RNAi Knock-Down of LHCBM1, 2 and 3 Increases Photosynthetic H2 Production Efficiency of the Green Alga Chlamydomonas reinhardtii

Single cell green algae (microalgae) are rapidly emerging as a platform for the production of sustainable fuels. Solar-driven H2 production from H2O theoretically provides the highest-efficiency route to fuel production in microalgae. This is because the H2-producing hydrogenase (HYDA) is directly coupled to the photosynthetic electron transport chain, thereby eliminating downstream energetic losses associated with the synthesis of carbohydrate and oils (feedstocks for methane, ethanol and oil-based fuels). Here we report the simultaneous knock-down of three light-harvesting complex proteins (LHCMB1, 2 and 3) in the high H2-producing Chlamydomonas reinhardtii mutant Stm6Glc4 using an RNAi triple knock-down strategy. The resultant Stm6Glc4L01 mutant exhibited a light green phenotype, reduced expression of LHCBM1 (20.6% ±0.27%), LHCBM2 (81.2% ±0.037%) and LHCBM3 (41.4% ±0.05%) compared to 100% control levels, and improved light to H2 (180%) and biomass (165%) conversion efficiencies. The improved H2 production efficiency was achieved at increased solar flux densities (450 instead of ∼100 µE m−2 s−1) and high cell densities which are best suited for microalgae production as light is ideally the limiting factor. Our data suggests that the overall improved photon-to-H2 conversion efficiency is due to: 1) reduced loss of absorbed energy by non-photochemical quenching (fluorescence and heat losses) near the photobioreactor surface; 2) improved light distribution in the reactor; 3) reduced photoinhibition; 4) early onset of HYDA expression and 5) reduction of O2-induced inhibition of HYDA. The Stm6Glc4L01 phenotype therefore provides important insights for the development of high-efficiency photobiological H2 production systems.


Introduction
The development of clean fuels for the future is one of the most urgent challenges facing our society for three reasons; to reduce CO 2 emissions, increase energy security and enable sustainable economic development. The importance of fuels is emphasized by the fact that, they directly supply ,80% of global energy demand while electricity provides only ,17% [1]. In terms of fuel security it is notable that in 2010 global energy use rose to ,0.5 ZJ (1 ZJ = 10 21 J) [2] and that the total global fuel reserves (oil, coal, gas and uranium) were reportedly at only ,82 ZJ [3].
The capability to produce sufficient renewable fuel to supply global demand is dependent on the availability of a renewable energy source that is large enough to drive this process. Solar energy is by far the largest and most equitably distributed renewable energy source available, supplying ,3020 ZJ yr 21 to the Earth's surface [4]. Photosynthetically active radiation (PARconventionally 400# l #700 nm [5]) makes up 43% of the irradiance delivered by the standard AM1.5 reference solar spectrum [6]. Therefore ,1300 ZJ yr 21 is available globally to drive photosynthesis. Photosynthesis is a solar-to-chemical-energy conversion process that can be used to produce biomass and biofuels. Economic feasibility of photosynthetic fuel production systems is not limited by global solar irradiance, but rather by the areal productivities currently achievable cost-competitively. These depend strongly on the energy conversion efficiency of photosynthesis, which can be improved through engineering.
Microalgae systems are rapidly emerging as one of the most promising platforms for the production of renewable fuels and have a number of advantages compared to higher plants. In particular, they can theoretically address the key areas of concern related to the food vs. fuel conflicts of 1 st generation biofuel systems (e.g. corn ethanol) [7,8] and assist with the development of a more sustainable food and fuel future. This is because well-designed microalgae systems can be located on non-arable land, cultivated at least in part using saline and waste water, enable increased nutrient recycling and achieve higher yields than crop plants, due to the ability to optimize light distribution, CO 2 supply and production conditions [9]. Importantly, algae, besides being able to produce biomass and oil which can be converted into bio-fuels such as bio-diesel, methane and ethanol, also have the ability to produce hydrogen from water [10]. Solar powered H 2 production is theoretically the most efficient method of biofuel production as it is closely coupled to the photosynthetic electron transport chain and additionally offers the potential to produce a CO 2 -negative bio-fuel, reducing CO 2 emission and assisting in carbon sequestration [11].
The first step of photosynthesis and all biofuel production is light capture. As a result, its optimization is essential for the development of all high-efficiency microalgal processes. In algal cells, light capture is performed by the chlorophyll-binding proteins of the Light Harvesting Complexes I (LHCI) and II (LHCII), which transfer the derived excitation energy to Photosystems I (PSI) and II (PSII), respectively. A secondary but very important role of the LHCII (LHCBM) proteins is that under high light conditions, they are involved in the photoprotective processes of non-photochemical quenching (NPQ) [12][13][14][15].
Given the need for microalgae to balance light capture and protection against photodamage, wild type algae (wt) have generally evolved large LHC antenna systems to capture incident light efficiently under low light conditions, and the ability to dissipate the excess (,80%) under high light conditions to avoid photodamage [16]. Wt cells appear to have a competitive advantage, as they capture the bulk of the incident light at an illuminated surface and dissipate the excess, with the result that cells deeper in culture suffer from light limitation. Even though microalgae down-regulate LHCII levels naturally to a degree, under high light conditions, at 25% of full sunlight (full sunlight typically being 2000 mE m 22 s 21 midday at equatorial region) the photosynthetic capacity of most algal species is usually saturated [17] and energy dissipation is required. This can be broadly likened to a 'selfish organism strategy'; i.e. each cell uses the light it needs and wastes the rest as heat, which cannot be used to drive photosynthesis in other cells.
The overall efficiency of the culture can theoretically be improved, by artificially reducing the light-harvesting antenna size. This strategy is designed to enable each cell to capture the light that it requires but allow the excess light that would otherwise be wasted, to penetrate further into the culture. This illuminates the cells that would otherwise be shaded. This strategy can be therefore broadly be likened to 'social engineering' in that the light is more evenly distributed between all cells in the culture. Thus, while the maximum light harvesting efficiency of individual cells should theoretically remain the same as long as the incident light exceeds saturating levels at the surface, the overall PBR efficiency can be increased. Due to the decreased energy wastage through NPQ, the concentration of algae cells can be increased, thereby improving the yield per unit volume of culture and reducing PBR costs. This offers a theoretical route to increase the efficiency and cost competitiveness of photo-biological fuel production.
In the green microalga C. reinhardtii, at least 25 LHC genes have been identified [18]. These exhibit a high degree of gene and protein sequence similarity across plant and algal families [14]. The LHCII proteins can be divided into the major and minor LHCII proteins. The major LHCII proteins are reportedly transcribed from 9 different genes LHCBM 1-9 which are numbered according to the relative expression levels initially observed [19,20]. These LHCBM proteins are reported to trimerize and their primary role is to capture solar energy and funnel it via the minor LHCII proteins, CP29 (LHCB4) and CP26 (LHCB5), to the PSII core (CP47, CP43, D1, D2, cytb559, PSBO and additional small and extrinsic subunits). In PSI solar energy is captured by nine LHCA proteins (LHCA1-9) and the derived excitation energy channeled to the PSI reaction center (psaA and psaB) [14,[21][22][23]. In higher plants the LHCI subunits reportedly form a crescent shaped belt on one side of the PSI monomer. This is termed the ''LHCI-belt'', and reportedly consists of two dimeric subcomplexes. These subcomplexes are referred as LHCI-730 and LHCI-680 in higher plants and LHCI-705 and LHCI-680 in Chlamydomonas, where the numbers refer to their characteristic fluorescence peaks [24,25].
Melis and coworkers and later others [16,[26][27][28] tested the hypothesis that artificial antenna reduction could improve light-tobiomass conversion efficiency and showed improvement under lab conditions. However, none of these strains were optimized for improved levels of H 2 production in liquid culture.
Hydrogen production is typically induced by sulfur deprivation [10,29]. Under these conditions the repair of the photo-inhibited D1 protein of PSII which contains methionine (and so S) is inhibited, reducing O 2 production. This in turn lifts the inhibition of HYDA expression, inducing H 2 production. Down regulation of PSII however also reduces the flow of H + and efrom H 2 O to H 2 .
Here we report the specific triple knock-down of the three most abundant LHCII proteins (LHCBM1, LHCBM2 and LHCBM3) in the high H 2 producing background of Stm6Glc4 [30], with the aim of further increasing the efficiency of photobiological H 2 production. Stm6Glc4 is based on the Stm6 mutant, which under H 2 -producing conditions was shown to have upregulated Alternative Oxidase (AOX) activity [31] and to be locked into linear electron transport [32], resulting in increased O 2 consumption and linear electron pathway from water to HYDA. The results show that reducing the LHC antenna system allows the rate of O 2 production by photosynthesis to be brought into balance with the rate of O 2 consumption by respiration. This in turn facilitates the induction of HYDA and provides a mechanism for continuous photosynthetic H 2 production from water.

