Proteomic analysis of Chlorella vulgaris: Potential targets for enhanced lipid accumulation

J O U RN A L OF P ROT EO M IC S 9 3 ( 2 01 3 ) 2 4 5 –2 53

Available online at www.sciencedirect.com

ScienceDirect www.elsevier.com/locate/jprot

Proteomic analysis of Chlorella vulgaris: Potential targets for enhanced lipid accumulation☆ Michael T. Guarnieria,⁎, Ambarish Nagb , Shihui Yanga , Philip T. Pienkosa a

National Bioenergy Center, National Renewable Energy Laboratory, Golden, CO 80401, USA Computational Science Center, National Renewable Energy Laboratory, Golden, CO 80401, USA

b

AR TIC LE I N FO

ABS TR ACT

Available online 5 June 2013

Oleaginous microalgae are capable of producing large quantities of fatty acids and triacylglycerides. As such, they are promising feedstocks for the production of biofuels

Keywords:

and bioproducts. Genetic strain-engineering strategies offer a means to accelerate the

Microalgae

commercialization of algal biofuels by improving the rate and total accumulation of

Lipid

microalgal lipids. However, the industrial potential of these organisms remains to be met,

Proteomics

largely due to the incomplete knowledgebase surrounding the mechanisms governing the

Cell cycle

induction of algal lipid biosynthesis. Such strategies require further elucidation of genes

Metabolic engineering

and gene products controlling algal lipid accumulation. In this study, we have set out to

Biofuels

examine these mechanisms and identify novel strain-engineering targets in the oleaginous microalga, Chlorella vulgaris. Comparative shotgun proteomic analyses have identified a number of novel targets, including previously unidentified transcription factors and proteins involved in cell signaling and cell cycle regulation. These results lay the foundation for strain-improvement strategies and demonstrate the power of translational proteomic analysis. Biological significance We have applied label-free, comparative shotgun proteomic analyses, via a transcriptome-toproteome pipeline, in order to examine the nitrogen deprivation response in the oleaginous microalga, C. vulgaris. Herein, we identify potential targets for strain-engineering strategies targeting enhanced lipid accumulation for algal biofuels applications. Among the identified targets are proteins involved in transcriptional regulation, lipid biosynthesis, cell signaling and cell cycle progression. This article is part of a Special Issue entitled: Translational Plant Proteomics. © 2013 Elsevier B.V. All rights reserved.

1.

Introduction

Microalgae are currently being explored as bio-production platforms for hydrocarbon and lipid-based bioproducts and biofuels. In particular, microalgal triacylglycerides (TAGs) offer a promising feedstock for biodiesel production [1]. However, despite the historical and recently renewed interest in algae-based fuels, our understanding of regulatory mech-

anisms governing algal lipid metabolism, particularly the regulation of fatty acid and TAG accumulation, remains incomplete. Identification of key regulators of genes, proteins, and metabolites triggering algal lipid biosynthesis and accumulation opens the door for genetic and metabolic engineering strategies targeting increased rates and absolute quantity of lipid accumulation. As such, molecular examination of microalgal lipid accumulation mechanisms has recently intensified.

☆ This article is part of a Special Issue entitled: Translational Plant Proteomics. ⁎ Corresponding author at: 15013 Denver West Parkway, MS 3323, Golden, CO 80401, USA. Tel.: + 1 303 724 3510. E-mail address: [email protected] (M.T. Guarnieri). 1874-3919/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jprot.2013.05.025

246

J O U RN A L OF P ROTE O M IC S 9 3 ( 2 01 3 ) 2 4 5 –25 3

The nitrogen (N)-deprivation response is perhaps the bestcharacterized inducer of lipid accumulation in microalgae [2–9]. In addition to increased total lipid content, N-deprivation can also induce changes in fatty acid chain length and saturation, with resultant TAGs more favorable for biofuel conversion [10,11]. Transcriptomic profiling has been widely applied to examine this N-deprivation response, yielding characteristic mRNA expression signatures associated with deprivation and concurrent lipid accumulation [2,5,12]. Such analyses offer powerful insights into transcriptional regulation of lipid accumulation, but these analyses do not fully define metabolic regulatory control points. This is especially true in algae and higher plants where posttranscriptional and post-translational regulation plays a critical role in protein expression and metabolic regulation [13–15,10]. Furthermore, generally employed endpoint analyses (N-replete vs. N-deplete) fail to fully elucidate the kinetics of expression, which can both illuminate underlying molecular mechanisms governing phenotypic responses and inform strain-engineering strategies necessitating induced gene expression. Time-course proteomic analyses offer a means to examine post-transcriptional responses to N-deprivation and concurrent lipid accumulation, which can both complement and further the mechanistic insights gained from transcriptional analyses. However, to date, the availability of such proteomic data is relatively limited, in particular for oleaginous microalgae, compared to that of transcriptomic analyses [4,10,16]. Our group recently implemented a transcriptome-to-proteome pipeline to examine the proteome of the oleaginous green alga, Chlorella vulgaris, specifically focusing our analysis upon changes in fatty acid (FA) and TAG biosynthetic machinery under nitrogenreplete (low lipid) and -deplete (high lipid) conditions [10]. The results identified a low abundance, yet dramatic fold-change increase in TAG biosynthetic proteins under high lipid conditions, while the FA biosynthetic components, present in higher abundance, increased to a lesser extent, suggesting these may offer promising targets for strain-engineering pursuits. Herein, we expand our analysis and examine the global proteome of C. vulgaris under N-deprivation-induced lipid accumulation, with the goal of further defining the temporal regulation of previously identified targets, and identifying novel strain-engineering targets. In addition to previously proposed regulators of fatty acid and TAG accumulation, proteins involved in cell signaling and cell cycle control and previously unidentified transcription factors are implicated as major players in the lipid accumulation response. We have also explored the temporal regulation of protein expression for these targets in order to further our understanding of mechanisms controlling microalgal lipid accumulation. Combined, these data lay the foundation for hypothesis-driven strain improvement strategies, targeting an enhanced microalgal lipid accumulation phenotype.

