The 5' Untranslated Region of Protein Kinase C{delta} Directs Translation by an Internal Ribosome Entry Segment That Is Most Active in Densely Growing Cells and during Apoptosis

Protein kinase C ${delta}$ (PKC ${delta}$ ) is a member of the PKC family of phospholipid-dependentserine/threonine kinases and is involved in cell proliferation,apoptosis, and differentiation. Previous studies have suggestedthat different PKC isoforms might be translationally regulated.We report here that the 395-nt-long 5' untranslated region (5'UTR) of PKC ${delta}$ is predicted to form very stable secondary structureswith free energies ( ${Delta}$ G values) of around -170 kcal/mol. The 5'UTR of PKC ${delta}$ can significantly repress luciferase translationin rabbit reticulocyte lysate but does not repress luciferasetranslation in a number of transiently transfected cell lines.By using a bicistronic luciferase reporter, we show that the5' UTR of PKC ${delta}$ contains a functional internal ribosome entrysegment (IRES). The activity of the PKC ${delta}$ IRES is greatest indensely growing cells and during apoptosis, when total proteinsynthesis and levels of full-length eukaryotic initiation factor4G are reduced. However, the IRES activity of the 5' UTR ofPKC ${delta}$ is not enhanced during serum starvation, another conditionshown to inhibit cap-dependent translation, suggesting thatits potency is dependent on specific cellular conditions. Accumulatingdata suggest that PKC ${delta}$ has a function as proliferating cellsreach high density and in early and later events of apoptosis.Our studies suggest a mechanism whereby PKC ${delta}$ synthesis can bemaintained under these conditions when cap-dependent translationis inhibited.

INTRODUCTION

Top

Introduction

Protein kinase C (PKC) is a key mediator of signal transductionpathways that regulate a vast array of physiological functions,including cell proliferation, differentiation, and apoptosis(40). Members of the mammalian PKC family of phospholipid-dependentserine/threonine kinases fall into three subfamilies based ontheir amino acid sequences and enzymatic properties. The conventionalPKCs (PKC ${alpha}$ , PKCßI, PKCßII, and PKC ${gamma}$ ) are activatedby Ca²⁺ and diacylglycerol (DAG), the novel PKCs (PKC ${varepsilon}$ , PKC ${delta}$ ,PKC ${eta}$ , and PKC ${theta}$ ) are DAG dependent and Ca²⁺ independent, and theatypical PKCs (PKCµ, PKC ${zeta}$ , and PKC ${iota}$ / ${lambda}$ ) are Ca²⁺ and DAG independent(7, 33). Although the heterogeneity of PKC function is postulatedto be a consequence of the multiple isoforms of PKC that existwithin cells, the specific roles of the various isoforms havenot been clearly established.

When overexpressed, the novel PKC isoforms PKC ${varepsilon}$ and PKC ${delta}$ haveopposing effects on cell growth and proliferation. Overexpressionof PKC ${varepsilon}$ stimulates growth and promotes transformation of rat-6fibroblasts, NIH 3T3 fibroblasts, and rat colonic epithelialcells (2, 36, 42). However, overexpression of PKC ${delta}$ inhibits growthof NIH 3T3 fibroblasts, results in cell cycle arrest of CHOcells, and is lethal to normal and neoplastic keratinocytes(24, 36, 53). PKC ${varepsilon}$ behaves as an oncogene because of its abilityto induce tumors in nude mice (2). However, PKC ${delta}$ is considereda tumor suppressor, as its overexpression reverses the transformedphenotype of D/src rat colonic epithelial cells and depletionof PKC ${delta}$ promotes c-src-induced transformation of rat 3Y1 fibroblasts(28, 43). Moreover, transgenic mice overexpressing PKC ${delta}$ are resistantto tumor promotion by phorbol esters (45). PKC ${delta}$ is also implicatedin the induction and execution of programmed cell death, orapoptosis, in response to a number of signals, including ionizingradiation, tumor necrosis factor alpha, agonistic Fas antibody(10), H₂O₂ (51), etoposide (29, 31), and phorbol esters (13,24).

Most cellular mRNAs possess 5' untranslated regions (UTRs) ofless than 100 nucleotides (nt) that are likely to allow efficientinitiation of translation by cap-dependent ribosomal scanning(21). However, the 5' UTRs of many regulators of cell proliferationand apoptosis, including PKC ${varepsilon}$ , are long and structured and containupstream AUG codons so that scanning ribosomes are unlikelyto efficiently initiate translation (22, 39). Efficient initiationof the translation of a number of these mRNAs is achieved bya cap-independent mechanism mediated by internal ribosome entrysegments (IRESs) (reviewed in reference 52). Translation mediatedby IRESs is usually favored when cap-dependent initiation isinhibited. The IRESs of XIAP, p97/DAP5/NAT1, and c-myc are activeduring apoptosis (16, 17, 49). The IRES of vascular endothelialgrowth factor is active during hypoxia (47), and the IRES ofornithine decarboxylase is active during the G₂/M phase of thecell cycle (44).

Nucleotide sequencing of PKC ${delta}$ isolated from rat and mouse cDNAlibraries has revealed that it has a long 5' UTR (35, 41). Weshow here that the 395-nt-long 5' UTR of rat PKC ${delta}$ can potentiallyform secondary structures with free energies expected to significantlyimpede cap-dependent ribosome scanning and may thus be subjectto translational regulation. We show that the 5' UTR of PKC ${delta}$ can repress luciferase translation in rabbit reticulocyte lysatebut not in transiently transfected cells. With a bicistronicreporter assay, we show that the 5' UTR of PKC ${delta}$ contains an IRESand that the IRES activity of the 5' UTR of PKC ${delta}$ is strongestin cells grown to a high density and during apoptosis, conditionsunder which cap-dependent translation is reduced. However, theIRES activity of the 5' UTR of PKC ${delta}$ is not enhanced during serumstarvation, another condition known to inhibit cap-dependenttranslation.

