Construction Of Warfarin Dose Prediction Algorithm And Study Of Its Drug Resistance Mechanism Based On VKORC1 Gene Polymorphism

Objective:Warfarin is one of the most representative coumarin anticoagulants,which has been widely used in clinical due to sufficient evidence-based medical evidence,obvious curative effect,and lower price.However,there are still some pharmacological difficulties during the warfarin therapy,including a narrow safe dose range and significant dose differences among individuals.The construction of predictive models for precise warfarin dosing,combining VKORC1（Vitamin K epoxide reductase complex 1）and CYP2C9（Cytochrome P450,family 2,subfamily C,polypeptide 9）genotype polymorphism with clinical factors,has become a research hot spot in the field of individualized warfarin therapy.VKORC1,the target of warfarin,which contains 163 amino acids and is encoded by VKORC1 gene,is a kind of rate-limiting enzyme that maintains vitamin K circulation and activation of clotting factors.And warfarin,a vitamin K analogue,competitively binds to VKORC1 to inhibit the activation of clotting factors for anticoagulation.CYP2C9 is an important drug metabolizing enzyme.Warfarin used clinically is a racemate mixture,of which S-warfarin is the main active ingredient,mainly metabolized by CYP2C9,with the 3～5 times anticoagulant activity compared with R-warfarin.Therefore,VKORC1 and CYP2C9 genes are considered to be the most important genetic factors that significantly affect the warfarin dose.Specifically,VKORC1 gene polymorphism affects the combination of warfarin and the target,and the metabolic rate varies from CYP2C9 gene polymorphism.At the same time,researchers have established many warfarin pharmacogenomics dose prediction models in combination with age,gender,body weight and other non-genetic factors of patients.Nevertheless,when these models are applied to different ethnic groups,the accuracy varies greatly.The reason why this probably happens is that the genetic heterogeneity among different races causes a problem that the prediction accuracy of race-specific algorithms is reduced when they are extrapolated to other races,it is necessary to develop a localized algorithm for a specific population.In addition,the weight of VKORC1 genotype on warfarin dose was much higher than other factors.The complexity of the interaction between warfarin and its target protein VKORC1 is manifested in that certain Single nucleotide polymorphism（SNP）mutations in VKORC1 gene leading to warfarin drug resistance partially or completely.The mixed presence of these patients seriously affected the accuracy of warfarin dose prediction algorithm based on regression analysis.Whereas,warfarin resistance associated with SNP mutations in the VKORC1 gene has only been reported in rare case reports or brief observational clinical studies,the specific mechanism remains to be further elucidated.The purpose of this study was to construct a warfarin dose prediction model for the patients in Shanxi province,and to explore the relationship between SNP mutation of VKORC1 gene and warfarin resistance.Firstly,the clinical data of patients（n=217）in Shanxi province were used to evaluate the accuracy of six main warfarin dose prediction algorithms,then the influence weight of each factor on warfarin dose was investigated,to provide a theoretical basis for finding and establishing an algorithm more suitable for the patients of Shanxi province.Secondly,in order to further improve the prediction accuracy of the self-constructed algorithm and quest for a more suitable modeling balance point,the univariate and multiple linear regression analysis were performed again for the dependent variable"Warfarin Optimal Dose"（WOD）and its three transformation values,and then the prediction accuracy of each construction model was verified by the validation set data.Thirdly,the SNP mutation that can lead to the change of the VKORC1 amino acid sequence in the exon region of VKORC1 gene for the patients（n=124）of Shanxi was screened.Then Molecular Docking（MD）technique was used to simulate the changes in the binding of warfarin to VKORC1before and after mutation,and to search for mutants with reduced binding energy.Fourth,an analysis model characterizing the combination of VKORC1 and warfarin was constructed based on HEK293T cell culture using F9-VKORC1 co-expression transfection system,in order to avoid investigating the inhibition of VKORC1 activity by warfarin with the DL-Dithiothreitol（DTT）driven assay.There are two reasons for co-transfection of F9 and VKORC1 genes.On the one hand,the activation of F9 must depend on the vitamin K cycle,while VKORC1 