Search

Human kininogen belongs to the plasma kallikreinkinin system. High molecular weight kininogen is the precursor for two-chain kinin-free kininogen and bradykinin. It has been shown that the two-chain kinin-free kininogen has the properties of anti-adhesion, anti-platelet aggregation, and anti-thrombosis, whereas bradykinin is a potent vasodilator and mediator of inflammation. In this study we show that the human kininogen gene is strongly up-regulated by agonists of the farnesoid X receptor (FXR), a nuclear receptor for bile acids. In primary human hepatocytes, both the endogenous FXR agonist chenodeoxycholate and synthetic FXR agonist GW4064 increased kininogen mRNA with a maximum induction of 8-10-fold. A more robust induction of kininogen expression was observed in HepG2 cells, where kininogen mRNA was increased by chenodeoxycholate or GW4064 up to 130-140-fold as shown by real time PCR. Northern blot analysis confirmed the up-regulation of kininogen expression by FXR agonists. To determine whether kininogen is a direct target of FXR, we examined the sequence of the kininogen promoter and identified a highly conserved FXR response element (inverted repeat, IR-1) in the proximity of the kininogen promoter (-66/-54). FXR/RXRalpha heterodimers specifically bind to this IR-1. A construct of a minimal promoter with the luciferase reporter containing this IR-1 was transactivated by FXR. Deletion or mutation of this IR-1 abolished FXR-mediated promoter activation, indicating that this IR-1 element is responsible for the promoter transactivation by FXR. We conclude that kininogen is a novel and direct target of FXR, and bile acids may play a role in the vasodilation and anti-coagulation processes.

High molecular weight kininogen (HK) binds specifically, saturably, and reversibly to neutrophils and also reciprocally inhibits the binding of fibrinogen to neutrophils. Since fibrinogen binds to the leukocyte integrin CD11b/18 (Mac-1, alpha M beta 2), we investigated whether HK bound to Mac-1 and whether the binding site was similar to that for factor X. We also examined whether one or both chains of cleaved HK (HKa) were involved. Two monoclonal antibodies, 2B5 (0.29 microM) to HK heavy chain domains 2 (D2) and 3 (D3), and C11C1 (0.26 microM) to HK light chain domain 5 (D5), inhibited by 99 and 93% the binding, respectively, of 125I-HK (8.3 nM) to neutrophils. To minimize steric hindrance, we further demonstrated that the Fab' fragments of 2B5 and C11C1 were able to inhibit the binding of this ligand to virtually the same extent as the intact antibody, indicating that, as in binding of HK to platelets and endothelial cells, both chains are involved. To directly demonstrate the involvement of each chain, we showed that the reduced alkylated light chain derived from HK and low molecular weight kininogen, which contains the same heavy chain as HK, each markedly inhibited the binding of HK to neutrophils. We localized the domain responsible for the binding in each chain by showing that recombinant D3 and D5 decreased the binding of HK to neutrophils. To define the receptor for HK, we employed three monoclonal antibodies to Mac-1: OKM1 and OKM10 to epitopes on the alpha M subunit and IB4 to an epitope on the beta 2 chain. OKM1, which can inhibit fibrinogen binding to neutrophils, inhibited HK binding by 79%, whereas the other antibodies inhibited HK binding less than 25%. Coagulation factor X also binds to Mac-1 on monocytes at a similar site to C3bi. Synthetic peptides which define noncontiguous surface loops in factor X that interact with Mac-1, failed to inhibit 125I-HK binding to neutrophils. We conclude that HK binds, via domains on its heavy chain, D3, and light chain, D5, to Mac-1 on the neutrophil surface, and HK occupies a site overlapping with fibrinogen and different from factor X.

