Heterozygosity of genes on the sex chromosome regulates Candida albicans virulence

Authors


E-mail david-soll@uiowa.edu; Tel. (+1) 319 335 1117; Fax (+1) 319 335 2772.

Summary

In the mouse model for systemic infection, natural a/α strains of C. albicans are more virulent and more competitive than their spontaneous MTL-homozygous offspring, which arise primarily by loss of one chromosome 5 homologue followed by duplication of the retained homologue (uniparental disomy). Deletion of either the a or α copy of the MTL locus of natural a/α strains results in a small decrease in virulence, and a small decrease in competitiveness. Loss of the heterozygosity of non-MTL genes along chromosome 5, however, results in larger decreases in virulence and competitiveness. Natural MTL-homozygous strains are on average less virulent than natural MTL-heterozygous strains and arise by multiple mitotic cross-overs along chromosome 5 outside of the MTL region. These results are consistent with the hypothesis that a competitive advantage of natural a/α strains over MTL-homozygous offspring maintains the mating system of C. albicans.

Introduction

The fungal pathogen Candida albicans is diploid and possesses a single mating-type locus, MTL (Hull and Johnson, 1999), which is heterozygous (a/α) in a majority (∼90%) of natural strains (Lockhart et al., 2002; Legrand et al., 2004; Tavanti et al., 2005). To become mating competent, a/α strains must undergo MTL-homozygosis (Hull et al., 2000; Magee and Magee, 2000). In vitro, spontaneous MTL-homozygosis is achieved primarily by the loss of one homologue of chromosome 5, which carries the MTL locus, followed by duplication of the retained homologue, resulting in uniparental disomy (Wu et al., 2005). This scenario differs from that of Saccharomyces cerevisiae, where a/α strains undergo meiosis to form haploid a and α cells (Herskowitz and Oshima, 1981; Haber, 1998). Unlike C. albicans, S. cerevisiae contains three mating type loci, one which is expressed (MAT) and dictates mating type, and two (HMR, HML) which are silent and contain a and α genes respectively. Haploid a and α cells therefore each contain information for the opposite mating type, and can undergo mating type switching at the MAT locus by gene conversion (Klar et al., 1979; Haber et al., 1980; Haber, 1992). This configuration assures conservation of the mating system. C. albicans cells, on the other hand, lose opposite mating type information when they undergo MTL-homozygosis (Hull and Johnson, 1999; Lockhart et al., 2002). This results in a paradox regarding conservation of the mating system. A significant proportion of natural C. albicans strains undergo MTL-homozygosis to a/a or α/α spontaneously and at relatively high frequency (Lockhart et al., 2002; Pujol et al., 2003; Wu et al., 2005). Studies of the population structure of C. albicans, however, indicate that recombination and hence mating are rare events (Pujol et al., 2005). Therefore, while a robust avenue appears to exist in nature for the transition from the a/α to a/a or the α/α genotype, there appears to be no similar robust avenue of return to the a/α genotype. Why then do a/α strains remain a majority in nature? Lockhart et al. (2005) hypothesized that a/α strains predominate in nature because they have an advantage over their MTL-homozygous offspring in colonizing hosts. If a/a and α/α offspring do not re-establish MTL-heterozygosity through mating and a subsequent return to the diploid a/α state, they would be diluted from the population as a result of diminished competitiveness with parent a/α cells for host niches. To test this hypothesis, Lockhart et al. (2005) compared the virulence of four spontaneous MTL-homozygous offspring with that of their parent a/α strains in the murine model for systemic infection. They then tested competitiveness between eight spontaneous MTL-homozygous offspring and their parent a/α strains in mixed infections. The a/α strains used in this study had been isolated from patients within a 5 year period before analysis and maintained with relatively few transfers. Lockhart et al. (2005) found that the natural a/α strains were more virulent than their a/a or α/α offspring, and when coinfected with their a/a or α/α offspring, were more competitive in colonizing the host kidney. Wu et al. (2005) demonstrated that the majority of the spontaneous MTL-homozygous offspring analysed by Lockhart et al. (2005) had been generated by the loss of one of the two chromosome 5 homologues, followed by duplication of the retained one, leading to uniparental disomy of chromosome 5. Ibrahim et al. (2005) compared the virulence of two a/α strains (SC5314, 3567) and sorbose-induced a/a and α/α offspring in the mouse model. Although their studies employed a different injection protocol and mouse line, and did not include competition experiments, they found a similar, although less pronounced, trend in virulence between parent a/α and MTL-homozygous offspring.

Here we first present data confirming that spontaneous MTL-homozygous offspring are less virulent than their natural a/α parents. Based on these findings, alternative hypotheses are then entertained for the decrease in virulence and competitiveness accompanying spontaneous MTL-homozygosis. First, the decrease would be due to homozygosis at the MTL locus through loss of the a1-α2 corepressor complex. In this hypothesis, the a1-α2 complex, which represses mating and switching (Hull et al., 2000; Magee and Magee, 2000; Miller and Johnson, 2002; Lockhart et al., 2002; 2003), would also confer virulence and competitiveness to natural a/α strains. Second, the decrease would be due to the homozygosis of genes along chromosome 5 other than those at the MTL locus. In this hypothesis, the heterozygosity of these ‘other’ (i.e. non-MTL) genes would confer virulence and competitiveness to natural a/α strains. In a third scenario, the heterozygosity of the MTL locus and ‘other’ genes along chromosome 5 would combine to confer maximum virulence and competitiveness.

The second of the two proposed hypotheses in turn would explain the apparent success of natural MTL-homozygous strains that have established themselves in host niches. In contrast to the majority of spontaneous MTL-homozygous offspring, which arise through uniparental disomy, natural MTL-homozygous strains would be expected to arise from natural a/α strains through gene conversion at the MTL locus, or through multiple cross-overs outside the MTL locus that leave heterozygous those ‘other’ genes along chromosome 5 that conferred virulence.

To discriminate between these alternatives, the virulence and competitiveness of deletion derivatives of natural a/α strains lacking MTLa1, MTLα2, the entire MTLa locus or the entire MTLα locus were compared with parent a/α strains in the mouse model for systemic infection by single or multiple strain injection experiments respectively. In addition, natural MTL-homozygous strains were analysed for heterozygosity along chromosome 5 as well as for levels of virulence. The results indicate that both the heterozygosity of the MTL locus as well as that of other genes along chromosome 5 contribute to the virulence and competitiveness of natural a/α strains, the latter providing the larger contribution. These results are consistent with the hypothesis that the dominance of a/α strains in nature and the associated maintenance of the mating system in C. albicans are the result of an advantage of natural a/α strains over MTL-homozygous offspring for host colonization.

