A direct proofreader–clamp interaction stabilizes the Pol III replicase in the polymerization mode
2013; Springer Nature; Volume: 32; Issue: 9 Linguagem: Inglês
10.1038/emboj.2012.347
ISSN1460-2075
AutoresSlobodan Jergic, Nicholas P Horan, Mohamed M. Elshenawy, Claire E. Mason, Thitima Urathamakul, Kiyoshi Ozawa, Andrew Robinson, Joris M. H. Goudsmits, Yao Wang, Xuefeng Pan, Jennifer L. Beck, Antoine M. van Oijen, Thomas Huber, Samir M. Hamdan, Nicholas E. Dixon,
Tópico(s)DNA and Nucleic Acid Chemistry
ResumoArticle22 February 2013free access A direct proofreader–clamp interaction stabilizes the Pol III replicase in the polymerization mode Slobodan Jergic Slobodan Jergic School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Nicholas P Horan Nicholas P Horan School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Mohamed M Elshenawy Mohamed M Elshenawy Division of Biological and Environmental Sciences and Engineering, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia Search for more papers by this author Claire E Mason Claire E Mason School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Thitima Urathamakul Thitima Urathamakul School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Kiyoshi Ozawa Kiyoshi Ozawa School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Research School of Chemistry, Australian National University, Canberra, Australian Capital Territory, Australia Search for more papers by this author Andrew Robinson Andrew Robinson School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Zernike Institute for Advanced Materials, Groningen, The Netherlands Search for more papers by this author Joris M H Goudsmits Joris M H Goudsmits Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA Search for more papers by this author Yao Wang Yao Wang School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Xuefeng Pan Xuefeng Pan School of Chemistry, University of Wollongong, Wollongong, New South Wales, AustraliaPermanent address: School of Life Science, Beijing Institute of Technology, Beijing 100081, China. Search for more papers by this author Jennifer L Beck Jennifer L Beck School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Antoine M van Oijen Antoine M van Oijen Zernike Institute for Advanced Materials, Groningen, The Netherlands Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA Search for more papers by this author Thomas Huber Thomas Huber Research School of Chemistry, Australian National University, Canberra, Australian Capital Territory, Australia Search for more papers by this author Samir M Hamdan Samir M Hamdan Division of Biological and Environmental Sciences and Engineering, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia Search for more papers by this author Nicholas E Dixon Corresponding Author Nicholas E Dixon School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Slobodan Jergic Slobodan Jergic School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Nicholas P Horan Nicholas P Horan School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Mohamed M Elshenawy Mohamed M Elshenawy Division of Biological and Environmental Sciences and Engineering, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia Search for more papers by this author Claire E Mason Claire E Mason School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Thitima Urathamakul Thitima Urathamakul School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Kiyoshi Ozawa Kiyoshi Ozawa School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Research School of Chemistry, Australian National University, Canberra, Australian Capital Territory, Australia Search for more papers by this author Andrew Robinson Andrew Robinson School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Zernike Institute for Advanced Materials, Groningen, The Netherlands Search for more papers by this author Joris M H Goudsmits Joris M H Goudsmits Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA Search for more papers by this author Yao Wang Yao Wang School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Xuefeng Pan Xuefeng Pan School of Chemistry, University of Wollongong, Wollongong, New South Wales, AustraliaPermanent address: School of Life Science, Beijing Institute of Technology, Beijing 100081, China. Search for more papers by this author Jennifer L Beck Jennifer L Beck School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Antoine M van Oijen Antoine M van Oijen Zernike Institute for Advanced Materials, Groningen, The Netherlands Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA Search for more papers by this author Thomas Huber Thomas Huber Research School of Chemistry, Australian National University, Canberra, Australian Capital Territory, Australia Search for more papers by this author Samir M