The serum concentration of rat T1 kininogen increases 20-30-fold in response to acute inflammation. This increase, induced in the liver, is regulated primarily at the transcriptional level. In contrast, synthesis of a homologous K kininogen is not induced. In this study, we further analyzed a 321-base pair interleukin (IL)-6 response element in the T1 kininogen promoter and showed that it consists of at least three functionally distinct sequences (A, B, and C boxes). All three sequences were required for full promoter activity. The B box, a strong C/EBP-binding site, was crucial for T1 kininogen's basal expression, whereas A and C boxes resembled the type II IL-6 response elements and were critical for the cytokine response. C/EBP alpha, -beta, and -delta interacted with the B box sequence; however, upon IL-6 stimulation, C/EBP delta binding activity was dramatically induced and became the predominant factor binding to this site. Consistent with these binding studies were the cotransfection experiments, revealing that C/EBP delta was the most potent transactivator under induced conditions and that its transactivation on the T1 kininogen promoter required an intact B box. These findings substantiated the importance of the B box in eliciting the full acute-phase response. A sequence comparison showed the K kininogen promoter contained identical A and B boxes but differed from the T1 kininogen promoter by two nucleotides at the C box. This divergence reduced the IL-6 response by approximately 4-fold, thus contributing to the differential inflammatory response. Our studies demonstrate that evolutionary divergence of a few nucleotides at a critical sequence in the promoter regions can profoundly alter the expression patterns of downstream genes.

High and low molecular weight kininogens (HK and LK) modulate inflammatory responses, serve as precursors of kinins, inhibit cysteine proteinases, and modulate thrombin activation of platelets. Differential splicing yields two different mRNAs for HK and LK. We report the first molecular characterization of a kininogen gene mutation in a patient lacking LK and HK. No gross DNA deletion or insertion was detected by Southern blots, and LK and HK mRNAs were normal size on Northern blots. Exon sequences amplified by polymerase chain reaction showed a C-->T transition at nucleotide 587, resulting in a CGA (Arg)--> TGA (Stop) mutation in exon 5 before the splice site in exon 10, thus preventing synthesis of both HK and LK. The mutation eliminated the recognition site of the restriction enzyme Csp45I. Results of Csp45I digestion of the polymerase chain reaction-amplified exon 5 DNA fragment of the patient (lacking kininogens), three daughters (approximately 50% HK), and 1 granddaughter (normal HK) revealed that the patient was homozygous, while the three daughters were heterozygous, and the granddaughter was normal. We conclude that a single base mutation in the kininogen gene exon 5 was responsible for kininogen deficiency in the Williams family.

To clarify the mechanism of the secretion defect of high molecular weight kininogen (HK) and low molecular weight kininogen (LK) by the liver of Brown Norway (B/N) Katholiek rats causing plasma kininogen deficiency, we cloned cDNAs for HK from cDNA libraries of the livers of B/N Katholiek and B/N Kitasato rats. A point mutation of G to A at nucleotide 487 was found in the cDNA of B/N Katholiek rats by sequence analysis of the cDNAs (including the entire HK-coding region) obtained from both strains. Both B/N Katholiek and B/N Kitasato rat cDNA fragments were introduced into a eukaryotic vector, pRc/CMV, to construct their respective expression plasmid, which was used to transfect COS-1 cells. At 24 h of incubation, the culture medium of COS-1 cells transfected with the B/N Katholiek rat cDNA contained only 10% of the HK antigen that was found in COS-1 cells transfected with the B/N Kitasato rat cDNA. More HK antigen was retained in the former cells. Moreover, cells transfected with B/N Katholiek rat cDNA, in which the A at nucleotide 487 was artificially replaced by G, secreted a significant amount of HK into the medium. These results suggest that a point mutation of G to A at nucleotide 487, which causes a substitution of Ala163 to Thr in the heavy chain of HK and LK, is responsible for the defective secretion of HK and LK by the liver of B/N Katholiek rats.

