Biochimica et Biophysica Acta 1824 (2012) 359–365

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbapap

Structural characterization of the RNA chaperone Hfq from the nitrogen-fixingbacterium Herbaspirillum seropedicae SmR1

Marco Antonio Seiki Kadowaki a, Jorge Iulek c, João Alexandre Ribeiro Gonçalves Barbosa b,Fábio de Oliveira Pedrosa a, Emanuel Maltempi de Souza a, Leda Satie Chubatsu a, Rose Adele Monteiro a,Marco Aurélio Schüler de Oliveira a, Maria Berenice Reynaud Steffens a,⁎a Department of Biochemistry and Molecular Biology, Universidade Federal do Paraná, PR, Brazilb Postgraduate Programme in Genomic Sciences and Biotechnology, Universidade Católica de Brasília, Brasília, DF, Brazilc Department of Chemistry, Universidade Estadual de Ponta Grossa, PR, Brazil

⁎ Corresponding author. Tel.: +55 41 3361 1667; faxE-mail address: steffens@ufpr.br (M.B.R. Steffens).

1570-9639 © 2011 Elsevier B.V.doi:10.1016/j.bbapap.2011.11.002

Open access under the Else

a b s t r a c t

a r t i c l e i n f o

Article history:Received 21 June 2011Received in revised form 18 November 2011Accepted 21 November 2011Available online 2 December 2011

Keywords:HfqHerbaspirillum seropedicaeCrystal structureSAXS

The RNA chaperone Hfq is a homohexamer protein identified as an E. coli host factor involved in phage Qβreplication and it is an important posttranscriptional regulator of several types of RNA, affecting a plethoraof bacterial functions. Although twenty Hfq crystal structures have already been reported in the ProteinData Bank (PDB), new insights into these protein structures can still be discussed. In this work, the structureof Hfq from the β-proteobacterium Herbaspirillum seropedicae, a diazotroph associated with economically im-portant agricultural crops, was determined by X-ray crystallography and small-angle X-ray scattering (SAXS).Biochemical assays such as exclusion chromatography and RNA-binding by the electrophoretic shift assay(EMSA) confirmed that the purified protein is homogeneous and active. The crystal structure revealed a con-served Sm topology, composed of one N-terminal α-helix followed by five twisted β-strands, and a novel π-πstacking intra-subunit interaction of two histidine residues, absent in other Hfq proteins. Moreover, thecalculated ab initio envelope based on small-angle X-ray scattering (SAXS) data agreed with the Hfq crystalstructure, suggesting that the protein has the same folding structure in solution.

© 2011 Elsevier B.V. Open access under the Elsevier OA license.

1. Introduction

Hfq is an RNA chaperone involved in posttranscriptional generegulation in many bacterial species. Inactivation of the hfq gene inEscherichia coli led to a variety of phenotypes, indicating that thisprotein is involved in the expression control of several aspects ofbacterial metabolism [1–4]. Hfq has also been implied as an impor-tant virulence factor in pathogenic bacteria such as Brucella abortus,Legionella pneumophila, Pseudomonas aeruginosa, Vibrio cholerae and Yer-sinia enterocolitica [5–8] and appears to be involved in a type of RNA si-lencing mediated by sRNA (small RNA) in prokaryotes [9]. sRNAs arenoncoding RNAs, usually smaller than 300 nucleotides, involved inposttranscriptional regulation of a variety of cellular functions. Hfqregulates gene expression by several mechanisms: a) it can favorthe interaction between regulatory sRNA and its target mRNA, pro-moting either mRNA stability or degradation depending on thespecific target; b) it can change the accessibility of the target RNAto RNAses; c) Hfq binding can promote base pairing between the

: +55 41 3266 2042.

vier OA license.

regulatory sRNA and the target mRNA, leading to either repressionor activation of translation [10].

This protein is widely distributed and belongs to the Sm and Sm-like protein family, whose members are able to bind RNA and havea characteristic fold consisting of an N-terminal α-helix followed byfive β-strands [11]. Two highly conserved motifs (Sm1 and Sm2) areshared with human spliceosomal proteins [12]. The crystal structuresof twenty Hfq proteins from nine different organisms were solved byX-ray crystallography and deposited in the Protein Data Bank (PDB).All structures are homohexamers except that fromMethanotermobacterthermoautotrophicum (1MGQ), which crystallized as a homoheptamer.Of these, six models of Hfq (1KQ2 from Staphylococcus aureus, 3GIBand 3RER from E. coli, 3HSB and 3AHU from Bacillus subtilis, and 2YLCfrom Salmonella typhimurium) are complexed with RNA. Analyses ofS. aureus Hfq structure showed that bound RNA interacts with its prox-imal face and showed the role of the central pore in RNA-binding [13].In contrast, the homohexamer distal face of E. coli Hfq is involved incomplexing poly(A)-RNA [13]. The latter binding mode leads to an in-crease of RNA stability, unlike the uridine rich RNA bound by S. aureusHfq [14]. These two structures showed that Hfq can distinguish betweendifferent RNAs. TheHfq-U6 RNA complex from Salmonella thyphimurium(2YLC) also reveals a distinct RNA conformation due to a preferredrecognition of uridine-rich RNA 3′ end [15]. The recent Hfq65–AU6A–