is the rate-limiting enzyme of vitamin K cycle.The binding strength of warfarin to VKORC1 changes the activation rate of F9 factor when intervening with different concentrations of warfarin,and the binding of warfarin to VKORC1 can be indirectly characterized by examining the activity of F9 factor.On the other hand,the expression level of F9 factor of HEK293T cells was low,and significant overexpression of F9 factor could be achieved in cells by constructing overexpression plasmid to ensure that the activation rate of F9 factor can be examined more accurately.Finally,in order to verify the accuracy of MD simulation results,the co-expression transfection system of mutant F9-VKORC1 gene was constructed.The established analysis model based on HEK293T cell culture that can characterize the combination of VKORC1 and warfarin was utilized,and the changes in the binding of warfarin to the mutant VKORC1 after corresponding SNP mutations in VKORC1gene were investigated at the cellular level,so that to explore the relationship between warfarin resistance and SNP mutations in the exon region of VKORC1 gene,providing a research basis for clarifying the mechanism of warfarin resistance.Methods:1.Prediction performance evaluation of six main warfarin dose prediction models in the patients of Shanxi provincePatients receiving warfarin anticoagulant therapy in a large general tertiary hospital in Shanxi from January 2017 to December 2019 were selected as the research subjects.A total of 217 patients were included in the study after exclusion criteria screening and informed consent.Genotyping of VKORC1（rs9923231）,CYP2C9*2（rs1799853）and CYP2C9*3（rs1057910）was performed for each patient.In addition,15 other non-genetic factors such as age,gender,weight（kg）,height（cm）,and concomitant medication were also collected.WOD was defined as the dose of warfarin that kept INR within the target range during hospitalization or follow-up.The selection principles of the six main warfarin dose prediction algorithms are as follows:algorithms had to be designed to calculate the maintenance dose rather than the initial dose to be considered;the minimum acceptable number of study subjects for algorithm selection was greater than or equal to 120;at the same time,algorithms established for Asian populations were also considered.This resulted in a total of six algorithms from different references.Among them,HUANG and MIAO are originated from China,IWPC is an internationally recognized algorithm,OHNO and KIM are established by Japanese and Korean scholars respectively,and BRESS is an algorithm established for African Americans.Six warfarin dose prediction algorithms were applied to the dose estimation of 217patients,and each patient received six Predicted doses（PD）.Paired sample t-test and Pearson’s correlation coefficient were used to evaluate the statistical difference between PD and WOD in the six groups and to evaluate the accuracy of the prediction algorithms.Additionally,the mean absolute error(MAE;=¹₉)∑⁹)₄=1)|4)-（24）|)calculated from the predicted PD and WOD values was also adopted to evaluate the accuracy of the six algorithms.In dose stratified analysis,patients were divided into 3 dose groups according to daily dose,namely low dose group（≤3 mg/day）,medium dose group（3-5 mg/day）and high dose group（>5 mg/day）,the prediction accuracy of each algorithm for different dose groups was evaluated by calculating mean relative error(MRE;=¹₉)∑₄=1（24）)⁹|4)-（24）|)).Finally,univariate analysis was served to investigate the influence of each clinical factor on warfarin dose.Independent sample t-test was taken to analyze the statistical difference between variables for binary variables,and One-way ANOVA was performed to test the statistical differences among variables for multi-categorical variables.2.Construction of warfarin dose prediction algorithm based on VKORC1 gene polymorphism in the patients of Shanxi provinceIn order to fit the linear regression equation of the influence of various factors on warfarin dose more accurately and enhance the prediction accuracy,various forms of data conversion can be carried out for the dependent variable WOD.In this study,the above-mentioned 217 patients were taken as the research objects,and the WOD data of the patients were transformed by logarithm（lg WOD）,reciprocal（1/WOD）and square root(√（2）.Then WOD and the three transformation values were served as dependent variables for univariate analysis.For binary variables,independent sample t-test was applied to analyze the statistical differences between variables,while for multi-categorical variables,one-way ANOVA was adhibited to test the statistical