Journal of Bioscience and Bioengineering VOL. 120 No. 2, 187e192, 2015 www.elsevier.com/locate/jbiosc
Phase analysis in single-chain variable fragment production by recombinant Pichia pastoris based on proteomics combined with multivariate statistics Yuya Fujiki, Yoichi Kumada, and Michimasa Kishimoto* Department of Chemistry and Materials, Kyoto Institute of Technology, Goshokaidocho, Sakyoku, Kyoto 606-8585, Japan1 and Department of Bio-molecular Engineering, Kyoto Institute of Technology, Goshokaidocho, Sakyoku, Kyoto 606-8585, Japan2 Received 26 August 2014; accepted 15 December 2014 Available online 28 January 2015
The proteomics technique, which consists of two-dimensional gel electrophoresis (2-DE), peptide mass fingerprinting (PMF), gel image analysis, and multivariate statistics, was applied to the phase analysis of a fed-batch culture for the production of a single-chain variable fragment (scFv) of an anti-C-reactive protein (CRP) antibody by Pichia pastoris. The time courses of the fed-batch culture were separated into three distinct phases: the growth phase of the batch process, the growth phase of the fed-batch process, and the production phase of the fed-batch process. Multivariate statistical analysis using 2-DE gel image analysis data clearly showed the change in the culture phase and provided information concerning the protein expression, which suggested a metabolic change related to cell growth and production during the fed-batch culture. Furthermore, specific proteins, such as alcohol oxidase, which is strongly related to scFv expression, and proteinase A, which could biodegrade scFv in the latter phases of production, were identified via the PMF method. The proteomics technique provided valuable information about the effect of the methanol concentration on scFv production. Ó 2014, The Society for Biotechnology, Japan. All rights reserved. [Key words: Phase analysis; Single-chain variable fragment; Pichia pastoris; Proteomics; Cancerd2; Two-dimensional gel electrophoresis; Peptide mass fingerprinting]
Two-dimensional electrophoresis (2-DE) is a valuable tool for precise protein separation and can be effectively applied to proteome analysis based on peptide mass fingerprinting (PMF) using MALDI-TOF MS (1). In previous work, proteome analysis with the aid of artificial neural networks was successfully applied to recognize the sufficiency of chemical elements based on the protein spots resolved in 2-DE (2). The recognition of culture phases in a culture process would be important if the determination of different culture operations in each phase had a critical advantage for the improvement of bioproduction (3,4). We have developed a novel image analysis method combined with recently developed powerful computing hardware, in which the time series of the image data of silver staining gel taken during the developmental process was used for analysis (5). This image analysis method can check the spots of the same protein in each image at different times of development, and was modified in the present study in order to compare corresponding protein spots from two different gel images. The modified image analysis also contains multivariate analysis, which examines the relationship between the spots of two gel images. In the present study, a proteomic technique with a modified image analysis was applied to phase recognition during the fedbatch culture production of a single-chain variable fragment antibody (scFv). Pichia pastoris is widely used as a host strain for the
* Corresponding author. Tel./fax: þ81 75 724 7539. E-mail address: [email protected] (M. Kishimoto).
production of foreign proteins, and many eukaryotic proteins have been produced in the supernatant outside cells (6e11). We also used a strain of P. pastoris (GS115) expressing the scFv of anti-CRP antibody as a model strain in the present study, and the fedbatch cultures were carried out under the control of methanol concentration during the production phase. In the fed-batch culture, the control of methanol feeding is important because methanol is used not only for the induction of protein expression but also for the growth of the strain. The starvation of the carbon source mortally inhibits both the growth of the strain as well as the production of foreign protein. Furthermore, an excessively high methanol concentration in the culture broth severely affects the metabolism in the strain. Therefore, we constructed a methanol feeding control system with the aid of a semi-conductor gas sensor (7,8). As a result of the fed-batch culture using a 2 l jar-fermentor with a control system, we obtained 73 g/l of the dry cell weight (DCW) and 2.1 g/l of the scFv concentration in the supernatant of the culture broth, which was 5 times higher than the maximum product concentration obtained in the flask cultivation. The obtained scFv concentration in the present study is one of the highest concentrations reported in scFv production (6,8,11e15). In the fedbatch culture with a methanol control system, the scFv production started soon after the startup of the methanol feeding operation. The production continued for approximately 100 h, during which the production rate was kept almost constant. We analyzed the proteins contained in the cell in the growth and production phases using two-dimensional electrophoresis (2-DE) in order to investigate the effect of the methanol feeding operation on production.
1389-1723/$ e see front matter Ó 2014, The Society for Biotechnology, Japan. All rights reserved. http://dx.doi.org/10.1016/j.jbiosc.2014.12.015
188
FUJIKI ET AL.
J. BIOSCI. BIOENG.,
We also examined the protein expression amounts in each phase and compared them statistically based on multivariate analysis. The images obtained from the cells in the different phases were remarkably different from one another, which suggested that the expression protein of the cells had seriously changed with the phase transition. Furthermore, many proteins in the cell were identified using a MALDI TOF MASS with a MASCOT database system, and some of them were strongly related to the expression of scFv with an AOX1 promoter and the scFv biodegradation process. These results could provide crucial information for the determination of an optimum operation for scFv production. MATERIALS AND METHODS Culture conditions P. pastoris GS115 was obtained from Invitrogen (Carrlsbad, CA, USA). P. pastoris is particularly well-suited for fermentative growth, and has the ability to reach very high cell densities during fermentation, which may improve overall protein yields. The transformation of the strain for the scFv production was carried out based on the information from references (7,10), as follows. The anti-CRP scFv gene with a His tag was inserted into the Snab I site of a vector plasmid pPIC9 (Invitrogen), which contained the AOX1 promoter and a-factor signal genes for expression in P. pastoris. The inserted plasmid was digested with Sal I in a HIS 4 gene and introduced into P. pastoris GS115 by electroporation after competent cell preparation, as shown in Supplementary Fig. S1. The transformed strain was stocked in a 1 ml vial in a freezer at 80 C, and was used as a fermentation inoculum for the seed culture, which was carried out for 24 h in 500 ml flasks containing 50 ml YPD medium at 30 C with circular shaking at 170 rpm. A seed culture broth was used for the inoculum of the fed-batch culture using a 2-l Jar fermentor (BMJ-O2PI; Biott, Tokyo) with a 1-l medium solution (glycerol, 40 g/l; K2SO4, 18.2 g/l; CaSO4$2H2O, 0.93 g/l; MgSO4$7H2O, 14.9 g/l; KOH, 4.13 g/l; H3PO4, 26.7 ml/l) and 4.4 ml of a trace metal solution (CuSO4$5H2O, 6 g/l; KI, 0.08 g/l; MnSO4$5H2O, 3 g/l; H3BO3, 0.02 g/l; MoNa2O4$2H2O, 0.20 g/l; CoCl2$6H2O, 0.916 g/l; ZnCl2, 20 g/l; FeSO4$7H2O, 65 g/l; D-biotin, 0.2 g/l; H2SO4, 5.0 ml/l). The pH of the medium was automatically controlled at 5.0 with the addition of an ammonium aqueous solution, and the dissolved oxygen (DO) level was maintained above 20% air saturation by control of the agitation speed (400e1000 rpm). On-line control of the methanol concentration Semi-conductor gas sensor models TGS813 and TGS822 (Figaro Engineering Inc., Osaka) were used for the on-line monitoring of the methanol concentration in the exhausted gas from the fermentor (7,8). The output signal from the sensor was amplified and converted to digital signals and transferred to a personal computer, which could graphically display the methanol concentration in the culture broth, and could control the feed rate of the methanol tubing pump with a proportional-integral-derivative control system. TGS813 was used for the fed-batch culture at 5, 10 and 25 g/l of methanol concentration, and TGS822 was used for the fed-batch culture at 3 g/l of methanol concentration. Operation of the fed-batch culture The fed-batch culture process was divided into the following operational phases. In the first phase, the batch operation was carried out for initial growth using glycerol as a carbon source. When the glycerol in the initial medium was perfectly consumed, and the DO value jumped up for the exhaustion of the carbon source, the second operation phase was started as follows. The glycerol fed-batch culture was carried out for high cell density, and an 80% glycerol aqueous solution with a 2.5% volume of trace element solution was used for the feeding medium. The feed pump was on/
