Strong morphological defects in conditional Arabidopsis abp1 knock-down mutants generated in absence of functional ABP1 protein

Plant material and growth conditions

Arabidopsis thaliana mutants used in this study were: abp1-c1, abp1-TD1 (Gao et al., 2015), abp1-AS, SS12K9, SS12S6 (Braun et al., 2008; David et al., 2007). A. thaliana Col-0 wild type seeds were obtained from The Nottingham Arabidopsis Stock Centre (NASC, http://www.arabidopsis.info). For in vitro experiments, seeds were surface-sterilized with chlorine vapor, vernalized for 2 days in the dark at 4°C and grown on 1/2 MS 0.8% agar medium with or without 1% w/v sucrose (pH 5.9) on vertical Petri dishes under long day conditions (16 h light/8 h dark) or in complete darkness at 21°C. A sterilized microtube with 500 µl 5% ethanol was placed at the bottom of the plate to induce expression of abp1-AS, SS12K9 and SS12S6 constructs before germination. Plates with 5-day old etiolated or 7-day old light-grown seedlings were scanned on a flatbed scanner, phenotyped by visual examination and used for DNA/RNA extraction.

Genotyping mutants

Ethanol-inducible ABP1 down-regulating lines (abp1-AS, SS12K9, SS12S6) were genotyped for the presence of the alcR gene encoding the transcriptional regulator of the ethanol-inducible system using primers alcR_for and alcR_rev (Table 1). Fragments amplified from abp1-c1 with primer pairs ABP1-U409F + ABP1-586R or ABP1-5P + ABP1-586R were digested with BslI, which cuts the WT fragment once and does not cut the mutant fragment; abp1-TD1 was genotyped as described previously (Gao et al., 2015). Genomic DNA was isolated using the CTAB extraction method. GoTaq G2 polymerase (Promega) and Bio-Rad T100 Thermal Cycler were used for PCR under following conditions: initial denaturation 5 min 98°C; 35–45 cycles (denaturation 30 s at 98°C; annealing 30 s at 55°C, elongation 1 min at 72°C); final elongation 5 min at 72°C. Restriction analysis was performed by adding the restriction enzyme directly to unpurified PCR reaction. Alternatively, Phire Plant Direct PCR Kit (Thermo Scientific by Finnzymes) and QIAquick Gel Extraction Kit (QIAGEN) were used following manufacturer’s instructions to genotype the SS12K9 x abp1-c1 line.

Quantitative RT-PCR

Total RNA from approximately twenty 8-day old seedlings frozen in liquid nitrogen was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and purified using RNeasy Mini Kit (Qiagen) according to manufacturer’s instructions. 2 µg of purified total RNA were used for a reverse transcription reaction using the iScript cDNA Synthesis Kit (BioRad). qRT-PCR was performed using the LightCycler 480 SYBR Green I Master chemistry (Roche) in a LightCycler480 II thermal cycler (Ser. no. 5659, Roche) according to manufacturer’s instructions. cDNA diluted 1:10 in water was used as a template to prepare 5 µL reaction mixture (final volume). Primers used for the quantitative RT-PCR were designed using QuantPrime (http://www.quantprime.de). The ABP1 cDNA fragment (84 bp in length) was amplified with ABP1-2E and ABP1-586R primers. Arabidopsis Tubulin beta chain 2 (TUB2, At5g62690) amplified with TUB2-F and TUB2-R primers was used as a reference gene (Dataset 1). Gene expression was calculated with the 2^-ΔΔCT method (Livak & Schmittgen, 2001). Results are expressed as the average +/- standard deviation of 2 biological and three technical replicates. Sequences of primers used for genotyping and qRT-PCR analysis are listed in Table 1.

