Gibberellins and Heterosis in Crops and Trees: An Integrative Review and Preliminary Study with Brassica

Heterosis, or hybrid vigor, has contributed substantially to genetic improvements in crops and trees and its physiological basis involves multiple processes. Four associations with the phytohormone gibberellin (GA) indicate its involvement in the regulation of heterosis for shoot growth in maize, sorghum, wheat, rice, tomato and poplar. (1) Inbreds somewhat resemble GA-deficient dwarfs and are often highly responsive to exogenous GA3. (2) Levels of endogenous GAs, including the bioeffector GA1, its precursors GA19 and GA20, and/or its metabolite GA8, are higher in some fast-growing hybrids than parental genotypes. (3) Oxidative metabolism of applied [3H]GAs is more rapid in vigorous hybrids than inbreds, and (4) heterotic hybrids have displayed increased expression of GA biosynthetic genes including GA 20-oxidase and GA 3-oxidase. We further investigated Brassica rapa, an oilseed rape, by comparing two inbreds (AO533 and AO539) and their F1 hybrid. Seedling emergence was faster in the hybrid and potence ratios indicated dominance for increased leaf number, area and mass, and stem mass. Overdominance (heterosis) was displayed for root mass, leading to slight heterosis for total plant mass. Stem contents of GA19,20,1 were similar across the Brassica genotypes and increased prior to bolting; elongation was correlated with endogenous GA but heterosis for shoot growth was modest. The collective studies support a physiological role for GAs in the regulation of heterosis for shoot growth in crops and trees, and the Brassica study encourages further investigation of heterosis for root growth.


Gibberellins and Heterosis in Crops and Trees: Four Lines of Evidence
Heterosis, or hybrid vigor, represents hybrid performance that exceeds that of both parents [1]. Heterosis is common in crop plants and in some trees and has been responsible for a substantial proportion of the genetic improvement through plant breeding over the past century [2,3]. However, while heterosis is widely utilized for agriculture, horticulture and silviculture, and despite a century of investigation, its physiological basis is poorly understood [4].
Studies indicate that there are multiple contributing genetic and physiological factors [5][6][7] and phytohormones should be involved since these provide master regulators of plant form and function [8,9]. Of the phytohormones, there is substantial evidence that the promotive regulator gibberellin (GA) is especially involved in heterosis for shoot growth. There have been preliminary reviews of this association [10,11] and we extend the analyses to incorporate recent findings, expand the range of study approaches and increase the number of plant species.

Hybrids Have Higher Levels of Endogenous GAs
The hypothesis that inbreds are GA deficient was subsequently tested with advancing techniques including selective GA bioassays, GA immunoassays, and definitive physicochemical analyses with combined gas chromatography-mass spectrometry (GC-MS). For GC-MS analyses, internal standard [ 2 H]GAs are added to account for variable losses during the extensive purification [15,16]. Studies with different species and in different research labs supported the hypothesis, as heterotic hybrids contain higher levels of endogenous GAs (Tables 1 and 2). This provides the second line of evidence supporting a regulatory role of GAs in heterosis for shoot growth. Table 1. Studies of gibberellins (GAs) and heterosis for growth in maize, or corn, Zea mays L., sequenced chronologically (GA 3 = gibberellic acid).

Findings
Reference Exogenous GA 3

Gene expression Other
Higher sensitivity of inbreds to exogenous GA 3 [12] X Higher levels of GA-like substances in shoot apices from a hybrid than parental inbreds [  Multiple GAs are native in plants and GA 1 is probably the bioeffector GA in most angiosperms for shoot elongation and some other processes (Figure 1a). The biosynthetic pathways have been determined and the "early 13-hydroxylation pathway" is probably the primary pathway in most crops and broad-leaved trees ( Figure 1b). Another biosynthetic pathway leads to GA 4 , a bioactive GA [42] or an alternate GA 1 precursor (Figure 1b) [43]. With 2-hydroxylation, GA 1 is inactivated to form GA 8 and different GAs are often analyzed to provide a record of present (GA 1 ), past (GA 8 ) and future (GA 19 , GA 20 , GA 4 ) GA biosynthesis ( Figure 1b). Researchers have primarily analyzed these endogenous GAs which may display different patterns across the genotypes (Tables 1 and 2). crops and broad-leaved trees (Figure 1b). Another biosynthetic pathway leads to GA4, a bioactive GA [42] or an alternate GA1 precursor (Figure 1b) [43]. With 2-hydroxylation, GA1 is inactivated to form GA8 and different GAs are often analyzed to provide a record of present (GA1), past (GA8) and future (GA19, GA20, GA4) GA biosynthesis (Figure 1b). Researchers have primarily analyzed these endogenous GAs which may display different patterns across the genotypes (Tables 1 and 2). The early 13-hydroxylation GA biosynthetic pathway, including GAs and enzymes (blue; ox = oxidase) that were assessed in this study with Brassica or in the studies listed in Tables 1 and 2. GA19 and GA20 are the primary GA1 precursors and GA4 provides an alternate precursor [16]. Following 2hydroxylation, GA29 and GA8 (red) are biologically inactive.

