YEAST VOL.12: 199-205 (1996) Selection of Yeast Cells with a Higher Plasmid Copy Number in a Saccharomyces cerevisiae Autoselection System CONCETTA COMPAGNO, DANILO PORRO*, STEFANIA RADICE, ENZO MARTEGANI AND BIANCA MARIA RANZI Dipartimento di Fisiologia e Biochimica Generali, Sez. Biochimica Comparata, Universita di Milano, via Celoria 26, 20133 Milano, Italy Received 11 May 1995; accepted 7 August 1995 Autoselection systems allow the selection of a genetically engineered population independently of the growth medium composition. The structure of a Saccharomyces cerevisiae population transformed with an autoselection plasmid, in which a carbon-source-dependentmodulation of the plasmid copy number occurs, was analysed. By means of flow cytometric procedures we tested the cell viability, dynamics of growth and heterologous protein production at single cell level. Such analyses allow the identification and the tracking of a specific cellular sub-population with a higher plasmid copy number which arises after the carbon source shift. The effects of the cellular plasmid distribution on the dynamics of growth are also discussed. KEY WORDS - S. cerevisiae; autoselection plasmid; flow cytometry INTRODUCTION The first key factor required for successful production based on genetically engineered host cells is the stability of the plasmid. Autoselection systems represent a special class of host-expression vectors which offer many advantages for the improvement of processes related to the production of heterologous proteins. In such systems, selection of the transformed population occurs independently of the growth medium composition, allowing the use of relatively inexpensive complex media which are preferred in industry for commercial-scale fermentations. Different autoselection systems have been developed for Saccharomyces cerevisiae host cells (Loison et al., 1986; Piper and Curran, 1990; Unternahrer et al., 1991; Rech et al., 1992; Ayub et al., 1992; Ludwig et al., 1993). The efficiency of these systems is related to the opportunity to use rich media which yield increased recombinant protein productions (Loison et al., 1989; Napp and Da Silva, 1993). Recently an autoselection system has also been described for Kluyveromyces lactis (Fleer, 1992). *Corresponding author. CCC 0749-503X/96/030199-07 0 1996 by John Wiley & Sons Ltd Since autoselection systems are relatively new, little information is available on the plasmid distribution in the cellular population. More knowledge on the state(s) of the population and on the cellular distribution of the plasmid could be used to improve the production processes. Flow cytometry allows the study of the behaviour of a cell population following the analysis of bioparameters at the single-cell level. Each cellular population is heterogeneous with respect to cellular states such as mass, age and biochemical composition. The distributions of these properties in the whole population provide information that usually is lost when average properties of the population are considered. The analysis of population growth dynamics (Alberghina and Porro, 1993), of the propagation of recombinant plasmids (Wittrup et al., 1990) and of the metabolic state of a cellular population (Porro et al., 1994) are some examples of the application of this technique to the studies of growing yeast populations. We have previously reported the development of an autoselection system based on the use of the gene FBAZ (i.e. encoding the FDP aldolase enzyme) to stabilize expression vectors in S. cerevisiae host cells bearing a disruption of the 200 chromosomal FBAl gene. By using the inducible UAS,,, promoter to regulate the expression of the FBAI gene, we have also obtained a system in which the plasmid copy number is strongly modulated by the carbon source used for growth (Compagno et al., 1993). In this work we report the studies performed to gain insight into the population structure and the cellular plasmid distribution in yeast populations transformed with the autoselection plasmid described above. By means of flow cytometric procedures, we were able to study dynamics of selection of a sub-population with a higher plasmid copy number during transient state(s) of growth. MATERIALS AND METHODS Strain and growth conditions