Caulobacter crescentus CdnL is a non-essential RNA polymerase-binding protein whose depletion impairs normal growth and rRNA transcription

Abstract

CdnL is an essential RNA polymerase (RNAP)-binding activator of rRNA transcription in mycobacteria and myxobacteria but reportedly not in Bacillus. Whether its function and mode of action are conserved in other bacteria thus remains unclear. Because virtually all alphaproteobacteria have a CdnL homolog and none of these have been characterized, we studied the homolog (CdnL_Cc) of the model alphaproteobacterium Caulobacter crescentus. We show that CdnL_Cc is not essential for viability but that its absence or depletion causes slow growth and cell filamentation. CdnL_Cc is degraded in vivo in a manner dependent on its C-terminus, yet excess CdnL_Cc resulting from its stabilization did not adversely affect growth. We find that CdnL_Cc interacts with itself and with the RNAP β subunit, and localizes to at least one rRNA promoter in vivo, whose activity diminishes upon depletion of CdnL_Cc. Interestingly, cells expressing CdnL_Cc mutants unable to interact with the RNAP were cold-sensitive, suggesting that CdnL_Cc interaction with RNAP is especially required at lower than standard growth temperatures in C. crescentus. Our study indicates that despite limited sequence similarities and regulatory differences compared to its myco/myxobacterial homologs, CdnL_Cc may share similar biological functions, since it affects rRNA synthesis, probably by stabilizing open promoter-RNAP complexes.

Introduction

Bacteria adapt to changing conditions by controlling gene expression, and transcription initiation is the most frequently regulated step. Bacterial transcription begins when the RNA polymerase (RNAP) holoenzyme, consisting of the core (α₂ββ′ω) and a specific dissociable σ factor, recognizes a target promoter to form a closed complex (RP_c), which undergoes a series of conformational changes leading to open complex (RP_o) formation and RNA synthesis from the DNA template^1,2,3. Defined promoter elements and the choice of a given σ factor are fundamental determinants of target specificity. Nonetheless, numerous accessory factors often play key roles in modulating, positively or negatively, the various steps of transcription initiation.

An emerging class of bacterial RNAP-interacting transcriptional factors are members of the CarD-CdnL family. The global transcriptional regulator CarD, composed of an N-terminal RNAP-interacting domain and a DNA-binding domain resembling eukaryotic HMGA proteins, is found only in myxobacteria, where it is dispensable for cell viability, and has been linked to the action of the alternative extracytoplasmic function (ECF) σ factors^4,5,6,7. By contrast, CdnL (for CarD N-terminal like) is a standalone version of the N-terminal domain of CarD, whose homologs (sometimes referred to as CarD) occur in several bacterial taxonomical groups^5,8,9,10. CdnL from M. xanthus (CdnL_Mx) and its Mycobacterium tuberculosis homolog (CdnL_Mt) are essential for cell viability^9,10, as may also be the case for the homolog in the spirochaete Borrelia burgdorferi¹¹. On the other hand, CdnL is reportedly not essential in Bacillus subtilis^12,13 or in Bacillus cereus¹⁴, suggesting that its function is not always conserved. In fact, many members of the phylum Firmicutes, to which B. subtilis and B. cereus belong, lack a CdnL homolog.

The critical function of CdnL is likely related to its role in promoting gene expression by stabilizing RP_o formation at promoters, such as those for rRNA, that depend on the primary σ factor^{15,16,17,18,19}. Like the CarD N-terminal domain, CdnL_Mx interacts with the RNAP β subunit (RNAPβ) but, while this interaction is dispensable for CarD, it is crucial for CdnL_Mx^16,20. The CdnL-RNAPβ interaction has also been demonstrated for CdnL_Mt²¹, CdnL_Tt in Thermus thermophilus^10,22, CdnL_Bb in Bdellovibrio bacteriovorus (a deltaproteobacterium like M. xanthus), CdnL_Cg in Corynebacterium glutamicum and CdnL_Sc in Streptomyces coelicolor, both actinobacteria like M. tuberculosis¹⁶. The B. subtilis homolog CdnL_Bs (also named YdeB), however, does not appear to interact with RNAPβ¹³.

To establish if CdnL has conserved functions, and to gain insights into structure-function relationships in this large family of proteins, it is imperative to characterize CdnL from diverse classes of bacteria. Caulobacter crescentus, a Gram-negative oligotrophic fresh water alphaproteobacterium, is an important model system for studies on bacterial cell cycle, division and differentiation²³. The alphaproteobacteria class includes members with very varied lifestyles and, although most of them have a CdnL homolog, none had been specifically studied. This prompted us to characterize the homolog in C. crescentus, CdnL_Cc (26–34% identical to homologs studied in other bacterial species; Supplementary Table S1, Supplementary Fig. S1). Our findings indicate that CdnL_Cc is required for normal growth in C. crescentus but is not essential for viability, in contrast to CdnL_Mt and CdnL_Mx. Our data show that CdnL_Cc is degraded in vivo and that this depends on its C-terminal AA motif, although stabilizing CdnL_Cc had no apparent effect on growth. We also show that CdnL_Cc binds to RNAPβ and localizes to at least one rRNA promoter in vivo, whose activity diminishes on limiting intracellular CdnL_Cc levels. Given these parallels with CdnL_Mx, CdnL_Mt, and CdnL_Tt, we propose that CdnL_Cc may also stabilize open rRNA promoter-RNAP complexes. Interestingly, we find that missense mutations of conserved residues of CdnL_Cc that impair the interaction with RNAPβ lead to a cold-sensitive in vivo phenotype, suggesting that CdnL_Cc interaction with RNAP is functionally more important at lower temperatures in C. crescentus. Our results extend the requirement of CdnL for normal cell growth and its possible functional roles in alphaproteobacteria and contribute to elucidating structure-function relationships underlying its mode of action.

Results

CdnL is widespread and highly conserved in alphaproteobacteria

A CdnL homolog was found in all representative alphaproteobacteria, except for the early diverging genus Magnetococcus, a clade of Rickettsiales comprising the genera Neorickettsia, Wolbachia, Ehrlichia and Anaplasma, and a single Rhodospirillales (Thalassobaculum_L2). A phylogenetic tree of these proteins (Supplementary Fig. S2) is globally consistent with the accepted phylogeny of alphaproteobacteria²⁴, suggesting that CdnL has mostly been transmitted vertically along the phylogeny without long-range lateral gene transfer. Given that other classes of proteobacteria have a CdnL homolog, the most parsimonious hypothesis is that the common ancestor of alphaproteobacteria had a CdnL homolog that was lost along one of the main branches in the Rickettsiales group and, independently, in Magnetococcus. The presence of CdnL in most alphaproteobacteria, with high overall amino acid sequence conservation (>50% identity and >90% coverage relative to CdnL_Cc) suggests strong purifying selection and an important cellular function. Studies with CdnL_Cc could thus help understand its role not only in C. crescentus but also in other alphaproteobacteria.

CdnL_Cc is not essential for viability but is required for normal cell growth

A recent global transcriptional start site (TSS) mapping study in C. crescentus²⁵ listed a single TSS for cdnL_Cc and assigned a putative −35 to −10 promoter segment (TTCATAG-x₁₂-GCTATTGT, where x is A, T, C, or G) similar to the consensus for promoters dependent on σ⁷³, the primary σ factor in C. crescentus (5′-TTGaCg(c/g)-x_11-14-GCtAxA(a/t)C-3′^26,27). We found that a reporter lacZ fusion to the 352-bp intergenic region upstream of cdnL_Cc, which includes the predicted σ⁷³-dependent promoter, showed high and increasing reporter β-galactosidase activity during exponential growth that leveled off in stationary phase (Supplementary Fig. S3). Interestingly and notably different from cdnL_Mx and cdnL_Mt, the mapped cdnL_Cc TSS suggests a long (233 nt) 5′untranslated region (5′-UTR), comparable to the 266-nt one reported for the alphaproteobacterium Sinorhizobium meliloti in another global TSS study²⁸.

