Alternative splicing in plants: current knowledge and future directions for assessing the biological relevance of splice variants

Abstract

Alternative splicing is an important regulatory process that produces multiple transcripts from a single gene, significantly modulating the transcriptome and potentially the proteome, during development and in response to environmental cues. In the first part of this review, we summarize recent advances and highlight the accumulated knowledge on the biological roles of alternative splicing isoforms that are key for different plant responses and during development. Remarkably, we found that many of the studies in this area use similar methodological approaches that need to be improved to gain more accurate conclusions, since they generally presume that stable isoforms undoubtedly have coding capacities. This is mostly done without data indicating that a particular RNA isoform is in fact translated. So, in the latter part of the review, we propose a thorough strategy to analyze, evaluate, and characterize putative functions for alternative splicing isoforms of interest.

Alternative splicing, Arabidopsis, crops, environmental stress, plant development

Introduction

Climate change is altering the environments in which all organisms grow and develop. As sessile organisms rooted to the soil, plants cannot walk, run, fly, or swim toward food or away from environmental stresses (i.e., drought, salinity, and extreme temperatures). Nonetheless, plant species respond plastically to those changes in their environment and adapt to novel conditions (Bradshaw, 1965, 2006; Nicotra et al., 2010). In this sense, the ability to process environmental information and adjust phenotypes optimally within each generation to face daily, seasonal, and annual changes (phenotypic plasticity) is critical for plant survival, growth, and reproduction (Liancourt et al., 2013). Environmental changes can elicit plastic responses and shifts in the timing of developmental transitions. There is an increasing body of evidence showing that post-transcriptional processes substantially increase transcriptome complexity and play an important role in modulating gene expression in response to internal and external cues. Gene expression can be regulated through transcriptional and co-transcriptional or post-transcriptional mechanisms. Among these, alternative splicing, a co-transcriptional mechanism by which more than one mRNA can be produced from a single gene, plays multiple roles in plant responses by integrating endogenous developmental and exogenous environmental signals (Lee et al., 2013; Posé et al., 2013), and is the most important contributor to transcriptome diversification in both plants and animals. Furthermore, changes in chromatin (epigenetics) and/or modifications in RNA bases can also have profound effects on the processing and stability of RNA molecules (Shen et al., 2016; Xu et al., 2022).

Even though most of the published studies are mainly focused on transcriptomic analyses showing that numerous genes are alternatively spliced in response to changes in environmental conditions (light, temperature, salinity, drought), in different cells or tissues (seeds, roots, leaves), and at different times in the life cycle (seed maturation, flowering), a significantly minor proportion actually assess the biological implications and functional relevance of these environmentally regulated alternative splice variants in plant responses. For example, a hypothetical question we may ask is whether the splicing isoforms X.1, X.2, and X.3 from a gene X, shown by a transcriptomic approach (i.e. RNA-seq) to change their expression in response to ambient changes, really have a biological role in regulating specific plant responses. Relatively few researchers take a further step after the bioinformatic assessment and molecular characterization of specific isoforms and analyze whether the isoform(s) really has a role in a physiological response. For example (and to mention just one of the many examples we will discuss in this review), FLOWERING LOCUS M (FLM) undergoes alternative splicing (Scortecci et al., 2003) and produces different ratios of splicing variants in response to shifting ambient temperature, and in this way the ratio of the two most abundant splicing isoforms (FLMβ and FLMδ), encoding proteins with antagonistic effects, determines floral transition (Posé et al., 2013).

In this review we explore, summarize, and highlight the accumulated knowledge on the biological relevance and function of alternative splicing isoforms functioning in plant responses, and provide insights into the latest findings on this topic, focusing not only on the model plant Arabidopsis thaliana (Arabidopsis) but also on other species. In the first part of the review, we describe general aspects of alternative splicing in plants. In the second part, we focus on how specific isoforms from several genes are involved in the regulation of plant responses during plant growth and development and upon environmental changes. Finally, we present some prospects for future research.

Alternative splicing in plants

Environmental signals can regulate the transcription of thousands of genes while also affecting layers of gene expression other than transcription initiation, such as 5ʹ end capping, splicing, 3ʹ end processing, export, translation, and RNA stability/degradation. These regulatory mechanisms can be directly related to transcription, that is, being co-transcriptional or post-transcriptional. Moreover, changes in chromatin (epigenomic) and/or modifications in RNA bases (epitranscriptomic) can have profound effects on the processing and stability of RNA molecules (Shen et al., 2016; Xu et al., 2022). Environmental cues such as temperature and light shape the plant transcriptome by impacting on each possible level of gene expression.

RNA splicing, a largely co-transcriptional molecular event, is one of the steps in RNA processing that involves the removal of some regions (introns) while joining others (exons). This process requires more than 200 proteins and five small RNAs, which are associated with the spliceosome (Wahl et al., 2009). Alternative splicing produces multiple mRNAs from a single gene by differentially regulating the selection of splice sites, significantly modulating the transcriptome and, to some extent, the proteome, during development and in response to environmental cues (Yu et al., 2016; Chaudhary et al., 2019). The genes that are subjected to alternative splicing encode proteins with regulatory functions, and this mechanism impacts on protein functions such as binding properties, enzymatic activity, intracellular localization, post-translational modification, or protein stability (Zhu et al., 2017). This requires precise regulation to guarantee plasticity while maintaining high specificity and fidelity. The selection of alternative splice sites is mediated by trans-acting splicing factors, such as serine/arginine-rich (SR) proteins. These factors bind to cis-elements on the pre-mRNA to promote or inhibit the recruitment of spliceosome components to the adjacent alternative splice sites (Kornblihtt et al., 2013). Therefore, the regulation of alternative splicing depends on the expression level and post-translational modification of SR proteins and other splicing factors (Syed et al., 2012). In addition, since splicing reactions occur mainly while transcription takes place (co-transcriptionally), the regulation of RNA polymerase II transcription affects the splicing outcomes (Beyer and Osheim, 1988; Jabre et al., 2019; Zhu et al., 2020). The splicing machinery and mechanisms are shared across all eukaryotic phyla to some extent, such as the RNA sequences that define exon/intron boundaries, spliceosome components, and splicing factors, which are indeed conserved across all eukaryotes (Reddy, 2007; McGuire et al., 2008; Ling et al., 2019). Nonetheless, there are some pronounced differences in splicing among eukaryotes; for example, the average length of introns and exons varies dramatically between plants and animals, and between different species as well. Moreover, the target branch point sequence of the splicing factor 1 (SF1 or branch point bridging protein, BBP) in plants is not conserved compared with other eukaryotic organisms (Lorković et al., 2000). Consistent with this, SF1 may have a different mechanism of 3ʹ splice site (ss) recognition since its homologs contain a different RNA recognition motif domain (Worden et al., 2009; Park et al., 2019). The frequency of each alternative splicing event also differs, with intron retention being the most conspicuous in plants and exon skipping the most frequent in animals (Chaudhary et al., 2019).

An intron-containing gene can potentially produce several different isoforms, depending on the intron–exon structure. The most common types of alternative splicing are: (i) exon skipping (an exon is alternatively skipped and results in a shorter mature mRNA), (ii) alternative 5ʹ ss (the spliceosome may recognize an alternative 5ʹ ss, within an exon, to produce a shorter transcript), (iii) alternative 3ʹ ss (a 3ʹ ss within an exon is recognized by the spliceosome), (iv) intron retention (the spliceosome retains a full intron in the mature mRNA), (v) exitron splicing (alternatively spliced internal regions of protein-coding exons. Transcripts with retained exitrons exit the nucleus for translation into different proteins in a tissue-specific manner; Marquez et al., 2015), and (vi) mutually exclusive exons (an exon-skipping-derived mechanism where an exon is included and another exon is excluded, so the mature mRNAs never contain both exons) (Fig. 1). According to the high-throughput transcriptomic dataset for Arabidopsis (AtRTD2), IR is the most common event (40%), followed by alternative 3ʹ ss, alternative 5ʹ ss, exon skipping, and mutually exclusive exons (Zhang et al., 2017). More recently, Zhang et al. (2022) reported the construction of a new Arabidopsis transcriptome (AtRTD3), based on a wide range of tissues and treatments including the use of long reads from Iso-seq (PacBio), and showed that although the number of transcripts is more than doubled in AtRTD3, the number of multi-exonic protein-coding genes that have alternative splicing is similar to that found in AtRTD2 (60.4%). The increased number of protein variant transcripts in AtRTD3 includes transcripts from the same genes with alternative transcription start sites (TSS) and alternative polyadenylation sites, and the identification of novel alternative splicing events that alter coding sequences. In AtRTD3, authors also found an increased number of unproductive transcripts, including transcripts with the same premature termination codon (PTC)-generating alternative splicing event but with alternative TSS and polyadenylation sites (Zhang et al., 2022).

Fig. 1.

Schematic representation of the major types of alternative splicing events. Alternative splicing generates multiple mRNAs from a single gene by differentially regulating the selection of splice sites, significantly modulating the transcriptome and, to some extent, the proteome, during development and in response to environmental cues (Yu et al., 2016; Chaudhary et al., 2019). In Arabidopsis, intron-retention splice variants containing premature termination codons (PTCs) can escape the nonsense-mediated mRNA decay (NMD) pathway due to nuclear retention. Most of the transcripts containing PTCs are neither degraded nor translated as they never reach the cytosol. However, some transcripts might reach the cytosol and could be substrates of the degradation (NMD pathway) or translation machineries. UTR, untranslated region.

Alternative splicing is also associated with the regulation of mRNA stability. The most usual mechanism that regulates the abundance of transcripts is nonsense-mediated mRNA decay (NMD), a cytoplasmic RNA degradation system. Generally, isoforms that are degraded through this pathway contain intron retentions, PTCs, and upstream open reading frames, among others (Breitbart et al., 1987). Approximately 10–15% of splice variants are coupled with NMD in Arabidopsis (Kalyna et al., 2012), supporting the notion that alternative splicing mediates the controlled turnover of gene transcripts. However, these transcripts can also bypass this pathway and code for truncated proteins, which regulate key functions under stress and/or control the abundance of their full-length counterparts (Filichkin and Mockler, 2012; Kalyna et al., 2012; Drechsel et al., 2013; Filichkin et al., 2015).Thanks to recent advances in sequencing technologies, it is possible to estimate the proportion of genes subjected to alternative splicing more accurately. It has been reported that approximately 95% of human and over 60% of plant (the percentage varies depending on the plant species) undergo alternative splicing (Pan et al., 2008; Wang et al., 2008; Chamala et al., 2015; Marquez et al., 2015; Zhang et al., 2017). Moreover, alternative splicing substantially increases transcriptome complexity and plays an important role in modulating gene expression in response to internal and external cues (Lightfoot et al., 2008; Matsumura et al., 2009; Sanchez et al., 2010; Martín-Trillo et al., 2011; James et al., 2012; Jones et al., 2012; Rosloski et al., 2013).

