Description
While studying the effects of RALF peptides on reproduction in Arabidopsis, we noted reduced seed set in F1 lines containing a
RALF17
(
At2G32885;
as per (Abarca
et al.
, 2021)) GABI-Kat T-DNA insertion (GK_699G03.03, N467083) in the hemizygous state, but no effect on seed set in homozygous lines. Given reports of genomic rearrangements causing similar phenotypes (Alling and Galindo-Trigo, 2023), we describe here our phenotypic observations and genotyping in order to alert the Arabidopsis research community to issues with this specific plant line.
A
ralf17-1 ralf20-1
cross was performed due to potential functions of the cognate proteins in pollen grain hydration and an unexpected reduction in seed number per silique was noted in the F1 generation (i.e. plants hemizygous for both T-DNA insertions) (Lau, 2023). Production of a homozygous double mutant was not possible due to the linkage of the two genes (
AT2G32885
and
AT2G34825
). However, normal seed set was observed for each individual T-DNA insertion line in a homozygous state, indicating the observed phenotype was not due to lethality of one or other of the gene knockouts during gamete development.
To assess the extent of the seed development phenotype in the
RALF17/ralf17-1 RALF20/ralf20-1
F1 plants, siliques were cleared and seeds counted in comparison to the homozygous single T-DNA lines and a Col-0 wildtype control. The F1 plants had significantly fewer seeds than the Col-0,
ralf17-1/ralf17-1
or
ralf20-1/ralf20-1
controls (
Fig. 1A,
B
; Kruskal-Wallis test; Col-0 vs F1 p = 0.01,
ralf17-1
vs F1 p = 0.002,
ralf20-1
vs F1 p = 0.01), with a mean of 25.6 seeds per F1 silique compared to 62.7 in Col-0, 64.0 in
ralf17-1
and 63.4 in
ralf20-1
. While the F1 plants produce fewer than half as many seeds, it was unclear which aspect of reproduction was affected (e.g. gamete development, pollen tube growth, pollen tube reception or seed development), and whether the
RALF17
or
RALF20
T-DNA insertion was the genetic determinant.
To determine which of the
RALF17
or
RALF20
T-DNA insertions was responsible for the observed reduction in seed set, the
ralf17-1
and
ralf20-1
homozygous lines were each backcrossed to Col-0 and seed numbers in the F1 generation counted. With seed counts in the
RALF17/ralf17-1
plants averaging 22.0 compared to a mean of 58.8 in the
RALF20/ralf20-1
plants (
Fig. 1A,
B
), the
RALF17
T-DNA insertion was determined to be causal.
Crosses with the wildtype Col-0 line further explored the genetics of the phenotype. As the two T-DNAs are linked, non-Mendelian inheritance in a 1:1 ratio is expected in the seeds of a cross to Col-0. Genotyping of 99 seedlings from the Col-0 (female) x
RALF17/ralf17-1 RALF20/ralf20-1
(male)
cross demonstrated equal transmission of the two T-DNAs via pollen (47 and 52 progeny hemizygous for the
ralf17-1
and
ralf20-1
T-DNAs, respectively; Chi-squared test p = 0.62). In contrast, transmission of the two T-DNA insertions via the female lineage in a
RALF17/ralf17-1 RALF20/ralf20-1
x
Col-0 cross indicated reduced inheritance of the
ralf17-1
T-DNA (35 and 56 progeny inheriting the
ralf17-1
and
ralf20-1
T-DNAs, respectively; Chi-squared test p = 0.03). Thus, viable pollen grain formation is not significantly affected in the F1
RALF17/ralf17-1 RALF20/ralf20-1
plants, while
ralf17-1
haploid ovules may have reduced viability.
To further understand the
ralf17-1
hemizygous phenotype, we examined gamete development, including pollen grains despite the lack of transmission effect. Pollen structure was examined through imaging with a scanning electron microscope. The wildtype Col-0 line and both single mutants had elongated pollen grains with three evenly spaced apertures and a regular exine pattern. A high proportion of the pollen grains of the
RALF17/ralf17-1 RALF20/ralf20-1
plants were much smaller and more rounded with wider, less regular apertures and an irregular, less defined exine pattern (
Fig. 1C
). The F1
RALF17/ralf17-1
had pollen grains similar to the double mutant, with many pollen grains having the smaller, irregular shape.
