Carotenogenesis Is Regulated by 5'UTR-Mediated Translation of Phytoene Synthase Splice Variants

Abstract

Phytoene synthase (PSY) catalyzes the highly regulated, frequently rate-limiting synthesis of the first biosynthetically formed carotene. While PSY constitutes a small gene family in most plant taxa, the Brassicaceae, including Arabidopsis (Arabidopsis thaliana), predominantly possess a single PSY gene. This monogenic situation is compensated by the differential expression of two alternative splice variants (ASV), which differ in length and in the exon/intron retention of their 5'UTRs. ASV1 contains a long 5'UTR (untranslated region) and is involved in developmentally regulated carotenoid formation, such as during deetiolation. ASV2 contains a short 5'UTR and is preferentially induced when an immediate increase in the carotenoid pathway flux is required, such as under salt stress or upon sudden light intensity changes. We show that the long 5'UTR of ASV1 is capable of attenuating the translational activity in response to high carotenoid pathway fluxes. This function resides in a defined 5'UTR stretch with two predicted interconvertible RNA conformations, as known from riboswitches, which might act as a flux sensor. The translation-inhibitory structure is absent from the short 5'UTR of ASV2 allowing to bypass translational inhibition under conditions requiring rapidly increased pathway fluxes. The mechanism is not found in the rice (Oryza sativa) PSY1 5'UTR, consistent with the prevalence of transcriptional control mechanisms in taxa with multiple PSY genes. The translational control mechanism identified is interpreted in terms of flux adjustments needed in response to retrograde signals stemming from intermediates of the plastid-localized carotenoid biosynthesis pathway.

Figure 1.

Selective expression of alternative AtPSY variants. Transcript levels of two AtPSY ASVs with different 5′UTRs were investigated by qRT-PCR in Arabidopsis wild type. A scheme on different exon/intron retentions and the positions of amplicons used to discriminate both variants is shown in A (ASV1: 403 nucleotides, black square/dotted line; ASV2: 252 nucleotides, black circle/dashed line). Both ASVs were captured by amplification within the AtPSY ORF (ASV1/2: white triangle/solid line). Transcript levels in deetiolating seedlings (B), in roots of hydroponically grown seedlings transferred to medium containing 200 mm NaCl (C) and in leaves dark-adapted for 2 h and upon illumination with white light for 4 h (D). E, PSY protein levels and F, phytoene levels indicating PSY activity, determined by HPLC in illuminated dark-adapted leaves, incubated with norflurazon. G, Cycle threshold (C_t) values ± sd of ASV1 and ASV2 from samples with comparable 18S rRNA levels. Normalized transcript levels were expressed relative to levels prior to illumination or salt treatment, respectively; results are the average ± sd of three biological replicates; *significant difference to the wild type (Student’s t test, P < 0.05).

Figure 2.

Sequence-dependent 5′UTR-mediated control of AtPSY ORF translation in Arabidopsis leaves. A, Phylogenetic tree generated from PSY 5′UTR sequences. The Arabidopsis PSY ASV1 5′UTR was used; for full taxa names and accession numbers, see Supplemental Table S2. Bootstrap values are reported next to the branches (only bootstraps above 50% are shown). A consensus structure of PSY 5′UTR predicted from Brassicaceae PSY 5′UTRs is shown below. Different colors indicate whether base pairs are formed by one (red), two (yellow), or three (green) different combinations of nucleotides; hyphens indicate gaps within the alignment. 5′UTR truncations used in B are indicated by arrows. TS, transcription start. B, Analyses of Arabidopsis leaves overexpressing the AtPSY ORF with full-length (403 nucleotides), 330 nucleotides, and 280 nucleotides of the 5′UTR of ASV1. AtPSY transcript (top) and AtPSY protein levels (bottom); actin-normalized protein levels relative to the wild type (Wt) are shown above. Data are the average ± sd of three biological replicates; *significant difference (Student’s t test, P < 0.05).

Figure 3.

5′UTR-regulated AtPSY translation efficiency determines carotenoid levels in roots. The AtPSY ORF with different lengths of the 5′UTR from ASV1 was overexpressed in Arabidopsis. Root carotenoid content (A) increased only in lines with only 280 nucleotides or no AtPSY 5′UTR. Chloroform extracts are shown below. AtPSY protein amounts are shown in (B) with actin as loading control. C, AtPSY transcript levels. Data are the average ± sd of three biological replicates; *significant difference to the wild type (Wt; Student’s t test, P < 0.05).