GA Metabolism is Rapid in Heterotic Hybrids
Analyses of endogenous GAs provide instantaneous measurements, while studies of GA metabolism capture physiological effects over hours or days. The early 13-hydroxylation GA biosynthetic pathway, including GAs and enzymes (blue; ox = oxidase) that were assessed in this study with Brassica or in the studies listed in Tables 1 and 2. GA 19 and GA 20 are the primary GA 1 precursors and GA 4 provides an alternate precursor [16]. Following 2-hydroxylation, GA 29 and GA 8 (red) are biologically inactive.

GA Metabolism is Rapid in Heterotic Hybrids
Analyses of endogenous GAs provide instantaneous measurements, while studies of GA metabolism capture physiological effects over hours or days.

Expression of GA Biosynthetic Genes is Increased in Some Hybrids
After the GA biosynthetic sequence was determined, the responsible genes were identified and cloned [42,44]. The multi-gene GA 20-oxidase family is responsible for sequential conversions from GA 53 through GA 44 and GA 19 to GA 20 ( Figure 1b). GA 19 is often an abundant endogenous GA and its conversion to GA 20 is regarded as rate-limiting and hence, regulatory [44]. This gene has been especially studied and its expression has been associated with heterosis in some plants (Tables 1 and 2). The final conversion from GA 20 to GA 1 is enabled by GA 3-oxidase, another multi-gene family (formerly 3β-hydroxylase), and the expression of this gene has also been found to be correlated with heterosis ( Table 2). This association between the expression of the GA biosynthetic genes and hybrid vigor provides the fourth line of evidence supporting the role of GAs in the regulation of heterosis for shoot growth in crops and trees. With multiple studies over a half century, by different researchers and with different plants, there has been substantial support for the association between GAs and heterosis for shoot growth. However, there has also been opposing evidence. In some studies, there have been no correspondences but there are confounding influences. A challenge is the tissue sampling since GAs are differentially distributed within plants, commonly with higher concentration in the shoot apices and subtending elongation zones [45]. Below those zones, the concentrations are lower and if whole stems are sampled there could be greater dilution by the larger stems of hybrids, thus obscuring and even inverting the endogenous GA pattern. This challenge can be increased for trees, which are much larger, and consequently bark scrapings are often sampled (Table 2). However, the locations relative to the growth zones remain a complexity and the timing of sampling further challenges these studies [36,45].
Another complexity relates to the type of growth. The association between GA and stem height is well established across plants, and there are also correspondences with other elongation measures such as leaf lengths, but there may be weaker associations with some other growth measures. A study by Auger et al. [22] applied an interesting approach with GA dwarfing genes in maize, and heterosis persisted. Auger et al. [22] considered this as opposing a regulatory role for GAs but the dwarfing genes would lead to GA reduction [46] rather than elimination, which would be lethal. Consequently, without analyses of the GA contents, that study was less complete.

Gibberellins and Heterosis in Brassica
To further explore the growth components that contribute to heterosis, and to investigate their association with endogenous GAs, Brassica was used as another study system. Heterosis is observed in some Brassica hybrids [47,48] and genetic diversity is incorporated with the allotetraploid genotypes of some rapeseed and canola cultivars [49]. Brassica is a close crop relative to Arabidopsis, allowing for the application of knowledge from that primary model plant system, which also displays heterosis [40,50,51]. Consequently, this study was undertaken with a triplet family consisting of two inbreds, AO533 and AO539, and their F 1 hybrid, and we expected: (1) heterosis in the hybrid and (2) correspondence between GA content and stem growth within and across the genotypes.

Growth and Dominance Patterns
Initial seedling emergence was more rapid in the Brassica hybrid (Table 3). Subsequently, the parent AO539 and the F 1 displayed similar seedling emergence percentages, while AO533 displayed slower and less complete emergence. This pattern of dominance or slight overdominance for the increased trait was also observed for most of the subsequent shoot characteristics; the F 1 was slightly superior to AO539, while AO533 was consistently less vigorous (Tables 3 and 4). This was the case for leaf number while leaf sizes were similar. As the product of these two traits, total leaf areas and leaf masses were greater in the F 1 and AO539 than AO533 (Tables 3 and 4, Figure 2).  The root rinsing with a water spray appeared complete and we expect similar root recoveries across the harvests and genotypes, providing a valid comparison. The challenge with root sampling should be recognized but the results indicate substantial heterosis for root growth with this triplet of genotypes.

Endogenous Gibberellins
Generally, similar patterns across the genotypes were observed for GA1 and its precursors, GA20 and GA19 (Figure 3). There were often lower GA contents at the early harvests and increases by day 42, when stem bolting was commencing. This was also during the transition to the reproductive phase ( Figure 3) and it is likely that endogenous GA would be involved in that developmental process as well as stem elongation. There were no significant differences across the genotypes for GA content or concentration at day 21 or 35, and at day 42 GA contents per stem were significantly higher in the F1 hybrid than in either parent ( Figure 3; ANOVA with combined GAs: F(2,4) = 11.3, p = 0.023). Concentrations were higher in the F1 than in AO539 but not significantly different from AO533 (3.16, 2.60 and 3.83 ng/g for GA19; 2.50, 1.59 and 2.36 for GA20; and 2.65, 1.90, and 3.39 for GA1 in AO533, AO539 and the F1, respectively; ANOVA: F(2,4) = 15.49, p = 0.013). These results support the involvement of endogenous GAs in the regulation of stem elongation but are uncertain relative to the relationship with heterosis for shoot growth.  The sequential harvests displayed the increasing growth of the Brassica plants, with the F 1 displaying similar or slightly greater growth than AO539, and both showing greater growth than AO533 (Figure 2). There was apparently slight heterosis for stem mass and the largest difference was for root mass ( Figure 2). The F 1 root mass was consistently larger and there was a lag in root growth in AO533. This provided the greatest contribution to the total plant mass, which displayed slight heterosis ( Figure 2). The analyses of covariance included the combined results from harvests on days 21, 35 and 42, with day as the covariate. The primary factor of genotype provided significant effects (p < 0.05) for seven of the eight growth measures (Table 4). Subsequent post-hoc pairings with the marginal means confirmed the pattern displayed in the growth plots, with the F 1 being similar to or slightly larger than AO539, while AO533 was smaller (Table 4, Figure 2).
Potence ratios (PR) quantified the degree of dominance with values substantially greater than 1 indicating overdominance, or heterosis (Table 4). PR averages from the three harvests are provided (Table 4) and were highest for juvenile height, which extended from the early advantage of the hybrid, and root mass, which contributed to the moderate overdominance for overall plant mass. Root sampling is imprecise due to some adhering substrate particles and the loss of some fine roots. The root rinsing with a water spray appeared complete and we expect similar root recoveries across the harvests and genotypes, providing a valid comparison. The challenge with root sampling should be recognized but the results indicate substantial heterosis for root growth with this triplet of genotypes.

Endogenous Gibberellins
Generally, similar patterns across the genotypes were observed for GA 1 and its precursors, GA 20 and GA 19 (Figure 3). There were often lower GA contents at the early harvests and increases by day 42, when stem bolting was commencing. This was also during the transition to the reproductive phase ( Figure 3) and it is likely that endogenous GA would be involved in that developmental process as well as stem elongation. There were no significant differences across the genotypes for GA content or concentration at day 21 or 35, and at day 42 GA contents per stem were significantly higher in the F 1 hybrid than in either parent ( Figure 3; ANOVA with combined GAs: F (2,4) = 11.3, p = 0.023). Concentrations were higher in the F 1 than in AO539 but not significantly different from AO533 (3.16, 2.60 and 3.83 ng/g for GA 19 ; 2.50, 1.59 and 2.36 for GA 20 ; and 2.65, 1.90, and 3.39 for GA 1 in AO533, AO539 and the F 1 , respectively; ANOVA: F (2,4) = 15.49, p = 0.013). These results support the involvement of endogenous GAs in the regulation of stem elongation but are uncertain relative to the relationship with heterosis for shoot growth.

Heterosis and Gibberellins in Brassica
In this study, hybrid vigor, superior performance of the hybrid, was slight. As occurs for other plants [14,53,54], seedling emergence was faster in the hybrid. Subsequently, for the different growth Gibberellin Content (ng/stem) Figure 3. Contents of endogenous gibberellin A 1 (GA 1 ) and its precursors GA 19 and GA 20 in stems, and heights (Means ± SE provided for day 49, n = 10) of two Brassica inbreds and their F 1 hybrid grown in greenhouse conditions. The inset plot displays the developmental stages at day 42, in accordance with Harper and Berkencamp [52].

Heterosis and Gibberellins in Brassica
In this study, hybrid vigor, superior performance of the hybrid, was slight. As occurs for other plants [14,53,54], seedling emergence was faster in the hybrid. Subsequently, for the different growth traits, the hybrid performance matched or slightly exceeded the superior inbred parent. The expression of heterosis through the multiple traits supports a polygenic foundation and the hybrid would benefit from superior performance for many traits, both independent and coordinated [55]. We expected heterosis for shoot growth, but an unexpected finding was that heterosis was greatest for root growth. There have been fewer studies of heterosis and root growth but early research with tomato hybrids recognized heterosis for root growth [56] and this pattern has been subsequently demonstrated for sorghum and maize [53,57]. At the cellular level, heterosis for gene expression in roots has also been demonstrated, but the function of some genes is less clear [58].
Root mass is a product of root numbers and their sizes, including lengths. The observed overdominance for total root mass might reflect the additive benefit from dominance for higher root numbers and lengths, and it would be useful to resolve the root components and even cell sizes and numbers. If GAs were involved, this might relate to root lengths, since GAs contribute strongly to the regulation of elongation growth and GA manipulation through recombinant GA 2-oxidase resulted in denser but shorter root systems in hybrid poplar [59]. Heterosis for root growth might contribute to yield stability and could influence adaptation to drought or salinity [60].
Relative to endogenous GAs, GA 1 and GA 19 contents per stem were apparently higher in the hybrid on day 42, when the bolting stage commenced, and the hybrid displayed slight heterosis for increased stem mass. However, this Brassica study was preliminary and future research should focus on hybrid versus parent combinations that display greater hybrid vigor in the growth patterns [61,62], as has been useful for maize inbreds versus hybrids [15,17,23,63]. A comparison for Brassica could include family combinations with increased hybrid vigor [64] and it would be useful to analyze GA contents and especially GA 1 along with its precursors and metabolite, as well as with the expression of the key GA biosynthesis genes [39,42] (Table 1). A genetic approach could also be undertaken such as investigating the influence on heterosis from dwarfing genes in Brassica that block GA biosynthesis or action [22,46,65]. Due to the genetic, molecular and physiological similarity, directions from studies of heterosis in Arabidopsis should also be productive and these have supported the involvement of phytohormonal regulation [50]. And while GA could provide a focus, there are likely to be coordinated interactions with other phytohormones and especially auxins [58,66], along with other signaling substances such as brassinosteroids [25].

A Final Consideration for GAs and Heterosis: Cause or Correlate
While this study with Brassica provided provisional outcomes regarding heterosis, it is clear that GAs are involved in the regulation of shoot growth in Brassica, as is observed for the other plants studied: maize, sorghum, wheat, rice and poplar (Tables 1 and 2). There have been numerous studies that provide direct or indirect support for the role of GAs in the regulation of heterosis for shoot growth, including stem and leaf growth, in crops and trees. However, while the findings indicate that slow-growing inbreds are limited due to relative GA deficiency, or conversely that hybrids display increased GA biosynthesis and contents, the subsequent question arises, "how are the GA biosynthetic genes regulated?" Thus, GAs may serve as an intermediate between heterotic factors and especially genetic heterozygosity and the phenotypic outcome of hybrid vigor. Also, rather than individual genes being central, as is the case for single gene dwarf mutants of maize, Brassica and other plants, a number of heterozygous genes would probably contribute, providing a polygenic foundation. The consequent up-regulation of the GA biosynthesis genes would provide a mechanism to amplify the outcome, since this would promote a range of morphological and physiological traits that contribute to the phenotypic outcome of hybrid vigor.

