C-terminal low-complexity sequence repeats of Mycobacterium smegmatis Ku modulate DNA binding Open Access

Many proteins in both prokaryotes and eukaryotes have been identified that contain stretches of simple amino acid sequence repeats that have low information content due to biased amino acid composition and a lack of amino acid diversity. These segments are referred to as LCRs (low-complexity regions). These sequences can either be homopolymers or they can be composed of a few different amino acids, often classified as intrinsically disordered regions [1]. Within a protein, these LCRs have been found to evolve more rapidly than flanking sequences such that their length and amino acid content may differ widely between homologues encoded by different species. These sequences are also characterized by a lack of identifiable three-dimensional structure and are therefore underrepresented in the protein data bank [2]. Because of compositional plasticity and lack of three-dimensional structure, the functional role of low-complexity sequences is not properly understood. However, studies have suggested that position (terminal or central) of the LCRs within a protein sequence plays an important role in determining their function. Proteins with terminal LCRs are important in stress responses, translation and transport processes, and those with central LCRs have been implicated in transcription [3].

In vitro characterization of DNA-binding proteins that contain LCRs at either their N- or C-termini has shown that the LCRs modulate functional properties. For example, Deinococcus radiodurans HU contains proline-, alanine- and lysine-rich PAKKA repeats at its N-terminus that affect the binding-site size and mode of binding to four-way junction DNA [4,5]. Similar PAKKA repeats are present at the C-termini of HU homologues encoded by some members of the actinomycetes and by a member of the genus Kineococcus. In mycobacteria, the HU homologues, also referred to as Hlps (histone-like proteins), contain a particularly extensive C-terminal tail composed of the repeated PAKKA units. In vitro Mycobacterium smegmatis Hlp promotes DNA end-joining by T4 DNA ligase and this ability has been attributed to the lysine-rich C-terminal domain [6]. Streptomyces coelicolor likewise encodes Hlp with PAKKA repeats, and its deletion was shown to be associated with decreased heat resistance and an expanded nucleoid [7]. What is particularly intriguing is that these proteins contain LCRs that resemble those found within the C-terminus of eukaryotic histone H1; in the case of histone H1, the basic repeat region is important for chromatin condensation [8–11].

Ku protein encoded by M. smegmatis is another example of a protein with LCRs composed of PAKKA units (Figure 1B). Ku is an important component of the NHEJ (non-homologous end-joining) DSB (double-strand break) repair pathway in eukaryotes and select prokaryotes such as Bacillus, Mycobacterium and Pseudomonas [12–18]. Mycobacterium tuberculosis Ku has been speculated to play a major role in repairing DSBs induced by genotoxic defence of human cells [12,19], and it has been reported that Ku specifically interacts with the polymerase domain of the multifunctional LigD (ligase D) protein to facilitate DSB repair by NHEJ thereby protecting M. smegmatis against DSBs accumulating during stationary phase [16,17,20,21].

Eukaryotic Ku proteins are heterodimers consisting of two subunits Ku70 and Ku80 that together form a functional unit [22]. In contrast, prokaryotic Ku proteins are homodimers and much smaller (30–40 kDa), being composed of just the central core domain of eukaryotic Ku [17,23]. The β-barrel structure of this core domain is conserved despite limited sequence conservation. In vitro analyses of eukaryotic Ku have shown that it binds non-specifically to both blunt and cohesive DNA ends. Its binding affinity varies from picomolar to nanomolar and is independent of the sequence and the structure of DNA ends, but it is affected by the length of DNA duplex [24–27]. Stoichiometric measurements of eukaryotic Ku have indicated that it requires 14–25 bp of DNA for binding [25,28,29]. M. tuberculosis Ku also binds DNA non-sequence specifically irrespective of the kind of DNA ends and it is dependent on the DNA length [17]. However, little is known about the stoichiometry and binding affinity of mycobacterial Ku, and the role of the lysine-rich C-terminal tail, exclusively seen in Ku encoded by soil-dwelling mycobacterial species, remains unexplored.

