Further Biochemical and Kinetic Characterization of Human Eukaryotic Initiation Factor 4H*

A cDNA encoding human eukaryotic initiation factor (eIF) 4H was subcloned into a bacterial expression plasmid for purification of recombinant protein. Recombinant human eIF4H (heIF4H) was purified to greater than 95% homogeneity and shown to have similar physical characteristics to eIF4H purified from rabbit reticulocyte lysate as described previously. Functional studies have revealed that recombinant heIF4H functions identically to rabbit eIF4H in stimulating protein synthesis, and the ATP hydrolysis and helicase activities of eIF4A. More detailed enzymatic studies revealed that eIF4H increases the affinity of eIF4A for RNA by 2-fold, but has no effect on the binding of ATP by eIF4A. eIF4H stimulates the helicase activity of eIF4A at least 4-fold, and it is postulated that this stimulation occurs through increasing the processivity of eIF4A. Northern blot analysis shows that eIF4H is expressed ubiquitously in human tissues, and displays different levels of expression in given tissues relative to eIF4B. Secondary structure analysis of heIF4H by circular dichroism suggest that eIF4H has a mostly β-sheet structure, which appears similar to other RNA recognition motif-containing proteins. Finally, it is suggested that eIF4H functions in translation initiation through protein-protein interactions that possibly stabilize conformational changes that occur in eIF4A during RNA binding, ATP hydrolysis, and RNA duplex unwinding.

A cDNA encoding human eukaryotic initiation factor (eIF) 4H was subcloned into a bacterial expression plasmid for purification of recombinant protein. Recombinant human eIF4H (heIF4H) was purified to greater than 95% homogeneity and shown to have similar physical characteristics to eIF4H purified from rabbit reticulocyte lysate as described previously. Functional studies have revealed that recombinant heIF4H functions identically to rabbit eIF4H in stimulating protein synthesis, and the ATP hydrolysis and helicase activities of eIF4A. More detailed enzymatic studies revealed that eIF4H increases the affinity of eIF4A for RNA by 2-fold, but has no effect on the binding of ATP by eIF4A. eIF4H stimulates the helicase activity of eIF4A at least 4-fold, and it is postulated that this stimulation occurs through increasing the processivity of eIF4A. Northern blot analysis shows that eIF4H is expressed ubiquitously in human tissues, and displays different levels of expression in given tissues relative to eIF4B. Secondary structure analysis of heIF4H by circular dichroism suggest that eIF4H has a mostly ␤-sheet structure, which appears similar to other RNA recognition motif-containing proteins. Finally, it is suggested that eIF4H functions in translation initiation through protein-protein interactions that possibly stabilize conformational changes that occur in eIF4A during RNA binding, ATP hydrolysis, and RNA duplex unwinding.
The initiation of protein synthesis in eukaryotes is a complex process involving almost a dozen initiation factors that work in combination to bring the mRNA, the initiating methionyl-tRNA (Met-tRNA i ), 1 and the 40 S ribosomal subunit together into a 48 S translation complex. This 48 S complex can then associate with the 60 S ribosomal subunit to complete the initiation phase of protein synthesis (for recent reviews on translation initiation, see Refs. [1][2][3]. Eukaryotic initiation factor (eIF) 4A, eIF4B, and eIF4F specifically interact with the mRNA and facilitate its binding to the 43 S ribosome complex (40 S subunit⅐eIF2⅐GTP⅐Met-tRNA i ). eIF4H has been identified as a new protein translation initiation factor that enhances the activities of initiation factors eIF4A, eIF4B, and eIF4F in various protein synthesis assays (4).
Previous studies have shown that eIF4H stimulates the translational activities of eIF4B and eIF4F in a reconstituted reticulocyte lysate system, as well as the RNA-dependent ATP hydrolysis (ATPase) activity of the initiation factor eIF4A. Also, it was shown that eIF4H binds weakly to RNA, potentially through its conserved RNA recognition motif (RRM), which shares a 45% identity to the RRM in eIF4B (4,5). The presence of this RRM suggests that eIF4H may function similarly to eIF4B to increase the helicase activity of eIF4A, since this motif in eIF4B was shown to be essential for stimulation of duplex unwinding (6). Recent studies confirm that eIF4H stimulates the RNA helicase activity of eIF4A (7).
Previously, it was reported that eIF4H is not an abundant protein, and that ϳ1 mg can be purified from 6 liters of rabbit reticulocyte lysate. It was also reported that the human eIF4H protein is encoded by the human cDNA sequence KIAA0038 (GenBank™ accession number D26068) (4,8,9). Therefore, this human cDNA sequence was obtained for expression and purification of recombinant human eIF4H protein (heIF4H) from a bacterial system. Here, methodology is presented for easily obtaining large amounts of very pure heIF4H that allow for further biochemical and kinetic characterization of eIF4H. Results from studies described here show that recombinant heIF4H has physical properties similar to those found in eIF4H purified from rabbit reticulocytes, with activities identical to those previously reported for rabbit eIF4H (4). The availability of large quantities of active and pure protein made it feasible to examine more closely the functional properties of heIF4H in translation. In addition, by using recombinant heIF4H in the studies described below, it was possible to begin examining the differences between eIF4B and eIF4H, which are known to be similar in both activity and amino acid sequence (human eIF4H is 39% identical and 62% similar to human eIF4B). From these studies, it may be possible to develop a model of how eIF4H functions during the initiation of protein synthesis. and phosphocellulose (type P-11) from Whatman Inc.; IEF and high molecular weight SDS-PAGE standards from Bio-Rad; pH 3.5-10 ampholytes from Amersham Pharmacia Biotech; pH 7-9 and pH 8

Methods
Subcloning of Recombinant Human eIF4H-A clone of heIF4H cDNA was obtained as a generous gift from Dr. Nobuo Nomura at the Kazusa DNA Research Institute in Japan. This was provided in the pBluescript SKϩ vector as an insert between the EcoRV and NotI sites oriented to allow for transcription of the sense strand from the T7 promoter (named pBS4H). DNA mini-preps were performed using the alkaline lysis technique (10). PCR primers were designed to engineer a unique NdeI restriction site at the AUG start codon for heIF4H (4H1: 5Ј-TGAT-GACGGCATATGGCGGACTTC-3Ј, NdeI site underlined), and a unique SpeI site 18 nucleotides 3Ј of the UGA stop codon (4H2: 5Ј-CCCCACG-CACTAGTCCCTCCCAAC-3Ј, SpeI site underlined) for subcloning into the pET-17b expression vector. The PCR product and pET-17b plasmid were digested with NdeI and SpeI. The pET-17b fragment was gel purified and recovered by electroelution, treated with shrimp alkaline phosphatase, and then ligated to the digested PCR fragment. The ligation reaction was transformed directly into NM522 competent cells, and selected by ampicillin resistance. Restriction digestion and DNA sequencing confirmed the presence of the heIF4H coding region within the plasmid, which was named pET4H. Subsequently, pET4H was transfected into bacterial strain BL21(DE3) for expression of recombinant human eIF4H upon IPTG induction.
