The Saccharomyces cerevisiae Hal3 protein is a moonlighting protein, able to function both as an inhibitory subunit of the Ppz1 protein phosphatase and as a constituent protomer of an unprecedented heterotrimeric PPCDC (phosphopantothenoylcysteine decarboxylase), the third enzyme of the CoA biosynthetic pathway. In the present study we initiated the dissection of the structural elements required for both disparate cellular tasks by using a combination of biochemical and genetic approaches. We show that the conserved Hal3 core [PD (PPCDC domain)] is necessary for both functions, as determined by in vitro and in vivo assays. The Hal3 NtD (N-terminal domain) is not functional by itself, although in vitro experiments indicate that when this domain is combined with the core it has a relevant function in Hal3's heteromeric PPCDC activity. Both the NtD and the acidic CtD (C-terminal domain) also appear to be important for Hal3's Ppz1 regulatory function, although our results indicate that the CtD fulfils the key role in this regard. Finally, we show that the introduction of two key asparagine and cysteine residues, essential for monofunctional PPCDC activity but absent in Hal3, is not sufficient to convert it into such a homomeric PPCDC, and that additional modifications of Hal3's PD aimed at increasing its resemblance to known PPCDCs also fails to introduce this activity. This suggests that Hal3 has undergone significant evolutionary drift from ancestral PPCDC proteins. Taken together, our work highlights specific structural determinants that could be exploited for full understanding of Hal3's cellular functions.
The capacity of a growing number of proteins to ‘moonlight’ – that is, to develop several functions within a single polypeptide chain that often are apparently unrelated – represents a way to expand the functional capabilities of an organism in spite of the limited protein-encoding capacity of its genome [1,2]. The discovery of such types of proteins also represented the end of a paradigm in biology.
Previously we reported that in the yeast Saccharomyces cerevisiae, Hal3 (also known as Sis2) and Vhs3 (either separately or in combination) interact with Ykl088w (since renamed as Cab3) to form a heterotrimeric protein that functions as the PPCDC [PPC (phosphopantothenoylcysteine) decarboxylase] enzyme in this organism . PPCDCs are flavoproteins that catalyse the third step of CoA biosynthesis, i.e. the decarboxylation of 4′-PPC to form 4′-phosphopantetheine. This is accomplished through an innovative two-step mechanism in which the substrate is first oxidatively decarboxylated through intervention of the flavin cofactor and an essential histidine residue to form an enethiol intermediate. This intermediate is subsequently reduced in a second step that is mediated by an essential cysteine residue, which returns the flavin cofactor to its oxidized state and releases the final product [4,5]. PPCDC enzymes have been characterized in bacteria and eukaryotic organisms, such as humans and the plant Arabidopsis thaliana. However, in contrast with the exceptional yeast protein, all other eukaryotic PPCDCs are homotrimers [6–9]. Moreover, in these enzymes three active sites are formed at the protomer interfaces with the catalytically essential histidine and cysteine residues (His90 and Cys175 in AtHal3a, the PPCDC from A. thaliana) provided by opposing protomers. In contrast, each respective constituent member of the heterotrimeric yeast protein has only either the histidine or the cysteine residue, resulting in the formation of a single functional active site per trimer .
The discovery of the involvement of Hal3 and Vhs3 in CoA biosynthesis was even more remarkable in light of their initial description as negative regulatory subunits of the yeast serine/threonine protein phosphatases Ppz1 and Ppz2 [10,11], which modulate salt tolerance, cell integrity and cell cycle progression, among other processes (see  for a review). Although little is known about Vhs3 [11,13], S. cerevisiae Hal3 has been studied in some detail previously. Initially identified as a halotolerance determinant  and as a high-copy suppressor of the growth defect of cells deficient in the Sit4 protein phosphatase , it was subsequently found to interact in vivo and in vitro with the C-terminal catalytic domain of Ppz1 and to strongly inhibit its phosphatase activity . The inhibition of Ppz1 by Hal3 presented an explanation for the phenotype derived from HAL3 mutation or overexpression, since decreased activity of Ppz1 leads to salt tolerance, compromised cell integrity, altered translation efficiency and accelerated G1/S transition [16–22].
This involvement of the Hal3 and Vhs3 proteins in such disparate cellular functions provided an explanation for the enigmatic synthetically lethal phenotype of the hal3 vhs3 double deletion mutant , and clearly qualified them for consideration as new examples of moonlighting proteins in yeast. However, the identification and characterization of their unrelated functions also raised the question of the molecular basis for such diversity. Hal3 is a 562-residue protein that can be divided into three segments (Figure 1A): (i) an N-terminal region of approximately 250 residues that shows no evident similarity to other proteins outside yeasts, here dubbed the NtD (N-terminal domain); (ii) a core region of roughly equivalent size, which shows strong similarity to PPCDC enzymes such as AtHal3a from A. thaliana [PD (PPCDC domain)]; and (iii) a C-terminal tail of approximately 80 residues, highly enriched in acidic residues [CtD (C-terminal domain)]. Interestingly, both Vhs3 and Cab3 (the third member of the PPCDC heterotrimer in yeast) are structurally related to Hal3, having 49 and 28% sequence identity to the Hal3 protein respectively . All three proteins also share this three-segment motif, except that the PD of Hal3 and Vhs3 contain additional non-homologous insertions. The residues essential for PPCDC activity are also located in this central domain, with Hal3 and Vhs3 each containing the required histidine residue, whereas Cab3 contains the cysteine residue and an asparagine residue (Asn141 in AtHal3a), which has also been shown to be essential, although its catalytic role still needs to be elucidated [7,8,23]. This analysis, and our previous results which demonstrated that the expression of the plant or human PPCDC is sufficient to provide PPCDC function to yeast cells lacking all three Hal3, Vhs3 and Cab3 proteins, suggest that the structural determinants for PPCDC activity are restricted to the PD of the yeast proteins . On the other hand, random mutagenesis of the central region of Hal3 (from Arg256 to Ile480) revealed nine single amino acid changes that affected the ability of Hal3 to bind or inhibit Ppz1, although none of them prevented rescue of the hal3 vhs3 synthetically lethal phenotype . This indicated that these changes do not interfere with the PPCDC functions of these proteins, and suggested that the structural determinants for Ppz1 regulation and PPCDC activity are independent.
Schematic diagram showing the comparison of AtHal3, Hal3 and Hal3 constructs prepared in the present study
To further explore the molecular basis of the moonlighting functions of Hal3, we set out to dissect the Hal3 structure in the present study, and to determine the involvement and role of each of its constituent parts in its two known functions: as an inhibitor of Ppz1, and as one of the protomers of the yeast PPCDC enzyme. This allowed us to investigate in a systematic way the structural requirements of these unrelated activities in Hal3, with the goal to gain further insight into the origin and development of this intriguing protein's ability to perform such exceedingly distinct cellular functions.
Escherichia coli and yeast growth conditions
Yeast cells were grown at 28°C in YPD medium (10 g/l yeast extract, 20 g/l peptone and 20 g/l dextrose) or, when carrying plasmids, in synthetic minimal medium lacking the appropriate selection requirements . Yeast strains used in the present study are described in Table 1. Strain MAR30 (1788 hal3::LEU2/HAL3 vhs3::URA3/VHS3 cab3::kanMX4/CAB3) was generated by transforming the cab3::kanMX4 cassette, described in  into the MAR6 (hal3/HAL3 vhs3/VHS3) strain. The MAR6 strain was previously described in . E. coli DH5α or Mach1 cells (Invitrogen) were used as plasmid DNA host and were grown at 37°C in LB (Luria–Bertani) broth supplemented, if necessary, with 100 μg/ml ampicillin or 30 μg/ml kanamycin (Roche). Bacterial and yeast cells were transformed using standard methods . Recombinant DNA techniques were performed following standard protocols as described elsewhere .
|JA100||MATa ura3-52 leu2-3,112 his4 trp1-1 can-1r|||
|JC002||JA100 tet0:HAL3 sit4::TRP1|||
|IM021||JA100 ppz1::kanMX4 hal3::LEU2|||
|BY4741||MATa his3Δ 1 leu2Δ met15Δ ura3Δ|||
|1788||MATa/α ura3-52 leu2-3,112 his4 trp1-1 can-1r||D. Levin*|
|AGS4||1788 hal3::LEU2/HAL3 vhs3::kanMX4/VHS3|||
|MAR30||1788 a/α hal3::LEU2/HAL3 vhs3::URA3/VHS3 cab3::kanMX4/CAB3||The present study|
|AGS31||1788 hal3::LEU2/HAL3 vhs3::nat1/VHS3 cab3::kanMX4/CAB3|||
|JA100||MATa ura3-52 leu2-3,112 his4 trp1-1 can-1r|||
|JC002||JA100 tet0:HAL3 sit4::TRP1|||
|IM021||JA100 ppz1::kanMX4 hal3::LEU2|||
|BY4741||MATa his3Δ 1 leu2Δ met15Δ ura3Δ|||
|1788||MATa/α ura3-52 leu2-3,112 his4 trp1-1 can-1r||D. Levin*|
|AGS4||1788 hal3::LEU2/HAL3 vhs3::kanMX4/VHS3|||
|MAR30||1788 a/α hal3::LEU2/HAL3 vhs3::URA3/VHS3 cab3::kanMX4/CAB3||The present study|
|AGS31||1788 hal3::LEU2/HAL3 vhs3::nat1/VHS3 cab3::kanMX4/CAB3|||
Department of Molecular and Cell Biology and Department of Microbiology, Boston University School of Medicine, Boston, MA02118, U.S.A.
HAL3 constructs (Figure 1B) were amplified by PCR and subcloned into pWS93  for high-copy expression in S. cerevisiae under the strong ADH1 promoter, and into pGEX6P-1 (Amersham Biosciences) for bacterial expression of GST (glutathione transferase)-fusion proteins. The PD of Hal3 was amplified by PCR using the oligonucleotides 5′-Hal3_Core (EcoRI) and 3′-Hal3_Core (XhoI, SalI) with pGEX-Hal3 [11,24] – which was subcloned with EcoRI and XhoI restriction sites – as template, and subsequently subcloned into the EcoRI/SalI sites of pWS93 to yield pWS93-Hal3_core. The XhoI site introduced into this plasmid was utilized in the subcloning of the other versions of Hal3. The NtD of Hal3 was amplified by using the oligonucleotides 5′-pGEX and 3′-Hal3_N-terminus (XhoI), whereas the N-terminal and PDs (NtD+PD) were amplified by using 5′-pGEX and 3′-Hal3_Core (XhoI, SalI). The PD and CtD (PD+CtD) were amplified by using 5′-Hal3_Core (EcoRI) and 3′-pGEX. The deletion and mutations for Hal3 Δ312–350, Hal3 HVC (Hal3N466C) and Hal3 HNC (Hal3V430N, N466C) was introduced by means of SOE (single overlap extension) PCR using the pGEX sequencing primers (5′-pGEX and 3′-pGEX), as well as primers designed to introduce the desired deletion and mutations: the removal of the insert region (934–1050 bp, amino acids 312–350 of the native Hal3 sequence), and the introduction of the mutations Asn466 to cysteine (AAT to TGT) and Val430 to asparagine (GTG to AAC). The plasmid pGEX-Hal3 was used as template in the construction of all the of above plasmids, except for Hal3 HNC where the plasmid pGEX-Hal3_HVC was used as template. The obtained PCR products were subcloned into the EcoRI/XhoI restriction sites of pWS93 Hal3_PD. The inserts were subsequently subcloned from the pWS93 plasmids into the EcoRI/XhoI restriction sites of pGEX6P-1 (Amersham Biosciences). To construct the plasmid pSK93-Hal3_HNC (equivalent to pWS93 but carrying a TRP1 gene instead of the URA3 marker), an EcoRI/SalI fragment obtained from digestion of pWS93-Hal3_HNC was cloned in the same sites of pSK93.
The various PD constructs Hal3_PD and Hal3_PD_HNC were amplified by PCR from the appropriate pGEX6P-1 vectors with the oligonucleotides 5′-Hal3_CoaC (NdeI) and 3′-Hal3_CoaC (XhoI). The Hal3_PD_HNC_Δ312–350 construct was prepared by SOE PCR using pET28a-Hal3_PD_HNC as template, and the T7 sequencing primers (5′-T7 promoter and 3′-T7 terminator) and the primers described above for the preparation of the pGEX-Hal3_Δ312–350 construct. Hal3_PD_HNRepC and Hal3_PD_HNRepC_Δ312–350 were prepared by SOE PCR using the T7 sequencing primers and the oligonucleotides 5′-Hal3Core_HNRepC and 3′Hal3Core_HNRepC, with pET28a-Hal3_PD_HNC and pET28a-Hal3_PD_HNC_Δ312–350 serving as templates, respectively. All the PD constructs described above were subcloned into the NdeI/XhoI restriction sites of pET28a (Novagen) yielding the vectors pET28aHal3_PD, pET28a-Hal3_PD_HNC, pET28a-Hal3_PD_HNC_Δ312–350, pET28a-Hal3_PD_HNRepC and pET28aHal3_PD_HNRepC_Δ312–350, respectively for expression as His6-fusion proteins. All newly generated constructs were verified by DNA sequencing. Oligonucleotides used in the present study are listed in Supplementary Table S1 (available at http://www.BiochemJ.org/bj/442/bj4420357add.htm). The construction of plasmids for bacterial expression of Ppz1 and Ppz1 Δ1–344 (pGEX-PPZ1 and pGEX-PPZ1Δ1–344) was described previously . All plasmids used in the present study are listed in Supplemental Table S2 (available at http://www.BiochemJ.org/bj/442/bj4420357add.htm).
Assessment of yeast phenotypes
Sensitivity of yeast cells to LiCl, NaCl, caffeine (Merck), hygromycin, TMA (tetramethylammonium) and spermine (all from Sigma) was evaluated by growth on agar plates (drop tests) as described previously . Growth under limiting external potassium concentrations was carried out using Translucent K+-free medium , which carries a negligible potassium concentration of approximately 15 μM, supplemented with 1 mM KCl. The effect of high-copy expression of different versions of Hal3 on sit4 cells was studied by monitoring the growth of JA110 (sit4) cultures transformed with the various pWS93 plasmids. The cultures were inoculated on synthetic minimal medium lacking uracil at a D660 of 0.01 and growth was monitored for 24 h. Sporulation and random spore analysis was performed essentially as described previously [25,29].
Recombinant protein expression and purification
The described pGEX6P-1 plasmids were transformed into E. coli BL21 (DE3) RIL or E. coli BL21 (DE3) co-transformed with pRARE2 (Novagen), to allow for the expression of rare codons. Expression and purification of Hal3, Cab3, Ppz1 and Ppz1 Δ1–344 were essentially performed as described previously [3,11]. Hal3 NtD, Hal3 NtD+PD, Hal3 Δ312–350, Hal3 HVC and Hal3 HNC were expressed under the same conditions as Hal3 [induced with 1 mM IPTG (isopropyl β-D-thiogalactopyranoside)] at a D660 of 0.6 and expressed for 3 h at 37°C. Hal3 PD and Hal3 PD+CtD were induced with 0.1 mM IPTG at a D660 of ~0.6 and expressed overnight at 25°C. Briefly, the collected cells were disrupted by sonication in lysis buffer [50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100 and a protease inhibitor cocktail (Thermo Scientific or Sigma)], and allowed to bind to equilibrated glutathione–Sepharose 4B resin (GE Healthcare), followed by extensive washing. For the expression and purification of Ppz1 and Ppz1 Δ1–344, 2 mM MnCl2 was added to all media and buffers. When necessary, the untagged proteins were eluted by incubating the resin with PreScission protease (GE Healthcare) overnight at 4°C according to the manufacturer's instructions. In all cases the concentration of the protein of interest was estimated by densitometry of samples run on SDS/PAGE (8 or 12% gels) stained with Coomassie Brilliant Blue, by comparison with known amounts of BSA run on the same gel. Densitometry analyses were performed using UN-SCAN-IT gel (Silk Scientific Corporation).
The described pET28a plasmids were transformed into E. coli BL21 (DE3), and expressed in LB broth. The cultures were induced with 0.2 mM IPTG at a D660 of ~0.6 and expressed overnight at 25°C. The His6-fusion proteins were purified by Ni2+-affinity chromatography using an ÄKTAprime (GE Healthcare) system and HiTrap chelating columns (GE Healthcare). Briefly, cells collected from the 500 ml expressions were resuspended in binding buffer (50 mM Tris/HCl, pH 8.0, 500 mM NaCl and 10% glycerol), sonicated and subsequently centrifuged to remove cellular debris. The obtained cell extract was loaded on to the column, followed by washing with binding buffer and 15% elution buffer (50 mM Tris/HCl, pH 8.0, 150 mM NaCl, 500 mM imidazole and 10% glycerol) to elute non-specifically bound proteins, followed by elution of the protein of interest in 100% elution buffer. Fractions containing the protein of interest were combined and subsequently exchanged to gel-filtration buffer (50 mM Tris/HCl, pH 8.0, 150 mM NaCl and 10% glycerol) using HiTrap desalting columns (GE Healthcare). The buffer-exchanged proteins were stored at −80°C. Protein concentration was determined using the Bradford method  modified for use in 96-well plates.
Preparation of yeast protein extracts
Yeast cells from strain IM021 (ppz1 hal3) transformed with the pWS93 multicopy plasmid carrying the different HA (haemagglutinin)-tagged versions of Hal3 under study were grown in 40 ml of synthetic minimal medium lacking uracil to a D660 of 1.5–2 at 28°C. Cells were collected by centrifugation for 5 min at 1620 g, and resuspended in 800 μl of extraction buffer containing 50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 0.1% Triton X-100, 1 mM DTT (dithiothreitol), 10% glycerol, 2 mM PMSF and Complete™ protease inhibitor mixture (Roche Applied Science). Then 300 μl of acid-washed glass beads (Sigma) was added and cells were broken at 4°C by vigorous shaking in a Fast Prep cell breaker (Bio 101) at setting 5.5 for 25 s, four times in total. After sedimentation at 500 g for 10 min at 4°C, the cleared lysate was recovered and the protein concentration was quantified by a Bradford assay.
Interaction between the diverse versions of Hal3 and Ppz1 or Ppz1Δ1–344
A 50 μl portion of the affinity beads containing 4 μg of Ppz1 was mixed with 600 μg of each protein extract, except for PD (6 mg), PD+CtD (6 mg) and NtD (7.2 mg). Similarly, 50 μl of the beads containing 4 μg of Ppz1 Δ1–344 was mixed with 300 μg of each protein extract, except for PD (3 mg), PD+CtD (3 mg) and NtD (3.6 mg). The mixtures were incubated for 1.5 h at 4°C with gentle shaking. Samples were centrifuged and the supernatant was removed. Beads were resuspended in 200 μl of extraction buffer, transferred to MultiScreen filter plates (Millipore), and extensively washed with the same buffer (without protease inhibitors). The beads were finally resuspended in 100 μl of 2×SDS sample buffer and boiled for 5 min. A 20 μl volume of the samples was fractionated by SDS/PAGE (12% gels) and transferred to Immobilon-P membranes (Millipore). Membranes were incubated for 1 h with anti-HA antibody (Covance) at 1:1000 dilution, followed by the secondary HRP (horseradish peroxide)-conjugated anti-mouse IgG antibody (Amersham Biosciences) at 1:20000 dilution. The immunocomplexes were visualized using an enhanced chemiluminescence Advance Western blotting detection kit (GE Healthcare). Chemiluminescence was detected using an LAS-3000 instrument (Fuji) and quantified using Multi Gauge version 3.0 software.
In vitro phosphatase assays
Ppz1 activity assays were performed essentially as described previously . Briefly, increasing amounts of the Hal3 constructs (actual values are indicated on the x-axis of Figures 5A and 5B) were incubated with Ppz1Δ1–344 (3.7 nmol Pi·min−1·μg−1 Ppz1 Δ1–344) and Ppz1 (3.27 nmol Pi·min−1·μg−1 Ppz1) for 5 min at 30°C, and the reaction was started by adding 10 mM PNPP (p-nitrophenyl phosphate; Sigma). A reaction buffer of 50 mM Tris/HCl, pH 7.5, 2 mM MnCl2 and 1 mM DTT was used. The reactions were incubated at 30°C for 20 min, after which phosphatase activity was measured at 405 nm using a microplate UV spectrophotometer (Varioskan, Thermo Labsystems). The measured background absorbance reading was subtracted from the measured absorbance values.
In vitro PPCDC assays
PPCDC assays were performed as described previously, using HsCoaC as positive control . For assays based on the formation of the 4′-phosphopantetheine product, proteins (either individually or in mixtures, with a total concentration of 60 nM) were added to the individual wells of a 96-well plate, followed by initiation of the reaction by addition of the rest of the assay mixture [60 μl of PPi reagent (Sigma), 0.5 mM PPC (synthesized as described in ), 2 mM ATP, 1 mM DTT, 10 mM MgCl2, 20 mM KCl and ~1.5 μM His6–EcPPAT  in 50 mM Tris/HCl, pH 7.6]. The final reaction volume was 150 μl. The reaction progress was followed at 340 nm and 37°C using a microplate-based UV spectrophotometer (Varioskan, Thermo Labsystems).
For assays based on CO2 formation, assay mixtures (150 μl) contained 0.5 mM PPC, 5.0 mM MgSO4, 1.0 mM PEP (phosphoenolpyruvate), 0.16 mM NADH, 1.0 mM DTT, 1.8 units malate dehydrogenase, 5.5 units LDH (lactate dehydrogenase), 0.375 unit PEPC (PEP carboxylase) and the putative PPCDCs in 50 mM Tris/HCl, pH 7.6. The PPCDC proteins were loaded into the individual wells of a 96-well microplate, followed by a short incubation period at 37°C. The reaction was subsequently initiated by addition of the rest of the assay mixture at 37°C and monitored by following the changes in absorbance at 340 nm.
Cross-linking experiments and size-exclusion chromatography
Cross-linking of recombinant Hal3 and Hal3 HNC proteins was carried out using glutaraldehyde as follows. GST–Hal3 and GST–Hal3 HNC constructs were expressed as described previously , purified using glutathione–agarose beads and treated with PreScission protease (GE Healthcare) using 50 mM NaH2PO4 (pH 7.0) instead of Tris/HCl buffer as described in . A 5 μl volume of a 0.05% stock solution of glutaraldehyde was added to 45 μl of a solution containing 1 μg of each Hal3 or Hal3 HNC to make 0.005% glutaraldehyde. Then, samples were incubated for 20 min at 24°C. The cross-linking reaction was stopped by adding 2 μl of 1 M Tris/HCl (pH 7.5) and subsequent incubation for 5 min at 24°C. After addition of the appropriate volume of sample buffer (4 ×), samples were boiled for 5 min and subjected to SDS/PAGE (6% gel). An equivalent cross-linking reaction was carried out with 1 μg of BSA as a negative control.
Size exclusion chromatography was performed on an ÄKTApurifier (GE Healthcare) with a Superose 12 column (Amersham), using 50 mM phosphate, pH 7.0, and 150 mM NaCl as eluant at a flow rate of 0.5 ml/min. Approximately 100 μl of each the respective proteins were loaded on to the column. The column was calibrated using carbonic anhydrase (29 kDa), BSA (66 kDa), alcohol dehydrogenase (150 kDa), β-amylase (200 kDa), apoferritin (443 kDa) and thyroglobulin (669 kDa) obtained from Sigma as a calibration kit. The column's void volume was determined with Dextran Blue.
Design of Hal3 constructs
To allow the study of the involvement of Hal3's constituent parts in its two known functions, i.e. as inhibitor of Ppz1 activity and as one of the protomers of the yeast PPCDC enzyme, constructs were designed based on a systematic dissection of its structure, thus leading to the preparation of versions containing only the NtD (Hal3 NtD), only the PD (Hal3 PD), and constructs containing the PD fused to either the NtD or CtDs (Hal3 NtD+PD and Hal3 PD+CtD respectively). Moreover, a further version (Hal3 Δ312–350) was prepared in which the non-homologous insertion sequence had been removed. Finally, since the PD of Hal3 only contains one of the three essential residues that are associated with PPCDC activity (His378, equivalent to AtHal3 His90), two Hal3 mutants were prepared to systematically introduce the two other catalytically essential residues (Asn141 and Cys175 in AtHal3a) into the protein at their expected positions. This was done to determine whether these exchanges were functionally relevant in the context of Ppz1 inhibition (note that Cab3, which contains both of these residues in its native version, is unable to inhibit Ppz1 ) and would also establish whether re-introduction of the essential residues converts Hal3 into a monofunctional PPCDC. The introduction of the essential residues was achieved by first exchanging Asn466 for cysteine, since this residue is known to be required for the second step in the PPCDC reaction. Next, the Val430 residue in this mutant was exchanged for an asparagine residue, which is also required for PPCDC activity although its exact role in the mechanism is still not known. The two mutants were named Hal3 HVC and Hal3 HNC respectively, to denote the identity of the residues occupying the positions where the three essential PPCDC catalytic residues are expected to occur. All of the constructs are schematically represented in Figure 1(B).