MATERIALS AND METHODS

Top

Materials and Methods

DNA constructs. The PKC ${delta}$ cDNA clone (41) was derived from rat brain and was agift from Y. Ono (Kobe University, Kobe, Japan). The luciferasereporter plasmids pGL3-promoter (called pGL here) and pGL3-control(called pGLenh here), the pSV-ß-galactosidase plasmid,and cloning plasmid pGEM7Zf(-) were purchased from Promega.Restriction and modifying enzymes were purchased from New EnglandBiolabs or Promega and used in accordance with the manufacturer'sinstructions. Plasmid DNA was prepared for transfection witha plasmid midi prep kit (QIAGEN).

Plasmid constructs for transient transfection were generatedas follows. A 3-kb EcoRI fragment comprising the cDNA of PKC ${delta}$ was subcloned into pGEM7. The 5' UTR was cloned by PCR withan SP6 primer (5' end of the 5' UTR) and a PKC ${delta}$ -specific primerfrom the start codon, introducing an NcoI site (CCATGG) as thestart codon. The PCR product was cleaved with HindIII/NcoI torelease a 390-bp fragment that was subcloned into the pGL3-promoterplasmid and the pGL3-control plasmid to make pGL ${delta}$ and pGLenh- ${delta}$ ,respectively. The following plasmids were generously providedby A. Willis (Leicester University, Leicester, United Kingdom)and M. Stoneley (Leeds University, Leeds, United Kingdom): pGL3R(called pR-F here), pGL3RutrH (48), pGL3RHRV (called pR-HRV-Fhere) (49), and pHpL (called pH here) (50). The 5' UTR of PKC ${delta}$ was introduced between the Renilla and firefly luciferase codingregions with the EcoRI and NcoI restriction sites to make pR- ${delta}$ -F.Bicistronic plasmid pGL3RutrH, containing a stable hairpin 5'of Renilla luciferase, was used to make pHR- ${delta}$ -F by exchangingthe 5' UTR of c-myc with the 5' UTR of PKC ${delta}$ with the SpeI andNcoI restriction sites. The pHpL monocistronic luciferase plasmid,containing a stable hairpin 5' of the luciferase start, wasused to make pH- ${delta}$ with the EcoRI and NcoI restriction sites.

For the plasmids used for in vitro transcription/translation,HindIII-to-XbaI fragments from the pGL3-promoter plasmid (containingthe entire luciferase coding region) and the same restrictionfragment from pGL ${delta}$ (containing the 5' UTR of PKC ${delta}$ linked to theluciferase open reading frame) were treated with the Klenowfragment of DNA polymerase to generate blunt ends and subclonedinto the SmaI site of the T7 promoter pA90 vector pBKSpA (agift from James Malter) to make pT7pA and pT7 ${delta}$ pA, respectively.The integrity of all clones was confirmed by DNA sequencing.

In vitro transcription and translation. In vitro-transcribed, capped RNA was made with T7 RNA polymerase(Gibco BRL) in a 20-µl reaction mixture comprising 1 µgof HindIII-linearized pT7pA or pT7 ${delta}$ pA plasmid DNA; 0.5 mM (each)CTP, UTP, and ATP; and 0.05 mM GTP with 0.5 mM 7mG cap analogue(Promega) in 1x T7 transcription buffer (40 mM Tris/HCl [pH8.0], 8 mM MgCl₂, 2 mM spermidine, 25 mM NaCl), 5 mM dithiothreitol,and 30 U of RNasin (Promega) at 37°C for 2 h. The reactionmixture was treated with DNase I and purified through G-50 Sephadexcolumns (Pharmacia) to remove unincorporated nucleotides andfree 7mG cap. The RNA was quantified spectrophotometrically,and its quality was verified on nondenaturing agarose gels.Equal amounts of RNA (typical final concentration, 20 ng/µl)were used in a 25-µl in vitro translation reaction mixturecomprising 70% rabbit reticulocyte lysate (Promega), 0.02 mMeach amino acid, and 30 U of RNasin. The in vitro translationwas allowed to proceed at 28°C for 90 min, and the reactionmixture was finally treated with 1 µl of RNase A at 10mg/ml for 10 min at 28°C. One microliter of each reactionmixture was used in a luciferase assay (with 50 µl ofluciferase assay reagent [described below]). A 1-µl samplewas taken at the start and end of the translation reaction,and the RNA was extracted for Northern blot analysis. The translationreactions were done at least five times with different batchesof in vitro-transcribed RNA and rabbit reticulocyte lysate.

Cell culture and transient transfection. The 3T3 Swiss albino (mouse fibroblast), 3T6 Swiss albino (mousefibroblast), BALB/3T12-3 (mouse tumorigenic fibroblast), C6(rat glioma), HeLa (human cervical carcinoma), and MCF7 (humanbreast carcinoma) cell lines were purchased from ECACC (Salisbury,United Kingdom) and maintained at 37°C in 5% CO₂ in Dulbecco'smodified Eagle medium (DMEM; Gibco BRL, Life Technologies) containingpenicillin/streptomycin and supplemented with 10% fetal bovineserum (FBS; Gibco BRL, Life Technologies). For a typical transienttransfection, cells were grown to 50 to 70% density in six-wellplates and then cotransfected with 1 µg of pSV-ß-galactosidaseplasmid and 1 µg of the luciferase reporter plasmid with10 µl (30 µg) of Superfect reagent (QIAGEN) in accordancewith the manufacturer's instructions. Low, medium, and highcell densities were achieved by seeding cells at 4 x 10³/cm²(approximately 10% density), 2 x 10⁴/cm² (approximately 50%density), or 4 x 10⁴/cm² (approximately 100% density), respectively.Cells were transfected at these densities and then harvestedafter 24 h. For serum starvation experiments, cells growingat 50 to 70% density in 10% FBS-containing DMEM were transfected,and then 6 h later, the cells were washed twice with serum-freeDMEM and grown for a further 24 h in serum-free DMEM beforebeing harvested. To induce apoptosis, 0.5 µg of tumornecrosis factor-related apoptosis-inducing ligand (TRAIL) wasadded per ml of medium 24 h after transfection and the cellswere harvested 6 h later. For the RNA analysis, a 75-cm² flaskof cells was transfected for each construct with 10 µgof plasmid DNA and 15 µl of FuGene 6 (Roche) in accordancewith the manufacturer's instructions.

