Concurrent interactions of heme and FLU with Glu tRNA reductase (HEMA1), the target of metabolic feedback inhibition of tetrapyrrole biosynthesis, in dark- and light-grown Arabidopsis plants


(fax +41 01 632 1239; e-mail


The regulation of tetrapyrrole biosynthesis in higher plants has been attributed to metabolic feedback inhibition of Glu tRNA reductase by heme. Recently, another negative regulator of tetrapyrrole biosynthesis has been discovered, the FLU protein. During an extensive second site screen of mutagenized flu seedlings a suppressor of flu, ulf3, was identified that is allelic to hy1 and encodes a heme oxygenase. Increased levels of heme in the hy1 mutant have been implicated with inhibiting Glu tRNA reductase and suppressing the synthesis of δ-aminolevulinic acid (ALA) and Pchlide accumulation. When combined with hy1 or ulf3 upregulation of ALA synthesis and overaccumulation of protochlorophyllide in the flu mutants were severely suppressed supporting the notion that heme antagonizes the effect of the flu mutation by inhibiting Glu tRNA reductase independently of FLU. The coiled-coil domain at the C-terminal end of Glu tRNA reductase interacts with FLU, whereas the N-terminal site of Glu tRNA reductase that is necessary for the inhibition of the enzyme by heme is not required for this interaction. The interaction with FLU is specific for the Glu tRNA reductase encoded by HEMA1 that is expressed in photosynthetically active tissues. FLU seems to be part of a second regulatory circuit that controls chlorophyll biosynthesis by interacting directly with Glu tRNA reductase not only in etiolated seedlings but also in light-adapted green plants.


In plants tetrapyrroles are involved in various light-dependent processes such as photosynthesis and photomorphogenesis (Beale, 1999). Usually they are bound to proteins and in this state may use various quenching mechanisms to dissipate absorbed light energy. Free tetrapyrroles, however, may act as highly potent photosensitizers that upon illumination may generate singlet oxygen that can lead to severe photooxidative damage (op den Camp et al., 2003; Rebeiz et al., 1988). The most highly evolved group of plants, angiosperms, face the particular problem that in the absence of light chlorophyll biosynthesis leads only to the biosynthesis of its immediate precursor, protochlorophyllide (Pchlide). The subsequent reduction of Pchlide to Chlide requires light and is catalyzed by the light-dependent NADPH-Pchlide oxidoreductase (POR) (Griffiths, 1978). Overaccumulation of Pchlide in dark-grown seedlings is suppressed by negative feedback control of the first enzyme committed to tetrapyrrole biosynthesis, Glu tRNA reductase, by heme (Beale, 1999). This conclusion was based on the following experiments. Inactivation of a heme oxygenase gene in the yellow-green-2 mutant of tomato suppresses Pchlide accumulation in etiolated seedlings and this effect has been ascribed to enhanced levels of heme in the mutant (Cornah et al., 2003; Terry and Kendrick, 1999). Conversely, a reduction in heme by the removal of free Fe2+/3+ causes an increase in the Pchlide content (Duggan and Gassman, 1974). Finally, the activity of the purified Glu tRNA reductase is inhibited in vitro by heme (Huang and Wang, 1986; Pontoppidan and Kannangara, 1994; Weinstein and Beale, 1985).

Although these data emphasize the importance of free heme as an inhibitor of ALA synthesis, they do not exclude the possibility that other effectors of feed back control of ALA synthesis may exist that together with heme regulate tetrapyrrole biosynthesis (Yaronskaya et al., 2003). One way to identify such regulators has been the search for mutants with a defect in the control of Chl biosynthesis. The first of these mutants were identified in barley (von Wettstein et al., 1974). Several different tigrina mutants were described that in the dark accumulate more Pchlide than wild-type plants (Nielsen, 1974a,b). Three of these mutants, tig b, tig n, and tig o, displayed pleiotropic effects that included not only increases in Pchlide levels but also differences in the composition of carotenoids and the fine structure of etioplast membranes (Nielsen, 1974a). In the tig d mutant, however, only the control of Pchlide accumulation seemed to be affected (Kahn et al., 1976). More recently, the flu mutant of Arabidopsis was identified that resembles tig d (Meskauskiene et al., 2001). Subsequent work has shown that the TIGRINA d gene of barley is an ortholog of the FLU gene (Lee et al., 2003). In the present work the likely target of feedback control, Glu tRNA reductase, was found to interact with FLU. We have used a combined genetic and biochemical approach to assess the role of this interaction during ALA synthesis. The results of this study support the existence of a second regulatory circuit with FLU interacting with Glu tRNA reductase independently of heme that operates not only in etiolated seedlings but also in light-adapted green plants.


Light-induced changes of Glu tRNA reductase and Glu 1-semialdehyde aminotransferase mRNA levels in the flu mutant

