CRISPR/Cas9-Mediated Knockout of HOS1 Reveals Its Role in the Regulation of Secondary Metabolism in Arabidopsis thaliana

In Arabidopsis, the RING finger-containing E3 ubiquitin ligase HIGH EXPRESSION OF OSMOTICALLY RESPONSIVE GENES 1 (HOS1) functions as a main regulator of the cold signaling. In this study, CRISPR/Cas9-mediated targeted mutagenesis of the HOS1 gene in the first exon was performed. DNA sequencing showed that frameshift indels introduced by genome editing of HOS1 resulted in the appearance of premature stop codons, disrupting the open reading frame. Obtained hos1Cas9 mutant plants were compared with the SALK T-DNA insertion mutant, line hos1-3, in terms of their tolerance to abiotic stresses, accumulation of secondary metabolites and expression levels of genes participating in these processes. Upon exposure to cold stress, enhanced tolerance and expression of cold-responsive genes were observed in both hos1-3 and hos1Cas9 plants. The hos1 mutation caused changes in the synthesis of phytoalexins in transformed cells. The content of glucosinolates (GSLs) was down-regulated by 1.5-times, while flavonol glycosides were up-regulated by 1.2 to 4.2 times in transgenic plants. The transcript abundance of the corresponding MYB and bHLH transcription factors, which are responsible for the regulation of secondary metabolism in Arabidopsis, were also altered. Our data suggest a relationship between HOS1-regulated downstream signaling and phytoalexin biosynthesis.


Introduction
Clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) technology has become a versatile approach of targeted genetic manipulations in plants [1,2]. In contrast to the classical loss-of-function approach introducing random mutations, Cas9 protein driven by an engineered 20-nt RNA sequence (short guide RNA; sgRNA) performs strand scission of a specific target sequence within the genome, enabling precise corrections [3]. Mutations introduced by Cas9/sgRNA are inherited in the germline and have a classical pattern of Mendelian segregation in the next generations, as has been demonstrated in the transformants of Arabidopsis, rice and tomato [4][5][6]. Amino acid substitutions in HNH or RuvC nuclease domains of Cas9 result in gaining nickase activity of Cas9 variants, while mutations introduced in both domains simultaneously result in catalytically dead Cas9 (dCas9), lacking its endonuclease activity [7]. Nevertheless, dCas9 is still capable of binding to its target sites of DNA [7]. Cas9 nickase and dCas9, coupled with other functional enzymes, may be used as carriers to a certain region in the genome targeted by sgRNA. Using this technique, gene activation or repression, single nucleotide substitutions and epigenetic editing of chromatin status have been reported [8].
Exposure to adverse environmental conditions, including extreme temperatures, water deprivation, excessive soil salinity, high radiation and chemical pollution, can negatively impact plant quality, accumulation of biomass and yield [9,10]. Different strategies in

Transient Transformation of A. thaliana Seedlings
Transient transformation of Arabidopsis seedlings was performed by vacuum infiltration, in accordance with the protocol described by Reference [43]. The day before infiltration, A. tumefaciens harboring pPZP-RCS2-NPTII/pcoCas9/gRNA1/EGFP or pPZP-RCS2-NPTII/pcoCas9/gRNA2/EGFP constructions were picked from the plate, inoculated into 20 mL of LB medium with antibiotics and cultivated overnight at 28 • C with moderate shaking (150 rpm). The optical density (OD 600 ) of Agrobacterium suspension was measured using BioSpec-nano spectrophotometer (Shimadzu Corporation, Kyoto, Japan). Then overnight liquid culture was centrifuged at 4000× g for 10 min at room temperature (24 • C), the pellet was then resuspended in the appropriate volume of infiltration buffer (10 mM MES, pH 5.6, 10 mM MgCl 2 , 100 µM acetosyringone) to give a final OD 600 of 0.3. Three-four-day seedlings were immersed into a bacterial suspension placed in a 50-mL Falcon sterile tube and a short-term vacuum treatment was applied twice. Immediately after infiltration, the seedlings were washed three times with sterile water. Afterward, plants were transferred to half-strength W medium without antibiotics for cocultivation. After 3 days of cocultivation, the seedlings were washed in sterile distilled water and used for confocal GFP fluorescence analysis and DNA isolation.