off controlled in such a manner that its switch would be on when the DO value was higher than 60%, and its switch would be off when the DO value was less than 60%. The temperature, pH, agitation speed, and aeration rate were maintained at 30 C, 5.0, 900 rpm, and 1.2 vvm, respectively, during cultivation. The transition process was carried out as follows. Glycerol feeding was terminated when the optical density at 600 nm (OD600) exceeded 130. Initiation of the induction of the protein production was carried out by adding 10 g of a 100% methanol solution containing 25 ml PTM trace salts per liter of methanol 30 min after the DO sudden increase, which indicated glycerol starvation. After 1.5 h, 10 g of methanol was added again, and after another 1.5 h the methanol feeding was started using the control system of methanol concentration with the on-line detection system of methanol concentration, as shown in Fig. 1. We carried out the culture experiment with various methanol concentrations in order to examine the effect on the growth and the scFv production. Measurement of scFv concentration in the culture broth A sample solution from the culture broth was centrifuged at 4500 g for 10 min. The supernatant solution was harvested and centrifuged again at 4500 g for 5 min and the 2nd supernatant was harvested. The His-tagged scFv was purified from the 2nd supernatant solution using the affinity chromatography system (ÄKTA system) with a His Trap HP affinity column (GE Healthcare). The supernatant was applied to the affinity column, which was pre-equilibrated with buffer A (pH 7.2) containing 20 mM imidazole and 2 PBS buffer containing 16 g/l NaCl, 0.4 g/l KCl, 5.8 g/l Na2HPO4$12H2O and 0.4 g/l KH2PO4. After washing the column with buffer A, the bound proteins were eluted with buffer B (pH 7.2) containing 400 mM imidazole and 2 PBS buffer, and the eluates were harvested using a fraction collector. The harvested solution was treated by dialysis using a cellulose membrane tube with 1 PBS buffer as the dialysate. The protein amount in the dialysis-treated solution was measured based on the Lowry’s method using a Dc protein assay kit (Bio-Rad). The cell concentration in the culture broth was calculated from the optical density at 600 nm using a V-630 spectrophotometer (JASCO, Tokyo). The methanol concentration was intermittently measured using a gas chromatograph (model GC14B, Shimadzu Kyoto), which was equipped with a flame ionization detector and a Chromatopac C-R8A (Shimadzu). The initial column temperature was 80 C, the temperature was increased at 4 C/min up to 120 C, and the temperature was held at 120 C for 2 min. The temperature of the injector and detector were maintained at 140 C and 120 C, respectively. N2 gas (0.2 MPa) was used as a carrier gas. A propanol solution was used as the internal standard liquid. SDS-PAGE SDS-PAGE was carried out in order to investigate the protein mixture content in the supernatant of the culture broth using 13 cm 13 cm acryl amide gels, which contained a stacking gel zone and a separating gel zone. The electric currency during electrophoresis was set at 30 mA/gel. Coomassie Brilliant Blue G-250 (0.1%) was used to stain the separated protein. 2-DE analysis The overall procedure for 2-DE was based on the protocol listed in a review by Görg et al. (16) and on the technical information for a 2-D PAGE in SWISS-2DPAGE (http://world-2dpage.expasy.org/swiss2dpage/docs/ protocols/). The culture broth in the fed-batch culture was diluted to 30 g/l DCW, and 1.6 ml of this diluted solution was centrifuged for 5 min at 800 g. Precipitated cells were washed 9 times in 1/15 M phosphate buffer (pH 7.0) with centrifugation for 5 min at 4500 g after each washing. Cell washing using a phosphate buffer should be carried out carefully in order to thoroughly remove the ionized components. The cell pellet was resuspended in 0.8 ml of lysis buffer containing 8 M urea, 2 M thiourea, 2％ w/v 3-[(3-cholamidopropyl) dimethylammonio] propanesulfonate (CHAPS), 0.3 M dithiothreitol (DTT), 32 ml of IPG buffer (GE Healthcare Japan, Tokyo) and trace amounts of Bromophenol Blue. The sample solution was transferred into a screw-capped tube with 0.92 g of glass beads (diameter, 1.0 mm). Cell disruption should be carried out
FIG. 1. Time courses of a fermentation system with methanol concentration controlled at 5 g/l.
VOL. 120, 2015
PHASE ANALYSIS IN scFv PRODUCTION
189
completely using 3 repetitions of the bead Smash and freezing of the cell sample as follows. Void space that equaled at least 1/3 the total volume of the tube was provided for good performance of the homogenization. The tube was set on a Bead Smash12, and homogenization was carried out by strong shaking at 4200 spm for 80 s followed by freezing at 80 C and a natural thawing procedure. The homogenization was repeated 3 times. The homogenized suspension harvested from the tube was centrifuged for 10 min at 20,000 g and the supernatant solution was applied to 2D electrophoresis as follows. We loaded 300 ml of the solution onto an IPG gel strip (13 cm, pH 4-7, Amersham Biosciences, USA), which was set in an Amersham reswelling cassette, and the strip was covered with 2 ml of silicone oil. The strip was left undisturbed and hydrated overnight. After hydration, the IPG strips containing the supernatant solution were transferred to a strip tray. Isoelectric focusing (IEF) was carried out as follows: voltage was linearly increased from 0 to 100 V within 0.5 h, and was maintained at 100 V for 2 h followed by a linear increase from 100 V to 3500 V within 1.5 h. The voltage was then maintained at a constant 3500 V for 3.0 h. The other conditions of 2-D electrophoresis were the same as those reported by Izawa et al. (2). The 2-DE gels were stained by soaking in 250 ml of CBB Stain One (Nakarai tesque) for 1 h with shaking at 35 rpm. The gels were destained 3 times by soaking in deionized water for 5 min with shaking at 35 rpm, and were left undisturbed until the background color had become almost transparent and only the color of the protein remained.
solution containing 1 pmol/ml trypsin, and were then incubated at 37 C for 16 h. Peptides produced in the gel were extracted using an extract solution that contained 50% acetonitrile, 1% trifluoro acetic acid, and 49% extra-pure water, and was dried in a vacuum centrifuge. One ml of extracted solution and 1 ml of saturated a-cyano-4hydroxycinnamic acid (CHCA) solution were mixed, and 1 ml of the mixture was placed on the target plate of a mass spectrometer (Autoflex speed TOF/TOF) with the peptide calibration standard solution. Protein identification was performed using the MASCOT database. The parameters were as follows: database, NCBInr; taxonomomy, fungi; enzyme, trypsin; fixed modifications, carbamidomethyl (C); mass values, MHþ, monoisotopic; peptide mass tolerance, 100 ppm; peptide change state, 1þ; and, max missed cleavages, 1.
Image analysis Gel images were captured using a digital scanner, and were pre-treated using Paint Shop Pro 8 to trim the image in order to define the analysis area and for conversion to the bmp file image from gray scale. The converted images were analyzed with Cancerd2, which is software that discriminates the spot area, calculates each spot area, and compares the two images statistically. To compare the two gel images, we adjusted the coordinate using landmark spot pairs, which were manually selected. The adjustment of the coordinate was carried out as follows. We assumed that the gel in the CBB staining tray would move parallel and perpendicular to the tray movement, and its expansion and contraction would also happen in the tray. Then, we adjusted the second gel coordinate so that the centroids of the landmark spots in the second gel had almost the same location as those in the first gel, as follows. First, we adjusted the rotation angle of the gel coordinates in order to minimize the distance between the centroids of the landmark spots in the first gel and those of the corresponding landmark spots in the second gel. Second, we adjusted the coordinate axes according to the expansion and contraction to minimize the distance between the centroids in the first gel and those in the second. The rolling ball algorithm was used for background normalization (17). Then the software calculated the volume of each spot separated by 2DE using the following equation.
