2.1 Plant material and growth condition
Seeds of Nicotiana tabacum (L.) were germinated and cultivated in
a substrate composed by a mixture of vermiculite, sand and soil
(2:1:0.5) for 30 days. Plants were kept well-watered and nourished with
Hoagland nutrient solution every week (Hoagland & Arnon 1950) under
non-controlled greenhouse conditions with natural 12 h of photoperiod,
ambient temperature 30 ± 4 °C, relative humidity 62 ± 10% and
photosynthetic photon flux density (PPFD) which reached a maximum value
of 500 µmol photons m-2 s-1.
2.2 Guard cell isolation
A pool of guard cell-enriched epidermal fragments (simply referred here
as guard cells) was isolated following a protocol that was optimized for
metabolite profiling analysis (Daloso et al. 2015a). Guard cells
were isolated at pre-dawn by blending approximately three expanded
leaves per replicate in a Warring blender (Philips, RI 2044 B.V.
International Philips, Amsterdam, The Netherlands), that contains an
internal filter to remove excess mesophyll cells, fibres and other
cellular debris. The viability of the guard cells was analysed by
staining with fluorescein diacetate and propidium iodide dyes, as
described earlier (Huang et al. 1986). This analysis demonstrated
that only the guard cells are alive in the epidermal fragments (Dalosoet al. 2015a; Antunes et al. 2017). All guard cell
isolations were carried out in the dark in order to maintain closed
stomata and simulate opening upon illumination, following the natural
stomatal circadian rhythm (Daloso et al. 2016b; Antunes et
al. 2017).
2.3 13C-isotope labelling
experiment
After isolation, guard cells were transferred to light or dark
conditions and incubated in a solution containing 50 μM
CaCl2 and 5 mM MES-Tris, pH 6.15 in the presence of 5 mM13C-NaHCO3. The experiment was
initiated at the beginning of the light period of the day. Guard cell
samples were rapidly harvested on a nylon membrane (220 µm) and
snap-frozen in liquid nitrogen after 0, 10, 20 and 60 minutes under
light or dark conditions.
2.4 Metabolomics analysis
Approximately 30 mg of guard cells were disrupted and transformed into a
powder by maceration using mortar, pestle and liquid nitrogen. The
powder was then used for metabolite extraction. The extraction and
derivatization of polar metabolites were carried out as described
previously (Lisec et al., 2006), except that 1 ml of the polar
(upper) phase, instead of 150 µl, was collected and reduced to dryness.
Metabolites were derivatized by methoxyamination, with subsequent
addition of tri-methyl-silyl (TMS) and finally analysed by gas
chromatography electron impact – time of flight – mass spectrometry
(GC-EI-TOF-MS), as described previously (Lisec et al., 2006). The
mass spectral analysis were performed using the software Xcalibur® 2.1
(Thermo Fisher Scientific, Waltham, MA, USA), as described earlier (Limaet al. 2018). The metabolites were identified using the Golm
Metabolome Database (http://gmd.mpimp-golm.mpg.de/) (Kopkaet al. 2005). The metabolite content is expressed as relative to
ribitol (added to each biological replicate during the extraction) on a
fresh weight (FW) basis. The data is reported following recommendations
for metabolomics analysis (Supplemental Data S1) (Fernie et al.2011; Alseekh et al. 2021).
2.5 13C-enrichment
analysis
The relative abundance of the isotopologues (RIA) (M, M1, M2…Mn)
and the fractional 13C-enrichment
(F13C) was calculated as described previously (Limaet al. 2018). For instance, in a fragment of three carbons, the
sum of the intensity of the isotopologues M, M1, M2 and M3 was set to
100%, and the intensity of each isotopologue is then relative to this
sum. In this hypothetical fragment, the F13C is
calculated according to the following equation: F13C =
((M1*1)+(M2*2)+(M3*3))/3. The relative 13C-enrichment
(R13C) was obtained by normalizing the
F13C by the time 0 of the experiment.
2.6 Metabolic network analysis
Correlation-based metabolic networks were created using metabolite
profiling data, in which the nodes correspond to the metabolites and the
links to the strength of correlation between a pair of metabolites. The
correlation was calculated using debiased sparse partial correlation
(DSPC) analysis using the Java-based CorrelationCalculator software
(Basu et al. 2017). The networks were designed by restricting the
strength of the connections to a specific limit of DSPC coefficient
(r ) (-0.5 > r > 0.5) using
Metscape on CYTOSCAPE v.3.7.2 software (Shannon et al. 2003;
Karnovsky et al. 2012). Network-derived parameters such as
clustering coefficient, network heterogeneity, network density and
network centralization were obtained as described previously (Assenovet al. 2008). We further determined the preferential attachment
and the appearance of new hubs, which highlight the nodes that are
considered hubs before the start of the experiment (time 0) and maintain
its degree of connection and demonstrate nodes that become highly
connected after the beginning of the treatments, respectively (Freireet al. 2021).