VI. Genetic Transformation in Plants

62



W.R. SCOWCROFT



phase chromosomes from a different species, and enzymes specified by the

donor can be detected in the treated recipient cells (Burch and McBride, 1975).

In plants both observational and experimental evidence indicate that transformationlike phenomena do occur. Most of these studies have involved whole plants,

and these will be reviewed briefly because of their possible relevance to the

development of novel techniques for genetic modification in plants.



A. MODIFICATION BY HOMOLOGOUS DNA



The treatment of mutant plants with wild-type DNA can be associated with

the correction of the lesion at a frequency which is significantly greater than

spontaneous correction in the appropriate control plants. The observations

which have been reported include the correction of anthocyanin deficiency in

petals of Petunia (Hess, 1969), the modification of the effects of the waxy locus

in pollen of barley (Turbin er af., 1975), and the modification of genetically

determined fruiting characters in Capsicum annuum following DNA treatment

(Nawa et al., 1975). This latter observation was similar to graft-induced genetic

alteration observed in red pepper (Ohta and van Chuong, 1975). Pandey (1975)

has also observed gene transfer following the use of “lulled” irradiated pollen to

overcome intraspecific incompatibility in Nicotiana hybrids. Genes which were

apparently transferred from the mentor pollen parent modified the incompatibility genotype or the flower color that characterized the maternal parent. Among

the many reasons Pandey (1975) advanced to exclude rare pollen mentor nuclei,

which might have escaped the very high lethal dose of irradiation, as the cause of

such rare genetic events, is that the transformed plants were otherwise phenotypically maternal and distinctively different from the expected hybrid.



B. MODIFICATION BY HETEROLOGOUS DNA



The uptake, integration, and possible transforming ability of foreign DNA has

been studied using seedlings, seeds, cultured cells, and protoplasts (Ledoux,

1975; Markham et al., 1975). For some years now Ledoux and co-workers (see

Ledoux, 1975) have examined the uptake and integration of bacterial DNA

following its application to germinating seedlings. The evidence for the pseudointegration (covalent binding) of the foreign DNA with that of the host DNA is

based on the occurrence of DNA, isolated from the treated plants, with a

buoyant density intermediate between that of the higher density of the plant

DNA and the lower density of the donor bacterial DNA. This intermediate form

subsequently could be separated into components which corresponded to the

respective buoyant densities of recipient and donor DNA. It has also been



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



63



claimed that a thamine deficiency in Arabidopsis can be corrected with DNA

isolated from bacteria prototrophic for thiamine (see Ledoux et al., cited in

Ledoux, 1975). However, Ledoux’s buoyant density evidence has not been

confirmed. Kleinhofs et al. (1975) using sedimentation analysis were unable to

confirm Ledoux’s findings. No intermediate peak was found when axenic plants

of pea, tomato, and barley were treated with foreign DNA according to

Ledoux’s procedure. Bacterial DNA covalently bound to recipient DNA was

found when the plants were not treated axenically. Kleinhofs et al. (1975)

argue that the observations of Ledoux are artifactual, resulting from contamination by other bacteria. Lurquin and Hotta (1975) treated Arabidopsis cell

cultures with bacterial DNA but were unable t o find any evidence for the

intracellular presence of bacterial DNA sequences, either in an integrated or in a

free state. With the development of techniques for the characterization of DNA

it has emerged that analysis based solely on buoyant density sedimentation is

not a sufficient criterion to identify the origin of the DNA. It is essential that

base sequence homology be established by DNA hybridization techniques such

as can be done on nitrocellulose filters or in solution. The fidelity of base pairing

can only be established by thermal renaturation studies. Using these more

sensitive techniques, Kleinhofs (cited in Ledoux, 1975) was still unable to

demonstrate the presence of donor DNA sequences in the DNA of treated barley

seedlings. Therefore, until further evidence is provided, the results of Ledoux

claiming integration of bacterial DNA into plant DNA must be viewed cautiously.