Enzyme assays. For the monocistronic luciferase assays, cells were harvestedby being scraped into 120 µl of reporter lysis bufferand assayed with the Luciferase Assay System (Promega) in accordancewith the manufacturer's instructions. For the bicistronic luciferaseassays, cell were harvested with 120 µl of passive lysisbuffer and assayed with the Dual Luciferase Reporter Assay System(Promega) in accordance with the manufacturer's instructions.The peak light emission for 10 s at 25°C due to luciferaseactivity was measured as luciferase relative light units (LRLUs)with a Berthoid luminometer. ß-Galactosidase activitywas measured with the substrate o-nitrophenyl-ß-D-galactopyranoside(ONPG). ß-Galactosidase activity absorbances weremeasured at 420 nm. For each transfection, the luciferase activityis expressed relative to the ß-galactosidase activityto control for transfection efficiency. The mean values of relativeluciferase activity shown in Results were derived from the analysisof at least five independent cotransfections of each luciferaseplasmid with pSV-ß-galactosidase with at least twodifferent preparations of plasmid DNA.

RNA analysis. RNA secondary structures were predicted with mFOLD (32), whichis available at http://bioinfo.math.rpi.edu/ $~$ mfold/rna/form1.cgi.Total RNA was extracted with the Trizol reagent (Gibco BRL,Life Technologies) and treated with RQ1 DNase (Promega). PolyadenylatedRNA was purified with oligo(dT) cellulose with the PolyA-spinmRNA isolation kit (New England Biolabs). RNA from transientlytransfected cells [total: 20 µg of luciferase RNA, 10µg of glyceraldehyde-3-phosphate dehydrogenase (GAPDH)RNA, or 1 µg of poly(A) RNA] was denatured with formamide/formaldehydeand subjected to electrophoresis on a 1% agarose-6% formaldehyde-morpholinepropanesulfonicacid (MOPS) gel. The RNA was transferred onto Hybond N⁺ membranes(Amersham). The membranes were hybridized with [ ${alpha}$ -³²P]UTP (AmershamRadiochemicals)-labeled single-stranded RNA antisense probesderived from the entire coding regions of firefly luciferaseor GAPDH (12) in ULTRAhybe hybridization buffer (Ambion). Theluciferase and GAPDH RNA signals were quantified by phosphorimagingwith a Molecular Dynamics phosphorimager with ImageQuant (version5.1) software.

PKC ${delta}$ and GAPDH RNA levels were assessed in cells growing at differentdensities with semiquantitative reverse transcription (RT)-PCR.One microgram of total RNA was reverse transcribed with 0.5µg of 5'-pd(T)_12-18-3' (Pharmacia) with SuperScript IIreverse transcriptase (Gibco BRL, Life Technologies) in a 25-µlreaction volume in accordance with the manufacturer's instructions.The PCR was performed in a 50-µl volume with 2 µlof the RT reaction mixture with Taq polymerase (Promega) inaccordance with the manufacturer's instructions. The numberof amplification cycles chosen for analysis of PKC ${delta}$ and GAPDHRNA levels by semiquantitative PCR was determined for each primerpair to be within the exponential range for cells growing atdifferent densities. PKC ${delta}$ was amplified with primers specificto the C-terminal variable region (sequences: 5' GAC ATC ATGGAG AAG CTA TTC and 5' TCG ATG AGG TTC TTG TCA CTG) with 20cycles of 94°C for 1 min, 55°C for 1 min, and 72°Cfor 1 min, and GAPDH was amplified with specific primers (sequences:5' AGA CAG CCG CAT CTT CTT GTG C and 5' CTC CTG GAA GAT GGTGAT GG) with 15 cycles of 94°C for 1 min, 55°C for 1min, and 72°C, for 1 min. Ten microliters of the PCR mixturewas electrophoresed on 2% agarose gels and stained with ethidiumbromide. The intensity of ethidium bromide staining was quantifiedwith the AlphaImager 1200 documentation and analysis system(Alpha Innotech Corporation). The specificity of the PCR productswas confirmed by Southern blotting with oligonucleotide probesdesigned to hybridize to internal sequences of the predictedPCR products (data not shown).

Metabolic labeling of cells. Cells were grown in serum-free methionine/cysteine-free DMEMfor 30 min prior to the addition of 60 µCi of [³⁵S]Met/Cys(Promix; Amersham) per ml and incubation for 1 h. Cells werethen washed twice with phosphate-buffered saline and harvestedin lysis buffer A (150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid,0.1% sodium dodecyl sulfate [SDS], 50 mM Tris-HCl, 1 mM EDTA,pH 8.0) containing 10 µg of aprotinin per ml, 10 µgof leupeptin per ml, and 2 mM 4-(2-aminoethyl)benzenesulfonylfluoride (AEBSF). Total protein concentrations were estimatedwith the bicinchoninic acid assay (Pierce), and equal amountsof protein were added to Laemmli loading buffer and heated to100°C for 5 min before being subjected to denaturing polyacrylamidegel electrophoresis with 10% polyacrylamide gels. Proteins weretransferred to Hybond-C (Amersham) membranes. The membraneswere stained with Ponceau (Sigma) to visualize the protein,individual lanes were excised, and the amount of incorporated³⁵S in each lane was quantified by scintillation counting.

