Interaction of three Caenorhabditis elegans isoforms of translation initiation factor eIF 4 E with mono-and trimethylated mRNA 5 cap analogues

Translation initiation factor eIF4E binds the m 7 G cap of eukaryotic mRNAs and mediates recruitment of mRNA to the ribosome during cap-dependent translation initiation. This event is the rate-limiting step of translation and a major target for translational control. In the nematode Caenorhabditis elegans, about 70% of genes express mRNAs with an unusual cap structure containing m3 2,2,7 G, which is poorly recognized by mammalian eIF4E. C. elegans expresses five isoforms of eIF4E (IFE-1, Vol. 49 No. 3/2002


IFE-2, etc.). Three of these (IFE-3, IFE-4 and IFE
The cap structure m 7 GpppN (where N is any nucleoside, Fig. 1) is present at the 5¢ ends of RNA polymerase II transcripts, including primary transcripts of all eukaryotic mRNAs and most U-type snRNAs (uridine-rich small nuclear RNAs) (Furuichi & Shatkin, 2000).The cap is added enzymatically to RNA precursors in the nucleus during the initial phases of transcription and acts as one of the signals for nuclear export (Hamm & Mattaj, 1990).After export, mRNAs remain in the cytoplasm where they serve as templates for protein synthesis.The cap both stimulates translation and protects against nucleolytic degradation (Rhoads, 1985).By contrast, snRNAs bind Sm proteins in the cytoplasm, their m 7 G cap is hypermethylated to m 3 2,2,7 GpppN (Fig. 1), and then they are imported back to the nu-cleus where they play a catalytic role in the splicing of pre-mRNAs (Mattaj, 1986).The m 3 2,2,7 G cap and Sm proteins constitute a bipartite nuclear targeting signal (Hamm et al., 1990).The m 7 G cap in mRNA is specifically recognized by a 25-kDa protein -termed eIF4E.Binding of eIF4E to the m 7 G cap occurs during formation of the 48S initiation complex, which is rate-limiting for translation initiation under normal (e.g., non-stressed) conditions (Raught et al., 2000).eIF4E is in turn bound to the 180-kDa initiation factor eIF4G, which also binds eIF4A, a 46-kDa ATP-dependent RNA helicase, eIF4B, a 70-kDa protein that stimulates eIF4A helicase activity, the 550-kDa complex eIF3, and poly(A)-binding protein (Raught et al., 2000).This complex serves to recruit mRNA to the 43S initiation complex and unwind secondary structure in the mRNA 5¢ untranslated region.The stimulatory effect of eIF4E on translation is regulated through the interactions with the 4E-binding proteins 4E-BP1, 4E-BP2 and 4E-BP3, as well as through phosphorylation of those proteins (Gingras et al., 1999;2001) and eIF4E itself (Rhoads, 1993;Sonenberg, 1996).
eIF4E levels in cells are also regulated at the transcriptional level (Rosenwald et al., 1993).Abnormal overexpression of eIF4E has a profound effect on cellular growth and tumorigenesis (Rhoads, 1991;Sonenberg, 1996;De Benedetti & Harris, 1999).eIF4E overexpression activates the ras oncogene whereas overexpression of GAP, a negative regulator of Ras, reverses the transformed phenotype (Lazaris-Karatzas et al., 1992).The aggressive growth phenotype of transformed cells is also suppressed by eIF4E antisense RNA (De Benedetti et al., 1991) or by overexpressing 4E-BP1 and 4E-BP2 (Rousseau et al., 1996).eIF4E is involved also in regulating apoptosis.Elevated eIF4E levels prevent apoptosis of NIH-3T3 cells subjected to serum deprivation (Polunovsky et al., 1996).Overexpression of 4E-BP1 or treatment with rapamycin of ras-transformed fibroblasts -both factors suppressing the activity of eIF4E -increases their susceptibility to apoptosis (Tan et al., 2000).
The primary structure of eIF4E is known for a wide variety of plant, animal and fungal organisms (McKendrick et al., 1999).Interestingly, multiple isoforms of eIF4E differing in primary structure have been found in a single species, e.g., two isoforms in humans, eIF4E-1 (Rychlik et al., 1987) and 4E-HP (Rom et al., 1998), and three isoforms in plants (Metz et al., 1992;Allen et al., 1992;Rodriguez et al., 1998;Ruud et al., 1998).The amino-acid sequences in all these proteins exhibit conservation of a central core domain, suggesting a similar tertiary structure of the protein.The cocrystal structure of mouse eIF4E bound to m 7 GDP (Marcotrigiano et al., 1997) and NMR structure of yeast eIF4E bound to m 7 GDP (Matsuo et al., 1997) showed that the purine moiety is sandwiched between the indole rings of two Trp residues .Glu-103 in mouse eIF4E (equivalent to Glu-105 in yeast eIF4E) interacts with the N1 and N2 protons of m 7 G by hydrogen bonding, whereas the N7 methyl group makes a van der Waals contact with the side chain of Trp-166.
Although the mRNAs of most eukaryotic organisms contain exclusively m 7 G caps, those of the nematode Caenorhabditis elegans contain both m 3 2,2,7 G caps (transcribed from about 70% of the genes) and m 7 G caps (transcribed from about 30% of the genes) (van Doren & Hirsh, 1990;Zorio et al., 1994).The m 3 2,2,7 G cap is acquired along with a conserved 22-nucleotide RNA sequence during trans-splicing of pre-mRNA (Blumenthal, 1998).Both types of C. elegans mRNA (m 7 G-capped and m 3 2,2,7 G -capped) are capable of associating with polysomes and being translated (Liou & Blumenthal, 1990).All eIF4E proteins from higher eukaryotes characterized to date strongly prefer the m 7 G over the m 3 2,2,7 G cap.This is based on several types of experiments: inhibition of in vitro translation by cap analogues (Cai et al., 1999), translation of mRNAs capped with modified structures (Dar¿ynkiewicz et al., 1988), and quenching of intrinsic eIF4E fluorescence by titration with cap analogues (Carberry et al., 1990;Wieczorek et al., 1999).The fact that C. elegans expresses and translates m 3 2,2,7 -capped mRNAs suggests that eIF4E of this organism may be different.Recently five isoforms of eIF4E were identified in C. elegans (Jankowska-Anyszka et al., 1998;Keiper et al., 2000).Experiments employing affinity chromatography on m 7 GTP-and m 3 2,2,7 GTP-Sepharose provided qualitative evidence that the eIF4E isoforms differed in the ability to discriminate between m 7 G-and m 3 2,2,7 G-containing caps.The aim of the present study was to obtain quantitative information on the cap binding properties of C. elegans eIF4E isoforms by fluorescence quenching.
UV absorption and fluorescence measurements.UV absorption spectra were recorded on a Cary 300 spectrophotometer and fluorescence spectra on a Perkin Elmer LS 50B luminescence spectrometer, applying the conventional right-angle geometry of detection in a 0.4 cm ´1.0 cm cell.The temperature of the cell-housing block was adjusted to 20 ± 0.1°C.For all fluorescence measurements, the excitation wavelength of 280 nm, corresponding to the absorption maximum of eIF4E, was used along with the excitation bandwidth of 5 nm and the emission bandwidth of 8 nm.The intensity of the excitation beam was reduced 50 fold with a special filter to protect the protein during titration.Fluorescence quenching was performed by adding 1.0 to 60 ml aliquots of the cap analogue at 150 mM concentration to 1200 ml of IFE solution (concentration 0.2 to 0.6 mM) in 20 mM Hepes (pH range 6.6 to 8.4) containing 100 mM KCl, 0.2 mM EDTA and 1 mM dithiothreitol.Iodide titrations were performed the same way by adding 3 M KI to IFE solution in a pH 7.2 buffer.
Due to partial overlap of the IFE and cap fluorescence spectra, the fluorescence emission due solely to the protein was obtained by subtracting the ligand contribution using least-square fitting.The spectra were subsequently corrected for dilution of the sample during the course of titration and for absorption of the cap analog (inner-filter effect) as described previously (Wieczorek et al., 1993).Fluorescence intensities were obtained by integrating the whole, corrected fluorescence spectrum.The equilibrium association constant, K as , of the IFE-cap complex was determined by non-linear least-square fitting of the theoretical curve to the experimental data points, according to the equation.where C L,tot and C P,tot are the total concentration of the cap analogue and IFE protein, respectively; DF = F 0 -F is the difference between the initial IFE fluorescence intensity (F 0 ) and that for the C L,tot concentration of the cap analogue (F); and DF max = F 0 -F eq is the difference between the initial and the final (F eq ) fluorescence intensities, after saturation of the protein with the cap analogue.Data analysis and calculations were performed by means of the programs Prism 2 (GraphPad Software, Inc.) and Origin 6 (Microcal Software, Inc.).NMR spectroscopy. 1 .2, 7.7, 7.9, or 8.2.The strong water signal was suppressed by presaturation.
Modelling of the IFE tertiary structures by homology.Three-dimensional molecular models of IFE-3, IFE-4 and IFE-5 were built with the MODELLER program (Sali & Blundell, 1993) using the crystal structure of the mouse eIF4E-m 7 GDP complex (PDB code: 1EJ1) (Marcotrigiano et al., 1997) as a template.Initially, the CLUSTAL W program (Thompson et al., 1994) was used to generate multiple sequence alignments for the proteins.Taking into account the general requirements of the eIF4E fold, some corrections were performed manually to preserve conservation of specific residues and to adjust the exact positions of inserted and deleted aminoacid sequences.In some cases, 3D evaluation procedure (Venclovas et al., 1999) was used to obtain the final structure-based sequence alignment.The final models were subjected to visual inspection to remove improper packing of residues, cavities and steric clashes.
The water-and iodide-accessible surfaces of the proteins were determined by means of the program NACCESS (S.J. Hubbard, J.M. Thornton, Department of Biochemistry and Molecular Biology, University College, London, GB, 1993), using van der Waals radii of 1.4 Å and 2.15 Å, respectively.

RESULTS AND DISCUSSION
The five isoforms of eIF4E existing in C. elegans may be divided into three classes (Keiper et al., 2000) on the basis of their primary sequences, cap-binding specificities, and requirements for embryonic viability (Jankowska-Anyszka et al., 1998;Keiper et al., 2000).Class A contains IFE-3, the isoform most similar to human eIF4E and the only isoform that is alone required for viability.IFE-3, like mammalian eIF4E, is retained on m 7 GTP-but not m 3 2,2,7 GTP-Sepharose.Class B consists of IFE-1, IFE-2 and IFE-5, proteins with high primary sequence identity, which are retained on both m 7 GTP-and m 3 2,2,7 GTP-Sepharose.None of these three proteins alone are required for viability, but the absence of all of them (caused by interfering RNA) causes 99% embryonic lethality.Class C contains IFE-4, the form most different in primary structure from the others; it is not essential for viability and is retained by m 7 GTP-but not m 3 2,2,7 GTP-Sepharose.
We chose IFE-3, IFE-4 and IFE-5 as representatives of the three classes for further physicochemical characterization and quantitative determination of cap binding specificity.The molecular structures of eIF4E isoforms from C. elegans have not yet been experimentally determined.In order to have a preliminary basis to interpret the spectroscopic data, the structures of IFE-3, -4 and -5 were modelled by sequence homology as described in Materials and Methods, taking advantage of the high similarity of the IFE primary structures (50% for IFE-3, 25% for IFE-4 and 42% for IFE-5) to that of mouse eIF4E.The known crystallographic structure of the mouse eIF4E-m 7 GDP complex was used as template (Marcotrigiano et al., 1997).According to expectations, the modelled structures of all three isoforms (not shown) appeared to have very similar tertiary folds, and were similar to that of the template eIF4E.The model structures gave no clear explanation for the observed differences in affinities of mono-and trimetyloguanosine caps for the eIF4E isoforms.

Fluorescence spectra of IFEs
The values of emission maximum and quantum yield of Trp residues in proteins can vary greatly due to variations in protein structure (Lakowicz, 1999).Fluorescence intensities of equimolar solutions of the IFE proteins (Fig. 2A) reflect to some extent the number of Trp residues in each protein: ten in IFE-5, nine in IFE-3, and five in IFE-4 (Jankowska-Anyszka et al., 1998;Keiper et al., 2000).The proportionality between fluorescence emis-sion and Trp content holds for IFE-5 and IFE-3, but a large deviation is observed for IFE-4; one would have expected about 50% signal intensity for IFE-4 cf.IFE-5 if each Trp contributed equally to the total emission.The data suggest that more of the Trp residues in IFE-4 are buried inside the protein, leading to fluorescence quenching by charged or aromatic amino-acid side chains.Our computer modelling of the three IFE structures bound to m 7 GDP showed that the fluorescence of the buried Trp residues is likely to be quenched by Arg, Lys, or Phe residues situated 3-5 Å from the indole rings.The fraction of unquenched, solvent-accessible Trp residues in the model structure of IFE-4 (Table 1) is smaller than in IFE-3 and IFE-5, thus leading to a substan-tially lower fluorescence emission of IFE-4.Iodide quenching experiments (Fig. 3) yielded the fraction of Trp fluorescence accessible to quenching: 65% for IFE-3, 57% for IFE-4, and 68% for IFE-5 (Table 1).These values independently confirm the lower contribution of buried Trp residues to the total fluorescence for IFE-4.In the normalized spectra of the IFE proteins (Fig. 2B), the positions of the emission maxima are in good agreement with the iodide quenching data.The IFE-3 spectrum is slightly blue-shifted comparing to that of IFE-5, while the IFE-4 emission maximum is blue-shifted even more (Table 1).
The number of solvent-exposed Trp residues and their solvent-accesible surfaces, as calculated from the modelled structures, roughly agree with the corresponding fractions of total fluorescence quenched by the iodide ions (Table 1).However, the area of solvent-accessible Trp surface is greater in IFE-3 than in IFE-5, whereas the proportion of fluorescence quenched by iodide was in the reverse order.The discrepancies could be ascribed, at least partially, to differences between the structures of the apo and ligand-bound proteins, due to a probable conformational rearrangement upon binding of the ligand.

Affinity of mono-and trimethylated cap analogues for the IFEs
Association of eIF4E with a cap analogue results in quenching of intrinsic Trp fluores-  cence in the protein (Wieczorek et al., 1999;McCubbin et al., 1988;Carberry et al., 1989;Ueda et al., 1991), presumably due to stacking of m 7 G between two Trp residues (Marcotrigiano et al., 1997;Matsuo et al., 1997).The equilibrium association constant, K as , can be derived from titration of eIF4E with increasing amounts of a cap analogue.High K as values of 10 6 M -1 to 10 7 M -1 reflect the tight, specific binding of the ligand to the protein.A 30% decrease of the fluorescence intensity was observed in the case of IFE-5 (Fig. 4A) as well as IFE-3 and -4 (not shown) on binding of the monomethylated cap analogues m 7 GTP and m 7 GpppG.Similar fluorescence quenching was observed for IFE-5, when titrated with the trimethylated caps m 3 2,2,7 GTP (Fig. 4B) and m 3 2,2,7 GpppG (not shown).The spectra shown in Figs.4A and 4B are uncorrected for the emission of the free ligand.The decrease in fluorescence intensity due to the ligand quenching was extracted from the spectra and corrected for sample dilution and inner-filter effect to yield the resulting DF.The theoretical curve was used to fit numerically the DF values (Fig. 4A and B, insets) as described in Materials and Methods, providing the equilibrium association constants K as of the IFE-cap analogue complexes (Table 2).The titration experiments were performed at pH 7.2, optimal for the specific binding of the cap analogues to the IFE proteins (see below).The K as values for the specific binding of the cap analogues to the IFE proteins are in the range of 2 ´10 6 M -1 to 7 ´10 6 M -1 .Nonspecific IFE-cap interactions are characterized by K as values   more than two orders of magnitude lower.In the latter case, only an upper limit of about 0.06 ´10 6 M -1 can be estimated from such titrations.
In chromatographic experiments (Jankowska-Anyszka et al., 1998;Keiper et al., 2000), the proteins retained on and eluted from either m 7 GTP-Sepharose or m 3 2,2,7 GTP-Sepharose were subsequently subjected to SDS/polyacrylamide gel electrophoresis and stained with Coomassie blue.All the isoforms considered (IFE-3, -4 and -5) were retained on the former column whereas only IFE-5 was captured on the latter.Although the intensity of staining need not always reflect the amount of a protein, the relative intensities of the bands corresponding to IFE-3, IFE-4 and IFE-5 correlate with the amounts of the isoforms bound to m 7 GTP or m 3 2,2,7 GTP during the titration experiments.Thus, the quantitative, fluorometrically determined values of K as are in good agreement with the qualitative results of affinity chromatography.
The K as values for the association of m 7 GTP and m 7 GpppG with the full-length human eIF4E, 1.2 ´10 6 and 5.7 ´10 5 , respectively (Wieczorek et al., 1999), are close to those measured for C. elegans eIF4E isoforms with the same analogues.The dinucleotide analogue m 7 GpppG is bound to each IFE protein weaker than its mononucleotide counterpart, similarly as in the case of human eIF4E.This effect can be attributed to two factors: 1) disadvantageous entropic contribution to the free energy of binding in the case of the longer molecule, and 2) more negative charge of the m 7 GTP phosphate chain in comparison with m 7 GpppG, the former being better stabilized by the positively charged Arg and Lys residues in the cap binding center (Marcotrigiano et al., 1997;Matsuo et al., 1997).IFE-3 and IFE-4 discriminate strongly between monoand trimethylated cap analogues, as does mammalian eIF4E.Double methylation at N 2 of the m 7 G ring disrupts at least one hydrogen bond between the cap and eIF4E.This feature may be responsible for the 4-fold decrease in translation activity of b-globin mRNA capped with m 3 2,2,7 G compared with m 7 G (Dar¿ynkiewicz et al., 1988), and the 17-fold decrease in the inhibitory activity of m 3 2,2,7 GTP in comparison to m 7 GTP in a rabbit reticulocyte translation system (Cai et al., 1999).By contrast, the affinity of mono-and trimethylated cap analogues for IFE-5 is similar (Table 2).Since IFE-5 has a high sequence homology with eIF4E as well as with IFE-3 and IFE-4, this raises a fundamental question concerning the differences in the cap recognition pattern among those closely related isoforms.The basis for the discrimination between the two cap structures by the active centers of these proteins may be subtle, in light of the structural similarities of the isoforms derived from homology modelling.

K as as a function of pH
The spectrophotometrically determined apparent K as values depend on environmental parameters like pH, ionic strength, and types of ions in solution (Record et al., 1991).The pH dependence of K as for the binding of m 7 GTP and m 3 2,2,7 GTP to IFE-5 (Fig. 5) shows that optimal affinity occurs at pH 7.2, lower than pH 7.6 that was observed in the case of full-length eIF4E from human erythrocytes (Carberry et al., 1989).A pH optimum of 7.6 was also found for cell-free translation of globin mRNA in reticulocyte lysate (Rhoads et al., 1983).Similarly, the degree of inhibition of in vitro translation by m 7 GTP increased from 48% at pH 6.8 up to 82% at pH 7.8 at 200 mM concentration of the cap analog.In this pH range m 7 G is a mixture of the positively charged and zwitterionic forms due to dissociation of the N1 proton (pK about 7.2-7.5,Wieczorek et al., 1990).Although preliminary spectroscopic (Carberry et al., 1989) and translation data (Rhoads et al., 1983) pointed to the zwitterionic form of cap that specifically binds eIF4E, only the cationic form of m 7 GDP was postulated to be recognized by eIF4E on the basis of the structural, crystallographic studies (Marcotrigiano et al., 1997;Matsuo et al., 1997).The N1 proton is required for hydrogen bond donation to Glu-103.Additionally, the cap analogues at elevated pH in solution undergo opening of the five-membered ring of the 7-substituted guanine, followed by hydrolysis of the glycosidic bond (Dar¿ynkiewicz et al., 1990).The stabil-ity of the cap analogues was tested by NMR (not shown).m 7 GMP was stable between pH 7.2 and 8.2 but generated breakdown products at higher pH values, where the zwitterionic form dominates.
Our results also suggest that the cationic form of cap analogs is recognized by IFEs.The observed differences in the pH dependence of the IFEs and eIF4E binding can be rather attributed to differences in procedures of the protein sample preparation, e.g. protein isolation and purification.
We are indebted to Mrs. Anna NiedŸwiecka for the assistance in analysis of the fluorescence titration data and Dr. Krzysztof Ginalski for help in homology modelling.Folding of the IFE forms was carried out using computer resources at the Interdisciplinary Centre for Mathematical and Computational Modelling (ICM), Warsaw University.The NMR spectra were recorded in the Laboratory of Nuclear Magnetic Resonance, Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Warsaw.

R E F E R E N C E S
Allen ML, Metz AM, Timmer RT, Rhoads RE, Browning KS. (1992) Isolation and sequence of the cDNA encoding the subunits of the isozyme form of wheat protein synthesis initiation factor 4F. J Biol Chem.; 267: 23232-6.
confirm earlier qualitative data derived from affinity chromatography.The dependence of K as on pH indicated optimal cap binding of IFE-3, IFE-4 and IFE-5 at pH 7.2, lower by 0.4 pH units than that of eIF4E from human erythrocytes.These results provide insight into the molecular mechanism of recognition of structurally different caps by the highly homologous IFEs.

Figure 2 .
Figure 2. Fluorescence emission spectra of IFE isoforms.Fluorescence spectra (excitation wavelength 280 nm) in 20 mM Hepes, pH 7.2, 100 mM KCl, 0.2 mM EDTA, 1 mM DTT, at 1 mM protein concentration (A) and after normalization to unity at the wavelengths corresponding to the maximal fluorescence intensities (B).
mRNA cap analogues as inhibitors of in vitro translation.Biochemistry.;38: 8538-47.Carberry SE, Rhoads RE, Goss DJ. (1989) A spectroscopic study of the binding of m 7 GTP and m 7 GpppG to human protein synthesis

Figure
Figure 5. Association equilibrium constant K as of binding of m 7 GTP and m 3 2,2,7 GTP to IFE-5 as a function of pH.