Multi-scale architecture of archaeal chromosomes

Naomichi Takemata; Stephen D Bell

doi:10.1016/j.molcel.2020.12.001

. Author manuscript; available in PMC: 2022 Feb 4.

Published in final edited form as: Mol Cell. 2020 Dec 30;81(3):473–487.e6. doi: 10.1016/j.molcel.2020.12.001

Multi-scale architecture of archaeal chromosomes

Naomichi Takemata ^1,^2,^3,⁴, Stephen D Bell ^1,²

PMCID: PMC7867652 NIHMSID: NIHMS1654324 PMID: 33382983

SUMMARY

Chromosome conformation capture (3C) technologies have identified topologically associating domains (TADs), and larger A/B compartments as two salient structural features of eukaryotic chromosomes. These structures are sculpted by the combined actions of transcription and structural maintenance of chromosomes (SMC) superfamily proteins. Bacterial chromosomes fold into TAD-like chromosomal interaction domains (CIDs) but do not display A/B compartment-type organization. We reveal that chromosomes of Sulfolobus archaea are organized into CID-like topological domains in addition to previously described larger A/B compartment-type structures. We uncover local rules governing the identity of the topological domains and their boundaries. We also identify long-range loop structures and provide evidence of a hub-like structure that colocalizes genes involved in ribosome biogenesis. In addition to providing high resolution descriptions of archaeal chromosome architectures, our data provide evidence for multiple modes of organization in prokaryotic chromosomes and yield insight into the evolution of eukaryotic chromosome conformation.

Graphical Abstract

graphic file with name nihms-1654324-f0001.jpg

eToc

Takemata and Bell reveal that chromosomes of archaea of the genus Sulfolobus organize into A and B compartments and, independently, into local domains akin to bacterial CIDs. Additionally, the genomes possess loop structures that can join loci separated by up to half the length of the circular chromosomes.

Introduction

The spatial organization of chromosomes plays critical roles in various DNA-related activities. Chromosome conformation capture (3C) combined with deep sequencing (Hi-C) has revealed two organizational principles of eukaryotic chromosomes. First, linear eukaryotic chromosomes fold into arrays of self-interacting domains called TADs (also called “contact domains”)(Dixon et al., 2012; Nora et al., 2012; Rao et al., 2014; Sexton et al., 2012). TADs are tens to hundreds of kilobases in size and are often nested by finer-scale subdomains (Phillips-Cremins et al., 2013). An emerging model is that TADs are formed via progressive extrusion of DNA loops by the SMC complex cohesin (Alipour and Marko, 2012; Davidson et al., 2019; Kim et al., 2019). Boundaries of TADs and subdomains are specified by roadblocks for cohesin translocation or active transcription (Sanborn et al., 2015; van Steensel and Furlong, 2019). Metazoan chromosomes show an additional and independent layer of organization, in which transcriptionally active and inactive chromatin segments at the megabase scale segregate into distinct spatial compartments (A-type and B-type respectively)(Lieberman-Aiden et al., 2009; Nora et al., 2017; Schwarzer et al., 2017). Transcription and SMC proteins are dispensable for, but can affect, this compartmentalization (Gibcus et al., 2018; Rao et al., 2017; Rowley et al., 2017). Bacterial chromosomes are generally circular and fold into TAD-like structures called chromosomal interaction domains (CIDs), the boundaries of which are formed by highly transcribed genes (Le et al., 2013). Although additional higher-order folding has been found in some bacteria (Le et al., 2013; Lioy et al., 2018; Marbouty et al., 2015; Wang et al., 2017), to date A/B compartment-type organization has not been described in this domain of life.

There is growing evidence that the Eukarya and the prokaryotic domain Archaea represent sister groups in the tree of life (Imachi et al., 2020; Spang et al., 2017). For a long time, essentially nothing was known about higher-order chromosome organization in archaea (Takemata and Bell, 2020). We have recently shown that chromosomes in members of the hyperthermophilic crenarchaeal genus Sulfolobus are organized into A/B compartment-type structures (Takemata et al., 2019). This folding involves the Sulfolobus-specific SMC protein that we named coalescin (ClsN). Curiously, we were unable to detect TAD- or CID-like structures except for the large-scale domains that accrete to form compartments. In the current work we describe high-resolution 3C analyses of Sulfolobus chromosomes, generating contact maps at the gene and operon levels. Our data reveal, in addition to compartmentalization, two hitherto undetected levels of structure in Sulfolobus chromosomes. First, we observe CID-like domains in both the A and B compartments. In addition, we detect loop structures that can bridge over one million base pairs. Notably, genes encoding ribosomal components are preferentially enriched at loop anchor points and engage each other, suggesting spatial coordination of ribosome biogenesis.

Results

Our previous Hi-C study used the 6-base cutter HindIII to yield genomic contact maps of 15-kb resolution at maximum (Takemata et al., 2019). We were concerned that this relatively low resolution may have prevented us from detecting TAD- or CID-type domains. To improve resolution, in the current study we used the 4-base blunt cutter AluI to conduct 3C-seq of asynchronously growing Sulfolobus acidocaldarius and Sulfolobus islandicus. AluI recognizes their genomes at every ~230 bases on average, allowing us to generate contact maps at 2-kb resolution and Pearson correlation maps at 10-kb resolution (Figure S1). 3C-seq data were highly reproducible between replicates and congruent with our previous Hi-C data (Figure S1).

The contact maps and Pearson correlation maps from the 3C-seq experiments exhibit plaid patterns formed by regional enrichment and depletion of long-range interactions, confirming the existence of A/B compartments (Figure S1). In the following, we shall refer to the large-scale regions that interact to form compartments as compartmental segments. At 2-kb resolution, the contact maps additionally show on-diagonal arrays of triangles formed by shorter-range interactions (Figures 1A and 1B, upper panels). These triangles are smaller than most compartmental segments and thus seem to correspond to TAD- or CID-like domains. Furthermore, we could observe signals indicative of loop structures (Figure S1 and below).

Figure 1. — Two biological replicates of 3C-seq data and three biological replicates of RNA-seq data were used for each condition.

(A) (Top) 3C-seq contact map of the *S. acidocaldarius* chromosome at 2-kb resolution. Black bars indicate CID boundaries determined from directional preference of contacts within a ±40 kb window (also see Figure S2A and STAR Methods). B-compartment regions were determined at 10-kb resolution and shaded in gray. (Bottom) RNA-seq profile of *S. acidocaldarius*. Reads Per Kilobase of exon per Million mapped reads (RPKM) value is plotted for each 2-kb bin.

(B) A 3C-seq contact map and an RNA-seq profile were generated for *S. islandicus* as in Figure 1A. Filtered-out bins are indicated in gray (see STAR Methods for more detail).

(C) Average insulation scores around CID boundaries in the A compartment (red) and in the B compartment (blue).

(D) Examples of CID boundary formation at long, highly expressed operons in *S. islandicus*. (Top) 3C-seq contact maps at 2-kb resolution. (Middle) Magnified view of strand-specific RNA-seq profiles. Normalized coverage is shown for each 100-bp bin. Operons of interest are shaded in gray. (Bottom) Magnified views of directional preference plots in which contact biases toward downstream and upstream regions are indicated by positive and negative values respectively. CID boundaries are indicated by dotted lines. Black horizontal lines indicate statistical thresholds.

(E) The percentage of CID boundaries proximal (≤ 2 kb) to highly expressed genes (top 10% in the genome) was calculated for A and B compartments. Expected percentages and the statistical significance of the observed data were determined by random permutation of CID boundary positions (n = 10,000, see STAR Methods for more detail).

CID-like structures in Sulfolobus chromosomes

First, we sought to characterize further the CID-like features. We quantified directional preference of short-range interactions as performed in studies of bacterial CIDs (Le et al., 2013; Lioy et al., 2018; Marbouty et al., 2015) but with a finer scale (40 kb in this study versus 100 kb in the cited studies). Directional preference plots from pooled 3C-seq reads were used to determine CID boundaries (Figures S2A and S2B). We identified 32 and 47 domains in S. acidocaldarius and S. islandicus respectively (Figures 1A and 1B). Domains in the two species ranged in size from 14 kb to 196 kb with a mean size of 60 kb. Given the comparable size to bacterial CIDs (Le et al., 2013; Trussart et al., 2017) and apparent lack of TAD-forming factors such as cohesin and CTCF (Nora et al., 2017; Rao et al., 2017) in archaea, and in light of the data presented below, we will refer to the identified domains as CIDs. Additionally, we quantified a metric termed insulation score, in which the sum of interactions occurring across intervals is defined by a square window sliding along the principle diagonal of the contact matrix (Crane et al., 2015). Minima in insulation scores, therefore, correlate with domain boundaries. Plotting averaged insulation scores across the CID boundaries detected in the directional preference analyses, we observe that CIDs within the A compartment have more highly-defined boundaries than those in the B compartment (Figure 1C).

For many bacterial CIDs and certain kinds of topological domains in eukaryotes, boundaries are formed at highly transcribed genes (Le et al., 2013; Lioy et al., 2018; van Steensel and Furlong, 2019). Additionally, in bacteria, these boundary-associated genes tend to comprise long operons (Le and Laub, 2016). In line with these findings, our RNA-seq analyses detected noticeably high levels of transcripts derived from boundary-proximal genes in S. acidocaldarius and S. islandicus (Figures 1A and 1B). For example, particularly well-defined CID boundaries were found at the highly transcribed 16S-23S rRNA and ribosomal protein operons in both species (Figures 1D and S2C). We then tested, more generally, whether archaeal CID boundaries are associated with gene expression using the RNA-seq data. In the A-type compartment, where gene expression is active on average, ~80% of CID boundaries were found at highly expressed (top 10% in the genome) genes. This colocalization was seen much more frequently than expected from random shuffling of boundary positions (p ≤ 4 × 10⁻⁴) (Figures 1E). CID boundaries within the A compartment also frequently (~30%) colocalized with large (≥ 3 kb), highly expressed operons (Figure S2D). In contrast, in the B-type compartment, we did not observe consistent enrichment of highly expressed genes at CID boundaries (Figures 1E and S2D). More specifically, while significant enrichment was seen in S. acidocaldarius, it was not observed in S. islandicus. We next tested whether CID boundaries within B-type compartment segments are associated with genes whose expression levels are not among the top rank in the whole genome but which are relatively high compared to other B compartment genes. To do so, we calculated RNA levels (as adjudged by RNA-seq) within 2 kb of the CID boundaries or control loci in either A or B compartment (Figure S2E). As shown above, A compartment CID boundaries were highly significantly associated with elevated transcription. In the B-type compartment, RNA levels of boundary-proximal genes were substantially higher in S. acidocaldarius and modestly higher in S. islandicus when compared to those of boundary-distal genes (Figure S2E). However, we note that boundary-proximal genes for CIDs in the B compartment were expressed at lower levels than CID-internal genes in the A compartment.

