Chromosome architecture in an archaeal species naturally lacking structural maintenance of chromosomes proteins

Abstract

Proteins in the structural maintenance of chromosomes (SMC) superfamily play key roles in chromosome organization and are ubiquitous across all domains of life. However, SMC proteins are notably absent in the Desulfurococcales of phylum Crenarchaeota. Intrigued by this observation, we performed chromosome conformation capture experiments in the model Desulfurococcales species Aeropyrum pernix. As in other archaea, we observe chromosomal interaction domains across the chromosome. The boundaries between chromosomal interaction domains show a dependence on transcription and translation for their definition. Importantly, however, we reveal an additional higher-order, bipartite organization of the chromosome-with a small high-gene-expression and self-interacting domain that is defined by transcriptional activity and loop structures. Viewing these data in the context of the distribution of SMC superfamily proteins in the Crenarchaeota, we suggest that the organization of the Aeropyrum genome represents an evolutionary antecedent of the compartmentalized architecture observed in the Sulfolobus lineage.

Fig. 1. A. pernix primary chromosome organization.

a, MFA in the exponential phase; the red line is a moving point average. b, Core gene localization along the chromosome, for core genomes determined with different datasets resulting in different stringency levels (see Supplementary Table 1 for a dataset description). Proviruses are also indicated in grey, rRNA genes in blue and CRISPR loci in green. c, The distance to the nearest origin of replication of core and accessory genes, at the most stringent level (AAPD). A two-sided Wilcoxon test P value is indicated. d, Gene transcriptional level, expressed as RPKSP, in the exponential phase. e, Gene transcriptional level plotted in function of the distance to the nearest origin of replication, for the exponential phase. A two-sided Pearson correlation P value is indicated.

Fig. 2. A. pernix chromosome is organized into CIDs and a HEID with a higher transcriptional level.

a, Contact score heat map generated at a bin size of 3 kb. b, Heat map of the distance-normalized contact score indicating the localization of the CIDs as black triangles. c, Directional preference score used to determine the CID boundaries. Positive and negative values of directional preference are in green and orange, respectively. d, Aggregate insulation score around CID boundaries. e, Aggregate heat map around CIDs. f, Pearson correlation heat map at a bin size of 3 kb. g, The compartment index (PC1) defines the ‘high-expression insulated domain’ (HEID) and the ROC; see text for the definition of these features. h, The gene transcriptional level (RPKSP) with the HEID highlighted in orange. i, Violin plot of the transcriptional level for the HEID and ROC genes. The P value of the two-sided Wilcoxon test is indicated, and the horizontal line represents the median. j, Number of genes in the HEID, expected from a random distribution of the domains along the chromosome (grey) and observed (black), for different gene groups. An empirical P value is indicated (Methods). k, The number of core genes in the HEID, expected from a random distribution of the domains along the chromosome (grey) and observed (black), for the various core genomes determined. An empirical P value is indicated (Methods).

Fig. 3. Chromosomal loops formed between specific loci.

a, Heat map of the distance-normalized contact score at a 3 kb resolution. Loop-type interactions identified by Chromosight are indicated by circles. b, Number of loop clusters and number of loops with one or both anchors in a cluster, expected from a random distribution of loops along the chromosome and observed. An empirical P value is indicated (Methods). c, Top, the loops are represented as a curve joining the two anchors on a circular chromosome representation, for various loop types. The curve colour represents the loop score. Bottom, aggregate contact maps showing average values of distance-normalized interaction scores around the loop anchors. d, Number of loops, expected from a random distribution of loops along the chromosome (grey) and observed (black), for different gene types found at one or both loop anchors. An empirical P value is indicated (Methods). e, Correlation between the local transcriptional level at the two anchor bins of the loops. A two-sided Pearson correlation P value and coefficient are indicated. f, Loop score plotted in function of the local transcriptional level at the loop anchor. A two-sided Pearson correlation P value and coefficient are indicated. Cor, correlation coefficient.

Fig. 4. Effect of transcriptional reconfiguration (left, ActD treatment) and translation disruption (right, chloramphenicol treatment) on CIDs.

a, Contact score LFD heat maps between the treatment and the control. b,c, Contact score fold difference between treatment with actinomycin D (b) or chloramphenicol (c) and control in function of the distance between the interacting bins. Cam, chloramphenicol; EtOH, ethanol. d, RNA level (RPKSP) LFC between the treatment and the control. Colours indicate the significance of the change. For the chloramphenicol treatment, triangles indicate translation-related genes. NS, not significant. e, Heat maps of the distance-normalized contact score for the treatment and control and heat map of the contact score LFD between the treatment and the control. The heat maps are focusing on contacts between bins that are less than 150 kb apart on the chromosome. The positions of the CIDs are indicated as black triangles for the treatment and grey triangles for the control. f, Aggregate heat maps around the CIDs showing the distance-normalized score for the various treatments and controls and the LFD between treatments and controls. g,h, Aggregate insulation score around the ActD CID boundary (g) and the EtOH CID boundary (h) for the treatment (purple) and the control (green).

