The Association of Grain Yield and Agronomical Traits with Genes of Plant Height, Photoperiod Sensitivity and Plastid Glutamine Synthetase in Winter Bread Wheat (Triticum aestivum L.) Collection

The reduction in plant height caused by mutations in Rht-B1 or Rht-D1 (Reduced height-1) genes in combination with day-length-independent early flowering associated with the Ppd-D1 (Photoperiod-D1) gene were the main factors of the drastic yield increase in bread wheat in the 1960s. Increasing nitrogen use efficiency as well as maintaining high yields under conditions of global climate change are the modern goals of wheat breeding. The glutamine synthetase (GS) enzyme plays a key role in ammonium assimilation in plants. In previous studies, the TaGS2-A1 gene, coding the plastid isoform of GS, was shown to be connected with nitrogen use efficiency in wheat. Using the polymerase chain reaction (PCR) markers, the association of yield and agronomical traits with haplotypes of Rht-B1, Rht-D1, Ppd-D1 and TaGS2-A1 genes was studied in a diverse collection of winter bread wheat cultivars grown in Krasnodar (Russia). In the three-year experiment, semidwarfism and photoperiod insensitivity were confirmed to be highly favorable for the grain yield. The TaGS2-A1b haplotype had a tendency for increased grain yield and lodging resistance, but mainly in plants not possessing the ‘green revolution’ alleles. Thus, TaGS2-A1b may have potential in breeding wheat cultivars with alternative dwarfing genes or tall cultivars, which may be optimal for growing under certain environments.


Introduction
Wheat is the staple crop for an estimated 35% of the world's population [1]. During the "green revolution" of the 1960s, wheat grain yields were nearly doubled due to the introduction of the new short-straw varieties, which better supplied assimilates for the developing spike at the expense of stem biomass and allowed for the application of higher doses of nitrogen fertilizers and irrigation without causing lodging [2,3]. The new varieties were shorter due to limited responsiveness to the growth-promoting gibberellin hormones, which was conferred by mutant alleles at one of the two Reduced height-1 loci (Rht-B1b or Rht-D1b) [4]. These genes, located on chromosomes 4B and 4D, correspondingly, encode DELLA proteins, which act in the cell nucleus as regulators of gene transcription. The DELLA proteins consist of a conservative C-terminal domain, responsible for growth repression; and a N-terminal DELLA domain, which is involved in GA perception and targeted plant survival under photorespiration conditions. During photorespiration, the ammonium is released, while GS2 traps it and incorporates in glutamine, preventing its toxicity and depletion of organic nitrogen in the plant organism [31]. Increased plastid GS expression was shown to be favorable for plant growth and yield as it improves, among other traits, the nitrogen use efficiency. The lines overexpressing GS2 were shown to be more tolerant to abiotic stresses such as drought or soil salinity, as higher GS activity favors the synthesis of osmolytes such as proline or polyamines [31]. Moreover, plants overexpressing GS2 showed increased tolerance to high-intensity light [32].
In bread wheat, the plastid glutamine synthetase (GS2) genes located on chromosomes 2A, 2B and 2D were isolated and sequenced [28]. Two haplotypes were distinguished for TaGS2-A1 (a and b), along with six haplotypes for TaGS2-B1 (a-f ) and two haplotypes for TaGS2-D1 (a and b). TaGS2-A1b was considered as a favorable one, along with TaGS2-B1a, B1b and D1a, by their association with nitrogen use and agronomic traits in a mini core collection of Chinese bread wheat varieties. In that collection, 65% of accessions possessed TaGS2-A1b, and this haplotype was more frequent in modern varieties than in landraces [28]. In another study of durum wheat collection, the variation at TaGS2-A1 was significantly associated with grain protein content [33]. Expression of the wheat GS2 in Escherichia coli showed that an enzyme encoded by the TaGS2-A1 gene is the most active isoform among others [34].
The transgenic experiment showed that expression of TaGS2-A1b under its own promoter in winter bread wheat increases grain yield and its components both under low and high nitrogen supply conditions. In addition, nitrogen acquisition by roots and remobilization was improved in transgenic plants [34]. Thus, TaGS2-A1b is valuable in wheat breeding for improved nitrogen use efficiency and grain yield.
In this study, we try to estimate the benefits of the TaGS2-A1b haplotype in a minicollection of winter bread wheat varieties grown in southern Russia and evaluate the prospects of its combined use with well-studied 'green-revolution' alleles of Rht-B1 and Ppd-D1 genes. We hypothesize that TaGS2-A1b can improve nitrogen-use efficiency and further increase grain yields of semidwarf and photoperiod-insensitive wheat plants.

