Can double PPO mutations exist in the same allele and are such mutants functional?

BACKGROUND: Resistance to protoporphyrinogen oxidase (PPO)-inhibiting herbicides is endowed primarily by target-site mutations at the PPX2 gene that compromise binding of the herbicide to the catalytic domain. In Amaranthus spp. PPX2, the most prevalent target mutations are deletion of the G210 codon, and the R128G and G339A substitutions. These mutations strongly affect the dynamic of the PPO2 binding pocket, resulting in reduced affinity with the ligand. Here we investigated the likelihood of co-occurrence of the most widespread target site mutations in the same PPX2 allele.
RESULTS: Plants carrying R128G+/+ ΔG210+/-, where + indicates presence of the mutation, were crossed with each other. The PPX2 of the offspring was subjected to pyrosequencing and E. coli-based Sanger sequencing to determine mutation frequencies and allele co-occurrence. The data show that R128G ΔG210 can occur in one allele only; the second allele carries only one mutation. Double mutation in both alleles is less likely because of significant loss of enzyme activity. The segregation of offspring populations derived from a cross between heterozygous plants carrying ΔG210 G399A also showed no co-occurrence in the same allele. The offspring exhibited the expected mutation distribution patterns with few exceptions.
CONCLUSIONS: Homozygous double-mutants are not physiologically viable. Double-mutant plants can only exist in a heterozygous state. Alternatively, if two mutations are detected in one plant, each mutation would occur in a separate allele.

protoporphyrinogen oxidase

INTRODUCTION

Protoporphyrinogen IX oxidase (PPO, EC. 1.3.3.4) pertains to a highly conserved family of membrane‐bound enzymes in the tetrapyrrole biosynthetic pathway that is found in mammals, plants, bacteria and fungi. It catalyzes the six‐atom oxidation of protoporphyrinogen IX to protoporphyrin IX, which is the last common step in the production of chlorophylls and heme. In plants, two isoforms encoded by two distinct nuclear genes exist: PPO1 in chloroplasts and PPO2 in the mitochondria. Some plant species can have PPO2 targeted to both organelles.

PPO inhibitors have been used as herbicides for >60 years. Their high efficacy, broad weed control spectrum and residual activity, as well as the evolution of glyphosate‐ and acetolactate synthase (ALS)‐resistant weeds contributed to their large‐scale adoption and intensive use by farmers. Four distinct chemical families of PPO‐inhibiting herbicides have been developed commercially so far: diphenyl ethers, N‐phenyl‐imides, N‐phenyl‐oxadiazolones and N‐phenyl‐triazolinones. Two active ingredients have not been assigned to a chemical family yet, according to the latest HRAC update: pyraclonil and pyraflufen‐ethyl. Much of the variation in herbicide efficacy across PPO inhibitors can be attributed to their chemical structures, as these molecules bind to PPO by resembling its substrate.

Although the selection of weeds resistant to PPO inhibitors was a slow process compared to herbicides with other modes of action, biotypes from 13 species have developed resistance mechanisms to these herbicides so far. The involvement of herbicide metabolism has been reported but not fully characterized, , and a polymorphism in the PPX1 gene responsible for resistance of Eleusine indica to oxadiazon was discovered recently. Nevertheless, mutations in the PPX2 gene have been identified as the main resistance mechanism to PPO inhibitors. Among those, the G210 codon‐deletion (∆G210) was first reported in 2001, and is so far the most prevalent resistance‐conferring mutation among Amaranthus tuberculatus and A. palmeri species. , , In 2019, a Gly‐to‐Ala substitution at the A. palmeri PPX2 399th position (G399A) was reported, and has not been found in other species yet. Interestingly, neither of the native residues (G210 and G399) are directly related to protogen binding to PPO2, , so the resistance caused by mutation at these loci is unexpected. It was later understood that ∆G210 caused an enlargement of the binding pocket compared to the wild‐type (WT) protein, allowing concomitant binding of inhibitor and substrate. Protogen binding to ∆G210 PPO2 was not negatively affected, which is a favorable condition to minimize related fitness costs. , However, the same cannot be said about G399A. The authors concluded that the additional methyl group from Ala in relation to Gly protruded into the binding‐site, creating repulsive interactions towards the inhibitor (and consequently, towards the substrate). This mutation led to a 97% reduction in enzyme activity compared to the WT PPO2. Lastly, mutations at the Arg128 position (homologous to Arg98, in Ambrosia artemisiifolia) has been reported in several species, such as A. tuberculatus, A. palmeri, A. retroflexus, , Am. artemisiifolia and Euphorbia heterophylla. The most common substitution is R128G (four species), whereas R128M, R128I and R128L are present in only one species each (A. palmeri, A. tuberculatus and Am. artemisiifolia, respectively). The importance of this residue for protogen binding relates to the salt bridge formed between the positively charged Arg with the negatively charged carboxyl group of ring C of protogen. , , ,

