Emerging views of genome organization in Archaea

ABSTRACT

Over the past decade, advances in methodologies for the determination of chromosome conformation have provided remarkable insight into the local and higher-order organization of bacterial and eukaryotic chromosomes. Locally folded domains are found in both bacterial and eukaryotic genomes, although they vary in size. Importantly, genomes of metazoans also possess higher-order organization into A- and B-type compartments, regions of transcriptionally active and inactive chromatin, respectively. Until recently, nothing was known about the organization of genomes of organisms in the third domain of life – the archaea. However, despite archaea possessing simple circular genomes that are morphologically reminiscent of those seen in many bacteria, a recent study of archaea of the genus Sulfolobus has revealed that it organizes its genome into large-scale domains. These domains further interact to form defined A- and B-type compartments. The interplay of transcription and localization of a novel structural maintenance of chromosomes (SMC) superfamily protein, termed coalescin, defines compartment identity. In this Review, we discuss the mechanistic and evolutionary implications of these findings.

Summary: A review of recent work on archaea of the genus Sulfolobus that has revealed intriguing parallels between genome organization in these prokaryotes and in metazoan organisms.

Introduction

Although it has been recognized for many years that archaea and eukaryotes share common core information processing machineries, such as those for DNA replication and gene transcription, very little is known about the subcellular organization of archaea. In part, this limited knowledge is due to the relatively low number of cultivated archaeal species, their small size and often extreme growth conditions. Phylogenetic studies of many archaeal species have generated a diverse range of viewpoints about the finer structure of the archaeal branch of the tree of life (Da Cunha et al., 2018; Dombrowski et al., 2019; Spang et al., 2017). There is even debate about whether eukaryotes arose within the archaeal domain or whether archaea and eukaryotes are sister lineages (Rivera and Lake, 2004; Williams et al., 2020, 2013; Woese and Fox, 1977; Zhou et al., 2018). Studies of archaeal chromosome biology have principally focused on members of the euryarchaeal and crenarchaeal lineages and have revealed a striking disparity in cell cycle organization, chromosome ploidy and cell division mechanisms. Euryarchaea (including methanogens, which produce methane in hypoxic environments, and halophilic archaea, which grow in high-salt conditions) typically possess multiple copies of their genomes, a state referred to as oligoploidy, with up to 55 copies being reported for Methanococcus maripaludis (Hildenbrand et al., 2011). Interestingly, studies in Haloferax volcanii and Thermococcus kodakarensis have revealed that the replication of their genomes is not dependent on active DNA replication origins and can be affected by recombination-based processes. Presumably, this is an effective strategy when tens of copies of the genome are present (Gehring et al., 2017; Hawkins et al., 2013). This rather promiscuous replication mode is mirrored by a lack of overt partitioning of the cell cycle into distinct phases. It appears that these euryarchaeal organisms can undergo overlapping cycles of genome duplication and cell division, perhaps in a manner akin to Escherichia coli rapidly growing in rich medium. In contrast, members of the crenarchaea possess cell cycles that contain defined G1 and G2 phases after cell division and DNA replication, respectively. Further, crenarchaea typically contain a single copy of the chromosome in G1 and two copies in G2, following DNA replication (Lundgren et al., 2008; Samson and Bell, 2011). Fluorescence in situ hybridization studies in Sulfolobus (the most intensively studied genus of crenarchaea) demonstrated that loci encompassing the DNA replication origins remained cohesed for the majority of the G2 phase, following completion of DNA replication – a stark contrast with the situation in bacteria, in which rapid segregation of loci follows their replication (Reyes-Lamothe and Sherratt, 2019; Robinson et al., 2007). The observed cohesion correlated with junctions between double-stranded (ds)DNA molecules, in the form of post-replicative hemicatenane structures, at all tested origin and non-origin loci around the chromosome (Robinson et al., 2007). It is not known whether the phenomenon of cohesion extends to other members of the archaea, but it is notable that 2D neutral-neutral DNA electrophoresis analyses, a technique that resolves structured DNA molecules, performed on the crenarchaeon Aeropyrum pernix also demonstrated the existence of DNA junctions between daughter strands following replication (Robinson and Bell, 2007). In contrast, analogous studies of the euryarchaeal species Methanothermobacter thermautotrophicus and Pyrococcus abyssi did not detect such junctional molecules (Majerník and Chong, 2008; Matsunaga et al., 2001). It is possible, therefore, that the presence or absence of post-replicative sister chromatid cohesion is another aspect of the cell cycle that differentiates euryarchaea and crenarchaea.

