Exploring karyotype diversity of Argentinian Guaraní maize landraces: Relationship among South American maize

In Argentina there are two different centers of maize diversity, the Northeastern (NEA) and the Northwestern (NWA) regions of the country. In NEA, morphological studies identified 15 landraces cultivated by the Guaraní communities in Misiones Province. In the present study we analyzed the karyotype diversity of 20 populations of Guaraní maize landraces through classical and molecular cytogenetic analyses. Our results demonstrate significant intra and inter-populational variation in the percentage, number, size, chromosome position and frequencies of the heterochromatic blocks, which are called knobs. Knob sequence analysis (180-bp and TR-1) did not show significant differences among Guaraní populations. B chromosomes were not detected, and abnormal 10 (AB10) chromosomes were found with low frequency (0.1≥f ≤0.40) in six populations. Our results allowed karyotypic characterization of each analyzed population, defining for the first time the chromosomal constitution of maize germplasm from NEA. The multivariate analysis (PCoA and UPGMA) of karyotype parameters allowed the distinction between two populations groups: the Popcorn and the Floury maize populations. These results are in agreement with previously published microsatellite and morphological/phenological studies. Finally, we compared our karyotype results with those previously reported for NWA and Central Region of South America maize. Our data suggest that there are important differences between maize from NEA and NWA at the karyotype level, supporting the hypothesis that there are two pathways of input of South America maize. Our results also confirm the existence of two centers of diversification of Argentinian native maize, NWA and NEA. This work contributes new knowledge about maize diversity, which is relevant for future plans to improve commercial maize, and for conservation of agrobiodiversity.


Introduction
The Northern region of Argentina is one of the southernmost areas of maize landraces cultivation, where Cámara Hernández et al. [1] described 51 morphological native landraces In this region, there are two different centers of diversity: the highland region or Northwestern (NWA) and the Mesopotamic and Chaco plains or Northeastern (NEA) [2]. In the NEA region, 23 native maize landraces have been described, of which 15 are cultivated by the Guaraní communities from the subtropical forests in Misiones province. These landraces have agronomic characteristics that can greatly enrich current agriculture [1,3]. Analysis of molecular markers and morphological/phenological parameters suggest that these landraces can be divided into different gene pools, corresponding with the type of kernels, the Floury and the Popcorn landraces [3,4].
Previous studies in lines and landrace of maize from Argentina have shown that the genome size ranges from 4.4 to 6.9 pg [5][6][7][8]. Realini et al. [7] reported a 2C DNA variation from 4.62 to 6.29 pg in 20 Guaraní populations, revealing significant differences among them. In the Zea genus, the variation in DNA content is mainly due to differences in the amount of heterochromatin at distal chromosome blocks called knobs, as well as the presence of B chromosomes [5,9]. Furthermore, there are differences in the amount of interspersed DNA, such as retrotransposon families [10].
Variations in the number, size and position of knobs were reported in American maize landraces [5,6,7,8,11,12,13]. Kato 1976 [11] reported that knobs could be found in 34 different chromosomal positions; a knob may be present or absent at these positions, and its presence and size is a heritable trait. Due to their conserved character, knobs have played an important role in studying the origin and relationships among races of American maize [12]. Knobs consist mainly of highly repeated tandem arrangements of two families of satellite-type repeat sequences of 180-bp and 350-bp (TR-1). Both sequences are repeated thousands to millions of times in different proportions relative to each other, forming the different types of knobs; those knobs formed exclusively by 180-bp or TR-1 repeats, and those formed by both sequences in different ratios [14][15][16]. The knobs sequence composition has been also used to characterize maize landraces [6,8,17]. FISH experiments using simultaneously 180-bp and TR-1 as probes, revealed the sequence composition of each knob in NWA landraces and 12 Guaraní population from NEA [6,8,17]. In NWA landraces the proportion of knobs that hybridized with the 180-bp were positively correlated with cultivation altitude in contrast with the proportion of knobs that hybridized with the TR-1 that were negatively correlated [18]. The joint analyses of different karyotype parameters (morphology and chromosome size, number, size and position of knob and frequency of B chromosomes), including the knob sequence composition, allowed the cytogenetic characterization of the NWA landraces [6]. Some cytological parameters were preliminary described in Guaraní populations to further explore their relationship with the intra-specific genome size variation [7]. Therefore, there is not a comprehensive study that allows the karyotype characterization of NEA maize.
Another karyotypic parameter that was used for maize landraces characterization is the presence and frequencies of B chromosomes (Bs). Maize Bs are widely distributed and large intra-and inter-populational differences in their numbers and frequencies have been reported in NWA landraces [5,6,13,18,19]. Because Bs have not been previously detected in Guaraní landraces from NEA [20], exhaustive analysis of their occurrence will be a significant contribution to understanding the role of these chromosomes in the diversification of maize landraces. Moreover, the presence of abnormal chromosome 10 (AB10) was also reported and used for cytogenetic characterization of maize and teosinte landraces from America [12,21]. AB10 is characterized for a knob on its long arm that have about the size of their short arm. Maize neocentromeres observed during meiotic anaphases are associated with AB10 [22][23][24]. Until now, AB10 chromosomes have not been reported in Argentinian landraces.
McClintock et al. 1981 [12], based on knobs chromosome constitution of American landraces, suggested that different types of maize were introduced early at two initial centers of cultivation: Northern South America and the Central Andean highlands. They proposed that maize germplasm from the Northern South America region had a vast influence on the races found in the Caribbean Islands and on those in Eastern South America, whereas races from the Andean Center spread extensively throughout the American Southwest. Using microsatellites analysis, Lia et al. 2009 [25] established the affiliation between NWA landraces and Andean complex maize. Recently, an extensive analysis of molecular markers determined the relationship between NEA landraces and those of American Continent lowlands, revealing the existence of three different genetic groups: Tropical Lowland, NEA Flourys and NEA Popcorns [4].
The aim of this work was to analyze the intra-and inter-populational karyotypic variability of Guaraní maize landraces from NEA, to study the chromosome positions, frequencies and sequences composition of knobs, and to define for the first time the chromosomal constitution of this Argentinian maize germoplasm. Our karyotype parameters analysis allowed us to explore the relationships among NEA maize landraces and their affiliation with NWA landraces and with those cultivated in the vicinity regions of lowland South America. This work contributes to knowledge of maize global diversity, an indispensable requirement for its integration into future plans for the improvement of commercial maize and for conservation of agrobiodiversity.