2.

Methods

2.1.

Algal strain and culture conditions

C. vulgaris strain UTEX 395 was grown in 1 L Roux bottles using modified Bold's Basal Media (mBBM), as described previously [10]. Cultures were inoculated at a cell density of approximately 3.5 × 106 cells/mL. Cultures were supplemented with

2% CO2/air and mixed with a magnetic stir bar at 500 rpm. To induce nitrogen deprivation, cells were harvested via centrifugation for 5 min at 5000 ×g upon entry into log phase (cell density approximately 7.8 × 107 cells), washed once in nitrogen-free mBBM, and resuspended in nitrogen-free mBBM for continued growth. Cell growth was monitored via optical density at 750 nm (OD750), and cell count measurements using a 0.1 mm depth, Hy Lite hemocytometer (Hausser Scientific). For time-course fatty acid, RNA, and protein isolation, cells were harvested from nitrogen-deplete media 24, 72, and 144 h post-deprivation, corresponding to OD750 values of approximately 4, 6, and 8, respectively, and cell densities of approximately 3.0 × 108, 5.0 × 108, and 5.3 × 108 cells per mL, respectively. For all analyses, two biological culture replicates (defined as samples generated from separate photobioreactors) were utilized, and identical samples were examined for each proteomic, RNA, and FAME analysis.

2.2.

FAME analysis

Total fatty acid content was determined via transesterification of glycerolipids followed by gas chromatography-flame ionization detector (GC-FID) analysis, as described previously [17,10]. Briefly, 50 mL samples of cell culture at the aforementioned nitrogen-replete and -deplete cell densities were harvested via centrifugation for 5 min at 5000 ×g. Cell pellets were quenched in liquid nitrogen and lyophilized overnight. Approximately 5 mg of dry biomass was suspended in chloroform–methanol (2:1, v/v), and glycerolipids were transesterified in HCl-methanol (5%, v/v) for 1 h at 85 °C in the presence of a tridecanoic acid methyl ester as an internal standard (Sigma Aldrich). Fatty acid methyl esters were extracted in hexane (Sigma Aldrich) at room temperature for 1 h and analyzed by GC-FID. For all FAME analyses, two biological and three technical replicates (defined as samples derived from the same photobioreactor) were examined.

2.3.

Protein isolation and proteomic analysis

The soluble protein fraction was isolated as described previously [10]. Briefly, cells were harvested via centrifugation for 2 min at 3000 ×g. Cell pellets were immediately quenched in liquid nitrogen, thawed and solubilized on ice in 2 mL of lysis buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM DTT, 10% glycerol, supplemented with 1× cOmplete Protease Inhibitor Cocktail solution (Roche Diagnostics Corporation, Indianapolis, IN)). The cells were then sonicated on ice at 4 °C, at 90% power setting for 6 cycles of 30 s each, with a one minute cool-down period between sonication cycles using a Braun-Sonic-L ultrasonicator. Lysates were cleared via two cycles of centrifugation at 16,000 ×g at 4 °C for 30 min, and the supernatants were isolated for use in subsequent proteomic analysis. Two biological culture replicates were utilized for protein isolation in all subsequent analyses. Gel-based liquid chromatography–mass spectrometry (GeLC/MS) was employed for comparative shotgun proteomic analysis, as described previously [10]. Briefly, 20 μg soluble proteins, as determined by Qubit fluorometry (Life Technologies, CA) following the manufacturer's instructions, were resolved using one-dimensional SDS-PAGE, followed by robotic gel excision, reduction, alkylation, and tryptic digestion,

J O U RN A L OF P ROT EO M IC S 9 3 ( 2 01 3 ) 2 4 5 –2 53

using a ProGest protein digestion station (DigiLab, Inc., MA). Peptide-containing fractions were analyzed via nanoLC/MS/MS on a Waters NanoAcquity HPLC system interfaced to a ThermoFisher LTQ Orbitrap Velos mass spectrometer. Peptides were loaded on a trapping column and eluted over a 75-μm analytical column at 350 nL/min; both columns were packed with Jupiter Proteo resin (Phenomenex, CA). Mascot was used to perform in silico six-frame translation of an annotated C. vulgaris transcriptome assembled from both N-replete and N-deplete conditions [10], and the product ion data were searched against the resultant database [18]. Databases were appended with commonly observed background proteins (cRAP) to prevent false assignment of peptides derived from those proteins. Mascot DAT output files were parsed into the Scaffold program (Proteome Software, OR). Scaffold parameters were set to a minimum of 2 peptides per protein with minimum probabilities of 95% at the protein level and 50% (Prophet scores) at the corresponding peptide level in order to ensure <1% false discovery rates (FDR). Data were normalized based upon the total number of spectral counts under nitrogen-deplete conditions 144 h postN-deprivation, followed by application of normalized spectral abundance factors (NSAF), as described by Zybailov et al. [19], and as described previously [10].

2.4.

Statistical analysis

Normalized quantitative proteomic data were subjected to one-way ANOVA statistical analysis, principal component analysis, and K-mean clustering using ArrayTrack data analysis software suite [20], as described previously [10]. Volcano plots were generated by plotting −log2(p-value) vs. log2(fold-change). Only those proteins for which p-values ≤0.05 and fold-change increases or decreases ≥ |2| were considered statistically significant and examined further for the data presented herein, unless stated otherwise. Four-way Venn diagrams and tables of unique proteins within datasets were generated using the VENNY platform [21].

