Unlocking the Constraints of Cyanobacterial Productivity: Acclimations Enabling Ultrafast Growth

ABSTRACT Harnessing the metabolic potential of photosynthetic microbes for next-generation biotechnology objectives requires detailed scientific understanding of the physiological constraints and regulatory controls affecting carbon partitioning between biomass, metabolite storage pools, and bioproduct synthesis. We dissected the cellular mechanisms underlying the remarkable physiological robustness of the euryhaline unicellular cyanobacterium Synechococcus sp. strain PCC 7002 (Synechococcus 7002) and identify key mechanisms that allow cyanobacteria to achieve unprecedented photoautotrophic productivities (~2.5-h doubling time). Ultrafast growth of Synechococcus 7002 was supported by high rates of photosynthetic electron transfer and linked to significantly elevated transcription of precursor biosynthesis and protein translation machinery. Notably, no growth or photosynthesis inhibition signatures were observed under any of the tested experimental conditions. Finally, the ultrafast growth in Synechococcus 7002 was also linked to a 300% expansion of average cell volume. We hypothesize that this cellular adaptation is required at high irradiances to support higher cell division rates and reduce deleterious effects, corresponding to high light, through increased carbon and reductant sequestration.

I t has long been established that photoautotrophic growth is dependent on the combined rates of light, carbon, and macronutrient acquisition and the efficiencies by which these resources are directed toward biomass synthesis (1). The inherent differences between the rates of light and dark processes highlight an important paradigm that defines cyanobacterial productivity constraints, whereby photosynthetic energy acquisition and CO 2 fixation are interdependent but maintain distinct enzymatic mechanisms and kinetics (2,3). Under high irradiances, photosynthesis is adversely affected by damage (photoinhibition) to the light-harvesting machinery (1) and/or by photorespiration as RuBisCO selectivity shifts toward O 2 (4). Notably, most of what is known about these effects as a function of irradiance and O 2 tension concerns photosynthesis (5-7) rather than cell growth.
The typical range of doubling times for well-characterized unicellular cyanobacteria (e.g., Synechocystis sp. strain PCC 6803 and Synechococcus elongatus PCC 7942) is between 7 and 12 h (8,9). Several hypotheses concerning mechanisms constraining cyanobacterial growth rates have been proposed; these include spatial restrictions within the cell that limit diffusion processes (10) as well as metabolic costs that determine partitioning of cellular resources and resulting fitness (11,12). The optimization model, developed to simulate the partitioning of material and energy within a photoautotrophic cell (11), has extended the concept of growth as a function of proteome allocation between adaptation to niche-specific environments and cell division resources. An important implication of this is the ability to minimize the requirement for niche-adaptive responses that may be a key for cyano-bacteria to redirect energy and nutrients efficiently toward the biosynthesis of biomass (13). This also holds significance for understanding regulatory switches governing central metabolism and secondary biosynthetic pathways, which are primary targets for bioengineering of cyanobacteria (14,15).
In this study, we systematically dissected the growth and photophysiological performance of Synechococcus sp. strain PCC 7002 (hereafter Synechococcus 7002), a fast-growing euryhaline unicellular cyanobacterium (16) that has become a promising biotechnological platform (17)(18)(19)(20). Although the physiological behavior of Synechococcus 7002 has been investigated under a wide range of irradiance, temperature, and salinity conditions (1,9,(21)(22)(23), there is still a paucity of information concerning principles underlying cyanobacterial growth efficiency and robustness. The photosynthetic potential, thresholds of inhibition, and growth are not explicitly equivalent, and significant variation is expected between different environments (24). Herein, we compared the growth and photosynthetic rates of Synechococcus 7002 across varied incident irradiance and dissolved O 2 with a slower growing unicellular cyanobacterium, Cyanothece sp. strain ATCC 51142 (hereafter Cyanothece 51142) (25). Cyanothece 51142 was chosen as a comparative organism because it is a well-characterized strain that is amenable to continuous cultivation (15,17,26). The growth rates of Cyanothece 51142 are similar, compared to those displayed by other model cyanobacterial strains, such as Synechocystis strain 6803, and Synechococcus elongatus strain 7942 (8,9). The irradiance-dependent response was further investigated through global RNA sequencing analysis to correlate the ultrafast growth with the photophysiological dynamics and gene expression of Synechococcus 7002. Through integration of state-of-the-art cultivation with photophysiological kinetic analyses and transcriptomic measurements, this genome study provides a new level of insight into the mechanisms guiding the energy and resource partitioning in cyanobacteria and sheds light on the phenomenon of ultrafast photoautotrophic growth.