The co-occurrence of CID boundaries in the A compartment with highly expressed genes can be explained by a mechanism of domain formation akin to that seen in bacteria (Le et al., 2013). However, loci associated with CID boundaries in the B compartment are transcriptionally much less active than those in the A compartment and therefore additional factors may be contributory. Enrichment of the Sulfolobus SMC-superfamily protein, ClsN, is another hallmark of the B-type compartment in Sulfolobus (Takemata et al., 2019). Therefore, we speculated that ClsN might play a role in establishing or maintaining B compartment CIDs. We explored this possibility by examining ChIP-seq profiles of ClsN around CID boundaries. This revealed that a noticeable number of CID boundaries reside at “ClsN cliffs,” near which ClsN occupancy increases abruptly to form peaks (Figures 2A and S3). Some CID boundaries are nestled between two ClsN cliffs (“ClsN valleys”). These patterns are observed whether or not the boundary colocalizes with highly expressed genes.

Figure 2. — Two biological replicates of ChIP-seq and 3C-seq data were used for each condition.

(A) ChIP-seq profiles of ClsN at 2-kb resolution (top panels) and plots showing distribution bias of ClsN in a 40-kb window (bottom panels, see STAR Methods for more detail). Also indicated are positions of CID boundaries, some of which are labeled according to the ClsN distribution patterns around them. B compartment loci are shaded in gray.

(B) Schematic model illustrating how the short-range attraction between ClsN-bound regions can affect the directional bias of contacts.

(C) Schematic model illustrating the relationships between ClsN distribution, ClsN bias, and the directional bias in ClsN-mediated short-range contacts.

(D) Plots showing the correlation between ClsN bias and contact bias. Before the plotting, ClsN occupancy was compared between 2-kb bins in the B compartment, and the top 50% of them were used for the analysis. Spearman rank correlation coefficients (r) and corresponding p-values are also shown. See STAR Methods for more detail.

(E) Schematic model illustrating how the magnitude (strength) of ClsN bias can affect boundary strength.

(F) Plots showing the correlation between the strength of ClsN bias (represented in Figure 2E) and boundary strength. Spearman rank correlation coefficients (r) and corresponding p-values are also shown. See STAR Methods for more detail.

How could the patterns of ClsN distribution observed at CID boundaries contribute to domain formation? Our previous study revealed that ClsN promotes interactions between its binding sites (Takemata et al., 2019). Thus, we hypothesized that ClsN-bound loci are attracted more toward a ClsN-enriched side than toward the other side that harbor less ClsN (Figure 2B). This directional bias in ClsN-mediated contacts will peak at ClsN cliffs and shift from left to right around ClsN valleys (Figure 2C), leading to the formation or reinforcement of CID boundaries at these regions. To measure the bias in ClsN distribution, we quantified the mean ClsN occupancy at 2-kb intervals in the right 40-kb region minus the mean ClsN occupancy in the left 40-kb region (“ClsN bias”). The window size was determined in accordance with the distance range used for directional preference analyses. Plotting ClsN bias confirmed that this metric peaks/bottoms around cliff-type CID boundaries and shifts from negative to positive at valley-type CID boundaries (Figures 2A and S3). To test whether the bias in ClsN distribution correlates with the direction of genomic contacts, we compared CID bias in the B compartment with the directional bias of contacts in a ±40-kb window (a metric akin to directional preference, see STAR Methods for more detail). We filtered out loci with low ClsN occupancy (smaller than the median in the B compartment) to focus on ClsN-mediated interactions. The two metrics showed highly significant correlations in both S. acidocaldarius (p < 2.2 × 10⁻¹⁶) and S. islandicus (p = 5.9 × 10⁻⁷) (Figure 2D).

We also tested whether the magnitude of ClsN bias change could be correlated with the strength of CID boundaries in the B compartment. For example, if a CID boundary is located between high peaks of ClsN, the directional bias of ClsN-mediated contacts will sharply shift from left to right between the peaks. This large transition will enhance the strength of the boundary (Figure 2E). To test this idea, we measured the magnitude of ClsN bias change (“strength of CID bias”) around each CID boundary in the B compartment. We defined this metric as the difference between the local maximum of ClsN bias on the right side and the local minimum of ClsN bias on the left side. The strength of ClsN bias was positively correlated with boundary strength (calculated from the insulation score plot) in both S. acidocaldarius and S. islandicus (Figure 2F). Taken together, these results suggest a role for ClsN in establishing or maintaining B compartment CIDs.

The impact of inhibiting transcription on Sulfolobus CIDs

From the results obtained so far, it appears that CIDs in the A-type compartment are formed by boundaries generated by highly active transcription, whereas formation of those in the B-type compartment may involve the combined effects of relatively strong transcription at boundaries and CID-internal elevated ClsN occupancy. To explore whether transcription is causally related to CID formation, we conducted 3C-seq using S. acidocaldarius cells treated with the transcription inhibitor actinomycin D. This treatment results in delocalization ClsN from the B-type compartment and diminishes chromosome compartmentalization (Figures S4A and S4B)(Takemata et al., 2019). High-resolution 3C-seq allowed us to observe a previously undetected increase in signals that spread from a number of spots on the main diagonal (Figures 3A and S4C). The corresponding loci often overlapped with or were close to genomic regions where CID boundaries were located in untreated cells (Figure 3A, indicated by black bars). The alterations in short-range interactions over the CID boundaries were also manifested as a relative increase in insulation score at boundary-proximal regions. This increase, meaning reduced insulation activities of CID boundaries, was observed for both the A-type and B-type compartments (Figures 3B and 3C). These results suggest that transcription promotes the formation of CID boundaries in both compartments.

Actinomycin D increased insulation scores of A compartments CID boundaries for all the insulation square sizes tested (≤ 150 kb)(Figure 3D). However, for B compartment CIDs, increases in insulation score were only statistically significant for short-range interactions (square size < 50 kb). The difference between the two compartments was also displayed in aggregate plots of insulation score (Figure 3B). The A compartment was characterized by a dramatic increase in the score at boundary-proximal regions. In contrast, the B compartment was characterized by a decrease in the score at boundary-distal regions. The difference between the two compartments prompted us to further investigate the role of ClsN as a B-compartment-specific factor for CID formation. Accordingly, we examined the ChIP-seq profile of ClsN for S. acidocaldarius cells treated with actinomycin D. Although actinomycin D delocalizes ClsN from most B compartment loci, the drug causes the appearance of a broad and high peak of ClsN in a 200-kb region ranging from 1.6 to 1.8 Mb (Figure 3E, middle right panel)(Takemata et al., 2019). We found that this region is enriched for short-range interactions (Figure 3E, top right panel) and coincident with two CID boundaries that were not observed in the untreated condition (Figure 3E, bottom panels). These new boundaries were located adjacent to subpeaks reminiscent of ClsN cliffs. Furthermore, in the 200-kb region encompassing the ClsN peak, ClsN bias and contact bias were correlated with each other as observed for B compartment loci in the untreated condition (Figure 3F). Thus, in agreement with the data from unperturbed cells (Figure 2), ClsN may help structure CIDs.

Modulation of local gene expression influences CID boundaries

We addressed the effects of deleting a highly-expressed operon, that encoding the S-layer proteins SlaA and SlaB in S. islandicus. This A compartment-localized 4925-bp operon generates the 5th highest levels of mRNA in the organism. Recent work has revealed that the S-layer genes are dispensable for Sulfolobus viability (Zhang et al., 2018). However, cells lacking the S-layer show significant changes in cell size, morphology and chromosome content and an enhanced propensity to form aggregates (Zhang et al., 2019). In wild-type cells, a strong CID boundary was detected at the slaAB locus (Figure 4A). Upon disruption of the locus, the boundary was abolished (Figure 4A).

Figure 4. — Two biological replicates of 3C-seq and RNA-seq data were used for each condition.

(A) Magnified views of a locus containing *S. islandicus slaAB*. The dotted line indicates a CID boundary in wild-type cells. (Top) 3C-seq contact maps. The resolution was set to 5 kb to avoid complete loss of signals around the deleted *slaAB* operon. (Middle) Strand-specific RNA-seq profiles. Normalized coverage is shown for each 100-bp bin. (Bottom) Directional preference plots at 2-kb resolution. Black horizontal lines indicate statistical thresholds.

(B) The fold change in RNA expression in Δ*slaAB* cells of *S. islandicus*. Differentially genes (absolute value of log2 fold change ≥ 1, adjusted p-value < 0.05) were indicated by red circles. B compartment regions in wild-type cells are indicated by gray shading.

(C) 3C-seq contact maps of the wild-type (top) and Δ*slaAB* (bottom) strains at 2-kb resolution. CID boundaries in the two strains are indicated by black bars below and above the corresponding contact map respectively.

(D) Pearson correlation maps at 10-kb resolution.

(E) Boxplot showing the fold change in RNA expression in Δ*slaAB* cells. Genes were classified according to their compartment identities in wild-type cells.

Given the highly pleiotropic nature of the mutation, we observed additional extensive changes to the gene expression profile in the mutant cells with 357 genes showing 2-fold or greater changes in expression relative to the wild-type cells (Figure 4B). Correspondingly, we also observed that the locations of a number of boundaries CIDs were altered in the mutant cells (Figures 4C). However, we emphasize that CID formation per se was still observed in the mutant. Notably, one unanticipated consequence of disrupting the S-layer genes was a loss of compartmentalization in the mutant cells (Figure 4D). The basis of this is unresolved at this time but could be linked to the morphological phenotypes of the mutant cells (Zhang et al., 2019) or possibly the elevated gene expression within the B compartment (Figure 4B and E).

To take a complementary and less phenotypically consequential approach, we tested the effect of inducing gene expression on CID boundaries. We generated an artificial construct containing the arabinose-inducible araS promoter (ParaS)(Peng et al., 2009) driving expression of the lacS β-galactosidase. This construct was inserted into an intergenic A compartment region, which is located between the SiRe_0163 and SiRe_0164 genes and less than 5 kb away from a CID boundary in wild type. Transcription of the transgene was modulated by the addition of sucrose (basal expression) or arabinose, resulting in a 6-fold induction (Fgure 5A). In comparison with the wild-type strain, even the basal expression of the transgene strengthened the nearby boundary (Figure 5B). Importantly, arabinose elevated transcription of the lacS transgene and further strengthened the boundary (Figure 5A and 5B). A collateral benefit of this experiment was that arabinose addition also induced expression of several endogenous chromosomal loci, including a cluster of genes (SiRe_2125- SiRe2130) involved in arabinose metabolism (Figure 5C). Significantly, induction of transcription of these B compartment-located genes resulted in generation of a CID boundary at this locus (Figure 5D). Thus, for both A and B compartment loci, induction of transcription promotes CID boundary formation, underscoring the causal role for transcription in this phenomenon.

Figure 5. — Two biological replicates of 3C-seq data were used for each condition. Three and two biological replicates of RNA-seq data were used for sucrose-fed wild-type cells and the other conditions respectively.

(A) Strand-specific RNA-seq profiles (normalized coverage plotted for every 100-bp bin) from wild-type and *SiRe*_0133::ParaS-*lacS* strains of *S. islandicus*. Cells were grown under non-inducing (sucrose) and inducing (arabinose) conditions for the *ParaS*-*lacS* transgene, whose position is indicated by gray shading.

(B) (Top) 3C-seq contact maps at 2-kb resolution. (Middle) The fold change in interaction score relative to that in the wild-type strain supplied with sucrose. (Bottom) Directional preference plots at 2-kb resolution. Black horizontal lines indicate statistical thresholds. The integration site of the transgene is indicated by white arrowheads.

