. Interestingly, the PAV1 genome shows homology to several different plasmids and integrating elements from other archaeal species; these genes are all encoded on approximately the same half of the viral genome. Genes encoding capsid proteins are located on the other half of the genome. This genetic organization could be due to a fusion of a plasmid and a virus, or otherwise, PAV1 might have given rise to plasmids and integrating elements. As was described for SSV1 and AFV1 viruses, PAV1-like viruses might have been involved in the horizontal transfer of plasmids (Krupovic et al. 2010).

In general, it seems clear that virus infections contributed significantly to HGT between viruses as well as their hosts. In the course of evolution this has led to a highly diverse virus population as well as dynamic prokaryotic genomes. Future studies might give more detailed insights in the impact that viruses have on the evolution of their hosts especially in hyperthermophilic environments.

Membrane vesicles, nanopods and nanotubes

The release of membrane vesicles (MVs) is an important physiological process for organisms from all domains of life. MVs mediate the intercellular transfer of several different biological compounds including DNA. They might therefore have contributed to HGT. Both bacterial and archaeal MVs are produced from the cell surface (Deatherage and Cookson 2012). Not much is known about how this is achieved in Gram-positive bacteria, but in Gram-negative bacteria the release of vesicles is thought to be promoted by temporal disruptions of the interaction between the outer membrane and the peptidoglycan (Deatherage and Cookson 2012). In archaea on the other hand, in Sulfolobales the release of MVs appears to occur by membrane scission events that are facilitated by ESCRT-III proteins (Ellen et al. 2009). In other archaeal species that do not contain ESCRT-III homologues, such as Thermococcus, the mode of vesicle formation is unknown. For recent reviews about MV formation in bacteria and archaea, see Deatherage and Cookson (2012), Kulp and Kuehn (2010) and Mashburn-Warren and Whiteley (2006).

Several bacteria were shown to produce DNA-containing vesicles (Dorward et al. 1989; Kahn et al. 1982; Kolling and Matthews 1999; Pérez-Cruz et al. 2013; Renelli et al. 2004; Rumbo et al. 2011; Yaron et al. 2000). However, no MV release has been observed for hyperthermophilic bacteria. Hyperthermophilic archaea on the other hand do generate MVs. Sulfolobus species were shown to produce MVs harboring toxins killing other Sulfolobus strains (sulfolobicins) (Ellen et al. 2009; Prangishvili et al. 2000). Moreover, Ignicoccus species produce many periplasmic MVs (Rachel et al. 2002). Euryarchaeota from the order of Thermococcales commonly release MVs that were previously thought to be viruses. Some of these MVs were shown to be associated with DNA that is highly resistant to DNAse treatment and high temperatures (Soler et al. 2008). MVs from Thermococcus nautilus contain the endogenous plasmid pTN1 (Soler et al. 2011). If a derivative from this plasmid, shuttle vector pLC70 (Santangelo et al. 2008), is transformed to T. kodakarensis, the cells start to release MVs containing the same plasmid. These MVs can be subsequently used to again transfer pLC70 into plasmid-free cells. As the transfer of DNA was insensitive to DNAse treatment, DNA must have been present inside the MVs. MVs might therefore function in both the protection as well as the transfer of DNA (Gaudin et al. 2012). Comparative genomics showed that HGT has taken place between Thermococcus and Thermotoga. Moreover, it is thought that certain plasmids have been horizontally transferred between the Thermococcales and Methanococcales (Krupovic et al. 2013). Future studies need to determine whether or not MVs can also be transferred efficiently between different species. This would imply an important role of MVs in the above-described HGT events (Marguet et al. 2013).

