Cap-Independent Translation Is Required for Starvation-Induced Differentiation in Yeast

See allHide authors and affiliations

Science  31 Aug 2007:
Vol. 317, Issue 5842, pp. 1224-1227
DOI: 10.1126/science.1144467


Cellular internal ribosome entry sites (IRESs) are untranslated segments of mRNA transcripts thought to initiate protein synthesis in response to environmental stresses that prevent canonical 5′ cap–dependent translation. Although numerous cellular mRNAs are proposed to have IRESs, none has a demonstrated physiological function or molecular mechanism. Here we show that seven yeast genes required for invasive growth, a developmental pathway induced by nutrient limitation, contain potent IRESs that require the initiation factor eIF4G for cap-independent translation. In contrast to the RNA structure-based activity of viral IRESs, we show that an unstructured A-rich element mediates internal initiation via recruitment of the poly(A) binding protein (Pab1) to the 5′ untranslated region (UTR) of invasive growth messages. A 5′UTR mutation that impairs IRES activity compromises invasive growth, which indicates that cap-independent translation is required for physiological adaptation to stress.

Translation initiation is a crucial point of regulation of eukaryotic gene expression, allowing cells to adapt rapidly to changing environmental conditions. In response to glucose deprivation, haploid Saccharomyces cerevisiae cells dramatically down-regulate translation of most cellular messages, while also exhibiting striking morphological changes leading to invasive growth (1, 2). Whereas the global translational repression requires the mRNA 5′ decapping machinery (3), the developmental switch requires new protein synthesis, which suggests that proteins required for invasive growth might be translated by a cap-independent mechanism.

Many invasive growth genes have unusually long 5′ untranslated regions (5′UTRs) with the potential to form stable RNA secondary structures (table S1) (4). Furthermore, one gene required for invasive growth, YMR181c (5), is the downstream open reading frame (ORF) of a naturally occurring bicistronic cellular message (6), suggesting that invasive growth genes might be translated by a mechanism that depends on an internal ribosome entry site (IRES). To test whether the 5′UTRs of invasive growth genes are capable of internal translation initiation, we inserted these sequences into a firefly luciferase reporter (F-luc) containing a stable stem-loop structure [change in Gibbs free energy (ΔG°) = –58 kcal/mol] at the 5′ end to inhibit scanning (7) and capped with a nonphysiological ApppG cap that reduces binding of the cap-binding initiation factor (eIF4E) by three orders of magnitude (7). With no IRES inserted, the ApppG-capped hairpin RNA is poorly translated, yielding 0.4% in vitro and 0.04% in vivo compared with an m7GpppG-capped mRNA (at 100%), with an unstructured 18-nucleotide (nt) 5′UTR (Fig. 1A) (8). We confirmed by Northern blots that the reporter mRNAs were stable in both extracts and cells (Fig. 1B), which ruled out differential RNA stability as a source of differences in luciferase activity. Insertion of the 5′UTR sequences from YMR181c or from the invasive growth genes GPR1, BOI1, FLO8, NCE102, MSN1, or GIC1, resulted in efficient translation of ApppG-capped hairpin mRNAs compared with length-matched negative control constructs containing reverse-complement 5′UTR sequences. In addition, the 5′UTRs from TPK2, HMS2, and YEL033w lacked IRES activity (Fig. 1C). The invasive growth cellular IRESs' activities ranged from 8 to 33% of that of the control m7GpppG-capped mRNA (table S2). We tested the IRES-containing 5′UTRs for activity in vivo by elec-troporating the reporter mRNAs into yeast cells to avoid any possibility of mistaking cryptic promoter or splicing activity for IRES activity. All seven 5′UTRs promoted efficient cap-independent translation in vivo (Fig. 1D). Experiments with bicistronic reporters, in which the 5′ ORF is translated via cap-dependent initiation and the 3′ ORF is efficiently translated only when an active IRES is inserted between the two ORFs, corroborated our finding that the 5′UTRs of seven invasive growth genes mediate internal translation initiation (fig. S1).

Fig. 1.

Invasive growth genes have IRES activity. (A) In vitro (left) and in vivo (right) translation of m7G- and ApppG-capped poly(A)+ F-luc mRNAs. For all luciferase assays, values represent the average of three independent experiments and error bars show the standard deviation (SD). m7GpppG-capped reporters were set equal to 100%. (B) Northern blots of F-luc mRNAs isolated from (A) after incubation in vitro (left) or in vivo (right). Relative levels of full-length F-luc RNA are indicated. (C) In vitro translation of F-luc reporters containing various 5′UTR inserts from YMR181c sequences or the reverse complement (rc) or from invasive growth genes. F-luc activity was normalized to mRNA levels as determined by Northern blot. (D) In vivo translation of F-luc reporter mRNAs, normalized to mRNA levels as in (C).