360 M.A.S. Kadowaki et al. / Biochimica et Biophysica Acta 1824 (2012) 359–365

ADP (3RER) complex from E. coli shows a differential binding mode ofAU6A, an internal segment of DsrA sRNA. This structure revealed thatthis specific Hfq substrate interacts simultaneously on the distal andproximal face [16].

The Hfq sequences from the thermophilic archaeon Methanococcusjanaschii (2QTX) and the cyanobacteria Synechocystis sp. PCC 6803(3HFO) and Anabaena PCC 7120 (3HFN) are less conserved. The Hfqstructure from M. janaschii has the most particular features: short α-helix, small diameter, larger negatively charged distal face and a distinctstructure of the L4 loop and C-terminus compared with other bacterialHfqs [17]. In spite of these differences, the Sm fold and protein functionare conserved. Hfq proteins from the cyanobacteria Synechocystis sp.PCC 6803 and Anabaena PCC 7120 bind weakly to E. coli RNAs knownto interact with Hfq, presumably because of an unusual RNA-bindingsite and surface charge differences in the binding pocket [18]. Com-parison of the crystal structures of P. aeruginosa wild type Hfq anda mutant protein in the histidine of the conserved motif YKHAI(3INZ and 3M4G) [19] showed the importance of intermonomer hy-drogen bonds. Another Hfq structural feature is the non-conservedC-terminal. This flexible extension was reported to be not essentialfor riboregulation [20]. Recently, the importance of this region forthe interaction with longer RNA fragments as well as the first struc-ture information about this disordered region in solution by small-angle X-ray scattering (SAXS) was reported [21].

A

B 0

sRNA-protein complex

Free sRNA

Template DNA

Fig. 1. (Panel A) Gel filtration chromatography on Superose12 HR 10/30. The arrow indicaweight (square). Standard proteins (closed circles) were α-amylase (200 kDa), alcohol d(29 kDa) and cytochrome C (12.4 kDa). The Hfq MW was calculated as 54 kDa. Right inser1: purified Hfq. Proteins were Coomassie blue stained (15 μg). (Panel B) Hfq RNA binding actRNA band-shift assay was carried out as described in Material and Methods.

Thus, additional Hfq structures are necessary to provide detailsabout its chaperone mechanism as well as protein–protein interac-tions. In this work we report the crystal structure of Hfq fromH. seropedicae at 2.6 Å resolution as well as structural aspects ofthe protein in solution.

2. Materials and methods

2.1. Cloning of the H. seropedicae hfq gene

The hfq gene was amplified from H. seropedicae strain SmR1 (Z78,SmR) genomic DNA using the primers NdeI-hfq (5′-CACCGGAGCTCA-TATGAGCAAC-3′) and HindIII-hfq (5′-GGCGTCAAGCTTGCCAGGTAG-3′)which introduced restriction sites (underlined) to allow cloning. The280 bp DNA fragment was cloned into the vector pCR2.1 and thendigested with NdeI and HindIII. This DNA fragment was subcloned intopT7-7 yielding the plasmid pKADO1.

2.2. Expression, purification and characterization of the Hfq protein

Hfq was expressed in E. coli BL21 (DE3) carrying plasmid pKADO1upon induction with 1 mM isopropyl β-D-thiogalactopyranoside(IPTG) for 3 h at 37 °C. The purification procedure was based on thosedescribed [22,23] with modifications. Cells were lysed by sonication

Hfq

0.1 0.2 0.5 1.0

tes the Hfq peak elution. Left insert: calibration curve for determining Hfq molecularehydrogenase (150 kDa), bovine serum albumin (BSA) (66 kDa), carbonic anhydraset: SDS-PAGE profile of purified Hfq. Lane M: molecular weight standards (kDa). Laneivity. 50 nM DsrA sRNAwas incubated with the indicated amount (μM) of Hfq hexamer.

Table 1Data collection and refinement statistics.