DNA Sequencing-DNA sequencing was performed by either the Molecular Biology Core Laboratory at Case Western Reserve University, or the Molecular Biology Core Laboratory at the Cleveland Clinic Foundation.
Purification of Initiation Factors from Rabbit Reticulocyte Lysate-Purification of eIF4H from rabbit reticulocytes was described previously (4). Purification of other eukaryotic initiation factors has been described previously (11,12).
Expression and Purification of Recombinant Human eIF4H from Bacteria-One liter of BL21(DE3)/pET4H culture was induced at midlog phase with 50 M IPTG for 4 h. Bacteria were pelleted, washed once in cold 25 mM Tris (pH 8.0), 10 mM EDTA, and then resuspended in cold standard buffer (SB ϭ 20 mM Tris (pH 7.5), 1 mM DTT, 0.1 mM EDTA) with 25% glycerol and protease inhibitors (0.5 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, 2.2 mg/liter aprotinin). Bacteria were lysed by sonication, and insoluble material was removed by centrifugation for 1 h at 50,000 ϫ g.
Purification of recombinant heIF4H follows the standard procedures used to purify rabbit eIF4H (4) with some modifications and additions as outlined below. All steps in the purification of heIF4H were performed at 4°C. Soluble protein was applied by batch wash sequentially to DEAE-cellulose and CM-cellulose, both equilibrated in SB ϩ 50 mM KCl ϩ 25% glycerol. Unabsorbed protein was collected by filtering the resin using a small column, and then washing with an equal volume of SB ϩ 50 mM KCl ϩ 25% glycerol. The protein that did not bind to either DEAE-or CM-cellulose (flow-through) was then applied to a column containing phosphocellulose equilibrated in SB ϩ 50 mM KCl ϩ 25% glycerol. The column was washed with at least 1 column volume of equilibrating buffer at ϳ0.5 ml/min collecting 10-min fractions. A 50 -500 mM KCl gradient (50 mM KCl/column volume) in SB ϩ 25% glycerol was applied at ϳ0.5 ml/min collecting fractions every 10 min and monitoring absorbance at 280 nm. Fractions were analyzed on a 15% SDS-polyacrylamide gel, and those containing eIF4H were pooled and dialyzed against a saturated (NH 4 ) 2 SO 4 solution with 20 mM Tris (pH 7.5) overnight to precipitate the protein. Precipitated protein was pelleted, resuspended in a small volume of SB ϩ 500 mM KCl ϩ 25% glycerol, and dialyzed overnight against the same. The sample was then applied to a Superdex-75 FPLC column and developed with SB ϩ 500 mM KCl ϩ 25% glycerol at 0.1 ml/min collecting 0.5-ml fractions and monitoring absorbance at 280 nm. Fractions were analyzed on 15% SDS-polyacrylamide gels, and those containing eIF4H were analyzed in an ATPase assay both with and without eIF4A. Fractions that were free of contaminating Escherichia coli proteins with ATPase activity, that stimulated the RNA-dependent ATPase activity of eIF4A, and with the highest concentration of eIF4H were pooled and dialyzed against SB ϩ 100 mM KCl ϩ 25% glycerol. Protein concentration was determined by Bradford assay (eIF4H has no tryptophans and gives a poor UV reading at 280 nm), and stored at the liquid nitrogen vapor temperature.
Liquid Chromatography Mass Spectrometry (LC/MS)-Samples of rabbit and recombinant eIF4H were analyzed by LC/MS at the Cleveland Mass Spectrometry Facility, Cleveland State University. Approximately 50 pmol of sample (20 -50 l) was injected onto a 2.1 mm ϫ 25-cm C-18 reverse phase HPLC column (Vydac) coupled to a Micromass Quattro-II triple quadrapole electrospray mass spectrometer. Peaks from the HPLC were analyzed by mass spectrometry and molecular weights determined.
Two-dimensional Isoelectric Focusing/Sodium Dodecyl Sulfate (Twodimensional IEF/SDS)-Polyacrylamide Gel Electrophoresis-Purified eIF4H was analyzed by two-dimensional IEF/SDS electrophoresis as described previously (4). Where noted, samples were dephosphorylated by treatment with shrimp alkaline phosphatase for 2.5 h at 37°C.
RNA-dependent ATP Hydrolysis Assay-Combinations of the initiation factors eIF4A, eIF4F, eIF4B, and eIF4H were used to investigate RNA-dependent ATP hydrolysis by measuring the release of 32 P i from [␥-32 P]ATP as described previously (4,15). Briefly, 20-l reactions contained 15 mM HEPES (pH 7.5), 80 mM KCl, 2.5 mM Mg(CH 3 CO 2 ) 2 , 1 mM DTT, 0 -400 M [␥-32 P]ATP (specific activity of 2,000 -4,000 cpm/pmol), 0 -100 M poly(A) (concentration determined using 100-mer units), and various concentrations and combinations of initiation factors. Reactions were incubated at 37°C for 15 min, and 32 P i release was quantitated by organic extraction. The amounts of initiation factors eIF4A, eIF4B, eIF4F, and eIF4H added are shown in the legend of Table I. The amounts of eIF4A and eIF4H used to determine kinetic constants were 1.25 g each.