RESULTS
Ultrafast growth under high irradiance. Optically thin (optical density at 730 nm [OD 730 ] of 0.082 Ϯ 0.003) cultures of Synechococcus 7002 and Cyanothece 51142 were grown under turbidostat control to maintain nutrient-replete steady states operating at the maximum specific growth rate obtainable for a given environmental condition (27,28). Under these conditions, Synechococcus 7002 displayed specific growth rates () that increased with irradiance to a maximum value of 0.20 Ϯ 0.01 h Ϫ1 . Growth was not inhibited even at the highest incident irradiance (I i ), 760 mol photons · m Ϫ2 · s Ϫ1 , examined in this study (Fig. 1A). In contrast, specific growth rates for Cyanothece 51142 reached a maximum of 0.07 Ϯ 0.01 h Ϫ1 at 430 mol photons · m Ϫ2 · s Ϫ1 and then declined at higher irradiances. The -I i relationships displayed saturating and/or peak functional trends similar to typical photosynthesisto-irradiance (P-I) curves (see Fig. S1 in the supplemental material) (29,30).
The steady-state net specific rates of photosynthesis (q O2 ) followed a similar increase as (Fig. 1B) Ϫ1 and subsequently decreased at higher irradiances. These results confirm that unlike Synechococ-cus 7002, Cyanothece 51142 is susceptible to photoinhibition during steady-state growth at I i of Ͼ430 mol photons · m Ϫ2 · s Ϫ1 . The observed effects on growth occurred at low dissolved O 2 concentrations (Յ8.0 and Յ2.5 M for Synechococcus 7002 and Cyanothece 51142, respectively).
Based on the growth and photosynthetic relationships with I i , the physiological responses can be categorized into two distinguishable regimes: (i) a "light-limited" regime, characterized by an increasing linear response to I i ; and (ii) a "light-saturated" regime, which varied from asymptotic growth in Synechococcus 7002 to inhibition in Cyanothece 51142. The two regimes were separated by a "transitional" phase defined here as the responses at or near the theoretical saturating irradiance I k (31). The observed bimodal behavior was further supported by net growth-tophotosynthesis yields that are proxies for the photosynthetic quotient (Q) (Cmmol biomass/mmol of O 2 ), which was defined here as the ratio of the net rate of carbon fixation into biomass to the net rate of oxygenic photosynthesis. For Synechococcus 7002, the photosynthetic quotients were higher during light-limited growth (Q lim ϭ 1.3 Ϯ 0.3) than light-saturated growth (Q sat ϭ 0.6 Ϯ 0.1) (Fig. 1C). In contrast, the Q values calculated for growth of Cyanothece 51142 varied only from 1.3 Ϯ 0.2 to 1.1 Ϯ 0.4 during light-limited and light-saturated conditions, respectively (Fig. 1D). These results show that Q is greatest during lightlimited growth of Synechococcus 7002 and also indicate that more reductant is required to support fast growth at higher incident irradiances. The same cannot be concluded for Cyanothece 51142, as much slower (Ն3-fold) growth rates are supported by an effectively constant amount of net oxygenic photosynthesis regardless of whether I i is limiting or saturating (i.e., inhibiting) relative to growth.