Sequence Comparison and PCR Marker Development
To investigate the diversity of the TaGS2-A1 gene using modern genomic information, we extracted the sequences of this gene from 18 wheat accessions having sequenced and assembled genomes, listed in our previous publication [35], most of which are the part of the 10+ Wheat Genomes Project [36]. In the region of the miniature inverted transposable element (MITE) insertion in the second intron, which was previously used as diagnostic for the TaGS2-A1b haplotype, some of the genomic sequences were lacking information, and were represented by indefinite nucleotide bases (e.g., 'NNN'). For further conclusions, we take these regions to be identical to the TaGS2-A1b haplotype. Compared to the other sequences, the sequence of the TaGS2-A1b haplotype of Xiaoyan 54 (GenBank accession GQ169685.1) has a rare single-nucleotide variant (SNV) within the MITE insertion that differs it from all other genomic sequences having the MITE insertion. If we consider this SNV to be a sequencing error, then only two haplotypes will be found in bread and cultivated durum wheat, which were previously reported as TaGS2-A1a (found in Chinese Spring and Lancer bread wheat and in spelt wheat PI 190962) and TaGS2-A1b (found in ArinaLrFor, Cadenza, CDC Landmark, CDC Stanley, Claire, Jagger, Julius, Mace, Norin 61, Paragon, Robigus and SY Mattis bread wheat accessions and in durum wheat Kronos). The haplotype TaGS2-A1b was obviously prevailing among these accessions of cultivated wheat. In wild emmer wheat Zavitan (T. turgidum ssp. dicoccoides), two unique SNPs-one in the promotor and another in the sixth intron-were found. The haplotype of Zavitan did not contain the MITE insertion in the second intron. The haplotype of the wild einkorn wheat Triticum urartu G1812 (PI428198) contained more unique SNPs and insertions in the promoter and introns, but also did not contain the MITE insertion characteristic for TaGS2-A1b (see the supplementary FASTA-format file).
TaGS2-A1b differs from TaGS2-A1a not only by the 239 bp MITE insertion in the second intron, but also by a 10 bp insertion in the 5 flanking sequence, which obviously serves as a gene promotor. For this 10 bp insertion, we developed the subgenome-specific primers and made a codominant PCR marker suitable for agarose gel electrophoresis. The 172 bp product is diagnostic for haplotype TaGS2-A1a, while 182 bp is diagnostic for haplotype TaGS2-A1b (Figure 1).
TaGS2-A1b (found in ArinaLrFor, Cadenza, CDC Landmark, CDC Stanley, Claire, Jagger, Julius, Mace, Norin 61, Paragon, Robigus and SY Mattis bread wheat accessions and in durum wheat Kronos). The haplotype TaGS2-A1b was obviously prevailing among these accessions of cultivated wheat. In wild emmer wheat Zavitan (T. turgidum ssp. dicoccoides), two unique SNPs-one in the promotor and another in the sixth intron-were found. The haplotype of Zavitan did not contain the MITE insertion in the second intron. The haplotype of the wild einkorn wheat Triticum urartu G1812 (PI428198) contained more unique SNPs and insertions in the promoter and introns, but also did not contain the MITE insertion characteristic for TaGS2-A1b (see the supplementary FASTA-format file).
TaGS2-A1b differs from TaGS2-A1a not only by the 239 bp MITE insertion in the second intron, but also by a 10 bp insertion in the 5′ flanking sequence, which obviously serves as a gene promotor. For this 10 bp insertion, we developed the subgenome-specific primers and made a codominant PCR marker suitable for agarose gel electrophoresis. The 172 bp product is diagnostic for haplotype TaGS2-A1a, while 182 bp is diagnostic for haplotype TaGS2-A1b (Figure 1).

One-Way Analysis
Using this and other molecular markers described in the 'Materials and Methods' section, we genotyped the collection of 195 winter bread wheat accessions for Rht-B1, Rht-D1, Ppd-D1, and TaGS2-A1 genes (Table S13). This collection was grown and evaluated for agronomic traits at the National Center of Grain in Krasnodar (Russia) for three years (2018-2020). Below, we describe the results of genotyping and the associations of the genotypes with the best linear unbiased estimates (BLUEs) of grain yield and agronomic traits for the groups of genes separately and in the triple interaction ((Rht-B1 + Rht-D1) × Ppd-D1 × TaGS2-A1).

Rht-1 Effects
In 71% of the bread wheat accessions, one of the three gibberellin-insensitive reduced-height alleles was found: Rht-B1b was found in 47%, Rht-D1b was found in 14%, and Rht-B1e was found in 9% of accessions. In 29% of accessions, none of these reducedheight alleles were detected (Table S13). In addition, there were no double dwarfs combining Rht-D1b with Rht-B1b or Rht-B1e in the collection studied. The effects of these reduced-height alleles on plant height were slightly different: The accessions carrying Rht-D1b or Rht-B1e had a lower average plant height (91 ± 3 and 90 ± 5 cm, respectively) than those carrying Rht-B1b (98 ± 1 cm; here and further the means of BLUEs with the 95% confidence intervals are given). The average height of accessions without gibberellin-insensitive reduced-height alleles was 112 ± 3 cm (Figure 2a). The effects of the three Rht alleles were also a bit different for the other agronomical traits. Accessions carrying Rht-D1b had a reduced 1000-kernel weight, but at the same time they were the most resistant to lodging, substantially differing in this feature from all other genotypes. The grain yield per hectare of the three gibberellin-insensitive reduced-height genotypes did not differ

One-Way Analysis
Using this and other molecular markers described in the 'Materials and Methods' section, we genotyped the collection of 195 winter bread wheat accessions for Rht-B1, Rht-D1, Ppd-D1, and TaGS2-A1 genes (Table S13). This collection was grown and evaluated for agronomic traits at the National Center of Grain in Krasnodar (Russia) for three years (2018-2020). Below, we describe the results of genotyping and the associations of the genotypes with the best linear unbiased estimates (BLUEs) of grain yield and agronomic traits for the groups of genes separately and in the triple interaction ((Rht-B1 + Rht-D1) × Ppd-D1 × TaGS2-A1).

