Fe²⁺-Tetracycline-Mediated Cleavage of the Tn10 Tetracycline Efflux Protein TetA Reveals a Substrate Binding Site near Glutamine 225 in Transmembrane Helix 7

Abstract

TetA specified by Tn10 is a class B member of a group of related bacterial transport proteins of 12 transmembrane alpha helices that mediate resistance to the antibiotic tetracycline. A tetracycline-divalent metal cation complex is expelled from the cell in exchange for a entering proton. The site(s) where tetracycline binds to this export pump is not known. We found that, when chelated to tetracycline, Fe²⁺ cleaved the backbone of TetA predominantly at a single position, glutamine 225 in transmembrane helix 7. The related class D TetA protein from plasmid RA1 was cut at exactly the same position. There was no cleavage with glycylcycline, an analog of tetracycline that does not bind to TetA. The Fe²⁺-tetracycline complex was not detectably transported by TetA. However, cleavage products of the same size as with Fe²⁺ occurred with Co²⁺, known to be cotransported with tetracycline. The known substrate Mg ²⁺-tetracycline interfered with cleavage by Fe²⁺. These findings suggest that cleavage results from binding at a substrate-specific site. Fe²⁺ is known to be able to cleave amide bonds in proteins at distances up to approximately 12 Å. We conclude that the α carbon of glutamine 225 is probably within 12 Å of the position of the Fe²⁺ ion in the Fe²⁺-tetracycline complex bound to the protein.

The TetA(B) protein (Fig. 1) encoded by the tetA(B) gene on Tn10 represents class B of a set of seven related TetA tetracycline efflux pumps from gram-negative bacteria (9, 20). All TetA proteins produce resistance to the antibiotic tetracycline, an inhibitor of protein synthesis, by exporting it out of the cell as a divalent cation chelate (48) in exchange for a proton (16, 45). Transcription of tetA is blocked by the repressor TetR and is inducible by low levels of tetracycline, which reversibly binds to and inactivates the repressor (12). TetA is located in the cytoplasmic membrane and has been extensively characterized in relation to its topology, structure, and the importance of various residues to its function (26, 39, 41, 46).

FIG. 1.

Model of TetA protein from Tn10. The protein has 401 residues. Transmembrane alpha helices are represented by gray rectangles numbered TM1, TM2, etc., which each span the membrane. The helices are connected by putative extramembrane loops; the boundaries of the helices are as given by Tamura et al. (41). The circled residue in TM7 is Trp231, and the boxed residue in TM9 is Asp285. Residues in or bordering TM7 that are identical in the class B and D TetA proteins are shown in boldface. The arrow designates the site between Gly224 and Gln225 cleaved in the FeTc reaction.

Along with a few other well-studied transporters, such as lactose permease LacY (36), melibiose carrier MelB (2), sugar phosphate carrier UhpT (7), and glucose transporter GLUT1 (42), TetA is representative of transport proteins in the major facilitator superfamily, each typically having 12 transmembrane alpha helices configured as two related halves of six transmembrane helices each (30, 35). A model for LacY based on the average predicted tertiary structure of many such proteins has been proposed (see reference 5). A model for TetA has also been put forward (41). A trimeric structure for TetA has been seen in two-dimensional crystals at low (17 Å) resolution (49). Possible substrate binding regions of TetA have been defined by mutations that alter substrate specificity or affinity (6, 38, 43, 46). Mutations at other regions that may be involved in substrate binding have also been described (17, 18, 41). However, the location of the TetA substrate binding site is not known.

Chelated ferrous ion (Fe²⁺) in the presence of oxygen and reducing agent can lead to cleavage of an amide bond of a protein backbone, resulting in fragments with (in some cases) unblocked amino termini that can be sequenced (15, 31, 33). The cleavage sites indicate regions in close proximity to the Fe²⁺. This technique was used with Fe²⁺ chelated to tetracycline to map the inducer binding site in the tet repressor TetR (3). The results agreed with the location of the divalent cation in the Mg²⁺-tetracycline (MgTc) ligand obtained earlier by X-ray crystallography (3, 14), confirming the validity of the method. We now describe similar experiments on the TetA(B) efflux pump protein.

MATERIALS AND METHODS

Chemicals and antibodies.

9-(N,N-Dimethylglycylamido)-6-demethyl-6-deoxytetracycline (DMG or glycylcycline) and 13-(cyclopentylthio)-5-hydroxy-6-α-deoxytetracycline (Cptc) were gifts of Lederle and Paratek Pharmaceuticals, respectively. FeSO₄ · 7H₂O, Fe(NH₄)₂(SO₄)₂·6H₂O, CoCl₂·6H₂O, and tetracycline hydrochloride came from Sigma, and n-dodecyl-β-d-maltoside came from Anatrace. Anti-Ct, rabbit immune serum to the C-terminal 14 amino acids of TetA (44), was a kind gift of A. Yamaguchi; it does not react with Tet-6H (23) or TetB1-6H. The rabbit immune serum anti-Tet recognized an unknown epitope between residues 127 and 201 of TetA (23), possibly that in the central loop described for a mouse monoclonal antibody (27). Anti-6H was a mouse monoclonal antibody to 6H (Novagen). Anti-rabbit-horseradish peroxidase (HRP) and anti-mouse-HRP were HRP conjugates (Cell Signaling/New England Biolabs), as was anti-T7 tag-HRP (Novagen).

