The WOR1 5' untranslated region regulates white-opaque switching in Candida albicans by reducing translational efficiency.

The human fungal pathogen Candida albicans undergoes white-opaque phenotypic switching, which enhances its adaptation to host niches. Switching is controlled by a transcriptional regulatory network of interlocking feedback loops acting on the transcription of WOR1, the master regulator of white-opaque switching, but regulation of the network on the translational level is not yet explored. Here, we show that the long 5' untranslated region of WOR1 regulates the white-opaque phenotype. Deletion of the WOR1 5' UTR promotes white-to-opaque switching and stabilizes the opaque state. The WOR1 5' UTR reduces translational efficiency and the association of the transcript with polysomes. Reduced polysome association was observed for additional key regulators of cell fate and morphology with long 5' UTR as well. Overall, we find a novel regulatory step of white-opaque switching at the translational level. This translational regulation is implicated for many key regulators of cell fate and morphology in C. albicans.

Introduction

The diploid yeast Candida albicans is a major fungal pathogen of humans. C. albicans exists as a commensal in healthy individuals but can cause localized infection of mucosal membranes as well as systemic infection in immunocompromised patients. The phenotypic plasticity of C. albicans contributes to its success as a commensal and pathogen. In addition to the yeast‐hyphal transition, C. albicans also undergoes switching between two epigenetically heritable phenotypic states, white and opaque (Slutsky et al., 1987). Switching is stochastic and reversible and occurs approximately every 104 generations (Rikkerink et al., 1988). White‐to‐opaque switching can be induced by environmental conditions, such as genotoxic stresses, N‐acetyl‐glucosamine (GlcNAc) as a carbon source and high CO2 concentration (Ramirez‐Zavala et al., 2008; Alby and Bennett, 2009; Huang et al., 2010). Growth at 37oC converts opaque cells to white (Bergen et al., 1990). White and opaque cells exhibit distinct gene expression profiles, with 1306 genes transcribed differentially between the two states (Tuch et al., 2010). In addition, white and opaque cells adapt to host defenses differently, with white cells more capable of causing infections in the bloodstream, whereas opaque cells are more efficient in colonizing skin, and in escaping macrophage detection (Kvaal et al., 1999; Lachke et al., 2003; Lohse and Johnson, 2008). Opaque cells are mating‐competent, and MTLa or MTLα opaque cells mate with 106 greater efficiency compared with white cells. In MTLa/α cells, the a1‐α2 heterodimer represses opaque formation (Miller and Johnson, 2002). Although the majority of natural C. albicans isolates are MTLa/α, a recent study has shown that passage through the mammalian gut promotes a phenotypic switch of MTLa/α cells to opaque‐like cells that are more fit for GI colonization and commensalism (Pande et al., 2013). In addition, MTLa/α cells are capable of opaque formation under high CO2 and with GlcNAc as a carbon source, conditions that mimic the host environment (Xie et al., 2013). These findings support the notion that white‐opaque switching can occur in a majority of C. albicans strains and is highly relevant to Candida–host interactions.

The master regulator of white‐opaque switching is WOR1, which is highly expressed in opaque cells and required for switching to opaque (Huang et al., 2006; Srikantha et al., 2006; Zordan et al., 2006). Wor1 protein contains a novel DNA‐binding domain (Lohse et al., 2010; 2014; Zhang et al., 2014) and promotes its own expression by binding to the WOR1 promoter up to 8 kb upstream of its transcription start site (Zordan et al., 2007). White‐opaque switching and WOR1 transcription are regulated by a circuit of interlocking transcriptional feedback loops, consisting of regulators WOR1, EFG1, CZF1, WOR2, WOR3 and AHR1 (Downs et al., 2004; Levchenko and Jackson, 2004; Zordan et al., 2007; Hernday et al., 2013; Lohse et al., 2013). Notably, most of the regulatory genes in this circuit have a long 5′ untranslated region (5′ UTR): for example, the WOR1 5′ UTR is 1997 bp, EFG1 1139 bp, CZF1 1662 bp (Srikantha et al., 2006; Bruno et al., 2010; Tuch et al., 2010). In contrast, 5′ UTRs of other C. albicans genes that are not in the regulatory circuits for cell fate or yeast‐hyphal regulation are mostly under 100 bp in length. Most yeasts, such as Saccharomyces cerevisiae, do not have genes with long 5′ UTRs (Nagalakshmi et al., 2008; Bruno et al., 2010). Although the transcriptional regulation of WOR1 and white‐opaque switching has been extensively studied, functions of and regulation by long 5′ UTRs remain to be explored.

Although long 5′ UTRs are rare in yeasts, they are common in higher eukaryotes and in viral genes and are frequently linked to translational regulation, in particular translational repression (Pickering and Willis, 2005). A common mechanism for translational repression at the 5′ UTR is through RNA‐binding proteins. For example, a developmentally regulated translational control at 5′ UTR by a meiosis‐specific RNA‐binding protein is critical for establishing the meiotic chromosome segregation pattern in S. cerevisiae (Berchowitz et al., 2013). Additional elements that have been shown to control translational efficiency at 5′ UTRs include upstream open reading frames (uORFs) and internal ribosome entry sites (IRESs) (Morris and Geballe, 2000; Hellen and Sarnow, 2001). In S. cerevisiae, cap‐independent translation at IRESs in the 5′ UTR of a small subset of genes is required for invasive growth (Gilbert et al., 2007). A recent study has shown that C. albicans GCN4 is translationally regulated by a uORF (Sundaram and Grant, 2014b). The length and structure of the 5′ UTR also affect microRNA‐mediated translational repression (Meijer et al., 2013). Finally, secondary structures in the 5′ UTR can inhibit translation by stalling translational initiation (Jackson et al., 1996; Jackson and Gorovsky, 2000). In C. albicans, translational inhibition by a long 5′ UTR has been observed for the hyphal‐specific transcriptional regulator UME6 (Childers et al., 2014). The long 5′ UTR of WOR1 makes it a promising candidate as a cis‐regulatory element of WOR1 translation, and therefore of white‐opaque switching.

In this study, we find that the WOR1 5′ UTR regulates the white‐opaque phenotype by reducing translational efficiency of WOR1. We demonstrate that the 5′ UTR is repressive toward both opaque formation and stability. Deletion of the 5′ UTR greatly increases white‐to‐opaque switching while reducing opaque‐to‐white switching. We further show that translational repression at long 5′ UTRs is pervasive and is observed for several genes in C. albicans. As these genes are pivotal to yeast‐hyphal and/or white‐opaque transitions, we speculate that translational regulation at their 5′ UTR may also play an important role in cell fate commitment and maintenance.

