Improving the stability of a foreign protein in the periplasmic space of Escherichia coli

Biochimie 70 (1988) 727-733 (~) Soci6t6de Chimie biologique/Elsevier, Paris

727

R e s e a r c h article

Improving the stability of a foreign protein in the periplasmic space of Escherichia coli Jamila A N B A , Alain B E R N A D A C , Claude L A Z D U N S K I and Jean-Marie PAGI~S Centre de Biochimie et de Biologie Moldculaire du CNRS, 31, Chemin Joseph-Aiguier, B.P. 71, 13402 Marseille Cddex 9, France (Received 31-8-1987, acceptedafter revision 10-12-1987)

Summary - - An efficient expression / export vector comprising the entire phoS (phosphate binding protein) gene fused to a synthetic gene encoding the human growth hormone releasing factor (mhGRF) has recently been constructed [ 1]. The hybrid protein ( P h o S - m h G R F ) was exported to the periplasmic space. However, in this location proteolytic degradation occurred at the C-terminal region. Phenylmethylsulfonyl fluoride (PMSF) increased the stability of the hybrid protein indicating that a serine protease may be involved in the proteolytic cleavage. The correct export and subsequent degradation of the recombinant protein in the periplasmic space were demonstrated in situ using double immunogold labeling on ultrathin sections. Using a phoS-based expression / export vector in the presence of PMSF, 2 - 4 mg of hybrid protein per liter of culture could be obtained.

fusion protein/human growth hormone releasingfactor/periplasmic protein/protein export/double immunogold labeling

Introduction Because the knowledge of Escherichia coli genetics is well-advanced, in recent years this organism has been used extensively to express foreign proteins of potential biotechnological interest. These proteins can be exported and bacterial, eukaryotic or hybrid signal sequences can be used in order to direct foreign proteins out of the cytoplasm (for recent reviews, see [2-4]). We have previously reported the construcr~on of an expression/secretion vector allowing the production of a hybrid protein, the periplasmic phosphate binding protein (PhoS) fused to the human growth hormone releasing factor (hGRF) [1]. In cells grown in phosphate-limiting medium, this hybrid was produced as a fusion protein of 370 amino acid residues comprising the total sequence of mature PhoS (321 amino acids), the m h G R F (44 amino acids) and the linker peptide (5 amino acids). It is exported to the periplasmic space where as much as 104 molecules / cell (700/z g / 1 ) can accumulate.

However, some proteolytic degradation of the hybrid protein occurred in this cell compartment as previously demonstrated [1]. In this paper, we report molecular and electron microscopic studies aimed at improving the stability of the hybrid protein ( P h o S - m h G R F ) in the periplasmic space.

Materials and methods Bacterial strains, plasmid and culture media E. coli strains ANCC75 (F , leu, purE, trp, his, argG, rpsL, met, thi, phoS64) and AC42 (F-, gal, lac, rpsL, A[recC-thyA] 238) were previously described [5, 6]. The plasmid pAA1, which carries the hybrid gene phoS-mhGRF has recently been described I1]. ANCC75 or AC42 were transformed by pAAt according to Lederberg and Cohen [7]. Tris/glucose medium, a minimal salt solution buffered with Tris/HCI at pH 7.2 and containing 0.2% glucose supplemented with either 0.64 mM KH2PO4 (referred to as TGHP) or 0.064 mM KH2 PO4 (referred to as TGLP), was used as previously reported [8]. These media were supplemented with required nutrients as well as tetracycline (10/~g/ml). When

728

J. Anba et al.

mentioned, protease inhibitors were added during growth in TGLP at 1 mM for phenylmethylsulfonyl fluoride (PMSF), 1 mM for aprotinin or 10 mM for ethylenediaminetetraacetate (EDTA).

Expression of the hybrid protein The induction of P h o S - m h G R F synthesis was carried out as previously described for the PhoS protein [9]. During growth in TGLP, various samples were removed at intervals to analyze the cell content in hybrid protein either by polyacrylamide gel electrophoresis (PAGE) or by electron microscopy.