The S. cerevisiae haploid strain WA6[pIA10] ( M A T a, ade2-1, canl-100, Jbal::URA3, leu2-3, trpl-1, his3-11,15), in which the plasmid pIAlO (carrying the wild-type gene for aldolase) complements the disruption of the corresponding chromosomal gene, was used in this study (Compagno et al., 1993). Yeast cells were grown at 30°C in minimal medium containing 0.67% (wh) Yeast Nitrogen Base without amino acids (Difco Laboratory, Detroit, MI, U.S.A.) supplemented with the required amino acids (50 pg ml- I). The carbon source was 2% (w/v) glucose or 2% (wh) galactose. In all cases, the cultures were grown in flasks in a shaking incubator. For solid media, 2% (w/v) agar was added. The cell number concentration was determined, after sonication and appropriate dilution with Isoton (Coulter Electronics, England), with a Coulter Counter ZBI equipped with a 70 pm orifice. Plasmid construction and shufling Standard DNA manipulations were performed according to Sambrook et al. (1989). Plasmid pIlOG was constructed from pIAlO (Compagno et al., 1993). The Escherichia coli LacZ gene was obtained by the BamHI-StuI digestion of the plasmid pEl (Compagno et al., 1991) and was inserted into the BamHI site of the plasmid pIA10, obtaining the plasmid pI1OG. The plasmid pIlOG was used to transform WA6[p13A] cells and plasmid shuffling was induced, selecting Trp+lLeu- cells in which the plasmid p13A was substituted by the PI 1OG plasmid. C . COMPAGNO ET AL. Determination of the plasmid copy number Total yeast DNA was prepared by the method described by Nasmyth and Reed (1980). Total DNA was digested with HindIII, run on agarose gel and stained with ethidium bromide. The most prominent bands seen were those due to plasmid restriction fragments and those due to the restriction fragments of ribosomal DNA. The gels were photographed and plasmid copy number was determined by comparing the band density of plasmid DNA with that of ribosomal DNA taken as internal control (100 copies per cell; Broach, 1983) using a scanning densitometer. The plasmid copy number determination related to t=O in Table 2 was determined by a classical Southern approach (Compagno et al., 1993). Enzyme assays Aldolase and P-galactosidase assays were performed as previously described (Compagno et al., 1993). ImmunoJiuorescent staining The P-galactosidase immunofluorescent staining was based on the procedure described by Eitzman and Srienc (1991) with some modifications. Each sample containing approximately 3 x lo7 cells was collected, centrifuged and resuspended in 5 ml of a fixation solution (3.7% formaldehyde; 1.3 Msorbitol in saline phosphate buffer (PBS: 3.3 mMNaH,PO,, 6.7 ~ M - N ~ , H P O ,0.2 mM-EDTA, 130 mM-NaC1, pH 7.5)) and incubated for 90 min at room temperature (25°C). After recovery, the cells were washed twice in 1-3M-sorbitol in PBS and resuspended in a digestion solution containing 8 mg ml of Zymolyase 20T (ICN, Biomedicals), 27 mM-P-mercaptoethanol in 1.3 M-sorbitol-PBS. The mixture obtained was vortexed gently and incubated at 37°C for 30 min. The cells were then centrifuged (2000 rpm, 10 min), washed once with cold PBS and resuspended in 2.5 ml of cold PBS. A 100 p1 aliquot of each digested cell suspension was resuspended in a microcentrifuge tube with 100 pl of a warm (37°C) PBS solution containing 10 mg ml - of bovine serum albumin (Sigma) and the mouse anti-P-gal antibody (diluted 1:lOOO from the stock solution; Sigma, Immuno Chemicals). The cell suspension was incubated at 37°C for 1 h, recovered by centrifugation (2000 rpm, 10 min) and washed with 200pl of PBS. A second wash was performed for 20 min at 37°C in PBS. After recovery, cells were resuspended in a solution ~ 20 1 AUTOSELECTION OF HIGH PLASMID COPY NUMBER containing 2 pg/ml affinity-purified goat antimouse F(ab’)2 fragment conjugated with fluorescein isothiocyanate (FITC; Boehringer Mannheim, Biochemicals) in PBS and incubated for 75 min at 37°C. After centrifugation, cells were washed twice in 200yl of PBS, resuspended in PBS, sonicated and analysed by flow cytometry. Ethidium bromide, Concanavalin A (ConA)-FITC staining andJlow cytometric analyses Staining of dead cells with ethidium bromide, staining of cell walls with conjugated ConA-FITC and flow cytometric analyses were performed as previously