Results

Spontaneous MTL-homozygosis results in a loss of virulence

Because the difference in virulence between natural a/α strains and their spontaneous MTL-homozygous offspring were found to be pronounced in our previous analysis (Lockhart et al., 2005), but less pronounced in the analysis by Ibrahim et al. (2005), we deemed it necessary to verify our original observations, given that they provide the basis for the hypotheses developed and tested here. The three previously tested natural a/α strains (P37039, P75063, P37037) and an additional unrelated natural a/α strain (P34048) were compared with their spontaneous MTL-homozygous offspring for virulence in the mouse model for systemic infection. All of the tested offspring arose through uniparental disomy. For each a/α parent strain and offspring, a 250 μl inoculum containing 106 cells was injected into each of a set of 10 outbred ND4 mice and host survival monitored. All four natural a/α strains killed 50% of the mice between 2 and 10 days, and 100% within 22 days (Fig. 1A–D). In contrast, none of their spontaneous MTL-homozygous offspring killed more than 30% of the mice in a 22 day period (Fig. 1A–D), verifying our previous observations (Lockhart et al., 2005).

Figure 1.

Reductions in the virulence of MTL-homozygous offspring and MTLa1, MTLα2, the a copy of the MTL locus and the α copy of the MTL locus. Each strain was individually injected into the tail vein of a set of 10 outbred mice, and host survival monitored.
A–D. The survival curves for four unrelated natural a/α strains and an MTL-homozygous offspring. Parent and offspring strains were injected in parallel into sets of outbred ND4 mice from the same batch.
E–J. The survival curves for six natural a/α strains and MTLa1 or MTLα2 deletion derivatives.
K and L. The survival curves for two natural a/α strains and deletion derivatives of the entire a or α copy of the MTL locus. Symbols: filled circle, natural a/α strains; filled square and triangle, spontaneous a/a or α/α offspring; open circle, Δa1 or Δa; open square, Δα2 or Δα.

Deletion of MTLa1 and MTLα2 causes a small decrease in virulence

Because spontaneous MTL-homozygous offspring arise in vitro primarily through the loss of one chromosome 5 homologue followed by duplication of the retained homologue, all genes along chromosome 5 are homozygous (Wu et al., 2005). Three alternative hypotheses were therefore entertained for the associated loss of virulence. First, the reduction in virulence could be due solely to homozygosis at the MTL locus, which results in the loss of the a1-α2 corepressor complex. Second, it could be due solely to homozygosis of one or more non-MTL genes along chromosome 5. Third, it could be due to a combination of the two. To test whether the a1-α2 complex contributes to virulence, a dominant selection method (Reuss et al., 2004) was used to delete MTLa1 in three natural a/α strains (P76067, P37039 and P37037) and MTLα2 in four natural a/α strains (P37039, P37037, P75063, P78042). This strategy resulted in deletion derivatives that did not harbour auxotrophic markers. Each parent strain and respective deletion derivative was individually injected into each of a set of 10 mice and survival of host animals monitored.

Deletion of MTLα2 (Δα2) resulted in a dramatic loss in virulence in the a/α strain P78042 (Fig. 2I and J). P78042a/α was the least virulent of the natural a/α strains tested, causing death in 22 days in seven and five of the 10 mice injected in each of two repeat experiments (Fig. 1I and J respectively). Only one of 20 mice injected with the Δα2 derivative of P78042 in the two repeat experiments died after 22 days (Fig. 1I and J). For three of the strains (P37039, P75063, P76067), deletion of MTLa1 and/or MTLα2 resulted in a slight or modest loss of virulence (Fig. 1E, F and H respectively). For one a/α strain (P37037), the survival curves of mice injected with Δa1 or Δα2 derivatives were indistinguishable from that of the a/α parent strain (Fig. 1G).

Figure 2.

On average, natural MTL-homozygous strains are less virulent than natural MTL-heterozygous strains. Each strain was individually injected into the tail veins of a set of 10 outbred mice, and host survival monitored.
A–J. Natural a/α strains.
K–T. Natural a/a or α/α strains. For two a/α strains and all a/a or α/α strains, data from two independent experiments using different batches of outbred mice are presented in order to demonstrate reproducibility.

The same dominant selection method was used to delete either the a or α copy of the entire MTL locus in the two natural a/α strains P37039 and P76067. For bothstrains, Δα derivatives were slightly less virulent, but Δa derivatives were indistinguishable (Fig. 1K and L respectively). When the 11 offspring survival plots (Fig. 1E–L) were scored for a shorter time period to 50% host survival or a higher trailing edge of survivors, in nine of the cases (82%) the parent a/α strain proved slightly more virulent than deletion derivative, supporting the suggestion of a weak trend.

Deleting either copy of the MTL locus causes a decrease in competitiveness

In previous studies (Lockhart et al., 2005), we found that competition experiments between natural a/α and MTL-homozygous offspring could be more sensitive for distinguishing differences than comparisons of the survival plots of sets of mice injected with parental or offspring strains (Lockhart et al., 2005). We therefore performed competition experiments between four natural heterozygous strains and derivatives of those strains in which the entire a or α copy of the MTL locus was deleted. Mice were injected with 50:50 mixtures of the parent a/α strain and either the MTLa or MTLα deletion derivative. In each case the parent a/α strain was transformed with a DNA fragment containing the green fluorescent protein (GFP) gene under regulation of the constitutive actin promoter, targeted to the MET3 locus. For strains P37039 and P76067, both an MTLa and MTLα deletion derivative were tested, and for strains P37037 and P75063, only an MTLα or MTLa deletion derivative, respectively, was tested. Controls were performed in which 50:50 mixtures of the GFP-expressing parental a/α cells and untransformed (non-expressing) parental a/α cells were injected into mice. Upon death or euthanization at the time of extreme morbidity, the kidneys of each mouse were macerated and plated, and the proportion of fluorescent and non-fluorescent colonies counted. Of the 19 mice injected with test mixtures, the kidneys of 15 (79%) contained significantly more parental a/α cells than cells of the MTL deletion derivatives (a or α) (Table 1). For the four remaining mice, the difference in colonization of cells of the two strains was non-significant (Table 1). In no case did the kidneys of a test animal contain significantly more cells of the MTL deletion derivative than cells of the parental a/α strain (Table 1). Of the 11 mice injected with control mixtures of GFP-expressing and untransformed (non-expressing) parental a/α cells, the kidneys of only five contained a significant difference between coinjected cells, in two a majority of GFP-expressing cells and in three a majority of untransformed cells (Table 1). The results of these control experiments indicate that transformation with the GFP cassette and the CaSAT1 marker did not affect competitiveness. These experiments indicate that cells that are heterozygous at the MTL locus are more competitive than cells homozygous at the MTL locus.

