Aquaporin-Mediated Improvement of Freeze Tolerance of Saccharomyces cerevisiae Is Restricted to Rapid Freezing Conditions

Previous observations that aquaporin overexpression increasesthe freeze tolerance of baker's yeast (Saccharomyces cerevisiae)without negatively affecting the growth or fermentation characteristicsheld promise for the development of commercial baker's yeaststrains used in frozen dough applications. In this study wefound that overexpression of the aquaporin-encoding genes AQY1-1and AQY2-1 improves the freeze tolerance of industrial strainAT25, but only in small doughs under laboratory conditions andnot in large doughs under industrial conditions. We found thatthe difference in the freezing rate is apparently responsiblefor the difference in the results. We tested six different coolingrates and found that at high cooling rates aquaporin overexpressionsignificantly improved the survival of yeast cells, while atlow cooling rates there was no significant effect. Differencesin the cultivation conditions and in the thawing rate did notinfluence the freeze tolerance under the conditions tested.Survival after freezing is determined mainly by two factors,cellular dehydration and intracellular ice crystal formation,which depend in an inverse manner on the cooling velocity. Inaccordance with this so-called two-factor hypothesis of freezinginjury, we suggest that water permeability is limiting, andtherefore that aquaporin function is advantageous, only underrapid freezing conditions. If this hypothesis is correct, thenaquaporin overexpression is not expected to affect the leaveningcapacity of yeast cells in large, industrial frozen doughs,which do not freeze rapidly. Our results imply that aquaporin-overexpressingstrains have less potential for use in frozen doughs than originallythought.

INTRODUCTION

Top

Introduction

Although much correlative evidence is available, the determinantsof freeze resistance in Saccharomyces cerevisiae are largelyunknown (for a review see reference 34). The injury sustainedduring freezing and thawing is caused by a combination of multipletypes of stress imposed on the cells, such as changes in temperature,water content, water state, pH, and free radical, ion, and soluteconcentrations. The concomitant loss of survival is not attributableto any one form of injury and depends on the cooling and warmingrates, the final temperature, the duration of freezing, andthe suspending medium (23, 24).

When a cell suspension is cooled to a temperature of 0°Cor less, both the suspending medium and the cells initiallysupercool. Extracellular ice crystal formation precedes intracellularfreezing and is determined by the freezing point of the suspendingmedium and the presence of ice-nucleating agents. Based onlyon osmolarity, the freezing point of the cytoplasm is predictedto be approximately –1°C. Nevertheless, the cell interiortypically remains unfrozen until the temperature is –10to –15°C. This intracellular supercooling may be dueto prevention of growth of ice into the cell interior by thecell membrane and due to the lack of nucleators of supercooledwater within the cell (23, 24).

Following external freezing and internal supercooling, a chemicalgradient for free water is established between the extracellularice-containing water and the intracellular supercooled water.This thermodynamically unstable situation can be resolved eitherby water outflow or by internal freezing. At low cooling rates,ice crystal formation remains mainly extracellular, and wateroutflow from the cell is sufficient to restore the chemicalequilibrium and minimize the intracellular supercooling. Ifcells are cooled rapidly, extensive intracellular supercoolingoccurs, and the chemical gradient is eliminated by freezingcellular water. The critical cooling rate at which intracellularice crystals are formed is determined by the cell type, particularlythe surface-to-volume ratio of the cell and the water permeability;larger, spherical cells and cells that are less permeable towater have a lower critical cooling rate (23, 24). Althoughthe characteristics of primary lesions resulting from slow andrapid freezing are different, both processes are detrimentalto the cell. This is the so-called two-factor hypothesis offreezing, which postulates that survival after freezing is mainlyaffected by two factors, cellular dehydration and intracellularice crystal formation, which depend in opposite ways on thecooling velocity (23, 24).

