The Variability of Puroindoline-Encoding Alleles and Their Inﬂuence on Grain Hardness in Modern Wheat Cultivars Cultivated in Poland, Breeding Lines and Polish Old Landraces ( Triticum aestivum

: Wheat ( Triticum aestivum L.) grain hardness is determined mainly by variations in puroindoline genes ( Pina-D1 and Pinb-D1 ), which are located on the short arm of chromosome 5D. This trait has a direct e ﬀ ect on the technological properties of the ﬂour and the ﬁnal product quality. The objective of the study was to analyze the mutation frequency in both Pin genes and their inﬂuence on grain hardness in 118 modern bread wheat cultivars and breeding lines cultivated in Poland, and 80 landraces from Poland. The PCR products containing the Pin gene coding sequences were sequenced by the Sanger method. Based on detected the SNPs (single-nucleotide polymorphisms) we designed CAPS (cleaved ampliﬁed polymorphic sequence) markers for the fast screening of Pinb alleles in a large number of genotypes. All analyzed cultivars, breeding lines, and landraces possess the wild-type Pina-D1a allele. Allelic variation was observed within the Pinb gene. The most frequently occurring allele in modern wheat cultivars and breeding lines (over 50%) was Pinb-D1b . The contribution of the remaining alleles ( Pinb-D1a , Pinb-D1c , and Pinb-D1d ) was much less (approx. 15% each). In landraces, the most frequent allele was Pinb-D1a (over 70%), followed by Pinb-D1b (21% frequency). Pinb-D1c and Pinb-D1g were found in individual varieties. SKCS (single-kernel characterization system) analysis revealed that grain hardness was strictly connected with Pinb gene allelic variation in most tested cultivars. The mean grain hardness values were signiﬁcantly greater in cultivars with mutant Pinb variants as compared to those with the wild-type Pinb-D1a allele. Based on grain hardness measured by SKCS, we classiﬁed the analyzed cultivars and lines into di ﬀ erent classes according to a previously proposed classiﬁcation system.


Introduction
In Europe, common wheat (Triticum aestivum L.) is cultivated on almost 61 million hectares of land, and, in Poland, on 2.416 million hectares, which accounts for about 31% of all cereals cultivated in Poland [1]. Wheat is an important component of food and feed. Wheat grains, depending on their technological properties, are used for various food products [2]. Grain hardness, one of the most important technological traits, is considered as significant as gluten protein composition [3]. It has a direct effect on the grinding and baking characteristics, flour particle sizes, and water absorption Agronomy 2020, 10, 1075 3 of 12 2018-19 in a greenhouse, with eight plants in a twelve liter pot using standard agronomic practices including preventative spraying with plant protection products. The full list of wheat genotypes is presented in Supplementary Tables S1 and S2.

DNA Extraction
The seedlings were grown in beakers with 4 mm glass balls and autoclaved tap water, in darkness at 23 • C. Genomic DNA was extracted from coleoptile fragments and leaves of six-day old seedlings using a modified cetrimonium bromide (CTAB) method devised by Murray and Thompson [35]. DNA was extracted from the pooled tissue samples of at least six individuals from the same variety. The DNA purity and concentration were checked using a Nanodrop 1000 (NanoDrop Technologies) and diluted to 50 ng/µL for further PCR analysis.
The amplification products were purified using the GeneJET PCR Purification Kit (Thermo Scientific), according to the manufacturer's instructions, and then sequenced by the Sanger method using the same primers that were used for amplification. The samples were sequenced by Genomed S.A. Sequences, which were aligned using MegAlgin (DNASTAR) software.

Grain Hardness Measurement
The grain hardness was measured using the Single Kernel Characterization System (SKCS) 4100 (Perten Instruments). Single kernel characterization was measured according to Approved Method 55-31.01 [38]. The hardness of each wheat kernel was determined by the instrumental measurement of the force required to crush the kernel. The SKCS instrument was calibrated to compute the hardness index including the weight, moisture, and diameter of each kernel. For each sample, 300 individual kernels were analyzed.

Percentage of Protein, Starch, and Wet Gluten
The percentage of protein, starch, and wet gluten was measured using the Infratec Nova Grain Analyser (Foss). Analyses were performed for landraces in two independent replicates. Samples of 30 g grains were used.

