Aerenchyma formation


  • David E. Evans

    Corresponding author
    1. Research School of Biological and Molecular Sciences, Oxford Brookes University, Oxford, UK
      Author for correspondence: David E. Evans Tel: +44 1865483968 Fax: +44 1865483242 Email:
    Search for more papers by this author

Author for correspondence: David E. Evans Tel: +44 1865483968 Fax: +44 1865483242 Email:



  • Introduction 
  • II Schizogenous aerenchyma formation 
  • III Lysigenous aerenchyma formation 
  • IV Regulators of lysigenous aerenchyma formation
  • Key questions in lysigenous aerenchyma formation 
  • VI Sensing hypoxia; early events in aerenchyma formation 
  • VII Ultrastructural changes associated with lysigenous aerenchyma formation 
  • VIII Late events in cell death in aerenchyma formation 
  • IX Comparative evidence on programmed cell death in aerenchyma formation 
  • Comparison with other abiotic initiators of cell death in plants 
  • XI Acknowledgements 
  • XII References 


Aerenchyma – tissue containing enlarged gas spaces – occurs in many plants. It is formed either as part of normal development, or in response to stress (e.g. hypoxia). Two mechanisms of aerenchyma formation have been described; schizogeny, in which development results in the cell separation and lysigeny, in which cells die to create the gas space. While schizogenous aerenchyma provides a fascinating system for study and has been described in detail at a morphological and ultrastructural level, little is known about the molecular genetics of its formation. The ultrastructure and morphology of lysigenous aerenchyma has also been researched in detail, and considerable progress has been made in describing the cell death processes involved, particularly in relation to programmed cell death. Once again, the molecular genetics of the process are not well understood. Aerenchyma is of great importance in crop survival in waterlogging. It is also important in being a major pathway for the release of the global warming gas methane to the atmosphere in flooded soils. Understanding the regulation of its development is therefore a research priority.

I. Introduction

Aerenchyma is the term given to plant tissues containing enlarged gas spaces exceeding those commonly found as intracellular spaces. It is formed in the roots and shoots of wetland species and in some dryland species in adverse conditions, either constitutively or because of abiotic stress. While usually associated with hypoxia resulting from waterlogging, it may also be caused by other forms of stress, including high temperature, drought and nutrient deficiency. Aerenchyma occurs as one of two basic types, first identified by Sachs more than 120 years ago (Sachs, 1882) and named lysigenous and schizogenous (Arber, 1920, originally ‘lysogenetic and schizogenetic’, De Bary, 1884). Figure 1a summarizes the differences between the formation of lysigenous and schizogenous aerenchyma. Lysigenous aerenchyma formation involves cell death. Cells formed during earlier development, die and are ‘removed’, leaving a gas space. Lysigenous aerenchyma is found in many important crop species, including barley (Arikado & Adachi, 1955), wheat (Trought & Drew, 1980), rice (Justin & Armstrong, 1991), and maize (He et al., 1996a; Gunawardena et al., 2001a). Schizogeneous aerenchyma is common in wetland species like Rumex and is formed by cell separation, without the cells involved dying. Some species (like Saggitaria lancifolia; Schussler et al., 1997) form both schizogenous and lysigenous aerenchyma, though in different tissues.

Figure 1.

Processes in aerenchyma formation. The general processes involved in lysigenous and schizogenous aerenchyma formation are represented in (a) while the pathway for the initiation of induced aerenchyma, or for the stimulation of the production of constitutive aerenchyma is represented in (b).

In many species, root aerenchyma is a component of a pathway of gas conducting tissue extending to the shoot. Gas transport in the pathway may be simple diffusion or may be due to pressure flow (see Raven, 1996 and Jackson & Armstrong, 1999 for reviews). Formation of aerenchyma decreases oxygen demand effected by the removal of some cells of the cortex, but its importance extends beyond this (Armstrong, 1979). It delivers oxygen to the root tip and to the rhizosphere and removes gases (carbon dioxide, ethylene, methane) from the root and soil (Shannon et al., 1996; Colmer, 2003). Transport of oxygen to the root tip is enhanced in wetland species by the formation of a barrier to its loss at the surface of the basal end of the root (Armstrong et al., 2000; Visser et al., 2000; Soukoup et al., 2002). This barrier may be inducible and occurs at the hypodermis/exodermis boundary (Colmer, 2003). Oxygenation of the rhizosphere around the growing tip both reduces the harmful effects of anoxic soils on roots and supplies the demand of soil organisms that would compete with the root tip for oxygen. The oxygen supplied may come from photosynthesis or from the atmosphere.

Removal of gases from the soil and root are also very significant functions of aerenchyma. Carbon dioxide from the root tip and from soil microorganisms is vented to atmosphere through the aerenchyma, as is ethylene. Possibly most significant in global terms, however, is the fact that aerenchyma provides a low resistance pathway for the venting of methane from the rhizosphere to the atmosphere. Methane is an important global warming agent, believed to contribute to about 20% of the effect. Its concentration in the atmosphere has doubled over the last two centuries and flooded rice paddies represent a major source. In a study comparing two rice cultivars, Buterbach-Bahl et al. (1997) estimated that more than 80% of methane emitted from waterlogged soils reaches the atmosphere via aerenchyma. Perhaps more importantly from a research perspective, they observed that the two cultivars differed significantly in methane loss to the atmosphere and that this could be correlated to the nature and structure of the aerenchyma formed.

Both schizogenous and lysigenous aerenchyma are fascinating developmental systems. While both produce the same end result – a tissue filled with gas spaces – and schizogeny appears to be a more ordered process, both are under developmental control. While the process of schizogenous aerenchyma formation has been described in depth in a number of species at a histological level, large gaps remain in our understanding of its regulation as a developmental process. More is known about lysigenous aerenchyma at the cellular level, but once again many questions remain about its developmental regulation.