Segregation of strong morphological defects in conditional abp1 knock-down alleles crossed with abp1-TD1 and abp1-c1 knock-out alleles

To investigate the contradiction between missing phenotypic defects in the loss-of-function abp1 alleles and strong morphological defects of conditional ABP1 down-regulating lines (knock-down; KD), we decided to cross both types of lines to test three possible scenarios: 1) The absence of the strong morphological defects in the abp1-c1 or abp1-TD1 alleles is caused by an adaptation of the plants to the permanent loss of the ABP1 function, which compensates for this deletion; 2) the strong morphological phenotypes induced in the KD lines do not require functional ABP1 and are caused by off-target effects; or 3) both abp1-TD-1 and abp1-c1 lines contain background mutation(s) that suppress the phenotypes caused by the absence of ABP1.

We crossed each of the conditional lines with abp1-TD1 and abp1-c1 null mutants and with an ABP1-WTc1 line as a control and analyzed seedling phenotypes of ethanol induced F2 segregating plants (Figure 1a). We hypothesized that in case of an adaptive process, the conditional abp1 KD phenotypes (short wavy roots and epinastic cotyledons) would not be manifested in homozygous abp1 null background, resulting in a 9/16 KD and 7/16 WT phenotype segregation ratio. If the inducible phenotypes in the KD lines are independent of ABP1, these phenotypes will be manifested even in the absence of the functional ABP1 gene, thus resulting in a classic Mendelian 3/4 KD and 1/4 WT phenotype segregation ratio. In case of the presence of background suppressive mutation(s), the KD phenotype segregation ratio would lie anywhere between 3/16 (dominant suppressor mutation closely linked to the ABP1 locus) and 3/4 (recessive mutation with low penetrance and no linkage to ABP1) (Supplementary Figure 1).

Segregation of the morphological phenotypes in the F2 plants from different crosses is summarized in Figure 1b. These observations show that strong phenotypes in both the abp1 antisense-based and the scFv12-based conditional knock-down lines segregate approximately 75% in the F2 crosses with abp1-c1. This observation favors the scenario that the strong morphological defects in the KD lines are not influenced by the presence or absence of the functional ABP1 gene copy. The F2 phenotypic segregation is however shifted in favor of WT-looking plants in all three KD lines crossed to abp1-TD1. This segregation shift may be ascribed to partial transcriptional silencing of the ethanol-inducible constructs due to the presence of multiple 35S promoters/enhancers in the constructs themselves as well as the tandem T-DNA insertion in abp1-TD1.

We genotyped all analyzed F2 plants for the presence of the alcR transcriptional regulator, which is an integral part of the ethanol-inducible system and verified that the observed morphological defects were indeed correlating with the presence of the ABP1 KD constructs (Figure 1c). About 5% of seedlings from all lines showed WT phenotype despite being positive for the presence of alcR or vice versa. As this phenomenon was independent of ABP1 genetic background and could not be reproduced in F3 progeny (Supplementary Figure 2), we put it down to biological variability and/or occasional silencing of the ethanol-inducible constructs.

Strong morphological defects in conditional abp1 knock-down alleles can be manifested in homozygous abp1 knock-out alleles

To investigate whether the abp1 KD phenotypes can be observed in the absence of a functional copy of the At4g02980 ABP1 gene we further genotyped the respective abp1 mutations in F2 seedlings of all crosses (Figure 2). As summarized in Figure 2c, in all crosses we were able to identify multiple homozygous abp1 mutants that showed the strong KD phenotype following ethanol induction. This analysis demonstrates that strong morphological phenotypes in abp1 antisense-based (abp1-AS) and scFv12 antibody-based (SS12S6, SS12K9) conditional KD lines can be generated also in a null abp1 background.

