Ethylene† and fruit softening

Abstract

This review is concerned with the mechanisms controlling fruit softening. Master genetic regulators switch on the ripening programme and the regulatory pathway branches downstream, with separate controls for distinct quality attributes such as colour, flavour, texture, and aroma. Ethylene plays a critical role as a ripening hormone and is implicated in controlling different facets of ripening, including texture change, acting through a range of transcriptional regulators, and this signalling can be blocked using 1-methylcyclopropene. A battery of at least seven cell-wall-modifying enzymes, most of which are synthesized de novo during ripening, cause major alterations in the structure and composition of the cell wall components and contribute to the softening process. Significant differences between fruits may be related to the precise structure and composition of their cell walls and the enzymes recruited to the ripening programme during evolution. Attempts to slow texture change and reduce fruit spoilage by delaying the entire ripening process can often affect negatively other aspects of quality, and low temperatures, in particular, can have deleterious effects on texture change. Gene silencing has been used to probe the function of individual genes involved in different aspects of ripening, including colour, flavour, ethylene synthesis, and particularly texture change. The picture that emerges is that softening is a multi-genic trait, with some genes making a more important contribution than others. In future, it may be possible to control texture genetically to produce fruits more suitable for our needs.

Introduction

Fruit ripening involves a series of changes in colour, flavour, texture, aroma, and nutrient content, which affect quality, post-harvest life, and value. Most fleshy fruits undergo some, if not all, of these changes during ripening, which, in evolutionary terms, are designed to make them attractive to eat and therefore aid seed dispersal. Perhaps, the best-studied fruit is tomato, where it has been established that ripening is controlled by the interaction between genetic and hormonal factors, and the general features of this ripening model seem applicable to many fruits. Essential features are the involvement of both transcriptional regulators, such as RIN, which was discovered by analysis of the rin tomato mutant (Vrebalov et al., 2002), and structural genes that control biochemical changes. In addition, the action of hormones, particularly ethylene, is essential, and downstream of ethylene the signalling pathway branches to control different facets of ripening separately (Figure 1; Grierson, 2013). Textural change is one of the major characteristic features of fruit ripening and, although the underlying mechanisms are complex, and the details vary between different fruits, textural change is generally recognized to involve both softening and changes in crispness and juiciness of the fruit (Harker et al., 1997; Chaïb et al., 2007). These changes result from major alterations in the structure and composition of the cell wall components during ripening, which are considered as major contributory factors in the softening process (Brummell, 2006; Vicente et al., 2007). However, other mechanisms may also be active in determining the overall textural characteristics of the fruit. Banana, for example, is packed with starch, and part of the texture change during ripening is also due to the conversion of insoluble starch to soluble sugars (see Kojima et al., 1994). In addition, partly because of starch solubilization and changes in cell wall structure, and perhaps also due to other factors, such as changes in activity of water channels and cuticle properties, there can be substantial alterations in the distribution of water molecules within fruits during ripening. All of these changes can have a profound effect on juiciness, texture, and overall softening. Various aspects of ripening, including softening, can also be influenced by the post-harvest environment, for example high and low temperature, the presence of applied or naturally occurring gases, all of which can influence changes in gene expression and the activities of enzymes determining overall ripening behaviour.

In this review, we shall focus on fruit cell wall structure, the mechanisms that bring about texture- and softening-related changes during ripening and post-harvest storage, and consider the contribution and role of gene expression, ethylene, and other factors in controlling these changes.

Cell wall composition and structure

Plant cell walls are very diverse and complex structures, and there are significant differences between different groups of plants, and between different cells within a plant. Growing plant cells are surrounded by a dynamic, carbohydrate-rich primary cell wall and, in dicots, in particular, the space between adjacent cells is occupied by a pectin-rich middle lamella (Popper et al., 2011). In those tissues that have ceased growth (such as the woody tissue of trees), the cell walls undergo a process of lignification in which an aromatic polymer (lignin) is laid down to form a relatively inert secondary cell wall.

Fruit in general have primary walls with very little, if any, lignification (Toivonen and Brummell, 2008; but see the later discussion on loquat fruit post-harvest lignification).

Several general models of the primary cell wall have been constructed during the last 30 years (Carpita and Gibeaut 1993; Carpita, 1996; Somerville et al., 2004; Cosgrove, 2005; Doblin et al., 2010; Albersheim et al., 2011). These all have the general feature of cellulose fibrils being embedded in a matrix of hemicellulose and pectic polymers (Figure 2). The wall also contains a certain amount of structural proteins, such as extensin, hydroxyproline-rich, and arabinogalactan proteins. Cell walls are generally assigned to one of the two general groups—type I or type II. Type I cell walls are normally associated with dicotyledonous plants and type II are associated with monocotyledonous species. Cellulose fibrils are a feature of both types of wall with the major differentiating factors being the nature of hemicellulosic polymers present and the relative proportion of pectic polymers, type I walls having a much higher proportion of pectin than type II. There are, however, a number of plants that appear to have cell wall compositions that are intermediate between these two types, one example being the pineapple (Smith and Harris, 1995).

