Asymmetric distribution of glucose transporter mRNA provides a growth advantage in yeast

Abstract

Asymmetric localization of mRNA is important for cell fate decisions in eukaryotes and provides the means for localized protein synthesis in a variety of cell types. Here, we show that hexose transporter mRNAs are retained in the mother cell of S. cerevisiae until metaphase–anaphase transition (MAT) and then are released into the bud. The retained mRNA was translationally less active but bound to ribosomes before MAT. Importantly, when cells were shifted from starvation to glucose‐rich conditions, HXT2 mRNA, but none of the other HXT mRNAs, was enriched in the bud after MAT. This enrichment was dependent on the Ras/cAMP/PKA pathway, the APC ortholog Kar9, and nuclear segregation into the bud. Competition experiments between strains that only expressed one hexose transporter at a time revealed that HXT2 only cells grow faster than their counterparts when released from starvation. Therefore, asymmetric distribution of HXT2 mRNA provides a growth advantage for daughters, who are better prepared for nutritional changes in the environment. Our data provide evidence that asymmetric mRNA localization is an important factor in determining cellular fitness.

Introduction

Cells have to respond to an ever‐changing environment. This is true not only for single cell organisms like the yeast Saccharomyces cerevisiae but also for multicellular organisms, including humans. Variations in the availability of nutrients, in particular glucose, are one of the major challenges and cells have evolved a number of strategies to counteract glucose depletion. For example, under glucose‐rich conditions, insulin promotes the plasma membrane expression of the glucose transporter GLUT4 in adipocytes allowing glucose uptake (Karnieli et al, 1981; Martin et al, 2000). In contrast, another glucose transporter, GLUT1, is stabilized at the plasma membrane under energy stress conditions (Wu et al, 2013). Alterations in glucose transporter expression have been observed in numerous diseases, including a variety of cancers. For example, GLUT1 and GLUT3 are overexpressed in solid tumors, and their expression levels are used as prognostic and predictive markers. Overexpression has been associated with poor survival (Barron et al, 2016) as it may help to increase biomass production and tumor progression (Yun et al, 2009; Calvo et al, 2010).

The yeast Saccharomyces cerevisiae is sensitive to alterations of nutrient availability in the environment. Because of its inability to actively move toward a food source, it has developed strategies to adapt quickly to local changes. Depending on the accessibility of glucose for example, yeast expresses a suitable set of its 17 hexose transporters to ensure an optimal growth pattern (Bisson et al, 2016). It has been well established that the transcription of the hexose transporters is modulated by glucose concentrations. In addition, the abundance of glucose transporters at the plasma membrane is regulated by endocytosis and subsequent degradation, if they are no longer needed (Snowdon & van der Merwe, 2012; Roy et al, 2014; O'Donnell et al, 2015; Llopis‐Torregrosa et al, 2016; Hovsepian et al, 2017). Even though these processes were initially studied mostly in yeast, they are conserved up to humans, indicating that S. cerevisiae is an excellent model organism for these types of studies.

Responses to changes in the environment can occur at both the transcriptional and the post‐transcriptional level. Whereas our understanding of global transcriptional responses to environmental dynamics has vastly expanded from the deluge of next‐generation sequencing data, much less is known about post‐transcriptional processes. This is partly due to the complexity of regulatory processes occurring at the levels of both RNA and protein. In the case of mRNA, a multitude of factors determines its stability, whether it is translated or stored, and how and where it is localized. All these mechanisms contribute to the regulation of protein expression and can be modulated in response to specific stresses (Wang et al, 2018).

In this regard, subcellular localization of mRNA has been established as one important mechanism to control the timing and extent of protein expression (Parton et al, 2014). During development and stem cell division for example, asymmetric localization of mRNA determines cell fate (Tsunekawa et al, 2014; Ramat et al, 2017). Additionally, for neurotransmission, axonal and dendritic mRNA localization and localized translation are essential (Sigrist et al, 2000; Zhang & Poo, 2002). Yet, mRNA localization is not only limited to specialized cells. Localized mRNAs have been detected in a wide range of organisms including bacteria, yeast, plants, and animals (Medioni et al, 2012). Moreover, mRNA can be localized to a variety of intracellular organelles. In fact, there are thousands of mRNAs that exhibit specific subcellular localization patterns in mammals and Drosophila (Lecuyer et al, 2007; Mili et al, 2008). It is generally assumed that mRNA localization correlates with regulated translation, suggesting that this highly conserved process is very efficient.

Here, we investigated the mRNA localization of the hexose transporter Hxt2. HXT2 mRNA is asymmetrically localized during the cell cycle, showing a strong retention in the mother cell until mitosis at which point equal portioning between mother and future daughter cell is achieved. The retained mRNA is presumably translationally inactive during early phases of the cell cycle and only becomes translated upon entry into mitosis. Importantly, refeeding of starved cells with glucose shifted the equal distribution of HXT2 mRNA to an enrichment in the bud. This process was dependent on the Ras/cAMP/PKA signaling pathway, spindle positioning, and the nuclear pore components Nup2 and Mlp1/2, which were reported to be involved in nuclear pore complex inheritance. The asymmetric enrichment of HXT2 mRNA during mitosis may contribute to a growth advantage of those daughters over cells that are unable to mount a similar response.

Results

Hxt2 protein is asymmetrically localized in yeast cells

We and others have used Hxt2 protein as a marker of the plasma membrane (Bagnat & Simons, 2002; Walther et al, 2006; Zanolari et al, 2011; Estrada et al, 2015). We observe, however, that in small‐ and medium‐sized buds, Hxt2 levels were always rather low but increased as the cell cycle progressed until cytokinesis, at which equal levels of Hxt2 were present in the mother and the daughter cells (Fig 1A). These differences in protein levels might have been brought about by transcriptional and/or translational control. Indeed, several high‐throughput studies indicate that HXT2 mRNA expression occurs at restricted times in the cell cycle (Cho et al, 1998; Spellman et al, 1998; Pramila et al, 2006). However, HXT2 mRNA was reported to be expressed at the mitosis to G1 boundary, an expression pattern that does not match well with the observed protein expression pattern.

Figure 1. The localization of both HXT2 mRNA and Hxt2 protein is cell‐cycle dependent.

1. The distribution of the GFP‐tagged Hxt2 changes during cell division. In small and medium‐budded cells (white arrowheads), Hxt2 is mainly present in the membrane of the mother cell. In large‐budded cells, Hxt2 is distributed over the entire plasma membrane (yellow arrowheads).
2. HXT2 mRNA localization correlates with that of the protein. FISH with an HXT2‐specific probe. Early in the cell cycle the mRNA is predominantly found in the mother cell (white arrowheads) while it is equally distributed after meta‐to‐anaphase transition (MAT) (yellow arrowheads).
3. Quantification of (B). The fluorescence intensities from mother and bud were independently measured and the ratio of the signal from the mother over the signal from the bud was calculated. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. Three independent experiments with 50 mother cells and 50 buds per experiment were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively.
4. Changes of HXT2 mRNA localization appear to be independent of the cytoskeleton. Cells were either treated for 15 min with 30 μg/ml latrunculin A or with 30 μg/ml benomyl. White arrowheads point to small‐budded and yellow arrowheads to large‐budded cells.
5. Quantification of (D) of three independent experiments with at least 50 cells each. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively.

Data information: Scale bars in (A, B and D) correspond to 10 μm.Source data are available online for this figure.

HXT2 mRNA is asymmetrically localized early and equally distributed late in the cell cycle

As pointed out above, mRNA localization is a potential mechanism to regulate protein expression. Therefore, we determined HXT2 mRNA cellular distribution by fluorescence in situ hybridization (FISH; Fig 1B). The probe is specific for HXT2 mRNA (Fig EV1D). In small‐ and medium‐budded cells, HXT2 mRNA was restricted to the mother cell (Fig 1B, white arrowheads), but in large‐budded cells (Fig 1B, yellow arrowhead), HXT2 mRNA became equally distributed between mother and daughter cells. One explanation for this observation is that the mRNA distribution was connected to DNA segregation onto the two poles or in other words to the metaphase–anaphase transition (MAT). To investigate whether HXT2 mRNA localization is indeed correlated to cell‐cycle progression, we abrogated mitosis by treatment with nocodazole. Under these conditions, HXT2 mRNA remained restricted to the mother cell, suggesting a link between HXT2 mRNA localization and cell‐cycle stage (Fig EV1A). For a more quantitative measure as readout from the FISH experiments, we determined the fluorescence intensity in the mother and the bud. The quotient of the mean fluorescence intensity of the mother cell over the bud/daughter cell reflects the relative mRNA distribution. A quotient of >1 indicates enrichment in the mother, and <1 in the bud (Fig 1C). We scored cells with a bud and containing either one (before metaphase–anaphase transition [MAT]) or two nuclei (after MAT). We conclude that HXT2 mRNA localization changes over the cell cycle and that this change in localization is very robust and reproducible.