Laser Confocal Imaging of EGFP in Living Cells
The detection of the GFP fluorescence in transformed seedlings was performed using the LSM 710 LIVE and LSM 510 META scanning confocal laser microscopes (Carl Zeiss, Jena, Germany) with an excitation wavelength of 488 nm and an emission filter of 505-530 nm. Non-transformed seedlings were used as negative control. To monitor emission maximum of GFP, spectral imaging in the 505-720 nm range with a 10.4 nm detection bandwidth was performed in all examined transformed plants. Wavelength scanning was employed to check the specificity of the GFP signal. Video files and single frames of the captured images were recorded and analyzed with LSM 510 Release version 4.2 and ZEN 2011 software. The absence of autofluorescence was evaluated using the λ-scan mode ( Figure S2).

Obtaining of Stable hos1 Cas9 Mutant Plants
The pPZP-RCS2-nptII/pcoCas9/gRNA1 and pPZP-RCS2-nptII/pcoCas9/gRNA2 constructs were transformed into A. thaliana by the floral dip method [44]. T1 plants were selected in Petri dishes on a half-strength of W medium containing 50 µg/L kanamycin. Antibiotic-resistant transformants were further transplanted on antibiotic-free medium in glass tubes for continuous growth, molecular analysis and seed harvesting. Homozygous T3 transgenic plants were used in this study. Plant growth conditions for transgenic plants were the same as for wild-type (WT) counterparts.

Genotyping and Sequencing of the HOS1 Gene Mutations
DNA extraction from leaves was performed using the cetyl trimethylammonium bromide (CTAB) protocol (Molecular Cloning, 3rd edition). Gene-specific primer pairs HOS1-1-target and HOS1-2-target (Supplementary Table S1) were used to amplify 381-and 372-bp long genomic regions containing the corresponding target sites, respectively. Obtained amplicons were ligated into a pJET (Thermo Fisher Scientific Inc., MA, USA) and sequenced using an ABI 3500 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).
To genotype mutations by a heteroduplex mobility assay (HMA), polyacrylamide gel electrophoresis (PAGE) analysis of amplicons subjected to denaturation/renaturation cycle was performed following the procedure of Reference [45]. To screen for mutations using high-resolution melting (HRM) analysis, we performed polymerase chain reaction (PCR)-HRM reactions with the same primer sets and 2.5 x SYBR green PCR master mix (Evrogen, Moscow, Russia) using CFX96 thermocycler (Bio-Rad Laboratories). Melting curves were determined by incubating the reaction mixes from 60 • C to 95 • C with an increment 0.2 • C for 10 s with a plate reading. Precision Melt Analysis software (Bio-Rad Laboratories) was utilized to discriminate between native and mutant HOS1 alleles.

Gene Expression Analysis
Total RNA from leaves of A. thaliana plants was isolated using LiCl protocol [46]. The technique of RNA analysis and first strand cDNA synthesis was described by us previously [47]. Quantitative PCR (qPCR) analysis was conducted using CFX96 thermocycler (Bio-Rad Laboratories) with 2.5 x SYBR green PCR mix (Evrogen) as described [48]. The primer sets used in the qPCR are listed in Table S1. Analysis for three biological replicates from three separate RNA extractions and three technical replicates for each qPCR experiment were performed. CFX Manager Software (Version 3.1; Bio-Rad Laboratories) was used for data processing. Melting curve analysis was conducted after each run to verify the absence of primer-dimer artefacts or non-specific amplicons. An expression-based heat map was created using web server Heatmapper [49].

