Dissecting the biosynthetic pathway for the bypass1 root-derived signal


(fax +1 801 581 4668; e-mail sieburth@biology.utah.edu).


The Arabidopsis BYPASS1 (BPS1) gene is required for normal root and shoot development. In bps1 mutants, grafting and root excision experiments have shown that mutant roots produce a transmissible signal that is capable of arresting shoot development. In addition, we previously showed that growth of bps1 mutants on the carotenoid biosynthesis inhibitor fluridone resulted in partial rescue of both leaf and root defects. These observations suggest that a single mobile carotenoid-derived signal affects both root and shoot development. Here, we describe further characterization of the bps1 root-derived signal using genetic and biosynthetic inhibitor approaches. We characterized leaf and root development in double mutants that combined the bps1 mutant with mutants that have known defects in genes encoding carotenoid processing enzymes or defects in responses to carotenoid-derived abscisic acid. Our studies indicate that the mobile signal is neither abscisic acid nor the MAX-dependent hormone that regulates shoot branching, and that production of the signal does not require the activity of any single carotenoid cleavage dioxygenase. In addition, our studies with CPTA, a lycopene cyclase inhibitor, show that signal production requires synthesis of β-carotene and its derivatives. Furthermore, we show a direct requirement for carotenoids as signal precursors, as the GUN plastid-to-nucleus signaling pathway is not required for phenotypic rescue. Together, our results suggest that bps1 roots produce a novel mobile carotenoid-derived signaling compound.


Carotenoids are abundant isoprenoid-derived molecules. In plants, carotenoid biosynthesis occurs in the plastids, where carotenoids are incorporated into the light-harvesting and photosynthetic reaction center complexes. In these complexes, carotenoids serve both to absorb light energy and to dissipate excess energy (photo-protection) (Demmig-Adams and Adams, 1992, 2002; Niyogi et al., 1997). In addition, carotenoids can be processed into apocarotenoids (reviewed in Giuliano et al., 2003), some of which have diverse roles in growth and development of both plants and animals. For example, in animals, the β-carotene derivatives retinol and retinoic acid are important nutritional and signaling molecules. In plants, apocarotenoids contribute to fragrance and flavor, and can also serve as mobile signaling molecules (Booker et al., 2005; Rock and Zeevaart, 1991; Simkin et al., 2004a,b).

Abscisic acid (ABA) and the MAX-dependent hormone that regulates shoot branching are plant signaling molecules that are derived from β-carotene (Figure 1). ABA is a well-characterized plant hormone that has diverse roles in plant growth and development, including seed dormancy, germination and vegetative growth (Arenas-Huertero et al., 2000; Bewley, 1997). In addition, mobile and dynamic pools of ABA mediate plant responses to drought by inducing stomatal closure (Harris and Outlaw, 1991). By contrast, the MAX-dependent hormone was characterized more recently through analysis of moreaxillary branching mutants (max1–max4) (Stirnburg et al., 2002). Double mutant analyses and molecular characterizations indicate that MAX1–MAX4 function in a single pathway to produce an apocarotenoid signal. Furthermore, grafting studies indicate that this signal is graft-transmissible, and that it negatively regulates shoot branch outgrowth (Booker et al., 2004, 2005; Sorefan et al., 2003).

Figure 1.

 Carotenoid biosynthesis (simplified).
This simplified scheme of carotenoid biosynthesis includes the sites of action for inhibitors and the positions where mutations disrupt carotenoid biosynthesis (gray). pds1 and aba3 are indicated in parentheses as they do not encode the carotenoid biosynthetic enzymes, but these mutations result in disruption of carotenoid biosynthesis at the indicated positions. For simplicity, not all intermediates or enzymes are shown.

The production of mobile apocarotenoid signaling molecules requires cleavage of carotenoid precursors. Carotenoid cleavage dioxygenases (CCDs) are a major class of carotenoid cleaving enzymes. The first CCD gene identified came from molecular analysis of maize viviparous14 (vp14) mutants, which showed precocious seed germination (Tan et al., 1997). The VP14 gene product was found to catalyze oxidative cleavage of 9-cis-epoxycarotenoids, which is the rate-limiting step in ABA biosynthesis in maize (Schwartz et al., 2003; Tan et al., 1997, 2003). Subsequently, molecular cloning of Arabidopsis MAX3 and MAX4 revealed that they also encode CCDs (Booker et al., 2004; Sorefan et al., 2003). Subsequent biochemical characterization of MAX3 and MAX4 indicated that they can catalyze β-carotene cleavage reactions (Booker et al., 2004; Schwartz et al., 2004). Genes encoding CCDs have been found in plant, animal and bacterial genomes, suggesting that they carry out important functions (reviewed in Giuliano et al., 2003). In Arabidopsis, there are nine genes with sequence homology to VP14 making up the CCD gene family, and these genes are further divided into two subfamilies, the nine-cis-epoxycarotenoid dioxygenases (NCEDs) and the carotenoid cleavage dioxygenases (CCDs) (Schwartz et al., 2001; Tan et al., 2003). The five NCEDs (NCED2, 3, 5, 6, 9) have the highest sequence similarity to VP14, and four (NCED2, 3, 6, 9) have been shown to have the same enzymatic activity as VP14 in vitro. These enzymes are thought to play roles in ABA biosynthesis in response to developmental or environmental cues (Iuchi et al., 2001; Lefebvre et al., 2006; Tan et al., 2003; Xiong and Zhu 2003).

The four CCDs (CCD1, 4, 7, 8) are more distantly related to VP14. CCD7 and CCD8 correspond to MAX3 and MAX4, respectively, and have been shown to be required for the sequential cleavage of β-carotene to form a mobile signal that controls shoot branching (Booker et al., 2004; Schwartz et al., 2004; Sorefan et al., 2003). CCD1 has been shown to cleave a variety of carotenoids (Schwartz et al., 2001), but the cleavage substrates of CCD4 remain unknown. While the carotenoid cleavage reactions catalyzed by these enzymes have been characterized, the biological function(s) of many of these enzymes, as well as their cleavage products, remain poorly understood.

Carotenoids, in their function as photo-protective pigments, prevent production of a retrograde plastid-to-nucleus signal. This retrograde signal induces changes in nuclear gene expression as a result of photo-oxidative damage to plastids (reviewed in Rodermel and Park, 2003; Strand, 2004). Analysis of Arabidopsis mutants defective in plastid-to-nucleus retrograde signaling, the genome uncoupled (gun) mutants, has revealed that perturbations of specific chlorophyll biosynthesis enzymes disrupt signal production (Nott et al., 2006; Strand et al., 2003).

Recently, we described the Arabidopsis bypass1 (bps1) mutant, which has growth and development defects in both the root and shoot (Van Norman et al., 2004). Grafting and root excision experiments indicated that the bps1 root constitutively produces a graft-transmissible signal that is sufficient to inhibit wild-type shoot growth. BYPASS1 encodes a plant-specific protein of unknown function that we proposed negatively regulates production of a root-derived mobile signal such that, in bps1 mutant seedlings, there is constitutive signal production.

Here we report our investigations into the biosynthesis of the bps1 root-derived signal. Previously we showed both leaf and root phenotypes of bps1 mutants were partially rescued by growth on medium containing fluridone, a carotenoid biosynthesis inhibitor (Bartels and Watson, 1978; Van Norman et al., 2004). We also showed that bps1 aba1 double mutants have an enhanced phenotype. Together these data suggest that the signal may be derived from a carotenoid that is distinct from ABA. In this study, we further explored the requirement for specific carotenoids in biosynthesis of the bps1 signal using inhibitors and double mutants. Our data suggest that the bps1 signal may be a novel β-carotene derivative.


bps1 signal production requires carotenoid biosynthesis

We previously showed that developmental defects in bps1 roots and leaves were partially rescued when seedlings were treated with fluridone (Van Norman et al., 2004). Fluridone inhibits phytoene desaturase, an early step of carotenoid biosynthesis (Figure 1) (Bartels and Watson, 1978). To test whether the bps1 phenotype was also rescued when carotenoid biosynthesis was interrupted genetically, we characterized double mutants betweenbps1 and phytoene desaturase1 (pds1). pds1 mutants are deficient in tocopherol and plastoquinone biosynthesis, and, as a secondary consequence, fail to carry out carotenoid desaturation (Norris et al., 1995, 1998). We reasoned that if synthesis of the bps1 signal required carotenoid precursors, then the bps1 pds1 double mutant would be white (like the pds1 single mutant) and have broad flattened leaves and more elongated smooth roots (like the fluridone-rescued bps1 mutant). In contrast, if synthesis of the bps1 root-derived signal did not require carotenoid precursors, then the bps1 pds1 double mutant would be white (like the pds1 single mutant) and have short knotted-looking roots and radial leaves (like the bps1 mutant) (Figure 2c,d,i,j).

Figure 2.

bypass1 phenotype and partial rescue with CPTA treatment.
(a–d) Plants grown at 22°C, 11 days post-imbibition (DPI), under moderate light intensity (approximately 100 μE m−2 sec−1). (a) Columbia seedling and (b) leaf vein pattern; (c) bps1 seedling and (d) leaf vein pattern.
(e–h) Plants treated grown at room temperature, 15 DPI, under high light intensity (approximately 200 μE m−2 sec−1) and treated with 10 μm CPTA. (e) Columbia leaf vein pattern and (f) root, (g) bps1 leaf vein pattern and (h) root.
(i,j) Untreated plants grown at room temperature, 15 DPI, under high light intensity (approximately 200 μE m−2 sec−1). (i) bps1 leaf; (j) root showing the ‘knotted’ appearance. Arrowheads (f,h,j) indicate the hypocotyl–root junction. Scale bars = 500 μm (a–d) and 250 μm (e–j).

Analysis of the bps1 pds1 double mutant was complicated due to linkage between the two genes (BPS1, At1g01550; PDS1, At1g06570). To have a reasonable expectation of observing the double mutant, we determined recombination rates in this region (see Experimental procedures) to estimate the frequency of double mutants. We predicted that one bps1 pds1 double mutant would be present among every 125 pds1 mutants derived from F2 segregating for both bps1 and pds1. We screened more than 4000 F2 seedlings, and examined 1027 white pds1 mutants. By our calculations, these 1027 pds1 mutants should have included eight double mutants. All pds1 mutants (control and F2) showed a range of phenotypes, including variable leaf size, but all had smooth elongated roots. Among the F2 that segregated for both pds1 and bps1, the white pds1 mutants all produced flattened leaves and smooth elongated roots. Because we observed no white pds1 mutants with bps1 leaf and root morphology, when we expected to see eight, we conclude that loss of pds1 led to rescue of the bps1 phenotype. This observation supports a carotenoid origin for the bps1 root-derived signal.

The bps1 signal is neither abscisic acid nor the MAX-dependent hormone

We next examined whether the bps1 root-derived signal was ABA. Previously, we showed that the bps1 root-derived signal was still produced in bps1 aba1 and bps1 aba2 double mutants, suggesting that the signal was not ABA (Van Norman et al., 2004). However, because small amounts of ABA may still be synthesized in these mutants (Xiong et al., 2002), we revisited the possibility of an ABA signal by analyzing double mutants between bps1 and aba3. ABA3 encodes the molybdenum co-factor (MoCo) sulfurase, and MoCo is required for activity of all four of the ABA aldehyde oxidase gene products (Figure 1), thus a more complete block of ABA biosynthesis is expected in aba3 mutants (Bittner et al., 2001; Schwartz et al., 1997; Xiong et al., 2001).

We analyzed F2 lines segregating for both bps1 and aba3 with the hypothesis that, if the bps1 root-derived signal was ABA, double mutants would show a suppressed (or wild-type) phenotype. Instead, the F2 comprised 25%bps1 mutants that were indistinguishable from bps1 single mutants (Table 1). This result indicates that ABA3 was not required for synthesis of the bps1 signal, and suggests that ABA is not the bps1 root-derived signal.

Table 1.   Genetic analysis of ABA as a candidate for the bps1 signal
Plants analyzedNumber of plants observed
Total (n)bps1 (n)χ2
  1. The critical χ2 value at 95% confidence is 3.841 when degrees of freedom = 1.

aba3-2 bps1-2 F28882453.18
abi1-1 bps1-1 F210652650.01
abi4 bps1-2 F210752800.63

We also used ABA-insensitive mutants to confirm that the root-derived signal was not ABA. ABI1 encodes a 2C-type protein phosphatase, and ABI4 encodes an AP2-domain protein, and mutations at either of these genes results in decreased ABA responses (Finkelstein et al., 1998; Leung et al., 1994). We reasoned that if the bps1 signal was ABA, then the abi mutations might condition a reduced response, and the double mutant would exhibit a suppressed phenotype. Instead, F2 plants segregatedbps1 mutants that were indistinguishable from bps1 single mutants (Table 1). Taken together, these genetic analyses confirm that the bps1 signal is not ABA.

To test whether the bps1 root-derived signal was either the same as the MAX-dependent hormone or derived from the same pathway that produces the MAX-dependent hormone, we characterized double mutants between bps1 and max3 and between bps1 and max4. MAX3 and MAX4 both encode carotenoid cleavage dioxygenases (CCD7 and CCD8, respectively), and are required for MAX-dependent hormone synthesis (Booker et al., 2004; Sorefan et al., 2003). We reasoned that, if production of the bps1 signal required similar carotenoid cleavage steps as the MAX-dependent hormone, then the double mutants would show either a suppressed or wild-type phenotype. We analyzed both F2 and F3 populations (Table 2), and found 25% segregation of typical bps1 mutants. These results indicate that synthesis of the bps1 signal does not require MAX3 or MAX4, and thus the bps1 signal is distinct from the MAX-dependent hormone. Furthermore, these data suggest that the bps1 root-derived signal may be a novel carotenoid-derived compound.

Table 2.   Genetic analysis of the MAX-dependent hormone as a candidate for the bps1 signal
Plants analyzedNumber of plants observed
Total (n)bps1 (n)χ2
  1. The critical χ2 value at 95% confidence is 3.841, when degrees of freedom = 1.

max3-11 bps1-2 F212803190.004
max4-5 bps1-2 F212813130.22
max3-11 bps1-2 F3383950.0008
max4-5 bps1-2 F34511031.124

The α-carotene branch of the pathway is not required for bps1 signal production

Because our data suggested that the bps1 root-derived signal might be a novel carotenoid derivative, we carried out further experiments in an attempt to identify its biosynthetic precursor(s). A major branch in carotenoid biosynthesis occurs at lycopene cyclization, with lycopene feeding into either α-carotene or β-carotene and their derivatives (Figure 1). To determine whether the bps1 signal was derived from the α-carotene branch, we characterized bps1 lut1 and bps1 lut2 double mutants (Pogson et al., 1996; Tian et al., 2003, 2004). We reasoned that if the bps1 signal was derived from either α-carotene or lutein, then the mobile signal would not be produced in the corresponding double mutant, and the bps1 phenotype would appear either suppressed or wild-type. For each double mutant analysis, F2 populations segregated bps1 mutants at 25%, and these bps1 mutants exhibited root and shoot phenotypes identical to bps1 single mutants (Table 3). These data indicated that neither LUT1 nor LUT2 is required for synthesis of the bps1 mobile signal, and suggest that the bps1 signal might be derived from the β-carotene branch of this pathway.

Table 3.   Genetic analysis of α-carotene requirement for production of bps1 signal
Plants analyzedNumber of plants observed
Total (n)bps1 (n)χ2
  1. The critical χ2 value at 95% confidence is 3.841, when degrees of freedom = 1.

lut2-1 bps1-2 F210332610.04
lut1-1 bps1-2 F217424230.48

The β-carotene branch of the pathway is required for bps1 signal production

To further analyze the contribution of carotenoid biosynthesis to bps1 signal production, we examined bps1 mutants grown on medium containing the lycopene cyclase inhibitor 2-(4-chlorophenylthio)-triethylamine hydrochloride (CPTA) (Figure 1). In various fruits and tobacco seedlings, CPTA treatment led to accumulation of lycopene and a reduction in downstream carotenoids (Bouvier et al., 1997; Coggins et al., 1970; Corona et al., 1996). Consistent with a block in biosynthesis of photo-protective carotenoids, Arabidopsis seedlings sown on medium supplemented with CPTA were photo-bleached. Also as predicted, bleaching of the seedlings was dependent on light intensity and concentration of the inhibitor in the medium.

To determine the specific effect of CPTA on Arabidopsis carotenoids, we carried out HPLC analysis of pigment extracts (Figure 3). Similar carotenoid profiles were obtained for bps1 mutants and wild-type untreated control plants. For wild-type and bps1 seedlings grown on 10 μm CPTA, we observed incomplete photo-bleaching of seedlings (not shown); consistent with this observation, the pigment profile (Figure 3) indicates a decrease, but not complete loss, of β-carotene (β), violaxanthin (V) and neoxanthin (N), with accumulation of lycopene (L). As expected for an incomplete block of carotenoid biosynthesis, we were still able to detect chlorophyll a and b (Ca, Cb). Note also that levels of lutein showed little change, and that a new peak, that is probably δ-carotene, appeared.

Figure 3.

 Reverse-phase HPLC analysis of carotenoids in bps1 and wild-type seedlings grown on media supplemented with the carotenoid biosynthesis inhibitor CPTA, under low light conditions (30 μE m−2 sec−1).
Peaks representing neoxanthin (N), violaxanthin (V), lutein (Lu), chlorophyll a and b (Ca, Cb), β-carotene (β), lycopene (L) and δ-carotene (δ) are indicated. Untreated wild-type (Columbia) and bps1 seedlings exhibited similar carotenoid profiles (top panel). Treatment with 10 μm CPTA resulted in a decrease in carotenoids downstream of lycopene cyclization, especially those derived from the β-carotene branch (middle panel, see Figure 1 for pathway details). Treatment with 50 μm CPTA resulted in complete loss of β-carotene and its derivatives, and a strong decrease in lutein, an α-carotene derivative (lower panel). Treatment with either concentration of the inhibitor led to accumulation of lycopene, which is not detectable in the untreated seedlings.

For seedlings grown on 50 μm CPTA, we observed complete photo-bleaching of the seedlings (not shown), and pigment profiles show stronger accumulation of lycopene, a complete loss of carotenoids downstream of β-carotene, and loss of chlorophyll pigments. As with the 10 μm treatment, traces of lutein and δ-carotene are also observed at this CPTA concentration, suggesting that CPTA is a more effective inhibitor of β-cyclase than ɛ-cyclase. Results similar to these were obtained for growth under moderate light conditions (100 μE m−2 sec−1, not shown). These data indicate that CPTA is an effective lycopene cyclase inhibitor in Arabidopsis.

Growth of bps1 mutants on CPTA-supplemented medium resulted in partial rescue of both leaf and root defects (Figure 2). bps1 mutants grown on control growth medium have small radialized leaves with very little vascular tissue (Figure 2c,d,i), and very short mis-shapen ‘knotted-looking’ roots (Figure 2j). By contrast, bps1 mutants grown on CPTA-supplemented medium produced larger flattened leaves that contained primary and secondary veins, and smooth elongated roots (Figure 2g,h). We compared the effect of light intensity on bps1 mutants treated with 10 μm CPTA. Under high light conditions (approximately 200 μE m−2 sec−1; Figure 2e–h), treated seedlings were completely photo-bleached and bps1 mutants showed rescue of both leaf and root development (Figure 2g,h). Under moderate (approximately 100 μE m−2 sec−1, not shown) and low (30 μE m−2 sec−1) light intensities, seedlings are incompletely photo-bleached, but leaf and root development continue to show partial rescue (Figure 2). Therefore, it is unlikely that rescue is simply due to photo-oxidation because we still observe rescue when the plants are somewhat green. We also compared the effect of CPTA concentration (10, 25, 50 and 100 μm) on bps1 mutants grown at moderate and low light intensities. Increased concentrations resulted in greater rescue of the bps1 root phenotype, suggesting that the partial rescue is dose-responsive. Thus, the effects of CPTA on the bps1 phenotype match our previous observations on bps1 mutants grown on fluridone (Van Norman et al., 2004), and are consistent with the bps1 signal being derived from a carotenoid downstream of lycopene cyclization. In combination with our lut1 and lut2 double mutant analyses, these data suggest that the bps1 signal is derived from the β-carotene branch of the pathway.

No single NCED or CCD is required for production of the bps1 signal

To further explore bps1 signal biosynthesis, we extended our genetic analysis to include the remaining NCED/CCD gene family members. NCED3 functions in ABA biosynthesis under drought stress conditions (Iuchi et al., 2001). Tissue-specific roles in ABA biosynthesis have been proposed for NCED2, NCED5, NCED6 and NCED9; however, products of their cleavage reactions may not be limited to production of ABA (Lefebvre et al., 2006; Tan et al., 2003). To determine whether production of the bps1 signal required an NCED enzyme, we characterized double mutants between bps1 and a mutant allele for each of the NCED genes. If one of these NCED enzymes was essential for synthesis of the bps1 signal, then we expected the double mutant to show a suppressed or possibly wild-type phenotype. Analysis of F2 populations segregating for both bps1 and each nced mutant revealed, in each case, that bps1 mutants segregated at 25% and that the bps1 nced double mutants were indistinguishable from bps1 single mutants (Table 4).

Table 4.   Genetic analysis of requirement for NCEDs in bps1 signal production
Plants analyzedNumber of plants observed
Total (n)bps1 (n)χ2
  1. The critical χ2 value at 95% confidence is 3.841 when degrees of freedom = 1.

nced2 bps1-2 F210182590.11
nced3 bps1-2 F29002280.05
nced5 bps1-2 F213273131.41
nced6 bps1-2 F211002710.08
nced9 bps1-2 F217744181.95
nced2 bps1-2 F3372890.23
nced6 bps1-2 F37481880.01
nced9 bps1-2 F39442300.20

Four CCD genes (CCD1, CCD4, MAX3/CCD7 and MAX4/CCD8) encode a distinct clade of the NCED/CCD gene family, and are thought to catalyze distinct carotenoid cleavage reactions (Schwartz et al., 2001). We have already shown that production of the bps1 signal persists in max3 and max4 mutant backgrounds (Table 2); to determine whether either CCD1 or CCD4 was required for bps1 signal biosynthesis, we analyzed double mutants. In both F2 and F3 populations, bps1 mutants segregated at 25% of the total plants (Table 5) and were indistinguishable from bps1 single mutants. Together these data indicate that the activity of a single NCED or CCD is not required for production of the bps1 root-derived signal.

Table 5.   Genetic analysis of requirement for CCD1 and CCD4 in bps1 signal production
Plants analyzedNumber of plants observed
Total (n)bps1 (n)χ2
  1. The critical χ2 value at 95% confidence is 3.841 when degrees of freedom = 1.

ccd1-1 bps1-2 F27241982.13
ccd4 bps1-2 F2178343.30
ccd4 bps1-2 F36281490.54

bps1 rescue on CPTA is not due to photo-oxidation-induced plastid-to-nucleus signaling

We have shown that production of the bps1 mobile root-derived signal requires carotenoid biosynthesis, specifically the β-carotene branch. The simplest explanation for this requirement is that the bps1 root-derived signal is synthesized from β-carotene or one of its derivatives. However, in addition to removing pools of potential bps1 signal precursors, blocking carotenoid biosynthesis also leads to plastid photo-oxidation and production of a retrograde plastid-to-nucleus signal (Mayfield et al., 1986; reviewed in Nott et al., 2006; Rodermel and Park, 2003; Strand, 2004). This retrograde signal has been characterized through analysis of genome uncoupled (gun) mutants, which abolish synthesis of the signal. This retrograde signal is known to alter nuclear gene expression in response to plastid damage (Mochizuki et al., 2001; Strand et al., 2003); thus an alternative explanation for the partial rescue of the bps1 phenotype upon loss of carotenoids could be the effect of the retrograde signal on nuclear gene expression.

To distinguish between these possibilities, we generated and characterized bps1 gun2 and bps1 gun5 double mutants. We reasoned that if fluridone- and CPTA-induced plastid-to-nucleus signaling were responsible for partial rescue of the bps1 phenotype, then double mutants with gun2 or gun5 grown in the presence of CPTA would fail to exhibit partial rescue. In bps1 gun2 and bps1 gun5 seedlings grown on CPTA, we observed partial rescue of the bps1 phenotype. Because gun mutants produce small leaves, we found root development to be the most robust indicator of phenotypic rescue. We characterized primary root length (Figure 4a) and development of lateral roots (Figure 4b,c) in bps1 control seedlings and bps1, bps1 gun2 and bps1 gun5 mutants treated with 50 μm CPTA. The observation that loss of GUN2 or GUN5 does not prevent rescue indicates that rescue of the bps1 phenotype on CPTA is not due to retrograde signaling from photo-oxidized plastids. Taken together, our data suggest that bps1 partial rescue in response to carotenoid inhibition is directly due to loss of carotenoids, and that bps1 roots produce a novel carotenoid derivative.

Figure 4.

 CPTA rescue of the bps1 phenotype persists in gun mutant backgrounds.
(a) Length of the primary root in bps1 untreated seedlings (n = 21), and in bps1 (n = 20), bps1 gun5 (n = 35) and bps1 gun2 (n = 38) mutants treated with CPTA.
(b) Number of emerged lateral or adventitious roots in bps1 untreated seedlings (n = 21), and in bps1 (n = 20), bps1 gun5 (n = 35) and bps1 gun2 (n = 38) mutants treated with CPTA.
(c) Length of emerged lateral or adventitious roots in bps1 untreated seedlings (n = 21), and in bps1 (n = 38), bps1 gun5 (n = 57) and bps1 gun2 (n = 82) mutants treated with CPTA.


bps1 mutant roots constitutively produce a graft-transmissible signal that causes defects in both shoot and root development (Van Norman et al., 2004). As a first step toward investigating the biological role for this root-to-shoot signal, we are working to identify the signal. Identification of the mobile signal may allow us to begin to understand how BPS1 controls its production, as well as how it affects target tissues.

Several pieces of data suggest that the bps1 signal is a carotenoid derivative. Inhibitors of carotenoid biosynthesis (fluridone and CPTA) partially rescue the bps1 phenotype, suggesting that these compounds inhibit synthesis of the signal. Double mutant analysis with pds1 was also consistent with a carotenoid-derived signal. Furthermore, the bps1 aba1 double mutant shows an enhanced phenotype, suggesting that the aba1 mutant may condition greater production of the bps1 signal (Van Norman et al., 2004).

Inhibition of the early steps of carotenoid biosynthesis either chemically (fluridone and CPTA) or genetically (pds1) results in photo-bleached plants, as carotenoids function as important photo-protective pigments in addition to their roles as mobile signal precursors. Thus, an alternative explanation for rescue of bps1 under carotenoid-limited conditions results from altered gene expression due to retrograde signaling induced by photo-oxidized plastids. However, we showed that bps1 mutants grown on 10 μm CPTA under low light conditions showed a partially rescued phenotype. Because the bps1 seedlings under these conditions were somewhat green and partially rescued, it suggests rescue is not due retrograde signaling but instead is due to decreased carotenoid production. Additionally, we were able to distinguish between these possibilities by showing that CPTA continues to rescue the bps1 phenotype in gun2 and gun5 mutant backgrounds where retrograde signaling is impaired.

Two known carotenoid derivatives, ABA and the MAX-dependent hormone, were attractive candidates for the signal, because both have been implicated in root-to-shoot signaling (Harris and Outlaw, 1991; Turnbull et al., 2002; Booker et al., 2005; Sorefan et al., 2003). However, our genetic analyses indicated that bps1 mutants continue to produce the mobile signal in genetic backgrounds where ABA and MAX-dependent signals are not produced.

If the bps1 signal is not one of the conventional apocarotenoid signals, then what is it? The signal appears to be generated downstream of lycopene cyclization, as CPTA, a lycopene cyclase inhibitor, confers partial rescue of the bps1 mutant phenotype. However, the signal is unlikely to be an α-carotene derivative, as mutations in lut1 and lut2 failed to confer rescue to bps1 mutants. Thus our data are consistent with the bps1 signal being derived from the β-carotene branch of this pathway. NCED/CCD enzymes were attractive candidates in bps1 signal synthesis. The production of both ABA and the MAX-dependent hormone requires NCED/CCD activity, and functions for some members of this gene family in Arabidopsis (especially CCD1, CCD4) remain unknown. However, no single NCED/CCD mutant was sufficient to prevent synthesis of the bps1 signal. It is still possible that bps1 signal requires CCD/NCED activity, and that our double mutants still produce the mobile signal due to a redundant activity supplied by another gene family member. Testing for suppression by treatment with abamine, which is a competitive inhibitor of NCEDs, would allow us to examine the requirement for all five NCED enzymes simultaneously (Han et al., 2004); however, we were unable to procure this substance.

Cytochrome P450 enzymes are another attractive candidate for bps1 signal synthesis. Both the production of lutein and the production of the MAX-dependent hormone require cytochrome P450 enzymes (Booker et al., 2005; Tian et al., 2004). However, a comprehensive genetic approach to test all the cytochrome P450 genes is impractical, as 246 cytochrome P450 genes are encoded by the Arabidopsis genome.

Root-to-shoot signaling has been implicated in controlling whole-plant responses to environments perceived by the plant root. BPS1 appears to be a negative regulator of a root-derived signal because bps1 mutant roots constitutively produce a graft-transmissible signal that affects shoot development. An important step toward understanding the physiological role of BPS1 signaling is identification of the signal itself. Our data suggest that the bps1 signal is a novel, and so far unidentified, apocarotenoid. How this molecule affects target tissues, and the circumstances under which the signal is normally produced, are important future research goals.

Experimental procedures

Materials used here are available upon request.

bps1 pds1 double mutant analysis

To estimate recombination between bps1 and pds1, we started with 0.25 cM being equivalent to 50 kb (Weigel and Glazebrook, 2002). The BPS1 and PDS1 genes are separated by 1810 kb, which would be equivalent to 9% recombination. To determine whether recombination in the BPS1PDS1 region is similar to average rates, we measured recombination in this region using the L112 CAUT line (Furner et al., 1996; http://arabidopsis.info/CollectionInfo?id=13). This insertion carries the CH42 gene (a dominant marker in the ch42 mutant) inserted into BAC F13M7 (between bps1 and pds1). We constructed homozygous ch42 mutants segregating for bps1, and crossed them to CAUT line L112 (ch42 homozygote with CH42 insertion). This generated F1 plants with bps1 and the CH42 insertion (dominant marker) in repulsion. Any green bps1 F2 plants arose due to recombination. We found 5.7% recombination (n = 44), which is consistent with the L112 insertion being closer to BPS1 than PDS1. This recombination rate indicates normal levels of recombination in this region. Because the homozygous double mutant (bps1 pds1) requires recombination in both the male and female gametes, we expected 0.81% double mutants (i.e. out of every 125 pds1 mutants, approximately one should be a bps1 pds1 double mutant).