DNA compaction systems are also very variable across the archaeal domain of life (for reviews of archaeal chromatin proteins, see Sanders et al., 2019; Visone et al., 2014; White and Bell, 2002). The majority of archaeal lineages, with the exception of crenarchaea, encode orthologs of histones H3 and H4 of eukaryotes. Crenarchaea possess a number of small basic proteins, including the phylogenetically broadly conserved Cren7 and Alba superfamilies, that have been proposed to play roles in genome compaction (Driessen et al., 2013). However, the genomic distribution and physiological roles of these proteins remain very poorly understood. At the topological level, the genomes of hyperthermophilic archaea (which grow optimally in temperatures in excess of 80°C) are characterized by the presence of positive supercoiling (over-winding of DNA molecules), presumably to stabilize the duplex DNA at elevated temperatures. This property is generally believed to be conferred by the activity of reverse gyrase, an enzyme that is universally conserved in hyperthermophilic archaea and bacteria (Bouthier de la Tour et al., 1990). However, the distribution of supercoiling density across a given genome is largely unknown. Notably, studies of plasmids in crenarchaea have revealed that topology is modulated during temperature stress – more specifically, cold shock results in the introduction of negative supercoils (under-winding) in Sulfolobus plasmids. This presumably facilitates gene transcription at reduced temperatures. In contrast, heat shock led to enhanced positive supercoiling of plasmids (Lopez-Garcia and Forterre, 1997, 1999). In the following sections, we describe the basic organizational principles of bacterial and eukaryotic chromosomes and compare these to the conclusions from recent work revealing the structure of archaeal chromosomes from species in the genus Sulfolobus. Additionally, we will discuss the nature of chromosome-organizing protein complexes in the three domains of life. Finally, we will speculate on the possible selective forces that may have driven the evolution of compartmentalized chromosome conformations.

Higher-order structures of archaeal genomes

In addition to local DNA folding, higher-order structures of the genome play critical roles in genomic maintenance and function. Our knowledge about the spatial organization of genomes has been increasing for decades, and this advance has been driven further since the development of Hi-C, a combination of chromosome conformation capture (known as 3C) and deep sequencing techniques (Lieberman-Aiden et al., 2009).

Hi-C and its derivative technologies have revealed that eukaryotic genomes are structurally organized according to two major principles. First, eukaryotic chromosomes fold into arrays of topologically associating domains (TADs, also called contact domains), which are characterized by higher frequencies of genomic contacts within them than between neighboring TADs (Dixon et al., 2012; Nora et al., 2012; Rao et al., 2014; Sexton et al., 2012). TADs therefore appear as squares of high contact frequencies along the main diagonal in the contact matrix (Fig. 1). Second, genomes of higher eukaryotes are organized into A-type and B-type compartments, genomic regions that are enriched for transcriptionally active chromatin and inactive chromatin, respectively (Lieberman-Aiden et al., 2009; Rowley and Corces, 2018). This organization is manifested as a plaid pattern of interactions in the contact map. As a corollary, DNA regions segmented by the plaid pattern show a self-interacting nature in a similar way that TADs do (Fig. 1). Although there has been debate regarding how to term self-interacting domains arising from different mechanisms and defined by different analytic methodologies (Rowley and Corces, 2018), hereafter we will use the term TADs to refer to eukaryotic self-interacting domains other than those formed by A- and B-type compartments. More generally, TADs are local structures that form in cis, while compartments arise from longer range interactions that can be in cis or trans.

Fig. 1.

Overview of chromosome organization in the three domains of life. Eukaryotic chromosomes are organized into two types of self-interacting domains; TADs and larger compartmental domains. Transcriptionally active and inactive compartmental domains segregate from each other to form the A compartment and the B compartment, respectively. Bacterial chromosomes are folded into self-interacting domains called CIDs but lack A- or B-type compartment-like structure. Another feature of many bacterial chromosomes is that the two chromosomal arms that connect the origin and terminus of replication are aligned in parallel. This so-called ori-ter configuration is manifested as the secondary diagonal perpendicular to the main diagonal in the Hi-C heatmap. Chromosomes of the archaea Sulfolobus form compartmental domains like eukaryotic chromosomes but apparently do not possess TAD- or CID-like structure. Thus, to the best of our knowledge, the two organizational principles (formation of TADs, CIDs and of A and B compartments) have different phylogenetic distributions.

