Osmotic regulation of root system architecture


(fax +1 773 702 9270; e-mail jmalamy@bsd.uchicago.edu).


Although root system architecture is known to be highly plastic and strongly affected by environmental conditions, we have little understanding of the underlying mechanisms controlling root system development. Here we demonstrate that the formation of a lateral root from a lateral root primordium is repressed as water availability is reduced. This osmotic-responsive regulatory mechanism requires abscisic acid (ABA) and a newly identified gene, LRD2. Mutant analysis also revealed interactions of ABA and LRD2 with auxin signaling. Surprisingly, further examination revealed that both ABA and LRD2 control root system architecture even in the absence of osmotic stress. This suggests that the same molecules that mediate responses to environmental cues can also be regulators of intrinsic developmental programs in the root system.


Most growth and differentiation in plants occurs post-embryonically. A plant germinates with only a very rudimentary root and shoot system – a single embryonic root and a pair of embryonic leaves. The extensive roots, branches, leaves and flowers of a mature plant are added as the plant grows. Hence, plants have the unique potential to incorporate information from the environment into the developmental decisions that determine their morphology. Although the patterning of the organs themselves is genetically pre-programmed, a plant can optimize its body plan by modifying organ size and shape, producing more or less organs, and placing these organs in proximity to nutrients and light, and away from unfavorable conditions. This developmental plasticity is of critical importance to the sessile plant, allowing it to survive non-ideal conditions and compete for resources (Callahan et al., 1997; Grime et al., 1986; Malamy, 2005).

The root system is formed through a reiterative program in which lateral roots are produced along the length of other roots. Despite this apparent simplicity, plant species show a wide variation in the architectural form of the root system (Kramer and Boyer, 1995). Furthermore, even genetically identical plants can have vastly different root system morphologies, due to responses to the unique microenvironment each plant encounters as it grows. Soil type, moisture and nutrients all strongly influence the architecture of the root system (Fitter, 1991; Kramer and Boyer, 1995; Lopez-Bucio et al., 2003). Architectural changes can be caused by changes in lateral root number, distribution, growth rate, or orientation (Malamy, 2005).

The process of lateral root formation has been extremely well studied in many plants, including Arabidopsis. The first visible evidence of lateral root formation is the appearance of anticlinal divisions in a small subset of ‘founder cells’ in the root pericycle (Casimiro et al., 2003; Dubrovsky et al., 2000; Malamy and Benfey, 1997a,b). This is referred to as ‘initiation’. Once initiation has occurred, the founder cells begin a well-conserved program of cell divisions to form a lateral root primordium (Casimiro et al., 2003; Charleton, 1991; Malamy and Benfey, 1997a,b). Primordia then emerge from the parent root largely by expansion of the existing cells. After emergence, the lateral root apical meristem is activated and the newly autonomous lateral root begins to grow (Malamy and Benfey, 1997a,b). The initiation, development, and emergence of the primordium and the subsequent activation of the lateral root apical meristem appear to be regulated by distinct but overlapping developmental mechanisms. In addition, each of these events can be affected by environmental signals (Malamy, 2005).

The osmotic potential of the soil alters the depth of the root system, its overall mass, the rate of root elongation and the number of lateral roots in many plants, including Arabidopsis (Fitter, 1991; Kramer and Boyer, 1995; Russell, 1961; Spollen et al., 1993; van der Weele et al., 2000). Nevertheless, nothing is known about the mechanisms that control root system responses to water availability. In this study, we show that reducing water availability dramatically represses the formation of autonomous lateral roots from lateral root primordia in Arabidopsis, while lateral root initiation is not greatly affected. Abscisic acid (ABA) and a newly identified gene, LRD2, are involved in osmotic repression of lateral root formation. Interestingly, these regulators are also involved in establishing root system architecture in the apparent absence of osmotic stress. The lrd2 mutant has an altered response to exogenous ABA, while both ABA-deficient mutants and the lrd2 mutant show altered responses to auxin polar transport inhibitors, suggesting an interplay of hormone signaling pathways in the regulation of lateral root formation. We propose a model in which promotive and repressive hormone signaling pathways and regulators such as LRD2 determine the fate of a lateral root primordium and coordinate root system architecture with environmental cues.


Root system architecture is regulated by osmotica

We observed that seedlings of Arabidopsis var. Columbia maintained on our standard growth media, full-strength Murashige and Skoog (MS) salts with 4.5% sucrose (MS4.5), for 12 days produced a very small root system, consisting of a long primary root but almost no lateral roots. As we are interested in understanding factors that influence root system architecture, we examined the components of the growth media to determine which one(s) were having such a strong repressive effect on lateral root formation. Previous reports have described repression of lateral root formation when plants are grown on high concentrations of nitrate (≥10 mm) (Zhang and Forde, 1998; Zhang et al., 2000). MS4.5 contains 40 mm nitrate in the form of 20 mm KNO3 and 20 mm NH4NO3. Indeed, when we reduced the concentration of nitrogen salts in our media to 10 mm (5 mm NH4NO3 and 5 mm KNO3), the lower end of the repressive nitrate range, 12-day-old seedlings produced a much larger root system with many lateral roots. MS4.5 with 5 mm NH4NO3 and 5 mm KNO3 is hereafter referred to as the ‘control’ condition, as opposed to ‘repressive’ conditions where few lateral roots are formed. The difference in root system size was quantified by tracing and measuring the total lengths of all lateral roots in the root system (Figure 1a,b).

Figure 1.

