Mutually opposing activity of PIN7 splicing isoforms is required for auxin-mediated tropic responses in Arabidopsis thaliana

Summary

• Advanced transcriptome sequencing has revealed that the majority of eukaryotic genes undergo alternative splicing (AS). Nonetheless, little effort has been dedicated to investigating the functional relevance of particular splicing events, even those in the key developmental and hormonal regulators.
• Combining approaches of genetics, biochemistry and advanced confocal microscopy, we describe the impact of alternative splicing on the PIN7 gene in the model plant Arabidopsis thaliana. PIN7 encodes a polarly localized transporter for the phytohormone auxin and produces two evolutionarily conserved transcripts, PIN7a and PIN7b.
• PIN7a and PIN7b, differing in a four amino acid stretch, exhibit almost identical expression patterns and subcellular localization. We reveal that they are closely associated and mutually influence each other's mobility within the plasma membrane. Phenotypic complementation tests indicate that the functional contribution of PIN7b per se is minor, but it markedly reduces the prominent PIN7a activity, which is required for correct seedling apical hook formation and auxin-mediated tropic responses.
• Our results establish alternative splicing of the PIN family as a conserved, functionally relevant mechanism, revealing an additional regulatory level of auxin-mediated plant development.

Introduction

Auxin is an essential phytohormone, which plays a role in nearly all aspects of plant development. To flexibly adapt to rapidly changing environmental cues, directional auxin transport represents a highly dynamic means for triggering downstream morphogenetic processes. PIN-FORMED (PIN) auxin efflux carriers are among the key regulators in this respect. In recent years, many studies have revealed multiple mechanisms operating transcriptionally or posttranslationally on the capacity and directionality of PIN-mediated transport. However, little progress has been made in exploring the contribution of posttranscriptional regulation (Adamowski & Friml, 2015; Hrtyan et al., 2015).

Advances in high-throughput sequencing have revealed unexpected complexity within eukaryotic transcriptomes by alternative splicing (AS). Although the majority of AS transcripts may be functionally neutral (Darracq & Adams, 2013; Reddy et al., 2013; Chamala et al., 2015; Blencowe, 2017; Tress et al., 2017), several detailed studies have highlighted a plausible role for numerous AS events in physiologically relevant contexts, including those involved in plant developmental and hormonal pathways (Staiger & Brown, 2013; Hrtyan et al., 2015; Shang et al., 2017; Szakonyi & Duque, 2018). Pervious studies have described auxin-related defects resulting from the aberrant function of several regulators of AS (Kalyna et al., 2003; Casson et al., 2009; Retzer et al., 2014; Hrtyan et al., 2015; Tsugeki et al., 2015; Bazin et al., 2018). AS changes the subcellular localization of the auxin biosynthetic gene YUCCA 4 (Kriechbaumer et al., 2012), and differential splicing of an exitron (Marquez et al., 2015) inside the AUXIN RESPONSE FACTOR 8 (ARF8) results in developmental changes of generative organs (Ghelli et al., 2018). Likewise, a splice variant of MONOPTEROS (MP/ARF5) acts as a noncanonical regulator of ovule development (Cucinotta et al., 2020). In addition, AS of the Major Facilitator Superfamily transporter ZIFL1 interferes with auxin transport, influencing the mobility of PINs in the plasma membrane (PM) (Remy et al., 2013). These lines of evidence indicate that AS is an important player in auxin-dependent processes. However, no coherent functional model of any auxin-related AS event has been provided so far.

Here, we have investigated the functional significance of the protein isoforms arising through AS of the Arabidopsis thaliana PIN7 gene. PIN7 is, together with PIN3 and PIN4, a member of the PIN3 clade of PIN auxin efflux carriers (Bennett et al., 2014), which are required for a broad range of morphogenetic and tropic processes (Adamowski & Friml, 2015). We provide evidence that the evolutionarily conserved PIN7 isoforms mutually influence each other's dynamics within the PM and demonstrate that the coordinated action of both splice variants is required for auxin-mediated differential growth responses.

Materials and Methods

Plant material and plant growth conditions

All plant material, except tobacco cell cultures, was in the A. thaliana (L.) Heynh., Col-0 ecotype. The mutant and transgenic lines used were described previously: PIN3::PIN3-GFP (Zadnikova et al., 2010), PIN4::PIN4-GFP, PIN7::PIN7-GFP (Blilou et al., 2005), pin7-2 (Friml et al., 2003), pin3-3 pin4-101 (pin34), and pin3-3 pin4-101 pin7-102 (pin347) (Willige et al., 2013). The primers used in the study are listed in Supporting Information Table S1.

For in vitro cultivation, the seeds were surface-sterilized for 5 h with chlorine gas, plated on 0.5× Murashige & Skoog medium with 1% sucrose and 1% agar, and then stratified for 2 d at 4°C in the dark. Unless indicated otherwise, the seedlings were grown on vertically oriented plates for 4–6 d under a 16 h : 8 h photoperiod, 22 : 18°C, light : dark.

The following chemicals were used for treatments: brefeldin A (BFA), cytochalasin D (CytD), oryzalin (Ory), β-estradiol and 2,4-dichlorophenoxyacetic acid (2,4-D), all from Sigma. Radioactively labeled auxin accumulation assays were performed with [³H]NAA (naphthalene-1-acetic acid; 20 Ci mmol⁻¹; American Radiolabeled Chemicals, St Louis, MO, USA).

Plant phenotype analysis

Dynamic seedling development was tracked in a custom-made dynamic morphogenesis observation chamber equipped with blue and white LED unilateral light sources, infrared LED backlight and a camera for imaging in the infrared spectrum, controlled by a Raspberry Pi3B microcomputer (Raspberry Pi Foundation, Cambridge, UK). Image acquisition was controlled by a custom-written python script (https://github.com/lamewarden/RaPiD-boxes-software). For hypocotyl bending assays, seeds on the plate were first illuminated for 6 h with white light. The plates were then transferred to the observation chamber for 3–4 d. For hypocotyl phototropic bending experiments (Friml et al., 2002), the dark-grown seedlings were then illuminated for 20 h with unilateral white light and imaged every 20 min. For hypocotyl gravitropic bending experiments (Rakusová et al., 2011), the dark-grown seedlings were rotated by 90° clockwise and imaged for 30 h every 60 min. For tracking of apical hook development, the seeds were first illuminated for 6 h with white light. They were then transferred to the observation chamber and their development was recorded every 4 h for a total 150–200 h in the infrared spectrum. Germination time was set as time 0, when first traces of the main root were observed, individually for each seedling analyzed. Apical hook development was determined as a dynamic change of the angle between the immobile (nonbending) part of hypocotyl and the distal edge of cotyledons (Zadnikova et al., 2010), using ImageJ software (Rueden et al., 2017). At least 15 seedlings were analyzed for each line. Each experiment was done at least three times.

