In the past two decades, Kluyveromyces lactis has developed into a second eukaryotic model and production organism as an alternative to the baker's yeast, Saccharomyces cerevisiae. One reason for the growing interest in K. lactis is its commercial use as an enzyme source for lactose degradation, which is employed in the dairy industry (Breunig et al., 2000). K. lactis is also frequently employed as a host for heterologous protein production (van Ooyen et al., 2006). Basic research on this simple eukaryote has greatly profited from the completion of the whole genome sequence (Sherman et al., 2004), and the fact that most classical and molecular genetic techniques established for S. cerevisiae can be applied (Rubio-Texeira, 2006). A congenic strain series of K. lactis derived from the sequenced type strain, with a variety of auxotrophic markers, plasmids and PCR deletion constructs, is now available (Heinisch et al., 2010).
Their different natural habitats (i.e. milk, with lactose as the major carbon source for K. lactis; and fruits, with glucose, fructose and saccharose for S. cerevisiae) explains their different physiologies. Thus, K. lactis is a Crabtree-negative yeast with a predominantly respiratory metabolism, whereas S. cerevisiae is Crabtree-positive and widely known for its fermentative capacity (Gonzalez-Siso et al., 2000). In fact, K. lactis has been suggested as a good model system for the oxidative stress response in neurons, since they also rely on an active respiration (Gonzalez-Siso et al., 2009; Ocampo and Barrientos, 2008). To our knowledge, no systematic analysis has been performed of the cell wall composition of K. lactis, until now.
As in all fungi, cell morphology and integrity in K. lactis is ensured by a rigid, yet highly adaptive, cell wall. In general, yeast cell walls contribute approximately 10–30% to the cellular dry weight and are composed of approximately 5% of proteins and 95% of polysaccharides (Klis et al., 2002). Of the polysaccharides, about 40% are mannan, covalently bound to cell wall proteins. The other 60% consist mainly of long 1,3-β-glucan chains connected to shorter 1,6-β-glucan chains. Chitin amounts to only 1–3% of the polysaccharides in yeast cells growing on rich medium, but its level can greatly vary between different yeast species and within one species, depending on the growth conditions (Aguilar-Uscanga and Francois, 2003). In most yeast species analysed so far, chitin preferentially accumulates at the bud neck and in the bud scars, but some chitin is also found in the lateral walls of the mother cell (Molano et al., 1980). Known covalently linked cell wall proteins can be grouped into two classes: The majority (approximately 30 different proteins in S. cerevisiae) are characterized by a GPI anchor, which first provides a connection to the plasma membrane but is processed further for covalent attachment of such proteins to cell wall polysaccharides (Kollar et al., 1997). The second class can be solubilized in mild alkali and include the Pir proteins, which are named after the presence of internally repeated sequences, through which they are attached to the 1,3-β-glucan network (five proteins in S. cerevisiae; Ecker et al., 2006).
In transmission electron microscopy (TEM), ascomycetous yeast cell walls generally appear as a two-layered structure. The inner one is less electron-dense, surrounds the plasma membrane and mainly constitutes the glucan network. The outer layer is in contact with the medium, is more electron-dense and contains most of the mannoproteins (Zlotnik et al., 1984). This is also true for K. lactis, where Klpmt1 and Klvga mutants have been shown to increase their cell wall thickness, presumably by altering the glycosylation pattern of the mannoproteins (Uccelletti et al., 1999, 2000). In S. cerevisiae, cell wall synthesis is governed by the so-called cell wall integrity (CWI) signalling pathway (Heinisch et al., 2010; Levin, 2005). A mutant carrying a defect in the endocytosis of one of its sensors, Wsc1, also has a thicker cell wall (Piao et al., 2007). Although two Wsc-type sensors exist in K. lactis (Rodicio et al., 2008) and many other components of the CWI pathway are conserved (Jacoby et al., 1999; Kirchrath et al., 2000; Rodicio et al., 2006), none of the mutants has yet been investigated for an altered cell wall composition.
