A novel missense mutation responsible for factor VII deficiency in research Beagle colonies


Katherine A. High, The Children's Hospital of Philadelphia, 3615 Civic Center Blvd, Philadelphia, PA 19104, USA.
Tel.: + 1 215 590 4521; fax: +1 215 590 3660; e-mail: high@email.chop.edu


Summary. Background: Canine factor VII (cFVII) deficiency, an autosomal recessive trait originally identified in research Beagles, is associated with a mild to moderate bleeding tendency. Objective: Our aim was to identify and characterize the mutation causing cFVII deficiency. Methods: In order to sequence the coding regions of the cFVII gene, we cloned the cFVII cDNA. Genomic DNA and plasma from FVII-deficient Beagles and obligate carriers were utilized. Results: In all FVII-deficient dogs, we identified a single causative G to A missense mutation in exon 5, encoding the second epidermal growth factor-like domain, resulting in substitution of glycine 96 by glutamic acid, with plasma FVII coagulant activity of ≤ 4% in affected Beagles. In vitro expression indicated that the majority (96%) of cFVII-G96E protein was retained intracellularly. In addition, analysis of purified recombinant wild-type and mutant cFVII proteins demonstrated reduced activity of the mutant (< 2%) compared with wild-type. Rotational thromboelastometry revealed a severe impairment of clotting activity in affected Beagles, and heterozygotes also exhibited changes in coagulation-based assays. Using a mutation-specific polymerase chain reaction/restriction digest that allows rapid identification of the G96E mutation, we surveyed a US research Beagle colony and identified a mutant allelic frequency of 31%. Conclusions: We have identified a single causative mutation for cFVII deficiency that may have implications for pharmacotoxicologic research, because reduced FVII coagulant activity may alter hemostatic and/or cardiovascular endpoints in this commonly used animal species.


Factor (F) VII is a vitamin K-dependent glycoprotein synthesized in the liver and secreted into the circulation as a single-chain zymogen that, once activated, plays a pivotal role in the initiation of coagulation. Following vascular injury, FVII, in combination with the membrane protein tissue factor (TF) (FVIIa/TF complex) and in the presence of calcium, cleaves FIX and FX to their activated forms, leading to the generation of thrombin [1,2]. FVII deficiency in humans is a rare autosomal recessive coagulopathy, with an estimated frequency of 1:500 000, which results in a bleeding diathesis ranging from mild to severe [3]. More than 130 mutations in the human FVII (hFVII) gene, with varying effects on protein expression and function, have been reported (http://europium.csc.mrc.ac.uk).

Naturally occurring hereditary FVII deficiency has been previously reported in Beagle dogs maintained in research colonies in the USA, Canada, and a few countries in Europe, as well as in the companion animal population [4–8]. This disorder is associated with a mild to moderate bleeding tendency, but in most research animal colonies FVII deficiency has been noted incidentally, when routine coagulation screens performed prior to experimental studies revealed prolonged prothrombin times (PT) [4,5]. However, it is not clear if any research colonies currently screen Beagles for FVII deficiency.

A canine model of FVII deficiency is desirable for translational research in coagulation and investigation of novel therapeutic approaches, including gene therapy. In this report, we describe the cloning of the canine FVII (cFVII) cDNA and the molecular characterization of a single missense mutation in the second epidermal growth factor-like (EGF-2) domain as the cause of FVII deficiency in Beagles in research colonies and in the companion animal population. Through in vitro expression studies, we demonstrate that the mutation is associated with impaired secretion as well as reduced activity of the activated form. We also describe a DNA-based screening test for FVII deficiency in Beagles. Using this test on a small sample from a US Beagle research colony, we found that the allelic frequency of the FVII deficiency (G96E) is approximately 31%. This rapid DNA screening test will aid in the identification of dogs that are homozygous or heterozygous for this coagulation defect, allowing investigators to effectively eliminate this trait from dogs commonly utilized in biological, pharmacological, and toxicological research studies.

