Tissue Kallikrein Actions at the Rabbit Natural or Recombinant Kinin B2 Receptors
We have examined whether exogenous human tissue kallikrein exerts pharmacological actions via the bradykinin B2 receptor; specifically, whether the protease can bind to, cleave, internalize, and/or activate a fusion protein composed of the rabbit B2 receptor conjugated to the green fluorescent protein (B2R-GFP). The enzyme partially digested the fusion protein at 1 μmol/L, but not 100 nmol/L, and promoted B2R-GFP endocytosis in HEK 293 cells (≥50 nmol/L). Trypsin and endoproteinase Lys-C, but not plasma kallikrein, also cleaved B2R-GFP. Phospholipase A2 was activated by 50 nmol/L tissue kallikrein in HEK 293 cells expressing B2R-GFP, and this was mediated by the receptor, as shown by the effect of a B2 receptor antagonist and by the lack of response in untransfected cells. However, 500 nmol/L kallikrein elicited a strong receptor-independent activation of phospholipase A2. Tissue kallikrein competed for [3H]bradykinin binding to B2R-GFP only at 1 μmol/L. A simulation involving kallikrein treatment of HEK 293 cells, pretreated or not with human plasma, evidenced the formation of immunoreactive bradykinin. The enzyme (50 nmol/L) contracted the rabbit isolated jugular vein via its endogenous B2 receptors, but the effect was tachyphylactic, and there was no cross-desensitization with bradykinin effects. Aprotinin prevented all pharmacological responses to tissue kallikrein, indicating that the enzyme activity is required for its effect. The local generation of kinins is a plausible mechanism for the pharmacological effects of lower concentrations of tissue kallikrein (50 to 100 nmol/L); higher levels (0.5 to 1 μmol/L) can not only initiate the degradation of rabbit B2 receptors but also exert nonreceptor-mediated effects.
Kallikreins are a heterogeneous group of serine proteases capable of releasing bradykinin (BK)-related peptides (kinins) from kininogens.1 The classic human tissue kallikrein KLK1 (hK1), notably present in renal tissue and released into urine, has been used as a pharmacological agent in animals, lately under the form of DNA expression vectors used in gene therapy strategies or transgenic animals.2 The mechanism of the observed effects (hypotension, cardioprotection, angiogenesis, etc) frequently involved the BK B2 receptor (B2R), as evidenced by the preventive effect of a peptide B2R antagonist, icatibant. In some models involving direct tissue injury, the kinin B1 receptor (selective for a class of kinin metabolites) was also involved in the effect of human tissue kallikrein.3
The mode of action of kallikrein at the cell level is not entirely clear. The proteinase-activated receptors (PARs) are not believed to possess other ligands than proteases, but tissue kallikrein does not activate PAR-1 or PAR-2.4 The human B2R has been proposed to be a specific binding site for serine proteases like human or porcine tissue kallikrein and trypsin, with pharmacological activation of these receptors on binding.5 Moreover, extracellular proteases can initiate the destruction of G protein–coupled receptors (GPCRs), as shown by the effects of chymotrypsin on the receptor for anaphylatoxin C5a6 or of trypsin on the BK B2R.7
We have previously shown the feasibility of using a construct composed of the rabbit B2R fused to the green fluorescent protein (GFP) to identify treatments that lead to B2R degradation.8 The B2R-GFP fusion protein retains the pharmacological profile, the functional properties, and the nanomolar affinity for [3H]BK, virtually indistinguishable from the wild-type B2R.9 These investigations failed to support the concept that BK-induced endocytosis is a mechanism for B2R downregulation, as the receptors were restricted to the recycling endosome compartment, did not enter lysosomes, and were eventually completely recycled to the cell surface. However, B2R-GFP exposure to extracellular trypsin rapidly (10 minutes) produced a small immunoreactive protein similar to GFP, suggesting that limited proteolysis of B2R was followed by rapid degradation by mechanisms endogenous to the cell.8
We have examined whether human tissue kallikrein exerts pharmacological actions via the B2R; specifically, whether the protease can bind to, cleave, internalize, and/or activate the B2R-GFP. The physiological consequences of kallikrein interaction with the rabbit wild-type B2R have been verified by using the contractility of the rabbit jugular vein.
