Download Chromosome numbers in female and male gametes: One

INDUCTION O F CHROMOSOME DOUBLING A T NIEIOSIS
BY THE ELONGATE GENE IN MAIZE
M. M. RHOADES
Department
of
AND
ELLEN DEMPSEY
Botany, Indiana University, Bloomington
Received March 15, 1966
EIOSIS is a highly integrated system characterized by the orderly procession
of events such as pairing, recombination, chiasma formation, and disjunction, which culminates in a reduction o€ chromosome number in the gametes.
That this system is under genic control has long been evident from the many
mutant genes reported in plants and animals which affect specific stages of the
meiotic system. These meiotic genes are of unusual interest in that a thorough
study of the physicochemical basis of their action gives promise of leading to a
better understanding of the physiological conditions which bring about the transsition from mitosis to meiosis as well as of the cellular environment essential for
such critical events as synapsis, recombination, and disjunction. A discussion of
the genetic control of meiosis is given in REES(1961) .
A new meiotic gene is reported in this paper which will be a useful tool in
cytogenetical studies with maize. This mutant was found in the open-pollinated
variety Hays Golden. It is a simple recessive and has not yet been placed in a
linkage group. This new meiotic gene is called elongate (el) because of the elongated appearance or despiralization of the chromosomes at both meiotic anaphases,
but it has a number of other effects. The most significant of these is the production
of unreduced eggs which occur in varying proportions with haploid eggs. Evidence will be presented that the diploid eggs originate by chromosome doubling
at the second meiotic division. The recovery of two of the four chromatids of a
bivalent in the diploid eggs of elongate plants enabled a half-tetrad analysis and
data were obtained which bear on the questions of chromatid interference and
centromere mapping.
EXPERIMENTAL RESULTS
Chromosome numbers in female and male gametes: One of the striking manifestations of the action of the el gene is expressed in the female flowers. Ears
borne on el plants have plump and shriveled kernels as well as aborted ovules.
The relative frequencies of these vary greatly on different ears and are evidently
affected by modifying genes but, with the exception of one progeny, no difficulty
was encountered in classifying segregating normal and el plants on the basis of
ovule abortion and shriveled kernels. One-to-one ratios of EL to el were obtained
in testcrosses and 3:l in F, populations. The plump kernels gave rise to diploid
plants, all with 20 chromosomes. A considerable fraction of the shriveled kernels
Gcrietics 54: 505-522 August 1966.
506
M. M. RHOADES A N D E. DEMPSEY
are so poorly developed that they do not germinate, but chromosome counts were
made on root tip cells of 825 individuals from shriveled kernels. All were near the
triploid level, indicating that diploid or near diploid eggs were formed by el
plants. Although 82% were euploids with 30 chromosomes, as shown in the
tabulation below, 18% had aneuploid numbers ranging from 25 to 33.
Chromosomenumber in
plants from shriveled seed
25
26
27
28
29
30
31
32
33
Frequency
1
1
2
17
71
676
46
10
1
A variable amount of pollen abortion is found in el plants but the functional
pollen is haploid. When elongate plants were used as male parents in crosses
with diploid females, the 961 offspring were all diploid. Crosses of 4n egg parents
by 2n males customarily give only shriveled seed. Pollen from elongate stocks on
4n silks gave rise to 1842 shriveled seeds and 18 plump kernels which proved to
have triploid embryos. No plump kernels with 4n zygotes were found. These
would occur if the pollen of el plants consisted of a mixture of diploid and haploid
grains and the diploid grains effected fertilization. However, haploid grains have
a competitive advantage over diploid pollen in pollen mixtures, and the failure
to find 4n kernels might mean that the el plants had a low proportion of diploid
pollen grains which, together with their disadvantageous handicap, could account
for the absence of 4n kernels. This is unlikely. Diploid grains are larger than
haploid, and the two can readily be distinguishedin pollen samples from tassels
chimeric for 4n and Qin tissue. Although there is some abortion, pollen samples
from elongate tassels show no indication of diploid grains. Furthermore, in the
cytological studies of microsporogenesis all cells observed at metaphase I (MI)
and MI1 contained either ten bivalents or ten dyads respectively and none had
double that number. It may be concluded that no viable diploid pollen is produced.
Among the effects found at microsporogenesis in el plants are the following:
( 1 ) The chromonemata are relatively uncoiled at anaphase and telophase of both
meiotic divisions giving elongated or stretched chromosomes (Figure la, d) . (2)
Misdivision of the centromere occurs occasionally at MII. (3) Neocentric
regions, resembling those produced by abnormal chromosome 10 (RHOADESand
VILKOMERSON
1942), occur sporadically at MI1 (Figure l f ) but have not been
observed at M I or anaphase I (AI). (4) Immediately after the quartet stage the
haploid spores may undergo another division without further chromosomal
replication. The ten chromosomes are distributed randomly to one or the other
pole (Figure lg). This behavior is similar to that produced by the plymitotic
gene described by BEADLE(1931) except that in polymitotic plants the super-
ELONGATE MAIZE
507
numerary mitoses take place while the spores are in quartets. Of these four cytological effects of the elongate gene, only the despiralization of the chromonemata
is consistently found in all strains of elongate, Also observed in some elongate
stocks were duets of cells at interphase between the first and second meiotic
divisions that possessed two nucleoli in each sister cell and presumably had an
unreduced chromosomal complement (Figure 1b,c) . These putative unreduced
cells were never seen to undergo division. They apparently degenerate and may
account for some of the pollen abortion observed in el plants,
Hypotheses on the action of the el gene: It is evident from the above presentation that el plants produce both haploid eggs with ten chromosomes and unreduced eggs at the diploid level although some have less than 20 chromosomes
and some have more. These unreduced eggs could arise in various ways. Among
the possibilities are (1 ) somatic doubling in the sporogenous cells to form tetraploid tissue, ( 2 ) doubling in the gametophytic generation, ( 3 ) suppression of the
first meiotic division followed by a normal second division (as used in this paper,
“suppression of the first meiotic division” means that the meiotic events proceed
normally up to AI, but disjunction fails to occur and a restitution nucleus is
formed), (4) a normal first division with omission of the second, and ( 5 ) a
normal first division with chromosomal replication occurring during interphase
to form 20 dyads in each sister cell, followed by the second meiotic division.
