Pleiotropy of the de novo-originated gene MDF1

MDF1 is a young de novo-originated gene from a non-coding sequence in baker's yeast, S. cerevisiae, which can suppress mating and promote vegetative growth. Our previous experiments successfully demonstrated how Mdf1p binds to the key mating pathway determinant MATα2 to suppress mating. However, how Mdf1p promotes growth and fulfills the crosstalk between the yeast mating and growth pathways are still open questions. Thus, the adaptive significance of this new de novo gene remains speculative. Here, we show that Mdf1p shortens the lag phase of S. cerevisiae by physically interacting with SNF1, the governing factor for nonfermentable carbon source utilization, and thereby confers a selective advantage on yeasts through the rapid consumption of glucose in the early generational stage in rich medium. Therefore, MDF1 functions in two important molecular pathways, mating and fermentation, and mediates the crosstalk between reproduction and vegetative growth. Together, our results provide a comprehensive example of how a de novo-originated gene organizes new regulatory circuits and thereby confers a selective advantage on S. cerevisiae to allow exquisite adaptation to the changing environment.

T he origination of new genes provides pivotal genetic novelties for adaptive innovation of organisms [1][2][3] . Among all of the molecular mechanisms of gene origination, a number of protein-coding genes that have originated de novo from non-coding sequences have been shown to reshape the landscape of molecular evolution [4][5][6] . In particular, we previously reported that the de novo gene MDF1 from baker's yeast plays a key role in regulating the yeast mating pathway by binding to MATa2 7 . MDF1 was originally referred to as FYV5, performing a putative function in response to the K1 killer toxin 8 . Interestingly, we also observed that MDF1 can promote growth, though the detailed mechanism of growth promotion was beyond the scope of our previous study 7 . The dual functions of MDF1 in two pathways (mating and vegetative growth) indicate a pleiotropic role for this de novo gene. Sexual reproduction benefits a species by eliminating deleterious mutations through meiotic recombination, but mating is energetically costly for vegetative growth 9 . It has been shown that a growth-rate advantage can be gained by losing signaling at multiple points in the mating pathway 10 . However, there is little information available regarding the intrinsic molecular mechanism accounting for the trade-off of these two options. MDF1 provides a unique opportunity for studying how two conflicting pathways are coordinated to maximize the adaptive advantage.
In the present work, we filled in the missing piece of the MDF1 growth promotion puzzle and formed a comprehensive picture of how a de novo gene confers an evolutionary advantage by incorporating mating and growth pathways into one molecular network. We clearly demonstrated how MDF1 enhances the effect of the active glucose signaling pathway to achieve advantageous vegetative growth in a rich glucose environment. The complete description of the dual functions of MDF1 provides the first persuasive explanation for the dynamic crosstalk between mating and growth pathways from the aspect of a molecular mechanism.

Results and Discussion
Mdf1p shortens the lag phase of S. cerevisiae. Our previous results showed that Mdf1p is a newly originated key regulator of the mating pathway that acts by binding MATa2 and promoting vegetative growth in S. cerevisiae 7 . However, the molecular mechanism underlying how MDF1 promotes growth was beyond the scope of our previous study. In this study, to investigate this question in depth, we therefore measured the growth rate of yeast cells every 20 min until growth plateau was reached. Then, we compared OD values between a wild-type strain and an MDF1 strain in which expression was driven by the MDF1 promoter (Methods). Interestingly, in the first two hours, the MDF1 strain significantly outperformed the wild-type strain (Fig. 1A). Thereafter, the growth curves of the MDF1 and wild-type strains paralleled with each other (Supplementary Fig. 1A), indicating that the OPEN SUBJECT AREAS: EVOLUTION GENETICS growth acceleration induced by MDF1 may be restricted to the time window of the first two hours. Indeed, MDF1 began to be expressed at approximately 1 h and reached a peak at approximately 1.5 h, but the expression level of MDF1 was reduced after 2 h ( Supplementary  Fig. 1B). The fluctuation of the expression of MDF1 was synchronized with a significant increase in the OD at 80 min (Two-tails Ttest, P-value,0.001; Fig 1B) in the growth curve, implying a causal link between MDF1 and the growth rate. These results suggest that MDF1 mainly exerts its function in the lag phase when yeasts are transferred to fresh fermentative medium.