In case of the crosses SS12K9 × abp1-c1 and SS12K9 × abp1-TD1 we observed a lower level of allelic segregation between the abp1 mutations and the KD construct in their F2 progeny (Figure 2c). Out of 28 genotyped plants with WT phenotype, 24 (85.7%) were homozygous for abp1 mutation and did not contain the ethanol-inducible KD cassette. These results point towards genetic linkage between these two loci, most likely caused by the positional effect of the KD cassette located close to the ABP1 locus on the chromosome 4. Nevertheless, some level of genetic recombination was happening between the two loci in the crosses as demonstrated by the identification of three F2 SS12K9 × abp1-c1 plants showing KD phenotype that were homozygous for abp1-c1 mutation (Figure 2c). This analysis confirms that also SS12K9 conditional KD construct can generate strong morphological phenotypes in the homozygous abp1 knock-out alleles despite the insertion position being linked to the ABP1 locus. Altogether these data are consistent with results obtained by the other crosses and further support that morphological phenotypes in the abp1 knock-down lines can be generated in the absence of the functional ABP1.

Analysis of F3 generation confirms SS12K9-induced strong morphological defects in absence of ABP1 function

Next we tested the occurrence of the strong KD-induced morphological phenotypes in the absence of the functional ABP1 in the next generation by analyzing the F3 progeny of two SS12K9 × abp1-c1 plants showing strong KD phenotype. We confirmed that the F3 progeny was homozygous for the abp1-c1 mutation and segregated the ethanol-inducible construct approximately in a 3:1 ratio (Figure 3b). After induction with ethanol, the analyzed F3 population of the SS12K9 × abp1-c1 plant A segregated into 27 plants (67.5%) with KD phenotype and 13 WT looking plants (32.5%) (Figure 3). The F3 population of plant B segregated into 18 plants with KD phenotype (81.2%) and 4 WT looking plants (18.2%) (data not shown). Genotyping of all F3 plants with ethanol-inducible phenotypes revealed that they contain KD construct in the homozygous abp1-c1 background (Figure 3b). Notably, among the 17 analyzed WT looking F3 seedlings we also identified two plants that contain the ethanol-inducible construct in homozygous abp1-c1 background (Figure 3b) suggesting that in these plants the functionality of the construct was affected, most probably by its silencing. Nonetheless, the majority of the plants containing the ethanol-inducible construct generated the strong morphological phenotypes even in the abp1^-/- homozygous background.

We also analyzed the ABP1 expression in WT, abp1-c1 and SS12K9 × abp1-c1 F3 seedlings by quantitative RT-PCR just to verify that introducing KD alleles does not influence, in any way, the ABP1 expression (Figure 3c). We observed ca. 80% decrease in ABP1 transcript levels in abp1-c1. We assume that this difference - somewhat surprising, since the CRISPR-induced small deletion does not necessarily decrease transcript levels - is probably caused by the decreased stability of the mutant mRNA. SS12K9 × abp1-c1 F3 plants positive for the KD phenotype and homozygous for abp1-c1 showed the same 80% decrease in ABP1 transcription.

In summary, the phenotypic, genotypic and expression analyses consistently showed that all three conditional abp1 knock-down alleles can generate strong morphological defects also in the absence of the functional ABP1 protein.

Strong morphological phenotypes in abp1 conditional knock-down alleles are not caused by ABP1 down-regulation

All three available conditional abp1 knock-down alleles have been extensively characterized and used to link number of developmental and cellular processes to the ABP1-mediated signaling (for overview, see Grones & Friml, 2015). They are based on two unrelated strategies for down-regulation of the protein’s functionality: the antisense (abp1-AS) and the scFv12 monoclonal antibody expression (SS12S6, SS12K9), which suppress the protein functionality by entirely different mechanisms and at different levels (Tromas et al., 2009). All three lines showed consistent and reproducible results in a number of different laboratories and a number of developmental, physiological and cellular processes.

Nonetheless, our analysis, made possible by the newly available abp1 knock-out lines (Gao et al., 2015), strongly suggests that these observed and described effects are not caused by conditional down-regulation of the ABP1. This is supported by the fact that all three constructs show the same strong conditional phenotypes in two different homozygous abp1 null alleles. This means that even in the absence of the functional ABP1 protein, the ethanol-inducible constructs are inducing phenotypic defects that were originally ascribed to the down-regulation of ABP1. Therefore, results generated using these lines need to be critically re-interpreted.

