Candida albicans, a major human fungal pathogen, usually contains a diploid genome, but controls adaptation to a toxic alternative carbon source L-sorbose, by the reversible loss of one chromosome 5 (Ch5). We have previously identified multiple unique regions on Ch5 that repress the growth on sorbose. In one of the regions, the CSU51 gene determining the repressive property of the region was identified. We report here the identification of the CSU53 gene from a different region on Ch5. Most importantly, we find that CSU51 and CSU53 are associated with novel regulatory elements, ASUs, which are embedded within CSUs in an antisense configuration. ASUs act opposite to CSUs by enhancing the growth on sorbose. In respect to the CSU transcripts, the ASU long antisense transcripts are in lesser amounts, are completely overlapped, and are inversely related. ASUs interact with CSUs in natural CSU/ASU cis configurations, as well as when extra copies of ASUs are placed in trans to the CSU/ASU configurations. We suggest that ASU long embedded antisense transcripts modulate CSU sense transcripts.
GENOME-WIDE surveys of eukaryotic transcriptomes have led to the identification of abundant noncoding transcription in plants, mammals (Lee et al. 2009), and fungi (David et al. 2006; Samanta et al. 2006; Dutrow et al. 2008; Nagalakshmi et al. 2008; Sellam et al. 2010). The noncoding transcriptome of higher eukaryotes includes a class of various small RNAs that were extensively studied and that were shown to directly modulate gene expression (reviewed by Lee et al. 2009; Olejniczak et al. 2010; Taft et al. 2010). It was not initially appreciated that another class of long noncoding RNAs appears in 10- to 12-fold larger amounts than the coding transcripts (Nagano and Fraser 2011). Initially neglected, the long noncoding RNAs are now given much attention, being recognized as important regulators that are implicated in a large range of various functions, including epigenetic control, enhancing or mediating long-range chromatin interactions, as well as serving as scaffolds of chromatin-modifying complexes (Hongay et al. 2006; Camblong et al. 2007; Houseley et al. 2008; Guttman et al. 2009; Khalil et al. 2009; Mahmoudi et al. 2009; reviewed by Morris 2009; Nagano and Fraser 2011).
Characterization of transcription in Candida albicans, which is considered to be the most common fungal opportunistic pathogen of humans, is in its infancy. RNA interference (RNAi) in this organism has been recently discovered (Drinnenberg et al. 2009). First reports on the transcriptome have been published, revealing an extensive antisense transcriptome (Bruno et al. 2010; Sellam et al. 2010; Tuch et al. 2010).
C. albicans is a single-cell organism with a diploid genome organized into eight pairs of chromosomes. Survival in various adverse environments is an important property of this pathogen. Integral parts of C. albicans adaptation and survival strategies are alterations of large portions of the genome, including monosomy or trisomy of entire chromosomes. Similar alterations of the same chromosomes occur in the same environments, allowing survival (Rustchenko 2007, 2008). A well-known example, which is the topic of this article, is survival on the toxic sugar L-sorbose, when it is available as a sole source of carbon. The loss and gain of chromosome 5 (Ch5) up- and downregulates, respectively, the SOU1 (sorbose utilization) gene (orf19.2896) on Ch4 and confers growth, Sou+, and no growth, Sou−, on sorbose (Rustchenko and Sherman 2002; Rustchenko 2007, 2008).
Our early studies already indicated the complexity of regulation by copy number of Ch5. This chromosome carries multiple unique regions for negative control of growth on sorbose; the final number of regions is yet to be established (Kabir et al. 2005). A total of five regions A–C, 135, and 139 (Figure 1) have been rigorously confirmed by an analysis of Ch5 deletions. The regions were proposed to each encompass at least one unique negative controlling element CSU (control of sorbose utilization). The first CSU, CSU51 (orf19.1105.2), has been identified in region A (Kabir et al. 2005). Another CSU, CSU53 (orf19.3931) from region 135, is presented in this work.
We report here a new genetic element ASU (activation of sorbose utilization), adding an additional layer of control from Ch5. ASUs are embedded in CSU51 and CSU53 in the opposite orientation and are associated with a distinct, albeit weak, overexpression phenotype of the enhanced growth on sorbose, thus counteracting the repressive phenotype by CSUs. Antisense ASU transcripts are long, can be capped and polyadenylated, and seem to act as noncoding transcripts. We present evidence that, as expected, complementary sense CSU and antisense ASU transcripts interact. The final number of the CSU/ASU configurations on Ch5 has yet to be determined.