The cellulose fibril structure is relatively conserved. Cellulose polymers occur as long (up to 5000 residues) linear chains of β-1,4 linked glucose residues. These are found grouped together in fibrils (Somerville, 2006). The precise number of polymers in each fibril is still a matter of debate and may vary between cell walls, but is generally considered to be between 24 and 36. The polymers are aligned parallel to each other, held together by hydrogen bonds to give a crystalline structure that is very resistant to degradation. The degree of crystallinity may vary along the length of the fibril (Viëtor et al., 2002; Šturcová et al., 2004) and again between cell walls. The fibrils themselves also tend to be aligned parallel to each other within each plane of the cell wall.

Hemicelluloses are represented by a very diverse range of structural polymers (Scheller and Ulvskov, 2010). In type I walls of dicotyledonous species, a major hemicellulose is xyloglucan. This polymer also has a backbone of β-1,4 linked glucose, but in this instance, this is substituted by single xylose residues, some of which may be further substituted by galactose and fucose residues (Figure 3). In contrast, the predominant hemicelluloses in type II cell walls are based on xylans. In this case, there is often a backbone of β-1,4 linked xylose residues again, with substitutions, except that a major sugar substituent is arabinose. The xylose residues can also be acetylated in some cases. Other hemicelluloses often present in type II walls include galactomannans. Although these hemicelluloses have very different monomeric compositions, they often seem to share a common structural feature in that they are thought to be capable of hydrogen bonding to the surface of the cellulose fibrils and also embedding themselves in the para-crystalline regions of the cellulose fibril (Pauly et al., 1999; Rose and Bennett, 1999). It has also been shown that, in some instances, the hemicellulose can traverse the gap between adjacent cellulose fibrils and may serve to act as tethers between the fibrils.

The structure and synthesis of pectins has been reviewed (Mohnen, 2008). The three major pectic polymers homogalacturonan (HMG), rhamnogalacturonan I (RhaI), and rhamnogalacturonan II (RhaII) have a generally conserved structure (Figure 4). HMG consists of chains of α-1,4 linked galacturonic acids, the individual monomers of which may exist as either a the free acid or a methyl ester. Where there are blocks of de-esterified residues on adjacent polymers, these may interact through calcium chelation to form so-called egg box structures (Grant et al., 1973). RhaI consists of a backbone of alternate galacturonic acid and rhamnose monomers, with rhamnose being substituted by side chains of galactose and arabinose residues. Individual walls can vary in the proportion and fine structure of these polymers. RhaII is a very complex polymer consisting of 12 sugars and 20 different glycosidic bonds. It is only present in small amounts compared with HMG and RhaI, and seems to be highly conserved. More recently, a fourth pectic polymer, xylogalacturonan, has been identified, and this has been found in several fruits, including watermelon and apple (Thibault and Ralet, 2001).

Changes in cell walls during ripening

There have been several reviews covering the structural and compositional changes that occur in the cell walls of ripening fruits (Rose et al., 2003; Brummell, 2006; Vicente et al., 2007; Toivonen and Brummell, 2008; Li et al., 2010; Ruiz-May and Rose, 2012). There are changes to the structure and chemical composition of most of the fruit cell wall polymers during ripening (Brummell, 2006), and although these may vary between species (Toivonen and Brummell, 2008), there are some common features. These changes affect the cell wall ultrastructure and can be detected by light and electron microscopy (Figure 5). During ripening of fruit, the middle lamella can be seen to swell (Redgwell, 1997), indicating hydration. This is thought to be brought about by Donan forces resulting from increases in the fixed charge on the pectin polymers.

At a molecular level, the earliest detectable change seems to be a loss of neutral sugars, galactose, and arabinose, (Gross and Wallner, 1979) presumably from the side chains of RhaI. This can occur slightly before the onset of other aspects of ripening. There is also a decrease in the degree of methylation of the pectin. In tomato, this reduces from 90% in green fruit down to about 35% in ripe fruit (Koch and Nevins, 1989). This reduction in methylation increases the negative charge on the pectin and is presumably the major cause of the Donan forces that result in swelling of the middle lamella. There is also a significant depolymerization of the pectin polymers.

These changes are brought about by enzymatic action within the wall; indeed, Hobson (1964) demonstrated that ripe tomato fruit contained sufficient enzyme activity to completely de-esterify and depolymerize the pectin in the cell wall within about 4 min of the fruit pericarp being homogenized. This suggests that the enzyme activities in situ must be very tightly regulated and suppressed.

Enzymes involved in cell wall metabolism during ripening

These ultrastructural changes are due to the activity of a range of enzymes that modify the structure, physical, and chemical properties of various cell wall components. It would appear that a similar range of wall-modifying enzymes are present in most fleshy fruit, but that their relative activities may vary (Brummell, 2006). A number of enzyme activities have been identified as being associated with the cell walls of ripening fruit and some key examples are shown in Table 1 along with their proposed modes of action. This list is not exhaustive but represents the most intensively studied enzymes believed to have a potential role in fruit softening.