Bacterial genomes fold into TAD-like domains called chromosomal interaction domains (CIDs), but, to date, A- and B-type compartment-like structures have not been found in bacteria (Le et al., 2013; Lioy et al., 2018; Marbouty et al., 2015; Trussart et al., 2017; Val et al., 2016; Wang et al., 2015). Bacteria exhibit additional layers of genome organization depending on the species. The most commonly-described conformation of the bacterial genome is a so-called ori-ter configuration. In this organization, the single origin of replication and the replication terminus localize at different cell poles, with the two chromosomal arms between them aligned in parallel along the long cell axis (Badrinarayanan et al., 2015; Viollier et al., 2004). This arrangement is manifested as the secondary diagonal perpendicular to the main diagonal in the contact matrix (Fig. 1) (Le et al., 2013). Curiously, and in agreement with previous microscopy observations, similar inter-arm interaction was not observed in a 3C-seq study of E. coli (Lioy et al., 2018; Wang et al., 2006). The E. coli genome is also unique in that it forms megabase-sized domains called macrodomains in addition to smaller CIDs (Lioy et al., 2018; Niki et al., 2000; Valens et al., 2004).

A number of studies have described gross morphologies of archaeal genomes by observing stained nucleoids in fixed cells. They have suggested that archaeal genomes are structurally organized in a non-random fashion (Lundgren et al., 2008; Malandrin et al., 1999; Poplawski and Bernander, 1997). However, further delineation of archaeal genome conformation was hampered for a long time. This is partly because most of the commonly studied archaea thrive under extreme environments and are therefore hard to study by in vivo techniques that have been successful in determining genome organization in eukaryotes and bacteria (e.g. live-cell imaging of loci using fluorescent proteins). Moreover, many of the experimentally available archaea contain tens of copies of the genome per cell (Soppa, 2011), and this polyploid nature makes it difficult to assess the structure of each chromosome experimentally.

We have recently succeeded in establishing a Hi-C protocol for members of the monoploid thermophilic archaea Sulfolobus (Takemata et al., 2019). Unexpectedly, the contact matrices we obtained showed plaid patterns, revealing that Sulfolobus genomes are organized into two compartments (Fig. 1). As is the case with eukaryotic A- and B-type compartments, the two compartments in Sulfolobus are characterized by different levels of gene expression. We therefore labeled the active compartment in Sulfolobus A-type and the less active compartment B-type. We did not observe domain-like structures other than those that contribute to the compartmental organization.

Molecular players sculpting the higher-order organization of archaeal genomes

In both prokaryotes and eukaryotes, members of the structural maintenance of chromosomes (SMC) protein family are crucial for higher-order genome organization (Nolivos and Sherratt, 2014; Uhlmann, 2016). SMC complexes form ring-shaped structures in which two SMC ATPase subunits are joined at their hinge domains (Fig. 2). Using its ring structure and ATPase activity, the SMC complex, in conjunction with accessory subunits, is believed to encircle two segments on the same DNA strand and then progressively extrude a DNA loop (Alipour and Marko, 2012; Hassler et al., 2018; Nasmyth, 2001). This loop extrusion model explains how the SMC complex cohesin promotes the formation of TADs (Sanborn et al., 2015). It is also believed that another SMC complex, condensin, extrudes DNA loops to shape condensed chromosomes. This is a prerequisite for faithful chromosome segregation and accomplishes the bacterial ori-ter configuration (Gibcus et al., 2018; Tran et al., 2017; Wang et al., 2017). Condensin is broadly conserved in bacteria and eukaryotes, and the majority of archaeal genomes encode clear homologs of condensin subunits. Indeed, structural studies of the euryarchaeal Pyrococcus condensin SMC subunit were instrumental in building a structural model of Bacillus subtilis condensin (Diebold-Durand et al., 2017).

Fig. 2.

Schematic representations of SMC complexes. (A) SMC monomers form a rod-like structure with a dimerization domain on one end, an ATPase head domain on the other and a coiled-coil arm between them. In canonical SMC complexes (cohesin, condensin, etc.), SMC proteins dimerize via their hinge domains and associate with accessory subunits to form a ring-shaped structure. (B) Rad50, which possesses a zinc hook instead of a hinge domain, associates with the nuclease protein Mre11.