Photosynthetic performance as a function of growth phase and irradiance. Interrogation of the photosynthetic apparatus via chlorophyll (chlorophyll a [Chl a]) fluorescence techniques identified further differences between how Synechococcus 7002 and Cyanothece 51142 acclimate to increasing irradiance. As a general trend, the parameters associated with photosystem performance and photosynthetic electron transfer demonstrated distinct bimodal distributions as a function of light availability (i.e., limited versus saturated regimes). The most remarkable difference between Synechococcus 7002 and Cyanothece 51142 was the contrasting trends of cyclic electron flow (rate of cyclic electron transport [rCEF]), determined by postillumination fluorescence (Fig. 2A). Synechococcus 7002 displayed a positive rCEF rate trend across all I i values, with the exception of the highest illumination (760 mol photons · m Ϫ2 · s Ϫ1 ). In contrast, the rCEF rates in Cyanothece 51142 followed both growth and photosynthetic trends and displayed three distinct phases: a fast increase in rCEF associated with light-limited growth (I i Ͻ 430 mol photons · m Ϫ2 · s Ϫ1 ), deceleration during the transition I k (~430 mol photons · m Ϫ2 · s Ϫ1 ), and steep decline under light-saturated conditions (I i Ͼ 430 mol photons · m Ϫ2 · s Ϫ1 ) associated with inhibition of growth and photosynthesis.
The maximum relative rate of electron transport (rETR max ) (Fig. 2B) shifted from a positive trend to a negative trend at the transition from light-limited to light-saturated growth for both cyanobacteria. This result indicates that the photosynthetic potential increased with increasing growth under light-limited regimes but leveled off and even decreased under light-saturated growth. Notably, the rETR max values measured in Synechococcus 7002 were approximately 3-fold higher than those displayed by Cyanothece 51142. Furthermore, the relative maximal quantum yield of photochemistry (␣ r ) was the highest during light-limited growth and decreased with increasing I i for both cyanobacteria (Fig. 2C). It should be noted that some of the photosynthesis parameters measured for Synechococcus 7002 (e.g., rCEF and ␣ r ) during the steady state corresponding to the highest I i (760 mol photons · m Ϫ2 · s Ϫ1 ) deviated from the predominant trends observed during light-saturated growth (see Fig. S2 in the supplemental material).
Growth tolerance to increasing O 2 tension. To further define the boundaries of cyanobacterial growth robustness and to simu-  electron transport rate (rETR max ); (C) relative maximal quantum yield of photochemistry (␣ r ). Data points specific to Synechococcus 7002 and Cyanothece 51142 are represented by red circles and blue squares, respectively. Linear regression was used to establish positive or negative trends of each parameter during light-limited growth (solid lines), saturated/peak growth (dotted lines), and photoinhibited growth (dashed lines). f.u., fluorescence units; r.u., relative units. late the effects of elevated O 2 tensions that can occur under intense photosynthetic activity, the steady-state turbidostat-grown cultures were subjected to increasing partial O 2 pressure (pO 2 ). While growth of both Synechococcus 7002 ( Fig. 3A) and Cyanothece 51142 (Fig. 3B) was inhibited by O 2 to a greater degree under high irradiance relative to low irradiance, there were stark differences between the sensitivities of the two organisms to elevated pO 2 . Synechococcus 7002 was much more resistant to the increasing pO 2 levels, with only a 28% decline in growth rate at a pO 2 of 0.78 (840 M) and I i of 760 mol photons · m Ϫ2 · s Ϫ1 . In contrast, the growth rates of Cyanothece 51142 decreased severely with increasing O 2 tension under the same irradiance. At a pO 2 of 0.78 and I i of 760 mol photons · m Ϫ2 · s Ϫ1 , growth of Cyanothece 51142 was nearly completely (~98%) inhibited. Interestingly, growth rates for Cyanothece 51142 increased in response to increasing pO 2 values up to 0.1 (107 M), but only at irradiances of Ͻ400 mol photons · m Ϫ2 · s Ϫ1 .