II. Schizogenous aerenchyma formation

Schizogenous aerenchyma occurs when intercellular gas spaces form during tissue development without cell death taking place. Spaces are formed by differential growth, with adjacent cells separating from one another at the middle lamella. The process is therefore a facet of normal development, usually involving the formation of specialized cortical cells that divide and enlarge differentially to create ordered gas spaces by cell separation. Schizogenous aerenchyma occurs in many species, not only wetland plants. It is usually constitutive, though in some species it can be increased in extent in hypoxic conditions (e.g. Laan et al., 1989, 1991; Visser et al., 2003).

Schizogenous aerenchyma is often complex and well ordered. In the petiole of Sagittaria lancifolia, for instance, regular cylinders of aerenchyma develop (Schussler et al., 1997). The walls of the cylinders are formed by cells that divide both perpendicular to, and parallel with, the petiole axis. This creates a gas space within the cylinder of cells and gives increasing volume with development. A second cell type, termed diaphragm cells, which are much smaller than the cylinder cells, form single cell thick partitions across the cylinder. The gaps between these partitions increase as the cylinder cells divide and enlarge. Similar diaphragms have been described that originate from mother cells identifiable in intercalary meristems in aerial internodes of the aquatic species Scirpus validus and in leaves of other aquatics (e.g. Kaul, 1971, 1972, 1973). In other species, schizogenous aerenchyma may involve less complex structures, where the gas spaces are separated files of cells sometimes only one cell across. In their recent review, Jackson & Armstrong (1999) conclude that the developmental regulation of schizogeny is not well understood. Investigation of the developmental control mechanism of cell separation and the ordered growth and division of the cells surrounding and forming a gas space is clearly ripe for further investigation.

III. Lysigenous aerenchyma formation

Lysigeny results in the formation of an aerenchyma with conspicuous gas spaces, often with a less regular or ‘ordered’ structure than that seen for an aerenchyma formed by schizogeny. A transverse section of a root of maize in which lysigenous aerenchyma has been induced to form is shown in Fig. 2. Strands of surviving cells separate gas spaces in the cortex, formed where cells have died. This forms a tissue structure resembling a somewhat disorganized ‘spoked wheel’ arrangement. Survival of some cells in the structure is clearly important for the structural integrity of the root and for both apoplastic and symplastic nutrient transport (Drew & Fourcy, 1986). A similar pattern is observed in the formation of lysigenous aerenchyma in other species; while the tissue appears to show a degree of disorganization, with gas spaces of varying size and regularity, they are always bounded by regions of living cells, like the remaining ‘spokes’ in the maize root system.

Figure 2.

Aerenchyma in a transverse wax section of a maize root, grown in 3% oxygen. Note the gas spaces in the cortex separated by bridges of cells spanning the space between the stele and epidermis, giving the appearance of spokes in a wheel. Section stained with haematoxylin and eosin; scale bar = 100 µm. Reproduced with permission from Gunawardena et al. 2001a.

IV. Regulators of lysigenous aerenchyma formation

Constitutive lysigenous aerenchyma forms in normoxia as well as hypoxia (Jackson et al., 1985), hypoxia being one of the physiological initiators of aerenchyma in species showing induced lysigenous aerenchyma (see below). The process of cell death in these species must therefore be assumed to be initiated internally as part of development. As with constitutive schizogenous aerenchyma, however, the development of constitutive lysigenous aerenchyma may be increased by waterlogging (Das & Jat, 1977). In some instances, it is possible to identify the cells that are to die in constitutive lysigenous aerenchyma formation at an early stage. For instance, in a study of Hydrocharis morsus-ranae L. (an aquatic stoloniferous plant forming a very ordered aerenchyma), small groups of cells formed in the outer cortex by anticlinal and periclinal cell division were identified immediately beneath the hypodermis that subsequently die, leaving radial gas spaces (Seago & Enstone, 1998). However, changes are not always apparent; in rice roots (Oryza sativa L.), aerenchyma forms as cells in the mid cortex, die. These cells were characterized by Kawai et al. (1998) as being shorter and larger in radial diameter than other cortical cells. In another, detailed electron microscope study, however, Inada et al. (2002) were not able to observe any structural features that set these cells apart from their neighbours in the cortex.

Formation of induced, lysigenous aerenchyma is initiated by internal and external stimuli. In maize roots in waterlogged soils, for instance, aerenchyma forms as hypoxia develops. Low partial pressures of oxygen therefore initiate the development of the aerenchyma. In anoxia, however, the process is arrested and the tissue remains intact until death by necrosis occurs. It has been known for some time that the major internal initiator of lysigeny in hypoxia is ethylene (ethene), which accumulates within the tissue (e.g. Jackson et al., 1985). Ethylene induces aerenchyma formation in a number of species either when produced endogenously or applied exogenously (Drew et al., 1981; Jackson et al., 1985; Jackson & Armstrong, 1999; Gunawardena et al., 2001a). The development of the aerenchyma is identical whether induced by hypoxia or ethylene. The important conclusion from this is that the cellular changes are not merely a direct consequence of oxygen starvation, but result from the initiation of a cell death pathway (Gunawardena et al., 2001a). The fact that aerenchyma can also be formed in roots by mechanical impedance and by deficiency of nitrate, phosphate (Drew et al., 1989) or sulfate (Bouranis et al., 2003) also strongly suggests that a variety of stresses may act to initiate the same cell death programme. Ethylene is also involved in aerenchyma development in these cases, where it has been shown that although the rate of ethylene synthesis declines, the root tissue becomes more sensitive to ethylene and formation of aerenchyma is inhibited by inhibitors of ethylene action (He et al., 1992). Figure 1b summarizes the processes involved in lysigenous aerenchyma formation.