The continued use of PPO inhibitors, the dioecious character of some Amaranthus species and their hybridization capacity, has led to an accumulation of mutations at the population‐, plant‐ and allele‐levels, , , , although the latter is very rare. Computational data suggested that both combinations ∆G210 + G399A and G399A + R128G would confer high resistance to fomesafen, but the very low frequency of such genotypes, and the absence of double homozygous plants could also indicate that the resultant protein might be inactive. In addition, it was predicted that the combination ∆G210 + R128G would not be viable as a consequence of the loss of 3D integrity of the binding pocket, which was backed up by the absence of such genotype among the fomesafen‐survivors. To assess the herbicide resistance risk arising from double mutations, the aim of this study was (i) to verify which mutation‐combinations would be tolerable in PPO2 from A. palmeri; and (ii) study the inheritance pattern of a double‐heterozygous ∆G210 G399A mutant in A. palmeri, which was the most prevalent genotype among the double‐mutation carriers.

MATERIALS AND METHODS

Detection of target site mutations by pyrosequencing

In order to select plants for this study, seeds were sown on a plastic tray filled with potting soil, and upon emergence, ≈400 seedlings were individually transplanted to 8 cm× 8 cm pots filled with the same potting mix. Plants were grown in a glasshouse maintained at 32/28 °C day/night temperature, with supplemental light during the 14 h 10 h, light:dark photoperiod. When seedlings were ≈5–7 cm tall, fomesafen (Flexstar 1.88 EC; Syngenta Crop Protection, Greensboro, NC, USA) at 280 g ai ha−1 was applied using a laboratory sprayer calibrated to deliver 375 L ha−1 of spray mix. A nonionic adjuvant (Dash HC; BASF SE, Ludwigshafen, Germany) was added to the spray mix at 1% v v−1. Plants were returned to the glasshouse immediately after application, and tissue collection from survivors was done at 21 days post‐application (DPA).

DNA extraction, amplification of PPX2 fragments and pyrosequencing were carried out in the laboratory of IDENTXX GmbH (Stuttgart, Germany, www.identxx.com). Leaf tissues (0.5 cm2) of A. palmeri plants were collected and placed individually into microtubes (Qiagen, Hilden, Germany). The leaf samples then were homogenized in a shaking mill (TissueLyser II; Qiagen) using steel beads. The DNA extraction was carried out in the KingFisher™ Flex Magnetic Particle Processors (Thermo Fisher Scientific, Schwerte, Germany) using the Chemagic™ Plant 400 kit (Perkin Elmer, Rodgau, Germany) according to the manufacturer's instructions (modified by IDENTXX GmbH). The Endpoint PCR amplification (gDNA concentration of 20–50 ng μL−1 per sample) for each target region were performed using a MyFi™ DNA Polymerase Kit (Bioline GmbH, Luckenwalde, Germany) and specific primers (IDENTXX GmbH) in a PCR thermal cycler (T100 PCR thermal cycler, Bio‐Rad Laboratories Inc.) under the following conditions: 3 min at 95 °C and 42 cycles of 10 s denaturation at 95 °C; 35 s annealing at 60 °C and 45 s elongation at 72 °C; and a final elongation step at 72 °C for 5 min. The successful amplification (225 bp for the ∆G210, 250 bp for the R128G product) was checked per gel electrophoresis on a 1.5% agarose gel.

The PCR products were analyzed for SNPs at the target positions via pyrosequencing on a PyroMark Q24 (Qiagen) using specific sequencing primers (IDENTXX GmbH). During the sequencing reaction, all incorporated nucleotides of a short region that encompasses the SNP position of interest were detected and reported by creating a pyrogram in a pyrorun file. The file was read using pyromark Q24 software (v2.0.7) and visually examined for mutations, both heterozygous and homozygous.