For analysis of protoxylem defects, 5-d-old light-grown seedlings were cleared and analyzed as described previously (Bishopp et al., 2011). For examining lateral root density, 8-d-old light-grown seedlings were cleared and observed with a differential interference contrast microscope. Lateral root density was calculated by dividing the total number of lateral roots and lateral root primordia by the length of the main root as described by Dubrovsky et al. (2009). Vertical growth index (VGI), defined as a ratio between main root ordinate and main root length, was quantified on 5-d-old seedlings as given previously (Grabov et al., 2005). For measuring the gravitropic set-point angle (GSA), plates with 14-d-old light-grown seedlings were scanned, and the angles between the vertical axis and five innermost 0.5 mm parts of the lateral root were determined as described previously (Roychoudhry et al., 2017). In all cases, 12–20 seedlings were analyzed for each line. Each experiment was done at least three times.

For decapitation experiments, 4-wk short-day grown plants were moved into long-day conditions to induce flowering. After the primary bolt reached 10–15 cm, the plant was decapitated. The number of rosette branches was recorded at 7, 10 and 14 d after decapitation (Greb et al., 2003; Waldie & Leyser, 2018). Rosette size was inspected in 18-d light-grown plants. Ten plants were analyzed for each line, and the experiment was done three times.

Tobacco cell lines and auxin accumulation assays

Tobacco cell line BY-2 (Nicotiana tabacum L. cv Bright Yellow 2) was cultivated as described by Müller et al. (2019). BY-2 cells were transformed with the pMDC7 constructs by cocultivation with Agrobacterium tumefaciens strain GV2260 as previously described (Petrasek et al., 2006; Müller et al., 2019). Transgene expression was maintained by cultivation on media supplemented with 40 μg ml⁻¹ hygromycin (Roche) and 100 μg ml⁻¹ cefotaxime (Sigma). Accumulation assays of radioactively labeled auxin were performed as previously described (Delbarre et al., 1996; Petrasek et al., 2006) with cells cultured for 2 d in β-estradiol or DMSO (dimethyl sulfoxide) mock treatments. Presented results are from three biological replicates for each representative PIN7a and PIN7b line and were confirmed for each on two other independent genotypes. Independent β-estradiol inductions were done within 3–9 months after establishing the cell suspension on lines, which showed comparable levels of the expressed PIN7-GFP signal.

Microscopy

Bright-field microscopy differential interference contrast (DIC) was conducted on an Olympus BX61 instrument (Olympus, Shinjuku, Tokyo, Japan) equipped with a DP50 camera (Olympus). Routine confocal microscopy was performed on an inverted Zeiss Axio Observer.Z1 containing the standard confocal LSM880 and Airyscan modules with ×20/0.8 DIC M27 air, ×40/1.2 W Kor FCS M27 air and ×63/1.40 Oil DIC M27 objectives (Zeiss). Gravity-induced polarity change experiments were carried out on a Zeiss Axio Observer.Z1 with vertically oriented sample position and the ×40/0.75 glycerol objective.

To observe the light-induced polarity change of PIN7a-GFP and PIN7b-RFP, 4-d dark-grown seedlings were irradiated for 4 h with unilateral white light and then imaged with a Zeiss Axio Observer.Z1 LSM880 with a vertically oriented sample position, as described (Willige et al., 2013). For analysis of gravity-induced polarity change, 4-d dark-grown seedlings were reoriented 90° clockwise and imaged 6 and 24 h after rotation, as described by Rakusová et al. (2011).

For BFA treatments, 5-d light-grown seedlings were transferred to liquid 0.5× MS media containing 50 μM BFA. The membrane : cytosol ratio was determined with ImageJ, and defined as the mean membrane signal intensity divided by the mean fluorescence in the cytosol. For cytoskeleton depolymerizing drug treatments (Geldner et al., 2001), 5-d light-grown seedlings were transferred to liquid 0.5× MS media supplemented with 20 μM CytD or 20 μM Ory. The cotreatments were done by the direct addition of BFA. For removal of BFA, before adding the cytoskeleton-depolymerizing compounds, seedlings were twice washed out with fresh media and then transferred to that supplemented with the respective cytoskeleton-depolymerizing drug.

For imaging of the expression of DR5v2::GFP in the apical hook, the green fluorescent protein (GFP) signal in 4-d-old etiolated seedlings was fixed in 4% paraformaldehyde (PFA; Sigma) overnight. Then, the tissue was hand-sectioned with a razor blade and observed.

Fluorescence recovery after photobleaching

For fluorescence recovery after photobleaching (FRAP) experiments, a Zeiss Axio Observer.Z1 equipped with the LSM880 confocal and Airyscan modules and the ×40/1.2 W Kor FCS M27 air objective was used. FRAP assays were done and evaluated as described previously (Sprague & McNally, 2005; Laňková et al., 2016). Normalization and time averaging were done with a python script available via GitHub (https://github.com/lamewarden/FRAP-normalization-script). The Slices Alignment plugin (Tseng et al., 2012) for ImageJ was used to eliminate cell movement caused by root growth.

FLIM-FRET analysis

For assessing the pairing specificity in Arabidopsis, the lines harboring all four combinations of G10-90::XVE>>PIN7a/b-GFP (donors) and PIN7a/b-RFP (acceptors) were generated. Before imaging, 4-d light-growth seedlings were transferred for 2 d to solid ½ Murashige & Skoog medium (½MS) supplemented with 5 μM β-estradiol. Transient expression in tobacco leaves was determined as previously published (Horák et al., 2008), omitting the β-estradiol induction to keep the transgene expression low (Bashandy et al., 2015). The abaxial epidermis of tobacco leaves was observed 3 d after infiltration.

