Stimulation of poly(A) synthesis by Escherichia coli
poly(A)polymerase I is correlated with Hfq binding to
poly(A) tails
Marc Folichon, Fre
´
de
´
ric Allemand, Philippe Re
´
gnier and Eliane Hajnsdorf
UPR CNRS 9073, conventionne
´
e avec l’Universite
´
Paris 7 – Denis Diderot, Institut de Biologie Physico-Chimique, Paris, France
Host factor I (Hfq) is an abundant protein of
Escherichia coli, which was first identified as a host
factor required for the replication of Qb bacterio-
phage [1]. It has then been established that disrup-
tion of the hfq gene causes pronounced pleiotropic
phenotypes in uninfected E. coli [2] and that in other
bacteria Hfq permits the adaptation to multiple envi-
ronmental stresses [3,4].
There is now an accumulation of data that shows
Hfq is an RNA-binding protein that is associated with
RNA replication, translation and stabilization [5–8].
In the case of phage Qb replication, Hfq acts directly
by bringing into close proximity the 3¢ terminal and
internal regions of the genomic RNA [9]. Hfq was also
reported to weaken base-pairing in stem loops of
OxyS sRNA, to mask the ribosome binding site of

Fax: +33 1 58 41 50 20
Tel: +33 1 58 41 51 26
E-mail:
(Received 16 September 2004, revised 15
November 2004, accepted 16 November
2004)
doi:10.1111/j.1742-4658.2004.04485.x
The bacterial Lsm protein, host factor I (Hfq), is an RNA chaperone
involved in many types of RNA transactions such as replication and stabil-
ity, control of small RNA activity and polyadenylation. In this latter case,
Hfq stimulates poly(A) synthesis and binds poly(A) tails that it protects
from exonucleolytic degradation. We show here, that there is a correlation
between Hfq binding to the 3¢ end of an RNA molecule and its ability to
stimulate RNA elongation catalyzed by poly(A)polymerase I. In contrast,
formation of the Hfq–RNA complex inhibits elongation of the RNA by
polynucleotide phosphorylase. We demonstrate also that Hfq binding is
not affected by the phosphorylation status of the RNA molecule and
occurs equally well at terminal or internal stretches of poly(A).
Abbreviations
Hfq, host factor I; PAP I, poly(A)polymerase; PNPase, polynucleotide phosphorylase.
454 FEBS Journal 272 (2005) 454–463 ª 2004 FEBS
ATPase activity and to affect polyadenylation of bac-
terial RNAs [25–27].
Polyadenylation is ubiquitous in all organisms but
has opposite effects on mRNA stability in prokaryotes
and eukaryotes. These two opposed functions of poly-
adenylation appear to be represented in the mitochon-
dria of different organisms [28]. In the chloroplast,
however, there is no equivalent of poly(A)polymerase
and the same enzyme, polynucleotide phosphorylase,

of poly(A) tails, but it is not known whether this
reflects a modification of the RNA substrate due to
the formation of an Hfq–RNA complex or to a direct
interaction of Hfq with PAP I. To better understand
the mechanism of activation, we have investigated
whether stimulation of PAP I activity is correlated
with the affinity of Hfq for the 3¢ end of RNA. We
used an RNA fragment corresponding to the 3¢ end of
the rpsO transcript, which was shown to be polyaden-
ylated both in vivo and in vitro [26,30,31]. For that
purpose, we have first examined Hfq binding to 3¢rpsO
RNA fragments harboring different homopolymeric
tails of 18 nucleotides or a stretch of 18 encoded
nucleotides lying downstream of the transcription ter-
minator of the polycistronic rpsO-pnp mRNA (5¢-AA
GCUGACGGCAGCAAUU). As seen in Fig. 1A, we
find that Hfq binds much more efficiently RNAs that
harbor poly(A) or poly(U) tails than RNA harboring
poly(G) or poly(C). These results are in agreement
with previous data obtained with homopolymers [22].
Comparison with tail-less RNA suggests that poly(G),
poly(C) tails and the natural stretch of 18 nucleotides
encoded downstream of the rpsO transcription termi-
nator are either not, or only inefficiently, bound by
Hfq. In order to determine which RNA substrate is
more efficiently bound by Hfq, we performed competi-
tion experiments using poly(A) and poly(U) tailed
3¢rpsO RNA. The gel-shift experiments of Fig. 1B
clearly show that Hfq exhibits a preference for poly-
adenylated molecules over RNA tailed with a poly(U)