Hamdan Samir M Hamdan Division of Biological and Environmental Sciences and Engineering, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia Search for more papers by this author Nicholas E Dixon Corresponding Author Nicholas E Dixon School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia Search for more papers by this author Author Information Slobodan Jergic1, Nicholas P Horan1, Mohamed M Elshenawy2, Claire E Mason1, Thitima Urathamakul1, Kiyoshi Ozawa1,3, Andrew Robinson1,4, Joris M H Goudsmits5, Yao Wang1, Xuefeng Pan1, Jennifer L Beck1, Antoine M van Oijen4,5, Thomas Huber3, Samir M Hamdan2 and Nicholas E Dixon 1 1School of Chemistry, University of Wollongong, Wollongong, New South Wales, Australia 2Division of Biological and Environmental Sciences and Engineering, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia 3Research School of Chemistry, Australian National University, Canberra, Australian Capital Territory, Australia 4Zernike Institute for Advanced Materials, Groningen, The Netherlands 5Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, Boston, MA, USA *Corresponding author. School of Chemistry, University of Wollongong, Northfields Avenue, Wollongong, New South Wales 2522, Australia. Tel.:+61 2 42214346; Fax:+61 2 42214287; E-mail: [email protected] The EMBO Journal (2013)32:1322-1333https://doi.org/10.1038/emboj.2012.347 PDFDownload PDF of article text and main figures. Peer ReviewDownload a summary of the editorial decision process including editorial decision letters, reviewer comments and author responses to feedback. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Processive DNA synthesis by the αεθ core of the Escherichia coli Pol III replicase requires it to be bound to the β2 clamp via a site in the α polymerase subunit. How the ε proofreading exonuclease subunit influences DNA synthesis by α was not previously understood. In this work, bulk assays of DNA replication were used to uncover a non-proofreading activity of ε. Combination of mutagenesis with biophysical studies and single-molecule leading-strand replication assays traced this activity to a novel β-binding site in ε that, in conjunction with the site in α, maintains a closed state of the αεθ–β2 replicase in the polymerization mode of DNA synthesis. The ε–β interaction, selected during evolution to be weak and thus suited for transient disruption to enable access of alternate polymerases and other clamp binding proteins, therefore makes an important contribution to the network of protein–protein interactions that finely tune stability of the replicase on the DNA template in its various conformational states. Introduction The Escherichia coli replicase provides a well-characterized system to discover design principles for evolution of structure and function of Nature's dynamic molecular machines. At its heart is the DNA polymerase III holoenzyme (Pol III HE), a complex of at least 17 subunits that include two (or three; McInerney et al, 2007) αεθ cores, two (or three) β2 sliding clamps, and a δδ′τ3ψχ clamp loader assembly in which one (or non-functionally, two or three) of the τ subunits may be substituted by a C-terminally truncated form called γ (McHenry, 2011). The Pol III replicase is dynamic in that many of its subunits change conformation and even binding partners as it carries out coordinated synthesis of both DNA strands at replication forks. Work over the past two decades (reviewed by Johnson and O'Donnell, 2005; Schaeffer et al, 2005; Hamdan and Richardson, 2009; McHenry, 2011) has resulted in (i) determination of static high resolution structures of essentially all of the replicase components, (ii) identification of many pairwise protein–protein interactions that show a finely tuned hierarchy of binding energies, (iii) demonstration that many of the dynamic protein–protein interactions are mediated by intrinsically unstructured segments of subunits that become structured upon interaction with partner proteins (e.g., see Ozawa et al, 2005, 2008; Jergic et al, 2007), and (iv) revelation that many of these interactions occur at sites at which binding partners change places in a particular order during the replication cycles, especially during Okazaki fragment synthesis on the lagging strand. These attributes, when combined with irreversible chemical steps involving dNTP incorporation and ATP hydrolysis, provide the underlying design rules for replicase function as a dynamic machine. In particular, the many weak interactions within the Pol III replicase allow it to transit rapidly from one conformational state to another by breaking and remaking of interactions, without risk of the whole complex dissociating from the template DNA. The proximity effects of nearby interactions effectively reduce the apparent dissociation constants (KD) of protein–protein complexes in the context of the whole replisome relative to values for pairwise interactions. It is thus more likely to uncover weak but nonetheless important interactions using functional assays, rather than by biophysical studies of protein–protein interactions in a pairwise manner. Here, we focus on the complex of the αεθ replicase core with the β2 sliding clamp on primer-template DNA, required for both leading- and lagging-strand synthesis. This complex is expected to have at least two (and probably more) major conformational states, one of which ensures the efficiency and processivity of DNA synthesis (polymerization mode) and the other contributes to fidelity by exonucleolytic editing of polymerase insertion errors (proofreading mode). Protein interactions that affect the transition between these two modes are not currently understood, and are the focus of this article. The β2 sliding clamp is a ring-shaped dimer (Kong et al, 1992) that has to be opened by interactions with a clamp loader complex to be loaded onto a primer-template DNA in a first step to initiate primer extension DNA synthesis (reviewed by Bloom, 2009). The clamp ensures replicase processivity by interaction with the α polymerase subunit of the core. In the polymerase mode, α interacts at only one of the two symmetry-related protein-binding sites in the β dimer (Dohrmann and McHenry, 2005) via a short peptide motif. Related clamp binding motifs (CBMs) occur in disordered segments or loops in the many β-binding proteins (Dalrymple et al, 2001). Thus, having two equivalent sites in the β2 ring enables it to bind two different proteins at the same time. This is suggested to be important for reversible handover of a primer-template from α to a repair polymerase (e.g., Pol II, IV, or V) during bypass of a lesion in the template DNA (López de Saro et al, 2003a; Indiani et al, 2005). The Pol III core contains the polymerase subunit α (1160 residues in E. coli), the proofreading 3′–5′ exonuclease ε (243 residues), and the small θ subunit of unknown function (McHenry and Crow, 1979); α interacts with ε but not with θ, while θ interacts only with ε (Studwell-Vaughan and O'Donnell, 1993). The E. coli α subunit contains two β-binding sites, a conserved internal CBM used for processive DNA synthesis (Dohrmann and McHenry, 2005) and a C-terminal site (Kim and McHenry, 1996b) that may have a role in polymerase recycling from the ends of completed Okazaki fragments (López de Saro et al, 2003b). X-ray crystal structures of a large N-terminal portion of E. coli α (that terminates just before the internal CBM at residue 917; Lamers et al, 2006), of the closely related full-length Thermus aquaticus (Taq) α by itself (Bailey et al, 2006) and bound to primer-template DNA (Wing et al, 2008), and of β2 bound separately to CBM peptides (e.g., Georgescu et al, 2008a) and double-stranded (ds) DNA (Georgescu et al, 2008b) allow construction of a plausible model of the α–β2–DNA complex in the polymerization mode (Wing et al, 2008). The precise location of the ε proofreading subunit in the replicase is uncertain. It has two domains (Figure 1A): its N-terminal exonuclease domain (residues 2–180; Hamdan et al, 2002) interacts with θ (Pintacuda et al, 2006), and residues following Ala209 in its intrinsically unstructured C-terminal segment (εCTS, Gly181–Ala243) interact with the N-terminal PHP domain of α (Wieczorek and McHenry, 2006; Ozawa et al, 2008). The location of the proofreader is a little more clearly defined in the PolC replicase of Firmicutes, where it is integrated as an insertion into the PHP domain, but was removed for PolC structure determination (Evans et al, 2008). Figure 1.A clamp-binding motif (CBM) is located in the C-terminal segment of the ε subunit of Pol III (εCTS). (A) Two-domain organization of ε. The flexible εCTS that interacts with α (Ozawa et al, 2008) extends from the N-terminal exonuclease domain. (B) Sequence alignment of εCTS from representative α-, β-, and γ-proteobacteria shows conservation of the CBM. Residue numbering is based on the E. coli sequence, and boxes denote putative CBMs. Download figure Download PowerPoint We envisaged that a useful strategy to uncover new protein–protein interactions in the replicase would be to challenge it to make DNA under difficult conditions, so that even the weakest interactions become essential. For example, there has not previously been an assay that depends absolutely on the presence of ε in the replicase; ε was observed to stimulate the rate (Kim and McHenry, 1996a) and processivity (Studwell and O'Donnell, 1990) of DNA synthesis in replication assays under conditions where proofreading is not expected to be limiting, and had more subtle effects on coupled leading- and lagging-strand synthesis