Flowers and Yeo (1995) suggested five possible ways, which wereappropriate at that time, to develop salt-tolerant crops: (1)develop halophytes as alternative crops; (2) use interspecifichybridization to raise the tolerance of current crops; (3) usethe variation already present in existing crops; (4) generatevariation within existing crops by using recurrent selection,mutagenesis or tissue culture, and (5) breed for yield ratherthan tolerance. These all remain possible solutions to the problem.Although conventional forms of mutagenesis have not, in general,delivered salt-tolerant genotypes (Flowers and Yeo, 1995; butsee Tester and Davenport, 2003), mutagenesis has unearthed anumber of salt-sensitive types (Borsani et al., 2002; Zhu, 2002).Bohnert and Jensen (1996) claimed that an important approachhad been missed by Flowers and Yeo: they wrote ‘tolerancebreeding must be accompanied by transformation’; and that‘successful releases of tolerant crops will require large-scale"metabolic engineering" which must include the transfer of manygenes’. While such an approach was not feasible in theearly 1990s (Flowers and Yeo, 1996); this approach is now beingwidely advocated. Some 13 species (Table 1A) have been transformedwith nearly 40 genes in experiments reported between 1993 and2003 (Table 1B). The majority of experiments have used rice,tobacco and arabidopsis; transformations involving the synthesisof compatible solutes have been more popular than any other,with those involving glycine betaine the most commonly performed(Table 1A). There is an increasing number of claims in thisliterature that overall tolerance can be manipulated throughalteration in the activity of one or two genes (see below),which was not something claimed by Bohnert and Jensen (1996).For a trait as complex as salt tolerance this seems intuitivelyunlikely. The fundamental issue to be resolved is the importanceof individual components or sub-traits of salt tolerance andwhether the manipulation of individual or of many genes is requiredto alter complex traits. If altering a single gene can altertolerance, this suggests either that changing the concentrationof a few key components has a substantial effect on a wide rangeof other processes or that salt tolerance is not as complexas it appears or that a key limit to tolerance might be alteredin any given species (or genotype). Substantiating, or otherwise,claims that tolerance is altered by transformation is clearlyof major importance both for our understanding of complex traitsand for the practicalities of their manipulation.
Genetic engineering of salt tolerance: evaluation of success
The evaluation of transgenic material requires some comment.The material to be tested should be genetically stable (it hasbeen suggested that it should be in its fourth or fifth generationby Bajaj et al., 1999) and a comparison of as many transformedlines as possible made with the performance of the parental(wild-type) line under saline and non-saline conditions (Table2). It is important to know whether or not the overall growthof the transgenic plant has been affected, as vigour itselfis an important determinant of salt tolerance. For crops, claimsof enhanced tolerance should be made on the basis of yield.Unfortunately, there were no such reports by 1999 (Bajaj etal., 1999) and the situation had changed little by early 2003.Given the paucity of data on crop yield (just five reports ofestimates of crop yields, Guo et al., 1997; Wang et al., 2000;Zhang and Blumwald, 2001; Zhang et al., 2001; Li et al., 2002),the success, or otherwise, of a transformation in altering salttolerance has generally to be evaluated against the nature ofthe data that is presented. Those claims based on quantitativeestimates of the growth of fourth or fifth generation transgeniclines should be seen as stronger than claims based upon photographicevidence of the performance of plants of the primary transformantsgrown in salt alone. In the following analysis, papers are allocatedto one of five categories (Table 3). Only those data relatingto the growth of plants under conditions in which transpirationoccurs have been evaluated: it is transpiration that transportsions to the shoots, where their presence brings about injuryand death. Photographs of plants in culture medium are unconvincingas evidence for a successful alteration of crop yield. Quantitativemeasures of growth are required for plants grown in the presenceand absence of salt: the ability to germinate in salt is, ingeneral, a poor indicator of performance in the field. It isalso important that salt be added in such a way that it is notthe effect of water or osmotic stress that is being evaluatedand this generally requires an increase of salt concentrationof 50 mM or less per day and determination of the consequencesdays or weeks later, depending on the salt tolerance of thespecies (Munns, 1993, 2002).
