Electrophoretic Isoenzyme Variation in Kluyveromyces Populations and Revision of Kluyverornyces Marxianus (Hansen) Van Der Walt

Home , Kluyveromyces marxianus

INTERNATIONALJOURNAL OF SYSTEMATICBACTERIOLOGY, Jan. 1986, p. 94102 Vol. 36, No. 1 0020-7713/86/010094-09$02.0010 Copyright 0 1986, International Union of Microbiological Societies

Electrophoretic Isoenzyme Variation in Kluyveromyces Populations and Revision of Kluyverornyces marxianus (Hansen) van der Walt

DEBORAH GAYLE SIDENBERG AND MARC-ANDRE LACHANCE" Department of Plant Sciences, University of Western Ontario, London, Ontario, N6A 5B7, Canada

Natural isolates of Kluyveromyces from a variety of habitats were compared on the basis of seven isoenzyme patterns. Some strains of Candida considered to be Kluyveromyces anamorphs, as well as one strain each of Candida sake, Saccharomyces cerevisiae, and Pichia JEuxuum, were also examined. The isoenzyme patterns readily substantiated the relationships between the anamorphs and their putative teleomorphs and the nonmember status of strains not belonging to the genus. A multivariate analysis of the electrophoretic patterns indicated that isolates belonging to the Kluy veromyces lactis and Kluy veromyces marxianus deoxyribonucleic acid reassociation groups are not phenotypically continuous with one another or with the former species Kluyveromyces dobzhanskii, and thus we propose to reinstate these taxa as separate species. Taxa previously described as Kluyveromyces thermotokrans and Kluyveromyces waltii also appeared to constitute reproductively isolated natural populations. When the results of isoenzyme electrophoresis were compared with deoxyribonucleic acid reassociation data and with mating compatibility patterns determined by other workers, we found that deoxyribonucleic acid relatedness gives a more accurate view of relationships among Kluyveromyces species.

In 1984, the genus Kluyveromyces van der Walt emend. MATERIALS AND METHODS van der Walt underwent a major revision (15). Based on Microorganisms. The type strains of Kluyveromyces strain hybridization studies (5), Kluyveromyces bulgaricus, species (Table 1) were obtained from H. J. Phaff, Department Kluyveromyces cicerisporus, Kluyveromyces dobzhanskii, of Food Science and Technology, University of California, Kluyveromyces drosophilarum, Kluyveromyces fragilis, Davis. Strains isolated from cheese were kindly provided by Kluyveromyces lactis, Kluyveromyces phaseolosporus, Kluyveromyces vanudenii, and Kluyveromyces wikenii were J. L. Schmidt, Institut National Agronomique, Paris-Gignon, France. S. cerevisiae strain 79-11 was subcultured from relegated to seven varieties of Kluyveromyces marxianus. Furthermore, K. cicerisporus was united with Kluy- commercial bakers' yeast. All other strains were obtained veromyces marxianus var. bulgaricus, and K. phase- through various ecological studies conducted in our olosporus was united with Kluyveromyces marxianus var. laboratory. Table 2 shows the sources and locations of drosophilarum. Controversy surrounds this classification. isolation of the strains and their accession numbers in the Some yeast taxonomists (9, 12) feel that deoxyribonucleic yeast culture collection of the Department of Plant Sciences, acid (DNA) reassociation, in addition to mating compatibil- University of Western Ontario, London, Ontario, Canada. ity, must be taken into account in delimiting the largely All strains were maintained on yeast extract-malt extract- self-fertile species in the genus Kluyveromyces. In a prelim- glucose-peptone agar at 4°C in screw-cap vials. For clarity, inary study of Kluyveromyces type strains (12), isoenzyme we used the nomenclature of van der Walt (14). patterns suggested that yeasts considered to be varieties of Polyacrylamide gel electrophoresis. Discontinuous K. marxianus (15) constitute discrete taxa which would be nondenaturing polyacrylamide gels were prepared as recom- better classified as separate species. mended by Davis (2). The conditions used were those In this paper, we describe the results of an electrophoretic described previously (12). Electromorphs were scored on a characterization of seven enzymes to assess the degrees of presence or absence basis, and no assumptions concerning variability within and among Kluyveromyces populations the individual genetic bases were made. For each enzyme and to determine the taxonomic relationships of these pop- activity, the bands were numbered consecutively as a func- ulations based on reproductive isolation as it appears to have tion of increasing electrophoretic mobility. All samples were occurred in nature. Anamorphs from the genus Candida and studied at least twice for each enzyme. one strain each of Candida sake, Pichia fluxuum, and Histochemical staining. The procedures used for histo- Saccharomyces cerevisiae were also examined. The inclu- chemical staining of electrophoretic gels have been de- sion of a large number of natural isolates from diverse scribed elsewhere (12). P-Glucosidase (PGL) was an ex- habitats (black knots, tree exudates, Drosophila, and ception; the staining system proposed by Sidenberg and cheese) seems to have resolved the disagreement as to which Lachance (11) for exo-P-glucanase (EPG) was adapted for criterion, DNA reassociation c)r mating compatibility pat- this enzyme, which previously (12) was visualized by an- terns, should prevail in delineating Kluyveromyces species. other technique (3). We redefine the varieties of K. marxianus proposed by van Data analysis. Reciprocal averaging, a form of correspon- der Walt and Johannsen (15). dence analysis (4), was used to ordinate the strains as a function of correlated electromorphs. For this analysis, the results were expressed in a matrix having dimensions of n x p, where n strains were described by the presence (coded as 1) or absence (coded as 0) of each of p electromorphs. * Corresponding author. Reciprocal averaging ordinated rows and columns of a