Helicase Assay-The ability of eIF4H to stimulate the helicase activity of eIF4A was performed as described in detail in Ref. 7. Duplexes consisted of a 50-nucleotide RNA (RNA-1) made by in vitro transcription (Megashortscript kit) that is hybridized at its 3Ј end to RNA oligonucleotides that are 10 -15 bases in length (RNA-10 through 15), which are 5Ј end-labeled with 32 P (using [␥-32 P]ATP and T4 polynucleotide kinase). Preparation and sequences of RNA duplexes, as well as their stabilities are described in Ref. 7. In general, 20-l reactions contained 20 mM HEPES (pH 7.5), 70 mM KCl, 2 mM DTT, 1 mg/ml bovine serum albumin, 1 mM Mg(CH 3 CO 2 ) 2 , 1 mM ATP, 2.0 nM duplex, 0.2-0.4 M eIF4A, and 0 -0.4 M eIF4H. Reactions were incubated at 35°C for 0 -15 min, and terminated by adding 5 l of stop solution (50% glycerol, 2% SDS, 20 mM EDTA, 0.01% bromphenol blue, and 0.01% xylene cyanol). Reactions were analyzed on a 12% native polyacrylamide gel (19:1, acrylamide:bisacrylamide) (pre-electrophoresed at 4°C for 20 -30 min) for 1.5 h at 200 V, 4°C. Gels were then scanned directly using an Ambis radioanalytical scanner for 30 -60 min. Results were quantitated as described previously (7). Curves in Fig. 4B were fit using a single phase exponential equation as described in Ref. 7, and the curve in Fig. 4C was fit using a one-site binding equation, both using the GraphPad graphing software (Prism).
Northern Blot-Probes were made from cDNA fragments of human eIF4H, human eIF4B, and ␤-actin by the random primer method using the Prime-a-Gene labeling kit as described by the manufacturer. These 32 P-labeled probes (internally labeled with [␣-32 P]CTP) were used to identify eIF4H and eIF4B mRNA separately on a Human Multiple Tissue Blot containing 2 g of poly(A) ϩ RNA per lane from eight different human tissues following procedures described by the manufacturer. Blots were exposed to x-ray film, and bands were quantitated using a SciScan 5000 densitometer (U. S. Biochemical Corp.) and OS-Scan Image Analysis System software (Oberlin Scientific Corp.). Quantitation of each band by densitometry was standardized to that in liver tissue on the same blot where the level of mRNA expression in liver tissue was set to a value of 1.0 (arbitrary units), and all other values were calculated relative to that within the same blot. The cDNA fragments of human eIF4H and human eIF4B used to make probes were created by PCR amplification from pBS4H and pET4B (a generous gift from Drs. T. Pestova and C. Hellen), respectively. Primers used for eIF4H (4H1 and 4H2) are described above. Primers used for eIF4B are as follows: 4B1, 5Ј-ATGGCGGCCTCAGCAAAAAAGAAGA-3Ј; 4B3, 5Ј-GAGAAAAGCACAGGATGTAGAGGTC-3Ј. The cDNA used for the ␤-actin probe was supplied by the manufacturer.
Circular Dichroism-A 200-l 6 M sample of heIF4H in a 20 mM KPO 4 (pH 7.2), 50 mM KCl, 10% glycerol buffer was analyzed at 10°C using a Jasco J-600 spectropolarimeter. Secondary structure was determined using the Jasco J-700 for Windows secondary structure estimation software. Denaturation studies were performed using 6 M heIF4H diluted into urea solutions at concentrations shown in Fig. 7B. Curve fitting was performed using the GraphPad graphing software (Prism).

Subcloning and Expression of Human eIF4H
The sequence of the cDNA encoding human eIF4H (heIF4H) and the aligned protein sequence, is shown in Fig. 1. This cDNA was subcloned into an expression vector to facilitate purification of large quantities of recombinant protein from bacteria. PCR primers 4H1 and 4H2 (shown in Fig. 1) were used to amplify the coding sequence for heIF4H with a unique NdeI restriction site engineered at the AUG start codon, as well as a unique SpeI site 18 base pairs from the UGA stop codon at the 3Ј end of the coding sequence. This PCR product was then subcloned into the pET-17b expression vector between the NdeI and SpeI sites, and this plasmid was named pET4H. Restriction digestion and DNA sequencing confirmed the presence and integrity of the heIF4H coding sequence. Fig. 2A shows the levels of recombinant heIF4H expressed using 50 M IPTG The open reading frame for heIF4H is shown in capital letters in the cDNA sequence. Positions of PCR primers are shown above the cDNA sequence with arrows designating the 5Ј 3 3Ј direction of the primers. The 5Ј most portion of the primer 4H1 is complementary to sequence within the pBluescript SKϩ plasmid (not shown). Nucleotides mutated after PCR amplification to create the desired restriction sites are in shaded boxes (wild type cDNA sequence is shown, refer to "Methods" section for sequences of primers). The RRM is located between amino acid residues 36 and 117, and the RNP-2 (A 44 YVGNL) and RNP-1 (K 82 GFCYVEF) domains are shown in hatched boxes. The putative poly(A) addition signal is between nucleotides 2457 and 2463 and is underlined.
(lane 4), and the amounts in the soluble and insoluble fractions (lanes 5 and 6, respectively). As can be seen, high quantities of heIF4H are expressed (compare lanes 3 and 4), and a significant portion of the heIF4H was in the soluble fraction (at least 75%). This was determined by comparing the ratios of protein "X" to heIF4H in the total protein fraction, soluble fraction, and insoluble fraction.

