Towards engineering hormone-binding globulins as drug delivery agents

Abstract

The treatment of many diseases such as cancer requires the use of drugs that can cause severe side effects. Off-target toxicity can often be reduced simply by directing the drugs specifically to sites of diseases. Amidst increasingly sophisticated methods of targeted drug delivery, we observed that Nature has already evolved elegant means of sending biological molecules to where they are needed. One such example is corticosteroid binding globulin (CBG), the major carrier of the anti-inflammatory hormone, cortisol. Targeted release of cortisol is triggered by cleavage of CBG's reactive centre loop by elastase, a protease released by neutrophils in inflamed tissues. This work aimed to establish the feasibility of exploiting this mechanism to carry therapeutic agents to defined locations. The reactive centre loop of CBG was altered with site-directed mutagenesis to favour cleavage by other proteases, to alter the sites at which it would release its cargo. Mutagenesis succeeded in making CBG a substrate for either prostate specific antigen (PSA), a prostate-specific serine protease, or thrombin, a key protease in the blood coagulation cascade. PSA is conspicuously overproduced in prostatic hyperplasia and is, therefore, a good way of targeting hyperplastic prostate tissues. Thrombin is released during clotting and consequently is ideal for conferring specificity to thrombotic sites. Using fluorescence-based titration assays, we also showed that CBG can be engineered to bind a new compound, thyroxine-6-carboxyfluorescein, instead of its physiological ligand, cortisol, thereby demonstrating that it is possible to tailor the hormone binding site to deliver a therapeutic drug. In addition, we proved that the efficiency with which CBG releases bound ligand can be increased by introducing some well-placed mutations. This proof-of-concept study has raised the prospect of a novel means of targeted drug delivery, using the serpin conformational change to combat the problem of off-target effects in the treatment of diseases.

Figure 1. Thrombin specificity assay.

Wild type and engineered CBG were incubated with thrombin. Only CBG-antitrypsin_Pittsburgh and CBG-Thr349Arg were successfully cleaved by thrombin, while wild type CBG remained unreactive. Lane 1: wild type CBG (negative control); Lane 2: wild type CBG + human neutrophil elastase; Lane 3: wild type CBG + thrombin; Lane 4: CBG-antitrypsin_Pittsburgh (negative control); Lane 5: CBG-antitrypsin_Pittsburgh + human neutrophil elastase; Lane 6: CBG-antitrypsin_Pittsburgh + thrombin; Lane 7: CBG-Thr349Arg (negative control); Lane 8: CBG-Thr349Arg + human neutrophil elastase; Lane 9: CBG-Thr349Arg + thrombin; Lane 10: thrombin; Lane 11: human neutrophil elastase. Lanes 5 and 8 show that further engineering would be required to remove elastase cleavage.

Figure 2. Prostate specific antigen (PSA) assay.

Engineered CBG variants were incubated with PSA. In both gels, Lane1: uncleaved CBG; Lane 2: 2 hours incubation with PSA; Lane 3: 4 hours incubation with PSA; Lane 4: 8 hours incubation with PSA; Lane 5: 16 hours incubation with PSA; Lane 6: 24 hours incubation with PSA; Lane 7: 48 hours incubation with PSA; Lane 8: 72 hours incubation with PSA. (A) Of the two variants of PSA-sensitive CBG, the one with the reactive centre loop sequence EEGVDTAGSSYYSGNLTSKPII is cleaved the more rapidly. (B) CBG-EEGVDTAGSALLSSDITSKPII.

Figure 3. Ligand binding pocket of the hormone binding globulins.

(A) The ligand binding pocket (solid surface, PDB entry 2VDY [19]) is formed by Helix A (yellow), Helix D (blue), β-sheet B (red), the s2B/s3B loop (orange) and the s4B/s5B loop (green). (B) R-state human CBG (PDB 2VDY , chain A) showing the interactions of pocket residues with cortisol. (C) Corticosteroid binding globulin (PDB 2VDY, chain A), and (D) Thyroxine binding globulin (PDB 2CEO , chain A). Electrostatic map: negative potential (red); positive potential (blue), uncharged/hydrophobic (white). (C) and (D) (inset). Orientation represented in the electrostatic maps, with the key secondary structural elements coloured as in (A).

Figure 4. Reactive centre loop P14 residue.

The P14 residue in CBG (PDB entry 2VDY [19]), Val 336, shown in yellow spheres, is believed to displace Tyr 235 (blue spheres) on the s2B/s3B loop when the reactive centre loop is inserted into the central β-sheet A (red), leading to conformational changes in the binding pocket (pale green). CBG's interaction with cortisol is provided by a network of side chains (bright green ball-and-stick representation), which can be perturbed by any slight movement in the pocket and its surrounding regions.

Figure 5. CBG-S100C-V236C-T349R structure.

(A) Under non-reducing conditions, residues 100 and 236 form a disulphide bridge. Ligand binding pocket elements are identified by the same colour scheme as Figure 3. In this figure, cortisol from a previous structure (PDB 2VDY [19]) was superimposed on our unliganded structure (PDB 4C41, this work) to highlight the position of the pocket. It is represented as a white silhouette bound by broken lines. (B) When the disulphide bridge is formed, there is an obvious distortion of the steroid binding pocket, as shown here by superimposing the R-state structures of CBG (pale blue, transparent) and CBG-S100C-V236C-T349R (coloured as in A).

Figure 6. The conserved serpin fold.

The canonical serpin fold comprises three β-sheets (yellow), eight to nine α-helices (red) and the reactive centre loop (blue). The proposed sites of the new ligand binding pocket are highlighted in blue (bounded by β-sheet A and helix D) and magenta (bounded by β-sheet A and helix F). (A) The reactive centre loop intact S-state of a typical serpin. (B) The reactive centre loop cleaved R-state of a serpin, showing the insertion of the loop into the middle of β-sheet A.