Results

Deletion of the 1 5′ UTR enhances white‐opaque switching and opaque stability

In order to examine the role of the WOR1 5′ UTR in white‐opaque switching, we constructed strains in which the 5′ UTR was deleted, as shown in Fig. 1A and described in detail in Experimental procedures. In brief, the WOR1p‐Δ5'WOR1‐HA plasmid was constructed, in which the 5′ UTR sequence was removed by placing a 3 kb fragment of the WOR1 promoter directly upstream of the WOR1 coding sequence. A 5′ UTR‐WOR1/Δ5′‐WOR1 strain was generated by integrating the WOR1p‐Δ5'WOR1‐HA at the WOR1 promoter upstream of the wor1Δ locus of a WOR1 heterozygous deletion mutant (5′ UTR‐WOR1/wor1). Δ5′‐WOR1/Δ5′‐WOR1 and Δ5′‐WOR1/wor1 strains were generated by transforming the WOR1p‐Δ5'WOR1‐HA into a wor1Δ/wor1Δ strain once and twice respectively (see Experimental procedures). In all three strains, the Δ5′ copy of WOR1 was 3′‐tagged with HA to facilitate additional assays. These Δ5′‐WOR1 strains were compared with 5′ UTR‐WOR1 strains for white‐opaque switching frequency and opaque phase stability. White cells were grown on synthetic complete dextrose (SCD) plates at room temperature to assay spontaneous switching to opaque. After 7 days, all strains carrying Δ5′‐WOR1 displayed increased white‐to‐opaque switching, compared with corresponding strains carrying the same copy number of 5′ UTR‐WOR1 (Fig. 1B). The switching rate of 5′ UTR‐WOR1/Δ5′‐WOR1 approached 100%, in contrast to 2% for wild‐type cells of 5′ UTR‐WOR1/5′ UTR‐WOR1 strain. Switching of 5′ UTR‐WOR1/Δ5′‐WOR1 occurred as multiple opaque sectors per white colony, with entire colonies becoming opaque by 7 days. Intriguingly, the Δ5′‐WOR1/Δ5′‐WOR1 strain had a 35.7% white‐opaque switching rate, still substantially elevated from wild type but lower than 5′ UTR‐WOR1/Δ5′‐WOR1. Notably, the highest switching rate appeared to be conferred by heterozygosity for the WOR1 5′ UTR, suggesting both a positive and negative role for the 5′ UTR in regulation of switching. For cells carrying only one copy of WOR1, the absence of the 5′ UTR still conferred a higher switching rate. Compared with 5′ UTR‐WOR1/wor1, which had a switching rate of only 1%, Δ5′‐WOR1/wor1 was able to switch to opaque at 12%. However, this was much lower than the Δ5′‐WOR1/Δ5′‐WOR1 strain carrying two copies of WOR1 (Fig 1B). The high white‐opaque switching frequency raised the question of whether opaque state itself is more stably maintained in Δ5′‐WOR1 strains. Stability was assayed by incubating opaque cells of control and Δ5′‐WOR1 strains on solid media at room temperature or 37°C. We find that the presence of Δ5′‐WOR1 enhances opaque stability and reduces opaque‐to‐white switching (Fig. 1C). 5′ UTR‐WOR1/Δ5′‐WOR1 and Δ5′‐WOR1/Δ5′‐WOR1 showed less spontaneous opaque‐white switching at room temperature than wild type, whereas Δ5′‐WOR1/wor1 also switched to white less frequently than 5′ UTR‐WOR1/wor1. At 37°C, where opaque wild‐type and 5′ UTR‐WOR1/wor1 cells switched en masse to white, nearly all 5′ UTR‐WOR1/Δ5′‐WOR1 cells as well as 65.7% of Δ5′‐WOR1/Δ5′‐WOR1 cells remained opaque after 7 days. However, all Δ5′‐WOR1/wor1 cells switched to white (Fig 1C). The highly opaque‐switching and opaque‐stable phenotype of cells carrying Δ5′‐WOR1 appears to be enhanced by increased copy number of WOR1, and in particular by having both 5′ UTR‐WOR1 and Δ5′‐WOR1. As Δ5′‐WOR1 is HA‐tagged, we needed to exclude the possibility that the HA tag produced this phenotype by affecting the stability/activity of Wor1. We compared the white‐to‐opaque switching rate of wild type strains with and without HA‐tagged WOR1 and found the presence of the HA tag on WOR1 did not increase switching (Fig. S1, lanes 1 and 2). Having an additional untagged Δ5′‐WOR1 increased switching (lanes 2 and 3) to levels similar to HA tagged Δ5′‐WOR1 (lanes 3 and 4).

Figure 1

Deletion of the 5′ UTR of 1 promotes white‐opaque switching and opaque stability.

A. Schematics of WOR1 loci of the Δ5′‐WOR1 strains constructed for this study: 5′ UTR‐WOR1/Δ5′‐WOR1, Δ5′‐WOR1/Δ5′‐WOR1 and Δ5′‐WOR1/wor1.

B. Percentage of spontaneous white‐opaque switching in the following strains; 1: 5′ UTR‐WOR1/5′ UTR‐WOR1 (wild type; JYC1), 2: 5′ UTR‐WOR1/Δ5′‐WOR1 (HLY4212), 3: Δ5′‐WOR1/ Δ5′‐WOR1 (HLY4214), 4: 5′ UTR‐WOR1/wor1(HLY3903) and 5: Δ5′‐WOR1/wor1 (HLY4213). White cells grown at 30°C were plated to SCD media and grown at room temperature (25°C) for 7 days before colony phenotype was scored.

C. Opaque stability of the strains described in (B). Opaque cells grown at 25°C were plated to SCD media and grown at 25°C and 37°C for 7 days. Stability of opaque state was measured as percentage of colonies that stay in opaque.