differences among variables.Multiple linear regression analysis was conducted for the influencing factors of P<0.05 screened out by univariate analysis.Stepwise regression analysis was adopted to screen variables(α_in=0.10,α_out=0.15),and finally four regression equations were obtained.Two regression equations with high R² values were selected as the alternative warfarin dose prediction algorithms for patients in Shanxi.A validation set was comprised of the clinical data for patients receiving warfarin anticoagulant therapy in a large general tertiary hospital from January to June 2020（n=24）,and the above two alternative warfarin dose prediction algorithms and published models（the optimal model screened out in the first part）were respectively used to estimate warfarin PD values.Three PD values were obtained for each patient,and Paired sample t-test and Pearson’s correlation coefficient were applied to evaluate the statistical difference between PD and WOD in the three groups of predicted doses,and to evaluate the prediction accuracy of each algorithm.In addition,the MAE(=¹₉)∑⁹)₄=1)|4)-（24）|)calculated from the predicted PD and WOD values were also used to evaluate the accuracy of the three algorithms.After determining the optimal model,weight analysis was carried out for each influencing factor in the algorithm,and the weight of each factor was compared.3.SNP screening in exon region of VKORC1 gene and simulation of interaction between VKORC1 and S/R-warfarin in patients of Shanxi provinceHigh-throughput gene sequencing technology was operated to sequence the exon region of VKORC1 gene in 124 patients taking warfarin to screen for significant SNP mutations that could lead to alterations in the amino acid sequence of the VKORC1.Subsequently,the amino acid sequences of VKORC1 before and after mutation were imported into Swiss-model（https://swissmodel.expasy.org/）database to construct protein models of VKORC1gene before and after mutation,respectively.Software Desmond（Schr?dinger Release 2021-1）was operated for 20 ns molecular docking（MD）simulation of the protein models before and after mutation.A TIP3P explicit solvent model was adopted and the protein model was placed in a solvent box at least 10?away from the protein boundary.Furthermore,0.15n M Na Cl was added into the system to simulate physiological conditions.The optimization of the system sets OPLS-3e force field to minimize the energy.The recording time interval of1ps was selected to obtain the MD trajectory with 2000 unique conformations.With an isothermal isobaric（NPT）harness,the temperature was fixed at 300 K and the pressure is1.01 bar.The integral time step was set to 2fs.The MD simulation was carried out using the Py Rx-v0.8 virtual screening tool combined with Auto Dock-Vina.All molecules were converted to PDBQT format after energy minimization by openbabel in the Py Rx software package.All ligand pairs were attached to a square box in which the active pocket was located（center＿x=-9.7,center＿y=26.6,center＿z=56.6,size＿x=17.5,size＿y=15.3,size＿z=18.3）.The choice of docking boxes allowed the ligand to move freely within specified values in the grid box.In Lamarckian genetic algorithm,the state variables of ligand,including initial position,orientation,and torsion,were randomly set.Eight docking sessions were set for each ligand,and the results were cluster analyzed according to the Root Mean Square Deviation（RMSD）standard.Through the MD simulation study above,the SNP mutation site of VKORC1 gene that could lead to warfarin resistance was further speculated.4.Construction of an analytical model of HEK293T cells to characterize the function of the target protein VKORC1HEK293T cells were transfected with a plasmid vector co-expressing human F9 and VKORC1 genes taking FIX（F9）activity as the characterization of the binding of warfarin and VKORC1,and the two proteins were overexpressed in the cells.After transfection,fluorescence microscopy,Real Time Quantitative（RT-q PCR）and Western blot（WB）were applied to evaluate the transfection effect and to investigate the expression of m RNA and target proteins of F9 and VKORC1.Subsequently,vitamin K（5μg/m L）as well as different concentrations of warfarin（0,0.01,0.05,0.1,0.5,1μΜ）were added to the cell culture medium for intervention after 24h transfection with the empty plasmid（F9-VKORC1zero,FV_Z）and overexpression plasmid（F9-VKORC1,FV）,respectively.And the cells were continued to be cultured for 48h before detecting the relative activity of F9 in the culture medium（The F9 factor activity in the cell culture wells without warfarin was considered as the denominator）.The relative activity of F9 factor and