Purified human plasma kallikrein and tissue (urinary) kallikrein were purchased from Calbiochem. Sodium dodecyl sulfate–polyacrylamide gel electrophoresis of tissue kallikrein, followed by silver nitrate staining, revealed a single protein band. Sequencing-grade proteases were from Sigma (bovine trypsin) or Roche (endoproteinase Lys-C from Lysobacter enzymogenes). BK was purchased from Bachem. LF 16.0687, a competitive antagonist of BK at the rabbit B2R,9 was a gift from Laboratoires Fournier (Daix, France). When aprotinin (Sigma) was used as an inhibitor of the pharmacological actions of tissue kallikrein (50 to 500 nmol/L), its final concentration at the cell level was 10 μmol/L, but the enzyme and aprotinin had been preincubated together (50-fold more-concentrated stock) for 1 hour at 37°C before application (the kallikrein stock was also preincubated at 37°C in control experiments).
The derivation of an HEK 293 cell line stably expressing B2R-GFP and its properties are described elsewhere.8,9 Nontransfected HEK 293 cells were used in control experiments. Confocal microscopy was applied to the cells as described.8,9 An arachidonic acid release assay was performed as described elsewhere to evaluate the activation of phospholipase A2 (PLA2) after stimulation with tissue kallikrein in untransfected HEK 293 cells or cells stably expressing B2R-GFP (24-well plates).9
Protease Digestion of Surface Molecules in Intact Cells
The protease or BK treatments were applied to confluent 25-cm2 flasks of HEK 293 cells stably expressing B2R-GFP or of untransfected cells. The culture medium was removed from the cell flasks, which were rinsed and filled with 500 μL serum-free α-minimal essential medium containing 71 μmol/L cycloheximide to avoid interference from newly synthesized receptors. The cell flasks were further incubated for 30 minutes at 37°C before extraction. The protease or BK treatments were applied for the last 10 minutes of the incubation period.
Relative to other anti-GFP antibodies previously used by us for immunoblots,8,9 the monoclonal antibodies to GFP (Clontech; clone JL-8, used at dilution 1/1000) exhibited an exceptionally low background in total cell extracts of untransfected HEK 293 cells. For the analysis of B2R-GFP in total cell extracts, immunoblots were generally performed as described previously.8 Immunoblots of low- and high-molecular-weight kininogens were also conducted on HEK 293 cells exposed or not to citrated human plasma; the procedure outlined above was applied by using characterized polyclonal antibodies raised against human high-molecular-weight kininogen10 (dilution 1/25 000) and the appropriate secondary antibodies.
The plasma membranes from HEK 293 cells stably expressing B2R-GFP were recovered after cell fractionation by using a sucrose/tricine buffer system as described elsewhere; the pellet of the third centrifugation was used as the source of material (150 000g, 3 hours).9 The binding of 2 nmol/L [3H]BK (Perkin Elmer Life Sciences; 90 Ci/mmol) to these membranes (10 μg/tube, suspended in 500 μL phosphate-buffered saline [PBS], pH 7.4, supplemented with 0.02% NaN3, 0.1% bovine serum albumin, 1 mmol/L phenylmethylsulfonyl fluoride, and 1 μmol/L captopril) was established in the presence or absence of a cold competitor (tissue kallikrein, BK, or the antagonist LF 16.0687). All tubes were incubated on ice for 90 minutes, and then the [3H]BK-membrane complexes were adsorbed to glass fiber filters (GF/C, Whatman, presoaked in polyethylenimine, 0.5% in PBS, 2 hours) by using a 24-channel cell harvester (Brandell Corp). The filters were washed 5 times with cold PBS and were subsequently analyzed for bound radioactivity by scintillation counting.
Enzyme Immunoassay of Kinins in the Culture Medium of HEK 293 Cells
To test whether tissue kallikrein could release BK-like peptides from B2R-GFP–expressing or untransfected HEK 293 cells, confluent 75-cm2 flasks were rinsed with Earle’s balanced salt solution containing 1 μmol/L orthophenanthroline, filled with 5 mL undiluted human citrated plasma for 30 minutes, and then rinsed and incubated in Earle’s solution (5 mL) for 30 minutes. Then, tissue kallikrein (50 nmol/L) was added to some of the flasks. After a further 10 minutes’ incubation at 37°C, 3 mL of the medium was removed and transferred in 15 mL of ice-cold ethanol. A variation of the protocol aimed at measuring very low amounts of kinins released by kallikrein in cells not treated with plasma in larger flasks (175-cm2, 10 mL Earle’s solution filling, of which 6 were transferred in 30 mL ethanol). The suspensions were incubated on ice for 1 hour, centrifuged to remove the precipitated proteins, and stored at −80°C until assayed. For that purpose, the ethanol extract was evaporated to dryness in a Speed vac system and then processed precisely as described for the separate measurements of immunoreactive BK and des-Arg9-BK.11,12 The results are expressed as the estimated kinin concentration in the initial medium.
