Angiostatin, a proteolytic fragment of plasminogen, is a potent inhibitor of angiogenesis and suppresses primary and metastatic tumor growths (1,2). Native angiostatin occurs in some glycoforms, whereas recombinant angiostatin (Lys⁷⁸-Val³³⁸ of plasminogen) used in clinical trials lacks glycosyl modifications (3). Recent report has suggested the importance of a certain form of Thr³⁴⁶-Oglycosylation (plasminogen numbering) in angiostatin activity (4), thus emphasizing the significance of the production of properly glycosylated angiostatin. The production of such a natural glycoform can be achieved by the proteolytic cleavage of plasminogen present in plasma through processes including the purification of plasminogen from plasma, the cleavage of plasminogen by elastase, and the purification of the resulting angiostatin, which required careful quality control (2) (possibly owing to protein denaturation during the cleavage and subsequent purification and to the formation of multiple forms due to the lower specificity of elastase cleavage). Although immobilized enzymes are widely used, the practical application of immobilized protease is limited owing to the instability and low efficiency (due to restricted accessibility to the substrate) of the immobilized protease and to the need for the purification of digested protein fragments. Moreover there are considerable problems with the use of a crude mixture of proteins such as plasma as substrate. To overcome such problems we have developed an immobilized protease reactor with an affinity ligand for the substrate and product. We presumed that the introduction of an affinity ligand could enable a specific, efficient cleavage of the desired protein in a crude mixture by capturing and concentrating the substrate on matrices where the protease is immobilized, and that this would enable reactor operation even at a low temperature to prevent enzyme and substrate inactivation or denaturation. We also expected that product purification could be achieved simultaneously.

Bacillolysin MA (BL-MA) is a metalloproteinase that preferentially cleaves at Ser⁴⁴¹-Val⁴⁴² in plasminogen to produce an antiangiogenic fragment, BL-angiostatin (Glu¹Ser⁴⁴¹ of plasminogen) (5). We immobilized BL-MA together with lysine, which binds to plasminogen (6), on CNBr-activated Sepharose™ 4B or HiTrap™ NHS-activated HP (Amersham Biosciences, Piscataway, NJ, USA). One volume of CNBr-activated Sepharose 4B [pre-equilibrated with buffer A (0.1 M NaHCO₃, 0.5 M NaCl, and 0.65 M isopropanol, pH 8.3)] was incubated with two volumes of the enzyme solution (2.5 mg/mL in buffer A) for 2 h at room temperature. After draining the enzyme solution, the gel was further incubated with two volumes of a mixture containing 0.2 M lysine and 0.65 M isopropanol, pH 8.0, for 2 h at room temperature. Finally, the gel was washed with 50 volumes of buffer B (25 mM sodium phosphate and 0.65 M isopropanol, pH 7.4) and stored in buffer B containing 3.1 mM sodium azide at 4°C until use. One milliliter of the gel contained 26.9 nmol (0.91 mg) BL-MA and 4.75 µmol lysine in a typical experiment. After several preliminary experiments using HiTrap NHS-activated HP as matrix, we noted that the reactor containing as low as 0.2 mg immobilized BL-MA/mL gel is as effective as the reactor described above in terms of BL-angiostatin yield and operational stability.

The BL-MA/Lys affinity-capture reactor was successful in producing and purifying BL-angiostatin from plasma at 4°C in the presence of 0.65 M isopropanol, conditions that inhibit protease activity by >90% and prevent protease degradation while allowing the specific cleavage of plasminogen. Such conversion and purification, which takes about 3 h, entails plasma application, reactor washing, and BL-angiostatin elution with 6-aminohexanoic acid ((Figure 1)). An example of a practical reactor operation, which was carried out at 4°C, is as follows: human citrated plasma, to which 0.65 M isopropanol was added and which was centrifuged at 15,000× g for 20 min (105 mL; actual plasma volume of 90 mL), was loaded onto a 10-mL reactor (preequilibrated with buffer B) for 1 h. The reactor was washed with 300 mL buffer B containing 0.5 M NaCl for 1.5 h. BL-angiostatin was eluted for 0.5 h with 50 mL 0.2 M 6-aminohexanoic acid containing 0.65 M isopropanol, and 10-mL fractions of the eluate were collected. The major portion of BL-angiostatin was recovered in the second and third fractions. After its operation, the reactor was washed at 4°C with 50 mL buffer B containing 1 M NaCl and 0.2 M 6-aminohexanoic acid and then with 50 mL buffer B containing 3.1 mM sodium azide each for 0.5 h. The reactor was stored at 4°C until use.

Figure 1.

A 10-mL reactor afforded 4–6 mg BL-angiostatin (40%–60% molar recovery from plasminogen) with 80%–95% purity (plasminogen as a major impurity) from 90 mL human plasma ((Figure 2)A). The identity of the purified protein as BL-angiostatin (Glu¹Ser⁴⁴¹ of plasminogen) was confirmed by the result that the protein consisted of two species with molecular masses of 65.4 and 62.0 kDa in reduced sodium dodecyl sulfate poly aery 1 amide gel electrophoresis (SDS-PAGE) ((Figure 2)B; glycoforms I and II, respectively), each of which had an N-terminal amino acid sequence identical to that of Glu-plasminogen ((Figure 2)A) and was positively detected by immunoblotting with anti-plasminogen immunoglobulin G (IgG) (data not shown). The reactor could be used repeatedly for 16 days with slight reductions in recovery and purity ((Figure 2)C). Since the major impurity is plasminogen, the reduction in the purity of BL-angiostatin may be due to a decrease in the catalytic activity of immobilized BL-MA. The antiangiogenic activity of BL-angiostatin was evaluated by in vitro angiogenesis assay using a co-culture system of human dermal fibroblasts and human umbilical vein endothelial cells (Kurabo, Osaka, Japan). In this assay, BL-angiostatin inhibited vascular endothelial growth factor (VEGF)-induced angiogenesis with a median inhibitory concentration (IC₅₀) of 22±11 nM. A pooled BL-angiostatin preparation, which was confirmed to show biological activity in vivo using a mouse dorsal air sack angiogenesis model (7) and a mouse implanted tumor model (see below), was used as standard.

Figure 2.

The reactor enabled the preparation of large amounts of BL-angiostatin that has, as compared with recombinant angiostatin (3), native oligosaccharides and a 77-amino acid N-terminal hydrophilic peptide (NTP), each of which may confer pharmacokinetic advantages on BL-angiostatin, because nonglycosylated plasminogen (8) or plasminogen without NTP (9) is much more rapidly cleared from circulation than native plasminogen. BL-angiostatin exhibits antitumor activity in a mouse Lewis lung carcinoma implantation model at a dose as low as 0.3 mg/kg body weight per day, which is approximately 100 times lower than the effective dose of recombinant angiostatin (3,10).

A rapid, simple conversion and purification using the BL-MA/Lys affinity-capture reactor enables a large-scale production of native BL-angiostatin for clinical application. Moreover, the use of the reactor may enable the bedside preparation of BL-angiostatin for self-treatment by cancer patients using their own plasma. The concept of the affinity-capture reactor can be extended to other combinations of an enzyme and a substrate ligand.