Labeling of ceihdar proteins and cell fractionation procedure 3 h after transfer to low phosphate medium, pulse-chase experiments were carried out as previously described [10]. Cells (2.5 x 10'J) were pulselabeled for l min with [35S]methionine (100 pCi / ml) and chased with unlabeled methionine (70 mM) for 1 h. When mentioned, 1 mM PMSF, 1 mM aprotinin or 10 mM EDTA was added during the pulse or only during the chase. Aliquots of cells were removed at various chase times and were solubilized prior to immunoprecipitation [9, 10]. The immunoprecipitates were then analyzed by S D S - P A G E and fluorography. Cell fractionations (2.5 x 10. cells) were performed 3 h after transfer to TGLP. After rapia centrifugation, the cell pellets were washed with 200 mM Tris/HCI, pH 8.0, and converted into spheroplasts via lysozyme / E D T A treatment [9]. The proteins of the various compartments were directly analyzed by electrophoresis in S D S - PAGE. hnmunoprecipitations, western-blotting and S D S PAGE S D S - P A G E analyses, immunoprecipitations, fluorography and densitometer scannings were carried out as previously described [10]. Western blotting and immunostaining with various antisera were performed according to PagEs et al. [11]. immunolabeling of proteins on ultrathin sections of fi'ozen cells The detailed procedure for fixation, sectioning, immunolabeling and staining has been previously described [12]. lmmuno-gold labeling of mhGRF and PhoS on ultrathin sections was carried out according to Geuze et al. [13]. The average diameters of small and large gold particules were 6 and 13 nm respectively. Antisera The antiserum directed against the PhoS protein was a generous gift from Dr. H. Shinagawa [8]. The antisera specific to complete hGRF or directed against synthetic amino acid 10-15 and amino acid 4 0 - 4 4 peptides of hGRF were prepared by F.-X. Coudr, Sanofi- Elf Biorecherches.

Results

Evidence for COOH-terminal degradation of the PhoS-mhGRF hybrid protein in the periplasmic space Synthesis and export of P h o S - m h G R F to the periplasmic space have recently b e e n r e p o r t e d . In this location, in addition to the m a t u r e f o r m , products corresponding to d e g r a d e d forms were d e t e c t e d by p u l s e - c h a s e e x p e r i m e n t s [1]. In the work p r e s e n t e d here, the proteolytic products were f u r t h e r characterized by immunological analysis after electrotransfer. Various antisera were used: 1) an antiserum directed against PhoS [8] which can recognize all the hybrid forms; 2) an antiserum directed against the complete G R F peptide; 3) an antiserum directed against a synthetic peptide comprising amino acids 1 0 - 1 5 of G R F ; and 4) an antiserum specific to amino acids 4 0 - 4 4 o f G R F which can recognize only the intact C O O H - t e r m i n u s o f the hybrid protein [14]. With these probes, the presence o f the various forms was confirmed both in total cells (TC) and the periplasmic fluid ( P E ) (Fig. l), after 3 h of growth in phosphate-limit-

TC PE TC PE ~,~!

d

A

d

0

2

~ 4

E

B

C

6~ 4- 0

2

....

4

ll~llpJb

6

lmll==~

F

Fig. l. Identification of the various forms of PhoS-mhGRF. At intervals after transfer to low phosphate medium as previously reported [1], cell samples of ANCC75 (pAAI) [1,5] were removed and analyzed by SDS-PAGE (E, F) according to Pages et al. [il]. The times of sampling were 0, 2, 4 and 6 h (lanes 0-6. respectively). After 3 h of growth at 37"C in phosphate limiting medium, cell fractionation was carried out as previously reported [ 1,9] and proteins from total cells (TC), and periplasm (PE) were analyzed by SDS-PAGE. The immunoblots obtained with an antiserum directed against PhoS are shown in A and E, with an antiserum directed against complete GRF in B and F, with an antiserum specific to amino acids 10-15 of GRF in C and with an antiserum specific to amino acids 40-44 of GRF in D, p, m and d indicate the migration of precursor, mature and degraded forms of PhoS- mhGRF, respectively.

Stabilization of a foreign protein in the periplasmic space of E. coli ing medium to induce the synthesis of the fusion protein [1]. Only the mature and degraded forms of P h o S - m h G R F were detected in the periplasmic fraction with the antiserum specific to PhoS (Fig. 1A lane PE). The antiserum directed against complete GRF recognized the precursor and mature forms (Fig. 1B). Several products were detected with the antiserum specific to amino acids 10-15 of G R F (Fig. 1C). The most abundant degraded form present in the periplasm was recognized indicating the presence of the sequence 10-15 of GRF in this proteolytic product. Only the precursor and intact forms of P h o S - m h G R F were recognized by the antiserum directed against amino acids 40-44 of the G R F (Fig. 1D). Thus, the degradation occurred at the COOH-terminus of the hybrid protein. The apparent molecular weights of the 3 forms of the protein (precursor, mature and degraded) were 41 000, 39 000 and 37 000 Da, respectively, in agreement with those expected in this hypothesis. During the induction, the various forms of the hybrid protein were detected in total cells by the antiserum specific to PhoS (Fig. 1E). Under the same conditions, the precursor and mature forms of P h o S - m h G R F were easily observed with the antiserum specific to complete GRF, while the degraded proteins reacted very weakly (Fig. 1F).

A B C Di

729

F

oL.I

0

pmd pmd pmd pmd pmd

Fig. 2. Effect of protease inhibitors and of the ptr deletion on the stability of the P h o S - m h G R F fusion protein. 3 h after transfer to low phosphate medium, cells of ANCC75 (pAAt) (2 × 10~) were pulse-labeled for 1 min with [3sSImethionine and chased with unlabeled methionine in the absence (A) or in the presence of 1 mM aprotinin (B), l0 mM E D T A (C), or l mM PMSF (D). In E, the pulse was also performed in the presence of PMSF. 3 h after transfer to low phosphate medium, cells of AC42 (pAAI) (F) were pulse-labeled for 1 min with [35S]methionine and chased with unlabeled methionine at 37°C. At various chase times (5, 30 or 45 min, top to bottom), aliquots were removed, solubilized and immunoprecipitated with the antiserum directed against PhoS as previously described [10]. The immunoprecipitates were analyzed by S D S - P A G E . The percentages of precursor (p), mature (m) and degraded (d) forms of the hybrid protein were determined by densitometer scan-

Stabilization of the P h o S - m h G R F hybrid protein The effect of various protease inhibitors on the proteolytic process described above was assayed to determine whether or not the stability of the periplasmic P h o S - m h G R F could be increased. The experimental approach was to compare the kinetics of degradation of pulse-labeled hybrid protein in the presence of 10 mM EDTA, which inhibits metalloproteases, 1 mM aprotinin or 1 mM PMSF both of which are serine protease inhibitors. Since the low molecular mass of PMSF (174 Mr) allows rapid diffusion into the cell, it was added either during the pulse or only for the chase. The only methionine residue present in the mature P h o S - m h G R F is that inserted between PhoS and G R F [1] which allowed us to evaluate the stability of the hybrid by [ass]methionine labeling. The results are presented in Fig. 2. Aprotinin and E D T A had no detectable effect on the kinetics of proteolysis (Fig. 2B, C). In contrast, PMSF was found to stabilize

the hybrid protein (Fig. 2D, E). The percentage of inta,vt P h o S - m h G R F was increased from nearly 12% after 45 rain of chase in the control cells to 50% when PMSF was added during the pulse (Fig. 2A, E). Aprotinin, which cannot diffuse into the cells due to its molecular mass (6500 Mr in monomer form), had no protective effect even at a concentration of 10 mM. Since the addition of PMSF had some adverse effect on protein synthesis and cell growth, a protocol was designed in which the protease inhibitor was added only 1 h after transfer into phosphatelimiting medium. In the presence of the inhibitor, intact P h o S - m h G R F (39 000 Mr) was detected after 2 h of growth in low-phosphate medium. The hybrid protein appeared as a major cell protein in the presence of PMSF (Fig. 3). The production of the elongation factor EF-Tu is stable during cell growth and can be