Polygalacturonase (PG) has been shown to be absent in green tomato fruit and increases dramatically during ripening (Hobson, 1964). The enzyme is capable of cleaving the α-1,4 linkage between galacturonic acids in HMG but only if the two adjacent galacturonic acids are both de-esterified. The enzyme has been shown to exist in at least two isomeric forms (PG1 and PG2) in tomato fruit (Pressey and Avants, 1973) with a sequential expression during ripening (Tucker et al., 1980). These two isoforms were purified and both, when analysed by denaturing electrophoresis, contained a common 46 kDa polypeptide (Tucker et al., 1980). It was further shown that green fruit contained a heat-stable non-dialysable factor that could convert PG2 into an isoform with the same physical properties as PG1 (Tucker et al., 1981; see also Pressey 1988), and this was termed the converter and later the β-subunit of PG by others. The role of this is unclear but has been reviewed (Brummell and Harpster, 2001). The converter protein was later identified as being an integral component of the fruit cell wall in unripe fruit (Pogson and Brady, 1993), and results of its down-regulation suggest that it may restrict PG activity (DellaPenna et al., 1995; Chun and Huber, 2000). One theory is that the β-subunit restricts PG activity, either directly or by acting as a target for the action of PG during the initial stages of ripening and thus limiting its distribution throughout the wall. This may provide one potential mechanism for the tight regulation of cell wall degradation seen during ripening. Increased polygalacturonase activity, and/or transcripts of the corresponding genes, has been detected in a large number of fruit other than tomato, including strawberry (Pose et al., 2013), pear (Hiwasa et al., 2003), banana (Asif and Nath, 2005), papaya (Fabi et al., 2009), guava (Abu-Gourkh et al., 2003), peach (Ghiani et al., 2011), and apple (Costa et al., 2010). Several fruits have been shown to express another enzyme—pectate lyase (PL)—that is also capable of depolymerizing the HMG. However, unlike PG, which acts as a hydrolase, PL acts through a β-elimination mechanism. This enzyme activity has also been detected in banana (Marín-Rodrígues et al., 2002), strawberry (Jiménez-Bermúdez et al., 2002), and tomato (Uluisik et al., 2016).

Pectinesterase (PE) is capable of de-esterifying the galacturonic acid residues in HMG to form the free acid. This functionality implies an obvious potential synergy with the action of PG, which requires adjacent de-esterified residues for action.

Unlike PG, this enzyme has been shown to be active in tomato fruit throughout both development and ripening. However, as with PG, PE has also been shown to occur in several isoforms. Pressey and Avants (1972) identified four PE isoforms in tomato fruit, whereas Tucker et al. (1982) reported three in the Ailsa Craig tomato cultivar. One of these isoforms (PE2) appeared to be fruit-specific and increased in activity during fruit development (Tucker et al., 1982). Gaffe et al. (1997) separated five PE isoforms from tomato fruit and demonstrated that three of these were fruit-specific and the other two appeared to be found ubiquitously in different tissues.

As with PG, PE activity has been detected in a very wide range of fruit and, indeed, this enzyme activity is associated with a much wider range of plant tissues. The PE enzyme may be involved in the generation of blocks of de-esterified galacturonic acid residues in HMG and thus contribute to wall stability through the formation of calcium ‘egg box’ structures (Alonso et al., 1995; Micheli, 2001). Thus, separate isoforms may be contributing to both wall strength and degradation during different phases of plant development. It has been suggested that PE activity may also be influenced by the presence of specific pectinesterase inhibitors (D’Avino et al., 2003), and these have been isolated from kiwifruit (Irifune et al., 2004).

β-Galactosidase (β-Gal) is an enzyme activity that is found in many plant tissues and has the potential to remove galactose residues from the side chains of RhaI. However, this enzyme activity is normally assayed by its ability to release nitrophenol from an artificial para-nitrophenol galactoside substrate and this catalytic activity does not necessarily imply that the enzyme would be active against a natural galactan substrate as found in RhaI. That said, this enzyme activity has been detected in at least three isoforms in tomato fruit, only one of which was capable of degrading a native galactan substrate (Pressey, 1983). Since then, a number of β-galactosidase/exo-(1,4)-β-d-galactanases have been isolated from persimmon (Kang et al., 1994), kiwifruit (Ross et al., 1993), apple (Ross et al., 1994), and tomato (Carey et al., 1995).

An increase in β-1,4-glucanase activity, or the expression of the corresponding genes, is often correlated with fruit softening (Brummell et al., 1994; Urbanowicz et al., 2007). However, the substrate for β-1,4-glucanase in situ is not entirely clear and again this is partially due to the fact that this activity is normally determined using artificial β-1,4-glucan polymeric substrates. Although this enzyme is capable of hydrolyzing β-1,4-linked glucose polymers, its action against cellulose is thought to be limited. Initial studies indicated that its actual substrate may be xyloglucan, but this has been questioned because this hemicellulose is not readily hydrolyzed, at least in vitro (Urbanowicz et al., 2007).

Xyloglucan transglycosylase-hydrolases (XTH) are enzymes that can exhibit dual catalytic activity (Fry et al., 1992). They can either act as hydrolases to cleave the bond between adjacent glucose residues in xyloglucan resulting in depolymerization. Alternatively, they can cleave the xyloglucan, but only when an alternate bond with another xyloglucan oligomer or polymer can be formed (Rose et al., 2002). Different subclasses of this enzyme have been shown to be predominant in either the hydrolase or glycosylase activity in vitro. It is thus postulated that in situ this enzyme may be capable of ‘breaking and making’ bonds between adjacent polymers tethering the cellulose fibrils, thus allowing polymer creep within the wall. This activity has been identified in several fruits including tomatoes (Arrowsmith and de Silva, 1995), pears (Fonseca et al., 2005), and grapes (Ishimaru and Kobayashi, 2002).

