Báo cáo khoa học: Leishmania infantum LeIF protein is an ATP-dependent RNA helicase and an eIF4A-like factor that inhibits translation in yeast docx

We show that a His-tagged, recombinant, LeIF protein of Leishmania infantum, which was puri-fied from Escherichia coli, is both an RNA-dependent ATPase and an ATP-dependent RNA helicase i

Leishmania infantum LeIF protein is an ATP-dependent

RNA helicase and an eIF4A-like factor that inhibits

translation in yeast

Mourad Barhoumi1, N K Tanner2, Josette Banroques2,3, Patrick Linder2and Ikram Guizani1

1 Laboratoire d’Epide´miologie et d’Ecologie Parasitaire, Institut Pasteur de Tunis, Tunisia

2 De´partement de Microbiologie et Me´dicine Mole´culaire, Centre Me´dical Universitaire, Gene`ve, Switzerland

3 Centre de Ge´ne´tique Mole´culaire, CNRS, Gif-sur-Yvette, France

The leishmaniases constitute a group of diverse,

world-wide-distributed, parasitic diseases caused by

proto-zoan parasites of the genus Leishmania that are

transmitted by female sandflies Leishmania are

Tryp-anosomatidae protozoans having two main stages in

their life cycle: intracellular amastigotes in the

macro-phage of mammalian host and motile promastigotes

in the sandfly midgut [1] At least 20 species of

Leish-mania are pathogenic to humans Leishmaniases range

from mild, often self-healing, cutaneous lesions to

mucocutaneous, severely mutilating lesions, to fatal

vis-ceral leishmaniasis The clinical outcome of leishmanial

infections depends on a complex interplay involving the host, vector, parasite and environmental determi-nants The annual incidence is two million cases in 88 countries The mainstay therapy is based on the use of pentavalent antimonials; no efficient vaccine is yet available [2]

A number of Leishmania antigens have been cloned and characterized with respect to the immune responses elicited during experimental murine or nat-ural human infections [3–13] Among these antigens, LeIF was described originally as an antigen that indu-ces an IL12-mediated Th1 response in the peripheral

Keywords

ATPase; DEAD box; eIF4AIII; Leishmaniasis;

unwindase

Correspondance

I Guizani, Laboratoire d’Epide´miologie et

d’Ecologie Parasitaire, Institut Pasteur de

Tunis, 13 Place Pasteur, BP74, 1002 Tunis,

Tunisia

Fax: +216 71 791 833

Tel: +216 71 844 171

E-mail: ikram.guizani@pasteur.rns.tn

(Received 7 July 2006, revised 15 September

2006, accepted 18 September 2006)

doi:10.1111/j.1742-4658.2006.05506.x

LeIF, a Leishmania protein similar to the eukaryotic initiation factor eIF4A, which is a prototype of the DEAD box protein family, was origin-ally described as a Th1-type natural adjuvant and as an antigen that indu-ces an IL12-mediated Th1 response in the peripheral blood mononuclear cells of leishmaniasis patients This study aims to characterize this protein

by comparative biochemical and genetic analysis with eIF4A in order to assess its potential as a target for drug development We show that a His-tagged, recombinant, LeIF protein of Leishmania infantum, which was puri-fied from Escherichia coli, is both an RNA-dependent ATPase and an ATP-dependent RNA helicase in vitro, as described previously for other members of the DEAD box helicase protein family In vivo experiments show that the LeIF gene cannot complement the deletion of the essential TIF1 and TIF2 genes in the yeast Saccharomyces cerevisiae that encode eIF4A In contrast, expression of LeIF inhibits yeast growth when endog-enous eIF4A is expressed off only one of its two encoding genes Further-more, in vitro binding assays show that LeIF interacts with yeast eIF4G These results show an unproductive interaction of LeIF with translation initiation factors in yeast Furthermore, the 25 amino terminal residues were shown to enhance the ability of LeIF to interfere with the translation machinery in yeast

Abbreviations

eIF, eukaryotic initiation factor; EJC, exon junction complex; 5-FOA, 5-fluoro-orotic acid; GST, glutathione S-transferase; PABP, polyA-binding protein; PBMC, peripheral blood mononuclear cells; SD, synthetic dextrose; SF2, superfamily 2.

blood mononuclear cells (PBMC) of leishmaniasis

patients, which also acts as a Th1-type natural

adju-vant [8,10,11,14] Its importance in host–parasite

inter-actions is not clear yet; several studies have highlighted

its immunomodulatory properties on cells of healthy

donors [8] Along with two other antigens, stress

indu-cible protein 1 (ST11) and thiol-specific antioxidant

(TSA), LeIF is part of a trifusion recombinant protein

vaccine, leish-111f, which proved efficient in

signifi-cantly reducing the parasite load and size of the lesion

in mice and in primate models [15] These recombinant

proteins, when administered as a cocktail, were

effi-cient for immunotherapy [16] Immunomodulatory

activity leading to production of IL12 is thought to

occur via a yet unknown receptor [14], as supported

by the existence of a polarity in the molecule with

respect to the levels of cytokine induced; the 226

amino terminal residues are sufficient for this activity

[8,11,14] LeIF protein contains 403 residues and it

shows high sequence similarity to the mammalian

translation initiation factor eIF4A and to other

homo-logues in lower and higher eukaryotes It is expressed

both in the promastigote and amastigote parasite

forms of all the different Leishmania species tested [8]

