High-Level Production of Recombinant Human Parathyroid Hormone 1-34

Abstract

Expression of the synthetic human parathyroid hormone 1-34 [hPTH(1-34)] gene by a gene fusion strategy was demonstrated. hPTH(1-34) was produced at the C terminus of the partner peptides involving amino acids 1 to 97, 1 to 117, or 1 to 139 of a modified Escherichia coli β-galactosidase by linker peptides containing oligohistidine of different lengths. The fusion proteins in the inclusion bodies were rendered soluble with urea and subjected to site-specific cleavage with the secretory type yeast Kex2 protease. Optimal expression and enzymatic processing were achieved in the fusion protein βG-117S4HPT, constructed from amino acids 1 to 117 of β-galactosidase and the linker of HHHHPGGSVKKR. The fusion protein accumulated more than 20% of the E. coli total protein. The hPTH(1-34) was purified up to 99.5% with a good yield of 0.5 g/liter of culture. The purified product was identified as intact hPTH(1-34) by amino acid analysis and N-terminal sequencing.

Synthetic human parathyroid hormone 1-34 [hPTH(1-34)] is recognized to cover most of the hormonal actions of the intact human parathyroid hormone [hPTH(1-84)] regulating calcium/phosphate homeostasis and controlling bone resorption (4, 10). Intermittent administration of hPTH(1-34) to patients can increase bone mass (12). The whole mechanism is still under discussion, but low-dose hPTH triggers cyclic AMP-dependent protein kinase in some populations of bone cells bearing PTH receptors, which stimulates the proliferation of osteoblasts (2). At present, hPTH is undergoing clinical trials for use in osteoporosis treatment; however, it may be hard for patients to continue the treatment for years in compliance with periodic subcutaneous or intramuscular injection. This drawback can be overcome by nasal or oral delivery, although the bioavailability has been estimated to be as low as a few percent of that of subcutaneous delivery. A method which would enable the mass production of hPTH at low cost is keenly awaited. Chemical synthesis often involves high risk and cost, and although production via recombinant genetic technology has been expected to replace this process, the yield has so far been insufficient (5). We have developed production methods for the human arterial natriuretic peptide (hANP), human C-type natriuretic peptide (hCNP), and human calcitonin (11) up to a pilot or commercial level. In these methods, we used gene fusion for efficient inclusion body formation of the fusion proteins, which suppressed proteolytic damage by the host cells (8). The fusion proteins were constructed from a truncated Escherichia coli β-galactosidase derivative, a linker peptide, and the target peptides. The linker peptide was designed to supply a proteolytic cleavage site and to improve the productivity of the fusion protein (10 to 30% of total host cell proteins). The insoluble fusion proteins could be easily purified from the cell lysate by a few rounds of centrifugation and resuspension. Enzymatic site-specific processing was performed to release the target peptide, which was subsequently applied to the downstream purification processes. Proteases with strict specificity to the processing site were chosen, which enabled a relatively simple and efficient purification process because there was less contamination of the short peptides degraded from the fusion protein. Achromobacter protease I and Staphylococcus aureus V8 protease were used for hANP and human calcitonin production, respectively. In this paper, we describe an application of this method to hPTH(1-34) production and downstream processing that promises the efficient production of hPTH(1-34) with high purity.

MATERIALS AND METHODS

Bacterial strains and the construction of expression vectors.

E. coli JM109 was used for genetic manipulation, and E. coli M25 (W3110 OmpT⁻) (7) was used for expression. The general procedures for DNA manipulation, cloning, and PCR were as described previously (6) and as recommended by the respective manufacturers.

Genes of the fusion proteins were expressed under the control of an E. coli lac promoter-operator system on a pBR322-based plasmid (Fig. 1). An SmaI-StuI site was introduced at the 3′ end of the partner peptide gene by PCR, generating pG97SPT, pG117SPT, and pG139SPT. The sense and antisense oligonucleotides encoding HHHHPG, HHHHPGHHHHPG, and KKKKPG (abbreviated 4H, 8H, and 4K, respectively) were then synthesized. Each of them (or a mixture of 4H and 8H) was introduced into the SmaI sites to generate a wide variety of fusion proteins. The orientation and length of each insert were confirmed by restriction endonuclease analysis and DNA sequencing with an ALF DNA sequencer (Pharmacia Biotech, Uppsala, Sweden), generating pG97SnHPT, pG117SnHPT, and pG139SnHPT, where “n” indicates the number of introduced His or Lys residues.