The logical way to decide among these five alternatives would be by direct
cytological examination of megasporogenesis and gametogenesis, but this is
technically difficult to do in maize. Some observations have have been made of
megasporogenesis in sectioned ovules. They are in accord with the conclusions
reached by the genetic analysis but by themselves are not conclusive. However,
a cytogenetic study did permit discrimination among the various hypotheses.
Observations of megasporogenesis disclosed only ten pairs at diakinesis and
MI, but a limited number of cells were found in these stages and, since part of
the eggs are diploid and part haploid, it is possible that only normal cells were
examined. Nevertheless, the observations indicated that nucellar doubling is not
responsible for the diploid eggs. Confirmation of this conclusion came from the
ratio of dominant to recessive phenotypes found in the diploid eggs of el plants
heterozygous for the wx gene close to the centromere of chromosome 9. The
observed ratio of 3Wz:2wz is far from the approximate 5:1 expected if the diploid
eggs are derived from duplex tetraploid cells.
Cytological evidence of unreduced megaspores is based on observations on the
nuclei of embryo sacs. Several ovules were found in which the nuclei contained
two nucleoli and were undoubtedly diploid. Diploid nuclei were also observed
at the interphase between the first and second meiotic divisions and, if 2n megaspores arise from these cells, doubling in the gametophyte generation (hypothesis
2 ) would be ruled out. Genetic studies have revealed that many of the diploid
eggs are heterozygous for one or more marked loci. Obviously such heterozygous
eggs could not have arisen by chromosome doubling in the gametophytic generation because, on such a mechanism, both homologues would carry the same alleles.
I n order to differentiate between hypothesis 3 and 4, plants homozygous for
508
M. M. RHOADES AND E. DEMPSEY
el and heterozygous for Sh Wx/shwx were crossed by pollen from recessive
plants. The Wx locus is in the short arm of chromosome 9 and is known to lie
within a few crossover units of the centromere (ANDERSON
and RANDOLPH1945).
Single exchanges may occur between Wx and the centromere, but double exchanges should be rare. The Sh, locus is also in the short arm of 9, approximately 20 crossover units distal to Wx. The frequencies of wx homozygosis in
the diploid eggs expected from populations of megasporocytes with different frequencies of single-exchange tetrads is given in the tabulation below on the hypothesis of suppression of the first division and on the hypothesis of omission of
the second division.
Percentage of diploid eggs homozygous for wx coming from megasporocyte populations
with different frequencies of single-exchange tetrads
Megasporocyte population
Suppression of
first division
100% no-exchange Wx-centromere
80% no-exchange,20%single exchanges
100% single exchanges
0
50
5
25
41)
Omission of
second division
0
If no exchanges occur, 50% of the diploid eggs will be homozygous for the dominant allele of the segregating locus and 50% will be homozygous for the recessive
allele on the hypothesis of omission of the second division. If, for example, 20%
of the bivalents have a single exchange between the centromere and the tested
locus and 80% have no exchange, then 40% of the diploid eggs will be homozygous for the recessive allele. Quite different expectations will come from the
hypothesis of first division failure according to the above tabulation. No homozygosity will be found if no recombination occurs between the segregating locus
and its centromere and only 5% of the diploid eggs will be homozygous for the
recessive allele if single exchanges take place in 120% of the cells. Even with
100% single chiasmata between the locus and the centromere, the frequency of
the homozygous recessive would be only 25 %.
In a population of 509 diploid eggs, 39.5% were homozygous for the recessive
wx allele while among 156 diploid eggs whose genotypes were determined,
19.2% were homozygous for the recessive sh allele. The high percentage of wx
cannot be explained on the hypothesis of first division suppression. Moreover,
the more distally located sh allele would be expected to give a higher and not a
lower frequency of homozygosis if homozygosis was dependent on crossing over
between the locus and the centromere. These data clearly indicate that the
hypothesis of first division suppression is incorrect; they are wholly in accord
with second division doubling as the mechanism by which diploid eggs originate.
The validity of this conclusion was tested by an analysis of the genotypes of 2n
gametes arising from elongate plants heterozygous for the Zg, and al loci in the
long arm of chromosome 3. (Plants with the dominant Lg allele have a ligule at
the base of the leaf while Zg plants are liguleless; the dominant A allele produces
anthocyanin color in the aleurone and plant, while aa plants have colorless
509
ELONGATE MAIZE
aleurone and are green.) The lg locus is situated near the middle of the arm and
the A gene lies more distally. The standard recombination value between Lg
and A is 36%.
Both the plump and shriveled kernels on the testcrossed ears were planted
and the progenies scored for the Zg and a phenotypes. The plump kernels gave
rise to diploid offspring. The shriveled kernels required special germination techniques and even then many failed to grow. The viable seedlings from shriveled
seeds were transplanted to the field and classified for the segregating phenotypes.
The chromosome number of each plant was determined; all were at the triploid
level with the great majority having 30 chromosomes. The euploid plants with
30 chromosomes were testcrossed both as the male and female parent in order to
ascertain the genotypic constitution of the diploid eggs from the phenotypes and
their proportions in the testcross progenies. Because of sterility, a great number
of triploids failed to produce sufficient seed to allow the elucidation of the genotypic constitution of the diploid eggs from which they arose. However, in the
crosses involving chromosome 3, the number of genotypic determinations was
sufficient for a genetic analysis.
Listed below are the genotypes of diploid eggs coming from diploid elel
Lg A/Zg a plants used as the female parent in testcrosses.