Its role in the biology of the parasite is unknown

In silicopredictions and expression levels seem to

indi-cate an involvement in the translation process [17],

although recent alignments of the LeIF protein from

Leishmania braziliensis and Leishmania major with

eIF4A from other organisms show some divergence

[18] The purpose of this work is to characterize the

LeIF protein by a comparative biochemical and

gen-etic analysis with its apparent homologue in yeast,

eIF4A, in order to assess its potential as a target for

drug development

The eIF4A-like proteins are the archetype of the

DEAD box family of proteins [19] The DEAD box

helicases belong to superfamily 2 (SF2) in the

classifi-cation of Gorbalenya and Koonin [20] All members

of the DEAD box family share nine conserved amino

acid motifs [21–24], including the sequence

Asp-Glu-Ala-Asp (D-E-A-D) that inspired their name Members

of the DEAD box family are found in a wide range of

organisms, including bacteria and eukaryotes ranging

from yeast to humans, and they are implicated in

vir-tually every cellular process involving RNA These

include transcription, ribosomal biogenesis, pre-mRNA

splicing, RNA export, translation, and RNA

degrada-tion [25–27] In vitro analyses of purified proteins show

an RNA-dependent ATPase activity and in some cases

ATP-dependent unwinding activity [28–31] The solved

crystal structures of various DEAD box proteins,

including yeast eIF4A, show a core structure that

consists of two RecA-like domains connected by a flexible linker [21,32–34] The tertiary structure of this core can be largely superimposed on the solved crystal structures of other SF1 and SF2 helicases, which suggests a common mechanistic theme among these helicases [21,34] The eIF4A-like helicases are close to the minimal size constituting the core structure alone [21,24,34]

Translation initiation in eukaryotes involves a series

of steps that result in the recruitment of a transla-tion-competent 80S ribosome to the initiation codon

of an mRNA The process is catalyzed by a large number of eukaryotic initiation factors (eIFs) Among these factors, eIF4A is part of the translation initi-ation complex eIF4F that binds to the cap structure

of mRNAs, in conjunction with eIF4E and eIF4G, to promote the binding of the 40S ribosomal subunit to the mRNA and the subsequent scanning for the initi-ation AUG codon [35,36] eIF4A has been proposed

to facilitate the ‘melting’ of secondary structures in the 5¢ untranslated region of the mRNA during the scanning process [35–38] Translation initiation in trypanosomatidae protozoans is not well character-ized; translation factors were identified according to their sequence similarities to known factors in other organisms Among these factors, polyA-binding pro-tein (PABP) from Trypanosoma cruzi, Trypanosoma brucei and L major have been identified [39–41] The eIF4F components of L major have been predicted [17], and the analysis of the eIF4E component of the eIF4F complex has been initiated [42] However, little

is known regarding the role of these factors in trans-lation

In this work we studied the biochemical properties

of purified, recombinant, LeIF protein from Leishma-nia infantum, and we demonstrate that it is an RNA-dependent ATPase and an ATP-dependent RNA helicase Sequence alignments show that LeIF is closely related to known eIF4A factors, but its closest homo-logue in humans is DDX48, also known as eIF4AIII, which plays a role in nonsense-mediated mRNA decay and nuclear mRNA splicing [43–46] Genetic studies in the yeast Saccharomyces cerevisiae provided evidence that LeIF can impair cell growth and can associate with yeast proteins involved in translation initiation, although it is not able to complement the deletion of the yeast-encoded eIF4A Finally, in vitro coimmuno-precipitation experiments show that LeIF interacts with the yeast translation initiation factor eIF4G Our results also point to the importance of the 25 amino terminal residues in enhancing the ability of the pro-tein to interfere with the translation machinery of yeast All this confirms an unproductive interaction of

LeIF with translation initiation factors in yeast and

interest for it as a potential drug target

Results

Sequence analysis

LeIF protein of L infantum has 98% and 100%

iden-tity with LeIF proteins of L braziliensis and L major,

respectively [11] The alignment shown in Fig 1 and

summarized in Table 1 compares the LeIF protein of

L infantum with eIF4A-like proteins from humans,

mouse and yeast IF41 and IF42 are identical between

mice and humans while DDX48 shows only three

differences IF41 and IF42, also called eIF4AI and eIF4AII, are known translation initiation factors in mammalians, as is eIF4A in yeast [35] IF42 is func-tionally equivalent to IF41 but its tissue-specific expression and developmental regulation is somewhat different DDX48 is involved in splicing and nonsense-mediated mRNA decay [44–46] It cannot substitute for IF41 in ribosome binding assays, it inhibits transla-tion in vitro in reconstitutransla-tion experiments, and its affinity for eIF4G is somewhat different from that of eIF4AI [47] Fal1 (for eIF4A-Like) is a nucleolar pro-tein involved in ribosomal biogenesis [48] It has 56% identity with yeast eIF4A, and it cannot substitute for eIF4A in vivo

Fig 1 Sequence comparison of L infantum LeIF with eIF4A homologues CLUSTALW alignment shows the comparison of the predicted amino acid sequences of L infantum eIF4A (LieIF), with the eIF4A-like proteins from human (Hu), and yeast (Sc) The mouse equivalents are essentially the same as the human Conserved motifs found in RNA helicases, are as indicated in light blue (Q, I–VI) Identical amino acids shared between the proteins are shown in magenta and green Asterisk indicates fully conserved residues; colon means that substitu-tions are conserved; period means that substitusubstitu-tions are semiconserved.