J. Anba et al.

730

0 1 2 3 4 5 6

type strain (ANCC75) to that in a strain with a

ptr deletion (E. coli K12 AC42) that does not produce protease III [6]. No significant difference was found between these two strains, demonstrating that this endopeptidase is not involved in the degradation of the P h o S m h G R F hybrid protein (Fig. 2F).

Direct evidence for export to the periplasm provided by immunolabeling of P h o S mhGRF on ultrathin sections of frozen cells

Fig. 3. Stabilization ol the h xbrid protein. I I1 0 a n t ()) ~Lftcr transfer into phosphate limiting medium, I mM PMSF was added to cells of ANCC75 ( p A A 0 . Cell samples were removed at interv'als and proteins were analyzed by S D S PAGE. The sampling times were 0, I, 2, 3, 4, 5 and 6 h of growth (lanes 0 - 6 , respectively) in the presence of PMSF. The arrow (m) and the asterisk indicate the migration of intact P h o S - m h G R F . Molecular weight standards ( a - f ) are indicated in kDa a: 94; b: 67; c: 46; d: 36; e: 30; f: 20.

used as a reliable standard [15, 16]. The hybrid protein / EF-Tu ratio was evaluated from densitometer scanning of Coomassie blue-stained gels [11]. We estimated that the production of PhoS - m h G R F in the presence of PMSF reached about 5 × 104 molecules / cell which corresponds to 2 - 4 mg / 1 of culture. This result agrees with the evaluation obtained by reference to standard amounts of PhoS loaded in the same gel. When compared to similar experiments in the absence of the inhibitor [1], this represented an increase of 4-5-fold in the yield of the fusion protein.

Stability of the hybrid protein in a mutant deficient in the major periplasmic protease Protease III (product ofptr gene) is a 110 000 Mr endopeptidase located in the periplasmic space !6]. It has been shown to have proteolytic activnty on a limited number of small peptides of less than 6000 Mr [17]. This protease might be involved in the degradation of the periplasmic hybrid protein, since the G R F portion of P h o S m h G R F is specifically digested. We have compared the stability of the hyb.rid protein in a wild

The hybrid protein was visualized 3.5 h after transfer to low phosphate medium in the presence of PMSF. lmmunodetection was performed using immuno-gold labeling of cryosections with two different antisera directed against G R F according to Geuze et al. [13]. In control cells (Fig. 4a, c) which do not carry the plasmid pAAI encoding for the phoS-mhGRF gene, only a very few gold particles corresponding to the background of non-specific labeling were observed. In strains harboring pAAx, P h o S m h G R F was detected predominantly in the periplasmic space of the cells (Fig. 4b) using an antiserum directed against complete GRF. This protein was also recognized by an antiserum specific to amino acids 4 0 - 4 4 of G R F (Fig. 4d). In an attempt to directly demonstrate the degradation of the hybrid protein in the periplasmic space, double immuno-gold labelings were performed. Cryosections were first labeled with an antie ~ r n m

dir~ot~A

~croinef th~

~arnnl.ta

~DII~

anA

stained with 6 nm protein A - g o l d particles. A second run of immunorecognition was carried out with an antiserum directed against PhoS and then 13 nm protein A - g o l d particles. Large and small particles were detected in this process (Fig. 4e). However, when the first incubation was performed with the antiserum specific to PhoS, an intense labeling was obtained, whereas no significant staining was observed with the second antiserum specific to G R F (data not shown). These results suggest that the antigenic determinants of G R F were not accessible in this case. Thus, the presence of small and large gold particles in Fig. 4e indicates the periplasmic location of intact and degraded forms of the hybrid protein. The different affinities of the antibodies directed against PhoS and those directed against G R F cannot explain the differential labeling of the hybrid protein because both antisera were used at saturating conditions during the immunorecognition assay. In a second approach (Fig. 4f), a first labeling was