An interesting structure, first described for the Gram-negative soil bacterium Delftia sp. Cs1, is the so-called nanopod (Shetty et al. 2011). Nanopods are prokaryotic organelles, used for projecting MVs several micrometers away from the cell. Very recently, both T. gammatolerans and T. kodakarensis were shown to have tubular structures with a row of internal vesicles resembling bacterial nanopods (Marguet et al. 2013). Similar to the nanopods from Delftia sp. Cs1, the nanopods from Thermococcus species might also project MVs and could thereby increase the distance and perhaps also efficiency of MV transfer between cells (Marguet et al. 2013). Next to nanopods, also nanotubes can be found in Thermococcales (Marguet et al. 2013). Nanotubes are intra- or inter-species tubular cytoplasmic bridges between neighboring cells and were first described for B. subtilis, S. aureus and E. coli. As the structures are between 30 and 100 nm wide, all sorts of molecules including DNA can be transferred between connected cells (Dubey and Ben-Yehuda 2011). Very recently, nanotubes with a diameter of 60–80 nm have been observed in the hyperthermophilic Thermococcus sp. 5-4 (Marguet et al. 2013) (Fig. 1b). Given their similar appearance, these archaeal structures might also be involved in exchange of biomolecules, as is observed for bacterial nanotubes. In that way they might facilitate DNA transfer between cells from similar or different species. As nanotubes now have been observed in a range of bacteria and archaea, they appear to be more widespread than previously thought. Other tubular structures observed in hyperthermophilic archaea are the cannulae found between Pyrodictium cells (König et al. 1988). Cryo-electron tomography, however, revealed that the cannulae enter the periplasmic space, but not the cytoplasm (Nickell et al. 2003). It is therefore unlikely that they transfer DNA.

UV-inducible pili of Sulfolobales (Ups)

A further DNA transfer mechanism that involves type IV pili is a UV-induced system in hyperthermophilic Sulfolobales (Schmidt et al. 1999). The type IV pili that are involved in this DNA exchange are the so-called Ups-pili (UV-inducible pili of Sulfolobales) (Fröls et al. 2008). These pili were shown to mediate UV-induced cellular aggregation during which chromosomal marker exchange could be observed. With marker recombination frequencies of up to 10−2 the efficiency of DNA transfer is very high (Ajon et al. 2011). Next to UV stress, the DNA damaging agent bleomycin could also induce cellular aggregation. Thus, the trigger for pili formation and subsequent cellular aggregation is therefore probably the formation of DNA DSBs (Fröls et al. 2008). Since Ups-knockout strains showed decreased survival rates upon UV treatment (Ajon et al. 2011), the Ups-pili are proposed to be part of a unique HR-based “community” DNA repair mechanism (Fröls et al. 2009). Correspondingly, the formation of cellular aggregates only takes place between cells from the same Sulfolobus species ensuring species-specific DNA exchange essential for DNA repair (Ajon et al. 2011). Recent data suggest that the intraspecies recognition of Sulfolobus species is determined by S-layer glycosylation patterns and the Ups-pilin subunits (van Wolferen and Albers, unpublished). Additional studies on the molecular basis of this self-recognition will give more insights in how intraspecies communication is achieved and what are the barriers that drive speciation (Cadillo-Quiroz et al. 2012).

The Ups-gene cluster is conserved among all Sulfolobales and encodes five proteins: UpsX, a hypothetical protein; UpsE, a secretion ATPase; UpsF, an integral membrane protein; and UpsA/B, two putative pilin subunits containing class III pre-pilin signal peptides (Fig. 2b). Directly downstream to the ups gene cluster, three genes are present encoding predicted DNA processing proteins. These are: an endonuclease III homologue, a ParB-like protein and an ATP-dependent helicase. Deletion of any of these genes in S. acidocaldarius results in lower survival rates upon UV treatment (van Wolferen and Albers, unpublished), we therefore speculate that they are involved in either in the DNA uptake or the HR-mediated DNA repair pathway that is linked to the Ups-system. A role in HR seems more likely as the helicase ortholog from S. solfataricus (Sso0112) was shown to catalyze the processing of Holliday junctions (Valenti et al. 2012).