(C) Effects of different carbon sources on RNA expression in the *SiRe*_0163::ParaS-*lacS* strain. Differentially genes (absolute value of log2 fold change ≥ 1, adjusted p-value < 0.05) were indicated by red circles. B compartment regions in the presence of sucrose are indicated by gray shading.

(D) Arabinose-induced formation of a CID boundary at the SiRe_2125-SiRe_2130 gene cluster (highlighted by a dotted rectangle), whose expression is upregulated in the presence of arabinose (also see Figure 5C). (Top) The fold change in interaction score (2-kb resolution) caused by the carbon source shift. (Middle and bottom) Directional preference plots at 2-kb resolution.

Loop formation in Sulfolobus chromosomes

In addition to the CIDs described above, the improved resolution of our 3C data revealed the presence of loop structures (Figure S1A). To quantify these, we employed the recently-described Chromosight algorithm (Matthey-Doret et al., 2020). We identified 77 and 67 loops in S. acidoaldarius and S. islandicus respectively, ~80% of which linked loci within the same compartment (Figures 6A, S5A, and S5B). Most loops were smaller than 100 kb (the median size of all loops were 82 kb in S. acidocaldarius and 62 kb in S. islandicus), but the largest loops approached half the size of the circular chromosomes for both species (~1 Mb). The aggregate plots of loops shown in Figure 6A reveal horizontal and vertical lines emanating from the central loop point, indicative of one anchor point of the loop having heterogeneous interactions with the other side, albeit with a preference for interaction with a unique locus (that gives rise to the initial loop signal).

Figure 6. — Wild-type cells of *S. acidocaldarius* and of *S. islandicus* were cultured in standard conditions. Two biological replicates of 3C-seq data were used for each condition.

(A) Loop-type interactions identified by Chromosight (Matthey-Doret et al., 2020) are indicated by black circles on distance-normalized contact maps (leftmost and third left panels). These loops were used to generate aggregate contact maps showing median values of distance-normalized interaction scores around the loop interactions (second left and fourth left panels). These data were generated using 3C-seq data at 2-kb resolution.

(B) Magnified views of distance-normalized contact maps (2-kb resolution) covering all ribosomal genes (genes encoding ribosomal protein subunits or ribosomal RNAs). Loop-type interactions are indicated by black circles. A/B compartments and the number of ribosomal genes in each 2-kb bin (the midpoint of the gene was used for the assignment) are shown above and on the left side of the contact map.

(C) The percentage of ribosomal-gene (RG) loops, which were defined as loops either of whose anchors is proximal (≤ 5 kb) to at least one RG. Expected percentages and the statistical significance of the observed data were determined by random permutation of loop positions (n = 10,000, see STAR Methods for more detail).

(D) Distributions of RGs loops. Each red line represents a pair of anchor sites forming a RG loop. The number of ribosomal genes is also plotted as in Figure 6B. The 16S-23S rRNA operon and the 5S rRNA gene are indicated by black and white arrowheads respectively.

(E) Model for the spatial organization of RG loops in *Sulfolobus*.

To determine the relationship between loops and CIDs, we quantified the coincidence of loop anchor points with CID boundary positions. 22 anchor sites in S. acidocaldarius (22%) and 45 anchor sites in S. islandicus (43%) colocalized with CID boundaries (Figure S5C). These boundary-associated anchors were often connected with each other to form “boundary-to-boundary” loops, which were significantly enriched (p < 10⁻⁴) in both genomes (Figure S5D). Most of the boundary-to-boundary loops spanned multiple CIDs (Figure S5E), suggesting that loop domain-type structures prevalent in eukaryotes (Rao et al., 2014) are rare in Sulfolobus, at least at the scale we adopted to detect CIDs and loops.

In eukaryotes, enrichment of the SMC complex cohesin at anchor sites is a hallmark of loops (Rao et al., 2014). We tested whether there was any correlation between loop anchor position and occupancy by the Sulfolobus SMC protein ClsN. However, no significant enrichment or depletion of ClsN occupancy was observed at loop anchors in either A or B compartment (Figure S5F).

Examination of the identity of genes in the proximity of loop anchors revealed an enrichment of genes encoding components of the ribosome (rRNAs and protein subunits) in both S. acidocaldrius and S. islandicus (Figure 6B). Loops involving these genes, hereafter referred to as Ribosomal-Gene (RG) loops, were significantly enriched (p ≤ 5 × 10⁻⁴) in both genomes (Figure 6C). We note that a given RG can form loops with multiple other RGs (Figure 6D). While we cannot eliminate the possibility that this represents heterogenous contacts within the population of cells examined, it raises the intriguing possibility that multiple RGs can be gathered together into a hub-like structure (Figure 6E).

Since RGs are among the most highly expressed genes in cells, we explored the role of transcription for tethering of loop anchors. Except for RGs, loop-anchor loci did not show a clear difference in gene expression relative to non-anchor loci (Figure S5G). Nevertheless, transcription appears important for formation of loops, as loop interactions found in actively growing S. acidocaldarius cells were attenuated after the treatment with actinomycin D, leading to ~90% loss of the loops (Figures 7A and 7B). This attenuation was observed even after normalizing for the global weakening of compartmentalization caused by the drug (Figures S6A and S6B)(Takemata et al., 2019). It should also be noted that all RG loops found in untreated cells disappeared after the treatment (Figure 7B). Quantification of loop strength by Chromosight revealed that actinomycin D reduced the strength of RG loops more profoundly than that of other loops (Figure 7C). Thus, transcription seems to be especially important for the formation of RG loops.

Figure 7. — Two biological replicates of 3C-seq data were used for each condition. Three and two biological replicates of RNA-seq data were used for control *S. acidocaldarius* cells and the other conditions respectively.

(A) An aggregate contact map centered at loop-forming bin pairs were generated as in Figure 6A using 3C-seq data of untreated, logarithmically growing *S. acidocaldarius* cells (left panel, labeled as control). For the same bin pairs, aggregate contact maps were generated using 3C-seq data from actinomycin D-treated (middle panel) and stationary-phase cells (right panel) of *S. acidocaldarius*.

(B) Venn diagram showing common and specific loops before and after the treatment of *S. acidocaldarius* with actinomycin D. Parenthesized numbers indicate numbers of RG loops.

(C) The change in loop strength after the treatment of *S. acidocaldarius* with actinomycin D (actinomycin D minus control). 2-kb bin pairs forming loops in the control condition were analyzed. Loop strength was calculated using Chromosight (see STAR Methods for more detail). p-value was calculated using Wilcoxon rank sum test.

(D) Venn diagram showing common and specific loops before and after the entry of *S. acidocaldarius* into stationary phase. Parenthesized numbers indicate numbers of RG loops.

(E) The change in loop strength after the entry of *S. acidocaldarius* into stationary phase (stationary phase minus control). 2-kb bin pairs forming loops in the control condition were analyzed. p-value was calculated using Wilcoxon rank sum test.

(F) The fold change in RNA expression after the entry of *S. acidocaldarius* into stationary phase. Genes were classified according to their compartment identities and proximity (Y: ≤ 2 kb, N: > 2 kb) to loop anchors in the control condition. The anchor-proximal genes were further classified according to whether they are RGs (Y) or not (N). p-values were calculated using Wilcoxon rank sum test.

(G) The data in Figures 6C and 6D were used to generate a scatter plot comparing the effects of actinomycin D and stationary phase on loop strength. Spearman correlation coefficient was calculated separately for RG loops and the others.

(H) An aggregate contact map centered at loop-forming bin pairs were generated as in Figure 6A using 3C-seq data from wild-type *S. islandicus* cells (left panel). For the same bin pairs, an aggregate contact map was generated using 3C-seq data from Δ*slaAB* cells of *S. islandicus* (right panel).

(I) Venn diagram showing common and specific loops between wild-type and Δ*slaAB* cells of *S. islandicus*. Parenthesized numbers indicate numbers of RG loops.

(J) The change in loop strength caused by the *slaAB* deletion in *S. islandicus* (Δ*slaAB* minus wild type). 2-kb bin pairs forming loops in wild-type cells were analyzed. p-value was calculated using Wilcoxon rank sum test.

(K) The fold change in RNA expression caused by the *slaAB* deletion in *S. islandicus*. Genes were classified according to their compartment identities and proximity (Y: ≤ 2 kb, N: > 2 kb) to loop anchors in wild-type cells. The anchor-proximal genes were further classified according to whether they are RGs (Y) or not (N). p-values were calculated using Wilcoxon rank sum test.

To gain more insight into the transcription-dependent formation of RG loops, we grew S. acidocaldarius into stationary phase, upon which cells downregulate ribosome biogenesis to save energy. In stationary phase, positions of CID boundaries were markedly altered, and they did not show a clear correlation with highly expressed genes (Figure S7). Similar loss of correlation has been reported for stationary-phase cultures of the bacterium E. coli (Lioy et al., 2018). The entry of S. acidocaldarius into stationary phase also led to global impairment of loop formation (Figures 7D and 7E). This inhibitory effect was still visible after cancelling the global weakening of compartmentalization that occurs in stationary phase (Figure S6C)(Takemata et al., 2019). Among the affected loops, RG loops most strikingly decreased their strength in response to stationary phase as seen in the stationary-phase experiments (Figures 7C and 7E). This pronounced effect on RG loops was observed concurrently with RG-specific downregulation of gene expression at loop anchors (2.3-fold reduction on average, Figure 7F). Furthermore, transcription inhibition by actinomycin D and the entry into stationary phase reduced the strength of RG loops in a correlated manner (r = 0.52)(Figure 7G). Such correlation was not seen for other loops (r = −0.10, significance of the difference in the correlation coefficients: p = 6.4 × 10⁻³). Taken together, these results support that transcription of RGs promotes loop formation at these regions.

As shown above, loss of RG loops in stationary phase can be explained by repression of their anchor loci. However, loss of other loops in stationary phase was seemingly unrelated to altered gene expression (Figure 7F). Given that gene clustering is linked to micro-compartments such as liquid droplets (Hnisz et al., 2017), it is possible that the reduced chromosome compartmentalization in stationary phase is another factor responsible for loop attenuation. To further explore the potential role of A/B compartments for loop formation, we characterized loop structures in the ΔslaAB mutant of S. islandicus, which has lost A/B compartments (Figure 4D). Loops present in wild-type S. islandicus reduced their number and strength in the mutant (Figures 7H–7J). Interestingly, the slaAB deletion lowered the strength of RG loops and other loops to a similar extent (Figure 7J), while the mutation by and large did not alter the expression of anchor-proximal genes for either type (Figure 7K). More specifically, RGs at loop anchors decreased their expression levels very slightly (1.3-fold on average), and the expression of other anchor-proximal genes did not change at all on average (1.04-fold). These results are in agreement with the idea that compartmentalization is required for robust formation of a majority of loops, whereas local transcription serves as an additional factor driving the formation of RG loops.