A subset of viral IRESs recruits the translation machinery by providing high-affinity internal binding sites for the translation initiation factor eIF4G (9). To test whether invasive growth IRESs require eIF4G, we prepared extracts genetically depleted of eIF4G (8). Reducing eIF4G levels to 37% decreased translation from invasive growth IRES reporters to ∼25% activity (Fig. 2A), whereas translation from the eIF4G-independent cricket paralysis virus (CrPV) IRES was only slightly affected. Further depletion abolished activity (fig. S2). In cell extracts containing 9-fold overexpressed eIF4G, invasive growth IRES activity increased 10- to 20-fold (Fig. 2B), which indicated that eIF4G is limiting for IRES activity.

Fig. 2.

eIF4G promotes invasive growth IRES activity. (A) Translation in extracts from control (black bar, +) or eIF4G-depleted (white bar, –) cells. Reporters contain an m7GpppG cap (cap) or an ApppG cap plus a stable hairpin followed by IRES sequences. Average activity for each mRNA in the control extract is set equal to 1. Relative levels of eIF4G were determined by Western blot. (B) Translation of F-luc reporters from (A) in extracts from control (black, –) or eIF4G-overexpressing (gray, +) cells. Average activity for each mRNA in the control extract is set equal to 1. Relative levels of eIF4G are indicated below. (C) IRES activity in vitro (a) and in vivo (b) of the 5′UTR of eIF4G2 compared with that of YMR181c, normalized to mRNA levels.

eIF4G is the least abundant initiation factor in yeast (10) and becomes unstable in nutrient-limited cells (11), which suggests a need for continued synthesis of eIF4G protein in glucose-starved cells in order to maintain invasive growth IRES activity. We therefore tested whether the 5′UTRs from either yeast eIF4G gene are themselves capable of internal initiation using the ApppG-capped hairpin reporter. The 461-nt 5′UTR of eIF4G2 showed IRES activity similar to invasive growth genes' 5′UTRs both in vitro and in vivo (Fig. 2C). Thus, cap-independent synthesis of eIF4G2 could support eIF4G-dependent IRES activity in starved cells.

Well-characterized viral IRESs use disparate strategies for cap-independent recruitment of eIF4G, but many share a requirement for the formation of stable RNA structures (9). We performed ribonuclease-mediated structural probing of YMR181c's298-nt5′UTR, revealing extensive regions of secondary and tertiary structure (fig. S3). To determine which RNA elements are important for IRES activity, we constructed a series of deletion mutants. Deletion of the structured 5′-most 224 nt had little effect on IRES activity (fig. S4). The 60 nt immediately upstream of the AUG initiation codon were sufficient for robust internal translation initiation (Fig. 3A). This minimal IRES is almost completely unstructured (fig. S3), which suggests a sequence requirement for IRES activity. The minimal IRES sequence includes a polyadenosine [poly(A)] tract (12 out of 13 residues) preceding the AUG initiation codon, reminiscent of the leaders of vaccinia viral mRNAs that are capable of efficient cap-independent translation (12, 13). Deletion of this poly(A) sequence from the YMR181c 5′UTR reduced IRES activity both in vitro and in vivo (Fig. 3A). Deletion of the A-tract had no effect on RNA integrity or stability (Fig. 3B).

Fig. 3.

An unstructured poly(A) tract mediates Pab1-dependent IRES activity. (A) Translation of ApppG-capped F-luc reporters containing YMR181c sequences or the reverse complement (rc) in yeast extracts (a) or cells (b) as described in Fig. 1. The 12-nt sequence deleted in ΔA is underlined. (B) Northern blots of F-luc mRNAs isolated from (A). (C) In vitro translation of ApppG-capped F-luc mRNA [lacking a poly(A) tail] ±100 ng exogenous poly(A) RNA ±100 ng of recombinant Pab1. The activity from the reaction without any additions was set equal to 1. Attempts to test the effects of Pab1 depletion using antibodies required high concentrations of antibody and prolonged incubations that resulted in nonspecific inactivation of extracts. (D) Translation of m7GpppG-capped (black), ApppG-capped + IRES (white) or ApppG-capped+IRESΔA (gray) F-luc mRNAs [lacking poly(A) tails] in the presence of 100, 200, or 400 ng rPab1 or buffer. Average activity from the reaction + buffer was set equal to 1 for each construct. (E) Filter-binding assay for binding of rPab1 to various YMR181c 5′UTR sequences showing the fraction bound for each concentration of rPab1. (F) Translation in vitro (a) and in vivo (b) of F-luc reporters containing 5′UTRs of YMR181c and PAB1, normalized to mRNA levels by Northern blot.