The light chain (LC) of cleaved high molecular weight kininogen (HK) binds to anionic surfaces as well as the zymogens prekallikrein and factor XI and thus accelerates activation of the kallikrein-kinin, fibrinolytic, and coagulation pathways. The binding sites on HK LC for factor XI (amino acid residues 574-631) and prekallikrein (residues 583-613) have been localized to domain 6. Domain 5 (residues 438-520) has been postulated to contain the anionic surface binding subdomain. In order to define this subdomain we have expressed HK LC (residues Lys438-Ser644) as a fusion protein with glutathione-S-transferase (GST) in Escherichia coli and generated various HK LC deletion mutants. The recombinant HK LC (rHK LC) and various HK LC fragments were purified as GST fusion proteins by glutathione-Sepharose affinity chromatography from bacterial cell extracts. The rHK LC and recombinant fragments His459-Ser644, Glu466-Ser644, Leu483-Ser644, His493-Ser644, Lys438-Asp492, Lys438-Ser531, and His493-Lys520 inhibited 125I-HKa binding to kaolin, a model anionic surface used in the contact system, in a concentration-dependent manner. Deletion mutant proteins lacking domain 5, Thr521-Ser644 and Ser583-Ser644, did not inhibit the radiolabeled HKa binding to kaolin. The rHK LC and recombinant fragments Lys438-Asp492, Lys438-Ser531, His493-Ser644, His493-Lys520, Thr521-Ser644, and Ser583-Ser644 were radiolabeled with 125I and were then tested for their ability to bind to kaolin in the presence of fibrinogen and albumin. Except for the Thr521-Ser644 and Ser583-Ser644 fragments, all other radiolabeled HK LC deletion mutant proteins and rHK LC bound to kaolin in a concentration-dependent manner. This binding to kaolin was specific since it was inhibited by the addition of excess unlabeled HKa. The rHK LC, His493-Ser644 and delta 493-520 HK LC have coagulant activity, while other deletion mutant proteins did not exhibit coagulant properties. We conclude that there are at least two anionic surface binding subdomains, one in the histidine-glycine-rich region (Lys438-Asp492) and the other in the histidine-glycine-lysine-rich region (His493-Lys520), in the domain 5 of HK LC. Either subdomain, in the presence of the zymogen binding domain 6, is sufficient to impart coagulant activity to HK LC, while the presence of both did not increase the coagulant activity of HK LC additively.

Interleukin-6 is a pleiotropic cytokine that has a major role in the coordination of the hepatic acute phase response. In order to more fully understand this role, we have examined the interleukin-6 induction of T kininogen, a cysteine protease inhibitor and a major acute phase reactant in the rat. Using deletional analysis and site-directed mutagenesis of T kininogen-chloramphenicol acetyltransferase fusion constructs transfected into HepG2 hepatoma cells, we have identified two similar interleukin-6 response elements within 250 base pairs of the transcription start site. These two response elements are functionally interdependent. The sequences of these two elements match the consensus sequence for the previously described Type B interleukin-6 response element. Interleukin-6 signal transduction via two Type B elements has not been observed previously in vivo. A DNA fragment encompassing these response elements forms the same protein complex with nuclear extracts from both untreated and interleukin-6-treated cells.

The serum concentration of rat T1 kininogen increases 20-30-fold in response to acute inflammation, an induced hepatic synthesis regulated primarily at the transcriptional level. To analyze the cis-regulatory elements responsible for the induced transcription, we fused a 1.6-kilobase segment of the rat T1 kininogen promoter to a reporter gene, chloramphenicol acetyltransferase (CAT). The resultant chimeric DNA was transfected into cultured cells. In transient transfection assays, this 5'-flanking sequence was sufficient to confer cell-specific expression: CAT activity was readily detectable when the construct was transfected into liver-derived cells, but it was not detectable in nonliver cells. Furthermore, when liver cells (Hep3B) transfected with this construct were treated with conditioned medium prepared from activated mixed lymphocyte cultures or with recombinant interleukin-6 (IL-6), a 5-fold increase in CAT activity was detected. Addition of dexamethasone to the conditioned medium or to IL-6 showed synergistic effects and resulted in a 10-fold increase in CAT activity. In contrast, when IL-1 was used with IL-6, induction of CAT activity was inhibited. Deletion analyses revealed two regions important for tissue-specific and induced regulation of T1 kininogen: sequences proximal to base pair -73 conferred enhanced expression in liver-derived cells and a distal region that conferred responsiveness to conditioned medium, recombinant IL-6, and dexamethasone. This responsive element had properties of an inducible transcriptional enhancer, and it was functional in both liver and nonliver cells when placed immediately upstream of a thymidine kinase promoter.

