ABSTRACT
The thermostable alkaline protease from Bacillus stearothermophilus F1 has high potential for industrial applications, and attempt to produce the enzyme in yeast for higher yield was undertaken. Secretory expression of F1 protease through yeast system could improve enzyme’s capability, thus simplifying the purification steps. Mature and full genes of F1 protease were cloned into Pichia pastoris expression vectors (pGAPZαB and pPICZαB) and transformed into P. pastoris strains (GS115 and SMD1168H) via electroporation method. Recombinant F1 protease under regulation constitutive GAP promoter revealed that the highest expression was achieved after 72 h cultivation. While inducible AOX promoter showed that 0.5% (v/v) methanol was the best to induce expression. It was proven that constitutive expression strategy was better than inducible system. The α-secretion signal from the plasmid demonstrated higher secretory expression level of F1 protease as compared to native Open Reading Frame (ORF) in GS115 strain (GE6GS). Production medium YPTD was found to be the best for F1 protease expression with the highest yield of 4.13 U/mL. The protein was expressed as His-tagged fusion protein with a size about 34 kDa.
INTRODUCTION
Proteases are degradative enzymes that catalyze the hydrolysis of proteins. It can be isolated from various microbial sources (Bajaj and Sharma, 2011; Fu et al., 2003; Gohel and Singh, 2012; Purohit and Singh, 2011). Among all types of proteases, alkaline protease is the largest subgroup of serine protease enzymes and highly active at extreme alkaline pH. Thermostable alkaline proteases are highly demanded for high temperature processing thus, resulting faster reaction rates, increasing the solubility of substrate and reducing microbial contaminations (Akel et al., 2009; Purohit and Singh, 2011). In this study, thermostable alkaline protease F1 was isolated from Bacillus stearothermophilus. B. stearothermophilus strain F1 is a thermophilic organism isolated from decomposed oil palm branches by Rahman (1993). The enzyme has an optimum pH value at 9.0 and active at 85℃ (Rahman et al., 1994). In addition, F1 protease is more thermostable than a currently reported protease isolated from salt tolerant alkaliphilic actinomycetes (Gohel and Singh, 2012). Previously, the thermostable F1 protease was successfully cloned and expressed in Escherichia coli expression system extracellularly. A simple heat-treatment method was used to purify the expressed protein (Fu et al., 2003). The lack of secretory pathway in E. coli system has encouraged the researchers to use of bacteriocin release protein (BRP) to facilitate the protein into the culture medium. Secretory expression system offers many advantages over intracellular expression where there would be fewer impurities existed in the protein of interest. Thus, it would simplify the purification steps.
However, E. coli expression system promoted to several disadvantages where the overexpressed proteins are often form as inclusion bodies and inactive. Therefore, denaturation and refolding are required to activate the proteins which are costly and the recovery is usually low (Choi and Lee, 2004). To overcome those bottlenecks, yeast expression system has offered many advantages to express protein of interest extracellularly by using its secretory pathway.
Yeast was found to be a versatile host to express various recombinant proteins. Saccharomyces cerevisiae was proven to be one of the most versatile platforms by creating stable expression of multi-gene pathways (Mikkelsen et al., 2012). Alternatively, methylotrophic yeast namely Pichia pastoris became more favorable for industrial applications when it was proven to regulate foreign protein expression to the highest level. These particular well known yeasts (S. cerevisiae and P. pastoris) and other yeast expression systems such as Hansenula polymorpha, Kluyveromyces lactis, Schizosaccharomyces pombe, Yarrowia lipolytica and Arxula adeninivorans, are able to use different kind of carbon sources to generate high cell density, thus could improve the production of heterologous proteins (Çelik and Çalik, 2012; Kimata et al., 2012). In addition, most yeast expression vectors are designed to be integrated in the genome; therefore, they do not require antibiotic selection during cultivation.
Glyceraldehydes-3-phosphate dehydrogenase (GAP) and alcohol oxidase (AOX) are the most successful promoters in P. pastoris which used to regulate the protein expression constitutively or inducible. Furthermore, the secretory pathway of the cloned gene is headed by α-secretion signal of S. cerevisiae. This study highlights the expression of thermostable alkaline F1 protease in P. pastoris by using different kind of cloning strategies in order to improve the expression level extracellularly.