Statistical Analysis
Statistical analysis was performed using the R v. 3.6.2 and the FSA v. 0.8.27 package [39]. The normality distribution was verified by the Shapiro-Wilk test. The Kruskal-Wallis test was performed to compare variable distributions of Pinb allele groups and hardness index. The Dunn test was performed in order to establish which groups were different from each other. Pearson correlationa between hardness and protein or starch or gluten content were calculated separately for landraces with the wild-type Pinb alleles and mutant alleles, as well as for all landraces, regardless of their allelic status. The chart of hardness index ranges was performed using gglot2 v. 3.2.1 package [40].

Results
The allelic variation of Pina and Pinb was determined by Sanger sequencing. All tested cultivars, breeding lines, and landraces had the wild-type allele of the Pina gene (Pina-D1a). The frequencies of Pinb-D1 alleles are shown in Table 1. The most frequent allele among modern cultivars and breeding lines was Pinb-D1b (57%). The wild type Pinb-D1a allele occurred in 19 lines (16%) and the frequencies of the remaining alleles, Pinb-D1c and Pinb-D1d, were at a similar level (12% and 15%, respectively). In landraces, the Pinb frequencies were different to modern cultivars and breeding lines. Over 71% had the Pinb-D1a allele, while 26% (21) had the Pinb-D1b allele. Pinb-D1c and Pinb-D1g were each found in one tested landrace. Alignment of the nucleotide and amino acid sequences of Pinb alleles with the marked mutations is shown in Figure 1, and chromatogram fragments of five alleles found in tested wheat genotypes are shown in Supplementary Figure S1.

Results
The allelic variation of Pina and Pinb was determined by Sanger sequencing. All tested cultivars, breeding lines, and landraces had the wild-type allele of the Pina gene (Pina-D1a). The frequencies of Pinb-D1 alleles are shown in Table 1. The most frequent allele among modern cultivars and breeding lines was Pinb-D1b (57%). The wild type Pinb-D1a allele occurred in 19 lines (16%) and the frequencies of the remaining alleles, Pinb-D1c and Pinb-D1d, were at a similar level (12% and 15%, respectively).  Next, we examined whether the mutations in the Pinb coding sequence could be detected by CAPS analysis. Restriction enzyme digestion of the amplified full length Pinb coding sequences produced allele-specific fragment patterns, which were unique for each enzyme. The Pinb-D1b, Pinb-D1c, and Pinb-D1d alleles can be distinguished from the wild-type Pinb-D1a allele by restriction analysis with MbiI, PvuII, and MnlI, respectively ( Figure 2). The SKCS hardness values were grouped by each Pinb allele and by genotype (modern vs. landraces) ( Figure 3). In modern cultivars and breeding lines the hardness index for each allele group ranged as follows: for Pinb-D1a, from 6 to 57; for Pinb-D1b, from 35 to 80; for Pinb-D1c, from 42 to 77; for Pinb-D1d from 40 to 82. In landraces, the hardness index for the Pinb-D1a group ranged from 9 to 37 and for Pinb-D1b from 18 to 63. The landrace with Pinb-D1c had a hardness index of 66, and another with Pinb-D1g had a hardness index of 61 (Supplementary  Tables S1 and S2).  In landraces, the Pinb frequencies were different to modern cultivars and breeding lines. Over 71% had the Pinb-D1a allele, while 26% (21) had the Pinb-D1b allele. Pinb-D1c and Pinb-D1g were each found in one tested landrace. Alignment of the nucleotide and amino acid sequences of Pinb alleles with the marked mutations is shown in Figure 1, and chromatogram fragments of five alleles found in tested wheat genotypes are shown in Supplementary Figure S1.
Next, we examined whether the mutations in the Pinb coding sequence could be detected by CAPS analysis. Restriction enzyme digestion of the amplified full length Pinb coding sequences produced allele-specific fragment patterns, which were unique for each enzyme. The Pinb-D1b, Pinb-D1c, and Pinb-D1d alleles can be distinguished from the wild-type Pinb-D1a allele by restriction analysis with MbiI, PvuII, and MnlI, respectively ( Figure 2). The SKCS hardness values were grouped by each Pinb allele and by genotype (modern vs. landraces) (Figure 3). In modern cultivars and breeding lines the hardness index for each allele group ranged as follows: for Pinb-D1a, from 6 to 57; for Pinb-D1b, from 35 to 80; for Pinb-D1c, from 42 to 77; for Pinb-D1d from 40 to 82. In landraces, the hardness index for the Pinb-D1a group ranged from 9 to 37 and for Pinb-D1b from 18 to 63. The landrace with Pinb-D1c had a hardness index of 66, and another with Pinb-D1g had a hardness index of 61 (Supplementary Tables S1 and S2).