Immunoblotting. Total protein was extracted from cells by scraping them intolysis buffer A containing 10 µg of aprotinin per ml, 10µg of leupeptin per ml, and 2 mM AEBSF for eukaryoticinitiation factor 4G (eIF4G) or by scraping them into lysisbuffer B (0.5% NP-40, 50 mM Tris, 0.5 mM EDTA, 2 mM EGTA, pH7.5) containing 10 µg of aprotinin per ml, 10 µgof leupeptin per ml, and 2 mM AEBSF for PKC ${delta}$ . Samples containing30 µg of protein were subjected to denaturing polyacrylamidegel electrophoresis and transferred to Hybond-C as describedabove. The membranes were incubated with polyclonal rabbit antiserumagainst PKC ${delta}$ (used at 1:5,000; a gift from N. Groom, Oxford BrookesUniversity), with polyclonal rabbit antiserum against the Cterminus (amino acids 920 to 1396) of eIF4GI (used at 1:2,000;a gift from S. Morley, Sussex University), or with polyclonalrabbit antiserum against poly(ADP-ribose) polymerase (PARP),which detects both the 116-kDa full-length PARP molecule andthe 89-kDa caspase cleavage product (used at 1:1,000; purchasedfrom Cell Signaling Technology). The membranes were then incubatedwith an anti-rabbit peroxidase-conjugated secondary antibody(used at 1:5,000; Sigma). Immunoreactive bands were detectedby chemiluminescence assay.

RESULTS

Top

Results

The 5' UTR of PKC ${delta}$ is predicted to form highly stable secondary structures and is resistant to cap-dependent translation in vitro but not in vivo. The 5' UTR of PKC ${delta}$ isolated from a rat brain cDNA library is395 nt long and has a G+C content of 68%. With the computeralgorithm of Zuker (32), which estimates the minimum free energyof a folded RNA molecule, the 5' UTR of PKC ${delta}$ was predicted toform highly stable secondary structures with free energies ( ${Delta}$ Gvalues) exceeding -170 kcal/mol (data not shown). As an RNAstem-loop structure with a free energy of -50 kcal/mol is inhibitoryto cap-dependent ribosomal scanning (20), it was expected thatthe 5' UTR of PKC ${delta}$ would inhibit cap-dependent translation initiation.To test this, the 5' UTR of PKC ${delta}$ was introduced immediately upstreamof the luciferase start in either T7 RNA polymerase promoter-basedplasmid pT7pA or simian virus 40 promoter-based plasmid pGLto create pT7 ${delta}$ pA and pGL ${delta}$ , respectively, as shown in Fig. 1A.Capped and polyadenylated luciferase RNA generated in vitrofrom pT7 ${delta}$ pA with T7 RNA polymerase was poorly translated in rabbitreticulocyte lysate compared to luciferase RNA generated fromthe control, pT7pA (Fig. 1B). In transiently transfected Swiss3T3 cells, the levels of luciferase conferred by the pGL andpGL ${delta}$ constructs were similar (Fig. 1C). However, in transientlytransfected Swiss 3T6, C6, BALB/c, HeLa, and MCF7 cells, thepGL ${delta}$ construct conferred significantly more luciferase activitythan did pGL (t test, P < 0.01) (Fig. 1C). Northern blotanalysis revealed that comparable amounts of luciferase-specificRNA of the expected size could be recovered after 90 min fromthe in vitro translation reactions programmed with in vitro-transcribedpT7pA and pT7 ${delta}$ pA (Fig. 1D). In transiently transfected cells,the pGL and pGL ${delta}$ constructs produced similar levels of luciferasemRNA of the expected size. These results suggest that, in contrastto the pT7pA and pGL constructs, the pT7 ${delta}$ pA and pGL ${delta}$ constructsdo not affect steady-state levels of luciferase mRNA, nor arethey likely to contain splicing sequences. Taken together, thesedata suggest that the 5' UTR of PKC ${delta}$ can repress luciferase translationin rabbit reticulocyte lysate but enhance translation of luciferasein several cell lines.

View larger version (44K):
[in this window]
[in a new window]
FIG. 1. The 5' UTR of PKC ${delta}$ represses translation in vitro but not in vivo. (A) Schematic diagram of the T7 RNA polymerase-and pGL3-promoter-based luciferase constructs used for in vitro and in vivo analyses, respectively. The 5' UTR of PKC ${delta}$ was introduced immediately upstream of the luciferase start codon in the pGL3-promoter plasmid with the HindIII and NcoI restriction sites to make pGL ${delta}$ . The luciferase-coding region from pGL and the 5' UTR of PKC ${delta}$ linked to luciferase from pGL ${delta}$ were introduced into the pBKSpA plasmid to make pT7pA and pT7 ${delta}$ pA, respectively, as described in Materials and Methods. (B) Capped and polyadenylated in vitro-transcribed RNAs were generated with T7 RNA polymerase from the linearized pT7pA and pT7 ${delta}$ pA plasmids. Equal amounts of purified RNA were used to program rabbit reticulocyte in vitro translation reactions, which were subsequently assayed for luciferase activity. The luciferase activity conferred by pT7pA was normalized to 100%. Results shown are the mean and standard error of the mean from at least five independent translation reactions with different in vitro-transcribed RNAs. (C) Rodent and human cell lines were cotransfected with either the pGL or the pGL ${delta}$ luciferase construct and pSV-ß-gal to control for transfection efficiency. For each cell line, the luciferase activity shown is relative to the ß-galactosidase activity conferred by pSV-ß-gal. The relative luciferase/ß-galactosidase activity conferred by pGL is normalized to a value of 1 for each cell line. Results are shown as the mean and standard error of the mean of at least five independent transfections. (D) Northern blot of pT7pA and pT7 ${delta}$ pA luciferase RNAs during in vitro translation (top left, 0-min time point; bottom left, 90-min time point). Northern blot of RNA derived from cells transfected (in vivo) with the pGL and pGL ${delta}$ constructs and hybridized with a ³²P-labeled firefly luciferase (top right) or GAPDH (bottom right) riboprobe. SV40, simian virus 40.