by full replisomes (Marians et al, 1998). This is in spite of genetic evidence that the very poor growth phenotype of disruption of the chromosomal dnaQ gene (encoding ε) can be rescued by suppressor mutations in dnaE (encoding α) like spq2 (αV832G) that do not relieve the dnaQ mutator phenotype. It was argued that this indicates an additional role for ε in stabilizing the replicase that does not depend on its proofreading capability (Lancy et al, 1989; Lifsics et al, 1992; Slater et al, 1994). Here, we report situations where replication of DNA templates by replisomes assembled in vitro becomes highly dependent on ε, but do not require it to be active as an exonuclease. This non-proofreading activity is traced to a relatively weak interaction of a CBM we identify in ε with one of the protein-binding sites in β2. We show using single-molecule (SM) replication experiments that it also makes an important contribution to both rate and processivity in helicase-coupled leading-strand synthesis, without affecting the lifetimes of active replisomes. We conclude that the ε–β interaction is maintained in the polymerization mode of DNA synthesis by the full replicase and that it is disrupted in transitions to other conformational states. Results Efficient strand-displacement synthesis by Pol III HE requires ε We set up a simplified assay for DNA synthesis by Pol III HE on oligonucleotide-primed circular single-stranded (ss) M13 DNA (6.4 kb); the products were separated on an agarose gel and stained with a dye that detects both ss and dsDNA. In addition to the expected strand-extension synthesis of the fully ds circular product (TFII), we observed robust helicase-independent synthesis of products greater than unit length (Figure 2A). These long products arise from strand-displacement (SD) DNA synthesis, a process studied by Yuan and McHenry (2009). Although the conditions we used are somewhat different (e.g., physiological ionic strength), we confirmed that SD synthesis requires high concentrations of dNTPs, works also with isolated pre-filled TFII, and is dependent on ssDNA-binding protein (SSB) with an intact C-terminal protein-binding motif. It also requires loading of β2 on the primer template and interaction of αεθ with at least one τ subunit in a clamp loader that also contains ψ and χ for interaction with SSB. Figure 2.Pol III strand-displacement (SD) DNA synthesis provides functional evidence for the ε–β interaction. (A) Efficient SSB-dependent SD DNA synthesis by the complete Pol III HE under standard conditions (see Materials and methods). Time course of flap-primer extension on M13 ssDNA shows larger than unit length dsDNA (tailed form II, TFII) products produced by SD synthesis. (B) The ε subunit, but not θ, contributes to SD synthesis. Assays (20 min) used Pol III HE with purified αεθ or assembled in situ with α±ε±θ. (C) The ε contribution to SD synthesis does not require proofreading and determinants of it are located in the εCTS. SD DNA synthesis by Pol III HE containing indicated core sub-complexes assembled in situ. (D) Mutations within the β-binding motif of ε affect SD synthesis. Time course of DNA synthesis by Pol III HE with purified wild-type or mutated αεθ cores; εQ does not bind significantly to β while εL has a strengthened binding site. (E) Both protein interaction sites in β2 are engaged during SD synthesis. Assays had wild-type (βwt)2 substituted by modified clamps His6-βwt/βwt (no site occluded), His6-βC/βwt (one site occluded) or (His6-βC)2 (both sites occluded), or no clamp. (F) The spq2 suppressor (αV832G) polymerase is capable of more efficient SD synthesis than wild-type α. Assays (20 min) were under standard conditions except that NaCl was at 100 mM, with versions of Pol III HE assembled in situ. All panels show photographic negative images of gels that had been stained with SYBR gold nucleic acid stain (Invitrogen). Download figure Download PowerPoint The SD reaction is demanding in terms of its requirement for all but one of the HE subunits. Indeed, hoping to detect a novel function of the θ subunit, we examined the dependence of SD synthesis on the Pol III core subunits. Although we found θ to be dispensable under our standard conditions, SD synthesis was absolutely dependent on the ε proofreading subunit (Figure 2B). This is the first report of a replication assay that is so strongly dependent on ε. SD synthesis does not require the exonuclease domain of ε The ε subunit contains a binuclear Mn2+ or Mg2+ metallocentre at its active site coordinated by carboxylates of Asp12, Glu14, and Asp102 (Hamdan et al, 2002), and its D12A and D12A/E14A mutants have no residual 3′–5′ exonuclease (proofreading) activity (Fijalkowska and Schaaper, 1996). Nevertheless, we found that both mutants were capable of sustaining