III. Anther Culture and Haploids

SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



45



demonstrated that isolated microspores of tobacco can be cultured to produce

haploid plants. Extensive cytological analyses in Nicotiana and Datura have

elucidated the events which lead to the production of haploids (see Sunderland

and Dunwell, 1974; Sunderland, 1974). Under appropriate culture conditions

the normal process of pollen development is arrested between the tetrad stage

and the conclusion of the first pollen mitosis. Subsequently the generative cell

degenerates while the normally quiescent vegetative cell divides to form an

embryolike structure. Cytological events indicate that haploidy follows one of

either of two developmental pathways, each of which leads to the ultimate

degeneration of the generative partner (Sunderland, 1974). Following induction

of cell division in the pollen, embryos develop which give rise to plantlets

actually growing out of the anther, or callus may be formed which then has to

be differentiated to regenerate a plant. The former sequence is characteristic

primarily of tobacco and Datura, and the latter is the more general consequence

in other species. In cases where callus, hopefully with a haploid complement, is

formed, controlling the ploidy level in the callus may be a serious limitation to

its practical use for generating haploids. On the basis of the use of p-fluorophenylalanine (PFP) to increase frequency of haploid segregation in fungi (Day

and Jones, 1971), Gupta and Carlson (1972) claimed that PFP inhibited the

growth of diploid, but not haploid, cells of tobacco. This claim has not been

sustained (Zenk, 1974; Dix and Street, 1974) or at best is not very reproducible

(Chaleff and Carlson, 1974). An additional problem, as pointed out earlier, is the

difficulty of regenerating plants from callus. This varies from species to species,

and indeed from genotype to genotype within a species. With haploid callus I

would judge that the problem of regenerating a representative sample of haploid

genotypes might be even more difficult.

Notwithstanding, haploid plants have been successfully produced in species

other than tobacco and Datura. Catalogs of species in which haploids have been

produced are provided by Smith (1974), McComb (1974), and Sunderland

(1974). Although specific reports cannot be cited, a haploid information exchange service, edited by the Haploid Project Group, Max Planck Institute fur

Biologie, Rosenhof, Germany (see Kasha, 1974, p. 41 l), carries reports of both

successful and unsuccessful attempts to induce haploids. Apart from tobacco,

other important crop species in which haploids have been produced by anther

culture include rice (Niizeki and Oono, 1968; Oono, 1975; Wang e t al., 1974;

Laboratory of Genetics, 1975; Woo and See, 1975), wheat (Ouyang et al., 1973;

C. Wang et aZ., 1973;Picard and de Buyser, 1973), barley (Clapham, 1973; Dale,

1975), triticale (Y. Wang et al., 1973; Sun et al., 1974), tuberous Solanum

species (Irikura and Sakaguchi, 1972; Dunwell and Sunderland, 1973), and

turnip rape (Brassica campestris) (Keller e t al., 1975).

The use of haploids in plant improvement, or indeed any research where a

gametophyte is required from the haploid sporophyte, requires that the chro-



46



W.R. SCOWCROFT



mosome complement be doubled. Fortunately, chromosome doubling techniques are already efficient. An excellent account of their status and

methodology is given by Jensen (1974).



B. THEORY AND APPLICATION

In a breeding program designed to produce pure line varieties in a self-fertilized

species, or inbreds for hybrid production, the advantages of using haploids is

obvious (Kimber and Riley, 1963; Nei, 1963; Scowcroft, 1975). Normally five

or six generations of selfing are required to produce a homozygous line from a

genetically heterogeneous population. Inbreeding depression causes inviability or

sterility in many of the lines, the cost of which may not be apparent until the

third or fourth generation of selfing. The use of the doubled haploid technique

automatically selects against any inviable gene combinations and immediately

exposes mutations causing sterility. In addition to producing homozygous lines,

the doubled haploid technique may have considerable advantage in recurrent

selection programs where inheritance is not particulate. Griffing (1975) compared the efficiency of standard recurrent selection methods with those modified by the inclusion of doubled haploid and cloning techniques. With the first

of these modifications, individual phenotypic performance of the doubled haploids was evaluated, the population was subjected to truncation selection, and

selected individuals were randomly mated to provide the breeding population

for the next cycle. The inclusion of cloning techniques provides more precision

in evaluating the genotype of a doubled haploid where environmental variance is

significant. The comparisons were made where heritability was high (dphenotypic variance was additive genetic), moderate, or low (each separately with

environmental variance equal to the additive genetic or with n o environmental

variance). The efficiency comparisons showed that genetic gain per cycle of

selection is considerably improved by the inclusion of the doubled haploid

technique, particularly when total plant numbers are restricted. Given similar

cycle lengths the haploid selection procedures can be up to six times as efficient

as selection based on diploids.