Transcriptional responses reveal potential mechanisms of growth robustness. The Synechococcus 7002 gene-level responses to light-controlled growth were further interrogated by RNA sequencing. Hierarchical clustering of relative transcript abun-dances (reads per kilobase per million reads [RPKM]) across a 2,732-gene data set identified four major clusters, whose eigengenes aligned with the growth and photosynthetic performance of Synechococcus 7002 across the irradiance scale ( Fig. 4; see Table S1 in the supplemental material). Of the four clusters, cluster I contained the largest group of transcripts (29.6%), the relative abundance of which increased in direct proportion with growth rate. Cluster I was functionally enriched (P Ͻ 0.05) for genes involved in translation (i.e., ribosomal proteins and tRNA aminoacylation), purine biosynthesis, iron-sulfur cluster assembly, and ATP synthesis (Table 1; Table S2). In concert with the putative upregulation of growth-related functions, there was also a broad increase in transcripts involved in amino acid biosynthesis, protein folding, and iron transport and acquisition. Cluster I also contained genes involved in central carbon metabolism reactions such as the anaplerotic pathways containing bifunctional fructose-1,6-bisphosphatase-II/sedoheptulose-bisphosphatase regulated by photosystem I (PS I) activity (32). Furthermore, reflecting an increased demand for CO 2 as an electron sink, the relative mRNA levels of genes of the NADH-plastoquinone oxidoreductase complex involved in CO 2 uptake (ndhD3, ndhF3, and  cupS) (33) were elevated in response to increasing irradiance. While not displaying significant functional enrichment within cluster I, the relative transcript abundance of genes involved in peptidoglycan biosynthesis (mreB and mreC) and cell division (minCDE) increased with growth rate. Remarkably, confocal microscopic analysis of Synechococcus 7002 turbidostat cultures revealed that the average cell volume changed proportionally with growth rate (Fig. 5). Specifically, the transition of Synechococcus 7002 from light-limited (I i ϭ 98 mol photons · m Ϫ2 · s Ϫ1 ) to light-saturated growth (I i ϭ 395 mol photons · m Ϫ2 · s Ϫ1 ) resulted in doubling of the average cell volume from 4.84 to 9.69 m 3 .
Gene expression levels in cluster II, which contained 26% of the total transcripts, were inversely correlated with the growth rate ( Fig. 4 and Table 1; see Table S3 in the supplemental material). a The percentage of genes of a given function within a given functional category. b The percentage of a particular functional category within the cluster versus the percentage of genes of that functional category within the genome as a whole. c The P value represents the probability that the number of genes associated with a specific pathway/regulon occurs by chance. This cluster was functionally enriched in genes encoding photosystems I and II antenna proteins and contained most of the critical genes required for photosynthesis. Among other transcripts were those encoding key putative signal transduction and regulatory proteins such as the thioredoxin-responsive regulator of photosynthesis and carbon fixation (pedR), the ferric uptake regulator (fur), and the positive phototaxis signal transduction system (pixL-pixGH), which was downregulated in excess light. Notably, the increase in specific growth rates was associated with a relative decline in transcript abundance of Synechococcus 7002 group 2 and 3 RNA polymerase factors, which are typically required for the expression of stress-induced pathways or niche-adaptive protein functions (34). The eigengene profiles of clusters III and IV exhibited minima and maxima, respectively, at the steady state associated with the transition from light-limited to light-saturated growth (I k of Ϸ164 mol photons · m Ϫ2 · Ϫ1 ). Cluster III (Fig. 4 and Table 1; see Table S4 in the supplemental material), which contained 23% of the expressed genome, included genes associated with photodamage and stress responses, including PS I and PS II stability and repair. Among the genes displaying increased relative mRNA abundance during light-saturated growth were two carboxysome structural genes (ccmK2 and ccmK4), the carbonic anhydrase gene (ccaA), the RuBisCO subunits (rbcSL), and the CO 2 fixation transcriptional regulator (ccmR). The genes encoding the core subunits of the Ndh-1 complex of Synechococcus 7002 (ndhHIJK and ndhA-ndhB-ndhCG) along with the NdhD1 subunit responsible for cyclic electron transfer around PS I also showed irradiancedependent increases under light-saturating conditions (33). Cluster IV, containing 20.2% of the transcriptome, exhibited gene expression trends inverse to cluster III; its eigengene displayed increase under light-limiting conditions and then decreased for each successive steady state within the light-saturated regime (Fig. 4, Table 1, and Table S5). Cluster IV was functionally enriched for polysaccharide and glycolipid biosynthesis, the oxidative branch of the pentose phosphate pathway, nickel-iron hydrogenase biogenesis, and genes involved in electron transport processes such as cyclic electron flow at low light intensities (psaE) (35) and oxidative phosphorylation (coxAB and sucC). While many of these pathways can serve as potential reductant sinks (36), they were downregulated under light-saturating conditions, suggesting the presence of alternative mechanisms of reductant partitioning and oxidative stress mitigation in Synechococcus 7002.