Several cellular factors affect survival of yeast cells followingfreezing and thawing (for a review see reference 34). High trehaloselevels (1, 6, 12, 13, 36) and high levels of heat shock proteinexpression (17) largely explain the freeze resistance acquiredfollowing exposure to mild forms of other stresses (22) andthe influence of growth conditions on yeast freeze tolerance(10, 27, 30, 36). Data on the importance of cold shock proteins(18, 19) for freeze resistance are contradictory (16, 30). Giventhe oxidative and osmotic stress component of freeze stress,superoxide dismutases (29) and compatible solutes (28, 31, 32)also seem to protect yeast cells upon freezing. The importanceof plasma membrane characteristics, such as fluidity and permeability,in relation to yeast freeze resistance has not been resolved(4, 20, 21). Finally, anti-freeze proteins (25) and hydrophylins(14, 15) are known to increase freeze tolerance.

Freeze tolerance and expression of the aquaporin-encoding genesAQY1 and AQY2 have been correlated based on a genome-wide geneexpression analysis of the freeze-resistant industrial mutantstrain AT25 (33). Overexpression of AQY1-1 or AQY2-1 in thisstrain increased freeze tolerance, both in cell suspensionsand in small doughs, without negatively affecting growth orfermentation characteristics. Thus, the modification of aquaporinexpression held promise for improving the freeze tolerance ofcommercial baker's yeast strains used in frozen dough applications.Unfortunately, aquaporin-overexpressing strains in industrialconditions were no more freeze tolerant than the controls were.

The goal of this work was to identify the cause of the differencebetween the results obtained with small doughs under laboratoryconditions and the results obtained with large doughs underindustrial conditions for freeze protection by aquaporin overexpression.We evaluated the cultivation conditions, freezing rate, andthawing rate as possible contributing factors. The results areimportant for understanding the molecular basis of freeze damageand the protective effect of aquaporin overexpression, as wellas for improving industrial baker's yeast strains used in thepreparation of frozen doughs.

MATERIALS AND METHODS

Top

Materials and Methods

Strains, plasmids, and culture conditions.
The strains used in this study were the laboratory strain BY4743(MATa/MAT ${alpha}$ his3 ${Delta}$ 1/his3 ${Delta}$ 1 leu2 ${Delta}$ 0/leu2 ${Delta}$ 0 ura3 ${Delta}$ 0/ura3 ${Delta}$ 0) (3) and the industrialstrain AT25 (± diploid, probably aneuploid) (35). Inthese strains, pYX012 KanMX (vector), pYX012 KanMX/AQY1-1 (pAQY1-1),and pYX012 KanMX/AQY2-1 (pAQY2-1) have previously been integratedat the TPI locus, resulting in geneticin-resistant strains (33).In laboratory conditions, cells were routinely grown in YP (1%[wt/vol] yeast extract, 2% [wt/vol] Bacto Peptone [Difco]) with2% glucose (YPD) at 30°C in an orbital shaker at 200 rpm.In addition, production under pilot-scale conditions was performedin fed-batch mode (20-liter fermentor containing 10 liters ofmolasses medium) (36).

RGC after freezing.
The residual glucose consumption (RGC) after fermentation andfreezing was determined by comparing the average glucose consumptionof two frozen samples (FGC) to the average glucose consumptionof two control samples (IGC), as described previously [RGC =(FGC/IGC) x 100] (33). Briefly, equal amounts of stationary-phasecells (40-µl cell suspensions at an optical density at600 nm of 20 for laboratory strains and at an optical densityat 600 nm of 15 for industrial strains) were frozen 30 min afterthe addition of glucose in order to mimic the production offrozen dough, in which the dough is frozen after mixing, kneading,and eventually a short prefermentation period. After thawing,400 µl of YP containing 33 mM glucose was added to boththe control cells and the stressed cells, and after incubationat 30°C for 3 h (laboratory strains) or for 2.5 h (industrialstrains) the amount of glucose taken up was measured.