A crude determination of mannoproteins and glucans in a variety of yeast species after growing them for 48 h in batch cultures, which included K. lactis, revealed no major differences to S. cerevisiae (Nguyen et al., 1998). On the other hand, studies on the so-called ‘secretome’ of K. lactis, albeit identifying some cell wall proteins, were not aimed to provide details on the cell wall composition (Brustolini et al., 2009; Madinger et al., 2009). We therefore investigated the cell wall of K. lactis in logarithmically growing and stationary phase cells, after growth on different carbon sources. The cell walls were visualized by TEM and key parameters, such as wall thickness, glucan, mannan and chitin content, glycoprotein composition and also sensitivity of growth to the cell-wall degrading enzyme preparation zymolyase, were determined under various growth conditions.
Materials and methods
Strains, media and culture conditions
In this study, the haploid type strain CBS2359 from K. lactis was compared to the S. cerevisiae strain CEN.PK111-61A (MATα his3-11,15 leu2-3,112 ura3-52 MAL3 SUC2 GAL; Entian and Kötter, 2007). CBS2359 is the type strain, whose genome has been sequenced and for which a congenic series is now available (Heinisch et al., 2010); thus, it is the most representative K. lactis strain available. The CEN.PK-series has been employed in functional analysis of the yeast genome (albeit not as widely as the BY-series) and was reported amongst all laboratory strains tested to be most closely related in its physiological properties to industrial strains.
Rich media were based on 1% yeast extract and 2% bacto peptone (Difco). As carbon sources, 2% glucose, 2% lactose or 3% ethanol were added. Cells grown in liquid media at 30 °C with shaking (180 rpm/min) were either harvested in the logarithmic growth phase at OD600 ≈ 2.0 (which is well within the logarithmic phase also for K. lactis; Heinisch et al., 1993) or in the stationary phase after 72 h of incubation.
30 OD600 units of exponentially growing cells were prefixed with 3× FIX [6% glutaraldehyde, 300 mM Pipes–HCl, 3 mM MgCl2, 3 mM CaCl2, 600 mM sorbitol, pH 6.8 (Brizzio et al., 1996; Wright, 2000)] for 5 min at room temperature. The cells were harvested by centrifugation and incubated with 1× FIX for another 30 min at room temperature. The cells were then washed three times with water and fixed with 3% w/v freshly prepared KMnO4 for 15 h at 8 °C. After four further washing steps with water, the cells were incubated in 1% w/v NaIO3 for 15 min at room temperature (Brizzio et al., 1996), washed once more, resuspended in 50 mM (NH4)3PO4 and incubated for 15 min at room temperature (Brizzio et al., 1996).
The fixed cells were dehydrated with increasing concentrations of ethanol (10–100% v/v) and transferred to a 1 : 1 mixture of ethanol and acetone for 10 min, followed twice by incubation in acetone for 5 min. The cells were then infiltrated overnight in a 3 : 1 mixture of acetone: Epon812. This mixture was replaced by Epon812 for 7–8 h and the cells were resuspended in fresh Epon812, which was polymerized for 72 h at 60 °C. Sections (70 nm thick) were cut with a diamond knife on a Leica Ultracut UCT ultramicrotome, transferred to single slot grids, stained with 2% uranyl acetate and lead citrate (Reynolds, 1963) and examined with a Zeiss TEM 902A electron microscope at 50 kV.
Preparation of yeast cell walls
Cell walls were isolated according to de Groot et al. (2004). Briefly, 100 ml logarithmically growing cells or 50 ml stationary culture were harvested, washed once with cold water and once more with 10 mM Tris–HCl, pH 7.5. The cells were resuspended in 2.5 ml 10 mM Tris–HCl, pH 7.5, with the addition of protease inhibitors (Protease Inhibitor Mix FY in DMSO, Serva) and disintegrated by vigorous shaking with glass beads (0.5 mm diameter) twice for 30 min at 8 °C. The debris containing the cell walls was washed extensively with 1 M NaCl and extracted four times by boiling for 10 min in a detergent solution (50 mM Tris–HCl, pH 7.8, 2% SDS, 100 mM Na–EDTA, 150 mM NaCl and 8 µl β-mercaptoethanol per ml extraction buffer). After seven washing steps with water, the cell walls were freeze-dried and stored at − 20 °C until use.