Verification of double mutants by cloning in Escherichia coli

The following work was carried out in the laboratory of IDENTXX GmbH (Stuttgart, Germany, www.identxx.com). A 0.5‐cm2 leaf sample of each plant (continuously cooled at −20 °C) was transferred into microtubes. The samples then were homogenized at room temperature in a shaking mill. The RNA was extracted in the KingFisher™ Flex Magnetic Particle Processors (Thermo Fisher Scientific Inc.) using the MagMAX™ Plant RNA Isolation Kit (Thermo Fisher Scientific Inc.) according to the manufacturer's instructions (modified by IDENTXX GmbH). The cDNA transcription was done using the High‐Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific Inc.) according to the manufacturer's instructions (modified by IDENTXX GmbH). The Endpoint PCR amplification for the target region encompassing both SNPs positions were performed using MyFi™ DNA Polymerase Kit and specific primers (IDENTXX GmbH) in a PCR thermal cycler under the following conditions: 5 min at 95 °C and 42 cycles of 20 s denaturation at 95 °C; 35 s annealing at 60 °C and 90 s elongation at 72 °C; and a final elongation step at 72 °C for 5 min. The successful amplification of the 1850‐bp PCR product was checked per gel electrophoresis on a 1.5% agarose gel.

PCR products were cloned using StrataClone PCR Cloning Kit (Agilent, Waldbronn, Germany). Positive white colonies were randomly picked and verified with colony PCR. For each clone, ten positive PCR fragments were randomly selected and verified via Sanger sequencing (SeqLab‐Microsynth, Göttingen, Germany). Sequences were analyzed using geneious prime software v9.1.8 (Biomatters, Auckland, New Zealand).

Evaluation of mutant PPO enzyme activity

The complete description of expression and purification of Amaranthus PPO2 variant proteins and the enzymatic assay to determine protein activity was referenced to the method described by Rangani et al. To calculate the percentage of remaining protein activity, the enzyme activity of WT PPO2 was divided by the activities of R128G, ∆G210 and R128G ∆G210 PPO2 variants and multiplied by 100. This was further normalized to the amount (ng) of protein used in the assay.

Arabidopsis transgenics growth and herbicide treatment

Donor plant material and growth conditions

Arabidopsis thaliana seeds (stock MC24, from the Max Planck Institute for Molecular Plant Physiology at Golm) were sown into a substrate composed of GS90 soil + 5% sand. Plants were subjected to stratification for 5 days at 4 °C, followed by a short‐day growth period of 10 days (10 h:14 h, day:night photperiod at 20/18 °C ± 1 °C, ~120 μmol PAR). After that, plants were transplanted into 8 cm × 8 cm pots filled with GS90 soil and cultivated under the same conditions for 14 days, under long‐day growth conditions (16 h:8 h, day:night at 20/18 °C ± 1 °C, ~200 μmol PAR) and maintained until seed harvest. Plants were fertilized with 0.3% Hakaphos Blau (15‐10‐15 NPK) twice a week until flowering. Relative humidity was not controlled, but kept between 40% and 70% during all growth stages, except during stratification.

Transgene preparation

In order to prepare the transgene, the mutant PPX2 was inserted into RTP6557 transformation vector, which was then inserted into Agrobacterium tumefaciens strain C58C1pMP90. The gene insert also included an ALS‐herbicide‐resistance trait as a selectable marker to identify transformed Arabidopsis seedlings. This would ensure that plants eventually tested for resistance to PPO herbicide all expressed the transgene.

Bacterial culture and dipping medium

Agrobacterium culture containing the plasmid was prepared one day before dipping by inoculating 1 mL glycerol stock into 250 mL YEB medium (1 g L−1 yeast, 5 g L−1 beef extract, 5 g L−1 peptone, 5 g L−1 sucrose, 0.49 g L−1 MgSO4·7H2O) + appropriate antibiotic. The bacteria were cultured for 12 h at 28 °C with continuous agitation at 150 rpm. The next day, after adjusting the Agrobacterium culture density to optical density at 600 nm (OD600) = 1.0 (with YEB medium), the culture was collected by centrifugation at 4000 rpm for 10 min and re‐suspended in 150 mL infiltration medium composed of 2.2 g L−1 MS (Murashige & Skoog medium), 50 g L−1 sucrose, 0.5 g L−1 MES hydrate and 10 μL L−1 BAP (benzylaminopurin, 1 mg mL−1). The pH then was adjusted to 5.7–5.8.