Strains.

All strains were Escherichia coli. Strain D1-209 was host ML308-225 (see reference 24) containing plasmid R222 (see below). Other strains are cited elsewhere in the text.

Plasmids and encoded fusion proteins.

Most experiments were performed using the large, very-low-copy-number, naturally occurring plasmid R222, which bears Tn10, which in turn contains the divergent genes for the class B TetR (repressor) and TetA (efflux pump) proteins (see reference 24). The sequence that we have found for the Tn10 tetA was identical to one reported (29) and differed from another (13) at four nonsilent codons. pET21b-tet6, derived from pET21b (Novagen) (1), bears genes for ampicillin resistance and the repressor LacI and specifies the fusion protein Tet-6H (1, 23). pETtetB1, specifying the fusion protein TetB1-6H, was constructed from pET21b-tet6 by replacement of the NdeI-EcoRI restriction enzyme fragment (beginning just upstream from the T7 tag translational start of Tet-6H and ending at the EcoRI site in the center of the tetA gene) with an NdeI-EcoRI PCR product encoding the first half of TetA with the native amino terminus (synthesized from a pLR1068 template bearing tetA (25). pETtetA2 was like pET21b-tet6, except the cloned gene was the class A tetA gene from naturally occurring plasmid RP1 from our laboratory collection. pETtetC1 bearing the class C gene was constructed in similar fashion using pCR2 (37). pETtetD4 bearing the class D gene was made similarly using naturally occurring plasmid RA1; the fusion protein was called TetD-6H. The transcription of all fusion proteins was regulated by a T7 promoter/lac operator. pACT7 (1), having no lacI gene and specifying T7 RNA polymerase regulated by a lac promoter/operator, was compatible with pET21b derivatives and was transformed into cells already bearing those plasmids. DNA sequencing to verify all clones was performed by the Tufts University Core Facility using an ABI 37X DNA sequencer.

All fusion proteins except TetB1-6H each had an 11-residue T7 tag epitope at the N terminus and LEHHHHHH appended to the native carboxy terminus of TetA to allow purification by affinity to an Ni²⁺ resin. TetB1-6H had a native TetA amino terminus and LEHHHHHH attached to the native carboxy terminus.

pGG9, pGCR10, and pGCR17 (6) were kind gifts of G. G. Guay. They encode, respectively, the class B wild type and mutant W231C and W231Y TetA proteins; they were used in strain MC1061, which was induced with 1.4 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 1.5 h.

Growth of cells and preparation of membranes.

Cells were grown at 37°C in Luria-Bertani broth (23). Synthesis of TetA(B) in strain D1-209 was induced by continuous growth of the strain in 2 μg of tetracycline/ml. Synthesis of the -6H fusion proteins (in strain DH5α [see reference 1] bearing pACT7) was induced when the cells reached late logarithmic growth phase (A₅₃₀ = 0.8) by addition of 300 μM IPTG for 2 h. Cells were washed and resuspended in 50 mM Tris-HCl, pH 7.5, and were lysed by sonication (Branson) in 50 mM Tris-HCl, 1 mM EDTA, and 100 μg of lysozyme/ml at an A₅₃₀ of 80. Unless noted, D1-209 cells were harvested at an A₅₃₀ of 0.8 and were washed once in 50 mM MOPS (morpholinepropanesulfonic acid, potassium salt), pH 6.6. D1-209 cells were lysed in MOPS at an A₅₃₀ of 80 (∼24 mg of cell protein/ml) using a French pressure cell (24); unlysed cells were removed by centrifugation for 10 min at 12,000 × g. In either method, membranes were isolated by sedimentation at 180,000 × g for 1 h at 4°C and were eventually resuspended at 20 to 50 mg of protein/ml. For lysis in MOPS, final resuspension of membranes was in MOPS; for lysis in Tris, the membranes were washed once by sonication in Tris to remove EDTA before final resuspension in the same buffer. Membranes were either used immediately or stored at −80°C. For Fig. 3A and B, cells harvested at an A₅₃₀ of 0.8 were washed and concentrated to an A₅₃₀ of 30. A 0.2-ml sample was lysed by the sonication method. Conditions for sedimentation and washing of membranes were 30 min at 15,000 × g in 1.5-ml Eppendorf tubes.

FIG. 3.