Purification of Recombinant Human eIF4H
To purify recombinant heIF4H from E. coli, techniques that were successful in purifying eIF4H from rabbit reticulocytes (4) were employed. As was done with rabbit eIF4H, all buffers used during the purification contained 25% glycerol to stabilize the protein. A detailed protocol for purification of heIF4H is described under "Methods." Soluble heIF4H was obtained upon clarification of the cell lysate at 50,000 ϫ g (Fig. 2B, lane 3), applied to DEAE-and CM-cellulose columns by batch wash in 50 mM KCl, and heIF4H was recovered in the flow-through (Fig. 2B, lane 4). This fraction was now enriched for heIF4H (compare lanes 3 and 4), and the majority of nucleic acids that bind tightly to DEAE-cellulose at low salt concentrations have been removed (data not shown). heIF4H from the DEAE-and CM-cellulose columns was then applied to a phosphocellulose column in 50 mM KCl, and eluted using a 50 -500 mM KCl gradient. Fig. 2B, lane 5, shows that the flow-through from phosphocellulose chromatography contained most of the contaminating E. coli proteins, while the majority of heIF4H was found to elute between 200 and 225 mM KCl (pool from gradient shown in lane 6). At this point the heIF4H preparation was ϳ70 -80% pure, yet contained high amounts of contaminating RNA-dependent and -independent ATP hydrolysis activity (data not shown), and therefore would not be useful in some functional studies. Previous studies suggested that eIF4H by itself has no ATPase activity (4). To further purify heIF4H, it was applied to a Superdex-75 FPLC column in 500 mM KCl, and column fractions containing heIF4H were analyzed in an ATP hydrolysis assay. The final pool was made from fractions that were most concentrated for heIF4H, were free of contaminating ATPase activity, and stimulated the RNA-dependent ATPase activity of eIF4A. A sample from this pool is shown in Fig. 2B, lane 7. After this purification step, heIF4H was Ͼ95% pure, and was used in physical analysis and functional activity assays as described below.

Physical Characterization of heIF4H Relative to Rabbit eIF4H
Determination of Molecular Weight-Recombinant human eIF4H, when compared with the rabbit protein by SDS-PAGE, has a similar apparent molecular weight (refer to Fig. 2). The native molecular weight of heIF4H was determined empirically by size exclusion chromatography using the Superdex-75 FPLC column. Profiles from purifications of both rabbit and human eIF4H showed that both proteins eluted at the same position from this column (data not shown).
To determine their actual molecular weight, samples of both rabbit and recombinant human eIF4H were analyzed by LC/ MS. Mass spectrometry analysis of the main reverse phase HPLC peak showed that the heIF4H preparation contained two species (data not shown). The major species of heIF4H has a molecular mass of 25,072 Da, and the minor species has one of 24,340. The molecular mass of human eIF4H calculated from its amino acid sequence is 25,200 Da. Removal of the initiating formylated methionine (fMet) after protein synthesis would yield a molecular mass of 25,069 Da, which is essentially identical to that of the major species determined by LC/MS within error (Ϯ2-4 Da). This result confirms, along with the complete DNA sequencing of the coding region in pET4H, that the expressed heIF4H protein has the expected amino acid sequence. The minor species seen in the heIF4H sample (24,340 Da) was most likely one of the contaminating E. coli proteins seen in Fig. 2B, lane 7, which may have co-eluted by HPLC with heIF4H.
Analysis of rabbit eIF4H by LC/MS revealed three species within the major HPLC peak with the following molecular masses: 25,166 Da (major species), 25,277 Da, and 25,387 Da (minor species) (data not shown). The calculated molecular mass of human eIF4H, as stated above, is 25,200 Da, yet it is known that the N terminus is blocked in the rabbit protein (4) so that the first initiating methionine is most likely removed and the new N-terminal amino acid subsequently acetylated (16 -18). When the molecular mass is recalculated for the Nterminal blocked human eIF4H, the resulting size is 25,111 Da. However, all of the molecular masses determined by LC/MS for rabbit eIF4H were slightly higher than that calculated, and the differences between the three species were 110 -111 Da. The increase in molecular mass of the rabbit protein relative to that calculated for the human eIF4H sequence (55 Da) was most likely due to minor differences in amino acid sequence. The differences between the three species in the rabbit eIF4H preparation may be due to addition or removal of an amino acid(s) (average amino acid ϳ120 Da), or post-translational modification(s) of the protein.
Determination of Isoelectric Point-The isoelectric point for heIF4H was determined using two-dimensional IEF/SDS-polyacrylamide gel electrophoresis and results are shown in Fig.  3A. The heIF4H migrated on two-dimensional IEF/SDS gels as two distinct spots with the same apparent molecular weight, but separate pI values. The major band has a pI of ϳ8.4 and the minor ϳ8.2. The calculated pI of heIF4H with removal of the N-terminal fMet (as determined by mass spectrometry above) is 8.0. The minor band seen in the heIF4H preparation may be due to random loss of an amide nitrogen, which would decrease the protein's pI, but not significantly affect its molecular weight. The migration of heIF4H on two-dimensional IEF/SDS gels was similar to that seen for rabbit eIF4H shown in Fig. 3B. Rabbit eIF4H migrated as two or more distinct bands, with the major band migrating at a pI of ϳ8.7, and a minor band migrating at an apparently higher molecular weight and lower pI close to 8.5. These results are similar to what was previously reported for rabbit eIF4H (4). It is possible that the slight discrepancies between rabbit and recombinant human eIF4H seen by two-dimensional IEF/SDS/PAGE are due to differences in amino acid sequence or protein modification, as is also suggested by the mass spectrometry data (see above).
Results from mass spectrometry analysis show that there are indeed multiple species within the rabbit eIF4H preparation. The major species seen by LC/MS had the lower molecular mass (25,166 Da), and may have been represented by the major band seen on two-dimensional IEF/SDS gels. The minor band migrating at the higher molecular weight and more acidic pI on two-dimensional IEF/SDS gels may have represented one of the minor species of rabbit eIF4H seen by LC/MS (25,277 or 25,387 Da), which could be a post-translationally modified form of eIF4H. A phosphorylation would increase the molecular masses of eIF4H by 78 Da and decrease the pI by ϳ0.2 pH units. To determine if rabbit eIF4H is phosphorylated, the protein was treated with alkaline phosphatase and reanalyzed by two-dimensional IEF/SDS/PAGE. Results of this experiment showed that there was no difference in the migration of bands between the untreated and phosphatase-treated samples (data not shown). Therefore rabbit eIF4H does not appear to be phosphorylated, but the possibility that it is modified in some other way cannot be ruled out at this time.

