Plant-derived antifungal agent poacic acidtargets β-1,3-glucanJeff S. Piotrowskia,1,2, Hiroki Okadab,1, Fachuang Lua, Sheena C. Lic, Li Hinchmana, Ashish Ranjand, Damon L. Smithd,Alan J. Higbeee, Arne Ulbriche, Joshua J. Coone, Raamesh Deshpandef, Yury V. Bukhmana, Sean McIlwaina,Irene M. Onga, Chad L. Myersf, Charles Boonec,g, Robert Landicka, John Ralpha, Mehdi Kabbaged, and Yoshikazu Ohyab,2

aGreat Lakes Bioenergy Research Center, University of Wisconsin–Madison, Madison, WI 53703; Departments of dPlant Pathology and eChemistry, Universityof Wisconsin–Madison, Madison, WI 53706; bDepartment of Integrated Biosciences, Graduate School of Frontier Sciences, University of Tokyo, Kashiwa,Chiba, Japan 277-8561; cRIKEN Center for Sustainable Resource Science, Wako, Saitama, Japan 351-0198; fDepartment of Computer Scienceand Engineering, University of Minnesota–Twin Cities, Minneapolis, MN 55455; and gTerrence Donnelly Centre for Cellular and Biomolecular Research,University of Toronto, Toronto, ON, Canada M5S 3E1

Edited by Diter von Wettstein, Washington State University, Pullman, WA, and approved February 11, 2015 (received for review June 13, 2014)

A rise in resistance to current antifungals necessitates strategies toidentify alternative sources of effective fungicides. We report thediscovery of poacic acid, a potent antifungal compound found inlignocellulosic hydrolysates of grasses. Chemical genomics usingSaccharomyces cerevisiae showed that loss of cell wall synthesisand maintenance genes conferred increased sensitivity to poacicacid. Morphological analysis revealed that cells treated with poacicacid behaved similarly to cells treated with other cell wall-target-ing drugs and mutants with deletions in genes involved in pro-cesses related to cell wall biogenesis. Poacic acid causes rapid celllysis and is synergistic with caspofungin and fluconazole. The cel-lular target was identified; poacic acid localized to the cell wall andinhibited β-1,3-glucan synthesis in vivo and in vitro, apparently bydirectly binding β-1,3-glucan. Through its activity on the glucanlayer, poacic acid inhibits growth of the fungi Sclerotinia sclerotiorumandAlternaria solani as well as the oomycete Phytophthora sojae. Asingle application of poacic acid to leaves infected with the broad-range fungal pathogen S. sclerotiorum substantially reduced le-sion development. The discovery of poacic acid as a natural anti-fungal agent targeting β-1,3-glucan highlights the potential sideuse of products generated in the processing of renewable bio-mass toward biofuels as a source of valuable bioactive com-pounds and further clarifies the nature and mechanism offermentation inhibitors found in lignocellulosic hydrolysates.

chemical genomics | high-dimensional morphometrics |lignocellulosic hydrolysates | fungal cell wall | Saccharomyces cerevisiae

Lignocellulosics are a potential sugar feedstock for biofuelsand bio-based chemicals. Before plant materials can be con-

verted to biofuels by fermentation, their cell wall polysaccharidesmust be hydrolyzed to sugar monomers for microbial conversion(1). The hydrolysis processes generates, in addition to the sugars,small acids, furans, and other compounds that affect microbialgrowth and inhibit fermentation (2–5). The inhibitory effects ofthese compounds represent a challenge to efficient microbialbioconversion. The primary focus of lignocellulosic-derived in-hibitor research has been to understand, evolve, and engineertolerance in fermentative microbes (2). However, as naturalantimicrobial agents, lignocellulosic fermentation inhibitors offeran untapped reservoir of bioactive compounds.One increasingly important potential use of these inhibitors is as

antifungal agents. Worldwide, fungicide-resistant pathogens posea threat to agricultural sustainability. Pathogen resistance to con-ventional fungicides affects multiple crops (6, 7). Copper-basedfungicides are effective in organic agriculture but facing restrictionsbecause of copper accumulation in soils (8, 9). Furthermore, cli-mate change is altering the global distribution of fungal pathogens(10, 11). New sources of fungicides are a necessity to keep pacewith the evolution of resistant strains and emerging pathogens (12).The antifungal activities of many of the inhibitors (e.g., ferulic

acid and furfural) in hydrolysates have been described (13, 14),

but new compounds continue to be discovered (15). One under-studied class of compounds found in grasses and their hydro-lysates is the dehydrodiferulates and compounds derived fromthem (hereafter all simply termed diferulates) (16, 17). Ingrasses, diferulates are produced by radical dimerization of fer-ulates that acylate arabinoxylan polysaccharides and function aspowerful cell wall cross-linkers (16); derivatives of diferulates arereleased during the hydrolysis of biomass (16, 18, 19). At present,the structures of a range of diferulates have been described (16,18), but the biological activities of isolated diferulates (beyondtheir function in the plant cell wall) have not been explored.Diferulates may be expected to have effects on organisms otherthan plants. One study found a negative correlation betweendiferulate concentration and colonization by corn-boring insects(20), but a direct effect of diferulates is unknown. Despite thewell-documented antifungal activity of ferulic acid and itsderivatives (13, 21, 22), no studies on the antifungal properties ofdiferulates have been described.We screened a collection of diferulates found in lignocellulosic

hydrolysates for antifungal activity using the yeast Saccharomycescerevisiae as a discovery system for antifungal agents. We focusedon the diferulate 8–5-DC (16) derived during hydrolysis froma major diferulate in grasses; we name this compound here as

Significance

The search for new antifungal compounds is struggling to keeppace with emerging fungicide resistance. Through chemo-prospecting of an untapped reservoir of inhibitory compounds,lignocellulosic hydrolysates, we have identified a previouslyundescribed antifungal agent, poacic acid. Using both chemicalgenomics and morphological analysis together for the first time,to our knowledge, we identified the cellular target of poacic acidas β-1,3-glucan. Through its action on the glucan layer of fungalcell walls, poacic acid is a natural antifungal agent against eco-nomically significant fungi and oomycete plant pathogens. Thiswork highlights the chemical diversity within lignocellulosichydrolysates as a source of potentially valuable chemicals.

Author contributions: J.S.P., H.O., A.R., D.L.S., R.L., M.K., and Y.O. designed research; J.S.P.,H.O., F.L., L.H., A.R., A.J.H., A.U., and M.K. performed research; J.S.P., H.O., F.L., S.C.L.,D.L.S., J.J.C., C.L.M., C.B., J.R., M.K., and Y.O. contributed new reagents/analytic tools; J.S.P.,H.O., A.J.H., R.D., Y.V.B., S.M., I.M.O., and C.L.M. analyzed data; and J.S.P., H.O., C.B., R.L.,J.R., and Y.O. 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.1J.S.P. and H.O. contributed equally to this work.2To whom correspondence may be addressed. Email: jpiotrowski@wisc.edu or ohya@k.u-tokyo.ac.jp.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1410400112/-/DCSupplemental.

E1490–E1497 | PNAS | Published online March 9, 2015 www.pnas.org/cgi/doi/10.1073/pnas.1410400112

poacic acid, because it is found primarily in grasses (Poaceae).By applying both chemical genomics and morphological ana-lysis, we predicted and confirmed that poacic acid binds to cellwall β-1,3-glucan. We showed its biological activity against notonly yeast but also, the economically important fungal andoomycete pathogens Sclerotinia sclerotiorum, Alternaria solani,and Phytophthora sojae.

ResultsDiferulates with Antifungal Activity. We tested a collection of ninediferulates known to occur in hydrolysates from corn stover fortheir effects on S. cerevisiae growth (Fig. 1A and Table S1, de-tailed nomenclature). Of these, only two had detectable bio-activity at the tested concentration of 1 mg/mL. In particular,poacic acid had the greatest antifungal activity, with an IC50 of111 μg/mL (324 μM) against our control yeast (Fig. 1B). Thisinhibition is comparable with that of the widely used fungicidespicoxystrobin (IC50 of 308 μM) and polyoxin D (IC50 of 340 μM)and substantially lower than that of the primary fungicide used inorganic agriculture, copper sulfate (IC50 of 2.4 mM) (23–25).

Chemical Genomics Predict That Poacic Acid Targets the Fungal CellWall. To gain insight into the mode of action and the cellulartarget of poacic acid, we conducted chemical genomic analysis, amethod that uses genome-wide collections of viable gene-deletion mutants to identify genes with deletions that confersensitivity or resistance to bioactive compounds (26, 27). Theresulting set of sensitive and resistant gene-deletion mutantsassociated with a response yields functional insight into the modeof action (26). We challenged a pooled mixture of ∼4,000 differentyeast gene-deletion mutants with either poacic acid or a solventcontrol (DMSO). Sequencing of strain-specific DNA barcodesenabled us to decipher the relative fitness of each yeast mutant inthe presence of the drug relative to the solvent control (28).Deletion mutants for genes involved in cell wall and glyco-

sylation-related processes were present in the top significantlysensitive and resistant strains (Fig. 2A, blue circles and TableS2). Among the top 10 deletion mutants sensitive to poacic acid,we detected enrichment for genes involved in the gene ontology(GO) category fungal-type cell wall organization (P < 0.01).These mutants included deletion alleles of BCK1 (bypass of Ckinase), which encodes an MAPKKK in the Pkc1p (protein ki-nase C) cell wall integrity signaling pathway (PKC pathway);CWH43, which encodes a membrane protein involved in cell wallbiogenesis and its null mutation is synthetically lethal with PKC1mutants (29); ROM2 (Rho1 multicopy suppressor), a GDP/GTPexchange factor for Rho1p and another component of the PKCpathway; and ACK1, which seems to encode an upstream activatorof Pkc1p and has a physical interaction with Rom2p. Overall, thePKC pathway was the most sensitive pathway to poacic acid(Pathway z score = −7.85) (Fig. 2B). This profile is similar to thechemical genomic profiles of other agents that target the cell walland related processes (e.g., caspofungin) (26). Deletion mutants ofBCK1 are hypersensitive to agents that compromise glycosylation(tunicamycin) and cell wall β-1,3-glucan biosynthesis (caspo-fungin) (26). We confirmed the sensitivity of the individual bck1Δmutant and found a sixfold reduction in the IC50 against poacicacid compared with the control strain (WT) (Fig. 2C).Among the top significantly resistant strains, we detected

significant enrichment for deletions of genes involved in the GOcategory glycosphingolipid biosynthetic process (P < 0.01) drivenby csg2Δ (calcium sensitive growth) and sur1Δ (suppressor ofRvs161 and rvs167 mutations) (Fig. 2A, yellow circles and TableS2). Deletion of glycosphingolipid genes has been shown to ac-tivate the PKC pathway and cell wall biogenesis (Fig. 2B) (30).Involvement of SUR1 in poacic acid sensitivity was confirmedwhen we isolated a spontaneous chain-termination mutant inSUR1 able to form colonies on agar with 500 μg poacic acid/mL

