Cytological study on haploid male fertility in maize

YANG Ji-wei ,LlU Zong-hua ,QU Yan-zhi ,ZHANG Ya-zhou ,Ll Hao-chuan

1 College of Agronomy,Henan Agricultural University/State Key Laboratory of Wheat and Maize Crop Science/Henan Grain Crop Collaborative Innovation Center,Zhengzhou 450046,P.R.China

2 Institute of Cereal Crops,Henan Academy of Agricultural Sciences,Zhengzhou 450002,P.R.China

Abstract Doubled haploid (DH) breeding technology,which relies on haploid genome doubling,is widely used in commercial maize breeding. Spontaneous haploid genome doubling (SHGD),a more simplified and straightforward method,is gaining popularity among maize breeders. However,the cytological mechanism of SHGD remains unclear. This study crossed inbred lines RL36 and RL7,which have differing SHGD abilities,with inducer line YHI-1 to obtain haploid kernels. The meiotic processes of pollen mother cells (PMCs) in the haploid plants were compared with diploid controls. The results suggested that three main pathways,the early doubling of haploid PMCs,the first meiotic metaphase chromosomal segregation distortion,and anomaly of the second meiosis,are responsible for SHGD. Furthermore,flow cytometry analysis of ploidy levels in leaves and PMCs from haploids and diploid controls revealed that somatic cell chromosome doubling and germ cell chromosome doubling are independent processes. These findings provide a foundation for further studies on the underlying mechanism of SHGD,aiding the application of DH technology in maize breeding practices.

Keywords: maize,haploid,male fertility,spontaneous genome doubling,meiosis

1.lntroduction

Maize is one of the most successful crops utilizing heterosis,and it is widely used for multiple purposes.Breeding the superior homozygous inbred lines is the basis of new hybrids. Development of homozygous lines usually requires 8-10 generations using traditional breeding methods;in other words,4-5 years are required,even with two generations generated per year (Hallaueret al.2010). However,in doubled haploid (DH) breeding technology,only two generations are required to obtain a completely homozygous line. Therefore,DH breeding technology can significantly shorten the breeding period and improve breeding efficiency,and it has been widely used in maize and other crop species such as wheat,rice,and barley (Germanàet al.2011;Dwivediet al.2015).

Haploid plants possess only one set of chromosomes from either male or female gametes of diploid parents.As a result,there are no homologous chromosomes in haploid plants,and homologous chromosome synapsis does not occur during the process of meiosis,usually resulting in sterile gametes. In maize,only if all 10 univalent chromosomes move into one cell during anaphase I at the first meiosis stage can regular gametes be produced;however,the probability of this is very low at about 1/512. DH production requires both male and female parts to produce fertile gametes. Previous studies have reported a significant difference between male and female fertility of maize haploids,with a high fertility rate in females at over 90%,allowing haploid plants to produce a seed set after pollination with viable pollen (Chalyk 1994;Geigeret al.2006). However,haploid male fertility (HMF)is relatively rare (Zabirova 1993;Geiger and Sch?nleben 2011;Wuet al.2014) and,as a result,is considered a limiting factor and the most critical step in DH production(Kleiberet al.2012;Renet al.2017;Wuet al.2017).Haploid plants thus tend to be sterile and have difficulty producing normal fertile gametes. Therefore,haploid genome doubling is an important bottleneck to application in DH breeding. At present,the main genome doubling method of haploid is the utilization of chemical agents such as cochicine (Prasannaet al.2012;Weber 2014),APM (Beaumont and Widholm 1993),oryzalin (H?ntzschel and Weber 2010),and trifluralin (Wanet al.1991).However,this method requires transplantation of the treated haploid plantlets into the field,requiring a large amount of manpower and financial resources,not to mention the harm caused to the environment (Chaikam and Mahuku 2012;Melchingeret al.2016). Although nitrous oxide gas can replace toxic chemical reagents,it can also be used to double the chromosomes of maize haploids,which requires large equipment and is impractical (Kato and Geiger 2002;Molenaaret al.2018).In contrast,spontaneous haploid genome doubling(SHGD) is straightforward and efficient and has therefore become a promising technique of haploid doubling.

