Identification of canonical tobacco PIN genes, their phylogeny and cloning

To identify the genes encoding PIN proteins in tobacco genome, sequences of A. thaliana canonical PIN genes (AtPIN1, AtPIN2, AtPIN3, AtPIN4 and AtPIN7) were used as a query for tblastx (NCBI) against tobacco methyl‐filtered database (Opperman et al., 2003). Based on these alignments, full‐length NtPIN genomic and coding sequences were identified. A majority of corresponding sequences were cloned from Nicotiana tabacum L. cv. Bright Yellow 2 plants and cultured cells, and deposited in NCBI GenBank (Table S1). Phylogenetic analysis of tobacco canonical PIN sequences (Figures 1 and S1 and Table S3) showed that they cluster together with other Solanaceae PIN homologues, forming three major groups, PIN1‐type (NtPIN1, NtPIN11), PIN2‐type (NtPIN2) and PIN3‐type (NtPIN3a, NtPIN3b). Nicotiana tabacum (Nt) genome contains always two closely related copies of each PIN gene, which represent forms originating from its two ancestral species Nicotiana sylvestris (Nsyl) and Nicotiana tomentosiformis (Ntom) (Sierro et al., 2014). Based on the alignment of Nt, Nsyl and Ntom sequences (Figure S2), we named individual NtPINs according to their Nsyl and Ntom homologues, as exemplified here for NtPIN1, i.e. NtPIN1S (Nsyl descendant) and NtPIN1T (Ntom descendant).

Our successful cloning of coding sequences from cDNA suggested that all canonical NtPINs are transcribed, either in cell culture or plant tissues (Table S1). Therefore, we have decided to test their transcriptions in detail, both in planta and in cultured cells.

Tobacco canonical PINs are transcribed in dividing BY‐2 cells and in planta

To determine which canonical NtPINs are expressed in cultured cells, qRT‐PCR‐based expression profiling for NtPIN1, NtPIN11, NtPIN2, NtPIN3a and NtPIN3b was performed in suspension‐grown cells cultured for 2 days, leaves, stems and roots of seedlings grown in vitro for 3 weeks, and anthers and ovaries collected from mature plants grown in greenhouse. As shown in Figure 2, NtPINs were all transcribed in the exponentially grown BY‐2 cells, except NtPIN1. NtPIN11 and NtPIN1 were both transcribed in vegetative and generative tissues, although NtPIN1 showed the highest transcript level in anthers in contrast to NtPIN11, which showed there even lower transcript abundance compared with the cell culture (Figure 2, upper two panels). NtPIN2 mRNA was present in exponential BY‐2 cells in levels comparable with roots, while in leaves as well as in the reproductive organs the transcript levels were very low, or even absent (anthers). The opposite results were shown for NtPIN3a and NtPIN3b transcripts, which were again comparable with BY‐2 cells in roots, but were significantly more abundant in all green tissues (Figure 2, lower two panels). To understand whether BY‐2 cells during their life cycle express also noncanonical PINs, i.e. homologues of A. thaliana PIN5, PIN6 and PIN8, qRT‐PCR‐based expression profiling was used with specific primers (Figure S3 and Table S4). Importantly, all of these genes showed no expression in cultured cells, providing the evidence that their protein products are probably not involved in the regulation of the overall auxin homeostasis in suspension‐cultured cells under standard conditions.

Altogether, transcript levels of NtPIN11, NtPIN2, NtPIN3a and NtPIN3b in exponential BY‐2 cells suggest an interplay between several members of the tobacco auxin efflux carrier family during the proliferation of suspension‐cultured cells.