Deletion of the 1 5′ UTR enhances 1 positive feedback at 37o C

As WOR1 transcription reflects the white‐opaque phenotypic state of the cell, WOR1 expression in strains carrying Δ5′‐WOR1 could illuminate the processes behind the higher opaque stability of Δ5′‐WOR1 strains. We examined the effects of the 5′ UTR on WOR1 transcript levels under conditions where wild‐type opaque cells switch to white, but Δ5′‐WOR1 opaque cells remain mostly in opaque. To this end, we incubated at 37°C opaque cells of two strains that carry only one copy of WOR1: 5′ UTR‐WOR1/wor1 and Δ5′‐WOR1/wor1. Northern blotting probing for WOR1 was performed to compare transcript level in these two strains, sampled every 4 h (Fig. 2A). Cells from each time point were plated and grown on YPD plates at room temperature to determine percent of opaque cells. Prior to transfer to 37°C, WOR1 transcript was more abundant in opaque Δ5′‐WOR1/wor1 cells than in 5′ UTR‐WOR1/wor1. After cells were incubated at 37°C, WOR1 transcript disappeared from the 5′ UTR‐WOR1/wor1 strain immediately but persisted for longer in the Δ5′‐WOR1/wor1 strain, with some transcript detectable 12 h after temperature shift (Fig. 2A). The opaque stability of cells removed from 37°C during the experiment was also higher for Δ5′‐WOR1/wor1 (Fig. 2A). The elevated level of Δ5′‐WOR1 transcript even at high temperature, assuming transcription levels are similar, presents the possibility of either heightened transcript stability or increased translation efficiency from the Δ5′‐WOR1, which in turn promotes WOR1 transcription through positive feedback of Wor1. To further examine the effect of the Δ5′‐WOR1 on WOR1 positive feedback, we conducted Northern blotting under the same conditions on strains carrying two copies of WOR1: 5′ UTR‐WOR1/5′ UTR‐WOR1 and 5′ UTR‐WOR1/Δ5′‐WOR1 (Fig. 2B). At room temperature, 5′ UTR‐WOR1 transcript levels were similar in opaque cells of both strains. In the 5′ UTR‐WOR1/Δ5′‐WOR1 strain, Δ5′‐WOR1 was present at a lower level than 5′ UTR‐WOR1. This result suggested that the 5′ UTR of WOR1 does not reduce its transcript level and in fact is correlated with more abundant transcript. Upon shifting to 37°C, WOR1 transcript rapidly disappears in wild‐type 5′ UTR‐WOR1/5′ UTR‐WOR1 cells, whereas both 5′ UTR‐WOR1 and Δ5′‐WOR1 persist strongly for 8 or more hours in 5′ UTR‐WOR1/Δ5′‐WOR1 and are still detectable at 12 h (Fig. 2B). Consistent with persistent WOR1 transcription, a high percentage of cells remained opaque after treatment at 37°C compared with the wild‐type control. We suggest that the stronger and more persistent expression of WOR1 in the 5′ UTR‐WOR1/Δ5′‐WOR1 strain is due to a stronger positive feedback from Wor1 protein, which is likely translated more efficiently from Δ5′‐WOR1 than from 5′UTR‐WOR1. The transcription from the Δ5′‐WOR1 construct likely initiated close to the promoter upstream of the annotated start codon, as the transcript from this construct was stable and efficiently translated into a functional Wor1 (Arribere and Gilbert, 2013). Transcriptional initiation at a downstream AUG will render the product non‐functional as the next AUG is downstream of a significant portion of the DNA binding domain of Wor1.

Figure 2

Stabilization of 1 expression at 37°C in Δ5′‐ 1 strains.

Northern blot probing for a 500 bp region of the WOR1 coding sequence against total RNA of A. initially opaque 5′ UTR‐WOR1/wor1 and Δ5′‐WOR1/wor1 cells, or B. initially opaque wild type (5′ UTR‐WOR1/5′ UTR‐WOR1) and 5′ UTR‐WOR1/ Δ5′‐WOR1 cells. Stably opaque cells were grown to log phase at 25°C, then transferred to 37°C. RNA was extracted every 4 h for Northern blotting, and white‐opaque phenotype was assessed by removing cells to SCD plates, incubating at 25°C, then scoring colony phenotype after 4 days.

5′ UTR‐ 1 transcripts are enriched in monosome

We suspected that 5′ UTR‐WOR1 is less efficiently translated than Δ5′‐WOR1. Transcripts that are actively translated in the cell are associated with multiple ribosomes (polysomes), whereas those with stalled or less efficient translation are associated with ribosomal subunits or single 80S ribosomes (monosomes). Polysomes can be isolated by sedimentation of cell lysates (McQuillen et al., 1959; Warner et al., 1963). To determine the effect of the WOR1 5′ UTR on translational efficiency, we conducted a polysome gradient assay on the 5′ UTR‐WOR1/Δ5′‐WOR1 strain, using opaque cells in which WOR1 is expressed. Whole cell extract was separated by centrifugation on a sucrose gradient into fractions enriched for monosomes or polysomes, and absorbance profile was taken to visualize polysomal peaks (Fig. 3A). Ethylenediaminetetraacetic acid (EDTA) dissociated polysomes as expected (Fig. 3A). Monosomal and polysomal fractions were identified both based on absorbance profile and levels of 40S and 60S rRNA in each fraction as determined by reverse transcription quantitative PCR (RT‐qPCR) on RNA extracted from each fraction (Fig. 3B). RT‐qPCR was then used to determine the level of monosomal and polysomal association of 5′ UTR‐WOR1 and Δ5′‐WOR1 transcripts. As Δ5′‐WOR1 is tagged at the 3′ end with ‐HA in this strain, primers amplifying a region extending from the 3′ end of the WOR1 coding sequence to the ‐HA tag were used for detection of Δ5′‐WOR1 transcript. Primers amplifying a region of the WOR1 5′ UTR were used for detection of 5′ UTR‐WOR1 transcript. We found the distribution of 5′ UTR‐WOR1 to be highest in the monosome fractions and present at lower levels in the polysomal fractions. Conversely, Δ5′‐WOR1 transcript level showed higher association with polysomes, similar to ACT1 (Fig. 3C). These results suggest less efficient translation of WOR1 transcripts bearing the 5′ UTR, whereas Δ5′‐WOR1 transcripts are bound by multiple ribosomes and actively translated like ACT1.

Figure 3

5′ UTR‐ 1 transcripts are enriched in monosomes, whereas Δ5′‐ 1 transcripts are enriched in polysomes.

A. Polysome profile of 5′ UTR‐WOR1/Δ5′‐WOR1 opaque cells grown to log phase in YPD at room temperature (25°C). Cell lysate was applied to a sucrose gradient and centrifuged to separate monosomal and polysomal fractions. Fractions were collected in order of least to greatest density on an ISCO gradient fractionator, and absorbance profile taken at A254 nm. Absorbance profile of lysate applied to a gradient containing 20 mM EDTA is shown as negative control.