The critical parameter therefore is the relative length of time required per

cycle of selection. The use of the doubled haploid technique to increase the

efficiency of selection depends solely on the development of rapid doubled

haploid extraction procedures. Considerable success has been achieved with

tobacco where indeed the value of haploids in breeding programs is being

realized. Recent work in the People’s Republic of China has also acheved a

substantial improvement in the frequency of haploid production in rice and

wheat.



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



47



Lines of tobacco differing in alkaloid content (Collins et al., 1974) and

nullisomics for use in genetic analysis have been developed (Mattingly and

Collins, 1974). As with many crops, breeding for disease resistance in tobacco is

a major objective. Within a relatively short time, compared with conventional

procedures, promising disease-resistant lines have been derived from haploids

produced by anther culture (Nakamura et al., 1974; Cooperative Group, 1974;

Wark, 1977). In yield and quality tests doubled haploid derivatives performed as

well as or better than the parental cultivars. Wark (1977 and personal communication) has utilized the doubled haploid technique t o introduce sources of

resistance to blue mold (Peronospora tabacina) from related species of Nicotiuna

into commercial cultivars. Similarly, resistance to tobacco mosaic virus has been

transferred from Nicotiana glutinosa to N. tabacum. Following mutagenesis,

haploids have also been screened for resistance to black shank (Phytophthora

nicotianae var. nicotianae) (Wark, personal communication). Preliminary tests

indicate that some resistant haploids have been obtained.

The use of anther-derived haploids in plant improvement appears to have

begun in China about 1971. An intensive effort to produce haploids in wheat

and rice has led to a substantial increase in the frequency of haploid green plants

derived from anther culture. For both wheat and rice the object has been to

recover superior haploid segregants primarily from F1 and F2 hybrids. Initial

studies on haploid culture in wheat (Ouyang et al., 1973; Wang et al., 1973a)

reported a low frequency (11%) of callus formation in cultured anthers and of

these less than 30%were capable of regenerating green haploid plants. In 1976 a

group from the Institute of Genetics, Academia Sinica (301 Research Group,

1976) reported a dramatic increase in the frequency of wheat haploids by

culturing anthers on a medium containing 20%potato water extract, 9%sucrose,

2,4-D (2.0 mg/liter), kinetin (0.5 mg/liter), and iron chelate. When anthers were

induced t o form callus on this medium, and then differentiated on Murashige

and Skoog medium, the overall frequency of green anther-derived plants was

3-17 times greater than for the controls, the highest frequency being 13.6% of

anthers differentiating green plantlets.

Initial attempts at haploid culture of rice were successful, but a low frequency

(less than 3%) of green plantlets were obtained. Altering the nitrogen source

from 10 mM KN03, 12.5 mM NH4N03, and 1.5 mM Ca ( N 0 3 ) 2 , to 3.5 mM

(NH4)2 SO4 and 28 mM KN03 (N6 medium) more than doubled the percentage

of anthers which produced callus (Chu et al., 1975). Variation in (NH4)2S04

concentration had substantially more effect on callus formation than did variation in KN03 concentration. With the addition of appropriate growth regulators

to the N6 medium, a high frequency (75-80%) of plantlet regeneration from

callus was found, of which approximately half were albinos. In the two varieties

examined in detail the production of anther-derived green plants was 16% and



48



W.R. SCOWCROFT



12% respectively. Recent work has further indicated that satisfactory induction

of plants from anther callus of rice (and wheat) can be achieved without the

addition of growth regulations (Chu e t al., 1976).

In a recent publication (Yin e t aZ., 1976), a cooperative group of Chinese

workers have evaluated a number of lines from anther-derived haploids of rice

for agronomic characteristics, disease resistance, yield, etc. Several promising

lines are being further evaluated and one line has been named as the variety

“Tanfeng” (haploid-derived high-yielding No. 1).

The theoretical and practical advantages of the use of haploids in plant

improvement are clear. For a given species it is not sufficient merely to

demonstrate that haploid plants can be derived from anther culture. This is

simply analogous to the occurrence of spontaneous haploids from malfunctions

in the process of fertilization and zygote formation. Rather haploidy can only be

of value provided haploids can be produced rapidly and in large numbers. An

additional limitation may result from competition between pollen-derived embryos during the induction process. Obviously, inviable gene combinations will

cause the elimination of many developing embryos and the extent of inviability

will depend on the genetic heterogeneity of the breeding population.

Competition between developing haploid plantlets must be minimal to ensure

that genetic segregation for traits of interest to the plant breeder is fully

represented. The ability to culture isolated pollen grains (Nitsch, 1974) largely

eliminates this problem. On the basis of realistic assumptions, Nitsch estimated

that some 7000 plants could be obtained from a single flower bud of tobacco.

This represents immense segregation potential and, for example, could allow the

isolation from a heterozygote of an individual carrying up to six recessive alleles.

A similar number of F2 segregating genotypes would at best permit the isolation

of a homozygous recessive for no more than three loci, which were heterozygous

in the parent. Techniques are required to enable this potential to be realized in

major agronomic crops. I wish to echo Riley’s (1974) plea that in experiments

on anther culture the behavior of the developing gametic sporophyte be monitored closely. In this way principles of wider application may emerge. In this

context the correlation of a cytological dimorphism in barley with the propensity t o form pollen callus (Dale, 1975) is noteworthy, as are the earlier observations that ethrel stimulated additional nuclear divisions in pollen grains (Bennett

and Hughes, 1972) and that the ribosomal populations of the meiocytes change

as they enter meiosis (Mackenzie e l aZ., 1967).

I V . Mutant isolation and Selection



The utilization of mutants in understanding biochemical and developmental

processes in microorganisms is an obvious paradigm for their potential value in



SOMATIC CELL GENETICS AND PLANT IMPROVEMENT



49



plant biology. Moreover, defined mutants greatly facilitate the recognition of

rare genetic events such as might result from genetic recombination, mutation,

somatic hybridization, and genetic transformation. Apart from these more

fundamental uses of biochemical mutants, selecting mutants which cause lesions

or alterations in biochemical pathways may be of importance in several aspects

of plant improvement. For example, biochemical mutants could be selected for

disease resistance, improvement of nutritional quality, adaptation of plants to

stress conditions such as occurs in saline soils, elimination of toxins and antimetabolites deleterious to man and animals, and to increase the biosynthesis of

plant products used for medicinal or industrial purposes.

There are only a few cases where mutants which cause a block in a particular

biosynthetic pathway have been recovered in whole plants. These include

thiamine-deficient mutants in Arubidopsis (Langridge, 1955) and tomato (Langridge and Brock, 1961), nitrate reductase deficiency in Arubidopsis (OostindierBraaksma and Feenstra, 1973), and a proline auxotroph in maize (Gavazzi et al.,

1975). Slightly more success has been achieved in isolating mutants which affect

photosynthesis primarily because they affect chloroplast development and can

be readily selected (Levine, 1969; Miles and Daniel, 1974; Miles, 1976). Such

mutants have been valuable in analyzing basic processes in photosynthesis. The

relatively depauperate collection of biochemical mutants in plants probably

results from the expense of screening large populations of whole plants for

relatively rare mutants. As pointed out by Chaleff and Carlson (1974), the

organizational complexity of plants with morphologically and biochemically

different, yet interdependent, cells and structures also hinders the isolation of

defined biochemical mutants.

The ability to manipulate large populations of homogeneous plant cells provides the opportunity to isolate biochemical mutants. Technically it is relatively

simple to screen 106-107 cells in culture; screening a similar number of whole

plants is very resource-consuming. Because plants can be regenerated from cells

of some species the effect of such mutants may be evaluated in mature plants.

Dominant and co-dominant mutants can be isolated from diploid, or indeed,

polyploid cells. It might appear axiomatic that haploid cell lines would be

required to isolate recessive biochemical mutants. However, this might not be

the case. Recessive mutants occur in diploid animal cell lines at a frequency

considerably greater than would be expected from the frequency of a double

mutation event (Terzi, 1974). Recently, Williams (1976) found in the slime

mold Dictyostelium discoideum that the frequency of spontaneous mutation to

the recessive state at a single locus was only an order of magnitude greater in

diploids relative to that in haploids.

Indeed, plant cell cultures have been used to successfully isolate biochemical

mutants. A discussion of some of these mutants can be found in Chaleff and

Carlson (1974, 1975), Widholm (1974b), and Zenk (1974). The only report