Genetic mechanisms of age regulation of protein C and blood coagulation.

Blood coagulation activity in humans increases with age. We previously identified two genetic elements, age-related stability element (ASE; GAGGAAG) and age-related increase element (AIE; unique stretch of dinucleotide repeats), which were responsible for age-related stable and increasing expression patterns, respectively, and together recapitulated normal age regulation of the human factor IX (hFIX) gene. Here we report the age-regulatory mechanisms of human anticoagulant protein C (hPC), which shows an age-stable pattern of circulatory levels. The murine protein C gene showed an age-related stable expression pattern in general agreement with that of the hPC. Through longitudinal analyses of transgenic mice carrying hPC minigenes, the hPC gene was found to have a functional age-related stability element (hPC ASE; CAGGAAG) in the 5'-upstream proximal region but was found to lack any age-related increase element. Three other ASE-like sequences present in the hPC gene, GAGGAAA and (G/C)AGGATG, also bound nuclear proteins but were not active in the age regulation of the hPC gene. Functional hPC ASE and hFIX ASE were apparently generated through convergent evolution, and hFIX ASE can fully substitute for the hPC ASE in conferring age-related stable expression pattern of the hPC gene. In the presence of the hPC ASE, hFIX AIE can convert the age-stable expression pattern of the hPC gene to a hFIX-like age-related increase pattern. These results support the universality of ASE and AIE functions across different genes. Clearance of hPC protein from the circulation was not significantly affected by age. We now have established the basic mechanisms responsible for the age-related increase of blood coagulation activity.

Blood coagulation activity in normal humans increases with age (1)(2)(3). This is apparently due to age-related increase in disparity between procoagulation and anticoagulation activities and may have critical contributions to age-related increases in frequency of thrombotic and cardiovascular diseases (4 -7). Human PC 1 is an important component of the anticoagulation pathway, while hFIX is a key blood procoagulation factor playing critical roles in blood coagulation (8,9). Epidemiological studies showed that the activity and plasma levels of hFIX increased significantly with advancing age in normal human populations (3,10). In comparison with hFIX, the plasma levels of hPC showed only marginal age-associated changes (3,11). We previously identified two genetic elements, ASE and AIE (renamed from AE 5Ј and AE 3Ј, respectively), which are required for age-related stability and increase of hFIX expression, respectively (12). ASE, originally identified as a footprint sequence between nucleotide (nt) Ϫ802 and nt Ϫ784 of the hFIX gene, has a core sequence, GAGGAAG, matching the transcriptional factor PEA-3 consensus sequence ((G/C)AG-GA(A/T)G) (13,14). AIE, present in the middle of the hFIX 3Ј-untranslated region (UTR), is composed of 102-bp dinucleotide repeats (AT, GT, and CA) and has the potential to form three distinct stem loop structures in its RNA form (15). Murine factor IX (mFIX) shows an age-related increase in both circulatory protein levels and gene expression in the liver (16,17), and the mFIX gene also contains ASE-like sequences in the 5Ј-upstream and a stretch of 106-bp dinucleotide repeats in the 3Ј-UTR (18).
New critical questions have emerged, including whether the age-regulatory elements ASE and AIE can play roles in other genes, especially anticoagulation factor genes, and what are the fundamental genetic mechanisms responsible for the agerelated increase of blood coagulation activity. To address these issues, we focused our efforts on the hPC gene. Human PC and hFIX share similarities in protein structure and gene coding region organization, but the 5Ј-flanking regions and 3Ј-UTRs of their genes are grossly dissimilar (15, 19 -21). For example, the 5Ј-flanking region of the hFIX gene, from approximately beyond nt Ϫ350, where the functional ASE is located, was derived from retrotransposed LINE-1 sequences (22), while the corresponding 5Ј-flanking region of the hPC gene was not (20). Furthermore, the 3Ј-UTR of the hFIX gene is about 1.4 kb in length, whereas that of the hPC gene is only 295 bp, and no AIE-like element or dinucleotide repeats are present (15,21). These differences between the hFIX and hPC genes may result in the substantial dissimilarity in their age-related expression patterns.