Rht-1 Effects
In 71% of the bread wheat accessions, one of the three gibberellin-insensitive reducedheight alleles was found: Rht-B1b was found in 47%, Rht-D1b was found in 14%, and Rht-B1e was found in 9% of accessions. In 29% of accessions, none of these reduced-height alleles were detected (Table S13). In addition, there were no double dwarfs combining Rht-D1b with Rht-B1b or Rht-B1e in the collection studied. The effects of these reducedheight alleles on plant height were slightly different: The accessions carrying Rht-D1b or Rht-B1e had a lower average plant height (91 ± 3 and 90 ± 5 cm, respectively) than those carrying Rht-B1b (98 ± 1 cm; here and further the means of BLUEs with the 95% confidence intervals are given). The average height of accessions without gibberellin-insensitive reduced-height alleles was 112 ± 3 cm (Figure 2a). The effects of the three Rht alleles were also a bit different for the other agronomical traits. Accessions carrying Rht-D1b had a reduced 1000-kernel weight, but at the same time they were the most resistant to lodging, substantially differing in this feature from all other genotypes. The grain yield per hectare of the three gibberellin-insensitive reduced-height genotypes did not differ significantly from each other (Figure 2b), but their average yield (9.1 ± 0.1 t/ha) was significantly higher than that of 'tall' accessions (7.9 ± 0.3 t/ha). The grain protein content was decreased by semidwarfism; however, the grain protein yield per hectare was increased. Unexpectedly, the lodging tolerance of accessions carrying the gibberellin insensitivity alleles was not substantially higher (Table S1). significantly from each other (Figure 2b), but their average yield (9.1 ± 0.1 t/ha) was significantly higher than that of 'tall' accessions (7.9 ± 0.3 t/ha). The grain protein content was decreased by semidwarfism; however, the grain protein yield per hectare was increased. Unexpectedly, the lodging tolerance of accessions carrying the gibberellin insensitivity alleles was not substantially higher (Table S1).

Ppd-D1 Effects
The allele Ppd-D1a, conferring photoperiod insensitivity, was observed in 76% of wheat accessions; 19% of accessions had the Ppd-D1b; the remaining 5% of accessions were heterogenous (Table S13). Only the homogenous accessions were used for calculating any effects on agronomical traits. On average, accessions with Ppd-D1a proceeded to heading 7 days earlier than accessions with Ppd-D1b (Table S2). In addition, Ppd-D1a increased 1000-kernel weight by 3.8 g. The grain yield in photoperiod-insensitive accessions was higher by 0.7 t/ha. Ppd-D1a decreased the grain protein content by 0.7%; however, the grain protein yield slightly increased by 0.5 t/ha. In photoperiod-insensitive plants carrying Ppd-D1a, the leaf area damaged by brown rust (Puccinia triticina f. sp. tritici) was around twice lower. Despite a little reduction of the plant height, the Ppd-D1a allele noticeably decreased the lodging tolerance of the wheat plants.

GS2-A1 Effects
Using the developed marker, the TaGS2-A1b haplotype, previously considered as more favorable for nitrogen assimilation, was detected in 43% of accessions, while TaGS2-A1a was observed in 53% of them. A total of 4% of accessions were heterogenous for TaGS2-A1. By the results of one-way ANOVA, TaGS2-A1 did not influence any of the traits significantly. However, there was a tendency for TaGS2-A1b to increase lodging resistance and to delay heading.

Ppd-D1 Effects
The allele Ppd-D1a, conferring photoperiod insensitivity, was observed in 76% of wheat accessions; 19% of accessions had the Ppd-D1b; the remaining 5% of accessions were heterogenous (Table S13). Only the homogenous accessions were used for calculating any effects on agronomical traits. On average, accessions with Ppd-D1a proceeded to heading 7 days earlier than accessions with Ppd-D1b (Table S2). In addition, Ppd-D1a increased 1000-kernel weight by 3.8 g. The grain yield in photoperiod-insensitive accessions was higher by 0.7 t/ha. Ppd-D1a decreased the grain protein content by 0.7%; however, the grain protein yield slightly increased by 0.5 t/ha. In photoperiod-insensitive plants carrying Ppd-D1a, the leaf area damaged by brown rust (Puccinia triticina f. sp. tritici) was around twice lower. Despite a little reduction of the plant height, the Ppd-D1a allele noticeably decreased the lodging tolerance of the wheat plants.

GS2-A1 Effects
Using the developed marker, the TaGS2-A1b haplotype, previously considered as more favorable for nitrogen assimilation, was detected in 43% of accessions, while TaGS2-A1a was observed in 53% of them. A total of 4% of accessions were heterogenous for TaGS2-A1. By the results of one-way ANOVA, TaGS2-A1 did not influence any of the traits significantly. However, there was a tendency for TaGS2-A1b to increase lodging resistance and to delay heading.

Multifactor Analysis
In the studied collection, the wheat accessions that combine any of the reduced-height Rht-1 alleles with the allele of insensitivity to the photoperiod Ppd-D1a are prevailing. Since both the reduced-height and photoperiod insensitivity genes have a strong effect on many agronomical traits, and the combinations of alleles of these genes in the collection are not strictly random (slightly more accessions combining insensitivity to photoperiod with gibberellin-insensitive dwarfism, p = 0.02 for χ 2 ), it is better to view the effects of these genes in interaction, based on the results of multivariate analysis. Since, in general, the effects of various gibberellin-insensitive Rht alleles have the same direction, to simplify the interpretation of the results we decided to unite the data of three gibberellin-insensitive alleles markers to divide the collection in two groups: Those with such alleles (Rht-B1b, Rht-B1e, Rht-D1b) will be called 'dwarf', and those without them will be called 'tall'. Further, the results of the three-factor analyses ((Rht-B1 + Rht-D1) × Ppd-D1 × TaGS2-A1) accounting for the gene interactions are presented.

Heading Date
As expected, Ppd-D1 explained most of the heading date genetic variance in the winter bread wheat collection (Table S4). Accounting for the Bonferroni correction, other factors did not significantly influence the heading date. The accessions carrying Ppd-D1a, the allele conferring photoperiod insensitivity, headed 6 to 7 days earlier than accessions with Ppd-D1b (Table S12, Figure 3).