Materials and Methods
The co-overexpressing system
We used an important tool, a low copy number replicative plasmid pCA88 overexpressing the metabolic gene SOU1, thus causing the Sou− recipient strain to utilize sorbose, Sou− → Sou+ (Wang et al. 2004). This plasmid was previously used for preparing a Ch5 DNA library and subsequently cloning a negative regulatory gene CSU51, as well as unique regions carrying other putative CSUs on Ch5, as based on the reversal of sorbose utilization, Sou+ → Sou− (Kabir et al. 2005). In this work, similarly, we used pCA88 to co-overexpress SOU1 and different Ch5 sequences, as presented in the nine diagrams in Figure 2. Care was taken to assure that inserted genes had up to 1.5 kb of the upstream regions, as C. albicans is known to have long promoters (Gaur et al. 2004; Srikantha et al. 2006; Vinces et al. 2006). In addition, genes contained ∼100 bp of the downstream region.
Strains, media, plasmids, and primers
We used the C. albicans Sou− sequencing strain SC5314 and its relatively genetically stable Sou− Ura− derivative CAF4-2 (Ahmad et al. 2008). Also, we used the well-characterized prototrophic Sou− strain 3153A (Rustchenko-Bulgac and Howard 1993). The strains carrying in their genome either a control empty vector pAK156 or a vector with one ASU53, pEA249, or two ASU53s, pEA254, were prepared by individually integrating the plasmids into the LEU2 locus on Ch7.
Yeast extract/peptone/dextrose (YPD) and synthetic dextrose (SD) media were previously described (Sherman 2002). A total of 1 M sorbitol was added in SD medium, when growing transformants. Synthetic sorbose or sorbitol media were the same as SD medium, but contained 2% of either L-sorbose or sorbitol, as a sole carbon source (Rustchenko et al. 1994). To prepare solid medium, 2% (wt/vol) agar or agarose was added. Uridine (50 µg/ml) was added when needed. The proper growth and handling of cells preventing chromosomal instability was previously reported (Rustchenko-Bulgac 1991; Perepnikhatka et al. 1999; Wang et al. 2004; Ahmad et al. 2008).
All plasmids or primers that were used in this study are described in Supporting Information, File S1 and are also presented in Table S1 and Table S2, respectively. All replicative plasmids are derivatives of pCA88 (see The co-overexpressing system in Results), which is pRC2312 carrying SOU1 (Wang et al. 2004).
RNA preparation and Northern blot analysis
C. albicans cells were grown for independent colonies at 37° on plates with sorbitol medium (see above) and opened with glass beads. Total RNA was isolated according to Russo et al. (1991) or with RNeasy Midi kit (Qiagen) and additionally treated with RNase free DNase to remove all traces of genomic DNA. mRNA was isolated from total RNA using Oligotex mRNA kit (Qiagen), as recommended by the manufacturer. Either 15 μg of total RNA or 3–4 μg of mRNA was denatured and size fractioned on 1% formaldehyde gel and then transferred to positively charged nylon membrane from Ambion according to Russo et al. (1991). For loading controls, blots were hybridized with double stranded (ds) DNA probes that were prepared from the PGK1 gene (orf19.3651) or 18S rRNA and labeled with 32P (Russo et al. 1991). 32P-labeled strand-specific single-stranded RNAs (riboprobes) were generated by in vitro transcription with MAXIscript T7 kit (Ambion) according to the manufacturer’s instructions. Riboprobes were hybridized with blots, according to Clements et al. (1988). Briefly, blots were incubated in QuikHyb Hybridization solution from Stratagene overnight at 68° and washed with low- and high-stringency solutions at room temperature and at 68°, respectively. Images were processed using PhosphorImager Storm-820 (Molecular Dynamics). Individual bands were quantified using ImageQuant 5 software (Molecular Dynamics). Transcript sizes were estimated with RNA Millenium size markers–formamide (Ambion). General approaches for performing Northern analyses were adopted from Ding et al. (2007). For example, each experiment was repeated several times with different batches of RNA that were prepared from independently grown cultures.
RT–PCR and semiquantitative analysis
To synthesize strand-specific cDNA, we set the RT reaction as a duplex with gene-specific primers for a gene of interest and a control gene. The following genes were used as controls: CDC6 (orf19.5242), EMP24 (orf19.6293), TPK2 (orf19.2277), and PGK1 (orf19.3651). Synthesis was conducted with the MonsterScript Reverse Transcriptase (Epicentre Biotechnologies), as recommended by the manufacturer. The PCR amplification was also set as a duplex and was conducted for different number of cycles with Phusion Hot Start High-Fidelity DNA Polymerase (New England BioLabs), as recommended by the manufacturer (also see Cerazin-Leroy et al. 1998; Kuai et al. 2004). Prior to that, pilot duplex PCR amplifications were undertaken to optimize reaction conditions for efficiency of amplification and the lack of nonspecific bands.
