Deciphering the Mechanism of Glyphosate Resistance in Amaranthus palmeri by Cytogenomics

Amaranthus palmeri (AP) is an annual herbaceous plant, which is a troublesome and economically damaging agronomic weed in the agricultural production system of the USA. Chemical control of AP offers a cost-effective control, however, overuse of and sole reliance on glyphosate for AP control has resulted in the evolution of glyphosate resistance (GR) in AP [Duke and Powles, 2008]. Modelling predicts that 5 applications of glyphosate each year with no other herbicides would result in resistance evolving in 74% of the simulated AP populations [Neve et al., 2011]. The primary mechanism of GR in AP is the amplification of the 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) gene [Gaines et al., 2010]. A similar GR mechanism was reported in Italian ryegrass [Salas et al., 2012], Kochia [Jugulam et al., 2014; Wiersma et al., 2015], waterhemp [Lorentz et al., 2014; Chatham et al., 2015], goose grass [Chen et al., 2015], ripgut brome [Malone et al., 2016], and windmill grass [Ngo et al., 2018]. FISH, a molecular cytogenetic method, was used to study the mechanism of EPSPS gene amplification in AP [Gaines et al., 2010], kochia [Jugulam et al., 2014], and waterhemp [Dillon et al., 2017]. With the advances in sequencing technologies and the availability of robust bioinformatics platforms, cytogenetic approaches combined with genomic approaches have given rise to cytogenomics, uniting ideas from classical, new cytogenetic, molecular-genetic/genomic and bioinformatic approaches under one roof [Liehr 2021]. Cytogenomics is a very powerful approach to address GR in AP [Gaines et al., 2010; Molin et al., 2017; Koo et al., 2018]. The first report that amplified copies of the EPSPS gene in GR in AP are present in the form of extrachromosomal circular DNA molecules (eccDNAs) conferring GR was given by Koo et al. [2018].

The concept that DNA might be organized as series of rings in the chromosomes of higher organisms was put forth by Franklin Stahl [Cairns, 1963] and experimentally proved by Hotta and Bassel [1965]. They referred to the circles of DNA of various sizes in mammalian cells as Double Minutes (DMs). In eukaryotes, eccDNA size ranges from <100 bases up to several hundred thousand base pairs. Based on the different sizes and sequences of eccDNA, Liao et al. [2020] categorized them into small polydispersed DNA (spcDNA; 100 bp–10 kb), telomeric circles (t-circles; multiples of 738 bp), microDNA (100–400 bp), and extrachromosomal DNA (ecDNA; the largest one with millions of bp). Based on the cell type, tissue type, and genetic background, these DNA molecules show heterogeneity in size and copy number. eccDNAs can replicate and propagate in cells as stable genetic elements [Shoura et al., 2017; Turner et al., 2017]. eccDNA may originate from nuclear DNA by homologous recombination between adjacent repeats, such as amplified genes [Mukherjee and Storici, 2012] or tandem repeats (satellite, telomeric, centromeric, and ribosomal repeats) [Cohen et al., 2008; Diaz-Lara et al., 2016] or they can result from linear extrachromosomal forms of active transposable elements [Gaubatz 1990]. In plants, eccDNA derived from satellite repeats are common [Navratilova et al., 2008]. In eukaryotes, eccDNAs contain not only complete or partial genes but also intergenic sequences [Møller et al., 2018; Yerlici et al., 2019], playing a crucial role in genome evolution in response to selection pressure [Gresham et al., 2010]. eccDNAs can serve as templates for the reinsertion of amplified genes into chromosomes in yeast [Demeke et al., 2015]. In this review, a case study of eccDNA-mediated GR in AP is discussed.