V. Key questions in lysigenous aerenchyma formation

While consideration of the development of schizogenous aerenchyma poses interesting questions for the developmental regulation of cell shape, enlargement and attachment, the development of lysigenous aerenchyma suggests equally important questions. The most fundamental of these is why some cells within the tissue die, while other, adjacent and apparently identical cells, do not. To answer this question the nature of the cell death process, and in particular its regulation and initiation, must be understood. While evidence for this is most abundant for induced lysigenous aerenchyma, the same questions also apply for constitutive lysigenous aerenchyma. We must also ask the converse question: what, if anything, makes the surviving cells resistant to the initiation of cell death? One possibility is that some or all of the cells die by necrosis. Necrosis is an uncontrolled form of cell death occurring when cells are exposed to variations from physiological conditions sufficient to overcome their tolerance mechanisms. It may also be induced under physiological conditions by agents causing membrane damage. The process is passive, and does not require energy usage. Necrotic cell death is frequently initiated by anoxia through oxygen starvation and acidification of the cytoplasm (Vartapetian & Jackson, 1997) and so appears relevant to aerenchyma formation. Key features of necrosis include loss of membrane integrity; loss of ionic homeostasis; swelling and disintegration of organelles; and random degradation of cellular contents including DNA. Necrosis may be localized with small necrotic lesions at growing points and in leaves, stems, etc. In animals, it is associated with an inflammatory response and the removal of the cell contents by phagocytosis. The term ‘accidental cell death’ has been suggested as an alternative to ‘necrosis’ in plants (Gray & Johal, 1998).

The alternative to necrosis is programmed cell death (PCD). Comparison of cell death in plants with PCD in animal cells is inevitable and has stimulated a number of publications in the last decade (e.g. Wang et al., 1996; McCabe et al., 1997; Mittler & Lam, 1997; Morgan et al., 1997; Miller et al., 1999, 2002; Mitsuhara et al., 1999; Lam & del Pozo, 2000; McCabe & Leaver, 2000; Mergemann & Sauter, 2000; Loake, 2001). To what extent is cell death in aerenchyma formation comparable with PCD in animals and if the process is a form of PCD, what happens to the cells that die? Thus far in this review the term ‘removed’ has been used to describe their apparent total disappearance. There are no macrophages or phagocytes in a plant system to digest and remove the debris; plants have no circulatory system to carry it away. So what is the fate of the cell debris? An added problem, compared with the removal of the contents of a dead animal cell, is the removal of the cell wall. Investigation of the late stages of the cell death pathway therefore has an additional significance for plant scientists.

Some of the questions above have been answered, at least for some species and some types of aerenchyma. Others require or suggest further research. This review concentrates on cell death processes in induced lysigenous aerenchyma in roots to address topics on the initiation and development of the gas spaces and the removal of the cell debris. While comparison with PCD systems in animal cells has been made in the past, use of different criteria and terminology has undoubtedly created confusion. An attempt will therefore be made to clarify this by consideration of all the available evidence, without trying to force plant data to fit into a mould established for quite different systems. Maize roots will be used as the major example, but evidence from other species, and from constitutive lysigenous and schizogenous aerenchyma. will be introduced where relevant.

VI. Sensing hypoxia; early events in aerenchyma formation

Maize roots in oxygen-deficient conditions show a number of responses, depending on the oxygen concentration and the suddenness of its deficiency. As already indicated, aerenchyma does not form in anoxia. However, root tip cells suddenly placed in anoxia, without time to acclimate, will die within a few hours (Roberts et al., 1984; Johnson et al., 1989; Subbaiah & Sachs, 2003). Subbaiah & Sachs (2003) propose this as a survival mechanism for the plant as a whole, and probably involves necrosis. Behind the root apex, internal oxygen deficiency is most pronounced at the stele (Gibbs et al., 1998) and it may well be that this is the first point at which hypoxia is sensed. When maize roots are placed in an environment where anoxia develops steadily over a number of hours, rapid changes in gene expression take place. Anoxia induces a rapid decrease in the cytosolic pH of root tip cells (Fox et al., 1995) and intracellular calcium concentrations rise (see below). Protein synthesis in general diminishes, but a family of 20 anaerobic proteins (ANPs) are synthesized. Most encode proteins involved in a switch to anaerobic metabolism and enhance survival of the tissue (see Subbaiah & Sachs, 2003, for a recent review). Calcium release from intracellular stores (probably mitochondria) precedes the synthesis of ANPs (Subbaiah et al., 1994; Subbaiah & Sachs, 2000). Subbaiah and Sachs therefore propose that the first site of sensing low oxygen may be within the mitochondrion, possibly a component of the electron transport chain. They also note, however, that the kinetics of the system may suggest that a lower affinity system like plasma membrane redox may be responsible. Within 5 h of anoxia, synthesis of a calcium pumping ATPase, CAP1 (Subbaiah & Sachs, 2000) is also initiated, likely to be involved in the calcium response. Indirect evidence for a role for calcium in aerenchyma formation in maize roots is provided by inhibitor studies (He et al., 1996b). Aerenchyma did not develop in roots treated with Ca chelator EGTA or with Ruthenium red, an inhibitor of Ca release from intracellular stores.

The role of ethylene as a regulator of aerenchyma formation in inducible systems has been studied extensively. Readers are directed to an early but important review of the field (Jackson, 1985) and to more recent publications (Drew et al., 1989; He et al., 1992, 1996a, 1996b; Gunawardena et al., 2001a, 2001b). Ethylene is produced from S-adenosyl methionine in a pathway involving the enzymes 1-aminocyclopropane-1-carboxylate (ACC) synthase and ACC oxidase. An important question therefore in ethylene controlled aerenchyma formation, is how the signal of low oxygen availability is transduced into ethylene biosynthesis. Formation of the enzyme ACC synthase is induced by hypoxia within a few hours of treatment (He et al., 1996a, 1996b). However, the signal transduction pathway between hypoxia and ethylene biosynthesis remains to be fully investigated. It seems paradoxical that ethylene biosynthesis, which is an oxygen requiring process (the final enzyme in the ethylene biosynthetic pathway, ACC oxidase requires oxygen) should increase in hypoxia. Upregulation of ACC oxidase, and the effect of the entrapment of ethylene in the tissue may both be important in this, as is the fact that some oxygen is required for a response; aerenchyma does not form in anoxic tissue. The formation of a permeability barrier to ethylene is also significant (Armstrong et al., 2000 and see Introduction). Presence of a barrier prevents ethylene efflux and may explain the fact that aerenchyma does not develop where lateral roots form as they provide a ‘window’ for ethylene efflux. Once released, ethylene is likely to be detected by an ethylene receptor in the cortical cells that die to form the gas space. While the role of an ethylene receptor (RP-ERS1) has been demonstrated in the response of Rumex palustris to submergence (Vriezen et al., 1997), similar direct evidence is yet to be obtained for aerenchyma formation. In Rumex, waterlogging results in a very rapid upregulation of the receptor within 20 min of exposure, and this may be one of the first events in aerenchyma formation. Signal transduction subsequent to ethylene sensing will be discussed later in this review.

