All living plant tissues consume oxygen and release carbon dioxide (CO2) within the process of mitochondrial respiration, which has, therefore, a fundamental role that ranges from cellular bioenergetics to the functioning of whole ecosystems. Because the CO2 geochemical cycle is in dynamic equilibrium and has dramatic effects on the evolution of the Earth's climate, global plant respiration is often viewed first as a major CO2 source. Importantly, for some plant tissues, a great concern is access to oxygen, which at first glance seems paradoxical owing to the natural abundance of molecular oxygen in the atmosphere. Oxygen availability should not be a problem for leaves which are aerial organs and which have a large surface : volume ratio and considerable intercellular space to facilitate CO2 photosynthetic assimilation. Moreover, chloroplasts produce large amounts of oxygen during photosynthesis. However, in contrast to the situation in leaves, roots and other underground organs are likely to be exposed to hypoxia depending on oxygen availability in soil, which can be dramatically reduced during flooding. Such stress situations lead to diverse metabolic adaptations, including fermentation, and are accompanied by substantial modifications in gene expression (van Dongen et al., 2009). Interestingly, hypoxia can occur within organs that are naturally exposed to normoxic conditions, but which share compact tissues with high respiratory activity: a situation that generates hypoxia because diffusion of oxygen cannot match its mitochondrial consumption. Depending on their morphology, many seeds, fruits and storage organs can face such internal hypoxic situations. For instance, in imbibed pea seeds, the steady-state oxygen concentration within cotyledon tissues was found to range from 0 to 3 μM until germination (i.e. for > 25 h; Benamar et al., 2008). Neither apples, potatoes nor germinating pea seeds placed in normoxic conditions show signs of fermentation, which suggests mitochondria metabolism is adapted to such conditions, with potential roles for nitric oxide, nitrite and nonsymbiotic leghemoglobins in the case of seeds (Borisjuk et al., 2007; Thiel et al., 2011). A striking question is how oxygen actually diffuses within the rather compact tissues to supply internal cells and their mitochondria. Although large air spaces are typically present in the cortex of roots or pome fruits (Verboven et al., 2008), seeds are evolutionarily driven toward a small size to facilitate desiccation, storage, and dispersal, and their tissues are, as a result, rather compact. Seeds thus combine dense tissues with high metabolic activities; this combination of metabolic demand and tissue structure represents a real challenge for respiration oxygen supply.
In this issue of New Phytologist, Verboven et al. (pp. 936–947) have used state-of-the-art image analysis to unveil the architecture of the void space network in developing rape seeds. Then, by combining gas diffusion modeling and metabolic analysis they provide strong evidence for a major role of the network in channeling the precious oxygen resource within the tissues. The occurrence of void spaces in seed tissues (as well as in other organs) was noted long ago by histologists, identified simply as empty spaces separating cell walls on the corner of cells. To quote Smith & Flinn (1967) writing about the storage parenchyma of pea seed cotyledons: ‘The cells of both tissues are compactly but irregularly arranged, with small triangular intercellular spaces confined to the corners of the cells’. The architecture of a seed void space was first unveiled by Cloetens et al. (2006) who examined the structure of Arabidopsis dry seeds using synchrotron X-ray scanning computer tomography. They revealed the complex architecture of the void space, which appeared filled with air in the dry seed, which led the authors to posit that void space could represent an oxygen reserve for seed respiration during imbibition or facilitate water diffusion.