MATERIALS AND METHODS
Host strains and plasmid
Thermostable F1 protease gene from B. stearothermophilus was amplified from the recombinant plasmid pTricHis-F1 (Fu, 2001). The expression plasmids (pGAPZαB and pPICZαB) and P. pastoris strains [GS115 – (his4, His–) and SMD1168H – (his4 pep4)] were purchased from Invitrogen, USA. E. coli TOP10 [F’ f mcrA, Δ(mrr-hsdRMS-mcrBC), Φ80lacZΔM15, ΔlacX74, deoR, recA1, araD139, Δ(ara-leu)7697, galU, galK, rpsL(StrR), endA1, nupG λ-] were used for subcloning and plasmid propagation (Invitrogen, USA).
Media compositions
Luria Bertani (LB) broth containing 50 μg/mL ampicilin and 35 μg/mL chloramphenicol was used to grow the E. coli strain XL-1 blue/pTricHis-F1. LB broth supplemented with 25 μg/mL Zeocin was used to grow P. pastoris plasmids. The bacterium was cultivated at 37℃ for 16–18 h with 200 rpm agitation. Yeast Peptone Dextrose (YPD) medium containing [YPD: 1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) dextrose] was used to grow the yeast strains at 30℃ with agitation rate of 250 rpm for overnight.
The production media were YPD, BYPD [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (w/v) dextrose, 100 mM potassium phosphate pH 6.0], YPG [1% (w/v) yeast extract, 2% (w/v) peptone, 2% (v/v) glycerol], BMD [1% (w/v) yeast extract, 2% (w/v) peptone, 1% (w/v) dextrose, 100 mM potassium phosphate pH 6.0, 1.34% (w/v/) yeast nitrogen base – with ammonium sulphate and without amino acids, 4 × 10–5% (w/v) biotin], BMDH [1% (w/v) dextrose, 100 mM potassium phosphate pH 6.0, 1.34% (w/v/) yeast nitrogen base – with ammonium sulphate and without amino acids, 4 × 10–5% (w/v) biotin, 0.004% (w/v) histidine] and YPTD [0.5% (w/v) yeast extract, 2% (w/v) peptone, 1% (w/v) dextrose, 4 × 10–5% (w/v) biotin, 0.04% (w/v) tryptic soy broth].
Construction of recombinant plasmid
Open Reading Frame gene primers flanked with XhoI and XbaI restriction sites were designed at forward and reverse primers, respectively. The primers used were FORORF (5’-GTATCTCTCGAGATGAAGTTTAAAGCG-3’) and REVORF (5’-TTAAATCTCTAGAAAATATGTTACAGC-3’). Mature part of F1 protease gene was amplified by using primers incorporated with restriction endonuclease sites, EcoRI for FORPSEQ (5’-GCTGCAGAATTCACTCTAATGATGGGGTA-3’) and XbaI for REVPSEQ (5’-TTAAATCTCTAGAAAATATGTTACAGC-3’) (restriction endonuclease sites are underlined). Both primers had the annealing temperature at 55℃. Cloning of F1 protease genes into the plasmids were conducted according to protocols suggested by EasySelectTM Pichia Expression Kit manual, Invitrogen (USA). N-glycosylation of the recombinant F1 protease was predicted by using NetNGlyc 1.0 Server (http://www.cbs.dtu.dk/services/NetNGlyc/).
Transformation into P. pastoris strains
Transformation into P. pastoris strain was conducted by using electroporation method. Electrocompetent cells were prepared according to protocols suggested from EasySelectTM Pichia Expression Kit manual, Invitrogen (USA) with minor modification. The Pichia strains were cultivated in 10 mL YPD medium for overnight at 30℃ with agitation rate 250 rpm. Then, 200 μL cultures were transferred into 100 mL of fresh YPD until OD600 reached 1.3–1.5. The cells were harvested at 1500 × g at 4℃ for 5 min. The pellets were resuspended gently in an equal volume of sterile, chilled dH2O, and then spun again at the mentioned condition followed by resuspended in 0.5 volume of dH2O. Next, the cells were harvested and resuspended with 5 mL of ice-cold 1 M sorbitol. Finally, they were harvested and resuspended in 400 μL of 1 M sorbitol.
