Carrier frequencies of antithrombin, protein C, and protein S deficiency variants estimated using a public database and expression experiments.

BACKGROUND: Genetic deficiencies of antithrombin (AT), protein C (PC), and protein S (PS) are risk factors for venous thromboembolism. In the general population, the prevalence of heterozygous deficiency of AT, PC, and PS are reported as approximately 0.02%-0.2%, 0.2%-0.4%, and 0.03%-0.5%, respectively. The Exome Aggregation Consortium (ExAC) provides a public database containing reference data for over 60 000 exomes.
OBJECTIVE: This study aimed to determine the frequency of AT, PC, and PS deficiencies using the ExAC database and transient expression experiments.
METHODS: In total, 133, 157, and 221 variants of SERPIN1 (encoding AT), PROC (PC), and PROS1 (PS), respectively, were registered as missense and putative loss-of-function variants in the ExAC database. Variants with relatively high allele frequencies were selected and randomly sampled. Recombinant proteins were expressed in human embryo kidney 293 cells and their secretion and anticoagulant activities examined. RESULTS AND
CONCLUSION: We assessed 9 AT, 4 PC, and 14 PS variants with relatively high allele frequencies and randomly sampled 12 AT, 15 PC, and 19 PS missense variants. All 21 AT variants showed normal or mildly reduced secretion, and 6 showed reduced total activity (specific activity × antigen level). Of the 19 PC variants, 11 showed impaired total activity. All 33 PS variants showed normal or mildly reduced secretion, and 4 showed reduced total activity. Based on allele frequencies in the ExAC database, we calculated the frequencies of AT, PC, and PS genetic deficiency as 0.36%, 0.63%, and 0.39%, respectively.

Antithrombin (AT), protein C (PC), and protein S (PS) deficiencies are risks for thrombophilia.

Exome Aggregation Consortium data and expression experiments were used to estimate deleterious allele frequencies.

Secretion and activity of 21 AT, 19 PC, and 33 PS selected variants were assessed.

Estimated carrier frequencies of the deleterious alleles were 0.36% (AT), 0.63% (PC), and 0.39% (PS).

INTRODUCTION

Venous thromboembolism (VTE) is a common and serious disorder afflicting millions of individuals annually worldwide. VTE is a multifactorial disease in which acquired or inherited predispositions of thrombosis interact with various risk factors. Deficiencies in natural anticoagulant proteins antithrombin (AT), protein C (PC), and protein S (PS) severely increase the risk of VTE. The serine protease inhibitor AT regulates blood coagulation and inactivates multiple coagulation factors, such as thrombin, activated factor XIa, activated factor Xa (FXa), activated factor IXa, and activated factor VIIa. In contrast, PC and PS are vitamin K–dependent plasma glycoproteins that exert their anticoagulant actions through the degradation of coagulation factors activated factor VIIIa, and activated factor Va. PS also serves in the efficient inhibition of FXa as a cofactor of tissue factor pathway inhibitor.

Although the pathophysiology and risk factors of VTE have been long studied, it remains difficult to estimate the precise prevalence of inherited AT, PC, and PS deficiencies. In patients with suspected hereditary VTE (juvenile and recurrent events), the prevalence of AT, PC, and PS deficiencies have been reported as 2%‐5%, 5%‐10%, and 5%‐10%, respectively. Among the general Caucasian population, the prevalence of AT, PC, and PS deficiencies are estimated to be 0.02%‐0.2%, 0.2%‐0.4%, and <0.5%, respectively. The AT protein is encoded by SERPINC1 (7 exons), PC by PROC (9 exons), and PS by PROS1 (15 exons). Although more than 200 missense and nonsense variants have been registered in the Human Gene Mutation Database, their allele frequencies in the general population are unclear.

The Exome Aggregation Consortium (ExAC) is a coalition of investigators seeking to aggregate and harmonize exome sequencing data from a wide variety of large‐scale sequencing projects and to make summary data available to a wider scientific community. The data provided through ExAC included 60 706 unrelated individuals whose genomes were sequenced for various disease‐specific and population genetic studies with frequencies provided for each variant.