Plant materials
Twenty Guaraní maize populations from Northeastern Argentina (NEA) were collected from Guaraní farmers in Misiones Province, Argentina (Table 1). The specimens were deposited at the seed bank of the Vavilov Laboratory, FA-UBA.
In Table 1, the type of grain is indicated for each population, Floury (F), Floury with corneal periphery (F-Pc) and Popcorn (Pc). The term population is used here to refer to a set of individuals belonging to a morphological race, which are cultivated by a farmer or family group in a location.

Methods
Seeds were germinated at 28˚C for 2-3 days in Petri dishes containing wet filter paper. Primary root tips, 0.5-1 cm in length, were pre-treated with 8-hydroxiquinoline (0.02 M, Sigma) for 5 h at room temperature. Then they were fixed in 3: 1 (ethanol: acetic acid) and stored at 4˚C until use.

Chromosomal preparations
Mitotic metaphase preparations were performed in 20 maize populations. Fixed root tips were treated with an enzymatic solution (2% cellulose, Onozuka R10 Merck, and 20% Pectinase, Sigma P4716) for 1 h at 37˚C. After freezing to remove the coverslips, the slides were air-dried and stored at 4˚C until use.

DNA probes
For FISH analysis, the maize knob sequences, 180-bp, TR-1(350 bp) and rDNA 5S and 18S were obtained from GenBank (http://www.ncbi.nlm.nih.gov/), the centromeric sequences of maize CentC probe was designed by a CentC consensus repeats provided by Dr. Kelly Dawe of the University of Georgia, USA. The primers were designed using the Primer3 program (version 0.6) provided by the Whitehead Institute for Biomedical Research & Howard Hughes Medical Institute, USA (http://bioinfo.ut.ee/primer3/). These sequences were isolated and amplified from total genomic DNA of maize by polymerase chain reaction (PCR) methods [27]; The PCR cycling conditions were 4 min at 94˚C, followed by 30 cycles of 1 min at 94˚C,