Effects of analogs and mutations on cleavage and efflux activity. TetA proteins were expressed from the IPTG-inducible promoter of pGG9 or its derivatives. Both immunoblots were reacted with anti-Ct. (A) Immunoblot showing effect of Tc and its analogs Cptc and DMG on cleavage of wild-type TetA (pGG9). Concentrations (above lanes) are in micromolars. Fe²⁺ was in excess at 200 μM; cleavage was for 2 h. Arrows locate the intact protein (TetA, 36 kDa), the major cleavage fragment (18 kDa), and a minor fragment representing a second, less efficient cleavage site (8 kDa). (B) Immunoblot showing effect of a W231C mutation (pGCR10) in TetA on cleavage. Fe and Tc were each 100 μM, cleavage was for 3 h, and a wild-type TetA control was included. (C) Fluorescence assay for efflux activity of vesicles bearing wild-type and mutant TetA (Materials and Methods). Vesicle protein was 0.2 mg/ml, MgCl₂ was 10 mM, lithium lactate (10 mM) was added to energize, 100 μM Tc was added to initiate proton antiport, and 20 μM carbonyl cyanide m-chlorophenylhydrazone (CCCP) was added to deenergize.

Cleavage reaction.

Membranes resuspended in 50 mM Tris-HCl, pH 7.5, at 0.5 to 2.7 mg of protein/ml were incubated with sodium ascorbate (10 mM) and additives as desired (tetracycline hydrochloride or its analogs, FeSO₄, MgCl₂, and CoCl₂) at 22°C for 2 h unless noted. When the ferrous ion concentration was above 200 μM, the more soluble Fe(NH₄)₂(SO₄)₂ was used in the cleavage reaction. All stock solutions of additives were prepared immediately before use. The reaction was stopped by adding sodium EDTA to 10 mM; in some cases, electrophoresis sample buffer was also added.

Competition between MgTc and FeTc.

The K_d for the chelation of Fe²⁺ by tetracycline is about 5 μM, and that for Mg²⁺ is about 250 μM (22). In experiments analyzing the ability of the Fe²⁺-tetracycline (FeTc) complex to inhibit Mg-dependent tetracycline transport or the ability of the MgTc complex to prevent the cleavage reaction by competing with FeTc for the same binding site on TetA, it was important that competition between the two cations at the level of chelation to tetracycline did not play a role. Therefore the concentrations of the three species (Fe, Mg, and Tc) had to be such that the concentration of the MgTc chelation complex was lowered very little by the addition of Fe²⁺ and vice versa (see “Calculations” below).

Calculations.

To calculate the different species present in a mixture of Mg, Tc, and Fe, we defined the following: a = total Fe; b = unbound Fe; c = FeTc; d = total Mg; e = unbound Mg; f = MgTc; x = total Tc; and y = unbound Tc. We have five equations that describe the relationship between these variables: a = b + c; d = e + f; x = y + c + f; by/c = 5 μM; and ey/f = 250 μM. Since the values of a, d, and x are known, this leaves us with five unknown variables (b, c, e, f, and y) with five independent equations. Successive elimination of variables other than b through combinations of the various equations leads to a cubic equation: hb³ + ib² + jb + k = 0, where h = 49; i = d − 48a + 49x + 245; j = −240 + ax − ad − a²; and k = −5a². The equation was solved for b for different sets of values for a, d, and x by the Cubic Equation Calculator on the Internet at www.1728.com/cubic.htm. The validity of any solution of the cubic equation can be seen because all of the primary equations (e.g., by/c = 5 μM) are obeyed.

Purification of cleavage fragments derived from TetB1-6H protein.

Membranes containing TetB1-6H that had been treated in the Fe²⁺-mediated cleavage reaction either with or without tetracycline were resuspended in 10 mM Tris-HCl, 150 mM NaCl, 8% glycerol, pH 8.0, and solubilized by addition of n-dodecyl-β-d-maltoside to 1.5%, and the undissolved material was sedimented by centrifugation as described earlier (1) and discarded. Intact TetB1-6H and carboxy-terminal fragments thereof were purified on a column of nickel ion-nitrilotriacetic acid (Ni-NTA) agarose (Qiagen) as described earlier (1) via the affinity of the polyhistidine -6H for the Ni-NTA. The loaded column was washed with 5, 25, and 40 mM imidazole and was eluted with 500 mM imidazole. The eluted proteins were concentrated by precipitation with 10% trichloroacetic acid.

SDS-PAGE, immunoblotting, and amino-terminal sequencing.