A genome-wide transposon insertion analysis suggested that cdnL_Cc is essential, or at least has a strong fitness impact, when C. crescentus is cultivated in rich medium²⁹. Similar studies in other alphaproteobacteria have also listed cdnL as essential in Brevundimonas subvibrioides (Caulobacterales, like C. crescentus)³⁰ and Rhodopseudomonas palustris (Rhizobiales)³¹, but it was unclear in Agrobacterium tumefaciens³⁰ and in Rhizobium leguminosarum³² (both Rhizobiales). We therefore probed the functional importance of cdnL_Cc by first attempting to generate a markerless, in-frame cdnL_Cc deletion (ΔcdnL_Cc) with a two-step allele exchange strategy (see Supplementary Methods). Inability to obtain haploid cells with the ΔcdnL_Cc allele (none out of ∼75 colonies analyzed that grew after two days had the ΔcdnL_Cc allele) hinted that cdnL_Cc may be essential for viability. This was further supported by the ability to delete cdnL_Cc when a second functional copy, expressed under the control of the vanillate-inducible P_van promoter, was supplied at a heterologous chromosomal site: approximately a third of the colonies analyzed after the normal two-day growth now had the ΔcdnL_Cc allele (strain ME4). However, ME4 grew without vanillate in the medium, albeit slower than in the presence of the inducer (doubling time ∼140 min versus ∼90 min; Fig. 1a,b), suggesting leaky P_van expression in the absence of vanillate or that cdnL_Cc is not essential. To resolve this issue, we attempted to replace cdnL_Cc by a ΔcdnL_Cc::Ω allele (conferring spectinomycin/streptomycin resistance) using the two-step allele exchange strategy in the presence of a complementing plasmid, followed by generalized transduction of the Spec^R/Strep^R Ω cassette into the wild-type strain (see Supplementary Methods). A significant number of Spec^R/Strep^R colonies (30–50) grew ∼4–5 days after transduction, which were confirmed to have the ΔcdnL_Cc::Ω allele (strain JC784), suggesting that CdnL_Cc is in fact dispensable but its absence slows down growth (Fig. 1a,b). We therefore repeated our attempt to obtain a markerless ΔcdnL_Cc strain, plating increasing dilutions to enable detection of slow-growing colonies. Such colonies emerged after ∼4–5 days, and all of them had the ΔcdnL_Cc allele (strain ME50). It is unlikely that these ΔcdnL_Cc colonies appeared as a consequence of supressor mutations, since they were isolated at a frequency comparable to that observed when a complementing cdnL_Cc copy was present. Furthermore, both ΔcdnL_Cc strains (JC784 and ME50), even though generated independently, exhibited the same slow-growth behavior in rich medium (Fig. 1a), with similar doubling times (∼230 min; Fig. 1b), which were higher than for the wild type (∼90 min) or for ME4 in the absence of vanillate (∼140 min). Many JC784 and ME50 cells, and ME4 cells grown in the absence of vanillate, were filamentous indicating cell division defects, and the DNA visualized by DAPI fluorescence was often unevenly distributed (Fig. 1c).

Figure 1: Growth and cell morphology upon CdnL_Cc depletion.

(a) Growth of the indicated C. crescentus strains. Liquid cultures (OD₆₆₀ ∼ 0.5) were serially diluted, spotted (8 μl) on PYE plates with (+van) or without (−van) 0.5 mM vanillate, and incubated for 2 days at 30 °C. (b) Growth curves for the strains indicated (with symbols in parentheses). Freshly plated cells were innoculated into 10 ml of PYE (with 0.5 mM vanillate for ME4 and ME5) and grown at 30 °C to OD₆₆₀ ∼ 0.8. 50 μl were aliquoted (for ME4 and ME5, after washing three times with PYE to remove the vanillate) into 10 ml of fresh PYE (for ME4 and ME5, one with 0.5 mM vanillate and one without). Growth was monitored at the indicated times, and the average and error of three independent measurements is shown. (c) Cellular morphology of cells from (b). Samples (concentrated ten-fold in the case of ME5 cultivated in the absence of vanillate) of each culture in (b) after ∼20 h of growth were DAPI-stained and examined by DIC (differential interference contrast; top panels) and fluorescence microscopy (bottom panels), as described in Methods. Scale bar: 5 μm.

The faster growth rate of ME4 (with no vanillate) relative to JC784 or ME50, suggests leaky P_van-cdnL_Cc expression. To achieve tighter repression, we introduced a high-copy number plasmid expressing vanR, pBVMCS-6²⁷, into ME4. Growth of the resulting strain (ME5; Fig. 1a–c) in the absence of vanillate was more severely affected than that of ME4 but, curiously, even more so than that of the two ΔcdnL_Cc strains (JC784 or ME50). This was also observed when a high-copy number plasmid bearing vanR with a different antibiotic resistance marker (pBVMCS-2) was used, or even when such plasmids lacked vanR (Supplementary Fig. S4)²⁷. Thus, CdnL_Cc depletion appears to undermine the ability of cells to cope with the fitness burden due to the presence of these plasmids. The possibility of turning CdnL_Cc expression on or off in ME5, together with its more severe growth phenotype upon depleting CdnL_Cc, led us to use this strain for assessing functionality of CdnL_Cc variants in C. crescentus (below).

In sum, CdnL_Cc is not essential under standard growth conditions on rich medium but its absence or its depletion causes growth and morphological defects, and compromises cell fitness and ability to deal with stresses such as maintaining high-copy number plasmids in C. crescentus.

CdnL_Cc is degraded in vivo in a manner dependent on its C-terminus, but its stabilization does not impair cell growth

CdnL_Cc has a C-terminal AA motif (Supplementary Fig. S1), a hallmark of many substrates of the energy-dependent ClpXP protease, essential in C. crescentus, where it degrades several proteins implicated in replication, cell cycle or development^33,34. We therefore examined the stability of CdnL_Cc in vivo. For this, CdnL_Cc was fused to a FLAG epitope tag, to enable detection by immunoblot analysis. Since the tag can mask potential protease recognition, both N- and C-terminally tagged versions were tested. We could generate a normally growing strain expressing the C-terminally FLAG-tagged CdnL_Cc (CdnL_Cc-FLAG) as the sole CdnL_Cc copy, but not one with only the N-terminally FLAG-tagged CdnL_Cc (FLAG-CdnL_Cc), suggesting that an N-terminal tag impairs CdnL_Cc function. We therefore used strains expressing cdnL_Cc from its native locus (to ensure normal growth) and the given FLAG-tagged CdnL_Cc from a tightly controlled vanillate-inducible P_van promoter. To test if CdnL_Cc is degraded in a ClpX-dependent manner in vivo, we introduced, as reported previously³⁵, a plasmid (pM088) allowing xylose-inducible expression of the dominant-negative ClpX* mutant chaperone (which is altered in its ATP-binding site and produces a catalytically dead ClpX form)³⁶. Immunoblot analysis of FLAG-CdnL_Cc degradation after turning off its expression (Fig. 2a) revealed that its stability was enhanced about two-fold when ClpX* expression was induced (estimated half-life of 45 ± 3 min, versus 18 ± 4 min when ClpX* was not induced). Moreover, CdnL_Cc-FLAG and FLAG-CdnL_Cc(DD), with the C-terminal AA motif masked or replaced by two aspartates, respectively, were considerably more stable whether or not ClpX* expression was induced (half-lives of ∼300–580 min in vivo; Fig. 2a). Taken together, these results indicate that the AA motif at the C-terminus of CdnL_Cc is an important determinant for its degradation in vivo and suggest that this might be dependent, at least in part, on ClpX.

Figure 2: Analysis of CdnL_Cc protein stability in vivo.