Multifactor Analysis
In the studied collection, the wheat accessions that combine any of the reducedheight Rht-1 alleles with the allele of insensitivity to the photoperiod Ppd-D1a are prevailing. Since both the reduced-height and photoperiod insensitivity genes have a strong effect on many agronomical traits, and the combinations of alleles of these genes in the collection are not strictly random (slightly more accessions combining insensitivity to photoperiod with gibberellin-insensitive dwarfism, p = 0.02 for χ 2 ), it is better to view the effects of these genes in interaction, based on the results of multivariate analysis. Since, in general, the effects of various gibberellin-insensitive Rht alleles have the same direction, to simplify the interpretation of the results we decided to unite the data of three gibberellininsensitive alleles markers to divide the collection in two groups: Those with such alleles (Rht-B1b, Rht-B1e, Rht-D1b) will be called 'dwarf', and those without them will be called 'tall'. Further, the results of the three-factor analyses ((Rht-B1 + Rht-D1) × Ppd-D1 × TaGS2-A1) accounting for the gene interactions are presented.

Heading Date
As expected, Ppd-D1 explained most of the heading date genetic variance in the winter bread wheat collection (Table S4). Accounting for the Bonferroni correction, other factors did not significantly influence the heading date. The accessions carrying Ppd-D1a, the allele conferring photoperiod insensitivity, headed 6 to 7 days earlier than accessions with Ppd-D1b (Table S12, Figure 3).

Plant Height
Most part of the plant height variance was explained by gibberellin-insensitive dwarfing genes Rht-B1 and Rht-D1, and their interaction with Ppd-D1 (Table S5). As expected, the height of the wheat plants carrying one of the gibberellin-insensitivity alleles was from 12 to 30 cm lower depending on the genetic background. The photoperiod-insensitive allele Ppd-D1a also reduced the height of plants by [8][9][10][11][12][13][14][15] cm, but only in accessions not carrying the Rht-B1 or Rht-D1 dwarfing alleles ( Figure 4, Table S12). Variation at the TaGS2-A1 locus did not have a significant impact on plant height, neither by itself nor in interaction with Rht-1 or Ppd-D1.

Plant Height
Most part of the plant height variance was explained by gibberellin-insensitive dwarfing genes Rht-B1 and Rht-D1, and their interaction with Ppd-D1 (Table S5). As expected, the height of the wheat plants carrying one of the gibberellin-insensitivity alleles was from 12 to 30 cm lower depending on the genetic background. The photoperiod-insensitive allele Ppd-D1a also reduced the height of plants by [8][9][10][11][12][13][14][15] cm, but only in accessions not carrying the Rht-B1 or Rht-D1 dwarfing alleles ( Figure 4, Table S12). Variation at the TaGS2-A1 locus did not have a significant impact on plant height, neither by itself nor in interaction with Rht-1 or Ppd-D1.

Grain Yield
The yield of grain (t/ha) was significantly affected by the Rht-1 and Ppd-D1 genes, while TaGS2-A1 had a marginally significant effect if the interaction of the genes had been accounted (Table S6). The semidwarf, insensitive to photoperiod accessions, as well as those carrying TaGS2-A1b, showed higher three-year BLUEs of grain yield than those carrying wild-type alleles. On average, semidwarfism increased the grain yield by 1.4 t/ha, insensitivity to photoperiod increased the grain yield by 0.8 t/ha and TaGS2-A1b increased the grain yield by 0.6 t/ha. Tall, long-day accessions carrying TaGS2-A1a had the lowest grain yield (6.4 ± 0.8 t/ha), significantly differing from other genotypes. The presence of at least one of the favorable alleles of Rht-1, Ppd-D1 or TaGS2-A1 increased the yield substantially. Further yield increase due to the combination of favorable alleles of the two or three genes in one genotype was not so strong, but statistically significant. The highest grain yield, 9.3 ± 0.3 t/ha, was observed in accessions combining "positive" alleles of all three genes (Rht-B1b or Rht-B1e or Rht-D1b together with Ppd-D1a and TaGS2-A1b) ( Figure 5, Table S12).

Grain Yield
The yield of grain (t/ha) was significantly affected by the Rht-1 and Ppd-D1 genes, while TaGS2-A1 had a marginally significant effect if the interaction of the genes had been accounted (Table S6). The semidwarf, insensitive to photoperiod accessions, as well as those carrying TaGS2-A1b, showed higher three-year BLUEs of grain yield than those carrying wild-type alleles. On average, semidwarfism increased the grain yield by 1.4 t/ha, insensitivity to photoperiod increased the grain yield by 0.8 t/ha and TaGS2-A1b increased the grain yield by 0.6 t/ha. Tall, long-day accessions carrying TaGS2-A1a had the lowest grain yield (6.4 ± 0.8 t/ha), significantly differing from other genotypes. The presence of at least one of the favorable alleles of Rht-1, Ppd-D1 or TaGS2-A1 increased the yield substantially. Further yield increase due to the combination of favorable alleles of the two or three genes in one genotype was not so strong, but statistically significant. The highest grain yield, 9.3 ± 0.3 t/ha, was observed in accessions combining "positive" alleles of all three genes (Rht-B1b or Rht-B1e or Rht-D1b together with Ppd-D1a and TaGS2-A1b) ( Figure 5, Table S12).