RESULTS AND DISCUSSION
Spot volume ¼
All pixels in the spot X
ðSignal intensity of each pixel e background intensityÞ (1)
Spot volume means the integral value of the pixel color minus the background color within the spot area. Using the spot volumes, Cancerd2 can calculate the correlation coefficient between the spot volumes of two 2DE-gel images using the following equation. Pn ðx xÞðy yÞ i ¼ 1 i Pn i j ¼ 1 xj x k ¼ 1 ðyk yÞ
r ¼ Pn
(2)
where r is the correlation coefficient, n is the number of pixels in the spot area, x is the spot volume in the first gel image, y is the spot volume in the second image, and the number with a superscript bar is the average value. The software for 2DE image analysis was developed by improving the software of the previous version (5) in order to examine the correlation between a pair of 2DE images, as follows. The software provides a figure that graphically presents the relationship of the spot volumes in a pair of 2-DE gel images, in which the x coordinate represents the spot volume in the first 2-DE gel image, and the y coordinate represents the corresponding spot volume in the other image. Moreover, it provides the correlation coefficient between them. Identification of protein spots The identification of protein spots were carried out based on peptide mass finger print data using matrix assisted laser desorption ionization (MALDI-TOF-MS) as follows (1,2). Protein spots were excised from the CBB stained gel and placed into a 1.5 ml siliconized tube. To perfectly destain CBB staining, 150 ml of 0.1 M ammonium bicarbonate in 40% acetonitrile solution was added to the tube, and was then incubated for 45 min with strong shaking using a vortex mixer. The supernatant was then discarded. The gel was then washed 3 or 4 times with 500 ml of extra-pure water and was incubated for 5 min with 100 ml of acetonitrile. The acetonitrile was discarded and the gel fragments of each of the protein spots were dried in a vacuum centrifuge. The fragments were rehydrated with an ammonium bicarbonate
Analysis of protease activity in the supernatant of culture broth Samples of culture broth harvested during cultivation were centrifuged at 4500 g for 10 min, and the supernatants were used for the second centrifugation at 4500 g for 5 min. Three ml of the supernatant from the second centrifugation was mixed with 3 ml of the scFv solution purified by liquid chromatography from the culture solution, and the mixed solution was incubated with shaking at 300 rpm and 30 C in a Deep Well Maximizer (Taitec, Saitama). Samples were taken at incubation times of 1, 4, 8, and 18 h, and were applied to western blotting analysis as follows. Proteins blotted on the PVDF membrane were defected using Rabbit anti-6 His antibody as a first antibody and alkaline phosphatase-conjugated anti rabbit IgG antibody as a second antibody.
Culture experimental results Fig. 1 shows the time courses of fermentation wherein the methanol concentration was maintained at 5 g/l. During growth phase 1 with glycerol as a carbon source in the batch culture and growth phase 2 with glycerol feeding, an exponential growth of cells was observed, as shown in Fig. 1, and the cell concentration reached 50 g DCW/l after 26 h of cultivation. Then, the methanol feeding was started in place of the glycerol feeding, which gave the AOX1 promoter a chance to function and scFv production was started, as shown in Fig. 1. In the early period of the production phase in the methanol feeding, the cell growth rate became slow, and scFv production was observed. The production continued until 100 h of cultivation, and ceased gradually. The concentration of scFv decreased after 150 h of cultivation. Methanol concentration after 26 h of cultivation was maintained at approximately 5 g/l, and the feedback control system for methanol concentration in the fermentation system worked well throughout the culture experiments. In the other culture experiments, the methanol concentration was controlled well at each of the set levels, i.e., 3, 10 and 25 g/l. Fig. 2 shows the effect of methanol concentration on cell growth. The methanol concentration was maintained at each of the set levels, 3, 5, 10, and 25 g/l, after methanol feeding had been started. The patterns in the growth phase were almost the same except for the lag time at a methanol concentration of 25 g/l during run 2. In the production phase, cell concentration increased gradually for 3, 5 and 10 g/l of methanol concentration, but almost ceased in the latter production phase of a 25 g/l methanol concentration. We assumed that the high methanol concentration caused some type of stress to the cells and damaged the cell growth. Fig. 3 shows the effects of different methanol concentrations on the production of scFv. The production amount was higher with an increase in the methanol concentration. In the case of methanol concentrations of 10 g/l and less, the scFv production had almost ceased after 100 h of cultivation. However, for 25 g/l of methanol concentration in run 1, the production continued after 100 h, and the maximum concentration of scFv was 2090 mg/l, which was the highest level of protein production using P. pastoris. In the case of run 2, the lag period (about 10 h) was longer than that in run 1 as well as in other experiments. Therefore, we assumed that the metabolic activity of the cell in run 2 might be a bit lower, and in fact the production was lower than that in run 1 wherein the methanol concentration was controlled at the same level (25 g/l).
190
FUJIKI ET AL.
J. BIOSCI. BIOENG.,
FIG. 2. Effects of methanol concentration on cell growth.
FIG. 3. Effects of methanol concentration on scFv production.
However, the production level in run 2 was still higher than those under lower methanol concentration levels, and the methanol concentration at 25 g/l was more suitable for scFv production in the presented culture system than the lower methanol concentration. The other highlight was that the concentration dropped sharply after the attainment of the highest concentration in the case of the fed-batch culture wherein the methanol concentration was maintained at more than 10 g/l during the production phase. The reason for the drop in production was examined in order to prepare a strategy for the improvement of productivity in a fedbatch culture operation. As a preliminary step, we used SDSpolyacrylamide gel electrophoresis (SDS-PAGE) to investigate the protein content in the supernatant of the culture broth, and the results are shown as gel images in Supplementary Fig. S2. In the case of a high methanol concentration, i.e., 25 g/l, various protein contents in the supernatants increased remarkably after 90 h of cultivation. On the other hand, such a remarkable increase was not observed in the case of a methanol concentration of 10 g/l. We assumed that the high methanol concentration had resulted in autolysis in the latter phases of cultivation. Supplementary Fig. S3 shows the scFv degradation in the supernatant of the culture broth, as analyzed by western blotting. In the case of 25 g/l of methanol concentration, a degradation of scFv in the supernatant was observed after 116 h of cultivation. Degradation was not observed in the case of 5 g/l of methanol concentration. The degradation observed in the western blotting
corresponded to the culture experiment results shown in Fig. 3. Therefore, we checked the supernatant for the presence of bioactive materials, which can degrade the scFv. Fig. 4 shows the results of affinity chromatography for scFv biodegradation in the mixture of 1.0 (g/l) scFv solution purified from the culture broth and in the supernatant from the culture broth, both of which were analyzed by western blotting. When the supernatant of a 25 g/l methanol concentration was used, the added scFv was degraded remarkably compared with that in the case of 5 g/l. These results presented strong evidence that the supernatant
FIG. 4. ScFv degradation in the mixture of 1.0 g/l scFv solution, and the supernatant prepared from the culture broth sample taken from the fed-batch culture at the given cultivation time.
VOL. 120, 2015
PHASE ANALYSIS IN scFv PRODUCTION
191
FIG. 5. (A) Relationship of the spot volumes between two 2DE-gel images prepared from the same sample. (B) Relationship between the spot volumes in the gel prepared from the growth phase sample and those from the production phase sample.
contained some proteases that had leaked from the cells by autolysis, which could degrade the scFv. This was observed by SDS-PAGE, as shown in the Supplementary Fig. S3. The experimental results of 2DE In order to investigate the culture conditions using the correlation coefficients from the 2DE analysis, we checked the reliability, or reproducibility, of the gel image, and we used the 3 gel images for each culture sample, as shown in Supplementary Fig. S4. Fig. 5A shows the relationship between the spot volumes in two 2DE-gel images prepared from the same sample. The data of the spot volumes are almost linearly placed on the figure, and the correlation factor is 0.93. In the case of 2-DE in this study, when using yeast, we found the following precautions should be followed: (i) Cell washing using a phosphate buffer should be carried out carefully in order to thoroughly remove the ionized components. (ii) Cell disruption should be carried out completely using 3 repetitions of the bead Smash and freezing of the cell sample. Fig. 5B shows the relationship between the spot volumes in the gel prepared from the growth phase sample and those from the production phase sample. The value of the correlation factor was 0.0485, which was much lower than the value calculated from the same sample. The results suggested that the correlation coefficient could represent the change of the cell internal proteins with the phase change. Moreover, we were able to recognize the difference in the two gel images by visual inspection, and some spots appeared in the gel of the production phase, but not in the gel of the growth phase. We identified the proteins in the spots based on peptide mass fingerprinting using MALDI-TOF-MS. Supplementary Fig. S5 shows the identified spot protein results in the gels with 25 g/l methanol concentration after 74 h of cultivation, 5 g/l methanol concentration after 74 h of cultivation, and 5 g/l methanol concentration after 186 h of cultivation. Supplementary Fig. S5b and S5c shows only the identified number of proteins that are not identified in Supplementary Fig. S5a. Supplementary Table S1 shows the details of the identified results, i.e., the identified number, protein name, molecular weight, and so on. The spots that are indicated by the arrows in Fig. S5b are identified in Fig. S5a, and have the same identified protein number as that in Fig. S5a. The spots, which are indicated by arrows in Fig. S5c, are identified in Fig. S5a or S5b. Approximately 400 spots for each gel were isolated using the
Cancerd2 program, and approximately 70 spots were identified as the specified proteins spots, which include the enzymes related to methanol metabolism [alcohol oxidase (AOX), catalase A (CAT), formaldehyde dehydrogenase (FLD), and formate dehydrogenase (FDH)], ribosomal protein, glycolysis enzymes, TCA cyclic enzymes, thiamine synthesis enzymes, heat shock enzymes, superoxide dismutase, and so on. Table S1 presents the detailed identified protein data, and the corresponding spots of NCBI ID numbers are shown in Fig. S5. Among the identified proteins, we paid attention to vacuolar aspartyl protease (proteinase A), because it was assumed that the enzyme of methanol-utilizing yeast could biodegrade the antibody. Fig. 6 shows the spot volume of proteinase A in the 2DE gel image, where the applied samples were taken from the culture supernatant at different culture times and under different methanol concentrations. Proteinase A contents under 25 g/l methanol concentration were much higher than those under 5 g/l methanol concentration. According to the phase analyses of 2DE gels, only proteinase A among 400 protein spots identified on 2DE gel was detected as a protease and it was increased at a higher concentration of methanol (25 g/l), while another type of protease was not identified on the gels. Therefore, proteinase A would potentially degrade scFv secreted in supernatant by cell lysis, which
FIG. 6. The spot volumes of proteinase A in 2DE gel images prepared from the fed-batch culture at 5 g/l of methanol concentration, and those at 25 g/l of methanol concentration.