PDB code 3SB2

Data statisticsSpace group P21Unit cell dimensions a=59.64, b=65.32, c=60.99 Å, β=105.5°Asymmetric unit content One homohexamerWavelength (Å) 1.54 (rotating anode)Resolution range (Å) 66.00–2.63 (2.70–2.63)No. of unique reflections 13212 (977)Redundancy 6.02 (5.78)Completeness (%) 97.2 (94.9)I/σ(I) 29.21 (6.52)Rsym (%) 4.5 (30.5)R-meas(%) 4.9 (33.5)Mosaicity (°) 0.402Wilson B factor (A2) 59.9

Refinement statisticsResolution range (Å) 23.47–2.63Rwork (%) 0.20Rfree (%) 0.28RMSD bond distances (Å) 0.002RMSD bond angles (°) 0.541Protein atoms 3024Water molecules 210Average B factor (protein) (A2) 57.1Average B factor (water) (A2) 48.6

Residues in Ramachandran plot regions (%)Most favored 85.5Additionally allowed 12.8Generously allowed 1.4Disallowed 0.3

Ramachandran statistics were calculated with PROCHEK [33].

361M.A.S. Kadowaki et al. / Biochimica et Biophysica Acta 1824 (2012) 359–365

(Ultrasonic Processor XL, Heat Systems) in buffer containing 50 mMTris–HCl pH 8.0, 1 M NaCl, 250 mM MgCl2, 5 mM EDTA and 0.1 mMPMSF. After clarification (20 min, 14,000 g, 4 °C), the protein extractwas incubated for 15 min at 80 °C. The protein precipitate was removedby centrifugation (40 min, 14,000 g, 4 °C) before adding (NH4)2SO4

(1.7 M final concentration). Protein solution was loaded onto a 1 mLButyl-Sepharose column (GE Healthcare) equilibrated with 1 M NaCl,1.5 M (NH4)2SO4 and 50 mM Tris–HCl pH 8.0. Elution was performedby decreasing the concentration of (NH4)2SO4 (1.5 to 0 M) and NaCl(1.0 to 0.1 M). Fractions containing Hfq were pooled, dialyzed againstbuffer A (50 mM Tris–HCl pH 8.0 with 100 mM NaCl) with 0.1 mMPMSF and then loaded onto a 40 mL DEAE-Sepharose column (GEHealthCare). The flow-through fraction containing Hfq was concentrat-ed by ultrafiltration (Centricon 5 K, Millipore). Protein molecular massand homogeneity were determined by gel filtration chromatographyon Superose12 HR 10/30 column (GE HealthCare) equilibrated withbuffer A.

2.3. Protein analyses

Proteins were analyzed by Tricine–SDS-PAGE [24] and CoomassieBrilliant Blue R-250 staining. Densitometric analyses were performedusing a Personal Densitometer SI (Molecular Dynamics). Protein con-centrations were determined as described [25] using bovine serumalbumin as standard. The concentration of Hfq was determined byconsidering it a hexamer. The expressed Hfq was confirmed by massspectrometry using tryptic digestion [26].

2.4. Electrophoretic mobility shift assay (EMSA)

The DNA template pUCT7DsrA was linearized with DraI prior torun-off in vitro transcription [27]. The RNA produced was analyzedby 2% agarose gel electrophoresis and the concentration was deter-mined with the Qubit Quant-iT RNA Assay Kit (Invitrogen). The RNA-binding reaction was performed as described [27]. Gels were stainedwith SYBR Gold (Invitrogen) and images registered using a Biochemibioimaging system (UVP).

2.5. Protein crystallization, data collection and structure analyses

Initial crystallization trials were performed by the sitting dropvapor method diffusion at two protein concentrations (13.0 and6.5 mgmL−1) on a Honeybee 963 robot (Genomic Solution®) using544 different conditions [Crystal Screen, Crystal Screen 2 and SaltRx(Hampton), Wizard I, Wizard II and Precipitant Synergy (Emerald Bio-systems) and PACT and JCSG (Nextal/Qiagen) kits]. For each condition,0.5 μL of protein solution was mixed with 0.5 μL of the crystallizationsolution and equilibrated against 80 μL of the latter. Plates wereplaced at 18 °C and monitored. Hfq was crystallized in fifteen condi-tions with polyethylene glycol (PEG) as a common precipitant. Crys-tals were formed in a few days and all of them diffracted, althoughonly one at 2.6 Å resolution [grown in 0.1 M PCB (sodium propio-nate, sodium cacodylate, BIS-TRIS propane) at pH 7.0 and 25% (w/v)PEG 1500, condition number 16 of PACT]. After cryosoaking withcrystallization solution containing 5% (v/v) glycerol, the crystal wasmounted in a loop and flash-frozen within a continuous stream ofgaseous nitrogen at 100 K. X-ray diffraction data were collectedwith a Rigaku UltraX-18HF diffractometer equipped with a Mar345detector using a 2 min exposure time per frame at LNLS (BrazilianSynchrotron Light Laboratory). 304 images with an oscillation angleof 1° were integrated, scaled and merged with the XDS package [28].Initial phases were calculated using the Phaser program [29] with theE. coli Hfq (PDB code 1HK9) hexamer edited by the program Chainsaw[30] as template. The model was adjusted manually with Coot [31]and submitted to refinement using the programPHENIX [32] for severalcycles. The model quality was checked using the PROCHECK program