In the present study, we show that lysine-rich LCRs are characteristic of Ku proteins from free-living mycobacterial species found in soil and natural reservoirs. DNA-binding experiments suggest a role for the C-terminal tail in DNA interaction. Unexpectedly, removal of the C-terminal lysine-rich repeats from M. smegmatis Ku enhances the DNA-binding affinity. Consistent with the role of the lysine-rich repeats of Hlp in promoting DNA end-joining, only full-length Ku promotes DNA end-joining by a heterologous ligase. We propose that lysine-rich LCRs have evolved repeatedly to modulate the function of unrelated DNA-binding proteins, in the case of M. smegmatis Ku and Hlp to respond more efficiently to environmental stresses with the potential to damage genomic DNA.

The gene encoding Ku (JCVI Locus: MSMEG_5580) was amplified from M. smegmatis genomic DNA using primers 5′CACCATGACGGGTGCGTCAGTTATG-3′ and 5′-TGCGAAGGTGCCCTGAGTTACGAC-3′ and a gene fragment encoding TKu (truncated Ku) lacking the C-terminal lysine-rich repeats was amplified using primers 5′-CACCATGACGGGTGCGTCAGTTATG-3′ and 5′-GCGGGCTAGGAATCCGACTTGG-3′. Both genes were cloned into the Champion pET100/D-TOPO vector (Invitrogen). Fidelity of the constructs was verified by DNA sequencing. The resulting constructs were transformed into Escherichia coli Rosetta Blue cells. Cultures were grown in LB (Luria–Bertani) broth with 50 μg/ml ampicillin at 37°C to a D₆₀₀ of 0.5, and expression of proteins was induced with 1 mM IPTG (isopropyl-β-D-thiogalactopyranoside) for 1 h, following which cells were pelleted at 4°C and stored at −80°C.

For purification of both full-length Ku and TKu, the cell pellets were resuspended in lysis buffer, pH 8.0 [50 mM sodium phosphate (pH 8.0), 2 mM 2-mercaptoethanol, 300 mM NaCl, 5% (v/v) glycerol, 1 mM PMSF, 300 μg/ml lysozyme, 0.05% Triton X-100] and the mixture was incubated on ice for 1 h. DNA was precipitated by slow addition of 13% (v/v) polymin P (BASF) to a final concentration of 0.05%. The cell lysate was centrifuged at 4°C for 40 min at 8000 g. The supernatant was mixed with 1 ml of nickel beads (Sigma) and incubated at 4°C for 1 h. The mixture was loaded on to a gravity flow column and washed with 10 column volumes of lysis buffer and eluted with 150 mM imidazole-containing lysis buffer. The purest fractions were pooled and dialysed overnight at 4°C against low salt Tris buffer, pH 8 [50 mM Tris/HCl, 2 mM 2-mercaptoethanol, 30 mM NaCl, 5% (v/v) glycerol, 1 mM EDTA, 1 mM PMSF] and passed through a Q-Sepharose column equilibrated with the same buffer, and proteins were eluted and analysed as described above. Both Ku and TKu were concentrated and concentrations determined using the Micro BCA Protein Assay Kit (Pierce) using BSA as standard and further confirmed by UV absorbance. Purity was determined by SDS/PAGE, followed by Coomassie Brilliant Blue staining.

All steps of gel filtration were carried out at 4°C using a HiLoad 16/60 Superdex 30 preparative grade column (bed length 60 cm, inner diameter 16 mm; GE Healthcare). The column was equilibrated with 2 column volumes of Tris buffer, pH 8.0 [50 mM Tris/HCl, 2 mM 2-mercaptoethanol, 200 mM NaCl, 10% (v/v) glycerol and 1 mM EDTA]. The gel filtration standard (Bio-Rad), which is a mixture of bovine thyroglobulin (670 kDa), bovine γ-globulin (158 kDa), chicken ovalbumin (44 kDa), horse myoglobin (17 kDa) and vitamin B-12 (1.35 kDa), was run to calibrate the column. The concentration of protein applied to the gel filtration column was 1 mg/ml for both Ku and TKu. The proteins were run independently under the same conditions and were eluted with a flow rate of 0.5 ml/min.