Plant cell cultures and protoplasts have been used to study the uptake,

expression, and possible integration of foreign DNA. Uptake and maintenance of

integrity of foreign DNA has been reported in plant protoplasts (Vasil, 1976). In

studies such as this, plant nucleases present a real hazard to the integrity of the

donor DNA. Moreover, the enzyme preparations used to prepare protoplasts also

have considerable nuclease activity (Langridge, personal communication). Nuclease activity, particularly exonuclease, is pH-dependent and activity is substantially impaired at pH 9-10. Another precaution to avoid nuclease activity is to

thoroughly wash plant cells or protoplasts prior to treatment.

Studies on the expression and possible integration of bacterial DNA in plant

cells is at best equivocal. Carlson (1973) infected barley protoplasts with a

bacteriophage of Escherichia coli and was able to detect two phage-specific

enzymes, S-adenosylmethionine cleaving enzyme and RNA polymerase. The

expression of these functions was rapid and transitory. Analogous work with

haploid cell cultures of tomato and Arabidopsis suggested that bacterial genetic

information for the utilization of lactose as a carbon source could be transferred

to plant cells using a transducing phage as vector (Doy et al., 1973). Apparent

expression of the transferred genetic information enabled the plant cells to grow

on lactose, whereas untreated cells could not, and immunological studies on the



64



W.R. SCOWCROFT



surviving plant cells indicated that the P-galactosidase was of bacterial, rather

than plant, origin. Similar work with the same phage, but with sycamore cells,

was initially confirmatory (Johnson et al., 1973). However, subsequent experiments (Smith et al., 1975) have been unable to demonstrate bacterial P-galactosidase. The growth response of the treated sycamore plant cells showed an initial

burst of cell division which was not maintained. Smith et al. (1975) doubt that

their results provide direct evidence for the uptake and expression of bacterial

genetic information.

Research in our own laboratory by Dr. V. E. Merriam has also examined the

expression of bacterial genes in plant cells, and while some success has been

achieved, the results are still equivocal. Tobacco cells, which are sensitive to

growth inhibition by the antibiotic kanamycin, were treated with plasmid DNA

from Escherichia coli which carried kanamycin resistance. Plant cells were

treated under conditions which minimized the degradation of the plasmid by

plant nucleases and also included concomitant low levels of irradiation, hopefully

to aid integration of the plasmid. Two types of resistant clones were recoveredthose which survived for only a few subcultures in the presence of kanamycin,

and those which have been serially transferred many times with no apparent loss

of resistance. The majority of these stable clones were derived from experiments

where cells were irradiated following exposure to plasmid DNA. Spontaneous

kanamycin-resistant mutants have also been recovered. However, in experiments

using similar numbers of tobacco cells, the frequency of resistant clones was

higher using plasmid DNA than in the controls where plasmid DNA was excluded or replaced by nonplasmid DNA. Similar results have also been reported

following treatment of the green alga Chlamydomonasreinhardtii with bacterial

plasmid DNA and selection for kanamycin resistance (Gresshoff and Hess,

1977).

None of the observations or studies reported in the preceding paragraphs

provide unequivocal evidence that foreign DNA can be utilized to genetically

modify plants. However, other somewhat more sophisticated evidence indicates

that DNA of bacterial origin can be replicated in (Ganem et al., 1976) and

translated by (Wang et al., 1976) eukaryote systems and, as mentioned earlier,

the stable transfer and expression of genetic information of isolated metaphase

chromosomes into mammalian cell cultures of a different species has been

achieved. In plants it would seem that the technology has not yet developed to

effect the transfer and subsequently detect the presence of foreign DNA. The

solution may come from the very recent, and indeed exciting, developments

commonly referred to as “genetic engineering.” In the sense of genetic modification by hybridization, selection, and mutation, genetic engineering has been

widely practiced by plant breeders. These new techniques however involve the

isolation and restructuring of DNA and the reinsertion into a cellular environment where indeed it will function. These techniques are being currently



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



65



explored as a means of generating genetic variation for use in plant improvement. It is appropriate therefore to consider them in this review.



C. MOLECULAR GENE MANIPULATION



The discovery and utilization of two natural phenomena in bacteria have made

the in vitro rearrangement of DNA possible. First, circles of double-stranded

DNA were found which replicated independently of the bacterial chromosome.

