Reversion of the mutable allele o2-m(r) (formed by an insertion of the receptor element rbg in the o2 locus) to normal O2 occurs in the presence of the regulatory element Bg as a result of rbg excision from the o2 locus, and phenotypically is expressed in restoration of the normal (vitreous) endosperm structure. In cases when such a reversion takes place during gametophyte development or at the early stages of endosperm development it leads to formation of kernels which are indistinguishable from normals: whole endosperm revertants (WER).

Changing frequency of reversions by selection. The frequency of reversion of o2-m(r) in our initial source of instability designated as 3449o2, derived from selfed generations of a simple hybrid of the o2-m(r)/o2-m(r) +Bg/+Bg x O2/O2 Bg/Bg genotype, can be significantly and rapidly changed by ear selection. Disruptive selection was started in 1990 from ear 89-2911-1 (F5 kernels of this ear contained 8.0 % of WER), and carried out for two generations of selfing. Ears were obtained with WER frequencies from 0.8 to 53.7%, i. e. with gametic frequency equal to 0.3 and 17.9%, respectively (Koterniak V. V., Russian Journal of Genetics, 1995, 31:950-954). The lines with low and high frequency of WER formation were designated respectively as 3449 o2-lf Bg-lf and 3449 o2-hf Bg-hf (or LFWER and HFWER). In the following it will be shown that the additional specification of o2-m(r) alleles and Bg elements is justified.

In 1993-1997 selection for WER frequency was looser since we did not try to obtain HFWER families with WER frequency significantly higher than 50% (in this case it would be difficult to distinguish ears homozygous and heterozygous for o2-m(r)). We also did not try to obtain LFWER lines with WER frequency equal to zero to have the possibility of studying normal descendants of these lines. Thus for planting in 1997 ears were selected with gametic frequencies of reversion of the receptive alleles within the limits of 0.18 - 0.30 and 14.64 - 17.00 % for o2-lf and o2-hf respectively.

Reversion frequency of the mutable o2-m(r) allele is determined by its state. To determine the effect of selection on the state of responsive allele o2-m(r) we analyzed the F1 kernels obtained in 1997 by crossing an o2-R Bg line (courtesy of F. Salamini) with LFWER and HFWER lines lacking the regulatory element as a result of its loss or inactivation (Table 1). The possibility of the loss or inactivation of the regulatory element was reported earlier (Salamini F. et al., Heredity, 1982, 49: 111-115; Koterniak V.V., 1995, Russian Journal of Genetics 31: 950-954). Since the o2-R allele is not mutable, somatic instability observed on the ears obtained in these crosses is conditioned by interaction of the regulatory element Bg present in the o2-R Bg line and the receptive alleles o2-lf and o2-hf. In this experiment besides the normal (WER), variegated and opaque kernels we also counted the kernels, one side of which was variegated while the other was normal (these kernels were designated as 1/2n or 1/2WER). Their formation was reported earlier (Montanelli C. et al., Mol. Gen. Genet., 1984, 197: 209-218; Koterniak V. V., Karaivanov G. P., Genetica, 1991, 27: 1814-1819) as a result of reversion of a mutable allele during first mitotic division of the primary endosperm nucleus.

Table 1. F1 kernel segregation on ears of crosses of an o2-R Bg tester with 3449 o2-lf +Bg and 3449 o2-hf +Bg lines.
 

	Number of kernels+
Ear No	n (WER)	1/2n (1/2WER)	v	o	WER, %++	1/2WER,%
1	2	3	4	5	6	7
o2-R o2-R Bg Bg x o2-lf o2-lf +Bg +Bg crosses
9258p252/o2-lf	-	3	129	4	0	2.27
9264p253/o2-lf	-	2	107	3	0	1.83
(9266/o2-lf)-1	-	10	193	1	0	4.93
(9266/o2-lf)-2	-	2	61	-	0	3.17
9267/o2-lf	-		60	-	0	0
9269/o2-lf	1	4	105	3	0.91	3.64
Total	1	21	655	11	0.15**	3.10***
o2-R o2-R Bg Bg x o2-hf o2-hf +Bg +Bg crosses
9258/o2-hf	1	10	89	3	1.00	10.00
9258p12/o2-hf	1	9	109	-	0.84	7.56
9258p13/o2-hf	9	39	149	11	4.57	19.80
9263p269/o2-hf	3	22	100	-	2.40	17.60
9263/o2-hf	1	31	129	-	0.62	19.25
(9264/o2-hf)-1	3	17	153	-	1.73	9.83
(9264/o2-hf)-2	2	22	121	6	1.38	15.17
(9266/o2-hf)-1	5	17	133	-	3.23	10.97
(9266/o2-hf)-2	2	22	100	-	1.61	17.74
(9267/o2-hf)-1	1	24	131	2	0.64	15.38
(9267/o2-hf)-2	2	17	114	-	1.50	12.78
9269/o2-hf	2	6	85	-	2.15	6.45
(9270/o2-hf)-1	4	20	103	9	3.15	15.75
(9270/o2-hf)-2	3	31	88	1	2.46	25.41
Total	39	287	1604	32	2.02	14.87