The mammalian protein with the closest similarity

to LeIF is DDX48, although the predicted pKi of

LeIF is intermediate between the IF proteins and

DDX48 The differences on the sequence level seem to

be randomly distributed on the carboxyl terminal

RecA-like domain (domain 2) while they tend to be

more clustered in the amino terminal domain (domain

1) In particular, the most notable differences are seen

in the sequence upstream of the isolated, highly

con-served phenylalanine of the recently identified Q motif

[49] and between motifs I and II The LeIF protein

has all the conserved motifs characteristic of DEAD

box helicase (motifs Q, I, Ia, Ib, II, III, IV, V, and VI)

that are known to be important for ATP binding and

hydrolysis, for RNA binding and for RNA unwinding

This prompted us to characterize its biochemical

activ-ities and compare them to yeast eIF4A

LeIF protein has an RNA-dependent ATPase

activity

We subcloned the LeIF gene into a pET22b plasmid

containing a carboxyl terminal His6 tag, expressed the

protein in the Origami Escherichia coli strain and

puri-fied the soluble protein by nickel-nitrilotriacetic acid

agarose chromatography (Fig 2) We estimated the

protein to be greater than 90% pure after this column

We also cloned, expressed and purified a mutant in

motif I (K76A) as a control; a similar mutation in

eIF4A disrupts ATP binding and ATPase activity

[49,50] The identity of the proteins was verified using

antibodies raised against His and LeIF (data not

shown)

The purified recombinant proteins were used in

ATP-ase assays that measured the free phosphate releATP-ased,

in the presence of commercially available total yeast

RNA, with a colorimetric assay based on

molybdate-Malachite Green [49,51] The optimal reaction condi-tions were determined for the wild-type LeIF and yeast eIF4A proteins LeIF showed a sharp peak around

pH 6.0, and there was little activity at pH 5.0 or below and a gradual decrease at pH 6.5 and above A similar profile was obtained for eIF4A Likewise, both LeIF and eIF4A were more active with acetate ions than chloride, with a peak activity around 10–20 mm The divalent cation optimum was 1–5 mm for LeIF and 1–

2 mm for eIF4A The ATPase activity for both pro-teins was saturated at the RNA concentration typically used (500 ngÆlL)1), but LeIF showed saturation at a

Table 1 Protein characteristics and sequence homology to LeIF The L infantum LeIF protein sequence was used to find similar proteins using BLAST2.0 on the EMBnet web site (http://www.ch.embnet.org) using the SwissProt and TrEMBL databases and the default settings CLUSTALW analyses were also carried out on the EMBnet site with the default setting All values are relative to LeIF Molecular mass (m) and

pKiwere calculated through ExPASy web site (http://www.expasy.org/) Hu, human; Mus, mouse; Sc, S cerevisiae %Similarity includes conserved and semiconserved substitutions E value, a measure of the expected random matches.

IF42_Mus (P10630)

IF41_Mus (P60843)

124 K 80.0K 49.0K 34.8K 28.9K 20.6K

209 K

MW GST-elF4G LeIF Δ 25LeIF elF4A

Fig 2 Expression and purification of the proteins used Aliquots of purified His6-LeIF, His6-D25LeIF, His6-eIF4A and GST-eIF4G protein were resolved by SDS polyacrylamide gel and stained with Coo-massie brilliant blue The positions of the Bio-Rad prestained mark-ers (in kDa) are indicated at the left The K76A mutant of LeIF had purity similar to LeIF (not shown).

lower concentration (around 100–200 ngÆlL)1 RNA)

than eIF4A, which suggested a higher affinity for

RNA This was consistent with electrophoretic

mobility shift assays (EMSA) that indicated LeIF had

roughly a two-fold higher affinity (data not shown)

ATPase activity was directly proportional to the

enzyme concentration for both proteins, which showed

that they were probably functional as monomers As

expected the LeIF mutant with a substitution in motif

I (K76A) showed no significant ATPase activity The

amount of ATP hydrolyzed for LeIF and eIF4A

increased in a time-dependent manner in the presence

of saturating concentrations of total yeast RNA

(Fig 3A) Thus, the LeIF protein exhibited an

RNA-dependent ATPase activity that is characteristic of

pro-teins from the DEAD box family

The nucleotide specificity of LeIF protein was assessed using different NTPs and dNTPs Both ATP and dATP were efficiently hydrolyzed in the presence

of RNA, as was found for eIF4A [49] The other NTPs and dNTPs had no effect As shown in Fig 3, the Michaelis–Menten parameters were determined with variable concentrations of ATP at saturating con-centrations of RNA We determined the Km for ATP

of LeIF to be 350 ± 120 lm, the kcatwas 72 ± 9 s)1 and the kcat⁄ Km was 0.21 ± 0.7 s)1Ælm)1 LeIF was inhibited by ADP, which had a binding affinity similar

to that for ATP We also determined the kinetic parameters for eIF4A However, ADP binds eIF4A with a higher affinity than ATP [49], which made our measurements less reliable, especially at higher ATP concentrations Nevertheless, the values were in the same range as those for LeIF with a Km of

