Antagonistic cell surface and intracellular auxin signalling regulate plasma membrane H+-fluxes for root growth


 Growth regulation tailors plant development to its environment. A showcase is growth adaptation to gravity, where shoots bend up and roots down. This paradox is based on different responses to the phytohormone auxin, which promotes cell expansion in shoots, while inhibiting it in roots via a yet unknown cellular mechanism. Here, by combining microfluidics, live imaging, genetic engineering and phospho-proteomics in Arabidopsis thaliana, we reveal how auxin inhibits root growth. We show that auxin activates two distinct, antagonistically acting signalling pathways that converge on the rapid regulation of the apoplastic pH, which is the direct growth-determining mechanism. Cell surface-based TRANSMEMBRANE KINASE 1 (TMK1) interacts with and mediates phosphorylation and activation of plasma membrane H+-ATPases for apoplast acidification, while intracellular TIR1/AFB-mediated signalling triggers net cellular H+-influx, causing apoplast alkalinisation. The simultaneous activation of these two counteracting mechanisms poises the root for a rapid, fine-tuned growth modulation to subtle changes in the environment.

Auxin, as a major growth regulator in plants, acts oppositely in shoots and roots. In shoots, it promotes cell elongation by activating H + -pumps to acidify the apoplast 1 in agreement with the classical Acid Growth Theory, which postulates that low apoplastic pH promotes growth [2][3][4] . This relatively slow process relies on the canonical/intracellular TIR1/AFB receptors-mediated pathway of transcriptional regulation [5][6][7] . In many species including the model Arabidopsis, however, auxin inhibits growth in roots 8 by an unclear cellular mechanism. This contrasting response to the same signal is the basis for the positive versus negative bending of roots and shoots in response to gravity and light. The inhibitory auxin effect in roots is very rapid, but also involves TIR1/AFB receptors 9 . The timing indicates the existence of a rapid, non-transcriptional branch of this pathway 9 , nonetheless with unclear downstream molecular, cellular and physiological mechanisms. Besides intracellular TIR1/AFB auxin signalling, a cell surface-based pathway mediated by TRANSMEMBRANE KINASE 1 (TMK1) has been shown to regulate differential growth in the apical hook 10 , while its contribution to a general growth regulation, remains unclear. Hence, the auxin signalling mechanism and the downstream cellular mechanisms for regulating root growth remain elusive.
In this study, we revealed simultaneous antagonistic activities of TIR1/AFB intracellular and TMK1-based cell surface auxin signalling converging on regulation of apoplastic pH, which we identified as the key cellular mechanism allowing immediate and sensitive root growth regulation.
implicated, the causal mechanism mediating rapid auxin inhibition of root growth remains unidentified.
The microfluidic vRootchip 9 setup in combination with vertical confocal microscopy 17 allowed us to critically re-evaluate the contributions of these processes by comparing their kinetics. Growth inhibition was observed within 30 seconds after 10 nM of the natural auxin indole-3-acetic acid (IAA) 9 . Using the microtubule plus-end marker pEB1b::EB1b-GFP, we observed that even using 100 nM IAA, less than 5% of the CMT in elongating epidermal cells reoriented after 1 minute (Extended Data Fig. 1a-b). Furthermore, inhibition of CMT dynamics by taxol treatment blocked auxin-mediated CMT reorientation (Extended Data Fig. 1c-d), but had no effect on auxin-mediated growth inhibition (Extended Data Fig. 1e). We visualized vacuolar morphology using tonoplast marker pSYP22::SYP22-YFP. By live imaging the same cells, we could not detect obvious changes in vacuolar morphology even after 30 minutes of 100 nM IAA treatment (Extended Data Fig. 1f), when root growth was already strongly inhibited. These results show that both CMT reorientation and vacuole constriction occur well after the rapid auxin-mediated growth inhibition, arguing against their involvement in this process.
To evaluate the kinetics of apoplastic pH changes, we applied a membrane-impermeable ratiometric pH indicator: 8-hydroxypyrene-1,3,6-trisulfonic acid (HPTS) 14 and imaged apoplastic pH in elongating epidermis cells while simultaneously tracking root tip growth in the vRootchip. We detected a gradient in the apoplastic pH, decreasing from transition to elongation zone, in root epidermal cells (Extended Data Fig. 1g). Regardless of their position, all cells showed a strong apoplastic pH increase within 30 seconds of 5 nM IAA treatment (Fig. 1a, b and Extended Data Fig. 1g). This auxin-mediated alkalinisation was observed at the same time as root growth inhibition (Fig. 1b). The pH increase was very robust as we also detected alkalinisation of the external medium along the root tip surface (Extended Data Fig. 1h). In a complementary approach, we analysed the cytosolic pH by using the PM-Cyto reporter 18 . The cytosolic pH adjacent to the PM decreased within 30 seconds of 5 nM IAA treatment (Fig. 1c). The concomitant apoplastic pH increase and intracellular decrease in elongating root epidermal cells suggest that auxin causes a H + -influx into the cell. This notion was confirmed in direct experiments using non-invasive ion microelectrodes monitoring the net H + -exchange across the PM of root elongating epidermis cells (Extended Data Fig. 1i).
Overall, auxin triggers rapid apoplast alkalinisation by increasing the net H + -flux into cells. Strong spatial and temporal correlation with auxin-mediated root growth inhibition suggests that apoplast alkalinisation may be the key cellular mechanism, by which auxin rapidly inhibits root growth.