94 VOL. 36, 1986 VARIATION IN KLUYVEROMYCES POPULATIONS 95

TABLE 1. Nomenclature of Kluyveromvces species used in this study

Taxonomic designation of van der Walt" Taxonomic designation of van der Walt and Johannsen" UCD (FS&T) no. K. cicerisporus van der Walt, Nel, et K. marxianus (Hansen) van der Walt var. bulguricus (Santa Maria) 71- 14T" Kerken 1966 Johannsen et van der Walt in Johannsen (1980) K. dobzhanskii (Shehata, Mrak et PhaE) van K. marxianus (Hansen) van der Walt var. dobzhunskii (Shehata, 50-45= der Walt 1955 Mrak et Pham Johannsen et van der Walt in Johannsen (1980) K. drosophilarurn (Shehata, Mrak et PhaE) K. marxianus (Hansen) van der Walt var. drosuphilarum (Shehata, 51-130* van der Walt 1955 Mrak et Pham Johannsen et van der Walt in Johannsen (1980) K. lads (Dombrowskii) van der Walt 1910 K. marxianus (Hansen) van der Walt var. lactis (Dombrowski) 71-59T Johannsen et van der Walt in Johannsen (1980) K.fragilis (Jorgensen) van der Walt 1909 K. marxianus (Hansen) van der Walt var. murxiunus van der Walt 71-58T (1970) K. marxianus (Hansen) van der Walt 1888 K. marxianus (Hansen) van der Walt var. murxiunus van der Walt 55-82T (1970) K. vanudenii (van der Walt et Nel) van der K. marxiunus (Hansen) van der Walt var. vunudenii (van der Walt 70-4= Walt 1963 et Nel) Johannsen et van der Walt in Johannsen (1980) K. thermotolerans (Philippov) Yarrow 1932 K. thermotolerans (Philippov) Yarrow 1932 55-41T K. waltii Kodama 1974 K. waltii Kodama 1974 72-13T 'See reference 13. 'See reference 14. ' UCD (FS&T), Culture Collection of the Department of Food Science and Technology, University of California, Davis. T = type strain. frequency matrix and simultaneously revealed correspon- relationships among groups of strains based on electromorph dences between two kinds of information (i.e., strains and frequencies pooled by taxon (the species recognized by van electrornorphs). der Walt in 1970 [14]).This technique reveals the correlation A principal component analysis was used to assess the structure among continuous variables (in this case the

TABLE 2. List of strains and their sources of isolation Taxonomic Source" Culture collection no." designation C. sake 7 79-228 K. cicerisporus 11 80SM1-4, 80SM2-4, 80SM3-10, 80SM10-2, 80SM 10-5, 80SM12-2, 80SM 12-12, 80SM17-1, 80SM21-7, 80SM53-2 K. dobzhanskii 2 79-37, 79-133, 79-199, 79-265, 79-267. 80-28, 80-29, 80-37, 80-87, 80-88 3 82-32, 82-232, 82-244 7 79-183, 79-187, 79-188, 79-189 K. drosophilarurn 2 79-261. 80-45 3 82-233, 82-237, 82-241, 82-245 4 80-89, 82-12, 82-17, 82-45 7 79-169 9 80-48 K.fragifis 11 80SM 16-1, 80s M 16-10, 80SM27-8, 80s M46-5 K. lactis 11 80SM1-11, 80SM2-5', 80SM3-8, 80SM6-2, 80SM3-13, 80SM5-6, 80SM5-8, 80SM6-10. 80SM6-7, 80SM13-5, 80SM16-9, 80SM30-7, 80SM32-5', 80SM35-14', 80SM32-6, 80SM48-7 K. marxianus 11 80 S M 3-4' K. thermotolerans 1 79-255 2 79-110, 70-112, 79-114, 79-116, 79-117, 79-118, 79-119. 79-126, 79-191, 79-192, 79-193, 79-194, 79-195, 79-196, 79-249, 79-250. 79-251, 79-252, 79-253. 79-254, 80-19, 80-84 3 80-070, 82-204, 82-206, 82-209. 82-215, 82-216, 82-218, 82-220, 82-223, 82-225, 82-231, 82-292 7 79-164 8 79-162, 81-125, 81-126 10 79-139 K. vanudenii 3 82-235, 82-236, 82-239, 82-210 5 79-127, 79-168 6 80- 12 9 80-49 K. waltii 3 82-227, 82-228 7 79-163 8 79-160, 79-161, 81-127, 81-128 P. fluxuum 4 80-109

I' Isolation sources: 1, black knot, Prrrnus pumilri. Pinery Provincial Park, Ontario, Canada: 2, black knot, Priinirs i,ir,qinkimi, Pinery Provincial Park, Ontario, Canada; 3, Drosophiki, Pinery Provincial Park, Ontario. Canada; 4, exudate, Qrierciis rribrri, Pinery Provincial Park, Ontario. Canada; 5. black knot, Priinirs virgininna, Coldstream Conservation Area. Ontario, Canada; 6, black knot. Prrrnirs i,irginimri, Medway Creek. London, Ontario. Canada; 7. black knot, Pritnus virginiana, Melbourne, Ontario, Canada; 8, ball, Qrrerciis rirbrci, Melbourne. Ontario, Canada: 9. black knot. Prrrnris i*irginilinri,Dingman Creek, London, Ontario, Canada; 10, black knot, Pritnus serotinri, St. Anicet. Quebec, Canada; 11. Camembert cheese, J. L. Schmidt, Institut National Agronomique, Paris- Grignon, France. Culture collection of the Department of Plant Sciences, University of Western Ontario. London, Ontario, Canada. ' Anamorphs. 96 SIDENBERG AND LACHANCE INT. J. SYST. BACTERIOL.