The Förster resonance energy transfer by fluorescence lifetime imaging (FRET-FLIM) assay was done as previously described (Krejčí et al., 2015). Imaging was performed on an inverted Zeiss LSM 780 AxioObserver.Z1 equipped with the M27 Plan-Apochromat 20× air objective, external In-Tune laser (< 3 nm width, pulsed, 40 MHz, 1.5 mW) and GaAsP detectors. For conducting the FLIM-FRET assays, an HPM-100-40 hybrid detector associated with the photon counting module Simple-Tau 150 (compact TCSPC system based on the SPC-150N device) was utilized, using a DCC-100 detector controller card (Becker & Hickl GmbH, Berlin, Germany).

Statistics and sequence analysis

For comparison of means between the two groups, Student's t-test was applied. Statistical analysis of multiple groups was performed by one-way ANOVA with subsequent Tukey HSD post-hoc test. For temporal analysis of seedling development, regions with approximately linear courses were selected. Those parallel to the x-axis were evaluated as mean angles of each seedling within the sample by one-way ANOVA. Regions with an apparent slope were fitted by linear regression (Table S2). The regression line intercepts with the y-axis and the slope values were evaluated with one-way ANOVA (Table S3). All statistical tests, including the Shapiro–Wilk test for assessing the normal distribution of variables, were performed in Rstudio IDE (R-Studio, Boston, MA, USA). In the box plots, the whisker length was set as Q ± 1.5 × IQR, where Q is the corresponding quartile and IQR is the interquartile range. For the multiple sequence alignments, the protein sequences were assembled with the Clustal Omega algorithm (Sievers et al., 2011) and graphically outlined by Mega-X, using the default ClustalX color code (Kumar et al., 2018).

Additional methods

Details of DNA manipulation, quantitative reverse transcription PCR (qRT-PCR), protein extraction, immunoblotting, and coimmunoprecipitation methods and procedures are provided in Methods S1.

Results

Arabidopsis PIN7 and PIN4 produce two evolutionary conserved AS transcripts

Our previous survey (Hrtyan et al., 2015) revealed that several genes involved in auxin-dependent processes undergo AS. Among them, closely related paralogs from the PIN3 clade of auxin transporters, PIN4 and PIN7 (but not PIN3), are regulated by the same type of AS (Petrasek et al., 2006; Bennett et al., 2014; Hrtyan et al., 2015). The resulting transcripts, denoted as a and b, differ in the AS donor site position in the first intron (Fig. 1a). The differentially spliced region corresponds to a four amino acid motif inside the large internal hydrophilic loop (Ganguly et al., 2014; Nodzyński et al., 2016) of the integral PM transporter (Fig. 1a,b). We examined the quantities of individual reads spanning the exon junctions in the respective gene region from the Arabidopsis root tip and in several other available transcriptomes from different tissues and organs (Klepikova et al., 2016; Cheng et al., 2017; Ruzicka et al., 2017) (Fig. 1c; Table S4). In accordance with other data sets (Martín et al., 2021), we found that each of the PIN4 and PIN7 splice isoforms is expressed abundantly in all tissues, independently of the inspected data set. We also identified a minor PIN4c (Marquez et al., 2012; Hrtyan et al., 2015) transcript (but not corresponding PIN7c), which comprised c. 3–7% of the PIN4 exon1–exon2 spanning reads (Fig. 1a; Table S4). Other occasionally observed transcripts (also corresponding to the other exon junctions) were not seen among different RNA-sequencing (RNA-seq) data sets. It thus appears that PIN7 and PIN4 are processed into two and three splice isoforms, respectively, and that PIN7a and b (or PIN4a and b) are expressed in most plant organs at comparable levels, possibly with a moderate prevalence of PIN7b over PIN7a.

Fig. 1

Alternative splicing (AS) of the Arabidopsis thaliana PIN7 and PIN4 genes leads to two evolutionarily conserved functionally different transcripts. (a) Scheme of coding regions of the PIN3 clade genes in A. thaliana. The alternative donor splice site at the end of the first exon of PIN7 and PIN4, respectively, but not PIN3, results in two transcripts differing in 12 nucleotides. This sequence (orange) corresponds to the protein motif located in the long internal hydrophilic loop of the transporter. There is also an additional PIN4c transcript present in publicly available transcriptomes. (b) Amino acid alignment of the region around the four amino acid motif changed by AS in the PIN3 clade proteins (boxed in pink) in A. thaliana, including the closest PIN paralogs. (c) Table showing the number of RNA-seq reads spanning the exon1–exon2 junction corresponding to the detected PIN7 transcripts in selected A. thaliana tissue sources. Their ratio was calculated as a percentage of total reads mapped to this area, as assessed from the genome browser graphic interface. (d) Protein sequence alignments showing the conservation of AS in the example PIN7 isoform orthologs among eudicots. For the sequences of PIN7s from Nicotiana tabacum, no homologous AS was seen in the corresponding region. (e, f) Phototropic bending of the etiolated pin347 seedlings carrying the PIN7a-GFP and PIN7b-RFP (e), and the PIN7a-RFP and PIN7b-GFP constructs (f). (g, h) Quantification of the primary root protoxylem defects (g) and lateral root primordia initiation (h) of the pin347 seedlings harboring the PIN7a-RFP and PIN7b-GFP constructs. In (e), (f) and (h), the middle line corresponds to the median, the box corresponds to the 25% and 75% quantiles, the whiskers are the minima and maxima, and dots represent single data points. Asterisks indicate a significant difference between the respective line and the pin347 mutant (**, P < 0.01; ***, P < 0.001 by one-way ANOVA). For each line in each experiment, at least 15 seedlings were analyzed and confirmed on eight independent transgenic lines.

Functionally relevant AS events are commonly evolutionarily conserved (Keren et al., 2010; Reddy et al., 2013). Therefore, we looked for available validated transcripts to determine whether similar splicing events occur in orthologous PIN3 clade genes in other dicot species (Bennett et al., 2014; O’Leary et al., 2016). We found examples of such mRNAs besides members of the family Brassicaceae, for instance in Hibiscus syriacus (Malvaceae) and Abrus precatorius (Fabaceae), which document the conservation of this AS event among rosids, but not, for example, in Nicotiana tabacum (Fig. 1d). PIN4c did not show any deeper evolutional conservation. Thus, at least some genes of the PIN3 clade are regulated by the same type of AS across several plant families, which suggests that these AS events may have a biologically relevant function.

