An activator for pyruvoyl-dependent l-aspartate α-decarboxylase is conserved in a small group of the γ-proteobacteria including Escherichia coli

In bacteria, β-alanine is formed via the action of l-aspartate α-decarboxylase (PanD) which is one of the small class of pyruvoyl-dependent enzymes. The pyruvoyl cofactor in these enzymes is formed via the intramolecular rearrangement of a serine residue in the peptide backbone leading to chain cleavage and formation of the covalently-bound cofactor from the serine residue. This reaction was previously thought to be uncatalysed. Here we show that in Escherichia coli, PanD is activated by the putative acetyltransferase YhhK, subsequently termed PanZ. Activation of PanD both in vivo and in vitro is PanZ-dependent. PanZ binds to PanD, and we demonstrate that a PanZ(N45A) site-directed mutant is unable to enhance cleavage of the proenzyme PanD despite retaining affinity for PanD. This suggests that the putative acetyltransferases domain of PanZ may be responsible for activation to enhance the processing of PanD. Although panD is conserved among most bacteria, the panZ gene is conserved only in E. coli-related enterobacterial species including Shigella, Salmonella, Klebsiella and Yersinia. These bacteria are found predominantly in the gut flora where pantothenate is abundant and regulation of PanD by PanZ allows these organisms to closely regulate production of β-alanine and hence pantothenate in response to metabolic demand.


Introduction
Pantothenate (vitamin B5) is required in all organisms for the synthesis of coenzyme A (CoA) and acyl carrier proteins, however its biosynthesis is limited to prokaryotes, fungi and plants in which it is produced from D-pantoate and b-alanine (Webb et al. 2004;Fig. 1A). In bacteria, b-alanine is synthesized via decarboxylation of L-aspartate, a reaction catalyzed by the pyruvoyl-dependent enzyme L-aspartate a-decarboxylase (ADC). ADC-dependent synthesis of b-alanine is the sole source of b-alanine and deficiency in ADC b-alanine therefore leads to b-alanine auxotrophy which can be relieved with either b-alanine or pantothenate. In Escherichia coli, supplementation of culture media with pantothenate or b-alanine markedly increases the cellular level of CoA, whereas supplementation with excess pantoate is much less effective (Cronan et al. 1982). This demonstrates that b-alanine supply is limiting for pantothenate biosynthesis in this organism. Regulation of ADC activity is therefore central to the regulation of intracellular pantothenate and CoA concentrations.
The enzyme is a member of the small class of pyruvoyldependent enzymes, which contain a covalently-bound pyruvoyl cofactor (Williamson and Brown 1979). This enzyme class includes human S-adenosyl methionine decarboxylase, human phosphatidylserine decarboxylase and histidine decarboxylase from Lactobacillus 30a (van Poelje and Snell 1990). For all of these enzymes, the protein is initially translated as an inactive proenzyme (designated the p-protein) subsequently is activated via posttranslational modification. A serine residue in the peptide backbone rearranges to generate an ester intermediate (Albert et al. 1998) which is subsequently cleaved to generate the a-subunit and b-subunit observed in the active enzyme. Cleavage of the peptide chain generates a dehydroalanyl residue at the N-terminus of the a-subunit which hydrolyses to generate the pyruvoyl group (van Poelje and Snell 1990;Schmitzberger et al. 2003). In all cases, cleavage of the p-chain is autocatalytic; the p-chain of purified E. coli PanD is activated to generate the a-subunit and the b-subunit of ADC by auto-activation in vitro (Ramjee et al. 1997). Auto-activation of purified E. coli PanD is however slow; although activation occurs to some extent after incubation at 37°C for 24 h, stoichiometric activation requires higher temperatures or longer incubation times. Considering the doubling time of E. coli cells, which is about 20 min at 37°C under optimal conditions, this rate for activation of PanD to generate ADC is relatively slow compared with in vivo. Naturally, it was therefore proposed that there might be a specific catalyst for proenzyme processing in vivo (Ramjee et al. 1997). Activation of pyruvoyl-dependent S-adenosylmethionine decarboxylase is stimulated by addition of putrescine and spermidine, these molecules react with the product of S-AdometDC, and thereby can regulate its production. More recently it has been found that cleavage of histidine decarboxylase, is activated by a protein activator, HdcB, which acts enzymatically in maturation (Trip et al. 2011). However, no such activator has been identified for ADC. Reinvestigation of the genetic analysis of mutants requiring pantothenate suggested the existence of an activator in Salmonella typhimurium and E. coli. Mutations generating b-alanine auxotrophy were identified in the 1970s and 1980s. These mutations were mapped to two different chromosomal positions at 5 min and 89 min in Salmonella typhimurium (Ortega et al. 1975;Primerano and Burns 1983). The gene encoding ADC, PanD, was mapped at 5 min in S. typhimurium and genetically linked to panB and panC, which encode the two other essential enzymes for pantothenate biosynthesis (Primerano and Burns 1983). In E. coli, PanD was mapped at 3 min and was closely linked to panC (Cronan 1980) corresponding, therefore, to the mutation in S. typhimurium at 5 min. However, the second mutation at 89 min was forgotten for decades without further study, during which time it had been thought that the metabolic pathway of pantothenate was completely elucidated in bacteria (Webb et al. 2004). Here we report that this "forgotten mutation" is required for activation of ADC.