TABLE 3. Observed and expected entropy values for the RESULTS enzymes examined in this study Electrophoretic variation. Studies performed with type Enzyme Observed I Expected I strains (12) indicated that polymorphism in Kluyveromyces EST 1,129" 1,186 species was best revealed by the following enzymes: alcohol ADH 618 509 dehydrogenase (ADH; EC 1.1.1.1) , malate dehydrogenase SOD 577 540 (MDH; EC 1.1.1.37), superoxide dismutase (SOD; EC MDH 545 525 1.15.1.1), esterase (EST;EC 3.1.1.1), EPG (EC 3.2.1.58), aGL 45 1 414 a-glucosidase (aGL; EC 3.2.1.20), and PGL (EC 3.2.1.21). EPG 231 197 92 75 These enzymes varied in the amount of information (r) which BGL they contained (Table 3), and thus different enzymes con- ' The values are given in order of decreasing heterogeneity. tributed differently to the elucidation of group structure within the populations. As anticipated (12), EST (I = 1,129) was the most heterogeneous enzyme. In addition, it was the only enzyme for which the observed entropy value was less pooled relative frequencies of electromorphs) and dimin- than the expected value, which indicates that the EST ishes their dimensionality. Besides eliminating insignificant banding patterns were very disorderly to start with. Stabi- variables, new less obvious but more meaningful variables lizing patterns only began to emerge as sampling size in- may be identified (1). creased. By contrast, the heterogeneity of ADH was sub- The traditionally accepted (8) measures of diploid variabil- stantially higher than anticipated. As more strains were ity would have been difficult to apply in this study. introduced into the study, new ADH patterns appeared. Kluyverornyces strains are usually mixtures of haploid and PGL (I = 92) was the least heterogeneous enzyme, due in diploid cells, and assumptions about the allelic nature of the part to the fact that several strains expressed no activity for electromorphs would have been unjustified. Genetic diver- this enzyme. All of the other enzymes revealed intermediate sity was assessed by the entropy or information measure I, amounts of information. which serves as a measure of both richness (the number of Group structure among individual strains. Simple visual distinct kinds of bands encountered in the populatiofi of a inspection of the presence or absence data (not shown) particular size) and evenness (distribution of band frequen- indicated that important discontinuities existed within the cies). The formula proposed by Sneath and Sokal (13) to electrophoretic patterns of the yeasts. Correspondence anal- calculate I is given elsewhere (12). The expected entropy ysis clearly identified the group structure, as shown by values for the enzymes of the populations were calculated by ordinations of the yeast strains (Fig. 1) and of the using the same formula and artificial populations made up electromorphs (Fig. 2). Figures 1 and 2 are superimposable, entirely of type strains. but they are presented separately for clarity.

0 y2

4 - 0

0 0 W. 0 F I m :0 1 0

0 V 0 I2 231 a2 2040 , 0 79 1390 0 0

FIG. 1. Ordination of strains of Kluyveromyces by reciprocal averaging. Type strains are indicated by squares and labeled as follows: C, K. cicerisporus; DO, K. dobzhanskii; DR, K. drosophilarltm; F, K. fragilis; L, K. lactis; M, K. marxianus; T, K. thermotolerans; V, K. vanudenii; W, K. waltii. Anamorphic strains (Candida) are indicated by triangles. Culture collection numbers are given for atypical strains of K. thermotolerans. The bars show two standard deviations about the mean for each taxon as defined by van der Walt (14). VOL. 36, 1986 VARIATION IN KLUYVEROMYCES POPULATIONS 97

SOD 1

SODS. T y2

EST 140 ADH 2. EST160

EST8 0 SOD2 oooMDH1 0 MDH4. .EST17 0 MDH3 0, ADH 9 -EST25j)jf - .SOB9 - b OMDH 7 s oADH5 MDH5. Y1 0 aGL6 0 EST24 aGL3. E/3G 3 MDH2. aGLl0 0 EST23 aGL5 I SOD20 OaGL7 ADH3.

SOD80

FIG. 2. Ordination of electromorphs of Kluyverornyces by reciprocal averaging. Unlabeled points locate electromorphs with near-zero coordinates.