MbiI
PvuII MnlI  The tested genotypes were classified, as proposed by Corona et al. [41] into three classes, namely soft, medium hard, and hard types with the SKCS values in the range of 15-40, 41-70, and 71-95 respectively. In the modern cultivars group, 20 genotypes were classified as soft type, 74 as medium hard, and 24 as hard type. In the group of landraces, 61 genotypes were classified as soft and 19 as medium hard.
The protein content in the tested landraces ranged from 13.10% to 23.15%. The starch content ranged from 55.85% to 67.70%, and the wet gluten content ranged from 28.05% to 53.15% (Supplementary Table S3). None of the above values were significantly correlated at p < 0.05 with the hardness measured by SKCS. The correlation coefficient between grain hardness and protein, starch and wet gluten content calculated for all landraces was −0.18, 0.19, and −0.16, respectively. The same correlation coefficients calculated for the wild-type Pinb landraces were −0.18 (SKCS vs. protein content), 0.15 (SKCS vs. starch content), and −0.03 (SKCS vs. wet gluten content), and for the mutant landraces, these values were −0.41, 0.36, and −0.45, respectively. The normality distribution of hardness values in each allele group was verified by the Shapiro-Wilk test. The test was positive only for the Pinb-D1b group in modern wheats. The Kruskal-Wallis test was performed to compare variable distributions between Pinb allele groups with a breakdown into modern wheat and landraces, and hardness, and excluding single observations (Pinb-D1c and Pinb-D1g in landraces). The results proved to be statistically significant (p-value < 2.2 × 10 −16 ). In the next step, the Dunn test was performed in order to establish which groups were different from each other. The SKCS values of modern wheats and landraces with Pinb-D1b, Pinb-D1c, or Pinb-D1d were significantly different compared to genotypes with wild-type Pinb allele (Table 2). There were no significant differences between genotypes carrying the Pinb-D1b, Pinb-D1c, or Pinb-D1d alleles ( Table 2). The tested genotypes were classified, as proposed by Corona et al. [41] into three classes, namely soft, medium hard, and hard types with the SKCS values in the range of 15-40, 41-70, and 71-95 respectively. In the modern cultivars group, 20 genotypes were classified as soft type, 74 as medium hard, and 24 as hard type. In the group of landraces, 61 genotypes were classified as soft and 19 as medium hard.
The protein content in the tested landraces ranged from 13.10% to 23.15%. The starch content ranged from 55.85% to 67.70%, and the wet gluten content ranged from 28.05% to 53.15% (Supplementary Table S3). None of the above values were significantly correlated at p < 0.05 with the hardness measured by SKCS. The correlation coefficient between grain hardness and protein, starch and wet gluten content calculated for all landraces was −0.18, 0.19, and −0.16, respectively. The same correlation coefficients calculated for the wild-type Pinb landraces were −0.18 (SKCS vs. protein content), 0.15 (SKCS vs. starch content), and −0.03 (SKCS vs. wet gluten content), and for the mutant landraces, these values were −0.41, 0.36, and −0.45, respectively. The normality distribution of hardness values in each allele group was verified by the Shapiro-Wilk test. The test was positive only for the Pinb-D1b group in modern wheats. The Kruskal-Wallis test was performed to compare variable distributions between Pinb allele groups with a breakdown into modern wheat and landraces, and hardness, and excluding single observations (Pinb-D1c and Pinb-D1g in landraces). The results proved to be statistically significant (p-value < 2.2 × 10 −16 ). In the next step, the Dunn test was performed in order to establish which groups were different from each other. The SKCS values of modern wheats and landraces with Pinb-D1b, Pinb-D1c, or Pinb-D1d were significantly different compared to genotypes with wild-type Pinb allele ( Table 2). There were no significant differences between genotypes carrying the Pinb-D1b, Pinb-D1c, or Pinb-D1d alleles ( Table 2).