In the current study, ExAC data and expression experiments were used to estimate the allele frequencies of AT, PC, and PS genetic deficiencies that may cause thrombophilia.

METHODS

Sampling of ExAC data for nonsynonymous variants

In total, 133 SERPIN1, 157 PROC, and 221 PROS1 missense and putative loss‐of‐function variants with allele counts have been registered in the ExAC database version 1.0. The variants come from seven regional populations, including African, East Asian, European (non‐Finnish), Finnish, Latino, South Asian, and Other. Variants with relatively high allele frequencies (≥0.1%) and allele counts ≥2 in each region were selected as a representative subgroup for subjection to expression experiments and included 9 SERPIN1, 4 PROC, and 14 PROS1 variants. In addition, we randomly selected approximately 10% of the remaining variants of each gene, regardless of frequency and allele count. This included 12 SERPIN1, 15 PROC, and 19 PROS1 variants.

Mutagenesis and construction of expression vectors

The human AT expression vector ORF‐NM_000488‐pcDNA3 was generously provided by Dr Eriko Morishita (Kanazawa University). The human PC expression vector ORF‐NM_000312‐pCMV‐SPORT6 was purchased from RIKEN BRC (Tsukuba, Japan). The human PS expression vector was previously described. In total, we constructed 73 variants using a site‐directed mutagenesis technique and verified their sequences using an Applied Biosystems 3500xL genetic analyzer (Life Technologies, Rockville, MD, USA). Two clones of each variant were selected for expression experiments and are described below.

Preparation of recombinant AT, PC, and PS

Human embryo kidney 293 cells were transiently transfected with each expression vector using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA, USA). For measuring expression levels in the culture supernatants, transfected cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; Wako Pure Chemical, Osaka, Japan) supplemented with 10% fetal bovine serum (FBS) for 72 hours. For PC and PS expression, 10 µg/mL vitamin K1 was added to the culture media. Culture supernatants were centrifuged to remove cell debris. For measuring AT and PC activity, the cells were cultured for 72 hours in OptiMEM Ⅰ Reduced Serum Medium (Thermo Fisher Scientific). The culture supernatants were centrifuged to remove cell debris, concentrated using Amicon Ultra Centrifugal Filters (Sigma‐Aldrich, St. Louis, MO, USA), and exchanged against reaction buffer (50 mmol/L Tris‐HCl, 100 mmol/L NaCl, pH 7.4). For measuring PS activity, the cells were cultured in DMEM supplemented with 10% FBS for 72 hours, and His‐tagged PS molecules in the supernatant were purified using Talon Metal Affinity Resin (Takara Bio, Mountain View, CA, USA). PS was eluted with elution buffer (20 mmol/L Tris‐HCl, 300 mmol/L NaCl, and 250 mmol/L imidazole) and dialyzed against reaction buffer.

Measurement of AT, PC, and PS expression levels

Antigen levels of AT and PC in culture media were measured using Matched‐Pair Antibody Sets for ELISA (Affinity Biologicals, Ancaster, ON, Canada). Antigen levels of PS were measured using an in‐house sandwich ELISA with a polyclonal rabbit antihuman PS primary antibody (Agilent Dako, Carpinteria, CA, USA) and horseradish peroxidase–labeled anti‐human PS secondary antibody (Affinity Biologicals). Although the protein S ELISA used can detect both C4b‐bound and free (unbound) protein S, all recombinant protein S was free in the present study. ELISA standards were prepared using purified human AT, PC, and PS (Haematologic Technologies, Essex Junction, VT, USA). Results are shown as means of duplicate measurements relative to their respective recombinant wild‐type (WT) protein.