Possible modes of action of abp1 conditional knock-down lines

All three types of abp1 KD Arabidopsis lines generate indistinguishable morphological phenotypes. How it is possible that independent lines using fundamentally different approaches for functional down-regulation of a unique target would have in fact the same off-target effects; we do not know. One possible explanation is that the morphological defects are an artifact of the ethanol-inducible expression system. However, control lines generated in parallel using the same vector and expressing the UIDA reporter gene did not exhibit any significant growth and developmental alterations (Braun et al., 2008). Furthermore, a number of authors have used the same ethanol-inducible system to control the expression of distinct genes and to the best of our knowledge, there are no reports describing similar phenotypes by using the ethanol-inducible system for other genes in other studies (Battaglia et al., 2006; Deveaux et al., 2003; Laufs et al., 2003; Peaucelle et al., 2008; Roslan et al., 2001). This system was also used to successfully rescue mutant defects after ethanol induction of gene expression e.g. for LEAFY (Maizel & Weigel, 2004) or for N-myristoyltransferase (Pierre et al., 2007) indicating that it is not responsible per se of the phenotypes observed with the ethanol inducible ABP1 AS and scFv12 constructs. In tobacco plants and BY-2 cells, tetracycline de-repressible promoter-driven expression of the SS12S and SS12K constructs resulted in similar growth defects as their ethanol-inducible expression in Arabidopsis (Braun et al. 2008; David et al., 2007), suggesting that the observed phenotypes are tightly correlated to the scFv12 action. The expression of the scFv12 in the cytosol had however no effect on cell proliferation in BY2 cells indicating that expression of scFv12 per se is not sufficient to generate severe phenotypes whatever its cellular localisation and that scFv12 effects are correlated to its secretion and/or retention in the ER that are known location of ABP1 (David et al., 2007).

Another possibility is that both the antisense- and antibody-based lines have off-target(s) either on the very same gene(s) or elements of a common genetic pathway. Such a hypothesis would be supported by strict similarities in the phenotypes resulting from ABP1 antisense and scFv12 expression and by the fact that opposite and auxin-related defects were observed in both constitutive and conditional gain-of-function Arabidopsis transgenic plants as well as transitionally expressing tobacco cells (Grones et al., 2015; Robert et al., 2010). ABP1 is placed within the superfamily of cupins based on the presence of cupin-like motifs HXH(X)₁₁G and P(X)₄H(X)₃N (where X is any amino-acid residue) and a β-barrel jellyroll fold subunit structure (Dunwell et al., 2004; Woo et al., 2002). The epitope recognized by the scFv12 might be present in proteins belonging to this functionally highly diverse protein superfamily. On the other hand, the sequence similarity of even the closest ABP1 homologues in Arabidopsis does not seem to be sufficiently high to be targeted by the abp1-AS constructs, thus this explanation is unlikely as well.

We also cannot completely rule out that the WT phenotype of the abp1 knock-out mutants is caused by suppressor mutation(s). However, we do not consider it very likely, as this would imply that the similar mutation(s) or mutations with similar effects are present in the genetic background of both abp1-c1 and abp1-TD1, which are independent alleles from independent mutant collections.

In summary, we do not understand how it is possible that the used abp1 knock-down alleles generate the similar strong morphological phenotypes also in absence of the functional ABP1 protein. All possible explanations we could come up with are unlikely, including common off-targets in abp1 antisense and antibody KD lines or common suppressor mutations in two different abp1 knock-out alleles. Thus, more experimentation is needed to figure out what really happens in the different abp1 KD lines and how it is possible that they independently generate phenotypes that are so consistent. Whatever the explanation at the end will be, in light of the presented data it seems obvious that these lines do not act solely by down-regulating the ABP1 function, despite the accumulation of well-fitting data from independent and complementary approaches. It is a sobering realization that even when you use independent approaches with all standard controls performed, there is no real guarantee that the observations will not lead you amiss.