Ten years after the introduction of GR crops in the USA, the first case of GR in an AP population was reported in Georgia [Culpepper et al., 2006; Molin et al., 2017]. Various molecular mechanisms conferring GR in AP were reported [Gaines et al., 2011]. The GR AP population from Macon County, Georgia, was used to investigate the mechanism of resistance. The GR AP population contained 5- to >160-fold copies of the EPSPS gene than the glyphosate-susceptible (GS) population [Gaines et al., 2010]. Such a huge copy number increase contributed to the increase in genome size of GR AP [Molin et al., 2017]. A GS AP population from Iowa had a genome size of 0.82 pg/2C, whereas GR AP from Mississippi had various genome sizes depending on the number of copies of the EPSPS gene. Plants having 77, 79, 92, and 106 copies of the EPSPS gene had genome sizes of 0.88 pg/2C (6.8% genome increase), 0.85 pg/2C (3.5% genome increase), 0.87 pg/2C (5.75% genome increase), and 0.93 pg/2C (11.8% genome increase), respectively [Molin et al., 2017]. FISH analysis using the EPSPS gene as a probe revealed that amplified EPSPS gene copies were dispersed in every chromosome throughout the genome, suggesting that EPSPS gene amplification in GR AP was not caused by unequal crossing over or rolling circle replication-based mechanisms. Gaines et al. [2010] speculated that EPSPS amplification in GR AP could have originated via transposon-mediated elements that were activated during stress. He put forward the hypothesis that “the original EPSPS locus was associated with a mobile genetic element that activated and amplified the EPSPS gene”. Following this hypothesis, Gaines et al. [2013] constructed a fosmid library and generated a ∼30-kb sequence harboring 10 kb of the EPSPS gene flanked by a 1.5-kb upstream and 20-kb downstream region. Analysis of the genomic regions around the EPSPS gene revealed that it is bordered by miniature, inverted-repeat transposable elements (MITEs) and a putative Activator (Ac) transposase [Gaines et al., 2013].

To explore more genomic regions around the EPSPS gene, Molin et al. [2017] constructed a BAC tiling path covering the EPSPS locus and flanking regions in GR AP, which was sequenced using the PacBio platform. The de novo assembly of the sequence produced a final sequence length of 297,445 bp. This EPSPS cassette contained a single copy of the EPSPS gene flanked by nearly equal-sized upstream and downstream regions. Around 300 repeat sequences were identified in the EPSPS cassette, predominately simple repeats, followed by Class I and II transposons. Helitrons, a novel class of repeats associated with genome reshuffling and gene amplification, were detected downstream of the EPSPS gene [Molin et al., 2017]. Based on the genomic characterization of the upstream and downstream regions of EPSPS in AP, the structural genomic region of the EPSPS locus is amplified and distributed among all chromosomes mediated by transposons in GR AP, which was further supported by FISH analysis [Gaines et al., 2010]. However, the notion that the EPSPS cassette is integrated randomly into chromosomes through the action of transposons became untenable when fiber-FISH conclusively demonstrated that it is a distinct entity organized as an extrachromosomal circular DNA that mediated the amplification of the EPSPS gene in GR AP [Koo et al., 2018].

EccDNA containing the EPSPS gene in GR AP is termed EPSPS-eccDNA and is ca. 400 kb in size [Koo et al., 2018; Molin et al., 2020a]. Predominately, plant eccDNAs contain tandem repeats, which supports the concept of the origin of EPSPS-eccDNA by intrachromosomal homologous recombination. In natural selection, the mechanism for controlling the length of a nucleotide string is favored by the formation of tandem-repetitive structures rather than the sequence itself [Stephan and Cho, 1994]. Formation of eccDNA from tandem repeats was reported in Drosophila [Pont et al., 1987; Cohen et al., 2003], Xenopus [Cohen and Mechali, 2001], Arabidopsis [Navratilova et al., 2008; Wang et al., 2021], Oryza, Pisum, Secale, Triticum, and Vicia [Navratilova et al., 2008]. Hence, the predominance of complex arrays of repeats and transposon sequences [Molin et al., 2017] in EPSPS-eccDNA likely plays a role in controlling the size and dynamics of EPSPS-eccDNA in GR AP. Sequences containing 143 (AAAT)₂ or 186 (AATT)₂ motifs flank the EPSPS gene in EPSPS-eccDNA. In addition, EPSPS-eccDNA contains clustered long and short interspersed palindromic repeat sequences (CLiSPrs) flanking the EPSPS gene. These repeats may play a crucial role in the stability and dynamics of EPSPS-eccDNA [Molin et al., 2020a]. The EPSPS-eccDNA also have long terminal repeat (LTR) retroposons and non-LTR retroposons, which amplify by a copy and paste mechanism. Generally, LTRs carry a response element that directs the onset and termination of retrotransposon replication. Plant LTR promoters are activated by a variety of biotic and abiotic stresses [Ansari et al., 2007; Salazar et al., 2007]. In the life cycle of an LTR retroposon, the formation of eccDNA is a byproduct, which is an indicator for their continuing activity [Lanciano et al., 2017]. EccDNA is produced through heat-responsive, copia-like ONSEN retrotransposons in Arabidopsis and drug-induced activation of the Houba retrotransposons in Oryza sativa [Thieme et al., 2017]. The presence of repeats and transposons upstream and downstream of the EPSPS locus may facilitate the formation of EPSPS-eccDNA as a response to natural selection, conferring GR in AP.