Analysis of publications to date shows that of the 68 reportsproduced between 1993 and early 2003 (Table 3; see also supplementarydata online) only 19 describe quantitative estimates of plantgrowth. Of these, four papers (Table 4) contain quantitativedata on the response of transformants and wild type of six specieswithout and with salinity applied in an appropriate manner.About half of all the papers (35, Table 3 and supplementarydata online) report data on experiments conducted under conditionswhere there is little or no transpiration: such experimentsmay provide insights into components of tolerance, but are notgrounds for claims of enhanced tolerance at the whole plantlevel—in such a system, the fern Ceratopteris, where singlegene mutants alter the salt tolerance in the gametophytic generation(Warne et al., 1995) might be a useful genetic model.
Those experiments where the effects of transformation were determinedin saline and non-saline soil or hydroponic culture suggestthat real changes in salt tolerance can be effected, but generallynot without consequences for the growth that occurs in the absenceof salt. Over-expression of the gene Alfin1 in alfalfa increasedits salt tolerance and promoted root growth and shoot growth(Table 4A), under normal and saline conditions, producing largerplants than the wild type (Winicov, 2000). Alfin1 is a putativetranscription factor, but its mode of action in altering overallsalt tolerance is still unclear. Adding to the uncertainty ofhow some genes affect overall tolerance is the consequence oftransforming tomato with the yeast gene HAL1. HAL1 alters thesalt tolerance of tomato (Gisbert et al., 2000) and increasesthe K/Na ratio in transgenic plants. However, these transgenicplants, when grown in the absence of salt, had half the shootdry weight of the wild type (Table 4A). A similar effect ofan introduced gene on growth is also seen following the transformationof tobacco with mannitol-1-phosphate dehydrogenase. Here mannitolthat accumulated as a consequence of the transformation madeonly a small contribution to the osmotic potential of the transformedplants, which were smaller than the wild-type, although theywere less affected in relative terms by salinity (Table 4A).A similar situation was reported by Huang et al. (2000) forarabidopsis, canola and tobacco transformed to oxidize cholineto glycinebetaine (Table 4A). In all of these cases, the effectsof the genes are not simply on tolerance, making the evaluationof the effects complex. Tolerance, judged in relative terms(i.e. yield in the saline conditions expressed as a proportionof yield in non-saline conditions), although an important indicator,is unlikely to impress a farmer unless the absolute yield isadequate. A genotype whose yield is hardly affected by salinitymay well still be out-performed by a vigorous, high-yieldinggenotype which loses 50% of its yield under saline conditions,if the ‘salt-tolerant’ genotype is intrinsicallylow-yielding (Dewey, 1962). A similar situation has been previouslyreported for some hybrids between established crop varietiesand wild relatives (Table 4B).
There is other, albeit weaker, evidence that transformationof plants with genes whose products affect transcription doesappear to alter salt tolerance. Tobacco transformed with a geneisolated from a cDNA library prepared from salt-treated plantsof Atriplex hortensis, by screening with a fragment of arabidopsisAtDREB2A encoding a DNA-binding domain, apparently increasedthe tolerance of tobacco to salt (Shen et al., 2003) withoutcausing a dwarf phenotype, although quantitative data are lacking.A yeast kinase (a functional homologue of the yeast Dbf2 kinase)enhanced tolerance of tobacco cells to salt in tissue culture(Lee et al., 1999). However, the assessment of salt tolerancein tissue culture is a poor predictor of tolerance in the wholeplant and tolerance in cultured cells is not translated to tolerancein plants in the field (Flowers and Yeo, 1995). Transformationof arabidopsis with the protein kinase coded by AtGSK1 (a GSK/Shaggy-likeprotein kinase) induced anthocyanin synthesis, a symptom ofNaCl stress, in the absence of NaCl (Piao et al., 2001) andpromoted survival in soil irrigated with NaCl (but with 300mM, which is likely to have produced an initial osmotic shockrather than a salt stress).