The 5' UTR of PKC ${delta}$ enhances translation when cap-dependent ribosomal scanning is inhibited. HeLa cells were transiently transfected with a construct carryinga stable hairpin structure designed to block cap-dependent translation(Fig. 2A). Luciferase activity conferred by the pH constructcontaining the stable hairpin structure was 5% of that conferredby the control, pGLenh (Fig. 2B). The pH- ${delta}$ construct carryingthe 5' UTR of PKC ${delta}$ between the stable hairpin and the translationstart site conferred fourfold more luciferase activity thandid pH, suggesting that the 5' UTR of PKC ${delta}$ can promote translationinitiation when cap-dependent scanning is inhibited (Fig. 2B).However, luciferase activity conferred by the pH- ${delta}$ constructwas around fivefold less than that conferred by the control,pGLenh, and more than sixfold less than that conferred by pGLenh- ${delta}$ .This suggests that a component of translation initiation bythe 5' UTR of PKC ${delta}$ must be cap dependent and has been inhibitedby the introduced cap-proximal hairpin. Alternatively, the cap-proximalhairpin may have interfered with the function of the 5' UTRof PKC ${delta}$ in promoting cap-independent initiation.

View larger version (24K):
[in this window]
[in a new window]
FIG. 2. The 5' UTR of PKC ${delta}$ enhances translation when cap-dependent scanning is inhibited in a monocistronic construct. (A) Schematic diagram of the constructs used. The 5' UTR of PKC ${delta}$ was introduced immediately upstream of the luciferase open reading frame in the pGL3-control plasmid (called pGLenh) to create pGLenh- ${delta}$ . The stable hairpin structure with a free energy ( ${Delta}$ G) of -55 kcal/mol was introduced 5' of the firefly luciferase start in the pGLenh plasmid to create pH as previously described (50). The 5' UTR of PKC ${delta}$ was introduced between the hairpin and the luciferase start to create pH- ${delta}$ as shown. (B) Luciferase activities conferred by the monocistronic hairpin constructs in transiently transfected HeLa cells relative to the ß-galactosidase activity conferred by pSV-ß-gal to control for transfection efficiency. The relative luciferase or ß-galactosidase activity conferred by the pH construct is normalized to a value of 1. The results shown are the mean and standard error of the mean of at least five independent transfections. SV40, simian virus 40.

The 5' UTR of PKC ${delta}$ functions as an IRES in a bicistronic construct. The 5' UTR of PKC ${delta}$ was introduced between the Renilla and fireflyluciferase coding regions in a bicistronic reporter plasmidto test if it could function as an IRES in vivo (Fig. 3A). Thehuman rhinovirus (HRV) IRES was included in the transfectionsas a positive control (1, 49). In all of the cell lines tested,the presence of the 5' UTR of PKC ${delta}$ in the intergenic region resultedin downstream (firefly) luciferase activity two- to sevenfoldgreater than that of the control, pR-F (Fig. 3B). There wereno significant differences in the activity of the upstream (Renilla)luciferase conferred by the pR-F and pR- ${delta}$ -F constructs in anyof the transiently transfected cells (Fig. 3B). The 5' UTR ofPKC ${delta}$ was at least as efficient as the HRV IRES in initiatingtranslation of the downstream luciferase in cells of human originand was at least twofold more efficient in cells of rodent origin.The lack of activity of the HRV IRES in nonhuman cells is consistentwith previous studies (1). Northern blot analysis of mRNA isolatedfrom HeLa cells transiently transfected with the control, PKC ${delta}$ ,and HRV bicistronic constructs revealed an intact single speciesof luciferase-specific mRNA migrating at the expected size (Fig.3C). This supports the notion that the 5' UTRs of PKC ${delta}$ and HRVneither alter the steady-state levels of the bicistronic RNAnor result in cleavage of the bicistronic RNA into a monocistronicfirefly luciferase.

View larger version (53K):
[in this window]
[in a new window]
FIG. 3. The 5' UTR of PKC ${delta}$ acts as an IRES in a bicistronic construct. (A) Schematic diagram of bicistronic Renilla and firefly luciferase constructs used to test for IRES activity in transiently transfected cells. The 5' UTR of PKC ${delta}$ was inserted between the coding regions of the Renilla and firefly luciferases in the pR-F vector to make pR- ${delta}$ -F as detailed in Materials and Methods. The pR-HRV-F construct containing the HRV IRES was used as a positive control (49). (B) Firefly and Renilla luciferase activities conferred by the bicistronic constructs in transiently transfected rodent and human cell lines relative to the ß-galactosidase activity conferred by pSV-ß-gal. For each cell line, the relative firefly and Renilla luciferase activities conferred by pR-F are normalized to a value of 1. The results shown are the mean and standard error of the mean from at least five independent transfections. (C) Northern blot analysis of 1 µg of poly(A) RNA isolated from HeLa cells transiently transfected with no DNA or the pR-F, pR- ${delta}$ -F, or pR-HRV-F construct and hybridized with a firefly luciferase-specific ³²P-labeled riboprobe. The approximate positions (from top of gel to bottom) of 5-, 3-, and 2-kb RNA markers (M) are indicated on the left.

The 5' UTR of PKC ${delta}$ does not promote ribosomal readthrough in the bicistronic construct. The 5' UTR of PKC ${delta}$ may function in the bicistronic constructby stimulating readthrough of ribosomes through the intergenicregion. If this were the case, it would be expected that, inthe presence of a stable hairpin that inhibits translation fromthe upstream cistron, the level of luciferase activity fromthe downstream cistron would be proportionally decreased inconstructs carrying the 5' UTR of PKC ${delta}$ as the intergenic spacer.We found that in the control bicistronic construct, which lacksan intergenic spacer, a stable hairpin structure inserted 5'of the Renilla coding region reduced translation of both luciferasecistrons to around 20% of the levels conferred by the controllacking the hairpin (reference 48 and data not shown). We comparedthe Renilla and firefly luciferase activities in the bicistronicconstruct carrying the 5' UTR of PKC ${delta}$ as the intergenic spacerin the absence and presence of a stable hairpin 5' of the upstreamcistron (Fig. 4A). Although Renilla luciferase activity wassignificantly reduced in all of the cell lines tested, therewas no significant reduction in firefly luciferase activity(Fig. 4B). This confirms that the 5' UTR of PKC ${delta}$ does not functionby enhancing readthrough of the bicistronic construct and supportsthe notion that the 5' UTR of PKC ${delta}$ contains a bona fide IRES.

