The CsrA-FliW network controls polar localization of the dual-function flagellin mRNA in Campylobacter jejuni

Abstract

The widespread CsrA/RsmA protein regulators repress translation by binding GGA motifs in bacterial mRNAs. CsrA activity is primarily controlled through sequestration by multiple small regulatory RNAs. Here we investigate CsrA activity control in the absence of antagonizing small RNAs by examining the CsrA regulon in the human pathogen Campylobacter jejuni. We use genome-wide co-immunoprecipitation combined with RNA sequencing to show that CsrA primarily binds flagellar mRNAs and identify the major flagellin mRNA (flaA) as the main CsrA target. The flaA mRNA is translationally repressed by CsrA, but it can also titrate CsrA activity. Together with the main C. jejuni CsrA antagonist, the FliW protein, flaA mRNA controls CsrA-mediated post-transcriptional regulation of other flagellar genes. RNA-FISH reveals that flaA mRNA is expressed and localized at the poles of elongating cells. Polar flaA mRNA localization is translation dependent and is post-transcriptionally regulated by the CsrA-FliW network. Overall, our results suggest a role for CsrA-FliW in spatiotemporal control of flagella assembly and localization of a dual-function mRNA.

ARTICLE

Received 12 Dec 2015 | Accepted 18 Apr 2016 | Published 27 May 2016

DOI: 10.1038/ncomms11667 OPEN

The CsrA-FliW network controls polar localization of the dual-function agellin mRNA in Campylobacter jejuni

Gaurav Dugar1, Sarah L. Svensson1, Thorsten Bischler1, Sina Waldchen2, Richard Reinhardt3, Markus Sauer2

& Cynthia M. Sharma1

The widespread CsrA/RsmA protein regulators repress translation by binding GGA motifs in bacterial mRNAs. CsrA activity is primarily controlled through sequestration by multiple small regulatory RNAs. Here we investigate CsrA activity control in the absence of antagonizing small RNAs by examining the CsrA regulon in the human pathogen Campylobacter jejuni. We use genome-wide co-immunoprecipitation combined with RNA sequencing to show that CsrA primarily binds agellar mRNAs and identify the major agellin mRNA (aA) as the main CsrA target. The aA mRNA is translationally repressed by CsrA, but it can also titrate CsrA activity. Together with the main C. jejuni CsrA antagonist, the FliW protein, aA mRNA controls CsrA-mediated post-transcriptional regulation of other agellar genes. RNA-FISH reveals that aA mRNA is expressed and localized at the poles of elongating cells. Polar aA mRNA localization is translation dependent and is post-transcriptionally regulated by the CsrA-FliW network. Overall, our results suggest a role for CsrA-FliW in spatiotemporal control of agella assembly and localization of a dual-function mRNA.

1 Research Centre for Infectious Diseases (ZINF), University of Wrzburg, Josef-Schneider-Str. 2/D15, Wrzburg D-97080, Germany. 2 Department of Biotechnology and Biophysics, University of Wrzburg, Am Hubland, Wrzburg D-97074, Germany. 3 Max Planck Genome Centre Cologne, Max Planck Institute for Plant Breeding Research, Carl-von-Linn-Weg 10, Cologne D-50829, Germany. Correspondence and requests for materials should be addressed to C.M.S. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

Post-transcriptional control involves a complex interplay between mRNAs, small regulatory RNAs (sRNAs) and protein regulators. Although regulatory functions have

typically been attributed to proteins or sRNAs, mRNAs have canonically been considered as targets of this regulation. However, regulatory functions have recently also been described for mRNAs that either encode sRNAs in their untranslated regions (UTRs) or act as sponges that sequester other regulatory factors14.

The widespread bacterial Csr/Rsm (Carbon storage regulator/ Regulator of secondary metabolism) regulatory network5 is an ideal model system to study the complex post-transcriptional cross-talk between mRNAs, sRNAs and protein regulators. About 75% of all sequenced bacterial genomes encode a homologue of the central RNA-binding protein (RBP) of this system, CsrA (RsmA/E). CsrA is a pleiotropic regulator of global physiological phenomena in Gammaproteobacteria5 and considered the most conserved post-transcriptional virulence regulator6. CsrA mainly acts by repression of translation initiation via binding to 50 regions of mRNAs7. The homodimeric CsrA binds GGA-rich motifs that are often located in hairpin loops and/or overlap the Shine-Dalgarno (SD) sequence5. In Gammaproteobacteria, CsrA activity is regulated through the CsrB/C and RsmX/Y/Z families of sRNAs5,7. These antagonizing sRNAs are often induced by environmental signals6 and harbour multiple stem-loops with high-afnity GGA motifs that sequester CsrA/RsmA8. Despite the presence of CsrA, many bacteria lack homologues of these antagonizing sRNAs. Also, the global CsrA regulon and its general biological function outside the Gammaproteobacteria are unclear. In the Gram-positive Bacillus subtilis, the agellar assembly protein FliW antagonizes CsrA via direct binding9. Although FliW homologues are relatively widespread9, protein-mediated regulation of CsrA has not yet been shown outside B. subtilis. Whether FliW can cooperate with RNA-mediated regulation of CsrA is also unknown.

In the Gram-negative Epsilonproteobacterium Campylobacter jejuni, currently the leading cause of bacterial gastroenteritis in humans, CsrA affects motility, biolm formation, oxidative stress response and infection10. Despite several phenotypic analyses of csrA deletion strains1012, direct CsrA targets in Epsilonproteobacteria are largely unknown. Global transcriptome studies indicated that both C. jejuni and the related pathogen Helicobacter pylori1316, which both carry potential FliW homologues, lack the CsrA-antagonizing sRNAs.

Here we use co-immunoprecipitation (coIP) combined with RNA sequencing17,18 (RIP-seq) to globally determine the direct RNA-binding partners of C. jejuni CsrA and investigate whether RNA-based regulation of CsrA occurs in the absence of canonical antagonizing sRNAs. Our genome-wide approach reveals many mRNAs of agellar genes as potential CsrA targets and we demonstrate that aA mRNA, encoding the major agellin, has dual (coding and regulatory) function. As the most abundantly co-puried transcript, aA mRNA is the main target of CsrA translational repression. In addition, the aA leader can act as an mRNA-derived RNA antagonist of CsrA. Together with the main CsrA antagonist, the FliW protein, aA mRNA titrates CsrA to regulate expression of other agellar genes.

In addition, using confocal and super-resolution microscopy imaging, we show that aA mRNA is expressed in elongating cells and localizes to the cell poles of the amphitrichous C. jejuni. In contrast to eukaryotes19, RNA localization is so far only poorly understood in prokaryotes. Bacterial mRNAs can remain localized close to their genomic site of transcription20 or can migrate to places in the cell where their encoded products are required in a translation-independent manner involving cis-acting signals in the RNA itself21. Besides the mechanisms of

bacterial RNA localization, even less is known about how this process may be regulated and which, if any, RBPs are involved. Here we show, based on a variety of C. jejuni mutants that disrupt or maintain aA translation, that polar aA mRNA localization requires its translation. Furthermore, we demonstrate that FliW facilitates polar agellin mRNA localization by antagonizing CsrA-mediated translational repression of aA. The unexpected role of the CsrA-FliW system in spatial control of agellin mRNA expression provides new insight into the role of RBPs in bacterial mRNA localization, a process only recently described in prokaryotes.

ResultsGlobal RIP-seq reveals direct CsrA targets in C. jejuni. To globally identify C. jejuni CsrA targets and any RNA regulators of CsrA activity, we applied a RIP-seq approach17,18. The csrA (Cj1103) gene was chromosomally 3xFLAG-tagged at its C-terminus in strains NCTC11168 and 81-176. CsrA-3xFLAG is constitutively expressed during growth in rich medium, and neither introduction of the FLAG-tag nor deletion of csrA affectsC. jejuni growth under the examined conditions (Supplementary Fig. 1). We performed coIPs on mid-exponential-phase lysates of csrA-3xFLAG strains and, as control, their respective untagged wild-type (WT) strains (Fig. 1a and Supplementary Fig. 2a). After conversion of co-puried RNAs into cDNA and deep sequencing,93.295.8% of the 4.66.2 million sequenced reads for the individual libraries were mapped to the respective genomes (Supplementary Table 1). Most of the NCTC11168 control-coIP library reads mapped to presumably non-specically pulled-down abundant classes of RNA (rRNA, tRNA and housekeeping RNAs; Fig. 1b and Supplementary Table 2). In contrast, a B36-fold and B5-fold enrichment for reads mapped to 50UTRs or open reading frames (ORFs) of mRNAs, respectively, was observed in the CsrA-3xFLAG coIP library (Fig. 1b). No specic sRNA enrichment was detected. As the coIP of strain 81-176 showed similar enrichment patterns (Supplementary Fig. 2b), we focused on strain NCTC11168.

C. jejuni CsrA primarily binds agellar mRNAs. Functional enrichment analysis of the 154 top CsrA targets with 45-fold enrichment in the CsrA-3xFLAG- versus control-coIP (Supplementary Data 1) revealed an overrepresentation of mRNAs from the class Surface Structures, including agellar genes (Supplementary Fig. 3a,b). In fact, 90% of the reads mapping to the 45-fold-enriched CsrA targets belonged to agella- or motility-related genes (Fig. 1c). The alternative sigma factors RpoN (s54) and FliA (s28) hierarchically control agellar expression in Campylobacter22. Early genes are expressed from RpoD/s70-dependent promoters, whereas class 2 (middle) and class 3 (late) genes are RpoN- and FliA-dependent, respectively22. Most of the enriched transcripts belonged to either class 2 or class 3 (Table 1 and Supplementary Fig. 3c). The most abundantly co-puried transcript, with more than 300-fold enrichment, was aA mRNA, encoding the major agellin (Fig. 1c).

cDNA peaks reveal CsrA binds in diverse mRNA regions. Visual inspection of the cDNA read-patterns showed that numerous agellar mRNAs, including aA, aG and gI (encoding the major agellin, a gene involved in agellum formation, and a P-ring component, respectively) showed strong enrichment in their 50UTRs (Fig. 1d and Supplementary Fig. 4a).

CsrA binding was also observed between two genes in polycistronic mRNAs, such as the Cj0310c-Cj0309c and Cj0805-dapA operons. Analysis of the potential CsrA-binding sites in an Escherichia coli green uorescent protein (GFP) reporter-system,

2 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

a

d

WT (control) CsrA-3xFLAG

flaA (Cj1339c)

Cj0310c

coIP

CsrA-3xFLAG

Control

pgaA

pgaA

Supernatant

Wash

coIP

Culture

Lysate

0

csrA

csrA

Culture

Lysate

Supernatant

Wash

coIP

pBAD-csrACj

pBAD-csrACj

CsrA-3xFLAG

Control

++

++

++

+

Relative cDNA scores

Arabinose

15 kDa

CsrA-3xFLAG

25 kDa

15 kDa

25 kDa

FlaA-GFP

GroEL

1,000,000

0

1,000,000

0 0

100 101 67 5 1

%

%

GroEL

55 kDa

10 kDa CsrACj-Strep

0

75 100 %

0.1 OD

10 OD

0.1 OD

10 OD

55 kDa

55 kDa

b

Control coIP CsrA-3xFLAG coIP

sRNA (1x)

5'UTR (36x)

ORF (5x) rRNA, tRNA,

hkRNA (0.7x)

Cj0309c

coIP

0

1,000 0

1,000

Relative cDNA scores

++

++

++

+

Arabinose

FLAG-Cj0310c

100 95 98 73 48

57 62 %

Covarying mutations

Nucleotide

present

Nucleotide

identity

N

N

N

90%

80%

65%

13 19

100 76 76 28 7

Cj0309c-GFP

GroEL

c

flaG

flgH

flgI

flgE

flgG2 flaB

motA

nhaA1

pseB

ftsY

dccS

pglC

hydA

Motility-related genes

Other

functions

e

2

324/328 Targets

0 1 2 3 4 5 6

A

276/328 Targets

13%

10%

flaA

A

G G

A

YYY Y

R R R

77%

Bits

1

90%

80%

65%

50%

R

Figure 1 | RIP-seq analysis of C. jejuni CsrA. (a) Western blot analysis of coIP samples of C. jejuni NCTC11168 WT and csrA-3xFLAG strains using anti-FLAG antibody conrms a successful CsrA-3xFLAG pulldown in the tagged strain. The amount of samples loaded (OD600 of bacteria) is indicated. GroEL served as loading control. (b) Pie charts showing relative proportions of mapped cDNA reads of different RNA classes in the coIP libraries (hkRNA: housekeeping RNAs). Numbers in brackets indicate the relative enrichment of the respective RNA class in the CsrA-3xFLAG versus control coIPs. (c) Pie chart showing the percentages and enriched genes of mapped reads for all 45-fold enriched CsrA target genes. (d) (Left) Mapped RNA-seq reads for the control (black) and CsrA-3xFLAG coIP (blue) in strain NCTC11168. Grey arrows: ORFs; black arrows: transcriptional start sites (TSS). Examples of enrichment patterns in 50 UTRs (aA) and between genes in a polycistron (Cj0310c-Cj0309c operon; encoding two paralogous efux proteins). (Right)

Western blot analysis using anti-FLAG and anti-GFP antibodies of reporter fusions to potential C. jejuni CsrA target genes in E. coli DpgaA, DpgaA/DcsrA and DpgaA/DcsrA pBAD-csrACj (complementation with C. jejuni CsrA-Strep under control of an arabinose-inducible pBAD promoter) strains. Putative CsrA

targets from C. jejuni were fused in-frame (for example, 33 aa for aA) to GFP or a FLAG-lacZ tag (Supplementary Fig. 4b). As deletion of csrA dramatically enhanced biolm formation and led to poor growth in liquid culture in our E. coli strain, reporter experiments were performed in a DpgaA background. GroEL served as loading control. Protein samples corresponding to 0.1 OD600 were loaded. Quantications of reporter expression are given below the blots. (e) (Left) CsrA-binding motif predicted by MEME24 (E-value 2.1E-11). (Right) Consensus secondary structure motif of C. jejuni CsrA-binding sites

predicted by CMnder62.

originally developed to study sRNA-mediated regulation23, revealed all of the tested 50UTR targets (aA, aG, gI, aB, pseB and Cj1249) were highly upregulated (410-fold) in the absence of E. coli csrA as measured by western blot and FACS analyses (Fig. 1d and Supplementary Figs 4 and 5). Reduced reporter fusion expression was restored by complementation of DcsrA with C. jejuni CsrA. Using an operon reporter, where the

C-terminal part of the upstream gene is fused to FLAG-lacZ and the N-terminal part of the downstream gene to GFP, we observed that both E. coli and C. jejuni CsrA can repress the downstream genes in polycistrons (Cj0310c-Cj0309c and Cj0805-dapA). Expression of the upstream genes was only slightly affected and they do not contain any strong internal transcriptional start sites that could lead to uncoupled transcription of the downstream

genes14. As we observed that potential SD sequences right at the 30 end of the upstream genes are covered by CsrA target sites, CsrA probably interferes with ribosome binding and translation of the downstream genes and thereby might mediate discoordinate operon regulation.