Functional analysis of NtPINs after their inducible overexpression in BY‐2 cells

To test the function of individual tobacco PIN proteins, coding sequences of NtPIN2T (KP143727), NtPIN3bT (KC425459) and NtPIN11T (KC433528) were cloned, introduced into BY‐2 cells and overexpressed under β‐estradiol‐inducible promoter. Resulting BY‐2 cell lines XVE‐NtPIN2, XVE‐NtPIN3b and XVE‐NtPIN11 showed strong increase of NtPIN2, NtPIN3b and NtPIN11 transcripts, as shown by qRT‐PCR (Figure 3a). Auxin accumulation assays, performed after 48 h induction with 1 μm β‐estradiol, showed that the accumulation of good auxin efflux carrier substrate [³H]NAA (Delbarre et al., 1996), decreased significantly (to c. 60%) in the XVE‐NtPIN11 line in six biological repetitions, i.e. after an independent gene inductions (Figure 3b), thus confirming an increase in auxin efflux. This was also accompanied by significantly enhanced cell elongation, a symptom of auxin starvation, scored 4 days after induction (Figure 3c). The induction of NtPIN2 and NtPIN3b in the XVE‐NtPIN2 and XVE‐NtPIN3b lines decreased the accumulation of [³H]NAA in some experiments, but these lines did not display significant difference in auxin efflux (Figure 3b). Also, they had only weak stimulation of cell elongation (Figure 3c). The opposite approach, i.e. inducible silencing of one of the auxin efflux genes, was also employed with NtPIN3b in the XVE‐silNtPIN3b line. The expression of this gene decreased in the XVE‐silNtPIN3b line after 48 h induction with β‐estradiol (Figure 3a). In agreement, [³H]NAA accumulation increased (Figure 3b) and cell elongation was not stimulated at all (Figure 3c). To show that all tested NtPINs were localized at the PM, translational fusion of coding sequences of NtPIN2, NtPIN3b and NtPIN11 with green fluorescent protein (GFP) were used to generate β‐estradiol‐inducible lines XVE‐NtPIN2–GFP, XVE‐NtPIN3b–GFP and XVE‐NtPIN11–GFP. As shown in Figure 3(d), all three proteins were localized within the PM in cells cultured for 2 days with β‐estradiol. No alternative localization patterns were detected (i.e. endoplasmic reticulum (ER), cytosol, etc.) in three independent transformations with 50 independent lines tested for each NtPIN, suggesting that all three NtPINs are correctly positioned at the PM after β‐estradiol‐inducible overexpression.

These results suggest that individual members of tobacco PIN family might be differentially regulated at the post‐transcriptional or post‐translational level, affecting the function, stability or abundance of their protein products. Therefore, we focused in our further analysis on the XVE‐NtPIN11 cells that showed remarkable increase in auxin efflux and phenotypic response.

Experimental data supported by a mathematical model of cellular auxin transport show enhanced auxin influx in XVE‐NtPIN11 cells

To estimate changes in auxin transport parameters after the induction of NtPIN11 in XVE‐NtPIN11 cells, a specific series of [³H]2,4‐D accumulation assays were performed in both β‐estradiol‐induced cells and dimethyl sulfoxide (DMSO)‐treated controls according to our previous report (Hošek et al., 2012). Mathematical model of auxin transport (Appendix S1) was then fitted into the empirical data by optimizing its parameter values. After that, the assays with both non‐induced and induced cells were successfully reproduced by the solution of the mathematical model of 2,4‐D transport (Figure 4; Figure S4). Parameter estimates obtained by the optimisation of the model using the data from non‐induced and induced XVE‐NtPIN11 cells are shown in Figure S4 and their values stated in Appendix S2. While 2,4‐D diffusion (i.e. pI and kD) showed only minimal changes after induction of NtPIN11 (Appendix S2), the induction of NtPIN11 generated 50% increase of carrier‐mediated auxin efflux rate constant, but there was also even three‐fold increase of 2,4‐D carrier‐mediated auxin influx rate detected (Figure 4).

This empirically‐supported increase in carrier‐mediated auxin influx after the inducible overexpression of auxin efflux carrier NtPIN11 was rather unexpected, suggesting the existence of an auxin transport‐based compensatory mechanism balancing the intracellular auxin levels. Therefore, we have focused on this in further experimental analysis.

NtPIN11 overexpression results in the increased transcript levels of specific tobacco LAX gene

To understand which auxin influx carriers might be involved in the response of tobacco cells to increased auxin efflux, all tobacco homologues of AUX1/LAX genes have been identified from tobacco genome sequence database (Sierro et al., 2014). Phylogenetic analysis of recent evolution of selected angiosperm AUX1/LAX sequences and their tobacco homologues (Figures 5a, S5 and S6 and Table S5) showed three major groups of AUX1/LAX genes with their independent radiations for Solanaceae and selected angiosperms, i.e. AUX1/LAX, LAX2 and LAX3 group (Figure S5). As found for NtPIN sequences, due to the allotetraploid character of Nicotiana tabacum (Nt) genome, NtLAX genes are present in two copies with high similarity, each representing the form originating from one of the two ancestral species Nsyl and Ntom (Sierro et al., 2014). Based on the alignment of Nt, Nsyl and Ntom sequences (Figure S6), individual tobacco homologues were named according to their Nsyl and Ntom homologues.