Discussion
We analyzed grain hardness within a collection of various wheat cultivars, breeding lines currently applied in wheat breeding programs and old landraces collected during multiple expeditions between 1960 and 1976 in Poland. Cultivars were tested both at the molecular level (by determining Pin gene allelic status) and at the phenotype level (by measuring wheat kernel SKCS hardness index).
Surprisingly, no allelic variability within the Pina gene was observed, although mutant alleles of this gene were detected in many other European cultivars [15,24,42]. All tested cultivars, breeding lines, and landraces had the wild-type Pina-D1a allele. This allele, originally described by Gautier et al. [5], was proven to be the most common Pina allele among wheat genotypes [10,12]. The second Pina allele found in European cultivars was Pina-D1b, discovered by Giroux and Morris [9] which is extremely rare. Other Pina alleles mainly occurred in landraces from China or India [17,18]. The lack of Pina variability in the European modern wheat cultivars and breeding lines tested here indicates negative mutant gene selection during the breeding process.
More common allelic variability was observed for the Pinb gene. The Pinb-D1b allele, detected in 57% of tested modern cultivars and breeding lines, was the most frequent. The remaining Pinb alleles were the wild-type Pinb-D1a and the mutant Pinb-D1c and Pinb-D1d. They were represented by a similar variant number (11-16%). Similar distribution of the same Pinb alleles was observed previously in European wheat populations [15,24,43]. As for the landraces, the Pinb allele distributions were different. The main, wild-type Pinb-D1a allele was found in over 70% of tested landraces. The next most frequent was Pinb-D1b, detected in 26% of the tested landraces. Pinb-D1c was found only in one variety, and none contained the Pinb-D1d allele. It was surprising for us to find a landrace with the Pinb-D1g allele. To date, this allele has not been observed in European wheat cultivars. It was reported for the first time in historical varieties from North America [16]. We did not find such high variability among Pina and Pinb alleles, which were observed in hard wheat cultivars originating from other continents. For example, among those originating from Asia, seven various Pinb alleles, including four newly described, such as Pinb-D1ad, Pinb-D1ae, Pinb-D1af and Pinb-D1ag, have been found [17]. In another study, nine Pinb alleles were described [44]. Therefore, our observation confirms that these four alleles found among modern wheats cultivated in Poland dominate in modern European varieties. This is probably caused by the narrow genetic variability of breeding lines as well as being the result of breeding pressure.
The most commonly used methods of determination of grain hardness include the particle size index (PSI), near-infrared reflectance (NIR) with a properly prepared library, and the single kernel characterization system (SKCS). We decided to use SKCS due to the high reliability and accuracy of Agronomy 2020, 10, 1075 8 of 12 this method, and the possibility of analyzing 300 individual grains in a single run. Based on the SKCS value, wheats can be classified into different categories. The first classification was proposed by Morris et al. [16], who distinguished four groups according to the SKCS hardness index: (I) from 0 to 33, (II) from 34 to 46, (III) from 47 to 59, and (IV) above 60. Slightly different categories were proposed by Corona et al. [41] who divided wheat into three main groups: soft (SKCS , medium hard , and hard (71-95). The second wheat classification system, used also by us, is the most common. However, in many regions, other classes are used as well, depending on local needs or regulations. For example, Sharma et al. [45] proposed five classes for a more precise classification of soft and hard cultivars from India. They included very soft (SKCS 34 or less), soft (35)(36)(37)(38)(39)(40)(41)(42)(43)(44)(45)(46)(47)(48)(49)(50)(51)(52)(53)(54), medium hard (SKCS 55-74), and hard (90 or more).
It was noted that some of the tested cultivars were classified into groups contrary to ourexpectations based on their allelic status. For example, in case of the Pinb-D1a, a group of six cultivars and breeding lines had the SKCS indexes ranging from 31 to 56, with a mean value of 46, which was higher than usual for wild-type alleles. In previous reports, cultivars with both wild-type Pina and Pinb alleles had low grain hardness values ranging from 7 to 40, with a mean value of 27 [41], or from 15 to 29, with a mean value of 24.5 [16]. Moreover, some cultivars, breeding lines, and landraces with Pinb-D1b or Pinb-D1d mutant alleles, which were expected to have higher grain hardness values, were classified as soft, with a hardness index of 15 to 40. Similarly, one cultivar and one breeding line carrying Pinb-D1d were classified as soft.
