Evidence for binary Smc complexes lacking kite subunits in archaea

SMC complexes play a central role in chromosome organization in all domains of life. The bacterial Smc–ScpAB complex is a three-subunit complex composed of Smc, ScpA and ScpB. ScpA bridges the two ATPase domains of the Smc homodimer, while ScpB, which belongs to the kite family of proteins, interacts with ScpA. The three subunits are known to be equally important for the function of Smc–ScpAB in bacteria. From crystallographic and biochemical studies, evidence is provided that six archaeal ScpA proteins are unable to interact with the only putative ScpB found in these species. Structure-based sequence alignment reveals that these archaeal ScpAs lack the ScpB-binding segment that is commonly present in the middle of bacterial ScpA sequences, which is thus responsible for their inability to interact with ScpB. ScpA proteins lacking the ScpB-binding segment are found to prevail in archaea. Moreover, two archaeal ScpA proteins with a longer middle region also failed to bind their putative ScpB partner. Furthermore, all or most species belonging to five out of 14 euryarchaeotal orders contain Smc and ScpA but not a detectable ScpB homologue. These data support the notion that archaeal Smc-based complexes generally function as a two-subunit complex composed of only Smc and ScpA.

1. Introduction

During cell division, faithful chromosome segregation and partitioning into two daughter cells relies on large protein complexes called condensins in eukaryotes or condensin-like complexes in prokaryotes (Kamada et al., 2017; Nasmyth & Haering, 2005). All condensin complexes commonly contain two subunits belonging to the family of SMC (structural maintenance of chromosomes) proteins, which form a homodimer or heterodimer. The SMC proteins contain a hinge domain and an ATPase domain that are separated by an ∼50 nm long coiled-coil arm. The hinge domain serves as the dimerization interface, and the ATPase domain undergoes an engagement and disengagement cycle. In this cycle, ATP binding induces the engagement of two ATPase domains, and subsequent ATP hydrolysis results in their disengagement. In addition to the SMC subunits, prokaryotic condensin-like complexes typically have two non-SMC subunits, whereas eukaryotic condensins have three non-SMC subunits (Uhlmann, 2016). Three different types of condensin-like complexes have been discovered in bacteria. One is the Smc–ScpAB complex, which is composed of an Smc homodimer and two non-SMC subunits, ScpA and ScpB, that form a tight binary complex (usually referred to as ScpAB). This condensin-like complex is also found in archaea (Soppa et al., 2002; Mascarenhas et al., 2002; Barillà, 2016; Kamada & Barillà, 2018). The other two are MukBEF and MksBEF, which are composed of the SMC subunit MukB (or MksB) and two non-SMC proteins called MukE and MukF (or MksE and MksF) that also form a binary complex (Niki et al., 1991; Yamanaka et al., 1996; Yamazoe et al., 1999; Petrushenko et al., 2011). Smc–ScpAB is more closely related to the eukaryotic condensins and is much more widely spread in bacteria than MukBEF and MksBEF (Cobbe & Heck, 2004). In Smc–ScpAB, ScpA is a kleisin subunit that binds two distinct interfaces of Smc: one on the coiled-coil segment close to the ATPase domain (dubbed the neck) and the other at the bottom of the ATPase domain (dubbed the cap) (Bürmann et al., 2013). Binding on the neck and on the cap are through the N-terminal α-helical domain (NαHD) and the C-terminal winged-helix domain (cWHD) of ScpA, respectively. These interactions result in an asymmetric 2:1 complex between Smc and ScpA, despite two molecules of Smc forming a symmetric homodimer (Bürmann et al., 2013).

The bacterial ScpB and MukE proteins (and the eukaryotic Nse1 and Nse3 proteins in the Smc5/6 complex) comprise tandem WHDs and are classified into a new family of proteins called kite (kleisin-interacting tandem winged-helix elements of SMC complexes) proteins (Palecek & Gruber, 2015). In the structurally characterized ScpAB complexes derived from Streptococcus pneumoniae and Geobacillus stearothermo­philus, two molecules of ScpB bind to the middle region of ScpA between the NαHD and the cWHD (Kamada et al., 2013; Bürmann et al., 2013). This middle region is a linear segment that adopts a `rope-like' shape, and thus would be flexible by itself. The binding of two ScpB molecules renders the middle segment conformationally rigid and physically separates the NαHD from the cWHD by about 40 Å, indicating that one functional role of ScpB lies in shaping the central ScpA structure.

Disruption of ScpB is known to be as detrimental as null mutation of Smc or ScpA in Bacillus subtilis (Mascarenhas et al., 2002), implying that the kite subunit is an integral component of Smc–ScpAB. However, prokaryotes which harbor Smc and ScpA apparently do not always harbor ScpB (Soppa et al., 2002). In particular, some euryarchaeotal species lack obvious ScpB homologues in their genome, suggesting the existence of an Smc-based complex that functions without a kite subunit.

We purified six pairs of archaeal ScpA and ScpB homologues and investigated their intermolecular interactions to find strong evidence for a lack of direct association. Based on the structural characterization of Thermococcus onnurineus ScpA and the putative Pyrococcus yayanosii ScpB, the lack of interaction is ascribed to the absence of a sequence that corresponds to the ScpB-binding interface in bacterial ScpAs. Genome-wide sequence analyses reveal that the middle region of archaeal ScpAs is generally significantly shorter than that of bacterial ScpAs, and that a ScpB homologue is apparently absent from a range of archaeal species. We show that the N-terminal domain of T. onnurineus ScpA (ToScpA^N), however, interacts with the Smc neck but only weakly, which has similarly been observed in a number of bacterial Smc–ScpAB complexes. Thus, a large fraction of archaeal Smc-based complexes appear to function as a binary complex between Smc and ScpA.

