Introduction

Plastids in plants evolved from a free-living cyanobacterial ancestor possibly through an endosymbiotic event (Nott et al. 2006). After that, most genes in the previous cyanobacterial genome were either lost or transferred into the host nuclear genome. Consequently, the nuclear genome encodes for the majority of plastid proteins, which are translated in the cytosol and imported into plastids. Therefore, proper coordination of plastid and nuclear gene expression is necessary for plant development and plastid function (Nott et al. 2006).

Plastid gene expression is predominately under nuclear control, which is called “anterograde signaling”. However, coordination is also regulated by “retrograde signaling”, where signals generated in plastids control nuclear gene expression (Nott et al. 2006).

Environmental stresses perturb plastid functions and subsequently influence nuclear gene expression. Signals generated from plastids under environmental stresses are also considered a type of retrograde signal. Therefore, plastid retrograde signaling not only coordinates nuclear and plastid gene expression, but also maintains plastid function at optimal levels, which is important for normal development and optimal response to the environment (Nott et al. 2006).

Higher plants contain many tetrapyrroles, including chlorophyll (Chl), heme, siroheme, phytochromobilin and some of their precursors (from uroporphyrinogen to chlorophyllide) (Tanaka and Tanaka 2007). Chl precursors have a tetrapyrrole structure of porphyrin including four pyrrole rings connected with methylene bridges (Tanaka and Tanaka 2007). Heme is also a tetrapyrrole (four pyrrole rings joined with methylene bridges) that chelates a Fe2+ ion in the center of its porphyrin ring, rather than an Mg2+ as in chlorophyll. Heme is a component of cytochromes, hemoglobins and leghemoglobin (Tanaka and Tanaka 2007). Cytochromes are iron-containing proteins and involved in photosynthesis as well as in respiration (Jahns et al. 2002). Leghemoglobin is an analogue of blood hemoglobin, which attaches to molecular oxygen and has an important role in the metabolism of nodules and nodule bacteria (Verma and Bal 1976). Siroheme is an iron-containing tetrapyrrole. In plants, it is a cofactor of nitrite and sulfite reductases, key enzymes in nitrogen and sulfur assimilation (Tripathy et al. 2010). Siroheme deficiency affects not only nitrogen and sulfur metabolism, but also plant growth and development, responses to biotic and abiotic stresses, and other processes (Tripathy et al. 2010). Phytochromobilin is a linear tetrapyrrole and is the chromophore of phytochrome, absorbing red and far red light (Terry et al. 1993). The four pyrrole rings of chlorophyll are connected to form a tetrapyrrole ring with an Mg2+ in the center. The fourth ring of the tetrapyrrole is esterified with phytol. The early steps in chlorophyll biosynthesis are shared with the heme biosynthetic pathway. Heme is synthesized from protoporphyrin IX by the key enzyme Fe-chelatase. Heme is then converted to phytochromobilin by heme oxygenase and phytochromobilin synthase (Kohchi et al. 2001). The enzymes catalyzing tetrapyrrole biosynthesis are summarized in Fig. 1. Recent advances in these biosynthetic processes are also discussed in this review.

Fig. 1
figure 1

Tetrapyrrole biosynthesis pathway. Key enzymes are indicated in red. Mutants are shown in blue. Siroheme is synthesized in plastids from tetrapyrrole uroporphyrinogen III by two key enzymes Uroporphyrinogen III methyltransferase (UPM) and Sirohydrochlorin ferrochelatase (SIRB). Mg-chelatase has three subunits, CHLH, CHLD and CHLI. The mutant gun5 has a point mutation in the CHLH gene, while GUN4 is an Mg-chelatase coordinator that can enhance the Mg-chelatase activity. Heme is synthesized from protoporphyrin IX by the key enzyme Fe-chelatase (ferrochelatase, FC). gun2 and gun3 have mutations in the heme oxygenase and phytochromobilin synthase, respectively. Heme accumulation leads to negative feedback regulation of chlorophyll biosynthesis. CRD is an Mg-protoIX methyl ester cyclase. 3,8-Divinyl (proto)chlorophyll(ide) a 8-vinyl reductase (DVR) is suggested to catalyze the conversion of divinyl tetrapyrroles to monovinyl tetrapyrroles. NADPH–Pchlide oxidoreductase (POR) functions dependent on light. por mutants accumulate high levels of protochlorophyllide (Pchlide), which also negatively regulates Glutamyl-tRNA reductase (the negative regulation being mediated by FLU protein). Chl synthase (G4) was proposed to localize to the thylakoid membranes where esterification of Chlide a with phytol occurs. Chlorophyllide a oxygenase (CAO) converts Chl a to Chl b. The model is revised from the models presented in reference (Nott et al. 2006)

Full size image

A large number of studies have indicated that multiple retrograde signals derived from plastids control expression of photosynthesis-associated nuclear gene (PhANG) expression, such as the LHCB gene (encoding a light-harvesting chlorophyll a/b-binding protein). One of these pathways is related to Mg-protoporphyrin IX (Mg-protoIX) (Koussevitzky et al. 2007; Nott et al. 2006; Strand et al. 2003; Tanaka and Tanaka 2007; Zhang et al. 2011a), and another employs heme signal (Barajas-López Jde et al. 2013a; Enami et al. 2011; von Gromoff et al. 2008; Woodson et al. 2011; Zhang et al. 2013). Additionally, recent studies have shown that tetrapyrrole-derived signals strongly regulate plant resistance to environmental stresses, which are summarized in this review.