Are the cells that respond by cell death in hypoxia uniquely equipped to do so? This is likely. Cell death does not initiate at the point of lowest oxygen concentration and some cortical cells die when near neighbours do not. Taken together, this suggests that, early in development, the cells develop the necessary receptors and signal transduction pathways to undergo programmed cell death. Ultrastructural evidence (see section VII below) further suggests this.

VII. Ultrastructural changes associated with lysigenous aerenchyma formation

Describing the developmental progression of aerenchyma formation requires systems in which the chronology of the process can be followed. This is easier to achieve in roots (where sequential root marking can be used accurately to age tissue) than in shoots and in an inducible system like maize, where development can be directly related to time after induction. This approach was originally taken by Campbell & Drew (1983) and refined by Gunawardena et al. (2001a) to provide a detailed developmental progression.

The first point of aerenchyma formation in maize and rice is the death of cells in the cortex. Cell death in maize then progresses radially and tangentially into surrounding cells (Justin & Armstrong, 1987). When the onset of cell death was timed by marking roots on a daily basis, Gunawardena et al. (2001a) found that roots treated with 3% oxygen or 1 µl l−1 ethylene for 3 d at 18°C, formed aerenchyma in 0.5-d-old tissue. Aerenchyma was completely formed by 2.5-d-old tissue. A previous study (Campbell & Drew, 1983) observed initiation of cell death 10 mm from the root tip, with aerenchyma fully formed by 30–40 mm from the tip. It is therefore likely that cells become sensitive to ethylene and begin the cell death process when they are less than 0.5-d-old and within 10 mm of the root tip.

In the electron microscope study of Campbell & Drew (1983) the first signs of cell death detected in sections stained for electron microscopy were the appearance of deposits (suggested to be polyphenols) in the vacuole, together with invagination of the plasma membrane (PM). In our own study (Gunawardena et al., 2001a), we observed that the first events detectable in roots treated with either 1 ppm ethylene or 3% oxygen were invagination of the PM and the formation of numerous small vesicles just inside it. These features were present in cells that were less than 0.5 d old and were accompanied by electron opaque staining in the vacuole and a heavily stained (electron opaque) tonoplast. In 1.0 d tissue cytoplasmic changes continued, with further invagination of the PM. Organelles like mitochondria and Golgi bodies remained intact, and numerous, possibly secretory, vesicles were associated with the Golgi. Signs of stress were evident and the mitochondria appeared crescent-shaped. Campbell & Drew (1983) also noted the presence of intact organelles to a late stage. Other changes observed in 1.0-d tissue and of significance in the description of the cell death pathway occurred in the nucleus. Here, chromatin, which in control cells is evenly dispersed throughout the nucleus, became concentrated at the nuclear periphery, just within the nuclear envelope. Chromatin condensation did not precede the cytoplasmic changes described above as observation showed that cells showing it always also showed cytoplasmic changes, whilst some cells showing cytoplasmic changes did not show chromatin condensation.

Perhaps the most surprizing and significant result of our study (Gunawardena et al., 2001a) was the observation of the formation of membrane bodies enclosing organelles like mitochondria, ER and Golgi apparatus, and material resembling chromatin. This preceded the final loss of cell contents and collapse of the cell wall. Campbell & Drew (1983) also recognized cell wall collapse to be a very late event in the cell death process. The resemblance of these membrane-bounded structures to apoptotic bodies was striking. Some of them are shown in Fig. 3a–d. It is important to note, however, that until their origins, development and function are defined, the evidence for apoptotic bodies is only based on resemblance after preparation for electron microscopy. There are other possible explanations for their appearance, including artefacts in EM preparation. Use of fluorescent membrane markers in living cells, or membrane specific antibodies in electron microscopy, will provide the kind of information on the origins of the membranes forming the inclusions, and on the dynamics of membrane activity during cell death that will permit rigorous comparison with mammalian systems.

Figure 3.

(a–d) Structures present in 1.5–2.0 d tissue of maize roots forming aerenchyma in the presence of 3% oxygen (a,c) and 1 ppm ethylene (b,d). Note the presence of mitochondria (a,b,d), Golgi apparatus (d), endoplasmic reticulum (c) and other cellular contents surrounded by membrane in structures bearing some resemblance to apoptotic bodies. Scale bar = 500 nm (a), 100 nm (b,c,d). Reproduced with permission from Gunawardena et al. 2001a.

Changes in the maize cell wall during cell death were also investigated by Gunawardena et al. (2001b). Using antibodies to esterified and de-esterified pectin (JIM 7 and JIM 5, respectively), wall changes were detectable at a very early stage of cell death. Whereas in control (normoxic) roots, staining with JIM 5 was restricted to triangular junctions between cortical cells, 0.5-d tissue from roots treated with 1 ppm ethylene or 3% oxygen showed staining throughout the cell wall. These changes occur at the stage when abundant vesicles are present at the PM and secretory vesicles associated with the GA. A marked increase in the prominence of plasmodesmata was also observed at this stage (Gunawardena et al., 2001a). Thus while wall degradation was not apparent by electron microscopy until a late stage in the process, it is apparent that wall changes are initiated as an early event. Previous studies on cellulase activity in aerenchyma formation in the same species showed that cellulase activity increased after an increase in ACC synthase activity (see section VI above) and within 3 d of hypoxia (He et al., 1994). Wall degradation will be considered in more detail later in this review.