FIG. 1.

Expression vectors for the hPTH(1-34) fusion protein production. A schematic representation of the expression vectors of the partner peptides, linker peptides, and fusion proteins is given. “m” and “n” indicate the number of inserted oligohistidine linkers and the number of histidine residues, respectively. Plac, E. coli lac operator-promoter; Ttrp, E. coli trp terminator; pBR ori, replication origin of pBR322; Tet, tetracycline resistance gene; Sma I, the SmaI site diminished by linker insertion.

Expression of the fusion protein genes.

Every single colony of the fresh transformants of E. coli M25 harboring the expression vectors was inoculated and cultured in 50 ml of Terrific broth (6). Isopropyl-β-d-thiogalactopyranoside (IPTG) was then added to a concentration of 1.0 mM as the cell concentration reached an optical density at 660 nm of 1.0, and incubation was continued for an additional 4 h. For a large-scale culture, colonies of the fresh transformants were suspended in 10 ml of Luria broth and inoculated into 20 liters of Terrific broth supplemented with 2 mM methionine in a 30-liter fermentor. Expression was performed as indicated above, except that the cell concentration was 3.0 optical density at 660 nm units when IPTG was added.

Site-specific cleavage of the fusion protein with Kex2.

Kex2-660, a secretory type-Kex2 protease was designed, produced from recombinant Candida boidinii, and purified in our laboratory. Inclusion bodies were recovered by differential centrifugation from the cell lysate and successively washed with Tris-EDTA (TE), 1% Triton–5 mM EDTA, and 50 mM NaCl. An aliquot of the dense suspension was transferred to a test tube, dissolved by the addition of 10 M biochemical grade urea, and diluted with a buffer to give a reaction mixture of 20 mM Tris · HCl (pH 8.2), 50 mM NaCl, 2.0 mM CaCl₂, 3.0 M urea, and 2 to 3 mg/ml of the fusion protein. After being preincubated at 30°C for 10 min, the reaction mixture was provided with Kex2-660 at an enzyme-to-substrate molar ratio of 1:2,000, and the incubation was continued.

Quantification of hPTH(1-34) and the fusion protein.

Aliquots of the hPTH(1-34) solution were appropriately diluted in 1.0 ml of 2.0 M urea–1.0 M acetic acid, and the supernatant was subjected to chromatography on an ODS A302 column (YMC, Kyoto, Japan) for 20 min at a flow rate of 1.0 ml/min with a linear gradient of 28.4 to 40.4% acetonitrile containing 0.1% trifluoroacetic acid. The hPTH(1-34) content was calculated by using a standard consisting of hPTH(1-34) whose concentration was determined by amino acid analysis. The fusion protein concentration was evaluated from the amount of hPTH(1-34) released in a complete digestion reaction with Kex2-660.

Preparative hPTH(1-34) purification.

Inclusion bodies of βG-117S4HPT were recovered from a 20-liter culture of E. coli M25 harboring pG117S4HPT by differential centrifugation and were repeatedly washed with TE. The concentrated suspension of the inclusion bodies was dissolved in 8.0 M urea with buffering reagents, before the solution was diluted with deionized water, to give a 15-liter solution of 20 mM Tris · HCl (pH 8.2), 50 mM NaCl, 2.5 mM CaCl₂, 3.0 M urea, and 8 mg of the fusion protein per ml. Kex2-660 was added at an enzyme-to-substrate molar ratio of around 1:2,000, and the solution was incubated at 30°C for 1 h. The reaction mixture was then diluted with deionized water, and most of the peptide impurities were precipitated by adding acetic acid. The supernatant was applied to a Poros HS-50 (PerSeptive Biosystems, Framingham, Mass.) column (inner diameter [ID], 100 by 100 mm) that had been equilibrated with 10 mM sodium acetate (pH 5.0)–1.5 M urea, washed with 2 column volumes (CV) of the same buffer, and then washed with 2 CV of 20 mM sodium acetate (pH 6.5)–1.5 M urea. Then the hPTH(1-34) was eluted with a linear gradient of 0 to 0.3 M NaCl containing 20 mM sodium acetate (pH 6.5)–1.5 M urea. The eluate, supplemented with 3.0% acetic acid, was applied to a Poros R2-50 (PerSeptive Biosystems) column (100 by 100 mm [ID]) that had been equilibrated with 3% acetic acid, washed with 2 CV of the same buffer, and eluted with 2 CV of 3% acetic acid containing 30% acetonitrile. These chromatographic procedures were performed at a flow rate of 1.0 liters/min at room temperature. Acetonitrile in the eluate was removed by evaporation, and then the rested solution was passed through a 0.22-μm-pore-size filter (Millipore, Bedford, Mass.) and chromatographed on a TSKgel ODS-120T column (600 by 55 mm [ID]; Tosoh, Tokuyama, Japan) that had been equilibrated with 5% acetic acid, washed with 1 CV of the same buffer, and eluted at a flow rate of 40 ml/min with a linear gradient of acetonitrile containing 5.0% acetic acid. Fractions containing hPTH(1-34) of greater than 99% purity were collected.