With the exception of the aforementioned structural studies, the molecular functions of archaeal SMC proteins are poorly understood. An early study showed that disruption of a condensin SMC homolog causes aberrant chromosome segregation and cell morphology in the euryarchaeon Methanococcus voltae, suggesting a conserved role of condensin in structuring archaeal genomes for cell division (Long and Faguy, 2004). Another study characterized an SMC-like protein named Sph1 in the halophilic archaeon Halobacterium salinarum (Herrmann and Soppa, 2002). The cellular concentration of Sph1 is elevated when chromosome segregation, which is concomitant with DNA replication in this species, is nearly complete, leading the authors to propose that Sph1 plays a role in a late step of DNA replication. Members of the archaeal order Thermoproteales encode another SMC-like protein, arcadin-4 (Ettema et al., 2011). The arcadin-4-encoding gene is located in a gene cluster, together with a gene encoding the actin-like protein crenactin. Crenactin has been implicated in cell division of the Thermoproteales, leading to the proposal that arcadin-4 could also be involved in genome segregation prior to or concomitant with cell division (Ettema et al., 2011). However, it remains unknown whether these SMC or SMC-like proteins play roles in orchestrating archaeal genome conformation.

Another mystery regarding the archaeal SMC proteins is that condensin is widely conserved across the three domains of life with the notable exception of the Crenarchaeota, the archaeal phylum that contains Sulfolobus (Hirano, 2016; Kamada and Barilla, 2018). We have recently shown that Sulfolobus species instead use a hitherto uncharacterized SMC-like protein, called coalescin, to structure their genomes (Takemata et al., 2019). Coalescin preferentially localizes in the B compartment because coalescin binding is blocked at actively transcribed regions. Coalescin is found to be essential for viability, and its overexpression enhances interactions between coalescin-bound regions (Takemata et al., 2019). Taken together, these results demonstrate a condensin-independent mechanism of genome organization in Sulfolobus, in which coalescin binds to transcriptionally less active genes and promotes their coalescence to facilitate the compartmental organization of Sulfolobus genomes.

In sharp contrast to coalescin-mediated compartmentalization, loop extrusion by SMC complexes is a phenomenon distinct from formation of compartments. More specifically, while local domains are generated by loop extrusion, compartment identity has been proposed to be driven by local chromatin states, perhaps manifested by localized microphase separation (Nuebler et al., 2018; Rao et al., 2017). It is conceivable that coalescin depends at least partly on a different mechanism to accomplish genome compartmentalization. This unique function may be related to the fact that coalescin does not have a hinge domain, but instead harbors a conserved Cys-X-X-Cys motif at its center (Takemata et al., 2019). In the SMC-like protein Rad50, this motif forms a zinc hook that has been proposed to mediate inter-complex interactions of Rad50 dimers (Table 1; Fig. 2). The inter-complex association of Rad50 has been suggested to tether distant DNA segments for DNA repair (Hopfner et al., 2002). By analogy, coalescin molecules on distant loci may form a dimer or dimer of dimer complexes to bring the loci into proximity. However, it is also possible that coalescin serves as a DNA crosslinker independently of the zinc hook, as has been proposed for SMC complexes without a zinc hook (Cheng et al., 2015; Vickridge et al., 2017). Indeed, it is notable that the Gruber lab successfully replaced the globular hinge domain of a bacterial condensin with a zinc hook domain derived from an archaeal Rad50 with no discernible impact on the growth of Bacillus subtilis (Burmann et al., 2017). Further s­tructural and biochemical characterization of coalescin will be essential to elucidate its molecular function.

Table 1.

Comparison of the size of SMC subunits, nature of hinge domain and identity of partner proteins (where known) for coalescin, condensin and RAD50

Potential advantages of folding archaeal genomes into compartments

What are the biological consequences of folding the Sulfolobus genome into A- and B-type compartments? It is conceivable that the compartments are involved in the coordinated control of genes with different activities. The A compartment is enriched for genes involved in core metabolic processes (particularly those involved in protein biogenesis) and for essential genes (Fig. 3), suggesting that these housekeeping genes need to be highly expressed. By contrast, the B compartment is enriched for mobile genetic elements (Takemata et al., 2019) and seemingly non-essential or conditionally required genes functioning in diverse metabolic pathways (Fig. 3). Thus, most B-compartment genes need to be repressed when the growth conditions are optimal and constant, and their expression will be induced in response to certain environmental stresses. Under unstressed conditions, the physical and functional compartmentalization of loci may allow Sulfolobus cells to concentrate the transcriptional apparatus at housekeeping genes in the A compartment. This concentrating mechanism is analogous to the transcription factory model or the phase-separation model for eukaryotic transcriptional control (Hnisz et al., 2017; Papantonis and Cook, 2011). Meanwhile, the same mechanism will restrict the access of the transcriptional machinery to stress-responsive genes in the B compartment. Consistent with this idea, overexpression of coalescin not only enhances genome compartmentalization but also further represses B-compartment genes (Takemata et al., 2019). Additionally, Sulfolobus cells may decrease the extent of genome compartmentalization in response to transcriptional cues, so that transcriptional components in the A compartment will gain more access to B-compartment genes. Indeed, a weakening of compartment strength and a concomitant relative induction of B-compartment genes have been observed upon entry into stationary phase (Takemata et al., 2019). Thus, a finely balanced interplay of transcription and coalescin occupancy appears to be crucial for establishing the higher order architecture of the Sulfolobus chromosome.