Semiquantitative (sq)RT–PCR analysis was performed according to Kuai et al. (2004) and Cerazin-Leroy et al. (1998). Briefly, images of ethidium bromide-stained gels loaded with amplification products from different number of cycles were prepared with the AlphaImager IS2000 Digital Imaging system (Alpha Innotech). ImageQuant 5 software (Molecular Dynamics) was used to measure the brightness of the bands (Ahmad et al. 2008). At least three consecutive amplicons in exponential phase, as estimated by the determination coefficient value 0.97 or more were used to normalize the experimental values against the values of the control genes and then to calculate the ratios test/control genes from the averaged values.
Mapping of 5′- or 3′-untranslated region with rapid amplification of cDNA ends
Total RNA was used to analyze 5′- and 3′untranslated regions (UTRs) by rapid amplification of cDNA ends (RACE) with the gene-specific primers and with the FirstChoice RLM RACE kit (Ambion), according to the manufacture’s specifications. The RACE products were electrophoretically separated on agarose gel, purified, and subsequently ligated into pJET1.2/Blunt vector (Fermentas Life Sciences) for further transformation into Escherichia coli 5-alpha competent cells supplied by New England BioLabs. Plasmids from individual transformants were analyzed for the presence of inserts with expected size and sequenced using BigDye Terminator v3.1 from Applied Biosystems.
Assay for the Sou phenotype
Spot dilution assay was performed on solid sorbose medium (see above), as described by Wellington and Rustchenko (2005).
Handling transformants carrying replicative plasmids for the growth assay and for RNA isolation
Several transformants grown as colonies on solid SD medium supplied with sorbitol were combined and streaked as patches on solid SD medium. After incubation, some cells were taken for sorbose growth assays, whereas the other cells were suspended in distilled water, plated on solid SD medium for independent colonies, incubated, colonies grown, harvested, and total RNA isolated (see above).
Amplicons or plasmid inserts were routinely sequenced in the core facility at the University of Rochester using BigDye Terminator v3.1 Cycle Sequencing kit on ABI 3730 PRISM Genetic Analyzer. Transformation of C. albicans cells was conducted according to Kabir and Rustchenko (2005). Site-directed mutagenesis was performed with a QuickChange Site-Directed Mutagenesis kit (Stratagene), according to the manufacturer’s recommendations.
Identification of the CSU53 gene in region 135 of Ch5
In this work, we continue characterizing five well-established regions on Ch5 that are involved in the repression of the growth on sorbose (see the Introduction and Figure 1). One of the approaches used here is the replicative test-plasmid pCA88 carrying metabolic SOU1. As previously reported (see also The co-overexpressing system in Materials and Methods), this plasmid confers growth on sorbose medium to the Sou− recipient cells, Sou− → Sou+ (Figure 2, diagrams 1 and 2). However, the introduction of the 4.3-kb portion of Ch5 that is designated region 135 (Figure 1) into pCA88 results in the plasmid pCA135 (Figure 2, diagram 3) that shifts the Sou+ phenotype of the recipient cells due to SOU1 back to no growth, Sou+ → Sou− (Kabir et al. 2005). Compare the repressive Sou– phenotype due to region 135 with the control Sou+ phenotype due to SOU1, as shown with the spot assay in Figure 3A (see Materials and Methods for the assay and Figure S1 for the control growth on glucose medium). Analysis of the sequence of region 135 indicated a single large ORF of 912 bp previously annotated, as the SFC1 (orf19.3931) gene. We demonstrated that this ORF is implicated in the repression of growth on sorbose, as presented below, and thus designated this gene as CSU53. In figure 4A showing the diagram of region 135, the CSU53 ORF is presented with a shaded box.