192
FUJIKI ET AL.
J. BIOSCI. BIOENG.,
will frequently occur in the latter phases of cultivation under induction condition with a high concentration of methanol. Proteome analysis yields much information about protein expressions and metabolism. Supplementary Fig. S6 shows a typical example of the protein expression that appeared on the 2DE gel image, and Fig. S7 presents the spot volume data calculated from the gel images by Cancerd2. There is a wealth of information available on the status of many proteins and some of them might be directly related to the production or the expression of the objective materials. In our case, some information was used for the inference of metabolic change in the cultivation, i.e., the appearance of methanol oxidase spots suggested the expression of scFv under the AOX1 promoter, and the appearance of protease A spots showed that the enzyme might cause the degradation of scFv in the case of high methanol concentration in a latter production phase of cultivation. From these results, it was suggested that we should maintain a high methanol concentration in order to accelerate the scFv production, and should avoid biodegradation by the protease in order to increase the scFv production by a recombinant strain. From the experimental results shown in Fig. 3, we noticed a trend whereby a higher concentration promotes the biodegradation of scFv. These results show that the methanol concentration should be carefully controlled at the appropriate level during the production phase. Kuroda et al. (18) reported that the partial degradation of secreted antibodies was suppressed in the case of the use of a protease-deficient strain, and this should be taken into account. In the future, more information about bioprocesses will result from proteome analysis compared with the approach used in the current study, and we will be able to provide more precise inferences for the improvement of such bioprocesses. Furthermore, the collaboration of proteome analysis and metabolic engineering can provide information about the kinetics in the metabolic system inside the cell body, and suggest strategies of cell breeding and optimization of culture conditions. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.jbiosc.2014.12.015. ACKNOWLEDGMENT We would like to express our gratitude to Prof. Y. Katakura at Kansai University for provision of the scFv genes tested in this study. References 1. Henzel, W. J., Watanabe, C., and Stults, J. T.: Protein identification: origins of peptide mass fingerprinting, J. Am. Soc. Mass Spectrom., 14, 931e942 (2003).
2. Izawa, N., Kishimoto, M., Konishi, M., Omasa, T., Shioya, S., and Ohtake, H.: Recognition of culture state using two-dimensional gel electrophoresis with an artificial neural network, Proteomics, 6, 3730e3738 (2006). 3. Kishimoto, M., Kitsuta, Y., Takeuchi, S., Nakajima, M., and Yoshida, T.: Computer control of glutamic acid production based on fuzzy clusterization of culture phases, J. Ferment. Bioeng., 72, 110e114 (1991). 4. Horiuchi, J. and Hiraga, K.: Industrial application of fuzzy control to largescale recombinant vitamin B2 production, J. Biosci. Bioeng., 87, 358e364 (1999). 5. Kishimoto, M., Tatsumi, Y., Tamesui, N., Kumada, Y., Horiuchi, J., and Okumra, K.: Dynamic analysis of the silver staining gel image of 2-DE for protein quantification, J. Electrophor., 51, 1e7 (2010). 6. Cregg, J. M., Vedvick, T. S., and Raschke, W. C.: Recent advaces in the expression of foreign genes in Pichia pastoris, Biotechnology, 11, 905e910 (1993). 7. Guarna, M. M., Lesnicki, G. J., Tam, B. M., Robinson, J., Radziminski, C. Z., Hasenwinkle, D., Boraston, A., Jervis, E., MacGillvray, R. T. A., Turner, R. F. B., and Kilburn, D. G.: On-line monitoring and control of methanol concentration in shake-flask cultures of Pichia pastoris, Biotechnol. Bioeng., 56, 279e286 (1997). 8. Katakura, Y., Zhang, W., Zhuang, G., Omasa, T., Kishimoto, M., Goto, Y., and Suga, K.: Effect of methanol concentration on the production of human b2glycoprotein I domain V by a recombinant Pichia pastoris: a simple system for the control of methanol concentration using a semiconductor gas sensor, J. Biosci. Bioeng., 86, 482e487 (1998). 9. Ciaccio, C., Gambacurta, A., Sanctes, G. D. E., Spagnolo, D., Sakarikou, C., Petrella, G., and Colitta, M.: rhEPO (recombinant human eosinophil peroxidse): expression in Pichia pastoris and biochemical characterization, Biochem. J., 395, 295e301 (2006). 10. Lee, C. Y., Nakano, A., Shiomi, N., Lee, E. K., and Katoh, S.: Effects of substrate feed rate on heterologous protein expression by Pichia pastoris in DO-stat fedbatch fermentation, Enzyme Microb. Technol., 33, 358e365 (2003). 11. Yamawaki, S., Matsumoto, T., Ohnishi, Y., Kumada, Y., Shiomi, N., Katsuda, T., Lee, E. K., and Katoh, S.: Production of single-chain variable fragment antibody (scFv) in fed-batch and continuous culture of Pichia pastoris by two different methanol feeding methods, J. Biosci. Bioeng., 104, 403e407 (2007). 12. Freyre, F. M., Vazquez, J. E., Ayaya, M., Canaan-Haden, L., Bell, H., Rodriguez, I., Gonzalez, A., Cintado, A., and Gavilondo, J. V.: Very high expression of an anti- carcinoembryonic antigen single chain Fv antibody fragment in the yeast Pichia pastoris, J. Biotechnol., 76, 157e163 (2000). 13. Wan, L., Cai, H. W., Yang, H., Lu, Y. R., Li, Y. Y., Li, X. W., Li, S. F., Zhang, J., Li, Y. P., Cheng, J. Q., and Lu, X. F.: High-level expression of a functional humanized single-chain variable fragment antibody against CD25 in Pichia pastoris, Appl. Microbiol. Biotechnol., 81, 33e41 (2008). 14. Damasceno, L. M., Anderson, K. A., Ritter, G., Cregg, J. M., Old, L. J., and Batt, C. A.: Cooverexpression of chaperones for enhanced secretion of a singlechain antibody fragment in Pichia pastoris, Appl. Microbiol. Biotechnol., 74, 381e389 (2007). 15. Kharti, N. K. and Hoffmann, F.: Impact of methanol concentration on secreted protein production in oxygen-limited cultures of recombinant Pichia pastoris, Biotechnol. Bioeng., 93, 871e879 (2006). 16. Görg, A., Obermaier, C., Boguth, G., Harder, A., Scheibe, B., Wildgruber, R., and Weiss, W.: The current state of two-dimensional electrophoresis with immobilized pH gradients, Electrophoresis, 21, 1037e1053 (2000). 17. Sternberg, S. R.: Biomedical image processing, Computer, 16, 22e34 (1983). 18. Kuroda, K., Kitagawa, Y., Kobayashi, K., Tsumura, H., Komeda, T., Mori, E., Motoki, K., Kataoka, S., Chiba, Y., and Jigami, Y.: Antibody expression in protease-deficient strains of the methylotrophic yeast Ogataea minuta, FEMS Yeast Res., 7, 1307e1316 (2007).