Prior to transformation, the recombinant plasmids were linearized by using AvrII and SacI for pGAPZαB and pPICZαB, respectively. 10 μL (10 μg) of each recombinant plasmid was incubated in a pre-chilled electroporation cuvette prior to transformation into Pichia strains. Electroporation method was conducted according to S. cerevisiae protocols by using an electroporator (Gene Pulser, Biorad, USA). The parameters’ used were: charging voltage of 1500 V, capacitance of 25 μF and resistance of 400 Ω, which generated a pulse length of 5–10 ms with a field strength of ~7500 V/cm. The transformants were subsequently plated on YPDS agar containing 100 μg/mL Zeocin [same composition to YPD, 1 M sorbitol and 2% (w/v) agar]. The plates were incubated until the colonies were formed. The positive transformants were checked via colony PCR by using insert primers.
Screening of F1 protease expression in P. pastoris
Two sets of production media were prepared to screen the expression of recombinant F1 protease in pGAPZαB and pPICZαB. Transformants containing pGAPZαB/F1 plasmid was grown in 10 mL YPD broth until OD600 = 6. Then, the cells were pelleted and resuspended into 3 mL fresh YPD medium to OD600 = 5. The culture was cultivated for 3 days. Then, the cells were harvested and supernatants were used to determine the protease activity.
Transformants containing recombinant pPICZαB/F1 plasmid was grown in 10 mL BMGY [1% (w/v) yeast extract, 2% (w/v) peptone, 100 mM potassium phosphate pH 6.0, 1.34% (w/v) YNB, 4 × 10–5% (w/v) biotin, 1% (v/v) glycerol] at 30℃ with 250 rpm agitation until OD600 = 6. Then the cells were harvested and resuspended into 3 mL BMMY medium 1% (w/v) yeast extract, 2% (w/v) peptone, 100 mM potassium phosphate pH 6.0, 1.34% (w/v) YNB, 4 × 10–5% (w/v) biotin, 0.5% (v/v) methanol] to OD600 = 5. The methanol was added to a final 0.5% (v/v) every 24 h to induce F1 expression in recombinant strains.
Time course study
P. pastoris strains with the highest protease activity were selected and were further cultivated for time course study. Recombinant cultures with pGAPZαB and pPICZαB plasmids in GS115 and SMD1168H strains were cultivated until OD600 = 20 in 50 mL YPD and BMGY media, respectively. Then, the cells were harvested at 5000 × g, 4℃ for 5 min. Finally, the pellets were resuspended into 100 mL YPD and BMMY media to OD600 = 3 for enzyme production. Transformants containing pPICZαB were induced with absolute methanol to final concentration 0.5% (v/v) every 24 h. All the cells were cultivated for 5 days. A 3 mL culture was collected every 24 h and the supernatant was used for protease assay.
Media optimization
Different media compositions were formulated to obtain the optimum F1 protease expression in GE6GS. Initially, the cells were grown in 50 mL YPD medium to OD600 = 20. Then the cultures were harvested at 5000 × g, 4℃ for 5 min prior to resuspension with 100 mL of different production media to achieve OD600 = 3. The cells were cultivated for 3 days and the supernatants were harvested for assay.
Assay of protease activity
The protease activity was determined by using method suggested by Kocabiyik and Erdem (2002) with minor modification. One unit of protease activity on azocasein was defined as the absorbance change of 0.1 per min at 440 nm under the standard assay condition (Blumentals et al., 1990).
Detection of His-tagged fusion protein
His-tagged F1 protease fusion was detected by using SDS-PAGE analysis. 12% gel was prepared according to Laemmli (1976) method. The sample was concentrated with 10% (w/v) trichloroacetic acid (TCA) as described by Fu (2001). Then, the gel was stained with Coomassie brilliant blue and InVision™ His-tag (Invitrogen, USA) for detection.
RESULT AND DISCUSSION
Cloning of F1 protease gene into pGAPZαB and pPICZαB
Mature gene of F1 protease (1160 bp) and ORF F1 protease (1230 bp) were amplified by using the primers flanked with EcoRI - XbaI and XhoI - XbaI restriction sites, respectively. Both amplicons were cloned into the P. pastoris expression vectors (pGAPZαB and pPICZαB). The recombinant plasmids were confirmed via PCR and sequencing (data not shown). The F1 sequence was compared with the available F1 protease gene (AY028615). The cloned gene was fused with the secretion signal and his-tag of the plasmid. Recombinant mature gene was assigned as pGAPZαB/mature and pPICZαB/mature. While recombinant ORF was assigned as pGAPZαB/ORF and pPICZαB/ORF.