The rat K and T kininogen genes show different modes of mRNA production. The K gene encodes two distinct mRNAs for high molecular weight (HMW) and low molecular weight (LMW) kininogens. These two mRNAs are generated by differential usage of the 3'-terminal exon (LMW exon) and the exon next to and upstream from the LMW exon (HMW exon) through alternative splicing and polyadenylation. In contrast, the T gene generates one mRNA by using selectively the LMW exon, although the T gene is extremely homologous to the K gene. In this study, we constructed a series of chimeric kininogen genes by not only exchanging equivalent restriction fragments of the two genes but also replacing nucleotides that differ between the two genes. We then examined the sequences and the mechanisms governing the different expression patterns of the two genes by transfecting the chimeric genes into heterologous COS cells. The results indicated that the different expression patterns of the K and T genes are governed by two separate internal sequences of the HMW and LMW exons. The internal HMW sequence contains a set of five repetitive sequences, and these repetitive sequences are highly complementary to the 5' portion of U1 snRNA. Furthermore, the nucleotide differences in the U1 snRNA-complementary sequences between the K and T genes have marked effects on the relative formation of the HMW and LMW mRNAs; this indicates that the repetitive sequences complementary to U1 snRNA play a crucial role in determining the relative expression of the two mRNAs. Based on these findings, we discuss a novel mechanism for alternative RNA processing, in which splicing efficiency is controlled by the interaction of U1 small nuclear ribonucleoproteins and the U1 snRNA-complementary repetitive sequences of the kininogen pre-mRNA.

It has been proposed that a cysteine proteinase inhibitor (CPI) found in the ascitic fluid of Sarcoma 180 tumor-bearing mice is a kind of kininogen (Itoh, N., Yokota, S., Takagishi, U., Hatta, A., and Okamaoto, H. (1987) Cancer Res. 47, 5560-5565). The first 40 NH2-terminal residues and 54 residues of the COOH-terminal sequence, including the bradykinin moiety of highly purified ascites CPI, were determined and compared with those of mammalian low molecular weight kininogens (LMWK). The significant identity between these amino acid sequences with those of other mammalian LMWKs suggests that ascites CPI corresponds precisely to mouse LMWK. This kininogen has a light chain composed of 43 amino acid residues, which contains a unique Met-Ala-Arg-bradykinin sequence. Hydroxyproline, which was recently identified in the bradykinin sequence of kininogen from the ascitic fluid of a cancer patient, was not found in the kinin moiety of this mouse kininogen. Among purified glandular kallikreins from human, hog, rat, and mouse, only mouse submaxillary gland kallikrein was able to release bradykinin from this kininogen. Kinetic studies using a newly synthesized fluorogenic substrate, N-t-butoxycarbonyl-Met-Ala-Arg-MCA, revealed that mouse kallikrein hydrolyzes this substrate approximately 80-fold faster than does hog kallikrein, suggesting that the unique Met-Ala-Arg-bradykinin sequence is responsible for the varied susceptibility of mouse kininogen to different kallikreins.

Two major acute phase alpha 1-protein (alpha 1-MAP, also called thiostatin) genes, one kininogen gene and one structurally related pseudogene, were isolated from a Buffalo rat genomic library and characterized. Comparison of the nucleotide sequence in the 5' proximal flanking region (1 kilobase) of a strongly inducible alpha 1-MAP gene with that in the non-inducible rat kininogen gene showed an overall homology of 90%, with a small number of randomly distributed base changes. In the intron downstream of the exon coding for Ile-Ser-bradykinin, the two alpha 1-MAP genes contained sequence sections similar to the section in the human kininogen gene which codes for the carboxyl-terminal chain of high molecular weight kininogen. Various features of the DNA related to gene expression such as initiation and termination sites of transcription, receptor binding sites, a Z-DNA section, and exon/intron splicing sites were identified. mRNAs expressed by alpha 1-MAP and kininogen genes in liver were studied by RNA blot hybridization using specific oligonucleotide probes. The kininogen gene expressed two mRNA species, one coding for high molecular weight kininogen, the other coding for low molecular weight kininogen. The levels of the two kininogen mRNAs in liver did not increase during acute inflammation. Only one type of mRNA, similar in size to low molecular weight kininogen RNA, was expressed by each of the two alpha 1-MAP genes. The levels of both alpha 1-MAP mRNAs increased strongly during acute inflammation.