Expansin has an important effect on cell walls, although it is not an enzyme, in the strict definition of the term, because it does not catalyse a chemical reaction. Instead, it is thought to interact at the junction of the cellulose fibril and the coating hemicellulose polymers to disrupt the hydrogen bonds, although their exact molecular targets are unclear (Brummell et al., 1999; Georgelis et al., 2011; Tabuchi et al., 2011). The activity may thus induce polymer creep within the cell wall, allowing for growth or shape change.

Much of the enzyme activity data described above have come from studies on tomato fruit. Kiwifruit, a climacteric fruit in which ripening is also regulated by ethylene, is another of the typical models used for fruit softening studies, also involving, to varying degrees, the modification and degradation of cell wall polysaccharides (Ahmed and Labavitch, 1980). Softening is associated with changes in the activity of a variety of enzymes resulting in pectin and hemicellulose degradation leading to cell wall biochemical changes and fruit softening (Brummell, 2006). In contrast, there is little or no evidence for cellulose changes during kiwifruit softening (Newman and Redgwell, 2002).

Kiwifruit pectin solubilization, degradation and softening, results from the concerted action of a variety of enzymes, including PE, PG, rhamnogalacturonase (RGase), PL, β-Gal, and α-arabinofuranosidase (α-Af). These enzymes work mainly on methyl-esterfied GalA, the pectin backbone, and the galactan and arabinan side chains (Fullerton, 2015). It is thought that xyloglucan degradation may be one of the last events to occur in fruit ripening (Redgwell and Fry, 1993), and modifications of the xyloglucan structure are considered as an important aspect of kiwifruit softening (MacRae and Redgwell, 1992). The key enzyme acting on xyloglucan is XTH (Maclachlan and Brady, 1994). Although depolymerization of xylans and mannans may contribute to the loss of rigidity of cell wall; however, neither is found to any great extent in kiwifruit.

The activities of several cell wall-associated enzymes have been assayed during softening of kiwifruit. PG activity increased steadily as kiwifruit firmness decreased, while there was no significant change in β-Gal activity, and PE showed a transient increase in response to ethylene treatment (Wegrzyn and MacRae, 1992; Tavarini et al., 2009). XTH activity has also been shown to increase during kiwifruit softening (Redgwell and Fry, 1993; Atkinson et al., 2009), and EG (endo-1,4-β-glucanase), induced by propylene treatment (Bonghi et al., 1996), was also examined for its contribution to kiwifruit softening.

Ethylene (ethene) and control of cell wall enzymes during ripening

Ethylene plays a critical role in controlling ripening, particularly in climacteric fruits, and, in contrast to the traditional view, it also seems to be important in non-climacteric fruits, where ethylene biosynthesis genes and signalling components are also expressed, albeit at a much lower level. The biochemical pathway for ethylene synthesis involves a series of reactions described originally by Yang et al. (Yang and Hoffman, 1984; Grierson, 2013), where S-adenosyl methionine is converted by ACC synthase (ACS) into 1-amino cyclopropane-1-carboxylic acid synthase (ACS), which in turn is converted to ethylene by ACC oxidase (ACO) (Figure 6). Downstream of ethylene, there is a signalling pathway that involves multiple transcription factors, including ethylene response factors (ERFs), that regulate different ripening responses in fruit (Xie et al., 2016)

Recently, Mondher Bouzayen’s group has shown that there are 77 ERFs in the tomato genome, of which 27 show enhanced expression during ripening, while mRNA levels for another 28 decrease (Liu et al., 2016). This suggests that different ERFs have contrasting roles in fruit development and ripening. By examining altered ERF expression in the tomato ripening mutants rin, nor, and Nr, they went on to show that 11 ERFs are strongly down-regulated in the mutants, while three show enhanced mRNA accumulation. Three ERFs, members of sub-class E, were dramatically down-regulated in the mutants, indicating that they probably had important roles in controlling ripening events (Liu et al., 2016). Recently, some other tomato ERFs have been characterized and shown to be involved in fruit softening probably by mediating ethylene production (Chung et al., 2010; Li et al., 2007). AP2a, which belongs to the AP2/ERF subfamily, appears to be involved in repressing ethylene production, since reducing AP2a expression results in enhanced ethylene production and softer fruits (Chung et al., 2010). The roles of several ERF transcription factors identified in apple, banana, citrus, grape, kiwifruit, persimmon, and tomato, including factors regulating texture change, are shown in Figure 7, and recent work has identified several additional members of the AP2/ERF gene family, involved in controlling expression of genes related to softening in banana (Fan et al., 2016; Fu et al., 2016; Han et al., 2016).

One or more family members of each type of cell wall modifying enzyme listed in Table 1 are expressed in response to ethylene during ripening of different fruits. Examples are taken mainly from studies on tomato, melon, peach, and kiwifruit. Further information about the species, enzyme isoforms, or gene family members induced by ethylene can be obtained from Alexander and Grierson (2002), Cara and Giovannoni (2008), and Pech et al. (2008, 2012).

