Assessing the components of the eIF3 complex and their phosphorylation status

Abstract

The eukaryotic initiation factor 3 (eIF3) is an essential, highly conserved multi-protein complex that is a key component in the recruitment and assembly of the translation initiation machinery. To better understand the molecular function of eIF3, we examined its composition and phosphorylation status in Saccharomyces cerevisiae. The yeast eIF3 complex contains five core components: Rpg1, Nip1, Prt1, Tif34, and Tif35. 2-D LC-MS/MS mass spectrometry analysis of affinity purified eIF3 complexes showed that several other initiation factors (Fun12, Tif5, Sui3, Pab1, Hcr1, and Sui1) and the casein kinase 2 complex (CK2) co-purify. In vivo metabolic labeling of proteins with ³²P revealed that Nip1 is phosphorylated. Using 2-D LC-MS/MS analysis of eIF3 complexes, we identified Prt1 phosphopeptides indicating phosphorylation at S22 and T707 and a Tif5 phosphopeptide with phosphorylation at T191. Additionally, we used immobilized metal affinity chromatography (IMAC) to enrich for eIF3 phosphopeptides and tandem mass spectrometry to identify phosphorylated residues. We found that three CK2 consensus sequences in Nip1 are phosphorylated: S98, S99, and S103. Using in vitro kinase assays, we showed that CK2 phophorylates Nip1 and that a synthetic Nip1 peptide containing S98, S99, and S103 competitively inhibits the reaction. Replacement of these three Nip1 serines with alanines causes a slow growth phenotype.

1 Introduction

Translation initiation is the ribosome-mediated pairing of the methionyl initiator tRNA (met-tRNAi) anti-codon with the AUG start codon of the mRNA. This essential and highly conserved process is mediated by interactions of numerous multi-subunit eukaryotic initiation factors (eIFs) with the ribosome, mRNA, and tRNAs. In the first step, a ternary complex consisting of eIF2 bound to GTP and met-tRNAi is recruited to the 40S ribosomal subunit to form the 43S pre-initiation complex ^1,2. In vitro experiments suggest that this interaction is promoted by eIF1, eIF1A, and eIF3 ^3–9. In the next step, eIF4F, which is composed of eIF4A (RNA helicase), eIF4E (5´ m⁷GTP cap binding protein), and eIF4G (scaffold protein) ¹⁰ facilitates interaction of the 43S pre-initiation complex with the 5’ cap structure of the mRNA producing the 48S pre-initiation complex. Efficient formation of the 48S pre-initiation complex is also dependent on eIF3 ¹¹. The 40S complex scans the mRNA from the 5’ end for the start codon in a process mediated by eIF1 and eIF1A ¹². Once the met-tRNAi anti-codon pairs with the AUG codon, eIF2-bound GTP is hydrolyzed in a reaction stimulated by eIF5 ¹³. The GTP hydrolysis triggers the release of the bound eIFs and the joining of the 60S subunit to create the 80S ribosome ¹⁴. At this point, the 80S complex accepts the next aminoacyl-tRNA in the A site of the ribosome and the process of translation elongation begins.

A key player in the assembly and function of the 48S pre-initiation complex is the multi-protein eIF3 complex. While the mammalian eIF3 complex contains thirteen subunits, much of what is known about the mechanisms of eIF3 function have been characterized in studies of the five-subunit eIF3 complex in S. cerevisiae ¹¹. The yeast eIF3 subunits Rpg1, Nip1, Prt1, Tif34, and Tif35 are homologs of the human eIF3 subunits eIF3a, eIF3c, eIF3b, eIF3i, and eIF3g, respectively ¹⁵. Yeast Hcr1 is homologous to human eIF3j, and appears to associate with the core comples of eIF3 at non-stoichiometric levels ¹⁶. The eIF3 complex, as well as eIF1, eIF5, and the eIF2 ternary complex, have been identified as part of a multifactor complex (MFC) that mediates the assembly of the 48S pre-initiation complex ^4,11. Within eIF3, Rpg1 and Hcr1 interact with Prt1 via an N-terminal RNA recognition motif (RRM) ¹⁶, which mediates protein-protein interactions despite the name. These interactions are crucial for the structural integrity of the eIF3 complex and its stable association with the MFC and 40S ribosome complex ¹⁶. Nip1 and Prt1 play key roles in both the assembly and maintenance of the 48S pre-initiation complex ¹¹. Nip1 interacts directly with eIF1 and eIF5, suggesting that it coordinates both AUG start codon recognition (in conjunction with eIF1, which interacts with the met-tRNAi) and eIF2-GTP hydrolysis (in conjunction with eIF5) ^6,17. The smallest of the eIF3 subunits, Tif34 and Tif35, have recently been shown to promote the linear scanning of mRNA ¹⁸.

Protein kinases are key regulators of cell function and are believed to directly affect translation efficiency ^19,20. By phosphorylating a protein, kinases can alter the substrate’s activity, interactions, and stability. The functions of several eIF complexes are regulated by phosphorylation. The best characterized of these phosphorylation events involve regulation of eIF2 and eIF4E ²¹. eIF2 is bound to GDP in an inactive state at the end of each round of translation initiation. To reconstitute a functional ternary complex for a new round of translation initiation, eIF2B catalyzes the exchange of GDP for GTP ². Phosphorylation of the eIF2 subunit eIF2α effectively blocks eIF2B-mediated GDP-GTP exchange, thus inhibiting protein synthesis ²². Phosphorylation also regulates association of eIF4E with the eIF4F mRNA-cap binding complex. This interaction is required for cap-dependent mRNA translation initiation and is inhibited by the eIF4E-binding protein (4E-BP) ¹⁰. Phosphorylation of 4E-BP attenuates this inhibition ²³. In its hypophorphorylated state, 4E-BP binds to eIF4E and prevents its association with eIF4F ²⁴. In its hyperphosphoryated state, 4E-BP binds poorly to eIF4E. 4E-BP phosphorylation is stimulated by many different external stimuli including growth factors, hormones, and mitogens ²⁴. Finally, eIF4E is also phosphorylated. Phosphorylation of S209 is stimulated by different mitogen-activated protein kinase pathways and correlates with increased translation rates ^23,25,26.

We have used mass spectrometry techniques to identify the components of eIF3 and specific residues that are phosphorylated in vivo. Mass spectrometry has become the method of choice for identifying the composition of a protein complex ²⁷. However, although protein identification by mass spectrometry has become routine, protein phosphorylation analysis remains a challenging problem. Low phosphorylation stoichiometry, heterogeneous phosphorylation sites, and low protein abundance contribute to the difficulty of phosphoprotein analysis ²⁸. In addition, phosphopeptides are generally difficult to analyze by mass spectrometry for a number of reasons ²⁹. Reduced ionization efficiency in positive ionization mode and suppression by nonphosphorylated peptides exacerbate the problems. To improve the efficiency of phosphopeptide analysis, various strategies have been developed to enrich for phosphopeptides including strong cation exchange chromatography (SCX), immunoaffinity capture using antiphosphotyrosine antibodies, and immobilized metal affinity chromatography (IMAC) ^30–32. IMAC was initially plagued by the nonspecific retention of peptides rich in acidic amino acids, but improvements have increased both the specificity and sensitivity of IMAC ^33–35. The direct coupling of IMAC with high sensitivity reversed-phase liquid chromatography and tandem mass spectrometry has enabled us to identify phosphopeptides starting with small amounts of material.