In pGAPZαB and pPICZαB vector system, the same α-secretion signal is used. Therefore, the N-glycosylation was predicted for both cloned genes (mature and ORF). NetNGlyc 1.0 Server is the NetNglyc server which used to predict N-glycosylation sites in human proteins using artificial neural networks that examine the sequence context of Asn-Xaa-Ser/Thr sequins (Gupta et al., 2004). The results showed that mature and ORF genes have six potential sites to be glycosylated where they have the reading more than the threshold value 0.5 (Fig. 1). However, only three of them have slightly higher value. The sequences before the gene started were omitted because they were obtained from the α-secretion signal of the vector (first 83 amino acids).
Transformation into P. pastoris strain GS115 and SMD1168H
The recombinant plasmids were transformed into P. pastoris genome via electroporation method. The plasmids were designed to be integrated inside the yeast genome at AOX or GAP promoter. Both plasmids did not have the P. pastoris-specific autonomous replication sequence (PARS) to maintain the plasmid in yeast genome. Furthermore, this episomal feature of plasmid was found to be unstable as compared to integrative vectors (Higgins and Cregg, 1998). Integration could occur at the homologous promoter sequence or at auxotrophic marker. The recombinant plasmids containing F1 protease gene were checked via colony PCR.
Time course study
The effect of using a constitutive and inducible promoters were studied for both F1 protease genes (mature and ORF) in different host strains. Cereghino and Cregg (2000) reported that P. pastoris expression system was able to produce heterologous proteins range from 10–100 fold higher than their native hosts. The positive transformants from PCR analysis were cultivated and tested for protease expression by using method described by previous study (Fu, 2001; Kocabiyik and Erdem, 2002). The enzyme was pre-incubated at 70℃ prior to assay in order to activate the F1 protease and eliminate other native proteins from the host. TCA was added to stop the reaction by precipitating the uncatalyzed azocasein.
Table 1 shows the expression of F1 protease from each transformant. The result revealed that recombinant genes in plasmid pGAPZαB showed higher protease activity as compared to pPICZαB. However, longer time was taken when using constitutive GAP system (72 h). While 48 h for by using AOX promoter regulation. The production rate for all recombinant strains was within 0.02 to 0.03 U/mL/h. Most of the recombinant from protease deficient strain (SMD1168H) showed lower expression level than GS115 strain.
Table 1.Expression of F1 protease in different secretory
P. pastoris system
| Colony | Recombinant plasmid | Protease activity (U/mL)e |
|---|
| GE10SMa | pGAPZαB/F1 gene without signal sequencec | 2.083±0.05 |
| GE6GSb | | 2.208±0.05 |
| PE56SMa | pPICZαB/F1 gene without signal sequenced | 1.361±0.05 |
| PE16GSb | | 1.417±0.05 |
| GX7SMa | pGAPZαB/F1 ORF genec | 1.458±0.05 |
| GX17GSb | | 1.458±0.05 |
| PX57SMa | pPICZαB/F1 ORF gened | 1.000±0.05 |
| PX20GSb | | 1.125±0.05 |
P. pastoris SMD1168Ha and P. pastoris GS115b
The protease activity was taken from the highest peak at 72 hc and 48 hd.
e Each recombinant strain was monitored at the optimum time. ± indicates the 5% confidence interval.
In comparison to protein secretory, the used of α-secretion signal from the plasmid which is originated from S. cerevisiae was more efficient as compared to native secretion signal from B. stearothermophilus (ORF gene). Prokaryotic signal peptide did not work efficiently in P. pastoris because the host preferred to use eukaryotic signal. In this case, double secretion signals were found in recombinant plasmid carrying ORF gene (GX7SM, GX17GS, PX57SM and PX20GS). From the screening, recombinant mature F1 protease in GS115 strain with constitutive expression (GE6GS) was chosen for optimization study.