(a) Stability of FLAG-CdnL_Cc, CdnL_Cc-FLAG and FLAG-CdnL_Cc(DD) in vivo. C. crescentus strains ME27, ME28 and ME29 were grown in PYE with vanillate, and subjected to a previously described protocol³⁵ (see Methods) prior to immunoblot analysis (−xyl: no xylose; +xyl: 0.3% xylose to induce clpX* expression). Total protein from 1 ml aliquots withdrawn every 20 min was detected in immunoblots using anti-FLAG antibodies (left). On the right is a semi-log plot of the relative band intensities (mean of three independent experiments) versus time for FLAG-CdnL_Cc (−xyl, circles; +xyl, diamonds), CdnL_Cc-FLAG (−xyl, triangles) and FLAG-CdnL_Cc(DD) (−xyl, squares). Slopes of the linear fits shown yield the decay rate constants used to estimate half-lives. (b) FLAG-CdnL_Cc levels during the cell cycle. Swarmer cells (SW) from strain ME24 grown in M2G with vanillate were isolated and used for synchronized cell cycle progression (∼150 min doubling time). Total protein from 1 ml aliquots taken every 20 min was subjected to immunoblot analysis using anti-FLAG antibodies. The control CtrA was probed on a separate blot (since its gel mobility is close to that of FLAG-CdnL_Cc) using anti-CtrA antibodies³⁸ and equivalent samples from the same experiment processed in parallel. Samples at 150 min correspond to SW and ST (with a small proportion of PD) cells isolated from the culture remaining at the end of this assay. A plot of band intensities (% of the maximum value) versus time is shown (right). In (a,b), positions of molecular size markers (kDa) are shown in blue to the right of cropped immunoblots. Note that FLAG-tagged CdnL_Cc (∼19.5 kDa calculated Mw) migrates slower than expected. (c) Growth of C. crescentus strains (ME41, ME39, ME8) expressing CdnL_Cc, CdnL_Cc(DD) or CdnL_Cc-FLAG, respectively, at the endogenous site. Liquid cultures (OD₆₆₀ ∼ 0.5) were serially diluted, spotted (8 μl) on PYE plates with or without vanillate, and incubated for 2 days at 30 °C. (d) Growth curves of strains in (c) expressing CdnL_Cc (squares, red), CdnL_Cc-FLAG (triangles, blue) or CdnL_Cc(DD) (circles, green) in PYE without vanillate at 30 °C.

C. crescentus undergoes asymmetric cell division to produce a non-replicative swarmer cell (SW) and a stalked cell (ST) that reinitiates the replicative cell cycle. Proper cell cycle progression depends on regulated proteolysis, often mediated by ClpXP^33,35. The aberrant cell division phenotype caused by CdnL_Cc depletion and the observation that it is subject to proteolysis prompted us to examine the levels of FLAG-CdnL_Cc (expressed from P_van) along the cell cycle. As a control for synchronization and cell cycle progression, we also monitored the changes in the levels of CtrA, a master cell-cycle regulator and known ClpXP substrate, which is stable in SW cells, then gets degraded in ST cells and again reappears in predivisional (PD) cells^34,37,38. FLAG-CdnL_Cc increased from barely detectable levels in SW cells to peak at ∼80 min (Fig. 2b). Interestingly, the clearance of FLAG-CdnL_Cc from SW cells resembles that reported for the cell division protein FtsZ, a ClpXP and ClpAP substrate that, like CdnL_Cc, is stabilized by tagging its C-terminus or by mutating it to a DD motif³⁵.

The normal growth in the absence of vanillate of strain ME8, which expresses CdnL_Cc-FLAG from the native promoter and also bears a copy of cdnL_Cc under P_van control, suggests that enhancing the stability of CdnL_Cc and its resulting accumulation in vivo are not toxic to the cells (Fig. 2c). We confirmed this further with cells expressing only untagged CdnL_Cc(DD) that, like equivalent strains expressing only CdnL_Cc or CdnL_Cc-FLAG, grew normally in the absence of vanillate on plates (Fig. 2c) or in liquid media (Fig. 2d). Thus, curiously, even though CdnL_Cc is targeted for degradation in vivo, preventing this degradation by mutating the AA motif to DD or masking it with a C-terminal tag does not appear detrimental to C. crescentus growth or viability.

CdnL_Cc interacts with itself and with RNAPβ

Full-length CdnL_Cc is overall basic like CdnL_Tt, and not acidic like CdnL_Mx or CdnL_Mt (Supplementary Table S1). High-resolution tertiary structures determined for CdnL_Tt, CdnL_Mt and CdnL_Mx revealed a ∼70-residue N-terminal β-sheet module and a C-terminal α-helical domain comprising the rest of the protein^{16,18,19,22,39,40} that correspond well with sequence-based predictions of secondary structure (PSIPRED; http://bioinf.cs.ucl.ac.uk/psipred). Compared to its homologs, the putative CdnL_Cc N-terminal domain is 32–39% identical and is also acidic (only in CdnL_Tt it is basic), whereas its C-terminal domain, only 22–31% identical to the rest, is markedly basic, as in CdnL_Bs (Supplementary Table S1). Despite the low sequence identity and the divergent overall domain charge distribution, the predicted CdnL_Cc secondary structure (Supplementary Fig. S5) mirrors those in the high-resolution structures of CdnL_Tt, CdnL_Mt and CdnL_Mx.

Interaction with cognate RNAPβ, a hallmark of most CdnL homologs studied thus far, was mapped to a solvent-exposed surface on the N-terminal β1 lobe of RNAPβ and to the ∼70-residue N-terminal module in CdnL^9,16,21,39, which also mediates self-interactions^16,40. Bacterial two-hybrid analysis (BACTH) confirmed that CdnL_Cc conserves the ability to self-interact (Fig. 3a). A ~120 residue fragment of RNAPβ corresponding to the β1a subdomain (where residues critical for interaction with other CdnL homologs are located) was sufficient to detect interaction between CdnL_Mx or CdnL_Tt and their respective RNAPβ in BACTH^9,10. However, in similar assays with other CdnL homologs, a longer ∼500-residue fragment corresponding to the whole β1 domain (subdomains β1a and β1b) was required, likely because β1a alone does not always constitute a stable, well-folded domain^16,21. Thus, to test the interaction between CdnL_Cc and RNAPβ, a fragment encompassing only β1a (β_16-214) or the whole β1 domain (β_16-523) was used. Increased β-galactosidase activity with the β_16-523 fragment (Fig. 3b) suggests that CdnL_Cc conserves the interaction with RNAPβ. To corroborate this interaction with RNAP in C. crescentus, we performed coimmunoprecipitation experiments with cells expressing CdnL_Cc-FLAG and, as a control, with equivalent cells expressing untagged CdnL_Cc (see Methods). Detection of a band corresponding to RNAPβ by immunoblot analysis of cells expressing CdnL_Cc-FLAG confirmed the CdnL_Cc-RNAP interaction in C. crescentus (Fig. 3c).

Figure 3: CdnL_Cc interacts with itself, with RNAP, and with DNA.

(a) BACTH analysis of CdnL_Cc self-interaction in E. coli BTH101 transformed with plasmids pUT18-cdnL_Cc and pKT25-cdnL_Cc. (b) BACTH analysis of the interaction between CdnL_Cc (in pKT25) and C. crescentus RNAP β subunit fragments β_16-214 and β_16-523 (in pUT18C). In (a,b), the negative control (-) was pKT25 without insert. (c) Western blot of immunoprecipitated CdnL_Cc-FLAG probed for the presence of coprecipitating RNAPβ. Cells expressing CdnL_Cc-FLAG (strain ME17: ΔcdnL_Cc, P_van::cdnL_Cc-flag, vanR) or, as the negative control, untagged CdnL_Cc (strain ME5: ΔcdnL_Cc, P_van::cdnL_Cc, vanR) were immunoprecipitated with anti-FLAG agarose and processed in parallel, as described in Methods. Equal amounts of sample were then resolved by SDS-PAGE for immunoblot analysis. Monoclonal anti-RNAP β antibodies were used to detect RNAPβ (top) that coimmunoprecipitated with CdnL_Cc-FLAG, which was detected using anti-FLAG antibodies (bottom). Molecular size markers are shown to the right of the cropped immunoblots by the lines and corresponding values in kDa. (d) EMSA for the DNA binding of CdnL_Cc, CdnL_Tt, and CdnL_Mx. Reactions were performed as described in Methods with a 350-bp P_rrnA DNA probe, and with 1 μg of poly[dG-dC] or poly[dI-dC] added as nonspecific competitor.