Alternative splicing variants and their roles in plant growth and development

Transition from embryogenesis to seed germination

The ‘life’ of a seed is characterized by two major phase transitions: embryogenesis to seed maturation, and dry seed to seed germination. One of the most important pathways regulating grain size during embryogenesis is the G-protein signaling pathway. This pathway comprises three proteins, including Gγ. In wheat, Ta-GS3 produces five variants by alternative splicing: the constitutive Ta-GS3.1, an alternative 3ʹ ss Ta-GS3.2, an alternative 5ʹ ss Ta-GS3.3, an intron retention Ta-GS3.4, and an exon skipping Ta-GS3.5 splice variant (Ren et al., 2021). These variants play different roles in determining the weight and size of the grain when overexpressed. While Ta-GS3.1 overexpression significantly reduces grain weight and length, Ta-GS3.2–3.4 overexpression does not alter them and Ta-GS3.5 overexpression significantly increases grain weight and length. Furthermore, analysis of the protein–protein interactions between the isoforms of Ta-GS3 and WGB1 (another Gβ pathway protein) reveals that Ta-GS3.2, Ta-GS3.3, and Ta-GS3.4 interfere with the function of Ta-GS3.1 by competitively forming functional Gβγ heterodimers but Ta-GS3.5 cannot form a Gβγ heterodimer with WGB1. Taken together, these observations support the idea that the TaGS3-mediated signal transduction pathway regulates grain weight and size through alternative splicing (Ren et al., 2021).

Regulation of seed dormancy and germination must be precisely timed so that they occur under environmental conditions that permit growth and survival (Donohue et al., 2005; Baskin and Baskin, 2014; Graeber et al., 2014). In recent years, our knowledge on the relevance of alternative splicing in seeds has increased. Even though little is known about the role of alternative splicing during seed germination, growing evidence suggests that alternative splicing regulates seed germination by mainly targeting the abscisic acid (ABA) signaling pathway. In the absence of ABA, group A protein type 2C phosphatases (PP2Cs) are negative regulators of ABA signaling and plant adaptation to stress. PP2Cs interact with subclass III SNF1-related protein kinases (SnRK2.2, SnRK2.3, and SnRK2.6) to dephosphorylate and inhibit their kinase activity, thereby turning off ABA signaling. HYPERSENSITIVE TO ABA1 (HAB1) is among the best-characterized PP2C genes in ABA signaling (Rodriguez et al., 1998). HAB1 is broadly expressed in various tissues and organs and is induced by ABA treatment (Saez et al., 2004). In Arabidopsis, At-HAB1 yields two main isoforms: the full-length At-HAB1.1, which contains four coding exons (six in total) and gives rise to a protein that interacts with SnRK2.6, inhibiting its activity and turning off ABA signaling; and At-HAB1.2, which retains the fourth intron and encodes a putative smaller protein due to a PTC resulting in a premature translation arrest and loss of phosphatase activity (Wang et al., 2015; Xue et al., 2018). The proteins encoded by At-HAB1.1 and At-HAB1.2 play opposing roles in the regulation of seed germination and post-germination developmental arrest induced by ABA (Fig. 2). Whereas At-HAB1.1 promotes seed germination, At-HAB1.2 is a positive regulator of ABA signaling and prevents seed germination (Wang et al., 2015). RBM proteins are important regulators in pre-mRNA splicing, and many of them are involved in the regulation of development and the stress response (Reddy et al., 2013). Interestingly, RBM25 is a key regulator of At-HAB1 alternative splicing. The increased ABA sensitivity of rbm25 seedlings is caused by reduced expression of the At-HAB1.1 isoform and an increased HAB1.2/HAB1.1 ratio. This high ratio is associated with delayed seed germination and post-germination developmental arrest, suggesting that At-HAB1.2 may facilitate ABA-induced germination delay and post-germinative growth arrest (Wang et al., 2015). More recently, by using a Cas9-directed base editor, Xue et al. (2018) converted the 5ʹ ss in At-HAB1 from the active GT form to the inactive AT form. Silencing the alternative splicing of At-HAB1.1 validated its function in ABA signaling (Xue et al., 2018).

Fig. 2.

Alternative splicing changes during plant growth and development. Alternative splicing acts as a crucial regulatory mechanism that fine-tunes plant responses across the life cycle of Arabidopsis. This co-transcriptional mechanism regulates seed germination mainly by targeting the ABA signaling pathway. In this scenario, the alternative splicing of At-HAB1, a PP2C involved in ABA signaling, influences the germination response of the seeds. While At-HAB1.1 promotes seed germination, At-HAB1.2 represses it, since it is a positive regulator of ABA signaling (Wang et al., 2015). Seed germination and post-germination development are different but interconnected processes with distinct regulatory mechanisms. One of the most studied light-regulated developmental pathways in plants is photomorphogenesis. At-HYH functions in hypocotyl elongation and is subjected to alternative splicing, giving rise to four isoforms (At-HYH.1–At-HYH.4). All At-HYH variants are induced by light and have similar roles in hypocotyl development, although higher transcript levels of At-HYH.2–HYH.4 than those of At-HYH.1 are found in these tissues (C. Li et al., 2017). The transition from vegetative growth to flowering is the most drastic change for plant reproductive success. Among other auxin response factors (ARFs), At-ARF8, which plays a key role in flower organ development, is subjected to alternative splicing, giving rise to four splice variants (At-ARF8.1–At-ARF8.4). The splice variant At-ARF8.4 has a specific role in the control of stamen elongation (Ghelli et al., 2018). Exons are shown as colored boxes; introns are shown as straight lines; untranslated regions are shown as dashed vertical black lines inside exons. Solid lines represent canonical splicing and dashed lines represent alternative splicing.

More examples of how alternative splicing variants play key roles in the regulation of seed dormancy and germination (e.g. At-ABI3, At-DOG1, and At-PIF6, among others) have been reviewed elsewhere (Tognacca et al., 2020; Liu et al., 2021; Tognacca and Botto, 2021; Kashkan et al., 2022b).

Post-germination development and seedling establishment

Seed germination and post-germination development are different and interconnected processes with distinct regulatory mechanisms. Post-germination seedling establishment denotes the developmental window after germination that involves the opening, greening, and expansion of the cotyledons, marking the switch to autotrophic development (Weitbrecht et al., 2011). Light is an important environmental cue that influences root growth (Yokawa et al., 2014) and auxin is the main signaling molecule in this pathway (Laxmi et al., 2008; Sassi et al., 2012). Auxin is actively distributed within the plant by efflux-dependent cell-to-cell movement (Friml, 2003). Due to the directional auxin transport, plants adapt to rapid changes in environmental cues. This is mediated by key regulators, mainly the PIN-FORMED (PIN) auxin efflux carriers. PIN3, PIN4, and PIN7 are required for a broad range of morphogenetic and tropic processes (Adamowski and Friml, 2015). Several genes involved in auxin-dependent processes undergo alternative splicing (Hrtyan et al., 2015), as is the case for At-PIN7. The alternative 5ʹ ss at the end of the first exon of At-PIN7 results in two transcripts (At-PIN7a and At-PIN7b) that differ by the presence of a four-amino-acid stretch. This sequence corresponds to the protein motif located in the long internal hydrophilic loop of the transporter (Kashkan et al., 2022a). By using the fluorescent reporters PIN7a-GFP and PIN7b-RFP, which allow monitoring of the activity of the alternative splicing of At-PIN7 in planta and in situ, Kashkan et al. (2022a) showed overlapping PIN7a-GFP and PIN7b-RFP reporter expression in the root tip, in the etiolated hypocotyl following 4 h of unilateral light stimulation, and in the area covering the apical hook. Surprisingly, At-PIN7a and At-PIN7b regulate apical hook development and hypocotyl bending in a mutually antagonistic manner. Moreover, At-PIN7a, but not At-PIN7b, is required for proper formation of auxin maxima in planta. As the authors discuss, the example of the alternative splicing of At-PIN7 is among rarely described instances where the mutually antagonistic effects of two splice isoforms are observed and can be placed in a developmental context (Kashkan et al., 2022a).

INDOLE-3-BUTYRIC ACID RESPONSE5 (IBR5) is another gene that regulates plant auxin responses. In Arabidopsis, At-IBR5 undergoes alternative splicing, generating two isoforms, At-IBR5.1 and At-IBR5.3. The At-IBR5.1 isoform exhibits phosphatase catalytic activity that is required for both proper degradation of Aux/IAA proteins and auxin-induced gene expression. Interestingly, the At-IBR5 isoforms exhibit different localization patterns that are correlated with their functions: while At-IBR5.1 localizes to the cytosol and the nucleus, At-IBR5.3 is exclusively localized to the nucleus. When the At-IBR5.1 and At-IBR5.3 isoforms are overexpressed in wild-type and ibr5 (ibr5-1 null mutant, ibr5-4 a catalytic site mutant, and ibr5-5 a splice site mutant) backgrounds, only At-IBR5.1 complements the auxin-insensitive primary root elongation phenotype and the interdigitation defect of leaf epidermal cells of the three mutant alleles. However, both At-IBR5.1 and At-IBR5.3 isoforms are needed to complement the defective lateral root phenotype of ibr5-1 and ibr5-4 (Jayaweera et al., 2014).

Photomorphogenesis is one of the best-studied light-regulated developmental pathways in plants. In Arabidopsis, LONG HYPOCOTYL5 (HY5) and HY5 HOMOLOG (HYH) function in root development during de-etiolation. The transcription factor HY5 promotes photomorphogenesis (Ang and Deng, 1994; Lee et al., 2007; Li et al., 2010). At-HYH is subjected to alternative splicing, giving rise to four isoforms (At-HYH.1–At-HYH.4) generated by intron retention and alternative 3ʹ ss (Fig. 2). At-HYH.2 is the longest transcript, and At-HYH.1 contains one more codon than At-HYH.4 at the 3ʹ ss. In addition, both isoforms contain an evolutionarily conserved NAGNAG acceptor splicing site, with transcripts spliced after the first and second NAG, respectively. All At-HYH isoforms retain the bZIP DNA-binding domain (C. Li et al., 2017). In the hypocotyls of 5-day-old light- and dark-grown seedlings, the expression of the four At-HYH variants is induced by light in a similar manner to At-HY5, and overexpression of the four At-HYH transcripts revealed no obvious effects on wild-type hypocotyl elongation. However, all variants complemented the hy5 elongated-hypocotyl phenotype, pointing to their functional conservation and redundancy.