View larger version (58K):
[in this window]
[in a new window]
FIG. 4. The 5' UTR of PKC ${delta}$ does not promote ribosomal readthrough in the bicistronic construct. (A) Schematic diagram of the bicistronic constructs used for transient transfections. The stable hairpin structure with a free-energy of -55 kcal/mol introduced 5' of the Renilla luciferase start (48) was used to create pHR- ${delta}$ -F. (B) Renilla and firefly luciferase activities conferred by the bicistronic pR- ${delta}$ -F and pHR- ${delta}$ -F constructs in transiently transfected rodent and human cell lines relative to the ß-galactosidase activity conferred by pSV-ß-gal. For each cell line, the relative Renilla and firefly luciferase activities conferred by the pR- ${delta}$ -F construct were normalized to a value of 1. The results shown are the mean and standard error of the mean from at least five independent transfections.

The IRES activity of the 5' UTR of PKC ${delta}$ is greatest in cells growing at a high density, when cap-dependent translation is reduced. Previous studies have indicated that IRES activity of a 5' UTRcan promote translation initiation during times of cellularstress and other occasions when cap-dependent translation isinhibited. Cells grown to a high density, or confluency, enterG₀ and growth arrest, which is marked by changes in the expressionof many cellular factors (46, 54). PKC ${delta}$ protein levels are maximalwhen RPTE cells, C6 rat glioma cells, and 3T6 fibroblasts aregrown to a high density rather than a subconfluent cell density(9, 27, 38). We therefore speculated that the IRES activityof the 5' UTR of PKC ${delta}$ might be required to translate the mRNAwhen cells are grown to a high density. To test this, HeLa cellswere transiently transfected with the control, PKC ${delta}$ , and HRVbicistronic constructs at low, medium, and high cell densities.At low and medium cell densities, the level of firefly luciferaseactivity conferred by pR- ${delta}$ -F was around eightfold greater thanthat of the control, pR-F (Fig. 5A). However, when cells weretransfected at a high density, the level of firefly luciferaseactivity conferred by pR- ${delta}$ -F was 18-fold greater than that conferredby the control. Elevated firefly luciferase activity conferredby pR- ${delta}$ -F compared to the control was also observed when thecells were transfected at a medium density and then harvestedafter growth to a high density (data not shown). There was amodest, but not statistically significant, improvement in theIRES activity of the HRV construct in cells growing at a highdensity compared to that of cells growing at a medium density.These data therefore suggest that the IRES activity of the 5'UTR of PKC ${delta}$ is specifically elevated in cells growing at a highdensity. It is likely that cap-dependent translation is reducedin cells growing at a high density, as the relative amount ofthe upstream (Renilla) luciferase was significantly reducedunder these growth conditions compared to that in low- and medium-densitycells (Fig. 5B). However, it is also possible that the half-lifeof Renilla luciferase is shorter in cells growing at a highdensity and that cap-dependent translation is not reduced indensely growing cells. To test whether cap-dependent translationis reduced in densely growing cells, HeLa cells growing at differentdensities were labeled with [³⁵S]methionine for 1 h prior tobeing harvested (Fig. 5C). Analysis of the cell lysates revealedreduced incorporation of [³⁵S]methionine at a high cell density,supporting the idea that global cell protein synthesis is reduced.A significant reduction in global protein synthesis was alsoobserved in 3T6 cells grown to a high density (data not shown).Proteolytic cleavage of eIF4G has previously been observed underconditions in which global translation is reduced (5, 30). AsHeLa cells grow to a high density, immunoblotting for eIF4Grevealed a predicted 76-kDa cleavage product (Fig. 5D). Takentogether, these results suggest that when cells are growingat a high density, global translation is reduced and the IRESactivity of the 5' UTR of PKC ${delta}$ is significantly enhanced.

View larger version (29K):
[in this window]
[in a new window]
FIG. 5. The IRES activity of the 5' UTR of PKC ${delta}$ is greatest in densely growing cells. (A) Firefly luciferase activities conferred by the bicistronic pR-F, pR- ${delta}$ -F, and pR-HRV-F constructs in HeLa cells transiently transfected at approximately 10, 50, and 100% cell density relative to the ß-galactosidase activity conferred by pSV-ß-gal. Relative firefly luciferase activity conferred by pR-F is normalized to a value of 1 for each cell density. (B) Relative Renilla luciferase activity conferred by the bicistronic constructs, normalized to a value of 1 for each construct at 10% cell density. The results shown are the mean andstandard error of the mean from at least five independent transfections. (C) Global protein synthesis was determined by labeling HeLa cells growing at a low (10 to 20%), medium (50 to 70%), or high (90 to 100%) density with [³⁵S]methionine as described in Materials and Methods. (D) Total protein extracted from cultures parallel to those in panel C were electrophoresed through SDS-10% acrylamide gels, and the proteins were transferred to nitrocellulose and probed with antisera raised against eIF4G. The approximate locations of protein molecular size markers (M) in kilodaltons are indicated at the left.

Endogenous PKC ${delta}$ protein levels are elevated as cells reach a high density, while its mRNA levels do not change. To determine the status of endogenous PKC ${delta}$ in HeLa cells growingat different densities, we performed immunoblotting and semiquantitativeRT-PCR. Immunoblotting revealed that PKC ${delta}$ protein levels peakin cells growing at a maximal density (Fig. 6A), while therewas no change in the PKC ${delta}$ mRNA levels (Fig. 6B). This suggeststhat as HeLa cells grow to a high density, endogenous PKC ${delta}$ proteinlevels are modulated posttranscriptionally.

View larger version (54K):
[in this window]
[in a new window]
FIG. 6. Levels of endogenous PKC ${delta}$ protein, but not its mRNA, increase in densely growing cells. (A) Equal amounts of total protein extracted from HeLa cells growing at approximately 20, 50, 80, and 100% density were electrophoresed through SDS-10% acrylamide gels and then transferred to nitrocellulose and probed with antiserum against PKC ${delta}$ . (B) PKC ${delta}$ RNA levels were determined with semiquantitative RT-PCR with total RNA isolated from HeLa cells growing at approximately 20, 50, 80, and 100% density (upper band, PCR product; lower band, primer). The same reverse-transcribed RNA samples were amplified with primers specific for GAPDH. A negative control of amplified RNA is shown in the left lane of each gel.

The IRES activity of the 5' UTR of PKC ${delta}$ is not enhanced by serum starvation but is enhanced during apoptosis. Previous studies have demonstrated that serum starvation andapoptosis reduce global translation (5, 49). We therefore examinedwhether the IRES activity of the 5' UTR of PKC ${delta}$ is improved underthese conditions of cellular stress. Cells were transfectedwith the bicistronic constructs and then subjected to serumstarvation or apoptosis induced by TRAIL. During serum starvation,there was a significant reduction in both firefly (cap-independent)and Renilla (cap-dependent) luciferase activities (Fig. 7A andB). This suggests that the IRES activity of the 5' UTR of PKC ${delta}$ is diminished during serum starvation.

View larger version (44K):
[in this window]
[in a new window]
FIG. 7. The IRES activity of the 5' UTR of PKC ${delta}$ is not enhanced in serum-starved cells. (A) HeLa cells were grown in 10% FBS-containing DMEM to approximately 50% density and transiently transfected with the bicistronic pR-F or pR- ${delta}$ -F construct and pSV-ß-gal to control for transfection efficiency. After 6 h, the cells were washed twice with serum-free DMEM and incubated in serum-free DMEM for a further 24 h before being harvested and assayed for firefly luciferase and ß-galactosidase activities. Relative firefly luciferase activity conferred by pR-F is normalized to a value of 1 for each condition. (B) Relative Renilla luciferase activities from the transfected cells described above were determined and normalized to a value of 1 for each construct in cells growing in normal medium. The results shown are the mean and standard error of the mean from at least five independent transfections.

The IRES activity of the 5' UTR of c-myc is maintained duringapoptosis induced by TRAIL (49). In HeLa cells transfected withpR- ${delta}$ -F, we observed a significant improvement in relative fireflyluciferase activity, compared to that of the pR-F control, 6h after 0.5 µg of TRAIL per ml was added to the culturemedium (Fig. 8A). Cap-dependent initiation was modestly butsignificantly reduced, as shown by analysis of the Renilla luciferaseactivity (Fig. 8B). Induction of apoptosis by 0.5 µg ofTRAIL per ml is clearly evident after 4 h by the appearanceof the 89-kDa caspase-specific cleavage product of PARP (Fig.8C). Further evidence that TRAIL reduces cap-dependent translationis shown by a time-dependent reduction of the full-length eIF4G-specificsignal and the appearance of the 76-kDa cleavage product 2 hafter the addition of TRAIL to HeLa cells (Fig. 8C). These analysessuggest that the IRES activity of the 5' UTR of PKC ${delta}$ is enhancedunder specific, rather than general, cellular conditions thatreduce cap-dependent translation.

View larger version (37K):
[in this window]
[in a new window]
FIG. 8. The IRES activity of the 5' UTR of PKC ${delta}$ is enhanced in cells treated with TRAIL to induce apoptosis. (A) HeLa cells were transiently transfected with the bicistronic pR-F or pR- ${delta}$ -F construct and pSV-ß-gal to control for transfection efficiency. After 24 h, 0.5 µg of TRAIL per ml was added to the medium and the cells were harvested 6 h later. Relative firefly luciferase activities are shown, with the levels conferred by pR-F normalized to a value of 1. (B) Relative Renilla luciferase activities from the transiently transfected cells above were determined and normalized to a value of 1 for each construct in cells in normal medium. (C) Equal amounts of total protein extracted from HeLa cells growing in the presence of 0.5 µg of TRAIL per ml for 0, 1, 2, 4, 6, and 8 h were electrophoresed through SDS-10% acrylamide gels and then transferred to nitrocellulose and probed withantisera against PARP or eIF4G. The positions of full-length PARP and eIF4G are indicated, and the position of specific cleavage products is shown by the asterisk. The approximate locations of protein molecular size markers (M) in kilodaltons are indicated at the left.

DISCUSSION

Top

Discussion

PKC ${delta}$ has been implicated in the response of cells to a myriadof signals, including those that promote differentiation, celltransformation, growth arrest, and apoptosis (8, 10, 25, 37,43). While many studies have addressed the posttranslationalactivation and action of PKC ${delta}$ in cells overexpressing nativeor dominant-negative forms of PKC ${delta}$ , relatively little is knownof the mechanisms that result in de novo PKC ${delta}$ synthesis.