250 ± 90 lm, a kcat of 39 ± 7 s)1 and a kcat⁄ Km of 0.16 ± 0.06 s)1Ælm)1 These values of eIF4A are sim-ilar to those obtained by other workers [30,52]

LeIF protein has an ATP-dependent RNA helicase activity in vitro

To test whether LeIF has an RNA unwinding activity

in vitro, we constructed two RNA⁄ DNA

heteroduplex-es containing 44 or 45 nucleotide long RNAs and a 16 nucleotide long DNA that could hybridize on either the 5¢ or 3¢ end of the RNAs (Fig 4A) It was previ-ously shown that RNA⁄ DNA duplexes are substrates for RNA helicases as long as the single-stranded region is RNA; it functions as the initial binding site for the proteins [21,51,53] As shown in Fig 4B,C, LeIF and eIF4A were able to unwind both the 5¢ and the 3¢ duplexes when they were in 20-fold excess of the substrate There was significant unwinding in the absence of ATP, which probably reflected the intrinsic affinity of the protein for the RNA at these high pro-tein concentrations However, there was approximately 30% more unwinding activity in the presence of ATP This relatively poor ATP-dependent helicase activity of eIF4A proteins has been noted previously [53] The 5¢ duplex was unwound more efficiently than the 3¢ duplex with both proteins, but we do not consider this evidence for directionality Rather, this probably reflects the intrinsic properties of the duplexes themselves Although the same oligonucleotide was hybridized on both RNAs (calculated DG ¼ –19.8 kcalÆmol)1 under standard conditions), the 5¢ duplex had a slightly lower Tm, which probably resulted because the 5¢ duplex RNA (K06) could form a moderately stable (calculated DG ¼ –4.5 kcalÆmol)1) intramolecular hairpin that could compete for the

A

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

[ATP] m M

B

0

10

20

30

40

50

Time (min)

Fig 3 Kinetic measurements of the ATPase activity of LeIF (A) An

example of a time course for the ATPase activity of 540 n M LeIF

with 0 n M (n), 50 n M (s), 100 n M (+), 400 n M (m), 1 m M (r), or

3 m M (n) ATP in the presence of 500 ngÆlL)1RNA The control

con-sisted of 3 m M ATP and no RNA (h) The K76A mutant control of

LeIF showed ATPase activity comparable to the control (not

shown) (B) Michaelis–Menten plot of the medium values of three

independent experiments.

oligonucleotide binding site (NK Tanner, unpublished

data) Thus, it is important to incorporate the

proper-ties of the substrates when interpreting the unwinding

activity of the helicases

LeIF cannot complement the deletion of eIF4A

Our biochemical analyses showed that LeIF had very

similar properties to yeast eIF4A However, this

provi-ded only circumstantial evidence that LeIF is a

transla-tion initiatransla-tion factor Consequently, we used genetic

studies in the yeast S cerevisiae to understand the potential role of LeIF in the translation initiation pro-cess In order to test whether the LeIF gene can com-plement the deletion of the essential TIF1 and TIF2 genes in yeast, which encode eIF4A, we subcloned the LeIF gene into both low and high copy number yeast plasmids, p415-PL-ADH and p424-PL-ADH, respect-ively, containing strong, constitutively expressed, ADH promoters [49] We also cloned the LeIF gene in an equivalent plasmid containing a galactose-inducible promoter p424-PL-GAL As a control, the yeast eIF4A gene was cloned into the same vectors

The various constructs were transformed into the yeast strain SS13-3A, where both chromosomal copies

of the essential eIF4A genes were deleted and eIF4A was expressed off the YCplac33-TIF1 (CEN-URA3) plasmid [49] Because this plasmid contained a URA3 marker we could selectively eliminate it from trans-formed cells by plating them on 5-fluoro-orotic acid (5-FOA)-containing medium Thus, the protein enco-ded by the transforming plasmid could ensure growth

of the yeast only if it had the ability to complement for the missing function Protein expression was veri-fied by western blot analysis of cell extracts separated

on 12% SDS Laemmli gels and revealed with anti-HA IgG (data not shown) None of the LeIF-containing plasmids were able to support yeast growth at any temperature tested (18C, 30 C, and 36 C) These data showed that LeIF could not substitute for the yeast eIF4A Likewise, purified LeIF protein did not support translation in an in vitro reconstitution assay using rabbit reticulocytes (M Altmann, University of Bern, Switzerland, unpublished data)

In experiments similar to those previously described,

we also transformed a yeast strain deleted for the FAL1 gene, encoding the Fal1 protein, which has a clearly different function from eIF4A, with the various LeIF constructs None of them supported growth on 5-FOA-containing medium (data not shown) Thus, LeIF cannot substitute for the Fal1 protein

