|
|
||||||||
a USDA–ARS National Small Grains Germplasm Research Facility, 1691 S. 2700 W., Aberdeen, ID 83210
b USDA–ARS, Northern Crop Science Lab., 1307 N. 18th St., Fargo, ND 58105
c Dep. of Plant Pathology, Walster Hall, North Dakota State Univ., Fargo, ND 58105
d Dep. of Plant Sciences, Loftsgard Hall, North Dakota State Univ., Fargo, ND 58105
e Dep. of Agriculture, Univ. of Arkansas at Pine Bluff, AR 71601
* Corresponding author (phil.bregitzer{at}ars.usda.gov).
| ABSTRACT |
|---|
|
|
|---|
Abbreviations: BC, backcross FAN, free amino nitrogen PCR, polymerase chain reaction
| INTRODUCTION |
|---|
|
|
|---|
The productivity and quality characteristics of modern cultivars are the result of a lengthy, incremental process of selection for particular allelic combinations at multiple loci. The introduction of random somaclonal variation would be expected, and has been demonstrated, to have primarily negative consequences. In barley (Hordeum vulgare L.), simple passage through tissue culture has been shown to significantly reduce yield and malting quality in a majority of derived lines (Bregitzer and Poulson, 1995; Bregitzer et al., 1995), although lines with negligible changes can also be recovered (see Bregitzer et al. [2002] and references therein). The process of transformation (in barley as well as in other cereals) imposes additional stress that exacerbates somaclonal variation (Bregitzer et al., 1998; Choi et al., 2000a; Choi et al., 2000b; Choi et al., 2001; Horvath et al., 2001; Schuh et al., 1993), and transgenic lines derived by self-pollination of the original transgenic event typically have significant and sometimes severe (greater than 50%) yield reductions.
In the context of mainstream agricultural production, even slight performance reductions are generally unacceptable. Furthermore, somaclonal variation complicates evaluations of transgene-encoded phenotypes by introducing multiple unrelated sources of variability that are confounded with the effects of transgene insertion and expression. Transgenic germplasm thus will be of most value for genetic investigations and for plant breeding if somaclonal variation is minimized.
The generation of somaclonal variation can be reduced by technical improvements to the transformation process (Bregitzer et al., 2002), but a more generally applicable approach may be simply to use backcrossing to eliminate variant alleles in favor of wild-type alleles present in the recurrent parent. A lengthy process involving many backcrossing cycles would be necessary to completely eliminate variant alleles, but even a short process involving two crosses to the original cultivar used to produce the transgenic parent should produce significant improvements. This would theoretically eliminate 75% of the variant alleles that differentiate the transgenic line from its wild-type parent, thus facilitating meaningful comparisons. Furthermore, it is possible that the reduction of epigenetic alterations may be more rapid than predicted based on quantitative genetic models. A particularly intriguing example is that of a dwarf mutant of rice (Oryza sativa)—putatively associated with altered methylation patterns—that was stably transmitted by self-pollination but which could not be recovered on outcrossing to a parent of normal height (Oono, 1985).
Several studies have reported performance improvements via crosses and backcrosses of transgenic parents (Horvath et al., 2001; Shao et al., 2006), but these studies did not use the original wild-type cultivar as the recurrent parent. This study was conducted to determine the utility of a simple and rapid breeding scheme—a single backcross of transgenic barley plants to the original parent used in their production—for the recovery of recurrent parent phenotype as measured by variation in agronomic and malting quality characteristics.
| MATERIALS AND METHODS |
|---|
|
|
|---|
Eight lines (primary transgenic lines), two derived from each of four transgenic events, were derived directly from transgenic plants via self-pollination. The two lines from each event were designated as lines "A" and "B" and represented random selections among T0 plants arising from different sectors of the callus derived from that event. The data in this report were collected from field tests of T2:4 and T2:5 lines in 2005 and 2006, respectively. Seed for these tests was produced from field increases at Aberdeen, ID, of T2:3 and T2:4 seed in 2004 and 2005, respectively. T2:3 seed was produced in the greenhouse at Aberdeen, ID, from small populations (5 to 10 plants) in the winter of 2003–2004. Prior generation advance had been conducted via single seed descent in the greenhouse in Fargo, ND. These lines were produced by cotransformation with pAHC25 (Christensen and Quail, 1996), which includes the selectable marker bar, in combination with either maize-ubiquitin-driven TRI101 (primary transgenic lines pUBR1-II and pUBR1-III) (Manoharan et al., 2006) or PDR5 (primary transgenic lines pUBR2-I and pUBR2-II). TRI101, pUBR1, and PDR5 have been previously described (Balzi et al., 1994; Kimura et al., 1998; Okubara et al., 2002). pUBR2 is identical to pUBR1 except for the substitution of PDR5 for TRI101.