3.

Results and discussion

3.1.

Proteins unique to differential nitrogen states

We investigated the proteome of C. vulgaris under nitrogen replete (NRep) conditions, followed by subsequent time-course nitrogen deplete (NDep1, 2, 3) conditions, as described above, corresponding to total fatty acid content of 11% ± 2%, 20% ± 4%, 35% ± 6%, and 60% ± 6% of dry cell weight, respectively (data not shown). GeLC/MS analysis yielded identification of 2942 proteins, with 2650 (90%) returning positive BLAST results (corresponding to positive transcriptome blastx results, as described previously [10]). 1569 proteins (~60%) were found under both NRep and NDep conditions, with 173, 30, 31, and 51 unique to each condition, respectively (Fig. 1A). One-way ANOVA statistical analysis was applied across the nitrogen replete and three, nitrogen deplete datasets in order to filter proteins based upon statistical significance. Four-way Venn diagram analysis of filtered protein datasets shows 1094 proteins had resultant p-values ≤ 0.05, with 722

247

proteins common among all four datasets (Fig. 1B). 89 proteins were unique to the NRep state, whereas 9, 7, and 18 proteins were unique to the NDep1, NDep2, and NDep3 states, respectively (Fig. 1B, Supplemental Table 1).

3.1.1. Cell cycle and cell signaling regulators implicated in nitrogen deprivation-induced lipid accumulation Examination of proteins unique to the low-lipid, nitrogen replete state yielded an abundance of proteins with gene ontology classification indicating functional involvement in cell signaling and cell cycle control (Table 1). Interestingly, halting cell cycle progression has been shown to be an effective means to induce lipid accumulation in microalgae, and as such, the mechanism of N-deprivation-induced lipid accumulation is proposed to function in part through cell cycle inhibition [22]. Among the proteins only present in the nitrogen replete state was a RIO (right open reading frame) kinase [23,24]. This kinase family has been implicated in cell cycle progression, specifically promoting exit from mitosis into S phase. In yeast, knockdown of Rio1 results in G1 or mitosis cell cycle arrest, and Rio1 knockouts are lethal, though it is unknown if this is the case in microalgae [23]. The absence of RIO from our nitrogen deplete samples suggests cells lacking RIO are viable, though it is possible that protein abundance fell below limits of detection upon nitrogen deprivation. A homolog of the cut4 (cell untimely torn) gene was also present only under N-replete conditions. Cut4 is a cell cycle checkpoint regulator associated with the anaphasepromoting complex (APC), which drives the transition from metaphase to anaphase [25]. In yeast, disruption of the cut4 gene inhibits entry into anaphase [25]. The absence of cut4 from nitrogen deplete samples suggests APC breakdown may be initiated, and as such, cell signaling associated with cell cycle halting. The APC and its role in cell cycle regulation and association with the circadian clock have been investigated in other microalgae such as Chlamydomonas reinhardtii [26] and Ostreococcus tauri [27], though its role in lipid accumulation remains to be examined. A ctr1 (constitutive triple response) homolog was also among the proteins unique to the nitrogen replete state (Fig. 1B, Table 1). Ctr1 is a Raf-family kinase with similar functionality to MAP-kinase kinase kinase (MAPKKK), and serves as a negative regulator of ethylene signaling in terrestrial plants [28]. To our knowledge, ethylene signaling has not been implicated as an associated regulator of lipid accumulation in C. vulgaris. However, due to its pleiotropic effector nature in higher plants, inducing both stress responses and senescence, the C. vulgaris ctr1 homolog may play a similar role in inhibition of stress-responsive signal transduction, removal of which may activate a subset of N-deprivation responsive genes. To date, the role of cell signaling and cell cycle regulators upon cell cycle-mediated microalgal lipid accumulation remains largely unexplored, though cell cycle inhibition is an established inducer of lipid accumulation. Elucidation of the interplay between cell cycle regulation and lipid accumulation may unveil a number of strategies to mimic and/or bypass the nitrogen-deprivation response. The identification of a number of proteins involved in cell signaling and cell cycle progression only under N-replete conditions presents an exciting finding,

248

J O U RN A L OF P ROTE O M IC S 9 3 ( 2 01 3 ) 2 4 5 –25 3

A

B

Fig. 1 – Protein distribution under varied nitrogen states. A) Venn diagram illustrating the number of unique and common proteins among nitrogen replete (N Rep), and time-course-deplete (N Dep1, NDep 2, NDep 3) conditions. B) Venn diagram showing groups of unique and common proteins with p-values ≤ 0.05 across N Rep and N Dep datasets.

such that knockdown or inducible repression of the genes coding for these proteins may present novel mechanisms to mimic N-deprivation induced cell cycle arrest and, in turn, lipid accumulation, even in actively growing N-replete cells.

3.2. Proteins differentially regulated under N-replete and N-deplete states We next examined the change in protein abundance for proteins common to both nitrogen replete and deplete states. Volcano plots were generated by plotting −log2(p-value) vs. log2(fold-change), where fold-change reflects the NSAF ratios from NRep to NDep1 (Fig. 2A), NDep1 to NDep2 (Fig. 2B), and NDep2 to NDep3 (Fig. 2C). Of those proteins meeting statistically significant criteria, the greatest changes were observed from NRep to NDep1, with 175 proteins down-regulated and 282 proteins up-regulated at statistically significant levels (Fig. 2A, Supplemental Table 2). From NDep1 to NDep2, 75 proteins were down-regulated and 95 proteins were upregulated at statistically significant levels (Fig. 2B, Supplemental Table 2). From NDep2 to NDep3, 117 proteins were

down-regulated and 82 were up-regulated at statistically significant levels (Fig. 2C, Supplemental Table 2).