Genotype
Number
l g a __
L g A __
L g A -1gA LgA Lga L
1gA
--g a L g A lga lga Lga lga
1gA l g a 1 g A 1gA
LgA
__
8
5
66
26
37
21
16
22
3
2=204
The frequencies of homozygosis for Zg and for a do not yield useful information
in support of either first or second division doubling since these loci in chromosome 3 are too far removed from the centric region. However, if the first division
is normal, the frequency of tetrads with different exchanges can be determined
from the kinds and numbers of half-tetrads recovered in the diploid eggs, They
are as follows: combinations Lg A/Lg A and lg a/lg a come from no-exchange
tetrads, Lg A/Zg a from singles in region 1 and from one half of the 3-strand
doubles in regions 1 and 2, L g A / L g a and l g A / l g a from single exchanges in
region 2, Lg A/lg A and Lg a/lg a from 2- and 4-strand doubles in regions 1 and
2, L g a / l g A from one hali of the 3-strand doubles in regions 1 and 2 and
ZgA/lgA from 4-strand doubles in region 2. Region 1 is the segment from the
centromere to Lg; region 2 is the Lg-A interval. When allowance is made for
undetected combinations arising from doubles in region 2, the tetrads giving rise
to the 204 diploid eggs consisted of 4.9% with no exchanges in either regions 1
or 2, 21.6% with a single exchange in region 1. 27.9% with a single exchange
in region 2, 39.7% with one exchange in both regions 1 and 2, and 5.9% with
two exchanges in region 2.
On the assumption that doubling took place by formation of a restitution
nucleus at the first division, a n entirely different array of tetrads is obtained.
Double exchange tetrads are estimated at 59.3%, no-exchange tetrads are 16.7%
and the single exchanges in (1) are 9.8%. Singles in (2) cannot be distinguished
from doubles in (2); these would total 2.3%. Triple exchange tetrads with two
510
M. M. RHOADES A N D E. DEMPSEY
exchanges in region 2 and one exchange in (1 ) would represent the remaining
12%. The tetrad frequencies calculated on the basis of first or second division
doubling may be compared with those derived from G1, LghA,/gl, lgaal heterozygotes (RHOADESand DEMPSEY 1966) where the Gl-Lg interval is comparable
to the centromere-Lg segment in the elongate material. Data from plants with
knobless chromosomes 3 gave the following frequencies: no exchange tetrads
6.0%, single exchanges in Gl-Lg 20.2%, single exchanges in Lg-A 43.0%, and
double exchanges 30.8%. The figures based on hypothesis 4 are in closer agreement with the above distribution than those calculated on hypothesis 3, especially
when the doubles in region 2 are combined with the singles in that region.
At this stage in the analysis it could be claimed that the diploid eggs originate
from the omission of the second meiotic division (hypothesis 4) and that the
problem of their origin was solved. However, the same array of half tetrads in the
diploid eggs would be expected if the megasporocytes had undergone two normal
meiotic divisions with an extra chromosomal replication occurring at interphase
after the first division. Hypothesis 5 is the most unorthodox since it invokes a
phenomenon that is believed to be unique. An extra replication of univalent
chromosomes presumably occurs during meiosis in some triploids and species
hybrids (see AVERS1954) since these chromosomes divide equationally at both
meiotic divisions. The postulated behavior on hypothesis 5 differs from the previous studies since it presumes a replication of dyads which have come from
paired bivalents. The ten dyads found in each cell at the end of TI may give rise
to 20 monads which then replicate to produce 20 dyads in each sister cell of the
duet. Interphase cells were observed at microsporogenesis and megasporogenesis
in which each sister nucleus possessed two nucleoli (Figure Ib,c). They could
have arisen either by interphase replication or by the ten dyads separating into
20 monads without replication (hypothesis 4). Although the interphase cells in
microsporogenesis were never seen to divide and their diploid constitution was
inferred from the number of nucleoli, the occurrrence of interphase replication
during megasporogenesis would, if followed by a normal second meiotic division,
lead to the formation of a linear set of four diploid megaspores. These have not
been found but the number of ovules examined at the crucial stage was not great.
On hypothesis 4 there would be two diploid spores at the end of meiosis which
would resemble the diploid interphase cells already described. Although neither
the cytological nor the genetic data permit an unequivocal choice between hypothesis 4 and 5, they clearly indicate that doubling of chromosome number takes
place at the second meiotic division. The postulated interphase replication could
be tested by a comparison of the DNA content in the duets of microspores with
one nucleolus in each interphase nucleus and in the putative diploid duets with
two nucleoli in each nucleus. According to hypothesis 4,the nuclei of the diploid
duets would have a 2n level of DNA while on hypothesis 5 the amount of DNA
should be at the 4n level following replication.
I n addition to production of diploid megaspores, elongate plants also show
variable pollen abortion, and considerable ovule abortion is characteristic of el
ears. Any explanation of the action of the el gene must account for the inviability
ELONGATE MAIZE
511
FIGURE1.-(a) Anaphase I in microsporogenesis of elongate showing stretched or uncoiled
chromonemnta. (b) and (c) Duets of cells fr3m elongate microsporocytes at interphase after
the first meiotic division. The two nucleoli in each putative diploid nucleus are indicated by
arrows. (d) Anaphase I1 in elongate microsporogenesis illustrating the relatively uncoiled condition of the chromosomes. (e) Anaphase I1 in a normal plant showing the contracted chromosomrs characteristic of this stage. Compare with (d). ( f ) Neocentromeres a t MI1 in elongate
microsporogmesis. ( g ) Supernumerary mitoses occurring in young microspores from a n elongate
plant. No replication of the chromosomes has occurred and the ten chromosomes pass a t random
to the two poles. This supernumerary mitosis is the primary cause of the pollen abortion found
in elongate plants.
512
M. M. RHOADES A N D E. DEMPSEY
of pollen and ovules as well as for the aneuploid chromosome numbers observed
in some eggs at the diploid level. Inasmuch as misdivision of the centromeres was
occasionally observed at MI1 in microsporogenesis, it was assumed that a similar
phenomenon took place in megasporogenesis to form telocentric chromosomes.