PIN7 splice isoforms are functionally distinct

We expressed fluorescently tagged cDNA versions of the respective PIN7 and PIN4 transcripts under their native promoters in A. thaliana. Their overall expression patterns resembled those of the PIN7-GFP and PIN4-GFP lines made on the basis of the genomic sequence (Fig. S1a–d). Next, we tested their ability to complement the phenotypes associated with the PIN7 locus. Phototropic hypocotyl bending is among the classical assays for testing the activity of PIN3 clade proteins (Friml et al., 2002; Willige et al., 2013). As pin7-2 loss of function mutants show a weak phenotype under laboratory conditions (Friml et al., 2003; Blilou et al., 2005; Willige et al., 2013), we employed a triple pin3-3 pin4-101 pin7-102 knockout (pin347) as a genetic background, which almost completely lacks the phototropic response (Willige et al., 2013). Here, PIN7a-GFP cDNA rescued the phototropic bending defects, while the PIN7b-RFP cDNA did not show any effect, regardless of whether the native (Fig. 1e) or a stronger endodermal SCR promoter (Rakusová et al., 2011) (Fig. S1e) was used. Tag choice does not appear to have any effect in these assays, as evidenced by the lines where the fluorophore sequences have been swapped (Fig. 1f). Together, these data indicate that the motif changed by AS alters the function of the PIN7 protein in Arabidopsis.

The PIN3 clade auxin efflux carriers have been implicated in many other instances of auxin-mediated development (Adamowski & Friml, 2015). We therefore hypothesized whether the role of the particular isoform could be prevalent in some of them. These processes include: determination of root protoxylem formation (Bishopp et al., 2011) (Fig. 1g), lateral root density (Swarup et al., 2008) (Fig. 1h), vertical direction of root growth (Friml et al., 2002; Kleine-Vehn et al., 2010; Pernisova et al., 2016) (Fig. S1f), lateral root orthogravitropism (Rosquete et al., 2013) (Fig. S1g), gravity-induced hypocotyl bending (Rakusová et al., 2011) (Fig. S1h), number of rosette branches after decapitation (Bennett et al., 2016) (Fig. S1i) and overall rosette size (Bennett et al., 2016) (Fig. S1j). In all assays, we found that the PIN7a-GFP cDNA usually almost completely rescues the pin347 phenotypes, while the contribution of PIN7b-RFP is smaller or even undetectable.

Fluorescent reporters reveal highly overlapping PIN7a and b expression

The levels of PIN7 (and PIN4) AS transcripts seem to be comparable in most organs (Fig. 1c; Table S4). However, this may not describe the actual situation at the resolution of individual cells. To address this, we analyzed the P7A₁G and P7BR fluorescent reporters (Kashkan et al., 2020), which allow for monitoring the activity of the AS of PIN7 in planta and in situ (Fig. 2a). Indeed, in most cells, including the primary root tip (Fig. 2b) or the hypocotyl during the light bending assay (Fig. 2c), we observed generally overlapping expression of both isoforms without any apparent tissue preference. However, there were several instances in the vegetative tissue where the ratio of reporter signals appears to be uneven. These include early lateral root primordia (Fig. 2d), the stomatal lineage ground cells of the cotyledons (Fig. 2e), and the concave side of the opening apical hook (Fig. 2f), where the PIN7b-RFP signal prevailed over that of PIN7a-GFP. Overall, these data corroborate the presence of both isoforms in most cells and suggest that they may function in a coordinated manner.

Fig. 2

The PIN7 AS reporter system shows overlapping PIN7a and PIN7b expression in most Arabidopsis tissues. (a) A scheme of the PIN7 splicing reporter, which consists of two separate constructs. The PIN7a-GFP part (PIN7A₁G) contains two point mutations (dashed arrows) and leads to the frameshift and its subsequent restoration when the PIN7a transcript is solely produced. The PIN7b-RFP reporter (PIN7BR) carries a premature stop codon inside the protruding PIN7a region (dashed arrow below). (b, c) Overlapping PIN7a-GFP and PIN7b-RFP reporter expression in the root tip (b) and in the etiolated hypocotyl following 4 h of unilateral light stimulation (c). (d–f) Differential expression of the P7A₁G/P7BR reporter in the lateral root primordia (d), in the stomatal lineage ground cells of the cotyledon epidermis (e) and in the area covering the apical hook (f). Yellow arrows, direction of unilateral light; white arrows, areas of different reporter expression. For each tissue, at least nine images were analyzed. Bars: (b–e) 50 μm; (f) 100 μm.

The combined activity of both PIN7a and PIN7b is required for apical hook formation and tropic responses

The generally overlapping expression of both isoforms and an occasionally exaggerated response of the pin347 mutants containing the PIN7a-GFP transgene (Fig. S1h) prompted us to record the temporal dynamics of the processes linked to PIN3 clade function. We were able to capture subtle phenotypes of the single pin7-2 knockout allele (Friml et al., 2003) during hypocotyl phototropic bending (Fig. 3a) or apical hook formation (Fig. 3b) (Zadnikova et al., 2010), where we again observed accelerated development conferred by expression of PIN7a-GFP. Thus, temporal analysis of the complemented pin7-2 lines underlines that AS of PIN7 has a role in plant development.

Fig. 3

PIN7a-GFP and PIN7b-RFP regulate Arabidopsis apical hook development and hypocotyl bending in a mutually antagonistic manner. (a, b) Subtle phenotypic differences between the wild-type and pin7-2 mutants expressing the PIN7a-GFP and PIN7b-RFP cDNAs, documented by temporal analysis of phototropic bending (a) and apical hook development (b), where even a tendency of mild gain-of-function phenotypes resulting from PIN7a-GFP expression was detected. (c, d) Temporal analysis of etiolated pin347 seedlings carrying both PIN7a-GFP and PIN7b-RFP cDNA transgenes during apical hook development (c) and hypocotyl phototropic bending (d). Inset in (c): levels of the transcript corresponding to PIN7a-GFP cDNA in the pin347 background, normalized to the native PIN7 levels in the pin34 lines as determined by qRT-PCR. The statistical evaluation, summarized below each panel, was conducted on the regions showing an approximately linear temporal course (‘lin.’). lin. 1 (hypocotyl opening plateau phase) on (a) and (d) and lin. 2 (apical hook maintenance phase) on (b) were evaluated as mean angles of each individual by one-way ANOVA. For lin. 2 on (c), and lin. 3 (apical hook opening phase before full opening) on (b) and (c), the regions were fitted with linear regression, and the y-axis intercepts for each seedling were assessed by one-way ANOVA. For all experiments, at least 15 seedlings were analyzed for each data point. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by one-way ANOVA; ns, non-significant. Data are means ± SE.

