Biology HW

Wang et al. Biotechnol Biofuels (2019) 12:59 https://doi.org/10.1186/s13068-019-1398-7

R E S E A R C H

QTL analysis reveals genomic variants linked to high-temperature fermentation performance in the industrial yeast Zhen Wang1,2†, Qi Qi1,2†, Yuping Lin1*, Yufeng Guo1, Yanfang Liu1,2 and Qinhong Wang1*

Abstract Background: High-temperature fermentation is desirable for the industrial production of ethanol, which requires thermotolerant yeast strains. However, yeast thermotolerance is a complicated quantitative trait. The understanding of genetic basis behind high-temperature fermentation performance is still limited. Quantitative trait locus (QTL) map- ping by pooled-segregant whole genome sequencing has been proved to be a powerful and reliable approach to identify the loci, genes and single nucleotide polymorphism (SNP) variants linked to quantitative traits of yeast.

Results: One superior thermotolerant industrial strain and one inferior thermosensitive natural strain with distinct high-temperature fermentation performances were screened from 124 Saccharomyces cerevisiae strains as parent strains for crossing and segregant isolation. Based on QTL mapping by pooled-segregant whole genome sequencing as well as the subsequent reciprocal hemizygosity analysis (RHA) and allele replacement analysis, we identified and validated total eight causative genes in four QTLs that linked to high-temperature fermentation of yeast. Interestingly, loss of heterozygosity in five of the eight causative genes including RXT2, ECM24, CSC1, IRA2 and AVO1 exhibited posi- tive effects on high-temperature fermentation. Principal component analysis (PCA) of high-temperature fermentation data from all the RHA and allele replacement strains of those eight genes distinguished three superior parent alleles including VPS34, VID24 and DAP1 to be greatly beneficial to high-temperature fermentation in contrast to their inferior parent alleles. Strikingly, physiological impacts of the superior parent alleles of VPS34, VID24 and DAP1 converged on cell membrane by increasing trehalose accumulation or reducing membrane fluidity.

Conclusions: This work revealed eight novel causative genes and SNP variants closely associated with high-temper- ature fermentation performance. Among these genes, VPS34 and DAP1 would be good targets for improving high- temperature fermentation of the industrial yeast. It also showed that loss of heterozygosity of causative genes could contribute to the improvement of high-temperature fermentation capacities. Our findings would provide guides to develop more robust and thermotolerant strains for the industrial production of ethanol.

Keywords: High-temperature fermentation (HTF), Pooled-segregant whole-genome sequence analysis, QTL mapping, Reciprocal hemizygosity analysis, Allele replacement, Saccharomyces cerevisiae

© The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat iveco mmons .org/ publi cdoma in/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Open Access

Biotechnology for Biofuels

*Correspondence: lin_yp@tib.cas.cn; wang_qh@tib.cas.cn †Zhen Wang and Qi Qi contributed equally to this work 1 CAS Key Laboratory of Systems Microbial Biotechnology, Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, Tianjin 300308, China Full list of author information is available at the end of the article

Page 2 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

Background Saccharomyces cerevisiae has been widely used for the production of various fuels and chemicals, more recently, eco-friendly bioethanol [1, 2]. Although robust indus- trial S. cerevisiae strains produce ethanol from agricul- tural wastes with high yield and productivity, the urgent demand of larger production and minimum costs is still challenging. Improved thermotolerance performance can address this obstacle to some extent, since high-tem- perature fermentation can greatly reduce cooling costs, increase cell growth, viability and ethanol productivity via facilitating the synchronization of saccharification and fermentation [3, 4]. However, thermotolerance is a complex quantitative trait and determined by a compli- cated mechanism referring to the interaction of many genes [5]. Thus, it is very challenging to develop robust S. cerevisiae strains with enhanced thermotolerance to meet industrial requirement.

Many efforts have been made to understand the molec- ular mechanisms and genetic determinants underlying yeast thermotolerance, but most of them focused on laboratory strains, which display much lower thermal tolerance than the robust industrial and natural yeast strains [6]. Previous study indicated that industrial yeast has evolved complex but subtle mechanisms to protect the organism from high-temperature lesion by activat- ing and regulating of specific thermal tolerance-related genes to synthesize specific compounds [7]. To identify novel genes and elucidate the intricate mechanism of thermotolerance, many methods were developed [8–12]. Although these approaches have disclosed a number of causative genes and revealed some compounds, e.g. sterol composition, for responding to the thermal stress, identification of quantitative trait genes still faced with tremendous challenges, including variable contributions of quantitative trait loci (QTL), epistasis [13], genetic heterogeneity [14], etc.

With the rapidly development of high-throughput genome sequencing, pooled-segregant whole genome sequencing technology has been developed for efficiently mapping QTLs related to complex traits [15, 16]. Sub- sequent genetic approaches, such as reciprocal hemizy- gosity analysis (RHA) and allele replacement analysis, accelerated identification of the causative genes linked to superior phenotypes [17]. S. cerevisiae as a model organism is renowned for the acquisition of abundant genetic markers [18], the ease of introduction of precise genetic modification and the convenience of perform- ing experimental crosses [19], thus perfectly suitable for the application of QTL methodology to disclose complex traits. The efficient methodology has facilitated identifi- cation of several genomic regions and causative genes related to the complex traits in S. cerevisiae, including

thermotolerance, ethanol tolerance, glycerol yield, etc. [5, 20–22]. However, up to now, the underlying molecular mechanisms of thermotolerance in S. cerevisiae are still unclear, and the identification of novel causative genes continues to be of interest to accelerate the breeding of robust yeast strains with improved high-temperature fer- mentation performance.

In this study, to uncover genetic determinants linked to high-temperature fermentation performance of the industrial yeast, QTL mapping by pooled-segregant whole genome sequence analysis and subsequent valida- tion by RHA and allele replacement analysis were per- formed. The scheme of this work was shown in Fig.  1. Total eight genes containing nonsynonymous SNP vari- ants in two major QTLs linked to the superior parent and two minor QTLs linked to the inferior parent were iden- tified and validated to be causative genes tightly associ- ated with thermotolerance. Among these genes, loss of heterozygosity in RXT2, ECM24, CSC1, IRA2 and AVO1 seemed to play beneficial roles in developing thermotol- erance; meanwhile, the superior parent alleles of VPS34, VID24 and DAP1 were distinguished to be greatly benefi- cial to high-temperature fermentation in contrast to their inferior parent alleles, due to their positive effects on improving protective function of cell membrane against thermal stress. This study improved our understanding of genetic basis behind thermotolerance, and identified more new causative genes linked to yeast thermotoler- ance, thus providing more guidance to enhance thermo- tolerance of industrial yeast strains.

Results Selection of parent strains for genetic mapping of thermotolerance Total 124 natural, laboratory and industrial isolates of S. cerevisiae collected in our lab (Additional file 1: Table S1) were evaluated for their high-temperature fermentation performances. The OD600 values representing cell growth at 42 °C for 36 h ranged from 0.66 to 6.24 (Fig. 2a, Addi- tional file  2: Table  S2), showing that the strain ScY01 had the highest cell growth while W65 had the low- est cell growth under thermal stress conditions. Mean- while, ScY01 consumed the highest amount of glucose (116.0  g/l) and produced the highest amount of ethanol (57.3 g/l) at 42 °C (Fig. 2a, Additional file 2: Table S2). By contrast, W65 almost had no glucose consumption and ethanol production at 42  °C. ScY01 derived from the industrial strain Ethanol Red through adaptive evolu- tion at high temperature [11], whereas W65 is a natural isolate. Cell growth profiles at 42  °C and 30  °C further confirmed that ScY01 was significantly more thermotol- erant than W65 at elevated temperature (Fig.  2b), while both strains had no significant differences of cell growth

Page 3 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

at normal temperature. Therefore, ScY01 and W65 were chosen as the superior and inferior strains for genetic mapping of thermotolerance, respectively.

Both ScY01 and W65 were separately sporu- lated to generate the MATα and MATa haploid seg- regants, named ScY01α and W65a (Additional file  1:

Table  S1), respectively. To obtain stable haploids for genetic mapping, the HO gene in ScY01α and W65a were further knocked out by inserting zeocin- or geneticin-resistance cassettes. The resulting haploid parent strains were named ScY01α-tp and W65a-sp, respectively.