HPLC-MS Analysis
Identification of secondary metabolites in dry aerial parts of 6-week-old control and transgenic plants was performed on a 1260 Infinity analytical high-performance liquid chromatography (HPLC) instrument (Agilent Technologies, Santa Clara, CA, USA) coupled with an electrospray ionization (ESI) ion trap mass spectrometer (MS) (Bruker HCT ultra PTM Discovery System, Bruker Daltonik GmbH, Bremen, Germany). Samples were separated into fractions on a Zorbax C18 column (Agilent Technologies). Kaempferol, quercetin, sinigrin hydrate and sulfatase (H-1 from Helix pomatia) standards were obtained from Sigma-Aldrich (St. Louis, MO, USA). Experimental procedure of sample preparation and glucosinolate desulfation, as well as chromatographic and mass spectrometric parameters, has been previously described in detail [48]. Quantitative analysis was performed with a photodiode array detector using commercially available reference standards. Agilent OpenLAB CDS software (v.01.06.111) was used for data processing. Representative MS spectra (acquired in negative ionization mode) were obtained for each sample to confirm target profiling of metabolites. The Bruker Daltonics Compass 1.3 esquire control software (v.6.2.581.3 and v.4.0.234.0) was used for collecting and processing MS data.

Statistical Analysis
Statistical tests were performed using Statistica 10.0 (StatSoft Inc., Tulsa, OK, USA) with the threshold for statistical significance set at p < 0.05. PAST 4.03 software was used for principal component analysis (PCA) of secondary metabolites detected by HPLC-MS [51]. Two independent categories were compared using the Student's t-test, while ANOVA analysis followed by a multiple comparison method was applied when comparisons among multiple groups have been performed. The inter-group comparison was made using Fisher's protected least significant difference (PLSD) post-hoc test.

Assessment of sgRNA Activity Using Transient Transformation
We constructed plant CRISPR/Cas9 all-in-one binary vectors harboring components for HOS1 mutagenesis, including the plant codon-optimized Cas9 (pcoCas9) and enhanced green fluorescent protein (EGFP), both controlled by the 35S CaMV promoter, a sgRNA expression cassette driven by the AtU6-1 promoter and the NPTII gene as a negative selectable marker ( Figure 1A,B). In these vectors, Cas9 contains nuclear localization signals (NLSs) at N-and C-termini, which ensure efficient nuclear targeting [40].
A. thaliana HOS1 gene (GenBank No. NM_129540) is located in chromosome 2 and is 5346 bp in size, with 10 exons. Two gRNAs targeting exon 1 and 2 of the HOS1 gene were designed ( Figure 1C,D), CRISPR-OFFinder was used for off-target scoring [52]; no potential off-targets were detected with 0 to 2 mismatches in the core sequence.
Pre-evaluation of genetic constructions in a particular plant is an important step for validation of genome engineering experiment. Although computational tools are widely used to predict gRNA efficiency based on the sequence analysis, the calculated on-target activity often does not match the results obtained in planta [53,54]. To test the effectiveness of CRISPR/Cas9, a system of transient expression of genetic constructs in protoplasts or plant seedlings is utilized for the screening of DNA mutations [55].
We performed Agrobacterium-mediated transient transformation of A. thaliana seedlings by vacuum infiltration method. Successful transformants were selected by detecting GFP fluorescence in leaves ( Figure 1E,F), which could potentially indicate that Cas9 and sgRNA cassettes were also expressed at the appropriate levels. To determine the mutation efficacy of two selected HOS1 gRNAs, we further conducted a heteroduplex mobility assay (HMA) using PAGE separation of heteroduplexes between the native allele of the gene and its mutant variant [45]. A total of 25 GFP-positive seedlings were screened for each of the gRNA construction and the typical PAGE results are depicted in Figure 1G,H. Heteroduplex bands were detected in PCR products of the gRNA1-and gRNA2-transformed seedlings, whereas a single band was observed in the wild-type sample, indicating that the CRISPR/Cas9mediated mutations were introduced at the target site of the HOS1 gene. According to the number of mutated bands, the efficiency for gRNA1 and gRNA2 in Arabidopsis was about 36% (9 mutant seedling) and 4% (1 mutant seedling), respectively. To verify the presence of mutant alleles of the HOS1 gene in transformed seedlings, genotyping by the high-resolution melting (HRM) analysis of the PCR products was also performed [56]. Figure 1I,J shows an example of such an analysis, according to which the PCR products of mutant plant lines (denoted with red, orange and green curves) differed from the PCR products of control plants (blue curves). These results suggested that the obtained vectors were suitable to be used for plant transformation. However, gRNA1 showed almost 10-fold higher effectiveness of genomic editing than gRNA2. Different levels of sgRNA activity is a common phenomenon observed during CRISPR/Cas9-mediated genome editing. For example, in wheat protoplasts treated with seven EPSPS-specific sgRNAs, the mean frequency of indels varied from 0.0 to 23.3% [57]. An intensive screening of 195 gRNAs in transiently transformed tomato leaves revealed that 49% of tested gRNAs showed 0-1% mutation efficiency [58]. It has been previously demonstrated that sgRNAs have similar editing efficiency in the transient assay and in stable transgenic plants [53].