Figure 4.

Rice PSY1 5′UTR has no translation inhibitory function in Arabidopsis. The AtPSY ORF containing the full-length rice PSY1 5′UTR (OsPSY1 UTR) was expressed in Arabidopsis. One line expressing the translation-permissive 280-nucleotide truncation of the AtPSY ASV1 5′UTR was included as a control (AtPSY-UTR280). Roots and leaves were analyzed for transgene transcript levels (A), AtPSY protein levels (B; actin-normalized AtPSY protein levels are shown above), and carotenoid levels, determined by HPLC (C). Root chloroform extracts are shown below. Data are the average ± sd of three biological replicates; *significant difference to wild type (wt; Student’s t test, P < 0.05). DW, Dry weight.

Figure 5.

AtPSY 5′UTR translation inhibition in vitro and effect on GUS expression. A, In vitro translation. The AtPSY ORF with various lengths of its 5′UTR was in vitro transcribed and in vitro translated in wheat germ lysate in presence of [³⁵S]Met. Incorporation into [³⁵S]AtPSY was determined by SDS-PAGE followed by autoradiography and expressed relative to the [³⁵S]AtPSY amount obtained with the 5UTR403-AtPSY mRNA version. Results are the average ± sem of five biological replicates; *significant difference to 5UTR403-AtPSY (Student’s t test, P < 0.05). B, GUS translation efficiency. The GUS ORF with different lengths of the AtPSY 5′UTR was expressed in Arabidopsis. GUS enzyme activity was normalized to corresponding GUS transcript levels in leaves of 4-week-old plants and expressed relative to one reference line. Results are average ± sd of three biological replicates from three events per transformation; *significant difference to 5UTR403-GUS-expressing line (Student’s t test, P < 0.05).

Figure 6.

Conformational switch analysis of AtPSY and OsPSY1 5′UTRs. Energy barrier (EB in kcal mol⁻¹) distances of 5′UTR RNA secondary structures to respective consensus structures were predicted with paRNAss. Two clouds of energetically separable RNA structures are predicted for translation-inhibitory AtPSY 5′UTR versions (A, full length, 403 nucleotides; B, 330 nucleotides), while only one cloud is predicted for translation-permissive AtPSY 5′UTR truncation (C, 280 nucleotides). In contrast, distance calculations for the full-length OsPSY1 5′UTR spread over the diagram and do not form distinct clusters, which is in agreement with the absence of a translation-regulatory function (D).

Figure 7.

Carotenoid flux negatively regulates Arabidopsis PSY protein levels in leaves. PSY ORFs from maize (Zm), rice (Os), and Arabidopsis (At) were expressed in Arabidopsis. In seed-derived callus (A) and roots (B), this results in increased carotenoid levels compared to wild-type control (wt). Representative calli are shown above A. In contrast, in leaves, carotenoids were as in the wild type (C), although pathway flux was increased, as determined by in vitro PSY activity assays with isolated chloroplast membranes, incubated with DMAPP, [1-¹⁴C]IPP, and GGPP synthase (D). In leaves, this did not affect transcript levels of both AtPSY ASVs determined by qRT-PCR (E) but down-regulated protein levels of endogenous AtPSY. An immunoblot using AtPSY-specific antibodies and actin as loading control is shown in F; quantification of normalized AtPSY protein levels relative to wild type is shown above. As expected, AtPSY levels are strongly increased in the AtPSY-overexpressing line. Data are average ± sd of three biological replicates. DW, dry weight; *significant difference to the wild type (Wt; Student’s t test, P < 0.05).

Figure 8.

Model for the posttranscriptional regulation of PSY in Arabidopsis. Regulation of PSY protein amount entails two translational control elements realized by ASVs differing in their 5′UTR length. ASV1 contains 403-nucleotide 5′UTR, which coordinates translation with the carotenoid pathway flux and is suggested to be regulated by plastid-derived apocarotenoids. Translation inhibition requires a hairpin loop the elimination of which results in a translation-permissive 5′UTR. This structural element is absent in the 252-nucleotide 5′UTR of ASV2, which therefore bypasses feedback-regulated translational control and is able to quickly increase PSY protein levels, e.g. for higher carotenoid consumption by photo-oxidation in illuminated leaves or for ABA biosynthesis in salt-stressed roots.