Analyses of cell wall polysaccharides
The glucose and mannose content of the K. lactis cell walls was analysed by HPLC. Freeze-dried walls (4 mg) were solubilized by treatment with 100 µl v/v 72% H2SO4 for 3 h at room temperature. After the addition of 575 µl distilled water to obtain a 2 M H2SO4 solution, the samples were boiled for 4 h and cooled on ice. Proteins were precipitated by adding 67.5 µl 35% perchloric acid and 37 µl 7 M KOH. After centrifugation, the supernatant was filtered and the sugar composition was analysed by HPLC on a REZEX organic acid analysis column (Phenomenex, Torrance, CA, USA) at 45 °C with 7.2 mM H2SO4 as the eluent, using an RI 1530 refractive index detector (Jasco, Tokyo, Japan). AZUR chromatography software was used for data integration. Solutions with different glucose and mannose concentrations were used as standards.
The chitin content of cell wall preparations was determined after acid and alkali treatment in the acid–alkali-insoluble fraction (Bahmed et al., 2002; Kapteyn et al., 2001). The cell walls were resuspended in 100 µl 1 M NaOH and boiled for 10 min. After cooling to ambient temperature, 100 µl 1 M HCl were added, followed by a washing step with water. The pellets were resuspended in 6 M HCl and boiled for 17 h. Dried samples were resuspended in 1 ml water. The chitin content was determined using a colorimetric method (Elson and Morgan, 1933). Briefly, 100 µl sample was mixed with 100 µl solution A (1.5 N Na2CO3 in 4% acetylacetone). The mixture was boiled for 20 min. After cooling to ambient temperature, 100 µl solution B (1.6g p-dimethyl aminobenzaldehyde in 30 ml concentrated HCl and 30 ml 96% ethanol) were added, incubated for 1 h at room temperature, and the absorbance at 520 nm was determined. The chitin content of the samples was determined relative to a glucosamine standard.
Zymolyase growth inhibition assay
The sensitivity assay to a β-1,3-glucanase preparation (ImmunO zymolyase-20T, MP Biomedicals, Illkirch, France) was performed according to Alonso-Monge et al. (2001). Cell cultures were inoculated with OD600 = 0.025 in 3 ml of YPD medium. Different amounts of zymolyase were added and the OD600 was determined after 16 h of cultivation at 30 °C on a shaker (180 rpm). Relative sensitivities were calculated as the percentage of cell densities compared to the samples grown without enzyme addition.
Analyses of covalently linked cell wall proteins by mass-spectrometry
Lyophilized cell walls were reduced with 10 mM dithiothreitol in 100 mM NH4HCO3 by incubation for 1 h at 55 °C. After cooling to room temperature and centrifugation, the reduced proteins in the pellet were alkylated with 65 mM iodoacetamide in 100 mM NH4HCO3 for 45 min at room temperature in the dark. The samples were quenched with 55 mM dithiothreitol in 100 mM NH4HCO3 for 5 min, washed six times with 50 mM NH4HCO3 and either frozen in liquid nitrogen for storage at − 80 °C or directly subjected to trypsin digestion. Trypsin Gold (2 µg; Promega, Madison, WI, USA) were added from a 1 µg/µl stock solution containing 0.1% Rapigest SF surfactant (Waters, Milhouse, MA, USA) and incubated for 18 h at 37 °C. The tryptic digests were desalted using a C18 tip column (Varian, Palo Alto, CA, USA) according to the manufacturer's instructions. After evaporation of acetonitrile in a Speedvac (Genevac, Ipswich, UK) the peptide concentration was determined at 205 nm using a NanoDrop ND-1000 (Isogen Life Science, Ijsselstein, The Netherlands) (Desjardins et al., 2009). Each sample was diluted with 0.1% trifluoroacetic acid to a final concentration of 75 ng/µl, and 10 µl per run were injected onto an Ultimate 2000 nano-HPLC system (LC Packings, Amsterdam, The Netherlands) equipped with a PepMap100 C18 reversed-phase column (75 µm i.d. × 25 cm length; Dionex, Sunnyvale, CA, USA). An elution flow rate of 0.3 µl/min was applied along a linear gradient with increasing acetonitrile