B. Relative abundance of 40S and 60S ribosomal RNA in each fraction.

C. Relative abundance of 5′ UTR‐WOR1, Δ5′‐WOR1‐HA, and ACT1 transcripts in each gradient fraction as determined by RT‐qPCR. All experiments were performed in triplicate. For each transcript, values were normalized to the fraction containing the highest value, set as 1. Error bars represent standard deviation.

The 1 5′ UTR reduces translational efficiency

Having established that transcripts containing the WOR1 5′ UTR are enriched in monosomes and less translated, we next sought to demonstrate that less Wor1 protein is made from the 5′ UTR‐WOR1 transcript than from WOR1 transcript without the 5′ UTR. To this end, the WOR1 gene with or without its 5′ UTR was tagged with –HA and placed under the maltose‐inducible MAL2 promoter, resulting in the MAL2p‐5'UTR‐WOR1‐HA or MAL2p‐WOR1‐HA plasmid. This allows for conditional expression without positive feedback of Wor1 onto its own promoter. Cells of WT + MAL2p‐5'UTR‐WOR1‐HA or WT + MAL2p‐WOR1‐HA were transferred from dextrose to maltose to turn on expression from the MAL2 promoter, upon which WOR1 mRNA levels were assayed by RT‐qPCR, and HA‐tagged Wor1 protein by Western blotting (Fig. 4). At the transcript level, little difference was observed between 5′ UTR‐WOR1 or WOR1, indicating that the 5′ UTR did not function to reduce transcription. As the MAL2 promoter is more active at higher temperature, transcript levels were elevated at 37°C, but again this was observed for both the 5′ UTR‐WOR1 and WOR1 constructs (Fig. 4A). At the protein level, however, much higher levels of Wor1 were detected from MAL2p‐WOR1‐HA than MAL2p‐5'UTR‐WOR1‐HA, pointing to significantly reduced translation when the 5′ UTR is present (Fig. 4B).

Figure 4

2‐driven 5′ UTR‐ 1 produces less Wor1 protein than 2‐driven 1.

A. RT‐qPCR of WOR1 and B. Western blot detecting HA‐tagged Wor1, in WT carrying MAL2p‐5'UTR‐WOR1‐HA (HLY4215) or MAL2p‐WOR1‐HA (HLY3569). White cells were grown to log phase in YEP + dextrose, then inoculated to YEP + maltose to induce transcription from the MAL2 promoter for 4 h at RT (25°C) or 37°C. For qPCR, all experiments were performed in triplicate, and values were normalized to ACT1 signal.

To determine whether the WOR1 5′ UTR was sufficient for the observed translational inhibition effect and the WOR1 coding region did not contribute, we cloned (WOR1) 5′ UTR‐GFP under the MAL2 promoter to produce the MAL2p‐5'UTR‐GFP plasmid. A MAL2p‐myc‐GFP plasmid was used as a 5′ UTR‐less control. White and opaque wild‐type cells transformed with either plasmid were again transferred from dextrose to maltose to turn on expression from the MAL2 promoter. Transcript levels of GFP with and without the WOR1 5′ UTR were similar in maltose, as shown by qPCR (Fig. 5A). The MAL2 promoter is not entirely shut off in dextrose in the MAL2p‐5'UTR‐GFP strain; however, this appears to be unrelated to the WOR1 5′ UTR as it was also observed for both the MAL2p‐5'UTR‐WOR1‐HA and MAL2p‐WOR1‐HA strains (Fig. S2). In white cells, Western blots probing for GFP showed that higher protein levels were reached more quickly upon induction with MAL2p‐myc‐GFP than with MAL2p‐5'UTR‐GFP, demonstrating that the WOR1 5′ UTR is sufficient for reducing translational efficiency. Notably, opaque cells of MAL2p‐5'UTR‐GFP expressed more GFP protein than their corresponding white cells upon maltose induction (Fig. 5B). Thus, the difference in translational efficiency imparted by the 5′ UTR is less stark in opaque cells, suggesting less regulation by the 5′ UTR in this state. Similar results were seen by observing GFP fluorescence using fluorescence‐activated cell sorting (not shown).

Figure 5

The 1 5′ UTR represses translation of a downstream reporter at both room temperature and 37°C.

A. RT‐qPCR of GFP mRNA and B. Western blot detecting GFP protein in white and opaque cells of WT + MAL2p‐WOR1 5'UTR‐GFP (HLY4216) and WT + MAL2p‐myc‐GFP (HLY4217).

C. Translational efficiency of GFP as a ratio of GFP fluorescence detected by FACS vs. GFP mRNA level as quantified by qPCR. Strains were grown to log phase in YEP + dextrose at RT, then inoculated to YEP + maltose at 37°C for 4 h to induce expression from the MAL2 promoter. qPCR experiments were performed in triplicate and values were normalized to ACT1.

Because of the increased opaque stability by the Δ5′‐WOR1 at 37oC, we investigated whether the 5′ UTR regulation is temperature dependent. As growth at 37°C favors white state, a temperature‐sensitive translational regulation by the 5′ UTR could provide a mechanism disfavoring opaque formation at this temperature. We compared translational efficiency (protein/mRNA ratio) of GFP between MAL2p‐5'UTR‐GFP and MAL2p‐myc‐GFP strains while inducing the MAL2 promoter at room temperature or 37°C. The WOR1 5′ UTR had a repressive effect on translation at both RT and 37°C, with the effect being moderately stronger at 37°C. In the absence of the WOR1 5′ UTR, translational efficiency of MAL2p‐myc‐GFP was not affected by temperature (Fig. 5C). We found similar results for opaque cells (not shown); however, the MAL2 promoter is poorly induced in opaque state at RT, so a different system would be needed to confirm this finding. Our data suggest that the WOR1 5′ UTR is not the major point of regulation by temperature in opaque stability.