The transcription factors B-BOX DOMAIN PROTEIN 20 (BBX20), BBX21, and BBX22 are essential partners of HY5 to promote photomorphogenesis in Arabidopsis (Bursch et al., 2020). Other BBX members, such as BBX24 and BBX25, function as negative regulators to fine-tune photomorphogenesis by interacting with HY5 to down-regulate the expression of BBX22 (Gangappa et al., 2013). A recent study by Huang et al. (2022) sought to identify full-length transcriptomes in early Arabidopsis etiolated and de-etiolating seedlings by using Iso-seq, and aimed to assess whether alternative splicing variants constitute regulatory roles in photomorphogenic development. The authors found 212 transcription factors with intron retention variants, and variants showing evidence of sequences protected by ribosomes. Among them, the intron retention variants At-BBX22IR and At-BBX24IR encode truncated protein products lacking the VP-like motif (a COP1-interacting motif that regulates the protein stability of BBX24). Hypocotyls of transgenic plants overexpressing At-BBX22IR are significantly shorter than the wild type when grown under continuous white light for 4 d, but no noticeable phenotypic differences for 4-day-old dark-grown seedlings were found, suggesting that the intron-retention At-BBX22IR variant positively regulates the light-mediated inhibition of hypocotyl elongation during photomorphogenesis (Huang et al., 2022). On the contrary, At-BBX24IR suppresses the function of BBX24 in photomorphogenic development, since seedlings of transgenic HA-BBX24IR show a light-hypersensitive phenotype, suggesting that the intron-retention At-BBX24IR variant functions as a positive regulator in this process, in contrast to the negative role of BBX24 in photomorphogenic development (Huang et al., 2022).

We have previously mentioned the important roles that the SR protein family plays in various cellular processes such as hormonal signaling, developmental processes, and adaptation to abiotic stresses (Zhang and Mount, 2009; Li et al., 2021). Even though the pre-mRNAs of SRs undergo alternative splicing in response to environmental change (Palusa et al., 2007; Tanabe et al., 2007; Reddy and Shad Ali, 2011; Ding et al., 2014), the biological functions of most splice variants are still unknown. In Arabidopsis, At-SR45, encoding an SR-like splicing regulator, undergoes alternative splicing and generates two isoforms, the long At-SR45.1 and the short At-SR45.2. In At-SR45.1, an alternative 3ʹ ss selection event at the beginning of the seventh exon results in the inclusion of an additional 21 nucleotide in-frame sequence. The proteins encoded by the two isoforms differ in eight amino acids (Palusa et al., 2007; Zhang and Mount, 2009). Interestingly, both isoforms have distinct biological functions in plant development. Studies using transgenic plants overexpressing different At-SR45 constructs in the sr45-1 mutant background for in vivo functional studies show that although both isoforms are expressed in all tissues in the overexpressing lines, At-SR45.1 function is more important for flower development while At-SR45.2 function is more important for root growth, in both cases by directly regulating the alternative splicing pattern of splicing factor genes (Zhang and Mount, 2009). As the authors discuss, it is possible that some type of post-translational modification selectively represses At-SR45.1 activity in root and At-SR45.2 activity in inflorescence, or activates At-SR45.2 in root and At-SR45.1 in inflorescence, since both transcript variants were detected in both tissues (Zhang and Mount, 2009).

Transition from vegetative growth to flowering

The transition from vegetative growth to flowering is the most drastic change for plant reproductive success. Plants initiate the floral transition in response to environmental and endogenous cues. The formation of gametes in flowering plants (angiosperms) takes place at the final stage of plant development, unlike in metazoans, where it takes place at an early embryonic stage (Wylie, 1999). Plant gametogenesis occurs in defined organs: anthers, in the case of male gametes (microspores), through microsporogenesis; and ovules, in the case of female gametes (megaspores), through megasporogenesis. The male gametes produce haploid pollen grains that fertilize the embryo sac, the female gametophyte, and this is the progenitor of the seed (Wilson and Yang, 2004). Regarding the signaling mechanisms underlying gametogenesis, the Plant Intracellular Ras-group Leucine-rich repeat (PIRL) genes that encode a plant-specific class of leucine-rich repeat proteins related to Ras-interacting LRRs are of relevance (Forsthoefel et al., 2013). At-PIRL6 is exclusively expressed in flowers and has an important role in both male and female gametogenesis. It has a three-exon structure and undergoes alternative splicing to produce seven transcripts, comprising a functional transcript and six other variants with residual intron sequence(s) and consequent PTCs. All are located within the highly conserved LRR domain, suggesting that the alternative transcripts go untranslated and/or encode non-functional truncated products. At-PIRL6 undergoes unproductive alternative splicing outside the gametophytes. In the gametophytes, functional transcripts are detected in wild-type but not in sporocyteless (spl) mutant flowers, which do not produce gametophytes. The splice variants expressed in vegetative organs and spl mutant flowers are accumulated in the NMD mutant At-upf3, indicating that they are normally subjected to degradation. At-PIRL6 knockdown lines have a higher percentage of aborted ovules and pollen than the wild type. Hence, unproductive isoforms are produced either in non-reproductive organs or in organs where they have no function and then undergo NMD. This evidence suggests that alternative splicing serves a negative regulatory function to minimize At-PIRL6 protein expression outside gametophytes and may reflect the need for strict control in tissues where it is not required (Forsthoefel et al., 2018).

CYCLIN-DEPENDENT KINASE G1 (CDKG1) is also relevant for microsporogenesis and microgametogenesis (Cavallari et al., 2018), and to maintain fertility in Arabidopsis (Nibau et al., 2020). The alternative splicing of At-CDKG1 is temperature dependent. Two isoforms of At-CDKG1 have been found in wild type Arabidopsis, the long (At-CDKG1L) and short (At-CDKG1S) variants, which are distinguished by the retention or removal of intron 1. If the intron is retained, the nuclear-localized full-length At-CDKG1L protein is produced. The removal of the first intron produces the shorter nuclear- and cytoplasmic compartment-localized At-CDKG1S isoform, which lacks two out of four SR domains for protein–protein interaction and the nuclear localization signal (Nibau et al., 2020). At-CDKG1L is dominant at low temperatures, whereas At-CDKG1S increases in abundance at high temperatures (Cavallari et al., 2018). Furthermore, the alternative splicing of At-CDKG1 is also regulated by At-CDKG2 and CYCLIN L1 (CYCL1), and the ratio between the two variants is temperature dependent. When At-CDKG1L is expressed in the cdkg1 background, the mutant phenotype is rescued, restoring fertility, but it is not rescued when At-CDKG1S is overexpressed. In cdkg1 mutant lines, pollen grains are unviable, and viability is recovered when At-CDKG1L is expressed. However, increased expression of At-CDKG1L in the wild type Col-0 background decreases pollen formation and fertility. Hence, this regulatory mechanism may be important for the down-regulation of the signaling pathway where this protein is involved (Nibau et al., 2020).

Auxin regulates the activity of the auxin response factors (ARFs) by modulating the degradation of Aux/IAA transcriptional repressors. ARFs bind to auxin response elements in auxin-regulated promoters of downstream target genes to control their expression (Ulmasov et al., 1997). The Aux/IAA transcriptional repressors can form heterodimers with ARFs and prevent their binding to the auxin response elements (Kim et al., 1997). Among other ARFs, ARF8 plays a key role in flower development (Fig. 2). In Arabidopsis, At-ARF8 is subjected to alternative splicing giving rise to four splice variants, At-ARF8.1–At-ARF8.4. The full-length At-ARF8.1 contains 14 exons; alternative splicing leads to At-ARF8.2 with a PTC, while At-ARF8.3 lacks exon 1 and 34 nucleotides from exon 2, and At-ARF8.4 retains intron 8 along with the usage of an alternative 5ʹ ss in the last intron that generates a PTC. Genetic and molecular analyses using transgenic lines that overexpress the individual splice variants in a wild-type background revealed that the stamens of the At-ARF8.4 transgenic lines were significantly longer than those of untransformed wild-type and mock-treated flowers, but the length of the stamens in ARF8.1- and ARF8.2-overexpressing lines did not differ, suggesting a specific role for the At-ARF8.4 splice variant in the control of stamen elongation (Ghelli et al., 2018). ARF5/MONOPTEROS (MP) is an important integrator of auxin signaling in plant development and activates transcription in cells with elevated auxin levels. However, in ovules, MP is expressed in cells with low levels of auxin and can activate the expression of direct target genes. In Arabidopsis, the alternative splicing of At-MP generates a biologically functional isoform that retains intron 11 (At-MP11ir). This intron introduces a PTC into the MP open reading frame, resulting in the translation of a truncated protein that lacks the Aux/IAA interaction domain. By polysome profiling using pre-fertilization inflorescences, Cucinotta et al. (2021) suggested that At-MP11ir transcripts might escape NMD to produce a functional truncated protein. At-MP11ir is essential for the growth of the integument, implying that its function is tissue- and developmental-stage-specific in ovules. Moreover, this isoform can complement inflorescence, floral, and ovule developmental defects in mpS319 mutants, suggesting that the At-MP11ir protein is fully functional. In a similar manner, it restores the fertility, plant, and inflorescence morphology in the arf5.1 background. Overall, these results suggest that despite being uncoupled from classical auxin regulation due to the absence of the Aux/IAA interaction domain, At-MP11ir can still accomplish several native MP functions during ovule development (Cucinotta et al., 2021). Taken together, these findings propose an alternative scenario to the classical MP auxin regulation in which an isoform can function as a transcriptional activator in regions of subthreshold auxin concentration during ovule development (Cucinotta et al., 2021).

The transition to flowering is influenced by many environmental changes, such as temperature, dark/light cycles, and stress. As mentioned before, the At-SR45.1 isoform functions in flower development by directly regulating the alternative splicing pattern of splicing factor genes. The narrow flower petal phenotype in sr45-1 mutants is rescued in transgenic plants overexpressing the At-SR45.1 isoform, but not in plants overexpressing the At-SR45.2 isoform (Zhang and Mount, 2009).