To evaluate whether CSU53 is relevant to the Sou phenotype, we interfered with the CSU53 translation product by creating three independent frameshift mutations in region 135 within the CSU53 ORF at positions +66, +112, or +233, as indicated by stars in Figure 4A (see supporting information File S1 for details). The mutations were verified by sequencing and each mutated region 135 was individually subcloned into pCA88 resulting in plasmids pEA227, pEA158, and pEA201, respectively. Note that in Figure 4A the plasmid names are given to the corresponding inserts, a nomenclature that will be kept throughout the text. CAF4-2 cells were individually transformed with the above plasmids and tested for the growth on sorbose with the spot assay. For the handling of C. albicans transformants, see Materials and Methods. Each mutation consistently abolished the repressive Sou− phenotype of region 135 and restored the control Sou+ phenotype due to SOU1, as presented in Figure 3B (compare pEA227, pEA158, and pEA201 with pCA135 and pCA88; also see Figure S2A for the control growths on glucose medium). These data strongly implicated the CSU53 (SFC1) gene with the repression of growth on sorbose and indicated that the putative Csu53p is important for the Sou phenotype.
To produce other evidence of the repressive function of CSU53, we interfered with its transcription. Portions of region 135 encompassing the CSU53 ORF with various sizes of the sequence in front of the start codon, were PCR amplified, individually co-overexpressed with SOU1, and tested for growth on sorbose, as above. We found that diminishing the sequence upstream to the ORF led to diminution of the repressive property. For example, an entire upstream sequence of ∼2 kb, pEA144, rendered the same repressive phenotype, as the entire region 135 (Figures 3A, bottom, and 4A; also Figure 2, diagram 4). Diminishing the upstream sequence to 1.4 kb, pEA105, or to 0.5 kb, pEA143 (Figure 4A) decreased the repression, as indicated by multiple Sou+ colonies, Sou−*, which appeared after prolonged incubation (Figure 3A, top or bottom, respectively). On the other hand, various portions of region 135 lacking the CSU53 ORF, as, for example, pEA145, pEA146, pEA141, pEA140, pEA130, pEA132, or pEA133 (Figure 4A) consistently displayed the control Sou+ phenotype (see Figure 3A, bottom, for representative pEA145 and pEA140), thus demonstrating no relevance to the repressive property. These results provided clear evidence for a single CSU53 in region 135.
Identification of new elements, ASUs, which are associated with CSUs, and which act opposite to CSUs by enhancing the growth on sorbose
By co-overexpressing SOU1 and different portions of region 135, as described above, we found a phenotype that was different from the repressive Sou− or the control Sou+. This phenotype was an increased growth on sorbose medium, Sou++, which occurred when the region upstream to CSU53 was removed, thus resulting in pEA104, pEA156, and pEA155 (Figure 4A), as exemplified with the growth assay of pEA104 in Figure 3A (top).
We addressed the question of whether the Sou++ phenotype depends on the insert orientation toward SOU1 by preparing plasmid pEA234, which carried a 2158-bp insert from plasmid pEA104 in opposite orientation (see File S1 and Table S1 for the pEA234 construction). As exemplified in Figure 3C, plasmids with different orientation of the insert, pEA104 and pEA234, rendered the same increase of growth, which was independent of insert orientation, in multiple experiments. See Figure S3A for the control growth on glucose medium.
We next addressed the question of whether the Sou++ phenotype depends on the large insert, acting, for example, as a stabilizing factor. We prepared and analyzed plasmid pEA261 carrying an insert of the same size, 2158 bp, as the above pEA104, encompassing the sequence upstream to CSU53 with an adjacent 169 bp extending to outside the region 135 (Figure 4A). Unlike pEA104, Sou++, new pEA261 conferred the control Sou+ phenotype.
It did not seem that the Sou++ phenotype was an artifact due to a certain sequence, as various shorter portions encompassing the region downstream to CSU53 alone or with a portion of the CSU53 ORF, such as pEA145, pEA146, pEA141, pEA140, and pEA130 (Figure 4A) showed the control Sou+ phenotype (see Figure 3A, bottom, for the representative pEA145 and pEA140).
Importantly, all three portions conferring the Sou++ phenotype, pEA155, pEA156, and pEA104 contained a relatively small ORF of 120 bp embedded in the CSU53 ORF of 912 bp in opposite orientation, 461 bp downstream from the start codon (Figure 4A, solid box). Apparently, there is a critical 20-bp region, which causes the difference between pEA155 (Sou++) and pEA142 (Sou+). This critical region lies in front of a putative transcription start site of the embedded ORF. We thus designated the ORF embedded in CSU53 of region 135, as ASU53 (activation of sorbose utilization).