Strains and Culture Conditions
The high H 2 producing C. reinhardtii strain Stm6Glc4 described in [30]

Growth Rate m max
To identify conditions yielding maximum microalgae growth rates in TAP and PCM, a number of factors were tested including light intensity (low light: 35 mE m 22 s 21 and high light: 450 mE m 22 s 21 ), culture volume (100, 112.5, 125, 137.5 and 150 mL trials corresponding to a culture depth of 2.7, 3, 3.5, 4 and 5 mm, respectively) and inoculation density (OD 750 = 0.1 and 0.3). These experiments were conducted in a specially designed robotic Tecan Freedom Evo system which was fitted with three microwell shakers each capable of holding six 96 well microwell plates, a bank of fluorescent lights positioned 1.2 m from the microwell surface and a CO 2 controller which maintained CO 2 concentrations at 160.2% CO 2 during photoautotrophic conditions. The lights consist of 12 cool white fluorescent lights (Phillips PL-L55W/840 Cool White, Phillips International B.V. Netherland) and 11 warm white fluorescent lights (Phillips PL-L55W/830 Warm White, Phillips International B.V. Netherland) which were installed in an alternating arrangement. The chamber temperature was 3361uC. Algal cultures were adjusted to the desired OD 750 (OD 750 = 0.1 and 0.3, respectively) in TAP or PCM (supplemented with 1.5 mM tryptophan and 60 mM 5-FI for Stm6Glc5L01) and aliquoted into microwell plates. OD 750 measurements were recorded at 3 h intervals using a dedicated spectrophotometer (Tecan Infinite M200 PRO microwell plate reader). Growth curves were generated using Microsoft Excel and Graph Pad Prism software was used for growth curve fitting. An R 2 threshold of 0.85 was used to exclude false positives. The maximum value of the specific growth rate (m max ) was determined using a linear fit to ln OD 750 vs. time, based on equation 1: m = ln (DOD 750 )/Dt.
To exclude that the chosen 5-FI concentration of 60 mM impaired growth of the mutant strain, growth rate was monitored as described above in 150 mL TAP supplemented with tryptophan and different 5-FI concentrations (0, 30, 60, 90, 120, 240, 360, 480 mM) at 100 mE m 22 s 21 .

Biomass Determination
To determine direct biomass yields, Stm6Glc4 and Stm6Glc4L01 (grown in TAP medium) were adjusted to OD 750 of 0.3 in 144 mL PCM (OD 750 measurements were conducted using Tecan Freedom Evo system and a 96 well plate filled with 150 mL culture). Six 6 well microwell plates (total of 36 wells) for each strain were filled with 4 mL of adjusted cell suspension to a depth of 5 mm and incubated at a 1% CO 2 atmosphere for 43 h with 450 mE m 22 s 21 illumination. OD 750 was measured at the starting point and at 19 h and 43 h in the 6 well plates. After 43 h samples were harvested, pooled and washed in water to remove salt residues. Cells were dried at 60C until no further mass change could be observed and obtained dry mass was calculated as g per Liter culture.