3.2.1. Regulators of fatty acid and TAG biosynthesis as strainengineering targets Among the proteins found to be most differentially expressed between N-replete and N-deplete states were components of the fatty acid biosynthetic machinery. Microalgal fatty acid biosynthesis is primarily plastidial, driven by conversion of acetyl-CoA to malonyl-CoA, followed by four successive condensation reactions, penultimately leading to the synthesis of an acyl-ACP, and ultimately short- and long-chain fatty acids (summarized in [10]; extensive review in [29]). Committed entry into fatty acid biosynthesis is governed by the catalytic activity of acetyl-CoA carboxylase (ACCase). Two isoforms of ACCase are found in plants and microalgae; cytosolic homomeric and plastidial heteromeric forms (homACCase and hetACCase, respectively). AMP-activated kinase (AMPK) is a proposed accessory protein in this initial catalysis, believed to inhibit ACCase activity through phosphoregulation [10,30]. Due to the critical role ACCase plays in fatty acid biosynthesis, overexpression of genes coding for ACCase has been extensively examined in plant

Table 1 – Cell signaling and cell cycle regulators unique to nitrogen replete conditions. Protein ID

p-Value

No. of unique peptides

Total no. of peptides

BLAST expect value

Score (bits)

Checkpoint regulator cut 4 SAC9 phosphatase UGP3 (UDP-glucose pyrophosphorylase 3) Dual-specificity protein phosphatase 6 Dual-specificity protein phosphatase 7 PAS domain protein PP2A-twin subunit Serine/threonine-protein kinase ATR Dual specificity protein phosphatase, putative Ctr1 serine/threonine-protein kinase F-box family protein RIO kinase Ribosomal protein S6 kinase Glucose-inhibited division gene, gidA

0.016 0.031 0.00001 0.001 0.017 0.067 0.013 0.0084 0.0044 0.0031 0.00001 0.022 0.017 0.053

4 4 12 7 4 4 2 8 3 2 3 11 3 2

7 7 19 5 3 5 4 13 5 2 5 11 5 2

1.00E−34 2.00E−10 1.00E−120 1.00E−61 6.00E−50 2.00E−65 2.00E−93 1.00E−56 1.00E−22 1.00E−48 2.00E−10 1.00E−143 1.00E−118 2.00E−11

144 52 432 235 196 249 340 220 105 194 65 505 425 70

J O U RN A L OF P ROT EO M IC S 9 3 ( 2 01 3 ) 2 4 5 –2 53

249

A

-log2(p-value)

B

C

log2(fold-change) Fig. 2 – Changes in protein abundance under varied nitrogen states. Volcano plots of significance (− log2p-value) vs. abundance fold-change (log2fold-change) from A) N Rep to N Dep1 states, B) N Dep1 to N Dep2 states, and C) N Dep2 to N Dep3 states. Each diamond represents a single unique protein identified in our proteomic analysis, color coded as p-value ≤ 0.05 and ≥ |2|-fold change (red), p-value ≤ 0.05 and ≤ |2|-fold change (black), p-value > 0.05 and ≥ |2|-fold change (orange), and p-value > 0.05 and ≤ |2|-fold change (gray).

systems as a means to induce fatty acid biosynthesis (detailed review in [31]), and was among the first strain-engineering targets examined in microalgae [32,33]. Prior transcriptomic analysis of C. reinhardtii revealed upregulation of mRNAs coding for FA biosynthetic components under N deprivation [2]. In agreement with these results, we previously reported higher protein abundance of fatty acid and lipid biosynthetic machinery, with concurrent down-regulation of AMPK, under late-stage N-deprivation (corresponding to NDep3) in C. vulgaris [10]. However, it remained unclear at what rate, and at what stage in the lipid accumulation cycle the machinery accumulated. Furthermore, it was unclear what role, if any, homACCase played in fatty acid accumulation. In order to elucidate the temporal regulation of the fatty acid biosynthetic machinery, we examined the C. vulgaris proteome in a time-course manner, under short- and long-term nitrogen deprivation. Specifically, we examined the expression profiles, as reflected by normalized spectral abundance factors (NSAF) of both hetACCase and homACCase, respectively, as well as AMPK, under both nitrogen replete and nitrogen deplete conditions (Fig. 3A).

Relative abundance of AMPK (p = 0.047) and hetACCase (p = 0.021) were roughly equivalent under nitrogen replete conditions, and approximately 2-fold greater than homACCase (p = 0.014) (Fig. 3A). Upon nitrogen deprivation, levels of both AMPK and hetACCase increased minimally (~1.1-fold), while levels of homACCase decreased approximately 1.4-fold. Following 72 h of nitrogen deprivation, levels of homACCase fall below the limits of detection. Concurrently, AMPK and hetACCase abundance trends in opposite directions, with hetACCase accumulating to levels nearly two-fold greater than those observed under nitrogen replete conditions, and AMPK abundance decreasing over 1.6-fold 144 h post nitrogen deprivation. The hetACCase protein was found to be in higher abundance than homACCase under both nitrogen-replete and -deplete conditions, suggesting plastidial fatty acid synthesis governed by hetACCase is the preferred means of fatty acid biosynthesis in C. vulgaris in both low- and high-lipid states (Fig. 3A). An increase in hetACCase abundance was observed at our earliest N-deprivation time-point (24 h post-deprivation), demonstrating a rapid response and tight regulatory linkage between N-deprivation and fatty acid accumulation. No trace of