The genetic basis of HMF has been revealed. Four QTLs,qhmf1,qhmf2,qhmf3,andqhmf4,were identified using a haploid population derived from Yu87-1/Zheng58,and these QTLs were subsequently located on chromosomes 1,3,4,and 6,respectively (Renet al.2017). Maet al.(2018) used a mapping population from a diversity panel of 431 inbred lines crossed with inbred lines Mo17 and Zheng58 to reveal 14 significant SNPs related to SHGD in genome-wide association analysis.Nine additional QTL responsible for SHGD were also detected,distributed on chromosomes 1,3,4,7,and 8 (Yanget al.2019). Furthermore,Renet al.(2020)identified three QTLs,qshgd1,qshgd2,andqshgd3,using selective genotyping. A major QTL for SHGD was identified on chromosome 5,and the introgression of this QTL can replace the need for colchicine in DH lines development (Trampeet al.2020). These findings have provided a basis for the analysis of the molecular mechanism of SHGD. However,in maize,the cytological mechanism of SHGD leading to haploid male fertility remains unknown. To this end,two inbred lines,RL36 with a high rate of SHGD and RL7 with a low rate,were selected as experimental materials for the analysis of the meiosis of haploid pollen mother cells (PMCs) and ploidy of anthers at different developmental stages and in mature leaves. The objectives were to obtain cytological evidence of SHGD restoration.

2.Materials and methods

2.1.Plant materials

Two inbred lines,RL36 and RL7,were selected based on their HMF ability in multiple environments (Appendix A). The tassels of haploids from RL36 show very high haploid male fertility (97.59%),while those from RL7 show poor haploid male fertility (0%). Both inbred lines were obtained from a cross between inbred lines K22 (developed by Northwest Agriculture and Forestry University,China,has 56.4% HMF rate) and Zheng58 (developed by Henan Academy of Agricultural Science,China,has 5.8% HMF rate) (Yanget al.2019). A large number of plants of two inbred lines as females were crossed with the male inducer line YHI-1 and putative haploids were identified using theR1-njcolor marker (Nanda and Chase 1966).In summer 2017,putative haploid kernels of RL36 and RL7 were grown in the field at Zhengzhou Experimental Station,Henan Agricultural University (34°48′N,113°42′E),Zhengzhou,China. Haploids from each line were sown in a plot consisting of 10 rows,with 25 plants per row.The misclassified hybrid plants were removed based on superior plant vigor at the six-leaf stage,which ensured the true haploid plants to be used for this experiment materials.

2.2.Observations of pollen viability and anther morphology

Anthers were collected from glumes at the trumpet stage. Each male floret contained three anthers,one of which was placed on a clean slide with 2-3 drops of acetocarmine solution to examine meiosis using a light microscope. The stages of anther development were determined as described by Rosset al.(1966).The remaining two anthers were placed in a fixing solution consisting of 5 mL of 38% formaldehyde,5 mL of glacial acetic acid,and 90 mL of 70% alcohol (FAA)for morphological and cytological observations. At the flowering stage,pollen was placed on a clean glass slide and stained with 1% I2-KI (potassium iodide) for observations of pollen viability under a microscope.

2.3.Flow cytometry ploidy analysis

The anthers were divided into different groups according to the stage of meiosis,and then 400 μL of cell lysis buffer aid (provided by Sysmex Corporation,Japan)was added before using a sterilized treated pipette to crush the anther cells. The DNA was stained for 10-20 min using an equal volume of staining buffer,and any remaining anther wall was removed by sieving through a 100-mesh sieve. The resulting substrate was transferred to a dedicated ploidy tube and then diluted with deionized water. Analysis of cell ploidy at different stages was carried out using a diploid anther as a control. Ploidy analysis of the leaves was also carried out,as described by Palominoet al.(2008). Ploidy of the PMCs was carried out by flow cytometry using the CyFlow Cube8 Cytometer (Sysmex Corporation Kobe,Japan) to determine the DNA content.

2.4.Cytological observations of PMCs

Anthers at different stages of meiosis were carefully removed from the FAA solution,placed on a clean glass slide using tweezers,and then washed three times in deionized water. Next,5-10 μL of 1 μg μL-14′,6-diamidino-2-phenylindole (DAPI) was added to dye the anthers. The anther wall was carefully broken using two anatomical needles to release the PMCs,and the broken anther wall was removed. After 3-5 min,the samples were covered with a cover slide,and PMC chromosome behavior during meiosis was observed under a fluorescent microscope.