Mdf1p inhibits Snf1p to promote fermentation. When yeast cells are inoculated into fresh medium, they initially enter a brief lag phase during the first two hours, which is a crucial time for yeast cells to acclimatize to the new environment 11 . In the lag phase, the yeast cell experiences a delay or a pause of growth on the surface, but rapid adjustment of the yeast physiology to the fresh medium is in fact taking place to ensure optimal growth initiation 12 . The physiological and metabolic changes that occur during the lag phase are thought to prepare for the upcoming fermentation process because hundreds of fermentation-related genes are up-regulated in the genome 13 . Having established that MDF1 shortens the lag phase, we hypothesized that MDF1 may enhance fermentation. Indeed, there was no expression of MDF1 detected under non-fermentative conditions in our experiments (Supplementary Fig. 2A), which was consistent with our previous observation that the MDF1 strain presents no growth superiority in non-fermentable medium 7 . To confirm that MDF1 functions in fermentative medium, our next task was to reveal the molecular mechanism underlying the enhancement of fermentation by MDF1 during the lag phase. Co-IP assays were conducted using the MDF1-overexpressing strain to search for the factor(s) interacting with Mdf1p. Interestingly, a potential interaction partner, Snf1p, was repeatedly identified via MALDI-TOF/TOF (expect Pvalue ,1.3 3 10 210 ) (Supplementary Figure 2B). To verify the interaction between Mdf1p and Snf1p, the products of the Mdf1p Co-IP assay were tested using a Snf1p antibody in vivo. Western blotting revealed the existence of Snf1p in the Co-IP products (Figure 2A), which firmly demonstrated that Mdf1p interacts with Snf1p in the lag phase. The interaction was further validated in a yeast two-hybrid assay, in which Mdf1p was fused to the DNAbinding domain of Gal4 (DB) and Snf1p to the activation domain of Gal4 (AD) (Supplementary Figure 2C). Taken together, the results obtained demonstrate the interaction between Mdf1p and Snf1p.
Snf1p is a governing factor in the transition from fermentative to respiratory growth in S. cerevisiae 14,15 . Snf1p leads to efficient derepression of downstream glucose-repressed genes when glucose becomes limited 16 . Yeast cells thereby switch from fermentative growth to respiratory growth to utilize alternative carbon sources 17,18 . Given that Mdf1p facilitates fermentation, whereas Snf1p guides the non-fermentative pathway, we hypothesized that Mdf1p imposes negative regulation on Snf1p. To determine whether the Mdf1p-Snf1p interaction could inhibit the activity of Snf1p, we measured the expression levels of glucose-repressed genes, including ICL1, FBP1, PCK1, ACS1, and MLS1, which are activated by Snf1p 16 . All of the glucose-repressed genes displayed decreased expression levels in quantitative RT-PCR assays conducted in the MDF1 strain (Fig. 2B), suggesting that the binding of Mdf1p and Snf1p produces an inhibitory effect on glucose-repressed genes. Taken these findings together, we proposed that the newly evolved MDF1 gene integrated into the yeast glucose signaling pathway to speed up glucose utilization and fermentative growth via physically suppressing Snf1p. Fermentation as a preferred mode of resource consumption has particularly important significance in promoting the growth of yeasts. It has been shown that under aerobic conditions with an ample available carbon source such as glucose, baker's yeast predominantly conducts fermentation to generate energy for rapid cellular growth, though the yield of ATP is lower than during respiration 19,20 . The speed of resource intake matters much more than the efficiency of energy production when glucose is sufficient. Therefore, by accelerating glucose consumption, Mdf1p puts yeasts in an advantageous position in competing with other microbes present in the same environment.
MDF1 physically connects the growth pathway with the mating pathway. In our previous study, we revealed that Mdf1p was capable of suppressing yeast mating behavior by binding one of the yeast mating type determinants, MATa2, which would subsequently shut down the mating pathway 7 . In the present study, we discovered that Mdf1p could speed up vegetative growth through physically inhibiting the fermentation-suppressing factor Snf1p. Based on the evidence obtained thus far, we extended the previously proposed model 7 concerning the established functions of Mdf1p in the mating and growth pathways ( Figure 3): initially, especially after the recovery of haploid cells from growth arrest under unfavorable conditions, vegetative proliferation is advantageous for rapid resource consumption. Mdf1p enhances the effect of activating the glucose signaling pathway to speed up glucose metabolism and simultaneously suppresses the activity of the opposing mating pathway to limit the cost of mating.
