Uncoupling reproduction from metabolism extendschronological lifespan in yeastSaisubramanian Nagarajana,b, Arthur L. Kruckeberga,c, Karen H. Schmidta, Evgueny Krolla, Morgan Hamiltona,Kate McInnerneyd, Ryan Summerse, Timothy Taylore, and Frank Rosenzweiga,1

aDivision of Biological Sciences, University of Montana, Missoula, MT 59812; bSchool of Chemical and Biotechnology, Shanmugha Arts, Science,Technology and Research Academy (SASTRA), Thanjavur, Tamilnadu 613041, India; cBiotechnology, DuPont Central Research and Development, ExperimentalStation, Wilmington, DE 19880; dMSU Functional Genomics Core Facility, Department of Microbiology, Montana State University, Bozeman,MT 59171; and eDepartment of Biological Engineering, Utah State University, Logan, UT 84322

Edited by Robert Haselkorn, University of Chicago, Chicago, IL, and approved March 7, 2014 (received for review January 6, 2014)

Studies of replicative and chronological lifespan in Saccharomycescerevisiae have advanced understanding of longevity in all eukar-yotes. Chronological lifespan in this species is defined as the age-dependent viability of nondividing cells. To date this parameterhas only been estimated under calorie restriction, mimicked bystarvation. Because postmitotic cells in higher eukaryotes oftendo not starve, we developed a model yeast system to study cellsas they age in the absence of calorie restriction. Yeast cells wereencapsulated in a matrix consisting of calcium alginate to form∼3 mm beads that were packed into bioreactors and fed ad libi-tum. Under these conditions cells ceased to divide, became heatshock and zymolyase resistant, yet retained high fermentative ca-pacity. Over the course of 17 d, immobilized yeast cells main-tained >95% viability, whereas the viability of starving, freelysuspended (planktonic) cells decreased to <10%. Immobilized cellsexhibited a stable pattern of gene expression that differed mark-edly from growing or starving planktonic cells, highly expressinggenes in glycolysis, cell wall remodeling, and stress resistance, butdecreasing transcription of genes in the tricarboxylic acid cycle, andgenes that regulate the cell cycle, including master cyclins CDC28and CLN1. Stress resistance transcription factor MSN4 and its up-stream effector RIM15 are conspicuously up-regulated in the immo-bilized state, and an immobilized rim15 knockout strain fails toexhibit the long-lived, growth-arrested phenotype, suggestingthat altered regulation of the Rim15-mediated nutrient-sensingpathway plays an important role in extending yeast chronologicallifespan under calorie-unrestricted conditions.

cell longevity | Ca-alginate encapsulation

Unicellular microbes senesce and die (for reviews, see refs.1, 2) and can therefore be used as models to study the ge-

netic control of lifespan in their multicellular relatives. Bakers’yeast cells have a replicative lifespan (RLS), the number of timesa mother cell undergoes mitosis (3), and a chronological lifespan(CLS), the length of time a nondividing cell remains alive (4).Under standard conditions, wild and laboratory yeast exhibitcharacteristic and highly variable RLS (5) and CLS (6), indi-cating that each trait has a genetic component. Indeed, we nowknow that genes such as SIR2, CYR1, and the protein kinaseSCH9 influence yeast RLS and that each has a functional homologin higher eukaryotes (7). Although yeast CLS has been reported tovary over a fourfold range among wild and laboratory strains (6), todate estimates of this parameter have been based on starvation as ameans to keep cells from dividing. Perhaps not surprisingly, thisapproach has identified genes acting in stress-response pathways asregulators of cell lifespan, notably TPK and BCY1, which togetherencode cAMP-dependent protein kinase TOR, which encodes target-of-rapamycin protein kinase and SCH9 (4, 8–10). One virtue of theapparent need for this approach is that calorie restriction (CR)promotes longevity in all species studied (11).However, many terminally differentiated metazoan cell types rarely

starve and some, such as neurons, can consume a disproportionate

share of an animal’s lifetime energy budget (12). Therefore, it isimperative to develop a different yeast model to study CLS, one inwhich nutrients are plentiful, but cell division ceases and metabolismremains highly active. Such a model would enable yeast to betterserve as a surrogate for postmitotic cells in higher eukaryotes, in-cluding humans. To model CLS under nutrient-replete conditions, weeschewed the traditional approach of culturing yeast as free-floating,planktonic cells in liquid media, in favor of encapsulating cells ina matrix made of calcium alginate. We reasoned that such a matrixwould more nearly approximate the physical milieu typically experi-enced by postmitotic metazoan cells. Previous studies have shown thatrelative to planktonic cells, encapsulated (i.e., immobilized) yeastexhibits increased glycolytic flux but greatly diminished production ofbiomass (13). These observations led us to investigate the possibilitythat encapsulated yeast could be used as a tool to study CLS in theabsence of CR.Immobilized yeast has been used for decades in brewing and

bioethanol production (14, 15), and indeed virtually all studies ofimmobilized yeast have been aimed at optimizing its use in in-dustrial processes. However, to the best of our knowledge, nostudy to date has brought the tools of functional genomics tobear on understanding the mechanism(s) by which metabolismmight be uncoupled from reproduction in such cells, much lessthe consequences of this uncoupling for yeast’s life history. Here,for the first time according to the authors, we compare global

Significance

All cells age and do so in relation to how many times a celldivides (replicative aging) and how long a nondividing cell canlive (chronological aging). Bakers’ yeast has been used to studyboth, but because yeast divides when nutrient levels permit,the genetics of its chronological lifespan has only been studiedunder calorie restriction, mimicked by starvation. Becausemany terminally differentiated animal cells are long-lived andrarely starve, we developed a model of cell lifespan undercalorie-unrestricted conditions. When encapsulated and fed adlibitum, yeast goes into cell cycle arrest, continues to be met-abolically active, and remains viable for weeks, offering a newexperimental paradigm to study chronological lifespan in theabsence of calorie restriction.

Author contributions: S.N., A.L.K., E.K., T.T., and F.R. designed research; S.N., A.L.K., K.H.S.,E.K., M.H., K.M., R.S., T.T., and F.R. performed research; A.L.K. contributed new reagents/analytic tools; S.N., K.H.S., K.M., T.T., and F.R. analyzed data; and S.N., K.H.S., E.K., and F.R.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The data reported in this paper have been deposited in the Gene Ex-pression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE21187).1To whom correspondence should be addressed. E-mail: frank.rosenzweig@mso.umt.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1323918111/-/DCSupplemental.

E1538–E1547 | PNAS | Published online March 31, 2014 www.pnas.org/cgi/doi/10.1073/pnas.1323918111

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

gene expression of immobilized yeast to that of planktonic cells,from the perspective of yeast chronological aging. We find thatnondividing yeast encapsulated in a permeable matrix and fedad libitum resembles in key respects postmitotic metazoan cells(16, 17), offering an alternative paradigm for studying mechanismsthat control CLS in all eukaryotes.

ResultsTo provide context for transcriptional profiling, we first comparedthe population structure and physiology of well-fed immobilizedyeast to that of planktonic yeast growing at different rates andnutrient levels. In batch culture, log-phase planktonic cells expe-rience excess nutrients and divide rapidly, whereas stationary-phase cells experience progressive nutrient depletion and ulti-mately undergo cell cycle arrest; in a chemostat, planktonic cellsare nutrient-limited and divide at a rate fixed by the dilution rate(18). Yeast was cultured in Synthetic Minimal (SM) medium (19)using glucose as the sole carbon source in batch, in chemostats,and in immobilized cell reactors (ICRs) (for reactor schematic andphoto, see Fig. S1 A and B). Because ICR yeast produces copiousCO2, purging the bioreactor of oxygen, planktonic cells were alsocultured under anoxic conditions, unless otherwise indicated.Unlike planktonic cells in batch, immobilized yeast cells were fedcontinuously: every 48 h as glucose fell below 5 g·L−1, the ICRfeed was replaced with fresh media containing 100 g·L−1 glucoseand an excess of micronutrients (Fig. S2).

Encapsulated Yeast Stops Dividing Under Nutrient-Sufficient Conditions.Compared with batch- and chemostat-grown cells, the density ofimmobilized yeast cultures plateaued after 6–7 cell divisions.Planktonic batch cultures were inoculated at a starting densityof ∼106 cells·mL−1, attained midlog phase 12 h postinoculation,then entered the stationary phase (∼108 cells·mL−1) 24 h later.Cells in chemostats reached the steady state (∼108 cells·mL−1)24 h postinoculation, then continued to divide at the set di-lution rate. Yeast encapsulated in ∼3 mm calcium alginatebeads (∼2.3 × 106 cells·bead−1) and packed into ICRs underwentmultiple cell divisions in the first 72 h of culture, formingmicrocolonies (Fig. S1D). Thereafter, mean cell density remainedconstant at ∼108 cells·bead−1. To discern replicative age structurein planktonic and immobilized populations, budding and bud scarindices were estimated. After 5 d, immobilized yeast populationshad fourfold fewer budded cells than planktonic yeast in batchculture (4.3% vs. 17.2%) (Fig. S3A). At that time, a majority(∼70%) of the immobilized populations consisted of virgin,unbudded cells, and even older mothers having >2 bud scars wereunbudded (Fig. S3B), further indicating that immobilized yeastproliferates only during the first few days of culture, then ceasesto divide.