Automated peak-detection reveals a CsrA-binding motif. To automatically identify CsrA-binding regions and a binding motif from coIP cDNA enrichment patterns, we developed a peak-detection algorithm based on a sliding window approach (see the Methods for details). This approach predicted 328 potential CsrA-binding sites with 45-fold enrichment in the

NCTC11168 coIP (Supplementary Data 2). As a control, peak

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

Table 1 | Enrichment of genes involved in agellar biosynthesis in the CsrA coIP data.

C. jejuni NCTC11168 C. jejuni 81-176

Enrichment (reads) 50UTR ORF 50UTR ORF Regulation of expression (class 1)

rpoN (Cj0670) 1.5 (26) 25 (1,747) 8 (634)

iA (Cj0061c) 1.1 (121) 0.7 (79)

gS (Cj0793) 1.7 (43) 1.3 (1) 1.7 (15)

gR (Cj1024) 1.2 (4) 1.1 (137) 1.3 (4) 1.2 (101) Flagellar protein secretion (class 1)gM (Cj1464) 5 (1429) 2.3 (380)

iF (Cj0318) 6.1 (1,005) 7.8 (1,105)

hA (Cj0882c) 1.2 (82) 0.9 (53)

hB (Cj0335) 0.6 (3) 1.2 (99) 0.7 (1) 1.0 (67)

iO (Cj0352) 1.4 (41) 0.7 (1)

iP (Cj0820c) 1.2 (29) 0.5 (15)

iQ (Cj1675) 1.3 (45) 0.9 (29)

iR (Cj1179c) 1.2 (6) 0.5 (5)

iH (Cj0320) 4.8 (202) 1.3 (100)

iI (Cj0195) 3.6 (442) 0.6 (86) Basal body components (classes 1 and 2)iE (Cj0526c) 2.4 (268) 1.4 (120)

gC (Cj0527c) 1.2 (397) 0.9 (217)

gB (Cj0528c) 1.6 (7) 1.6 (181) 0.7 (2) 0.9 (61)

gG2 (Cj0697) 43.9 (8,133) 77.4 (9,670)

gG (Cj0698) 1.2 (1) 1.4 (253) 5.3 (4) 1.3 (165)

gJ (Cj1463) 4.4 (180) 1.0 (26)

gI (Cj1462) 170.5 (5,750) 52.7 (12,087) 157.1 (1,666) 61.5 (5,401)

gA (Cj0769c) 0.8 (2) 15.8 (410) 0.9 (4) 3.7 (104)

gH (Cj0687c) 200.9 (1,911) 20.1 (3,288) 110.1 (917) 27.6 (2,487) Flagellar hook components (class 2)gE (Cj1729c) 68.2 (104,324) 7.1 (3,967)

gD (Cj0042) 3.6 (1015) 2.0 (284)

gE2 (Cj0043) 2.5 (1045) 1.1 (250)

iK (Cj0041) 4.1 (613) 1.2 (89)

Cj0040* 356.2 (3,389) 110.6 (9,277) 38 (230) 20.4 (727)

gK (Cj1466) 0.7 (4) 1.0 (141)

gL (Cj0887c) 2.0 (484) 2.3 (74) 0.9 (193) Flagellar lament components (classes 2 and 3)aA (Cj1339c) 304.5 (693,471) 111 (473,588) 324.7 (158,590) 45.3 (138,159)

aB (Cj1338c) 58.8 (915) 14.1 (17,880) 59.4 (1,170) 14.9 (29,530)

iD (Cj0548) 6.8 (4,348) 5.4 (3,929)

iS (Cj0549) 1.6 (149) 1.3 (165)

aC (Cj0720) 1.2x (344) 1.2 (1,298) 1.3 (239) 1.2 (1,237) Other enriched genes (45x) involved in agella formationpseB (Cj1293) 119.7 (2,298) 9.5 (2,280) 34.5 (470) 4.1 (759)

pseI (Cj1317) 1.7 (22) 7.2 (864) 2.3 (14) 0.8 (111)

aG (Cj0547) 346.1 (11,077) 72.4 (18,150) 168.5 (3,701) 84.2 (16,012)

motA (Cj0337c) 10.3 (89) 1.8 (660) 1.4 (16) 0.8 (271)

Cj0951c 15.2 (79) 2 (3) 1.3 (194)

Cj0248 5.5 (120) 1.8 (387) 1.1 (38) 0.9 (257)

hX (Cj0848c) 7.5 (13) 1.5 (7)

CoIP, co-immunoprecipitation; UTR, untranslated region; ORF, open reading frame.

Classication of agellar genes is based on ref. 75. Transcripts with 45-fold enrichment in cDNA read counts in the CsrA-3xFLAG versus control coIP libraries are highlighted in bold. Numbers in

brackets indicate the absolute cDNA read counts in the CsrA-3xFLAG coIP libraries.

*Cj0040 (unknown function) is the rst gene of the hook gene operon.

detection was performed in reverse manner by scanning for enriched regions in the control- versus CsrA-3xFLAG-coIP. This analysis revealed only ve peaks, without a common motif, indicating a high specicity of the peaks detected in the CsrA-3xFLAG-coIP. MEME24 analysis of the 328 enriched sequences revealed a (C/A)A(A/T)GGA motif in 324/328 input sequences (Fig. 1e). Analysis of the 81-176 coIP led to a similar motif (Supplementary Fig. 2c). To check if a similar motif can be found in non-enriched regions, we conducted the peak-detection

in reverse manner using a cutoff of only 41-fold enrichment in the control- versus CsrA-3xFLAG-coIP. This revealed 448 enriched sites in the control library. Subsequent motif prediction did not yield any signicant motifs, further supporting high specicity of the coIP approach. Consensus-structure motif screening of the enriched CsrA-coIP sequences revealed an AAGGA motif in a hairpin-structure loop in 276/328 input sequences (Fig. 1e). These C. jejuni sequence/structural motifs agree with binding sites of other CsrA homologues25.

4 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

a

b

% FlaA-3xFLAG expression

U

(M2)

flaA 5'UTR

A A

(M1)

**

% flaA-3xFLAG mRNA

300

0

NS

300

0

NS

NS

G

G

U

1 2

A

G

U

G (M3)

200

200

-A-A-G-U-U-C-A

U-G-A-G-C-U-U-

G

-A-A-U-U-U-

30

SD

SD

G

A

-U-U-U-A-A-A

U-U-U-A-A-A-

RNase T1 cleavage

100

100

SL1

SL2

40

M1 csrA

3

M2

5'

U-A-A-

C

U

A

-G-G-A-U-U

-U-G

...3'

WT

csrA

M1

M2 csrA

M2/M3

M2/M3 csrA

CsrA/flaA

(Molar excess)

45

20

15

+1

Start codon

+8

c

e

1,000

2,000

0

10

20

50

100

200

flaA WT control

flaA M2/M3 control

flaA WT + CsrA (1/100)

T1 Ladder (flaA WT)

T1 Ladder (flaA M2/M3)

CsrA (nM)

flaA M2/M3 + CsrA (1/100)

500

OH Ladder

0 20 50 100

flaA WT flaA M2/M3

0 20 50100

flaA WT leader* ( 4 nM)

d

100

80

5G

GG AGGAGGA

4G

3G

% Bound

60

80

60

40

20

0

20 40 60 80 100

M3

8G

40

9G

SL2

SL1

20

0

500 1,000 1,500 2,000

23G

CsrA (nM)

flaA WT

flaA M1 (GGA to AAA in SL1)

flaA M2 (GGA to UGA in SL1)

flaA M3 (GGA to GGG in SL2)

flaA M2/M3 (GGA to UGA in SL1; GGA to GGG in SL2)

27G

29G

32G

33G

M2

Figure 2 | CsrA represses aA translation by binding to its 50UTR. (a) Predicted secondary structure of the aA leader using Mfold74. Blue bars indicate

GGA motifs; grey: SD sequence. Black triangles indicate RNase T1 cleavages from the structure probing in c. (b) Western blot quantication (n 5

biological replicates) of FlaA with a C-terminal 3xFLAG epitope tag integrated at its native locus (FlaA-3xFLAG) and northern blot analysis of aA mRNA (n 3 biological replicates) in DcsrA and various aA 50UTR mutant strains. Shown is the means.e.m (**Po0.01 using Students t-test, NS: not

signicant). Mutations are depicted in red in a. (c) Gel-shift assays using B0.04 pmol in vitro-transcribed and 50 end-labelled aA leader ( 45 to 99

relative to the start codon) with increasing concentrations of CsrA. (d) Afnity binding curves determined by gel-shift assays for 32P-labelled aA WT and mutant leaders (r4 nM) based on three replicates. The inset represents an enlargement of the binding curves for low CsrA concentrations. Shown is the means.d. (e) Footprinting assays of B0.2 pmol 32P-labelled aA WT and aA M2/M3 mutant leaders in the absence or presence of increasing CsrA concentrations (molar excess of 0, 20, 50 and 100 CsrA) using RNase T1. Untreated aA leader alone or incubated with 100-fold excess of CsrA served as controls and RNase T1- or alkali (OH)-digested aA leader as ladders, respectively. Blue lines: GGA motifs; green lines: protection from RNA cleavage upon addition of CsrA. The secondary structure of the aA leader according to a is depicted on the right.

aA mRNA is translationally repressed by CsrA. The agellar lament, consisting mainly of the FlaA agellin, is among the last components produced during agellum assembly. In our coIP, 77% of the reads from 45-fold enriched genes mapped to aA,

indicating it as the main CsrA target (Fig. 1c). Secondary-structure predictions revealed that the 45-nt-long aA 50UTR can fold into two stem-loops (SL1 and SL2), both of which harbour an ANGGA motif in their loops (Fig. 2a). The second ANGGA motif

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

covers the ribosome-binding site and a third GGA is present as the second codon. The aA 50UTR secondary structure is conserved and supported by compensatory base-pair changes in other Campylobacter species (Supplementary Figs 6 and 7, and Supplementary Methods). A chromosomally 3xFLAG-tagged FlaA was B3-fold upregulated in a DcsrA strain compared with

WT on western blots (Fig. 2b and Supplementary Fig. 8a, lanes 1 and 2). To show that CsrA affected translation by binding to the aA leader, we introduced chromosomal point-mutations into the two putative GGA CsrA-binding motifs (M1: SL1GGA-AAA,

M2: SL1GGA-UGA, and M3: SL2GGA-GGG; Fig. 2a,b and

Supplementary Fig. 8a, lanes 38). Like deletion of csrA, mutation of the GGA motifs resulted in two- to threefold elevated FlaA-3xFLAG protein expression. FlaA-3xFLAG levels were not affected by deletion of csrA in the aA leader mutants, indicating CsrA binding was abolished in these strains. Northern blot analysis showed aA-3xFLAG mRNA levels are only mildly affected in the different mutant strains, further indicating post-transcriptional regulation of aA by CsrA (Fig. 2b and Supplementary Fig. 8a).

In vitro gel-shift assays using recombinant C. jejuni CsrA-Strep and T7-transcribed, 50-end radiolabelled aA WT leader showed strong CsrA binding (Kd B50 nM) with two dened shifts,

indicating at least two CsrA-binding sites (Fig. 2c). In contrast, aA leaders with GGA point-mutations in either SL1 (M1 and M2), SL2 (M3) or both SL1 and SL2 (M2/M3) showed four- to tenfold higher Kd values (200500 nM), conrming that the mutations reduced CsrA binding (Fig. 2d and Supplementary Fig. 9a). To map CsrA-binding sites on the aA leader, we performed in-vitro footprinting assays with labelled aA leader in the absence or presence of CsrA using enzymatic and chemical cleavage (RNase T1; single stranded G-residues and lead(II) acetate; single-stranded RNA). Cleavage patterns without CsrA conrmed the predicted aA leader structure (Fig. 2e and Supplementary Fig. 8b). A clear protection was observed at the SL1 and SL2 GGA motifs of the WT leader upon addition of increasing CsrA amounts, but not for a aA M2/M3 mutant with disrupted binding motifs. The third GGA downstream of the start codon was not protected. Overall, our data suggest C. jejuni CsrA represses aA translation by high-afnity binding to the two GGA-containing stem-loops SL1 and SL2 in the aA leader.

The agellar assembly factor FliW binds CsrA in C. jejuni. The constitutive expression of CsrA during routine culture (Supplementary Fig. 1) suggested modulation of its activity rather than its expression. Because homologues of the CsrB/C sRNAs are absent in C. jejuni, we hypothesized that other RNAs, or even proteins, might control CsrA activity in Campylobacter. One candidate (Cj1075, 129 aa) is a potential homologue of the agellar assembly factor, FliW, which has a role in motility26,27 but is otherwise uncharacterized. In B. subtilis, FliW binds CsrA and antagonizes CsrA-mediated translational repression of hag mRNA, encoding the major agellin9. FliW can also bind Hag, which accumulates in the cytoplasm before agellar hook completion. Hag thus sequesters FliW from CsrA, allowing CsrA to repress Hag synthesis. Upon completion of the hook, Hag is secreted, FliW is released and CsrA repression of aA translation is relieved. Thus, this Hag-FliW-CsrA partner-switch mechanism ensures appropriate temporal agellin synthesis. In Epsilonproteobacteria, iW homologues are present, but, unlike Bacillus, are not encoded adjacent to csrA (Fig. 3a). To investigate whether FliW can interact with CsrA and FlaA in C. jejuni, we performed proteinprotein coIP experiments using chromosomal C-terminal 3xFLAG-tag fusions as bait. The anticipated interaction partners were tagged with mCherry at their

C-terminus to allow detection by western blotting. In a FliW-3xFLAG-coIP, CsrA-mCherry was successfully co-puried, indicating the two proteins can interact (Supplementary Fig. 10). Similarly, FliW-mCherry was co-puried in a FlaA-3xFLAG-coIP, indicating conserved interactions between all three proteins. As control, none of the proteins was co-puried in coIPs with strains that carry the mCherry-fusion proteins but not the FLAG-tagged proteins.