Cell death to form aerenchyma in rice roots is constitutive and it is therefore more difficult to establish a precise chronology, with gas spaces already being formed in tissue as young as 1.0 d. In their electron microscope study, Webb & Jackson (1986) described some cortical cell disruption in cells that were 6-h-old, together with the first stages of wall disruption, particularly breakdown of the middle lamella. Extensive vacuolation was also noted. Wall breakdown was apparent by 12 h with loss of cell turgor; this preceded the lysis of the vacuole. Few intact organelles remained. In a subsequent study, Kawai et al. (1998) observed gas spaces in tissue 8–10 mm from the root tip. The spaces originated from the death of cells in the outer mid cortex that were larger than their neighbours. Schussler & Longstreth (1996) also observed cell enlargement before cell death in Saggitaria lancifolia. In rice, Kawai et al. (1998) noted that cell death was preceded by acidification of the cytoplasm and loss of integrity of the plasma membrane, with the gas space spreading radially. The most recent study of aerenchyma development in rice roots (Inada et al., 2002) noted the rapidity of the initiation of cell death, both in coleoptiles and roots. While suggesting tonoplast rupture was the initial event, Inada et al. did not observe any characteristic cell morphology differentiating the cells that were going to die from those around them (and see previous section above). Tonoplast rupture was followed by cytoplasmic swelling, plasma membrane rupture, loss of cellular contents and wall degradation, all in rapid succession.

While there are differences between the descriptions of the early events in aerenchyma formation in rice, possibly highlighting cultivar differences or the problems of working with a non-inducible system, there is no doubt that: (1) cytoplasmic events occur at an early stage and (2) that cell wall degradation begins with separation at the middle lamella, followed by collapse of the cells, leaving wall material surrounding the gas space. Kawai et al. (1998) also noted that the cell nuclei remained intact to a late stage. There are, however, significant differences to the process described in maize.

Changes in the nucleus are particularly significant in mammalian PCD. Deacon et al. (1986) observed ‘distinctive changes to the nucleus’ of cells dying to form aerenchyma in maize. Gunawardena et al. (2001a) also observed changes in nuclear morphology as part of (though not as the earliest stage of) cell death in the same species. Chromatin condensation to the nuclear periphery is typical of PCD in animal cells and has also been observed in cell death in plant systems. Another key change in the nucleus dying by PCD is oligonucleosomal DNA cleavage in which DNA is cleaved into oligomeric fragments which, when separated by agarose gel electrophoresis, gives a characteristic DNA ladder pattern. In addition, the nuclei of cells in which oligonucleosomal cleavage has taken place show positive staining with the TUNEL (tDNA dUTP nick end labelling) procedure (Gavrieli et al., 1992). In this procedure free OH termini in DNA strand breaks are labelled using a terminal deoxynucleotide transferase. Great care needs to be taken to ensure that DNA degradation by other means does not take place as this can result in false positive results. Gunawardena et al. (2001a) used the TUNEL procedure successfully with maize and obtained staining of the nuclei of cells of the mid cortex from roots that were treated with either 1 ppm ethylene or 3% oxygen. This was first observed in 0.5-d tissue and was also apparent in tissue > 1-d-old. It proved difficult to confirm DNA cleavage using agarose gels, as the tissue samples inevitably contained both healthy cells and those dying at different stages. However, suggestion of DNA laddering was obtained (Gunawardena et al., 2001a). The TUNEL positive cells were also observed in the epidermis (but not the stele) of cells treated with ethylene or low oxygen (Fig. 4). As these epidermal cells did not appear to subsequently die, the reason for TUNEL staining was unclear. However, it was clearly initiated by both hypoxia and ethylene, and was not observed in control cells (Gunawardena et al., 2001a). The first TUNEL positive nuclei were observed in cells before chromatin condensation was evident by electron microscopy. Interestingly (Kawai et al., 1998) did not obtain a positive result for TUNEL staining in rice aerenchyma formation. This may reflect the short time course of aerenchyma formation in rice, or a major difference in the nature of the process.

Figure 4.

TUNEL (tDNA dUTP nick end labelling) of 0.5 d tissue sections from maize roots treated with 21% oxygen (a,b,c), 3% oxygen (d,e,f) an 1 ppm ethylene (g,h,i). TUNEL positive nuclei are present in sections treated with 3% oxygen (d) or 1 ppm ethylene (g) but not 21% oxygen (a). Images b, e and g show equivalent sections stained with propidium iodide (PI); co-localization (orange) of TUNEL and PI staining is shown in c, f and i. Reproduced with permission from Gunawardena et al. 2001a.

VIII. Late events in cell death in aerenchyma formation

The most useful studies of the late stages of cell death in aerenchyma formation are those that have used cryo-techniques to attempt to preserve the cellular contents and debris in situ for microscopy, as conventional fixation removes material from severely disrupted cells and tissues. van der Weele et al., 1996) conducted a cryo-SEM study of maize roots in which the late stages of the progress of aerenchyma development was observed. This study differed from those described previously in that the aerenchyma was formed in older tissue, 5–10 cm from the tip of mature roots of soil grown plants that were not waterlogged. X-ray microanalysis was used to determine ionic concentrations within the tissue. The first sign of aerenchyma formation in this system was the presence of groups of cortical cells, which were beginning to show signs of turgor collapse, still attached to their neighbours. These cells were fluid filled, but markedly depleted in nutrient ions, whilst the surrounding tissue was ion-rich. In the next stage, a fluid filled space was detected. The ionic content if this fluid resembled that of the cells in the previous stage and was assumed to be derived from them. The walls of these cells were collapsed and torn. It was suggested (van der Weele et al., 1996) that cell walls were finally degraded in this fluid. Finally, the fluid was lost from the space and cell degradation concluded at the edges of a ‘dry’ gas space. This suggests that ionic reabsorption into surrounding tissues occurs at an early stage, with turgor loss, and that the absorption of the cell wall and contents occurs at least initially from a fluid-filled space. There is no evidence, from our own study (Gunawardena et al., 2001a), or from any other for the persistence of the membrane-bounded organelle inclusions to this stage as the techniques employed would not have preserved them in situ. The topic would be ideal for investigation by rapid-freeze high resolution transmission electron microscopy and the use of probes for different membrane or other cellular components would greatly enhance information gained.