2. Methods

2.1. Protein and gene-locus search

The existence and gene loci of Smc, ScpA and putative ScpB proteins were searched for using a Python script (provided as supporting information) and a BLAST search. Each order in the phylum Euryarchaeota and other phyla from the archaea was searched in the NCBI Assembly database (https://www.ncbi.nlm.nih.gov/assembly/), and the reference sequence (RefSeq) data were downloaded in genomic general feature format (GFF). The existence of smc, scpA and scpB genes was determined by searching for the strings `chromosome segregation protein SMC', `segregation/condensation protein A' or `ScpA' and `segregation and condensation protein B' or `ScpB', respectively, in each GFF file. The output files created by the script were composed of three parts: Results, Analyses and Outliers. The Results part contains the taxonomic identification, locus tags and product names of the genes flanking scpA. The Analyses part contains the number of species containing smc, scpA or scpB genes, the number of species in which smc and scpA neighbor each other, the number of species in which scpA and scpB neighbor each other, and the numbers of each of the three genes that are marked as a `pseudogene'. The Outliers part contains taxonomic identifications of species in which any of the three genes is not found in the GFF file.

If any one of the three genes was not found in a GFF file, a BLAST search was performed in two steps to clarify whether the gene(s) is (are) actually missing or present but annotated with other names such as `hypothetical protein'. In the first step, T. onnurineus Smc, ScpA or ScpB was used to identify the closest homologues in each of the archaeal orders or phyla, which are listed in Supplementary Table S1. In the second step, one of these proteins was selected as a query for each BLAST search against a specific genome where one of the three genes is undetected in the string searches. In the BLAST searches, proteins exhibiting higher than 20% sequence identity with greater than 70% coverage of the query sequence were counted as Smc, ScpA or ScpB. Additionally, proteins exhibiting higher than 50% identity with greater than 25% coverage were counted as ScpA, since only the N-terminus of the query was often aligned.

Independently, an HMM search was performed against the `nr_arc_1_Oct' database using the sequence of Smc, ScpA or ScpB from T. onnurineus as a query. Sequences exhibiting E-values of less than 1.0 × 10⁻⁴, 1.0 × 10⁻⁴ or 1.0 × 10⁻¹⁰⁰ were counted as ScpA, ScpB or Smc, respectively. These E-values were chosen as sequences with a higher E-value are often annotated as an unrelated protein. The presence or absence of the three proteins was compared using the BLAST search for the source archaeal organisms in the RefSeq database.

2.2. Protein production

All of the protein constructs used in this study are listed in Supplementary Table S2. Each construct was cloned and transformed into the Escherichia coli BL21 (DE3) RIPL strain. The cells were grown in Luria–Bertani medium and the proteins were expressed at 18°C for 18 h after induction with 0.1 mM isopropyl β-D-1-thiogalactopyranoside.

For all of the proteins that contain a CPD (cysteinyl protease domain)-(His)₁₀ tag (Shen et al., 2009), the cells were harvested and sonicated in a buffer solution (buffer A) consisting of 20 mM Tris–HCl pH 7.5, 100 mM NaCl, 3 mM β-mercaptoethanol (β-ME) and the supernatant was applied onto a column filled with HisPur cobalt resin (Thermo Scientific). The resin was washed with buffer A containing an additional 10 mM imidazole. The CPD-(His)₁₀ tag was removed by on-gel digestion with 0.1 mM phytate, which activates CPD. The proteins were further purified using a HiTrap Q anion-exchange column (GE Healthcare) and a HiLoad 26/60 Superdex 75 gel-filtration column (GE Healthcare) in buffer A.

For the purification of proteins containing a GFP-(His)₁₀, (His)₁₀-GST or (His)₁₀-MBP tag, the cell lysates were obtained in buffer A and the supernatant was applied onto the cobalt resin. The proteins were eluted from the resin using buffer A containing an additional 150 mM imidazole. If necessary, the eluted proteins were treated with Tobacco etch virus protease at 4°C for 4 h to cleave the tag, dialyzed against buffer A and applied onto the cobalt resin again to remove the tag. The proteins were further purified using a HiTrap Q anion-exchange column and a HiLoad 26/60 Superdex 75 gel-filtration column in buffer A. All of the purified proteins in buffer A were concentrated, flash-frozen in liquid nitrogen and stored at −80°C.

For co-expression experiments, vectors encoding P. yayanosii ScpA (PyScpA) with an N-terminal (His)₁₀-MBP tag and PYCH_12850 (the putative P. yayanosii ScpB) without a tag were introduced into the E. coli BL21 (DE3) RIPL strain. The proteins were expressed as described above. The cell lysate was obtained in buffer A and the supernatant was applied onto the cobalt resin. The proteins were eluted from the resin using buffer A containing an additional 150 mM imidazole. The eluents were subjected to denaturing polyacrylamide gel electrophoresis (PAGE) to determine whether the two proteins eluted together or separately.

2.3. Crystallization, data collection and structure determination

Selenomethionine (SeMet)-substituted ToScpA^N and PYCH_12850 were expressed in the E. coli B834 (DE3) RIL strain (Novagen) and purified as described above. SeMet-substituted ToScpA^N (at 4.5 mg ml⁻¹ in buffer A) was crystallized using a solution consisting of 3.5 M sodium formate, 0.1 M sodium malonate. The crystals were grown at 22°C in sitting drops consisting of 1.5 µl protein solution and 1.5 µl reservoir solution. The crystals were dehydrated by adding 30 µl reservoir solution consisting of 3.5 M sodium formate, 0.1 M sodium malonate, 10% glycerol to the crystal-containing drops followed by air exposure for 2 h at 22°C (Heras & Martin, 2005). SeMet-substituted PYCH_12850 (82.1 mg ml⁻¹ in buffer A) was crystallized in a solution consisting of 35%(v/v) pentaerythritol propoxylate (5/4 PO/OH, Hampton Research), 0.1 M MES pH 5.5, 0.4 M sodium chloride. The crystals were grown at 22°C in hanging drops consisting of 1.5 µl protein solution and 1.5 µl reservoir solution. Single-wavelength anomalous dispersion (SAD) data sets for ToScpA^N and PYCH_12850 were collected on beamlines 5C and 11C at Pohang Accelerator Laboratory, Republic of Korea. The SAD data set for ToScpA^N was processed using the HKL-2000 suite (Otwinowski & Minor, 1997). Four SAD data sets for PYCH_12850 were integrated using MOSFLM (Battye et al., 2011) and were merged and scaled using BLEND (Foadi et al., 2013; Evans & Murshudov, 2013; Evans, 2011), which are included in the CCP4 suite (Winn et al., 2011). Phasing and initial model building were performed using AutoSol in Phenix (Terwilliger et al., 2009). Manual model building and structure refinement were performed using Coot (Emsley et al., 2010), CNS (Brünger et al., 1998) and Phenix (Afonine et al., 2012; Liebschner et al., 2019). Crystallographic data statistics for the two SeMet-substituted proteins are summarized in Table 1.

