An abscisic acid-sensitive checkpoint in lateral root development of Arabidopsis


  • Ive De Smet,

    1. Centre for Plant Sciences, Leeds Institute of Biotechnology and Agriculture (LIBA), School of Biology, University of Leeds, Leeds LS2 9JT, UK,
    2. Vlaams Instituut voor Biotechnologie (VIB), Department of Plant Systems Biology, University of Gent, K. L. Ledeganckstraat 35, B-9000 Gent, Belgium, and
    Search for more papers by this author
  • Laurent Signora,

    1. Centre for Plant Sciences, Leeds Institute of Biotechnology and Agriculture (LIBA), School of Biology, University of Leeds, Leeds LS2 9JT, UK,
    2. Division of Crop Performance and Improvement, IACR-Rothamsted, Harpenden, Hertfordshire AL5 2JQ, UK
    Search for more papers by this author
  • Tom Beeckman,

    1. Vlaams Instituut voor Biotechnologie (VIB), Department of Plant Systems Biology, University of Gent, K. L. Ledeganckstraat 35, B-9000 Gent, Belgium, and
    Search for more papers by this author
  • Dirk Inzé,

    1. Vlaams Instituut voor Biotechnologie (VIB), Department of Plant Systems Biology, University of Gent, K. L. Ledeganckstraat 35, B-9000 Gent, Belgium, and
    Search for more papers by this author
  • Christine H. Foyer,

    1. Division of Crop Performance and Improvement, IACR-Rothamsted, Harpenden, Hertfordshire AL5 2JQ, UK
    Search for more papers by this author
  • Hanma Zhang

    Corresponding author
    1. Centre for Plant Sciences, Leeds Institute of Biotechnology and Agriculture (LIBA), School of Biology, University of Leeds, Leeds LS2 9JT, UK,
      * For correspondence (fax +44 113 343 3144; e-mail
    Search for more papers by this author

* For correspondence (fax +44 113 343 3144; e-mail


Lateral root (LR) formation displays considerable plasticity in response to developmental and environmental signals. The mechanism whereby plants incorporate diverse regulatory signals into the developmental programme of LRs remains to be elucidated. Current concepts of lateral root regulation focus on the role of auxin. In this study, we show that another plant hormone, abscisic acid (ABA), also plays a critical role in the regulation of this post-embryonic developmental event. In the presence of exogenous ABA, LR development is inhibited. This occurs at a specific developmental stage, i.e. immediately after the emergence of the LR primordium (LRP) from the parent root and prior to the activation of the LR meristem, and is reversible. Interestingly, this inhibition requires 10-fold less ABA than the inhibition of seed germination and is only slightly reduced in characterised abi mutants, suggesting that it may involve novel ABA signalling mechanisms. We also present several lines of evidence to support the conclusion that the ABA-induced lateral root inhibition is mediated by an auxin-independent pathway. First, the inhibition could not be rescued by either exogenous auxin application or elevated auxin synthesis. Secondly, a mutation in the ALF3 gene, which is believed to encode an important component in the auxin-dependent regulatory pathway for the post-emergence LR development, does not affect the sensitivity of LRs to ABA. Thirdly, ABA and the alf3-1 mutation do not act at the same developmental point. To summarise, these results demonstrate a novel ABA-sensitive, auxin-independent checkpoint for lateral root development in Arabidopsis at the post-emergence stage. In addition, we also present data indicating that regulation of this developmental checkpoint may require novel ABA signalling mechanisms and that ABA suppresses auxin response in the LRPs.


abscisic acid


lateral root


lateral root primordium


abscisic acid insensitive


indole-3-acetic acid


α-naphthaleneacetic acid


2,4-dichlorophenoxyacetic acid


days after germination.


The root system of most ‘higher’ plants consists of three types of roots, i.e. the main (also known as tap or primary), lateral and adventitious root (Fitter, 1991). The meristem of the main root is formed during embryogenesis. In contrast, the meristems of lateral and adventitious roots are formed post-embryogenically during the lifetime of a plant and their numbers vary according to the age and the growth conditions (Malamy and Benfey, 1997a; Van den Berg et al., 1998).

Lateral roots (LR) arise from a specific differentiated layer of cells encircling the vascular stele, the pericycle. The developmental process begins in most plants with asymmetric transverse divisions of a limited number of pericycle cells adjacent to the two xylem poles of the parent root (Casimiro et al., 2001). Some of the newly divided cells undergo further periclinal divisions leading to the formation of a dome-shaped LR primordium (LRP), which then grows through the outer layers of the parent root and eventually emerges. Soon after the emergence, the LRP is activated and forms a functional LR meristem (Malamy and Benfey, 1997b).

Auxin plays a key role in LR development. It has been known for more than 50 years that auxin stimulates LR formation (Torrey, 1950). Increasing auxin supply via either exogenous auxin application or an enhancement of endogenous auxin synthesis results in a significant increase in the number of LRs (Boerjan et al., 1995; Kares et al., 1990; Klee et al., 1987; Laskowski et al., 1995). A large body of evidence indicates that auxin regulates at least three developmental stages: initiation, LRP establishment, and the activation of LR meristems. Cell divisions in the pericycle represent the first step of LR development and are stimulated by auxin. In Arabidopsis, radish and many other plant species, exogenous auxin application can induce LRP formation from almost all the pericycle cells adjacent to the xylem poles (Blakely et al., 1982; Laskowski et al., 1995).

A requirement for auxin during LRP establishment is demonstrated in experiments with excised Arabidopsis or radish root segments (Laskowski et al., 1995). In the presence of exogenous auxin, LRPs of any developmental stage in the excised segments continued their development to form LRs. However, in the absence of auxin, only primordia that had at least 3–5 cell layers developed into LRs, and no further development occurred in primordia of less than three cell layers. These observations suggested that LR development prior to the 3–5 cell layer stage was auxin dependent and that development beyond this stage was either auxin independent or auxin self-sufficient (i.e. the primordium is capable of synthesising the required amount of auxin de novo).

The phenotype of the aberrant lateral root formation 3 (alf3-1) mutant indicates that auxin is also required for lateral root development beyond the 3–5 cell layer stage. In the absence of exogenous auxin, LRs in alf3-1 arrest soon after emergence from the parent roots. This developmental arrest can be rescued by exogenous IAA or indole, a precursor of IAA biosynthesis. This observation suggests that auxin is required at the post-emergence stage (Celenza et al., 1995).

Two recent studies suggest that auxin may also play an important role in LR emergence (Bhalerao et al., 2002; Marchant et al., 2002). During early seedling development (3–10 days after germination (DAG)) the emergence of the first LR coincides with the formation of the first true leaves (Marchant et al., 2002). The removal of apical tissues prior to the formation of the first true leaves has very little effect on LR initiation but inhibits LR emergence (Bhalerao et al., 2002), indicating that the emergence of LRs requires leaf-derived IAA.

In addition to auxin, many environmental factors also affect LR development. An excellent example is the effect of nutrients (Drew, 1975; Drew and Saker, 1975; Drew et al., 1973; Leyser and Fitter, 1998; Malamy and Ryan, 2001). It has been known for decades that, in soil or growth medium with patchy nutrient distribution, LRs preferentially proliferate in the nutrient-rich zone (Drew and Saker, 1975; Drew et al., 1973). The biological significance of this developmental plasticity is easy to understand as it enables the plant to explore the available nutrients in the surrounding soil environment more efficiently. However, the mechanisms by which plants incorporate such signals into the LR developmental process are poorly understood.

We reported previously that, in Arabidopsis thaliana, nitrate availability affects LR development in two different ways, i.e. a localised stimulatory effect and a systemic inhibitory effect. The localised stimulatory effect acts mainly on the elongation of lateral roots and is mediated by a putative MADS-box transcriptional factor, ANR1 (Zhang and Forde, 1998). The systemic inhibitory effect is only observed at high concentrations of nitrate and occurs at a specific developmental point, immediately after the emergence of the LRP from the parent root (Zhang et al., 1999, 2000). Interestingly, this inhibitory effect was significantly reduced in all known ABA synthesis mutants (aba1-1, aba2-3, aba2-4 and aba3-2) and two ABA-insensitive mutants, abi4 and abi5 (Signora et al., 2001). The results led us to investigate whether ABA plays a direct role in LR regulation. In this paper, we present evidence for the existence of a novel ABA-sensitive, auxin-independent checkpoint during LR development in Arabidopsis.


Exogenous ABA inhibits LR development

The presence of ABA in the growth medium has a dramatic effect on the morphology of Arabidopsis roots. Figure 1 shows a comparison of seedlings grown on an ABA-free (ABA) and an ABA-containing (ABA+, 1.0 µm ABA) medium. On the ABA medium, each seedling produced several visible LRs at 9 DAG. However, no easily visible (with naked eye) LRs were formed in seedlings of the same age grown on the ABA+ medium. This inhibitory effect of ABA is LR-specific, as primary root growth on both the ABA and ABA+ media remained similar.

Figure 1.

A comparison of roots in the ABA-treated and control seedlings showing the inhibitory effect of ABA on LR development.

Arabidopsis Col wild-type seedlings were grown on media containing either no ABA (a) or 1.0 µm ABA (b) at 9 days after germination. Bar = 1.0 cm.

To assess the dosage effect of ABA on LRs, seedlings were grown on a range of ABA concentrations (0, 0.1, 0.25, 0.5 and 1.0 µm). While primary root growth was not affected by ABA within the above range of (0–1.0 µm, Figure 2a) or even higher concentrations (up to 10 µm, data not shown), 0.1 µm ABA was sufficient to cause a 68 or 89% reduction in the number (Figure 2b) or the length (Figure 2c) of visible LRs, respectively. The LR inhibition was further increased with higher ABA concentrations, which resulted in LR length dropping to 0 at 0.25 µm and no LRs being visible at 0.5 µm (which was subsequently used as the standard ABA concentration in the ABA+ medium). For comparison, the ABA dosage effect on seed germination was also measured under our standard growth conditions. Complete inhibition of seed germination (defined as the emergence of the radicle after 5 days of incubation) required at least 3.0 µm ABA, indicating a 10-fold difference in ABA sensitivity between seeds and LRs (Figure 2d).

Figure 2.

ABA dose–response characteristics on LR morphology.

(a–c) The effect of ABA concentrations on the length of primary roots (a), the number (b) and the total length of visible LRs (c) in wild-type (Col) seedlings at 9 DAG. Each data point represents an average of 15 seedlings.

(d) Dose effect of ABA on seed germination. Each data point was calculated from a total of 150 seeds. Error bar = SD.

ABA-induced LR inhibition occurs at a specific developmental stage

The absence of visible LRs in the ABA-treated seedlings suggested that the inhibition took place either prior to or immediately after LR emergence (LRs are not visible to the naked eye immediately after emergence because of the liquid layer surrounding the root surface). To establish when the inhibition actually occurred, we examined the effect of ABA on LRPs at various developmental stages. At first, we checked whether ABA reduced the total number of LRs (including LR initials, LRPs and LRs). The rationale for this was that an inhibition of LR initiation will lead to a reduction of the total number of LRs (including LRPs). In order to make an accurate count of early LRPs, we used a marker line, END199 (Malamy and Benfey, 1997b), in which the β-glucuronidase (uidA) gene is expressed under the control of the 5′-upstream promoter sequence of the SCARECROW gene and through all the developmental stages of LRPs from the two-cell layer stage onwards. As in Col wild-type seedlings, the ABA treatment did not cause significant changes in primary root growth in this marker line (Figure 3a). Both the ABA-treated and the control seedlings produced very similar number of LRs (including LRPs) (Figure 3b) indicating that the ABA treatment did not significantly affect LR initiation in this marker line. We then examined the effect of the ABA treatment on the progression of early LR development in END199 seedlings using a previously described approach (Zhang et al., 1999). This analysis is based on the fact that LRP formation occurs in an acropetal sequence, i.e. LRPs are more developmentally advanced with increasing distance from the root tip. We divided the early LR developmental process into four key stages (for details of the classification of the stages, see the legend of Figure 3). The developmental status of LRPs was assessed under a light microscope and the frequency of LRPs at each stage (per centimetre primary root segment) is presented in Figure 3(c). The results showed that the ABA treatment did not alter the distribution of stage A and stage B LRPs along the primary roots. However, there were proportionally more stage C LRPs (immediately after emergence) and less stage D (longer than 0.5 mm) LRs in the ABA-treated roots in the distal regions. For example, the frequency of stage C LRPs in the 4.1–5, or 5.1–6 cm from tip region is, respectively, 1.6- or 8.5-fold higher in roots grown on the ABA+ medium than in those on the ABA medium (Figure 3c). Consequently, hardly any stage D LRs were developed in seedlings on the ABA+ medium. These results indicate that the ABA-induced LR arrest occurs at a specific developmental stage (during the transition from stage C to D).

Figure 3.

Stage-specific inhibitory effects of ABA on LR development.

Comparisons of primary root length (a), the number of LRs (b) and the distribution of LRPs are shown at each of the four developmental stages (classification of the stages: stage A, initials with up to two layers of cells; stage B, between three layers of cells just prior to emergence; stage C, immediately after emergence but less than 0.5 mm long; stage D, LRs longer than 0.5 mm) (c) of END199 seedlings on media containing either no ABA (open bar) or 0.5 µm ABA (filled bar) at 9 DAG. Each data point represents the average of 15 seedlings. Error bar = SD.

ABA blocks the activation of LR meristems

According to Malamy and Benfey (1997b), the LRP forms a characteristic 8-8-8 cell pattern (8–10 basal cells on either side of the 8–10 central cells) in the layer termed ‘OL1’ (the outermost layer behind the root cap) immediately before the emergence process (see Figure 2i in Malamy and Benfey, 1997b). This cell pattern is invariable and is maintained during the emergence process. After emergence, the cell pattern in the OL1 layer changes to an 8-10+-8 pattern (more than 10 central cells) as a result of divisions near the LR apex (see Figure 2j–l in Malamy and Benfey, 1997b). The number of the central cells in the OL1 layer is therefore a useful indication for the activation ‘status’ of a LR meristem.

To establish whether the ABA-induced arrest occurred before or after meristem activation, we compared the cell patterns in the OL1 layer of the arrested and the non-arrested LRs using median longitudinal sections. Figure 4 shows the sections of a typical ABA-arrested LRP (Figure 4a), a non-arrested LRP at the similar developmental stage (Figure 4b), an activated meristem (Figure 4c) on the ABA medium and the schematic drawings traced from the above sections showing the cellular patterns around the tips (Figure 4d–f). The arrested (Figure 4a,d) LRP had, in the OL1 layer, the typical 8-8-8 cell pattern (with clearly identifiable 8 basal cells and 10 central cells) of a pre-activation LRP (Figure 4b,e) and clearly fewer central cells (10 rather than 18) than the activated LR meristem (Figure 4c,f), indicating that the arrest occurred before the activation of the LR meristem.

Figure 4.

Morphology of ABA-arrested LRPs.

The median longitudinal sections of a typical arrested LRP on the ABA+ (0.5 µm) media (a), a control LRP (on the ABA medium) at the similar stage (b), or a slightly advanced developmental stage (c) than the arrested LRP. The schematic drawings in (d–f) were traced from (a) (the arrested LRP), (b) (the non-arrested LRP) and (c) (the LR with an activated meristem), respectively, to show the cellular patterns around the apex. Cell patterns of the OL1 layer in (d–f) are indicated by the numbers in white (basal cells) and black (central cells). Arrows in (a–f) indicate the boundaries between the basal and central cells. Bar = 0.2 mm.

The ABA-induced LR arrest is reversible

A ‘transferring’ experiment was carried out to test whether the ABA-induced LR arrest was reversible. Seedlings of the Col wild type were first grown on the ABA+ medium for 9 days and then transferred to either the ABA medium or fresh ABA+ medium. The development of LRs on both media was then monitored at 24 h intervals for 3 days (Figure 5). On the ABA+ medium, LRs remained arrested at the end of the 3-day period. However, on the ABA medium, LRs started to grow within 24 h of the transfer indicated by the clearly longer LRs. Most of the LRs reached 3–5 mm within 48 h and were 5–10 mm in length within 72 h following the transfer. These results indicate that the ABA-induced LR inhibition is reversible.

Figure 5.

The reversibility of ABA inhibition of LR development.

Wild-type Col seedlings were initially grown on the ABA+ medium (0.5 µm) for 9 days and then transferred to the ABA (top row) media, or to fresh ABA+ (bottom row) medium. They were photographed at times 0 (left), 24 (middle) and 72 h (right) after the transfer. Close-up images, inserted in the middle column, were obtained under a dissecting microscope. They show the differences in LRs on the ABA and ABA+ media relative to the primary roots 24 h after transfer. Dashed line boxes indicate the region of the primary roots where the images were inserted. Arrows indicate the position of the root tips. Bar = 4 mm.

ABA suppresses the transcription of two cell cycle-related genes, Arath:CycD3;1 and Arath:CDKB1;1, and auxin response in the LRPs

In an attempt to establish the mechanism underlying the LR arrest, we investigated the effect of ABA on cell cycle regulation. We investigated whether the ABA-induced LR arrest was related to the transcriptional regulation of specific cell cycle genes. Promoter-driven β-glucuronidase (GUS) reporter lines were used to examine the effect of ABA on the expression of a number of cell cycle-related genes, including Arath:CDKA;1 (Hemerly et al., 1993), Arath:CDKB1;1 (Burssens et al., 2000; Segers et al., 1996), Arath:CycA2;1 (Burssens et al., 2000), Arath:CycB1;1 (Ferreira et al., 1994) and Arath:CycD3;1 (Soni et al., 1995).

In three of the lines, i.e. Arath:CDKA;1::GUS, Arath:CycA2;1::GUS and Arath:CycB1;1::GUS, the ABA treatment had very little effect on the expression (measured by histochemical staining) of the GUS marker gene (data not shown). In the other two lines, Arath:CDKB1;1::GUS (Figure 6a–f) and Arath:CycD3;1::GUS, the ABA treatment caused a decreased GUS staining. In the Arath:CDKB1;1::GUS line, GUS activity is detectable in LRPs of all developmental stages on the ABA-free medium (Figure 6a–e). The presence of 0.5 µm ABA caused a general reduction of GUS expression in all LRPs and a doubling of the number of LRPs with no detectable GUS activity (Figure 6f).

Figure 6.

The effect of ABA on the expression of Arath:CDKB1;1::GUS and DR5::GUS.

(a–e) The expression of the Arath:CDKB1;1::GUS marker on the ABA medium at the two cell layer (a), five cell layer (b), prior to emergence (c), immediately after emergence (d) and established LR with a functional meristem (e) stages.

(f) The proportion (%) of LRPs/LRs with no detectable GUS staining in the Arath:CDKB1;1::GUS seedlings on the ABA (open bar) or ABA+ (0.5 µm) (grey bar) media at each of the four stages (see the legend of Figure 3 for the classification of the stages). Data were collected and calculated from 12 seedlings on each medium. Error bar = SD.

(g) Expression of the DR5::GUS marker on the ABA (top row) or ABA+ (bottom row) media. Seedlings were examined 9 DAG and stained for 2 h at 37°C for GUS activity. Bar = 0.2 mm.

However, this reduction in GUS activity in the Arath:CDKB1;1::GUS line was not developmentally stage-specific, i.e. occurred at all developmental stages. In the Arath:CycD3;1::GUS (unpublished, gift from J. A. Murray) seedlings, weak GUS activity was detected in the basal 2–3 cell layers of LRPs/LRs of all developmental stages on the ABA medium. In the presence of ABA, more than 80% of LRPs/LRs lost expression of the marker gene in the LRPs/LRs (data not shown). The non-specific nature of the ABA-induced reduction of both Arath:CDKB1;1::GUS and Arath:CycD3;1::GUS indicates that the reduced expression of these two genes is unlikely the direct cause of the stage-specific LR arrest.

Examination of genomic sequences of the Arath:CycD3;1 or Arath:CDKB1;1 genes revealed the presence of a known auxin responsive element, TGTCTC, in the 5′-upstream region of both genes (about 2.0 or 2.7 kb from the first ATG codon of Arath:CycD3;1 or Arath:CDKB1;1 gene, respectively) that was absent from the other three genes. This finding led to the hypothesis that the down-regulation of Arath:CDKB1;1::GUS and Arath:CycD3;1::GUS in the LRPs could reflect a general suppression of the auxin response by ABA in the LRPs. To test this hypothesis, we examined the effect of ABA on the expression of an auxin-responsive marker, DR5::GUS, which contains several copies of a synthetic auxin-responsive element (TGTCTC) fused to a 35S minimum promoter and the GUS encoding sequence (Ulmasov et al., 1997). The expression of GUS marker in this line is known to reflect the level of auxin response (including both the level of and the sensitivity to auxin) in plant cells (Zhao et al., 2001). In the absence of ABA, GUS activity was detectable (by histochemical staining) in some LRPs of all developmental stages (Figure 6g, top row). However, in the presence of 0.5 µm ABA, no GUS activity was detectable in LRPs (Figure 6g, bottom row). The absence of the DR5::GUS expression was found at all developmental stages. This result demonstrates that ABA has a negative effect on the auxin response in the LRPs.

The inhibitory effect of ABA on LRs is likely to be mediated by an auxin-independent pathway

Arrested LR development in ABA-treated wild-type seedlings and in alf3-1 appeared to occur at a very similar developmental stage. As auxin rescues the LR arrest in alf3-1 (Celenza et al., 1995), it is important to establish whether the ABA-induced LR arrest is mediated by auxin. To this end, we examined whether exogenous auxin could rescue the ABA-induced LR inhibition. Col seedlings were grown on media containing NAA alone (Figure 7a,c) or NAA plus 0.5 µm ABA (Figure 7b,d). Two concentrations of NAA were used, 100 nm (a subtoxic level, Figure 7a,b) and 200 nm (a toxic level, Figure 7c,d). ABA-induced LR arrest could not be rescued by NAA at either concentrations (Figure 7b,d). Similar results were obtained with other auxins, 2,4-D (at 50 and 100 nm), IAA (at 100 and 200 nm), or indole (at 20 and 40 µm). Neither the auxins nor indole could rescue the ABA-induced LR arrest phenotype (data not shown).

Figure 7.

The effect of auxin on ABA-induced LR inhibition in wild type (Col) and the superroot1 (sur1) mutant and the expression of DR5::GUS in the arrested LRPs.

(a–d) Col wild-type seedlings grown on media containing (a) 100 nm NAA (b) 100 nm NAA + 0.5 µm ABA (c) 200 nm NAA or (d) 200 nm NAA + 0.5 µm ABA.

(e,f) sur1 seedlings grown on media containing no ABA (e) or 0.5 µm ABA (f). Seedlings were at 9 (a, b), 12 (c, d), or 20 (e, f) DAG. Bar = 5 mm.

(g–i) The effect of ABA on the expression of the DR5::GUS marker in the LRPs. Seedlings were grown on the ABA(g) and ABA+(h) media (at 10 DAG), or on ABA+ medium for 9 days and then transferred to the ABA+ medium supplemented with 100 nm NAA (i) for 24 h. Seedlings in (g–i) were stained for GUS activity for 4 h at 37°C.

We also tested whether elevated endogenous auxin synthesis could rescue the ABA-induced LR arrest in the sur1 (superroot1) mutant (Boerjan et al., 1995) which is known to overproduce auxin. The morphology of a sur1 seedling grown on the ABA and the ABA+ medium is presented in Figure 7e and Figure 7f, respectively, showing that the elevated auxin synthesis in the mutant did not overcome the ABA-induced LR arrest.

Since we also observed that ABA has a negative effect on auxin response, one possible explanation for the failure of auxin to rescue the arrest could be that the LRPs were unable to respond to the increased auxin supply in the presence of ABA. To test this, we first grew the DR5::GUS seedlings on the ABA+ medium for 9 days and then transferred some of them to a medium containing both 0.5 µm ABA and 100 nm NAA for 12 h. As shown before, the arrested LRPs in the seedlings on the ABA medium showed clear GUS staining (Figure 7g), while the ones on the ABA+ medium had no GUS staining (Figure 7h). However, the addition of 100 nm NAA to this medium could restore GUS expression in the LRPs (Figure 7i), even though it could not rescue the LR arrest. This result indicates that the arrested LRs are able to respond to auxin in the presence of ABA and that the LR arrest is likely mediated by an auxin-independent pathway. Consistent with the above conclusion is the observation that 200 nm NAA caused similar inhibition on primary root growth on both the ABA and ABA+ media (comparing Figure 7 a,b with Figure 7 c,d), resulting in a 3.5-fold reduction of primary root length in both cases. This observation indicates that 200 nm NAA can overcome the suppression of ABA and cause specific morphological response such as root inhibition. The failure to rescue the arrest by 200 nm NAA, therefore, also implies that the arrest is auxin independent.

We also investigated whether the LRs of alf3-1 seedlings were sensitive to ABA. The mutant seedlings were grown on three different media: ABA, ABA plus 50 nm 2,4-D or ABA+ plus 50 nm 2,4-D. On the ABA medium, the mutant seedlings produced short primary roots and stunted lateral roots (Figure 8a). The presence of 50 nm 2,4-D restored normal growth in both the primary and lateral roots in more than 50% of the seedlings (Figure 8b). When 50 nm 2,4-D was added in combination with 1.0 µm ABA, primary root growth remained similar to that observed on media with 2,4-D alone. In marked contrast, LR development was inhibited (Figure 8c), indicating that LR development in this mutant remains sensitive to ABA.

Figure 8.

ABA-induced LR inhibition in the alf3-1 mutant.

(a–c) alf3-1 seedlings grown on the ABA(a), ABA plus 50 nm 2,4-D (b), or 0.5 µm ABA plus 50 nm 2,4-D (c) media at 9 DAG. Bar = 5 mm.

(d–i) The images show the whole mount (d, g), FDA staining (e, h) and median longitudinal sections (f, i) of a typical arrested LR on either ABA (d–f) or ABA+ (g–i) media. Bar = 0.1 mm.

The relative timing of the two LR arrests caused by ABA and the alf3-1 mutation was also investigated using a vital staining. This approach is based on the following: (i) the ABA-induced arrest is reversible (Figure 6) indicating that arrested LRPs are still alive; (ii) Celenza et al. (1995) showed that the arrested LRs in alf3-1 were no longer viable. If alf3-1 seedlings were grown on the ABA+ medium, the viability of the arrested LRs would indicate which of two inhibitions occurred first. For example, if ABA acted earlier than the alf3 mutation, the arrested LRs would be viable. On the other hand, if ABA acts after the alf3-1 mutation, the arrested LRs would be non-viable. The alf3 seedlings were grown on the ABA+ and ABA medium for 9 days and then stained with FDA. Figure 8(d,g) shows the whole mount images of the arrested LRs on the ABA (Figure 8d) and ABA+ (Figure 8g) media. The arrested LRs on both media are morphologically very similar, except that the arrested LRs on the ABA medium were usually slightly bigger in size than those on the ABA+ medium. However, the arrested LRs on the two media showed a striking difference in cell vitality as indicated by the FDA staining. On the ABA medium, most cells (except a few in the centre) in the arrested LRs had no FDA staining (Figure 8e). On the other hand, strong FDA staining was observed in the LRPs arrested on the ABA+ medium (Figure 8g). All LRPs at the pre-emergence stages showed strong FDA staining on both the ABA and ABA+ media (data not shown), indicating that the cell death in the arrested LRs on the ABA medium probably occurred after the arrest. These results indicate that ABA-induced LR arrest is likely to occur before the effect of the alf3-1 mutation.

Median longitudinal sections revealed that there was an unusual group of smaller cells with much thickened cell walls in the centre of the arrested LRs on the ABA medium (Figure 8f). There was often a layer of crushed cells surrounding this central cell group, indicating that these cells were produced after the arrest, and therefore crushed the cells in the nearest outer layer. On the ABA+ medium, the arrested LRs in alf3-1 seedlings did not have this layer of small cells in the centre and were morphologically similar to those in the ABA-treated wild-type seedlings (Figure 8i). The above observation provided further morphological evidence to support the hypothesis that the LR arrests induced by ABA and the alf3-1 mutation are likely mediated by different mechanisms.

ABA-induced LR arrest occurs in existing abi mutants

To establish whether ABA-induced LR inhibition is mediated by previously characterised components in ABA signalling (ABI1-5, ERA1) (Cutler et al., 1996; Finkelstein and Lynch, 2000; Finkelstein et al., 1998; Giraudat et al., 1992; Leung et al., 1997), the inhibitory effect of ABA was tested in six ABA response mutants, including abi1-1, abi2-1, abi3-1, abi4-1, abi5-1 and era1-2. None of the above mutants produced clearly visible LRs on 0.5 µm ABA+ at 10 DAG, suggesting that the genes which are mutated in the above lines may not play a prominent role in the ABA-induced LR arrest. Nevertheless, experiments with lower ABA concentrations showed a slight reduction in ABA sensitivity in some abi lines (Figure 9). For example, two of the abi mutants, abi1-1 and abi3-1, produced short but visible LRs on 250 nm ABA, while no visible LRs were produced in Ler seedlings grown on this medium (Figure 9a). The reduced LR ABA sensitivity is more evident when ABA concentration is further reduced (Figure 9b). The presence of 10 or 25 nm ABA caused a 49 or 69% reduction in the mean total LR length in the wild type. These figures were reduced to 12 or 40%, 10 or 60% and 40 or 50% in the abi1-1, abi2-1 and abi3-1 mutants, respectively. However, statistic analysis (t-test) indicated that only four data points, abi1-1 at 10 (P = 0.018) and 25 nm (P = 0.027), abi2-1 at 10 nm (P = 0.020) and abi3-1 at 25 nm (P = 0.042), showed significant difference from the Ler control. For all the other data points, the differences between the mutants and Ler control are not statistically significant (i.e. P > 0.05). The dosage responses of abi4-1 and abi5-1 (data not shown) showed some reduced ABA inhibition on LR length when compared with their wild-type controls (Col for abi4-1 and Ws for abi5-1), but the reduction is not statistically significant in both mutants at the above ABA concentration range. These results suggest that some of the ABI proteins, i.e. ABI1, ABI2, ABI3, may play some role, but not a prominent one, in the ABA-induced LR inhibition.

Figure 9.

ABA-induced LR inhibition in the abi mutants.

(a) Morphology of Ler wild type, abi1-1, abi2-1 and abi3-1 seedlings on 0.25 µm ABA at 9 DAG. Bar = 5 mm.

(b) ABA dosage effect on LR length in the Ler wild type (●), abi1-1(□), abi2-1(○) and abi3-1 (▵) seedlings (11 DAG). To calculate the relative LR length, the LR length on the ABA medium of each line was defined as 100% and used to divide the LR length on the ABA-containing media. Each point represents data from 20 seedlings. Points marked with an asterisk differed significantly from Ler control (P < 0.05 in a t-test).

The era1-2 mutants produced much less visible lateral root on both ABA+ and ABA media (data not shown), but did not show any significant alteration in the ABA sensitivity of LRs (data not shown). Recently, it has been revealed that high ABA concentrations (>1.0 µm) inhibit primary root growth and this inhibition requires functional ethylene response pathways (Beaudoin et al., 2000; Ghassemian et al., 2000). We examined the ABA dose–response in the ein2-1 and etr1-3 mutants and did not observe any clear difference in LR ABA sensitivity between the mutants and their wild-type (Col) control (data not shown).


The inhibition of LR development represents a novel function of ABA

ABA was first described as a ‘growth-inhibiting’ hormone in the 1960s because of its functions in bud dormancy and fruit abscission (Salisbury and Ross, 1991). Extensive studies in the past few decades have revealed many more functions of this hormone in a wide range of processes, including embryo and seed development, seed and shoot dormancy, senescence, stress responses and drought tolerance (Giraudat, 1995; Leung and Giraudat, 1998). The inhibitory effect of ABA on LR development reported in this study represents a novel ABA function.

Interestingly, we noticed that ABA did not completely prevent the radicles from breaking through the seed coat, but rather to slow down this process and to suppress the subsequent growth of the radicles (unpublished data). This observation suggests that the main effect of ABA on both seed germination and LR development is quite similar, i.e. to prevent the roots from active growth immediately after their emergence from the parent roots or the seeds. However, there is a 10-fold difference in ABA sensitivity between seeds and LRs. The higher ABA requirement for inhibiting seed germination is not due to a sheltering effect of the seed coat, as longer incubation (4 days) in ABA solution at 4°C prior to germination did not alter the sensitivity of seeds to ABA (data not shown).

The post-emergence activation of LR meristems is an important developmental checkpoint

Based on the auxin requirement for LR development in excised root segments, Laskowski et al. (1995) proposed that LR development has two distinct developmental stages: an auxin-dependent stage (LRP formation) and an auxin-autonomous stage (the formation of a meristem). The boundary between the two stages lies at the 3–5 cell layer stage. After this stage, according to their model, LRPs behave as a meristem and are capable of maintaining lateral root growth.

The phenotype of alf3-1, however, indicates a developmental checkpoint beyond the 3–5 cell layer stage, i.e. after the emergence of LRPs from the parent roots. The morphological analysis described by Malamy and Benfey (1997b) also suggested the presence of a developmental checkpoint at the post-emergence stage. They observed a sharp reduction of mitotic activity in LRPs prior to and during the emergence process. Active LR meristems are established after emergence. The activation process coincides with a resumption of cell division of the initials. A recent report also shows that the emergence of LRPs is regulated by shoot-derived IAA (Bhalerao et al., 2002). The data presented in this article, together with our previous observations concerning the inhibitory effect of high nitrate supplies (Zhang et al., 1999, 2000), support the existence of a developmental checkpoint at the post-emergence stage.

The inhibitory effect of ABA reveals a novel regulatory pathway for LR development

The LR arrests in the ABA-treated seedlings and alf3-1 take place after the emergence step. Nevertheless, our results show that these two types of arrest are unlikely mediated by the same mechanism. First, we showed that exogenous auxin application, which rescues the arrest in alf3-1, and elevated auxin synthesis could not rescue the ABA-induced LR arrest. We also presented evidence showing that the failure of elevated auxin to rescue the ABA-induced arrest is not because the LRPs can not respond to auxin. Secondly, the LR arrest in alf3-1 is accompanied by an inhibition of primary root growth, while the ABA treatment does not affect the growth of primary roots. Thirdly, the alf3-1 mutation does not alter the response of LRs to ABA. In this case, LRs remain sensitive to the ABA-induced inhibition. In addition, we have also shown that ABA and the alf3-1 mutation did not act at the same point of LR development. We therefore conclude that ABA regulates a novel, auxin-independent developmental checkpoint in LR development.

The ABA-induced LR inhibition may require novel ABA-signalling components

The differential ABA sensitivities of LRs and seeds suggest that different signalling pathways are likely to mediate the two types of ABA-induced inhibition in LRs and seeds. The fact that the ABA-induced LR arrest was not rescued in all the abi mutants tested (abi1-1, abi2-1, abi3-1, abi4-1 and abi5-1) at 0.5 µm ABA also supports the above hypothesis. In addition, we have isolated several mutants which can produce visible LRs on 0.5 or 1.0 µm ABA, but are sensitive to ABA during seed germination (data not shown). Nevertheless, our data do suggest that some of the characterised ABI proteins, i.e. ABI1, ABI2 and ABI3, play some role in the LR inhibition, as mutations in their encoding sequences result in a slight reduction in LR ABA sensitivity.

Cell cycle regulation and the ABA-induced LR arrest

The LR arrest undoubtedly involves an inhibition of mitotic activity in the LRP. It has also been reported that ABA inhibits DNA replication in tobacco suspension cells (Swiatek et al., 2002) and induces the expression of a putative Arabidopsis cyclin-dependent kinase inhibitor, ICK1 (Wang et al., 1998). Surprisingly, none of the cell cycle regulatory genes examined in this study are transcriptionally switched off in the arrested LRPs. There are two facts arguing against a direct link between the observed downregulation of the Arath:CDKB1;1 and Arath:CycD3;1 genes by ABA and the LR arrest. First, the arrest takes place in all LRPs, but complete downregulation (of both Arath:CDKB1;1 and Arath:CycD3;1 genes) only occurs in a small proportion of LRPs. Secondly, both Arath:CDKB1;1 and Arath:CycD3;1 genes are regulated by auxin, but the LR arrest is auxin independent. The downregulation of these two genes likely reflect a general suppression of auxin response in the LRPs. Obviously, the cell cycle machineries are inhibited in the arrested LRPs. Our results suggest that this inhibition either involves cell cycle genes not examined in the present study or is mediated by the examined genes via post-transcriptional regulation.

In conclusion, we propose a model to summarise our current understanding of the regulation of early LR development (Figure 10). There are five main checkpoints during early LR development regulating the initiation, LRP establishment, emergence, meristem activation and maintenance of meristem function. Auxin plays a stimulatory role at four of the checkpoints, the first (Boerjan et al., 1995; Kares et al., 1990; Klee et al., 1987; Laskowski et al., 1995), second (Laskowski et al., 1995), third (Bhalerao et al., 2002) and fifth (Celenza et al., 1995). ALF4 and ALF3 mediate the regulatory roles of auxin in the first (initiation) and the fifth (maintenance of meristem function) checkpoint, respectively (Celenza et al., 1995). The fourth (meristem activation) checkpoint is sensitive to both ABA and high nitrate. The regulatory role of high nitrate is mediated via both an ABA-dependent and an ABA-independent pathway (Signora et al., 2001). It is likely that the regulation of LR development involves cell cycle regulation, especially at the initiation, LRP establishment and meristem activation steps. In addition to the regulatory role at the fourth checkpoint, ABA may also affect LR development by acting on the other checkpoints indirectly, i.e. through a suppression of auxin response.

Figure 10.

A model for the regulation of early LR development.