Most vertebrate transcripts contain 5' UTRs 20 to 100 nt longthat are unlikely to impede translation initiation from theAUG codon (21). We report here that the 395-nt-long 5' UTR ofPKC ${delta}$ is potentially highly structured and has the ability tosignificantly repress the translation of a linked luciferasereporter gene with rabbit reticulocyte lysate as a source oftranslation factors. From this, we speculate that PKC ${delta}$ mightbe translationally regulated in cells. In support of this, PKC ${delta}$ was recently identified as a translationally regulated genein a differential screen comparing resting and antigen-activatedhuman T lymphocytes (34). An in vivo study suggests that PKC ${delta}$ might be posttranscriptionally regulated in response to experimentallyinduced colitis in the rat as PKC ${delta}$ protein levels increase withoutan accompanying increase in mRNA abundance (4). We also reporthere that the levels of PKC ${delta}$ protein increase as HeLa cells growto a high density, while its mRNA levels remain unchanged. Thisis consistent with other studies comparing PKC ${delta}$ protein and mRNAlevels in confluent and subconfluent cultures of RPTE and 3T6cells (9, 27) and may thus represent a general feature of PKC ${delta}$ function in response to cell density-dependent growth.

A subset of transcripts that encode growth factors, signalingmolecules, and other genes involved in cell proliferation andapoptosis possess long, structured 5' UTRs that function asIRESs in a bicistronic assay (reviewed in reference 3). We provideevidence here that the 5' UTR of PKC ${delta}$ acts as an IRES both ina monocistronic construct, where cap-dependent initiation isblocked by the introduction of a stable hairpin structure, andin a bicistronic construct. The monocistronic and bicistronicluciferase plasmids used in this study have previously beenused to study the c-myc and Apaf-1 IRESs in a number of celllines (6, 48, 49, 50). Compared to the control lacking an intergenicspacer, the IRES activity of the 5' UTRs of c-myc, Apaf-1, andPKC ${delta}$ was greatest in HeLa cells and weakest in BALB/c cells,which supports the idea that some cell lines may lack noncanonicalfactors required for IRES function. However, we found that theIRES activity of the 5' UTR of PKC ${delta}$ was comparable in MCF7 andHeLa cells, whereas MCF7 cells do not support the activity ofthe c-myc or Apaf-1 IRES (6, 50). This suggests that the utilizationof cellular factors by different IRESs varies in particularcell types.

Cellular IRESs are considered likely to have an important functionin maintaining synthesis of gene products when cap-dependenttranslation is inhibited, such as during viral infection (18),hypoxia (47), apoptosis (16, 49), gamma irradiation (17), andamino acid starvation (11) and in the G₂/M stage of the cellcycle (44). We show here that the IRES activity of the 5' UTRof PKC ${delta}$ is greatest during TRAIL-induced apoptosis and when cellsgrow to a high density. A reduction in global protein synthesisand proteolytic cleavage of eIF4G is well documented duringapoptosis (5, 30, 49). With increasing cell density of normalskin fibroblasts, eIF4E protein levels decrease and there isa concomitant increase in the unphosphorylated (eIF4E-sequestering)form of 4E-BP1, suggesting a loss of cap-dependent translationin densely growing cells (14). We also show here that the IRESactivity of PKC ${delta}$ is diminished in response to serum starvation,another condition known to reduce global protein synthesis andpromote eIF4G degradation (5). These results suggest that theIRES activity of the 5' UTR of PKC ${delta}$ is not enhanced under generalconditions of cellular stress but is specifically enhanced inresponse to particular stimuli.

A cap-independent mechanism that would allow continued synthesisof PKC ${delta}$ when cells grow to a high density and during apoptosisis consistent with PKC ${delta}$ function under these conditions. Althoughan increase in PKC ${delta}$ protein levels appears to be a general phenomenonof cells growing to a high density, its role here is only speculated.PKC ${delta}$ may be required in densely growing cells to promote growtharrest and/or to prepare the cells for apoptosis. In supportof this, many studies have implied a role for PKC ${delta}$ in the inductionof growth arrest (36, 43, 53) and loss of PKC ${delta}$ is associatedwith cellular transformation and malignancy of gliomas (28,29). In most apoptotic cells, PKC ${delta}$ is activated by caspase 3-mediatedproteolysis (10). The active proteolytic product of PKC ${delta}$ is sufficientto induce apoptosis of HeLa cells (15). However, PKC ${delta}$ -mediatedapoptosis can occur in the absence of proteolytic cleavage (13).Furthermore, in response to mitochondrion-targeting toxins,a dominant-negative form of PKC ${delta}$ inhibits caspase 3 activity,suggesting that PKC ${delta}$ is also involved in a pathway leading tocaspase 3 activation (31).

In apparent contrast to its role in growth arrest and apoptosis,PKC ${delta}$ is also implicated in proliferative pathways. PKC ${delta}$ promotescellular transformation via the insulin-like growth factor Ireceptor (25), and elevated PKC ${delta}$ levels are observed in simianvirus 40-transformed rat fibroblasts and in highly metastaticmammary tumors (19, 26). Intriguingly, serum-stimulated PKC ${delta}$ phosphorylates the repressor of eIF4E, 4E-BP1, in vivo and thusis involved in the activation of cap-dependent translation (23).This perhaps serves to highlight the important roles that PKC ${delta}$ plays in many aspects of cell life, in promoting cell proliferationin response to some signals and in stimulating growth arrestand apoptosis in response to others. PKC ${delta}$ may thus representa switch for cell fate.

ACKNOWLEDGMENTS

This work was supported by a grant (049296) from the WellcomeTrust.

We are grateful to Y. Ono (Kobe University, Kobe, Japan) forthe generous gift of the PKC ${delta}$ cDNA, to James Malter (Universityof Wisconsin, Madison) for the gift of the pBKSpA plasmid, andto Simon Morley (Sussex University) for the gift of eIF4G antiserum.We are especially grateful to Anne Willis (Leicester University)and members of her group for advice and many helpful discussionsand for the gifts of TRAIL, the HRV construct, and most of theluciferase reporter plasmids used here.

FOOTNOTES

* Corresponding author. Present address: Molecular Development Laboratory, Murdoch Childrens Research Institute, Royal Children's Hospital, Parkville, Victoria 3052, Australia. Phone: 61-3-9345-6607. Fax: 61-3-9345-6000. E-mail: morrishb{at}cryptic.rch.unimelb.edu.au.

REFERENCES

Top