concentrations over 45 min. The eluting peptides were directly ionized by electrospray in a Q-TOF (Micromass, Whyttenshawe, UK). Survey scans were acquired from m/z 350–1200. For low-energy collision-induced dissociation (MS/MS), the most intense ions were selected in a data-dependent mode. After processing with the MaxEnt3 algorithm, included in the Masslynx Proteinlynx software, the spectra were converted into pkl (peak list) files. Submitting these to an internally licensed version of MASCOT (Matrix Science, UK) led to the identification of proteins by comparison to a complete ORF translation of the K. lactis genome (http://www.genolevures.org/). In MASCOT two miscleavages and a tolerance of 0.6 Da for peptides and MS/MS were allowed. Based on probabilistic MASCOT scoring, p < 0.05 was considered significant for peptide identification. At least two independently obtained biological samples were analysed for each condition (biological replicates). Each biological sample was subjected to two runs with MS/MS selection switching times of 1.5 and 1.25 s, respectively. All identified proteins were subjected to signal peptide prediction using SignalP3.0 (Bendtsen et al., 2004) and prediction of a GPI anchor sequence using the BIG-PI fungal predictor (Eisenhaber et al., 2004). For a semi-quantitative analysis of the data, the mean of the total number of peptide identifications per biological replicate was calculated for each growth condition.
Transmission electron microscopy (TEM) of K. lactis and S. cerevisiae grown on glucose or ethanol as carbon sources
Although numerous studies have been performed on S. cerevisiae regarding the ultrastructure of its cell wall (Osumi, 1998), similar data on K. lactis are scarce. In fact, TEM has been performed in K. lactis only for some mutants with an altered cell wall, but since these works had a different focus, cell wall thickness was not systematically determined (Uccelletti et al., 1999, 2000, 2005). Here, we first established a preparation method for S. cerevisiae cells designed to yield maximal information on the cell wall and applied the same procedure for K. lactis cells (see Materials and methods for details). As apparent from the TEM images in Figure 1, cell wall thickness varies significantly at different points of the yeast cells, especially within growing buds. In order to obtain reliable comparative values, we therefore determined cell wall thickness only in mother cells and measured 10 different data points for each cell (Figure 1F). For cells grown with glucose as carbon source, the cell wall of K. lactis is approximately 40% thinner than the wall from S. cerevisiae (Table 1). When grown on 3% ethanol, K. lactis increases its cell wall thickness by approximately 65%, whereas no significant difference is observed for ethanol-grown cells of S. cerevisiae.
Table 1. Cell wall thickness after growth on different carbon sources
Cell wall thickness (nm)
Values given are the average of 45 cells measured for each condition ± SD. For each cell, wall thickness was determined at 10 different points of the mother cell, as exemplified in Figure 1E, F.
64 ± 10
105 ± 18
102 ± 14
100 ± 15
Regarding the overall envelope structure, no significant differences can be observed between K. lactis and S. cerevisiae. Thus, a ‘fuzzy’ electron-dense outer layer, presumed to contain most mannoproteins (Osumi, 1998), is succeeded in both yeasts by a less electron-dense inner layer (most likely constituted by the glucan network; Zlotnik et al., 1984; Figure 1). The glucan network appears to directly surround the plasma membrane. As expected from a respiratory yeast, more mitochondrial structures are observed within glucose-grown cells of K. lactis as compared to glucose-grown cells of S. cerevisiae. A more similar picture regarding these structures is obtained if both yeasts are grown under respiratory conditions (i.e. on ethanol; Figure 1B, D).