Sequence Studies
All genome sequences were obtained from NCBI (National Center for Biotechnology Information, http://www.ncbi.nlm.nih. gov/, accessed March 2013) and analyzed using the BLAST tool provided by the Chlamydomonas Center (http://www.chlamy.org/, accessed March 2013). The cross BLAST analysis of all mRNA sequences against each other and the identification of nonhomologous parts of the sequence led to a map of unique gene regions of the LHCBM. Linker and sense sequence of the inverted repeat construct were picked from these regions; the anti-sense region was determined using the ''reverse complement'' function of Vector NTI and confirmed by BLAST alignment.

Plasmid Construction and Transformation
All nucleic acid work was conducted using standard procedures [34]. The LHCBM1 target sequence was synthesized based on a 95 bp sequence of LHCBM1 (bp 81-175), followed by an XbaI site. A subsequent 106 bp linker chosen from LHCBM1 39UTR (bp 831-915) and the reverse orientated sequence of LHCBM1 (bp 81-190) completed the RNAi target sequence. The LHCBM2 target sequence was synthesized from two fragments. Fragment A consisted of a part of the LHCBM2 39UTR (bp 1033-1132) followed by a SacI site. Fragment B contained the reverse orientated sequence to fragment A (bp 1033-1160) with subsequent EcoRI and SacI restriction sites. A linker region taken from LHCBM2 (bp 11-103) with a preceding SacI site and fragment B were amplified and joined via multiple template PCR. The linker and fragment B combination was cloned as a SacI fragment into SacI site downstream of fragment A to complete the inverted repeat sequence. The LHCBM3 target sequence was synthesized from bp 628-728 of the LHCBM3 59UTR followed by a BamHI site, a linker of LHCBM3 59UTR (bp 255-404) and the reverse orientated 59UTR sequence (bp 255-404) was synthesized. Additional cloning details are provided in the supplementary materials ( Figure S1).The inverted repeat sequences were flanked by EcoRI restriction sites to facilitate subsequent cloning into the transformation vector pBDH-R, and yielded vectors pMO52, pMO53 and pMO75, respectively. To create pBDH-R the Ribulose-1,5-bisphosphate-carboxylase/2oxygenase small subunit (RBCS) promoter, the intron fragment (with the start codon deleted) of RBCS was produced via fusion PCR and cloned into the XhoI and MluI site of pALKK1 [27] to replace its RBCS promoter/ ATG/intron/BLE region. This deletes an undesired BLE gene fragment present in pALKK1. To produce the triple knock-down strains, equal amounts of each vector pMO52, pMO53 and pMO75 were mixed and transformed into the strain Stm6Glc4 [30] using biolistic bombardment [35].

RNA Isolation, cDNA Synthesis and Quantitative Realtime PCR (qRT-PCR)
Isolation of total RNA was performed using a PureLink TM RNA Mini kit (Invitrogen) following the protocol optimized for animal and plant cells according to manufacturer's directions. As Stm6Glc4 is a cell wall deficient cell line the homogenization step was not carried out. DNase treatment was performed according to the PureLink TM RNA Mini kit manual using RQ1 RNase-free DNase (Promega). First strand cDNA synthesis was performed using iScript TM cDNA Synthesis kit (Bio-Rad Laboratories) and 1 mg total RNA according to manufacturer's instructions. Quantitative real-time PCR analysis was carried out in triplicate using an Applied Biosystem 7500 Real Time PCR System with software SDS version 1.2.3 and the standard method 'absolute quantification' two step RT-PCR for thermal profile and dissociation stage (stage 1:1 replication at 50uC, 1 min; stage 2:1 replication at 95uC, 10 min; stage 3:40 replications at 95uC, 15 sec; stage 4: dissociation; 1 replication at 95uC, 0.15 sec; 60uC, 1 min; 95uC, 0.15 sec). The experiment was carried out on a 96 well plate and each reaction contained 7.5 ml SYBRH Green Master Mix (Applied Bioscience), 5 ml of cDNA [1.6 ng mL 21 ], 2 mL of each primer [2 mM] and 3.5 mL of water. Relative expression levels were normalized using the C. reinhardtii reference gene CBLP [36]. Primer sequences for LHCBM1, LHCBM2, LHCBM3 and CBLP were designed as reported previously [27,36].

FACS Analysis and Chlorophyll Measurements
Cells were grown to late log phase in TAP liquid culture (OD 750 = ,1.0), 1 mL of the culture was used for analysis in a FACSCanto II Flow Cytometer (Becton Dickson). Comparative chlorophyll fluorescence data was recorded using a 488 nm blue excitation laser and the 670 nm long pass detector to measure emission.
Total chlorophyll concentration was measured according to previously reported methods [37]. 1 mL of algae culture was sampled at late log phase (OD 750 = ,1.0). The cells were then pelleted (5006g, 5 min, 20uC) and the clear supernatant discarded before being resuspended in 1 mL of ice cold 80% acetone. After vortexing the sample, the precipitate was pelleted (141006g, 4 min) before measuring the absorption (A) of the supernatant. The A 750 was set to zero and A 664 and A 647 measured subsequently. Chlorophyll content was calculated as previously described [37].

Demonstration of Algal Fluorescence at Different Depths
Stm6Glc4L01 and Stm6Glc4 were adjusted to same cell density (7.5*10 5 cells mL 21 ) and transferred to a 100 mL glass cylinder positioned on a Safe Imager TM blue-light transilluminator (Invitrogen) fitted with light emitting diodes producing a narrow emission peak centered at ,470 nm. Chlorophyll fluorescence was visualized by an 'amber' (530 nm long path) filter (Invitrogen).