In this paper, we report the basic genetic mechanisms of age regulation of the hPC gene and the functional universality of the hFIX-derived ASE and AIE, thus laying the foundation for comprehensive understanding of the age-related increase in blood coagulation activity.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes and DNA modification enzymes were purchased from Invitrogen and New England Biolabs. Radioactive nucleotides, [␣-32 P]dCTP and [␥-32 P]ATP, were obtained from Amersham Biosciences. Mouse anti-hPC monoclonal antibody and rabbit polyclonal anti-hPC antibody were purchased from Celsus Laboratories. Horseradish peroxidase-linked goat anti-rabbit IgG was purchased from Invitrogen. Antibodies used for hFIX-specific enzyme-linked immunosorbent assay (ELISA) were described previously (12). Medium, fetal calf serum, penicillin, and streptomycin for mammalian cell cultures were obtained from Invitrogen. Fugene-6 transfection reagent was purchased from Roche Molecular Biochemicals. HepG2 cells, a human hepatoma cell line, were obtained from ATCC. C57BL/6, SJL, and CD-1 mice used for generating transgenic animals were purchased from the Jackson Laboratory. All other reagents were of the highest quality commercially available.
Construction of hPC Minigenes-Human PC minigene construct, Ϫ1462hPCm1, was first constructed and subsequently used to generate other hPC minigene constructs. The hPC gene nucleotide numbering system used was that with the major transcriptional initiation site defined as nt ϩ1 (20). A 3022-bp sequence spanning nt Ϫ1462 through nt ϩ1560 of the hPC gene was amplified by polymerase chain reaction (PCR) using human genomic DNA as a template, and 5Ј forward primer with SphI linker (5Ј-CAAGCATGCGAATTCTGTAAGCATTTCCT-3Ј) and 3Ј reverse primer with MscI linker (5Ј-GAAATTCCCCAGGTGGC-CAC-3Ј). The amplified fragment was then inserted into pUC119-hPC (kindly provided by Dr. Francis Castellino, University of Notre Dame) to replace its 5Ј portion released by SphI/MscI double digestion. The 3Ј-end portion of the resultant intermediate construct, from the internal Sse8387I site at nt ϩ10497 in the 3Ј-UTR to the EcoRI site outside of the poly(A) attachment site (nt ϩ10783), was released by Sse8387I/ EcoRI double digestion. This portion was then replaced by a PCRamplified fragment with Sse8387I/EcoRI sticky ends (612 bp in length, spanning nt ϩ10497 through ϩ11108 with a 325-bp extension into the 3Ј-flanking region sequence of the hPC gene), thus generating minigene Ϫ1462hPCm1. By replacing the 5Ј portion of Ϫ1462hPCm1 (nt Ϫ1462 through ϩ1560) with PCR-amplified SphI/MscI fragments, spanning nt Ϫ849, Ϫ802, or Ϫ82 to nt ϩ1560, new minigenes Ϫ849hPCm1, Ϫ802hPCm1, or Ϫ82hPCm1 were generated, respectively. ASE/ Ϫ802hPCm1 or ASE/Ϫ82hPCm1 were similarly constructed by inserting a single copy of a 32-bp fragment containing hFIX ASE (nt Ϫ802 through Ϫ771 of the hFIX gene) with SphI sticky ends into Ϫ802hPCm1 or Ϫ82hPCm1 at the 5Ј-end SphI site, respectively. Minigene pseudoASE/Ϫ802hPCm1 was generated by inserting a 23-bp fragment containing pseudo-ASE (nt Ϫ878 through Ϫ856 of the hPC gene) with SphI sticky ends into Ϫ802hPCm1 at the 5Ј-end SphI site. Minigene Ϫ1462hPCm1/AIE or Ϫ1462hPCm1/AIEr was constructed by inserting a 102-bp hFIX AIE fragment (nt ϩ32142 through ϩ32243 of the hFIX gene) with Sse8387I linker into the 3Ј-UTR of Ϫ1462hPCm1 at the Sse8387I site in a normal or reversed orientation, respectively. ASE/Ϫ1462hPCm1/AIE was generated by inserting above hFIX ASE into Ϫ1462hPCm1/AIE at the 5Ј-end SphI site. Human FIX minigenes, ASEmin/Ϫ416FIXm1 and AIE/Ϫ416FIXm1/ASE, were constructed by inserting a single copy of the 17-bp minimum ASE sequence (ASEmin: nt Ϫ795 through Ϫ779 of the hFIX gene) or a copy of the above AIE into minigenes Ϫ416FIXm1 and Ϫ416FIXm1/ASE (12) at the 5Ј-end SphI site, respectively. Minigene Ϫ802FIXm1/mDR was constructed by inserting a 106-bp dinucleotide repeat sequence of the mFIX gene (nt ϩ2018 through ϩ2123 of the mFIX cDNA) (18) at the BamHI site of Ϫ802FIXm1 (12). PCR-amplified sequences and ligation sites of all of the new constructs were confirmed by automated dideoxy sequencing.