Responses of the rat liver prekininogen mRNAs after induction of acute inflammation were examined by blot-hybridization and S1 nuclease protection analyses with the aid of cDNA probes specific for rat kininogens. Marked changes in the relative levels of the low molecular weight (LMW) prekininogen mRNAs were observed after administration of Escherichia coli lipopolysaccharide, and the mRNA levels increased with a half-maximal dose of approximately 100 ng of lipopolysaccharide/100 g body weight. At maximum level of induction, the LMW prekininogen mRNAs comprised about 1% of total liver mRNA, thus representing a major component of the liver mRNA in the acutely inflamed rat. Differences in the inflammatory responses of various forms of the prekininogen mRNAs were then investigated by S1 nuclease protection analysis with the use of three different cDNA probes, each specific for either K-prekininogen or two types of T-prekininogens. Both of the T-prekininogen mRNAs increased progressively during the first 24 h after induction of inflammation, and at maximum level of induction, these two mRNAs increased about 10- and 13-fold over their normal level. In contrast, neither of the high molecular weight and LMW K-prekininogen mRNAs exhibited such an increase after induction of inflammation. Thus, the expressions of the rat T- and K-prekininogen mRNAs are differentially regulated in response to the induction of acute inflammation.

The rat major acute phase protein (alpha 1-MAP) is a cysteine protease inhibitor. The stoichiometry of the interaction between the inhibitor and enzyme was shown to be 1:2. A cDNA clone specific for rat alpha 1-MAP was isolated from a cDNA library prepared from an inflamed rat liver RNA template. The 1458-base pair insert was sequenced and positively identified by alignment with a partial amino acid sequence obtained by radiosequence analysis of the primary translation product for alpha 1-MAP. Complete sequence analysis determined the alpha 1-MAP cDNA coded for the entire protein with the exception of the first four amino acids of the signal peptide, all of which were identified by radiosequencing. The coding sequence spans 1282 nucleotides, followed by 115 base pairs of a 3' untranslated region. Two putative active sites, suggested by the enzyme-inhibitor ratio, have been identified by analysis of internal duplications of the alpha 1-MAP sequence and the alignment of these regions with the sequences of several low molecular weight cysteine protease inhibitors. A computer homology analysis of the protein sequence revealed a 59.3% overall identity between rat alpha 1-MAP and bovine low molecular weight (LMW) kininogen. The homology included the signal peptide regions. LMW kininogen is a precursor of bradykinin. alpha 1-MAP does contain a bradykinin sequence; the flanking amino acids are different, however. Evidence for the expression of the LMW and a high molecular weight kininogen from the same gene, and the high degree of homology between these proteins and the rat acute phase protein suggest that all three proteins belong to a precisely regulated gene family.