The ordination of strains on the first two axes resolved listed. None of the three atypical strains exhibited MDH four major groups. The first axis (Yl),%hich had a canonical electromorph 2 (MDH2), SOD2, EST8, EST14, EST16, correlation value of 0.79, not only separated K. dobzhanskii aGL7, and EPG3 activities, each of which was detected in at and the K. lactis and K. marxianus DNA subgroups from least one-third of the other strains of K. thermotolerans. One Kluyveromyces waltii and Kluyveromyces thermotolerans, would normally expect at least one of the three atypical but also provided a small amount of discrimination between strains to possess these bands. In addition, most atypical the two DNA subgroups. The second axis (Y2) had a strains exhibited SOD3, SOD6, EST18, and EPG2 activities, canonical correlation value of 0.68, and it resolved the K. which were not found in the typical strains. One of these marxianus DNA subgroup and K. waltii from K. bands (EST18) was unique, and the other three were abun- thermotolerans, K. dobzhanskii, and the K. lactis DNA dant in other taxa. Strain 82-231 was the most unusual, and subgroup. The bars in Fig. 1 show two standard deviations its position on the ordination might have suggested that it is about the mean for each taxori and clearly illustrate that more closely related to the K. lactis DNA subgroup. Closer little, if any, overlap exists between groups. Interestingly, examination revealed that strain 82-231 lacked, in addition to several type strains (Fig. 1, squares) tended to be closer to the activities listed above, ADH3, ADH4, SODS, and aGL1 the origin. One explanation for this is that the maintenance activities, which are normally found in K. thermotolerans. of these strains in culture collections over many years may Therefore, the position of this strain nearer the ordinate (Fig. have caused some enzyme activities to disappear. Axis Y3 1) is better explained by default than by the possibility that it (data not shown) further separated K. dobzhanskii from the acquited other genes by hybridization. Strains 82-204 and K. lactis DNA subgroup. 82-231 were isolated from Drosophila at the sanie time, and Figure 2 shows that the first axis (Yl) had a broad strain 79-139 came from black knot. Strains 79-139 and representation in all seven enzymes used in this study. The 82-231 differed from the rest of the species in their ability to second axis (Y2) was more heavily weighted by certain utilize melibiose as a sole carbon source (data from our electromorphs of SOD, ADH, MDH, and aGL. Altogether, laboratory records), but strain 82-204 was indistinguishable the first three canonical variables accounted for 35% of the from other K. thermotolerans strains. data structure. This would generally be considered low, until Anamorphs. A number of Kluyveromyces species are it is realized that the original binary data matrix had a lesser considered to have anamorphs in the genus Candida (15). dimension of 68 and, therefore, the potential for generating Species af Candida have been examined electrophoretically as many non-zero eigenvalues. by other researchers, but few comparisons of the Outliers. Among the Kluyveromyces strains included in isoenzymes of anamorphs and their putative teleomorphs this analysis, the most striking outliers were three atypical have been made (19). Three strains identified as Candida strains of K. thermotolerans, strains 79-139, 82-204, and sphaerica (presumed to be the imperfect form of K. lactis) 82-231 (Fig. 1). The pooled relative frequencies of these (strains 80SM2-5, 80SM32-5, and 80SM35-14) and one rep- organisms are compared with those of other Kluyveromyces resentative of Candida kefir (strain 80SM3-4), which is taxa in Table 4. Only those electromorphs that were present physiologically indistinguishable from K. marxianus, were at least in typical or atypical strains of K. thermotolerans are included in this study. Although the anamorphs exhibited 98 SIDENBERG AND LACHANCE INT. J. SYST.BACTERIOL.

TABLE 4. Comparison of popled electromorph frequencies of atypical strains of K. thermotolerans and other Kluyveromyces taxav K. luctis K. m~irxianrrs K. thermotoleruns Band K. dobzhanskii DNA subgruup DNA subgroup (n = 18)' Atypical Typica I (n = 39) (n = 18) (n = 3) (n = 38) ADH3 0.67 0.97 ADH4 0.66 0.03 0.67 0.95 ADH5 0.16 0.95 0.78 1.oo 0.95 ADH6 0.06 0.22 MDH2 0.05 0.50 MDY3 0.67 0.08 h4DH4 0.33 0.37 MDHS 0.16 0.05 1.oo 0.55 h4DH6 1.oo 0.92 0.72 0.67 0.81 SOD2 0.68 SOD3 0.88 0.21 0.95 0.6'1 SOD6 1.00 0.82 1.00 SOD8 0.16 0.28 0.33 0.94 EST6 0.03 EST8 0.36 0.28 0.42 EST14 1.00 0.73 EST16 0.95 0.86 EST18 1.oo EST23 0.67 0.11 0.33 EST26 0.06 0.10 0.33 0.05 ctGLl 0.11 0.67 0.79 aGL3 0.89 0.06 0.67 0.53 ctGL5 0.03 0.67 0.63 ctGL6 0.24 ctGL7 0.82 E$G2 0.03 0.84 1.00 E$G2 0.68 EPG2 0.03 0.89 0.03 '' No value indicates a frequency of 0.00. n, Number of samples tested. some electrophoretic differences when they were compared patibility group and even beyond (ADHS, MDH6, and with other strains, their overall patterns resolved by PGL1) or electromorphs found exclusively in K. dobzhanskii multivariate analysis (Fig. 1, triangles) agreed with the (SOD7 and EST4), K. thermotolerans (SOD2, EST18, and putative anamorph-teleomorph relationships. aGL6), K. waltii (MDW1 and ESV), the K. lactis DNA Group structure among taxa. The relative frequencies of subgroup (ADH9, ADH12, SODS, EST1, EST13, EST25, electromorphs pooled by taxon confirmed the group stFc- and (xGL9), or the K. marxianus DNA subgroup (ADH8, t ure determined by reciprocal averaging of the individual MDH8, EST21, and PGL3). Note that many of these enzyme strains. It was relatively easy to identify (Table 5) activities are not likely to be connected with the nutritional electromorphs shared across the K. marxianus mating com- criteria used to recognize these taxa. The group structure was substantiated by a principal component analysis of the pooled frequencies. Figure 3 shows the first three principal axes, which accounted for 66% of the total variation in frequencies shown in Table 5. lr K. dobzanskii The first component (27% of the variation) separated the K. PC 3(16%) marxianus DNA subgroup from the other taxa. Elec- I K. t hQrmotolQrans tromorphs EpG2, EPG4, MDHS, SODl, SOD5, ESTS, ESTl4, EST21, and ADHll had high loading levels on this K. drosophilarum, T' component. The second component (24% of the variation) I K. thermotolerans waltii isolated and K. from the other taxa and was correlated with higher frequencies for electromorphs EPG3, MDH4, and MDHS. The third principal component accounted for 16% of the variation; it resolved K. dobzhanskii from the K. factis DNA subgroup, a trend summarizing high frequencies for bands ADH7, SOD7, and EST4. ADH7 was shared by K. dobehanskii and K. drosophilarurn, but the other two electromorphs were unique to K. dobzhanskii. Yeasts in other genera. Most Kluyveromyces taxa included -PC2 (244614 in this study were once classified in the genus Saccharomy- FIG. 3. Ordination of Kluyveromyces taxa by principal compo- ces (14). Saccharomyces strains have been investigated nent analysis of pooled frequencies. The percentage of variation extensively by using isoenzyme electrophoresis (17, 18), but represented is indicated for each &xis.Taxa are presented according the patterns reported in the literature are not comparable to to the classification of van der Walt (14). those found in this study because the choice of enzymes and TABLE 5. Electromorph frequencies pooled by taxon" K. lactis DNA subgroup K. marxicinus DNA subgroup K. dobzhanskii K. waltii K. thermotolerans Band K. vanudenii K. lactis K. drosophilarum = 18) K.fralgilis K. marxianus K. cicerisporus (n = 9) (n = 41) (n = 9)b (n = 17) (n = 13) (n = 5) (n = 2) (n = 11) ADH2 0.18 1.00 ADH3 0.22 0.95 ADH4 0.08 0.66 0.93 ADH5 1.00 0.94 0.92 0.16 0.80 0.50 0.82 0.95 