Measurement of activity of AT, PC, and PS

Anti‐FXa activity of AT was determined using a Testzym S ATIII kit (Sekisui Diagnostics, Framingham, MA, USA) with a modified protocol. Briefly, 1.2 µL of test solution was mixed with 40 µL of FXa solution and 95 µL of reaction buffer, incubated for 3 minutes, and 20 µL of substrate solution then added. Absorbance was measured at 405 nm against a reference wavelength of 505 nm at 1‐minute intervals for 10 minutes at 37°C. Anti‐thrombin activity of AT was similarly determined using a Berichrom ATIII kit (Siemens Healthcare Diagnostics, Deerfield, IL, USA). Briefly, 2.5 µL of test solution was mixed with 20 µL of thrombin solution and 130 µL of reaction buffer, incubated for 30 seconds, and 25 µL of substrate solution then added. Absorbance was measured at 405 nm at 1‐minute intervals for 10 minutes at 37°C. PC activity was determined using a Staclot Protein C kit (Diagnostica Stago, Parsippany, NJ, USA) according to the manufacturer’s protocol. APC cofactor activity of PS was determined using a Staclot Protein S kit (Diagnostica Stago) using the manufacturer’s protocols. Standard curves for all assays were generated using recombinant WT proteins of AT, PC, and PS. Results are shown as means of duplicate measurements relative to their respective WT proteins. Specific activities were determined by dividing total activity in culture supernatant by antigen level.

In silico analysis

Three bioinformatics tools, PolyPhen‐2 (http://genetics.bwh.harvard.edu/pph2/), PROVEAN Protein (http://provean.jcvi.org/seq_submit.php), and MutationTaster2 (http://www.mutationtaster.org) were used for in silico analysis.

RESULTS AND DISCUSSION

We retrieved 131 missense, 1 in‐frame deletion, and 1 start loss variants of SERPINC1; 147 missense, 1 in‐frame insertion, 2 in‐frame deletion, 4 nonsense, 1 frameshift, 1 splicing defect, and 1 stop loss variant of PROC; and 206 missense, 4 nonsense, 5 frameshift, and 6 splicing defect variants of PROS1 (Table 1). We then selected 21 AT, 19 PC, and 33 PS variants using the criteria described in the Methods section.

Table 1

The numbers of variants retrieved from ExAC data

Protein	Gene	Total	Allele frequency		Type
			≥0.1% and ≥2 counts	Less	Missense	In‐frame insertion	In‐frame deletion	Nonsense	Frame shift	Splicing defect	Start loss	Stop loss
AT (464 a.a.)	SERPINC1	133	9	124	131	0	1	0	0	0	1	0
PC (461 a.a.)	PROC	157	4	153	147	1	2	4	1	1	0	1
PS (676 a.a.)	PROS1	221	14	207	206	0	0	4	5	6	0	0

Abbreviation: ExAC, Exome Aggregation Consortium.

Variant expression and activity

Culture‐supernatant antigen levels of all AT variants were >50% compared to that of WT except for AT‐D141V (Table 2), suggesting the variants had no severe expression defects. Anti‐FXa–specific activity of AT variants R56C and E413K were <50% of WT and anti‐thrombin–specific activity of R56C and A416S were <50% of WT. Both expression (supernatant antigen levels) and specific activities of the variants should be considered when evaluating their function in plasma. Therefore, we emphasized total activity (specific activity × antigen level) relative to WT AT. Anti‐FXa total activities of R56C, D141V, E413K, and T418I were <50% of WT and anti‐thrombin total activities of R56C, D141V, T312I, and A416S were <50% of WT (Table 2). Therefore, these six AT variants were considered loss‐of‐function alleles.