PG is synthesized de novo during ripening onset (Tucker and Grierson, 1982); the PG gene is transcriptionally activated during ripening (DellaPenna et al., 1989; Montgomery et al., 1993); and the mRNA is very abundant in ripening fruit. These observations suggested that this gene, and perhaps others involved in softening, may be regulated by the fruit ripening hormone, ethylene. There has been disagreement as to whether or not PG is induced by ethylene. Early work had shown that treating tomatoes with inhibitors of ethylene action inhibited PG mRNA (Davies et al., 1988, 1990; Lincoln et al., 1987) and antisense inhibition of ACC oxidase certainly reduced PG mRNA accumulation (Picton et al., 1993), whereas inhibiting ethylene synthesis by antisense-inhibition of tomato ACS2 was reported as having no effect (Oeller et al.,1991). The conclusion from the ACC synthase antisense results was inconsistent with the other results, particularly in view of the earlier finding that tomato ripening is inhibited by Ag⁺, an inhibitor of ethylene action (Davies et al., 1988, 1990), and the detailed structure of the PG promoter region (Nicholass et al., 1995) which suggested that PG is under ethylene control. This issue was resolved by Sitrit and Bennett (1998), who showed that induction of PG mRNA occurs at very low ethylene levels. This indicates that PG expression is probably under both ethylene and developmental control. Polygalacturonase activity is also induced by ethylene (and cold treatment) in other climacteric fruit such as apple (Tacken et al., 2010), and both PG and PL expressions have been shown to increase after ethylene treatment of kiwifruit (Atkinson et al., 2011). Treatment with ethylene inhibitors such as Ag⁺ or more recently 1-methylcyclopropene (1-MCP) has also been used to investigate the role of ethylene in the induction of these cell wall degrading enzymes. Zhang et al. (2011) demonstrated that 1-MCP treatment of avocado reduced PG activity. The situation is more complicated in melon, which has three PG genes. CmPG1 expression is totally dependent on ethylene, whereas regulation of CmPG2 expression is ethylene-independent and expression of CmPG3 is regulated by both ethylene-dependent and ethylene-independent factors (Pech et al., 2008, 2012). The situation in non-climacteric fruit, which generally produce little ethylene during ripening, is also interesting. Villarreal et al. (2010) demonstrated that both ethephon (to increase ethylene) and 1-MCP treatment (which inhibits ethylene perception and signalling) of the non-climacteric strawberry fruit were found to influence the expression of both PG and β-Gal, suggesting that even in this case there may be sufficient ethylene produced by the fruit to induce certain aspects of ripening such as wall degradation. Thus, ethylene would seem to play a significant role in the induction of cell wall enzyme gene expression in many fruit, of both the climacteric and non-climacteric types. However, this may not always be the case. Nardi et al. (2016) have shown that expansin expression in strawberry fruit, whilst appearing to be influenced by both ABA and auxin, was unaffected by either ethylene or 1-MCP.

Moreover, different cultivars also showed different responses to ethylene at the biochemical and molecular levels. Our recent results indicated that commercially mature ‘Hongyang’ kiwifruit (red/yellow-fleshed, Actinidia chinensis) had higher enzyme activities for β-Gal than ‘Hayward’ fruit (green-fleshed, Actinidia deliciosa); yet, β-Gal activities showed lower responses to external ethylene treatment. Highly homologous β-Gal genes, with very similar ORFs designated as AdGal5 (‘Hayward’) or AcGal5 (‘Hongyang’), show very different ethylene responses in the two cultivars, with AcGal5 unresponsive to ethylene and AdGal5 being up-regulated by ethylene treatment (Unpublished data by Yin et al.).

Inhibiting genes for cell wall degrading enzymes with antisense and RNAi gene constructs

The identification of enzyme activities in ripening fruit prompted the search for their corresponding genes. Once they had been identified, it was possible to either up-regulate (by over-expression) or down-regulate (using antisense genes, RNA-interference, or clustered regularly interspaced short palindromic repeats, or CRISPR) these genes in transgenic plants: firstly to confirm their relationship with specific enzyme activities, and in particular isoforms of the enzymes, and secondly to explore their impact on cell wall metabolism and fruit softening. Given that there were major changes observed in the pectic polymers of the fruit, early studies targeted this fraction of the cell wall. It was postulated that at least four enzymes may be responsible for the major changes in the pectin polymers observed during ripening: PG, and/or PL, being responsible for the depolymerization, PE for the increase in demethylation, and β-Gal for the loss of galactose from the sidechains (Brummell and Harpster, 2001). Much of the early investigation was carried out in tomato fruit. One of the earliest genes identified was that for PG in tomato (Grierson et al., 1986). A single cDNA was identified which when subjected to antisense-gene silencing resulted in fruit with much reduced PG activity (Sheehy et al., 1988; Smith et al., 1988), with all isoforms being impacted. The pectin isolated from these antisense fruit showed, as expected, a much-reduced depolymerization, but with no impact on the degree of esterification (Tucker et al., 1992). Interestingly, there was only a limited effect on the softening of these antisense PG fruit during the initial phases of ripening, but an effect was seen later in ripening (Kramer et al., 1992; Langley et al., 1994) and a reduction in the cracking of the fruit during transportation was also observed (Schuch et al., 1991). In contrast, silencing of PG genes in several other fruit has been found to elicit a more significant effect on texture, but not necessarily firmness, for example as seen in peach (Ghiani et al., 2011), whilst firmer fruit were observed when the PG gene was suppressed in apple (Atkinson et al., 2012). It would appear that PG activity alone is insufficient to cause softening as exemplified by Giovannoni et al. (1989) who showed that expressing PG in the rin tomato mutant was able to induce pectin degradation but was not sufficient to cause softening. This latter finding is not really surprising; however, because rin is a pleiotropic mutation and the RIN MADS-box transcription factor (Vrebalov et al., 2002) has hundreds of gene targets (Fujisawa et al., 2012, 2013), lots of other enzymes, including some that modify cell wall structure, are deficient in the mutant.