Oligodeoxyribonucleotides used to generate duplex DNA constructs were purchased and purified by denaturing PAGE. The sequences of different DNA substrates used are available in Table 1. The top strand was ³²P-labelled at the 5′-end with phage T4 PNK (polynucleotide kinase). Equimolar amounts of complementary oligonucleotides were mixed, heated to 90°C and cooled slowly to room temperature (22°C) to form duplex DNA. The concentrations of DNA were determined spectrophotometrically.

For binding assays under stoichiometric conditions, 40 or 5 nM of ³²P-labelled DNA was titrated with Ku or TKu, respectively, in a total reaction volume of 10 μl in binding buffer [25 mM Tris/HCl (pH 8), 50 mM NaCl, 0.1 mM Na₂EDTA, 0.05% Triton X-100, 5 mM DTT (dithiothreitol) and 2% (v/v) glycerol]. Reactions were incubated at room temperature for 1 h. A non-denaturing 8% polyacrylamide gel was prerun for 30 min at 175 V in 0.5× TBE [Tris/borate/EDTA (1×TBE=45 mM Tris/borate and 1 mM EDTA)] buffer [45 mM Tris borate (pH 8.3), 1 mM Na₂EDTA], and samples were loaded with power on. After electrophoresis, gels were dried, and protein–DNA complexes and free DNA were quantified by phosphorimaging using software supplied by the manufacturer (Image Quant 1.1). Percentage complex formation was plotted against [protein]/[DNA]. The stoichiometry of the protein–DNA complex was determined by algebraically calculating the value of x at the intersection of the tangents to the linear portions of the graph. Experiments were performed in triplicate.

EMSAs for affinity determination were performed as described above, except that binding reactions contained 5 or 0.5 nM of ³²P-labelled DNA, titrated with Ku or TKu, respectively. For TKu, the binding buffer was modified to contain 300 mM NaCl, keeping the concentration of other components the same. Percentage complex formation was plotted as a function of protein concentrations and fitted to the Hill equation:

where [Ku] is the protein concentration, f is the fractional saturation, K_d reflects the apparent equilibrium dissociation constant and n is the Hill coefficient. All bands corresponding to protein–DNA complexes, including the area between the fastest migrating complex and the free DNA were considered as complex. Fits were performed using the program KaleidaGraph. The K_d value is reported as the mean±S.D. Experiments were performed in triplicate.

Plasmid pUC18 was digested with EcoRI to obtain DNA with cohesive ends. Fifty nanograms of linear pUC18 was incubated with Ku or TKu at room temperature for 1 h. To this reaction, 1 μl of 40 units/μl of T4 DNA ligase was added and incubated at room temperature for 1 h. To one of the reactions, 1 μl of exonuclease III (100 units/μl) was added and incubated at room temperature for 1 h. The reactions were terminated by adding 1 μl of stop buffer [5 mM EDTA, 1.1% (v/v) glycerol and 0.2 mg/ml proteinase K] and 1 μl of 10% (w/v) SDS. Samples were run on 0.8% (w/v) TBE agarose gels and visualized by ethidium bromide staining.

A 105 bp DNA duplex with cohesive ends was generated as described [30]. The 105 bp DNA was labelled with ³²P at the 5′-ends using T4 PNK. Five nanomolar 105 bp DNA was incubated with Ku or TKu at room temperature for 1 h. An aliquot (1 μl) of T4 DNA ligase of concentrations 40 units/μl and 80 units/μl was added to the reaction containing Ku and TKu, respectively, and incubated at room temperature for 1 h. Reactions were treated with exonuclease III and terminated as described above, following which they were phenol extracted and ethanol precipitated and loaded on a prerun 8% polyacrylamide gel and electrophoresed using 0.5% TBE running buffer. Complexes were visualized by phosphorimaging.

