The Arabidopsis translocator protein (AtTSPO) is regulated at multiple levels in response to salt stress and perturbations in tetrapyrrole metabolism

Background The translocator protein 18 kDa (TSPO), previously known as the peripheral-type benzodiazepine receptor (PBR), is important for many cellular functions in mammals and bacteria, such as steroid biosynthesis, cellular respiration, cell proliferation, apoptosis, immunomodulation, transport of porphyrins and anions. Arabidopsis thaliana contains a single TSPO/PBR-related gene with a 40 amino acid N-terminal extension compared to its homologs in bacteria or mammals suggesting it might be chloroplast or mitochondrial localized. Results To test if the TSPO N-terminal extension targets it to organelles, we fused three potential translational start sites in the TSPO cDNA to the N-terminus of GFP (AtTSPO:eGFP). The location of the AtTSPO:eGFP fusion protein was found to depend on the translational start position and the conditions under which plants were grown. Full-length AtTSPO:eGFP fusion protein was found in the endoplasmic reticulum and in vesicles of unknown identity when plants were grown in standard conditions. However, full length AtTSPO:eGFP localized to chloroplasts when grown in the presence of 150 mM NaCl, conditions of salt stress. In contrast, when AtTSPO:eGFP was truncated to the second or third start codon at amino acid position 21 or 42, the fusion protein co-localized with a mitochondrial marker in standard conditions. Using promoter GUS fusions, qRT-PCR, fluorescent protein tagging, and chloroplast fractionation approaches, we demonstrate that AtTSPO levels are regulated at the transcriptional, post-transcriptional and post-translational levels in response to abiotic stress conditions. Salt-responsive genes are increased in a tspo-1 knock-down mutant compared to wild type under conditions of salt stress, while they are decreased when AtTSPO is overexpressed. Mutations in tetrapyrrole biosynthesis genes and the application of chlorophyll or carotenoid biosynthesis inhibitors also affect AtTSPO expression. Conclusion Our data suggest that AtTSPO plays a role in the response of Arabidopsis to high salt stress. Salt stress leads to re-localization of the AtTSPO from the ER to chloroplasts through its N-terminal extension. In addition, our results show that AtTSPO is regulated at the transcriptional level in tetrapyrrole biosynthetic mutants. Thus, we propose that AtTSPO may play a role in transporting tetrapyrrole intermediates during salt stress and other conditions in which tetrapyrrole metabolism is compromised.


Background
Higher plants synthesize four major tetrapyrroles (chlorophyll, haem, sirohaem and phytochromobilin) via a common branched pathway [1][2][3] (Additional file 1). In metazoans, heme and siroheme are synthesized in mitochondria, but in plants tetrapyrrole biosynthesis is plastid-localized, suggesting that tetrapyrroles are transported from the chloroplast to the mitochondria. This suggests that late stages of the heme biosynthetic pathway are present in both chloroplasts and mitochondria (Additional file 1). The concentration of tetrapyrrole intermediates is tightly controlled because these compounds are photoreactive and can generate reactive oxygen species (ROS). Additionally, many of the enzymes in this pathway are regulated by environmental stimuli and development signals [4,5].
In mammals, an 18-kDa peripheral-type benzodiazepine receptor (TSPO/PBR) is localized in the outer mitochondrial membrane [6] where it binds other proteins, such as the 34-kDa voltage-dependent anion channel and the inner membrane adenine nucleotide carrier [7]. TSPO was originally named the "peripheral benzodiazepine receptor" (PBR), however, it has more recently been renamed "TSPO" reflecting its structural and functional similarity to the bacterial tryptophan-rich sensory protein [8].
TSPO primarily functions to transport heme, porphyrins, steroids and anions [8][9][10][11]. However, TSPO proteins are also important for cellular respiration [12], cell proliferation [13] and apoptosis [14]. For example, in erythroids, in response to stress, TSPO is important for transporting porphyrins, which induce the expression of heme biosynthesis genes. Likewise, in mouse erythroleukemia cells TSPO has been shown to transport protoporphyrin IX playing a key role in tetrapyrrole and heme biosynthesis [15].