RESULTS AND DISCUSSION

Design of the fusion protein.

Fusion proteins should be designed to exploit the advantages of high expression and production in inclusion bodies for easy recovery and of strict site-specific cleavage for efficient purification of target peptides. A wide variety of fusion proteins were constructed by combining three partner peptides of truncated E. coli β-galactosidase, the linker peptides providing a proteolytic processing site, and hPTH(1-34) (Fig. 1). The length of the partner peptide is always a significant factor in determining productivity and solubility. First, three fusion proteins, βG-139SPT, βG-97SPT, and βG-117SPT, were constructed. Second, an oligohistidine-containing peptide (HHHHPG)_m (m = 1 to 4) was systematically incorporated into the fusion protein to generate βG-97SnHPT, βG-117SnHPT, and βG-139SnHPT, as described in Materials and Methods, where “n” indicated the number of incorporated histidine residues. It was expected that some of the fusion proteins would exhibit high productivity with improved solubility potential in the processing reaction mixture (1, 9). βG-97S4KPT, βG-117S4KPT, and βG139S4KPT were also constructed to investigate the effect of a large pI shift on their productivity.

For specific processing, the proteolytic site was designed as follows. Because hPTH(1-34) contains Lys, Arg, Glu, Asp, and some aromatic amino acid residues, proteases recognizing these amino acid residues should be avoided. It is desirable that the processing occur only at the processing site to eliminate contamination by degraded peptide fragments, which would make the purification process complicated and result in low recovery. Proteases fulfilling this requirement can be found among those recognizing sequential amino acids like prohormone convertases. Kex2-660 had the same substrate specificity as Kex2 in strictly recognizing the RR, KR, and PR sequences (3) not involved in hPTH(1-34), and the inactivation of Kex2-660 became moderate during the processing reaction with the addition of more than 1.0 mM of calcium ion (data not shown). With these considerations, we adopted the original cleavage site of the hPTH precursor (SVKKR) as the processing site of Kex2-660. The other recognition site for Kex2-660, Arg-Arg (amino acids 14 to 15 of each partner peptide), was eliminated by substituting Lys for Arg15 (data not shown).

Improving the production level of the fusion protein.

E. coli cells harboring the expression vectors were cultured as described in Materials and Methods, with the production of the fusion proteins being improved by oligohistidine incorporation, as shown in Fig. 2. All the fusion proteins were accumulated in inclusion bodies (data not shown). In the case of βG-97SnHPT, the productivity increased with an increase in the length of the incorporated oligohistidine linker (lanes 1 to 4). In the same manner, significantly enhanced productivity was observed in every βG-117SnHPT and βG-139SnHPT with oligohistidine linkers (lanes 6 to 11 and lanes 13, 14, and 16, respectively). The range of calculated pI (pI_cal) values resulting in optimum production seemed to depend on the length of the fusion proteins, with the longer fusion proteins having more potential for productivity at around the optimum pI_cal value. Incorporation of four Lys residues (4K) was less effective, as predicted from the high pI_cal value (lanes 5, 12, and 15); however, βG-139S4KPT, with pI_cal close to the neutral value and with the longest partner peptide, resulted in increased productivity (lanes 13 and 15). These results demonstrate that pI shift by oligohistidine incorporation could provide a useful tool for controlling the productivity of fusion proteins. βG-117S4HPT resulted in high productivity with just one four-His residue cluster (4H), whereby the content of hPTH(1-34) was relatively high. This could be due to a pI shift as well as due to the flat and stable net charge at around neutral pH by oligohistidine, which might facilitate inclusion body formation in the host cell by suppressing inter- and intramolecular electric expulsion. Our strategy of systematic combination of partner peptides with different lengths and the incorporation of oligohistidine will probably be applicable to the production of a wide variety of short peptides, although there are also some other factors affecting the productivity of the fusion proteins such as the stability of mRNA, the intramolecular charge distribution, and different hydropathy patterns.

FIG. 2.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the productivity of the fusion proteins. Aliquots of the total lysate of each strain containing the same number of cells were loaded onto a sodium dodecyl sulfate–16% polyacrylamide gel (TEFCO, Chino, Japan) as described previously (11). The identity of each lane number, with the putative isoelectric point (pI_cal), is as indicated. The positions of the molecular mass standards in kilodaltons are shown to the left of the figure. pI_cal was calculated with the DNASIS program (Hitachi Tokyo, Japan).

Besides the production level, it is also important that the fusion proteins have sufficient solubility at a low denaturant concentration, because the activity of Kex2-660 decreases to half its original level in 3 M urea (pH 8.0) within 20 min, even in the presence of 2.5 mM CaCl₂. Oligohistidine incorporation increased the potential solubility to more than 15 g/liter in βG-117SnHPT, while that of βG-139SPT was less than 8.0 g/liter (data not shown).

Processing of fusion proteins with Kex2-660.

Urea (3 M), required to maintain the solubility of the fusion proteins in the reaction mixture, significantly inactivated Kex2-660. On the other hand, 2.5 mM CaCl₂, required to suppress Kex2-660 inactivation, facilitated the precipitation of the fusion proteins. These antipodal phenomena restricted optimization of the processing conditions (in which the approximate enzyme-to-substrate molar ratio was as large as 1:2,000), with a low conversion rate of 80%. Although a good yield was produced, Kex2-660 still accounted for too large aproportion of the primary cost of the production process. Therefore, a fusion protein which required a low enzyme-to-substrate molar ratio for complete processing was preferable. Site-specific processing of Kex2-660 was carried out to determine whether the release of hPTH(1-34) from the fusion proteins was affected by the introduction of oligohistidine (Table 1). The length of the partner peptide hardly affected the k_cat/K_m ratio, although K_m and k_cat slightly increased in βG-139S8HPT, the longest of the partner peptides. A secondary-structure prediction of the recognition site revealed an increase in flexibility (turn or random coil) by oligohistidine, which could explain the decreased K_m value of βG-117S4HPT or βG-117S8HPT in terms of enzyme accessibility. Unfortunately, the oligohistidine incorporated in βG-117SPT decreased both K_m and k_cat, resulting in a small increase of k_cat/K_m, which needs to be overcome by other strategies such as optimizing the upstream subsites of the recognition site (7a).

TABLE 1.

Kinetic parameters of typical fusion proteins in Kex2-660 processing

Fusion protein	Km (μM)	kcat (s−1)	kcat/Km (s−1 μM−1)
βG-97S8HPT	92	23.3	0.25
βG-117SPT	186	31.2	0.17
βG-117S4HPT	118	27.5	0.23
βG-117S8HPT	86	21.6	0.25
βG-139S8HPT	140	30.3	0.22

Purification of hPTH(1-34).

Purification was carried out as described in Materials and Methods. The strict and precise processing reaction generated a single sharp peak of hPTH(1-34) (Fig. 3). Most of the undigested fusion protein, partner peptides, and large amounts of host cell-derived impurities could be efficiently removed by acid precipitation followed by press filtration. This treatment was a key step to recovery of hPTH(1-34) in the clear supernatant that enabled smooth chromatographic operations. Most impurities, such as short peptide fragments from βG-117S4HRH, which had been generated in the inclusion bodies were effectively separated by strong cation-exchange chromatography (Poros HS 50) with a linear NaCl gradient. After exchanging buffer and concentrating hPTH(1-34) through a reversed-phase column (Poros R2 50), derivatives of hPTH(1-34), dyes, and other impurities were separated by fractionation through a reversed-phase high-performance liquid chromatography column (TSKgel ODS-120T). The purity, yield, and overall recovery were 99.5%, 0.5 g/liter of culture, and 48%, respectively. A single sharp peak was detected when the synthetic product was cochromatographed with standard hPTH(1-34), and the amino acid composition analysis and N-terminal sequencing of the purified product revealed close correlation with those of authentic hPTH(1-34) (data not shown).

FIG. 3.

Site-specific processing of βG-117S4HPT with Kex2-660 as represented by the high-performance liquid chromatography elution profiles. Peaks: 1, hPTH(1-34); 2, β-Gal-117S-4H; 3, βG-117S4HPT. Traces: A, before processing; B, after processing.