described (Martegani et al., 1993; Porro and Srienc, 1995, respectively). E J -I w 10’ v 0 10 20 30 40 50 TIME (hours) Figure 1. Growth curve of WA6[pIA10] transformed cells during carbon source shift from galactose- to glucosecontaining medium. Cells precultured on galactose medium were inoculated on glucose medium ( t = O ) . After 24 h of growth, cells were reinoculated at a lower cell density in fresh glucose medium. An unusually long lag phase can be observed. RESULTS AND DISCUSSION Analysis of cellular viability and growth properties of transformed yeast cells during a carbon source s h f t The effect of the expression level of the selective marker on the plasmid copy number has been documented (Erhart and Hollemberg, 1983; Piper and Curran, 1990). We have previously shown that a carbon-source-dependent modulation of the plasmid copy number can be also obtained by cloning of the FBAl gene, the autoselective marker, under the control of the UASGAL,.IO promoter (i.e. plasmid pIA10) in cells bearing a disruption of the chromosomal FBAl gene (i.e. strain WA6; Compagno et al., 1993). WA6[pIA10] S. cerevisiae transformed cells growing on galactose media exhibited a high level of aldolase (35 U mg - ’) in spite of a low plasmid copy number (about five copies per cell), due to the effect of UASGA, induction on FBAl gene expression. On the other hand, in glucose media the repression of the FBAl gene expression resulted in a low aldolase level (0.2 1 U mg - I ) and a high plasmid copy number (about 80 copieskell). These observations led us to wonder how yeast cells succeed in amplifying the plasmid copy number. At least two different mechanisms could occur. In one case, the plasmid copy number in each cell could increase over time; on the other hand, the increase of plasmid copy number should be related to the selection of transformed cells bearing a higher plasmid copy number. In order to discriminate between the two hypotheses, we studied the viability of the population and the dynamics of growth during a typical transient state of growth (Figure 1). The viability of the population during shift from galactose- to glucose-containing media was assessed by flow cytometry. The method to estimate yeast cell viability relies on the uptake from dead/damaged cells of ethidium bromide, a dye which is normally excluded from viable cells (De la Fuente et al., 1992; Martegani et al., 1993). Analysis of staining patterns over time after the carbon source shift (Figure 2) showed a deep alteration of the cellular permeability during the lag phase (B=27 h). In parallel with a restored growth of the population, a cellular subpopulation with a low associated fluorescence gradually rose (C=42 h, D=48 h). Such analysis indicates that during the transient state of growth, a relevant fraction of the transformed WA6[pIA10] yeast cells died. Such a phenomenon could be related to the long lag phase observed (see Figure l), while the new growth following the lag phase should be related to the accumulation of viable cells. In parallel, cell viability was estimated by the classical approach of the plate-forming units obtained by plating on glucose samples taken from the culture at different times during the lag phase. In agreement with the flow cytometry analysis, we observed that at the beginning of the lag phase, most of the population was unable to grow on glucose (i.e. only 0.5% of the population is viable), while the number of colonies increased in the course of this phase (at 31 h the fraction of viable cells was 2% and at 46 h it was 9%). It is interesting to note that the total amount of plasmid in all the individual colonies analysed was high (60-80 copieskell, data not shown). 