Fig. 1 Scheme of identification of causative genes linked to the thermotolerance phenotype. One superior thermotolerant strain ScY01 and one inferior thermosensitive strain W65 were selected from 124 S. cerevisiae strains based on evaluation of thermotolerance. The haploid segregants of two parent strains with HO gene deletion were generated and crossed to create the hybrid diploid strain tp/sp. Total 277 segregants were sporulated from the hybrid strain and selected for the superior, random and inferior pools based on evaluation of thermotolerance. Genomic DNA was extracted from these three pools as well as two parent strains and the best and worst spores in the superior pool and subjected to genome resequencing. QTL mapping analyses were performed using the EXPloRA and MULTIPOOL methods. To identify candidate causative genes, the SNPs in QTLs were annotated, and the nonsynonymous SNPs in coding regions were sorted out according to their existences in G28 and Z118 and their manually checked frequencies using Integrative Genomics Viewer (IGV ). Two major QTLs and two minor QTLs were identified to originate from the superior and inferior parent strain, in which five and three genes contained nonsynonymous single nucleotide polymorphism (SNP) variants, respectively. Reciprocal hemizygosity analysis (RHA) and allele replacement analysis further revealed two causative genes in major QTLs and one causative gene in minor QTLs

Page 4 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

Screening of the superior, inferior and random pools of segregants for genome sequencing The parent haploid strains ScY01α-tp and W65a-sp were crossed to obtain the hybrid diploid strain tp × sp and then sporulated. Since ScY01α-tp and W65a-sp had zeocin- or geneticin-resistance cassettes at HO locus, successfully segregated haploid spores should only inherit one drug resistance capacity of either zeocin or gene- ticin. Combining with the subsequent diagnostic PCR for the MAT locus, we isolated 107 haploid segregants on geneticin selective plates and 170 haploid segregants on zeocin selective plates. Total 277 haploid segregants were isolated and tested for their thermotolerance capacities to screen the ten most thermotolerant or thermosensitive segregants for the superior pool and the inferior pool, respectively, as well as ten random segregants for the ran- dom pool for genome sequencing.

The distribution of the stress tolerance index (STI) values (calculated as the ratio of the OD600 at 42  °C ver- sus the OD600 at 30 °C measured at the 16-h time point) in 277 haploid segregants is shown in Fig.  3a. Apparent continuous variation as well as normal frequency distri- bution of STI in the haploid segregants from the hybrid

tp × sp indicated yeast thermotolerance as a quantitative trait. Among them, 49 segregants showed lower STI val- ues (< 0.22) than the inferior parent W65a-sp, while 77 segregants showed higher STI values (> 0.38) than the superior parent ScY01α-tp. Thus, ten segregants show- ing the 10 lowest STI values (0.09 to 0.11) were selected as the most thermosensitive segregants and assembled in the inferior pool (Fig. 3a). To further narrow down supe- rior segregants, cell growths of those 77 segregants at 42  °C were compared with ScY01α-tp (Fig.  3b). Among them, 31 segregants showed higher cell growth than ScY01α-tp at 42  °C. Thus, ten segregants showing the ten highest OD600 ratios (1.37 to 2.17) than ScY01α-tp were selected as the most thermosensitive segregants and assembled in the superior pool. Finally, excluding the segregants in superior and inferior pools, ten of the rest segregants were randomly selected and assembled in the random pool.

Additionally, fermentation capacities of the ten seg- regants in the superior pool as well as parent strains were evaluated (Fig.  3c). After 36  h incubation at 42  °C, the thermotolerant parent strain ScY01α-tp consumed 68.6 ± 1.5  g/l glucose, produced 28.6 ± 1.1  g/l ethanol and resulted in cell growth of 4.12 ± 0.04 OD600. By con- trast, the thermosensitive parent strain W65a-sp, which consumed 14.4 ± 0.3  g/l glucose, produced 6.0 ± 0.2  g/l ethanol, and resulted in cell growth of 0.50 ± 0.01 OD600, showing much lower fermentation capacity in contrast to ScY01α-tp. The hybrid strain tp × sp exhibited higher fermentation capacity than both the haploid parent strains, which might be partially due to ploidy-driven adaptation in cell physiology as previously reported [23]. Remarkably, two segregants G29 and G28 showed higher capacities of glucose consumption and ethanol accu- mulation than the hybrid strain tp × sp and the superior parent ScY01α-tp, implicating unknown genetic factors beyond the impacts of ploidy and the superior parent on cell physiology. In addition, G28 showed slightly higher ethanol accumulation than G29. On the other hand, the segregant Z118 showed the worst fermentation capac- ity. Thus, to facilitate QTL mapping based on pooled- segregant whole-genome sequence analysis, the best and the worst spores (G28 and Z118) from the superior pool were also selected for genome sequencing.

Identification of QTLs and candidate causative genes by pooled‑segregant whole‑genome sequence analysis To identify the genetic basis underlying yeast thermo- tolerance, seven samples, which were two parent strains ScY01α-tp and W65a-sp showing distinct thermotol- erance capacities, three segregant pools including the superior, inferior and random pools derived from these

Fig. 2 Thermotolerance of 124 S. cerevisiae strains including two parent strains ScY01 and W65. a Cell growth (black bar), consumed glucose (red bar) and produced ethanol (blue bar) at 42 °C for 36 h. Cells were grown in YP medium containing 200 g/l glucose. b Cell growth of two parent strains ScY01 and W65 at 42 °C and 30 °C. Cells were grown in 100 ml Erlenmeyer flasks containing 50 ml YP medium with 200 g/l glucose. Data represent the mean and standard error of duplicate cultures at each condition (error bars are covered by symbols). Initial OD600 of 0.5 was used for all the fermentations. In panel b, data represent the mean and standard error of duplicate cultures at each condition

Page 5 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

two parents, and two individual segregants that were the best and worst segregants G28 and Z118 in the supe- rior pool, were subjected to whole-genome sequencing for QTL mapping analysis. Single nucleotide polymor- phisms (SNPs) in these seven samples were separately extracted from their genomic alignments with the sequence of the reference S288c genome. Total 35,459 quality-filtered and discordant SNPs from two parent strains were used as genetic makers (Additional file  3: Dataset S1). Usually, thermotolerance-related SNP vari- ants in the superior pool are expected to dominantly

inherit from the superior parent. However, previous studies have demonstrated the presence of recessive mutations linked to yeast stress tolerance in the infe- rior parent [5, 24], suggesting that the inferior parent could also pass thermotolerance-related SNP variants on to segregants in the superior pool. Therefore, we detected the major QTLs originating from the superior parent and the minor QTLs inherited from the inferior parents, respectively. Correspondingly, the SNP variant frequencies in the three segregant pools were calculated as the percentages of the SNP nucleotides originating

Fig. 3 Selection of superior, inferior and random pools for genome sequencing by evaluating thermotolerance capacities of segregants. a The distribution of the STI values in 277 haploid segregants from the hybrid of the two parent haploid strains ScY01α-tp and W65a-sp. Seventy-seven segregants showing higher STI values (> 0.38) than the superior parent ScY01α-tp were selected as superior segregants. Forty-nine segregants showing lower STI values (< 0.22) than the inferior parent W65a-sp were selected as inferior segregants. Ten segregants showing the ten lowest STI values (0.09 to 0.11) were selected as the most thermosensitive segregants and assembled in the inferior pool. Cell growth experiments were carried out in triplicates for each strain in 96-well plates with 1 ml YPD medium at 42 °C and 30 °C. b Cell growth comparison of the 77 segregants and W65a-sp with ScY01α-tp at 42 °C. Thirty-one segregants showed higher cell growth than ScY01α-tp at 42 °C. Ten segregants showing the ten highest OD600 ratios (1.37 to 2.17) than ScY01α-tp were selected as the most thermosensitive segregants and assembled in the superior pool. c Fermentation capacities of ten segregants in the superior pool at 42 °C. Fermentation experiments were conducted in 100 ml Erlenmeyer flasks containing 50 ml YP medium with 200 g/l glucose at 42 °C. Consumed glucose, produced ethanol and cell growth were measured after incubation for 36 h. Data represent the mean and standard error of duplicate cultures at each condition. Excluding the segregants in superior and inferior pools, ten of the rest segregants were selected and assembled in the random pool

Page 6 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

from the superior or inferior parent for mapping major or minor QTLs. The raw SNP frequencies were plotted against the chromosomal position and smoothened by using a Linear Mixed Model [25] (Fig.  4, upper panel). Linkage analysis of QTLs was further performed using the EXPLoRA and MULTIPOOL methods [26] (Fig.  4, bottom panel). Overall, the numbers of QTLs identified by these two methods were similar (Table  1; Additional file  4: Dataset 2; Additional file  5: Dataset 3). However, the average lengths of major and minor QTLs identified by EXPLoRA were 47-kb or 69-kb, which were refined to 20-kb or 12-kb by MULTIPOOL (Table  1). Meanwhile, the numbers of nonsynonymous variants and affected genes were narrowed down.

To subtly identify candidate causative genes linked to thermotolerance, all the SNP variants in the QTLs identified by MULTIPOOL were localized to coding and non-coding regions and annotated to be synony- mous and nonsynonymous (Additional file  5: Dataset S3). Furthermore, the nonsynonymous SNPs in cod- ing regions, which in two sequenced individual spores G28 and Z118 were similar to those in the parent strains and also consisted with those in the superior pool, were sorted out and manually checked for their frequencies using Integrative Genomics Viewer (IGV) [27–29]. We estimated that the SNP frequencies (count of SNP-containing reads/total count of mapped reads) in QTLs linked to thermotolerance could be high in the superior pool, but low in the inferior pool, and simul- taneously at the median value of around 0.5. Only the variants in major QTLs meeting the criteria of allele frequencies with ≤ 10% in the inferior pool, ≥ 75% in the superior pool and around 50% in the random pool as well as the variants in minor QTLs meeting the cri- teria of allele frequencies with ≤ 25% in the inferior pool, ≥ 75% in the superior pool and around 50% in the random pool were considered to be causative vari- ant candidates related to thermotolerance (Additional file  5: Dataset S3). Therefore, the genes affected by these causative variants were considered as candidate causative genes, and the QTLs containing these can- didate causative genes were fine-mapped (Table  2). In total, two major QTLs, QTL1 and QTL2, were local- ized on chromosome II and XII (Fig.  4a) and con- tained two (RXT2 and VID24) and three affected genes (ECM22, VPS34 and CSC1) by nonsynonymous causa- tive variant candidates. Two minor QTLs, QTL3 and QTL4, were localized on chromosome XV and XVI (Fig.  4b) and contained two (IRA2 and AVO1) and one (DAP1) affected genes by nonsynonymous causa- tive variant candidates (Table 2). Total eight candidate causative genes were identified by pooled-segregant whole-genome sequence analysis.

