FK506 Binding Protein from the Hyperthermophilic Archaeon Pyrococcus horikoshii Suppresses the Aggregation of Proteins in Escherichia coli

Abstract

The 29-kDa FK506 binding protein (FKBP) gene is the only peptidyl-prolyl cis-trans isomerase (PPIase) gene in the genome of Pyrococcus horikoshii. We characterized the function of this FKBP (PhFKBP29) and used it to increase the production yield of soluble recombinant protein in Escherichia coli. The PPIase activity (k_cat/K_m) of PhFKBP29 was found to be much lower than that of other archaeal 16- to 18-kDa FKBPs by a chymotrypsin-coupled assay of the oligo-peptidyl substrate at 15°C. Besides this low PPIase activity, PhFKBP29 showed chaperone-like protein folding activity which enhanced the refolding yield of chemically unfolded rhodanese in vitro. In addition, it suppressed thermal protein aggregation in a temperature range of 45 to 100°C. When the PhFKBP29 gene was coexpressed with the recombinant Fab fragment gene of the anti-hen egg lysozyme antibody in the cytoplasm of E. coli, whose expressed product tended to form an inactive aggregate in E. coli, it improved the yield of the soluble Fab fragments with antibody specificity. PhFKBP29 exerted protein folding and aggregation suppression in E. coli cells.

FK506 binding protein (FKBP) is a family of peptidyl-prolyl cis-trans isomerases (PPIase) that accelerates the isomerization of the proline imide bond in polypeptides, the rate-limiting step in protein folding (2). Archaeal FKBPs have been classified into two groups: 16- to 18-kDa FKBPs (short-type FKBPs) and 26- to 33-kDa FKBPs (long-type FKBPs) (20). It has been reported that the short-type FKBP from archaea possesses not only PPIase activity but also chaperone-like protein folding activity in vitro (4, 13). The long-type FKBPs consist of an N-terminal FKBP domain, which is homologous to that in the short-type FKBPs, and an adjacent C-terminal domain comprising ca. 100 amino acids (20). A long-type FKBP from Methanobacterium thermoautotrophicum (MbtFKBP28, gene number MTH1125 [http://www.ncbi.nlm.nih.gov/]) has shown detectable but very low PPIase activity, in contrast to the short-type FKBPs, while it showed weak but significant chaperone-like protein folding activity with protein aggregation suppression in vitro (12). A genome sequence analysis has indicated that the hyperthermophilic archaea Pyrococcus horikoshii, Archaeoglobus fulgidus, and Aeropyrum pernix have only the long-type FKBP gene as a PPIase in their genomes (20). It is interesting to know whether these long-type archaeal FKBPs possess little PPIase activity but significant chaperone-like activity in the cell. In the present study, we focused on the long-type FKBP (PhFKBP29, gene number PH1399 [http://www.ncbi.nlm.nih.gov/]) of the hyperthermophilic archaeon, P. horikoshii (7, 17), which is 43% identical to MbtFKBP28 in the amino acid sequence (for the amino acid sequence alignment, see references 12 and 20), and investigated its protein folding activity in vitro. To examine the effect of PhFKBP29 on protein folding in vivo, we used the Escherichia coli expression system of a recombinant protein, which forms an insoluble and improperly folded aggregate in the cell. Since the Fab fragment of the antibody is expressed as an insoluble aggregate in the cytoplasm of E. coli (6), we used this as the model recombinant target protein for an in vivo study. We demonstrate that PhFKBP29 significantly suppressed protein aggregation in vitro and that it improved the expression of the soluble form of the Fab fragment in the cytoplasm of E. coli.

MATERIALS AND METHODS

Microbial strains and chemicals.

Rhodanese (from bovine liver), human 12-kDa FKBP (HsFKBP12), E. coli GroEL/ES mixture and N-succinyl-Ala-Ala-Pro-Phe-p-nitroanilide (N-suc-AAPF) were purchased from Sigma Chemical Co. (St. Louis, Mo.). N-Succinyl-Ala-Leu-Pro-Phe-p-nitroanilide (N-suc-ALPF) was purchased from Peptide Institute, Inc. (Osaka, Japan). MbtFKBP28 was prepared as described previously (12). The rabbit antiserum against FKBP from P. horikoshii (PhFKBP29) was prepared by Takara Shuzo Co. (Kyoto, Japan). The protein concentration was determined by the Bradford dye-binding method (1) with a Bio-Rad (Richmond, California) protein assay kit, with bovine serum albumin (BSA) as the standard.