Phase analysis in single-chain variable fragment production by recombinant Pichia pastoris based on proteomics combined with multivariate statistics Yuya Fujiki, Yoichi Kumada, and Michimasa Kishimoto* Department of Chemistry and Materials, Kyoto Institute of Technology, Goshokaidocho, Sakyoku, Kyoto 606-8585, Japan1 and Department of Bio-molecular Engineering, Kyoto Institute of Technology, Goshokaidocho, Sakyoku, Kyoto 606-8585, Japan2 Received 26 August 2014; accepted 15 December 2014 Available online 28 January 2015
The proteomics technique, which consists of two-dimensional gel electrophoresis (2-DE), peptide mass fingerprinting (PMF), gel image analysis, and multivariate statistics, was applied to the phase analysis of a fed-batch culture for the production of a single-chain variable fragment (scFv) of an anti-C-reactive protein (CRP) antibody by Pichia pastoris. The time courses of the fed-batch culture were separated into three distinct phases: the growth phase of the batch process, the growth phase of the fed-batch process, and the production phase of the fed-batch process. Multivariate statistical analysis using 2-DE gel image analysis data clearly showed the change in the culture phase and provided information concerning the protein expression, which suggested a metabolic change related to cell growth and production during the fed-batch culture. Furthermore, specific proteins, such as alcohol oxidase, which is strongly related to scFv expression, and proteinase A, which could biodegrade scFv in the latter phases of production, were identified via the PMF method. The proteomics technique provided valuable information about the effect of the methanol concentration on scFv production. Ó 2014, The Society for Biotechnology, Japan. All rights reserved. [Key words: Phase analysis; Single-chain variable fragment; Pichia pastoris; Proteomics; Cancerd2; Two-dimensional gel electrophoresis; Peptide mass fingerprinting]
Two-dimensional electrophoresis (2-DE) is a valuable tool for precise protein separation and can be effectively applied to proteome analysis based on peptide mass fingerprinting (PMF) using MALDI-TOF MS (1). In previous work, proteome analysis with the aid of artificial neural networks was successfully applied to recognize the sufficiency of chemical elements based on the protein spots resolved in 2-DE (2). The recognition of culture phases in a culture process would be important if the determination of different culture operations in each phase had a critical advantage for the improvement of bioproduction (3,4). We have developed a novel image analysis method combined with recently developed powerful computing hardware, in which the time series of the image data of silver staining gel taken during the developmental process was used for analysis (5). This image analysis method can check the spots of the same protein in each image at different times of development, and was modified in the present study in order to compare corresponding protein spots from two different gel images. The modified image analysis also contains multivariate analysis, which examines the relationship between the spots of two gel images. In the present study, a proteomic technique with a modified image analysis was applied to phase recognition during the fedbatch culture production of a single-chain variable fragment antibody (scFv). Pichia pastoris is widely used as a host strain for the
* Corresponding author. Tel./fax: þ81 75 724 7539. E-mail address: [email protected] (M. Kishimoto).
production of foreign proteins, and many eukaryotic proteins have been produced in the supernatant outside cells (6e11). We also used a strain of P. pastoris (GS115) expressing the scFv of anti-CRP antibody as a model strain in the present study, and the fedbatch cultures were carried out under the control of methanol concentration during the production phase. In the fed-batch culture, the control of methanol feeding is important because methanol is used not only for the induction of protein expression but also for the growth of the strain. The starvation of the carbon source mortally inhibits both the growth of the strain as well as the production of foreign protein. Furthermore, an excessively high methanol concentration in the culture broth severely affects the metabolism in the strain. Therefore, we constructed a methanol feeding control system with the aid of a semi-conductor gas sensor (7,8). As a result of the fed-batch culture using a 2 l jar-fermentor with a control system, we obtained 73 g/l of the dry cell weight (DCW) and 2.1 g/l of the scFv concentration in the supernatant of the culture broth, which was 5 times higher than the maximum product concentration obtained in the flask cultivation. The obtained scFv concentration in the present study is one of the highest concentrations reported in scFv production (6,8,11e15). In the fedbatch culture with a methanol control system, the scFv production started soon after the startup of the methanol feeding operation. The production continued for approximately 100 h, during which the production rate was kept almost constant. We analyzed the proteins contained in the cell in the growth and production phases using two-dimensional electrophoresis (2-DE) in order to investigate the effect of the methanol feeding operation on production.
1389-1723/$ e see front matter Ó 2014, The Society for Biotechnology, Japan. All rights reserved. http://dx.doi.org/10.1016/j.jbiosc.2014.12.015
188
FUJIKI ET AL.
J. BIOSCI. BIOENG.,
We also examined the protein expression amounts in each phase and compared them statistically based on multivariate analysis. The images obtained from the cells in the different phases were remarkably different from one another, which suggested that the expression protein of the cells had seriously changed with the phase transition. Furthermore, many proteins in the cell were identified using a MALDI TOF MASS with a MASCOT database system, and some of them were strongly related to the expression of scFv with an AOX1 promoter and the scFv biodegradation process. These results could provide crucial information for the determination of an optimum operation for scFv production. MATERIALS AND METHODS Culture conditions P. pastoris GS115 was obtained from Invitrogen (Carrlsbad, CA, USA). P. pastoris is particularly well-suited for fermentative growth, and has the ability to reach very high cell densities during fermentation, which may improve overall protein yields. The transformation of the strain for the scFv production was carried out based on the information from references (7,10), as follows. The anti-CRP scFv gene with a His tag was inserted into the Snab I site of a vector plasmid pPIC9 (Invitrogen), which contained the AOX1 promoter and a-factor signal genes for expression in P. pastoris. The inserted plasmid was digested with Sal I in a HIS 4 gene and introduced into P. pastoris GS115 by electroporation after competent cell preparation, as shown in Supplementary Fig. S1. The transformed strain was stocked in a 1 ml vial in a freezer at 80 C, and was used as a fermentation inoculum for the seed culture, which was carried out for 24 h in 500 ml flasks containing 50 ml YPD medium at 30 C with circular shaking at 170 rpm. A seed culture broth was used for the inoculum of the fed-batch culture using a 2-l Jar fermentor (BMJ-O2PI; Biott, Tokyo) with a 1-l medium solution (glycerol, 40 g/l; K2SO4, 18.2 g/l; CaSO4$2H2O, 0.93 g/l; MgSO4$7H2O, 14.9 g/l; KOH, 4.13 g/l; H3PO4, 26.7 ml/l) and 4.4 ml of a trace metal solution (CuSO4$5H2O, 6 g/l; KI, 0.08 g/l; MnSO4$5H2O, 3 g/l; H3BO3, 0.02 g/l; MoNa2O4$2H2O, 0.20 g/l; CoCl2$6H2O, 0.916 g/l; ZnCl2, 20 g/l; FeSO4$7H2O, 65 g/l; D-biotin, 0.2 g/l; H2SO4, 5.0 ml/l). The pH of the medium was automatically controlled at 5.0 with the addition of an ammonium aqueous solution, and the dissolved oxygen (DO) level was maintained above 20% air saturation by control of the agitation speed (400e1000 rpm). On-line control of the methanol concentration Semi-conductor gas sensor models TGS813 and TGS822 (Figaro Engineering Inc., Osaka) were used for the on-line monitoring of the methanol concentration in the exhausted gas from the fermentor (7,8). The output signal from the sensor was amplified and converted to digital signals and transferred to a personal computer, which could graphically display the methanol concentration in the culture broth, and could control the feed rate of the methanol tubing pump with a proportional-integral-derivative control system. TGS813 was used for the fed-batch culture at 5, 10 and 25 g/l of methanol concentration, and TGS822 was used for the fed-batch culture at 3 g/l of methanol concentration. Operation of the fed-batch culture The fed-batch culture process was divided into the following operational phases. In the first phase, the batch operation was carried out for initial growth using glycerol as a carbon source. When the glycerol in the initial medium was perfectly consumed, and the DO value jumped up for the exhaustion of the carbon source, the second operation phase was started as follows. The glycerol fed-batch culture was carried out for high cell density, and an 80% glycerol aqueous solution with a 2.5% volume of trace element solution was used for the feeding medium. The feed pump was on/