H 2 Volume and Gas Composition Measurements
For H 2 production Stm6Glc4L01 and Stm6Glc4 were grown to late log phase in TAP medium. Cells were harvested, washed twice with sulfur-free TAP medium and resuspended in sulfur-free TAP medium containing glucose [1 mM] [30] and adjusted to the same chlorophyll concentration (14.5 mg mL 21 ). H 2 measurements were performed under continuous white light (450 mE m 22 s 21 ) using a custom-built PBR system with a dedicated gas collection tube mount [32]. The evolved gas was sampled using a gas-tight Hamilton syringe and injected at regular intervals into a gas chromatograph (Agilent Micro GC3000) fitted with a PlotU precolumn (3 m x 0.32 mm) and MolSieve 5APlot column (10 m x 0.32 mm). Argon (32.5 psi, pound per square inch) was used as the carrier gas and H 2 , O 2 and N 2 concentrations were monitored.

Identification of LHCBM Target Sequences
To develop specific LHCBM knock-down strains, the identification of highly specific target sequences is essential. As all LHCBM genes display a high degree of sequence similarity [27], genomic BLAST analysis of the three target mRNAs, encoding the major light harvesting complex proteins of PSII (LHCBM1, 2 and 3), was performed. This analysis aimed to identify target regions that had minimal similarities to other genes and so prevent their co-suppression. Based on this analysis, discrete regions for each single LHCBM target (between 95-100 bp) were chosen for the development of RNAi constructs (sequences see Figure S1). To minimize non-specific RNAi effects further, the LHCBM1-, 2and 3-specific RNAi constructs were engineered so that the linkers positioned between the sense and anti-sense regions of the construct were specific to the respective target genes. The plasmid pALKK1 [27] was used to create the transformation vector pBDH-R ( Figure 1A), additionally providing an RNAi target site for the endogenous tryptophan synthase. This allowed direct selection of positive transformants on media plates containing 5-fluoroindole (5-FI), as tryptophan synthase converts 5-FI into the toxic tryptophan analogue 5-fluorotryptophan [38].

Development and Selection of Positive Transformants
Single LHCBM knock-out or knock-down transformants have not to date shown any marked phenotype in terms of chlorophyll concentration. This suggests that the knock-down of one LHCBM may be compensated for by the over expression of the remaining LHCBM proteins [39,40]. Consequently a triple LHCBM1/2/3  [38]). Inverted repeat (IR) sequences used to target LHC genes (Lhcbm IR) and linker (Linker), which spaces the inverted repeats, are located in between two Maa7 inverted repeats (Maa7 IR). In this study 'Lhcbm IR' and 'Linker' were replaced with the sequences from the target LHCBM genes to minimize non-specific knock-down effects. EcoRI restriction sites used for cloning are marked. (B) mRNA levels of the three targeted LHCII genes (LHCBM1 to LHCBM3) were determined in triplicate via quantitative real-time PCR and normalized to CBLP mRNA [36]. knock-down strategy was tested. LHCBM1, 2 and 3 were selected as they are reported to be the most abundant LHCII associated light harvesting proteins and so would be expected to achieve the greatest amount of antenna reduction [19,20]. To develop triple knock-down mutants of LHCBM1, LHCBM2 and LHCBM3, equal amounts of the target vectors were mixed and used to transform the high H 2 producing C. reinhardtii strain Stm6Glc4 [30] using biolistic bombardment [35]. Positive transformants were selected on agar plates containing 5-FI since, as based on the vector design, it was expected that transformants with knocked-down tryptophan synthase, and thus resistance to 5-FI, would co-suppress the desired LHC protein. This strategy enabled the testing of all possible combinations of LHCBM1, LHCBM2 and LHCBM3 knock-downs, with the low-antenna (light green) phenotype as the secondary selection criterion. Consistent with our hypothesis that the simultaneous knock-down of these major LHCs could result in a marked reduction in chlorophyll concentration, a number of light green LCHBM1/2/3 multiple knock-down transformants were identified, of which the mutant with the palest phenotype (designated Stm6Glc4L01 to signify light harvesting mutant combination 01) was chosen for further characterization.

LHC mRNAs and Phenotypic Characterization
Total mRNA levels of LHCBM1, LHCBM2 and LHCBM3 in Stm6Glc4L01 were determined using quantitative real-time PCR (qRT-PCR) to establish which of the target LHCBM genes were primarily affected. To confirm that equal amounts of cDNA were used in this experiment, data were compared with those of 18S RNA and the mRNA for endogenous C. reinhardtii gene CBLP (Chlamydomonas ß-subunit-like polypeptide), which was previously shown to be stably expressed throughout growth and H 2 production [36]. Data normalized to CBLP ( Figure 1B) showed that LHCBM1 has been down-regulated to 20.6% 60.27% and LHCBM3 to 41.4% 60.05% of their original levels, respectively. The down-regulation of LHCBM2 was less dramatic (81.2% 60.037%) but importantly, LHCBM2 levels did not increase to compensate for the knock-down of LHCBM1 and LHCBM3. As gene dosage compensation has been suggested to be responsible for the unaffected phenotype of single antenna protein knock-outs [39][40][41], it is possible that the RNAi construct targeting LHCBM2 functions primarily to limit its expression level, preventing such compensation. Data with absolute expression levels in addition to the fold-differences relative to CBLP ( Figure 1B) are provided (Table S1). The similarity of expression levels for 18S and CBLP genes between Stm6Glc4 and Stm6Glc4L01 suggests that the reduction in LHCBM1 and -3 levels is both meaningful and significant.
Light micrographs of Stm6Glc4 and Stm6Glc4L01 recorded under identical conditions ( Figure 1C) showed that Stm6Glc4L01 cells contained considerably lower chlorophyll levels than the Stm6Glc4 parental control, consistent with the observed downregulation of LHCBM1, LHCBM2 and LHCBM3 mRNA ( Figure 1B) in a manner similar to that previously reported for other antenna mutants [16,26,27]. Auto-fluorescence images ( Figure 1D), support the optical microscopy results and confirm that Stm6Glc4L01 accumulates less chlorophyll per cell, as fluorescence levels are significantly lower than in Stm6Glc4. The increased chlorophyll a/b ratio (Mean: Stm6Glc4:2.2960.05 and Stm6Glc4L01:2.6260.09; P-value ,0.0021 [37] is indicative of a specific reduction in the ratio of ChlB-rich LHC proteins compared with the ChlA-rich PSII core complexes [42] confirming LHC antenna size reduction.