202 C. COMPAGNO ET AL. d L -1 ii FSC d K W 8 -I CONFI-FITC W FSC LIN( FL2 1 Figure 2. Analysis of cell viability of WA6[pIAlO] transformed cells during carbon source shift from galactose- to glucose-containing medium. Histograms A-D refer to the samples withdrawn at 24 h (A), 27 h (B), 42 h (C) and 48 h (D) of the experiment shown in Figure 1, treated with ethidium bromide and analysed by flow cytometry. Lin(FL2) is the fluorescence signal (linear scale) related to ethidium bromide. Recently a novel flow cytometric procedure has been developed to obtain information on the growth properties of individual S. cerevisiue cells in asynchronous culture. The method is based on labelling the cell walls of growing cells with the lectin ConA conjugated to a fluorescent marker such as FITC. Cells growing in balanced growth conditions are collected, stained with ConA-FITC and resuspended in fresh medium. Because formation of new cell wall material in budded cells is restricted to the bud tip, exposure of the stained cells to growth conditions results in the production of newborn daughter cells with a gradual decrease in surface staining, which can thus easily be identified from the overall growing population (Porro and Srienc, 1995). Transformed WA6[pIA 101 yeast cells, precultured on galactose medium, were inoculated on glucose medium and, 24 h after the nutritional shift, were stained with ConA-FITC and resuspended in fresh glucose medium. At this time all the cell walls were completely stained (Figure 3A). Figure 3B shows the staining behaviour 7 h later (corresponding to t=31 h of growth on glucose in Figure 1). On the right the evolution of partially stained newborn daughter cells is Figure 3. Dynamics of the cell-wall-tag for cells grown on glucose. WA6[pIAIO] cells, cultured as described in Figure 1, were stained with ConA-FITC after 24 h of growth on glucose (A) and resuspended in fresh glucose medium. (B) 7 h after resuspension partially stained cells (on the right) evolve as newborn daughter cells; this point corresponds to 31 h of growth on glucose. (C) 18 h after resuspension (42 h of growth on glucose). (D) 24 h after resuspension (48 h of growth). FSC is flow cytometric determination of cell volume (linear scale); ConA-FITC is fluorescence signals related to Concanavalin A conjugated to fluorescein (logarithmic scale). The arrows indicate the direction of axes. clearly evident. If it is assumed that the area of the measured light-scattering signals at low angles (FSC) reasonably reflects the size of cells, one can directly estimate from these data some growth properties of the growing transformed yeast population. For example, it is interesting to note that the average size of the overall population increased during the first 7 h (compare FSC signals in Figure 3B and A). Even if during this time it is not possible to note an appreciable increase of the cell number concentration, all the above considerations taken together indicate that the lag time observed is only apparent. In fact, from the area of the fluorescence signals associated with the newborn cells, it is possible to calculate that this population evolves with a doubling time of about 5 h. Figures 3C and D are related to the staining pattern observed 18 h (t=42 h of growth on glucose in Figure 1) and 24 h (t=48 h in Figure 1) later. The figure clearly shows the selection of a new population and its increase over time. 203 AUTOSELECTION OF HIGH PLASMID COPY NUMBER - Table 2. Time-course of the specific 0-galactosidase I activity and of the plasmid copy number during the E 2 growth of the WA6[pIAlOG] population on glucose media. W m I 3 ~ Z Time (h)* -1 -I ~~ P-gal activity (U I*g - '> COPY number W 0 0 10 20 30 40 50 TIME (hours) Figure 4. Growth curve of WA6[pIlOG] transformed cells during carbon source shift from galactose- to glucosecontaining medium. Cells precultured on galactose medium were inoculated on glucose medium (t=O). After 24 h of growth, cells were reinoculated at a lower cell density in fresh glucose medium. 0 24 32 48 1 5 15 17 5 20 nd 80 *The values reported at the t=O refer to cells cultured on galactose medium. At this time, cells were inoculated on glucose medium, as described in Figure 4. Table 1. Physiological parameters of WA6[pIA10] and WA6[pIlOG] strains after 24 h of growth on glucose medium. WA6[pIA101 Cell number mi - ' Aldolase (U mg ') Plasmid copy number ~ P-gal (U I*g- '1 5 x 106 0.05 10 - WA6[pI1OG] 107 0.2 20 5 Analysis of cellular plusmid distribution during the carbon source shvt In order to correlate more tightly the dynamics of growth to the increased plasmid content in the population growing on glucose, we analysed the distribution of the plasmid copy number at singlecell level during the nutritional shift. For this purpose, the E. coli LacZ gene was inserted in the plAlO plasmid under the control