CdnL_Cc interacts nonspecifically with dsDNA

CdnL_Mt has been reported to bind DNA nonspecifically in vitro through a positively charged patch at its C-terminal domain³⁹, although in vivo it was not found on the genome in the absence of RNAP¹⁸. However, direct DNA binding has never been observed for CdnL_Mx, which was proposed to associate with DNA exclusively via its interaction with RNAP^9,16. CdnL_Mx is overall acidic, as is CdnL_Mt (Supplementary Table S1), whereas CdnL_Cc is basic and could possibly bind to DNA and other polyanions. We therefore compared the DNA binding behaviour of CdnL_Cc, CdnL_Tt (also basic) and CdnL_Mx to a 350-bp double-stranded DNA probe that includes the promoter region of rrnA (P_rrnA), one of the two rRNA operons in C. crescentus. The probe incubated with CdnL_Cc or CdnL_Tt and without nonspecific competitor DNA present precipitated in the loading well (not shown) suggesting nonspecific DNA binding. In the presence of nonspecific competitor DNA, a smeared retarded band could be discerned at high protein concentrations (5–10 μM) with CdnL_Cc or CdnL_Tt, but not with CdnL_Mx, but most of the labeled DNA probe remained free (Fig. 3d). We mutated to Ala three basic residues in the C-terminal domain of CdnL_Cc (Arg92, Arg93, Arg130) that align with those in the basic patch mentioned above (Supplementary Fig. S1). Binding to the DNA probe was weakened on mutating Arg130, and abolished on mutating both Arg92 and Arg93 (Supplementary Fig. S6a). Similar results were observed with a randomly chosen intragenic DNA probe in vitro (Supplementary Fig. S6b). Altogether, these data suggest that CdnL_Cc can bind nonspecifically to DNA in vitro, largely via electrostatic interactions.

CdnL_Cc depletion impairs rRNA transcription in C. crescentus

Our results thus far indicate that CdnL_Cc is required for normal cell growth and that it interacts with RNAPβ, like most of its homologs that have been studied thus far. These have been shown to localize at promoters dependent on the primary σ factor, such as those for rRNA, and to activate them^{15,16,17,18,19}. Hence, we used quantitative chromatin immunoprecipitation (ChIP) with anti-FLAG antibodies to probe if CdnL_Cc localizes to P_rrnA (the rrnA operon promoter region) in C. crescentus by using cells expressing CdnL_Cc-FLAG (strain ME17) or CdnL_Cc as negative control (strain ME5). This assay demonstrated that, relative to an intragenic region, CdnL_Cc was indeed enriched at P_rrnA, as were the primary σ factor in C. crescentus (σ⁷³) or RNAPβ used as positive controls (Fig. 4a). We also found that CdnL_Cc was enriched at two other σ⁷³-dependent promoters but not at one requiring σ^F, an alternative ECF σ factor in C. crescentus^25,41,42 (Supplementary Fig. S7). This suggests that CdnL_Cc, like the homologs studied, is associated with promoters that depend on the primary σ factor.

Figure 4: CdnL_Cc localizes at an rRNA promoter and affects its transcription in vivo.

(a) CdnL_Cc binds to P_rrnA in vivo. ChIP-qPCR analysis using an anti-FLAG antibody on cells expressing CdnL_Cc-FLAG (strain ME17: ΔcdnL_Cc, P_van::cdnL_Cc-flag, vanR; unfilled bars) or CdnL_Cc as negative control (strain ME5: ΔcdnL_Cc, P_van::cdnL_Cc, vanR; black bars) showing CdnL_Cc-FLAG enrichment at P_rrnA in vivo, relative to an intergenic region. As positive controls, ChIP-qPCR analysis was carried out using anti-RNAP β or anti-σ^A monoclonal antibodies for enrichment of RNAP or σ⁷³, respectively, at P_rrnA. (b) Effects of CdnL_Cc depletion on P_rrnA promoter activity in vivo. CdnL_Cc was expressed from the P_van promoter in strain ME42. Cells grown in M2G with vanillate to OD₆₆₀~0.4 were washed and then grown in vanillate-free medium to block cdnL_Cc expression, and activity was measured by qRT-PCR at the times indicated. At each time point following vanillate withdrawal, one-half of the sample was washed and resuspended in medium without glucose and the other half remained untreated. Following 15 min incubation at 30 °C, the samples were collected for qRT-PCR analysis. (c) qRT-PCR analysis carried out with strain ME40, which expresses CdnL_Cc-FLAG from the P_van promoter, using a procedure identical to that in (b). Data shown in (a–c) correspond to the mean and standard error from three biological replicates. (d) Immunoblot analysis of CdnL_Cc-FLAG corresponding to samples in (c) with and without glucose deprivation and at the times indicated following vanillate withdrawal (top). As loading control, the same blot was probed using polyclonal anti-RNAP antibodies; the band corresponding to the RNAP β, β′ subunits is shown (bottom). Molecular size markers are shown to the right of the cropped immunoblots by lines and corresponding values in kDa.

Next, we examined if CdnL_Cc affects transcription in vivo, focusing on P_rrnA. We used a C. crescentus strain (ME42) that expresses cdnL_Cc under P_van control and bears a transcription reporter plasmid with P_rrnA and the first 81 nucleotides of its leader fused to a lacZ gene fragment. This design is based on a previous study⁴³, which showed that transcription from this P_rrnA reporter was rapidly downregulated on glucose starvation, and proposed that this was mediated by an unknown factor. Such a factor could, in principle, be one that activates rRNA transcription but disappears rapidly on glucose starvation to swiftly lower rRNA transcription. CdnL_Cc could potentially be this factor if it activates rRNA transcription, and if it becomes unavailable on glucose deprivation. Hence, we tested the effects of depleting CdnL_Cc on P_rrnA transcription in vivo and whether this was affected on glucose deprivation.

Cells were grown to mid-log phase with vanillate, after which the inducer was eliminated to restrict cdnL_Cc expression. This would result in a progressive drop in CdnL_Cc levels over time due to intracellular degradation. Reporter P_rrnA transcription, estimated by qRT-PCR, decreased gradually after removal of vanillate, suggesting that CdnL_Cc is directly or indirectly required for P_rrnA expression (Fig. 4b). Glucose deprivation caused a sharp drop in rRNA transcription, even right after removal of the inducer (Fig. 4b). To test if this is due to rapid degradation of CdnL_Cc, somehow triggered by glucose starvation, we repeated the above analysis in a strain (ME40) expressing the considerably more stable, yet functional, CdnL_Cc-FLAG. Rather than gradually decreasing over time, relative transcript levels observed in the presence of glucose now remained fairly steady over the 8 h period (Fig. 4c), consistent with CdnL_Cc-FLAG persisting even 8 h after removal of the inducer, albeit at lower levels (Fig. 4d). P_rrnA transcription again dropped dramatically on glucose starvation (Fig. 4c), even though CdnL_Cc-FLAG levels, at all time points, were comparable with or without glucose deprivation (Fig. 4d). Taken together these results indicate that depletion of CdnL_Cc impairs rRNA expression, suggesting that CdnL_Cc directly or indirectly promotes rRNA transcription in vivo, but that factors and mechanisms other than a rapid loss of CdnL_Cc likely determine the sharp fall in rRNA expression occurring on glucose deprivation.

Missense mutations at conserved CdnL_Cc residues cause cold sensitivity

The results thus far establish that CdnL_Cc interacts with itself and with RNAPβ, localizes at P_rrnA and affects its activity. To test if the interaction with RNAPβ is required for CdnL_Cc function in C. crescentus, two classes of mutations were generated, mimicking those reported previously in other CdnL homologs: mutations in the N-terminal module that disrupt the interaction with RNAP and those in the C-terminal domain that leave the interaction with RNAP intact^16,18,21,22.