Specific primers were designed for qRT‐PCR analysis of NtLAX genes in a way similar to that for NtPINs. Transcript levels of individual genes were compared in suspension cells grown for 2 days, leaves, stems and roots of seedlings grown in vitro for 3 weeks and anthers and ovaries collected from mature plants grown in greenhouse, i.e. in a setup similar to the comparison of quantification of NtPINs transcripts (Figure 2). As shown in Figure 5(b), transcripts of NtLAXes were present in all vegetative and generative tissues as well as in exponentially grown BY‐2 cells, with NtLAX1c, NtLAX2a and NtLAX2b transcripts being very low or absent in BY‐2 cells. In general, BY‐2 cells transcribed significantly less NtLAX1a, NtLAX1c, NtLAX2a and NtLAX2b, and more NtLAX1b (with the exception of leaves, where the mRNA levels were comparable with NtLAX1b and NtLAX2b). Interestingly, in contrast with the transcript levels of NtPINs, which had transcript levels comparable with root cells, for NtLAXes, there was just NtLAX3 expressed in comparable levels in cell culture and vegetative tissues.

To test whether NtLAXes might be involved in the above‐mentioned auxin transport‐based compensatory mechanism balancing intracellular auxin levels, we have followed their transcript levels after inducible overexpression of NtPIN11 in XVE‐PIN11 cells. These experiments showed two important facts. Firstly, there was just one NtLAX with significantly increased transcript abundance, NtLAX1c, the member of the NtLAX family that is not transcribed in wild type BY‐2 cells. Secondly, after the induction of NtPIN11, the transcript level of all other endogenous NtPINs decreased (Figure 5c), suggesting that the compensatory mechanism reacting to increased auxin efflux might involve both reduction of endogenous auxin efflux and increased auxin influx. These results suggested the existence of mechanisms driving auxin influx and efflux rates at the cellular level, which could be tuned by intracellular auxin levels. Therefore, we decided to test whether the removal of auxin from culture media can induce similar changes.

Auxin starvation increases transcript levels of NtLAX1c and decreases transcript levels of NtPINs

To test which NtPINs and NtLAXes are expressed under auxin starvation conditions, wild type BY‐2 cells were cultured for 48 h from the beginning of the subculture interval in auxin‐free medium, i.e. without 2,4‐D, following the conditions used previously for auxin starvation experiments (Sakai et al., 2004). After 48 h, the transcript level of just one member of NtPIN family (NtPIN3b) did not decrease and all other NtPINs had remarkably lower levels of their mRNAs (Figure 5d). In contrast, two NtLAXes (NtLAX1b and NtLAX3) decreased their transcript abundances; three of these showed no changes (NtLAX1a, NtLAX2a and NtLAX2b) and one (NtLAX1c) showed strong transcript accumulation (Figure 5d). These results suggest that auxin starvation might trigger mechanisms similar to those observed in NtPIN11 overexpressing cells. To test this hypothesis, individual auxin transport parameters of auxin‐starved cells were measured by means of auxin accumulation assays with mathematical modelling.

Experimental data supported by mathematical model of cellular auxin transport in tobacco cells grown in auxin‐free medium show decreased auxin efflux

To estimate the changes in auxin transport parameters in cells cultured in control and auxin‐free conditions, a specific series of [³H]2,4‐D accumulation assays were performed according to the protocol used for XVE‐NtPIN11 cells. Mathematical model of auxin transport was successfully fitted into the empirical data (Appendix S1, Figure 6a; Figure S7), thus providing transport parameter estimates for control and auxin‐starved cells (Appendix S3). As shown in Figure 6(a), decreased carrier‐mediated auxin efflux was detected, but, in contrast with NtPIN11 overexpression, there was no increase of 2,4‐D carrier‐mediated auxin influx. These results suggested that the transcription‐based compensatory mechanism, decreasing the expression of NtPINs and stimulating the expression of NtLAX1c, was clearly manifested in auxin‐starving cells. However, under these conditions, this mechanism was not able to increase carrier‐driven auxin influx in our radiolabelled 2,4‐D assays.