Table 1.  Natural a/α strains are more competitive than MTLa2 and MTLα1 derivative strains in mixed infections.
StrainCombinationMouseNumber cells analysedPercent colonizing kidneySignificanceaCompetitiveness
Fluorescent (GFP)Non-fluorescent
  • a. 

    Significance was assessed by the chi-square test. A P-value < 5 × 10−2 was considered significant.

  • NS, not significant.

P37037a/α (GFP); Δα126877235 × 10−19a/α > a
219285151 × 10−22a/α > a
a/α (GFP); a/α
(control)
1644753NS 
2484456NS 
39238622 × 10−2[a/α > a/α (GFP)]
43335347NS 
P37039a/α (GFP); Δα1955842NS 
211672283 × 10−6a/α > a
314472281 × 10−7a/α > a
a/α (GFP); Δa12770303 × 10−2a/α > α
27873275 × 10−5a/α > α
38464369 × 10−3a/α > α
41604357NS 
a/α (GFP); a/α
(control)
11974654NS 
210770303 × 10−5[a/α (GFP) > a/α]
346436645 × 10−9[a/α > a/α (GFP)]
P76067a/α (GFP); Δa13977238 × 10−4a/α > α
26862384 × 10−2a/α > α
3142366346 × 10−33a/α > α
476584166 × 10−81a/α > α
a/α (GFP); Δα114363372 × 10−3a/α > a
a/α (GFP); a/α
(control)
1604555NS 
268658427 × 10−5[a/α (GFP) > a/α]
P75063a/α (GFP); Δα15024951NS 
228965356 × 10−7a/α > a
32005347NS 
421670302 × 10−9a/α > a
542061393 × 10−6a/α > a
a/α (GFP); a/α
(control)
139743573 × 10−3[a/α > a/α (GFP)]
2715149NS 

Virulence of natural MTL-homozygous strains

For strains P37039, P75063 and P37037, uniparental disomy had a far greater effect on the reduction of virulence than deletion of a1, α2 or either copy of the MTL locus (Fig. 1), indicating that the contribution to the virulence of a strain by heterozygosity of non-MTL genes along chromosome 5 is greater than the contribution of the heterozygosity of the MTL locus. One would therefore expect a/a and α/α strains established in nature to (i) be less virulent on average than natural a/α strains and (ii) be heterozygous for non-MTL genes along chromosome 5 (i.e. to have arisen from gene conversion or multiple cross-overs outside of the MTL locus rather than uniparental disomy). To test the first of these predictions, a comparison of virulence was made between 10 natural unrelated a/α strains and 10 natural unrelated MTL-homozygous strains (6 a/a, 4 α/α). Nine of the 10 (90%) natural a/α strains caused death or extreme morbidity in 90–100% of injected mice within 22 days (Fig. 2A–I); one, P78042, caused death or extreme morbidity in only 70% of mice in that time period (Fig. 1J). Only three (30%) of the 10 natural MTL-homozygous strains were as virulent as the majority of natural a/α strains (Fig. 2K–M). The majority of natural MTL-homozygous strains were either weakly virulent (Fig. 2N and O) or relatively avirulent (Fig. 2P–T). No difference was observed between natural a/a and natural α/α strains. Repeat experiments performed 6 months apart for two natural a/α strains (Fig. 2B and C) and all of the MTL-homozygous strains (Fig. 2K–T) demonstrated reproducibility of the results. In Table 2, comparisons are provided of the virulence of natural a/α, natural a/a or α/α, and spontaneous a/a or α/α offspring of natural a/α strains. It is clear that the average virulence of natural MTL-homozygous strains falls between natural a/α strains and their spontaneous MTL-homozygous offspring.

Table 2.  Natural a/a and α/α strains are on average less virulent than natural a/α strains, but more virulent than spontaneous a/a or α/α strains.
MTL genotype and originNo. of strainsPercent strains causing death or extreme morbidity in 90–100% of injected animals in 22 daysaPercent strains causing death or extreme morbidity in 30–90% of injected animals in 22 daysaPercent strains causing death or extreme morbidity in 0–30% of injected animals in 22 daysa
  • Virulence was assessed by single strain injection in the mouse model for systemic infection.

  • a. 

    Data for natural a/α strains can be found in Figs 1 and 2, for natural a/a or α/α strains in Fig. 2, and spontaneous a/a or α/α offspring in Lockhart et al. (2005) and Fig. 1. The eight spontaneous a/a or α/α strains originated from four natural a/α strains.

  • b. 

    Mean of two independent experiments.

  • c. 

    Spontaneous a/a and α/α offspring of natural a/α strains.

Natural a/α1090100
Natural a/a or α/α1020b40b40b
Spontaneous a/a or α/αc800100

The heterozygosity of non-MTL genes in natural MTL-homozygous strains

To test the second prediction concerning natural MTL-homozygous strains, namely that they are heterozygous for non-MTL genes along chromosome 5, we first identified 18 sites distributed along the length of the chromosome that were polymorphic in the natural a/α strain P37037, which exhibited the average level of virulence for natural a/α strains (Fig. 2E). These sites were on average 500 bp and contained on average 9 polymorphic bases. The zygosities of these sites were analysed in six natural a/α strains (Table 3) and 11 natural a/a or α/α strains (Table 4). The proportion of natural a/α strains that were polymorphic at the test sites ranged between 50 and 100% for 17 of the 18 sites (Table 3). One site (2284B) was polymorphic in only P37037 (Table 4). The proportion of natural a/a and α/α strains that were polymorphic at each of the 18 sites ranged between 0 and 100% (Table 3). The mean (±standard deviation) for the 18 sites was 80% (±26%) for natural a/α strains (Table 3) and 53% (±28%) for natural MTL-homozygous strains (Table 4). The difference was significant (Student's t-test, P = 2.7 × 10−3). Of the 18 sites analysed, nine were 100% polymorphic among natural MTL heterozygous strains (Table 3), while only one was 100% polymorphic among natural MTL-homozygous strains (Table 4). The one site that was monomorphic among MTL-homozygous strains was the YAP3 sequence (Table 4). YAP3 was the closest to the MTL locus among the initial 18 sites tested (Fig. 3A). It was located approximately 7 kb from the right end of the MTL locus. Among the six natural MTL-heterozygous strains, YAP3 was polymorphic in five (83%) (Table 3).

Table 3.  Natural a/α strains are highly polymorphic along chromosome 5.
CladeStrain, clade, MTL genotypeaProportion (%) of strains polymorphic for site
P37037P37039P75063P76035P76055P52084
IISAIIISA
Sites on chromosome 5ba/αa/αa/αa/αa/αa/α
  • a. 

    Clade groupings after Soll and Pujol (2003).

  • b. 

    Sites sequenced averaged 500 bp and contained on average 9 polymorphic bases. See Fig. 3A for test site locations.

  • c. 

    Means (±standard deviation) computed for the proportions of strains polymorphic for sites other than MTL.

  • Polymorphic sites identified in the natural a/α strain P37037 were then analysed in five additional natural a/α strains selected randomly.

  • P, polymorphic; N, non-polymorphic.

10080APPNNNP50
104PPNPPN67
103PPNPPP83
10137APPPPPP100
10170APPPPPP100
10170DPPPPPP100
(MTL)c(P)(P)(P)(P)(P)(P)(100)
YAP3PNPPPP83
1854APPPPPP100
1990CPPPPPP100
1990APPPPPP100
10170BPPPPPP100
2466CPPPPPP100
2466APPPPPP100
2284BPNNNNN17
116PNNPNP50
2222APNNPPN50
2222CPNNPNP50
YIR12PNNPPP67
         = 80 (±26)c
Table 4.  Natural a/a and α/α strains are less polymorphic along chromosome 5 than natural a/α strains.
CladeStrain, clade, MTL genotypeaProportion (%) strains polymorphic for site
12CL26P94015P87P37005P60002WO-119FGC75P78048P57072
IIISAISAIIISAIII
Sites on chromosome 5ba/aa/aa/aa/aa/aa/aα/αα/αα/αα/αα/α
  • a. 

    Clade groupings after Soll and Pujol (2003).

  • b. 

    Sites sequenced averaged 500 bp and contained on average 9 polymorphic bases. See Fig. 3A for test site locations.

  • c. 

    Means (±standard deviation) computed for the percents of strains polymorphic for sites other than MTL.

  • Polymorphic sites identified in the natural strain P37037 were analysed in 11 natural a/a and α/α strains.

  • P, polymorphic; N, non-polymorphic.

10080ANNNNPNNNPPN27
104PPNNPPNNNNN36
103PPNPPPNNPNN55
10137APPPPPPPNPPP91
10170APPPPPPPPPPP100
10170DNPNPPNNNPPN45
(MTL)(N)(N)(N)(N)(N)(N)(N)(N)(N)(N)(N)(0)
YAP3NNNNNNNNNNN0
1854APPPNPPNPPNN64
1990CPPPNPNNPPPN64
1990ANPPPPPPPPPN80
10170BNPPPPNPNPPP73
2466CNPPPNNPNNPP55
2466ANNPNPNNPNPN36
2284BNNPNNPNNNPN27
116NPPPNNNNPPP55
2222ANPPPPNNNPPP64
2222CPPPPPPPPPPN90
YIR12NNNPNNPNPPN36
              = 53 (±28)c
Figure 3.

A map is presented of the polymorphic sites along chromosome 5 analysed in natural MTL-heterozygous and natural MTL-homozygous strains.
A. The map for the first 18 sites distributed throughout chromosome 5. The locations of the MTL locus, the centromere and the SfiI site in the RPS repetitive sequence are indicated.
B. An expanded region of the map that includes additional sites close to the MTL locus. The map was derived from the Stanford Genome Technology Centre C. albicans sequencing project database and the contig assembly version 20 website constructed by Whiteway and colleagues.

The heterozygosity of non-MTL sites in natural MTL-homozygous strains could have evolved either before or after MTL-homozygosis. The former would be consistent with gene conversion or multiple cross-overs outside of the MTL locus along chromosome 5; the latter would be consistent with uniparental disomy. We therefore compared polymorphisms between natural MTL-heterozygous and natural MTL-homozygous strains. The great majority of polymorphisms were shared (data not shown), indicating that the markers employed were good indicators of ancestral heterozygosities, not recent mutations.

When the percent of natural a/α strains that were polymorphic at each site was plotted along the length of chromosome 5, a central, highly polymorphic region was evident, which we will refer to as region II (Fig. 4A). It included the MTL locus and the putative centromere (Sanyal et al., 2004) (Fig. 4A). Region II, which was approximately 350 kb in length, included 10 sites, nine of which were 100% polymorphic among natural a/α strains (Table 3; Fig. 4A). Region II terminated midway between sites 2466 A and 2284B, a stretch containing the highly recombinational RPS locus (Iwaguchi et al., 1992; Chibana et al., 1994; Lockhart et al., 1996; Pujol et al., 1999) (Fig. 4A). The 300 kb region (region I) preceding region II contained three sites which were 50–83% polymorphic among natural a/α strains (Table 4; Fig. 4A). The region (region III) following region II contained five sites that were 17–67% polymorphic among natural a/α strains (Fig. 4A).

Figure 4.

Three regions of chromosome 5 are delineated by the extent of polymorphisms in natural MTL-heterozygous strains as region I, II and III. Natural MTL-homozygous strains exhibit dramatic reductions in percent polymorphisms in region II.
A. The percent polymorphisms of natural MTL-heterozygous and natural MTL-homozygous strains are plotted as a function of chromosome location for the original 18 sites distributed throughout chromosome 5 (data derived from Tables 4 and 5).
B. An expansion of the region of the map around the MTL locus that reveals the dramatic decrease in polymorphisms for sites close to the MTL locus in natural MTL-homozygous strains (data derived from Table 6). See Fig. 4 for chromosome distances.

The plot for the proportion of natural MTL-homozygous strains polymorphic for sites in region I was similar to that for MTL-heterozygous strains, although the values were consistently lower (Fig. 4A). The proportions of natural MTL-homozygous strains polymorphic at the first two test sites in region II (10137A, 10170A) were close to or equal to 100%, respectively, similar to that of natural MTL-heterozygous strains. However, the proportions of natural MTL-homozygous strains polymorphic at subsequent sites along region II were far lower than that of MTL-heterozygous strains (Fig. 4A). YAP3, the closest marker to the MTL-locus, was, as previously noted, homozygous for all 11 natural MTL-homozygous strains, while polymorphic for five of the six MTL-heterozygous strains. The plot for sites in region III of MTL-homozygous strains was similar to that for natural MTL-heterozygous strains (Fig. 4A).

These results were more consistent with a model for MTL-homozygosis that involved multiple mitotic cross-overs along chromosome 5 in non-MTL regions. They were not consistent with either a specific gene conversion at the MTL locus or uniparental disomy of chromosome 5. If multiple mitotic cross-overs outside of the MTL locus was the cause of MTL-homozygosis in natural MTL-homozygous strains, one would expect sites very close to the MTL-locus to be less polymorphic on average than sites further away, as was observed for YAP3. We therefore identified, four additional polymorphic sites close to the MTL locus in the natural a/α strain P37037, andthen analysed them in nine additional natural MTL-heterozygous strains and 11 natural MTL-homozygous strains. U4 and U5 were 5540 and 50 bp, respectively, to the left, and D1 and D2 were 360 and 4770 bp, respectively, to the right of the MTL locus (Fig. 3B). Among the 10 natural MTL-heterozygous strains, the percent that were heterozygous at these four sites and YAP3 ranged between 80 and 100%, with a mean of 94 ± 9% (Table 5). Among the 11 natural MTL-homozygous strains, the proportion ranged between 0 and 36%, with a mean of 20 ± 13% (Table 5). When the proportion of strains polymorphic at each of these MTL-associated sites was plotted along chromosome 5, it was clear that the most monomorphic sites along chromosome 5 in natural MTL-homozygous strains clustered around the MTL locus (Fig. 4B).

Table 5.  Natural a/a and α/α strains are far less polymorphic than natural a/α strains at sites very close to the MTL locus.
Sites on chromosome 5bNatural a/α strain, cladea and MTL genotypePercent strains polymorphic for site
P22078P48086P52084P57003P75038P75071P76023P76035P76055P78038
SAISAIISAIIIIII
a/αa/αa/αa/αa/αa/αa/αa/αa/αa/α
U4PPPPPPPPPP100
U5PPPPPPPPPP100
(MTL)(P)(P)(P)(P)(P)(P)(P)(P)(P)(P) 
D1PNPPPPPPPN80
D2PPPPPPPPPP100
YAP3 PPPPPPPPPP100
             = 96 (±9)c
Percent polymorphic1008010010010010010010010080 
          = 93 ± 12   
Sites on chromosome 5bNatural a/a and α/α strain, clade and MTL genotypePercent strains polymorphic for site
12CL26P94015P87P37005P60002WO-119FGC75P78048P57072
IIISAISAIIISAIII
a/aa/aa/aa/aa/aa/aa/aa/aa/aa/aa/a
U4PPNPPNNNNNN36
U5NPNPNNNNNNN18
(MTL)(N)(N)(N)(N)(N)(N)(N)(N)(N)(N)(N) 
D1PNNNPNNPNNN27
D2NPNPNNNNNNN18
YAP3 NNNNNNNNNNN0
              = 20 (±13)c
Percent polymorphic4060060400020000 
          = 27 (±31)c    

Polymorphisms on other chromosomes

Because spontaneous a/a and α/α offspring arise primarily by the loss of one homologue of chromosome 5 followed by duplication of the retained homologue, we tentatively concluded that the loss of virulence resulted from uniparental disomy of chromosome 5. However, the possibility existed that in these offspring uniparental disomy may occur for chromosomes other than chromosome 5. We therefore tested whether polymorphisms at 14 sites distributed on the seven non-sex chromosomes of three natural a/α strains (P37037, P37039, P75063) were retained in their MTL-homozygous offspring. Offspring and parental strains proved identical for all 14 sites tested (Table 6).

Table 6.  Strains that undergo spontaneous MTL-homozygosis through uniparental disomy do not undergo similar changes in other chromosomes, and natural a/α and a/a or α/α strains exhibit similar polymorphisms throughout the chromosomes.a
 StrainsChromosome RChromosome 1Chromosome 2Chromosome 3Chromosome 4Chromosome 6Chromosome 7
1210057A1021610205A1012510196A799020166100481231014020099140
  • a. 

    Sites sequenced averaged 500 bp and contained on average 9 polymorphic bases.

  • P, polymorphic; N, non-polymorphic.

Natural a/α strains and their spontaneous a/a or α/α offspringP37037a/αNPNPPPPNPNPNNP
P37037α/αNPNPPPPNPNPNNP
P37039a/αPPPPPPPNPPNNNP
P37039α/αPPPPPPPNPPNNNP
P75063a/αPPNPPPPPNNNPNP
P75063a/aPPNPPPPPNNNPNP
Natural a/α strainsP22078a/αPPPPPPPPPPNPNP
P48086a/αPPPPPPPNNPNNPP
P52084a/αPPNPNPPNNPNPPP
P57003a/αNPPPPPPNPNNNNP
P75038a/αNPPPPNPPNPNNNP
P75071a/αNPPPPNPPNPNNPP
P76023a/αPPNPNNPPNPNNNP
P76035a/αNPNPPPPNPPNNNP
P76055a/αPPNPPPPPNPNPNP
P78038a/αPPNPPPPNPPNNNP
P37037a/αNPNPPPPNPNPNNP
P37039a/αPPPPPPPNPPNNNP
P75063a/αPPNPPPPPNNNPNP
Percent polymorphic6110046100857710047467783023100
Natural a/a and α/α strainsWO-1α/αNPNPPPPPNPPNNP
P57072α/αNPPPNNPPNPNPNP
L26a/aNPPPPPPNPNNNNP
P37005a/aNPPPPPPNNPNNNP
12Ca/aNPPPPPNNNPNPNP
P87a/aPPNPNPPNNNNNNP
GC75α/αPPPPNPPPPNNNPP
P60002a/aNPNPPNNPNPNNNP
19Fα/αPPPPPPPNPNNNNP
P78048α/αNPPPPPNNNPNNPP
P94015a/aPPNPPNPNNPNNNP
Percent polymorphic36100631007373733627649818100

We also compared these sites among a collection of 13 natural a/α strains and 11 natural a/a or α/α strains for polymorphisms. The proportion of strains polymorphic at each site was similar between the two groups – i.e. similarly high, moderate or low (Table 6). On average, however, natural a/α strains were slightly more polymorphic than natural a/a or α/α strains.