Plants with 31 chromosomes were, on this supposition, expected to have arisen
from eggs with two telocentric chromosomes. Plants with 32 and 33 chromosomes
could be accounted for by centromere misdivision of more than one chromosome.
These suppositions are reasonable, but cytological examination of 31-chromosome
plants revealed that they had nine trivalents and one quadrivalent at pachynema;
no telocentrics were present and three intact homologous chromosomes had been
contributed to the zygote by the 2n 4-1 eggs from the el parent. When plants with
32 chromosomes were studied, it was found that they possessed eight trivalents
and two quadrivalents; in these cases two nonhomologous chromosomes were
present in a trisomic condition in the 2n 2 eggs. There is no evidence for nondisjunction of individual chromosomes at the, first division since all plump kernels on el ears had zygotes with 20 chromosomes.
Unlike the putative diploid interphase cells observed in microsporogenesis
which do not develop further, diploid megaspores continue development to form
embryo sacs although it is uncertain whether there is a second meiotic division
following interphase replication (hypothesis 5 ) or whether they pass directly
into the first of the three mitotic divisions of gametogenesis (hypothesis 4). I n
either case, nondisjunction may occur when the 20 chromosomes congress onto
a mitotic spindle which normally accommodates ten chromosomes. Failure of
some of the 20 chromosomes to attain proper orientation owing to crowding on
the spindle plate could lead to nondisjunction and thus give eggs with more and
with less than 20 chromosomes. Ovule abortion would come from aneuploid cells
having extreme chromosome unbalance. This explanation of the cause of nondisjunction is a mechanical one. It is perhaps more probable that there is a disturbance in the timing of the normally ordered mitotic events which leads to
faulty orientation and consequently to nondisjunction. That unusual chromosome
behavior does occur in microsporogenesis is evident from the modification of the
coiling cycle, sporadic neocentromere formation, and centromere misdivision.
Pollen abortion varies widely in different el strains but in those with high
abortion a considerable fraction of the haploid microspores were observed to
undergo division without chromosome replication having occurred (Figure lg) .
Inasmuch as the ten chromosomes were distributed to the two poles in an apparent random manner, a vast majority of the resulting spores would be deficient
for one or more chromosomes and hence abort. The inability of the diploid duets
to undergo further development would be an additional cause of poor pollen
production.
A gene causing omission of the second meiotic division was reported in Datura
by SATINAand BLAKESLEE(1935). I n their material a low percentage of both
mega- and microsporocytes underwent a normal second division but the great
majority gave rise to two spores instead of a quartet at the end of meiosis. The
abnormal sporocytes had a prolonged telophase and resting stage following the
+
ELONGATE MAIZE
513
first division, during which time chromosome replication must have occurred in
preparation for the first gametophytic mitosis. In the elongate material the chromosomes at telophase I resemble those of an interphase nucleus. The premature
loss of condensation and the early despiralization may provide the extended resting condition needed for chromosome replication at the interphase between meiotic divisions postulated to occur by hypothesis 5.
The cytological studies of megasporogenesis do not provide critical evidence
for or against the hypothesis of omission of the second meiotic division and that
of an extra chromosomal replication accompanied by a second division. However,
the omission of the second division is the simpler of the two hypotheses and
accounts for the cytogenetic data in as satisfactory a manner as does the hypothesis of an extra replication; perhaps it should be the favored mechanism even
though there are at present no data which afford an unequivocal decision between
the two schemes.
Chromatid interference: The genotypic constitutions of 204 diploid eggs from
el plants have already been presented. They permit a study of chromatid interference. Plants of el el, Lg A / l g a constitution produce diploid eggs of L g A / l g A
and Lg a/lg a genotypes from both 2- and 4-strand doubles in regions 1 and 2.
Three-strand doubles in the same regions yield equal numbers of L g a/lg A and
L g A / l g a diploid eggs. The latter type cannot be distinguished from diploid eggs
originating from single exchanges in region 1 but the origin of the former class
is unambiguous. If there is no chromatid interference, the number of combined
2- and 4-strand combinations should be twice as frequent as the recognizable
%strand doubles. There were 21 L g A/lg A , 16 Lg a/lg a and 22 Lg a/lg A diploid eggs, a proportion consistent with that expected with no chromatid interference.
The crossover regions followed in this study are long enough to permit the
occurrence of undetected double exchanges since 30% or more recombination
occurs in both regions 1 and 2. That double exchanges do take place in region 2
is evidenced by the diploid eggs of lg A / l g A constitution which arise only from
4-strand doubles in (2) and cannot originate from other types of exchanges.
Before accepting the evidence for random chromatid involvement in double exchanges in regions 1 and 2, it is necessary to consider the effect of double exchanges in (1) o r in (2). The diploid eggs from double exchanges in region 1
consist of equal numbers of L g A / L g A and lg a/lg a noncrossover combinations
and of Lg A / l g a eggs which are identical to those coming from single exchanges
in (1 ) . No combinations classifiable as 2, 4, or the identifiable 3-strand doubles
in (1-2) are formed. Likewise double exchanges in region 2 yield no diploid
eggs which could be classified as coming from 2-, 3-, or 4-strand doubles in (1-2).
However, 4-strand doubles in region 2 will lead to the formation of L g a/Lg a
and lg A / l g A compounds; three of the latter were found. These are unique types
which come only from 4-strand doubles. No combinations are produced that are
similar to the double exchanges in 1 and 2. It follows that doubles in either region
1 or in region 2 will have no effect on the question of chromatid interference in
1-2 doubles.
5 14
M. M. RHOADES AND E. DEMPSEY
Triple exchanges with a double in region 1 and a single exchange in region E2
should be exceedingly rare events but if they do occur the ratio of recognizable
2-, 3-, and 4-strand doubles is identical to that coming from double exchanges in
regions 1 and 2. Only from triples with a single exchange in region 1 and a
double in region 2 is there a deviation from the ratio of 2: Ifor the combined
2- and 4-strand doubles to one class of 3-strand doubles. These infrequent triples
produce a ratio of 2: 3 for the above combinations. However, it is improbable that
the number of these triples is high enough to have a measureable effect.