Materials and methods


Five FVII-deficient dogs, three companion animals and two from research colonies, were studied (Table 1). One companion animal (dog #1) was evaluated at the Veterinary Hospital of the University of Pennsylvania, while EDTA-anticoagulated blood or buccal mucosal swabs for DNA extraction (QIAamp DNA blood mini kit, Qiagen, Valencia, CA, USA or Puregene DNA isolation kit, Gentra Systems, Minneapolis, MN, USA, respectively) were obtained from known FVII-deficient companion animals in Michigan (dog #2) and Canada (dog #5) [8], as well as six Beagles closely related to dog #2 for mutation analysis. A dog research colony in Switzerland (Ciba-Geigy, Basel) provided stored (frozen) EDTA-anticoagulated blood from a FVII-deficient Beagle (dog #3) and 17 related dogs for further studies [9]. In addition, blood was obtained from a Beagle (dog #4) and subsequently an additional 104 dogs from a research colony in the USA. Finally, blood was collected from 12 clinically normal non-Beagle dogs and six unrelated Beagles for evaluation of possible polymorphisms in the cFVII gene and for use as study controls. These studies were approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania.

Table 1.   Clinical and laboratory data from factor (F) VII-deficient dogs homozygous for G96E mutation
DogBreedSourceAge (years)/sexPhenotypeProthrombin times (s)FVII:C (%)
  1. *FVII:C and prothrombin times measured at outside laboratory: sample not available for confirmation at authors’ laboratory.

2BeaglePet1/FPostsurgical hemorrhage
5Mixed breedPet7/MPostsurgical hemorrhage29.5< 1*
Reference range    5.5–7.957–181

Coagulation studies and rotational thromboelastometry

PTs on plasma, cell culture supernatants, and purified cFVII proteins were determined using one-stage clotting assays, as previously described [10]. The modified PT assay was performed using 1:2.5 dilution of Innovin (TF) reagent (Dade Behring, Newark, DE, USA) in Tris-buffered saline-bovine serum albumin (TBS-BSA) buffer (0.05 mol L−1 Tris, pH 7.5, 0.15 mol L−1 NaCl, 0.1% BSA). Plasma FVII coagulant activity was determined in a PT assay using pooled plasma from 12 control dogs serially diluted with TBS-BSA buffer to create a standard curve. The rotational thromboelastometry analysis was performed in a ROTEM® (BioDis, Signes, France) using citrated plasma and Innovin reagent (final concentration 4 pmol L−1) and was instantly recorded using the EXTEM test for a period of 30 min.

Cloning and sequencing of cFVII cDNA

Total RNA was extracted from the liver of a mixed-breed dog using Trizol® (Invitrogen, Carlsbad, CA, USA), and full-length enriched cDNA was made using the SMARTTM cDNA synthesis kit (Clontech, Mountain View, CA, USA) following the manufacturer's protocol. The full-length cFVII cDNA was amplified using primers that were designed based on conserved amino acid residues among human, mouse, and rat FVII mature cDNA. The polymerase chain reaction (PCR) product (∼1.4 kb) was cloned into pcDNA3.1 vector (Invitrogen) and verified by sequencing. The nucleotide sequence (GenBank accession number DQ223901) was used to search the non-redundant EMBL/GenBank database utilizing the BLAST search tool (http://www.ncbi.nlm.nih.gov/blast; accessed 31 August 2006).

Isolation and characterization of mutant cFVII gene

The cFVII cDNA sequence was utilized to search the dog genome database at NCBI (http://www.ncbi.nlm.nih.gov/genome/guide/dog), and the eight exons of the cFVII gene were determined. Seven pairs of oligonucleotide primers (sequences and PCR conditions available upon request) were used to amplify the coding regions and exon–intron boundaries of the cFVII gene using genomic DNA from two affected (dogs #1 and #3) and two closely related colony Beagles with plasma FVII coagulant activities of 18 and 45%. Exons 3 and 4 were amplified as a single fragment. Target sequences were amplified in 50 μL reaction mixtures containing approximately 50 ng genomic DNA under standard reaction conditions with 0.2 μmol L−1 of each primer. Sequencing of PCR-amplified products was performed following electrophoresis and gel recovery (QIAquick gel extraction kit, Qiagen).

During DNA sequence analysis of the cFVII gene from two affected and two closely related Beagles, a suspected disease-causing mutation in exon 5 and two other amino acid changes in exon 8 were observed. In order to evaluate whether these amino acid changes were associated with the disease, exons 5 and 8 of the cFVII gene were sequenced for all affected dogs, five of the remaining Swiss colony Beagles, two relatives of dog #2, three additional affected Beagles from a US research colony, six Beagles unrelated to the dogs studied, and 11 non-Beagle dogs.