The existence of two types of circulating bovine plasma high molecular weight kininogen (HMWK) was predicted from analyses of complementary DNAs coding for this protein (Kitamura, N., Takagaki, Y., Furuto, S., Tanaka, T., Nawa, H., and Nakanishi, S. (1983) Nature 305, 545-549). The present protein-based study provided evidence in support of the proposed amino acid sequence derived from analysis of the cDNA clone, and the results confirm the existence of two types of circulating HMWK. Type I HMWK contains a heavy chain composed of 361 residues, while the heavy chain of type II HMWK contains 359 residues. The amino acid sequences of type I and type II HMWK determined in this study were identical to that inferred from the cDNA sequence with the exception of microheterogeneity observed in the cDNA at position 87 (Glu/Gln) and 168 (Lys/Arg). The heavy chain of type I HMWK contains 4 asparagine-linked carbohydrate chains at Asn-69, -150 (or -151), -179, and -186, while the heavy chain of type II HMWK contains these and an additional carbohydrate chain at Asn-264. In addition, a carbohydrate chain was found to be O-glycosidically linked to Thr-118 in both chains. Among nine disulfide linkages found in HMWK, eight intrachain disulfide pairs were established in the heavy chain. One interchain disulfide bridge occurs between the heavy chain and the light chain. This disulfide pairing, as well as repeating amino acid sequences observed in the heavy chain, provides strong evidence for the existence of three homologous domains in the heavy chain of bovine HMWK.

Three types of cloned cDNA sequences for rat low molecular weight prekininogens were isolated and determined by molecular cloning and sequence analysis. The deduced amino acid sequences indicated that one, termed K-prekininogen, represents the counterpart of the known low molecular weight prekininogen present in other mammals, while the other two, called T-prekininogens, contain a novel T-kinin sequence which was recently identified from rat plasma. Although T- and K-prekininogens are highly homologous with each other, both of the T-prekininogens contain methionine, instead of arginine or lysine, as an amino acid preceding T-kinin and exhibit two consecutive amino acid deletions in the preceding region of T-kinin as compared with K-prekininogen. The former finding accounts for the previous observation of strong resistance of T-kininogens to cleavage with trypsin or kallikreins, while the latter finding has been explained by the structural analysis of genomic clones in which T-kinin-coding exon is contracted at its intron junction. A partial nucleotide sequence reported recently for the rat major acute phase protein (alpha 1-MAP) mRNA was found to be extremely related to the corresponding portion of the rat T-prekininogen mRNA. Furthermore, consistent with the previous report of the structural identity of major acute phase protein and alpha 1-cysteine proteinase inhibitor, kininogen closely resembles not only the former but also the latter in the amino acid compositions. The interrelationship among the triad of these proteins has been discussed.

The entire human kininogen gene has been isolated as a set of overlapping genomic DNA fragments, and the 11 exons encompassing approximately 27 kilobase pairs have been mapped by restriction enzyme analysis and nucleotide sequence determination. The nine 5'-terminal exons encode the 5'-untranslated region and the protein-coding region for the signal peptide and the heavy chain, which are common for high molecular weight (HMW) and low molecular weight (LMW) prekininogen mRNAs. Exon 10 consists of the common sequence for bradykinin and the immediately following unique sequence for HMW prekininogen mRNA. Exon 11 is then located following a 90-nucleotide sequence downstream from exon 10 and precisely specifies the sequence unique to LMW prekininogen mRNA. This, together with the hybridization analysis of total human cellular DNA, leads us to conclude that human HMW and LMW prekininogen mRNAs are produced from a single gene as a consequence of alternative RNA processing events. The structural analysis of the kininogen gene also shows that each of the nine 5'-terminal exons discretely specifies the nine protein domains observed in the amino-terminal portion of the kininogens. Furthermore, these nine genetic domains can be characterized by a thrice repeated pattern of three genetic segments, and two sets of these three domains, encompassing exons 3-5 and exons 6-8, are most closely related to each other. Therefore, we have proposed two successive duplication mechanisms as a model for the generation of the structure of the kininogen gene.

Two types of rat kininogen mRNAs exhibit marked differences in the regulation of their expressions. The two T kininogen (T) mRNAs considerably increase under acute inflammation conditions, although no such increase occurs in the high molecular weight and low molecular weight K kininogen (K) mRNAs encoded by the same gene. This investigation examines the sequences of the 5' portions of the K and the two T genes and analyzes transcription initiation sites and their utilization at differently regulated conditions of the K and T genes. The sequence analysis indicates that the K and the two T genes are extremely homologous at the 5' portions as well as the 5'-flanking regions of at least 1.0 kilobase pairs. The S1 nuclease and primer extension analyses show that both T and K mRNAs start with three sets of the homologous initiation sites, and the utilization of the two 3'-side initiation sites of the T genes markedly increases under inflammation conditions. In contrast, no such elevation is observed for the three initiation sites of the K gene. Thus, although the sequences involved as transcription initiation sites are extremely homologous, their utilizations in the expressions of the K and T genes are regulated differently under inflammation conditions.

