Characterization of inhibitor binding sites on VKOR enzymes in

DKO cells were generated to abolish endogenous VKOR activity and were proven to be unable to activate FIX at a given vitamin K concentration (4.1.1). To study the effect of mutant VKORC1 or VKORC1L1, the bicistronic vector pIRES was used which allows the simultaneous expression of two proteins separately but from the same mRNA transcript.

The cDNA of F9 was cloned into multiple cloning site A (MCS A), whereby MCS B har-bors cDNA of VKORC1 or VKORC1L1 and the respective variants. Cloning was per-formed as described in sections 3.1.1 and 3.1.2.

For warfarin and fluindione, OACs representing the two groups of investigated com-pounds, dose-response curves were generated for overexpression of VKORC1 and VKORC1L1. In addition, loop swap variants were made. The alignment of loop regions of both enzymes showed 48 % identity (as indicated by asterisk in Figure 31) and 62 % similarity (as indicated by bold letters). The VKORC1 loop swap comprised amino acids Lys30 to Ser79 changed to Glu37 to Pro86 of VKORC1L1 and vice versa (Figure 31).

Figure 31: Alignment of loop region of VKORC1 and VKORC1L1.

Human VKORC1 (NP_076869.1) protein sequence was aligned with sequence of VKORC1L1 (NP_775788.2) using NCBI blast tool. Amino acids 30 to 79 of VKORC1 aligned to amino acids 37 to 86 of VKORC1L1. Stars indicate identical amino acids, bold letters (grey background) rep-resent amino acids with similar chemical properties.

Endogenous VKORC1 showed higher susceptibility to warfarin and fluindione compared to VKORC1L1. In DKO HEK 293T cells, overexpression of VKORC1 presented highest susceptibility to warfarin and fluindione and VKORC1L1 required higher concentration for half-maximal inhibition (Figure 32). Calculations of IC₅₀ values are given in Table 4.

Thus, VKORC1 was about 27-fold more sensitive to warfarin treatment compared to VKORC1L1. Fluindione showed 6-fold higher potency to inhibit VKORC1. VKORC1 loop swap gave similar curves like those of VKORC1L1 WT for both compounds investigated which is also represented by similar IC₅₀ values (e.g. VKORC1L1 WT: 474 nM and VKORC1 loop swap: 215 nM for warfarin). When VKORC1 loop was swapped into VKORC1L1 similar results were obtained, i.e. VKORC1L1 loop swap acted like VKORC1 WT with respect to warfarin and fluindione.

10^-3 10^-2 10^-1 10⁰ 10¹

0 20 40 60 80 100

IC50

Warfarin concentration in µM

Normalized FIX activity (%)

10^-3 10^-2 10^-1 10⁰ 10¹

0 20 40 60 80 100

IC50

Fluindione concentration in µM

Normalized FIX activity (%)

Figure 32: Inhibition curves for warfarin and fluindione assessed in DKO cells.

Activity of VKORC1 (red) and VKORC1 loop swap (black), VKORC1L1 (blue) and VKORC1 loop swap (green) determined in DKO cells by overexpression of different constructs. Measurements were performed in triplicates, values are shown as mean and error bars are represented as SEM.

Table 4: IC50 values of warfarin and fluindione.

DKO HEK 293T cells transfected with VKORC1 and VKORC1L1 and its respective loop swaps were supplemented with 12 µM K1 and warfarin to assess IC50 valuesby means of FIX cell-based assay.

VKOR variant IC50 (nM)

Warfarin Fluindione

VKORC1 WT 16.8 121

VKORC1 loop swap 251 572

VKORC1L1 WT 589 738

VKORC1L1 loop swap 16.8 309

The previous experiment showed that the loop is important at least in warfarin and flu-indione binding. To further limit the drug binding site, different modifications in VKORC1 and VKORC1L1 were tested. First, it was proven that Phe55 in VKORC1 is essential in warfarin binding [21]. Therefore, phenylalanine 55 was substituted by alanine (Phe55Ala). As presented in Figure 33, this variant was resistant to warfarin treatment as concentrations up to 5 µM resulted only in ~40% reduction of normalized FIX activity.

10^-3 10^-2 10^-1 10⁰ 10¹

0 20 40 60 80 100

IC50

Warfarin concentration in µM

Normalized FIX activity (%)

Figure 33: Dose-response for warfarin of VKORC1 Phe55Ala (red curve) assessed in DKO cells.

For comparison, VKORC1 WT is included (black curve). Measurements were performed in tripli-cates, values are shown as mean and error bars are represented as SEM.

By sequence alignment the corresponding amino acid of VKORC1 Phe55 in VKORC1L1 was identified (Figure 34). To test the hypothesis that by introducing the “same” amino acid responsible for warfarin binding in VKORC1 would lead to a more sensitive en-zyme, a mutant of VKORC1L1 at Leu62 was generated (Leu62Phe). Furthermore, the amino acids 60 to 62 of VKORC1L1 were changed to corresponding amino acids 53 to 55 of VKORC1 (Ala60Arg, Ala61Val, Leu62Phe, see Figure 34), which represent the putative contact surface II of VKORC1 identified by previously published in silico analy-sis [21].

Figure 34: Alignment of loop region of VKORC1 with VKORC1L1.

Amino acids 37 to 78 of VKORC1 aligned to amino acids 44 to 85 of VKORC1L1. Star indicates identical amino acids, bold letters represent amino acids with similar chemical properties. Red box shows contact surface II in VKORC1 identified by in silico analysis as published by 21.

VKORC1 Phe55 is highlighted in yellow.

The inhibition curves for VKORC1L1 Leu62Phe and the triple mutant comprising amino acids 60 to 62 are shown in Figure 35. Single and/or triple variant of VKORC1L1 was anticipated to move IC₅₀ towards VKORC1, since more sensitive enzyme inhibition was expected. On the contrary, both constructs showed shift of dose-response curves to higher warfarin concentration. VKORC1L1 Leu62Phe required 3-fold higher warfarin for half-maximal inhibition compared to VKORC1L1 WT (Table 5). Moreover, the VKORC1L1 triple variant needed almost 4 times higher warfarin concentration for inhibi-tion compared to VKORC1L1 WT.

10^-2 10^-1 10⁰ 10¹

0 20 40 60 80 100

IC50

Warfarin concentration in µM

Normalized FIX activity (%)

Figure 35: Inhibition curves for warfarin of VKORC1L1 variants assessed in DKO cells.

Normalized FIX activity is shown for VKORC1L1 Leu62Phe (blue) and Ala60Arg+Ala61Val+Leu62Phe (red). For comparison, VKORC1L1 WT is included (black curve) Measurements were performed in triplicates and values are shown as mean +/- SEM.

Table 5: IC50 values of warfarin for different VKORC1L1 variants.

DKO HEK 293T cells transfected with VKORC1L1 variants were supplemented with 12 µM K1

and warfarin to assess IC50 valuesby means of FIX cell-based assay.

VKORC1L1 variants IC50 (nM)

WT 474

Leu62Phe 1543

Ala60Arg+Ala61Val+Leu62Phe 1789

The analysis of different OACs revealed different susceptibilities of the two enzymes investigated. By alignment of protein sequence and further in vitro analysis it was obvi-ous that to some extent the loop plays an important role in drug binding. Furthermore, these results suggest that warfarin binding site of VKORC1L1 is different from VKORC1.

On this occasion in silico analysis was performed to identify regions or single amino ac-ids being responsible for different drug binding. Details of in silico analysis and subse-quent in vitro experiments are described in the next section.