RALF20/ralf20-1
pollen grains, however, were comparable to Col-0 or the single homozygous mutants.
Therefore, although transmission of the T-DNA insertions via pollen appears unaffected by whether the haploid genome contains the
ralf17-1
or
ralf20-1
T-DNA insertion, pollen grain development is nonetheless affected by hemizygousity of the
ralf17-1
T-DNA insertion in the parental plant.
Female gamete development was examined through callose staining of unfertilised carpels (48 hrs after emasculation to ensure mature ovules). Unfertilised, mature ovules in the
RALF17/ralf17-1 RALF20/ralf20-1
plants could be separated into two categories - those with limited/no callose deposition in the embryo sac that are comparable to ovules of wildtype Col-0 and those with increased deposition of callose in the embryo sac (
Fig. 1D,
E
). These two categories of embryo sacs were observed at 45% (normal) and 55% (increased callose) frequency in
RALF17/ralf17-1 RALF20/ralf20-1
plants (
Fig. 1F
). Examining earlier stages of ovule development in Col-0, we did not observe callose deposition in the embryo sacs, indicating that this is ectopic callose deposition rather than a delay or cessation of ovule development. Combined with the T-DNA insertion transmission rates and seed set numbers presented above, the embryo sac phenotype does not appear to strictly correlate with a lack of viability of the ovules, as otherwise we would expect a much stronger effect on T-DNA insertion transmission via the female lineage and an inability to obtain homozygous
ralf17-1
plants. Aberrant embryo sac development could alternatively be a function of genomic rearrangements which lead to developmental defects rather than arising from hemizygousity of the
ralf17-1
T-DNA insertion per se.
Whole genome Illumina sequencing was undertaken to identify if a second T-DNA insertion is present in the
ralf17-1
GABI-Kat line. As well as confirming the T-DNA insertion in
RALF17
(
Fig. 1G,
H
), a second insertion on chromosome 1 at
AT1G51370
(encoding an uncharacterised F-box protein) was identified based on a gap in reads mapped to the
Arabidopsis thaliana
genome (
Fig. 1I
) and mapping of read contigs to the pGABI1 plasmid. This insertion, confirmed by PCR (
Fig. 1J
), consists of the plasmid backbone between the right and left T-DNA borders of the pGABI1 plasmid, explaining why it was not identified during standard T-DNA line characterisation. Our mapping approach cannot address whether there is a genomic rearrangement incorporating these two T-DNA insertions as individual reads do not span the insertions.
The ovules and pollen grain phenotypes we have observed in plants hemizygous for the
ralf17-1
T-DNA insertion are consistent with phenotypes previously reported for other T-DNA insertion lines with genomic rearrangements. Reports of chromosomal translocations in non-T-DNA-containing plant lines of Datura and maize, that result in meiotic and gametophytic defects, date back a century (Belling and Blakeslee, 1926). With the advent of T-DNA lines for genetics research, genomic rearrangements became a more frequent pitfall. A number of studies have reported gametophytic defects and chromosomal translocations in individual Arabidopsis T-DNA lines (e.g. (Curtis
et al.
, 2009, Alling and Galindo-Trigo, 2023)). One study of 64 SALK T-DNA lines found a 19% rate of chromosomal translocations (Clark and Krysan, 2010) indicating the prevalence of this issue. A smaller study which sequenced the genomes of 14 GABI-Kat T-DNA insertion lines identified chromosomal translocations in four lines, along with numerous T-DNA insertion sites that had not been annotated, concluding that GABI-Kat lines may contain an average of two T-DNA insertions (Pucker
et al.
, 2021). The presence of multiple T-DNA insertions is not however required for chromosomal translocations to occur in insertion lines (reviewed in (Thomson
et al.
, 2024)). As recently highlighted by (Raabe
et al.
, 2024), the gametophytic effects of potential chromosomal rearrangements are of particular concern when studying plant reproduction, with screening of gametophyte phenotypes in F1 crosses of particular relevance. Here we would add our voice to those of others, urging caution when gametophytic phenotypes are observed in hemizygous T-DNA insertion lines but not cognate homozygous lines, with particular warning regarding the RALF17 GABI-Kat T-DNA insertion line GK_699G03.03, NASC N467083.
Methods
Plant growth
The
ralf17-1
and
ralf20-1
T-DNA insertion lines (GK_699G03.03 and SALK_089411C, respectively; (Alonso
et al.
, 2003, Kleinboelting
et al.
, 2012)) were acquired from the Nottingham Arabidopsis Stock Centre. Plants were grown in a 3:1 mix of Levington M3 compost:sand in a Conviron growth chamber. Set growth conditions were 22ºC, ~150 µmol
-1
m
-2
light from LEDs for 16 hrs per day and 60% humidity. Genotyping used Qiagen multiplex PCR mix (catalogue number 206143) as per the manufacturer’s guidance, with primers as per the sequences below.
Developmental imaging
For seed counts, mature siliques (7-8 per plant from 6 plants per genotype) were cleared for 2-4 days in 0.4 M NaOH 1% Triton-X 100 at room temperature before photographs were taken. Seed counts were scored from the photographs.
For pollen imaging, anthers were mounted onto aluminium stubs with carbon tabs (Agar Scientific) and imaged using a Hitachi TM3030Plus benchtop scanning electron microscope in secondary electron mode at 15kV.
For aniline blue/sirofluor staining of ovules (Conrath, 2024), anthers were removed from the flowers (stage 16) prior to dehiscence and unpollinated carpels collected 48 hrs later to ensure ovule maturity. Staining and imaging with a Leica DM6 epifluorescence microscope were as described in (Galindo-Trigo
et al.
, 2020). Ovule phenotypes were categorised for 2-3 carpels per plant and 3 plants per genotype.
Statistics
Seed counts were compared using a Kruskal-Wallis test with correction for multiple testing (Dunn’s multiple comparisons test) performed in GraphPad Prism v 10.0.2. T-DNA insertion transmission rates were analysed in Microsoft Excel (v 16.90.02) using a Chi-square test.
T-DNA insertion mapping
To identify T-DNA insertion sites in the
ralf17-1
mutant, genomic DNA was extracted from 3-week-old plants using the DNeasy Plant Maxi Kit (Qiagen). The DNA sample was submitted to Novagene for Illumina sequencing to generate 150-bp paired-end reads. The raw sequencing reads were processed using Fastp (Chen
et al.
, 2018) to remove adapter sequences and trim low-quality bases. The resulting filtered reads were initially mapped to the
Arabidopsis thaliana
reference genome (TAIR10) using the Burrows-Wheeler Aligner (BWA; (Li and Durbin, 2010)). The resulting SAM files were converted to sorted BAM files using SAMtools (Li
et al.
, 2009). The sorted BAM files were then visualised using Integrative Genomics Viewer (IGV), version 2.19.8 (Robinson
et al.
, 2011), allowing the T-DNA integration site in
ralf17-1
to be identified.
To determine whether additional T-DNA integrations were present in the mutant, the filtered Illumina sequencing reads were used to generate
de novo
genome assemblies for
ralf17-1
using SPAdes Genome Assembler, version 4.3.0 (Prjibelski
et al.
, 2020). The
ralf17-1
assembly consisted of 4,728 contigs, with a total assembly length of 124,456,250 bp and an N50 of 100,402 bp.
The genome assembly was subsequently searched for sequences derived from the full-length pGABI1 plasmid sequence using BLAST+ (Camacho
et al.
, 2009). BLAST hits corresponding to plasmid-derived sequences were then examined to identify contigs containing T-DNA sequences and to assess the possibility of additional T-DNA integration sites.