Identification of canonical tobacco PIN genes and their role in BY‐2 cells

Phylogenetic analysis of PIN gene family evolution (Bennett et al., 2014; Bennett, 2015), which was accomplished by aligning 473 sequences from 109 species, significantly improved our understanding of PIN radiation in the green plant lineage. According to this study, there are three distinct lineages of PIN proteins among bryophytes, lycophytes and euphyllophytes, and each lineage progresses through several independent radiations, giving rise to at least 21 clades. This radiation is apparent in the euphyllophyte‐specific lineage Eu‐3 that contains radiations of canonical PIN1‐, PIN2‐ and PIN3‐type sequences. Interestingly, while there are three members of the PIN3 subclade in A. thaliana (AtPIN3, AtPIN4, and AtPIN7), the tobacco genome only contains two independently derived genes (NtPIN3a and NtPIN3b), which are each present in both the Nsyl and Ntom versions. Conversely, the tobacco genome contains homologues from the NtPIN1 and NtPIN11 (SoPIN1) subclades (O'Connor et al., 2014; Bennett, 2015). qRT‐PCR transcription profiling showed that all NtPINs, except for NtPIN1, were transcribed in exponentially grown BY‐2 cells. The absence of NtPIN1 expression in cultured cells supports the notion of sub‐functionalization of PIN1 and PIN11/SoPIN1 in monocots (O'Connor et al., 2017) in terms of the preferential usage of NtPIN11 in proliferating tobacco cells. Similar sub‐functionalization has been previously reported in Solanaceae for the Solanum lycopersicum homologue of SoPIN1 that specifically acts in the epidermis of leaves and generates auxin transport convergence points (Martinez et al., 2016). In general, the transcript levels of NtPIN11, NtPIN2, NtPIN3a and NtPIN3b in exponentially grown BY‐2 cells are comparable with their levels in roots, from which the BY‐2 cell line was previously established (Kato et al., 1972).

One of the main questions that arose after we identified individual members of the NtPIN gene family was whether the protein products of all members of the family are functional, i.e. whether increased auxin efflux can be detected after their inducible overexpression. To address this question, we performed auxin transport assays using synthetic radioactive‐labelled auxin [³H]NAA, which is a very good substrate for auxin efflux carriers (Delbarre et al., 1996; Petrášek et al., 2006). This assay was optimized for the detection of the overall auxin accumulation in cells. Therefore, increased auxin efflux was indicated by a decreased [³H]NAA pool in the cells. Although the transcript levels of all NtPINs were elevated in their specific transgenic lines after β‐estradiol induction, only the XVE‐NtPIN11 line showed repeatedly lower accumulation of [³H]NAA. This discrepancy could be explained by post‐translational regulation of the transport activity of PIN proteins or by different localization of the carrier on PMs between individual cells within cell chains. This has been previously indicated by enhanced auxin efflux in an apolar‐localized ubiquitylation mutant version of AtPIN2 in BY‐2 cells (Leitner et al., 2012). As we showed in this study, when tagged with GFP, all investigated NtPINs correctly localized to the PM. Therefore, based on the fact that there was no alternative localization, we suggest that non‐tagged versions also be localized to the PM. Since there is a lack of antibodies against NtPINs, we could not provide more direct immunological evidence on the presence of the investigated NtPINs in the PM as we have previously shown for heterologously expressed PINs from A. thaliana (Petrášek et al., 2006). Overall, although all three investigated NtPINs induced hallmarks of auxin starvation (Miyazawa et al., 1999), to identify the auxin homeostasis‐driven mechanism, we further analyzed NtPIN11, which showed the most notable increases in auxin efflux and cell elongation.

Enhanced carrier‐mediated auxin efflux stimulates the expression of specific auxin influx carrier genes

Surprisingly, the application of a mathematical model to the [³H]2,4‐D accumulation kinetics (Hošek et al., 2012) in XVE‐NtPIN11 cells estimated a significant increase in both carrier‐mediated auxin efflux and auxin influx. Importantly, 2,4‐D diffusion (i.e. pI and kD) only showed minimal changes after the induction of NtPIN11, providing good evidence for the validity of the obtained model parameter estimates.

Our results suggest the existence of an auxin homeostasis‐based mechanism for the maintenance of optimal endogenous auxin concentrations in situations involving excess auxin efflux. A similar compensatory mechanism has been previously shown at the tissue level in A. thaliana (Vieten et al., 2005), where the ectopic expression of individual members of the PIN protein family was observed after loss of a specific PIN protein. In our study, using the opposite approach, i.e. strong inducible overexpression of one of the auxin efflux carriers, we detected enhanced auxin flow at the cellular level. This possibly triggered changes in endogenous auxin levels with subsequent differential transcription of both auxin efflux and influx carriers. To understand this regulation, members of the tobacco AUX1/LAX gene family were identified, and their transcripts were quantified in cultured tobacco cells and plants as well as after the inducible expression of NtPIN11. In agreement with a recent phylogenetic study (Singh et al., 2018), there were three major groups of AUX1/LAX genes identified in Solanaceae, i.e. the AUX1/LAX, LAX2 and LAX3 groups. qRT‐PCR transcription profiling of all NtLAXes after inducible expression of NtPIN11 in BY‐2 cells showed that there was only one member of the NtLAX gene family with increased transcript abundance, which was NtLAX1c. Importantly, this is a gene that has not been transcribed in wild type BY‐2 cells. In these cells, NtLAX1a, NtLAX1b and NtLAX3 were preferentially transcribed. The differential transcript abundances of individual members of the NtLAX family that were shown in our experiments in planta and in BY‐2 cells supports a previous report on the sub‐functionalization of the A. thaliana AUX1/LAX family (Péret et al., 2012). Moreover, the preferential increase in the NtLAX1c transcript levels in cells with strongly enhanced auxin efflux is also in good agreement with results reported by Péret et al. (2012), who showed that the expression of not all members of the A. thaliana AUX1/LAX family is altered in the presence of auxin. The compensatory mechanism that was triggered upon the stimulation of auxin efflux by overexpressing a specific auxin efflux carrier (NtPIN11) also included an effective decrease in the transcription levels of all other members of the NtPIN family. This is consistent with the fact that the transcript levels of all canonical PIN genes have been previously shown to be stimulated by auxin (Paponov et al., 2008). Based on our results, we cannot exclude the option that overexpressed NtPIN11 might serve as an auxin influx carrier itself. In theory, this activity might resemble auxin uptake activities of ABCB4 (Kubeš et al., 2012) and ABCB21 (Kamimoto et al., 2012) at low intracellular auxin concentrations.