Routinely, cells were frozen in standard 1.5-ml Eppendorf microcentrifugetubes (VWR International, Leuven, Belgium) (referred to as smalltubes or single plastic layer below) in an ethanol bath at –30°Cfor 1 h, stored in a freezer at –30°C for 1 day, andthawed at room temperature (between 18 and 22°C). To evaluatethe effect of several successive rapid freezing treatments,the tubes were frozen in an ethanol bath at –30°Cfor 1 h and thawed at room temperature for 30 min, and thisprocedure was repeated up to 10 times.

Occasionally, cells were frozen slowly by putting the smalltubes in a standard 50-ml plastic centrifuge tube (Falcon; BectonDickinson, Erembodegem, Belgium) (referred to as large tubeor double plastic layer below) stuffed with paper tissue (tocreate an insulating layer) or by putting the small tubes ina freezer at –30°C, as individual tubes or in a largetube. To evaluate the effect of several successive slow freezingtreatments, the tubes were frozen in a freezer at –30°Cfor 1 h and thawed at room temperature for 30 min, and thisprocedure was repeated up to 30 times. Occasionally, cells werefrozen very rapidly (by immersion of the small tubes in liquidnitrogen, as individual tubes or in a large tube). Unless indicatedotherwise, cells were thawed at room temperature. Occasionally,cells were thawed more slowly (on ice) or more rapidly (in awater bath at room temperature or at 30°C). During freezing,temperature changes for four of these freezing conditions weremonitored by using a thermocouple (type PT; diameter, 3 mm;Fisher Bioblock Scientific, Illkirch, France) connected to adata collector (Quick; Bioblock).

Survival in frozen dough. (i) Laboratory conditions.
The level of survival (expressed as a percentage of CFU) afterfermentation and freezing was determined by comparing the averagenumber of CFU isolated from two frozen doughs to the averagenumber of CFU isolated from two nonfrozen control doughs, asdescribed previously (33). Most doughs were frozen rapidly at–30°C in an ethanol bath or slowly in a freezer at–30°C; the exceptions were two nonfrozen controlsthat were analyzed immediately. After 1 h, the samples thatwere frozen in the ethanol bath were stored in the freezer at–30°C.

(ii) Industrial conditions.
At Lesaffre Développement, lean doughs were preparedby using the following formulation (3,000 g of flour per batch):100 g of flour (14.5 to 15.0% protein on a 14% moisture basis),63 g of water (with crushed ice), 2 g of salt (NaCl), 6 g (1.6%on a dry weight basis) of compressed yeast (from pilot-scaleproduction), and 1 g of improver (Ibis Bleu; Lesaffre InternationalIngrédients, Marquette Lez Lille, France). For sugar-containingdoughs, 10% saccharose was added. The target dough temperaturewas 19°C at all stages of preparation. The dough was kneadedmechanically for 15 min with a spiral mixer (FPI 50; VMI, Montaigu,France) and subsequently divided and molded. After prefermentationfor 30 min in an air-conditioned room at 19°C, dough portionswere frozen at –35°C with ventilation until the coretemperature was –20°C. The temperature profiles inthe cores of the dough portions were measured with a thermocouple(type K; diameter, 1.3 mm; Bioblock) connected to a data collector(HotBox; Bioblock). The doughs were subsequently stored at –20°C.At different times a dough portion was thawed at 0°C, theyeast gassing power was measured with a Zymotachigraph at 27°Cfor 2 h, and the dough proofing time was determined at 35°C.

RESULTS

Top

Results

Effect of aquaporin overexpression on freeze tolerance in industrial conditions.
For large, industrial frozen doughs, no significant effect ofaquaporin overexpression on the gassing power of the yeast (Fig.1A) or the proofing time of the dough (Fig. 1B) was observed.This was the case for lean doughs (no saccharose) (Fig. 1),as well as for sugar-containing doughs (10% saccharose) (resultsnot shown). These freezing conditions resulted in an averagecooling rate of 0.6 to 0.8°C per min in the core of thedough (Fig. 1C).