CdnL_Cc N-terminal residues Val39, Arg52, and Pro54 (Supplementary Fig. S1) were mutated to Ala, since equivalent mutations in CdnL_Mx, CdnL_Mt or CdnL_Tt caused loss of interaction with RNAPβ and, where tested, impaired cell growth^16,22,39. BACTH indicated significantly reduced interaction of β_16-523 with mutants V39A and P54A, but not with R52A (Fig. 5a), indicating that CdnL_Cc conserves at least two of the expected contacts with RNAPβ. To test the effect of the mutations in C. crescentus, plasmids bearing a given allele (expressing the protein with a C-terminal FLAG tag) flanked by ∼500 bp genomic DNA upstream and downstream of cdnL_Cc were introduced into strain ME5 (Fig. 5b). Transformants with plasmids integrated by homologous recombination at the endogenous cdnL_Cc site were isolated in the presence of vanillate and then examined after removal of the inducer. On plates lacking vanillate, all three mutants grew normally at 30 °C (Fig. 5c). Notably, V39A and P54A functioned in C. crescentus despite their inability to interact with RNAPβ. By contrast, such mutations were lethal in M. xanthus¹⁶ and in M. tuberculosis, but apparently not in M. smegmatis, closely related to M. tuberculosis²¹. Interestingly, we noticed that lowering the growth temperature from 30 °C to 25 °C under vanillate-free conditions caused significant growth arrest of V39A and P54A, while R52A, which continues to interact with RNAPβ, still grew normally (Fig. 5c). Consistent results were obtained for growth in liquid media without vanillate (Fig. 5d), with several cells of the poorly growing V39A and P54A mutants at 25 °C having the aberrant elongated cell morphology of a ΔcdnL_Cc strain (Fig. 5e). All of the mutant proteins were stable in vivo at both growth temperatures (Fig. 5f), implying that loss of function of V39A and P54A at the lower temperature does not stem from protein instability. This behavior does not appear to be caused by increased stability of CdnL_Cc due to the C-terminal FLAG tag, since the cold sensitivity was also observed with the representative untagged P54A mutant (Supplementary Fig. S8). Thus, interaction with RNAPβ does not appear to be critical for CdnL_Cc function in C. crescentus except at lower than standard growth temperatures.

Figure 5: Mutational analysis of CdnL_Cc-RNAP interaction in vivo.

(a) BACTH analysis of the interaction of CdnL_Cc mutants V39A, R52A and P54A (in pKT25) with C. crescentus RNAPβ fragment β_16-523 (in pUT18C). The negative control (–) bears empty pKT25 and pUT18C-β_16-523. (b) Schematic for the strategy employed to check for cdnL_Cc complementation in C. crescentus. A pMR3552 derivative with the required cdnL_Cc-flag allele (* indicates mutant) flanked by DNA segments upstream (grey) and downstream (black) of cdnL_Cc in the genome was introduced into strain ME5, which bears the ΔcdnL_Cc allele at the endogenous site and P_van-cdnL_Cc at a heterologous site. Merodiploids resulting from plasmid integration by recombination express both CdnL_Cc*-FLAG and CdnL_Cc in the presence of vanillate (+van) and only the former in the absence of vanillate (−van). (c) Complementation analysis in C. crescentus of cells bearing the ΔcdnL_Cc allele or ones expressing at the endogenous site C-terminal FLAG-tagged wild-type CdnL_Cc (WT) or the indicated N-terminal CdnL_Cc variants. PYE plates with (+van) or without (−van) vanillate were spotted with 8 μl of liquid cultures (OD₆₆₀ ∼ 0.5) at the dilutions indicated and incubated at 30 °C for two days or at 25 °C for three days. (d) Growth curves at 30 °C or 25 °C of C. crescentus expressing C-terminal FLAG-tagged CdnL_Cc (WT) or its indicated variants cultivated in liquid PYE without vanillate using the procedures described in Fig. 1b. (e) Cellular morphology examined by DIC microscopy of the wild-type (WT; scale bar: 5 μm) and the indicated mutant cells from (d) grown at 30 °C or 25 °C. (f) Immunoblot analysis to probe the stability of N-terminal CdnL_Cc variants. Cell extracts of strains expressing C-terminal FLAG-tagged CdnL_Cc (WT) or its indicated variants grown at 30 °C or 25 °C in PYE with vanillate were probed using anti-FLAG antibodies (top). As loading control, the same blot was probed using polyclonal anti-RNAP antibodies; the band corresponding to the RNAP β, β′ subunits is shown (bottom). Molecular size markers are shown to the right of the cropped immunoblots by lines and corresponding values in kDa (in blue).

We next examined the effect of mutating Trp90, Arg92, Arg93, Tyr127 or Arg130 (Supplementary Fig. S1) in the CdnL_Cc C-terminal domain, which is markedly basic and shares low sequence identity with its acidic counterparts in CdnL_Mx, CdnL_Mt and/or CdnL_Tt (Supplementary Table S1). Nonetheless, CdnL_Cc residues Trp90, Arg93, and Arg130 are conserved in all four homologs, Arg92 is conserved in CdnL_Mx and CdnL_Mt, and Tyr127 is a Phe in CdnL_Mx. With CdnL_Mx, CdnL_Mt or CdnL_Tt, mutational data had shown these residues to be functionally important in at least one of the homologs; and high-resolution structures revealed the residues to be part of a solvent-exposed basic-hydrophobic patch^16,18,39,40. Moreover, in the crystal structure of the RP_o-CdnL_Tt complex, Trp86 was in a position to interact with a highly conserved thymine (T₁₂) at the upstream edge of the DNA bubble, and act as a wedge to prevent bubble collapse; and CdnL_Tt activity in vitro was shown to require Trp86 and T₁₂¹⁹. Interestingly, mutating this highly conserved tryptophan impaired growth in M. tuberculosis but not in M. smegmatis⁴⁴ or in M. xanthus¹⁶.

The CdnL_Cc C-terminal mutants tested grew normally at 30 °C in the absence of vanillate, except for the double R92A/R93A mutant, which did show a growth defect (Fig. 6a). In comparison, the M. xanthus mutant corresponding to CdnL_Cc R92A/R93A also grew very poorly while that equivalent to Y127A showed somewhat impaired growth¹⁶, and single mutations of a number of these residues caused lethality in M. tuberculosis but not in the closely related M. smegmatis⁴⁴. Lowering the C. crescentus incubation temperature to 25 °C exacerbated the phenotype of the R92A/R93A mutant, caused somewhat deficient growth of W90A and Y127A, and a marked decrease in growth for R130A (Fig. 6a). Consistent results were obtained in liquid cultures lacking vanillate (Fig. 6b), and several cells of the mutants growing poorly at 25 °C had the abnormal elongated morphology (Fig. 6c). Again, the cold sensitivity does not appear to be due to protein instability (Fig. 6d), or entirely to protein stabilization resulting from the presence of the C-terminal tag, since the R130A mutant without the tag remained cold sensitive, albeit to a lower extent (Supplementary Fig. S8). Relative levels of reporter rRNA estimated by qRT-PCR were comparable for R130A and wild type at 30 °C but dropped significantly for the mutant at 25 °C (Fig. 6f), suggesting that impaired growth correlates with diminished rRNA transcription.

Figure 6: Analysis of C-terminal CdnL_Cc mutations in vivo.