Agrobacterium‐mediated transformation of A. thaliana by floral dip

Plant transformation was performed following a protocol established previously. Briefly, plants with immature floral buds were dipped in the bacterial suspension for 10 s after adding 75 μL Silwet‐L77 per 150 mL infiltration medium to a jar. After dipping, plants were kept overnight in a cabinet under high humidity and low light intensity, and were grown under long‐day conditions until maturity. When siliques turned yellow, plants were placed inside paper bags to collect the seeds. T1 seeds then were transferred to falcon tubes and stored at 4 °C.

Selection of putative transformants with Imazamox

After ≥ 14 days of storage at 4 °C, T1 seeds were sown to select putative transgenic Arabidopsis plants. Sowing and stratification was performed as described previously. Following that, seeds were treated with 20 mg L−1 imazamox solution and cultivated under short‐day growth conditions for 12–14 days, when resistant seedlings (four‐leaf stage) were transplanted into 6‐cm pots filled with GS90 soil and grown for another 10 days. One day before herbicide application, growth conditions were set to ‘long‐day’ and maintained throughout the duration of the test. Herbicide treatments consisted of two concentrations of saflufenacil (Kixor, BASF Corporation, at 10 and 25 g ai ha−1) foliar‐applied when plants reached the ten‐leaf stage, using a spray chamber calibrated to deliver 375 L ha−1 of spray solution. Herbicide efficacy was assessed visually 7 days after herbicide treatment.

Verifying PPO2 double mutants from heterozygous dG210 × heterozygous G399A

In a previous study we learned that ∆G210 −/+ G399A −/+ was the most common genotype among the double mutant plants identified. The double mutants referred to here were plants carrying two PPX2 mutations, but without confirmation on whether the two mutations occurred in one allele. To study the inheritance pattern of this genotype, seeds from the selected Palmer amaranth populations were sown in a plastic tray containing potting mix (Sunshine LC1; SunGro Horticulture, Agawam, MA, USA). After one week, 400 seedlings were transplanted to 50‐cell trays at one seedling per cell. When plants reached the 6‐ to 8‐cm stage, fomesafen (264 g ai ha−1, Flexstar® 1.88 EC) was applied in a spray chamber, equipped with two flat‐fan 1 000 065 nozzles, calibrated to deliver 187 L ha−1 of spray mix at a speed of 3.6 km h−1. At 21 DPA, leaf tissues from ≈200 survivors were individually collected into 1.5‐mL microtubes (VWR International LLC, Radnor, PA, USA) and stored at −80 °C until processing. DNA was extracted and the plants were genotyped via pyrosequencing as described previously. Four males and 14 females of the same genotype (∆G210 −/+ G399A −/+) were individually transplanted to 8‐L pots and grown together to interbreed. Seeds were harvested separately from each female plant and cleaned. The germination capacity of 14 F1 lines was evaluated. Six F1 lines with the higher germination capacity were selected, and ≤100 plants from each female were selected for genotyping via pyrosequencing.

Because Palmer amaranth is diploid, a plant that is heterozygous for both mutations can either: (i) have one allele containing two mutations, whereas the other allele is WT; or (ii) have each allele carrying one mutation only. The Palmer amaranth parent population used in this study exhibited the latter condition. A cross between two double‐heterozygous is expected to result in 25% of the offspring carrying a homozygous ∆G210, 25% carrying a homozygous G399A and 50% of the offspring having the same parental genotype. A Pearson's chi‐squared test was used to verify this hypothesis, where the calculated chi‐squared value [χ 2 Calc, as shown in Eqn (1)] is compared to a tabulated value (χ 2 Tab) where Oi is the observed number of plants of a genotype and Ei is the expected number of plants of a genotype.

Lastly, the occurrence of double mutants was verified by cloning the PPX2 gene into E. coli as described previously. A total of 26 samples were chosen for this procedure, and around ten clones from each sample were Sanger‐sequenced.

RESULTS AND DISCUSSION

PPO mutation profile of R128G+/+ ΔG210−/+ offspring

A male and a female plant LIH11640A and LIH11640F carrying R128G+/+ ΔG210+/− (where + indicates presence of the mutation), were detected in a large biotype screening of putative resistant A. palmeri. The R128G+/+ ΔG210+/− plants carried homozygous R128G, indicating that these mutations can occur on the same allele. This finding answers the question that we posed in a previous study, where this genotype was not observed among the double mutants. In that study, we hypothesized that: (i) either the occurrence of this genotype was too rare and would require a larger sample size to be detected; or (ii) the resultant protein would be inactive and, therefore, the fitness cost associated with this double‐mutation would be lethal.