LeIF protein inhibits cells growth While our in vivo complementation assays failed to reveal a role for the LeIF protein, we did notice that cells expressing the protein were less vigorous after transformation It is possible that LeIF was interfering with the cellular machinery by interacting with, and sequestering, yeast factors involved in translation We tested this by transforming the various LeIF constructs into the yeast SS3 strain that has the TIF2 gene replaced by a cassette carrying the URA3 gene and a second TIF1 gene under the control of the

CYC1-C

0

20

40

60

80

100

15 30 60

0 5

15 30 60

0 5

15 30 60

0 5

15 30 60

0 5

no ATP

Time(min)

LeIF eIF4A

LeIF eIF4A

3 Duplex

5 Duplex

A

Duplex

Olgo

0 5 15 30 60

LeIF

0 5 15 30 60

eIF4A

0 5 15 30 60 15 30 60 LeIF

0 5 eIF4A

5 Duplex

3 Duplex

B

Fig 4 Unwinding activity of LeIF (A) The same 5¢ [ 32 P] end-labeled

DNA oligonucleotide was hybridized to two RNA transcripts that

yielded 3¢ and 5¢ duplexes (B) Time course for ATP-dependent

unwinding of 3¢ and 5¢ duplexes by LeIF protein Briefly, 50 n M of

duplex were incubated with 1 l M protein with or without 1 m M

ATP, at 30 C, for the times indicated in minutes To prevent

rean-nealing of the displaced [ 32 P]-labeled oligonucleotide, 1 l M cold

DNA oligonucleotide was added as a competitor Products were

separated on a 15% polyacrylamide gel, which was then subject to

autoradiography and quantification (C) Comparison of the relative

helicase activities of LeIF to yeast eIF4A.

GAL promoter Because the expression of the TIF1

gene under its own promoter is several-folds lower

than that of the TIF2 gene [54], this strain produces

less eIF4A protein on glucose-containing medium than

a normal strain, but it can be induced for higher

eIF4A production on galactose-containing medium

This strain was previously used to see

dominant-negat-ive phenotypes of eIF4A mutations [54] Cells

expres-sing the full-length LeIF showed strongly reduced

growth on glucose-containing medium compared to

the cells transformed with the vector alone or with the

plasmid carrying the TIF1 gene (Fig 5) The difference

in growth however, was not observed on

galactose-containing medium (data not shown) Cells

constitu-tively expressing yeast eIF4A also showed slightly

reduced growth relative to cells with the plasmid alone,

but not nearly as strongly as with LeIF (Fig 5) This

presumably reflected the altered stoichiometry of the

translation initiation factors that caused inefficient

assembly of the initiation complex

The first 25 amino terminal residues interfere

with translation machinery in yeast

In order to identify the part of LeIF protein that is

implicated in this inhibition, we cloned a construct of

LeIFthat was missing the first 25 amino terminal

resi-dues (D25LeIF) This construct was made because the

amino termini showed the most differences between

proteins (Table 1) and because a similar construct of

yeast eIF4A could complement growth (NK Tanner, unpublished data)

As shown in Fig 5, expression of the D25LeIF pro-tein showed the same growth profile as overexpression

of eIF4A Thus these amino terminal residues enhanced the ability of LeIF to interfere with the cellu-lar machinery

We verified this result by measuring the doubling time of cells expressing the various constructs in liquid culture containing glucose Cells were grown at 30C with continuous shaking in minimal medium lacking tryptophan [synthetic dextrose (SD)-Trp] The absence

of revertants or loss of plasmids was verified at the end of the incubation by streaking culture aliquots on SD-Trp plates Three independent cultures were made for full-length LeIF and two independent cultures were made for the other constructs The cells expressing full-length LeIF grew about 50% less rapidly than the cells transformed with the plasmid alone, with a doub-ling time of 5.0 h versus 2.5 h, respectively Overex-pression of eIF4A showed a slight inhibitory effect (3.0 h) as did the D25LeIF (3.3 h) To rule out the possibility that the deletion of the amino terminus affected the expression or stability of the protein, total cellular proteins were extracted from exponentially growing cells (D600¼ 0.8), separated on an SDS Laemmli gel, transferred to nitrocellulose membrane and analyzed by a western blot analysis using anti-HA and anti-LeIF IgG The results showed that the recom-binant HA-tagged D25LeIF protein had a stable expression comparable to the HA-tagged LeIF protein (data not shown)

Interaction between LeIF protein and GST-eIF4G

in vitro The dominant-negative phenotype that we observed with the LeIF protein suggested that it was capable of interacting nonproductively with the yeast translation initiation factors, which resulted in translation inhibi-tion However, a more trivial explanation was that expression of the LeIF protein had a general toxic effect on the cells that was unrelated to translation per

se This possibility was unlikely because increased expression of yeast eIF4A largely reversed the effect Nevertheless, we decided to verify that LeIF could interact with components of the eIF4F complex Previ-ous studies in yeast have shown that the 542–883 frag-ment of eIF4G interacts with eIF4A in vitro [55] We purified this fragment, which was expressed in E coli

as a glutathione S-transferase (GST) fusion protein (Fig 2) and used it to generate a glutathione-sepharose affinity column The LeIF recombinant proteins