Identification and Analysis of Stable Arabidopsis Mutants
Agrobacterium-mediated transformation of flower buds was performed to obtain A. thaliana transgenic plants [44]. Only the gRNA1 construct was used for transformation, since the gRNA2 activity was very low in the transient assay. T1 transformed plants were selected on media containing 50 µg/L kanamycin. Antibiotic-resistant seedlings were analyzed for the presence of the mutant HOS1 allele using automated Sanger sequencing of cloned target PCR fragments. 246 clones were sequenced from 12 T1 transgenic plants and six types of mutations were revealed: two insertions, three deletions and one transition ( Figure 2).  We should note that not all detected mutational events led to the inactivation of the HOS1 at the protein level (Table 1). In particular, the transition led to the replacement of alanine (A) with valine (V) and a deletion of 30 bp leads to the elimination of 10 aa without shifting the reading frame. Single nucleotide insertions and deletions of 35 and 71 bp resulted in a frameshift mutation producing premature termination codon. Only two homozygous mutants, hos1-1 Cas9 and hos1-2 Cas9 , were revealed by genotyping and both of them We should note that not all detected mutational events led to the inactivation of the HOS1 at the protein level (Table 1). In particular, the transition led to the replacement of alanine (A) with valine (V) and a deletion of 30 bp leads to the elimination of 10 aa without shifting the reading frame. Single nucleotide insertions and deletions of 35 bp and 71 bp resulted in a frameshift mutation producing premature termination codon. Only two homozygous mutants, hos1-1 Cas9 and hos1-2 Cas9 , were revealed by genotyping and both of them carried 1-bp insertion (Table 1).

Plant Phenotype and Expression Profiles of Flowering Control Related Genes
The HOS1 protein negatively regulates the CO protein levels thus modulating flowering time [34]. Consistent with previous results [35], the SALK hos1-3 plants initiated flowering at leaf number 9, while wild-type plants it was at number 14. Like previously established hos1 mutant alleles, the newly isolated hos1-1 Cas9 and hos1-2 Cas9 mutant alleles displayed early flowering at the same number of leaves as hos1-3 mutants ( Figure 3A). This result indicates that of hos1 Cas9 mutation did not provoke additional effect on the phenotype of transgenic plants. In accordance with previous observations [34], FLC transcriptional level was considerably decreased in hos1-3 plants ( Figure 3B). On the contrary, the transcription of flowering integrator genes, FT and TSF, was up-regulated in the hos1-3 line. Our hos1-1 Cas9 and hos1-2 Cas9 mutant plants showed similar patterns of FLC, FT and TSF expression ( Figure 3B). Activation of FT was slightly less effective in hos1-1 Cas9 and hos1-2 Cas9 plants, compared with the hos1-3 line. TSF was activated slightly stronger in the hos1-2 Cas9 plants, compared to hos1-1 Cas9 and hos1-3 mutants. It has been previously proposed that HOS1 could potentially mediate the crosstalk between photoperiod and temperature regulatory factors of flowering, thus facilitating plant adaptation to environmental signals [34,35].