Discussion

Previously, we have provided evidence that Sulfolobus chromosomes are organized into discrete compartments; a transcriptionally active A compartment and a less transcriptionally active B compartment. Our current work confirms the compartmentalization and further reveals that Sulfolobus chromosomes possess self-interacting domains – similar in scale and, at least in the A compartment, similar in properties to bacterial CIDs. The Sulfolobus A compartment CIDs are typically bounded by highly transcribed genes or operons. This relationship appears causal in nature as inhibition of transcription or deletion or insertion of highly transcribed genes influences CID boundary localization. In archaea, as in bacteria, transcription and translation are believed to be coupled processes. This coordination will lead to an enrichment of macromolecular complexes (ribosome, etc.) at highly transcribed genes, which could serve as steric hindrance for inter-domain contacts. We note, however, that translation is not an absolute requirement for CID barrier formation, as the very highly transcribed rRNA locus serves as a robust CID boundary in both Sulfolobus species examined in the current work. CID formation independent of concurrent translation has also been reported for the bacterium Caulobacter crescentus (Le and Laub, 2016).

The less transcriptionally active B compartment also contains CIDs. In the B compartment CIDs, boundaries are associated with locally-high transcription levels and also with local biases in occupancy of ClsN. We additionally observe a correlation between a ClsN peak and local chromosomal folding in actinomycin D-treated cells. Thus, it is plausible that ClsN is maintaining a locally folded region independent of transcription to help organize transcriptionally less active regions into CID arrays. ClsN does not appear to be enriched at CID boundaries, suggesting that ClsN is acting in a manner distinct from that observed with eukaryotic cohesin. Rather, ClsN’s distribution is compatible with it creating local domains by facilitating accretion or coalescence of sequences in the vicinity of high occupancy binding sites. Such a mechanism could involve second strand capture events by the SMC protein but we speculate that additional local loop extrusion may occur.

In addition to CIDs, the high-resolution data presented in this study have allowed us to identify tens of loops, many of which involve RGs. On the population-based 3C-seq data, single RG-loop anchors are spatially linked to multiple other RG-loop anchors, hinting at the existence of a hub-like structure that colocalizes RGs in a single Sulfolobus cell. This proposed gene cluster is reminiscent of the eukaryotic nucleolus, a well-known subnuclear compartment involved in ribosome biogenesis. A number of microscopic studies have described a similar nucleolus-like structure in bacteria, which is thought to contain multiple rRNA operons (Cabrera and Jin, 2003; Gaal et al., 2016; Lewis et al., 2000). To our knowledge, however, spatial clustering of bacterial RGs has not been detected by 3C techniques (Cagliero et al., 2013; Lioy et al., 2018; Marbouty et al., 2015). By taking advantage of 3C-seq complementary to microscopic approaches, and of the close phylogenetic relationship between archaea and eukaryotes, future studies of Sulfolobus will provide insight into the spatial organization of ribosome biogenesis and its evolution.

The correlation between attenuation/abolishment of chromosome compartmentalization and disruption of loops in Sulfolobus suggests that the two phenomena are causally linked. Approximately 80% of the Sulfolobus loops tether loci within the same compartment (Figure S5B), and it is reasonable to think that compartmentalization helps this engagement. Reciprocally, intra-compartment loops may serve as physical links that facilitate chromosome compartmentalization. In our previous study, it was unclear whether the A compartment is actively defined, or simply defined by exclusion of active, ClsN-less genes from the B compartment. Loop structures within the A and B compartments could provide a basis for maintaining the integrity of both compartments.

Finally, our data reveal that the chromosomes of the crenarchaeon Sulfolobus and eukaryotes are organized into similar, multi-scale structural entities – compartments, domains, and loops. Recent work on the euryarchaeon Haloferax volcanii has also identified CIDs and loops in the chromosome of that organism (Romain Koszul, personal communication), raising the possibility that these structures are common to a broad range of archaeal species. Interestingly, the aforementioned study did not observe A/B-type compartments in H. volcanii. The “more-eukaryotic” higher-order structure of the Sulfolobus chromosome may reflect the fact that the crenarchaea shared more recent common ancestor with eukaryotes than did the euryarchaea.

Limitations

The contact data derived from the 3C-Seq analyses have their resolution defined by the frequency of occurrence of restriction enzyme sites and the efficiency of digestion of the genomic DNA. Accordingly, we have conservatively binned our data in 2 kilobase windows. The data from exponential cells are derived from asynchronous cell populations. There may be modulation of local conformation and boundaries during the cell cycle. Similarly, it remains to be determined whether the network of RG loops, which are detected in the population-based 3C-seq data, colocalizes the RG-loop anchors to form a hub-like structure in a single cell. Finally, although our data reveal a correlation between ClsN distribution and CID boundaries in the B compartment, direct causality has yet to be established.

STAR★METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Steve Bell (stedbell@indiana.edu).

Materials Availability

Plasmid and cell lines generated are available from the Lead Contact without restrictions with reasonable compensation by requestor for its processing and shipping.

Data and Code Availability

All sequencing data used in this study has been deposited to the NCBI Gene Expression Omnibus (GEO). The accession number for 3C-seq and RNA-seq data generated in this study is GEO: GSE159537.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Archaeal strains

The S. acidocaldarius wild-type strain DSM639 was grown as described previously (Takemata et al., 2019) and used for all experiments with S. acidocaldarius. The strain was grown shaking at 78°C in Brock’s medium (containing 0.2% sucrose and 0.1% tryptone, pH 3.2) until the culture reached log phase (OD₆₀₀ ≈ 0.2) or stationary phase (OD₆₀₀ ≈ 1). For treatment of DSM639 with actinomycin D, 1 mg/ml of the drug (dissolved in DMSO) was added to a log-phase culture to a final concentration of 15 mg/ml. The treated cells were collected after 30 min.

The S. islandicus strains used in this study were derived from the wild-type strain REY15A. If not stated otherwise, the S. islandicus strain E233S (ΔpyrEF ΔlacS) (Deng et al., 2009) was grown shaking at 78°C in TSVY medium (containing mineral salts, 0.2% sucrose, 0.1% tryptone, 0.05% yeast extract, and 1 × vitamin solution) until the culture reached mid-log phase (OD600 ≈ 0.2) after overnight culture. The medium was also supplemented with 20 μg/ml of uracil. For experiments using S. islandicus ΔslaAB (ΔpyrEF ΔlacS ΔargD ΔslaAB::argD_sto) and control (E233S) cultures, cells were grown to OD₆₀₀ ≈ 0.05 (early log phase) without shaking to reduce mechanical stress that could exacerbate phenotypes of ΔslaAB (Zink et al., 2019). For arabinose induction experiments using E233S and the SiRe_0163::ParaS-lacS strain (ΔpyrEF ΔlacS ΔargD SiRe_0163::ParaS-lacS_sso-argD_sto), sucrose was replaced with 0.2% D-arabinose.

METHOD DETAILS

Strain constructions

To delete slaAB in S. islandicus by homologous recombination, the Sulfolobus tokodaii argD gene (ST1348) was amplified from the pCR-ArgD plasmid (Takemata et al., 2019) using the primers slaABdelta_F and slaABdelta_R, which contain flanks that are homologous to 75 bp of sequence upstream and downstream of the deletion target. 1 μg of the deletion cassette was used for transformation of the agmatine auxotroph strain E235 (Zhang and Whitaker, 2018) as described previously (Takemata et al., 2019). ΔslaAB transformants were confirmed by colony PCR and sequencing of the locus.

The SiRe_0163::ParaS-lacS strain of S. islandicus was constructed as follows. To fuse the coding sequence of S. solfataricus lacS to a promoter sequence of the S. solfataricus araS gene (ParaS), PCR was carried out using genomic DNA of S. solfataricus P2 as a template and the primers lacS_NdeI_F and lacS_SalI_R. Following digestion with NdeI and SalI, the PCR fragment was ligated into the NdeI and SalI sites of the pSSRgD plasmid (Takemata et al., 2019). In the resultant plasmid pSSRgD-lacS, the lacS sequence has been integrated between ParaS and the marker gene argD (from S. tokodaii). To integrate the ParaS-lacS-argD construct into the intergenic region between the SiRe_0163 and SiRe_0164 genes, the construct was amplified using the primers SiRe0163_Para and SiRe0164_argD that contain flanks homologous to 75 bp of sequence upstream and downstream of a target integration site. 1 μg of the PCR product was used for transformation of E235 as described for the construction of ΔslaAB. Insertion of the ParaS-lacS-argD sequence was confirmed by colony PCR and sequencing of the locus.

The primer sequences used for strain construction are shown in Table S1.

3C-seq

3C-seq was carried out following by and large the same procedure as described for our previous Hi-C experiments (Takemata et al., 2019). To preserve DNA-DNA contacts, 20 ml of cell culture were mixed with (80-X) ml of ambient PBS buffer and X ml of 37% formaldehyde. X was 10.8 for S. acidocaldarius (final formaldehyde concentration of 4%) and 16.2 for S. islandicus (final formaldehyde concentration of 6%). The reaction was incubated for 30 min at 25°C before quenching with glycine (a final concentration of 250 mM) for 10 min at room temperature. The fixed cells were collected by centrifugation and washed twice with PBS buffer. The cell pellet was stored at −80°C until use.

A frozen cell pellet was resuspended and lysed as described in the previous work (Takemata et al., 2019), and 25 μL of the lysate (corresponding to ~4 × 10⁸ cells) were mixed with 41.8 μl of 1 × NEBuffer 2, 3.2 μl of 10 × NEBuffer 2, 20 μl of 10% Triton X-100. Chromosomal DNA was digested by adding 10 μl of 10 U/μl AluI (NEB) to the mixture and incubating the reaction for 3.5 h at 37°C. The reaction was quenched by adding 11.1 μl of 10% SDS and incubating for 30 min at 37°C. For ligation, the reaction was mixed with 1469 μl of nuclease-free water, 200 μl of 10 × T4 DNA Ligase Reaction Buffer (NEB), 200 μl of 10% Triton X-100, and 20 μl of 400 U/μl T4 DNA ligase (NEB). The ligation reaction was incubated for 4 h at 16°C with occasional inversion of the tube (every 30 min). After the ligation was completed, the reaction was mixed with 200 μl of 10% SDS, 100 μl of 0.5 M EDTA, and 10 μl of 20 mg/ml proteinase K. To reverse crosslinks, the mixture was incubated for 6 h at 65°C and then for at least 6 h at 37°C. The DNA was extracted twice with phenol:chloroform:isoamyl alcohol and isopropanol-precipitated together with 40 mg of glycogen. The purified DNA was dissolved in 40 μl of 1 × NEBuffer 2 containing 0.1 mg/ml RNase A and incubated for 30 min at 37°C. Successful proximity ligation was confirmed by running 10 μl of the DNA on a gel. The remainder of the DNA was further extracted with phenol:chloroform:isoamyl alcohol and ethanol-precipitated. The DNA was dissolved in 90 μl of Buffer EB (QIAGEN) and sheared with a Bioruptor (Diagenode) at low power for 40 cycles (30-s on, 30-s off). 55.5 μl of the sheared DNA were used for library construction with NEBNext Ultra DNA Library Prep Kit for Illumina and NEBNext Multiplex Oligos for Illumina (NEB) according to the manufacturer’s instructions. DNA was purified using AMPure XP Beads (Beckman Coulter) to enrich adaptor-ligated DNA molecules of 400–500 bp. 8 or 9 cycles of PCR were conducted for library amplification. DNA libraries were paired-end sequenced (43 bp × 2) on the Illumina NextSeq platform at the Center for Genomics and Bioinformatics at Indiana University.