Polysaccharide composition of K. lactis cell walls after growth on different carbon sources
Yeast cell wall polysaccharides are mainly composed of glucans, mannan from glycosylated proteins and only minor amounts of chitin (Klis et al., 2002). We here determined the relative contents of these constituents after growing K. lactis cells to logarithmic phase with either 2% glucose or 3% ethanol as a carbon source. In addition, cells grown to stationary phase were examined. As a reference, S. cerevisiae grown under the same conditions were also employed. As evident from Table 2, the glucan content of K. lactis cell walls did not significantly vary between the different growth conditions. In contrast, S. cerevisiae displayed a slight reduction in the amount of glucan in preparations from cells grown on ethanol, as compared to those grown on glucose. This reduction was even more pronounced when stationary phase cells of S. cerevisiae were compared to logarithmically growing cells on glucose.
Table 2. Polysaccharide content of cell walls obtained under different growth conditions
Carbohydrate contents are given in µg/mg of cell wall dry weight ± SD from at least three independent determinations, with the sole exception of K. lactis cells growing logarithmically on glucose, for which only two determinations were done. Log, logarithmic growth phase; Stat, stationary phase.
556 ± 11
326 ± 24
34 ± 3
557 ± 21
272 ± 13
43 ± 8
569 ± 40
263 ± 3
34 ± 4
463 ± 34
317 ± 20
29 ± 4
427 ± 61
340 ± 27
47 ± 3
373 ± 69
368 ± 53
24 ± 1
Conversely, the mannan content increased from logarithmically growing cells on glucose to those growing on ethanol, with the highest value obtained in stationary S. cerevisiae cells. In contrast, in K. lactis a decrease in cell wall mannan content can be observed in the respective order of growth conditions. Regarding chitin, similar relative amounts can be detected in K. lactis and in S. cerevisiae, constituting < 1% of the total cell wall polysaccharides (Table 2). Cells of both yeast species grown on glucose, either into logarithmic or stationary phase, contain approximately 30 µg chitin/mg cell wall dry weight. The chitin content increases by approximately 30% in both species when the cells are grown on ethanol.
Apart from the overall polysaccharide content, sensitivity to the 1,3-β-glucanase-containing enzyme preparation zymolyase is frequently employed to detect in vivo alterations in the cell wall structures of growing yeast cells (de Nobel et al., 2004). We therefore incubated cultures from K. lactis and S. cerevisiae in the presence of increasing amounts of zymolyase and determined their ability to grow overnight in rich medium with either glucose or ethanol as carbon sources (Figure 2). S. cerevisiae displays a slightly increased resistance to zymolyase treatment when grown on ethanol as a carbon source, as compared to cultures grown on glucose. In contrast, K. lactis is more sensitive to zymolyase when grown on ethanol as compared to glucose.
Cell wall proteins in K. lactis
Since mannoproteins dominate the outer layer of the cell wall, and the wall protein composition in various yeast species has been shown to depend on environmental conditions, the overall composition of the cell wall proteome was determined for K. lactis under different growth conditions. For this purpose, cells were harvested during logarithmic growth on different carbon sources, as well as during stationary phase for glucose-grown cells, and used for preparing cell walls. Peptides from the proteins contained in these preparations were analysed by mass spectrometry, as described in Materials and methods. Table 3 shows a representative selection of proteins detected and relates them to the S. cerevisiae proteome (for a complete listing of the mass spectrometry data, see Supporting information, Tables S1–S4).
Table 3. Selected K. lactis cell wall proteins identified by mass spectrometry (for complete listings, see Supporting information, Tables S1–S4)
The closest S. cerevisiae homologue is listed; amino acid identity of the deduced protein sequence given in %; ws, weak similarity (mentioned in NCBI database, but not found by blasting against the same database).
The lengths of the K. lactis proteins are given, with the regions covered by the identified peptides given in parentheses.
CWP, cell wall protein; CWI, involved in cell wall integrity and stability.
Taken from the SGD and NCBI databases.