[wild-LeIF

25LeIF

eIF4A

Control

1 10 100 1000 10 000 100 000

Fig 5 Dominant-negative phenotype of the LeIF gene Yeast SS3

cells were transformed with the plasmids containing the LeIF gene,

the D25LeIF gene, yeast TIF1 (eIF4A) or the p424-PL plasmid alone

(control) Cells were grown in liquid minimum (SD) medium lacking

tryptophan to the same density, serially diluted and 5 lL of each

dilution was spotted onto SD minus Trp-containing plates The

plates were incubated at 30 C The numbers refer to the amount

of dilution.

type (wt) and D25LeIF] were then loaded onto

col-umns with the bound eIF4G and washed The retained

proteins were eluted with reduced glutathione,

separ-ated on an SDS Laemmli gel, transferred to a

nitrocel-lulose membrane and then subjected to western blot

analysis using anti-GST, anti-LeIF and anti-His-tag

IgGs As a control, we carried out the same

experi-ment with recombinant yeast eIF4A

The results showed that recombinant LeIF and

D25LeIF were capable of binding to the column with

the yeast GST-eIF4G fusion, but not to GST alone

(Fig 6) Similarly, the yeast eIF4A was retained on

the GST-eIF4G column Interestingly, a minor

degra-dation product of LeIF was preferentially retained on

the column by the GST-eIF4G in some experiments,

probably as a result of protease cleavage while bound

to the matrix The visible contaminants on the

Coo-massie blue-stained gel were extracted and sequenced

with a MALDI-TOF mass spectrometer; the 23 kDa

fragment corresponds to the carboxyl terminal region

consisting of domain 2 and residues just amino

ter-minal to motif III Although previous studies showed

it is the amino terminal domain of eIF4A that binds

to eIF4G [56], recent NMR studies indicate that,

although both domains 1 and 2 interact with the

middle domain of eIF4G, it is the carboxy terminal domain 2 that forms the main interactions [52] The result that the LeIF carboxyl terminal domain was selectively retained in some experiments would imply that the LeIF interactions with eIF4G are similar to those of eIF4A Regardless, these results show that LeIF protein can interact with yeast eIF4G in vitro, and they suggest that a similar interaction could occur

in vivo

We used the purified eIF4G to determine whether it would enhance the ATPase activity of LeIF as previ-ously observed with eIF4A [56] The eIF4G elution buffer, probably the glutathione, was strongly inhibi-tory in the ATPase assay, and the eIF4G required extensive dialysis against the binding buffer We found

up to a 50% enhancement of the ATPase of eIF4A and a smaller enhancement with LeIF However, the primary effect of eIF4G is to enhance the affinity of eIF4A for the RNA [56], and our conditions may not have been optimized to see this More extensive kinetic analyses are needed, but these preliminary experiments show a small eIF4G-dependent enhancement of the ATPase activity of both eIF4A and LeIF

Discussion

The antigenic properties of Leishmania LeIF protein are well characterized Indeed all studies highlight the peculiar and unique characteristics of this protein that lead researchers to consider it as a Th1-type natural adjuvant and as an immunotherapeutic molecule against intracellular pathogens However, little is known about its biological role The sequence homol-ogy with eIF4A implies a role as a translation initi-ation factor [10,11] although other sequence analysis shows a more distant relationship [18] The Leishmania genome encodes for two genes annotated as eIF4A (http://www.genedb.org/) These identical isoforms, borne by chromosome 1, are identified in L infantum

as LinEIF4A1 (LinJ01.0780 and LinJ01.0790), which encode for LeIF protein Another gene, LinEIF4A2 (LinJ28.1600) on chromosome 28, encodes for a sim-ilar protein that has only 49% identity with LeIF, and

it is predicted to be 14 amino acids shorter This work was undertaken to characterize the biochemical prop-erties of the LeIF protein and to compare its bio-chemical and genetic properties with its counterpart in yeast, eIF4A

The in vitro biochemical studies show that LeIF pro-tein is an RNA-dependent ATPase that has the ability

to unwind RNA⁄ DNA heteroduplexes in an ATP-dependent manner As is true of the other DEAD box proteins characterized, nucleotide binding and

hydro-124 K

80.0K

49.0K

34.8K

28.9K

25 LeIF LeIF eIF4A

Fig 6 Interaction between recombinant His6-LeIF, His6-D25LeIF

or His6-eIF4A with eIF4G in vitro Five micrograms of

GST-eIF4G (+ GST-eIF4G) or buffer alone (Control + GST) were incubated

with GSH-Sepharose beads and 5 lg of LeIF, D25LeIF or eIF4A as

described in Experimental procedures Proteins retained by the

matrix were eluted with glutathione and resolved by SDS ⁄ PAGE.