+ n (WER), v, o - normal (whole endosperm revertants), variegated and opaque kernels respectively. 1/2n (1/2 WER) - kernels 1/2 of which is normal and the other half is variegated.

++ - % of WER was calculated without considering opaque kernels.

**, *** - significance of differences between o2-R o2-R Bg Bg x o2-lf o2-lf +Bg +Bg and o2-R o2-R Bg Bg x o2-hf o2-hf +Bg +Bg crosses at P=0.01 and P=0.001, respectively. Here and in the following tables significance of differences was determined by t-criterion.

Analysis of F1 kernels on the ears obtained by crossing o2-R Bg with o2-lf +Bg and o2-hf +Bg lines showed significant differences both in WER frequency and in 1/2WER content. Since in these crosses the regulatory element was represented by the standard Bg but not by the regulatory elements existing in the lines undergoing disruptive selection for WER frequency, it is possible to make the conclusion that frequency of formation of WER is conditioned by the receptive alleles o2-lf and o2-hf present respectively at LFWER and HFWER lines. This means that selection for WER frequency resulted in the changes in the state of the initial o2-m(r) allele. The change in state phenomenon is known for different systems of transposable elements and for the o2-m(r) allele and was reported earlier by Montanelli C. et al (1984) under selection for a different pattern of kernel variegation.

Data obtained also demonstrate that some ears have small numbers of opaque kernels (Table 1). Their presence can be explained by the loss or inactivation of the regulatory element, the possibility of which was mentioned above. It is interesting to note that in some cases we observed kernels 1/2 of which were variegated and the other half opaque, showing that this loss or inactivation can take place at the same stages of endosperm development at which the 1/2WER formation occurs.

Differences in reversion frequencies of receptive alleles contained in LFWER and HFWER lines seem not to be connected with presence or absence of other genes (modifiers). This conclusion can be reached by comparing gametic frequencies of mutable alleles obtained when analyzing selfed ears of the hybrids from the crosses of LFWER and HFWER lines with the opaque2 and normal testers carrying alleles which did not pass through selection for WER frequency (Table 2). The testers were homozygous for o2-m(r) or O2 alleles and did not contain regulatory element Bg.

In Table 2 total data of kernel segregation are presented on selfed ears of crosses of LFWER and HFWER lines with normal and o2-m(r) testers of 502 line background. Here for analysis ears were used with a segregation ratio of normal and variegated kernels to opaque ones not significantly different from 3:1 (for descendants of o2-m(r) +Bg tester) or from 15:1 (for descendants of O2 +Bg tester). These ratios indicate ears that were obtained by selfing plants heterozygous for one copy of the regulatory element, and allow calculating gametic frequencies (gf, %) by the formula gf=100n/(2(n+v)) for descendants of the o2-m(r) +Bg tester, and by the formula gf=100(n-12/15(n+v))/(6/15(n+v)) for descendants of the O2 +Bg tester, where n and v are the number of the normal and variegated kernels respectively.

Table 2. Results of kernel segregation on ears obtained by selfing the hybrids of 502O2 +Bg and 502o2-m(r) +Bg testers with 3449 o2-lf Bg-lf and 3449 o2-hf Bg-hf lines (1997 year data).
 

