Biotin deficiency causes spontaneous cell death and activation of defense signaling

Authors

  • Jing Li,

    1. Division of Genetics, Department of Biosciences, Viikki Biocenter, University of Helsinki, 00014 Helsinki, Finland
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Günter Brader,

    1. Division of Genetics, Department of Biosciences, Viikki Biocenter, University of Helsinki, 00014 Helsinki, Finland
    2. Austrian Institute of Technology GmbH, Health & Environment Department, Bioresources Unit, 3430 Tulln, Austria
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Elina Helenius,

    1. Division of Genetics, Department of Biosciences, Viikki Biocenter, University of Helsinki, 00014 Helsinki, Finland
    Search for more papers by this author
  • Tarja Kariola,

    1. Division of Genetics, Department of Biosciences, Viikki Biocenter, University of Helsinki, 00014 Helsinki, Finland
    Search for more papers by this author
  • E. Tapio Palva

    Corresponding author
    1. Division of Genetics, Department of Biosciences, Viikki Biocenter, University of Helsinki, 00014 Helsinki, Finland
    Search for more papers by this author

*(fax +358 9 19159076; e-mail tapio.palva@helsinki.fi).

Summary

In addition to its essential metabolic functions, biotin has been suggested to play a critical role in regulating gene expression. The first committed enzyme in biotin biosynthesis in Arabidopsis, 7-keto-8-aminopelargonic acid synthase, is encoded by At5g04620 (BIO4). We isolated a T-DNA insertion mutant of BIO4 (bio4-1) with a spontaneous cell death phenotype, which was rescued both by exogenous biotin and genetic complementation. The bio4-1 plants exhibited massive accumulation of hydrogen peroxide and constitutive up-regulation of a number of genes that are diagnostic for defense and reactive oxygen species signaling. The cell-death phenotype was independent of salicylic acid and jasmonate signaling. Interestingly, the observed increase in defense gene expression was not accompanied by enhanced resistance to bacterial pathogens, which may be explained by uncoupling of defense gene transcription from accumulation of the corresponding protein. Characterization of biotinylated protein profiles showed a substantial reduction of both chloroplastic biotinylated proteins and a nuclear biotinylated polypeptide in the mutant. Our results suggest that biotin deficiency results in light-dependent spontaneous cell death and modulates defense gene expression. The isolation and molecular characterization of the bio4-1 mutant provides a valuable tool for elucidating new functions of biotin.

Introduction

Biotin (vitamin B8 or H) is an essential cofactor in the metabolism of all organisms, but its biosynthesis is restricted to bacteria, plants and some fungi (Alban et al., 2000; Roje, 2007). An essential role for biotin is its attachment to the active site of carboxylases such as acetyl CoA carboxylase (ACCase) and methycrotonyl CoA carboxylase (MCCase), which catalyze carboxylation reactions in crucial metabolic processes such as the synthesis and catabolism of amino acids, fatty acids and isoprenoids (Nikolau et al., 2003; Alban et al., 2000). Homozygous biosynthesis mutants in Arabidopsis (bio1, bio2 and bio3) are embryo-lethal (Muralla et al., 2008; Patton et al., 1996, 1998; Pinon et al., 2005; Schneider et al., 1989).

In animal cells, biotin has been proposed to have functions additional to its role in catalyzing carboxylation reactions (Kothapalli et al., 2005; Rodriguez-Melendez and Zempleni, 2003; Zempleni, 2005). In human cells, a large proportion of the enzyme holocarboxylase (HCS) localizes to the nucleus, and is associated with chromatin and the nuclear lamina (Narang et al., 2004). HCS biotinylates histones in vitro, and it has been suggested that histone biotinylation plays a role in silencing of chromatin and gene expression in animal cells (Beckett, 2009; Camporeale et al., 2007; Gravel and Narang, 2005; Narang et al., 2004; Kothapalli et al., 2005). Moreover, biotin deficiency stimulates nuclear factor κB (NF-κB) survival pathways (Griffin and Zempleni, 2005; Rodriguez-Melendez and Zempleni, 2003).

Programmed cell death (PCD) in the form of the hypersensitive response (HR) is a central feature in plant–pathogen interactions (Toth and Birch, 2005; Nomura et al., 2005), and is accompanied by the rapid death of plant cells in close proximity to invading pathogens (Greenberg, 1997). An oxidative burst generating reactive oxygen species (ROS) is a characteristic early feature of HR-PCD. ROS function in executing cell-death and as protective agents, and drive oxidative cross-linking in the cell wall, but also trigger defense gene expression and play a role in systemic acquired resistance (SAR) (Lamb and Dixon, 1997; Van Breusegem and Dat, 2006). The protein NON-EXPRESSOR OF PATHOGENESIS-RELATED1 (NPR1) has a central role in the establishment of defense responses and SAR, and is required for the expression of several pathogenesis-related (PR) genes. NPR1 is activated and localizes to the nucleus after alterations in the cellular redox status triggered by pathogens, salicylic acid (SA), ROS or elicitors (Mou et al., 2003), and activates defense gene expression by interaction with TGACG motif-binding factors (TGA) (Dong, 2004; Pieterse and Van Loon, 2004).

To unravel pathogen-associated cell death pathways and ROS signaling, a number of lesion-mimic mutants have been isolated (Lorrain et al., 2003), which exhibit accelerated or spontaneous cell death phenotypes. The majority of lesion-mimic mutants are also more resistant than wild-type plants to a number of pathogens, although enhanced lesion formation and induced defense gene expression are uncoupled from resistance in some cases (Greenberg et al., 2000; Ishikawa et al., 2001; Liang et al., 2003). Identification of mutants that misdirect cell death in plants has revealed substantial overlap or cross-talk between multiple signaling pathways controlling HR-PCD and defense responses (Liang et al., 2003; Lorrain et al., 2003, 2004; Meng et al., 2010; Mosher et al., 2010; Overmyer et al., 2003; Yao and Greenberg, 2006).

Here we report the isolation, characterization and identification of a spontaneous cell-death mutant of Arabidopsis. Molecular cloning and genetic complementation revealed that the phenotype is caused by a T-DNA insertion in At5g04620 (BIO4), which encodes the first committed enzyme of the biotin biosynthesis pathway. Unlike the other three genes involved in biotin biosynthesis, no mutants have yet been described for this gene (Muralla et al., 2008; Patton et al., 1996, 1998; Pinon et al., 2005; Schneider et al., 1989). The mutant was consequently named bio4-1, and accumulates H2O2 and shows up-regulation of defense- and ROS signaling-related genes. Interestingly, the mutation uncouples the constitutively elevated levels of H2O2 and PR gene expression from PR1 protein accumulation and bacterial disease resistance. The induced cell death phenotype appears to be SA/NPR1-independent, although maximal expression of SAR marker genes requires the presence of functional NPR1. Characterization of biotinylated proteins by streptavidin assays showed that the bio4-1 mutation leads to a significant decrease or even absence of a nuclear biotinylated polypeptide and two chloroplast-localized biotin-containing proteins.

Results

Isolation and identification of a T-DNA mutant with a spontaneous cell death phenotype

We screened a collection of Arabidopsis ecotype Columbia (Col-0) T-DNA lines generated using the pBIN19-derived vector pCP60 (Li et al., 2004), which harbors a constitutively expressed methionine sulfoxide reductase B9 gene (MsrB9; At4g21850), for lesion phenotypes. Seeds from more than 100 kanamycin-resistant T1 lines were individually selected to generate T2 and T3 soil-grown plants. This screen led to isolation of a lesion mutant exhibiting a spontaneous cell-death phenotype (Figure 1). The phenotype was not observed in the T1 generation, but 17 of 54 T2 plants showed spontaneous cell death in rosette leaves (1:3, χ2 = 0.422; 0.9 > > 0.5). No genetic segregation existed in progeny (T3–T6) derived from T2 lesion plants. These results indicate that lesion formation results from a recessive mutation at a single locus, and appears to be independent of over-expressed MsrB9.

Figure 1.

 Lesion formation in mutant plants under standard growth conditions.
(a) Five-week-old wild-type (left) and mutant (right) plants grown in soil.
(b) Three-week-old wild-type (left) and mutant (right) plants axenically grown on half-strength MS medium.
(c) Trypan blue staining of a representative leaf of axenically grown wild-type (left) and mutant (right) plants shows dead plant cells (indicated by a red arrow).

Homozygous mutants germinated with normal cotyledons, but typically developed visible lesions 2–3 weeks after germination (Figure 1a). Despite the increasing size of scattered lesions during leaf development and expansion, no rosette leaves became completely chlorotic. Like most lesion-mimic mutants of Arabidopsis characterized so far (Lorrain et al., 2003), mutants exhibited reduced size (Figure 1a) and had deeply serrated and slightly rolled-up leaf edges. Reducing the light cycle from 12 to 8 h without changing the light intensity suppressed the formation of visible necrotic lesions in rosette leaves of mutant plants (data not shown), suggesting that lesion formation was dependent on the day length. Lesions also appeared consistently in axenically grown mutant plants (Figure 1b), showing that lesion formation does not require any pathogen stimulus or infection. The lesions were clearly necrotic and contained substantial amounts of dead cells, as indicated by trypan blue staining (Figure 1c). Compared with wild-type, mutants initiated earlier reproductive growth, but seed yield was drastically reduced (data not shown).

Molecular characterization and complementation of the mutant

To identify the interrupted gene, we performed genome walking (Siebert et al., 1995). The T-DNA insert is located in the 5′ UTR of At5g04620 (Figure S1A). At5g04620 has been named both AtbioF (Pinon et al., 2005) and BIO4 (Muralla et al., 2008), so the mutant was named bio4-1. This gene encodes 7-keto-8-aminopelargonic acid synthase, the first committed enzyme of the biotin biosynthesis pathway (Pinon et al., 2005). To assess the effect of the insertion on BIO4 expression, a 1075 bp full-length fragment from a leaf cDNA library (Li et al., 2004) was used as a probe in RNA gel-blot analysis. BIO4 transcripts were detectable in wild-type but not mutant leaves (Figure S1B). RT-PCR revealed very low levels of BIO4 expression (Figure S1C). These results indicate that the T-DNA interrupts the 5’ UTR of the BIO4 gene, resulting in a drastic down-regulation of BIO4 expression and a possible reduction of biotin levels in bio4-1.

To test the hypothesis that bio4-1 mutants suffer from a lack of d-biotin, we transferred 1-week-old bio4-1 seedlings to half-strength MS medium containing d-biotin. Visible cell death symptoms in bio4-1 were significantly reduced with increasing concentrations of d-biotin, and completely disappeared at final concentration of 1 μM d-biotin (Figure S1D). Soil-grown bio4-1 plants were also rescued by the addition of d-biotin (data not shown). To confirm that the bio4-1 phenotype exclusively results from reduced expression of the BIO4 gene, we transformed a 3699 bp genomic clone of At5g04620, including a 1087 bp promoter fragment, into bio4-1 plants. The complemented bio4-1 plants exhibited similar BIO4 expression and morphological traits as vector controls and wild-type (Figure S1E), confirming the hypothesis that the observed phenotype of the bio4-1 plants is the consequence of the decreased production of biotin caused by reduced expression of At5g04620.

Accumulation and subcellular localization of H2O2 in the bio4-1 mutant

To determine the nature of bio4-1-associated lesions, we examined whether ROS levels are increased in this mutant. Leaves were stained with either diaminobenzidine (DAB) or nitroblue tetrazolium (NBT). Enhanced H2O2 accumulation irrespective of day length was clearly apparent in DAB-stained leaves of the bio4-1 mutant but absent in control plants (Figure 2a). Co-infiltration with catalase significantly reduced DAB staining, confirming H2O2 generation in bio4-1. Interestingly, the blue formazan precipitation produced by reaction of NBT with O2 (Figure 2a) was not reduced even when very high concentrations of a commercial superoxide dismutase (SOD) were co-infiltrated into the leaves (data not shown), suggesting that either the amount of exogenous SOD is not sufficient to catalyze the conversion of O2 or that uncharacterized radicals accumulate in bio4-1 mutants. Positive DAB and NBT staining was abolished in bio4-1 mutants complemented either genetically with BIO4 or by 1 μM d-biotin added to the growth medium (data not shown).

Figure 2.

 H2O2 accumulation and SOD gene expression in the bio4-1 mutant.
(a) Diaminobenzidine (DAB) staining for H2O2 and nitroblue tetrazolium (NBT) staining for O2 in leaves of 4-week-old bio4-1 mutant (upper panels) and wild-type (WT) (lower panels) plants. ‘+’ indicates co-infiltration with catalase. The day length during plant growth and staining method is indicated below.
(b) Transmission electron micrograph of cross-sections of 4-week-old wild-type (WT, left) and bio4-1 plants (right). The images in the bottom panels are of the regions that are boxed in the upper panels. CV, central vacuole; CW, cell wall; ER, endoplasmic reticulum; IS, intercellular space; M, mitochondia; XV, xylem vessel.
(c) RNA gel-blot analysis of expression of superoxide dismutase (SOD) genes in 4-week-old wild-type (WT) and bio4-1 mutant plants.

To elucidate the subcellular localization of the accumulated H2O2 in bio4-1, we detected deposits of cerium perhydroxides by electron microscopy, indicating the presence of H2O2 (Bestwick et al., 1997). Precipitates were predominantly located within and along the xylem cell walls of vascular bundles in bio4-1 (Figure 2b, right panels). Moreover, deposits of cerium perhydroxides were also found within the endoplasmic reticulum of xylem cells in the mutant. In contrast, only low levels of cerium perhydroxides were occasionally found in the cell walls of vascular bundles in wild-type (Figure 2b, left panels). Plants contain a set of genes encoding three classes of SOD activity that differ in terms of the active site metal cofactors (Fe, Mn or Cu/Zn). These plant isoenzymes appear in different subcellular locations, and are often induced at mRNA and protein levels in response to various pathogen-associated stimuli and oxidative stress (Kliebenstein et al., 1998, 1999). To determine whether expression of specific SODs is implicated in the cell death phenotype of bio4-1, we characterized the transcript levels of four major leaf SOD genes (CDS1, CDS2, FSD1 and MSD). The mRNA levels did not show any substantial alteration in the bio4-1 mutant (Figure 2c), indicating that the bio4-1 lesion phenotype and ROS accumulation do not correlate with expression of these ROS detoxification enzymes.

Role of defense-related hormones in bio4-1 lesion formation

SA, jasmonic acid (JA) and ethylene (ET) are three major hormones implicated in plant defense responses (Kunkel and Brooks, 2002; Glazebrook, 2005). To determine whether the spontaneous cell-death phenotype of the bio4-1 mutant is accompanied by hormonal imbalances, we determined the endogenous levels of hormones involved in plant defense. The levels of JA and ET were not significantly altered, and the level of free SA was slightly elevated in 4-week-old bio4-1 plants in comparison to wild-type (Figure S2). In contrast, total SA levels were clearly increased in the bio4-1 mutant (Figure S2A), indicating a significant increase in the content of conjugated SA-glucoside. Similar results were also obtained with younger 2-week-old bio4-1 plants (data not shown). These data suggest that cell death in the bio4-1 mutant do not result from alterations in the balance of defense-related hormones, and that the moderate change in free hormone levels may instead be a consequence of lesion formation.

SA signaling is required for cell death in many lesion-mimic mutants (Lorrain et al., 2003; Noutoshi et al., 2006). To determine whether SA or JA signaling is essential for the spontaneous cell-death phenotype in the bio4-1 mutant, we generated the following double mutants: bio4-1/NahG, which is deficient in SA accumulation (Delaney et al., 1994), bio4-1/npr1-1, which shows impaired SA-mediated induction of SAR marker gene expression (Cao et al., 1997), and bio4-1/coi1-1, which is defective in JA signaling (Feys et al., 1994). Morphologically, all homozygous double mutants resemble the bio4-1 mutant, with visible spontaneous lesions, serrated leaves and reduced size compared to wild-type (Figure S3A). Also, H2O2 accumulation appears to be independent of functional NPR1, as indicated by comparable H2O2 accumulation in bio4-1 and bio4-1/npr1-1 double mutants (Figure S3A). To obtain a more quantitative measure of lesion formation in bio4-1, we determined the total chlorophyll content. As shown in Figure S3B, both bio4-1/npr1-1 and bio4-1/NahG exhibited similar levels of chlorophyll in comparison to the single mutant bio4-1, confirming that the lesion phenotype is independent of NPR1 and SA. Taken together, these results indicate that the cell-death phenotype of the bio4-1 mutant is independent of SA and JA signaling.

Patterns of overall gene expression in the bio4-1 mutant

To obtain a better understanding of the overall gene expression pattern in the bio4-1 mutant, we compared expression in bio4-1 and wild-type plants using an ATH1 Gene Chip array with more than 22 500 probe sets. Genes with a mean differential expression greater than or equal to twofold in three replicates and a t test P value < 0.05 were assigned to two groups (activation or repression). To identify genes that were differentially expressed specifically due to the bio4-1 mutation and avoid any false positives due to over-expression of MsrB9 in the bio4-1 mutant line, we omitted genes that were also differentially expressed between the bio4-1 mutant and the mutant complemented with d-biotin. Taking this into account, 153 probe sets (corresponding to 152 genes) were significantly up-regulated (Table S1) and 26 probe sets (26 genes) were significantly down-regulated (Table S2).

To identify possible over-represented Gene-Ontology (GO) attributes we were using the GO Term Enrichment tool (http://amigo.geneontology.org/cgi-bin/amigo/term_enrichment). In the up-regulated group, various GO terms related to pathogen and defense responses can be found (Table S3). Up-regulated genes include pathogenesis-related (PR) genes and genes for several glutathione S-transferases, indicative of enhanced oxidative stress (Table S1). As expression of defense-related genes and the establishment of SAR are dependent on NPR1, we were interested to determine the number of up-regulated genes controlled by NPR1. Of 187 described NPR1 primary targets (Wang et al., 2005), seven (3.7%) are up-regulated in bio4-1 (Table S1) compared to a total of 0.7% up-regulated genes in bio4-1, which indicates significant over-representation (P < 0.001) of direct NPR1 targets among the up-regulated genes. Among the significantly down-regulated genes in bio4-1, GO attributes corresponding to abiotic stress (response to cold and hormone stimulus) are over-represented (Table S3). No direct NPR1 target was detected in the down-regulated genes. Taken together, these data suggest that the BIO4-1 locus may control and differentially regulate multiple biological processes, particularly those related to biotic and abiotic stresses.

Expression of SAR marker genes in the bio4-1 mutant is enhanced but does not require SA/NPR1

The up-regulation of defense-related genes in the bio4-1 mutant observed in the microarray analysis, and the fact that a number of lesion-mimic mutants constitutively express defense-related PR genes (Lorrain et al., 2003, 2004; Noutoshi et al., 2006), prompted us to verify the up-regulation of the SAR marker genes (Uknes et al., 1992) PR1, PR2 and PR5 using RNA blots, and to characterize whether this expression is dependent on the known defense pathway elements NPR1 and COI1 and the ability to accumulate SA (Figure 3). As shown in Figure 3(a), expression of PR1, PR2 and PR5 is clearly enhanced in the leaves of the bio4-1 mutant, confirming the results obtained in the microarray. In addition, GST1 transcripts encoding a glutathionine S-transferase implicated in stress responses such as pathogen attack and oxidative stress accumulated in the mutant (Figure 3a).

Figure 3.

 Epistasis analysis of defense-related gene expression.
(a) PR1, PR2, PR5 and GST1 expression in leaves of 3-week-old Col-0, bio4-1, npr1-1 and bio4-1/npr1-1 plants.
(b) PR1 and GST1 expression in leaves of 3-week-old Col-0, bio4-1, NahG, coi1-1, bio4-1/NahG and bio4-1/coi1-1 plants.
RNA samples were extracted from 4-week-old soil-grown plants. For each sample, 20 μg of total RNA was loaded. A probe against 18S rRNA is shown as a loading control. Similar results were obtained in three independent experiments.

To determine the requirement for functional NPR1 for constitutive expression of SAR markers in bio4-1 plants, we characterized expression of these genes in the double mutant bio4-1/npr1-1 by RNA gel-blot analysis. Interestingly, up-regulation of PR1, PR2 and PR5 was still evident in the npr1-1 background, albeit substantially reduced (Figure 3a). Consistent with these results, the constitutive expression of PR1 in bio4-1 was significantly reduced but not completely abolished in bio4-1/NahG plants that are unable to accumulate SA (Figure 3b). Furthermore, bio4-1-dependent up-regulation of PR1 is further enhanced in bio4-1/coi1-1 double mutants, which are defective in JA signaling, consistent with possible negative feedback regulation in JA- and SA-related signaling (Li et al., 2004, 2006; Koornneef and Pieterse, 2008). Unlike the PR genes, expression of GST1 was barely altered in the bio4-1/npr1-1 and bio4-1/coi1-1 double mutants (Figure 3), indicating that the up-regulation of GST1 expression in bio4-1 is independent of functional NPR1 and JA signaling. Taken together, these results indicate that the up-regulation of SAR markers in the bio4-1 mutant is enhanced by but does not require SA/NPR1-mediated signaling.

Susceptibility of the bio4-1 mutant to bacterial pathogens

To explore whether constitutive activation of PR genes and massive accumulation of hydrogen peroxide in cell walls of bio4-1 plants are associated with disease resistance, we assessed the growth of two bacterial phytopathogens, Pseudomonas syringae pv. tomato DC3000 (Pst), which results in chlorosis in Arabidopsis wild-type plants (Katagiri et al., 2002), and Pectobacterium carotovorum (Erwinia carotovora subsp. carotovora) SCC1 (Ecc), which causes tissue maceration (Li et al., 2004). Strikingly, the bio4-1 mutant displayed disease susceptibility similar to that of wild-type (Figure 4a). No differential disease resistance was observed between wild-type and bio4-1 plants following bacterial inoculation using various inoculum sizes for both Pst DC3000 and Ecc SCC1 suspensions.

Figure 4.

 Bacterial growth in the bio4-1 mutant and PR1 protein analysis.
(a) Growth of bacterial pathogens Pst DC3000 and Ecc SCC1 in planta. Wild-type (WT) and bio4-1 leaves were analyzed for bacterial numbers at various time points after bacterial infiltration. Values are means ± SE of three pools from eight replicate samples. Experiments were repeated three times with similar results.
(b) Immunoblot of PR1 protein in wild-type (WT) and bio4-1 plants. PR1 protein levels before and 24 h after treatment with 5 mM SA were determined from 100 μg total proteins separated on SDS gels and blotted to nitrocellulose using polyclonal PR1 antibody. Coomassie staining confirmed equal loading of the total proteins (bottom panel). This experiment was repeated three times with similar results. hpt, hours post-treatment.

This apparent discrepancy between defense gene expression and disease resistance was observed with some previously isolated lesion mutants that showed enhanced PR1 expression but unaltered disease resistance (Lorrain et al., 2003). Furthermore, it has been shown that successful establishment of SAR requires activation of the protein secretory pathway (Wang et al., 2005). To assess the potential causes of the lack of enhanced disease resistance in bio4-1 plants, we therefore determined whether the bio4-1 mutants show increased levels of secreted PR1 proteins in addition to elevated expression of the PR1 gene (Wang et al., 2005). In agreement with our observations in the pathogen assays, but unlike the results from PR1 expression analysis, bio4-1 plants did not show constitutively enhanced levels of PR1 protein. As a control, we showed that SA, a known inducer of PR1, induce accumulation of PR1 proteins to similar levels in both the wild-type and bio4-1 mutants (Figure 4b). These results indicate that defects in biotin accumulation and the accompanying production of H2O2 and expression of PR genes in bio4-1 are not sufficient to confer enhanced disease resistance against two pathogens. Moreover, only a single gene of the secretory pathway is significantly up-regulated in the bio4-1 mutant (Table S1), which suggests that bio4-1 does not constitutively activate the machinery for build-up or accumulation of PR proteins.

Alterations in the biotin-containing protein profile in the bio4-1 mutant

Biotin, as a catalytic cofactor covalently bound to lysine residues of a group of enzymes catalyzing carboxylation, plays a vital role in the catabolic and anabolic metabolic processes (Alban et al., 2000). Several biotin-containing proteins have been characterized in Arabidopsis plants, including the chloroplast-localized biotin carboxyl carrier proteins AtBCCP1 (approximately 35 kDa) (Choi et al., 1995) and AtBCCP2 (approximately 25 kDa) (Thelen et al., 2001), as well as a biotinyl subunit (also referred to referred as to AtMCC-A; Che et al., 2003) of mitochondrial-localized methylcrotonoyl CoA carboxylase (MCCase) of approximately 80 kDa. To investigate whether the T-DNA insertion in the bio4-1 locus leads to reduced levels of biotin-containing proteins, we detected biotin-containing proteins with streptavidin peroxidase based on the streptavidin–biotin interaction. Western blot analysis revealed that a polypeptide with an approximate size of 80 kDa was predominant in mutant and wild-type leaf extracts, and an additional 35 kDa polypeptide was only weakly present in the mutant. A 25 kDa polypeptide was only detectable in wild-type plants (Figure 5a). These results are consistent with previous observations in wild-type leaves of Arabidopsis (Che et al., 2003; Thelen et al., 2001). Interestingly, the amounts of proteins of the same size as the chloroplast-localized biotin-containing proteins AtBCCP1 and AtBCCP2 were strongly reduced in the bio4-1 mutant. In contrast, the levels of protein of the size of mitochondrial-localized AtMCC-A were barely affected in bio4-1. We also considered the possibility that reduced levels of hitherto uncharacterized biotin-containing proteins may contribute to the observed phenotype in the bio4-1 mutant. To assess this, we isolated nuclei and concentrated nuclear proteins (Turck et al., 2004). Three bands of sizes 80 kDa (data not shown) and 37 and 35 kDa (Figure 5b) were present in the wild-type, but only the 80 kDa band was visible in the bio4-1 mutant. Most likely, the bands with sizes of 80 and 35 kDa represent contamination with the non-nuclear biotin-containing proteins AtMCC-A and AtBCCP1. Interestingly, a protein of size 37 kDa was present in the wild-type but absent in bio4-1 (Figure 5b). Our data therefore indicate the existence of an additional, so far uncharacterized, biotinylated nuclear-localized protein.

Figure 5.

 Biotin-containing proteins in wild-type (WT) and bio4-1 plants.
(a) Total proteins (100 μg) were separated on a 12% SDS–PAGE gel, blotted to nitrocellulose, conjugated with streptavidin peroxidase, and visualized by chemiluminescence. Exposure time was 5 or 15 min (for the lower part containing the 25 kDa protein.) As a loading control, the SDS–PAGE gel was visualized by Coomassie blue staining (bottom panel).
(b) After fractionation, cytosolic and nuclear proteins were separated on a 15% SDS–PAGE gel and blotted and visualized as above.
This experiment was repeated three times with similar results.

Discussion

Biotin is essential for all living organisms, and must be either taken up in the diet or synthesized de novo (Alban et al., 2000). In plants, the requirement for biotin is illustrated by the fact that all homozygous biotin biosynthesis mutants of Arabidopsis isolated so far are embryo-lethal (Muralla et al., 2008; Patton et al., 1996, 1998; Pinon et al., 2005; Schneider et al., 1989). Here, we characterize a recessive and viable mutant in the committed step (7-keto-8-aminopelargonic acid synthase; BIO4) of the biotin biosynthesis pathway. First, we show that a T-DNA insertion in the 5’ UTR of BIO4 leads to reduced expression of this gene. This in turn results in a decrease in the fraction of biotinylated proteins, most likely due to biotin starvation. Second, we show that the biotin deficiency results in accumulation of H2O2 and a lesion-mimic phenotype, independently of SA and JA signaling. Third, our studies show that biotin deficiency leads to a reprogramming of the transcriptome, resulting in enhanced expression of genes encoding defense-related proteins and down-regulation of genes involved in abiotic stress responses. Fourth, unlike the lesion-mimic phenotype itself, expression of PR genes in the bio4-1 mutant is partly dependent on NPR1 and SA signaling. Fifth, despite enhanced PR gene expression, bio4-1 mutants do not have elevated levels of PR-1 protein and do not exhibit enhanced resistance to the bacterial pathogens Ecc SCC1 and Pst DC3000. Both genetic complementation with a genomic clone of BIO4 and external application of biotin can suppress the lesion-mimic phenotype, H2O2 accumulation and a substantial proportion of the observed transcriptome reprogramming, confirming that low biotin levels or lower availability of biotin in specific compartments is responsible for the observed phenotypes.

The bio4-1 mutation is caused by insertion of a T-DNA harboring a constitutively expressed methionine sulfoxide reductase B9 gene (MsrB9; At4g21850). Thus, the question arises whether over-expression of MsrB9 plays a role in the formation of the observed phenotype. This is unlikely for three reasons. First, of 16 independent over-expression lines with similar levels of MsrB9 expression, only the bio4-1 mutant line shows the characteristic growth and lesion phenotype (data not shown). Second, the growth and lesion phenotype is recessive, with bio4-1 heterozygotes over-expressing MsrB9 not showing any detectable phenotype. Third, a substantial proportion of the observed changes in the transcriptome are due to biotin depletion, as 46% of all the genes that are significantly and more than or equal to twofold differentially regulated (and 91% of the genes that are significantly and more than or equal to 0.5 (log2)-fold up- or down-regulated) in the bio4-1 mutant compared to wild-type are also differentially regulated in the same way in the bio4-1 mutant compared to the bio4-1 mutant supplemented with external biotin (Tables S1 and S2; GEO accession GSE26631). However, as Msr proteins may play a role in oxidative stress tolerance by repairing peptide damage (Bechtold et al., 2004; Vieira Dos Santos et al., 2005; Tarrago et al., 2009), we cannot exclude the possibility that over-expression of MsrB9 may confer elevated oxidative stress tolerance that may contribute to the survival of biotin-depleted plants. Together with the requirement for biotin in all living organisms and the position of the T-DNA insertion in the 5’ UTR of BIO4 allowing some residual expression of the gene (Figure S1), it is very likely that the survival of the bio4-1 mutant is due to leakiness of the mutation.

In Arabidopsis, biotin is known to be an essential cofactor for the functionality of three enzymes (Nikolau et al., 2003): it is part of the cytosolic homomeric ACCase, the plastidial heteromeric ACCase and mitochondria-localized MCCase. While biotinylation of MCCase and the homomeric ACCase appear not to be altered in bio4-1, as indicated by equally intense biotinylated protein bands at 80 kDa (Figure 5a) and a large band of > 200 kDa, respectively (data not shown), two other biotinylated bands at 25 and 35 kDa are clearly reduced in the bio4-1 mutant. The size of these two bands corresponds to the two chloroplastic Arabidopsis biotin carboxyl-carrier proteins (BCCPs) (Choi et al., 1995; Thelen et al., 2001), which are part of the heteromeric ACCase enzyme complex. As the final step of the biotin biosynthetic pathway is localized within mitochondria (Picciocchi et al., 2001, 2003), free biotin must be transported to other subcellular organelles forming biotin-containing proteins. Our interpretation is that biotin biosynthesized in mitochondria is limited in the bio4-1 mutant, especially in target organelles such as the chloroplast and the nucleus, while the mitochondrial MCCase is more easily saturated with biotin at the site of synthesis. Both homomeric and heteromeric ACCase catalyzes the ATP-dependent carboxylation of acetyl CoA to form malonyl CoA, which is the precursor for a wide range of fatty acids, secondary metabolites and some amino acids (Alban et al., 2000; Nikolau et al., 2003). As the plastid envelope is not permeable to malonyl CoA (Alban et al., 2000), plastidial ACCase activity is required to sustain fatty acid synthesis in the chloroplast. Plants without functional AtBCCP2 do not show visible phenotypes (Li et al., 2011; Thelen and Ohlrogge, 2002), but mutants of BCCP1 are embryo-lethal and antisense plants are severely affected in vegetative growth and show decreased fatty acid accumulation (Li et al., 2011). In the bio4-1 mutant, the content and ratio of C16 and C18 fatty acids of leaves are not affected (data not shown). The most likely explanation is that reduced biotinylation of plastidic ACCase leads to slower de novo fatty acid synthesis, resulting in the slower growth and dwarfism observed (Figures 1 and 2).

In addition to its catalytic function as a coenzyme for carboxylases, biotin plays a role in regulating gene expression (Alban et al., 2000; Beckett, 2009; Che et al., 2003; Zempleni, 2005) Non-enzymatic functions for biotin in animals and bacteria have been documented and discussed previously (Beckett, 2009; McMahon, 2002; Rodriguez-Melendez and Zempleni, 2003), including nuclear translocation of NF-κB and chromatin remodeling by histone biotinylation (Camporeale et al., 2007; Zempleni, 2005; Pestinger et al., 2011). It has been suggested that the plant protein NPR1 is a homolog of mammalian signal transduction factor IκB (Ryals et al., 1997), which regulates the activity and subcellular localization of NF-κB in response to biotin (Zempleni, 2005). Do our results with the bio4-1 mutant suggest the existence of biotin-dependent regulation of NPR1-controlled pathways in plants? Both lesion formation and H2O2 accumulation in the bio4-1 background are independent of NPR1, as indicated by results obtained with bio4-1/npr1-1 double mutants (Figure S3A), suggesting that biotin-dependent re-localization of NPR1 is not responsible for the observed phenotypes. Up-regulation of PR genes in the bio4-1 mutant is clearly at least partly dependent on functional NPR1, but expression of the oxidative stress marker GST1 (Vranova et al., 2002) is not (Figure 3a). The redox activation of NPR1 is a prerequisite for expression of PR genes (Mou et al., 2003; Tada et al., 2008). This suggests that the elevated levels of H2O2 in bio4-1 plants may provide a cellular redox environment for re-localization and activation of NPR1, leading to enhanced PR gene expression. Establishment of SAR and enhanced PR gene expression in response to pathogen recognition is associated with elevated levels of SA at both the site of infection and in systemic tissues (Lamb and Dixon, 1997). Interestingly, the decreased levels of biotin in the bio4-1 mutant only trigger moderate changes in the level of free SA, although the levels of total SA (free SA and SA-glucoside) are increased threefold (Figure S2A). This is partly in agreement with previous studies showing that increased H2O2 levels by endogenous over-production of H2O2 lead to induction of PR1 gene expression without alteration in levels of free SA (Vranova et al., 2002; Wu et al., 1997). The observed elevated levels of total SA in the bio4-1 plants may indicate previous high levels of SA in or around developing lesions, which is subsequently detoxified to SA-glucoside (Seo et al., 1995). Interestingly, lesion formation does not appear to require SA or JA signaling: removal of SA by salicylate hydrolase and blocking of JA signaling does not alter lesion formation in bio4-1 plants (Figure S3). Taken together, our results show that lesion formation is independent of NPR1, SA and JA signaling, but up-regulation of PR genes in bio4-1 plants occurs at least partly through SA- and NPR1-dependent mechanisms.

Despite the enhanced generation of H2O2 and constitutive expression of PR genes in the bio4-1 mutant, these plants are unable to resist the growth of virulent pathogens (Figure 4a). This is in agreement with results obtained for some lesion-mimic mutants, such as accelerated death 5 (acd5) and lesion initiation 2 (lin2), which do not show increased resistance to Pseudomonas syringae despite elevated PR1 expression (Greenberg et al., 2000; Ishikawa et al., 2001; Liang et al., 2003). What is the difference between bio4-1 and the majority of lesion-mimic mutants with elevated PR gene expression and enhanced disease resistance? In contrast to other lesion-mimic mutants such as cpr5 and vad1 (Bowling et al., 1997; Lorrain et al., 2004), bio4-1 plants only accumulate slightly elevated levels of free SA (Figure S2A). Importantly, although PR1 gene expression is up-regulated, PR1 protein levels are not altered in the bio4-1 mutants (Figure 4b). However, application of external SA induces PR1 accumulation in both wild-type and the bio4-1 mutant (Figure 4b), indicating that additional SA is required to fully activate PR1 protein production. The NPR1-regulated expression of ER-resident genes is essential for establishment of SAR (Wang et al., 2005). The bio4-1 mutant shows clearly enhanced expression of a large subset of NPR1 targets, but only one gene encoding a component of the secretory pathway is significantly up-regulated (Table S1) (Wang et al., 2005, 2006), and expression of the remaining secretory pathway genes is largely unchanged. Based on our results, it is evident that either some components of the secretory pathway or the translation machinery are limiting in the bio4-1 mutant and require an additional trigger (such as SA). It is also possible that H2O2 accumulation in the ER apparatus may cause ER damage or block the secretory pathway for extracellular PR proteins, and again additional triggers may be needed to release the block.

How does biotin deficiency cause accumulation of H2O2, lesion formation and transcriptional alterations? Previously, it has been shown that expression of certain biotin-containing protein genes is regulated by biotin itself in plants (Che et al., 2003), but so far no mechanism as to how biotin could regulate gene expression has been described. Althought the exact mechanism is still unclear, there are three possibilities. First, our data show the existence of a nuclear-localized biotin-containing protein (Figure 5b), and it is possible that biotin modulates the function of this uncharacterized nuclear-localized protein, which in turn could be directly responsible for the subsequent expression of target genes or the modulation of histones resulting in epigenetic regulation, as discussed above for animals. Alternatively, based on the known function of biotinylated proteins in lipid biosynthesis (Alban et al., 2000), we cannot exclude the possibility that the mutant has an altered composition of specific, possible uncharacterized lipids, which may act as messengers in oxidative stress signaling. In this respect, it is interesting to note that the dicarboxylic fatty acid azaleic acid was recently shown to be a component of long-distance stress signaling in plants (Jung et al., 2009). Third, the extensive reprogramming of the transcriptome, lesion formation and hydrogen peroxide generation may be a result of pleiotropic effects due to perturbation of cellular processes requiring the well-characterized biotin-containing enzymes. The isolation of a viable biotin mutant provides a unique opportunity to study the function of biotin in all aspects of plant development and plant stress responses. Furthermore, identification of the nuclear-localized biotin-containing protein is likely to advance our understanding of the biological functions of biotin.

Experimental procedures

Mutant isolation

A full-length fragment of MsrB9 (At4g21850) was amplified using primers 5’-ACACACGACCATCTTTCCGCTGTCTTC-3′ and 5′-CTTGCATTGAACATATCAAGGGTCAATGG-3′, and cloned into the pBIN19-derived vector pCP60 (Li et al., 2004) containing the CaMV 35S promoter. A. thaliana plants were transformed by floral dip (Clough and Bent, 1998). The locations of T-DNA insertions were determined by genome walking using a GenomeWalker kit (Clontech, http://www.clontech.com/) according to the manufacturer’s instructions and as described by Siebert et al. (1995). In addition to DraI, EcoRV, PvuII and StuI DNA libraries a XmnI library was constructed. 5% DMSO was added to the PCR reactions. The left border T-DNA-specific primers were 5′-TGGTTCACGTAGTGGGCCATCG-3′ and 5′-GCGTGGACCGCTTGCTGCAACT-3′.

Plant material, biotin application and growth conditions

Arabidopsis thaliana (Col-0) plants were grown under conditions as previously described (Li et al., 2008). d-biotin (Sigma-Aldrich, http://www.sigmaaldrich.com/) solution (pH 6.5) was directly added to MS medium (Sigma-Aldrich) or to soil by watering weekly with 200 μM d-biotin until Agrobacterium-mediated transformation or harvesting for isolation of RNA.

Genetic complementation

The genomic DNA sequence of BIO4 (At5g04620) was amplified by PCR using primers 5′-TTCCTTGATTTTGATTTCACCATCATCC-3′ and 5′-GTGAAGAAGAAGAAGACATAATTGGAGA-3′. The product was inserted into pCR2.1 (Invitrogen, http://www.invitrogen.com/) and cloned using BamHI/NotI into the binary vector pCPBAR [a pCP60-derived vector (Kariola et al., 2005), in which nptII is replaced by the bialophos resistance gene (BAR) conferring resistance to phosphinothricin]. This plasmid was transfered to Agrobacterium tumefaciens strain GV2260, which was used for floral-dip transformation of bio4-1 plants that had been treated with 200 μM d-biotin weekly until dipping. Transformants were selected on half-strength MS medium containing 15 mg l−1 phosphinothricin (BASTA) (Duchefa Biochemie, http://www.duchefa.com).

Histochemistry

Trypan blue, nitroblue terazolium and diaminobenzidine staining were performed as described by Bowling et al. (1997), Jabs et al. (1996) and Kariola et al. (2005), respectively. Catalase (Sigma, C3515) and superoxide dismutase (Sigma, S4636) were used at 100 units μl−1 and 100–500 units μl−1, respectively.

Transmission electron microscopy

Leaves from 2-week-old soil-grown plants were used for localization of H2O2 generation. Samples were stained, fixed, post-fixed, dehydrated, embedded and sectioned as described by Bestwick et al. (1997) with the following modifications. Leaves were infiltrated under vacuum with 5 mM CeCl3 in 50 mM MOPS (pH 7.0), and fixed in 2% glutaraldehyde/2% paraformaldehyde in 50 mM sodium cacodylate (Sigma) buffer, pH 7.0. Tissue pieces (1–2 mm2) were excised after fixation.

DNA and RNA analyses

RNA gel-blot hybridizations with digoxigenin (DIG) were performed as described previously (Li et al., 2004). RT-PCR analysis was performed using 1 μg of total RNA treated with DNaseI (GE Healthcare, http://www.gelifesciences.com) and SuperScript II RNase H reverse transcriptase (Invitrogen) and oligo(dT)16 according to the manufacturer’s instructions. PCR was performed for 35 cycles of 94°C for 45 sec, 58°C for 1 min and 72°C for 1 min, followed by a final extension at 72°C for 5 min. The following primer pairs were used: 5′-CAGTCAAAGAATATGGTATGGGACCTA-3′ and 5′-CGACAACAGAAAGAAGTTTTATAATTTGGG-3′ for BIO4, and 5′-GACATGGAAAAGATATGGCATCACAC-3′ and 5′-AGATCCTTCCTGATATCGACATCAC-3′ for actin (control).

Microarray analysis

Total RNA from a pool of five 5-week-old soil-grown Col-0 wild-type and bio4-1 plants and bio4-1 plants supplemented with d-biotin was extracted using the RNeasy Plant mini kit protocol (Qiagen, http://www.qiagen.com/), followed by an additional sodium acetate/isopropanol precipitation step. Total RNA (3 μg) was submitted to the Nottingham Arabidopsis Stock Centre International Affymetrix Service, where synthesis of biotin-labeled cRNA, microarray hybridization to Arabidopsis ATH1 Gene Chip arrays (Affymetrix, http://www.affymetrix.com/) and scanning procedures were performed.

Three independent biological experiments were performed, and CEL files were analyzed using affylmGUI (http://bioinf.wehi.edu.au/affylmGUI/) using robust multi-array average (RMA) normalization (Irizarry et al., 2003). To identify differentially regulated genes in ATH1 Gene Chip arrays, probe sets with more than or equal to twofold expression differences and a P value < 0.05 for an empirical Bayes two-group test between bio4-1 mutants and wild-type were selected and listed in Tables S1 and S2. The genes that showed also significant differences between bio4-1 mutants and bio4-1 mutants complemented with d-biotin to correct for expression differences resulting from over-expressing MsrB9 alone are listed in Tables S1A and S2A. Gene functions and identities are based on Arabidopsis Information Resource annotations. Over-represented Gene Ontology (GO) attributes in the up- or down-regulated genes were searched using FuncAssociate (http://llama.mshri.on.ca/funcassociate/) as described previously (Berriz et al., 2003).

The microarray data were deposited in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/; accession number GSE26631).

Pathogen infection

Pathogen infection with Pseudomonas syringae pv. tomato DC3000 and Pectobacterium carotovorum SCC1 was performed as described previously (Li et al., 2008).

Protein extraction and Western blot analyses

Total protein extraction and Western blotting were performed as previously described (Wang et al., 2005), and nuclear and non-nuclear protein extracts were obtained essentially as described by Turck et al. (2004). Detailed protocols are described in Appendix S1.

Acknowledgments

We thank J. Turner (University of East Anglia, Norwich, UK) for coi1-1, J. Ryals (Research Triangle Park, NC, USA) for NahG seeds and X. Dong (Duke University, Durham, NC, USA) for discussion and PR1 antibodies. We thank the Electron Microscopy Unit of the University of Helsinki for the transmission electron microscopy service. This work was supported by the Academy of Finland (grants number 38033, 42180, 49905, 44252 and 44883; Finnish Center of Excellence Program 2000–2005 and 2006–2011), Biocentrum Helsinki, and the Helsinki Graduate School in Biotechnology and Molecular Biology.

Ancillary