The recombinant expression of RTD-1 was carried out by using a modified protein splicing unit to assist the intramolecular native chemical ligation (NCL) required for backbone cyclization.23, 27
RTD-1 has six Cys residues that may be used for cyclization. In order to facilitate the cyclization reaction, we decided to use the Cys residues located in positions 3 and 7 (). These Cys residues are the only ones in the sequence of RTD-1 that do not have a charged or β-branched residue N-terminally adjacent, which should facilitate the kinetics of the cyclization reaction without affect the splicing activity of the intein. Accordingly, two different RTD-1 linear precursors (RTD-C3 and RTD-C7) were cloned in frame with a modified Mxe
Gyrase intein (). The N-terminal Met residue in both constructs is efficiently removed by endogenoeus methionine aminopeptidases when expressed in Escherichia coli21, 22
therefore yielding the required N-terminal Cys for intramolecular NCL. At the same time, the modified intein allows the generation of the required α-thioester at the junction between the C-terminal end of RTD-1 and the intein ()
Both RTD-1 intein fusion protein precursors (RTD-C3 and RTD-C7) were expressed in E. coli BL21(DE3) cells at 30°C for 3 h, and purified by affinity chromatography using chitin-sepharose beads. The RTD-intein precursors have a chitin binding domain (CBD) fused at the C-terminus of the intein domain to facilitate purification. As shown in , both intein precursors had comparable levels of expression in E. coli cells (≈ 5 mg/L as estimated by UV spectroscopy), and showed similar rates and propensities for in-vivo cleavage (≈ 65%). It is worth noting that under these conditions most of the intein fusion precursors in both cases were expressed as soluble proteins ().
Figure 2 In vitro production of RTD-1. A. SDS-PAGE analysis of cell lysates from BL21 E. coli cells expressing precursors RTD-C3 and RTD-C7. The identity of the bands corresponding to intein precursors and in vivo cleaved proteins are shown on the right. The positions (more ...)
We next tested the ability of the different precursors to be cleaved in vitro
by using reduced glutathione (GSH) to produce folded RTD-1. GSH has been shown to promote cyclization and concomitant folding when used in the biosynthesis of Cys-rich cyclic polypeptides.21, 25, 26
The cyclization/folding reaction was performed on the chitin beads where the corresponding precursors had been purified. The best cleavage/cyclization conditions were accomplished using 100 mM of GSH in phosphate buffer at pH 7.2 for 48 h at room temperature. Under these conditions both RTD-intein precursors were completely cleaved (). HPLC analysis of the crude cyclization mixture revealed that in both cases the main peptide product was the corresponding folded RTD-1 () as revealed by HPLC and ES-MS analysis (). Other peptide minor peaks in the HPLC chromatograms were identified as not correctly folded GSH-adducts. Precursor RTD-C3 gave the best cyclization/folding yield producing around 10 μg/L of purified RTD-1. This yield corresponds to ≈10% of the theoretical value. The in vitro
cyclization/folding of precursor RTD-C7 gave a slightly lower yield (≈7 μg/L of purified RTD-1). The lower yield observed for this precursor can be attributed to the slightly more complex cyclization/folding crude ()
Recombinant RTD-1 was purified by HPLC and its biological activity tested using an Anthrax lethal factor (LF) protease inhibitory assay.28
produced RTD-1 was able to inhibit anthrax LF with a IC50
of 384 ± 33 nM (), which corresponds to a Ki
≈ 0.4 μM under the conditions used in the inhibitory assay.36,37
value closely parallels previously reported values,20, 29
thus confirming the biological activity of the recombinantly produced θ-defensin.
Figure 3 Characterization of recombinantly produced θ-defensin RTD-1. A. Inhibition assay of RTD-1 against Anthrax lethal factor (LF). Different concentrations of RTD-1 were tested against LF. At each concentration, residual LF activity was measured and (more ...)
We also used NMR spectroscopy to confirm that recombinant RTD-1 adopted a native θ-defensin fold. The structure of native RTD-1 has been previously reported by Craik in 10% MeCN aqueous buffer at pH 4.5.6
We decided to carry out the NMR experiments in more physiological conditions using an aqueous buffer containing no organic solvents at pH 6.5. The rationale to use more physiological conditions was to allow the future study of biologically relevant interactions between RTD-1 and potential biomolecular targets by NMR. The chemical shifts of the assigned backbone amide and alpha protons (HNα and HCα) for recombinant RTD-1 at pH 6.5 were very similar to those reported earlier by Craik at pH 4.5 (Table S2
). No significant (≥0.3 ppm) differences were found in the backbone amide protons. We also saw a uniform shift rather than variable changes for the backbone alpha protons (≈0.2 ppm) at pH 6.0 (Table S2
). Since backbone alpha protons usually reflect the secondary structure of the peptide backbone, this uniform offset in the resonances of the backbone HCα protons could be attributed to the different buffer conditions used in the two samples rather than changes in secondary structure.
We also produced recombinant 15N-labeled RTD-1 by expressing RTD-C3 precursor in minimal M9 medium containing 15NH4Cl as the only source of nitrogen. Under these conditions the expression yield was around 7 μg/L of 15N-labeled RTD-1 after purification by HPLC. The HSQC spectra of recombinant RTD-1 was very well dispersed, indicating a well-folded structure (). Having access to the recombinant expression of θ-defensins allows the introduction of NMR active isotopes (15N and/or 13C) in a very inexpensive fashion, thus facilitating the use of heteronuclear NMR to study intermolecular interactions between θ-defensins and their biomolecular targets.
Encouraged by the results obtained with the in vitro
GSH-induced cyclization/folding of the RTD-intein precursors we also decided to explore the expression of folded RTD-1 inside E. coli
cells. RTD-1 has been shown to be antimicrobial against both Gram-positive and Gram-negative bacteria. However, the antimicrobial activity of RTD-1 has been shown to strongly depend on the conditions used in the antimicrobial assays. The presence of 10% human serum has been shown to significantly decrease the antimicrobial properties of RTD-1 especially against Gram-negative bacteria such as E. coli
. We anticipated that the high molecular complexity of the bacterial cytosol could decrease the antimicrobial activity of RTD-1 when produced intracellularly and therefore allow its production in the cellular cytosol.
In order to test this hypothesis we used the RTD-C7 precursor (). In-cell expression of RTD-1 was accomplished in Origami2(DE3). These cells have mutations in the thioredoxin and glutathione reductase genes, which facilitates the formation of disulfide bonds in the bacterial cytosol.30
We have recently used these cells for the in vivo
production of several disulfide-containing backbone cyclized polypeptides.22, 25, 26
Precursor RTD-C7 was expressed in Origami2(DE3) overnight at room temperature giving a total yield of precursor protein of ≈3 mg/L. Under these conditions the precursor was completely cleaved in vivo
(). In contrast when precursor RTD-C3 was expressed under these conditions only ≈74% of the precursor protein was cleaved in vivo
(). This difference could be attributed to the proximity of Cys5
to the RTD-intein junction in the precursor RTD-C7, which is only one residue away from the RTD-intein junction therefore facilitating the intramolecular cleavage. In the precursor RTD-C3, the closest Cys residue within the sequence (Cys16
) is four residues away from the RTD-intein junction (). After lysing the cells expressing RTD-C7, both the insoluble and soluble cellular fractions were analyzed by HPLC-MS/MS in multiple reaction mode to identify and quantify the amount of folded RTD-1. Interestingly, almost all of the folded RTD-1 produced inside living E. coli
cells was found in the insoluble fraction (≈22 μg/L, ). θ-Defensins are known to interact with model phospholipid membranes, and have been shown to have low nM affinity for glyco-proteins (gp120 and cd4) and glyco-lipids (galactosylceramide).19
This could explain the higher affinity of recombinant RTD-1 for the insoluble cellular lysate fraction, where it is likely to find insoluble membrane fragments containing glyco-lipids and/or membrane-bound peptidoglycan precursors. In fact, peptidoglycan precursor and membrane-anchored lipid II has been shown to bind to human α-and β-defensins.31
We are now also studying the potential interaction between several peptidoglycan precursors and θ-defensins.
Figure 4 In vivo production of RTD-1 in E. coli cells. A. SDS-PAGE analysis of the soluble cell lysate fraction (S) of Origami E. coli cells expressing precursors RTD-C3 and RTD-C7. Expression was induced as described in the text at room temperature for 18 h (S: (more ...)
In summary, we report in this work the first recombinant expression of the θ-defensin RTD-1 in E. coli
cells. RTD-1 can be produced either in vitro
, by processing of the corresponding RTD-intein precursor, or in vivo
by using longer expression times in combination with Origami or similar E. coli
cells. The yield of purified RTD-1 was greater in vivo
than in vitro
(≈ 22 μg vs. ≈ 10 μg of purified peptide/L of bacterial culture). Although the yields are modest, our approach represents an improvement over the approach reported by Suga for the ribosomal synthesis of RTD-1 using a cell-free expression system in combination with genetic code reprogramming.29
The bacterial expression yields could be easily improved by using richer media and/or fermentors to allow expression at higher cellular densities. The bacterial expression of RTD-1 allows the incorporation of NMR-active isotopes in a very inexpensive manner to facilitate the structural study of the interactions between θ-defensin and potential biomolecular targets through heteronuclear NMR. For example, we are using 15
N-labeled RTD-1 to study the potential interaction of this defensin to a soluble form of lipid II. More interestingly, in-cell expression of RTD-1 gave a yield of ≈22 μg/L, which corresponds to an intracellular concentration of ≈4 μM. This result opens the intriguing possibility for the generation of genetically-encoded libraries using the θ-defensin scaffold. These libraries could be used for the development of in-cell screening and directed evolution technologies to further study the biological activity of θ-defensins or to engineer new biological activities on the defensin scaffold.