Bidirectional C and N transfer and a potential role for sulfur in an epiphytic diazotrophic mutualism

In nitrogen-limited boreal forests, associations between feathermoss and diazotrophic cyanobacteria control nitrogen inputs and thus carbon cycling, but little is known about the molecular regulators required for initiation and maintenance of these associations. Specifically, a benefit to the cyanobacteria is not known, challenging whether the association is a nutritional mutualism. Targeted mutagenesis of the cyanobacterial alkane sulfonate monooxygenase results in an inability to colonize feathermosses by the cyanobacterium Nostoc punctiforme, suggesting a role for organic sulfur in communication or nutrition. Isotope probing paired with high-resolution imaging mass spectrometry (NanoSIMS) demonstrated bidirectional elemental transfer between partners, with carbon and sulfur both being transferred to the cyanobacteria, and nitrogen transferred to the moss. These results support the hypothesis that moss and cyanobacteria enter a mutualistic exosymbiosis with substantial bidirectional material exchange of carbon and nitrogen and potential signaling through sulfur compounds.

with on different levels of 33 S enrichment. The cyanobacterial cells were then analyzed for 33 S enrichment (here, f33S) by spot analysis (blue points). The linear regression through the data (dotted line) was used to estimate the contribution of surrounding phyllid to the cyanobacterial spot analysis (~2.4%), which is the basis for our mixing model for the spot analyses: 0.024*f33Smoss-meas + (1 -0.024)*f33S-cyano-est = f33S-cyano-meas. We used the measured 33 S fractions for the moss and cyanobacterial (f33S-moss-meas and f33S-cyano-meas, respectively) and solved this mixing model for the estimate the true 33S fraction in each cyanobacterial cell (f33S-cyano-est ).

Acetylene reduction assay
Nitrogen fixation rates were measured on triplicate 1 mL samples from 2 week old cultures placed in 10 mL chromatography vials equipped with rubber septa. 10% of the volume was replaced with acetylene and incubated for 4 hours, with above growth conditions, and ethylene production measured on Shimadzu GC-8A gas chromatograph (Shimadzu Corporation, Japan).
Triplicate negative controls with BG110 media blanks were also run. Another set of triplicate 1 mL samples from the same cultures were pelleted and chlorophyll a extracted in cold absolute methanol and OD measured (Nagarkar and Williams 1997) using an Ultrospec 3000 spectrophotometer (Pharmacia Biotech, Cambridge, England).

Mutant construction
A plasmid to replace the putative alkanesulfonate monooxygenase (AMSO, Npun_F5882) was constructed as follows: Regions of the N. punctiforme chromosome 1-kb upstream and 1-kb downstream of approximately the midpoint of the targeted gene were amplified by PCR (Takara Bio, Mountain View, CA, USA) using primers specified in Supplementary Table S1. The neomycin resistance gene cassette was amplified using primers NpunKO-1 and NpunKO-2 and plasmid pRL448 as a template (Elhai and Wolk 1988). The 3' end of the 1-kb upstream PCR product and the 5' end of the 1-kb downstream PCR product had sequence homology to the neomycin resistance cassette to result in a final assembly product with the neomycin resistance flanked by the two regions of the gene to be interrupted. The vector, plasmid pRL2948a (C. P. Wolk, unpublished), provided an origin of transfer (oriT) for RP4-based conjugation, and a counter selectable sacB-erythromycin resistance gene (EmR). It was amplified using primers NpunKO-4 and NpunKO-3.2. Each product was purified by PCR cleanup kit (QIAquick, Qiagen). Each of the plasmids to knockout an N. punctiforme gene was assembled from four pieces (1-kb upstream, neomycin resistance gene, 1-kb downstream, and vector) using Gibson Assembly (

NanoSIMS analysis
All samples were stored in an argon dry box. Samples were mapped first with epifluorescence microscopy to identify Nostoc M2 cells, and then coated with ~5 nm of gold and imaged with a FEI Inspect F scanning electron microscope (Hillsboro, OR) to identify areas of interest for NanoSIMS imaging. A focused 2pA 150nm diameter 16 keV 133 Cs + primary ion beam was scanned in a raster pattern (for images: 225-625 µm 2 analyses areas, 256-by-256 pixels with a dwell time of 1 ms/pixel for 10-30 cycles; for spot analyses: 1 µm 2 areas, 32-by-32 pixels with a dwell time of 3 ms/pixel for 10 cycles). To quantify 33 S/ 32 S, 15 N/ 14 N, and 13 C/ 12 C at each location, [ 12 C 14 N -, 12 C 15 N -, 32 S -, 33 S -] and [ 12 C 14 N -, 13 C 14 N -, 32 S -, 33 S -] were alternately collected using electrostatic peak jumping ("combined analysis"). Secondary electrons were also simultaneously collected. For imaging analysis, samples were first sputtered with 90 pA of Cs + current to an approximate depth of 50 nm to reach sputtering equilibrium (Ghosal et al 2008).
Samples without any introduced stable isotope label (unlabeled control) were run to ensure no sources of extraneous label was introduced through samples preparation and analysis, to make sure samples of that material ran properly, and for calculations of atom percent excess (APE).
The data are presented as atom percent excess (APE) (Pett-Ridge and Weber 2012, Popa et al where ! and " are the final and initial fractions of the spiked isotope in the sample. For N and C, which are two isotope systems, these abundances are calculated using = ( + 1) ⁄ , where R is the measured ratio. To accurately calculate 33 S APE from the measured ratio 33 S/ 32 S, the abundance of 34 S must be included: where $ !+ = 0.042 and $ !" = 0.95. This formulation is based on the relative abundance of 34 S to 32 S being fixed, 34 S being negligible in the Na 33 SO4, and 36 S being negligible in all cases.