(a) Complementation analysis in C. crescentus of cells bearing the ΔcdnL_Cc allele or ones expressing at the endogenous site C-terminal FLAG-tagged wild-type CdnL_Cc (WT) or the indicated C-terminal CdnL_Cc variants. The analysis was carried out using the same procedures and conditions described in Fig. 5c. (b) Growth curves at 30 °C or 25 °C of C. crescentus strains expressing the C-terminal FLAG-tagged CdnL_Cc (WT) or its indicated variants cultivated in liquid PYE without vanillate using the procedures described in Fig. 1b. (c) Cellular morphology examined by DIC microscopy of the wild-type (WT; scale bar: 5 μm) and the indicated mutant cells from (b) grown at 30 °C or 25 °C. (d) Immunoblot analysis to probe the stability of C-terminal CdnL_Cc variants. Cell extracts of strains expressing C-terminal FLAG-tagged wild-type CdnL_Cc (WT) or the indicated CdnL_Cc mutants grown at 30 °C or 25 °C in PYE with vanillate were probed using anti-FLAG M2 antibodies (top). The negative control “Δ” corresponds to the strain (ME5) with the ΔcdnL_Cc allele at the endogenous site and expressing untagged CdnL_Cc under P_van. As loading control, the same blot was probed using polyclonal anti-RNAP antibodies; the band corresponding to the RNAP β, β′ subunits is shown (bottom). Molecular size markers are shown to the right of the cropped immunoblots by lines and corresponding values in kDa (in blue). (e) P_rrnA promoter activity in vivo at 30 °C or at 25 °C in cells expressing CdnL_Cc-FLAG (WT) or its variant with the R130A mutation under P_van control (strains ME40 and ME38, respectively). Cells grown overnight at 30 °C in M2G with vanillate were diluted into the same medium to OD₆₆₀ ∼ 0.1. One-half was grown at 30 °C and the other at 25 °C to OD₆₆₀ of 0.3–0.4, and RNA was quantitated using qRT-PCR. Data shown correspond to the mean and standard error from three biological replicates.

In sum, our data suggest that both classes of functionally important mutations are conserved in CdnL_Cc but they cause cold-sensitive phenotypes in C. crescentus.

Discussion

Experimental validation of function for different proteins of the same family is required not only to gain insights into structure-function relationships but also because of several examples of apparently sequence-related proteins shown to be functionally distinct. Thus, CarD and CdnL have different functions, despite various shared structural features and interactions. CdnL_Mx and CdnL_Mt are similar in their essentiality, structure, and function in stabilizing RP_o formation and activating rRNA transcription, whereas CdnL_Bs lacks these properties. The fact that nearly all alphaproteobacteria have a CdnL homolog, with those of C. crescentus and a few other alphaproteobacteria being listed as essential based on genome-wide Tn-Seq studies^29,30,31, prompted us to study CdnL_Cc. We have shown here that CdnL_Cc is not essential, unlike CdnL_Mt and CdnL_Mx, but that its depletion leads to slow growth and cell filamentation in C. crescentus. The slow growth probably explains why CdnL_Cc was listed among the essential C. crescentus proteins²⁹, just as in the case of SsrA, GcrA and CcrM, which were initially deemed to be essential in C. crescentus but subsequently shown to be dispensable^45,46,47. Despite this key difference from its myco/myxobacterial homologs, CdnL_Cc nevertheless conserves the interaction with RNAP and other functional determinants, and directly or indirectly affects rRNA transcription, reinforcing the evidence for an important, broadly conserved regulatory role of CdnL in bacteria.

Both cdnL_Mx and cdnL_Mt are expressed from a primary σ-dependent promoter with a short 5′-UTR. While cdnL_Mt is sharply and quickly upregulated upon starvation or exposure to some other stresses¹⁰, this has not been observed for cdnL_Mx⁹. On the other hand, cdnL_Bs and B. burgdorferi cdnL expression was reportedly induced at lower temperatures from promoters that remain uncharacterized^11,48. We found that expression of cdnL_Cc increases during exponential growth and appears to be maintained in stationary phase. Global 5′-RACE data for C. crescentus²⁵ and S. meliloti²⁸ list their cdnL homologs among those genes dependent on the primary σ factor, but unlike cdnL_Mx or cdnL_Mt, both alphaproteobacterial cdnL have 5′-UTRs >200-bp long. The S. meliloti cdnL 5′-UTR was reported to contain also a heat-shock σ^H promoter^28,49, but we did not find any appreciable change in cdnL_Cc expression upon heat shock (Supplementary Fig. S9), suggesting absence of a σ^H promoter in the cdnL_Cc 5′-UTR. We found that the intergenic region at the 5′ end of cdnL tends to be unusually long in alphaproteobacteria: >200 bp for >60% of the homologs with an overall median of 312 bp. Short 5′ intergenic sequences are only found in the Rickettsiales, which are well known for reductive evolution⁵⁰. A long 5′-UTR may therefore be a common feature for cdnL in alphaproteobacteria. Future studies could shed light on the exact role of the long 5′-UTR, if any, on cdnL_Cc expression.

Our data indicate that CdnL_Cc is targeted for proteolysis in vivo in a manner dependent on its C-terminal AA motif, and is stabilized against intracellular degradation if this motif is masked by a C-terminal tag or is mutated to DD. This is a characteristic of many ClpXP-dependent substrates, and CdnL_Cc degradation was reduced about two-fold upon expressing a dominant-negative ClpX variant that is known to inhibit ClpXP activity^35,36. While this suggests that CdnL_Cc may be degraded in a ClpXP-dependent manner in C. crescentus, we cannot rule out other mechanisms also operating, such as through the ClpAP protease, as has been observed for other proteins³⁵. CdnL_Cc levels (in cells expressing FLAG-CdnL_Cc under P_van control) were barely detectable in non-replicative SW cells but easily detectable in ST and PD cells, suggesting that CdnL_Cc may be subject to cell cycle-dependent proteolytic control, with a possible role in actively dividing cells. Interestingly, changes in CdnL_Cc levels during the cell cycle are akin to those reported for the cell division protein FtsZ, a ClpXP and ClpAP substrate that also becomes stabilized if its C-terminus is tagged or mutated to DD³⁵. We found that >80% of the alphaproteobacterial CdnL homologs have a C-terminal AA or VA, including those found in early diverging groups like the Rickettsiales. Hence, proteolysis of CdnL might be a common and conserved ancestral mechanism in alphaproteobacteria. CdnL_Mt, with a C-terminal AAAS has been shown to be a target of the essential Clp protease in M. tuberculosis, a phylum distinct from alphaproteobacteria⁵¹, suggesting that Clp-mediated CdnL turnover might be a general feature. If so, this would hint at a crucial role for degradation of CdnL for post-translational control of its intracellular levels. However, it appears that degradation of CdnL_Cc is not required for normal growth under standard conditions, but might instead be more important under some other untested conditions. Increased intracellular levels of CdnL in M. tuberculosis⁵¹ or M. xanthus⁹ are not harmful either. Cells thus appear to cope well with high CdnL concentrations under standard growth conditions.

Depleting CdnL_Cc correlates with decreasing rRNA transcription and hence rapid degradation of CdnL_Cc could, in principle, be a mechanism for the sharp drop in rRNA transcription that occurs in vivo upon glucose starvation in C. crescentus⁴³. Our finding that this decrease occurs even when CdnL_Cc is present at significant levels (due to dysregulation of its proteolysis) points to other mechanisms, as yet unidentified, for downregulating rRNA expression upon glucose limitation. An interesting parallel to this has been reported for E. coli DksA, which is a ClpXP substrate⁵² that inhibits rRNA transcription. Here, stabilization and a rapid accumulation of DksA on starvation could in theory swiftly reduce rRNA expression, but this is actually achieved by modulating the amounts of small cofactors like (p)ppGpp and NTPs⁵³.

An interesting finding of our study is the cold-sensitive phenotype associated with missense mutations in CdnL_Cc. Cold sensitivity was observed with N-terminal CdnL_Cc mutations that disrupt interaction with RNAPβ, as well as with C-terminal ones that retain this interaction. By contrast, equivalent mutations in CdnL_Mx or CdnL_Mt impaired the essential function of CdnL even under standard growth conditions^16,21,44. In vitro and structural studies have linked CdnL function to RP_o stabilization by preventing collapse of the transcriptional bubble^{15,16,17,18,19}. Defective RP_o stabilization by CdnL mutants with the N-terminal mutations could be attributed to their lack of RNAP-binding and hence poor recruitment to RP_o (or RP_c), and by CdnL mutants with C-terminal mutations to their inability to accelerate DNA opening and inhibit bubble collapse (they interact with RNAP and are recruited to RP_o and RP_c like wild-type CdnL)¹⁷. Since CdnL_Cc, like CdnL_Mt or CdnL_Mx, binds to an rRNA promoter and affects its expression in vivo, it may also function in stabilizing RP_o formation. If so, temperature-dependent effects on RP_o formation and intrinsic differences between CdnL homologs and/or RNAP could possibly underlie the cold sensitivity displayed by CdnL_Cc point mutants. Lower temperatures are known to disfavor RP_o, which forms via a series of conformational changes, starting from RP_c assembly to melting of the promoter region between positions −11 and +2 relative to the TSS^2,3. Also, the stabilizing effect in vitro of CdnL_Mt on RP_o is more noticeable at 25 °C than at lower (10 °C) or higher (37 °C) temperatures, presumably because at 25 °C the energy landscape between RP_c and RP_o is more balanced and the additional binding energy provided by CdnL_Mt significantly drives the equilibrium towards RP_o formation^15,17.