Functional Characterization of heIF4H
Relative to Rabbit eIF4H Globin Synthesis Assay-Since the recombinant heIF4H was determined to be physically similar to rabbit eIF4H as described above, it was necessary to assess if heIF4H is functionally similar to rabbit eIF4H. Initially, the recombinant protein was examined in the assay first used to identify eIF4H, the globin synthesis assay. A reconstituted rabbit reticulocyte lysate system deficient for eIF4B, eIF4F, and eIF4H was used to compare the activities of rabbit and human eIF4H. It was found that heIF4H stimulated translation to the same degree as an equivalent amount of rabbit eIF4H (1.9-and 1.7-fold over background, respectively, data not shown).
ATP Hydrolysis Assay-Next, the activity of heIF4H was compared with that of rabbit eIF4H in the RNA-dependent ATP hydrolysis (ATPase) assay. heIF4H alone had no detectable RNA-dependent (Table I) or -independent (data not shown) ATPase activity. However, as shown in Table I, heIF4H enhances the ATPase activity of eIF4A. These results are similar to those previously reported for rabbit eIF4H (4). In addition, when compared with rabbit eIF4H, recombinant heIF4H had comparable activities. Differences seen were most likely due to the higher level of contaminating ATPase activity in the rabbit eIF4H preparation used (compare background levels of heIF4H only and rabbit eIF4H only in Table I). These results using recombinant heIF4H then prove that eIF4H has no ATPase activity of its own, but can stimulate the ATPase activity of eIF4A.
Since the recombinant heIF4H was as active as rabbit eIF4H and had less contaminating ATPase activity, it was used in further enzymatic analyses of the stimulation by eIF4H of eIF4A in the ATPase assay. It has been reported previously using this assay, that the activation constant (K act ) of eIF4A for poly(A) is 15 M, and the addition of either eIF4B or eIF4F decrease this K act to 60 and 20 nM, respectively (19). In those experiments, K act was defined as the concentration of activating RNA required to achieve one-half the maximal velocity (V max ), and is similar to a binding constant. In this study, analogous experiments were performed using heIF4H to ascertain if this initiation factor also affects the K act of eIF4A for poly(A). Results showed that the measured K act of eIF4A for poly(A) was 18 M, which is essentially the same as previously reported (15 M), and the addition of heIF4H decreased the K act by approximately 2-fold from 18 to 8 M, and showed a slight decrease (25%) in the V max (data not shown). Previously, both eIF4B and eIF4F were shown to have little effect on the V max of the reaction (19). Thus, as opposed to eIF4B and eIF4F, which decrease the K act of eIF4A for poly(A) 250-1000-fold, the effect FIG. 3. Two-dimensional isoelectric focusing/SDS-polyacrylamide gel electrophoresis. The migration of recombinant heIF4H (A) was compared with that of rabbit eIF4H (B). Standards used are labeled numerically: 1, hen egg white conalbumin, 76 kDa, pI 6.0, 6.3, 6.6; 2, bovine serum albumin, 66.2 kDa, pI 5.4, 5.5, 5.6; 3, bovine muscle actin, 43 kDa, pI 5.0, 5.1; 4, rabbit muscle glyceraldehyde-3-phosphate dehydrogenase, 36 kDa, pI 8.3, 8.5; 5, bovine carbonic anhydrase, 31 kDa, pI 5.9, 6.0; 6, soybean trypsin inhibitor, 21.5 kDa, pI 4.5; 7, Equine myoglobin, 17.5 kDa, pI 7.0. Molecular mass (in kDa) and isoelectric points of the standards are marked along the axes of the gel. eIF4H has on the K act is minimal. It was found previously that eIF4B did not change the K m of eIF4A for ATP (80 M), but did increase the V max by 2-fold when using 16 M poly(A) (19). Here, the K m and V max values of eIF4A for ATP using 100 M poly(A) were determined to be 286 M and 187 fmol/s, respectively, and the addition of heIF4H had no effect on either the K m (261 M) or V max (181 fmol/s) (data not shown). Differences between the K m and V max of eIF4A for ATP reported in this and previous studies will be discussed later.
Helicase Assay-Recently, it was reported that rabbit eIF4H stimulates the helicase activity of eIF4A (7). To determine if recombinant heIF4H is similar to rabbit eIF4H, both were analyzed in a RNA helicase assay using 2.0 nM RNA-1/RNA-12 duplex as substrate (described under "Methods" and Ref. 7). Fig. 4A shows a schematic representation of the helicase assay showing the RNA duplex being converted to single-stranded RNA monomers by incubation with eIF4A, ATP, and Ϯ eIF4H. Results of these experiments are shown in Fig. 4B. When 0.2 M eIF4H was used, both rabbit and recombinant proteins stimulated the helicase activity of 0.4 M eIF4A by a factor of 2 when measuring total amount of duplex unwound, and by a factor of 3.5 when measuring the initial rate of the reaction (femtomole of duplex unwound per min). In addition, heIF4H proved to be as active as rabbit eIF4H in this assay, and displayed less contaminating helicase activity than the rabbit protein.
To determine the optimal ratio of eIF4H relative to eIF4A, 0 -0.4 M heIF4H was titrated into reactions containing 0.2 M eIF4A and 2 nM RNA-1/RNA-12 duplex substrate. Fig.  4C shows that half-maximal unwinding was achieved when 0.11 M heIF4H was used. Therefore, under the conditions used in this assay, maximal stimulation of eIF4A's activity occurs when eIF4A and eIF4H are present in equimolar amounts.
To determine how eIF4H affects the helicase activity of eIF4A, the initial rates of unwinding were measured in the helicase assay using RNA duplexes of varying length and stability and equimolar amounts of eIF4A and eIF4H (0.2 M). Fig.  5A shows the determination of initial rates using 0.4 M eIF4A alone, while Fig. 5B shows the initial rates of 0.2 M eIF4A with 0.2 M heIF4H added. The addition of heIF4H increased the initial rate of unwinding by eIF4A for all duplexes analyzed (see below). Fig. 5C shows that there was a linear relationship between the ln(initial rates of unwinding) and the ⌬G values of the RNA duplexes for both eIF4A only and the eIF4A ϩ heIF4H combination. The slope of the line obtained for eIF4A equals 0.31, which is similar to that reported earlier using 0.4 M eIF4A (7). When heIF4H was added in equimolar amounts to eIF4A (0.2 M each), the slope of the line changed to 0.19. The significance of this change will be discussed below.