GPI cell wall proteins
Two conserved domains: DUF2403 (peptide 1), DUF2401 (peptide 2), which are homologous to ScTos1
Pir repeat contains the diagnostic sequence DGQJQ at position 207–211; Q209 is the predicted attachment site for β-1,3-glucan
All identified peptides in the N-terminal region belong to a superfamily domain involved in hyphal regulation; homologous to CaHyr1 in the N-terminal region
Fatty acid release
Peptides of the cPLA2_fungal_PLB domain
Transglucosylase using β-1,3-glucan as substrate
Peptides of the glyco_hydro_72 superfamily domain
Transglucosylase using β-1,3-glucan as substrate
Peptides of the glyco_hydro_72 superfamily domain
CWI and mannan assembly
Peptides of the receptor_L domain
Peptides of the Flo_11_superfamily domain
Peptides of the pepsin_like domain; closest homologue from A. gossypii (37% amino acid identity)
Transglucosylase using β-1,3-glucan as substrate
Peptides of the glyco_hydro_72 superfamily domain
Contains two Flo11_superfamily domains in the N-terminal region
GPI protein turnover
SAP_like domain includes the identified peptide
Peptides are located in the N-terminal CFEM domain
Peptides in the N-terminal region with a PA14 superfamily domain
Closest homologue in Z. rouxii with 34% amino acid identity
No Pir repeat, in contrast to ScCwp1
Linkage of chitin to 1,6-β-glucan
Glyco_hydrolase_16 superfamily domain identified by some peptides
Linkage of chitin to 1,6-β-glucan
Glyco_hydrolase_16 superfamily domain identified by most peptides
Non-GPI cell wall proteins
Contains five Pir repeats, with all identified peptides located downstream of the repeat region
Contains eight Pir repeats, with all identified peptides located downstream of the repeat region
Peptides of the glyco_hydro_17 superfamily domain
Peptide within the pepsin_retropepsin_like_superfamily domain
Peptides within the GLpA-domain
Peptides within the glyco_32 domain
Metalloreductase (plasma membrane)
Peptide in the NAD binding pocket within the NOX_Duox_like_FAD_NADP domain
K. lactis cell wall proteins detected include homologues of abundant GPI-anchored proteins of S. cerevisiae, such as Muc1/Flo11 and Ecm33, as well as enzymes with a presumed transglucosidase function, such as Gas1, Gas3 and Gas5. Whereas most of the proteins listed in Table 3 can be detected under all growth conditions tested, Flo5 and Utr2 are found exclusively in logarithmically growing cells on either glucose or lactose, with Utr2 also found on ethanol. Interestingly, a putative GPI cell wall protein of unknown function and with no apparent homologue in S. cerevisiae (KLLA0E24893g) is exclusively found in the wall of ethanol-grown cells of K. lactis. Three examples of proteins most likely not involved in cell wall composition and/or remodelling, which were also detected in the cell wall preparations, are given in the lower part of Table 3. One is a predicted ferric reductase located in the plasma membrane (Fre1) and another is a secreted invertase (KlInv1; Georis et al., 1999), mostly retained in the periplasmic region. This probably explains their detection in cell wall preparations. The reason for the third protein, the mitochondrial Gut2, to appear in such preparations is not known.
More importantly, we wanted to know which cell wall proteins are preferred during a certain growth condition. For this, we used a semi-quantitative approach based on the observation that the total number of peptide identifications of a particular protein is correlated with the concentration of that protein (Liu et al., 2004). As an exclusion limit, Table 4 summarizes only proteins, whose estimated abundance varies by at least a factor of two regarding the relative number of peptide hits detected in the pairwise comparison of different conditions. Cwp1 and Muc1/Flo11 are especially interesting in this context. For either protein, there is only one encoding gene present in the S. cerevisiae genome, whereas the K. lactis homologue has apparently duplicated, giving rise to two isoforms. Whereas both isoforms of Cwp1 in K. lactis are predominantly found in the stationary phase, KlMuc1a and KlMuc1b are oppositely regulated, in that KlMuc1a predominates in logarithmic growth and KlMuc1b is highly represented in the stationary phase (Table 4A). The content of other cell wall proteins varies in a carbon source-dependent manner, with many proteins more readily found in the walls of glucose-grown cells (Table 4B). This includes two proteins with no apparent homologue in S. cerevisiae (KLLA0B14498g and KLLA0E24959g). In contrast, KlCcw14 seems to be ethanol-specific. Another difference in the regulation of the KlMuc1 isoforms becomes apparent in the comparison between glucose-grown cells to those grown on lactose as a carbon source (Table 4C). Whereas KlMuc1a is preferentially found on glucose, KlMuc1b seems to be lactose-induced. Two other proteins, KlFlo5 and KlCcw14, are also preferentially found in the presence of lactose, i.e. the ‘natural’ carbon source of K. lactis.