As discussed above, another enzyme activity (PL) may also be able to depolymerize the pectin and appears to play a role in the softening of strawberry (Jiménez-Bermúdez et al., 2002; Santiago-Doménech et al., 2008), although in this case even though PG activity is low it may also play a role in the determination of firmness (Quesada et al., 2009). The PL enzyme is also thought to be involved in texture change in banana fruit (Marín-Rodrígues et al., 2002) and more recently has been implicated in the softening of tomato fruit (Uluisik et al., 2016). Uluisik et al. (2016) reported that a tomato PL gene plays a critical role in fruit softening. Silencing this PL using RNAi and by CRISPR altered fruit texture without affecting other aspects of ripening and fruits retained their integrity following storage for 14 days at 20°C, indicating the potential for improved shelf life. Light and transmission electron-microscopy using specific probes showed that linear de-esterified HMG is concentrated in the tricellular junctions between cells (see Figure 5D and E) and that there was more of this material present in low PL RNAi transgenic fruit. In addition to the higher amount of HMG when PL was inhibited, PL-silenced fruits also had reduced amounts of water-soluble pectins and both observations are consistent with a role for PL in breaking down cross-linked HMG polymers in both tricellular junctions and the middle lamella. It is possible that this enables the pectic polysaccharides in the cell wall to be further degraded by enzymes such as PG.

Two cDNA clones thought to encode the major fruit-specific PE were identified in the tomato cultivars, Ailsa Craig (Hall et al., 1994) and Rutgers (Harriman et al., 1991; Turner et al., 1996a). The equivalent genes were found to be part of a small gene family, consisting of at least three genes, present in a tandem repeat (Turner et al., 1996b), one of which was a pseudogene. The major fruit-specific PE isoform in the Ailsa Craig cultivar has been silenced using antisense gene constructs, and although PE activity was reduced, there were no major differences in either fruit development or ripening detected (Hall et al., 1993). The pectin in these fruits remained more heavily esterified than wild-type controls at all stages of fruit development (Tucker et al., 1992). Tieman et al. (1992) generated similar PE antisense lines in Rutgers cultivar tomato fruit. In this case, the reduced PE activity caused an almost complete loss of tissue integrity during fruit senescence, but there was little effect on fruit firmness during ripening (Tieman et al., 1992).

A single copy gene encoding a ubiquitously expressed PE isoform has also been identified (Gaffe et al., 1996; Tiznado-Hernandez et al., 2004). This gene has been silenced and shown to encode the major ubiquitously expressed PE isoform in tomato fruit (Phan et al., 2007). Antisense silencing in this instance resulted in reduced PE activity in both leaf and mature green (MG) fruit. Again, the phenotype of these antisense fruit showed very little impact on the softening process. More recently, both the major PE isoforms in the tomato have been silenced together but again with minimal impact on softening (Wen et al., 2013)

Expression of the PE isoforms during strawberry ripening seems to follow a similar pattern to those in tomato. Strawberry PE cDNAs have been isolated from red-ripe fruit. One was found to be fruit-specific and another two were ubiquitously expressed (Castillejo et al., 2004). Three PE genes have also been identified from Valencia orange fruit (Nairn et al., 1998).

Although much research has focussed on PG and PE activities, other cell wall-modifying enzymes are likely to be important. Seven potential genes for β-Gal were originally identified in tomato fruit (Smith and Gross, 2000), whilst a more recent paper suggests that there may be as many as 17 genes in tomato (Chandrasekar and van der Hoorn, 2016). Of these, one (TBG4) has been implicated in softening in tomato (Smith et al., 2002). Silencing of a related gene in strawberry fruit has also been shown to inhibit softening (Paniagua et al., 2016). The manipulation of 1,4-β-glucanase in transgenic pepper suggested that this enzyme acts on as yet unidentified matrix glycans (Harpster et al., 2002a, 2002b), and overexpression of XTH activity in transgenic tomatoes results in increased softening (Miedes et al., 2010).

Although these antisense experiments to inhibit or over-express specific genes served to demonstrate the link between genes and individual enzyme activities and isoforms, they did not individually result in any major inhibition of softening. This led to the postulation that softening is a multi-genic trait and that silencing of several genes simultaneously would be required to have a major effect on softening. Simultaneous down-regulation of PG and PE2 was not that much more effective, however, and suppressing both PG and expansin resulted in only a small increase in firmness (Kalamaki et al., 2003; Powell et al., 2003). These, and other findings discussed above, suggest that the concurrent inhibition of PG, PE2, β-Gal, and PL may well be necessary in order to bring about a major inhibition of softening, although the effect on PL silencing alone is quite impressive (Uluisik et al., 2016).

Effects of fruit storage and post-harvest treatments on softening

The biotechnological control of ripening, which is extremely important for post-harvest physiology and biochemistry, was revolutionized by the work of the late Ed Sisler and colleagues on 1-methylcyclopropene (1-MCP) (see Reid and Staby, 2008, for a brief history). 1-MCP is a volatile cycloalkene with the molecular formula C₄H₆ and a boiling point of ~12°C. It is widely used for fruit and vegetables as an effective inhibitor of ethylene perception, to improve post-harvest storage. 1-MCP is thought to interact with ethylene receptors, thereby preventing their signalling responses (Sisler and Blankenship, 1996; Sisler and Serek, 1997, 2003; Watkins, 2006). 1-MCP has been used with many fruit and vegetables, such as apple (Fawbush et al., 2009), kiwifruit (Koukounaras and Sfakiotakis, 2007), papaya (Manenoi et al., 2007), peach (Hayama et al., 2008), pear (Trinchero et al., 2004), persimmon (Harima et al., 2003), and tomato (Wills and Ku, 2002), and not only provides a very efficient and relatively inexpensive way to maintain fruit and vegetable quality after harvest, but has also been important for understanding the mechanism of fruit ripening.