Our analysis highlights differences between CdnL_Cc and its homologs, consistent with CdnL-RNAP interactions being species-specific¹⁶, but how these translate into the observed cold-sensitive phenotype remains elusive. Significant differences in the stability of the complexes formed by different RNAP on the same promoters reveal mechanistic differences across bacterial species. Thus, the occurrence of CdnL has been correlated to the cognate RNAP forming an unstable RP_o and, tellingly, CdnL is absent in E. coli whose RNAP appears to form more stable RP_o than mycobacterial RNAP¹⁵. Relative to M. xanthus and mycobacteria, C. crescentus RNAP could form intrinsically more stable RP_o, such that lower temperature, on its own unfavorable for RP_o formation, is necessary to detect the debilitating effects of CdnL_Cc mutations on RP_o stabilization. In vitro analyses using purified C. crescentus RNAP and CdnL_Cc, an aim for future work, should elucidate this further. Species-specific variations will necessarily have to be invoked to rationalize why B. subtilis, with an RNAP that forms unstable RP_o⁵⁴, has a CdnL that is not essential and that does not bind to RNAP or regulate rRNA transcription^12,13.

In conclusion, our study indicates that C. crescentus CdnL is not essential but is nonetheless required for normal growth and morphology, and is likely to be involved in rRNA transcription. The functional role of CdnL therefore appears to be fairly conserved in bacteria. It also highlights species-specific mechanistic differences for this factor relative to some of its homologs, within an otherwise preserved mode of action, that will be useful in understanding the structure-function relationships governing this class of important, RNAP-binding transcriptional regulators.

Methods

Strains, plasmids, and growth conditions

Strains and plasmids used in this study are listed in Supplementary Tables S2 and S3, respectively. Growth conditions, and strain and plasmid construction are detailed in Supplementary Methods in the Supplementary information file.

Bacterial two-hybrid (BACTH) analysis

The E. coli BACTH system used is based on functional complementation of the T25 and T18 fragments of the Bordetella pertussis adenylate cyclase catalytic domain when two test proteins interact⁵⁵. Coding regions selected were PCR-amplified and cloned into the XbaI and BamHI sites of pKT25, pUT18 or pUT18C (Supplementary Table S3). Given pairs of pKT25 and pUT18/pUT18C constructs were electroporated into E. coli strain BTH101 (cya⁻), a pair with a vector lacking an insert was the negative control. Interaction was assessed from measurements (mean and standard error of at least three experiments) of β-galactosidase specific activity (β-gal activity, in nmol of o-nitrophenyl β-D-galactoside hydrolysed/min/mg protein) from liquid cultures, as described⁵⁶.

Protein purification

His₆-tagged CdnL_Cc and its variants were overexpressed from pET15b (Novagen) constructs as soluble, native proteins using procedures described for CdnL_Mx and CdnL_Tt¹⁶. After thrombin digestion to remove the His₆-tag, the sample was passed through a phosphocellulose column equilibrated with 100 mM NaCl, 50 mM phosphate pH 7.5, 2 mM β-mercaptoethanol, eluted at 0.4 M NaCl, purified by size-exclusion (Superdex200, GE Health Sciences), and concentrated with Amicon Ultra (10000 MWCO from Millipore). Protein concentrations were estimated from absorbance at 280 nm using ε₂₈₀(M⁻¹cm⁻¹) determined from the sequence (http://web.expasy.org/protparam/).

Microscopy

At different times during growth, samples were withdrawn and the fluorescent dye 4′-6-diamino-2-phenylindole (DAPI; 350 nm excitation maximum, 461 nm emission maximum) was added to a final concentration of 1 ng/μl. A 1 μl drop applied on 1% agarose pads of M2 salts was examined under a Nikon Eclipse 80i microscope equipped with a Plan Apo VC 100×/1.40 oil immersion objective, and a Hamamatsu ORCA-AG CCD camera. A Nikon UV-2E/C filter set was used for DAPI fluorescence. Images were processed with Metamorph 4.5 (Universal Imaging Group) and Photoshop 6.0 (Adobe Systems).

Western blot and in vivo degradation analysis

Immunoblot analysis in whole cell extracts of CdnL_Cc tagged with the FLAG epitope was carried out using standard procedures⁹. Total protein was estimated prior to the analysis and aliquots with equal amounts of protein were resolved in 10% SDS-PAGE gels, transferred to Hybond-ECL membranes, and probed using the ECL system and anti-FLAG M2 monoclonal antibodies (F3165, Sigma-Aldrich). As loading control, the same blot was probed for RNAP (subunits β/β′) using polyclonal B. subtilis RNAP holoenzyme antibodies⁵⁷. To test the possibility that CdnL_Cc is degraded in a ClpX-dependent manner in vivo, we employed a previously established protocol³⁵. Basically, the C. crescentus strain was grown in PYE with appropriate antibiotics and 0.5 mM vanillate overnight at 30 °C, diluted to an OD₆₆₀ = 0.1 into 20 ml of the same medium, grown to an OD₆₆₀ of 0.4 and divided into two 10 ml cultures. To one, glucose was added to 0.2% final concentration and, to the other, 0.3% xylose to induce clpX*. After 2 h, cells were harvested by centrifugation, washed with 10 ml inducer-free PYE, and suspended in 10 ml fresh inducer-free media. Aliquots (1 ml) withdrawn at 20-min intervals were subjected to immunoblot analysis using anti-FLAG M2 antibodies. Relative band intensities were quantified by densitometry using the ImageJ software program and recommended protocols (NIH). Briefly, bands plus background were selected and a profile plot was obtained for each band (peaks). The straight-line tool was used to minimize background noise by closing off each peak above the baseline of the corresponding profile plot and to adjust the closing at the base of the peak in the case of spill-over signals, and the wand tool was used to quantify the closed peaks. The average and standard error of three independent experiments were used in further analysis. The slope of a linear fit to a plot of the natural logarithm of the relative band intensity (in % normalized to the zero time point) versus time using Sigmaplot (Systat Software Inc) yielded the decay rate and error, and the half-life was determined as ln(2)/-rate. To examine changes during the cell cycle (see Supplementary Methods), aliquots were withdrawn from synchronized cell cultures expressing FLAG-CdnL_Cc (strain ME24 grown in the presence of 0.5 mM vanillate) every 20 min, inspected by microscopy for progression of the cell cycle, and analyzed in Western blots using anti-FLAG M2 or anti-CtrA antibodies³⁸, as a control for the synchronization protocol.

Electrophoretic mobility shift assays (EMSA)

EMSA was carried out as described previously¹⁶. EMSA samples (20 μl) contained 1 nM ³²P-5′-end radiolabeled, double-stranded DNA probe (13,000 cpm) obtained by PCR and 5 or 10 μM protein in EMSA buffer (80 mM KCl, 25 mM Tris pH 8.0, 5 mM MgCl₂, 1 mM dithiothreitol, 10% glycerol, 200 ng/ml bovine serum albumin) with 1 μg of poly[dG-dC] or poly[dI-dC] as nonspecific competitor, as indicated. Samples were incubated for 30 min at 37 °C and electrophoresed at 200 V for 1.5 h in 4% nondenaturing PAGE gels in TBE buffer (45 mM Tris-boric acid, 1 mM EDTA) at 10 °C, after which the gel was vacuum dried and analyzed by autoradiography.