Abiotic Stress Treatment and Expression of CBF Regulon Genes
Hos1 mutants exhibited transcriptional activation of cold-responsive genes and higher tolerance to cold stress [28]. In order to evaluate if Cas9-mediated HOS1 mutations affect the cold signaling, we subjected hos1-3, hos1-1 Cas9 and hos1-2 Cas9 mutants to continuous cold stress at 12 • C exposure for 2 weeks ( Figure 4A). All hos1 mutants showed similar flowering time under normal and cold conditions (at leaf number 12), while the flowering time of Col-0 plants was considerably delayed (at leaf number 23). There was no visible difference in hos1 plants grown in high NaCl concentration and without watering compared with control plants ( Figure 4A). However, the seed germination test showed that hos1-3 and hos1 Cas9 plants had higher seed germination rates under cold stress treatment (12 • C) on medium containing NaCl (300 mM) and mannitol (300 mM), compared to the corresponding control conditions ( Figure 4B).
The CBF/DREB1 regulon, which controls the transcription of hundreds of genes activated by cold, is known to be one of the major stress-signaling pathways affecting cold tolerance [59]. The CBF/DREB1 genes are controlled partially by ICE1, which was considered to be the main regulator during cold acclimation in A. thaliana [60]. Transgenic plants with increased expression of ICE1 and CBF genes showed considerable resistance to cold temperatures [61].
transcription of flowering integrator genes, FT and TSF, was up-regulated in the hos1-3 line. Our hos1-1 Cas9 and hos1-2 Cas9 mutant plants showed similar patterns of FLC, FT and TSF expression ( Figure 3B). Activation of FT was slightly less effective in hos1-1 Cas9 and hos1-2 Cas9 plants, compared with the hos1-3 line. TSF was activated slightly stronger in the hos1-2 Cas9 plants, compared to hos1-1 Cas9 and hos1-3 mutants. It has been previously proposed that HOS1 could potentially mediate the crosstalk between photoperiod and temperature regulatory factors of flowering, thus facilitating plant adaptation to environmental signals [34,35]. We studied the expression of key components of the CBF/DREB1 regulon in hos1-3, hos1-1 Cas9 and hos1-2 Cas9 mutants subjected to continuous 12 • C exposure for 2 weeks ( Figure 4C). The expression of ICE1 in the hos1 mutant plants was almost equal to control. The hos1 mutation caused significant activation of the CBF2 and CBF3 genes, while the level of CBF1 expression was activated to a lesser extent. As previously reported, no alterations in HOS1 transcript abundance was detected at ambient temperatures and the expression of CBF genes was not considerably affected by altered HOS1 activity without cold treatment [28].  When exposed to cold stress, hos1-3 and hos1-1 Cas9 plants exhibited strong transcriptional activation of the cold-responsive genes, with the exception of ICE1 ( Figure 4C). Among the CBF genes, CBF2 and CBF3 were up-regulated in hos1 mutant lines up to 8-12 times compared with the control plant. The transcriptional abundance of COR genes was also considerably higher in both hos1-3 and hos1 Cas9 plants: their expression was activated by 4 to 5-fold.
COR proteins play a major role in physiological rearrangements, leading to an increase of cold resistance. About 4000 COR genes are known in Arabidopsis [62], while factors CBF1, CBF2 and CBF3 regulate only about 10% of them [63]. Of the CBF-regulated COR genes of Arabidopsis, the most studied are COR6.6, COR15a, COR15b, COR47 and COR78 [64]. We examined the expression of these genes in Cas9-edited and SALK mutant Arabidopsis plants ( Figure 4C). The levels of transcription of most of the COR genes in the studied lines were up-regulated compared to control. Expression of such genes as the COR6.6, COR15a and COR78 in hos1 mutant plants exceeded that of the control plant by almost 3-fold.