The activation of wall degrading enzymes has already been mentioned. In addition to our study of changes in esterified and de-esterified pectins (Gunawardena et al., 2001b), the role of other wall degrading enzymes has been suggested. Enzymes involved are likely to include expansins, cellulases, xyloglucan endo-transglycosylase (XET) and pectinases (Jackson & Armstrong, 1999). Ethylene induced upregulation of expansin has been shown (Rose et al., 2000) and XET transcripts increase in maize roots after 12 h of hypoxia due to flooding (Saab & Sachs, 1996), the rise being inhibited by ethylene biosynthesis inhibitors. In a study of sunflowers, Kawase (1979) demonstrated that an increase in cellulase activity accompanied aerenchyma formation. In rice, where aerenchyma formation occurs constitutively, cell wall degradation is the first visible sign of cell death and is not stimulated by hypoxia or ethylene.

Less is known about the degradation of other cellular components, though the action of proteases, lipases and other hydrolases, either from a vacuolar store or synthesized during cell death seems likely. This would leave the cellular contents as a collection of low molecular mass solutes to be absorbed by surrounding cells. The activity of transporters for both mineral and organic nutrients in cells surrounding the dying cells during this process has yet to be studied.

Another important aspect of the late stages of aerenchyma formation is limitation of the spread of the lesion such that surrounding cells do not die. Once again, little is known with certainty. As indicated above, the surrounding cells are likely to be nutrient-rich, but until the aerenchyma is effective in delivering oxygen it will still be hypoxic. It is known that cell packing is important, with aerenchyma developing less frequently and less extensively where cells are packed hexagonally than where they are packed cubically (Justin & Armstrong, 1987; Drew et al., 2000). Otherwise, it seems likely that there are differences in the sensitivity of these cells to the stimuli initiating cell death – hypoxia and ethylene – or in the response pathways present for subsequent events. Such differences may be a subtle difference in sensitivity rather than an absolute response/no response; this would explain the further development of aerenchyma in more extreme conditions. The fact that in rice (for instance) cell death begins a specific file of cortical cells strongly suggests specific developmental regulation.

IX. Comparative evidence for programmed cell death in aerenchyma formation

It has already been commented that much scientific activity has been dedicated to identifying processes involving programmed cell death in plants. However, confusion has been generated by the loose application of terms and the desire to make plant systems ‘fit’ with an animal cell paradigm. A variety of methods is available to identify programmed cell death in plants, and by taking several of them together, programmed cell death may be implicated with some confidence. The fact that diagnosis needs to be made in multicellular tissues further complicates identification and some methods are not appropriate for intact tissues or tissue sections.

The following criteria were suggested for description of plant programmed cell death resembling apoptosis (Evans, 2003):

  • 1Description of the factor or factors initiating the cell death process. Can cell death showing similar characteristics be initiated by several independent triggers? Can an endogenous trigger be identified (e.g. a hormone)?
  • 2Is cell death limited to a group of cells or a tissue? Once initiated, does it spread unchecked or is its progress limited?
  • 3Does gene expression, metabolism or other cellular activity continue to a late stage in the process?
  • 4Is there evidence for oligonucleosomal fragmentation of DNA?
  • 5Is chromatin condensation to the nuclear periphery observed?
  • 6Is membrane integrity maintained to a late stage?
  • 7Are membrane bound inclusions formed?

Criteria (1)–(3) all appear to be true for aerenchyma development in maize. Criteria (4)–(7) are key indicators of processes resembling apoptosis. Sensu strictu, apoptosis is described by the development of characteristic cell morphology (Kerr et al., 1972; Kitanaka & Kuchino, 1999). It was originally characterized by chromatin condensation to the nuclear periphery and by the formation of membrane structures surrounding cellular organelles. These latter structures, termed ‘apoptotic bodies’ contain chromatin and intact organelles including mitochondria and Golgi apparatus. Cell membranes and organelles do not lose integrity until a late stage in the process. Finally, the cell fragments into small, membrane-bounded fragments and is rapidly digested by macrophages. Apoptosis is not, however, the only form of PCD in animal cells. Membrane changes at an early stage also occur in a second form of PCD, known as cytoplasmic cell death (CCD; Clarke, 1990). In this process, nuclear changes do not occur until vesiculation and the formation of autophagic vacuoles has taken place. However, in contrast to apoptosis, organelles are frequently degraded in specific sequence, before chromatin condensation. Another form of cell death, termed autophagic cell death, features the activity of autophagosomes derived from lysosomes (Kitanaka & Kuchino, 1999).

Table 1 summarizes comparison of the stages of cell death described for maize with the stages of apoptosis. It is apparent that two pieces of evidence stand out to link the process with apoptosis; the retention of intact organelles surrounded by membrane and chromatin condensation, and oligonucleosomal cleavage of DNA. The latter has been detected in a number of plant cell death systems, including tracheary element differentiation (Mittler & Lam, 1997), pathogen induced cell death (Ryerson & Heath, 1996) and root cap cell shedding (Wang et al., 1996). The former has been described only for root cap cell shedding (Wang et al., 1996). The process is not, however, identical to apoptosis, as cytoplasmic events precede nuclear events. This is more like cytoplasmic cell death, but not identical to it. It is therefore appropriate to consider the cell death process in aerenchyma to be unique, different from both apoptosis and cytoplasmic cell death, though with properties resembling both. Perhaps, in deference to De Bary, 1884) the term ‘lysogenetic cell death’ might be deemed appropriate, though care would have to be taken to avoid confusion with lysogenic cell death – of a host cell as a result of infection.