Sequence alignment of Ku proteins from mycobacterial species reveals very significant sequence conservation within the core domain, but variation at the C-termini. Soil-dwelling mycobacterial species such as M. smegmatis, M. gilvum, Mycobacterium sp. JLS, Mycobacterium sp. KMS and other free-living mycobacterial species such as M. avium, M. ulcerans, M. marinum and M. kansasii, which are found in natural reservoirs, encode Ku homologues with low-complexity regions characterized by conserved lysine, alanine and proline residues. Strikingly, this LCR is entirely absent in Ku proteins from obligate parasites such as M. tuberculosis and M. bovis (Figure 1A), indicating that only Ku proteins encoded by free-living mycobacterial species inhabiting soil or natural reservoirs contain these rapidly evolving LCRs.

Annotation of the M. smegmatis genome (JCVI) indicates that Ku consists of 358 amino acids, which includes several PAKKA repeats at the C-terminus (Figure 1A). To determine the role of the C-terminal LCR, M. smegmatis Ku and Ku truncated for the C-terminal region (TKu) were purified to apparent homogeneity as judged by Coomassie Brilliant Blue staining of SDS/PAGE gels (Figure 1C, inset); TKu was created by placing a stop codon after residue 327 (Figures 1A and 1B). Analysis of Ku and TKu by gel filtration chromatography indicated that both proteins exist as a homodimer in solution (Figure 1C); this observation was further confirmed by glutaraldehyde cross-linking, which showed no trace of residual monomeric Ku or TKu (results not shown).

We expected the lysine-rich LCR to participate in DNA contacts based on its charge and the previous observation that similar repeats in HU and Hlp homologues modulate DNA binding. However, while Ku binds to 37 bp DNA with K_d=8.6±0.5 nM, TKu binds with much higher affinity (K_d=4.2±0.7 nM using a buffer with significantly higher ionic strength) (Figures 2A and 2B, and Figures 3A and 3B). The Hill coefficients of 1.6±0.1 and 1.1±0.1 for Ku and TKu, respectively, suggest modest positive cooperativity of DNA binding for full-length Ku, reflecting preferred binding of a second Ku protomer to the DNA. Considering that Ku self-associates to bring together DNA ends, this observation can be readily reconciled with its normal function. No sequence preference of Ku and TKu is evident, as indicated by the equivalent affinity for other 36 and 37 bp duplexes (results not shown).

M. smegmatis Ku is a homodimer and removal of the C-terminal extension has no effect on oligomeric assembly (Figure 1C). From EMSA performed under stoichiometric conditions, where proteins were titrated with 37 bp DNA, both TKu and Ku were found to bind 37 bp DNA at a ratio of 4:1 (Figures 2C and 2D, and Figures 3C and 3D), calculated by considering the molecular weight of monomeric protein, which suggests that a dimer requires ~18 bp for binding and is consistent with eukaryotic Ku that has been shown to require 14–25 bp of double-stranded DNA for binding [25,28,29]. The formation of two discrete complexes is consistent with this interpretation.

As a further test of the duplex length required for optimal Ku binding, we used a 21/34 nt hairpin DNA substrate that can accommodate only one Ku dimer [31]. This DNA, which was used for Ku–DNA structure determination, was designed to form a 14 bp duplex that is separated from 7 bp of duplex by a short stem-loop that prevents Ku from sliding along the DNA; the 7 bp duplex is designed to be too short for stable complex formation, thus restricting Ku binding to the 14 bp segment (Figure 4). A binding assay with this construct showed that TKu forms the expected single complex, whereas Ku forms two complexes, most probably due to an interaction between the C-terminal tail of full-length Ku with the 7 bp region of the hairpin substrate (Figure 4). That TKu fails to saturate this DNA construct even at 40 nM protein suggests reduced affinity compared with the 37 bp DNA, perhaps reflecting that 14 bp is insufficient for optimal complex formation.