The role of FLOWERING LOCUS M (FLM) in temperature-dependent flowering in Arabidopsis has been well studied. Mutants of this gene exhibit partial temperature insensitivity and an early-flowering phenotype (Scortecci et al., 2001). At-FLM has two splice variants, namely At-FLM-β and At-FLM-δ, resulting from the alternative splicing of At-FLM pre-mRNA in a mutually exclusive event involving the second and third exons. The abundance of At-FLM-β increases at low temperature (16 °C) and decreases at high temperature (27 °C) (Steffen et al., 2019); in contrast, At-FLM-δ is induced at high temperatures. The overexpression of At-FLM-β represses flowering, while overexpression of At-FLM-δ activates flowering (Posé et al., 2013). Under long-day conditions, At-FLM-β and the MADS-domain transcription factor SHORT VEGETATIVE PHASE (SVP) form a complex that inhibits floral activators, such as SUPPRESSOR OF CONSTANS OVEREXPRESSION 1 (SOC1) and FLOWERING LOCUS T (FT), to control the floral transition. By contrast, At-FLM-δ acts as a dominant negative isoform in flowering by competing with At-FLM-β for interaction with SVP. Therefore, the FLM-β/FLM-δ ratio fine-tunes flowering time in response to ambient temperature (Lee et al., 2013). However, several recent studies have revealed that the level of At-FLM-δ does not change in response to increased temperature. Additionally, besides At-FLM-β, At-FLM creates a large number of non-canonical splice variants at high temperature (27 °C) (Sureshkumar et al., 2016; Steffen et al., 2019). These splice variants are targeted to the NMD pathway, thereby resulting in down-regulated At-FLM expression. Wang et al. (2019) studied the involvement of the heterodimeric splicing factor U2 snRNP auxiliary factor (U2AF), which consists of two subunits, U2AF35 and U2AF65, and functions in 3ʹ ss recognition. These splicing factors are key for flowering under the influence of ambient temperature and ABA in Arabidopsis (Hrtyan et al., 2015; Zhu et al., 2017). Interestingly, the alternative splicing of At-U2AF65a in response to changes in temperature is controlled by At-CDKG1. Three splice isoforms of At-U2AF65a, namely mRNA1–3, have been identified in Arabidopsis. At 12 °C, mRNA1 is the most abundant isoform, while levels of mRNA2 and mRNA3 are increased significantly at 27 °C, with isoform mRNA3 being the most abundant. By contrast, in the null mutant cdkg1, the levels of mRNA2 and mRNA3 are decreased and increased, respectively, because of the retention of intron 11 (Cavallari et al., 2018). These findings suggest that At-CDKG1 is involved in the alternative splicing of At-U2AF65a. High levels of mRNA2 and mRNA3 at high temperature may be translated to truncated proteins functioning in the splicing of non-canonical FLM isoforms (Steffen et al., 2019). Wang et al. (2019) propose a signaling cascade from At-CDKG2 to At-FLM in the regulation of flowering in response to changes in ambient temperature. In this model, both At-CDKG2 and At-CYCL1 regulate the alternative splicing of At-CDKG1L in cold temperatures and At-CDKG1S in warm temperatures (Wang et al., 2019). At-CDKG1S induces an intron retention in At-U2AF65a that allows exon 3 of At-FLM to be included, producing At-FLM-δ. By contrast, under the At-CDKG1L isoform induced by cold, At-U2AF65a is fully spliced, leading to the exclusion of exon 3 and exon 8 of At-FLM, producing At-FLM-β, which inhibits the expression of flowering genes, such as At-FT and At-SOC1, by interacting with At-SVP (Steffen et al., 2019).

The Arabidopsis RNA-binding protein (RBP) GLYCINE-RICH RNA-BINDING PROTEIN 7 (At-GRP7) is under the control of the circadian clock, with maximal expression in the evening. It autoregulates its expression by altering its alternative splicing, with subsequent degradation via NMD (Schöning et al., 2007). Steffen et al. (2019) showed that the RBP At-GRP7 affects flowering time via at least two different pathways. In the autonomous pathway, At-GRP7 regulates the transcriptional levels of the MADS-box transcription factor gene At-FLC. In the thermosensory pathway, At-GRP7 regulates the alternative splicing of At-FLM and shifts the balance of the isoforms At-FLM-β and At-FLM-δ toward the floral repressive isoform At-FLM-β. In contrast, loss of At-GRP7 together with a reduction in At-GRP8 leads to higher levels of At-FLM-δ and numerous non-canonical isoforms that harbor intron 3, which are subsequently degraded by NMD (Sureshkumar et al., 2016). Thus, it appears that At-GRP7 and At-GRP8 are required for the correct splicing of At-FLM and that loss of both results in unproductive splicing. This effect is mediated through the direct binding of At-FLM by either one or both proteins (Steffen et al., 2019).

The circadian clock control of CONSTANS (CO) transcription and the light-mediated stabilization of its encoded protein coordinately adjust photoperiodic flowering by triggering rhythmic expression of the floral integrator At-FT. Diurnal accumulation of At-CO is modulated sequentially by distinct E3 ubiquitin ligases, allowing the peak of At-CO to occur in the late afternoon under long-day conditions (Gil et al., 2017). At-CO undergoes alternative splicing to produce the full-size At-COα and the C-terminally truncated At-Coβ, which lacks DNA-binding affinity. At-COα and At-COβ form homo- and heterodimers in the nucleus. At-COβ inhibits At-COα function by forming heterodimers in photoperiodic flowering. At-COβ is resistant to ubiquitin/proteasome-dependent degradation and, most importantly, facilitates the degradation of At-COα by enhancing its interaction with destabilizing factors (Gil et al., 2017). CIRCADIAN CLOCK-ASSOCIATED 1 (CCA1) is one of the core components of the day–night rhythm response and is associated with ambient temperature responses in Arabidopsis. At-CCA1 has two different isoforms (α and β), which differ in a MYB domain on exon 2 and exon 3. Both isoforms contain dimerization domains, and dynamic dimer formation plays a role in regulating transcription factor activities; At-CCA1 can form heterodimers and homodimers, but only At-CCA1α homodimers are responsible for transcriptional regulation (Filichkin et al., 2010). At-CCA1β alternative expression is, in this case, responsible for heterodimerization by diminishing the influence of At-CCA1α–At-CCA1α homodimers in the transcriptional landscape. At-CCA1 has an intron retention event on intron 4; this alternative splicing event creates a PTC in the reading frame, inducing NMD degradation. Since At-CCA1 has two alternative protein isoforms, this raises the question of how the At-CCA1β protein is translated; downstream AUGs (start codons) are rare and potentially inefficient, leading to the conclusion that this event is not really an alternative splicing product but an alternative transcription start site (Brown et al., 2015).

Fruit ripening and leaf senescence

During plant senescence, the nutrients in the leaves are recycled either to be translocated for the development of flowers and/or fruits (annual plants) or to be stored in a protein body to withstand the cold seasons (perennial plants). This process is mediated by programmed cell death and influenced by both external and internal factors that induce a massive genetic reprogramming (Lim et al., 2007; Woo et al., 2016). Bioinformatic analyses have revealed that programmed cell death is regulated by several groups of transcription factors. Among them, members of the NAC (NAM, no apical meristem; ATAF1–2, Arabidopsis thaliana activating factor; and CUC2, cup-shaped cotyledon) and WRKY families regulate the transcription of genes involved in leaf senescence (Balazadeh et al., 2008). In Arabidopsis, the NAC transcription factor RESPONSIVE TO DESICCATION 26 (At-RD26) acts as a positive transcriptional regulator of genes involved in chloroplast protein degradation and other pathways related to the leaf senescence process (Kamranfar et al., 2018). In Populus tomentosa, Pt-RD26 induces leaf senescence by activating a variety of senescence-related pathways and genes, such as ORESARA1 (Pt-ORE1), ARABIDOPSIS NAC DOMAIN CONTAINING PROTEIN 29 (Pt-NAP), and ETHYLENE INSENSITIVE 3 (Pt-EIN3), encoding three important transcription factors in leaf senescence regulation (Guo and Gan, 2006; Li et al., 2013). Through an alternative splicing event that occurs in a senescence-associated manner, Pt-RD26 produces the Pt-RD26^IR variant, which retains the first intron. Analysis of Pt-RD26 expression throughout leaf development reveals that Pt-RD26^IR expression increases in early and late senescence and transcripts gradually accumulate upon leaf aging. Since Pt-RD26^IR mRNA harbors a PTC that is predicted to be degraded by NMD, Wang et al. (2021) transiently expressed a construct containing the genomic sequence of Pt-RD26 and its promoter and untranslated regions (UTRs) in Arabidopsis upf1-5 (a loss-of-function mutant of UP FRAMESHIFT1, a core component of the NMD machinery) and found no increase in the expression level of Pt-RD26^IR mRNA, suggesting that the PTC might not be efficiently targeted by NMD machinery. When transiently overexpressing leaves are treated under dark conditions, Pt-RD26 promotes and Pt-RD26^IR delays dark-induced leaf senescence, suggesting that Pt-RD26^IR effectively represses Pt-RD26-induced leaf senescence by disrupting its binding to downstream targets. Furthermore, Pt-RD26^IR is translocated from the cytoplasm into the nucleus by physically interacting and forming heterodimers with Pt-RD26 or its homologs, which interfere with the DNA-binding activity of Pt-RD26 or other NAC-TF homodimers. This regulation by Pt-RD26^IR might be part of an interconnected loop to fine-tune the progression of leaf senescence (Wang et al., 2021).

As mentioned above, the senescence process provides nutrients that favor the transition from flower to fruit. The last stage of fruit development is ripening, during which the fruit color is modified and the sweetening and softening of its internal tissues takes place. A series of complex biological processes are involved in fruit ripening, including the metabolism and signaling of phytohormones, cell wall degradation, the formation of flavor and aroma compounds, and the biosynthesis and degradation of pigments and reserves (Gapper et al., 2013). Transcription factors and alternative splicing play a very important role in this process. Ma-MYB16L is a banana (Musa acuminata AAA) R1-type MYB transcriptional repressor that generates two isoforms by alternative splicing: the full-length isoform, Ma-MYB16L, and a truncated isoform, Ma-MYB16S (Jiang et al., 2021). At the time of the banana harvest, the long isoform negatively regulates fruit ripening by directly inhibiting the transcription of many genes associated with starch degradation. During fruit ripening, the mRNA levels of Ma-MYB16L and the Ma-MYB16L protein levels are reduced, while the levels of the truncated isoform are increased. In interaction experiments between both proteins, Ma-MYB16S also exerts a negative regulatory effect on the function of Ma-MYB16L through competitive binding. Ma-MYB16S, lacking a DNA-binding domain, can attenuate the activity of Ma-MYB16L by forming non-functional heterodimers with it and favoring the transcription of target genes associated with starch degradation. Therefore, the alternative splicing of Ma-MYB16 is implicated in the regulation of fruit ripening in banana through competitive inhibition and modification of the ratio of active to non-active isoforms (Jiang et al., 2021).