RNA-seq

RNA extraction and library construction were carried out as described previously (Takemata et al., 2019). RNA-seq experiments using S. acidocaldarius and sucrose-fed S. islandicus E233S were conducted once. Obtained data were analyzed together with the corresponding RNA-seq data in the above study, resulting in a total of three biological replicates per condition. Other RNA-seq experiments were performed in biological duplicates. All datasets are available at the Gene Expression Omnibus database, and the accession numbers of the datasets from the previous study are as follows. S. acidocaldarius (log phase): GSM3662081 and GSM3662082. S. acidocaldarius (stationary phase): GSM3662083 and GSM3662084. S. islandicus: GSM3662085 and GSM3662086.

QUANTIFICATION AND STATISTICAL ANALYSIS

Generation of 3C-seq matrices

3C-seq reads were processed using HiC-Pro version 2.9.0 (Servant et al., 2015) as performed previously for Hi-C (Takemata et al., 2019). To alleviate a self-ligation issue resulting from the circular nature of Sulfolobus genomes (Takemata et al., 2019), we re-defined the genomic coordinates of each Sulfolobus species so that they started from the first base of the first restriction site in the original definition (S. acidocaldarius DSM639: 820 bp from the start, S. islandicus REY15A: 817 bp from the start). These modified genome sequences were then binned at different resolutions to generate raw contact matrices. Valid read pairs from biological replicates (summarized in Table S2) were pooled and used for iterative correction implemented in HiC-Pro. When 3C-seq mapping was carried out for S. islandicus at 2-kb resolution, bins with extremely low coverage were filtered out by setting the FILTER_LOW_COUNT_PER parameter to 0.01. The low coverage over the filtered-out bins was due to a large number of repeat sequences (transposons, etc.) in the S. islandicus genome, whereas the number of such repeats is very small in the S. acidocaldarius genome. After the iterative correction, all values were multiplied to make the sum of interaction scores equal to 1,000 for each row and column. In some cases, reads from biological replicates were processed separately to evaluate the reproducibility of data.

Generation of distance-normalized contact maps, Pearson correlation maps, and compartment index plots

Distance-normalized contact matrices, Pearson correlation matrices, and compartment index plots were generated as described previously using HiTC (Servant et al., 2012; Takemata et al., 2019). Compartment identity of each gene was determined according to which compartment the midpoint of the gene belongs to.

Analyses of published Hi-C data for comparison with 3C-seq data

Published Hi-C sequence data were processed as performed in the original study (Takemata et al., 2019). Reads from all available replicates were pooled for the analysis. The datasets are available at the Gene Expression Omnibus database, and their accession numbers are as follows. S. acidocaldarius (untreated): GSM3662029, GSM3662030, and GSM3662031. S. acidocaldarius (treated with actinomycin D): GSM3662034 and GSM3662035. S. islandicus: GSM3662039, GSM3662040, and GSM3662041.

Directional preference

Directional preference was calculated essentially according to a previous study (Le et al., 2013). To calculate the directional preference for a given bin, we collected 3C-seq interaction scores between the bin and bins located either downstream or upstream within the distance of 40 kb. We then compared log2 values of the upstream and downstream vectors by paired t test. Directional preference was defined as t-value of the test, and a p-value of 0.05 was used as a threshold to assess statistical significance. When either of the two vectors contained an interaction involving a filtered-out bin, the corresponding pair was not used to calculate t-value. Taking advantage of the circular nature of Sulfolobus genomes, we calculated directional preference also for bins that are located within 40 kb from the start/end of the chromosome.

To locate CID boundaries, we divided the genome into segments where direction preference value was consistently positive or consistently negative. We then retained the segments where the maximum of absolute values of direction preference was higher than the threshold. Boundaries between the upstream segments with negative values (that is, loci preferentially interacting with upstream regions) and the downstream segments with positive values (that is, loci preferentially interacting with downstream regions) were defined as CID boundaries. If two such segments were separated by loci whose directional preference was below the threshold, the CID boundary was located at the start site of the downstream segment.

Insulation score analysis

Insulation score was determined according to a previous study (Crane et al., 2015). Using 3C-seq contact matrices at 2-kb resolution, we calculated interaction frequencies between the left side and right side of each bin. This was done by aggregating values within a square, one of whose corner was on the matrix diagonal. The side length of the square was defined as “insulation square size.” The sum of the interaction frequencies within the square was calculated for each bin, and this value was further divided by the mean value of the sums in the genome. log2 of the resultant value was defined as insulation score. Taking advantage of the circular nature of Sulfolobus genomes, we calculated insulation scores also for bins that are located within the insulation square size from the matrix start/end.

To quantify the effect of actinomycin D on insulation activities of CID boundaries in Figures 3C and 3D, we classified 2-kb bins according to which compartment they belong to and whether they are located within ±6-kb windows centered at CID boundaries. For each compartment, the change in insulation score (the score in the actinomycin D-treated sample minus the score in the untreated sample) was compared between the boundary-proximal and -distal groups. In Figure 3D, the difference in the medians of the two groups (boundary-proximal minus boundary-distal) is shown as relative change in insulation score.

Analysis of loop structure by Chromosight

To identify DNA loops, we processed 3C-seq data using Chromosight (Matthey-Doret et al., 2020). This algorithm uses a template image representing a 3D structure (loop, domain border, etc.) and searches the contact map of interest for sub-images resembling the template. Chromosight also outputs the metric named loop score, which represents the similarity between a detected pattern and the template (calculated as the Pearson correlation coefficient). This value was used as the strength of the loop in this study.

To prepare inputs for Chromosight, we first generated raw 3C-seq contact matrices binned at 2-kb resolution using HiC-Pro (Servant et al., 2015) and pooled reads. After removing intra-bin reads, we converted the matrices into cool files using the hicpro2higlass.sh script in HiC-Pro and the COOLER package version 0.8.9 (Abdennur and Mirny, 2020). Normalization was also performed at this stage using the -n option for hicpro2higlass.sh. The generated files were then processed using the “detect” function in Chromosight using default parameters. This initial analysis detected more than 200 loops in both S. acidocaldarius and S. islandicus. To focus on the most robust loops, we omitted loops whose loop scores were less than 0.4 or which were shorter than 20 kb. We also omitted loops formed at low-coverage bins in S. islandicus. The low-coverage bins were determined the same way as we did to generate iteratively-normalized contact maps with HiC-pro. The center of a bin that forms loop(s) was defined as the position of the loop anchor. To quantify the loop score change caused by treatments, we implemented the “quantify” function of Chromosight. The coordinates of loop interactions in a control sample were used as a reference. The cool file from the treated sample and the list of loops from a control sample were used as inputs. The cocor website was used for the statistical analysis of loop score changes in Figure 7G (Diedenhofen and Musch, 2015).

When counting the number of common loops between two datasets (pooled data vs. single replicate, mutant vs. wild type, etc.), we treated two loops identical if the upstream ends of the two loops are located ≤ 4 kb from each other and the downstream ends of the two loops are located ≤ 4 kb from each other. When identifying loops in each biological replicate, we did not set the threshold for loop score. This is because the threshold of 0.4, which was used for the pooled data, was too stringent for the single replicate (it seems that loop score is sensitive to read depth).

Aggregate contact maps

Aggregate contact maps in Figures 6 and 7 were generated from distance-normalized contact matrices (2-kb resolution) by collecting 26 × 26 sub-metrices centered at bin pairs of interest. If the sub-matrix overlaps with the main diagonal, the values below/above the diagonal were set to NA. Values at the same coordinates in the sub-matrices were gathered to calculate the median. The heatmap of these medians is shown as an aggregate contact map.

Aggregate contact maps in Figure S6 were generated as follows. We first prepared two iteratively-corrected contact matrices (2-kb resolution) to be compared (M_a and M_b) and used them to generate a ratio matrix (M_a/M_b). From this ratio matrix, we extracted a 26 × 26 sub-matrix centered at a loop-forming bin pair of interest (M’). We then collected all bin pairs that have the same distance and compartment type as the loop-forming bin pair. These pairs and the ratio matrix were used to generate an aggregate 26 × 26 sub-matrix (M’_ctrl) as performed for Figure 6A. We generated a log2 ratio matrix from M’ and M’_ctrl (M’/M’_ctrl) to visualize the change in the loop interaction while normalizing global effects on chromosome conformation. This log2 ratio matrix was generated for the loop-forming bin pairs of interest, and the resultant matrices were piled up to generate an aggregate matrix in which each value represents the median of the piled-up values at the corresponding coordinate.

ChIP-seq data analysis

ChIP-seq data that we published before (Takemata et al., 2019) were used for analysis. Protein enrichment was calculated as described in the same study using reads pooled from biological replicates. The aggregate plot of ClsN occupancy around loop anchors was generated using deepTools version 3.0.0 (Ramirez et al., 2016). After mapping, ChIP reads and input reads were compared using bamCompare (parameters: −bs 100, – scaleFactorsMethod readCount). The output file was processed using computeMatrix and plotProfile. The datasets are available at the Gene Expression Omnibus database, and their accession numbers are as follows. S. acidocaldarius (untreated): GSM3662057, GSM3662058, GSM3662070, and GSM3662071. S. acidocaldarius (treated with actinomycin D): GSM3662060 and GSM3662073. S. islandicus: GSM3662063, GSM3662064, GSM3662076, and GSM3662077.

RNA-seq data analysis

To generate RNA-seq profiles, reads from biological replicates were pooled and mapped in a stranded or un-stranded manner using Bowtie 2 version 2.3.2 (parameter:–maxins 1000)(Langmead and Salzberg, 2012) and the genome sequences of S. acidocaldarius DSM639 (GenBank ID: CP000077.1) and S. islandicus REY15A (GenBank ID: CP002425.1). After removal of multi-mapping reads, the remaining reads were processed using bamCoverage implemented in deepTools version 3.0.0 (Ramirez et al., 2016) to calculate read coverage. When reads were mapped in an un-stranded manner, read coverage was normalized using the option --normalizeUsing RPKM. When reads were mapped in a strand-specific manner, read coverage was normalized to the total number of mapped reads (the number of reads mapped to the Watson strand + the number of reads mapped to the Crick strand).

To quantify expression levels for protein-coding genes and noncoding RNA genes, reads from each biological replicate were mapped using Salmon version 0.8.2 (Patro et al., 2017). The genome coordinates of these genes were downloaded from UCSC Archaeal Genome Browser (http://archaea.ucsc.edu/index.html). Genes whose length is ≤ 100 bp were omitted for the downstream analyses. When Transcripts Per Million (TPM) was used, the value was averaged for biological replicates. When calculating the log2 fold change to determine differentially expressed genes, the data generated by Salmon was further processed using DEseq2 (Love et al., 2014). For this, the number of mapped reads that was determined by Salmon was rounded to the nearest integer for each replicate.