Construction of the expression plasmids for PhFKBP29.

Cells of P. horikoshii were harvested by centrifugation from 300 μl of a cell suspension obtained from the Japan Collection of Microorganisms (Riken, Saitama, Japan). The genomic DNA was prepared from the cells according to a previously described procedure (12) and used for the PCR template. The gene for FKBP from P. horikoshii (PhFKBP29) was amplified by PCR with the primer set of PhFK-F1 and PhFK-R1 (Table 1). The amplified DNA fragment was recovered and cloned into a pT7 blue-T vector (Novagen, Madison, Wis.). The gene sequence was confirmed with an ABI PRISM 3700 DNA sequencer (Perkin-Elmer, Norwalk, Conn.). The cloned FKBP gene was digested with the restriction enzymes and ligated into the corresponding sites of the plasmid vector, pET21a (Novagen), for expression. The resulting expression vector for PhFKBP29 is named pEPhFK-1.

To construct an expression plasmid of PhFKBP29 that was compatible with the Fab expression plasmid in E. coli, the gene for PhFKBP29, together with the adjacent T7 promoter and terminator in plasmid pEPhFK-1, was amplified by PCR with the primer set of PhFK-F2 and PhFK-R2 (Table 1). The amplified DNA fragment was ligated into the corresponding sites of the plasmid vector, pACYC184 (Nippon Gene, Tokyo, Japan), as described above. The resulting expression vector for PhFKBP29 is named pACPhFK-1.

Expression and purification of recombinant PhFKBP29.

The expression vector for PhFKBP29, pEPhFK-1, was introduced into E. coli BL21(DE3). The transformant was grown in 700 ml of 2xYT medium (yeast extract, 10 g; tryptone, 16 g; NaCl, 5 g/liter of medium) containing 100 μg of ampicillin/ml at 35°C for 24 h. The harvested cells were sonicated in 25 mM HEPES-KOH buffer (pH 6.8) containing 1 mM EDTA and, after centrifugation to remove the cell debris, the resulting cell extract was loaded onto a DEAE Toyopearl column (16 mm by 60 cm; Tosoh Co.) and then eluted with a linear gradient of 0.5 M NaCl in 25 mM HEPES-KOH buffer (pH 6.8) at a flow rate of 1 ml/min. The eluted PhFKBP29 fractions were pooled, concentrated, applied to a HiLoad Superdex 200-pg column (26 mm by 60 cm; Amersham Pharmacia Biotech, Uppsala, Sweden) that had been equilibrated with 100 mM sodium phosphate (pH 7.0) containing 0.15 M NaCl, and then eluted at a flow rate of 3 ml/min. The eluted PhFKBP29 was concentrated, loaded onto a TSKgel SuperQ-5PW column (7.5 mm by 7.5 cm; Tosoh Co.), and eluted with a linear gradient of NaCl (0 to 0.5 M) in 25 mM HEPES-KOH buffer (pH 6.8) at a flow rate of 1 ml/min. The resulting fractions containing FKBP were combined and applied to a TSKgel G3000 SW_XL column (7.5 mm by 30 cm; Tosoh Co.) and then eluted with 100 mM sodium phosphate (pH 7.0) containing 0.15 M NaCl. PhFKBP29 was identified by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) with Coomassie brilliant blue R250 staining.

CD spectroscopy.

To examine the thermostability of PhFKBP29, the FKBP was dissolved in 25 mM sodium phosphate (pH 7.0) at 0.05 mg/ml, and the temperature-dependent circular dichroism (CD) change was monitored at 222 nm with a Jasco J-725 spectrometer. Temperature was raised from 30 to 87°C at a heating rate of 1°C/min.

PPIase assay.

After we confirmed the stability of the FKBP to chymotrypsin digestion, we measured the PPIase activity by protease-coupled assay with the oligopeptides N-suc-ALPF and N-suc-AAPF as substrates (3, 23). The PPIase assay of 5 to 80 μM PhFKBP29 was carried out as described previously (12).