Particularly interesting is the landrace with the accession number I31930, (Pinb-D1b) of which the hardness index is 17.43 (±18.59). Such a low value has never been reported in the case of the Pinb-D1b allele. In five breeding lines and seven landraces, the standard deviation of SKCS hardness was above 20, which might suggest that the lines represented a mixture of hard and soft genotypes. This is possible because landraces have been sourced from farmland or local markets and may be contaminated by seeds from other genotypes. In turn, the breeding lines may not yet be genetically homogenous.
Statistical analyses do not show differences in grain hardness between genotypes with the mutant alleles of Pinb in both groups. Statistically significant differences occurred only between genotypes carrying the wild-type allele Pinb-D1a and genotypes with mutated alleles, which proved that the mutation causing the amino acid sequence change leads to increased grain hardness.
Differences in grain hardness between the same alleles may be caused by other biochemical components of the grain [11,43,46]. Therefore, we also investigated the total protein content, starch, and wet gluten measured by the NIR method. Unfortunately, no correlation between any of these factors was found. This contradicts the results reported by Salmanowicz et al. [47], who found a strong correlation between grain hardness and protein content. It should be noted however, that in the aforementioned work the grain hardness was measured using the NIR and the PSI methods, whereas in the current study the SKCS was applied. As mentioned earlier, the method of measurement influences the results because each method defines the grain hardness in a slightly different way [46,48,49]. The influence of protein content on grain hardness was also reported in other papers [11,50,51]. Moreover grain hardness might be affected by other grain components such as pentosans, primarily arabinoxylans and arabinogalactans [11,52], and gliadin compositions [50], as well as gluten proteins [53]. Other reasons for differences in grain hardness may be the result of differences in genetic background other than variability in Pin genes.
Nirmal et al. [34] described significant differences between the expression of Pin genes in hard and soft wheat and concluded that higher expression at earlier stages of grain development causes a softer grain, but this does not fully explain entirety of the variability between the genotypes studied. On the other hand, increased expression of Pin genes in rice, durum wheat, or in common wheat caused a significant decrease in grain hardness [27,54,55]. The opposed relationship was shown by the silencing of the Pin genes, which resulted in increased grain hardness [25,26]. Another known genetic factor, which might affect grain hardness is nucleotide polymorphisms of the transcription activator SPA (storage protein activator) [56]. In addition many QTLs (quantitative trait locus) have been described Agronomy 2020, 10, 1075 9 of 12 that affect the grain hardness. The QTLs related to grain hardness are located on chromosomes 1A, 1B, 2B, 2D, 3B, 4A, 4B, 5AL, 5BL, 5D, 6A, 6BL, 7A, and 7B [57][58][59][60][61][62]. It has also been proven that grain hardness is affected by environmental factors [49,50,63]. However, in our investigations all cultivars, breeding lines and landraces grew under the same optimal conditions, so we assume that this factor was eliminated. Another factor, which was proven to influence kernel texture is the grain moisture [11,51]. However, in our research, this parameter was almost identical in all tested lines, with the standard deviation values below 1%.

Conclusions
In conclusion, no new Pin alleles have been discovered however, the Pinb-D1g allele, unique to European wheat, has been found among landraces from Poland. In addition, we found many interesting objects whose grain hardness seemed to be determined not only by Pina and Pinb alleles but also by other genetic or biochemical factors. In further studies, we aim to unravel the genetic background of these inconsistences.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2073-4395/10/8/1075/s1, Table S1: Means for kernel hardness, weight, moisture and diameter of modern wheat. Table S2: Means for kernel hardness, weight, moisture and diameter of Polish landraces. Table S3: Means for protein, starch, and wet gluten contents of Polish landraces. Figure S1: Chromatogram fragments of five alleles found in tested wheat genotypes.