Similar to CSU53 from region 135, we previously reported that CSU51 from region A, Figure 1, also has a smaller ORF of 105 bp embedded in opposite orientation, 32 bp downstream from the start codon (Kabir et al. 2005) (Figure 4B). Also similarly, the authors introduced frameshift mutations in region A to target either CSU51 or the embedded ORF and demonstrated that the embedded ORF lacked the repressive property. Importantly, when the repressive property of CSU51 was abolished, the phenotype shifted to Sou+, i.e., no Sou++ growth occurred. In this work, we changed the approach and analyzed portions of region A, instead of introducing mutations in the intact region. We found that the portions pEA219 of 792 bp, pEA221of 802 bp, and pEA209 of 404 bp encompassing CSU51 with the embedded ORF, but lacking the region upstream to the CSU51 ORF, rendered, as expected, the Sou++ phenotype (Figure 4B and Figure S3B). Increased growth was independent of the insert orientation, as shown with plasmid pEA236 carrying the same insert as in pEA209, but in opposite orientation (Figure S3B). We thus designated the ORF embedded in CSU51 of region A as ASU51, by analogy with ASU53 of region 135.
We confirmed the Sou++ phenotype of ASU51 by trimming the region upstream to its ORF to 20 bp, a portion pEA240 (Figure 4B) that shifted the phenotype Sou++ → Sou+. We also extended the region downstream to the ASU51 ORF from a portion of pEA209 (Sou++) into the region upstream to the CSU51 ORF, a portion of pEA183, which shifted the phenotype Sou++ → Sou+ (the phenotypes are indicated on the schematics in Figure 4B). The 44-bp sequence downstream to the ASU51 ORF on pEA209 or 75 bp on pEA183 included 11 bp or 44 bp, respectively, from the region upstream to CSU51. Sequence analysis of the critical 33-bp difference between pEA209 (Sou++) and pEA183 (Sou+) revealed a putative TATA box between position −12 to −15 in the 5′-UTR of CSU51, suggesting that this TATA box was sufficient for some transcriptional activity of CSU51 that prevailed over the weaker Sou++ phenotype.
Importantly, the Sou++ phenotype consistently occurred in multiple independent experiments, as well as in the series of experiments that were conducted at 37°, 30°, and 22°, as established with pEA104 or pEA234 carrying ASU53 or with pEA104 or pEA234 carrying ASU51. This phenotype could be clearly observed within approximately the first 3–5 days of incubation, until the control cells caught up with the growth, thus obscuring a convenient comparison. At 37°, the phenotype due to ASU51 was not as pronounced, as those at two other temperatures. Also, overall, the phenotype due to ASU51 was not as pronounced as the phenotype due to ASU53 (Figure S3).
In conclusion, we have found that each studied locus, CSU51 or CSU53, contains a CSU element and an ASU element, which is embedded in CSU in the opposite orientation. In the natural CSU/ASU configurations, the Sou− repressive phenotype of CSUs dominates the weaker Sou++ phenotype of the ASUs (Figure 2, diagram 3 or 4, Figure 3A, Figure S3, and Figure 4). The ASU element can be revealed by interfering with the transcription, but not the translation of the CSU element. A simple explanation would be that mutations destroying the Csu proteins leave the corresponding sense RNAs nearly intact. However, the elimination of the upstream regions, hence promoters, interferes with the production of the sense CSU transcripts, which subsequently affects translation products. Such phenotypic differences are expected if sense CSU and antisense ASU transcripts interact, for example, by forming dsRNA molecules that subsequently either inhibit translation or lead to degradation. Then, the lack of the CSU transcript from the CSU/ASU configuration from the plasmid would lead to the increased abundance of the ASU transcript from the plasmid and ultimately would lead to the increase of combined amount of plasmid and chromosomal ASU transcripts. This would increase the interaction with the chromosomal CSU transcripts and cause more depletion of CSU inhibitory transcripts, thus upregulating SOU1, and finally resulting in the better growth.
Visualization of sense CSU and antisense ASU transcripts with Northern blots
We determined CSU and ASU transcripts from chromosomal CSU/ASU configurations in the strains SC5314, 3153A, and CAF4-2 with Northern blots from three or four independent cultures of each strain (Materials and Methods), as exemplified by SC5314 and 3153A in Figure 5A. Highly abundant CSU51 sense transcript was revealed with total RNA and dsDNA probe prepared from CSU51. However, the low-abundance complementary ASU51 antisense transcript could be revealed only with mRNA and a riboprobe prepared from ASU51 (Figure 5B). The low-abundance sense CSU53 and antisense ASU53 were also revealed with mRNA and riboprobes prepared from the corresponding genes (Figure 5B). Each probe repeatedly produced hybridization signal(s) of the pattern, which was identical in all three examined strains. Specifically, the CSU51 probe revealed more abundant transcript of ∼700 nt and less abundant transcript of ∼1200 nt. The ASU51 probe revealed a pool of transcripts that were distributed around a 1000-nt size. A CSU53/ASU53 pair produced one sense and one antisense transcript of ∼1300 nt and 1200 nt, respectively.