Validation of causative genes in the QTLs Reciprocal hemizygosity analysis (RHA) and allele replacement analysis were, respectively, employed to validate the eight candidate causative genes in the QTLs (Table 2) based on the lethality and unavailability of their gene deletions. RHA was used for five non-essential genes including RXT2, VID24, ECM22, IRA2 and DAP1, since their deletions were non-lethal. Allele replacement was used for two essential genes including VPS34 and AVO1, whose null alleles are inviable, as well as the CSC1 gene, whose deletion mutant was unavailable after sev- eral rounds of attempts. For RHA, five pairs of hemizy- gous diploid tp × sp hybrid strains were constructed (Additional file  1: Table  S1), in which each pair retained a single copy of the superior (ScY01α-tp) or inferior (W65a-sp) parent allele of RXT2, VID24, ECM22, IRA2 and DAP1, respectively, while the other copy of the gene was deleted. For allele replacement analysis, three pairs of allele homozygotes of diploid tp × sp hybrid strains were constructed (Additional file  1: Table  S1), in which each pair contained two homogeneous gene allele from the superior (ScY01α-tp) or inferior parent (W65a-sp) allele of VPS34, AVO1 and CSC1, respectively. The fer- mentation profiles of RHA and allele replacement strains at high temperature were shown in Additional file  1: Figure S1, and the diploid hybrid (tp × sp) of two parent strains was used as a control. To have better quantitative comparisons of fermentation capacities, the fermenta- tion rates including maximum cell growth rate (μmax), glucose-consumption rate (qsmax) and ethanol produc- tivity (PEtOH) were calculated according to the fermen- tation data in Additional file  1: Figure S1, and shown in Fig. 5. From the results of RHA, compared with the con- trol strain tp × sp, one of two hemizygotes for VID24 and DAP1 showed significantly decreased cell growth or/ and fermentation capacities at high temperature (Fig. 5), whereas both two hemizygotes for RXT2, ECM22 and IRA2 showed increased thermotolerances to different extent (Fig.  5). As for allele replacement analysis, com- pared with the control strain tp × sp, one of two allele homozygotes for VPS34 but both two allele homozygotes for CSC1 and AVO1 showed significantly increased cell growth or/and fermentation capacities at high tempera- ture (Fig. 5). As for VPS34 and CSC1 in the major QTL2, the allele homozygotes containing the variants from the superior parent ScY01α-tp were expected to have higher thermotolerance than the control stain or the allele homozygotes containing the variants from the infe- rior parent W65a-sp. As for AVO1 in the minor QTL3, the homozygote containing the variant from W65a-sp was expected to have higher thermotolerance than the control stain or the allele homozygote containing the variant from ScY01α-tp. Unexpectedly, the homozygote

Page 7 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

Fig. 4 Mapping of major (a) and minor (b) thermotolerance-related QTLs by pooled-segregant whole-genome sequence analysis. a In major QTL mapping, the SNP frequencies refer to the percentage of the SNP nucleotide in the pools originating from the thermotolerant parent strain ScY01α-tp. b In minor QTL mapping, the SNP frequencies refer to the percentage of the SNP nucleotide in the pools originating from the thermosensitive parent strain W65a-tp. In the upper panels of a and b, scatter plots of SNP frequency versus chromosome are shown. The raw data of SNP frequencies are shown as dots, smoothened using a Linear Mixed Model [30] and shown in bold lines. Green, red and purple dots and lines represent the raw data and smoothed data of SNP frequencies in superior pool, inferior pool and random pool, respectively. In the bottom panels of a and b, QTL detections using the EXPLoRA and MULTIPPOL methods are shown. The green line represents the probability of linkage obtained by EXPLoRA, where peak regions showed higher SNP frequencies than 0.5 and were, therefore, detected as QTLs. The red line represents LOD scores in superior pool versus inferior pool calculated by MULTIPOOL, whereas the purple line represents LOD scores in superior pool versus random pool. When both maximum LOD scores were higher than 5, this locus was detected as a QTL by MULTIPOOL. QTLs were further narrowed down by analysing whether nonsynonymous amino acid changes were present. Eventually, two major QTLs named QTL1 and QTL2 as well as two minor QTLs named QTL3 and QTL4 were identified

Page 8 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

containing the CSC1 allele from W65a-sp showed oppo- site results due to increased PEtOH (Fig.  5). Overall, all these eight genes seemed to have impacts on high-tem- perature fermentation performance.

The detailed results were as follows: VID24 was local- ized in the major QTL of QTL1 (Table  2). Deletion of the superior (ScY01α-tp) parent allele of VID24 in the reciprocal hemizygote resulted in decreased cell growth at high temperature but not significantly, and had sig- nificant effects on qsmax and PEtOH at high-temperature (Fig.  5). This result suggested the VID24 allele from the superior strain might act as a causative and positive gene in thermotolerance. VPS34 was in the major QTL of QTL2 (Table  2). The allele homozygote containing two copies of the VPS34D591E allele from the superior parent showed significantly higher fermentation rates and capac- ities at high temperature than the one containing two copies of the inferior parent allele as well as the control hybrid strain tp × sp (Fig. 5, Additional file 1: Figure S1). Furthermore, our previous genome sequencing showed that the diploid superior parent strain ScY01 has two homogenous copies of the VPS34 D591E allele [30]. There- fore, the VPS34D591E allele might be a causative gene in thermotolerance. DAP1 was in the minor QTL of QTL4 (Table  2). The DAP1V39I mutant allele inheriting  from

the inferior parent strain W65a-sp was  found in the superior thermotolerant pool (Table  2, Additional file  5: Dataset S3). We estimated that the reciprocal hemizy- gote containing the inferior parent allele of DAP1 might have higher thermotolerance than the one containing the superior parent allele of DAP1. Unexpectedly, the result is quite the opposite. Compared with the control strain tp × sp, the reciprocal hemizygote containing the inferior parent allele of DAP1 showed significantly decreased fer- mentation rates and capacities at high temperature, while the one containing the superior parent allele of DAP1 exhibited significantly increased thermotolerance (Fig. 5, Additional file  1: Figure S1). This result implicated that the inferior parent allele of DAP1V39I might be a reces- sive deleterious mutation in segregants of the superior pool, while DAP1 might act as a recessive beneficial gene in the superior thermotolerant parent. In terms of the other five genes except for VID24, VPS34 and DAP1, the hybrid control strain tp × sp containing their heterogene- ous alleles showed lower high-temperature fermentation performance than either the reciprocal hemizygotes only retaining a single copy of allele or the allele homozygotes containing two homogeneous copies of allele (Fig.  5, Additional file  1: Figure S1). The extensive loss of hete- rozygosity in S. cerevisiae genomes have been reported to enable the expression of recessive alleles and generating

Table 1 QTL mapping by EXPLoRA and MULTIPOOL methods

Method Number of QTL Average length (kb) Number of nonsynonymous variants

Number of affected genes

Major QTL EXPLoRA 24 47 744 297

MULTIPOOL 22 20 292 119

Minor QTL EXPLoRA 13 69 521 233

MULTIPOOL 11 12 86 40

Table 2 Genes with nonsynonymous variants in two major and two minor QTLs

QTLs Chr Start (bp) End (bp) Length (bp) LOD score Affected gene Mutation (S288c genome as a reference)

ScY01α‑tp W65a‑sp

Major QTLs

QTL1 II 408,800 553,700 144,900 17 RXT2 332 G>C (R111G) Wild type

VID24 154 C>T (P52S) Wild type

QTL2 XII 595,800 633,500 37,700 300 ECM22 1954 G>A (G652S) Wild type

VPS34 1773 C>G (D591E) Wild type

CSC1 1126 C>A (Q376K) Wild type

Minor QTLs

QTL3 XV 174,500 184,900 10,400 272 IRA2 Wild type 7222 C>A (P2408T )

AVO1 Wild type 2558 T>C (V853A)

QTL4 XVI 228,200 238,100 9900 155 DAP1 Wild type 115 G>A (V39I)

Page 9 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

novel allele combinations with potential effects on phe- notypic diversity [31]. Thus, loss of heterozygosity in the five gene alleles might play a similar function in contrib- uting to high-temperature fermentation performance.

Overall, all the results suggested that these eight genes were probably causative genes that linked to high-tem- perature fermentation performance in S. cerevisiae, although in different ways and to different extent.