250

J O U RN A L OF P ROTE O M IC S 9 3 ( 2 01 3 ) 2 4 5 –25 3

1.5

1

D

0 NREP

Normalized Spectral Abundance Factor

B

NDEP1

NDEP2

NDEP3

1

0.5 0.25 0 NREP

NDEP1

NDEP2

NDEP3

NDEP2

NDEP3

0.6

E

0.4

0.3

0.2 0.25 NREP

Normalized Spectral Abundance Factor

0.75

0.8

0

C

Normalized Spectral Abundance Factor

1 0.5

NDEP1

NDEP2

NDEP3

0.12 0.1 0.08 0.06

Normalized Spectral Abundance Factor

Normalized Spectral Abundance Factor

A

0.2 0.15 0.1 0.05

0.04 0.02

0

0 NREP

NDEP1

NDEP2

NDEP3

NREP

NDEP1

Fig. 3 – Temporal profiling for proteins differentially expressed under nitrogen deprivation. Normalized spectral abundance factors × 103 (NSAF) vs. nitrogen state for: A) heteromeric acetyl-CoA carboxylase (blue), homomeric acetyl-CoA carboxylase (red), and AMP-activated kinase (green); B) Malic enzyme (ME) C) Diacylglyceride acyltransferase (DGAT); D) Upstream-element binding protein (blue), universal stress protein (USP, red), and putative transcriptional repressor (green); E) zinc-finger protein (blue), putative transcriptional regulator (red), F-box family protein (green), zinc-finger protein (purple), PTAC14 (teel), zinc-finger protein (orange), non-specific DNA-binding protein (light blue).

homACCase was detected after 24 h of N-deprivation, further implicating hetACCase as the dominant isoform controlling the observed lipid accumulation. Interestingly, abundance of AMPK is roughly equal to that of hetACCase, and the expression profile of AMPK trends with that of hetACCase through the first 24 h of N-deprivation. However, a dramatic shift is observed for AMPK thereafter, as its abundance decline nearly mirrors the increase in hetACCase. Though correlation does not imply causation, the established regulatory relationship between AMPK and ACCase, combined with these results, lead us to speculate that coupling ACCase overexpression and AMPK knockdown or knockout may indeed offer an effective strain-engineering path. We have also previously observed a dramatic increase in diacylglyceride acyltransferase (DGAT) abundance in late stage nitrogen-deprivation (NDep3). DGAT catalyzes the transfer of a third and final acyl chain to glycerol-3-phosphate (G3P) in the process of triacylglyceride biosynthesis. Overexpression of genes coding for DGAT have led to significant increases in TAG accumulation in both plant and yeast systems ([34,35] extensive review in [31]), and acyltransferases have recently been shown to be required for maximal lipid

accumulation in the model microalga, C. reinhardtii [5]. Additionally, recent transcriptomic analysis of C. reinhardtii revealed increases in mRNAs coding for putative DGATs under N-deprivation [2]. The expression profile of DGAT shows that, similar to ACCase, the abundance increase is rapid (observed at N Dep1, Fig. 3C)), indicating that TAG biosynthesis is not a late-stage N-deprivation phenomenon. As observed previously, DGAT is present in relatively low abundance (Fig. 3C), though abundance fold-change increases upon N-deprivation are more than an order of magnitude greater than those observed for fatty acid biosynthetic components (90-fold vs. 2-fold). Overexpression of the gene encoding DGAT may thus present another strong target for strain-engineering.

3.2.2. Increasing NADPH generation as a means to increase lipid biosynthesis capacity Driven by the aforementioned proteins, lipid biosynthesis is an energy intensive process, requiring large quantities of reducing equivalent in the form of reduced nicotinamide adenine dinucleotide phosphate (NADPH). Additionally, lipids

J O U RN A L OF P ROT EO M IC S 9 3 ( 2 01 3 ) 2 4 5 –2 53

may serve as a sink for excess reducing equivalents in N-starved cells that continue to convert light energy but cannot undergo cell division, as a means of photo-detoxification. Malic enzyme (ME), which generates NADPH through the reduction of NADP+ upon conversion of malate to pyruvate, has been shown to enhance both quantity and rate of lipid accumulation in oleaginous yeast and fungi [36]. As such, it has been proposed that overexpression of ME in microalgae presents a potential means to increase lipid biosynthesis by generating the reducing equivalents required for maximal lipid production [31]. We identified a ME isoform among the most differentially expressed proteins in our proteomic analysis. A 2.5-fold increase in observed upon N-deprivation (Fig. 3B), followed by relatively constant abundance thereafter. The abundance fold-increase is similar to that observed for fatty acid biosynthetic components (Fig. 3A, [10]). These results strengthen the case for ME as a target for strain-engineering, and a combinatorial approach, in which both ME and components of the fatty acid and TAG biosynthetic machinery are concurrently overexpressed, may offer a means to provide maximal reductant for these energy-intensive processes, or alternatively, may serve as an inducer of lipid biosynthesis in the photodetoxification process.

3.2.3. Transcriptional control factors as a means to induce lipid accumulation Overexpression of one or more native genes such as those discussed above, and heterologous expression of non-native genes involved in fatty acid biosynthesis has been explored as a means to induce and enhance lipid accumulation in microalgae [33]. Though these approaches have yielded promising results in plants and yeast, they have had limited success in microalgae, likely due to a number of factors, including poorly understood checkpoint and feedback inhibition, as well as post-transcriptional regulatory mechanisms such as gene silencing and post-translational modification [10,33]. Manipulation of transcription factors, governing global gene expression cascades as opposed to single genes, offers a potential means to avoid such post-transcriptional regulation, and trigger lipid biosynthesis in the absence of external stimuli (such as N-deprivation) [31]. Recently, comparative transcriptomics approaches in C. reinhardtii, have identified two transcription factors implicated in lipid accumulation; one that is requisite for lipid accumulation under N-deprivation [5] and one that is sufficient for phospholipid accumulation (though, not TAG accumulation) in absence of N-deprivation [37]. These results suggest transcription factors are indeed strong targets for strain-engineering strategies targeting enhanced lipid accumulation. Neither of the aforementioned C. reinhardtii transcription factor sequences returned a positive BLAST result when queried against the C. vulgaris proteome, possibly due to species-specific regulation. However, a diverse set of DNAbinding transcription factors and transcriptional regulators have been identified among the most differentially regulated proteins in our comparative proteomics analysis from the N Rep state to the N Dep1 state. These include members of the PHD finger, Zinc finger, and TFIIH families (Fig. 3D,E, Table 2).