Table 2

Characteristics of AT (SERPINC1) variants sampled from ExAC data

Protein	Transcript	Ag (%)	Anti‐FXa sAct (%)	Anti‐thrombin sAct (%)	Anti‐FXa tAct (%)	Anti‐thrombin tAct (%)	PolyPhen‐2	PROVEAN	MutationTaster2	Reported to be causative
WT	…	100	100	100	100	100	…	…	…	…
V30E a	c.89T>A	127.6 (± 22.5)	97.9 (± 0.8)	111.4 (± 4.2)	124.9	142.1	Possibly damaging	Neutral	Polymorphism	10
D38N	c.112G>A	101.3 (± 10.9)	92.7 (± 0.9)	104.9 (± 3.7)	93.9	106.3	Probably damaging	Neutral	Disease causing
I54V	c.160A>G	105.4 (± 15.3)	93.8 (± 0.1)	96.5 (± 6.5)	98.8	101.7	Benign	Neutral	Disease causing
R56C	c.166C>T	56.7 (± 10.7)	29.0 (± 1.3)	8.8 (± 0.1)	16.5	5.0	Probably damaging	Deleterious	Disease causing	11
P73L a	c.218C>T	103.4 (± 6.2)	96.7 (± 3.8)	56.3 (± 4.7)	99.9	58.2	Probably damaging	Deleterious	Disease causing	12
N105K	c.315T>A	88.8 (± 3.4)	104.8 (± 2.2)	109.4 (± 6.4)	93.1	97.2	Benign	Neutral	Polymorphism
D141V	c.422A>T	45.3 (± 1.8)	62.1 (± 3.3)	54.8 (± 5.3)	28.1	24.8	Probably damaging	Deleterious	Disease causing
T147A a	c.439A>G	94.3 (± 4.9)	100.7 (± 1.2)	92.3 (± 6.2)	95.0	87.0	Possibly damaging	Deleterious	disease causing	13
R177C a	c.529C>T	57.4 (± 2.6)	105.3 (± 2.0)	107.5 (± 0.0)	60.5	61.7	Probably damaging	Deleterious	Disease causing
T185A a	c.553A>G	102.4 (± 3.6)	105.4 (± 2.0)	103.8 (± 8.7)	108.0	106.3	Benign	Neutral	Polymorphism
E212K	c.634G>A	76.9 (± 0.4)	92.2 (± 2.1)	73.8 (± 1.2)	71.0	56.8	Benign	Deleterious	Disease causing
N240S a	c.719A>G	88.8 (± 4.5)	105.4 (± 2.0)	95.7 (± 2.7)	93.6	85.0	Benign	Neutral	Polymorphism	14
T250A	c.748A>G	129.5 (± 7.0)	84.0 (± 1.5)	106.1 (± 7.2)	108.8	137.4	Benign	Neutral	Disease causing
Y292F	c.875A>T	103.5 (± 4.4)	105.4 (± 2.4)	106.6 (± 9.7)	109.1	110.3	Probably damaging	Neutral	Disease causing
R293Q a	c.878G>A	112.1 (± 17.5)	106.0 (± 1.9)	108.5 (± 7.5)	118.9	121.7	Possibly damaging	Neutral	Polymorphism
P305H a	c.914C>A	63.9 (± 15.1)	105.3 (± 1.8)	91.5 (± 4.1)	67.2	58.4	Probably damaging	Deleterious	Disease causing
T312I	c.935C>T	56.3 (± 8.5)	94.9 (± 10.4)	82.8 (± 3.4)	53.5	46.7	Probably damaging	Deleterious	Disease causing
R356C	c.1066C>T	78.1 (± 11.8)	103.0 (± 0.6)	97.1 (± 7.4)	80.5	75.9	Possibly damaging	Deleterious	Disease causing
E413K	c.1237G>A	76.9 (± 4.4)	47.3 (± 0.5)	115.4 (± 1.9)	36.4	88.8	Probably damaging	Deleterious	Disease causing
A416S a	c.1246G>T	99.1 (± 7.1)	54.2 (± 6.5)	13.9 (± 5.0)	53.6	13.8	Probably damaging	Deleterious	Disease causing	15
T418I	c.1253C>T	72.3 (± 15.5)	57.9 (± 3.8)	79.0 (± 0.6)	41.9	57.1	Probably damaging	Deleterious	Disease causing

All data are the means (± range) of duplicate measurements.

