Bacterial PerO Permeases Transport Sulfate and Related Oxyanions

ABSTRACT

Rhodobacter capsulatus synthesizes the high-affinity ABC transporters CysTWA and ModABC to specifically import the chemically related oxyanions sulfate and molybdate, respectively. In addition, R. capsulatus has the low-affinity permease PerO acting as a general oxyanion transporter, whose elimination increases tolerance to molybdate and tungstate. Although PerO-like permeases are widespread in bacteria, their function has not been examined in any other species to date. Here, we present evidence that PerO permeases from the alphaproteobacteria Agrobacterium tumefaciens, Dinoroseobacter shibae, Rhodobacter sphaeroides, and Sinorhizobium meliloti and the gammaproteobacterium Pseudomonas stutzeri functionally substitute for R. capsulatus PerO in sulfate uptake and sulfate-dependent growth, as shown by assimilation of radioactively labeled sulfate and heterologous complementation. Disruption of perO genes in A. tumefaciens, R. sphaeroides, and S. meliloti increased tolerance to tungstate and, in the case of R. sphaeroides, to molybdate, suggesting that heterometal oxyanions are common substrates of PerO permeases. This study supports the view that bacterial PerO permeases typically transport sulfate and related oxyanions and, hence, form a functionally conserved permease family.

IMPORTANCE Despite the widespread distribution of PerO-like permeases in bacteria, our knowledge about PerO function until now was limited to one species, Rhodobacter capsulatus. In this study, we showed that PerO proteins from diverse bacteria are functionally similar to the R. capsulatus prototype, suggesting that PerO permeases form a conserved family whose members transport sulfate and related oxyanions.

INTRODUCTION

The oxyanions sulfate (SO₄^2–) and molybdate (MoO₄^2–) are the preferred sources for sulfur and molybdenum, respectively, in bacteria. To specifically import these structurally similar oxyanions, Escherichia coli synthesizes the ATP-binding cassette (ABC) transporters CysPTWA and ModABC (1, 2). ABC uptake systems consist of a periplasmic substrate binding protein, a transmembrane channel protein, and a cytoplasmic ATP-binding protein, which provides the energy for active substrate transport across the cytoplasmic membrane against a concentration gradient (3). In addition to molybdate, ModABC imports tungstate (WO₄^2–) but not sulfate (4, 5). In addition to sulfate, CysPTWA imports molybdate, albeit less efficiently than its primary substrate sulfate (6, 7).

Recently, we described the permease PerO mediating transport of sulfate, molybdate, and tungstate in the photosynthetic purple nonsulfur bacterium Rhodobacter capsulatus (8). PerO belongs to the ArsB/NhaD ion transporter superfamily, whose members encompass 10 to 13 transmembrane domains and have diverse functions such as arsenite export, citrate/succinate antiport, and Na⁺/H⁺ antiport. PerO contains two central TrkA_C domains, making this permease considerably larger than the eponymous transporters ArsB and NhaD. TrkA_C domains have been implicated in the binding of unidentified ligands (9, 10). Biocomputational searches suggest that PerO-like proteins are widely distributed in bacteria and archaea and possibly form a subgroup of oxyanion-specific transporters (8). To date, however, R. capsulatus PerO is the only member of this proposed subgroup that has been functionally analyzed.

In the present study, we investigated the functions of PerO-like proteins in distantly related bacteria. We showed that perO genes from five selected species complement the sulfate uptake deficiency of a ΔcysTWA ΔperO double mutant of R. capsulatus. Furthermore, we disrupted perO genes in three of these species and showed enhanced tolerance to high concentrations of tungstate and/or molybdate, which were toxic for the corresponding wild-type strains. Together, these findings suggest that bacterial PerO proteins commonly transport sulfate and related oxyanions, as is the case for R. capsulatus PerO.

RESULTS AND DISCUSSION

Heterologous perO genes support sulfate-dependent growth of R. capsulatus.

R. capsulatus imports sulfate by two transporters, the ABC transporter CysTWA and the permease PerO (8). Accordingly, ΔcysTWA and ΔperO strains grew with 1 mM sulfate as the sole sulfur source, whereas growth of the cysTWA perO double mutant was strongly impaired (Fig. 1). Growth of the ΔcysTWA strain was somewhat delayed compared to the wild type and the ΔperO strain, indicating that PerO only partially compensates for lack of CysTWA. All strains grew comparably well with the alternative sulfur source taurine (data not shown), which is taken up by a distinct transporter, TauABC (11).

FIG 1.

Sulfate-dependent growth of R. capsulatus wild-type and mutant strains. R. capsulatus strains were grown in RCV minimal medium containing 1 mM MgSO₄ as a sulfur source. The R. capsulatus strains used were as follows: Rc_WT (1), wild-type strain B10S; Rc_ΔcysTWA (2), cysTWA mutant YP91; Rc_ΔperO (3), perO mutant YP152; Rc_ΔcysTWA-ΔperO (4), cysTWA perO double mutant YP91-YP152; Rc_ΔΔ + vector (5), YP91-YP152 carrying pMT1; Rc_ΔΔ + Ds_perO (6), YP91-YP152 carrying pMT48 (D. shibae perO); Rc_ΔΔ + Rs_perO1 (7), YP91-YP152 carrying pMT42 (R. sphaeroides perO1); Rc_ΔΔ + Rs_perO2 (8), YP91-YP152 carrying pMT63 (R. sphaeroides perO2); Rc_ΔΔ + Sm_perO1 (9), YP91-YP152 carrying pMT24 (S. meliloti perO1); Rc_ΔΔ + Sm_perO2 (10), YP91-YP152 carrying pMT22 (S. meliloti perO2); Rc_ΔΔ + At_perO (11), YP91-YP152 carrying pMT83 (A. tumefaciens perO); and Rc_ΔΔ + Ps_perO (12), YP91-YP152 carrying pMT25 (P. stutzeri perO). The upper part of the figure shows growth curves (based on duplicate cultures) over 4 days of incubation under phototrophic conditions. The lower part of the figure shows representative test tubes at day 4 for each of the 12 strains tested.

To test whether PerO-like proteins from different species commonly function in sulfate uptake, we examined sulfate-dependent growth of the R. capsulatus ΔcysTWA ΔperO strain carrying perO genes from the alphaproteobacteria Rhodobacter sphaeroides, Sinorhizobium meliloti, Agrobacterium tumefaciens, and Dinoroseobacter shibae and the gammaproteobacterium Pseudomonas stutzeri (Fig. 1). R. sphaeroides is a photosynthetic purple bacterium closely related to R. capsulatus (12). S. meliloti is a symbiotic soil bacterium that induces root nodule formation in legume host plants and provides its host with fixed nitrogen (13). A. tumefaciens is a plant-pathogenic soil bacterium that induces tumor formation in a large variety of host plants by transferring part of its tumor-inducing plasmid into host cells (14). D. shibae is a marine algal symbiont that provides its host with vitamins (15). P. stutzeri is a nutritionally versatile strain that occupies diverse environments and can act as an opportunistic pathogen in humans (16). R. sphaeroides and S. meliloti have two perO-like genes each, whereas A. tumefaciens, D. shibae, and P. stutzeri have single perO homologs.

The heterologous perO genes were cloned into a broad-host-range plasmid downstream of a gentamicin cassette to drive perO expression in the R. capsulatus ΔcysTWA ΔperO strain (see Materials and Methods; Table 1). For reasons unknown, all attempts to clone the full-length R. capsulatus perO gene failed, although we tried perO fragments with different flanking regions, medium-copy-number and low-copy-number vector plasmids, and different vector insertion sites. All but one of the heterologous perO genes analyzed in this study supported sulfate-dependent growth of the R. capsulatus ΔcysTWA ΔperO strain (Fig. 1), suggesting that sulfate is a common substrate of bacterial PerO permeases. The only exception was the S. meliloti perO2 gene (for possible explanations, see below).

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Relevant characteristic(s) or descriptiona	Source or reference(s)
Strains
Rhodobacter capsulatus
B10S	Wild type	35
YP91	Polar cysT::Gm mutant of B10S (ΔcysTWA)	8
YP152	Markerless perO mutant of B10S (ΔperO)	This study
YP91-YP152	ΔcysTWA ΔperO	This study
Rhodobacter sphaeroides
2.4.1	Wild type	12, 36
MT32	perO1::Km mutant of 2.4.1 (ΔperO1)	This study
MT59	perO2::Gm mutant of 2.4.1 (ΔperO2)	This study
MT32-MT59	ΔperO1 ΔperO2	This study
Sinorhizobium meliloti
1021	Wild type	13
MT23	perO1::Km mutant of 1021 (ΔperO1)	This study
MT53	perO2::Gm mutant of 1021 (ΔperO2)	This study
MT23-MT53	ΔperO1 ΔperO2	This study
Agrobacterium tumefaciens
C58	Wild type	14, 37
MH25	perO::Km mutant of C58 (ΔperO)	This study
Plasmids
pBBR1MCS-2	Mobilizable broad-host-range vector	38
pMT1	pBBR1MCS-2 derivative carrying Gm	This study
pMT22	pMT1 derivative carrying S. meliloti perO2	This study
pMT24	pMT1 derivative carrying S. meliloti perO1	This study
pMT25	pMT1 derivative carrying Pseudomonas stutzeri perO	This study
pMT42	pMT1 derivative carrying R. sphaeroides perO1	This study
pMT48	pMT1 derivative carrying Dinoroseobacter shibae perO	This study
pMT63	pMT1 derivative carrying R. sphaeroides perO2	This study
pMT83	pMT1 derivative carrying A. tumefaciens perO	This study

^aGene locus identifiers: R. capsulatus cysT, Rcc02536; R. capsulatus perO, Rcc00882; R. sphaeroides perO1, RSP_0656; R. sphaeroides perO2, RSP_3330; S. meliloti perO1, SMc04179; S. meliloti perO2, SMa1916; A. tumefaciens perO, Atu4475; P. stutzeri perO, PST_0569; and D. shibae perO, Dshi_2766. Antibiotic resistance cassettes: Gm, gentamicin; Km, kanamycin.

Heterologous perO genes mediate sulfate assimilation in R. capsulatus.

To further characterize sulfate uptake by heterologous PerO proteins in R. capsulatus, we examined the incorporation of radioactively labeled sulfate into protein (see Materials and Methods). The R. capsulatus ΔperO and ΔcysTWA strains assimilated [³⁵S]sulfate up to 86% and 6% of the wild-type level, respectively (Fig. 2), indicating the particular importance of CysTWA in sulfate uptake in R. capsulatus. The cysTWA perO double mutant, which was strongly affected in sulfate-dependent growth, assimilated sulfate to only 2% of the wild-type value.

FIG 2.

Sulfate uptake by R. capsulatus wild-type and mutant strains. To estimate sulfate uptake by R. capsulatus wild-type and mutant strains, incorporation of [³⁵S]sulfate in protein was determined (see Materials and Methods). The R. capsulatus strains used and their numbering (1 to 12) are the same as described for Fig. 1. The dashed line indicates the lowest level of sulfate assimilation still supporting significant growth with sulfate as sole sulfur source (Fig. 1). The results are the means and standard deviations from three independent measurements.

Sulfate assimilation by R. capsulatus ΔcysTWA ΔperO strains harboring heterologous perO genes differed greatly (Fig. 2). Sulfate uptakes mediated by the perO genes from D. shibae and P. stutzeri were within the same order of magnitude as those observed for the R. capsulatus wild-type strain, whereas uptakes mediated by the other perO genes were considerably lower. Despite rapid sulfate uptake by the R. capsulatus ΔcysTWA ΔperO strain complemented with the perO genes from D. shibae and P. stutzeri, these strains showed the same lag in sulfate-dependent growth as strains that transported sulfate poorly (Fig. 1). At present, we can only speculate that activity of PerO from D. shibae and P. stutzeri is downregulated during growth by increasing cellular sulfate concentrations. Since the S. meliloti genes perO1 and perO2 differed in their ability to restore sulfate uptake in the R. capsulatus ΔcysTWA ΔperO strain, we looked for structural differences between the two permeases. Comparison of deduced protein sequences, however, revealed remarkably high overall similarity between S. meliloti PerO1 and PerO2 (56% identity, 72% similarity, no gaps). Nevertheless, the central TrkA_C domains are less well conserved than the flanking transmembrane domains, and hence, one might speculate that these domains determine the differences in sulfate uptake by PerO1 and PerO2. We aligned the sequences of all PerO proteins but could not identify conserved amino acids distinguishing the TrkA_C domains of S. meliloti PerO2 from those of other PerO proteins. The TrkA_C domains of the PerO-like sulfur acclimation protein 1 (SAC1) from the green alga Chlamydomonas reinhardtii were predicted to form a large intracellular loop; however, SAC1 regulates the sulfur-deprivation response in C. reinhardtii rather than acting as a sulfate transporter (17). If the TrkA_C domains of bacterial PerO proteins form cytoplasmic loops, too, these domains would not serve as initial sulfate binding sites in the periplasm.

Although all perO expression plasmids were constructed according to the same scheme (see Materials and Methods), expression levels may well differ between the respective perO genes. Hence, differences in perO-based sulfate assimilation do not necessarily reflect differences in the transport efficiency of the heterologous permeases but may rather reflect differences in gene expression in the heterologous host R. capsulatus. Cumulatively, our experiments corroborated that sulfate is a common substrate of bacterial PerO permeases. Apart from that, PerO permeases may have a different impact on sulfate uptake in their homologous hosts.

Bacterial PerO permeases mediate uptake of heterometal oxyanions.

High concentrations of molybdate (MoO₄^2–) and tungstate (WO₄^2–) are toxic to cells, because these heterometal oxyanions (HMOs) compete with sulfate (SO₄^2–) as substrates of ATP sulfurylase, thus catalyzing futile ATP-depleting cycles (8, 18, 19). In R. capsulatus, tolerance to molybdate and tungstate is increased by disruption of the perO permease gene, indicating that these HMOs are PerO substrates (8).

Here we investigated the effects of perO disruption on HMO tolerance in R. sphaeroides, S. meliloti, and A. tumefaciens. As mentioned above, R. sphaeroides and S. meliloti have two perO genes each, whereas A. tumefaciens contains only one perO homolog. Single and double perO mutants were generated as described in Materials and Methods (Table 1). Serial dilutions of wild-type and mutant cultures were spotted onto plates containing molybdate or tungstate (Fig. 3) as described earlier (8).

FIG 3.

Growth of bacterial perO mutants at high concentrations of molybdate and tungstate. Serial (1:5) dilutions of bacterial cultures were spotted on nutrient agar plates with the indicated molybdate and tungstate concentrations. R. capsulatus and R. sphaeroides strains were grown on PY plates, S. meliloti strains were grown on TY plates, and A. tumefaciens strains were grown on LB plates (see Materials and Methods). Plates without added molybdate or tungstate served as controls. The bacterial strains used were as follows: Rc_WT, R. capsulatus wild-type strain B10S; Rc_ΔperO, perO mutant YP152; Rs_WT, R. sphaeroides wild-type strain 2.4.1; Rs_ΔperO1, perO1 mutant MT32; Rs_ ΔperO2, perO2 mutant MT59; Rs_ΔperO1-ΔperO2, perO1 perO2 double mutant MT32-MT59; Sm_WT, S. meliloti wild-type strain 1021; Sm_ΔperO1, perO1 mutant MT23; Sm_ ΔperO2, perO2 mutant MT53; Sm_ΔperO1-ΔperO2, perO1 perO2 double mutant MT23-MT53; At_WT, A. tumefaciens wild-type strain C58; At_ΔperO, perO mutant MH25. Growth of A. tumefaciens was documented after 2 days of incubation, whereas growth of the other strains was documented after 3 days.

All perO mutants grew on control plates (with no heterometal addition) like their parental strains, ruling out general defects of ΔperO strains in viability or fitness (Fig. 3). Hence, PerO permeases are dispensable in R. sphaeroides, S. meliloti, and A. tumefaciens as is the case for R. capsulatus (8). Strikingly, R. sphaeroides, S. meliloti, and A. tumefaciens grew well at 100 mM molybdate and 100 mM tungstate (data not shown) and therefore tolerated much higher HMO concentrations than did R. capsulatus (Fig. 3). Since R. capsulatus and R. sphaeroides were grown on the same medium, peptone-yeast extract (PY), we conclude that differences in natural HMO tolerance are strain specific rather than medium specific at least for the two Rhodobacter strains. In other words, even closely related strains like R. capsulatus and R. sphaeroides substantially differ in HMO susceptibility. The molecular mechanism(s) underlying these differences remains unknown to date. R. sphaeroides, S. meliloti, and A. tumefaciens possibly produce siderophores that bind and thus detoxify excess HMO in the medium, as shown for Azotobacter vinelandii (20). It is noteworthy that none of these strains has a homolog of the A. vinelandii Mo/W storage protein MosAB, which binds up to 100 Mo or W atoms as polyoxometalates (21, 22). The two Rhodobacter strains have Mop-type Mo storage proteins (23), which are absent in S. meliloti and A. tumefaciens. Taken together, there is no correlation between HMO susceptibility and the presence of storage proteins.

The R. capsulatus ΔperO strain tolerated 10 mM molybdate and 5 mM tungstate, concentrations otherwise toxic for the wild type (Fig. 3) (8), suggesting that PerO is the main HMO importer in this species. In addition to PerO (Rcc00882), R. capsulatus is capable of synthesizing a second PerO-like protein (Rcc02286), which exhibits only low similarity to PerO (27% identity, 49% similarity, 2% gaps) and apparently does not substitute for PerO in HMO uptake. In contrast, R. sphaeroides PerO1 and PerO2 are highly similar to each other (65% identity, 81% similarity, no gaps). The disruption of either perO1 or perO2 in R. sphaeroides had only moderate effects on molybdate and tungstate susceptibility, whereas the double mutant was more tolerant than the wild type (Fig. 3). This suggests that R. sphaeroides PerO1 and PerO2 are functionally redundant and able to substitute for each other in HMO uptake, well in line with the high degree of similarity between the two permeases.

The wild-type strains of S. meliloti and A. tumefaciens exhibited residual growth at 300 and 200 mM molybdate, respectively (Fig. 3). In contrast to the situation in R. capsulatus and R. sphaeroides, however, disruption of the perO genes in S. meliloti and A. tumefaciens did not increase molybdate tolerance, suggesting that PerO permeases do not play major roles in molybdate uptake in these bacteria. The S. meliloti ΔperO1 strain exhibited clear tungstate tolerance, whereas the ΔperO2 strain was as sensitive as the wild type, indicating that PerO2 does not substitute for PerO1 in tungstate transport. Like S. meliloti PerO1, A. tumefaciens PerO appeared to import tungstate.

Conclusions.

PerO proteins belong to the ArsB/NhaD superfamily of transporters but differ structurally from other members of this family by having two central TrkA_C domains (8). Despite the widespread distribution of PerO proteins in bacteria, only R. capsulatus PerO has been experimentally characterized to date (8). In this study, we provided evidence that PerO permeases from distantly related bacteria transport sulfate, tungstate, and/or molybdate like R. capsulatus PerO. Hence, structural conservation correlates well with functional conservation, supporting the view that PerO permeases form a subclass of ArsB/NhaD transporters specific for sulfate and related oxyanions.

With the exception of S. meliloti perO2, all of the other heterologous perO genes tested in this study mediated sulfate uptake in R. capsulatus, indicating that import of sulfate is the major physiological function of PerO permeases. This may be especially relevant at high sulfate concentrations inhibiting the high-affinity sulfate transporter CysTWA (1, 24).

There are no known tungstoenzymes in R. capsulatus, R. sphaeroides, S. meliloti, and A. tumefaciens, suggesting that PerO-mediated tungstate uptake has no physiological relevance in these bacteria. In contrast, these bacteria synthesize different molybdoenzymes (12–14, 25, 26). Since molybdate represses modABC transcription in these bacteria via ModE proteins and hence limits synthesis of the high-affinity transporter ModABC to molybdate-limiting conditions (1, 25, 27, 28), PerO-mediated molybdate uptake may well be relevant at high molybdate concentrations. Indeed, PerO supports molybdenum nitrogenase activity in R. capsulatus at high micromolar molybdate concentrations (8).

Finally and most importantly, PerO-like proteins are often misannotated as citrate, di/tricarboxylate, arsenite, or potassium transporters in the databases. This study, however, suggests that PerO permeases instead transport sulfate and chemically related oxyanions.

MATERIALS AND METHODS

Strains, plasmids, and growth conditions.

The bacterial strains and plasmids used in this study are listed in Table 1. The following rich media were used: peptone-yeast extract (PY) medium for Rhodobacter capsulatus and Rhodobacter sphaeroides, tryptone-yeast extract (TY) medium for Sinorhizobium meliloti, and Luria broth (LB) medium for Agrobacterium tumefaciens (29, 30). In addition, chemically defined minimal Rhodobacter capsulatus V (RCV) medium was used for R. capsulatus (30). When required, 1 mM MgSO₄ or 1 mM taurine was added as a sulfur source.

Construction of perO mutant strains in selected alphaproteobacteria.

Knockout strains of R. capsulatus, R. sphaeroides, S. meliloti, and A. tumefaciens were constructed by standard procedures making use of mobilizable narrow-host-range suicide plasmids, which replicate in Escherichia coli but not in alphaproteobacteria (8, 25, 31). PCR-amplified perO genes were cloned into suicide vectors prior to insertion of kanamycin (Km) and gentamicin (Gm) cassettes into appropriate restriction sites within the perO genes. In detail, the following restriction sites were used to disrupt the respective perO genes given in parentheses: SalI (R. sphaeroides perO1, RSP_0656), PstI (R. sphaeroides perO2, RSP_3330), SphI (S. meliloti perO1, SMc04179), EcoRI (S. meliloti perO2, SMa1916), and BclI (A. tumefaciens perO, Atu4475). The resulting mutagenesis plasmids were transferred by biparental mating into R. sphaeroides and by electroporation into S. meliloti and A. tumefaciens. Knockout strains were identified by selection for Km or Gm resistance and loss of the vector-encoded resistance indicating marker rescue by double recombination events. The resulting knockout strains were verified by PCR (Table 1; data not shown). Using a similar procedure, a 252-bp BclI fragment from the R. capsulatus perO gene (Rcc00882) was replaced with a lacZ-Km cassette, resulting in a strain producing red-blue colonies on X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) plates. Subsequently, a suicide plasmid containing the perO gene but lacking the BclI fragment was introduced into the perO::lacZ-Km strain. Screening for red colonies on X-Gal plates identified the markerless knockout strain YP152 carrying an in-frame perO deletion (Table 1). Finally, mutagenesis plasmid pYP91 carrying cysT::Gm (8) was used to generate the R. capsulatus double mutant YP91-YP152 lacking both the high-affinity sulfate transporter CysTWA and PerO (Table 1).

Construction of broad-host-range plasmids carrying selected perO genes.

The following perO genes were cloned into the broad-host-range plasmid pBBR1-MCS: Dinoroseobacter shibae perO (Dshi_2766), R. sphaeroides perO1 (RSP_0656) and perO2 (RSP_3330), S. meliloti perO1 (SMc04179) and perO2 (SMa1916), A. tumefaciens perO (Atu4475), and Pseudomonas stutzeri perO (PST_0569). For this purpose, the respective coding regions and 21 to 24 nucleotides of the upstream regions containing the ribosomal binding sites were PCR amplified. In each case, the upper PCR primer started with an XbaI recognition site, which was subsequently used for insertion of a Gm cassette to drive perO gene expression. In previous studies, we used the Gm cassette to drive transcription of R. capsulatus nitrogen fixation genes, which require strong promoters (32, 33). The perO expression plasmids are listed in Table 1.

Sulfate assimilation by R. capsulatus ΔcysTWA ΔperO strains carrying heterologous perO genes.

Incorporation of radioactively labeled sulfate-sulfur into protein was determined essentially as described previously (34). Precultures of R. capsulatus strains were grown in RCV minimal medium with taurine-sulfur until logarithmic growth phase. After cells were washed with sulfate-free RCV medium, approximately 10⁸ cells were resuspended in 500 μl sulfate-free RCV medium, before 1 μCi [³⁵S]sodium sulfate (Hartmann Analytic, Braunschweig, Germany) was added. After 1 h of incubation, chloramphenicol was added to stop further translation. Cells were collected by centrifugation and washed with Tris-EDTA buffer (100 mM Tris, 1 mM EDTA, pH 7.5). Subsequently, cells were resuspended in 400 μl lysis buffer (10 mM Tris, 1.4 μM phenylmethylsulfonyl fluoride, pH 7.5) prior to disruption by ultrasonication in a VialTweeter UIS250V (Hielscher, Teltow, Germany). Aliquots of cell extracts were spotted onto 0.34-mm Whatman blotting paper (Roth, Karlsruhe, Germany), and proteins were precipitated with 20% trichloroacetic acid (TCA). Filter papers were washed twice with 10% TCA and once with 96% ethanol before the amounts of incorporated [³⁵S]sulfur were measured in Ultima Gold liquid scintillation cocktail (PerkinElmer, Rodgau, Germany) in a Tri-Carb 2800TR liquid scintillation analyzer (PerkinElmer).