Fluorescence in situ hybridization
The FISH procedure was performed as Realini et al. [7], with minor modifications. Slide preparations were incubated in 100 μg mL -1 of RNAse (Sigma) in 2× saline sodium citrate (2 × SSC) for 1 h at 37˚C in a humidified chamber and washed three times in 2 × SSC for 5 min each at room temperature. The slides were post-fixed in freshly prepared 4% (w/v) paraformaldehydein (Fluka) for 10 min and then washed in 2 × SSC for 15 min at room temperature. Then, the preparations were dehydrated in a graded ethanol series and air-dried. To each preparation was added 30 μL of the hybridization mixture, which contained 50 ng of each labelled probe. The hybridization mixture was denatured for 15 min at 75˚C. The slides were placed on a thermocycler at 75˚C for 7 min, 45˚C for 10 min and 38˚C for 10 min. Then the slides were incubated overnight at 37˚C. Post-hybridization washes were carried out [7]. The slides were incubated in the detection buffer containing 2.5% bovine serum albumin and the corresponding detection antibodies Streptavidine-CY3 conjugate (Sigma) or antidigoxigeninfluorescein isothiocyanate-FITC (Sigma), for 1 h at 37˚C. Slides were washed three times in 4 × SSC/Tween buffer for 10 min at room temperature, counterstained with 1 μg mL -1 of DAPI in 4 × SSC/Tween buffer for 40 min at room temperature and mounted in Vectashield antifade solution (Vector Laboratories, Burlingame, CA, USA). Slides were examined with a Carl Zeiss Axiophot epifluorescence microscope (Carl Zeiss, Germany), with appropriate Carl Zeiss filters coupled with a Leica DC 250 digital camera and with an image analyzer Leica IM 1000. The location of hybridization signals and DAPI positive bands were based on the observation of at least 20 complete metaphases for each analyzed individual.

Karyotype parameters
For characterization and detection of Bs and AB10 chromosomes, 4 to 9 individuals were studied for each population sampled. At least 20 similar condensed metaphases for each individual were chosen, photographed and used for karyotype parameters estimation. The identification of each pair of chromosomes was based on the maize chromosomal morphology described by McClintock et al. 1981 [12], as well as on ribosomal and centromeric sequences mapping by FISH. Karyotype parameters were determined using MicroMeasure V.3.3 (www.colostate.edu/ Depts/Biology/MicroMeasure). Centromeric index (CI) and total chromosome length (TCL) were estimated. The symmetry of karyotypes was studied using the A 1 [28], CV CL , CV CI [29] and M CA indexes [30]. Consensus idiograms were performed for each Guaraní maize landrace using Adobe Photoshop CS3 version 10.0. Chromosome position, size and appearance frequencies of each knob were indicated. The knob frequencies were represented in a histogram for each population. In the histograms and idiograms, the most frequent (f! 0.6) positions were indicated with black bars/block, while the least frequent positions (f< 0.6) were indicated with white bars/blocks. For all data consensus idiogram and frequencies appearance of knob histogram were constructed, representing the karyotypes from Guaraní maize from NEA. Knob frequencies appearance of knob histograms were constructed for the populations with different type of grains, Floury (F), Floury grains with corneal periphery (F-Pc) and Popcorn (Pc) maize, respectively.