Table 4. Differential detection of K. lactis cell wall proteins
(A) Stationary vs. logarithmic growth phase for cells grown on YEPD
K. lactis ORF
(B) Growth on glucose vs. growth on ethanol
K. lactis ORF
(C) Growth on glucose versus growth on lactose
K. lactis ORF
Previously, the outer surface of different yeast species was shown to have a very similar three-layered ultrastructure, i.e. the electron-dense plasma membrane is covered by a less dense network of crosslinked β-1,3- and β-1,6-glucans, which in turn is surrounded by an electron-denser layer of mannoproteins. This was first observed for S. cerevisiae (Zlotnik et al., 1984) but is also true for the yeast forms of the opportunistic pathogen Candida albicans (Kapteyn et al., 2000) and other ascomycetous yeasts (Klis et al., 2010). As observed by us and by others (Uccelletti et al., 2000, 2005), a similar composition is found in K. lactis. However, they did not systematically determine the actual cell wall thickness of K. lactis, due to a different focus of interest. Note that although we tested only one specific strain for each species of K. lactis and S. cerevisiae (for reasons explained in Materials and methods), we have no reason to believe that the data obtained here would not be respresentative. In fact, our TEM data agree well with what can be deduced from the literature on other strains of these two species (Uccelletti et al., 2000; Klis et al., 2002). We found that K. lactis cell walls are considerably thinner than those of S. cerevisiae when the cells are grown on glucose as a fermentable carbon source. For S. cerevisiae, a cell wall thickness of approximately 100 nm correlates well with data from other TEM studies (Klis et al., 2002), as well as with the finding that a modified Wsc1 sensor, which reaches out only 86 nm from the plasma membrane, cannot be detected at the cell surface by atomic force microscopy in vivo (Dupres et al., 2009). It is feasible that the smaller cells of K. lactis would accordingly adjust their cell wall thickness as compared to the larger S. cerevisiae cells. On first sight, this coincides with the increased sensitivity of the thinner K. lactis cell wall to zymolyase treatment, as compared to S. cerevisiae. However, we also observed that the K. lactis cell wall is considerably fortified when cells where grown on ethanol as a carbon source, with a similar thickness to that of S. cerevisiae, but a significantly increased zymolyase sensitivity. Moreover, the thickness of the S. cerevisiae cell wall does not increase on 3% ethanol as compared to glucose-grown cells, yet they are slightly more resistant to the treatment. In conclusion, it is not the overall content of 1,3-β-glucan targets that determines the susceptibility of a yeast to the enzyme. It has been suggested that the degree of crosslinking with other cell wall polysaccharides, such as 1,6-β-glucan and chitin, determines the sensitivity to zymolyase (Aguilar-Uscanga and Francois, 2003). Surprisingly, reference data on cell wall thickness of S. cerevisiae (or any other yeast) grown on ethanol are not available. This may be attributed to technical difficulties in preparing the cells for TEM, which needed considerable optimization in this work. Similar technical problems have been reported to prevent investigations of stationary phase cells by TEM in S. cerevisiae, which were therefore not attempted here (Wright, 2000).
A moderate difference between K. lactis and S. cerevisiae was observed in the relative contribution of glucan and mannan to the overall polysaccharide content under different growth conditions. Whereas S. cerevisiae shows a tendency to decrease the amount of glucan under stress conditions (i.e. growth on ethanol or into stationary phase), K. lactis keeps its glucan content constant. Nevertheless, the changes in growth sensitivity to zymolyase between glucose- and ethanol-grown cells of K. lactis show that cell wall properties have changed considerably. Apparently, glucan and mannan content are crude parameters for monitoring such changes.