of the constitutive FBAI promoter. The resulting plasmid pI1OG was used to obtain the WA6[pIlOG] strain (as described in Materials and Methods). In this way the LacZ gene expression should be proportional to the plasmid copy number. WAG[pIlOG] transformed cells were cultured under the same conditions as shown in Figure 1 (Figure 4). On one hand, a higher cell density than that observed for the WA6[pIA10] cells was achieved after 24 h of growth on glucose, while on the other hand a shorter lag phase was observed when cells were resuspended in fresh glucose medium. Table 1 summarizes some cellular parameters of the two transformed yeast populations. The aldolase levels L.OG( FL Figure 5. Time-course analysis of fluorescence signals related to P-galactosidase levels in WA6[pIlOG] cells, cultured as described in Figure 4. (A) Cells grown on galactose; (B) cells 24 h after the nutritional shift; (C) cells after 48 h of growth on glucose. Log(FL1) is the logarithm of fluorescence signals. from WA6[pIlOG] cells after 24 h in glucose were higher than from WA6[pIA10] cells. In addition, the pIlOG plasmid content after 24 h of growth on glucose was higher than the level of the pIAlO plasmid. Table 2 compares the behaviour of the specific P-galactosidase activity with the behaviour of the plasmid copy number during 48 h of growth on glucose. Analysis of the data reported supports the assumption that the P-galactosidase content is proportional to the plasmid copy number. Histogram A in Figure 5 reports the fluorescence signals related to the P-galactosidase levels in WA6[pI 10G] transformed cells growing on galactose. The plasmid copy number in these cells is 204 about five, as determined by Southern hybridization of the total D N A content (time t=O in Table 2). Histogram B reports the fluorescence signals related to the P-galactosidase level in WA6[pI 10G] transformed cells growing for 24 h on glucose. A quantitative analysis of this bimodal histogram indicated that for about 60% of the transformed yeast population, the level of heterologous enzyme is similar to that of the galactose culture (histogram A), while for the remaining yeast population, a higher content is clearly visible. Finally, histogram C is related to the p-galactosidase content from transformed yeast cells growing for 48 h on glucose. Analysis of the staining pattern shown in Figure 5 clearly indicates that the level of the heterologous enzyme rose over time as a consequence of the transient state of growth (i.e. shift from galactose- to glucose-containing media). Therefore, such increase of the heterologous activity (i.e. of the plasmid copy number) seems to be related to the selection of transformed cells bearing a higher plasmid copy number. The unequal segregation of plasmid molecules a t cellular division (Futcher, 1988; Morrissey and Cashmore, 1992) could be the mechanism responsible for the selection of cells with a higher plasmid content and thus with an aldolase level enabling growth on glucose. The higher plasmid copy number observed in the case of plasmid pIlOG in comparison to plasmid pIAlO (see Table l), could allow a more rapid selection of the sub-population, as indicated by the shortened lag phase after 24 h of growth o n glucose (see Figures 1 and 4). A t this time, at least 30% of the population (Figure 5B) is already able to grow because of the higher plasmid content. In conclusion, in this work we have shown that the singular feature of a carbon-source-dependent modulation of the plasmid copy number coupled to flow cytometric investigations allows the correlation of the distribution of a cellular component, the plasmid, to dynamics of growth and selection of a specific sub-population. Finally, the potential of this kind of autoselection system should be further tested with regard to both the long-term stability of the amplified copy number and the yield of heterologous protein. Studies along these lines are in progress in our laboratory. ACKNOWLEDGEMENTS This project was supported by CNR, Progetto Finalizzato Biotecnologie e Biostrumentazioni, subproject no. 3 (B.M.R.). C . COMPAGNO ET AL. REFERENCES Alberghina, L. and Porro, D. (1993). Quantitative flow cytometry: analysis of protein distribution in budding yeast: A mini review. Yeast 9, 815-823. Ayub, M. A. Z . , Astolfi-Filho, S., Mavituna, F. and Oliver, S. G. (1992). Studies on plasmid stability, cell metabolism and superoxide dismutase production by strains of Saccharomyces cerevisiae. Appl. Pgk Microbiol. Biotechnol. 