Grain Yield
The yield of grain (t/ha) was significantly affected by the Rht-1 and Ppd-D1 genes, while TaGS2-A1 had a marginally significant effect if the interaction of the genes had been accounted (Table S6). The semidwarf, insensitive to photoperiod accessions, as well as those carrying TaGS2-A1b, showed higher three-year BLUEs of grain yield than those carrying wild-type alleles. On average, semidwarfism increased the grain yield by 1.4 t/ha, insensitivity to photoperiod increased the grain yield by 0.8 t/ha and TaGS2-A1b increased the grain yield by 0.6 t/ha. Tall, long-day accessions carrying TaGS2-A1a had the lowest grain yield (6.4 ± 0.8 t/ha), significantly differing from other genotypes. The presence of at least one of the favorable alleles of Rht-1, Ppd-D1 or TaGS2-A1 increased the yield substantially. Further yield increase due to the combination of favorable alleles of the two or three genes in one genotype was not so strong, but statistically significant. The highest grain yield, 9.3 ± 0.3 t/ha, was observed in accessions combining "positive" alleles of all three genes (Rht-B1b or Rht-B1e or Rht-D1b together with Ppd-D1a and TaGS2-A1b) ( Figure 5, Table S12).

Leaf Rust
The genotype of the Ppd-D1 gene showed near-significant association with the percentage of leaf area damaged by brown rust (Table S7). Wheat accessions having the allele of photoperiod neutrality, Ppd-D1a, were less affected by leaf rust compared to long-day plants. On average, the percentage of leaf area affected by leaf rust was 5% in day-length-neutral accessions and 12% in day-length-sensitive accessions. The genes TaGS2-A1 and Rht-1 and genetic factor interactions showed no significant effects on this trait in the three-factor analysis. For the factor analysis, the near-absolutely resistant accessions were excluded for data to fit the normal distribution. These near-absolutely resistant accessions were more frequent among those that carried the TaGS2-A1b allele, as we can see from Figure 6, where all the data were used.

Leaf Rust
The genotype of the Ppd-D1 gene showed near-significant association with the percentage of leaf area damaged by brown rust (Table S7). Wheat accessions having the allele of photoperiod neutrality, Ppd-D1a, were less affected by leaf rust compared to long-day plants. On average, the percentage of leaf area affected by leaf rust was 5% in day-lengthneutral accessions and 12% in day-length-sensitive accessions. The genes TaGS2-A1 and Rht-1 and genetic factor interactions showed no significant effects on this trait in the threefactor analysis. For the factor analysis, the near-absolutely resistant accessions were excluded for data to fit the normal distribution. These near-absolutely resistant accessions were more frequent among those that carried the TaGS2-A1b allele, as we can see from Figure 6, where all the data were used.

The 1000-Kernel Weight
The 1000-kernel weight was significantly affected only by the Ppd-D1 gene (Table S8). In photoperiod-insensitive accessions, the 1000-kernel weight averaged 39.2 ± 0.5 g, while in long-day accessions it was 35.4 ± 1.0 g. The Rht-1 and TaGS2-A1 genes did not influence 1000-kernel weight significantly (Figure 7).

The 1000-Kernel Weight
The 1000-kernel weight was significantly affected only by the Ppd-D1 gene (Table S8). In photoperiod-insensitive accessions, the 1000-kernel weight averaged 39.2 ± 0.5 g, while in long-day accessions it was 35.4 ± 1.0 g. The Rht-1 and TaGS2-A1 genes did not influence 1000-kernel weight significantly (Figure 7).

Leaf Rust
The genotype of the Ppd-D1 gene showed near-significant association with the percentage of leaf area damaged by brown rust (Table S7). Wheat accessions having the allele of photoperiod neutrality, Ppd-D1a, were less affected by leaf rust compared to long-day plants. On average, the percentage of leaf area affected by leaf rust was 5% in day-lengthneutral accessions and 12% in day-length-sensitive accessions. The genes TaGS2-A1 and Rht-1 and genetic factor interactions showed no significant effects on this trait in the threefactor analysis. For the factor analysis, the near-absolutely resistant accessions were excluded for data to fit the normal distribution. These near-absolutely resistant accessions were more frequent among those that carried the TaGS2-A1b allele, as we can see from Figure 6, where all the data were used.

The 1000-Kernel Weight
The 1000-kernel weight was significantly affected only by the Ppd-D1 gene (Table S8). In photoperiod-insensitive accessions, the 1000-kernel weight averaged 39.2 ± 0.5 g, while in long-day accessions it was 35.4 ± 1.0 g. The Rht-1 and TaGS2-A1 genes did not influence 1000-kernel weight significantly (Figure 7).

Grain Protein Content
Both Rht-1 and Ppd-D1 genes showed a significant effect on the grain protein content (Table S9). In both semidwarf and photoperiod-neutral forms, the protein content was significantly lower (Figure 8). The gibberellin-insensitive semidwarf alleles reduced the grain protein content by about 1% (from 15.4 to 14.4%), while day-length insensitivity reduced it by about 0.6% (from 15.2 to 14.6%). The TaGS2-A1 gene, however, did not significantly affect the protein content. In addition, there was no statistically significant interaction of the three genetic factors, which indicates their additive effect on the trait.

Grain Protein Content
Both Rht-1 and Ppd-D1 genes showed a significant effect on the grain protein content (Table S9). In both semidwarf and photoperiod-neutral forms, the protein content was significantly lower (Figure 8). The gibberellin-insensitive semidwarf alleles reduced the grain protein content by about 1% (from 15.4 to 14.4%), while day-length insensitivity reduced it by about 0.6% (from 15.2 to 14.6%). The TaGS2-A1 gene, however, did not significantly affect the protein content. In addition, there was no statistically significant interaction of the three genetic factors, which indicates their additive effect on the trait.

Grain Protein Yield
The grain protein yield per hectare was significantly influenced by Rht genes, and was marginally significant by the TaGS2-A1 gene (Table S10). Despite the decrease in protein content, the grain protein yield per hectare was significantly higher in semidwarf accessions compared to tall ones. The reason for that was an overcompensating increase in grain yield. The presence of one of the reduced-height alleles (Rht-B1b, RhtB1e or Rht-D1b) gave an increase in protein yield by 0.13 t/ha (from 1.16 to 1.29 t/ha), and the presence of the TaGS2-A1b allele increased the protein yield by 0.08 t/ha (from 1.18 to 1.26 t/ha). The lowest protein yield, significantly different from other genotypes, was observed in 'tall' accessions carrying Ppd-D1b and TaGS2-A1a alleles ( Figure 9, Table S12).