Characterization of key causative gene alleles for improving high‑temperature fermentation of the industrial yeast To further distinguish good targets from the eight causative gene alleles for improving high-temperature fermentation of the industrial yeast, we performed principal component analysis (PCA) for high-temper- ature fermentation data from all the RHA and allele replacement strains at 42  °C in Additional file  1: Fig- ure S1, including cell growth, glucose consumption and ethanol production at all the time points during fermentation. As shown in Fig. 6a, the first and second

accounted for 75.7% (PC1) and 10.5% (PC2) of the total variation, respectively. The RHA and allele replace- ment strains harbouring gene allele of VID24, VPS34 or DAP1 from the superior industrial parent ScY01α-tp (red, blue and pink triangles in Fig.  6a), showing enhanced high-temperature fermentation capacities in contrast to the control stain tp × sp, were clearly sepa- rated by the PCs from the RHA and allele replacement strains containing those gene alleles from W65a-sp (red, blue and pink circles in Fig.  6a). This result con- firmed that the alleles of VID24, VPS34 and DAP1 in the industrial yeast ScY01 could be greatly beneficial to high-temperature fermentation. By contrast, as for the rest five genes including RXT2, ECM24, CSC1, IRA2 and AVO1, the RHA and allele replacement strains har- bouring their alleles from the parents were relatively closely grouped by the PCs, although showing higher high-temperature fermentation capacities than the con- trol strain tp × sp. This result suggested that the alleles of RXT2, ECM24, CSC1, IRA2 and AVO1 in the indus- trial yeast ScY01 might play minor roles in supporting

Fig. 5 Identification of the causative genes using RHA and allele replacement methods. a Maximum cell growth rate (μmax). b Glucose-consumption rate (qsmax). c Ethanol productivity (PEtOH). The RHA and allele replacement strains are detailed in Additional file 1: Table S1. High-temperature fermentation capacities were evaluated at 42 °C in 100 ml Erlenmeyer flasks containing 50 ml YP medium with 200 g/l glucose at 220 rpm. Data represent the mean and standard error of duplicate cultures at each condition. Statistical analysis for each group of three strains including the control strain tp × sp and two hemizygotes or homozygotes of each gene was performed using one-way ANOVA followed by Tukey’s multiple-comparison posttest (***P < 0.001, **P < 0.01, *P < 0.05)

Page 10 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

high-temperature fermentation. Most strikingly, the PCs highly distinguished the RHA and allele replace- ment strains harbouring the superior gene alleles of VPS34 or DAP1 (Fig. 6a), suggesting that they would be good targets for improving high-temperature fermenta- tion of the industrial yeast.

VPS34 and VID24 have been reported to be involved in the degradation of FBPase [32, 33], thus possibly affecting

trehalose accumulation. Furthermore, trehalose is required on both sides of the lipid bilayer of membranes for effective protection against thermal stress in S. cerevi- siae [34]. Thus, we measured the trehalose levels in cells of the RHA and allele replacement strains of VPS34 and VID24 and the control strain tp × sp, which were grown at thermal stress conditions (42  °C). Compared to the control strain tp × sp, the allele homozygote containing

Fig. 6 Principal component analysis of high-temperature fermentation data and physiological impacts of key causative genes. a Principal component analysis (PCA) of high-temperature fermentation data from all the RHA and allele replacement strains, including cell growth (orange lines), glucose consumption (purple lines) and ethanol production (turquoise lines) during fermentation (hours 0, 8, 12, 18, 24 30 36 42 and 48). Means of biological repeats (in duplicates) are used. The gene alleles originating from the superior (ScY01α-tp, triangle symbols) and inferior (W65a-sp, circle symbols) parents in the RHA and allele replacement strains were colour-coded. b Trehalose accumulation in the RHA and allele replacement strains of VID24 and VPS34 at high temperature. c Membrane fluidity of the RHA strains of DAP1 at high temperature. The Membrane fluidity is determined by the steady-state anisotropy of fluorescent probe 1-[4-(trimethylamino)pheny]-6-phenyl-1,3,5-hexatriene (TMA-DPH). Yeast cells were grown at 42 °C in 100 ml Erlenmeyer flasks containing 50 ml YP medium with 200 g/l glucose at 220 rpm. For measuring trehalose accumulation, cells were harvested after incubation for 36 h. For determining membrane fluidity, cells were harvested after incubation for 8 h, 16 h and 36 h at the early-exponential, mid-exponential and stationary phases, respectively. Data represent the mean and standard error of duplicate cultures at each condition. Statistical analysis in b and c was performed using one-way ANOVA followed by Tukey’s multiple-comparison posttest (***P < 0.001, **P < 0.01, *P < 0.05)

Page 11 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

two copies of the VPS34D591E allele from the superior parent had significantly higher trehalose levels (Fig.  6b), which was positively correlated with its enhanced high- temperature fermentation capacities (Fig.  5, Additional file  1: Figure S1). Similarly, the reciprocal hemizygote containing the superior (ScY01α-tp) parent allele of VID24 showed significantly higher trehalose levels than the control strain, while the reciprocal hemizygote con- taining the inferior (W65a-sp) parent allele of VID24 had significantly lower trehalose levels, positively corre- lating with their enhanced high-temperature fermenta- tion capacities (Fig. 5, Additional file 1: Figure S1). These results indicated that the superior alleles of VPS34 and VID24 might achieve beneficial effects on the high-tem- perature fermentation capacities of the industrial yeast by increasing trehalose levels.

DAP1 mutation leads to defects in sterol synthe- sis, and thus influencing membrane fluidity [35, 36]. Cell wall and membrane are the first defence barrier against environmental stresses. Negative correlation between stress tolerance and membrane fluidity has been observed for ethanol stress [37]. Therefore, we determined the membrane fluidity of the reciprocal hemizygotes of DAP1 and the control strain tp × sp by measuring steady-state anisotropy of membrane-incor- porated 1-[4-(trimethylamino)pheny]-6-phenyl-1,3,5- hexatriene (TMA-DPH). High anisotropy values indicate low membrane fluidity, allowing strong pro- tection against environmental stresses, and vice versa. The reciprocal hemizygote containing the superior (ScY01α-tp) parent allele of DAP1 exhibited enhanced high-temperature fermentation capacities (Fig.  5,

Additional file  1: Figure S1). Positively correlated, this strain showed significantly higher anisotropy levels at the early-exponential (8  h), mid-exponential (16  h) phases than the control strain, indicating lower mem- brane fluidity (Fig.  6c), thus providing effective pro- tection against thermal stress to support active cell metabolism, especially at the mid-log phase. By con- trast, membrane fluidities of these cells at the station- ary phase among the reciprocal hemizygotes of and the control strains. These results suggested that the supe- rior allele of DAP1 might achieve a beneficial effect on the high-temperature fermentation capacities of the industrial yeast by inhibiting membrane fluidity.

Based on characterization of key causative gene alleles, we generated an overarching model integrat- ing good targets for improving high-temperature fer- mentation of the industrial yeast (Fig.  7). Remarkably, we found that the physiological beneficial effects of the superior (ScY01α-tp) parent alleles converged on cell membrane. Vps34 and Vid24 from the superior par- ent can contribute to trehalose accumulation at a high level, thus providing more trehalose on both sides of the lipid bilayer of membranes for effective protection against thermal stress. On the other hand, Dap1 con- taining Valine instead of Isoleucine at position 39 can contribute to reduce membrane fluidity, thus providing a strong defense barrier against thermal stress. Taken together, our model supported the previous under- standings that trehalose accumulation and reduced membrane fluidity could promote high-temperature fermentation in industrial yeast [34, 37], meanwhile revealing VPS34 and DAP1 as good targets for further

Fig. 7 An overarching model integrating good targets for improving high-temperature fermentation of the industrial yeast

Page 12 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

enhancing high-temperature fermentation of the indus- trial yeast.

Discussion Elevated thermotolerance is a highly valuable trait of industrial yeasts that can substantially reduce the pro- duction costs. Previous studies have identified several causative genes and gained some insights into the under- lying mechanism of this complex trait via various effi- cient approaches, especially QTL methodology [5, 10, 12]. A major challenge of QTL analysis is to efficiently identify minor QTLs linked to the inferior parent strain. Since the phenotype is often masked by many subtle fac- tors, for instance, epistasis [13], it is difficult to character- ize the linkage between minor QTLs and the phenotype. However, minor QTLs are unignorable, because they may cause synergistic or additive effect, thus resulting in sig- nificant effects on the related phenotype as major QTLs. An efficient strategy has been used to reveal minor QTLs by eliminating candidate QTLs in both superior and infe- rior parent strains and repeatedly mapping the QTL with pooled-segregant whole-genome sequence analysis [5]. This approach was further upgraded to be carried out using relatively low numbers of segregants [20].

Based on the extensive pooled-segregant whole genome sequence analysis, we successfully identified two major QTLs (QTL1 and QTL2) and two minor QTLs (QTL3 and QTL4) localized on chromosome II, XII, XV, XVI, respectively (Fig.  4, Table  2). Similar to previous study [20], our work confirmed that relatively low numbers of segregants can be used for successful QTL mapping using pooled-segregant whole-genome sequence analy- sis. Besides two methods of EXPLoRA and MUTIPOOL used to detect QTLs, we also sequenced two individual segregants from the superior pool and used IGV to man- ually check SNP frequencies to facilitate more accurate detection of QTLs closely associated with thermotoler- ance. Four QTLs and eight nonsynonymous gene alleles were narrowed down from dozens of QTLs and hun- dreds of nonsynonymous SNP variants after QTL map- ping, and finally validated to be causative factors related to yeast thermotolerance (Additional file  4: Dataset 2, Additional file 5: Dataset 3, Figs. 5, 6). Thus, the workflow used in this study could be feasible and effective for QTL mapping and identification of candidate causative genes.