The inference that Ku may bind DNA shorter than 14 bp prompted us to investigate binding to 37 bp DNA using a 6% polyacrylamide gel, which yields higher resolution. In this gel system, TKu still formed two complexes with 37 bp DNA, consistent with the estimated site size; however, three complexes could be detected with full-length Ku (Figure 5A). The detection of a third complex is intriguing, and it might be a result of protein–protein interactions, leading to two Ku–DNA complexes associating, or due to interaction of the C-terminal lysine-rich tail with the DNA. To examine the presence of protein–protein interaction, an assay was performed in which equimolar concentrations of ³²P-labelled 37 bp and non-radioactive 50 bp DNA was mixed and titrated with increasing concentrations of Ku and TKu (Figure 5B) with the idea that the migration of a complex consisting of two Ku–DNA complexes would be different if one 37 bp DNA duplex is replaced with a 50 bp duplex. However, no such change in the mobility was observed, suggesting that if Ku–DNA complexes do associate in solution, such junctions are not stable during electrophoresis. We therefore surmise that the additional complex observed when full-length Ku interacts with 37 bp DNA is due to interaction of its C-terminal tail with the DNA.

Ku participates in NHEJ repair of DNA DSBs. Earlier studies have reported that M. tuberculosis Ku specifically interacts with and stimulates the ligation activity of LigD protein from M. tuberculosis and that it inhibits end-joining by T4 ligase, reflecting its preferred binding to DNA ends [17]. In contrast, end-joining assays with M. smegmatis Ku using linearized pUC18 or radiolabelled 105 bp DNA substrate showed that M. smegmatis Ku promotes end-joining by T4 ligase as can be seen by the appearance of end-joined products with increasing concentration of Ku (Figure 6). Treatment with exonuclease III digested the end-joined products, which shows that Ku promotes formation of linear multimers and not circularization of the DNA (Figure 6A, lane 6). In contrast to full-length Ku, TKu, at similar and even lower concentrations, prevented the formation of end-joined products and also protected DNA from exonucleolytic cleavage, most likely reflecting its higher affinity binding (Figure 7). Taken together, these data show that while TKu is similar to M. tuberculosis Ku in inhibiting DNA end-joining by a heterologous ligase, full-length M. smegmatis Ku promotes such end-joining, implying that this feature is a property of the C-terminal extension.

The multiple sequence alignment of Ku from various mycobacterial species revealed the presence of lysine-rich LCRs only in Ku encoded by soil-dwelling mycobacterial species such as M. smegmatis, whereas Ku encoded by obligate parasites including M. tuberculosis completely lack these LCRs (Figure 1A, and Supplementary Figure S1 at http://www.bioscirep.org/bsr/033/bsr033e016add.htm). Considering the phylogenetic relationship between mycobacterial species and the clustering and reduced genome size of obligate parasites, the LCR may have evolved in an ancestral species and subsequently been lost in parasitic species. The presence of terminal LCRs in Ku proteins, whose function is in DSB repair, is in agreement with an observation according to which LCRs have position-dependent roles and proteins with terminal LCRs participate in stress responses [3]. Notable examples of proteins that contain terminal LCRs characterized by the same PAKKA repeats include mycobacterial Hlps, which are up-regulated during anoxia or cold shock-induced dormancy and proposed to be involved in DNA DSB repair [6,32]. Similarly, HupS protein from S. coelicolor, which is up-regulated during sporulation and plays a role in DNA packaging and protection, also contains lysine-rich LCRs at its C-terminus while HU from D. radiodurans contains such repeats at its N-terminus [4,7]. Furthermore, these LCRs tend to evolve rapidly [2], suggesting that they have evolved in response to the stress conditions that the bacteria encounter. It is also notable that the PAKKA repeats significantly alter DNA-binding properties, for example, conferring on Hlp the ability to promote DNA end-joining and directing D. radiodurans HU to an unusual binding mode with four-way junction DNA [4,6].