DISCUSSION
This study yields insight into an important biological principle, which allows unicellular cyanobacteria to achieve ultrafast growth by having ultrahigh growth rates and manage cellular resource under different irradiance-controlled growth regimes. Herein, we provide direct evidence that the bimodal transition around the theoretical saturating irradiance (I k ) extends not only to the adjustment of the photosynthesis and growth rates but is also coupled to regulation of specific metabolic reactions and cellular functions. The mechanisms by which Synechococcus 7002 mitigates the negative effects of high irradiance that typically inhibit the growth and photosynthesis processes of unicellular cyanobacteria are likely to have broader implications for understanding the metabolic and regulatory underpinnings of photosynthetic growth (2,3).
Consistent increases in relative transcript abundance with in-creasing irradiance was observed for genes encoding translational machinery, amino acid and nucleotide biosynthesis, the ATPase complex, and the anaplerotic pathways of central carbon metabolism (Fig. 4, cluster I; see Table S2 in the supplemental material). At the same time, the irradiance-driven increase in the relative growth rate of Synechococcus 7002 coincided with the broad decrease in transcripts encoding light acquisition machinery and photosystem I and II reaction centers. These coupled transcriptional responses suggest that the level of resources expended for biomass (i.e., protein synthesis) and energy (ATP) synthesis continue to increase with irradiance, while biomass production and net photosynthesis rates are essentially constant due to increased resource expenditure required. Furthermore, our data provide direct experimental support to earlier calculations (11) positing that energy demand increases to sustain growth, even when growth is light saturated, across increasing irradiance inputs. Reduced net growth-tophotosynthesis yields (Q lim Ͼ Q sat ) during light-replete steady states (Fig. 2) confirm that, while Synechococcus 7002 growth is not inhibited at the high-I i treatments, the ratio of biomass production to energy acquisition decreases once irradiance exceeds an optimum observed near I k . This is most likely due to the decrease in the optical cross section of the photosynthetic apparatus, as Synechococcus 7002 transitions from light-limited to lightsaturated growth (1). This transition has negative effects upon the quantum efficiencies of PS II, and in other oxygenic phototrophs, it is linked to increased photoinhibition. Here, we suggest that the reduction in antenna size also allows the cells to redirect carbon and energy fluxes toward biosynthetic processes which fuel cell division. Notably, this metabolic redistribution in Synechococcus 7002 occurs in conjunction with a tripling in average cell volume (Fig. 5), a physiological phenomenon that is known to also occur in heterotrophic organisms (i.e., Salmonella enterica serotype Typhimurium) in response to increased rate of division (37). These increases in cell mass and size are thought to accommodate the changes in the number of nuclei/cell and can relieve molecular crowding that limits cell growth (11). To that end, removing physical constraints may increase the intracellular capacity needed to accommodate the biosynthetic machinery supporting higher growth rates and alleviate photoinhibition through increased reductant sequestration capacities.