The conclusion that chromatid interference does not exist in the regions of
chromosome 3 investigated appears to be justified although the demonstration
suffers from the fact that 2- and 4-strand doubles could not be differentiated from
one another and were grouped into one class.
Centromere mapping: Centromere mapping in Ascomycetes with ordered spore
arrangement is determined from the frequency of second division segregations.
Since a single crossover between gene and centromere results in metaphase I1
segregation, the frequency of such segregations may be equated to chiasma frequency or to twice the recombination frequency. With unordered tetrads, centromere distances may be calculated from frequencies of tetratype asci if three
1949; PERKINS
1949; W H I T E independent loci are segregating ( LINDEGREN
HOUSE 1950).
The transformation of MI1 segregation frequencies into recombination values
becomes less accurate as distance from the centromere increases since recombination values approach a maximum of 50% while MI1 frequencies approach 67%
in the absence of chiasma interference. SPIEGELMAN’S
(1952) equation for conversion of MI1 frequencies into crossover percentages is based on a modification
of the formula derived by RIZET and ENGELMANN
(1949) and PAPAZIAN
(1951)
which assumes a Poisson distribution of chiasmata. Centromere-gene distances
determined from one half the frequency of MI1 segregation are underestimated
whereas those derived from equations assuming no chiasma interference are overestimated. This relationship has been graphically presented by PERKINS
(1962),
who plotted the observed frequencies of tetratypes in Neurospora and four times
the homozygosis values from attached-X Drosophila against map distance. Theoretical expectancies of crossover distances based on complete interference (one
half of MI1 segregations) and with no chiasma interference were compared with
the observed values. These fell between the two theoretical extremes, indicating
a considerable amount of chiasma interference. With the exception of Aspergillus,
chiasma interference has been reported in all organisms where tetrad or halftetrad analyses have been made.
HOWE’S( 1963) determination of centromere distances in Neurospora tetrasperma, where the asci contain four binucleate ascospores, represents an interesting variant of the method used for those species with &spored asci where the
frequency of MI1 segregation is considered to be twice the crossover distance
from the centromere. In tetrasperma the overlapping spindles of both the second
meiotic division and the third (mitotic) division result in the juxtaposition of
pairs of nonsister nuclei which are cut out into the same ascospore. When no
515
ELONGATE MAIZE
exchange occurs between a given locus and its centromere, all of the ascospores
of an ascus are heterocaryotic (each of the two nuclei carrying a different allele).
Following a single exchange, one half of the asci would be identical to those
coming from no-exchange tetrads (Type I) while the remaining half would have
homocaryotic ascospores (Type 11). Since only one half of the asci have homocaryotic ascospores following an exchange, the frequency of homocaryotic asci
equals the recombination value.
In species of filamentous fungi in which a sexual stage is unknown, it has
been possible to locate centromeres because of the occurrence of haploidization
and mitotic crossing over (PONTECORVO
and KAFER 1958). Single chromosome
arms have been mapped using the relative frequencies of homozygosis of recessive
markers following crossing over. All of the genes on one chromosome are inherited
as a unit in haploidization and the junction of the linkage maps of the two arms
represents the centromere location.
In sexually reproducing organisms where the haploid gametes contain a single
chromatid of the four comprising each bivalent and there is no mechanism for
sequestering the four products of meiosis, the location of the centromere is not
so readily accomplished. It has, however, been achieved in a number of ways.
In the early days of Drosophila genetics the location of the centromeres on the
linkage maps of the two large autosomes was estimated by the absence of chiasma
interference in specific short, adjacent regions. Another method of locating centromeres involves the determination of the linkage of mutant loci with a structural
aberration which has one breakpoint close to the centric region. It was not possible to measure recombination values between the centromere and a given locus
in Drosophila until the discovery of females with attached-X chromosomes (L. V.
MORGAN
1922). Since two X chromosomes were regularly recovered in the eggs,
half-tetrad analyses were possible and the amount of recombination as well as
map distance between the centromere and mutant loci could be calculated from
the genotypic constitutions of the exceptional daughters (ANDERSON
1925; EMERSON and BEADLE1933; BEADLEand EMERSON
1935).
The opportunity provided by the el gene for half-tetrad analysis is comparable
to that furnished by attached-X’s in Drosophila. Only part of the eggs are diploid
in el maize, but a sufficient number are produced for half-tetrad studies. The
array of diploid eggs from tests involving genes on chromosome 3 has already
been presented in another section. Elongate plants heterozygous for sh wx on
chromosome 9 were also testcrossed and gave rise to the following diploid eggs:
ShWx
~
_
Sh Wx
24
shwx_
_
shwx
S_
h W x_
sh wx
28
30
S-h W-x - shwx
_ _ShWx
sh W x
Shwx
Sh wx
35
33
4
sh wx
shWx
2
x=156
Another group of diploid eggs was obtained from el el, Ws, Lgr Gle/ws, lg, gl,
plants, but the triploid progeny was not analyzed for genotypic constitution and
only the phenotypic classification is available. The data are given below:
516
M. M. RHOADES AND E. DEMPSEY
Ws Lg G1
ws 1g gl
ws Lg G1
ws lg G1
371
28
10
12
ws Lg gl
Ws Lg gl
Ws lg gl
2
30
23
2 = 476
The wss (white sheath), Zgl (liguleless), and gl, (glossy) loci lie in the short
arm of chromosome 2 in the order listed with ws near the distal end and gl
closest to the centromere. Analyses of these three sets of data have made possible
the tentative location of the centromeres on the genetic maps of chromosomes
2 , 3 , and 9.
There is a direct relationship between recombination values and the percentage
of homozygosis in diploid eggs from el plants. With no recombination between a
marked locus and the centromere, 50% of the diploid eggs are homozygous for
the recessive allele and 50% are homozygous for the dominant allele. With 10%
recombination, 40% of the diploid eggs would be homozygous for the recessive
allele, 40% homozygous for the dominant allele, and 20% heterozygous. Thus,
for short regions, recombination is equal to 50 minus the percentage of the homozygous recessive class, or one half the frequency of the heterozygous class. With
recombination values greater than 1 OX, the percentage of equational segregation
( A a gametes) no longer represents twice the recombination value because multiple exchanges give rise to undetected crossovers. Table l gives the frequencies
of homozygosis observed in the 3n progenies of el plants heterozygous for the
gene loci listed and the calculated frequencies of equational segregation on the
assumption that the homozygous a a class is equal to the A A . The recombination
percentage between each locus and its centromere is based on one half the frequency of equational disjunction.
The best estimation of centromere position is that determined for chromosome
9 as based on wx homozygosis. The Wx-centromere interval is short and no
distortion of the recombination value by double or triple exchanges should occur.
It may be concluded that the centromere lies approximately 11 map units from
W x . This value is considerably greater than that reported by ANDERSON
and
RANDOLPH(1945) from translocation studies, but crossing over is reduced in
TABLE 1
Recombination values in the gene-centromere segment for seven loci based on
observed percentages of homozygosis in diploid eggs
Gene
,wx
Shl
Lg,
AI
G4
Lg,
WS,
Percent
homozygous recessive
Percent
equational disjunction
38.5*
18.6*
19.4*
13.0*
17.5
13.3
10.9
23.0
63.6
61.2
74.0
65.0
73.4
78.2
* Average of homozygous dominant and homozygous recessive classes.
Percent recombination
(Equational disjunction X l/)
11.5
31.8
30.6
37.0
32.5
36.7
39.1
517
ELONGATE MAIZE
translocation heterozygotes and their value of 2% is certainly less than the
amount occurring in structural homozygotes.
Lower percentages of homozygosis (and higher frequencies of equational segregation) were found for the other loci, which are known to be more distally located. In three cases, the frequencies of equational disjunction were greater than
67%, the theoretical limit when chiasma frequencies follow a Poisson distribution. The complete genotypic analysis of the 2n eggs segregating for genes on
chromosome 3 and for genes on chromosome 9 made possible a calculation of the
distribution of chiasmata in the A-centromere and the Sh-centromere regions.
In neither case did the calculated distribution conform with the Poisson distribution expected with the average chiasma frequencies indicated for each set of data.
The high frequencies of equational disjunction are therefore due to deviations
from a Poisson distribution in which single chiasmata are more frequent than
expected and no-exchange tetrads are less frequent. Comparable deviations in
chiasma frequencies were reported by PERKINS
(1962) for Neurospora and by
KNAPP (1960) for the liverwort Sphaerocarpus. Although the four spores in
Sphaerocarpus are unordered, the regular disjunction of the X and Y chromosomes to opposite poles at AI permitted identification of the equational or reductional segregation of loci in other chromosomes.
Chiasma frequencies were also calculated for the chromosome 2 data from
elongate plants. Since only phenotypes were scored, a number of the recombinant
types are included in the triple dominant class and their frequencies must be
estimated on the basis of detectable crossovers. Although this places undue emphasis on some of the less frequent classes, a reasonable distribution of chiasma
frequencies is obtained. These are listed in Table 2, as are the values determined
for the chromosome 9 and chromosome 3 data. Double chiasmata in the Wxcentromere, Lg,-centromere, and GL-centromere regions cannot be detected and
have been ignored. On the basis of these chiasma frequencies, a corrected recombination value may be obtained. Bivalents with no exchanges give only homozygous diploid eggs and no recombinant strands. Single exchanges lead to 100%
A a eggs and 50% recombinant chromatids. Double exchanges give 50% A a eggs
and 50% recombination. The corrected recombination exceeds the value obtained
from the frequency of equational disjunction by the percentage of 4-strand douTABLE 2
Calculated chiasma frequencies and corrected recombination values in
four gene-centromere intervals
Calculated chiasma frequencies
Region
None
Single
Double
Ws,-Centromere
Lg,-Centromere
Sh,-Centromere
A,-Centromere
11.7
68.1
68.5
60.9
49.5
20.2
Corrected
percent recombination
~
21 .4
33.3
4.9
10.1
5.8
45.6
~~
44.2
39.3
33.4
47.6
518
M. M. RHOADES AND E. DEMPSEY
bles since these are considered to be nonrecombinant on the basis of their reductional segregation but actually contain all recombinant strands.
The objection may be raised that the abnormal chromosome behavior in megasporocytes giving rise to diploid eggs could also affect recombination and the
values obtained would be unreliable. That this is not the case is indicated by a
comparison of recombination in the haploid and diploid eggs from the same ears.
The amount of recombination between Lg and A in the 408 chromosomes of the
204 diploid eggs was 36.7%, while 34.1 % was found in a population of 452 plants
arising from the haploid eggs. A crossover value of 23.7% was found for the
Sh-Wx region in diploid eggs and 99% in the haploid eggs. Moreover, the percentages of recombination in both haploid and diploid eggs of elongate plants
agree well with the standard values. It may be concluded that crossing over is
normal in megasporocytes which are destined to give rise to diploid eggs and the
estimations of centromere distances are considered to be valid.
Centromere mapping has been attempted in other higher organisms. LINDSLEY,
FANKHAUSER
and HUMPHREY
(1956) determined the frequency of homozygosis is diploid eggs of the Mexican axolotl produced when the second meiotic
division was suppressed as a result of refrigeration. Heterozygous females were
mated to recessive males and the triploid offspring arising from the diploid eggs
were scored for three loci. The frequency of equational separation was determined from the percentage of homozygous recessive off spring. The value of approximately 67% equational separation found for all three loci was interpreted
to mean that the frequency of crossing over in the studied regions was high
enough to lead to random assortment. The suggestion was made that all three
genes lie near the ends of their respective arms. This need not be the case if the
number of chiasmata in different tetrads are not in accord with that expected by
the Poisson distribution. For example, the Sh, locus in chromosome 9 is not more
than 35 map units from the centromere region, yet in 64% of the diploid eggs
from elongate plants there was an equational separation of the Sh:sh alleles. This
value is very close to the 67% equational separation which would be realized if
the Sh locus was segregating independently of the centric region but this obviously cannot occur for a region 35 map units long. The conclusion reached by
LINDSLEY
et al. that the three tested loci in the axolotl segregated independently
of their centric regions should not be unequivocally accepted.
A situation somewhat similar to that in the axolotl has been studied in the
honey bee. Diploid progeny are occasionally produced by unmated queens.
TUCKER
(1958) believes these arise by fusion of two of the four products of
meiosis, specifically those arising from different secondary oocytes. The cytological scheme proposed was based on observed percentages of homozygous recessive phenotypes in the offspring of queens heterozygous f o r three unlinked
loci. The very low frequency of homozygous ivory was attributed to a combination of (1) its location close to the centromere, (2) union of nuclei derived from
different secondary oocytes and (3) reductional centromere division at AI. Interference with oviposition led to increased frequencies of impaternate workers on
resumption of egg laying. Since the egg remains at the AI stage until after it is
ELONGATE MAIZE
519
laid, it n'as argued that aging of the egg by inhibition of oviposition might cause
a failure of spindle reorientation which normally occurs at this stage. This would
lead to the union of the two adjacent nuclei resulting from different second
division spindles. The observed frequencies of homozygosis could not be explained
by the occurrence of meiosis in tetraploid tissue or by fusion of the products of
the first cleavage after meiosis. However, TUCKER'S
data could be more simply
accounted for if there was a failure to complete the first meiotic division and a
restitution nucleus was formed. On this hypothesis, a gene close to the centromere
would show very low frequencies of homozygosis. Cytological studies are needed
to discriminate between the failure of the first division and the spindle reorientation hypothesis.
Although the effective suppression of first meiotic division in the unisexual
wasp Devorgilla should lead to estimation of valid centromere distances, it is
uncertain whether or not unambiguous conclusions can be reached with the I
gene studied by SPEICHER,SPEICHER
and ROBERTS
(1965). Distinction between
+/I and
females depends on the percentage of abortion in their progenies.
This varied from 0 to 23.9%. Females with 0-2% egg lethality were considered
to be
while those whose egg abortion ranged from 3.77 to 23.9% were scored
as heterozygotes. The difference between the two classes is not of great magnitude.
In polyploid plants which tolerate a deficiency of one chromosome arm, it is
possible to determine gene-centromere distances by the use of telocentric chromosomes. Plants having a heteromorphic pair consisting of a normal chromosome
and a telocentric and heterozygous for a gene located in this pair of chromosomes
map be testcrossed and the progeny classified for phenotype and for the presence
of the telocentric. Centromeres have been located with respect to genes on chromosomes 6B and 3B of wheat by this method (SEARS1962) and a modification
allowing use of F, data has resulted in placement of the centromere of chromosome 2A (DRISCOLL
1964).
Essentially the same method has been used in maize to locate the centromere
of chromosome 3. Plants heterozygous for mutant loci and for a normal chromosome 3 and two chromosome fragments consisting of the short and long arms were
testcrossed. The progenies were scored for the segregating alleles and for chromosome constitution (RHOADES,
unpublished). I n determining centromere-gene recombination values, the telocentric method has an advantage over the other
methods in that the centromeres are distinguishable by virtue of one being terminal and the other interstitial. Recognition of 4-strand doubles is possible with a
telocentric chromosome, but comparable doubles in tetrad or half-tetrad analysis
produce an apparent noncrossover tetrad. The recombination value for the centromere-A interval determined by the elongate method should be smaller than the
correct value by the number of 4-strand doubles. Such doubles comprise 11.4%
of the tetrads. The uncorrected centromere-A recombination value from elongate
half-tetrads was 37.0%. The sum of 11.4 and 37.0 or 48.4% is a more accurate
estimate of the recombination percentage between the centromere and A. Significantly, the centromere-A recombination value determined by the telocentric
method indicated that A segregated independently of the centromere.
+/+
+/+
520
M. M. RHOADES AND E. DEMPSEY
It is evident from the above discussion that centromere mapping is possible
in higher forms by employing unusual cytogenetic mechanisms. However, as
LINDSLEY
et al. (1956) have emphasized, caution should be exercised before accepting these centromere distances unless it is possible to establish that the amount
of recombination has not been modified by the genetic system itself.
Polyploid series: The el gene has been used to obtain an interesting series of
polyploids. Triploids come irom the cross of el x 2n. Crosses of el by 4n gave
some plump kernels which were tetraploids. By self-pollination 4n el stocks were
obtained. When these were self-pollinated or crossed by 4n plants, 6n plants
came from the union of unreduced 4n eggs with 2n pollen. The 6n plants came
from shriveled kernels while the plump kernels on the same ear gave rise to 4n
plants. Hexaploid plants homozygous for el have been produced but no seed was
obtained when they were self-pollinated. However, plants at the 7n level have
arisen from the cross of 4n el x 6n when unreduced 4n eggs were fertilized by
3n pollen. Pentaploids (5n) have come from crosses of 4n el x 6n where reduced
2n eggs were fertilized by 3n pollen. Vigorous plants are obtained up to the 5n
level but beyound this level the plants are dwarfed and female sterile. Accurate
chromosome counts are difficult to make at high levels of ploidy and the statement
that a plant is hexaploid means only that it is at the 6n level.
It is of some interest that the chloroplasts were of the same size in haploid
leaves as in hexaploid and heptaploid leaves. They are more numerous per cell
in the latter but no size difference is apparent. Whether this represents genetic
autonomy of the plastids or is an adaptation to optimum size for photosynthesis
is uncertain.
This manuscript was prepared while the senior author was a guest investigator in the Department of Genetics of the Australian ,National University and in the Division of Plant Industry,
C.S.I.R.O.,Canberra, Australia. Appreciation of their hospitality is gratefully acknowledged.
SUMMARY
Reduced and unreduced eggs are produced by plants homozygous for the recessive elongate ( e l ) gene. When el plants are crossed by pollen from a 2n plant,
plump and shriveled kernels as well as aborted ovules are found. The plump
kernels give rise to 2n plants and the shriveled kernels to plants at the 3n level.
Eighty-two percent of the unreduced eggs had 20 chromosomes while 18% had
aneuploid numbers ranging from 15 to 23. The aborted ovules presumably come
from megaspores with so extreme a chromosomal unbalance that lethality results.
-All functional pollen of el plants is haploid. Duets of cells with an unreduced
chromosome number are found in some el microsporocytes at interphase. They
do not develop further and are believed to contribute to the poor pollen production. Pollen abortion results from a supernumerary cell division of the microspores at a time before the chromosomes have replicated.-The elongated or
stretched appearance of the chromosomes in microsporocytes of el plants at AI
and AI1 indicated that the chromonemata are relatively uncoiled. Centromere
misdivision and neocentromere formation sporadically occur at MI1.-Half-tetrad
ELONGATE MAIZE
521
analysis of the 20-chromosome eggs indicated random chromatid involvement in
double exchanges.-The percentage of recombination between a gene and its
centromere can be determined from the percentage of diploid eggs homozygous
for the recessive or dominant allele. Centromere recombination values were obtained for the wx and sh, loci in chromosome 9, for the lg, and a, loci in chromosome 3, and for the ws3,lg, and gl, loci in chromosome 2.-Unreduced eggs are
produced by 4n plants homozygous for el. This made possible a polyploid series
up to the heptaploid level-The unreduced eggs originate either by omission of
the second meiotic division or from an extra chromosomal replication during interphase between the first and second meiotic divisions. The simpler hypothesis of
second division suppression is favored over that of an additional replication but
the genetic data are consistent with either hypothesis.
LITERATURE CITED
ANDERSON,
E. G., 1925 Crossing over in a case of attached X chromosomes in Drosophila
melanogaster. Genetics 10: 403-41 7.
ANDERSON,
E. G., and L. F. RANDOLPH,
1945 Location of centromeres on the linkage maps of
maize. Genetics 30: 518-526.
AVERS,C., 1954 Chromosome behavior in fertile triploid aster hybrids. Genetics 39: 117-126.
BEADLE,G. W., 1931 A gene i n maize for supernumerary cell divisions following meiosis.
Cornel1 Univ. Agric. Sta. Mem. 135: 1-12.
BEADLE,
G. W., and E. EMERSON,1935 Further studies of crossing over in attached-X chromosomes of Drosophila melanogaster. Genetics 20 : 192-!206.
DRISCOLL,
C. J., 1964 Wheat centromere mapping by aneuploid F, observations. Agron. Abstr.
13: 65.
EMERSON,S., and G. W. BEADLE,1933 Crossing over near the spindle fiber in attached X
chromosomes of Drosophila melanogaster. Z. Ind. Abst. Vererb. 6 5 : 129-140.
HOWE,H. B., JR., 1963 Markers and centromere distances in Neurospora tetrasperma. Genetics
48: 121-131.
KNAPP,E., 1960 Tetrad analyses in green plants. Canad. J. Genet. Cytol. 2: 89-95.
LINDEGREN,
C. C., 1949 Chromosome maps of Saccharomyces. Proc. 8th Intern. Cong. Genet.
(Hereditas Suppl.) 338-355.
LINDSLEY,D. L., G. FANKHAUSER,
and R. R. HUMPHREY,
1956 Mapping centromeres in the
axolotl. Genetics 41 : 58-64.
MORGAN,
L. V., 1922 Non-criss-cross inheritance in Drosophila melanogaster. Biol. Bull. 42:
267-274.
PAPAZIAN,
H. P., 1951 The incompatibility factors and a related gene in Schizophyllum
commune. Genetics 36 : 441-459.
PERKINS,
D., 1949 Biochemical mutants in the smut fungus Ustilago maydis. Genetics 34:
607-626. - 1962 Crossing-over and interference i n a multiply marked chromosome
arm of Neurospora. Genetics 47: 1253-1274.
PONTECORVO,
G., and E. &FER, 1958 Genetic analysis based on mitotic recombination. Advan.
Genet. 9: 71-104..
R m , H., 1961 Genotypic control of chromosome form and behavior. Botan. Rev. 27: 288-318.
RHOADES,
M. M., and E. DEMPSEY,1966 The effect of abnormal chromosome 10 on preferential
segregation and crossing over in maize. Genetics 53: 989-1020.
522
M. M. RHOADES AND E. DEMPSEY
RHOADES,
M. M., and H. VILKOMERSON,
1942 On the anaphase movement of chromosomes.
Proc. Natl. Acad. Sci. U.S. 28: 433-436.
RIZET,G., and C. ENGELMANN.
19W Contribution i YCtude gbnbtique d u n Ascomyct.te t6traspo&: Podospora anserina (Ces.) Rehm. Rev. Cytol. Biol. VbgCtales 11: 201-304.
SATINA,S., and A. F. BLAKESLEE,
1935 Cytological effects of a gene in Datura which causes
dyad formation in sporogenesis. Botan. Gaz. %: 521-532.
SEARS,E. R., 1962 The use of telocentric chromosomes in linkage mapping. Genetics 47: 983.
SPEICHER,
B. R., K. G. SPEICHER,and F. L. ROBERTS,
1965 Genetic segregation in the unisexual
wasp Devorgilla. Genetics 52: 1035-1M1.
SPIEGELMAN,
S., 1952 Mapping functions in tetrad and recombinant analysis. Science 116:
510-5 12.
TUCKER,
K., 1958 Automictic parthenogenesis in the honey bee. Genetics 43: 299-316.
WHITEHOUSE,
H. L. K., 1950 Mapping chromosome centromeres by the analysis of unordered
tetrads. Nature 165: 893.