Argonaute, Ssd1 and uncapped 5′ UTR‐ 1 transcripts are not involved in translational regulation at the 1 5′ UTR

Several possibilities exist for the mechanism of translational regulation by the WOR1 5′ UTR. One such possibility is through micro‐RNA‐directed translational repression. It has been demonstrated that translational repression by miRNAs interacts with the 5′ UTR of genes through the initiation factor eIF4A2, which unwinds 5′ UTR structure to allow binding of the 60S ribosomal subunit (Meijer et al., 2013). Although RNAi is absent from S. cerevisiae, many other budding yeasts, including C. albicans, have functional Argonaute and Dicer proteins as well as small non‐coding RNAs, suggesting a possible RNAi system (Liu et al., 1996; Drinnenberg et al., 2009) To test the possibility of translational repression of RNA transcripts by Argonaute, we examined whether WOR1 5′ UTR‐associated translational repression is present in the absence of Argonaute (AGO1). An ago1 mutant was constructed and transformed with MAL2p‐5'UTR‐GFP or MAL2p‐myc‐GFP. We found that in an ago1 mutant, levels of GFP protein from the 5′ UTR‐GFP were still greatly reduced compared with myc‐GFP absent the WOR1 5′ UTR, similar to data observed in wild type (Fig. 6A). This result is corroborated by fluorescence‐activated cell sorting (FACS) measuring GFP fluorescence (Fig. S3). In addition, GFP mRNA transcripts from 5'UTR‐GFP and myc‐GFP are found at similar levels in the ago1 mutant (Fig. 6B), indicating that the difference in protein levels is due to translational and not transcriptional regulation. Therefore, Argonaute is not necessary for translational repression by the WOR1 5′ UTR. This is consistent with our observeration of the lack of an effect by AGO1 deletion on white‐opaque switching (not shown).

Figure 6

Ago1 is not required for the translational repression by the 1 5′ UTR.

A. Western blot detecting GFP and B. GFP/ACT1 transcript levels as detected by qRT‐PCR from white cells of WT + MAL2p‐WOR1 5'UTR‐GFP (HLY4216), WT + MAL2p‐myc‐GFP (HLY4217), ago1 (HLY3673) + MAL2p‐WOR1 5'UTR‐GFP (HLY4218) or ago1 + MAL2p‐myc‐GFP (HLY4219). Two individual transformants from ago1 mutant background were used (ago1‐1 and ago1‐2). Strains were grown to log phase in YEP + dextrose, then inoculated to YEP + maltose for 4 h at 37°C to induce transcription from the MAL2 promoter. GFP was detected with Clontech Living Colors anti‐GFP antibody.

To examine alternate mechanisms of translational regulation, we considered the role of RNA‐binding proteins that may potentially interact with the WOR1 5′ UTR. In S. cerevisiae, Ssd1 is known to bind at the 5′ and 3′ UTR of specific mRNAs, reducing their translational efficiency (Cao et al., 2006). Ssd1 activity is repressed by Cbk1, which is part of the RAM network critical for the regulation of cellular morphogenesis, polarized growth and septum destruction (Brooks and Jackson, 1994; Kolonko et al., 2010; Lopes da Rosa et al., 2010). Homologues of Cbk1, Ssd1 and downstream genes are present in C. albicans, where they are also critical for cell morphogenesis and mother–daughter separation (Jackson et al., 1994; Lopes da Rosa et al., 2013; Watanabe et al., 2013). Considering that opaque cells are better separated than white cells between mother and daughters, the Cbk1‐Ssd1 pathway could potentially be differentially regulated between white and opaque cells, and Ssd1 could be a candidate for translational repression at the WOR1 5′ UTR. If translational repression through the WOR1 5′ UTR is indeed regulated through Ssd1, deletion of SSD1 should reduce this repression. To investigate this, we transformed a ssd1 deletion mutant (Zentner and Henikoff, 2013) with MAL2p‐myc‐GFP or MAL2p‐UTR‐GFP. FACS was performed to assess population‐level GFP fluorescence as a reporter of translation in maltose‐containing medium. We find that fluorescence of MAL2p‐5'UTR‐GFP is lower than that of MAL2p‐myc‐GFP in the ssd1 mutant as in the wild‐type control (Fig. S3). Thus, it appears that translational regulation through the WOR1 5′ UTR is not regulated by Ssd1.

Another possible mechanism of translational regulation by the WOR1 5′ UTR is cap‐independent translation initiation. To explore this possibility, we sought to determine whether uncapped 5′ UTR‐WOR1 transcripts are normally found in the cell and at what levels. The splint‐ligation‐based qSL‐RT‐PCR method described by Blewett et al. was used to detect uncapped mRNA (Blewett et al., 2011). RNA was extracted from wild‐type opaque cells, and ligation reactions performed to ligate the 5'UTR‐WOR1 RNA to an RNA anchor oligo, facilitated by a DNA splint homologous to both. Only RNA transcripts without the 5′ cap would be ligated to the anchor. RT‐qPCR was then performed to detect both an amplicon within the original transcript (WOR1 5′ UTR) and one spanning the anchor and 5′ UTR. Detection of the second amplicon is indicative of successful ligation of an uncapped transcript. A ligase‐free reaction was performed as a control. We find that with the addition of ligase, the detection of uncapped 5′ UTR‐WOR1 transcripts are 0.58‐fold of ligase‐free reaction, indicating that few, if any, 5′ UTR‐WOR1 transcripts exist in uncapped form. In comparison, detection of uncapped ACT1 increases by 9.13‐fold when ligase is added, demonstrating that the ligation reaction proceeds efficiently when uncapped transcripts exist (Table 1). Although the high level of uncapped ACT1 transcript appears reflective of non‐specific background decapping activity, this level of decapping is not seen for 5′ UTR‐WOR1 transcripts. These results indicate that the mechanism behind the lower translational efficiency of the WOR1 5′ UTR does not involve translation of uncapped transcripts. We also tested the possibility of alternative transcription originating within the WOR1 5′ UTR, resulting in shorter WOR1 transcripts lacking all or part of the 5′ UTR. To assess whether these transcripts were produced and whether they appeared transiently during switching or in stable opaque cells, we used a strain carrying the WOR1p‐5′UTR‐GFP and MAL2p‐WOR1. This would allow for induction of white‐opaque switching by ectopically expressing WOR1 under the MAL2 promoter, whereas transcription from the WOR1 promoter could be assayed through GFP RNA levels, without interference from ectopically expressed WOR1. Northern blotting was conducted to probe for GFP‐containing RNA. As expected, the appearance of 5′ UTR‐GFP RNA was correlated with switching to opaque. Although alternate transcription would have produced both a long (5′ UTR‐GFP) and short (GFP only) transcript, we found GFP transcript of only one size, correlating to 5′ UTR‐GFP at 2.7 kbp (Fig. S4). Thus, alternative transcription does not appear to initiate in the WOR1 5′ UTR.

