Retinol binding protein IV purified from Escherichia coli using intein-mediated cleavage as a suitable replacement for serum sources

Abstract

Retinol binding protein IV (RBP) functions as the principal carrier of retinol (Vitamin A) in the blood, where RBP circulates bound to another serum protein, transthyretin. Isolation of pure RBP from the transthyretin complex in human serum can be difficult, but expression of RBP in recombinant systems can circumvent these purification issues. Human recombinant RBP has previously been successfully expressed and purified from E. coli, but recovery of active protein typically requires extensive processing steps, such as denaturing and refolding, and complex purification steps, such as multi-modal chromatography. Furthermore, these methods produce recombinant proteins, often tagged, that display different functional and structural characteristics across systems. In this work, we optimized downstream processing by use of an intein-based expression system in E. coli to produce tag-free, human recombinant RBP (rRBP) with intact native amino termini at yields of up to ~15 mg/L off column. The novel method requires solubilization of inclusion bodies and subsequent oxidative refolding in the presence of retinol, but importantly allows for one-step chromatographic purification that yields high purity rRBP with no N-terminal Met or other tag. Previously reported purification methods typically require two or more chromatographic separation steps to recover tag-free rRBP. Given the interest in mechanistic understanding of RBP transport of retinol in health and disease, we characterized our purified product extensively to confirm rRBP is both structurally and functionally a suitable replacement for serum-derived RBP.

Keywords: Inteins; Oxidative refolding; Retinol; Retinol binding protein; Transthyretin.

Figure 1:

PDB Entry 5NU7. X-ray structure of serum RBP with bound retinol (ROH). Both ROH and RBP tryptophans are shown in black.

Figure 2:

Process flow diagram for rRBP expression and purification.

Figure 3:

A) SDS-PAGE stained with Coomassie Brilliant Blue R-250. Lanes 1 and 7, molecular weight standard; lane 2, cell lysate; lane 3, lysate flow through; lane 4, wash effluent; lane 5, ROH-rRBP after intein cleavage; lane 6, chitin bead regeneration effluent. The approximate expected position of the fusion protein (45 kDa) and rRBP (21 kDa) are indicated by labels. B) Western blot of apo-rRBP, serum-derived RBP and human albumin controls (75 ng of protein loaded per well). Lanes 5 – 8 were run on the same gel as lanes 1 – 4 but were separated after blocking to serve as no-primary controls processed in parallel with lanes 1 – 4. Lanes 1 and 8, MagicMark molecular weight standard; lanes 2 and 5, serum-derived RBP; lanes 3 and 6, human recombinant albumin; lanes 4 and 7, apo-rRBP. The expected molecular weight of human recombinant albumin is ~ 67 kDa.

Figure 4:

DEAE chromatogram of eluted ROH-rRBP at 280 nm (solid line) and 330 nm (dashed line). The absorbance data were collected independently by two successive injections of the same stock of eluted ROH-rRBP, and each chromatogram was normalized by the maximum signal for that particular run. The dotted line indicates the salt gradient applied to elute the protein fractions.

Figure 5:

Absorbance spectra of DEE and reconstituted rRBP: 12.5 μM apo-rRBP (dashed line); 12.5 μM RA-rRBP (dotted line); 12.5 μM ROH-rRBP (solid line).

Figure 6:

Circular Dichroism (CD) spectra of apo-rRBP at 25° C. The sample was prepared in phosphate buffer (10 mM sodium phosphate, 150 mM NaF, pH 7.4) at a final protein concentration of 0.1 mg/mL (near UV) or 0.05 mg/mL (far UV). Data were corrected for concentration by conversion to molar ellipticity and merged.

Figure 7:

Guanidine hydrochloride unfolding of apo-rRBP (open circle), RA-rRBP (filled circle) and ROH-rRBP (open circle with cross hatch). The solid curves are least-square regression fits to a two-state unfolding model (apo-rRBP and RA-rRBP, Eq 2), or a three-state unfolding model (ROH-rRBP, Eq 3).

Figure 8:

Retinol (ROH) binding to apo-rRBP monitored by emission of ROH at 460 nm as an acceptor in resonance energy transfer (RET) from donor rRBP tryptophan. Data are fit by nonlinear regression to a mass action model (Eq 4), assuming one binding site of ROH per molecule of rRBP, yielding a dissociation constant, $K_D^{ROH}$, of 100 ± 30 nM.

Figure 9:

ROH-rRBP binding to TTR measured by fluorescence anisotropy. Data are fit to a mass action model (Eq 5) with n independent, identical ROH-rRBP binding sites per TTR molecule. Fitted parameters are $K_D^{TTR}$ = 250 ± 60 nM and n = 1.8 ± 0.2.