The present investigation using molecular cloning and sequence analysis concerns the examination of the molecular basis for different expression patterns of two types of the rat kininogen genes. We show that the low molecular weight and high molecular weight forms of K kininogens are produced from a single gene through alternative usage of two 3'-coding regions, whereas only the low molecular weight forms of T kininogens are generated as a result of several mutational changes in the high molecular weight-specifying regions of both T-I and T-II kininogen genes. The mutational changes include a nucleotide substitution at the polyadenylation/processing signal site, nucleotide deletions resulting in the frame-shift mutation, and an insertion of the type 2 Alu-equivalent sequence. Because kininogens represent a multifunctional protein comprising the proteinase-inhibitory activity, the kinin moiety, and the clotting activity, these results present evidence indicating the molecular basis for the disappearance of a part of the gene functions. We also show that the K and T kininogen genes as well as the two T kininogen genes are extremely homologous, excluding and including the above mutational changes, respectively. These structural relationships allow us to envisage evolutionary processes for the generation of the rat kininogen gene family, particularly for the disappearance of a part of the gene functions.

The two types of the rat kininogen genes show different modes of mRNA production. The K gene encodes two distinct mRNAs for high molecular weight (HMW) and low molecular weight (LMW) kininogens. These two mRNAs are generated by differential usage of the 3'-terminal exon (LMW exon) and the one next to this exon (HMW exon) through alternative polyadenylation and splicing. In contrast, the two T genes selectively generate the LMW form of the mRNA, although the T genes are extremely homologous to the K gene, including the sequence (psi HMW region) corresponding to the HMW exon of the K gene. In this study, we constructed a series of chimeric kininogen genes by exchanging equivalent restriction fragments of the K and T genes and examined the sequences and the mechanisms governing the different expression patterns of the kininogen genes by introducing the chimeric genes into heterologous COS cells. The results indicate that the formation of the two forms of the mRNA is controlled by two separate 3' sequences of the kininogen genes. One is located within the internal sequence of the HMW/psi HMW region, whereas the other is within the LMW exon and its preceding region. Our data also suggest that the different expression patterns of the kininogen genes are primarily governed by differing splicing efficiency.

High molecular weight kininogen (HMWK) functions as a cofactor for activation of plasma serine zymogens and as an inhibitor of tissue cysteine proteases. Cell surfaces to which HMWK binds may provide sites for regulation of these systems. Localization of these HMWK-dependent processes at sites of vascular injury may depend on its binding to specific receptors on endothelial cells. In culture, passaged human umbilical vein endothelial cells (HUVEC) bind anti-HMWK antibody to the cell surface and contain 171 +/- 75 ng of HMWK/10(8) cells. [35S]Methionine-labeled HUVEC in culture synthesize a 120-kDa protein immunoisolated using an anti-kininogen antibody, and a 3500-nucleotide message for human HMWK was detected by Northern blot in RNA extracted from HUVEC. HUVEC also express unoccupied binding sites for HMWK on their surface. 125I-HMWK specifically binds to HUVEC in a reaction requiring Zn2+. 125I-HMWK binding to HUVEC is saturable at 4 degrees C but not at 23 degrees C. 125I-HMWK binds to HUVEC with equal affinity as unlabeled HMWK. Kallikrein, factor XII, fibrinogen, fibronectin, and thrombin do not inhibit 125I-HMWK binding to HUVEC. 125I-HMWK-HUVEC binding remains fully reversible at 60 min following the addition of a 50-fold molar excess HMWK. HUVEC express 9.3 +/- 2.0 X 10(5) (mean +/- S.E.) HMWK binding sites/cell (Kd = 52 +/- 13 nM). Both added and cell-bound 125I-HMWK migrate at 120 kDa on sodium dodecyl sulfate gel electrophoresis, suggesting that the protein remains uncleaved upon binding to the HUVEC surface. These studies indicate that HUVEC synthesize HMWK and the HUVEC surface has a site for its expression. By synthesizing and localizing HMWK to the cell surface, endothelial cells may contribute to the activation of plasma's contact serine zymogens and regulation of tissue cysteine proteases.