Secondary Metabolites Content and Expression of the bHLH and MYB Transcription Factors in hos1-3 and hos1 Cas9 Mutants
HPLC analysis revealed that the hos1 mutant Arabidopsis plants showed remarkably altered secondary metabolite profiles ( Table 2). We found that the levels of total flavonoids were up-regulated by 1.2 to 4.2 fold in hos1 plants and this effect was mainly due to the rise of kaempferol derivatives content. Principal component analysis based on the content of the flavonol glycosides identified two significant components; PC 1 explained 97.4% of the variance, while PC 2 had a variation of 1.9%. The loading plot of PC 1 was dominated by kaempferol hexose dideoxyhexose, kaempferol hexose deoxyhexose and kaempferol dideoxyhexose. PC 2 was mainly contributed by kaempferol hexose deoxyhexose. The content of glucosinolates (GSLs) in the mutant lines was reduced up to 1.5 times (Table 2). At the same time, clear patterns of change in certain metabolites of this group in hos1 mutant lines were not obvious. The strongest decrease was noted for aliphatic GSLs (methyl-thioalkyl and methyl-sulfinylalkyl GSLs), however, the reason for this selective action is not clear. Principal component analysis based on the content of glucosinolates identified three significant components; PC 1 had a variation of 52.1%, while PC 2 explained 40.4% of the variance and PC 3 had 7.5%. The loading plot of PC 1 was mainly contributed by 3-methylsulfinylpropyl glucosinolate, 4-methylsulfinylbutyl glucosinolate and 5-methylsulfinylpentyl glucosinolate. PC 2 was dominated by 4-methylthiobutyl glucosinolate, 8-methylthiooctyl glucosinolate and 8-methylsulfinyloctyl glucosinolate.
We next examined the level of expression of the bHLH and MYB transcription factors that regulate secondary metabolism in Arabidopsis ( Figure 5C). The expression of the gene encoding the flavonoid-specific transcription factor MYB12 was increased by 3 to 6 times the hos1-3 and hos1 Cas9 mutant lines, compared with the control line ( Figure 5A). Expression of MYB11 in hos1-3 and hos1 Cas9 mutant lines did not differ from the control plants. It has been reported that enhanced expression of MYB12 in Arabidopsis led to increase in flavonoids production [65]. Probably, activation of flavonoid biosynthesis in hos1 mutants was caused by the up-regulation of MYB12. The expression analysis of the MYB-bHLH-WD40 components showed that TTG1, an important regulator of late structural genes in biosynthesis of flavonoids [66], was expressed in a similar manner in WT, hos1-3 and hos1 Cas9 plants ( Figure 5A). Expression of the transcription factor genes TT8 and PAP1 (synonym: MYB75) was strongly activated in hos1-3 and hos1 Cas9 mutants. It was shown that TT2 and PAP1 overexpression is sufficient for late flavonoid gene activation in A. thaliana protoplasts with comparatively low levels of TTG1 [67,68]. Similar results were obtained in A. thaliana seedlings with knockout of the MYBL2 gene, which encodes an R3-MYB-related protein. Two independent mybl2 mutants showed enhanced accumulation of anthocyanins and this effect was accompanied by the increase in transcriptional activity of TT8 and PAP1 genes, while TTG1 was not affected [69]. Therefore, our results indicate that transcriptional activity of MYB12, TT8 and PAP1 in hos1 mutants was sufficient to mediate overproduction of flavonoids.  1 Aerial part of 6-week-old control and transgenic plants from different passages were analyzed using tree biological replicates and two technical replicates. Data are presented as mean ± standard error. Different superscript letters indicate statistically significant differences of means (P < 0.05) in the columns, Fisher's LSD. tr, trace amounts (below 0.01 µmol/g of dry weight).
The expression of aliphatic GSLs related genes, MYB28 and MYB76, was found to be considerably decreased in hos1 mutants compared to the control ( Figure 5B). MYB28 and MYB29, together with MYB76, are the main regulators of aliphatic GSLs biosynthesis in Arabidopsis. Among these genes, MYB28 and MYB29 are equally important for activation of GSLs biosynthesis genes, because myb28myb29 double mutation completely abolished accumulation of aliphatic GSLs and the expression of MYB76 alone did not neutralize the loss of their function [70]. However, genes involved in the production of indolic GSLs responded differently. In the hos1-3 and hos1 Cas9 mutants, expression of MYB34 and MYB122 decreased, whereas the expression of MYB51 was generally stable ( Figure 5B). It is known that MYB34 and MYB51 play pivotal roles in the biosynthesis of indolic GSLs in A. thaliana, while MYB122 has an accessory role and its function becomes evident during specific environmental stimuli [71]. Thus, the expression pattern of TFs regulating the biosynthetic pathways of aliphatic and indolic GSLs in the hos1 mutant plants is consistent with the content of secondary metabolites in them.