3.Results

3.1.Characteristics of HMF for RL36 and RL7

The morphological characteristics from haploid plants RL36 and RL7 before the anthesis stage were observed (Fig.1-A).At the flowering stage,some of the anthers in the haploid RL36 plants were extruded from the glumes and appeared plump (Fig.1-B and D). Meanwhile,in RL7,the haploid tassels were all shriveled,some spikelets appeared wilted,and no anthers emerged (Fig.1-C and E). At the flourishing flowering stage,tassel samples were randomly selected from haploid individuals of each line for I2-KI staining. More than 80% of pollen grains from the extruded anthers of RL36 were darkly stained under I2-KI (Fig.1-F),indicating viable pollen. Anthers that were not extruded also showed some staining indicative of viable pollens (Fig.1-H);however,the tassels from RL7 were completely sterile,indicating no fertile pollen production (Fig.1-G and I).

3.2.Dynamic changes in anther morphology at different stages

Furthermore,according to the correlation of anther length and cytological stages (Kelliher and Walbot 2011,2014;Nelms and Walbot 2019),the anther morphology of RL36 and RL7 were subsequently compared. Haploid anthers of RL36 and RL7 became gradually elongated from prophase to the mature anther stage (Fig.2-AD). The RL36 anthers reached the maximum size and were at their fullest at the mature anther stage (Fig.2-D).Meanwhile,at prophase,there were no significant differences in anther size between RL36 and RL7.Overall,the growth rate of the RL7 anthers was slower than that of the RL36 anthers until the mature anther stage,when a significant difference in size was observed,and RL7 appeared shriveled (Fig.2-B-D). These results demonstrated that anther development differs between haploid RL36 and RL7,which could be related to PMC development.

3.3.Ploidy analysis in anther cells

In order to analyze the ploidy changes in haploid PMCs,anthers and leaf tissues were sampled from RL7 diploid plants as a control. Comparisons of the ploidy level of leaf tissue from diploid and haploid plants subsequently revealed that cells from both RL36 and RL7 were haploid,with no diploid cells detected,while diploid plants showed only diploid cells (Fig.3-A,E,and I). A diploid peak (200)and a tetraploid peak (400) were observed,and in diploid cells the DNA has replicated in the interphase when they entered the first meiosis (Fig.3-B). Two daughter cells were formed during the dyad stage,and the number of chromosomes decreased by 50%. In each daughter cell,each chromosome contains two sister chromatids,suggesting no haploid cells (Fig.3-C). Meanwhile,at anaphase,separation of the two sister chromatids was observed,resulting in four daughter cells,each containing only half of the mother cell genome. At this time,the number of tetraploid cells gradually decreased,while the haploid cells began to increase,causing a significant increase in the haploid peak and a reduction in the tetraploid peak (Fig.3-D). Throughout the process of meiosis,the PMCs of RL36 contained diploid as well as haploid cells (Fig.3-F,G,and H),which indicated that the genome of some haploid cells had been doubled. In addition,a few tetraploid cells were found (Fig.3-F,G,and H);it intimated that diploid cells also exist at the early stage of PMC development in RL36 haploid anthers,potentially proceeding through normal meiosis to produce normal microspores. However,the PMCs of RL7 tended to show haploid at all stages of meiosis,accompanied by mixoploid cells between the haploid peak (100) and diploid peak (200) (Fig.3-J,K,and L),suggesting the absence of normal meiosis. Moreover,according to these results,no relationship between ploidy of somatic cells and spontaneous doubling of generative cells was determined.