Nondividing, Well-Fed Yeast Shows No Loss in Viability DuringProlonged ICR Culture. Yeast CLS is conventionally reported asthe survivorship of nondividing cells in extended batch culture,estimated by their reproductive potential—that is, their ability toform colonies when transferred to solid medium (8). Planktoniccells in glucose-limited chemostats divide as a function of dilutionrate (18), and under these conditions yeast showed >95% survi-vorship. In discontinuous batch culture, yeast quickly lost viability(Fig. 1A): after 5 d of anaerobic batch culture in SM medium,only 27% of cells could form colonies, and by day 10 and day 17,only 16% and ∼0.1% could do so, respectively. When yeast wasbatch-cultured under aerobic conditions, the decline in survivor-ship began ∼5 d later because these yeast could respire fermen-tation products following exhaustion of glucose. After 17 d,however, only ∼5% of cells in aerobic SM cultures could stillproduce colonies on solid medium, a value consistent with pre-vious reports (20–22). By contrast, >95% of immobilized yeastremained reproductively competent over a 17-d experimental

time course (Fig. 1A), indicating that division-arrested, meta-bolically active yeast cells are chronologically long-lived.To compare cell cycle status and DNA integrity in immobi-

lized and planktonic yeast, we stained cells with SYTOX Greenand performed flow cytometry (23). SYTOX fluorescence in-tensity is a direct measure of cellular DNA content and thus canbe used to track progress through the cell cycle. Consistent withbudding index data, some planktonic yeast in anaerobic batchculture were still dividing at day 5, as is evident from the secondhistogram peak corresponding to gap 2 (G2) in the cell cycle(Fig. 2A). The overall leftward shift in fluorescence intensity instarving planktonic populations (e.g., Fig. 2A, 17-d) may be evidenceof DNA degradation (24); indeed, it has been argued that such cellsbecome apoptotic (25). By contrast, flow cytometry confirmed thatimmobilized yeast fed ad libitum ceased replicating by day 5,was in G1-like arrest, and was not apoptotic (Fig. 2B).

Nondividing Immobilized Yeast Exhibits High Fermentative Capacityand Hyperaccumulates Glycogen. Even though ICR yeast ceaseddividing, extracellular glucose and ethanol concentrations con-tinuously changed throughout each experiment (Fig. S2B), in-dicating that these cells continued to be metabolically highlyactive. As the glucose concentration fell only briefly below5 g·L−1 and the ethanol concentration never exceeded 50 g·L−1,immobilized yeast experienced nutrient-sufficient conditionswith little or no ethanol inhibition. In triplicate short-term (72 h)experiments, ethanol yield on glucose by immobilized cells

Fig. 1. (A) Survivorship of immobilized yeast, anaerobic planktonic (freelysuspended) yeast, and aerobic planktonic yeast over 17 d culture. Survivor-ship was estimated as the ratio of CFUs on rich solid medium relative to totalcell number estimated by hemocytometry. Each experiment was repeatedtwice; error bars represent 1 SD from the mean value. (B) Glycogen contentof immobilized and planktonic yeast (milligram of glucose equivalents pergram of dry weight biomass). Each experiment was repeated twice; errorbars represent 1 SD (closed circle, immobilized, anaerobic; open triangle,planktonic, aerobic; closed triangle, planktonic, anaerobic).

Nagarajan et al. PNAS | Published online March 31, 2014 | E1539

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

(0.414 ± 0.012 g·g−1) was greater than that of log-phase planktoniccells (0.396 ± 0.007 g·g−1) (ANOVA, P < 0.05), whereas biomassyield on glucose was much less (0.006 ± 0.005 g dry wt ·g−1 vs.0.027 ± 0.001 g dry wt ·g−1). Ethanol productivity per gram of cellbiomass was therefore nearly fivefold greater in immobilized than inplanktonic cells, consistent with previous reports of very high gly-colytic flux in immobilized yeast (26). Thus, although immobilizedwell-fed yeast ceases to divide, it retains high fermentative capacity.Plentiful glycogen provides yeast with survival and reproductive

advantages (27). Indeed, CLS has been dramatically extended instarving planktonic cells by periodically exposing them to glucoseconcentrations that are sufficient to replenish glycogen reserves,but insufficient to support reproduction (28). Glycogen content inimmobilized yeast (Fig. 1B) greatly exceeded that observed ineither starving planktonic cells or in planktonic cells grown an-aerobically in glucose-limited chemostats. Glycogen levels rosefrom 0.7 mg of glucose equivalents per gram of dry wt at day 1 to4 mg of glucose equivalents at day 17, a >fivefold change. Thesteep increase in cells’ glycogen content commenced after day 3,the point at which immobilized yeast ceased to divide (Fig. 2B),suggesting that these cells reallocate resources to storage thatwould otherwise be directed toward growth and reproduction. Nocomparable increase was detected in yeast’s other major storagecarbohydrate, trehalose, whose increase under aerobic conditions

reportedly protects calorie-restricted planktonic cells from oxi-dative damage (29).In summary, population, physiological, and cytometric data

demonstrate that metabolism and reproduction are uncoupled inencapsulated, continuously fed yeast and that such cells remainviable for extended periods of time, features reminiscent of manytypes of postmitotic metazoan cells.

Patterns of Gene Expression in Immobilized Yeast Markedly Differfrom Those in Planktonic Yeast. To better understand how ICRyeast uncouples metabolism from reproduction and remainsviable in prolonged culture, we performed microarray analysisusing the Affymetrix platform. GeneSpring was first used toanalyze transcript levels in immobilized cells and planktonic cellsgrown either in chemostat or in batch culture, the latter har-vested at midlog and stationary phases. Overall, 4,308 of 5,804genes assayed were found to be differentially expressed (Fig.S4A). Considering only immobilized yeast, 2,375 transcripts weredifferentially expressed, although most of these differences couldbe attributed to using stationary-phase planktonic cells for en-capsulation. Immobilized yeast at day 0 was indistinguishablefrom stationary-phase yeast for >90% of transcripts assayed. Day10 immobilized cells were harvested just before the medium wasreplenished, which did not change the sign of but rather exag-gerated expression changes seen after day 3. Pairwise volcanoplot comparisons showed that only 79 genes varied by more thantwofold over the course of 2 wk of continuous immobilizedculture (Fig. S4B). Over this interval, immobilized yeast con-sumed ∼1.2 kg of glucose with little or no cell division.Next, the GeneSpring dataset was imported into GenMapp

and criteria set to uncover genes differentially expressed bytwofold or greater between immobilized cells and any of theplanktonic states. Day 0 immobilized cells were excluded fromGenMapp analyses, as their profile was essentially identical tothat of planktonic stationary-phase yeast, as were day 1 immo-bilized cells, which were still dividing. GenMapp data wereoverlayed onto Kyoto Encyclopedia of Genes and Genomes(KEGG) pathway maps for glycolysis, the tricarboxylic acid (TCA)cycle, and the yeast cell cycle. Consistent with observations thatcontinuously fed immobilized yeast is highly fermentative, anaer-obic, and not dividing (30), we found that many glycolytic genes,such as HXK2, PFK2, and PGK, were expressed highly, whereasTCA cycle genes, such as CIT1 and ACO1, were expressed at lowlevels (Fig. 3A). Strikingly, multiple genes whose products helpdrive transitions between G1, S, and G2 phases of the cell cyclewere down-regulated relative to every planktonic state assayed(Fig. 3B and the full KEGG cell cycle pathway in Fig. S5). Lowtranscript levels of cyclin CLN1 and master cyclin-dependentkinase CDC28 were consistent with flow cytometric and de-mographic analyses indicating that immobilized yeast is in a stateof G1-like arrest.Two-class SAM reveals immobilization-specific changes in gene expression.A two-class Significance Analysis of Microarrays (SAM) (31) wasundertaken by grouping batch-log, batch-stationary, and chemostatexpression data into a single planktonic group, then comparingtheir expression profiles to that of an immobilized group consistingof days 1 through 17. Although planktonic cells in chemostat,batch-log, and batch-stationary cultures obviously differ in theirnutritional states, growth rates, and gene expression profiles (Fig.S4A), comparing their aggregated expression patterns to those ofimmobilized cells uncovered classes of genes whose altered regu-lation could be attributed to immobilization per se. SAM identi-fied 379 significantly up-regulated and 204 down-regulated genesin immobilized relative to planktonic yeast (Table S1). To discernits dominant features, the two-class SAM dataset was further in-terrogated by K-means clustering and by Gene Ontology (GO)analysis using Database for Annotation, Visualization, and In-tegrated Discovery (DAVID).