Refolding of unfolded rhodanese.

Chemical denaturation and the refolding of rhodanese in the presence of 0 to 60 μM PhFKBP29 was carried out as described previously (12). The rhodanese activity was measured according to the method of Horowitz (10). The refolding yield was calculated as the percent activity compared to that of native rhodanese, whose activity was considered as 100%.

Effect of PhFKBP29 on thermal aggregation of E. coli proteins in vitro.

To investigate the effect of PhFKBP29 on the thermal aggregation of proteins, the cell homogenate of E. coli BL21(DE3) was used as target proteins. PhFKBP29 was added to the cell homogenate at a final concentration of 1.0 mg/ml, the final concentration of E. coli protein being 2.2 mg/ml. The mixture was heated at various temperatures (60 to 100°C) for 30 min. After centrifugation at 34,000 × g for 20 min to precipitate the denatured E. coli proteins, the supernatant was analyzed by SDS-PAGE with an 18% polyacrylamide gel. The band density was analyzed with Bio-Rad Multi-Analyst software.

Protection of the thermal aggregation of rhodanese.

Native rhodanese (0.8 μM) was incubated at 45°C for 10 min in 50 mM sodium phosphate buffer (pH 7.8) containing 0 to 30 μM PhFKBP29 in a final volume of 1.5 ml. The thermal aggregation of rhodanese was continuously monitored by light scattering at 320 nm with a Jasco-777 fluorescence spectrophotometer.

Coexpression of PhFKBP29 and the recombinant Fab antibody fragment in E. coli.

To investigate the effect of PhFKBP29 on protein folding in vivo, the anti-hen egg lysozyme (HEL) Fab antibody fragment was used as the model protein (11). The genes of the heavy chain (D1.3VH-CH1) and the light chain (D1.3Vκ-Cκ) of the Fab fragment were amplified by PCR by using pAALFab (11) as the template. Their signal sequences were removed for cytoplasmic expression by using two PCR primer sets: HELVH-F1 and HELVH-R1 for the heavy chain and HELVL-F1 and HELVL-R1 for the light chain (Table 1). The amplified DNA fragments were digested with the restriction enzymes and ligated into the pET21a plasmid vector (Novagen). The resulting expression vector for the anti-HEL Fab fragment is named pEHELFab-1. Both pACPhFK-1 and pEHELFab-1 were introduced into E. coli JM109(DE3), and the transformant was grown in the 2xYT medium containing 100 μg of ampicillin and chloramphenicol/ml at 35°C. When the absorbance at 600 nm of the culture had reached 0.8, isopropyl-β-d-thiogalactopyranoside (IPTG) was added at 1 mM. The harvested cells were resuspended in 30 ml of the 25 mM HEPES-KOH buffer (pH 6.8) containing 1 mM EDTA. The cells were disrupted by sonication for 5 min on ice and centrifuged at 10,200 × g for 30 min to separate the supernatant (soluble fraction) and the precipitate. The precipitate was washed twice with and then resuspended in 30 ml of the same buffer (insoluble fraction). As a negative control, the transformant with only pEHELFab-1 was examined.

Western blot analysis of the recombinant anti-HEL Fab fragment and PhFKBP29 expressed in E. coli.

A total of 40 μg of E. coli proteins in the soluble fraction and the equivalent volume of insoluble fraction were dissolved in the sampling buffer and subjected to reducing SDS-PAGE analysis. The proteins were then blotted onto a polyvinylidine difluoride membrane (Bio-Rad). Anti-HEL Fab was detected with the horseradish peroxidase (HRP)-conjugated goat anti-mouse Fab polyclonal antibody (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, Md.) by using 3,3"-diaminobenzidine-4HCl as the substrate. To detect the expressed PhFKBP29, 60 ng of soluble protein and an equivalent volume of the insoluble fraction were analyzed by SDS-PAGE and Western blotting with the rabbit anti-PhFKBP29 serum. The bound antibody was visualized by using HRP-conjugated goat anti-rabbit antibody (ICN Biomedicals, Irvine, Calif.) with 3,3"-diaminobenzidine-4HCl as the substrate. The band densities were estimated as described above.

Enzyme-linked immunosorbent assay (ELISA) detection of the antigen by using recombinant anti-HEL Fab.