View larger version (21K):
[in this window]
[in a new window]
FIG. 1. (A and B) Evaluation of yeast gassing power (A) and dough proofing time (B) during frozen storage of large industrial doughs made with AT25 overexpressing AQY1-1 (•) or AQY2-1 (x) and with AT25 ( ${blacksquare}$ ) and a control strain containing an empty plasmid ( ${circ}$ ). The means and standard deviations for two independent tests are shown; the standard deviations for data points without error bars are less than 10% of the values of the points. (C) Temperature in the core of large, industrial doughs during freezing at –35°C with ventilation at Lesaffre Développement. The curve represents the average values for 64 measurements (two doughs, eight horizontal positions, and four vertical positions).

Importance of growth conditions.
The freeze tolerance of industrial strain AT25, as determinedby calculating the RGC after the cells were frozen in an ethanolbath at –30°C, was significantly improved by overexpressionof AQY1-1 or AQY2-1, both with the cells cultured under laboratoryconditions and with the cells cultured under industrial conditions(results not shown). Essentially the same results were obtainedwhen survival was determined for small doughs frozen in an ethanolbath at –30°C and then stored at –30°C ina freezer; the strain overexpressing AQY2-1 survived the initialfreezing step and frozen storage better than the control strain,and it made no difference whether the strains were culturedunder laboratory or industrial conditions (Fig. 2A).

View larger version (21K):
[in this window]
[in a new window]
FIG. 2. Freeze tolerance of AT25 and AT25 overexpressing AQY2-1 in small frozen doughs, either cultured under laboratory conditions and harvested from liquid medium ( ${lozenge}$ and ${circ}$ , respectively) or produced at a pilot scale and resuspended from yeast cake ( ${blacklozenge}$ and •, respectively), frozen in an ethanol bath at –30°C (A) or in a freezer at –30°C (B). Survival is expressed as the percentage of CFU isolated from two frozen doughs based on the number of CFU in two nonfrozen control doughs. The standard deviations for replicate samples are indicated; the standard deviations for data points without error bars are less than 10% of the values of the points.

Importance of freezing conditions.
When survival was determined in small doughs that were slowlyfrozen by placing them in a freezer at –30°C, no differencewas observed between the control strains and the strains overexpressingaquaporins (Fig. 2B). Thus, we evaluated the effect of the coolingrate in more detail. The cells generally remained highly viableunder slow freezing conditions, whereas under rapid freezingconditions the RGC was significantly or even dramatically decreased(Table 1). Cells that were frozen in a small tube in liquidnitrogen had a low RGC rate that was significantly improvedby overexpression of AQY1-1 or AQY2-1. Reduction of the coolingrate by use of a double plastic layer significantly improvedthe RGC of strains expressing AQY1-1 or AQY2-1 (Table 1). Theimprovement in the freeze tolerance due to aquaporin overexpressionupon rapid freezing by immersion of a small tube directly inan ethanol bath at –30°C is consistent with the resultsobtained with other strain backgrounds (33). These freezingconditions resulted in an average cooling rate of 30 to 35°Cper min (Fig. 3). Use of a double plastic layer and immersionin an ethanol bath at –30°C resulted in a lower coolingrate (10 to 15°C per min) (Fig. 3) and much higher RGC,especially in case of the control strains (Table 1). Althoughthe RGC was still less than 100%, there was no significant differencebetween the controls and the strains overexpressing aquaporins.Similarly, when the cells were frozen by placing the vials ina freezer, the RGC was less than 100% and was not affected byaquaporin overexpression (Table 1).