Results
Requirement of b-alanine for pantothenate synthesis in the yhhK gene deletion mutant A gene-knockout library of E. coli strains revealed a number of gene knock-out mutants which, while essential for growth of E. coli in minimal synthetic media Joyce et al. 2006), are dispensable in nutrient-rich media. This media-dependent growth inhibition suggests that these knock-out mutations cause nutritional deficiencies. However, in most cases, the biological functions of these genes remain unknown. One such gene knock-out mutant is ΔyhhK. Supplementation of growth media with pantothenate restored the growth of ΔyhhK cells even in synthetic minimal media (Adams et al. 1990); it seemed likely therefore that ΔyhhK cells are deficient in the biosynthetic pathway for pantothenate. We examined which intermediate during pantothenate synthesis was required for normal growth (Fig. 1B). Pantothenate is generated via the condensation of b-alanine and pantoate. Supplementation of the growth medium with b-alanine, but not with pantoate, restored the growth of ΔyhhK cells, whereas supplementation with L-aspartate, the precursor of b-alanine, did not. Complementation of ΔyhhK cells, using a plasmid encoding a his-tagged YhhK, efficiently restored the normal growth of ΔyhhK cells. We therefore concluded that biosynthesis of b-alanine is prevented by the knock-out mutation of yhhK. In bacteria, b-alanine is formed by decarboxylation of L-aspartate. This enzymatic reaction is catalyzed by L-aspartate-a-decarboxylase (ADC), encoded by PanD (Webb et al. 2004). In E. coli, the locus of yhhK does not correspond to the locus of PanD. In S. typhimurium, two loci for b-alanine auxotrophy have been identified, one corresponding to PanD and the other located at 89 min. The genetic configuration of E. coli and S. typhimurium chromosomes resemble each other, and so we theorized that the map position of the latter mutant in S. typhimurium might correspond to that of yhhK in E. coli, which was indeed the case. We will therefore henceforth refer to yhhK as PanZ.
Activation of cleavage of PanD by PanZ (YhhK) for the maturation of ADC How does the PanZ gene product influence the decarboxylation of L-aspartate to generate b-alanine? We hypothesized that the conversion of PanD to an active enzyme could be accelerated by PanZ. To assess whether PanZ is an activator for the maturation of ADC, we examined the cleavage of PanD in ΔPanZ cells. To detect cleavage of PanD in vivo, FLAG peptides were added to the C-terminus of PanD (PanD-FLAG). The growth rate of the cells with PanD-FLAG was slower on minimal synthetic medium as compared to wild type cells, but they did not require b-alanine as a supplementary nutrient (Fig. 1C) demonstrating that PanD-FLAG is a functional enzyme. In PanZ + cells, two bands at 18 kDa and 15 kDa corresponding to the unactivated PanD gene product and the a-chain of the activated ADC were detected by Western blotting 300 analysis ( Figure 1D). The 15 kDa band was however not observed in extracts of ΔPanZ cells. Complementation of the ΔPanZ cells with his-tagged PanZ (his-PanZ) led to detection of the 15 kDa band. High-level induction of His-PanZ using 100 lmol/L of isopropyl b-D-1-thiogalactopyranoside (IPTG) led to near disappearance of the 18 kDa band corresponding to the inactive PanD and the presence of only the 15 kDa band corresponding to the activated protein. These results indicate that PanZ stimulates cleavage of the PanD zymogen in cells and that it is essential for this reaction. We next determined whether PanZ was sufficient for the activation of PanD to form ADC by in vitro analysis of cleavage of purified PanD-FLAG by His-PanZ (Fig. 1E). After incubation for 3 h at 37°C, PanD-FLAG on its own is unactivated (Fig. 1E). However, addition of His-panz leads to complete activation of PanD-FLAG. On the timescale shown (3 h) however, addition of 16 equivalents of His-PanZ is insufficient to fully activate PanD-FLAG. Thus, while PanD is cleaved in a PanZ-dependent manner in vivo and in vitro, it appears that this process is not catalytic.

Involvement of the GNAT domain in PanD cleavage
Bioinformatics analysis showed that PanZ is a putative N-acetyltransferase and a member of the GNAT (Gcn5related N-acetyltransferase) superfamily of acetyltransferases (Fig 2 A). Indeed, the structure of PanZ (YhhK) bound to CoA has been solved using nuclear magnetic resonance (NMR, PDB ID: 2k5t). We used ITC to confirm the interaction between PanZ and CoA, and also acetyl CoA (data not shown). Several residues in the acetyltransferase domain are well conserved in the GNAT superfamily between bacteria and humans. Residue Asn45 is one such residue, and it is completely conserved in those bacterial species which have the PanZ gene. We constructed a single mutation in PanZ at one of these positions to generate a N45A site-directed mutant, this mutant PanZ(N45A) could not compensate for the requirement for b-alanine in ΔPanZ cells ( Fig 2B).
To confirm that this was due to a deficiency in the catalytic activity of PanZ(N45A), we tested cleavage of PanD using purified proteins. Native PanD was purified from cell lysates that overproduced PanD ( Fig 2C). In contrast, PanZ and PanZ(N45A) proteins were purified using histags and affinity purification ( Fig 2D). After incubation of a mixture including purified PanD and PanZ-His, cleaved fragments of PanD were detected by SDS gel electrophoresis (Fig. 2E). This cleavage reaction was independent of CoA or acetyl-CoA, and no change in the cleavage reaction could be detected by the addition of either (data not shown). However, the cleavage reaction could not be detected when PanZ(N45A)-His was used in place of PanZ-His; the single amino acid substitution in the GNAT domain prevents PanZ from activating PanD. Hence, the GNAT domain is responsible for maturation of PanD to ADC.
yhhK is conserved only among c-proteobacteria Orthologous genes encoding ADC are conserved among almost all bacterial species, whereas homologous genes of PanZ are found only in some members of the c-proteobacteria including E. coli (Fig. 3A).
The observation of PanD in bacterial species lacking PanZ homologues suggests that other activating factors might be involved in the activation of PanD or that, in these organisms, auto-activation of PanD alone might produce active ADC. We tested whether PanD derived from Bacillus subtilis, which lacks a PanZ homologue, could suppress the b-alanine auxotrophy of E. coli ΔPanZ cells. The cloned PanD gene of B. subtilis (PanD BS ) was substituted for PanD on the E. coli chromosome. E. coli PanZ + cells in which PanD is replaced with PanD BS grew as well as wild type E. coli in M9 glucose medium (Fig. 3B). The same substitution in E. coli ΔPanZ generates cells that can also grow without pantothenate supplementation. These results suggest that activation of PanD BS does not require PanZ. We confirmed the cleavage of PanD BS in E. coli. A C-terminal FLAG tag was added to the chromosomal PanD BS and cell extracts were monitored by western blotting. Two bands at~17 kDa (corresponding to the PanD BS proenzyme) and~14 kDa (the activated a-subunit) were observed in both panZ + and ΔPanZ cells (Fig. 3C). Although the proenzyme form of PanD BS -FLAG was observed in the cell extract, almost all of the proenzyme form was completely changed to the cleaved form during purification (data not shown) preventing study of the activation of PanD BS in vitro.

Physical interaction of PanD with PanZ
Acceleration of PanD activation by PanZ suggests that non-covalent interaction of the proteins may be required. We examined the protein-protein interaction between PanZ and PanD in vivo using a bacterial two-hybrid assay (Karimova et al. 1998). Each of PanZ and PanD were fused to fragments of the catalytic domain of adenylate cyclase (T25 or T180). Association between PanZ and PanD leads to association of the fragments of adenylate cyclase, further leading to cAMP production which subsequently induces expression of lacZ. Homotetramerization of PanD was used as a positive control (Ramjee et al. 1997;Albert et al. 1998); cells that co-expressed T25-PanD and T18-PanD fusions appeared dark blue when grown on X-gal media (Fig. 4). This intense color reflects the strong interactions within the homotetramer -boiling in the presence of SDS is required to fully dissociate the PanD tetramer (Ramjee et al. 1997). Co-expression of either the T25-PanD and the T18-PanZ fusion, or the T18-PanD and the T25-PanZ fusion led to generation of a pale blue color (Fig. 4A) indicating an interaction between PanZ and PanD (though weaker than the very strong PanD tetramerization). Finally, co-expression of T25-PanZ and T18-PanZ leads to a very pale blue color suggesting multimerisation of PanZ is occurring, although however this could be a PanD tetramer mediated interaction.
Physical interaction between PanZ and PanD in vitro was then confirmed by gel filtration with purified PanZ and PanD. Although PanZ-His and PanD have similar molecular weights, 15.6 kDa and 13.8 kDa respectively, retention time of the purified protein, PanZ-His and PanD are distinctive ( Fig. 5A and B). Rather, the retention time of PanD corresponds to a tetramer. After co-incubation of PanZ-His and PanD, both proteins co-eluted together with an earlier retention time (Fig. 5D). Analysis by gel electrophoresis indicated that PanD had been cleaved and the PanZ-His was co-purified with the cleaved PanD. This suggests that PanZ tightly binds to the PanD tetramer even after maturation of the inactive proenzyme. In order to confirm the affinity of PanZ(N45A)-His for the unactivated proenzyme form of PanD, the purified PanZ(N45A)-His was also eluted using gel filtration chromatography ( Fig. 5C and E). PanZ (N45A)-His co-purified with PanD, suggesting that the mutant PanZ still retains the ability to bind to PanD. However, the processing of the unactivated proenzyme form of PanD was completely inhibited. Thus, the mutated residue is critical for cleavage of the proenzyme 302 rather than perturbing its affinity for the substrate. Addition of CoA or acetyl-CoA did not affect the elution of PanD, PanZ and the mixture as before (data not shown).

Discussion
The putative acetyltransferase PanZ is crucial for the activation of the proenzyme PanD in vivo. This function of PanZ appears to not be catalytic, but rather to provide a cofactor to stimulate self-activation of PanD. Although PanZ can bind to CoA and acetyl-CoA, we could not detect any influence of the addition of CoA and acetyl-CoA to the in vitro reaction. PanZ with the single mutation of N45A demonstrates that the well conserved feature of this domain between the GNAT family domain, which can bind CoA, is involved in enhancing the selfprocessing of PanD. This suggests that binding of a CoA derivative to PanZ might regulatactivation of the processing of proenzyme PanD in vivo. It is possible that a CoA derivative has been co-purified with PanZ and is therefore present in our experiments to investigate the interaction between PanZ and PanD. More precise biochemical measurements are therefore needed to observe the influence of CoA and acetyl-CoA on the in vitro reaction.
Is it possible and physiologically significant that somehow CoA or acetyl CoA binding regulate the activation of PanD in the generation of ADC? As CoA is derived from pantothenate, we would therefore expect CoA to negatively regulate the synthesis of pantothenate. Such a regulatory pathway would provide a second negative feedback in the pantothenate synthetic pathway. In most organisms except archaea, pantothenate kinase, CoaA, phosphorylates pantothenate to produce phosphopantothenate (Brown 1959) (Fig. 6A). CoaA activity is regulated by CoA in a negative feedback manner (Vallari and Rock 1987;Vallari and Jackowski 1988). Inhibition of CoaA leads to accumulation of pantothenate and the excess pantothenate is secreted from the cells via the sodium-dependent transporter PanF (Jackowski and Rock 1981;Vallari and Rock 1985). This

304
means that ΔPanZ cells can grow by using pantothenate secreted by wild type cells (Fig. 6B). Indeed, slight growth of ΔPanZ cells was often seen in proximity to growing PanZ + cells. When the ΔPanZ cells were inoculated near pantothenate transport-deficient cells (ΔpanF) however, no growth was observed (Fig. 6B). Thus, secreted pantothenate can be used by those bacteria that cannot produce or otherwise do not produce pantothenate.
In the gut flora, vast numbers of microorganisms live and produce pantothenate. However feedback control of CoA on CoaA activity does not effectively regulate this pathway. The production of pantoate from a-ketoisovaleric acid via ketopantoate is essentially reversible. ATPdependent synthesis of pantothenate from pantoate and b-alanine therefore represents the first committed step in the biosynthetic pathway. In E. coli, the flux through this step is limited by the supply of b-alanine, formed by decarboxylation of L-aspartate. Control of b-alanine supply via control of ADC activation would therefore prevent depletion of L-aspartate and intermediates in branched chain amino acid biosynthesis.
As PanZ is a putative acetyltransferase, we therefore hypothesize that acetylation of PanD may regulate PanD activation in response to the ratio of cellular acetyl-CoA and CoA. Both small molecules bind tightly to PanZ. Therefore, if acetylation inhibits formation of ADC then acetylation will only occur when acetyl-CoA accumulates in the cell. Such accumulation could be a signal to repress pantothenate biosynthesis, despite the presence of numerous potential acetylation sites in PanD, we have not been able to obtain evidence of acetylation suggesting that a more complex mechanism is required for regulation of PanD activation in vivo.
In contrast to PanD from E. coli, PanD BS does not require PanZ for its activation. Overexpression of PanD from Corynebacterium glutamicum in E. coli has previously been reported as a high yielding strategy for production of b-alanine and hence pantothenate (Dusch et al. 1999). This is, perhaps, consistent with our findings: an analogue of the PanZ gene is not present in C. glutamicum and so, like PanD BS , PanD from C. glutamicum might be fully activated by auto-processing. Previous attempts to engineer the pathway in plants by Fouad et al. (Fouad and Rathinasabapathi 2006)and Chakauya et al. (Chakauya et al. 2008) have used PanD EC . As PanD EC requires PanZ for its activation, PanD BS would therefore be more effective in these applications. It remains unclear which structural differences between these orthologs lead to rapid activation of PanD BS without the addition of PanZ. Alignment of orthologs of PanD (Fig. S1) reveals that the N-terminal (1-111) portions of the proteins are highly conserved among all bacterial species. However, the last 20 aa are poorly conserved among most bacterial species, but are well conserved in those bacterial species in which orthologs of PanZ are found. This region is distant from the site of cleavage (Gly24-Ser25) so it is not clear how this region could prevent activation of PanD in these species. Recently, it has been reported that this portion of the protein can bind intermolecularly to the active site of adjacent proteins as a result of crystal packing (Webb et al. 2012); it is therefore possible that such intramolecular interactions might act to inhibit activation in solution. We are currently investigating the role of this and other semi-conserved residues in the activation of PanD.
In summary, we have identified an essential new component in the pantothenate biosynthetic pathway. The pathway of pantothenate biosynthesis has been previously proposed as a target for antimicrobial chemotherapy (Spry et al. 2008). But as the pathway is common to all bacteria, the opportunity for selective chemotherapy is limited. PanZ homologues are however only found in the genomes of E. coli and related bacteria, such as Shigella, Salmonella, Klebsiella and Yersinia, all of which are resident in the animal gut. PanZ therefore provides a potential target by which to inhibit only pantothenate biosynthesis in these pathogenic bacteria, while sparing the gut flora where pantothenate is abundant.

Experimental Procedures
Bacterial strains and plasmids Table 1 shows bacterial strains and plasmids used in this report. To construct the PanD-flag gene of SN208, we used a DNA fragment encoding the FLAG octapeptide repeated three times in tandem (39 flag) with the cat gene inserted between frt sites (frt-cat-frt). The frt sites are recognition sequences of the sequence specific recombinase FLP. The DNA fragment (39 flag-frt-cat-frt) was amplified by polymerase chain reaction (PCR) using primers PanDflagU and PanDflagL. We used high-fidelity KOD-plus DNA polymerase (TOYOBO, Osaka) for PCR. The resulting fragment was used for transformation into MG1655 according to the Datsenko and Wanner's procedure (Datsenko and Wanner 2000). Then the 39 flag tag was thereby fused with the carboxyl terminus of the PanD gene. If necessary, the cat genetic marker was removed using the FLP-FRP recombination (Datsenko and Wanner 2000).
To construct the PanD BS strain (SN219), the PanD gene of B. subtilis 168 was amplified using BSPD1 and BSPD21 as PCR primers. The frt-cat-frt fragment was also amplified from template DNA of PanD-flag-frt-cat-frt using primers of BSPD31 and PanDflagL. As the sequence of BSPD21 is complementary to the sequence BSPD31, the amplified DNA fragment encoding PanD of B. subtilis could anneal with the PanD-flag-frt-cat-frt DNA fragment to, yielding the PanD BS -frt-cat-frt DNA fragment. This DNA fragment was used to substitute E. coli PanD with PanD BS by recombination according to Datsenko and Wanner's procedure (Datsenko and Wanner 2000). Similarly a strain with PanD BS -flag (SN220) was constructed using primers BSPD1 and BSPD4 for amplification of PanD BS , and primers of BSPD5 and PanDflagL for that of flag-frt-cat-frt.
To construct ΔyhhK (PanZ) derivatives, ΔyhhK::kan, which is one of the deletion mutants in the Keio collection , was used to transduce each parental strain by P1 transduction, resulting in SN202, SN216, SN223 and SN224. The kan gene marker was eliminated according to Datsenko and Wanner's procedure (Datsenko and Wanner 2000). To construct the ΔpanF strain (SN225), ΔpanF::kan of the KEIO collection was introduced to MG1655 by P1 transduction.
pCA24N(-GFP)-his-PanZ is one of the expression plasmids of the ASKA clone (-) library, in which each cloned gene can be expressed by addition of IPTG (isopropylb-D-thiogalactopyranoside) with the gene product having an N-terminal His tag (Kitagawa et al. 2005).

wild-type
panF::kan panZ A B Figure 6. A model of feedback regulation of pantothenate synthesis by CoA and effect of PanF on cell growth. (A) Regulation of pantothenate synthesis and its cellular content in E. coli. Pantothenate is produced by condensation of b-alanine and pantonate. The following phosphorylation of pantothenate by CoaA is negatively regulated by CoA. After inhibition of the metabolic step of CoA synthesis, pantothenate is still produced in the cells and then an excess of pantothenate is secreted toward the living environment by PanF, which is a sodium dependent transporter. PanF can also incorporate extracellular pantothenate. Activation of PanD is dependent on PanZ. We speculate that CoA binds to PanZ when activation of PanD might be down. This secondary feed back loop allow cells to stop the pantothenate synthesis, in addition to CoaA. (B) The phenotype of the panZ mutant is characterized by a requirement of b-alanine that suppressed by neighbors of the wild type cells in a panF dependent manner. Cells were grown on the M9 medium with or without b-alanine and incubated at 37°C for 2 days. Open, closed and grey circles indicate the spot of wild-type cells, ΔpanF::kan cells and ΔpanZ cells, respectively. To construct pBAD24-PanD-flag, the PanD-flag fragment was amplified by PCR using the primers ECPanD-BADU and PanDL. The PCR product was digested with EcoRI and SphI and inserted into the EcoRI-SphI site of the pBAD24 vector plasmid (Guzman et al. 1995). Expression of the PanD-flag gene is induced by addition of arabinose. However, leaky, uninduced expression could complement the phenotype of ΔPanD. The vector pBAD24-PanZ-his was constructed similarly except that the PanZ gene was amplified using the primers PanZ-BADU and PanZBADL.
To construct pBAD24-PanZ(N45A)-his, a single nucleotide mutation was introduced into the PCR primers PanZN45AL and PanZN45AU. The 5' portion of the PanZ-his gene was amplified using the primers of PanZ-BADU and PanZN45AL, and the 3' portion using primers of PanZN45AU and PanZBADL. As the primer sequence of PanZN45AL is complementary to that of PanZN45AU, the PCR products could be annealed and thus the mutated PanZ gene was obtained. This PanZ (N45A)-his fragment was cloned into the pBAD24 vector as described above.
To construct an expression plasmid for PanZ, the PanZ gene was amplified using primers THPanZU and THPanZL. The PCR product was digested with EcoRI and BamHI and inserted into the EcoRI-BamHI site of pUT18C and pKT25 (Karimova et al. 1998), respectively, resulting in pUT18C-PanZ and pKT25-PanZ. pUT18C-PanD and pKT25-PanD were constructed similarly except that the PanD gene was amplified by primers THPanDU and THPanDL.
We converted the gene selection marker of pET28 (Novagen), which is an IPTG-inducible expression vector, from a kanamycin resistance gene to an ampicillin resistance gene. To construct pET28Ap, pET28 plasmid DNA was digested with AlwNI and ClaI to eliminate the kanamycin resistant gene. pBR322 plasmid DNA was digested with AlwNI and ClaI to obtain the ampicillin resistance gene, and the DNA fragment was ligated into the above pET28 plasmid DNA. To construct pET28Ap-PanD, the PanD gene was amplified from MG1655genomic DNA using the primers PanDPETU and PanDPETL. The PCR product was digested with NcoI and HindIII and introduced into the NcoI-HindII site of pET28Ap.
Strains of the KEIO collection were obtained from the National BioResource Project of the National Institute of Genetics, Mishima, Japan.

Purification of His-PanZ
ΔPanZ cells harboring pCA24N(-GFP)-his-PanZ were incubated in 1000 ml of L broth containing 10 lg/ml of chloramphenicol at 30°C until OD 600 reached 0.5, and then isopropyl-b-D-thiogalactopyranoside (IPTG) was added to a final concentration of 1mmol/L in order to express His-PanZ protein. The culture was incubated for an additional 4 h. Cells were harvested by centrifugation and resuspended in 10 ml of Lysis Buffer (50 mmol/L NaH 2 PO 4 , 300 mmol/L NaCl, 10 mmol/L imidazole 1 mg/ml lysozyme, pH 8.0) and incubated on ice for 30 min. Cells were disrupted by sonication and cell debris was removed by centrifugation to obtain a cleared lysate. Two ml of 50% slurry of Ni-NTA agarose (QIAGEN) was added to the cleared lysate and mixed by shaking at 4°C for 1 h. The lysate-Ni-NTA mixture was loaded into a column and washed twice with 10 ml of Wash Buffer (50 mmol/L NaH 2 PO 4 , 300 mmol/L NaCl, 20 mmol/L imidazole, pH 8.0). His-PanZ protein was eluted with 49 1 ml of Elution Buffer (50 mmol/L NaH 2 PO 4 , 300 mmol/L NaCl, 250 mmol/L imidazole, pH 8.0). Eluted His-PanZ protein was dialyzed against TBSE Buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1 mmol/L EDTA, pH 7.5) and used for subsequent assays.

Purification of PanD-FLAG
ΔPanZ cells harboring pBAD24-PanD-flag were incubated in 100 ml of L broth containing 50 lg/ml of ampicillin at 30°C until the OD 600 reached 0.5, and then arabinose was added to a final concentration of 0.2% in order to express PanD-FLAG. The culture was incubated for an additional 3 h. Cells were harvested by centrifugation and resuspended in 2 ml of Lysis Buffer and incubated on ice for 30 min. Cells were disrupted by sonication and cell debris was removed by centrifugation to obtain a cleared lysate. The cleared lysate was loaded onto a column with 200 ll of Anti-FLAG M2 affinity gel (SIGMA) and washed twice with 5 ml of TBS (50 mmol/L Tris-HCl, 150 mmol/L NaCl, pH 7.5). PanD-FLAG protein was eluted with 59 250 ll of TBS containing 100 lg/ml 39FLAG peptide. Eluted PanD-FLAG was dialyzed against TBSE Buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1 mmol/L EDTA, pH 7.5) and used for subsequent assays.

Purification of PanD
SN227 harboring pET28Ap-PanD was cultured in L medium containing 50 lg/ml ampicillin until the OD 600 reached 0.6, and then IPTG was added to a final concentration of 1 mmol/L. The culture was incubated for an additional 3 h. Cells were harvested by centrifugation and resuspended in Lysis buffer (50 mmol/L HEPES/KOH, 100 mmol/L potassium acetate, 1 mmol/L EDTA, 1 mg/ ml lysozyme, pH 7.6). Cells were lysed by freeze-thawing three times with liquid nitrogen and a single cycle of sonication and cell debris removed by ultracentrifugation.
Ammonium sulfate was added to the supernatant to 40% saturation and the precipitate was removed by ultracentrifugation. Further ammonium sulfate was added to the supernatant to 50% saturation. After ultracentrifugation, the precipitate was resuspended in K0 buffer (50 mmol/L HEPES/KOH, 1 mmol/L EDTA, pH 7.6) and dialyzed against K0 buffer for more than 3 h to remove ammonium sulfate. The dialyzed solution was loaded on a Resource Q (GE Healthcare, Uppsala, Sweden) anion-exchange column using an AKTA prime plus liquid chromatography system (GE Healthcare). Gradient elution was carried out from K0 buffer to K1000 buffer (50 mmol/L HEPES/ KOH, 1 mmol/L EDTA, 1 mol/L potassium acetate, pH7.6). Each eluted fraction was analyzed by SDS-PAGE visualized using Coomassie Brilliant Blue. Protein-containing fractions were collected and dialyzed against K100 buffer (50 mmol/L Hepes/KOH, 1 mmol/L EDTA, 100 mmol/L potassium acetate, pH 7.6) containing 10% glycerol for more than 3 h. After dialysis, purified PanD protein was stored at -80°C.
Purification of PanZ-His and PanZ(N45A)-His SN202 harboring pBAD24-PanZ-his or panZ(N45A)-his was cultured in L medium containing 50 lg/ml ampicillin until the OD 600 reached 0.6, and then L-arabinose was added to a final concentration of 0.2%. After further incubation for 3 h, cells were harvested by centrifugation and resuspended in Lysis buffer. Cells were lysed by three cycles of freeze-thawing in liquid nitrogen and a single cycle of sonication and cell debris removed by ultracentrifugation. Ammonium sulfate was added to the supernatant to a final 50% saturation. After the precipitate was removed by centrifugation, further ammonium sulfate was added to 80% saturation. After ultracentrifugation, the precipitate was resuspended in His-binding buffer (50 mmol/L HEPES/KOH, 100 mmol/L potassium acetate, 20 mM imidazole, pH 7.6) and loaded on HisTrap FF column (GE Healthcare) using the AKTA prime plus liquid chromatography system (GE Healthcare). Stepwise elution chromatography was carried out using Elution buffer (50 mmol/L HEPES/KOH, 100 mmol/L potassium acetate, 500 mmol/L imidazole, pH 7.6). Eluted proteins were analyzed by SDS-PAGE visualised using CBB. Peak fractions were collected and dialyzed against K100 buffer containing 10% glycerol. After dialysis, purified proteins were stored at -80°C.

Assay of PanD-FLAG cleavage in vivo and in vitro
To assay in vivo PanD-FLAG cleavage, exponentially growing cells in LB broth at 37°C were harvested, and washed once with 0.85% (w/v) NaCl. The cells were lysed in 19 SDS-PAGE loading buffer and subjected to SDS polyacrylamide (20%) gel electrophoresis. After blotting of the proteins to a nitrocellulose membrane (GE Healthcare), imumunodetection using anti-FLAG M2 antibody (SIGMA) was carried out. To express His-PanZ protein in SN216 cells harboring pCA24N(-GFP)-his-PanZ, cells were incubated in LB broth at 37°C until the OD 600 reached 0.2, then IPTG was added to a final concentration of 0, 1, 10 and 100 lM. The cells were incubated for an additional 60 min. SN216 cells harboring pCA24N(-GFP) were used as control.

Phylogenetic tree of panD
A phylogenetic tree of PanD was generated using ClustalX (Thompson 1997) and drawn using TreeView X software (Page 1996).