Media optimization
Recombinant GE6GS was selected for optimization study where it was cultivated in various media composition containing different nitrogen and carbon sources to be used in protein expression. BMD, BMDH and YPTD were modified from BMGY, BMGH and YPTG media, respectively. Glycerol in each medium was replaced with dextrose. Glycerol was used to generate cell mass when using inducible AOX1 expression system. No inducer was needed for GAP expression, therefore, dextrose was chosen as the carbon source. Buffered medium was used to control the pH of the culture (EasySelectTMPichia Expression Kit manual). In addition, when the pH was being controlled, native protease expression from the host was decreased. BYPD was used by Li et al. (2001) to express the antifreeze protein. YPG was modified from YPD medium to see the effect of glycerol on F1 protease expression under regulation GAP promoter. Sabri et al. (2009) suggested that YPTM medium was the best medium for L2 lipase production under the control of AOX promoter. Therefore, YPTD was generated by replacing the 0.5% (v/v) methanol with 1% (w/v) dextrose.
Figure 2 shows that YPTD medium exhibited the highest F1 protease expression as compared to other suggested media. Tryptic soy broth (TSB) was used as nitrogen source instead of yeast nitrogen base (YNB). TSB was much more economical as compared to YNB. On the other hand, YPG showed lower expression level as compared to YPD where the glycerol was not preferable for GAP regulation system. Furthermore, BMDH could not be used as production medium when there were no peptone and yeast extract to generate mass and energy. Yeast extract could increase the recombinant protein secretion and accumulation (Sreekrishna et al., 1997).
According to Fu et al. (2003) the enzyme activity was calculated as the total activity of 500 mL (11,000 U) that equal to 22 U/mL activities (result not shown). Nevertheless, the actual U/mL was 2.2 U/mL by comparing to the one unit definition used in current study (0.1/min/mL). Therefore, the F1 protease expression in P. pastoris has achieved 4–5 U/mL indicating 2-fold higher than E. coli.
Confirmation of F1 protease expression
The expression of F1 protease in GE6GS was confirmed via SDS-PAGE analysis and stained with InVisionTM His-tag in gel. The result showed that about 34 kDa protein band was observed in Fig. 3. The staining with InVisionTM His-tag in gel could detect the His-tagged fusion proteins (Invitrogen, USA). The empty pGAPZαB plasmid showed no band on the gel. The poly His-tag was located at the C-terminal of pGAPZαB. This particular staining strategy could reduce the time and cost to detect the recombinant His-tagged protein. Western blot could be a tedious and costly method to detect the recombinant protein in GE6GS. The size of mature F1 protease in E. coli XL1-Blue expression was 27 kDa. Bigger protein band was detected from P. pastoris expression system could be due to post-translation modification which occurred within the protein. This result was supported by previous N-glycosylation prediction which showed that the F1 gene could have glycosylations.
Conclusion
Thermostable alkaline F1 protease from B. stearothermophilus was successfully expressed in P. pastoris expression system extracellularly. The result proved that mature F1 protease exhibited better expression level as compared to ORF gene. On the other hand, constitutive GAP promoter could offer higher expression level than the inducible promoter (AOX). Effect of the medium composition revealed that YPTD could express the gene optimally in GE6GS. GS115 strain showed significantly higher expression as compared to SMD1168H with 34 kDa of protein size.
ACKNOWLEDGMENT
This project was supported financially by the Malaysia Genome Institute project number 07-05-MGI-GMB003 from the Ministry of Science, Technology and Innovation, Malaysia.
REFERENCES
- Akel, H., Al-Quadan, F., and Yousef, T. K. (2009) Characterization of a purified thermostable protease from hyperthermophilic Bacillus strain HUTBS71. Eur. J. Sci. Res. 31, 280–288.
- Bajaj, B. K., and Sharma, P. (2011) An alkali-thermotolerant extracellular protease from a newly isolated Streptomyces sp. DP2. New Biotechnol. 28, 725–732.
- Blumentals, I. I., Robinson, A. S., and Kelly, R. M. (1990) Characterization of sodium dodecyl sulphate resistant proteolytic activity in the hyperthermophilic archaebacterium Pyrococcus furiosus. Appl. Environ. Microbiol. 56, 1992–1998.
- Çelik, E., and Çalik, P. (2012) Production of recombinant proteins by yeast cells. Biotechnol. Adv. 30, 1108–1118.