FliW antagonizes CsrA-mediated translational repression. To determine whether the FliWCsrA interaction could antagonize CsrA function in Epsilonproteobacteria, we used FlaA protein levels as a read-out for CsrA activity (Fig. 3b). Whereas FlaA-3xFLAG was B3-fold upregulated in DcsrA, deletion of iW led to B6-fold downregulation, consistent with further repression of aA translation by additional CsrA released upon deletion of its protein antagonist (Fig. 3b). A DcsrA/DiW double deletion conrmed that the observed downregulation was indeed mediated through CsrA, as FlaA-3xFLAG levels increased back to those in the DcsrA mutant. Despite strong reduction of FlaA-3xFLAG protein levels, a B2-fold higher aA mRNA level was observed upon deletion of iW, indicating additional effects of FliW on aA expression (Supplementary Fig. 11a). Thus, we constructed a transcriptional reporter composed of the unrelated Cj1321 50UTR and its early coding region (Cj1321_mini) under the control of the aA promoter. This reporter was, like the endogenous aA mRNA, B2-fold upregulated in the DiW mutant (Supplementary Fig. 11b). As Cj1321 is independent of CsrA-mediated control, FliW seems to have a negative effect (direct or indirect) on aA transcription.

To uncouple transcriptional control of aA from its translational regulation, we replaced the s28-dependent aA promoter in the FlaA-3xFLAG strain with a constitutive s70-dependent metK promoter. Upon deletion of csrA in this strain, a B3-fold increase in FlaA-3xFLAG level was observed, further conrming post-transcriptional regulation of FlaA-3xFLAG protein expression by CsrA (Fig. 3b). Like for the strain expressing FlaA-3xFLAG from its native promoter, FlaA-3xFLAG expressed from the metK promoter was strongly downregulated upon deletion of iW and was restored to DcsrA levels in the DcsrA/DiW double mutant. This further indicates FliW antagonizes CsrA-mediated translational repression of aA in a promoter-independent manner. In addition, decreased aA mRNA stability was observed upon iW deletion in rifampicin stability assays. This is consistent with increased translational repression of aA in the absence of iW, despite overall higher steady-state aA mRNA levels because of FliW-dependent increased transcription (Supplementary Fig. 11c).

In line with strong downregulation of the FlaA protein upon iW deletion, transmission electron microscopy revealed shorter agella on DiW bacteria compared with those of the WT strain (Fig. 3c,d). In fact, the agella of DiW appeared similar to those of a DaA mutant strain and of bacteria lacking s28 (DiA), required for aA transcription. In contrast, the DcsrA and DcsrA/ DiW strains expressed normal agellar laments. The short agella of the DiW strain are probably composed mainly of the minor agellin FlaB, which is transcribed from an RpoN (s54)-dependent promoter. Upon deletion of both agellin genes (DaA/DaB), the bacteria no longer had laments but the hook structure was visible at the poles (black arrowheads, Fig. 3c). Furthermore, a DrpoN mutant strain had neither agella nor hooks. Motility assays revealed that the DcsrA or DiW strains showed a halo-radius reduction to 78% and 72% of WT, respectively (Fig. 3d). Likely due to its shorter agella, DiW also showed slower autoagglutination than WT, but greater than

6 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

a b

FliW flaA mRNA

Activity (binding)

CsrA

Translation

csrA

50 0 WT fliW

fliW csrA

C. jejuni

C. lari

C. curvus

W. succinogenes

H. pylori

T. maritima

T. lettingae

L. intracellularis

B. burgdorferi

B. subtilis

E. coli

comL

fliW

proC

uvrD truB

csrA ispE smpB

Epsilonproteobacteria

comL comL

fliW

proC

uvrD truB

csrA ispE smpB

fliW

uvrD truB

csrA ispE smpB

comL

fliW

proC

truB

csrA ispE smpB

**

**

** **

% FlaA-3xFLAG expression

300

250

150

50

0

300

250

150

fliW 1

proC

comL

fliW 2

csrA ispE smpB

flgK

flgL

csrA gatC

fliW smpB

200

200

flgK

flgL

csrA gatC

fliW smpB

100

100

flgK

flgL

fliW

csrA

flgN

flgK

flgL

fliW

csrA

***

***

flgN yviE

flgK

flgL

fliW

csrA

hag

WT

fliW csrA

csrA

fliW

alaS

csrA

tRNAs

PflaA PmetK

c

d

WT

csrA

4

12

***

***

9

csrA fliW

Flagellumlength (m)

***

Mean swim zone radius (mm)

3

2

***

6

fliW

fliA

flaA

1

3

***

***

rpoN

***

flaA flaB

0

0 WT

csrA

fliW

fliW csrA

fliA

rpoN

flaB

Figure 3 | The agellar assembly factor FliW binds and antagonizes CsrA. (a) Genomic context of csrA and iW homologues in diverse bacterial species (Campylobacter spp: C. jejuni, C. lari, C. curvus; Wolinella succinogenes; Helicobacter pylori; Thermotogales: T. maritima, T. lettingae; Lawsonia intracellularis; Borrellia burgdorferi; Bacillus subtilis; Escherichia coli). Blue: csrA homologs; dark or light red: iW homologues; shades of green: agellar genes. (b) (Top) Scheme of the antagonizing effect of FliW on CsrA-mediated translational repression of aA mRNA by direct binding of FliW to CsrA. (Bottom, left) Quantication of FlaA-3xFLAG using western blot in C. jejuni WT, DcsrA, DiW and DcsrA/DiW strains in mid-log phase (n 3 biological replicates).

Plotted is the means.e.m (**Po0.01, ***Po0.001 using Students t-test). (Bottom, right) Quantication of FlaA-3xFLAG using western blot in WT, DcsrA, DiW and DcsrA/DiW strain backgrounds where the aA promoter has been exchanged with the constitutive metK promoter. Please note that FlaA-3xFLAG levels expressed from the PmetK promoter represent B70% compared with the expression from its native PaA promoter. (c) Transmission electron micrographs of indicated strains harvested from MH agar. Black triangles indicate hook structures. (d) Average agella length (dark grey bars) of indicated strains from transmission electron micrographs using ImageJ (n425 measurements). Plotted is the means.d. (***Po0.001 versus WT using Students t-test). Motility was measured as average swimming distance (light grey bars) in soft agar. Bars show the means.e.m (***Po0.001 versus WT using

Students t-test).

the non-motile DiA and DrpoN mutants (Supplementary Fig. 12). Overall, these data suggest that, besides a mild effect on aA transcription, FliW affects post-transcriptional control of FlaA, and therefore lament assembly and motility, in a CsrA-dependent manner.

Expression of agellar mRNAs is not affected in DcsrA. Besides aA mRNA, many other agellar targets, such as the 50UTRs of aG, aB and gI, were strongly enriched in the CsrA-3xFLAG-coIP (4346-, 458- and 4170-fold, respectively; Table 1).

The aG, aB and gI leaders also have one or more GGA-containing motifs near their SD (Fig. 4a). In vitro gel-shift assays of in vitro transcribed aG, aB and gI leaders, and several other co-puried agellar mRNAs (Cj0040, gA and gM), conrmed CsrA binding (Fig. 4b and Supplementary Fig. 9b). The non-enriched Cj1324 mRNA, encoding a gene involved in agellin modication, or an unrelated mRNA fragment from H. pylori did

not shift with CsrA, conrming specic binding of CsrA to coIP-enriched transcripts (Supplementary Fig. 9c). However, CsrA afnity for aG, aB and gI leaders was lower (Kd 4350 nM)

than for the aA WT leader (Kd B50 nM, Fig. 4b). Although

FlaA-3xFLAG was upregulated upon csrA deletion (Fig. 2b), chromosomally tagged FlaG-3xFLAG, FlaB-3xFLAG and FlgI-3xFLAG levels did not change substantially (Fig. 4a).

FliW and aA mRNA titrate CsrA-mediated repression. The observed strong CsrA-mediated regulation of aG, aB and gI in the E. coli reporter system (Supplementary Figs 4 and 5) indicates that CsrA can, in principle, regulate these targets. Thus, we hypothesized that FliW, or even abundant mRNAs, might sequester CsrA under the examined routine growth conditions, obscuring any regulatory effect on these low-afnity targets. Because aA mRNA is highly abundant14 and expressed at the end of the agellar cascade, we reasoned aA mRNA might itself

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

a

140

**

G

G

**

% FlaG-3xFLAG expression

120

100

60

20

0

SD

SD

SD

SD

A

A

flaG leader

flaB leader

-U-U-U-U-A-A-A-A

U-U-U-U-A-A-A--

*

WT

csrA

80

FlaG-3xFLAG

10 kDa

55 kDa

***

A

100 84 %

40

***

- -G-A-A-A-T

G

5'- ...-3'

G-A-A-

A

U

Start codon

SD

SD

55 kDa

55 kDa

35 kDa

GroEL

***

M1

WT tagged

M1 fliW csrA

G

G

csrA

fliW

fliW csrA

M1 csrA

M1 fliW

A

A

-A-U-U-U-U-G-A-A

U-U-U-A-A-A-A-U-

G

G

WT

csrA

GroEL

A

A

140

*

U

U

% FlaB-3xFLAG expression

120

100

60

20

0

FlaB-3xFLAG

*

-U-U-U

A-A-A-

100 122 %

***

-G-

5'- ...-3'

...C-A-A-

U

G

G

C

-A-C

Start codon

80

* * * *

40

G

A

WT

csrA

*** ***

A

A

FlgI-3xFLAG

flgI leader

-G-A

U-C-

100 108 %

5'-...

U-A-A-A-

G

C

- -A

A-U-G ...-3'

WT tagged

M1 fliW csrA

GroEL

M1

M1 fliW

M1 csrA

start codon

55 kDa

csrA

fliW

fliW csrA

b

100

140

NS

80

**

60

% Bound

flaA

flgI

flaG

% FlgI-3xFLAG expression

120

100

60

20

0

80

60

40

20

0 20 40 60 80 100

80

**

***

40

flaB

40

20

0

WT tagged

csrA

fliW

M1

M1 fliW

M1 fliW csrA

500 1,000 1,500 2,000

fliW csrA

M1 csrA

CsrA (nM)

Figure 4 | CsrA binds to other agellar target mRNAs but csrA deletion does not affect their translation. (a) (Left) Predicted secondary structures of aG, aB and gI leaders using Mfold74 with putative GGA binding-sites of CsrA (blue) and SD sequences (grey). (Right) Western blot analyses of FlaG-3xFLAG, FlaB-3xFLAG and FlgI-3xFLAG in C. jejuni WT or DcsrA strains. (b) CsrA-binding afnities of agella mRNA leaders (r4 nM)

determined by in vitro gel-shift assays. The inset represents an enlargement of the binding curves for low CsrA concentrations. Shown is the means.d.

Figure 5 | The aA 50UTR and FliW inhibit CsrA-mediated regulation of agella genes. Quantication of FlaG-3xFLAG, FlaB-3xFLAG and FlgI-3xFLAG levels using western blot of the indicated C. jejuni NCTC11168 strains grown to mid-log phase (M1: GGA-AAA in SL1 of aA 50UTR).

Values were calculated based on at least three biological replicates. Shown is the means.e.m (*Po0.05, **Po0.01, ***Po0.001, using Students t-test). NS, not signicant.

titrate CsrA activity. To investigate the role of FliW and the aA mRNA as CsrA antagonists, we analysed FlaG-3xFLAG, FlaB-3xFLAG and FlgI-3xFLAG protein expression in loss-of-function strains of both antagonists. In line with FliW acting as a general CsrA antagonist that limits CsrA activity, deletion of iW led to a B3-fold decrease in FlaG-3xFLAG level, which was restored to

WT level in a DcsrA/DiW double mutant (Fig. 5 and Supplementary Fig. 13a).

Because aG and aA are primarily transcribed from s28-dependent promoters28 and are thus expressed at the same time, monitoring FlaG-3xFLAG might reveal the potential role of aA 50UTR as a CsrA antagonist. The chromosomal

M1 aA leader mutation (GGA - AAA in SL1, Fig. 2a), which leaves the coding region intact but abolishes CsrA binding (Fig. 2d), decreased FlaG-3xFLAG levels B3-fold (Fig. 5 and

Supplementary Fig. 13a). Upon introduction of DcsrA, FlaG-

8 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

3xFLAG expression was restored to WT levels, indicating decreased FlaG expression in the aA-M1 mutant is dependent on CsrA, and suggesting that the aA leader can also titrate CsrA. Combining both DiW and aA-M1 led to a tenfold reduction in

FlaG-3xFLAG levels, showing their cumulative effect in antagonizing CsrA. In line with this, the M1/DiW/DcsrA triple mutant restored FlaG-3xFLAG levels back to WT levels (Fig. 5 and Supplementary Fig. 13a). Growth curves showed that there was no major impact on growth of the individual mutations under the examined conditions (Supplementary Fig. 13b). Although the DiW and M1/DiW mutants showed a slightly increased growth rate compared with WT, this increase was less than a non-motile DiA strain.

To further conrm the role of the aA 50UTR as a CsrA antagonist, a B250-nt long aA_mini transcript comprising the aA leader and rst 17 codons followed by a stable ribosomal rrnB terminator was ectopically expressed from the native aA promoter (Supplementary Fig. 14a). Expression of the aA_mini transcript in a DiW mutant, which has strong CsrA-mediated aA translational repression, increased FlaA-3xFLAG levels around 2.6-fold (Supplementary Fig. 14b). This indicates aA_mini can bind and antagonize CsrA and partially relieve CsrA-mediated repression of aA translation. A smaller, yet signicant, complementation of the effect of a iW deletion was also observed for FlaG-3xFLAG levels.