Virulence is not associated with isochromosome formation

Recently Selmecki et al. (2006) found that isochromosomes of chromosome 5 were gained or lost in association with increased and decreased azole drug resistance respectively. These isochromosomes were composed of two identical chromosome 5 left arms flanking the centromere (Selmecki et al., 2006). Chromosome 5 harbours TAC1, which regulates the ABC transporter genes CDR1 and CDR2 (Rustad et al., 2002; Coste et al., 2006). Hence, the formation of isochromosomes changes gene dosage and azole resistance. We tested whether natural MTL-homozygosity is associated with the formation of isochromosomes of chromosome 5. Chromosomes of nine natural MTL-homozygous strains exhibiting the full range of virulence in the mouse model were first separated by CHEF and Southern blots probed for four markers along chromosome 5, 1990C, 10137A, del29 and 10080A. Hybridization was restricted exclusively to the chromosome 5 homologues (Supplementary material– Fig. S1A). There was no indication of a chromosome 5 isochromosome in any of the nine strains (Supplementary material– Fig. S1A). We then digested total genomic DNA of the 11 natural MTL-homozygous strains and three spontaneous MTL-homozygous offspring, P37037α/α-1, P37039α/α-1 and P75063a/a-1, with EcoNI and probed Southern blots with a probe for the chromosome 5 binding site for centromeric Cse4p (Sanyal et al., 2004; Selmecki et al., 2006). The EcoNI fragment containing Cse4p binding sites is approximately 30 kb in normal chromosome 5, but approximately 10 kb in the isochromosome (Supplementary material– Fig. S1B) (Selmecki et al., 2006). None of the natural MTL-homozygous strains or the MTL-homozygous offspring contained a hybridizing 10 kb fragment. These results demonstrate that loss of virulence and competitiveness in spontaneous MTL-homozygous offspring is not associated with chromosome 5 isochromosome formation, and that our analysis of polymorphisms along chromosome 5 was not biased by the presence of isochromosomes.

Discussion

Spontaneous MTL-homozygosis results in a decrease in virulence and competitiveness

We previously hypothesized that the reason the majority of natural C. albicans strains were predominantly a/α was because they had an advantage over their spontaneous a/a or α/α offspring in colonizing a host, and presented evidence supporting the hypothesis (Lockhart et al., 2005). Ibrahim et al. (2005), using sorbose-induced random chromosome loss (Janbon et al., 1998; Bennett and Johnson, 2003), found far less of a loss of virulence in MTL-homozygotes than we did in spontaneous MTL-homozygous offspring (Lockhart et al., 2005). Because the hypotheses we developed in the present investigation stemmed from our initial findings (Lockhart et al., 2005), we compared again the virulence of the three original unrelated natural a/α strains and their spontaneous MTL-homozygous offspring, and added a new natural a/α strain and its offspring to the comparison. The spontaneous MTL-homozygous offspring again proved in every case to be less virulent than their a/α parent strain, confirming our previous results (Lockhart et al., 2005). The difference between our results and those of Ibrahim et al. (2005) in the extent of the effect on virulence is due most likely to dissimilar protocols, including differences in the mice employed (ND4 in the former versus BALB/c in the latter), differences in the volume of the inoculum (250 μl in the former versus 500 μl in the latter), differences in the origin of MTL-homozygous offspring (spontaneous in the former versus sorbose induction in the latter) and the origin of strains. Additionally, our previous study uniquely included competition experiments between natural a/α strains and spontaneous MTL-homozygous offspring, the results of which supported our conclusions.

The MTL locus contributes to virulence and competitiveness

We previously presented evidence that spontaneous MTL-homozygosis in natural a/α strains resulted in a significant reduction in virulence and competitiveness (Lockhart et al., 2005), but because this occurred primarily through uniparental disomy (Wu et al., 2005), we could not distinguish between contributions resulting from the heterozygosity of the MTL locus and contributions from the heterozygosity of non-MTL genes along chromosome 5. In a competition study, we found that an a/a/α2 derivative was more competitive than its natural a/a parent, suggesting that the a1-α2 repressor complex may be involved in conferring competitiveness (Lockhart et al., 2005). Here, we have tested this hypothesis in a more critical fashion. We found that in single strain-injection experiments, natural a/α strains with specific deletions of MTLa1, MTLα2, the entire a locus or the entire α locus exhibited a slight reduction in virulence when compared with parent strains in the majority of cases. This trend was suggestive but not overly convincing. We therefore performed competition experiments by injecting equal numbers of cells from the natural parent a/α strain and cells from deletion derivatives of either the a or α copy of the entire MTL locus. The results were more convincing. The data indicated that in a natural a/α strain, heterozygosity of the MTL locus, presumably through the a1-α2 repressor complex, contributes to competitiveness. However, because the reductions in virulence and competitiveness were far greater in a/a and α/α offspring that arise through uniparental disomy of chromosome 5 than in deletion derivatives for MTLa1, MTLα2, the entire MTLa locus or the entire MTLα locus, we conclude that the heterozygosity of non-MTL genes contributes to the chromosome 5-associated advantage natural a/α strains have over spontaneous MTL-homozygous offspring to a greater extent than heterozygosity at the MTL locus.

Natural MTL-homozygous strains

Approximately 10% of natural strains are MTL-homozygous (Lockhart et al., 2002; Legrand et al., 2004; Tavanti et al., 2005). We found that on average natural MTL-homozygous offspring were not as virulent as natural MTL-heterozygous strains. Indeed in 50% of cases, they were relatively non-virulent in the mouse model for systemic infection. Because the heterozygosity of non-MTL genes appears to be the major contributor to virulence and competitiveness, we expected natural MTL-homozygous strains to be on average less heterozygous for non-MTL genes along chromosome 5 than natural MTL-heterozygous strains. This proved to be the case. A significant decrease in polymorphisms was observed along chromosome 5 in natural MTL-homozygous strains primarily in a central region (region II), which harboured the MTL locus and the centromere. Interestingly, the three most virulent natural MTL-homozygous strains (L26, GC75 and P37005; Fig. 5) exhibited some of the highest levels of heterozygosity for that group. While there was this loose correlation between the level of virulence and the general level of heterozygosity along chromosome 5, attempts to correlate virulence and heterozygosity at a particular region of chromosome 5 have so far failed (data not shown), suggesting that the heterozygosity of more than one non-MTL gene may be involved in chromosome 5-associated virulence and competitiveness. Interestingly, Coste et al. (2006) demonstrated that homozygosis of hyperactive alleles of TAC1, which encodes a regulator of the ABC transporter genes CDR1 and CDR2, confers azole resistance. TAC1 resides approximately 15 kb from the MTL locus in region II of chromosome 5 (Stanford Candida Genome Project), and hence, TAC1 homozygosis and azole resistance have been found associated with MTL-homozygosis (Rustad et al., 2002; Coste et al., 2006). The increase in azole resistance associated with TAC1 homozygosis may also be associated with a decrease in virulence, which would explain why drug-resistant strains do not accumulate in nature (Pfaller et al., 1999; 2004).