To use the operon as a unit for gene expression analysis, operons were predicted using Operon-mapper (Taboada et al., 2018). Genome sequences and GFF files were used as input (accession for S. acidocaldarius: CP000077.1 and GCA_000012285.1, accession for S. islandicus: CP002425.1 and GCA_000189555.1). TPM values of these operons were calculated as done for protein-coding genes and noncoding RNA genes.

When mapping RNA-seq reads for the arabinose induction experiments, we integrated the sequence of the ParaS-lacS-argD construct into the genome sequence of S. islandicus REY15A. This modified version of the genome sequence was used for mapping.

Colocalization of highly expressed genes and CID boundaries

To investigate the relationship between CID boundary formation and transcription in each compartment, CID boundaries were classified according to their compartment identities, and the percentage of CID boundaries colocalizing with highly expressed gene (top 10% in the genome, ranked based on TPM) was calculated for each group. A CID boundary and a highly expressed gene were regarded as colocalizing with each other if the gene overlaps with a window from −2 kb to +2kb relative to the boundary.

As performed in previous studies (Lioy et al., 2018; Trussart et al., 2017), we evaluated the significance of colocalization by conducting permutation tests in which all boundary positions were randomly shuffled across the genome. The number and sizes of CIDs were maintained before and after this randomization. After the shuffling, we calculated the percentage of CID boundaries colocalizing highly expressed genes. This procedure was repeated 10,000 times, and the mean percentage in the repetition was represented as an expected value. We also counted the number of permutation procedures where the value was the same or higher/lower than the observed value. This number was divided by the number of the trials to compute an empirical p-value.

Comparing ClsN bias with contact bias

To explore the role of ClsN in CID boundary formation, we calculated and compared two metrics “ClsN bias” and “contact bias.” ClsN bias represents which side has more ClsN within a ±40-kb window. To compute ClsN bias for each 2-kb bin, we measured the mean ClsN occupancy of the right 20 bins and that of the left 20 bins. The right mean minus the left mean was used as ClsN bias at the bin of interest. For the same bin, we also quantified which side in a ±40-kb window the bin interacted more frequently with. This metric was used as contact bias and calculated as follows. We first collected 3C-seq interaction scores between the bin and bins located either downstream or upstream within the distance of 40 kb. We paired these scores by the interaction distance and calculated the log2 ratio of each pair (right/left). The mean of the log2 values were used as contact bias of the bin.

Comparing the strength of ClsN bias and of boundary

To estimate the magnitude of ClsN bias change around a given CID boundary (“strength of CID bias”), we first calculated the maximum of ClsN bias in the right CID and the minimum of ClsN bias in the left CID. The maximum value minus the minimum value was defined as the strength of ClsN bias around the boundary. We compared this metric with boundary strength, which was quantified from the insulation score plot as described in a previous study (Crane et al., 2015). In this quantification, square size was set to 40 kb for both S. acidocaldarius and S. islandicus. Insulation delta span (the window size to quantify delta vector) was set to 20 kb and 10 kb for S. acidocaldarius and S. islandicus respectively.

Features associated with loop anchors

To explore loop formation at ribosomal genes (RGs), we investigated whether a ±5-kb window centered at the loop anchor overlaps with any RGs (genes with KEGG Brite ID 03011). If the overlap was observed, the loops involving the anchor site were referred to as “RG loops.” To evaluate the enrichment of RG loops, we conducted permutation test by shuffling all loop positions across the genome. The number, sizes, and compartment types of loops (A-to-A, B-to-B, and A-to-B) were maintained before and after this randomization. After the shuffling, we calculated the percentage of ribosomal-gene loops relative to all loops. This procedure was repeated 10,000 times, and the mean percentage in the repetition was represented as an expected value. We also counted the number of permutation procedures where the value was the same or higher/lower than the observed value. This number was divided by the number of the trials to compute an empirical p-value.

We also explored the relationship between loop formation and CID formation by investigating whether a ±5-kb window centered at the loop anchor overlaps with any CID boundary. If the overlap was observed for both anchors of a loop, the loop was referred to as a “boundary-to-boundary loop.” Enrichment of boundary-to-boundary loops were evaluated as performed for RG loops.

Quantification of gene expression at boundary/anchor-proximal regions and -distal regions

To compare gene expression between genes proximal and distal to CID boundaries, we classified genes according to which compartment the gene belongs to and whether or not the gene overlaps with a window from −2 kb to +2kb relative to a CID boundary. TPM or the log2 fold change in TPM was compared between resultant groups. We followed the same procedures to characterize gene expression features around loop anchors.

Data visualization

Data visualization was carried out using R software (http://www.R-project.org).

Supplementary Material

NIHMS1654324-supplement-1.pdf^{(46.7MB, pdf)}

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, Peptides, and Recombinant Proteins
Actinomycin D	Sigma Aldrich	Cat# A1410–2MG
Restriction enzymes	New England Biolabs	Cat# R0104M
T4 DNA ligase	New England Biolabs	Cat# M0202L
Phenol:chloroform:isoamyl alcohol	Sigma Aldrich	Cat# 77618–100ML
AMPure XP Beads	Beckman Coulter	Cat# A63880
37 % Formaldehyde	Thermo Fisher Scientific	Cat# BP531–500
Glycogen	Invitrogen	Cat# 10814–010
Critical Commercial Assays
NEBNext Ultra DNA Library Prep Kit for Illumina	New England Biolabs	Cat# E7370S
NEBNext Multiplex Oligos for Illumina	New England Biolabs	Cat# E7335S, E7500S, E7710S
NEBNext Ultra II Directional RNA Library Prep Kit for Illumina	New England Biolabs	Cat# E7760S
Deposited Data
Raw data and processed data	This work	GEO: GSE159537
Published Hi-C data	(Takemata et al., 2019)	GEO: GSE128063
Experimental Models: Organisms/Strains
S. acidocaldarius DSM639	DSMZ	N/A
S. islandicusE233S (ΔpyrEF ΔlacS)	(Deng et al., 2009)	N/A
S. islandicus E235 (ΔargD)	(Zhang and Whitaker, 2018)	N/A
S. islandicus ΔslaAB (ΔpyrEF ΔlacS ΔargD ΔslaAB::argD_sto)	This work	N/A
S. islandicus SiRe_0163::ParaS-lacS (ΔpyrEF ΔlacS ΔargD SiRe_0163::ParaS-lacS_sso-argD_sto)	This work	N/A
Oligonucleotides
For primers, see Table S1	This work	N/A
Recombinant DNA
pCR-ArgD	(Takemata et al., 2019)	N/A
pSSRgD-lacS	This study	N/A
Software and Algorithms
HiC-Pro	(Servant et al., 2015)	https://github.com/nservant/HiC-Pro
HiTC	(Servant et al., 2012)	https://bioconductor.org/packages/release/bioc/html/HiTC.html
Bowtie 2	(Langmead and Salzberg, 2012)	http://bowtie-bio.sourceforge.net/bowtie2/index.shtml
deepTools	(Ramirez et al., 2016)	https://deeptools.readthedocs.io/en/develop/
Salmon	(Patro et al., 2017)	https://combine-lab.github.io/salmon/
Operon-mapper	(Taboada et al., 2018)	http://biocomputo.ibt.unam.mx/operon_mapper/
DEseq2	(Love et al., 2014)	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
Cocor	(Diedenhofen and Musch, 2015)	http://comparingcorrelations.org/
Chromosight	(Matthey-Doret et al., 2020)	https://github.com/koszullab/chromosight
COOLER	(Abdennur and Mirny, 2020)	https://github.com/open2c/cooler

Open in a new tab

Highlights.

Chromosomes of Sulfolobus archaea possess A/B compartments, CIDs and loop structures.

CID boundaries are formed by high transcription, most notably in the A compartment.

The Sulfolobus SMC protein, ClsN, may facilitate CID bundling in the B compartment.

Long-range loop interactions occur between genes encoding ribosome components.

Acknowledgements

This project was funded by a JSPS Overseas Research Fellowship (Japan Society for the Promotion of Science) and by NIH R01GM135178. We thank the Center for Genomics and Bioinformatics at Indiana University for high-throughput sequencing. We thank Romain Koszul for sharing unpublished data.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of Interests

The authors declare no competing interests.