Using tandem mass spectrometry analysis of tandem affinity-purified (TAP) eIF3 complexes, we detected the core yeast components Rpg1, Nip1, Prt1, Tif34, and Tif35. Several other initiation factors co-purified, including Fun12, Tif5, Sui3, Pab1, Hcr1, Sui1, and the casein kinase 2 complex (CK2). Using phosphorylation assays and IMAC enrichment prior to mass spectrometry, we found that Prt1, Nip1, and Tif5 are phosphorylated in vivo and identified three specific sites on Nip1, two on Prt1 and one on Tif5. We used an in vitro kinase assay to test individual kinases for their ability to phosphorylate eIF3 in vitro. CK2, which associates with the eIF3 complex, phosphorylates the eIF3 component Nip1. A synthetic NIP1 peptide containing the three serine residues identified as phosphorylated inhibits in vitro phosphorylation of Nip1 by CK2. Mutagenesis of these three phosphorylation sites in Nip1 indicates that they are important for a normal growth rate.

2 Materials and Methods

Plasmids

Plasmid pNip1entry (ScCD00008839) containing the wild-type Nip1 gene in a Gateway entry vector was obtained from the Harvard Institute of Proteomics (Boston, MA) and sequenced verified. It was recombined with the destination clone pAG415GPD-ccdB-TAP from Addgene (Cambridge, MA) as previously described ³⁶ to generate pNip1a. The identified sites of phosphorylation, S98, S99 and S101, were converted to alanines by site directed mutagenesis ³⁷ using the primers nip1mutfwd (GCTCTAACTATGATGCCGCTGATGAAGAAGCCGATGAAGAAG) and nip1mutrev (CTTCTTCATCGGCTTCTTCATCAGCGGCATCATAGTTAGAGC) to generate pNip1b. The endogenous stop codon was removed from both pNip1a and pNip1b with the primers nip1stopfwd (CCATCAAATCGTCGTGCGACCCAGC) and nip1stoprev (GCTGGGTCGCACGACGATTTGATGG) to generate pNip1wtTAP and pNip1mutTAP. All clones were sequence verified. pNip1comp, expressing Nip1 with a URA3 selectable marker ¹⁷, was obtained from Dr. Alan Hinnebusch’s laboratory.

Yeast strains

TAP-tagged yeast strains used in this study have been previously described ³⁸. We tested pNip1a (wild-type) and pNip1b (with three phosphorylation sites replaced by alanines) by transforming the strain H2997 (MATa, ura3-52, leu2-3, 112, his4-303, (AUU)SUI1, nip1Δ [pNIP1+(B3940)] ¹⁷, obtained from Dr. Alan Hinnebusch and selecting for loss of the parent strain’s plasmid with 5-FOA. Strains containing pNip1wtTAP and pNip1mutTAP were similarly generated. pNip1comp was used to transform the pNip1b strain, generating a complemented strain.

Purification and Identification of eIF3 Components

The eIF3 complex was purified using a previously described protocol from TAP-tagged yeast strains grown to early stationary phase (O.D.₆₀₀ 2–4) in YPD medium ^39,40. For identifying the protein composition of eIF3, 2 L cultures of strains expressing TAP-tagged alleles of RPG1, NIP1, PRT1, and TIF5 were analyzed. BY4741, the parental strain for the TAP clones, was used as a negative control since it lacks any affinity tagged gene. For analysis of the association of casein kinase 2 (CK2) with eIF3, a TAP-tagged allele of CKB2 was used for a similar purification. For validation experiments, TAP-tagged FUN12 and PAB1 strains were used. Purified proteins were identified using two different methodologies. In one approach, proteins were separated by 10% acrylamide SDS-PAGE. Stained bands were excised and in-gel trypsin digested, and purified proteins were identified using reversed-phase LC-MS/MS analysis ^39,41. In the second approach, components of eIF3 were identified directly in solution with Multidimensional Protein Identification Technology (MudPIT) using an LTQ quadrupole ion trap (Thermo Electron, Inc) ^39,42,43. Protein samples were reduced with 1/10 volume of 50 mM DTT at 65°C for 5 min, cysteines were alkylated with 1/10 volume of 100 mM iodoacetamide at 30°C for 30 min, and proteins were trypsinized with modified sequencing grade trypsin at ~25:1 subtrate:enzyme ratio (Promega, Madison, WI) at 37°C overnight. A fritless, microcapillary (100 µm-inner diameter) column was packed sequentially with the following: 9 cm of 5 µm C₁₈ reverse-phase packing material (Synergi 4 µ Hydro RP80a, Phenomenex), 3 cm of 5 µm strong cation exchange packing material (Partisphere SCX, Whatman) and 2 cm of C₁₈ reverse-phase packing material. The trypsin-digested samples were loaded directly onto the triphasic column equilibrated in 0.1% formic acid, 2% acetonitrile, which was then placed in-line with an LTQ linear ion trap mass spectrometer (Thermo, Inc.). An automated six-cycle multidimensional chromatographic separation was performed using buffer A (0.1% formic acid, 5% acetonitrile), buffer B (0.1% formic acid, 80% acetonitrile) and buffer C (0.1% formic acid, 5% acetonitrile, 500 mM ammonium acetate) at a flow rate of 300 nL/min. The first cycle was a 20-min isocratic flow of buffer B. Cycles 2–6 consisted of 3 min of buffer A, 2 min of 15–100% buffer C, 5 min of buffer A, followed by a 60-min linear gradient to 60% buffer B. In cycles 2–6, the percent of buffer C was increased incrementally (from 15, 30, 50, 70 to 100%) in each cycle. During the linear gradient, the eluting peptides were analyzed by one full MS scan (200–2000 m/z), followed by five MS/MS scans on the five most abundant ions detected in the full MS scan while operating under dynamic exclusion.

Mass Spectrometry Data Analysis

To process and analyze the mass spectrometry data, the program extractms2 was used to generate an ASCII peak list and identify +1 or multiply charged precursor ions from native *.RAW mass spectrometry data files (Jimmy Eng and John R. Yates III, unpublished). Tandem spectra were searched with no protease specificity using SEQUEST-PVM ⁴⁴ against the SGD yeast_orfs database containing 6,000 entries with a static modification of +57 on C (addition of a carbamidomethyl group). For multiply charged precursor ions (z ≥ +2), an independent search was performed on both the +2 and +3 mass of the parent ion. Data were processed and organized using the BIGCAT software analysis suite⁴⁵. A weighted scoring matrix was used to select the most likely charge state of multiply charged precursor ions^42,45. From the database search, tryptic peptide sequences with SEQUEST cross-correlation scores (Cn) ≥ 1.5 for +1 ions, ≥ 2 for +2 ions, and ≥ 2 for +3 ions were considered significant and used to create a list of identified proteins.