It should be noted that the initial rates for eIF4A alone were determined using 0.4 M protein to allow for comparison with previously published results, while those measured for the eIF4A ϩ heIF4H combination used half as much eIF4A (0.2 M). This was necessary since the helicase activity using 0.4 M of both initiation factors was too fast for accurate measurement of the initial rate of unwinding of the less stable substrates (data not shown). Therefore, the initial rates for the eIF4A ϩ heIF4H combination (0.2 M each) should be compared with those obtained using 0.2 M eIF4A. Since previous studies have shown that there is a linear relationship between the concentration of eIF4A and the initial rate of duplex unwinding (7), the slope of the line shown in Fig. 5C will be unchanged at 0.2 M eIF4A, only shifted to lower ln(initial rate) values at the same ⌬G values. The calculated values for 0.2 M eIF4A are shown in Fig. 5C as a calculated curve (dashed line). Therefore, when comparing the value calculated for 0.2 M eIF4A alone to that obtained experimentally for eIF4A ϩ heIF4H (0.2 M each) using the RNA-1/RNA-10 duplex, heIF4H would increase the initial rate of eIF4A-dependent duplex unwinding by a factor of 2. Therefore, heIF4H does increase the initial rate of unwinding by eIF4A for all duplexes analyzed, and by examining the lines shown in Fig. 5C, the fold stimulation by heIF4H increases with the stability of the duplexes.

Expression of eIF4H mRNA in Human Tissue
Several translation initiation factors have been shown to be expressed ubiquitously in human tissues (20 -22), 2 and it is expected that all translation factors are expressed in all tissue types. To determine if eIF4H is also expressed ubiquitously in human tissues, a radiolabeled probe was created from the cDNA sequence of human eIF4H to hybridize to a human multiple tissue blot. Fig. 6A shows that eIF4H was expressed in all tissues analyzed. This confirms results from previous studies performed by Nomura et al. (8), who stated that HU-MORFU_1 (heIF4H) was present in all human tissues and cell lines they examined. Also, the mRNA for human eIF4H was found to be approximately the same size as the reported length for the human cDNA, HUMORFU_1 (2,477 nucleotides) (8).
Because eIF4H and eIF4B both stimulate the ATPase and helicase activities of eIF4A, the question arises why are there two translation initiation factors with related functions. eIF4H and eIF4B may function differently in the translation of divergent mRNAs, in other non-standard initiation events (i.e. internal initiation, reinitiation, etc.), during development, or in particular tissues. To begin answering these questions, the expression of eIF4H relative to eIF4B was compared on the human multiple tissue blot to determine if there is tissuespecific expression of either mRNA. A cDNA probe for eIF4B was created and used to examine the same blot as eIF4H (Fig.  6A). eIF4B was also expressed ubiquitously in all tissue types, and the size of eIF4B mRNA was determined to be approximately the reported length for human eIF4B cDNA (3,877 nucleotides) (5). Fig. 6B shows a quantitative comparison of eIF4B and eIF4H expression in these tissues using ␤-actin (blot shown in Fig. 6A)     eIF4H and eIF4B mRNA in a given tissue, only their relative ratios. When the expression of a given mRNA in each tissue was standardized to that in liver on the same blot, it can be seen that eIF4B and eIF4H were expressed differently in several of the tissues examined. eIF4B expression was approximately 1.7-fold higher than eIF4H in pancreas, 3.7-fold higher in skeletal muscle, and 4-fold higher in kidney (this latter value may be slightly high due to background signal on the blot). Conversely, eIF4H was expressed approximately 2-fold higher than eIF4B in brain, and about 1.4-fold higher in placenta. It should be noted that the signal for the 2,000-nucleotide ␤-actin mRNA in skeletal muscle was obscured by the very intense signal from the muscle-specific 1,800-nucleotide ␤-actin mRNA (seen in heart and skeletal muscle), and therefore accurate quantitation of ␤-actin mRNA cannot be performed for this tissue. These results show that there was a difference between the expression of eIF4B and eIF4H mRNA in several tissues, FIG. 4. Stimulation of the helicase activity of eIF4A by rabbit and recombinant human eIF4H. A, schematic representation of helicase assay. The RNA duplex substrate is represented at the top with the 50-nucleotide long RNA-1 (shown 5Ј to 3Ј) base paired at its 3Ј end to the shorter RNA oligonucleotide (RNA-10 through -15, shown 3Ј to 5Ј). The shorter RNA oligonucleotide is 5Ј end-labeled with 32 P, which is designated by an asterisk (*). Upon incubation with eIF4A and ATP (Ϯ eIF4H), the RNA duplex is unwound and the single-stranded RNAs are released. B, unwinding of RNA-1/RNA-12 duplex substrate by eIF4A only (filled circle), rabbit eIF4H only (open square), recombinant heIF4H only (open triangle), eIF4A ϩ rabbit eIF4H (filled square), and eIF4A ϩ heIF4H (filled triangle). Concentrations used are 0.4 M eIF4A, 0.2 M eIF4H, 2 nM RNA duplex, and 1 mM ATP. C, concentration curve of heIF4H in helicase assay containing eIF4A. The plot shows the initial rates of RNA-1/RNA-12 duplex unwinding by eIF4A (0 M eIF4H value), and the increase in this value as heIF4H is titrated in. Initial rates were determined at each concentration of heIF4H in individual experiments (not shown) using 2 nM RNA-1/RNA-12 duplex substrate, 0.2 M eIF4A, and 1 mM ATP by measuring unwinding over the first 2 min of the reaction. Linear fits were applied to each, and the initial rate was taken from the slope of each line.