Fluorescence Properties and Chlorophyll Measurements
Flow cytometry (Figure 2A) Figure 2A and displayed in Figure 2B as percentage of Stm6Glc4L01 (100%). This suggests that the control cell line Stm6Glc4 exhibited ,175% higher levels of fluorescence losses than those observed in Stm6Glc4L01 mutant (100%). Furthermore as the red fluorescence emission observed is due to the transition from the lower energy state of chlorophyll to the ground state, a concomitant heat loss is also be expected due to the transition from the higher to the lower energy state transition. Therefore Stm6Glc4L01 is expected to display less heat losses than parental strain Stm6Glc4.
Though in practice, fluorescence measured by flow cytometry may not be strictly linear with respect to chlorophyll content [43], it indicates that Stm6Glc4L01 exhibits a significant reduction compared to the parent strain. As the mean cell diameter of the two strains was similar ( Figure 1C) a reduction in cell size cannot explain the reduced mean fluorescence in Stm6Glc4L01. The reduction in fluorescence in Stm6Glc4LO1 is however consistent with the increased chlorophyll a/b ratio (Stm6Glc4:2.2960.05; Stm6Glc4L01:2.6260.09, Figure 2C) which is indicative of a reduction in the ratio of ChlB-rich LHC proteins compared with the ChlA-rich PSII core complexes [42] as well as the total reduction in chlorophyll concentration per cell ( Figure 2D). The observation that the ChlA/B ratio has only increased from 2.29 to 2.62 is completely consistent with many independent observations including the following. The first is the fact that purified thylakoid membranes containing PSII-LHCII and PSI-LHCI typically have a ChlA/B ratio of about 1.9-2.3. The second is that purified PSIIcore complexes from higher plants binding CP29, CP26 and CP24 (all of which bind ChlB), but which are almost completely devoid of LHCII and LHCI, have a ChlA/B ratio of ,7 [44]. The fact that the ChlA/B ratio of Stm6Glc4L01 cells is closer to 2.3 than 7 is therefore completely consistent with the fact that in addition to the ChlA/B binding proteins CP26 and CP29, the Stm6Glc4L01 cells also contain the full complement of LHCA Chla/b proteins of which there are ,9 different types, as well as LHCBM4-9 and residual levels of LHCBM1-3. Collectively this explains why the ChlA/B ratio of Stm6Glc4L01 cells is significantly higher than that of the control (2.29+/20.05 vs 2.62+/20.09) but closer to 2.3 (wt thylokoids) than 7 (purified PSII core complexes).
Furthermore a determination of the total chlorophyll content [mg mL 21 ] showed that at the same cell number, total chlorophyll content was reduced by 50% in Stm6Glc4L01 compared to Stm6Glc4 ( Figure 2D) (total chlorophyll content: Stm6Glc4:13.65 mg mL 21 , Stm6Glc4L01:6.77 mg mL 21 , P-value ,0.0006.

Light Penetration
To measure whether the decrease in LHCBM protein expression level increased light penetration through a liquid culture column, thereby making more light available for cells deeper in the culture, auto-fluorescence of Stm6Glc4 and Stm6Glc4L01 at the same cell density was measured ( Figure 3A). As expected, Stm6Glc4 cultures were a darker green color compared to those of Stm6Glc4L01, due to their larger LHC antenna. The cultures were then placed onto a blue light illuminator and chlorophyll auto-fluorescence imaged using an orange filter to block the blue background light. As illumination was provided from below, the height of the fluorescent column was indicative of the light penetration distance. The fluorescence throughout the Stm6Glc4L01 culture column illustrates that light penetrated ,4 times further than in Stm6Glc4 at the same cell density ( Figure 3B).