Statistical analysis
Correlation analyses were performed using the Spearman coefficient. The variables correlated were: percentages of heterochromatin, number of knobs, total chromosome length (TCL) and asymmetric indexes (A 1 , CV CL , CV CI and M CA ). A linear regression analysis was performed between percentages of heterochromatin and the number of knobs. These statistical analyses were considered significant at P-values 0.05. For each continuous karyotype variable (percentage of heterochromatin, numbers of knobs, TCL and CV CI , CV CL , M AC and A 1 indexes), an analysis of variance (ANOVA) was performed to test the karyotype differences among populations with different types of grain. The variance was modulated with VarIdent. The seven ANOVAs were applied with a global level of 0.05 using the Bonferroni´s method, so the differences were considered significant at P-values 0.007. These analyses were followed by multiple comparison Fisher's least significant difference test [31].
A principal coordinated analysis (PCoA) and a UPGMA multivariate analysis were performed using the continuous and binary (presence/absence of knob positions) karyotype variables. The 6S and 6-Sat knob positions were excluded from these analyses if they were either always absent or always present in the studied individuals, respectively. In both analyses the data was standardized. Since a mixture of variable types (binary and continuous) was used, Gower distance-based similarity matrix was employed [32]. For the PCoA analysis Biplots, two-dimensional graphs, axis1 vs. axis 2, axis1 vs. axis 3 and axis 1 vs. axis 4 were represented. In these graphs the contours correspond to 80% prediction ellipses for individuals belonging to F and Pc maize.
All statistical analyses were performed in 17 Guaraní populations using the program Infostat, FCA, National University of Córdoba [33]. For the unbalanced design, the R package [34] was used.
The percentages of heterochromatin were calculated as the percentage of the total chromosomal complement occupied by the knob heterochromatin. A variation range from 5% to 20% was found among populations. For example, the VAV6574 population had the lowest mean number of knobs ( " X = 9.83) and the lowest percentage of heterochromatin ( " X = 6.37%), while the VAV6563 population showed the highest mean number of knobs ( " X = 19.75) and the highest percentage of heterochromatin ( " X = 16.71%). Intra-populational variation in the number of knobs and percentage of heterochromatin was also detected ( Table 2). For example, the VAV6564 population varied from 8 to 19 knobs, and the percentage of heterochromatin ranged from 5.57% to 11.51%. A significant positive relationship was found between the number of knobs and percentages of heterochromatin (p 0.0001, Spearman coefficient (SC) = 0.875, df = 95), and a positive linear relationship was detected between both parameters (Y = -2.97 + 0.94X, r = 0.71, n = 101, F = 239.96, p <0.0001, where Y represents the percentages and X indicates the number of knob).
The CI, TCL and karyotype asymmetry indexes (A 1 , M AC , CV CI and CV CL ) were calculated for each studied individual (Table 2 and S1 Table). The chromosomal asymmetry indexes showed inter-and intra-populational variation ( Table 2). These karyotype parameters were correlated with percentages of heterochromatin and numbers of knob, respectively. The CV CI was significantly correlated with percentages of heterochromatin and number of knobs, while TCL, A 1 , M AC and CV CL did not show significant correlation with both karyotype parameters (S2 Table).

Karyotype differences among Floury, F-Pc and Popcorn populations
The Popcorn and Tupí (F-Pc) populations showed the highest values in number of knobs and percentage of heterochromatin, while the Floury populations showed the lowest values for these two parameters ( Table 2). Analysis of variance (ANOVA) performed on the karyotype parameters (number of knobs, percentage of heterochromatin, TCL, A 1 , M AC CV CI, and CV CL indexes) showed significant differences in the number of knobs, percentage of heterochromatin and CV CI index (F 2,14 = 26.55, P< 0.0001; F 2,14 = 85.47, p< 0.0001; F 2,14 = 9.98, p< 0.0022; respectively) among the different types of grain populations. The Pc and F populations showed significant differences in these three karyotype parameters. The F-Pc populations did not show differences with Pc or F populations in the number of knobs and percentage of heterochromatin, but they did show significant differences with F populations in their CV CI index values.

FISH sequences mapping
Fluorescent in situ hybridization (FISH) assays on mitotic metaphases, using 180-bp and TR-1 knob sequences as probes, showed variations in the hybridization signals of the different knobs. In the studied individual, the majority of their knobs exhibited 180-bp and TR-1 signal overlapped (mixed knobs). The knobs that were hybridized with one single knob sequence showed, in general, the 180-bp signal (Figs 1D, 1F, 2B, 2D, 2H and 2I). Although in VAV6592 the majority of knobs were mixed, in some chromosome pairs both signals did not overlapped,  The ribosomal sequences 18S and 5S hybridized on the secondary constriction of the short arm of chromosome 6 (6S) and on the long arm of chromosome 2 (2L), respectively, with no differences in intensity of hybridization signals between the analyzed individuals. Hybridization signals from the centromeric sequence CentC allowed a precise identification of chromosomal morphology, and showed variation in intensity and size among chromosomes from the same metaphase and among individuals from different populations (Fig 2F).