Interestingly, among the eight validated causa- tive genes, both VID24 and VPS34 were found to be involved in translocation and degradation of fructose- 1,6-bisphosphatase (FBPase) in the vacuole. VID24 encodes a peripheral protein on vacuole import and degradation (Vid) vesicles [38], which is required to transfer FBPase from the Vid vesicles to the vacu- ole for degradation [32]. VPS34 encodes the sole

phosphatidylinositol (Pl) 3-kinase in yeast, which is essential for autophagy [39], which is also required for the degradation of extracellular FBPase in the vacuole import and degradation (VID) pathway [33]. When yeast cells are out of glucose feeding for a long time, Vps34 is induced and co-localized with actin patches in starved cells. Once Vps34 is absent, FBPase and the Vid24 associated with related actin patches before and after re-feeding glucose. Strikingly, VID24 null mutation leads to FBPase accumulation in the vesi- cles, thus affecting trehalose synthesis [32, 40]. VPS34 null mutant also arrests FBPase with high levels in the extracellular fraction. A previous study indicated treha- lose is beneficial to protect cells from thermal stress in S. cerevisiae [34]. Hence, we speculated that VID24 and VPS34 might affect trehalose synthesis by controlling the degradation of FBPase and thus be closely linked to thermotolerance. As expected, we observed the posi- tive correlation between the accumulation of trehalose and the improvement of ethanol production due to the existent of VID24 and VPS34D591E originating from the thermotolerant parent strain ScY01α-tp (Fig.  6b). In terms of testing the relationship between the degrada- tion of FBPase and the improvement of ethanol produc- tion, it would be worthwhile to be further investigated in the future.

DAP1 was identified to be linked to thermotolerance by minor QTL mapping (Fig.  4b). DAP1 encodes Heme- binding protein and mutations lead to defects in mito- chondria, telomeres, and sterol synthesis [35, 36], which was closely associated with thermotolerance [10, 12]. The abundance and composition of sterol plays a significant modulatory role in yeast response to thermal stress by affecting membrane fluidity [41]. Furthermore, the recip- rocal hemizygote containing the superior allele of DAP1 showed increased high-temperature fermentation and lower membrane fluidity in contrast to the control strain (Fig.  6c). Thus, DAP1 might be involved in thermotoler- ance by affecting sterol synthesis and membrane fluidity. Furthermore, our results suggested DAP1 to be a reces- sive causative gene linked to thermotolerance, which was influenced by the genetic background. The mutant allele of DAP1V39I from the inferior parent was validated to be a recessive deleterious mutation for thermotolerance, since the hemizygote containing the DAP1V39I allele showed decreased high-temperature fermentation performance compared to the hybrid control strain tp × sp (Fig.  5c). Meanwhile, the wild-type DAP1 allele was validated to be a recessive beneficial gene in the superior parent, since the hemizygote containing the wild-type DAP1 allele showed increased high-temperature fermentation per- formance compared to the hybrid control strain tp × sp (Fig.  5c). A previous study reported that mechanisms of

Page 13 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

hydrolysate tolerance are very dependent on the genetic background, and causal genes in different strains are dis- tinct [24]. Our results confirmed that the effect of reces- sive alleles or variants might be covered by different genetic backgrounds and complementation of recessive alleles could also contribute to the strain improvement.

Recent genome-wide association study revealed an extensive loss of heterozygosity (LOH) associated with phenotypic diversity across 1011 S. cerevisiae isolates [31]. LOH could provide a driving force of evolution dur- ing the adaptation of the hybrid strain to novel or stress- ful environments by enabling the expression of recessive alleles to potentially support the robustness of cells [42, 43]. In this study, based on RHA and allele replacement analysis, positive effects of LOH on high-temperature fermentation were observed for five causative genes including RXT2, ECM24, CSC1, IRA2 and AVO1 identi- fied by QTL mapping (Figs. 5, 6). Furthermore, we found that the heterozygous forms of these five genes in the control strain tp × sp seemed to have negative effects on thermotolerance. This was different from the findings that the beneficial mutations in heterozygous form seem- ingly confer no benefit at the cellular level in nystatin [44]. These results suggested that LOH would be an inter- esting focus for QTL analysis studies.

Conclusions We evaluated high-temperature fermentation perfor- mances of 124 industrial, natural or laboratory S. cer- evisiae strain and selected one superior thermotolerant strain and one inferior thermosensitive strain as parent strains. Pooled-segregant whole-genome sequence analy- sis was performed for the selected three segregant pools including the superior, inferior and random pools from the hybrid of those two parent strains. Two individual segregants in the superior pool were also sequenced to facilitate the detection of nonsynonymous variants linked to thermotolerance. Candidate causative genes were vali- dated by RHA and allele replacement. Finally, two major QTLs and two minor QTLs as well as eight causative genes containing nonsynonymous SNP variants were identified to be closely linked to yeast thermotolerance. Strikingly, the superior parent alleles of VPS34, VID24 and DAP1 converged on cell membrane by increasing trehalose accumulation or reducing membrane fluidity, and thus beneficial to high-temperature fermentation of the industrial yeast. Furthermore, LOH of five causative genes including RXT2, ECM24, CSC1, IRA2 and AVO1 had positive effects on high-temperature fermentation, suggesting that LOH would be an interesting focus for QTL analysis studies. Overall, we identified novel caus- ative genes linked to high-temperature fermentation

performance of yeast, providing guidelines to develop more robust thermotolerant strain for the industrial pro- duction of ethanol.

Methods Strains, cultivation conditions and sporulation All the strains used in this study are listed in Additional file 1: Table S1. Yeast cells were grown in YPD media (per litre, 10  g yeast extract, 20  g peptone, 20  g glucose) or on YPD agar plates supplemented with 20 g/l agar. Gene knockout transformants or segregants were selected on YPD agar plates containing 400  µg/ml geneticin, 70  μg/ ml zeocin or 200  μg/ml hygromycin B as specified in the text. Mating, sporulation and isolation of haploid segregants were conducted by following standard pro- cedures [45]. The MATα and MATa haploid segregants of parent strains were isolated from strains ScY01 and W65, and named ScY01α and W65a, respectively. To avoid mating-type switch [46], the HO gene in ScY01α and W65a were further knocked out using the previously reported method based on PCR amplification and one- step gene replacement [47]. Zeocin- and geneticin-resist- ance cassettes were PCR amplified from the plasmids pREMI-Z [48] and pFA6-kanMX4 [49] (Additional file 1: Table S1), flanked with 500-bp homologous sequences to the HO gene by fusion PCR and transferred into ScY01α and W65a using the electrotransformation method [50], respectively. Positive transformants were sepa- rately selected on zeocin and geneticin selective plates. To confirm successful knockout, diagnostic PCR reac- tions with primers designed on the HO locus as well as zeocin- and geneticin-resistance cassettes were used (for primers, see Additional file  6: Table  S3). The resulting haploid parent strains were named ScY01α-tp and W65a- sp, respectively, and then crossed and sporulated. Since zeocin- and geneticin-resistance cassettes were allelic in the hybrid diploid strain tp × sp, successfully segregated haploid spores should only inherit one drug resistance capacity of either zeocin or geneticin. Thus, to select haploid segregants, sporulated cells were first isolated on YPD agar plates, and then replica plated on both zeocin and geneticin selective plates. Cell patches only grown on zeocin or geneticin selective plate were further subjected to diagnostic PCR for the MAT locus to determine the mating type of segregants and confirm haploidy [51].

Thermotolerance phenotyping The thermotolerance phenotypes of yeast cells were determined by three evaluation ways as specified in the text: (1) cell growth at 42 °C monitored by measuring the OD600 at the 24-h time point, (2) stress tolerance index (STI) based on cell growth (calculated as the ratio of the OD600 at 42  °C versus the OD600 at 30  °C measured

Page 14 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

at the 16-h time point), and (3) fermentation capacity at 42  °C by measuring cell growth, glucose consumption and ethanol production. Cell growth experiments were performed either using high-throughput growth assays in 96-well plates containing 1  ml YPD medium or using 50-ml Falcon tubes containing 10  ml YPD medium with shaking at 220 rpm. Fermentation experiments were con- ducted in 100-ml Erlenmeyer flasks containing 50 ml YP medium (per liter, 10 g yeast extract, 20 g peptone) with 200  g/l glucose at 220  rpm. Cells were pre-cultured in YPD medium at 30  °C overnight before applying to cell growth or fermentation experiments. Starting OD600 used in all the experiments was 0.2. Optical density (OD) at 600  nm was measured using a platereader (Molecu- lar Devices SpectraMax M2e, San Jose, CA, USA). Con- centrations of glucose and ethanol were monitored by high-performance liquid chromatography (HPLC) on an Agilent 1260 system (Agilent, Santa Clara, CA, USA) equipped with a refractive index detector and a Phe- nomenex RFQ fast acid column (100  mm × 7.8  mm ID) (Phenomenex Inc., Torrance, CA, USA). The column was eluted with 0.01  N H2SO4 at a flow rate of 0.6  ml  min

−1 at 55 °C.