To validate candidate phosphopeptides, tandem mass spectra were manually evaluated to identify evidence of neutral loss of phosphoric acid caused by fragmentation of the precursor ion (98 (z=+1), 49 (z=+2), or 32.7 (z=+3) m/z units relative to the precursor). For all IMAC mass spectrometry data, the acquired mass spectrometry data were searched against the yeast ORF database using the Sequest algorithm assuming static modifications of +14 on D, E, and the C-termini, static modification of +57 on C, and variable modifications of +80 on S, T, and Y ⁴⁶. Our criteria in identifying phosphorylations were: 1) a series of at least 3 y or b ions with an appropriate +80 Dalton shift, and 2) ions indicating a neutral loss of phosphoric acid from the precursor ion. Annotated MS/MS spectra illustrating such evidence are presented in Figure 3 and Supplemental Figure 1.

Figure 3. Nip1 is phosphorylated at S98, S99, and S103.

An MS/MS spectrum generated after IMAC-enrichment of phosphopeptides from trypsin-digested eIF3 showing in vivo phosphorylation of a Nip1 peptide at S98, S99, and S103. The spectrum significantly matched the trypsin-generated peptide of Nip1 (Sequest Xcorr = 3.1). The spectrum shows the addition of 80 Da at S98, S99, and S103. The m/z values of the most prominent ions in the MS/MS spectrum correspond to the neutral loss of one and two phosphoric acids from the precursor ion of 1131.6 (+2).

A protein abundance factor (PAF) representing relative protein abundance, was calculated for each identified protein by normalizing the total number of non-redundant spectra that correlated significantly with each cognate protein to the molecular weight of the protein and multiplying by 10^{4 39}.

In Vivo ³²P Labeling of the eIF3 Complex

12 mL cultures of strains expressing TAP-Rpg1, TAP-Nip1, and TAP-Nip1S98,99,101A were grown in low-phosphate YPD at 30°C to an O.D.₆₀₀ of 1.0 ⁴⁷. The cultures were centrifuged, and the cells were resuspended in 5 mL low-phosphate YPD containing 5 mCi carrier-free ³²P-orthophosphate (ICN, Costa Mesa, CA). An unlabeled culture was grown in parallel. The cells were labeled for 30 min at 30°C. The cultures were centrifuged, and the cells were resuspended in 20 mL fresh low-phosphate YPD and incubated for 4 h at 30°C. Cells were harvested by centrifugation, washed twice with 10 mL ice-cold water and resuspended in 1 mL lysis buffer containing 10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% NP-40, and 1x Complete^R protease inhibitor (Roche Diagnostics Corp., Indianapolis, IN). Cells were lysed by vortexing with 500 µL 0.5 mm zirconia/silica beads (BioSpec Products, Inc., Bartlesville, OK) for 6 min with cycles of 30 sec of active vortexing followed by 30 sec off. The lysates were centrifuged at 20,000xg for 15 min to remove particulate material. The proteins were purified using the TAP method ^39,40 with the volumes reduced proportionally to the amount of cells isolated. Denatured proteins were separated by 10% acrylamide SDS-PAGE. ³²P labeled-proteins were detected by autoradiography. Densitometric measurements were quantified using Scion image software ⁴⁸.

Direct Mapping of Phosphorylated Sites on eIF3 Components with Tandem Mass Spectrometry

The eIF3 complex was purified from whole-cell lysates prepared from 4 L overnight cultures of the TAP-Rpg1 strain ^39,40. Purified complexes were digested directly in solution with sequencing grade trypsin (Promega, Madison, WI) or chymotrypsin (Roche Diagnostics Corp., Indianapolis, IN) ⁴³ and analyzed by MudPIT using an LTQ linear ion trap mass spectrometer (Thermo Electron, Inc).

Mapping eIF3 Phosphopeptides using IMAC Enrichment and Tandem Mass Spectrometry

The S. cerevisiae strain expressing a TAP-Rpg1 was grown overnight (O.D.₆₀₀ 2–4) in 16 L of YPD medium. eIF3 complexes were TAP-isolated from whole cell lysates ⁴⁰. The pH of the eluted eIF3 mixture was adjusted to 8.0 using 1 M Tris pH 8.0. Cysteine residues were reduced with DTT and alkylated with iodoacetamide. The proteins were digested with 20 µg of sequencing grade trypsin ⁴³ (Promega, Madison, WI). The tryptic peptide mixture was divided into 1 mL aliquots and lyophylized. To methyl esterify the peptides, a pre-mix of 1 mL anhydrous methanol and 50 µL thionly chloride (Sigma-Aldrich, St. Louis, MO) was added to the dried peptides. The mixture was sonicated for 15 min at room temperature and incubated at room temperature for 2 h. Esterified peptides were lyophilized to dryness.

An IMAC column was constructed using a fritted 20 cm long piece of 200 × 360 µm fused silica capillary (FSC). The column was packed with POROS MC 20 beads (Applied Biosystems, Foster City, CA) to a length of 15 cm. The column was rinsed with 100 mM EDTA (Sigma-Aldrich, St. Louis, MO), pH 8.5 and then with deionized water. The column was charged with 100 mM Iron(III) chloride solution (Sigma-Aldrich, St. Louis, MO). Finally, the IMAC column was rinsed with 0.1% acetic acid.

The dried, esterified peptides were resuspended in 180 µL of 33% methanol, 33% acetonitrile, and 0.034% acetic acid and loaded onto the IMAC column. Following loading, the column was rinsed with a solution of 25% acetonitrile, 100 mM NaCl, and 1% acetic acid.

A reversed-phase capture column was constructed from 20 cm of 100 × 360 µm FSC fritted at one end using an inline MicroFilter Assembly (Upchurch Scientific, Inc, Oak Harbor, WA). The RP-capture column was packed with POROS 10 R2 (Applied Biosystems, Foster City, CA) to a length of 10 cm. Using a 0.012 × 0.060 inch piece of Teflon tubing (Zeus, Orangeburg, SC), the RP-capture column was butt-jointed to the IMAC column. Phosphopeptides were eluted from the IMAC column onto the RP-capture column with 250 mM sodium phosphate, pH 8.0 solution.

The RP-capture column was rinsed overnight with 5% acetonitrile and 0.1% formic acid and attached to an analytical RP chromatography column (100 × 365 µm FSC with an integrated laser-pulled emitter tip packed with 10 cm of Synergi 4 µm RP80A (Phenomenex)) via the inline MicroFilter Assembly for LC-ESI-MS/MS analysis. Peptides were eluted using the following linear gradient: 0 min: 0% B, 120 min: 40% B, 140 min: 60% B, 160 min: 100% B at a flow rate of 400 nL/min (mobile phase A: 5% acetonitrile, 0.1% formic acid and mobile phase B: 80% acetonitrile, 0.1% formic acid).

Spectra were acquired with a LTQ linear ion trap mass spectrometer (Thermo Electron, Inc). During LC-MS/MS analysis, the mass spectrometer performed automated data-dependent acquisition with a full MS scan followed by three MS/MS scans on the most intense ions while operating under dynamic exclusion.