Poly(A)+ tracts at the 3′ end stimulate eukaryotic translation through conserved poly(A)–binding proteins (PABPs) that enhance translation by at least two mechanisms: stabilizing eIF4G's interaction with the 5′ end of the mRNA through direct interactions between PABP and eIF4G and stimulating 60S ribosome subunit joining (14, 15). In principle, both of these functions of PABP could be performed by binding to poly(A) in the 5′UTR. To test the hypothesis that the yeast PABP, Pab1, stimulates cap-independent translation of YMR181c specifically through binding to the 5′UTR, we examined translation of mRNAs lacking poly(A) tails. Addition of exogenous poly(A) RNA dramatically inhibited translation of an ApppG-capped hairpin mRNA containing the 5′UTR of YMR181c, a defect that was rescued by the addition of recombinant Pab1 (Fig. 3C). Increasing Pab1 concentration specifically enhanced translation from the wild-type YMR181c 5′UTR compared with the ΔA mutant IRES or an m7GpppG-capped message lacking a poly(A) tail (Fig. 3D). To test whether Pab1 binds the YMR181c 5′UTR directly, we performed filter-binding assays with recombinant protein. Pab1 bound tightly and specifically to the YMR181c RNA with an apparent dissociation constant (Kd) of 0.25 ± 0.05 μM, compared with the reverse complement control RNA (Kd = 1.52 ± 0.13 μM); deletion of the poly(A) tract eliminated specific binding (Kd = 3.29 ± 0.41 μM) (Fig. 3E).

Taken together, these data support a mechanism for YMR181c IRES activity requiring specific binding of Pab1 to the 5′UTR and suggest that binding of Pab1 to the 5′UTR can functionally substitute for a cap and eIF4E in recruiting eIF4G. This mechanism is not unique to YMR181c: The 176-nt 5′UTR from BOI1 contains a similar AUG-proximal poly(A) tract and requires binding to Pab1 for IRES activity (fig. S5). Notably, we found that the 171-nt 5′UTR of yeast PAB1 had robust IRES activity in vitro and in vivo (Fig. 3F), and like the invasive growth IRESs, the PAB1 IRES was strongly eIF4G-dependent (fig. S6). Cap-independent production of both Pab1 (Fig. 3F) and eIF4G (Fig. 2C) could sustain necessary translation during prolonged periods of decreased cap-dependent initiation.

The discovery of IRESs in numerous eukaryotic regulatory genes suggests that IRES-dependent initiation plays a crucial role in cellular adaptation (1619). To test this hypothesis directly, we determined whether the FLO8 IRES is required for FLO8 function by creating a yeast strain in which the FLO8 gene was replaced with a mutant version lacking residues –60 to –11 in the 167-nt 5′UTR (flo8 ΔIRES). This internal deletion, which removes two poly(A) tracts (fig. S7), reduced IRES-dependent translation of a reporter gene by 60% in vitro (Fig. 4A). Deletion of nucleotides –60 to –11 from the endogenous FLO8 locus similarly reduced Flo8 protein levels in vivo (Fig. 4B) without affecting FLO8 mRNA levels (Fig. 4C), consistent with a requirement for IRES activity to maintain wild-type translation. No other feature of the FLO8 gene was altered. To test whether this reduction in Flo8 is physiologically significant, we assayed the flo8 ΔIRES mutant strain for invasive growth and for expression of FLO11 mRNA, which requires FLO8 (20). The flo8 ΔIRES mutant is defective for both invasive growth (Fig. 4D) and for transcription of FLO11 (Fig. 4C). These data strongly argue for a physiological requirement for IRES-dependent translation in yeast invasive growth. Our findings demonstrate a direct connection between IRES activity and cellular differentiation and suggest a coherent molecular mechanism for internal initiation on cellular messages that may be conserved in higher eukaryotes.

Fig. 4.

IRES-dependent translation of FLO8 is required for invasive growth. (A) In vitro translation of ApppG-capped F-luc mRNAs containing the full-length 5′UTR from FLO8 wild-type (WT) or a deletion mutant lacking nucleotides –60 to –11(Δ), normalized to mRNA levels by Northern blot (inset). (B) Western blot of Flo8 from two independent isolates each of WT FLO8 or –60 to –11Δ mutant strains containing a C-terminal hemagglutinin (HA) epitope tag. The reference band (*Ref.) is a cross-reacting protein detected in untagged strains. Relative levels of Flo8-HA are indicated below. (C) Northern blots of RNA isolated from the extracts prepared in (B). (D) Invasive growth assay for WT, Δflo8, and –60 to –11Δ strains. Portions of the cultures assayed in (B) and (C) were spotted onto rich medium and photographed after growth at 30°C. Invasive growth was photographed after washing the same plate under a gentle stream of water.

Supporting Online Material

Materials and Methods

Figs. S1 and S7

Tables S1 to S3


References and Notes

View Abstract

Navigate This Article