251

Expression profiles for these proteins show subsets that are both up-regulated (Fig. 3D) and down-regulated (Fig. 3E) following N-deprivation. Among those upregulated with discernible homologous function is a protein with a universal stress protein (USP) domain and homology to a USP in Arabidopsis thaliana. Proteins of this family have been shown to mediate cell survival under an array of stress conditions, including nutrient deprivation [38], and thus the overabundance of USP under N-deplete conditions implies a similar function in C. vulgaris. Among those transcriptional regulators downregulated with discernible predicted homologous function is a homolog to A. thaliana pTAC14, a member of the plastid transcriptionally active complex. pTAC is a chloroplast-localized transcriptional regulating multimeric complex, involved in an array of biological functions, ranging from replication and transcription to plastidial metabolism [39]. Though the functionality of the remaining differentially expressed transcriptional regulators in our organism is currently unknown, the large differential regulation at early-stage N-deprivation (Fig. 3D,E), coupled with the successful implementation of transcription factor engineering discussed above, suggest that these offer exciting and promising targets for future exploration. Lastly, a number of SNF2 (sucrose non-fermenting) gene family members, which encode a transcription factor comprising part of the SWI/SNF chromatin remodeling complex, are among the most differentially expressed proteins from N Rep to N Dep1. Interestingly, disruption of Snf2 in yeast leads to increased lipid accumulation [40]. Snf2 has also been linked to increase stress tolerance under N-deprivation in plant systems [41]. Lastly, recent transcriptomic analyses in C. reinhardtii have implicated SNF1-related kinases in the sulfur-deprivation response and sulfolipid biosynthesis [42], demonstrating the global regulatory nature of these genes, and again suggesting a target warranting further investigation.

4.

Concluding remarks

Oleaginous microalgae are promising feedstocks for the production of biofuels and bioproducts, due to their ability to accumulate large quantities of fatty acids and storage lipids. However, in order to maximize the industrial potential of these organisms, strain-engineering approaches that bypass conventional lipid induction methods will likely be required. Herein, we have implemented comparative shotgun proteomic analyses on the oleaginous microalga, C. vulgaris, under a time-course N-deprivation regime, with the goal of further elucidating the mechanisms governing lipid induction and identifying novel targets for strain-improvement. Examination of the C. vulgaris proteome before and after nitrogen deprivation unveiled a number of potential strategies for strain-engineering, targeting enhanced lipid accumulation. Components of the FA biosynthesis machinery display rapid induction upon N-deprivation. Analysis of the nitrogen deprivation response also implicates a number of transcription factors and cell signaling and cell cycle regulators in the lipid accumulation response, all of which have promise as strain-engineering targets. Many of these factors are only present in a low-lipid state, offering targets for knockdown or knockout, while others show dramatic abundance changes

252

J O U RN A L OF P ROTE O M IC S 9 3 ( 2 01 3 ) 2 4 5 –25 3

Table 2 – Transcription factors and transcriptional regulators differentially-regulated under N-deprivation. Protein ID

p-Value

Total no. of peptides

− log2(NSAF fold-change, NRep:NDep1)

BLAST E-value

BLAST score (bits)

Putative transcriptional repressor DNA-binding protein smubp-2 Putative zinc finger protein PHD finger family protein DNA binding (zinc ion binding) Transcription factor TFIIH, 44 kDa subunit GC-rich sequence DNA-binding factor SNF2 histone linker SNF2 family chromodomain-helicase KAK isoform 2 Far upstream transcription regulator Putatitve zinc finger protein DNA binding protein, putative Zinc finger protein, putative pTAC14 SNF2 family DNA-dependent ATPase SNF2 superfamily chromatin remodeling protein Universal stress response protein Zinc finger (C2H2 type) family protein

0.0002 0.0005 0.0005 0.0008 0.0017 0.0031 0.0049 0.0062 0.0079 0.013 0.016 0.017 0.017 0.018 0.032 0.035 0.039 0.043 0.045

43 9 9 13 14 4 12 22 80 6 61 35 31 2 5 30 27 20 17

1.27 −5.89 −5.67 −7.41 −7.08 −6.78 −7.12 −7.40 −1.21 −7.13 2.04 −1.54 −7.85 −6.25 −5.98 −1.68 −1.90 1.61 −6.46

7.00E−84 5.00E−48 4.00E−17 2.00E−28 1.00E−126 5.00E−91 9.00E−39 1.00E−30 2.00E−89 1.00E−24 1.00E−32 6.00E−34 5.00E−14 1.00E−79 2.00E−11 3.00E−37 1.00E−48 1.00E−10 2.00E−21

310 192 90 127 451 333 160 134 330 114 139 144 80 294 69 156 194 62 102

upon nitrogen deprivation, offering targets for inducible overexpression. Comparative proteomic analyses have thus provided insight into a complex regulatory network controlling N-deprivation induced lipid accumulation in C. vulgaris, providing a means to identify strain-engineering targets and inform strain-engineering strategies aimed at enhanced lipid accumulation for biofuel production for industrial applications. We note that this analysis has focused upon a subset of targets putatively involved in lipid biosynthesis. Many additional strain-engineering targets may lie in pathways and critical biological processes other than those examined herein. Recent transcriptomic and proteomic analyses in C. reinhardtii have also identified a number of additional differentially regulated components of the FA and TAG biosynthesis under N deprivation [2,43], suggesting other strain-engineering targets may lie within these pathways. Statistically significant (p ≤ 0.05) proteins from other pathways and processes can be found in Supplemental Table 3 and complete proteomic data is deposited in the PRIDE repository: http://www.ebi.ac.uk/pride. Additionally, we note that a true systems biology approach, employing integrated proteomic and transcriptomic analyses, may help to identify other promising targets and strain-engineering strategies for enhanced microalgal lipid accumulation.

Acknowledgements Funding for this work was provided by the Air Force Office of Scientific Research, grants WFC31000 and WFL21000, and by the Laboratory Directed Research and Development (LDRD) Program at the National Renewable Energy Laboratory (NREL), LDRD #06510901. The authors thank Drs. David Allen and Richard Jones at MS Bioworks for contributing their expertise in MS data collection and analysis; Lieve Laurens at NREL, for method development and technical expertise with FAME

analysis; and Sharon Smolinski for technical expertise with RNA isolation.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jprot.2013.05.025.

REFERENCES [1] Pienkos PT, Darzins A. The promise and challenges of microalgal-derived biofuels. Biofuels Bioprod Bioref 2009;3: 431–40. [2] Miller R, Wu G, Deshpande RR, Vieler A, Gärtner K, Li X, et al. Changes in transcript abundance in Chlamydomonas reinhardtii following nitrogen deprivation predict diversion of metabolism. Plant Physiol 2010;154(4):1737–52. [3] Rismani-Yazdi H, Haznedaroglu BZ, Bibby K, Peccia J. Transcriptome sequencing and annotation of the microalgae Dunaliella tertiolecta: pathway description and gene discovery for production of next-generation biofuels. BMC Genomics 2011;12:148. [4] Hockin NL, Mock T, Mulholland F, Kopriva S, Malin G. The response of diatom central carbon metabolism to nitrogen starvation is different from that of green algae and higher plants. Plant Physiol 2012;158(1):299–312. [5] Boyle NR, Page MD, Liu B, Blaby IK, Casero D, Kropat J, et al. Three acyltransferases and nitrogen-responsive regulator are implicated in nitrogen starvation-induced triacylglycerol accumulation in Chlamydomonas. J Biol Chem 2012;287(19): 15811–25. [6] Recht L, Zarka A, Boussiba S. Patterns of carbohydrate and fatty acid changes under nitrogen starvation in the microalgae Haematococcus pluvialis and Nannochloropsis sp. Appl Microbiol Biotechnol 2012;94(6):1495–503. [7] Cakmak T, Angun P, Ozkan AD, Cakmak Z, Olmez TT, Tekinay T. Nitrogen and sulfur deprivation differentiate lipid accumulation

J O U RN A L OF P ROT EO M IC S 9 3 ( 2 01 3 ) 2 4 5 –2 53

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23] [24]

[25]

targets of Chlamydomonas reinhardtii. Bioengineered 2012;3(6): 343–6. Msanne J, Xu D, Konda AR, Casas-Mollano JA, Awada T, Cahoon EB, et al. Metabolic and gene expression changes triggered by nitrogen deprivation in the photoautotrophically grown microalgae Chlamydomonas reinhardtii and Coccomyxa sp C-169. Phytochemistry 2012;75:50–9. Li YJ, Fei XW, Deng XD. Novel molecular insights into nitrogen starvation-induced triacylglycerols accumulation revealed by differential gene expression analysis in green algae Micractinium pusillum. Biomass Bioenergy 2012;42: 199–211. Guarnieri MT, Nag A, Smolinski SL, Darzins A, Seibert M, Pienkos PT. Examination of triacylglycerol biosynthetic pathways via de novo transcriptomic and proteomic analyses in an unsequenced microalga. PLoS One 2011;6(10):e25851. Stephenson ALDJ, Howe CJ, Scott SA, Smith AG. Influence of nitrogen-limitation regime on the production by Chlorella vulgaris of lipids for biodiesel feedstocks. Biofuels 2010;1: 47–58. Radakovits R, Jinkerson RE, Fuerstenberg SI, Tae H, Settlage RE, Boore JL, et al. Draft genome sequence and genetic transformation of the oleaginous alga Nannochloropis gaditana. Nat Commun 2012;3:686. Kirk MM, Kirk DL. Translational regulation of protein synthesis, in response to light, at a critical stage of Volvox development. Cell 1985;41(2):419–28. Gillham NW, Boynton JE, Hauser CR. Translational regulation of gene expression in chloroplasts and mitochondria. Annu Rev Genet 1994;28:71–93. Mayfield SP, Yohn CB, Cohen A, Danon A. Regulation of chloroplast gene-expression. Annu Rev Plant Physiol 1995;46: 147–66. Le Bihan T, Martin SF, Chirnside ES, van Ooijen G, Barrios-Llerena ME, O'Neill JS, et al. Shotgun proteomic analysis of the unicellular alga Ostreococcus tauri. J Proteomics 2011;74(10):2060–70. Laurens LM, Quinn M, Van Wychen S, Templeton DW, Wolfrum EJ. Accurate and reliable quantification of total microalgal fuel potential as fatty acid methyl esters by in situ transesterification. Anal Bioanal Chem 2012;403(1):167–78. Perkins DN, Pappin DJ, Creasy DM, Cottrell JS. Probability-based protein identification by searching sequence databases using mass spectrometry data. Electrophoresis 1999;20(18):3551–67. Zybailov B, Mosley AL, Sardiu ME, Coleman MK, Florens L, Washburn MP. Statistical analysis of membrane proteome expression changes in Saccharomyces cerevisiae. J Proteome Res 2006;5(9):2339–47. Tong W, Cao X, Harris S, Sun H, Fang H, Fuscoe J, et al. ArrayTrack-supporting toxicogenomic research at the U.S. Food and Drug Administration National Center for Toxicological Research. Environ Health Perspect 2003;11(15):1819–26. Oliveros JC. VENNY. An interactive tool for comparing Venn Diagrams. http://bioinfogp.cnb.csic.es/tools/venny/index. html; 2007. Guckert JB, Cooksey KE. Triglyceride accumulation and fatty-acid profile changes in Chlorella (Chlorophyta) during high pH-induced cell-cycle inhibition. J Phycol 1990;26(1): 72–9. LaRonde-LeBlanc N, Wlodawer A. A family portrait of the RIO kinases. J Biol Chem 2005;280(45):37297–300. LaRonde-LeBlanc N, Wlodawer A. The RIO kinases: an atypical protein kinase family required for ribosome biogenesis and cell cycle progression. Biochim Biophys Acta 2005;1754(1–2):14–24. Yamashita YM, Nakaseko Y, Samejima I, Kumada K, Yamada H, Michaelson D, et al. 20S cyclosome complex formation and

[26]

[27]

[28] [29] [30]

[31]

[32]

[33]

[34]

[35]

[36]

[37] [38]

[39]

[40]

[41]

[42]

[43]

253

proteolytic activity inhibited by the cAMP/PKA pathway. Nature 1996;384(6606):276–9. Goto K, Johnson CH. Is the cell division cycle gated by a circadian clock? The case of Chlamydomonas reinhardtii. J Cell Biol 1995;129(4):1061–9. Moulager M, Monnier A, Jesson B, Bouvet R, Mosser J, Schwartz C, et al. Light-dependent regulation of cell division in Ostreococcus: evidence for a major transcriptional input. Plant Physiol 2007;144(3):1360–9. Chen YF, Etheridge N, Schaller GE. Ethylene signal transduction. Ann Bot 2005;95(6):901–15. Ohlrogge J, Browse J. Lipid biosynthesis. Plant Cell 1995;7(7): 957–70. Hardie DG, Pan DA. Regulation of fatty acid synthesis and oxidation by the AMP-activated protein kinase. Biochem Soc Trans 2002;30(Pt 6):1064–70. Courchesne NM, Parisien A, Wang B, Lan CQ. Enhancement of lipid production using biochemical, genetic and transcription factor engineering approaches. J Biotechnol 2009;141(1–2): 31–41. Dunahay TG, Jarvis EE, Dais SS, Roessler PG. Manipulation of microalgal lipid production using genetic engineering. Appl Biochem Biotechnol 1996;57–8:223–31. Sheehan J, Dunahay T, Bennemann J, Roessler P. DOE Aquatic Species Program Closeout Report. www.nrel.gov/docs/legosti/ fy98/24190.pdf; 1998. Jako C, Kumar A, Wei Y, Zou J, Barton DL, Giblin EM, et al. Seed-specific over-expression of an Arabidopsis cDNA encoding a diacylglycerol acyltransferase enhances seed oil content and seed weight. Plant Physiol 2001;126(2):861–74. Bouvier-Nave P, Benveniste P, Oelkers P, Sturley SL, Schaller H. Expression in yeast and tobacco of plant cDNAs encoding acyl CoA:diacylglycerol acyltransferase. Eur J Biochem 2000;267(1):85–96. Zhang Y, Adams IP, Ratledge C. Malic enzyme: the controlling activity for lipid production? Overexpression of malic enzyme in Mucor circinelloides leads to a 2.5-fold increase in lipid accumulation. Microbiology 2007;153(Pt 7):2013–25. Yohn C, Mendez M, Behnke C, Brand A. Stress-induced lipid trigger. Patent No. WO/2011 97261(11); 2011. Kerk D, Bulgrien J, Smith DW, Gribskov M. Arabidopsis proteins containing similarity to the universal stress protein domain of bacteria. Plant Physiol 2003;131(3):1209–19. Gao ZP, Yu QB, Zhao TT, Ma Q, Chen GX, Yang ZN. A functional component of the transcriptionally active chromosome complex, Arabidopsis pTAC14, interacts with pTAC12/HEMERA and regulates plastid gene expression. Plant Physiol 2011;157(4): 1733–45. Kamisaka Y, Tomita N, Kimura K, Kainou K, Uemura H. DGA1 (diacylglycerol acyltransferase gene) overexpression and leucine biosynthesis significantly increase lipid accumulation in the Deltasnf2 disruptant of Saccharomyces cerevisiae. Biochem J 2007;408(1):61–8. Allen SM, Aukerman M, Loussaert D, Luck S, Sakai H, Simmons CR, et al. Plants having altered agronomic characteristics under nitrogen limiting conditions and related constructs and methods involving gene encoding SNF2 domain-containing polypeptides. US Patent: 2012/0023617 A1; 2010. González-Ballester D, Casero D, Cokus S, Pellegrini M, Merchant SS, Grossman AR. RNA-seq analysis of sulfur-deprived Chlamydomonas cells reveals aspects of acclimation critical for cell survival. Plant Cell Online 2010;22(6):2058–84. Lee DY, Park JJ, Barupal DK, Fiehn O. System response of metabolic networks in Chlamydomonas reinhardtii to total available ammonium. Mol Cell Proteomics 2012;11(10): 973–88.