Figure EV1. Controls for HXT2 mRNA localization and translation experiments.

1. HXT2 mRNA release from the mother cell is coupled to cell‐cycle progression and nuclear segregation respectively. Cells arrested in G2/M‐phase with nocodazole show still retention of HXT2 mRNA in the mother even in large‐budded cells (arrows). Cells were treated with 15 μg/ml for 3 h, subsequently fixed, and mRNA was visualized by FISH.
2. Rhodamine‐phalloidin staining. Cells were either treated with 30 μg/ml Latrunculin A (LatA) or as a solvent control with DMSO for 30 min. After fixation, actin was stained with rhodamine‐phalloidin. LatA‐treated cells show no actin cables or patches anymore.
3. Benomyl treatment leads to the depolymerization of cytoplasmic microtubules but cells are still able to segregate the nuclei. Immunofluorescence of microtubules with an antibody against α‐tubulin.
4. The FISH probes against HXT2 mRNA are specific. FISH with a HXT2 deletion strain or the no probe control shows no signal.
5. Two additional independent polysome profiling experiments show the same trend as Fig 2D.

Data information: Scale bars in (A–D) represent 10 μm.Source data are available online for this figure.

We wondered whether the HXT2 mRNA was transported to the bud through an active transport pathway, which would involve the cytoskeleton, or through passive diffusion. We depolymerized actin cables with latrunculin A (LatA; Fig EV1B) and microtubules (MTs) with benomyl (Fig EV1C) and assessed the localization of HXT2 mRNA. A benomyl concentration was chosen such that the number of cytoplasmic MTs was strongly reduced but cells could still undergo mitosis (Fig EV1C). However, neither mother cell retention nor release after MAT of HXT2 mRNA was affected when the cytoskeletal drugs were applied (Fig 1D and E). Our data suggest that under these conditions, HXT2 mRNA diffuses into the bud after MAT.

HXT2 mRNA localization in the bud is independent of de novo transcription

Even though HXT2 mRNA was reported to be expressed at the M/G1 boundary, we cannot exclude that the change in distribution is due to transcriptional activity. To distinguish between mRNA movement in the cytoplasm from the mother to the bud and transcriptional increase of HXT2 mRNA concentration, we blocked transcription with 1,10‐phenanthroline. Under these conditions, the overall HXT2 mRNA signal was weaker, consistent with a decrease in transcriptional activity. Nevertheless, we still observed equal distribution of the mRNA after anaphase (Fig EV2A and B). Therefore, the change in HXT2 mRNA localization is largely independent of de novo mRNA synthesis. To corroborate our findings, we replaced the HXT2 promoter by the constitutive ADH and GPD promoters, which vary in their strength and led to a slight to moderate overexpression of HXT2 mRNA (Fig EV2C). Interestingly, under those conditions, HXT2 mRNA was still retained in the mother (Fig EV2D and E), indicating an active retention mechanism. In contrast, after mitosis, most of the HXT2 mRNA was still present in the mother cell, suggesting that the mRNA signal in the daughter is independent of increased mRNA levels. Taken together, these results suggest that transcription is not a major contributor to HXT2 mRNA localization changes in logarithmically growing cells.

Figure EV2. Transcription contributes to HXT2 mRNA enrichment under glucose‐shift conditions but not in logarithmically growing cells.

• A
  Cells were treated with 100 μM 1,10‐phenanthroline or the solvent control ethanol (EtOH) for 1 h. Scale bar represents 10 μm.
• B
  Quantification of (A). Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. At least 150 cells per condition from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test
• C–E
  Increased transcription of HXT2 per se does not lead to its enrichment in the bud. (C) Exchanging the 5′‐UTR with either the ADH‐ or GPD promoter leads to the overexpression of HXT2. Quantification of HXT2 expression compared to WT with qPCR. The mean of three independent experiments is plotted; bars represent standard deviation. (D) FISH experiments revealed that HXT2 mRNA is still retained in the mother cell after MAT under glucose‐rich conditions when the native 5′‐UTR is swapped for the ADH or GPD promoter. (E) Quantification of (D). Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. 150 cells per condition and cell‐cycle stage from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test. **P < 0.05 in a two‐tailed, unpaired t‐test.

Source data are available online for this figure.

HXT2 mRNA is associated with stalled ribosomes in the mother cell

Next, we asked whether these changes in mRNA localization were connected to translation. To this end, we treated cells with two translational inhibitors. Cycloheximide inhibits the elongation cycle during translation and therefore the mRNA is covered with stalled ribosomes (“polysomes”), while verrucarin A is a translation initiation inhibitor, leading to accumulation of monosomes and “free” RNA. In both cases, we observed a FISH signal reduction in the bud after MAT (Fig 2A). In addition, we noticed that the overall FISH signal was stronger in cycloheximide‐treated cells and somewhat weaker in the presence of verrucarin A, when compared to control (Fig 2B), indicating that HXT2 mRNA is stabilized by binding to ribosomes. To test this hypothesis, we performed qPCR. Indeed, the HXT2 mRNA levels were slightly increased in the presence of cycloheximide (Fig 2C). Conversely, HXT2 mRNA levels were reduced when translation initiation was attenuated using the temperature‐sensitive initiation factor prt1‐1 mutant (Fig 2C). Our data imply that HXT2 mRNA requires ribosome association for protection from decay and suggests that HXT2 mRNA is ribosome‐associated in the mother cell early in the cell cycle. Since we detected Hxt2 protein only much later in the cell cycle, we infer that the ribosomes on HXT2 mRNA would be stalled early in the cell cycle and that this block would be released at a later time point.

Figure 2. Active translation is important for HXT2 mRNA release to the bud.

• A, B
  Inhibiting translation by either applying Cycloheximide (CHX) or Verrucarin A (VerruA) leads to the retention of HXT2 mRNA in the mother cell after MAT. FISH of CHX or VerruA‐treated cells. Chloro: chloroform; solvent control for VerruA; NT: not treated; control for CHX.
• C
  CHX treatment leads to increased transcript stability. Quantitative PCR of HXT2 mRNA in CHX‐treated cells in comparison with prt1‐1 mutant cells. The data represent 3 independent experiments and the standard deviation is given.
• D
  Translation of HXT2 mRNA is more active late in the cell cycle. The HXT2 mRNA level bound to either di‐/trisomes and polysomes (≥ 4 ribosomes) over monosomes was determined by extraction of mRNA from cells elutriated and harvested at G1/S or M‐phase followed by qPCR. At G1/S similar levels of the HXT2 mRNA were found in the different ribosome fractions, whereas in M‐phase more mRNA was bound to polysomes, suggesting a more active translation in M‐phase. A single experiment is shown. Two additional experiments are provided in Fig EV1.
• E
  The localization of HXT2 mRNA is independent of the signal recognition particle (SRP). Deletion of SRP54 does not affect the retention in the mother cell or the equal distribution of HXT2 mRNA.

Data information: Quantification of the FISH experiments in (A) and (E) was performed as in Fig 1C. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. 150 cells per condition and cell‐cycle stage from three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test. Scale bars in (B) correspond to 5 μm.Source data are available online for this figure.

To test this notion, we performed polysome profiling followed by qPCR of the different ribosome fractions from cell population that were synchronized by elutriation to be in G1/S or M‐phase of the cell cycle. In G1/S, HXT2 mRNA was present in similar levels in monosomes, disomes, trisomes, and polysomes (Figs 2D and EV1E). In M‐phase, an increase of HXT2 mRNA was detected in the di/trisome and polysome fractions, consistent with higher translation rates of HXT2 mRNA in mitosis compared to G1/S phase.