Plants 2021, 10, x FOR PEER REVIEW 14 of 20 specific environmental stimuli [71]. Thus, the expression pattern of TFs regulating the biosynthetic pathways of aliphatic and indolic GSLs in the hos1 mutant plants is consistent with the content of secondary metabolites in them. Although the effect of hos1-3 and hos1 Cas9 mutations on flowering time and abiotic stress responses was equal, secondary metabolism was modified differently in these lines. In particular, accumulation of flavonol glycosides was mainly increased in hos1 Cas9 mutants, while inhibition of indolic glucosinolates level was less pronounced in hos1-3 plants ( Table 2). The hos1-3 mutation is located in the 5th exon and disrupts HOS1 reading frame at position 912 bp of mRNA leading to the production of a truncated protein which, however, retains 304 aa of a native form ( Figure S4). This protein lacks an important C-terminal portion of the polypeptide including protein-protein interaction domain and a nuclear localization signal but still contains the intact RING finger E3 ligase domain located at aa 34-100. It has been previously noted that different hos1 variants exhibit pleiotropic effects on many processes of plant physiology [72]. Moreover, in vitro ubiquitination assay revealed that activity of a truncated HOS1 having 1-550 aa was similar to native protein, while another mutant form having only 1-210 aa was not able to generate ubiquitinated products [28]. Currently there is no information about potential ubiquitin ligase activity Although the effect of hos1-3 and hos1 Cas9 mutations on flowering time and abiotic stress responses was equal, secondary metabolism was modified differently in these lines. In particular, accumulation of flavonol glycosides was mainly increased in hos1 Cas9 mutants, while inhibition of indolic glucosinolates level was less pronounced in hos1-3 plants ( Table  2). The hos1-3 mutation is located in the 5th exon and disrupts HOS1 reading frame at position 912 bp of mRNA leading to the production of a truncated protein which, however, retains 304 aa of a native form ( Figure S4). This protein lacks an important C-terminal portion of the polypeptide including protein-protein interaction domain and a nuclear localization signal but still contains the intact RING finger E3 ligase domain located at aa 34-100. It has been previously noted that different hos1 variants exhibit pleiotropic effects on many processes of plant physiology [72]. Moreover, in vitro ubiquitination assay revealed that activity of a truncated HOS1 having 1-550 aa was similar to native protein, while another mutant form having only 1-210 aa was not able to generate ubiquitinated products [28]. Currently there is no information about potential ubiquitin ligase activity of a 1-304 aa truncated HOS1 in hos1-3 Arabidopsis mutant line. However we cannot fully exclude that it may possess some nonspecific function thus conferring changes in biosynthesis of secondary metabolites different from those of hos1 Cas9 lines.
The molecular mechanism by which loss of HOS1 function mutation cause transcriptional perturbations of the bHLH and MYB TFs and subsequent secondary metabolite perturbations in hos1-3 and hos1 Cas9 mutants is not clear yet. HOS1 has a pivotal role in light and cold signaling and thus, its effect on phytoalexin biosynthesis may be attributed to these two main functions. Recently FLC was found to mediate the integration between flowering time regulation and biosynthesis of GSLs in Arabidopsis [73]. More specifically, FLC knockout mutation led to increased transcript levels of MYB15, while MYB29 was down-regulated. The presence of sulfate transporter SULTR2;1 and FLC loci in two Aethionema arabicum ecotypes (CYP and TUR) is responsible for up to 10-fold difference in GSL concentrations, with the earlier-flowering CYP genotype always have the higher aliphatic GSLs concentration [74]. In addition, direct regulation of Arabidopsis indolic GSL biosynthetic gene, CYP79B3, is also possible because FLC binds to its promoter region [75].
Another possible mechanism can be realized in hos1 mutants via participation of HOS1 in the modulation of CBF regulon function. In Arabidopsis, HOS1 triggers ubiquitination ICE1 [28] and this degradation is inhibited by hos1-3 and hos1 Cas9 mutation. Cold treatment facilitates the binding of ICE1 to CBF3 promoter to induce its expression and at the same time induces the degradation of ICE1 by HOS1 [28]. CBF3, in turn, activates GA 20-oxidases leading to a reduction of the gibberellin level and provokes the accumulation of DELLA proteins [76]. Consequently, DELLAs regulate jasmonate pathway signaling leading to the releasing of MYC2 transcriptional activity by interaction with JASMONATE ZIM-domain proteins [77]. Finally, MYC2, a key regulator of jasmonatemediated responses, positively regulates the transcription of flavonoid biosynthetic genes and negatively regulates biosynthesis of indole GSLs [78]. Moreover, ICE1 physically interacts with ABSCISIC ACID INSENSITIVE5 (ABI5) to antagonize its transcriptional function [79]. ABI5 is the central regulator of abscisic acid (ABA) functions, participating in diverse abiotic stress responses, including biosynthesis of phytoalexins [80]. In particular, ABI5 regulates the transcriptional activity of MYB28 and MYB29, pivotal transcriptional regulators of aliphatic GSLs biosynthesis [81]. ABI5 also participates in indolic GSLs metabolism by modulating the transcriptional activity of cytochrome P450 enzymes, CYP79B2 and CYP83B1, while having no effect on MYBs [82]. Moreover, ABI5 is implicated in the biosynthesis of flavonoids/anthocyanins as a positive regulator of PAP1 and PAP2 expression [83]. Similarly, in tomato, Anthocyanins11-mediated induction of secondary metabolites accumulation involves ABI5 gene up-regulation [84]. It is noteworthy that DELLAs suppress the transcriptional function of ICE1 and its antagonistic effect on ABI5 to fine-tune ABA signaling [79]. Thus, several pathways can be affected either simultaneously or sequentially in hos1 mutants. Besides, their interaction and intersection are also possible. In this regard, further research is needed to better understand the role of HOS1 in the regulatory relationships among light signaling, cold stress responses and of secondary metabolism.