extensive SD synthesis (Figure 2C), indicating that proofreading is not required for this process. To determine if the εCTS (Figure 1A), decoupled from the exonuclease domain, is itself sufficient to support SD synthesis, we fused it to the C-terminus of human ubiquitin (as a small soluble tag) to produce ubq-εCTS (Supplementary Figure S1A). Although ubq-εCTS was partly proteolysed during expression in E. coli (Supplementary Figure S1B), the intact protein in this preparation still interacted strongly with α (Supplementary Figure S1C), and a mixture of α and ubq-εCTS could still sustain robust SD synthesis (Figure 2C). This provides clear evidence for a non-proofreading role of ε in DNA replication, dependent only on residues within the εCTS. A potential clamp-binding site in the C-terminal segment of ε Sequence alignment of the ε subunit in species of the α-, β-, and γ-proteobacteria shows conservation of the structured nuclease domain, but much greater variability in the εCTS. An exception is a short moderately conserved region immediately following the structured domain (Figure 1B) that resembles a CBM in other proteins (Dalrymple et al, 2001); CBMs are either penta- (optimally QLS/DLF) or hexapeptides (QxxΦxΦ, where x is any residue and Φ is hydrophobic). The regions in the ε subunits of various species mostly resemble the hexapeptide (QTSMAF in E. coli), but some also have pentapeptide motifs. Studies of binding of many synthetic peptides to β2 enable reliable prediction of binding strengths of CBMs (Wijffels et al, 2004, 2011). Thus, we made two mutants of wild-type ε (εwt) that we call εQ, in which the conserved first residue (Gln182) of the motif is changed to Ala, and εL, with the motif changed to QLSLPL. The εQ motif is predicted to bind more weakly to β2 than the εwt motif, while the εL motif is one of the tighter binding CBMs, from the replication initiation factor Hda (Wijffels et al, 2004). We confirmed that both εQ and εL interact with α as expected, and could be used to assemble stable isolable αεθ core complexes. These isolated Pol III cores containing εQ, εL, and wild-type εwt were compared for their ability to support Pol III SD DNA synthesis (Figure 2D). The results were consistent with the predicted ε–β binding strength; the αεQθ complex could no longer sustain SD synthesis, while αεLθ promoted more extensive synthesis than the wild-type complex. Therefore, the putative CBM in ε is required for SD DNA synthesis. Efficient SD DNA synthesis by Pol III HE requires two β-binding sites We predicted that SD synthesis by the Pol III HE would require that the two protein interaction sites in the β dimer be occupied simultaneously by a CBM of α and the newly discovered site in ε. To test this, we prepared a hemi-mutant β dimer made up of one native subunit that contains an intact protein interaction site and one that does not (βC, a mutant lacking five C-terminal residues that comprise part of the CBM-binding cleft). Scouten Ponticelli et al (2009) had earlier made a hemi-mutant βC/βwt dimer in vivo, isolated and characterized it. They showed that it could be efficiently loaded on a primer-template DNA by the clamp loader, and was proficient for DNA strand extension. We used subunit exchange and chromatography to isolate a similar His6-βC/βwt hemi-mutant dimer, and its composition and absence of contamination by (βwt)2 was confirmed by electrospray ionization mass spectrometry (ESI-MS) under native conditions (Supplementary Figure S2). The hemi-mutant β was inactive in the Pol III SD reaction (Figure 2E), showing that both protein interaction sites in β2 are utilized, presumably being bound simultaneously to the CBMs in α and ε. In contrast, and consistent with previous work showing that a single binding cleft in the clamp is sufficient to stimulate replication by Pol III both in vitro (Scouten Ponticelli et al, 2009) and in vivo (Sutton et al, 2010), the hemi-mutant β was still able to support efficient primer extension to form TFII (Figure 2E). The αV832G mutation suppresses the requirement for ε in SD DNA synthesis The phenotype of the spq2 mutation in dnaE (αV832G) argues for an important non-proofreading function of ε in vivo, so we studied the activity of αV832G in SD synthesis (Figure 2F) and found that (i) it is still active even in the absence of ε and (ii) its activity is stimulated by ε to a level higher than wild-type α. These observations suggest that αV832G forms more stable interactions with proteins or DNA that can partially compensate for the absence of ε, and that ε normally makes an important contribution to the protein and DNA interaction network that stabilizes the wild-type polymerase on the DNA template. Physical evidence for interaction of β2 with ε in the αεθ core complex We next used surface