This mechanistic concept is further corroborated by the absence of observed photoinhibition within the Synechococcus 7002 steady states (Fig. 3), as increases in irradiance correlated with the generally decreased mRNA levels of known light-sensitive photosystem reaction centers (psbAD and psaAB). Interestingly, Synechococcus 7002 may be unusual, as the lack of elevated transcription of psb genes under increasing irradiance contrasts with other transcriptional studies of various slower-growing cyanobacteria exposed to differential and/or stress-inducing light regimes (38)(39)(40)(41)(42). The signatures of PS II inhibition were also not observed in the Chl a fluorescence analyses performed on Synechococcus 7002 (data not shown). For example, the relative maximum rate of electron transport (rETR max ) increased through the light-limited regime and showed no significant change during light-saturated steady-state growth (transitional saturated states). Furthermore, the regulation of genes mediating reactive oxygen species (ROS) scavenging does not occur uniformly at the transcriptional level. This is consistent with previous gene expression studies (23,43,44), which suggested that lack of strong concerted upregulation of stress response machinery under high irradiance levels may reflect other mechanisms employed by Synechococcus 7002 for dealing with excess reductant to avoid photoinhibition.
A key observation in support of the above conclusion is based on substantially elevated levels of rCEF displayed by Synechococcus 7002 under saturating high irradiances (Fig. 2), indicating reduction of the plastoquinone (PQ) pool from electron donors not associated with PS II (22). While the exact mechanisms and role of CEF are not fully understood in cyanobacteria, it is postulated that this process contributes to balancing reductant and ATP pools, especially under low irradiance levels (45). Interestingly, psaE and ndhOP genes, encoding the PS I subunit and PQ-oxidoreductase, respectively, which were previously implicated in CEF (35,46), displayed maximum relative expression only during the transitional steady state but were downregulated at saturating irradiances (cluster IV in Table S5 in the supplemental material). In contrast, genes encoding the core subunits of the Ndh-1 complex along with the NdhD1 subunit responsible for cyclic electron transfer around PS I (ndhHIJK and ndhA-ndhB-ndhCG) showed irradiance-dependent increase in abundance under lightsaturating conditions (cluster III in Table S4). Thus, our data implicate new genes (i.e., ndhHIJK and ndhA-ndhB-ndhCG) with potentially important roles in the cyclic electron transport of Synechococcus 7002 that are inherently linked to its ability to effectively partition reductant fluxes and avoid detrimental effects of oxidative stress using multiple strategies.
In summary, a coordinated, functionally grouped gene expression was observed which broadly supports inferences about the role that transcriptional regulation may play in processes such as photosynthesis, carbon fixation, electron transport, and stress response. Interestingly, the kinetics of growth and photosynthesis revealed bimodal growth regimes with respect to carbon uptake to chemical energy production yields, providing evidence of increased cellular energy expenditure during growth at high (but not inhibiting) irradiances. This effect was corroborated by observations that the relative expression of genes involved in biomass synthesis and chemical energy production continued to increase even when the growth and photosynthetic rates were essentially constant, showing strong evidence for an increased energy requirement with increasing light energy input. Only a few other studies have ever demonstrated distinct saturating and/or inhibition kinetics of specific growth rate as a function of irradiance (47)(48)(49)(50). Finally, this study sets a bench mark for specific growth rate ( ϭ 0.2 h Ϫ1 ) achieved by mesophilic cyanobacteria (16,43,51) and provides a foundation for understanding the linkages between photosynthetic performance and growth which can bring fundamental new insights that are broadly applicable to other photosynthetic and nonphotosynthetic biological systems.