Recent research has shown that rice, transformed to overexpressgenes that brought about the synthesis of trehalose, containeda reduced concentration of Na in the shoot and grew better thannon-transformed (control) plants when in the presence of 100mM NaCl (Garg et al., 2002). Trehalose concentrations in theplants grown in 100 mM NaCl were relatively low, at less than0.1 mg g–1 fresh weight (approximately 5 mM in the cytoplasmif this were 10% of the water volume; Flowers et al., 1991).It has been suggested that solutes such as trehalose are likelyto function through their ability to scavenge reactive oxygenspecies (Zhu, 2001) and the protection afforded to the machineryof protein synthesis may be particularly important for normalrepair processes (Chen and Murata, 2002). Another possible explanationfor at least some of the compounds such as trehalose is thatthey act in a signalling cascade. Although trehalose is commonlypresent in bacteria, fungi and insects, its concentration inplants is very low and it may even be toxic: recent evidencesuggests that this toxicity may stem from its role in the regulationof carbon metabolism (Muller et al., 1999; Wingler, 2002). Earlierresearch on tobacco had shown that transformants producing trehalosewere stunted in growth (Romero et al., 1997) and experimentson rice had shown that treatment of plants with exogenous trehalosereduced sodium accumulation, but had a significant effect onroot morphology (Garcia et al., 1997b). In the more recent experimentof Garg et al. (2002), the synthesis of trehalose was underthe influence of a stress-inducible promoter, so that growthunder control conditions was presumably no different from thewild type (the authors note that non-stressed plants appearednormal, but did not, unfortunately, support this with quantitativedata). The use of stress-inducible promoters may be an importantway in which to avoid inhibition under non-stressed conditions(Kasuga et al., 1999), if there are yield penalties from expressinggenes under a constitutive promoter.
Although the targets of genetic engineering have largely beencompatible solutes, there have been some attempts to manipulateone of the underlying causes of salt damage, the net accumulationof sodium ions. Down-regulation of HKT1 in wheat increased resistanceto salinity under conditions of low K supply (Laurie et al.,2002) and transformation of the cyanobacterium Synechococcuswith a Na/H antiporter increased its tolerance to salt (Waditeeet al., 2002). For higher plants, however, any enhancement ofantiporter activity would have to be targeted to root cells,for in the aerial parts of the plant enhanced Na/H antiporteractivity would only exacerbate the consequences of ion accumulationin cell walls (Oertli, 1968; Flowers et al., 1991; Munns, 2002).Even in the roots, it is likely that ions removed from corticalcell walls would have osmotic consequences (Yeo, 1998): onlyin situations where there was a large volume of external solution(e.g. marine algae) would there be a chance of ions effluxingfrom the cytoplasm being washed from cell walls. Manipulatingthe vacuolar proton gradient to enhance ion accumulation hasalso led to claims of enhanced salt tolerance in transgenicplants (Gaxiola et al., 2002). However, there is only qualitativeevidence for Arabidopsis (Apse et al., 1999) and the evidenceobtained with Brassica napus (Zhang et al., 2001) and tomato(Zhang and Blumwald, 2001) does not include (other than a photograph)the effects of salt (200 mM) on the wild type. In other experiments,B. napus continued to yield in 200 mM NaCl (Ashraf et al., 2001)as does the tomato cultivar Moneymaker (TJ Flowers and SA Flowers,unpublished data). While the ability to accumulate sodium inleaf vacuoles is clearly a trait that is important for dicotyledonoushalophytes, in such species this ability is coupled with othertraits such as the regulation of transpiration, the synthesisof compatible solutes and an ability to function with low cytoplasmicpotassium concentrations (Flowers and Dalmond, 1992).
Conventional breeding programmes
Strategies for breeding for salt tolerance in cross-pollinatingspecies by cycles of recurrent selection were described longago (Dewey, 1962): for a self-pollinating species the same processwould require the use of male-sterile lines to facilitate out-crossing(Ramage, 1980). These approaches depend on adequate heritabilityof the overall trait, for which there is evidence for wild grasses(Ashraf et al., 1986), sorghum (0.74, Maiti et al., 1994), maize(0.4, Maiti et al., 1996), and tomato (Saranga et al., 1992,1993). In both rice (Yeo et al., 1988) and Trifolium (Rogers and Noble, 1992;Rogers et al., 1997), it has proved possibleto select lines whose ion contents, when grown under salineconditions, are either higher or lower than those of the parentaltypes. By way of contrast, Saranga et al. (1992) concluded thatfor tomato (a cross between L. esculentum and L. pennellii),selection for ion contents would not improve the breeding process.