Following a cleavage reaction, the membrane resuspensions or the purified proteins were dissolved in sodium dodecyl sulfate (SDS)-β-mercaptoethanol-Tris-HCl (pH 6.8) electrophoresis sample buffer at 22°C with sonication and fractionated by SDS-12% polyacrylamide gel electrophoresis (PAGE) as described earlier (1). The gel was then agitated 15 min in 10 volumes of 25 mM Tris base-192 mM glycine pH 8.3 to partially remove SDS and was electroblotted in the same buffer onto a polyvinylidene difluoride membrane. For routine immunoblotting, the membrane was Immobilon P (Millipore), the blotter was usually a semidry model, and the resulting Western blot was blocked in 0.5% bovine serum albumin and probed with the appropriate primary antibody and secondary antibody-HRP conjugate. The Renaissance system (New England Nuclear) was used to obtain the HRP chemiluminescent signal. For protein sequencing, the acrylamide gel was polymerized overnight prior to use, the membrane was Sequiblot (Bio-Rad), the blotter was a tank model, and the blot was stained with 0.1% Coomassie brilliant blue in 50% methanol and 7% acetic acid. Particular bands found only in the samples that had received tetracycline were excised for amino-terminal protein sequencing by the Tufts University Core Facility using an ABI 477 Protein Sequencer.

Quantitation of immunoblots.

When the chemiluminescence signal was detected by BioMax-MR film (Kodak), the film was scanned using Adobe Photoshop with default settings. Using NIH Image 1.62, the relative amounts of bands were then determined from the linear region of a standard curve of a dilution series of one sample. For determination of K_d values and of cleavage in the presence of Mg²⁺, the (more efficient) tank electroblotter was used; the signal was detected and analyzed by a Kodak 440 CL Imaging System, and standard curves were prepared from a dilution series in which the diluent was ML308-225 vesicles on which a mock cleavage procedure had been performed through the SDS sample buffer step. The standard curve was linear, and duplicate lanes gave comparable values.

Transport of tetracycline.

Transport of [³H]tetracycline in everted membrane vesicles, prepared by lysis in 50 mM potassium phosphate, pH 6.6, in a French pressure cell and energized by 10 mM lithium lactate, was assayed as described earlier (24). Antiport of a tetracycline-divalent cation complex in exchange for a proton was assayed using an acridine orange fluorescence method (25), in which dequenching of the fluorescence of energized vesicles by substrate addition showed antiport.

RESULTS

TetA was cleaved into two fragments by Fe²⁺ in the presence of tetracycline and ascorbate. Everted membrane vesicles bearing wild-type TetA protein (from strain D1-209) were incubated in 50 mM Tris-HCl at pH 7.5 in the presence of 10 mM ascorbate (as reducing agent) together with 50 μM FeSO₄ or 50 μM tetracycline or both for 2 or 30 min at 22°C. EDTA was then added to stop the reaction by chelating the Fe²⁺, and the samples were fractionated by SDS-PAGE and electroblotted onto a polyvinylidene difluoride membrane. The latter was probed with anti-Ct to detect the native carboxy terminus or with anti-Tet to detect an unknown epitope in the amino-terminal half between residues 127 and 201 of the 401-residue protein.

The intact TetA protein at 36 kDa reacted with both antibodies as expected (Fig. 2A). By 30 min, two cleavage products were clearly seen, one of ∼18 kDa reactive only with anti-Ct and another of ∼21.5 kDa reactive only with anti-Tet (Fig. 2A). Both tetracycline and Fe²⁺ were required for cleavage (Fig. 2A).

FIG. 2.

Cleavage of TetA and Tet fusions by tetracycline (Tc) plus ferrous ion (Fe). Parts A, B, and C represent different immunoblots. As in other figures, film images were scanned using Adobe Photoshop 4.0 at default settings. Apparent molecular weights of bands are shown in kilodaltons. The antibody used is indicated beneath each blot. The type of Tet protein present in each sample is labeled above the lanes. All cleavage incubations were for 30 min except those marked **, which were for 2 min. If present (+), Fe was 40 μM and Tc was 20 μM (A) or 80 μM (B and C). The absence of Fe or Tc is indicated by (−).

The smaller product (18 kDa) could be immunoprecipitated with anti-Ct (data not shown). When the membrane vesicles bore the TetB1-6H fusion protein (see next paragraph) instead of TetA, the smaller fragment reacted with anti-6H (Fig. 2B). The reactivity of the smaller fragment with anti-Ct (TetA) or anti-6H (TetB1-6H) showed that it contained the carboxy terminus of the protein.

The larger product (21.5 kDa), reactive with anti-Tet, was found with all tested Tet fusion proteins that retained the native amino terminus, including TetB1-6H, Tet390-LacZ (from pRKH40 [11]), and Tet390-PhoA (from pRKH12 [10] (data not shown). When the membrane vesicles contained the Tet-6H fusion protein, which had a T7 tag at the amino terminus, the anti-Tet-reactive 21.5-kDa fragment was slightly larger (22 kDa) than seen for TetB1-6H (Fig. 2C) and was reactive with anti-T7 tag. For Tet-6H, the fragment detected with both anti-T7 tag and anti-Tet was the same size, 22 kDa, as expected (data not shown). The larger fragment therefore came from the amino-terminal portion of the protein.

As tetracycline was raised from 0 to 320 μM (with Fe²⁺ at 100 to 400 μM, in excess of tetracycline), the amount of 18-kDa band formed in 2 h began to saturate. The point at which this occurred varied in different experiments. Hence, the K_d calculated from Scatchard plots (assuming a hyperbolic curve, with cleavage proportional to binding of FeTc) ranged from less than 20 to 125 μM, with a median of 35 μM from five determinations. The amount of the 18-kDa cleavage product at the higher levels of tetracycline was 20% or more of the total. If ascorbate was omitted, no cleavage of TetA occurred. The reaction was prevented by 10 mM EDTA (to chelate Fe²⁺) or by heating the vesicles at 95°C for 10 min (to denature TetA) before the addition of tetracycline and Fe²⁺. The radical scavengers glycerol and mannitol (500 mM) did not inhibit cleavage of TetA, nor did hydrogen peroxide enhance the reaction. The reaction could be extended overnight at 22°C for some increase in cleavage (data not shown) with few obvious side reactions (see Fig. 6).

FIG. 6.

Immunoblot comparing tetracycline-dependent cleavage of TetA by Co²⁺ with that by Fe²⁺. Cleavage was for 18 h at 22°C in the presence (+) or absence (−) of the various additives. Membranes were prepared from D1-209 cells by the sonication method. The acrylamide concentration of the gel was 14%, and a tank blotter was used. The amount of protein loaded in Co²⁺ samples 5 and 6 was increased fivefold to better show the relative fragment yield compared to that with Fe²⁺. The left and right panels have identical samples probed with different antibodies to show the carboxy-terminal (left) and amino-terminal cleavage fragments (right). Both antibodies react with intact TetA. Two sets of cleavage products can be seen: the major cleavage yields a carboxy-terminal 18-kDa fragment and its amino-terminal partner of 21.5 kDa; a minor cleavage yields probable 8- and 28-kDa partners.

A second, less abundant carboxy-terminal cleavage product of approximately 8 kDa was also often seen (Fig. 3; see Fig. 5 and 6), as well as a presumably complementary, less abundant amino-terminal fragment of about 28 kDa (see Fig. 6).

FIG. 5.

Immunoblot (probed with anti-Ct) showing the inhibitory effect of Mg²⁺ on cleavage of TetA by FeTc. Three different sets of concentrations (in micromolars) were used: A, B, and C. Lanes in which the control tube (+FeTc; no Mg²⁺) was diluted 1/3 and 1/9 were used to help quantitate cleavage in the presence of Mg²⁺. On the left are marked the major (18 kDa) and minor (8 kDa) cleavage products. On the right are marked other bands of unknown significance that appeared with Mg²⁺ only in the presence of Fe²⁺.

Effect of two tetracycline analogs in the cleavage reaction.

Cptc, an analog of tetracycline that binds competitively to TetA (28), also caused cleavage, at times more efficiently than did tetracycline (Fig. 3A). The glycylcycline DMG, an analog that does not interact with TetA (40), did not cause Fe²⁺-mediated cleavage at 125 μM in 2 h (Fig. 3A and B) or even at 175 μM after 22 h (data not shown). The same preparation of DMG was shown to be bioactive by its efficient and similar growth inhibition of both tetracycline-resistant (TetA-containing) and -susceptible cells. DMG was known to chelate other divalent cations (48); we found these to include Fe²⁺, since there was an Fe²⁺-dependent bathochromic shift in the DMG absorbance peak from 361 to 376 nm (maximal at ∼40 μM Fe²⁺ for DMG at 50 μM). The results with the analogs supported the hypothesis that the cleavage of TetA occurred near its substrate binding site.

The cleavage site in TetB1-6H was located between Gly224 and Gln225 in TM7.

The sum of the apparent molecular weights of the two major fragments of TetA produced by the cleavage reaction was consistent with that of the intact protein (36 kDa for TetA). The intensities of the two fragment bands (each detected with one of the antibodies) were similar to each other, suggesting that the bands were similar in amount, since the concentrations of anti-Ct and anti-Tet were chosen to produce signals of similar intensity with the intact protein (Fig. 2A; see also Fig. 6). Therefore, the two fragments probably had resulted from a single scission of the TetA protein. From the relative molecular weights of the two fragments, the cut site was predicted to be near TM7 or TM8.

To locate the site of the cleavage more precisely, the amino terminus of the smaller product was sequenced. To do this, we performed the cleavage reaction with membranes containing the TetB1-6H fusion protein; as a control, we omitted tetracycline from a portion of the reaction. Polypeptides containing the polyhistidine carboxy terminus were then purified on Ni-NTA, fractionated by SDS-PAGE, and electroblotted (see Materials and Methods). Coomassie-stained protein bands that were present only if FeTc had been present in the reaction were excised from the blot and sequenced.

Two independent experiments were performed. In the first experiment, Fe²⁺ was 50 μM, tetracycline was 100 μM, and the control lacked both. The 18-kDa carboxy-terminal fragment and, unexpectedly, the 21.5-kDa amino-terminal fragment were purified in approximately equal amounts on Ni-NTA agarose, representing about 5% of the also-purified 36-kDa band (Fig. 4, right lane). The 21.5-kDa band proved to be a single species beginning at the first residue (Met 1) of TetB1-6H (MNSSTKIA). Probably this amino-terminal fragment had copurified with intact TetB1-6H on Ni-NTA due to the Tet-Tet dimerization domain present in the first half of the protein (23). Amino-terminal sequencing of the 18-kDa band revealed the presence of two polypeptides similar in amount. One had the amino-terminal sequence of TetA, as above; this was probably a copurifying endogenous breakdown product slightly larger than the 18-kDa cleavage product, since such a slightly larger fragment was also seen for protein purified in parallel from control membranes not treated with tetracycline (Fig. 4, left lane). The second polypeptide in the 18-kDa band had an amino-terminal glutamine (Q) followed by IPATVWVLF. From the sequence of TetA (29) it was clear that this was the true carboxy-terminal cleavage product and that its amino terminus was Gln225, believed to be located in the center of TM7 (41) (Fig. 1).

FIG. 4.

Coomassie-stained gel from SDS-PAGE of TetB1-6H extracted from membranes treated without (−) or with (+) 50 μM Fe²⁺ plus 100 μM tetracycline. Intact TetB1-6H and its -6H cleavage fragments were purified by Ni-NTA chromatography. The uncleaved protein (36 kDa), the copurifying N-terminal (21.5 kDa) cleavage product, and the C-terminal (18 kDa) cleavage product are labeled.

In the second experiment, the entire procedure was repeated. In this case the tetracycline analog Cptc (see above) at 175 μM replaced tetracycline, Fe²⁺ was 175 μM, and the control lacked only Cptc. The 18-kDa band was excised more carefully to avoid the nearby endogenous band mentioned above. The 18-kDa band now consisted almost entirely of a single sequence, QIPATVW; only a small amount of the amino-terminal TetA contamination (MNS) was detected. Therefore, the cleavage had occurred at precisely the same position as in the first experiment.

We interpreted these results to mean that the peptide bond between Gly224 and Gln225 in the middle of TM7 of TetB1-6H had been cut uniquely in the presence of FeTc. This position was likely close to Fe²⁺ in the binding site for the FeTc complex.

Effect on the cleavage reaction of mutations at Trp231 in TM7.

Mutations converting Trp231 at the periplasmic end of TM7 (Fig. 1) to Cys or Ser (W231C or W231Y) allowed TetA to confer low-level DMG resistance to cells while somewhat lowering the resistance to tetracycline, thereby altering substrate specificity and so suggesting involvement in substrate binding (6). We expected that DMG, as a substrate for the mutant proteins, might cleave these mutant TetA proteins. However, DMG caused no cleavage of the W231C mutant TetA protein specified by pGCR10 (Fig. 3B). In fact, even tetracycline, which cleaved the wild-type TetA encoded by plasmid pGG9 as expected (Fig. 3B), did not cause cleavage of W231C (Fig. 3B) or W231Y (data not shown). These results were understood when the fluorescence assay showed that everted membrane vesicles prepared from cells synthesizing the mutant proteins did not transport tetracycline, while vesicles containing the wild-type protein encoded by the same plasmid did (Fig. 3C). The reason for this lack of function is not known; perhaps the mutant proteins are inactivated during vesicle preparation. In any case, loss of activity of the protein correlated with loss of cleavage.

FeTc was probably not transported by TetA.

We asked whether the cleavage by FeTc represented action at the substrate binding site of TetA. To study whether FeTc could be transported by TetA, since this was not known, we used an acridine orange fluorescence assay (25) (Materials and Methods) to look for the FeTc-dependent proton efflux (antiport) that would be expected to accompany the uptake of FeTc by energized everted membrane vesicles. While antiport was seen as expected for MgTc, it was not seen for FeTc, although fluorescence quenching caused by FeTc made these assays imperfect. We also directly measured the lactate-dependent uptake of [³H]tetracycline (at 2 μM) into everted membrane vesicles in the presence of Mg²⁺ or Fe²⁺. There was some active tetracycline transport even in the absence of any added divalent cation; with the addition of 2 mM Mg²⁺, this endogenous uptake increased 3.5-fold (data not shown). With 50 μM Fe²⁺, which would chelate an amount of [³H]tetracycline similar to that chelated by the 2 mM Mg²⁺ (∼1.8 μM), there was no change in the endogenous active uptake (data not shown). We tentatively concluded that the FeTc complex was not transported by TetA.

Transported substrate MgTc reduced cleavage by FeTc.