DNA screening tests

A MnlI (New England BioLabs, Beverly, MA, USA) restriction digest was developed to determine the genotype of Beagles with regard to the disease-causing mutation. Exon 5 was amplified by PCR using appropriate primers to generate a 311 bp fragment, digested with MnlI, and visualized with ethidium bromide staining after electrophoresis in an 8% polyacrylamide gel.

Site-directed mutagenesis and in vitro expression

The identified mutation (G96E) was introduced into the cFVII cDNA by site-directed mutagenesis (QuikChange Site-Directed Mutagenesis Kit, Stratagene, La Jolla, CA, USA). A 12-amino acid epitope (thrombin cleavage site of human protein C, EDQVDPRLIDGK, HPC4) for purification was placed at the C terminus of these transgenes by PCR [10,11]. Stable human embryonic kidney-293 (HEK-293) cell lines expressing cFVII wild-type (cFVII-wt) and cFVII-G96E were generated. Site-directed mutagenesis was also utilized to introduce R227Q into the cFVII-wt and cFVII-G96E cDNA using mutagenic primers.

Purification of wild-type and mutant cFVII from conditioned medium

Recombinant cFVII proteins were purified from conditioned medium supplemented with vitamin K (6 μg mL−1) as previously described [11]. Protein concentrations were calculated using a molecular weight of 50 000 kDa and an extinction coefficient (Einline image) of 1.39 [10]. cFVII protein was assumed to have an extinction coefficient (Einline image) similar to hFVII [10].

Detection of wild-type and mutant cFVII proteins from HEK-293 cell supernatant and lyzate by Western blotting

Forty-eight hour supernatants from stable clones expressing cFVII-wt and cFVII-G96E were collected, and cell lyzates were prepared by sonication. Supernatants (30 μL) or lyzates (30 μg) from stable clones expressing cFVII-wt and cFVII-G96E were subjected to reducing SDS–PAGE, followed by Western blotting. Sheep anti-hFVII (1:500 dilution) (Cedarlane Laboratories, Ontario, Canada) and antisheep immunoglobulin G-horseradish peroxidase (1:3000 dilution) (Rockland Immunochemicals Inc, Gilbertsville, PA, USA) were utilized as primary and secondary antibodies, respectively, and cFVII was detected by chemiluminescence (Pierce, Rockford, IL, USA).

Activation of purified recombinant cFVII-G96E with human TF and human FXa

Two μmol L−1 cFVII or cFVII-G96E was incubated with either soluble human TF (shTF) or human FXa (Hematologic Technologies, Inc., Essex Junction, VT, USA) in the presence of 5 mm CaCl2, as previously described [12]. Aliquots of the autoactivated samples were removed following incubation for determination of clotting activity. The remainder of the autoactivated samples was then subjected to SDS–PAGE, followed by Coomassie Brilliant Blue staining.

Statistical analysis

Student's t-test was used for statistical analysis. A P-value < 0.05 was considered significant. Data are presented as mean ± SD.


Cloning and sequencing of cFVII cDNA

We used a PCR strategy based on conserved amino acid sequences among known FVII proteins from a variety of species to isolate the cFVII cDNA from canine liver RNA. The protein sequence predicted from the cFVII cDNA results in a mature protein of 406 amino acids, with 22 and 18 residues as a putative signal peptide and propeptide, respectively, and sharing 75% identity with the hFVII mature protein. A ClustalW alignment of the cloned cFVII cDNA with the recently available dog genome sequence database at NCBI is shown in Fig. 1. Using our cloned cDNA, we located the cFVII gene to chromosome 22 of the dog genome by BLAST, with the gene comprised of eight exons, with exon–intron boundaries shown in Fig. 2. Canine chromosome 22 shows conservation of synteny with human chromosome 13 [13], where the hFVII gene is located [14].

Figure 1.

 ClustalW alignments of the conceptual translation of canine factor VII (cFVII) from our cloned cDNA, assembled exons from the cFVII gene in the dog genome, and sequencing and assembling of all exons of the cFVII cDNA from one FVII-deficient dog. Boxes show the sites for the G96E mutation in exon 5 and two polymorphisms in exon 8.

Figure 2.

 Schematic representation of the exon–intron distribution in the canine factor VII gene.