Cell Culture, Transfection, and Transient Expression Assays-HepG2 cells were cultured in Dulbecco's modified eagle medium supplemented with L-glutamine, 25 mM HEPES buffer, 110 mg/liter sodium pyruvate, and 10% fetal bovine serum. Cell lines were maintained in a 5% CO 2 atmosphere at 37°C. Transfection was carried out by using Fugene 6 reagent according to the manufacturer's instructions. Transient expression activities of minigene constructs were determined with HepG2 cells as follows. Mixtures of expression constructs (total of 3 g/well, six-well plate) composed of an hPC minigene construct and pCH110 (␤galactosidase expression vector) (19) at a 3:1 (w/w) ratio were transfected into HepG2 cells by using Fugene-6 reagent and serum-free medium (3:97, v/v). Human PC produced was quantified by hPC-specific ELISA using a three-antibody system (24): mouse anti-hPC monoclonal antibody for antigen catching, rabbit polyclonal anti-hPC antibody as the second antibody, and horseradish peroxidase-linked goat anti-rabbit IgG for the detection. Average net hPC production levels from four or five independent assays were obtained after subtracting the background hPC produced by HepG2 cells transfected only with pCH110.
Construction and Longitudinal Analysis of Transgenic Mice-Transgenic mice were constructed and subjected to longitudinal analyses as previously described (12) with minor modifications. Human PC minigenes released from minigene expression vectors by HindIII/EcoRI digestion were purified and microinjected into fertilized eggs of C57BL/ 6 ϫ SJL mice (1-2 ng of DNA/egg) and implanted into foster mother animals (CD-1). Founder animals were genotyped at 2-3 weeks of age by quantitative multiplex PCR analyses of tail tissue DNA samples. Two pairs of primer were used: one specific to the 746-bp hPC transgene fragment corresponding to exons 2-9 (nt ϩ1577 through ϩ9911 in the genomic nucleotide number (20,21)) and one that produced a 494-bp mouse ␤-globin gene fragment (endogenous control) (nt ϩ2590 through ϩ3083 (25)). To be quantitative, 50 ng of genomic DNA from the tail tissue, 200 ng of hFIX primers, and 100 ng of mouse ␤-globin primers were applied. The transgene copy number was determined based on the relative ratio between the hFIX-specific band and that of the ␤-globin gene. Animals carrying Ն5-10 copies of the hPC transgene were selected for further analyses. Founders were back-crossed with nontransgenic mice (C57BL/6 ϫ SJL) to generate F1 progeny animals. Homozygous F2 animals were generated by crossing among heterozygous F1 littermates. The zygosity status of animals was determined by quantitative multiplex PCR analysis as described above. Longitudinal blood collection via snipped tails was started at 1 month of age from individual animals of 4 -8 representative founders for each minigene construct. Animals of subsequent generations of representative lines were similarly subjected to longitudinal analyses. Circulatory hPC or hFIX levels of transgenic mice at each age point were quantified by duplicated ELISA. Human FIX ELISA was applied as previously described (12). To minimize experimental fluctuations from assay to assay in the longitudinal analysis, overlapped serum samples from the previous assay group were included in each assay. Transgene positional effect can substantially affect absolute levels of transgene expression among individual animals. Therefore, unless otherwise mentioned, our longitudinal animal analyses were focused on determining the age-related pattern of minigene expression for each individual animal. All animal experiments were carried out under the institutional guidelines for ethical animal use (Office for Protection from Research Risk No. A3114-01).