A local ethics committee approved the procedures based on rabbits. Rabbit jugular vein contractile responses to tissue kallikrein, BK (mediated by B2Rs), or histamine (mediated by H1 receptors) were measured as reported,9 except for captopril in the Krebs’ buffer, which was omitted in the present experiments.
Assay for B2R Cleavage Based on Detection of the B2R-GFP Immunoreactive C-Terminal Fragments
When untransfected HEK 293 cell extracts were immunoblotted with the anti-GFP monoclonal antibodies, essentially no background was observed (Figure 1, lane 2). Transfection with a GFP-coding vector produced the expected strong immunoreactive band at ≈27 kDa (Figure 1, lane 1). As previously reported with different anti-GFP antibodies,8,9 B2R-GFP–specific immunoreactivity was expressed as a wide band of 101 to 105 kDa; faint additional bands suggested that spontaneous degradation of the fusion protein occurs in the cells, with a protein resembling GFP as 1 of the main immunoreactive metabolites (Figure 1, lane 3). Short trypsin digestion (10 minutes, 1 μmol/L) of untransfected cells did not reveal immunoreactive bands (data not shown) but produced ≈31- and 27-kDa C-terminal immunoreactive fragments in cells expressing B2R-GFP (Figure 1, lanes 5 or 10). Treatment of transfected cells with an alternate serine protease, endoproteinase Lys-C (10 minutes, 0.3 μmol/L), produced somewhat different results, with the reinforcement of an ≈75-kDa band that existed in transfected cells not exposed to enzymes, and the appearance of lower-molecular-weight bands (Figure 1, lane 6). Plasma kallikrein (10 nmol/L to 1 μmol/L, 10 minutes) failed to digest B2R-GFP (Figure 1, lanes 7, 11, and 12), whereas tissue kallikrein (1 μmol/L, 10 minutes) produced a third pattern of digestion, with a 28-kDa major product (Figure 1, lane 8). However, tissue kallikrein failed to digest B2R-GFP at lower concentrations (10 or 100 nmol/L, Figure 1, lanes 13 and 14). All observed digestions were partial. B2R-GFP fragmentation does not occur when cells are stimulated with BK (100 nmol/L, 10 minutes, Figure 1, lane 4), suggesting that receptor degradation is not secondary to B2R signaling. The B2R antagonist LF 16.0687 (10 μmol/L) did not inhibit the digestion of B2R-GFP by trypsin, endoproteinase Lys-C, or tissue kallikrein (data not shown).
Subcellular Redistribution of Fluorescence
Ten-minute treatments with tissue kallikrein promoted the internalization of B2R-GFP membrane-associated fluorescence in cells stably expressing the fusion protein (Figure 2). The effect of kallikrein was concentration dependent (50 and 500 nmol/L tested). The internalized fluorescence was either finely granular or concentrated into ill-defined structures, similar to what is observed after BK treatment (Figure 2).9 The effects of kallikrein on fluorescence distribution were at least partially inhibited by the protease inhibitor aprotinin (final concentration, 10 μmol/L) or the receptor antagonist LF 16.0687 (the latter drug being active only against the lower concentration level of the enzyme).
Functional Response to Tissue Kallikrein in Cells
As the action of tissue kallikrein on B2R has been proposed,5 we have investigated the functional effect of the enzyme on B2R-GFP by using a PLA2 assay (Figure 3). Both BK (1 nmol/L) and tissue kallikrein (50 nmol/L) significantly increased arachidonate release from cells expressing B2R-GFP (Figure 3A) but not from untransfected cells (Figure 3B). Further insight into the role of the receptor in kallikrein action is provided by combining treatments with the competitive nonpeptide B2R antagonist LF 16.0687: a concentration (1 μmol/L) that is sufficient to significantly reduce the effect of BK (1 nmol/L) also significantly decreased the tissue kallikrein effect (50 nmol/L; Figure 3A). This enzyme level is in a concentration range that does not promote detectable digestion of the fusion protein as assessed by immunoblotting (Figure 1). The 500 nmol/L concentration is closer to the level where digestion is observed (Figure 1). At this concentration, tissue kallikrein profusely released arachidonate from cells expressing B2R-GFP or untransfected cells (Figures 3A and 3B). LF 16.0687 did not prevent the effect of the 500 nmol/L enzyme concentration in the first type of cells. Aprotinin inhibited arachidonate release induced by tissue kallikrein (50 nmol/L) in cells expressing B2R-GFP (Figure 3C), indicating that the enzyme activity is required for its action. Aprotinin did not inhibit BK-induced PLA2 activation.