RNA rather than its interaction with PAP I. Alternat-
ively, it is also possible that PAP I stimulation of
poly(A) synthesis is correlated with the ATPase activ-
ity of Hfq [25].
M. Folichon et al. Hfq binding to RNA stimulates elongation by PAP I
FEBS Journal 272 (2005) 454–463 ª 2004 FEBS 455
The position of the AU rich region at the 5¢ end
is not critical for Hfq binding
The idea that stimulation of poly(A) synthesis is cor-
related with the association of Hfq to the 3¢ end of
RNA prompted us to investigate whether 3¢ terminal
AU rich regions are preferentially bound by Hfq. If
true, this may indicate that a function of Hfq in vivo
could be to favor polyadenylation of RNAs harbor-
ing such 3¢ elements. It is already known that Hfq
binds the OxyS109 sRNA as well as mRNA frag-
ments such as 3¢rpsO and ompA105. Moreover, it
has been shown that appending a 3¢ terminal poly(A)
or a single stranded stretch of nucleotides to these
mRNA fragments (thus giving rise to 3¢rpsO-A
18
and
ompA117, respectively) strongly enhances Hfq affinity
(Fig. 3) [10,12]. We performed a series of gel-shift
experiments in order to generate a hierarchy of Hfq
binding efficiency to these different RNAs harboring
internal or 3¢ terminal presumptive binding sites.
Interestingly, Fig. 3 shows that Hfq exhibits roughly
the same affinity for the 3¢rpsO-A
18

18
or 3¢ N
18
tailed rpsO mRNAs. 5¢ end labeled RNA and RNA 3¢ end tailed with
various homopolymer and heteropolymer sequences were mixed with Hfq and analyzed on native gel. N
18
represents the 18 nucleotide
sequence 5¢-AAGCUGACGGCAGCAAUU. (A) Hfq binding to various RNA substrates. The different 5¢ end labeled RNAs (20 p
M) indicated at
the bottom of the graph were mixed with 20 p
M Hfq-His6 and formation of complexes was analyzed by gel-shift assay. The quantification of
the gel is shown. (B) Competition assay. The 5¢ labeled RNA indicated at the bottom of the autoradiograph (20 p
M) was incubated without
(–) (lane 1) and with 10 p
M Hfq-His6 (+) (lanes 2–7) and increasing amounts of the competitor RNA indicated at the top of the autoradio-
graph; 20 p
M (lane 3), 40 pM (lane 4), 100 pM (lane 5), 200 pM (lane 6) and 400 pM (lane 7).
Hfq binding to RNA stimulates elongation by PAP I M. Folichon et al.
456 FEBS Journal 272 (2005) 454–463 ª 2004 FEBS
AB
Fig. 2. Hfq stimulates addition of A residues
by PAP I but not of C residues. (A) Addition
of various residues by PAP I. 5¢ end labeled
p-3¢rpsO mRNA was incubated for 15 min
at 37 °C, without PAP I (–) and with the
NTP indicated at the top of the panel.
Samples were analyzed on a sequencing
6% acrylamide gel along with radioactive
DNA size markers. (B) Hfq stimulation of
PAP I elongation in the presence of ATP or

structures does not modify the affinity of Hfq for
RNA.
These data prompted us to examine whether any
RNA fragment containing AU rich single-stranded
sequences and secondary structures may be bound by
Hfq. For that purpose, we synthesized an RNA frag-
ment which corresponds to the 5¢ part of rpsO mRNA
(5¢rpsO). Examination of its secondary structure
reveals a bulge loop containing a UUUUAAAAUGU
sequence and an 11 nucleotide long linker containing 6
Us and 3 As [36]. It is interesting that this RNA frag-
ment, which is not known to interact with Hfq in the
cell, is bound by Hfq as efficiently as the ompA117
mRNA fragment which contains an Hfq binding site
implicated in translation and stability of this mRNA
(Fig. 3). These data may indicate either that structural
determinants different from the AU rich stretch of
nucleotides are required for efficient Hfq binding
in vivo or that Hfq interacts with the 5¢rpsO RNA
fragment which contains the translational operator of
the rpsO messenger [37].
The phosphate number at the 5¢ end of the RNA
does not affect Hfq binding
We have also examined whether the phosphorylation
status of the 5 ¢ end of the RNA fragment, which was
shown to affect poly(A) synthesis of RNA I, a highly
folded unstranslated regulatory RNA [38], poly(A)
dependent decay [39,40] as well as RNase E and RNase
G processing [41–43] also modulates Hfq binding. To
address this question, two forms of the (3¢rpsO-A