Next, the effect of the two antagonists on CsrA-mediated regulation of the RpoN-dependent genes aB and gI was evaluated. A similar, yet less pronounced effect compared with FlaG-3xFLAG, was observed for FlaB-3xFLAG upon single or double mutations of iW and M1. In contrast, FlgI-3xFLAG levels were only signicantly reduced upon iW deletion (Fig. 5 and Supplementary Fig. 13a). Overall, this reveals FliW as the major CsrA antagonist under the examined growth conditions that titrates, along with the aA mRNA antagonist, CsrA from lower afnity agellar targets such as aG.

aA mRNA localizes to the poles of elongating cells. As aA mRNA can titrate CsrA activity, we wondered when aA mRNA levels change to modulate CsrA activity. Expression of aA mRNA appeared constitutive during growth (Supplementary Fig. 13c). However, in the amphitrichously agellated C. jejuni, after every cell division, a new agellum has to be synthesized at the new pole of each daughter cell. As bacteria in batch culture are not synchronized in cell cycle, differences in aA mRNA expression might be obscured because of the population-based northern analysis. To monitor aA mRNA expression in single bacteria, we performed RNA-FISH (uorescence in situ hybridization) in xed C. jejuni cells from exponential phase. Although the control RNA, 16S rRNA (Fig. 6a, green), was visible in all

a

b

aA mRNA (Cy5)

Merge

aA mRNA (Cy5)

Merge

16S rRNA (FITC)

1.4 m

2.1 m

16S rRNA (FITC)

1.3 m

Fluorescence

intensity

Fluorescence

intensity

Long axis length

of bacteria

Long axis length

of bacteria

c

***

Cell length in m

2.0

1.0

0

1.5

Localized

Non-localized

n = 85 n = 56

0.5

Figure 6 | aA mRNA localizes to the poles of shorter cells. (a) RNA-FISH analysis of 16S rRNA (FITC-labelled DNA oligonucleotide probe, green) and aA mRNA (14 Cy5-labelled single-stranded DNA oligonucleotide probes, red) in C. jejuni WT cells in mid-log phase using confocal microscopy (scale bar, 1 mm). (b) A magnied RNA-FISH image showing the distribution of uorescence signals. aA mRNA (Cy5) and 16S rRNA (FITC) signals were quantied along the long axis length of bacteria using ImageJ software and were subsequently merged as shown at the bottom of the panel (scale bar, 1 mm). The length of individual cells was also quantied using ImageJ. Statistical analysis for average aA mRNA and 16S rRNA signals over the cell length is provided in Supplementary Fig. 15. (c) Average C. jejuni WT cell lengths in bacteria where aA mRNA is localized (56 cells) or non-localized (85 cells), ***Po10 15 using Students t-test.

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

cells, aA mRNA (Fig. 6a, red) was detected in only some of the cells. As a negative control, we also performed aA mRNA FISH on a DiA mutant strain (Fig. 7a), which showed no expression of aA (Supplementary Fig. 11a). Whereas 16S rRNA was equally distributed throughout the cell, aA mRNA was specically detected at the cell poles in B20% of WT cells (Fig. 6a,b).

Quantication of cell length across the population showed that cells with localized aA mRNA were signicantly shorter than cells without aA expression (Fig. 6b,c). Live-cell imaging of a non-motile C. jejuni strain (DiA) over two or three division cycles showed regular patterns of an increase in cell length until cells divide at mid-cell, resulting in short daughter cells (Supplementary Fig. 16). This indicates shorter cells likely correspond to cells that have divided and are elongating. Together, these data suggest differential expression of aA mRNA during the cell cycle and accumulation in elongating cells at the required site of its encoded protein.

FliW impacts aA mRNA localization via CsrA. To investigate whether CsrA-FliW impacts aA mRNA localization, we next performed RNA-FISH in DiW, DcsrA and DiW/DcsrA mutant strains. Although csrA deletion had no effect on aA localization, it was completely abolished in a DiW mutant (Fig. 7a). Instead of a polar localization, aA mRNA was now dispersed throughout the cell. The loss of aA mRNA localization upon iW deletion was not due to lower transcript abundance as its mRNA level is increased despite strong repression at the protein level (Supplementary Fig. 11a). Strikingly, aA mRNA localization was

restored to the cell poles in the DiW/DcsrA double mutant, showing CsrA affects localization of aA mRNA. As a further conrmation of aA mRNA localization, we performed super-resolution imaging of aA mRNA FISH in WT and mutant strains using direct stochastic optical reconstruction microscopy (dSTORM)29, which has only recently been applied for bacterial RNA localization30. dSTORM analysis fully supported and complemented the observations from confocal microscopy analysis (Fig. 7b and Supplementary Fig. 17). Overall, this suggests a model where aA translation is required for polar localization: upon deletion of iW, CsrA is released and in turn strongly represses aA mRNA translation to impede its localization to the poles.

Polar aA mRNA localization requires its translation. To support the translation-dependent model of aA localization, we constructed several point mutants in the native aA gene that either maintain or disrupt aA translation (Fig. 8a). Mutation of the start codon of aA (AUG - AAG (X1) or AUU (X2)) to abolish translation initiation resulted in dispersed aA mRNA (Fig. 8b). In contrast, when the start codon was changed to an alternative start codon (AUG - GUG (X3)), aA mRNA still localized to the cell poles, indicating translation of aA mRNA is indeed required for polar localization. Mutation of the third aA codon to a stop codon (UUU - UAG (X4)) also resulted in a completely dispersed aA mRNA signal (Fig. 8b and Supplementary Fig. 17). In contrast, aA mRNA with a synonymous silent mutation (UUU - UUC (X5); both encoding Phe)

a

10 Cells (averaged)

b

WT

WT csrA

csrA iW

csrA

iW

Fluorescence intensity

iW

csrA iW

aA mRNA

16S rRNA

Long axis length of bacteria

iA

aA mRNA (Cy5)

16S rRNA (FITC)

Merge

Figure 7 | CsrA and FliW inuence aA mRNA localization to the poles. (a) RNA-FISH analysis (Left: confocal microscopy images; Right: averaged uorescence intensity along the long axis based on 10 cells) of 16S rRNA (green) and aA mRNA (red) in C. jejuni NCTC11168 WT, DcsrA, DiW, DcsrA/

DiW and DiA strains in mid-log phase. FITC and Cy5 channels were merged in the microcopy images in the third lanes (scale bar, 1 mm). (b) Super-resolution microscopy imaging of aA mRNA RNA-FISH (14 Cy5-labelled oligos) in the indicated C. jejuni strains using dSTORM imaging. Cell boundaries from bright-eld images are depicted by white dotted lines.

10 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

a

G

G

A

G

G

A

AAG (X1), AUU (X2), GUG (X3)

UAG (X4), UUC (X5)

UAA (X6)

5'

AUG

UUU

CAA

UAG

3'

Stop codon

10 Cells (averaged)

Start codon

3rd Codon

101st Codon

b

X1 Start codon (AUG > AAG)

X2 Start codon (AUG > AUU)

X3 Start codon (AUG > GUG)

X43rd Codon (UUU > UAG)

X6 101st Codon (CAA > UAA)

Fluorescence intensity

X53rd Codon (UUU > UUC)

aA mRNA

16S rRNA

aA mRNA (Cy5)

16S rRNA (FITC)

Merge

Long axis length of bacteria

Figure 8 | Translation is required for aA mRNA localization to the cell poles. (a) Point mutations in aA mRNA that were introduced at the native aA locus. Mutations X1, X2, X4 and X6 abolish or prematurely stop aA translation, whereas X3 and X5 represent silent mutations. (b) RNA-FISH analysis (Left: confocal microscopy images; Right: averaged uorescence intensity along the long axis based on 10 cells) of C. jejuni point mutant strains depicted in a. FITC and Cy5 channels were merged in the third rows of the microscopy images (scale bar, 1 mm).

at the third codon localized similarly to the WT mRNA. Some of the mutations that abolish translation (X1, X2) lead to reduced (5080% of WT) aA mRNA levels (Supplementary Fig. 18). Nonetheless, as the strain expressing the aA mRNA with a stop mutation at the third codon (X4), which also showed abolished polar mRNA localization, had even higher (B170%) aA

expression levels than WT, it is unlikely that reduced (or increased) aA mRNA levels lead to loss of localization. To determine the effect of terminating translation at a downstream position, we introduced a stop codon at the 101st codon of aA (CAA - UAA (X6)). This mutant showed partial polar aA mRNA localization, suggesting the N-terminal peptide might be

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

required for recruiting aA mRNA to the cell poles. Overall, these data support a role of the FliW/CsrA post-transcriptional network in controlling translation-dependent polar aA mRNA localization in C. jejuni.

DiscussionUsing genome-wide RIP-seq, we have identied direct RNA targets of the translational regulator CsrA in a bacterium that lacks the canonical antagonizing sRNAs. Our study revealed the major agellin mRNA is both the main CsrA target and a dual-function mRNA, which can titrate CsrA activity together with the FliW protein, the main CsrA antagonist (Fig. 9). Compared with microarray-based transcriptome analyses of csrA loss-of-function strains31,32, which might reveal indirect effects or miss targets because of a lack of changes in target mRNA levels despite translational repression, a coIP approach facilitates the identication of direct targets and binding sites. Sanger sequencing of cDNAs from an RsmA-coIP identied six target mRNAs in P. aeruginosa32. RNA-seq of a CsrA-coIP inE. coli revealed 721 co-puried transcripts33, and in vivo ultraviolet crosslinking combined with RNA-seq (CLIP-seq) revealed 467 potential CsrA-binding sites in Salmonella typhimurium, including binding sites in many virulence mRNAs34. In our RIP-seq approach, we used untagged WT strains as a negative control to allow for elimination of nonspecically bound transcripts. Our peak-detection tool conrmed the high specicity of this approach, as it detected an ANGGA sequence in 324/328 targets, which resembles the CsrA consensus-motif determined by in-vitro selection25. Besides canonical binding to 50UTRs or early codons5,35, our coIP also revealed CsrA binding within coding regions or between genes in polycistrons to mediate discoordinate operon regulation.

Our coIP approach revealed many mRNAs of agellar genes as direct CsrA targets. The motility defect of DcsrA suggests that tight regulation of agellar genes by CsrA, and especially of the major agellin FlaA, is required for proper motility. Balancing CsrA activity through the antagonizing protein FliW also appears crucial for agellar assembly, as we observed that a C. jejuni NCTC11168 DiW mutant expresses short agella, as also reported in other strains27,36, and is defective for autoagglutination and motility in both B. subtilis and C. jejuni9,26. Although CsrA impacts motility by directly controlling agellin expression in C. jejuni, B. subtilis and Borrelia, the strong motility defect of an E. coli csrA mutant37 is due to a requirement of CsrA for stabilization of the mRNA encoding the master regulator FlhDC38. The agellum also plays an essential, multi-factorial role in C. jejuni colonization and pathogenesis, including secretion of Cia/Fed effectors28,39, and is required for proper cell division40. Future studies might reveal CsrA-affected phenotypes beyond motility.

Instead of CsrA-activity control by antagonizing sRNAs5, we demonstrated that the aA mRNA itself can titrate CsrA. This represents a new mode of CsrA activity control by a target mRNA-derived antagonist. The aA leader has higher afnity for CsrA compared to other agellar targets. It has two GGA motifs in adjacent hexaloops, resembling high-afnity CANGGANG-containing apical hexaloop structures targeted by CsrA/ RsmE25,41. The 21-nt spacing between the aA GGA motifs is close to the 18-nt optimal intersite distance for binding of a CsrA dimer42. Whereas aA mRNA probably only binds one CsrA dimer, multiple RsmE dimers are cooperatively assembled on RsmZ sRNA8,41. CsrA titration by a 50UTR has recently been shown to mediate hierarchical control of mbriae expression in Salmonella typhimurium43. The mAICDHF mRNA leader, which in contrast to aA mRNA is not itself a CsrA target, cooperates with the CsrB/C sRNAs to antagonize CsrA-mediated activation of plasmid-encoded mbriae. Small RNAs other than CsrB/C can also sequester CsrA in addition to functioning as antisense RNAs44. Global approaches such as RIP-seq are ideally suited to identify additional antagonizing sRNAs or members of the emerging class of dual-function, cross-regulating mRNAs2,3.

Analysis of aA mRNA expression in single bacteria using RNA-FISH showed that this transcript localizes to the poles of shorter, and presumably elongating, cells. As a new agellum is synthesized after each cell division at the new pole of the amphitrichous C. jejuni, polar aA mRNA localization might facilitate this process. This temporal and spatial modulation of aA mRNA expression might also affect CsrA-mediated regulation of other agellar genes through mediating varying levels of this CsrA RNA antagonist. Mutations that either abolish or maintain translation showed aA translation is required for its polar localization. Bacterial mRNA localization has only recently been described and unlike eukaryotes the underlying mechanisms and regulation of this process are poorly understood45,46. Besides co-translational targeting of mRNAs to the required sites of their encoded products, translation-independent mechanisms of RNA localization have also been described20,21, including spatial expression according to chromosome organization. We observed that a aA mRNA variant with a premature stop-codon mutation at the 101st codon partially localizes, suggesting a role of the N-terminus in directing the nascent peptide along with the mRNA to the secretion apparatus. Little is known how agellar substrates are selected for secretion, as they do not share a secretion-signal sequence or cleavable signal peptide. N-terminal domains are required for secretion of agellar proteins in diverse bacteria, including C. jejuni36, and both 50UTR and N-terminal peptide secretion signals have been shown to contribute to secretion efciency47. In addition, agellar

FlaA

Activity

Translation

flaA mRNA

FliW

Activity

Translation and localization

Activity

CsrA

Translation

Flagellar mRNAs

Figure 9 | Model depicting the C. jejuni CsrA-FliW regulatory network. Schematic representation of the regulatory circuit and the putative roles of CsrA, FliW and FlaA proteins along with aA mRNA in the CsrA-FliW regulon of C. jejuni. The post-transcriptional regulatory protein CsrA represses translation of multiple agellar mRNAs including aA mRNA, encoding the major agellin, by direct binding to the mRNAs. The FliW protein can directly bind and titrate CsrA activity and in-turn affects CsrA-mediated post-transcriptional regulation of agellar genes. FliW can also bind to the FlaA protein, which releases FliW-mediated sequestration of CsrA. The abundant aA mRNA is the main target of CsrA translational repression but can also act as a regulatory sponge and titrate CsrA activity together with the main CsrA antagonist FliW. Furthermore, aA mRNA localizes to the cell poles of elongating cells. Polar localization of aA mRNA itself is dependent on its translation, which is controlled by the CsrA-FliW regulatory network.

12 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

chaperones play a role in regulating the coupling of translation to secretion of agellar substrates48. In Yersinia, cis-encoded RNA-localization elements in the early coding region are required for secretion of effector proteins by type III secretion systems49. Future studies will identify and clarify the role of elements, either in the protein N-terminus or the mRNA 50UTR, as well as potential interaction partners that are crucial for directing the peptide and/or mRNA to the cell poles and secretion apparatus. Besides the requirement of aA translation for localization, other factors such as the aA genomic location or the transcriptional complex might also contribute to polar aA mRNA localization.

Our study revealed an unexpected function for the CsrA-FliW network in spatial and temporal gene-expression control, and specically FliW affects translation-dependent polar localization of the agellin mRNA by antagonizing CsrA-mediated translational repression. The limited CsrA activity in WT cells under standard growth conditions, because of sequestration by the FliW protein antagonist, probably allows sufcient translation of aA mRNA for its polar localization. Strong CsrA-mediated translation repression of aA upon iW deletion is probably responsible for the diffuse aA localization in the DiW mutant. CsrA binding might mediate storage of translationally inactive aA mRNA until synthesis of FlaA is required or proper localization is achieved, similar to mRNP granules in eukaryotes50. Future studies will show whether other agellar mRNAs also polarly localize and if the CsrA-FliW regulatory network also impacts their localization. CsrA-mediated regulation of mRNA localization might also occur in B. subtilis and B. burgdorferi, where CsrA overexpression represses the major agellin5153. An analogous system might have also evolved in the Alphaproteobacterium Caulobacter crescentus, which encodes two proteins with opposing activities on agellin regulation, FlaF and FlbT, whereby FlbT post-transcriptionally regulates agellin expression54.

Our identication of C. jejuni CsrA titration by FliW indicates that CsrA-activity control by a protein antagonist, a mechanism rst identied in the Gram-positive B. subtilis9, is more widespread than previously appreciated. Besides the post-transcriptional effect of FliW on aA and other agellar genes by antagonizing CsrA, deletion of iW directly or indirectly increases aA transcription. Transcription of hag is also twofold upregulated in B. subtilis upon iW deletion9,55. Although FliW appears to be the main CsrA antagonist, its synergistic interplay with the aA mRNA antagonist affects other agellar genes showed that RNA-based regulation can also impact CsrA activity in this type of Csr network. Gammaproteobacterial genomes encode CsrA56 as well as the antagonizing sRNAs5 and an anti-correlation between the presence of the CsrB/C sRNAs and FliW has been observed57. As the csrA gene is located next to a tRNA cluster in E. coli, this strongly suggests the pleiotropic function of CsrA in Gammaproteobacteria might have been horizontally acquired, followed by evolution of the antagonizing sRNAs. Thus, the conserved or possibly more ancient function of the CsrAFliW system might be to mediate temporal and spatial control of proper agellum assembly. During our conservation analysis we observed that certain non-agellated Campylobacter species, such as C. hominis, C. gracilis and C. ureolyticus, lack csrA and iW homologues, further supporting their conserved function in agellar regulation. Further studies are required to unravel the full complexity of the CsrA-FliW regulatory network and its impact on RNA localization.

Methods

Bacterial strains, oligonucleotides and plasmids. All C. jejuni and E. coli strains used in this study are listed in Supplementary Table 3 and DNA oligonucleotides in

Supplementary Table 4, respectively. Plasmids are summarized in Supplementary

Table 5.

Bacterial growth conditions. C. jejuni strains were routinely grown on Mller-Hinton agar plates or with shaking in Brucella broth (BB), both supplemented with 10 mg ml 1 vancomycin, at 37 C under microaerobic (10% CO2, 5% O2) conditions as described previously14. The agar was further supplemented with marker-selective antibiotics (20 mg ml 1 chloramphenicol, 50 mg ml 1 kanamycin , 20 mg ml 1 gentamicin or 250 mg ml 1 hygromycin B) where appropriate. E. coli strains were grown aerobically at 37 C in Luria-Bertani (LB) medium supplemented with appropriate antibiotics. For induction of arabinoseinducible pBAD promoter, 0.001% ( ) or 0.003% ( )

L-arabinose was added

to LB media.

Construction of bacterial mutant strains. All C. jejuni mutant strains (deletion, chromosomal 3xFLAG-tagging, chromosomal point mutations) were constructed using double-crossover homologous recombination. Cloning strategies and the generation of constructs are described in detail in the Methods and Supplementary Methods. Oligonucleotides used to amplify regions of upstream/downstream homology and resistance cassettes for homologous recombination, as well as recipient strains and oligonucleotides for validation of mutant strains by colony PCR, are listed in Supplementary Table 6 for each generated strain. Introduction of PCR products with 500 bp homologous ends or genomic DNA with mutant constructs into C. jejuni was performed by electroporation or natural transformation, respectively, as described previously14.

Construction of 3xFLAG epitope-tagged proteins in C. jejuni. C. jejuni genes were chromosomally tagged at their C-terminus either by cloning of constructs for C-terminal epitope tagging on plasmids or by construction of 3xFLAG constructs by overlap PCR.

Tagging of proteins using PCR products amplied from plasmid constructs. The CsrA, FlaA, FlgI and FlaB proteins were fused to a 3xFLAG epitope at their C-termini by cloning regions encoding B500 bp of their C-terminal coding region (C-term) and B500 bp downstream of the stop codon (DN) into plasmid pGG1 to ank a 3xFLAG tag and aphA-3 KanR cassette. Afterwards, the 3xFLAG-tag constructs were amplied by PCR and introduced into the chromosome of C. jejuni strains by electroporation and double-crossover homologous recombination. An example of this plasmid cloning strategy is described for csrA. Approximately 500 bp of the region downstream of csrA was amplied from genomic DNA (gDNA) with primers CSO-0173/-0174. These primers included XbaI and EcoRI sites, respectively. Following cleanup, the PCR product was digested with EcoRI and XbaI and ligated into a similarly digested pGG1 backbone, generated by inverse PCR with primers CSO-0074/-0075, to create pGD2-1. The plasmid was veried by colony PCR with primers JVO-0054/CSO-0173 and the sequence was veried using JVO-0054. Next, the backbone of this plasmid, including the csrA DN region, was amplied by PCR with primers CSO-0073 (XhoI) and JVO-5142 (blunt). The C-terminal coding region of csrA (B500 bp) without the stop codon was amplied with primers CSO-0171/-0172 from NCTC11168 WT gDNA. The sense primer (CSO-0172) included an XhoI site, whereas the antisense primer (CSO-0171) contained a 50-phosphate. Both the plasmid backbone with the DN

insert and the C-term insert were digested with XhoI and ligated to create plasmid pGD4-1. Integration of the PCR product was conrmed by colony PCR using primers CSO-0172/-0023 and the plasmid was validated by sequencing using CSO-0023. The entire integration cassette was then amplied with Phusion High-Fidelity DNA polymerase (NEB) using primers CSO-0172/-0173 and electroporated into C. jejuni and selected on kanamycin plates. Mutants were conrmed by colony PCR with primers CSO-0196/-0023 and western blot analysis with an anti-FLAG antibody.

3xFLAG tagging of proteins by overlap PCR. Construction of a C-terminal 3xFLAG translational fusion at its native locus was performed by overlap PCR for aG as described in Supplementary Methods for gene deletions, but with the following modications. The nal overlap PCR product contained B500 bp of the

C-terminal coding region of aG minus the stop codon (C-term) and B500 bp downstream of aG (DN) for homologous recombination. These regions anked an in-frame 3xFLAG tag and stop codon followed by an aphA-3 KanR cassette. For example, for tagging aG, the 3xFLAG tag and KanR cassette was amplied from plasmid pGG1 with primers JVO-5142 and HPK2. The C-term region of aG was amplied using primers CSO-1002/-1098, where CSO-1098 is antisense and contains region of complementarity at its 50 end to the 3xFLAG tag/JVO-5142, from NCTC11168 gDNA. The DN region was amplied using primers CSO-1099/-1003, where CSO-1099 is sense to aG DN and contains a region of complementarity to the 30 end of the KanR cassette/primer HPK2. If the coding region of the target gene contained sequences required for expression of a downstream ORF (that is, SD sequence or codons), these sequences were included in the DN amplicon. The three PCR products were then used for overlap PCR with primers CSO-1002/-1003, and the resulting amplicon was electroporated intoC. jejuni, followed by selection of positive clones on kanamycin plates. Mutants

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

were checked by colony PCR with primers CSO-1005/HPK2 and western blot analysis with an anti-FLAG antibody.

Introducing chromosomal point mutations into the aA leader. To introduce point mutations into the 50UTR of aA at the native locus, a 1,100-bp region around the aA promoter was amplied using oligos CSO-0752/-0753. These primers introduced XhoI and XbaI sites, respectively, into the resulting PCR product. After XhoI and XbaI digestion, the product was then ligated into a similarly digested plasmid pJV752-1, resulting in plasmid pGD70-5. Plasmid pGD70-5 was checked by colony PCR using primers pZE-A/CSO-0753 and sequencing with pZE-A. Next, plasmid pGD70-5 was amplied by inverse PCR using primers CSO-0754/-0755, thereby introducing NdeI and BamHI restriction sites 40 nt upstream of the aA transcriptional start site (TSS). An aac(3)-IV gentamicin resistance cassette with its own promoter and terminator was amplied using CSO-0483/-0576 and introduced into PCR-amplied pGD70-5 in the reverse orientation to aA, just upstream of its promoter, using the NdeI/BamHI restriction sites, resulting in plasmid pGD76-1. Plasmid pGD76-1 was checked by colony PCR using primers CSO-0576/-0753 and sequencing with CSO-0753.

Point mutations were then introduced into the aA 50UTR by inverse PCR on pGD76-1 using complementary oligos harbouring the desired mutation, followed by DpnI digestion and transformation of the resulting puried PCR product intoE. coli TOP10. For introduction of the aA M1 mutation (GGA4AAA in stem-loop SL1 of the aA leader), oligonucleotides CSO-1114/-1115 were used for PCR on pGD76-1. The mutation was conrmed in the resulting plasmid pGD92-1 by sequencing with CSO-0753. Similarly, the aA M2 (GGA4UGA in stem-loop

SL1 of the aA leader), M3 (GGA4GGG in stem-loop SL2 of the aA leader), X1 start codon (AUG4AAG), X2 start codon (AUG4AUU), X3 start codon (AUG4GUG), X4 3rd codon (UUU4UAG), X5 3rd codon (UUU4UUC) and X6 101st codon (CAA4UAA) mutations were introduced using primer pairs CSO-0757/-0758, CSO-1116/-1117, CSO-2019/-2020, CSO-2827/-2828, CSO-2825/-2826, CSO-2829/-2830, CSO-2831/-2832 and CSO-2833/-2834, respectively, resulting in plasmids pGD77-1, pGD93-1, pGD114-2, pGD205-1, pGD204-1, pGD206-1, pGD207-1 and pGD208-1, respectively. For combination of the aA M2 and M3 mutations, a similar mutagenesis approach was performed based on PCR amplication of the M2 plasmid pGD77-1 using oligonucleotides CSO-1116/-1117, resulting in pGD95-1 harbouring both the mutations. To introduce the aA 50UTR mutations into C. jejuni, a PCR product covering the homologous ends and the gentamicin resistance cassette was amplied from the respective WT (pGD76-1) or mutant plasmids using CSO-0752/-0850 and electroporated into C. jejuni as described above. To conrm introduction of point mutation in C. jejuni, colony PCR was performed using CSO-0576/-0753 and sequencing with CSO-0850.

Construction of E. coli mutants. The E. coli DpgaA and DpgaA DcsrA deletion strains were constructed in the TOP10 background using the l Red protocol58. Briey, a kanamycin resistance gene, amplied from plasmid pKD4 using primers CSO-0652/-0653, was used to replace the entire pgaA ORF excluding the start and stop codon. The mutant strain was veried by colony PCR using the primer pairs CSO-0654/-0653 and CSO-0652/-0655. After verication, helper plasmid pCP20 containing FLP recombinase was introduced to remove the kanamycin resistance marker58. The helper plasmid, which is temperature-sensitive and carries an ampicillin resistance marker, was then cured by recovering colonies at 37 C and conrming ampicillin sensitivity, resulting in strain CSS-0556. Similarly, the ORF of the csrA gene excluding the start and stop codon was then replaced by the kanamycin resistance marker (amplied using CSO-0611/-0612) in the DpgaA strain resulting in strain CSS-0557, harbouring both pgaA and csrA deletions. The csrA deletion was veried by colony PCR using primer pairs CSO-0639/-0612 and CSO-0611/-0640.

RIP-seq of C. jejuni CsrA-3xFLAG. coIP combined with RNA-seq (RIP-seq) to identify direct RNA-binding partners of CsrA-3xFLAG in C. jejuni was performed as previously described18,59 with minor modications.

CoIP of RNA with CsrA-3xFLAG. CoIP of chromosomally epitope-tagged C. jejuni CsrA with an anti-FLAG antibody and Protein A-Sepharose beads was performed from lysates of C. jejuni NCTC11168 and 81-176 WT (control) and isogenic csrA-3xFLAG strains grown in 100 ml (50 ml 2 asks) BB containing 10 mg ml 1

vancomycin to mid-exponential phase (OD600 0.6) at 37 C as described

previously for H. pylori18. Cells were harvested by centrifugation at 6,000g for15 min at 4 C. Afterwards, cell pellets were resuspended in 1 ml Buffer A(20 mM Tris-HCl, pH 8.0, 150 mM KCl, 1 mM MgCl2, 1 mM dithiothreitol (DTT))

and subsequently centrifuged (3 min, 11,000g, 4 C). The pellets were shock-frozen in liquid nitrogen and stored at 80 C. Frozen pellets were thawed on ice and

resuspended in 0.8 ml Buffer A. An equal volume of glass beads was then added to the cell suspension. Cells were then lysed using a Retsch MM40 ball mill(30 s 1, 10 min) in pre-cooled blocks (4 C) and centrifuged for 2 min at 15,200g, 4 C. The supernatant was transferred to a new tube, and an additional 0.4 ml of Buffer A was added to the remaining un-lysed cells with beads. Lysis of the remaining cells was achieved by a second round of lysis at 30 s 1 for 5 min.

Centrifugation was repeated and this second supernatant was combined with

the rst one. The combined supernatant was centrifuged again for 30 min at 15,200g, 4 C for clarication and the resulting supernatant (lysate fraction) was transferred to a new tube. The lysate was incubated with 35 ml anti-FLAG antibody (Monoclonal ANTI-FLAG M2, Sigma, #F1804) for 30 min at 4 C on a rocker. Next, 75 ml of Protein A-Sepharose (Sigma, #P6649), prewashed with Buffer A, was added and the mixture was rocked for another 30 min at 4 C. After centrifugation at 15,200g for 1 min, the supernatant was removed. Pelleted beads were washed ve times with 0.5 ml Buffer A. Finally, 500 ml Buffer A was added to the beads and RNA and proteins were separated by phenol-chloroform-isoamyl alcohol extraction and precipitated as described previously18. From each coIP, 7001,000 ng of RNA was recovered. 100 ml of 1 protein loading buffer (62.5 mM

Tris-HCl, pH 6.8, 100 mM DTT, 10% (v/v) glycerol, 2% (w/v) SDS, 0.01% (w/v) bromophenol blue) was added to the nal protein sample precipitated along with beads. This sample was termed the coIP sample. For verication of a successful coIP, protein samples equivalent to 1.0 OD600 of cells were obtained during different stages of the coIP (culture, lysate, supernatant, wash and coIP (beads)) for further western blot analysis. One hundred microlitres of 1 protein loading

buffer was added to the protein samples and boiled for 8 min. Protein sample corresponding to an OD600 of 0.1 or 0.15 (culture, lysate, supernatant and wash fraction) and 10 or 5 (for proteins precipitated from beads) were used for western blot analysis.

RIP-Seq cDNA library preparation. Residual gDNA was removed from thecoIP RNA samples isolated from the control (WT) and CsrA-3xFLAG coIPs of the two strains C. jejuni NCTC11168 and 81-176 using DNase I treatment. cDNA libraries for Illumina sequencing were constructed by vertis Biotechnologie AG (http://www.vertis-biotech.com

Web End =http://www.vertis-biotech.com) in a strand-specic manner as described previously14. In brief, equal amounts of RNA samples were poly(A)-tailed using poly(A) polymerase. Then, 50-triphosphates were removed using tobacco acid pyrophosphatase, and an RNA adapter was then ligated to the resulting 50-monophosphate. First-strand cDNA was synthesized with an oligo(dT)-adapter primer using M-MLV reverse transcriptase. In a PCR-based amplication step, using a high-delity DNA polymerase, the cDNA concentration was increased to 2030 ng ml 1. For all libraries, the Agencourt AMPure XP kit (Beckman Coulter

Genomics) was used to purify the DNA, which was subsequently analysed by capillary electrophoresis.

A library-specic barcode for multiplex sequencing was included as part of a 30-sequencing adapter. The following adapter sequences ank the cDNA inserts:

TrueSeq_Sense_primer 50-AATGATACGGCGACCACCGAGATCTACACTC TTTCCCTACACGAC

GCTCTTCCGATCT-30TrueSeq_Antisense_NNNNNN_primer (NNNNNN 6nt barcode for

multiplexing)

50-CAAGCAGAAGACGGCATACGAGAT-NNNNNN-GTGACTGGAG TTCAGACGTGTGCTCTTCCGATC(dT25)-30.

The samples were sequenced on an Illumina HiSeq instrument with 100 cycles in single-read mode. The resulting read numbers are listed in Supplementary Table 1.

Analysis of deep sequencing data. To assure high sequence quality, the Illumina reads in FASTQ format were trimmed with a cutoff phred score of 20 by the programme fastq_quality_trimmer from FASTX toolkit version 0.0.13. After trimming, poly(A)-tail sequences were removed and a size ltering step was applied in which sequences shorter than 12 nt were eliminated. The collections of remaining reads were mapped to the C. jejuni NCTC11168 (NCBI Acc.-No: NC_002163.1) and 81-176 (NCBI Acc.-No: NC_008770.1, NC_008787.1, NC_008790.1) genomes using segemehl60 with an accuracy cutoff of 95%. Mapping statistics are listed in Supplementary Table 1. Coverage plots representing the numbers of mapped reads per nucleotide were generated. Reads that mapped to multiple locations contributed a fraction to the coverage value. For example, reads mapping to three positions contributed only one-third to the coverage values. Each graph was normalized to the number of reads that could be mapped from the respective library. To restore the original data range, each graph was then multiplied by the minimum number of mapped reads calculated over all libraries.

The overlap of sequenced cDNA reads to annotations was assessed for each library by counting all reads overlapping selected annotations on the sense strand. These annotations consist of strain-specic NCBI gene annotations complemented with annotations of previously determined 50UTRs and small RNAs14. Each read with a minimum overlap of 10 nt was counted with a value based on the number of locations where the read was mapped. If the read overlapped more than one annotation, the value was divided by the number of regions and counted separately for each region (for example, one-third for a read mapped to three locations).

Enrichment analysis of CsrA targets. Enrichment of transcripts in the CsrA-3xFLAG coIP versus control coIP libraries was determined based on mapped cDNA read counts for annotations provided in NC_002163.gff (NCBI) for NCTC11168 using GFOLD version 1.0.9 (ref. 61) but with manually dened normalization constants based on the number of reads that could be mapped to the respective libraries. For determination of genes enriched in the CsrA-3xFLAG-tagged library, log2 fold changes (FCs) rather than GFOLD values were used.

14 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

Similar analysis was done for strains 81-176 using annotations provided in NC_008787.gff (chromosome), NC_008770.gff (pVir plasmid) and NC_008790.gff (pTet plasmid).

Peak detection and CsrA-binding motif analyses. To automatically dene CsrA-bound RNA regions or peaks from the CsrA-3xFLAG coIP data sets, an in-house tool sliding_window_peak_calling_script was developed based ona sliding window approach. A detailed description of the tool will be described elsewhere. The script has been deposited at Zenodo (https://zenodo.org/record/49292

Web End =https://zenodo.org/record/ https://zenodo.org/record/49292

Web End =49292 ) under DOI 10.5281/zenodo.49292 (http://dx.doi.org/10.5281/zenodo.49292

Web End =http://dx.doi.org/10.5281/ http://dx.doi.org/10.5281/zenodo.49292

Web End =zenodo.49292 ). The script is written in Python 3 and requires installation of the Python 3 packages numpy and scipy for execution.

In brief, the sliding_window_peak_calling_script software uses normalized wiggle les of the CsrA-3xFLAG and control coIP libraries as input to determine sites showing a continuous enrichment of the CsrA-3xFLAG-tagged library compared with the control. The identication of enriched regions is based on four parameters: a minimum required fold change (FC) for the enrichment, a factor multiplied by the 90th percentile of the wiggle graph, which reects the minimum required expression (MRE) in the tagged library, a window size in nt (WS), for which the previous two values are calculated in a sliding window approach, and a nucleotide step size (SS), which denes the steps in which the window is moved along the genomic axis. All consecutive windows that fulll the enrichment requirements are assembled into a single peak region. The peak detection is performed separately for the forward and reverse strand of each replicon. For the CsrA-3xFLAG coIP data set, the following parameters were used: FC 5, MRE 3,

WS 25 and SS 5.

For the prediction of consensus motifs based on the peak sequences, MEME24 and CMnder 0.2.1 (ref. 62) were used. For MEME24 predictions, the following settings were applied: Search 0 or 1 motif of length 47 bp per sequence in the given strand only. To search for the presence of a structural motif, CMnder 0.2.1 (ref. 62) was run on the enriched peak sequences with default parameters except for allowing a minimum single stem loop candidate length of 20 nt. The top-ranked motif incorporated 276 of the 328 sequences and was visualized by R2R63.

Functional classes enrichment analysis. To check for overrepresentation of functional classes of CsrA-bound genes, we considered genes with at least vefold enrichment in their 50UTR and/or coding sequence in the CsrA-3xFLAG coIP library (versus control) as CsrA-bound and the remaining genes as unbound. We applied an existing functional classication64 of genes from strain NCTC11168 to determine statistically enriched functional classes. Because a similar classication was not available for strain 81-176, a table with orthologue mappings between the two strains was downloaded from OrtholugeDB65 and used to assign the NCTC11168 functional classes to their respective 81-176 counterparts. Genes in our annotation lists without an existing functional classication in NCTC11168 or without an orthologue match were assigned to class 5.I, dened as Unknown, in the original classication scheme. Genes encoded on the pVir and pTet plasmids of strain 81-176 were assigned to new pVir and pTet classes, respectively. Functional overrepresentation was analysed for each functional class via a two-sided Fishers exact test followed by multiple-testing correction using the BenjaminiHochberg method. An adjusted P-value of 0.05 was selected as signicance threshold for functional overrepresentation.

Proteinprotein coIP. The FliW and CsrA proteinprotein coIP was performed exactly as described for the RIP-seq coIP protocol (see above) until the step where beads were washed ve times with Buffer A. After washing, the beads were suspended in 200 ml of 1 protein loading buffer (62.5 mM Tris-HCl, pH 6.8,

100 mM DTT, 10% (v/v) glycerol, 2% (w/v) SDS, 0.01% (w/v) bromophenol blue) and boiled for 8 min. Lysate samples corresponding to an OD600 of 0.05 and 2 (for proteins precipitated from beads) were used for western blot analysis.

SDSPAGE and immunoblotting. Protein analyses were performed on cells collected from C. jejuni in mid-exponential phase (OD600 0.50.6) or E. coli cultures in late-exponential phase (OD600 1.01.5). Cells were collected by centrifugation at 11,000g for 3 min. Cell pellets were resuspended in 100 ml of 1 protein loading

buffer (62.5 mM Tris-HCl, pH 6.8, 100 mM DTT, 10% (v/v) glycerol, 2% (w/v) SDS, 0.01% (w/v) bromophenol blue) and boiled for 8 min. For western blot analysis, samples corresponding to an OD600 of 0.02 to 0.1 were separated by 12, 15 or 18% (v/v) SDS-polyacrylamide (PAA) gels and transferred to a nitrocellulose membrane by semidry blotting. Membranes were blocked for 1 h with 10% (w/v) milk powder/TBS-T (Tris-buffered saline-Tween-20) and incubated overnight with primary antibody at 4 C. Membranes were then washed with TBS-T, followed by 1 h incubation with secondary antibody. After washing, the blot was developed using enhanced chemiluminescence-reagent. GFP-, FLAG- and Strep-tagged proteins of interest were detected with monoclonal anti-GFP (1:1,000 in 3% BSA/TBS-T; Roche, #11814460001), monoclonal anti-FLAG (1:1,000 in 3% BSA/TBS-T; Sigma-Aldrich, #F1804-1MG) or monoclonal anti-Strep (1:10,000 in 3% BSA/TBS-T; IBA GmbH, #2-1507-001) primary antibodies and anti-mouse IgG (1:10,000 in 3% BSA/TBS-T; GE-Healthcare, #RPN4201) secondary antibody. mCherry-tagged proteins were detected using a polyclonal anti-mCherry

(1:4,000 in 3% BSA/TBS-T; Acris, #AB0040-20) primary antibody and an anti-goat (1:10,000 in 3% BSA/TBS-T; Santa Cruz Biotechnology, #sc2020) secondary antibody. A monoclonal antibody specic for GroEL (1:10,000 in 3% BSA/TBS-T; Sigma-Aldrich, # G6532-5ML) and an anti-rabbit IgG (1:10,000 in 3% BSA/TBS-T; GE-Healthcare, #RPN4301) secondary antibody were used asa loading control. Images of full blots that were cropped in main Figures are shown in Supplementary Fig. 19.

Validation of CsrA targets with a GFP reporter system. Validation of CsrA targets was performed using a heterologous E. coli system previously developed for validation of sRNAmRNA interactions23. Selected candidate C. jejuni CsrA target sequences from the coIP were cloned as translational fusions to GFP or FLAG in plasmids pXG-10 or pXG-30 as listed in Supplementary Tables 5 and 7. Levels of FLAG or GFP translational fusions were then determined by western blotting or FACS in E. coli DpgaA, DpgaA DcsrA and a DpgaA DcsrA strain harbouring plasmid pGD72-3 with C. jejuni CsrA-Strep under the control of an arabinoseinducible promoter.

Flow cytometric analysis. For FACS analysis of GFP reporter uorescence inE. coli, cells corresponding to 1 OD600 were collected from LB cultures in log phase and resuspended in 0.25 ml PBS. Cells were then xed for 10 min with 0.25 ml of 4% paraformaldehyde, collected by centrifugation and washed twice with 0.5 ml PBS before nal resuspension in 0.5 ml PBS. A 1/100 dilution of the xed sample in PBS was used for measurement. Measurements (50 000 counts per sample) were performed on a BD FACSCalibur machine and analysed using FlowJo (V10).

Purication of C. jejuni CsrA. Recombinant, C-terminal Strep-tagged C. jejuni CsrA (Cj1103) was overexpressed and puried from E. coli TOP10 DpgaA/DcsrA using Strep-Tactin Sepharose (IBA GmbH, #2-1202-001). Primers and plasmids used for cloning are listed in Supplementary Tables 4 and 5 The csrA gene, including its SD sequence, was fused to a C-terminal Strep-tag in the arabinoseinducible plasmid pBAD/Myc-His A (Invitrogen) for overexpression and afnity purication. The csrA-coding region and SD were amplied from C. jejuni NCTC11168 genomic DNA using primers CSO-0746/-0747, and the pBAD/Myc-His A plasmid was amplied by inverse PCR with JVO-0900/-0901 as previously described66. CSO-0747 and JVO-0901 introduce an XbaI site to the insert and vector, respectively, whereas CSO-0746 has a 50-phosphate to facilitate blunt-end ligation. XbaI-digested insert and vector were then ligated, resulting in pGD68-1. Plasmid pGD68-1 was checked by colony PCR using primers pBAD-FW/CSO-0747 and sequencing with pBAD-FW. A Strep-tag (WSHPQFEK) was then added at the C-terminus of csrA by inverse PCR using oligonucleotides CSO-0852/-0853, resulting in plasmid pGD72-3. Plasmid pGD72-3 was checked by sequencing with pBAD-FW. Plasmid pGD72-3 was then introduced into an E. coli TOP10 DpgaA/ DcsrA deletion strain resulting in strain CSS-0931. CSS-0931 was grown in 500 ml LB broth with 100 mg ml 1 of ampicillin at 37 C and shaking at 220 r.p.m. to an

OD600 of 0.3, at which time L-arabinose was added to a nal concentration of0.01%. The culture was then incubated for an additional 8 h at 18 C. Cells were harvested by centrifugation at 7,000g for 30 min at 4 C. The pellet was resuspended in 5 ml of Buffer W (IBA GmbH, #2-1003-100). The rest of the protocol was followed as per the manufacturers instructions using 1 ml Gravity ow Strep-Tactin Sepharose. After washing steps, the CsrA-Strep protein was nally eluted using Buffer E (IBA GmbH, #2-1000-025) in three successive steps (E1: 0.8 ml, E2: 1.4 ml and E3: 0.8 ml). The majority of CsrA-Strep was concentrated in the E2 fraction. Concentration was quantied using Roti-Quant (Carl ROTH, #K015.3), and the protein was stored at 20 C in 50 ml aliquots.

RNA isolation. Bacteria were grown to the indicated growth phase and culture volume corresponding to a total amount of 4 OD600 was harvested and mixed with0.2 volumes of stop-mix (95% ethanol and 5% phenol, vol/vol). The samples were snap-frozen in liquid nitrogen and stored at 80 C until RNA extraction. Frozen

samples were thawed on ice and centrifuged at 4 C to collect cell pellets. Cell pellets were lysed by resuspension in 600 ml of a solution containing 0.5 mg ml 1 lysozyme in TE buffer (pH 8.0) and 60 ml of 10% SDS. The samples were incubated for 12 min at 65 C to ensure lysis. Afterwards, total RNA was extracted using the hot-phenol method as described previously13,14.

Northern blot analysis. For northern blot analysis, 510 mg RNA sample was loaded per lane. After separation on 6% PAA gels containing 7 M urea, RNA was transferred to Hybond-XL membranes (GE-Healthcare) by electroblotting. After blotting, the RNA was ultraviolet cross-linked to the membrane and hybridized with g32P-ATP end-labelled DNA oligonucleotides (Supplementary Table 4).

Rifampicin RNA stability assays. To determine the stability of aA mRNA inC. jejuni NCTC11168 WT, DcsrA, DiW and DcsrA DiW strains, cells were grown to an OD600 of 0.45 (mid-log phase) and treated with rifampicin to a nal concentration 500 mg ml 1. Samples were harvested for RNA isolation at indicated time points following rifampicin addition (0, 4, 8, 16 and 32 min) as described

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 15

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

above. After RNA isolation, 10 mg of each RNA sample was used for northern blot analysis as detailed above.

In-vitro T7 transcription and RNA labelling. DNA templates containing the T7 promoter sequence were generated by PCR using oligos and DNA templates listed in Supplementary Table 8. T7 in-vitro transcription of RNAs was carried out using the MEGAscript T7 kit (Ambion) and sequences of the resulting T7 transcripts are listed in Supplementary Table 8. In vitro transcribed RNAs were quality checked and 50 end-labelled (g32P) as previously described66,67.

Gel mobility shift assays. Gel-shift assays were performed using B0.04 pmol 50-labelled RNA (4 nM nal concentration) with increasing amounts of puriedC. jejuni CsrA in 10 ml reactions. In brief, 50-radiolabelled RNA (32P, 0.04 pmol in 6 ml) was denatured (1 min, 95 C) and cooled for 5 min on ice. Yeast tRNA (1 mg) and 1 ml of 10 RNA Structure Buffer (Ambion: 10 mM Tris, pH 7, 100 mM KCl,

10 mM MgCl2) was then added to the labelled RNA. CsrA protein (2 ml diluted in 1 Structure Buffer) was added to the desired nal concentrations (0 mM, 10 nM,

20 nM, 50 nM, 100 nM, 200 nM, 500 nM, 1 mM or 2 mM CsrA). Binding reactions were incubated at 37 C for 15 min. Before loading on a pre-cooled native 6% PAA,0.5 TBE gel, samples were mixed with 3 ml native loading buffer (50% (v/v)

glycerol, 0.5 TBE, 0.2% (w/v) bromophenol blue). Gels were run in 0.5 TBE

buffer at 300 V at 4 C for 3 h. Gels were dried and analysed using a PhosphoImager (FLA-3000 Series, Fuji).

In vitro structure probing assays. In vitro structure probing of aA WT and aA M1/M2 leaders with RNase T1 and lead(II) acetate was performed as previously described68. For each reaction, 0.1 pmol of a labelled aA leader variant was denatured for 1 min at 95 C and chilled on ice for 5 min. One microgram yeast tRNA as competitor and 10 RNA Structure Buffer was added (provided together

with RNase T1, Ambion). Unlabelled recombinant C. jejuni CsrA protein was then added at 0-, 20-, 50- or 100-fold molar excess. After incubation for 15 min at 37 C, 2 ml RNase T1 (0.01 U ml 1) or 2 ml freshly prepared lead(II)-acetate solution(25 mM) were added and reactions were incubated for 3 min or 90 s, respectively.

As a control, B0.1 pmol labelled RNA with 100-fold excess CsrA was also prepared without nuclease/lead(II) treatment. The reactions were stopped by addition of 12 ml Gel loading buffer II (#AM8546G, Ambion). For RNase T1 ladders,B0.1 pmol labelled RNA was denatured in 1 Structure Buffer for 1 min at 95 C

and afterwards incubated with 0.1 U ml 1 RNase T1 for 5 min. The OH ladder was generated by incubation of B0.1 pmol labelled aA WT leader RNA in 1

alkaline hydrolysis buffer (Ambion) for 5 min at 95 C. Ladders and samples were then separated on 10% (v/v) PAA/7M urea gels in 1 TBE buffer. Gels

were dried, exposed to a screen and analysed using a PhosphorImager (FLA-3000 Series, Fuji).

Transmission electron microscopy. C. jejuni WT and mutant strains were grown for 14 h on MH plates supplemented with vancomycin (10 mg ml 1). Cells were resuspended gently in PBS using a cotton swab and centrifuged at 5,000g for 5 min.

The cell pellet was resuspended in 2% glutaraldehyde in 0.1 M cacodylate and incubated at 4 C overnight. The next day, samples were stained with 2% uranyl acetate and imaged using a Zeiss EM10 transmission electron microscope.

Motility assays. C. jejuni strains were inoculated from the appropriate selective MH agar plates into 20 ml BB containing 10 mg ml 1 vancomycin and grown microaerobically with shaking at 37 C to an OD600 of B0.5. Cells were harvested by centrifugation at 6,500g for 5 min and resuspended at an OD600 of 0.5 in BB. For each strain, 0.5 ml of bacterial suspension was inoculated into motility soft-agar plates (MH broth 0.4% agar) poured the day before. Plates were incubated right-

side-up for B24 h microaerobically at 37 C. Three measurements of each motility halo were made for each inoculation, which were averaged to give the mean swim distance for each strain on a plate. All strains were inoculated together on six replicate plates and the mean swim distancestandard error on these plates was used to the compare motility of each strain.

Autoagglutination assay. Autoagglutination was determined as described previously26. Briey, strains grown in liquid cultures for motility assays were resuspended in PBS, pH 7.4, to an OD600 of 1.0. Two millilitres were placed into three replicate tubes and the OD600 was measured. Tubes were incubated at 37 C microaerobically without shaking, and at indicated time points, 100 ml was carefully removed from the top of the suspension, diluted tenfold in PBS, and the OD600 was measured. Measurements were normalized to the optical density of each strain at the zero time point.

Time-lapse microscopy to monitor cell division. C. jejuni DiA mutant cells corresponding to an OD600 of 0.5 were collected from BB culture in log phase by centrifugation and resuspended in 0.5 ml BB. The cells were further serially diluted 100- and 1,000-fold in BB. Five microlitres of the diluted samples were spotted on a BB-agarose (1%) plate. The plate was incubated under microaerophilic conditions

at 37 C for 10 min. The agarose patch was excised and inverted onto a Petri dish with a glass bottom. Single cells were then monitored over time using several bright-eld images in a uorescence microscope (Leica DMI6000 B) maintained at 37 C under aerobic conditions.

RNA FISH. RNA-FISH was performed as previously described69 with some modications. A total amount of cells corresponding to two OD600 was collected from BB cultures in mid-log phase (OD600 0.4) and resuspended in 0.5 ml PBS.

Cells were then xed for 3 h with 0.5 ml 4% paraformaldehyde at room temperature, collected by centrifugation and washed twice with 0.5 ml PBS before nal resuspension in 0.5 ml 70% ethanol. After 10 min, cells were collected by centrifugation and resuspended in 95% ethanol and incubated at room temperature for 1 h. Cells were again collected by centrifugation, completely dried in a laminar ow hood and then washed once with 2 SSC before nal resuspension in 0.5 ml

of 2 SSC containing 10% formamide. Fluorescently labelled DNA oligos

(14 Cy5-labelled oligos to detect aA mRNA and one FITC-labelled oligo specic for 16 S rRNA, Sigma, Supplementary Table 4) were then added at a concentration of 10 ng ml 1 and incubated at 37 C overnight. The next day, cells were collected by centrifugation and washed three times for 1 h at 37 C with 0.5 ml of 2 SSC

containing 10% formamide before nal resuspension in 2 SSC (50250 ml). Cells

were then imaged in a Leica Confocal TCS SP5 II microscope using sequential scanning mode.

dSTORM. For super-resolution imaging, C. jejuni cells were grown, xed and labelled using the above-described RNA-FISH protocol (14 Cy5-labelled DNA oligonucleotides to detect aA mRNA and a FITC oligo to label 16S rRNA, Sigma, Supplementary Table 4). Labelled cells were immobilized on poly-D-lysine (Sigma-Aldrich)-coated eight-well chambered cover glasses (Sarstedt). For uorophore photo switching, a buffer with a pH of 8.38.5 was used30,70 containing 50 mM Tris-HCl (pH 8), 10% glucose, 1% 2-mercaptoethanol(Carl Roth), 3 U ml 1 pyranose oxidase (Sigma-Aldrich) and 90 U ml 1 catalase (Sigma-Aldrich) in 2 SSC.

dSTORM was performed on a wide-eld setup for localization microscopy71. An optically pumped semiconductor laser (Genesis MX STM-Series, Coherent) with a wavelength of 639 nm (maximum power of 1 W) was used for excitation of Cy5 and a diode laser (iBeam smart Family, TOPTICA Photonics) with a wavelength of 405 nm (maximum power of 120 mW) was used for reactivation of Cy5. Laser beams were cleaned-up by bandpass lters (Semrock/Chroma) and combined by appropriate dichroic mirrors (LaserMUX lters, Semrock). Afterwards they were focused onto the back focal plane of the high numerical oil-immersion objective (Olympus APON 60XO TIRF, numerical aperture 1.49), which is part of an inverted uorescence microscope (Olympus IX71). To separate the excitation light from the uorescence light, suitable dichroic beam splitters (Semrock) were placed into the light path before the laser beams enter the objective. Fluorescence light collected by the objective was ltered by appropriate detection lters (Semrock/Chroma) and was detected by an EMCCD camera with 512 512 pixels (iXon Ultra 897, Andor Technology). The pixel size in the image

was 129 nm px 1. Cy5 was excited with the 639-nm laser at a maximum intensity of 4.19 kW cm 2. During imaging, the 405-nm laser was switched on to keep up a suitable switching ratio. Its laser power was increased successively to a maximum intensity of 0.04 kW cm 2. For every image, 5,00025,000 frames were taken with an integration time of 15 ms per frame. For every imaged area, additionallya bright-eld image was taken to identify single bacteria. Data analysis was performed using rapidSTORM open source software72.

Statistical analysis. All data for western, northern blot or FISH analysis are presented as means.e.m. Statistical analysis was carried out using Students t-test. For statistical comparison of two groups, a two-tailed paired Students t-test was used. A value of Po0.05 was considered signicant and marked with an asterisk (*)

as explained in the legends. For FISH analysis, uorescence data curves from10 cells from a single image were merged as a single averaged curve after cell length normalization. The data were acquired and normalized over cell length using ImageJ and subsequently the merged average curve was generated using Microsoft Excel.

Code availability. The sliding_window_peak_calling_script for identication of CsrA-binding sites based on RIP-seq data has been deposited at Zenodo (https://zenodo.org/record/49292

Web End =https://zenodo.org/record/49292) under DOI: 10.5281/zenodo.49292 (http://dx.doi.org/10.5281/zenodo.49292

Web End =http://dx.doi.org/10.5281/zenodo.49292).

Data availability. The raw, de-multiplexed reads as well as coverage les of the RIP-seq libraries have been deposited in the NCBI Gene Expression Omnibus73 under the accession number GSE58419. The authors declare that all other data supporting the ndings of this study are available within the article and its supplementary information les, or from the corresponding author upon request.

16 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667 ARTICLE

References

1. Kartha, R. V. & Subramanian, S. Competing endogenous RNAs (ceRNAs): new entrants to the intricacies of gene regulation. Front. Genet. 5, 8 (2014).

2. Figueroa-Bossi, N., Valentini, M., Malleret, L., Fiorini, F. & Bossi, L. Caught at its own game: regulatory small RNA inactivated by an inducible transcript mimicking its target. Genes Dev. 23, 20042015 (2009).

3. Miyakoshi, M., Chao, Y. & Vogel, J. Cross talk between ABC transporter mRNAs via a target mRNA-derived sponge of the GcvB small RNA. EMBO J. 34, 14781492 (2015).

4. Miyakoshi, M., Chao, Y. & Vogel, J. Regulatory small RNAs from the 3 regions of bacterial mRNAs. Curr. Opin. Microbiol. 24C, 132139 (2015).

5. Romeo, T., Vakulskas, C. A. & Babitzke, P. Post-transcriptional regulation on a global scale: form and function of Csr/Rsm systems. Environ. Microbiol. 15, 313324 (2013).

6. Heroven, A. K., Bohme, K. & Dersch, P. The Csr/Rsm system of Yersinia and related pathogens: a post-transcriptional strategy for managing virulence. RNA Biol. 9, 379391 (2012).

7. Babitzke, P. & Romeo, T. CsrB sRNA family: sequestration of RNA-binding regulatory proteins. Curr. Opin. Microbiol. 10, 156163 (2007).

8. Duss, O. et al. Structural basis of the non-coding RNA RsmZ acting as a protein sponge. Nature 509, 588592 (2014).

9. Mukherjee, S. et al. CsrA-FliW interaction governs agellin homeostasis and a checkpoint on agellar morphogenesis in Bacillus subtilis. Mol. Microbiol. 82, 447461 (2011).

10. Fields, J. A. & Thompson, S. A. Campylobacter jejuni CsrA mediates oxidative stress responses, biolm formation, and host cell invasion. J. Bacteriol. 190, 34113416 (2008).

11. Barnard, F. M. et al. Global regulation of virulence and the stress response by CsrA in the highly adapted human gastric pathogen Helicobacter pylori. Mol. Microbiol. 51, 1532 (2004).

12. Kao, C. Y., Sheu, B. S. & Wu, J. J. CsrA regulates Helicobacter pylori J99 motility and adhesion by controlling agella formation. Helicobacter 19,443454 (2014).

13. Sharma, C. M. et al. The primary transcriptome of the major human pathogen Helicobacter pylori. Nature 464, 250255 (2010).

14. Dugar, G. et al. High-resolution transcriptome maps reveal strain-specic regulatory features of multiple Campylobacter jejuni isolates. PLoS Genet. 9, e1003495 (2013).

15. Porcelli, I., Reuter, M., Pearson, B. M., Wilhelm, T. & van Vliet, A. H. Parallel evolution of genome structure and transcriptional landscape in the Epsilonproteobacteria. BMC Genomics 14, 616 (2013).

16. Taveirne, M. E., Theriot, C. M., Livny, J. & DiRita, V. J. The complete Campylobacter jejuni transcriptome during colonization of a natural host determined by RNAseq. PLoS One 8, e73586 (2013).

17. Sittka, A. et al. Deep sequencing analysis of small noncoding RNA and mRNA targets of the global post-transcriptional regulator, Hfq. PLoS Genet. 4, e1000163 (2008).

18. Rieder, R., Reinhardt, R., Sharma, C. M. & Vogel, J. Experimental tools to identify RNA-protein interactions in Helicobacter pylori. RNA Biol. 9, 520531 (2012).

19. Martin, K. C. & Ephrussi, A. mRNA localization: gene expression in the spatial dimension. Cell 136, 719730 (2009).

20. Montero Llopis, P. et al. Spatial organization of the ow of genetic information in bacteria. Nature 466, 7781 (2010).

21. Nevo-Dinur, K., Nussbaum-Shochat, A., Ben-Yehuda, S. & Amster-Choder, O. Translation-independent localization of mRNA in E. coli. Science 331, 10811084 (2011).

22. Lertsethtakarn, P., Ottemann, K. M. & Hendrixson, D. R. Motility and chemotaxis in Campylobacter and Helicobacter. Annu. Rev. Microbiol. 65, 389410 (2011).

23. Urban, J. H. & Vogel, J. Translational control and target recognition by Escherichia coli small RNAs in vivo. Nucleic Acids Res. 35, 10181037 (2007).

24. Bailey, T. L. et al. MEME SUITE: tools for motif discovery and searching. Nucleic Acids Res. 37(Web Server issue): W202W208 (2009).

25. Dubey, A. K., Baker, C. S., Romeo, T. & Babitzke, P. RNA sequence and secondary structure participate in high-afnity CsrA-RNA interaction. RNA 11, 15791587 (2005).

26. Golden, N. J. & Acheson, D. W. Identication of motility and autoagglutination Campylobacter jejuni mutants by random transposon mutagenesis. Infect. Immun. 70, 17611771 (2002).

27. de Vries, S. P. et al. Motility defects in Campylobacter jejuni dened gene deletion mutants caused by second-site mutations. Microbiology 161, 23162327 (2015).

28. Barrero-Tobon, A. M. & Hendrixson, D. R. Identication and analysis of agellar coexpressed determinants (Feds) of Campylobacter jejuni involved in colonization. Mol. Microbiol. 84, 352369 (2012).

29. Heilemann, M. et al. Subdiffraction-resolution uorescence imaging with conventional uorescent probes. Angew. Chem. 47, 61726176 (2008).

30. Fei, J. et al. RNA biochemistry. Determination of in vivo target search kinetics of regulatory noncoding RNA. Science 347, 13711374 (2015).

31. Lawhon, S. D. et al. Global regulation by CsrA in Salmonella typhimurium. Mol. Microbiol. 48, 16331645 (2003).

32. Brencic, A. & Lory, S. Determination of the regulon and identication of novel mRNA targets of Pseudomonas aeruginosa RsmA. Mol. Microbiol. 72, 612632 (2009).

33. Edwards, A. N. et al. Circuitry linking the Csr and stringent response global regulatory systems. Mol. Microbiol. 80, 15611580 (2011).

34. Holmqvist, E. et al. Global RNA recognition patterns of post-transcriptional regulators Hfq and CsrA revealed by UV crosslinking in vivo. EMBO J. 35, 9911011 (2016).

35. Yakhnin, H. et al. CsrA represses translation of sdiA, which encodes the N-acylhomoserine-L-lactone receptor of Escherichia coli, by binding exclusively within the coding region of sdiA mRNA. J. Bacteriol. 193,61626170 (2011).

36. Barrero Tobon, A. M. & Hendrixson, D. R. Flagellar biosynthesis exerts temporal regulation of secretion of specic Campylobacter jejuni colonization and virulence determinants. Mol. Microbiol. 93, 957974 (2014).

37. Wei, B. L. et al. Positive regulation of motility and hDC expression by the RNA-binding protein CsrA of Escherichia coli. Mol. Microbiol. 40, 245256 (2001).

38. Yakhnin, A. V. et al. CsrA activates hDC expression by protecting hDC mRNA from RNase E-mediated cleavage. Mol. Microbiol 87, 851866 (2013).

39. Konkel, M. E. et al. Secretion of virulence proteins from Campylobacter jejuni is dependent on a functional agellar export apparatus. J. Bacteriol. 186, 32963303 (2004).

40. Balaban, M. & Hendrixson, D. R. Polar agellar biosynthesis and a regulator of agellar number inuence spatial parameters of cell division in Campylobacter jejuni. PLoS Pathog. 7, e1002420 (2011).

41. Duss, O., Michel, E., Diarra Dit Konte, N., Schubert, M. & Allain, F. H. Molecular basis for the wide range of afnity found in Csr/Rsm protein-RNA recognition. Nucleic Acids Res. 42, 53325346 (2014).

42. Mercante, J., Edwards, A. N., Dubey, A. K., Babitzke, P. & Romeo, T. Molecular geometry of CsrA (RsmA) binding to RNA and its implications for regulated expression. J. Mol. Biol. 392, 511528 (2009).

43. Sterzenbach, T. et al. A novel CsrA titration mechanism regulates mbrial gene expression in Salmonella typhimurium. EMBO J. 32, 28722883 (2013).

44. Jorgensen, M. G., Thomason, M. K., Havelund, J., Valentin-Hansen, P. & Storz,G. Dual function of the McaS small RNA in controlling biolm formation. Genes Dev. 27, 11321145 (2013).45. Keiler, K. C. RNA localization in bacteria. Curr. Opin. Microbiol. 14, 155159 (2011).

46. Buskila, A. A., Kannaiah, S. & Amster-Choder, O. RNA localization in bacteria. RNA Biol. 11, 10511060 (2014).

47. Singer, H. M., Erhardt, M. & Hughes, K. T. Comparative analysis of the secretion capability of early and late agellar type III secretion substrates. Mol. Microbiol. 93, 505520 (2014).

48. Karlinsey, J. E., Lonner, J., Brown, K. L. & Hughes, K. T. Translation/secretion coupling by type III secretion systems. Cell 102, 487497 (2000).

49. Sorg, J. A., Miller, N. C. & Schneewind, O. Substrate recognition of type III secretion machines--testing the RNA signal hypothesis. Cell Microbiol. 7, 12171225 (2005).

50. Buchan, J. R. mRNP granules: Assembly, function, and connections with disease. RNA Biol. 11, 10191030 (2014).

51. Yakhnin, H. et al. CsrA of Bacillus subtilis regulates translation initiation of the gene encoding the agellin protein (hag) by blocking ribosome binding.

Mol. Microbiol. 64, 16051620 (2007).

52. Sze, C. W. et al. Carbon storage regulator A (CsrA(Bb)) is a repressor of Borrelia burgdorferi agellin protein FlaB. Mol. Microbiol. 82, 851864 (2011).

53. Ouyang, Z., Zhou, J. & Norgard, M. V. CsrA (BB0184) is not involved in activation of the RpoN-RpoS regulatory pathway in Borrelia burgdorferi. Infect. Immun. 82, 15111522 (2014).

54. Anderson, P. E. & Gober, J. W. FlbT, the post-transcriptional regulator of agellin synthesis in Caulobacter crescentus, interacts with the 5 untranslated region of agellin mRNA. Mol. Microbiol. 38, 4152 (2000).

55. Mukherjee, S., Babitzke, P. & Kearns, D. B. FliW and FliS function independently to control cytoplasmic agellin levels in Bacillus subtilis.J. Bacteriol. 195, 297306 (2013).56. Snel, B., Lehmann, G., Bork, P. & Huynen, M. A. STRING: a web-server to retrieve and display the repeatedly occurring neighbourhood of a gene. Nucleic Acids Res. 28, 34423444 (2000).

57. Zere, T. R. et al. Genomic targets and features of BarA-UvrY (-SirA) signal transduction systems. PLoS One 10, e0145035 (2015).

NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 17

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11667

58. Datsenko, K. A. & Wanner, B. L. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl Acad. Sci. USA 97, 66406645 (2000).

59. Sittka, A., Sharma, C. M., Rolle, K. & Vogel, J. Deep sequencing of Salmonella RNA associated with heterologous Hfq proteins in vivo reveals small RNAs as a major target class and identies RNA processing phenotypes. RNA Biol. 6, 266275 (2009).

60. Hoffmann, S. et al. Fast mapping of short sequences with mismatches, insertions and deletions using index structures. PLoS Comput. Biol. 5, e1000502 (2009).

61. Feng, J. et al. GFOLD: a generalized fold change for ranking differentially expressed genes from RNA-seq data. Bioinformatics 28, 27822788 (2012).

62. Yao, Z., Weinberg, Z. & Ruzzo, W. L. CMnder--a covariance model based RNA motif nding algorithm. Bioinformatics 22, 445452 (2006).

63. Weinberg, Z. & Breaker, R. R. R2R--software to speed the depiction of aesthetic consensus RNA secondary structures. BMC Bioinformatics 12, 3 (2011).

64. Gundogdu, O. et al. Re-annotation and re-analysis of the Campylobacter jejuni NCTC11168 genome sequence. BMC Genomics 8, 162 (2007).

65. Whiteside, M. D., Winsor, G. L., Laird, M. R. & Brinkman, F. S. OrtholugeDB: a bacterial and archaeal orthology resource for improved comparative genomic analysis. Nucleic Acids Res. 41(Database issue): D366D376(2013).

66. Papenfort, K. et al. SigmaE-dependent small RNAs of Salmonella respond to membrane stress by accelerating global omp mRNA decay. Mol. Microbiol. 62, 16741688 (2006).

67. Sittka, A., Pfeiffer, V., Tedin, K. & Vogel, J. The RNA chaperone Hfq is essential for the virulence of Salmonella typhimurium. Mol. Microbiol. 63, 193217 (2007).

68. Sharma, C. M., Darfeuille, F., Plantinga, T. H. & Vogel, J. A small RNA regulates multiple ABC transporter mRNAs by targeting C/A-rich elements inside and upstream of ribosome-binding sites. Genes Dev. 21, 28042817 (2007).

69. Russell, J. H. & Keiler, K. C. Subcellular localization of a bacterial regulatory RNA. Proc. Natl Acad. Sci. USA 106, 1640516409 (2009).

70. Swoboda, M. et al. Enzymatic oxygen scavenging for photostability without pH drop in single-molecule experiments. ACS Nano 6, 63646369 (2012).

71. van de Linde, S. et al. Direct stochastic optical reconstruction microscopy with standard uorescent probes. Nat. Protoc. 6, 9911009 (2011).

72. Wolter, S. et al. rapidSTORM: accurate, fast open-source software for localization microscopy. Nat. Methods 9, 10401041 (2012).

73. Edgar, R., Domrachev, M. & Lash, A. E. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. 30, 207210 (2002).

74. Zuker, M. Mfold web server for nucleic acid folding and hybridization prediction. Nucleic Acids Res. 31, 34063415 (2003).

75. Hendrixson, D. R. Regulation of Flagellar Gene Expression and Assembly. Campylobacter 3rd edn (ASM, 2008).

Acknowledgements

We thank Sandy R. Pernitzsch, Erik Holmqvist, Jrg Vogel, Gisela Storz and Stan Gorski for critical comments on our manuscript and/or fruitful discussions. We thank Konrad U. Frstner for help with RNA-seq analysis, Lars Barquist for help with CMnder motif analysis, Hilde Merkert for help with electron microscopy and Belinda Aul for technical assistance. G.D. is supported by the Graduate School for Life Sciences (GSLS), Wrzburg. Research in the Sharmas laboratory is supported by the Young Investigator program at the Research Center for Infectious Diseases in Wrzburg, Germany, the Bavarian Research Network for Molecular Biosystems (BioSysNet), the Bavarian Academy of Science and Humanities and the Daimler and Benz foundation.

Author contributions

G.D., S.L.S. and C.M.S. designed the study. G.D., S.L.S., T.B., S.W., M.S. and C.M.S. performed the experiments and analysed the data. R.R. provided reagents and deep sequencing and T.B. performed the RNA-seq data analysis. G.D., S.L.S. and C.M.S. wrote the manuscript with input from all the authors.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Dugar, G. et al. The CsrA-FliW network controls polar localization of the dual-function agellin mRNA in Campylobacter jejuni. Nat. Commun. 7:11667 doi: 10.1038/ncomms11667 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

18 NATURE COMMUNICATIONS | 7:11667 | DOI: 10.1038/ncomms11667 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications