Biosensing Tacrolimus in Human Whole Blood by Using a Drug Receptor Fused to the Emerald Green Fluorescent Protein

Tacrolimus (FK506) is an immunosuppressant drug (ISD) used to prevent organ rejection after transplantation that exhibits a narrow therapeutic window and is subject to wide inter- and intra-individual pharmacokinetic fluctuations requiring careful monitoring. The immunosuppressive capacity of FK506 arises from the formation of a complex with immunophilin FKBP1A. This paper describes the use of FKBP1A as an alternative to common antibodies for biosensing purposes. Bioassays use recombinant FKBP1A fused to the emerald green fluorescent protein (FKBP1A–EmGFP). Samples containing the immunosuppressant are incubated with the recombinant protein, and free FKBP1A–EmGFP is captured by magnetic beads functionalized with FK506 to generate a fluorescence signal. Recombinant receptor–drug interaction is evaluated by using a quartz crystal microbalance and nuclear magnetic resonance. The limit of detection (3 ng mL–1) and dynamic range thus obtained (5–70 ng mL–1) fulfill therapeutic requirements. The assay is selective for other ISD usually coadministered with FK506 and allows the drug to be determined in human whole blood samples from organ transplant patients with results comparing favorably with those of an external laboratory.


Figure S1
S2 Table S1 S2 Figure S2 S2 Figure S3 S3 Figure S4 S3 Protein identification by MALDI-TOF mass spectrometry S4 Table S2 S4 Figure S5 S5 Preparation of the FK506-protein platform on the QCM sensor S6 Figure S6 S8 Table S3 S8 Figure S7 S10 Figure S8 S10 Figure S9 S11 Table S4 S11 Table S5 S12 Table S6 S14 References S15 Figure S1. Chemical structures of the immunosuppressant drugs tacrolimus (FK506), mycophenolic acid (MPA) and sirolimus (Sir), and the carboxylated derivative of tacrolimus (FK506-CO2H). Table S1. Primer sets used for PCR amplification of the pQE-T7-2 vector and gene fragments. The oligo tail added to the target gene sequence is in bold face, the segment identical with the target gene in italics and the overlapping regions for the Gibson assembly reaction underlined.

Sir FK506
FK506-CO 2 H MPA Figure S3. SDS-PAGE analysis of the purified fusion protein with Coomassie brilliant blue protein staining: lane 1 and 9, molecular marker (Thermo Scientific™ PageRuler™ Prestained Protein Ladder, 10 to 180 kDa); 2, lysate sample; 3, HisTrap column throughput with non-retained proteins; 4-7, aliquots from the HisTrap column elution; 8, purified protein obtained after PD-10 column. Prior to analysis, a 5-minute boiling step at +95 °C have been used. The thickest band corresponds to the molecular weight of the fusion protein (~40.6 kDa).  MALDI-TOF MS analysis of the electrophoretically resolved protein was performed, at the   Proteomics Unit of the Complutensian University of Madrid, on a 4800 Plus Proteomics   Analyzer MALDI-TOF/TOF mass spectrometer (Applied Biosystems, MDS Sciex, Toronto, Canada). The protein was excised from the SDS-PAGE gel, in-gel reduced, alkylated and digested with trypsin. The MALDI-TOF/TOF MS technique provides peptide mass fingerprints ( Figure S5A), those peptides were collated and represented as a list of monoisotopic molecular weights and compared with the masses of theoretical trypsin digestion of the recombinant protein ( Figure S5B and Table S2). The results showed a coverage of 63% of the protein, having the FKBP1A domain a coverage of more than 87% of the primary sequence. Furthermore, the EmGFP domain was also observed except the C-terminal end that contains low number of Lys and Arg avoiding the digestion by Trypsin that produces too large peptides to be identified by MS.

Preparation of the FK506-protein platform on the QCM sensor and data analysis
QCM data analysis. The sensed mass of adsorbed protein layers was estimated using the Sauerbrey equation 1 (Eq. S1) and the Voigt model 2 supplied by QCMBrowser 2.30 software (KSV Instrument Ltd, Helsinki, Finland).
where n is the overtone number and ρq and μq are the density and the shear modulus of the quartz crystal, respectively.
While the Sauerbrey equation was used for calculating the sensed mass of the rigid film with small or no energy dissipation, the Voigt model was applied in the case of larger dissipation values corresponding to soft and elastic layers. 3 The surface density of the sample at saturation, and the relative molecular area occupied on the surface, Ā, was obtained from Eq. S2 where MW is the molecular weight of the adsorbed molecule, (Δm/A)Sat is the surface mass density at monolayer saturation and NA is the Avogadro number.
Thickness of the adsorbed layer was determined from the detailed analysis of all overtones for the QCM-Z set-up using the QCMBrowse software and by simple estimation of the layer thickness using an average density of the adsorbed proteins obtained from QCMBrowse data fitting.
The data were analyzed with Igor 6.10A software (Wavemetrics, Lake Oswego, OR, USA).
The calculation of the kinetic parameters was done with a Langmuir binding model (Eq. S3): Eq. S3 where (Δm/A)Sat is the saturation surface mass density, k'd is the pseudo binding constant for FK506 binding to FKP1A containing proteins.
Formation of dithiobis(C2NTA-Ni 2+ ) nanoplatforms. Self Assembled Monolayer (SAM) of dithiobis(C2NTA)-Ni 2+ complex was formed on the QCM sensors using a procedure previously described. 4 Briefly, Au coated AT-cut 5 MHz sensor (Nordtest.srl, Serravalle Scrivia (AL), Italy) were cleaned with a drop of chromic acid solution for a few seconds, rinsed copiously with water and dried with a nitrogen flux. Then, the SAM was obtained with a two-step procedure: the chemisorption of dithiobis(C2NTA) covalently immobilized on the sensor surface via sulfur-gold binding, followed by the addition of a Ni2SO4 solution resulting in complexation of Ni 2+ to the water-exposed NTA groups.

S7
The average orientation of the dithiobis(C2NTA-Ni 2+ ) covalently immobilized on the sensor surface is reported in Figure S6. The procedure provides densely packed and rigid SAM of dithiobis(C2NTA-Ni 2+ ) layers on gold-coated QCM sensors, in such nanoplatform Ni-NTA functional groups protrude exposing the two chelating aquoions available for further complexation.
The principal parameters obtained from the adsorption (summarized in Table S3  shows that adsorption of the protein does not significantly affect the viscoelastic properties of the SAM layer. This finding suggests that a compact protein layer is formed on the surface. Figure 3C, the adsorption of the FKBP1A-EmGFP takes around 2.5 h to be completed but the kinetics does not follow a simple mono-exponential dependence on time as expected for Langmuir adsorption. The presence of different kinetic regimes for protein adsorption was observed also by other authors. 7 In the first regime, the protein is arranged in a flat conformation occupying a large surface area, as adsorption proceeds the flip-up from the flat-lying conformation to a tilted one is promoted, further His-tagged proteins bind to the platform saturating the binding sites. However, the application of this model to our systems was not completely satisfactory, and the best agreement was found using the tri-exponential function (Eq. S4)

As shown in
where y0 is the saturation value, x0 is the x offset, A1, A2 and A3 are the amplitude and τ1, τ2 and τ3 are the time constants expressed in seconds.
The reported data show that the first two-steps for protein adsorption usually proposed in the literature, namely chemisorption in flat-lying configuration (τ1 = 13.569 s) and tilting up (τ2 = 168.71 s) of the molecule for dense monolayer packing, is followed by a slower reorganization of the two protein subunits in the surface layer (τ3 = 1539 s) to reach the maximum packing density.

S8
The adsorbed mass of the protein, the area occupied by each molecule at surface saturation, the thickness increment caused by the adsorption of the protein and the change in dissipation obtained from the experimental data are reported in Table S3.
The average molecular area for the protein is in accordance with theoretical calculations reported in literature for similar systems containing EmGFP histidine tagged protein, where the average area ranges between 500 and 1250 Å 2 /molecule depending on the orientation of the protein on the surface. 8 Further insight in the protein conformation at the interface was obtained from the comparative analysis of the structural experimental data (thickness and molecular area at saturation) with the physical dimensions estimated from the chemical structure of the proteins. Given that the kinetic measurements proved that the proteins do not lay flat at the interface at maximum coverage, such conformations were excluded in the computations.
Under this assumption, comparison of the experimental value of Table S3 and the theoretical calculation allows us to draw some general conclusions. Figure S6C reports the most likely conformation of the system, compatible with the multistep kinetic. The FKBP1A protein is vertically oriented while the long axis of EmGFP is oriented with a tilt angle of 10° with respect to the normal to the interface.         TRFIA: time-resolved fluorescence immunoassay; WB: whole blood.