All these cases constitute excellent examples showing that different isoforms generated by alternative splicing can have distinct tissue- and developmental-stage-specific activities and biological functions (Table 1). However, in most of the examples mentioned above, there is a lack of proper controls to determine whether the different isoforms have coding capacities. In plants, different RNA polymerase II-derived transcripts, mostly mRNAs, are retained in the nucleus due to particular features of their sequences, such as keeping introns completely or partially retained. This avoids the translation of these RNA molecules and, concomitantly, it gives them the chance to escape different regulatory pathways, with NMD being one of relevance for their stability. We will return to this point in the Concluding remarks and future directions section.

Table 1.

List of alternative splicing isoforms with relevant physiological roles in plants

Gene name	Species	Isoforms	Description	Functions in	References
Ta-GS3	Wheat	Ta-GS3.1 Ta-GS3.2 Ta-GS3.3 Ta-GS3.4 Ta-GS3.5	Reduces grain weight and length No effect on grain weight and length No effect on grain weight and length No effect on grain weight and length Increases grain weight and length	Embryogenesis	Ren et al. (2021)
At-HAB1	Arabidopsis	Two main isoforms: At-HAB1.1 At-HAB1.2	Promotes seed germination Prevents seed germination	Seed germination	Wang et al. (2015)
At-PIN7	Arabidopsis	At-PIN7a At-PIN7b	Regulate apical hook development and hypocotyl bending in a mutually antagonistic manner	Post-germination seedling establishment	Kashkan et al. (2022a)
At-IBR5	Arabidopsis	At-IBR5.1 At-IBR5.3	IBR5.1 phosphatase activity is necessary for both proper Aux/IAA degradation and auxin-induced gene expression. IBR5.1 and IBR5.3 isoforms may have distinct as well as overlapping functions in growth and development	Root growth and lateral root development	Jayaweera et al. (2014)
At-HYH	Arabidopsis	At-HYH.1 At-HYH.2 At-HYH.3 At-HYH.4	Higher transcript levels of HYH.2-HYH.4 than those of HYH.1 are found in hypocotyl tissues. All HYH variants are induced by light, have similar roles in hypocotyl development, and may function redundantly with HY5	Hypocotyl elongation	C. Li et al. (2017)
At-BBX22	Arabidopsis	At-BBX22IR	Positively regulates the light-mediated inhibition of hypocotyl elongation during photomorphogenesis	Hypocotyl elongation	Huang et al. (2022)
)At-BBX24	Arabidopsis	At-BBX24IR	Positive regulator of photomorphogenic development by suppressing the function of At-BBX24	Hypocotyl elongation	Huang et al. (2022)
At-SR45	Arabidopsis	At-SR45.1 At-SR45.2	Functions in flower development Functions in root growth	Plant development Response to salt stress	Zhang and Mount (2009)
At-JAZ10	Arabidopsis	At-JAZ10.1 At-JAZ10.3 At-JAZ10.4	Plants overexpressing JAZ10.1 remain fully responsive to jasmonic acid (JA) Overexpressing JAZ10.3 causes moderate JA insensitivity Overexpressing JAZ10.4 causes severe JA insensitivity	JA signaling pathway	Moreno et al. (2013)
At-PIRL6	Arabidopsis	Seven transcripts, only one functional	The functional isoform maintains fertility in Arabidopsis	Gametogenesis	Forsthoefel et al. (2018)
At-CDKG1	Arabidopsis	Two main isoforms: At-CDKG1L At-CDKG1S	In reproductive tissues, only the At-CDKG1L form is necessary to maintain fertility	Meiosis and pollen formation	Nibau et al. (2020)
At-ARF8	Arabidopsis	At-ARF8.1 At-ARF8.2 At-ARF8.3 At-ARF8.4	At-ARF8.4 is involved in the control of stamen elongation	Flower organ development	Ghelli et al. (2018)
At-ARF5/MP	Arabidopsis	At-MP11ir	At-MP11ir is essential for the integument growth during ovule development	Ovule development	Cucinotta et al. (2021)
At-FLM	Arabidopsis	At-FLM-β At-FLM-δ	Overexpression of At-FLM-β represses flowering Overexpression of At-FLM-δ activates flowering The FLM-β/FLM-δ ratio fine-tunes flowering time in response to ambient temperature	Flowering	Posé et al. (2013)
At-CO	Arabidopsis	At-COα At-COβ	At-COβ inhibits At-COα function by forming heterodimers in photoperiodic flowering	Flowering	Gil et al. (2017)
Pt-RD26	Populus tomentosa	Pt-RD26IR	Pt-RD26IR effectively represses Pt-RD26-induced leaf senescence by disrupting its binding downstream targets	Leaf senescence	Wang et al. (2021)
Ma-MYB16L	Banana	Ma-MYB16L Ma-MYB16S	Alternative splicing of Ma-MYB16 is implicated in the regulation of fruit ripening in bananas through competitive inhibition and modification of the ratio of active to non-active isoforms	Fruit ripening	Jiang et al. (2021)
At-HsfA2	Arabidopsis	At-HsfA2-II At-HsfA2-III At-HsfA2-IV	At-HsfA2-III can be translated into the small protein S-HsfA2 during severe heat shock (42 °C for 1 h) to activate its own gene expression	Response to heat	Liu et al. (2013)
Ll-HSFA3	Lilium spp.	LlHSFA3B-III	Heterologous expression of LlHSFA3B-III in Arabidopsis and Nicotianabenthamiana increases plant tolerance of salt and prolonged heat at 40 °C	Response to heat	Wu et al. (2019)
Os-HSFA2d	Rice	Three isoforms; only Os-HSFA2dI is specifically induced by heat shock	Functions in the heat-stress-induced unfolded protein response in rice	Response to heat	Cheng et al. (2015)
Sl-HSFA2	Tomato	Seven isoforms, one of them appears during recovery from 42.5 °C	The splice isoform HsfA2-II is implicated in the early stress response at the expense of HsfA2-I, which is involved in short-term thermotolerance	Response to heat	Hu et al. (2020)
Os-bZIP58	Rice	Os-bZIP58α Os-bZIP58β	OsbZIP58β is involved in seed quality related to grain weight and thickness, and starch and lipid content	Response to heat	Xu et al. (2020)
At-MBD4L	Arabidopsis	At-MBD4L.1 At-MBD4L.2. At-MBD4L.3 At-MBD4L.4	Heat stress conditions (37 °C) increase the relative abundance of the nucleolar variant	Response to heat	Cecchini et al. (2022)
At-SR45a	Arabidopsis	At-SR45a-1a At-SR45a-1b	Involved in post-transcriptional regulation of salt tolerance	Response to salt	Li et al. (2021)
At-SRAS1	Arabidopsis	At-SRAS1.1 At-SRAS1.2	Plants overexpressing At-SRAS1.1 are more tolerant to salt stress Plants overexpressing At-SRAS1.2 are more sensitive to salt stress	Response to salt	Zhou et al. (2021)
Sc-MYBAS1	Sugarcane	Sc-MYAS1-2 Sc-MYBAS1-3 Sc-MYBAS1-4 Sc-MYBAS1-5	Sc-MYBAS1-3 transgenic lines show increased biomass, whereas Sc-MYBAS1-2 transgenic lines have a decrease in biomass under both well-watered and drought conditions	Response to drought	Fávero Peixoto-Junior et al. (2018)
Ta-DREB3	Wheat	Ta-DREB3-I Ta-DREB3-II Ta-DREB3-III	Only Ta-DREB3-I shows important roles in responses to stress (plants show better growth status and higher survival rates after dehydration, high temperature, and salt stress)	Response to drought	Niu et al. (2020)
Zm-CCA1	Maize	Zm-CCA1.1 Zm-CCA1.2 Zm-CCA1.3	When overexpressing each isoform in Arabidopsis, Zm-CCA1.1 and Zm-CCA1.3 show significantly higher drought tolerance whereas Zm-CCA1.2 exhibited only slightly higher drought tolerance	Response to drought	Tian et al. (2019)
Os-RLI1	Rice	Os-RLI1a Os-RLI1b	Only Os-RLI1a regulates brassinolide (BL) biosynthesis and signaling by directly activating BL biosynthesis and signaling genes. Both splice variants modulate Pi starvation signaling	Pi starvation signaling and growth	Guo et al. (2022)
Os-Rad9	Rice	Os-Rad9.1 Os-Rad9.2	Os-Rad9.1 is involved in the response to genotoxic, drought, salt, and heavy metal stress, and auxin stimuli. Os-Rad9.2 shows different preferential stimulation effects on these stresses and pollen development duration	Response to genotoxic stress and pollen development	R. Li et al. (2017)
At-RS31A	Arabidopsis	At-RS31A.1 At-RS31A.2 At-RS31A.3	At-RS31A.2 plays a significant role in genotoxic responses	Response to genotoxic stress	Xue et al. (2018)
At-IDD14	Arabidopsis	At-IDD14α At-IDD14β	At-IDD14α-overexpressing plants contain reduced starch levels and stunted growth phenotypes At-IDD14β-overexpressing plants contain high starch contents	Starch metabolism in response to cold	Seo et al. (2011)
At-SGR5/IDD15	Arabidopsis	At-SGR5α At-SGR5β	Heat stress conditions induce more SGR5β proteins, and this regulates the gravitropic response of inflorescence stems	Gravitropic response of inflorescence stems at high temperatures	Kim et al. (2016)
At-ERF7/HRE1	Arabidopsis	At-HRE1α At-HRE1β	At-HRE1α functions in root development through regulation of root meristem cell division At-HRE1β plays a more relevant role in the hypoxia response than At-HRE1α	Response to hypoxia and primary root development	Seok et al. (2020)

Stress-induced alternative splicing variants and their biological implications in response to environmental changes

Several studies have shown that environmental changes modulate alternative splicing patterns at both gene-specific and transcriptome-wide levels. Plant molecular responses to stress have often been considered a complex process based mainly on the modulation of the transcriptional activity of stress-related genes. However, recent findings suggest further layers of regulation and complexity (Mazzucotelli et al., 2008). According to multiple transcriptomic approaches (i.e. RNA-seq datasets), abiotic stress induces remarkable changes in the alternative splicing patterns of many genes in plants (Filichkin et al., 2010; Ding et al., 2014; Thatcher et al., 2016; Jiang et al., 2017). Salt stress induces the expression of many stress-responsive genes, including those encoding transcription factors (Yamaguchi-Shinozaki and Shinozaki, 2006). In Arabidopsis, for example, SNW/SKI-INTERACTING PROTEIN (At-SKIP) confers osmotic tolerance during salt stress by controlling alternative splicing (Feng et al., 2015), and a similar function has been described in grapes under high temperature (Jiang et al., 2017). These are only some of the examples reflecting the important role that alternative splicing plays in adaptive responses to stress. Indeed, plant genes encoding known stress-responsive regulators are particularly prone to generating multiple transcripts through this mechanism (Ner-Gaon et al., 2004). In this section, we discuss the functional roles of alternative splicing variants under stress conditions.

Heat stress

Heat shock (HS) is one of the abiotic stresses that most affects plant growth and survival. The main regulators of plant cell responses to HS are the heat shock transcription factors (Hsfs), which have been identified in different plant species such as alfalfa (Medicago sativa) (He et al., 2007), Arabidopsis (Sugio et al., 2009), and Potamogeton spp. (Amano et al., 2012). Liu et al. (2013) found that in Arabidopsis under severe HS a new splice variant of the HEAT SHOCK TRANSCRIPTION FACTOR A2 (At-HsfA2-III) is generated by using a cryptic 5ʹ ss in the intron. At-HsfA2-III can be translated into the small protein S-HsfA2 during severe HS (42 °C for 1 h) to activate its own gene expression, thus constituting a positive autoregulatory loop (Liu et al., 2013). Something similar happens in Lilium spp. With the homolog LI-Hsfa3. When seedlings are exposed to HS (37 °C), the heat-inducible splice variant Li-Hsfa3B-III is generated (Wu et al., 2019). Li-Hsfa3B-III contains a PTC and can be translated as a small, truncated protein that negatively regulates another Hsf isoform (Li-Hsfa3A-i) in the short-term heat shock response (HSR) by preventing the formation of the functional oligomer and thus positively regulating long-term HSR (Fig. 3A). A similar mechanism occurs in Oryza sativa. As shown by Cheng et al. (2015), Os-HSFA2d produces three isoforms by alternative splicing, of which only one is specifically induced by HS (Os-HSFA2dI) and encodes a protein related to HSR and the unfolded protein response (or response to misfolded proteins). In tomato seedlings, a splice variant of the Hsf Sl-HsfA2-I increases during short-term acclimation to high temperature, enhancing thermotolerance, whereas the Si-HsfA2-II variant, whose truncated C-terminal activation motif lacks a nuclear export signal, is rapidly degraded (Hu et al., 2020).

Fig. 3.

Stress-induced alternative splicing changes in different plant species. (A) In Lilium spp., Li-Hsfa3B produces three isoforms that differ in the inclusion of an exon or an intron. When leaves are exposed to high temperatures, Li-Hsfa3B-III is induced without altering the expression of the other two isoforms. The heat-inducible Li-Hsfa3B-III variant promotes interaction with other proteins involved in the heat shock response (HSR), favoring a long over a short HSR (Wu et al., 2019). (B) In Arabidopsis, the alternative splicing of At-SRAS1 generates two variants that differ in the inclusion of one intron. Increased salt concentrations antagonistically alter the expression of both splice variants, inducing the expression of the At-SRAS1.1 isoform while reducing that of At-SRAS1.2. The expression of At-SRAS1.1 promotes salt tolerance by interacting with other stress-response proteins (Zhou et al., 2021). (C) Alternative splicing of Zm-CCA1 in maize generates three splice variants that differ in the retention of an intron and an alternative 5ʹ ss. The splice variants perform different tolerance abilities in response to drought (Tian et al., 2019). Exons are shown as colored boxes; introns are shown as straight lines; untranslated regions are shown as dashed vertical black lines inside exons. Solid lines represent canonical splicing and dashed lines represent alternative splicing. Up arrows indicate an increase in expression and down arrows indicate a decrease. The thickness of the arrows thickness indicates the intensity of expression relative to other splice variants. N/A, No changes in expression levels.

High temperatures can impair rice grain quality by significantly reducing the accumulation of storage compounds. Xu et al. (2020) found that the alternative splicing of Opaque2-like transcription factor (Os-bZIP58) is altered at high temperatures. When rice plants were grown under a cycle of 12 h light at 35 °C and 12 h dark at 28 °C for several days after pollination, the activity of the protein Os-bZIP58β encoded by the alternatively spliced transcript was reduced; this could explain the decrease in grain quality related to grain weight and thickness, and starch and lipid content (Xu et al., 2020).

The Arabidopsis MBD4-like (At-MBD4L) DNA glycosylase improves tolerance to genotoxic stress. This enzyme is encoded by a single gene carrying an exitron at its 5ʹ region and has four predicted alternative transcripts: the larger exitron-retained version At-MBD4L.3, the shorter spliced isoform At-MBD4L.4 (both present in leaves and flowers), At-MBD4L.1, and At-MBD4L.2. At-MBD4L.3 and At-MBD4L.4 differ in an exitron and encode proteins differing in the number of amino acids. Both isoforms conserve the catalytic domain but are directed to the nucleoplasm (At-MBD4L.3) or to the nucleolus (At-MBD4L.4). Exposing wild-type Col-0 seedlings to heat stress conditions (37 °C) induces changes in the alternative splicing of the MBD4L exitron and increases the relative abundance of the nucleolar variant, a process that depends, at least in part, on the splicing factors RS31 and NTR1. Interestingly, these changes in the relative abundance of both isoforms are recovered after stress release (Cecchini et al., 2022).

Salt stress

High salinity severely affects plant growth and development, impairing crop production worldwide. Many splicing factors have been reported in plants but there are not many examples of alternative splicing variants playing roles in salt stress tolerance (Singh and Roychoudhury, 2021). At-SR45 isoforms, described earlier, also have a role in salt stress tolerance. Arabidopsis seedlings lacking the At-SR45 gene showed enhanced sensitivity to salt stress, but the overexpression of only the long isoform At-SR45.1, and not At-SR45.2, is able to reverse salt sensitivity. In addition, the long isoform is essential for the expression and alternative splicing of salt-stress-responsive genes, suggesting that At-SR45 is a positive regulator of salt tolerance in Arabidopsis (Albaqami et al., 2019). Likewise, a conserved SR-like protein, At-SR45a, is also involved in the post-transcriptional regulation of salt tolerance in Arabidopsis. Two alternatively spliced variants of At-SR45a are induced by salt stress: the full-length At-SR45a-1a and the truncated isoform At-SR45a-1b. The full-length protein At-SR45a-1a interacts with the cap-binding protein 20 (CBP20) subunit of the cap-binding complex to regulate the expression of salt-related genes and splicing events, while the truncated At-SR45a-1b promotes the interaction between them (Li et al., 2021). This is another piece of evidence suggesting a positive self-regulation loop, as seen with Hsfs.

A recent study in Arabidopsis showed that SALT-RESPONSIVE ALTERNATIVELY SPLICED GENE 1 (At-SRAS1) is alternatively spliced, generating two variants, At-SRAS1.1 and At-SRAS1.2 (Zhou et al., 2021). The full-length At-SRAS1.1 protein has E3 ubiquitin ligase activity, while the truncated At-SRAS1.2 lacks this activity. Interestingly, these variants respond antagonistically to salt tolerance: At-SARS1.1 expression levels increase under high salt conditions whereas At-SARS1.2 levels decrease (Fig. 3B). Plants overexpressing At-SRAS1.1 are more tolerant to salt stress, while those overexpressing At-SRAS1.2 are more sensitive to salt (Zhou et al., 2021).

Drought stress

Since drought is one of the biggest challenges for the crop industry, finding genes and regulatory pathways related to drought tolerance are of great importance. Stress-responsive genes are mostly regulated by various transcription factors. Several transcription factors participate in plant responses to drought (Seki et al., 2003). For example, various Myeloblastosis (MYB) genes from Arabidopsis and other plants encode transcription factors that are involved in drought responses (Li et al., 2015). In sugarcane, one of the MYB factors is encoded by the R2R3-MYB (Sc-MYBAS1) gene. Sc-MYBAS1 undergoes alternative splicing giving rise to four transcripts whose proteins differ in their MYB domain repeats and act differently in the regulatory pathways underlying drought tolerance. Overexpression of the alternative transcripts Sc-MYBAS1-2 and Sc-MYBAS1-3 in rice resulted in a higher relative water content than the wild type before maximum stress under drought conditions. However, both splice variants play opposite roles in the amount of total biomass production: while the overexpression of Sc-MYBAS1-2 results in a reduction in biomass (total dry weight), the overexpression of Sc-MYBAS1-3 results in higher biomass relative to wild-type plants (Fávero Peixoto-Junior et al., 2018).

The Dehydration-responsive-element-binding (DREB) gene in wheat that responds differentially to abiotic stress is Ta-DREB3 (Niu et al., 2020). Ta-DREB3 contains three introns and generates three isoforms: Ta-DREB3-I and Ta-DREB3-II encode one long and one short intact amino acid sequence, respectively, and Ta-DREB3-III has a frameshift that abolishes translation. The expression of both Ta-DREB3-I and Ta-DREB3-II is induced in response to various abiotic stresses, such as drought, salt, and heat. Overexpression of each isoform ectopically in Arabidopsis revealed that Ta-DREB3-I transgenic plants have a better growth status and higher survival rates after dehydration, high temperature, and salt stress compared with the wild type. Although the expression of the three isoforms of Ta-DREB3 increases after various abiotic stresses, only Ta-DREB3-I shows important roles in the response to stress, suggesting that Ta-DREB3-II and Ta-DREB3-III can maintain the constitutive activation of TaDREB3 transcription without affecting plant growth (Niu et al., 2020).

