Riboswitch-dependent gene regulation and its evolution in the plant kingdom

Abstract

Riboswitches are natural RNA sensors that affect gene control via their capacity to bind small molecules. Their prevalence in higher eukaryotes is unclear. We discovered a post-transcriptional mechanism in plants that uses a riboswitch to control a metabolic feedback loop through differential processing of the precursor RNA 3' terminus. When cellular thiamin pyrophosphate (TPP) levels rise, metabolite sensing by the riboswitch located in TPP biosynthesis genes directs formation of an unstable splicing product, and consequently TPP levels drop. When transformed in plants, engineered TPP riboswitches can act autonomously to modulate gene expression. In an evolutionary perspective, a TPP riboswitch is also present in ancient plant taxa, suggesting that this mechanism is active since vascular plants emerged 400 million years ago.

Figure 1.

TPP-dependent alternative splicing in the 3′ UTR of Arabidopsis and tomato THIC genes. (A) Schematic representation of the THIC pre-mRNA and the two RNA splicing products, marking the TPP riboswitch, introns 1 and 2 (dark gray), and exons (light gray). Numbers indicate oligonucleotides used for experiments described in B (see Supplemental Material). (B) Transcript levels in TPP-treated Atthi1 mutant seedlings resolved by qPCR experiments (n = 3; P < 0.01; SEM; see Materials and Methods for details). (var.) Variant. The oligonucleotides (see A) used were #21 and #22 (coding seq.), #23 and #24 (intron-retention var.), and #25 and #26 (intron-spliced var.), and AtACTIN2 was used as endogenous control (oligonucleotides #34 and #35). (C) End-point analysis RT–PCR products that monitor the levels of the THIC RNA splicing products in the tomato thiamin auxotroph mutant strain grown under increasing TPP concentrations. The two splicing variants were amplified with specific oligonucleotides (#8 and #9, and #8 and #10), and the coding region was amplified with oligonucleotides derived from conserved regions (#8 and #11). The intron-retention variant, the intron-spliced variant, and the coding sequence were amplified for 31, 35, and 33 cycles, respectively. The effect of TPP addition on whole plants is shown at the bottom of each panel. (D) Stability of the AtTHIC splice variants and coding sequence during variable time points after the addition of the transcription inhibitor actinomycin D to Arabidopsis cell suspension culture supplemented with 1 mM TPP, as resolved by qPCR using the same oligonucleotides as in B (n = 3; P < 0.01; SEM). AtUBIQUITINC was used as the endogenous control (oligonucleotides #27 and #28). The decay of the intron-spliced variant was on average 63% per hour.

Figure 2.

TPP riboswitch-dependent alternative splicing in transgenic Arabidopsis plants. (A) Schematic representation of the constructs used to generate transgenic Arabidopsis plants and protoplasts expressing a YFP reporter gene that was transcriptionally fused to the native AtTHIC 3′ UTR (i), the same 3′ UTR but with a point mutation disabling riboswitch ligand binding (A515G) (ii), the same 3′ UTR but with a point mutation in the 5′ splice (spl.) site of int2 that prevents splicing in this region (iii) (Fig. 1A), and YFP transcriptionally fused to the NOPALINE SYNTHASE (NOS) terminator (iv). (B) RT–PCR end-point analysis experiments in which the transcript levels of the YFP RNA splicing products and the coding region were monitored in transgenic plants grown with or without additional TPP. The intron-retention variant, the intron-spliced variant, and the coding sequence were amplified for 18, 22, and 18 cycles, respectively. (C) RT–PCR end-point analysis experiments in which transcript levels of the YFP RNA splicing products and the coding region were detected in Arabidopsis protoplasts transiently transformed with the corresponding constructs, and incubated with or without additional TPP. The intron-retention variant, the intron-spliced variant, and the coding sequence were amplified for 22, 22, and 27 cycles, respectively. (D) YFP expression in transgenic Arabidopsis plants determined by fluorescence and Western blot analyses.

Figure 3.

The TPP riboswitch is present in two genes of the thiamin biosynthesis pathway in ancient plants. The presence of the TPP riboswitch in the 3′ UTR of the THIC and THI1 genes across the plant kingdom is depicted. (np) Not present, based on analysis of whole genome sequences; (nd) not detected in PCR assays.

Figure 4.

(A) Schematic representations of the TPP-binding domain of P. patens (Pp) located in the 3′ UTRs of the PpTHIC and PpTHI1-2 genes. Exon and intron sequences are indicated by uppercase and lowercase letters, respectively. Arrows indicate the adenine base that forms the crucial assembly platform of the TPP riboswitch (Serganov et al. 2006; Thore et al. 2006); the arrowhead marks the 3′ splice site. Shadowed nucleotides are mostly conserved among TPP riboswitches (Sudarsan et al. 2003). (B) RT–PCR end-point analysis products that monitor the transcript levels of the PpTHIC and PpTHI1 genes in P. patens (Pp) protomemata grown without or with additional TPP. (PDS) Phytoene desaturase transcript was used for calibration. PpTHIC and PpTHI1 genes were amplified for 25 cycles. (C) Apparent Kd for TPP binding by the TPP riboswitch of Arabidopsis (AtTHIC) and P. patens (PpTHI1-2), determined by in-line probing (Sudarsan et al. 2003). (D) Fragmentation pattern of the in vitro-transcribed 80-bp RNA molecule that comprises the P. patens THI1-2 TPP–riboswitch. The in-line probing assay was performed in the presence of increasing TPP concentrations (0–10 μM). (NR) Nonreacted RNA; (OH⁻) RNA subjected to alkaline degradation; (T1) RNA submitted to RNase T1 digestion.

Figure 5.

A model for TPP riboswitch-mediated regulation of gene expression via alternative splicing in vascular plants. The 3′ splice site is highlighted with a star.