Recent advances in tetrapyrrole biosynthesis

Enzymes catalyzing the formation of the early hydrophilic intermediate of chl biosynthesis [from ALA to coproporphyrinogen III] are localized in the stroma. Enzymes catalyzing reactions involving late hydrophobic intermediates of chl biosynthesis (from protoIX to chls) are bound to membranes (Fig. 1). Most enzymes catalyzing Chl formation are localized in thylakoid membranes and inner membranes of the plastid envelope (Nott et al. 2006; Tanaka and Tanaka 2007). The localization of Mg-chelatase depends on Mg2+ concentration in plastids. At a high concentration, the CHLH subunit of Mg-chelatase is bound to membranes of the chloroplast envelope, whereas at a low concentration the enzyme detaches from the membrane (Nakayama et al. 1998). CHLH spans the plastid envelope and its C terminus interacts with WRKY transcription factors (WRKY40, WRKY18 and WRKY60) in the cytosol (Shang et al. 2010). The CHLI subunit of Mg-chelatase is always found within the stroma, but The CHLD subunit is localized in both the stroma (predominantly) and the thylakoid membranes (Zhang et al. 2006).

GUN4 is an Mg-chelatase cofactor, also binding protoIX and Mg-protoIX (Larkin et al. 2003). It is thought that GUN4 controls porphyrin distribution between the branches of the pathways leading to chl and heme (Larkin et al. 2003). The expression of GUN4 is most active in young and greening tissues. GUN4 correlates with the rate of ALA and Mg-protoIX synthesis. Overexpression of GUN4 results in a general activation of chl biosynthesis enzymes. GUN4 deficiency in this period suppresses ALA synthesis and chl accumulation. However, ALA addition does not compensate for GUN4 shortage during growth in the light/dark regime, but instead results in chl accumulation under weak illumination (Peter and Grimm 2009). The porphyrin binding activity of GUN4 and GUN5 affects their associations with the plastid membranes and regulates reactive oxygen species (ROS)-induced gene expression (Adhikari et al. 2011).

NADPH-protochlorophyllide (Pchlide) oxidoreductase (POR) is one of the key enzymes responsible for de-etiolation of higher plants. Light-harvesting POR–Pchlide complex (PORA:PORB = 5:1) called LHPP was found in barley etioplasts, but not in other higher plants (Reinbothe et al. 1999). POR complexes may be assembled by dithiol oxidation of cysteines of two adjacent proteins. The LHPP complex can dissipate excess light and decrease Pchlide-derived singlet-oxygen generation, which leads to photoprotection for de-etiolated seedlings (McCormac and Terry 2002). POR proteins only formed dimers in Arabidopsis and formation of the dimer was not related with its photo-protective function (Yuan et al. 2012). Nevertheless, down-regulation to POR protein levels in barley decreases the density of prolamellar bodies, although their sizes are not changed (Yuan et al. 2010).

3,8-Divinyl (proto)chlorophyll(ide) a 8-vinyl reductase (DVR) has a role in catalyzing the conversion of divinyl tetrapyrrole to monovinyl tetrapyrrole (Nagata et al. 2005). DVR proteins from different species have different substrate specificities (Fig. 1). Monocotyledon DVR can convert divinyl Mg-protoIX (methyl esters) to monovinyl Mg-protoIX (methyl esters) (Wang et al. 2010, 2013).

A key enzyme in higher plant tetrapyrrole biosynthesis, Glu tRNA reductase, is controlled by negative feedback by some tetrapyrroles, such as heme and Pchlide. A negative sensor, FLUORESCENT (FLU; flu mutant shows strong Pchlide fluorescence, characterized by Meskauskiene et al. 2001) was identified in Arabidopsis (Meskauskiene et al. 2001). FLU works independently of heme, but suppresses tetrapyrrole biosynthesis by inhibiting Glu tRNA reductase selectively (Kauss et al. 2012; Meskauskiene et al. 2001) (Fig. 1). Pchlide overaccumulation in flu mutant generates singlet oxygen. Plastid proteins EXECUTER1 (EX1) and EXECUTER2 (EX2) are responsible for the transduction of singlet oxygen-derived plastid signals to the nucleus (Lee et al. 2008; Wagner et al. 2004). Protein-to-protein binding between FLU and POR, Glu tRNA reductase and Mg-protoIX methyl ester oxidative cyclase (CHL27/CRD) have been identified (Kauss et al. 2012), indicating direct down-regulation of tetrapyrrole biosynthetic enzymes by FLU.

Putative signaling pathways triggered by tetrapyrroles

LHCB is a nuclear-encoded marker gene required for plastid-to-nucleus signaling (Susek et al. 1993). Mutants have been isolated when LHCB expression was uncoupled from the functional state of the chloroplast. Norflurazon, a specific inhibitor of phytoene desaturase (Nott et al. 2006) has been used for the mutant screening. When wild-type plants were treated with norflurazon under continuous light, LHCB expression was repressed due to photobleaching of the plastid. However, LHCB expression levels in the mutants were relatively high and less affected by plastid photobleaching. After this screening, some mutants impaired in retrograde signaling were identified and named genome uncoupled (gun) mutants (Susek et al. 1993).