‘The small empty corners between cells that were noted by cell biologists in the 1960s are now back on the scene …’
On the functional side, knowledge about oxygen status and energy metabolism in developing seeds has greatly increased in the last decade thanks to the development of oxygen sensors, transcriptomics and metabolic analysis (Borisjuk & Rolletschek, 2009; Thiel et al., 2011). It became clear that a hypoxic state was the rule in embryonic tissues, and that seed photosynthesis had a primordial role in increasing oxygen concentration in the light (Rolletschek et al., 2005). It was thus logical to investigate whether a void space network was also present during seed development and could facilitate oxygen diffusion in the hypoxic seed tissues. Verboven et al. submitted oilseed rape seeds at mid-storage stage to high resolution micro X-ray computed tomography, using a benchtop system to reach an image resolution of 1.7 μm per pixel which was further increased by half using synchrotron radiation. This nondestructive image analysis approach enables the construction of three-dimensional (3D) images based on the electron density of materials. It revealed the presence of void spaces in most tissues, which were especially prominent and connected in the testa and hypocotyl. In hypocotyls, the highly connected void space network in the cortex appeared highly connected in the axial direction (Fig. 1). Of great importance is the fact that the void space in these well hydrated seeds appeared not to be filled with water, but with air, immediately suggesting it could have a role in oxygen storage and transport. Indeed, although oxygen is fairly soluble in water, there is still > 30 times more oxygen in air than in air-saturated water, but more significantly, oxygen diffuses 10 000-fold faster in air than in solution. This means that connected void space is likely to offer a highway for oxygen diffusion in tissues. The authors then combined a sophisticated mathematical modeling of an oxygen reaction diffusion model (Ho et al., 2011) with the high-resolution seed structure data to model the porosity and oxygen diffusivity of all compartments. This allowed, for the first time, publication of an estimation of the distribution of gaseous and dissolved oxygen pools in a whole developing seed and in its main organs and tissues. Taking respiratory data into account, simple calculation then revealed that the void space oxygen content could only sustain 1 min of seed respiration, a turnover that would be extended by 20 s if taking into account the dissolved oxygen. Hence, the mid-development stage oilseed rape seed needs to replace its oxygen fuel almost every minute! The reaction diffusion model was validated experimentally by establishing oxygen profiles using needle-type and planar oxygen sensors. The axial distribution of the void network in hypocotyl cortex indicated a major channeling of oxygen toward the radicle, which is immersed in the hypoxic liquid endosperm. In an endeavor to relate oxygen dynamics to storage metabolism, the authors used magnetic resonance imaging, another noninvasive technique, to examine the distribution of water and lipids. At the mid-growth stage, oil was predominantly accumulating in the hypocotyl cortex and peripheral parts of the outer cotyledon. Further analysis of the hypocotyl using histochemistry and metabolic analysis of laser microdissected tissues highlighted heterogeneity between the cortex where lipids and proteins accumulate, and the central stele where storage is low. This can be related to the void space axial architecture in the cortex that allows funneling of oxygen which is then consumed by the metabolically active cells. Since radial diffusion of oxygen to the stele is limiting because of a lack of void space and due to local consumption, little reserve accumulation occurs in the hypoxia-prone tissue. Interestingly, void space is not developed enough in cotyledons to ensure such a distribution of oxygen. This implies that in cotyledons, which are the major site of reserve accumulation, energy metabolism is likely to operate under quite hypoxic conditions, at least at night. It is noteworthy that according to this analysis at mid-maturation of oilseed rape seeds, accumulation of reserves proceeds first in the hypocotyl and later in the cotyledons. Therefore one can expect at later stages a decrease in metabolic activity (and oxygen consumption) in the reserve-filled hypocotyl, thus improving the oxygen balance in the cotyledons.
One of the aphorisms from the 1943 book entitled ‘Voces’ by the illustrious Argentinian poet Antonio Porchia (1885–1968) reads: ‘Percibimos el vacío, llenándolo’, usually translated as ‘We become aware of the void as we fill it’ which is a beautiful fit with this research. The authors indeed provide a convincing demonstration that void space in seeds is of tremendous importance for energy metabolism and reserve accumulation in seeds. The small empty corners between cells that were noted by cell biologists in the 1960s are thus now back on the scene, and new and intriguing questions follow. Because of its indirect role in sustaining energy metabolism and hence reserve deposition, seed void space is a potential target for crop improvement. Whether it also plays a major role in oxygen balance of the germinating seed is not known, so far. The big question behind these issues is what determines the importance and architecture of the void space, which strongly differ between organs and tissues. While one cannot exclude the possibility of a direct genetic control of void space architecture, it is much more likely that the formation and connection of void space is driven by cell shape and expansion within tissues, itself under genetic and environmental control. The architecture of void space has certainly a crucial importance in natural evolution and agronomical selection at the seed level, and further exploration with genetic approaches should shed light on its genetic determinants.