Competition of [3H]BK Binding to B2R-GFP
As tissue kallikrein apparently stimulates the B2R (Figure 3A) and as the enzyme is reported to bind with high affinity to this receptor,5 we tested whether it could displace the specific binding of [3H]BK (2 nmol/L) to the membranes of HEK 293 cells expressing B2R-GFP (Figure 4). No competition was seen for the 1 to 100 nmol/L concentration range of the protease, but some competition was seen at 1 μmol/L. By contrast, BK and the antagonist LF 16.0687 were effective competitors of the radioligand binding.
Immunoreactive BK Release From HEK 293 Cells Exposed to Tissue Kallikrein
An alternative explanation for the stimulation of B2R-GFP by tissue kallikrein is the formation of kinins in the vicinity of the receptors from kininogen produced or taken up by the cells. To test this possibility, immunoreactive BK and des-Arg9-BK (1 of the BK metabolites) were measured in the supernatant of cell flasks exposed or not to tissue kallikrein (50 nmol/L, Table). Values close to the detection limits were observed in flasks without cells but containing both the Earle’s solution and tissue kallikrein. Confluent HEK 293 cells (75-cm2 flasks) expressing B2R-GFP first exposed to undiluted human plasma and then washed and treated with tissue kallikrein released large quantities of immunoreactive BK and des-Arg9-BK into the supernatant medium (Table), supporting the notion that kinin generation in the system is dependent on the presence of kininogen. Even higher concentrations were recorded in untransfected cells pretreated with plasma and then treated with kallikrein, suggesting that the receptor presence reduces the diffusion of the newly formed kinins into the supernatant. The conversion of immunoreactive BK into the metabolite des-Arg9-BK was ineffective. Further use of the immunoassays in 175-cm2 flasks not pretreated with human plasma (allowing sample concentration and a higher sensitivity) showed that HEK 293 cells release small quantities of immunoreactive BK under the effect of tissue kallikrein.
To validate the procedure wherein HEK 293 cells are exposed to human plasma to load them with kininogens, we extracted proteins from cells exposed or not to plasma and ran immunoblots for both the low- and high-molecular-weight kininogens (60 and 110 kDa, respectively, Figure 5). The procedure was highly effective, whether the cells expressed B2R-GFP or not. The lower-molecular-weight band was further identified as a kininogen by its reaction with anti-BK antibodies (same antibodies used as in the enzyme immunoassay applied to immunoblotting; data not shown). The assay was not sensitive enough to detect kininogen presence on cells not exposed to plasma.
Effect of Tissue Kallikrein on Rabbit Jugular Vein Contractility
These experiments were performed to verify that a submicromolar concentration of tissue kallikrein could stimulate wild-type rabbit B2R expressed at a physiological level. Tissue kallikrein (50 nmol/L) rapidly induced a sizeable contraction in the isolated rabbit jugular vein, but the tissue was completely desensitized when a second application of the protease was done at 30-minute interval (Figure 6A, top tracing). The tissues desensitized to kallikrein were not desensitized to BK (10 nmol/L, Figure 6A, top tracing); in fact, the agonist BK applied at 30-minute intervals did not desensitize the preparation to itself or to kallikrein (Figure 6A, bottom tracing). Pretreatment of naive tissues with the B2R antagonist LF 16.0687 (1 μmol/L) prevented the contractile response to tissue kallikrein (50 nmol/L, in tissues exposed for the first time to the protease) and, in fact, allowed a very small relaxant response to kallikrein in most pretreated tissues (Figure 6B). LF 16.0687 also reduced the response to BK (10 nmol/L) subsequently recorded but did not influence the contractile effect of histamine (100 μmol/L; Figure 6B). Similarly, aprotinin selectively prevented the contractile effect of tissue kallikrein (Figure 6C).