Because type IV pili are often involved in the competence systems, one could speculate that uptake of DNA from lysed cells takes place rather than direct DNA exchange. However, the exchange of DNA was shown to be insensitive to DNAse treatment. Moreover, efforts to obtain recombinants by mixing cells with supernatant, lysed cells or marker DNA failed (Grogan 1996). This demonstrates that exchange of DNA indeed occurs directly from one living cell to another. Nevertheless, the mode of DNA transfer and the role of the Ups-pili in this process are still unclear. One could imagine that the Ups-pili are only involved in recognizing other cells from the same species and the initiation of a physical interaction. DNA transfer could subsequently occur via a yet unknown mechanism. Conjugation-like DNA transfer could for instance occur or cell fusion events might take place such as those observed during pNOB8 transfer (Schleper et al. 1995). However, when plating Ups-gene inactivation mutants directly on top of each other, which theoretically would also bring cells in a close proximity, no recombinants can be obtained (Ajon et al. 2011). It can therefore be speculated that the Ups-pili fulfill an active role in DNA exchange. The pili might function in a similar manner as type IV pili involved in natural transformation. In competent bacteria, however, retraction of the pili is almost always essential and in Sulfolobus species so far no retraction ATPase could be found. Furthermore, an unsolved puzzle is the directionality of the DNA transfer in Sulfolobus species. All cells appear to be serving as donor and as recipient (Grogan 1996) which makes sense as all cells are genotypically similar. Following the DNA damage hypothesis, it would be most logical for a cell with damaged DNA to acquire DNA from another cell, possibly functioning as a template for HR. As it seems that there is little to no specificity in the chromosomal DNA that can be transferred between cells, it is hard to use chromosomal markers to study directionality in more detail.

Besides a role in cellular aggregation, Ups-pili were also shown to be involved in biofilm formation (Koerdt et al. 2010). Moreover, Ups-pili may play a role in HGT. Even though Sulfolobus species appear to aggregate only species specifically, in nature rare intra-species DNA exchange might have occurred, thereby promoting HGT. Overall, we can say that Ups-mediated DNA transfer has proven to be a very interesting and unique mechanism of the Sulfolobales. So far, evidences in favor of a role in DNA repair are strong. Future studies might improve our understanding about intraspecies recognition, the mode of DNA transfer and confirm the function of DNA repair.

Summary and concluding remarks

Transfer of DNA has been shown to occur in all domains of life although this process can take place by means of different mechanisms. The widespread occurrence stresses the universal significance of DNA transfer for life on earth. Here, we focused on why and how DNA is transferred among hyperthermophilic prokaryotes. DNA transfer at high temperatures may serve an important role in evolution. HGT has proven to be a powerful driving force for prokaryotic adaptation to changing environments such as rising temperatures. Moreover, as DNA degradation is increased at higher temperatures, specialized DNA repair mechanisms involving DNA uptake may play an important role in cell survival. A third but more controversial role of DNA uptake is the use of nucleotides as a nutrient source.

Numerous DNA transfer mechanisms have been described for hyperthermophilic microorganisms. Some are based on the direct contact of neighboring cells, such as conjugation and the production of nanotubes or UV-induced pili, others involve the direct uptake of DNA from the environment (natural transformation). Lastly, vehicles such as viruses or vesicles might function in DNA transfer. One could speculate that especially at higher temperatures, DNA transfer methods that involve direct cellular contact or DNA-protecting vehicles are favored. In that way, unprotected DNA is not exposed directly to the DNA damaging surroundings. Nevertheless, uptake of free DNA has been described for several (hyper)thermophilic microorganisms, including the Thermus thermophilus competence system. The latter mechanism is the most efficient competence system found thus far in nature, which might relate to its proposed role in the adaptation to the high temperatures. In this respect, a high uptake rate also ensures limited exposure of free DNA to the surroundings.

Future research will undoubtedly expand the array of DNA transfer mechanisms among hyperthermophiles. Though more importantly, observed transfer processes need to be unravelled mechanistically, as their exact functioning is often still far from well understood. Interesting examples include conjugation and natural transformation among hyperthermophilic archaea which are both thought to differ extremely from their bacterial counterparts.

As genetic manipulation of hyperthermophilic bacteria and archaea has shown to be very challenging, further studies on their DNA transfer methods may advance the development of such tools. For example, the natural competence of T. thermophilus (Koyama et al. 1986), T. kodakarensis (Santangelo et al. 2008, 2010; Sato et al. 2003, 2005) and P. furiosis (Basen et al. 2012; Lipscomb et al. 2011) enabled the development of genetic tools. Moreover, the SSV1-based shuttle vector pMJ05 enabled reporter gene studies and the expression of proteins in S. solfataricus (Jonuscheit et al. 2003). Future studies will therefore not only increase our knowledge about the extremely diverse DNA transfer mechanisms functioning at high temperatures, but will also promote the study of other aspects in hyperthermophiles using these genetic systems.