Fig. 5. HEID and loop changes upon transcriptional reconfiguration (ActD treatment).

a, Pearson correlation heat map for the DMSO-treated control. b, Compartment index (PC1) for the DMSO-treated control. c, Pearson correlation heat map for the ActD treatment. d, Compartment index (PC1) for the ActD treatment defining a different domain named HEID′ (orange). e, RNA levels (RPKSP) after ActD treatment. The HEID′ is highlighted in orange. f, RNA (PRKSP) LFC between the ActD and DMSO treatments. The HEID′ is highlighted in orange. g, Violin plot of the RNA level for the HEID′ and ROC genes. h, Violin plot of the RNA LFC for the HEID′ and ROC genes. i, Loops detected for the DMSO control and ActD treatments and their score. Various chromosomal structures are indicated as an outside ring: the HEID and HEID′ in orange, proviruses in grey and CRISPR loci in green. j, Aggregate heat map in DMSO and ActD conditions around the loop anchor, for various categories of loops. k, Venn diagram of A. pernix K1 and Sulfolobus acidocaldarius DSM 639 genes and their chromosomal domain location with the HEID and A compartments highlighted in orange and red for A. pernix and S. acidocaldarius, respectively. For the violin plots, the P value of the two-sided Wilcoxon test is indicated and the horizontal line represents the median.

Extended Data Fig. 1. Distribution of SMC superfamily proteins in archaeal species.

Adapted from. The positions of the Sulfolobales and Desulfurococcales (of which Aeropyrum pernix K1 is a member) are indicated in red within the yellow shaded box indicating the crenarchaea. Aeropyrum was selected as our species of choice due to the ease of growth in an aerobic environment and the availability of a high-quality, fully-closed genome sequence for the species. For a comprehensive analysis of the distribution of ClsN-like proteins in the archaea the reader is directed to ref. .

Extended Data Fig. 2. Primary chromosome organization in stationary phase Aeropyrum pernix.

a Marker Frequency Analysis in stationary phase. b. Core gene localization along the chromosome, for core genomes determined with different datasets resulting in different stringency levels (see Table S2 for dataset description). Proviruses are also indicated in grey (Mochizuki et al.), rRNA genes in blue and CRISPR loci in green. c Gene transcriptional level, expressed as RPKSP stationary phase. d. Gene transcriptional level plotted in function of the distance to the nearest origin of replication, for stationary phase. Two-sided Pearson-correlation p-value and coefficient are indicated.

Extended Data Fig. 3. Individual contact score heatmaps.

Data generated at a bin size of 3 kb for all the conditions presented in this study.

Extended Data Fig. 4. Violin plot of the GC content (%) in 3 kb bins between the HEID domain and ROC.

The p-value of two-sided Wilcoxon test is indicated and the horizontal line represents the median.

Extended Data Fig. 5. ChIPseq analysis of RAD50 binding to Aeropyrum pernix chromosome.

a. Reproducibility between the two experimental conditions used. b. RAD50 enrichment per 3 kb window along the chromosome with the HEID domain highlighted in orange. c. Violin plot of RAD50 enrichment in the HEID domain and ROC. The p-value of two-sided Wilcoxon test is indicated and the horizontal line represents the median. d. Correlation between RAD50 enrichment and the transcriptional level (RPSP) per 3 kb window. e. Correlation between RAD50 enrichment and the GC content per 3 kb window. f. Correlation between RAD50 enrichment and the distance to the origin of replication. Two-sided Pearson correlation p-value and coefficient are indicated. For A, D, E and F, Two-sided Pearson correlation p-values and estimates were indicated.

Extended Data Fig. 6. The 15 loop clusters.

A The loops of each cluster are represented as a curve joining the two anchors on a circular chromosome representation. The HEID domain in indicated in orange. For three clusters, an interesting genetic element is located at the cluster anchor and is indicated. B Number of clusters, expected from a random distribution of clusters along the chromosome (grey) and observed (black), for different genetic elements found at the cluster anchor. An empirical p-value is indicated (see material and method).

Extended Data Fig. 7. HEID domain and loop changes between untreated exponential phase sample and DMSO-treated exponential phase sample.

a. Pearson correlation heatmap for the untreated sample. b. Compartment index (PC1) for the untreated sample. The conservation of the HEID domain segments with the DMSO-treated sample is indicated in orange shades. c. Pearson correlation heatmap for the DMSO treated sample. d. Compartment index (PC1) for the DMSO treatment. The conservation of the HEID domain segments with the untreated sample is indicated in the same orange shades as in B. e. RNA (PRKSP) Log2 Fold Change (LFC) between the DMSO-treated and untreated samples. The conservation of the untreated HEID domain segments with the DMSO-treated sample is indicated in the same orange shades as in B. f. RNA (RPKSP) Log2 Fold Change (LFC) between the DMSO-treated and untreated samples. The conservation of the DMSO-treated HEID domain segments with the untreated sample is indicated in the same orange shades as in B. g. Violin plot of the RNA LFC in function of the HEID domain segment conservation between the untreated and DMSO-treated samples. The p-value of two-sided Wilcoxon test is indicated and the horizontal line represents the median. h. Aggregate heatmap in untreated and DMSO-treated conditions around the loop anchor, for various categories of loops.

Extended Data Fig. 8. HEID domain and loop changes upon translation disruption (Chloramphenicol treatment).

a. From top to bottom. Pearson correlation heatmap for the ethanol (EtOH)-treated control. Compartment index (PC1) for the ethanol-treated control. Pearson correlation heatmap for the Chloramphenicol (Cam) treatment. Compartment index (PC1) for the Chloramphenicol treatment. b. Loops detected for the ethanol control and chloramphenicol treatments and their score. Various chromosomal structures are indicated as an outside ring: the HEID domain in orange, proviruses in grey and CRIPSR loci in green. c. Aggregate heatmap in Ethanol and chloramphenicol conditions around the loop anchor, for various categories of loops.