View this table:
[in this window]
[in a new window]
TABLE 1. Freeze tolerance of laboratory strain BY4743 and industrial strain AT25, with and without AQY1-1 or AQY2-1 overexpression, grown under laboratory conditions and tested in different freezing conditions^a

View larger version (19K):
[in this window]
[in a new window]
FIG. 3. Temperature courses in cell suspensions during freezing in the ethanol bath (single plastic layer [ ${blacklozenge}$ ] or double plastic layer [ ${blacksquare}$ ]) and in the freezer at –30°C (single plastic layer [ ${blacktriangleup}$ ] or double plastic layer [x]) (average of two measurements). For comparison, the first 10 min of the temperature course in the core of large, industrial dough during freezing at –35°C with ventilation at Lesaffre Développement (Fig. 1C) also is shown.

After only one cycle of slow freezing the RGC was still veryhigh, which made any improvement due to aquaporin overexpressiondifficult to detect. Successive slow freezing treatments (upto 30 treatments) were used to lower the overall RGC and tomore accurately measure any protective effect of aquaporin overexpressionin the slow freezing conditions (Table 2), and the results werecompared to the RGC for successive rapid freeze treatments (upto 10 treatments) (Table 2). After 1, 5, and 10 cycles of rapidfreezing, there was a two- to threefold increase in freeze tolerancein the aquaporin-overexpressing strains, whereas after the samenumber of slow freezing cycles no significant effect of aquaporinoverexpression was observed (Table 2). Thus, in general, strainsoverexpressing aquaporin were more freeze tolerant than thecontrol strains, but the degree of improvement depended on thefreezing rate.

View this table:
[in this window]
[in a new window]
TABLE 2. Freeze tolerance of laboratory strain BY4743 and industrial strain AT25, with and without AQY1-1 or AQY2-1 overexpression, grown under laboratory conditions and tested after successive rapid or slow freezing treatments^a

Importance of thawing conditions.
As described above, overexpression of aquaporin significantlyimproved freeze tolerance in all the conditions tested, butno significant differences in RGC were observed based on thethawing conditions. The RGC means ± standard deviationsfor the different conditions were as follows: BY4743 plus vector,7.0% ± 3.0%; BY4743 plus pAQY1-1, 69% ± 4.8%;BY4743 plus pAQY2-1, 38% ± 5.8%; AT25 plus vector, 27%± 2.0%; AT25 plus pAQY1-1, 90% ± 3.9%; and AT25plus pAQY2-1, 100% ± 2.5%.

DISCUSSION

Top

Discussion

The starting point of this study was the observation that incontrast to results obtained in small, laboratory doughs (33),no effect of aquaporin overexpression on the leavening capacityof yeast cells could be observed in large, industrial frozendoughs (Fig. 1). Laboratory and industrial conditions differin the yeast cultivation conditions, the thawing rate, and thefreezing rate. Additional differences between laboratory testsand industrial frozen dough bread-making conditions includethe addition of protective substances to the dough formulation,the methods used for mixing and kneading the dough, the temperatureused for dough preparation, and the osmotic pressure of thedough. Our results show that laboratory growth conditions aloneare not a prerequisite for improvement of freeze tolerance byaquaporin-overexpressing strains at the start of the fermentation(Fig. 2A), nor is the thawing process a significant factor indetermining cell survival. These results are consistent withthe observation that the main damage to cells caused by freeze-thawstress results from freezing rather than from thawing (30).

Our results for the effect of different freezing rates on yeastsurvival in the range of cooling rates that we studied are consistentwith the results of other workers (7, 8, 23, 24). The levelsof survival (expressed as RGC) generally remained high at coolingrates of 10 to 15°C per min but decreased dramatically atcooling rates of 30 to 35°C per min or higher (Table 1 andFig. 3). The cooling rate for samples frozen in liquid nitrogencould not be determined with the thermocouple used, althoughit was presumed to be very high (7, 8). Aquaporin overexpressionwas somewhat advantageous when small tubes were put directlyin liquid nitrogen, but it was even more advantageous when thecooling rate was slowed by use of a double plastic layer (Table1).