Guaraní landraces chromosomal characterization
In S1 and S2 Figs, the relative size, chromosome position, and the most and least frequent positions of knobs are summarized in representative idiograms for each population. The frequencies for each chromosomal position of knob are shown in histograms (S1 and S2 Figs).
To represent the chromosomal constitution of Guaraní maize from NEA, we constructed a consensus idiogram and histogram with data from all analyzed populations (Fig 3A and 3B). The principal components analysis (PCoA) showed that the first 10 axis account for 72% of the total variability, and the first four account for 43% of total variability. Biplots graphs showed that individuals from Floury populations were more closely related within each other than with those of Popcorn populations. The individuals from F showed different distributions than those of Pc populations, showing karyotype differences between individuals belonging to these 2 groups. Individuals from the F-Pc populations showed a wide dispersion (S3 Fig). The UPGMA cluster analysis (cophenetic correlation = 0.746) showed two groups, G1 and G2, at an approximately Gower distance of 0.6. The G1 group included all Pc populations and F-Pc populations of Tupí (VAV6563 and VAV6592). The G2 group included all the F populations and two F-Pc populations of Colorado (VAV6837 and VAV6573) (Fig 4).

Discussion
The joint analysis of karyotype parameters provides the first thorough cytogenetic characterization of Guaraní Argentinian maize from Northeastern Argentina (NEA). The cytogenetic variability of 20 populations of Guaraní landraces was analyzed. Furthermore, our karyotype results were compared with those previously reported from Northwestern Argentina (NWA) and Central Region of South America maize.

Karyotype parameters: Intra-and inter-populational variability
In maize Guaraní landraces, high intra-and inter-populational karyotype variability were detected. The inter-populational variation in the number of knobs and percentage of heterochromatin was significant, varying from 8 to 23 knobs and from 5.06% to 20.02%, respectively. Although a significant positive relationship was found between the number of knobs and percentage of heterochromatin, individuals of the same population with the same number of knobs, not always exhibited similar percentages of heterochromatin. This suggests that the percentage of heterochromatin in each individual depends not only in the number but also in the size of knobs, given mainly by the number of copies of the satellite DNA repeats that conform them. Considering all the populations that were analyzed, knobs were observed at 20 different chromosomal positions, with inter-populational differences in their frequencies. Knob positions 6L 2 and 6L 3 , described in pachytene chromosomes by McClintock et al. 1981 [12], were only observed in the VAV6560 population, which was a distinctive karyotype trait this population. It is interesting to note that populations of the same race exhibited similarities in the percentage of heterochromatin as well as in the number and the frequent positions of knobs. In addition, chromosomes with AB10 morphology were detected, with low frequencies (0.1! f 0.40), in only six populations. On this basis it could be concluded that the populations of the Guaraní landraces could be characterized by the set of karyotype parameters presented here, defining for the first time their chromosomal constitution. Therefore, the karyotype Exploring karyotype diversity of Argentinian Guaraní maize landraces parameters could be used as another index of variability and taken into account in maize landraces studies for sure.
The percentage of heterochromatin and the number of knobs did not show a significant correlation with the CV CL , A 1 and M AC indexes. This could be due to the great intra-and inter-populational variability detected in position and size of knobs, and to differences in the content and location of other repetitive sequences scattered throughout the genome. However, the relationship found between the CV CI index and the number of knobs, and the percentage of heterochromatin could be explained because centromeric index and karyotype asymmetry depend on chromosomal heterochromatin distribution. In addition, the amplification of scattered sequences also affects the chromosomal arm size and consequently, the centromeric index [35].
The FISH experiments allowed detecting the sequence composition of each knob. The majority of studied individuals had all their knobs mixed. As the 180-bp hybridization signals were more intense than TR-1 signals, it is possible to infer that the mixed knobs would be composed of greater numbers of 180-bp repeats than TR-1 repeats. It has been proposed that TR-1 arose by duplication and divergence of the 180-bp sequence [15,16]. Therefore, it is possible that the 180-bp repeat formed knobs before TR-1 appeared. In this scenario, TR-1 might have invaded 180-bp loci, optimizing its chances of being transmitted preferentially through the Exploring karyotype diversity of Argentinian Guaraní maize landraces AB10-mediated meiotic impulse [23,36]. On the other hand, the sequence composition of each knob detected here did not allow discrimination among Guaraní landraces.
Microsatellite studies and multivariate analysis of morphological, phenological and reproductive traits identified two different genetic groups within the NEA maize, Floury and Popcorn [3,37]. In the present study, the number of knobs and the percentages of heterochromatin showed that the Popcorn and Tupí populations differ from the Floury populations. In addition, the Floury populations have only six knobs positions exhibiting high knob frequency, whereas F-Pc and Popcorn populations showed a higher number of knob positions with high frequency. The UPGMA and the PcoA analyses supported the differences between the Floury and Popcorn populations at the karyotype level. F-Pc populations exhibited wide dispersion, where the Tupí populations were more closely related to the Popcorn populations. The congruence between the clusters obtained from the analysis of molecular markers, morphological / phenological traits and karyotype parameters confirm that Popcorn and Floury populations from NEA are clearly different genetic groups.