Table 1
X-ray data-collection and structure-refinement statistics

Values in parentheses are for the highest resolution shell.

Crystal	ToScpA^N (SeMet)	PYCH_12850 (SeMet)
Data collection
Space group	P61	P4122
a, b, c (Å)	77.073, 77.073, 39.942	92.300, 92.300, 111.817
α, β, γ (°)	90, 90, 120	90, 90, 90
Wavelength (Å)	0.9794	0.97941
Resolution (Å)	50–2.5 (2.54–2.50)	92.3–3.04 (3.25–3.04)
Rmerge (%)	9.1 (28.8)	16.3 (38.1)
〈I/σ(I)〉	36.9 (4.2)	34.3 (20.2)
Completeness (%)	99.3 (97.9)	100 (100)
Multiplicity	12.5 (5.4)	113.1 (118.9)†
Refinement
Resolution (Å)	38.5–2.50	92.3–3.04
Total No. of reflections	180880	3148552
No. of unique reflections	8905	9775
Rwork/Rfree (%)	21.5/26.0	23.6/26.5
R.m.s. deviations
Bond lengths Å)	0.011	0.002
Angles (°)	1.238	0.495
Average B value (Å^2)	40.17	58.32
Ramachandran plot (%)
Favored	92.0	94.7
Additionally allowed	8.0	5.3
Generously allowed	0	0
PDB code	6ivh	6juv

†The unusually high value is owing to the merging of four data sets that are dissimilar in the data completeness in resolution shells.

2.4. Pull-down assay

For (His)₁₀ pull-down assays, (His)₁₀-MBP-tagged ScpAs and untagged putative ScpBs were used. The pairs of proteins (1 µM each) were incubated in 100 µl buffer solution consisting of 20 mM Tris–HCl pH 7.5, 100 mM NaCl at 25°C for 10 min. The reaction mixtures were then incubated with 15 µl Ni–NTA resin (Thermo). The resin was washed three times with buffer consisting of 20 mM Tris–HCl pH 7.5, 100 mM NaCl, 30 mM imidazole and subjected to denaturing PAGE to determine whether untagged putative ScpB was retained on the resin together with tagged ScpA.

2.5. Size-exclusion chromatography

Size-exclusion chromatography was performed using a Superdex 200 Increase 10/300 GL column (GE Healthcare) in a running buffer consisting of 20 mM Tris–HCl pH 7.5, 100 mM NaCl, 1 mM DTT at 4°C. Each protein or protein mixture (10 µM each in 100 µl) was injected into the column.

2.6. Bismaleimidoethane (BMOE)-mediated cysteine cross-linking

Four constructs (wild type, Q185C, Q994C and A1110C) of the T. onnurineus Smc (ToSmc) head domain with an 80-residue coiled-coil stretch (ToSmcHd-CC80), which were co-purified with ToScpA(E69C)-GFP-(His)₁₀, were dialyzed in a buffer solution consisting of 20 mM Tris–HCl pH 7.5, 100 mM NaCl. BMOE (Thermo Scientific) was then added to a final concentration of 200 µM to each mixture at 2 µM. The reaction mixtures were incubated for 10 min at 25°C, quenched by adding β-ME (14 mM) and subjected to denaturing PAGE. The protein bands were visualized by both GFP signal and Coomassie Blue staining.

3. Results

3.1. Absence or incomplete presence of smc, scpA and scpB in archaeal genomes

We performed a genome-wide search for the presence of smc, scpA and scpB and their co-occurrence in archaeal organisms using a Python script (see Section 2), the STRING database (https://string-db.org/; Szklarczyk et al., 2015), BLAST (Mount, 2007) and HMM (Zimmermann et al., 2018) searches. The BLAST search was restricted to organisms for which the species has been identified and the genome sequence is registered in the NCBI RefSeq database (a total of 640 genomes), and the HMM search was compared for these organisms. Our search did not detect smc, scpA and scpB genes in the archaeal phylum Crenarchaeota and in the classes Methanobacteria and Methanopyri of the phylum Euryarchaeota [Fig. 1(b) and Table 2], suggesting that these organisms are unlikely to encode an Smc-based complex. Notably, Sulfolobus species belonging to the Crenarchaeota branch have recently been found to utilize a novel SMC-like protein called coalescin for chromosome organization (Takemata et al., 2019). Members of the other branches of Euryarchaeota have both smc and scpA. These two genes are next to each other on the genome in most of these euryarchaeotal species (433 out of 469 species) and in the species from Korarchaeota (three species). The two genes, however, are separated from each other in all Thaumarchaeota species (14 species), in all of the species belonging to the order Methanococcales (22 species), in three out of 62 species belonging to the order Methanosarcinales and in 11 out of 44 species in the order Thermococcales [Fig. 1(b) and Table 2].