Analysis of Secondary Structure Using Circular Dichroism
With milligram quantities of highly pure and active heIF4H, it was now possible to perform preliminary structural analysis of the protein. The heIF4H preparation used in these studies was the same preparation used in all enzymatic assays described above. Examination of the amino acid sequence of human eIF4H using the PHD structural prediction program suggested that eIF4H is an ␣-␤ protein (4). To assess the accuracy of this prediction, determination of heIF4H's secondary structure was performed using circular dichroism and a representative spectra is shown in Fig. 7A. Deconvolution of the heIF4H spectra revealed that the protein had a high content of ␤-sheet and random coil (ϳ50 and 45%, respectively), and very little ␣-helix (ϳ1%) or turn (ϳ4%). Denaturation studies using urea are seen in Fig. 7B with the change in molar ellipticity at 212 nm (␤-sheet) shown as a function of urea concentration. Secondary structure was lost upon treatment with urea, proving that heIF4H was indeed structured, and the inflection point of the denaturation curve was at 5.8 M urea. DISCUSSION In this study, the cDNA for human eIF4H (HUMORFU_1) (4,8,9) was overexpressed in E. coli, and the recombinant protein, heIF4H, was purified without the use of an affinity tag to greater than 95% homogeneity using standard chromatography procedures. The recombinant protein was shown to have many of the same physical characteristics as rabbit eIF4H, and slight differences suggested there are minimal discrepancies in the amino acid sequence of human and rabbit eIF4H. Results obtained for the recombinant heIF4H in the globin synthesis and ATPase assays proved that the recombinant heIF4H has activities similar to rabbit eIF4H shown both in these and previous studies (4). When rabbit eIF4H and heIF4H were compared in the helicase assay, it was also found that both proteins had the same activities. In conclusion, all of the assays indicated that the rabbit and the recombinant human eIF4H had identical activity, confirming that the cDNA for HU-MORFU_1 is indeed eIF4H, and supporting the conclusion that eIF4H is a translation initiation factor (4). In addition, heIF4H purified from E. coli was free of other contaminating eukaryotic translation initiation factors, had less contaminating ATP hydrolysis and helicase activities than the rabbit eIF4H preparation used, and therefore was the better candidate for more detailed studies.
Since heIF4H was functionally identical to rabbit eIF4H, it was used in the ATP hydrolysis assay to determine how eIF4H affects the ability of eIF4A to bind RNA and utilize ATP. It was found that heIF4H decreased the K act of eIF4A for poly(A) by a factor of 2, which suggests that eIF4H has only a mild effect on the affinity of eIF4A for RNA. This is also supported by previous studies that report eIF4H does not enhance the ATP-dependent binding of RNA by eIF4A in nitrocellulose filter binding assays (4). When heIF4H was added to similar experiments that measured the K m and V max of eIF4A for ATP, it had no effect on either the K m or V max of eIF4A for ATP. FIG. 6. Northern blot analysis of eIF4H and eIF4B RNA expression in human tissues. A, a human multiple tissue blot containing 2 g/lane of poly(A) ϩ RNA from different tissues was probed for eIF4H, eIF4B, and ␤-actin mRNA with their specific cDNA probe (designated to the left of the corresponding autoradiograph). The migration of the 2,000-and 1,800-nucleotide muscle-specific forms of ␤-actin are marked to the right of the corresponding autoradiograph. B, graphic representation of eIF4H (white bars), eIF4B (black bars), and 2,000-nucleotide ␤-actin (hatched bars) mRNA expression relative to that in liver tissue. Bands were quantitated by densitometry and standardized to that in liver tissue on the same blot where the level of expression in liver tissue was set to a value of 1.0 (arbitrary units). The ␤-actin signal in skeletal muscle was not determined due to the intense signal from the 1,800nucleotide muscle-specific form. The K m of eIF4A for ATP measured in these studies, 290 M, was different than that reported previously in similar assays (80 M) (19). This can be explained by the fact that previous experiments were performed using subsaturating levels of poly(A) (16 M, equal to the K act of eIF4A), while those performed here used saturating amounts (100 M). The K act value obtained here was similar to that obtained by Lorsch et al. (23) of 330 M under similar conditions. This would also explain why previous studies showed eIF4B to increase the V max by 2-fold (19), since experiments including eIF4B were done using saturating amounts of poly(A) (K act of eIF4A decreases to 60 nM in presence of eIF4B) and those with eIF4A only were not.
Despite the minimal effect seen by heIF4H in the ATPase assays, heIF4H appeared to enhance the RNA helicase activity of eIF4A to a much greater degree. To better understand how heIF4H enhanced this activity, the initial rate of unwinding by eIF4A with and without heIF4H was measured using different RNA duplexes, and the ln(initial rate of unwinding) for each of these was then plotted against the ⌬G values of the RNA duplexes. Results showed that heIF4H increased the initial rate of unwinding by eIF4A for all duplexes, and that the slope of the line in Fig. 5C (ln(initial rate of unwinding) versus ⌬G) decreased upon addition of heIF4H. By examining how the addition of heIF4H affected the slope of the line in this experiment, it may be possible to determine the mechanism by which eIF4H stimulates the activity of eIF4A.
If the addition of eIF4H increased only the affinity of eIF4A for the RNA substrate, this would increase the number of active eIF4A-duplex complexes and would be analogous to increasing the concentration of eIF4A. It would then be expected that the resulting line from the ln(initial rate of unwinding) versus ⌬G plot would be parallel to the eIF4A only line, but with higher initial rate values at each ⌬G value (refer to Fig.  5C comparing 0.2 M eIF4A calculated to 0.4 M experimental). This was not seen when heIF4H was added to the reaction. Therefore, the increase in helicase activity by eIF4H is not merely due to an increase in the affinity of eIF4A for substrate, which is consistent with the results from ATPase studies (heIF4H decreases K act only 2-fold). If the addition of eIF4H increases the activity of eIF4A by making it more processive (hydrolyze more than one ATP and/or actively unwind more RNA base pairs per RNA binding event), the initial rate would be expected to increase at all ⌬G values, but the linear relationship might be different. Fig. 5C shows that the addition of heIF4H lead to a decrease in the slope of the line by allowing eIF4A to efficiently unwind more stable substrates, and unwind more RNA base pairs in the RNA duplex before dissociating. This is also seen in the ability of heIF4H to increase the maximum amount of duplex unwound over the course of the reaction (refer to Fig. 4B). These results all suggest that eIF4H increases the processivity of eIF4A.