REFERENCES

Abdennur N, and Mirny LA (2020). Cooler: scalable storage for Hi-C data and other genomically labeled arrays. Bioinformatics 36, 311–316. [DOI] [PMC free article] [PubMed] [Google Scholar]
Alipour E, and Marko JF (2012). Self-organization of domain structures by DNA-loop-extruding enzymes. Nucleic Acids Res 40, 11202–11212. [DOI] [PMC free article] [PubMed] [Google Scholar]
Cabrera JE, and Jin DJ (2003). The distribution of RNA polymerase in Escherichia coli is dynamic and sensitive to environmental cues. Mol Microbiol 50, 1493–1505. [DOI] [PubMed] [Google Scholar]
Cagliero C, Grand RS, Jones MB, Jin DJ, and O’Sullivan JM (2013). Genome conformation capture reveals that the Escherichia coli chromosome is organized by replication and transcription. Nucleic Acids Res 41, 6058–6071. [DOI] [PMC free article] [PubMed] [Google Scholar]
Crane E, Bian Q, McCord RP, Lajoie BR, Wheeler BS, Ralston EJ, Uzawa S, Dekker J, and Meyer BJ (2015). Condensin-driven remodelling of X chromosome topology during dosage compensation. Nature 523, 240–244. [DOI] [PMC free article] [PubMed] [Google Scholar]
Davidson IF, Bauer B, Goetz D, Tang W, Wutz G, and Peters JM (2019). DNA loop extrusion by human cohesin. Science 366, 1338–1345. [DOI] [PubMed] [Google Scholar]
Deng L, Zhu H, Chen Z, Liang YX, and She Q (2009). Unmarked gene deletion and host-vector system for the hyperthermophilic crenarchaeon Sulfolobus islandicus. Extremophiles 13, 735–746. [DOI] [PubMed] [Google Scholar]
Diedenhofen B, and Musch J (2015). cocor: a comprehensive solution for the statistical comparison of correlations. PLoS One 10, e0121945. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dixon JR, Selvaraj S, Yue F, Kim A, Li Y, Shen Y, Hu M, Liu JS, and Ren B (2012). Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature 485, 376–380. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gaal T, Bratton BP, Sanchez-Vazquez P, Sliwicki A, Sliwicki K, Vegel A, Pannu R, and Gourse RL (2016). Colocalization of distant chromosomal loci in space in E. coli: a bacterial nucleolus. Genes Dev 30, 2272–2285. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gibcus JH, Samejima K, Goloborodko A, Samejima I, Naumova N, Nuebler J, Kanemaki MT, Xie L, Paulson JR, Earnshaw WC, et al. (2018). A pathway for mitotic chromosome formation. Science 359. [DOI] [PMC free article] [PubMed] [Google Scholar]
Hnisz D, Shrinivas K, Young RA, Chakraborty AK, and Sharp PA (2017). A Phase Separation Model for Transcriptional Control. Cell 169, 13–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
Imachi H, Nobu MK, Nakahara N, Morono Y, Ogawara M, Takaki Y, Takano Y, Uematsu K, Ikuta T, Ito M, et al. (2020). Isolation of an archaeon at the prokaryote-eukaryote interface. Nature 577, 519–525. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kim Y, Shi Z, Zhang H, Finkelstein IJ, and Yu H (2019). Human cohesin compacts DNA by loop extrusion. Science 366, 1345–1349. [DOI] [PMC free article] [PubMed] [Google Scholar]
Langmead B, and Salzberg SL (2012). Fast gapped-read alignment with Bowtie 2. Nat Methods 9, 357–359. [DOI] [PMC free article] [PubMed] [Google Scholar]
Le TB, Imakaev MV, Mirny LA, and Laub MT (2013). High-resolution mapping of the spatial organization of a bacterial chromosome. Science 342, 731–734. [DOI] [PMC free article] [PubMed] [Google Scholar]
Le TB, and Laub MT (2016). Transcription rate and transcript length drive formation of chromosomal interaction domain boundaries. EMBO J 35, 1582–1595. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lewis PJ, Thaker SD, and Errington J (2000). Compartmentalization of transcription and translation in Bacillus subtilis. EMBO J 19, 710–718. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T, Telling A, Amit I, Lajoie BR, Sabo PJ, Dorschner MO, et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326, 289–293. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lioy VS, Cournac A, Marbouty M, Duigou S, Mozziconacci J, Espeli O, Boccard F, and Koszul R (2018). Multiscale Structuring of the E. coli Chromosome by Nucleoid-Associated and Condensin Proteins. Cell 172, 771–783 e718. [DOI] [PubMed] [Google Scholar]
Love MI, Huber W, and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550. [DOI] [PMC free article] [PubMed] [Google Scholar]
Marbouty M, Le Gall A, Cattoni DI, Cournac A, Koh A, Fiche JB, Mozziconacci J, Murray H, Koszul R, and Nollmann M (2015). Condensin- and Replication-Mediated Bacterial Chromosome Folding and Origin Condensation Revealed by Hi-C and Super-resolution Imaging. Mol Cell 59, 588–602. [DOI] [PubMed] [Google Scholar]
Matthey-Doret C, Baudry L, Breuer A, Montagne R, Guiglielmoni N, Scolari V, Jean E, Campeas A, Chanut PH, Oriol E, et al. (2020). Computer vision for pattern detection in chromosome contact maps. bioRxiv. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nora EP, Goloborodko A, Valton AL, Gibcus JH, Uebersohn A, Abdennur N, Dekker J, Mirny LA, and Bruneau BG (2017). Targeted Degradation of CTCF Decouples Local Insulation of Chromosome Domains from Genomic Compartmentalization. Cell 169, 930–944 e922. [DOI] [PMC free article] [PubMed] [Google Scholar]
Nora EP, Lajoie BR, Schulz EG, Giorgetti L, Okamoto I, Servant N, Piolot T, van Berkum NL, Meisig J, Sedat J, et al. (2012). Spatial partitioning of the regulatory landscape of the X-inactivation centre. Nature 485, 381–385. [DOI] [PMC free article] [PubMed] [Google Scholar]
Patro R, Duggal G, Love MI, Irizarry RA, and Kingsford C (2017). Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods 14, 417–419. [DOI] [PMC free article] [PubMed] [Google Scholar]
Peng N, Xia Q, Chen Z, Liang YX, and She Q (2009). An upstream activation element exerting differential transcriptional activation on an archaeal promoter. Mol Microbiol 74, 928–939. [DOI] [PubMed] [Google Scholar]
Phillips-Cremins JE, Sauria ME, Sanyal A, Gerasimova TI, Lajoie BR, Bell JS, Ong CT, Hookway TA, Guo C, Sun Y, et al. (2013). Architectural protein subclasses shape 3D organization of genomes during lineage commitment. Cell 153, 1281–1295. [DOI] [PMC free article] [PubMed] [Google Scholar]
Ramirez F, Ryan DP, Gruning B, Bhardwaj V, Kilpert F, Richter AS, Heyne S, Dundar F, and Manke T (2016). deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res 44, W160–165. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rao SS, Huntley MH, Durand NC, Stamenova EK, Bochkov ID, Robinson JT, Sanborn AL, Machol I, Omer AD, Lander ES, et al. (2014). A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell 159, 1665–1680. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rao SSP, Huang SC, Glenn St Hilaire B, Engreitz JM, Perez EM, Kieffer-Kwon KR, Sanborn AL, Johnstone SE, Bascom GD, Bochkov ID, et al. (2017). Cohesin Loss Eliminates All Loop Domains. Cell 171, 305–320 e324. [DOI] [PMC free article] [PubMed] [Google Scholar]
Rowley MJ, Nichols MH, Lyu X, Ando-Kuri M, Rivera ISM, Hermetz K, Wang P, Ruan Y, and Corces VG (2017). Evolutionarily Conserved Principles Predict 3D Chromatin Organization. Mol Cell 67, 837–852 e837. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sanborn AL, Rao SS, Huang SC, Durand NC, Huntley MH, Jewett AI, Bochkov ID, Chinnappan D, Cutkosky A, Li J, et al. (2015). Chromatin extrusion explains key features of loop and domain formation in wild-type and engineered genomes. Proc Natl Acad Sci U S A 112, E6456–6465. [DOI] [PMC free article] [PubMed] [Google Scholar]
Schwarzer W, Abdennur N, Goloborodko A, Pekowska A, Fudenberg G, Loe-Mie Y, Fonseca NA, Huber W, C HH, Mirny L, et al. (2017). Two independent modes of chromatin organization revealed by cohesin removal. Nature 551, 51–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
Servant N, Lajoie BR, Nora EP, Giorgetti L, Chen CJ, Heard E, Dekker J, and Barillot E (2012). HiTC: exploration of high-throughput ‘C’ experiments. Bioinformatics 28, 2843–2844. [DOI] [PMC free article] [PubMed] [Google Scholar]
Servant N, Varoquaux N, Lajoie BR, Viara E, Chen CJ, Vert JP, Heard E, Dekker J, and Barillot E (2015). HiC-Pro: an optimized and flexible pipeline for Hi-C data processing. Genome Biol 16, 259. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sexton T, Yaffe E, Kenigsberg E, Bantignies F, Leblanc B, Hoichman M, Parrinello H, Tanay A, and Cavalli G (2012). Three-dimensional folding and functional organization principles of the Drosophila genome. Cell 148, 458–472. [DOI] [PubMed] [Google Scholar]
Spang A, Caceres EF, and Ettema TJG (2017). Genomic exploration of the diversity, ecology, and evolution of the archaeal domain of life. Science 357. [DOI] [PubMed] [Google Scholar]
Taboada B, Estrada K, Ciria R, and Merino E (2018). Operon-mapper: a web server for precise operon identification in bacterial and archaeal genomes. Bioinformatics 34, 4118–4120. [DOI] [PMC free article] [PubMed] [Google Scholar]
Takemata N, and Bell SD (2020). Emerging views of genome organization in Archaea. J Cell Sci 133. [DOI] [PMC free article] [PubMed] [Google Scholar]
Takemata N, Samson RY, and Bell SD (2019). Physical and Functional Compartmentalization of Archaeal Chromosomes. Cell 179, 165–179 e118. [DOI] [PMC free article] [PubMed] [Google Scholar]
Trussart M, Yus E, Martinez S, Bau D, Tahara YO, Pengo T, Widjaja M, Kretschmer S, Swoger J, Djordjevic S, et al. (2017). Defined chromosome structure in the genome-reduced bacterium Mycoplasma pneumoniae. Nat Commun 8, 14665. [DOI] [PMC free article] [PubMed] [Google Scholar]
van Steensel B, and Furlong EEM (2019). The role of transcription in shaping the spatial organization of the genome. Nat Rev Mol Cell Biol 20, 327–337. [DOI] [PMC free article] [PubMed] [Google Scholar]
Wang X, Brandao HB, Le TB, Laub MT, and Rudner DZ (2017). Bacillus subtilis SMC complexes juxtapose chromosome arms as they travel from origin to terminus. Science 355, 524–527. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang C, Phillips APR, Wipfler RL, Olsen GJ, and Whitaker RJ (2018). The essential genome of the crenarchaeal model Sulfolobus islandicus. Nat Commun 9, 4908. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang C, and Whitaker RJ (2018). Microhomology-Mediated High-Throughput Gene Inactivation Strategy for the Hyperthermophilic Crenarchaeon Sulfolobus islandicus. Appl Environ Microbiol 84. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zhang C, Wipfler RL, Li Y, Wang Z, Hallett EN, and Whitaker RJ (2019). Cell Structure Changes in the Hyperthermophilic Crenarchaeon Sulfolobus islandicus Lacking the S-Layer. mBio 10. [DOI] [PMC free article] [PubMed] [Google Scholar]
Zink IA, Pfeifer K, Wimmer E, Sleytr UB, Schuster B, and Schleper C (2019). CRISPR-mediated gene silencing reveals involvement of the archaeal S-layer in cell division and virus infection. Nat Commun 10, 4797. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS1654324-supplement-1.pdf^{(46.7MB, pdf)}

Data Availability Statement

All sequencing data used in this study has been deposited to the NCBI Gene Expression Omnibus (GEO). The accession number for 3C-seq and RNA-seq data generated in this study is GEO: GSE159537.

[R1] Abdennur N, and Mirny LA (2020). Cooler: scalable storage for Hi-C data and other genomically labeled arrays. Bioinformatics 36, 311–316. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R2] Alipour E, and Marko JF (2012). Self-organization of domain structures by DNA-loop-extruding enzymes. Nucleic Acids Res 40, 11202–11212. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R3] Cabrera JE, and Jin DJ (2003). The distribution of RNA polymerase in Escherichia coli is dynamic and sensitive to environmental cues. Mol Microbiol 50, 1493–1505. [DOI] [PubMed] [Google Scholar]

[R4] Cagliero C, Grand RS, Jones MB, Jin DJ, and O’Sullivan JM (2013). Genome conformation capture reveals that the Escherichia coli chromosome is organized by replication and transcription. Nucleic Acids Res 41, 6058–6071. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] Crane E, Bian Q, McCord RP, Lajoie BR, Wheeler BS, Ralston EJ, Uzawa S, Dekker J, and Meyer BJ (2015). Condensin-driven remodelling of X chromosome topology during dosage compensation. Nature 523, 240–244. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] Davidson IF, Bauer B, Goetz D, Tang W, Wutz G, and Peters JM (2019). DNA loop extrusion by human cohesin. Science 366, 1338–1345. [DOI] [PubMed] [Google Scholar]

[R7] Deng L, Zhu H, Chen Z, Liang YX, and She Q (2009). Unmarked gene deletion and host-vector system for the hyperthermophilic crenarchaeon Sulfolobus islandicus. Extremophiles 13, 735–746. [DOI] [PubMed] [Google Scholar]