Plant Material and Growth Conditions
Seeds of two inbred parental lines of Brassica rapa L. (syn. B. campestris L.), AO533 and AO539, and their F 1 hybrid, AO533 × AO539 (from the Agriculture Canada canola breeding program, Saskatoon, Canada), were sown in pots containing Metro-mix (Ajax, ON, Canada), a Sphagnum peat moss and vermiculite-containing potting medium. Plant pots were randomly arranged in the University of Lethbridge greenhouse (49 • 70 N, 112 • 86 W), with a 16 h photoperiod that was provided by a combination of natural sunlight and light from Na vapour lamps (260 µmol m −2 s −1 ; Reflector PL90M [medium] N400; P.L. Light Systems Canada, Inc., Beamsville, ON, Canada), and with temperatures maintained between 21-24 • C. Pots were watered daily to saturation and fertilized monthly with a water-soluble 20-14-14 (N-P-K) fertilizer with supplemental trace elements (Professional Gardener Co., Ltd., Calgary, AB, Canada).

Growth Measurements
Seedling emergence was noted and height and developmental stages [52] were recorded biweekly. Ten individual plants of each genotype were harvested at each of five harvest dates, at 14, 21, 35, 42 and 70 days after planting and for each plant, leaf number, total leaf area using a portable LI-COR photometric area meter (Lincoln, NB, USA), and organ dry weights (leaf, stem and root) were measured. Prior to weighing, leaf and root tissues were oven dried, while stem tissues were flash frozen in liquid nitrogen and then freeze dried for dry weight determinations and to permit analysis of endogenous GAs.

Potence Ratios
Inheritance patterns were quantified by comparing the performance of the hybrids with that of the two parents [17,67]. The additive component of variation (H) was calculated as the absolute value of either parent (P 2 represents the superior parent for that trait) minus the mean of the two parents, the midparent value (mp). The dominance component of variation (D) was calculated as the hybrid value minus mp and the potence ratios represent the overall degree of dominance as:

Analyses of Endogenous Gibberellins
Gibberellins (GAs) were quantified in one of the two replicates, with each consisting of 5-6 stems from a genotype at a harvest. Analyses were undertaken as described by Zanewich and Rood [68] and others [16,69]. Briefly, tissue was extracted in aqueous methanol with internal standards of [ 2 H 2 ]-GA 1 , -GA 19 and -GA 20 (from L. Mander, Australian National University), filtered, and buffered (pH 9). Following in vacuo reduction, the aqueous phase was partitioned twice with water-saturated ether. The pH was then adjusted to 3 and the sample was further partitioned 3 times with ethyl acetate. Extracts were frozen, filtered to remove ice, and dried in vacuo. Residues were resuspended and transferred to glass microfiber discs and trace quantities of [ 3 H]GA 1 and [ 3 H]GA 4 were added for tracking. Samples were further purified using silica columns [68] and respective GA-containing fractions were separated using reversed-phase C 18 high performance liquid chromatography. Grouped fractions containing the three GAs of interest were methylated and trimethylsilylated, and GAs were detected using gas chromatography-mass spectrometry (GC-MS) with selected ion monitoring. Quantities were calculated based in the paired ion abundances from the internal standard [ 2 H 2 ]GAs and endogenous forms [16,43].

Statistical Analyses
Analyses of variance (ANOVA) with SPSS v.19 (IBM, Armonk, NY, USA) provided the primary assessment for each growth measure with genotype and harvest day as fixed factors. These analyses used the data collected from harvest days 21, 35 and 42, which consistently displayed strong linear or log-linear patterns. Gibberellin content and concentrations were also analyzed by ANOVA, for all harvest days (three-factor: GA, genotype and day) or individual days (two-factor: GA and genotype).
Author Contributions: Literature review, S.B.R. and K.P.Z.; Brassica study, conceptualization K.P.Z. and S.B.R.; plant cultivation and harvesting, gibberellin sample preparation and quantification, K.P.Z.; data analyses, K.P.Z. and S.B.R.; manuscript writing and editing, S.B.R. and K.P.Z.; funding acquisition, S.B.R. All authors have read and agreed to the published version of the manuscript.
Funding: This study was supported by funding from Alberta Innovates and the Natural Sciences and Engineering Research Council of Canada.