Abbreviations: AT, antithrombin; Ag, antigen; ExAC, Exome Aggregation Consortium; sAct, specific activity; tAct, total activity (Ag X sAct).

Variants with relatively high allele frequencies in ExAC data (≥0.1% and ≥2 allele counts).

For PC, 10 variants (F181V, P210L, R220Q, A251V, D297N, A309T, V368I, N371D, G392R, and M406I) showed impaired secretion (<50% of WT; Table 3). The specific activity of E205_Q207dup was <50% of WT, and those of 7 variants (F181V, P210L, R220Q, A251V, A309T, G392R, and M406I) could not be measured because of severely impaired secretion. Therefore, the total activities of 11 variants (F181V, R189W, P210L, R220Q, A251V, D297N, A309T, V368I, N371D, G392R, and M406I) were considered to be <50% of WT, that is, loss‐of‐function.

Table 3

Characteristics of PC (PROC) variants sampled from ExAC data

Protein	Transcript	Ag (%)	sAct (%)	tAct (%)	PolyPhen‐2	PROVEAN	MutationTaster2	Reported to be causative
WT	…	100	100	100	…	…	…	…
K70R	c.209A>G	61.7 (± 19.4)	96.0 (± 3.8)	59.2	Benign	Neutral	Polymorphism
F181V	c.541T>G	21.3 (± 1.1)	ND	…	Benign	Deleterious	Disease causing	16
R189W a	c.565C>T	73.3 (± 1.9)	60.6 (± 0.7)	44.4	Benign	Neutral	Polymorphism	17
K193Q	c.577A>C	137.0 (± 4.3)	55.5 (± 0.0)	76.1	Possibly damaging	Neutral	Polymorphism
K193del a	c.577_579delAAG	151.6 (± 0.9)	51.1 (± 0.6)	77.5	‐	Deleterious	Polymorphism	18
L197P	c.590T>C	61.3 (± 15.5)	106.3 (± 3.1)	65.1	Possibly damaging	Neutral	Polymorphism
E205_Q207dup	c.614_622dup AAGACCAAG	141.1 (± 0.2)	44.0 (± 1.7)	62.1	…	Neutral	Polymorphism
P210L	c.629C>T	11.7 (± 2.8)	ND	…	Possibly damaging	Neutral	Disease causing	17
R220Q	c.659G>A	14.6 (± 6.6)	ND	…	Benign	Neutral	Disease causing	18
A251V a	c.752C>T	19.5 (± 0.6)	ND	…	Probably damaging	Deleterious	Disease causing	19
D297N	c.889G>A	47.4 (± 1.5)	92.3 (± 3.0)	43.7	Probably damaging	Deleterious	Disease causing
A309T	c.925G>A	7.2 (± 1.4)	ND	…	Possibly damaging	Neutral	Disease causing	20
E349D	c.1047G>C	84.0 (± 1.4)	98.1 (± 7.2)	82.4	Benign	Neutral	Polymorphism
V368I	c.1102G>A	41.4 (± 6.1)	117.4 (± 15.9)	48.6	Benign	Neutral	Disease causing
N371D a	c.1111A>G	35.1 (± 11.2)	102.1 (± 7.3)	35.8	Benign	Deleterious	Polymorphism
G392R	c.1174G>A	3.9 (± 0.1)	ND	…	Probably damaging	Deleterious	Disease causing	21
M406I	c.1218G>A	3.7 (± 0.5)	ND	…	Probably damaging	Deleterious	Disease causing	22
V434I	c.1300G>A	68.1 (± 2.6)	101.3 (± 2.9)	69.0	Possibly damaging	Neutral	Polymorphism
S458R	c.1374C>G	71.3 (± 7.4)	130.5 (± 7.4)	93.0	Benign	Neutral	Polymorphism

All data are the means (± range) of duplicate measurements.

Abbreviations: Ag, antigen; ExAC, Exome Aggregation Consortium; ND, not detectable; PC, protein C; sAct, specific activity; tAct, total activity (Ag X sAct).