The blot was then probed with His-tag (shown), LeIF,

anti-eIF4A, and anti-GST IgG Lanes Load correspond to the purified

proteins loaded onto the matrix, + eIF4G correspond to proteins

bound to eIF4G and subsequently eluted with glutathione, and

Con-trol is LeIF protein eluted from the matrix without GST-eIF4G The

positions of marker proteins (in kDa) are indicated at the left.

lysis activity of LeIF is dependent on the presence

of RNA, and it is specific to ATP and dATP

[21,22,24,34] This ATPase activity can be abolished by

a mutation of the conserved lysine (K76A) in motif I,

which is consistent with studies of other helicases, such

as eIF4A [50] and yeast Has1 [57]; this confirms the

importance of this motif in nucleotide binding Indeed,

crystallographic analyses of yeast eIF4A [32] and viral

NS3 [58] have shown that this residue contacts the a,

b and sometimes c phosphates of the bound NTP

The LeIF protein has a Kmfor ATP binding around

350 lm, which is similar to that reported for other

DEAD box proteins such as human p68 [59], yeast

eIF4A ([30,52] and this study), yeast Has1 [57] and

E coliDbpA [60] This value, which is below the

cellu-lar concentration of ATP (5–10 mm), indicates that

LeIF can bind and hydrolyze ATP in the cell

cyto-plasm The kcatmeasured for ATP hydrolysis by LeIF,

1.2 min)1, is in the range of kcat values for eIF4A

(1 min)1for the mammalian factor [61] and 0.65 min)1

for the yeast factor; this study), E coli SrmB

(1.2 min)1 [62]) and RNA helicase II (1.9 min)1 [63])

but is much lower than that of yeast Ded1 (300 min)1

[28]), yeast Prp22p (400 min)1 [31]) and E coli DbpA

(600 min)1 [60]) This relatively weak ATPase activity

measured in vitro could reflect low intrinsic catalytic

activity Alternatively, the lack of post-translational

modifications in the recombinant protein or the

absence of specific substrates may contribute to the

low activity Of the DEAD box proteins that have

been studied biochemically, only DbpA from E coli

shows a strong RNA substrate specificity [60,64] It

also may be due to the absence of protein cofactors;

the ATPase and helicase activities of eIF4A purified

from rabbit reticulocyte lysates are increased in the

presence of eIF4F, eIF4B and eIF4H [30,53] Recently,

it was shown that cpc3 – the central domain of eIF4G

that binds eIF4A – stimulates the ATPase activity by

about 40-fold by lowering the KRNA

m by 10-fold and by raising the kcat by 4-fold [56] We see only a slight

eIF4G-specific enhancement of the ATPase activity

with eIF4A and LeIF, but our assay conditions were

probably not optimized Nevertheless, our results are

consistent with the published data, which implies a

functional interaction between eIF4A and LeIF with

eIF4G

The recombinant LeIF protein exhibits poor

ATP-dependent duplex unwinding activity in vitro as shown

previously for eIF4A [30] The unwinding in the

absence of ATP is found significant, which is

consis-tent with an intrinsic (ATP-independent) affinity of the

protein for RNA We demonstrate that LeIF protein

can exert its activity in a bidirectional way and unwind

RNA⁄ DNA heteroduplexes that have either a 3¢ duplex relative to the loading strand or a 5¢ duplex This suggests that LeIF acts nonprocessively, and it is only capable of unwinding short RNA duplexes The majority of RNA helicases studied so far are thought

to have directional unwinding Nevertheless, Ded1, eIF4A and p68 were reported to unwind duplexes in both directions in vitro [49,51,59] Although LeIF has similar biochemical properties to the eIF4A proteins from other organisms, there are some differences between LeIF and the yeast eIF4A that include a wider range for the optimum magnesium concentra-tion, a similar affinity for ATP and ADP, and a higher affinity for RNA These differences could reflect fundamental differences in the dynamics of the interac-tion of the protein within the eIF4F complex or within the translation machinery In this regard, the eIF4B protein has not been described so far and was not uncovered by the Leishmania or Trypanosoma sp gen-ome projects (http://www.genedb.org/)

Translation initiation in mammals and yeast is well studied; it involves many RNA–RNA, protein–RNA, and protein–protein interactions In contrast, know-ledge about the process of protein synthesis in Trypan-osomatidae protozoans is inferred by indirect evidence, such as sequence similarities between individual trans-lation factors with homologues from higher eukaryo-tes Recently, Dhalia et al [17] reported the in silico identification of multiple potential homologues of the three eIF4F components, eIF4E, eIF4A, and eIF4G These putative eIF4F components are expressed at similar levels and relative stoichiometry as those des-cribed for yeast and other eukaryote systems [17] In particular, the L major LmEIF4A1, which shows 100% identity with LeIF, is readily detected in the promastigote as a very abundant protein, which also is true for eIF4A from mammals and yeast [65,66] Nev-ertheless, our results show that LeIF cannot substitute for the yeast eIF4A in spite of the high sequence iden-tity between the two proteins Moreover, it does not support translation in vitro in reconstitution assays (M Altmann, unpublished data) However, these results are not surprising because the mammalian proteins do not support growth in yeast either [67]