Apoplastic pH is causative for rapid root growth regulation
To investigate the causal relationship between apoplast alkalinisation and root growth inhibition, we manipulated the apoplastic pH by changing the pH of the medium and monitoring the impact on root growth using the vRootchip. Indeed, pH manipulation of the medium had a rapid impact on the root apoplastic pH (Extended Data Fig. 1j-k).
Replacement of the standard medium at pH 5.8 by a more alkaline (pH 6.15) medium caused instant reduction of root growth; the growth rate restored rapidly after washout with the original pH 5.8 medium (Fig. 1d, e). Gradual alkalinisation of the medium resulted in gradual root growth inhibition (Extended Data Fig. 1l). On the other hand, replacing the alkaline medium by more acidic pH 5.1 medium increased root growth instantly and washout restored the original growth rate (Fig. 1f, g).
Thus, exogenous manipulation of apoplastic pH has an immediate and reversible effect on root growth, with alkaline pH inhibiting and acidic pH promoting growth. This resembles the root growth inhibition effects of auxin in both speed and reversibility and strongly supports that auxin-induced apoplast alkalinisation is the key downstream cellular mechanism for rapid root growth inhibition.

Auxin rapidly phosphorylates and activates plasma membrane H + -ATPases in the root
The auxin effect on apoplast alkalinisation occurs within seconds, thus being too fast to involve transcriptional regulation. To gain insights into the underlying mechanism, we mined recent datasets from Mass Spectroscopy (MS)-aided phospho-proteomics in Arabidopsis root tips treated for 2 minutes with 100 nM IAA (Han et al., manuscript in preparation). Among the top differentially phosphorylated targets were two Plasma Membrane (PM) H + -ATPases encoded by AT2G18960 (AHA1) and AT4G30190 (AHA2). Multiple putative phosphorylation sites were identified in the auto-inhibitory C-terminal region as being regulated by auxin, potentially leading to both activation and deactivation of H + -pump activity 19,20 (Fig. 2a and Extended Data Table 1). This suggested that auxin may regulate H + -pump activity in roots via phosphorylation, providing a plausible mechanism for the observed rapid auxin effect on apoplastic pH.
To test whether and how auxin changes the activity of PM H + -ATPases in roots, we performed an ATP hydrolysis assay measuring the hydrolytic release of inorganic phosphate from the ATP, which has been shown to correlate with the activity of the H + -ATPases and with the H + -extrusion 21,22 . After 1 hour 100 nM IAA treatment, we detected increased ATP hydrolysis activity in root protein extracts (Fig. 2b), similarly as observed in shoots following auxin treatment (Lin et al., submitted). This supports the notion that auxin activates H + -pumps also in root cells, which should lead to apoplast acidification contrasting with the observed alkalinisation (see Fig. 1b).
Hereafter, we reanalysed the phospho-proteomics data specifically for the phosphorylation of Thr 947 in AHA2, a known activation site 23,24 . Thr 947 was significantly more phosphorylated after IAA treatment (Fig. 2c), suggesting that auxin activates AHAs in the root by stimulating this phosphorylation. To confirm the phosphorylation state of AHA2, we used antibodies that specifically recognize the catalytic domain of AHA2 and phosphorylated Thr 947 . We pre-treated seedlings with 50 μM of the auxin biosynthesis inhibitor kynurenine for 24 hours, followed by 10 nM IAA treatment and harvested roots at given time points. At such low concentration, IAA strongly induced phosphorylation of Thr 947 of AHA2 within 10-20 minutes and additionally led to an increase of AHA2 protein levels at later time points (Fig. 2d). Thus, auxin mediates AHA2 phosphorylation leading to its activation.
Our results show that, consistently with previous findings in shoots 7,22 , auxin rapidly induces AHA phosphorylation leading to H + -pump activation in the root. This, however, should lead to apoplast acidification and not the observed auxin-induced apoplast alkalinisation (see Fig. 1b) suggesting that in roots H + -pump activation may act antagonistically, presumably as a negative feed-back to auxin-mediated apoplast alkalinisation.

H + -ATPases activation counteracts auxin-mediated apoplast alkalinisation and growth inhibition
To better understand the role of H + -pump activation during auxin-mediated root growth inhibition, we used the fungal toxin Fusicoccin (FC). FC stabilizes binding of 14-3-3 proteins to H + -ATPases, thus rapidly and specifically activating them 23,25,26 , without otherwise affecting auxin signalling (Extended Data Fig. 2a). This allowed us to dissect the effects of auxin and H + -ATPase activation on pH and growth simultaneously. FC caused rapid apoplast acidification and promoted growth in roots (Extended Data Fig. 2b, c), opposite to the auxin effect. When FC and IAA were applied simultaneously, we observed an intermediate outcome on apoplastic pH and root growth proportional to the ratio of auxin versus FC (Fig. 2e, f and Extended Data Fig. 2f, g). Similar observations were made when FC and IAA were added sequentially (Extended Data Fig. 2b-e). These observations suggest that FC-mediated H + -ATPase activation and IAA-mediated apoplast alkalinisation act antagonistically during apoplastic pH and root growth regulation.
To test this notion genetically, we analysed auxin response of loss-and gain-of-function aha mutants. Single aha1 and aha2 mutants have no consistent root phenotypic defects, when grown on auxin (Extended Data Fig. 2h), while the double mutant is embryo-lethal 27 . To overcome redundancy within the AHA family and test for AHA function specifically in cells where root growth is controlled, we used a synthetic trans-acting siRNA 28 targeting AHA1/2/7/11 (AtTAS1c-AHA), expressed from the PIN2 promoter in the outer root tissues 29 . Expression of all four AHAs was verifiably downregulated in both independent transgenic lines (Extended Data Fig. 2i). Both AtTAS1c-AHA lines were hypersensitive to auxin in terms of apoplast alkalinisation (Fig. 2g) and root growth inhibition (Fig. 2i). In contrast, constitutive activation of AHA1 in the ost2-3D mutants 30 resulted in decreased auxin sensitivity of apoplastic pH (Fig. 2h) and root growth (Fig. 2i). These aha mutant phenotypes further confirm that auxin-mediated H + -ATPase activation antagonizes auxininduced apoplast alkalinisation in the root.
These observations show that H + -pump activation in roots acts against the observed auxinmediated apoplast alkalinisation. The auxin effect on root growth is thus the net result of auxin-induced AHA activation, which presumably acts as negative feed-back against the more dominant auxin-mediated alkalinisation.

TMK1 interacts with H + -ATPases
To gain insights into the mechanisms, by which auxin signalling regulates apoplastic pH, we performed co-immunoprecipitation pull-down assays followed by MS-assisted identification of proteins associated with either the intracellular TIR1/AFB1 receptor or the cell surface TMK1 Leucine-rich repeat receptor-like kinase 31 . For TIR1 and AFB1 this approach did not reveal any components that have yet been linked to control of H + -transport (Extended Data Fig. 3a and Extended Data Table 2). For TMK1, although enrichment was not strong and the identified peptides did not allow discriminating between the two proteins, we found peptides of AHA1/AHA2 among the top enriched associated peptides for TMK1 (Extended Data Fig. 3b and Extended Data Table 3).
These observations show that TMK1, a presumable component of cell surface auxin signalling, interacts with the PM H + -ATPase AHA2.

TMK1 activity mediates auxin effect on H + -ATPases phosphorylation and activation
To test the role of TMK1 in H + -ATPase phosphorylation, we performed phospho-proteomic analysis in roots of tmk1-1 compared to WT, and detected strong hypophosphorylation of AHAs in the tmk1-1 ( Fig. 3d and Extended Data Table 1). This stipulates that TMK1 is involved in H + -ATPases phosphorylation.
To further confirm this notion, we cloned p35S::TMK1-HA and two kinase-dead versions with mutations in the ATP binding site, in which K 616 is exchanged to either E or R (TMK1 K616E or TMK1 K616R ). Transient overexpression of the wild type (WT), but not the kinase-dead TMK1 constructs in tobacco resulted in rapid wilting of the leaves (Extended Data Fig. 3e), an effect consistent with PM H + -ATPases activation 32 . To test this directly in Arabidopsis roots, we generated dexamethasone (DEX)-inducible TMK1 gain-of-function lines and assessed the phosphorylation status of AHA2. In the root extract expressing TMK1 K616R , no phosphorylation of AHA Thr947 was observed after 100 nM IAA treatment compared to TMK1 WT (Extended Data Fig. 3f).
In summary, these independent approaches show that active TMK1 kinase is required for the auxin-mediated phosphorylation and activation of H + -ATPases in roots.

TIR1/AFB and TMK1 signalling converge antagonistically on apoplastic pH and growth regulation
Our results show that TMK1 interacts with, and mediates the phosphorylation and activation of PM H + -ATPases (see Fig. 3), which results in apoplast acidification (see Fig. 2). How can we reconcile this with the observed auxin-mediated apoplast alkalinisation leading to growth inhibition (see Fig. 1)?
Previously, it was proposed that the auxin influx transporter AUXIN RESISTANT1 (AUX1) and the intracellular TIR1/AFB auxin receptors mediate auxin-induced membrane depolarisation associated with changes in H + -fluxes across the PM 15,33 . Therefore, we set out to assess the interplay of these components with TMK1 action in regulation of apoplastic pH. First, we used aux1-100 mutants defective in the AUXIN RESISTANT 1 (AUX1), which mediates cellular uptake of IAA and, in particular, of the synthetic auxin analogue 2,4-dichlorophenoxyacetic acid (2,4-D) 34 . Consistently with previous reports 9 , aux1-100 roots were less sensitive to 5 nM IAA in the vRootchip (Extended Data Fig. 4a, b) or to 0.1 μM 2,4-D (Extended Data Fig. 4c, d) in the steady-state conditions, both for the apoplastic pH and the root growth inhibition suggesting intracellular auxin perception is required for auxin-mediated apoplast alkalinisation.
Given that intracellular TIR1/AFB receptors mediate auxin-mediated growth inhibition 9 , we evaluated apoplastic pH in parallel to growth in the tir1-1, afb2-1, afb3-1 (tir triple) mutant. The tir triple roots were resistant to 5 nM IAA in both apoplast alkalinisation and growth inhibition ( Fig. 4a and Extended Data Fig. 4e). As a pharmacological alternative, we applied PEO-IAA, which acts as an anti-auxin and blocks downstream signalling following binding to TIR1/AFB receptors 35 . Simultaneous addition of 10 μM PEO-IAA and 5 nM IAA prevented apoplast alkalinisation and growth inhibition (Extended Data Fig. 4f, g). Both approaches further corroborate the involvement of the TIR1/AFB auxin perception in auxin-mediated apoplast alkalinisation.
To test the TIR1/AFB requirement, definitively, we took advantage of the cvxIAA/ccvTIR1 system, in which the engineered concave (ccv) TIR1 receptor cannot interact with the natural IAA, but only with a synthetic convex (cvx) IAA analogue, thus allowing specific activation of TIR1/AFB signalling 7 . Application of 50 nM cvxIAA in the vRootchip resulted in apoplastic alkalinisation in the ccvTIR1 plants, but not in the control (Fig. 4b), confirming that specific activation of the TIR1/AFB pathway is sufficient to induce root growth inhibition 9 and apoplast alkalinisation.
These observations suggest that intracellular TIR1/AFB signalling mediates the dominant auxin effect: apoplast alkalinisation and root growth inhibition. This is then counteracted by the cell surface TMK1-mediated H + -ATPase activation for apoplast acidification. Indeed, in the steady state, TMK1 function is redundantly required for root growth as demonstrated by shorter roots in tmk mutants (Fig. 4c) 36 . In response to low concentration of auxin, tmk1-related mutants were more sensitive (Fig. 4d), while overexpressing TMK1 (pUBQ10::TMK1-3HA) led to a slight auxin resistance (Extended Data Fig. 4h). This resembles the corresponding loss-and gain-of-function aha mutants (see Fig. 2) providing additional support for the antagonistic, growth-promoting role of TMK-mediated AHA activity.
To explore further the antagonism of TIR1/AFB and TMK1, we created double mutant tmk1-1, tir1-1 (tmk1,tir1) and analysed the auxin effect on apoplastic pH and root growth. As expected, the tmk1,tir1 mutant showed intermediate auxin sensitivity compared to the single mutants both for growth and apoplastic pH (Fig. 4e, f and Extended Data Fig. 4i, j).
Collectively, these results strongly suggest that auxin activates two antagonistically acting signalling pathways: (i) the cell surface, TMK1-mediated H + -efflux acidifying apoplast and (ii) an intracellular TIR1/AFB-dependent apopast alkalinisation leading to the rapid growth inhibition (Fig. 4g). It seems that the TIR1/AFB-mediated apoplast alkalinisation is dominant, in particular at conditions of higher auxin after its exogenous application. But at low auxin levels, TMK1-mediated acidification becomes more apparent and fine-tunes the growth inhibition outcome.

Conclusions
Our findings provide novel insights into a long-standing question on how root growth is regulated in plants. In particular, we addresses the old mystery of opposite growth regulation in shoots and roots by the phytohormone auxin and we also clarify the downstream cellular mechanism of auxin-mediated root growth inhibition.
Auxin regulates root growth very rapidly, utilizing a non-transcriptional branch of a signalling pathway downstream of intracellular TIR1/AFB receptors 9 . This pathway mediates apoplast alkalinisation, which we show as the direct, causative cellular mechanism for root growth regulation. This finding extends the classical Acid Growth Theory proposing that low apoplastic pH promotes growth 2,3 . We demonstrate validity of this mechanism also for growth inhibition, showing that root growth rate is immediately and reversibly determined by apoplastic pH changes.
Remarkably, the auxin-mediated apoplast alkalinisation in roots does not occur through the regulation of PM H + -ATPases as observed in shoots, where the same TIR1/AFB auxin perception mechanism leads to PM H + -ATPases activation and apoplast acidification 5,7,22 .
Instead, in roots, PM H + -ATPases are phosphorylated and activated via the cell surface TMK1 receptor like kinase-based auxin signalling, which leads to apoplast acidification and root growth promotion. This mechanism, acting antagonistically to the more dominant TIR1/ AFB-mediated alkalinisation, constitutes a negative feed-back, presumably to fine-tune root growth to be able to rapidly respond to subtle changes of environmental stimuli.
Key remaining open questions are: (i) if not by PM H + -ATPases, how does TIR1/AFB signalling mediate alkalinisation in roots? A plausible scenario would be a rapid increase in H + -permeability across the PM, which may be mediated by TIR1/AFB-activated Ca 2+ signalling 33 and would impact on the apoplastic pH and membrane potential. (ii) What is the auxin perception mechanism for the TMK1 pathway? Is it direct activation of TMK1 by auxin or through another yet to be identified associated auxin receptor?
With cell surface-based TMK1 activating H + -pumps and intracellular TIR1/AFB signalling causing the net cellular H + -influx, two auxin-triggered antagonistic mechanisms converge on the regulation of extracellular pH, which directly determines root growth rates. This seemingly counterproductive simultaneous 'gas and brake' action presumably poises the root tip for rapid and flexible changes of the growth rate and direction in response to numerous stimuli during the challenging task to navigate through the complex soil environments.

Microfluidics
The microfluidic vRootchip was used to analyze root tip growth and apoplastic pH in real-time. The manufacturing of the chip, sample preparation procedure and data analysis of root tip growth were performed as described previously 22 . Our new design contains an additional valve in the control layer that closes the ends of the root channels (Extended Data Fig. 1m). In case of air bubbles in the root channels, the additional valve allows pressurizing the channel and air will be absorbed into the Polydimethylsiloxane (PDMS) chip material within 2-10 minutes. Afterwards, experiments started after adaptation of at least two hours. Besides, we introduced a graphical user interface (Extended Data Script 1) using the Processing software (https://processing.org/) with the ControlIP5 package (http://www.sojamo.de/libraries/controlP5/) that sends serial commands to the Arduino. A sketch (Extended Data Script 2) runs on the Arduino to operate the electronics and receive commands. For one vRootchip, maximum 8 samples were used. When comparing two genotypes, 3-4 seedlings were used for each genotype and mounted in alternating channels to minimize the time difference between imaging the two genotypes. For each root, we imaged one ROI containing early elongating epidermal cells and the other ROI covering the root tip. As these two ROIs were captured sequentially, we imaged the apoplastic pH and the growth of the same root close to simultaneously.

Scanner growth assay
To complement the real-time imaging in the vRootchip, growth analysis was performed on a vertical scanner with bigger sample sizes allowing more conditions to be evaluated. This growth measurement we called steady-state. 4-day-old seedlings were transferred to 60 × 15 mm petri dishes filled with 5 ml of ½ MS medium with treatments as indicated. The petri dishes were placed on a vertically mounted flatbed scanner (Epson perfection V370) and seedlings were imaged through the layer of medium. Either wet black filter paper or ½ MS medium containing activated charcoal was placed in the lid of the petri dishes to improve background contrast. The samples were automatically imaged every 10 or 30 minutes using the AutoIt script described previously 23 and scans were taken at 1200 dpi. The resulting image series were analyzed using StackReg stabilization and the Manual Tracking plugin in ImageJ. Li 24 . Fluorescent signals for protonated HPTS (excitation 405 nm, emission 514 nm, visualized in red) and deprotonated HPTS (excitation, 488 nm, emission 514 nm, visualized in green) were detected with a 20x/0.8 air objective. Image analysis was performed on a cropped region of elongating epidermis cells using a batch processing modification of a previously described the ImageJ macro 25 . The relative pH value is calculated as the background-subtracted intensity of the deprotonated intensity divided by that of the protonated intensity to represent the relative pH. The resulting relative pH data were plotted over time and statistically evaluated in GraphPad Prism 6. Note that we did not transform the relative pH value to the absolute pH values, which would require the generation of a calibration curve for each experiment.

Imaging and measuring cytosolic pH with the PM-cyto reporter line
Real-time imaging of the cytosolic pH near the PM was done by using the PM-cytp reporter line in vRootchip containing medium (¼ MS + 0.1% sucrose), and was imaged on the inhouse established vertical Zeiss LSM 800 confocal microscope 24 . Sequential illumination at 488 and 405 nm with emission 514 nm for both, corresponding to the two absorption peaks of pHluorin, were taken with a 20x/0.8 air objective. For each root in the vRootchip, two ROI with one containing the elongating epidermal cell for measuring the cytosolic pH and one containing the root tip for measuring the root growth rate were tracked over time.
Image analysis was performed similar to the HPTS analysis described above.

Imaging microtubule orientation and vacuolar morphology
The pEB1b::EB1b-GFP maker line was used to track the dynamics of CMT orientation in vRootchip. Images were obtained every 6.25 s and the analysis of the CMT orientation was done in ImageJ by max Z-projection on every 10 frames and quantification by a for batch processing modified version of the Fibril Tool macro 26 . The p35S::MAP4-GFP marker line was used for capturing the CMT orientation after treatment for the indicated time period. The CMT orientation angle was processed using the Bioline script 27 . For both marker lines, the GFP (excitation 488 nm, emission 514 nm) signal was detected by Plan-Apochromat 20x/0.8 air objective in the vertical Zeiss LSM 800 confocal microscope 24 .
The pSYP22::SYP22-YFP marker line was used for imaging vacuolar morphology. We used a mounting system 28 , which allows the injection of new liquid medium during imaging. Images were taken before and 30 minutes after Mock or 100 nM IAA treatment in liquid medium and the YFP (excitation 488 nm, emission 527 nm) intensity was detected with C-Apochromat 40x/1.20 W Korr objective in an inverted Zeiss LSM 800 confocal microscope.

Non-invasive microelectrode (MIFE) ion flux measurements
Net H + flux was measured using non-invasive microelectrode ion flux estimation (MIFE) technique essentially as described elsewhere 29 . Roots of intact 6-day-old Arabidopsis WT seedlings were immobilised in a measuring chamber using Perspex holders. The composition of the solution was 0.5 mM KCl and 0.1 mM CaCl 2 ; pH 5.8, unbuffered. After 30-40 minutes of conditioning, the H + microelectrode was positioned 20 μm from the root surface in the elongation zone (~450 μm from the tip) (from which ion fluxes were measured). Steady-state H + fluxes were recorded for 5-10 minutes, and then 10 nM auxin treatment was applied following by another 30-40 minutes of recording. At least 9 individual plants from several batches were used. The sign convention is "influx positive". The resulting time-lapse video was analysed in ImageJ as described previously 23 .

Identification of TMK1-interacting proteins using IP/MS-MS
Immunoprecipitation (IP) experiments were performed in three biological replicates as described previously 30 using 1 g of roots of 7-day-old seedlings from the p35S::TMK1-eGFP transgenic line and 1 g of roots from WT. Interacting proteins were isolated by incubating total protein extracts with 100 μL anti-GFP coupled magnetic beads (Miltenyi Biotech). Three replicates of p35S::TMK1-eGFP were compared to three WT replicates.
Tandem mass spectrometry (MS) on a Q-Exactive device (Thermo Fisher) and statistical analysis using MaxQuant and Perseus software was performed as described previously 31 .

Identification of TIR1-and AFB1-interacting proteins using IP/MS-MS
For immunoprecipitation, ground plant material of pTIR1::TIR1-VENUS in tir1-1 and pAFB1::AFB1-VENUS in afb1-3 transgenic lines was lysed in mild lysis buffer (50 After Stagetip processing, peptides were applied to online nanoLC-MS/MS using a 60 minutes acetonitrile gradient from 8-50%. Spectra were recorded on a LTQ-XL mass spectrometer (Thermo Scientific) according to 31 . Statistical analysis using MaxQuant and Perseus software was performed as described previously 31 .

Phospho-proteomics of auxin-treated roots
Roots from 5-day-old plants were treated and immediately harvested and flash frozen in liquid nitrogen. They then were ground to fine powder in liquid nitrogen. Powder was suspended in SDS lysis buffer (100 mM Tris pH8.0, 4%SDS and 10 mM DTT) and sonicated using a cooled Biorupter (Diagenode) for 10 minutes using high power with 30  After Stagetip processing, peptides were applied to online nanoLC-MS/MS using a 120 minutes acetonitrile gradient from 8-50% for phospho-proteomics. Spectra recording and statistical analysis were as previously described, with the addition of phosphorylation as a variable modification 31 . Filtering of datasets was done in Perseus in as described 32 .

Europe PMC Funders Author Manuscripts
Phospho-proteomics in WT and tmk1-1 roots 4 biological samples of Col-0 WT and tmk1-1 roots were prepared and treated as indicated above. They were submitted to the phospho-proteomic pipeline 31,32 and differentially phosphorylated peptides belonging to H + -ATPases were specifically filtered out of the big dataset (Extended Data Table 1).

Protein extraction and Western blot analysis for co-IP and determination of AHA2 phosphorylation state
To isolate PM H + -ATPases and potential interactors, 5-7 day-old plant roots were harvested at the indicated time points after 10 or 100 nM IAA auxin treatment. 24 hour prior to the evaluation of auxin effects, these seedlings were sprayed with ½ AM solution containing 30 μM kynurenine. The root samples were flash frozen in liquid nitrogen and ground (Retsch mill, 2x 1 minute at 20 Hz). The root powder was then resuspended in a 1:1 (w/v) ratio in protein extraction buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1x Roche cOmplete™ Protease Inhibitor Cocktail, 1x Roche PhosSTOP™, 1 mM EDTA, 1 mM DTT and 0.5 mM PMSF). The samples were incubated on ice for 30 minutes, followed by a centrifuging step at 10,000g to discard the plant debris. The cleared supernatant containing the proteins of interest was collected and the total protein content was determined using Quick Start Bradford reagent (Bio-Rad). This could further be used for co-immunoprecipitation analysis or for SDS-PAGE analysis. In order not to lose relevant proteins, protein samples were not boiled in the presence of reducing Laemmli buffer and no harsher PM extraction or membrane enrichment was attempted.
For co-immunoprecipitation, root extracts (obtained by extraction in the Lysis buffer supplied in the Miltenyi μMACs kit, supplemented with 1x Roche cOmplete™ Protease Inhibitor Cocktail, 1 mM DTT and 0.5 mM PMSF), were incubated with magnetic beads from the Miltenyi anti-GFP, anti-HA or anti-FLAG μMACs kits (depending on the tags of the proteins of interest) and kept rotating for 4 hour at 4°C. Elution was performed with room-temperature denaturing elution buffer and the proteins were analyzed by SDS-PAGE and Western blot.
Following separation of proteins by SDS-PAGE in a 10% acrylamide gel (Protean ® TGX™, Bio-Rad), proteins were transferred to PVDF membranes by electroblotting (Transblot ® Turbo™, Bio-Rad). The membranes were then incubated in blocking buffer (0.05% Tween-20, 5% milk powder or 3% BSA, 20 mM Tris-HCl, pH 7.5 and 150 mM NaCl) for at least 60 minutes and incubated with antibody solution against the protein of interest.

Antibodies
The anti-AHA2 and anti-Thr 947 AHA2 antibody were shared by Toshinori Kinoshita and used as described previously 33 at final dilution of 1:5000 in TBST buffer + 3% BSA, followed by anti-rabbit IgG secondary antibody conjugated to horseradish peroxidase (HRP) (GE Healthcare, NA934) at a dilution of 1:10000 and chemiluminescence reaction (SuperSignal West Femto, Thermo Scientific). To allow multiple antibody detections using the same PVDF membrane, mild stripping was performed using 15 g/L glycine, 1 g/L SDS, 10 mL/L Tween-20 buffer at pH 2.2 for 2-5 minutes.

ATP hydrolysis in root samples
To deplete endogenous auxin levels in the seedlings, 14-day-old plants were pre-treated with 30 μM kynurenine for 24 hour in the dark. Then, the pretreated seedlings were incubated in the presence and the absence of 100 nM indole-3-acetic acid for 60 minutes under dark condition. The roots excised from the seedlings were homogenized in the homogenization buffer (50 mM MOPS-KOH [pH 7.0], 100 mM KNO 3 , 2 mM sodium molybdate, 0.1 mM NaF, 2 mM EGTA, 1 mM PMSF and 20 μM leupeptin) and the homogenates were centrifuged at 10,000 g for 10 minutes; the obtained supernatant was further ultracentrifuged at 45,000 rpm for 60 minutes. The resultant precipitate (microsomal fraction) was resuspended in the homogenization buffer. ATP hydrolytic activity in the microsomal fraction was measured by the release of inorganic phosphate from ATP in a vanadatesensitive manner following the method published 34

Bimolecular Fluorescence Complementation (BiFC)
Following the method described 35 , the full-length coding sequences of AHA2 and TMK1 without stop codons were amplified by PCR (primers in Extended Data Table 4), cloned into pENTR/D-TOPO or pDONR207 and recombined in pSPYNE and pSPYCE 36 to generate BiFC expression constructs. The resulting binary vectors were introduced in Agrobacterium GV3101 by electroporation and these were cultured until OD 600 0.8. Syringe infiltration was performed in Nicotiana benthamiana leaves as described in 37 . For the constructs of interest, final OD 600 of 0.2 was used and p19 was co-infiltrated at OD 600 0.1 to avoid gene silencing. Infiltration buffer of pH5.8 contained: 10 mM MgSO 4 , 10 mM MES-KOH and 0.15 mM acetosyringone. TMK1 overexpression, even transiently, has a strong effect on the viability of the leaves, so samples were taken daily after infiltration to determine the optimal balance between expression level and viable leaf cells. To visualize protein interactions, sections of the leaves were imaged using a Zeiss LSM 700 confocal microscope.

Quantitative RT-PCR
RNA was extracted from 5-day-old light-grown root tips with the RNAeasy Plant Mini Kit (Qiagen), with three biological replicates for each genotype. Two μg of RNA was used for cDNA synthesis (Qiagen). Samples were pipetted in three technical replicates using an automated JANUS Workstation (PerkinElmer) and measured by the Real-time PCR Roche LightCycler 480 using Luna® Universal qPCR mastermix (NEB, M3003S). Primers utilized for assessing gene expression are listed in Extended Table 4. Expression levels were normalized to Elongation factor 1-alpha (At5G60390) 37 .

Statistical analysis
All graphs were generated using GraphPad Prism 6 or 8. For statistical analysis of vRootchip data, Two-way ANOVA was performed for the entire time of x-axis except for indicated ones. Welch ANOVA analysis was applied for the scanner growth assays with multiple time points, and one-way ANOVA assays were used for steady state (one time point) pH and scanner growth datasets. Stars indicate significant differences on all graphs with ns for p>0.05, * for p≤0.05, ** for p≤0.01, *** for p≤0.001 and **** for p≤0.0001. Li   i, PM H + -net influx measured by a non-invasive microelectrode before and after 10 nM IAA treatment in the elongating zone of WT roots. Means of 9 roots ± SD. j-k, The medium pH (j) and apoplastic pH (k) changed rapidly after the exchange of the medium of different pH in vRootchip. Following media were used sequentially: basal medium at pH 5.8, auxin-containing medium at pH 5.8, gradually more acidic medium of pH 5.6, followed by pH 5.4 and lastly again basal medium at pH 5.8. l, Quantification of root growth rate in response to the gradual addition of KOH in the medium in the vRootchip. The greener the shade, the more KOH was added followed by washout with the initial basal pH 5.8 medium. still decreased the apoplastic pH (d) and promoted root growth (e) in presence of IAA in vRootchip. n = 4, the shaded area represents the duration time of the indicated treatments (d-e). Root growth after FC, IAA and co-treatment for 1 hour in steady state by scanner. 1 μM FC and 10 nM IAA were used in (f) while 10 μM FC and 2 nM IAA were used in (g). n > 9 for both graphs, ns p>0.05, *p≤0.05, **p≤0.01, ****p≤0.0001, One-way ANOVA (f-g). h, Dose-response of auxin-induced root growth inhibition of aha single mutants. n > 22.
Relative GR is ratio between auxin-affected growth in the mutant to mock growth for the same genotype. ns p>0.05, **p≤0.01, ***p≤0.001, Welch ANOVA. i, Quantitative Real-time PCR on the AHA1,2,7,11 expression in root tips of AtTAS1c-AHA#2 and #4. The expression level was normalized to EF1α as housekeeping gene. Means of 3 biological replicates + SD.  x-axis) and significance (p-value, y-axis). Red dots are in the range of ratio > 10 and -LOG p-value > 2. P-values are calculated based on the three replicates of 35S::TMK1-GFP vs WT using a Two-sided t-test. c, co-IP of pTMK1::TMK1-GFP roots, followed by Western blot detection of AHA2 and Thr 947 -phosphorylated AHA2. The interaction was not regulated by 30 minutes of 100 nM IAA treatment, but the phosphorylation state of the interacting AHA2 was increased.
Extended Data Table 1 Phospho-proteomic data of rapid auxin effects in root and phospho-proteomic analysis of H + -ATPases in tmk1-1 mutants

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. b, Quantitative analysis of the auxin-induced apoplast alkalinisation in the EZ reported by HPTS dye and corresponding inhibition and recovery of root tip growth rate (GR) following 5 nM IAA and washout as in (a). Means of 4 roots + SD. c, Quantitative analysis of the auxin-induced cytoplasm acidification in the EZ, visualized by the PM-cyto reporter and the corresponding inhibition and recovery of GR upon 5 nM IAA treatment and washout in vRootchip. Means of 3 roots +SD. d-g, Time-lapse of root growth response to a pulse of alkaline (pH 6.15) (d) or acidic medium (pH 5.10) (f). The slope of the white dotted line that tracks the root tip indicates the GR. Quantifications of GR in d (n = 8) (e) and f (n = 7) (g). The shaded areas represent the duration of the indicated treatments. Mean +SD. c, Phospho-proteome detection of auxin-induced AHA2-Thr 947 phosphorylation after 2 minutes of 100 nM IAA treatment. n= 4. Box plot depicts minimum to maximum, mean ± SD. Two-sample t-test (part of MaxQuant-Perseus analysis). **p≤0.01. d, Time-course western blot analysis of auxin-induced Thr 947 -phosphorylated AHA2 levels in roots treated with 10 nM IAA using AHA2 and pThr 947 -specific antibodies. Total AHA2 protein levels were monitored in the same samples as a control. Band intensities of the different lanes are quantified by the Gels Analysis function in ImageJ. e-f, Apoplastic pH (e) and root growth (f) analysis upon pharmacological activation of H + -pumps by 10 μM FC and 10 nM IAA in vRootchip. Apoplastic pH was monitored using HPTS. The shaded area represents the duration of the indicated treatment. Mean of 4 roots for each treatment + SD. p≤0.0001 between IAA and IAA+FC, from 0 -32 min (e), from 0 -31 min (f), Two-way ANOVA. g-h, Apoplastic pH response in two independent AHA knock-down lines (AtTAS1c-AHA#2 and #4) (n > 9) (g), and the ost2-3D gain-of-function allele (n > 5) (h) in response to 30 minutes 5 nM IAA treatment. Apoplastic pH changes are determined by the HPTS dye and IAA treatment was normalized to the corresponding mock-treatment. **p≤0.01, ***p≤0.001, ****p≤0.0001, One-way ANOVA. i, Dose-response of auxin-induced root growth inhibition of AtTAS1c-AHA lines and ost2-3D mutants reveals hypersensitivity and resistance respectively to IAA in comparison to WT (n > 15) (i). The relative GR is calculated by the ratio of GR for each IAA concentration relative to mock-treated GR of the same genotype. *p≤0.05, **p≤0.01, ***p≤0.001, ****p≤0.0001, Welch ANOVA.
g, Western blot detection of the AHA2 levels and its Thr 947 phosphorylation in full seedlings shows reduced AHA2 phosphorylation in tmk1,4 compared to WT. h, Auxin-induced ATP hydrolysis activity is impaired in tmk mutants relative to WT roots (1 hour mock or 100 nM IAA treatment). The IAA-treated sample was normalized to the mock-treated WT. Means of 3 biological replicates + SD. *p≤0.05, ns p>0.05, One-way ANOVA.