1-MCP (as EthylBloc) was approved by the U.S. Environmental Protection Agency for use on ornamental crops to prevent premature wilting, leaf yellowing, premature death, etc. 1-MCP (as SmartFresh) is used by growers and the fruit packing and transportation industries to maintain the quality of fruits and vegetables by preventing or delaying the ripening process. The use of 1-MCP in apples, kiwifruits, tomatoes, bananas, plums, persimmons, avocados, and melons has been approved and accepted for use in more than 34 countries, including the European Union and the USA (Watkins, 2006).

The effects of 1-MCP vary between different fruits, application periods, and also storage environments (Watkins and Nock, 2005), and detailed studies have shown that for different fruit (even different cultivars), the effects of 1-MCP need to be investigated and optimized (e.g. concentration, duration, etc.). The differential responses of apple and peach, both Rosaceae species, provide outstanding examples. Dal Cin et al. (2006) found that 1-MCP treatment delayed apple fruit ripening at room temperature for several days, but had very limited effects on peach fruit. In most fruit, 1-MCP is usually applied prior to, or separately from, further storage or other treatments. We found that combination of high CO₂ (95%) treatment and 1-MCP was very effective at accelerating the deastringency process, while maintaining firmness of persimmon fruit (Wang et al., 2017).

Low temperature is also frequently used to slow the rate of ripening during transport and to prolong shelf life, particularly for tomato, peach, and apricot fruits. In tomato, and probably other fruits, low temperature can affect methylation of the RIN gene, which has a detrimental effect on flavour (Zhang et al., 2016). Furthermore, chilling can reduce the softening necessary to produce an appealing product after low temperature storage and can also induce other chilling injury (Emond et al., 2005; Pan et al., 2017). In apricot, calcium treatment combined with cold storage significantly improved fruit quality and shelf life, and cell wall pectins and hemicelluloses were disassembled and degraded slowly during storage (Liu et al., 2017). In peach, transcriptomics and metabolomics analyses indicated that chilling-induced changes in lipid and cell wall metabolism-related genes could be ameliorated by low temperature conditioning (LTC), which appears to involve ethylene, and LTC was capable of ameliorating the deleterious effects of low temperature by reducing the inhibition of softening, allowing it to proceed, thereby producing a product more acceptable to consumers (Wang et al., 2017).

Controlled atmosphere (CA) storage, typically involving reduced O₂ and high CO₂, is frequently used to prolong storage life in fruits. For instance, litchi fruits stored under 1% O₂ + 5% CO₂ showed a reduction in weight loss, pericarp browning, membrane leakage, and malondialdehyde content compared with control fruit (Ali et al., 2016); peach fruits in gas concentrations of 2% O₂ and 5% CO₂ showed lower weight loss, greener ground colour, higher flesh firmness, and anthocyanin content and had a more intense characteristic taste when compared with control fruit (Cantillano et al., 2010). Usually, ethylene production is inhibited during CA storage; however, in some of these situations, trace amounts of ethylene may still influence fruit ripening, as in kiwifruit, which is one of the most ethylene-sensitive fruits, where 0.1 µl/L ethylene could accelerate softening under CA storage (McDonald and Harman, 1982).

During normal ripening, metabolite gradients can occur in relation to the in situ hypoxic areas generated inside the fruit (Pedreschi et al., 2009; Biais et al., 2010). Several studies have indicated that an anoxic atmosphere can be beneficial for post-harvest fruit quality, and high CO₂ treatment (95%) can accelerate astringency removal in persimmon fruit (Yin et al., 2012; Min et al., 2012, 2014). However, the anaerobic environment can also promote detrimental excessive softening during deastringency (Arnal and Del Río, 2004; Yin et al., 2012). This is opposite to the effects of CA (with reduced oxygen level), which normally prolongs post-harvest storage and maintenance of firmness in various fruit (Lara et al., 2011; Cukrov et al., 2016). A recent report indicated that persimmon DkXTH8 is an important regulator of fruit ripening (Han et al., 2016), which indicated a potential difference between regulation of cell wall enzymes during normal softening and anaerobic-induced softening. Further analysis of the mechanism of anaerobic enhanced persimmon fruit softening indicated that three ethylene response factor genes (DkERF8/16/19) and eight cell wall metabolism genes (Dkβ-gal1/4, DkEGase1, DkPE1/2, DkPG1, DkXTH9/10) were involved (Wang et al., 2017).