Electrophoretic Mobility Shift Assay (EMSA)-EMSAs of pseudo-ASE, hPC ASE, and other similar elements were carried out as previously described (26) with minor modifications. Nuclear extracts (NEs) were prepared from HepG2 cells or liver tissues of normal mice (C/ 57BL/6) at 5 months of age. Single-stranded oligonucleotides were chemically synthesized and used to prepare double-stranded (ds; see Fig. 4) oligonucleotides (sequences shown in the figure legend). These oligonucleotides were labeled with 32 P to a specific activity of ϳ1.9 ϫ 10 9 cpm. Aliquots (20,000 cpm) were incubated with 10 g of NEs in the presence of 1 g of double-stranded poly(dI-dC) in DNA binding buffer for 20 min at room temperature and subjected to vertical polyacrylamide gel electrophoresis analysis. For competition assays, unlabeled oligonucleotides (double-stranded) in the amount of 100-fold excess were added to the labeled ones.
Human PC Clearance Assay-Half-clearance times of circulatory hPC were determined as previously described for hFIX (12) with appropriate modifications. Aliquots of plasma-derived hPC preparation (4 g/0.1 ml of phosphate-buffered saline; Hematologic Technologies, Inc.) were injected via tail vein into three separate groups of C57BL/6 ϫ SJL mice of 2, 8 -10, and 18 -20 months of age (n ϭ 3 per group), respectively. Circulatory hPC levels were determined by ELISA using serum samples, which were prepared from blood samples (ϳ50-l aliquots) collected via snipped tails at 10 min and at 2, 6, 12, 18, 24, 30, 36, and 48 h after protein injection.

Age-related Expression
Patterns of the mPC Gene-To provide the rationale for utilizing transgenic mice for analyzing age regulation of hPC gene expression, we first examined the age-related expression pattern of the mPC gene (Fig. 1A). Mouse PC liver mRNA levels rapidly increased during the perinatal stage (Fig. 1A, lanes 2-8), reaching the young adult level around the time of weaning (ϳ3 weeks of age), followed by relative stable levels through old age (20 months) (Fig. 1A,  lanes 8 -14 and B).
Genetic Control of Age-related Expression Patterns of hPC Minigenes-The first series of hPC minigenes were constructed to identify genetic elements responsible for age-related regulation of the hPC gene, specifically focusing on the PEA-3-like elements ( Fig. 2A). There were two PEA-3-like elements present in the proximal 5Ј-upstream of the gene. The first element was located at nt Ϫ832 through Ϫ826 (CAGGAAG; designated as hPC ASE), which is consistent with the PEA-3 consensus motif and different from the hFIX ASE (GAGGAAG) by one nucleotide. The second element was present at nt Ϫ871 through Ϫ865 (GAGGAAA; designated as pseudo-ASE), differing from the PEA-3 consensus motif by one nucleotide. All of the hPC minigenes showed similar transient expression activities as assayed with HepG2 cells (Fig. 2A). The second series of hPC minigenes were constructed with hFIX-derived ASE or AIE (Fig. 2B). The presence of ASE did not have any significant effect on transient hPC expression in HepG2 cells, while the presence of AIE in hPC minigenes lowered their transient expression activities by ϳ30% (Fig. 2B) in agreement with our previous observations on hFIX minigenes containing AIE (12).
Transgenic mice carrying minigenes Ϫ1462hPCm1 (n ϭ 40), Ϫ849hPCm1 (n ϭ 31), Ϫ802hPCm1 (n ϭ 35), or Ϫ82hPCm1 B, human PC minigene constructs containing hFIX ASE and AIE. The structure is depicted with the promoter regions (thick horizontal lines at left) with the 5Ј-end nt number. Transcribed hPC regions (gray rectangles with thin lines representing the first intron) are followed by 3Јflanking sequence regions (thick horizontal lines at right). Arrow, transcription start site; thin vertical line in the large rectangular box, translation stop codon; pA, polyadenylation signal; ASE, age-stable expression element of the hFIX gene (black rectangular box at the 5Ј-end); psASE, pseudo-ASE of the hPC gene (white rectangular box at the 5Ј-end); AIE, age-related increase element of the hFIX gene (black