[R8] Diedenhofen B, and Musch J (2015). cocor: a comprehensive solution for the statistical comparison of correlations. PLoS One 10, e0121945. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] Dixon JR, Selvaraj S, Yue F, Kim A, Li Y, Shen Y, Hu M, Liu JS, and Ren B (2012). Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature 485, 376–380. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] Gaal T, Bratton BP, Sanchez-Vazquez P, Sliwicki A, Sliwicki K, Vegel A, Pannu R, and Gourse RL (2016). Colocalization of distant chromosomal loci in space in E. coli: a bacterial nucleolus. Genes Dev 30, 2272–2285. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R11] Gibcus JH, Samejima K, Goloborodko A, Samejima I, Naumova N, Nuebler J, Kanemaki MT, Xie L, Paulson JR, Earnshaw WC, et al. (2018). A pathway for mitotic chromosome formation. Science 359. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] Hnisz D, Shrinivas K, Young RA, Chakraborty AK, and Sharp PA (2017). A Phase Separation Model for Transcriptional Control. Cell 169, 13–23. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] Imachi H, Nobu MK, Nakahara N, Morono Y, Ogawara M, Takaki Y, Takano Y, Uematsu K, Ikuta T, Ito M, et al. (2020). Isolation of an archaeon at the prokaryote-eukaryote interface. Nature 577, 519–525. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] Kim Y, Shi Z, Zhang H, Finkelstein IJ, and Yu H (2019). Human cohesin compacts DNA by loop extrusion. Science 366, 1345–1349. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] Langmead B, and Salzberg SL (2012). Fast gapped-read alignment with Bowtie 2. Nat Methods 9, 357–359. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] Le TB, Imakaev MV, Mirny LA, and Laub MT (2013). High-resolution mapping of the spatial organization of a bacterial chromosome. Science 342, 731–734. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R17] Le TB, and Laub MT (2016). Transcription rate and transcript length drive formation of chromosomal interaction domain boundaries. EMBO J 35, 1582–1595. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] Lewis PJ, Thaker SD, and Errington J (2000). Compartmentalization of transcription and translation in Bacillus subtilis. EMBO J 19, 710–718. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] Lieberman-Aiden E, van Berkum NL, Williams L, Imakaev M, Ragoczy T, Telling A, Amit I, Lajoie BR, Sabo PJ, Dorschner MO, et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326, 289–293. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] Lioy VS, Cournac A, Marbouty M, Duigou S, Mozziconacci J, Espeli O, Boccard F, and Koszul R (2018). Multiscale Structuring of the E. coli Chromosome by Nucleoid-Associated and Condensin Proteins. Cell 172, 771–783 e718. [DOI] [PubMed] [Google Scholar]

[R21] Love MI, Huber W, and Anders S (2014). Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R22] Marbouty M, Le Gall A, Cattoni DI, Cournac A, Koh A, Fiche JB, Mozziconacci J, Murray H, Koszul R, and Nollmann M (2015). Condensin- and Replication-Mediated Bacterial Chromosome Folding and Origin Condensation Revealed by Hi-C and Super-resolution Imaging. Mol Cell 59, 588–602. [DOI] [PubMed] [Google Scholar]

[R23] Matthey-Doret C, Baudry L, Breuer A, Montagne R, Guiglielmoni N, Scolari V, Jean E, Campeas A, Chanut PH, Oriol E, et al. (2020). Computer vision for pattern detection in chromosome contact maps. bioRxiv. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] Nora EP, Goloborodko A, Valton AL, Gibcus JH, Uebersohn A, Abdennur N, Dekker J, Mirny LA, and Bruneau BG (2017). Targeted Degradation of CTCF Decouples Local Insulation of Chromosome Domains from Genomic Compartmentalization. Cell 169, 930–944 e922. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] Nora EP, Lajoie BR, Schulz EG, Giorgetti L, Okamoto I, Servant N, Piolot T, van Berkum NL, Meisig J, Sedat J, et al. (2012). Spatial partitioning of the regulatory landscape of the X-inactivation centre. Nature 485, 381–385. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] Patro R, Duggal G, Love MI, Irizarry RA, and Kingsford C (2017). Salmon provides fast and bias-aware quantification of transcript expression. Nat Methods 14, 417–419. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] Peng N, Xia Q, Chen Z, Liang YX, and She Q (2009). An upstream activation element exerting differential transcriptional activation on an archaeal promoter. Mol Microbiol 74, 928–939. [DOI] [PubMed] [Google Scholar]

[R28] Phillips-Cremins JE, Sauria ME, Sanyal A, Gerasimova TI, Lajoie BR, Bell JS, Ong CT, Hookway TA, Guo C, Sun Y, et al. (2013). Architectural protein subclasses shape 3D organization of genomes during lineage commitment. Cell 153, 1281–1295. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] Ramirez F, Ryan DP, Gruning B, Bhardwaj V, Kilpert F, Richter AS, Heyne S, Dundar F, and Manke T (2016). deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res 44, W160–165. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] Rao SS, Huntley MH, Durand NC, Stamenova EK, Bochkov ID, Robinson JT, Sanborn AL, Machol I, Omer AD, Lander ES, et al. (2014). A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell 159, 1665–1680. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] Rao SSP, Huang SC, Glenn St Hilaire B, Engreitz JM, Perez EM, Kieffer-Kwon KR, Sanborn AL, Johnstone SE, Bascom GD, Bochkov ID, et al. (2017). Cohesin Loss Eliminates All Loop Domains. Cell 171, 305–320 e324. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] Rowley MJ, Nichols MH, Lyu X, Ando-Kuri M, Rivera ISM, Hermetz K, Wang P, Ruan Y, and Corces VG (2017). Evolutionarily Conserved Principles Predict 3D Chromatin Organization. Mol Cell 67, 837–852 e837. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] Sanborn AL, Rao SS, Huang SC, Durand NC, Huntley MH, Jewett AI, Bochkov ID, Chinnappan D, Cutkosky A, Li J, et al. (2015). Chromatin extrusion explains key features of loop and domain formation in wild-type and engineered genomes. Proc Natl Acad Sci U S A 112, E6456–6465. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] Schwarzer W, Abdennur N, Goloborodko A, Pekowska A, Fudenberg G, Loe-Mie Y, Fonseca NA, Huber W, C HH, Mirny L, et al. (2017). Two independent modes of chromatin organization revealed by cohesin removal. Nature 551, 51–56. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] Servant N, Lajoie BR, Nora EP, Giorgetti L, Chen CJ, Heard E, Dekker J, and Barillot E (2012). HiTC: exploration of high-throughput ‘C’ experiments. Bioinformatics 28, 2843–2844. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R36] Servant N, Varoquaux N, Lajoie BR, Viara E, Chen CJ, Vert JP, Heard E, Dekker J, and Barillot E (2015). HiC-Pro: an optimized and flexible pipeline for Hi-C data processing. Genome Biol 16, 259. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R37] Sexton T, Yaffe E, Kenigsberg E, Bantignies F, Leblanc B, Hoichman M, Parrinello H, Tanay A, and Cavalli G (2012). Three-dimensional folding and functional organization principles of the Drosophila genome. Cell 148, 458–472. [DOI] [PubMed] [Google Scholar]

[R38] Spang A, Caceres EF, and Ettema TJG (2017). Genomic exploration of the diversity, ecology, and evolution of the archaeal domain of life. Science 357. [DOI] [PubMed] [Google Scholar]

[R39] Taboada B, Estrada K, Ciria R, and Merino E (2018). Operon-mapper: a web server for precise operon identification in bacterial and archaeal genomes. Bioinformatics 34, 4118–4120. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R40] Takemata N, and Bell SD (2020). Emerging views of genome organization in Archaea. J Cell Sci 133. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] Takemata N, Samson RY, and Bell SD (2019). Physical and Functional Compartmentalization of Archaeal Chromosomes. Cell 179, 165–179 e118. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] Trussart M, Yus E, Martinez S, Bau D, Tahara YO, Pengo T, Widjaja M, Kretschmer S, Swoger J, Djordjevic S, et al. (2017). Defined chromosome structure in the genome-reduced bacterium Mycoplasma pneumoniae. Nat Commun 8, 14665. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] van Steensel B, and Furlong EEM (2019). The role of transcription in shaping the spatial organization of the genome. Nat Rev Mol Cell Biol 20, 327–337. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] Wang X, Brandao HB, Le TB, Laub MT, and Rudner DZ (2017). Bacillus subtilis SMC complexes juxtapose chromosome arms as they travel from origin to terminus. Science 355, 524–527. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] Zhang C, Phillips APR, Wipfler RL, Olsen GJ, and Whitaker RJ (2018). The essential genome of the crenarchaeal model Sulfolobus islandicus. Nat Commun 9, 4908. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] Zhang C, and Whitaker RJ (2018). Microhomology-Mediated High-Throughput Gene Inactivation Strategy for the Hyperthermophilic Crenarchaeon Sulfolobus islandicus. Appl Environ Microbiol 84. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] Zhang C, Wipfler RL, Li Y, Wang Z, Hallett EN, and Whitaker RJ (2019). Cell Structure Changes in the Hyperthermophilic Crenarchaeon Sulfolobus islandicus Lacking the S-Layer. mBio 10. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] Zink IA, Pfeifer K, Wimmer E, Sleytr UB, Schuster B, and Schleper C (2019). CRISPR-mediated gene silencing reveals involvement of the archaeal S-layer in cell division and virus infection. Nat Commun 10, 4797. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Multi-scale architecture of archaeal chromosomes

Naomichi Takemata

Stephen D Bell

SUMMARY

Graphical Abstract

eToc

Introduction

Results

Figure 1. High-resolution 3C-seq Identifies CID-like Domains in Sulfolobus Archaea (see also Figures S1 and S2 and Table S2).

CID-like structures in Sulfolobus chromosomes

Figure 2. Correlation between CID Formation and Coalescin Distribution in the B Compartment (see also Figure S3).

The impact of inhibiting transcription on Sulfolobus CIDs

Figure 3. Global Inhibition of Transcription Affects CIDs Differently Depending on Their Compartment Identity (see also Figure S4).

Modulation of local gene expression influences CID boundaries

Figure 4. A Highly Expressed Operon slaAB Drives CID Formation.

Figure 5. Inducible Transcription of a Transgene Reinforces a Nearby Boundary.

Loop formation in Sulfolobus chromosomes

Figure 6. Identification of DNA Loops in Sulfolobus (see also Figure S5).

Figure 7. Transcription of RGs and A/B Compartments Are Potential Drivers of Loop Formation (see also Figures S6 and S7).

Discussion

Limitations

STAR★METHODS

RESOURCE AVAILABILITY

Lead Contact

Materials Availability

Data and Code Availability

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Archaeal strains

METHOD DETAILS

Strain constructions

3C-seq

RNA-seq

QUANTIFICATION AND STATISTICAL ANALYSIS

Generation of 3C-seq matrices

Generation of distance-normalized contact maps, Pearson correlation maps, and compartment index plots

Analyses of published Hi-C data for comparison with 3C-seq data

Directional preference

Insulation score analysis

Analysis of loop structure by Chromosight

Aggregate contact maps

ChIP-seq data analysis

RNA-seq data analysis

Colocalization of highly expressed genes and CID boundaries

Comparing ClsN bias with contact bias

Comparing the strength of ClsN bias and of boundary

Features associated with loop anchors

Quantification of gene expression at boundary/anchor-proximal regions and -distal regions

Data visualization

Supplementary Material

Highlights.

Acknowledgements

Footnotes

REFERENCES

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases