A key developmental regulator controls the synthesis of the antibiotic erythromycin in Saccharopolyspora erythraea

Abstract

Saccharopolyspora erythraea makes erythromycin, an antibiotic commonly used in human medicine. Unusually, the erythromycin biosynthetic (ery) cluster lacks a pathway-specific regulatory gene. We isolated a transcriptional regulator of the ery biosynthetic genes from S. erythraea and found that this protein appears to directly link morphological changes caused by impending starvation to the synthesis of a molecule that kills other bacteria, i.e., erythromycin. DNA binding assays, liquid and affinity chromatography, MALDI-MS analysis, and de novo sequencing identified this protein (M_r = 18 kDa) as the S. erythraea ortholog of BldD, a key regulator of development in Streptomyces coelicolor. Recombinant S. erythraea BldD bound to all five regions containing promoters in the ery cluster as well as to its own promoter, the latter with an order-of-magnitude stronger than to the ery promoters. Deletion of bldD in S. erythraea decreased the erythromycin titer in a liquid culture 7-fold and blocked differentiation on a solid medium. Moreover, an industrial strain of S. erythraea with a higher titer of erythromycin expressed more BldD than a wild-type strain during erythromycin synthesis. Together, these results suggest that BldD concurrently regulates the synthesis of erythromycin and morphological differentiation. The ery genes are the first direct targets of a BldD ortholog to be identified that are positively regulated.

Erythromycin, an antibiotic made by Saccharopolyspora erythraea, kills Gram-positive bacteria that infect humans. The industrial importance of erythromycin is significant, with worldwide sales of erythromycin and its derivatives reaching billions of dollars every year. Two strategies, classical strain improvement and rational engineering, have boosted titers of erythromycin from S. erythraea. In classical strain improvement, a strain undergoes multiple rounds of random mutagenesis and screening. Although tedious and time-consuming, the strategy can increase titers 10- to 100-fold (1). In rational engineering, knowledge of biochemical pathways guides genetic manipulations to increase titers, such as through modification of the methylmalonyl-CoA metabolite node (2). Often, overexpression of a gene that regulates the biosynthetic cluster of an antibiotic increases titers. For instance, overexpression of sanG, a transcriptional regulator of the nikkomycin biosynthetic genes in Streptomyces ansochromogenes, increased titers of the antibiotic (3). Normally, such regulatory genes, like actII-ORF4 for actinorhodin (4) or redD for undecylprodigiosin (5) in Streptomyces coelicolor, lie within the antibiotic biosynthetic cluster. However, in S. erythraea, the biosynthetic cluster for erythromycin lacks a regulatory gene.

In actinomycetes, synthesis of antibiotics often coincides with complex morphological changes during the life cycle. Compounds produced during these stages of differentiation comprise nearly two-thirds of bioactive molecules synthesized by microorganisms, including antibiotics, antitumor agents, and immunosuppressants. S. coelicolor, among the most well studied of actinomycetes, produces several antibiotics and has a sequenced genome (6). Apart from the pathway-specific regulators for antibiotic biosynthesis, other genes of S. coelicolor have pleiotropic roles in development and the synthesis of antibiotics. For example, mutation or deletion of bldD causes defects both in the formation of aerial hyphae and the production of antibiotics (7). To our knowledge, none of the genes with characterized roles in the development of S. coelicolor directly regulate an antibiotic biosynthetic gene cluster.

Previously, DNA microarrays revealed differences in transcription between two S. erythraea strains, a wild-type (WT) strain and a classically improved strain that overproduces erythromycin (OVP). In particular, the OVP strain expressed the entire biosynthetic cluster of erythromycin (ery) several days longer than the WT strain (8). The altered and coordinated expression of nearly all of the ery genes suggested the existence of a common regulator for the ery cluster, although the genome sequence of S. erythraea confirmed the absence of such a regulator within the cluster (9). In this study, we purified and identified a regulatory protein, BldD, which binds to all of the promoters in the ery cluster. This work reveals multiple cellular roles of BldD and makes progress toward the rational manipulation of actinomycetes to overproduce antibiotics.

Results and Discussion

Protein Binding to the Promoters of the Erythromycin Biosynthetic Gene Cluster.

Sustained, coordinate expression of the biosynthetic gene cluster of erythromycin by the OVP strain indicated the existence of a global regulator of these genes (8). To begin to identify the protein, we prepared fluorescent DNA probes of the five regions containing ery promoters (Fig. 1A) (10, 11). When incubated with crude lysates of both WT and OVP strains after 22 and 65 h of growth in liquid medium, all five promoter probes showed shifts in electrophoretic mobility shift assays (EMSAs), with eryAI-BIV and eryBVI probes producing the strongest shifts with the 22 h lysates (Fig. 1B, lanes c and e). For both WT and OVP strains, there was no shift observed after 65 h of growth (Fig. 1B, lanes d and f). The eryK and eryCI-ermE probes exhibited only modest shifts, whereas a smeared shift was observed for the eryBI-BIII probe with all lysates tested [supporting information (SI) Fig. S1]. Moreover, we obtained similar EMSA results by using DNA probes prepared from genomic DNA of both WT and OVP strains (data not shown).

Fig. 1.

EMSAs with promoters of the ery cluster and lysates of S. erythraea. (A) The biosynthetic gene cluster of erythromycin (ery). Gray lines show regions that contain promoters and were used as probes for EMSAs. For eryBVI and eryK, probes included the start site of transcription and the start codon. For divergent promoters (eryCI-ermE, eryBI-BIII, and eryAI-BIV), probes included start sites of transcription and start codons of the divergent genes. (B) EMSAs for promoters eryAI-BIV and eryBVI. Each probe has lanes b–f. Lanes show DNA ladder (a), probe only (b), probe + WT 22-hr lysate (c), probe + WT 65-hr lysate (d), probe + overproducer 22-hr lysate (e), and probe + overproducer 65-hr lysate (f). (C) EMSAs with the eryBVI probe and lysates of the WT and OVP strains at 12, 24, 30, and 48 h. Results are a representative example of three independent experiments.

Several tests indicated that the protein(s) causing the shifts bound to ery promoters specifically. In competition with the eryBVI probe, excess unrelated DNA (a fragment of a plasmid that contains an oriT site) did not affect the shift of the probe, but excess unlabeled eryBVI probe effectively competed with the labeled probe for binding to BldD (data not shown). In addition, incubation of a crude lysate of S. erythraea with a promoter from S. coelicolor (actVI-Ap) produced no shift (data not shown). Together, these initial results suggested that the protein(s) causing the EMSA shifts bind specifically to promoters of the ery cluster.

Presence of BldD Correlates with Expression of the ery Cluster.

A time course revealed more precisely when the WT and OVP strains produce the protein(s) causing the EMSA shifts in vegetative cultures. The two strains had similar growth curves in a liquid, rich medium (Fig. S2). In EMSAs with the eryBVI probe, lysates from both strains produced a shift at 12 h, the earliest time point (Fig. 1C). For the WT strain, the shift was observed until 24 h. In contrast, for the OVP strain, the shift remained detectable at 30 h (Fig. 1C), roughly correlating with the lengthened expression of the ery cluster (8).

Regulator Is a BldD Ortholog.

A protein designated BldD (see above) was purified based on its ability to cause an EMSA shift by using the eryBVI probe, and its structural gene was cloned as described in Materials and Methods. The bldD gene encodes a protein of 162 aa and molecular mass of 17.7 kDa (Fig. S3). The predicted BldD sequence yielded strong matches to the BldD homologs of several actinomycetes. For instance, BldD is 77% identical to the BldD of Streptomyces coelicolor. This finding was unexpected because, although mutations in S. coelicolor bldD influence the synthesis of secondary metabolites (12), we found no published evidence indicating that a member of the BldD family directly regulates antibiotic biosynthetic genes. In S. coelicolor, BldD negatively regulates expression of the developmental σ factor genes bldN and whiG during vegetative growth (13), in addition to repressing vegetative expression of the stress response σ factor gene sigH (14) and an as-yet uncharacterized regulator termed bdtA (13). It also negatively regulates its own synthesis (15). Thus, all of the BldD targets so far identified in S. coelicolor are repressed by BldD, whereas the ery genes are positively regulated by BldD in S. erythraea.

BldD Resides Apart from the ery Cluster in the Circular Chromosome of S. erythraea.

In the 8.2-Mb S. erythraea chromosome, the ery cluster is centrally located within the core region, which encodes most of the essential genes (9), whereas bldD lies near the edge of the core region, ≈1.5 Mb from the ery cluster (Fig. 2A). The separate positions of bldD and the ery cluster in the chromosome of S. erythraea contrast with many biosynthetic clusters of antibiotics such as actinorhodin (4) and undecylprodigiosin (5) that contain regulatory genes. Analysis of sequences for Aeromicrobium erythreum, which also synthesizes erythromycin, and Micromonospora megalomicea, which synthesizes the related molecule megalomicin, suggests that the ery cluster once might have contained a regulatory gene. Both species have biosynthetic gene clusters extremely similar to the ery cluster in S. erythraea but contain putative regulatory genes. In A. erythreum, a transcriptional regulator of the MarR family (ery-ORF25) is encoded at one end of the ery cluster adjacent to eryCI (16), and in M. megalomicea, a putative regulator is encoded at the end adjacent to megDVI (17).

Fig. 2.

Chromosomal locus of bldD. (A) Location of bldD in the chromosome of S. erythraea. Gray bar indicates the core region. (B) Alignment of the locus of bldD (S. ery) to the locus of bldD in S. coelicolor (S. coe). Black arrows show bldD (S. ery) and bldD (S. coe). Alignments represent genes homologous between the two bacteria. The dotted line represents a discontinuity of one gene in S. coelicolor in the alignment. The numbers above the genes are gene numbers in the genome sequence for both species.

The gene organization around the bldD gene is very similar to that of bldD in S. coelicolor (Fig. 2B). In particular, the three genes downstream of bldD (SACE_2074–2076) resemble three genes downstream of S. coelicolor bldD (SCO1492–1490), and the six genes upstream of bldD (SACE_2078–2084) resemble six genes upstream of S. coelicolor bldD (SCO1488–1483 and SCO1481). Combined with the bald phenotype of the bldD mutant (see above), this synteny reinforces the idea that S. erythraea BldD and S. coelicolor BldD are orthologous.

Recombinant BldD Binds to ery Promoters and to Its Own Promoter.

To confirm that bldD encodes a DNA binding protein, we expressed recombinant BldD in Escherichia coli BL21(DE3). In EMSAs, a lysate of E. coli induced to express BldD shifted the eryBVI probe completely, whereas a lysate of an uninduced strain caused no shift (data not shown). Further, a 56-bp footprint of BldD binding at the eryBVIp region was obtained (Fig. 3 and Fig. S4). The footprint includes the transcriptional start site of eryBVIp (11), and although it is an unusual binding location for an activator, it has been observed before (18). By visual inspection of the sequence within this protected region, we identified a possible binding sequence of AGTGC(n)₉TCGAC for BldD, based on the S. coelicolor BldD consensus binding sequence of AGTgA(n)_mTCACc (13). Consistent with our data, S. coelicolor BldD binds upstream or across the transcriptional start sites of its targets (13–15). However, S. coelicolor BldD acts as a transcriptional repressor of each of these target genes (12–14), whereas S. erythraea BldD appears to be acting as a transcriptional activator of the ery gene cluster. The mechanism by which BldD can act as a transcriptional activator in some cases and a transcriptional repressor in others is unclear and warrants further investigation.

Fig. 3.

Identification of the sequence of the BldD-protected regions of the eryBVI promoter by DNase I protection footprinting. (A) Electrophoregram for DNase I (0.1U) digest of the eryBVI probe without BldD. (B) Electrophoregram for DNase I (0.1U) digest of the eryBVI probe after incubation with 54 μM recombinant BldD. The boxed region indicates a region protected by BldD. (C) Underlined sequence indicates footprint. Gray nucleotides show putative BldD binding sites. Asterisk denotes start sites of transcription for eryBVI, and bracketed sequences indicate putative −10 and −35 promoter sequences (11).

BldD shifted all five probes of the ery promoters (Fig. 4), indicating that BldD regulates the entire biosynthetic cluster of erythromycin. To assess the affinity of BldD for each probe, we purified recombinant BldD by using a Ni-NTA resin (data not shown). Titrations of BldD with a fixed concentration of each probe, followed by measurements of the fraction of probe bound in EMSAs, yielded equilibrium dissociation constants (K_d) (Table 1 and Fig. S5). BldD binds with similar affinities to the probes of eryAI-BIV, eryBVI, and eryBI-BIII, and ≈3- to 5-fold less strongly to the probes of eryK and eryCI-ermE.

Fig. 4.

EMSAs with recombinant BldD and the five regions of ery promoters, designated eryCI-ermE, eryBI-BIII, eryAI-BIV, eryBVI, and eryK. ⁻, probe only; +, probe and BldD.