Increases in osmotica repress root system size.
(a) Seedlings were grown for 12 days on control media, or control media supplemented with 30 mm N salts [in the form of NH4NO3 and KNO3 (MS4.5)], 30 mm KCl or 60 mm mannitol. The total nitrate concentration of the control media is 10 mm; nitrate concentrations for the supplemented media are 40, 10 and 10 mm, respectively. One representative seedling is shown. In this experiment repression of root system size is related to osmotic concentration, not nitrate concentration.
(b) The osmotic repression of root system size was quantified by measuring and summing the lengths of all the lateral roots on each plant to find a total lateral root length. This gave us a reasonable indication of root system size, and is more feasible in Arabidopsis than obtaining a mass measurement. Total lateral root length in cm was quantified for at least 10 seedlings in each growth condition. Shown are means of 10 plants ± SD.

The fact that reducing nitrogen salts in the media results in seedlings with larger root systems suggests that nitrogen salts are the repressive components of MS4.5. To determine whether the repression of lateral root formation was specific to nitrogen salts or was due to the increased osmotic strength of the media caused by the presence of these salts, we supplemented control media with 30 mm KCl. This created conditions with an equivalent concentration of osmotica as the repressive MS4.5 but with only 10 mm nitrogen salts. KCl supplementation appeared to reduce root system size as effectively as nitrogen salt supplementation (Figure 1a), suggesting that repression is caused by either the high osmotic or high ionic strength of the media. To distinguish between these possibilities, control media was supplemented instead with 2 m equivalents of the non-ionic, non-metabolized sugar mannitol (60 mm). Mannitol also effectively repressed lateral root formation (Figure 1a). Using total lateral root length to quantify root system size, we found that root systems of seedlings grown on control media supplemented with nitrogen salts, KCl or mannitol were 100, 98.8, 92.7% repressed, respectively, as compared with seedlings grown on the control media (Figure 1b). Hence, within the osmotic potential range of this assay, the concentration of osmotica is a critical determinant of root system size.

Sucrose can act as a signaling molecule in certain situations, and therefore it was possible that the observed effect of osmotica would only be seen in the presence of high sucrose, such as the 4.5% used in our experiments. However, when we reduced sucrose concentrations to 1%, mannitol still had a repressive effect on total lateral root length (Figure 2, compare first and second bars, Student's t-test, P ≤ 0.01). The high concentration of sucrose in the media makes a significant contribution to the total osmotic strength. When the osmotic contributions of sucrose and mannitol were calculated (in parentheses in Figure 2), an inverse correlation was seen between the concentration of these sugars and total lateral root length (Figure 2). This suggests that sucrose is not acting as a signaling molecule in this assay, and that the root system architecture is related to the osmotic potential of the media.

Figure 2.

Mannitol represses root system development at high and low sucrose concentrations. The sucrose concentration was reduced in the control media to 1%. Seedlings were grown for 12 days on this media with or without mannitol (bars 1 and 2). Seedlings grown on control media with or without mannitol were included in the experiment for comparison (bars 3 and 4). Total concentration of added sugars (sucrose plus mannitol, mm) are shown in parenthesis. Mannitol represses root system size at both sucrose concentrations relative to controls. Shown are means from at least 15 plants ± SD.

Osmotic stress is not required for osmotic repression of lateral root formation

Direct measurements were taken of the osmotic potential of our media to determine the level of osmotic stress simulated by our conditions (Table 1). These readings show that the osmotic potentials of media that repress lateral root formation vary from approximately −0.64 to −0.68 MPa, while control media had an osmotic potential of approximately −0.51. Although it is difficult to make a direct correlation between these osmotic potential readings and soil drought conditions, osmotic potentials in this range have been described previously as a mild level of drought stress (Kramer and Boyer, 1995; van der Weele et al., 2000). van der Weele et al. (2000) found only slight decreases in shoot dry weight and no difference in root dry weight in this osmotic range, and all seedlings in our experiments appeared green and healthy. The permanent wilting point for most plants is −1.5 to −2.0 MPa, significantly lower that that of our media (Kramer and Boyer, 1995). In addition, a decrease in root system size is still observed when mannitol is added to media containing 1% sucrose; the osmotic potential range in this experiment is much higher than when 4.5% sucrose is used (1% sucrose media without mannitol approximately −0.2 MPa, 1% sucrose media with 60 mm mannitol >−0.4 MPa, Christmann et al., 2004). Recent experiments showed that increases in ABA levels and/or signaling, usually associated with osmotic stress, were not apparent at an osmotic potential of −0.4 MPa. Hence, our experiments suggest that small increases in the concentration of osmotica, under conditions where plants are unstressed or experiencing only mild osmotic stress, have a repressive effect on lateral root formation.

Table 1.  Osmotic potentials (MPa) of control and repressive mediaa
  1. Values are averages of readings from two plates.

  2. aThe osmotic potential was determined in the solidified media used for seedling growth.

Control (MS with 4.5% sucrose and 10 mm nitrogen salts)−0.514a
Control plus 30 mm N salts−0.6577
Control plus 30 mm KCl−0.6415
Control plus 60 mm mannitol−0.6746

Osmotic repression of lateral root formation does not reflect decreased growth rate or delayed plant development

The repression of lateral root formation by osmotica might have been the result of an overall reduction in growth rate, as primary roots of seedlings on increased osmotica grew more slowly and were noticeably shorter (Figure 1a). If this were the case, we would expect lateral roots to appear a few days later, when primary roots of seedlings on repressive osmotic conditions had reached the lengths of primary roots grown on control conditions. However, seedlings grown on MS4.5 media or control media plus mannitol (mannitol) still had few, if any, lateral roots when primary roots had reached the length of the 12-day-old control plants (data not shown). Therefore, differences in root system size under the two growth regimes are not caused by a difference in primary root growth rate. The size of the shoot system appeared to increase in direct proportion to the root system size under all conditions tested (Figure 1a). The coordination of root and shoot system sizes is a well-documented phenomenon; osmotic regulation of the root system may be indirect, occurring at least in part as a result of osmotic effects on the shoot system. However, the reduction in root system size does not reflect an overall slower development of the plant, as shoot systems contained equivalent numbers of leaves under each growth condition (Figure 1 and data not shown).

Osmotic repression targets the formation of lateral roots from lateral root primordia

Lateral root formation can be divided into several broad stages: (i) Initiation of divisions in the pericycle; (ii) Development of a patterned primordium via further division of these cells; (iii) Emergence of the primordium from the parent root through cell expansion; and (iv) Activation of the new lateral root apical meristem (Malamy and Benfey, 1997a,b). Differences in the frequency or rate of any of these events could account for the differences in total lateral root length that we observe in the presence of increased osmotica.

To ask whether increases in osmotica affect lateral root initiation or a later stage of lateral root formation, we compared seedlings grown on control conditions and on repressive osmotic conditions, using nitrogen salts, KCl or mannitol as the osmotica. All roots were cleared and examined microscopically to visualize lateral root primordia, which can be easily detected as early as stage I (Malamy and Benfey, 1997a). All primordia looked morphologically similar under all growth conditions. The total number of all lateral roots and lateral root primordia was not affected by N or KCl, but was significantly lower in seedlings grown on 60 mm mannitol, indicating that lateral root initiation is unaffected or only slightly affected by increases in osmotica (Figure 3a). We then categorized all lateral roots as being primordia (including primordia just after initiation, during development or at the emergence stage), or as autonomous lateral roots with activated meristems (see Experimental procedures). Strikingly, we found that a dramatically reduced percentage of primordia progressed through meristem activation under repressive osmotic conditions, suggesting that osmotica downregulate a post-initiation step in lateral root formation (Figure 3b). To directly test this hypothesis, individual primordia in 7-day-old seedlings were identified and marked under a dissecting microscope without perturbing growing roots. Although it is impossible to stage a primordium precisely with this method, only developing, unemerged primordia (those that had pushed through the endodermal layer but had not entered the epidermal layer) were chosen for marking. Eleven to 12 days later, over 60% of the marked primordia had emerged and become autonomous lateral roots in seedlings grown on control conditions, while <13% had done so in the presence of mannitol (Table 2). The above findings indicate that repressive osmotic conditions have a weak effect on initiation and a strong effect on development, emergence and/or activation of lateral roots. In other words, increasing concentrations of osmotica repress or delay the formation of autonomous lateral roots from lateral root primordia.

Figure 3.

Osmotic repression of root system development targets the formation of lateral roots from lateral root primordia. Twelve-day-old seedlings were cleared and analyzed microscopically. All initiation events were counted and categorized as primordia or lateral roots (see Experimental procedures).
(a) Initiation events per cm primary root were similar in control, N salts and KCl and were slightly reduced on mannitol.
(b) The percentage of initiation events that resulted in lateral roots was calculated from the same seedlings analyzed in (a). All osmotica strongly repressed development of lateral roots from primordia.
Shown are means from 10 plants ± SD. *Significant difference from control based on Student's t-test, P < 0.01.

Table 2.  Mannitol repression of lateral root formation from lateral root primordiaa
ExperimentMediaAge of plant when primordia are markedDays from marking until primordia are scoredLateral root formation from marked primordia (%)Number of primordia scored
  1. aLateral root formation from primordia was determined directly by marking primordia in growing seedlings and observing their development 11–12 days later. In both experiments, a much greater percentage of primordia became autonomous lateral roots under control conditions than under mild osmotic stress conditions.

1Control plus 60 mm mannitol91112.58
2Control plus 60 mm mannitol91212.1233

Molecular players in the osmotic regulation of root system architecture

Repressive osmotic conditions block or delay the formation of an autonomous lateral root from a lateral root primordium. It is reasonable to assume that some critical signaling molecules are required for this response to take place; therefore, it should be possible to genetically disrupt osmotic regulation. We took a candidate gene approach and a forward genetic approach to identify key molecules in the osmotic regulation of root system architecture.

Abscisic acid biosynthetic genes and ABA.  Mutants in ABA biosynthesis are available, and have been shown to be ABA-deficient (Leon-Kloosterziel et al., 1996). We used these mutants to ask whether ABA biosynthetic genes, and by extension ABA, are essential for correct regulation of root system architecture. ABA is known to play a key role in many osmotic responses. In addition, exogenous ABA application is reported to arrest lateral root development immediately after emergence of the primordium (De Smet et al., 2003). We found that both aba2-1 and aba3-1 seedlings exhibited dramatically increased root system size when grown on mannitol, as compared to wild-type seedlings (Figure 4). These mutants also showed increased root system sizes when grown under control conditions (Figure 4). An increased root system size in these mutants was also reported by Signora et al. (2001) under both high and low nitrate conditions. This result demonstrates that ABA is essential for constraining the development of lateral roots under all conditions tested.

Figure 4.

Root systems of ABA biosynthetic mutants are larger than wild type and are less repressed by osmotica. Wild-type and mutant seedlings were grown for 12 days on control media with or without 60 mm mannitol. Shown are means from 10 plants ± SD. *Significant difference from control conditions for the same genotype.

To directly assess whether ABA plays a role in the osmotic repression of root system size, we calculated the proportional change in total lateral root length in response to osmotica in wild-type and mutant seedlings. Root systems of ABA-deficient mutants were less affected by the addition of mannitol than root systems of wild-type seedlings. In the experiment shown, the total lateral root length for wild-type root systems were reduced 97.7% by mannitol, whereas the total lateral root lengths of aba2-1 and aba3-1 root systems were reduced 52.2 and 28.4%, respectively. Although levels of repression differ from experiment to experiment, in three repetitions ABA-deficient mutants always showed a significantly decreased response to mannitol when compared with wild type (GXE interaction based on anova using log scale-converted values for total lateral root lengths, P < 0.01). Hence, normal levels of ABA are essential for lateral root repression by mannitol. From this data, we conclude that ABA is a critical component of the mechanism that represses lateral root formation in response to osmotica, in addition to its role in constraining lateral root formation during normal growth and development. However, we cannot rule out the possibility that the difference in mannitol response is a secondary effect of the overall increase in root system size.

LRD2, a new regulatory gene.  The consistently small root system and lack of lateral roots in wild-type plants grown on repressive osmotic conditions provides a sensitive assay for detecting even subtle alterations in root system architecture. We took advantage of this to screen for mutants with defects in regulators of root system architecture. One such mutant, lateral root development 2 (lrd2), produced many lateral roots on repressive conditions (Figure 5a,b). Although originally isolated in a screen that used nitrogen salts as the osmoticum, lrd2 also exhibited an increased root system size on KCl and mannitol (Figure 5a; compare to wild type in Figure 1a). Indeed, like the ABA-deficient mutants, lrd2 showed an increased root system size even under control conditions (Figure 5a,b). Therefore, it appears that the LRD2 gene product, like ABA, plays a role in constraining lateral root development under all conditions tested.

Figure 5.

The lrd2 mutant exhibits increased root system size.
(a) lrd2 seedlings were grown for 12 days on the indicated media. For media details, see legend of Figure 1. One representative seedling is shown. Root systems were manually spread out for easier visualization.
(b) Total lateral root lengths were quantified for wild-type and lrd2 seedlings grown on control media with or without the addition of 60 mm mannitol. Shown are averages of 10 or more plants ± SD.

To directly assess whether LRD2 plays a role in the osmotic repression of root system size we calculated the proportional change in total lateral root length in response to osmotica in wild-type and mutant seedlings. The root system size were less affected by the addition of mannitol in lrd2 mutants than in wild-type seedlings. In the experiment shown, the lrd2 mutant showed a 47.7% reduction in total lateral root length versus 97.7% reduction in wild type; in three experiments, lrd2 was significantly less responsive to the repressive effects of osmotica (GXE interaction based on anova using log scale-converted values for total lateral root length, P < 0.001). Therefore, we conclude that LRD2 is a critical component of both the mechanism that represses lateral root formation in response to osmotica and the mechanism that constrains lateral root formation during normal growth and development. However, it cannot be ruled out that the apparent difference in mannitol response is a secondary effect of the overall increase in root system size.

We previously demonstrated that osmotica repress or delay the formation of lateral roots from primordia in wild-type seedlings. If LRD2 is required for this regulatory mechanism, as suggested by the increase in total lateral root length in lrd2, we would expect to see a higher percentage of primordia develop into lateral roots in the mutant. As the lrd2 root systems are larger than that of wild type under both control and mild osmotic stress conditions, this increase in lateral root formation from primordia should occur under both conditions. Indeed, when all primordia and lateral roots were examined in 12-day-old wild-type and lrd2 mutant seedlings grown with or without mannitol, we found that the percentage of primordia that had emerged to form lateral roots was always higher in the mutant (Figure 6). Furthermore, the affect of mannitol on lateral root formation was much less in the mutants than in the wild-type seedlings (GXE interaction based on anova, using log scale-converted values for % lateral root formation from primordia P < 0.0001 in three experiments). Indeed, the percentage of primordia that had emerged at 12 days was not significantly affected by mannitol in lrd2 in three experiments (P > 0.06). These findings suggest that LRD2 is involved in an intrinsic pathway that determines the percentage of primordia that become lateral roots under all conditions tested, and is also involved in mediating the effect of mild osmotic stress on this pathway.

Figure 6.

A higher percentage of lateral root primordia develop into lateral roots in the lrd2 mutant. Wild-type and mutant seedlings were grown for 12 days on control media with or without 60 mm mannitol. Shown are means from 10 plants ± SD. *Significant difference from control conditions for the same genotype.

Interestingly, lateral root initiation levels also increased significantly in lrd2 under both conditions, and this was not significantly affected by mannitol (data not shown). Therefore, it is possible that LRD2 plays additional roles in intrinsic regulation of root system development.

Genetic analysis of the lrd2 mutant

The lrd2 mutant was backcrossed to wild-type Columbia plants. The segregation ratio in the F2 progeny was consistent with a single, recessive mutation (79 unbranched:23 branched; χ2 = 0.32, P = 0.99). lrd2 was also crossed to WS for genetic mapping using standard approaches (see Experimental procedures). The mutation in lrd2 was mapped to a 762 kb region of Chromosome 1 between two homemade markers, JJ2 and JJ4 (see Experimental procedures). The region is predicted to contain approximately 205 ORFs.

lrd2 does not phenocopy ABA-deficient mutants

The lrd2 mutant appears phenotypically similar to aba2-1 and aba3-1; both form larger root systems than wild type under all conditions tested, and both have decreased osmotic repression of lateral root formation. Therefore, lrd2 may have an overall defect in ABA synthesis or accumulation. We reasoned that if this scenario were true lrd2 should phenocopy the ABA-deficient mutants in other assays. One diagnostic assay for ABA-deficient mutants is the response of seedlings to NaCl. NaCl inhibits germination at high concentration, but at intermediate concentrations it induces an arrest of seedling development and prevents cotyledon greening; this has been proposed to be a protective response (Lopez-Molina et al., 2001). NaCl-induced developmental arrest requires ABA, and ABA-deficient mutants are therefore less able to arrest growth in response to NaCl (Gonzalez-Guzman et al., 2002; Leon-Kloosterziel et al., 1996). We placed wild-type, lrd2, aba2-1 and aba3-1 seeds in MS4.5 liquid media with increasing concentrations of NaCl. All wild-type seedlings germinated and cotyledons greened at 100 mm NaCl, but seedlings arrested development and remained white at NaCl concentrations of 150 mm and above. As expected, ABA-deficient aba2-1 and aba3-1 mutant seedlings showed a much lower frequency of arrest, even at 200 mm NaCl (Figure 7a and data not shown). Surprisingly, lrd2 showed an opposite phenotype to aba2-1 and aba3-1. Seedlings appeared hypersensitive to NaCl, with approximately 50% arresting development at 100 mm NaCl (Figure 7a). These results clearly demonstrate that lrd2 does not phenocopy ABA-deficient mutants, and therefore is unlikely to have a defect in general ABA biosynthesis or accumulation.

Figure 7.

The lrd2 mutant is hypersensitive to NaCl and ABA.
(a) The lrd2 and ABA-deficient mutants have opposite NaCl sensitivities in germination assays. Approximately 100 wild-type, lrd2, or aba2-1 seeds were placed in MS4.5 liquid media in six-well plates with 0, 100, 150, or 200 mm NaCl. After 12 days the number of seedlings in each condition that arrested at an early developmental stage without greening were counted. Shown are percentages that exhibited developmental arrest.
(b) lrd2 and ABA-deficient mutants are both hypersensitive to ABA. Approximately 100 seeds of the above genotypes were placed in MS4.5 in six-well plates with 0.5, 1.0, or 2.0 μm ABA or equivalent volume of 0.1 m NaOH, which was the ABA solvent.

lrd2 exhibits defects in ABA responses

When wild-type seeds are exposed to ABA, seedling development arrests as described for the NaCl response (Lopez-Molina et al., 2001). Interestingly, lrd2 shows hypersensitivity to ABA in this response (Figure 7b), as does aba2-1 (Figure 7b) and aba3-1 (data not shown). The fact that the lrd2 response to exogenous ABA differs from wild-type indicates that, although lrd2 does not appear to be defective in ABA biosynthesis or accumulation, it is compromised in some aspect(s) of ABA perception and/or signaling.

Osmotic repression of lateral root development can be overcome by auxin

The phytohormone auxin stimulates lateral root initiation and is essential for the formation of autonomous lateral roots from lateral root primordia (reviewed in Casimiro et al., 2003; Malamy, 2005). Therefore, auxin appears to have the opposite effects of osmotica. To ask if auxin can overcome the osmotic repression of lateral root formation, we transferred 7-day-old seedlings grown on mannitol to similar plates with or without auxin [50 nm 1-naphthalenacetic acid (NAA)]. At 7 days, primordia are already present in the seedling root. Hence, we can assess the fate of these primordia in the presence of auxin by examining the region of the root already formed at the time of transfer. The transferred region of wild-type seedlings maintained on mannitol formed few, if any, autonomous lateral roots 8 days later, as expected. However, the transferred region of seedlings shifted to mannitol + NAA media had large numbers of visible lateral roots (Table 3). Microscopic analysis of the transferred root regions showed that when maintained on mannitol, a smaller percentage of the primordia became lateral roots, while in the presence of NAA this percentage rose dramatically (Table 3). Hence, auxin stimulates lateral root formation from primordia and is clearly able to overcome the repressive effect of osmotica.

Table 3.  Auxin stimulates the formation of lateral roots from lateral root primordiaa
ExperimentMedia after transferNo. of lateral rootsP-value (Student's t-test)Lateral root formation from lateral root primordia (%)P-value (Student's t-test)Number of plants scored
  1. aSeedlings were grown on media containing 60 mm mannitol for 7 days to allow primordia formation, and then transferred to the same media with or without 50 nm NAA. Although few primordia developed into lateral roots on mannitol alone, the repression of lateral root formation was overcome by NAA in two experiments.

1Mannitol2.6 + 1.8  32.8 + 27.1 13
Mannitol + 50 nm NAA18 + 2.7<0.00175.74 + 11.5<0.00110
2Mannitol1.2 + 1.1  13.3 + 14.25  5
Mannitol + 50 nm NAA13.7 + 2.7<0.001 72.6 + 7.4<0.00110

To further test the effect of auxin signaling on the osmotic repression of lateral root formation, wild-type seedlings were transferred from mannitol plates to control plates in the presence or absence of the polar auxin transport inhibitors 2,3,5-triiodobenzoic acid (TIBA) or N-1-naphthylphtalamic acid (NPA). When shifted from mannitol to control conditions, lateral root development from primordia rapidly recovered in the transferred region. In contrast, the presence of TIBA in control media completely blocked the formation of autonomous lateral roots from primordia (Figure 8a). Similar results were seen with another transport inhibitor, NPA (data not shown). Hence, auxin transport is essential for formation of autonomous lateral roots from primordia, and reduced auxin transport resembles the effect of increased concentrations of osmotica.

Figure 8.

Lateral root development from primordia is inhibited by TIBA in wild-type but not lrd2, aba2-1 or aba3-1 mutant plants.
(a) Seedlings of the indicated genotypes were grown for 5 days on control media supplemented with 60 mm mannitol. They were then transferred to control media with TIBA or an equivalent volume of the TIBA solvent DMSO. The positions of primary root tips were marked (black line in photographs). Seven days later the seedlings were photographed; representative seedlings are shown. Agravitropic growth of roots after transfer is an effect of the TIBA. aba2-1 mutants resembled aba3-1 mutants.
(b) Average number of lateral roots produced in the transferred region of the seedlings described in (a). Data from seedlings transferred to DMSO (solvent control) are shown in solid bars; data from seedlings transferred to TIBA are shown in gray bars. Shown are averages of 14 or more plants ± SD.

The lrd2 and ABA-deficient mutants exhibit altered auxin signaling

Our results confirm that auxin signaling is both necessary and sufficient for stimulating the formation of lateral roots from primordia: TIBA inhibits lateral root formation on control conditions and exogenous auxin promotes it on repressive conditions (see previous section). This is in accordance with previous studies that reached a similar conclusion (Bhalerao et al., 2002; Casimiro et al., 2001). The lrd2 and ABA-deficient root system phenotypes could therefore be caused by increased auxin signaling in these mutants.

To test whether auxin signaling is altered in the mutants, we grew wild-type, lrd2, aba2-1 and aba3-1 mutant roots on TIBA. When lrd2 mutant seedlings were transferred from mannitol to control conditions in the presence of TIBA, a significant number of lateral roots still formed in the transferred region as compared with a complete absence of lateral root formation in wild type (Figure 8a,b). A similar phenotype was seen when ABA-deficient mutants were transferred to TIBA (Figure 8a,b). The mutants’ reduced sensitivity to TIBA indicates that both LRD2 and ABA signaling interact with auxin signaling.


Root system architecture is modified in response to water availability

In this paper, we show that osmotic potential can regulate the formation of lateral roots from lateral root primordia. Conditions that reduce water availability dramatically repress this process. Transfer of seedlings to conditions where water is more available reverses the effect of osmotica, and lateral roots rapidly form (Figure 8). There is no clear stage at which primordia arrest under repressive osmotic conditions, nor is the morphology of the primordia abnormal under repressive conditions. Instead, it is likely that increased osmotica reduces the overall rate of primordia development into lateral roots. From these observations, we can speculate that Arabidopsis plants optimize their root systems by repressing root proliferation into regions where less water is available. Instead, slow-developing primordia are maintained, creating a ‘primed’ root system that can respond rapidly to water when and where it is sensed.

It is significant that the shift from control conditions, where lateral roots form, to repressive osmotic conditions where their formation is repressed, represents only a small decrease in osmotic potential, in the absence of what is usually described as osmotic stress conditions. Indeed, the modulation mechanism for lateral root formation might be expected to respond to such small changes within a non-stress context, as the root system is constantly adjusted and re-optimized to the transient conditions in its environmental niche. It is exciting that this response has a clear dependence on ABA signaling. ABA has previously been implicated in many plant responses to severe drought stress. From our results, it now appears that an ABA-mediated signaling mechanism functions to adapt root system development to the osmotic conditions that plants see during normal growth (Kramer and Boyer, 1995). This is consistent with the demonstration that ABA signaling occurs in well-watered plants and is modulated by changes in osmotic potential that do not induce stress responses (Christmann et al., 2004).

Osmotic versus nitrate signaling in the root system

The repression of lateral root development by osmotica demonstrated here is highly reminiscent of the effects reported for nitrate. In fact, both signals appear to repress lateral root formation via an ABA-mediated pathway, and to act at the level of lateral root formation from primordia rather than lateral root initiation (this paper, Signora et al., 2001). Nevertheless, our results clearly demonstrate that in the osmotic range of our assays the effects of nitrate can be mimicked by other osmotica, such as KCl and mannitol. Hence, in our assay N salts are acting as osmotica. Despite this observation, we do see differences between plant responses to high nitrate and mannitol, suggesting that there may be distinct pathways. The high nitrate repression of lateral root formation appears to be stronger, with fewer ‘escaped’ lateral roots (data not shown). There is considerable literature discussing the interplay between the effects of nitrogen and water deficits on root systems (Wilkinson and Davies, 2002), and it is possible that the nitrate and osmotic response pathways converge at an early point – indeed, ABA may integrate nitrate and osmotic sensing. It is also important to note that addition of nitrate to media always changes the osmotic potential, and the concentration of osmotica reduces the uptake of nitrogen (Wilkinson and Davies, 2002). Therefore, great care will have to be taken in sorting out these regulatory pathways.

Molecular regulation of root system architecture

Several genes have recently been described that may be regulators of root system architecture. BRX encodes a transcription factor, and elimination of this gene results in a greatly reduced root system size (Mouchel et al., 2004). ANR1 encodes a transcription factor of the MADS-box family and is essential for coordinating lateral root development with external nitrate (Zhang and Forde, 1998). LIN1 coordinates lateral root initiation with carbon:nitrogen ratios (Malamy and Ryan, 2001), and PDR2 coordinates root system development with phosphate levels. NIT3, a nitrilase that forms auxin from a precursor, is expressed in lateral root primordia, and is induced in response to sulfate deprivation (Kutz et al., 2002). Despite these recent findings, we still have only a rudimentary understanding of how root system architecture is regulated, and how it is adjusted in response to environmental conditions.

lrd2, aba2-1 and aba3-1 all have dramatically increased root system size when compared with wild type, under both our mild osmotic stress and control conditions. This is also true when sucrose concentration is reduced to 1% (data not shown). In the case of lrd2, we directly observed that the formation of lateral roots from primordia is de-repressed. Thus, ABA and LRD2 are both necessary to constrain lateral root formation from primordia under all conditions tested.

Root system size is not as strongly affected by osmotica in lrd2 mutant plants as in wild-type plants. This suggests that the lrd2 mutation causes a defect in either perception or response to this environmental cue. ABA-deficient mutants aba2-1 and aba3-1 show a similar phenotype. Therefore, we postulate that both LRD2 and ABA also play a role in osmotic repression of lateral root formation. This suggests that the same molecules that mediate responses to environmental cues are regulators of intrinsic developmental programs in the root system.

It is important to note that although osmotic repression of root system size is compromised in all three mutants it is not eliminated. In the ABA-deficient mutants this might be explained by the fact that there is still ABA production in these plants, and therefore they might be expected to be hypomorphs (Leon-Kloosterziel et al., 1996). lrd2 may also not be a null mutant; alternatively, it may act redundantly with other molecules, or may alter sensitivity to osmotica rather than completely eliminating the response pathway.

It initially appears puzzling that neither the initiation of primordia nor the percentage of primordia that develop into lateral roots are affected by osmotica in lrd2, as we know that total lateral root length is affected (Figures 5 and 6). This could be explained if our repressive conditions do not block the development of lateral root primordia into lateral roots, but slow it down. Lateral roots that have just emerged from the parent root and become autonomous are much smaller, and therefore make a smaller contribution to the root system size. In 12-day-old lrd2 seedlings, a similar percentage of lateral roots primordia have become lateral roots in the presence and absence of mannitol, but in the presence of mannitol many of these roots are newly emerged and are too small to make a significant contribution to the root system size. In 12-day-old wild-type seedlings, in contrast, mannitol has caused a greater delay in primordia development such that a smaller percentage of primordia have emerged from the parent root.

An interplay between ABA and auxin signaling?

Our data demonstrate that ABA, LRD2, and increases in osmotica all repress lateral root formation, and that there is an interaction among these signaling components. Osmotic repression requires ABA and LRD2, and the lrd2 mutant is hypersensitive to ABA in certain assays. In contrast, auxin promotes lateral root formation from primordia, even in the presence of repressive osmotica. It is tempting to speculate that the balance between these promotive and repressive signaling mechanisms plays a role in determining the fate of a lateral root primordium. An interplay between auxin and ABA signaling has been previously suggested. The ABA-insensitive abi3-6 mutant has altered responses to auxin and NPA in terms of lateral root initiation (Brady et al., 2003). Another mutant, ibr5, shows defects in both auxin and ABA responsiveness (Monroe-Augustus et al., 2003). Christmann et al. (2004) propose an interplay between auxin and ABA in the regulation of cell division in the root quiescent center. Hence, it is becoming clear that there is cross-talk between ABA and auxin signaling pathways.

A working model

A working model that is consistent with all the data is presented below, with arrows representing lateral root primordia and lateral roots (Figure 9). This model postulates that a balance between promotive and repressive signaling pathways determines the fate of lateral root primordia under all growth conditions. These pathways could involve auxin and ABA signaling, respectively. ABA represses lateral root formation from primordia, and hence reduction of ABA levels in the ABA-deficient mutants shifts the balance toward promotive signaling, as does the addition of exogenous auxin. TIBA/NPA treatment reduces auxin signaling, shifting the balance toward repressive signaling. The fact that ABA-deficient mutants show a decreased sensitivity to TIBA/NPA is consistent with this model. Environmental cues, such as a reduction in water availability, alter the fate of lateral root primordia by influencing the promotive and repressive pathways in unknown ways. As osmotic repression requires normal ABA biosynthesis, it is possible that it acts through increasing ABA signaling, although there is no direct data to support this. LRD2 may fit into this model either by negatively regulating promotive (auxin) signaling or by positively regulating repressive (ABA) signaling. Note that in this model ABA plays a role in development even in the absence of osmotic stress. This is consistent with recent demonstrations that ABA signaling can occur in a stress-independent manner (Christmann et al., 2004 and references therein).

Figure 9.

A model for the regulation of root system architecture.

This working model is almost certainly oversimplified. Indeed, De Smet et al. (2003) found that the arrest of lateral root formation with exogenous ABA could not be overcome with exogenous auxin, confirming that there are complexities in the system that have not been accounted for by our model. Nevertheless, our model provides a framework for understanding how the intrinsic root system architecture of a plant is determined, and how it is altered in response to environmental growth conditions. Furthermore, two key regulatory molecules, ABA and LRD2, are identified that play roles in both fundamental processes. The ability of a plant to produce an optimal root system is critical for survival, especially in challenging environments. Therefore, there is tremendous potential to increase the fitness of crop plants by altering pathways that regulate lateral root formation. Understanding the molecular mechanisms that regulate root system architecture is an essential step toward this important goal.

Experimental procedures

Upon request, all novel materials described in this publication will be made available in a timely manner for non-commercial research purposes. No restrictions or conditions will be placed on the use of any materials described in this paper that would limit their use in non-commercial research purposes.

Plant growth conditions

Seeds were surface sterilized in 100% bleach plus Tween-20 for 3 min, and then rinsed four times with distilled water. The seeds were allowed to vernalize and imbibe for 2–3 days at 4°C. For all root branching assays, five to 10 seeds were sown on Petri dishes containing various media as described below and in the text. Plates were oriented vertically to allow roots to grow on the surface of the media in a growth chamber set at 22°C, 16 h light, 8 h dark.

Control media was composed of 0.33 g l−1 CaCl2·6H2O; 0.1807 g l−1 MgSO4, 1.7 g l−1 KH2PO4, 10 ml l−1 micronutrients (100× stock from Gibco BRL, Grand Island, NY, USA, cat. no. 11155), 5 ml l−1m NH4NO3, 5 ml l−1m KNO3, 0.5 g l−1 MES [2-(N-morpholino)ethanesulfonic acid] and 45 g sucrose. The pH was adjusted to 5.7 using 1 n KOH and 0.7% BRL Ultrapure Agar (Fisher#B-11849; Fisher Scientific, Pittsburgh, PA, USA) was added before autoclaving. Osmotica were added to the control media before adjustment of the pH. ‘N salts’ are composed of equimolar amounts of NH4NO3 and KNO3. ‘K salts’ refers to KCl. NAA (Sigma N-1641; Sigma, St Louis, MO, USA) was added to autoclaved media at a final concentration of 50 nm. TIBA (Sigma T-5910) and NPA (Chemical Services PS-343, West Chester, PA, USA) were dissolved in DMSO and added to autoclaved media at a final concentration of 20 and 10 μm, respectively. Equivalent volumes of DMSO were added to plates as a control.

Determination of total lateral root length

Digital images of plants were traced using ImageJ 1.29j (Rasband W. National Institutes of Health, USA) and individual lateral root lengths were determined. Total lateral root length represents the sum of the length of all lateral roots per plant.

Osmotic potential readings

Water potential was determined by isopiestic thermocouple psychrometry (Boyer and Knipling, 1965).

Statistical analysis

GXE interactions were determined calculated using the anova function in StatView 5.0.1 (SAS Institute, Cary, NC, USA). This allowed us to predict if plants of different genotypes were responding differently to osmotica. The log of total lateral root length or so lateral root formation from primordia values for each plant were used in making this calculation to account for differences in scale between genotypes.

Microscopic analysis of lateral root initiation and development

All tissues were cleared by incubating sequentially, 5–15 min each, in (i) 20% methanol acidified with 4% concentrated hydrochloric acid, 55°C; (ii) 7% NaOH in 60% ethanol. Tissues were then re-hydrated by 10-min incubations in 40, 20 and 10% ethanol, and then infiltrated for 10 min in 50% glycerol/5% ethanol. Cleared tissues were then mounted in 50% glycerol and visualized using DIC optics on a Leica DMR microscope (Leica Microsystems, Deefield, IL, USA). Even very early lateral root primordia can be easily visualized using this method.

Lateral root initiation numbers include all lateral root primordia and lateral roots at any developmental stage. Autonomous lateral roots are defined as any primordia that show signs of growth from the lateral root apical meristem. This is seen by the increased number of small cells near the lateral root tip (Malamy and Benfey, 1997a).

To mark lateral root primordia in growing seedlings, plants were visualized using a Zeiss M2 Bio dissecting microscope (Zeiss, Thornwood, NY, USA). Marks were scratched onto the back of the Petri dish to identify the location of primordia.

Isolation of mutants

Approximately 5000 M2 seeds (ecotype Columbia) that had been mutagenized with ethyl methanesulfonate were planted on MS4.5. At 12 days, plants that produced four or more lateral roots were transferred to soil. Progeny were then re-tested under similar conditions. Fifteen lrd mutants were isolated; the characterization of one of these mutants is presented here.

Genetic analysis and mapping

The lrd2 mutants were crossed to WS and the F2 progeny were used for mapping the genes using standard positional mapping techniques. Standard SSLP and CAPS PCR conditions were used. Map positions of the markers were taken from the public chromosome maps (http://www.arabidopsis.org). New markers that were polymorphic between Col and WS ecotypes were identified in our laboratory. Breakpoint analysis of recombinant F2 plants from an lrd2 × WS cross delimited the mutations to an approximately 760 kb region flanked by JJ2 and JJ4 markers. The PCR primers used to identify the JJ2 polymorphism (2 bp INDEL) were F: caccattagcaacatacggg; R: taaacgtggtcgtggtcg. For the JJ4 polymorphism (1 bp INDEL) F: actgtgttaagaataccggc; R: attgcctttgcgatgttacc.


The authors are grateful to T. Baskin and R.E. Sharp for assistance with the osmotic potential measurements, and J. Jung and P. Brannon for mapping the LRD2 gene. We also thank P.N. Benfey, L. Mets, J. Greenberg and Melissa Lehti-Shiu for critical discussions and reading of the manuscript. The project was supported by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service, grant number 2001-35100-10650.