off controlled in such a manner that its switch would be on when the DO value was higher than 60%, and its switch would be off when the DO value was less than 60%. The temperature, pH, agitation speed, and aeration rate were maintained at 30 C, 5.0, 900 rpm, and 1.2 vvm, respectively, during cultivation. The transition process was carried out as follows. Glycerol feeding was terminated when the optical density at 600 nm (OD600) exceeded 130. Initiation of the induction of the protein production was carried out by adding 10 g of a 100% methanol solution containing 25 ml PTM trace salts per liter of methanol 30 min after the DO sudden increase, which indicated glycerol starvation. After 1.5 h, 10 g of methanol was added again, and after another 1.5 h the methanol feeding was started using the control system of methanol concentration with the on-line detection system of methanol concentration, as shown in Fig. 1. We carried out the culture experiment with various methanol concentrations in order to examine the effect on the growth and the scFv production. Measurement of scFv concentration in the culture broth A sample solution from the culture broth was centrifuged at 4500 g for 10 min. The supernatant solution was harvested and centrifuged again at 4500 g for 5 min and the 2nd supernatant was harvested. The His-tagged scFv was purified from the 2nd supernatant solution using the affinity chromatography system (ÄKTA system) with a His Trap HP affinity column (GE Healthcare). The supernatant was applied to the affinity column, which was pre-equilibrated with buffer A (pH 7.2) containing 20 mM imidazole and 2 PBS buffer containing 16 g/l NaCl, 0.4 g/l KCl, 5.8 g/l Na2HPO4$12H2O and 0.4 g/l KH2PO4. After washing the column with buffer A, the bound proteins were eluted with buffer B (pH 7.2) containing 400 mM imidazole and 2 PBS buffer, and the eluates were harvested using a fraction collector. The harvested solution was treated by dialysis using a cellulose membrane tube with 1 PBS buffer as the dialysate. The protein amount in the dialysis-treated solution was measured based on the Lowry’s method using a Dc protein assay kit (Bio-Rad). The cell concentration in the culture broth was calculated from the optical density at 600 nm using a V-630 spectrophotometer (JASCO, Tokyo). The methanol concentration was intermittently measured using a gas chromatograph (model GC14B, Shimadzu Kyoto), which was equipped with a flame ionization detector and a Chromatopac C-R8A (Shimadzu). The initial column temperature was 80 C, the temperature was increased at 4 C/min up to 120 C, and the temperature was held at 120 C for 2 min. The temperature of the injector and detector were maintained at 140 C and 120 C, respectively. N2 gas (0.2 MPa) was used as a carrier gas. A propanol solution was used as the internal standard liquid. SDS-PAGE SDS-PAGE was carried out in order to investigate the protein mixture content in the supernatant of the culture broth using 13 cm 13 cm acryl amide gels, which contained a stacking gel zone and a separating gel zone. The electric currency during electrophoresis was set at 30 mA/gel. Coomassie Brilliant Blue G-250 (0.1%) was used to stain the separated protein. 2-DE analysis The overall procedure for 2-DE was based on the protocol listed in a review by Görg et al. (16) and on the technical information for a 2-D PAGE in SWISS-2DPAGE (http://world-2dpage.expasy.org/swiss2dpage/docs/ protocols/). The culture broth in the fed-batch culture was diluted to 30 g/l DCW, and 1.6 ml of this diluted solution was centrifuged for 5 min at 800 g. Precipitated cells were washed 9 times in 1/15 M phosphate buffer (pH 7.0) with centrifugation for 5 min at 4500 g after each washing. Cell washing using a phosphate buffer should be carried out carefully in order to thoroughly remove the ionized components. The cell pellet was resuspended in 0.8 ml of lysis buffer containing 8 M urea, 2 M thiourea, 2％ w/v 3-[(3-cholamidopropyl) dimethylammonio] propanesulfonate (CHAPS), 0.3 M dithiothreitol (DTT), 32 ml of IPG buffer (GE Healthcare Japan, Tokyo) and trace amounts of Bromophenol Blue. The sample solution was transferred into a screw-capped tube with 0.92 g of glass beads (diameter, 1.0 mm). Cell disruption should be carried out
FIG. 1. Time courses of a fermentation system with methanol concentration controlled at 5 g/l.
VOL. 120, 2015
PHASE ANALYSIS IN scFv PRODUCTION
189
completely using 3 repetitions of the bead Smash and freezing of the cell sample as follows. Void space that equaled at least 1/3 the total volume of the tube was provided for good performance of the homogenization. The tube was set on a Bead Smash12, and homogenization was carried out by strong shaking at 4200 spm for 80 s followed by freezing at 80 C and a natural thawing procedure. The homogenization was repeated 3 times. The homogenized suspension harvested from the tube was centrifuged for 10 min at 20,000 g and the supernatant solution was applied to 2D electrophoresis as follows. We loaded 300 ml of the solution onto an IPG gel strip (13 cm, pH 4-7, Amersham Biosciences, USA), which was set in an Amersham reswelling cassette, and the strip was covered with 2 ml of silicone oil. The strip was left undisturbed and hydrated overnight. After hydration, the IPG strips containing the supernatant solution were transferred to a strip tray. Isoelectric focusing (IEF) was carried out as follows: voltage was linearly increased from 0 to 100 V within 0.5 h, and was maintained at 100 V for 2 h followed by a linear increase from 100 V to 3500 V within 1.5 h. The voltage was then maintained at a constant 3500 V for 3.0 h. The other conditions of 2-D electrophoresis were the same as those reported by Izawa et al. (2). The 2-DE gels were stained by soaking in 250 ml of CBB Stain One (Nakarai tesque) for 1 h with shaking at 35 rpm. The gels were destained 3 times by soaking in deionized water for 5 min with shaking at 35 rpm, and were left undisturbed until the background color had become almost transparent and only the color of the protein remained.
solution containing 1 pmol/ml trypsin, and were then incubated at 37 C for 16 h. Peptides produced in the gel were extracted using an extract solution that contained 50% acetonitrile, 1% trifluoro acetic acid, and 49% extra-pure water, and was dried in a vacuum centrifuge. One ml of extracted solution and 1 ml of saturated a-cyano-4hydroxycinnamic acid (CHCA) solution were mixed, and 1 ml of the mixture was placed on the target plate of a mass spectrometer (Autoflex speed TOF/TOF) with the peptide calibration standard solution. Protein identification was performed using the MASCOT database. The parameters were as follows: database, NCBInr; taxonomomy, fungi; enzyme, trypsin; fixed modifications, carbamidomethyl (C); mass values, MHþ, monoisotopic; peptide mass tolerance, 100 ppm; peptide change state, 1þ; and, max missed cleavages, 1.
Image analysis Gel images were captured using a digital scanner, and were pre-treated using Paint Shop Pro 8 to trim the image in order to define the analysis area and for conversion to the bmp file image from gray scale. The converted images were analyzed with Cancerd2, which is software that discriminates the spot area, calculates each spot area, and compares the two images statistically. To compare the two gel images, we adjusted the coordinate using landmark spot pairs, which were manually selected. The adjustment of the coordinate was carried out as follows. We assumed that the gel in the CBB staining tray would move parallel and perpendicular to the tray movement, and its expansion and contraction would also happen in the tray. Then, we adjusted the second gel coordinate so that the centroids of the landmark spots in the second gel had almost the same location as those in the first gel, as follows. First, we adjusted the rotation angle of the gel coordinates in order to minimize the distance between the centroids of the landmark spots in the first gel and those of the corresponding landmark spots in the second gel. Second, we adjusted the coordinate axes according to the expansion and contraction to minimize the distance between the centroids in the first gel and those in the second. The rolling ball algorithm was used for background normalization (17). Then the software calculated the volume of each spot separated by 2DE using the following equation.