cDNA sequences for both human high molecular weight (HMW) and low molecular weight (LMW) prekininogens have been isolated by molecular cloning and determined by sequence analysis. The sequence determination together with the S1 nuclease mapping and RNA blot-hybridization analyses indicate that human HMW and LMW prekininogen mRNAs share an identical sequence throughout the 5'-untranslated region and the protein-coding region up to the sequence encoding the 12 amino acids distal to the bradykinin sequence, and the two mRNAs then completely diverge from each other. The signal peptide, the heavy chain (H chain), and the bradykinin moiety, which are common between the two prekininogens, consist of 18, 362, and 9 amino acids, respectively, while the light chains (L chains) of the HMW and LMW prekininogens are composed of 255 and 38 amino acids, respectively. All 17 cysteine residues present in the human and bovine H chains are located at exactly equivalent positions, indicating that the human H chain, like the bovine counterpart, can form 8 loop structures, each connected by two adjacent cysteine residues. The L chains of human and bovine kininogens differ in the protein lengths as well as in some amino acids crucial for the processing of the kininogens by kallikrein. Based upon this finding, we have discussed the molecular basis for the different modes of processing of human and bovine HMW kininogens and for the different kinetics of contact activation reactions exhibited by the two HMW kininogens.

The complete disulfide loop structure of human alpha 2-HS glycoprotein has been elucidated. alpha 2-HS glycoprotein isolated from human plasma was found to be a two-chain protein composed of a heavy and a light chain. The heavy chain comprises the A-chain of alpha 2-HS glycoprotein (Yoshioka, Y., Gejyo, F., Marti, T., Rickli, E. E., Bürgi, W., Offner, G. D., Troxler, R. F., and Schmid, K. (1986) J. Biol. Chem. 261, 1665-1676) and part of the connecting peptide which has been predicted from the corresponding cDNA sequence (Lee, C. C., Bowman, B. H., and Yang, F. (1987) Proc. Natl. Acad. Sci. U.S.A. 84, 4403-4407), whereas the light chain corresponds to the beta-chain of alpha 2-HS glycoprotein (Gejyo, F., Chang, J. L., Bürgi, W., Schmid, K., Offner, G. D., Troxler, R. F., Van Halbeek, H., Dorland, L., Gerwig, G. J., Vliegenthart, J. F. G. (1983) J. Biol. Chem. 258, 4966-4971). Twelve half-cystine residues are present in the alpha 2-HS glycoprotein molecule, and 11 of them are positioned in the heavy chain and a single one in the light chain of the molecule; they form six disulfide bridges. The first and the last half-cystine residues of the amino acid sequence of alpha 2-HS glycoprotein are engaged in the formation of a loop spanning the extreme NH2- and COOH-terminal portions of the molecule, thereby connecting the heavy and light chains. The other 10 half-cystines residues are linked consecutively in the heavy chain and form five loops which span 4-19 amino acid residues. Among them are two pairs of loops which are characterized by mutual sequence homology. The particular arrangement of disulfide loops in alpha 2-HS glycoprotein is similar to the patterns of linearly arranged and tandemly repeated disulfide loops of cysteine proteinase inhibitors, i.e. the cystatins and the kininogens. It is concluded that alpha 2-HS glycoprotein represents a structural prototype of a novel family among the cystatin superfamily, characterized by the presence of two cystatin-like building blocks. Extensive similarity among the NH2-terminal sequences of alpha 2-HS glycoprotein and human histidine-rich glycoprotein suggest that the latter protein is another candidate protein of this new family.