ADH6 0.50 0.20 ADH7 0.08 0.72 ADH8 0.50 0.18 ADH9 0.67 0.71 0.15 ADHll 0.11 0.06 0.08 0.60 0.50 0.55 ADH12 0.11 0.18 ADH13 0.12 0.09 MDHl 0.44 MDH2 0.22 0.46 MDHf 1.00 0.94 0.08 0.22 0.07 MDH4 0.89 0.37 MDH5 0.11 0.08 0.16 0.89 0.61 MDH6 0.78 0.94 1.00 1.00 0.60 1.00 0.73 0.80 MDH7 0.83 0.60 1.00 0.55 MDH8 1.00 1.00 0.82 SOD1 0.41 1.00 1.00 0.91 0.78 SOD2 0.63 SOD3 0.89 0.88 1.oo 1.00 0.91 0.11 0.07 SOD4 0.09 0.22 SOD5 1.00 1.00 0.91 0.33 SOD6 0.44 0.88 1.00 1.00 0.07 SOD7 0.11 SOD8 0.85 0.16 0.90 SOD9 1.00 0.59 0.46 EST1 0.44 EST2 0.06 0.20 EST3 0.33 0.77 0.80 1.00 0.64 EST4 0.06 EST5 0.22 0.06 1.00 1.00 0.82 EST6 0.22 0.02 EST7 0.11 EST8 0.82 0.40 1.00 0.09 0.67 0.39 EST9 0.11 0.56 EST10 0.38 0.44 EST11 0.33 EST12 0.11 0.29 0.62 1.oo 0.50 EST13 0.22 EST14 1.00 1.00 1.oo 0.44 0.68 EST16 1.00 1.OO 0.91 0.67 0.80 EST17 0.44 0.94 0.50 0.11 EST18 0.05 EST19 0.33 0.94 EST20 0.06 0.50 EST21 0.20 0.50 0.09 EST22 0.89 0.59 1.00 0.80 0.50 0.91 EST23 0.67 0.50 0.09 0.02 EST24 0.12 0.33 0.50 EST25 0.22 0.53 EST26 0.22 0.12 0.06 0.07 aGLl 0.20 0.50 0.33 0.78 aGL3 0.89 0.88 0.92 0.50 0.22 0.54 aGL4 0.94 0.20 aGL5 0.11 0.63 aGL6 0.22 aGL7 0.11 0.76 aGL8 0.67 0.88 0.77 0.94 aGL9 0.22 PGLl 1.00 0.88 0.62 0.06 0.60 1.00 0.82 PGL2 0.06 0.69 0.06 0.40 0.50 0.82 PGL3 0.50 EPG2 0.08 1.OO 1.00 0.73 047 EPG3 0.44 0.63 EPG4 0.08 1.00 1.00 0.82 0.02 EPG5 0.89 0.77 1.00 0.83 '' No value indicates a frequency of 0.00. n, Number of samples tested. 99 ‘LOO SIDENBERG AND LACHANCE INT.J. SYST.BACTERIOL. the experimental conditions were not identical. As an exter- TABLE 6. Electromorphs identified in strains of S. cerevisiae, C. nal control and out of pure curiosity, the isoenzyme patterns sake, and P. fluxuum (of S. cerevisiae strain 79-11 were examined (Table 6). Of the Yeast strain two-band ADH pattern associated with this strain, one Band S. cerevisiae C. sake P.fluxuum electromorph was unique, and the other was present in about 79-11 79-228 80-109 one-third of the Kluyveromyces strains examined. The single ~~ ADH*“ + MDH electromorph was also shared with some strains of ADHl + Kluyveromyces. Although the two-band SOD pattern was ADH2 + not found as a pattern in any other strain, neither band was ADH3 + unique to S. cerevisiae when it was considered individually. ADH4 + Its EPG band (EpG1) was unique. Considered as a whole, ADH8 + the banding pattern of this strain of S. cerevisiae is suffi- ADH9 + ciently different to preclude a close relationship with any of MDH4 + the Kluyveromyces species considered in this study. MDHS + + Two strains not belonging to the genus Kluyveromyces MDH6 + were accidentally introduced into the survey. A strain later MDH9 + SOD3 + + identified correctly as C. sake (strain 79-228) was originally SOD5 + mislabeled as K. dobzhanskii. Only 3 bands in its 13-band SOD6 + pattern (Table 6) could be considered typical of K. EST7 + dobzhanskii. Only one electromorph was unique when it was EST8 + considered by itself, but many of the isoenzyme patterns EST12 + were unique (ADH, MDH, SOD, EST, and aGL). EST13 + A second mislabeled strain (P.fruxuum 89-109) was intro- EST15 + duced into the study as K. drosophilarum. This discrepancy EST17 + EST18 + was easily detected by electrophoresis (Table 6). This strain + fluxuum EST19 + of P. produced a strikingly unique 14-electromorph aGLl + ADH pattern (Table 6, ADH*). Its two-band MDH pattern aGL2 + was found in a number of K. thermotolerans strains. The aGL4 + SOD activity band found in P.fluxuum was also present in a aGL5 + number of the other strains studied. Although the four-band EPGl + EST pattern of this strain was not found in any other strain, EPG2 + only one of its bands was unique. Overall, only 3 of the 21 ADH* represents 14 bands not equated with bands in Kluyveromyces. electromorphs of this strain corresponded to electromorphs found in K. drosophilarum.