Variants with relatively high allele frequencies in ExAC data (≥0.1% and ≥2 allele counts).

Culture‐supernatant antigen levels of all PS variants were >50% compared to that of WT PS (Table 4). The specific activities of D129G and F273L were <50% of WT. The total activities of four variants (E60K, V191D, F273L, and V510M) were <50% of WT and considered loss‐of‐function alleles.

Table 4

Characteristics of PS (PROS1) variants sampled from ExAC data

Protein	Transcript	Ag (%)	sAct (%)	tAct (%)	PolyPhen‐2	PROVEAN	MutationTaster2	Reported to be causative
WT	…	100	100	100	…	…	…	…
R40L a	c.119G>T	270.5 (± 6.7)	53.9 (± 2.8)	145.8	Probably damaging	Deleterious	Disease causing	23
E60K	c.178G>A	66.0 (± 5.1)	58.0 (± 0.5)	38.3	Probably damaging	Deleterious	Disease causing
P76L a	c.227C>T	91.9 (± 3.3)	85.8 (± 1.7)	78.9	Probably damaging	Deleterious	Disease causing	8
D129G	c.386A>G	107.6 (± 0.2)	47.0 (± 2.8)	50.5	Benign	Neutral	Disease causing	8
T144N a	c.431C>A	87.7 (± 5.7)	77.9 (± 3.9)	68.3	Benign	Neutral	Polymorphism	23
Q150R	c.449A>G	101.0 (± 1.4)	93.4 (± 0.4)	94.3	Benign	Neutral	Disease causing
N166H	c.496A>C	93.0 (± 6.0)	88.1 (± 2.4)	82.0	Probably damaging	Neutral	Polymorphism
N168S a	c.503A>G	109.2 (± 1.5)	71.1 (± 0.8)	77.6	Possibly damaging	Deleterious	Disease causing
D176H	c.526G>C	99.0 (± 10.9)	143.5 (± 3.4)	142.0	Possibly damaging	Neutral	Polymorphism
P179R	c.536C>G	89.6 (± 5.8)	81.0 (± 2.1)	72.6	Possibly damaging	Deleterious	Disease causing
V191D	c.572T>A	84.1 (± 6.9)	56.7 (± 3.3)	47.7	Benign	Neutral	Polymorphism
R233K a	c.698G>A	83.7 (± 2.7)	80.7 (± 1.1)	67.5	Benign	Neutral	Polymorphism	24
F273L	c.819C>A	79.2 (± 5.3)	36.3 (± 1.4)	28.8	Benign	Neutral	Disease causing	25
E283D	c.849G>C	91.8 (± 15.3)	130.3 (± 0.4)	119.6	Possibly damaging	Neutral	Disease causing
A307S	c.919G>T	98.7 (± 9.5)	88.9 (± 2.4)	87.7	Benign	Neutral	Polymorphism
R316C	c.946C>T	82.1 (± 10.1)	62.9 (± 0.8)	51.6	Probably damaging	Deleterious	Disease causing	26
A341T a	c.1021G>A	80.2 (± 12.4)	81.0 (± 1.5)	64.9	Probably damaging	Deleterious	Disease causing
I344F	c.1030A>T	82.1 (± 2.0)	104.4 (± 1.5)	85.7	Benign	Neutral	Polymorphism
A348V	c.1043C>T	81.4 (± 2.4)	82.3 (± 0.8)	67.0	Possibly damaging	Neutral	Polymorphism
N365K a	c.1095T>G	65.2 (± 3.9)	122.3 (± 0.0)	79.8	Probably damaging	Neutral	Disease causing	9
P410H a	c.1229C>A	88.9 (± 7.8)	81.7 (± 0.3)	72.6	Possibly damaging	Deleterious	Disease causing	9
P416L a	c.1247C>T	98.9 (± 0.9)	104.8 (± 0.3)	103.6	Benign	Neutral	Disease causing
V425I	c.1273G>A	94.5 (± 2.4)	80.2 (± 6.2)	75.8	Benign	Neutral	Disease causing
R432W	c.1294C>T	95.4 (± 4.7)	67.7 (± 1.8)	64.6	Probably damaging	Neutral	Disease causing
E435G	c.1304A>G	87.7 (± 5.8)	88.2 (± 2.1)	77.4	Possibly damaging	Deleterious	Polymorphism
P441L	c.1322C>T	62.5 (± 6.0)	88.2 (± 0.8)	55.1	Possibly damaging	Deleterious	Disease causing
R445C	c.1333C>T	109.8 (± 8.8)	81.0 (± 2.9)	88.9	Probably damaging	Deleterious	Disease causing
S501P a	c.1501T>C	90.5 (± 6.0)	80.8 (± 1.9)	73.1	Possibly damaging	Neutral	Disease causing	27
V510M a	c.1528G>A	50.2 (± 3.3)	99.4 (± 1.2)	49.96	Possibly damaging	Neutral	Disease causing	28
N530I	c.1589A>T	78.3 (± 4.4)	84.5 (± 0.7)	66.1	Possibly damaging	Neutral	Polymorphism
T588A a	c.1762A>G	97.4 (± 8.8)	70.9 (± 0.9)	69.0	Possibly damaging	Neutral	Polymorphism
D599H a	c.1795G>C	107.4 (± 1.6)	94.6 (± 0.1)	101.6	Probably damaging	Neutral	Polymorphism	29
T630I a	c.1889C>T	81.8 (± 11.8)	98.5 (± 1.9)	80.6	Probably damaging	Deleterious	Disease causing	30