Growth Rates Under Mixotrophic and Photoautotrophic Conditions
As light is both necessary and damaging for photosynthesis [45] as it is required to drive charge separation but in excess causes photoinhibition, the maximum rate of photochemistry is achieved at the point at which the photosystems are just saturated, assuming that light is the limiting factor. To minimize the area required for microalgae production and thereby also PBR costs, in theory systems should operate at the maximum flux density (high light) and the system poised at this optimal point by adjusting culture depth and cell density. Theoretically small antenna cell lines could allow more cells to be packed into a given unit volume before light becomes limiting and decrease both cell shading and photoinhibition. To test this hypothesis biomass production (as a proxy for chemical energy/biofuels) in form of maximum growth rates (m max ) of Stm6Glc4 and Stm6Glc4L01 cultures were measured under a range of culture depths and cell densities.
The growth rate of Stm6Glc4 and Stm6Glc4L01 was next determined under mixotrophic conditions (TAP medium) which are the best-established conditions for photobiological H 2 production. Experiments were conducted in microwell plates using three different volumes (100 mL, 125 mL and 150 mL) to simulate different culture depths (average depths of 2.7, 3.5 and 5 mm, respectively). Since an advantage of Stm6Glc4L01 was expected primarily in high light conditions, growth was measured under 450 mE m 22 s 21 illumination (optimal light intensity for the parental control and wt strains is 50-100 mE m 22 s 21 ) with an initial inoculation OD 750 of 0.1. Under all conditions Stm6Glc4L01 showed a higher growth rate than Stm6Glc4 ( Figure 4A). In the shortest path length cultures (100 mL = 2.7 mm) exposed to the highest light intensity (450 mE m 22 s 21 ) Stm6Glc4L01 showed ,3 times the growth rate of Stm6Glc4 based on OD 750 measurements. The increased growth rate could be due to reduced photoinhibition (due to the reduced LHC antenna size), improved light penetration and therefore increased total light-to-biomass conversion efficiency of the culture, or a combination of both factors. As the highest m max h 21 values for Stm6Glc4L01 were observed in the shortest path length cultures (2.7 mm) and declined with increasing culture depth ( Figure 4A) we concluded that path length and light distribution had a more dominant impact than photoinhibition. In contrast Stm6Glc4 showed no such trend. It was previously reported [31] that Stm6 (and hence Stm6Glc4) is light sensitive, demonstrated here by the poor growth displayed at very low cell densities, where no self-shading occurs at 450 mE m 22 s 21 ( Figure 4A).
To investigate the growth behavior of Stm6Glc4 and Stm6Glc4L01 under photoautotrophic conditions, we performed a parallel experiment ( Figure 4B Previous experiments in TP medium (TAP medium minus acetate) indicated that N and P were limiting and led to an early culture growth plateau. Consequently for this experiment, an improved in-  house medium was developed (Photoautotrophic Chlamydomonas Medium, PCM) which supplied more nitrogen and phosphate, to ensure that these did not limit growth. All cultures were inoculated at a starting OD 750 of 0.3 to reduce light stress for Stm6Glc4 at 450 mE m 22 s 21 .
At a culture depth of 2.7 mm and 450 mE m 22 s 21 illumination, Stm6Glc4L01 showed ,120% of the maximum growth rate observed for the parental strain Stm6Glc4. Further increases in depth accentuated the advantage of the Stm6Glc4L01 phenotype, as the growth rate at 5 mm depth was .185% compared to parental strain Stm6Glc4. This result shows that a reduction in antenna size improved growth across the conditions tested but particularly in deeper and denser cultures. In contrast, under low light levels (35 mE m 22 s 21 ) growth rates of Stm6Glc4 and Stm6Glc4L01 were comparable, indicating that the reduced antenna did not disadvantage the cells.
To confirm the OD 750 measurements, direct biomass yields were also determined both for Stm6Glc4 and Stm6Glc4L01 grown in PCM under the best high light conditions tested (450 mE m 22 s 21 , 5 mm culture depth, 1% CO 2 atmosphere). OD 750 was recorded at the starting point (0.113 for Stm6Glc4) and 0.169 forStm6Glc4L01) and at time intervals (19 h and 43 h) ( Figure 4C). To produce sufficient biomass to allow accurately biomass determination, samples were harvested at 43 h (OD 750 : Stm6Glc4:0.692 g and Stm6Glc4L01:0.803 g) and yielded 2.3 g L 21 (Stm6Glc4) and 2.5 g L 21 (Stm6Glc4L01) biomass dry weight, respectively. From this the biomass dry weight/OD 750 nm ratios of Stm6Glc4 and Stm6Glc4L01 were calculated to be 3.3 (2.3/0.692) and 3.1 (2.5/0.803) respectively. Based on these ratios, biomass yields at earlier time points were calculated Figure 4C. Although it appears initially that the differences in biomass yield at the 43 h time point are only approximately 10%, this was because the Stm6Glc4L01 growth rate had started to plateau presumably due to exhausted nutrient resources. In contrast at the point of maximum growth (19 h) biomass concentrations were by this method calculated to be 0.818 g L 21 (Stm6Glc4) and 1.35 g L 21 (Stm6Glc4L01). This represents a 165% increase for Stm6Glc4L01 over the control Stm6Glc4 level (100%).

Hydrogen Production
As photobiological H 2 production is the most efficient light-tobiofuel conversion strategy [46] the effect of specific LHC gene down-regulation in Stm6Glc4L01 on H 2 generation was investigated. For this purpose a time course measuring H 2 production rates (mL H 2 L 21 h 21 ) in Stm6Glc4 and Stm6Glc4L01 cultures, adjusted to the same chlorophyll concentration (14.5 mg mL 21 ), was carried out in sulfur deprived media to establish anaerobiosis ( Figure 5). At constant chlorophyll concentration and identical volumes, the cell number of Stm6Glc4L01 was approximately double the cell number of Stm6Glc4 (Stm6Glc4:1610 7 cells mL 21 ; Stm6Glc4L01:2610 7 cells mL 21 ). H 2 production in the Stm6Glc4L01 trials started almost immediately after transfer into sulfur deprived media in marked contrast to the ,22 h lag time of the Stm6Glc4 control ( Figure 5A). This suggests that intracellular O 2 concentrations were lower in Stm6Glc4L01 than in Stm6Glc4 and close to or below the induction threshold for HYDA expression. The reduced lag phase has marked operational benefits, and significantly increases the yield. The H 2 production rates of both strains peaked at 47 h after sulfur deprivation, with Stm6Glc4L01 producing 6.25 mL L 21 h 21 compared to 3.31 mL L 21 h 21 for Stm6Glc4 under the conditions tested. Both strains produced H 2 until 188 h after sulfur deprivation with a similar decline in both strains after peak production. Under the conditions tested Stm6Glc4L01 produced a total volume of 361 mL 627.6 mL H 2 per liter culture compared to 198 mL 621.2 mL H 2 per liter culture produced by Stm6Glc4. Stm6Glc4L01 therefore produced .180% more H 2 per unit chlorophyll than the Stm6Glc4 control ( Figure 5B), which via gas chromatography was determined to be .95% pure. These results clearly show that Stm6Glc4L01 exhibited improved characteristics in terms of early onset and higher rates of H 2 production.