Table 1

Absence of uncapped 5′ UTR‐ 1 transcripts as detected by qSL‐RT‐PCR

		Ct transcript only	SD	Ct anchor‐transcript	SD	Fold above background
ACT1	Ligase	14.08	0.04	32.15	0.15	9.13
	No ligase	15.08	0.10	36.34	0.21
5'UTR‐WOR1	Ligase	20.02	0.18	34.98	0.78	0.58
	No ligase	19.64	0.12	33.81	0.40

Splint ligation was performed on RNA extracted from opaque wild‐type cells, using DNA splints that facilitate the ligation of an RNA anchor oligo to uncapped ACT1 or 5'UTR‐WOR1 RNA transcript. Reactions for which ligase was omitted were used as control for background. RNA was purified from ligation reactions and used for RT‐PCR. Mean Ct values and standard deviation are shown for the amplification of each transcript as well as the ligated anchor‐transcript region. Fold change above background was calculated as ΔCt. Experiments were performed in triplicate.

Monosome association of 5′ UTR‐containing transcripts extends to other key regulators of morphology in . albicans

Although long 5′ UTRs are uncommon in C. albicans, they are notably present on a handful of genes regulating morphological switches (Bruno et al., 2010; Tuch et al., 2010). Key regulators of the yeast‐hyphal transition, such as RFG1 and FLO8, and of the white‐opaque circuitry, such as CZF1 and EFG1 (also a hyphal regulator), all have 5′ UTRs in excess of several hundred to several thousand base pairs (Table 2). The length of the 5′ UTR in these genes raises the question of whether they possess a translational repressive activity like the WOR1 5′ UTR. To address this question, we conducted a polysome profile assay on wild‐type opaque cells (Fig. 7A). The identification of monosome and polysome fractions was corroborated by 40S and 60S subunit content in each fraction as measured by RT‐qPCR (Fig. 7B). Further RT‐qPCR revealed that transcripts with long 5′ UTRs, such as CZF1, EFG1, RFG1 and FLO8, were consistently enriched in monosomal fractions compared with polysomes, suggesting many of these transcripts are stalled in translational initiation or otherwise less efficiently translated. ACT1, which is actively translated and has a short 5′ UTR, displayed the opposite pattern and was more enriched in polysomes (Fig 7C). This result demonstrates that a long 5′ UTR is associated with reduced translational efficiency in key regulators of morphology in C. albicans.

Table 2

5'UTR length of key regulatory genes in . albicans, based on transcriptome analysis by RNA‐seq (Tuch et al., 2010)

Gene name	5′ UTR length (bp)
CZF1	2071
WOR1	1978
RFG1	1267
EFG1	1139
FLO8	835
ACT1	72

Figure 7

Transcripts with long 5′ UTR are more enriched in monosome in . albicans.

A. Polysome profile of wild‐type (JYC5) opaque cells grown to log phase in YPD at room temperature.

B. 40S and 60S ribosomal subunit abundance in each fraction as determined by RT‐qPCR.

C. Relative abundance in each fraction of transcripts of white‐opaque regulatory genes CZF1, FLO8, RFG1, EFG1, as well as ACT1, as determined by RT‐qPCR. Abundance of each transcript was normalized to the fraction containing the highest value.

Discussion

Extensive work has been done to characterize the network governing white‐opaque switching in C. albicans, involving the master regulator WOR1 and numerous other transcriptional regulators. However, much remains to be understood regarding the regulation of white‐opaque switching at the translational level. In this study, we identify the WOR1 5′ UTR as a regulator of white‐opaque switching that reduces the translational efficiency of WOR1. The WOR1 5′ UTR negatively regulates white‐opaque switching and reduces the stability of the opaque cell state. We show evidence of reduced translational efficiency of 5′ UTR‐WOR1 both through polysome analysis, in which 5′ UTR‐WOR1 transcripts are more enriched in monosomes than polysomes, and through assays which demonstrate lower translation from 5′ UTR‐WOR1 than Δ5′‐WOR1 transcripts, even as transcription of the two remain the same. This effect is independent of the WOR1 promoter and also of the WOR1 coding region; placing the WOR1 5′ UTR upstream of another gene under an ectopic promoter produces the same reduction in translation. Therefore, the WOR1 5′ UTR is sufficient for the observed translational repression.

The mechanisms by which the WOR1 5′ UTR regulates translation leave ample room to be explored. There are numerous known mechanisms through which a 5′ UTR may affect translational efficiency. Within 5′ UTR regions themselves, uORFs and upstream AUG start codons (uAUGs) have been identified to regulate translation. Work in yeast has demonstrated that out‐of‐frame uAUGs can significantly alter the translational efficiency of a gene (Dvir et al., 2013). In C. albicans, an example of translational regulation by a uORF has been demonstrated for the transcription factor GCN4 (Sundaram and Grant, 2014b). The WOR1 5′ UTR contains no uORFS starting with AUG; however, uORFs with non‐AUG start codons are another possibility, as ribosome profiling in S. cerevisiae has demonstrated pervasive translation initiating at non‐AUG codons in 5′ UTRs (Ingolia et al., 2009). Although potential non‐AUG start codons such as GUG and UUG do exist in the WOR1 5′ UTR, ribosome profiling in C. albicans suggests that near‐AUG codons have little effect on translation (Muzzey et al., 2014). Therefore, it is unlikely that translational regulation by the WOR1 5′ UTR is mediated by uORFs. Secondary structures in the 5′ UTR are also known to negatively regulate translational efficiency, as stable structures can inhibit translational initiation. The secondary structure of the WOR1 5′ UTR and its effects on translation have yet to be determined due to the length of the 5′ UTR. Additional work has shown that the length and structure of 5′ UTRs impact translational repression by miRNAs through RNA interference (Meijer et al., 2013). C. albicans potentially has an intact RNAi system (Briggs et al., 1989). We tested the effect of Argonaute on 5′ UTR translational regulation and found that deletion of AGO1 did not increase the translational efficiency of transcripts bearing the WOR1 5′ UTR. Thus, regulation through Argonaute appears an unlikely mechanism. It should be noted that Argonaute and Dicer appear insufficient for RNA interference in C. albicans (Liu et al., 1996), and other components of RNA silencing pathways may still affect WOR1 transcript regulation.

Cap‐independent translation initiating in the 5′ UTR provides an alternate mechanism for translational regulation. It is possible that cap‐independent translation of 5′ UTR‐WOR1 occurs at a slower rate than cap‐dependent translation and that this may be a dominant means of WOR1 translation under some conditions. As a preliminary test, we assayed whether uncapped 5′ UTR‐WOR1 transcripts exist in opaque cells. We find that such transcripts are low, if not undetectable. Thus, the translation of uncapped RNAs does not appear to be a major mechanism regulating translation of 5′ UTR‐WOR1.

Finally, RNA‐binding protein activity at the 5′ UTR is another mechanism for translational regulation. One such example is Ssd1, which has been demonstrated to repress translation of specific genes by binding to their 5′ UTR and 3′ UTR in S. cerevisiae (Cao et al., 2006). Notably, Ssd1 represses translation of a specific subset of genes involved in cell morphogenesis and mother–daughter separation (Brooks and Jackson, 1994). Ssd1 targets genes expected to be highly expressed in opaque cells, making it a candidate for regulation at the WOR1 5′ UTR. We find that translational efficiency is reduced by the WOR1 5′ UTR in ssd1 deletion mutant to a similar extent as wild type, so Ssd1 does not appear to regulate translation through the WOR1 5′ UTR. However, the role of numerous other RNA‐binding proteins on the WOR1 5′ UTR remain to be tested. Whether the observed translational regulation at the 5′ UTR occurs through one of these proteins will require further exploration.

The identification of the WOR1 5′ UTR as a translational regulator of WOR1 raises many questions for further investigation. The role of the 5′ UTR in translational regulation in higher eukaryotes has been extensively characterized, and in recent years studies in yeast have explored instances of UTR‐mediated regulation as well. However, in C. albicans, the characterization of translational effect of 5′ UTRs has only begun. Recent work has shown that the UME6 5′ UTR, which is long and has predicted complex structure, reduces the translational efficiency of UME6 and thereby negatively regulates filamentation (Childers et al., 2014). RNAseq studies in C. albicans have identified long 5′ UTRs on genes regulating white‐opaque switching, hyphal morphogenesis and key processes in pathogenesis and commensalism. Our findings on WOR1 identify a novel level of regulation for white‐opaque switching through translation and support the possibility that key regulators of cell fate and cell morphology in C. albicans are translationally regulated through their 5′ UTRs. Further work may demonstrate roles for 5′ UTR on numerous other genes and characterize the mechanisms through which regulation occurs.

The ability of the WOR1 5′ UTR to regulate translational efficiency also raises questions on the biological impact of the 5′ UTR on white‐opaque switching, as well as its role in helping C. albicans adapt to environmental changes. Of particular note is its relevance to the timescale of switching and the processes behind cell fate determination. White‐opaque switching is slow, stochastic and infrequent. Even with ectopic expression of WOR1, many hours still pass before endogenous Wor1 protein is expressed at high levels, and switching is able to occur. The WOR1 5′ UTR provides one means through which opaque phase commitment is delayed. The reduced translational efficiency of WOR1 conferred by the 5′ UTR may be a mechanism for preventing rapid switching to opaque during transient increases of WOR1 transcription, such that commitment to opaque only occurs when long‐term conditions are favorable to opaque state. Consistent with this notion, we find that the translational repression by the WOR1 5′ UTR is more pronounced in white than opaque cells. In addition, the WOR1 5′ UTR reduces stability of opaque state, which may allow opaque cells to switch back to white under conditions that no longer favor opaque. We have begun to test whether the WOR1 5′ UTR is a sensor of environmental conditions that affect white‐opaque regulation. We found only a moderate change in 5′ UTR‐mediated translational repression in response to 37°C, so the 5′ UTR is not a major thermosensor. In addition, we did not observe synergy between the switching phenotype of Δ5′‐WOR1 strains with CO2 and GlcNAc, which are known to induce white‐opaque switching, making the 5'UTR less likely to be a sensor for these conditions. However, the role of the 5′ UTR in response to in vivo conditions that induce the expression of Wor1, such as those in the GI tract that promote formation of the opaque‐like GUT phenotype (Pande et al., 2013), remain to be tested. Inhibition of translational initiation has recently been shown to be crucial to the response of C. albicans to oxidative stresses (Sundaram and Grant, 2014a). It will be of interest to explore further relationships between translational regulation and environmental signals in C. albicans. In this light, the 5′ UTR provides another layer of regulation that facilitates adaptation to long‐term environmental changes without responding inappropriately to fluctuations in growth conditions. Regulation by the WOR1 5′ UTR promotes white phase stability and opaque‐to‐white switching, adding a layer of regulation to WOR1, which itself promotes opaque. It will be intriguing to investigate whether 5′ UTR of other genes regulate cell fate through reducing or delaying switching and cell fate commitment.

Experimental procedures

Strains and culturing conditions

Strains used in this study are listed in Table S1. Strains were grown in YEP (1% yeast extract, 2% peptone) + 2% dextrose or maltose, or SCD (synthetic complete + 2% dextrose) media unless otherwise described. Colonies were maintained at room temperature (opaque cells) or 30°C (white cells).

Strain construction

Plasmid pMAL2‐WOR1‐HA was constructed in this lab as described by Huang et al. (Huang et al., 2006). Plasmid WOR1p‐Δ5′WOR1‐HA was modified from pMAL2‐WOR1‐HA by replacing the MAL2 promoter with a fragment of the WOR1 promoter 5081 to 1975 base pairs upstream of the start codon, excluding the 5′ UTR sequence. WOR1p‐Δ5'WOR1‐HA was digested with ClaI for integration at the ClaI site 4647 bp upstream of the WOR1 start codon, and transformed into the strain WOR1/wor1::HIS1 to produce 5′ UTR‐WOR1/Δ5′‐WOR1, or wor1Δ to produce Δ5′‐WOR1/wor1. Transformants were selected for growth on SCD‐ura. Integration of Δ5′‐WOR1 into the correct allele in the WOR1/wor1::HIS1 background was verified by Southern blotting. To construct the Δ5′‐WOR1/Δ5′‐WOR1 strain, the URA3 marker in Δ5′‐WOR1/wor1 was replaced with the SAT1 marker to confer nourseothiricin (NAT) resistance. The SAT1 marker driven by the ACT1 promoter was amplified from plasmid pNIM1 (Ramirez‐Zavala et al., 2008), with flanking sequences homologous to URA3 introduced during amplification, then transformed into Δ5′‐WOR1/wor1. NATR and ura3‐ strains were selected from colonies capable of growth on YPD + 200 mM NAT and incapable of growth on SCD‐ura, then transformed again with plasmid WOR1p‐Δ5'WOR1‐HA, resulting in the Δ5′‐WOR1/Δ5′‐WOR1 strain. Nourseothiricin was obtained as ClonNAT from WERNER BioAgents. The MAL2p‐5'UTR‐WOR1‐HA plasmid was constructed by inserting a PCR amplification product containing the WOR1 5′ UTR into the XbaI site at the 5′ end of WOR1 in pMAL2‐WOR1‐HA.

The MAL2p‐myc‐GFP plasmid was modified from the MAL2p‐myc‐HGC plasmid (Levchenko et al., 2005), by replacement of HGC with a 700 bp GFP fragment (Wunsch and Jackson, 2005). The MAL2p‐5'UTR‐GFP plasmid was modified from the MAL2p‐myc‐GFP plasmid by excision of the 13‐myc tag via digestion with XbaI and MluI, and replacement with a PCR product containing the WOR1 5'UTR. All plasmids containing the MAL2 promoter described in this study, unless otherwise specified, were digested with AscI for integration at the ADE2 locus for transformation into C. albicans. The WT + WOR1p‐GFP + MAL2p‐WOR1 strain was constructed as described (van Daal and Elgin, 1992).

The ago1 mutant was constructed by deleting AGO1 from the wild‐type strain SN148 using the two‐step PCR deletion method described by Noble and Johnson (Noble and Johnson, 2005). One copy of AGO1 was replaced with C. dubliniensis HIS1 from pSN52. Due to difficulty of directly replacing the second copy, a MAL2p‐AGO1‐URA3 plasmid was transformed into the resulting AGO1/ago1 strain and integrated at the ADE2 locus. Subsequently, the remaining copy of AGO1 at its native locus was replaced with Candida maltosa LEU2 amplified from pSN40, and the MAL2p‐AGO1‐URA3 insert eliminated by growth on 5‐fluorootic acid (5'FOA). Complete deletion of AGO1 was verified by PCR. The MAL2p‐AGO1‐URA3 plasmid was modified from BES116 (Jackson, 1987). All primers used in strain construction are listed in Table S2.

Switching assays

Stable white or opaque cells from log‐phase cultures were plated to SCD agar plates at a density of 100 cells plate−1, using 100 μl of a dilution of 1000 cells ml−1. Colony morphology was scored after 7 days. Both whole‐colony and sectored switching events were counted. For white‐opaque switching assays, plates were incubated at 25°C. For opaque stability assays, plates were incubated at 25 and 37°C.

Northern blotting

RNA was extracted from log‐phase cultures using the Qiagen RNeasy kit. Northern blotting was conducted as described (Chen et al., 2002). A 500 bp PCR product of WOR1 or GFP was used to probe for RNA containing the region of interest. Loading was assessed using a fragment of ACT1 as probe, or with visualization of ribosomal RNA. Primers used for amplification of the WOR1 probe are listed as WOR1probeF/R in (Huang et al., 2006).

Western blotting

Protein extraction was modified from (Meimoun et al., 2000). After precipitation with TCA as described, samples were washed once with acetone and boiled 5 min with equivolume urea buffer (8M urea, 4M 2‐mercaptoethanol, 125 mM Tris pH 6.8 and 10% SDS), then centrifuged 5 min at 13 krpm. The supernatant was mixed with equivolume 20% glycerol + bromophenol blue and run on SDS‐PAGE gels for Western blotting. The HA tag was detected with monoclonal rabbit anti‐HA antibody from Abcam. GFP was detected with Living Colors monoclonal mouse anti‐GFP antibody from Clontech. PSTAIRE (Cdc28) antibody was obtained from Invitrogen. Secondary IgG antibodies (anti‐mouse or anti‐rabbit) were obtained from Bio‐Rad.

Polysome profiling

Polysome profiling was modified from (Masek et al., 2011). C. albicans cultures were grown to OD600 = 0.6, treated with 1 mg ml−1 cycloheximide, and incubated at 4°C for 5 min. Fifty milliliters of culture were collected for each sample. Extraction buffer was supplemented with 1 mg ml−1 cycloheximide. 7–47% sucrose gradients were prepared by stepwise freezing of 2.4 ml each of 7%, 17%, 27%, 37% and 47% sucrose in Beckman Coulter polyallomer centrifuge tubes at −80°C, then thawing at 4°C overnight. For EDTA control samples, each gradient layer contained 20 mM EDTA. Lysate samples were applied to gradients and centrifuged at 35,000 r.p.m. at 4°C for 2.5 h in a Beckman SW41 rotor. Gradients were fractionated, absorbance was read at 254 nm, and polysome profile was created using the ISCO Gradient Fractionator (Teledyne). Fractions were collected using the Foxy Jr Fraction Collector. Fractions of 500 μl each were added to 800 μl 8 M guanidine HCl and 700 μl 100% EtOH, and placed at −20°C overnight for RNA precipitation. Pellets were washed with 70% EtOH and resuspended in 30 μl RNase‐free water for subsequent use in RT‐qPCR.

Quantitative PCR

cDNA was synthesized from 2 μg total RNA using the Bio‐Rad iScript Reverse Transcription Kit. Quantitative PCR using the Bio‐Rad SYBR Green mix was performed on the Bio‐Rad iCycler. Cycle conditions were 95°C for 1 min, then 39 cycles of 95°C for 10 s, 56°C for 45 s, and 68°C for 20 s. Oligos used are listed in Table S2. All reactions were carried out in triplicate.

Splint ligation for qSL‐RT‐PCR

Ten micrograms of total RNA, 20 pmol of DNA splint and 30 pmol of RNA anchor were mixed and incubated at 70°C, 60°C, 42°C and 25°C, for 5 minutes at each step. The 20 U T4 DNA ligase (Invitrogen), T4 DNA ligase buffer and 20 U RNase Inhibitor Plus (Promega) were then added and ligation allowed to proceed overnight at 15°C. Ligase was omitted for control samples. Reactions were then treated with DNase using the Qiagen DNAse kit and protocols described therein. RNA was extracted using phenol/chloroform, precipitated with 1 ml 100% EtOH and 0.3M sodium acetate for 2 h at −20°C and washed once with 70% EtOH. RNA was then resuspended in 13.3 μl RNase‐free water and used for RT‐qPCR.

FACS

Fluorescence‐activated cell sorting was conducted using the BD FACSCalibur system. FL1‐H channel was used for GFP detection, and 10,000 cells were counted per sample. Results were analyzed with FlowJo cytometry analysis software.

Supporting information

Click here for additional data file.