All data are the means (± range) of duplicate measurements.

Abbreviations: Ag, antigen; ExAC, Exome Aggregation Consortium; PS, protein S; sAct, specific activity; tAct, total activity (Ag X sAct).

Variants with relatively high allele frequencies in ExAC data (≥0.1% and ≥2 allele counts).

In silico prediction of variant impact

Three in silico tools were used to predict whether the variants could be pathogenic due to loss of function (Tables 2, 3, 4). Of the 73 variants examined, 32 showed unanimous predictions using all three tools, with 12 predicted as normal (benign, neutral, or polymorphism) and 20 as pathogenic (probably damaging, deleterious, or disease causing). The concordance rate between the expression experiments and in silico analyses was low. Frequencies of AT, PC, and PS deficiencies were estimated using the expression experiment results and ExAC data.

Allele frequency estimation of AT, PC, and PS deficiencies

First, we estimated allele frequencies of AT deficiency. Of 133 AT variants in the ExAC data, 9 had relatively high allele frequencies of ≥1% with allele counts ≥2 (Table 1). Only one of the 9 frequent variants showed loss‐of‐function effects in the expression experiments (A416S) and had an allele frequency in the ExAC data of 0.0007172 (87 of 121 312 alleles). The 124 rare (nonfrequent) variants contained one start loss variant (c.3G>A) with an allele frequency of 0.0000083, which was predicted to be pathogenic. For the remaining 123 rare variants the mean allele frequency was 0.0000205. Of the randomly selected 12 variants, 5 (R56C, D141V, T312I, E413K, and T418I) showed loss‐of‐function effects (Table 2). The total frequency of “loss‐of‐function alleles” was estimated as follows:

Therefore, carriers of AT loss‐of‐function alleles in the general population were estimated as 35.5 of 10 000 individuals.

The frequency of PC deficiency was similarly estimated. Of 157 PC variants in the ExAC data, 4 had relatively high allele frequencies of ≥1% with allele counts ≥2. In the expression experiments, 3 of the frequent variants showed loss‐of‐function effects (R189W, A251V, and N371D) and had allele frequencies of 0.0007187, 0.0002246, and 0.0003802, respectively. The 153 rare variants included four nonsense (W83*, C101*, E274*, R348*), one frameshift (H149Pfs*13), one splicing defect (c.678+1G>A), and one stop loss (c.1384T>C) variants that were predicted to be pathogenic with a mean allele frequency of 0.0000142. The remaining 146 rare missense variants had a mean allele frequency of 0.0000219. Of the 15 randomly selected variants, 8 (F181V, P210L, R220Q, D297N, A309T, V368I, G392R, and M406I) showed loss‐of‐function effects (Table 3). The total frequency of “loss‐of‐function alleles” was estimated as follows:

Therefore, carriers of PC loss‐of‐function alleles in the general population were estimated as 62.5 of 10 000 individuals.

The frequency of PS deficiency was also estimated. Of 221 PS variants in the ExAC data, 14 had relatively high allele frequencies of ≥1% with allele counts ≥2. In the expression experiments, only one of the frequent variants showed loss‐of‐function effects (V510M) and had an allele frequency of 0.0010630. The 207 rare variants included 4 nonsense [L317*, Y485*, Y560* (c.1680T>A), Y560* (c.1680T>G)], 5 frameshift (C14Sfs*24, Y54Lfs*17, L45Yfs*42, N188Mfs*20, E437Gfs*3), and 6 splicing defect (c.77‐1G>C, c.728‐1G>A, c.728‐1G>T, c.850‐1G>A, c.1645‐2A>G, c.1870+1G>T) variants with a mean allele frequency of 0.0000100. The remaining 192 rare missense variants had a mean allele frequency, 0.0000246. Of the 19 randomly selected variants, 3 (E60K, V191D, F273L) showed loss‐of‐function effects (Table 4). The total frequency of “loss‐of‐function alleles” was estimated as follows:

Therefore, carriers of PC loss‐of‐function variants in general were estimated as 39.1 of 10 000 individuals.

Limitations

In the current study, we estimated the frequencies of AT, PC, and PS deficiencies using a simplified approach, which had several limitations. First, the ExAC database does not represent a general population and may cause some bias. Second, recombinant proteins expressed using cDNA vectors, and cell lines do not always properly reflect plasma proteins in humans. Some predicted normal functioning variants in our study have been reported in case reports as genetic causes of VTE (Tables 2, 3, 4). Third, expression experiment results may differ depending on materials and procedures used. In fact, our results of some variants differed from previous reports. , Fourth, the total activity threshold of 50% may not be appropriate to evaluate the variants. Since the cutoff value “50%” is for the total activity of each variant but not for the estimated plasma activity, the plasma activity of an individual with one “50%” mutant allele and one “100%” normal allele is theoretically 75% (25% + 50% activity in plasma). Fifth, sampling of 10% variants for expression experiments and extrapolating the results of the remaining 90% may cause some inaccuracy. Finally, other functional aspects that we did not examine, such as clearance from the circulation and binding affinity of partner proteins, may affect anticoagulant activities.

The present study highlighted the low concordance rate between expression experiments and in silico analyses, suggesting that clinical sequencing results may be difficult to interpret. There is growing interest in the clinical use of next‐generation sequencing panels to identify inherited thrombophilia. How does clinical practice determine if a detected mutation is pathogenic? Neither phenotypic assays, gene sequencing, and phenotypic‐genotyping correlations (interpretation of sequencing results) are complete. Moreover, the penetrance of these heritable thrombophilia seems to be far from 100%. Therefore, it is important to combine sequencing data with patient phenotypic assay data. In addition, it would be useful to obtain and examine sequencing and phenotypic assay data from the patient’s family. These data can provide helpful information for a definitive diagnosis.

CONCLUSION

Although our study had several limitations, our simplified approach was able to estimate the allele frequencies of AT, PC, and PS genetic deficiencies 0.36%, 0.63%, and 0.39%, respectively. In addition to this genetic basis, other factors such as penetrance, genetic modifiers, and environmental conditions may affect the onset of VTE.

RELATIONSHIP DISCLOSURE

The authors have no conflicts of interest to disclose.

AUTHOR CONTRIBUTIONS

KM performed the experiments and wrote the manuscript. KK conceived this study and wrote the manuscript. Both approved the final version of the manuscript.