		Number of kernels
Genotype	Number of ears	n	v	o	Gametic frequency (gf),%*
(O2/o2-lf Bg-lf/+Bg)@	16	5442	1227	364	4.00a
(o2-m(r)/o2-lf Bg-lf/+Bg )@	9	294	2693	893	4.92a
(O2/o2-hf Bg-hf/+Bg) @	11	3595	530	232	17.88b
(o2-m(r)/o2-hf Bg-hf/+Bg)@	13	900	2510	954	13.20c

* - a common letter at the means indicates insignificance of the differences between them (P=0.05).

In the F2 generation of the crosses of LFWER lines with O2 +Bg tester the number of WER was lower, though insignificantly, as compared with selfed progenies of the crosses of the same lines with the o2-m(r) +Bg tester. This can be explained by the fact that in the case of o2-m(r) +Bg tester a portion of gametes which took part in WER formation carried alleles (brought in by the 502 o2-m(r) +Bg tester) not changed by selection (i. e. with a higher frequency of receptive allele reversion). Accordingly in the F2 generation of the crosses of HFWER lines with O2 +Bg tester the number of WER was significantly higher as compared with selfed progenies of the same lines with o2-m(r) +Bg tester. This also can be explained by the fact that in the case of o2-m(r) +Bg tester a portion of gametes which took part in WER formation carried alleles (brought in by the 502 o2-m(r) +Bg tester) not changed by selection (i. e. with a lower frequency of receptive allele reversion).

In case a high frequency of WER was determined by modifier genes not linked with the o2 locus, it was expected that the selfed progeny of the O2 /o2-hf Bg-hf/+Bg crosses would have lower WER frequency in comparison with the same trait of selfed HFWER lines. However, the calculated genetic frequency of WER on the selfed O2 /o2-hf Bg-hf/+Bg crosses (17.88%, see table 2), not only was not lower than the gametic frequency of WER on selfed HFWER ears (in 1997 its value was equal to 11.40%), but even exceeded the latter (probably due to genotypic differences between HFWER lines and their crosses with 502 O2 +Bg line).

Effect of the dose of the regulatory element Bg on frequency of reversion. Expression of mutability at the o2 locus is not connected with dosage effects for the standard Bg (Montanelli C. et al., 1984). However, a regulatory element (Bg-7b3) was described, one dose of which was insufficient for rbg excision (Motto M. et al., Maydica, 1989, 34: 107-122).

Analysis of the crosses obtained with participation of LFWER and HFWER lines permitted us to make some conclusions about the effect of selection for WER frequency on activity of regulatory elements Bg-lf and Bg-hf, expressed in effects of different doses of these elements on reversion frequencies of the mutable o2 alleles studied.

To test the effect of different doses of the regulatory elements in LFWER and HFWER lines we made crosses of these lines with opaque2 testers which contained the receptive alleles and lacked the regulatory element (Table 3). By comparing the WER frequency on selfed ears of LFWER and HFWER lines with that on ears obtained from crossing these lines (taken as male parent) with opaque2 tester it was possible to compare the effect of the regulatory element in 3 and 1 doses respectively. By comparing the frequency of WER on the ears obtained from crossing LFWER and HFWER lines (used as male parent) with o2-m(r) +Bg tester with the frequency of WER on ears of reciprocal crosses, we compared the effect of regulatory element in 1 and 2 doses respectively. In this analysis only exactly reciprocal ears were included. Accordingly in studying effects of 1 and 3 gene doses only exactly paired ears were used. Besides the o2-m(r) +Bg tester which contained the o2 allele not exposed to selection (502 o2-m(r) +Bg line), a 3449 o2-hf +Bg line (i.e. the HFWER line lacked regulatory element Bg) was used as a tester. Under selfing, this line was characterized by stable opaque endosperm structure.

Data obtained showed significant differences between LFWER and HFWER lines in the dependence of a receptive allele reversion frequency on the dose of the regulatory element. Different doses of the regulatory element Bg-lf present in LFWER lines did not significantly affect frequency of WER formation. Insignificant differences in WER frequency in the crosses of varied doses of this regulatory element were observed both for each year separately and in comparison between the years in which crosses with varied composition of receptive alleles were studied (in Table 3 significance of differences in WER frequency between the years is not shown). This indicates that by the dosage effect for Bg-lf resembles standard Bg. We also note the resemblance between the receptive allele o2-lf of LFWER lines and the receptive allele o2-m(r) of the standard source of Bg: the mean gametic frequency of reversion in endosperm tissue is 0.86% for the first (see Table 4) and 0.78 for the latter (Salamini F., Cold Spring Harbor Symp. Quant. Bio., 1981, 45: 467-476).

In contrast with the Bg-lf element, frequency of WER formation depends strongly and positively on the dose of Bg-hf, the regulatory element present in HFWER lines. Thus on the ears obtained by the crosses of HFWER lines with 502o2-m(r) +Bg tester, 2 and 3 doses of Bg-hf conditioned a WER frequency respectively 5.4 and 6.2 times higher in comparison with that observed under 1 dose of the same regulatory element.

Results of the crossing of LFWER and HFWER lines with 3449 o2-hf +Bg tester were analogous to the results obtained by crossing with 502 o2-m(r)+Bg tester. Moreover, results of crossing with the 3449 o2-hf +Bg tester showed that observed differences in number of WER are conditioned not by the dose of the receptive allele (or by dose of the receptor element) but by the dose of the regulatory element. Thus the gametic frequency of reversion of the mutable allele o2-hf which leads to formation of the WER on the selfed ears of 3449 o2-hf Bg-hf lines (3 doses of o2-hf and 3 doses of Bg-hf) was equal to 12.34%, which is 3.7 times higher than the gametic frequency of reversion of o2-hf on the ears obtained by crossing 3449 o2-hf +Bg (female parent) with 3449 o2-hf Bg-hf (3 doses of o2-hf and 1 dose of Bg-hf).

Table 3. Kernel segregation on ears of selfed LFWER and HFWER lines and their crosses with 502 o2-m(r) +Bg and 3449 o2-hf +Bg testers (1994-1997 years data).
 

			Number of kernels
Selfing or crossing	Bg dose	No. of ears	n (WER)	v	o	Gametic frequency of WER, %+	Gametic frequency of "o", %
1	2	3	4	5	6	7	8
1994
502o2-m(r) +Bg x LFWER	1	12	15	2054	7	0.72a++	0.34
LFWER@	3	12	70	2384	-	0.95ab	-
502o2-m(r) +Bg x HFWER	1	16	77	4002	39	1.89b	0.95
HFWER@	3	16	1232	2255	-	11.78c	-
1995
3449 o2-hf +Bg x LFWER	1	11	12	1443	-	0.89a	-
LFWER@	3	11	55	1866	-	0.95a	-
3449 o2-hf +Bg x HFWER	1	13	61	1768	13	3.34b	0.71
HFWER@	3	13	937	1595	-	12.34c	-
1997
502o2-m(r) +Bg x LFWER	1	14	32	2947	7	1.07a	0.23
LFWER x 502 o2-m(r) +Bg	2	14	59	3042	-	0.95a	-
502o2-m(r) +Bg x HFWER	1	10	62	2249	13	2.68b	0.56
HFWER x 502o2-m(r) +Bg	2	10	677	1641	31	14.60c	0.81+++

+- % of WER was calculated without considering opaque kernels.

++ - - a common letter at the means indicates insignificance of the differences between them for each year separately (P=0.05).

+++ - for HFWER x 502o2-m(r) +Bg ears calculation was performed by the formula gf=100(o/2(n+v+o).

It is interesting to note that in progenies of lines with high WER frequency in comparison with the progenies of lines with low WER, the frequency of emergence of derivatives lacking the active regulatory element Bg was also higher. Data in Table 3 show that the frequency of formation of opaque kernels (as a result of the loss or inactivation of the regulatory element) in the crosses of HFWER lines with the o2-m(r) +Bg tester was higher than in the crosses of LFWER lines with the same tester. The frequency of ears which were heterozygous for the regulatory element Bg was also higher in HFWER lines. Thus from 114 ears of HFWER lines and 95 ears of LFWER lines studied in 1992-1996 we found 4 heterozygous ears of HFWER lines and 1 heterozygous ear of LFWER lines, giving gametic frequency of their formation equal to 1.75 and 0.53 respectively. Among the causes which determined the enhanced formation of derivatives lacking the regulatory element at the HFWER lines could be: i) changes in Bg-hf which enhance frequency of its inactivation, and ii) higher frequency of the loss of Bg-hf, e.g. as a result of its nonreplicative transposition. Since the receptor element rbg is a deletion derivative of the regulatory element Bg which lost its ability for autonomous transposition (Hartings H. et al., MNL, 1992, 66: 20-21), we can assume that the product of the regulatory element Bg-hf not only increases the transposition of the receptor element but also the transposition of the regulatory element itself. This in turn increased the possibility of the loss of the latter as a result of nonreplicative transposition.

Proceeding from the above it is possible to conclude that disruptive selection for WER frequency resulted in changes of state of both the receptive allele and the regulatory element. Change in state of the initial receptive allele o2-m(r) led to new receptive alleles o2-lf and o2-hf which determine low and high frequency of rbg excision respectively. Change in state of the regulatory element Bg resulted in change of its ability to induce excision of the receptive element from the o2 locus, expressed by dependence of the reversion frequency or the mutable alleles on the dose of the regulatory element. Frequency of rbg excision depends weakly and insignificantly on the dose of Bg-lf. In contrast with this, the frequency of rbg excision is characterized by strong positive dependence on the dose of Bg-hf.

Rapid derivation (during two generations of selfing) of the changed forms of transposable elements as a result of disruptive selection for WER frequency shows existence of significant heterozygosity in the state of the transposable elements in our initial source of instability used for selection. Since parental lines of this source were characterized by a high level of inbreeding and were homozygous for the receptive allele (the genotype of the female parent was o2-m(r)/o2-m(r) +Bg/+Bg), or for regulatory element (the genotype of the male parent was O2/O2 Bg/Bg), it is possible to conclude that the indicated heterogeneity is the result of inherent instability of transposable elements. Among the causes which determined this heterogeneity and led to the changes of the transposable elements could be self-inflicted intra-element deletions and changes in their pattern of methylation (Schiefelbein J. W. et al., Proc. Natl. Acad. Sci. USA, 1985, 82: 4783-4787; Schwarz-Sommer Zs. et al., EMBO J., 1985, 4: 2439-2443; Schwartz D., Dennis E., Mol. Gen. Genet., 1986, 205: 476-482).

Expressivity of some quantitative traits in lines with high and low frequency of reversion. For evaluation of the effect of disruptive selection for WER frequency on quantitative traits we studied some quantitative traits of LFWER and HFWER lines: kernel weight, volume and density; number of kernels per ear; number of leaves on the main stalk and the length of the period from emergence of seedlings to flowering of male inflorescences (Tables 4 and 5).

Kernel weight, density and volume was determined on samples of 50 variegated kernels taken from the seed remnants of the selfed ears studied in 1992-1996. Kernel volume was determined by liquid (purified petrol) displacement. For determination of the number of leaves and date of flowering, the LFWER and HFWER families were planted in 1997 in two terms with an interval of 6 days. To facilitate leaf counting the fifth and tenth leaves were marked.

Data in Table 4 show that on average gametic frequency of WER in HFWER lines was 12.6 times higher than in LFWER lines. Kernel weight and volume in lines with a high frequency of WER was more than 30% higher in comparison with lines with a low frequency. Kernel density also was higher in HFWER lines though the differences in this trait were expressed less clearly and on average were equal to only 2%. Number of kernels on the ears of LFWER and HFWER lines was approximately the same.

Table 4. Expressivity of some quantitative traits in lines with low (LFWER) and high (HFWER) frequency of WER formation.
 

Year	Line	Number of ears	WER, %	WER gametic frequency, %	50 kernel weight, g	50 kernel volume, cm3	Kernel density, g/cm3	Number of kernels per ear
1992	LFWER	3	1.09*	0.36*	5.19	4.10	1.262	263.7
	HFWER	2	48.10	16.03	5.53	4.30	1.284	278.5
1993	LFWER	27	2.22*	0.74*	4.89*	3.84*	1.269*	220.1
	HFWER	11	30.19	10.06	6.71	5.16	1.302	256.2
1994	LFWER	12	3.39*	1.13*	5.28*	4.16*	1.271*	220.5
	HFWER	17	33.41.	11.14	6.77	5.20	1.303	220.1
1995	LFWER	12	2.95*	0.98*	5.19*	4.03*	1.290	192.1
	HFWER	13	33.29	11.10	7.20	5.34	1.298	177.0
1996	LFWER	4	2.52*	0.84*	5.64*	4.45*	1.268	113.3
	HFWER	6	27.41	9.14	7.17	5.60	1.281	134.5
Total	LFWER	58	2.57*	0.86*	5.10*	4.00*	1.273*	209.3
	HFWER	49	32.52	10.84	6.87	5.29	1.298	208.7
HFWER as % of LFWER			1265.4	1260.5	134.7	132.3	102.0	99.7

* - significance of the differences between LFWER and HFWER lines (P(0.05)

Lines with a high frequency of WER had more leaves and were characterized by later flowering of the male inflorescences (Table 5).