Relative amounts of antisense ASU and sense CSU transcripts
We estimated comparative amounts of sense CSU and antisense ASU transcripts from chromosomal CSU/ASU configurations in the strains SC5314, 3153A, and CAF4-2 using sqRT–PCR analysis with gene-specific primers (Materials and Methods). The transcript levels of ASU51, as compared to CSU51, were 1.5% in SC5314, 4% in CAF4-2, and 3.8 and 4% in 3153A in two independent experiments, thus ranging from 1.5 to 4%. Also, the transcript levels of ASU53, as compared to CSU53, were 16% in SC5314, 8% in CAF4-2, and 37 and 44% in 3153A in two independent experiments, thus ranging from 8 to 44%.
The RT–PCR data are in agreement with the Northern blot analyses that indicated the lower abundance of the antisense transcript for at least CSU51/ASU51 configuration (see above and Figure 5A). Furthermore, the low amounts of the ASU transcripts are consistent with their phenotype being recessive in respect to the phenotype of CSUs (see above and Figure 3A). It should be pointed out that sense and antisense transcripts were analyzed using the same RNA preparations, implying that sense and antisense transcripts may exist in the same cells.
Mapping UTRs of sense and antisense transcripts
To clarify how sense and antisense transcripts overlap, we mapped UTRs of each CSU and ASU transcript using the RACE technique with a total of 3–6 clones from each of two different strains, as described in Materials and Methods. This approach, although designed to recognize 5′ cap or 3′ polyadenylated [poly(A)] structures, fails, however, to determine the sizes of poly(A) tails.
We found that CSU/ASU loci in both of the examined strains 3153A and CAF4-2 produced multiple sense or antisense 5′- and 3′-UTRs. The sizes of the largest CSU51, CSU53, and ASU53 transcripts, 945 nt, 1318 nt, and 1003 nt, respectively, that were calculated from the largest 5′- and 3′-UTRs, as well as the corresponding ORF, but not the poly(A) (see above), never exceeded, but approximately corresponded to the transcript sizes on the Northern blots in Figure 5A. A large difference occurred with the ASU51 transcript: 395 nt by RACE vs. ∼1000 nt by Northern blot. This difference could be, for example, due to a large poly(A) tail. In fact, large poly(A) tails of up to 365 nt, 400 nt, or 650 nt have been reported in eukaryotes (Carrazana et al. 1988; Salles and Strickland 1995).
Mutational analysis demonstrates that ASU ORFs are not critical for the Sou++ phenotype
We addressed the question of whether the ASU ORF is necessary for the Sou phenotype by introducing mutations in the ORFs of ASUs (Figure 2, diagram 7). We used a 1.2-kb portion, pEA104, from region 135 that carries ASU53 and that confers the Sou++ phenotype (Figures 3, A and C and 4A) to create two mutations: a frameshift mutation at positions +4 and +7 of the ASU53 ORF, pEA232, and a stop codon at position +12, pEA205, as indicated by stars in Figure 4A. Also, we used a 400-bp portion, pEA209, from region A that carries ASU51 (Figure 4B), which confers the Sou++ phenotype (Figure S3B) to create a frameshift mutation at position +39 of the ASU51 ORF, pEA243, as indicated by a star in Figure 4B. (See Table S1 and File S1 for the description of mutations and plasmids). We found no difference in the Sou++ growth of the cells carrying plasmids with intact or mutated ASUs, as exemplified in Figure 3D with the ASU53 mutations carried on pEA232 and pEA205. See Figure S2B for the control growth on glucose medium.
ASUs interact with CSUs at the phenotypic and the transcriptional levels
We directly addressed the question of interaction between the corresponding CSU and ASU elements at the phenotypic level. SOU1, a natural CSU/ASU configuration (Sou−), and an extra ASU (Sou++); were co-overexpressed on a replicative plasmid (Figure 2, diagrams 8 and 9), the CSU/ASU ratio, thus resulting in 1:2. The plasmid pEA238 included CSU51/ASU51 and ASU51 carried, respectively, on region A and the portion pEA209 from region A (Figure 4B). The plasmid pEA162 included CSU53/ASU53 and ASU53 carried, respectively, on the portions pEA105 and pEA104 of region 135 (Figure 4A). (See Table S1 and File S1 for plasmid preparations. See Figure 3, Figure S1, and Figure S3 for the spot assays on plates of the individual elements).