[33] and the molprobity webserver [34]. Structure factors and coordi-nates have been deposited in the PDB with ID 3SB2.

2.6. Small-angle X-ray scattering and low resolution model building

The Hfq small-angle X-ray scattering data were collected using theSAXS2 beamline (wavelength, 1.488 Å) at LNLS using a 2D detector(MAR-165). The experiments were performed at 25 °C with a detec-tor to a sample distance of 1534.5 mm and a total exposure time foreach scattering pattern of 10 min in 5 frames of 2 min each. Two pro-tein concentrations were tested (0.90 and 4.35 mg/mL). The lowerprotein concentration was used since a good signal-to-noise ratiofrom SAXS data measurements, and a non-aggregate pattern was ob-served. Data were scaled for beamline intensity variations and com-pensated for the buffer A background scattering with the programFIT2D [35]. The pair-distance distribution function and radius of gyra-tion were calculated with the program GNOM [36] using a maximumdiameter of 80 Å. The experimental molecular weight was determinedwith the program SAXS MoW [37]. Twenty low resolution envelopemodels were generated with the program GASBOR [38] using an im-posed six-fold symmetry and averaged with the program DAMAVER[39]. The superposition of the low resolution and crystallographicmodels was performed by the program SUPERCOMB [40]. The maxi-mum distance and radius of gyration of the models were calculated bythe program CRYSOL 2.6 [41].

3. Results

3.1. Purification and RNA-binding activity of Hfq from H. seropedicae

Upon expression in E. coli BL21 (DE3), about 90% of the expressedHfq protein (79 amino acid residues and a molecular weight of8.8 kDa) was found to be soluble, and a yield of 7.4 mg Hfq per L ofculture was obtained (99% purity). Hfq identity was confirmed by

β3

C

N

90°

C

N

C

Loop4

B

90°

A

H. seropedicae (3SB2) E. coli (1HK9)

Fig. 2. The crystallographic model of H. seropedicae Hfq at 2.6 Å resolution. (A) Different Hfq crystal packing profile represented by 3SB2 and 1HK9. (B) Overall Hfq structure. Mono-mer A is colored according to secondary structure. (C) The Hfq monomer A with histidines 20 and 36 highlighted in blue. Figures were prepared with Coot [31] and Pymol [42].

362 M.A.S. Kadowaki et al. / Biochimica et Biophysica Acta 1824 (2012) 359–365

Maldi-ToF mass spectrometry of the tryptic digestion. Four peptideswith m/z ratios of 1484.464, 1567.480, 1612.540 and 1902.537 werefound to correspond to Hfq residues 5–17, 5–18, 19–32 and 33–48(coverage of 37%). The gel-filtration elution profile showed one sym-metrical peak corresponding to 54 kDa, indicating H. seropedicae Hfqas a homohexamer (Fig. 1A). The purified Hfq protein was activeshowing RNA-binding activity to E. coli DsrA sRNA (Fig. 1B).

3.2. Overall structure determination and description

The dataset indexing was initially accomplished in point group222 but with some distortion and later it became clear that the truepoint group was 2 yet for the former point group. Therefore the hex-amer was not composed of two structurally equal trimers to accountfor the extra symmetry. Molecular replacement and further refine-ment indicated that the space group was P21, with one homohexamer

Fig. 3. (A) Loop 4 comparison. The loops are grouped by number of residues and representeloop. N: N-terminal. C: C-terminal. (B) Amino acid sequence and secondary structure alignhighlighted according to red color intensity. Histidine residues are indicated by blue arrowsaeruginosa, 1KQ1 for Staphylococcus aureus, 3HSB for Bacillus subtilis, 3HFN for Anabaena PCCforMethanotermobacter thermoautotrophicum. The secondary structure alignment is shown bby ESPript [44]. (C) 2mFo-DFc Fourier map contoured at 1.0 σ around histidine residues 20

per asymmetric unity (solvent content 43.3%). The final 2.6 Å resolu-tion structure model converged to R-work and R-free values of 0.20and 0.28, respectively. The data collection and refinement statisticsare shown in Table 1.

In the crystal packing, two monomers from one hexamer inter-act with the pore region of other hexamer through its distal face(Fig. 2A), which allowed the refinement of a longer ordered C-terminus for one of them (Fig. 2B).

Hfq from H. seropedicae has a classic hexameric ring shape (majordiameter of 72 and minor diameter of 64 Å) showing, in the crystalstructure, a slightly distorted round shape (Fig. 2B). The ring thick-ness is 34 Å with the central pore having a maximum diameter of13 Å. The first eight residues at the N-terminal and the last fourteenresidues at the C-terminal are disordered.

The conserved Sm topology in each monomer comprises one N-terminal α-helix (residues 8–19) followed by five twisted β-strands

d as sticks colored by atom type. Dashed lines represent interactions that stabilize thement of the Hfq structures available in the PDB. Conserved amino acids are boxed and. Crystal structure deposition codes are 1HK9 for Escherichia coli, 1U1S for Pseudomonas7120, 3HFO for Synechocystis sp. PCC 6803, 2QTX forMethanococcus janaschii and 1MGQelow the sequence alignment. The sequences were aligned by ClustalW [43] and drawnand 36.

A

B

N49

H. seropedicae(3SB2)

N48

E. coli(1HK9)

P. aeruginosa (1U1T)

R48

N

C

N

C

T50

T49

V51

V50

S. aureus(1KQ1)

B. subtilis (3HSB)

N

S47E48

G49

K50C

M. jannaschii (2QTX)

Synechocystis sp.(3HFO)

N

C

D50

M. thermoautotrophicum(1MGQ)

K47

3 amino acid residues 4 amino acid residues

D51

S52

E53R54

N

C

N

C

N

C

S48 Q49

G50

K41

V54G55

D56R57

E58

D59

G60

E61

C

H H

L1 L2 L3 L4 L5

His36

His20

N

N48

K47

C

T49

50V

363M.A.S. Kadowaki et al. / Biochimica et Biophysica Acta 1824 (2012) 359–365

364 M.A.S. Kadowaki et al. / Biochimica et Biophysica Acta 1824 (2012) 359–365

(β1, residues 23–27; β2, residues 30–40; β3, residues 44–48; β4, res-idues 52–56; β5, residues 61–65) (Fig. 2C). The longest β-strand (2)forms an arch and passes between the β1 and β3 strands. This topol-ogy is present in all available Hfq structures. A hydrophobic coreformed by the five β-strands is capped by the α-helix (Fig. 2B). Fiveloops connect the secondary structure elements and vary in aminoacid sequence, except loop 5. This loop links the β4 and β5 strandswhose highly conserved motif YKHAI contributes to the central poreformation in the quaternary structure (Fig. 2B). On the other hand,the Asn49 residue in the variable loop 4 is in a non-conventional con-formation according to the Ramachandran plot. This conformation ismaintained by a hydrogen bond between the Asn49 amide and theVal51 carbonyl groups (Fig. 3A).

H. seropedicae Hfq crystal structure also revealed interaction be-tween the His20 and His36 residue side chains that is absent in theother available Hfq structures (Fig. 3B). These residues located inloop 1 and in the β2 strand, respectively, interact through π-π stack-ing and could further stabilize Hfq conformation (Fig. 3C).

3.3. Hfq low resolution structure by small-angle X-ray scattering

SAXS experiments were performed to characterize Hfq in solu-tion. The scattering curve profile with appropriate corrections wasobtained at low protein concentration (0.9 mg/mL) (Fig. 4A). Thecalculated radius of gyration (Rg), obtained through Guinier analysisof the scattering data, andmaximumdistance (Dmax), obtainedwiththe pair-distance function (p(r)), were 2.68±0.01 nm and 8 nm,

A

90

B

Fig. 4. Small-angle X-ray scattering curve for Hfq. (A) Experimental data (gray circles) overbution function. (B) Average ab initio low resolution envelope (gray) superimposed on the

respectively. Dmax is in agreement with the crystal structure; how-ever, Rg shows some difference due to a poor fit of the experimentalX-ray scattering curve to the predicted crystal structure curve,which leads to a discrepancy of 5.31. This discrepancy may reflectthe lack of structural information of the disordered residues at Cand N-terminuses and their orientations, as the small “hat” observedon the top of the SAXS envelope. A molecular weight of 55.1 kDa wasestimated based upon the SAXS curve relative scale method [37], inagreement with both gel filtration assay and crystal structure.