Even if it were not transported, the FeTc complex might bind to TetA at the same site used by MgTc, a known substrate of TetA. Were this so, Fe²⁺ should interfere with the Mg²⁺-dependent active tetracycline transport. However, studies of this question proved to be technically difficult for a number of reasons. One was that the concentration of tetracycline had to be high so that addition of Fe²⁺ would not titrate much tetracycline away from Mg²⁺; this resulted in saturation of the Mg²⁺-dependent active uptake of tetracycline by vesicles. Consequently, uptake became more difficult to detect, since the specific activity of [³H]tetracycline was low. Other problems were a nonspecific sticking of tetracycline to vesicles caused by Fe²⁺ and the quenching of the fluorescence assay by the FeTc complex.

We therefore took the opposite approach, to see whether Mg²⁺ would interfere with the Fe²⁺-dependent cleavage of TetA. The addition of Mg²⁺ to the cleavage reaction could inhibit in one of two ways. The first would be that Mg²⁺ would simply compete with Fe²⁺ for chelation to tetracycline. The second, the one of interest to us here, would be that a MgTc chelation complex would compete with a FeTc chelation complex for binding to the TetA protein. To be sure that the second was the major effect, we chose concentrations of total Mg²⁺, Tc, and Fe²⁺ such that addition of Mg²⁺ would not significantly alter the concentration of FeTc and that the ratio of MgTc to FeTc would be at least 10 (Materials and Methods). Three experiments were done at the sets of concentrations shown in Fig. 5. Reagents were added in the following order: ascorbate, Mg²⁺, Tc, and Fe²⁺. In all cases the amount of cleavage product was reduced by Mg²⁺ (Fig. 5). The amount of the 18-kDa product was quantitated (Materials and Methods); the ratio of the amount of 18-kDa product formed with Mg²⁺ to that without Mg²⁺ was 0.28, 0.22, and 0.33 for experiments A, B, and C, respectively. Three bands of about 3, 20, and 24 kDa appeared with an increasing concentration of Mg²⁺ in the presence of Fe²⁺ (Fig. 5). The origin of these bands is not known; their sum was too low (12 to 30% of the control 18-kDa band) to account for the 67 to 78% Mg²⁺-dependent loss from the 18-kDa band. The degree of inhibition by MgTc suggests that MgTc may bind to TetA several times less tightly than does FeTc. The inhibition of the FeTc-mediated cleavage reaction by MgTc, a known substrate of TetA, is consistent with binding of FeTc at the substrate binding site.

A Co²⁺-tetracycline (CoTc) complex, known to be transported by TetA, caused cleavage at or near Gln225.

Because cobalt, like iron, is a transition metal, it can also cause Fe²⁺-like reactions (4, 8, 32). The affinity of Co²⁺ for tetracycline is similar to that of Fe²⁺ (22), and CoTc is known to be transported by TetA (48). We found that, when Co²⁺ replaced Fe²⁺, tetracycline-dependent cleavage of TetA also occurred, albeit less efficiently (Fig. 6). As with FeTc, the CoTc reaction required ascorbate (data not shown). The cleavage with Co²⁺ after 18 h at 22°C was ∼10% of that with Fe²⁺ (Fig. 6, compare samples 2 and 6). We were unable to obtain enough carboxy-terminal product from Co²⁺-mediated cleavage of TetB1-6H to verify that the site cleaved by CoTc was exactly that found for FeTc. It probably was, since the molecular weights of the two major CoTc cleavage products were indistinguishable by SDS-PAGE from those of the two major FeTc products (Fig. 6), even when the Fe²⁺ reaction samples were mixed with the Co²⁺ samples prior to SDS-PAGE (data not shown).

Studies of TetA proteins from other classes.

All experiments described heretofore have dealt with the class B TetA(B) efflux protein from Tn10. Several other classes of TetA protein exist in nature, having related sequences. We thought that they too might be cleaved by FeTc. When the entire second half of the class C protein was replaced by the second half of class B (34) (producing the inactive “TetC/B”), a carboxy-terminal cleavage product was seen (using anti-Ct) that migrated on SDS-PAGE like the class B product (data not shown). We then tested a class C protein TetA(C) bearing at its carboxy terminus the anti-Ct-reactive 14-amino-acid epitope from the class B protein (37). This protein gave no 18-kDa cleavage product but did give one of about 8 kDa similar to the secondary product from TetA(B) (data not shown). We next cloned the class A, C, and D genes so that each protein had an amino-terminal T7 tag epitope and a 6H (polyhistidine) carboxy terminus, the latter both an epitope and a tag useful for purification by metal affinity. The (nonfusion) proteins from these classes are 45 to 46% (classes A and C) and 59% identical (class D) to TetA(B); identity between classes A and C is 78%. By Western analysis using antibodies to each of the two epitopes, we saw that the amount of full-length fusion protein from classes A and C was low and that there was no cleavage by FeTc. The class D protein was more plentiful and cleaved into two fragments similar in size to those from the class B protein (data not shown). By the method used in the second sequencing trial for the class B protein, we determined that cleavage in the class D protein occurred at precisely the same location as in the class B protein, leaving a free amino terminus at Gln225 in TM7.

DISCUSSION

The substrate of the bacterial tetracycline efflux protein TetA is a tetracycline-divalent metal ion chelation complex such as MgTc (48). Although we could not show that FeTc was transported by the protein, we found an Fe²⁺-mediated, tetracycline-dependent cleavage of the TetA protein backbone between Gly224 and Gln225 in TM7. This finding suggested that FeTc bound to the region used by substrate during transport, since (i) FeTc-dependent cleavage occurred with great precision and fair efficiency at one site; (ii) cleavage occurred at the same site in TetA proteins from both classes B and D; (iii) the known substrate MgTc, added in excess over FeTc, decreased cleavage; (iv) cleavage did not occur with glycylcycline, a tetracycline analog known not to bind to TetA (40); (v) a CoTc complex, known to be transported by TetA, cleaved the protein to give two major products of the same apparent molecular weights as those produced by FeTc; (vi) the median of several values obtained for K_d for half-saturation of TetA cleavage with FeTc was 35 μM (the highest was 125 μM), which compares with the K_m values of 22 to 49 μM for transport in vesicles for other divalent cation-tetracycline complexes for TetA (48); and (vii) cleavage was prevented by heating the sample that presumably inactivated the protein or by mutations in TetA that caused loss of efflux ability during membrane preparation. We also note that (viii) Fe²⁺-mediated cleavage with the analog Cptc was usually greater than that with tetracycline (Fig. 3A), a finding that may be caused by the greater affinity of Cptc for TetA (28) or, alternatively, by a more effective cutting by the drug once bound.

Prior evidence that TM7 may have a role in substrate binding is that, as mentioned earlier, mutations at Trp231 alter substrate specificity in favor of glycylcyclines (6). Also, an Asp285Asn mutation in the middle of TM9 was suppressed by an Ala220Glu mutation in the middle of TM7, indicating a TM7-TM9 interaction (39); the K_m for the double mutant was four times that of the wild type (46), suggesting an (altered) substrate binding site involving TM7-TM9. Finally, a barrier to the nonspecific permeation of hydrophilic compounds is located between Gly224 and Ala228 (41), very close to the cleavage site. Perhaps binding of the substrate near this barrier poises it for passage through the protein.

On the other hand, TM7 has no residues whose replacement by Cys leads to a significant loss of activity (41). Most Cys mutant residues whose reactivity with sulfhydryl reagents is inhibited by tetracycline are found in the amino-terminal half of the protein, leading to the suggestion that the substrate binding site might be located there (41). A gating function (48) for the cytoplasmic loop connecting TM2 and TM3 has also been proposed (47).

It is possible that the cleavage position that we found is not within the substrate binding site. Fe²⁺ ion can cause cleavage at a distance of up to ∼12 Å, as shown by work on proteins with known structure (3, 21). Also, the tetracycline moiety of the complex may extend 7 Å or so from the metal ion. It is also clear that Fe²⁺ does not cut every nearby peptide bond (3, 21, 33). This has been attributed to interference by certain atoms or groups in the protein or to the absence of a critical water molecule involved in the reaction (3, 21, 33). Such effects may be greater in the TetA proteins of classes A and C, causing the difference from TetA of classes B and D.

The tetracycline-divalent cation complex that cleaves at Gln225 may thus bind to transmembrane or loop regions in TetA that are close neighbors of TM7. For example, one such region could be the loop at the periplasmic end of TM1, since a change of residue 30 in the loop to Cys allowed cross-linking to a Cys residue introduced by mutagenesis at residue 235 in the loop at the periplasmic end of TM7; the cross-linking was reduced in the presence of tetracycline (19).

We saw a secondary cleavage site, also in the carboxy-terminal half of the protein, represented by the cleavage product of ∼8 kDa seen in several figures. We have not been able to locate the second cleavage position because of the low cleavage efficiency in TetB1-6H. However, the size of the fragment suggested that the break occurred in or near TM10, which contains Leu308, whose mutation (like those in TM7) leads to altered substrate specificity (6). We found that one class C fusion protein also cut at a similar position. Thus, we have seen cleavages only in the second half of any Tet protein.

There are several mechanisms proposed for iron-mediated cleavage of proteins (see references 21, 31, and 33). Some involve radical formation and require hydrogen peroxide, while some do not. For one immobilized Fe-EDTA chelate, a simple hydrolysis could occur in the absence of hydrogen peroxide via oxygen in the presence of ascorbate (33). This may be the mechanism used in the cleavage of TetA protein described here, where ascorbate but not hydrogen peroxide is required, several quenchers of radicals did not prevent cleavage, and the amino terminus at the cleavage site was not blocked.

The FeTc cleavage method differs in kind from the genetic and biochemical techniques used previously to identify important regions in TetA. The results suggest that the divalent metal cation of the tetracycline chelation complex in the substrate binding site is probably within 12 Å of the α carbon of Gln225 in TM7.