The plasmids with extra ASUs, pEA238 and pEA162, were tested on sorbose medium in multiple spot assays, as described above. We found that these plasmids substantially diminished the dominating Sou− repressive phenotype of corresponding CSU elements. As exemplified in Figure 3E with the plasmid pEA162, an extra ASU53 resulted in the growth, which was almost indistinguishable from the control growth, Sou+. Consistently, an extra ASU51 resulted in either intermediate growth between no growth and control growth, i.e., neither Sou− nor Sou+ occurred, or high-frequency large Sou+ colonies (Figure S4A). We emphasize that the Sou+ colonies never occurred, when region A, i.e., CSU51/ASU51, was co-overexpressed alone with SOU1. This variability presumably occurred due to the highly expressed and thus strong CSU51 (see above). Also, see Figure S4, A and B for the control growths on glucose medium.
To substantiate the finding of the interaction between ASUs and CSUs, we integrated one or two copies of ASU53, vectors pEA249 or pEA254, respectively, in the LEU2 locus on Ch7 in the strain CAF4-2. The control integration was with no ASU53, pAK156 (Table S1). We have previously demonstrated that integration in LEU2 did not interfere with the Sou phenotype (Wang et al. 2004). The proper integration of pAK156, pEA249, and pEA254 was verified by PCR amplifications with a pair of primers, one for the LEU2 gene, AF237U, and another for the lacZ N-terminal sequence of integration vector, M-13FU (Table S2). All three amplicons had the expected sizes of ∼2.0 kb, 3.8 kb, and 6.2 kb, respectively, while no amplicon was produced with genomic DNA from the control strain, SC5314. The presence of ASU53 in corresponding amplicons was confirmed with primers AF110 and AF111.
We prepared three batches of total RNA from three independent cultures of each construct. We then synthesized cDNA and used it for PCR amplifications with gene specific primers (Table S2). Reactions were set in duplex with control genes and were conducted with different numbers of cycles followed by quantitation (Materials and Methods). See Figure S5 for the examples of amplicons that were quantitated. As expected, the amount of the ASU53 transcript increased, consistent with the increase of the copy number of ASU53, as compared to the control strain with two regular chromosome copies carrying the integration vector with no extra copies of ASU53 (Table 1). Introduction of one extra copy of ASU53 increased the ASU53 transcript to 1.4, 1.5, and 1.6, resulting in an average of 1.5 ± 0.1; introduction of two extra copies of ASU53 increased the ASU53 transcript to 1.7, 1.9, and 1.9, resulting in an average of 1.8 ± 0.1. We then asked whether the amount of the CSU53 transcript changed. We found a copy-dependent inhibition of CSU53 by ASU53. In the presence of one extra copy, the CSU53 transcript was 0.6, 0.7, and 0.8, resulting in an average of 0.7 ± 0.1, whereas in the presence of two extra copies, the CSU53 transcript was 0.5, 0.5, and 0.6, resulting in an average of 0.5 ± 0.1 (Table 1).
The effect of chromosomal extra copies of ASU51 on the expression of the corresponding CSU51 was not analyzed, because of a large difference in the expression of the two genes.
SOU1 is upregulated by ASU53 and downregulated by CSU51 or CSU53
We next determined the effect of genomic extra ASUs on the metabolic SOU1 expression. RT–PCR amplifications were conducted and analyzed as above. In the presence of one or two extra copies of ASU53, the amount of the SOU1 transcript increased in a copy-dependent fashion (Table 1). In the presence of one extra copy, the SOU1 transcript was 1.6, 1.8, and 1.9, resulting in an average of 1.8 ± 0.2, whereas in the presence of two extra copies, the SOU1 transcript was 2.6, 2.6, and 2.7, resulting in an average of 2.6 ± 0.1.
We next used a co-overexpression system to address whether CSU51 or CSU53 controls SOU1 expression. Northern blot analyses were carried out with independently prepared batches of total RNA that was extracted from CAF4-2 cells transformed with the replicative plasmids pEA105 or pCA135 (Figure 4A) co-overexpressing SOU1 with CSU53, as well as with the plasmid pAK65 (Figure 4B), co-overexpressing SOU1 with CSU51 (see also Figure 2, diagrams 3 and 4). Note that CSUs are represented by the natural configurations CSU/ASU, because CSUs cannot be separated from ASUs. However, in these natural configurations, CSUs dominate ASUs (see above). As exemplified in Figure 6, the co-overexpression with each CSU gene clearly diminished the amount of the SOU1 transcript, as compared to the control pCA88, which overexpressed SOU1, but lacked CSU. Specifically, SOU1 was downregulated 0.55-, 0.63-, 0.33-, and 0.36-fold, resulting in an average of 0.47 ± 0.15 by CSU53 in four independent experiments, as well as 0.18- and 0.29-fold by CSU51 in two independent experiments. As expected from the phenotypes and transcript abundances (see above), CSU51 repressed SOU1 more strongly than CSU53.