In maize, Tian et al. (2019) identified three CIRCADIAN CLOCK-ASSOCIATED 1 (Zm-CCA1) splice variants that are influenced by drought stress. Zm-CCA1 splice variants are generated by alternative 5ʹ ss, alternative polyadenylation, and a combination of two or more alternative splicing types. Zm-CCA1.1 encodes functional proteins, and both Zm-CCA1.2 and Zm-CCA1.3 encode truncated proteins. In this work, to test whether the expression of the isoforms is influenced by drought stress, maize seedlings were treated with 20% PEG-6000 for 1 d. After the drought treatment, when compared with the untreated control, Zm-CCA1.1 and ZmCCA1.3 expression were suppressed in leaves and increased in leaf sheaths while Zm-CCA1.2 expression did not change, suggesting that this isoform is not involved in the drought stress response in maize (Fig. 3C). Moreover, when each maize isoform is overexpressed in Arabidopsis, Zm-CCA1.1 and Zm-CCA1.3 confer significantly higher drought tolerance whereas Zm-CCA1.2 overexpression resulted in only slightly higher drought tolerance, suggesting that the three splice variants perform different tolerance functions in response to drought (Tian et al., 2019).

Together, this evidence shows that alternative splicing is a key regulatory mechanism to fine-tune signaling pathways involved in stress responses ensuring the survival of the plant, suggesting that it is a necessary mechanism for plants to adapt to a rapidly changing environment (Table 1). Once again, several examples cited here lack proper controls to determine whether the different isoforms have coding capacities. Transcripts with exitrons might be the exception, since these events are generally described to be actively translated and they are mostly devoid of stop codons, in opposition to canonical introns. We will revisit this issue in the Concluding remarks and future directions section.

Alternative splicing isoforms with roles in growth and development under stress conditions

In the previous sections, we have separately considered genes that undergo alternative splicing whose splicing isoforms play biological roles during plant growth and development or in response to environmental changes (i.e. abiotic stress). However, several alternative splice variants play dual roles as they work in both processes (Table 1). For example, the alternative splicing of REGULATOR OF LEAF INCLINATION 1 (RLI1) is an important adaptive mechanism in plants that modulates inorganic phosphate (Pi) starvation signaling, Pi homeostasis, and plant growth under Pi-deficiency stress (Guo et al., 2022). Alternative splicing of Os-RLI1 in rice produces two isoforms (Os-RLI1a and Os-RLI1b) that modulate Pi starvation signaling: under low-Pi stress, the expression of Os-RLI1a is repressed while the expression of Os-RLI1b is increased. Os-RLI1a protein has only a DNA-binding domain (MYB), whereas Os-RLIb has a coiled-coil domain as well. The absence of the coiled-coil domain in Os-RLI1a enables it to directly activate a broader range of target genes than Os-RLI1b, including genes involved in brassinolide biosynthesis and signaling, thereby modulating plant growth.

Rad9 is a key component of the checkpoint signaling pathway of the cell cycle and responds to DNA damage induced by stresses in both plants and animals (Xu et al., 2009). In rice, Os-Rad9 undergoes alternative splicing, generating two transcripts, Os-Rad9.1 and Os-Rad9.2. Under genotoxic conditions that induce DNA damage, Os-Rad9.1 expression increases more rapidly than Os-Rad9.2. Similar results are obtained in response to other environmental stresses, such as drought, salt, and heavy metals. Moreover, Os-Rad9.2 is highly expressed in developing pollen, while Os-Rad9.1 is expressed in almost all tissues examined (e.g. leaf, root, stem, and callus). Thus, Os-Rad9 may have multiple functions in the repair of DNA damage as well as in rice pollen development (R. Li et al., 2017). Also related to DNA damage, by using a Cas9-directed base editor, Xue et al. (2018) modulated the alternative splicing of the endogenous At-RS31A gene in Arabidopsis and revealed its functional involvement in the plant response to genotoxic treatment. The intron 2 of At-RS31A can be either totally or partially spliced, using two different 5ʹ ss, or completely retained, generating three mRNA isoforms that could encode three proteins with different N-terminals. The authors used an sgRNA to target the distal 5ʹ ss of At-RS31 to specifically prevent the production of At-RS31A.2, leaving At-RS31A.1 and At-RS31A.3 unaffected. When analyzing both cotyledon greening and true leaf development, the at-rs31a.2 knockout mutant plants were less sensitive to mitomycin C than the wild type, indicating that At-RS31A.2 plays a significant role in responses to genotoxic agents (Xue et al., 2018).

The Arabidopsis INDETERMINATE DOMAIN 14 (IDD14) transcription factor (Seo et al., 2011) functions as a homodimer in the regulation of genes involved in starch metabolism. Under cold conditions, At-IDD14 is alternatively spliced, producing, in addition to the functional At-IDD14α isoform, an At-IDD14β isoform that lacks a functional DNA-binding domain. Since At-IDD14β has a protein–protein interacting domain, it can form heterodimers with At-IDD14α, and this provokes a decrease in its participation in starch degradation. Consistent with this, At-IDD14α-overexpressing plants contain reduced starch levels and exhibit stunted growth phenotypes, whereas At-IDD14β-overexpressing plants contain high starch contents, supporting the dominant negative role of At-IDD14β. As the authors suggest, this strategy would be an adaptive advantage against cold conditions, helping the plant to maintain a certain amount of starch during the dark and light periods. A close paralog of At-IDD14 is SHOOT GRAVITROPISM 5 (SGR5/IDD15), which encodes a transcription factor that, together with At-IDD14 and At-IDD16, regulates auxin biosynthesis and transport, and thus aerial organ morphogenesis and gravitropic responses. At-SGR5 undergoes alternative splicing by retention of the first intron, giving rise to two splice variants, At-SGR5α and At-SGR5β. The full-size SGR5α protein contains three copies of zinc finger motifs and the truncated SGR5β protein lacks the first zinc finger DNA-binding domain. SGR5β interferes with SGR5α function by forming non-DNA-binding heterodimers. Kim et al. (2016) showed that the alternative splicing of At-SGR5 is induced under HS conditions (37 °C, 42 °C, and 45 °C), leading to the production of more SGR5β proteins, and this regulates the gravitropic response of inflorescence stems in Arabidopsis, suggesting that this could be an adaptive strategy by which the shoots are protected from potential high-temperature-induced damage in their natural habitats (Kim et al., 2016).

ETHYLENE RESPONSE FACTOR (ERF) proteins play important roles in plant responses to biotic and abiotic stresses and in hormone responses in plants (Dubouzet et al., 2003; Gutterson and Reuber, 2004; Nakano et al., 2006; Yamaguchi-Shinozaki and Shinozaki, 2006). In Arabidopsis, At-ERF genes are induced at different stages of hypoxia treatment, which is a growth-limiting factor, leading to a decrease in ATP production and a subsequent energy crisis affecting numerous plant processes. Among these genes, At-ERF73/HRE1 has two distinct alternative splicing variants, At-HRE1α and At-HRE1β (Seok et al., 2014), both localized in the nucleus. At-HRE1α does not contain any introns, whereas At-HRE1β does have one, and At-HRE1α and At-HRE1β encode proteins that potentially have 211 and 262 amino acids, respectively. Both isoforms share 200 amino acids, including the AP2/ERF domain and C-terminal region, whereas the N-terminal region of At-HRE1β is longer than that of At-HRE1α (Seok et al., 2020). By using overexpressing lines, Seok et al. (2014) showed that At-HRE1α-overexpressing lines were more resistant to submergence than wild-type plants. Moreover, the primary root length was longer in At-HRE1α-overexpressing lines than in wild-type plants, suggesting a role in root development through the regulation of root meristem cell division (Seok et al., 2014). Another recent study showed that the At-HRE1β isoform plays a more relevant role in the hypoxia response than At-HRE1α, although both splicing variants are involved in the hypoxia response and root development (Seok et al., 2020). In this work, the authors showed that the transcript levels of At-HRE1β were significantly higher than those of At-HRE1α in 10-day-old wild-type seedlings treated under low-oxygen conditions. The primary root lengths of the overexpressing lines were longer than the wild-type plants but did not significantly differ between At-HRE1α- and At-HRE1β-overexpressing lines, suggesting that both splicing variants function in primary root development and differentially trans-activate downstream genes in the hypoxia response and root development of Arabidopsis.

Concluding remarks and future directions

The complex molecular networks that underlie the successful transitions in a plant’s life cycle involve the integration of environmental cues and endogenous signals into regulatory mechanisms that include transcriptional, translational, and epigenetic processes. During the past years, it has become evident that alternative splicing fine-tunes plant responses. Advances in RNA sequencing by using the RNA-seq Illumina short-read-approach shed light on the increasing frequency of alternative splicing events in plants. However, isoform identification is still a limiting factor for RNA-seq experiments. It is noteworthy that this limitation can be overcome by using third-generation sequencing technologies such as single-molecule real-time sequencing, developed by PacBio, and Oxford Nanopore Technologies, which uses nanopores in a membrane to sequence single-stranded DNA or RNA molecules (Jain et al., 2015; Rhoads and Au, 2015). While next-generation sequencing platforms generate relatively short reads (up to ~600 nucleotides), third-generation sequencing is characterized by improved sequencing chemistry, leading to the production of long reads (average length >10 kb) (Goodwin et al., 2016; van Dijk et al., 2018). The use of isoform-specific transcriptomes (i.e. Iso-seq) constitutes an excellent tool for the identification of novel regulators in plant development and stress responses, since it allows accurate isoform quantification, identification of transcription start sites and splice sites, characterization of polyadenylation sites, and detection of RNA-modified bases, and it also opens a new scenario to identify novel players regulating photomorphogenic development.

