Part:BBa_K3512002

CcdA Antitoxin

CcdA antitoxin is part of the CcdA-CcdB toxin-antitoxin system. The target of CcdB is the GyrA subunit of DNA gyrase, an essential type II topoisomerase in Escherichia coli. Gyrase alters DNA topology by effecting a transient double-strand break in the DNA backbone, passing the double helix through the gate and resealing the gaps. The CcdB toxin acts by trapping DNA gyrase in a cleaved complex with the gyrase A subunit covalently closed to the cleaved DNA, causing DNA breakage and cell death in a way closely related to quinolones antibiotics.

In absence of the antitoxin CcdA, the CcdB toxin traps DNA-gyrase cleavable complexes, inducing breaks into DNA and cell death.

Contribution: 2022 Bioplus-China

Toxicity and antitoxicity plate assays

To test the toxicity of the cloned CcdBO157 and CcdBF variants, the corresponding pBAD33-ccdB constructs were transformed in MG1655. The resulting transformants were plated on LB plates containing chloramphenicol with or without arabinose (1%). The CcdB variants were considered to be functional (toxic) when transformants were able to grow only in the absence of arabinose. To test the ability of the cloned CcdAO157 and CcdAF variantsto counteract the toxicity of CcdBO157 and CcdBF, respectively,the corresponding pKK-ccdA constructs were transformed in MG1655 expressing the reference ccdBO157 orccdBF genes from the pBAD33 vector. The resulting transformants were plated on LB plates containing chloramphenicol and ampicillin with arabinose (1%). Basal expression of ccdA from the pTac promoter of pKK223-3 in MG1655 is sufficient to test the antitoxicity phenotype. The CcdA variants were considered to be functional when the toxicity of CcdBO157 or CcdBF protein was counteracted, i.e., when strains coexpressing a ccdA variant with the ccdBO157 or ccdBF reference genes were able to grow in the presence of arabinose while strains expressing only the ccdBO157 or ccdBF reference gene were not.

Result

We identified a small sequence that presents identity with the 39 region of ccdAO157 and two putative ORFs (Figure 4, A and B). The 2704-bp IR corresponds to that of 1425 bp with the insertion of an IS621 in the sequence presenting identity with ccdAO157.

Table 3 shows the amino acid sequences of the corresponding CcdA proteins. The 47 antitoxin proteins presented very few variations, except in the case of CcdAO138. Overall, seven classes of alleles could be identified. The most prevalent one was identical to the ccdAO157 gene of the O157:H7 EDL933 reference strain (37/47 isolates). Five classes, representing 9/47 isolates, presented single point variations (S76R, A54T, D72E, T47I, or T34M). The capacity of 1 representative protein of each class to antagonize the toxic activity of the reference CcdBO157 protein was tested, using the antitoxicity plate assay. The single point variations did not affect the capacity of the CcdA variants to antagonize CcdBO157 activity (data not shown, Table 3). The last class (1 isolate) presented a frameshift mutation caused by a 1-nt deletion. This led to a major modification of the carboxy terminus of CcdAO138 that affects the antitoxic activity of this variant. Indeed, when co-expressed with CcdBO157, this protein was unable to restore viability, showing that CcdAO138 is inactive (Figure 5A). This result was expected since it has been shown that the carboxy-terminal domain of CcdAF is responsible for the antitoxin activity (Bernard and Couturier 1991).

Discussion

Interestingly, the evolution of ccdAO157-like antitoxins and of the flanking apaH and folA genes appears to be much more constrained. Inactivation of the toxin gene prior to the antitoxin gene presumably constitutes the first and safer step of TA systems degradation.

Indeed, the inactive CcdBO138 variant is coupled either to the inactive CcdAO138 as in the case of the ccdO138 variant or to an active antitoxin as in the case of the ccdO153 variant. Thus, the situation at present strongly indicates a decay of the ccdO157 system. An alternative hypothesis is that the antitoxin might play an antiaddictive role as shown for the ccdEch system (Saavedra De Bast et al. 2008) although not against CcdBF-like toxin since CcdAO157 antitoxin does not protect against ccdF addiction (Wilbaux et al. 2007).

Reference

Mine, N., Guglielmini, J., Wilbaux, M. and Van Melderen, L., 2009. The Decay of the Chromosomally Encoded ccdO157 Toxin–Antitoxin System in the Escherichia coli Species. Genetics, 181(4), pp.1557-1566.

Introduction

In light of the important regulatory roles performed by the intrinsically disordered C-terminal arms of CcdA, our goal is to characterize the thermally accessible states in the native ensemble of apo CcdA. In the present study, we build off of these advances to map the conformational free energy surface of CcdA, paying particular attention to the disordered C-terminal arms. Our free energy simulations are 5 compared to experimental data from single-pair Förster resonance energy transfer (spFRET) and existing NMR measurements. We then propose how the conformations within CcdA’s free-energy surface enable it to both bind CcdB and regulate cleavage by Lon protease.

Result

CcdA preferentially adopts closed and partially open states CcdA fluctuates between several conformational states when not bound to CcdB. Specifically, Madl et al. observed a folded NTD and multiple sets of resonances for several C-terminal residues in the 15N1H HSQC spectrum for a R70K mutant of CcdA, suggesting that CcdA adopts several conformations on the NMR timescale. One set of resonances had chemical shifts that were in the range expected for a folded protein and are associated with long-range NOEs, while the chemical shifts of the other two sets of resonances were in the range expected for a random coil and were associated with only trivial and short-range NOEs. Using these data, Madl et al. constructed two structural models for CcdA. We refer to these models as the closed-NMR structures and the extended-NMR structures (Figure 1A-B). In the closed-NMR structures, both C-terminal arms fold back against the structured NTD (Figure 1A). In the extended-NMR structures, the C-termini are modeled as having no contacts with the NTD and adopt an ensemble of extended conformations, corresponding to the random coil-like set of chemical shifts for which no long-range NOEs were observed (Figure 1B).

To determine the ensemble of structures sampled by CcdA in solution, we first calculated the free energy of CcdA as a function of radius of gyration (Rg) using umbrella sampling coupled with explicit-solvent molecular dynamics simulations. The resulting free-energy surface has a well-defined global energy minimum and several shallow local minima (Figure 1C). Structures within the lowest energy state are compact in the sense that both C-termini arms (C1 - residues Ala41-Trp72 monomer 1; and C2 – residues Ala41’-Trp72’ monomer 2) are folded back against the structured NTD (Figure 1C, Conformer IR). Outside of the free energy minima, several less compact conformations are sampled (e.g. Figure 1C, conformers IIR and IVR-VIR).

The free energy surface of CcdA clarifies under-determined NMR data To determine how the theoretical free-energy surface of CcdA compares to the aforementioned NMR studies, we computed ensemble-averaged chemical shifts for CcdA based on the freeenergy surface. Overall, the theoretical chemical shifts are in good agreement with the experimentally determined values (Figure 2A-B).

Additionally, because long-range NOEs were observed between C- and N-terminal residues, we examined the inter-residue contacts observed in the theoretical closed state models (Figure 2C-F). The closed-state contact maps agree qualitatively with the contacts deduced from the experimental NOEs (circled in Figure 2E-F). That is, the general trend of C-terminal residues Ala66-Arg70 contacting N terminal residues Ala19-Val22 is consistent with the experimentally observed NOEs, while the specific residues vary slightly from those residues for which strong NOEs were observed (residues Ala66 and Asp67 to residues Tyr20, Val22, and Leu39’, Figure 2C-D). We note that the NMR study was performed on a CcdAR70K mutant and, although this corresponds to a conservative mutation, Arginine to Lysine mutations can have destabilizing effects on protein structure27,28. Moreover, this substitution, which occurs near the C-terminal residues that are involved in long-range NOEs (i.e., residues A66 and A67), may affect the specific contacts involved in the closed state. Nevertheless, although the interaction with residue Leu39’ was not observed in the inter-monomer contact map, contacts were detected between C-terminal residues Asn62-Ser64 of one monomer with N-terminal residues Leu39’-Asn42’ of the other monomer (Figure 2F).

(A) Ensemble averaged 1H chemical shifts. The ensemble mean chemical shift is shown as a black dot for each residue, with error bars indicating standard deviation combined additively with the root mean squared error in the SHIFTX2 predictions. Red dots indicate the experimental NMR 1H chemical shifts extracted from the HSQC spectrum for CcdAR70K19. (B) Ensemble averaged 15N chemical shifts, analogous to (A). (C) Contacts corresponding to the experimentally observed NOEs 19. Residues involved in NOEs are shown as sticks and labeled. Orange lines indicate contacts with residue Leu39’, blue lines contacts with residue Tyr20, and pink lines contacts with residue Val22. (D) Close-up view of contacts corresponding to the experimentally observed NOEs 19 shown in Panel C. (E) Intra-monomer contact maps derived from the PMF for the closed state of CcdA. The residue-pairs corresponding to detected NOEs are circled in the contact maps in Panels E and F and colored as in Panel D. (F) Inter-monomer contact maps derived from the PMF for the closed state of CcdA, analogous to (E).Since the open state does not, by definition, have any contacts between residues 66 and 67 and residues in the NTD, no NOEs would be observed experimentally for this state. However, ensemble-averaged inter- and intra-monomer contact maps for the partially open state are similar to the contact maps from the closed state (Figure 3). Thus, the closed and partially open states are not distinguishable through measurement of NOEs alone.

Since the open state does not, by definition, have any contacts between residues 66 and 67 and residues in the NTD, no NOEs would be observed experimentally for this state. However, ensemble-averaged inter- and intra-monomer contact maps for the partially open state are similar to the contact maps from the closed state (Figure 3). Thus, the closed and partially open states are not distinguishable through measurement of NOEs alone.

The ensemble average contact maps for the partially open state do not fully capture the range of partially open structures that the protein can adopt. Since there are two disordered C-terminal arms, and each one can contact the folded NTD, we can, in principle, distinguish between two partially open conformations. In the first conformation C1 is open and only the C2 arm contacts the NTD (partially-open substate A shown in Figure 4A), and in the second conformation C2 is open and only the C1 arm contacts the NTD (partially-open substate B shown in Figure 4B). In both substates, however, the C-terminal arm that contacts the NTD has interactions with the NTD that are similar to what is observed in the closed-NMR structures (Figure 4C and D).

Insights into the structure of CcdA using spFRET

For the spFRET experiments, we added donor and acceptor fluorophores to the C-terminal arms of the protein. As there is no Cysteine naturally present in the C-terminal of CcdA to which a fluorophore could be attached, we first created a mutant protein in which one residue from each monomer was mutated to Cysteine. We selected residue F58 for this mutation due to its central location in the disordered C-terminal domain, in addition to the assumption that placing a fluorophore in a position that already accommodates a residue with a large side chain would minimize any perturbation to the structure (Figure 5A).

SpFRET analysis of F58C CcdA, with donor and acceptor fluorophores attached to residue C58, shows a bimodal distance distribution (Figure 5B, middle panel). For direct comparison with the free energy profile of CcdA, we transformed the PMF to an axis describing the inter-monomer distances between residue F58-F58’ Cα–atoms (Figure 5B, top panel). The low-FRET state samples distances between residue F58-F58’ Cα–atoms (Figure 5B, top panel;). The low-FRET state samples distances near 45Å, around which we also see a broad local energy minimum in the PMF (ID, Figure 5B). Representative structures from this local energy minimum correspond to closed conformations (Figure 5C). By contrast, the high-FRET state samples distances near 30Å, which corresponds another local energy minimum in the transformed PMF of CcdA (IID, Figure 5B), and this state is populated by both partially-open and closed conformations (Figure 5C). Altogether the spFRET data are wholly explained by the free energy simulations that demonstrate that CcdA preferentially adopts closed and partially-open states. For comparison, we also computed the inter-monomer F58 F58’ distances within the previously constructed NMR structures (Figure 5B bottom panel, Figure 1A-B). While both the closed-NMR and extended-NMR structure models sample a range of inter-monomer distances, the bimodality apparent from spFRET distance distribution is not apparent in the NMR models.

CcdB-binding-competent structures are enriched in apo CcdA’s partially open state

The CcdB toxin has two partially overlapping bindings sites for CcdA that can simultaneously bind two C-terminal arms from distinct CcdA molecules (Figure 6A-B).The binding sites have different affinities for CcdA. In the low-affinity binding site, a C-terminal arm from CcdA binds CcdB with residues R40-M61 and forms an alpha helix, which we refer to as S1, while residues N62-W72 remain disordered. In the high-affinity binding site, a C-terminal arm from a different CcdA dimer binds CcdB through both an extended S1 (R40-G63) and a second short structure with a turn involving residues S64-W72, which we call S2.

We explored whether either C-terminal arm of apo CcdA adopts conformations similar to S1 or S2. Average backbone root mean square deviation (RMSD) from S1 remains above 4Å for all radii of gyration (Figure 6C), indicating that neither C-terminal arm adopts conformations that are similar to S1. However, C2 (the arm that more frequently adopted open conformations during the free energy simulations) frequently samples structures within 3Å of S2 (Figure 6C). A representative structure with low RMSD to S2 is shown in Figure 6D. This representative structure has a Rg of approximately 18Å. Conformers within this simulation window have a mean RMSD to S2 of 3.1+-0.5Å and an associated relative free energy of 2.36 kcal/mol, indicating that 0.4 % of CcdA molecules in solution adopt conformations with this Rg (and the corresponding range of RMSDs to S2) at any given time.

Specific CcdA conformations are recognized by Lon protease

CcdA is cleaved by ATP-dependent Lon Protease at specific known sites. However, Lon-mediated cleavage occurs via a number of steps, beginning with substrate recognition, and CcdA’s Lon recognition sites are not known. It is known, however, that Lon Protease recognizes clusters of hydrophobic residues within sequences as short as seven to twenty residues and that these recognition sites have certain hallmarks, such as aromatic residues and high surface-burial scores (e.g. greater than 140).We thus screened the CcdA sequence for the seven to twenty residue long subsequence with the highest surface-burial score (Figure 7A, Figure S4A) Residues R37-W44 (RRLRAERW) had a surface-burial score of 154.8, suggesting that Lon may recognize this sequence.

Our data suggest that this potential Lon recognition site (residues R37-W44) is significantly more solvent exposed in monomer 1 than in monomer 2 (the monomer whose C-terminal more frequently adopted open states during the free energy simulations) (Figure 7B). Closure of a C terminal arm involves formation of contacts between residues N62-S64 of that arm with residues L39’-E42’ from the other monomer (Figure 2F). Therefore, opening of one C-terminal arm necessarily exposes the predicted Lon recognition region on the other monomer. This is illustrated in Figure 7C, which shows the solvent-accessible surface of a representative conformation of CcdA from the simulation window with free energy of 0.34kcal/mol. C1 partially blocks the predicted recognition site on monomer 2, while the position of C2 exposes the predicted recognition site on monomer 1 to solvent.

Reference

Burger, V., Vandervelde, A., Hendrix, J., Konijnenberg, A., Sobott, F., Loris, R. and Stultz, C., 2017. Hidden States within Disordered Regions of the CcdA Antitoxin Protein. Journal of the American Chemical Society, 139(7), pp.2693-2701.

Sequence and Features

Assembly Compatibility:

• 10
  COMPATIBLE WITH RFC[10]
• 12
  COMPATIBLE WITH RFC[12]
• 21
  COMPATIBLE WITH RFC[21]
• 23
  COMPATIBLE WITH RFC[23]
• 25
  COMPATIBLE WITH RFC[25]
• 1000
  COMPATIBLE WITH RFC[1000]