We found that the previously identified CSU51 and currently identified CSU53 each contain a genetic element, ASU, which is embedded in CSU in the opposite orientation. ASUs are manifested in several ways. They produce antisense transcripts that are, however, significantly less abundant than the corresponding CSU sense transcripts. Extra copies of ASU in the genome lead to an decrease or increase of the transcript of the regulatory CSU or the metabolic SOU1, respectively. ASUs are also manifested at the phenotypic level. A copy of ASU overexpressed from the low copy number plasmid, possesses a distinct, albeit weak phenotype, enhancing the growth on sorbose. This is, presumably due to a combined action of the plasmid and chromosomal copies of ASU on the chromosomal copies of CSU. Consistently, a copy of ASU co-overexpressed from plasmid with the corresponding natural CSU/ASU configuration greatly diminishes the CSU repressive phenotype.
The interaction between CSUs and ASUs is thus clearly revealed at the transcription and at the phenotypic levels. It is also clear that the levels of the CSU and ASU transcripts are in a reverse relationship. Conversely, the expression of metabolic SOU1 on Ch4 directly depends on the transcription of ASUs and reversely depends on the transcription of negative regulatory CSUs. It is reasonable to suggest that ASU elements act as modulators repressing CSUs, because the phenotype of CSUs dominates and because the ASU transcripts are less abundant and are completely imbedded in the sense transcripts. Although not specifically addressed, both CSU sense and ASU antisense transcripts are expected to be produced in the same cell, as this is required for their interaction. The complementarity of sense/antisense elements is also a strong indication of their interaction. As discussed in Results, we obtained preliminary evidence that, as expected for the complementary sequences, sense and antisense RNA interact. This is because ASU phenotype is manifested when the sequence upstream to the CSU ORF is abrogated, thus greatly interfering with the CSU transcription, but not when the Csu protein is mutated, leaving the CSU transcription normal.
All sense CSU and antisense ASU transcripts contain ORFs and are also capped and polyadenylated, which is indicative of translation. However, while both CSU ORFs are essential for the Sou− repressive function, ASU ORFs are not essential for the Sou++ enhancing function, as clearly demonstrated by mutational analysis. This could be interpreted, as Asu proteins are not produced or, alternatively, are not relevant for the Sou phenotype. In this respect, noncoding RNA can be capped and polyadenylated; furthermore, some RNAs can function both as mRNA and as noncoding RNA (Callahan and Butler 2008; Dinger et al. 2008; Rapicavoli and Blackshaw 2009). Future studies should establish whether proteins are translated from ASUs and, if so, how they function. Currently, we consider that ASU transcripts are implicated with the Sou phenotype as noncoding RNAs.
Our current model proposes that interactions between three kinds of elements CSU, ASU, and SOU1 are based on the transcript ratios. ASUs repress CSUs that, in turn, repress SOU1. Because the amount of CSU transcript diminishes, when corresponding ASU is overexpressed, a plausible scenario is that the complementary CSU and ASU RNAs form dsRNA molecules that are subsequently degraded.
In summary, identification of an additional CSU53 confirms the previous evidence of multiple CSUs on Ch5. The inverse relationship between CSU51 or CSU53 transcripts on one hand and SOU1 transcript on the other hand is consistent with the previous proposal that CSUs are negative regulators of SOU1 (Kabir et al. 2005) that determine the regulatory role of Ch5 copy number. Discovery of ASU positive elements that counteract CSU negative elements reveals an unanticipated layer of complexity in the negative regulation of growth on sorbose.
We thank Fred Sherman, Scott Butler, and Yi-Tao Yu for the inspiring discussions, as well as for the critical reading of the manuscript. We thank M. Anaul kabir for the plasmid pAK156. This work was supported in part by National Institutes of Health grant GM12702. We are also grateful to the University of Rochester funds that enabled this study.
Supporting information is available online at http://www.genetics.org/content/suppl/2011/12/01/genetics.111.136267.DC1.
Communicating editor: M. Hampsey
- Received June 21, 2011.
- Accepted November 16, 2011.
- Copyright © 2012 by the Genetics Society of America