Ability of constructs to perform the regulatory functions of Hal3 in vivo
The overexpression of Hal3 is known to confer tolerance to LiCl and NaCl , to improve the growth of Sit4-lacking strains  and to aggravate the lytic phenotype of a slt2 mutant . All of these phenotypes are known to be mediated by inhibition of the Ppz1 (and perhaps Ppz2) phosphatase by Hal3. In the present study we evaluated the ability of the Hal3 constructs (Figure 1B) to confer these phenotypic traits and thus determine which Hal3 domains are involved in its function as a phosphatase inhibitor. The results of these experiments are represented in Figures 2 and 3. Overexpression of the diverse Hal3 constructs in a wild-type strain exposed to toxic levels of LiCl demonstrated that Hal3 Δ312–350, Hal3 HVC and Hal3 HNC conferred hypertolerance similar to native Hal3, whereas expression of the rest of the constructs either did not confer tolerance at all, or showed only a very slight increase in tolerance compared with the empty plasmid. When the same experiment was performed in a strain lacking endogenous Hal3 (BY4741 hal3) the same three constructs were able to provide LiCl tolerance to levels similar to those of the native Hal3 polypeptide. In contrast, the plasmids carrying Hal3 PD and NtD behaved similarly to the empty plasmid control, whereas those carrying PD+CtD and NtD+PD were able to confer some tolerance to LiCl, albeit to a lesser extent compared with native Hal3 (Figure 2A).The same results were obtained in a different wild-type background (JA100) and in strain JA104, its hal3 isogenic derivative (results not shown).
Examples of phenotypes derived from overexpression of the different versions of Hal3
When the above-mentioned constructs were expressed in a Slt2 MAPK (mitogen-activated protein kinase)-deficient strain (JC010) at the standard growth temperature (28°C), we observed that overexpression of Hal3 Δ312–350, Hal3 HVC and Hal3 HNC, similarly to native Hal3, was able to induce cell lysis (note that growth is restored by addition of 1 M sorbitol) (Figure 2B). The rest of the constructs did not significantly affect growth at 28°C. It is known that slt2 mutant cells are hypersensitive to caffeine and that overexpression of Hal3 aggravates this phenotype . As expected, expression of these constructs also blocked growth, even in the presence of low amounts of caffeine. Remarkably, expression of the Hal3 PD version yielded cells hypertolerant to the drug (Figure 2B). The ‘anti-Hal3’ effect of the Hal3 PD version was also detected when slt2 cells were stressed at 37°C, as cells carrying this version were able to grow at this restrictive temperature. In this case, overexpression of the PD+CtD, NtD and NtD+PD forms was also able to support moderate growth. The anomalous effect of overexpression of Hal3 PD could also be observed when the constructs were expressed in a sit4 strain, since it aggravated the slow growth of the sit4 mutant, whereas native Hal3 alleviates this defect, as well as by monitoring growth of a hal3 strain under limiting potassium amounts in the medium (Supplementary Figure S1 available at http://www.BiochemJ.org/bj/442/bj4420357add.htm). This finding was confirmed in the conditional synthetic lethal strain JC002 (tetO:HAL3 sit4), in which expression of the Hal3 PD largely blocks growth even in the absence of doxycycline (Figure 2C).
A total of 15 phenotypic tests corresponding to functions known to be related to the inhibitory role of Hal3 on Ppz phosphatases were carried out for each of the Hal3 versions and the results were quantified on a scale from +5 (equivalent to overexpression of native Hal3) to −5 (totally opposed to Hal3 expression), with a value of zero being assigned to the phenotype of the strain carrying the empty plasmid. The data were subjected to cluster analysis to give the results shown in Figure 3 in the form of a heat map. As can be seen from these results, Hal3 Δ312–350, Hal3 HVC and Hal3 HNC behaved in a manner very similar to native Hal3. In contrast, the PD and NtD versions exhibited a phenotype contrary to that of native Hal3, except in those tests performed in the hal3 mutants, where they behaved similarly to the empty plasmid. NtD+PD and PD+CtD behaved similarly to the empty plasmid in the wild-type genetic background, whereas these constructs exhibit some Hal3 function in the absence of the chromosomal copy of HAL3.
Integration of phenotypic data from different strains overexpressing different versions of Hal3
Binding of the diverse Hal3 constructs to Ppz1 and Ppz1 Δ1–344 phosphatases
Previous studies have demonstrated that Hal3 binds to the C-terminal half of Ppz1, which contains the phosphatase catalytic domain . We tested the ability of the different constructs to bind to either the entire Ppz1 protein or to its C-terminal phosphatase domain. To this end, both forms of the phosphatase were expressed in E. coli as GST-fusion proteins and the recombinant polypeptide, once bound to a glutathione–agarose resin, was used as an affinity trap for the different Hal3 versions expressed in S. cerevisiae. The results, shown in Figure 4, confirm the previous evidence that native Hal3 binds stronger to the C-terminal half of Ppz1 than to the entire protein. The NtD+PD polypeptide shows a similar pattern. In contrast, the Hal3 Δ312–350 protein only binds somewhat stronger to the entire Ppz1 than to the protein containing only the Ppz1 C-terminal half. This differential behaviour was even more evident for the PD and the PD+CtD versions, which showed very strong binding to the complete Ppz1 protein. Finally, the NtD polypeptide was unable to bind to either phosphatase version.
Binding of the different versions of Hal3 to intact Ppz1 and to the Ppz1 C-terminal catalytic half
In vitro inhibition of Ppz1 and Ppz1 Δ1–344
The ability of the various Hal3 constructs to inhibit the phosphatase activity of Ppz1 and the Ppz1 C-terminal catalytic domain (Ppz1 Δ1–344) was evaluated with an in vitro assay using PNPP as substrate. The results show a clear differentiation in inhibitory effect among the constructs, which is more pronounced in the case of the C-terminal catalytic domain of Ppz1 (Figure 5A) than with the entire Ppz1 protein (Figure 5B). Among the active inhibitors are all of those constructs that contain a PD and a CtD, whereas constructs that lacked this motif, such as Hal3 NtD and Hal3 PD, behaved similarly to the negative control (BSA). Importantly, the human PPCDC protein HsCoaC, which has high sequence similarity to the PD of Hal3, also did not have any inhibitory effect on Ppz1. Finally, the NtD+PD version gave contrasting results: it also formed part of the group of inhibitory constructs when assayed with Ppz1 Δ1–344, but had no effect on the entire Ppz1 at low concentrations, and only an intermediate effect at high concentrations. Taken together, these results indicate that the presence of Hal3's CtD is an essential requirement for the inhibition of Ppz1's phosphatase activity. This is in agreement with early observations that the Hal3 acidic tail performs an important role in salt tolerance and cell cycle progression [14,15], and also supports the phenotypic results shown above.
Ability of the Hal3 constructs to inhibit either the Ppz1 C-terminal catalytic domain (Ppz1 Δ1–344) (A) or Ppz1 (B) in vitro
Hal3's PD is sufficient for its PPCDC-related function in vivo
A hal3 vhs3 deletion mutant is synthetically lethal , due to the essential role the Hal3/Vhs3 pair plays in the formation of an active PPCDC . The ability of the various constructs to rescue the double deletion mutant was evaluated by means of random spore analysis. The results, shown in Table 2, indicate that all of the constructs studied, except Hal3 NtD, are able to rescue the deletion mutant. This was expected since only constructs containing the PD, which exhibits high homology with known PPCDCs and conserves the essential histidine residue required for activity of the heteromeric PPCDC protein in yeast , should show PPCDC activity.
|Construct||hal3 vhs3||cab3||hal3 vhs3 cab3|
|Construct||hal3 vhs3||cab3||hal3 vhs3 cab3|
In vitro PPCDC activity of the Hal3 constructs in conjunction with Cab3
To confirm the in vivo results, the various Hal3 constructs were evaluated for their ability to join with Cab3 to form a catalytically active heterotrimeric PPCDC in vitro. When these constructs were combined with Cab3 in a 2:1 ratio (the optimal ratio to form an active trimer as determined in previous studies ), it was found that Hal3 NtD+PD, Hal3 Δ312–350, Hal3 HVC and Hal3 HNC all have levels of activity comparable with that of native Hal3 (Figure 6A). Interestingly, Hal3 PD and Hal3 PD+CtD only exhibited a small amount of PPCDC activity compared with the other constructs, whereas Hal3 NtD did not show any activity as expected. These results indicate that, apart from the obvious requirement of the PD, Hal3 also requires its NtD to exhibit full PPCDC activity in combination with Cab3. This suggests that the NtD of Hal3 is important in forming an active heterotrimeric complex with Cab3.
In vitro PPCDC activity assays based on product formation
Engineering a homotrimeric yeast PPCDC based on Hal3
Previously we described the engineering of a functional homomeric yeast PPCDC based on Cab3, which contains a non-functional histidine residue in combination with the two other essential residues required for PPCDC activity (asparagine and cysteine) . This was accomplished by replacing the amino acid sequence surrounding the histidine residue with the corresponding sequence surrounding the essential histidine of Hal3, to provide Cab3 with all of the catalytically essential residues required for PPCDC activity. Subsequent experiments demonstrated that the newly engineered Cab3 could rescue the lethal hal3 vhs3 cab3 mutant, indicating an ability to replace the PPCDC-related functions of Hal3 and Vhs3 in vivo. The purified protein also exhibited PPCDC activity in vitro, although at a much lower level than that shown by the native heterotrimeric yeast protein. This reduced activity was ascribed to the probable reduced ability of Cab3 to form homotrimers.
Similarly, the Hal3 HVC (N466C) and the Hal3 HNC (V430N; N466C) versions, which were prepared in the present study to interrogate the functional relevance of the loss of two of the essential PPCDC residues in Hal3, re-introduces the catalytically essential cysteine and asparagine residues in a stepwise manner. We therefore expected that Hal3 HNC would act as fully functional homomeric PPCDC and would be able to fulfil the roles of Hal3/Vhs3 and Cab3 in vivo and in vitro. However, in vivo experiments showed that neither Hal3 HVC nor Hal3 HNC was able to rescue the cab3 mutant; instead, these versions only allowed survival of the hal3 vhs3 strain (Table 2). Similarly, the purified Hal3 HNC protein was also unable to catalyse the in vitro PPCDC reaction, whether on its own or in combination with Hal3 (i.e. as a replacement for Cab3 in the native heterotrimer) (Figure 6B). This lack of activity is not due to structural defects in the mutant, since when Hal3 HNC (or Hal3 HVC) is assayed together with Cab3 these combinations display substantial PPCDC activity (Figure 6A). Hal3 HNC is also able to form the homomeric trimers required for activity, as demonstrated by the results of a cross-linking experiment (Figure 6C). This indicates that while the introduction of the essential residues to Hal3 HVC and Hal3 HNC does not confer PPCDC activity on them, both mutants can at least act as native Hal3 proteins by interacting with Cab3 and providing the essential histidine residue in an active heterotrimeric PPCDC protein.
Investigation of the evolutionary drift of Hal3's PD from known PPCDCs
The surprising result that Hal3 HNC does not show PPCDC activity was investigated further by focusing on the PD of Hal3, since this domain closely mimics the structures of known PPCDC enzymes. Consequently, investigation of Hal3 PD constructs should provide the clearest indication of which motifs are responsible for the lack of PPCDC activity seen in Hal3 HNC. Two hypotheses with regard to the lack of Hal3 HNC's activity were made. First, we proposed that the presence of the unique non-homologous insert in Hal3's PD could prevent catalysis by preventing access to the protein's active site. A construct, Hal3-PD HNC Δ312–350, was therefore prepared in which both the essential asparagine and cysteine residues were introduced and the non-homologous insert removed. Secondly, we considered the importance of the region surrounding the essential asparagine residue for PPCDC activity, since in all known PPCDC enzymes the asparagine forms part of a highly conserved PAMNX2M motif . In Hal3 and the related Vhs3 protein this motif is modified in three ways: first, the alanine residue is replaced by serine in both proteins; secondly, the essential asparagine residue is replaced by valine and glycine respectively; and thirdly, the terminal methionine residue is exchanged for threonine in both cases. Previous studies have suggested that this methionine is essential in AtHal3a  and the PPCDC-related enzyme EpiD , although a methionine mutant of the E. coli PPCDC is slightly active , and a similar mutation of the PPCDC in rice, OsHal3, was found to be active in a complementation assay . This suggests that the methionine residue may be structurally, although not necessarily catalytically, relevant for PPCDC function. To test the importance of the PAMNX2M motif with regard to Hal3's PPCDC functions, the mutations S170A and T175M were introduced into Hal3 PD HNC and Hal3 PD HNC Δ312–350 to give the constructs Hal PD HNRepC and Hal3 PD HNRepC Δ312–350 respectively, where the ‘NRep’ signifies the introduction of a PAMNX2M motif to the PD.
All of these constructs were subsequently expressed, purified and assayed for PPCDC activity using assays based both on the production of CO2 (the product of the first half-reaction catalysed by PPCDC) and product formation. Unfortunately, none of the described PD constructs showed any PPCDC activity, in spite of the fact that the Hal3 PD HNRepC Δ312–350 construct fully resembles known PPCDCs in all respects that have been shown to be relevant for activity (Figure 7A). To demonstrate that this lack of activity was not due to any of the introduced changes preventing the formation of catalytic trimers, the oligomeric state of all of the PD-based protein constructs was investigated by size-exclusion chromatography (Figure 7B). These analyses reveal that all of the studied constructs successfully form trimers. Interestingly, all proteins except those in which the non-homologous insert had been deleted also showed the formation of higher-molecular-mass complexes based on the presence of peaks that can be attributed to a hexameric species. This suggests that the presence of the insert may be necessary for the formation of such entities.
Investigation of the functionality and oligomerization ability of engineered PD constructs
The discovery that yeast Hal3 is a moonlighting protein, acting as an inhibitory subunit of the Ppz1 phosphatase  as well as playing a role in CoA biosynthesis by contributing to the heterotrimeric PPCDC enzyme in S. cerevisiae (and probably in other hemiascomycetes) , raises the question of the molecular basis for such disparate functions. Moreover, an investigation of the mechanism by which Hal3 regulates Ppz1 activity is especially relevant, since Hal3 appears devoid of the structural determinants known to govern binding and inhibition of PP1-related protein phosphatases [24,37]. The clarification of the basis for Ppz1 inhibition by Hal3 may therefore uncover novel phosphatase regulatory mechanisms. On the other hand, gaining full understanding of the PPCDC-catalysed reaction in yeast is also an important step in the quest to obtain a complete description of central metabolism in this important model eukaryotic organism. In the present study, we set out to determine the structural determinants for these unrelated functions in Hal3. Our results, summarized in Figure 8, indicate that these determinants are very different.
Schematic depiction of the functional domains of Hal3 summarizing the results generated in the present study in the form of SAR (structure–activity relationship) analysis
As far as the Ppz1-related function is concerned, we show that the NtD on its own does not bind or inhibit Ppz1 in vitro (Figures 4 and 5). Our data also confirm that Hal3's PD is necessary for binding to Ppz1 and inhibition of its phosphatase activity, as initially demonstrated by an approach based on random mutagenesis of this domain . This work revealed the Hal3 residues Val390, Ile446 and Trp452 as very important for binding to Ppz1, and residues Glu460 and Val462 as key for the inhibitory mechanism. Bioinformatic analysis of a number of hemiascomycetous yeasts carrying a single HAL3-like gene (i.e. lacking the HAL3 and VHS3 pair), such as Candida albicans or Kluyveromyces lactis, shows that all five of these residues are fully conserved or, if exchanged, the substitution is highly conservative. This suggests that the inhibitory mechanisms involving the PD is highly conserved among yeasts. However, the non-homologous insertion sequence in the PD apparently has no relevance in Ppz1 inhibition, as the Hal3 Δ312–350 construct shows effects similar to native Hal3. Moreover, the PD is not sufficient to provide full Hal3-mediated Ppz1 inhibition, as demonstrated by the observation that expression in a HAL3 wild-type background of those versions lacking the N- or C-terminal extensions yields a phenotype equivalent to that produced by the presence of an empty plasmid. However, these forms are not completely devoid of biological activity, since their expression in a hal3 background partly alleviates the defect associated with the absence of native Hal3 (Figures 2 and 3).
Remarkably, the expression of the PD in a wild-type strain in many cases leads to phenotypes completely contrary to those resulting from the expression of native Hal3 (i.e. it decreases salt tolerance, aggravates the growth defect of a sit4 strain and improves growth of a slt2 mutant under restrictive conditions, as shown in Figures 2 and 3). This can be explained if one considers that the PD alone is not able to inhibit Ppz1's phosphatase activity (Figure 5), but still retains the ability to interact with the Ppz1 protein. We have shown previously  and confirm in the present study that the interaction of native Hal3 with the catalytic domain of Ppz1 is more intense than with the entire phosphatase. In contrast, the interaction of the PD with the entire Ppz1 protein is stronger than that of native Hal3; in fact, the PD shows a preference for the intact phosphatase rather than for its C-terminal catalytic domain (Figure 5). On this basis, it can be assumed that expression of PD displaces native Hal3 from its union with Ppz1, thus resulting in phenotypes reminiscent of those observed upon deletion of HAL3. This would also explain why these phenotypes disappear when PD is expressed in a hal3 background. A similar situation was previously found to occur upon overexpression of Cab3, which binds to Ppz1 (albeit quite weakly) but is unable to inhibit its phosphatase activity .
The in vitro experiments (Figures 5 and 6) provide some additional insights into the structural determinants for Ppz1 inhibition and PPCDC activity. First, the results in Figure 6 show that the presence of the C-terminal tail is a major requirement for Ppz1 inhibition by Hal3. The relevance of the acidic C-terminal extension of Hal3 for inhibition of native Ppz1 activity suggests the possibility of a functional interaction between this domain and the N-terminal half of Ppz1 that would result in the exposure of the catalytic Ppz1 domain to the inhibitory residues located on Hal3. Moreover, the formation of such an interaction is supported by the opposing ionic natures of the two interacting domains: the C-terminal tail of Hal3 is highly acidic (the calculated pI of the terminal 66 residues is 3.04), whereas the N-terminal half of Ppz1 is strongly basic (the calculated pI of residues 1–339 of Ppz1 is 9.59). The finding that in all cases the CtD of Ppz1 is more intensely inhibited than the entire protein could be due to the N-terminal half of Ppz1 protecting the protein against excessive phosphatase inhibition by Hal3, a role that was previously proposed for this domain . This is clearly seen in the NtD+PD version, which acts as a rather poor inhibitor of the entire Ppz1 enzyme, although it has an inhibitory effect on Ppz1 Δ1–344 similar to that seen for the other constructs. The poor inhibition of Ppz1 would be explained by the inability of this version to form the stabilizing interaction described above, whereas its inhibition of Ppz1 Δ1–344 reiterates the finding that most of the inhibitory residues reside on the PD. Finally, it must be noted that although in Saccharomycotina species the corresponding Hal3 orthologues all possess an acidic CtD, this is not true for other fungi. In this regard, it is highly suggestive that in those species in which Hal3 lacks the acidic tail, the size of the Ppz1 N-terminal extension is clearly shorter and often less rich in basic residues. This correlation would further emphasize the functional link between these Ppz1 and Hal3 structural features. Secondly, the data summarized in Figure 6(A) confirm that, as expected, the presence of the PD is the minimum requirement for Hal3 to show in vitro PPCDC activity in combination with Cab3. The NtD shows no activity, and in combination with its observed lack of binding or inhibition of Ppz1, this result suggests that this domain has no inherent biological activity of its own. However, the NtD is clearly important in stabilizing the interactions between Hal3 and Cab3 in the heterotrimeric yeast PPCDC – much more so than the CtD – since among the shortened constructs only Hal3 NtD+PD and Hal3 Δ312–350 shows PPCDC activity approaching that of native Hal3 when combined with Cab3 (Figure 6A). The results indicate that neither the CtD nor the non-homologous insertion is directly involved in the PPCDC activity, or in the formation or stabilization of the heterotrimeric PPCDC complex.
Finally, in the present study, we also set out to determine whether Hal3's loss of the asparagine and cysteine residues normally required for PPCDC activity was relevant in the context of its role as Ppz1 inhibitor, and whether re-introduction of these residues at the appropriate positions would return PPCDC activity to Hal3 . Such a restoration of PPCDC activity was previously achieved in Cab3 by simply replacing the region surrounding its non-functional essential histidine residue with the corresponding sequence of residues from Hal3. Importantly, the results show that the Hal3 HVC and Hal3 HNC constructs do not show Ppz1-related behaviour significantly different from that of Hal3 (Figures 2–4). Although a previous study indicated that another mutation of the asparagine residue, namely the N466I mutant (i.e. Hal3 HVI according to the terminology used here) did affect Hal3's Ppz1-related functions to some extent , the data presented here specifically indicates that exchange of Asn466 for the catalytically required cysteine has no effect on Ppz1 inhibition. This suggests that the loss of the two essential PPCDC residues from Hal3/Vhs3-like proteins and the development of heterotrimeric PPCDCs in the hemiascomycetes were not driven by the acquisition of Ppz1 inhibitory activity. Even more remarkable is the finding that the Hal3 HNC construct, which contains all of the required PPCDC catalytic residues, is not able to act as a PPCDC on its own either in vivo or in vitro (Table 2 and Figure 6B). Additional attempts to introduce PPCDC activity to the PD, which focused on removal of the non-homologous insert in Hal3 and introduction of the PAMNX2M motif known to be functionally important, also failed (Figure 7A). In none of these cases could the lack of activity be related to the oligomerization status of the various protein constructs, as cross-linking studies or size-exclusion chromatography analyses clearly showed the formation of trimers in all cases (Figures 6C and 7B). Taken together, these results show that the simple re-introduction of the essential asparagine and cysteine residues, even when accompanied by additional supporting structural changes, is not sufficient to yield a monofunctional PPCDC. This suggests that the structural determinants required for full PPCDC activity is more complex than previously thought, supporting the notion that the divergence of Hal3/Vhs3-like proteins from monofunctional PPCDCs was also driven by other, as yet undetermined, factors.
In conclusion, the results of the present study (Figure 8) highlight the key role of both the NtDs and acidic CtDs of Hal3 in Ppz1 regulation, thereby providing a structure-based rationalization for the importance of the latter made in early reports [14,15]. Our data also confirm the requirement of Hal3's core PD for PPCDC activity to be exhibited in combination with Cab3, and unveils the relevant function of its NtD in stabilizing this interaction. Moreover, our results show that Hal3's loss of two essential residues associated with monofunctional PPCDC activity was not due to the acquisition of Ppz1-related functions, and that the structural divergence at the relevant positions at which these residues are expected is of such an extent that their simple re-introduction is an insufficient change to reactivate PPCDC activity in Hal3. This implies that the development of heterotrimeric PPCDCs in the hemiascomycetes that consist of separate Hal3/Vhs3 and Cab3-like proteins with complementing essential catalytic residues cannot solely be ascribed to the ascendance of the moonlighting ability of these proteins.
single overlap extension
J. Albert Abrie prepared most expression constructs and performed the expression and purification of the associated proteins, performed PPCDC assays and in vitro Ppz1 inhibition studies and carried out part of the trimerization analyses. Asier González carried out most in vivo experiments in yeast, performed the Ppz1-Hal3 binding analysis and developed part of the trimerization analyses. Erick Strauss and Joaquín Ariño jointly conceived the project, supervised its development and wrote the paper.
The excellent technical assistance of Montserrat Robledo and Anna Vilalta is gratefully acknowledged. We thank Amparo Ruiz for construction of strain MAR30 and Leisl Brand for initial cloning of the Hal3 PD domain.
This work was supported by the Ministry of Science and Innovation, Spain, and Fondo Europeo de Desarrollo Regional (FEDER) [grant numbers BFU2008-04188-C03-01 and BFU2011-30197-C3-01 (to J.A.)] and the National Research Foundation, South Africa [grant number FA2007041600013 (to E.S.)]. J.A. is the recipient of an ‘Ajut 2009SGR-1091’ and an Institució Catalana de Recerca i Estudis Avançats (ICREA) Academia 2009 Award (Generalitat de Catalunya). J.A. and E.S. were recipients of a Spain/South Africa Research Cooperation grant [grant numbers HS2007-0008 and HS2007-0022] from the Ministerio de Educación y Ciencia (Spain) and the National Research Foundation (South Africa). J.A.A. is a recipient of the Wilhelm Frank bursary (South Africa).
These authors contributed equally to this work.