DISCUSSION its ability to utilize lactose and clearly appears to occupy a Ecological aspects. In general, the species of Kluy- separate ecological niche. The geographic component of this veromyces are not unusually stringent in their habitat spec- observation is also noteworthy. All strains of K. lactis used ificities, and most are rather widely distributed (14, 15). The in this study originated from the European continent two major sources of isolation of the strains investigated in (France), and all K. vanudenii and K.drosophilarum isolates this study were black knots (fungal galls of trees belonging to were from Canadian habitats. the genus Prunus) and Camembert cheese. Less well- Evolutionary and taxonomic significance. Until recently, represented habitats included oak exudates and galls and the typological thinking has dominated the study of yeast spe- digestive tract of Drosophila, Because of industrial impor- cies. In the past, phenotypic or genetic variation has been tance, the ecology of yeasts associated with dairy products used as a basis to describe numerous species, each of which has received much attention (10). Many yeasts found as had little internal variation. The genus Kluy veromyces has contaminants in milk products ferment or utilize lactose, the been studied systematically by many workers using various principal sugar of milk. Less is known about the ecology of methods and usually type strains or small numbers of culture yeasts associated with black knots (6). K. thermotolerans collection isolates. The electrophoretic assessment of varia- dominates in the mature stages of black knot developmect; tion reported here indicates that Kluyveromyces species are this species is followed by other species of Kluyveromyces, not monolithic, but represent multidimensional phenotypic in particular K. dobzhanskii and lactose-negative members clusters. Each species is characterized not only by a centroid of the K. lactis DNA subgroup, in later decay stages. The (an “average” yeast strain), but also by a peripheral zone in reason for the succession of yeast species during black knot which atypical strains may be recognized. development is not known, but it seems unlikely that this According to the biological species concept, two taxa are temporal factor was sufficient to isolate genetically K. regarded as separate species if genetic exchange does not thermotolerans and other Kluyveromyces taxa. take place when the opportunity for it to occur exists in The lactose-utilizing yeasts K. marxianus, K. fragilis, and nature. The exchange of genetic material is apparently K. cicerisporus (the K. marxianus DNA subgroup) and K. possible, under artificial laboratory conditions, even be- lactis (the K. lactis DNA subgroup) shared French tween taxa that show very low degrees of DNA complemen- Camembert cheese as a source of isolation, and yet their tarity or that differ significantly in their DNA guanine-plus- electrophoretic patterns were consistent with the idea that cytosine contents (5). Therefore, the meaningful application the two DNA subgroups are evolutionarily distinct entities of this sole criterion to species delineation must be regarded (in other words, that they have speciated). Conversely, K. with suspicion. The question underlying this study was lactis, K. drosophilarum, and K. vanudenii, all members of whether independent characters that are unlikely to be the K. lactis DNA subgroup, are electrophoretically indis affected directly and in concert by natural selection can tinguishable. K. lactis differs from the other two species by reveal meaningful patterns of taxonomic variation. Do pop- VOL. 36, 1986 VARIATION IN KLUYVEROMYCES POPULATIONS 101 ulations of yeasts shown to be genetically unrelated, but still electrophoretic relationship with the K. lactis DNA sub- capable of mating, evolve as a continuum of phenotypic group, especially with K. drosophilarum, but it is sufficiently variation stabilized by genetic interaction, or do they sepa- different to warrant maintaining it as a separate species. The rate into variable, but distinct entities? Members of taxa low level of DNA relatedness (7,9) between K. dobzhanskii recently (15) considered conspecific varieties (e.g., K. lactis and other K. lactis DNA subgroup members supports this and K. marxianus), in spite of their major genetic divergence view. as measured by DNA comparisons, were found to retain a Proposed tax on o m i c change : K Zuy ve romyce s lac ?is few electrophoretic similarities, but, more importantly, they (Dombrowski) van der Walt var. drosophilarum (Shehata, formed distinct phenotypic clusters. Mrak et Pham Sidenberg et Lachance comb. nov. K. waltii. Isolates of K. bulgaricus were absent from our Basonym: Saccharomyces drosophilarum Shehata, Mrak collection of natural isolates, and consequently the electro- et Phaff in Mycologia 47:804 (1955). Synonyms: phoretic similarities which we have identified between the Kluyveromyces marxianus (Hansen) van der Walt var. type strains of K. waltii and K. bulgaricus (12) were neither drosophilarum (Shehata, Mrak et Ph@ Johannsen et van confirmed nor contradicted. The isolates of K. waltii were der Walt in Johannsen (1980) (5); Kluyveromyces marxianus quite variable, but they appeared to be distinct from mem- (Hansen) van der Walt var. vanudenii (van der Walt et Nel) bers of the K. marxianus DNA subgroup (Fig. 1). Extensive Johannsen et van der Walt in Johannsen (1980) (5). genetic interaction does not seem to occur between these In relation to the nomenclature used above, the new taxa, and K. waltii should be retained as a separate species. variety K. lactis var. drosophilarum is meant to include K. The results of mating compatibility studies (5) partially agree phaseolosporus in addition to the synonyms listed above. K. with this finding. phaseolosporus and K. drosophilarum were also understood K. thermotolerans. Our results support the general consen- to be synonyms by van der Walt and Johannsen (15). K. sus that K. thermotolerans represents a distinct species. The cicerisporus and K.fragilis were considered to be synonyms results of different taxonomic approaches (i.e., DNA com- of K. marxianus var. bulgaricus and K. marxianus var. plementarity, mating compatibility, and electrophoretic dis- marxianus, respectively (15), and as such, they are excluded similarity) agree perfectly. Spatially, K. thermotolerans from the revised definition of K. lactis. They were also shared its habitats with the K. dobzhanskii, K. vanudenii, considered to be synonyms of K. marxianus by PhafFet al. and K. drosophilarum strains used in this study. Its apparent (Abstr. 12th Int. Congr. Microbiol.). reproductive isolation from these other taxa might be ex- All of the evidence supports the hypothesis that the plained in part by the temporal succession of yeast types in varieties of K. lactis are members of the same biological natural habitats, but more likely, K. thermotolerans reached species. The proposed separation of these varieties is based the final steps of speciation prior to its mingling with other primarily on ecological grounds. K. lactis var. Kluyveromyces species. drosophilarum has been isolated repeatedly and almost K. lactis. Based on mating compatibilities, the taxa K. exclusively from habitats which involve interacting plants bulgaricus, K. cicerisporus, K. dobzhanskii, K. and invertebrates, such as Drosophila, tree exudates, or drosophilarum, K. fragilis, K. lactis, K. phaseolosporus, K. fermenting plant extracts. K. factis var. lactis is essentially a vanudenii, and K. wikenii were relegated to seven syngam- dairy organism. The other motive for proposing two varieties ous varieties of K. marxianus (15). Electrophoretic data do is convenience. The two yeasts are unambiguously distin- not support this classification completely. K. lactis and K. guished by the presence or absence of P-galactosidase activ- vanudenii (level of DNA relatedness, 100% [7; H. J. Phaff, ity. A similar case might have been made for retaining K. M. A. Lachance, and H. L. Presley, Abstr. 12th Int. Congr. wikenii as a separate variety. There are only a small number Microbiol., Munich, Federal Republic of Germany, abstr. of known isolates of this yeast, and it is nearly identical no. S27.3, 19781) were almost indistinguishable electropho- phenotypically to K. bulgaricus, which is itself hardly dis- retically; they evidently share a common gene pool. K. tinguishable from K. cicerisporus. K. wikenii represents one drosophilarum is more heterogeneous, but it cannot be extreme of a continuum of variation, and K.fragilis is at the distinguished unequivocally from K. lactis and K. vanudenii other extreme. by electrophoresis. These results agree with the results of From a pragmatic point of view, the varietal simplification explorations at the DNA level which support the view that brought about by the revision described above does not K. drosophilarum is a member of the heterogeneous species necessarily ease identification of the taxa in question. Ac- K. lactis and that these organisms are distinct from K. cording to van der Walt (14) and to our laboratory records, marxianus. According to the DNA relatedness data and to K. dobzhanskii does not grow at 37"C, in contrast to the other data (9), K. phaseolosporus belongs to this group as response reported by van der Walt and Johannsen (15). Lack well. of growth at 37°C sets K. dobzhanskii apart from all other K. marxianus. K. cicerisporus, K. fragilis, and K. revised taxa except K. lactis var. lactis, in which this growth marxianus form a cohesive group electrophoretically, which characteristic is usually negative also. Unlike K. lactis var. is in agreement with their high levels of DNA complemen- lactis, K. dobzhanskii does not assimilate lactose. K. tarity. The absence of K. bulgaricus or K. wikenii strains in marxianus combines the unequivocal inability to assimilate our collection of natural isolates precluded assessment of maltose with the strong utilization (and fermentation) of these taxa in terms of populations; on the basis of DNA inulin. K. lactis (the revised species) is phenotypically very relatedness (7, 9) and the electrophoretic behavior of the variable. In K. lactis var. lactis, aGL, PGL, and p- type strains (12), these taxa belong to K. marxianus. galactosidase activities are usually expressed strongly. Contrary to the suggestion made by Johannsen (3,unre- Inulin utilization and growth at 37°C are reduced or absent. stricted genetic sharing between K. lactis and K. marxianus K. lactis var. drosophilarum is lactose negative, usually does not seem to prevail in nature. The electrophoretic expresses aGL and PGL activities, and, in the absence of similarity of these taxa does suggest that they represent the the latter characteristic, it assimilates inulin poorly or not at result of a relatively recent speciation event, but they are all. The inulin response of K. vanudenii is problematic. It sufficiently different to be considered fully speciated entities. was reported as negative in the original description by van K. dobzhanskii. The K. dobzhanskii population has a close der Walt and Nel (16), and it was always negative when 102 SIDENBERG AND LACHANCE INT. J. SYST.BACTERIOL.