When the survival after different consecutive slow freezingtreatments was compared with the survival after the same numberof rapid freezing treatments, it became clear that the increasedfreeze tolerance of aquaporin-overexpressing strains of S. cerevisiaeseems to be restricted primarily to rapid freezing conditions(Table 2). Aquaporins mediate the rapid transport of water acrossthe plasma membrane (2, 26). Our hypothesis is that higher levelsof aquaporins in the plasma membrane allow rapid water efflux,especially at freezing temperatures, when water diffusion throughthe phospholipid layer of the membrane is limiting. Since thewater permeability of the plasma membrane increases greatlyat temperatures above the freezing point, the water flow probablyis not limiting and the presence of excess aquaporins is notadvantageous at low cooling rates.

This reasoning is consistent with the two-factor hypothesisof freezing injury. This hypothesis states that cells are mainlyliable to cellular dehydration and exposure to solution effectsat low freezing rates but that at high freezing rates mainlyintracellular ice crystal formation occurs (23, 24). The observationthat aquaporin-mediated improvement of freeze tolerance is restrictedto rapid freezing conditions strengthens the hypothesis thatimproved freeze tolerance in baker's yeast overexpressing aquaporinimplies that there is water export activity and results froma reduction in intracellular ice crystal formation (33). Theresults which we obtained with small doughs that were rapidly(Fig. 2A) or slowly (Fig. 2B) frozen and subsequently storedin the freezer also are consistent with this hypothesis. Dumontand coworkers relied on the same water permeability hypothesisthat we relied on to explain the effects of cooling rate onyeast viability observed for the range of cooling rates thatwe studied (8).

If aquaporins effectively improve freeze tolerance only at highrates of freezing, then this limitation might help explain whyaquaporin overexpression does not affect the leavening capacityof yeast cells in large, industrial frozen doughs (Fig. 1A andB). These industrial doughs usually are cooled at a rate of0.6 to 0.8°C per min in the core of the dough (Fig. 1C and3), which is much lower than the cooling rates used under laboratoryconditions (Fig. 3). It is unlikely that the industrial freezingprocess could be adjusted in such a way that all cells in thedough undergo rapid enough freezing for aquaporins to be beneficial.Moreover, rapid freezing generally results in lower survival.Hence, the results described in this paper indicate that aquaporinoverexpression is less useful for improvement of the maintenanceof dough rising capacity in frozen doughs than originally thought(33).

Use of aquaporin overexpression for improvement of freeze tolerancemight still have other potential applications (e.g., cryopreservation).The recent finding that artificial expression of an aquaporinimproves the survival of mouse oocytes (5, 9) and fish embryos(11) after cryopreservation is particularly interesting in thisrespect. Finally, the fact that aquaporin overexpression isadvantageous only for freeze tolerance in rapid freezing conditionsalso seems to limit the physiological importance of aquaporin-mediatedfreeze resistance of microorganisms under natural conditions.

ACKNOWLEDGMENTS

This work was supported by a fellowship from the Institute forScientific and Technological Research (IWT) and the KatholiekeUniversiteit Leuven (PDM/02/114) to A.T. and by grants fromthe Flemish Interuniversity Institute of Biotechnology (VIB/PRJ2),the Fund for Scientific Research-Flanders, and the ResearchFund of the Katholieke Universiteit Leuven (Concerted ResearchActions) to J.M.T.

We thank Renata Wicik and Kris Vandezande for technical assistance.

FOOTNOTES

* Corresponding author. Mailing address: Laboratorium voor Moleculaire Celbiologie and Vlaams Interuniversitair Instituut voor Biotechnologie, Institute of Botany and Microbiology, Katholieke Universiteit Leuven, Kasteelpark Arenberg 31, B-3001 Leuven-Heverlee, Flanders, Belgium. Phone: 32-16-32 15 07. Fax: 32-16-32 19 79. E-mail: Johan.Thevelein{at}bio.kuleuven.ac.be.

REFERENCES

Top