The first cytological evidence of EPSPS-eccDNA tethering to chromosomes mediating their inheritance to the daughter cells during cell division in plants was reported by Koo et al. [2018]. Typically, EPSPS- eccDNAs are randomly transmitted to daughter cells due to the lack of centromeres. The EPSPS-eccDNAs are transmitted to the next generation by tethering to meiotic chromosomes leading to a rapid spread and development of GR [Koo et al., 2018; Fig. 1]. The EPSPS-eccDNAs are not integrated back into the chromosome and exist as autonomous replicating structures that display unequal mitotic segregation and, thereby, produce soma cell heterogeneity for evolution of GR [Koo et al., 2018]. The sequencing and annotation of EPSPS-eccDNA revealed genetic elements that are responsible for EPSPS-eccDNA stability, replication, and tethering. Molin et al. [2020a] predicted that AP_R.00g000496 encodes a protein containing a helicase domain, which is involved in DNA replication and tethering of EPSPS-eccDNA to the nuclear chromatin. Molin et al. [2020b] described EPSPS-eccDNA as an autonomously replicating genetic element possessing autonomous replication sequences (ARS) and having homology with the ARS sequence found in yeast and other eukaryotes. The ARS sequence in AP contains 2 putative DNA unwinding elements rich in A + T content and displaying helical instability that can result in a unique DNA structure that can facilitate accessibility to the DNA replisome. The origin of replication (ori) on the eccDNA replicon occurs in a putative NAC-containing gene. The ARS sequence of the eccDNA replicon from AP enabled the autonomous replication in ARS-less yeast plasmid.

Fiber FISH analysis using 6 BACs representing the 399-kb EPSPS cassette [Koo et al., 2018] revealed the existence of structural polymorphisms in EPSPS-eccDNA. Predominately, EPSPS-eccDNA exist in circular (50.2%), followed by linear (21.8%), dimeric circular (11.8%), dimeric linear (8%), and atypical (8.2%) forms. The circularity of the EPSPS-eccDNA accounts for the stability and integrity of EPSPS-eccDNA, where the closed circular structure of eccDNA may impart resistance to exonuclease digestion. The purpose of structural polymorphism in EPSPS-eccDNA is not well understood. However, characterizing EPSPS-eccDNA structural polymorphisms may give clues regarding the pathway that gives rise to EPSPS-eccDNA. The copy number of EPSPS-eccDNA varies between the somatic cells, generating somatic mosaicism patterns. Using 2-color FISH with 5S rDNA and EPSPS-eccDNA as probes, 4 different patterns of EPSPS-eccDNA signals on 5S rDNA-labeled homologous chromosome pairs in different cells derived from a single root tip meristem were observed: (1) both chromosomes lack EPSPS-eccDNA signals (16.7%), (2) 1 of the 2 chromosomes lacks an EPSPS-eccDNA signal (25%), (3) both chromosomes have a similar signal intensity (33.3%), and (4) the 2 chromosomes vary in signal intensity (25%) [Koo et al., 2018]. We hypothesize that EPSPS-eccDNA is not associated with chromosomes during the interphase stage. EPSPS-eccDNAs in metaphase tend to randomly anchor to chromosomes and are associated with the chromosomes until the end of telophase, enabling their transmission to the daughter cells leading to somatic mosaicism (Fig. 2). EPSPS-eccDNAs also are inherited by tethering to meiotic chromosomes and confer copy number variation in somatic cells of the progeny. From metaphase cell spreads from a single root preparation of one plant, we observed that (1) EPSPS-eccDNAs were associated with most of the chromosomes similar to the GR AP, (2) EPSPS-eccDNAs were associated with half of the chromosomes, (3) EPSPS-eccDNAs were associated with only a few chromosomes, and (4) all chromosomes were free of EPSPS-eccDNAs. Similarly, a huge variation in EPSPS-eccDNA distribution in the cells was observed when analyzing the F₁ progeny derived from the cross between GS AP and GR AP [see Fig. 5 shown in Koo et al., 2018]. This copy number variation is due to the independent existence of EPSPS-eccDNA and its unequal segregation during mitotic divisions.

Genes that occur in eccDNA amplify much faster than those in chromosomes. Turner et al. [2017] speculated that the genes carried by eccDNA can be easily accessed by the transcription machinery compared with chromosomal genes. Wu et al. [2019] further demonstrated that eccDNA was packaged into chromatin with an intact domain structure but lacked the higher-order compaction that is typical of chromosomes and displayed significantly enhanced chromatin accessibility enabling high transcription of genes. Similarly, preliminary investigation of eccDNA from AP observed that EPSPS-eccDNAs, highly condensed in mitotic and meiotic metaphase I chromosomes, are barely visible by DAPI staining [see Fig. 1 shown in Koo et al., 2018]. In meiotic pachytene stage, where chromosomes are decondensed and highly elongated, EPSPS-eccDNA appear as large dots [see Fig. 3 shown in Koo et al., 2018]. This correlation of chromosome condensation progression from meiotic pachytene through metaphase stages suggests that the EPSPS-eccDNA of AP may consist of basic nucleosome units organized into chromatin, rather than as naked DNA, and needs further investigation. Characterizing the histones present in EPSPS-eccDNAs will shed light on the structure and organization of EPSPS-eccDNA.

The introduction of GR crops favored the agriculture production system for a short period. Extensive use of glyphosate for the control of weeds gave rise to GR weeds. Different mechanisms operate that confer GR in plants. A case study with GR AP demonstrated how cytogenomics can unravel the mystery of evolution of GR. Genomic characterization of a 399-kb EPSPS cassette derived from BACs identified the complex array of repeats and the transposons flanking the EPSPS locus. Cytological investigation with BACs representing the EPSPS cassette identified that the EPSPS in GR AP exist as eccDNA, in addition to the EPSPS gene at the native locus of the chromosome. eccDNA containing the EPSPS gene in AP is termed EPSPS-eccDNA and exists as an autonomously replicating genetic element able to transmit during cell division by tethering to the mitotic chromosome. EPSPS-eccDNA is inherited to the progeny by tethering to meiotic chromosomes during gamete formation. This EPSPS-eccDNA is highly heterogeneous, exhibiting both copy number variation and somatic mosaicism. This copy number variation is attributed to unequal EPSPS-eccDNA segregation during mitotic divisions. The role of EPSPS-eccDNA in conferring genome plasticity to glyphosate challenge is discussed.

We thank W. John Raupp for critical review of the manuscript and Duane Wilson for technical assistance.

The authors report no competing interests.

This research was supported by grants from the Kansas Wheat Commission and the Kansas Crop Improvement Association, Wheat Genetics Resource Center Industry/University Cooperative Research Center National Science Foundation Contract 1338897. This is contribution number 22-050-J from the Kansas Agricultural Experiment Station, Kansas State University, Manhattan, KS 66506-5502, USA.

D.-H.K., R.S., B.F., and B.S.G. wrote the manuscript and analyzed the data and helped to draft the final manuscript.