3.4.Cytological observations of meiosis in diploid PMCs

To further examine chromosome morphological behavior in the PMCs of haploid plants during meiosis,morphological observations of PMCs from normal diploids were carried out using DAPI staining. At leptotene (Fig.4-A),the chromosomes of the diploid mother cells appeared as abundant tenuous chromosome threads,coiled into clumps,while at zygotene,homologous chromosomes with a similar morphological structure began to converge(Fig.4-B) and prepare for pairing. On entering diakinesis (Fig.4-C),the chromosomes became condensed and scattered around the cell,accompanied by the crossover of homologous chromos-omes,with “O”or “V” shaped bivalents appearing as a result of the pairing of homologous chromosomes and recombination between non-sister chromatids. At metaphase I,all bivalents became neatly arranged and aligned along the equatorial plate,with the centromeres of each bivalent positioned on both sides of the equatorial plane(Fig.4-D). The hom-ologous chromosomes then moved simultaneously to opposite poles (Fig.4-E),each pole containing half of the mother cell chromosomes,and each chromosome containing two sister chromatids phys-ically connected to each other at the centromere and remaining joined. At telophase during first meiosis (Fig.4-F),a cell plate began to form at the equator,and the cytoplasm divided into two parts to form two daughter cells (dyad),each containing half of the mother cell chromosomes.Meanwhile,the chromosomes started to unwind,becoming loose and slender. During the second meiotic division,the chromosomes re-condensed andthickened. At metaphase II,all chromosomes became neatly arranged at the equatorial site of each daughter cell,and the pairs of sister chromatids separated due to the pulling of the spindle fiber at anaphase (Fig.4-G).Four microspores formed at telophase II (Fig.4-H),and the chromosomes unwound and became entangled in clumps.

3.5.Cytological observations of meiosis in RL7 haploid PMCs

Chromosomes of RL7 haploid PMCs appeared as tenuous chromosome threads,coiled in clumps at leptotene (Fig.5-A),while at diplotene (Fig.5-B),they became shorter and thicker,with no crossover,making all 10 chromosomes clearly visible. At diakinesis (Fig.5-C),the chromosomes became condensed to their shortest length and were easily distinguishable within the cells.There were no “O” or “V” shaped bivalents,suggesting a lack of homologous chromosomes. At metaphase I(Fig.5-D),10 univalents were clearly distinguishable,randomly distributed around the center of the cell rather than aligned on both sides of the equatorial plate. At anaphase I,the 10 univalents moved randomly to the two poles and gathered together,although some failed to move synchronously and showed a lag phenomenon(Fig.5-E). At telophase I (Fig.5-F),the chromosomes became slender and once again intertwined,presenting as a clump,with obvious lagging of some chromosomes in the cell. The equatorial plate became a cell plate dividing the cytoplasm into two daughter cells,each containing only part of the genome. At this time,the chromosomes became clustered together in the daughter cells,with the number of chromosomes randomly assigned between each cell. At metaphase II,the chromosomes failed to distribute to the central equatorial plane and instead gathered in central or bipolar regions of the cell. Meanwhile,at anaphase II,the univalents began to separate (Fig.5-G),with obvious lagging of some chromosomes still visible. Eventually,at telophase II,some unbalanced tetrads were produced with cytoplasm division (Fig.5-H),each containing only some of the chromosomes. Overall,the PMCs of RL7 haploid plants did not undergo normal meiosis,and as a result,chromosomes were not evenly distributed in each daughter cell at telophase II. These aneuploid microspore cells were not able to produce fertile male gametes,eventually leading to haploid male sterility.

3.6.Cytological observations of meiosis in RL36 haploid PMCs

The meiosis process for the PMCs from RL36 haploid plants was observed. At leptotene,during prophase I(Fig.6-A),the chromosomes appeared as tenuous chromosomal threads,coiled into clusters,and there were no obvious morphological differences compared with the PMCs from RL7. At the diakinesis stage(Fig.6-B),the chromosomes became condensed together and distributed in the cells,similar to diploid haploid chromosome behavior. Meanwhile,“O” or “V”shaped bivalents were observed in the PMCs of RL36 haploid (Fig.6-C) at a frequency of 3.2% (282 anther sacs were observed,of which the cells of nine anther sacs showed normal diploid cell characteristics). At metaphase I,the chromosomes in 57.7% of cells(1 732/3 002 cells,not including early doubled cells)were positioned around the equatorial plate,consistent with RL7 haploid PMCs at metaphase I (Fig.5-D).Meanwhile,at anaphase I,the chromosomes moved randomly to the two poles,with some univalents presenting a lagging phenomenon (Fig.6-D). At metaphase I,in 42.3% of cells (1 270/3 002 cells,not including early doubled cells),the chromosomes existed at only one pole (Fig.6-E),while at anaphase I,they did not move and remained at each pole,forming clumps and clustering together (Fig.6-F). In other words,the first meiotic division failed. At telophase I during meiosis I,a few cells produced one large and one small daughter cell(Fig.6-G). In this case,although the first meiotic division was abnormal,one of the daughter cells contained the entire genome and was able to continue through a normal second meiotic division to produce normal fertile male gametes. In addition,triads were observed at telophase II (Fig.6-H),and a special type of tetrad microspore formed as a result of abnormal meiosis II.These results suggest that the PMCs of RL36 haploid plants can undergo multiple types of abnormal meiosis,resulting in fertile gametes that differ from those of RL7 haploids and diploids.