Pooled‑segregant whole‑genome sequence analysis After crossing the two parent haploid strains ScY01α-tp and W65a-sp, the ten most thermotolerant segregants were assembled in the superior pool, the ten most ther- mosensitive segregants were assembled in the inferior pool and ten random segregants were used to assemble the ran- dom pool. The segregants were grown separately in 50 ml liquid YPD media at 30 °C to exponential phase. Each pool was made by mixing equal amounts of cells from the ten segregants based on OD600 as previously described [52] and subjected to whole-genome resequencing. Besides, the haploid parent strains ScY01α-tp and W65a-sp and two individual segregants including the best and worst spores (G28 and Z118) in the superior pool were sub- jected to whole-genome resequencing as well. Genomic DNA isolation and the sequencing libraries were con- structed and sequenced on Illumina HiSeq  4000 using 150-bp paired-end sequencing by the Beijing Genomics Institute (BGI) (Shenzhen, China). A mean of 15.9 mil- lion 150-bp clean reads was generated for each library. All the genome sequencing raw data were deposited in the Sequence Read Archive (SRA) at the National Center for Biotechnology Information (NCBI) under the BioProject ID PRJNA414133 with accession number SRP119879.

Variant detection, QTL mapping and identification of candidate causative genes Variants were detected using the Genome Analysis Toolkit (GATK v3.5) Best Practices pipeline [53, 54].

The S. cerevisiae S288c genome was used as a reference and downloaded from RefSeq at the NCBI (sequence assembly version R64, RefSeq assembly accession: GCF_000146045.2). Initially called SNPs were filtered with a minimum read depth of 20 or a minimum variant frequency of 80%. SNP frequency was initially defined by using the percentage of SNP-containing reads in total mapped reads spanning each locus as previously reported [55]. Variant annotation was performed using the package ANNOVAR [56], and variants were then called using GATK HaplotypeCaller to generate variant lists of sequenced samples, relative to the S288c reference genome. By comparing the variant lists of two parent strains ScY01α-tp and W65a-sp, total 35,459 segregat- ing discordant SNP sites were used as genetic makers for QTL analysis (Additional file 3: Dataset S1).

We used the EXPLoRA method [26], to identify large chromosomal regions containing QTLs, and subse- quently analysed those regions with the MULTIPOOL method [57] to obtain high-resolution predictions for causative QTL regions. First, variants of the superior pool were analysed by EXPLoRA (version 1.0) to iden- tify all the putative QTLs when the posterior probabil- ity assigned to the marker is larger than 0.95 (Additional file  4: Dataset S2). Second, for each QTL-containing region identified with the EXPLoRA method, SNP allele frequencies in superior pool versus inferior pool and superior pool versus random pool were compared using MULTIPOOL (-n 1000, -c 3300, -r 100 –m contrast) to generate the high-resolution QTL map (Additional file 4: Dataset S2). When one locus showed a peak of allele frequencies with the EXPloRA method and simultane- ously had a maximum LOD (log10 likelihood ratio) value higher than 5 in superior pool versus inferior pool and random pool with the MULTIPOOL method, this locus was identified as a candidate causative QTL. Additionally, since the inferior parent with low thermotolerance might contain recessive beneficial variants to thermotolerance, putative QTLs linked with the inferior parent were also analysed using the genome variants of the inferior parent as a reference. QTLs linked to the superior and inferior parents were named major and minor QTLs, respectively.

Variants in QTLs resulting in nonsynonymous muta- tion were annotated using the package ANNOVAR [56] (Additional file  5: Dataset S3). To further narrow down and identify candidate causative variants and their affected genes, we first sorted out the variants, which in two sequenced individual spores G28 and Z118 were sim- ilar to those in the parent strains and also consistent with those in the superior pool. Second, we manually checked the variant frequencies of these variants in the sequenced segregant pools using Integrative Genomics Viewer (IGV) [27–29]. Only the variants in major QTLs meeting

Page 15 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

the criteria of allele frequencies with ≤ 10% in the inferior pool, ≥ 75% in the superior pool and around 50% in the random pool as well as the variants in minor QTLs meet- ing the criteria of allele frequencies with ≤ 25% in the inferior pool, ≥ 75% in the superior pool and around 50% in the random pool were considered to be causative vari- ants related to thermotolerance (Additional file  5: Data- set S3).

Reciprocal hemizygosity analysis (RHA) and allele replacement To validate the causative genes within QTLs, RHA was used for non-essential genes including RXT2, VID24, ECM22, IRA2 and DAP1, whereas allele replacement was used for essential genes including VPS34 and AVO1 whose null alleles are inviable. Additionally, since CSC1 deletion mutant for RHA failed to be obtained after several rounds of attempts, allele replacement was also used for this non-essential gene. RHA was carried out as described previously [17]. PCR-mediated gene dis- ruption, based on homologous recombination, was used to generate gene null mutants [58]. The gene disruption cassettes containing hygromycin modules flanked by 500-bp homologous sequences to the target genes were obtained using fusion PCR. Hygromycin-resistance mod- ules were PCR amplified from the plasmid pRS426-hphB [59] (Additional file  1: Table  S1). The 500-bp homolo- gous sequences upstream and downstream the five non-essential genes were PCR amplified from ScY01- tp genomic DNA. The primers were supplemented in Additional file 6: Table S3. The gene disruption cassettes were transferred into ScY01α-tp and W65a-sp using the electrotransformation method [50], respectively. Posi- tive transformants were selected on hygromycin selec- tive plates. Successful gene disruptions were confirmed by diagnostic PCR reactions with primers designed on the target genes as well as hygromycin-resistance cas- sette (for primers, see Additional file 6: Table S3). Subse- quently, for each non-essential causative gene candidate, the gene disruption mutant of ScY01α-tp and the wild- type strain of W65a-sp or vice versa were crossed to construct the diploid hybrid, which was the reciprocal hemizygote that only contained one single gene allele from either ScY01α-tp or W65a-sp.

Allele replacement was achieved using PCR-based fragment through homologous recombination. The 5′ homologous sequence contained the region from nearly 500  bp upstream the identified SNP by QTL mapping to the stop codon in the target gene. The 3′ homolo- gous sequence contained the region 500  bp downstream the stop codon of the target gene. The hygromycin- resistance module was PCR fused between the 5′ and 3′ homologous sequences. The homologous recombination

fragment containing the identified SNP by QTL map- ping in the target gene from one parent was transformed into the other parent or vice versa. Positive colonies were screened on hygromycin selective plates and subjected to PCR amplification and Sanger sequencing to confirm allele replacement (for primers, see Additional file  6: Table  S3). Subsequently, the allele replacement mutant of ScY01α-tp and the wild-type strain of W65a-sp or vice versa were crossed to construct the diploid hybrid, which was the allele homozygote that contained two homoge- neous gene allele from either ScY01α-tp or W65a-sp. Fermentation capacities of all the reciprocal hemizygotes and allele homozygotes were evaluated using the hybrid diploid tp/sp as a control.

Determination of trehalose and membrane fluidity Yeast cells were grown at 42  °C in 100-ml Erlenmeyer flasks containing 50  ml YP medium with 200  g/l glu- cose at 220  rpm. For measuring trehalose accumula- tion, cells were harvested at stationary phase after incubation for 36  h, when cells accumulate high levels of trehalose as previously reported [60]. Trehalose lev- els were determined using trehalose content detection kit (BestBio, China) in accordance with the manufac- turer’s instructions. For determining membrane fluidity, cells were harvested after incubation for 8  h, 16  h and 36  h at the early-exponential, mid-exponential and sta- tionary phases, respectively. Membrane fluidity was assessed using steady-state fluorescence spectroscopy. Steady-state anisotropy of 1-[4-(trimethylamino)pheny]- 6-phenyl-1,3,5-hexatriene (TMA-DPH, MedChemEx- press, USA) following incorporation of the probe into yeast plasma membranes was measured, as previously described with a slight modification [61]. A Spark™ Mul- timode Microplate Reader (Spark 10  M, Tecan, Switzer- land) was used for the measurement of the steady-state anisotropy of TMA-DPH. Both labelling of cells with TMA-DPH and the measurement were conducted at 42 °C.

Calculation of fermentation rates, statistical significance tests and principal component analysis Fermentation parameters including maximum cell growth rate (μmax), maximum glucose consumption rate (qsmax) and ethanol productivity (PEtOH) were calcu- lated corresponding to the fermentation profiles using Originlab® Origin 8 as previously reported [62]. For comparison of high-temperature fermentation between the control strain tp × sp and RHA or allele replace- ment strains, one-way ANOVA was used, followed by Tukey’s multiple-comparison posttest with a 95% confidence interval. Statistics were performed using Origin (version 8.0). The differences were considered

Page 16 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

significant at three levels of P < 0.001, P < 0.01 and P < 0.05. Principal component analysis (PCA) was used to evaluate the impact of gene alleles in the RHA and allele replacement strains, respectively, originat- ing from the superior and the inferior parent, on cell growth, glucose consumption and ethanol production at high temperature (42 °C) during fermentation (hours 0, 8, 12, 18, 24 30 36 42 and 48). Packages FactoMineR and Factoextra [63] were used within R environment [64] for the PCA data analysis and ggplot2-based visu- alization, separately.