In the α-proteobacterium Rhodobacter sphaeroides TSPO is localized in the outer membrane and its expression is induced by oxygen [16]. Under conditions of high oxygen, TSPO negatively regulates the expression of photosynthetic genes by exporting excess intermediates of the tetrapyrrole pathway, such as Mg-Protoporphyrin IX (Mg-ProtoIX) and MgProtoIX Monomethyl ester [17]. The rat TSPO homologue complements the Rhodobacter tspo mutant, suggesting that the function of TSPO is conserved in R. sphaeroides and metazoans [18].
Evidence for a functional TSPO protein in Arabidopsis thaliana and other plants has been previously reported [19]. Transport studies with the recombinant Arabidopsis TSPO in Escherichia coli revealed a benzodiazepinestimulated high-affinity uptake of protoporphyrin and cholesterol, leading to the hypothesis that the Arabidopsis homologue functions in the transport of protoporphyrinogen IX to the mitochondria where heme can be synthesized. However, the role of AtTSPO in plant metabolism is still unknown.
In animals and yeast, TSPO is found in the outer membrane of the mitochondria [6,20]. However the localization of TSPO in plants remains controversial. Lindenman et al. [19] used immunogold staining to show that TSPO is localized in the outer membrane of plastids and mitochondria in Digitalis lanata leaves. However, follow up Western blot experiments could only detect TSPO in mitochondrial fractions. In a separate study, TSPO was found in nuclear fractions prepared from Solanum tuberosum meristematic tissues, while low levels were detected in chloroplast fractions [21]. In Physcomitrella patens, transient expression of TSPO fused to the N-terminus of GFP, PpTSPO:GFP, localized to the mitochondria [22]. In Arabidopsis, fusion of TSPO to the C-terminus of YFP resulted in YFP:TSPO being found in the endoplasmic reticulum and the Golgi stacks [23].
The Arabidopsis genome contains a single TSPOrelated gene (AtTSPO). The predicted protein shares a high degree of similarity to the central domain of its bacterial and mammalian homologs. However, AtTSPO has a 40 amino acid N-terminal extension that is not present in either bacteria or mammals. Moreover, within these 40 amino acids are three in-frame ATG-codons that could code for the first methionine (at positions M1, M21 and M42) (Additional file 2) [19]. To determine whether this region contains organellar targeting information, we developed a series of fusion proteins using the 3 different start sites. Our results demonstrate that AtTSPO was found in different organellar compartments depending on environmental stress. These results, along with analysis of an insertional mutation and expression studies, show that AtTSPO plays an important role in allowing Arabidopsis to cope with high salt stress.

Induction AtTSPO gene expression by abiotic stress
In Physcomitrella patens, the expression of PpTSPO-1 is induced by salt stress and abscisic acid (ABA) [22]. AtT-SPO is also induced by salt stress in Arabidopsis [24], as well as in Arabidopsis cell cultures [23]. We further defined the transcript abundance of AtTSPO in 5-day old seedlings treated with NaCl, mannitol, ABA and methyl viologen (MV), by extracting total RNA from these plants and performing quantitative real-time PCR (qRT-PCR).
Compared to untreated plants, 150 mM NaCl, 250 mM mannitol, 1 μM ABA and 0.2 μM methyl viologen (MV) resulted in increased AtTSPO expression ( Figure  1A). The kinetics of AtTSPO induction by NaCl and ABA stress were similar, peaking 3 hours after treatment and slowly decreasing between 6-25 h ( Figure 1A-a and 1A-b respectively). Addition of mannitol resulted in peak AtTSPO expression between 3-6 h and then slowly decreased in abundance ( Figure 1A-a, 1A-b and 1A-c).
However mannitol treatment showed a two-fold induction compared to treatment with NaCl for 3 h ( Figure  1A-a and 1A-c), which suggests that AtTSPO is induced by osmotic stress rather than salt stress. AtTSPO is rapidly induced by MV treatment, showing induction at  To determine if AtTSPO accumulation was a transcriptional response to NaCl stress, a construct, containing 437 bp upstream the putative translational start site of the AtT-SPO gene was fused to the uidA reporter gene (AtTSPO-437::GUS), and transformed into plants, allowing in vivo analysis of AtTSPO transcriptional response to stress conditions. AtTSPO-437::GUS was found to be induced by 150 mM NaCl within 3 h of treatment, which is similar to qRT-PCR results of the endogenous gene ( Figure 1B). In control experiments, 150 mM NaCl resulted in a small decrease of expression of LHCB::GUS ( Figure 1B). Together these results suggest that the 437 bp region of AtTSPO promoter is sufficient for transcriptional regulation of TSPO.

Identification and characterization of AtTSPO mutants
To determine the function of AtTSPO in vivo, we obtained a T-DNA insertional mutant (SALK_135023) [25] in AtTSPO. This line (tspo-1) was found to have two tandem T-DNA insertions, 123 bp upstream from the translational initiation codon of the AtTSPO gene ( Figure 2A). Homozygous lines were then confirmed to be knock-down mutants by quantitative real time PCR (qRT-PCR) analysis. In this mutant, TSPO mRNA levels are about 20% of wild type ( Figure 2B). AtTSPO fused or not to the N-terminus of GFP was constitutively overexpressed from the CaMV 35S promoter in transgenic Arabidopsis lines (OxM1TSPO and OxM1TSPO:eGFP). We obtained 10 over-expression lines, but focused on the two homozygous lines that exhibited~500 fold over-expression of AtTSPO ( Figure  2B). The tspo-1, OxM1TSPO:eGFP and wild-type lines were grown side-by-side on either Murashige & Skoog (MS) agar medium or soil, and were monitored for possible abnormal phenotypes. The knock-down plants had longer roots compared to the wild type and the overexpression lines ( Figure 2C). Moreover, tspo-1 accumu-lated~30% less chlorophyll than either the wild type or the overexpression lines in the presence of 150 mM NaCl ( Figure 2D).
The expression of stress-response genes is enhanced in tspo-1 AtTSPO expression was previously shown to be regulated by osmotic stress in germination and seedling growth assays [23]. Because TSPO regulates the expression of photosynthetic genes in R. sphaeroides [16], we hypothesized that tspo-1 or OxM1TSPO:eGFP mutants might have an impaired salt stress response. We examined the expression of some well-known salt stress-regulated genes (RAB18, ERD10 and DREB2A) [26]. As expected, stress marker genes were induced by 150 mM NaCl in wild-type plants ( Figure 3A, B and 3C). In AtT-SPO over-expression lines, the levels of DREB2A and RAB18 were lower but no significant change ERD10 expression was observed ( Figure 3A, B and 3C).
In tspo-1 mutants, 3 h of 150 mM NaCl treatment resulted in the increased expression of all three stress marker genes ( Figure 3A, B and 3C). Taken together these results show that AtTSPO plays an important role in regulating the expression of stress response genes.
Expression of light-regulated tetrapyrrole genes are repressed in the tspo-1 knock-down mutant Consistent with TSPO transporting tetrapyrroles [17,19,27], tspo-1 plants accumulated less chlorophyll than wild-type plants ( Figure 2D). Because we found that TSPO is involved in the salt stress response and because TSPO negatively regulates photosynthetic genes in R. sphaeroides [17]. We next analyzed the expression of a few key chlorophyll biosynthesis genes in tspo-1 plants.
Initially, we determined the mRNA levels of most of the key genes in the tetrapyrrole pathway (Additional file 1) in tspo-1 and gun5 mutants. GUN5 encodes the H subunit of chloroplastic Mg-chelatase, which is involved in the perception of altered levels of tetrapyrrolic intermediates [28]. All tetrapyrrole biosynthetic genes known to be light-dependent [29] were found to be down-regulated in tspo-1, as well as in gun5 mutants [28] ( Figure 4A, C, E, F, G and 4H), whereas the expression of the two light-independent genes were unaffected in wild-type and tspo-1 mutant ( Figure 4B and 4D).
Correlation of tetrapyrrole pathway flux and AtTSPO mRNA levels tspo-1 mutants present reduced levels of light-regulated tetrapyrrole metabolism genes ( Figure 4A, C, E, F, G,  and 4H) and also have low chlorophyll content ( Figure  2D). In order to investigate if decreasing flux of tetrapyrrole intermediates would affect AtTSPO expression in wild-type plants, we used two different drugs that interfere with tetrapyrrole biosynthesis, Gabaculine and Norflurazon. Gabaculine acts as a tetrapyrrole biosynthesis inhibitor by blocking the glutamate-1-semi aldehyde aminotransferase activity [30,31]. The herbicide Norflurazon inhibits carotenoid biosynthesis and indirectly affects enzymes in tetrapyrrole biosynthesis [32][33][34]. AtTSPO mRNA levels increased 2-fold in plants treated with 50 μM of gabaculine and up to 500-fold after 500 nM norflurazon treatment ( Figure 5A).
To explore if AtTSPO expression is affected by genetic alterations of the tetrapyrrole biosynthesis pathway, we analyzed the expression of AtTSPO in different mutant backgrounds (Additional file 1). We found that AtTSPO levels are differently altered in various tetrapyrrole pathway mutants. AtTSPO steady-state levels were increased in gun2 (allele of hy1 -required for phytochromobilin synthesis from heme) [35], gun4 (mutant in the Protoporphyrin IX-and Mg-Protoporphyrin IX-binding protein) [36], fc1 (mutant in the ferrochelatase) [37], hemA1hemA2 double mutant (mutant in both glutamyl-tRNA reductases genes) [38] and lin-2 (mutant in the coproporphyrinogen III oxidase) [39] ( Figure 5B). The increased expression of AtTSPO in these mutants with reduced tetrapyrrole levels is consistent with AtTSPO transporting tetrapyrroles for roles in other compartments. The only biosynthetic mutant that resulted in reduced AtTSPO levels was crd1 (mutant in the Mgprotoporphyrin IX monomethyl ester cyclase) [40] (Figure 5B). All these mutations, in exception of crd1 [41], inhibit somehow ALA synthesis, suggesting that disturbances in tetrapyrrole biosynthesis or accumulation affect AtTSPO mRNA expression.
AtTSPO localization depends on the translational start site used AtTSPO (At2g47770) encodes a protein with a predicted molecular weight of 18 kDa. This protein has three possible in-frame ATG-start codons (M1, M21 and M42) in its N-terminal extension region (Additional file 2) [19].
Since reports of plant TSPO localization have resulted in different findings subcellular localization of plant TSPO [19,21,23] we re-examined the subcellular location of AtTSPO and evaluated the roles of the N-terminal extension in targeting AtTSPO within the cell. Past studies [20,23] have utilized N-terminal GFP fusions that might block potential organellar targeting of AtT-SPO, particularly mitochondrial or plastid localization. To allow proper targeting of AtTSPO fusions to GFP, AtTSPO was placed on the N-terminus of GFP. Three constructs were made, representing each of the potential start codons M1 (OxM1TSPO:eGFP), M21 (OxM21T-SPO:eGFP) and M42 (OxM42TSPO:eGFP) and expressed from the CaMV 35S promoter in Arabidopsis.
AtTSPO:eGFP subcellular localization was observed in root, hypocotyls and cotyledons of these lines by confocal microscopy. Full-length AtTSPO:eGFP (OxM1TSPO: eGFP) was found in the endoplasmic reticulum (ER) of the root tip ( Figure 6A) and cotyledons ( Figure 6C) in five day-old seedlings. However, in the hypocotyls of these plants, the fusion protein was found in the ER and in vesicles of unknown identity ( Figure 6B). When M21 (OxM21TSPO:eGFP) or M42 (OxM42TSPO:eGFP) were used, the fusion proteins always co-localized with mitotracker, indicating a mitochondrial localization ( Figure  6D, E, F, G, H and 6I) (Additional file 3). These results corroborate the previous observations of mitochondrial localization of TSPO in D. Lanata leaves by immunogold staining and in Arabidopsis by western blot experiments [19], as well as the endoplasmic reticulum located protein [23], indicating that the alternative use of three initiation codons could be important for AtTSPO localization and its post-translational control.
OxM1TSPOeGFP becomes associated with plastids following high salt stress Having established a key role for AtTSPO in response to abiotic stress, we next examined the localization of AtT-SPO:eGFP fusion proteins in plants subjected to various stress conditions. 5 day-old seedlings were treated with 250 mM mannitol, 1 μM ABA, 0.2 μM MV and 150 mM NaCl. After 18 hours of treatment, OxM1TSPO: eGFP became localized to the plastid ( Figure 7G, H, I, J, K and 7L), while neither OxM21TSPO:eGFP nor OxM42TSPO:eGFP had altered localization even with 5 day extended NaCl treatment (data not shown). AtT-SPO:GFP localization did not change when plants were treated with mannitol, ABA or MV (data not shown).
To verify the expression levels of AtTSPO during salt stress, total protein from each lines was immunoblotted with antibodies to GFP ( Figure 8A). In all cases, AtT-SPO:GFP protein was found to increase significantly after 24h of salt treatment. Accumulation of AtTSPO: GFP was dependent on the presence of AtTSPO because empty vector controls using CaMV or Ubiquitin 10 [42] promoters to drive the expression of GFP did not change in response to salt stress ( Figure 8A and not shown). These results indicate that AtTSPO accumulation is regulated at the transcriptional, post-transcriptional and post-translational levels.
To confirm the location of AtTSPO we performed protease protection assays on isolated chloroplasts from OxM1TSPO:eGFP lines that were grown with or without 150 mM NaCl treatment. Following chloroplast isolation and protease protection, equal quantities of chloroplasts were subjected to immunoblotting with antibodies to GFP. AtTSPO was detected in chloroplast fractions near its predicted monomeric molecular mass (Additional file 4 and 5) in plants treated 18 hours with 150 mM NaCl, but not in untreated plants ( Figure 8B). Chloroplasts prepared from OxTSPO:eGFP lines occasionally displayed a lower molecular mass band that is approximately the mass of GFP. This band probably results from proteolysis between AtTSPO and the GFP tag during sample preparation, although we cannot rule out other possibilities since we do not have an antibody to AtTSPO protein itself. Antibodies to RuBisCo and D1 confirmed the integrity and presence of chloroplasts following protease protection. Antibodies to the cytosolic protein UGPase also verified these fractions were free of cytoplasmic contamination (Additional file 5). These data together with confocal microscopy indicate that the region between the M1 and M21 is important for targeting AtTSPO to chloroplasts during salt stress. Since AtTSPO was protected from trypsin digestion ( Figure  8B), AtTSPO may be integral to the chloroplast outer envelope.

Discussion
The localization of TSPO in both chloroplasts and mitochondria is consistent with its role in porphyrin trafficking. Plant TSPO has been proposed to participate in the interaction between plastid and mitochondrial tetrapyrrole biosynthetic pathways [19]. In higher plants, tetrapyrroles are synthesized almost exclusively in plastids, with the exception of the two last steps of heme synthesis that may occur in both chloroplasts and mitochondria. If AtTSPO is involved in tetrapyrrole transport [19], it is reasonable to assume that AtTSPO may translocate tetrapyrrole intermediates across organellar membranes, explaining why plants would need chloroplastic and mitochondrial isoforms of TSPO.
Consistent with this hypothesis, the AtTSPO protein is longer than its mammalian and bacterial counterparts. The targeting determinants for chloroplasts and mitochondria are usually located at the N-terminus of the protein; therefore, a fusion protein with GFP fused to the C-terminus of TSPO was made. Using this strategy we demonstrated that AtTSPO had different sub-cellular localization patterns depending on the translational start codon, tissue type or the abiotic stress to which the plant was subjected. Placement of the GFP fusion on either the N-or C-terminus of TSPO probably explains the inconsistencies in previous studies [19,21,23] compared to those presented here. The construct used by Guillaumot [23] had the YFP fused to the N-terminus of AtTSPO, which could potentially mask the transit peptide that targets TSPO to chloroplasts and mitochondria. Indeed, a similar case was observed recently for the Arabidopsis HEMERA protein (HMR). When HMR had CFP on its C-terminus, it was localized exclusively in chloroplasts, however, fusion of HMR to the C-terminus of YFP (YFP-HMR) was localized to the nucleus and cytoplasm but not chloroplasts [43].
Salt-stress of Arabidopsis results in movement of ERlocalized AtM1TSPO:eGFP to the chloroplast. We also demonstrated that the different start codons within the TSPO N-terminal extension could target the TSPO protein to different organelles. Other plant proteins such as MDAR (Arabidopsis Monodehydroascorbate Reductase) and tRNA nucleotidyltransferase [44,45] are also known to be targeted to different organelles owing to alternative transcriptionsal start sites. Thus, it is tempting to speculate that, cloroplastic AtTSPO may protect the chloroplast from salt stress damage and the mitochondrial AtTSPO may normally import chloroplast-synthesized porphyrins into the mitochondria. Alterations in the sub-cellular localization of TSPO have been observed in mammals. The mammalian TSPO localizes to the mitochondrial outer membrane but during fast cell proliferation, such as metastatic processes, it relocates to the nuclear membrane, suggesting developmental control of its sub-cellular localization [46].
Our results suggest the existence of a chloroplast targeting region in AtTSPO that operates during salt stress. Constructs lacking the N-terminus of AtTSPO (AtM2T-SPO:eGFP and AtM3TSPO:eGFP) are not able to be target to this organelle. These results suggest that the first twenty aminoacids of AtTSPO may be part of the chloroplast targeting peptide of this protein. Further experiments should be conducted to precisely characterize this chloroplast targeting determinant.
TSPO localization in plant cells is complex, involving a relocation of the protein from ER and vesicles to chloroplasts during salt-stress. In recent years, several new mechanisms for import of proteins into chloroplasts have been proposed. For example, it is hypothesized that close contacts between the envelopes of chloroplasts, mitochondria and other organelle membranes could allow protein movement between them [47]. Such fusions have been observed between the mitochondria and ER [48], where it was suggested that vesicle associated membrane protein 1 (VAMP-1) might be involved in the docking of mitochondria to target membranes [49]. This, in turn, could facilitate a re-localization of proteins from mitochondria to other compartments. The recent discoveries of close intracellular membrane contacts in plants, namely between chloroplasts and the ER [50], as well as between mitochondria and the nucleus [51], corroborates this hypothesis. At the present moment it is not clear which pathway is used during AtTSPO relocation during salt-stress. However our data indicate that AtTSPO changes its localization during stress, and that it is also possible that the mitochondrial isoform observed previously by Frank et al. [22]in P. patens and by Lindenman et al. [19] in D.
lanata and Arabidopsis could be generated in Arabidopsis by the use of alternative translation start codons.
Transcriptional levels of AtTSPO in wild-type Arabidopsis plants increase in response to salt, mannitol, ABA and paraquat. The promoter region of AtTSPO was also found to be sufficient for salt stress transcriptional response. The induction of AtTSPO by salt stress was also observed when constitutive promoters (35S and UBI10) were used to express AtTSPO, suggesting that the induction of AtTSPO occurs at both transcriptional and post-transcriptional levels.
AtTSPO over-expression lines have decreased levels of stress response genes (ERD10, DREB2A and RAB18), while tspo-1 mutants over express these genes. This suggests that AtTSPO expression and/or function is necessary for the proper regulation of these genes during stress conditions. These results also imply that AtTSPO is important for stress adaptation in Arabidopsis, and this idea is consistent with results from P. patens [22]. Since Rhodobacter TSPO is a negative regulator of photosynthetic genes [17], it is possible that AtTSPO operates similarly in regulating stress responsive genes in plants.
The precise function of AtTSPO in tetrapyrrole transport during salt stress remains to be established. There are, however, many reports suggesting that alterations in tetrapyrrole flow can be involved in salt tolerance. Exogenous 5-Aminolevulinate (ALA) can improve salt tolerance in higher plants [51][52][53][54][55][56]. It has been also shown that transgenic Arabidopsis, tobacco and rice that overproduce ALA have improved salt tolerance [57,58]. Abdelkader et al. [59] assumed that high salt stress inhibited chlorophyll accumulation mainly by reducing the rate of porphyrin formation, and Zhang et al. [58] showed that salt stress caused a significant decrease in heme content. Thus in higher plants, ALA and tetrapyrrole synthesis is sensitive to salt stress.
Additionally, we demonstrated that AtTSPO is important for tetrapyrrole flux and/or metabolism. The herbicide Norflurazon, a non-competitive inhibitor of phytoene desaturase, [31][32][33] and the neurotoxin Gabaculine, which inhibits tetrapyrrole biosynthesis by blocking glutamate-1-semi aldehyde aminotransferase activity [29,30] were used in this study to decrease the flux through the tetrapyrrole biosynthesis pathways. Our data showed that mutations in tetrapyrrole biosynthesis genes and the application of these two different drugs that decrease flux of tetrapyrrole intermediates affect AtTSPO expression. All mutations tested that inhibit the synthesis of ALA increase AtTSPO mRNA steady-state levels. The same was observed when the formation of ALA is inhibited by the norflurazon and the gabaculine. The only mutant tested with decreased AtTSPO expression is crd1, which accumulates Mg-Protoporphyrin monomethyl ester and this accumulation does not affect the inhibition of ALA synthesis [40]. Finally, it is possible that AtTSPO could be involved in the partitioning of different tetrapyrrolic signal molecules within plant cells. The steady state levels of several light-regulated mRNAs of tetrapyrrole metabolism genes are downregulated in the tspo-1 mutant, suggesting that, AtTSPO could act as a regulator of tetrapyrrole biosynthesis similar to its bacterial counterpart [16].

Conclusions
TSPO has been shown to transport a number of small molecules in multiple organisms, however its function in plants is not known. Here we demonstrate that Arabidopsis TSPO is regulated at the transcriptional, post-transcriptional and post-translational levels in response to abiotic stress conditions such as salt stress. Our results suggest that AtTSPO can localize to ER and mitochondria, but when plants are salt stressed AtTSPO is found in chloroplasts. Also our data suggest that under normal conditions AtTSPO may be important for the import of chloroplastic synthesized heme into the mitochondria. However, targeting AtTSPO to the chloroplast during salt stress may protect chloroplasts from damage. In addition, tetrapyrrole intermediates has been suggested to operate in the chloroplast-to-nucleus retrograde signaling [35,60]. It is possible that AtTSPO could be involved in the partitioning of different tetrapyrrole signal molecules within plant cells depending on environmental conditions. AtTSPO may play a role in re-directing tetrapyrrole intermediates during salt stress or under conditions where tetrapyrrole metabolism is compromised. This is suggested by our finding that mutation or inhibition of the tetrapyrrole biosynthesis pathway increases AtTSPO expression. At the same time, AtTSPO may directly contribute to the detoxification of highly reactive porphyrins in the cytoplasm. We are currently investigating these possibilities.

Plant material and growth conditions
Arabidopsis thaliana seeds ecotype Col-0 were surface sterilized and plated on MS 1/2 medium [61] with or without 50 mM kanamycin. Seedlings were maintained for three days at 4°C and than grown under 16/8 hours light/dark cycles at 23°C in growth chambers. Root length measurements were conducted using plants grown on vertically oriented in standard conditions for 10 days. For abiotic stress treatment, 150 mM NaCl, 250 mM mannitol, 1 μM ABA (Sigma; St Louis, MO) or 0.2 μM paraquat was added to MS 1/2 agar plates, and the 5day-old seedlings were incubated under normal growth condition. For Norflurazon or Gabaculine experiments seeds were plated on MS 1/2 containing 1 or 2% sucrose with or without 5 μM norflurazon (Sandoz Pharmaceuticals; Vienna, Austria) or 50 μM of gabaculine (Sigma, USA). All experiments were repeated three times independently and the average was calculated.

RNA extraction and qRT-PCR analysis
Total RNA was isolated using Spectrum™ Plant Total RNA Kit (Sigma #STRN250-1KT), according to manufacturer's instructions. One microgram of total RNA was added to each cDNA synthesis reaction using the First Strand cDNA Synthesis Kit (#K1611). For qRT-PCR, DNA amplification was performed in the presence of SYBR ® Green qPCR Detection (Invitrogen) in a MyIQ™ Single Color Real-Time PCR Detection System (BioRad), using the primer pairs at table 1. The cycle

Verification of TSPO knock-out
The tspo-1 T-DNA mutant, SALK_135023, was obtained from the Salk collection [24]. Homozygous mutants were isolated by PCR-based genotyping using gene specific PCR primers AtTSPO-LP and AtTSPO-BP together with LBa1 (Table 2). Only homozygous lines were used for the phenotypic investigation.

Construction of AtTSPO GUS Fusion Vector and GUS Assay
The 437 bp upstream of the translational star site of the AtTSPO gene (At2g47770) was translational fused into uidA gene in pKGWFS7 vector by Gateway ® (Invitro-gen™) [62] and introduced into Arabidopsis via Agrobacterium-mediated transformation [63]. For cloning primers and constructs information see Tables 2 and 3, respectively. For histochemical GUS expression plant samples were soaked at 37°C for 16 hours in GUS assay solution (1 mm 5-bromo-4-chloro-3-indolylglucronide, 0.5 mm K 3 Fe(CN) 6 , 0.5 mm K 4 Fe(CN) 6 , 0.3% (v/v) Triton X-100, 20% (v/v) methanol, and 50 mm inorganic phosphate-buffered saline). The reaction was further conducted at 37°C in the dark for a maximum of 16 hours.

Subcellular localization of AtTSPO fusion proteins
For the GFP fusion constructs, clones containing the coding region of AtTSPO as well as fusions starting at methionine 21 and 42 were generated and cloned into pK7FWG2 [62] (

Determination of chlorophyll contents
Seedlings at 10 days after germination were weighted, frozen in liquid nitrogen, and ground in 80% (v/v) acetone. Ground tissue was centrifuged at 2,000 g for 5 min to pellet any insoluble material. The absorbance of the extracted chlorophyll at 645 and 663 nm was then determined. Chlorophyll (a and b) contents of the samples were determined according to Lichtenthaler [64].

Chloroplast Isolation
Isolation of chloroplasts from plate-grown Arabidopsis seedlings was performed as described previously [65]. Final resuspension of chloroplast was in buffer (330 mM sorbitol, 50 mM HEPES-KOH, pH 8.0) at a concentration of 1 mg chlorophyll ml -1 .

Additional material
Additional file 1: Schematic representation of tetrapyrrole biosyntheses pathway in plants showing genes analyzed in this study. In blue, are the genes already described for each step in the pathway. The enzymes that correspond to these genes names and the AGI code are: HEMA1 (Glutamyl-tRNA reductase 1, At1g58290); HEMA2 (Glutamyl-tRNA reductase 2, At1g09940); HEMA3 (Glutamyl-tRNA   Authors' contributions EBP, JC and GSM conceived and designed the experiments. EBP performed all the experiments, analyzed the data and wrote the paper. YJ helped in the confocal microscopy analyses. BJSCO helped in the fractionation experiment. LRA and JGU gave technical support. JC and GSM were project supervisors, participated in the discussion of all experiments from the project and preparation of the manuscript. All authors read and approved the final manuscript.