The metalloregulatory zinc site in Streptococcus pneumoniae AdcR, a zinc-activated MarR family repressor

Abstract

Streptococcus pneumoniae D39 AdcR (adhesin competence repressor) is the first metal-sensing member of the MarR (multiple antibiotic resistance repressor) family to be characterized. Expression profiling with a ΔadcR strain grown in liquid culture (brain-heart infusion) under microaerobic conditions revealed upregulation of 13 genes, including adcR and adcCBA, encoding a high-affinity ABC uptake system for zinc, and genes encoding cell-surface zinc-binding pneumococcal histidine triad (Pht) proteins and AdcAII (Lmb, laminin binding). The ΔadcR, H108Q and H112Q adcR mutant allelic strains grown in 0.2 mM Zn(II) exhibit a slow-growth phenotype and an approximately twofold increase in cell-associated Zn(II). Apo- and Zn(II)-bound AdcR are homodimers in solution and binding to a 28-mer DNA containing an adc operator is strongly stimulated by Zn(II) with K(DNA-Zn)=2.4 × 10(8) M(-1) (pH 6.0, 0.2 M NaCl, 25 °C). AdcR binds two Zn(II) per dimer, with stepwise Zn(II) affinities K(Zn1) and K(Zn2) of ≥10(9) M(-1) at pH 6.0 and ≥10(12) M(-1) at pH 8.0, and one to three lower affinity Zn(II) depending on the pH. X-ray absorption spectroscopy of the high-affinity site reveals a pentacoordinate N/O complex and no cysteine coordination, the latter finding corroborated by wild type-like functional properties of C30A AdcR. Alanine substitution of conserved residues His42 in the DNA-binding domain, and His108 and His112 in the C-terminal regulatory domain, abolish high-affinity Zn(II) binding and greatly reduce Zn(II)-activated binding to DNA. NMR studies reveal that these mutants adopt the same folded conformation as dimeric wild type apo-AdcR, but fail to conformationally switch upon Zn(II) binding. These studies implicate His42, His108 and H112 as metalloregulatory zinc ligands in S. pneumoniae AdcR.

Fig. 1

Microarray analyses of relative transcript amounts in strains IU2594 (ΔadcR) and IU1781 (adcR⁺) grown exponentially in BHI. Microarray analyses were performed as described in Materials and Methods. A representative log-scale scatter plot of relative transcript amounts is shown, and the fold changes and Bayesian P values for transcripts changing at least 2-fold are listed in Supplementary Table S2.

Fig 2

Growth and cellular zinc content of adcR mutants. (a) Growth curves of adcR parent, ΔadcR, adcR C30A, adcR H108Q, adcR H111Q, and adcR H112Q strains of S. pneumoniae in the presence of 200 μM ZnSO₄. (b) Cellular zinc concentration in bacteria growing exponentially (OD₆₂₀ = 0.1 to 0.3). The average of three biological replicates are shown with SEMs (P < 0.05). See Materials and Methods for ICP-MS and calculation of cellular concentration.

Fig. 3

Representative binding isotherms obtained from titrating Zn(II) into a mixture of AdcR wild-type and competitor (mag-fura-2 or quin-2) at pH 6.0 (a–b) or pH 8.0 (c–d). In the mag-fura-2 experiments, filled symbols represent ε₃₆₆ values which are maximal in the apo-mf2 and open symbols represent ε₃₂₅ values that are maximal in the Zn(II)-mf2 complex. For quin-2 experiments, filled circles represent ε₂₆₅ values that are maximal in apo-quin-2. (a) 47.6 μM AdcR and 25 μM mf2. The solid red line represents non-linear least-square fit to a Zn(II):AdcR_dimer, 3:1 binding model with the K_Zn1 stepwise binding affinity fixed to ≥10⁹ M⁻¹ (a lower limit) and optimized to K_Zn2=1.2 (±0.3) × 10⁹ M⁻¹, K_Zn3=1.7 (±0.1) × 10⁶ M⁻¹. The black dot and dash lines are simulated data to K_Zn1 = 10¹⁰ and 10⁹ M⁻¹, respectively. (b) 41 μM AdcR and 16 μM quin-2. The solid red line represent non-linear least-square fit to a Zn(II):AdcR_dimer, 0.5:1 binding model, where K_Zn1 = 8.3 (±0.2) × 10⁹ M⁻¹. Simulated data is represented by discontinuous black lines, with K_Zn1 = 10¹², 10¹¹, 10¹⁰, 10⁹, 10⁸ M⁻¹, descending from right to left. (c) 42.5 μM AdcR and 17.7 μM mf2. The solid red line represent non-linear least-square fit to a Zn(II):AdcR_dimer, 5:1 binding model with K_Zn1=K_Zn2 = 10¹¹ M⁻¹, K_Zn3=1.4 (±0.2) × 10⁸ M⁻¹, K_Zn4=3.00 (±0.03) × 10⁷, K_Zn5=3.8 (±0.8) × 10⁶ M⁻¹. The dot and dash black lines are simulated curves one order of magnitude larger and lower, respectively. (d) 44 μM AdcR and 17.1 μM quin-2. The solid red line represent non-linear least-square fit to a Zn(II):AdcR_dimer, 2:1 binding model with K_Zn1=K_Zn2 = 1.4 (±0.2) × 10¹² M⁻¹. The black dot and dashed line are simulated data to K_Zn1=K_Zn2=10¹³ and 10¹² M⁻¹, respectively.

Fig. 4

Binding isotherms obtained from titrating Zn(II) into a mixture of AdcR variants and competitor mag-fura-2 at pH 6.0 (a–b) or pH 8.0 (c–d). Closed markers represent ε₃₆₆ values which are maximal in the apo-mf2 and open markers represent ε₃₂₅ values that are maximal in the Zn(II)-mf2 complex. (a) 34.4 μM H108A AdcR and 28.6 μM mf2 and (b) 30 μM H112A AdcR and 30 μM mf2. The red solid line represent a non-linear least-square fit to stoichiometric binding of Zn(II) to mf2. (c) 25.3 μM H108A and 22.1 μM mf2. The red solid line represents a non-linear least-square fit to Zn(II):AdcR_dimer, 5:1 binding model where K_Zn1=K_Zn2=1.17×10⁹ M⁻¹, K_Zn3=1.77 (±0.1) × 10⁷ M⁻¹, K_Zn4=9.0 (±0.9) × 10⁶ M⁻¹, K_Zn5=1.8 (±0.5) × 10⁶ M⁻¹. (d) 41 μM H112A and 27 μM mf2. The red solid line represent a non-linear least-square fit to Zn(II):AdcR_dimer, 5:1 binding model where K_Zn1= 3.5 × 10⁹ M⁻¹ (fixed), K_Zn2=1.2 (±0.5) × 10⁹ M⁻¹, K_Zn3=9 (±1) × 10⁷ M⁻¹, K_Zn4=9 (±1) × 10⁶ M⁻¹, K_Zn5=2.0 (±0.6) × 10⁶ M⁻¹.

Fig. 5

X-ray absorption spectroscopic data of Zn(II)-bound AdcR. (Top) Zn K-edge X-ray absorption spectra of Zn(II)-bound WT AdcR at pH 6 (black), C30A mutant (green), and WT AdcR at pH 8 (blue). (Bottom) EXAFS Fourier transform (k³-weighted, k = 2–13 Å⁻¹ of WT AdcR at pH 6. (Inset) k³-weighted EXAFS spectra for WT AdcR at pH 6. For bottom panel, experimental data are black and best-fit simulations (Table 3, Fit 1) are red.

Fig. 6

¹H-¹⁵N TROSY spectra of wild-type AdcR in the absence (a) and presence (b) of Zn(II). Sequence-specific resonance assignments are shown for each conformer, with the single-contour red crosspeaks in panel (b) corresponding to those of apo-AdcR. (c) Secondary structural analysis of apo- and Zn(II)-bound AdcR as determined by analysis of the backbone chemical shifts in TALOS. Backbone resonances for residues R2, T37, R136 and K146 are unassigned in apo-AdcR. Likewise, backbone resonances for residues L17, L46, P103, Q121, P128, Q135 and K146 are unassigned in Zn(II)-bound AdcR.

Fig. 7

Chemical shift (¹H, ¹⁵N) perturbation maps |apo – Zn| of AdcR variants (35 °C, pH 6.0).

Fig. 8

Representative isotherms of the binding of Zn(II)-AdcR variants to a fluorescein-labeled 28-mer double stranded DNA containing the AdcO sequence. The anisotropy change is normalized to that observed for wild-type AdcR (circles); for all the other variants, C30A (squares), H42A (diamonds), H108A (triangles) and H112A (gridded square) AdcRs, the fractional change in anisotropy was calculated relative to the total change observed for wild-type AdcR. The Zn(II)-activated binding is reversible at the end of the experiment by addition of 500 μM EDTA (not shown). The red continuous lines represent non-linear least-square fits for the binding of a non-dissociable dimer to DNA with the affinities (K_{DNA Zn}) complied in Table 4.