BAP1 and BAP2 are general inhibitors of programmed cell death 1

The Arabidopsis genes Abstract Here we identify the BAP1 and BAP2 genes of Arabidopsis thaliana as general inhibitors of programmed cell death across the kingdoms. These two homologous genes encode small proteins containing a calcium-dependent phospholipid binding C2 domain. BAP1 and its functional partner BON1 have been shown to negatively regulate defense responses and a disease resistance gene SNC1 . Genetic studies here reveal an overlapping function of the BAP1 and BAP2 genes in cell death control. The loss of BAP2 function induces accelerated hypersensitive responses but does not compromise plant growth or confer enhanced resistance to virulent bacterial or oomycete pathogens. The loss of both BAP1 and BAP2 confers seedling lethality mediated by PAD4 and EDS1 , two regulators of cell death and defense responses. Overexpression of BAP1 or BAP2 with their partner BON1 inhibits programmed cell death induced by pathogens, the proapototic gene BAX and superoxide-generating paraquat in Arabidopsis or Nicotiana benthamiana . Moreover, expressing BAP1 or BAP2 in yeast alleviates cell death induced by hydrogen peroxide. Thus, the BAP genes function as general negative regulators of programmed cell death induced by biotic and abiotic stimuli including reactive oxygen species. The dual roles of BAP and BON genes in repressing defense responses mediated by disease resistance genes and in inhibiting general programmed cell death has implications in understanding the evolution of plant innate immunity.


Introduction
the largest family in Arabidopsis contain a nucleotide binding (NB) domain and a leucine rich repeats (LRR) domain at the carboxyl terminus, with either a Coiled-coil (CC) domain or a Toll/Interleukin-1-Receptor (TIR) domain at the amino-terminus (Meyers et al., 2003). Although examples of direct physical interaction between Avr and R exist, emerging evidence suggests that the recognition could be indirectly mediated by other plant host proteins. In this 'guard hypothesis', R proteins may 'guard' or monitor the status of the host plant proteins that are targets of pathogen Avr effector proteins (Martin et al., 2003;Chisholm et al., 2006;Jones and Dangl, 2006).
Genetic studies have identified genes required for R gene signaling (Dangl and Jones, 2001;Glazebrook, 2001). EDS1 (ENHANCED DISEASE SUSCEPTIBILITY 1) and PAD4 (PHYTOALEXIN DEFICIENT 4) are required for the function of TIR-NB-LRR proteins while NDR1 (NONRACE-SPECIFIC DISEASE RESISTANCE 1) is normally required for the CC-NB-LRR proteins although there are exceptions (Wiermer et al., 2005). RAR1 (REQUIRED FOR
Genetic studies have also identified genes for cell death control. A number of mutants classified as lesion mimics induce spontaneous cell death which may result from defects in developmental PCD, HR control or from necrosis and chlorosis (Shirasu and Schulze-Lefert, 2000). Some of the lesion mimic mutants have mis-regulation of the initiation of cell death and form small, localized, necrotic spots. More than 30 such mutants have been isolated including some of those in acd (accelerated cell death), cpr (constitutive expressor of PR genes), lsd (lesion simulating disease) and ssi (suppressor of SA insensitivity) (Lorrain et al., 2003) in Arabidopsis and mlo (mutation-induced recessive alleles) in barley (Buschges et al., 1997). About half a dozen mutants, including some lsd and acd, are unable to individual gene modulates cell death is essential to deciphering cell death control and defense pathways.
The Arabidopsis BAP1 gene is involved in defense and cell death regulation. It encodes a membrane associated protein containing a C2 domain and has a calcium-dependent phospholipid binding activity (Hua et al., 2001;Yang et al., 2006). Biochemical and genetic data indicate that BAP1 is a functional partner of BON1, an evolutionarily conserved copine protein with two C2 domains at its amino-terminus (Hua et al., 2001;Yang et al., 2006). BAP1 and BON1 are negative regulators of defense responses. Similar to but less so than the bon1 mutants (Hua et al., 2001;Jambunathan et al., 2001), the bap1 loss-of-function mutants have an enhanced disease resistance to virulent pathogens and consequently dwarfed statures (Yang et al., 2006). The defense phenotype is mediated by SNC1/BAL, a TIR-NB-LRR type of gene in the RPP5 cluster (Yang and Hua, 2004;Yang et al., 2006). Though a cognate avr gene has not been identified, SNC1 is likely an R gene as its active mutants induce constitutive defense responses (Stokes et al., 2002;Zhang et al., 2003). The bap1 and bon1 phenotypes are reversed by loss-of-function mutations in SNC1, EDS1 and PAD4 as well as by nahG encoding a SA degrading enzyme (Yang and Hua, 2004;Yang et al., 2006) transgenic lines in bap1-1 are essentially wild-type in appearance ( Figure 1C), indicating that the BAP2 protein has a similar biochemical activity to BAP1.
Previous studies demonstrated that the BAP1 protein interacts with the BON1 protein in vitro and that they likely act as partners in vivo (Hua et al., 2001;Yang et al., 2006). We asked whether BAP2 can interact with BON1 as well by using the yeast two-hybrid system (Fields and Song, 1989). BAP1 and BAP2 were each fused to the DNA-binding domain of the GAL4 transcription factor to generate GBD:BAP1 and GBD:BAP2 fusion proteins respectively while the A domain of BON1 was fused with the activation domain of GAL4 to generate GAD:BON1A. Co-expression of GBD:BAP2 with GAD:BON1A conferred growth to the yeast host strain on medium selecting for protein-protein interactions, similarly to that of GBD:BAP1 and GAD:BON1A ( Figure 1D), indicating a direct interaction between the BON1 and BAP2 proteins.
Because BON2 and BON3 have overlapping functions with BON1 (Yang et al., 2006), we further determined whether BAP1 and BAP2 can interact with BON2 or BON3 in the yeast two-hybrid system. Co-expression of GBD:BAP1 or GBD:BAP2 with GAD:BON2A and GAD:BON3A respectively conferred yeast growth on the selection medium ( Figure 1D). It thus appears that each member of the BON family can interact with each member of the BAP family.
Assessed by yeast growth, the strength of interaction differs among these protein pairs, with the weakest interaction found between BAP2 and BON2 and the strongest one found between BON1 and BAP1. These differences are yet to be validated with the analysis of expression and stability of these proteins in yeasts.

The loss of BAP1 and BAP2 function confers seedling lethality
To elucidate the function of BAP2, we isolated a T-DNA insertion mutant of BAP2 (SALK_052789) from the SALK collection. The T-DNA was inserted in the nucleotide sequence corresponding to Gln 67 of the encoded BAP2 protein ( Figure 1A), and no BAP2 transcript was observed by RNA blot analysis (data not shown). This loss-of-function mutant, named as bap2-4 (referred as bap2 from now on), did not exhibit any obvious growth defects, in contrast to the bap1 mutant ( Figure 1E). However, an accelerated HR was observed in bap2 compared to Col for Pst DC3000 expressing AvrRpt2. Col wild type and bap2 were inoculated with a high concentration of Pst DC3000 carrying avrRpt2. At 8 hours post inoculation (hpi), none of the Col leaves showed HR, while 50% of the bap2 leaves already had HR at this time ( Figure 1F and G). At 12 hours 90% of the bap2 leaves exhibited HR while only 10% of the wild type leaves showed HR ( Figure 1G).
To reveal possible overlapping functions between BAP1 and BAP2, we attempted to generate double mutants between bap2 and bap1. However, plants with the bap1bap2 genotype could not be identified from the F2 progenies of a cross between bap1 and bap2, suggesting that the homozygous mutant is either embryonic or seedling lethal. We subsequently sowed the progenies of double mutants (one heterozygous and the other homozygous) on agar plates, and found 39 out of 164 bap1bap2/+ and 52 out of 194 from bap1/+bap2 seeds germinated but soon died at the cotyledon stage ( Figure 1H). Again, no surviving seedlings were bap1bap2, confirming that the double mutant is seedling lethal.
We observed dominant interactions between the bap1 and bap2 mutants. bap1 is a recessive mutant with a mild growth defect (Yang et al., 2006) and bap2 has no obvious growth defect. However, heterozygous mutants of bap1 and bap2 each enhanced the phenotypes of the homozygous mutants of the other ( Figure 1I). The bap1/+bap2 mutant had small and slightly curly leaves in contrast to the wild-type looking bap2 mutant. After bolting, its primary shoot frequently bended at the tip and died afterward. Multiple lateral shoots usually generated subsequently, giving a bushy phenotype. The bap1bap2/+ mutant exhibited a stronger phenotype than the bap1 single mutant. Its leaves are very curly with water soaked appearance, resembling those of bon1. The genetic interactions between BAP1 and BAP2 indicate that these two genes have overlapping functions and that their functions are dosage dependent.

Cell death occurs in mutant combinations between bap1 and bap2
We assessed cell death in different mutant combinations between bap1 and bap2 as their double homozygous mutant is seedling lethal. Trypan blue, a membrane impermeable reagent, was used to stain dead cells or cells with damaged cell membranes. This vital stain revealed various degrees of cell death in leaves of different mutants ( Figure 2A). None of the wild-type Col leaves (0/8) analyzed had any staining, neither did the bap1 (0/8) or the bap2 (0/8) single mutants. Strong staining was found in most of the leaves of bon1-1 (9/14), consistent with previous findings (Jambunathan et al., 2001;Yang et al., 2006). Very few leaves of bap1/+bap2 (1/8) were stained by trypan blue, while most of the bap1bap2/+ leaves (7/12) were stained.
Thus, extensive cell death occurs in leaves of bap1bap2/+ as in bon1, correlating with a severe morphological defect in leaves.
We further analyzed leaves of these mutants for autofluorescence indicative of accumulation of phenolic compounds from dead cells. No significant autofluorescence was observed in Col, bon1, bap1, bap2, or bap1/+bap2 ( Figure 2B). In contrast, strong autofluorescence was found in bap1bap2/+ ( Figure 2B), indicating extensive cell death in bap1bap2/+.
We then asked whether the cell death phenotype in bap1 and bap2 mutant combinations was associated with an accumulation of ROS. To this end, we determined the relative amount of and bap1bap2/+ both had a stronger staining than bap1 ( Figure 2C). Thus, H 2 O 2 accumulates at a moderate level in bap1 and at a higher level in the bap1and bap2 mutant combinations.

Modulation of the bap1bap2 double mutant phenotypes by eds1, pad4 and the environment
The lethal phenotype of bap1bap2 could result from a heightened defense response leading to extensive cell death at very early stage of development. We assessed whether the lethal phenotype of bap1bap2 is due to a stronger activation of SNC1 and higher accumulation of SA in the double mutant than in the bap1 single mutant, given that the loss-of-function mutant snc1-11 (referred as snc1 from now on) and the SA degrading nahG suppressed the phenotype of bap1. Analysis of progenies of a bap1bap2/+snc1/+ plant and those of a bap1bap2/+nahG/+ plant indicate that neither snc1 nor nahG could rescue the lethal phenotype of bap1bap2 (data not shown).
Strikingly, the lethality of bap1bap2 can be suppressed by mutations in PAD4 or EDS1.
From the F2 progenies of a cross between bap2 and bap1pad4 (Yang et al., 2006), we were able to obtain bap1bap2 plants and these plants were always pad4 homozygous, indicating that pad4 suppressed the lethal phenotype of bap1bap2. Not only was the triple mutant bap1bap2pad4 viable, it was also wild-type in appearance throughout its development ( Figure 2D). Similar rescue of lethality of bap1bap2 was observed with the eds1 mutation as well ( Figure 2D).
pad4 and eds1 suppressed all other mutant phenotypes observed in the bap1 and bap2 mutant combinations. No autofluorescence could be seen on leaves of bap1bap2pad4 or bap1bap2eds1, in contrast to the strong fluorescence on the bap1bap2/+ leaves ( Figure 2B). Nor was a higher level of DAB staining observed in bap1bap2pad4, indicating a suppression of H 2 O 2 accumulation in bap1bap2 by pad4 (Fig 2C).
We determined whether environmental factors can modulate the phenotypes of the bap1 and bap2 mutant combinations. A higher temperature of 28ºC alleviates the growth defects observed in all double mutants to different degrees. Both bap1bap2/+ and bap1/+bap2 were wild-type looking throughout the life cycle at 28ºC in contrast to the dwarf phenotype at 22ºC ( Figure 2E). The bap1bap2 homozygous mutant was partially rescued by a higher growth temperature. Instead of dying immediately after germination at 22ºC, the double mutant grew like the wild type at 28ºC for two weeks after germination. However, when the wild type started bolting at approximately three-week-old, the double mutant turned yellow and died ( Figure 2E).
A shorter photoperiod suppressed phenotypes of some of the mutant combinations as well. bap1/+bap2 and bap1bap2/+ grown under a cycle of 12 hr light and 12 hr darkness rather than constant light were wild-type looking (data not shown). However, no bap1bap2 could be found from progenies of bap1/+bap2 or bap1bap2/+ under this growth condition, indicating that the shorter photoperiod does not suppress the seedling lethality of bap1bap2.

The bap2 cell death phenotype is not associated with defense responses
Because bap1 has heightened disease resistance to virulent Pseudomonas syringae and Hyaloperonospora parasitica (Yang et al., 2006), we assessed whether bap2 has an abnormal defense response. We challenged the bap2 mutant with a virulent bacterial pathogen P. syringae pv tomato (Pst) DC3000 and found that it was as susceptible to this pathogen as the wild-type Col ( Figure 3A). Four days after infection, Pst grew to 4.2×10 5 colony forming unit (cfu) mg -1 fresh weight in bap2, similarly to the level of 3×10 5 in the wild type, while its growth was reduced to 1.1×10 4 in bap1. bap2 was also as susceptible to virulent H. parasitica as the wild type. While no sporangiphores were found on bap1 a week after spray inoculation, bap2 supported the same amount of growth of this pathogen as the wild-type Col ( Figure 3B).
Given that bap1 and bap2 enhanced each other's morphological and cell death phenotype in a dominant manner, we asked whether the same is true for the disease resistance phenotype.
Growth of Pst DC3000 was analyzed in the bap1/+bap2 and bap1bap2/+ mutants. Pst propagated to 1.8×10 4 cfu mg -1 fresh weight in bap1bap2/+, comparable to the level of 1.1x10 4 in bap1 ( Figure 3A), indicating that bap2 does not dominantly enhance disease resistance in bap1. Pst grew to 1.3×10 5 in bap1/+bap2, similar though slightly lower than the level of 4.2×10 5 in bap2 ( Figure 3A). No significant difference was observed in biological replica between bap1/+bap2 and bap2. Thus, bap1 and bap2 do not dominantly enhance each other's disease resistance phenotype in contrast to the growth and cell death phenotype. In addition, bap1bap2pad4 was as susceptible to Pst as pad4 and bap1pad4 (Fig 3A), indicating that the resistance phenotype is mediated by PAD4.

Overexpression of BAP and BON genes inhibits PCD induced by a variety of biotic and abiotic stimuli in plants
The loss-of-function phenotypes indicate that the BAP genes are negative regulators of cell death. To determine whether they have a direct role in suppressing cell death, we analyzed their overexpression effect on PCD. First we assayed HR induced by Pst DC3000 carrying avirulent effectors in Arabidopsis. Wild-type Col plants were infiltrated with Pst DC3000 observed until 8-9 hpi with simultaneous agroinfiltration of BAP1 and BON1 ( Figure 4A). The suppression for both avirulent strains was consistently seen in replicated experiments. Therefore, overexpression of BON1 and BAP1 together in Arabidopsis greatly delayed HR induced by avirulent bacterial pathogen Pst DC3000 with two different effector proteins.
We subsequently analyzed the effect of over-expression of BAP1 and BON1 on PCD induced by other R proteins. Transient co-expression of a potato NB-LRR type of R protein Rx and its elicitor PVX coat protein (CP) was shown to induce HR in Nicotiana benthamiana (Bendahmane et al., 1999). A collapse of cells indicative of HR was observed in leaf area agroinfiltrated with Rx and CP at 36 hpi. Co-agroinfiltration with the vector alone did not alter the onset or the progression of HR. However when p35S::BAP1 or p35S::BON1 were coagroinfiltrated, HR was either suppressed or greatly reduced at 36 hpi ( Figure 4B). In some repeats, no HR was ever developed over the following five days' observation. Co-agroinfiltration of p35S::BAP1 and p35S::BON1 together did not appear to have a stronger effect in HR suppression.
Given that BAP1 and BON1 inhibit HR induced by R proteins, we further tested whether the BAP1 and BON1 Figure   4C). Strikingly, when p35S::BAP1 and p35S::BON1 were simultaneously agroinfiltrated with pDEX::Bax, no obvious cell death was observed at 72 hpi when the control areas exhibited strong cell death ( Figure 4C). Similar suppression of cell death was observed when p35S::BAP2 and p35S::BON1 were co-agroinfiltrated. In both cases, cell collapse started at 90 hpi and occurred to a full extend at 114 hpi in BON1 and BAP1/BAP2 co-infiltrated areas. Therefore, Bax-induced cell death was delayed by one to two days with over-expression of BON1 together with BAP1 or BAP2.

BAP1 and BAP2 inhibit cell death induced by ROS in Arabidopsis and yeast
The BAP transcripts are induced by a number of biotic and abiotic stimuli and the common feature of these treatments is probably ROS. Considering that the BAP genes are capable of inhibiting PCD, we asked whether overexpression of the BAP genes can inhibit cell death induced by ROS. To this end, we compared Col Arabidopsis lines containing the   (Yang et al., 2006). Therefore, multiple protein complexes might form between BON and BAP proteins to provide robustness and/or specificities to the system.

Regulation of cell death and defense by the BAP and BON proteins
In this study, we identified a more direct role of the BAP family in the control of genes as members of the copine gene family found not only in plants but also in animals. It is unclear whether or not the BAP genes are evolutionarily conserved because the most significant signature of their encoded proteins is the C2 domain which is widely present in many signaling molecules. The striking feature of BAP1 is its extreme responsiveness to numerous biotic and abiotic stimuli ranging from singlet oxygen species, temperature variation, wounding from bacterial infection, to even butterfly egg oviposition (op den Camp et al., 2003;Little et al., 2006;Yang et al., 2006). BAP2 and BON1 respond to at least some of these stimuli but apparently to a lesser degree. The responsiveness to diverse stimuli suggests that the BAP and BON genes may serve as signaling molecules or maintain calcium or lipid homeostasis in stress responses, and the loss of these activities results in cell death. The suppression of the bap and bon phenotypes by eds1 or pad4 indicates that BAP and BON genes regulate a cell death pathway mediated by EDS1 and PAD4. Emerging evidence has strongly implicated EDS1 and PAD4 in transducing redox signals (Mateo et al., 2004;Ochsenbein et al., 2006). It is tempting to speculate that the BAP and BON genes are responsive to ROS and/or calcium signals and modulate ROS signaling in stress responses.
The BAP and BON molecules might become targets of pathogen effector proteins because of their ancestral role in cell death control during the evolution of plant innate immune system (Jones and Dangl, 2006). Indeed, the bon1 and bap1 mutants have heightened defense responses that are at least partially mediated by a TIR-NB-LRR type of R gene SNC1. It is possible that the loss of the BON1 or BAP1 proteins is 'interpreted' by plants as the result of the invasion of a pathogen and thus triggers the activation of R proteins to mount defense responses.
Multiple R genes in addition to SNC1 are likely regulated by the BON family and the BAP family, as the bon1bon2, bon1bon3 and bap1bp2 double mutants have stronger phenotypes independent of SNC1 than the bon1 or bap1 single mutants. In addition, the bon or bap mutant combinations exhibit phenotypic variations in different accession backgrounds ((Yang et al., 2006), and unpublished results), suggesting the involvement of multiple accession-specific R genes.
It has yet to be determined whether the regulation of BON and BAP proteins on R proteins is similar to that of RIN4 on RPM1 and RPS2. Current data do not distinguish models of regulation at the protein level or the RNA transcript level. Future studies on the general PCD sample were resolved on 1.2% agarose gels containing 1.8% formaldehyde. Ethidium bromide was used to visualize the rRNA bands to ensure equal loading. RNA gel blots were hybridized with gene-specific, 32 P labeled, single-stranded DNA probes.

Pathogen resistance assay
Bacterial growth in Arabidopsis was monitored as described with some modifications (Tornero and Dangl, 2001). Pseudomonas syringae pv tomato DC3000 was grown overnight on the KB medium and resuspended at 10 8 cfu ml -1 in a solution of 10 mM MgCl 2 and 0.02% Silwet L-77.
Two-week-old seedlings were dip inoculated with bacteria and kept covered for one hour. The amount of bacteria in plants was analyzed at one hour after dipping (day 0) and 4 days after dipping (day 4). The aerial parts of three inoculated seedlings were pooled for each sample and three samples were collected for each genotype at one time point. Seedlings were ground in 1 ml of 10 mM of MgCl 2 and serial dilutions of the ground tissue were used to determine the number of cfus per gram of leaf tissues.
For HR test, Pst DC3000 with avirulent genes were resuspended at 10 8 cfu ml -1 and infiltrated into leaves of 4-week-old Arabidopsis plants. Infiltrated leaves were monitored hourly for symptoms of cell collapse.