MATERIALS AND METHODS
Bacterial cultures and growth. Synechococcus sp. strain PCC 7002 and Cyanothece sp. strain ATCC 51142 were cultured in modified A-plus medium which contained 17 mM NH 4 Cl (52). Starter cultures for the bioreactor were initiated from frozen stocks and were grown as batch cultures in sealed serum bottles charged with growth media supplemented to 15 mM NaHCO 3 under 50 mol photons · m Ϫ2 · s Ϫ1 white-light illumination (tungsten incandescent "soft white"). Controlled cultivation was performed using the New Brunswick Bioflo 310 fermentor platform (Eppendorf, Inc., Enfield, CT) equipped with previously described custombuilt light-emitting diode (LED) photobioreactor (28). Photobioreactors were operated at a 5.5-liter volume with 250 rpm agitation and main-tained at 30°C and pH 7.5 (controlled via 2 M NaOH and HCl additions). Scalar incident irradiance (I i ) ranged from 33 to 759 mol photons · m Ϫ2 · s Ϫ1 and combined 680-and 630-nm narrow-band spectra from LED illuminator chips (Marubeni America Corporation, New York, NY). The reactor was sparged at a constant 4.1 liters min Ϫ1 . In-gas composition was set as a partial nitrogen pressure (pN 2 ) of 0.98 and a pCO 2 of 0.02 during typical growth conditions; however, O 2 stress conditions were tested by modulating pO 2 from 0 to 0.8 and decreasing pN 2 accordingly. Steady-state biomass concentrations were measured directly as ash-free weight (dry weight) (AFDW) (milligram/liter AFDW) (53) and compared in a standard curve to the indirect OD 730 measurements obtained using a Genesys 20 visible spectrophotometer (Thermo Scientific, Rockford, IL). Dissolved O 2 concentrations in the reactor was measured with a Clark type O 2 electrode (InPro 6800 series; Mettler Toledo International Inc., Columbus, OH). Physiological steady state was inferred from continuity (Յ3% variation between measurements) of the following growth readouts: OD 730 , pH, and dissolved O 2 concentration. Samples for all the analyses were taken after at least five residence times at steady-state conditions.
Turbidostat control. For turbidostat operation, constant optical densities (OD 730 ) of 0.082 Ϯ 0.003 (36 Ϯ 1 mg AFDW liter Ϫ1 ) were maintained for both cultures across the entire range of irradiances to effectively eliminate self-shading. Incident and transmitted irradiance was measured with six opposing 2 quantum sensors (LI-210SA photometric sensor; LI-COR Biosciences, Lincoln, NE) and intercalibrated with a 4 submerged quantum sensor (LI-193SA spherical underwater quantum sensor; LI-COR Biosciences). Hence, scalar incident irradiance (I i ) is reported here as quanta incident to the center of the reactor and has been confirmed to be both axially and radially isotropic within the liquid culture volume. In situ photosynthesis-to-irradiance curves (P-I i curves) were generated by temporarily stopping flow and subsequently increasing the I i output of the LEDs over 5-min intervals while concurrently data logging the dissolved O 2 (DO) response within the culture volume (see photosynthesis calculations below).