Identification and rapid screening of the missense mutation

We analyzed cFVII sequences from five unrelated dogs with severe FVII deficiency, having plasma FVII coagulant activities ranging from 1 to 4% (Table 1). Two dogs experienced postsurgical hemorrhage, one dog developed spontaneous bleeding, and two research Beagles had no history of bleeding. Comparison of the DNA sequences of the entire coding region and exon–intron junctions from dogs #1 and #3 with the wild-type cFVII cDNA revealed a G to A missense mutation (nucleotide position 6385) in exon 5, resulting in substitution of glycine 96 (GGA) by glutamic acid (GAA) in the EGF-2 domain. All affected dogs were homozygous for the G96E mutation in exon 5. Sequencing of exon 5 in 17 closely related Beagles from the Swiss research colony revealed nine dogs heterozygous for the G to A mutation and eight dogs with the wild-type sequence. In the US research colony from which dog #4 originated, an additional 104 Beagles were screened, and six dogs homozygous for the G96E mutation and 52 heterozygotes were found (allelic frequency of the G96E mutation was 31%). Amongst six close relatives of dog #2 tested, two were found to be homozygous and four heterozygous for the missense mutation.

In addition to the G96E mutation, comparison of DNA sequences from the five FVII-deficient dogs, the cFVII cDNA (from a mixed-breed dog), and the cFVII genome (from a Boxer in the NCBI database, accession number NW_876275) revealed two other amino acid changes in exon 8: R227Q (CGG → CAG at nucleotide position 9641) and H373Y (CAC → TAC at nucleotide position 10078). All five dogs homozygous for the G96E mutation in this study were homozygous for the R227Q polymorphism and homozygous for H373; this haplotype (Q227/H373) was found only in FVII-deficient dogs. Eleven normal non-Beagle dogs analyzed were homozygous Q227/Y373, homozygous R227/Y373 or homozygous R227/H373, and there did not appear to be an appreciable phenotypic effect for any polymorphism, with plasma FVII coagulant activities overlapping with a broad reference range (52–159%; reference range 57–181%) [15]. In transient transfection studies on HEK-293 cells, we observed a marked reduction in FVII coagulant activity in the mutant construct as measured by PT on conditioned media (cFVII-wt 16.1 ± 0.7 s vs. cFVII-G96E 108 ± 2.9 s). The presence of the Q227/H373 haplotype in the context of the wild-type or mutant constructs resulted in a minimal reduction in coagulant activity that is unlikely to be of any clinical significance (data not shown). We conclude that G96E is the causative mutation of FVII deficiency in the five Beagles tested in this study and that Q227 and H373 are polymorphisms that do not have any appreciable phenotypic significance.

In addition to DNA sequence analysis, we developed a rapid detection method for the G96E mutation. Exon 5 was amplified by PCR, and the resulting 311 bp fragment was subjected to restriction digestion with MnlI. cFVII-wt restriction digestion generates five fragments of 27, 25, 32, 27, and 200 bp. The G96E mutation abolishes the second restriction site for MnlI, indicated by generation of a 57 bp fragment (and absence of the 25 and 32 bp fragments) that is not present in the digest of the normal allele. As predicted, screening of related Beagles heterozygous for the G96E mutation from the Swiss research colony revealed the presence of the five expected fragments in dogs with the wild-type sequence and an additional 57 bp fragment (Fig. 3).

Figure 3.

 Restriction digestion of exon 5 of the canine factor VII (cFVII) gene by MnlI. (A) Schematic representation of restriction digestion of a 311 bp polymerase chain reaction fragment from exon 5 of the cFVII wild-type and G96E mutant proteins. The G96E mutation abolishes the second restriction site for MnlI, indicated by generation of a 57 bp fragment (and absence of the 25 and 32 bp fragments) that was not present in the digest of the normal allele. (B) Polyacrylamide (8%) gel of restriction digestion. M, molecular size DNA marker; lane 1, uncut DNA; lanes 2–3, homozygote G96E; lanes 4–5, heterozygote G96E; lane 6, wild-type.