A 96-well microtiter plate coated with HEL was treated with 100 μl of the soluble fraction of the E. coli transformant cell lysate (2 mg/ml) in Tris-buffered saline (25 mM Tris-HCl, pH 7.4; 140 mM NaCl; 2.7 mM KCl) for 3 h at room temperature. The plate was then washed with the same buffer and incubated with the HRP-conjugated goat anti-mouse Fab polyclonal antibody (Kirkegaard & Perry). The color was developed with 2,2"-azido-di(3-ethyl-benzthiazoline-6-sulfonate) as a substrate, and the absorbance was measured at 405 nm.

RESULTS AND DISCUSSION

Thermostability of PhFKBP29.

The stability of archaeal FKBPs was examined by determining the CD change at 222 nm (Fig. 1). The secondary structure of long-type FKBP from Methanobacterium thermoautotrophicum (MbtFKBP28) was shown to change with a cooperative thermal denaturing transition at a melting temperature (T_m) of ca. 65°C. In contrast, no change was observed for FKBP from P. horikoshii (PhFKBP29) at 30 to 87°C, indicating that the secondary structure of PhFKBP29 was much more stable than MbtFKBP28.

FIG. 1.

Effect of temperature on the secondary structure of PhFKBP29 and MbtFKBP28. CD of the FKBPs was measured at 222 nm in a 1-cm cuvette with temperature control at the increasing rate of 1°C/min. Two scans were averaged, and the mean residue ellipticity was calculated.

Catalytic efficiency of PhFKBP29.

The catalytic efficiency of PPIase activity of PhFKBP29 was similar to that of other long-type archaeal FKBPs and was significantly lower than those of short-type archaeal FKBPs and other bacterial and eukaryotic PPIases (5, 8, 12, 14-16, 22) (Table 2). In the previous studies, an amino acid substitution of F99 with Y in HsFKBP12 (24), as well as the substitution of corresponding F with Y in the E. coli trigger factor and FKBP from Methanococcus thermolithotrophicus, significantly reduced their PPIase activities (4, 24). In PhFKBP29, the amino acid residue corresponding to this F was Y (20). The low PPIase activity of PhFKBP29 may be attributable to this substitution and is a common feature of archaeal long-type FKBPs. The k_cat/K_m value for PhFKBP29 to N-suc-ALPF was ninefold higher than that to N-suc-AAPF. This substrate specificity is similar to that of other FKBP-type PPIases. The PPIase activity of 15 μM PhFKBP29 with N-suc-ALPF as the substrate was inhibited to 75% by 20 μM of FK506.

TABLE 2.

Comparison of the catalytic efficiency with respect to N-suc-ALPF (Leu) and N-suc-AAPF (Ala) between archaeal long-type FKBPs and other PPIases

Organism	Abbreviationc	Catalytic efficiency (kcat/Km [mM−1·s−1])a		Source or reference
		Leu	Ala
P. horikoshii	PhFKBP29	1.57 ± 0.08	0.18 ± 0.02	This study
M. thermoautotrophicum	MbtFKBP28	0.36	0.19	Ideno et al. (12)
H. cutirubrum	HcFKBP33	0.78	0.15	Iida et al. (16)
Thermococcus sp. strain KS-1	TcFKBP18	350	290	Iida et al. (15)
M. thermolithotrophicus	MtFKBP17	380	200	Furutani et al. (5)
H. cutirubrum	HcCyP19	ND	970b	Iida et al. (14)
Homo sapiens	HsFKBP12	640	53	Harrison and Stein (8)
E. coli	Trigger factor	430	160	Stoller et al. (22)

^aND, not determined.

^bThe value was recalculated from published data.

^cMbtFKBP28, Methanobacterium thermoautotrophicum 28-kDa FKBP; HcFKBP33, Halobacterium cutirubrum 33-kDa FKBP; TcFKBP18, Thermococcus sp. strain KS-1 18-kDa FKBP; MtFKBP17, Methanococcus thermolithotrophicus 17-kDa FKBP; HcCyP19, H. cutirubrum 19-kDa cyclophilin type PPIase.

Chaperone-like protein folding activity of PhFKBP29.