protein interactions account for the appearance of these
complexes. We concluded that, in contrast to the many
processes quoted above, Hfq binding does not depend
on the phosphorylation status of the 5¢ extremity
(Fig. 5A).
We also examined whether polyadenylation effic-
iency of an mRNA fragment is influenced by the phos-
phorylation state of its 5¢ extremity as previously
reported for RNA I of colE1 plasmid. As shown in
Fig. 5B, PAP I is more active on the mono- than on
the tri-phosphate RNA substrate; poly(A) tailed
p-3¢rpsO are, on average, 150 nucleotides longer after
30 min than poly(A) tailed ppp-3¢rpsO. Our result
extends to mRNA the previous data obtained with
small regulatory RNA (Fig. 5B).
Hfq inhibits poly(A) synthesis by polynucleotide
phosphorylase
Because Hfq binds to the 3¢ end of RNA harboring
a single stranded stretch of nucleotides, we have
investigated whether Hfq also affects the activity of
Fig. 4. Relative affinity of Hfq for rpsO mRNA tailed at its 3¢ end
with 18 A residues or containing an internal A
18
sequence. The 5¢
labeled RNA indicated at the bottom of the autoradiograph (10 p
M)
was incubated without (–) and with (+) 200 p
M Hfq protein. Com-
plex formation was analyzed on native gel. The arrows indicate the
position of the A

to inhibit both initiation and elongation of the tail.
Discussion
In this report, we show a correlation between Hfq
binding to the 3¢ end of an RNA molecule and its abil-
ity to stimulate PAP I which suggests that the Hfq–
RNA interaction facilitates RNA recognition by
PAP I. Moreover, we also demonstrate that Hfq binds
very efficiently to RNA harboring stretches of poly(A)
and poly(U) and that the location of this structural
feature in the molecule (i.e. whether it is internal or 3¢
terminal) does not affect the affinity of Hfq. Hfq bind-
ing does not depend upon the phosphorylation status
of the 5¢ end. Finally, we show that formation of a
Hfq–poly(A) complex which activates poly(A) synthe-
sis by PAP I, inhibits poly(A) synthesis by PNPase,
suggesting that the two enzymes interact differently
with 3¢ extremities.
Our data suggest that PAP I preferentially elongates
RNA harboring poly(A) tails bound by Hfq. It is poss-
ible that structural modifications resulting from Hfq
binding facilitate recognition of the 3¢ end or its adeny-
lation by PAP I [12]. It is worth recalling here, that
similarly, an interaction between the 5¢ extremity of an
RNA with its 3¢ end was proposed to explain why 5¢
monophosphorylated RNAs are more efficently adeny-
lated by PAP I than those harboring a triphosphoryl-
ated 5¢ end [38].
AB
Fig. 5. A monophosphate 5¢ end stimulates polyadenylation but not Hfq binding. (A) Hfq binding to mono- and tri-phosphorylated RNA sub-
strates. 5¢ end labeled p-3¢rpsO mRNA and of uniformly labeled ppp-3¢rpsO mRNA (50 p

ing when they prevent intramolecular annealing of
RNA sequences with AU rich regions preferentially
recognized by Hfq.
We have shown here and in our previous study [12]
that the Hfq–poly(A) complex, which facilitates elonga-
tion of the RNA by PAP I, prevents its recognition by
PNPase which inhibits both degradation and elongation
of the RNA by this latter enzyme. These data imply
that different structural features of RNA are recognized
by the two enzymes. In the case of PNPase, Hfq may
mask the secondary site that was proposed to facilitate
RNA recognition and processivity of the reaction [48].
Failure of Hfq to impair the processivity of the reaction
(Fig. 6) suggests that it does not compete with PNPase
already engaged in processive elongation of the RNA.
These data also suggest that PAP I does not interact
with single-stranded RNA upstream of the 3¢ end of the
molecule which is presumably masked by Hfq. In addi-
tion, one can also speculate that the synthesis of long
tails, which was attributed to PNPase, is presumably
strongly inhibited by Hfq in vivo. Finally, although our
data suggest that structural modification of RNA due
to Hfq binding, account for stimulation of PAP I medi-
ated poly(A) synthesis, the recent demonstration that
Hfq interacts with proteins such as ribosomal protein
S1 and RNA polymerase implies that Hfq also establi-
shes protein–protein interactions that may affect
physiological functions of its protein partner.
Materials and methods
Protein purification