Microscopic analysis. Microscopic images were acquired on a Zeiss LSM 710 scanning confocal laser microscope (Carl Zeiss MicroImaging GmbH, Jena, Germany) equipped with a W Plan-Apochromatic 63ϫ/1.0 M27 objective. The Synechococcus 7002 cells were visualized by phycocyanin autofluorescence measured at 640 nm. Images were processed with Volocity (PerkinElmer, Waltham, MA) and used to obtain the cell size measurements made along the major and minor axis. Cell volumes were calculated using the equation for an ellipsoid, V ϭ 4 3 a 2 b, where a is the diameter of the minor axis and b is the diameter of the major axis. RNA isolation and sequencing. Cells from a steady-state turbidostat cultures were collected via centrifugation at 7,000 rpm for 5 min at 4°C, flash frozen in liquid nitrogen, and stored at Ϫ80°C. Total RNA extraction was performed using standard methodology (54). The quality and integrity of the RNA was assessed on an Agilent 2100 bioanalyzer, and only samples with integrity numbers between 8 and 10 were selected. Template cDNA was prepared using the Applied Biosystems SOLiD total RNA-Seq (transcriptome sequencing) kit (Life Technologies, Carlsbad, CA) according to the manufacturer's protocol. Sequencing was carried out using the SOLiD 5500XL protocol (Life Technologies). The 50-base sequence reads were mapped to the genomes of Synechococcus 7002 (GenBank accession number NC_010475), and gene expression levels were determined using the Rockhopper software package as previously described (55). Genes that showed expression values in the bottom 20% of genes at each I i level for which mRNA was collected were removed from analysis. Hierarchical clustering was performed in the MatLab Bioinformatics Toolbox (Math-Works, Inc.) to identify six distinct profiles of genes with similar expression profiles across irradiance-controlled steady states. Gene expression values (reads per kilobase per million reads [RPKM]) were transformed to log base 2 values and standardized (mean ϭ 0 and standard deviation ϭ 1). Genes in each of the six clusters (profiles) chosen for further analysis were examined to determine whether certain functions were enriched in a given module. Enrichment is defined as the percentage of genes within the profile for which a function has been assigned being significantly higher than the percentage of genes of the same function in the entire genome with a P value of Ͻ0.05 according to Fisher exact test.
Photosynthetic performance. Chl a fluorescence measurements were performed on samples obtained from steady-state turbidostat cultures using pulse amplitude-modulated fluorometry (PAM) in a DUAL-PAM-100 instrument equipped with a photodiode detector and RG665 filter (Walz GmbH, Effeltrich, Germany). Samples were dark adapted for 1 min prior to analyses. Red measuring light (620 nm) was pulsed at the lowest power at 1,000 Hz in the dark and at 10,000 Hz during actinic illumination at 98 mol photons · m Ϫ2 · s Ϫ1 with 635-nm light. Transient fluorescent changes were measured after fluorescence induction through the following: (i) a 200-ms saturating pulse (2,000 mol photons · m Ϫ2 · s Ϫ 1 ), (ii) 5 s of only far-red light (730 nm), (iii) another 15 s of actinic light, and (iv) 30 s of darkness. The rate of change (rise) of postillumination fluorescence occurring in the dark during step iv results from reduction of plastoquinone (PQ) from NAD(P)H or other reductant accumulated during illumination and was interpreted as a proxy for the rate of cyclic electron transport (rCEF) around PS I (22,56). Calculations for determining the relative electron transport rates (rETR ϭ PAR ϫ ⌬F/F m =) have been previously described (57,58). Rapid light curves were generated by evaluating rETR as a function of increasing PAR values (1-min step intervals). The maximum values and initial slopes (rETR max and ␣ r , respectively) were determined and interpreted as the relative photosynthetic capacity and relative maximal photosynthetic quantum yield of photochemistry, respectively (58).
Calculations and regression. The net rate of O 2 production was calculated from the steady-state mass balance through the bioreactor control volume (equation 1).
The specific rate of O 2 production (q O2 ) multiplied by the biomass concentration (x) is interpreted here as the net rate of O 2 production during photosynthesis (22) and is a function of the dilution rate (D), the lumped mass transfer coefficient (k l a) (0.83 min Ϫ1 ), dissolved O 2 concentration ([O 2 ]), and Henry's law partitioning coefficient (k H ϭ 1.08 mM atm Ϫ1 ). The specific rate of biomass production (q x ; Cmmol biomass hour Ϫ1 · gram AFDW Ϫ1 ) was calculated by assuming the molecular weight of dry biomass to be 24.59 g AFDW · C mol Ϫ1 (59). Parameterization and curve fitting were performed via nonlinear least-squares regression using the fit function in MatLab with a weighted fit option. All saturating and photoinhibition relationships of growth and photosynthesis were fit with equation 2, where the inhibition index (␤) was simplified from the form originally reported by Platt et al. (60) to more clearly represent the theoretical irradiance threshold of inhibition; parameters I i and I k are the scalar incident and theoretical saturating irradiances, respectively.