Table 2
Occurrence and juxtaposition of smc, scpA and putative scpB genes in archaea

			No. of species containing (BLAST)			No. of species containing (HMMER)			No. of species containing
Phylum	Class	Order†	smc	scpA	scpB	Smc	ScpA	ScpB	smc-scpA	scpA-scpB
Euryarchaeota	Methanomicrobia	Methanocellales (3)	3	3	3	3	3	3	3	3
		Methanosarcinales (62)	62	62	62	61‡	62	62	59	59
		Methanomicrobiales (28)	28	28	3	26	27‡	3	28	3
	Halobacteria	Halobacteriales (92)	92	92	0	92	91‡	0	92	0
		Haloferacales (129)	126	126	0	124‡	126	0	126	0
		Natriabales (64)	64	64	0	64	64	0	64	0
	Methanococci	Methanococcales (22)	22	21	2	22	21	2	0	0
	Archaeoglobi	Archaeoglobales (9)	9	9	9	9	9	9	9	8
	Thermoplasmata	Thermoplasmatales (12)	12	12	12	12	12	12	12	1
		Methanomassiliicoccales (6)	6	6	6	5‡	5‡	6	6	6
	Thermococci	Thermococcales (44)	44	44	44	44	44	44	33	0
	Methanobacteria	Methanobacteriales (72)	0	0	0	0	0	0	0	0
	Methanonatronarchaeia	Methanonatronarchaeales (1)	1	1	1	1	1	1	1	1
	Methanopyri	Methanopyrales (1)	0	0	0	0	0	0	0	0
Crenarchaeota	Thermoprotei	(4 orders, 77 sp.)	0	0	0	0	0	0	0	0
Korarchaeota	(No class)	(No order, 3 sp.)	3	3	3	3	3	3	3	3
Micrarchaeota	(No class)	(No order, 1 sp.)	1	1	1	1	1	1	0	1
Thaumarchaeota	Nitrososphaeria	Nitrosocaldales (1)	1	1	1	1	1	1	0	1
		Nitrososphaerales (2)	2	2	2	2	2	2	0	2
	(No class)	Nitrosopumilales (11)	11	11	11	9‡	11	11	0	11

†Values in parentheses are the total number of species in the RefSeq assembly database.
‡One or two fewer counts in each order compared with the count by the BLAST search because each gene is annotated as a pseudogene.

Figure 1 The co-occurrence of smc, scpA and scpB across prokaryotes. (a) A schematic drawing of the bacterial Smc–ScpAB complex is shown on the left. (b) Phyla, orders and selected species are shown and color-coded. All of the orders in Archaea are shown if at least one RefSeq assembly is available. Species are shown if they were the sources of the proteins used in this study (underlined), if the genomic loci of smc, scpA and scpB vary in the same order and if an order is undefined.

An obvious scpB gene is not always found on the archaeal genomes where smc and scpA are found: it was not detected in any of the species belonging to the class Halobacteria (285 species) and most of the species belonging to the orders Methanomicrobiales (25 out of 28 species) and Methanococcales (20 out of 22 species) [Fig. 1(b) and Table 2]. In terms of number of species, 480 out of the total of 640 analyzed archaeal genomes do not contain a readily detectable scpB homologue. The result of the HMM search was basically identical to that of the BLAST search, uncovering no additional credible homologues (Table 2). Of note, in the orders Methanococcales, Thermoplasmatales and Thermococcales, where scpB is found, this gene is remotely located from scpA in all of the member species, in contrast to the juxtaposition of scpA and scpB in most bacterial genomes, as noted previously (Kamada & Barillà, 2018). In a total of 61 out of 160 archaeal species containing both scpA and scpB, scpB is remotely located from scpA in the genome, and this is observed in three out of nine euryarchaeotal orders [Fig. 1(b) and Table 2].

The absence of scpB in a range of euryarchaeotal orders indicates that the Smc-based complex in these organisms does not require the kite subunit ScpB. The separated positioning of scpB from scpA in many Euryarchaeota species highlights the possibility that at least some genes annotated as scpB in these organisms may encode functionally unrelated paralogues, rather than true orthologues, of ScpB.

3.2. Absence of interaction between ScpA and ScpB derived from euryarchaeal branches

In the order Thermococcales, to which the genera Pyrococcus and Themococcus belong, smc and scpA are present according to our genomic analysis (Table 2). Three quarters of the analyzed gene pairs are next to each other, while the rest of them are found to be separated from each other by two intervening genes. Remotely from smc and scpA, scpB appears to also be present in these organisms [Fig. 1(b)]. In the P. yayanosii proteome, PYCH_01210 and PYCH_12850 are the sole plausible homologues of ScpA and ScpB, respectively. We first tested the potential interaction between PYCH_01210 (denoted PyScpA) and PYCH_12850. (His)₁₀-MBP-tagged PyScpA and PYCH_12850 without a tag were purified separately, and the two proteins were subjected to a (His)₁₀ pull-down assay in the presence of untagged P. yayanosii Smc (PySmc) head domain with an 80-residue coiled-coil stretch (PySmcHd-CC80). The three proteins were incubated together, loaded onto immobilized Ni²⁺ resin and resin-bound proteins were visualized on a denaturing polyacrylamide gel. In this experiment untagged PYCH_12850 was not observed in the eluate, indicating that the protein did not interact with (His)₁₀-MBP-tagged PyScpA and thus was not retained on the resin [Fig. 2(a)]. In contrast, a protein band corresponding to PySmcHd-CC80 was clearly observed, which is consistent with the known interactions between ScpA and the Smc head domain [Fig. 2(a)]. The mixture of (His)₁₀-MBP-tagged PyScpA and PYCH_12850 was subjected to size-exclusion chromatography. No elution peak corresponding to complex formation was observed, also indicating that the proteins do not interact with each other. Co-expression of the two proteins in E. coli also did not result in complex formation between the two proteins (not shown), which is in sharp contrast to bacterial ScpA–ScpB pairs, which can readily be co-purified (Kamada et al., 2017; Bürmann et al., 2013).

Figure 2 Pull-down assay and size-exclusion chromatography for pairs of archaeal ScpA and ScpB homologues. The name of the organism from which the two proteins were derived is shown at the top of each panel. Denaturing gels and chromatograms are shown on the left and right, respectively. (His)10-MBP-tagged ScpA proteins were mixed with the indicated ScpB homologues (TON_1955, PYCH_12850, MCON_2432 and Mzhil_1996) in a 1:1 molar ratio (1 µM each). `I' and `R' stand for the input proteins loaded onto the Ni–NTA resin and the resin-bound proteins, respectively. Molecular weights are labelled in kDa. For chromatography, individual proteins or their mixture (100 µg each) were loaded onto an analytical gel-filtration column. The two black triangles at the top of the chromatograms indicate the elution positions of the molecular-weight markers: 158 kDa (left) and 43 kDa (right).

Likewise, ScpA from T. onnurineus (TON_1071; denoted ToScpA) and the putative ScpB from this organism (TON_1955) did not exhibit a detectable interaction with each other [Fig. 2(b)]. The data obtained for the purified recombinant proteins are consistent with the fractionation of the native proteome of P. furiosus, showing that peptides derived from PF1842 (ScpA) and PF1843 (Smc) were detected in the same fraction, while those of PF2021 (ScpB) were detected in different fractions (Menon et al., 2009; F. Poole, personal communication).

We also tested the potential interaction between ScpA and the putative ScpB protein from Methanosalsum zhilinae and Methanothrix soehngenii, which are evolutionarily distant, from the order Thermococcales [Fig. 1(b)]. Similar (His)₁₀ pull-down assays exhibited no significant interaction between these two pairs [Figs. 2(c) and 2(d)]. Of note, the scpA and scpB genes of M. zhilinae are direct neighbors, like those in bacterial genomes, while those of M. soehngenii are distantly located in the genome [Fig. 1(b)]. Thus, the lack of interaction is unrelated to the genomic loci of scpA and scpB.

3.3. Structure of the ScpB homologue from P. yayanosii

To understand the structural features that may hinder interaction between archaeal ScpA proteins and putative archaeal ScpB proteins, we next determined the crystal structure of full-length PYCH_12850 at 3.0 Å resolution [Fig. 3(a)]. The protein is composed of N- and C-terminal WHDs (nWHD and cWHD, respectively) that are connected by a long intervening loop. This loop appears to be flexible, because the two molecules of PYCH_12850 in the asymmetric units show that the orientation of the cWHD relative to the nWHD in one molecule is quite different from that in the other molecule [Fig. 3(a)]. PYCH_12850 forms a homodimer through the interaction between the nWHDs [Fig. 3(a)]. These structural features have been commonly observed in the structures of bacterial ScpB proteins (Kamada et al., 2013; Bürmann et al., 2013). Structural superposition of S. pneumoniae ScpB (bound to ScpA; Bürmann et al., 2013) and PYCH_12850 showed that the N-terminal domains of the two structures are similar to each other and so are the C-terminal domains [Fig. 3(b)], as expected from the ∼48% sequence similarity between the two proteins. We could not find a particular structural feature that may prevent the archaeal sequence homologue of ScpB from interacting with ScpA, indicating that features of archaeal ScpA may prevent interaction with ScpB.

Figure 3 Structures of the ScpB homologue PYCH_12850 and ToScpA^N. (a, b) Structural features of PYCH_12850. (a) Two independent dimers in the crystal (left). The protein forms a homodimer through the N-terminal WHD (nWHD), which is followed by a flexible linker segment and a dangling cWHD (middle and right). The relative positions of the two domains differ in the two dimers. (b) Superposition of PYCH_12850 onto S. pneumoniae ScpB. The nWHDs and cWHDs are separately superposed. The nWHD–nWHD interaction as well as the structures are closely similar. (c)–(f) Structural features of ToScpA^N. (c) Overall structure. The structure is composed of four stacked α-helices. Residues 75–94 are disordered and indicated by a dotted line. Shown below is a representation of full-length ToScpA. The C-terminal region forms the cWHD. (d) Structural alignment onto SpScpA^ΔC. The α-helices are labeled in their order of appearance in the structures. The structures are closely similar but only up to α3. (e) Sequence alignment of ToScpA^N and SpScpA^ΔC. Reflecting the structural difference shown in (d), the sequence between α3 and α4 of ToScpA^N is notably different from that between α3 and α5 of SpScpA^ΔC. (f) Intramolecular hydrophobic interaction of α4 with the rest of the protein. The interacting hydrophobic residues on α4 are shown as sticks and the rest of the protein is shown as an electrostatic surface potential map.

3.4. Structure of an N-terminal fragment of T. onnurineus ScpA

We next determined the crystal structure of a T. onnurineus ScpA fragment composed of residues 1–126, referred to as ToScpA^N, at 2.5 Å resolution. This construct lacks the cWHD and the preceding linker segment (residues 127–220). ToScpA^N folds into an all-α-helical tertiary structure containing four α-helices and connecting loops [Fig. 3(c)]. In comparison with the structure of S. pneumoniae ScpA lacking the cWHD (SpScpA^ΔC) in complex with ScpB (SpScpB; Bürmann et al., 2013), ToScpA^N is similar to SpScpA^ΔC up to the third α-helix [Fig. 3(d)]. In contrast, the fourth α-helix (α4) of ToScpA^N and the preceding loop appear to be unrelated to any part of SpScpA^ΔC. According to a multiple sequence alignment using HHpred (Söding et al., 2005), α4 of ToScpA^N (residues 107–121) corresponds to α5 of SpScpA^ΔC (residues 137–151) [Fig. 3(e)]. In the structural alignment, these two α-helices are far from each other, indicating that the fourth α-helix of ToScpA^N has to be detached from the α1–α3 bundle in order to occupy the same spatial position as α5 of SpScpA^ΔC. Such a separation is unlikely to happen because α4 is tightly packed against α1–α3 via hydrophobic interactions [Fig. 3(f)] and, consistently, a construct lacking α4 was expressed as an insoluble form in E. coli (not shown). Intriguingly, however, the possibility of α4 detachment in the functional cycle of the holo complex cannot be ruled out. The structure thus highlights the different organization in the middle region of a bacterial and an archaeal ScpA protein.

3.5. Archaeal ScpAs generally lack the ScpB-binding interface

A multiple sequence alignment of ScpAs from phylo­genetically remote archaeal species shows that archaeal ScpAs are homologous to bacterial ScpAs in the N- and C-terminal regions (Fig. 4). In contrast, the middle region in the archaeal ScpAs, corresponding to the segment between α3 and α4 in the ToScpA^N structure, is clearly different from that in the bacterial ScpAs, corresponding to the segment between α3 and α5 in the G. stearothermophilus ScpA^ΔC (GsScpA^ΔC) structure (Kamada et al., 2013). This middle region in most archaeal species is notably shorter than that in bacterial ScpAs and exhibits virtually no significant sequence homology throughout archaeal species (black boxes in Fig. 4 and Supplementary Figs. S1 and S2). In the archaeal species shown in the alignment, the number of amino acids in the middle region in archaeal ScpAs varies from 30 to 66, while that in bacterial ScpAs varies from 61 to 72 (black box in Fig. 4; Table 3). Critically, the crystal structures of ScpA^ΔC–ScpB complexes (PDB entries 4i98 and 3w6j) show that this middle region in bacterial ScpA contains the major ScpB-binding interface in the second half of α4 (Kamada et al., 2013; Bürmann et al., 2013). These observations provide a plausible explanation for why ScpAs derived from the four archaeal species failed to interact with the putative ScpB homologues (Fig. 2).

Table 3
The number of amino acids in the middle region of archaeal ScpAs

Phylum	Class	Order	Length	Reference species†
Euryarchaeota	Methano­microbia	Methanocellales	36	Methanocella arvoryzae
		Methanosarcinales	39	Methanosarcina barkeri
		Methanomicrobiales	49	Methanofollis liminatans
	Methano­cocci	Methanococcales	47	Methanotorris formicicus
	Archaeoglobi	Archaeoglobales	66	Geoglobus acetivorans
	Thermoplasmata	Thermoplasmatales	37	Thermoplasma acidophilum
		Methanomassiliicoccales	40	Candidatus Methanoplasma termitum
	Thermococci	Thermococcales	31	Thermococcus onnurineus
Korarchaeota	(No class)	(No order)	30	Candidatus Methanodesulfokores washburnensis
Thaumarchaeota	Nitrososphaera	Nitrosocaldales	40	Candidatus Mancarchaeum acidiphilum
		Nitrososphaerales	37	Nitrososphaera viennensis
	(No class)	Nitrosopumilales	33	Candidatus Nitrosopumilus sediminis

†These species were used to count the number of amino acids.

Figure 4 Multiple sequence alignment of ScpAs from archaea and bacteria. ScpA sequences derived from nine archaeal species and four bacterial species were aligned using Clustal Omega (Sievers et al., 2011). Secondary-structural elements of ToScpA^N and GsScpA^ΔC are indicated at the top and the bottom of the alignment, respectively. The black box indicates the region of ScpA^N that shows a clear difference between archaeal and bacterial ScpA proteins. T.onnurin, Thermococcus onnurineus (gi:212009177); M.arvoryz, Methanocella arvoryzae (gi:500971514); M.barkeri, Methanosarcina barkeri (gi:805410469); M.zhilina, Methanosalsum zhilinae (gi:335931289); M.liminat, Methanofollis liminatans (gi:395441815); M.formici, Methanotorris formicicus (gi:373561535); G.acetivo, Geoglobus acetivorans (gi:728876205); T.acidoph, Thermoplasma acidophilum (gi:851298378); M.termitu, Candidatus Methanoplasma termitum (gi 851220379); M.washbur, Candidatus Methanodesulfokores washburnensis (gi:1538762872); M.acidiph, Candidatus Mancarchaeum acidiphilum (gi:1214173397); N.viennen, Nitrososphaera viennensis (gi:647811207); N.sedimin, Candidatus Nitrosopumilus sediminis (gi:407047581); R.qingshe, Rhodococcus qingshengii (gi:1595910292); P.aerugin, Pseudomonas aeruginosa (gi:1440714951); T.roseus, Terriglobus roseus (gi:1124365071); G.stearot, Geobacillus stearothermophilus (gi:1017231538). For expanded and grouped sequence alignments, see Supplementary Figs. S1 and S2.

Some archaeal ScpAs have a longer middle region. For example, Geoglobus acetivorans ScpA (GaScpA) and Methano­follis liminatans ScpA (MlScpA) have 66- and 49-residue-long middle regions, respectively (bold letters in Fig. 4). While the two proteins do not share sequence homology with the middle region of bacterial ScpAs (Fig. 4 and Supplementary Fig. S1), we tested whether they may interact with the sole ScpB homologue GACE_1479 in G. acetivorans or Metli_0606 in M. liminatans. In both (His)₁₀ pull-down and size-exclusion chromatographic analyses, the ScpA and ScpB proteins derived from these two species did not interact with each other, indicating that their longer middle region does not support ScpB binding (Fig. 5).

Figure 5 Protein-binding assay for archaeal ScpAs with a longer middle region. The indicated ScpA and ScpB homologue were mixed together in a 1:1 molar ratio (1 µM each), loaded onto the resin and analyzed by denaturing gel electrophoresis (left) or on an analytical size-exclusion column (right). The name of the source organism is shown at the top. Molecular weights are labelled in kDa for the gels. The two black arrows indicate the elution positions of the molecular-weight markers. GACE_1479 (174 amino acids) appears as a monomeric protein and Metli_0606 (161 amino acids) as a smaller dimeric protein in comparison with the ScpB homologues in Fig. 2 (192–364 amino acids).