Photosynthetic Efficiency
Light capture is the first step of all microalgae-based fuel production and its optimization is therefore of fundamental importance for the development of high-efficiency processes. Wild type microalgae antennae have evolved intricate mechanisms to adapt to natural fluctuations in light levels and quality to maximize light capture under low light conditions and dissipate excess light under high light conditions. While algae possess the ability to down-regulate their antenna, even 25% of full sunlight is sufficient to saturate the photosynthetic capacity of most algal species [17]. Under such over-saturated light conditions for a well-mixed culture in an open raceway or closed PBR, energy losses of up to 80% can occur [47]. This is because cells near the illuminated surface dissipate excess solar energy via NPQ which is then no longer available to cells deeper in the culture and which are therefore light-limited. Consequently, while these mechanisms are finely tuned for adaptation to their natural environment, they are far from optimized for industrial microalgae production, which to minimize area requirement and photobioreactor (PBR) construction costs should ideally be established for high flux density conditions. Photobioreactors can be designed to achieve light dilution (e.g. to 10% of 2000 mE m 22 s 21 ), but due to increased complexity and material costs this increases photobioreactor expense. Antenna engineering has the advantage that it can theoretically improve production efficiency and reduce capital costs both of which are major economic drivers for this technology [48].
Given the large number of LHC proteins in eukaryotic photosynthetic organisms, a diversity of function among family members is implied. The generation of Stm3LR3 [27] involved knocking-down all LHC classes and resulted in a low chlorophyll phenotype. It was however recognized during this initial antenna engineering step that some vital LHC functions might also be affected. This assumption was borne out by subsequent work [49] and the observation that Stm3LR3 is moderately light-sensitive at low cell densities. For example, CP29 (LHCB4) is essential for state transitions and for docking the LHCBM1, 22 and 23 trimers to PSI [49], while a role for CP26 (LHCB5) in preventing photoinhibition has also been described [13,15,[49][50][51]. It is therefore likely that the pattern of LHC down-regulation displayed in Stm3LR3 does not preserve all of the functions required to support growth at high light levels [40].
While the high degree of homology in LHCs reflects a common ancestry, it also suggests that specific functions likely reside in the small non-conserved regions. Consequently, precise engineering of specific LHC proteins is important to avoid undesired effects due to co-suppression of other genes. As PSII is the main site of lightinduced damage, there is a strong correlation between PSII antenna size and photosynthetic productivity [47,[52][53][54]. This supports the idea that the most abundant LHCII antenna proteins, LHCBM1, 22 and 23, are the most important targets for downregulation. In this context it is of note that while the exact molecular mechanisms for photodamage of PSII are still under investigation (for review see [54]), it has been shown that reduced chlorophyll content plays a crucial role in preventing photodamage, and thus allows the cells to grow under higher light conditions [55], supporting our findings. The above experiments show that LHCBM1 (20.6% 60.27%), LHCBM2 (81.2% 60.037%) and LHCBM3 (41.4% 60.05%) expression levels were effectively knocked-down ( Figure 1B), resulting in the light green mutant Stm6Glc4L01 ( Figure 1C), which had a reduced total chlorophyll content (50% of parental strain ( Figure 2D)) and an increased ChlA/B ratio (Stm6Glc4:2.2960.05 vs. Stm6Glc4L01:2.6260.09) ( Figure 2C), confirming LHCII depletion. Down-regulation of the LHCII antenna system was further supported by chlorophyll auto-fluorescence measurements obtained by flow cytometry (Figure 2A and B). Due to its reduced antenna size, cultures of Stm6Glc4L01 exhibited considerably improved light penetration ( Figure 3B) over those of the control (Stm6Glc4) at the same cell density. Our data also indicates that the improved light penetration is in part due to reduced fluorescence and heat losses at the illuminated culture surface ( Figure 2B) and in part due to the reduced chlorophyll content of each cell ( Figure 2D) which improves light transmission through the culture ( Figure 3B). Together these properties resulted in an increased fraction of cells in the culture being poised close to the optimum point of light saturation (achieving maximal rates of photochem-istry, without causing photo-inhibition), rather than within the broad range of illumination levels typically experienced by cells in the Stm6Glc4 control culture. This improved poise was achieved at light levels of 450 mE m 22 s 21 instead of the more typical level of 100 mE m 22 s 21 at which Stm6Glc4 antenna cell line is saturated and led to growth rates of .185% of Stm6Glc4 under photoautotrophic conditions ( Figure 4B). The importance of this is that under photoautotrophic conditions required for both biomass and ideally commercial scale H 2 production, at outside light levels (e.g. 2000 mE m 22 s 21 ), the light dilution factor could be reduced from 2000/100 = 20 to 2000/450 = ,4.4 [52], offering the potential for significant savings in photobioreactor costs.