GUN5 encodes the H-subunit of Mg-chelatase (CHLH), and LHCB expression in a gun5 mutant was less repressed by norflurazon, because of reduced Mg-protoIX synthesis (Strand et al. 2003). GUN4 is an Mg-chelatase cofactor (Larkin et al. 2003). GUN2 encodes a heme oxygenase, and GUN3 encodes a phytochromobilin synthase. Recently, a gain-of-function gun mutant gun6-1D was identified, which overexpresses the conserved plastid ferrochelatase 1 (FC1, heme synthase) (Woodson et al. 2011) (Fig. 1).

GUN1 is a plastid protein that may be involved in both generation of multiple plastid signals (upstream role) and integration of multiple plastid signals (downstream role), including tetrapyrrole-related signals, sugar signals, plastid gene expression (PGE) signals, photosynthetic electron transport chain (PET) signals and plastid redox signals (Koussevitzky et al. 2007; Zhang et al. 2010). The gun1 mutant was found in the same screenings as the other tetrapyrrole mutants (Susek et al. 1993). In response to the GUN1-mediated signals, ABI4 (an Apetala 2-type transcription factor) binds the promoter of LHCB1 and likely prevents GBF (a light-positive-responsible G-box binding factor) from binding, which results in LHCB repression (Koussevitzky et al. 2007) (Fig. 2).

Fig. 2
figure 2

Model of tetrapyrrole signaling. Heme and Mg-protoIX have been suggested to regulate nuclear gene expression. Tetrapyrroles (not so heme) act as negative regulators of PhANG expression, although the signaling components between the nucleus and plastids are not very clear. ROS de-activates cyclase complex (CRD), which resulted in Mg-protoIX over-accumulation. Plastid-encoded photosynthesis genes are also regulated by the accumulation of Mg-protoIX. Plastid-encoded RNA polymerase (PEP) subunits, the Sigma factors, participate in this process. Pchlide overaccumulation in flu mutant generates singlet oxygen. Plastid proteins EXECUTER1 (EX1) and EXECUTER2 (EX2) are responsible for the transduction of singlet oxygen-derived plastid signals to the nucleus. PAPP5 acts downstream of the GUN5 and Mg-protoIX(ME). PAPP5 negatively regulates transcription factor GLK1/2 expression, which is necessary for PhANG expression and Chl synthesis. HSP90 ATPase activity can be inhibited by Mg-protoIX binding and then results in a HY5-dependent PhANG repression. PTM is a chloroplast envelope-bound plant homeodomain transcription factor. During signaling transduction, PTM is cleaved and the amino-terminal PTM is transferred into the nucleus, where histone modifications to the ABI4 promoter occur. In response to the GUN1-mediated signals and ABA-signals, ABI4 binds the promoters of LHCB and CBFA to prevent GBF (a light-inducible CCACGT binding factor) from binding, and therefore prevents their transcription persistently. CBFA (a repressor) and LEC1 (an activator) are two CCAAT-binding factors (HAP3 subunits), which interact with HAP2 and HAP5 to form a trimeric complex working at the CBP (the transcription cofactor) promoter. Down-regulation to CBFA results in more LEC1 binding to the promoter and the transcriptional efficiency of HAP complex is enhanced. Increased CBP transcripts promote all RNA polymerase activities and the global RNA multiplication occurs. The model is revised from the models presented in reference (Barajas-López Jde et al. 2013a)

Full size image

ABI4 is a key component of the ABA signaling pathway. Germination suppression sensitivity of gun4 and gun5 mutant seeds to ABA was increased compared to wild-type seeds (Voigt et al. 2010). The gun1 mutant was more sensitive to ABA than wild-type plants, implying that GUN1 participates in ABA signaling (Cottage et al. 2010). All of the data imply that tetrapyrrole, plastid signaling and ABA signaling may be interconnected. However, the potential role of GUN5 as an ABA receptor (Shen et al. 2006) is controversial because some laboratories have reported that GUN5 does not bind ABA (Zhang et al. 2015), and the bona fide ABA receptors have been identified with solid genetic evidences on ABA signaling (Ma et al. 2009; Park et al. 2009). The role of GUN5 in ABA-regulated processes is distinct from its role in chlorophyll biosynthesis and in plastid-to-nucleus signaling.

Multiple signals, such as Mg-protoIX, heme or a high level of sugar (over 7 %) doubled total cellular RNA within 48 h. All types of RNA (mRNA, tRNA and rRNA) were increased after Mg-protoIX, heme or high-level sugar treatments. The transcription cofactor CBP (CAAT binding protein) is responsible for this rapid RNA increase (Zhang et al. 2011b). Both GUN1 and ABI4 are required for this rapid RNA increase (Zhang et al. 2011b). Another signaling component, CBFA (CCAAT-binding factor A), was found. CBFA is a redundant subunit of the HAP (Heme activator protein)2/HAP3/HAP5 trimeric transcription complex. CBFA is an ABI4-repressible gene. The master transcription factor ABI4 inhibits CBFA upon exposure to environmental stresses (such as drought, high-light stress or herbicide treatment). Other TF subunits (e.g., LEAFY COTYLEDON1 LEC1) then enter the transcription complex and stress-responsive genes are promoted (Zhang et al. 2013) (Fig. 2).

Transcription factors Golden 2-like 1 and 2 (GLK1/2) up-regulate tetrapyrrole biosynthesis gene expression. The glk1,glk2 double mutant results in an abnormal photosynthetic apparatus and low leaf chl levels (Waters et al. 2009). PhANG expression was significantly suppressed in glk1,glk2 mutants compared to the wild type. The reduced levels of chlorophyll intermediates in the glk1,glk2 mutant suggest a role for perturbed tetrapyrrole pools in plastid signaling (Waters et al. 2009).

Retrograde signaling associated with Mg-protoIX

Defects in earlier steps of chl biosynthesis before Mg-chelatase (Strand et al. 2003), or POR over-expression with reduced Mg-protoIX levels (McCormac and Terry 2002) result in a GUN phenotype (derepression of PhANG expression in the presence of noflurazon). Most gun mutants are located in the tetrapyrrole synthesis pathway, suggesting a putative signaling role for Mg-protoIX or its related tetrapyrroles.

Feeding experiments with Mg-protoIX provide more direct evidence. Mg-protoIX supplied to Chlamydomonas cell cultures led to induction of nuclear heat-shock genes HSP70 (von Gromoff et al. 2008). Mg-protoIX also coordinates organelle and nuclear DNA replication in both algal cells and higher plants (Kobayashi et al. 2011). Application of exogenous Mg-protoIX to Arabidopsis protoplasts under low-light conditions (without generating any ROS) inhibited LHCB expression, whereas porphobilinogen, ProtoIX and heme had no effect (Strand et al. 2003). Such feeding experiments suggested that Mg-protoIX and Mg-protoIX methyl esters might be the signaling molecules controlling nuclear gene expression (Strand et al. 2003).

The Mg-protoIX-derived signaling in higher plants is more complicated than in green alga. A plastid response element responsive to both Mg-protoIX/heme and light signals was identified in Chlamydomonas (von Gromoff et al. 2008). However, in higher plants, Mg-protoIX, heme and light signaling pathways are partly diverged, and multiple tetrapyrrole-signaling components are specific to higher plants (Barajas-López Jde et al. 2013a; Zhang et al. 2013). Furthermore, some observations in some barley and Arabidopsis mutants are incompatible with the “Mg-protoIX model” that Mg-protoIX works as a signaling molecule to repress PhANG expression (Gadjieva et al. 2005). Arabidopsis Mg-chelatase I subunit single mutants (cs or ch42) do not have a GUN phenotype even though these mutants have less potential for Mg-protoIX production than the gun5 mutant (Nott et al. 2006). Only the chli1/chli2 double mutant has a GUN phenotype upon treatment with norflurazon (Huang and Li 2009). A POR-less barley mutant accumulated much higher level of Mg-protoIX, but its LHCB expression level was almost the same as the wild type (Zhang et al. 2011b).

The role of Mg-protoIX as a signaling molecule was greatly challenged (Mochizuki et al. 2008; Moulin et al. 2008). There have been some discrepancies between accumulated pools of tetrapyrroles and PhANG expression levels. No tetrapyrrole (including Mg-protoIX) could be detected in norflurazon-treated plants, when PhANG expression was largely repressed (Moulin et al. 2008). All tetrapyrrole biosynthesis genes were significantly repressed in norflurazon-treated plants. Unfortunately, no plausible counter-model has been presented. Those authors speculated that perhaps the effects of norflurazon were due to the ROS burst. However, the expression of ROS-specific genes were not different in the gun5 mutant compared to the wild type following norflurazon treatments, leading to the conclusion that the gun5 phenotype is not related to an altered ROS accumulation (Voigt et al. 2010). A latter report suggested that in seedlings grown from the start in the presence of NF, ROS-dependent signaling was not detectable, whereas in seedlings first exposed to NF after light-dependent chloroplast formation had been completed, enhanced ROS production occurred (especially when the 1O2-mediated and EXECUTER-dependent retrograde signaling was induced) (Kim and Apel 2013). Hence, depending on the developmental stage at which plants are exposed to NF, different retrograde signaling pathways may be activated, some of which are also active in non-treated plants under light stress (Kim and Apel 2013).

Tetrapyrrole synthesis is inhibited when phytoene desaturase is blocked by norflurazon. However, not all tetrapyrrole synthesis enzymes are equally inhibited by norflurazon treatment. Mg-protoIX methyl ester oxidative cyclase and POR activities were greatly inhibited by short-term norflurazon treatment, while other tetrapyrrole synthesis enzymes were less affected (Zhang et al. 2011a). A short norflurazon treatment also resulted in LHCB repression. However, after a long norflurazon treatment (>96 h), tetrapyrrole levels varied widely (e.g., Mg-protoIX fluctuated markedly among individual plants), but LHCB repression continued (Zhang et al. 2011a). These results may explain the discrepancy of different reports. Moreover, LHCB repression cannot be fully reversed by ROS quenchers (such as dimethylthiourea and tiron). Therefore, the effects of norflurazon treatment could not be simply attributed to some kind of ROS signal. Alternatively, a transient accumulation of plastid Mg-protoIX induced by norflurazon treatment may transmit a signal to the nucleus to regulate gene expression (Zhang et al. 2011a).