Yeasts can reproduce both sexually and asexually. Natural selection can favor either vegetative fitness or the mating ability under different conditions and a negative correlation between these two traits has been proposed 21 . The dual roles of Mdf1p in mating and growth may help to explain the molecular link between these two conflicting options. Additionally, by mediating the trade-off between mating and vegetative growth, MDF1 potentially provides an example of antagonistic pleiotropy, in which a gene is beneficial in one condition, but detrimental in another [22][23][24][25] . A genome-wide survey of antagonistic pleiotropy genes was recently performed in yeasts 26 . The prevalence of antagonistic pleiotropy genes in the yeast genome was beyond previous expectations. Regulatory resolution at the transcriptional level was proposed to explain the antagonistic pleiotropy genes found in the genomes. The identification of protein-protein interactions is another key index of pleiotropic characteristics 27,28 . We found that Mdf1p could retain antagonistic pleiotropy by interacting with different partners in two functionally conflicting pathways, where it interacts with MATa2 to repress the mating pathway and binds to Snf1p to promote the growth pathway. Delineation of the functional evolution of the MDF1 gene not only provides a vivid example of how a de novo gene can play pivotal roles in important biological processes, resulting in adaptation, but also demonstrates a novel molecular mechanism to explain the antagon-istic pleiotropy between the mating and growth pathways according to a molecular mechanism.

Methods
Yeast strains and growth conditions. The strain of S. cerevisiae used for all genetic manipulations in this study was BY4742 (MATa his3D1 leu2D0 lys2D0 ura3D0), provided by Prof. Jin-Qiu Zhou of the Shanghai Institutes for Biological Sciences. Cells were grown in standard fermentative YPD medium containing 1% yeast extract, 2% peptone, and 2% glucose at 3uC. For the marker-bearing strains, SD-Ura, SD-Trp or SD-His 29 was used for selection.
Growth rate assay. The details of the generation of the MDF1 strain driven by its own promoter (previously referred to as the M1A-strain) were described by Li et al. 7 . Growth curves for the yeast strains were generated using a Bio photometer (Eppendorf). The cultured MDF1 strain and the wide-type strain were diluted to 0.1 6 0.01 OD independently. OD values were measured every 30 min for the first two hours and then every 1 h until 4 h. The experiments were repeated five times. Student's t-test was used to test the difference of growth rates among multiple timepoints.
Co-immunoprecipitation assays. A total of 100 OD yeast cells overexpressing either 6xHis or HA-tagged Mdf1p were lysed in 500 ml of lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1 mM PMSF, protease inhibitor cocktail (Roche), 0.2 mM Na3VO4, 100 mM NaF, 0.2% NP-40) via beating with glass beads, followed by centrifugation at 12000 g for 5 min at 4uC. The 1 ml supernatant was pre-cleared with 5 ml of protein-G agarose (Santa Cruz). An anti-His or HA mouse monoclonal antibody (R&D systems) was added at a dilution of 15100, followed by overnight incubation at 4uC with mild shaking. Then, 20 ml of protein-G agarose was added, followed by incubation for 1 h and centrifugation at 850 g for 10 min at 4uC. The beads were washed three times with lysis buffer and an additional three times with wash buffer. After the final wash, the beads were boiled for 5 min in 50 ml of SDS-  converging pathways: Mdf1p binds to the MATa2 protein to cooperatively inhibit haploid-specific genes that encode key components of the mating pathway (MAPK pathway) and thereby suppresses the mating behavior of budding yeast to limit the cost of mating, as revealed in a previous study by our group; and Mdf1p binds to Snf1p to turn off glucose-repressed genes and turn on glucose-induced genes, thus enhancing the activity of the glucose signaling pathway.
www.nature.com/scientificreports SCIENTIFIC REPORTS | 4 : 7280 | DOI: 10.1038/srep07280 PAGE loading buffer. As a negative control, wild-type yeast cells (100 OD) were also treated according to the method described above. A 20 ml immunoprecipitation sample obtained from the 6xHis-tagged Mdf1p strain and a control sample were analyzed via SDS-PAGE. The gel was visualized by silver staining with the PlusOneTM silver stain kit (GE healthcare). Compared with the control, the significant and repeatable unique band was analyzed through MALDI-TOF/TOF. The samples obtained from the HA-tagged Mdf1p strain and the control strain were analyzed by western blotting using an Snf1p antibody (Santa Cruz).
Quantitative RT-PCR. Quantitative RT-PCR primers were designed for the glucoserepressed genes, ICL1, FBP1, PCK1, ACS1, and MLS1, respectively. Quantitative RT-PCR assays were performed in triplicate in a 96-well plate in a Bio-Rad iCycler iQ Real-Time PCR Detection System (Bio-Rad) using iQSYBR Green Supermix (Bio-Rad), according to the manufacturer's recommended protocol. The CT values of each sample were acquired with iCycler iQ software 3.0 (Bio-Rad). Specific primers for ACT1 were used as an internal control. The initial 150 bp of the coding sequence were employed to quantify each of the selected genes.