Abscisic acid regulates root growth under osmotic stress conditions via an interacting hormonal network with cytokinin, ethylene and auxin

Summary Understanding the mechanisms regulating root development under drought conditions is an important question for plant biology and world agriculture. We examine the effect of osmotic stress on abscisic acid (ABA), cytokinin and ethylene responses and how they mediate auxin transport, distribution and root growth through effects on PIN proteins. We integrate experimental data to construct hormonal crosstalk networks to formulate a systems view of root growth regulation by multiple hormones. Experimental analysis shows: that ABA‐dependent and ABA‐independent stress responses increase under osmotic stress, but cytokinin responses are only slightly reduced; inhibition of root growth under osmotic stress does not require ethylene signalling, but auxin can rescue root growth and meristem size; osmotic stress modulates auxin transporter levels and localization, reducing root auxin concentrations; PIN1 levels are reduced under stress in an ABA‐dependent manner, overriding ethylene effects; and the interplay among ABA, ethylene, cytokinin and auxin is tissue‐specific, as evidenced by differential responses of PIN1 and PIN2 to osmotic stress. Combining experimental analysis with network construction reveals that ABA regulates root growth under osmotic stress conditions via an interacting hormonal network with cytokinin, ethylene and auxin.


Construction of hormonal crosstalk networks under osmotic stress conditions
We previously developed a hormonal interaction network for a single Arabidopsis cell by iteratively combining modelling with experimental analysis (Liu et al., 2010). We described how such a network regulates auxin concentration in the Arabidopsis root, by controlling the relative contribution of auxin influx, biosynthesis and efflux; and by integrating auxin, ethylene and cytokinin signalling. Recently, we have developed this hormonal interaction network to include PIN1 or PIN2 activities in a single Arabidopsis cell (Liu et al., 2013, and moved on to study the spatiotemporal dynamics of hormonal crosstalk in a multi-cellular root structure (Moore et al., 2015). Here we show that, after now incorporating abscisic acid (ABA) into the existing hormonal crosstalk network, a novel hormonal crosstalk network for osmotic stress conditions can be constructed. Fig. 7 in the main text describes how ABA, cytokinin, ethylene, auxin, PIN1 and AUX1 interplay in a single stele cell under osmotic stress conditions. The rationale for the network construction is now described.

Abscisic acid and ethylene biosynthesis and crosstalk
Contemporary and classic studies show that ABA biosynthesis increases under osmotic stress, and is responsible for many stress responses (Wright & Hiron, 1969;Zhang & Davies, 1987;Bray, 1997;Lee et al., 2006;Verslues & Bray, 2006;Jones et al., 2014;Waadt et al., 2014). This ABA increase is larger in shoot tissues than roots (Christmann et al., 2005) but is important for root growth under stress. There are several putative osmosensors (Urao et al., 1999;Reiser et al., 2003;Wohlbach et al., 2008;Kumar et al., 2013;Yuan et al., 2014) but the full signalling pathway leading to increased ABA biosynthesis is unknown. Therefore, we assume that ABA increases in response to a more negative external osmotic pressure (Fig. 7 in the main text).
There are several reports of increased ethylene biosynthesis under osmotic stress and ethylene signalling has been shown to be important in many drought stress responses (Spollen et al., 2000;Skirycz et al., 2011;Cheng et al., 2013;Cui et al., 2015). Moreover, ABA and ethylene can act either antagonistically or synergistically to affect root growth and development, but phenotypic analysis of ethylene-ABA mutant crosses reveals little crosstalk between the signalling cascades (Cheng et al., 2009). Furthermore, it has been proposed that ABA represses ethylene biosynthesis to help maintain root growth under stress (Spollen et al., 2000;Sharp, 2002;Li et al., 2011).
Although ABA represses the expression of ethylene response genes such as ERF1, as well as preventing ethylene-induced quiescent centre cell division, there is now growing evidence that ABA promotes ethylene biosynthesis to inhibit root growth (Ortega-Martinez et al., 2007;Zhang et al., 2010;Cheng et al., 2013;Luo et al., 2014). An intact ethylene signalling cascade is required for ABA inhibition of root growth and this requires ethylene-induced basipetal auxin transport components such as PIN2 and AUX1 (Beaudoin et al., 2000;Ghassemian et al., 2000;Luo et al., 2014;Thole et al., 2014).
A combination of these experimental and other observations indicates that ABA can promote ethylene biosynthesis but negatively regulates aspects of its response downstream of the main signalling cascade, such as PIN1 and ERF1 gene expression. In addition, ethylene insensitive/deficient mutants have been shown to hyperaccumulate ABA, so it is possible that ethylene inhibits ABA biosynthesis, providing a negative feedback (Beaudoin et al., 2000;Ghassemian et al., 2000;Wang et al., 2007;Cheng et al., 2009;Dong et al., 2011).

Regulation of auxin transport by both ABA and ethylene
Whilst it has been reported that endogenously applied ABA can increase PIN2 protein levels and basipetal auxin transport in the root (Xu et al., 2013), other work indicates that PIN1, PIN2 and AUX1 levels are reduced by ABA, probably through changes in PLETHORA gene expression (Belin et al., 2009;Shkolnik-Inbar & Bar-Zvi, 2010;Yang et al., 2014). Our experimental data show that AUX1 expression is repressed under osmotic stress (Fig. 5). Thus, in the hormonal crosstalk network (Fig. 7 in the main text), we consider AUX1 expression to be repressed by ABA.
In addition, the overriding effects of ABA on the regulation of PIN1 by ethylene imply that ABA acts on components downstream of ethylene signalling. We therefore assume that ABA negatively regulates PIN1 downstream of the ethylene signalling cascade (Fig. 7 in the main text). This overriding mechanism would allow PIN1 levels to decrease under osmotic stress, even as ethylene levels increase.
However, the regulation of auxin efflux carriers by ABA is tissue-specific. PIN1, which is expressed in the stele, and PIN2, which is expressed in the epidermis/cortex cells, differentially respond to osmotic stress (Fig. 5 in the main text). Moreover, PIN2 protein accumulation can increase under osmotic stress due to the ethylene response (Fig. 5 in the main text), implying that ABA does not override the regulation of PIN2 by ethylene. Therefore, the crosstalk between PIN2, ABA and ethylene is different from the crosstalk between PIN1, ABA and ethylene, as described in Fig. S4. In addition, decreased PIN1 protein levels and increased PIN2 protein levels ( Fig. 5 in the main text) can explain reduced meristem auxin levels under osmotic stress.

Cytokinin, abscisic acid and osmotic stress
Drought and ABA negatively affect trans-zeatin-type cytokinin levels by modulating expression of cytokinin biosynthesis/metabolism enzymes (Dobra et al., 2010;Nishiyama et al., 2011). As it is unclear whether osmotic stress affects cytokinin levels directly or through ABA signalling, our networks (Figs 7 (in the main text), S4) have included both effects that can limit cytokinin biosynthesis.
Cytokinin deficient/insensitive mutants display reduced ABA levels but increased ABA sensitivity, and drought induction of ABA biosynthesis has been shown to be similar to wildtype (Nishiyama et al., 2011). Lower basal levels of ABA could either be due to an increase in auxin and ethylene signalling in these mutants, suppressing ABA biosynthesis or cytokinin could be directly regulating ABA biosynthesis. Thus, no direct regulation of ABA biosynthesis by cytokinin is considered in the hormonal crosstalk networks (Figs 7 (in the main text), S4).

Gene expression
Expression of the relevant genes we have studied has been included in the hormonal crosstalk networks (Figs 7 (in the main text), S4). These genes are either hormone signalling reporter constructs (DR5 and DII VENUS for auxin) or the components important for hormonal crosstalk (PIN1, PIN2, AUX1 and PLS). Inclusion of gene expression enables us to establish the relationship between hormones, gene expression and osmotic stress.

Hormonal crosstalk networks for a single stele cell and a single epidermis/cortex cell
By integrating the biological knowledge described above with the hormonal crosstalk network we previously developed, a hormonal crosstalk network can be constructed for a single stele cell ( Fig. 7 in the main text) and a single epidermis/cortex cell (Fig. S4), respectively. The hormonal crosstalk networks describe how ABA, cytokinin, ethylene, auxin, PIN1 or PIN2, and AUX1 interplay in a single cell under osmotic stress conditions. The key difference between Fig. 7 in the main text and Fig. S4 is ABA could override ethylene induction of PIN1 gene expression, whilst still allowing PIN2 expression to increase.