Auxin starvation‐triggered changes include increased transcript levels of a specific auxin influx carrier and decreased transcript levels of a majority of auxin efflux carriers

As BY‐2 cells displayed hallmarks of auxin starvation upon the induction of the NtPIN11 auxin efflux carrier, including substantially enhanced auxin efflux and significantly decreased transcript levels of all endogenous NtPINs, we further investigated whether the removal of auxin from the culture medium caused similar changes. When the cells were incubated for 48 h in auxin‐free medium, a significant increase in NtLAX1c transcript levels and a decrease in the majority of NtPIN transcript levels were observed. Interestingly, the transcript levels of a gene coding for an auxin efflux carrier and three genes coding for auxin influx carriers remained unchanged. In addition, auxin transport assays combined with a mathematical modelling approach clearly showed that the transcription‐based auxin compensatory mechanism was stimulated, and auxin efflux decreased. However, auxin influx rates did not change, possibly because a low amount of endogenous auxin IAA (Nakajima et al., 1979; Shimizu et al., 2006) was not enough to stimulate sufficient expression of other auxin influx carriers. Moreover, the lack of exogenous auxin during the starvation period might result in low PM abundance or instability of the NtLAX1c protein itself (Figure 6b).

Although the results that were primarily obtained from the experimental model of tobacco BY‐2 cells have obvious interpretation limitations, they could greatly help to understand the homeostatic mechanisms of auxin at the cellular level, as we have shown in our previous studies. As shown in a schematic (Figure 6b), in this study, we revealed an auxin‐driven transcription‐based compensatory mechanism that decreases the expression of NtPINs and stimulates the expression of NtLAX1c in the situation of misbalanced auxin levels.

Our results could help to understand previously suggested auxin homeostatic mechanisms at the tissue and organ levels in A. thaliana. First, the results are consistent with the fact that the compensatory mechanism for tuning endogenous auxin levels is based on auxin transport because transport is approximately one order of magnitude faster than biosynthesis, as calculated and shown in numerous developmental contexts by Kramer and Ackelsberg (2015). BY‐2 cells resemble auxin‐recycling cells in the root meristem, where the interplay between individual PINs within the auxin reflux loop (Blilou et al., 2005) and between individual AUX/LAXes (Péret et al., 2012) occurs. The low biosynthesis of auxin in BY‐2 cells (Shimizu et al., 2006) is most likely counterbalanced by auxin influx and efflux, which provide enough auxin to trigger specific gene expression. Second, our results support the notion of the collaboration between AUX1/LAX auxin influx carriers and PIN auxin efflux carriers in the maintenance of auxin levels, as shown in lateral root cap cells (Band et al., 2014) and during embryogenesis (Robert et al., 2015). Third, stimulated expression of a specific auxin influx gene (NtLAX1c) caused by stimulated expression of an auxin efflux carrier gene (NtPIN11) might be explained by changes in endogenous auxin levels, as previously shown for auxin‐induced expression of LAX3 auxin influx carrier during lateral root emergence (Swarup et al., 2008). LAX3 is under the control of ARF7 through the transcription factor LBD29 and is preceded by the coordinated expression of the PIN3 auxin efflux carrier, which serves as the primary auxin‐regulated gene (Porco et al., 2016b). As previously suggested by the modelling approach (Mellor et al., 2015), this mechanism helps to confine the maximum amount of auxin to a restricted area during lateral root emergence or during the early phases of nodulation in legumes (Roy et al., 2017), in which the expression of only one AUX1 homologue has been reported.