Table 1.

Dissociation constants for BldD binding its own promoter and the five regions with promoters of the ery cluster

Promoter	Kd, μM
bldD	0.32 ± 0.024
eryBVI	3.04 ± 0.47
eryAI-BIV	3.95 ± 0.61
eryBI-BIII	6.03 ± 1.06
eryCI-ermE	17.9 ± 3.78
eryK	15.7 ± 1.98

Because S. coelicolor BldD binds to its own promoter (15), we asked whether S. erythraea BldD bound to a sequence upstream of bldD. EMSAs with the S. erythraea bldD promoter and purified recombinant BldD resulted in two shifted fragments (Fig. 5A). Both shifts required the presence of BldD, but we were unable to determine whether they arose from different conformations or multimers of BldD. Note that S. coelicolor BldD binds to its own promoter as a dimer, also resulting in two distinct bands in EMSA experiments (7). In addition, K_d of 0.32 μM for BldD binding to its own promoter was determined, which approximates the K_d of BldD for its own promoter (19). Therefore, BldD binds to the bldD promoter an order of magnitude more strongly than it does to promoters of the ery cluster.

Fig. 5.

EMSAs with the bldD promoter and purified, recombinant BldD (A) and lysates of the WT and OVP strains of S. erythraea (B). See text for details. (C) Western blot of lysates (4 μg of total protein) of the WT and OVP strains with polyclonal antibody for BldD. As a positive control, the far right lane shows a Western blot of purified, recombinant BldD. Results are a representative example of three independent experiments. (D) Titers of erythromycin of the WT (black diamonds) and OVP (open squares) strains grown in a liquid, rich medium.

S. erythraea Overproducer Strain Has More BldD.

Because BldD binds to its own promoter with submicromolar K_d, we used the bldD promoter to examine how BldD binding varied with a time course of lysates from S. erythraea. Cultures of the WT and OVP strains in a liquid, rich medium were sampled at 27, 40, 62, 87, and 111 h. EMSAs using lysates from both strains revealed two shifted probe fragments at 27 h, the initial time point (Fig. 5B). By 40 h, the WT strain lysate failed to produce most of the upper band, whereas the OVP strain lysate still revealed that band at the last time point tested (111 h). Also, the reactions with WT-strain lysates resulted in more unbound probe than with the OVP strain lysates at all times (Fig. 5B). To determine whether the greater shifts observed with OVP strain lysates were because of more abundant BldD, we assessed the expression of BldD in the WT and OVP strains by using Western blots. Polyclonal anti-BldD antibody detected BldD in the first four lysates of each previous time course. BldD was observed in both strains after 27 h of growth, when they were producing erythromycin (Fig. 5 C and D). However, BldD abundance in the WT strain decreased after 40 h, coinciding with a decrease in erythromycin production. In contrast, the OVP strain maintained relatively constant BldD levels up to 87 h, as production of erythromycin continued. The amounts of BldD detected by Western blot analysis matched well with the intensities of shifts in EMSAs (Fig. 5B), indicating that a higher abundance of BldD, rather than a more active form, caused stronger shifts from the OVP strain. How the OVP strain acquires the phenotype of extended BldD expression is not known, but we postulate that during classical strain improvement, mutations were introduced in genes that regulate BldD expression, as the sequences for the promoter and coding region of bldD are identical for the WT and OVP strains.

Deletion of bldD Generates a “Bald” Phenotype.

We deleted bldD in S. erythraea strain AML315–638 (which derives from the WT strain) (20) as described in Materials and Methods. The deletion strain (ΔbldD) failed to form aerial mycelium and to sporulate on three different media, M1 (21), SFM, and R5 agar (Fig. 6A and Fig. S6). This bald phenotype, which is also characteristic of S. coelicolor ΔbldD strains (7, 22), suggests that S. erythraea BldD and S. coelicolor BldD have similar functions. Complementation of the bldD deletion with a single copy of bldD restored the WT phenotype, whereas complementation with a plasmid lacking bldD maintained the bald phenotype (Fig. 6A). When grown in a liquid, rich medium for 5 days, the ΔbldD strain produced 7-fold less erythromycin than the WT strain (Fig. 6B). Complementation of the bldD deletion with a single copy of bldD restored normal titers, whereas complementation with a plasmid lacking bldD left titers low (data not shown). Together, these data suggest that BldD positively regulates the ery genes.

Fig. 6.

Phenotypes and titers of different S. erythrea strains. (A) Phenotypes of S. erythraea lacking bldD. Strains were grown on M1 agar. WT, wild-type S. erythraea; ΔbldD, WT strain with bldD deleted; ΔbldD::bldD, ΔbldD strain with bldD integrated on a vector next to the ery cluster (see text for details); and ΔbldD::empty vector, ΔbldD strain with an empty vector integrated next to the ery cluster, which served as a negative control. (B) Titers of the WT (black diamonds) and ΔbldD (open squares) strains.

However, because the mutation left some synthesis of erythromycin intact, it differs from deletions of activators in many streptomycetes that abolish the production of an antibiotic completely (3, 23). Our observations more closely resemble the case of S. noursei, in which a strain with a deletion of a nystatin synthesis regulator still produced some antibiotic (24).

Our data show that not only is BldD important for erythromycin biosynthesis, it is also necessary for morphological differentiation. We report the discovery of a developmental transcription factor that directly activates expression of the enzymes of an antibiotic biosynthetic pathway. The findings here provide a starting point for understanding the regulation of erythromycin biosynthesis. The complex mechanisms that generate antibiotics in actinomycetes involve factors such as small signaling molecules (25) and hierarchies of transcriptional proteins (26). Attempts to understand how proteins such as BldD work with these factors should promote progress toward new strategies for strain improvement. The same efforts should reveal further how bacteria connect the synthesis of small molecules to their morphogenesis.

Materials and Methods

Bacterial Strains and Media.

S. erythraea WT NRRL2338 and OVP strain K41–135 were obtained from the American Type Culture Collection and Kosan Biosciences, respectively. All S. erythraea strains were grown and stored as described in ref. 20. E. coli strains XL1-Blue and One Shot TOP10 (Invitrogen) were used for DNA cloning. E. coli BL21(DE3) (Novagen) was used for heterologous production of BldD.

Generation of Fluorescent Probes for the Promoters of the ery Cluster.

Promoter fragments were generated by PCR with Taq polymerase by using the primers listed in Table S1. The PCR products were purified and concentrated by gel purification and end-labeled with cyanine 3-dCTP. The end-labeling reaction consisted of ≈1 μg PCR product, 1X TdT Buffer, 0.25 mM CoCl₂, 30 units of TdT (NEB), and 1 nmol of cyanine 3-dCTP (Perkin–Elmer) in a final volume of 20 μl. The reaction was incubated at 37°C for 1 h and then heat inactivated at 70°C for 10 min. Unincorporated fluorescent nucleotides were removed by washing the reactions twice with 400 μl of TE [10 mM Tris, 1 mM EDTA (pH 7.5)] on Microcon YM-10 filters (Amicon). The labeled probe was concentrated to a final volume of 20 μl.

Protein Preparation and Quantification.

S. erythraea strains were grown in R5 medium for protein harvesting experiments. Cells were harvested by centrifugation and cell pellets were resuspended in TA Buffer (14) with 50 mM NaCl and Complete protease inhibitor mixture (Roche). Cells were lysed by sonication with a Branson Digital Sonifier (Model 450). Total protein concentrations were measured by using the Bradford method (27) with the Bio-Rad Protein Assay Kit II. Proteins were visualized by SDS-PAGE on 12% Tris·HCl Ready gels (Bio-Rad) that were run according to standard procedures (28).

Electrophoretic Mobility Shift Assays.

The binding reaction consisted of 10 mM Tris (pH 7.5), 5 mM MgCl₂, 50 mM EDTA, 60 mM KCl, 10 mM DTT, 10% glycerol, 1 μg of poly dI/dC, 1 μl of labeled probe (≈50 ng), and 10–50 μg of crude or partially purified protein. Reactions were incubated on ice for 10 min and then run on a 5% TBE polyacrylamide gel (Bio-Rad) buffered in 0.5X TBE (45 mM Tris-borate, 1 mM EDTA) at 30 mA. Gels were imaged on a Typhoon 9410 Variable Mode Imager (GE Healthcare).

Purification and Cloning of BldD.

We purified BldD from the OVP strain by using a series of chromatographic steps, monitoring BldD activity by EMSA with the eryBVI probe. The final purification step used the eryBVI probe as an affinity matrix, enriching the binding activity to almost a single ≈18-kDa band on a SDS-PAGE gel, which was characterized by de novo sequencing and MALDI-TOF MS/MS. From the peptide sequences obtained, a ligation mediated PCR (LMPCR) protocol (29) was used to obtain the bldD DNA sequence from the S. erythraea overproducer genome. See SI Materials and Methods for a detailed protocol.

Production of BldD in E. coli.

The bldD gene was amplified with NcoI and BamHI restriction sites from the S. erythraea overproducer genome by using primers 5′-CCATGGGCGACTACGCCAAG-3′ and 5′-GGATCCTCACTCCTCCCGGGC-3′ with Pfu polymerase (Stratagene) and cloned into pET15b (Novagen). pET15b/bldD was digested with NdeI and EcoRI and the resulting 500-bp fragment was purified and ligated into pET21b (Novagen). The vector was introduced by transformation into E. coli BL21(DE3) for protein production. Fifty mililiters of LB containing 100 mg/liter carbenicillin was inoculated with a single BL21(DE3) colony. Cultures were grown at 30°C to an OD₆₀₀ of ≈0.4–0.6. The inducer isopropyl β-d-thiogalactopyranoside was added to each culture to a final concentration of 0.4 mM. Cultures were induced for 3 h before lysis. Cells were harvested by centrifugation and washed with chilled TE [50 mM Tris (pH 7.5), 1 mM EDTA]. The pellets were resuspended in Buffer D (200 mM sodium phosphate, 200 mM NaCl, 2.5 mM DTT, 1.5 mM benzamidine, 2.5 mM EDTA, 2 mg/liter pepstatin, 2 mg/liter leupeptin, 30% vol/vol glycerol, pH 7.1) and sonicated. Lysates were harvested by centrifugation.

Purification of Recombinant BldD.

E.coli BL21(DE3) containing pET15b/bldD was grown for induction and harvested as described in paragraph 6 of Materials and Methods. We used the cleavage site for thrombin adjacent to the histidine tag in the vector to remove the tag after the protein had been purified with a nickel column. The thrombin was removed by using p-aminobenzamidine agarose. BldD was purified to almost single band purity in SDS-PAGE. See SI Materials and Methods for more details.

Gel Shift Assays for Binding Constants.

The primers that were used have similar sequences to those used in EMSAs but with a 6-FAM molecule at the 5′ end of the forward primers. PCR using WT S. erythraea genomic DNA (gDNA) as template with Pfu polymerase (Stratagene) was used to generate the probe, and the PCR product was purified on a Qiagen Gel Purification column. For binding reactions, an increasing amount of purified BldD was loaded into each lane. The binding reaction was performed and scanned as described in paragraph 4 of Materials and Methods. The free and bound probes were quantified by using ImageQuant software from Molecular Dynamics. Each measurement was performed in triplicate.

DNaseI Footprinting of eryBVI.

The DNaseI footprinting experiment was performed as described in ref. 30. See SI Materials and Methods for more details.

Gene Deletion and Complementation of bldD.

The REDIRECT protocol for S. coelicolor targeted gene replacement was used (31), with slight modifications at the conjugation step (O. Lazos, personal communication) as described in SI Materials and Methods. To reintroduce bldD into the ΔbldD strain, primers 5′-gaattcGACGCCACGGTGGAACC-3′ and 5′-ggtaccTCACTCCTCCCGGGCC-3′ were used to amplify bldD from WT S. erythraea gDNA and cloned into the EcoRI/KpnI sites of pKOS460–18 to generate pKOS460–18/bldD. Subsequently, this plasmid was introduced by conjugation into the ΔbldD strain. Table S2 lists the primers used to generate the apramycin resistance cassette.

Western Blot.

Crude cell lysates containing 4 μg of protein were mixed with SDS-PAGE sample buffer, boiled, run on a 12% Tris·HCl Ready Gel (Bio-Rad), and then transferred to a PVDF membrane (Bio-Rad) by using standard protocols. For immunodetection, anti-BldD polyclonal antibody from rabbit was used as the primary antibody (1:10,000 dilution), and horseradish peroxidase-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc.) (1:10,000 dilution) was used as the secondary antibody. The reaction was visualized by using the ECL Plus system (Amersham) on a Typhoon 9410 Variable Mode Imager.

Erythromycin Measurements.

Erythromycin titers were measured by using a Micrococcus luteus bioassay as described in ref. 20.