It is well known that the o2 gene significantly affects kernel weight and the mutant forms are characterized by reduced kernel weight. The effect of the o2 gene on kernel volume is less definitive and depending on the material used can also be significant (see for example Arnold J. M. et al., Crop Sci., 1977, 17: 362-366; 421-425). Proceeding from this we can assume that different alleles of the o2 gene, i.e. o2-lf and o2-hf, influence kernel weight and volume differently, contributing to existing differences on these traits between LFWER and HFWER lines.

Table 5. Number of leaves and length of period from emergence to flowering in lines with low (LFWER) and high (HFWER) frequency of WER formation.
 

Lines	Number of plants	Number of leaves per plant	Number of days to flowering
First term planting
LFWER	17	12.9***	57.6*
HFWER	19	14.0	58.9
Second term planting
LFWER	69	12.6***	56.4***
HFWER	74	13.3	59.3

* ,*** - significance of the differences between LFWER and HFWER lines at P=0.05 and P=0.001 respectively.

Influence of o2-lf and o2-hf on other quantitative traits also can not be excluded if it is presupposed that the product of the o2 gene, being the strong transcriptional activator of the b-32 gene (Lohmer S. et al., EMBO J., 1991, 10:617-624), also takes part in regulation of activity of other genes. In this connection it is necessary to mention the report of Genga A. et al. (MNL, 1995, 69:102), in which it was established that both the structural zein genes and the regulatory o2 gene are expressed not only in the endosperm but also in male inflorescences.

Besides the indirect influence of the transposable elements Bg and rbg on quantitative traits studied (via the o2 gene), the possibility also exists of their direct involvement in the expression of these traits if the genes which participate in the formation of quantitative traits have insertions of Bg and rbg elements. In case of the changes in state of the transposable elements (e. g. as a result of selection) they will change the activity of these genes and as a result of this the expression of quantitative traits.

About the evolutionary role of transposable elements. Summing up the data obtained it is possible to conclude that disruptive selection for WER frequency resulted in changes of state both of the receptive allele and of the regulatory element which affected the frequency of WER formation. Disruptive selection also resulted in changes of some quantitative traits of the lines obtained, changes which at least partially can be connected to the state of transposable elements.

We can assume that analogous phemomena exist in the case of other genes (in the first place regulatory genes) controlled by different systems of transposable elements. This means that transposable elements in genomes of the organisms which constitute a certain population permit the factors exerting selection pressure on this population to cause significant and rapid changes of the genomes of the organisms. The basis of these changes is the instability of transposable elements, which determine the formation of new states of regulatory and receptor elements, i.e. their heterogeneity. The consequence of heterogeneity of transposable elements is, on the one hand, significant diversity in the expression of the genes under their control and, on the other hand, the possibility of rapid genetic changes in case a selection factor is present in the medium.

Earlier it was proposed that transposable elements are generators of genetic diversity in life-threatening stress conditions (McClintock, Science, 1984, 226: 792-801). Stress conditions lead to the activation of cryptic regulatory elements which in turn result in a high level of excision and transposition of receptor elements. The main source of genetic variability in this scenario is the changes in nucleotide sequences of coding and regulatory regions resulting from imprecise excision of receptor elements (see review of Wessler, Science, 1988, 242: 399-405).

Undoubtedly stress conditions enhance genetic variability of organisms. However, available data show that the formation of new states of transposable elements, as a result of their inherent instability, causes significant genetic diversity and gives sufficient material for evolutionary changes of organisms even in the absence of stress factors. The necessary condition for such changes is the presence of a selection factor affecting the traits whose expression is controlled by transposable elements. In this scenario the main source of genetic variability is the changes of transposable elements and changes in state of mutable alleles caused by the action of the transposable elements.
 

Return to the MNL 73 On-Line Index
Return to the Maize Newsletter Index
Return to the MaizeGDB Homepage