Expression of LeIF in genetically engineered yeast strains where endogenous eIF4A is expressed off only one its two encoding genes results in severe growth inhibition Our experimental results exclude the possi-bility of a general toxic effect, or a difference in the expression levels or stability of LeIF in yeast; this sug-gests that LeIF can interact with the endogenous yeast factors within the translation initiation complex Inter-estingly, our results also emphasized the role of the

25 amino terminal residues of LeIF in its interactions

with the cellular machinery Deletion of this part

(D25LeIF), which is the part most divergent from

eIF4A, abolishes the severe dominant-negative

pheno-type of LeIF However, this variant also did not

complement the eIF4A double-deletion strain on

5-FOA plates The simplest explanation for our results

is that LeIF protein can assemble with the yeast

pro-teins to form stable, but nonproductive, interactions

that inhibit translation initiation The stability or

severity of these interactions are correlated with the

25 amino terminal residues because deletion of them

gives a slight dominant-negative phenotype that is

comparable to that obtained with overexpression of

the yeast eIF4A on the same ADH promoter Thus,

both D25LeIF and the excess eIF4A sequester the

translation initiation factors in a more transient, or

less inhibitory, fashion This implies that full-length

LeIF also could act as a translational inhibitor of the

mammalian host cells

In higher eukaryotes, eIF4A is assumed to be

recrui-ted to the mRNA through its interaction with eIF4G,

which acts as a molecular adapter that coordinates all

steps in translation initiation [68] It was also shown

that interactions between this fragment and eIF4A are

important for translation initiation and cell growth in

yeast [55] Our in vitro binding assay demonstrated

that LeIF can interact with the central domain of yeast

eIF4G, preferentially through its carboxy terminal

domain, as has been previously noted for eIF4A [52]

It is likely that this interaction occurs in vivo as well

and that this is, at least partially, the cause of the

dominant-negative phenotype This is further

suppor-ted by data showing that Leishmania LmEIF4G

pro-tein can bind both LmEIF4A1 and human eIF4A

in vitro[17] The role of the 25 amino terminal residues

is unclear, but they may form interactions with other

factors such as eIF4E

In our sequence comparisons, LeIF shows the

clo-sest similarity with DDX48 However, with the

excep-tion of two to four spliced genes, the vast majority of

trypanosomatid mRNA processing involves

trans-spli-cing; no exon junction complex (EJC) has been

identi-fied [69,70] Nonsense-mediated mRNA decay, which

is associated with the very early steps of translation,

has been described in yeast to humans [71], but it is

unknown so far in trypanosomatids [69] Furthermore,

a recent study indicates that TbEIF4AIII in T brucei,

which is similar to LmEIF4A2 in Leishmania, is the

closest orthologue to eIF4AIII [72] Taken together, it

is unlikely that LeIF plays the same role as DDX48

within the promastigotes and amastigotes of

Leishma-nia Yeast too lacks an EJC, although the downstream

sequence element probably serves a similar role in non-sense-mediated decay [71] Although there is evidence that DDX48 is more closely related to Fal1 than to eIF4A in yeast [18], it is unlikely that they play the same roles because Fal1 is located predominantly in the nucleolus, and it is thought to be involved in ribo-some biogenesis [48] Thus, a DDX48-like function probably does not exist in yeast either It is therefore intriguing that it is the amino terminus that shows the highest sequence divergence among these proteins (LeIF, DDX48, eIF4A and Fal1; Fig 1 and data not shown) Because it is the amino terminus of LeIF that confers the strong dominant-negative phenotype in yeast, it is possible that this short sequence modifies the function of the RecA domains or alters their inter-actions with other factors

Our results provide evidence for the potential involvement of LeIF in the translation machine in Leishmania This is further supported by data recently published that used RNAsi in T brucei [72] The high identity scores of Leishmania sp LeIF with proteins from other Trypanosomatidae species, such as T bru-cei and T cruzi (http://www.genedb.org/), which are pathogens responsible for human African trypanosom-iasis and Chagas disease, respectively, provides evi-dence that LeIF could be functional homologue of eIF4A, and that they all use similar mechanisms for translation initiation This is supported by the similar biochemical properties of LeIF and yeast eIF4A Nev-ertheless, definitive evidence must wait for the develop-ment of an in vitro translation system for Leishmania However, the potential interactions of these proteins with the host systems in the particular context of each infectious process also will need to be defined Anti-genic properties of LeIF, a cytosolic protein, could result from the infectious process when macrophages are lysed and the amastigotes, and the contents of the parasitophorous vacuoles, are released and scavenged

by macrophages LeIF could also be involved in direct interactions with the host cell and thereby constitute a virulence factor It will be important to see if LeIF expression affects translation in mammalian cells as it does in yeast, and whether it has cytotoxic effects because of its sequence similarity to eIF4AIII

In this regard, it is interesting that Leishmania EF-1a, which is another ubiquitous protein with antigenic properties [73], is able to diffuse into the cytosol of

L donovani infected macrophages and inactivate them [74] EF-1a plays an important role in eukaryotic pro-tein biosynthesis by binding aminoacyl-tRNAs and positioning them in the A site of ribosomes However, the cytoplamic Leishmania EF-1a binds the host’s Scr-homology-2-containing tyrosine phosphatase (SHP-1)