Thromboelastometry of normal and G96E heterozygote and homozygote dogs

To further analyze the effects of the FVII mutation on coagulation, we determined the plasma coagulation patterns amongst normal, carrier and affected dogs using rotational thromboelastometry. Six of 11 homozygous affected animals showed no clot even after 60 min. Quantitative analysis of those that did form clots indicated that several ROTEM parameters were significantly affected in FVII-deficient dogs (n = 5) compared with normal (n = 12) and carrier dogs (n = 9), including clotting time (time for initial clot formation), maximum velocity (MaxV), and time to reach maximum velocity (MaxV-t), all with P-values < 0.001 (Fig. 4). Both MaxV and MaxV-t are measurements of the rate of thrombin generation. There were no statistically significant differences in maximum clot firmness (amplitude of clot) between normal, carrier, and affected dogs (data not shown). Interestingly, a comparison of carrier and normal dogs revealed a trend towards decreased clotting activity in the carriers. This was further confirmed using a modified PT assay using limiting amounts of TF (data not shown).

Figure 4.

 Thromboelastometry analysis. (A) Clotting times, (B) time to reach maximum velocities, and (C) maximum velocities in normal, carrier, and affected dogs. Each symbol represents a single animal. Affected dogs that failed to clot (n = 6) are not displayed in this analysis.

Recombinant cFVII-wt and cFVII-G96E expression and purification

cFVII-wt and cFVII-G96E (both with R227) were used to generate stable HEK-293 cell clones. cFVII-wt and cFVII-G96E proteins were purified from conditioned medium but, surprisingly, yields of cFVII-G96E were consistently low (10% of cFVII-wt). Nonetheless, both purified proteins appeared as a single band on denaturing SDS–PAGE, with an approximate molecular weight of 50 kDa (Fig. 5A), indicating lack of autoactivation during the purification process. In a PT clotting assay, the cFVII-G96E construct had < 2% activity of cFVII-wt (Fig. 5A). Interestingly, neither complete (with human FXa) nor partial (with shTF) activation of the cFVII-G96E mutant improved its activity (< 2% of cFVII-wt) in a PT clotting assay (Fig. 5B,C), further indicating its inherent lack of appreciable extrinsic pathway activity even in a two-chain (enzymatic) form, in contrast to cFVII-wt, which was completely cleaved with the addition of either human FXa or shTF (Fig. 5B,C).

Figure 5.

 Purified canine recombinant wild-type and G96E mutant factor (F) VII proteins. (A) M, molecular size protein marker; 1, canine FVII wild-type (cFVII-wt); 2, cFVII-G96E, under reducing conditions. In vitro activity is based on prothrombin time, with recombinant wild-type protein being 100%. (B) Activation of purified canine proteins by human FXa. 1, cFVII-wt, unactivated; 2, cFVII-wt, activated; 3, cFVII-G96E, unactivated; 4, cFVII-G96E, activated. (C) Activation of purified canine proteins by human soluble tissue factor. 1, Human soluble tissue factor; 2, cFVII-wt, unactivated; 3, cFVII-wt, activated; 4, cFVII-G96E, unactivated; 5, cFVII-G96E, activated.

In the hFVII protein, two distinct mutations (G97C and Q100R) in the EGF-2 domain have been shown to result in markedly reduced antigen and activity levels associated with reduced secretion [16]. In order to determine whether the G96E mutation could affect cFVII secretion, we performed Western blot analysis on supernatants and cell lyzates from cultured cells (48 h) stably transfected with cFVII-wt or cFVII-G96E. Densitometric analysis of the specific bands indicated that 96% of cFVII-G96E protein was retained intracellularly, in contrast to cFVII-wt, which showed 15% intracellular retention (P < 0.0001) (Fig. 6).

Figure 6.

 Western blots of supernatant and cell lyzate from human embryonic kidney-293 stable clones expressing canine factor VII wild-type (cFVII-wt) and cFVII-G96E proteins, following gel electrophoresis under reducing conditions. 1–3, cFVII-G96E; 4–6, cFVII-wt; 7, purified cFVII-wt. Densitometric analysis indicated that 96% of the mutant protein (arrow) was retained intracellularly, compared with 15% of the wild-type protein (P < 0.001). Human β-actin was used as a loading control.