Stabilization of a foreign protein in the periplasmic space of E. coli

731

I|L

Fig. 4. Immunocytochemical localization of PhoS-mI~GRF hybrid protein in ultrathin cryosections. Cell samples of ANCC75 (a, c) or ANCC75 (pAA0 (b, d, e, 15 were removed 3.5 h after transfer to low phosphate medium. Ultrathin sections were prepared as described [12]. Single labelings were performed with antisera directed against complete GRF or against amino acids 40-44 of GRF (a, b and c, d, respectively). Two doable immuno-gold labelings were carried out: (e) the antiserum directed against complete GRF was first applied and then labeled with 6 nm protein A - g o l d particles, and the antiserum specific to PhoS and 13 nm protein A - g o l d particle conjugate were subsequently added; (15 immunorecognition with the antiserum directed against amino acids 40-44 of GRF (labeled by small gold particles) was followed by immunodetection with the antiserum specific to the complete GRF (labeled by large gold particles).

performed with an antiserum directed against amino acids 40-44 of GRF (small gold particles) and a second run was carried out with the antiserum specific to the total GRF (large gold particles). Associated double labeling was not clearly observed, presumably due to steric hindrance induced by the binding of the first immuno-gold complex to the antigen.

Discussion We have previously reported the construction of a hybrid protein. P h o S - m h G R F , which is synthesized in E. coli [I]. This hybrid is correctly exported to the periplasmic space. We present here evidence that a degradation process concerning the COOH-terminus of the hybrid

732

J. A n b a et al.

occurred after localization in the periplasm. With the antiserum directed against the last 5 amino acids of GRF, the intact forms were detected either in periplasmic, membrane or cytoplasmic fractions. The membrane and cytoplasmic associated forms have been characterized as precursors of the hybrid. These two pools of precursor containing the intact COOH-terminal region of G R F are accumulated during overproduction of P h o S - m h G R F due to saturation of export sites as previously observed during PhoS overproduction [9]. The presence of intact and degraded forms of P h o S - m h G R F in the periplasmic space of E. coli was also demonstrated by double immuno-gold labeling which indicates the presence of two forms in this compartment. This data clearly demonstrates that periplasmic degradation occurs in vivo and not during fractionation procedures. We have previously demonstrated that the cytoplasmic precursor of PhoS cannot be posttranslationally exported [9]. The kinetics of maturation of the hybrid precursor [1] and the localization of various forms presented here indicate that a protease cleaves at the C-terminus region of P h o S - m h G R F , degrading about 2 / 3 of the G R F sequence. The periplasmic degraded form still contains the 1 0 - 1 5 peptide, as indicated by immunoblotting, and the PhoS core of the protein, which is resistant to degradation. This result is consistent with the report [18] showing the resistance of periplasmic PhoS to trypsin, and suggests that the NH2-terminal i(~lldtli Ol UI[~F