The best ab initio models were obtained with the imposition ofsix-fold symmetry constrains into the DAMMIN program calculations.Twenty models were generated with a low discrepancy indicated bythe small NSD (normalized spatial discrepancy) of 0.6. The calculatedmodel curve fitted well the experimental X-ray curve with an ob-served χ value of 1.43. Both structures (X-ray and SAXS) were super-posed, thus showing an adequate match (Fig. 4B).

4. Discussion

H. seropedicae Hfq was purified using three steps including aheat treatment which improve purification without affecting itsactivity as shown by the sRNA binding assay. High purity and yieldwere obtained allowing its crystallization. Hfq crystallized within afew days in several screening conditions. The crystal structureshowed the conserved Sm fold observed in all available Hfq struc-tures. Analyses of the structure revealed two interesting character-istics. First, two histidines exclusive to H. seropedicae Hfq were

°

layed on crystal structure predicted curve (solid line). Upper inset, the distance distri-X-ray crystal structure (red). The figure was prepared with Pymol [42].

365M.A.S. Kadowaki et al. / Biochimica et Biophysica Acta 1824 (2012) 359–365

found to interact by π-π stacking. Second, the particular nature ofloop 4, which contains three residues with one (Asn49) in a non-conventional conformation. This residue is conserved in E.coli (PDBcodes: 1HK9 and 3GIB) and P. aeruginosa (PDB code: 1U1T) Hfq and isalso in a non-conventional conformation, suggesting the importanceof such conformation for Hfq activity/stability. These proteins havehigh structural similarity including a loop 4 with three residues, incontrast to other Hfq proteins which have four residues in loop 4.The crystal packing was also analysed. Amongst the structuresavailable in the PDB, there are two different packing profiles (Fig. 2A).The first involves the stacking of hexamer faces to form layers whereasthe second involves the interaction of one hexamer pore region withthe border region of another hexamer, as in the crystals of Hfq fromH. seropedicae and P. aeruginosa. SAXS measurements confirmed Hfqas a ring shaped protein with a maximum dimension of 80 Å and Rgof 26.8±0.10 Å. Moreover, the molecular weight (55.1 kDa) estimatedby the SAXS curve is in a good agreement with a hexameric structure.

5. Conclusions

This study describes the first Hfq structure from a non-pathogenicGram negative diazotrophic bacterium of agronomic interest, and dis-cusses two different conformations of loop 4 observed in Hfq crystal-lographic structures suggesting an importance for protein function.Furthermore, two histidine residues that occur only in H. seropedicaeHfq form a π-π stacking interaction. The structural envelope in solu-tion suggests that the crystallographic folding is maintained.

Acknowledgements

We thank Dr. Richard A. Lease for providing us with the pUCT7DsrAplasmid, Mário de Oliveira Neto (IFSC-USP) for helping in SAXS dataanalysis, Valter A. de Baura, Roseli Prado, Marilza Lamour and JulietaPie for technical support and M. G. Yates for reading the manuscript.This work was supported by INCT-FBN/CNPq, PRONEX, CNPq, FAPESP,CAPES and Brazilian Synchrotron Light Laboratory (LNLS).

References

[1] H.C. Tsui, H.C.E. Leung, M.E. Winkler, Characterization of broadly pleiotrophicphenotypes caused by an Hfq insertion mutation in Escherichia coli K-12, Mol.Microbiol. 13 (1994) 35–49.

[2] M. Wachi, A. Takada, K. Nagai, Overproduction of the outer-membrane proteinsFepA and FhuE responsible for iron transport in Escherichia coli hfq::cat mutant,Biochem. Biophys. Res. Commun. 264 (1999) 525–529.

[3] E. Sonnleitner, S. Hagens, F. Rosenau, S. Wilhelm, A. Habel, K. Jäger, U. Bläsi, Re-duced virulence of a hfq mutant of Pseudomonas aeruginosa O1, Microb. Pathog.35 (2003) 217–228.

[4] C. Ansong, H. Yoon, S. Porwollik, H. Mottaz-Brewer, B.O. Petritis, N. Jaitly, J.N.Adkins, M. McClelland, F. Heffron, R.D. Smith, Global systems-level analysis ofHfq and SmpB deletion mutants in Salmonella: implications for virulence andglobal protein translation, PLoS One 4 (2009) 1–13.

[5] G.T. Robertson, R.M.J. Roop, The Brucella abortus host factor I (HF-I) protein con-tributes to stress resistance during stationary phase and is a major determinant ofvirulence in mice, Mol. Microbiol. 34 (1999) 690–700.

