Multi-Omic Dynamics Associate Oxygenic Photosynthesis with Nitrogenase-Mediated H2 Production in Cyanothece sp. ATCC 51142

To date, the proposed mechanisms of nitrogenase-driven photosynthetic H2 production by the diazotrophic unicellular cyanobacterium Cyanothece sp. ATCC 51142 have assumed that reductant and ATP requirements are derived solely from glycogen oxidation and cyclic-electron flow around photosystem I. Through genome-scale transcript and protein profiling, this study presents and tests a new hypothesis on the metabolic relationship between oxygenic photosynthesis and nitrogenase-mediated H2 production in Cyanothece 51142. Our results show that net-positive rates of oxygenic photosynthesis and increased expression of photosystem II reaction centers correspond and are synchronized with nitrogenase expression and H2 production. These findings provide a new and more complete view on the metabolic processes contributing to the energy budget of photosynthetic H2 production and highlight the role of concurrent photocatalytic H2O oxidation as a participating process.

. Chl a and ash-free dry weight biomass measurements over the H 2 production profile. Time zero indicates onset of nitrogen-depletion and absence of media addition (dilution rate = 0). Times prior to t = 0 represent measurements taken during the ammonia limited chemostat controlled steady-state.

Measurements of chlorophyll fluorescence.
Variable Chl a fluorescence and fluorescence kinetic traces were measured by a previously reported methods 3 . Briefly, cells were sampled from the photobioreactor and dark-adapted for 5 min prior to taking measurements via pulse amplitude-modulated fluorometry (PAM) in a DUAL-PAM-100 equipped with a photodiode detector and RG665 filter (Walz GmbH, Effeltrich, Germany). Variable Chl a fluorescence parameters (rETR max and Fv/Fm) were measured and interpreted via standard approaches 4 . The slope of the post-illumination fluorescence rises (df/dt) are proxy measurements for the rate of cyclic electron transport (rCEF) 3,5 and were only observed (qualitatively) from the ammoniumlimited chemostat, precondition (Fig. S2). Figure S2. Chl a fluorescence traces measured from cells taken at different time points across the H 2 production transient. The post illumination rise in fluorescence is a result from reduction of plastoquinone from NAD(P)H or other reductants accumulated during illumination.

Mass balance of biological H 2 and O 2 production.
The specific rates of net H 2 production and O 2 production were calculated from non-steady state mass balances represented in the following equation progression (Equations S3-S4). Simplified versions for the following equations were used for the steady-state reaction rates; q H2 and q O2 . Changes to the control volume, due to sampling, were considered negligible. The solute (H 2 or O 2 ) is represented by the symbol x; the k l a values were experimentally measured to be 55 and 74 hr -1 for O 2 and H 2 gas, respectively.
The following assumptions and simplifications were made based on the reaction geometry to yield Equation S4: (i) changes to the control volume, due to sampling, were considered negligible; (ii) and the dilution rate (D) was zero; (iii) the variable x in gas (gas phase concentration) was zero; for both H 2 and O 2 .
A finite difference approximation was substituted into Equation S3 and both sides of the equation were normalized by the corresponding biomass (g-CDW) to yield the final working form of an equation for the specific rates production; q H2 or q O2 (Equation S4). The parameter h is defined as the time difference between sample steps normalized by the total number of samples.

Filtering of mRNA and protein expression data
All filtering and clustering was performed by custom Matlab (Mathworks) scripts that are available on request.

Standard mRNA expression filtering algorithm
Messenger RNA expression profiles were filtered with the custom Matlab script mRNA_preproc_1 which implements the following steps (in order): 1) genes with profile expression variances in the bottom 10% were masked, 2) genes with absolute expression profiles within the bottom 20% (RPKM values) were masked, 3) RPKM values equal to zero were assigned an arbitrarily low value of 0.5, 4) expression profiles with the 20% highest level entropy 6 were masked, 5) each expression value was normalized to its corresponding steadystate condition and log2 transformed.

Standard protein expression filtering algorithm
Protein expression profiles were filtered with the custom Matlab script mRNA_preproc_3 which implements the following steps (in order): 1) proteins with profile expression variances in the bottom 10% were masked, 2) protein expression profiles with the 20% highest level entropy 6 were masked, 5) each expression value was normalized to its corresponding steady-state condition and log2 transformed.

Synchronized profiles of protein and mRNA expression
These profiles were identified and calculated with the custom Matlab script systiter_V2 which implements the following steps (in order): 1) protein expression profiles were filtered (see 2.1.2.) and organized by K-means into six groups, 2) each cluster was assigned a numeric cluster ID (1-6), 3) the mRNA expression files and gene IDs were collected from each protein cluster, 4) the mRNA expression profiles exhibiting variance across the profile in the bottom 10% were masked, and the corresponding gene IDs were logged, 4) protein expression profiles corresponding to genes that were masked in from the gene variance filter were also masked. This algorithm output six clusters of protein and mRNA expression profiles containing identical coding genes.