Although PhFKBP29 showed detectable but very low PPIase activity, it significantly enhanced the yield of rhodanese refolding in a dose-dependent manner (t test, P < 0.05; Fig. 2). The refolding activity was weaker than that of GroEL/ES (0.4 μM). We found that a 1 μM concentration of short-type FKBP from the hyperthermophilic archaeon, Thermococcus sp. strain KS-1 (18-kDa FKBP), showed chaperone-like refolding activity comparable to that of GroEL/ES (85% ± 13% upon incubation for 45 min at 35°C [data not shown]). This indicates that the chaperone-like activity of PhFKBP29 was less than that of short-type FKBP. The chaperone-like activity of 30 μM PhFKBP29 was inhibited by 40 μM FK506 (Fig. 2). HsFKBP12 and BSA, as negative controls, did not affect the recovery of unfolded rhodanese. This indicates that the chaperone-like activity was independent of the PPIase activity.

FIG. 2.

Refolding of chemically unfolded rhodanese mediated by PhFKBP29. The refolding reaction was initiated by 60-fold dilution with 50 mM sodium phosphate (pH 7.8) containing 10 mM dithiothreitol, 50 mM sodium thiosulfate, and 0 to 60 μM PhFKBP29 in the presence or absence of 40 μM FK506. GroEL/ES (0.4 μM) was used as the positive control, and HsFKBP12 (5 μM) and BSA (13 μM) were used as negative controls. In the refolding experiment with GroEL/ES, 1 mM MgCl₂, 10 mM KCl, and 1 mM ATP were also included in the dilution buffer. The final concentration of rhodanese was 0.63 μM. The error bars indicate the standard deviation (n = 2 to 4).

Protection by PhFKBP29 of proteins from thermal aggregation in vitro.

Soluble cytoplasmic proteins of E. coli BL21(DE3) were precipitated after being heat-treated at 60 to 100°C (Fig. 3). A densitometric analysis of the bands from SDS-PAGE indicated that, after the 100°C treatment, only 21% of E. coli proteins remained soluble in the absence of PhFKBP29. In the presence of PhFKBP29, however, 63% of the E. coli proteins were soluble after the same treatment. The thermal aggregation of rhodanese at 45°C was also effectively prevented by PhFKBP29 in a dose-dependent manner (Fig. 4). A 30 μM concentration of PhFKBP29 completely prevented the thermal aggregation of 0.8 μM rhodanese, whereas BSA did not have any significant effect on the thermal aggregation (data not shown).

FIG. 3.

Effect of PhFKBP29 on the thermal aggregation of E. coli soluble cytoplasmic proteins in vitro. E. coli cell extract was heated at various temperatures (60 to 100°C) for 30 min with or without PhFKBP29. After centrifugation, the soluble fractions were analyzed by SDS-PAGE.

FIG. 4.

Effect of PhFKBP29 on the thermal aggregation of rhodanese. The thermal aggregation of rhodanese (0.8 μM) was continuously monitored by light scattering at 320 nm in the presence of 0 to 30 μM PhFKBP29.

Soluble expression of anti-HEL Fab fragments by the coexpression of PhFKBP29 in the cytoplasm of E. coli.

Because the mobility of PhFKBP29 and the Fab fragment were identical, as determined by SDS-PAGE, each band was identified by a Western blot with the corresponding antisera. When the anti-HEL Fab fragment was expressed in E. coli JM109(DE3) without the coexpression of PhFKBP29, it was detected only in the insoluble fraction (Fig. 5A, lanes 1 and 2), indicating that the Fab fragment was expressed as an insoluble aggregate in the E. coli cells. On the other hand, with the coexpression of PhFKBP29 (Fig. 5B, lane 3), it was detected only in the soluble fraction (Fig. 5A, lanes 3 and 4). While a densitometric analysis of the Western blot result indicated that the total expressed Fab had been reduced to ca. 30% of that obtained without the coexpression of PhFKBP29 (Fig. 5A, lanes 2 and 3), all expressed Fab was found in the soluble fraction.

FIG. 5.