It can be very difficult to revert Mg-protoIX signals once changes in nuclear gene expression have been triggered (Strand et al. 2003; Zhang et al. 2011a), which may imply that these signals are very important for plant growth or stress tolerance. Correspondingly, Mg-protoIX is a long-acting signaling molecule, which seems contradictory to the transient accumulation of Mg-protoIX as mentioned above. The turnover of tetrapyrroles is relatively fast (Zhang et al. 2011a). This discrepancy needs further investigation. ABI4 binds strongly to the G-box element CCACGT, which cannot be reversed by a high-light treatment (Zhang et al. 2013). We presume that ABI4 may transmit the signals by binding to the target promoters through a conformational change at the protein secondary structure. The change is probably irreversible, which may explain the long lasting efficacy of the Mg-protoIX signals. Koussevitzky and collaborators (2007) also reported that the promoters of some PhANGs possess ABI4 binding sites, but many others do not. Therefore, the role of ABI4 cannot fully explain global effects of plastid signals. Other transcriptional factors need to be identified.

Besides the regulation in PhANG expression, Mg-protoIX-derived signaling can also be involved in cell cycle coordination in algae (Kobayashi et al. 2011). Aside from nuclear genes, plastid-encoded photosynthesis genes are also down-regulated by blocking chloroplast development after norflurazon treatment. Plastid RNA polymerase subunits, the Sigma factors, participate in this process and negatively respond to Mg-protoIX accumulation (Ankele et al. 2007). Therefore, GUN5-mediated signals after the NF treatment also regulate plastid development by controlling its own gene expression (Table 1).

Table 1 Tetrapyrrole metabolism enzymes and signaling components and their related environmental stress and developmental processes
Full size table

The signaling role of the tetrapyrrole heme

Heme and its degradation products bilins up-regulate Chlamydomonas PhANG expression (Duanmu et al. 2013; Von Gromoff et al. 2008). There was no evidence that heme also works in higher plants until 2010 (Rockwell et al. 2006). Free heme is highly photodynamically cytotoxic and parallels ROS production (Rockwell et al. 2006). Therefore, heme may work as a ROS propagator but not directly as a signaling molecule. However, our results showed that 50–200 μM of heme feeding under low light doubled total cellular RNA and repressed LHCB expression without generating any ROS (Zhang et al. 2013). Thus, heme may work as a signaling molecule that controls PhANG expression independently. Heme synthesized by plastid ferrochelatase I (but not ferrochelatase II) up-regulated PhANGs, also indicating a putative signaling role of heme (Woodson et al. 2011). This observation seems contradictory to our results and the level of exogenous heme applied might be the reason. ALA feeding to seedlings at high concentrations (over 0.8 mM) has a negative effect on PhANG expression (Woodson et al. 2011), which might suggest that there are dual effects of heme on PhANG expression (both up-regulation and down-regulation). Exogenous heme treatments down-regulated nuclear starch biosynthesis gene expression (Enami et al. 2011). The dual role of heme in gene expression regulation and its signaling pathways needs further investigation.

Mg-protoIX signaling pathways and heme signaling pathways may share some common components, such as GUN1 and ABI4. Both Mg-protoIX signals and heme signals were largely abolished in gun1 or abi4 mutant (Zhang et al. 2013). Nevertheless, there are also some heme-specific transcription factors, such as subunits of HAP (Heme activator protein) (Zhang et al. 2013). Compared to 20 years of studies of Mg-protoIX signals, little is known about heme signals in higher plants.

The Chl (Mg2+) branch and heme (Fe2+) branches of tetrapyrrole biosynthesis pathways are regulated by Mg-chelatase and ferrochelatase, respectively. The inhibition of one branch would increase Mg-Proto IX flux through the other branch. However, the putative interplay between Mg-ProtoIX and heme in PhANG expression regulation or stress resistance is still unknown. Heme-related signal is the primary positive signal during plastid biogenesis, while GUN1 normally serves to restrict this heme-related signal. Aberrant plastid development may produce a negative signal due to accumulation of unbound tetrapyrroles such as Mg-ProtoIX. The accumulation of these tetrapyrroles results in a rapid light-dependent inhibition of nuclear gene expression that is most likely mediated via singlet oxygen generated by photo-excitation of Mg-ProtoIX (Terry and Smith 2013). However, the results of LHCB repression under exogenous Mg-ProtoIX feeding (Strand et al. 2003; Zhang et al. 2011a) or heme feeding (Zhang et al. 2013) in low light cannot be explained by this model. There is no evidence that singlet oxygen is generated upon Mg-ProtoIX accumulation in light. Experiments have only confirmed that Pchlide induces singlet oxygen in light. Both heme and Mg-ProtoIX induce ROS production, likely mainly superoxide and H2O2 (Strand et al. 2003; Zhang et al. 2013). It is unknown how heme works as a positive signal. Mg-ProtoIX accumulation triggers a negative signal. The putative interplay between Mg-protoIX and heme signals is still largely unclear.

Putative signaling components between the nucleus and plastids

PTM, a chloroplast envelope-bound plant homeodomain (PHD) transcription factor with transmembrane domains, connects the GUN1-pathway in the plastid and the ABI4-pathway in the nucleus (Sun et al. 2011). During signal transduction, PTM is cleaved and the amino-terminal PTM is transferred into the nucleus, where histone modifications to the ABI4 promoter occur (Sun et al. 2011) (Fig. 2). Nevertheless, the induction rates of ABI4 transcripts by plastid signals were much less than the increasing rates of ABI4 binding to the G-box, indicating some other regulatory mechanism besides transcriptional enhancement (Zhang et al. 2013). Expression of ABI4 in Arabidopsis seedlings was not changed after treatment with lincomycin, a plastid gene-expression signal inhibitor (Koussevitzky et al. 2007). Thus ABI4 may transmit signals through a conformational change (Sibéril et al. 2001), which is inconsistent with the signaling pathway of PTM as mentioned above. More importantly, there are some ABI4-independent signaling pathways, where the regulation of PhANG expression or plastid protein import is independent of ABI4 (Koussevitzky et al. 2007). GUN1–PTM–ABI4 cannot represent the entire signaling cascade.

There are also some other promising candidates for plastid retrograde signaling, such as SIGMA FACTOR BINDING PROTEINS (SIB1) and SIB2, who are localized in both plastids and nuclei (Krause et al. 2012). Among these dual-localized candidate proteins, a member of the Whirly family of nuclear transcriptional activators, Why1, is the most interesting one. Why1 can be exported from plastids and be translocated to the nucleus (Grabowski et al. 2008). However, further study should be performed, given that no GUN phenotype has been reported in the why1 mutant so far (Grabowski et al. 2008). Thus, Why1 may not be a component in plastid retrograde signaling.

The yeast HSP70–HSP90–HAP1 complex is responsible for heme accumulation and heme-induced oxidative stress (Mense and Zhang 2006). HSP90 and a transcription factor HY5 may also participate in the tetrapyrrole signal transduction in higher plant cells (Kindgren et al. 2012). ATPase activity of HSP90 can be inhibited by tetrapyrrole binding to HSP90 protein. Silencing of the HSP90 gene in the gun5 background abolishes the gun5 phenotype (Kindgren et al. 2012). Moreover, the hy5 mutant also had a GUN phenotype upon oxidative stress, and was insensitive to the HSP90 activity inhibitor geldanamycin. Thus, HY5 should act downstream of GUN5 (Kindgren et al. 2012). The GUN5–HSP90–HY5 pathway may be involved in the regulation of PhANGs in response to the tetrapyrrole signals (Kindgren et al. 2012) (Fig. 2).

Another putative signaling component is an Mg-protoIX-binding protein Phytochrome-Associated Protein Phosphatase 5 (PAPP5) (Kindgren et al. 2011). A defect of CHL27/CRD (encoding aerobic Mg-protoIX methyl ester cyclase) resulted in Mg-protoIX over-accumulation and PhANG repression. However, the papp5,crd double mutant or inhibition of PAPP5 phosphatase activity restored the crd phenotype, showing a derepression of PhANG transcription (Barajas-López Jde et al. 2013b). PAPP5 activity was tightly correlated with tetrapyrrole accumulation levels. Thus, PAPP5 should act downstream of GUN5 and Mg-protoIX in the plastid retrograde signaling pathways (Barajas-López Jde et al. 2013b) (Fig. 2).

The physiological roles of Mg-protoIX and Mg-protoIX-signaling components in plant resistance to environmental stress

The transcriptional network resulted by tetrapyrroles is a major component of plant adaptation to the environment. Transgenic rice expressing Myxococcus xanthus protoporphyrinogen oxidase (PPOX) accumulated more protoIX, Mg-protoIX, Pchlide and heme, and had greater tolerance to drought stress than wild-type plants (Phung et al. 2011) (Table 1).

We should pay attention to plastid-signal-inducible genes when studying the physiological significance of tetrapyrrole signaling. Many norflurazon-inducible genes were elucidated in micorarray analysis (Strand et al. 2003) and RT-PCR data (Zhang et al. 2011b). Interestingly, norflurazon-inducible genes are mostly involved in oxidative-stress resistance, such as mitochondrial alternative oxidase and some cytosol antioxidant enzymes (peroxidase; superoxide dismutase and ascorbate peroxidase). An alternative explanation is that ROS may be produced in response to Mg-protoIX accumulation (Strand et al. 2003; Zhang et al. 2013). Therefore, one could speculate that plastid-signaling-defective mutants adapt less effectively to environmental stresses.

The gun5, gun1 and abi4 mutants had impaired basal high-temperature tolerance (Miller et al. 2007). The gun3, gun5, gun1 and abi4 mutants had more oxidative damage than Col-0 plants under water stress (Cheng et al. 2011). The gun1 mutant showed apparent photo-bleaching under high-light and herbicide treatments together (Zhang et al. 2011b). Norflurazon pretreatment enhanced plant tolerance to environmental stresses by presumably inducing anti-oxidative gene expression (Cheng et al. 2011; Tang et al. 2014; Zhang et al. 2011b, 2013). The gun1, cbfA and cbp mutants had impaired basal drought tolerance, while the gun1, abi4, cbfA and cbp mutants had hampered norflurazon-induced drought tolerance (Zhang et al. 2013). The gun1, gun5 and abi4 mutants had impaired basal cold-tolerance and norflurazon-induced cold-tolerance (Tang et al. 2014). All of the studies indicated the physiological role of these key signaling components in stress adaption (Table 1).

The physiological roles of siroheme and heme-binding proteins in plant resistance to environmental stress

The reduction of sulfate to sulfite is mediated by sulfate reductase with siroheme as the prosthetic group in plastids. The reduced S is then integrated into cysteine and methionine (Tripathy et al. 2010). Glutathione is a key component of plant stress responses and counteracts oxidative damage induced by extreme temperatures, drought, pathogen attacks or chemicals/toxins (Tripathy et al. 2010). The reduced form of glutathione is a well-known antioxidant. Glutathione is also the substrate for metallothioneins, which chelate and detoxify excess heavy metals in the cell (Tripathy et al. 2010). Therefore, siroheme regulates glutathione biosynthesis and maintains the cellular redox state under environmental stress (Table 1).

According to affinity to oxygen, the nonsymbiotic hemoglobins can be divided into two groups: high-affinity class 1 hemoglobins (nsHb-1s) and lower-affinity class 2 hemoglobins (nsHb-2s). Hypoxia induces nsHb-1 expression, as do in osmotic and cold stresses, nutrient deficiency stress, fungal infection or nitric oxide (NO) treatments (Dordas 2009). Hemoglobin induction is an important adaptation that plants have evolved to overcome stress damage. The roles of hemoglobins in stress tolerance have been extensively reviewed elsewhere (Dordas 2009).

Overexpressing heme oxygenase greatly enhanced plant tolerance to several kinds of environmental stresses (Bose et al. 2013; Xie et al. 2008). Heme oxygenase is responsible for heme degradation, which then generates carbon monoxide and biliverdin. Biliverdin reductase then reduces biliverdin into bilirubin in mammal cells (Doré et al. 1999). Bilirubin is a potent antioxidant against endothelial cell injury and several other mammalian diseases (Doré et al. 1999). Increased bilirubin formation, due to activation of heme oxygenase, protects mammalian cells against hydrogen peroxide-induced neurotoxicity (Doré et al. 1999).

Environmental stress also induces an increase in endogenous carbon monoxide, another product of heme oxygenase, in plant roots. Overexpressing the HO1 gene or CO treatment alleviated salinity, osmotic stress and heavy metal stresses, which could be mimicked by the application of NO donors (Cui et al. 2013; Xie et al. 2008). NO or CO treatments significantly increased H+-pump and anti-ROS enzyme activities, and enhanced the cellular K/Na ratio, thereby maintaining cell membrane potential during the osmotic stress. Conversely, CO/NO scavengers or inhibitors hampered these anti-stress abilities. Moreover, CO treatments induced NO release in root tips, while the CO-induced NO signal was eliminated by NO scavengers (Xie et al. 2008). These results suggest that NO and CO play an important role in ion and redox homeostasis in plant tissues and may subsequently increase plant tolerance to multiple environmental stresses (Xie et al. 2008).

Heme oxygenase is triggered by multiple environmental stresses, including salt stress, heavy metal contaminants and UV radiation (Bose et al. 2013; Xie et al. 2008). Transgenic plants over-expressing the HO gene retained more K+ (higher H+ efflux activity), suffered less membrane damage in the root epidermis during salt stress, compared to HO-deficient mutants. Heme oxygenase modifies salt stress tolerance by controlling K+ retention and enhancing plasma membrane H+-ATPase activity and SOS1 gene expression (Bose et al. 2013) (Table 1). Furthermore, heme oxygenase also regulates plant development, such as lateral root development, adventitious rooting and leaf senescence (Cao et al. 2011; Huang et al. 2011) (Table 1).

A tryptophan-rich sensory protein (TSPO) is another heme-binding membrane protein. TSPO belongs to the TspO/MBR regulatory protein family, whose expression is induced by oxidative stress and ABA treatment (Vanhee et al. 2011). The TSPO level is positively correlated with cellular unbound heme levels, indicating its biochemical role in heme scavenging. Furthermore, TSPO overexpression attenuated ALA-induced cellular tetrapyrrole accumulation. TSPO may enhance plant tolerance to oxidative stresses by binding and scavenging heme (Vanhee et al. 2011).

The possible relationship between tetrapyrroles and stress-specific organelle-to-nucleus metabolic signal molecules

Several other stress-derived retrograde metabolic signals have recently been identified including Ca2+ signatures (through a calcium-sensing receptor CAS), unsaturated fatty acids, β-cyclocitral, 2-C-methyl-d-erythritol 2,4-cyclodiphosphate (MEcPP) and 3′-phosphoadenosine 5′-phosphate (PAP), and have been extensively reviewed elsewhere (Xiao et al. 2013).

Pathogen-associated molecular pattern (PAMP) signals are quickly relayed to plastids, and specific plastid Ca2+ signatures are involved (Nomura et al. 2012). A plastid-localized calcium-sensing receptor (CAS) is responsible for stromal Ca2+ transients, basal resistance and hypersensitive cell death. CAS-mediated plastid gene-expression regulation was correlated with singlet-oxygen-mediated retrograde signaling, suggesting that a chloroplast-mediated transcriptional reprogramming occurs during plant immune responses (Nomura et al. 2012).