Use of in vitro selection
The use of in vitro selection was widely advocated during the1980s, but did not result in cultivars in farmers’ fields(Rowland et al., 1989). More recently, selections for alfalfa(Winicov, 1991; Winicov and Bastola, 1997) look promising andthere may be a use for somaclonal variants within breeding programmes(Zhu et al., 2000).
Pooling physiological traits
The possibility of pooling physiological traits has been advocatedfor rice (Yeo et al., 1990), screening methods evaluated (Garciaet al., 1995) and the approach proved successful in generatingsalt-resistant lines (Gregorio et al., 2002). The methodologydoes not require a deep knowledge of the genetics of traits,merely that they display sufficient heritability and that suitablescreening procedures can be developed. The methods may be applicableto crops other than rice (Cuartero et al., 1992; Ellis et al.,1997; Foolad, 1997; Isla et al., 1998; Munns et al., 2002).
Interspecific hybridization
The introduction of genes from wild salt-tolerant species hasbeen explored for tomato (Rush and Epstein, 1981; Tal and Shannon, 1983;Saranga et al., 1991; Perez Alfocea et al., 1994), tomato/potato(Sherraf et al., 1994), wheat (Dvorak and Ross, 1986; Gorhamet al., 1986; Mahmood and Quarrie, 1993; Martin et al., 1993;William and Mujeebkazi, 1993; King et al., 1997a, b), and pigeonpea(Subbarao et al., 1990). However, the approach has not led tothe release of salt-tolerant crops, although there is a recentproposal for a new salt-tolerant cereal, tritipyrum (King etal., 1997b),
Halophytes as alternative crops
Historical evidence suggests that farmers shift from more sensitiveto more tolerant crops as salinity in their fields rises (Jacobsen and Adams, 1958).The natural end of such a succession wouldbe the use of halophytes, whose potential as crops has beenexplored (Malcolm, 1969; O’Leary, 1984; O’Learyet al., 1985; Lovett, 1993; Troyodieguez et al., 1994; Zahran, 1994;Brown and Glenn, 1999; Brown et al., 1999; Glenn et al.,1999), but is yet to be fully realized. Since the domesticationof wild species was, in the past, a successful strategy, thismust remain a useful approach for generating salt-tolerant cropsin the future, especially given the wide range of halophytesavailable.
Use of marker-aided selection
The multigenic nature of salt tolerance has clearly been establishedand quantitative trait loci associated with aspects of germination,ion transport and yield. One obvious use of QTL in plant breedingfor salt tolerance is in marker-aided selection (or marker-assistedselection, MAS). The drawbacks in using marker-assisted breedingare ‘linkage drag’ of undesirable traits due tothe large size of regions of chromosomes identified by QTL (Asins, 2002)and the fact that environment and genetic background havea significant influence on the QTL that are identified (seeabove). In a wider context, QTL might be used to identify genesthat are important in salt tolerance and it is noteworthy, giventhe complexity of salt tolerance, that so few QTL are identified(Yeo et al., 2000) within any given genome. This may be an indicationthat traits are determined by a limited number of sites and/orthat genes associated with physiological traits are clusteredon chromosomes. However, the fact that a QTL represents many,perhaps hundreds, of genes remains a problem to finding keyloci within a QTL. The easiest way forward may be through theidentification of candidate genes. Of the five QTL associatedwith the effects of salinity on vegetative growth in arabidopsis(Quesada et al., 2002) one was located close to the locationof SOS2 (which codes for a serine/threonine protein kinase)and another close to the positions of RD29A and RD29B (genescoding for hydrophilic proteins involved in ABA signal transduction).
Answers to all your Biology Questions