Short treatments with sequencing-grade extracellular trypsin or endoproteinase Lys-C should result in the cleavage of the rabbit B2R sequence after Arg (trypsin) or Lys (both enzymes) residues. The number of these residues is small (4) in the extracellular domains of the receptor,13 and they are not well conserved in sequences from other mammalian species. None of the predicted primary C-terminal products (≥38.5 kDa) can be identified with certainty from the digested cells (Figure 1). Tissue kallikrein cleavage site(s) are less predictable from the amino acid sequence,14 but subtle differences in the digestion patterns suggest that the 3 tested enzymes trigger sequential hydrolysis events that are partly different between them. It also seems that B2R-GFP exhibits a certain pattern of spontaneous degradation in HEK 293 cells, with the presence of several immunoreactive faint bands, and that limited proteolysis by extracellular proteases accelerates this process, with GFP-sized metabolites as major final products. GFP, a stable and compact globulin, has a long half-life in mammalian cells,15 and its accumulation is a sensitive indicator of partial B2R-GFP degradation (as previously observed with the endothelin ETB receptor-GFP conjugate).16 The observed digestions are far from complete (Figure 1), but we recently observed the complete disappearance of B2R-GFP replaced by a strong GFP product band in HEK 293 cells treated with enzymes secreted from human neutrophils.17 The receptor topography implies that the observed GFP-like products (≤31 kDa) are cytosolic if they include all of the GFP sequence (27 kDa); therefore, final GFP cleavage from the denatured receptor should have occurred well downstream of the transmembrane domain 7-cytosol boundary. The cellular mechanisms that complete the limited proteolysis initiated by extracellular proteases may include endocytosis, as evidenced by confocal microscopy. Membrane or vesicular structures labeled with fluorescence in the microphotographs should not be interpreted as free GFP, as the latter protein is diluted into the total cellular water.
A submicromolar level of tissue kallikrein (50 nmol/L) stimulates the recombinant B2R-GFP (PLA2 assay, Figure 3A) or the wild-type B2R contained in the rabbit jugular vein (contractility, Figure 5) in a manner sensitive to both the selective antagonist LF 16.0687 and the protease inhibitor aprotinin. Furthermore, such a low concentration of kallikrein does not stimulate HEK 293 cells devoid of B2Rs (Figure 3B). Hecquet et al5 suggested that tissue kallikrein exerts its effect on B2R after high-affinity binding and without the need for a catalytically active enzyme. At variance with them, we did not observe competition of [3H]BK binding to rabbit B2R-GFP by kallikrein at submicromolar concentrations. Such binding may occur at a site not conserved in the rabbit sequence. Limited proteolysis of the resting B2R may release a structural constraint of the molecule and mimic a conformation similar to that produced by agonist stimulation. However, receptor hydrolysis is not evident in cells expressing high levels of B2R-GFP and exposed to tissue kallikrein at 100 nmol/L in immunoblots (Figure 1). An entirely different interpretation of the pharmacological effect of kallikrein could be based on the cleavage of kininogen produced or taken up by cells with the generation of kinins. There is surface uptake of kininogen via more or less specific “receptors” in endothelial and other cell types.18 We have easily simulated kininogen cell uptake by exposing HEK 293 cells expressing or not B2R-GFP to human plasma (Figure 5), followed by washing and treatment with kallikrein: large quantities of immunoreactive BK and some of its metabolite des-Arg9-BK were measured (Table). Even under unfavorable conditions wherein HEK 293 cells were not pretreated with plasma and wherein the formed kinins were likely to be diluted when diffusing from the cell surface into the large volume of supernatant, kallikrein significantly released some immunoreactive BK from the cells (Table). The antibodies used in the immunoassays are not discriminative for N-terminal variants of the kinin sequence, so Lys-BK is included in the BK concentration, but not intact kininogen, as proteins are precipitated before the assay.11 Some receptor-mediated uptake of BK is likely, based on the difference of immunoreactive kinin concentrations measured in the culture media of cells that expressed or not recombinant B2R-GFP (Table; this is why kinin release by kallikrein has been tested in untransfected HEK 293 cells when plasma pretreatment was not applied). Rabbit blood was in contact with the jugular vein before tissue isolation in experiments reported in Figure 6. An indirect mechanism of action of kallikrein through the local generation of kinins would explain the drastic desensitization of the jugular vein repeatedly exposed to kallikrein, as the local kininogen stores would be consumed to form BK. There is no cross-desensitization between kallikrein and BK (Figure 6), which suggests that receptor presence is not limiting, and therefore, that there is no important degradation of B2Rs exposed to 50 nmol/L of kallikrein.