Free tetrapyrroles photodynamically generate ROS and then oxidative damage, decreasing the ratio of unsaturated fatty acids in membranes (Nott et al. 2006) and inducing the oxidation of the carotenoid β-carotene (Ramel et al. 2012, 2013). β-cyclocitral, an oxidative product of carotenoid, was reported to reprogram gene expression, especially the singlet oxygen-responsive genes, which was associated with an increased tolerance to photo-oxidative stress (Ramel et al. 2012, 2013). Accumulation of Pchlide in the flu mutant induced β-cyclocitral (Ramel et al. 2012). Thus, tetrapyrroles may be indirectly correlated with unsaturated fatty acids and β-cyclocitral. In plants, 3′-phosphoadenosine 5′-phosphosulfate (PAP) is a by-product of sulfur assimilation, produced by the transfer of the sulfate group to acceptor molecules in various sulfation reactions including the biosynthesis of thiols and glucosinolates (Wilson et al. 2009). In Arabidopsis, SAL1, also known as FIERY1 or FRY, encodes the major phosphatase that hydrolyzes PAP to AMP and Pi (Wilson et al. 2009). Mutants of SAL1 have elevated levels of foliar ABA and jasmonate and exhibit constitutive induction of nuclear genes responsive to high light, drought and salinity (Wilson et al. 2009; Estavillo et al. 2011). Given that siroheme is an important cofactor of sulfur assimilation (Tanaka and Tanaka 2007), studies of the relationship between siroheme and PAP in wild-type plants and sal1 mutants would be interesting. However, no direct link between tetrapyrroles and MEcPP (Xiao et al. 2012) could be found, therefore requiring further investigation.

Available microarray data and RT-PCR results from different signals (Gläßer et al. 2014; Lee et al. 2008; Miller et al. 2007; Strand et al. 2003; Wagner et al. 2004; Zhang et al. 2011b) were compared, but there is no apparent overlap in gene expression profiles in mutants and conditions affecting the different signals, except for some common stress-related genes, such as AOX, chalcone synthase, thioredoxin or some antioxidant enzyme genes. Therefore, comparative transcriptome analysis is required to uncover more tight relationships between tetrapyrrole signals and other stress-specific organelles-to-nucleus metabolic signals. A comparative study highlighted some plastid-signal-responsive cis-elements and classified ABA and auxin-based signaling as secondary components involved in response cascades following plastid signals (Gläßer et al. 2014). However, the common pathways suggested in this study are not directly related with the organelle-to-nucleus stress signals mentioned above. Additional in-depth transcriptome analyses are needed.

Conclusions and perspectives

Tetrapyrrole metabolism is involved in the regulation of nuclear gene expression, adjusts photosynthetic activities and has an important role in environmental stress adaptation. Nevertheless, accurate signaling roles of Mg-protoIX and heme are still controversial. To resolve the controversy, we should find a condition or treatment for which only tetrapyrrole signals are involved but ROS signals, plastid gene-expression signals, aberrant plastid development signals or photosynthetic electron transport chain signals are not induced. However, tetrapyrrole levels are usually tightly correlated with plastid developmental status and ROS production. Therefore, tetrapyrrole signals may be signals inseparable from other developmental signals or metabolite signals. Alternatively, we may find some tetrapyrrole-binding proteins that act as signal receptors, either in the plastid or in the cytosol, and show strong evidence of how they regulate signal sensing and transduction across the plastid envelope. Most importantly of all, the signal receptor responsive to tetrapyrrole specifically should be identified. Scientists sceptical about the Mg-protoIX model should also strive to identify the real signal molecules, which can fully replace all signaling roles of Mg-protoIX, and show strong evidence of why Mg-protoIX is not the direct signaling molecule.

A large number of other signaling molecules are involved in tetrapyrrole-signaling pathways, such as ROS, NO, CO, ABA and sugars. Therefore, the signaling network is very complicated. Most studies have focused on drought, high-light stress, water stress (osmotic stress), salinity and heavy metal stress. Research on the role of tetrapyrroles in cold stress is minimal. However, putative effects of tetrapyrrole signals on the control of the molecular adaptation to cold stress have been suggested by transcriptome analysis (Svensson et al. 2006). Thus, there should be some connection between tetrapyrrole metabolism and cold stress tolerance. Moreover, the physiological roles of Mg-protoIX-binding proteins (such as PAPP5, CRD and HSP90) and heme-binding proteins (such as HO and TSPO) in stress adaptations have not been well-understood, which needs further investigation.

Identification of additional proteins moving from the plastid to the nucleus or other potential carriers of chloroplast signals should be performed. PTM, HSP70–HSP90–HAP1 complex and PAPP5 may be only a part of the whole picture. Since PAPP5 is a protein phosphatase (Barajas-López Jde et al. 2013b), a protein phosphorylation–dephosphorylation cascade in the cytosol may be responsible for the signal transduction from the chloroplast to the nucleus. Since histone modifications to the ABI4 promoter were reported with lincomycin or norflurazon treatments (Sun et al. 2011), DNA methylation may be also involved in signal transduction. Real-time proteomic (phosphor-proteomic) and methylomic analyses of the signal transduction process may be useful approaches to reveal novel components and their interactions in chloroplast retrograde signaling.

In general, deficiencies in tetrapyrrole biosynthetic enzymes or plastid-signaling components reduce plant resistance to environmental stresses. Inversely, over-expression of tetrapyrrole biosynthetic enzymes or plastid-signaling components enhances plant resistance to environmental stresses. However, only a few mutants or transgenic plants were studied under stress conditions. More in-depth genetic and physiological studies might provide a more comprehensive explanation of the stress-responsive role of tetrapyrroles in the near future. This will open up new possibilities for creating future genetically modified plants that are more resistant to environmental stresses.

Author contributions statement

ZWZ and GCZ had a primary role in writing the manuscript. FZ read and corrected the manuscript extensively. SY contributed ideas to the article. DWZ made key contributions to the article, particularly in regard to tetrapyrroles signaling.