Construction of double mutants and genetic analysis

Double mutants were generated using standard approaches. The χ2 values indicated in all genetic analysis summary tables (Tables 1–5) represent values derived for the null hypothesis; the critical χ2 value is 3.841 (at 95% confidence), degrees of freedom = 1.

bps1-2 (Columbia) was the primary allele used for our genetic analysis; however, for crosses to aba3-2, which is in the Landsberg erecta background, we used bps1-1 (Landsberg erecta). Mutant alleles of NCED2, NCED9 and CCD4 correspond to Salk_090937, Salk_051969 and Salk_097984, respectively (Alonso et al., 2003). Each of these T-DNA alleles was annotated as an insertion within the coding region of each gene; the positions of our gene- specific primers (Table S1) and the sizes of the amplified products suggest that the annotation is correct. Identical results were obtained for both mutant alleles of NCED6 (WISC.DSLox 471G6 and 388C08) (Sussman et al., 2000); however, the results shown correspond to 388C08 (Table 4). These NCED6 alleles are annotated as insertions within the coding region, and this annotation is supported by the location of our gene-specific primers (Table S1) and the sizes of the PCR-amplified products. T-DNA alleles of NCED3 and NCED5 were obtained from Donald R. McCarty; the NCED3 insertion lies within the coding region and the NCED5 insertion is located just upstream of the ATG (at −45 nucleotides), both are thought to be knockout alleles (B.C. Tan, University of Florida, personal communication). Harry Klee generously provided ccd1-1, max3-11 and max4-5 seeds (Auldridge et al., 2006). Root development assays were carried out on F3 populations homozygous for either gun2 or gun5 and segregating bps1. Seedlings homozygous for gun2 were selected based on their light green color, and homozygous gun5 seedlings were identified by a PCR-based test for the genetic lesion (Table S1). We define ‘bps1-like knots’ (Figure 4b) as the disorganized swollen lateral roots that are present in bps1 single mutants (Figure 2j).

Plant growth conditions

Seeds were sown on plates of growth medium (GM) composed of 0.5 x MS salts (Caisson Laboratories, Rexburg, ID, USA), 0.5 g l−1 MES (J.T. Baker, Phillipsberg, NJ, USA), 1% sucrose and 0.8% agar (Phytablend, Caisson Laboratories). Following incubation at 4°C for 2–4 days, plants were grown under continuous light (90–120 μE m−2 sec−1) at room temperature (20–25°C) or 29, 22 or 16°C. Seedlings were scored between 8 and 16 days post-imbibition (DPI) depending on the growth temperature. Plants transplanted to soil for the generation of F3 seeds were grown at 22°C under continuous light. F3 generation seedlings were grown at 22°C under continuous light and analyzed at 11 DPI.

Inhibitor treatments

Seedlings were sown on GM containing CPTA (a generous gift from A. Trebst) at 10 or 50 μm, and grown at room temperature (20–25°C) for 15 DPI, then examined for partial rescue of the bps1 phenotype. The light intensities for these experiments were approximately 100 μE m−2 sec−1 (moderate light conditions), approximately 30 μE m−2 sec−1 (low light conditions) and approximately 200 μE m−2 sec−1 (high light conditions).

Pigment analyses

Seedlings were sown on control and CPTA-supplemented GM plates and grown at room temperature (20–25°C) under moderate or low light intensities for 15 DPI. Pigments were extracted as described by Pogson et al. (1996) and stored at −20°C until analysis by HPLC. Carotenoids were separated by reverse-phase HPLC as described by Norris et al. (1995). Carotenoids were identified by comparing retention times and spectra with those of standards and published data.


We are grateful for seeds and reagents received from many different people. We thank Professor A. Trebst (Ruhr-Universitat Bochum) for CPTA, D. DellaPenna (Michigan State University) for lut1, lut2 and pds1 seeds, D.R. McCarty (University of Florida) for nced5 and nced3 seeds, and H. Klee (University of Florida) for max3-11, max4-5 and ccd1-1 seeds. gun2 and gun5 mutants were generously provided by Dr Chory (Salk Institute). We also thank the Salk Institute Genomic Analysis Laboratory for providing the sequence-indexed Arabidopsis TDNA insertion mutants. We also thank Harry Klee and Dean Della Penna for useful discussions about carotenoid biosynthesis. We thank Dale Poulter and Marcus Babst for HPLC advice and use of equipment. Additionally, we thank the Arabidopsis Biological Resource Center for seed materials, and the donors of those materials (abi1-1 and aba3-2 donated by M. Koornneef and abi4 donated by R. Finkelstein). Additionally, we thank the Arabidopsis Knockout Facility at the University of Wisconsin Biotechnology Center for seeds for two alleles of NCED6. We also thank Josh Steffen and Jayson Punwani for critical reading of this manuscript. This work was funded by grants from the NSF (IBN-0445723) to L.E.S. and a National Institutes of Health training grant (5 T32 GM007464) to J.M.V.N.