Table 1.  Comparison of the stages of cell death described for maize (Zea mays) with the stages of apoptosis
StageEvents in maize aerenchyma formationEvents in mammalian apoptosisEvents in cytoplasmic cell death (Zakeri et al., 1995)
Stage 1Perception of hypoxia; initiation of ethylene biosynthesis  
Stage 2Perception of ethylene signal by cells of mid cortexPerception of cell death stimulusPerception of cell death stimulus
Stage 3Initiation of cell death; cell begins to lose ions to surroundingsOligonucleosomal DNA cleavage TUNEL positive nucleiVesiculation, vacuolation and lysosomal activity
Plasma membrane invagination and vesicles in cell cortex; vacuole and tonoplast staining electron opaque; first cell wall changesCondensation of chromatin to nuclear peripheryOrganelles degraded
Stage 4Chromatin condensation to nuclear periphery; TUNEL positive nucleiInvagination and shrinkage of plasma membraneOligonucleosomal DNA cleavage and chromatin condensation
Organelles intact; wall hydrolytic enzymes increase.Condensation of cytoplasm 
Organelles surrounded by membraneVesicle formation; organelles intact;apoptotic bodies formed 
Stage 5Cell lysis occurs leaving fluid filled space Absorption of cell contents and water by surrounding cellsPhagocytosis of cell membranes and contents by macrophagesPhagocytosis

It is pertinent to ask whether cell death in rice also shows the characteristics of PCD resembling apoptosis. The cell death process in rice commences with cytoplasmic events and cell wall changes; as yet there is no evidence for membrane inclusions resembling apoptotic bodies and an attempt to stain nuclei using the TUNEL procedure was negative (Kawai & Uchimiya, 2000). Two conclusions are possible; either the formation of constitutive aerenchyma in rice is different from that in maize in almost every respect, or the studies conducted to date have not identified features that are present, possibly momentarily. It seems at least surprizing (though by no means impossible) that the two systems should be so different, and the findings should certainly inspire research into the two systems, including a far wider range of characteristics of programmed cell death.

In contrast to ultrastructural/morphological changes, the biochemical events of the cell death pathway in aerenchyma formation are sparsely dealt with in the literature and do not yet permit detailed comparison with biochemical events described in animal PCD. In reviewing the topic in 2000, Drew, He and Morgan (Drew et al., 2000) cited their own largely unpublished data supporting the involvement of a signal transduction pathway involving inositol 1,4,5-trisphosphate, a rise in cytosolic calcium and protein phosphorylation in cell death. The fact that anoxia induces an increase in cytosolic free Ca in maize suspension cells has already been referred to (Subbaiah et al. 1994). PCD in animals also involves an increase in cytosolic calcium and the protein phosphorylation to initiate the activity of proteases and endonucleases.

The pathway of PCD in animal cells is now well described. It may be initiated by cell death receptors at the cell surface that are members of the tumour necrosis factor family, and/or by events that result in loss of cytochrome c and apoptotic induction factor from mitochondria. Activation of a family of cysteine proteases called caspases ensues in a cascade, resulting in both cytoplasmic and nuclear events. For instance, the characteristic changes in the nucleus are caused by caspase activation of DNase, inhibition of DNA repair, and lamin degradation. Evidence is lacking on these processes in aerenchyma formation. However, cytochrome c release from plant mitochondria during heat-induced PCD of cucumber cells has been observed (Balk et al., 1999) and a sunflower cytoplasmic male sterility (CMS) mutant showing premature PCD also showed cytochrome c release (Balk & Leaver, 2001). Readers are referred to a review article entitled ‘Do plant caspases exist?’ (Woltering et al., 2002) for a considered debate of the evidence for this family of proteins in plants. They conclude that while no functional homologues of caspases have been identified in plants, considerable indirect evidence suggesting their existence has accumulated. This includes results from experiments using a variety of classes of caspase inhibitors. Elbaz et al. (2002) also provide evidence for caspase-like proteins in PCD, this time in tobacco cells. Once again, comparable evidence is lacking for aerenchyma and considerably more is known about enzymes other than proteases in the process (see section VII above). Clearly there is considerable scope for investigation of the pathway initiating and executing cell death. Aerenchyma, as internal tissues, may however, prove hard to study and the use of comparable suspension cell lines is likely to initially be more fruitful.

X. Comparison with other abiotic initiators of cell death in plants

Programmed cell death in aerenchyma formation has seldom been compared with plant systems in which PCD is induced by abiotic stresses other than hypoxia. It has already been noted that in some instances, other stress factors can result in the development of aerenchyma. These include temperature stress, nutrient deficiency, and mechanical impedance. Abiotic stress can also induce forms of programmed cell death in other tissues. It is instructive to consider the mechanisms for the initiation of cell death by other abiotic factors and in other systems to establish whether there are commonalities of mechanism, or whether these systems can yield clues for as yet missing components of our understanding of aerenchyma formation. The role of reactive oxygen species (ROS) in initiating cell death has been suggested by work on several abiotic stress systems. ROS include molecules like hydrogen peroxide and ozone, ions like hypochlorite, and radicals like hydroxyl and superoxide. Gaseous pollutants can induce the production of ROS in the plant which are known to initiate cell death pathways and ROS formation is part of the hypersensitive response to pathogens (Sandermann et al., 1998). For instance, ozone (O3) reacts with cell wall phenolics, olefins, and unsaturated membrane lipids (Mudd, 1997) to give ROS. Exposure of a plant leaf to ozone may induce the formation of antioxidant defences, followed by a cell death response (Draper, 1997; Rao & Davis, 1999; Rao et al., 2000a). Jasmonic acid and ethylene signalling (Rao et al., 2000b) has also been implicated in inducing cell death in ozone exposure. ROS formation is also induced by sulfur dioxide (SO2). In the leaf, it hydrates giving HSO3 and SO32–, leading to a free radical chain reaction initiated by the interaction of SO32– and O2 · 2– (Madamanchi & Alscher, 1991). Clearly any stress that can induce ROS formation can potentially induce PCD. Intense illumination, coupled with stomatal closure, results in overloading of the electron transport chain resulting in electron leakage and oxidative damage. In these circumstances, oxygen acts as an electron acceptor, with the production of superoxide (O2 · ) ions which are dismutated by superoxide dismutase (SOD) to give the accumulation of high levels of other reactive oxygen species (ROS) like H2O2. While there is no evidence for PCD induced by visible light, intense illumination by ultraviolet (UV) light at wavelengths not normally encountered by plants (UV-C) initiates a cell death pathway (Danon & Gallois, 1998). This includes nuclear fragmentation detectable by TUNEL and DNA ‘laddering’. ROS involvement has also been suggested in PCD induced in some forms of chilling injury (Kratsch & Wise, 2000).