rectangular box at the 3Ј-end). AIEr, reversed AIE (shadowed gray rectangle box at the 3Ј-end). Transient expression activities assayed with HepG2 cells are presented relative to that of Ϫ1462hPCm1 (average 103 ng/10 6 cells/24 h). Negative regulatory activities previously reported for the region nt Ϫ1462 through Ϫ82 using a heterologous reporter chloramphenicol acetyltransferase gene (20) were not observed with native minigenes. (n ϭ 44) were constructed and subjected to longitudinal analyses of circulatory hPC. Age-related patterns of circulatory hPC levels in individual animals are shown in Fig. 3, A-D. Animals carrying either Ϫ1462hPCm1 or Ϫ849hPCm1 showed remarkable age-related stability in circulatory hPC at various levels as high as ϳ3 g/ml plasma, similar to natural hPC levels in humans (Fig. 3, A and B). The stability of circulatory hPC correlated with similar age-stable hPC mRNA levels in the liver (Fig. 3F), and was consistent with the absence of any AIE-like element in the hPC gene. These observations were reproducible in all animals under investigation, regardless of founder line, initial prepubertal hPC levels, sex, generation (including F2 animals), or zygosity status of the transgene. In contrast, circulatory hPC levels of animals carrying Ϫ802hPCm1 or Ϫ82hPCm1, which showed levels as high as ϳ590 or ϳ40 ng/ml serum, respectively, at 1 month of age, rapidly decreased through puberty to much lower or undetectable levels for the rest of their life spans (Fig. 3, C and D). This rapid age-related decline was observed in all animals, independent of founder line, initial prepubertal levels, sex, or zygosity status of the transgenes, and correlated with a similar decline in the steady state level of liver hPC mRNA (Fig. 3G). These results suggested that the region nt Ϫ849 through Ϫ803, where a PEA-3 element (CAGGAAG; hPC ASE) located, was critically required for age-stable expression of the hPC gene.
Another PEA-3 like element present in the region nt Ϫ871 through Ϫ865, GAGGAAA (pseudo-ASE), was also tested with transgenic mice carrying pseudoASE/Ϫ802hPCm1. Animals carrying pseudoASE/Ϫ802hPCm1 showed an age-related decline in hPC expression (n ϭ 23) (Fig. 3E) similar to those carrying Ϫ802hPCm1, indicating that pseudo-ASE does not function as an age stability element.
In EMSA, pseudo-ASE and hPC ASE bound different liver nuclear proteins, as evidenced by the absence of cross-competition (Fig. 4, lanes 7 and 14) and a subtle difference in the apparent sizes between them (Fig. 4, lanes 3 and 10). Furthermore, hFIX ASE efficiently out-competed hPC ASE for its binding nuclear protein (Fig. 4, lane 13), indicating that hFIX ASE (GAGGAAG) and hPC ASE (CAGGAAG) bind the same nuclear protein. Protein binding to pseudo-ASE was present in HepG2 NEs at a significant concentration level (Fig. 4, lane 2), while the hPC ASE-binding protein was present in HepG2 NEs at a very low level, if at all (Fig. 4, lane 9).
Human PC minigenes used in this study had two additional consensus PEA-3 sequences. One is in intron 1 (nt ϩ694 to ϩ700; GAGGATG; designated as Intron1PEA-3), and the other is in the exon 9 coding region (nt ϩ10286 to ϩ10292; CAG-GATG; designated as Exon9PEA-3) (21). Interestingly, they were not able to confer age-related stabilization effects on hPC gene expression as evidenced by age-unstable expression patterns of Ϫ802hPCm1 as well as Ϫ82hPCm1 in animals (Fig. 3,  C and D). EMSA indicated that both elements bound nuclear proteins present in both liver and HepG2 cell NEs (Fig. 5A,  lanes 2, 3, 8, and 9), but like in the case of pseudo-ASE above, these proteins were found to be different from those binding to functional hPC or hFIX ASE as shown by cross-competition assays (Fig. 5A, lanes 6 and 12). This was further supported by cross-competition assays using 32 P-labeled hPC ASE, unlabeled Intron1PEA-3 (GAGGATG), or Exon9PEA-3 (CAGGATG) (Fig. 5B, lanes 8 and 9). Cross-competition assays also showed that Intron1PEA-3 and Exon9PEA-3 bound a common nuclear protein (Fig. 5B, lanes 10 and 11).