In our estimation, the pharmacological activity of lower concentration levels (50 to 100 nmol/L) of tissue kallikrein on the rabbit B2R is highly tachyphylactic in isolated tissues, likely to be indirect, and dependent on the local formation of kinins. Local generation of kinins, which subsequently activate cell receptors, probably accounts for the effect of human kallikrein in various animal models.2,3 Transgenic mice with a liver-targeted expression of the enzyme and a hypotensive phenotype exhibited a maximal kallikrein serum level of 7 nmol/L.19 Only higher kallikrein levels (0.5 to 1 μmol/L) can cleave and degrade the B2Rs but also exert nonreceptor-mediated effects.
Results shown here and elsewhere17 based on the rabbit BK B2R suggest that this receptor does not efficiently function as a PAR when the pharmacological actions of kallikrein are considered. However, receptor degradation initiated by extracellular proteases (such as those secreted by activated neutrophils)17 is a possible mechanism for the downregulation of B2Rs observed in injured tissues.
This study was supported by the Canadian Institutes of Health Research (grant MOP-14077, studentship to S.H.).
- Received June 27, 2002.
- Revision received August 8, 2002.
- Accepted December 19, 2002.
Emanueli C, Minasi A, Zacheo A, Chao J, Chao L, Salis MB, Straino S, Tozzi MG, Smith R, Gaspa L, Bianchini G, Stillo F, Capogrossi MC, Madeddu P. Local delivery of human tissue kallikrein gene accelerates spontaneous angiogenesis in mouse model of hindlimb ischemia. Circulation. 2001; 103: 125–132.
Molino M, Woolkalis MJ, Reavey-Cantwell J, Pratico D, Andrade-Gordon P, Barnathan ES, Brass LF. Endothelial cell thrombin receptors and PAR-2: two protease-activated receptors located in a single cellular environment. J Biol Chem. 1997; 272: 11133–11141.
Hecquet C, Tan F, Marcic BM, Erdös EG. Human bradykinin B2 receptor is activated by kallikrein and other serine proteases. Mol Pharmacol. 2000; 58: 826–836.
Jagels MA, Travis J, Potempa J, Pike R, Hugli TE. Proteolytic inactivation of the leukocyte C5a receptor by proteinases derived from Porphyromonas gingivalis. Infect Immun. 1996; 64: 1984–1991.
Bachvarov DR, Houle S, Bachvarova M, Bouthillier J, Adam A, Marceau F. Agonist-induced rabbit bradykinin B2 receptor endocytosis and recycling assessed using green fluorescent protein conjugates. J Pharmacol Exp Ther. 2001; 297: 19–26.
Houle S, Larrivée JF, Bachvarova M, Bouthillier J, Bachvarov DR, Marceau F. Antagonist-induced intracellular sequestration of the rabbit bradykinin B2 receptor. Hypertension. 2000; 35: 1319–1325.
Adam A, Albert G, Calay J, Closset J, Damas P, Franchimont P. Low and high molecular weight human kininogens: quantification by a rapid radioimmunoassay and determination of reference values. Clin Chem. 1985; 31: 423–426.
Bachvarov DR, Saint-Jacques E, Larrivée JF, Levesque L, Rioux F, Drapeau G, Marceau F. Cloning and pharmacological characterization of the rabbit bradykinin B2 receptor. J Pharmacol Exp Ther. 1995; 275: 1623–1630.
Pimenta DC, Chao J, Chao L, Juliano MA, Juliano L. Specificity of human tissue kallikrein towards substrates containing Phe-Phe pair of amino acids. Biochem J. 1999; 339: 473–479.
Corish P, Tyler-Smith C. Attenuation of green fluorescent protein half-life in mammalian cells. Prot Eng. 1999; 12: 1035–1040.
Abe Y, Nakayama K, Yamanaka A, Sakurai T, Goto K. Subtype-specific trafficking of endothelin receptors. J Biol Chem. 2000; 275: 8664–8671.
Renne T, Dedio J, David G, Müller-Esterl W. High molecular weight kininogen utilizes heparan sulfate proteoglycans for accumulation on endothelial cells. J Biol Chem. 2000; 275: 33688–33696.