Iron uptake pathway of Escherichia coli as an entry route for peptide nucleic acids conjugated with a siderophore mimic

Abstract

Bacteria secrete various iron-chelators (siderophores), which scavenge Fe³⁺ from the environment, bind it with high affinity, and retrieve it inside the cell. After the Fe³⁺ uptake, bacteria extract the soluble iron(II) from the siderophore. Ferric siderophores are transported inside the cell via the TonB-dependent receptor system. Importantly, siderophore uptake paths have been also used by sideromycins, natural antibiotics. Our goal is to hijack the transport system for hydroxamate-type siderophores to deliver peptide nucleic acid oligomers into Escherichia coli cells. As siderophore mimics we designed and synthesized linear and cyclic N^δ-acetyl-N^δ-hydroxy-l-ornithine based peptides. Using circular dichroism spectroscopy, we found that iron(III) is coordinated by the linear trimer with hydroxamate groups but not by the cyclic peptide. The internal flexibility of the linear siderophore oxygen atoms and their interactions with Fe³⁺ were confirmed by all-atom molecular dynamics simulations. Using flow cytometry we found that the designed hydroxamate trimer transports PNA oligomers inside the E. coli cells. Growth recovery assays on various E. coli mutants suggest the pathway of this transport through the FhuE outer-membrane receptor, which is responsible for the uptake of the natural iron chelator, ferric-coprogen. This pathway also involves the FhuD periplasmic binding protein. Docking of the siderophores to the FhuE and FhuD receptor structures showed that binding of the hydroxamate trimer is energetically favorable corroborating the experimentally suggested uptake path. Therefore, this siderophore mimic, as well as its conjugate with PNA, is most probably internalized through the hydroxamate pathway.

Keywords: E. coli outer-membrane receptors; TonB-dependent transport system; iron coordination; peptide nucleic acid (PNA); siderophores.

Figure 1

A scheme of the TonB-dependent E. coli intake systems of catecholate (enterobactin, 2,3-dihydroxybenzoylserine—DHBS) and hydroxamate (ferrichrome, coprogen) siderophores (Andrews et al., 2003; Braun, 2003). Upon recognition by its specific TBDT, the ferric-siderophore is internalized into the periplasm. This is possible due to the TonB-ExbB-ExbD complex which transmits the energy to the Ton box of appropriate TBDT. This energy is generated by the proton-motive force across the inner membrane. Next, the siderophore binds its respective periplasmic binding protein (catecholate type siderophore - FepB and hydroxamate type - FhuD), which delivers it to the dedicated ABC transporter complex (FepCDG for catecholate and FhuBC for hydroxamate type siderophores) for transport across the inner membrane. Once in the cytoplasm, a dedicated reductase (Fes or FhuF) ‘unpacks’ the Fe³⁺ iron by reducing it to Fe²⁺ for which siderophores show lower affinity.

Figure 2

The synthesis scheme of azido-(Orn(Ac,OH))₃ – named S_L.

Figure 3

The synthesis scheme of azido-Cyc(Asp-(Orn(Ac‚OH))₃-GIy-Lys(N₃)) – named S_C.

Figure 4

CD spectra of the natural (ferrichrome and feroxamine) and synthetic (S_L and S_C) siderophores with Fe³⁺ salt solution in phosphate buffer (pH = 7.0). Optimal Λ (solid line) or Δ (dashed line) configuration was observed for siderophore-iron(III) complexes apart from S_C (dashed-dotted line).

Figure 5

Growth recovery assay. Various E. coli mutants were cultured in iron-limiting conditions with or without S_L (at a concentration of 16 μM). The growth of different strains was measured by OD₆₀₀ and normalized to OD₆₀₀ measured for a given strain in non-iron limiting conditions. The experiment was repeated three times on different days. The errors shown are SEM, n = 4–6. The statistical significance of the difference in growth was verified by a two-tailed t-test (**p < 0.01, *p < 0.05, ns: non-significant).

Figure 6

Schematic structure of the S_L-PNA conjugate with the PNA sequences used. The PNA_anti-acpP sequence was used to silence the expression of the acpP gene encoding the acyl carrier protein and PNA_anti-rfp to silence the mrfp1 gene encoding the red fluorescent protein (see text). The control scrambled PNA sequences for PNA_anti-acpP (PNA_control) and PNA_anti-rfp (PNA_control2) are shown with underlined mismatched bases.

Figure 7

Kinetic measurement of the growth recovery assay for the ΔfhuA strain cultured in iron-limiting conditions without or with S_L and its conjugate with PNA_anti-acpP or PNA_control (each at a concentration of 16 μM). The growth was normalized to the OD₆₀₀ measured without the addition of any compound. The experiment was performed in two biological replicates, and two technical replicates each. The errors shown are SEM, n = 4. Statistical significance was tested by the two-way ANOVA test with horizontal lines showing periods for which a significant difference (p < 0.05) was observed between S_L-PNA_anti-acpP and other compounds.

Figure 8

RFP fluorescence silencing in E. coli K-12 wild type and Δfur mutant. Bacteria were cultured in iron-limiting conditions and with the addition of various compounds (at a concentration of 16 μM). Bacterial cells were tested for RFP fluorescence using the flow cytometer with gates set up based on two control cultures: bacteria expressing RFP and bacteria lacking the RFP gene. The experiment was repeated three times on different days. Errors are SEM, n = 3. Statistical significance of the observed differences was determined by a two-way ANOVA test (****p < 0.0001, *p < 0.05, ns, non-significant).

Figure 9

Dose-dependent RFP fluorescence silencing in the E. coli K-12 Δfur mutant. Bacteria were cultured in iron-limiting conditions and with the addition of various compounds (at concentrations ranging from 0 to 16 μM) preloaded with iron. Bacterial cells were tested for RFP fluorescence using the flow cytometer with gates set up based on two control cultures: bacteria expressing RFP and bacteria lacking the RFP gene. Results were normalized to respective Growth Control values. The experiment was repeated three times on different days. Errors shown are SEM, n = 3.

Figure 10

The distances between the Fe³⁺ ion and hydroxamate oxygens or the center of mass (CoM) of the siderophore as a function of the simulation time. The plot shows distances to the six hydroxamate group O-atoms (OH or OZ in the deprotonated cis form with the S_L residue number in parenthesis). The distance to CoM is marked with red circles. The CoM was calculated for all non-hydrogen atoms.

Figure 11

Exemplary conformations in a ball and stick representation from MD simulations of S_L in the presence of Fe³⁺ with either two (A) or three (B) hydroxamate groups close to the Fe³⁺ ion (FE is shown as a brown sphere). The insets show the corresponding systems in van der Waals sphere representations. The Na⁺ ion is light pink. The distances between FE and oxygen atoms are between 2 and 4 Å.

Figure 12

Top scoring poses of S_L docked to FhuE (A) or FhuD (B) structures. Structures from docking are superposed with the crystallographic ligand (coprogen; green ball-and-stick model). Protein residues contacting S_L, defined as residues within 4 Å from the ligand, are shown as gray sticks. The light blue rectangle overlayed on FhuE depicts an approximate position of the E. coli outer membrane. Hydrogen atoms are omitted for clarity.