(Fig. 2C). Additionally, some other cell wall-related gene mu-tants were resistant to poacic acid. A deletion mutant of NBP2(Nap1 binding protein) was resistant to poacic acid (Fig. 2A);Nbp2p down-regulates cell wall biogenesis through an interac-tion with Bck1p, which is activated by Pkc1p (Fig. 2B) (31). Thus,it seems that defects in the PKC pathway confer sensitivity topoacic acid, whereas activation of the PKC pathway confers re-sistance. A deletion mutant of DFG5 (defective for filamentousgrowth) also was resistant to poacic acid; DFG5 encodes a GPI-anchored protein involved in cell wall biogenesis that also hasa genetic interaction with Bck1p (32, 33).Because the chemical inhibitor of a gene product tends to

mimic the loss-of-function phenotype of a mutant that inactivatesthe gene, the chemical–genomic profile for a bioactive compoundcan resemble the genetic interaction profile for the target (34).PKC1 has a genetic interaction profile that is most significantlycorrelated to the chemical genomic profile of poacic acid (Fig. 2B)(P < 0.0001). PKC1 is an essential gene required for growthand response to cell wall stress, and it has been implicated asa key mediator of cell wall-targeting drugs, such as the echi-nocandins (35). Together, these data narrowed our targetsearch to the fungal cell wall. Poacic acid could directly damagethe cell wall, inhibit a key cell wall synthesis enzyme, or disruptthe PKC pathway.

Morphological Analysis Revealed That Poacic Acid Affects the FungalCell Wall. We investigated the morphological changes inducedby poacic acid using high-dimensional microscopy (36, 37). Re-cently, two morphological features (a wide neck and morpho-logical heterogeneity) were reported as common phenotypes incells treated with agents known to affect the cell wall (38). Themorphologies of cells exposed to poacic acid had both features(Fig. 3 A and B); they displayed dose-dependent increase in budneck size (Fig. 3C) and heterogeneous morphologies (Fig. 3D).

A

B

Fig. 1. Bioactivity of diferulates. The bioactivity of nine diferulates againstS. cerevisiae at 1 mg/mL was tested. Poacic acid (8–5-DC) had (A) the highestbioactivity and (B) an IC50 of 111 μg/mL (mean ± SE).

Piotrowski et al. PNAS | Published online March 9, 2015 | E1491

MICRO

BIOLO

GY

PNASPL

US

Because mutants displaying a high correlation with a drug phe-notype can help identify targeted processes, we next compared themorphology of poacic acid-treated cells with the individual mor-phologies of 4,718 yeast deletion mutants (37, 39, 40). Forty-threedeletion mutants had morphological profiles statistically similar(P < 0.01) to those of poacic acid (Fig. 3E and Table S3). Withinthe top correlations, we found significant enrichment of genes inthe GO category transferase activity, transferring hexosyl groups(P < 0.001). This GO category contained genes responsible forkey processes in the cell wall biogenesis, such as OCH1, whichencodes a mannosyltransferase that initiates polymannose outerchain elongation, and FKS1 (FK506 sensitivity), which encodesa catalytic subunit of β-1,3-glucan synthase. Taken together,these data further indicate that poacic acid affects the yeast cellwall, consistent with the chemical genomic analysis.

Poacic Acid Is Synergistic with Drugs That Target the Cell Wall andMembrane Integrity.Given that poacic acid may directly target thecell wall or the integrity signaling pathway, we tested whether themode of action of poacic acid differed from that of the echino-candin caspofungin. Echinocandins damage the yeast cell wall bynoncompetitive binding of the β-1,3-glucan synthase complex atthe Fks1p subunit (41, 42). Synergistic interactions occur withdrugs targeting the same or a functionally related pathway butthrough different targets (43). We found significant synergisticeffects (Fig. 4A) between poacic acid and caspofungin (P < 0.05).This interaction suggests that poacic acid targets the cell wall butdoes so through a mechanism distinct from that of caspofungin.Because echinocandins are also synergistic with antifungal azolesthat target ergosterol biosynthesis and compromise membraneintegrity (44), we tested and determined that poacic acid alsodisplayed significant synergy with fluconazole (Fig. 4B) (P < 0.01).

Poacic Acid-Induced Morphologies Are Unique Compared with Thosefrom Other Cell Wall-Targeting Agents. Compounds that inducesimilar morphological responses can be indicative of similarmodes of action. To determine how similar the morphology induced

by poacic acid is to that induced by other cell wall-affecting drugs,we compared their morphological profiles. Two echinocandins(micafungin and echinocandin B), both of which bind Fks1p, hadmorphological profiles that were highly correlated with each other,whereas poacic acid-treated cells had lower morphological corre-lations with these and other cell wall-affecting compounds (Fig. 4C).Thus, although there is some morphological similarity with othercell wall agents, the morphological response to poacic acid suggeststhat it may have a mode of action that is different from that of othercell wall-targeting agents.

Poacic Acid Causes Rapid Cell Leakage. Cell wall-targeting agents,such as echinocandins, can lead to compromised cell integrityand ultimately, cell lysis from turgor pressure (45). We inves-tigated whether poacic acid caused cell lysis in a similar way. Wetested the extent of cell permeability after 4 h of treatment withpoacic acid, caspofungin, methyl methanesulfonate (MMS), orDMSO using a propidium iodide dye that is taken up only bycells with compromised cell integrity. MMS, an agent that doesnot cause rapid cell wall damage, was included as a negativecontrol. We found that both caspofungin and poacic acid causedrapid cell leakage, whereas MMS and DMSO did not (Fig. 4D).When growth was arrested by depriving cells of a carbon

source, the effect of caspofungin was diminished, supporting theknown mode of action of echinocandins, which inhibit glucansynthesis. The effects of poacic acid were reduced in nonactivelygrowing cells, but leakage was still significantly greater (P < 0.05)than in all other treatments (Fig. 4D). The mechanism by whichpoacic acid causes leakage is lessened without active growth,showing that the compound can still cause leakage in nonactivelygrowing cells, unlike with echinocandins. This result could in-dicate a general disruption of cell wall integrity rather than anenzymatic target and thus, a different mode of action.

Poacic Acid Localizes to the Cell Surface and Targets β-1,3-Glucan.Wenext sought to determine to which cell wall component poacicacid binds by localizing the compound in treated cells. As a ferulate

A

C

B

Fig. 2. Chemical genomics of poacic acid. (A) Treatment of the yeast deletion collection with poacic acid revealed that mutants involved in cell wall bio-synthesis and glycosylation were sensitive and resistant to the compound (blue). Mutants in genes involved in glycosphingolipid biogenesis were among themost resistant (yellow). (B) The PKC pathway, which governs cell wall integrity signaling, was the most sensitive pathway (Pathway z score = −7.85), with manymembers and interacting genes showing sensitivity to poacic acid (chemical genetic interaction score in parentheses). Comparison with the yeast genetic in-teraction network indicated that the genetic interaction profile of the essential gene PKC1 was most significantly correlated to the chemical genomic profile ofpoacic acid (P < 0.001). (C) Mutants of the gene encoding the cell wall signaling kinase BCK1 were sixfold more sensitive to poacic acid, whereas a poacic acid-resistant mutant (PAr) with an SNP in SUR1 had increased resistance compared with the control strain (mean and SE bars were removed for clarity).

E1492 | www.pnas.org/cgi/doi/10.1073/pnas.1410400112 Piotrowski et al.

derivative, poacic acid is fluorescent, enabling us to visualizeits accumulation at the cell surface (Fig. 5A). Based on thisresult together with poacic acid’s chemical genomic profile,morphological profile, phenotypic similarity to fks1Δ, and abilityto cause cell leakage, we hypothesized that poacic acid targets theβ-1,3-glucan layer and thus, rapidly compromises cell integrity,leading to cell lysis when turgor pressure bursts the weakened cellwall. The absence of chitin-related genes in both the chemicalgenomic and morphological profiles and the low correlation be-tween the morphological profile of poacic acid and the chitintargeting compound nikkomycin Z led us to believe that the chitinlayer is not the cell wall target of poacic acid. Furthermore,the uniform staining pattern with poacic acid is different from thecalcofluor white staining of chitin, which preferentially binds thebud neck and bud scar.We suggest that the mode of poacic acid action is distinct from

that of the echinocandins, acting through direct binding to theglucan fibrils rather than inhibition of glucan synthase. This hy-pothesis is supported by observations that poacic acid does notlocalize specifically to the site of bud growth, like Fks1p (46), butrather, binds across the entire cell surface (Fig. 5A). Poacic acidcan inhibit β-1,3-glucan synthesis in vivo as shown by significantlydecreased glucan staining in buds (Fig. 5B and Fig. S1) andsignificantly less 14C-glucose incorporation into the β-1,3-glucanlayer after poacic acid treatment (Fig. 5C). We also observedan in vitro inhibition of β-1,3-glucan synthesis after poacic acidtreatment (Fig. 5D). By incubating purified yeast glucan withpoacic acid and observing fluorescence, we found that poacicacid directly binds β-1,3-glucan (Fig. 5E). Furthermore, althoughpoacic acid can reduce aniline blue staining of β-1,3-glucan in

buds, it does not change mannoprotein staining with fluorescentdye-conjugated Con A (Fig. S1), which suggests that poacic acidacts primarily on the formation of the glucan fibrils rather thanby inhibiting mannoprotein assembly in the cell wall.