Previous studies have shown that eIF4A dissociates from its substrate faster than it can hydrolyze ATP, implying that the protein cannot function by itself as a processive helicase (23). Related studies have shown that eIF4A goes through a cycle of conformational changes upon substrate binding, ATP hydrolysis, and product dissociation (24), and that these conformational changes are believed to be the actual helicase "motor" of eIF4A that causes unwinding of the RNA duplexes. This mechanism has also been suggested for the NS3 RNA helicase from hepatitis C virus (25). Further support for this mechanism in eIF4A comes from recent crystallographic studies of the ATPase domain of eIF4A, which was shown to have structure nearly identical to the corresponding domain of NS3 (26). Since eIF4H has only a mild effect on the affinity of eIF4A for mRNA, and does not increase the catalytic step of ATP hydrolysis, it is possible that the ability of eIF4H to enhance the helicase activity of eIF4A may be through a stabilization of one or more of the conformational changes necessary for eIF4A to unwind RNA duplexes. This would increase the processivity of eIF4A and allow for more unwinding by eIF4A before product dissociation.
Although eIF4H and eIF4B have similar functions in stimulating translation initiation, one can notice significant differences in their mechanisms from results of this and previous studies. First, eIF4B functions as a dimer while eIF4H functions as a monomer. Second, eIF4B has a greater ability to stimulate the ATPase activity of eIF4A than eIF4H. Using equimolar amounts, the eIF4B dimer stimulates the ATPase activity of eIF4A almost 5 times more than eIF4H (4). Third, eIF4B has been shown to greatly increase the affinity of eIF4A for RNA (11,19,27), while eIF4H has only a mild effect on the affinity. This third difference could be due to the fact that eIF4B contains a second arginine-rich RNA-binding domain in addition to the RRM in its C terminus that is essential for RNA binding of eIF4B (6,28). This additional RNA-binding domain, which is not present in eIF4H, may play a role in the increased affinity of eIF4A for RNA in the presence of eIF4B. Fourth, results from Northern blots show that eIF4B and eIF4H mRNA are expressed at different levels in varying tissue types. Most interesting is the fact that expression of eIF4B mRNA is elevated relative to eIF4H mRNA in one tissue type, while the converse is true in another tissue type. Quantitation of protein levels within these same tissues would need to be performed to confirm these results. The significance of these differences is not apparent at this time, but it does suggest that there is tissue-specific expression of eIF4B and eIF4H, and that the two initiation factors may have different effects on varying mRNAs.
With large quantities of highly pure heIF4H, it was possible for the first time to perform structural analysis experiments. In this report, the secondary structure of heIF4H was analyzed using circular dichroism (CD), and compared with what was predicted from the amino acid sequence. Previous computer analysis of the protein sequence using the PHD structural prediction program predicted eIF4H to be an ␣-␤ protein (4) with 22.4% ␣-helix and 7.0% ␤-sheet. Eleven other secondary structure prediction programs gave differing results, and all 12 programs together provide predictions ranging from 9.2 to 31.6% ␣-helix and 5.7-38.6% ␤-sheet. Actual analysis of heIF4H showed that the protein is composed mostly of ␤-sheet and random coil (50 and 45%, respectively), which is quite different than what was predicted by all programs used.
It is known that there is an RRM in eIF4H (4) that delineates approximately 36% of the amino acid sequence of heIF4H. When the structures of other RRM-containing proteins were examined, it was found that RRMs typically have a ␤␣␤-␤␣␤ structure with the RNP-2 and RNP-1 domains found in structurally adjacent ␤-strands (␤ 1 and ␤ 3 , respectively) (29 -34). Two of the programs used (GOR I and SOPM) predicted a ␤␣␤-␤␣␤ structure in the RRM of eIF4H, and all other programs predicted ␤-sheet and ␣-helix in at least three of the six corresponding sections, suggesting that the RRM of eIF4H may exhibit typical secondary and tertiary structure. Two classes of RRMs have been described where one class has been shown to contain the canonical ␤␣␤-␤␣␤ structure and interacts well with RNA, while the second class contains less ␣-helix and does not bind RNA (33,35). Since eIF4H binds weakly to RNA (4) and has a low content of ␣-helix (as determined by CD) the RRM of eIF4H may belong to this second class of RRMs.
Although heIF4H has a high amount of random coil, this does not suggest that the protein is unstructured, as is confirmed by denaturation studies. It might be possible that part of eIF4H is unstructured when not complexed with other initiation factors or RNA. This may explain why eIF4H is unstable when stored in solutions containing less than 25% glycerol (4). Other proteins are known to be partially unstructured until they interact with other proteins or ligands (36 -43). An example is also found in eukaryotic protein translation where the eIF4E-binding proteins, 4E-BP1 and 4E-BP2, are completely unstructured in solution by themselves, but can inhibit translation by binding to eIF4E, which induces structural changes in the 4E-BPs (44,45). Since eIF4H has activity only upon interacting with other translation initiation factors and mRNA, further structural studies may reveal changes in eIF4H protein conformation upon interaction with these other components.
In conclusion, this study provides a method for purifying large amounts of active human eIF4H from E. coli, and begins to describe how eIF4H functions in the initiation of translation through more detailed enzymatic analysis. Studies have shown that eIF4H works in translation initiation during the steps of mRNA recognition and utilization by stimulating the ATPase and helicase activities of eIF4A, and enhancing the translational activities of the other initiation factors eIF4B and eIF4F. eIF4H increases the affinity of eIF4A for poly(A) by 2-fold, yet has no effect on the affinity of eIF4A for ATP, and results suggest that eIF4H stimulates the helicase activity of eIF4A by making it a more processive helicase. From helicase studies it was also determined that eIF4H and eIF4A function optimally under these conditions when both proteins are present in a 1:1 molar ratio, and one might extrapolate that this could also be true in vivo. Since its affinity for RNA is weak, eIF4H most likely functions in translation initiation through protein-protein interactions with the other eIF4 initiation factors. These interactions might stabilize the conformation of eIF4H, as well as conformational changes in eIF4A that occur upon RNA binding and ATP hydrolysis that may promote RNA helicase activity.