An HXT2 mRNA‐ribosome complex is retained in the mother cell in an SRP‐independent manner

Our data suggest that HXT2 mRNA is part of a translationally inactive ribosomal complex that is present in the mother cell before entry in mitosis. Since HXT2 mRNA is apparently synthesized in M/G1, one possible explanation for the mother cell localization is that HXT2 mRNA becomes ribosome‐associated and translation is initiated in G1. The Hxt2 protein contains twelve transmembrane domains (Kasahara et al, 2006) and hence must be co‐translationally translocated into the ER. The transfer of a translating ribosome from the cytoplasm to the ER membrane is mediated by the signal recognition particle (SRP; Hann & Walter, 1991). Thus, it is conceivable that during HXT2 mRNA translation, when the nascent polypeptide chain emerges from the ribosome, the HXT2 RNP is transferred to the ER by SRP, where the translation might be stalled. To test this possibility, we examined the HXT2 mRNA localization in a ∆srp54 mutant, in which SRP function and translocation into the ER are strongly impaired (Hann & Walter, 1991). We did not observe any difference in the distribution of HXT2 mRNA between wild‐type and ∆srp54 cells (Fig 2E), making it unlikely that SRP is actively involved in restricting HXT2 mRNA localization in the mother cell.

Loss of the mRNA binding protein Scp160 causes enrichment of Hxt2 mRNA in the bud after mitosis

The results presented above imply that translational efficiency might be key to the retention of HXT2 mRNA in the mother cell. The polysome‐associated mRNA binding protein Scp160 is ER‐localized and is involved in translational efficiency (Sezen et al, 2009; Hirschmann et al, 2014; Weidner et al, 2014). Therefore, we tested whether SCP160 deletion would affect HXT2 mRNA localization. While the localization was not altered early in the cell cycle, HXT2 mRNA unexpectedly was enriched in the bud after MAT (Fig 3A). ∆scp160 cells rapidly become polyploid (Wintersberger et al, 1995). To ensure that ploidy had no effect on the mRNA localization, we used a strain in which Scp160 expression is dependent on doxycycline (Hirschmann et al, 2014). Acute depletion of Scp160 also promoted HXT2 mRNA enrichment in the bud (Fig 3B + Doxy, Fig EV3A). Likewise, deleting 2 or 4 of the C‐terminal KH domains of Scp160 (Baum et al, 2004) was sufficient for HXT2 mRNA bud enrichment (Fig 3B). Thus, loss of Scp160 function promotes enrichment of HXT2 mRNA in the bud after anaphase. However, Scp160 does not appear to bind HXT2 mRNA efficiently (Fig EV3B and C), making it unlikely that Scp160‐HXT2 mRNA interaction contributes to HXT2 mRNA localization.

Figure 3. Loss of the Scp160/Bfr1/Asc1 complex and PKA causes enrichment of HXT2 mRNA in the bud after MAT.

1. Deletion of SCP160, ASC1, or BFR1 increases HXT2 mRNA signal in the bud. FISH experiment with deletions in various RNA binding proteins.
2. The effect of ∆scp160 on HXT2 mRNA localization is independent of its increased ploidy. Scp160 truncations that lack either the last two (∆C2) or the last four (∆C4) KH domains or a Tet‐off SCP160 construct, in which expression is blocked by the addition of doxycycline (Doxy) confirm the phenotype observed in (A).
3. Schematic depiction of the glucose responsiveness pathway.
4. Hyperactivation of PKA drives accumulation of HXT2 mRNA in the bud. Treatment of cells with forskolin (+Forsko) or using a ∆bcy1 strain, causes HXT2 mRNA to be bud‐enriched.
5. HXT2 mRNA enrichment upon refeeding is Ras2‐dependent.

Data information: Quantification of FISH experiments in (A, B, D and E) was performed as in Fig 1C. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. At least 150 cells per condition and cell‐cycle stage from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001; ***P < 0.0005 in a two‐tailed, unpaired t‐test.Source data are available online for this figure.

Figure EV3. HXT2 mRNA is enriched in the bud after glucose shift, independent of Scp160 or Eap1.

1. Protein levels of Scp160 under the control of a Tet‐off promoter markedly decrease when treated with 2 μg/ml Doxycycline (+Doxy) for 6 h. Immunoblot analysis. PGK1 serves as loading control.
2. GFP‐TRAP beads pull‐down Scp160‐GFP. Immunoblot of the Scp160‐GFP. PGK1 serves as a control.
3. Scp160 does not bind HXT2 mRNA efficiently. HXT2 mRNA is not enriched when pulling down Scp160‐GFP, while DHH1 mRNA, which served as a positive control, is strongly enriched. The mean of three independent experiments is plotted; bars represent standard deviation.
4. Deleting another component of the SESA complex does not have an influence on the localization of HXT2 mRNA as compared to Scp160 or Asc1. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud, and around 1 equal distribution. 150 cells per condition and cell‐cycle stage from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively.
5. Single molecule FISH (smFISH) shows the same distribution pattern for HXT2 mRNA under both, glucose‐rich and glucose‐shift conditions. The scale bar represents 10 μm.

Source data are available online for this figure.

The association of Scp160 with polysomes depends on Bfr1 and Asc1 (Lang et al, 2001; Baum et al, 2004). Deletion of either BFR1 or ASC1 showed a phenotype indistinguishable from scp160 mutants or depletion (Fig 3A). The phenotype was specific because deletion of two other RNA binding proteins involved in mRNA localization, Khd1 and Vts1, did not show HXT2 mRNA accumulation in the bud (Fig 3A).

Scp160 and Asc1 are part of the SESA complex, which functions in translational control of a subset of mRNAs (Sezen et al, 2009). Yet, deletion of another component of this complex, Eap1, had no effect on HXT2 mRNA localization, indicating that Scp160, Bfr1, and Asc1 act independently of the SESA complex in this process (Fig EV3D). Taken together, our results demonstrate that lack of Scp160, Bfr1, and Asc1 enhances asymmetric HXT2 mRNA distribution after MAT.

The HXT2 mRNA enrichment in the bud in scp160, bfr1, and asc1 mutants was rather unexpected and very surprising. Therefore, we decided to further explore the mechanism and function of HXT2 mRNA enrichment in the bud after MAT.

Increase of cAMP levels and PKA activity promote asymmetric distribution of HXT2 mRNA after metaphase/anaphase transition

Asc1 is the functional ortholog of mammalian RACK1 and was reported to function as a G‐protein β‐subunit coupled to glucose responsiveness (Zeller et al, 2007; Fig 3C). Since Hxt2 is a glucose transporter, it is conceivable that Asc1's function in repressing adenylate cyclase to keep cAMP levels low would prevent HXT2 mRNA accumulation in the bud. Conversely, in this scenario, raising cAMP levels should drive HXT2 mRNA accumulation in the bud. To test this idea, we treated cells with forskolin, an activator of adenylate cyclase. As expected, forskolin did not affect HXT2 mRNA localization early in the cell cycle (Fig 3D). However, we observed an increase in HXT2 mRNA signal in the bud comparable with that observed in the ∆asc1, ∆bfr1, and scp160 mutants. High cAMP levels cause the activation of PKA by binding to the inhibitory subunit Bcy1, which dissociates from PKA in the cAMP‐bound form (Fig 3C). We tested whether PKA signaling was involved in the HXT2 mRNA bud enrichment by deleting BCY1. In fact, HXT2 mRNA enrichment was even more pronounced in ∆bcy1 cells (Fig 3D), indicating that indeed activation of PKA was responsible for HXT2 mRNA enrichment in the bud after mitosis.