Relationship with South American Continent landraces
In order to contrast the Horowitz hypothesis of two different centers on diversification of Argentinian landraces, NEA and NWA [2], our results were compared with previous cytogenetic studies from NWA landraces [5,6,13,18]. Important karyotype differences were observed between maize from both regions, the maximum number of knobs in NEA (from 8 to 23 knobs) was higher than that reported for NWA landraces (from 5 to 19 knobs). There were also differences at the most frequent knob chromosome positions, as NEA had a higher number of knobs than those reported in NWA landraces, where only 6-Sat and 9S were founded as the most frequent positions [6]. In our study we detected, in low frequency, chromosomes with AB10 morphology in six NEA landraces cultivated from 98 to 600 m.a.s.l., but these chromosomes were not described in NWA that grown up to 3900 m.a.s.l [5,6]. These results are in agreement with the reported negative correlation between AB10 presence and altitude of cultivation associated to the selection for faster growth and against larger genome size at higher altitudes [21,36]. Moreover, Bs has not been detected in the Guaraní populations, but in NWA landraces numerical polymorphism (0-8) and frequencies up to 100% have been reported [5,6,13]. It could be concluded that these results allow discrimination between the NEA and NWA maize landraces, supporting the Horowitz hypothesis.
McClintock et al. 1981 [12], based on the knobs karyotype constitution from American races, proposed that maize from Northern South America extended to Eastern South America, while the races found in the Central Andean highlands extended along Western South America. In addition, these authors observed an almost total absence of B chromosomes for the eastern region of South America, and high frequency of Bs in the western region of South America. This hypothesis has been later supported by microsatellite analysis [4,[37][38][39]. Therefore, the karyotype differences between the NWA and the NEA landraces reported here support the hypothesis that the Southern South American maize would have been introduced from two routes, a highland route along the Andes and a lowland route along the Northeastern coast.
The relationship between NEA landraces and those of the lowlands in the American Continent was determined by analysis of molecular markers, revealing the existence of three different genetic groups: Tropical lowland maize, Floury maize, and Popcorn maize from NEA [4]. Also, it was proposed that Guaraní maize would be related to those of the Central Region described by McClintock et al. 1981 [12]. The results obtained here were compared with the maize from the Central Region. In this region 12 positions of knobs (1S, 2L, 3L, 4L, 5L, 6L 2 , 6L 3 , 7L, 8L 1 , 8L 2 , 9S and 9L) were reported, while in the Guaraní landraces the majority of those positions plus1L, 2S, 3S, 4S, 5S, 6-Sat, 7S, 8S and 10L were detected. This greater number of knob positions that we found in Guaraní maize could be due to the higher resolution cytogenetic techniques (DAPI banding and FISH) applied in the present work. In NEA and Central Region the B chromosomes were absent and AB10 chromosomes were reported with low frequency. These cytogenetic similarities allow us to infer a correspondence between maize from both regions. This fact, together with the significant differences founded between the NEA and NWA landraces support the hypothesis that maize was introduced in two ways in South America.
In summary, the joint analysis of karyotype parameters allowed us the first thorough karyotypic characterization of Guaraní maize from NEA and to identify two different karyotype groups, the Flourys and Popcorns. Furthermore, this work highlights the remarkable karyotype differentiation with NWA maize. The cytogenetic characterization of landraces from NEA contributes to the knowledge of the genetic variability of the Argentinian native maize and postulates this germoplasm as genetic resource to improve important agronomic traits.