Poacic Acid Is an Inhibitor of Fungal and Oomycete Plant Pathogens.As a plant-derived natural product, poacic acid may have a po-tential use in organic agriculture, which is presently lacking infungicide diversity beyond copper sulfate mixtures. We initiallytested the effects of poacic acid on S. sclerotiorum, an ascomycetefungal pathogen with an extremely broad host plant range (>400species) and worldwide distribution. In soybeans, S. sclerotiorumcauses Sclerotinia stem rot or white mold of soybean. The in-corporation of poacic acid into culture media caused a significant(P < 0.01) dose-dependent decrease in fungal growth both onagar plates and in liquid cultures, which was evidenced bydecreases in colony radial growth and fungal mass (Fig. 6A). Wefurther investigated whether poacic acid could inhibit lesion de-velopment in planta on detached soybean leaves. Solvent (DMSO)control or poacic acid (500 μg/mL) solutions were applied to de-tached leaves before inoculation with agar plugs containing activelygrowing mycelia of S. sclerotiorum. Lesion development was moni-tored daily up to 120 h postinoculation. Poacic acid treatmentmarkedly reduced lesion development over this time course com-pared with the control (Fig. 6 B and C). We also found poacic acidto be similarly effective against the ascomycete A. solani, whichcauses early blight in tomato and potato crops (Fig. S2). These datashow that poacic acid inhibits fungal growth in vitro and in planta,with promising agricultural applications.

A

E

B

C D

Fig. 3. Morphological characteristics of poacic acid-treated cells. Poacic acid treatment caused (A and B) abnormal cell morphology and (C and D) mor-phological characteristics similar to those caused by other cell wall-targeting agents. Poacic acid-treated cells had (C) a dose-dependent increased bud necksize and (D) heterogeneity of cell morphology. (E) The phenotype of poacic acid-treated cells was highly correlated with the phenotypes of mutants in genesinvolved in transferase activity transferring hexosyl groups (P < 0.001). The red line indicates an adjusted P value of <0.01.

Piotrowski et al. PNAS | Published online March 9, 2015 | E1493

MICRO

BIOLO

GY

PNASPL

US

Fungi generally have 30–80% glucan in their cell walls (47);similarly, oomycetes have a cell wall containing β-1,3-glucan andβ-1,6-glucan, but unlike fungi, oomycete walls contain a celluloselayer rather than chitin. Oomycetes are broadly distributed,economically significant pathogens, and a natural fungicide thatcould affect both true fungi and oomycetes by disruption of theglucan layer could be of high value. We found that poacic acidsignificantly reduces colony growth of the oomycete P. sojae (Fig.6D) (P < 0.01), a widespread pathogen that causes root and stemrot of soybeans. Given its effectiveness against both fungi andoomycetes, poacic acid may have potential as a plant-derivedfungicide with broad action.

DiscussionThrough chemoprospecting of lignocellulosic hydrolysates, wehave identified a promising antifungal agent. Combining chem-ical genomic and morphological analyses, we determined thatpoacic acid targets β-1,3-glucan within fungal cell walls. In-hibition of glucan synthesis in vivo and in vitro and cell-widelocalization and direct binding of purified glucan indicate thatthe compound can bind to β-1,3-glucans in the growing glucanfibrils as well as the mature wall. The cell wall dye Congo red mayalso bind growing glucan fibrils (48), but it also binds to chitin,and biochemical evidence indicates that the primary target ofCongo red is chitin (49). Poacic acid targets the β-1,3-glucanlayer of fungal cell walls in a manner distinct from that of othercell wall-affecting agents (e.g., caspofungin and nikkomycin Z)

and therefore, represents a previously undescribed compound tar-geting β-1,3-glucan. Although we found no effects of poacic acidon mannoprotein assembly, direct binding of glucan fibrilsoutside the plasma membrane may also result in inhibition of cellwall assemblages, such as the glucan-transglycolase Gas1p andthe chitin transglycosylases Crh1p and Utr2p, that connect chitinchains to glucans, which require glucan as a substrate or cosubstrate.In nature, diferulates strengthen monocot cell walls by cross-

linking polysaccharides (arabinoxylans) to each other and poly-saccharides to lignin—in both cases by radical coupling mecha-nisms (16, 17). It remains unknown if poacic acid has a similarinteraction with the fungal cell wall polysaccharides. However,this work raises the question about whether diferulates mayhave a dual role as antifungal secondary metabolites that conferprotection to certain plants. Other diferulates may have the samepotential to bind glucan, but the size/physical properties ofpoacic acid may contribute to its more optimal bioactivity.Against yeast, the bioactivity of poacic acid is similar to the

widely used fungicide picoxystrobin (IC50 of 308 μM), lower thanthiabendazole (IC50 of 607 μM), and considerably more toxicthan copper sulfate (IC50 of 2.4 mM) (25). Poacic acid may alsohave potential to be used combined with agricultural azolesthrough its documented synergism to slow the development ofazole resistance. Although there are conventional fungicideseffective at lower doses (e.g., captan at IC50 of 19 μM and pro-chloraz at IC50 of 132 μM) (25), most conventional agents arespecific to either the Eumycota or Oomycota, whereas poacicacid affects both. Options for organic agriculture are limited tocopper-based fungicides, which are facing increasing restrictionsbecause of copper accumulation in soil ecosystems (8, 9). Fur-thermore, as a natural plant-based phenolic acid, poacic acidwould likely be rapidly broken down in the soil and would notaccumulate (50). Field trials, application strategies (e.g., seedtreatment vs. foliar spray), and more diverse pathogen tests willbe necessary to determine its performance as an agricultural

A B

C D

Fig. 4. Synergisms and the mode of action of poacic acid. (A) Poacicacid (125 μg/mL) is significantly synergistic with caspofungin (12.5 ng/mL).(B) Poacic acid (125 μg/mL) is also synergistic with fluconazole (3.8 μg/mL).(C) Morphological similarity between poacic acid and other cell wall-affectingagents was measured based on the correlation coefficient value (R) of theirmorphological profiles. (D) Poacic acid causes cell leakage within 4 h oftreatment, similar to the cell wall-targeting compound caspofungin. Theleakage is most apparent in actively growing cells [yeast extract peptonedextrose (YPD)] compared with cells arrested without a carbon source [yeastextract peptone (YP)]. DMSO and MMS were included as control agents thatdo not directly affect cell wall integrity. In arrested cells, poacic acid hadsignificantly greater cell leakage than other treatments. One-way ANOVAand Tukey’s test were used to calculate the differences between treatments(mean ± SE). PA, poacic acid. A

D

B

E

C

Fig. 5. Poacic acid targets β-1,3-glucan. (A) Poacic acid is fluorescent andaccumulates on the cell wall. Poacic acid inhibits β-1,3-glucan in vivo asshown by (B) the decrease in signal from aniline blue staining (arrowheads)and (C) the incorporation of 14C-labeled glucose into the β-1,3-glucan layerof the cell wall (P < 0.05). Concentrations of poacic acid, echinocandin B, andhydroxyurea were 250 μg/mL, 4 μg/mL, and 30 mM, respectively. (D) Poacicacid inhibits β-1,3-glucan synthase activity in vitro with an IC50 of 31 μg/mL.(E) Poacic acid directly binds purified yeast glucan. Student’s t test was usedto determine significant differences (mean ± SD). DIC, differential inter-ference contrast.

E1494 | www.pnas.org/cgi/doi/10.1073/pnas.1410400112 Piotrowski et al.

fungicide and evaluate its persistence in the environment, but itmay also provide an important and abundant lead compoundthat could be modified for increased efficacy.Although it was identified from lignocellulosic hydrolysates

for bioethanol fermentation, poacic acid alone is not likely tobe a primary inhibitor affecting fermentation given its relativelylow concentration (0.1 μM) (Table S4). However, it could besynergistic with other inhibitory agents (e.g., furfural and phenolics)or other diferulates (e.g., 8–5-C) (2, 16). The combined effects ofdiferulates with other inhibitors could significantly affect lignocel-lulosic biofuel synthesis by fungi, but such combinatorial effectshave yet to be tested.The complementary profiling methodologies that we applied

to the analysis of poacic acid’s effects, including chemical ge-nomic profiling and morphological profiling, are powerful andcan provide high-resolution predictions of targeted processes;this work highlights the power of the combined approach. Giventhe increasing throughput of both techniques thanks to advancesin next generation sequencing and automated microscopy, theuse of both genetic and morphological approaches in large-scalescreening of drug libraries may allow unbiased whole-cell targetidentification with less reliance on target-centric high-through-put screening methods.This study was designed to identify novel bioactive compounds

from lignocellulosic hydrolysates. Given the goal of cellulosicethanol production [60 billion L/y by 2022, requiring 0.6–1.2trillion L hydrolysate/y assuming 5–10% (vol/vol) ethanol beforedistillation] (51, 52), even low-abundance compounds withinhydrolysates could be available in significant quantities. We havedetected monomeric ferulate in hydrolysates at markedly higherlevels (up to 1.7 mM in alkaline H2O2-treated corn stover). Ifsynthesized from the recovered ferulate component posthydrolysis,poacic acid may confer greater value to lignocellulosic conversion.Our results highlight the chemical diversity within lignocellulosichydrolysates as a source of potentially valuable chemicals.

MethodsDetailed methods are provided in SI Methods.

Chemical Genomic Analysis. Chemical genomic analysis of poacic acid wasperformed as described previously (26, 53). The tested yeast deletion col-lection had ∼4,000 strains using the genetic background described byAndrusiak (54). The optimal inhibitory concentration of poacic acid forchemical genomic profiling [70–80% growth vs. solvent control in yeastextract-peptone-galactose medium after 24 h of growth] was determinedusing an eight-point dose curve. A concentration of 88 μg/mL inhibitedgrowth within this range; 200-μL cultures of the pooled deletion collectionof S. cerevisiae were grown with 88 μg/mL poacic acid (n = 3) or a DMSOcontrol in triplicate for 48 h at 30 °C. Genomic DNA was extracted using theEpicentre MasterPure Yeast DNA Purification Kit. Mutant-specific molecularbarcodes were amplified with specially designed multiplex primers (28). Thebarcodes were sequenced using an Illumina MiSeq. Replicates of each con-dition, poacic acid (n = 3) or DMSO (n = 2), were sequenced. The barcodecounts for each yeast deletion mutant in the presence of poacic acid werecompared to the DMSO control conditions to determine sensitivity or re-sistance of individual strains (the chemical genetic interaction score) (26). Todetermine a P value for each top sensitive and resistant mutant, we used theEdgeR package (55, 56). A Bonferroni-corrected hypergeometric distributiontest was used to search for significant enrichment of GO terms among thetop 10 sensitive and resistant deletion mutants (57). To understand thepathways that were most affected by poacic acid, we developed a proteincomplex/pathway score based on the summation of the z scores for eachcomplex/pathway (Pathway z score). Correlation of the chemical genomicprofile of poacic acid with the yeast genetic interaction network was per-formed as previously described (34).

Multivariate Morphological Analysis. Cells of budding yeast S. cerevisiae(BY4741 his3Δ::KanMX; hereafter, his3Δ) were cultured in 2 mL 1% Bactoyeast extract (BD Biosciences), 2% Bacto peptone (BD Biosciences), and 2%glucose (YPD) with 0, 25, 50, 75, 100, or 125 μg/mL poacic acid or a DMSOcontrol at 25 °C for 16 h until the early log phase. The maximum concen-tration of the drug (125 μg/mL) was determined based on the growth in-hibition rates (10%). Cell fixation, staining, and observation were performed(n = 5) as described previously (37). Images of cell shape, actin, and nuclearDNA were analyzed using the image processing software CalMorph (version1.2), which extracted a total of 501 morphological quantitative values fromat least 200 individual cells in each experiment (37). Images were processedusing Photoshop CS2 (Adobe Systems) for illustrative purposes.

To assess the morphological similarity between the cells treated withpoacic acid and nonessential deletion mutants or cells treated with other cellwall-affecting drugs, their morphological profiles were compared as de-scribed previously (36). To identify functional gene clusters, the most sig-nificant similar mutants (43 genes, P < 0.01 after Bonferroni correction,t test) were selected as a query for GO term analysis (GO term finder in theSaccharomyces genome database).

To extract independent and characteristic features ofmorphology inducedby poacic acid, a two-step principal component analysis was performed asdescribed previously (39). To compare phenotypic noise in the yeast pop-ulation (poacic acid vs. DMSO), the noise score was calculated as describedpreviously (58).

Measurement of in Vivo β-1,3-Glucan Synthesis. Inhibition of in vivo β-1,3-glucansynthesis was measured as described previously with slight modification(n = 3) (59). Yeast cells (his3Δ) were grown in YPD to early log phase at 25 °C.The cultured cells were diluted to 1 × 107 cells per 1 mL with 1 mL ofYPD medium containing one-tenth the glucose containing 23.125 kBq[14C]glucose (ARC0122; American Radiolabeled Chemicals) and test com-pounds [250 μg/mL for poacic acid, 4 μg/mL for echinocandin B, 30 mM forhydroxyurea (negative control), or 0.4% (vol/vol) DMSO as a solvent control].The cells were radiolabeled by culturing at 25 °C for 2 h. The labeled cellswere harvested and incubatedwith 1 N NaOH at 80 °C for 30min. The insolublepellets were resuspended in 10 mM Tris·HCl, pH 7.5, containing 5 mg/mLzymolyase 100T (Seikagaku) and incubated at 37 °C for 18 h. After digestion,the zymolyase-resistant material was removed by centrifugation (15,000 × gfor 15 min), and the zymolyase-degradation product (mostly β-1,3-glucan)was purified by ultrafiltration with a centrifugal filter membrane (AmiconUltra 0.5 mL; molecular weight cutoff is 10,000; Millipore). The flow-throughfraction was mixed with scintillation mixture (Ultima Gold; PerkinElmer), andradioactivity was measured by a scintillation counter (LSC-6100; Aloka). The

A B

DC

Fig. 6. Poacic acid inhibits the growth of fungal and oomycete plantpathogens. (A) Colony growth on plates and mycelia weight of S. scle-rotiorum (strain 1980) in liquid culture were significantly inhibited by poacicacid in a dose-dependent manner. (B and C) A single aerosol treatment ofpoacic acid (500 μg/mL) before inoculation inhibited white mold lesion de-velopment on soybean leaves in planta. (D) Representative photographswere taken 96 h postinoculation. Poacic acid significantly inhibited colonygrowth of P. sojae (field isolate 7 d of growth). Dashed circles in Inset in-dicate the mycelium front after 2 d of growth. One-way ANOVA and Tukey’stest were used to calculate the difference between drug treatments amongtreatments (mean ± SE).

Piotrowski et al. PNAS | Published online March 9, 2015 | E1495

MICRO

BIOLO

GY

PNASPL

US

differences in the incorporation rates in samples were normalized by ΔOD600

measured before and after the labeling period.

Measurement of β-1,3-Glucan Synthase Activity in the Membrane Fraction.After the membrane fraction was prepared from S. cerevisiae BY4741 asdescribed previously (n = 3) (60), β-1,3-glucan synthase activity was measuredas described previously (n = 3) (61) with slight modification. Briefly, 20 μLmembrane fraction (∼70 μg total protein) was added to a reaction mixture(final volume of 100 μL) containing 50 mM Tris·HCl, pH 7.5, 10 mM potassiumfluoride, 1 mM EDTA, 0.2 mM UDP-Glc (with 89 Bq UDP-[Glucose-14C];NEC403; PerkinElmer), and different concentrations of poacic acid (1.25, 2.5,5, 10, 20, 40, 80, 160, or 320 μg/mL). The reaction mixture was incubated at25 °C for 30 min and stopped by the addition of ethanol. To trap reactionproduct (β-1,3-glucan polymer), the reaction mixture was filtered throughthe membrane filter (mixed cellulose esters; 0.2 μm in pore size; ADVANTEC),washed one time with 2 mL distilled water, and dried at room temperature.After addition of scintillation mixture (Econofluor-2; PerkinElmer), radioac-tivity was measured by a scintillation counter (LSC-6100; Aloka). Inhibitioncurves and IC50 values were determined using R software (ver. 3.0.1) bysigmoidal curve fitting with the glm function.

Growth Inhibition of Plant Pathogens. To test inhibition in liquid culture,a dose curve of 0, 125, 250, and 500 μg/mL poacic acid in 100 mL potatodextrose broth (n = 3) was used. Cultures of S. sclerotiorum strain 1980 wereinoculated with 100 μL homogenized mycelia and grown at 25 °C for 48 h.The mycelia in the liquid media were dried and weighed. The growth in-hibition of poacic acid on solid agar cultures (potato dextrose agar) wasassessed by generating replicate plates (n = 3) containing 0, 125, 250, and500 μg/mL poacic acid. Plates were inoculated with an actively growing plugof S. sclerotiorum and grown at 25 °C. The mycelial radial growth after 48 h

was measured. Inhibition of S. sclerotiorum in planta was tested by in-oculating detached soybean leaves of the commercial variety Williams 82with an agar plug of actively growing S. sclerotiorum mycelia. Leaves weretreated one time before inoculation with either an aerosol spray of waterwith DMSO (control) or a 500 μg/mL solution of poacic acid. Leaves were in-cubated in a moist environment, and lesion development was monitored up to120 h postinoculation. Field strains of P. sojae and A. solani were grown oncornmeal and potato dextrose agar plates, respectively, at room temperaturefor 7 and 5 d, respectively, before measurement. The growth inhibition ofpoacic acid was assessed at 0, 500, 1,000, and 1,500 μg/mL in replicate plates(n = 3). Agar plugs from actively growing cultures were placed at the center ofthe plates and allowed to grow at room temperature. Colony diameter wasmonitored for each treatment in a time-course experiment. One-way ANOVAand Tukey’s test were used to calculate the differences between drug treat-ments among treatments.

ACKNOWLEDGMENTS. We thank T. K. Sato, E. Hendel, S. Morford, andN. Keller for critical discussions of the manuscript. J.S.P., F.L., L.H., A.J.H.,A.U., J.J.C., S.M., I.M.O., and J.R. are funded by Department of Energy(DOE) Great Lakes Bioenergy Research Center DOE Biological and Environ-mental Research Office of Science Grant DE-FC02-07ER64494. J.S.P., F.L., andM.K. are supported by Wisconsin Alumni Research Foundation AwardMSN178899. H.O. is a research fellow of the Japan Society for the Promotionof Science. S.C.L. is supported by a RIKEN Foreign Postdoctoral Fellowship.A.R., D.L.S., and M.K. are supported by Wisconsin Soybean Marketing BoardGrant MSN172403. R.D. and C.L.M. are supported by National Institutes ofHealth Grants 1R01HG005084-01A1, 1R01GM104975-01, and R01HG005853and National Science Foundation Grant DBI 0953881. C.L.M. and C.B. aresupported by the Canadian Institute for Advanced Research Genetic NetworksProgram. M.K. is supported by United Soybean Board Grant MSN143317. Y.O.is supported by Ministry of Education, Culture, Sports, Science and Technology,Japan Grant for Scientific Research 24370002.

1. Sun Y, Cheng J (2002) Hydrolysis of lignocellulosic materials for ethanol production: A

review. Bioresour Technol 83(1):1–11.2. Piotrowski JS, et al. (2014) Death by a thousand cuts: The challenges and diverse

landscape of lignocellulosic hydrolysate inhibitors. Front Microbiol 5(2014):90.3. Palmqvist E, Hahn-Hägerdal B (2000) Fermentation of lignocellulosic hydrolysates. II:

Inhibitors and mechanisms of inhibition. Bioresour Technol 74(1):25–33.4. Skerker JM, et al. (2013) Dissecting a complex chemical stress: Chemogenomic pro-

filing of plant hydrolysates. Mol Syst Biol 9:674.5. FitzPatrick M, Champagne P, Cunningham MF, Whitney RA (2010) A biorefinery

processing perspective: Treatment of lignocellulosic materials for the production of

value-added products. Bioresour Technol 101(23):8915–8922.6. Avenot HF, Sellam A, Karaoglanidis G, Michailides TJ (2008) Characterization of mu-

tations in the iron-sulphur subunit of succinate dehydrogenase correlating with

Boscalid resistance in Alternaria alternata from California pistachio. Phytopathology

98(6):736–742.7. Leroch M, Kretschmer M, Hahn M (2011) Fungicide resistance phenotypes of Botrytis

cinerea isolates from commercial vineyards in south west Germany. J Phytopathol

159(1):63–65.8. Wightwick AM, Salzman SA, Reichman SM, Allinson G, Menzies NW (2013) Effects of

copper fungicide residues on the microbial function of vineyard soils. Environ Sci

Pollut Res Int 20(3):1574–1585.9. Mackie KA, Müller T, Zikeli S, Kandeler E (2013) Long-term copper application in an

organic vineyard modifies spatial distribution of soil micro-organisms. Soil Biol Bio-

chem 65(2013):245–253.10. Altizer S, Ostfeld RS, Johnson PTJ, Kutz S, Harvell CD (2013) Climate change and in-

fectious diseases: From evidence to a predictive framework. Science 341(6145):

514–519.11. Garrett KA, Dendy SP, Frank EE, Rouse MN, Travers SE (2006) Climate change effects

on plant disease: Genomes to ecosystems. Annu Rev Phytopathol 44(1):489–509.12. Alexander BD, Perfect JR (1997) Antifungal resistance trends towards the year 2000.

Implications for therapy and new approaches. Drugs 54(5):657–678.13. Sarma BK, Singh UP (2003) Ferulic acid may prevent infection of Cicer arietinum by

Sclerotium rolfsii. World J Microbiol Biotechnol 19(2):123–127.14. Heer D, Sauer U (2008) Identification of furfural as a key toxin in lignocellulosic hy-

drolysates and evolution of a tolerant yeast strain. Microb Biotechnol 1(6):497–506.15. Jayakody LN, Hayashi N, Kitagaki H (2011) Identification of glycolaldehyde as the key

inhibitor of bioethanol fermentation by yeast and genome-wide analysis of its toxicity.

Biotechnol Lett 33(2):285–292.16. Ralph J, Quideau S, Grabber JH, Hatfield RD (1994) Identification and synthesis of new

ferulic acid dehydrodimers present in grass cell walls. J Chem Soc 23:3485–3498.17. Ralph J, et al. (1998) Cell wall cross-linking in grasses by ferulates and diferulates.

Lignin and Lignan Biosynthesis, ACS Symposium Series, eds Lewis NG, Sarkanen S

(American Chemical Society, Washington, DC), pp 209–236.18. Vismeh R, et al. (2013) Profiling of diferulates (plant cell wall cross-linkers) using ultrahigh-

performance liquid chromatography-tandem mass spectrometry. Analyst (Lond) 138(21):

6683–6692.

19. Bunzel M, Ralph J, Marita JM, Hatfield RD, Steinhart H (2001) Diferulates as structuralcomponents in soluble and insoluble cereal dietary fibre. J Sci Food Agric 81(7):653–660.

20. Santiago R, et al. (2006) Diferulate content of maize sheaths is associated with re-sistance to theMediterranean corn borer Sesamia nonagrioides (Lepidoptera: Noctuidae).J Agric Food Chem 54(24):9140–9144.

21. Baranowski JD, Davidson PM, Nagel CW, Branen AL (1980) Inhibition of Saccharo-myces cerevisiae by naturally occurring hydroxycinnamates. J Food Sci 45(3):592–594.

22. Piotrowski JS, Morford SL, Rillig MC (2008) Inhibition of colonization by a native ar-buscular mycorrhizal fungal community via Populus trichocarpa litter, litter extract,and soluble phenolic compounds. Soil Biol Biochem 40(3):709–717.

23. Jo WJ, et al. (2008) Identification of genes involved in the toxic response of Saccha-romyces cerevisiae against iron and copper overload by parallel analysis of deletionmutants. Toxicol Sci 101(1):140–151.

24. Rogers B, et al. (2001) The pleitropic drug ABC transporters from Saccharomycescerevisiae. J Mol Microbiol Biotechnol 3(2):207–214.

25. Fai PB, Grant A (2009) A rapid resazurin bioassay for assessing the toxicity of fungi-cides. Chemosphere 74(9):1165–1170.

26. Parsons AB, et al. (2006) Exploring the mode-of-action of bioactive compounds bychemical-genetic profiling in yeast. Cell 126(3):611–625.

27. Ho CH, et al. (2011) Combining functional genomics and chemical biology to identifytargets of bioactive compounds. Curr Opin Chem Biol 15(1):66–78.

28. Smith AM, et al. (2009) Quantitative phenotyping via deep barcode sequencing.Genome Res 19(10):1836–1842.

29. Martin-Yken H, et al. (2001) Saccharomyces cerevisiae YCRO17c/CWH43 encodesa putative sensor/transporter protein upstream of the BCK2 branch of the PKC1-dependent cell wall integrity pathway. Yeast 18(9):827–840.

30. Jesch SA, Gaspar ML, Stefan CJ, Aregullin MA, Henry SA (2010) Interruption of inositolsphingolipid synthesis triggers Stt4p-dependent protein kinase C signaling. J BiolChem 285(53):41947–41960.

31. Ohkuni K, Okuda A, Kikuchi A (2003) Yeast Nap1-binding protein Nbp2p is requiredfor mitotic growth at high temperatures and for cell wall integrity. Genetics 165(2):517–529.

32. Kitagaki H, Wu H, Shimoi H, Ito K (2002) Two homologous genes, DCW1 (YKL046c)and DFG5, are essential for cell growth and encode glycosylphosphatidylinositol (GPI)-anchored membrane proteins required for cell wall biogenesis in Saccharomycescerevisiae. Mol Microbiol 46(4):1011–1022.

33. Surma MA, et al. (2013) A lipid E-MAP identifies Ubx2 as a critical regulator of lipidsaturation and lipid bilayer stress. Mol Cell 51(4):519–530.

34. Costanzo M, et al. (2010) The genetic landscape of a cell. Science 327(5964):425–431.35. Reinoso-Martín C, Schüller C, Schuetzer-Muehlbauer M, Kuchler K (2003) The yeast

protein kinase C cell integrity pathway mediates tolerance to the antifungal drugcaspofungin through activation of Slt2p mitogen-activated protein kinase signaling.Eukaryot Cell 2(6):1200–1210.

36. Ohnuki S, Oka S, Nogami S, Ohya Y (2010) High-content, image-based screening fordrug targets in yeast. PLoS ONE 5(4):e10177.

37. Ohya Y, et al. (2005) High-dimensional and large-scale phenotyping of yeast mutants.Proc Natl Acad Sci USA 102(52):19015–19020.

E1496 | www.pnas.org/cgi/doi/10.1073/pnas.1410400112 Piotrowski et al.

38. Okada H, Ohnuki S, Roncero C, Konopka JB, Ohya Y (2014) Distinct roles of cell wallbiogenesis in yeast morphogenesis as revealed by multivariate analysis of high-dimensional morphometric data. Mol Biol Cell 25(2):222–233.

39. Ohnuki S, et al. (2012) Analysis of the biological activity of a novel 24-memberedmacrolide JBIR-19 in Saccharomyces cerevisiae by the morphological imaging pro-gram CalMorph. FEMS Yeast Res 12(3):293–304.

40. Iwaki A, Ohnuki S, Suga Y, Izawa S, Ohya Y (2013) Vanillin inhibits translation andinduces messenger ribonucleoprotein (mRNP) granule formation in saccharomycescerevisiae: Application and validation of high-content, image-based profiling. PLoSONE 8(4):e61748.

41. Balashov SV, Park S, Perlin DS (2006) Assessing resistance to the echinocandin anti-fungal drug caspofungin in Candida albicans by profiling mutations in FKS1. Anti-microb Agents Chemother 50(6):2058–2063.

42. Johnson ME, Edlind TD (2012) Topological and mutational analysis of Saccharomycescerevisiae Fks1. Eukaryot Cell 11(7):952–960.

43. Cokol M, et al. (2011) Systematic exploration of synergistic drug pairs. Mol Syst Biol7(2011):544.

44. Kiraz N, et al. (2010) Synergistic activities of three triazoles with caspofungin againstCandida glabrata isolates determined by time-kill, Etest, and disk diffusion methods.Antimicrob Agents Chemother 54(5):2244–2247.

45. Cassone A, Mason RE, Kerridge D (1981) Lysis of growing yeast-form cells of Candidaalbicans by echinocandin: A cytological study. Sabouraudia 19(2):97–110.

46. Utsugi T, et al. (2002) Movement of yeast 1,3-β-glucan synthase is essential for uni-form cell wall synthesis. Genes Cells 7(1):1–9.

47. Free SJ (2013) Fungal cell wall organization and biosynthesis. Adv Genet 81:33–82.48. Kopecká M, Gabriel M (1992) The influence of congo red on the cell wall and (1----3)-

beta-D-glucan microfibril biogenesis in Saccharomyces cerevisiae. Arch Microbiol158(2):115–126.

49. Imai K, Noda Y, Adachi H, Yoda K (2005) A novel endoplasmic reticulum membraneprotein Rcr1 regulates chitin deposition in the cell wall of Saccharomyces cerevisiae.J Biol Chem 280(9):8275–8284.

50. Chen L, et al. (2011) Trichoderma harzianum SQR-T037 rapidly degrades allelo-chemicals in rhizospheres of continuously cropped cucumbers. Appl Microbiol Bio-technol 89(5):1653–1663.

51. Westbrook J, Barter GE, Manley DK, West TH (2014) A parametric analysis of futureethanol use in the light-duty transportation sector: Can the US meet its Renewable FuelStandard goals without an enforcement mechanism? Energy Policy 65(2014):419–431.

52. Lau MW, Dale BE (2009) Cellulosic ethanol production from AFEX-treated corn stoverusing Saccharomyces cerevisiae 424A(LNH-ST). Proc Natl Acad Sci USA 106(5):1368–1373.

53. Fung S-Y, et al. (2014) Unbiased screening of marine sponge extracts for anti-inflammatory agents combined with chemical genomics identifies girolline as an in-hibitor of protein synthesis. ACS Chem Biol 9(1):247–257.

54. Andrusiak K (2012) Adapting S. cerevisiae chemical genomics for identifying themodes of action of natural compounds. Masters thesis (University of Toronto, Toronto).

55. Robinson DG, Chen W, Storey JD, Gresham D (2014) Design and analysis of Bar-seqexperiments. G3 (Bethesda) 4(1):11–18.

56. Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: A Bioconductor package fordifferential expression analysis of digital gene expression data. Bioinformatics 26(1):139–140.

57. Boyle EI, et al. (2004) GO:TermFinder—open source software for accessing GeneOntology information and finding significantly enriched Gene Ontology terms asso-ciated with a list of genes. Bioinformatics 20(18):3710–3715.

58. Yvert G, et al. (2013) Single-cell phenomics reveals intra-species variation of pheno-typic noise in yeast. BMC Syst Biol 7(1):54.

59. Abe M, Qadota H, Hirata A, Ohya Y (2003) Lack of GTP-bound Rho1p in secretoryvesicles of Saccharomyces cerevisiae. J Cell Biol 162(1):85–97.

60. Abe M, et al. (2001) Yeast 1,3-β-glucan synthase activity is inhibited by phytosphingosinelocalized to the endoplasmic reticulum. J Biol Chem 276(29):26923–26930.

61. Inoue SB, et al. (1995) Characterization and gene cloning of 1,3-β-D-glucan synthasefrom Saccharomyces cerevisiae. Eur J Biochem 231(3):845–854.

Piotrowski et al. PNAS | Published online March 9, 2015 | E1497

MICRO

BIOLO

GY

PNASPL

US

Supporting InformationPiotrowski et al. 10.1073/pnas.1410400112SI MethodsCompounds, Initial Screening, and Growth.The diferulate compoundstested were synthesized as described by Lu et al. (1) and resus-pended in DMSO. Caspofungin, nikkomycin Z, and MMS werepurchased from Sigma-Aldrich. Echinocandin B was a gift fromO. Kondo (Chugai Pharmaceuticals, Tokyo, Japan). Micafunginwas provided by Astellas Pharma. Diferulates were initiallyscreened at a concentration of 1 mg/mL to determine bioactivity.Cells of Saccharomyces cerevisiae (MATα pdr1Δ::natMX pdr3Δ::KI.URA3 snq2Δ::KI.LEU2 can1Δ::STE2pr-Sp_his5 lyp1Δ his3Δ1leu2Δ0 ura3Δ0 met15Δ0), referred to as the control strain, weregrown in 200-μL cultures at 30 °C in YPD with a drug or DMSOcontrol. Plates were read on a TECAN M1000 over a 48-h growthperiod. The specific growth rate was calculated using GCAT anal-ysis software (https://gcat3-pub.glbrc.org/) (2). When presented,IC50 values for growth rate inhibition were calculated from trip-licate eight-point dose curves and SigmaPlot 12.0. When pre-sented, error bars are means ± SEs of at least three replicates.

Determining the Most Sensitive Pathway Through Chemical Genomics.A complex/pathway score based on chemical genomic data toidentify protein complexes or pathways was developed based onwhich members showed significant deviation in their chemicalgenetic interactions in the presence of a compound. For eachcomplex, the chemical genetic interaction score of the genes in thecomplex with the compound was summed. To determine signif-icance, the expectations for such a sum for random sets of genes ofequal size were calculated. The random sets of equal size wereexpected to have means equal to the background mean and SDsequal to the background SD/sqrt(n). With this information, a zscore (number of SDs from the expected mean) for each com-plex or pathway can be computed:

Pathway z  score= ðΣ=n− μÞ ðσ× sqrtðnÞÞ;=

where Σ = sum of the chemical genetic interaction scores ofgenes in the complex, μ = mean of the chemical genetic interac-tion scores of the compounds with all genes studies, σ = SD ofchemical genetic interaction scores of the compounds with allgenes in the study, and n = size of the complex.

Isolation, Sequencing, and Evaluation of Drug-Resistant Mutants.Agar containing 500 μg/mL poacic acid was inoculated with ∼1million cells of yeast (control strain). After 1 wk, two colonieswere found growing on the agar. Single-colony isolates wereobtained and found to be resistant to poacic acid. For whole-genome sequencing, single-colony isolates of poacic acid-resistant mutant, the caspofungin-resistant mutant, and the con-trol strain (WT) were grown in triplicate 200-μL cultures andpooled for genomic DNA extraction (Epicentre MasterPureYeast Kit; MPY80200). The genomic DNA was prepared forIllumina whole-genome sequencing using the Illumina TruSeqKkit(FC-121-3001) and sequenced by 150-bp paired-end reads on theMiSeq platform.To determine mutations in the drug-resistant mutants, read

quality analysis was performed using FastQC (www.bioinformatics.babraham.ac.uk/projects/fastqc/). Short reads were examined forquality and trimmed at the 3′ end when average base quality in a3-nt window fell below Q30. Short reads were mapped to thestandard S. cerevisiae reference genome, strain S288c (obtainedfrom the National Center for Biotechnology Information RefSeqrepository), using Burrows-Wheeler Alignment (BWA version

0.6.2) (3) using the default parameters, with the exception ofthe fraction of missing alignments threshold, which was set at0.08 (–n in bwa aln). SNP and indel detection were performedwith the Genome Analysis Toolkit (GATK version 1.4) (4)following their best practice variant-calling workflow (https://www.broadinstitute.org/gatk/). Duplicate reads were marked followed bybase quality recalibration using a single nucleotide polymorphismdatabase designed for S. cerevisiae. To minimize false-positive var-iant calls, stringent parameters were used: namely, the minimumbase quality required to consider a base for calling was 30, andthe minimum phred-scaled confidence threshold for genotypecalling was 50 (–mbq and –scc in the UnifiedGenotyper tool).Custom Perl scripts were used to further filter calls on the basisof read depth, mapping quality, and strand bias. This analysisrevealed an SNP in the gene SUR1 (glutamate > stop codon) inthe poacic acid-resistant mutant.

Cell Leakage Assays. A FungaLight Cell Viability Assay (L34952;Invitrogen) using a Guava Flow Cytometer (Millipore) was usedto determine if poacic acid caused membrane damage. Thepopulation of stained cells (damaged integrity) vs. nonstainedcells can be determined by flow cytometry. Caspofungin (50 ng/mL)was included as a positive control. MMS and DMSO were includedas a noncell wall-targeting control and a solvent control, respectively.To test the effects of the compounds on both active and arrestedcells, log-phase cultures were washed with 1× PBS and resuspendedto an OD0.5 in either YPDmedium or YP (no carbon source) in thepresence of the drugs (n = 3) for 4 h at 30 °C. The cells were thenstained and immediately read by flow cytometry. One-way ANOVAand Tukey’s test were used to calculate the difference between drugtreatments among cells with arrested growth.

Synergy Screening. To test for synergy, a 6 × 6-dose matrix wasinitially used to identify potentially synergistic dose combina-tions, and these points were then confirmed in triplicate. Cul-tures (200 μL) were grown with combinations of poacic acid(125 μg/mL), caspofungin (12.5 ng/mL), and fluconazole (3.8 μg/mL),and the ODs of relevant single-agent and solvent controls weremeasured after 24 h. Synergy was determined by comparing ac-tual OD in the presence of compound combinations with anexpected value calculated using the multiplicative hypothesis.This method assumes that, in the absence of an interaction, eachcompound would decrease the OD of the cell culture by thesame fraction in the presence of the other compound as it doeswhen applied alone (that is, E = A × B/C, where E is the ex-pected OD, A is OD when compound A is applied alone, B isOD when compound B is applied alone, and C is OD of thecontrol culture (DMSO). In the presence of synergy, the actualOD value is lower than the expected OD. A paired t test wasused to confirm statistical significance of this difference in threereplicates of the experiment.

Staining of Cells with Poacic Acid. Log-phase yeast cells (his3Δ)were harvested by centrifugation, washed two times with PBS,sonicated mildly, and then, incubated with 0.25% (wt/vol) poacicacid for 5 min. A small aliquot of the cells was mounted ona glass slide and observed under an Axioimager M1 Fluores-cence Microscope (Carl Zeiss) using the XF09 Filter Set (OptoScience; excitation wavelength, 340–390 nm; emission wave-length, 517.5–552.5 nm).

Mannoprotein and Glucan Staining. β-1,3-Glucan was stained withaniline blue (016-21302; Wako Chemicals) as described previously

Piotrowski et al. www.pnas.org/cgi/content/short/1410400112 1 of 6

(5) with slight modification. Briefly, log-phase yeast cells (his3Δ)were cultured in YPD with poacic acid (125 μg/mL) at 25 °C.Then, cells were collected at 0, 2, 4, and 6 h after treatment andstained with aniline blue without fixation as described previously(6). Cells mounted on a glass slide were exposed to UV for 30 sto bleach out poacic acid fluorescence before acquiring images.Staining of chitin or mannoproteins with calcofluor white (F3543;Sigma-Aldrich) or Alexa594-ConA (C11253; Life Technologies),respectively, was performed as described previously (6). Forcell-free glucan staining, yeast glucan (G0331; Tokyo Chem-ical Industry) was suspended to 0.125% (wt/vol) poacic acid

and observed under a fluorescent microscope using a regularDAPI filter set (Carl Zeiss).

Determination of Ferulate and Diferulates by Reverse-Phase HPLC–High-Resolution/Accurate MS in Hydrolysates. Ammonia fiber expan-sion treated corn stover hydrolysates samples were diluted 1:10,and 20-μL samples were analyzed by reverse-phase (C18) HPLC–high-resolution/accurate MS. Peak areas of peaks matching inretention time and accurate mass ± 10 ppm of authentic referencestandards were used to calculate concentrations by comparisonwith an external standard curve.

1. Lu F, Wei L, Azarpira A, Ralph J (2012) Rapid syntheses of dehydrodiferulates via biomimeticradical coupling reactions of ethyl ferulate. J Agric Food Chem 60(34):8272–8277.

2. Sato TK, et al. (2014) Harnessing genetic diversity in Saccharomyces cerevisiae for im-proved fermentation of xylose in hydrolysates of alkaline hydrogen peroxide pre-treated biomass. Appl Environ Microbiol 80(2):540–554.

3. Li H, Durbin R (2009) Fast and accurate short read alignment with Burrows-Wheelertransform. Bioinformatics 25(14):1754–1760.

4. DePristo MA, et al. (2011) A framework for variation discovery and genotyping usingnext-generation DNA sequencing data. Nat Genet 43(5):491–498.

5. Watanabe D, Abe M, Ohya Y (2001) Yeast Lrg1p acts as a specialized RhoGAP regu-lating 1,3-β-glucan synthesis. Yeast 18(10):943–951.

6. Okada H, Ohya Y (2015) Cold Spring Harbor Protocols (Cold Spring Harbor Lab Press,Plainview, NY).

Fig. S1. Poacic acid treatment reduces glucan staining with aniline blue but has no effect on mannoprotein straining. The control strain yeast cells (his3Δ)were grown in YPD at 25 °C until early log phase, transferred to fresh YPD medium containing poacic acid (125 μg/mL) or DMSO [0.125% (vol/vol)] as a solventcontrol, and cultured for 6 h. The cells were collected, and the cell wall components mannoproteins were stained with Alexa594-conjugated Con A followed byβ-1,3-glucan staining with aniline blue. The cells were observed under a fluorescent microscope, and over 150 budding cells were counted according to thestaining signal from three independent experiments. A Student’s t test was used to determine significant differences (mean ± SE; n = 3).

Piotrowski et al. www.pnas.org/cgi/content/short/1410400112 2 of 6

Fig. S2. Poacic acid significantly inhibits colony growth of Alternaria solani. Colony growth on plates of A. solani (field isolate) was significantly (P < 0.01)inhibited by poacic acid in a dose-dependent manner. One-way ANOVA and Tukey’s test were used to evaluate the difference between drug treatmentsamong treatments (mean ± SE; n = 3).

Table S1. Nomenclature, molecular weight, and IUPAC names of diferulate derivatives tested

Name DescriptionMolecularweight IUPAC

8–8-C 8–8-coupled cyclic diferulic acid 386 trans-7-Hydroxy-1-(4-hydroxy-3-methoxyphenyl)-6-methoxy-1,2-dihydronaphthalene-2,3-dicarboxylic acid

4–O–5 4–O–5-coupled diferulic acid 386 (E)-3-{4- [(E)-2-Carboxyvinyl]-2-methoxyphenoxy}-4-hydroxy-5-methoxycinnamic acid

8–5-C 8–5-coupled cyclic diferulic acid 386 trans-5-[(E)-2-carboxyvinyl]-2- (4-hydroxy-3-methoxyphenyl)-7-methoxy-2,3-dihydrobenzo-furan-3-carboxylic acid

8–8-O 8–8-coupled opened diferulic acid 386 4,4′-Dihydroxy-5,5′-dimethoxy-8,8′-bicinnamic acid8–8-THF 8–8-coupled tetrahydrofuran

diferulic acid404 2,5-bis-(4-Hydroxy-3-methoxyphenyl)-tetrahydrofuran-3,4-

dicarboxylic acid8–O–4 8–O–4-coupled diferulic acid 386 (Z)-8-{4-[(E)-2-Carboxyvinyl]-2-methoxyphenoxy}-4-hydroxy-3-

methoxy-cinnamic acid5–5 5–5-coupled diferulic acid 386 (E,E)-4,4′-Dihydroxy-5,5′-dimethoxy-3,3′-bicinnamic acid8–5-O 8–5-coupled opened diferulic acid 386 (E,E)-4,4′-dihydroxy-3,5′-dimethoxy-8,3′-bicinnamic acid8–5-DC

(poacic acid)8–5-coupled decarboxy diferulic acid 342 (E)-4-Hydroxy-3-{2-[(E)-4-hydroxy-3-methoxystyryl]}-5-

methoxycinnamic acid

IUPAC, International Union of Pure and Applied Chemistry.

Piotrowski et al. www.pnas.org/cgi/content/short/1410400112 3 of 6

Table S2. Top 10 sensitive and resistant deletion mutants among the poacic acid-treated deletion collection

Gene z Score P adjusted Description

Sensitive mutantsBCK1 −12.91 2.58E-28 MAPKKK acting in the PKC signaling pathway; the kinase C signaling pathway controls cell

integrity; on activation by Pkc1p phosphorylates downstream kinases Mkk1p and Mkk2pCWH43 −10.99 7.98E-7 Putative sensor/transporter protein involved in cell wall biogenesis; contains 14–16

transmembrane segments and several putative glycosylation and phosphorylation sites; nullmutation is synthetically lethal with pkc1 deletion

RGD1 −10.16 1.21E-7 GTPase-activating protein (RhoGAP) for Rho3p and Rho4p; possibly involved in control of actincytoskeleton organization

ROM2 −9.47 8.88E-8 GDP/GTP exchange factor (GEF) for Rho1p and Rho2p; mutations are synthetically lethal withmutations in rom1, which also encodes a GEF; Rom2p localization to the bud surface isdependent on Ack1p; ROM2 has a paralog, ROM1, that arose from the whole-genomeduplication

FYV8 −9.11 1.11E-14 Protein of unknown function; required for survival on exposure to K1 killer toxinACK1 −9.09 1.39E-5 Protein that functions in the cell wall integrity pathway; functions upstream of Pkc1p; GFP-fusion

protein expression is induced in response to the DNA-damaging agent MMS; nontagged Ack1pis detected in purified mitochondria

ALG6 −8.30 1.82E-11 α1,3-Glucosyltransferase; involved in transfer of oligosaccharides from dolichyl pyrophosphate toasparagine residues of proteins during N-linked protein glycosylation; mutations in humanortholog are associated with disease

EMC4 −7.92 1.90E-3 Member of conserved ER transmembrane complex; required for efficient folding of proteins inthe ER; null mutant displays induction of the unfolded protein response

SNG1 −7.91 8.96E-13 Protein involved in resistance to nitrosoguanidine and 6-azauracil; expression is regulated bytranscription factors involved in multidrug resistance; SNG1 has a paralog, YJR015W, that arosefrom the whole-genome duplication

ERG2 −7.80 4.76E-3 C-8 sterol isomerase; catalyzes the isomerization of the delta-8 double bond to the delta-7position at an intermediate step in ergosterol biosynthesis

Resistant mutantsCSG2 5.89 1.57E-3 Endoplasmic reticulum membrane protein; required for mannosylation of

inositolphosphorylceramide and growth at high calcium concentrations; protein abundanceincreases in response to DNA replication stress

LCL1 5.79 4.51E-3 Putative protein of unknown function; deletion mutant is fluconazole resistant and has longchronological lifespan

DFG5 5.51 1.34E-2 Putative mannosidase; essential GPI-anchored membrane protein required for cell wall biogenesisin bud formation, involved in filamentous growth, homologous to Dcw1p

NBP2 5.43 1.57E-3 Protein involved in the high osmolarity glycerol (HOG) pathway; negatively regulates Hog1p byrecruitment of phosphatase Ptc1p and the Pbs2p-Hog1p complex; interacts with Bck1p anddown-regulates the cell wall integrity pathway; found in the nucleus and cytoplasm, containsan SH3 domain and a Ptc1p binding domain

RTS1 5.39 6.26E-3 B-type regulatory subunit of protein phosphatase 2A (PP2A); Rts1p and Cdc55p are alternativeregulatory subunits for PP2A catalytic subunits, Pph21p and Pph22p; PP2A-Rts1p protectscohesin when recruited by Sgo1p to the pericentromere; highly enriched at centromeres in theabsence of Cdc55p; required for maintenance of septin ring organization during cytokinesis,ring disassembly in G1, and dephosphorylation of septin, Shs1p; homolog of the mammalianB subunit of PP2A

NUP170 5.33 2.79E-6 Subunit of the inner ring of the nuclear pore complex (NPC); contributes to NPC assembly andnucleocytoplasmic transport; both Nup170p and Nup157p are similar to human Nup155p;NUP170 has a paralog, NUP157, that arose from the whole-genome duplication

DSF2 5.32 6.68E-5 Deletion suppressor of mpt5 mutation; relocalizes from bud neck to cytoplasm on DNA replicationstress

SUR1 5.21 5.34E-3 Mannosylinositol phosphorylceramide synthase catalytic subunit; forms a complex withregulatory subunit Csg2p; function in sphingolipid biosynthesis is overlapping with that ofCsh1p; SUR1 has a paralog, CSH1, that arose from the whole-genome duplication

PIB2 4.96 2.79E-6 Protein of unknown function; contains FYVE domain; similar to Fab1 and Vps27RPL21B 4.81 1.11E-5 Ribosomal 60S subunit protein L21B; homologous to mammalian ribosomal protein L21, no

bacterial homolog; RPL21B has a paralog, RPL21A, that arose from the whole-genomeduplication

Piotrowski et al. www.pnas.org/cgi/content/short/1410400112 4 of 6

Table S3. Deletion mutants with significant morphological correlations with poacic acid-treated cells

Gene R valueP with Bonferroni

correction Description

COG1 0.61 1.52E-8 Essential component of the conserved oligomeric Golgi complex (Cog1p–Cog8p), a cytosolictethering complex that functions in protein trafficking to mediate fusion of transport vesiclesto Golgi compartments

NPY1 0.61 2.08E-8 NADH diphosphatase (pyrophosphatase) hydrolyzes the pyrophosphate linkage in NADH andrelated nucleotides; localizes to peroxisomes; nudix hydrolase family member

SUR4 0.61 2.25E-8 Elongase involved in fatty acid and sphingolipid biosynthesis; synthesizes very long-chain 20–26-carbon fatty acids from C18-CoA primers; involved in regulation of sphingolipid biosynthesis

OST4 0.61 3.31E-8 Subunit of the oligosaccharyltransferase complex of the endoplasmic reticulum (ER) lumen, whichcatalyzes protein asparagine-linked glycosylation; type I membrane protein required forincorporation of Ost3p or Ost6p into the OST complex

OST3 0.61 3.75E-8 γ-Subunit of the oligosaccharyltransferase complex of the ER lumen, which catalyzes asparagine-linked glycosylation of newly synthesized proteins; Ost3p is important for N-glycosylation ofa subset of proteins

YLR111W 0.60 4.20E-8 Dubious ORF unlikely to encode a protein based on available experimental and comparativesequence data

YAL058C-A 0.60 4.24E-8 Dubious ORF unlikely to encode a protein based on available experimental and comparativesequence data

SNC2 0.60 5.53E-8 Vesicle membrane receptor protein (v-SNARE); involved in the fusion between Golgi-derived secretoryvesicles with the plasma membrane; member of the synaptobrevin/VAMP family of R-type v-SNAREproteins; SNC2 has a paralog, SNC1, that arose from the whole-genome duplication

FKS1 0.59 1.01E-7 Catalytic subunit of 1,3-β-D-glucan synthase; functionally redundant with alternate catalytic subunitGsc2p; binds to regulatory subunit Rho1p; involved in cell wall synthesis and maintenance; localizes tosites of cell wall remodeling; FKS1 has a paralog, GSC2, that arose from the whole-genome duplication

BNI1 0.59 2.39E-7 Formin, nucleates the formation of linear actin filaments, involved in cell processes, such asbudding and mitotic spindle orientation, which require the formation of polarized actin cables,functionally redundant with BNR1

SWA2 0.59 2.41E-7 Auxilin-like protein involved in vesicular transport; clathrin-binding protein required for uncoatingof clathrin-coated vesicles

GAS1 0.58 5.56E-7 β-1,3-Glucanosyltransferase, required for cell wall assembly and also has a role in transcriptionalsilencing; localizes to the cell surface through a GPI anchor; also found at the nuclear periphery

PER1 0.57 1.14E-6 Protein of the endoplasmic reticulum, required for GPI-phospholipase A2 activity that remodels theGPI anchor as a prerequisite for association of GPI-anchored proteins with lipid rafts; functionallycomplemented by human ortholog PERLD1

OCH1 0.57 1.34E-6 Mannosyltransferase of the cis-Golgi apparatus, initiates the polymannose outer-chain elongationof N-linked oligosaccharides of glycoproteins

MNN11 0.55 6.42E-6 Subunit of a Golgi mannosyltransferase complex that also contains Anp1p, Mnn9p, Mnn10p, andHoc1p and mediates elongation of the polysaccharidemannan backbone; has homology to Mnn10p

CAX4 0.55 7.14E-6 Dolichyl pyrophosphate (Dol-P-P) phosphatase with a luminally oriented active site in the ERcleaves the anhydride linkage in Dol-P-P, required for Dol-P-P–linked oligosaccharideintermediate synthesis and protein N-glycosylation

MON2 0.54 1.13E-5 Peripheral membrane protein with a role in endocytosis and vacuole integrity, interacts with Arl1pand localizes to the endosome; member of the Sec7p family of proteins

KRE1 0.53 2.30E-5 Cell wall glycoprotein involved in β-glucan assembly; serves as a K1 killer toxin membrane receptorDFG5 0.52 4.40E-5 Putative mannosidase, essential GPI-anchored membrane protein required for cell wall biogenesis

in bud formation, involved in filamentous growth, homologous to Dcw1pGUP1 0.51 8.67E-5 Plasma membrane protein involved in remodeling GPI anchors; member of the MBOAT family of

putative membrane-bound O-acyltransferases; proposed to be involved in glycerol transport;GUP1 has a paralog, GUP2, that arose from the whole-genome duplication

TPM1 0.51 1.02E-4 Major isoform of tropomyosin; binds to and stabilizes actin cables and filaments, which directpolarized cell growth and the distribution of several organelles; acetylated by the NatB complexand acetylated form binds actin most efficiently; TPM1 has a paralog, TPM2, that arose from thewhole-genome duplication

YOL013W-A 0.51 1.09E-04 Putative protein of unknown function; identified by SAGERHO4 0.51 1.10E-4 Nonessential small GTPase; member of the Rho/Rac subfamily of Ras-like proteins; likely to be

involved in the establishment of cell polarity; has long N-terminal extension that plays an importantrole in Rho4p function and is shared with Rho4 homologs in other yeasts and filamentous fungi

ALG8 0.48 7.48E-4 Glucosyl transferase, involved in N-linked glycosylation; adds glucose to the dolichol-linkedoligosaccharide precursor before transfer to protein during lipid-linked oligosaccharidebiosynthesis; similar to Alg6p

VPS52 0.48 9.14E-4 Component of the Golgi-associated retrograde protein (GARP) complex, Vps51p-Vps52p-Vps53p-Vps54p, which is required for the recycling of proteins from endosomes to the late Golgi;involved in localization of actin and chitin

Piotrowski et al. www.pnas.org/cgi/content/short/1410400112 5 of 6

Table S3. Cont.

Gene R valueP with Bonferroni

correction Description

GDT1 0.47 1.26E-3 Protein of unknown function involved in calcium homeostasis; localizes to the cis- and medial-Golgiapparatus; GFP-fusion protein localizes to the vacuole; TMEM165, a human gene that causescongenital disorders of glycosylation is orthologous and functionally complements the nullallele; expression pattern and physical interactions suggest a possible role in ribosomebiogenesis; expression reduced in a gcr1 null mutant

UME1 0.47 1.42E-3 Negative regulator of meiosis; required for repression of a subset of meiotic genes duringvegetative growth, binding of histone deacetylase Rpd3p required for activity, contains an NEEbox and a WD repeat motif; homologous with Wtm1p; UME1 has a paralog, WTM2, that arosefrom the whole-genome duplication

CLC1 0.46 2.19E-3 Clathrin light chain; subunit of the major coat protein involved in intracellular protein transportand endocytosis; thought to regulate clathrin function; two Clathrin heavy chains (CHC1) formthe clathrin triskelion structural component; YGR167W

MMS2 0.46 2.87E-3 Subunit of an E3 ubiquitin ligase complex involved in replication repair; stabilizes proteincomponents of the replication fork, such as the fork-pausing complex and leading strandpolymerase, preventing fork collapse and promoting efficient recovery during replication stress;required for accurate meiotic chromosome segregation

IMP2 0.46 2.91E-3 Transcriptional activator involved in maintenance of ion homeostasis and protection against DNAdamage caused by bleomycin and other oxidants, contains a C-terminal leucine-rich repeat

PEP5 0.46 3.17E-3 Histone E3 ligase, component of CORVET tethering complex; peripheral vacuolar membraneprotein required for protein trafficking and vacuole biogenesis; interacts with Pep7p; involved inubiquitylation and degradation of excess histones

YPL184C 0.46 3.43E-3 RNA-binding protein that may be involved in translational regulation; binds specific categories ofmRNAs, including those that contain upstream ORFs and internal ribosome entry sites; interactsgenetically with chromatin remodelers and splicing factors, linking chromatin state, splicing andas a result, mRNA maturation

PEP3 0.46 3.63E-3 Component of CORVET tethering complex; vacuolar peripheral membrane protein that promotesvesicular docking/fusion reactions in conjunctionwith SNARE proteins, required for vacuolar biogenesis

CAP1 0.45 3.76E-3 α-Subunit of the capping protein heterodimer (Cap1p and Cap2p); capping protein binds to thebarbed ends of actin filaments, preventing additional polymerization; localized predominantlyto cortical actin patches; protein increases in abundance and relocalizes from bud neck to plasmamembrane on DNA replication stress

YFR016C 0.45 3.78E-3 Putative protein of unknown function; GFP-fusion protein localizes to the cytoplasm and bud;interacts with Spa2p; YFL016C is not an essential gene

PEA2 0.45 3.82E-3 Coiled-coil polarisome protein required for polarized morphogenesis, cell fusion, and low-affinityCa2+ influx; forms polarisome complex with Bni1p, Bud6p, and Spa2p; localizes to sites ofpolarized growth

BUD6 0.45 3.85E-3 Actin- and formin-interacting protein; participates in actin cable assembly and organization asa nucleation-promoting factor for formins Bni1p and Bnr1p; involved in polarized cell growth;isolated as bipolar budding mutant; potential Cdc28p substrate

VPS16 0.45 4.54E-3 Subunit of the vacuole fusion and protein-sorting HOPS complex and the CORVET tetheringcomplex; part of the class C Vps complex essential for membrane docking and fusion at Golgi-to-endosome and endosome-to-vacuole protein transport stages

POC4 0.45 5.86E-3 Component of a heterodimeric Poc4p-Irc25p chaperone involved in assembly of α-subunits into the20S proteasome; may regulate formation of proteasome isoforms with alternative subunitsunder different conditions

VPS33 0.45 6.49E-3 ATP-binding protein that is a subunit of the HOPS complex and the CORVET tethering complex;essential for protein sorting, vesicle docking, and fusion at the vacuole

OPT2 0.44 7.41E-3 Oligopeptide transporter; member of the OPT family, with potential orthologs inSchizosaccharomyces pombe and Candida albicans; also plays a role in formation of maturevacuoles

BNA1 0.44 8.62E-3 3-Hydroxyanthranilic acid dioxygenase, required for the de novo biosynthesis of NAD fromtryptophan through kynurenine; expression regulated by Hst1p

PPS1 0.44 8.99E-3 Protein phosphatase with specificity for serine, threonine, and tyrosine residues; has a role in theDNA synthesis phase of the cell cycle

Table S4. Diferulates and ferulate concentration in ammonia fiber expansion-treatedlignocellulosic hydrolysates (micromolar)

Pretreatment method 8–8-O 8–5-O 8–8-THF 5–5 8–O-4 8–5-C Poacic acid Ferulic acid

6% AFEX-treated corn stover 3.23 <0.2 0.55 0.16 0.06 8.58 0.10 76.6

AFEX, ammonia fiber expansion.

Piotrowski et al. www.pnas.org/cgi/content/short/1410400112 6 of 6