Together, these analyses demonstrate that archaeal ScpAs generally lack the ScpB-binding sequence found in bacterial ScpAs and thus they are unable to interact with ScpB.

3.6. Archaeal ScpA interacts with the Smc neck

The N- and C-terminal regions of archaeal ScpAs are conserved throughout archaea and exhibit high sequence homology to those of bacterial ScpAs. Previously, the cWHD of P. furiosus ScpA was shown to interact tightly with the head domain of P. furiosus Smc (Diebold-Durand et al., 2017; Bürmann et al., 2013), as similarly observed for the bacterial counterparts (Kamada et al., 2017; Diebold-Durand et al., 2017). We asked whether the N-terminal domain of archaeal ScpA interacts with the Smc neck, a head-proximal region of the coiled coil, as was observed for bacterial ScpA (Bürmann et al., 2013) and similarly for the kleisin subunit of yeast cohesin (Gligoris et al., 2014) and condensin (Hassler et al., 2019). ToScpA^N with a (His)₁₀-GST tag and a ToSmc head domain with an 80-residue coiled-coil stretch (ToSmcHd-CC80) were purified and subjected to a (His)₁₀ pull-down assay. For a control experiment, (His)₁₀-GST-tagged B. subtilis ScpA^N (BsScpA^N) and B. subtilis Smc (BsSmc) head domain with a 30-residue coiled-coil stretch (BsSmcHd-CC30) were purified. A fraction of ToSmcHd-CC80 was pulled down by (His)₁₀-GST-ToScpA^N, indicating an interaction, albeit weak, between the two proteins [Fig. 6(a)]. In comparison, a more robust pull-down of BsSmcHd-CC30 was observed [Fig. 6(a)], which is consistent with the high-affinity interaction observed with ScpA^N and SmcHd-CC30 from G. stearo­thermophilus (Kamada et al., 2017). We next probed whether ToScpA^N also interacts with the neck of ToSmc by thiol-specific bismaleimidoethane (BMOE) cross-linking. We prepared three mutant pairs of ToScpA and ToSmcHd-CC80, both of which contained a single cysteine substitution, based on a structure-based sequence alignment of ToScpA with BsScpA and of ToSmcHd-CC80 with BsSmcHd-CC30 [Fig. 6(b)]. The three mutant pairs contain a common E69C mutation on ToScpA and either a Q185C, a Q994C or an A1110C mutation on ToSmcHd-CC80. The Q185C and Q994C mutations are located on the Smc neck, whereas the A1110C mutation is located in the head domain [Fig. 6(c)]. In the ScpA^N–Smc neck interface of the aligned structure, the E69C–Q185C pair (inter-C^α distance of 7.6 Å) appeared to be cross-linkable, while the E69C–Q185C pair (inter-C^α distance of 13.0 Å) did not, considering the length of BMOE. The E69C–A1110C pair (inter-C^α distance of 68.9 Å) was selected as a negative control. Cross-linking of these mutant pairs by BMOE resulted in one outstanding and several weak cross-linked protein bands on a denaturing polyacrylamide gel [Fig. 6(d)]. The use of (His)₁₀-GFP-fused ToScpA [ToScpA-GFP-(His)₁₀] was important for identification of the cross-linked bands, which was further confirmed by mass spectrometry (Supplementary Table S3). The most slowly migrating bands [band 1 in Fig. 6(d)] and fast migrating bands [band 3 in Fig. 6(d)] were identified as cross-linked species between head domains (ToSmcHd–ToSmcHd) and between ScpA proteins (ToScpA–ToScpA), respectively, which are likely to arise from random encounters of the exposed cysteine residues. Cross-linking between ToScpA-GFP-(His)₁₀ and ToSmcHd-CC80 [band 2 in Fig. 6(d); ToSmcHd–ToScpA] was observed only in the ToScpA(E69C)–ToSmcHd(Q185C) pair, which is consistent with the distance (7.6 Å) between the two cysteine positions in the structural alignment. These analyses together suggest that the interaction between ScpA^N and the Smc neck is conserved in T. onnurineus and probably in other archaeal species, although this interaction appears to be weaker than that between ScpA^N and the Smc neck in bacterial species.

Figure 6 Interaction between ScpA^N and the Smc neck derived from T. onnurineus. (a) Pull-down assays using the indicated proteins and Ni–NTA resin. (His)10-GST-ToScpA^N and (His)10-GST-BsScpA^N were incubated with ToSmcHd-CC80 and BsSmcHd-CC30, respectively, in a 1:1 molar ratio (20 µM) and were applied to Ni–NTA resin. Proteins retained on the resin after washing with 30 mM imidazole solution were visualized on an SDS–polyacrylamide gel. (b) Modeling the interaction between ToScpA^N and the ToSmc neck. ToScpA^N was structurally aligned with BsScpA^N bound to BsSmcHd-CC30 (PDB entry 3zgx; Bürmann et al., 2013). Sequence alignments were performed using Clustal Omega (Sievers et al., 2011). The black arrows indicate the residues involved in interaction at the binding interface between BsScpA^N and BsSmcHd-CC30. Manual adjustment according to the structural alignment was unnecessary. (c) Cysteine positions introduced for BMOE cross-linking experiment. Based on the model of ToScpA–ToSmcHd-CC80 (left), Glu69 on ToScpA and Gln185, Gln994 and Ala1110 on ToSmcHd-CC80 were selected as the mutation sites (right three panels; cyan sticks). (d) BMOE cross-linking. The cross-linked bands are labeled from 1 to 3, and the identity of these bands were deduced based on the GFP signal (right) and mass spectrometry (Supplementary Table S3).

4. Discussion

In this study, we chose and purified six pairs of archaeal ScpA and ScpB homologues that reflect the variabilities in the genomic locations of scpA and scpB and in the length of the middle region in ScpA. In the source archaeal organisms, ScpB is the sole homologue of bacterial ScpB. None of the pairs exhibited a physical interaction between ScpA and the ScpB homologue. In these organisms, Smc and ScpA are likely to form a binary complex, which can be designated `Smc–ScpA', lacking the kite subunit ScpB.

4.1. Prevalence of the kite-less Smc-based complex in archaea

In total, only 160 out of 640 analyzed archaeal species contained both ScpA and ScpB. Of these, 148 species contain ScpAs with a middle region that is clearly shorter (30–47 residues) than that of bacterial ScpAs (61–72 residues) and they thus lack the polypeptide segment required for ScpB binding in bacteria. This segment is fairly hydrophobic and is conserved among bacterial ScpAs. The remaining 12 species, belonging to the orders Archaeoglobales or Methanomicrobiales, contain ScpAs with a middle region that is somewhat longer (49–66 residues) but lack meaningful sequence homology with the ScpB-binding segment found in bacterial ScpAs. Remarkably, all six tested archaeal ScpAs failed to interact with the sole putative ScpB partner, regardless of the genomic positions of the scpA and scpB genes or the length of the middle region in ScpA (Figs. 2 and 5). Our extensive experiments thus failed to identify a single pair of interacting ScpA and ScpB proteins. Therefore, archaeal ScpAs in general are unlikely to form a ternary complex with Smc and ScpB, and the kite-less Smc–ScpA complex appears to be prevalent in the archaeal domain of life.

4.2. What is the role of the kite subunit?

The ScpB subunit in the bacterial Smc–ScpAB complex is known to be as important as Smc and ScpA in supporting normal cell growth and is essential for the recruitment of Smc to the chromosome (Minnen et al., 2016). In GsSmc–ScpAB, the kite subunit ScpB was shown to negatively regulate the interaction of ScpA^N with the Smc neck via steric hindrance (Kamada et al., 2017). ScpB bound to ScpA clashes with the Smc head if ScpA^N (within the ScpAB subcomplex) were to simultaneously bind to the Smc neck, and this steric hindrance prevents ScpA^N from binding to the Smc neck. However, in contrast to this observation in vitro, cross-linking experiments showed that ScpA^N within the ScpAB subcomplex is allowed to bind to the Smc neck in vivo, conceivably as a result of a structural rearrangement in ScpA^N that removes the steric clash (Kamada et al., 2017). It is unknown how such a structural rearrangement could take place and why negative regulation by ScpB is required for the molecular mechanism of Smc–ScpAB. It may be needed for the assembly of asymmetric Smc–ScpA rings rather than Smc dimers with ScpA bound to only one of the two Smc monomers. If so, then archaeal Smc dimers may be able to assemble asymmetric rings by other means or they may form a mixture of both variants with only the asymmetric ring form being functional. The kite subunits of the Smc5/6 complex have been implicated in DNA binding (Zabrady et al., 2016). It is unclear whether ScpB has DNA-binding capabilities. Regardless, our results imply that DNA binding by ScpB is not essential in the archaeal Smc-based complex since it can function in the absence of kite subunits.

Thus, what would the function be of the ScpB sequence homologues that are found in archaeal species? A homology search against the PDB using the DALI server (Holm & Laakso, 2016) showed that PYCH_12850 aligns not only with S. pneumoniae ScpB (Z-score of 11.2, r.m.s.d. of 1.9 Å) but also with MTH313 from Methanobacterium thermoauto­trophicum (Z-score of 11.0, r.m.s.d. of 1.7 Å), which is a 146-residue DNA-binding protein belonging to the MarR family (Saridakis et al., 2008). Since the WHD is not only a protein–protein interaction domain but also a key component in establishing protein–DNA interactions (Teichmann et al., 2012), the function of the ScpB homologues in a large fraction of archaeal species might lie in a biological activity involving DNA. In a related manner, the genes encoding a kite subunit of the Smc5/6 complex (Nse3) have been duplicated several times and diversified in placental mammals, thus giving rise to a large family of proteins called MAGE proteins with a range of functions that are not directly related to those of Smc5/6 (Palecek & Gruber, 2015).

4.3. The interaction between archaeal Smc and ScpA is conserved

The observed interaction between archaeal ScpA^N and the Smc neck is weak, in contrast to the tight interaction between PyScpA^C (cWHD) and the Smc head domain, which enabled the co-purification of the two proteins (Bürmann et al., 2013). A similar binding property has been observed for bacterial ScpA and Smc pairs: while BsScpA^C and BsSmcHd could be co-purified (Diebold-Durand et al., 2017), the ScpA^ΔC–ScpB and SmcHd-CC pairs derived from B. subtilis, S. pneumoniae and G. stearothermophilus dissociated from each other in size-exclusion chromatography (not shown; Kamada et al., 2017). Notably, in the absence of GsScpB, GsScpA^N interacted with GsSmcHd-CC more tightly, with a dissociation constant of 0.5 µM (Kamada et al., 2017). Likely, the substantial difference in the binding affinities of ScpA or ScpAB for the head and for the neck of Smc might be commonly required for the function of the Smc-based complexes.

5. Perspective

The presented work identifies a class of Smc-based complexes that consist of Smc and ScpA but not ScpB. The Smc–ScpA complex might be the sole Smc-based complex in a large number of archaeal organisms. Our work suggests that the kite subunit might have been recruited to the Smc–ScpA complex to form the Smc–ScpAB complex, probably to meet physiological requirements in most bacterial species. Future work is necessary to delineate the detailed mechanistic roles of the kite subunit in the Smc–ScpAB complex and to address why such a role is unnecessary for chromosome condensation and segregation in many archaeal organisms. The eukaryotic condensin and cohesin complexes do not have kite subunits but hawk proteins (HEAT proteins associated with kleisins), the ancestral proteins of which are found in Lokiarchaeota, the closest relatives of the last eukaryotic common ancestor. One hypothesis is that ancestral kite subunits in ancestral Smc–kleisin–kite complexes were replaced by hawk proteins (Wells et al., 2017). Alternatively, ancestral Smc–kleisin complexes might have recruited hawk proteins.