There is evidence linking ethylene, ROS and low oxygen tensions in other systems. In nutrient-starved carrot suspension cells, ethylene was produced after 1 d, high levels of ROS were evident after 4 d and inhibitors of ROS production inhibited cell death (Chae & Lee, 2001). Internucleosomal DNA fragmentation was observed as part of this cell death process. In cell death in flower petals of some species, ethylene production also precedes ROS formation (Rubinstein, 2000). Perhaps most interesting for the study of aerenchyma formation is the finding that superoxide and hydrogen peroxide were detected in maize root cells dying to form aerenchyma as a result of sulfate starvation (Bouranis et al., 2003). A pathway in which ROS (hydrogen peroxide) formation is induced by oxygen deprivation has been implicated in the induction of ethanolic fermentation in Arabidopsis roots (Baxter-Burrell et al., 2002). The pathway involves the activation of a calcium-dependent NADPH oxidase by the action of Rop, a small GTPase.

Another form of abiotic stress in which programmed cell death has been implicated is that induced by mechanical stress. Here, a role for nitrous oxide (NO) has been suggested. When leaves of Kalanchoe diagramontiana were exposed to 10–50 g for 10–60 min, cell death was induced. TUNEL positive chloroplasts and then nuclei were observed at an early stage of the treatment. Chromatin migration to the nuclear periphery was followed later by disorganization of the nuclear envelope (Pedroso & Durzan, 2000). TUNEL positive material was observed in bodies surrounding the nucleus. Synthesis of nitric oxide (NO) occurred before DNA fragmentation as a result of gravity stimulation. Cell death was stimulated when a NO donor (sodium nitroprusside) was added and was inhibited when NG-monomethyl-l-arginine (NO synthesis inhibitor) was added. TUNEL positive cells were also observed in Arabidopsis (Garces et al. 2001), after exposure to gravity. The role of NO was investigated, using a nitrate assimilation mutant. Bursts of NO following both mechanical stimulation and wounding were shown not to be a by-product of nitrate reductase activity. Addition of the nitric oxide synthase inhibitor NG monomethyl arginine inhibited NO production and the NO-release agent sodium nitroprusside induced DNA fragmentation. Nitric oxide and ROS signalling are closely interrelated and stresses that induce H2O2 are also likely to induce phospholipase activity, giving stimulation of nitric oxide synthase and release of NO. Changes in NO and ROS are suggested to alter the levels of signalling molecules and hormones like jasmonic acid, salicylic acid and ethylene (Rao & Davis, 2001). It has been suggested that NO production in plants may also involve interactions with ethylene and ROS (Wojtaszek, 2000). An important area for future research must therefore be further analysis of the components of pathways of signalling involved in initiating cell death in aerenchyma formation.

XI. Conclusions and future prospects

Our current understanding of the formation of aerenchyma has resulted from detailed and innovative studies on a variety of systems, both lysigenous and schizogenous, induced and constitutive. It is now possible to thoroughly describe morphological and ultrastructural events for several model aerenchyma systems. This ultrastructural approach strongly suggests programmed cell death showing some (but not all) of the key features of apoptosis in maize (Evans, 2003). This is not true of all species studied. In addition, there remain gaps in knowledge of the developmental regulation and biochemistry of aerenchyma formation. Identification of cell death genes is required. This would most profitably initially be undertaken in an inducible system. The task is complicated by the induction of the ANP gene family (see section VI above) by anoxia; but analysis of the expression of ethylene-induced genes in the root cortex and comparison with hypoxia and anoxia-induced expression should permit the identification of major components of the cell death pathway. While precise plant homologues for the mammalian proteases involved in cell death seem unlikely, identification of proteins of equivalent or similar function should be a priority. Identification of the genes for the developmental regulation of the process should also be seen as a key research objective.

Several factors limit research of both induced and constitutive lysigenous aerenchyma. It occurs in some cells within a complex organ. It occurs in species that are genetically complex and useful mutants are lacking (Arabidopsis does not form aerenchyma). Use of suspension cultures provides one opportunity for studying cell death in a simplified system (McCabe & Leaver, 2000) and is likely to provide data transferable to the whole plant system. The progress of the rice genome project will also help in the identification of genes for aerenchyma development in this species. Development of maize mutants, together with our increased understanding of maize development, will also yield valuable information.

One study has investigated the possibility of introducing constitutive aerenchyma into maize (Ray et al., 1999). This was based on the fact that the progenitor of maize, Zea luxurians (teosinte) and a near relative, Tripsacum dactyloides (Eastern gamma grass) both form aerenchyma in well-aerated soil. Genetic analysis of the progeny and of back crosses suggested that constitutive aerenchyma is controlled by more than one gene locus, but that a major locus is present on the short arm of chromosome 16 of T. dactyloides. Clearly, adopting genetic and molecular genetic approaches, involving quantitative trait loci (QTL) and techniques like restriction fragment length polymorphism (RFLP) and amplified fragment length polymorphism (AFLP) would be promising approaches to identifying and working with aerenchyma genes in maize.

Another research goal for induced aerenchyma is the analysis of the signal transduction pathways from oxygen sensing to ethylene production and from ethylene to the initiation of the cell death process. He et al. (1996b) and the unpublished data cited in Drew et al. (2000) indicate the involvement of a signal transduction pathway involving inositol 1,4,5-trisphosphate, a rise in cytosolic calcium and protein phosphorylation in cell death. To date, evidence for this pathway is indirect – based on inhibitors and activators. Thorough biochemical and molecular identification of its components should be seen as a priority.

Cell death in aerenchyma formation is of both fundamental and economic significance. As part of the series of adaptations to stress possessed by plants it provides a target for crop improvement, especially in the light of degrading agricultural environments globally. Detailed molecular, genetic and ultrastructural understanding of the processes involved are therefore essential tools for progress.


The author wishes to acknowledge the work of Dr Arunika Gunawardena upon whose doctoral research in his laboratory the work on maize aerenchyma described in this review is based. Thanks are also due to Dr Julian Coleman who critically reviewed the final manuscript.