Despite all the technological and bioinformatics advances, we still lack information related to the biological effects of alternative splicing not only on mRNA levels but also on protein function(s). This reduces the knowledge on the real-world impact of splice variants on an organism phenotype, since the proportion of alternative splice isoforms that are indeed translated into functional protein(s) is largely unknown. Even when splice variants are expressed in different tissues and/or subcellular compartments, or at different developmental stages, the image of their functional role in a natural environment is still incomplete, since we do not know the extent to which the alternatively spliced transcripts are translatable. Here, it is worth pointing out that the methodological approaches used so far need to be improved in order to gain more accurate conclusions on the relevance of splice variants in plants. On the one hand, as we previously pointed out, most of the studies assessing the role of splice variants assume that the different isoforms indeed have coding capacities. In this review, we aimed to search for the best examples where different RNA isoforms seem to code for proteins with distinct functions; strikingly, we found that several of the studies reviewed here presumed that stable isoforms are coding, no matter whether they have PTCs and/or intron retention events. This is an issue that needs to be carefully addressed, as plants’ transcripts that retain introns, either partially or completely, are commonly retained in the nucleus and are relatively stable. In addition, other isoforms that present PTCs, even when they are able to leave the nucleus, are often subjected to active degradation by NMD in the cytosol. On the other hand, but intrinsically linked to this, most of the studies that evaluate isoform function are done using ‘classical’ methodologies such as constructing reporter or effector plasmids containing the coding sequences devoid of UTRs or other possible regulatory sequences. In this sense, it is important to mention that the transcripts’ export, stability, and degradation rely on those different RNA sequences or motifs, and on the factors that interact with them. By removing UTRs, other introns, or other regions of an RNA, we are neglecting their relevance and putatively generating artifacts, which in most cases mean that translation will definitely happen, but this translation does not reflect what would naturally occur. Hence, without data clearly indicating that a particular RNA isoform is translated, we are incurring big problems. In most cases, a reanalysis of the data considering these issues might lead to a completely different conclusion. What is commonly acknowledged as a consequence of alternative splicing, for example, the impact of a particular isoform on the organism’s phenotype, could be just an artifact with—by chance—potential biotechnological use, such as inhibiting the function of one coding isoform by overexpressing a truncated protein that has no real counterpart in nature. In this sense, CRISPR/Cas9 technology constitutes a concrete option to analyze the functions of the different splicing outcomes of a particular gene. By using this approach, we can generate particular splicing patterns for specific genes or events without affecting other possible regulatory sequences inside the same transcriptional unit. This approach can also be considered for crop improvement. For example, CRISPR/Cas9-mediated exon skipping induction has been recently used to improve fragrance in rice (Tang et al., 2021). Thus, CRISPR-directed base editors provide an interesting methodological approach to investigate the function of particular alternative splicing isoforms, even when the alternative events are fairly complex (Fig. 4). However, it is important to note that, in general, alternative splicing changes are mild (see examples in Figs 2, 3). Hence, overexpressing one isoform or avoiding the production of a particular one (see Fig. 4) could lead to extreme phenotypes that lack physiological relevance.

Fig. 4.

Base editing constitutes a powerful tool for investigating the functions of alternative splicing and splice isoforms. Hypothetical model representing base editing for GENE_A to prevent the generation of the splice isoform GENE_A.2. (A) Schematic diagram showing GENE_A alternative splicing in the wild type and the base editor target sequence. (B) Schematic diagram showing the alternative splicing in the gene_a knockout mutant and the base-edited sequence. In the pre-mRNA of GENE_A, intron 2 can be spliced at two different 5ʹ ss, or retained, generating three splice isoforms (GENE_A.1, GENE_A.2, and GENE_A.3). Designing an sgRNA to target the 5ʹ splice site of intron 2 by G-to-A conversions specifically prevents the production of the splice isoform GENE_A.2, leaving the GENE_A.1 and GENE_A.3 splice isoforms intact. (C) In this hypothetical example, gene_a knockout base-edited plants are less sensitive to drought conditions than wild-type plants, suggesting that the splice isoform GENE_A.2 might play a significant role in drought tolerance. AS, alternative splicing; PAM, protospacer-adjacent motif; sgRNA, single guide RNA; UTR, untranslated region.

As a corollary of this review, we propose a new strategy to properly analyze, evaluate, and characterize putative functions for the alternative splicing isoforms of interest (Fig. 5). Once a gene undergoing alternative splicing is identified, we should consider different features of the sequences of the individual isoforms before starting the evaluation of their function. First, as it is easy to determine the sequence of the different isoforms, we need to know the type of alternative splicing event underlying the observed outcome (see Fig. 1 for details of different types of alternative splicing events). At this point it is important to screen for PTCs, since they could lead to NMD. There are different mechanisms to determine whether a stop codon in a transcript is a PTC. In fact, particular features of the mRNAs, such as possessing very long 3ʹ UTRs or the presence of a stop codon positioned more than 50 nucleotides upstream of an exon junction (for a thorough review, see Popp and Maquat, 2013) are key factors that could label a stop codon as a PTC. In general, PTC-containing transcripts are targets of NMD. This can be experimentally tested by using cycloheximide, a translation inhibitor, as NMD depends on this process. Another option to block NMD could be the use of the photostable molecule cordycepin. These experimental approaches would allow us to analyze whether the transcripts of interest are sensitive to this degradation pathway. When available, as is the case for Arabidopsis, it is possible to use upf (UP-Frameshift) mutants. In particular, UPF1, UPF2, and UPF3 are core components of mRNA surveillance complexes and are essential for NMD (Kurihara et al., 2009). If an isoform is indeed an NMD target, the use of upf mutants or a translation inhibitor should result in an increase in the abundance of the isoform. If this is the case, analyzing the effect of the putative derived protein is pointless. By contrast, if the abundance of the isoform is not modified under these treatments or conditions, it could mean that the isoform is not actively degraded, although it does not imply that the transcript is translated into a protein. A transcript’s stability simply means that this RNA molecule avoids degradation.

Fig. 5.

Proposed working pipeline to analyze, evaluate, and characterize putative functions of alternative splicing isoforms of interest. When studying the outcomes of alternative splicing and the functions of different transcripts to understand their impact on an organism’s phenotype, we suggest considering the questions depicted in the workflow. *Bad luck! ⁺However, this might change when analyzing more conditions.

The question, then, is how. Since NMD is a cytosolic process, by staying away from this compartment, an isoform can be ‘safe’. Interestingly, escaping degradation is a common feature of transcripts with retained introns, as they are mostly retained in the nucleus (Göhring et al., 2014; Jia et al., 2020; Fuchs et al., 2021). Complete intron retention is the most prevalent event in plants (Marquez et al., 2012), so this can be a very common trend. Moreover, partial retention of an intron, in some cases, could also lead to nuclear retention, as has been shown for some splicing factors (Petrillo et al., 2014; Fuchs et al., 2021) where an alternative 5ʹ ss or 3ʹ ss is actually masking the retention of another, smaller intron that can be recognized and spliced out to give rise to a different isoform (e.g. RS31, RS31a, and SR30, among other splicing factors). Intron retention constitutes ~40% of total alternative splicing events in Arabidopsis, and stop codons in different reading frames are often present in intronic sequences. Taking these facts in consideration, we could think that most of the transcripts containing PTCs are neither degraded nor translated as they never reach the cytosol. However, some transcripts might indeed reach the cytosol and could be substrates of degradation or translation machineries (Fig. 1). Moreover, some transcripts that are degraded in a particular condition could be stable in others. Hence, we need to confirm whether an isoform that is stable is translated into a protein. There are different approaches and proxies that can be considered for achieving this goal. We propose some prospects for research as a starting point to evaluate whether an RNA is coding, although this list could be actively updated as new technological tools become available.

• One ‘cheap’ option is to use proteomic data. If the protein encoded by our isoform of interest has a distinctive domain or motif, we could simply search for representative peptides in publicly available databases as a starting point.

• Another option is to perform a Western blot experiment. In this case, antibodies against an epitope present are needed, at least, for the isoforms of interest. Since the availability of antibodies for plant epitopes is poor, we can overcome this obstacle by generating a reporter construct that expresses the whole (genomic) transcriptional unit and label the possible proteins with a commonly used tag (e.g. hemagglutinin or GFP). As the sequence of the isoforms is already known, we just need to analyze whether to place the tag at the N-terminus (immediately upstream of the transcriptional start site) or at the C-terminus (immediately downstream of the termination codon). We recommend the incorporation of both 5ʹ and 3ʹ complete UTR sequences in the reporter construct to avoid possible artifacts of stability, expression, or translation.

• We can also analyze the presence of the RNA isoform in association with ribosomes or polysomes. There are plenty of articles in the literature in which the authors performed ribosome profiling or analyzed polysome-associated RNAs, mainly in Arabidopsis, so we could rapidly check whether the isoforms in question are represented in the translatable pool of transcripts. If we are working with non-model organisms, we could perform a ribosome immunoprecipitation or density fractionation and an RT–PCR to assess whether the different isoforms of interest are represented in the translatable pool.

• We should also determine whether the isoforms of interest are present in the cytosolic fraction. This could be done by subcellular fractionation. Although this does not directly indicate that the isoforms are being actively translated, if the transcripts are not even present in the cytosol we can be certain that translation is not an option under the evaluated conditions.

To summarize, when studying the outcomes of alternative splicing and the functions of different transcripts to understand their impact on an organism’s phenotype (see Figs 2, 3), we need to ask at least seven questions:

1. 1) What type of alternative splicing event is under consideration?

2. 2) Does the isoform of interest contain a PTC?

3. 3) Is the isoform stable?

4. 4) Is the PTC-containing transcript retained in the nucleus?

5. 5) Is the isoform under study translated?

6. 6) Is the generated protein functional?

7. 7) If the RNA is not translated, could it have an active function?

This last question is of key relevance. We generally assume that molecular functions and impact on phenotypes are associated with proteins. Alternative splicing isoforms that are not substrates for translation are considered a mere by-product of gene expression that, most likely, reflect a down-regulation of the active/coding isoforms. However, RNAs can, and often do, perform active functions, like their protein counterparts. Specific RNAs act in different cellular machineries in the nucleus and in the cytosol. Moreover, it was recently reported that the human ASCC3 gene gives rise to both coding and non-coding transcript isoforms with opposite effects on transcriptional recovery after UV-induced DNA damage (Williamson et al., 2017). Hence, even though we often try to understand the role of a transcript by analyzing the putative encoded protein, we should be more open minded and consider possible active functions for the RNA molecules themselves.

Acknowledgements

We apologize to those authors whose work could not be discussed due to space limitations. We thank the whole community at IFIBYNE for the exciting and inspiring working atmosphere. We would like to thank Alexandra Elbakyan and all the friends and collaborators abroad who helped us to obtain the complete manuscripts to thoroughly read and analyze them. Science should be diverse, equal, inclusive, and open, always.

Author contributions

EP conceived the manuscript and, together with RST, guided its preparation; RST and FSR wrote most of the manuscript with contributions from FEA, CMC, LS, and EP; RST, FSR, FEA, and CMC prepared the figures; RST prepared the table. All authors read and approved the final version of the manuscript.

Conflict of interest

The authors declare no conflict of interest.

Funding

This work was supported by the Agencia Nacional de Promoción Científica y Tecnológica from Argentina (ANPCyT) grant to RST (PICT 2019 02274) and to EP (PICT 2017 1343, PICT 2019 01690, and PICT 2020 02865). RST is an ANPCyT postdoctoral fellow; FSR, FEA, and LS are fellows from Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET); CMC is an undergraduate student; EP is a career investigator from CONICET.

Data availability

No new data were generated in this work.

Author notes

These authors contributed equally to this work.