Rather than using mutants to up‐regulate cAMP/PKA signaling constitutively, we sought an approach to recapitulate transient cAMP production and hence transiently activate PKA. When starved cells are fed with glucose, they transiently increase cAMP levels (Jiang et al, 1998). We decided to use this regime to explore the changes in HXT2 mRNA localization upon transient PKA activation. We starved cells for 2 h and then shifted them to glucose‐rich medium for 30 min prior to FISH analysis. HXT2 mRNA was enriched in the bud after MAT (Fig 3E), consistent with the notion that the cAMP/PKA pathway is responsible for the enrichment of HXT2 mRNA in the bud. This enrichment was also confirmed by single molecule FISH (smFISH; Fig EV3E). The cAMP/PKA pathway can be activated through G‐protein‐coupled receptor activation or Ras (Jiang et al, 1998; Xue et al, 1998). To determine the upstream component of the adenylate cyclase, we deleted the receptor, GPR1, and the α‐subunit GPA2, and used a strain carrying the ras2 ^318S mutation (Jiang et al, 1998). While loss of the G‐protein‐coupled receptor branch did not interfere with HXT2 mRNA localization after refeeding, the ras2 ^318S mutant was defective in HXT2 mRNA enrichment in the bud after MAT (Fig 3E). We conclude that HXT2 mRNA enrichment in the bud after MAT is dependent on the Ras/cAMP/PKA signaling pathway. The strong enrichment of HXT2 mRNA could also at least partially depend on mRNA production. When we blocked transcription with 1,10 phenanthroline after glucose shift, we did not observe an HXT2 mRNA enrichment in the bud after MAT (Fig EV2A and B), indicating that de novo synthesis is required for the asymmetric mRNA distribution under these conditions.

Hexose transporter mRNAs are retained in the mother but not all of them are enriched in the bud when released from starvation

Seventeen HXT genes are encoded in the S. cerevisiae genome. We aimed to determine whether changes in mRNA localization are a common feature among HXTs. We concentrated on the 3 most important Hxts besides Hxt2: the low‐affinity transporters Hxt1 and Hxt3 and the high‐affinity transporter Hxt4 (Fig 4A). Hxt2 can adopt at least two conformations that have been proposed to reflect high‐affinity and low‐affinity transporter states (Reifenberger et al, 1997; Perez et al, 2005). Therefore, Hxt2 is special in that it cannot be clearly assigned to the low‐ or high‐affinity transporter group. HXT1, HXT3, and HXT4 mRNAs were each retained in the mother early in the cell cycle and equally distributed after mitosis (Fig 4B and C). However, HXT1, HXT3, and HXT4 mRNA were less responsive than HXT2 mRNA to high PKA levels in ∆bcy1 cells. HXT3 mRNA did not show any enrichment in the daughter cell (Fig 4D). In a ∆bcy1 strain, PKA is constitutively hyperactive. To investigate HXT mRNA localization in a more physiologically relevant context, we again shifted cells from starvation to glucose to induce a short cAMP peak, which in turn is sufficient to activate PKA. Under those conditions, HXT1, HXT3, and HXT4 mRNA did not show any enrichment in the daughter cell after mitosis, while HXT2 mRNA was still enriched (Fig 4E).

Figure 4. HXT mRNAs are retained in the mother before MAT, but only HXT2 mRNA is enriched in the bud after MAT upon refeeding.

1. Table of the glucose affinity of the four most important hexose transporters.
2. FISH of HXT1, 3, and 4 under glucose‐rich conditions. White arrowheads depict cells before and yellow arrowheads cells after MAT.
3. Quantification of (B).
4. Hyperactive PKA promotes enrichment of all HXTs after MAT.
5. HXT2 mRNA is the only HXT mRNA showing enrichment in the bud after MAT upon transient PKA activation.
6. Elutriation confirms enrichment of HXT2 mRNA in daughter cells upon refeeding. Daughters and mothers were separated through elutriation. qPCR was performed to determine the relative levels of different HXTs and ASH1 mRNA in both fractions.
7. Enrichment of HXT2 mRNA in the bud leads to higher Hxt2‐GFP transporter levels at the plasma membrane of the daughter cells. Daughter cells were enriched through elutriation. Quantification of the fluorescence intensity of the plasma membrane from live cell imaging from three independent experiments in which at least 50 cells were analyzed per condition and experiment. Western blot analysis confirms the higher amount of Hxt2‐GFP transporter protein present in daughter cells under glucose‐shift conditions

Data information: (C–E) Quantification of FISH experiments performed as described in Fig 1C. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud, and around 1 equal distribution. At least 100 cells per condition and cell‐cycle stage from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test. (F) The mean of three independent experiments is plotted; bars represent standard deviation. *P < 0.05 in a two‐tailed, unpaired t‐test. (G) The mean of three independent experiments is plotted. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test. Scale bar in (B); 5 μm.Source data are available online for this figure.

To corroborate our findings, we separated young, newborn daughters from mother cells through elutriation during which we switched from no glucose to glucose‐rich medium. We then extracted the total mRNA from the different cell populations and performed qPCR for HXTs. As a control, we used ASH1 mRNA, which is known to be enriched in daughter cells. ASH1 mRNA was about 2.3‐fold enriched in daughter cells, a value which is in good agreement with the 2–2.2‐fold increase reported previously (Takizawa et al, 2000). Among the HXT mRNAs, HXT2 was the only mRNA that accumulated in daughter cells (Fig 4F). Taken together, retention in the mother prior to entering mitosis appears to be a general feature among the most important hexose transporter mRNAs, while the enrichment in the bud late in the cell cycle after release from starvation is specific for HXT2 mRNA.

As a consequence of the asymmetric HXT2 mRNA localization, we would expect also the Hxt2 protein to be enriched in daughter cells upon glucose shift. We repeated the elutriation experiment with a strain in which Hxt2 was endogenously tagged with GFP. Indeed, Hxt2‐GFP was more abundant in daughters after glucose shift than in daughters from logarithmically growing cells as determined by mean fluorescence intensity of plasma membrane‐localized Hxt2‐GFP in live cells and immunoblot (Fig 4G). These results are consistent with the hypothesis that glucose‐shifted daughters can take up more glucose than logarithmically growing ones.

The spindle positioning protein Kar9 and nuclear segregation are essential for HXT2 mRNA bud enrichment under glucose‐shift conditions

Even though de novo HXT2 mRNA synthesis is required for bud enrichment (Fig EV2A and B), overexpression of HXT2 mRNA by itself was not sufficient to promote asymmetric mRNA distribution (Fig EV2C and D). Therefore, an additional layer of regulation must be required. It is conceivable that the enrichment of HXT2 mRNA in the bud late in the cell cycle could be dependent on three not mutually exclusive mechanisms: active transport of the mRNA into the bud, diffusion and retention, and selective degradation in the mother cell. P‐bodies are the primary site of mRNA decay in yeast, and the exonuclease Xrn1 is responsible for the degradation of most mRNAs (Muhlrad et al, 1994). Therefore, we determined HXT2 mRNA localization after shift from starvation to glucose containing medium in ∆xrn1 cells. We could not detect a significant difference in mRNA localization when compared to wild‐type cells (Fig 5A). Thus, localized mRNA decay is an unlikely mechanism for the asymmetric HXT2 mRNA localization. To distinguish between active and passive transport, we repeated the treatment of cells with benomyl and LatA under the glucose‐shift conditions. Both benomyl and LatA treatment abrogated the HXT2 mRNA bud enrichment (Fig 5B). In addition, inhibition of the Arp2/3 complex by CK‐666 inhibited HXT2 mRNA accumulation (Fig 5B). Our data suggest that actin and MTs both contribute to the asymmetric distribution of HXT2 mRNA. Accordingly, we sought a cellular process involving both MTs and actin. Spindle positioning and spindle pole body (SPB) movement to the bud is such a process (Miller et al, 1999). A key component in this process is the APC ortholog Kar9. Deleting KAR9 reduced HXT2 mRNA asymmetric localization to the bud after MAT upon refeeding (Fig 5C and D). In a ∆kar9 mutant, the old SPB is no longer preferentially segregated to the bud/daughter cell, and oftentimes both nuclei are retained in the mother cell after mitosis (Miller & Rose, 1998; Pereira et al, 2001). In large‐budded cells after MAT, we observed equal HXT2 mRNA distribution when mother and daughter received a nucleus. In contrast, in cells in which both nuclei remained in the mother, HXT2 mRNA was likewise retained (Fig 5C and D). Thus, Kar9 and potentially nuclear segregation into the bud play an essential role in asymmetric HXT2 mRNA distribution.

Figure 5. HXT2 mRNA enrichment in the bud depends on Kar9 and nuclear segregation.