the concentration dependence of warfarin intervention concentration were investigated,to evaluate the scientificity of taking the relative activity of F9 factor to characterize the binding of warfarin and VKORC1 in the F9-VKORC1 co-expression plasmid vector system based on HEK293T cell culture.5.Mechanism of SNP mutation in exon region of VKORC1 gene and warfarin resistanceAccording to the SNP mutations in the exon region of VKORC1 screened in Part III,the mutant F9-VKORC1 plasmid vector was constructed by point mutation technique,and the co-expression vector of corresponding mutant VKORC1 was obtained.Cells were transfected with FVz and mutant F9-VKORC1 plasmid vector,respectively.After transfection,fluorescence microscopy,RT-QPCR and WB were applied to evaluate the transfection effect,and the overexpression of m RNA and target proteins of F9 and VKORC1were investigated.After 24 hours of transfection,vitamin K（5μg/m L）and different concentrations of warfarin（0、0.01、0.05、0.1、0.5、1μΜ）were added to the cell culture medium for intervention,and the relative activity of F9 factor was detected after another 48hours.The binding of warfarin to wild-type and mutant VKORC1 enzymes was investigated at the cellular level,so that to explore the relationship between warfarin resistance and SNP mutation in the exon region of VKORC1 gene.Results:1.Prediction performance evaluation of six main warfarin dose prediction models in the patients of Shanxi province（1）Prediction accuracy of six warfarin dose prediction models:Paired sample t-test showed that OHNO（Constructed by Japanese scholars）and IWPC（International Warfarin Pharmacogenomics Consortium）models were not significantly different from WOD.Subsequently,when Pearson correlation analysis and MAE were treated for further analysis,the results showed that the prediction accuracy of OHNO model,which was slightly better than that of IWPC model（WOD-IWPC,r=0.606;WOD-OHNO,r=0.636）was the highest in the patients of Shanxi province.（2）Comparison of the predictive performance of six dose prediction models for low,intermediate,and high dose groups of warfarin:Stratified analysis showed that HUANG model provided better dose prediction accuracy for low-dose group（≤3 mg/d,56.2%of the total cohort）.In the intermediate dose group（3-5 mg/d,35%of the total cohort）,the OHNO model was the best choice.In patients requiring a higher dose（>5 mg/d,8.8%of the total cohort）,KIM was optimal.（3）Effects of genetic and clinical factors on warfarin dose:Univariate analysis showed that 7 factors such as VKORC1 genotype,concurrent atrial fibrillation,concurrent CHF/cardiomyopathy status,CYP2C9 genotype,body mass index（BMI）,gender,and lower serum albumin concentration,were significantly associated with WOD（P<0.05）.Therefore,clinicians should pay close attention to the above factors when taking warfarin for anticoagulant therapy.2.Construction of warfarin dose prediction algorithm based on VKORC1 gene polymorphism in the patients of Shanxi province（1）Univariate and multiple linear regression analysis with WOD and its three transformation values as dependent variables:The results of both univariate and multiple linear regression analyses with WOD,lg WOD,1/WOD and√(2as dependent variables respectively showed that the four factors that eventually entered the regression equation were:VKORC1 genotype,body weight/Body mass index（BMI）,CYP2C9 genotype and atrial fibrillation（yes/no）and sex,the R² of four groups of linear regression equations were 0.440,0.412,0.286 and 0.448,respectively.The two regression equations with higher R² were taken as alternative self-constructed models,self-constructed algorithm 1 was WOD=2.006+1.597X₁-0.613X₂-0.711X₃+0.055X₄-0.356X₅,R²=0.440,where X₁ represented VKORC1genotype,X₂ represented atrial fibrillation status,X₃ represented CYP2C9 genotype,X₄represented body mass index,and X₅ represented gender.The self-constructed algorithm 2was:√(2=1.265+0.408X₁+0.007X₂-0.166X₃-0.237X₄,R²=0.448,where X₁represented VKORC1 genotype,X₂ represented body weight,X₃ represented atrial fibrillation status,and X₄ represented CYP2C9 genotype.When VKORC1 genotype was AA,AG and GG,the values were 0,1 and 2,respectively;when CYP2C9 genotype was*1/*1,*1/*3 and*3/*3,the values were 0,1 and 2,respectively;assigned 1 if gender is female,0 otherwise;weight（kg）was an actual calculated value of the patients;the body mass index was also the actual calculated value.