Effect of the coexpression of PhFKBP29 on the expression of the anti-HEL Fab fragment in E. coli. The soluble fraction (S) and insoluble fraction (I) of the crude extract of the transformants were applied to a reducing SDS-18% polyacrylamide gel. Lanes 1 and 2, E. coli cell lysate in which only Fab gene expression was induced; lanes 3 and 4, both Fab and PhFKBP29 genes were coexpressed. (A and B) Gels immunologically stained with anti-mouse immunoglobulin G (A) and anti-PhFKBP29 (B) polyclonal antibodies. Rabbit anti-PhFKBP29 did not cross-react with insoluble mouse anti-HEL Fab (lane 2).

Antigen recognition of the recombinant anti-HEL Fab fragment.

The soluble fraction of the E. coli lysate containing the anti-HEL Fab fragment with or without coexpressed PhFKBP29 was subjected to an analysis of the HEL-binding activity by ELISA. The soluble fraction of the E. coli lysate without the coexpression of PhFKBP29 (described in Fig. 5A, lane 1) showed only a little HEL-binding activity (Fig. 6). On the other hand, the HEL-binding activity of the soluble fraction of the E. coli lysate with the coexpression of PhFKBP29 (described in Fig. 5A, lane 3) was 7- to 12-fold higher than that without the coexpression of PhFKBP29. No HEL-binding activity was detected in the soluble fraction of the E. coli lysate without the pEHELFab-1 expression vector. These results indicate that the Fab fragment coexpressed with PhFKBP29 was functional.

FIG. 6.

ELISA detection of HEL with the soluble fraction of the E. coli cell lysate containing the recombinant Fab fragment. A microtiter plate was coated with 50 μl of an HEL solution (0 to 100 μg/ml) in Tris-buffered saline at 4°C overnight. The antigen was detected with the soluble fraction of the E. coli cell lysate described in Fig. 5 as a primary antibody in ELISA. Symbols: ▪, Fab gene was expressed with the PhFKBP29 gene; □, only the Fab gene was expressed; ▵, no Fab and PhFKBP29 genes were expressed.

It has been reported that PPIases accelerated the speed of the rate-limiting isomerization of peptidyl-prolyl bonds of an Fab antibody and increased the yield of Fab refolding by reducing aggregation in a nonspecific binding manner in vitro (19). In general, however, the recombinant antibody fragment is expressed as an insoluble aggregate in the cytoplasm of E. coli because the intermolecular disulfide bond is not formed, and it is difficult to achieve a properly folded form in a reducing environment (6). Knappik et al. have reported that the overexpression of E. coli periplasmic PPIase was ineffective for increasing the yield of the soluble form of the Fab antibody fragment in the periplasm of E. coli K-12 (18), suggesting that the acceleration of peptidyl-prolyl cis-trans isomerization did not increase the production of the soluble form of the antibody fragment in vivo.

Although PhFKBP29 showed almost negligible PPIase activity, it had chaperone-like protein folding activity and strongly suppressed thermal protein aggregation in vitro. In the coexpression experiment, PhFKBP29 probably suppressed the irreversible aggregation of refolding anti-HEL Fab by binding to the hydrophobic region of refolding Fab via a mechanism similar to that of archaeal short-type FKBP (13). The chaperone-like folding activity, which is independent of the PPIase activity, may have simultaneously enhanced Fab folding in the cytoplasm of E. coli. Proba et al. succeeded in increasing the expression of the soluble form of the recombinant Fab antibody fragment with a disulfide bond in the cytoplasm of E. coli by using a mutant host E. coli lacking thioredoxin reductase (21). The mechanism for the expression of soluble Fab in this study is probably different from that underlying the enhanced expression of Fab in the mutant E. coli (21). While the cysteine residues of the Fab fragment were probably of the reduced form in the E. coli cytoplasm, they were oxidized to form disulfide bonds during extraction of the protein from the cells under air-oxidative conditions.

PhFKBP29 increased the expression of the properly folded form of the protein in E. coli cells. While the functions of PhFKBP29 in the cell of the hyperthermophile are still not clear, it appeared to contribute to protein folding in the cell. It has been reported that spontaneous isomerization of the peptidyl-prolyl bond is faster at a higher temperature (5, 9). Under high-temperature conditions such as the habitat of P. horikoshii, the chaperone-like activity and protein aggregation suppression function of PhFKBP29 may be more important than the PPIase activity in protein folding in the cell. PhFKBP29 may play the role of a chaperone in archaeal cells.