1. RNA degradation is not essential for asymmetric HXT2 mRNA localization. FISH of wild type and ∆xrn1 cells.
2. The cytoskeleton contributes to HXT2 mRNA enrichment in the bud. Cells were treated upon glucose addition with either 120 mg/ml benomyl for 15 min, 30 mg/ml latrunculin A for 30 min, or 200 μM CK‐666 for 30 min. HXT2 mRNA was detected by FISH.
3. Asymmetric loading of HXT2 mRNA onto the nuclear envelope and enrichment in the bud requires Kar9 but not Bim1 function. HXT2 mRNA FISH with ∆bim1 and ∆kar9 strains. Arrows point to a nucleus with HXT2 signal.
4. Quantification of the daughter cell enrichment from data displayed in (C).
5. Quantification of the HXT2 mRNA signal in cells with two nuclei in the mother cell from the experiment depicted in (C). Percentage of cells with asymmetric HXT2 mRNA staining is displayed.

Data information: Quantification of the FISH experiments in (A, B, and D) was performed as in Fig 1C. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud, and around 1 equal distribution. At least 150 cells per condition from at least three independent experiments were analyzed. For the sample ∆kar9 with 2 nuclei/cell in (D), 50 cells were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. Quantification in (E) is based on three independent experiments. The bars represent standard deviation. *P < 0.05, ****P < 0.0001 in a two‐tailed, unpaired t‐test. Scale bars in (C) correspond to 5 μm.Source data are available online for this figure.

Kar9 is required for asymmetric distribution of HXT2 mRNA on the nuclear envelope

To corroborate our results, we deleted another component of the spindle positioning pathway, the EB1 homolog Bim1 (Lee et al, 2000). In ∆bim1 cells, in which one of the nuclei reached the bud, HXT2 mRNA asymmetric distribution was largely unaffected (Fig 5C and D). However, when both nuclei were retained in the mother, HXT2 mRNA was likewise retained. Our results are consistent with HXT2 mRNA requiring nuclear segregation into the bud for its asymmetric enrichment after glucose shift. Moreover, this process appears to be dependent on the spindle positioning pathway. Yet, the effect of ∆kar9 and ∆bim1 on HXT2 mRNA localization in mothers containing two nuclei was different. While in ∆bim1 cells HXT2 mRNA was preferentially localized to one of two nuclei, in ∆kar9 cells, the mRNA was not closely associated with a particular nucleus and often not even recruited to the nuclear envelope (Fig 5C and E). These results suggest that Kar9 is required for asymmetric loading of HXT2 mRNA on the nuclear envelope.

The nuclear pore components Nup2 and Mlp1/2 are involved in the asymmetric localization of HXT2 mRNA

We have shown above that PKA activation and nuclear segregation in the bud are prerequisites for the asymmetric distribution of HXT2 mRNA upon refeeding. Even though Kar9 is phosphorylated, this phosphorylation event has not been attributed to PKA action. We consequently hypothesized that PKA might phosphorylate a protein at the nuclear envelope or the endoplasmic reticulum, which in its phosphorylated form might contribute to asymmetric HXT2 mRNA localization. To test this hypothesis, we searched databases for ER/nuclear envelope localized PKA substrates (Table EV1). Within this list, we found two components of the nuclear pore complex (NPC), which were promising candidates for asymmetric localization of mRNA:Nup2, which is involved in NPC inheritance (Suresh et al, 2017) and the FG‐repeat containing central core component Nup53. Indeed, when we deleted NUP2, the asymmetric localization of HXT2 mRNA to the bud was dampened (Fig 6A and B). In contrast, a deletion mutant of NUP53 and its paralog NUP59 did not affect the asymmetric HXT2 mRNA distribution (Fig 6A and B). The S. cerevisiae NPC components Mlp1/2 were shown to be enriched at NPC insertion sites during anaphase at the old SPB that enters the bud (Ruthnick et al, 2017). Deletion of MLP1/2 resulted in a similar defect on HXT2 mRNA enrichment to ∆nup2 (Fig 6A and B). Our data suggest a contribution by NPCs to the enrichment of HXT2 mRNA in anaphase in the bud upon refeeding, presumably through the NPC inheritance pathway. If our assumption was correct, we should be able to detect HXT2 mRNA on the nuclear envelope. To this end, we marked the nuclear envelope with Nup84‐HA and performed FISH‐IF. HXT2 mRNA was detected on the nuclear envelope in the bud (Fig 6C).

Figure 6. NPC components aid the enrichment of HXT2 mRNA in the bud upon refeeding.

1. Deletion of NUP2 and MLP1/2 reduced HXT2 mRNA enrichment in the bud. FISH under glucose‐shift conditions with strains in which different NPC components were deleted.
2. Quantification of (A).
3. HXT2 mRNA is on the nuclear envelope after MAT upon glucose shift. FISH‐IF experiment with a probe against HXT2 mRNA and an antibody against the HA tag of the nuclear pore complex component Nup84‐HA. White arrows point to HXT2 mRNA signals on the nuclear envelope.
4. HXT2 mRNA travels on the nuclear envelope into the bud during mitosis after glucose shift. Stills of time‐lapse movies. HXT2 mRNA was visualized using the MS2‐GFP system. Under glucose‐shift conditions, HXT2 mRNA is mostly localized to the bud together with the nucleus. Under glucose‐rich conditions, HXT2 mRNA moves mostly independent of the nucleus. Cyan: Nup84‐mCherry; white arrowhead: HXT2 mRNA‐MS2 + MCP‐GFP.

Data information: Quantification in (B) was performed as in Fig 1C. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud, and around 1 equal distribution. At least 175 cells per condition from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test. Scale bars in (A, C, and D) correspond to 5 μm.Source data are available online for this figure.

To further corroborate our findings, we aimed to image movement of HXT2 mRNA during mitosis using an improved MS2‐GFP‐based system (Tutucci et al, 2018). HXT2 mRNA was associated with the nucleus as it entered the bud in mitosis after refeeding (Fig 6D, Movie EV1). Such a movement was not observed in cells that did not experience glucose withdrawal (Fig 6D, Movie EV2). Taken together, our data provide strong evidence for an involvement of the nuclear envelope in the asymmetric HXT2 mRNA distribution during mitosis.

HXT2 expression provides a growth advantage

Why would the yeast cell specifically enrich HXT2 mRNA into the bud following starvation? To gain insights into the biological function of this process, we used yeast strains in which all seventeen HXT genes had been deleted and which were kept alive by the expression of a single HXT gene under its endogenous promoter (Wieczorke et al, 1999; Youk & van Oudenaarden, 2009). Those strains were either grown to stationary phase or cells were scraped off directly from a plate on which they were in a quiescent state. Serial dilutions were dropped on rich media plates. Growth of cells was assessed 2‐3 days later. Cells expressing only HXT2 grew better than cells expressing either only HXT1, HXT3 or HXT4 on a wide range of glucose concentrations (Figs 7A and B, and EV4). Next, we tested whether HXT2 only strains would also do better when they were directly competing with the other HXT only strains in culture. To this end, we mixed stationary cultures of the 4 HXT only strains, collected a sample for the starting point, and let them grow overnight. Then, we collected another sample and performed qPCR on the DNA. Under direct competition for resources, the HXT2 only strain grew faster and outcompeted the other strains (Fig 7C and D). HXT2 mRNA localization is responsible for the better growth as an HXT2 deletion grew less well when coming out of quiescence (Fig 7E) Collectively, our data suggest that asymmetric HXT2 mRNA distribution provides a growth advantage.

Figure 7. HXT2 mRNA enrichment in the bud under glucose‐shift conditions provides a growth advantage to daughter cells.

1. Schematic depiction of the growth assays. Cells expressing only one hexose transporter at a time were picked from a plate, serially diluted and plated on fresh YPD plates.
2. HXT2 only strains grow faster than the other HXT only strain. Growth was assessed 2–3 days after incubation at 30°C.
3. Workflow for the competition assay presented in (D). O/N cultures of HXT‐only strains were diluted to the same OD₆₀₀ and pooled. A sample was taken (“start”), and the rest of the culture incubated O/N at 30°C. Cells were harvested (“end”), DNA was isolated from both samples, and the amount of HXT DNA was determined by qPCR.
4. Quantification of competition assay. qPCR with primers against either HXT1, 2, 3, or 4 was performed. HXT gene abundance was normalized against actin.
5. Strains that lack HXT2 grow slower when coming from quiescence compared to WT.
6. Tagging the HXT2 mRNA with U1A stem loops leads to its artificial enrichment in the bud in 96% of cells independently of the cell cycle. Arrows point to examples of HXT2‐U1A mRNA in the bud early and late in the cell cycle.
7. Cells that always enrich HXT2 mRNA in the bud have a growth advantage in the first hours when coming from quiescence compared to WT or cells that only express HXT2.
8. Daughter cells do not retain a memory of the previous stress situation encountered by their mothers. Cells were re‐fed after starvation and analyzed for enrichment of HXT2 mRNA via FISH at different time points.
9. Proposed model for the regulation of the asymmetric HXT2 mRNA distribution when cells come out of starvation or leave the quiescent state. For details, please see text.