The suspension was injected onto a Q FF column (16 ⁄ 20
Amersham Biosciences Europe GmbH, Saclay, France).
Fractions eluting at 200 mm NaCl were pooled and dialyzed
Fig. 6. Hfq inhibits the addition of A residues by PNPase. Elonga-
tion of 5¢ labeled polyadenylated 3¢rpsO mRNA by PNPase was per-
formed in the absence and in the presence of 10 n
M and 100 nM
Hfq-His6 as described in Materials and methods. Aliquots were
removed at 0.5, 2, 3 and 5 min and analyzed on an acrylamide
sequencing gel.
Hfq binding to RNA stimulates elongation by PAP I M. Folichon et al.
460 FEBS Journal 272 (2005) 454–463 ª 2004 FEBS
overnight at room temperature in Binding Buffer. The solu-
tion was applied onto a Hi-trap Heparin column (Pharma-
cia) and eluted with a linear NaCl gradient between 500 and
600 mm NaCl. The pooled fractions were dialyzed overnight
at 4 °C into 50 mm Tris ⁄ HCl (pH 7.5), 50 mm NH
4
Cl, 5%
glycerol, 1 mm EDTA (Buffer A). The sample was loaded
onto a poly(A)-column [poly(A) Sepharose 4B Pharmacia]
equilibrated with Buffer A at 4 °C [49]. The run through
was collected and reloaded. The column was washed at 4 °C
with 50 mm Tris ⁄ HCl (pH 7.5), 1 m NH
4
Cl, 5% glycerol,
1mm EDTA (Buffer B) and then transfered to room tem-
perature. Hfq was eluted with Buffer B plus 8 m urea and
dialyzed at 4 °C into Buffer A with 0.5 mm dithiothreitol.
Purified Hfq was stored at 4 °C in the same buffer contain-

the 5¢ end was performed with [
32
P]ATP[a
˜
P] and T4 poly-
nucleotide kinase, labeled RNAs were separated from un-
incorporated nucleotides through a ProbQuant G-50
Microcolumn (Amersham Biosciences).
Electrophoretic mobility shift assays
Hfq protein was incubated with 5¢ [
32
P]RNA in 20 lL buffer
containing 10 mm Tris⁄ HCl (pH 8), 1 mm EDTA, 80 mm
NaCl, 1% glycerol (v ⁄ v). Reactions were incubated at 37 °C
for 30 min and complexes were resolved by electrophoresis
through native polyacrylamide gel [13]. A PhosphoImager
and the imagequant software (Amersham Biosciences
Europe) were used to view the gel and to quantify results.
Polyadenylation in vitro
Polyadenylation by PAP I was conducted as in [26] using 5¢
end labeled RNA or uniformly labeled 3¢rpsO RNA with
purified PAP I (77 fmol). Samples were analyzed on dena-
turing 6% polyacrylamide gel. Poly(A) tail synthesis by
PNPase was conducted using 5¢ end labeled 3¢rpsO-A
18
RNA (2 pmol) in 50 lL; 50 mm Tris ⁄ HCl (pH 8), 5 mm
MgCl
2
,50mm NaCl, 0.1 mm dithiothreitol, 0.5 mgÆmL
)1

tale en Microbiologie et Maladies Infectieuses et Paras-
itaires of the Ministe
`
re de l’Education Nationale de la
Recherche et de la Technologie. M. F. is recipient of a
grant from M. N. R. T.
References
1 Franze de Fernandez MT, Hayward WS & August JT
(1972) Bacterial proteins required for replication of
phage Qb ribonucleic acid. J Biol Chem 247, 821–824.
2 Muffler A, Fischer D & Hengge-Aronis R (1996) The
RNA-binding protein HF-I, known as a host factor for
phage Qb RNA replication, is essential for rpoS transla-
tion in Escherichia coli. Genes Dev 10, 1143–1151.
3 Roop RM II, Robertson GT, Ferguson GP, Milford LE,
Winkler ME & Walker GC (2002) Seeking a niche:
putative contributions of the hfq and bacA gene pro-
ducts to the successful adaptation of the brucellae to
their intracellular home. Vet Microbiol 90, 349–363.
4 Brown L & Elliott T (1996) Efficient translation of the
RpoS sigma factor in Salmonella typhimurium requires
Host Factor I, an RNA-binding protein encoded by the
hfq gene. J Bacteriol 178, 3763–3770.
5 Sledjeski DD, Whitman C & Zhang A (2001) HFq is
necessary for regulation by the untranslated RNA
DsrA. J Bacteriol 183, 1997–2005.
6 Takada A, Wachi M & Nagai K (1999) Negative regu-
latory role of the Escherichia coli hfq gene in cell divi-
sion. Biochem Biophys Res Com 266, 579–583.
7 Wachi M, Takada A & Nagai K (1999) Overproduction

´
E, Escorcia FE & Gottesman S (2003) Coupled
degradation of a small regulatory RNA and its mRNA
targets in Escherichia coli. Gene Dev 17, 2374–2383.
15 Geissmann TA & Touati D (2004) Hfq, a new chaper-
oning role: binding to messenger RNA determines
access for small RNA regulator. EMBO J 23, 396–405.
16 Majdalani N, Chen S, Murrow J, St John K &
Gottesman S (2001) Regulation of rpoS by a novel
small RNA: the characterization of RprA. Mol Micro-
biol 39, 1382–1394.
17 Zhang A, Wassarman KM, Ortega J, Steven AC &
Storz G (2002) The Sm-like HFq protein increases OxyS
RNA interaction with target mRNAs. Mol Cell 9, 11–22.
18 Moller T, Franch T, Hojrup P, Keene D, Ba
¨
chinger
HP, Brennan RG & Valentin-Hansen P (2002) HFq: a
bacterial Sm-like protein that mediates RNA–RNA
interaction. Mol Cell 9, 23–30.
19 Muffler A, Traulsen DD, Fischer D, Lange R &
Hengge-Aronis R (1997) The RNA-binding protein
HF-1 plays a global regulatory role which is largely, but
not exclusively, due to its role in expression of the
r
S
subunit of RNA polymerase in Escherichia coli.
J Bacteriol 179, 297–300.
20 Schuppli D, Miranda G, Tsui H-CT, Winckler ME, Sogo
JM & Weber H (1997) Altered 3¢-terminal RNA structure

Escherichia coli stimulates elongation of poly(A) tails by
poly(A) polymerase I. Proc Natl Acad Sci USA 97,
1501–1505.
27 Le Derout J, Folichon M, Briani F, Deho
`
G, Re
´
gnier P
& Hajnsdorf E (2003) Hfq affects the length and the fre-
quency of short oligo(A) tails at the 3¢ end of Escheri-
chia coli rpsO mRNAs. Nucleic Acids Res 31, 4017–
4023.
28 Gagliardi D, Stepien PP, Temperley RJ, Lightowlers
RN & Chrzanowska-Lightowlers ZM (2004) Messenger
RNA stability in mitochondria: different means to an
end. Trends Genet 20, 260–267.
29 Yehudai-Resheff S, Hirsh M & Schuster G (2001) Poly-
nucleotide phosphorylase functions as both an exonu-
clease and a poly(A) polymerase in spinach
chloroplasts. Mol Cell Biol 21, 5408–5416.
30 Hajnsdorf E, Braun F, Haugel-Nielsen J & Regnier P
(1995) Polyadenylylation destabilizes the rpsO mRNA
of Escherichia coli. Proc Natl Acad Sci USA 92, 3973–
3977.
31 Haugel-Nielsen J, Hajnsdorf E & Regnier P (1996) The
rpsO mRNA of Escherichia coli is polyadenylated at
multiple sites resulting from endonucleolytic processing
and exonucleolytic degradation. EMBO J 15, 3144–
3152.
32 Valentin-Hansen P, Eriksen M & Udesen C (2004) The

phorylation. Nature 374, 180–183.
40 Hajnsdorf E & Re
´
gnier P (1999) E. coli rpsO mRNA
decay: RNase E processing at the beginning of the cod-
ing sequence stimulates poly(A)-dependent degradation
of the mRNA. J Mol Biol 286, 1033–1043.
41 Mackie GA (1998) Ribonuclease E is a 5¢-end-depend-
ent endonuclease. Nature 395, 720–723.
42 Tock MR, Walsh AP, Carroll G & McDowall KJ
(2000) The CafA protein required for the 5¢-maturation
of 16S rRNA is a 5¢-end-dependent ribonuclease that
has context-dependent broad sequence specificity. J Biol
Chem 275, 8726–8732.
43 Jiang X, Diwa A & Belasco JG (2000) Regions of
RNase E important for 5¢-end-dependent RNA cleavage
and autoregulated synthesis. J Bacteriol 182, 2468–2475.
44 Mohanty BK & Kushner SR (2000) Polynucleotide
phosphorylase functions both as a 3¢fi5¢ exonuclease
and a poly(A) polymerase in Escherichia coli. Proc Natl
Acad Sci USA 97, 11966–11971.
45 de Haseth PL & Uhlenbeck OC (1980) Interaction of
Escherichia coli host factor protein with oligoriboadeny-
lates. Biochemistry 19, 6138–6146.
46 Brescia CC, Mikulecky PJ, Feig AL & Sledjeski DD
(2003) Identification of the HFq-binding site on DsrA
RNA: HFq binds without altering DsrA secondary
structure. RNA 9, 33–43.
47 Thore S, Mayer C, Sauter C, Weeks S & Suck D (2003)
Crystal structures of the Pyrococcus abyssi Sm core and