Functional Universality of ASE and AIE-To address the possible functional universality of ASE elements, another series of hPC minigenes constructed with hFIX-derived ASE (Fig.  2B) were longitudinally analyzed for age-related expression in transgenic mice. Unlike animals carrying Ϫ82hPCm1 (Fig. 3D) or Ϫ802hPCm1 (Fig. 3C), transgenic animals carrying ASE/ Ϫ82hPCm1 (n ϭ 32) or ASE/Ϫ802hPCm1 (n ϭ 30), which have the hFIX ASE inserted at the 5Ј-upstream position of hPC minigenes, reproducibly showed age-stable patterns of circulatory hPC at least up to 9 months of age (the last assay time points) (Fig. 6, A and B). The observed conversion in ageregulatory patterns was independent of founder animals, generation, initial pubertal levels, sex, or zygosity state. This age stability of the hPC circulatory levels paralleled similar agestable levels of hPC mRNA in the liver (Fig. 6F). Thus, we conclude that the hFIX-derived ASE can functionally replace the hPC endogenous ASE. ASEmin/Ϫ416FIXm1, a hFIX minigene with only a 17-bp nucleotide sequence containing the 7-bp essential PEA-3 motif sequence (GAGGAAG) in the middle inserted at the 5Ј-end of minigene Ϫ416FIXm1, also fully converted the age-unstable hFIX expression of Ϫ416FIXm1 (12) to age-stable expression (supplemental Fig. A), further supporting the importance of the motif sequence in vivo.
The functional universality of AIE was then tested in transgenic mice with a new set of hPC minigenes, Ϫ1462hPCm1/ AIE, ASE/Ϫ1462hPCm1/AIE, and Ϫ1462hPCm1/AIEr. These minigenes were constructed by inserting an hFIX-derived AIE, a minimal 102-bp core sequence of mostly dinucleotide repeats (15), in the middle of the 3Ј-UTR of Ϫ1462hPCm1 and ASE/ Ϫ1462hPCm1, respectively. These minigenes contained one or two units of functional ASE as described above. Animals with Ϫ1462hPCm1/AIE or ASE/Ϫ1462hPCm1/AIE clearly switched age patterns of circulatory hPC from the hPC age-stable pattern to an hFIX-like age-related increase pattern (Fig. 6, C and  D). This switch in circulatory hPC patterns also correlated with a similar switch in the age-related liver mRNA levels (Fig. 6G) and was reproducibly observed with all animals carrying either Ϫ1462hPCm1/AIE (n ϭ 30) or ASE/Ϫ1462hPCm1/AIE (n ϭ 28) regardless of founder line, initial pubertal hPC level, sex, generation, or zygosity. These results demonstrated that AIE can confer an age-related increase expression pattern on the hPC gene, which lacks any endogenous AIE-like element, and that the endogenous hPC ASE fully functioned in combination with AIE.
Minigene Ϫ1462hPCm1/AIEr, an hPC minigene with a reversed AIE sequence at the 3Ј-UTR, which is still capable of forming stem-loop structures in its RNA form, also showed an age-related increase expression pattern in transgenic animals (n ϭ 18) (Fig. 6E), indicating that AIE functions independent of its orientation in the 3Ј-UTR. However, AIE/Ϫ416FIXm1/ASE, an hFIX minigene with the AIE inserted at the 5Ј-end, showed only age-stable, but not an age-associated increase in circulatory hFIX expression in transgenic animals (supplemental Fig.  B), suggesting that AIE does not function as a transcriptional enhancer. A stretch of 106-bp dinucleotide repeats (mDR) present in the 3Ј-UTR of the mFIX gene was also tested for its possible age-related function. This dinucleotide repeat was composed of a GA repeat (H-DNA-forming sequence) and AT repeat (potential stem-loop-forming sequence in RNA form) (18). Transgenic mice carrying Ϫ802FIXm1/mDR, an hFIX minigene with one unit of mDR in place of the hFIX AIE in the 3Ј-UTR of Ϫ802FIXm1 (12), also showed substantial age-related increase in the circulatory levels (supplemental Fig. C).
Age-associated Effect on Circulatory hPC Clearance Time-Clearance of proteins from the circulation can also affect their circulatory levels. The clearance time of hPC in the circulation of mice was determined and summarized in Table I. Human PC clearance from the circulation was found to follow biphasic kinetics with rapid (␣) and slow (␤) phases. The clearance half-times (t1 ⁄2 ) of these two phases were ϳ6 and 16 h, respectively.

DISCUSSION
In this paper, we describe for the first time the genetic basis responsible for the age-related increase of blood coagulation activity and the mechanistic evidence for the functional universality of age-regulatory mechanisms identified in the hPC and hFIX genes.