Figure 5.

Diagrams of the three possible mechanisms for MTL-homozygosis in natural a/α strains. Our previous results (Wu et al., 2005) demonstrated that MTL-homozygous offspring spontaneously generated in vitro from natural MTL-heterozygous strains were predominantly due to uniparental disomy of chromosome 5 (A). The results presented here indicate that natural MTL-homozygous strains arise by multiple cross-overs outside of the MTL locus (C). There was no indication of natural MTL-homozygous strains arising by gene conversion (B).

The advantage that is conferred by non-MTL gene heterozygosity may be the result of (i) the individual functions of the distinct alleles, as has been demonstrated for ALS3, ALS9 and SAP2 in C. albicans (Staib et al., 2002; Zhao et al., 2003; Oh et al., 2005), (ii) transvection or transensing, demonstrated between wild-type alleles in Neurospora crassa (Aramayo and Metzenberg, 1996), (iii) interallelic complementation (Aramayo and Metzenberg, 1996; Grant et al., 1998; Gibson et al., 1999; Chandler et al., 2000; Mongelard et al., 2002), (iv) the formation of heteroallelic dimers (Steingrimsson et al., 2003) or (v) the presence of deleterious alleles. Experiments are now in progress to identify the non-MTL genes along chromosome 5 that confer chromosome 5-associated virulence and competitiveness through heterozygosity and the mechanisms through which these heterozygosities exert their effect.

The mechanism of MTL-homozygosis in natural strains

If natural MTL-homozygous strains were generated through uniparental disomy of chromosome 5, all markers along chromosome 5 would be homozygous (Fig. 5A), as they are in the great majority of spontaneous MTL-homozygous offspring of natural MTL-heterozygous strains (Wu et al., 2005). If natural MTL-homozygous strains were generated through a precise gene conversion event at the MTL locus, as is the case for the MAT locus in S. cerevisiae (Klar et al., 1979; Haber et al., 1980; Haber, 1992), genes along chromosome 5 should be as heterozygous as in natural MTL-heterozygous strains (Fig. 5B) (Wu et al., 2005). If natural MTL-homozygous strains were generated through multiple mitotic cross-overs outside the MTL locus, then non-MTL genes in the vicinity of the MTL locus should exhibit a reduction in heterozygosity (Fig. 5C). Our results are consistent with the last of these three possible mechanisms (Fig. 5C). Our results suggest that multiple cross-overs leading to MTL-homozygosity are not restricted to exact sites on either side of the MTL locus. Hsueh et al. (2006) recently demonstrated that meiotic recombination hotspots flank the mating-type locus (MAT) of Cryptococcus neoformans. Furthermore, they found that multiple recombination events frequently occur on both sides of MAT, and that a recombinational activator is located on the right side. They argue that heightened recombinational activity during meiosis may result in exchange of the MAT locus into different genetic backgrounds. In C. albicans, recombinational hot spots, if they exist, would function mitotically to generate MTL-homozygosity and mating competence, with the retention of sufficient non-MTL gene heterozygosity for a minimum degree of competitiveness in natural niches.

Maintenance of the sexual cycle through a competitive advantage in a/α strains

The results presented here support the hypothesis that a/α strains predominate in nature because they have a competitive advantage in host colonization over their natural and spontaneous MTL-homozygous offspring. The heterozygosity of one or more non-MTL genes along chromosome 5 to a greater extent and the heterozygosity of the MTL locus to a lesser extent contribute to chromosome 5-associated virulence and competitiveness. In nature, we suggest that a/a or α/α strains that spontaneously arise through uniparental disomy of chromosome 5, are rapidly diluted from the population due to the loss of competitiveness with their a/α parent strains. To survive, the offspring must mate and then return to the diploid a/α state in association with heterozygosity at the non-MTL genes that confer competitiveness. In addition, Ibrahim et al. (2005) have demonstrated that tetraploids undergo a reduction in virulence when compared with both MTL-homozygous and MTL-heterozygous diploids, a further selection pressure on the generation of a/α diploids. We further suggest that natural MTL-homozygous strains that appear to be established in nature are less competitive, on average, than natural MTL-heterozygous strains not only because of a decrease in the heterozygosity of non-MTL genes along chromosome 5, but also because of homozygosity of the MTL locus. Hence, a/α strains dominate in nature and this dominance in turn maintains the mating system. Demonstrating that natural a/α strains also are more virulent and competitive in additional animal models, including models for commensalism, would provide additional support for this hypothesis.

Experimental procedures

Strain maintenance and growth

The origins of the strains used in this study are presented in Supplementary material– Table S1. All natural strains were obtained from secondary cultures not passed more than three times after original isolation. Spontaneous MTL-homozygous offspring were identified by their capacity to switch to opaque, then verified for MTL-homozygosity by polymerase chain reaction (PCR). All strains were maintained in glycerol at −70°C. For Experimental purposes, cells from glycerol stocks were grown at 37°C on YPD (2% dextrose, 2% peptone, 1% yeast extract) agar plates. For mouse injection experiments, cells were grown at 25°C on agar containing modified Lee's medium (Bedell and Soll, 1979) supplemented with phloxine-B (Anderson and Soll, 1987) to distinguish between white and opaque colonies. The MTL genotype of each strain was verified by PCR using primers specific for MTLa1 and MTLα2, as previously described (Lockhart et al., 2005).