Kinase Purification and In Vitro Kinase Assays

Specific protein kinases were purified from S. cerevisiae strains expressing individual amino-terminally GST-tagged kinases ⁴⁹. The GST-tagged kinase strains were kindly provided by Dr. Mike Snyder. GST-tagged strains were cultured in 25 mL synthetic complete medium lacking uracil (SC - URA) with 2% raffinose at 30°C for 16 h. GST-kinase expression was induced by incubation in SC-URA with 2% galactose for an additional 2 h at 30°C. Cells were harvested by centrifugation and resuspended in 1 mL lysis buffer containing 10 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% NP-40, and 1x Complete^R protease inhibitor (Roche Diagnostics Corp., Indianapolis, IN). Cells were lysed using an equal volume of 0.5 mm zirconia/silica beads and a Mini-Beadbeater-8 (BioSpec Products, Inc., Bartlesville, OK). Lysates were centrifuged at 20,000xg for 15 min to remove particulate material. Supernatants were incubated with 100 µL glutathione-coupled agarose beads (Amersham Biosciences, Piscataway, NJ) for 1 h at 4°C. The beads were washed twice with 1 mL lysis buffer then twice with 1 mL buffer containing 50 mM HEPES pH 7.5, 100 mM NaCl, and 10% glycerol. The beads were then resuspended in 100 µL kinase buffer (25 mM HEPES, 25 mM β-glycerolphosphate, 25 mM MgCl₂, 2 mM DTT, 0.1 mM NaVO₃).

Phosphorylation reactions with the eIF3 complex were performed by incubating each purified kinase with 10 µCi [γ-³²P]ATP (ICN, Costa Mesa, CA) and 1 µg eIF3 complex (isolated by purification of TAP-Rpg1) in 30 µL kinase buffer for 1 h at 30°C. To eliminate endogenous kinase activity, eIF3 purifications were first denatured with 8 M urea, which was later diluted to 1 M prior to the kinase assays ⁴¹. Kinase reactions were terminated by the addition of Laemmli-SDS sample buffer and heating at 100 °C for 3 min. The kinase reaction products were separated by SDS-PAGE. The proteins were silver stained and phosphorylation was detected by autoradiography.

Inhibition of In Vitro Phosphorylation

A peptide inhibitor containing Nip1 phosphorylation sites (H2N-SSNYDSSDEEDSDDDGK-OH) was synthesized to mimic the tryptic peptide of Nip1 identified by mass spectrometry (New England Peptides Gardner, MA). A control peptide H2N-ISQAVAHAHAEINEAGR-OH that contains an identical number of amino acids but lacks CK2 consensus sequence sites for phosphorylation was synthesized (New England Peptides yes/no). The peptides were suspended at 1 mg/mL in 100 mM Tris, pH 8.0 to a concentration of 1 mg/mL. The CK2 inhibitor (E)-3-(2,3,4,5-Tetrabromophenyl)acrylic acid (TBCA) and the CK1 inhibitor 3-[(2,4,6-Trimethoxyphenyl)methylidenyl]-indolin-2-one (IC261) were obtained from Calbiochem (Gibbstown, NJ) and resuspended to 1 mg/mL in DMSO ^50,51. The peptides and kinase inhibitors were added to kinase reactions described above to a final concentration of 33 µg/mL.

Quantifying the Doubling time of Wild-type and Mutant of Nip1

We measured the growth rate of nip1 deletion mutants carrying either pNip1a (NIP1⁺), pNip1b (nip1-S98,99,101A) or pNIP1b plus pNip1comp (NIP1⁺). Single colonies from each strain were inoculated into YPD and grown overnight at 30°C. Cell density was calculated using a Countess automated cell counter from Invitrogen (Carlsbad, CA). 10,000 cells of each strain were transferred to fresh 6 mL cultures of YPD at 30°C. Aliquots from each strain were collected periodically, and cell density was calculated in triplicate until the cultures reached stationary phase. The triplicate density measurements were averaged. The doubling time (t_dub) for each strain was determined from the logarithmic growth phase of the cells using the following formula:

$${\mathrm{t}}_{\text{dub}}=\text{ln}2/((\text{ln}{\mathrm{D}}_2/{\mathrm{D}}_1)/\mathrm{\Delta }{\mathrm{t}}_{\text{min}})$$

Where D is the cell density and Δt_min is the change in time in min from D₁ to D₂. The experiment was repeated in triplicate and the resulting doubling times were averaged for each strain. A Student’s T-test (95% confidence interval) was performed on the data to determine its statistical significance.

3 Results

In Vivo Interaction of eIF3 Components

The goals of this study were to identify comprehensively the composition of the S. cerevisiae eIF3 complex and dissect the sites of in vivo phosphorylation. To this end, we first established an approach for high-quality purification of the eIF3 complex using epitope tags and tandem affinity purification (TAP). This approach enables the efficient recovery of proteins present at low cellular concentrations under native conditions ⁵². We used individually TAP-tagged yeast strains in which the fusion proteins are driven by native promoters, avoiding overexpression of the affinity tagged protein^38,53. A representative purification is shown for TAP-Rpg1 (Fig. 1A). eIF3 complexes were purified in duplicate from four different yeast strains, each expressing a different TAP-tagged protein (Rpg1, Nip1, Prt1, or Tif5) ¹⁵. To control for purification of non-specific proteins, three independent extracts were prepared and analyzed in parallel from an isogenic untagged yeast strain (BY4741). The isolated eIF3 complexes and control purifications were trypsin-digested in solution, and the proteins were identified using the 2-D LC-MS/MS mass spectrometry approach termed Multidimensional Protein Identification Technology (MudPIT) (Fig. 1B) ^42,54,55. To estimate the relative abundance of each protein, a Protein Abundance Factor (PAF) was calculated using the acquired mass spectrometry data ³⁹. A PAF is a label-free, semi-quantitative measure of a protein’s relative abundance and is based on the fact that a protein’s abundance is directly related to the frequency at which its peptides are selected for MS/MS analysis.

Figure 1. Purification and mass spectrometry identification of eIF3 components.

The eIF3 complex was isolated from S. cerevisiae by tandem affinity purification (TAP). Complexes and control extracts were prepared in duplicate or triplicate from strains expressing affinity tags specific for three core eIF3 components (Rpg1, Nip1, and Prt1) and the translation initiation factor Tif5, as well as an isogenic, untagged yeast strain. (A) Silver-stained SDS-PAGE separation of a TAP-Rpg1 purification. (B) MudPIT mass spectrometry analysis of the purifications from the four different TAP-tagged strains and the control strain (TAP-cntrl). The results from one of the replicate experiments are shown. The predicted molecular weights (MW) of the detected yeast proteins (rows) are shown in kDa. For each tagged strain and the control, protein abundance factors (PAFs) were calculated for the identified proteins (PAF = (the number of non-redundant spectra identifying the protein) / (MW of the protein) ×10⁴³⁹. The relative stoichiometry of each protein (in parentheses) was determined by normalizing its PAF to the PAF of the TAP-tagged protein targeted in the purification. (nd: not detected).