[6] T.L. McNealy, V. Forsbach-Birk, C. Shi, R. Marre, The Hfq homolog in Legionellapneumophila demonstrates regulation by LetA and RpoS and interacts with theglobal regulator CsrA, J. Bacteriol. 187 (2005) 1527–1532.

[7] Y. Ding, B.M. Davis, M.K. Waldor, Hfq is essencial for Vibrio cholera virulence anddowregulates sigma expression, Mol. Microbiol. 53 (2004) 345–354.

[8] H. Nakao, H. Watanabe, S. Nakayama, T. Takeda, yst gene expression in Yersiniaenterocolitica is positively regulated by a chromosomal region that is highly ho-mologous to Escherichia coli host factor 1 gene (hfq), Mol. Microbiol. 18 (2005)859–865.

[9] H. Aiba, Mechanism of RNA silencing by Hfq-binding small RNAs, Curr. Opin.Microbiol. 10 (2007) 1–6.

[10] G. Storz, J.A. Opdyke, A. Zhang, Controlling mRNA stability and translation withsmall, noncoding RNAs, Curr. Opin. Microbiol. 7 (2004) 140–144.

[11] C.J. Wilusz, J. Wilusz, Eucaryotyc Lsm proteins: lessons from bacteria, Nat. Struct.Mol. Biol. 12 (2005) 1031–1036.

[12] H. Hermann, P. Fabrizio, V.A. Raker, K. Foulaki, H. Hornig, H. Brahms, R. Luhrmann,snRNP Sm proteins share two evolutionarily conserved sequence motifs whichare involved in Sm protein–protein interactions, EMBO J. 14 (1995) 2076–2088.

[13] T.M. Link, P. Valentin-Hansen, R.G. Brennan, Structure of Escherichia coli Hfqbound to polyriboadenylate RNA, PNAS 106 (2009) 1–6.

[14] M.A. Schumacher, R.F. Pearson, T. Moller, P. Valentin-Hansen, R.G. Brennan, Struc-tures of the pleiotropic translational regulator Hfq and an Hfq-RNA complex: abacterial Sm-like protein, EMBO J. 21 (2002) 3546–3556.

[15] E. Sauer, O. Weichenrieder, Structural basis for RNA 3′-end recognition by Hfq,PNAS 108 (2011) 13065–13070.

[16] W. Wang, L. Wang, Y. Zou, J. Zhang, Q. Gong, J. Wu, Y. Shi, Cooperation of Escher-ichia coli Hfq hexamers in DsrA binding, Genes Dev. 25 (2011) 2106–2117.

[17] J.S. Nielsen, A. Bøggild, C.B.F. Andersen, G. Nielsen, A. Boysen, D.E. Brodersen, P.Valentin-Hansen, An Hfq-like protein in archaea: crystal structure and functionalcharacterization of the Sm protein fromMethanococcus jannaschii, RNA 13 (2007)2213–2223.

[18] A. Bøggild, M. Overgaard, P. Valentin-Hansen, D.E. Brodersen, Cyanobacteria con-tain a structural homologue of the Hfq protein with altered RNA-binding proper-ties, FEBS J. 276 (2009) 3904–3915.

[19] O. Moskaleva, B. Melnik, A. Gabdulkhakov, M. Garber, S. Nikonov, E. Stolboush-kina, A. Nikulin, The structures of mutant forms of Hfq from Pseudomonas aerugi-nosa reveal the importance of the conserved His57 for the protein hexamerorganization, Acta Cryst. F66 (2010) 760–764.

[20] A.S. Olsen, J. Møller-Jensen, R.G. Brennan, P. Valentin-Hansen, C-terminally trun-cated derivatives of Escherichia coli Hfq are proficient in riboregulation, J. Mol.Biol. 404 (2010) 173–182.

[21] M. Beich-Frandsen, B. Vecerek, P.V. Konarev, B. Sjöblom, K. Kloiber, H. Hämmerle,L. Rajkowitsch, A.J. Miles, G. Kontaxis, B.A. Wallace, D.I. Svergun, R. Konrat, U.Bläsi, K. Djinovic-Carugo, Structural insights into the dynamics and function ofthe C-terminus of the E. coli RNA chaperone Hfq, Nucleic Acids Res. 39 (2011)4900–4915.

[22] A. Nikulin, E. Stolboushkina, A. Perederina, I. Vassilieva, U. Blaesi, I. Moll, G.Kachalova, S. Yokoyama, D. Vassylyev, M. Garber, S. Nikonov, Structure of Pseu-domonas aeruginosa Hfq protein, Acta Cryst. D61 (2005) 141–146.

