Structural insights into RNA polymerase recognition and essential function of Myxococcus xanthus CdnL

Abstract

CdnL and CarD are two functionally distinct members of the CarD_CdnL_TRCF family of bacterial RNA polymerase (RNAP)-interacting proteins, which co-exist in Myxococcus xanthus. While CarD, found exclusively in myxobacteria, has been implicated in the activity of various extracytoplasmic function (ECF) σ-factors, the function and mode of action of the essential CdnL, whose homologs are widespread among bacteria, remain to be elucidated in M. xanthus. Here, we report the NMR solution structure of CdnL and present a structure-based mutational analysis of its function. An N-terminal five-stranded β-sheet Tudor-like module in the two-domain CdnL mediates binding to RNAP-β, and mutations that disrupt this interaction impair cell growth. The compact CdnL C-terminal domain consists of five α-helices folded as in some tetratricopeptide repeat-like protein-protein interaction domains, and contains a patch of solvent-exposed nonpolar and basic residues, among which a set of basic residues is shown to be crucial for CdnL function. We show that CdnL, but not its loss-of-function mutants, stabilizes formation of transcriptionally competent, open complexes by the primary σA-RNAP holoenzyme at an rRNA promoter in vitro. Consistent with this, CdnL is present at rRNA promoters in vivo. Implication of CdnL in RNAP-σA activity and of CarD in ECF-σ function in M. xanthus exemplifies how two related members within a widespread bacterial protein family have evolved to enable distinct σ-dependent promoter activity.

Figure 1. CdnL domain analysis.

(A) Limited proteolysis of CdnL and TtCdnL. Aliquots of the reaction mix (1∶100 w/w subtilisin:protein) at 30°C were withdrawn at the times (in min) indicated on top and analyzed in Coomassie-stained 15% SDS-PAGE gels (lane “M”: molecular weight markers). Arrowhead on the right points to the subtilisin-resistant fragment, which mass spectrometry and N-terminal sequencing identified as the ∼100-residue C-terminal region in both proteins. Note the slower mobility of TtCdnL, whose size (164 residues) is the same as CdnL. (B) Far UV-CD spectra of CdnL, TtCdnL, and their indicated fragments. (C) BACTH analysis of the interactions of CdnL and its domains showing reporter lacZ expression in E. coli transformed with plasmids pKT25 and pUT18 bearing cdnL or the gene for CdnLNt (left panel); or with a pUT18C construct of the gene for Mxβ_19–148 and pKT25 with cdnL or its indicated fragments (right panel). pKT25 without insert was used in negative controls (“−”).

Figure 2. NMR solution structures of CdnL and its domains.

(A) Schematic comparing CdnL and CarD domain architectures. Structured domains are shown as ellipsoids, intrinsically unstructured domains as wavy lines, and known interactions of each domain are indicated. Residues delimiting the CdnL and CarD domains are numbered. Note that CdnL resembles CarDNt in size and sequence, and lacks the HMGA-like DNA binding domain that is present in CarD. NMR structures of CdnLNt, CdnLCt, and full-length CdnL were determined in this study. (B) Superposition of the backbone traces for the 20 final NMR structures for CdnLNt (blue), CdnLCt (red), as indicated. (C) Same as in (B) for full length CdnL. (D) Average NMR solution structure of native CdnL in ribbon (left) and electrostatic surface representations (right). In the ribbon structure, the five β-strands (in cyan) of the CdnLNt domain are indicated, and the CdnLCt α-helices are colored red (α1), magenta (α2), yellow (α3), blue (α4) and green (α5). (E) Steady state backbone ¹⁵N-¹H NOE (from the ratio of cross-peak intensities with and without ¹H saturation at pH 7.0 and 25°C) plotted versus residue number obtained for 1 mM ¹⁵N, ¹³C-labeled CdnL. Errors for the heteronuclear NOE data were estimated to be ≤5% based on average noise levels in the NMR spectra.

Figure 3. Structural comparisons with M. xanthus CdnL NMR structure.

Backbone overlay of the CdnL NMR structure (red) onto the crystal structures of (A): TtCdnL (green; PDB: 4L5G); (B) one unit in the MtCdnL domain-swapped dimer (blue; PDB: 4ILU); (C) MtCdnL in complex with the RNAP β-lobe domain (cyan; PDB: 4KBM); (D) all four structures. The overlay shown was generated with optimal (maximum) superposition of the N-terminal domains of each structure. (E) Backbone overlay of CdnLNt (cyan; left) onto: (top) TtCdnLNt (gold; PDB ID: 2lqK); (bottom) E. coli TRCF-RID (magenta; PDB ID: 2eyq). (F) Backbone overlay of CdnLCt (red) onto the TPR domains of: (top) FK506-binding protein (blue; PDB ID: 1kt0); (bottom) CadC_pd (green; PDB ID: 3ly8). Structures were generated with MOLMOL.

Figure 4. Mutational analysis of CdnL interaction with RNAP-β.

(A) Native CdnLNt structure showing residue side chains that may contact RNAP-β, with the three labeled tested by mutational analysis (see text). (B) BACTH analysis showing that the F36A, M49A and P51A CdnL variants self-interact in cells with the pKT25 and pUT18 constructs of wild type or mutant cdnL, as indicated. (C) BACTH analysis of the interaction with Mxβ_19–148 (in pUT18C) of CdnL F36A, M49A, or P51A mutants in pKT25. (D) Sequence alignment of the M. xanthus RNAP-β segment, Mxβ_19–148 (NCBI code, YP_631280.1) and its equivalents in T. thermophilus, Ttβ_10–133 (WP_014630291), in E. coli, Ecβ_19–142 (P0A8V2), and in M. tuberculosis (CAB09390). Residues shaded black if identical, or grey if similar, in at least two of the aligned sequences. An asterisk in the consensus line below indicates conservation in all four sequences. Red arrows point to RNAP-β residues analyzed by site-directed mutagenesis in this study. (E) BACTH analysis of the interaction of Mxβ_19–148 (WT) or its indicated mutants in pUT18C versus wild-type CdnL in pKT25.

Figure 5. Consequences of disrupting CdnL interaction with RNAP in vivo.

(A) Scheme of the strategy employed to check for cdnL complementation in M. xanthus. A pMR2873 construct with the required cdnL allele (cdnL*) under P_cdnL control and DNA segments flanking cdnL upstream (grey) and downstream (black) in the genome, was introduced into strain MR1467, which bears the ΔcdnL allele and a copy of cdnL at a heterologous site under the control of the photoinducible P_B promoter (expressed in the light but repressed in the presence of B₁₂ in the dark). Merodiploids resulting from plasmid integration by recombination at either “1” or “2” would constitutively express cdnL* from P_cdnL and conditionally express the wild-type allele from P_B at a heterologous site. (B) Complementation analyses in M. xanthus with the F36A, M49A and P51A CdnL mutants. CTT/B₁₂ plates that were spotted with 5 µl of liquid cultures (OD₅₅₀ ∼1) grown under permissive conditions and diluted as indicated, and then incubated for 2 days at 33°C under restrictive (dark and B₁₂) or permissive (light, hence the red color) conditions. “WT” is the positive control derived from using pMR2873 with wild-type cdnL, and “ΔcdnL” is the recipient strain MR1467. Western blots (using polyclonal anti-CdnL antibodies) of M. xanthus cell extracts from strains expressing each CdnL variant and in which CdnL-eGFP supplied the essential CdnL function, as described in the text. (C) BACTH analysis of the interaction of CdnL homologs (BbCdnL, ScCdnL, and CgCdnL) with Bbβ_31–538, Scβ_29–426, and Cgβ_32–428 fragments, respectively. “+” below the open bars indicates that both members of the pair are present, while “−” is the negative control (without the indicated CdnL homolog). We have shown the interaction of TtCdnL with its cognate β-fragment elsewhere . (D) BACTH analysis of the interaction of BbCdnL, ScCdnL, CgCdnL or TtCdnL with M. xanthus Mxβ_19–537.

Figure 6. Consequences in vivo of mutating the CdnLCt solvent-exposed basic-hydrophobic patch.

(A) CdnLCt electrostatic surface (left) and ribbon representations (right; each helix is labeled and colored differently) generated using PyMOL. The zoom shows sidechains (as sticks) of the labeled nonpolar and basic residues composing the solvent-exposed basic-hydrophobic patch. (B) Complementation analyses in M. xanthus of the CdnL W88A, M96A, F125A, R128A/K129A, and R90A/R91A/R93A mutants versus wild-type CdnL (WT) and ΔcdnL, as in Figure 5B.

Figure 7. CdnL favors RP_o formation and transcription from σ^A-dependent promoters.

(A) Promoter sequences for the σ^A-dependent P_4rrnD and P_B, indicating the -35 and -10 elements (in boxes) and the transcription start site. For comparison, the ECF-σ CarQ-dependent P_QRS promoter is also included, with the conserved motifs at the -35 and -10 regions underlined. (B) ChIP-qPCR analysis of CdnL enrichment at P_4rrnD, P_B and P_QRS in M. xanthus carried out as described in the text. The mean and standard of three independent experiments are shown. (C)–(E) EMSA of the binding of the ³²P-labeled 130-bp P_B or 151-bp P_4rrnD DNA probes to 130 nM RNAP-σ^A holoenzyme, alone or in the presence of 10 µM of CdnL, CdnLNt, CdnLCt, or CarDNt (C), or a given CdnL variant (D, E), as indicated. After incubation for 30-min at 37°C, the complexes formed were challenged by adding 1 µg heparin. Note that any alteration in the migration of the shifted complexes due to CdnL binding to RNAP was not discernible, possibly because CdnL is considerably smaller than RNAP. (F) Single round, run-off in vitro transcription from P_4rrnD (see Materials and methods) with 130 nM RNAP-σ^A alone or with 10 µM CdnL (WT or mutant) followed by heparin challenge (1 µg) and then addition of the labeled NTP mix to initiate transcription. In C–F, representative data from three or more experiments are shown.