We routinely detected the five primary components of yeast eIF3 (Rpg1, Nip1, Prt1, Tif34, and Tif35) at greater than 75% coverage (Fig. 1B). The PAFs suggest that these proteins have equivalent stoichiometries. We also identified other translation initiation factors co-purifying in our eIF3 preparations, including the yeast orthologs of mammalian eIF5 (Tif5), eIF3j (Hcr1), eIF1 (Sui1), eIF2β (Sui3), eIF5B (Fun12), and PABC1 (Pab1) (Fig 1B). With the exception of Fun12 and Pab1, earlier studies have reported similar eIF3 interactions ¹¹. As expected, the PAFs values were lower for these six non-core, co-purifying translation initiation factors.

Interestingly, purifications based on different tagged targets yielded distinct subsets of co-purifying translation initiation factors, and the results were consistent in replicate purifications. We performed reciprocal purifications targeting the previously uncharacterized interactions of Fun12 and Pab1 and detected co-purifying eIF3 components but at lower abundances (see the TAP-Tif5 results in Fig. 1B, Table 1 and Supplemental Table 1). These results are consistent with a model in which these six translation initiation factors act as accessories in eIF3 function. Additionally, we consistently observed the yeast casein kinase 2 complex (CK2) copurifying with the TAP-Rpg1 isolation of eIF3 (Fig. 1B). In yeast, CK2 is a tetramer consisting of two catalytic subunits (Cka1 and Cka2) and two regulatory subunits (Ckb1 and Ckb2) ^56,57. However, the PAF values were relatively low, indicating that the CK2 subunits associate with eIF3 at substoichiometric levels, consistent with the transient nature of kinase-substrate interactions. The CK2 subunits were not detected in the controls or in a large number of independent TAP/MudPIT experiments using yeast strains with different TAP-tagged proteins (Link et al, manuscript in preparation), suggesting a specific interaction of CK2 with eIF3. We further investigated the physical association of CK2 with eIF3 by targeting TAP-Ckb2 for protein purification. LC-MS/MS analysis of purified TAP-Ckb2 confirmed that the isolate contains all components of CK2 as well as all five core subunits of eIF3. Control purifications from an untagged strain contained neither CK2 nor eIF3 (Table 3).

Table 1. Mass spectrometry identification of Fun12 and Pab1 interactions with eIF3.

Fun12 protein complexes was isolated from an S. cerevisiae strain expressing TAP-Fun12 and an isogenic, untagged yeast strain as a negative control. Mudpit mass spectrometry analysis of the purified complexes were performed in triplicate. PAF values were used to approximate protein abundances as in Fig. 1.

	Locus	TAP-Fun12	TAP-Pab1	negative control
Targeted Proteins
Fun12	YAL035W	5.1(1.0)		nda
Pab1	YER165W		12(1.0)	nd
eIF3 Core Proteins
Rpg1	YBR079C	0.18(0.035)	0.030(0.0025)	nd
Nip1	YMR309C	0.071(0.014)	0.036(0.0030)	nd
Prt1	YOR361C	0.038(0.0074)	nd	nd
Tif34	YMR146C	0.43(0.084)	nd	nd
Tif35	YDR429C	0.33(0.064)	nd	nd

^and = none detected

Table 3. Summary of Nip1 peptides identified by LC/MS/MS after IMAC-enrichment of phosphopeptides from trypsin-digested eIF3.

The table shows the precursor ion and the m/z of the most intense fragment ion (base peak) in the corresponding MS/MS spectrum. The observed neutral loss is given and corresponds to a loss of 98 Da from the precursor peptide. The predicted charge state of the precursor ion (Z) is listed. Sequest cross-correlation values (Cn) to the phosphopeptide sequences are shown. This value is a statistical measurement of how well the theoretical spectrum generated for the given peptide sequence corresponds to the experimental spectrum. The peptide sequences show the predicted upstream and downstream amino acids flanking the peptides.

Parent m/z	m/z of MS/MS base peak	Neutral Loss	Z	Cn	Peptide Sequence
1131.6	1081.4	50.2	+2	3.1	K.SSNYDpSpSDEEpSDEEDGK.K
796.8	764.1	32.7	+3	2.6	K.SSNYDpSpSDEEpSDEEDGKK.V
1090.5	1040.7	49.1	+2	2.8	K.SSNYDpSpSDEESDEEDGK.K
1151.1	1105.6	49.3	+2	2.2	K.SSNYDpSpSDEESDEEDGKK.V
770.1	737.0	33.1	+3	2.3	K.SSNYDpSpSDEESDEEDGKK.V
1090.3	1041.5	48.8	+2	2.3	K.SSNYDSpSDEEpSDEEDGK.K
1155.2	1105.1	50.1	+2	2.7	K.SSNYDSpSDEEpSDEEDGKK.V

The eIF3 complex was effectively purified using all four TAP-tagged eIF3 components, however the mass spectrometry results indicated that the TAP-Rpg1 strain produced the highest yield of both the eIF3 core components and other associated initiation factors. Therefore, from this point on eIF3 proteins were purified and analyzed using the TAP-Rpg1 strain.

Identification of Phosphorylated Components of eIF3

The observation that CK2 consistently copurifies with components of eIF3 and the important role of phosphorylation in a wide array of biological processes lead us to examine whether it can target phosphorylations on components of the eIF3 complex. We metabolically labeled the TAP-Rpg1 and an unlabeled strain with ³²P-orthophosphate at early stationary phase in rich media. eIF3 was purified, the protein components were separated by SDS-PAGE, and phosphorylation was detected by autoradiography (Fig. 2). One prominent band was observed that corresponded to the molecular weight of Nip1. The identity of the phosphorylated protein was confirmed to be Nip1 by mass spectrometry analysis of the corresponding band excised from the unlabeled eIF3 complex analyzed in parallel.

Figure 2. In vivo phosphorylation of the yeast eIF3 complex.

(A) SDS-PAGE separation and silver staining of eIF3 components isolated from a TAP-Rpg1 strain following in vivo ³²P metabolic labeling. In-gel trypsin digestion and LC/MS/MS were used to identify the bands (B) The corresponding autoradiograph indicates that Nip1 is phosphorylated in vivo. (C) Negative control lane using a TAP purification from an untagged strain metabolically labeled with ³²P.

Mass spectrometry Identification of In Vivo Prt1 and Tif5 Phosphorylation Sites

Mass spectrometry techniques can be used to pinpoint exact sites of protein modification, including phosphorylation. Our initial approach used MudPIT to identify eIF3 phosphopeptides directly from complexes purified from the TAP-Rpg1 strain. This direct approach avoided sample loss associated with strategies to enrich for phosphopepetides. TAP-isolated eIF3 samples were digested separately with either trypsin or chymotrypsin and analyzed using MudPIT. We used two different proteolytic enzymes to maximize peptide coverage and to facilitate identification of candidate phosphorylation sites on multiple overlapping peptides ⁵⁸.

Phosphorylations present several hallmarks in mass spectrometry analysis. First, the mass of a phosphorylated residue is increased by 80 Da. Second, when peptides containing phosphorylated serine or threonine residues are fragmented by resonance excitation in an ion trap they commonly undergo a gas-phase β-elimination reaction, resulting in the neutral loss of phosphoric acid (−H₃PO₄ or 98 Da) ²⁹. Depending on the charge of the precursor ion, neutral loss of phosphoric acid appears as a decrease of 98 (z=+1), 49 (z=+2), or 32.7 (z=+3) m/z units from the precursor. Finally, because a phosphorylated residue increases the net negativity of the peptide in solution, it is expected to have a shorter retention time during SCX chromatography compared to the unmodified peptide ³⁰. Our criteria in identifying phosphorylations were: 1) a series of at least 3 y or b ions with an appropriate +80 dalton shift for the addition of a phosphate group, and 2) ions indicating a neutral loss of 98 Da from the precursor ion

To identify candidate phosphorylation sites we used the Sequest algorithm to search for additions of 80 Da at serine, threonine, and tyrosine residues ⁵⁹. We anticipated that the in vivo phosphorylation would be incomplete (< 50%), and therefore we expected to identify the corresponding unmodified peptide as well. Once candidate sites were identified, we manually inspected the data for evidence of neutral loss events indicating phosphorylation. In this way, we identified two phosphorylation sites in Prt1 (Supplemental Figure 1). Phosphorylation of a high stringency CK2 consensus site at S22 (CK2 consensus sequence: PVDDIDFpSDLEEQYK) was identified in two independent trypsin and chymotrypsin experiments (http://scansite.mit.edu) (Supplemental Data). Phosphorylation of a low stringency CK2 consensus site at T707 (CK2 consensus sequence: DASSDDFTpTIEEIVEE) was identified following trypsin digestion, but not using chymotrypsin (Supplemental Data). The corresponding unmodified versions of these peptides were also identified (data not shown). As predicted, analysis of the SCX fractions in the MudPIT runs showed the phosphorylated peptides eluted first followed by the unmodified peptides (data not shown). Additionally, a single Tif5 phosphopeptide was detected that contained a phosphorylated T191 (SQNAPSDGTGSSpTPQHHDEDEDELSR). However, the site does not fit the CK2 consensus sequence. To our surprise, we did not detect phosphopeptides from Nip1.

Our results using metabolic labeling and mass spectrometry illustrate the need for a multifaceted approach to comprehensively identify phosphorylations. ³²P-metabolic labeling revealed that Nip1 is phosphorylated. Mass spectrometry analysis of eIF3 complexes identified specific sites of phosphorylation in Prt1 and Tif5, but not Nip1.

Mass spectrometry Identification of In Vivo Phosphorylation of Nip1 at S98, S99, and S103 using IMAC phosphopeptide enrichment

In considering our failure to identify Nip1 phosphopeptides in our direct mass spectrometry analysis, we speculated that either Nip1 phosphopeptides were not retained in either the 2-D LC’s SCX or RP chromatography steps or that the phosphopeptide ion signals were being repressed by co-eluting nonphosphorylated peptides. Therefore, we enriched for phosphorylated peptides from digested eIF3 complexes using an optimized immobilized metal affinity chromatography (IMAC) protocol ³⁵. Purified eIF3 complexes were digested with trypsin, and the peptides were converted to methyl esters. Phosphorylated peptides were enriched by Fe(III)-charged IMAC prior to LC-MS/MS analysis. Using this approach and the criteria described earlier for validating the MS/MS spectra, we detectedmultiple Nip1 peptides that were phoshorylated at S98, S99, and S103 (Fig. 3 and Table 3). All three sites are predicted to be high stringency substrates for CK2 (CK2 consensus sequence: LKSSNYDpSpSDEEpSDEEDGKK) ⁶⁰. S99 was phosphorylated in all the Nip1 phosphopeptides that we detected, suggesting constitutive phosphorylation, while the phosphorylation of S98 and S103 was variable (Table 1). The corresponding unmodified peptides were not detected in our LC-MS/MS analysis of unenriched samples. Thus the IMAC enrichment enabled us to analyze these particular peptides. However, the IMAC approach did not allow us to detect the singly modified peptides previously identified in this study for Prt1 and Tif5. This is a commonly cited limitation of the IMAC protocol in that it preferentially binds multiply phosphorylated peptides ⁶¹.

Casein Kinase 2 Phosphorylates Nip1, Prt1, and Tif5 In Vitro

To test whether CK2 directly phosphorylates eIF3 components, we used an in vitro kinase assay. We purified GST-tagged Cka1 and Ckb1 as well as eight randomly selected serine/threonine yeast protein kinases with GST tags as controls. Isolation of the kinases was inferred from observation of SDS-PAGE bands with appropriate molecular sizes. To identify autophosphorylation, control reactions with no eIF3 substrate were performed for each kinase. The Cka1 catalytic subunit significantly phosphorylated eIF3 components above background (Fig. 4). The CK2 regulatory subunit Ckb1 phosphorylated eIF3 to a lesser extent. The kinase activity of the Ckb1 preparation suggests that catalytic components of the CK2 complex co-purified with this regulatory subunit. The results of the Yck1 assays are ambiguous due to apparent Yck1 auto-phosphorylation. None of the other kinases had any effect, although we cannot say for certain they are active as no autophosphorylation was detected. Yet it is clear that Cka1 has a striking ability to phosphorylate eIF3. Following this positive result, we confirmed the identity of the purified Cka1 using LC-MS/MS analysis (data not shown).

Figure 4. In vitro kinase screen identifies CK2 as an eIF3 kinase.

GST-tagged Cka1, Ckb1, and eight control kinases were purified and incubated with [γ-³²P]ATP plus purified eIF3 treated with urea to reduce background kinase activity. Reaction products were separated on a 10% SDS-polyacrylamide gel and subjected to autoradiography. In the far right lane, purified eIF3 without urea treatment shows significant endogenous kinase activity. Arrows highlight phosphorylation of eIF3 components above background levels by the Cka1 catalytic subunit and to a lesser extent the Ckb1 regulatory subunit.

Analysis of the eIF3-Cka1 reaction products allowed us to determine which eIF3 components are phosphorylated by Cka1. SDS-PAGE showed the major phosphorylated product is in the size range of Nip1 and Prt1 (Fig. 5A). A gel with a lower percentage of acrylamide allowed separation of Nip1 and Prt1 and showed that both proteins are Cka1 substrates, with Nip1 the predominant target (Fig. 5B). Mass spectrometry analysis of these two bands confirmed that only Nip1 and Prt1 were present (data not shown). Finally, we also tested inhibitors of specific kinases. The addition of IC261, a CK1 inhibitor, had no affect on phosphorylation of Nip1. Conversely, addition of TBCA, a CK2 inhibitor, abolished the Nip1 signal (Fig. 6).

Figure 5. CK2 phosphorylates Nip1 and Prt1 in vitro.

SDS-PAGE separation was used to identify eIF3 components phosphorylated in in vitro kinase reactions with GST-tagged Cka1. (A) Separation on a 10% gel indicates that CK2 phosphorylates Nip1 and/or Prt1. (B) Separation on a 7.5 % gel resolves the most prominent phosphorylated band in (A) into a doublet and shows both Nip1 and Prt1 are phosphorylated by Cka1. Nip1 appears to be more prominently ³²P-labeled compared to Prt1.

Figure 6. Synthetic Nip1 peptide inhibits the in vitro phosphorylation of Nip1.

A peptide was synthesized that mimics the unmodified tryptic peptide of Nip1 identified as phosphorylated by mass spectrometry. Separation on a 4–12% gradient gel for an extended time allows for better resolution of the Nip1-Prt1 region of the gel. The ability of this peptide mimic to inhibit in vitro phosphorylation is shown (lane 2). A control peptide added to the kinase assay indicates that this inhibition is specific to the Nip1 peptide (lane 3). Addition of a CK2 inhibitor, TBCA, abolishes the in vitro phosphorylation of Nip1 (lane 5). An unrelated kinase inhibitor, IC261, does not affect the in vitro phosphorylation of Nip1 (lane 4).

Phosphorylation of Nip1 at Specific Sites

To test whether CK2 targets Nip1 specifically at S98, S99, and S103, the sites identified by mass spectrometry, a synthetic peptide with a sequence identical to the tryptic fragment was generated and added to the in vitro kinase assays. Addition of the competitor peptide reduced phosphorylation of Nip1 (Fig. 6, compare lanes 1 and 2), indicating that these specific sites identified on Nip1 are phosphorylated by CK2 in vitro. The residual phosphorylation of Nip1 in the lane with the peptide mimic may result from either incomplete competition or CK2 phosphorylation of Nip1 at other site(s). The addition of a control peptide of unrelated sequence had no effect on Nip1 phosphorylation (Fig 6, lane 3).

Phosphorylation of Nip1 at Additional Sites

The incomplete inhibition of in vitro Nip1 phosphorylation by the peptide containing S98, S99, and S103 suggested that there may be additional phosphorylation sites on Nip1. To address this possibility, we examined in vivo phosphorylation of a mutant version of Nip1 lacking these three CK2 sites. Using a TAP-tagged Nip1 plasmid, we replaced S98, S99, and S103 with alanines. Strains with the wild-type and Nip1S98,99,101A mutant plasmids in a nip1 null background were metabolically labeled with ³²P-orthophosphate. Following TAP purification of Nip1, autoradiography showed a considerable level of phosphorylation of Nip1S98,99,101A (Fig. 7A). The signal for the mutant strain represents a 41% reduction compared to the wild-type as determined by densitometry (Fig. 7A), consistent with the phosphorylation of other Nip1 sites.

Figure 7. Mutation of S98, S99 and S103 in Nip1 reduces in vivo phosphorylation and growth rate.

nip1 null mutants containing plasmids expressing either wild-type Nip1 or a Nip1 mutant protein in which S98, S99 and S103 were replaced by alanines were examined. (A) SDS-PAGE separation of TAP-Nip1 following in vivo ³²P metabolic labeling. The S98,S99, S103A mutation results in a 41% reduction of the mutant signal relative to wild-type based on densitometry using values normalized for the α-TAP loading control. As a control, western blotting with an α-TAP antibody shows equivalent amounts of TAP-Nip1 in each lane. (B) Doubling time for each strain (wild-type and mutant) was determined from the logarithmic growth phase of the cells in triplicate. Comparison of doubling times reveals that the strains harboring the mutation have a doubling time increased by 33%. A one-tailed t-test reveals a p-value between the wild-type and mutant doubling times of 0.00687. This defect is rescued with a wild-type copy of Nip1 in the mutant strain.

The Biological Significance of the Nip1 Phosphorylation at S98, S99, and S103

To test the functional importance of the S98, S99, and S103 phosphorylation sites, we closely examined a Nip1S98,99,101A mutant strain. We compared yeast strains with either wild-type Nip1 or Nip1S98,99,101A expressed from low-copy-number plasmids in a nip1 null background. Following a multitude of assays including drug selection, temperature sensitivity, polysome profiling, ³⁵S incorporation, and selective growth conditions, the most striking difference observed between the mutant and wild-type strains was a marked increase in doubling time for the Nip1S98,99,101A mutant (Fig. 7B). The wild-type strain doubled every 111 ± 12 min, within the normal range for S. cerevisiae. However, the Nip1S98,99,101A mutant strain doubled every 148 min ± 10 min, a 33% increase (p-value = 0.00687). Complementing the mutant strain with a wild-type Nip1 plasmid results in a restoration of the doubling time to normal levels. (112 min ±8 min).

4 Discussion

In this study, we have used variouis mass spectrometry approaches to study the molecular composition of the highly conserved translation initiation factor eIF3 in S. cerevisiae. We used tandem affinity purification to isolate eIF3 complexes from tagged yeast strains and direct LC-MS/MS analysis of the complexes to identify core components of eIF3 as well as more loosely associated factors. To optimize our purification strategy, we affinity purified eIF3 based on three different tagged core components as well as the associated factor Tif5. The TAP-Rpg1 and TAP-Prt1 purifications resulted in the highest yield of eIF3 subunits. This finding is consistent with the previously described roles of Rpg1 and Prt1 in scaffolding and structural maintenance of the core eIF3 complex and in mediating interactions with Hcr1, eIF1, and the eIF2 ternary complex ^3,4,16.

We confirmed that the yeast eIF3 complex is composed of five conserved core components: Rpg1 (eIF3a), Nip1 (eIF3c), Prt1 (eIF3b), Tif34 (eIF3i), and Tif35 (eIF3g). We also confirmed that other translation initiation factors co-purifying with eIF3. Similar to previous reports, we found Tif5 (eIF5), Sui1 (eIF1), Hcr1 (eIF3j), Sui3 (eIF2β) associated with eIF3 at substoichiometric levels ^3,11,16. Importantly, we also identified novel interactions of eIF3 with Fun12 (eIF5B) and Pab1 (PabC). Like the other eIF3-associated factors, Fun12 and Pab1 were found at low levels and were absent from some eIF3 preparations. Fun12 has been shown to facilitate the joining of the 40S and 60S ribosomal subunits in the terminal step of translation inititation ⁶². Δfun12 null strains have a slow growth phenotype and altered polysome profiles suggesting a defect in translation initiation ⁶³. Our finding of a possible interaction between Fun12 and eIF3 is consistent with the previous findings that both complexes play an active role in translation initiation. We postulate that the interaction of eIF3 with eIF5B links the formation of the pre-initiation complex with the binding of the 60S ribosome and the termination of translation initiation. Pab1 has also been shown to stimulate translation initiation ^64,65. Pab1 facilitates the joining of the 5’ and 3’ ends of mRNA by binding the initiation scaffold protein eIF4G ^66,67. Similar to Fun12, Δpab1 deleted strains have decreased levels of translation initiation ⁶⁸. Our finding that PABP interacts with the eIF3 complex complements the previous findings and suggests a role in this interaction in the efficient formation of the 48S pre-initiation complex. Taken together, we speculate that the novel interactions Fun12 and Pab1with the eIF3 complex facilitate their optimal biological function. We believe that our ability to identified novel interactions of a previously well characterized complex stems from our approach combining gentle affinity purification based on multiple eIF3 components with a sensitive and comprehensive protein identification using mass spectrometry.

As part of our characterization of eIF3, we have also identified several sites of phosphorylation. LC/MS/MS analysis can yield very specific and detailed information about protein modification, but we found it necessary to use a variety of approaches. Using ³²P metabolic labeling followed by TAP-Rpg1 purification of eIF3, we demonstrated in vivo phosphorylation of Nip1 and Prt1, two core eIF3 components. The findings are consistent with previous studies investigating the global phosphoproteome of S. cerevisiae ^33,69,70. To identify specific sites of phosphorylation, we analyzed TAP-isolated eIF3 complexes directly by LC/MS/MS. In this way, we identified two phosphorylation sites in Prt1 (S22 and T707) and one in Tif5 (T191). To our surprise, we were unable to identify any phosphorylation sites in Nip1 using this method. We enriched for eIF3 phosphopeptides using IMAC and then analyzed the peptides using LC/MS/MS. This approach enabled us to identify three phosphorylation sites in Nip1 (S98, S99, and S103). Using the Scansite algorithm (http://scansite.mit.edu), Nip1 is predicted to contain 5 sites that are high stringency CK2 targets. Of these five, S98, S99, and S103 were identified in this study. There was no evidence found to support the remaining two sites S56 and S62. However, our mass spectrometry data showed evidence of additional phosphorylations. In a number of tandem spectra, the most intense fragment ion was due to a neutral loss of 98, 49, or 32.6 Da, indicating phosphorylation. However, insufficient fragmentation prevented us from confidently identifying the sequences. Additionally, Rpg1 is the only other member of the eIF3 complex predicted to contain a high stringency CK2 site at T513. No evidence was found to support this and the lack of a signal in the Rpg1 region of the autoradiography (Fig. 2, lane C) supports the conclusion that this particular CK2 site is not targeted for phosphorylation.

To comprehensively identify the eIF3 phosphorylation sites, complete sequence coverage of proteins by mass spectrometry will be required. However, current mass spectrometry protocols seldom achieve complete coverage. A fraction of the proteolytic peptide fragments are either too small or too large to be confidently analyzed by mass spectrometry. In addition, the physiochemical properties of phosphopeptides compound the problems of obtaining comprehensive coverage. Because of the negative charge on phosphate groups, the ionization efficiency of phosphopeptides in positive ion mode is reduced. Furthermore, because a phosphopeptide is typically at a lower abundance relative to the equivalent nonphosphorylated peptide, the more prevalent mass spectrometry signal of nonphosphorylated peptide can suppress the signal of the phosphopeptides ^71–73.

We have found that casein kinase 2 (CK2) plays an important role in phosphorylating eIF3 components. Most of the phosphorylation sites we have identified fit the CK2 consensus sequence with high stringency: S98, S99, and S103 on Nip1 and S22 on Prt1. T707 on Prt1 fits the consensus with low stringency, while T191 on Tif 5 does not match. We found that CK2 subunits co-purify with eIF3 and that, in a reciprocal purification of TAP-Ckb1, eIF3 components co-purify with the CK2 complex. Our data are consistent with a previously described interaction between the CK2 kinase and eIF3 ⁵³. Using in vitro kinase assays, we found that CK2 phosphorylates both Nip1 and Prt1. Interestingly, the phosphorylated regions of Nip1 and Prt1 are predicted to stabilize both protein interactions within the eIF3 complex and interactions of eIF3 with other components of the multifactor pre-initiation complex ¹¹.

Casein Kinase 2 is a conserved serine/threonine protein kinase that is ubiquitous in eukaryotes. Despite numerous studies, its regulation and physiological role is still poorly understood ⁷⁴. Yeast strains with null alleles of either one of the two catalytic subunits (CKA1 or CKA2) are viable. However, deletion of both genes is lethal, showing that CK2 activity is essential for S. cerevisiae viability ⁷⁵. Yeast strains with null alleles of either or both regulatory subunits (CKB1 or CKB2) are viable but are salt sensitive ⁷⁶. A number of CK2 substrates have been reported, leading to the hypothesis that this kinase plays an essential role in several biological processes ⁵⁷.

Several other components of the yeast translation initiation machinery are CK2 targets. Phosphorylation of the eIF2-α subunit of the eIF2 translation initiation complex is required for its optimal function but is not essential for viability ⁷⁷. The yeast eIF4E cap binding protein and the 4E-BP protein Caf20/p20 are also phosphorylated in vitro by CK2⁷⁸.

To further our understanding of how phosphorylation might regulate translation initiation, we used mutational analysis to investigate the functional importance of the phosphorylated CK2 consensus sites in Nip1. Mutation of S98, S99 and S103 causes a 33% increase in the doubling times for the mutant strain. This observation suggests that the phosphorylation events occurring in the region of these sites S98, S99 and S103 are necessary for normal cell growth or division. In the absence of these three sites, in vivo phosphorylation of Nip1 was reduced by only 41%. This is consistent with our mass spectrometry analysis in which we were tantalized by several spectra that showed hallmarks of phosphorylation but did not contain enough fragment ions to provide sequence information. Therefore, we believe there are additional phosphorylation sites in the eIF3 complex yet to be identified. Additional genetic and biochemical experiments will be required to dissect the functional and mechanistic consequences of these phosphorylation events.

Collectively, our results provide a comprehensive characterization of the phosphorylation of eIF3 and its targeting by CK2. The work can serve as a blueprint for the discovery of kinases that phosphorylate components of other protein complexes in vivo. In addition, this approach can be employed in the discovery of enzymes responsible for other types of post-translational modification including ubiquination, acetylation, and methylation.

Table 2. Mass spectrometry identification of CK2 interactions with eIF3.

CK2 complexes were isolated from an S. cerevisiae strain expressing TAP-Ckb2 and an isogenic, untagged yeast strain as a negative control. MudPIT mass spectrometry analyses of the purified complexes were performed in triplicate. PAF values were used to approximate protein abundances as in Fig. 1.

Yeast Protein	Locus	TAP-Ckb2	negative control
Cka1	YIL035C	6.3(1.2)	nda
Cka2	YOR061C	5.6(1.0)	nd
Ckb2	YOR029C	5.5(1.0)	nd
Ckb1	YGL019W	3.6(0.66)	nd
Tif34	YMR146C	2.1(0.38)	nd
Rpg1	YBR079C	2.1(0.39)	nd
Tif35	YDR429C	1.9(0.34)	nd
Nip1	YMR309C	1.5(0.27)	nd
Prt1	YOR361C	1.5(0.28)	nd
Tif5	YBR041W	nd	nd

^and = none detected

Abbreviations

met-tRNAi

methionyl initiator tRNA

eIF3

eukaryotic Initiation Factor 3

eIFs

eukaryotic Initiation Factors

TAP

Tandem Affinity Purification

FSC

fused silica capillary

MudPIT

Multidimensional Protein Identification Technology

CK2

Casein Kinase 2

IMAC

Immobilized Metal Affinity Chromatography