37, 615-620. Broach, J. R. (1983). Construction of high copy yeast vector using 2-pm circle sequences. In Wu, R., Grossman, L. and Moldave, K. (Eds), Methods in Enzymology, vol. 101. Academic Press, Inc., New York, pp. 307-325. Compagno, C., Ranzi, B. M. and Martegani, E. (1991). The promoter of Saccharomyces cerevisiae FBAI gene contains a single positive upstream regulatory element. FEBS Lett. 293, 97-100. Compagno, C., Tura, A., Ranzi, B. M., Alberghina, L. and Martegani, E. (1993). Copy number modulation in an autoselection system for stable plasmid maintenance in Saccharomyces cerevisiae. Biotechnol. Prog. 9, 594-599. De la Fuente, J. M., Alvarez, A., Nombela, C. and Sanchez, M. (1992). Flow cytometric analysis of Saccharomyces cerevisiae autolytic mutants and protoplasts. Yeast 8, 3945. Eitzman, P. D. and Srienc, F. (1991). Dynamics of activation of a galactose-inducible promoter in Saccharomyces cerevisiae. J. Biotechnol. 21, 63-82. Erhart, E. and Hollemberg, C. P. (1983). The presence of a defective LEU2 gene in 2 pm DNA recombinant plasmid of Saccharomyces cerevisiae is responsible for curing and high copy number. J. Bacteriol. 156, 625-635. Fleer, R. (1992). Engineering yeast for high level expression. Current Opinion in Biotechnology 3, 486496. Futcher, A. B. (1988). The 2 pm circle plasmid of Saccharomyces cerevisiae. Yeast 4, 2740. Loison, G., Nguyen-Juilleret, M., Alouani, F. and Marquet, M. (1986). Plasmid-transformed URA3 FUR1 double mutant of Saccharomyces cerevisiae: an autoselection system applicable to the production of foreign proteins. BiolTechnology 4, 433437. Loison, G., Vidal, A,, Findelli, A., et al. (1989). High level of expression of a protective antigen of schistosomes in Saccharomyces cerevisiae. Yeast 5, 497-507. Ludwig, D. L., Ugolini, S. and Bruschi, C. V. (1993). High-level heterologous gene expression in Saccharomyces cerevisiae from a stable 2 pm plasmid system. Gene 132, 3340. Martegani, E., Brambilla, L., Porro, D., Ranzi, B. M. and Alberghina, L. (1993). Alteration of cell population structure due to cell lysis in Saccharomyces cerevisiae cells overexpressingthe GAL4 gene. Yeast 9, 575-582. ~ AUTOSELECTION OF HIGH PLASMID COPY NUMBER Morrissey, J. A. and Cashmore, A. M. (1992). Use of a cis-acting mutation to study the role of FLP-mediated recombination in the maintenance of native yeast 2 pm plasmids. Mol. Microbiol. 6, 3101-3107. Napp, S. J. and Da Silva, N. A. (1993). Enhancement of cloned gene product synthesis via autoselection in recombinant Saccharomyces cerevisiae. Biotechnol. Bioeng. 41, 801-810. Nasmyth, K. A. and Reed, S. I. (1980). Isolation of genes by complementation in yeast, molecular cloning of a cell-cycle gene. Proc. Natl. Acud. Sci. USA 77, 21 19-2123. Piper, P. W. and Curran, B. P. G. (1990). When a glycolytic gene on a yeast 2 p ORI-STB plasmid is made essential for growth its expression level is a major determinant of plasmid copy number. Curr. Genet. 17, 119-123. Porro, D., Smeraldi, C., Martegani, E., Ranzi, B. M. and Alberghina, L. (1994). Flow-cytometric determination of the respiratory activity in growing 205 Saccharomyces cerevisiae population. Biotechnol. Prog. 10, 193-197. Porro, D. and Srienc, F. (1995). Tracking of individual cell cohorts in asynchronous Saccharomyces cerevisiae population. Biotechnol. Prog. 11, 342-347. Rech, S. B., Stateva, L. I. and Oliver, S. G. (1992). Complementation of the Saccharomyces cerevisiae srbl-1 mutation: an autoselection system for stable plasmid maintenance. Curr. Genet. 21, 339-344. Sambrook, J., Fritsch, E. F. and Maniatis, T. (1989). Molecular Cloning. A Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Unternahrer, S., Pridmore, D. and Hinnen, A. (1991). A new system for amplifying 2 l m plasmid copy number in Saccharomyces cerevisiae. Mol. Microbiol. 5, 1539-1 548. Wittrup, K. D., Bailey, J. D., Ratzkin, B. and Patel, A. (1990). Propagation of an amplifiable recombinant plasmid in Saccharomyces cerevisiae. Flow cytometry studies and segregated modeling. Biotechnol. Bioeng. 35,565-577.