General characteristics of mouse blood coagulation factors parallel their human counterparts (27). Our findings on the mPC levels in the liver (Fig. 1) indicate that the overall agerelated expression pattern of the mPC gene correlates well with the known age-related pattern of circulatory hPC levels in humans (3,4). Both mPC and hPC rapidly increase their expression during the perinatal stage, followed by a general agestable expression pattern. This further supported the rationale of using a mouse model for studying age-related regulatory mechanisms of the hPC gene.
Age-related stable patterns of circulatory hPC levels observed for transgenic mice carrying hPC minigene Ϫ1462hPCm1 or Ϫ849hPCm1 recapitulated the natural patterns of mPC and hPC genes (Fig. 3, A and B). This suggests that these minigenes contained all of the necessary genetic elements required for age-stable expression of the hPC gene. Dramatic changes of age-related expression patterns of circulatory hPC in the animals carrying minigene Ϫ802hPCm1 or Ϫ82hPCm1 (Fig. 3, C and D) from those of Ϫ1462hPCm1 or Ϫ849hPCm1 allowed us to locate a critical element responsible for age-stable expression of the hPC gene to the region spanning nt Ϫ849 through Ϫ803. This region contains a single copy of the PEA-3 consensus element (CAGGAAG; hPC ASE) and no other known transcriptional elements, suggesting the critical role of hPC ASE in maintaining age-stable hPC expression. Although both hPC ASE and hFIX ASE are of the PEA-3 consensus elements ((G/C)AGGA(A/T)G), it is important to find that both elements actually bind the same liver nuclear protein of the ETS super family, as shown by DNA-protein binding analyses (Fig. 4, lanes 10 and 13). Very low levels, if any, of hPC ASE-binding protein observed in HepG2 cell NEs (Fig. 4, lane 9) may correlate with virtually no effects of hPC or hFIX ASE on transient minigene expression activities in HepG2 cells (Fig. 2).
A nuclear protein that bound to pseudo-ASE (GAGGAAA) was different from that binding to functional ASE of hPC or hFIX (Fig. 4, lanes 6 and 7). Since transgenic animals carrying pseudoASE/Ϫ802hPC showed age-related decline expression patterns of circulatory hPC, similar to those of the animals carrying Ϫ802hPCm1, pseudo-ASE and its binding protein are not active in conferring age-related stabilization of the hPC gene expression. Furthermore, as shown by ageunstable hPC patterns observed in the animal carrying Ϫ802hPCm1 or Ϫ82hPCm1 (Fig. 3, C and D), the other two PEA-3 consensus elements present in these minigenes, Intron1PEA-3 (GAGGATG) and Exon9PEA-3 (CAGGATG), do not function as age-related stabilization elements. This is consistent with the facts that these elements also bind a common nuclear protein (Fig. 5A), but the protein is different from that binding to functional hPC ASE as demonstrated by absence of cross-competition, size difference, and presence in both HepG2 cells and the liver (Fig. 5A, B). Together, these results indicate that specific single-base differences of PEA-3 motif (G/C)AGGA(A/T)G facilitate highly selective binding of specific nuclear proteins of the large ETS super family (14,28), resulting in distinctly different functions. Namely, not all of the known PEA-3 consensus sequences and binding proteins are active in age regulation of the genes. Efforts to identify the ASE-binding protein and to determine possible functions of GAGGAAA and (G/C)AGGATG are in progress.
As demonstrated by the age stability of hPC expression in animals carrying ASE/Ϫ82 hPCm1 or ASE/Ϫ802hPCm1 (Fig.  6, A and B), hFIX ASE can functionally substitute for hPC ASE in conferring the age-related stable hPC expression, strongly suggesting the functional universality of ASE. This is particularly interesting because hFIX ASE and hPC ASE have different evolutionary origins. Human FIX ASE was derived from a retrotransposed LINE-1 sequence (22), whereas hPC ASE was not (20). It is important to note that all animals carrying ASE/Ϫ82hPCm1 (Fig. 6A) showed agestable hPC expression but maintained absolute hPC levels very similar to or only marginally higher than the prepubertal hPC levels of animals carrying Ϫ82hPCm1 (Fig. 3D), even with transgene positional effects taken into account. The minigene ASE/Ϫ802hPCm1 also showed a similar phenomenon ( Fig. 3C; Fig. 6B). Together, these observations strongly suggest that ASE is a unique age-related transcription element essential for stabilizing gene expression at the prepu- bertal level, regardless of its absolute expression level (Fig.  7). ASE may exert its activity through alleviating unidentified negative effects on the promoter, an intriguing mechanism yet to be explored.