Fig. 2. Cell cycle status of (A) planktonic and (B) immobilized yeast through17 d cultures. DNA content of planktonic batch cultured yeast and immo-bilized yeast released from its alginate matrix was assayed by flow cytometryof ethanol-fixed, SYTOX Green-stained cells. SYTOX Green is measured inmean fluorescence intensity. Red and black lines indicate fluorescence pro-files of independent biological replicates. Insert displays fluorescence pro-files of actively dividing planktonic haploid and diploid yeast, which displayprominent G1 and G2 peaks indicative of mitotic activity.

E1540 | www.pnas.org/cgi/doi/10.1073/pnas.1323918111 Nagarajan et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

K-means clustering indicates that immobilizing yeast favors cell wallremodeling, anaerobic fermentation, and cell cycle arrest. Among genesup-regulated in immobilized but not in planktonic cells, K-meansclustering associated genes whose products act in spore wall as-sembly (Fig. S6A), whose products localize to the cell membrane(Fig S6B), and that lie near telomeres (Fig. S6C). Conspicuouslyup-regulated are subtelomeric loci in the PAU seripauperin genefamily, whose functions are unknown, and subtelomeric loci in the

TIR cell wall mannoprotein gene family; both families have beendescribed as being activated during anaerobic alcoholic fermen-tation (32, 33). Down-regulated clusters illuminate how immo-bilization uncouples reproduction from metabolism. One cluster(Fig. S6D) consisted of genes in nucleotide and amino acid syn-thesis, and the cell cycle, notably SWI5, a transcription factor thatcontrols genes expressed at the M/G1 boundary and in late G1(34). Another (Fig. S6E) included genes that act in DNA repli-cation, chromosome segregation, and the spindle checkpoint,among them the forkhead family transcription factor, FKH1,which plays a key role in cell cycle progression at G2/M (35). Wealso observed down-regulation of RAS1 (Fig. S6E), a G proteinessential for cell proliferation linked to the PKA-activating ade-nylate cyclase pathway (36). A third cluster of down-regulatedtranscripts (Fig. S6F) included those whose products play essen-tial roles in iron assimilation and metabolism and in cytokinesis(e.g., HOF1, UTR2) (37, 38).GO analysis further confirms that immobilized yeast remodels its cell wall,up-regulates glycolysis, and ceases to divide. The two-class SAM datasetwas further analyzed using DAVID, which clusters functionallysimilar GO terms into groups ranked by enrichment scores, witha score >1.3 being considered statistically significant (39). Thisprocedure uncovered seven gene clusters that were significantlyup-regulated in immobilized but not in planktonic yeast—cell wall(enrichment score, 6.3), cell membrane (3.5), spore wall (3.4),sporulation (2.5), zinc-finger proteins (2.0), GPI anchor (1.8), andthiamine biosynthesis (1.8)—as well as three clusters that weresignificantly down-regulated—cell cycle (2.0), chromosome segre-gation (1.6), and ribosome biogenesis (1.3) (Table S2).Of 50 genes up-regulated in the cell wall (Fig. 4A) and spore

wall (Fig. S7) clusters, 24 are involved in cell wall biogenesis.Altered expression of so many cell wall remodeling transcriptssuggests that immobilization induces the Cell Wall Integrity(CWI) pathway. In this regard, RPI1 up-regulation in immobi-lized but not in planktonic cells is especially noteworthy (Fig.S7C). Rpi1 acts as an antagonist to the RAS–cAMP pathway andprepares yeast for entry into the stationary phase by inducingtranscription of genes whose products fortify the cell wall (40).The TIR and DAN gene families, which encode cell wall man-noproteins and are induced under anaerobic conditions (41),were conspicuously up-regulated in ICRs. However, as plank-tonic yeast was also cultured under anaerobic conditions, andTIR and DAN genes were only up-regulated in immobilized cells,these genes must be specifically induced in response to beingencapsulated and maintained in ICRs.Dramatically increased expression of essential genes in asco-

spore biogenesis was observed in immobilized, but not in plank-tonic, yeast (Fig. 4A). These included SPR1 and SPS22, whoseproducts are required to synthesize the spore wall’s β-glucan layer(42); SPR3 and SPO19, which respectively encode meiosis-specific septin and prospore proteins (43); and CRR1, a glucosidehydrolase essential for spore wall assembly. Induction of thesegenes indicated overlap between the transcriptional programs ofwell-fed, nondividing vegetative cells and spores. Indeed, mostgenes belonging to the sporulation cluster (Fig. 4B) encode sporemembrane and spore wall proteins—for example, OSW1—whichis required for construction of the outer spore wall. The over-expression in immobilized haploid yeast of genes involved inspore formation was unexpected, as our strain carries mutationsin HO and MAT that prevent mating-type switching. Moreover,flow cytometry showed that ICR yeast remained haploid; thus,mating, diploidization, and meiosis did not occur en masse inICRs. However, the sporulation cluster also showed that immo-bilized yeast had elevated transcript levels for MSN4 and RIM15that encode, respectively, a stress-responsive transcription factorand a glucose-repressible protein kinase required for nutrientsignaling, especially during the onset of the stationary phase.RIM15 was originally identified in a genetic screen for mutants

Fig. 3. Comparison of gene expression in immobilized relative to plank-tonic yeast: (A) intermediary metabolism and (B) cell cycle. Differentiallyexpressed transcripts identified using GeneSpring were overlayed onto theS. cerevisiae KEGG biological pathway maps using GenMapp software. De-tailed KEGG pathway cell cycle results are presented in Fig. S5. Transcriptlevels in immobilized well-fed yeast for days 3, 5, 10, and 17 were averagedacross replicate experiments. Red indicates instances where those values wereat least twofold greater in immobilized cells than in planktonic cells. Greenindicates instances where those values were at least twofold less in immobi-lized cells than in planktonic cells. White indicates that the gene was not foundamong the significant gene list identified earlier using GeneSpring. Graydenotes instances where transcript levels differed by less than twofold.

Nagarajan et al. PNAS | Published online March 31, 2014 | E1541

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

having a reduced ability to undergo meiosis (44), and was latershown to transiently induce early meiotic genes (45). Thus,RIM15 up-regulation may help explain induction of certain mei-otic genes in well-fed, nondividing haploid yeast.Zinc-finger proteins formed another GO cluster (Fig. S7B),

members of which include MSN4, which acts in concert withMSN2 to activate transcription of stress-resistance genes (46).Significantly, the cytoplasmic-to-nuclear translocation of MSN2/MSN4 gene products depends not only on stress and on PKA(47) but also on Rim15 (48). Although TPK/BCY1 was not up-regulated in immobilized yeast, both RIM15 and MSN4 were(Fig. 4B), showing that immobilization per se activates these keystress response regulators (46, 49). Two other zinc-finger pro-teins in this cluster, NRG2 and MIG2, are transcription factorsthat negatively regulate expression of glucose-repressible genes.Up-regulation of NRG2, MIG2, and MIG3 in immobilized cells(Fig. S7B) likely reflects high glucose levels in their externalmilieu. Immobilization was also correlated with increased ex-pression of transcription factor TYE7 (Table S1). Although nota zinc-finger protein, TYE7 binds E-boxes of glycolytic genesand contributes to their activation (50). Thus, its differentialexpression helps to explain elevated levels of ENO2, PGK1, andGPM1 (Fig. 3A), whose products help support immobilized

yeast’s high rate of fermentation. Another gene of interest in thezinc-finger cluster is YPR015C, which encodes a protein of un-known function whose overexpression leads to cell cycle delay orarrest (51). Increased expression of YPR015C may be yet an-other factor contributing to cell cycle arrest in highly fermenta-tive ICR yeast.Consistent with other analyses, cell cycle (Fig. S8A) and chro-

mosome segregation (Fig. S8B) clusters were both down-regulatedin immobilized yeast, including genes such as HOF1 and BUD4,which encode structural proteins required for axial budding (52),and ACE2, which encodes a transcription factor required forcytokinesis (53). Expression of 22 genes related to ribosomebiogenesis (Fig. S8C) was decreased in immobilized relative toplanktonic yeast. As ribosomal gene expression is directlyproportional to growth rate (54), low transcript levels heremay result from G1-like arrest (54).To validate our microarray results, we performed quantitative

reverse-transcription PCR (qRT-PCR) to assay transcript levelsfor three key regulatory genes (RIM15,MSN4, and TYE7) and oneconstitutively expressed gene, ACT1, which encodes the structuralprotein actin. qRT-PCR results for all three regulatory genes,normalized to ACT1 expression, were consistent with array results,with correlation coefficients ranging from 0.68 to 0.93. If anything,microarray results tended to underestimate the magnitude ofchange in expression (Fig. S9).

Immobilized Yeast Is Thermotolerant and Zymolyase Resistant. Totest whether up-regulation of cell wall remodeling genes (e.g.,CRR1 and RPI1) and stress response regulators (e.g., RIM15 andMSN4) made immobilized yeast more stress resistant thanplanktonic yeast, we tested its susceptibility to heat shock andzymolyase digestion. Thermal tolerance is often used as a proxyfor stress resistance (55), which in yeast has been associated withCLS extension (56). Rapidly proliferating log-phase yeast isknown to be more heat-shock sensitive than either growth-arrested (57) or slowly dividing yeast (58). We therefore com-pared the thermotolerance of cells either confined to or releasedfrom their calcium–alginate matrix to planktonic cells in the logor stationary phase by exposing each to 48 °C for 2 h (Fig. 5A).Immobilized and stationary-phase planktonic yeasts were sig-nificantly more heat-shock tolerant than planktonic yeast in thelog phase (P = 0.00964 and P = 0.00637, respectively, by two-tailed t test). Thermal tolerance in immobilized cells was due notonly to their physiological status, but also to the presence of thealginate matrix, as evidenced by the intermediate survivorship ofcells released from beads.Yeast thermal tolerance is strongly correlated with changes in

cell wall structure that occur in planktonic cells following spor-ulation or entry into the stationary phase (57). To test whetherimmobilized yeast underwent similar changes in cell wall struc-ture, as suggested by its increased transcription of cell wall andsporulation genes, we assayed its resistance to zymolyase (Fig.5B), a glucanase–protease suspension that enables the state andintegrity of the yeast cell wall to be assessed qualitatively. Asexpected (59), midlog-phase planktonic cells were most susceptible,lysing at a higher rate (Vmax = –11.7) and to a greater extent overthe experimental time course. Late stationary planktonic and day5 immobilized yeast lysed at similar rates (Vmax = –5.49 and –5.81,respectively) and to the same extent. Immobilized cells were muchmore susceptible to zymolyase digestion on days 1 and 3, when theywere still dividing, than on days 5, 10, and 17, when they were not.

Rim15 Helps Mediate Cell Cycle Arrest in Immobilized Yeast. Immo-bilized yeast undergoes cell cycle arrest in the presence of excessnutrients. Because RIM15 was up-regulated in immobilized butnot in planktonic cells, because RIM15 can play a role in cellcycle arrest, because multiple nutrient-sensing pathways con-verge on Rim15p (48), and because it promotes chronological

Fig. 4. Functional annotation clustering using DAVID of gene expressiondifferences in immobilized relative to planktonic yeast. Groups with signif-icant (>1.3) enrichment scores include (A) cell wall (score, 6.3) and (B) spor-ulation (2.5).

E1542 | www.pnas.org/cgi/doi/10.1073/pnas.1323918111 Nagarajan et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

longevity in calorie-restricted yeast (8, 10), we hypothesized thatthis master regulator played a key role in uncoupling metabolismfrom reproduction. To test this hypothesis, we immobilized arim15Δ strain and cultured it in ICRs for 5 d under the sameconditions as wild type. Flow cytometry of DNA content revealedthat unlike wild type, rim15Δ cells exhibit a pronounced G2 peakat later time points, indicating that they continued to divide (Fig.6). Moreover, at day 5, only ∼25% of the rim15Δ cells were viable,compared with >90% of immobilized wild-type cells. We there-fore conclude that RIM15 helps mediate cell cycle arrest andstress resistance in well-fed immobilized yeast, attributes thatcontribute to its extraordinary chronological longevity.

DiscussionCLS in yeast is measured as the survivorship of nondividing cellsover time. Because yeast usually grows and divides when nutrientspermit, to date CLS has only been estimated under starvation orseverely nutrient-limited conditions (9, 22, 60). Thus, the CLS ofdivision-arrested yeast under nutrient-replete conditions is com-pletely unknown. Here, we describe a model system where cal-cium alginate-encapsulated yeast are placed in bioreactors andfed ad libitum. Under these calorie-unrestricted conditions, yeastceases to divide, remains metabolically active, and exhibits nodecline in viability over the course of 2 wk of continuous culture.Our findings are consistent with earlier studies suggesting thatimmobilized cells might be stress-resistant (61, 62) and long-lived(63). Furthermore, survivorship of nondividing immobilized yeast

exceeds that reported for calorie-restricted cells, even that ofseveral mutants known to prolong CLS (10, 64).

Growth-Arrested Immobilized Yeast Is Highly Fermentative. Becauseimmobilized yeast vigorously ferments glucose during prolongedculture, it has been used for ethanol production, reportedlyattaining, relative to planktonic cells, >10-fold increases in volu-metric ethanol productivity at high substrate concentrations (65).However, whereas high glycolytic flux typically triggers expressionof genes required to pass through G1 (66), immobilized fer-mentative yeast does not respond to this signal. Indeed, RAS1,which encodes a component of the RAS/cAMP/PKA glucose-signaling pathway, is down-regulated in immobilized relative toplanktonic cells, and the gene encoding the transcriptional reg-ulator Rpi1, which acts antagonistically to RAS/cAMP/PKA, isup-regulated (Fig. S6E and Fig. S7C). Because RPI1 also regu-lates the CWI pathway, helping to prepare cells for entry into thestationary phase (40), elevated expression of this gene in highlyfermentative ICR yeast may contribute to its heat-shock- andzymolyase-resistant phenotype.In starved batch cultures, yeast CLS depends on respiration,

not fermentation (67), as long-lived quiescent cells respire non-fermentable carbon that arises from autophagous nonquiescentcells (68). Immobilized yeast under calorie-unrestricted con-ditions exhibits long CLS in the absence of respiration, which isprecluded in ICRs by glucose repression and low O2 tension.Interestingly, long-lived, planktonic tor, sch9, and ras mutantsexhibit a metabolic shift away from respiration toward fermen-tation and increase glycerol production (69). Although encap-sulated well-fed yeast is also fermentative and produces glycerol,our experiments were performed using wild-type cells undercalorie-unrestricted conditions, not planktonic mutants undersevere CR. Finally, immobilized fermentative yeast also producesFig. 5. (A) Heat shock tolerance and (B) zymolyase resistance in immobi-

lized versus planktonic yeast. Survivorship following exposure to 48 °C wasestimated as the ratio of CFUs to total cell number enumerated byhemocytometry. Digestion of the cell wall was followed by a change inoptical density at 660 nm. Open triangle, planktonic stationary phase;closed triangle, planktonic log phase; closed circle, immobilized cells in calcium–

alginate matrix; open circle, immobilized cells released from the matrix.

Fig. 6. Flow cytometry of immobilized wild-type and rim15Δ yeast over 5 dof continuous culture: (A) wild type and (B) rim15Δ. Immobilized wild-typeand rim15Δ cells were released from their calcium–alginate matrix, stainedwith SYTOX Green, and analyzed by flow cytometry. SYTOX Green is mea-sured in mean fluorescence intensity. Insert displays fluorescence profiles ofactively dividing haploid and diploid yeast, which display prominent G1 andG2 peaks indicative of mitotic activity.

Nagarajan et al. PNAS | Published online March 31, 2014 | E1543

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

acetate, which has been implicated in causing planktonic yeastto senesce under standard conditions (70). However, in ICRexperiments, the “acetate effect” does not come into play, as themedium is exchanged frequently and its pH is controlled at 5.5.

Why Does Immobilized Yeast Cease Reproducing in the Presence ofExcess Nutrients? Possible causes for cell cycle arrest in ICR yeastinclude the following: copious CO2 production (71), ethanoltoxicity (72), substrate and product mass transfer limitations,contact inhibition within cell aggregates, and/or hyperbaric stressarising from expansion of microcolonies into an imperfectlyelastic bead matrix. We tested the possibility that copious CO2production caused cell cycle arrest by culturing planktonic cellsin medium sparged with CO2; these cells exhibited no evidenceof growth arrest. Concerning ethanol, this metabolite varies cy-clically in ICRs, but because the feed is replaced at least every48 h, its concentration never exceeds 50 g·L−1, well below thethreshold reported to cause growth arrest in fed-batch systems(80–100 g·L−1) (72). Mass transfer limitations are unlikely tocause growth arrest given the bead composition and geometry, aswell as the high rate of ICR void volume exchange. Calculationsusing diffusivity values for glucose and ethanol through algi-nate indicate there should be no mass transfer limitations inbeads <3.5 mm in diameter (73). In this regard, it is noteworthythat the biomass-specific rate of ethanol production by immo-bilized cells is several-fold greater than that of log-phaseplanktonic cells, and this high fermentation rate is maintainedfor the duration of the experiments. Concerning possible contactinhibition, immobilized cells highly express genes in the FLOfamily of cell wall glycoproteins (Fig. 4A and Table S1). BecauseFLO adhesion proteins are involved in cell aggregation and stressresistance (74), their increased expression may facilitate micro-colony formation in the alginate matrix, a feature that may favorincreased glycolytic flux (75). Indeed, scanning electron micros-copy provides evidence for the existence of such structures (Fig.S1D), wherein yet-to-be discovered quorum sensing pathway(s)may help transduce information about local cell density intosignals that mediate growth arrest.

Growth Arrest and Longevity in Immobilized Yeast May Be Related toDifferential Expression of RIM15. Even though the ultimate causefor cell cycle arrest in encapsulated cells remains elusive, ourdata indicate that key signal(s) may be transduced via Rim15.Relative to planktonic cells, immobilized wild-type yeast highlyexpress this gene and its downstream target MSN4. Unlike wildtype, immobilized rim15Δ cells continue to divide and exhibitmuch lower viability under nutrient-sufficient conditions. RIM15encodes a glucose-repressible serine–threonine protein kinasethat acts in signal transduction pathways responding to depletionof extracellular carbon, nitrogen, and phosphorus. Originallyidentified as a regulator of IME2, which stimulates early meioticgene expression (and which is also up-regulated in ICRs) (45),RIM15 is crucial to the establishment of the stationary phase,especially the quiescent G0 state (76). Multiple nutrient-sensingpathways (TOR, SCH9, RAS/PKA) converge on Rim15, which inturn positively regulates stress-resistance transcription factorsMsn2/Msn4 and Gis1 (48). Loss of RIM15 results in a pleiotropicphenotype that fails to enter the stationary phase; poorly accu-mulates trehalose and glycogen; derepresses HSP12, HSP26, andSSA3; and cannot induce either thermotolerance or starvationresistance (77).Whereas previous work has shown how RIM15 is induced

when nutrients are lacking, in ICR yeast we see RIM15 andits downstream target MSN4 induced when nutrients are abun-dant. If, as some suggest, Rim15 is activated by glucose uptakekinetics rather than by absolute glucose levels (76), then periodicvariation in extracellular glucose in ICRs may help explainRIM15’s unexpected pattern of expression. Alternatively, Rim15

up-regulation under nutrient-sufficient conditions may result fromsignal decoupling in one or more of the nutrient sensor kinasepathways, a hypothesis that can be tested by assaying the phos-phorylation status of kinases such as PKA, Sch9, the Pho80–Pho85complex, as well as that of Rim15 itself. RIM15 confers thermaland oxidative stress resistance on calorie-restricted, chronologi-cally long-lived planktonic yeast (10, 77). And indeed, the bene-ficial effects of CR on CLS can be negated either by deletion ofRIM15 or by deletion of MSN2/MSN4 (10). Thus, experimentalconditions that promote RIM15 overexpression could be expectedto increase stress resistance and extend CLS. Well-fed encapsu-lated rim15Δ cells exhibit greatly reduced survivorship relative tothat of encapsulated wild-type cells (25% vs. >90%). In fact,survivorship of encapsulated rim15Δmutants is comparable to thatof starved planktonic rim15Δ mutants in extended batch culture(20%) (77). Reduced viability of the deletion strain may be at-tributed to its inability to completely arrest, which leads to repli-cative stress (24, 78), and/or to diminished stress resistance, whichis essential for stationary-phase survival. Given that RIM15 ishighly expressed in encapsulated but not in planktonic yeast, andgiven that encapsulated wild-type but not rim15Δ cells cease toreproduce and show extended CLS, we conclude that differentialexpression of RIM15 facilitates cell cycle arrest and increasesstress resistance in immobilized yeast.

Glycogen Accumulation May Help Extend CLS in Immobilized Yeast. Inhigh-glucose medium, respiratory-deficient strains accumulateglycogen and mobilize it upon glucose depletion (79). AlthoughICR cells are not glucose-limited, they are effectively respiratory-deficient due to anoxic reactor conditions, which helps to explainwhy they hyperaccumulate glycogen. Also, because loss of RIM15prevents stationary-phase cells from accumulating glycogen (77),overexpression of RIM15 in immobilized cells may favor glyco-gen accumulation. The role that reserve storage carbohydratesplay in chronological cell aging remains controversial. Althoughsome (80) view their accumulation as a hallmark feature of se-nescence, others (28) have argued that periodic replenishment ofglycogen reserves can dramatically extend yeast CLS. Our find-ings are consistent with the latter hypothesis.

Other Conditions Where Reproduction Is Uncoupled from Metabolismin Yeast. Uncoupling of reproduction from metabolism has beenobserved in planktonic yeast during high-glucose, aerobic, fed-batch bioethanol fermentation (72). Here, uncoupling is attributedto ethanol (above an 80–100 g·L−1 threshold), which impairsethanol-sensitive nutrient transporters, including those for glu-cose and ammonium. In effect, transporter inhibition preventsnutrient acquisition. While encapsulated, continuously fed yeastalso uncouples reproduction from metabolism, the underlyingcause is unlikely related to ethanol stress, as its medium is fre-quently exchanged so that ethanol never exceeds 50 g·L−1. Further,only 45% of “ethanol-uncoupled” planktonic cells remain viablecompared with immobilized cells, which are essentially all viableeven after being “uncoupled” for ∼2 wk (81). Expression profilingreveals that ethanol-uncoupled yeast up-regulates SNF1, ADR1,ASP3-1, and ASP2-1, indicating that, unlike the yeast in our study,these cells are sugar- and nitrogen-limited (81). Finally, in ethanol-uncoupled cells, genes involved in ribosomal biogenesis are up-regulated, whereas genes involved in cell wall remodeling aredown-regulated. We observe exactly the opposite in immobi-lized cells; thus, the mechanisms that cause uncoupling in thesetwo systems must be fundamentally different.After planktonic cells exhaust glucose in the medium, they

differentiate into several subtypes, one of which is quiescent.Quiescent cells consist primarily of daughter cells that havenever budded, possess high stress tolerance, and retain highviability during prolonged periods of starvation (82). Thesecells seem poised to resume growth when resupplied with

E1544 | www.pnas.org/cgi/doi/10.1073/pnas.1323918111 Nagarajan et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

nutrients. Thus, quiescent cells in certain respects resembleencapsulated cells. However, whereas quiescent yeast meetsminimum energy requirements via respiration, encapsulatedyeast is highly fermentative. Both quiescent and encapsulatedcell types possess certain characteristics of metazoan cells,with quiescent cells being similar to stem cells (82) and well-fed encapsulated cells being similar to postmitotic cells thatare terminally differentiated.Metabolic uncoupling occurs in yeast colonies growing on

solid media, and does so in a manner that leaves older coloniesstratified with respect to cell types that differ in activity andlongevity (83). Similar to immobilized yeast, upper colony or“U-cells” are long-lived, resist stress, and accumulate glycogen.Also, like immobilized cells, but unlike quiescent cells, U-cellrespiratory metabolism is down-regulated. U-cells up-regulatenot only glycolysis but also pathways related to amino acidmetabolism that enable them to survive on resources released bythe lysis of lower “L-cells.” TOR pathway up-regulation inU-cells, but not in well-fed immobilized cells, indicates that theformer are nutrient-limited whereas the latter are not.Finally, yeast uncouples metabolism from reproduction when

cultured in retentostats. In this system, planktonic cells arephysically confined to a bioreactor operated under continuousglucose limitation at near-zero dilution rates (60, 84); such cellsshow exceptional longevity (∼80% viability after 2 wk). Sig-nificantly, although ICR yeast is metabolically very active andretentostat yeast is not, and although the expression profilesunderlying these two states differ in certain respects, bothhyperaccumulate glycogen but not trehalose and both highlyexpress RIM15. Thus, these two systems could provide com-plementary approaches to study division-arrested yeast CLS inthe presence and in the absence of CR.

An Alternative Paradigm for the Study of Aging in Postmitotic Cells.To date, all studies of yeast CLS have focused on postmitoticviability under starvation (22) or near-starvation conditions (60).Although yeast CLS is held to mimic that of postmitotic meta-zoan cells (85), such cells are rarely calorie-restricted and manyremain metabolically active throughout an animal’s lifetime.Terminally differentiated cells such as neurons and muscle arrestin G1 and uncouple metabolism from reproduction in the pres-ence of ample nutrients. Differentiated myotubes show dramaticloss of core factors in their transcriptional machinery, whichresults in large-scale silencing of genes that control cell pro-liferation and cell cycle progression (86). Ideally, to model thelifespan of such cells, yeast should be studied under calorie-unrestricted conditions. In practice, this has been difficult to ach-ieve as yeast divides when nutrient levels permit. Here we haveshown that well-fed yeast confined within a matrix shuts downtranscription of genes that drive the cell cycle, even as it con-tinues to perform glycolysis at near maximal rates. Becausereproduction is uncoupled from metabolism, resources thatwould otherwise be allocated to growth and cell division areredirected to storage and maintenance, a shift in yeast’s life his-tory that greatly extends its CLS. In fact, how long a nondividingyeast cell can live under nutrient-replete conditions remains anopen question. ICRs therefore offer an alternative experimentalparadigm for investigating yeast CLS in the absence of CR, onethat will yield fresh insights into the mechanisms that control celllifespan in higher eukaryotes, including humans.

Materials and MethodsStrains and Culture Media. Saccharomyces cerevisiae BY4722 (MATα leu2Δura3Δ0) was obtained from the American Type Culture Collection (ATCC No.200884); a MAT-a leu2Δ0 ura3Δ0 rim15Δ derivative congenic to BY4730 wasobtained from Open Biosystems. Culture media used in ICR, batch, andchemostat studies was the SM medium described by Verduyn et al. (19),modified for anaerobic culture. Because previous studies (21, 87) had shown

that starvation for auxotrophic requirements accelerates cell aging, thesewere provided in fivefold excess to planktonic cells. As the medium feedingICRs was regularly replenished, auxotrophic requirements were providedthere at recommended concentrations (88). Details of routine strain storageand maintenance are described in SI Materials and Methods.

Preparation of Alginate Encapsulated Cells. Early stationary yeast was encap-sulated in high-viscosity sodium alginate to form ∼3 mm beads, as described inSI Materials and Methods. Beads were hardened by incubation in fresh, sterile0.5 M CaCl2, then packed to a volume of 100 mL in a custom-built ICR.

Structural and Operational Features of the ICR. A schematic of the ICR used inthis study is shown in Fig. S1A. The Pyrex glass reactor column had a radius of2.7 cm and a working length of 40 cm. Overall, the apparatus was similar tothat described by ref. 89, modified by placing an in-line cell trap to removeplanktonic cells arising from ruptured beads, and a stainless-steel piston atthe base to permit bed recompression following sampling. Operationalfeatures are described in detail in SI Materials and Methods.

Planktonic Cell Studies. Planktonic cells were cultured in either batch orchemostat mode using an ATR SixFors fermentation apparatus (ATR Bio-technologies) with the reactor working volume set at 300 mL Anaerobic con-ditionsweremaintainedby sparging cultureswith sterile-filtered, humidifiedN2.Batch cultures were initiated with 2 g·L−1 glucose; chemostat cultivation wascarried out at 9 g·L−1 glucose with the dilution rate, D, fixed at 0.15 h−1.Temperature and pH were maintained at 32.5 °C and 5.5, respectively.

Sampling Procedures. Detailed procedures for sampling planktonic andimmobilized cell cultures are described in SI Materials and Methods.

Demographic Parameters: Population Density, Viability, Budding Index, andReplicative Age Structure. Planktonic cells were enumerated either byhemocytometry or spectrophotometry at A600. Viability was determined byplating known dilutions onto yeast extract, peptone, dextrose (YEPD) agar, thencounting colony-forming units (CFU) following 72 h of incubation at 30 °C.Following Calcofluor staining (90), budding index and bud scar analyses wereperformed by differential interference contrast and epifluorescence microscopy.For immobilized cells, demographic parameters were estimated on ICR yeastfollowing their release from the calcium–alginate matrix by using sodiummetaphosphate to chelate Ca2+, the cross-linking agent. Additional detailsare provided in SI Materials and Methods.

Assay of CLS. CLS was evaluated as the time-dependent survivorship ofnondividing cells, as in ref. 22. Survivorship was estimated in starvingplanktonic batch cultures and in ICRs as the ratio of CFUs to total cell countobtained by hemocytometry.

Metabolite Determinations. Glucose and ethanol were determined eitherusing a YSI 2700 analyzer (YSI Corp.) or, spectrophotometrically, usingR-BIOPHARM enzyme kits (no. 716251 and no. 176290). Glycogen and tre-halose were determined as described in ref. 91.

Resistance to Heat Shock and Zymolyase Digestion. Thermotolerance wasassayed as the change in cell survivorship over the course of a 2 h incubationat 48 °C, estimated as CFUs on YEPD agar. Susceptibility of the yeast cell wallto zymolyase digestion was assayed by the method of Ovalle et al. (92), butscaled for a 96-well format, as described in SI Materials and Methods.

Flow Cytometric Analysis of DNA Content and Integrity. Cell cycle status andDNA integrity were assessed in planktonic yeast and immobilized yeast re-leased from calcium alginate by staining ethanol-fixed cells with SYTOX Greenand performing flow cytometry as previously described (23). Flow cytometricdata were processed using FSC Express software (De Novo Software).

RNA Isolation. Total RNA was isolated from planktonic cells using the hot acidphenol method (93) and from immobilized cells using a protocol developedin our laboratory and described in ref. 94, then analyzed by UV spectroscopyand Bioanalyser 2100 (Agilent Technologies).

Microarray Hybridization and Scanning. Microarray analysis was performedusing Yeast GenomeChip 2.0 arrays (AFFYMETRIX Inc.), using standardmethods described in SI Materials and Methods. Microarray data havebeen deposited in the Gene Expression Omnibus repository under accessionno. GSE21187.

Nagarajan et al. PNAS | Published online March 31, 2014 | E1545

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

GeneSpring Analysis. Microarray data were first analyzed using GeneSpringsoftware version 7.3, as described in SI Materials and Methods. A scatterplotof robust microarray-normalized (95) expression data was generated frombiological replicates sampled on days 1, 3, 5, and 10 from ICRs. Analysis ofthis dataset showed that replicates from the same time point had Pearsoncorrelation coefficients ranging from 0.85 to 0.98 (Mean = 0.92, SE =0.0085). Minimal deviation was observed from the diagonal identity lineamong replicates, further indicating that our array results are highly re-producible (94).

GenMapp Analysis. The set of statistically significant genes identified usingGeneSpring 7.3 was imported into GenMapp software (GenMapp Version 2.1,Gladstone Institute, University of California, San Francisco) and criteria fixedto identify genes differentially expressed by a twofold-change thresholdbetween immobilized cells and any planktonic state. Expression data for ICRyeast on day 0 and day 1 were excluded from this analysis, as day 0 transcriptlevels were indistinguishable from those of planktonic stationary-phase cellsused for encapsulation and day 1 cells were still dividing. GenMapp data wereoverlayed onto KEGG pathway maps (96) for glycolysis, the TCA cycle, andthe yeast cell cycle (97).

SAM was performed using TIGR (MEV) TM4 software based on ref. 31. Anunpaired two-class SAM analysis was adopted to identify genes differentially

expressed between the planktonic and immobilized states. Additionaldetails are provided in SI Materials and Methods.

Functional Enrichment. Significant genes were subjected to GO term analysisand functional enrichment clustering using the DAVID software, available asa web application (http://david.abcc.ncifcrf.gov/) (39, 98). As DAVID does notsupport yeast systematic names found at the Saccharomyces GenomeDatabase (www.yeastgenome.org), Entrez GeneIDs were retrieved for sig-nificant genes and the analysis carried out using Entrez GeneIDs. qRT-PCR wasperformed on RIM15, MSN4, TYE7, and ACT1 using the BioRad MyiQ Real-Time PCR System (Biorad), using primers and conditions described in SIMaterials and Methods.

ACKNOWLEDGMENTS. We thank Matt Herron, Margie Kinnersley, PaulLevine, Gavin Sherlock, and Art Woods for useful discussions; GretchenMcAffrey for editorial comments; Pam Shaw for assistance in flowcytometry; and Jim Driver for electron microscopy. This work wassupported by grants from the National Institutes of Health (NIH)(GM079762-01 and R15 GM79762-01A1) (to F.R.), the University ofMontana’s Office of Sponsored Research, the Montana NIH–BiomedicalResearch Infrastructure Network program, and the University of MontanaFluorescence Cytometry Core Facility, which is in part supported by NIHGrant P20RR017670.

1. Fredriksson A, Nyström T (2006) Conditional and replicative senescence in Escherichiacoli. Curr Opin Microbiol 9(6):612–618.

2. Stephens C (2005) Senescence: Even bacteria get old. Curr Biol 15(8):R308–R310.3. Kaeberlein M, et al. (2005) Regulation of yeast replicative life span by TOR and Sch9 in

response to nutrients. Science 310(5751):1193–1196.4. Kaeberlein M (2010) Lessons on longevity from budding yeast. Nature 464(7288):

513–519.5. Stumpferl SW, et al. (2012) Natural genetic variation in yeast longevity. Genome Res

22(10):1963–1973.6. Qin H, Lu M (2006) Natural variation in replicative and chronological life spans of

Saccharomyces cerevisiae. Exp Gerontol 41(4):448–456.7. Smith ED, et al. (2008) Quantitative evidence for conserved longevity pathways be-

tween divergent eukaryotic species. Genome Res 18(4):564–570.8. Fabrizio P, Pozza F, Pletcher SD, Gendron CM, Longo VD (2001) Regulation of lon-

gevity and stress resistance by Sch9 in yeast. Science 292(5515):288–290.9. Powers RW, 3rd, Kaeberlein M, Caldwell SD, Kennedy BK, Fields S (2006) Extension of

chronological life span in yeast by decreased TOR pathway signaling. Genes Dev 20(2):174–184.

10. Wei M, et al. (2008) Life span extension by calorie restriction depends on Rim15 andtranscription factors downstream of Ras/PKA, Tor, and Sch9. PLoS Genet 4(1):e13.

11. Fontana L, Partridge L, Longo VD (2010) Extending healthy life span—From yeast tohumans. Science 328(5976):321–326.

12. Bourre JM (2006) Effects of nutrients (in food) on the structure and function of thenervous system: Update on dietary requirements for brain. Part 2: Macronutrients.J Nutr Health Aging 10(5):386–399.

13. Doran PM, Bailey JE (1986) Effects of immobilization on growth, fermentationproperties, and macromolecular composition of Saccharomyces cerevisiae attachedto gelatin. Biotechnol Bioeng 28(1):73–87.

14. Pacheco AM, Gondim DR, Goncalves LR (2010) Ethanol production by fermentationusing immobilized cells of Saccharomyces cerevisiae in cashew apple bagasse. ApplBiochem Biotechnol 161(1-8):209–217.

15. Verbelen PJ, De Schutter DP, Delvaux F, Verstrepen KJ, Delvaux FR (2006) Immobilizedyeast cell systems for continuous fermentation applications. Biotechnol Lett 28(19):1515–1525.

16. Blagosklonny MV (2003) Cell senescence and hypermitogenic arrest. EMBO Rep 4(4):358–362.

17. Flodin NW (1984) The senescence of postmitotic mammalian cells: A cell-clock hy-pothesis. Mech Ageing Dev 27(1):15–27.

18. Kubitschek HE, ed (1970) Introduction to ResearchWith Continuous Cultures (PrenticeHall, Englewood Cliffs, NJ), p 195.

19. Verduyn C, Postma E, Scheffers WA, Van Dijken JP (1992) Effect of benzoic acid onmetabolic fluxes in yeasts: A continuous-culture study on the regulation of respirationand alcoholic fermentation. Yeast 8(7):501–517.

20. Alvers AL, et al. (2009) Autophagy and amino acid homeostasis are required forchronological longevity in Saccharomyces cerevisiae. Aging Cell 8(4):353–369.

21. Gomes P, Sampaio-Marques B, Ludovico P, Rodrigues F, Leão C (2007) Low auxotro-phy-complementing amino acid concentrations reduce yeast chronological life span.Mech Ageing Dev 128(5-6):383–391.

22. Fabrizio P, Longo VD (2003) The chronological life span of Saccharomyces cerevisiae.Aging Cell 2(2):73–81.

23. Haase SB, Reed SI (2002) Improved flow cytometric analysis of the budding yeast cellcycle. Cell Cycle 1(2):132–136.

24. Weinberger M, et al. (2007) DNA replication stress is a determinant of chronologicallifespan in budding yeast. PLoS ONE 2(8):e748.

25. Herker E, et al. (2004) Chronological aging leads to apoptosis in yeast. J Cell Biol164(4):501–507.

26. Galazzo JL, Bailey JE (1990) Growing Saccharomyces cerevisiae in calcium-alginatebeads induces cell alterations which accelerate glucose conversion to ethanol. Bio-technol Bioeng 36(4):417–426.

27. François J, Parrou JL (2001) Reserve carbohydrates metabolism in the yeast Saccha-romyces cerevisiae. FEMS Microbiol Rev 25(1):125–145.

28. Samokhvalov V, Ignatov V, Kondrashova M (2004) Reserve carbohydratesmaintain the viability of Saccharomyces cerevisiae cells during chronologicalaging. Mech Ageing Dev 125(3):229–235.

29. Kyryakov P, et al. (2012) Caloric restriction extends yeast chronological lifespan byaltering a pattern of age-related changes in trehalose concentration. Front Physiol3:256.

30. Ulery TL, Jang SH, Jaehning JA (1994) Glucose repression of yeast mitochondrialtranscription: Kinetics of derepression and role of nuclear genes. Mol Cell Biol 14(2):1160–1170.

31. Tusher VG, Tibshirani R, Chu G (2001) Significance analysis of microarrays applied tothe ionizing radiation response. Proc Natl Acad Sci USA 98(9):5116–5121.

32. Abramova N, Sertil O, Mehta S, Lowry CV (2001) Reciprocal regulation of anaerobicand aerobic cell wall mannoprotein gene expression in Saccharomyces cerevisiae.J Bacteriol 183(9):2881–2887.

33. Luo Z, van Vuuren HJ (2009) Functional analyses of PAU genes in Saccharomycescerevisiae. Microbiology 155(Pt 12):4036–4049.

34. Moll T, Tebb G, Surana U, Robitsch H, Nasmyth K (1991) The role of phosphorylationand the CDC28 protein kinase in cell cycle-regulated nuclear import of the S. cerevisiaetranscription factor SWI5. Cell 66(4):743–758.

35. Zhu G, et al. (2000) Two yeast forkhead genes regulate the cell cycle and pseudo-hyphal growth. Nature 406(6791):90–94.

36. Thevelein JM (1992) The RAS-adenylate cyclase pathway and cell cycle control inSaccharomyces cerevisiae. Antonie van Leeuwenhoek 62(1-2):109–130.

37. Kamei T, et al. (1998) Interaction of Bnr1p with a novel Src homology 3 domain-containing Hof1p. Implication in cytokinesis in Saccharomyces cerevisiae. J Biol Chem273(43):28341–28345.

38. Rodríguez-Peña JM, Cid VJ, Arroyo J, Nombela C (2000) A novel family of cell wall-related proteins regulated differently during the yeast life cycle. Mol Cell Biol 20(9):3245–3255.

39. Huang W, Sherman BT, Lempicki RA (2009) Systematic and integrative analysis oflarge gene lists using DAVID bioinformatics resources. Nat Protoc 4(1):44–57.

40. Sobering AK, Jung US, Lee KS, Levin DE (2002) Yeast Rpi1 is a putative transcriptionalregulator that contributes to preparation for stationary phase. Eukaryot Cell 1(1):56–65.

41. Sertil O, Cohen BD, Davies KJ, Lowry CV (1997) The DAN1 gene of S. cerevisiae isregulated in parallel with the hypoxic genes, but by a different mechanism. Gene192(2):199–205.

42. Coluccio A, et al. (2004) Morphogenetic pathway of spore wall assembly in Saccha-romyces cerevisiae. Eukaryot Cell 3(6):1464–1475.

43. De Virgilio C, DeMarini DJ, Pringle JR (1996) SPR28, a sixth member of the septin genefamily in Saccharomyces cerevisiae that is expressed specifically in sporulating cells.Microbiology 142(Pt 10):2897–2905.

44. Su SS, Mitchell AP (1993) Identification of functionally related genes that stimulateearly meiotic gene expression in yeast. Genetics 133(1):67–77.

45. Vidan S, Mitchell AP (1997) Stimulation of yeast meiotic gene expression by theglucose-repressible protein kinase Rim15p. Mol Cell Biol 17(5):2688–2697.

46. Martínez-Pastor MT, et al. (1996) The Saccharomyces cerevisiae zinc finger proteinsMsn2p and Msn4p are required for transcriptional induction through the stress re-sponse element (STRE). EMBO J 15(9):2227–2235.

47. Görner W, et al. (1998) Nuclear localization of the C2H2 zinc finger protein Msn2p isregulated by stress and protein kinase A activity. Genes Dev 12(4):586–597.

48. Swinnen E, et al. (2006) Rim15 and the crossroads of nutrient signalling pathways inSaccharomyces cerevisiae. Cell Div 1:3.

E1546 | www.pnas.org/cgi/doi/10.1073/pnas.1323918111 Nagarajan et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0

49. Cameroni E, Hulo N, Roosen J, Winderickx J, De Virgilio C (2004) The novel yeast PASkinase Rim 15 orchestrates G0-associated antioxidant defense mechanisms. Cell Cycle3(4):462–468.

50. Sato T, et al. (1999) The E-box DNA binding protein Sgc1p suppresses the gcr2 mu-tation, which is involved in transcriptional activation of glycolytic genes in Saccha-romyces cerevisiae. FEBS Lett 463(3):307–311.

51. Niu W, Li Z, Zhan W, Iyer VR, Marcotte EM (2008) Mechanisms of cell cycle controlrevealed by a systematic and quantitative overexpression screen in S. cerevisiae. PLoSGenet 4(7):e1000120.

52. Sanders SL, Herskowitz I (1996) The BUD4 protein of yeast, required for axial bud-ding, is localized to the mother/BUD neck in a cell cycle-dependent manner. J Cell Biol134(2):413–427.

53. Laabs TL, et al. (2003) ACE2 is required for daughter cell-specific G1 delay in Sac-charomyces cerevisiae. Proc Natl Acad Sci USA 100(18):10275–10280.

54. Jorgensen P, et al. (2004) A dynamic transcriptional network communicates growthpotential to ribosome synthesis and critical cell size. Genes Dev 18(20):2491–2505.

55. Johnson TE, et al. (2001) Relationship between increased longevity and stress re-sistance as assessed through gerontogene mutations in Caenorhabditis elegans. ExpGerontol 36(10):1609–1617.

56. Harris N, MacLean M, Hatzianthis K, Panaretou B, Piper PW (2001) Increasing Sac-charomyces cerevisiae stress resistance, through the overactivation of the heat shockresponse resulting from defects in the Hsp90 chaperone, does not extend replicativelife span but can be associated with slower chronological ageing of nondividing cells.Mol Genet Genomics 265(2):258–263.

57. Werner-Washburne M, Braun E, Johnston GC, Singer RA (1993) Stationary phase inthe yeast Saccharomyces cerevisiae. Microbiol Rev 57(2):383–401.

58. Lu C, Brauer MJ, Botstein D (2009) Slow growth induces heat-shock resistance innormal and respiratory-deficient yeast. Mol Biol Cell 20(3):891–903.

59. Plesset J, Ludwig JR, Cox BS, McLaughlin CS (1987) Effect of cell cycle position onthermotolerance in Saccharomyces cerevisiae. J Bacteriol 169(2):779–784.

60. Boender LG, et al. (2011) Cellular responses of Saccharomyces cerevisiae at near-zerogrowth rates: Transcriptome analysis of anaerobic retentostat cultures. FEMS YeastRes 11(8):603–620.

61. Sun ZJ, et al. (2007) Differential role of microenvironment in microencapsulation forimproved cell tolerance to stress. Appl Microbiol Biotechnol 75(6):1419–1427.

62. Norton S, Watson K, D’Amore T (1995) Ethanol tolerance of immobilized brewers’yeast cells. Appl Microbiol Biotechnol 43(1):18–24.

63. McGhee JE, St Julian G, Detroy RW (1982) Continuous and static fermentation ofglucose to ethanol by immobilized Saccharomyces cerevisiae cells of different ages.Appl Environ Microbiol 44(1):19–22.

64. Matecic M, et al. (2010) A microarray-based genetic screen for yeast chronologicalaging factors. PLoS Genet 6(4):e1000921.

65. Najafpour G, Younesi H, Syahidah Ku Ismail K (2004) Ethanol fermentation in animmobilized cell reactor using Saccharomyces cerevisiae. Bioresour Technol 92(3):251–260.

66. Newcomb LL, Diderich JA, Slattery MG, Heideman W (2003) Glucose regulation ofSaccharomyces cerevisiae cell cycle genes. Eukaryot Cell 2(1):143–149.

67. Bonawitz ND, Chatenay-Lapointe M, Pan Y, Shadel GS (2007) Reduced TOR signalingextends chronological life span via increased respiration and upregulation of mito-chondrial gene expression. Cell Metab 5(4):265–277.

68. Davidson GS, et al. (2011) The proteomics of quiescent and nonquiescent cell dif-ferentiation in yeast stationary-phase cultures. Mol Biol Cell 22(7):988–998.

69. Wei M, et al. (2009) Tor1/Sch9-regulated carbon source substitution is as effective ascalorie restriction in life span extension. PLoS Genet 5(5):e1000467.

70. Burtner CR, Murakami CJ, Kennedy BK, Kaeberlein M (2009) A molecular mechanismof chronological aging in yeast. Cell Cycle 8(8):1256–1270.

71. Norton JS, Krauss RW (1972) The inhibition of cell division in Sachharomycescerevisiae (Meyen) by carbon dioxide. Plant Cell Physiol 13(1):139–149.

72. Cot M, Loret MO, François J, Benbadis L (2007) Physiological behaviour of Saccharo-myces cerevisiae in aerated fed-batch fermentation for high level production of bi-oethanol. FEMS Yeast Res 7(1):22–32.

73. Myers B (2005) Effects of cell immobilization on ethanol yield and cell growth rateswith Saccharomyces cervisiae. Master’s thesis (Utah State Univ, Logan, UT).

74. Smukalla S, et al. (2008) FLO1 is a variable green beard gene that drives biofilm-likecooperation in budding yeast. Cell 135(4):726–737.

75. Koschwanez JH, Foster KR, Murray AW (2011) Sucrose utilization in budding yeast asa model for the origin of undifferentiated multicellularity. PLoS Biol 9(8):e1001122.

76. Pedruzzi I, et al. (2003) TOR and PKA signaling pathways converge on the proteinkinase Rim15 to control entry into G0. Mol Cell 12(6):1607–1613.

77. Reinders A, Bürckert N, Boller T, Wiemken A, De Virgilio C (1998) Saccharomycescerevisiae cAMP-dependent protein kinase controls entry into stationary phasethrough the Rim15p protein kinase. Genes Dev 12(18):2943–2955.

78. Alcasabas AA, et al. (2001) Mrc1 transduces signals of DNA replication stress to acti-vate Rad53. Nat Cell Biol 3(11):958–965.

79. Enjalbert B, Parrou JL, Vincent O, François J (2000) Mitochondrial respiratory mutantsof Saccharomyces cerevisiae accumulate glycogen and readily mobilize it in a glucose-depleted medium. Microbiology 146(Pt 10):2685–2694.

80. Lin SS, Manchester JK, Gordon JI (2001) Enhanced gluconeogenesis and increasedenergy storage as hallmarks of aging in Saccharomyces cerevisiae. J Biol Chem276(38):36000–36007.

81. Benbadis L, Cot M, Rigoulet M, Francois J (2009) Isolation of two cell populations fromyeast during high-level alcoholic fermentation that resemble quiescent and non-quiescent cells from the stationary phase on glucose. FEMS Yeast Res 9(8):1172–1186.

82. Werner-Washburne M, Roy S, Davidson GS (2012) Aging and the survival of quiescentand non-quiescent cells in yeast stationary-phase cultures. Subcell Biochem 57:123–143.

83. Váchová L, Cáp M, Palková Z (2012) Yeast colonies: A model for studies of aging,environmental adaptation, and longevity. Oxid Med Cell Longev 2012:601836.

84. Boender LG, de Hulster EA, van Maris AJ, Daran-Lapujade PA, Pronk JT (2009)Quantitative physiology of Saccharomyces cerevisiae at near-zero specific growthrates. Appl Environ Microbiol 75(17):5607–5614.

85. MacLean M, Harris N, Piper PW (2001) Chronological lifespan of stationary phaseyeast cells: A model for investigating the factors that might influence the ageing ofpostmitotic tissues in higher organisms. Yeast 18(6):499–509.

86. Deato MD, Tjian R (2007) Switching of the core transcription machinery duringmyogenesis. Genes Dev 21(17):2137–2149.

87. Boer VM, Amini S, Botstein D (2008) Influence of genotype and nutrition on survivaland metabolism of starving yeast. Proc Natl Acad Sci USA 105(19):6930–6935.

88. Pronk JT (2002) Auxotrophic yeast strains in fundamental and applied research. ApplEnviron Microbiol 68(5):2095–2100.

89. Yadav BS, Rani U, Dhamija SS, Nigam P, Singh D (1996) Process optimization forcontinuous ethanol fermentation by alginate-immobilized cells of Saccharomycescerevisiae HAU-1. J Basic Microbiol 36(3):205–210.

90. Pringle JR (1991) Staining of bud scars and other cell wall chitin with calcofluor.Methods Enzymol 194:732–735.

91. Parrou JL, François J (1997) A simplified procedure for a rapid and reliable assay ofboth glycogen and trehalose in whole yeast cells. Anal Biochem 248(1):186–188.

92. Ovalle R, et al. (1998) A spheroplast rate assay for determination of cell wall integrityin yeast. Yeast 14(13):1159–1166.

93. Schmitt ME, Brown TA, Trumpower BL (1990) A rapid and simple method for prep-aration of RNA from Saccharomyces cerevisiae. Nucleic Acids Res 18(10):3091–3092.

94. Kruckeberg AL, Nagarajan S, McInnerney K, Rosenzweig F (2009) Extraction of RNAfrom Ca-alginate-encapsulated yeast for transcriptional profiling. Anal Biochem391(2):160–162.

95. Bolstad BM, Irizarry RA, Astrand M, Speed TP (2003) A comparison of normalizationmethods for high density oligonucleotide array data based on variance and bias.Bioinformatics 19(2):185–193.

96. Ogata H, et al. (1999) KEGG: Kyoto Encyclopedia of Genes and Genomes. NucleicAcids Res 27(1):29–34.

97. Dahlquist KD, Salomonis N, Vranizan K, Lawlor SC, Conklin BR (2002) GenMAPP,a new tool for viewing and analyzing microarray data on biological pathways. NatGenet 31(1):19–20.

98. Dennis G, Jr., et al. (2003) DAVID: Database for Annotation, Visualization, and In-tegrated Discovery. Genome Biol 4(5):3.

Nagarajan et al. PNAS | Published online March 31, 2014 | E1547

MICRO

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Nov

embe

r 18

, 202

0