The relative amount of these localizations varied between experiments. As a control of free

cytoplasmic GFP, a culture that has been previously reported to produce GFP in all cells of N. punctiforme was used (Fig. 3b) (Huang et al., 2010). This GFP control culture produced homogeneously distributed GFP in the cytoplasm of the heterocysts. In the cytoplasm of vegetative cells, the fluorescence was clearly obstructed by the presence of thylakoid membranes. The presence of thylakoids is seen in the red autofluorescence (magenta) stemming from the thylakoid-attached phycobilisome/photosystem II complexes (Cardona et al., 2009) and in the limited overlap between autofluorescence and GFP fluorescence (Fig. 3b). The full-length HupS–GFP protein required strong denaturing LY294002 mw conditions (2% SDS) for efficient

extraction (Fig. 1b), whereas most of the HupS–GFP degradation products could be extracted without detergent (Fig. S1). To examine the potential formation of a complete uptake hydrogenase by HupS–GFP and HupL, efforts were made to prepare native Buparlisib concentration extracts of these proteins from N2-fixing cultures of SHG. However, none of these attempts were successful. To examine the solubility of HupS–GFP, anti-GFP Western blots were performed with proteins extracted using buffers containing no detergents, mild nonionic detergents, or strongly denaturing additives. The results show that HupS–GFP could only be efficiently extracted using the strongly denaturing additives (Fig. 1b and Fig. S1). To identify any cell structure differences caused by potential HupS–GFP protein inclusions, isolated heterocysts

from N2-fixing Tolmetin cultures of SHG and WT were compared using TEM. The resulting images did not reveal any structural differences between SHG and WT heterocysts (Fig. S2). This study shows that the small subunit of the uptake hydrogenase, HupS, in N. punctiforme is solely localized to the heterocysts. The localization of the uptake hydrogenase in N. punctiforme has been under debate for a long time since previous immunolocalization studies have identified the large subunit, HupL, in both vegetative cells and heterocysts (Lindblad & Sellstedt, 1990; Tamagnini et al., 2002, 2007; Seabra et al., 2009). Interestingly, these studies used several different HupL antibodies and the results are not fully correlating. Both Seabra et al. (2009) and Lindblad & Sellstedt (1990) show vegetative cell localization of HupL, but the subcellular localization to the cytoplasmic membranes between vegetative cells clearly seen in Lindblad & Sellstedt (1990) is missing in Seabra et al. (2009), which argued for a subcellular localization of an inactive form to the vegetative cell thylakoid membranes as well as to what is described as the vesicular region of the heterocysts (Seabra et al., 2009).