Intermolecular ligation with eukaryotic Ku using 60 bp DNA and with M. tuberculosis Ku using 157 and 445 bp DNA have shown that Ku specifically stimulates the ligation activity of its cognate ligase, but not of unrelated ligases such as E. coli or T4 ligases [17,33]. In apparent contrast with these earlier observations, M. smegmatis Ku promotes end-joining by T4 DNA ligase (Figure 6). However, DNA end-joining is not promoted by TKu, regardless of DNA substrate (Figure 7), indicating that the ability to appose DNA ends for intermolecular ligation by a heterologous ligase is a property of the C-terminal lysine-rich extension. A similar phenomenon was reported for mycobacterial Hlp where the C-terminal lysine-rich domain of mycobacterial Hlp promotes DNA end-joining by T4 DNA ligase, while an Hlp mutant lacking the C-terminal repeats does not [6]. And consistent with its longer LCR, Hlp is more efficient than Ku in promoting end-joining. In analogy with the lysine-rich LCR of histone H1, we predict that the LCRs of Ku and Hlp are unstructured due to electrostatic repulsion, and that association with DNA may promote a helical conformation [34]. Ku and Hlp may therefore bind one DNA substrate via their core domain, while neighbouring DNA may be brought into proximity by interaction with the LCR. A recent study by Grob et al. [35] on yeast and human Ku has shown that Ku has a weak end-bridging activity contributing to end-to-end alignment during DSB repair by NHEJ. Our results suggest that the presence of PAKKA repeats in M. smegmatis Ku might enhance this activity. Moreover, the ability to bring distant DNA segments into proximity appears to be a shared feature of proteins with C-terminal PAKKA-type repeats.

Removal of the C-terminal lysine-rich repeats enhances the affinity of Ku for DNA. This increase in affinity is also manifest in the inability of both T4 DNA ligase and exonuclease III to access the TKu-bound DNA ends, which suggests that complexes with TKu fail to dissociate appreciably in solution during the time of incubation. The gain of stable binding to DNA on truncation of a C-terminal extension has also been reported for Pseudomonas aeruginosa Ku [13], which also contains an extended C-terminal tail, but it lacks PAKKA repeats.

The stoichiometry measurement suggests that the DNA-binding site size for both Ku and TKu is ~18 bp (Figures 2C and 2D, and Figures 3C and 3D). Consistent with the calculated stoichiometry, TKu formed two complexes with 37 bp DNA on a 6% polyacrylamide gel; in contrast, full length Ku formed three complexes with 37 bp DNA on the same gel (Figure 5A). Also, binding to the 21/34 nt hairpin substrate, which has 7 and 14 bp duplex regions separated by a hairpin structure, showed that Ku forms two complexes, whereas TKu forms one complex only with an apparent lower affinity compared with 37 bp DNA as evidenced by the failure to saturate this DNA construct (Figure 4) [36]. For TKu, this suggests that its optimal site size is >14 bp. The differences in the binding properties of full-length Ku and TKu could potentially be attributed to protein–protein interactions between DNA-bound Ku dimers or to the lysine-rich C-terminal LCRs interacting with DNA. Since the pattern of complexes seen when Ku is mixed with equimolar concentrations of ³²P-labelled 37 bp and non-radioactive 37 or 50 bp DNA is identical (Figure 5), we favour the latter interpretation. Interaction between the lysine-rich LCR and DNA would be expected to require only a few base pairs, potentially allowing such interaction to occur with the 7 bp duplex region of the 21/34 nt hairpin construct or with residual base pairs within the 37 bp DNA not occupied by Ku binding via its core DNA-binding motif.

In all, the lysine-rich C-terminus of M. smegmatis Ku significantly modulates DNA-binding properties and promotes DNA end-joining. Evidently, M. smegmatis Ku exhibits properties distinct from those characteristic of M. tuberculosis Ku, properties associated with its unique lysine-rich C-terminus. Low-complexity sequences, such as the PAKKA repeats found in M. smegmatis Hlp and Ku evolve rapidly and we suggest that M. smegmatis Ku has evolved in response to needs to cope with environmental stress such as desiccation.

DSB

double-strand break

EMSA

electrophoretic mobility-shift assay

Hlp

histone-like protein

LCR

low-complexity regions

LigD

ligase D

NHEJ

non-homologous end-joining

PNK

polynucleotide kinase

TBE

Tris/borate/EDTA

TKu

truncated Ku

Ambuj Kushwaha performed all of the experiments. Ambuj Kushwaha and Anne Grove contributed to experimental design and data analysis, and to writing the paper.

This work was supported by the National Science Foundation [grant numbers MCB-0744240 and 1051610 (to A.G.)] and an LSU Economic Development Assistantship.