The hPC gene lacks any AIE-like element. However, as demonstrated by the animals carrying Ϫ1462hPCm1/AIE, insertion of one unit of hFIX AIE into the 3Ј-UTR dramatically changed the age-stable expression pattern of Ϫ1462hPCm1 to an age-related increase expression pattern, similar to that of the hFIX gene (Fig. 6C). A similar conversion was also obtained in animals carrying ASE/Ϫ1462hPCm1/AIE, which has an additional ASE at the 5Ј-end of Ϫ1462hPCm1, further supporting the fact that presence of just one unit of ASE and AIE is sufficient to confer such a drastic change in age regulation of gene expression (Fig. 6D).
The hFIX-derived AIE and its reverse form have the potential to form distinct stem loop structures in its RNA form. A hPC minigene, Ϫ1462hPCm1/AIEr, which has a unit of AIE inserted in the 3Ј-UTR in the reverse orientation, showed an hFIX-like age-related increase expression pattern in animals (Fig. 6E), suggesting that the age-regulatory function of AIE is independent of its orientation in 3Ј-UTR of the hPC or hFIX gene and may correlate with its unique secondary stem loop structure. The failure of inducing any age-related hFIX expression increase in animals carrying minigene AIE/Ϫ416FIXm1/ ASE, which has AIE inserted at the immediate 5Ј-end position of the hFIX promoter (supplemental Fig. B), indicates that AIE apparently does not function as a transcriptional enhancer. The age-related increase of circulatory hPC in the animals carrying Ϫ1462hPCm1/AIE was accompanied by a similar agerelated increase pattern in the liver mRNA level (Fig. 6G), suggesting that AIE functions to induce age-associated elevation of mRNA levels, most likely through increasing mRNA stability. The underlying molecular mechanisms remain to be elucidated.
As demonstrated by animals carrying Ϫ802FIXm1/mDR, mDR actually functions as the murine counterpart of the hFIX AIE (supplemental Fig. C), supporting the possibility that the age-regulatory mechanisms involving ASE and AIE may be widely utilized across different animal species. Whether the GA or AT repeat alone or both are needed to confer the agerelated increase in expression remains to be tested. Together, these results demonstrate that the dinucleotide repeats pres- ent in the 3Ј-UTRs of the hFIX or mFIX genes are active AIEs in conferring an age-related expression increase and that they can universally function across different genes.
Human PC or hFIX protein clearance from the circulation with advancing age should also affect the age-related expression patterns of the hPC or hFIX gene. The clearance time of hPC in the circulation of mice did not change significantly with age (Table I), which was consistent with our observations on hFIX clearance time (12). These results further support the conclusion that the mechanisms involving ASE and AIE, but not circulatory protein stability, are primarily responsible for age-related regulation of hPC and hFIX.
Present studies have determined for the first time that the fundamental mechanisms responsible for age-associated increase in blood coagulation activity involve different uses of two critical age-related regulatory elements, ASE and AIE (Fig. 7). The hPC gene uses ASE only, resulting in its age-stable expression pattern, while the hFIX gene uses both ASE and AIE for its age-related increase pattern of expression.
We now have set a new stage toward comprehensive understanding of age-related regulation and homeostasis of the blood coagulation system. The regulatory mechanisms may also be involved in age-related regulation of genes of many other physiological systems. Genetic elements identified in the present studies may find their valuable utilities in modifying age regulation of various genes and in developing optimized gene transfer vector systems for gene therapy.  7. Schematic presentation of ASE and AIE function in age-associated regulation of gene expression. Perinatal and prepubertal gene expression are shown by the thick dotted line. Age-stable gene expression at high or low level is shown as thick black lines. Age-related increasing or decreasing expression patterns are shown by thick gray lines or curved gray lines, respectively. Relative positions of birth, weaning, and puberty are shown with vertical gray lines. ASE stabilizes gene expression at the prepubertal level, which may be high or low, and AIE confers an age-related increase in gene expression (thin dotted lines with arrows). Both ASE and AIE are required for conferring age-related increasing patterns of gene expression.