These extrachromosomal particles are called plasmids and they endow their host

with the capacity to adapt to new environments by conferring properties such as

drug or heavy metal resistance, the ability to metabolize long-chain hydrocarbons, and the ability to transfer genetic material by conjugation. Plasmids can

also be integrated into the bacterial chromosome. Some plasmids have a wide

host range and so genetic information can be transferred interspecifically.

Second, was the isolation of a class of bacterial enzymes, restriction endonucleases, which recognize a particular short sequence of DNA and cleave the

double-stranded DNA within this sequence. The enzymatic cut leaves singlestranded ends which are complementary. These endonucleases are normally

produced for degrading foreign DNA which may happen to enter the cell.

Concomitantly, a bacterium can modify its own DNA so that it will not be

degraded when nucleases are produced by that cell. Provided the specific

sequence is present and unmodified, the restriction nuclease will cut the DNA

no matter what its origin, be it prokaryote or eukaryote. Because the singlestrand ends of the staggered cut are complementary, an endonuclease fragment

from one species can be annealed with a fragment from another species that has

been degraded by the same endonuclease to form a ring which can be covalently

closed by incubating in appropriate enzymes (see Cohen, 1975). Hybrid DNA

molecules that have been produced to date usually contain a fragment from a

bacterial plasmid. This fragment has a very specific function, namely, a replication sequence, or replicon, i.e., a nucleotide sequence to which DNA polymerase

can attach so that the hybrid DNA molecule may be replicated. This plasmid

usually carries another gene, e.g., drug resistance, so that bacteria which are then

transformed by the hybrid molecule can be selected. In this way the hybrid

molecule containing the foreign DNA can be multiplied indefinitely and large

quantities of it can be purified.

Using this procedure of DNA fragmentation by endonuclease and hybridization with plasmid DNA, DNA sequences of Drosophila have been multiplied in

Escherichiu coli (Wensink el al., 1974). Some eukaryote DNA sequences specifying a particular function can be isolated, either because of the unique characteristics of that DNA, e.g., ribosomal DNA, or because a specific probe, e.g.,

mRNA, can be used to isolate the particular gene. These nucleotide sequences



66



W. R. SCOWCROFT



may also be hybridized with plasmid DNA and multiplied in bacteria. For

example, this has been done with the genes of the toad Xenopus laevis that

specify ribosomal RNA (Morrow et al., 1974), the histone gene from sea urchin

(Kedes et al., 1975), and the gene from rabbit that specifies &globulin synthesis

(Rougeon et al., 1975). While the replication sequence of plasmids is probably

the most versatile, other such sequences have been used. For example, Struhl et

ul. (1976) have hybridized the replication sequence of the bacteriophage h with

yeast DNA, and this hybrid molecule multiplies when inserted into bacterial

cells.

In applying these techniques to transfer specific genetic information t o higher

organisms, replication sequences must be found which will permit the hybrid

molecule to multiply in the eukaryote cell. The purified DNA of the mammalian

virus, SV40, can transform mammalian cells. This has been joined to phage h

DNA and the hybrid molecule can be multiplied in monkey cells (Ganem et al.,

1976).

Because of our interest in developing new techniques for generating genetic

variability of potential value to plant improvement, we are currently examining

the application of DNA hybridization techniques to plants (Langridge, 1977;

Langridge and Scowcroft, 1977). Plant cell and protoplast culture provide a

convenient experimental system since it has been established that plant cells,

particularly protoplasts, can take up macromolecules such as viral particles and

viral RNA which will multiply in the cultured protoplasts. It is more difficult

however to find the appropriate vector since the replication sequences that are

adapted to multiplication in plants tend to be limited. There are a few DNA

molecules which may provide a replication sequence which can be utilized for

the purposes of genetic engineering in plants.



I . Possible Molecular Vectors for Gene Transfer in Plants

The first of the molecular vectors for gene transfer is the DNA of plant viruses.

Most plant viruses are RNA, either single- or double-stranded, and a few have

double-stranded DNA. The latter class comprise the caulimoviruses, of which the

best characterized one is cauliflower mosaic virus (CaMV) (Shepherd; 1976).

Recently, the unrelated potato leafroll virus has also been classified as a doublestranded DNA virus (Sarkar, 1976). CaMV has a molecular weight of 4.7 X lo6

and the purified DNA is infective in plants, but as yet has not been shown to

multiply following “infection” of protoplast cultures.

Chloroplasts and mitochondrial DNA may also provide a suitable replication

sequence. They have respective molecular weights of 90 X lo6 and 40 X l o 6 .

Current research is attempting to isolate the replication sequence of the chloroplast DNA.

A third possible molecular vector is a plasmid of Agrobacteriurn rurnefaciens.

This bacterium is a plant pathogen responsible for a neoplastic disease, crown



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



67



gall. The plasmid of A . tumefuciens may represent the first documented case of a

natural example where genetic information is tranferred from bacteria to plants,

and in the context of this review warrants further discussion.

Infection of wounded plants by A . tumefuciens leads to the transformation of

the host cells to the neoplastic, crown gall state. Once transformation has been

established the neoplastic growth is self-proliferating in the absence of the

inciting bacterium. Ever since this fact was firmly established, considerable

effort has been devoted to attempting to understand the molecular nature of the

tumor-inducing principle (TIP) (see recent reviews by Drlica and Kado, 1975;

Lippincott and Lippincott, 1975). The evidence for a transmissible TIP, although

indirect, is compelling. Secondary tumors appear on some infected plant species

and these may be free of the inciting bacterium. Bacteria-free crown gall cell

cultures can also induce nonself-limiting growth when grafted onto normal,

healthy plants.

The experimental evidence that bacterial DNA, RNA, or bacteriophage DNA

of A . tumefaciens was tumorigenic has been challenged and, indeed, unconfirmed by others (for discussion and appropriate references, see Drlica and Kado,

1975; Lippincott and Lippincott, 1975). Because of the lack of an unequivocal

assay for the phenomenon of transformation, most recent studies on the role of

bacterial or phage nucleic acids in tumor induction have utilized nucleic acid

hybridization techniques. The initial experiments which claimed sequence

homologies between tumor DNA and bacterial or phage DNA were based on the

hybridization of tumor cell DNA with RNA complementary to A . tumefuciens

DNA or PS8 phage DNA. Confirmatory evidence that A. tumefaciens sequences

were represented in tumor cell DNA was based on DNA-DNA filter hybridization experiments. However, in these studies the fidelity of base pairing, which

can be evaluated by examining the thermal stability of the presumptive DNADNA duplexes, was not determined. When this was done, it was found that no

more than 0.02% of the crown gall genome could be homologous with A .

tumefuciens DNA. This amounts to less than one bacterial genome per diploid

tumor cell. Similarly, little or no homology was found between bacteriophage

PS8 DNA and A . tumefuciens DNA.

There is now compelling evidence that a large plasmid is involved in the

tumor-inducing process. There is a high, though not absolute, correlation between virulence and the possession of a large plasmid (Zaenen er d., 1974). The

early demonstration of the transfer of virulence to avirulent Agrobucterium

strains by an unknown mechanism of genetic exchange in plunru (Kerr, 1969),

now appears to be due to plasmid transfer (van Larabeke er ul., 1974; Watson et

ul., 1975). Moreover, many virulent strains lost oncogenicity following growth at

high temperatures (Watson er ul., 1975; Bomhoff et ul., 1976) and this was

correlated with loss of a large plasmid. Different strains of A. tumefuciens carry

different plasmids, and these vary in molecular weight from 96 X lo6 t o I56 X

lo6 (Zaenen et uL, 1974). From DNA hybridization studies Matthysse and



68



W.R. SCOWCROFT



Stump (1976) suggest that approximately 0.1% of bacteria-free tumor cell DNA

is A. tumefaciens plasmid DNA. This amounts to approximately 10 plasmids per

diploid tumor cell.

In addition to virulence there are a number of other properties which now

appear to be plasmid linked and they include bacteriophage exclusion (Schell,

1975), sensitivity to a bacteriocin produced by a nonpathogenic species, A.

radiobacter (Kerr and Htay, 1974; Schell, 1975), octopine or nopaline synthesis

and degradation (Bomhoff et al., 1976). These two unusual arginine derivatives,

octopine and nopaline, are normally found in tumorous tissue, and the production of one or other of these compounds is A. turnefaciens strain specific.

Moreover, strains with induced octopine in tumor cells are able t o utilize

octopine as a source of N for bacterial growth; strains which incude nopaline

tumors can utilize nopaline as a source of N. Genetic experiments (Bomhoff et

al., 1976) have examined the plasmid-linked nature of octopine or nopaline

utilization and systhesis. The plasmid-linked genes have an immense advantage

since cells containing this plasmid may be selected and characterized.

The crucial question of whether or not the plasmid is integrated into the host

cell DNA has not been answered. Neither has it been established that the plasmid

per se is capable of causing neoplastic growth of plant cells. Since cultured

crown gall cells are growth-regulator independent, and since other biochemical

properties are associated with the presumptive tumor-inducing plasmids, it

should be possible, using protoplasts, to establish unequivocally whether the

isolated plasmid can induce the tumorous state. We have begun research with

this specific object in mind.



2. Requirements of the Molecular Vector

There are other requirements which we believe essential to utilize such DNA

vectors for gene transfer in plants. First, it will be necessary to multiply the

hybrid molecules in some convenient organism and at present the bacterium

Escherichia coli is most suitable. Therefore the replication sequence that enables

the hybrid molecule to multiply in plants must be combined with a replication

sequence (i.e., of a bacterial plasmid) permitting multiplication in bacteria. Thus

the hybrid molecule could multiply in plants or bacteria. If efficient infection

and multiplication of the vectors previously discussed can be achieved in plant

cells, as has been obtained for tobacco mosaic virus, then indeed plant cells may

provide a convenient milieu in which to adequately multiply the hybrid molecule.

The hybrid DNA must aIso carry genetic information that is expressed in both

plant and bacterial cells in order to select those few cells which contain the

hybrid DNA, As mentioned earlier (Section VII, B), tobacco cells are sensitive to

similar concentrations of kanamycin (but not carbenicillin, tetracycline, or



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



69



neomycin) which inhibit protein synthesis in bacteria. No doubt other drugs or

antimetabolites will be found that equally inhibit growth of plant and bacterial

cells. The A . tumefaciens plasmids have the gene for octopine or nopaline

utilization which can be selected for, at least in bacteria. If the induction of

growth regulator autotrophy in plant protoplasts can be associated with infection by purified plasmid, then this provides another double selection system.

If the hybrid DNA molecule can be taken up, multiplied, and expressed in

plant cells, the ultimate object is to integrate i t into the plant genome. As

pointed out earlier (Section VI, B) kanamycin resistance appears t o have been

stabilized in tobacco cells following transformation by bacterial plasmid DNA.

The concomitant use of yirradiation may have led to integration brought about

by breakage and reunion. However, this evidence is only circumstantial.

3. Insertion Sequences and Transposons

It has generally been believed that integration of DNA into some specific

genome can be achieved only as a consequence of recombination between

segments of DNA having extensive nucleotide sequence homology or by breakage and reunion. Recently a new class of genetic elements has been characterized

in bacteria in which recombination only occurs at the termini of these elements

(see review by Cohen, 1976). The nucleotide sequences at these termini are

called insertion sequences and they have a defined length of 800-1400 base

pairs. There are several different insertion sequences. When such a sequence

inserts into a gene the function of that gene is abolished and indeed any other

genes distal to the point of insertion relative to the promoter for that operon.

When the sequence is excised from a gene the function of that gene, and any

other genes in the same operon which were affected, is restored. Any DNA

sequence can be contained between two insertion sequences and these termini

are arranged so that one is an inverted repeat of the other. Such complexes are

known as transposons because they can move from one hereditary element t o

another, e.g., from plasmid to plasmid, plasmid to chromosome, or plasmid to

bacteriophage, conferring on the receptor element the genetic function contained between the insertion sequences. These transposable genetic elements are

probably responsible for the immense genetic diversity of bacterial plasmids, and

indeed for the evolution of prokaryote genetic systems. Moreover, the transposable controlling elements that regulate phenotypic expression in maize (McClintock, 1956; Fincham and Sastry, 1974) are consequentially similar to

transposons in that they transpose to different chromosomal locations and in so

doing influence a variety of genetically controlled functions.

These insertion sequences may provide the mechanism for integrating the hybrid

molecules into the plant genome. Such a mechanism would mitigate the restraint of

the substantial need for sequence homology to effect integration. In the case of the



70



W. R. SCOWCROFT



association between the plasmid of A. tumefaciens and crown gall, Schell et al.

(1977) propose as a working model that insertion sequences are involved in the

mechanism of transformation of plant cells to the neoplastic state. If this

proposition is substantiated, then molecular genetic engineering in plant improvement is a very real possibility.

Provided that a molecular vector can be developed for transferring foreign

genetic information into plants, the remaining problem is to decide what genes,

or gene-controlled functions, might usefully be incorporated into a plant breeder’s population. A related problem is the identification of the nucleotide sequence for the particular gene in the hybrid molecule, because for the transfer of

specific genes it is essential that the hybrid containing the appropriate nucleotide

sequence be obtained in quantity. If a bacterium is to be used to multiply the

vector, and since eukaryote genes are poorly expressed by prokaryotic protein

synthesizing systems, it is probably essential that purified mRNA appropriate for

that nucleotide sequence be used as a probe (Kedes et al., 1975; Grunstein and

Hogness, 1975). Since such mRNAs are not readily available for plant genes, t h s

is unlikely to be of much value. Alternatively a “shotgun” approach can be used

where a large number of hybrid molecules, containing random pieces of DNA

from the donor genome, are made and inserted into plant cells. The specific

hybrid molecule containing the desirable nucleotide sequence is selected by

virtue of a property that the gene confers on the recipient plant cell. The specific

hybrid selected in plant cells, from among a random set, could then be multiplied in an appropriate bacterium. Even the integration of a vector, containing

random pieces of DNA from another plant genome, into a plant species of

choice would markedly increase the quantity, and hopefully the quality, of

genetic variability on which plant breeders could practice selection.

D. PLANT IMPROVEMENT AND DESIRABLE GENES FOR

MANIPULATION



In this review, and where experimental information is available, I have presented examples where the asexual transfer of genetic information may be of

value to plant improvement (Section IV). These examples have included aspects

of disease resistance, tolerance to stress conditions such as salinity or flooding,

and the possibility of selecting for amino acid overproducing mutants. Other

physiological characters may be subject to genetic modification in cellular

systems. For example, freezing injury in plants has long been thought to be due

to dehydration injury (Burke et al., 1976), and Towill and Mazur (1976) have

recently demonstrated a close correspondence between freezing injury and

osmotic injury to cultured plant cells. This would certainly be amenable to cell

culture and protoplast studies where resistance to osmotic shrinkage could be

selected. The consideration of other possibilities, such as specific gene(s) modifi-



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



71



cation which affect yield, would require the expertise of plant breeders, physiologists, biochemists, and geneticists which is beyond the scope of this review.

However, one of the commonly advertised aims of cell culture and genetic

engineering is to confer on nonlegume crop plants the ability t o fix atmospheric

nitrogen.

1. The Special Case of Nitrogen Fixation



If nonlegumes could be developed w h c h were able to meet even part of their

nitrogen requirement directly from biological nitrogen fixation, the benefits

from reduced fertilizer use might be enormous. In the best known agricultural

nitrogen fixing system, the legume-rhizobia symbiosis, it has been estimated

that grain legumes fix as much nitrogen (about 40 X lo6 tons/annum) as is

currently provided by the application of chemical nitrogen fertilizers (Hardy and

Havelka, 1975). In addition it is estimated that by the turn of the century the

demand for nitrogenous fertilizers will rise to 200 X l o 6 tons/annum. There is

an obvious need for an alternative.

From an examination of known systems for biological nitrogen fixation it is

clear that the critical process of reduction of nitrogen to ammonia by the

enzyme nitrogenase is restricted to the prokaryotes. There is n o unequivocal

example of a eukaryote which synthesizes this enzyme (see Postgate, 1974;

Dilworth, 1974). Among the prokaryotes the ability to fix atmospheric nitrogen

is fairly widespread, particularly among those classified as primitive. These

microorganisms inhabit the rhizosphere and phyllosphere of plants and are

found in the water of rice paddies. The contribution of free-living nitrogen-fixing

nicroorganisms t o the nitrogen nutrition of plants is poorly understood and

probably minimal. Nonpathogenic associations of nitrogen-fixation microorganisms with lower and higher eukaryotes are also widespread in nature, and

include the blue-green algal associations with fungi (lichens) and the aquatic fern

Azolla (see appropriate chapters in Quispel, 1974). In addition more than a

hundred species of nodulated nonlegume plants are known to fix nitrogen; the

microsymbiont is almost certainly an actinomycete (Bond, cited in Quispel,

1974). Recently, von Bulow and Dobereiner (1975) have described the natural

occurrence of a nitrogen-fixing association between the bacterium Spirillum

Zipoferum and roots of monocots. Apart from pasture grass species, this association has been found in roots of maize and sorghum. The extent to which this

association provides fixed nitrogen for the host plant is still a matter for

conjecture. Clearly then eukaryote plants, during the course of their evolution,

have taken advantage of biological nitrogen fixation by entering into associations

with microorganisms.

Because of its agricultural importance the best understood nitrogen-fixing

association is that of the legume-rhizobia symbiosis, where the contribution to

the nitrogen nutrition of the host, and t o the nitrogen status of agricultural soils



72



W. R. SCOWCROFT



is considerable. Recent evidence from our, and other, laboratories has shed

additional light on the relative roles of the bacterium and plant in legume

symbiosis which indicates that legume symbiosis is not as rigid as previously

believed. It was largely accepted that the legume host provided essential genetic

information for the synthesis of nitrogenase in legume nodules (Ddworth and

Parker, 1969). However, it is now known that at least some species of Rhizobium can fix nitrogen independently of the host (Pagan et al., 1975; McComb et

al., 1975; Kurz and LaRue, 1975). Moreover, these rhizobia can also fix nitrogen

when cultured with nonlegume plant cells such as tobacco (Scowcroft and

Gibson, 1975), wheat, rape, bromegrass (Child, 1975), carrot, and rice (Kurz and

LaRue, 1975). It seems therefore that the cellular environment of nonlegumes

does not inhibit the process of nitrogen fixation. Studies on the regulation of

nitrogen fixation in free-living rhizobia (Scowcroft et al,, 1976; Bergersen and

Turner, 1976) indicate that nitrogenase synthesis is not directly regulated by the

adenylylation/deadenylylation of glutamine synthetase as has been found for the

anaerobic nitrogen fixer Klebsiella pneumoniae (Shanmugam and Valentine,

1975).

A further index of the potential flexibility of nitrogen fixation by rhizobia is

apparent in its symbiosis with a nonlegume. Trinick (1973) showed that nodules

formed on the nonlegume Trema cannabina were due to a slow-growingstrain of

Rhizobium. In these nodules, leghemoglobin, which is normally required in

legume nodules to regulate the oxygen flux, was not found (Coventry et af.,

1976). It is possible that in these Trema nodules an oxypolyphenol oxidase may

act as an alternative 02-carrier, to maintain the nodule O2 flux required to

support N2 fixation. Alternatively, the rhizobia forming these nodules may have

undergone evolutionary adaptation to enable the nodule bacteria to survive and

fix N2 without an O2 stabilizing system. This could also account for the fact

that only some strains of rhizobia will fix nitrogen under free-living conditions

(Pagan et al., 1975). Trinick and Galbraith (1976) have further shown that

nodule development in T r e m is more rudimentary than that found in the

legumes. They suggest that nitrogen fixation occurs in the extensive infection

threadlike structures found in Trema roots as well as in the bacteria-filled

host cells.

Extensive studies have characterized the genetic regulation of nitrogen fmation in several bacteria, particularly Klebsiella pneumoniae (Brill, 1975). The

nif' gene(s) of Klebsiella can be mobilized and transferred to other bacteria

using bacterial plasmids. When such information is transferred to the closely

related species, Escherichia coli, the nif 'gene is expressed and can be integrated

into the E. coli chromosome (Cannon et al., 1974). Using a wide host range

plasmid, the Klebsiella nif' gene has also been transferred to nifAzotobacter

vinelandii cells (Cannon and Postgate, 1976), and to Rhizobium meliloti and

Agrobacterium tumefaciens (Dixon et al., 1976). It was only in the first

species that the Klebsiella nif' gene was expressed. In the latter two there