IIIglL~/ O~, DIOLI~;I.LI~V~ILI l l ) I l l

ldl~ltlUtl-

tion in the hybrid protein. Various proteases located in the envelope raay cause proteolysis of foreign proteins [2]. The cytoplasmic product of the Ion gene, protease La, has an important role in the degradation of heterologous proteins [19]. However, we failed to detect any stabilization of the P b o S - m h G R F in a Ion mutant or in cultures grown at low temperature (25oC) (data not shown). Among the various protease inhibitors tested, only PMSF caused a significant increase in the half-life of the hybrid protein, indicating that serine proteases could be involved in this case. Proteases such as these might be located on the periplasmic face of the inner membrane, on the inner face of the outer membrane or in the periplasm. Although we have no evidence pointing to a specific peptidase, we have shown that periplasmic protease III is not responsible for this degradation. To conclude, we have demonstrated that the degradation of the hybrid P h o S - m h G R F pro-

tein occurring in the periplasmic space involves the COOH-terminal region of the G R F part. Substantial stabilization of this hybrid protein can be achieved in the presence of PMSF and the yield of P h o S - m h G R F is improved by 4 - 5 fold, resulting in a production of 2 - 4 mg / liter. It seems reasonable to anticipate that the treatments described here should improve the stability of foreign proteins in general when they are exported to the periplasmic space in E. coli cells. Moreover, our results suggest that double immn- ,-gold labeling may be useful to characteri,, in situ the various fo. ; of a protein.

Acknowledgments We thank Dr. H. Shinagawa, F.-X. Coud6 and S. Kushner for generous gifts of antisera and strains. We are grateful to S. P. Howard for careful reading and to M. Payan for preparing the manuscript. This work was supported by the Centre National de la Recherche Scientifique, the lnstitut National de la Sant6 et de la Recherche Mrdicale (C.R.E. n ° 86.1020) and the Fondation pour la Recherche Mrdicale.

References 1 Anba J., Baty D., Lloubes R., Pagrs J.-M., Joseph-Liauzun E., Shire D., Roskam W. & Lazdunski C. (i987) Gene 53, 2i9-226 2 Martson F. A. O. (1986) Biochem. J. 240, 1-12 3 Nicaud J. M., Mackman N. & Holland I. B. (1986) J. Biotechnol. 3,255-270 4 Pugsley A . P . & Schwartz M. (1985) FEMS Microbiol. Rev. 32, 3-38 5 Amemura M., Shinagawa H., Makino K., Otsuji N. & Nakata A. (1982) J. Bacteriol. 152,692-701 6 Dykstra C. C. & Kushner S. R. (1985) J. Bacteriol. 163, 1055-1059 7 Lederberg E. M. & Cohen S. N. (1974) J. Bacteriol. 119, 1072-1074 8 Morita T., Amemura M., Makino K., Shinagawa H., Magota K., Otsuji N. & Nakata A. (1983) Eur. J. Biochem. 130, 427-435 9 Pag/~s J.-M., Anba J., Bernadac A., Shinagawa H., Nakata A. & Lazdunski C. (1984) Eur. J. Biochern. 143,499-505 10 Anba J., Lazdunski C. & Pages J.-M. (1986) J. Gen. MicrobioL 132,689-696 11 Pages J.-M., Anba J. & Lazdunski C. (1987) J. Bacteriol. 169, 1386-1390 12 Anba J., Bernadac A., Pages J.-M. & Lazdunski C. (1984) Biol. Cell 50, 273-278

Stabilization of a foreign protein in the peripiasmic space o f E. coli 13 Geuze H. J., Slot J. W., Van Der Ley P. A. & Scheffer R. C.T. (1981) J. Cell Biol. 89, 653-655 14 Pages J.-M., Belaich A., Anba J. & Lazdunski C. (1987) Eur. J. Biochem. 168, 239-243 15 Boyd A. & Holland I. B. (1979) Cell 18,287-296 16 Anba J., Lazdunski C. & Pag/:s J.-M. (1986)

733

FEBS Lett. 196, 9-13 17 Cheng Y. S. E. & Zipser D. (1979)J. Biol. Chem. 254, 4698-4706 18 Anba J., Pag/~s J.-M. & Lazdunski C. (1986) FEMS Microbioi. Lett. 34, 215-219 19 Goff S. A. & Goldberg A. L. (1985) Cell 41, 587-595