（2）Evaluation of accuracy of warfarin dose prediction by self-constructed model 1,self-constructed model 2 and OHNO model:Validation set data of 24 patients who received warfarin treatment were used to assess warfarin dose prediction algorithms of self-constructed model 1,self-constructed model 2 and OHNO model.Paired sample t-test showed that there was no significant difference between PD and WOD of self-constructed model 1,self-constructed model 2 and OHNO model.The results of Pearson correlation analysis and MAE analysis showed that the prediction accuracy rank of the above three models was self-constructed model 2>self-constructed model 1>OHNO.（3）Weight analysis of influencing factors in the self-constructed warfarin dose prediction algorithm:In self-constructed algorithm 2,VKORC1 genotype,body weight,atrial fibrillation status and CYP2C9 genotype entered the linear regression equation,and the above four factors could explain 44.8%of the differences among WOD individuals,with VKORC1 genotype having the greatest weight of 29.7%.At the same time,three abnormally high data of√(2in 217 patients were removed and the regression analysis was performed again,then the results showed that the factors entering the regression equation were still above four factors,but their weight to explain the difference among WOD individuals increased to 52.3%.3.SNP screening in exon region of VKORC1 gene and simulation of interaction between VKORC1 and R/S-warfarin in patients of Shanxi province（1）VKORC1 gene sequencing results and amino acid sequence analysis:High-throughput sequencing results of 124 patients with VKORC1 gene showed that 3 patients had SNP mutations in the exon region of VKORC1 gene,all of which were in the exon 2region,and their locations and types were chr:16:31093398（197 T>A）and chr:16:31093392（203 A>G）respectively.According to software comparative analysis,chr:16:31093398（197 T>A）mutation will cause the 66th amino acid of VKORC1 to change from valine（V）to glutamic acid（E）;chr:16:31093392（203 A>G）mutation will cause the 68th amino acid of VKORC1 to change from histidine（H）to arginine（R）.（2）Molecular docking simulation results of warfarin and VKORC1:MD study results showed that the binding energy of VKORC1 Wild type（WT）and 66 site mutants（V66E）was the same as that of（R/S-）warfarin,while the binding energy of 68 site amino acid mutants（H68R）was lower than that of WT.The binding energies of R-warfarin and S-Warfarin with H68R were-8 kcal/mol and-8.3 kcal/mol,respectively.In addition,the mutation of amino acid at 68 site（H68R）changed the original amino acid interactions,suggesting that the mutation of amino acid at 68 from histidine（H）to arginine（R）led to a decrease in the binding ability of（R/S-）warfarin to VKORC1.4.Construction of an analytical model of HEK293T cells to characterize the function of the target protein VKORC1（1）Effects of FV_Z and FV transfection on HEK293T cells and overexpression of m RNA and target protein:After transfection of HEK293T cells with FV_Z and FV plasmids for 24h,green fluorescence could be obviously observed in the cell culture flask under fluorescence microscope.The results of RT-q PCR revealed that the expression of F9 m RNA increased to 115.33 times（379.45±165.85/3.29±1.70）compared with FV_Z group in HEK293T cells transfected with FV for 48h.The expression of VKORC1 m RNA was increased to 12.91-fold（11.36±2.95/0.88±0.09）,and there was significant difference in both F9 and VKORC1 m RNA expression compared with FV_Z control group（P<0.05）.The Western blot assay demonstrated that after transfection of HEK293T cells for 48h with FV,the average overexpression of F9 protein and VKORC1 protein was 13.16±2.23 times and1.46±0.20 times,respectively,compared with the FV_Z control group（P<0.05）.（2）Detection of F9 factor in supernatant of HEK293T cells after FVz and FV transfection:The content of F9 in the supernatant of 6-well culture dishes did not change with increasing warfarin concentration in FV_Z-transfected and FV-transfected cells with different concentrations of warfarin（0,0.01,0.05,0.1,0.5,1μM）intervention,and there was no significant difference（P>0.05）.The content of F9 in supernatant of FV transfection group was significantly higher than that in FVz transfection group（P<0.05）,indicating that the constructed F9-VKORC1 co-expression plasmid vector successfully overexpressed F9 in the cells.