Finally, the results presented in this study using tobacco BY‐2 cells as a model support the hypothesis that a feedback mechanism drives the transcription of genes for individual auxin efflux and influx carriers, and auxin‐triggered gene expression occurs through the Aux/IAA pathway. When one potent auxin efflux carrier is induced (NtPIN11), symptoms of auxin starvation can be observed despite the presence of readily available auxin in the medium. This indicates that auxin‐stimulated gene expression is decreased. Cells are seemingly unable to properly counterbalance this extensive auxin efflux, and the expression of their endogenous auxin efflux carriers is decreased. However, even under these conditions, there is one auxin influx carrier that has the opposite reaction (NtLAX1c). We expect that within the promoter of this gene, some upstream elements might regulate the transcription of this gene so that it reacts in the opposite manner than the majority of other transporters. In the future, we aim to understand what factors drive the expression of this gene and conduct in planta studies.

Plant material and chemicals

Cells of tobacco cell line BY‐2 (Nicotiana tabacum L. cv. Bright Yellow 2) were maintained in liquid Murashige and Skoog (MS) medium (3% sucrose, 4.3 g L⁻¹ MS salts, 100 mg L⁻¹ myo‐inositol, 1 mg L⁻¹ thiamin, 0.2 mg L⁻¹ 2,4‐dichlorophenoxyacetic acid (2,4‐D), and 200 mg L⁻¹ KH₂PO₄; pH 5.8), in the dark, at 27°C, under continuous shaking (150 rpm; orbital diameter 30 mm) and subcultured every 7 days. Transgenic cells and calli were maintained on the same medium supplemented with 40 μg mL⁻¹ hygromycin and 100 μg mL⁻¹ cefotaxime. Plants of Nicotiana tabacum L., cv. Bright Yellow 2 were obtained from collections of the botanical garden of Faculty of Science, Charles University, Prague and propagated in greenhouse. For in vitro growth, seeds of BY‐2 were surface sterilized with 50% (v/v) household bleach (final concentration c. 2.5% sodium hypochlorite) and grown on vertical plates with ½MS supplemented with 1% (w/v) agar and 1% (w/v) sucrose, under long days conditions (16 h light/8 h dark at 25°C/18°C) for 3 weeks.

Unless otherwise stated, all chemicals and kits were obtained from Sigma‐Aldrich Inc. (St. Louis, MO, USA).

Phylogenetic analysis

PIN‐FORMED and AUX1/LAX genomic sequences were obtained from NCBI databases of fully‐sequenced genomes by tblastn algorithm. Exon−intron structure was defined based on homology with annotated A. thaliana genes. Exon regions were aligned at amino acid level using MAFFT algorithm. PIN‐FORMED family phylogenetic analysis was based on amino acid positions with coverage higher than 80% and was conducted in MEGA6 software (Tamura et al., 2013) by Maximum Likelihood approach using JTT+G model (Gamma shape parameter = 1.67485, number of substitution rate categories = 5). Maximum Likelihood phylogenetic analysis of AUX1/LAX family was computed by phyml software (Guindon and Gascuel, 2003) from all nucleotide position using GTR+G+I model (proportion of invariant sites = 0.3710, Gamma shape parameter = 1.0340, number of substitution rate categories = 4). Topologies were tested by bootstrap analysis with 500 replications.

Isolation of DNA and RNA

Genomic DNA and total RNA were isolated using NucleoSpin Plant II kit (Macherey‐Nagel, Düren, Germany) and Spectrum™ Plant Total RNA kit (Sigma‐Aldrich Inc., St. Louis, MO, USA), respectively, from approximately 100 mg of tobacco BY‐2 suspension‐grown cells harvested 48 h after inoculation, from tobacco BY‐2 seedlings grown in vitro for 3 weeks (roots, stems and leaves separately) and from mature BY‐2 plants from the greenhouse (anthers and ovaries). For experiments with auxin‐free culture medium, BY‐2 stationary cells culture for 7 days were inoculated to medium lacking 2,4‐D and harvested after 24 and 48 h. 1 μg of DNase‐treated RNA (DNA‐Free kit, Thermo Fisher Scientific, Vilnius, Lithuania) was reverse transcribed using M‐MLV reverse transcriptase, RNase H‐, point mutant (Promega, Madison, USA).