RESULTS AND DISCUSSION
Spot volume ¼
All pixels in the spot X
ðSignal intensity of each pixel e background intensityÞ (1)
Spot volume means the integral value of the pixel color minus the background color within the spot area. Using the spot volumes, Cancerd2 can calculate the correlation coefficient between the spot volumes of two 2DE-gel images using the following equation. Pn ðx xÞðy yÞ i ¼ 1 i Pn i j ¼ 1 xj x k ¼ 1 ðyk yÞ
r ¼ Pn
(2)
where r is the correlation coefficient, n is the number of pixels in the spot area, x is the spot volume in the first gel image, y is the spot volume in the second image, and the number with a superscript bar is the average value. The software for 2DE image analysis was developed by improving the software of the previous version (5) in order to examine the correlation between a pair of 2DE images, as follows. The software provides a figure that graphically presents the relationship of the spot volumes in a pair of 2-DE gel images, in which the x coordinate represents the spot volume in the first 2-DE gel image, and the y coordinate represents the corresponding spot volume in the other image. Moreover, it provides the correlation coefficient between them. Identification of protein spots The identification of protein spots were carried out based on peptide mass finger print data using matrix assisted laser desorption ionization (MALDI-TOF-MS) as follows (1,2). Protein spots were excised from the CBB stained gel and placed into a 1.5 ml siliconized tube. To perfectly destain CBB staining, 150 ml of 0.1 M ammonium bicarbonate in 40% acetonitrile solution was added to the tube, and was then incubated for 45 min with strong shaking using a vortex mixer. The supernatant was then discarded. The gel was then washed 3 or 4 times with 500 ml of extra-pure water and was incubated for 5 min with 100 ml of acetonitrile. The acetonitrile was discarded and the gel fragments of each of the protein spots were dried in a vacuum centrifuge. The fragments were rehydrated with an ammonium bicarbonate
Analysis of protease activity in the supernatant of culture broth Samples of culture broth harvested during cultivation were centrifuged at 4500 g for 10 min, and the supernatants were used for the second centrifugation at 4500 g for 5 min. Three ml of the supernatant from the second centrifugation was mixed with 3 ml of the scFv solution purified by liquid chromatography from the culture solution, and the mixed solution was incubated with shaking at 300 rpm and 30 C in a Deep Well Maximizer (Taitec, Saitama). Samples were taken at incubation times of 1, 4, 8, and 18 h, and were applied to western blotting analysis as follows. Proteins blotted on the PVDF membrane were defected using Rabbit anti-6 His antibody as a first antibody and alkaline phosphatase-conjugated anti rabbit IgG antibody as a second antibody.
Culture experimental results Fig. 1 shows the time courses of fermentation wherein the methanol concentration was maintained at 5 g/l. During growth phase 1 with glycerol as a carbon source in the batch culture and growth phase 2 with glycerol feeding, an exponential growth of cells was observed, as shown in Fig. 1, and the cell concentration reached 50 g DCW/l after 26 h of cultivation. Then, the methanol feeding was started in place of the glycerol feeding, which gave the AOX1 promoter a chance to function and scFv production was started, as shown in Fig. 1. In the early period of the production phase in the methanol feeding, the cell growth rate became slow, and scFv production was observed. The production continued until 100 h of cultivation, and ceased gradually. The concentration of scFv decreased after 150 h of cultivation. Methanol concentration after 26 h of cultivation was maintained at approximately 5 g/l, and the feedback control system for methanol concentration in the fermentation system worked well throughout the culture experiments. In the other culture experiments, the methanol concentration was controlled well at each of the set levels, i.e., 3, 10 and 25 g/l. Fig. 2 shows the effect of methanol concentration on cell growth. The methanol concentration was maintained at each of the set levels, 3, 5, 10, and 25 g/l, after methanol feeding had been started. The patterns in the growth phase were almost the same except for the lag time at a methanol concentration of 25 g/l during run 2. In the production phase, cell concentration increased gradually for 3, 5 and 10 g/l of methanol concentration, but almost ceased in the latter production phase of a 25 g/l methanol concentration. We assumed that the high methanol concentration caused some type of stress to the cells and damaged the cell growth. Fig. 3 shows the effects of different methanol concentrations on the production of scFv. The production amount was higher with an increase in the methanol concentration. In the case of methanol concentrations of 10 g/l and less, the scFv production had almost ceased after 100 h of cultivation. However, for 25 g/l of methanol concentration in run 1, the production continued after 100 h, and the maximum concentration of scFv was 2090 mg/l, which was the highest level of protein production using P. pastoris. In the case of run 2, the lag period (about 10 h) was longer than that in run 1 as well as in other experiments. Therefore, we assumed that the metabolic activity of the cell in run 2 might be a bit lower, and in fact the production was lower than that in run 1 wherein the methanol concentration was controlled at the same level (25 g/l).
190
FUJIKI ET AL.
J. BIOSCI. BIOENG.,
FIG. 2. Effects of methanol concentration on cell growth.
FIG. 3. Effects of methanol concentration on scFv production.
However, the production level in run 2 was still higher than those under lower methanol concentration levels, and the methanol concentration at 25 g/l was more suitable for scFv production in the presented culture system than the lower methanol concentration. The other highlight was that the concentration dropped sharply after the attainment of the highest concentration in the case of the fed-batch culture wherein the methanol concentration was maintained at more than 10 g/l during the production phase. The reason for the drop in production was examined in order to prepare a strategy for the improvement of productivity in a fedbatch culture operation. As a preliminary step, we used SDSpolyacrylamide gel electrophoresis (SDS-PAGE) to investigate the protein content in the supernatant of the culture broth, and the results are shown as gel images in Supplementary Fig. S2. In the case of a high methanol concentration, i.e., 25 g/l, various protein contents in the supernatants increased remarkably after 90 h of cultivation. On the other hand, such a remarkable increase was not observed in the case of a methanol concentration of 10 g/l. We assumed that the high methanol concentration had resulted in autolysis in the latter phases of cultivation. Supplementary Fig. S3 shows the scFv degradation in the supernatant of the culture broth, as analyzed by western blotting. In the case of 25 g/l of methanol concentration, a degradation of scFv in the supernatant was observed after 116 h of cultivation. Degradation was not observed in the case of 5 g/l of methanol concentration. The degradation observed in the western blotting
corresponded to the culture experiment results shown in Fig. 3. Therefore, we checked the supernatant for the presence of bioactive materials, which can degrade the scFv. Fig. 4 shows the results of affinity chromatography for scFv biodegradation in the mixture of 1.0 (g/l) scFv solution purified from the culture broth and in the supernatant from the culture broth, both of which were analyzed by western blotting. When the supernatant of a 25 g/l methanol concentration was used, the added scFv was degraded remarkably compared with that in the case of 5 g/l. These results presented strong evidence that the supernatant
FIG. 4. ScFv degradation in the mixture of 1.0 g/l scFv solution, and the supernatant prepared from the culture broth sample taken from the fed-batch culture at the given cultivation time.
VOL. 120, 2015
PHASE ANALYSIS IN scFv PRODUCTION
191
FIG. 5. (A) Relationship of the spot volumes between two 2DE-gel images prepared from the same sample. (B) Relationship between the spot volumes in the gel prepared from the growth phase sample and those from the production phase sample.
contained some proteases that had leaked from the cells by autolysis, which could degrade the scFv. This was observed by SDS-PAGE, as shown in the Supplementary Fig. S3. The experimental results of 2DE In order to investigate the culture conditions using the correlation coefficients from the 2DE analysis, we checked the reliability, or reproducibility, of the gel image, and we used the 3 gel images for each culture sample, as shown in Supplementary Fig. S4. Fig. 5A shows the relationship between the spot volumes in two 2DE-gel images prepared from the same sample. The data of the spot volumes are almost linearly placed on the figure, and the correlation factor is 0.93. In the case of 2-DE in this study, when using yeast, we found the following precautions should be followed: (i) Cell washing using a phosphate buffer should be carried out carefully in order to thoroughly remove the ionized components. (ii) Cell disruption should be carried out completely using 3 repetitions of the bead Smash and freezing of the cell sample. Fig. 5B shows the relationship between the spot volumes in the gel prepared from the growth phase sample and those from the production phase sample. The value of the correlation factor was 0.0485, which was much lower than the value calculated from the same sample. The results suggested that the correlation coefficient could represent the change of the cell internal proteins with the phase change. Moreover, we were able to recognize the difference in the two gel images by visual inspection, and some spots appeared in the gel of the production phase, but not in the gel of the growth phase. We identified the proteins in the spots based on peptide mass fingerprinting using MALDI-TOF-MS. Supplementary Fig. S5 shows the identified spot protein results in the gels with 25 g/l methanol concentration after 74 h of cultivation, 5 g/l methanol concentration after 74 h of cultivation, and 5 g/l methanol concentration after 186 h of cultivation. Supplementary Fig. S5b and S5c shows only the identified number of proteins that are not identified in Supplementary Fig. S5a. Supplementary Table S1 shows the details of the identified results, i.e., the identified number, protein name, molecular weight, and so on. The spots that are indicated by the arrows in Fig. S5b are identified in Fig. S5a, and have the same identified protein number as that in Fig. S5a. The spots, which are indicated by arrows in Fig. S5c, are identified in Fig. S5a or S5b. Approximately 400 spots for each gel were isolated using the
Cancerd2 program, and approximately 70 spots were identified as the specified proteins spots, which include the enzymes related to methanol metabolism [alcohol oxidase (AOX), catalase A (CAT), formaldehyde dehydrogenase (FLD), and formate dehydrogenase (FDH)], ribosomal protein, glycolysis enzymes, TCA cyclic enzymes, thiamine synthesis enzymes, heat shock enzymes, superoxide dismutase, and so on. Table S1 presents the detailed identified protein data, and the corresponding spots of NCBI ID numbers are shown in Fig. S5. Among the identified proteins, we paid attention to vacuolar aspartyl protease (proteinase A), because it was assumed that the enzyme of methanol-utilizing yeast could biodegrade the antibody. Fig. 6 shows the spot volume of proteinase A in the 2DE gel image, where the applied samples were taken from the culture supernatant at different culture times and under different methanol concentrations. Proteinase A contents under 25 g/l methanol concentration were much higher than those under 5 g/l methanol concentration. According to the phase analyses of 2DE gels, only proteinase A among 400 protein spots identified on 2DE gel was detected as a protease and it was increased at a higher concentration of methanol (25 g/l), while another type of protease was not identified on the gels. Therefore, proteinase A would potentially degrade scFv secreted in supernatant by cell lysis, which
FIG. 6. The spot volumes of proteinase A in 2DE gel images prepared from the fed-batch culture at 5 g/l of methanol concentration, and those at 25 g/l of methanol concentration.