Generating a1 and α2 deletion derivatives

A plasmid was constructed containing the CaSAT1 marker for Nourseothricin resistance in the dominant selection flipper cassette flanked by either the MTLa1 5′ and 3′ flanking regions or the MTLα2 5′ and 3′ flanking regions in the pGEM T-easy plasmid vector (Promega, Madison WI). The primers used for amplification (MTLa1del5′, MTLa1del3′, MTLα2del5′, MTLα2del3′) are described in Supplementary material– Table S2. The plasmid containing the CaSAT1 cassette (Reuss et al., 2004) was a generous gift from Dr Joachim Morschhäuser, University of Würzburg. The plasmid was cleaved with SacI and SphI to generate a linear DNA fragment, which was used to knock out the entire open reading frame (ORF) of either MTLa1 or MTLα2 in natural MTL-heterozygous strains, using the transformation protocol described by Reuss et al. (2004). Transformants were selected on agar containing YPD medium supplemented with nourseothricin (200 μg ml−1) (Reuss et al., 2004). Deletions in transformants were confirmed first by PCR and Southern analyses, and then by Northern blot hybridization using GPA1 expression as an assay for the elimination of the a1/α2 complex. MTLa1 was deleted in natural a/α strains P76067, P37037 and P37039. MTLα2 was deleted in a/α strains P75063, P37037, P78042 and P37039. To delete the entire MTLa or MTLα locus, the same strategy was employed. The primers used for amplification of the 728 bp 5′ flanking sequence (MTLdel5′F, MTLdel5′R) and the 826 bp 3′ flanking sequence (MTLdel3′F, MTLdel3′R) for MTLa and MTLα are described in Supplementary material– Table S2. The MTLa and MTLα deletion derivatives of strains P76067 and P37039 were identified by Southern analysis.

Generating GFP-expressing a/α strains

The actin promoter (800 kb region upstream of the ACTIN orf) was amplified by PCR, and then fused with the green fluorescent protein gene (GFP)-ORF-CAG1 at the 3′ end. This construct was then inserted into a derived pK91.6 plasmid (T. Srikantha and D.R. Soll, unpubl. data) containing the nourseothricin resistance gene CaSAT1 and two MET3 flanking DNA fragments, the 800 kb 5′ upstream and 320 kb 3′ downstream of the MET3 locus. The fragment 5′MET3-Actin promoter-GFP ORF-CaSAT1-MET3-3′ was cut from the plasmid with SacII and transformed into four a/α strains, P37037, P37039, P76067 and P75063 at the MET3 locus by homologous recombination. The transforments were selected on YPD medium plates supplemented with Nourceothricin (100 μg ml−1) and verified by fluorescence microscopy. Colony fluorescence was imaged with an IVIS 200 cooled CCD optical system (Xenogen, Alameda, CA). Using fluorescence microscopy, all cells in fluorescent colonies were found fluorescent and no cells in non-fluorescent colonies were found fluorescent. Grey-scale background photographic images of culture plates were overlaid with colour images of emitted fluorescent light using Living Image® software (Xenogen).

Virulence and competitiveness in a systemic mouse model

Cells from 5-day-old colonies were grown in suspension to late exponential phase in 10 ml of modified Lee's medium, washed twice in sterile phosphate-buffered saline (PBS) (8 mM Na2PO4, 15 mM K2PO4, 2.7 mM KCl, 0.14 M NaCl, pH 7.1) and resuspended in PBS at a concentration of 4 × 106 cells per ml. Only cultures containing > 99% white cells were employed. Six- to 8-week-old outbred female ND4 mice (Harlan Sprague, Madison, WI), weighing 21–26 g, were injected through their tail veins with 1 × 106 cells in a volume of 250 μl. Mice were examined at least once a day. When a mouse exhibited the first signs of extreme morbidity (i.e. tremors, hunched back), it was euthanized in a CO2 container (Lockhart et al., 2005). In competition experiments, all procedures were the same except that the injected mixture contained 50% GFP-a/α cells and 50% unlabelled cells. When a mouse showed the first signs of illness, it was euthanized, and both kidneys were removed and ground in 2 ml of sterile PBS in a sterile mortar and pestle. Aliquots of each kidney macerate were plated on modified Lee's medium and incubated at 25°C for 5 days. The plates containing C. albicans colonies were imaged in the IVIS 200 imaging system to identify GFP fluorescent colonies representing the heterozygous strains.

Polymorphisms identified by sequence analysis

Fifteen polymorphic regions along chromosome 5 that had previously been identified (Wu et al., 2005), and seven additional ones identified here in strain P37037, were employed for analyses of heterozygosity (polymorphisms) along chromosome 5 of test strains. Selection of the new marker sequences 103, 104 and 116 was based on the SNP polymorphism map of C. albicans (Forche et al., 2004). The polymorphic regions U4, U5, D1 and D2, which were located within approximately 7 kb of the MTL locus, were identified in the Candida Genome Database (http://www.candidagenome.org/). Sites were analysed directly for base polymorphisms on sequencing gels. All of the 22 sites were demonstrated to be polymorphic in the natural a/α strain P37037 by sequencing. Polymorphic sites were on average 500 bp in length and contained on average nine single nucleotide polymorphisms. The locations of the polymorphic sites on chromosome 5 were interpreted from data in the Stanford Genome Technology Center database and the contig assembly version 20 website constructed by Whiteway and colleagues (http://candida.bri.nrc.ca/). Three additional polymorphic sites, one on chromosome 1 (10205A), one on chromosome 2 (10196A) and one on chromosome R (10057A), were previously described (Wu et al., 2005). The selection of polymorphic sites on chromosomes R, 3, 4, 6 and 7 (12, 79, 90, 123 and 140) was based on the SNP polymorphic map of C. albicans (Forche et al., 2004). The other sites (10216, 10125, 20166, 10140 and 20099) were identified in the Stanford Genome Technology Centre C. albicans database. All of the marker sites were amplified and sequenced in 11 natural MTL-homozygous strains and 10 natural MTL-heterozygous strains. For PCR amplification and sequencing, primers were designed from the selected genomic regions (Supplementary material– Table S2). The methods for PCR amplification and sequencing have been previously described (Wu et al., 2005). Sequencing was performed in both directions with an ABI sequencing apparatus (PE-ABI, Foster City, CA), using the same primers as those used for PCR amplification. Sequence analysis and polymorphism identification were performed using ContigExpress of Vector NTI advance 10.1.1 software (Invitrogen, Carlsbad, CA).

Contour-clamped homogeneous electric field (CHEF) electrophoresis

Karyotypes of natural MTL-heterozygous and MTL-homozygous strains were analysed by CHEF electrophoresis according to methods previously described (Joly et al., 1999). Southern blots of the CHEF gels were hybridized with probes for the genomic sequences 10080A, 10137A, 1990C and del29, which are distributed along chromosome 5. Southern analysis was performed according to methods previously described (Wu et al., 2005).

Acknowledgements

We thank J. Collins and K. Daniels for assembling the manuscript and figures. This research was supported by NIH Grant AI2392.

Ancillary