Overaccumulation of Pchlide in dark-grown flu mutants of Arabidopsis has been attributed to the deregulation of ALA synthesis (Meskauskiene and Apel, 2002; Meskauskiene et al., 2001). There are only two enzymes known to be committed exclusively to ALA synthesis in plants, Glu tRNA reductase and glutamate-1-semialdehyde-2-1-aminotransferase (GSA) (O'Neill and Söll, 1990). Two different mechanisms have been described that are responsible for the regulation of these enzymes during a dark to light transition, light-induced changes in their mRNA concentrations (McCormac et al., 2001; Nadler and Granick, 1970; Papenbrock et al., 1999) and feedback control of enzyme activities (Castelfranco and Jones, 1975; Fluhr et al., 1975; Kahn et al., 1976). In tigrina d, a mutant in the barley ortholog of FLU, deregulation of Pchlide accumulation in etiolated seedlings initially had been ascribed to the inactivation of a repressor of genes that encode enzymes involved in the early steps of tetrapyrrole biosynthesis prior to ALA formation (Kahn et al., 1976). Hence, the enhanced levels of Pchlide in flu mutants transferred to the dark could be caused by a constitutive upregulation of mRNAs for Glu tRNA reductase and GSA. However, in the flu mutant the two transcripts showed the same fluctuation and reached similar levels during a light-to-dark transition as the wild-type plants (Ilag et al., 1994) (Figure 1). Similarly, in barley attempts to confirm the proposed function of TIGRINA d as a repressor have failed (Hansson et al., 1997). Thus, the accumulation of Pchlide and the increase in the rate of ALA formation in flu mutants do not seem to be due to a change in the biosynthesis of the two enzymes involved in ALA formation. Instead, FLU seems to take part in the metabolic feedback control of ALA synthesis by inhibiting one of these enzymes. This proposed role of FLU was tested first genetically by isolating and characterizing suppressor mutants that antagonize the effect of the flu mutation on Pchlide accumulation and second by analyzing the physical interaction of FLU with Glu tRNA reductase (HEMA1) that has been proposed to be the target of feedback control of tetrapyrrole biosynthesis (Cornah et al., 2003; Pontoppidan and Kannangara, 1994; Vothknecht et al., 1998).

Figure 1.

Changes in HEMA1 and GSA1 transcript levels in flu and wild-type Arabidopsis plants. flu(1–3) and wild-type (wt, 4–6) plants were either grown for 22 days under continuous light (1, 4) or were transferred after such a light treatment for 3 days to the dark without (2, 5) or with another 1 day light treatment following the dark period (3, 6). HEMA1 and GSA1 transcript levels were determined by Northern blot analysis. The HEMA1 and GSA1-specific probes were the same as described by Ilag et al. (1994). Ethidium bromide-stained total RNA showing 18S and 16SRNAs provides a sample loading control.

Identification and characterization of suppressor mutants of flu

Seedlings of flu growing in the dark overaccumulate Pchlide that in its free form acts as a potent photosensitizer (op den Camp et al., 2003; Rebeiz et al., 1988). When these seedlings are transferred from the dark to the light they rapidly bleach and die. The flu mutant can be rescued, however, under constant light such that Pchlide is immediately reduced to Chlide and no longer accumulates to toxic levels (Meskauskiene et al., 2001). The light sensitivity of flu seedlings grown under non-permissive light/dark cycles has been exploited for the identification and isolation of second-site mutants, which no longer accumulate the same high amounts of Pchlide in the dark as flu seedlings, but instead restrict Pchlide accumulation to similar low levels as wild-type seedlings.

The screening strategy was based on the assumption that tetrapyrrole biosynthesis is controlled by several modulators of ALA synthesis, some of which may antagonize the activity of the negative regulator FLU and stimulate ALA synthesis. Inactivation of such an effector molecule would be expected to reduce the overaccumulation of Pchlide in dark-grown flu seedlings. Seedlings of such second-site mutants of flu should no longer be susceptible to photooxidative damage and thus should easily be discernible under selective dark/light conditions. M2 seedlings of flu1-1 mutagenized with EMS were grown on MS agar plates. The flu1-1 line accumulates similar transcript levels as wild type, but does not contain detectable levels of the flu protein (Figure 2a). Wild-type seedlings grew normally under 8 h dark/16 h light conditions (Figure 2b), whereas seedlings of the original flu mutant rapidly died. In addition, most of the M2 descendants of mutagenized flu plants did not survive under these selective growth conditions except for a few that grew like wild type (Figure 2b). These plantlets were transferred to soil and were kept under continuous white light until seeds could be harvested.

Figure 2.

Identification of suppressor mutants of flu.
(a) Mutagenesis was carried out with the flu1-1 mutant that lacks detectable levels of the flu protein. Protein extracted from 4-day-old wild-type (wt) and flu1-1 seedlings were separated electrophoretically and probed with an anti-FLU polyclonal antibody.
(b) Seedlings of flu and wild type (wt) were grown under non-permissive 16 h light/8 h dark cycles together with M2 seeds of EMS-mutagenized flu plants. The majority of M2 seedlings of mutagenized flu plants died rapidly similar to the flu seedlings, whereas revertants of flu were recovered among the M2 seedlings that grew like wild type and contained much lower levels of Pchlide than flu, when grown in the dark (see c). These mutants were dubbed ulf. Of 120 000 M2 seedlings that were screened under these light/dark conditions, seven different ulf mutants were identified.
(c) Seedlings of flu, wild type and the selected second-site suppressor of flu, flu/ulf3, grown in the dark for 4 days. Etiolated seedlings were exposed to blue light (400–450 nm), and the emitted fluorescence was recorded. The bright red fluorescence emitted by the flu mutant is caused by the excitation of free Pchlide.

Seedlings of the selected second-site mutants of flu and wild-type plants were grown in the dark on MS agar plates for 4 days. Etiolated flu mutants can easily be distinguished from wild-type seedlings by the strong red fluorescence that they emit after being exposed to blue light (Figure 2c). Several selected second-site mutants with a reduced red fluorescence were identified. These mutants were named ulf (reversal of flu). Out of a total of 120 000 M2 seedlings seven ulf mutants were identified. Total porphyrins were extracted from etiolated seedlings and the fluorescence emission spectra of the extracts were recorded. The concentration of Pchlide in the flu mutant was up to 10-fold higher than in the selected second-site mutants and wild-type seedlings. The identity of Pchlide and the differences in concentrations were confirmed by HPLC analysis of these extracts (data not shown).

Each of the selected second-site mutants was backcrossed at least once to the original flu mutant. Segregation analysis of the F2 generation of these crosses verified that in each case the second site mutant's phenotype was caused by a single recessive gene mutation. The ulf mutants were characterized further by allelism tests. They comprised four different loci; for two of them, ulf1 and ulf2, three and two allelic mutant lines, respectively, were identified, whereas for each of the other loci, ulf3 and ulf4, only a single mutant line could be found. None of the ulf mutations was allelic to the flu mutation. Thus, they define distinct effector molecules, whose biological activities antagonize that of the flu mutation.

Characterization of the ulf3 mutant

The phenotype of ulf3 was different from that of the other ulf mutants in that the hypocotyl of this mutant grown under continuous light was significantly longer (Figure 3a). In this respect ulf3 resembled the hy mutants of Arabidopsis (Koornneef et al., 1980). Etiolated seedlings of one of the hy mutants, hy1, contain reduced levels of Pchlide (Montgomery et al., 1999) that have been ascribed to an increase in free heme in this mutant and inhibition of Glu tRNA reductase by heme-mediated feedback regulation (Cornah et al., 2003; Terry and Kendrick, 1999). If FLU forms part of a separate regulatory circuit that modulates ALA formation by inhibiting Glu tRNA reductase independently of heme, elevated levels of heme in hy1 are likely to antagonize the effect of the flu mutation and hence such a mutant could be expected to be selected during the suppressor screen. We therefore decided to test the possible allelism of ulf3 and hy1.

Figure 3.

Characterization of ulf3, hy1, and flu mutants.
(a) Allelism test of hy1 and ulf3 mutants. hy1 was crossed with ulf3 and the parental lines and also wild-type seedlings were compared with 6-day-old seedlings of the F1 generation of the cross hy1/ulf3. The phenotypes of the double mutant and the parental lines were indistinguishable.
(b) Genotyping of the flu mutant by PCR. PCRs were performed on wild-type DNA (2, 3) with wild type-specific (lane 2) and flu-specific primers, respectively (lane 3). A strong signal is obtained in lane 2, while only a weak one is visible in lane 3. PCRs performed on flu DNA (4, 5) with wild type-specific primers resulted in no signal (lane 4), while the same reaction performed with flu-specific primers led to a strong signal (lane 5). Lanes 1 and 6 are negative controls. M, marker fragments.

As ulf3 had been identified in a homozygous flu/flu background, the ulf3 mutation had to be separated first from the flu mutation by crossing the double mutant with wild-type Landsberg erecta (Ler). Within the segregating F2 progeny of this cross seedlings with long hypocotyls were pre-selected and descendants that no longer carried copies of the mutated flu gene were subsequently identified by using polymerase chain reaction (PCR) primers that either contained the wild-type sequence or the base substitution of the mutated flu gene. This difference in primer sequences led to a differential thermal stability of hetero and homo duplexes of amplification products that could be exploited for the distinction between the wild-type FLU gene and its mutated flu allele (Figure 3b). A similar differential PCR procedure was also used to identify homozygous hy1/flu double mutants among the F2 descendants of a cross between hy1 and flu (data not shown).

The identity of ulf3 and hy1 was established by mapping ulf3 on chromosome 2 at the same site as hy1 and by an allelism test in which the F1 progeny of a cross between homozygous ulf3 and hy1 mutant plants was analyzed (Figure 3a). Finally, the heme oxygenase gene of the ulf3 mutant was sequenced and its nucleotide sequence compared with that of wild-type plants. At position 326 of the heme oxygenase open reading frame derived from the cDNA sequence of ulf3 a G–A base exchange led to a premature stop codon (data not shown) that inactivated the gene.

Effects of ulf3 and hy1 on ALA formation and Pchlide accumulation in flu and wild-type seedlings kept in the dark

The rate of ALA formation was first determined in ulf3 and hy1 in wild-type background. In seedlings of ulf3 ALA formation seemed to be slightly lower than in hy1 and wild type (Figure 4a). In the flu mutant the rate of ALA formation was more than four times higher than in Ler wild type (Figure 4a). After introducing either ulf3 or hy1 into the flu line and isolating the homozygous double mutant lines, the rates of ALA synthesis were reduced to approximately one-third of the rate in flu alone (Figure 4a). Thus, both mutations, hy1 and ulf3, suppress upregulation of ALA formation in the flu mutant. Moreover, the accumulation of Pchlide in the flu mutant was downregulated in a similar way after the introduction of ulf3 and hy1, respectively (Figure 4b).

Figure 4.

ALA synthesis and Pchlide accumulation in hy1, ulf3, flu, flu/ulf3, flu/hy1, and wild-type (wt) seedlings.
(a, b) A comparison of the rates of ALA synthesis (a) and Pchlide contents (b). The rates of ALA synthesis were measured in seedlings grown for 6 days in continuous light and returned to the dark for 30 min. Pchlide was measured by HPLC in 4-day-old etiolated seedlings.
(c) Changes in Pchlide contents of wild-type (wt) and flu seedlings induced by exogenous ALA. Pchlide was measured in 4-day-old etiolated wt and flu seedlings after infiltration with exogenous ALA (hatched bars) or without ALA (white bars). SEM for at least three different experiments are shown.

The actual rates of Pchlide accumulation and ALA synthesis in the flu/ulf3 (or hy1) double mutant seem to result from a balance between two opposing forces, relief of Glu tRNA reductase from metabolic feedback inhibition by the inactivation of FLU and inhibition of the enzyme by increased levels of heme in hy1 (or ulf3). Residual amounts of heme in flu without the hy1/ulf3 genetic background should still prevent tetrapyrrole biosynthesis from reaching its maximum capacity. This proposal could be tested experimentally by feeding exogenous ALA to etiolated wild type and flu seedlings. In wild-type seedlings a 10- to 15-fold increase in Pchlide accumulation occurred after ALA feeding, and in flu seedlings the high levels of Pchlide were further increased two- to three-fold in response to added exogenous ALA (Figure 4c). Thus, in flu accumulation of Pchlide is still slightly suppressed, most likely by heme, whose inhibitory impact on tetrapyrrole accumulation could be evaded by adding exogenous ALA.

The inactivation of a heme oxygenase gene in the hy1 mutant may affect ALA formation not only through an increase in free heme that suppresses Glu tRNA reductase activity but also through a reduced phytochrome content of the mutant (Parks and Quail, 1991). The light-induced transcription of the Glu tRNA reductase gene (HEMA1) is mediated by phytochromes (McCormac and Terry, 2002; McCormac et al., 2001), and hence reduced amounts of these photoreceptors in hy1 could contribute to the suppression of tetrapyrrole biosynthesis in the flu/hy1 double mutant by restricting the de novo synthesis of Glu tRNA reductase. The concentration of mRNA for Glu tRNA reductase in hy1 was determined by real-time PCR and compared with that of wild type and flu. In etiolated seedlings the concentrations of these transcripts were very low and reached similar levels in all plant samples (Figure 5 ‘D’). In light-grown seedlings the concentrations of these transcripts increased approximately 12-fold and reached similar high levels (Figure 5‘L’). Hence, accumulation of Pchlide and formation of ALA in these plants is influenced primarily by the relative proportion of the two effector molecules FLU and heme, each of which seems to control tetrapyrrole biosynthesis separately by interacting with the common target of metabolic feedback control, Glu tRNA reductase.

Figure 5.

Quantification of HEMA1 mRNA expression by real-time PCR in 4-day-old etiolated flu, hy1, flu/hy1, and wild-type seedlings (D) and seedlings grown for 6 days in continuous light (L).
The amounts of mRNA measured in the mutants were normalized against the amount of HEMA1 mRNA in wild type (set as 1). No significant differences were found between the mutants. Each of the measurements was repeated at least 10 times. SEM of three (D) or six (L) different experiments are shown. The mRNA levels in light-grown wild-type seedlings were 12-fold higher than in etiolated seedlings.

Interaction of FLU with Glu tRNA reductase

In a previous study using the yeast two-hybrid system, physical interaction between FLU and Glu tRNA reductase could be demonstrated that required a functional tetratricopeptide repeat (TPR) domain of the FLU protein (Meskauskiene and Apel, 2002). In the present work we have extended these observations first by identifying the domain of Glu tRNA reductase that is needed for the interaction of the enzyme with FLU and second by showing this domain to be distinct from the site that is required for the inhibition of the enzyme by heme. Two different deletion mutants of Glu tRNA reductase were analyzed. Within the C-terminal end of the enzyme an additional coiled-coil motif has been predicted that seems to be unique for Glu tRNA reductases of higher plants (Brody et al., 1999). This domain was removed in a mutated version of Glu tRNA reductase (HEMA1-43). The N-terminal site of Glu tRNA reductase is necessary for the inhibition of the enzyme by heme. A truncated form of the enzyme that lacked the 30 N-terminal amino acids maintained its activity but was no longer inhibited by heme (Vothknecht et al., 1998). The second mutated form of Glu tRNA reductase that was used for the binding assays lacked this putative heme-inhibition site (HEMA1-noHI). In Arabidopsis there is a second Glu tRNA reductase gene, HEMA2, that is expressed only in roots and flowers in a light-independent fashion, whereas HEMA1 is regulated by light and expressed predominantly in leaves (Kumar et al., 1996; McCormac et al., 2001; Tanaka et al., 1996). This second Glu tRNA reductase was analyzed in the binding assay (HEMA2).

A truncated version of FLU which ranges from amino acid position 140 in the hydrophobic region to the C-terminal end of the protein was fused in frame with the GAL4 DNA binding domain (BD; vector pGBKT7) and the GAL4 activation domain (AD; vector GAD T7). The mutated forms of Glu tRNA reductase (HEMA1-43, HEMA1-noHI) and HEMA1 and HEMA2, which lacked their chloroplast signal peptides, were also fused in frame with BD and AD (Figure 6a). BD-FLU and AD-FLU were co-expressed with the various HEMA1 derivatives, as well as with HEMA2, in yeast. FLU interacted strongly with HEMA1, thus confirming previously published results, and also with HEMA1-noHI (Figure 6b). Removal of the coiled-coil domain at the C-terminal part of Glu tRNA, however, disrupted the physical interaction between FLU and Glu tRNA reductase (Figure 6b). In contrast to HEMA1, HEMA2 did not interact with FLU. The interaction between FLU and the truncated form of HEMA1-noHI was abolished, when FLU was replaced by the mutated flu1-1 variant that carried a single amino acid exchange within its TPR motif (data not shown) (Meskauskiene et al., 2001).

Figure 6.

The interaction of FLU with different variants of Glu tRNA reductase as revealed by the yeast two-hybrid analysis and co-precipitation.
(a) Constructs for the yeast two-hybrid analysis. HEMA1: Glu tRNA reductase which is encoded by the HEMA1 gene. HEMA1-43: Variant of HEMA1 that lacks the 43 C-terminal amino acids. HEMA1-noHI: Variant that lacks the 30 N-terminal amino acids of the mature HEMA1. HEMA2: Glu tRNA reductase encoded by the HEMA2 gene. None of the proteins contained its plastid transit peptide.
(b) The results of the yeast two-hybrid analysis. The indicated construct pairs were co-transformed into yeast strain AH109 and the yeast transformants were selected on medium lacking leucine and tryptophan. On this medium, only those cells that carry both pGBKT7 and pGADT7 vectors can grow (pGBKT7 and pGADT7 carry TRP1 and LEU2 selectable marker genes, respectively, that complement trp and leu mutations of AH109); however, there is no selection for interaction between the test proteins. For each pair of constructs, several clones of transformants were resuspended in water and plated as spots on medium lacking leucine and tryptophan (−Leu −Trp), as well as on medium lacking leucine, tryptophan, histidine, and adenine (−Leu −Trp −His −Ade). On the latter medium only those yeast cells can grow in which the transcription of two marker genes for interaction, HIS3 under GAL4-inducible GAL1 promoter and ADE2 under GAL4-inducible GAL2 promoter, is activated. Such transcriptional activation of the marker genes occurs, if the test proteins interact. As a positive control, the plasmids pGADT7-T and pGBKT7-53 (Clontech) were co-transformed, as a negative control the plasmids pGADT7-T and pGBKT7-Lam (Clontech) were co-transformed.
(c) Autoradiographs showing interactions of FLU with HEMA1 (1), HEMA1-43 (2), HEMA1-noHI (3), and HEMA2 (4). The upper panel shows the input proteins that were synthesized in an in vitro transcription/translation system and the lower panel shows proteins that interacted with FLU and were indirectly retained on an affinity column as described under Experimental procedures (arrow head). [35S]-labeled input proteins were separated electrophoretically on SDS polyacrylamide gels.

Interactions between FLU and the different variants of Glu tRNA reductase were also analyzed by co-precipitation of bound partners via a GST-tag attached to the truncated N-terminal part of the FLU protein. Each of the prey proteins was synthesized and labeled with [35S]-methionine by using an in vitro transcription/translation assay (Figure 6c). Equal amounts of radioactively labeled prey proteins were incubated with GST-FLU bait protein attached to an affinity resin. The bait-prey protein complexes were solubilized by adding an elution buffer containing 100 μm glutathione. The eluted proteins were subsequently separated electrophoretically on SDS polyacrylamide gels and analyzed by autoradiography. The truncated FLU protein coprecipitated only HEMA1 and the N-terminal deleted form HEMA1-noHI, whereas HEMA2 and the C-terminal-deleted variant HEMA1-43 were not retained by the FLU protein (Figure 6c). The interactions between FLU and the various Glu tRNA reductase derivatives as revealed by this indirect precipitation assay were identical to those found in yeast with the two-hybrid analysis.

Presence and function of FLU in light-adapted green plants

When etiolated flu seedlings are transferred from the dark to the light they rapidly bleach and die (Meskauskiene et al., 2001). This striking light sensitivity of the mutant emphasizes the importance of FLU-dependent control of tetrapyrrole biosynthesis for the survival of seedlings emerging from the dark. It is not known yet whether the FLU-dependent feedback control of tetrapyrrole biosynthesis is confined to etiolated seedlings or whether it also operates in light-adapted green plants. To address this question, we first determined mRNA and protein levels of FLU in wild-type seedlings that were either grown in the dark or in the light. In etiolated and light-grown 4-day-old seedlings similar amounts of FLU protein were detected immunologically, whereas FLU mRNA levels differed drastically between light-grown and etiolated seedlings (Figure 7). During illumination of etiolated seedlings the concentration of FLU mRNA increased steadily, whereas protein levels remained constant. When plants were shifted back from the light to the dark, both mRNA and protein levels did not decline during a dark period of 25 h (Figure 7). The constant level of the FLU protein in light- and dark-grown seedlings was in marked contrast to the light-dependent changes of Glu tRNA reductase concentrations. The enzyme protein was hardly detectable in etiolated seedlings and started to accumulate during illumination, leading to a drastic change in the ratio of Glu tRNA reductase to the FLU protein (Figure 7). These results indicate first that in etiolated seedlings the amounts of FLU protein seem to be far in excess of its target, Glu tRNA reductase, and secondly that the light-induced increase in Glu tRNA reductase levels contributes to an increased capacity of tetrapyrrole biosynthesis. The continuous presence of FLU and the light-induced increase in its mRNA levels suggest that FLU is still involved in controlling tetrapyrrole biosynthesis of light-adapted plants. This suggestion could be confirmed experimentally by first measuring the rate of ALA synthesis in light-adapted green wild-type seedlings of Arabidopsis. The rate was approximately three to four times higher than in seedlings transferred to the dark (Figure 8a). After inactivation of the FLU protein this ALA synthesis rate increased fourfold, whereas in light-grown hy1 mutants the rate was reduced to approximately half of that in wild-type seedlings. In light-grown seedlings of the hy1/flu double mutant, the upregulation of ALA synthesis in the flu background was suppressed (Figure 8a). When these plants were transferred back to the dark to block the light-dependent reduction of Pchlide to Chlide, different amounts of Pchlide accumulated that matched the different rates of ALA formation in light-grown plants (Figure 8b,c). Thus, both FLU and heme seem to control jointly the rate of ALA synthesis and the accumulation of Pchlide not only in etiolated but also in light-adapted green seedlings.

Figure 7.

Light-induced changes in protein and mRNA levels of FLU (a, b) and Glu tRNA reductase (c, d) of wild-type seedlings grown for various lengths of time in the dark (DD) or in the light (LL). Total proteins were separated electrophoretically on SDS polyacrylamide gels and FLU (b) and Glu tRNA reductase (d) proteins were detected immunologically on blots using polyclonal antibodies. Levels of mRNA (a, c) were determined by Northern blot analysis, using gene-specific hybridization probes. The different light/dark regimes to which plants had been exposed are indicated on top of the figure.

Figure 8.

FLU- and heme-dependent changes in the rate of ALA synthesis and Pchlide accumulation in light-grown seedlings of wild type (wt), flu, hy1, and hy1/flu.
(a) A comparison of the rates of ALA synthesis in 6-day-old seedlings grown under continuous light.
(b) Measurement of Pchlide by HPLC in green seedlings first grown for 4 days under continuous light and then transferred to the dark for 6 h.
(c) FLU- and heme-dependent differences in the amounts of Pchlide that accumulate following the transfer of 4-day-old light-grown seedlings to the dark for 6 h.
SEM for at least three different experiments are shown.


Metabolic feedback inhibition is often exerted on the first enzyme of a biosynthetic route by the final product of this pathway. Among the three enzymes involved in ALA formation, Glu tRNA reductase catalyzes the first step in the pathway that is unique to tetrapyrrole biosynthesis and thus appears to be the primary candidate for a target of feedback control by FLU. At the same time FLU may be controlled either by heme or Pchlide. The rate of ALA formation was found to be inversely proportional to the level of photoactive Pchlide, suggesting that Pchlide bound to the POR protein may mediate the control of ALA synthesis (Ford and Kasemir, 1980). However, after the putative target enzyme of feedback control, Glu tRNA reductase, had been purified, this enzyme was not inhibited by Pchlide but by heme (Pontoppidan and Kannangara, 1994; Vothknecht et al., 1998). Based on these in vitro studies, control of tetrapyrrole biosynthesis was subsequently ascribed to the regulatory role of heme.

In previous studies the N-terminal 30 amino acids of Glu tRNA reductase were necessary for the inhibition of the enzyme by heme (Vothknecht et al., 1998). A mutated form of Glu tRNA reductase that lacked this putative site was still able to interact with FLU. Removal of a coiled-coil domain at the opposite end of the protein, however, disrupted its interaction with FLU. These results are consistent with FLU acting independently of heme. In such a case the heme-dependent control of Glu tRNA reductase should antagonize the effect of the flu mutation and restrict the overaccumulation of Pchlide. In the flu mutant this predicted inhibitory effect of heme on ALA formation is difficult to test directly. However, after crossing flu with the hy1 mutant ALA synthesis and Pchlide accumulation in flu were suppressed. The reduction in Pchlide levels in yellow-green-2, a mutant in the tomato ortholog of HY1, is attributed to an inhibition of ALA formation by enhanced levels of heme due to the inactivation of heme oxygenase (Cornah et al., 2003; Terry and Kendrick, 1999). The combination of flu with hy1 in the homozygous double mutant may thus result in the downregulation of Pchlide accumulation and ALA synthesis with flu releasing ALA formation from feedback control and hy1 enhancing the inhibition of this reaction. This proposed antagonistic effect was supported independently by the outcome of the suppressor screen. One of the selected second-site mutants, ulf3, was allelic to hy1. Collectively these results suggest that the initial steps of tetrapyrrole biosynthesis leading to ALA formation are controlled concurrently by two different inhibitors, FLU and heme, that interact with different sites of the common target enzyme Glu tRNA reductase.

Pchlide would be an attractive candidate for a tetrapyrrole intermediate that activates the FLU-dependent metabolic feedback control. Photoactive Pchlide is localized in the hydrophobic environment of prolamellar bodies, plastid envelopes and thylakoid membranes and is thereby most likely excluded from interacting directly with Glu tRNA reductase in the stroma. The FLU protein could be necessary to bridge the gap between the membrane and the stroma and to facilitate the interaction between Pchlide and the hydrophilic target enzyme, Glu tRNA reductase (Meskauskiene et al., 2001). This involvement of FLU in the feedback control of the Mg2+ branch of tetrapyrrole biosynthesis would be in line with the reported failure of Pchlide to inhibit directly Glu tRNA reductase in vitro (Huang and Wang, 1986; Weinstein and Beale, 1985).

Whereas Pchlide may act as an effector molecule that autoregulates its own level in etiolated seedlings via FLU and Glu tRNA reductase, it does not seem to control ALA formation in light-adapted green plants. The FLU-dependent control of ALA formation persists in green plants kept under continuous light, although the concentration of Pchlide has dropped beyond the limit of detection and this pigment no longer poses a risk of photooxidative damage to the plant. In light-grown seedlings the reason for maintaining the heme- and FLU-dependent control of tetrapyrrole biosynthesis may be related to a continuous need of Chl biosynthesis to support the ongoing assembly of photosynthetic complexes. First, in Arabidopsis there are three structurally related but differentially regulated PORs, denoted PORA, PORB, and PORC. Unlike PORA and PORB, that are present in etiolated seedlings, PORC starts to accumulate only after etiolated seedlings are exposed to light (Su et al., 2001). This suggests that in light-adapted seedlings a need for additional Chl biosynthesis arises that can only be satisfied by the accumulation of an additional POR protein. Secondly, under steady-state photosynthesis conditions the D1 protein of the reaction center of PSII turns over rapidly even under low light conditions (Jansen et al., 1999; Keren et al., 1997). At the same time larger amounts of newly synthesized Chl molecules are directed toward the reassembly of the reaction center of PSII (Feierabend and Dehne, 1996). The biosynthesis of Chl and Chl-binding proteins have to be closely coordinated to avoid an imbalance between these two processes that may lead to photooxidative damage caused by excited free Chl molecules. Although the identity of the effector molecule that interacts with FLU and may assist in coordinating Chl and protein synthesis is not known yet, it seems likely that in light-grown plants it is different from Pchlide.

Experimental procedures

Plant materials

For the cultivation of mature plants seeds of Arabidopsis thaliana ecotype Landsberg erecta (Ler) were sown on soil under continuous light or 8 h dark/16 h light cycles (80–110 μmol quanta m−2 sec−1). Seedlings were cultivated by germinating surface-sterilized seeds on plates of Murashige and Skoog (1962) (MS) agar, in some cases supplemented with 0.5% sucrose. Plated seeds were either kept under continuous light or under 8 h dark/16 h light cycles (80–110 μmol quanta m−2 sec−1).

Isolation of the ulf3 mutant

Chemical mutagenesis of seeds from flu1-1 mutant plants (Meskauskiene et al., 2001) was performed as described by Runge et al. (1995) Although the mutated flu1-1 protein could not be detected in flu1-1 seedlings, the flu1-1 form was detected immunologically in yeast transformants carrying the flu1-1 cDNA and the amount of the mutated protein was similar to that of wild-type FLU in yeast transformants carrying FLU cDNA (data not shown). The ulf3 locus was genetically mapped in F2 plants from a cross of ulf3 in Ler flu/flu with Col-0 flu/flu plants. This latter line had been obtained by four backcrosses of flu in Ler with wild-type Col-0 plants. The single base pair exchange in the ulf 3 mutant was found by direct sequencing of the PCR product amplified from genomic DNA with primers derived from the heme oxygenase sequence of HY1.

Quantification of HEMA1 transcripts by real-time PCR

Two sets of seedlings were grown for 6 days under continuous light on MS agar containing 0.5% sucrose and another set was grown in the dark for 4 days on MS agar without sucrose. The seedlings were frozen with liquid nitrogen and total mRNA was isolated with the RNeasy® Plant Mini Kit (Qiagen, Hilden, Germany). cDNA was synthesized by using the ImProm-IITM Reverse Transcription System (random primers used; Promega, Madison, WI, USA). To determine the amount of HEMA1 mRNA, a real-time PCR assay was performed with the SYBR Green PCR Master Mix (ABI, Oslo, Norway). The samples were normalized against profilin mRNA (At2g9760).

Detection of Pchlide

Seedlings were grown in the dark for 4 days. Thirty seedlings were homogenized with a polytron homogenizer (Ultra-Turrax T25, IKA Labortechnik, Staufen i. Br., Germany) in 0.8 ml of 90% acetone (Spectranal; Riedel-de Haën, Seelze, Germany) containing 0.1% NH3 (32% stock solution). After centrifugation (>14 000 g) for 20 min at 4°C, the supernatant was recovered and stored at −20°C. To detect Pchlide, a gradient HPLC system with a fluorescence detector was used (Reverse phase column 25C18-25QK, Interchrom; Spectra System SCM400, Spectra System P4000, Spectra System FL3000; Thermo Separation Products, Allschwil, Switzerland). A 60-μl volume of the sample was injected into the 50 μl loop of the HPLC system. The gradient was as follows: 5 min 60% acetone, 20 min gradient from 60 to 100% acetone, 5 min gradient from 100 to 60% acetone. Acetic acid (50 μl) was added to 1 l running solution. The excitation wavelength was 430 nm and the emission wavelength was 630 nm.

Measurement of the rate of ALA synthesis

Seedlings were grown for 6 days under continuous light on MS agar containing 0.5% sucrose. After being shifted to darkness for 30 min, the seedlings were vacuum-infiltrated under green light with an 80-mm levulinic acid solution (10 mm KH2PO4; 0.5% Tween; pH 7.2). The seedlings were incubated in darkness. Samples were taken under green light every 15 min, immediately frozen in liquid nitrogen and stored at −80°C. Accumulated ALA was quantified as follows: Samples were homogenized in 4% TCA (v/v), boiled for 15 min, cooled on ice for 2 min and filtrated with 0.45 μm cellulose acetate Microfuge Tubes (Rainin Instrument, Emeryville, CA, USA). An aliquot of the filtrate was neutralized with the same amount of 0.5 m NaH2PO (pH 7.5). Ethylacetoacetate (1/5) was added and then the sample was boiled for 10 min. After cooling on ice for 5 min 1 volume freshly made Ehrlich's reagent (0.2 g p-dimethylaminobenzaldehyde; 8.4 ml acetic acid; 1.6 ml 70% perchloric acid) was added. The sample was then centrifuged at 4°C (>14 000 g) for 5 min and afterward kept for 10 min at room temperature. OD was determined at 553 nm (coefficient = 7.45 × 104 mol−1 cm−1) (Kruse et al., 1997; Mauzerall and Granick, 1956). Similar measurements were performed with seedlings kept under continuous light.

Identification of the flu genotype by PCR

The sequence of the flu-specific forward primer was 5′-GCA GCG CCA AGG GAA GTA TAG GGA AGT-3′ and the sequence of the wild type-specific forward primer was 5′-GCA GCG CCA AGG GAA GTA TAG GGA AGC-3′. The sequence of the reverse primer for the flu- and the wild type-specific reaction was 5′-GGC AAT TGG CAC TTA GCA AGA TGG C-3′. The initial denaturation of DNA was performed for 5 min at 94°C. During PCR denaturation of DNA was performed at 94°C for 30 sec, annealing at 63.5°C for 30 sec, and elongation at 72°C for 60 sec. These steps were repeated 34 times.

Feeding of seedlings with ALA

Seedlings were grown for 3 days in the dark and vacuum-infiltrated under green light with 10 mm ALA (25 mm KH2O4, pH 7.5, 0,5% Tween). They were then returned to darkness for 12 h, frozen in liquid nitrogen and stored at −80°C. The amount of accumulated tetrapyrroles was determined by extracting the tetrapyrroles and separating them by HPLC.

Construction of fusion proteins

The BD-HEMA1, AD-HEMA1, BD-FLU, and AD-FLU constructs were the same as described previously (Meskauskiene and Apel, 2002). For the construction of the truncated derivatives of HEMA1, BD-HEMA1-43, and AD-HEMA1-43, HEMA1 cDNA was amplified with primers 5′-GGA ATT CGC CAT GGC GTC TAA TGC AGC TAG CAT CTC TGC TC-3′ and 5′-TCC CCC GGG TTA TCT CAA ATG CTG CAT TGG ACC ATG-3′ using the BD-HEMA1 construct as a template. The amplified cDNA was cloned into EcoRI/SmaI-digested BluescriptR IISK+/− vector, sequenced and subcloned into EcoRI/SmaI-digested vectors pGBKT7 and pGADT7. For the construction of the truncated forms BD-HEMA1-noHI and AD-HEMA1-noHI, HEMA1 cDNA was amplified with primers 5′-GGA ATT CGA AAG AAG CAG TAT TGT TGT GAT TG-3′ and 5′-TCC CCC GGG TTA CTT CTG TTG TTG TTC CGC C- 3′ using the BD-HEMA1 construct as template. The cDNA was cloned into EcoRI/SmaI-digested BluescriptR IISK+/− vector, sequenced and subcloned into EcoRI/SmaI-digested vectors pGBKT7 and pGADT7, and sequenced. HEMA2 cDNA was amplified with primers 5′-GGA ATT CTC TAA TAA GGC AGC TAG CAT CTC A-3′ and 5′-TCC CCC GGG CTA CTT TTT TTC CAC CTT TGC TCT AA-3′ from total cDNA prepared from light-grown wild-type Arabidopsis seedlings. HEMA2 cDNA was cloned into EcoRI/SmaI-digested BluescriptR IISK+/− vector, sequenced and subcloned into EcoRI/SmaI-digested vectors pGBKT7 and pGADT7. The constructs AD-HEMA1, AD-HEMA1-43, AD-HEMA1-noHI, and AD-HEMA2 were used for in vitro translation of the corresponding [35S]-methionine-labeled proteins with a coupled transcription/translation system (Promega).

Yeast two-hybrid analysis

The Matchmaker GAL4 two-hybrid system 3 of Clontech (Palo, Alto, CA, USA) was used. The constructs in vectors pGBKT7 and pGADT7 were co-transformed into the yeast strain AH109 and yeast transformants were selected according to the Clontech protocol.

Pull-down experiments

The same constructs HEMA1, HEMA1-43, HEMA1-noHI, and HEMA2 were used for the synthesis of prey proteins as described for the yeast two-hybrid analysis. For the synthesis of the bait protein the same FLU cDNA encoding a truncated form of the FLU protein was amplified as described earlier (Meskauskiene and Apel, 2002) and subcloned into the pET42a vector (Novagen, Madison, WI, USA) adjacent to the GST sequence. The bait protein construct was first amplified in DH5αEscherichia coli cells and then retransformed into BL21(DE3) E. coli cells. Synthesis of the protein was induced by the addition of IPTG. [35S]-labeled prey proteins were synthesized in a coupled transcription/translation system (Promega). For the pull-down assay the ProFound pull-down GST protein: protein interaction kit (Pierce, Milwaukee, WI, USA) was used. Coprecipitated [35S]-labeled proteins were separated electrophoretically on 10% SDS polyacrylamide gels and detected by autoradiography.

Production of anti-FLU antibody and protein analysis

For expression of the FLU protein in E. coli, cDNA encoding the hydrophilic part of FLU of Arabidopsis was amplified from total cDNA, cloned into BamHI/EcoRV-digested BluescriptR IISK+/− vector, sequenced, and subcloned into NcoI/BamHI-digested expression vector pET-21d(+) (Novagen). The resulting construct was transferred into E. coli strain BL21(DE3). The FLU protein was expressed according to pET system manual TB055 (Novagen) and purified with HisBind Purification Kit (Novagen) according to TB054 1101 manual. The purified FLU protein was used for anti-FLU antibody production. Total protein extracts from seedlings were prepared according to Frick et al. (2003) except that 0.01% (v/v) β-mercaptoethanol was used in the homogenization buffer. Proteins were separated electrophoretically on SDS polyacrylamide gels and blotted onto PVDF membranes. For immunodetection of FLU, the anti-FLU polyclonal antibody was diluted 1:3000 and used in combination with GAR-HRP conjugate (Bio-Rad, Hercules, CA, USA) and Immun-StarTM HRP chemiluminescence detection kit (Bio-Rad). Anti-GLU-TR antibody was obtained from S. Beale (Brown University, Providence, RI, USA) and used at a dilution of 1:2000.


We are indebted to Samuel Beale for the gift of an antiserum against Glu tRNA reductase, to Dieter Rubli for artwork, to Martha Geier-Bächtold for editorial assistance, to André Imboden for taking care of the plants, and to Roel op den Camp for making some of the crosses between different mutant lines. This work was supported by grants from the Swiss Federal Institute of Technology (Zurich) and the Swiss National Science Foundation.