（3）The relative activity of F9 factor in supernatant of HEK293T cells transfected with FVz and FV was detected:The relative activity of F9 factor in the supernatant of HEK293T cells was measured after FV_Z or FV transfection for 72h.The results represented that the relative activity of F9 decreased with the increase of warfarin concentration in both FV_Z and FV transfected groups;the decreasing trend of F9 relative activity with the increase of warfarin intervention concentration in FV transfected group was less than that in FV_Ztransfected group.5.Mechanism of SNP mutation in exon region of VKORC1 gene and warfarin resistance（1）Construction and sequencing analysis of mutant F9-VKORC1 co-expression plasmid vector:FV-M1 was obtained by mutating the 8850 site T into A base by point mutation induction technique（corresponding to the mutation of site chr:16:31093398,197T>A of VKORC1 gene on plasmid）;and FV-M2 was obtained by mutating the 8856 site A into G base（corresponding to the mutation of site chr:16:31093392,203 A>G of VKORC1gene on plasmid）.Then,the three plasmids were sequenced,and the gene sequences of FV_Z,FV-M1 and FV-M2 plasmids were comparative analyzed by software Snape Gene（Version5.2.3）,and the results showed that FV-M1 and FV-M2 plasmids were successfully mutated according to the intended design,corresponding to the amino acid mutation of site 66（V66E）and site 68（H68R）on the VKORC1,respectively.（2）Effects of FV-M1 and FV-M2 transfection on HEK293T cells and overexpression of m RNA and target protein:After transfecting HEK293T cells with FV_Z,FV-M1 and FV-M2 plasmids for 24h,green fluorescence was observed evidently in cells under a fluorescent microscope.The results of RT-q PCR showed that the expression of F9 m RNA in HEK293T cells transfected with FV-M1 after 48h escalated to 124.25 times（168.98±18.65/1.36±0.28）compared with the FV_Z group;and the expression of VKORC1 m RNA was increased to 27.82 times（43.40±4.11/1.56±0.39）on average;so,there were significantly different（P<0.05）in the expression of F9 and VKORC1 m RNA compared with the FV_Z control group.After FV-M2 transfection of HEK293T cells for 48h,F9 m RNA expression increased255.64-fold on average（166.17±17.32/0.65±0.33）;VKORC1 m RNA expression also increased 10.83-fold（12.78±2.02/1.18±0.33）;therefore,compared with the FV_Z control group,the expression of both F9 and VKORC1 m RNA in cells was significantly different（P<0.05）.Western blot analysis results showed that the overexpression of F9 and VKORC1proteins in FV-M1 and FV-M2 transfected HEK293T cells for 48h was significantly different compared with the FV_Z control group（P<0.05）.Compared with the FV_Z group,the average overexpression of F9 protein was 4.48±1.08 times and 1.54±0.20 times for VKORC1protein after 48h transfection of FV-M1 in HEK293T cells;the average overexpression of F9 protein was 5.06±0.27 folds and 1.74±0.25 folds for VKORC1 protein after 48h transfection of FV-M2 in HEK293T cells.（3）Detection of F9 factor content in supernatant of HEK293T cells transfected with FV-M1 and FV-M2:The F9 levels in the supernatants of HEK293T cells transfected with FV_Z,FV-M1 and FV-M2 for 72h were measured.The results showed that the content of F9in the supernatant of each well of the 6-well culture dish in the three groups did not change with the increase of warfarin concentration after different concentrations of warfarin intervention（0,0.01,0.05,0.1,0.5 and 1μΜ）,and there was no significant difference（P>0.05）.In contrast,the content of F9 in the supernatant of each well of the FV-M1 and FV-M2 transfected group was significantly higher than that of the FVz transfected group（P<0.05）,indicating that the two constructed mutant F9-VKORC1 co-expression plasmid vectors FV-M1 and FV-M2 successfully overexpressed F9 in cells.（4）Detection of relative activity of F9 factor in HEK293T cell supernatant after FV-M1 and FV-M2 transfection:The relative activity of F9 factor was measured in the supernatants of HEK293T cells transfected with FV_Z,FV-M1 and FV-M2 for 72h.The results showed that the relative activity of F9 decreased with the increase of warfarin intervention concentrations in the FV_Z,FV-M1 and FV-M2 transfected groups;moreover,as the warfarin concentration increased,the decreasing trend of F9 relative activity in descending order was FV-M2,FV-M1 and then FV_Z transfected groups.Conclusion:1.When selecting warfarin dose prediction model empirically for patients in Shanxi Province,it is suggested that OHNO should be given priority to predict warfarin dose.For patients who take OHNO to predict warfarin dose（≤3 mg/d）or（>5mg/d）,HUANG or KIM may also be considered for more accurate predictions.Seven factors,namely VKORC1genotype,concurrent atrial fibrillation,concurrent CHF/cardiomyopathy status,CYP2C9genotype,BMI,gender,and lower serum albumin concentration can significantly affect the dosage of warfarin.Thus,physicians should pay close attention to these clinical factors to optimize warfarin dose adjustment strategies in Chinese patients.2.The results of paired sample t-test,Pearson correlation analysis and MAE analysis all revealed that the linear regression equation with√(2as the