Additional files

Additional file 1: Table S1. S. cerevisiae strains used in this study. Figure S1. Fermentation profiles of RHA and allele replacement strains of the causative genes.

Additional file 2: Table S2. All the data in Fig. 2a.

Additional file 3: Dataset S1. Markers for QTL analysis in two parent strains.

Additional file 4: Dataset S2. QTL lists detected by the EXPLoRA and MULTIPOOL methods.

Additional file 5: Dataset S3. Annotation of SNP variants in major and minor QTLs detected by MULTIPOOL.

Additional file 6: Table S3. Primers used in this study.

Abbreviations HTF: high-temperature fermentation; QTL: quantitative trait loci; RHA: reciprocal hemizygosity analysis; STI: stress tolerance index; SNPs: single nucleotide polymorphisms; IGV: Integrative Genomics Viewer; FBPase: fructose-1,6-bisphosphatase; Pl: phosphatidylinositol; VID: vacuole import and degradation; LOH: loss of heterozygosity; HPLC: high performance liquid chromatography; BGI: Beijing Genomics Institute; SRA: Sequence Read Archive; PCA: principal component analysis; TMA-DPH: 1-[4-(trimethylamino) pheny]-6-phenyl-1,3,5-hexatriene.

Authors’ contributions QW, YL, QQ and ZW conceived and designed the project. ZW, QQ and YL performed the experiments. YG, YL and ZW analysed the genome sequenc- ing data. YL, ZW and QW wrote the manuscript. QW supervised the research study and final approved of the version to be published. All authors read and approved the final manuscript.

Author details 1 CAS Key Laboratory of Systems Microbial Biotechnology, Tianjin Institute of Industrial Biotechnology, Chinese Academy of Sciences, Tianjin 300308, China. 2 University of Chinese Academy of Sciences, Beijing 100049, China.

Acknowledgements Not applicable.

Competing interests The authors declare that there are no competing interests.

Availability of data and materials All the genome sequencing raw data were deposited in the Sequence Read Archive (SRA) at the National Center for Biotechnology Information (NCBI) under the BioProject ID PRJNA414133 with Accession Number SRP119879. All other data generated or analysed during this study are included in this published article and its additional files.

Consent for publication Not applicable.

Ethics approval and consent to participate Not applicable.

Funding This work was supported by the National Science Foundation of China (31470214 and 31700077), the National Science Foundation of Tianjin (16JCY- BJC43100) and the Science and Technology Support Program of Tianjin, China (15PTCYSY00020), and funding from the Science and Technology Foundation for Selected Overseas Chinese Scholar of Tianjin to Yuping Lin.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub- lished maps and institutional affiliations.

Received: 30 November 2018 Accepted: 8 March 2019

References 1. Nielsen J, Larsson C, van Maris A, Pronk J. Metabolic engineering of

yeast for production of fuels and chemicals. Curr Opin Biotechnol. 2013;24(3):398–404.

2. Peralta-Yahya PP, Zhang F, del Cardayre SB, Keasling JD. Micro- bial engineering for the production of advanced biofuels. Nature. 2012;488(7411):320–8.

3. Abdel-Banat BM, Hoshida H, Ano A, Nonklang S, Akada R. High-tem- perature fermentation: how can processes for ethanol production at high temperatures become superior to the traditional process using mesophilic yeast? Appl Microbiol Biotechnol. 2010;85(4):861–7.

4. Shahsavarani H, Hasegawa D, Yokota D, Sugiyama M, Kaneko Y, Boonchird C, Harashima S. Enhanced bio-ethanol production from cellulosic materi- als by semi-simultaneous saccharification and fermentation using high temperature resistant Saccharomyces cerevisiae TJ14. J Biosci Bioeng. 2013;115(1):20–3.

5. Yang Y, Foulquie-Moreno MR, Clement L, Erdei E, Tanghe A, Schaerlaekens K, Dumortier F, Thevelein JM. QTL analysis of high thermotolerance with superior and downgraded parental yeast strains reveals new minor QTLs and converges on novel causative alleles involved in RNA processing. PLoS Genet. 2013;9(8):e1003693.

6. Jarolim S, Ayer A, Pillay B, Gee AC, Phrakaysone A, Perrone GG, Breiten- bach M, Dawes IW. Saccharomyces cerevisiae genes involved in survival of heat shock. G3. 2013;3(12):2321–33.

7. Edgardo A, Carolina P, Manuel R, Juanita F, Jaime B. Selection of thermo- tolerant yeast strains Saccharomyces cerevisiae for bioethanol production. Enzyme Microb Technol. 2008;43(2):120–3.

8. Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci USA. 1998;95(25):14863–8.

9. Auesukaree C, Damnernsawad A, Kruatrachue M, Pokethitiyook P, Boon- chird C, Kaneko Y, Harashima S. Genome-wide identification of genes involved in tolerance to various environmental stresses in Saccharomyces cerevisiae. J Appl Genet. 2009;50(3):301–10.

10. Caspeta L, Chen Y, Ghiaci P, Feizi A, Buskov S, Hallstrom BM, Petranovic D, Nielsen J. Biofuels. Altered sterol composition renders yeast thermotoler- ant. Science. 2014;346(6205):75–8.

11. Shui W, Xiong Y, Xiao W, Qi X, Zhang Y, Lin Y, Guo Y, Zhang Z, Wang Q, Ma Y. Understanding the mechanism of thermotolerance distinct from heat shock response through proteomic analysis of industrial strains of Sac- charomyces cerevisiae. Mol Cell Proteomics. 2015;14(7):1885–97.

12. Xiao W, Duan X, Lin Y, Cao Q, Li S, Guo Y, Gan Y, Qi X, Zhou Y, Guo L, et al. Distinct proteome remodeling of industrial Saccharomyces cerevisiae in response to prolonged thermal stress or transient heat shock. J Proteome Res. 2018;17(5):1812–25.

13. Carlborg O, Haley CS. Epistasis: too often neglected in complex trait stud- ies? Nat Rev Genet. 2004;5(8):618–25.

Page 17 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

14. Flint J, Mott R. Finding the molecular basis of quantitative traits: successes and pitfalls. Nat Rev Genet. 2001;2(6):437–45.

15. Ehrenreich IM, Torabi N, Jia Y, Kent J, Martis S, Shapiro JA, Gresham D, Caudy AA, Kruglyak L. Dissection of genetically complex traits with extremely large pools of yeast segregants. Nature. 2010;464(7291):1039–42.

16. Parts L, Cubillos FA, Warringer J, Jain K, Salinas F, Bumpstead SJ, Molin M, Zia A, Simpson JT, Quail MA, et al. Revealing the genetic structure of a trait by sequencing a population under selection. Genome Res. 2011;21(7):1131–8.

17. Steinmetz LM, Sinha H, Richards DR, Spiegelman JI, Oefner PJ, McCusker JH, Davis RW. Dissecting the architecture of a quantitative trait locus in yeast. Nature. 2002;416(6878):326–30.

18. Liti G, Louis EJ. Advances in quantitative trait analysis in yeast. PLoS Genet. 2012;8(8):e1002912.

19. Mancera E, Bourgon R, Brozzi A, Huber W, Steinmetz LM. High-resolution mapping of meiotic crossovers and non-crossovers in yeast. Nature. 2008;454(7203):479–85.

20. Pais TM, Foulquie-Moreno MR, Hubmann G, Duitama J, Swinnen S, Goovaerts A, Yang Y, Dumortier F, Thevelein JM. Comparative poly- genic analysis of maximal ethanol accumulation capacity and toler- ance to high ethanol levels of cell proliferation in yeast. PLoS Genet. 2013;9(6):e1003548.

21. Hubmann G, Mathe L, Foulquie-Moreno MR, Duitama J, Nevoigt E, Thevelein JM. Identification of multiple interacting alleles conferring low glycerol and high ethanol yield in Saccharomyces cerevisiae ethanolic fermentation. Biotechnol Biofuels. 2013;6(1):87.

22. Sinha H, Nicholson BP, Steinmetz LM, McCusker JH. Complex genetic interactions in a quantitative trait locus. PLoS Genet. 2006;2(2):e13.

23. Selmecki AM, Maruvka YE, Richmond PA, Guillet M, Shoresh N, Sorenson AL, De S, Kishony R, Michor F, Dowell R, et al. Polyploidy can drive rapid adaptation in yeast. Nature. 2015;519(7543):349–52.

24. Sardi M, Paithane V, Place M, Robinson E, Hose J, Wohlbach DJ, Gasch AP. Genome-wide association across Saccharomyces cerevisiae strains reveals substantial variation in underlying gene requirements for toxin tolerance. PLoS Genet. 2018;14(2):e1007217.

25. Swinnen S, Schaerlaekens K, Pais T, Claesen J, Hubmann G, Yang Y, Demeke M, Foulquie-Moreno MR, Goovaerts A, Souvereyns K, et al. Identification of novel causative genes determining the complex trait of high ethanol tolerance in yeast using pooled-segregant whole-genome sequence analysis. Genome Res. 2012;22(5):975–84.