Cloning of tobacco PIN genes, preparation of gene constructs and transformation of BY‐2 cells

Coding sequences of A. thaliana PIN genes were blasted against Nicotiana tabacum methylation filtered genome TGI: v.1 (Opperman et al., 2003) using tblastx. Primers for PCR amplification of full‐length genomic and coding sequences of tobacco PIN genes were designed (Table S2) based on homologous fragments. For every PIN homologue, PCR amplification was done using Phusion polymerase (Thermo Fisher Scientific) at T_a ranging from 58 to 62°C, depending on primer pair. The resulting fragments were cloned into pJET 1.2 vector (CloneJet kit, Thermo Fisher Scientific) and sequenced (GATC Biotech, Constance, Germany).

GFP‐tagged NtPINs were generated by insertion of GFP coding sequence into the cDNA of NtPINs using overlap extension PCR (Horton et al., 1989). Positioning of GFP insertion with the coding region of corresponding NtPIN was designed according to GFP‐tagging of Arabidopsis PIN genes, i.e. at positions 1212 bp from ATG for NtPIN2 (Xu and Scheres, 2005), 1197 bp from ATG for NtPIN3 (Žádníková et al., 2010) and 1269 bp from ATG for NtPIN11 (Benková et al., 2003). NtPINs–GFP were cloned into β‐estradiol‐inducible pER8 plant expression vector. Full‐length coding sequences for NtPIN2, NtPIN3b and NtPIN11 as well as their GFP‐tagged versions were cloned into pER8 plant expression vector (GenBank accession number AF309825.2, Zuo et al., 2000) to generate β‐estradiol‐inducible gene constructs, i.e. XVE‐NtPIN2 (NtPIN2T, GenBank accession number KP143727), XVE‐NtPIN3b (NtPIN3bT, GenBank accession number KC425459), and XVE‐NtPIN11 (NtPIN11T, GenBank accession number KC433528), XVE‐silNtPIN3b (hairpin RNA gene construct), XVE‐NtPIN2–GFP, XVE‐NtPIN3b–GFP and XVE‐NtPIN11–GFP.

BY‐2 cells cultured for 3 days were transformed with gene constructs mentioned above by co‐cultivation with Agrobacterium tumefaciens strain GV2260 as described earlier (An et al., 1985; Petrášek et al., 2003). Transgenic lines were harvested after 4 weeks, cultured on solid media with hygromycin and suspension cultures established from 10 independent lines that were tested by qRT‐PCR for the expression of the transgene. For the experiment, lines with comparable and stable expression were used.

qRT‐PCR expression analysis

As most genes are present in tobacco genome in two very similar copies inherited from two ancestors (Nicotiana sylvestris and Nicotiana tomentosiformis), primers for quantitative real‐time reverse transcription‐PCR (qRT‐PCR) were designed to always amplify both copies of studied PIN and LAX genes (Table S4). qRT‐PCR primers for NtLAX genes were designed according to the sequenced Nicotiana tabacum genome (Sierro et al., 2014). Quantitative real‐time PCR was performed using GoTaq qPCR Master Mix (Promega) at 58°C T_a on LighCycler480 instrument (Roche, Mannheim, Germany). PCR efficiency was estimated using serial dilution of template cDNA. Tobacco EF1a (GenBank accession number AF120093) was used as a reference for relative quantification. Relative transcript levels were calculated using equation:

where eff_ref and eff_target stand for qPCR efficiency of reference and targets genes, respectively, and CP_ref and CP_target stand for crossing point of reference and target genes. Positive transcript levels and the quality of PCR products were verified by melting curve analysis. Three biological replicates with two technical repetitions for each were performed. qRT‐PCR analysis in transformed cells was performed after induction of gene expression with 1 μm β‐estradiol for 24 or 48 h from the beginning of the subculture interval. An appropriate amount of DMSO was added into non‐induced cells.

Light microscopy and cell length measurements

Light microscopy was performed using Eclipse E600 microscope (Nikon, Tokyo, Japan) equipped with a colour digital camera 1310C (DVC, Austin, TX, USA) and NIS elements image analysis software (Laboratory Imaging, Prague, Czech Republic). Cell lengths were measured interactively using NIS elements in induced and non‐induced cells cultured for 4 days. 250–300 cells from every sample were analyzed in two independent biological repeats i.e. independent cultivations. Fluorescence microscopy was performed using a Nikon Eclipse Ti inverted microscope (Nikon) equipped with a spinning disc unit (CSU‐X1, Yokogawa, Japan) and water plan apochromatic objective 40× (NA 1.25). Fluorescence signals of GFP (excitation 488 nm, emission 495–555 nm) were recorded with Andor Zyla sCMOS camera (Andor Technology, Belfast, Northern Ireland) using Semrock Brightline single‐pass filters (Semrock, Rochester, NY, USA) for GFP.

Auxin accumulation assays

Radioactively labelled auxin accumulation assays were performed with [³H]NAA (naphthalene‐1‐acetic acid; 20 Ci/mmol; American Radiolabelled Chemicals, Inc., St. Louis, MO, USA) according to the protocol described previously (Delbarre et al., 1996; Petrášek et al., 2006). 1 μm β‐estradiol‐induced and non‐induced cells grown for 2 days were filtered using a 20‐μm nylon mesh, resuspended in uptake buffer (20 mm MES, 40 mm sucrose, 0.5 mm CaSO₄; pH 5.7) and equilibrated for 45 min under continuous shaking. After another washing step in fresh uptake buffer, cell densities were adjusted to around 7 × 10⁵ cells mL⁻¹ after counting the cells in Fuchs−Rosenthal haemocytometer and cells were incubated for additional 90 min in the dark at 25°C. Accumulation assays were performed with [³H]NAA added to a final concentration of 2 nm. 0.5 mL aliquots were collected by filtration under reduced pressure using cellulose filters. Filters with cell cakes were transferred to scintillation vials and extracted for 30 min with 96% (v/v) ethanol. After addition of scintillation cocktail for liquid samples (EcoLite™, MP Biomedicals, Santa Ana, CA, USA) and 30 min rigorous shaking, radioactivity was determined by liquid scintillation counting with automatic correction for quenching (Packard Tri‐Carb 2900TR scintillation counter; Packard Instrument, Fallbrooke, CA, USA). DPM values were converted to pmols of [³H]NAA per 1 million cells and expressed as the ratio between β‐estradiol‐induced cells and DMSO‐treated non‐induced controls at the time 10 min after the addition of [³H]NAA. Presented results are from six biological repetitions (independent β‐estradiol inductions in the particular line within 3–5 months after the establishment of the cell suspension).

To allow for the estimation of individual auxin transport parameters using a mathematical model, a modified protocol for auxin accumulation was used in β‐estradiol‐induced and DMSO‐treated non‐induced XVE‐NtPIN11 cells and in control cells grown in complete medium or in auxin‐free medium for 2 days. [³H]2,4‐D (20 Ci/mmol; American Radiolabelled Chemicals, Inc.) and specific inhibitors of auxin influx and efflux, 3‐chloro‐4‐hydroxyphenylacetic acid (CHPAA) and N‐1‐naphthylphthalamic acid (NPA), respectively, were used to trace auxin transport in modified accumulation assays, according to Hošek et al. (2012). Each assay consisted of a 10 min accumulation phase (sampled each 30 sec) followed by a wash‐out phase, where the tracer‐fed cells were transported into tracer‐free media and the subsequent decrease of intracellular tracer concentration was monitored (also sampled each 30 sec). Suspension density was decreased eight‐fold in the transition to the wash‐out phase in order to provide sufficient pool to dissolve high amounts of tracer released from the cells. The assays were performed using [³H]2,4‐D (2 nm) with CHPAA (10 μm) treatment (in two technical repeats), [³H]2,4‐D (2 nm) with NPA (10 μm) treatment (in two technical repeats) and control [³H]2,4‐D (2 nm) accumulation in untreated cells (one technical repeat). All these combinations were measured in β‐estradiol‐induced cells, DMSO‐treated non‐induced controls as well as in cells incubated in auxin‐free medium and their controls (standard 2,4‐D containing medium). Using these data the parameters of a mathematical model of 2,4‐D transport and metabolism were independently optimized for both induced and non‐induced samples, thus obtaining estimates of physiological auxin transport parameters for subsequent comparison.

Mathematical modelling of NtPIN11 cellular auxin transport

The mathematical model used in this study is based on the previous report (Hošek et al., 2012), with two modifications: Firstly, metabolism of accumulated [³H]2,4‐D resulting in the formation of a non‐transportable metabolite was introduced in the model using linear kinetics. Secondly, representation of cell surface contamination with the tracer from the media was changed from a constant to be now linearly proportional to the tracer concentration in the medium. The model was implemented in Simulink (graphical modelling and simulation module of Matlab, The MathWorks Inc., Natick, MA, USA), where it was numerically solved, and optimized in Matlab using the same approach as in Hošek et al. (2012). Overview of the differential equations of the model is provided in Appendix S1.

Statistics

Standard nonparametric tests were performed, details are specified in figure captions. The following software was used: Matlab (2017a, The MathWorks, Inc., Natick, MA, USA), Statistica (v12 Cz, Statsoft, www.statsoft.com), and BoxPlotR: a web tool for generation of box plots (Spitzer et al., 2014).