To gain detailed insight into the possible shared role of both isoforms, we analyzed apical hook development in the pin34 mutants (Wilige et al., 2013) along with the pin347 lines that carried the combinations of both PIN7 cDNA constructs. Presence of the PIN7b-RFP cDNA had generally little effect on the severe pin347 phenotype. Expression of PIN7a-GFP cDNA in pin347 indeed led to partial rescue of the phenotypic defects, surpassing the values observed for pin34. Interestingly, the simultaneous expression of both PIN7a-GFP and PIN7b-RFP suppressed the dominant effects conferred by PIN7a-GFP cDNA alone and phenocopied the pin34 mutant (Fig. 3c). It therefore appears that the two PIN7 isoforms act in a coordinated manner.

Relatively complicated apical hook development involves several bending steps (Zadnikova et al., 2010), and the splicing reporter shows differences in epidermal expression of PIN7a and b on its inner side (Fig. 2f). The hypocotyl photo- and gravitropic response requires only a single bending (Rakusová et al., 2011, 2016), and the reporter fluorescence ratio remains unchanged during stimulation (Fig. 2c). This can therefore suggest whether one can account for the antagonistic behavior of both isoforms by differential expression or by another mechanism, probably occurring at the cellular level. Similar to apical hook development, the presence of PIN7b-RFP cDNA had a marginal effect on the pin347 phenotype during the whole period recorded, while the expression of PIN7a-GFP led to faster bending than that observed for the pin34 mutant. Finally, the presence of both PIN7a-GFP and PIN7b-RFP cDNAs in pin347 was reminiscent of the pin34 phenotype in both phototropic and gravity assays (Figs 3d, S2a). The transcript levels attributed to PIN7a-GFP cDNA did not exceed the internal levels of the PIN7 transcript in the pin34 mutant, nor did the coexpression of PIN7b-RFP reduce the PIN7a-GFP levels in the pin347 seedlings (Figs 3c, S2b–d). We therefore conclude that the joint activity of both PIN7 isoforms located in the same group of cells is required for correct auxin-mediated tropic responses, and that these developmental changes cannot be ascribed to the altered balance of the PIN7 transcripts during tropic assays.

PIN7a, but not PIN7b, can form auxin maxima in planta

To further investigate the impact of AS on PIN7 protein function, we validated the formation of the downstream auxin response maxima in apical hooks by crossing the DR5v2::GFP transcriptional auxin reporter (Liao et al., 2015) into lines where the phenotypic changes have been temporally monitored (see Fig. 3). The reporter expression pattern in the forming apical hooks of pin34, PIN7a-GFP pin347 and PIN7a-GFP PIN7b-RFP pin347 resembled the situation in the wild-type (Zadnikova et al., 2016), whereas the distribution of DR5v2::GFP in the PIN7b-RFP pin347 line was indistinguishable from that of pin347 (Fig. 4a). These results underline that the PIN7b isoform alone is unable to establish the morphogenic auxin gradients and that the observed developmental changes occur in line with previously described mechanisms (Friml et al., 2002; Zadnikova et al., 2010, 2016; Willige et al., 2013).

Fig. 4

Functional analysis of the PIN7a-GFP and PIN7b-RFP interaction. (a) PIN7a but not PIN7b cDNA is able to promote expression of the DR5v2::GFP auxin reporter at the concave side of the Arabidopsis apical hooks (white semicircle) of pin347 seedlings. Note the asymmetric distribution of the nuclear-localized signal in the ground tissue and epidermis, instructive for proper apical hook development. (b) Comparable [³H]-NAA accumulation kinetics in tobacco BY-2 cells following induction of the G10-90::XVE>>AtPIN7a-GFP and AtPIN7b-GFP transgenes with decreasing doses of β-estradiol. (c) The effect of 50 µM brefeldin A (BFA) on the intracellular distribution of PIN7a-GFP and PIN7b-RFP in Arabidopsis root tips. Arrows mark incomplete colocalization of PIN7a and PIN7b in the BFA bodies. (d) Temporal dynamics of the BFA-mediated aggregation of PIN7a-GFP and PIN7b-GFP inside cells. The values were determined as a ratio of fluorescence intensities between the cytoplasm and PM. (e, f) Immunoblot of the anti-RFP-coupled agarose beads precipitated from protein extracts from the lines expressing G10-90::XVE>>PIN7a-GFP, G10-90::XVE>>PIN7b-GFP (negative controls) or a cross between G10-90::XVE>>PIN7a-GFP and G10-90::XVE>>PIN7-RFP, probed with anti-GFP and anti-RFP antibody, respectively. The electrophoretic gel was loaded with the crude protein extract (input) and immunoprecipitation eluate (eluate) from Arabidopsis seedlings treated with 5 µM β-estradiol (e); the loading control is shown on (f). (g, h) PIN7a and PIN7b can form homo- and heteromers on PM in vivo. GFP fluorescence lifetime (τ) was determined from the tagged PIN7a and PIN7b proteins and compared to that of PIN7a-GFP alone, presented as a representative image heat map (g) and quantified (h). Proteins were expressed under control of the G10-90::XVE promoter in Arabidopsis root tips and induced with 5 µM β-estradiol. On the box plots, the middle line corresponds to the median, the box corresponds to the 25% and 75% quantiles, and the whiskers represent the minima and maxima. On (d) and (h), asterisks indicate a significant difference between the respective protein pair and the PIN7a-GFP control (*, P < 0.05; ***, P < 0.001 by one-way ANOVA). On (d), data are means ± SE. For each line, on (a) and (b), n ≥ 9; on (c), (d), (g) and (h), n ≥ 12. Bars: (a) 50 µm; (c) 20 µm; (g) 10 µm.