Solid lines: confirmed regulatory paths; broken lines, unconfirmed (speculative) regulatory paths; (bsl00120): stimulation; (⊥): inhibition. The dotted lines represent the surface of the parent root and the star symbol indicates the activated status of the meristem. There are five checkpoints during early LR development at the initiation, LRP establishment, emergence, meristem activation and maintenance of meristem function steps. Auxin regulates four of the checkpoints, i.e. the first (initiation), second (LRP establishment), third (emergence) and fifth (maintenance of meristem function). ALF4 and ALF3 mediate the regulatory roles of auxin at the first and fifth checkpoint, respectively. High nitrate and ABA regulates the fourth (meristem activation) checkpoint. The inhibitory role of high nitrate on LR is mediated via both an ABA-dependent and an ABA-independent pathway. A link between meristem activation and specific cell cycle regulation has not yet been established.

Experimental procedures

Media and growth conditions

The basic mineral composition of all the growth media used in this study consisted of 100 µm KCl; 40 µm MgSO4, 20 µm CaCl2, 22 µm NaH2PO4, 0.9 µm MnSO4, 0.09 µm KI, 0.97 µm H3BO4, 0.14 µm ZnSO4, 2 nm CuSO4, 20.6 nm Na2MoO4, 2.1 nm CoCl2, 3.6 µm Fe-EDTA and 0.5 g l−1 MES. Unless otherwise stated, the standard ABA medium contained the basic mineral composition mentioned above, plus 0.1 mm KNO3 and 5 g l−1 sucrose. The standard ABA+ medium contained all the components of the ABA medium plus 0.5 µm ABA (± -cis-, trans-abscisic acid, Sigma, A-1049). All growth media were adjusted to a pH value 5.7 using 1.0 m KOH, solidified with 1% agar–agar (Fisher Chemicals, A/1080/53) and autoclaved at 121°C for 20 min. ABA was added to media prior to autoclaving. Seedlings were grown vertically on the surface of the agar-solidified media in a growth room at 20°C with a 16 h/8 h light/dark regime with overhead lighting. Root lengths were measured on plates using a stereoscopic binocular and a graded scale.

Plant materials

Following are the mutant lines and their genetic background (in bracket) used in this study : abi1-1(Ler), abi2-1(Ler), abi3-1(Ler), abi4-1 (Col), abi5-1 (Ws), alf3-1 (Col), era1-2 (Col), ein2-1 (Col) and etr1-3 (Col). The genetic backgrounds of the marker lines used in the present study are as following: END199 (Ws), DR5::GUS (Col), Arath:CycD3::GUS (Col), Arath:CDKA;1::GUS (C24), Arath:CDKB1;1::GUS (C24), Arath:CycA2;1:: GUS (C24), and Arath:CycB1;1::GUS (C24).

Histochemical staining of GUS activity

The histochemical stain for GUS reporter activity was carried out as previously described by Beeckman and Engler (1994). Briefly, seedlings were incubated in 90% acetone for 30 min at 4°C, followed by a rinse in the base buffer (0.1 m Tris, 50 mm NaCl, pH 7.0) and subsequent incubations first in the pre-staining buffer (base buffer plus 2.0 mm K3Fe(CN)6, pH 7.0) for 30 min at 37°C and then in the staining buffer (pre-staining buffer plus 1 mg ml−1 5-Bromo-4-Chloro-3-Indolyl-a-d-Glucuronide, X-Glu) for 2 h (unless otherwise stated) at 37°C. The seedlings were then transferred to the base buffer to terminate the reaction at 4°C.

Microscopy and sectioning

For anatomical analysis, seedlings were fixed in 1% (w/v) glutaraldehyde and 4% (w/v) paraformaldehyde dissolved in a 50-mm phosphate buffer (pH 7.0). Roots were embedded in Technovit 7100 according to Beeckman and Viane (2000), cut into 5–6-µm sections and stained with 0.05% rutheniumred (Sigma). The sections were mounted in DePeX (BDH) and photographed using either a Leica DMLB microscope equipped with a Leica MPS-30 camera or a Zeiss Axiovert 135TV microscope with the Openlab 2 imaging system (Improvison Ltd). The images were processed using either Photoshop 5 or Microsoft Photo Editor.

Vital staining of roots

The viability staining procedure was adopted from Celenza et al. (1995). Seedlings were incubated in 2 mg ml−1 fluorescein diacetate (FDA) solution for 30 min and rinsed four times in H2O. The roots were mounted in 50% glycerol and 0.01% Triton X-100 on a microscope slide and imaged using a Leitz DM RB fluorescence microscope and the leica qwin image analysing software.

Photography of seedlings

Arabidopsis seedlings were photographed on the agar plates (on which they were grown) using a Nikon COOLPIX950 digital camera. The images were processed using the Microsoft Photo Editor.


We thank Dr Finkelstein of University of California at Santa Barbara for seeds of the abi1-1, abi2-1, abi3-1 and abi5-1, Dr McCourt of University of Toronto for seeds of era1-2, Dr Celenza for seeds of alf3-1, Dr Guilfoyle of University of Missouri for the DR5::GUS (provided through Prof. Scheres of Utrecht University) and Dr Murray of Cambridge University for the Arath:CycD3;1::GUS marker lines. Thanks are also due to the Nottingham Arabidopsis Stock Centre for supplying seeds of Arabidopsis thaliana ecotype Columbia (Col-0), Landsberg (Ler-1), Wassilewskija (Ws) and the abi4-1 mutant. We acknowledge financial support of the University of Leeds for a research fellowship to HZ and the Royal Society for a research grant (HZ) and EU for a Marie Curie Postgraduate Training Fellowship (IDS).