Data information: (D) The results of three independent experiments are shown. Error bars represent standard deviation. Comparison of HXT levels at the beginning and the end of growth shows that cells expressing only HXT2 outgrow other HXT‐only cells in the range of 2–5‐fold. *P < 0.05 in a two‐tailed, unpaired t‐test. (E) Data from three independent experiments are plotted. The bars represent standard deviation. (G) Data from four independent experiments were plotted. The bars represent standard deviation. A two‐tailed unpaired t‐test was performed for T = 4 h. *P < 0.05, **P < 0.005. Quantification in (H) was performed as in Fig 1C. Values > 1 indicate a stronger signal in the mother, < 1 a stronger signal in the bud and around 1 equal distribution. At least 150 cells per condition from at least three independent experiments were analyzed. Boxes represent the interquartile range from the 25^th to the 75^th percentile with the median. Whiskers represent the 10^th and 90^th percentile, respectively. ****P < 0.0001 in a two‐tailed, unpaired t‐test.Source data are available online for this figure.

Figure EV4. At different glucose concentrations, HXT2 only cells still grow faster compared to other HXT only cells.

Growth assay as described in Fig 7A on plates with indicated glucose concentrations.Source data are available online for this figure.

Bud‐localized HXT2 mRNA provides a growth advantage

We sought a way to demonstrate that the asymmetric localization of HXT2 mRNA is the determining factor for the growth advantage. When we tagged HXT2 mRNA with U1A stem loops for live‐cell imaging (Brodsky & Silver, 2000), we noticed that the HXT2 mRNA was asymmetrically localized to the bud irrespective of the cell‐cycle stage and independent of the U1A binding protein (Fig 7F; in > 80% of cells with a FISH signal). This was a rather unexpected finding, and the underlying mechanism remains obscure. However, this finding provided a tool, which allowed us now to test directly, whether HXT2 mRNA is critical for the growth advantage. We compared the growth curves of the HXT2 only, the HXT2‐U1A only, and the isogenic wild‐type expressing all seventeen glucose transporters (Fig 7G). When coming out of quiescence, the HXT2 only strain grew as fast as the isogenic wild type, suggesting Hxt2 as sole transporter is sufficient to restart growth efficiently. However, when the HXT2 mRNA was partitioned effectively into bud with the U1A tag, those cells grew even faster than the wild type in the first cell cycle. Therefore, the asymmetric HXT2 mRNA enrichment in the bud is instructive of growth advantage of newborn daughters in the next cell cycle.

The cell does not retain a long‐lasting memory for HXT2 mRNA localization

We envisage that the increase in HXT2 mRNA and Hxt2 protein in newborn daughters will allow those daughters to take up more glucose and therefore grow faster than their counterparts that do not increase HXT2 expression. In this scenario, the daughters with high HXT2 will also reach the critical size earlier to enter the cell cycle and become a mother. We wondered whether asymmetric HXT2 mRNA distribution would be retained in future generations in order to not lose the competitive edge. In other words, we asked whether high HXT2 daughters remember that they received more HXT2 and will therefore also asymmetrically distribute the HXT2 mRNA during the next cell division, even though the daughters have not experienced nutrient limiting conditions themselves? Alternatively, there is no memory of the previous events transmitted to the daughters because entering the cell cycle faster comes at an expense in that errors that occurred during previous cell cycle may not be corrected. The results presented above would be more consistent with the latter possibility. To corroborate these data, we performed a glucose‐shift experiment and took samples at different time points after the shift. FISH revealed that a significant enrichment in the daughter was observed only up to 2 h after shift (Fig 7H). Considering the lag phase upon resuming the cell division cycle, these results indicate that a newborn daughter does not retain a memory in terms of HXT2 mRNA localization of its mother's starvation experience, when it becomes a mother itself.

Discussion

We have identified a mechanism in S. cerevisiae by which the mRNA of the glucose transporter Hxt2 is specifically enriched in the daughter cell when cells are coming out of quiescence or starvation. We assume that this enrichment provides a growth advantage to these young daughters as they are potentially able to take up more glucose and thus increase in size faster than daughters with less Hxt2. This mechanism allows those daughters to outcompete cells that cannot mount a similar response and provides an ecological advantage. This partitioning is not shared by other major glucose transporters, underscoring a unique role of Hxt2. Hxt2 differs from the other glucose transporters in that it can act as both high‐affinity and low‐affinity transporter (Perez et al, 2005), rendering it particularly suitable for such a regulation. We demonstrate that accumulation of HXT2 mRNA is dependent on the Ras/cAMP/PKA pathway and the nuclear envelope. We speculate that the short cAMP peak upon refeeding provides sufficient PKA activity to phosphorylate targets on the nuclear envelope such as Nup2 (Fig 7I). Our data are consistent with a model in which PKA phosphorylates the NPC component Nup2. This would then in turn allow the piggyback of HXT2 mRNA into the daughter cell. Whether this piggyback is through the NPC inheritance pathway, direct interaction with Nup2 or correlates to NPC function remains to be determined. However, we favor at this point the involvement of the NPC inheritance pathway rather than NPC function or direct binding to Nup2 based on the observation that deletion of another NPC PKA target, the FG‐repeat protein Nup53 and its paralog Nup59, did not affect HXT2 mRNA localization. These data argue against NPC function per se being essential for the asymmetric mRNA localization. In contrast, Mlp1/2, which are NPC components that are enriched in the daughter cell during mitosis, are likewise contributing to the accumulation of HXT2 mRNA in the daughter cell (Ruthnick et al, 2017). Importantly, NPCs have been shown to accumulate around the yeast centrosome (SPB) presumably through a mechanism involving Mlp2 and Nup2 (Winey et al, 1997; Niepel et al, 2005; Suresh et al, 2017).

Our data provide strong evidence that movement of the nucleus into the bud is critical for asymmetric enrichment of HXT2 mRNA. Both actin and microtubules as well as the yeast EB1 and APC homologs, Bim1 and Kar9, respectively, are essential for this process. Yet, it appears that Kar9 has an additional role in that it may help to generate the asymmetry on the nuclear envelope. While in a ∆bim1 background, HXT2 mRNA was present preferentially only on one of two nuclei that were retained in the mother, in ∆kar9 this asymmetry was abolished. During spindle formation early in the cell‐cycle Kar9 becomes enriched on the old SPB, which will move into the bud during mitosis in a process dependent on both actin and microtubules (Liakopoulos et al, 2003; Maekawa & Schiebel, 2004; Cepeda‐Garcia et al, 2010). During this process, Kar9 translocates from the SPB to the MT (+) ends in the bud by binding to Bim1 (Huls et al, 2012). Therefore, it is likely that Kar9 asymmetry is instructive for HXT2 mRNA localization. However, the underlying mechanism that controls Kar9 asymmetry remains unclear and highly controversial (Juanes et al, 2013).

We also found that HXT2 transcription contributed to the asymmetric enrichment in the bud upon refeeding. However, transcription per se was not sufficient because HXT2 overexpression did not influence mRNA localization. The HXT2 transcription during mitosis in the glucose‐shift experiments could potentially be driven by the same Ras and cAMP‐dependent peak of PKA activity that drives asymmetric mRNA localization. This scenario would elegantly couple gene transcription to asymmetric distribution of its mRNA.

We generally observed that HXT mRNA was retained in the mother cell before MAT. We assume that the mRNAs are engaged with ribosomes and membrane bound. This retention was also coupled, at least for HXT2, with translational stalling. The mechanism by which this is achieved remains unclear but does not appear to involve SRP. We suppose that at least a part of the stalling of the ribosomes is due to some cis‐elements on the mRNA itself. However, since this is a regulated process, trans‐acting factors are likely to play a role as well. How these stalled ribosomes would escape ribosome quality control also needs to be investigated. Further studies are needed to elucidate the mechanism of HXT2 retention.

We propose a model in which the asymmetric HXT2 mRNA distribution upon refeeding allows the daughters to uptake more glucose and therefore grow faster than daughters with less HXT2 mRNA. Since cell size is one of the hallmarks for entering a new cell cycle, the faster growing daughters will start a new division cycle before the others. The faster entry into the cell cycle appears to be sufficient to outcompete the slower growing cells. Yet, the faster growing daughters do not retain an obvious memory of the glucose limitation and refeeding because they do not provide more HXT2 mRNA for their daughters. This finding is consistent with the observation that a temporal cAMP/PKA activity peak is sufficient to drive the mRNA enrichment, and hence, a memory may not be needed. Moreover, it may not be advantageous for the cells to always enter the cell division cycle as fast as possible. Before passing START, the cells usually check whether the last cell division was successful on multiple levels and aim to repair defects that may have occurred. Therefore, the time between two cell division cycles needs to be balanced between speed and fidelity.

Maintaining such a balance is in contrast to what likely happens in cancer cells. In numerous cancer types, glucose transporters (GLUTs), in particular GLUT1 and GLUT3, are up‐regulated and often linked to oncogenic Ras (Adekola et al, 2012). Like HXTs in yeast, GLUTs are localized in polarized fashion in epithelial cells. Moreover, GLUT3 expression is up‐regulated by cAMP in a breast cancer cell line (Meneses et al, 2008) and increased GLUT1 level has been correlated with poor prognosis in colorectal carcinomas (Sakashita et al, 2001). Thus, cancer cells may employ a similar strategy in terms of accelerating growth, but lack check points to monitor damage that occurred in the previous cell cycle. This may be different in cancer stem cells (CSCs), which are metabolically more similar to stem cells in that they can use glycolysis over OXPHOS and are often in hypoxic environments (Wong et al, 2017). At least in embryonic stem cells GLUT1 and GLUT3 expression levels are correlated to maintaining pluripotency (Wu et al, 2017; Zhang et al, 2017). Moreover, GLUT1 is asymmetrically localized in dividing lymphocytes and becomes enriched in the differentiating daughter (Chen et al, 2018). Therefore, the controlled expression of specific glucose transporter at the plasma membrane is an important and conserved cellular feature to maintain competitiveness in difficult environments.

Materials and Methods

Yeast strains and growth conditions

Standard genetic techniques were employed throughout (Sherman, 1991). Unless otherwise noted, all genetic modifications were carried out chromosomally. Chromosomal tagging and deletions were performed as described (Knop et al, 1999; Gueldener et al, 2002; Janke et al, 2004). For amplification of C‐terminal tagging by chromosomal integration pYM (Knop et al, 1999) and for deletions, pUG plasmids (Gueldener et al, 2002) were used. For N‐terminal tagging or promoter exchange by chromosomal integration, pYM‐N plasmids were used (Janke et al, 2004). Primers, strains, and plasmids used in this study are listed in [Link], [Link], [Link]. The plasmids pMS449 and pMS450 carrying the ∆C2 and ∆C4 truncation of Scp160 were a kind gift from M. Seedorf (ZMBH Heidelberg, Germany). The plasmid pRJ1463 bearing Scp160 under the control of a Tet‐off promoter was a kind gift from Ralf‐Peter Jansen (Jansen et al, 2014). For glucose‐rich conditions, yeast cells were grown in either YPD (1% w/v Bacto yeast extract, 2% w/v Bacto‐peptone, 2% w/v dextrose) or selective medium (prepared as described; Kaiser et al, 1994) at 30°C, 200 rpm. For glucose‐shift conditions, cultures were first grown overnight in YPD, then washed two times in YP media without glucose diluted to an OD₆₀₀ of 0.3, and further grown for 2 h in YP media without glucose. After this incubation, glucose was added to a final concentration of 2%. Generally, yeast cells were harvested at mid‐log phase with an OD₆₀₀ of 0.4–0.8. OD₆₀₀ as determined with a UltroSpec 3100 pro Spectrophotometer (GE healthcare).

Live cell imaging

Yeast cells were grown in YPD or HC‐Leu (2% dextrose, 1× adenine) to early log phase and either analyzed directly or first starved without dextrose for 2 h then supplemented with 2% dextrose. The cells were taken up in HC complete or HC‐Leu medium and immobilized on 1% agar pads. Fluorescence was monitored with an Orca Flash 4.0 camera (Hamamatsu) mounted on an Axio M2 imager (Carl Zeiss) using a Plan Apochromat 63x/NA1.40 objective and filters for mCherry and GFP. ZEN software version 2.3 (blue edition) was used to acquire images (Carl Zeiss). Further image processing was performed using Fiji (Schindelin et al, 2012). All pictures from the same experiment were treated equally.

Live imaging fluorescence intensity measurement

Plasma membrane of Hxt2‐GFP expressing cells was outlined using the segmented lines tool in Fiji. Line width was set to 2. The mean fluorescence intensity of 50 cells per condition and experiment from three independent experiments was measured.

Fluorescence in situ hybridization and FISH combined with immunofluorescence

Fluorescence in situ hybridization and FISH‐IF were performed as described previously (Takizawa et al, 1997; Kilchert & Spang, 2011). For IF, we used as primary antibodies anti‐GFP (Torrey Pines, rabbit, polyclonal, 1:400) and anti‐HA (Covance, mouse, monoclonal, 1:400) and as secondary antibodies Alexa Fluor 488 goat‐anti‐rabbit or anti‐mouse, respectively (Invitrogen, polyclonal, 1:400). Images were acquired with an Axiocam MRm camera mounted on an Axioplan2 fluorescence microscope using a Plan Apochromat 63x/NA1.40 objective and filters for eqFP611, DAPI, and GFP. AxioVision software 3.1–4.8 was used to process images (Carl Zeiss).

FISH quantification—Image processing and analysis

Images were acquired on a Zeiss Axioplan2 microscope as described above. Segmentation and analysis were performed using a custom macro for Fiji (available upon request) carrying out the following steps: Based on the red (FISH) channel, the CLAHE (Zuiderveld, 1994) filter was executed to increase the local contrast with the goal of revealing enough contrast for being able to distinguish all cell bodies from the background. Parameters for the CLAHE filter were used with their defaults as provided by the plug‐in with a block size of 127, a default slope of 2 and 256 histogram bins. After filtering, the local thresholding method as described by Phansalkar et al (2011) was used to create a binary image marking the cell bodies (thresholding radius: 50, object size minimum: 50). To reduce artifacts in the segmentation, the binary image was then processed with Fiji's binary “Fill Holes” operation, followed by a “Close” operation. The resulting binary mask was then split into individual masks per cell by using the classical “Watershed” separation. As a last step for the segmentation, individual ROIs (regions of interest) were created from this binary mask. After creating the ROIs, they were used to quantify intensities in the red (FISH) channel by measuring the mean gray value per round cell shape. With those results, fluorescence intensities for mothers and their buds were determined separately and finally the ratio of fluorescence intensity in the mother over bud was calculated. At least 50 mother cells and their corresponding buds were counted per condition and experiment from at least three biologically independent experiments.

Elutriation

Elutriation was performed in a Beckman elutriation system (Avanti J‐26 XP centrifuge combined with a JE‐5.0 elutriator rotor and a 200 ml elutriation chamber, as well as standard elutriation accessories) according to (Marbouty et al, 2014) with modifications. For glucose‐shift experiments, first a colony was inoculated in 500 ml YPD medium and incubated overnight at 30°C. The next day, the cells were washed 2× with YP medium without glucose, then diluted to an OD₆₀₀ ~1 in 2 × 1 l fresh YP medium without dextrose, and further grown for 2 h. Dextrose was added to a final concentration of 2%, and the cultures were incubated for another 30 min at 30°C. The elutriation chamber was filled with the cell culture at a flow rate of ~25 ml/min and a rotor speed of 1,280 g at 4°C. The flow rate was gradually increased until the cells reached the top of the chamber. Equilibrium was allowed to settle in the chamber for 1 h while switching medium to PBS or YPD depending on the experiment. The flow through was checked for escaping cells by OD₆₀₀ measurement. The flow rate was carefully increased by increments of 2 ml/min until the OD₆₀₀ raised above 0.05, at which point cells were collected. The budding index of the recovered cells was checked by microscopy and only daughter cell fractions with less than 5% budded cells were used. After collection of the daughter cells, the centrifuge was stopped and the remaining cells were collected, providing the mother cell fraction. The fractions were spun down. The pellets were frozen in liquid nitrogen and stored at −80°C.

Polysome profiling

After elutriation, 40–50 OD₆₀₀ of young daughter cells were grown in YP 2%D at 30°C and 200 rpm until they started to bud for the G1/S‐fraction or until they were large‐budded for the M‐phase fraction. Cycloheximide was added to a final concentration of 100 μg/ml, and the flask was manually shaken for 1 min. The cells were filtered through a 0.45‐μm cellulose nitrate filter using a vacuum pump and once washed with 10–15 ml lysis buffer (20 mM Tris–HCl, pH 7.5, 100 mM NaCl, 10 mM MgCl₂, 1% Triton X‐100, 0.5 mM DTT, 100 μg/ml cycloheximide). Cells were scraped off of the filter and snap‐frozen in liquid N₂‐filled 50‐ml Falcon tubes. The cell pellets were lysed with a freezer mill (SPEXSamplePrep 6875, 2 cycles, 5 CPS). 0.5 ml of ice‐cold lysis buffer was added to the cell powder and thawed in a 30°C waterbath. The extract was centrifuged for 3 min at 3,000 g and 4°C. The supernatant was transferred to a pre‐chilled Eppendorf tube and centrifuged again at 10,000 g for 5 min. Ten A260 U of the supernatant were loaded on a 7–47% sucrose gradient. The gradient was centrifuged for 3 h at 4°C, 209,627 g in a Sorvall rotor (TH‐641) with brakes turned off. Fractionation was carried out with a Brandel BR‐188 Density Gradient Fractionating System (sensitivity of UV detector at 1, pump speed: 0.75 ml/min). 400 μl for each fraction was collected every 32 s in tubes. SDS was added to a final concentration of 1%. The tubes were snap‐frozen in liquid N₂ and stored at −80°C. RNA was isolated first by isopropanol precipitation and followed by hot phenol extraction. Finally, mRNA levels were analyzed by quantitative RT–PCR.

Pull‐down assay

100 OD₆₀₀ of logarithmically growing cells were resuspended in 2 ml of lysis buffer (10 mM Tris ‐HCl pH 7.5; 150 mM NaCl; 0.5 mM EDTA; 0.2% NP‐40; Roche protease inhibitor cocktail; RNasin). 200 μl of 150–212 μm and 200 μl of 450–600 μm acid‐washed glass beads were added per 1 ml of lysis buffer. Cells were lysed with a Fast prep (5 × 45 s, 4°C). Tubes were centrifuged 5 min at 1,300 g at 4°C. Supernatant was spun 10 min at 20,800 g at 4°C. Supernatant was mixed with 40 μl GFP‐Trap agarose beads (gta‐100 Chromotek) and incubated under inversion for 1–2 h. After thoroughly washing with wash buffer (10 mM Tris/HCl pH 7.5; 150 mM NaCl; 0.5 mM EDTA; RNasin), protease K (1.2 mg/ml) was added, and incubated beads at 55°C for 30 min. RNA was extracted, and mRNA was quantified with qPCR as described below.

Total RNA isolation

Total RNA was isolated from yeast essentially as described (Schmitt et al, 1990). Briefly, 10 OD₆₀₀ of yeast cells were harvested after different treatments. Cells were resuspended in 1 ml AE buffer (50 mM NaOAc pH 5.2, 10 mM EDTA), 100 μl 20% SDS and 1 ml PCI (125:24:1, pH 4.4) were added, and the tube vortexed for 10 s. After incubation for 10 min at 65°C, tubes were chilled in liquid nitrogen for 30 s, thawed, and then centrifuged (2 min, 16,000 g, RT). The upper aqueous phase was transferred to a fresh tube, and 1 ml PCI was added and vortexed 10 s. After phase separation by centrifugation, the upper aqueous phase was transferred to a fresh tube and 1/10 vol. of 3 M NaCl pH 5.2 was added. The RNA was precipitated with 1 vol. 100% EtOH at −80°C for at least 1 h. The precipitate was collected (30 min, 16,000 g, 4°C), washed with 70% EtOH, and again centrifuged. The pellet was resuspended in 200 μl RNase‐free water.

Quantitative RT–PCR

The 0.5–1 μg of total RNA was reversely transcribed with the Transcriptor reverse transcriptase kit (Roche, Cat# 03531287001), oligo‐dTs, and random hexamers. The mRNA levels were analyzed by SYBR green incorporation using the ABI StepOne Plus real‐time PCR system (Applied Biosystems). Primers used in qRT–PCR are listed in Table EV3.

Drop assays

Colonies were either directly taken from plate or grown overnight in liquid YPD medium to stationary phase (OD₆₀₀ ≥ 9). After adjusting to equal cell concentrations (OD₆₀₀~0.3), five serial dilutions (1:5) were dropped onto YP plates supplemented with either 0.1, 0.5, 2, or 4% glucose using a “frogger” stamp (custom‐built). Plates were incubated for 2–3 days at 30°C and photographed for documentation.

Growth test

Colonies were scraped from YP 2%D plates (stored at 4°C) and diluted in H₂O to OD₆₀₀~0.2. Cells were further grown in YP 2%D at 30°C, 200 rpm. Every 1–2 h, the OD₆₀₀ was measured.

Actin staining

Actin cytoskeleton staining was performed essentially as described previously (Adams & Pringle, 1991). Cells were grown overnight and fixed with 4% formaldehyde for 30 min at RT under gentle agitation. Cells were washed twice with PBS containing 1 mg/ml BSA (3 min, 1,000 g, RT). The cell pellet was resuspended in 25 μl PBS containing 1 mg/ml BSA and 5 μl of rhodamine‐phalloidin (Molecular Probes, 300 U/1.5 ml MeOH). After incubation for 1 h at RT in the dark, cells were washed three times and resuspended in 500 μl PBS containing 1 mg/ml BSA. An aliquot was allowed to settle for 30 min on polyethyleneimine‐treated multi‐well slides. The slides were washed briefly in PBS, Citifluor AF1 was added, and the coverslips were sealed with nail‐polish. Slides were stored at −20°C. Rhodamine fluorescence was observed by epifluorescence microscopy using the Cy3 channel on an Axioplan 2 fluorescence microscope from Zeiss. Pictures were taken with an Axiocam MRm camera using AxioVision software. Image processing was performed with Fiji.

Western blotting

Nine milliliters of a mid‐log grown culture was taken, immediately treated with cold trichloroacetic acid (10% final concentration), and incubated on ice for at least 5 min. Yeast extracts were prepared as described (Stracka et al, 2014). The protein concentration was determined using the DC Protein Assay (Bio‐Rad), and the total lysate was analyzed by SDS–PAGE. The primary antibody was rabbit anti‐Scp160 (Weidner et al, 2014, 1:1,000); the secondary antibody was goat‐anti‐rabbit, HRP‐conjugated (Pierce, polyclonal, 1:15,000). Enhanced Chemiluminescence solution (ECL; GE Healthcare) was used for detection.

PKA target prediction

To identify potential PKA targets, we intersected subcellular localization annotations and known phosphopeptides from yeast. The localization set of interest includes all UniProt entries for S. cerevisiae with at least one localization annotation of “endoplasmic reticulum”, “nucleus outer membrane”, or “nucleus envelope” (The UniProt C, 2017). Yeast phosphopeptides were identified from the ISB Library of the PeptideAtlas Project (Desiere et al, 2006) “The PeptideAtlas Project”, Nucleic Acids Research 34, D655‐D658]. Custom Python scripts were used to identify exact matches between phosphopeptides of the yeast PeptideAtlas and the protein sequences of the UniProt entries filtered by localization.

Author contributions

TS, SH, and AS conceived the study. TS and NM performed the experiments. SH provided preliminary data. NE wrote the macro for image analysis. GF was in charge of the PKA target prediction. TS and AS analyzed the data. AS wrote the manuscript. All authors were involved in editing of the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Table EV3

Table EV4

Movie EV1

Movie EV2

Source Data for Expanded View

Review Process File

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

Source Data for Figure 7

The EMBO Journal (2019) 38: e100373