4.Discussion

4.1.Spontaneous genome doubling in maize haploids

At present,SHGD is the most economical and straightforward method of haploid genome doubling in maize haploid breeding technology,and numerous studies have documented its genetic characteristics and genetic control (Kleiberet al.2012;Wuet al.2017). However,these tend to focus on the result of genetic studies,with little known about the cytological mechanisms of SHGD. Therefore,this study used two inbred lines,RL36 with a high rate of SHGD and RL7 with a low rate,to examine the characteristics of plant morphology,male fertility,and meiosis of haploid PMCs at different developmental stages. Comparisons of plant morphology,anther fertility,and anther development were also performed. As a result,the RL7 haploids were found to be completely sterile with shriveled anthers,while the RL36 haploids were confirmed as having high SHGD with numerous exposed plump anthers. The results of ploidy analysis of PMCs further revealed that diploidization also occurred in the RL36 haploid plants,viameiosis as in diploid plants. However,in contrast,no diploidization of the haploid PMCs was observed in RL7. In addition,according to ploidy analysis of haploid leaves from RL36,RL7,and normal diploids,no direct relationship was revealed between ploidy of somatic cells and spontaneous doubling of generative cells. This was further confirmed by cytological observations of meiosis of RL36 haploid PMCs,with chromosomal segregation distortion at metaphase I and the anomaly at telophase II.This abnormal meiosis process is thought to result in the fertility of haploid males.

4.2.Early SHGD

Early haploid genome doubling suggests that some PMCs undergo chromosome doubling during premeiosis,producing diploid PMCs before entering meiosis. These doubled PMCs can perform normal meiosis and eventually form normal gametes. However,Wuet al.(2014) reported that spontaneously doubled haploids occurred at a rate of only 1-3.5% during haploid embryo development after induction,not during haploid plant growth. These early doubled embryos were named EH lines;they possessed diploid and normal growth characteristics and completed fertility. In the present study,3.2% of anther sacs contained diploid cells under fluorescence microscopy and appeared to undergo normal meiosis during both the first and second meiotic divisions,resulting in normal gametes. These findings suggest that genome doubling occurs before meiosis,although they appeared haploid in cell ploidy analysis of the leaves,further suggesting that haploid somatic cell doubling and generative cell doubling occur separately. Several suggestions for genome doubling before meiosis have been reported. For example,early cytologists reported that 0.05-2.17% of doubled microspores were attributed to the presence of doubled PMCs in maleBrassicanigra(Heyn 1977). Moreover,Lam (1974) attributed 0.2% of giant microsporocytes from a trisomic ofSolanumchacoenseto premeiotic doubling,presumably resulting in four 2n microspores in the subsequent tetrads. Doubled megaspore productionviapremeiotic endomitotic chromosome duplication has also been observed in someAlliumspecies (Yamashitaet al.2012),while Mashkina (1979) suggested that premeiotic doubling resulted in 2n pollen formation inCerasusauium.Rouisset al.(2017) also confirmed that pre-meiotic doubling could lead to 2n gamete formation in lemon. In addition,inArabidopsis,mutation of DNA duplication before meiosis revealed that about 5% of gametes were double gametes (De Storme and Geelen 2013). In this study,the normal male gametes appeared,possibly because of the specificity of different cells. As a result,the doubled PMCs were formed.

4.3.The chromosome segregation defect of the first meiosis

During cell division,the spindle pulls the homologous chromosomes to opposite poles of the cell,evenly distributing them between the two daughter cells. Normal functioning of the spindle is necessary for this pulling process. In the presence of chromosomal doubling chemicals,binding to the tubulins occurs,preventing the formation of spindle microtubules during the metaphase stage of mitosis and,as a result,the chromosomes are unable to move to opposite poles during anaphase,instead staying in one cell (Prasannaet al.2012).Normal mitosis is subsequently disrupted. Abnormal spindle fibers could therefore result in chromosome doubling during mitosis as well as meiosis. Asymmetrical incomplete cytokinesis occurs due to spindle deformation(bending) during metaphase-anaphase I,and the chromosomes of daughter nuclei combine into a common spindle during the second meiotic division,resulting in doubled microspores at telophase II (Shamina and Shatskaia 2011). In this study,about 42.3% of PMCs in the RL36 haploid presented with chromosomes at the same pole during metaphase I,and these chromosomes did not move during anaphase I,resulting in two normal microspores after the second meiotic division. In Brazilian accessions ofPaspalum,Pagliariniet al.(1999) revealed that meiosis of PMCs is affected by the failure of the first meiotic division but remained normal during the second meiotic division,resulting in two microspores at telophase II. Furthermore,Sugiharaet al.(2013) observed a first division restitution1 (fdr1) mutant in maize,revealing that about 48% of anthers were haploid,producing equally divided dyads,with thefdr1phenotype expressed only in PMCs of haploid not diploid plants. In the present study,failure of the first meiosis division resulted in more doubled cells,possibly because of the formation of spindle fibers from only a unipolar tube,forcing the chromosomes to move into only one pole. In this case,the number of chromosomes did not decrease during the first meiotic division,and the second stage was normal,resulting in the formation of normal microspores.

4.4.Anomaly of the second meiosis

In the second meiosis,the haploid genome would be doubled when the abnormal division produces triads. It was also an important factor in the formation of doubled gametes (Ramanna 1979). Many dyads and triads were found at telophase II of Brazilian accessions ofPaspalum,which led to the doubled gametes formation by abnormal of cytokinesis (Pagliariniet al.1999). Cytological evidence was obtained by Bieliget al.(2003) for dyad and triad formation during microsporogenesis,2n male gametes formation were probably attributable to the abnormal of cytokinesis by the analysis of chromosome behavior at meiosis.Abnormal of cytokinesis was also detected among meiocytes in the second division,and monads,dyads,and triads with n or 2n nuclei were observed after telophase II in most meiocytes (Galloet al.2007).Rouisset al.(2017) confirmed that the second-division abnormal was the main mechanism for 2n gamete production in lemon occursviagenetic analysis with SSR and SNP markers. Lam (1974) considered that the production of triads was due to spindle deformation and may result in cytoplasmic fusion. However,in previous research,most crops were diploid or polyploid,with second-division abnormal resulting in doubled gametes. Although the second-division abnormal is an important mechanism of doubled gametes formation,little is available about second-division abnormal in haploid plants,and to date,few studies have reported the formation of double gametes as a result of seconddivision abnormal in maize haploid plants (Shamina and Shatskaia 2011). In this study,anomalies of cytokinesis were also revealed,resulting in the production of triads during telophase II. These anomalies may cause the production of doubling gametes,achieving the effect of chromosome doubling.

4.5.Model of SHGD

Dynamic comparisons of chromosome behavior were also carried out at different stages of meiosis in the RL36 haploid,which has a high rate of HMF,the RL7 haploid,which has a low rate of HMF,and a normal diploid. We concluded that three pathways during meiosis of RL36 haploid PMCs (i.e.,the early doubling,first meiotic metaphase chromosomal segregation distortion,and anomaly of the second meiosis) are the reason for SHGD and the high rate of HMF in the RL36 haploid (Fig.7).This is the first study to report chromosomal segregation distortion at metaphase I,suggesting that it could be regulated by a specific gene in the high-frequency doubling material RL36. In line with this,the PMCs of RL7 haploid plants did not produce double gametes during abnormal meiosis. However,SHGD is a complex trait,and whether other mechanisms are involved remains to be seen. In conclusion,these findings provide a cytological foundation for further analyses of the mechanism underlying the restoration of haploid fertility in maize.

5.Conclusion

Three main pathways,the early doubling of haploid PMCs,first meiotic metaphase chromosomal segregation distortion,and the anomaly of the second meiosis,could be responsible for SHGD. Furthermore,flow cytometry analysis revealed that somatic cell chromosome doubling and germ cell chromosome doubling were independent processes.

Acknowledgements

This work was supported by the Agricultural Seed Joint Research Project of Henan Province,China(2022010202),the Science and Technology Project of Henan Province,China (222102110276) and the China Postdoctoral Science Foundation (2020M682031).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendixassociated with this paper is available on http://www.ChinaAgriSci.com/V2/En/appendix.htm