tested in our laboratory. According to van der Walt (14), K. TABLE 7. Key for identification of Kluyveromyces species vanudenii assimilates inulin and ferments it slowly. In the Proceed latest published revision (15), the inulin response of this Characteristic to: yeast was not reported. A key to the species of Kluyveromyces is given in Table 7. la Galactose assimilated 2 b Galactose not assimilated 6 It was modified from that of van der Walt and Johannsen (15) to account for the revisions described above and to give 2a Raffinose assimilated 3 more emphasis to physiological characters. 7 b Raffinose not assimilated ACKNOWLEDGMENTS 3a Lactose assimilated 4 This work was funded by an operating grant from the Natural b Lactose not assimilated 9 Science and Engineering Research Council of Canada to M.-A.L. We thank C. P. Kurtzman for encouragement and expert advice. 4a Growth in the presence of cycloheximide" S b No growth in the presence of cycloheximide" LITERATURE CITED K. aestuarii 1. Chatfield, G., and A. J. Collins. 1980. Introduction to 5a Inulin assimilated strongly and fermented multivariate analysis. Chapman & Hall, New York. K. marxianus 2. Davis, B. J. 1964. Disc electrophoresis. 11. Method and applica- b Inulin not assimilated or assimilated weakly, not tion to human serum proteins. Ann. N.Y. Acad. Sci. fermented 1213404427. K. lactis var. lactis 3. Eilers, F. I., J. Allen, E. P. Hill, and A. S. Sussman. 1964. Localization of disaccharides in extracts of Neurospora after 6a Sucrose and raffinose assimilated electrophoresis in polyacrylamide gels. J. Histochem. K. waltii Cytochem. 12:448450. b Sucrose and raffinose not assimilated 4. Hill, M. 0. 1973. Reciprocal averaging, an eigenvector method K. delphensis of ordination. J. Ecol. 61:237-249. 5. Johannsen, E. 1980. Hybridization studies within the genus 7a Lactose assimilated 8 Kluyveromyces van der Walt emend. van der Walt. Antonie van b Lactose not assimilated 12 Leeuwenhoek J. Microbiol. Serol. 46:177-189. 6. Lachance, M. A. 1980. Yeasts associated with black knot 8a Melezitose assimilated disease of trees, p. 607-613. In G. G. Stewart and I. Russell K. lactis var. lactis (ed.), Current developments in yeast research. Pergamon Press, b Melezitose not assimilated Toronto. K. wickerhamii 7. Martini, A. 1973. Ibridazioni DNA/DNA tra specie di lieviti del genere Kluyveromyces. Ann. Fac. Agrar. Univ. Perugia 9a Maltose assimilated 10 28:157-171. b Maltose not assimilated 14 8. Nei, M. 1972. Genetic distance between populations. Am. Nat. 106:283-292. 10a Growth in the presence of cycloheximide" 11 9. Phaff, H. J. 1980. The species concept in yeast: physiologic, b No growth in the presence of cycloheximide" morphologic, genetic, and the ecological parameters, p. K. thermotolerans 635-643. In G. G. Stewart and I. Russell (ed.), Current developments in yeast research. Pergamon Press, Toronto. lla Growth at 37°C 10. Schmidt, J. L., C. Graffard, and J. Lenoir. 1979. Contribution a K. lactis var. drosophilarum l'etude de levures isolees du fromage Camembert. I. Essais b No growth at 37°C preliminaires. Lait 59:142-163. K. dobzhanskii 11. Sidenberg, D. G., and M. A. Lachance. 1982. Multiple molecular forms of exo-P-glucanase in Kluyveromyces species. Exp. 12a Ribose assimilated Mycol . 6:84-89. K. blattae 12. Sidenberg, D. G., and M. A. Lachance. 1983. Speciation, species b Ribose not assimilated 13 delineation, and electrophoretic isoenzyme patterns of the type strains of Kluyveromyces van der Walt emend. van der Walt. 13a Asci multispored (1 to 16 or more ascospores) Int. J. Syst. Bacteriol. 33:822-828. K. afiicanus 13. Sneath, P. H. A., and R. R. Sokal. 1973. Numerical taxonomy. b Asci one- to four-spored W. H. Freeman, San Francisco. K. phafii 14. van der Walt, J. P. 1970. Kluyveromyces van der Walt emend. van der Walt, p. 316-378. In J. Lodder (ed.), The yeasts, a 14a Growth at 37°C 1s taxonomic study. North-Holland Publishing Co., Amsterdam. b No growth at 37°C 16 1s. van der Walt, J. P., and E. Johannsen. 1984. Kluyveromyces van der Walt emend. van der Walt, p. 224-251. In N. J. W. 15a Inulin assimilated strongly and fermented Kreger-van Rij (ed.), The yeasts, a taxonomic study. North- K. marxianus Holland Publishing Co., Amsterdam. b Inulin not assimilated or assimilated weakly, not 16. van der Walt, J. P., and E. E. Nel. 1963. Saccharomyces fermented vanudenii nov. spec. Mycopathol. Mycol. Appl. 20:71-74. K. lactis var. drosophilavum 17. Yamazaki, M., S. Goto, and K. Komagata. 1982. Comparison of enzymes from strains of wine yeasts and their related yeasts on 16a Asci multispored, large electrophoresis. J. Inst. Enol. Vitic. Yamazaki Univ. 17:ll-19. K. polysporus 18. Yamazaki, M., S. Goto, and K. Komagata. 1983. An electropho- b Asci one- to four-spored retic comparison of the enzymes of Saccharomyces yeasts. J. K. lodderi Gen. Appl. Microbiol. 29:305-318. 19. Yamazaki, M., and K. Komagata. 1982. Asporogenous yeasts and their supposed ascosporogenous states; an electrophoretic comparison of enzymes. J. Gen. Appl. Microbiol. 28:119-138.