The molecular basis of hereditary FVII deficiency in the Beagle breed is a G to A missense mutation in exon 5, resulting in substitution of glycine 96 (GGA) by glutamic acid (GAA) in the EGF-2 domain. Affected dogs are homozygous for this mutation, and as expected for dogs of the same breed, Beagles from Switzerland, the USA, and Canada all have the same FVII mutation as a result of a founder and possibly popular sire effect. The presence of a single mutation causing FVII deficiency in Beagles, in contrast to the more than 130 mutations reported to cause FVII deficiency in humans, is advantageous for ease of mutation screening in research Beagle and companion animal populations. In addition to the G96E mutation, two polymorphisms, R227Q and H373Y, were detected in exon 8 of the cFVII gene. The haplotype with Q227/H373 was found exclusively in dogs homozygous for the G96E mutation. Our data from the transient transfection suggest that the negligible reduction in coagulant activity associated with this haplotype in the context of wild-type and mutant cFVII is not likely to be of clinical significance.

While plasma FVII coagulant activity was ≤ 4% in the homozygous FVII-deficient dogs in our study, plasma FVII antigen levels could not be determined because an antibody for a cFVII ELISA was not available. The G96E mutation identified in affected Beagles has not been previously reported in human patients, although a G96S substitution resulted in ∼2% FVII coagulant activity in a patient with a mild bleeding diathesis [3]. We confirmed by expression of the variant protein that canine G96E is the single causative mutation resulting in a marked reduction in FVII activity and secretion in vitro. The purified recombinant canine mutant protein was completely cleaved by FXa and partially cleaved in the presence of shTF, although both ‘activated’ proteins retained < 2% of the activity of activated cFVII-wt in a PT assay. Our data support a multifactorial defect with both impaired secretion and reduced activity. It is of interest that FVII protein sequence alignment for canine, human, mouse, rat, rabbit and zebrafish reveals that Gly96 is a conserved residue in all those species, suggesting its importance across a wide variety of species.

While several naturally occurring mutations in the EGF-2 domain of hFVII have been reported [3,16–19], few of these mutations have been expressed in vitro to determine the mechanism by which they result in the disease phenotype. Two of these mutations, hFVII-G97C and hFVII-Q100R, were associated with both markedly reduced FVII antigen levels and activities in affected patients (2% and < 1% for G97C and 12% and < 1% for Q100R, respectively), and hampered secretion of mutant proteins from stable clones [16]. Similar to the cFVII-G96E protein in our study, hFVII-Q100R exhibited an impaired response to shTF [18].

In addition to prolonged PTs, plasma from homozygous affected dogs showed impaired clotting profiles in rotational thromboelastometry, a global hemostatic test that was recently shown to strongly correlate with quantitative measurement of thrombin generation [20]. As expected, in the thromboelastometry assay, plasma from affected Beagles exhibited a markedly impaired clotting profile. Interestingly, the clotting profile of carrier dogs also appeared reduced when compared with normal canine plasma; the same was observed in a modified clotting assay.

Currently, the frequency of hereditary FVII deficiency in Beagle research colonies and in the companion animal population is unknown. However, using a small sample of 104 Beagles from a US research colony, we determined the allelic frequency of the G96E mutation to be approximately 31%. Obviously, hereditary FVII deficiency is also present in the companion animal Beagle population, with the same FVII-G96E mutation identified in three apparently unrelated dogs from different geographical areas in the USA and Canada. Heretofore, elimination of hereditary FVII deficiency from research colonies and breeding stock has been difficult based on measurement of plasma FVII coagulant activity, which does not reliably detect carrier dogs.

Characterization of the G96E mutation causing FVII deficiency in Beagles and development of a DNA-based screening test will aid in identification of dogs that are homozygous or heterozygous for this coagulation defect, allowing investigators to effectively eliminate this trait from dogs utilized in pharmacological and toxicological research studies. Conversely, affected animals may be useful for those interested in novel approaches to the treatment of inherited coagulation disorders. A breeding colony of FVII-deficient Beagles has been established at the University of Pennsylvania as a large animal model of FVII deficiency. Their response to low (5 μg kg−1) and standard (30 μg kg−1) dose rhFVIIa has been shown to be similar to that observed in hFVII-deficient patients [21], supporting the usefulness of this disease model in the study of therapeutic strategies.


The authors thank S. Krishnaswamy for his generous gift of soluble human tissue factor, and gratefully acknowledge helpful discussions with R. Toso and R. M. Camire, and the technical assistance of A. Seng.

Disclosure of Conflict of Interests

This work was supported in part by the NIH U01 HL66948 (to K. A. High) and RR02512 (to U. Giger) and the Howard Hughes Medical Institute (K. A. High).