Grain Protein Yield
The grain protein yield per hectare was significantly influenced by Rht genes, and was marginally significant by the TaGS2-A1 gene (Table S10). Despite the decrease in protein content, the grain protein yield per hectare was significantly higher in semidwarf accessions compared to tall ones. The reason for that was an overcompensating increase in grain yield. The presence of one of the reduced-height alleles (Rht-B1b, RhtB1e or Rht-D1b) gave an increase in protein yield by 0.13 t/ha (from 1.16 to 1.29 t/ha), and the presence of the TaGS2-A1b allele increased the protein yield by 0.08 t/ha (from 1.18 to 1.26 t/ha). The lowest protein yield, significantly different from other genotypes, was observed in 'tall' accessions carrying Ppd-D1b and TaGS2-A1a alleles ( Figure 9, Table S12).

Grain Protein Content
Both Rht-1 and Ppd-D1 genes showed a significant effect on the grain protein content (Table S9). In both semidwarf and photoperiod-neutral forms, the protein content was significantly lower (Figure 8). The gibberellin-insensitive semidwarf alleles reduced the grain protein content by about 1% (from 15.4 to 14.4%), while day-length insensitivity reduced it by about 0.6% (from 15.2 to 14.6%). The TaGS2-A1 gene, however, did not significantly affect the protein content. In addition, there was no statistically significant interaction of the three genetic factors, which indicates their additive effect on the trait.

Grain Protein Yield
The grain protein yield per hectare was significantly influenced by Rht genes, and was marginally significant by the TaGS2-A1 gene (Table S10). Despite the decrease in protein content, the grain protein yield per hectare was significantly higher in semidwarf accessions compared to tall ones. The reason for that was an overcompensating increase in grain yield. The presence of one of the reduced-height alleles (Rht-B1b, RhtB1e or Rht-D1b) gave an increase in protein yield by 0.13 t/ha (from 1.16 to 1.29 t/ha), and the presence of the TaGS2-A1b allele increased the protein yield by 0.08 t/ha (from 1.18 to 1.26 t/ha). The lowest protein yield, significantly different from other genotypes, was observed in 'tall' accessions carrying Ppd-D1b and TaGS2-A1a alleles ( Figure 9, Table S12).

Lodging Resistance
As expected, semidwarfism caused by the Rht-B1 and Rht-D1 dwarfing alleles was connected with higher resistance to lodging (higher values of the lodging score) (Table S11). Insensitivity to photoperiod caused by the Ppd-D1a allele, despite decreased plant height, was accompanied by lower values of lodging resistance. The TaGS2-A1b allele had a tendency to improve lodging resistance ( Figure 10, Table S12). The most lodging-resistant accessions were semidwarf, photoperiod-sensitive and those carrying the TaGS2-A1b allele.

Lodging Resistance
As expected, semidwarfism caused by the Rht-B1 and Rht-D1 dwarfing alleles was connected with higher resistance to lodging (higher values of the lodging score) (Table  S11). Insensitivity to photoperiod caused by the Ppd-D1a allele, despite decreased plant height, was accompanied by lower values of lodging resistance. The TaGS2-A1b allele had a tendency to improve lodging resistance ( Figure 10, Table S12). The most lodging-resistant accessions were semidwarf, photoperiod-sensitive and those carrying the TaGS2-A1b allele.

Discussion
The studied collection of winter bread wheat accessions tested in Krasnodar consists of old and new cultivars and advanced breeding lines. In our study, about 70% of the collection possessed one of the reduced-height gibberellin-insensitive alleles of either Rht-B1 or Rht-D1 genes, and the most frequent of them was Rht-B1b. These alleles provide optimal semidwarf plant height and about 15% higher grain yield. This shows that Rht-B1b, Rht-B1e and Rht-D1b dwarfing alleles are still current for winter wheat breeding in Russia, despite known poor adaptability of gibberellin-insensitive dwarf cultivars to dry environments. The climate of Krasnodar is characterized by prolonged, dry and hot summers. However, the winter wheat growing season spans autumn; winter, which in Krasnodar is relatively short and mild, without constant snow cover; and spring, which is usually favorable for plant growth. Thus, winter crops can escape the adverse conditions of heat and drought. Nevertheless, more efforts are being conducted by breeders recently to shorten the vegetation period of winter wheat. On the other hand, our results show that rather high yields could be achieved without gibberellin-insensitive dwarfism.
We also noted that reduced-height alleles, obviously having the same mechanism of action, cause various degrees of agronomical trait changes. Rht-D1b or Rht-B1e cause stronger height reduction than Rht-B1b, although the grain yield is not much different between groups of accessions having one or another allele of these three. This is consistent with previous findings [3,8].
Generally, the presence of gibberellin-insensitive reduced-height alleles in our study was associated with reduced grain protein content. That is consistent with other studies, and could be explained by the 'dilution' of grain protein by a larger amount of stored

Discussion
The studied collection of winter bread wheat accessions tested in Krasnodar consists of old and new cultivars and advanced breeding lines. In our study, about 70% of the collection possessed one of the reduced-height gibberellin-insensitive alleles of either Rht-B1 or Rht-D1 genes, and the most frequent of them was Rht-B1b. These alleles provide optimal semidwarf plant height and about 15% higher grain yield. This shows that Rht-B1b, Rht-B1e and Rht-D1b dwarfing alleles are still current for winter wheat breeding in Russia, despite known poor adaptability of gibberellin-insensitive dwarf cultivars to dry environments. The climate of Krasnodar is characterized by prolonged, dry and hot summers. However, the winter wheat growing season spans autumn; winter, which in Krasnodar is relatively short and mild, without constant snow cover; and spring, which is usually favorable for plant growth. Thus, winter crops can escape the adverse conditions of heat and drought. Nevertheless, more efforts are being conducted by breeders recently to shorten the vegetation period of winter wheat. On the other hand, our results show that rather high yields could be achieved without gibberellin-insensitive dwarfism.
We also noted that reduced-height alleles, obviously having the same mechanism of action, cause various degrees of agronomical trait changes. Rht-D1b or Rht-B1e cause stronger height reduction than Rht-B1b, although the grain yield is not much different between groups of accessions having one or another allele of these three. This is consistent with previous findings [3,8].
Generally, the presence of gibberellin-insensitive reduced-height alleles in our study was associated with reduced grain protein content. That is consistent with other studies, and could be explained by the 'dilution' of grain protein by a larger amount of stored carbohydrates [37]. However, grain protein yield per hectare is substantially higher in accessions carrying one of these dwarfing alleles than in 'tall' genotypes. Considering that all accessions in the field experiments were grown under equal conditions, this contradicts the states of some articles claiming weak nitrogen use efficiency in gibberellin-insensitive semidwarf cultivars [15].
Ppd-D1a, the photoperiod-insensitive allele, was observed in 76% of accessions in our collection, which is a relatively high proportion, showing its adaptability for winter wheat in warmer regions. Under conditions of southern Russia, Ppd-D1a gives improvement for several agronomic traits, including earlier heading date, lower plant height and higher 1000-grain weight. Above all, Ppd-D1a is favorable for higher grain yield per hectare. All these Ppd-D1 effects were reported in previous studies [10]. Shortening the vegetation period obviously allows them to escape the terminal drought and heat stresses, and it will be even more actual in the future, as according to the prognoses, global warming will be accompanied by shrinkage of autumn, winter and spring [38].
Despite mildly decreased plant height, the photoperiod-insensitive accessions carrying Ppd-D1a were less tolerant to lodging. That could be caused by the greater spike mass of these plants, which bends the stems to the ground. However, weaker stems due to faster development rate could be a reason. Further research should be conducted to elucidate this fact.
The leaf rust-affected area was significantly lower in accessions carrying Ppd-D1a. In our study, the leaf rust was scored 10 days after heading for each accession individually. Thus, a higher rate of development could allow the photoperiod-insensitive plants to escape the peak of disease development in the field. Previously, the connection of photoperiod insensitivity with disease resistance was reported in wheat, and was explained mainly by the shift of the development stages against the weather events [39]. Alternatively, we can assume co-occurrence of horizontal resistance genes with Ppd-D1a in the same cultivars due to breeding. The race-nonspecific resistance genes were widely introduced in breeding programs together with genes of reduced height and insensitivity to photoperiod, and co-occurrence of them in wheat cultivars of diverse origin could be expected [40].
The TaGS2-A1 gene encodes the most active isoform of plastid glutamine synthase in hexaploid wheat, which performs one of the key steps of nitrogen assimilation. Previously, two haplotypes were discovered (TaGS2-A1a and TaGS2-A1b) differing mainly within noncoding sequences and promotors, and two synonymous SNPs in protein-coding sequences. Previous studies showed advantage of the TaGS2-A1b haplotype in hexaploid wheat [28].
In our study, the TaGS2-A1b haplotype was weakly positively connected with higher grain yield, grain protein yield (but not grain protein content) and lodging resistance. The highest effects were observed within accessions that do not carry any of the 'green revolution' alleles of Ppd-D1, Rht-B1 or Rht-D1 genes. This is consistent with the study of the TaGS2-A1 gene in the Chinese wheat collection, where TaGS2-A1b showed positive effects on plant biomass under any nitrogen supply only among landraces, but not in bred cultivars (most of which are supposed to possess some of the 'green revolution' alleles, while landraces are not). Thus, we can hypothesize that the TaGS2-A1b haplotype could be effective, if one needs to refuse gibberellin-insensitive dwarfism. For example, in many regions of Central Asia, where drought begins earlier during vegetative growth and anthesis, taller genotypes perform much better than semidwarf ones [4]. On the other hand, it may be of value in more northern nonchernozem regions, where Ppd-D1a is not as favorable as in Krasnodar Krai.
A connection of glutamine synthetase with lodging was not expected, but as we know, lodging resistance of wheat depends not only on plant height, but also on the strength of the culm, anchorage power of the root system, and the weight of the spike [41]. Improved nitrogen assimilation in plants carrying TaGS2-A1b could result in the development of stronger stems, less prone to lodging.
Further studies should be aimed at glutamine synthetase biochemical activity, stem anatomy and disease immunity in wheat lines with various genotypes of Rht, Ppd and TaGS2 genes to elucidate our findings.

Plant Material and Phenotyping
The winter bread wheat (Triticum aestivum L.) accessions used in this study were a part of the collection of the National Center of Grain named after P.P. Lukyanenko in Krasnodar, Russia. The origin of accessions and references for their brief characteristics were published in our previous study [35].
Yield testing and phenotyping for agronomical traits were conducted during 2018-2020 harvest years in Krasnodar. The methodology of field experiments and weather conditions during growing seasons were published in our previous study [35]. The brown rust (Puccinia triticina f. sp. tritici) was scored 10 days after heading as a visual estimation of the percentage of the leaf area occupied by the disease.