RNA isolation and qRT-PCR analysis

Cells expressing P_van-cdnL_Cc or P_van-cdnL_Cc-flag were grown in 10 ml M2G medium with appropriate antibiotics and 0.5 mM vanillate (M2G-vanillate) overnight at 30 °C. 1 ml of this was innoculated into fresh 50 ml M2G-vanillate and grown to an OD₆₆₀ = 0.3–0.4. Two 4 ml aliquots were withdrawn (time “0”) and pelleted. One was stored at −80 °C until further processing. The other was washed twice with M2-vanillate (lacking glucose) to eliminate the glucose, resuspended in 4 ml M2-vanillate and grown at 30 °C for 15 min, then pelleted and stored at −80 °C until use (time “0”, no glucose). The rest of the 50 ml culture was washed twice with M2G to eliminate vanillate, resuspended in 42 ml of fresh M2G with appropriate antibiotics and incubated at 30 °C. 4 ml aliquots were withdrawn at specific times (2, 4, 8 hours) and treated as with time “0” samples with and without glucose starvation. RNA was extracted from each sample using PureLink RNA Mini Kit (Thermo Fisher Scientific), treated with Turbo DNase (Ambion) for 4 h at 37 °C, and purified using the PureLink RNA Mini Kit. RNA levels were assessed by gel electrophoresis and with NanoDrop ND-1000 (Thermo Fisher Scientific) using a 260 nm extinction coefficient of 40 ng-cm/μl. 2 μg of total RNA was reverse transcribed into cDNA using random hexamer primers (Promega) and Transcriptor Reverse Transcriptase (Roche) in 20 μl of reaction mix as per manufacturer’s instructions. 1 μl of cDNA was added to 10 μl SYBR Green PCR Master Mix (Applied Biosystems or BioRad), with the required primers (100 nM). Each reaction was performed in triplicate, with a control reaction using equivalent starting volume of RNA to verify absence of contaminating DNA. Primers to amplify an ∼50–150 bp region within each transcript were designed using Primer Express 3.0 software, and qRT-PCR was carried out in a StepOne instrument and software using the 1-Step RT-PCR program cycle without reverse transcription (Applied Biosystems). Melting and dissociation curves were determined from 60–95 °C, 30s and 95 °C, 15s. Primers to quantify 16S rRNA transcription from the P_rrnA::lacZ reporter plasmid pMR3769 and of ruvA (as endogenous control) are those described previously⁴³, and their ratio yields the ‘Relative transcript level’ for each sample. For each primer pair a standard RT-PCR curve was generated for five serial ten-fold dilutions of cDNA, and ones with near 100% efficiency were used.

Coimmunoprecipitation (CoIP)

Strains ME5 (ΔcdnL_Cc, P_van::cdnL_Cc, vanR) and ME17 (ΔcdnL_Cc, P_van::cdnL_Cc-flag, vanR) were grown in 500 ml PYE with appropriate antibiotics and 0.5 mM vanillate to OD₆₆₀ of ~0.3, harvested by centrifugation, washed thrice with CoIP buffer (20 mM HEPES pH 7.5, 50 mM NaCl, 20% glycerol), and the pellets frozen at −80 °C until further use. Frozen pellets were resuspended in 5 ml CoIP buffer, incubated with 10 mM MgCl₂, 50 mg lysozyme and 50 units DNase I (Promega) at 4 °C with shaking for 30 min, lysed with a French press at 16,000 psi, and clarified by centrifugation (18,000 g, 4 °C, 5 min). Cleared lysates were incubated with 20 μl of pre-equilibrated anti-FLAG agarose affinity gel (FLAGIPT-1, Sigma-Aldrich) overnight at 4 °C with rotation, washed thrice with CoIP buffer (100 mM NaCl), thrice with wash buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl) in SigmaPrep spin columns, incubated in 150 μl wash buffer containing 100 μg/ml 3xFLAG peptide for 1 h at 4 °C, and eluted. Samples were analyzed by Western blotting using anti-FLAG M2 (F3165, Sigma-Aldrich) or anti-RNAP β 8RB13 (Thermo Fisher Scientific) monoclonal antibodies. Co-IP results were checked for reproducibility in three independent experiments.

Quantitative chromatin immunoprecipitation (qChIP)

Cells grown to exponential phase (OD₆₆₀ = 0.4–0.6) in 50 ml M2G-vanillate were cross-linked with 1% final concentration (v/v) formaldehyde for 30 min at room temperature with shaking (100 rpm), quenched with 2.5 ml of 2.1 M glycine, pelleted, washed three times with phosphate-buffered saline, and stored at −80 °C until further use. Cell pellets were resuspended in 200 μl ChIP lysis buffer A (20% sucrose, 50 mM NaCl, 10 mM EDTA, 10 mM Tris pH 8, 1 mg/ml lysozyme), incubated for 30 min at 37 °C and cooled in ice. After mixing with 800 μl ChIP lysis buffer B (150 mM NaCl, 1 mM EDTA, 50 mM HEPES-KOH pH 7.5, 1% Triton X-100, 0.1% deoxycholate, 0.1% SDS, Roche Complete protease inhibitor cocktail), they were sonicated with 12 30s on-30s off cycles in a Bioruptor (Diagenode) to obtain ∼0.5 kb long fragments and clarified by centrifugation. 20 μl of the supernatant was kept aside for the input sample. The rest was added to 30 μl of anti-FLAG M2 (F3165, Sigma-Aldrich), anti-σ^A 2G10 (Thermo Fisher Scientific), or anti-RNAP β 8RB13 (Thermo Fisher Scientific) monoclonal antibodies previously immobilized (≥4 h incubation at 4 °C and two washes with PBS containing 5 mg/ml BSA) on protein A magnetic Dynabeads (Life Technologies), and incubated overnight at 4 °C with rotation. The beads were washed twice with ChIP lysis buffer B with 0.15 M NaCl, twice with the same buffer but with 0.5 M NaCl, and twice with wash buffer (250 mM LiCl, 10 mM Tris-HCl pH 8.0, 1 mM EDTA, 0.5% NP-40, 0.5% sodium deoxycholate). After a final wash with Tris-EDTA (TE) buffer, the beads were resuspended in 60 μl TE, 1% SDS, and incubated for 10 min at 65 °C. From this, 40 μl was mixed with 40 μl TE/1% SDS and 2.4 μl proteinase K (20 μg/μl), incubated at 42 °C for 2 h, then at 65 °C for 6 h, and the DNA isolated using the Roche High Pure PCR product Purification kit. The input sample was also subjected to the same cross-link reversal/DNA extraction protocols. qPCR was carried out with SYBR Green reaction mix (BioRad) in 0.1 ml MicroAMP FAST optical 48-well reaction plates and a StepOne qPCR apparatus (Applied Biosystems). Primers used were: 5′-TCCACGGGCGTCTGTTAAG-3′ and 5′-CCCCTCGCGACAATATAACG-3′ for P_rrnA; 5′-TGCTCGTGGACGTCAACAAC-3′ and 5′-GGGCGCATAGCCGAGAT-3′ for an intragenic nonpromoter control region (nucleotides 3542858 to 3542913 of gene CCNA_003364, whose σ^F-dependent expression is activated by heavy metal stress⁴²). Standard curves were obtained for each DNA region of interest with serially diluted input DNA sample and its primer pair. Signal enrichment at each promoter is the ratio of promoter-specific to intragenic signal of the ChIP fractions relative to that for the input sample, and is reported as the mean and standard error from three independent experiments.

Genome analysis

A database comprising 239 representative complete proteomes of alphaproteobacteria was searched for CdnL homologs using BLASTP and CdnL_Cc as a query. Results with an e-value < 0.001 were used for alignments. Protein alignments were performed using MUSCLE v3.8.31⁵⁸, and curated with Gblocks 0.91b⁵⁹ using default parameters. Phylogenetic trees were generated from the Gblocks output with FastTree version 2.1.7⁶⁰ using the Whelan and Goldman amino acid replacement matrix and a Gamma20-based likelihood calculation. The phylogenetic trees were visualized using FigTree v.1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/). 5′ intergenic regions were retrieved from the cognate genomic databases and analyzed independently.