dependent variable had better prediction accuracy,that is,√(2=1.265+0.408X₁+0.007X₂-0.166X₃-0.237X₄,R²=0.448,（X₁:VKORC1 genotype,X₂:body weight,X₃:atrial fibrillation status,X₄:CYP2C9genotype）.After converting√(2in the equation to WOD,the warfarin dose prediction algorithm for patients in Shanxi based on VKORC1 gene polymorphism was obtained,WOD=（1.265+0.408X₁+0.007X₂-0.166X₃-0.237X₄）².When VKORC1 genotype was AA,AG and GG,the values were 0,1 and 2,respectively;when CYP2C9 genotype was*1/*1,*1/*3and*3/*3,the values were 0,1 and 2,respectively;1 in the case of atrial fibrillation,0otherwise;body weight was the actual body weight of the patient.This algorithm explained44.8%of the inter-individual variation in WOD,with the VKORC1 genotype accounting for the largest weight of 29.7%,much higher than other factors.3.Although multiple linear regression analysis was carried out for WOD and three WOD transform values(lg WOD,1/WOD,√（2）,the prediction accuracy of the best regression algorithm was only 44.8%,which was in the middle level.However,when three points with abnormally high√(2in 217 patients’data were removed and the regression analysis was conducted again.The results illustrated that the factors entering the regression equation were still VKORC1 genotype,body weight（kg）,atrial fibrillation status and CYP2C9 genotype,but the prediction accuracy of the regression equation improved to 52.3%,suggesting that the patients with abnormally high demand for confounded warfarin dose were an important reason for the generally low predictive accuracy of the linear regression prediction algorithms.Among the above four factors,VKORC1 genotype accounted for the largest proportion of weight（29.7%）.Therefore,it is an important point for studying warfarin resistance to deeply explore the changes in protein spatial configuration caused by changes in VKORC1 amino acid sequence.4.By sequencing the VKORC1 gene of 124 patients from Shanxi,two SNPs were found to change the VKORC1 amino acid sequence.The location and types of the two mutations were chr:16:31093398（197 T>A）,with an incidence of 0.81%（1/124）and chr:16:31093392（203A>G）,with an incidence of 1.61%（2/124）.chr:16:31093398（197 T>A）mutation caused a change in amino acid at site 66 of VKORC1 from valine to glutamate（V66E）;chr:16:31093392（203 A>G）mutation caused a change in amino acid at site 68 of VKORC1from histidine to arginine（H68R）.In the subsequent MD simulation program,it was further speculated that the mutation of 68th amino acid（H68R）in the VKORC1 changed the original amino acid interaction relationships,which caused a decrease in the binding ability of（R/S-）warfarin to the VKORC1.Therefore,considering the complexity of human environment and the diversity of biochemical reactions in vivo,conducting experiments to verify MD simulation results is meaningful.5.In this study,a new cell culture-based co-expression system for F9-VKORC1 was established.HEK293T cells were transfected with a plasmid vector co-expressing human F9and VKORC1 genes to overexpress the two proteins separately.The concentration dependence of antigen level and relative activity of F9 factor on warfarin concentration,indicated that the F9-VKORC1 co-expression plasmid vector system based on HEK293T cell culture could indirectly characterize the binding strength of warfarin to VKORC1 by detecting the relative activity of F9 factor.This analytical model avoids the adoption of non-physiological reductant DTT for in vitro drive experiments,and can more accurately investigate the combination of warfarin and VKORC1.Furthermore,the analytical model can be applied to probe the influence of any individual amino acid change on the VKORC1enzyme activity caused by SNP mutation of VKORC1 gene,thus representing an important tool for future research on the relationship between VKORC1 gene SNP and warfarin resistance.6.The VKORC1 mutant plasmid vector was constructed by point mutation induction of plasmid vectors co-expressing human F9-VKORC1 genes（FV-M1,expressing VKORC166th amino acids from V→E;FV-M2,expressing VKORC1 68th amino acids from H→R）.Then,the constructed F9-VKORC1 co-expression plasmid vector system analysis model was to investigate the binding of VKORC1 with warfarin after V66E and H68R amino acid mutations at the cellular level.The results showed that the relative activity of F9 in the FV-M2 transfected group was lower than that in the FV-M1 transfected group and the FVz group,indicating that the mutation of amino acid 68th in VKORC1 would lead to the decrease of warfarin binding force,thus weakening the inhibition of warfarin on vitamin K cycle,and leading to the slowdown of the relative activ...