Conclusions
In this report, Cas9-mediated genomic loss-of-function mutations of the A. thaliana HOS1 genes were obtained for the first time. Several reading frame shifts occurred in the first exon resulting in premature termination of translation at codon positions 11, 23 and 24. Mutant plants, carrying hos1 allele generated via CRISPR/Cas9 chemistry (hos1 Cas9 ) and traditional T-DNA insertion mutagenesis (hos1-3) showed similar phenotypes, physiological and molecular responses to abiotic stress stimuli. Significant perturbations in phytoalexin biosynthesis have been detected, resulting in a reduced amount of GSLs and an elevated content of flavonoids. Moreover, hos1 mutations provoked significant changes in expression of the corresponding bHLH and MYB transcription factors. To our knowledge this is the first evidence of HOS1 impact in the context of secondary metabolism. The results suggest the interplay between HOS1-mediated light signaling, responses to cold and the regulation of secondary metabolism in Arabidopsis. contained protospacer sequences for both gRNAs. gRNA1 and gRNA2 were obtained using HiScribe T7 Quick High-Yield RNA synthesis Kit (NEB, Ipswich, MA, USA), mixed with Cas9 nuclease (NEB) and HOS1 DNA target. The mix was incubated at 37 • C for 15 min following fragment analysis using electrophoresis in 2% agarose gel, Figure S2: Visualization of GFP in A. thaliana seedlings subjected to Agrobacterium-mediated transformation using vacuum infiltration. Scanning by wavelengths (λ-scanning) to check the specificity of the fluorescence emission signal, Figure S3: Stability analysis of the candidate reference genes in A. thaliana calculated by RefFinder. Values above the bars indicate geomean of ranking values, Figure S4: Schematic illustration of the T-DNA insertion in hos1-3 mutant line of Arabidopsis thaliana (SALK_069312C) purchased from the collection of the Salk Institute. The detail underneath shows partial sequence of the mutated HOS1 allele obtained from the DNA template of hos1-3 plant using PCR with LP and LBb primer pair. Intronic region is marked with blue color, exon is yellow-colored and T-DNA region is highlighted in grey. The 29-bp insertion of unknown origin was also found between the exon and the T-DNA sequences. T-DNA left border (LB) as well as LP and LBb primers binding sites are underlined. The predicted sequence of the truncated protein product is also shown below. The portion of the native HOS1 is italicized, Table S1: List of primer sequences used in this study.