Several fruits, such as loquat and mangosteen, undergo post-harvest hardening, which has the opposite effect to softening. This has been most intensively studied in Loquat (Eriobotrya japonica Lindl.), which is a subtropical fruit native to China, but it also occurs in some pears. Post-harvest ripening in loquat is characterized by an unusual increase in firmness and toughness of the flesh resulting in a decrease in juiciness (Ding et al., 2002; Cai et al., 2006a). Recent analysis suggested that the increase in firmness is a consequence of tissue lignification, which can be trigged by chilling. Another factor appears to be ethylene, despite the fact that loquat has been reported to be a non-climacteric fruit, despite the fact that traditionally they are not considered to be responsive to ethylene. This post-harvest-hardening can be alleviated by low-temperature conditioning or heat treatment (Cai et al., 2006b; Zeng et al., 2015). In loquats, the activities of enzymes in the phenylpropanoid pathway, leading to lignin formation, such as phenylalanine ammonia lyase (PAL), 4-coumarate: coenzyme A ligase (4CL), caffeoyl-coenzyme A 3-O-methyltransferase (CcoAOMT), and cinnamyl alcohol dehydrogenase (CAD), show a positive correlation with fruit lignification (Shan et al., 2008; Liu et al., 2015, 2016), as did other genes such as EjCAD1 (Shan et al., 2008), EjCcoAOMT (Liu et al., 2015), and Ej4CL1 (Liu et al., 2016).

In recent years, the transcriptional regulatory mechanisms controlling lignification have begun to be understood through the characterisation of transcription factors that modulate the expression of genes involved in the process. Two novel EjMYBs, EjMYB1 and EjMYB2, were shown to have opposing roles in regulation of loquat lignification during different post-harvest treatments, through their competitive interaction with AC elements in their target gene promoter regions (Xu et al., 2014). Subsequently, EjMYB8 was isolated and its active role in lignification confirmed by means of transient over-expression in both N. tabacum and loquat leaves, which increased lignin content (Wang et al., 2016). Interestingly, EjMYB8 was able to interact with EjMYB1 to enhance its stimulated effect on their target promoter (Wang et al., 2016). Another partner of both EjMYB1 and EjMYB2 was identified as an AP2/ERF gene family member named EjAP2-1. Isolated from a group of 18 AP2/ERF members, EjAP2-1 was verified as an indirect transcriptional repressor of lignin biosynthesis, with EAR motifs in its structure being involved in protein-protein interaction with EjMYBs (Zeng et al., 2015).

Cell wall changes can have a major effect on the ease with which pathogens can infect fruits. Fruits of the tomato non-ripening mutants rin, nor, and Nr do not ripen fully and are very resistant to rotting, even after 1 year. Normal fruit, however, start to rot a few days after reaching full ripeness. This is partly because the cell wall solubilization and degradation that occur as fruit soften make it easier for some pathogens to infect the fruit. Thus, rotting is an inevitable consequence of ripening and the fruits become susceptible to invasion by pathogens, and in evolutionary terms, rotting may be considered ‘escape plan B’ for the seeds. Dynamic interactions between pathogens and plant cell walls have been reviewed by Bellincampi et al. (2014).

There are also other factors at work, however. Spores of the fungus anthracnose (Colletotrichum gleosporoides) become attached to the skin of a range of fruits when they are unripe. The fungal spores form an infection peg, which firmly fixes them to the outside of the fruits, but they do not invade until the fruits start to ripen. At the onset of ripening, three things happen; first, the fruit produces ethylene, which is detected by the fungus and the spores start to germinate (Swinburne, 1983; Jefferies et al., 1990). Second, softening enzymes are produced by the fruit, as described above, which makes it easier for the fungus to penetrate the fruit. Third, the fungus produces a battery of wall-modifying enzymes that cause further digestion of the fruit cell walls. Some of these enzymes are similar in catalytic activity and function to those produced by the fruit cells, but in other cases the fungal enzymes are more vigorous and cause destruction of the cell wall rather than just remodelling. Infection by Colletotrichum gloeosporioides of ACO-antisense tomatoes, engineered to produce less than 5% the normal level of ethylene, progressed more slowly than in the controls, and infection of rin fruit was greatly reduced compared with controls, but could be stimulated by supplying ethylene externally (Cooper et al., 1998). Ethylene also influences the course of infection by bacterial pathogens, and fruit of the (ethylene-insensitive) Nr mutant showed less-severe symptoms than wild type when infected by the fungus Fusarium oxysporum f.sp. lycopersici or by the bacteria Xanthomonas campestris pv. vesicatoria and Pseudomonas syringae pv. Tomato bacteria (Lund et al., 1998; Ciardi et al., 2000).

Conclusion

Experiments over the last 30 years support a general molecular model of ripening. Although much of this work has been carried out in the tomato, it has now been extended to many other fruits. The general features appear to be similar, but there are species and cultivar differences and the detailed aspects of the process vary in different fruits. The time is fast approaching when we will have sufficient knowledge to control texture and manipulate specific ripening attributes through biotechnology, without adversely affecting other aspects of quality.

Conflict of interest statement: None declared.

Author contributions

GAT wrote the account of cell wall structure and function of cell wall modifying enzymes; Xueren Yin, Aidi Zhang, MiaoMiao Wang, Qingang Zhu, Xiaofen Liu, Xiulan Xie, and Kunsong Chen contributed the sections on post-harvest texture changes and ERFs in fruit; DG designed the scope of the review, wrote several parts, and edited the final version.

Acknowledgements

We thank Philip Crookes for Figure 5A, B, and C from his PhD thesis (University of Nottingham, 1985), Graham Seymour for Figure 6D and E (see Uluisik et al., 2016), and Gong Ziyuan for drawing Figure 7.

References

Footnotes