Fig. 3.

Genomic features of Sulfolobus A- and B-type compartments. The Sulfolobus A compartment is enriched for housekeeping genes that need to be expressed constitutively, whereas the B compartment is enriched for conditionally-required genes. (A,B) Enrichment of gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways was analyzed using DAVID 6.8 for the A and B compartments in Sulfolobus acidocaldarius (Huang da et al., 2009a,b; Takemata et al., 2019). For the GO analysis, only non-redundant GO terms (GOTERM_BP_FAT, GOTERM_CC_FAT and GOTERM_MF_FAT) were considered. Shown are GO terms and KEGG pathways with a Benjamini–Hochberg adjusted P-value of <0.05. (C) The number of essential genes per Mb was calculated for the A and B compartments in S. acidocaldarius. Homologs of the essential genes in Sulfolobus islandicus M.16.4 (Zhang et al., 2018) were considered as essential in S. acidocaldarius.

Conclusions and perspectives

Current studies have demonstrated that self-interacting domains and A- and B-type compartments are widespread structural features of genomes in different lineages. However, these studies also pose possible evolutionary gaps in the prevalence of these structures. First, TADs and their bacterial counterpart CIDs are found in a variety of organisms, but Hi-C analysis of Sulfolobus archaea did not detect self-interacting domains other than those formed by A and B compartments (Le et al., 2013; Rowley and Corces, 2018; Takemata et al., 2019; Wang et al., 2015). The lack of TAD- or CID-like domains in Sulfolobus may be due to the absence of cohesin or to certain thermophile-specific features (e.g. positively supercoiled DNA). Alternatively, the apparent lack of self-interacting domains may result from the relatively low resolution (15–30 kb) of Hi-C experiments.

Second, the existence of A and B compartments has been documented for Sulfolobus archaea and higher eukaryotes but not for lower eukaryotes including Saccharomyces cerevisiae, Schizosaccharomyces pombe, Neurospora crassa, Plasmodium falciparum and Trypanosoma brucei (Ay et al., 2014; Duan et al., 2010; Galazka et al., 2016; Mizuguchi et al., 2014; Müller et al., 2018; Takemata et al., 2019). Instead, it is reported that heterochromatic regions in the N. crassa genome (centromeres, subtelomeres and smaller regions of interspersed heterochromatin) form a network of strong genomic contacts (Galazka et al., 2016). P. falciparum and T. brucei are also reported to form gene clusters containing transcriptionally inactive antigen variation genes (Ay et al., 2014; Müller et al., 2018). We speculate that the spatial gene clusters in these three species could be structural counterparts of the B compartment, whereas other genes may form A compartment-like structures. A and B compartments or an analogous compartmental organization may be more common in eukaryotes than currently thought.

Finally, how could we interpret the lack of A and B compartments in bacteria? Bacteria have evolved to segregate sister chromatids concomitantly with DNA replication (Badrinarayanan et al., 2015). We surmise that this mechanism allows for rapid proliferation but at the same time imposes strong conformational constraints on bacterial genomes. For example, in B. subtilis, condensin is recruited to duplicated origins and moves towards the terminus to extrude intrachromosomal DNA loops, leading to efficient sister chromatid resolution concurrent with replication (Gruber and Errington, 2009; Sullivan et al., 2009; Wang et al., 2017). It is hard to imagine that the resultant extensive alignment of the two chromosome arms are compatible with genome-wide segregation of active and inactive genes. In contrast, chromosome replication and segregation occur at temporally distinct stages in eukaryotes and crenarchaeal species including Sulfolobus (Lindas and Bernander, 2013). Furthermore, sister chromatids in Sulfolobus remain cohesed for the majority of the G2 period of the cell cycle (Robinson et al., 2007). We previously proposed that the acquisition of a structured cell cycle that separates replication from segregation may have loosened the constraints on genome conformation and thereby may have driven the evolution of compartmental genome organization in eukaryotes and Sulfolobus (Takemata et al., 2019).

As discussed above, studies of eukaryotes, bacteria and archaea have provided insights into common and lineage-specific rules of higher-order genome organization. However, in closing, it should be emphasized that our knowledge of genome architecture has been obtained from a limited number of organisms, especially regarding archaea. Caution must be urged for inter-species and inter-phylum generalizations.