STS Marker Development for the TaGS2-A1 Gene
The sequences of the glutamine synthetase gene TaGS2-A1 were obtained from the NCBI database (accession numbers GQ169684, GQ169685) [28], and further from the Chinese Spring wheat genome IWGSC RefSeq v1.0 [42], and from the sequences of the 10+ Wheat Genomes Project [36] using BLAST+ software [43] as described earlier [35]. The sequences were aligned and compared using GeneDoc2.7 software [44].
The gene-specific pair of primers flanking the 10 bp insertion/deletion in the promotor (pGS2-A1-F/pGS2-A1-R, Table 1) was designed using Primer-BLAST (NCBI) [45]. The specificity of the primers was checked using alignment of the three homoeologous genes of GS2 (TraesCS2A02G500400, TraesCS2B02G528300, TraesCS2D02G500600). (3) final extension step of 72 • C for 10 min. PCR products were separated in 2% agarose gels with TBE buffer for at least 1 h in an electric field intensity of 6 V/cm, stained with ethidium bromide and documented under UV light using the Gel Doc XR+ system (Bio-Rad Laboratories, Inc., Hercules, CA, USA).

DNA Extraction and PCR-Markers
Genomic DNA was extracted from ground dried leaves of seedlings using the CTABbased protocol [46]. The DNA samples of the two individual plants from each wheat accession were used for genotyping.
Detection of the dwarfing Rht-B1b, Rht-B1e (TraesCS4B02G043100) and Rht-D1b (TraesCS4D02G040400) alleles was performed using allele-specific polymerase chain reaction (ASPCR) markers [47]. For these markers, usually two separate PCRs are performed, and in each mixture a common primer and allele-specific primer with 3 end complementary to the SNP detected are used. The PCR products of each reaction are detected by gel electrophoresis on distinct lanes, and then results are combined to tell homozygotes for one or another allele from heterozygotes. Two pairs of primers were used for detection of Rht-B1b: BF + MR1 for finding the mutant allele, and BF + WR1 for the wild-type allele. Rht-D1b was detected using primers DF + MR2, and Rht-D1a using DF2 + WR2 primers. For Rht-B1e, two primer pairs were used: BF with WR3 to amplify wild-type sequences, while BF with MR3 to amplify a fragment only for Rht-B1e allele [48].
Detection of Ppd-D1 (TraesCS2D02G079600) alleles was performed using a PCR with 3 primers: common primers Ppd-D1_F and Ppd-D1_R1 for wild-type allele Ppd-D1b, giving a 414 bp product; and Ppd-D1_R2 for the mutant 'photoperiod-insensitive' allele Ppd-D1a with the deletion in a promotor of the gene, giving a 288 bp product [24].
The sequences of the primers used for PCR are represented in Table 1. The PCR conditions were used as described in the original protocols. The PCR products were separated in 1.5% agarose gel with TBE buffer for 30 min at electric field intensity of 6 V/cm, stained with ethidium bromide and documented under ultraviolet light using the Gel Doc XR+ system.

Statistical Analysis
The statistical distribution of the majority of the traits tested resembled the normal distribution, except for the values of the leaf rust damage. Before further processing, the percentages of the leaf rust damage were transformed using arcsin √ x , as this transformation keeps the zero values. The best linear unbiased estimates (BLUEs) of the three years of measurements were calculated using Tassel 5 software [49]. The BLUEs for the leaf rust were back-transformed to percentages to be presented in Table S13. To obtain the best fit for the normal distribution, the near-to-zero (<0.2%) BLUE values of the leaf rust damage were discarded (as absolute resistance is not assumed to be phenotype for the genes tested), and the remaining values were ln(x)-transformed. The BLUEs were used for the analysis of variance regarding the molecular markers in Statistica 6.0. We used one-way analysis of variance (ANOVA) for separate genes or a group of Rht-1 genes united through assumed phenotype (tall or semidwarf) as a single factor and separate traits as dependent variables. In addition, we used factorial ANOVA to estimate the gene interaction and to compensate for the allele imbalance in the wheat collection. Presence of GA-insensitive dwarfing alleles (1), alleles of Ppd-D1 (2) and TaGS2-A1 (3) were treated as factors. The factor analysis included each of the three factors (1), (2), (3), their double interactions (1 × 2), (1 × 3), (2 × 3) and the triple interaction (1 × 2 × 3) as fixed effects. Sigma-restricted parameterization and type VI (unique) sum of squares were used. Fisher's F-test was used to estimate the significance of the effects. The adjusted α levels were calculated using the Bonferroni-Holm correction for 8 comparisons (8 traits) in one-way ANOVA and 56 comparisons (8 traits × 7 factors, including double and triple interactions) for multifactorial ANOVA [50]. The least-square means were calculated as an estimation of the group means; the differences between means were accessed using Tukey's HSD (honestly significant difference) test. The boxplots for 3-year BLUEs were built using Rstudio 2021.09.1 Build 372 software, R 4.1.2 programming language and ggplot2 package [51].

Conclusions
Semidwarfism associated with Rht-B1 or Rht-D1 mutations and earliness connected to the Ppd-D1a allele are highly beneficial for the grain yield of winter bread wheat in the southern regions of Russia. The accessions carrying Rht-B1b, Rht-B1e or Rht-D1b lightly differ in plant height but not in the grain yield per unit area. Semidwarfism caused by these alleles was associated with lower grain protein content, but with higher protein yield per unit of area. Despite a mild decrease in plant height provided by the allele of photoperiod insensitivity Ppd-D1a, the accessions carrying it are less resistant to lodging. The TaGS2-A1b haplotype of the plastid glutamine synthetase gene has a tendency to improve lodging resistance and grain yield. However, in the presence of 'green revolution' alleles, the positive effects of TaGS2-A1b on agronomical traits are minimal. Thus, TaGS2-A1b may have potential in breeding taller wheat cultivars or cultivars with alternative dwarfing genes, which may perform better under drier environments.