PIN7a and PIN7b transport auxin with comparable rates in tobacco cells

Because PIN7a and PIN7b showed differential ability to generate auxin maxima in planta, we investigated whether both protein isoforms indeed function as auxin transporters. We expressed the PIN7a and b cDNA variants tagged with GFP under control of the β-estradiol-driven promoter in the tobacco BY-2 tissue culture system (Petrasek et al., 2006; Müller et al., 2019). We confirmed that PIN7a (Petrasek et al., 2006), however, as well as the PIN7b isoform, are functional auxin transporters. To this end, we also analyzed the quantitative aspects of PIN7-mediated transport, using multiple concentrations of β-estradiol. Following induction of both transgenes, we observed a comparable decrease of radioactively labeled auxin accumulation inside BY-2 cells for both lines in the whole β-estradiol dose range used (Fig. 4b), consistent with the similar PIN7a and PIN7b protein levels (Fig. S3a–c). The time course of the auxin accumulation drop appeared to be similar for both constructs (Fig. 4b). Hence, PIN7a and PIN7b code for true auxin exporters, and they seem to transport auxin in tobacco cell cultures at similar rates.

PIN7a and PIN7b differ in protein subcellular dynamics

Polarity and dynamic intracellular trafficking are essential functional attributes of PINs (Adamowski & Friml, 2015). However, the PIN7 isoforms did not prominently differ from each other in terms of polarity or general subcellular localization in the root tip at a given resolution limit (Fig. 4c). The anterograde trafficking of proteins towards the PM can be effectively blocked by the fungal toxin BFA. It leads to internal accumulation of the membrane-bound PINs into characteristic BFA bodies (Geldner et al., 2001; Kleine-Vehn et al., 2010). During time-lapse imaging, we observed that while PIN7a-GFP accumulated readily in these intracellular aggregates, the PIN7b-RFP BFA bodies were less pronounced and colocalized with PIN7a-GFP incompletely (Fig. 4c). To exclude the effects of diverse fluorescent tags, we compared the response of PIN7a-GFP to the PIN7b-GFP cDNA lines. The BFA-mediated aggregation of PIN7b-GFP indeed showed a moderate delay compared with PIN7a-GFP (Fig. 4d). This suggests that the PIN7 isoforms differ in the speed of their intracellular trafficking pathways or delivery to the PM, and the choice of the tag does not appear to interfere significantly with the subcellular dynamics of PIN7.

PIN polarity does not seem to strictly require the cytoskeleton (Glanc et al., 2019), but subcellular PIN trafficking has been proposed to be mediated by two distinct pathways (Geldner et al., 2001; Glanc et al., 2019). The first depends on actin filaments (CytD-sensitive) and occurs in most root meristem cells. The second (Ory-sensitive) utilizes microtubules and is linked to cytokinesis. Drugs that depolymerize actin filaments (CytD) and tubulin (Ory) (Geldner et al., 2001; Kleine-Vehn et al., 2008) showed only little effect on the intracellular localization of both PIN7 isoforms when applied alone (Fig. S3d,e). Pretreatment with CytD prevented the formation of the BFA bodies (Geldner et al., 2001) containing both PIN7 isoforms (Fig. S3f). Yet, when we applied Ory before the BFA treatment, the BFA compartments with PIN7a-GFP and PIN7b-RFP associated in only very weakly colocalizing structures (Fig. S3g). Next, we tested how the cytoskeleton is involved in the trafficking of both PIN7 isoforms from the BFA bodies to the PM by washing out BFA with CytD or Ory (Geldner et al., 2001). In the presence of CytD, PIN7a-GFP largely persisted inside the BFA bodies, while the PIN7b-RFP signal was almost absent in these aggregates (Fig. S3h). We generally did not see any difference between the two isoforms when BFA was washed out with Ory (Fig. S3i). These data thus suggest that both PIN7 isoforms use vesicle trafficking pathways assisted by a common cytoskeletal scaffold. These pathways differ in their dynamics and the endomembrane components involved and are consistent with previous findings that individual PINs can utilize multiple PM delivery routes (Boutté et al., 2006; Kleine-Vehn et al., 2008).

PIN7a and PIN7b do not differ in tropic stimuli-mediated polarity change or dimer formation

Auxin transporters of the PIN3 clade change their polar localization on the PM by reacting to various environmental cues, particularly by switching the light or gravity vectors (Friml et al., 2002; Ding et al., 2011; Rakusová et al., 2011). PIN3 relocation in response to gravity in columella root cells is seen in as little as 2 min, while the relocation of PIN7-GFP requires c. 30 min to be detected (Friml et al., 2002; Kleine-Vehn et al., 2010; Pernisova et al., 2016; Grones et al., 2018). We examined plants harboring both PIN7a-GFP and PIN7b-RFP cDNAs under short and long gravitropic stimuli. We did not find any difference in relocation dynamics between the two isoforms in these experiments (Fig. S4a–b). We also observed no difference in the polarity change between PIN7a-GFP and PIN7b-RFP in hypocotyl gravitropic (Rakusová et al., 2011, 2016) and phototropic (Ding et al., 2011) bending assays (Fig. S4c); due to the limited transparency of hypocotyls, we used lines expressing the cDNAs under control of a stronger SCR promoter (Rakusová et al., 2011). The different subcellular pathways driving both PIN7a-GFP and PIN7b-RFP cargos are thereby not connected with their ability to change polarity in response to tropic stimuli.

Alternative splicing frequently changes the protein function by altering their ability to multimerize (Kelemen et al., 2013), and PIN proteins have recently been found to associate in higher-order complexes in tissue cell cultures (Teale et al., 2020; Abas et al., 2021). To independently corroborate these findings, we examined extracts from both Arabidopsis seedlings and BY-2 cells, expressing the respective cDNAs, by native protein gel electrophoresis and western blotting (Leitner et al., 2012). Probing with anti-GFP antibody indeed revealed a faint protein band at the predicted size of tagged PIN7 dimers in both tissue sources (Fig. S4d–g). Next, using an anti-red fluorescent protein (RFP)-specific antibody coupled with agarose beads (Waidmann et al., 2018), we performed coimmunoprecipitation of crude extracts from Arabidopsis seedlings inducibly expressing both PIN7a-GFP and PIN7b-RFP cDNAs. Examining the precipitates by western blotting further indicated that isoforms can interact (Fig. 4e,f). We further validated these findings by FRET-FLIM, analyzing the cDNAs expressed stably in Arabidopsis seedlings and transiently in tobacco leaves (Figs 4g,h, S4h,i). Here, we also tested the combinations of the GFP- and RFP-tagged PIN7a and PIN7b cDNAs to determine their pairing preference. Shortening of the GFP fluorescence lifetime was seen regardless of the isoform interaction examined and also when expressed under the SCR promoter (Fig. S4j). However, no changes were observed when we used another protein located on the PM, aquaporin PIP2-GFP (expressed under a strong promoter) (Fig. S4k). These data therefore suggest that the PIN7 splice isoforms can mutually associate into dimers (or other higher order complexes), but it seems the strength of their mutual interaction remains comparable.

PIN7a and PIN7b show distinct mobility within PM

Numerous studies have used FRAP to investigate the dynamic turnover of various proteins, including PINs, on the PM. Some suggest that decreased lateral mobility may impact PIN-mediated auxin transport (Men et al., 2008; Martinière et al., 2012; Laňková et al., 2016; Glanc et al., 2021). We therefore bleached a region of the PM signal in the root meristem of the PIN7a-GFP and PIN7b-GFP cDNA lines and measured the FRAP in this area. Notably, PIN7a-GFP showed a slower recovery of fluorescence than PIN7b-GFP (Fig. 5a); the use of either a GFP or RFP tag did not markedly influence fluorescence recovery (Fig. S5a). We also generated plants carrying a PIN3::PIN3Δ-RFP cDNA construct, which lacks the GETK motif corresponding to the four amino acids absent in PIN7b (Figs 1b, 5b). The PIN3Δ-RFP signal displayed incomplete colocalization with the wild-type PIN3-GFP variant in the BFA bodies (Fig. S5b) and faster recovery on the PM, similar to that observed for PIN7a and PIN7b (Fig. 5a,b). It therefore appears that the motif altered by AS of PIN7 is required for regulation of the dynamics of individual isoforms within the PM.

Fig. 5

cDNA-encoded PIN7a and PIN7b mutually influence their stability within the PM, as assessed by fluorescence recovery after photobleaching (FRAP). (a) Differential FRAP of PIN7::PIN7a-GFP and PIN7::PIN7b-RFP on the PM of cells in Arabidopsis root tips. Example images of the bleached region (region of interest (ROI)) are shown in the inset. (b) FRAP analysis of PIN3Δ-RFP, which lacks the four amino acid motif to mimic the properties of the PIN7b isoform (inset), compared to the control PIN3-GFP cDNA line in Arabidopsis root tips. (c) FRAP of SCR::PIN7a-GFP and SCR::PIN7b-RFP expressed in the pin347 mutant background in Arabidopsis root tips. Lines with a single construct are marked as ‘single’, while lines carrying both transgenes are marked as ‘cross’. (d) FRAP analysis of BY-2 tobacco cell lines expressing G10-90::XVE>>AtPIN7a-GFP and AtPIN7b-GFP, following induction with 5 µM β-estradiol. Data are means ± SE. For each data point, at least 20 ROIs were measured. The normalized data points were fit with monoexponential curves. Bar, (a) 5 µm.

The fluorophore-tagged PIN7 cDNAs showed a relatively weak signal under the natural promoter, biasing the obtained data by extensive bleaching by prolonged confocal imaging (Fig. 5a). Therefore, we utilized the SCR promoter-driven lines, which displayed higher expression levels, but still lower intensities than the commonly used genomic DNA-derived PIN7::PIN7-GFP (Blilou et al., 2005) (Fig. S5c,d). Indeed, these lines revealed a finer difference in FRAP compared to those under control of the native promoter (Fig. 5c). The FRAP experiments (Fig. 5a) were conducted initially on the pin347 lines, which expressed the corresponding transgenes separately. Therefore, we compared the FRAP values of the single PIN7a-GFP and PIN7b-RFP with those carrying these constructs simultaneously in the pin347 genetic background. Strikingly, coexpression of PIN7a-GFP decreased the FRAP of PIN7b-RFP and, vice versa, the presence of PIN7b-RFP increased the FRAP of PIN7a-GFP (Fig. 5c). This shows that the PIN7 isoforms mutually influence their mobility within the PM, presumably by the association in a molecular complex.

In addition, PIN7a and PIN7b did not differ markedly in their ability to transport auxin in the BY-2 cells (Figs 4b, S3a–c). We were unable to detect homologous AS events in plant lineages related to tobacco (see Fig. 1d). We tested FRAP of the Arabidopsis PIN7a- and PIN7b-GFP on the PM in BY-2 cells. Interestingly, recovery of both isoforms showed a virtually identical course (Fig. 5d). Together, the observations here and above support the idea that both isoforms physically interact and mutually influence each other’s mobility within the PM, which leads to a differential impact on the plant phenotype.

Discussion

In this work, we present functional evidence that AS increases the diversity among PIN auxin transporters in A. thaliana and introduce AS as a valid regulator of auxin transport. Concentrating on PIN7, we reveal that both PIN7a and PIN7b isoforms (differing by the presence of a four amino acid stretch) are required together for proper apical hook formation and hypocotyl tropic responses. Genetically evidenced cooperative modes of action of splice isoforms have been previously proposed during Arabidopsis development (Szakonyi & Duque, 2018). The seed dormancy regulator DELAY OF GERMINATION 1 (DOG1) is processed into five mRNAs. Expression of only two or more DOG1 cDNAs under the native promoter rescues the dog1 phenotype by probably synergistic stabilization of the protein by its multimerization (Nakabayashi et al., 2015). The transcriptional factor HYH (HY5 HOMOLOG) possesses an isoform that lacks a domain required for proteasomal degradation, which leads to its increased stability and probably works as a semidominant splice variant (Sibout et al., 2006; Szakonyi & Duque, 2018); this is probably also the case of the AS of MONOPTEROS (Cucinotta et al., 2020). Together with, for example, a partly similar mechanism described for the temperature-mediated regulation of flowering (Lee et al., 2013; Posé et al., 2013), the example of AS of PIN7 is among rarely described instances where the mutually antagonistic effects of two splice isoforms are observed in the developmental context.

The PIN7a/b (PIN4a/b) AS event is conserved among rosids, a plant phylogenetic group that radiated more than 100 million years ago (Li et al., 2019). It appears that it has been present in the ancestral rosid PIN3-clade and, in Arabidopsis, it has remained in the PIN7 and PIN4 genes. The phenotypic manifestations of the individual PIN3-clade genes are weak but detectable under standard laboratory growth conditions (Robert et al., 2013; Simaskova et al., 2015; Rosquete et al., 2018; Ogura et al., 2019; Fig. 3a,b), and it seems that the subtle defects resulting from disruption of the single PIN3-clade genes are more pronounced under specific environmental cues (Adamowski & Friml, 2015; Ogura et al., 2019), where joint AS of PIN7 and PIN4 may play a role. Although publicly available transcriptomic resources do not indicate that expression of the individual PIN7 or PIN4 isoforms is dramatically changed under a wide range of experimental conditions (Martín et al., 2021), we observed that expression of PIN7a could be lowered by the application of exogenous auxin (Kashkan et al., 2020). Given that disabling several factors involved in AS leads to auxin-related defects (Kalyna et al., 2003; Casson et al., 2009; Retzer et al., 2014; Hrtyan et al., 2015; Tsugeki et al., 2015; Bazin et al., 2018), this mode of regulation of AS of PIN7 (and PIN4) suggests the existence of an unknown auxin-mediated pathway, active at the posttranscriptional level.

Alternative splicing modifies protein properties variably. It affects subcellular localization, ligand binding affinity, enzymatic or transporting activities, protein stability, or the presence of covalent posttranslational modifications (Stamm et al., 2005; Kelemen et al., 2013). Different covalent modifications alter the subcellular trafficking of most PINs. Phosphorylation sites on serine, threonine or tyrosine residues of various PINs have been identified; their phosphorylation status also changes PIN-mediated tropic responses (Rademacher & Offringa, 2012; Barbosa et al., 2018; Zwiewka et al., 2019). In addition, PIN ubiquitination (on lysines) and controlled proteolytic degradation act in auxin-mediated processes (Leitner et al., 2012). However, none of the candidate residues required for these modifications is present in the vicinity of the amino acid stretch changed by AS. This region in the PIN long hydrophilic loop is, inferred from the low amino acid conservation (Fig. 1b), probably intrinsically disordered. Therefore, one can speculate whether just the length of this motif is critical for functional interaction among internal structural domains (Buljan et al., 2013). These structural differences may modulate the affinity of PINs to bind the factors required for the entry and presence in the membrane secretory pathways, and thus participate in the PIN subcellular dynamics and directional auxin transport.

Here we have validated the most plausible hypotheses (Adamowski & Friml, 2015) to provide a mechanistic view on the observed mutual functionality of PIN7a and PIN7b. This AS event neither affects the ability of PIN7 carriers to transport auxin per se nor influences their ability to relocate after the tropic stimulus. Although we occasionally noted some expression irregularities of the reporter (Fig. 2d–f), it seems that the expression changes are unlikely to explain the distinct contributions of PIN7 isoforms to plant development. Instead, our data favor the scenario that PIN7a and PIN7b functionally interact at the protein level. Moreover, we demonstrate with orthogonal experimental approaches that PIN7 isoforms closely associate in native plant tissues and influence each other’s mobility within the PM. It has previously been suggested that PINs are complexed inside stable integral PM clusters (Feraru et al., 2011; Li et al., 2021). Although the proposed molecular model may be more complicated (Zourelidou et al., 2014; Weller et al., 2017), the mobility of PINs within the PM (described as lateral diffusion) has been recently linked with the activity of the AGC kinases and MAB4/MEL proteins (Glanc et al., 2021), whose loss-of-function mutations phenocopy pin mutants (Bennett et al., 1995; Furutani et al., 2007). We reveal that mobility on the PM is indeed required for PIN-mediated auxin transport. In this scheme, PIN7a, detained inside the PM microdomains, transports auxin, but dynamic PIN7b does not. Hence, PIN7b binds PIN7a, reducing its presence in these membrane domains and thereby impedes polar auxin flow to mediate auxin developmental responses (Fig. S5e).

Acknowledgements

We thank Claus Schwechheimer for the pin34 and pin347 seeds, Yuliia Mironova for technical assistance, Ksenia Timofeyenko and Dmitry Konovalov for help with the evolutional analysis, Konstantin Kutashev and Siarhei Dabravolski for assistance with FRET-FLIM, Huibin Han for advice with hypocotyl imaging, Karel Müller for the initial qRT-PCR on the tobacco cell lines, Stano Pekár for suggestions regarding the statistical analysis of the morphodynamic measurements, and Jozef Mravec, Dolf Weijers and Lindy Abas for their comments on the manuscript. This work was supported by the Czech Science Foundation (projects 16-26428S and 19-23773S to IK, MH and KRůžička, 19-18917S to JHumpolíčková and 18-26981S to JF), and the Ministry of Education, Youth and Sports of the Czech Republic (MEYS, CZ.02.1.01/0.0/0.0/16_019/0000738) to KRůžička and JHejátko. The imaging facilities of the Institute of Experimental Botany and CEITEC are supported by MEYS (LM2018129 – Czech BioImaging and CZ.02.1.01/0.0/0.0/16_013/0001775). The authors declare no competing interests.

Author contributions

IK, MH, KRetzer, JHumpolíčková, AJ, RF, ZV, DR and JP conducted experiments. SS and JF provided unpublished material. IK, MJF, JP and KRůžička conceptualized the research. IK, KRetzer, JHumpolíčková, JP, TBJ, JHejátko, JF, JP and KRůžička designed experiments. KRůžička and IK wrote the manuscript. All authors read and approved the final version of the manuscript.

Open Research

Data availability

The data that support the findings of this study and all generated material, such as cloned plasmids, BY-2 cell cultures and Arabidopsis seeds used in the study, will be made available from the corresponding author (KRů) upon request. Sequences of the primers used in the study are provided in Table S1. All protocols and methods used in the study are described in the Materials and Methods section and Methods S1. All programming code used in the study is available online on GitHub.com (https://github.com/lamewarden/RaPiD-boxes-software and https://github.com/lamewarden/FRAP-normalization-script).