26. Duitama J, Sanchez-Rodriguez A, Goovaerts A, Pulido-Tamayo S, Hub- mann G, Foulquie-Moreno MR, Thevelein JM, Verstrepen KJ, Marchal K. Improved linkage analysis of Quantitative Trait Loci using bulk segregants unveils a novel determinant of high ethanol tolerance in yeast. BMC Genomics. 2014;15:207.

27. Robinson JT, Thorvaldsdottir H, Wenger AM, Zehir A, Mesirov JP. Variant review with the Integrative Genomics Viewer. Cancer Res. 2017;77(21):e31–4.

28. Thorvaldsdottir H, Robinson JT, Mesirov JP. Integrative Genomics Viewer (IGV ): high-performance genomics data visualization and exploration. Brief Bioinform. 2013;14(2):178–92.

29. Robinson JT, Thorvaldsdottir H, Winckler W, Guttman M, Lander ES, Getz G, Mesirov JP. Integrative genomics viewer. Nat Biotechnol. 2011;29(1):24–6.

30. Gan Y, Lin Y, Guo Y, Qi X, Wang Q. Metabolic and genomic characterisation of stress-tolerant industrial Saccharomyces cerevisiae strains from TALENs- assisted multiplex editing. FEMS Yeast Res. 2018;18:5.

31. Peter J, De Chiara M, Friedrich A, Yue JX, Pflieger D, Bergstrom A, Sigwalt A, Barre B, Freel K, Llored A, et al. Genome evolution across 1,011 Saccha- romyces cerevisiae isolates. Nature. 2018;556(7701):339–44.

32. Chiang MC, Chiang HL. Vid24p, a novel protein localized to the fructose-1, 6-bisphosphatase-containing vesicles, regulates targeting of fructose- 1,6-bisphosphatase from the vesicles to the vacuole for degradation. J Cell Biol. 1998;140(6):1347–56.

33. Alibhoy AA, Giardina BJ, Dunton DD, Chiang HL. Vps34p is required for the decline of extracellular fructose-1,6-bisphosphatase in the vacuole import and degradation pathway. J Biol Chem. 2012;287(39):33080–93.

34. Magalhaes RSS, Popova B, Braus GH, Outeiro TF, Eleutherio ECA. The trehalose protective mechanism during thermal stress in Saccharomyces cerevisiae: the roles of Ath1 and Agt1. FEMS Yeast Res. 2018;18(6).

35. Hand RA, Jia N, Bard M, Craven RJ. Saccharomyces cerevisiae Dap1p, a novel DNA damage response protein related to the mammalian mem- brane-associated progesterone receptor. Eukaryot Cell. 2003;2(2):306–17.

36. Mallory JC, Crudden G, Johnson BL, Mo C, Pierson CA, Bard M, Craven RJ. Dap1p, a heme-binding protein that regulates the cytochrome P450 protein Erg11p/Cyp51p in Saccharomyces cerevisiae. Mol Cell Biol. 2005;25(5):1669–79.

37. Ishmayana S, Kennedy UJ, Learmonth RP. Further investigation of relation- ships between membrane fluidity and ethanol tolerance in Saccharomy- ces cerevisiae. World J Microbiol Biotechnol. 2017;33(12):218.

38. Brown CR, Dunton D, Chiang HL. The vacuole import and degradation pathway utilizes early steps of endocytosis and actin polymerization to deliver cargo proteins to the vacuole for degradation. J Biol Chem. 2010;285(2):1516–28.

39. Obara K, Noda T, Niimi K, Ohsumi Y. Transport of phosphatidylinositol 3-phosphate into the vacuole via autophagic membranes in Saccharomy- ces cerevisiae. Genes Cells. 2008;13(6):537–47.

40. Xu J, Zhang J, Guo Y, Zai Y, Zhang W. Improvement of cell growth and l-lysine production by genetically modified Corynebacterium glutamicum during growth on molasses. J Ind Microbiol Biotechnol. 2013;40(12):1423–32.

41. Dufourc EJ. Sterols and membrane dynamics. J Chem Biol. 2008;1(1–4):63–77.

42. Smukowski Heil CS, DeSevo CG, Pai DA, Tucker CM, Hoang ML, Dunham MJ. Loss of heterozygosity drives adaptation in hybrid yeast. Mol Biol Evol. 2017;34(7):1596–612.

43. Andersen MP, Nelson ZW, Hetrick ED, Gottschling DE. A genetic screen for increased loss of heterozygosity in Saccharomyces cerevisiae. Genetics. 2008;179(3):1179–95.

44. Gerstein AC, Kuzmin A, Otto SP. Loss-of-heterozygosity facilitates passage through Haldane’s sieve for Saccharomyces cerevisiae undergoing adapta- tion. Nat Commun. 2014;5:3819.

45. Sherman F, Hicks J. Micromanipulation and dissection of asci. Methods Enzymol. 1991;194:21–37.

46. Danhash N, Gardner DC, Oliver SG. Heritable damage to yeast caused by transformation. Biotechnology (N Y ). 1991;9(2):179–82.

47. Guldener U, Heck S, Fielder T, Beinhauer J, Hegemann JH. A new efficient gene disruption cassette for repeated use in budding yeast. Nucleic Acids Res. 1996;24(13):2519–24.

48. Mukaiyama H, Oku M, Baba M, Samizo T, Hammond AT, Glick BS, Kato N, Sakai Y. Paz2 and 13 other PAZ gene products regulate vacuolar engulfment of peroxisomes during micropexophagy. Genes Cells. 2002;7(1):75–90.

49. Wach A, Brachat A, Pohlmann R, Philippsen P. New heterologous modules for classical or PCR-based gene disruptions in Saccharomyces cerevisiae. Yeast. 1994;10(13):1793–808.

50. Becker DM, Guarente L. High-efficiency transformation of yeast by elec- troporation. Methods Enzymol. 1991;194:182–7.

51. Huxley C, Green ED, Dunham I. Rapid assessment of S. cerevisiae mating type by PCR. Trends Genet. 1990;6(8):236.

52. Hubmann G, Foulquie-Moreno MR, Nevoigt E, Duitama J, Meurens N, Pais TM, Mathe L, Saerens S, Nguyen HT, Swinnen S, et al. Quantitative trait analysis of yeast biodiversity yields novel gene tools for metabolic engineering. Metab Eng. 2013;17:68–81.

53. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, Garimella K, Altshuler D, Gabriel S, Daly M, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20(9):1297–303.

54. DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, Philip- pakis AA, del Angel G, Rivas MA, Hanna M, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43(5):491–8.

55. Pinel D, Colatriano D, Jiang H, Lee H, Martin VJ. Deconstructing the genetic basis of spent sulphite liquor tolerance using deep sequencing of genome-shuffled yeast. Biotechnol Biofuels. 2015;8:53.

56. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38(16):e164.

57. Edwards MD, Gifford DK. High-resolution genetic mapping with pooled sequencing. BMC Bioinf. 2012;13(Suppl 6):S8.

Page 18 of 18Wang et al. Biotechnol Biofuels (2019) 12:59

• fast, convenient online submission

•

thorough peer review by experienced researchers in your field

• rapid publication on acceptance

• support for research data, including large and complex data types

•

gold Open Access which fosters wider collaboration and increased citations

maximum visibility for your research: over 100M website views per year •

At BMC, research is always in progress.

Learn more biomedcentral.com/submissions

Ready to submit your research ? Choose BMC and benefit from:

58. Winzeler EA, Shoemaker DD, Astromoff A, Liang H, Anderson K, Andre B, Bangham R, Benito R, Boeke JD, Bussey H, et al. Functional characteriza- tion of the S. cerevisiae genome by gene deletion and parallel analysis. Science. 1999;285(5429):901–6.

59. Kaster KR, Burgett SG, Rao RN, Ingolia TD. Analysis of a bacterial hygromy- cin B resistance gene by transcriptional and translational fusions and by DNA sequencing. Nucleic Acids Res. 1983;11(19):6895–911.

60. Shima J, Hino A, Yamada-Iyo C, Suzuki Y, Nakajima R, Watanabe H, Mori K, Takano H. Stress tolerance in doughs of Saccharomyces cerevisiae trehalase mutants derived from commercial Baker’s yeast. Appl Environ Microb. 1999;65(7):2841–6.

61. Abe F, Hiraki T. Mechanistic role of ergosterol in membrane rigidity and cycloheximide resistance in Saccharomyces cerevisiae. Biochim Biophys Acta. 2009;1788(3):743–52.

62. Lin Y, Chomvong K, Acosta-Sampson L, Estrela R, Galazka JM, Kim SR, Jin YS, Cate JH. Leveraging transcription factors to speed cellobiose fermen- tation by Saccharomyces cerevisiae. Biotechnol Biofuels. 2014;7(1):126.

63. Lê S, Josse J, Husson F. FactoMineR. An R package for multivariate analysis. Stat Softw. 2008;25(1):1–18.

64. Team RC. R: A language and environment for statistical computing. R Found Stat Comput Vienna Austria. 2017.

BioMed Central publishes under the Creative Commons Attribution License (CCAL). Under the CCAL, authors retain copyright to the article but users are allowed to download, reprint, distribute and /or copy articles in BioMed Central journals, as long as the original work is properly cited.

Biology HW/HW1/Experiment Guide/page 1.JPG