192
FUJIKI ET AL.
J. BIOSCI. BIOENG.,
will frequently occur in the latter phases of cultivation under induction condition with a high concentration of methanol. Proteome analysis yields much information about protein expressions and metabolism. Supplementary Fig. S6 shows a typical example of the protein expression that appeared on the 2DE gel image, and Fig. S7 presents the spot volume data calculated from the gel images by Cancerd2. There is a wealth of information available on the status of many proteins and some of them might be directly related to the production or the expression of the objective materials. In our case, some information was used for the inference of metabolic change in the cultivation, i.e., the appearance of methanol oxidase spots suggested the expression of scFv under the AOX1 promoter, and the appearance of protease A spots showed that the enzyme might cause the degradation of scFv in the case of high methanol concentration in a latter production phase of cultivation. From these results, it was suggested that we should maintain a high methanol concentration in order to accelerate the scFv production, and should avoid biodegradation by the protease in order to increase the scFv production by a recombinant strain. From the experimental results shown in Fig. 3, we noticed a trend whereby a higher concentration promotes the biodegradation of scFv. These results show that the methanol concentration should be carefully controlled at the appropriate level during the production phase. Kuroda et al. (18) reported that the partial degradation of secreted antibodies was suppressed in the case of the use of a protease-deficient strain, and this should be taken into account. In the future, more information about bioprocesses will result from proteome analysis compared with the approach used in the current study, and we will be able to provide more precise inferences for the improvement of such bioprocesses. Furthermore, the collaboration of proteome analysis and metabolic engineering can provide information about the kinetics in the metabolic system inside the cell body, and suggest strategies of cell breeding and optimization of culture conditions. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.jbiosc.2014.12.015. ACKNOWLEDGMENT We would like to express our gratitude to Prof. Y. Katakura at Kansai University for provision of the scFv genes tested in this study. References 1. Henzel, W. J., Watanabe, C., and Stults, J. T.: Protein identification: origins of peptide mass fingerprinting, J. Am. Soc. Mass Spectrom., 14, 931e942 (2003).
2. Izawa, N., Kishimoto, M., Konishi, M., Omasa, T., Shioya, S., and Ohtake, H.: Recognition of culture state using two-dimensional gel electrophoresis with an artificial neural network, Proteomics, 6, 3730e3738 (2006). 3. Kishimoto, M., Kitsuta, Y., Takeuchi, S., Nakajima, M., and Yoshida, T.: Computer control of glutamic acid production based on fuzzy clusterization of culture phases, J. Ferment. Bioeng., 72, 110e114 (1991). 4. Horiuchi, J. and Hiraga, K.: Industrial application of fuzzy control to largescale recombinant vitamin B2 production, J. Biosci. Bioeng., 87, 358e364 (1999). 5. Kishimoto, M., Tatsumi, Y., Tamesui, N., Kumada, Y., Horiuchi, J., and Okumra, K.: Dynamic analysis of the silver staining gel image of 2-DE for protein quantification, J. Electrophor., 51, 1e7 (2010). 6. Cregg, J. M., Vedvick, T. S., and Raschke, W. C.: Recent advaces in the expression of foreign genes in Pichia pastoris, Biotechnology, 11, 905e910 (1993). 7. Guarna, M. M., Lesnicki, G. J., Tam, B. M., Robinson, J., Radziminski, C. Z., Hasenwinkle, D., Boraston, A., Jervis, E., MacGillvray, R. T. A., Turner, R. F. B., and Kilburn, D. G.: On-line monitoring and control of methanol concentration in shake-flask cultures of Pichia pastoris, Biotechnol. Bioeng., 56, 279e286 (1997). 8. Katakura, Y., Zhang, W., Zhuang, G., Omasa, T., Kishimoto, M., Goto, Y., and Suga, K.: Effect of methanol concentration on the production of human b2glycoprotein I domain V by a recombinant Pichia pastoris: a simple system for the control of methanol concentration using a semiconductor gas sensor, J. Biosci. Bioeng., 86, 482e487 (1998). 9. Ciaccio, C., Gambacurta, A., Sanctes, G. D. E., Spagnolo, D., Sakarikou, C., Petrella, G., and Colitta, M.: rhEPO (recombinant human eosinophil peroxidse): expression in Pichia pastoris and biochemical characterization, Biochem. J., 395, 295e301 (2006). 10. Lee, C. Y., Nakano, A., Shiomi, N., Lee, E. K., and Katoh, S.: Effects of substrate feed rate on heterologous protein expression by Pichia pastoris in DO-stat fedbatch fermentation, Enzyme Microb. Technol., 33, 358e365 (2003). 11. Yamawaki, S., Matsumoto, T., Ohnishi, Y., Kumada, Y., Shiomi, N., Katsuda, T., Lee, E. K., and Katoh, S.: Production of single-chain variable fragment antibody (scFv) in fed-batch and continuous culture of Pichia pastoris by two different methanol feeding methods, J. Biosci. Bioeng., 104, 403e407 (2007). 12. Freyre, F. M., Vazquez, J. E., Ayaya, M., Canaan-Haden, L., Bell, H., Rodriguez, I., Gonzalez, A., Cintado, A., and Gavilondo, J. V.: Very high expression of an anti- carcinoembryonic antigen single chain Fv antibody fragment in the yeast Pichia pastoris, J. Biotechnol., 76, 157e163 (2000). 13. Wan, L., Cai, H. W., Yang, H., Lu, Y. R., Li, Y. Y., Li, X. W., Li, S. F., Zhang, J., Li, Y. P., Cheng, J. Q., and Lu, X. F.: High-level expression of a functional humanized single-chain variable fragment antibody against CD25 in Pichia pastoris, Appl. Microbiol. Biotechnol., 81, 33e41 (2008). 14. Damasceno, L. M., Anderson, K. A., Ritter, G., Cregg, J. M., Old, L. J., and Batt, C. A.: Cooverexpression of chaperones for enhanced secretion of a singlechain antibody fragment in Pichia pastoris, Appl. Microbiol. Biotechnol., 74, 381e389 (2007). 15. Kharti, N. K. and Hoffmann, F.: Impact of methanol concentration on secreted protein production in oxygen-limited cultures of recombinant Pichia pastoris, Biotechnol. Bioeng., 93, 871e879 (2006). 16. Görg, A., Obermaier, C., Boguth, G., Harder, A., Scheibe, B., Wildgruber, R., and Weiss, W.: The current state of two-dimensional electrophoresis with immobilized pH gradients, Electrophoresis, 21, 1037e1053 (2000). 17. Sternberg, S. R.: Biomedical image processing, Computer, 16, 22e34 (1983). 18. Kuroda, K., Kitagawa, Y., Kobayashi, K., Tsumura, H., Komeda, T., Mori, E., Motoki, K., Kataoka, S., Chiba, Y., and Jigami, Y.: Antibody expression in protease-deficient strains of the methylotrophic yeast Ogataea minuta, FEMS Yeast Res., 7, 1307e1316 (2007).