plasmon resonance (SPR) to confirm the predicted strengths of interactions of β2 with the εwt, εQ, and εL peptides. Synthetic biotinylated peptides were immobilized on a streptavidin-coated SPR chip, and solutions of β2 were made to flow over it. Binding isotherms (Figure 3A; sensorgrams in Supplementary Figure S3A) showed β2 bound the εL peptide with KD=0.38±0.04 μM, consistent with previous data (0.38–0.45 μM) for very similar peptides (Wijffels et al, 2004). The εwt peptide was bound 550-fold more weakly, with KD=210±50 μM, while essentially no binding was detected with εQ (KD>2 mM). Figure 3.Physical interaction occurs between β2 and the clamp-binding motif (CBM) of ε in the αεθ–β2 complex. (A) β2 binds to a peptide containing the εwt CBM. SPR binding isotherms (R/Rmax) for the interaction of β2 with immobilized decapeptides containing CBMs from εL (diamonds), εwt (circles), and εQ (squares) are shown; sensorgrams are in Supplementary Figure S3A. Fits to data using a 1:1 binding model (εL and εwt peptides) are shown as solid lines. The small responses with the εQ peptide at the highest [β2] indicate KD>2 mM (see Supplementary Figure S3A). (B) The CBM in εL is accessible to β2. NanoESI mass spectra of 1 μM β2 alone or with 20 μM εwt, εQ, or εL show that εL interacts more strongly with β2 than does εwt. Proteins were in 140 mM NH4OAc, and ions due to free monomeric ε (1700–3400 m/z) have been omitted for clarity. (C) Wild-type ε contacts β2 in the αεθ–β2 complex, shown by a shift in the αεθ: αεθβ2 equilibrium (in excess β2) with progressive increase in ε–β binding strength. NanoESI-MS of 2.8 μM β2 with 1.8 μM purified αεθ cores in 140 mM NH4OAc. Ions due to free β2 (4000–5000 m/z) are not shown. Download figure Download PowerPoint Because of the proximity of the putative CBM to the structured domain of ε (Figure 1A), we thought it might not be accessible to a protein as large as β2. Gel filtration of a mixture of β2 with εL (Supplementary Figure S3B) confirmed CBM accessibility since much of the εL eluted in a peak coincident with β2. In the same conditions, however, neither εwt nor εQ showed much evidence of complex formation. To characterize the weak ε–β interaction with the native proteins, we turned to electrospray-ionization mass spectrometry (ESI-MS) in ammonium acetate (NH4OAc) buffers at neutral pH. While it is difficult to determine accurate KD values of protein complexes in solution by ESI-MS because different species ionize with different efficiencies, it can be used reliably to detect and rank the stabilities of complexes containing similar species in solution, for example, mutant proteins (Kapur et al, 2002). This follows from the argument: For two proteins A and B at equilibrium, KD=[A][B]/[AB]. If [A] is kept constant and in excess of [B], then the ratio of observed ions corresponding to B and AB is related to the KD of AB by a constant (c) determined by [A] and their relative ionization efficiencies, KD=c × [B]/[AB]. Thus, provided the reasonable assumption is made that mutations in B do not greatly affect its ionization efficiency or that of the AB complex, KD values of similar complexes can be ranked using the ratios of ions corresponding to free B and AB. The ESI-mass spectra of equilibrium mixtures of excess εQ or εwt with β2 showed only small amounts of εβ2 complexes (Figure 3B), providing little evidence for a significant ε–β interaction. In contrast, the εL protein, engineered to contain a strong β-binding site, shows clear evidence of species that contain one or two εL bound to one or both CBM-binding sites in the β dimer. However, although consistent with the gel filtration data, these data do not yet show that εwt interacts with β2. To demonstrate this, we used mixtures of β2 (in excess) with isolated αεθ complexes containing the three ε variants. The results (Figure 3C) clearly show the existence of an αεwtθ–β2 complex (ratio of ions αεwtθ–β2/αεwtθ is ∼1); the role of the putative CBM in εwt in this interaction over and above the α–β2 interaction (see middle panel in Figure 4A) is further demonstrated by the relative absence of complex formation with αεQθ, and the nearly quantitative formation of an apparently stable αεLθ–β2 complex. Figure 4.Physical evidence that the ε subunit and the internal CBM in α synergistically sequester β2 in the αεθ–β2 complex. (A) Of the two CBMs in α, the internal site interacts preferentially with β2. NanoESI-MS of 2 μM β2 and 0.9 μM αΔ7, αwt, or αL in 400 mM NH4OAc. Ions due to free β2 (4000–5000 m/z) are not shown. (B) Both ε and the internal CBM in α bind β2 in the αεθ–β2 complex. NanoESI-MS of 2 μM β2 and cores assembled in situ with 0.9 μM αΔ7, αwt, or αL, 2 μM εwt and 5 μM θ in 400 mM NH4OAc. Ions due to excess ε (1700–3400), θ (1000–3000), β2 (4000–5000)
Referência(s)