In order to determine whether homozygous double mutants R128G ΔG210 plants are viable, LIH11640A was crossed with LIH11640F and the PPX2 of LIH11640F1 offspring was subjected to pyrosequencing for the positions R128 and G210. If the R128G ΔG210 PPO2 is functional, and inheritance follows a Mendelian pattern, it is expected that two parents having the R128G+/+ ΔG210+/− genotype would result in 25% of the offspring being double‐homozygous, 25% being R128G+/+ G210 and 50% having the parental genotype (Fig. 1).

Figure 1

The expected segregation pattern of the offspring from a cross of homozygous R128G mutant (R128G+/+) and heterozygous ΔG210 mutant (ΔG210+/−).

Of 100 plants genotyped from the LIH11640F1 population, only 29% of the offspring carried the parental genotype; 71% of the plants were R128G +/+ G210; and none were homozygous double mutants (Table 1). The calculated chi‐squared value for this population was three‐fold higher than the tabulated (χ 2 Calc = 18.46, χ 2 Tab = 5.991), indicating that observed and expected ratios do not agree. Owing to the odd inheritance pattern in LIH11640F1, additional R128G+/+ ΔG210+/− male and female plants from this population were selected and crossed to generate another three independent LIH11640F2 populations, which also were pyrosequenced (Table 1). The mutation inheritance pattern among the three F2 populations conformed to that of the F1 population, where no homozygous double mutants were detected and most of the offspring carried a single, homozygous R128G mutation (R128G +/+ G210). All calculated chi‐squared values were higher than the tabulated. These results indicate that, although rare individuals containing both R128G ΔG210 mutations can occur, their co‐occurrence on both alleles is not possible. In addition, the percentage of F1 and F2 plants carrying R128G+/+ ΔG210+/− was significantly lower than the expected ratio (Table 1), suggesting that having R218G and ΔG210 in one allele has a significant impact on fitness cost.

Table 1

Segregation of mutant alleles in F1 and F2 generations of PPO‐herbicide‐resistant A. palmeri genotypes

F1 line	Number of plants							χ 2 Calc
	Genotyped	Expected			Observed
		128G +/+ ∆210 −/−	128G +/+ ∆210 +/−	128G +/+ ∆210 +/+	128G +/+ ∆210 −/−	128G +/+ ∆210 +/−	128G +/+ ∆210 +/+
LIH11640 F1	100	25	50	25	71	29	0	118.46
LIH11640 F2‐1	100	25	50	25	65	35	0	93.50
LIH11640 F2‐2	100	25	50	25	73	27	0	127.74
LIH11640 F2‐3	100	25	50	25	70	30	0	114.00
Grand total	400	100	200	100	279	121	0	451.62

Verification of double mutant allele combination by cloning in E. coli and Sanger sequencing

For further verification of the pyrosequencing results, the plants coded as LIH11640A and LIH11640F were used for RNA extraction, cDNA synthesis and amplification of the whole‐length PPX2 gene, which was later cloned into E. coli. For each plant, ten clones were selected for Sanger sequencing.

As expected, all clones from both plants carried the R128G mutation. The number of clones containing double mutations was similar for both plants studied: four and five of ten clones for plants LIH11640A and LIH11640F, respectively (Table 2). Because the PPX2 fragments inserted into E. coli were derived from RNA of the parent plants, the balanced distribution of double‐ and single‐mutants suggests that there is no difference in the expression of those alleles. In other words, LIH11640A and LIH11640F plants do not favor the expression of one allele over the other, and the amount of double‐ and single‐mutant proteins produced in vivo are equivalent.

Table 2

Confirmation of the occurrence of R128G and ΔG210 mutants in PPO2 clones in E. coli from PPO‐herbicide‐resistant A. palmeri plants

Plant ID	Clone ID	R128G	∆G210
AMAPA LIH11640A	Clone 1	+	+
	Clone 2	+	−
	Clone 3	+	+
	Clone 4	+	−
	Clone 5	+	−
	Clone 6	+	−
	Clone 7	+	+
	Clone 8	+	−
	Clone 9	+	−
	Clone 10	+	+
AMAPA LIH11640F	Clone 1	+	+
	Clone 2	+	−
	Clone 3	+	−
	Clone 4	+	+
	Clone 5	+	−
	Clone 6	+	−
	Clone 7	+	+
	Clone 8	+	−
	Clone 9	+	+
	Clone 10	+	+

Evaluation of double‐mutant R128G ΔG210 PPO2 enzymatic activity

The segregation of mutations in the offspring suggests that the double mutant (R128G ΔG210) protein activity is either severely compromised or nonexistent, and plants carrying such double‐mutant allele rely exclusively on the other, single‐mutant allele to survive. To confirm this hypothesis, the WT, single‐mutant and double‐mutant enzymes were heterologously expressed in E. coli, the PPO2 protein was purified, and the enzyme activity was quantified. Enzyme activity data are shown as percentages relative to the WT protein (Table 3). Of the three mutants tested, the R128G variant showed the highest remaining activity, followed by the ΔG210 and the double‐mutant variants. Using the same methodology, protein activity of the G399A mutant was reported as being 3% of the WT. The enzyme activity of the double mutant was too low to be detected, indicating that it was practically inactive. The inactivity of the double‐mutant enzyme agrees with our previous study, where we deduced that a R128G ΔG210 protein would lose its binding pocket integrity.

Table 3

Enzyme activity assay of A. palmeri PPO2 mutants

PPO variant	FU min−1	% remaining activity
WT PPO2 (25 ng)	584	100
∆G210 (500 ng)	1500	11
R128G (25 ng)	250	50
R128G ∆G210 (500 ng)	15	0.12

The effect of ΔG210 and R128G mutations on PPO activity already has been investigated and our results somewhat agree with those from previous studies. , , , These mutations are known to affect PPO activity and substrate binding differently. Although ΔG210 did not affect substrate binding (no differences in Km between WT and mutant), the mutation reduced enzyme activity ten‐fold. The deletion of G210 causes an enlargement of the substrate binding pocket and displaces G207, which has an important role in positioning protogen at the ideal distance and orientation from the co‐factor FAD, favoring the rate‐limiting hydride abstraction. The loss of that interaction results in lower protein activity. By contrast, R128G impairs protein binding efficacy, but doubles its activity. The decrease in substrate binding is caused by the loss of the salt bridge interaction between R128 and protogen; however, this also facilitates the release of proto from the binding site. In addition, the substitution of the bulky, charged Arg to the small, nonpolar Gly, increases the opening of the binding pocket, further speeding the proto egress. ,

Interestingly, despite the fact that R128G ΔG210 protein is inactive and its fitness cost is demonstrated by the odd inheritance pattern of this genotype, R128G+/+ ΔG210+/− plants do not have any apparent growth impairment or visible injury. Likewise, the G399A mutation was shown to cause a ~97% reduction in protein activity compared to the WT, but homozygous G399A plants exist , and are able to grow and develop normally. How plants are able to overcome the lower PPO2 activity is unknown and represents an interesting point to be studied.

In order to confirm the double‐mutant inactivity, A. palmeri PPO2 carrying R128G ΔG210 was overexpressed in Arabidopsis and the resulting transgenic plants were treated with two rates of saflufenacil (10 and 25 g ha−1). It is noteworthy that transgenic plants still carried their native PPO2, otherwise, growth of untransformed controls would not be possible. If double mutant protein was active, transgenic plants expressing R128G ΔG210 would show increased tolerance to saflufenacil. However, herbicide application killed the transgenic R128G ΔG210 Arabidopsis, and nontreated plants did not show any growth differences compared to the nontransformed plants (Fig. 2). These results indicate that the overexpressed protein neither conferred tolerance to saflufenacil nor affected the total PPO2 activity in those plants. Therefore, the only functional PPO2 in the transgenic Arabidposis was the native, WT protein, which was inhibited by the herbicide and led to plant death.

Figure 2

T1 35S:AMAPA PPO2 R128G ∆G210 Arabidopsis lines sprayed with saflufenacil. Pictures were taken 7 DPA. The bottom two pots contain WT plants, the upper five pots contain independent transgenic events (T1 plants, selected with Imazamox by confirming presence of resistance gene ALS).

PPO mutation profile of crosses between double heterozygous ∆G210 G399A

Among field populations of Palmer amaranth, the most common PPX2 mutation is ΔG210 and the most common mutations detected in the same population are ΔG210 and G399A. Therefore, it is only logical to assume that as a consequence of the obligate outcrossing behavior of this species, the co‐existence of single‐mutant plants (regardless of zygosity) in one population would only increase the frequency of ∆G210 G399A double mutants with time. This could then become the most common resistant genotype that would challenge current and future PPO‐inhibiting herbicides – that is, if such double mutant does not carry fitness penalty. Hence the follow‐up experiment presented herein. We expected that a cross between two double‐heterozygous ∆G210 G399A would result in 50% of the offspring carrying a single homozygous mutation (either ∆G210 or G399A), whereas the other 50% would carry both mutations concomitantly as heterozygous. To check if the observed values agreed with the expected inheritance pattern, a chi‐squared test was done. The χ 2 Calc values for individual F1 lines ranged from 1.02 to 4.73, and were lower than the tabulated threshold (χ 2 Tab = 5.991), confirming that the observed inheritance pattern agrees with the expected (Table 4). The same result was obtained when considering samples across all F1 lines (N = 376), where χ 2 Calc = 2.33.

Table 4

Genotyping of Palmer amaranth PPX2 mutations via pyrosequencing from six F1 lines produced by a cross between double‐heterozygous ∆G210/G399A parents

F1 line	Number of plants							χ 2 Calc
	Genotyped	Expected			Observed
		∆G210 −/− G399A +/+	∆G210 +/− G399A +/−	∆G210 +/+ G399A −/−	∆G210 −/− G399A +/+	∆G210 +/− G399A +/−	∆G210 +/+ G399A −/−
LIH19623	52	13	26	13	14	19	19	4.73
LIH19624	93	23.3	46.5	23.3	16	51	26	3.02
LIH19625	97	24.3	48.5	24.3	21	56	20	2.34
LIH19626	20	5	10	5	3	10	7	1.60
LIH19627	14	3.5	7.0	3.5	3	10	1	3.14
LIH19628	100	25	50	25	28	51	21	1.02
Grand total	376	94	188	94	85	197	94	2.33

A total of 26 plants double‐heterozygous from all of the F1 lines were selected for cloning of the PPX2 gene into E. coli. At least ten clones per plant were submitted for Sanger sequencing. In this case, half of the clones were expected to contain the G210 deletion, whereas the other half would carry G399A. This distribution was tested using the Pearson's chi‐squared goodness‐of‐fit test. Four of six F1 lines had observed distribution equal to the expected, but the lines LIH19626 and LIH19628 did not fit this pattern (Table 5). In these two cases, the higher number of clones containing the ∆G210 mutation indicates an unequal expression of this allele in comparison to the allele containing G399A. Sequence polymorphisms within gene regulatory elements (promoters, enhancers, silencers) can affect transcription rate and mRNA stability, leading to a biased allele expression. DNA methylation and/or histone modifications are known to regulate monoallelic expression, such as in genomic imprinting. The involvement of those mechanisms in what seems to be a preferential expression of the ∆G210 allele over the G399A allele is not understood, yet represents an interesting follow‐up question to be answered.

Table 5

Sanger sequencing of E. coli clones expressing the PPX2 gene from six F1 A. palmeri lines

F1 line	No. of plants cloned	Number of clones					χ 2 Calc
		Sequenced	Expected		Observed
			dG210	G399A	dG210	G399A
LIH19623	6	62	31	31	30	28	0.067
LIH19624	3	30	15	15	17	17	0.267
LIH19625	3	30	15	15	17	13	0.267
LIH19626	5	51	25.5	25.5	35	16	7.078
LIH19627	3	36	18	18	15	21	1.000
LIH19628	6	62	31	31	40	22	5.225

CONCLUSIONS

The co‐occurrence of two mutations in the same PPX2 allele, although rare, is possible. However, the coded protein has a severely reduced activity, which is supported by the fact that no double‐homozygous plants exist, and that its insertion in Arabidopsis does not confer resistance to PPO inhibitors. Plants carrying a double‐mutant allele must rely on the second allele to produce a functional enzyme. The fitness cost in higher‐order plants carrying a double‐mutant allele is yet to be determined.

CONFLICT OF INTEREST

Authors affiliated with BASF have contributed to the planning and implementation of research activities. All other authors declare no conflict of interest.