The remaining 35 lines were derivatives of a single backcross (BC) to Conlon (Conlon/T3//Conlon), and the data in this report were collected from field tests of BC1F2:5 and BC1F2:6 lines in 2005 and 2006, respectively. Seed for these tests was produced from field increases at Aberdeen, ID, of BC1F2:4 and BC1F2:5 seed in 2004 and 2005, respectively. BC1F2:4 seed was produced in the greenhouse at Aberdeen, ID, from small BC1F2:3 populations (8–16 plants) in the winter of 2003–2004. Ten transgenic and 7 null (nontransgenic) segregant lines were derived from crosses to pUBR1 lines, and 11 transgenic and 7 null segregant lines were derived from crosses to pUBR2 lines. Since both null segregant and transgenic backcross-derived populations include somaclonal variation inherited from the primary transgenic parent, comparisons of null-segregant and transgenic line performance enable estimation of phenotypic variability induced by transgene insertion and/or expression. For the purposes of this report, a primary transgenic parent and the backcross-derived lines derived from that parent are referred to as a family. A complete listing of transgenic lines and family relationships can be found in Tables 1 and 2 .
|
|
The expression of the selectable marker, bar, in all the backcross-derived lines, and by inference in all the primary transgenic lines, was demonstrated by the success of the selection process described above. In addition, the expression of TRI101 has been demonstrated in the primary transgenic parents and both of two backcross-derived lines tested via northern and immunoblot assays (Manoharan et al., 2006). Northern assays have shown PDR5 transcription in all primary transgenic lines (Dahleen and Manoharan, unpublished data).
Agronomic Analyses
Field trials were conducted in 2005 and 2006 at Aberdeen, ID, and in 2006 at Langdon, ND. Supplemental irrigation was used at both sites. The experimental design was a randomized complete block, with six replicates at Aberdeen and five replicates at Langdon. Each plot consisted of a single 3-m row. Planting was done with four-row headrow drills. At Langdon, rows were planted on 30.5-cm centers. At Aberdeen, rows were planted on 35.6-cm centers, and each plot was separated by a single row of wheat (T. aestivum L.). Seeding rates were 4 g per plot at Langdon and 8 g per plot at Aberdeen. Heading date (visual estimate of the date of 50% spike extrusion from the boot) and plant height (to the top of spikes not including awns) were recorded. Plots were harvested using single-row binders and small-plot threshers. Following harvest, seed samples were cleaned and analyzed for yield, test weight, and percentage plump kernels (defined as the percentage of kernels retained on a 2.38 x 19.1-mm sieve).
Malt Analyses
These analyses were conducted on the primary transgenic parents and on two transgenic backcross-derived lines from each family. Limited resources prevented analysis of grain from all plots, so at both locations two samples, each a composite of equal amount of grain from two replicates, were analyzed for each tested line.
Barley protein was determined with a Leco Nitrogen Analyzer model FP528 (Leco Corporation, St. Joseph, MI). Barley moisture was determined according to American Society of Brewing Chemists method Barley 5AC (ASBC, 1992). Thin kernels, passing through a sieve with 1.98 x 19.00-mm slotted openings, were removed before malting. Micromalting was performed in duplicate on each barley sample according to our standard method (Karababa et al., 1993). Time required to reach 45% steep-out moisture was first determined by pilot-steeping a 10-g sample. Steeping of 80-g samples was at 16°C with a 1 h air rest included for each 12 h of steeping. The steep water was aerated 6 min h–1. Germination was for 4 d at 16°C and
95% relative humidity. Samples were turned daily by hand to prevent matting, and sample weight was adjusted to 45% moisture with distilled water. Kilning was conducted in a forced-air laboratory kiln. Total kiln time was 24 h, during which temperatures were ramped from 49 to 85°C. Rootlets were removed from the kilned malt before analysis.
Malt moisture, extract, wort soluble protein, wort color, free amino nitrogen (FAN), wort viscosity, and wort β-glucan were determined by ASBC methods Malt-3, Malt-4, Wort-3, Malt-5A, Wort-9, Wort-12, Wort-13, and Wort-18, respectively (ASBC, 1992). The ratio of soluble/total protein was calculated using the value for barley protein. Alpha-amylase and diastatic power were determined by a ferricyanide-reducing sugar method using flow injection, as previously described (Karababa et al., 1993).
Data Analysis
Data were analyzed by SAS v. 8.0 Proc GLM (1999, SAS Institute Inc., Cary, NC). The models used to examine line performance data included the following sources of variability: environment (Aberdeen 2005, Aberdeen 2006, and Langdon 2006), replicate-within-environment, line, and line x environment. Replicate-within-environment, environment, and line x environment were considered random, and line x environment was used as the error term. The data were discarded from one replicate at Aberdeen in 2005 because of extreme soil variability and from 25 plots at Langdon that suffered damage from herbicide drift. The data are therefore presented as least-squares means.
The performance of primary transgenic or backcross-derived line performance was compared to Conlon performance using one-tailed Dunnett's comparisons for traits showing a clear, unidirectional change in the primary transgenic parents relative to Conlon (most traits); otherwise, they were analyzed with two-tailed Dunnett's tests. Since the objective of this study was to determine whether somaclonal variation present in the primary transgenic lines remained in their backcross-derived progeny, it was important to avoid a type II error (failure to reject the null hypothesis H0: mean of Conlon = mean of derived line), and a conservative p value of >0.2 was used as the basis for declaring that means were equal.
| RESULTS AND DISCUSSION |
|---|
|
|
|---|
Quantification of heading date, plant height, grain yield per plot, test weight, and plump kernel percentage from Aberdeen and Langdon showed results that were generally consistent with the visual observations conducted at Aberdeen (Tables 1 and 2). The primary transgenic lines pUBR2-IIA and -IIB showed the fewest instances of significant differences from Conlon, and the magnitude of significant differences was generally less than that seen for other lines. All primary transgenic lines except for pUBR2-IIA and -IIB headed later than Conlon. These two lines, plus pUBR2-IA, also did not differ from Conlon for the percentage of plump kernels. Plant height was less than Conlon except for pUBR1-IIB, pUBR1-IIIA, and pUBR2-IIA. All primary transgenic lines had lower grain yield and test weight than Conlon, but the smallest reductions were seen for pUBR2-IIA and -IIB. The trait showing the greatest reduction in the primary transgenic lines was yield, which ranged from 57 to 84% (with a mean of 69%) of that recorded for Conlon. This is also consistent with our previous assessments of tissue culture–derived and transgenic line performance (Bregitzer and Poulson, 1995; Bregitzer et al., 1998; Bregitzer et al., 2006). It is not surprising that yield would be more affected than other agronomic traits, since it is arguably dependent on the expression of a greater number of genes than most other traits.
In contrast to the performance of the primary transgenic lines, the backcross-derived lines showed few instances of significant differences from Conlon, even when taking the conservative approach to avoiding a type II error by declaring significant differences only at a p value of 0.2 or less (Tables 1 and 2). Relative to Conlon, four lines headed later, none were shorter, five yielded less, one had lower test weight, and none had a lower percentage of plump kernels. Despite the general lack of significant differences from Conlon, careful examination of the data suggested an overall trend for lower performance yields. There was a nonnormal distribution of individual backcross-derived line means about the mean of Conlon, and mean yields of the eight backcross-derived families ranged from 90 to 97% of Conlon (with an overall mean of 94%). Furthermore, perusal of the data presented in Tables 1 and 2 reveals rough correlation between the performance of transgenic parents and their backcross-derived lines. Consistent with such a correlation, the highest-yielding primary transgenic line (pUBR2-IIA) produced the best-yielding family of backcross-derived lines, and the lowest-yielding primary transgenic line (pUBR1-IIIB) produced the lowest-yielding family of backcross-derived lines. Thus, rather than being an artifact of sampling error, this trend may be viewed as the expected result if it is assumed that the yield depression in the primary transgenic parents is caused by alterations at many loci, each having small, equivalent, additive, and negative effects. If it is further assumed that the recovery of recurrent parent performance is proportional to the degree to which the recurrent parent genome replaces that of the primary transgenic parent, then the predicted yield loss after a single backcross would be 25% of the original yield loss relative to Conlon. Although these assumptions are undoubtedly simplistic, it is interesting that they predict a mean yield for backcross-derived lines of 548 g plot–1, very similar to the observed performance of 559 g plot–1.
Comparisons of the data in Tables 1 and 2 for backcross-derived line performance indicate no association of the presence or absence (null segregants) of transgenes on any measure of agronomic performance. Independent statistical analyses that excluded data for Conlon and the primary transgenic lines also failed to detect differences between transgenic and null segregant lines for grain yield (model and data not shown). These results suggest that the main source of variability in the primary transgenic lines relative to Conlon was somaclonal variation rather than effects of transgene insertion or expression. This is consistent with previous studies of transgenic barley and wheat (Barro et al., 2002; Bregitzer et al., 1998; Bregitzer et al., 2006; Zhou et al., 2003), although other studies have detected phenotypic alterations in transgenic plants as a result of transgene insertion or expression (Horvath et al., 2001; Lee et al., 2003).
Malt Analyses
The barley protein content of all primary transgenic lines showed significant increases relative to Conlon (Tables 3
and 4
). Soluble protein levels, the ratio of soluble to total protein, and FAN levels were significantly elevated in pUBR2-IA and pUBR2-IB. The FAN level of pUBR1-IIIA showed a slight increase.
|
|
-amylase and diastatic power). Malt extract levels showed a downward trend for the primary transgenic lines, although the decreases did not reach the 0.2 probability level for pUBR2-IIA and pUBR2-IIB. However, diastatic power showed significant increases only for the pUBR2 primary transgenic lines, suggesting that factors other than increased protein were involved in variability for enzymatic activity. A similar trend toward increased activity was seen for
-amylase but was significant only for pUBR2-IIB. β-glucan levels also showed distinct differences between the two groups of primary transgenic lines. All of the pUBR2 lines had significantly lower levels than Conlon, while none of the pUBR1 lines were significantly different from Conlon.
The backcross-derived families showed general improvements for characteristics that were changed in their primary transgenic parents, but there were several important exceptions. All families showed a trend toward greater similarity to Conlon for barley protein, soluble protein, the ratio of soluble to total protein, and FAN, but there were several instances of large and significant differences, primarily in backcross-derived lines derived from the primary transgenic lines, that showed the greatest differences from Conlon (pUBR2-IA, -IAB). Malt extract showed a similar trend of improvement in the backcross-derived lines, again with instances of significant differences from Conlon being associated with primary transgenic lines having the greatest reductions relative to Conlon.
Diastatic power and the levels of
-amylase and β-glucan, in particular, deviated from the trend toward recovery of Conlon levels for the backcross-derived lines. Data from the pUBR1 families, which showed no significant changes for these traits, have little informative value, although it is interesting to note that for diastatic power, one could argue that there is a trend toward higher levels (and thus greater deviations from Conlon) in the backcross-derived lines than in the primary transgenic parents. Of much more significance, however, is the general lack of recovery of Conlon levels of performance for these traits in the families derived from the transformation events pUBR2-I and pUBR2-II, which is evidenced both by a majority of lines showing significant differences from Conlon and the lack of a clear trend toward Conlon performance when compared with the performance of the primary transgenic lines. Consideration of the contrasting family performance (PDR5 vs. TRI101), the general failure to recover recurrent parent phenotype, and the lack of variability between the PDR5-containing families derived from different transgenic events suggests that the expression of PDR5 may influence these traits. Examination of the malting performance of null segregant lines derived from backcrosses to PDR5 lines would be necessary to make a definitive conclusion regarding PDR5-induced variability. However, the differences in phenotypic effects are consistent with the functions of the two genes: PDR5 encodes a transport protein with a wide variety of targets (Golin et al., 2007), while TRI101 encodes an acetyltransferase with just a few known substrates (Kimura et al., 1998).
| CONCLUSIONS |
|---|
|
|
|---|
The process described in this report is relatively simple and rapid and requires minimal resources. Phenotypic selection among large populations in the field was not necessary—selection was based exclusively on the presence of the transgene among very small populations in the greenhouse. When combined with improved in vitro techniques that generate less somaclonal variation and enable transformation of elite cultivars (Bregitzer et al., 2002; Dahleen and Bregitzer, 2002; Manoharan and Dahleen, 2002; Zhang et al., 1999), superior transgenic barley germplasm can be produced. The reduction of somaclonal variation in transgenic lines facilitates accurate determinations of the phenotypic effect(s) and commercial value of transgene expression by enabling more nearly isogenic comparisons with the nontransformed parent. Furthermore, the phenotype conferred by the transgenic trait can be more realistically evaluated if it expressed a wild-type background, free of any pleiotropic interactions with variant alleles that may have arisen by somaclonal variation during the transformation process.
| NOTES |
|---|
|
|
|---|
Received for publication June 29, 2007.
| REFERENCES |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
K. B. R. S. Visarada, K. Meena, C. Aruna, S. Srujana, N. Saikishore, and N. Seetharama Transgenic Breeding: Perspectives and Prospects Crop Sci., August 7, 2009; 49(5): 1555 - 1563. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||