- Cereghino, J. L., and Cregg, J. M. (2000) Heterologous protein expression in the methylotrophic yeast Pichia pastoris. FEMS Microbiol. Rev. 24, 45–66.
- Choi, J. H., and Lee, S. Y. (2004) Secretory and extracellular production of recombinant proteins using Escherichia coli: Mini-review. Appl. Microbiol. Biotechnol. 64, 625–635.
- Fu, Z. (2001) Cloning, sequencing and extracellular expression of alkaline protease gene from Bacillus stearothermophilus F1. Masters thesis, Universiti Putra Malaysia.
- Fu, Z., Hamid, S., Razak, C. N., Basri, M., Salleh, A. B., and Rahman, R. N. Z. R. A. (2003) Secretory expression in Escherichia coli and single-step purification of a heat-stable alkaline protease. Protein Expr. Purif. 28, 63–68.
- Gohel, S. D., and Singh, S. P. (2012) Purification strategies, characteristics and thermodynamic analysis of a highly thermostable alkaline protease from a salt-tolerant alkaliphilic actinomycete, Nocardiopsis alba OK-5. J. Chromatogr. B 889–890, 61–68.
- Gupta, R., Jung, E., and Brunak, S. (2004) Prediction of N-glycosylation sites in human proteins.
- Higgins, D. R., and Cregg, J. M. (1998) Introduction to Pichia pastoris. In: Methods in Molecular Biology: Pichia Protocols (eds.: D. R. Higgins, and J. M. Cregg). Humana Press, Totowa, New Jersey.
- Kimata, K., Yamaguchi, M., Saito, Y., Hata, H., Miyake, K., Yamane, T., Nakagawa, Y., Yano, A., Ito, K., and Kawarasaki, Y. (2012) High cell-density expression system: A novel method for extracellular production of difficult-to-express proteins. J. Biosci. Bioeng. 113, 154–159.
- Kocabiyik, S., and Erdem, B. (2002) Intracellular alkaline proteases produced by thermoacidophiles: detection of protease heterogeneity by gelatin zymography and polymerase chain reaction (PCR). Bioresource Technol. 84, 29–33.
- Laemmli, U. (1970) Most commonly used discontinuous buffer system for SDS electrophoresis. Nature 227, 680–686.
- Li, Z., Xiong, F., Lin, Q., d’Anjou, M., Daugulis, A. J., Yang, D. S. C., and Hew, C. L. (2001) Low temperature increases the yeild of biologically active herring antifreeze protein in Pichia pastoris. Protein Expr. Purif. 21, 438–445.
- Mikkelsen, M. D., Buron, L. D., Salomonsen, B., Olsen, C. E., Hansen, B. G., Mortensen, U. H., and Halkier, B. A. (2012) Microbial production of indolylglucosinolate through engineering of a multi-gene pathway in a versatile yeast expression platform. Metab. Eng. 14, 104–111.
- Purohit, M. K., and Singh, S. P. (2011) Comparative analysis of enzymatic stability and amino acid sequences of thermostable alkaline proteases from two haloalkaliphilic bacteria isolated from Coastal region of Gujarat, India. Int. J. Biol. Macromol. 49, 103–112.
- Rahman, R. N. Z. R. A. (1993) Production and characterization of thermostable protease from Bacillus stearothermophilus strain F1. Masters thesis, Universiti Pertanian Malaysia.
- Rahman, R. N. Z. A., Razak, C. N., Ampon, K., Basri, M., Yunus, W. M. Z. W., and Salleh, A. B. (1994) Purification and characterization of a heat-stable alkaline protease from Bacillus stearothermophilus F1. Appl. Microbiol. Biotechnol. 40, 822–827.
- Sabri, S., Rahman, R. N. Z. R. A., Leow, T. C., Basri, M., and Salleh, A. B. (2009) Secretory expression and characterization of a highly Ca2+-activated thermostable L2 lipase. Protein Expr. Purif. 68, 161–166.
- Sreekrishna, K., Brankamp, R. G., Kropp, K. E., Blankenship, D. T., Tsay, J.-T., Smith, P. L., Wierschke, J. D., Subramaniam, A., and Birkenberger, L. A. (1997) Strategies for optimal synthesis and secretion of heterologous proteins in the methylotrophic yeast Pichia pastoris. Gene 190, 55–62.