The cell wall proteome of K. lactis shares several characteristics with the wall proteomes of C. albicans, C. glabrata and S. cerevisiae:
1.The number of covalently linked proteins identified is about 20 in these four ascomycetous yeasts (Klis et al., 2010).
2.The composition of the wall proteome depends on the growth conditions (Table 4). KlMuc1a and KlMuc1b represent a fascinating example because the level of KlMuc1a is higher in walls from logarithmically growing cells, whereas KlMuc1b is strongly increased in stationary phase cell walls.
3.GPI proteins form the largest class of wall proteins and Pir proteins form a minor group. Interestingly, the GPI protein KlCwp1a, but not KlCwp1b, is a hybrid GPI protein in the sense that it also possesses a Pir repeat just like its counterpart ScCwp1. This suggests that KlCwp1a can bind to both β-1,6-glucan through its GPI structure at the C-terminus and to β-1,3-glucan through its Pir repeat (Ecker et al., 2006; Kapteyn et al., 2001).
4.The tryptic peptides of GPI proteins are generally found in the N-terminal region, where the predicted functional domain is found, whereas in Pir proteins the tryptic peptides are from the C-terminal half of the protein.
With respect to the differential detection of specific cell wall proteins under specific growth conditions, it is intriguing that Muc1 has only one member in S. cerevisiae but two in K. lactis. This draws attention, because as a consequence of the whole genome duplication S. cerevisiae usually displays a higher degree of redundancy in enzymatic functions than K. lactis (Gbelska et al., 2006). Apparently, ScMUC1 gene expression is subject to a variety of regulatory mechanisms. One may speculate that gene duplication in K. lactis would allow a more stringent control of each of the new genes without the need for a complex integration of different signals at the level of a single gene transcription. Judging from the data presented in Table 4, both KlMUC1b and KlCCW14 can be expected to be under the control of glucose repression, since the encoded proteins are preferentially detected in the absence of this carbon source. This observation also illustrates again the dynamic nature of the cell wall proteome. Finally, whereas several cell wall proteins are conserved in the four ascomycetous yeast species discussed above, such as the Crh proteins, Ecm33, the Gas proteins, phospholipase, Pir proteins, Scw4 and yapsins, we also detected a number of putative cell wall proteins in K. lactis, such as KLLA0B14498g, KLLA0E24893g and KLLA0E24959g, which lack a counterpart in the yeast databases. These provide potential candidates for an adaptation of K. lactis to its natural environment, as well as prospective targets for a specific containment in mixed yeast cultures.
It should be noted that one of the drawbacks of the techniques employed here is the information gap that exists between TEM studies and glycoproteome analyses. Whereas the first provide structural information but lack biochemical details, the second generate a wealth of data on protein composition but lack positional information as to where these proteins are located within the cell wall. This problem can be overcome for specific wall proteins by either employing specific probes in electron microscopy (Hardham and Mitchell, 1998) or by detecting single protein molecules by specified atomic force microscopy. The latter provides a nanoscale resolution, but is only applicable to proteins reaching the cell surfaces (Dupres et al., 2009). Clearly, such techniques are extremely interesting for future applications, but would be well beyond the scope of this study.
In summary, although we have tested only a limited number of growth conditions, our results convincingly show that cell wall properties and composition, and especially protein composition, can vary considerably. This indicates that cell wall formation in K. lactis is tightly controlled and underlines the importance of as-yet only partially investigated signalling pathways (see Rodicio et al., 2008, and references therein), which may regulate this process.
We are grateful to Bernadette Sander-Turgut for technical assistance and Rosaura Rodicio for critical reading of the manuscript. F.M.K. acknowledges financial support by the EU Programme, Grant No. FP7-214004-2 FINSysB. The work in Osnabrück was funded by the Deutsche Forschungsgemeinschaft (Grant No. SFB 431, to J.J.H).
Supporting information on the internet
The following supporting information may be found in the online version of this article: