Developmental roles of Auxin Binding Protein 1 in Arabidopsis thaliana

Auxin is a major plant growth regulator, but current models on auxin perception and signaling cannot explain the whole plethora of auxin effects, in particular those associated with rapid responses. A possible candidate for a component of additional auxin perception mechanisms is the AUXIN BINDING PROTEIN 1 (ABP1), whose function in planta remains unclear. Here we combined expression analysis with gainand loss-of-function approaches to analyze the role of ABP1 in plant development. ABP1 shows a broad expression largely overlapping with, but not regulated by, transcriptional auxin response activity. Furthermore, ABP1 activity is not essential for the transcriptional auxin signaling. Genetic in planta analysis revealed that abp1 loss-of-function mutants show largely normal development with minor defects in bolting. On the other hand, ABP1 gain-of-function alleles show a broad range of growth and developmental defects, including root and hypocotyl growth and bending, lateral root and leaf development, bolting, as well as response to heat stress. At the cellular level, ABP1 gain-of-function leads to impaired auxin effect on PIN polar distribution and affects BFA-sensitive PIN intracellular aggregation. The gain-of-function analysis suggests a broad, but still mechanistically unclear involvement of ABP1 in plant development, possibly masked in abp1 loss-of-function mutants by a functional redundancy.


Introduction
The phytohormone auxin is a major coordinator of plant growth that governs a multitude of developmental processes. Its versatility is related to its differential distribution within plant tissues and the ability of cellular auxin concentrations determine various cell fate decisions. The establishment of these morphogenic auxin gradients and local auxin maxima is achieved by a combination of local auxin biosynthesis [1] and synergetic, directional cell-to-cell polar auxin transport [2].
Auxin concentration affects cellular processes, mainly through a modulation of transcription. A broad range of auxin-responsive transcriptional regulators remodel the transcriptome of cells through tissue specific expression and thus trigger complex developmental changes [3]. On this transcriptional level, auxin controls processes such as embryogenesis, vascular tissues formation and organogenesis of the shoot apex or maintenance of the root apical meristem [4].
Nevertheless, some cellular auxin effects occur too fast to be a result of transcriptome remodeling and/or they were shown not to require functional transcription or de novo protein synthesis. Auxin triggers rapid hyperpolarization of the plasma membrane leading to protoplast swelling [5,6], induces calcium ion and proton fluxes across the plasma membrane and therefore alkalizes the apoplast [7,8], and inhibits clathrin-mediated endocytic trafficking processes [9,10].
The auxin signal is transduced via several mechanisms [11,12]. The canonical pathway is mediated by a nuclear-localized co-receptor complex comprising the TRANSPORT INHIBITOR RESISTANT1/AUXIN SIGNALING F-BOX (TIR1/AFB) F-box proteins and the AUXIN/INDOLE-3-ACETIC ACID (Aux/IAA) transcriptional repressors. Here, auxin promotes the interaction of TIR1/AFBs with Aux/IAAs that results in ubiquitin-dependent degradation of the Aux/IAA proteins. Aux/IAA proteins act as transcriptional repressors of AUXIN RESPONSE FACTORs (ARFs) transcription factors and thus their degradation activates auxin-responsive transcription [13].
Notably, recent findings suggest that TIR1/AFB signaling mediates both rapid transcriptional as well as even faster non-transcriptional auxin effects on growth. In shoots, auxin via the TIR1/AFB pathway induces fast apoplast acidification and growth promotion by a rapid transcriptional regulation of SMALL AUXIN UPREGULATED (SAUR) genes [14,15]. In contrast, auxin-mediated growth inhibition in roots occurs within 30 s and does not require de novo protein synthesis but is still strictly dependent on the TIR1/AFB pathway [16]. Furthermore, the auxin-mediated fast depolarization of the plasma membrane and Ca 2+ uptake were demonstrably linked with the TIR1/AFB signaling pathway [17].
Recently, two additional non-canonical auxin-sensing mechanisms were described. Auxin has been shown to bind directly to the atypical ARF ARF3/ETTIN to modulate chromatin states and interaction with other transcriptional regulators during gynoecium development [18,19]. The other mechanism involves TRANSMEMBRANE KINASE 1 (TMK1), a member of the plasma membrane-localized TMK receptor-like kinase family [20]. It was shown that auxin triggers cleavage of TMK1's intracellular kinase domain and its consequent translocation to the nucleus. There, the TMK1 kinase domain binds, phosphorylates and thus stabilizes two non-canonical Aux/IAAs, IAA32 and IAA34. Via this alternative transcriptional pathway, auxin regulates apical hook development [21]. TMK1 also regulates lateral root organogenesis and auxin biosynthesis by other cellular mechanisms [22,23]. Importantly, while the canonical TIR1/AFB receptors sense auxin predominantly in the nucleus, the TMKs located in the plasma membrane may perceive auxin from the apoplast by an unknown perception mechanism.
The accumulating developmental roles of TMKs in conjunction with their plasma membrane localization stir up a decades-lasting debate on the existence of a cell-surface auxin receptor. In the past, the best candidate appeared to be AUXIN BINDING PROTEIN 1 (ABP1). This evolutionally conserved 22-kDa glycoprotein [24,25] has been shown to bind auxin at apoplastic pH 5.5 [26][27][28] and although it predominantly localizes to the endoplasmic reticulum (ER), a small fraction has been proposed to reside in the apoplast [29]. ABP1 has been proposed to be mainly associated with rapid non-transcriptional auxin-mediated processes, but the genetic analysis has been hampered by the lack of viable loss-of-function mutants. It has also remained unclear how apoplastic ABP1 could transduce the auxin signal into the cell and therefore the requirement of a plasma membrane-localized docking partner was hypothesized. Later, ABP1 was found to interact with TMK1 in an auxin-dependent manner. It was proposed that ABP1 and TMK1 form an auxin-sensing complex at the plasma membrane that activates downstream cellular processes via small GTPases ROP2 and ROP6 and their effector proteins RIC1 and RIC4 [30][31][32]. Based on the phenotypes of ABP1 gain-of-function mutants, the weak abp1-5 allele and conditional abp1 knock-down lines [33] ABP1 was proposed to play a role in clathrin-mediated endocytosis [10,34,35], growth-correlating microtubule re-orientation [36], cell wall remodeling [37] or interdigitated growth of leaf pavement cells [30,36]. All these proposed roles were called into question by the failure to complement the alleged embryo lethal abp1 phenotypes, by the coding sequence of ABP1 [35] and by the identification of new abp1 knock-out alleles with no obvious morphological phenotypes [38]. These discrepancies were clarified by proofs that the original abp1 embryo lethal phenotypes were caused by disruption of a neighboring gene rather than ABP1 itself [39,40]. Furthermore, the abp1-5 line carries many additional mutations [41] and the conditional knock-down lines, despite independently targeting either ABP1 mRNA or protein [33], also have other targets [42]. Thus, with much of the previously used genetic material called into question and with only superficial phenotype analysis of the more recent, verified knock-out lines [38], the developmental and physiological roles of ABP1 still remain largely unclear.
Here we used the verified gain-and loss-of-function mutant lines in Arabidopsis to (re)evaluate the role of ABP1 in cellular processes, physiological responses and plant development.

Vector construction
All plasmids were constructed by the Gateway cloning technology (www.invitrogen.com). Previously generated constructs pDONR221-ABP1cDNA and pDONR221-ABP1cDNA-M2X containing cDNA sequence of ABP1 [35] were used to construct the final plasmids 35S:: ABP1 and 35S::ABP1-M2X by recombination into the p2GW7 destination vector. ABP1::ABP1 was constructed as follows: the 3.0 kb promoter, genomic coding region and 0.6 kb of 3' untranslated region for ABP1 was amplified and inserted into a pDONR-Zeo vector, then inserted into the pGWB401 destination vector. ABP1::GFP-ABP1 was constructed using a 1585 bp promoter fragment [43] and a N-terminal GFP fusion directly after the N-terminal signal peptide. The GFP insertion was flanked at the 5' end by a PKAPA linker (tested for cleavage using the SignalP-5.0 server) and at the 3' end by a PKPAPKPA linker. The ABP1 fragments were amplified from genomic DNA using primer pairs 1 and 2 (promoter, signal peptide and 5' linker), 3 and 4 (GFP and 3' linker) and 5 and 6 (gABP1 gene body including 3' UTR). All three fragments were fused in a single overlap PCR reaction and cloned into the pDONR221 entry vector. The resulting construct was cloned into the pKGW,0 destination vector and sequenced. All primers used in this study are listed in Supplemental Table 2. All plasmids used in this study are listed in Supplemental Table 3.

Growth conditions
Seeds were chlorine gas sterilized or sterilized with 70 % EtOH, sown on plates with ½ Murashige and Skoog (MS) medium supplemented with 1 % (w/v) sucrose and 0.8 % (w/v) Phytoagar (pH 5.9) and stratified for 2 days at 4 • C. For experiments using Arabidopsis seedlings, the seedlings were grown on plates at 21 • C under a long-day photoperiod ( Plants were always grown together within the growth chamber and with randomized positions within each tray in order to minimize the effect of environmental fluctuations. For etiolated growth, the plated and stratified seeds were exposed to light for 8 h and further covered with aluminum foil to cultivate them in the dark at 21 • C for 4 days (the shoot gravity response experiment) or in the dark chamber at 24 • C for 5 days (the etiolated growth experiment).

Quantitative real-time PCR
After treatment with 100 nM IAA, seedlings were sampled in 4 biological replicates at different times (t0, 5 min, 30 min, 1 h, 2 h and 4 h). Total RNA was prepared from max100 mg of shoots/roots of 5-d-old seedlings with the RNeasy Plant Mini Kit (Qiagen, 74904) according to the manufacturers's instructions. cDNA was synthesized from 2 μg of total RNA using the QuantiNova Reverse Transcription Kit (Qiagen, 205410). For the mutant expression analyses, 3 biological replicates of full seedlings were used. All samples were pipetted in 3 technical replicates in a 384 well plate using an automated JANUS Workstation (PerkinElmer). According to the manufacturer's instructions, 5 μL reaction volume contained 2.5 μL Luna® Universal qPCR mastermix (NEB, M3003S). RT-qPCR analyses were performed using the Real-time PCR Roche Lightcycler 480 and the expression of PP2AA3 (At1G13320) or EF1a (At5G60390) was used as a reference [47]. For ABP1, 5 different primer pairs were evaluated and one representative graph is included in the manuscript. The primers used for the presented analysis are listed in Supplemental Table 2.

Confocal laser scanning microscopy and image analysis
Confocal laser scanning microscopy for analysis of fluorescence intensity was performed on a Zeiss LSM800 microscope assisted with Zeiss Zen 2011 software. Images were acquired with 20x/0.8 NA air (DR5 evaluation) or 40x/1.2 NA water immersion objectives (immunostaining).

DR5-GFP intensity measurement
5-d-old seedlings were transferred from solid ½ MS media to plates supplemented with either DMSO or 1 μM IAA for 3 h and imaged using confocal microscopy. The fluorescence intensity of GFP (excitation wavelength: 488 nm) was measured in ImageJ.

Microfluidic vRootchip
A microfluidic chip, vRootchip was used to analyze root tip growth in real-time. The manufacturing of the chip, sample preparation procedure, and data analysis of root tip growth was performed as described previously [16] and according to Li and Verstraeten et al. (unpublished). vRoot-chip was used with 10 nM IAA treatment in ¼ MS and 0.1 % sucrose. For imaging, the vertical confocal microscopy setup was used as described previously [16,48] and according to Li and Verstraeten et al. (unpublished). The 10 nM IAA solution was supplemented with the cell-impermeable fluorescent dye Tetramethylrhodamine isothiocyanate-Dextran [16].
As a control, protoplasts co-transfected with DR5::LUC and 35S::Renilla were used. The protoplasts were incubated with either 100 nM NAA (Sigma-Aldrich, 86-87-3) for 16 h or without treatment for 12 h followed by 100 nM NAA for 4 h in the dark at room temperature. The corresponding amount of DMSO was used as mock treatment. Chemiluminescence measurement was performed with the Dual-Luciferase Reporter Assay System kit (Promega, E1910) using a Spectrophotometer BioTek SynergyH1 plate reader and Gen 5 software (both BioTek).

Root length analysis
Plates with 4-and 7-d-old seedlings were scanned using an Epson Perfection V370 Photo flatbed scanner and the root length was measured using ImageJ.

Root gravitropic assay
For measurements of root gravitropic curvature kinetics, 4-d-old seedlings were placed on plates with ½ MS and rotated 90 • and roots were imaged using a vertically placed flatbed scanner (Epson Perfection V370 Photo). Multiple plates were held in place on the scanner by a custom-made holder. Max. 12 ROIs of the seedlings were automatically imaged with a resolution of 1200 dpi in 10 min time intervals using an AutoIt script for 8 h. In ImageJ, the time-lapse movies of the seedlings were manually cropped and registered (stabilized) using the Fiji plugin "StackReg" in "Rigid body" mode.

Root growth (RG)-tracker
We developed a custom MATLAB application named RG-tracker (https://research-explorer.app.ist.ac.at/librecat/record/8294) with a graphical user interface that allows entirely automated root growth analysis and tracking of the root tips. Root tips were segmented based on the pixel classification workflow of Ilastik [52], which only requires manual retraining in case the imaging conditions change drastically. For each point in time, the positions of the root tips were determined by segmenting the tip-probability output, performing particle size filtering and calculating the center of mass. The root tips were then tracked over time by solving the linear assignment problem using the Hungarian algorithm (Munkres). The tracking algorithm can deal with gaps in the root tip detection and both the gap closing and the maximum linking distance can be specified in the GUI. Completed tracks are filtered by minimum track length, duration and maximum growth speed to remove miss-detections and then presented to the user as an overlay of raw data, tip segmentation and tip tracks. At this point the user can exclude additional tracks from further analysis and export the overlay of the tracks and the root time-lapse. The x/y coordinates of each root tip, growth speed, direction of growth, growth angle and root length are then calculated for each point in time and exported for further analysis. All experiment specific parameters such as the segmentation threshold, particle size, and track filters can be saved and together with the segmentation project file form the complete data set to clearly recapitulate the data analysis.

Lateral root analysis
For the analysis of lateral root primordia, samples of 6-d-old seedling were collected and the tissue was cleared as previously described [46]. To visualize the lateral root primordia DIC microscopy was performed using an Olympus BX53 microscope equipped with a 20x air objective. The lateral root primordia were staged according to Malam and Benfey (1997) [46]. For analysis of emerged lateral roots, 4-d-old seedlings were transferred from ½ MS plates to plates supplemented with 500 nM NAA or DMSO. After 3 days, the plates were scanned using an Epson Perfection V370 Photo flatbed scanner and the pictures were analyzed using ImageJ.

Etiolated hypocotyl growth
To analyze the growth of etiolated hypocotyls, the seedlings were recorded at 12 h intervals for 120 h in a dark chamber equipped with an infrared light source (880 nm LED; Velleman, Belgium) and a spectrumenhanced camera (EOS035 Canon Rebel Xti, 400DH) with built-in clear wideband-multicoated filter and standard accessories (Canon) and operated by the EOS utility software. The hypocotyl length was measured using ImageJ.

Etiolated hypocotyl bending
To determine hypocotyl gravitropism, the 3-d-old dark grown seedlings were rotated 90 • . The plates were scanned using an Epson Perfection V370 Photo flatbed scanner and the hypocotyl bending angle was measured after gravity stimulation in 6 h intervals for 24 h using ImageJ.

Rosette size analysis
Seeds were germinated and grown on horizontally placed plates for 12 days, scanned using an Epson Perfection V370 Photo flatbed scanner and the rosette size was measured manually in ImageJ.

Vasculature development analysis
10-d-old cotyledons were collected and the tissue was cleared as follows: 2 days incubation in 70 % ethanol with a subsequent incubation in 4 % HCl, 20 % methanol solution at 65 • C for maximum 15 min, followed by an incubation in 7 % NaOH, 60 % ethanol solution at RT for another 15 min. The cotyledons were then re-hydrated in a series of decreasing ethanol concentrations (60 %, 40 %, 20 % and 10 %) for 1 h in each concentration at RT. Before mounting the cotyledons in 50 % glycerol onto microscopy slides they were incubated for 1 h in 25 % glycerol, 5 % ethanol solution at RT. Imaging was done using an Olympus BX53 microscope equipped with a 4x air objective.

Hypocotyl growth under high temperature
Seeds were germinated and grown on ½ MS plates with or without 10 g/L sucrose under 28 • C, continuous light for 7 days. The plates were scanned using an Epson Perfection V370 Photo flatbed scanner and the hypocotyl length was measured using ImageJ.

Hyperosmotic stress assay
4-d-old seedlings were transferred on media supplemented with either 200 mM mannitol or 100 mM NaCl for 4 days. The plates were scanned using an Epson Perfection V370 Photo flatbed scanner and the root extension was measured using ImageJ.

UV laser ablation and periclinal division analysis
3-d-old seedlings were transferred from solid ½ MS medium to plates containing 10 μM propidium iodide (Sigma-Aldrich, 25535-16-4) supplemented with 1 μM NAA or DMSO. The subsequent sample preparation, UV laser ablation, imaging and periclinal cell division analysis was performed as described previously [53].

Bolting time, leaf number and branch number analysis
Seeds were suspended in 0.1 % agarose and spread out in soil. The number of plants, bolted and with the primary inflorescence stem grown 1 cm, was recorded every day. The number of rosette and cauline leaves was counted when the first flower of each plant bloomed. The rosette branch was referred to the branch directly attached to the rosette, while the cauline branch was defined as the branch on the primary stem. The number of cauline branches and rosette branches were counted 21 days after sowing.

Phyllotaxis and internode length measurement
Analyses of 25 plants per genotype were performed when the last flowers had appeared. Angles and internode lengths between two subsequent siliques were measured starting from the lowest one. For each individual of each genotype, the variance of the divergence angles was computed, and individual variances of divergence angles were compared between genotypes using a non-parametric Kruskal-Wallis test in R (version 3.5.1, r-project.org), since their values were not normally distributed.

PIN lateralization
3to 4-d-old seedlings were treated either with 10 μM NAA or DMSO as a control for 4 h in liquid ½ MS medium. Subsequently, immunostaining using PIN1 and PIN2 antibodies was performed. Samples were imaged using confocal microscopy. The fluorescence intensity of Cy3 (excitation wavelength: 548 nm) was measured using ImageJ.

BFA treatment
4-d-old seedlings were incubated in liquid ½ MS medium at a final concentration of 25 μM BFA (Sigma-Aldrich, 20350-15-6) for 1 h. For BFA/NAA treatment the seedlings were pre-treated with 5 μM NAA for 30 min followed by co-treatment with 25 μM BFA and 5 μM NAA for 1 h.
As control, seedlings were incubated in liquid ½ MS medium supplemented with DMSO substituting NAA. Subsequently, immunostaining using PIN1 and PIN2 antibodies was performed. Samples were imaged using confocal microscopy and the fluorescence signal of Cy3 (excitation wavelength: 548 nm) was detected. BFA body formation was scored from 0 (no BFA body formation) to 3 (maximal BFA body formation) for each image, reflecting both the number of cells with BFA bodies as well as size and number of BFA bodies per cell. To avoid cognitive bias, all images were encoded prior to analysis.

Global transcriptome data analysis
Tissue-specific expression pattern and expression following different perturbations were obtained using Genevestigator (www.genevesti gator.com) and were based on the 'AT_AFFY_ATH1-0' dataset.

Accession numbers
Sequences data from this article can be found in the GenBank/EMBL libraries under the following accession numbers: ABP1 (AT4G02980); PP2AA3 (At1G13320); EF1a (At5G60390).

ABP1 expression and regulation by auxin
To obtain indications regarding the developmental processes and conditions in which ABP1 might play a role, we analyzed the ABP1 expression pattern. The analysis of publicly available global transcriptome data in GENEVESTIGATOR® [56] suggested that ABP1 is expressed constitutively in different tissues during development. ABP1 transcription appears to be the highest in rosette leaves and roots, whilst lowest in pollen (Fig. S1A-B). In seedlings, ABP1 is expressed in cotyledons, hypocotyls and root tips as well as in lateral roots. Global transcriptomics data following different perturbations suggested that ABP1 expression is elevated in response to heat and slightly decreased following biotic stress (Fig. S1C).
To obtain more detailed insight into the ABP1 expression pattern and confirm the global transcriptome analysis-based notions, we used an ABP1::GUS line to report ABP1 promoter activity in vivo. GUS staining of 6-d-old seedlings confirmed the ABP1 expression in cotyledons in which we detected stronger ABP1 promoter activity in hydathodes and stomata ( Fig. 1A-B). In both light-and dark-grown hypocotyls, the ABP1 promoter activity was very low (Fig. 1C-D). Further, we confirmed ABP1 expression in the primary root, particularly in the root tip (Fig. 1E) and during different stages of lateral root development (Fig. 1F-H). We observed that ABP1 expression pattern in hydathodes, root tip and lateral roots largely overlaps with that of DR5 reporters for transcriptional auxin response [43,44,50,[57][58][59].
Therefore, we tested whether auxin regulates ABP1 promoter activity and transcription. We employed L-Kynurenine, an inhibitor of auxin biosynthesis [60], to decrease auxin levels in the ABP1::GUS seedlings.
We tested both, the effect of L-Kynurenine treatment alone or with subsequent auxin treatment, to study the effect of exogenously applied auxin. Overall, we detected no obvious changes in GUS reporter activity either after L-Kynurenine or after L-Kynurenine followed by auxin treatments (Fig. 1I).
To additionally verify these observations, we examined the auxin effect on ABP1 transcription using real-time quantitative PCR (RT-qPCR). We performed RT-qPCR with roots and shoots of 5-d-old wildtype seedlings after auxin treatment. Consistent to what we observed with the ABP1::GUS transgenic line, auxin treatment did not strongly affect ABP1 transcription (Fig. 1J).
These results show that ABP1 expression overlaps with auxin response maxima during seedling development, but that ABP1 promoter activity and ABP1 transcription are not significantly regulated by auxin.

Involvement of ABP1 in TIR1/AFB-mediated auxin responses
Considering that ABP1::GUS expression pattern largely overlaps with that of DR5 reporters for transcriptional auxin response [43,44,50,[57][58][59], we investigated whether ABP1 function is in any way linked to the transcriptional auxin signaling downstream of TIR1/AFB receptors [12,61]. First, we introduced DR5rev::GFP reporter into abp1 loss-of-function mutants (abp1-C1 and abp1-TD1). In the abp1 mutant backgrounds, DR5rev::GFP expression pattern in the root tip was not visibly altered and showed the typical maximum in the columella cells and quiescent center [57,58,62] (Fig. 2A). After auxin treatment, the DR5rev::GFP signals in abp1 mutants expanded to the lateral root cap and stele to the same extent as in the control (Fig. 2B). Quantification of the DR5 signal without and with auxin treatment in the root tips did not reveal any differences between the control and abp1 mutants (Fig. 2C). Taken together, these results show that the DR5 auxin response reporter's readout does not depend on a functional ABP1.
Recently it was demonstrated, that the TIR1/AFB pathway is required for a rapid non-transcriptional auxin response [16]. We used this experimental system to investigate TIR1/AFB-mediated non-transcriptional auxin effects on root growth in the mutant lines. Evaluation of root growth on the vertical imaging set-up with high spatio-temporal resolution [16,48] revealed a comparable auxin sensitivity of the abp1-C1 and abp1-TD1 mutants and the control line in terms of rapid inhibition of root growth (Fig. 2D) suggesting that abp1 loss-of-function does not affect the TIR1/AFB-mediated non-transcriptional response.
Next, we tested the effect of ABP1 gain-of-function on TIR1/AFBmediated transcriptional auxin signaling by performing a transient expression assay in Arabidopsis protoplasts. We derived protoplasts from root cell culture, co-transfected them with DR5::LUC reporter together with either 35S::ABP1 or 35S::ABP1-M2X carrying a mutation in the auxin-binding site [35] and measured the DR5::LUC signal with and without auxin. The DR5-driven luciferase activity increased after both short (4 h) and long (16 h) term auxin treatment, however neither ABP1 nor ABP1-M2X overexpression had any significant influence on this induction (Fig. 2E).
These results do not support a strict requirement of ABP1 function in the canonical, TIR1/AFB-mediated auxin signaling pathway.

Role of ABP1 in primary root growth and root gravity response
Since ABP1 is expressed in the primary root and root tip (Fig. 1C, E) and auxin is a major regulator of root growth [16,[63][64][65], we analyzed whether abp1 loss-of-function or the overexpression influences primary root growth. We used two independent loss-of function mutant lines, abp1-C1 and abp1-TD1 and a line expressing ABP1-GFP under the control of the ubiquitous 35S promoter (ABP1-GFP OE ) (Fig. S2). Visually, roots of all tested lines developed normally (Fig. 3A). We measured the root length of 4-and 7-d-old seedlings and found that the root growth of abp1 mutants was comparable to WT, while roots of ABP1-GFP OE were shorter Error bars denote standard deviation. Asterisks indicate significant differences according to Student's t tests (****, P < 0. 0001). The experiment was repeated 2 times with similar results. (Fig. 3A).
Asymmetric auxin distribution is involved in gravitropism, an important plant adaptive process manifested by shoot and root bending [66][67][68][69]. In order to describe a role of ABP1 during root bending, we gravistimulated (90 • reorientation) roots of 4-d-old abp1 and ABP1-GFP OE seedlings for 8 h and measured the root bending kinetics. We observed that abp1 mutants showed a normal root gravitropic response while the roots of ABP1-GFP OE bent significantly slower (Fig. 3B).
In summary, the abp1 loss-of-function mutants do not have any impact on either root growth or root bending, whereas gain-of-function leads to slower root growth and root bending.

Role of ABP1 during lateral root development
As ABP1 is expressed during lateral root development (Fig. 1F-H), and auxin promotes lateral root initiation and formation [59], we analyzed lateral root development in 6-d-old abp1 and ABP1-GFP OE seedlings. We counted and scored all lateral root primordia stages. The analysis revealed that both abp1 mutants and ABP1-GFP OE developed a comparable number of lateral root primordia (Fig. 4A). In addition, we could not find any differences in the frequency of individual primordial stages (Fig. 4B).
To test the auxin effect on lateral root emergence, we transferred 4-dold seedlings to media supplemented with auxin and 3 days later we counted the density of emerged lateral roots. We observed that the density of emerged lateral roots was comparable between abp1 mutants and WT, while ABP1-GFP OE developed less lateral roots (Fig. 4C).
Together, the results presented above demonstrate that both abp1 loss-of-function mutants do not have any impact on lateral root development, but that ABP1 overexpression leads to impaired auxin-induced lateral root development.

Role of ABP1 in etiolated growth and shoot gravity response
Auxin is required for a sustained rapid hypocotyl-elongation of plants grown in darkness [70][71][72]. The auxin-induced growth of etiolated hypocotyl segments is not altered in abp1 loss-of-function mutants [15]. To complement these observations in intact plants, we analyzed growth of etiolated hypocotyls for both abp1 loss-and gain-of-function lines and measured the hypocotyl length of the dark-grown seedlings every twelve hours (Fig. 5A). Initially, the hypocotyls of all tested lines elongated at the same speed. Later, starting 36 h after germination, etiolated hypocotyls of ABP1-GFP OE elongated faster and they were significantly longer than the control 120 h after germination. On the other hand, etiolated hypocotyls of both abp1 mutant alleles elongated comparably to the controls.
The gravitropic response of the hypocotyl is also regulated by auxin [67][68][69]. To investigate a possible function of ABP1 in hypocotyl gravitropism, we gravistimulated 3-d-old etiolated hypocotyls and measured the bending angle after 6, 18 and 24 h. The analysis revealed that the ABP1-GFP OE hypocotyls bend significantly less than WT (Fig. 5B). The difference was noticeable already 6 h after gravistimulation. Notably, both abp1 mutants showed a similar tendency towards slower bending, albeit not significant.
In summary, these observations unveiled that abp1 loss-of-function alleles do not show defects in etiolated hypocotyl growth and gravitropic responses, whereas gain-of-function of ABP1 leads to increased elongation and defective gravity-mediated hypocotyl bending.

Role of ABP1 in leaf development and vasculature formation
In cotyledons, auxin and its directional transport act as a positional cue for vasculature vein formation [73,74] and also regulate leave shape and serration [75]. We analyzed whether ABP1 plays a role in the young rosette growth and development as well as in cotyledon vasculature formation. Macroscopically, neither abp1 mutants nor ABP1-GFP OE showed any defects in cotyledon development (Fig. 6A). We measured the size of young rosettes consisting of both cotyledons and primary leaves. We found that ABP1-GFP OE had slightly bigger rosettes (Fig. 6A).
The vasculature of cotyledons typically consists of four formed closed loops (Fig. 6B). We scored the number and the completeness of these loops in abp1 mutants and ABP1-GFP OE . We observed a normal vasculature pattern in both abp1 mutants, but ABP1-GFP OE showed irregularities at higher frequency than WT (Fig. 6B). The most striking difference in ABP1-GFP OE were fewer loops (22 % in WT and 46 % in ABP1-GFP OE ) and loops that were opened at their upper end, which is almost never seen in WT (2 % in WT and 6.5 % in ABP1-GFP OE ).
The results show that, whilts abp1 loss-of-function has no impact on leaves growth and venation, ABP1 gain-of-function affects vasculature formation.

Role of ABP1 during stress
Abiotic stresses, such as salinity and osmotic stress, induce changes in turgor pressure and in polar auxin transport [76][77][78] and thus lead to root growth inhibition. On the other hand, an increase of auxin biosynthesis results in higher salt tolerance [79,80]. The regulation of ABP1 transcription by various stresses such as heat (Fig. S1) prompted us to test the requirement of ABP1 to adapt to stress.
We challenged abp1 mutants with osmotic stress using mannitol or sodium chloride treatments to assess the involvement of ABP1 in stress responses. Overall, following the treatments, root growth and lateral root formation of WT and abp1 mutants were inhibited (Fig. 7A-C). In addition, no obvious differences in root growth inhibition were observed after mannitol or sodium chloride treatment between the tested lines ( Fig. 7B-C).
High temperature promotes auxin biosynthesis, thereby leading to rapid hypocotyl growth [70]. To address a potential role of ABP1 in auxin-mediated rapid hypocotyl growth in response to high temperature and the presence of sugar, we characterized hypocotyl elongation of abp1 and ABP1-GFP OE seedlings grown under high temperature (28 • C), on media supplemented with or without sucrose. When grown in high temperature (28 • C) on the medium with sucrose, ABP1-GFP OE exhibited longer hypocotyls compared to WT, whereas the hypocotyl length of abp1 mutants was comparable to that of WT plants (Fig. 7D). At high temperature (28 • C), but in absence, of sucrose the hypocotyl elongation of ABP1-GFP OE line was less inhibited than in WT (Fig. 7E).
To test whether ABP1 plays a role in wound healing responses, we performed a targeted cell ablation in the root tips of abp1-TD1 and ABP1-GFP OE lines [53,81]. After cell ablation, the numbers of initiating periclinal cell divisions in abp1-TD1 and ABP1-GFP OE were similar to that in WT (Fig. S3).
Taken together, the results show that the root growth of abp1 loss-offunction mutants is not influenced differently by salt stress and high temperature. ABP1 gain-of-function seedlings show increased hypocotyl growth when grown at high temperature. (A) Representative images of the rosettes of 12-d-old of Col-0, abp1-C1, abp1-TD1 and ABP1-GFP OE seedlings. Scale bar =5 mm. The boxplot shows the size of the rosettes for Col-0, abp1-C1, abp1-TD1, complemented abp1-TD1 mutant (ABP1::ABP1;abp1-TD1) as control for abp1-TD1, and ABP1-GFP OE . For each genotype and experiment, more than 19 rosettes from 12-d-old seedlings were measured. For box plot, box defines the first and third quartiles, and the central lines in the box represent the median. Whiskers, from minimum to maximum. Asterisks indicate significant differences according to Student's t-tests (*, P < 0.05). The experiment was repeated 2 times with the similar result.

Role of ABP1 in rosette leaves and inflorescence development
The establishment of auxin maxima in the shoot apical meristem (SAM) and directed basipetal polar auxin transport are crucial for overall shoot development [59,[82][83][84][85][86]. ABP1 is expressed in both SAM and rosette leaves (Fig. S1B), therefore we investigated its possible function in shoot development.
First, we characterized leaf development. Visually, the size and shape of rosette leaves in abp1 mutants and ABP1-GFP OE plants were comparable to that of WT plants. We quantified the rosette leaves number at the stage when the first flower of each individual plant bloomed. We observed that the apb1-TD1 mutant developed slightly more, whereas the ABP1-GFP OE line developed significantly less rosette leaves in comparison to WT (Fig. 8A). However, the results for the abp1-TD1 mutant line were variable between the experimental repetitions. We found no difference in the number of cauline leaves for any of the analyzed lines (Fig. S4A).
Further, we studied the function of ABP1 during bolting. We measured the length of the first internode of abp1 mutants and ABP1-GFP OE and we recorded the timing to reach 1 cm. Compared to WT, both abp1 mutants and ABP1-GFP OE line bolt earlier, at 21 st and 22 nd day after sowing versus 23 rd day in WT (Fig. 8B).
To determine whether ABP1 is involved in phyllotaxis establishment, we measured the sequence of divergence angles between siliques in abp1 mutants. Visually, abp1 mutants developed normal inflorescence stems (Fig. 8C). WT plants typically exhibit a spiral phyllotaxis that leads to a distribution of the consecutive organs on the stem with a divergence angle close to 137.5 • [87]. Our analysis revealed that the distribution of divergence angles in abp1 mutants was not altered (Fig. 8D-F). We also analyzed the internode length between the siliques and counted the number of rosette and cauline branches of abp1 mutant and ABP1-GFP OE plants. However, we did not detect any differences (Fig. S4B-D).
The results show that overexpression of ABP1 affects the number of rosette leaves and that both ABP1 loss-and gain-of-function accelerate bolting.

Role of ABP1 in auxin-mediated PIN polarization and BFAvisualized PIN trafficking
The formation of organized vasculature requires coordinated cell polarization. The canalization hypothesis proposes that auxin acts as a polarizing cue in this process [88] and that auxin feed-back on PIN polarity, together with constitutive PIN endocytic trafficking are important features in this process [54,89,90]. Since overexpression of ABP1 results in defects in vascular tissue formation (Fig. 6B), we tested whether abp1 loss-or gain-of-function alleles show defects in these For each genotype and experiment, at least 25 hypocotyls were analyzed. Error bars denote standard error. Asterisks indicate significant differences according to Student's t tests (****, P < 0. 0001). The experiment was repeated 3 times with the similar results. (E) Quantification of the hypocotyl growth inhibition sucrose of Col-0, abp1-C1, abp1-TD1 and ABP1-GFP OE seedlings grown under continuous light, higher temperature (28 • C) and in absence of sucrose. For each genotype, at least 10 roots were measured per experiment. Asterisks indicate significant differences according to Student's t tests (****, P < 0. 0001). The experiment was repeated 3 times with the similar results and the pooled values are presented. processes. To evaluate the effect of auxin on PIN polarity, we analyzed the repolarization of PIN1 from the basal to the inner lateral side in root endodermis cells and the repolarization of PIN2 from the basal to the outer lateral side in root cortex cells [54] following auxin treatment in abp1 mutants and ABP1-GFP OE . Anti-PIN1 and anti-PIN2 immunolocalization revealed that PIN1 and PIN2 repolarization was not altered in abp1 mutants, while overexpression of ABP1 led to reduced or no repolarization of PIN1 and PIN2 respectively (Fig. 9A-B).
Further, we used the trafficking inhibitor Brefeldin A (BFA) to indirectly visualize PIN intracellular trafficking [91]. BFA treatment results in PIN internal aggregation manifested as BFA-body formation and this effect is decreased when BFA is used together with auxin [9]. The anti-PIN1 immunostaining in roots after BFA treatment showed that the intracellular aggregation of PIN1 was similar to that of WT in both abp1 mutants (Fig. 9C) and [92]. In ABP1-GFP OE we observed repeatedly that BFA affected PIN1 intracellular aggregation more severely (BFA bodies were more pronounced) (Fig. 9C). Anti-PIN1 immunostaining after auxin and BFA co-treatment confirmed that auxin inhibited BFA-body formation. Comparison of the abp1 mutants with the corresponding complemented lines did not reveal any consistent changes in the auxin effect on BFA-induced PIN1 aggregation, whereas ABP1-GFP OE showed again slightly more BFA-induced PIN1 aggregation even in presence of auxin (Fig. 9C). The analysis of the BFA effect on PIN2 intracellular aggregation revealed no consistent and reproducible differences in BFA-body formation between WT, abp1 mutants and ABP1-GFP OE (Fig. 9D). Accordingly, auxin and BFA co-treatment led to a comparable and variable decrease of PIN2 intracellular aggregation in WT, abp1 mutants and ABP1-GFP OE (Fig. 9D).
Taken together, the ABP1 overexpression interferes with auxininduced PIN repolarization and slightly affects BFA-induced, constitutive PIN1 but not PIN2 trafficking, while mutation in ABP1 does not show altered auxin feed-back on PIN polarity or constitutive PIN recycling.

Discussion
ABP1 has been identified in maize decades ago based on its potential ability to bind auxin [93,94]. Nonetheless, the developmental roles and cellular functions of ABP1 remain unclear due to problems with some of the genetic material [35,39,40,42] and due to the lack of obvious developmental defects after superficial analyzes of the verified knock-out lines [38].
Here, we assessed the function of ABP1 in various developmental processes and (re)evaluated its role in cellular processes related to trafficking and polar distribution of PIN auxin transporters.

ABP1 is not essential for or regulated by TIR1/AFB-mediated auxin responses
ABP1 promoter activity has been reported to overlap, to some extent, with that of the transcriptional DR5 auxin reporter during early seedling development [43]. Our analysis revealed a similar overlap in hydathodes, root tips and lateral root primordia as well as in older seedlings. The activity of the ABP1 promoter at places with high auxin response suggested either that auxin might regulate the transcription of ABP1 or that ABP1 is somehow linked to TIR1/AFB-mediated transcriptional auxin signaling.
Indeed, ABP1 was previously identified among early auxin-regulated genes. ABP1 transcription was upregulated by auxin in a dose dependent manner within 30 min in 19-d-old WT seedlings [95]. Our observations in 5-d-old WT roots and shoots did not reveal any changes in ABP1 expression following auxin treatment. These contradictory findings suggest that a potential auxin effect on ABP1 transcription could be tissue-and/or developmental stage-dependent.
Also, the connection between ABP1 and TIR1/AFB-mediated auxin signaling was previously investigated. Downregulation of the ABP1 activity was shown to affect transcription of auxin-responsive genes [33,95,96], to regulate Aux/IAA homeostasis and thus negatively impact on the SCF TIR1/AFB pathway [97]. However, these observations are inconclusive due to the potential off-targets in the conditional knock-down lines [42] and the inactivation of ABP1 did not have any significant effects on the DR5 auxin response reporter activity [33]. In the verified abp1 knock-out lines it was reported that auxin-regulated gene expression is unchanged [38] and our analysis in these lines and following ABP1 overexpression in protoplasts did also not reveal any changes in DR5 reporter activity. Furthermore, abp1 knock-out lines also showed normal TIR1/AFB-mediated non-transcriptional auxin effect on root growth. Overall, these observations suggest that ABP1 is not directly involved in the TIR1/AFB-mediated auxin response.

ABP1 loss-of-function mutants show minor defects in development
The initial analysis of CRISPR and T-DNA insertion abp1 knock-out mutants did not reveal any obvious defects during development under normal conditions leading to a conclusion that ABP1 is not required for Arabidopsis development [38]. We analyzed different auxin-related phenotypes of the corresponding abp1 knock-out mutants in more detail. We observed that both abp1 alleles exhibited normal root growth, etiolated hypocotyl, root and shoot gravitropic responses, lateral root and leaf development, including venation and phyllotaxis. Notably, both abp1 mutant alleles bolted earlier compared to the control lines. Accelerated bolting in abp1 mutants might be caused by changes in auxin levels caused by either impaired biosynthesis, auxin transport or eventually a change in auxin sensitivity. Nonetheless, it is unclear why such changes are not reflected also in other developmental processes regulated by auxin.

ABP1 gain-of-function lines show a plethora of auxin-related phenotypes
ABP1 overexpression has been shown previously to cause several postembryonic developmental defects [5,10,35,98,99]. Similarly, our analysis of a stable line expressing 35S::ABP1-GFP revealed that ABP1 gain-of-function leads to developmental changes. Seedlings overexpressing ABP1 have reduced root length, impaired auxin-induced lateral root development, enhanced elongation of both high temperature-and dark-grown hypocotyls, reduced root and shoot gravitropic response, defective vasculature development, increased size of young rosettes but decreased number of rosettes leaves. Additionally, similar to the abp1 mutants, ABP1 overexpressors also bolted earlier. At the cellular level, we confirmed the previous observations [10,35] that the ABP1 gain-of-function affects the BFA-sensitive PIN endocytic trafficking and newly showed that they also impair auxin effects on PIN polar distribution in root cells.
All aforementioned processes, which were found defective in ABP1 gain-of-function mutants are linked to auxin regulation. It is therefore conceivable that, in line with the importance of the auxin binding pocket for the ABP1 function [35], ABP1 plays so far a mechanistically unclear role in auxin perception and signaling.

Potential role and functional mechanism of Arabidopsis ABP1
Arabidopsis ABP1 was identified based on the orthology with ABP1 previously found in maize [98,100]. Auxin-binding properties of maize ABP1 are well characterized. Several biochemical studies along with the structural analysis of the ABP1-auxin co-crystal revealed that maize ABP1 binds auxin with the highest affinity at apoplastic pH 5.5, while binding at pH 7.2 corresponding to the ER lumen where the majority of protein is localized, is much lower [26][27][28][29]94,101]. In contrast, the auxin-binding properties of Arabidopsis ABP1 have not been characterized yet. Based on the high homology with the maize protein, it is assumed that Arabidopsis ABP1 binds auxin in a similar manner. This shows the scoring of an overall count of formed BFA bodies in Col-0, abp1-C1, statement is supported by the finding that the auxin-binding pocket of Arabidopsis ABP1 is important for its gain-of-function cellular and developmental roles [35].
The ABP1 binding optimum at pH 5.5 would imply that ABP1 is functional in the apoplast, further supported by auxin-dependent interaction between ABP1 and the plasma membrane-localized receptor-like kinase TMK1 [30,31]. TMK1 belongs to a four-member TMK receptor-like kinase family, that function redundantly and multiple mutants show severe reduction in organ size and substantial growth retardation [20]. Both TMK1 and TMK4 play roles in auxin-mediated developmental processes and in the control of local auxin biosynthesis [22,23,30]. Importantly, TMK1 mediates auxin signaling that regulates differential growth of the apical hook [21]. However, the mechanism of how TMK1 perceives auxin remains elusive.
The function of ABP1 as a part of the auxin perception machinery contributing towards TMK-based downstream signaling, is a tempting hypothesis consistent with a rather broad spectrum of auxin-related growth defects. But it is not supported by the rather mild phenotypic defects in the abp1 loss-of-function mutants, especially considering that ABP1 is a single copy gene in Arabidopsis [100]. On the other hand, ABP1 is evolutionary conserved and ubiquitous in vascular plants [102], suggesting that it has an important and conserved function. Structurally ABP1 belongs to an ancient group of germin and germin-like proteins that have a highly conserved tertiary structure despite low similarity in primary sequence among the members [28,103]. Therefore, it is possible that some other proteins from the germin family are functionally redundant with ABP1, thus masking the effect of the abp1 mutation. Nonetheless, to identify and characterize functional homologues within this large family will be a challenging task. An alternative explanation for the weak developmental defects in abp1 loss-of-function mutants is that ABP1 plays an important role in specific processes that provide competitive advantage in nature but are not easily manifested under laboratory conditions.

Conclusions
In conclusion, our detailed phenotypic analysis of both ABP1 gainand loss-of-function lines provides new insights into the developmental role of ABP1. Despite the overlap of ABP1 expression pattern with auxin response maxima during seedling development, none of our observations supports a direct involvement of ABP1 in the TIR1/AFB-mediated transcriptional auxin response. abp1 knock-out mutants show only mild phenotypic defects, whereas ABP1 overexpression generates a broad range of potentially auxin-related phenotypes. The previously described strong and related defects in conditional abp1 knock-down lines let us hypothesize that the discrepancy between the effects of loss-and gainof-function is due to the action of unknown germin family proteins that are functionally redundant with ABP1.