Molecular Mechanism of Improved H 2 Production
The results presented are summarized with the following mechanistic model (Figure 6). In wt cells, Stm6 and Stm6Glc4, light used by PSII is predominantly captured by LHCBM trimers bound to the PSII core complex. The major LHCII proteins are reportedly transcribed from 9 different genes LHCBM 1-9 which are numbered according to the relative expression levels initially observed (16,17). These LHCBM proteins are reported to trimerize and their primary role is to capture solar energy and funnel it via the minor LHCII proteins, CP29 (LHCB4) and CP26 (LHCB5), to the PSII core (CP47, CP43, D1, D2, cytb559, PSBO and additional small and extrinsic subunits) ( Figure 6). Down regulating LHCBM1-3 which are the most abundant of these, as expected, resulted in a marked reduction in antenna size. It should be noted that LHCBM1 has also been reported be to be involved in NPQ as well as the scavenging of radical oxygen species [40]. While the down regulation of these functions might be expected to have a detrimental effect at high light levels, our results indicate improved biomass and H 2 yields at 450 mE m 22 s 21 (.100 mE m 22 s 21 can saturate Stm6 and Stm6Glc4). This can be explained by the small antenna size of Stm6Glc4L01 resulting in lower levels of over excitation of PSII. This in turn is expected to result in the reduction of the formation of radical oxygen species and the requirement for the dissipation of excess energy via NPQ. Consequently the down regulation of the NPQ and ROS scavenging functions through the knock-down of LHCBM1 appears to be compensated for by the reduction of antenna size.
The light captured by the PSII antenna system drives the PSIImediated oxidation of water to H + , eand O 2 ( Figure 6A) [32]. Electron transport subsequently progresses via plastoquinone (PQ), cytochrome b 6 f, plastocyanin to PSI where electrons are again energized by light, to reduce ferredoxin (Fd). Ferredoxin nucleotide reductase (FNR) couples the reoxidation of Fd to the reduction of NADP + which under normal aerobic conditions is then used during carbon fixation to reduce CO 2 to sugars. During this process, protons are translocated to the lumen establishing the proton gradient which drives ATP production. Apart from water, products derived from starch metabolism can be used to provide electrons to PQ via the NADPH dehydrogenase (NDA2) [56]. Under illuminated anaerobic/micro-oxic conditions the derived H + and ecan also act as substrates for HYDA-mediated H 2 production. It is of note however that up to 80% of the H + and eused by HYDA are reportedly produced by PSII [57], providing a direct pathway for the solar driven H 2 production from water.
H 2 production requires both illumination (to drive photosynthesis) and anaerobic conditions (to induce the expression of HYDA and prevent HYDA inhibition) [10]. Under anaerobic conditions, mitochondrial respiration largely ceases as the cell is deprived of O 2 . The expression of HYDA however provides an alternative H + /erelease mechanism which recombines H + and eto produce H 2 gas that is excreted from the cell (i.e. instead of H 2 O from mitochondria). This allows photosynthetic electron transport to continue to operate, providing an alternative pathway for the production of ATP and NADPH to enable cell survival (11).
In the wt, only small amounts of H 2 are typically produced as the large antenna phenotype increases oxygen production in illuminated cells, which is not removed quickly enough by wt respiration levels and so leads to a reduction in HYDA transcription and function. Typically, anaerobiosis can be established in wt algae cultures through sulfur depletion. This is because the Methionine containing D1 protein of PSII, which is rapidly photo-damaged under high light conditions, can only be repaired in the presence of sulfur. Consequently, by limiting sulfur, PSII-mediated O 2 evolution is gradually reduced over time. When it drops below the rate of O 2 uptake by mitochondrial respiration, the system becomes micro-oxic or anaerobic and induces HYDA expression [58].
The parental strains of Stm6Glc4L01 are moc1 mutants (i.e. Stm6 and Stm6Glc4 [30][31][32]). Both have an increased respiration rate due to upregulated AOX activity (high O 2 consumption), but also have a large wt like PSII antenna system ( Figure 6A). As a result of their increased rate of respiration Stm6 and Stm6Glc4 exhibit a higher rate of oxygen consumption than the wt and so enter anaerobiosis faster under sulfur deprivation. This in turn results in earlier activation of HYDA expression [59] compared to the wt leading to the rapid onset of H 2 production. However the Figure 6. Mechanistic model of improved H 2 production in Stm6Glc4L01. (A) Stm6Glc4 has a large PSII antenna system consisting of LHCBM1-9. LHCBM1-3 are reported to be most abundant. Large antenna size results in increased PSII mediated O 2 production and NPQ losses. NPQ losses reduce system efficiency; intracellular O 2 levels inhibit expression of HYDA until the system is sulfur deprived (sulfur required for the repair of the PSII-D1 subunit. B: Stm6Glc4L01 has a reduced antenna size which is figuratively shown, and leads to reduced O 2 production and early onset of H 2 production. The light green phenotype allows higher cell densities to be used leading to increased rates of H 2 production. doi:10.1371/journal.pone.0061375.g006 relatively large antenna systems of Stm6 and Stm6Glc4 still supports rates of O 2 production which exceed the respiration rate and so sulfur depletion is required to induce H 2 production. In contrast in Stm6Glc4L01 the LHCBM1, LHCBM2 and LHCBM3 have been down-regulated within an upregulated AOX background ( Figure 6B). The importance of this is that the simultaneous reduction of O 2 production and increased O 2 consumption is thought to reduce intracellular O 2 levels below the induction threshold for HYDA. This hypothesis is supported by three observations. First, Stm6Glc4L01 cultures showed an almost immediate onset of H 2 production upon sulfur deprivation (Figure5A) indicating that the intracellular O 2 levels in sulfur replete medium are already poised closed to HYDA threshold level. Second, during the early stages of H 2 production the contaminating O 2 concentrations in the gas produced by Stm6Glc4LO1 cultures were considerably lower than those produced by Stm6Glc4. Third, the rate of H 2 production by Stm6Glc4L01 was almost twice as high as that observed for Stm6Glc4 ( Figure 5B) for a given culture volume adjusted to the same chlorophyll concentration, for reasons described above.
The finding that the Stm6Glc4L01 is so closely poised to HYDA expression opens up the exciting possibility of using this mutant for continuous H 2 production in sulfur replete medium. To date, to our knowledge continuous H 2 production in sulfur replete liquid medium has not been reported, presumably because under these conditions O 2 production usually exceeds O 2 consumption inhibition HYDA expression. For example the strains Stm3LR3 [27], Stm6Glc4T7 [26] and TLA1 [16,60] have not been reported to increase H 2 yield in liquid culture which would be desirable for the development of continuous H 2 production in scale PBR systems. However, important steps along this development path include the use of sulfur microdosing approaches [61] and the use of alginate solid phase systems which reduced O 2 inhibition [60].
In summary, Stm6Glc4L01 exhibited an improved H 2 and biomass production efficiency (165-180% improvement over parental strain). Importantly, this was achieved at increased solar flux densities (450 mE m 22 s 21 ) and high cell densities which are best suited for microalgae production as light should ideally be the limiting factor. Our data suggests that the overall improved photon to H 2 conversion efficiency is due to: 1) reduced loss of absorbed energy by non-photochemical quenching (fluorescence and heat losses) near the photobioreactor surface; 2) improved light distribution in the reactor; 3) reduced photoinhibition; 4) early onset of HYDA expression, 5) reduction of O 2 induced inhibition of HYDA. The Stm6Glc4L01 phenotype therefore provides important insights for the development of high efficiency photobiological H 2 production systems.