[23] I.M. Vassilieva, A.D. Nikulin, U. Blasi, I. Moll, M.B. Garber, Crystallization of Hfqprotein: a bacterial gene-expression regulator, Acta Cryst. D59 (2003) 1061–1063.

[24] H. Shägger, G. Jagow, Tricine-sodium dodecyl sulfate-polyacrylamide gel electro-phoresis for the separation of proteins in the range from 1 to 100 kDa, Anal. Bio-chem. 166 (1987) 368–379.

[25] M.M. Bradford, A rapid and sensitive method for the quantification of microgramquantities of protein utilization: the principle of protein-dye binding, Anal.Bichem. 72 (1976) 248–254.

[26] A.R. Westermeier, T. Naven, Proteomics in Practice, A Laboratory Manual of Pro-teome Analysis, Wiley-VCH, Weinheim, 2002.

[27] R.A. Lease, S.A. Woodson, Cycling of the Sm-like protein Hfq on the DsrA smallregulatory RNA, J. Mol. Biol. 344 (2004) 1211–1223.

[28] W. Kabsch, XDS, Acta Cryst. D66 (2010) 125–132.[29] A.J. McCoy, R.W. Grosse-Kunstleve, P.D. Adams, M.D. Winn, L.C. Storonia, R.J.

Reada, Phaser crystallographic software, J. Appl. Cryst. 40 (2007) 658–674.[30] N. Stain, CHAINSAW: a program for mutating pdb files used as templates in

molecular replacement, J. Appl. Cryst. 41 (2008) 641–643.[31] P. Emsley, K. Cowtan, Coot: model-building tools for molecular graphics, Acta

Cryst. D60 (2004) 2126–2132.[32] P.D. Adams, P.V. Afonine, G. Bunkoczi, V.B. Chen, I.W. Davis, N. Echols, J.J. Headd, L.

Hung, G.J. Kapral, R.W. Grosse-Kunstleve, A.J. McCoy, N.W. Moriarty, R. Oeffner, R.J.Read, D.C. Richardson, J.S. Richardson, T.C. Terwilligere, P.H. Zwarta, PHENIX: acomprehensive Python-based system for macromolecular structure solution,Acta Cryst. D66 (2010) 213–221.

[33] R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, PROCHECK: a programto check the stereochemical quality of protein structures, J. Appl. Cryst. 26 (1993)283–291.

[34] I.W. Davis, L.W. Murray, J.S. Richardson, D.C. Richardson, MOLPROBITY: structurevalidation and all-atom contact analysis for nucleic acids and their complexes,Nucl. Acids Res. 32 (2004) 615–619.

[35] A.P. Hammersley, FIT2D: An Introduction and Overview, ESRF Internal Report,ESRF97HA02T, , 1997.

[36] D.I. Svergun, Determination of the regularization parameter in indirect-transformmethods using perceptual criteria, J. Appl. Cryst. 25 (1992) 495–503.

[37] H. Fischer, M.O. Neto, H.B. Napolitano, I. Polikarpov, A.F. Craievich, Determinationof the molecular weight of proteins in solution from a single small-angle x-rayscattering measurement on a relative scale, J. Appl. Cryst. 43 (2010) 101–109.

[38] D.I. Svergun, M.V. Petoukhov, M.H.J. Koch, Determination of domain structure ofproteins from X-ray solution scattering, Biophys. J. 80 (2001) 2946–2953.

[39] V.V. Volkov, D.I. Svergun, Uniqueness of ab initio shape determination in small-angle scattering, J. Appl. Cryst. 36 (2003) 860–864.

[40] M.B. Kozin, D.I. Svergun, Automated matching of high- and low-resolution struc-tural models, J. Appl. Cryst. 34 (2001) 33–41.

[41] D.I. Svergun, C. Barberato, M.H. Koch, CRYSOL — a program to evaluate X-raysolution scattering of biological macromolecules from atomic coordinates, J.Appl. Cryst. 28 (1995) 768–773.

[42] L.L.C. Schrodinger, The PyMOL Molecular Graphics System, Version 1.3r1, , 2010.[43] J.D. Thompson, D.G. Higgins, T.J. Gibson, CLUSTAL W: improving the sensitivity of

progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice, Nucleic Acids Res. 11 (1994)4673–4680.

[44] P. Gouet, X. Robert, E. Courcel, ESPript/ENDscript: extracting and rendering se-quence and 3D information from atomic structures of proteins, Nucleic AcidsRes. 31 (2003) 3320–3323.