Novel CRISPR/Cas9-Derived mlo Alleles in Barley: Resistance to Powdery Mildew and Microbiome Implications

TL;DR

Scientists used CRISPR to create barley resistant to powdery mildew and found limited effects on root microbes.

Contribution

Generated first mlo mutants with extracellular amino acid changes and showed limited microbiome impact.

Findings

Five new mlo mutant barley lines showed 0.5–5% B. hordei colony formation.

Microbial alpha and beta diversity remained largely unchanged in mutant lines.

Some microbial taxa showed time-dependent abundance shifts in mutant plants.

Abstract

Barley grown in temperate regions is often challenged by powdery mildew disease. An effective solution is mildew resistance locus o (mlo)-based resistance, which is monogenic, durable, and broad-spectrum. While the pleiotropic effects of mlo mutations on above-ground tissues are well documented, their impact on the root-associated microbiome remains underexplored. We utilized CRISPR/Cas9 to generate novel mlo mutant lines and evaluated their resistance to causal fungus Blumeria hordei. We further examined if mlo knockout has any impact on the overall root microbiome diversity and composition under field-like conditions and applied DESeq2 to compare the abundance of microbial taxa between mutants and wild type. We created five novel resistant mlo lines, including the first mutants with amino acid alterations in the protein’s extracellular region. Mutant lines showed significantly reduced B. hordei colony formation (0.5–5%). While microbial alpha and beta diversity were not significantly altered, a few microbial taxa displayed time-dependent shifts in abundance. Overall, our study demonstrates the effectiveness of CRISPR/Cas9 in generating mlo-based resistance. Moreover, the study revealed functionally important residues in the protein’s extracellular region. Finally, we present the first evidence of limited mlo-associated effects on root microbiome diversity and relative abundance of microbial taxa.

Figures

Full text

1. Introduction

Powdery mildew is a common fungal disease in temperate climates, reducing crop yields and grain quality [1,2,3]. The fungus responsible for this disease in barley is the ascomycete obligate biotrophic pathogen Blumeria hordei (Bh), which is one of the most frequent barley pathogens [4,5]. Fortunately for barley farmers, an effective resistance to powdery mildew was discovered in the 1940s. It is attributed to recessive loss-of-function mutations in the Mlo (Mildew Resistance Locus O) and has been widely incorporated into barley breeding programs [6]. This resistance, which is widely prevalent in monocot and dicot plant species, is particularly attractive because it is durable and effective against most fungal isolates [2]. In general, immunity to powdery mildew is manifested at the stage of initial fungal penetration, by local cell-wall appositions known as papillae, and as programmed cell death response, referred to as the hypersensitive reaction (HR), of one or a few plant cells at the attack site. The collective action of these immunities determines the level of resistance of the plant [7]. mlo-based resistance is primarily characterized by unusually large papillae, which mediate early stop of the fungal attack [6,8].

More than 40 distinct mlo mutant alleles have been identified in barley, arising either naturally or through chemical and radiation-induced mutagenesis [2,9]. However, since mlo-based resistance represents a monogenic trait in barley, targeted gene-specific mutations can now be precisely recreated using New Genomic Techniques (NGTs) such as CRISPR/Cas, which has already been successfully applied in barley [10]. Unlike traditional mutagenesis, which introduces numerous background mutations, gene-specific approaches enable the direct assessment of mlo-mediated resistance and its effects such as those on the plant microsphere with unprecedented precision.

The Mlo gene codes for an integral membrane protein with seven transmembrane domains, with yet poorly understood molecular function [11,12,13]. Different protein parts and amino acid residues have been described as highly conserved, or variable, and more or less critical for the powdery mildew resistance [9,11,12]. Overall, cytoplasmatic loops 2 and 3, transmembrane proline, and four extracellular cysteine residues appear to be particularly sensitive to mutations [9,12]. Additionally, the majority of molecularly characterized barley mlo mutants have mutations in the cytoplasmatic parts of the protein, suggested to be the most crucial for powdery mildew resistance [9].

While mutations in the Mlo gene provide notable benefits in terms of disease resistance, they also come with specific adverse pleiotropic effects. Barley mlo mutants exhibit spontaneous mesophyll cell death, loss of chlorophyll, and accelerated leaf senescence, which leads to visible chlorotic and necrotic spots [14,15]. Moreover, mlo mutants show increased susceptibility to certain necrotrophic or hemibiotrophic pathogens, such as Magnaporthe grisea [16], Ramularia collo-cygni [17] and Bipolaris sorokiniana [18].

Furthermore, the pleiotropic effects of mlo mutations extend to below-ground plant parts, although these effects are significantly less explored. Studies have shown that mlo mutations can influence the colonization of root-associated microbes, particularly arbuscular mycorrhizal fungi (AMF) and endophytes, in inoculation experiments [19,20,21]. However, these experiments typically involve single microbial strains and are conducted in sterile conditions or germ-free substrates, lacking the complexity of natural soil microbiota, microbial interactions, and environmental variability. Such insights into microbiome stability are important, as the contribution of the microbiome to plant health, productivity, and stress resilience has become more evident in recent years [22,23,24,25]. Consequently, it is important to better understand if the mlo mutation affects the microbiome community as a whole within the natural soil environments.

This study aimed to develop clean, mlo-specific barley mutants using CRISPR/Cas9 technology and to validate resistance at both the macroscopic and microscopic levels. In addition, we investigated the impact of the mlo knockout mutation on root-associated microbial communities under natural, field-like growth conditions. Finally, we assessed whether the mlo mutation influences the relative abundance of specific bacterial and fungal taxa in the rhizosphere.

2. Results

2.1. Generating New CRISPR/Cas9-Induced mlo Mutants

We used A. tumefaciens with the final transformation construct (Figure 1) to transform cv. ‘Maythorpe’ immature embryos. The protospacer designed showed no obvious off-targets based on analysis using the CasOFFinder tool 2.4 (Supplementary Materials Figure S1). In total, 38 regenerated plants were obtained. They were PCR-screened for the presence of the hygromycin resistance gene, and 30 tested positive (Supplementary Materials). Mutations in the mlo gene were confirmed in all 30 plants by PCR amplification and sequencing of the target region within exon 2. Various indel mutations were detected, resulting in homozygous, heterozygous, biallelic, and chimeric genotypes.

Six T_0_ plants representing different genotypes were selected to generate the T_1_ generation. Their seeds were sown, and T_1_ offspring were genotyped by PCR amplification and sequencing of the targeted mlo region. Based on the results, five distinct homozygous mutant lines were selected for subsequent resistance screening (Figure 1). The lines will further be labeled as mlo_-3-3_, mlo_-6-6_, mlo_-24-24_, mlo_-5-5_, and mlo_-14-14_, based on the mutation they are carrying.

For the chosen five genotypes, protein sequences were predicted (Table 1; Supplementary Materials Figure S2). Three lines had in-frame mutations (mlo_-3-3_, mlo_-6-6_, and mlo_-24-24_) that deleted one (Tyr113del), two (Asp112_Tyr113del), and eight (Pro107_Cys114del) amino acid sequences from the protein, respectively. The rest of the protein was intact. Based on the instability index (II), the protein stability was unaffected (38.94, 39.28, and 38.88, respectively, compared to the wild type 38.88). The other two lines (mlo_-5-5_ and mlo_-14-14_) carried frameshift mutations, leading to a premature stop codon at the beginning of exon 5. The amino acid sequences were disrupted downstream of Asp112 and Tyr110, and encoded polypeptides were 173 and 170 aa long, respectively.

Three-dimensional (3D) modeling using the AlphaFold3 algorithm confirmed that for the mutant lines with early stop codon only, a short peptide is formed (Supplementary Materials), while the other mutations still could result in a protein comparable to the wild type, with minor changes to the intracellular fraction (Supplementary Materials). The short peptide model was expected by the impact of the frameshift mutation in the two lines mlo_-5-5_ and mlo_-14-14_ (Table 1). The three in-frame mutations had little effect on the protein model, only influencing the intracellular loop part of the 3D model.

2.2. CRISPR-Induced mlo Mutants Provided Powdery Mildew Resistance

To assess resistance in the T_2_ generation, five selected mutant lines and wild-type cv. ‘Maythorpe’ were inoculated with Bh using the settling tower method. Estimated spore density was four to six spores per square millimeter. Macroscopic resistance was visually scored 7 dpi.

None of the five mutant lines exhibited fungal growth, indicating complete resistance, while wild-type cv. ‘Maythorpe’ showed extensive powdery mildew symptoms (Figure 2). No difference in resistance was observed between mutant lines.

Microscopic analysis was conducted to assess resistance further. One hundred fungal penetration attempts were evaluated per replica, and averages were compared between mutant lines and the wild-type cv. ‘Maythorpe.’ The wild type exhibited a significantly higher number of successful penetration attempts compared to all mutant lines (p < 0.001), with an average of 62.5% (Figure 3). Among the mutants, the frameshift mutation lines mlo_-5-5_ and mlo_-14-14_ displayed the lowest frequencies of penetrating appressoria, with means of 22.5% and 18.5%, respectively. However, this did not differ significantly from in-frame mutant line mlo_-3-3_, which exhibited 24% successful penetration events. Lines mlo_-6-6_ and mlo_-24-24_ exhibited the highest penetration level among the mutants, with mean penetration counts of 32.5% and 28.75%, respectively. However, mlo_-24-24_ did not differ significantly from mlo_-5-5_ and mlo_-3-3._

The mean number of HRs in the wild type occurred in 9% of the attacked epidermal cells, which was significantly lower compared to all mutant lines (p < 0.001). Among mutants, mlo_-6-6_, mlo_-24-24_, and mlo_-5-5_ exhibited the highest average numbers of HRs, with 29.5%, 27%, and 22%, respectively. Nonetheless, the level of HR in mlo_-5-5_ was not significantly different from those observed in lines mlo_-3-3_ and mlo_-14-14_, which displayed 19 and 16 HRs, respectively (Figure 3).

Finally, the number of fungal colonies formed was significantly greater in the wild-type ‘Maythorpe’ compared to all mutant lines (p < 0.001). Wild type had mean of 53.5 colonies counted. On the other hand, the most successful colony establishments among mutants were observed in in-frame mlo_-3-3_ and mlo_-6-6_, with 5 and 3 colonies, respectively. These values were not significantly different from those of mlo_-14-14_ and mlo_-24-24_, which exhibited 2.5 and 1.75 colonies, respectively. The most pronounced resistance was seen in mlo_-5-5_, where only 0.5 colonies were formed (Figure 3).

In summary, these microscopy data shows that the obtained mlo-based resistance in all our CRISPR-/Cas9 lines is due to a combination of reduced penetration and HR.

2.3. Mlo Knockout Did Not Alter Overall Microbial Diversity and Only Affected a Few Specific Taxa

The knockout mutant mlo_-14-14_ and wild-type cv ‘Maythorpe’ were used in the field-like experiment to assess whether knocking out mlo influences the root-associated microbial community. In particular, we evaluated alpha and beta diversity metrics for bacterial and fungal communities at two time points (T1 and T2). No significant differences in alpha diversity, described by richness or evenness, were observed between the two genotypes in bacterial or fungal communities at any of the sampling times (Figure 4a,b). Moreover, beta diversity, assessed via principal coordinate analysis (PCoA) based on Bray–Curtis dissimilarity, revealed no distinct separation between the mutant and wild type at any sampling time regarding bacterial or fungal community composition (Figure 4c,d). PERMANOVA and ANOSIM further confirmed no significant differences (p > 0.05, R ≈ 0) in overall microbial community composition between the mutant and wild type at any time.

To study if the relative abundances of specific ASVs/OTUs were different between the mutant line and wild type, we performed DESeq2 analysis. We identified a few bacterial ASVs and fungal OTUs that were significantly enriched (P_adj_ < 0.05, Log^2^Fold change > 4) in one of the two lines at different sampling time points. At time point 1, two bacterial ASVs (i.e., ASV1356 and ASV1266), identified as Allorhizobium-Neorhizobium-Rhizobium group, and one fungal OTU (i.e., OTU_117), classified as Rhizopus, were enriched in the wild type. These were absent in the rhizosphere of the mutant line (Figure 5).

At time point 2, four bacterial ASVs (Allorhizobium-Neorhizobium-Rhizobium group ASV1264, Ornithinibacillus ASV365, Bacillus ASV 321, and Flavobacterium ASV725) exhibited significantly higher abundance in the rhizosphere of the mutant line. In contrast, one bacterial ASV, ASV818, classified as Promicromonospora, was enriched in the wild type. For fungi, three OTUs (i.e., Mortierellaceae OTU_449, Auriculariales OTU_22, and Pyrenochaetopsis OTU_405) were significantly more abundant in the rhizosphere of the wild type.

3. Discussion

3.1. Novel mlo Alleles with Mutations in Extracellular Loop 1 Confer Full Resistance to Blumeria hordei

Loss of function mutations in the mlo gene are known to provide durable, broad resistance to powdery mildew in barley [6,26]. Since we aimed to knock out the gene and recreate the loss of function phenotype, we designed the protospacer to target a sequence early in the gene, in exon 2, corresponding to the extracellular loop 1 part of the Mlo protein [11,27]. To date, no known barley mlo mutant carries a mutation in this position [9,27]. Our transformation pipeline was successful, and we proceeded with experiments using two lines with loss-of-function, frameshift mutations and three with in-frame mutations. The resistance scoring on macroscopic and microscopic levels confirmed the loss of function in all cases.

To analyze in-frame mutants, diving deeper into protein structure and mutation position is important. As mentioned, we generated mutations in the extracellular loop 1, a region described as highly variable and generally considered nonessential for the protein function [11,12]. In particular, mlo_-3-3_, mlo_-6-6_, and mlo_-24-24_ had one (Tyr113del), two (Asp112_Tyr113del), and eight (Pro107_Cys114del) amino acid residues deleted. As for the extracellular loop 1, amino acid residues 107 to 113 have been reported to be highly variable in the Mlo gene family [11,12]. Although extracellular loops have been described as highly variable and less critical for the function in powdery mildew resistance, four Cysteine (Cys) residues in those loops were described as invariable [11,12]. It is proposed that they have a role in disulfide bridge formation, which is vital for protein conformational stability, and even Mlo protein dimer/oligomerization, which is important for the protein function [11,12]. Among four invariable Cys residues, one was on the aa position 114, precisely where we targeted. Elliott, Müller [12] delivered the Mlo Cys114A variant through biolistic transformation into the mlo resistant genotype and showed no complementation and resistance preservation, emphasizing the significance of this residue for protein function. On the other hand, the wild-type, susceptible phenotype was successfully restored when the same complementation approach was applied to variable residues 103 and 105 [12].

Going back to our mutants, the invariable Cys114 residue was deleted in line mlo_-24-_24. Considering its vital role in powdery mildew resistance, we anticipated loss of function in that line. However, all our in-frame mutants showed complete resistance, raising the question about the role of highly variable residues deleted in lines mlo_-3-3_ and mlo_-6-6_. They have one or two aa residue(s) right upstream of the Cys114 deleted. Although those are known to be variable [12], we hypothesize that their deletion could affect the Cys position and conformation and impair its ability to form essential disulfide bridges or even oligomers. This would consequently lead to loss of function and a resistant phenotype, as observed. A study by Petersen, Jonson [28] reported on the importance and preference of Cys involved in disulfide bridges towards particular aa as neighbors, suggesting that this could be the mechanism. When comparing the 3D modeling results of our in-frame mutants with the wild-type protein, we can see that changes only occur in the intracellular loop (Supplementary Data). These changes could have an effect on the intracellular loop stability or tethering, leading to impaired disulfide bridges in our in-frame mutation. This is backed by other research, showing that cysteine replacement impairs protein function [29]. Therefore, deeper research on protein structure and biochemistry is necessary to understand the potential influence of mentioned aa residues on protein conformation, folding, disulfide bridge formation, and oligomerization.

As mentioned, all our mutants showed strong resistance to powdery mildew, with penetration success rates varying from 5% to 0.5%. There was no systematic difference between in-frame and frameshift mutant lines in the number of unsuccessful penetration attempts, HRs, or number of colonies. The in-frame mlo_-3-3_ and mlo_-6-6_ lines exhibited the highest number of colonies (5 and 3 per 100 attempts, respectively). However, the result was not significantly different from that observed for mlo_-24-24_ and mlo_-14-14_.

It is noteworthy that our mutants also show no typical symptoms of other barley mlo mutants. We could not detect any necrotic area under the Bh challenged sites at 7 dpi (Figure 2). While other mlo mutants, like mlo-5 or mlo-28, show these typical necrotic regions, our mutants are in barley the first mutants showing strong resistance without this side-effect [15]. There are reports of one allele from an Ethiopian landrace, showing a tempered resistance to powdery mildew without necrotic lesions, indicating the importance of location for the mutation [30]. There are similar reports from wheat, where usually the same phenotypic effect is observed [31], but novel CRISPR/Cas9-derived alleles show similar effects like our barley mutants [32].

We reported new CRISPR-induced mlo alleles that confer strong resistance to powdery mildew and the first-ever resistant mlo mutants in the extracellular part of the protein. Our study also highlights the potential of using CRISPR/Cas technology in breeding programs for the rapid and straightforward incorporation of valuable traits, such as disease resistance. Furthermore, we provided new insights into protein parts and amino acid residues essential for powdery mildew resistance, paving the way for further research on the topic.

3.2. mlo Knockout Preserves Overall Microbial Diversity but Influences a Few Rhizosphere Taxa

Several studies have examined the effects of the mlo mutation on below-ground plant–microbe interactions, specifically regarding its role in endophyte and mycorrhizal colonization [19,20,21]. However, no study has investigated how the mlo loss of function impacts the entire microbial community in the roots and rhizosphere. To address this knowledge gap, we used amplicon sequencing for analyzing microbial diversity, as well as specific taxa enrichment in mlo_-14-14_ and wild-type plants under field-like conditions.

We compared alpha and beta diversity between mlo_-14-14_ and wild type at two time points. Our results suggest no significant influence of the mlo loss of function mutation on either alpha or beta diversity at any time point. It is possible that the Mlo function is localized primarily in above-ground parts, while roots may have other genes playing a more dominant role in immune responses and plant–microbe interactions. Although Jacott, Charpentier [19] reported Mlo expression pattern changes during the interaction with beneficial root microbes, Le Fevre, O’Boyle [33] discovered that the role of Mlo in Phytophthora palmivora colonization was limited to leaves and did not have any detectable effect in roots, suggesting that its influence is exclusive for foliar tissue. Additionally, the root microbiome is known to be shaped by multiple factors, such as soil type, plant development stage, plant genotype, and secondary metabolites [34,35,36,37,38,39]. Hence, it is plausible that a loss of function mutation in this gene may be insufficient to override the mentioned dominant factors.

Recently, the plant microbiome has emerged as a key factor in sustainable agriculture. Microbiome management could soon be integrated into breeding programs, allowing breeders to select genotypes that recruit beneficial microbial communities [40]. With that in mind, understanding the factors that shape plant microbiomes is not only of fundamental scientific interest but also of practical relevance for the future of sustainable agriculture. Concerns about the influence of the mlo mutation, widely used in breeding, on microbial communities have already been raised in the barley scientific circles [41,42]. Our results provide initial indications that root microbiome diversity may not be substantially affected by the mlo knockout mutation.

However, we did detect alterations in relative abundances of a few specific taxa in the mlo mutant at different time points. At TP1, two ASVs within the Allorhizobium-Neorhizobium-Rhizobium group were more abundant in the wild type. At TP2, another ASV assigned to the Allorhizobium-Neorhizobium-Rhizobium group was significantly more abundant in the mlo mutant, suggesting that Mlo has a time-dependent role in the colonization.

Additionally, fungal OTU_117, classified as Rhizopus, was also enriched in the wild type at TP1. The Rhizopus genus is known to be pathogenic, with some members causing root rot disease in Plukenetia volubilis [43] or seedling blight in rice [44]. The Rhizopus enrichment in the wild type may indicate that the mlo knockout also provides resistance to this fungal genus. Studies on the effect of mlo mutation on Rhizopus infections are needed to better understand their interaction.

Additionally, other bacteria were enriched in the mutant at TP2. Those included agriculturally valuable genera Bacillus and Flavobacterium, known to promote growth [45,46,47], increased tolerance to abiotic stress [48,49,50], and disease resistance [51,52,53], and less explored but also growth-promoting Ornithinibacillus [54]. On the other hand, the only genus whose colonization was inhibited by mlo mutation was Promicromonospora, which was reported as growth promoter in some cases [55]. Overall, the mlo mutation appeared to facilitate the recruitment and colonization of roots by a few specific bacterial taxa, but there are no indications that the overall composition or functioning of the bacterial community is significantly modulated by the mutation.

Moreover, specific fungal taxa were enriched in the wild type at TP2. Mortierellaceae are well-described in the plant microbiome context, with reported growth-promoting [56,57] and disease-resistance properties [58]. The genus Pyrenochaetopsis is less documented but also reported to be beneficial, enhancing resistance to Fusarium oxysporum in tomatoes [59] and improving phosphorus uptake [60]. The last enriched fungal taxon was Auriculariales. This taxon is mainly described as a significant wood decomposer [61], especially in forests, but a study reported its presence in rhizosphere soil of switchgrass [62]. No information about its possible function in the context of root microbiome is available. Further research is needed to confirm and understand the nature of associations between Mlo and the mentioned fungal taxa.

Apart from the probable effect of the mlo mutation, the observed differences in specific taxa could be driven by powdery mildew infection that occurred naturally through airborne spores at TP2. The relationship between powdery mildew disease and the barley root microbiome has already been reported [63], and it is known that plant immunity, which is triggered by the Bh infection, also affects microbiome composition [64]. Although no effect was detectable on the overall microbial community, the disease could contribute to the observed changes in the relative abundance of the mentioned taxa. We propose multi-site replication studies to further confirm the observed results.

4. Materials and Methods

4.1. Plant Material and Agrobacterium-Mediated Transformation

Powdery mildew-susceptible barley (Hordeum vulgare ssp. vulgare) cv. ‘Maythorpe’ was used as donor material for the Agrobacterium-mediated transformation and genome editing of Mlo. Cv. ‘Maythorpe’ was sown into the substrate Hawita professional pH 5.2 (sphagnum) (HAWITA Gruppe GmbH, Vechta, Germany), and plants were grown in a growth chamber at 20 °C during the day and 16 °C at night, with a day/night cycle of 16 h/8 h. Water mixed with fertilizers (Azelis brun pioneer basis and NPK Makro 13-2-23+Mg+TE) (Tangaard, Bjæverskov, Danmark) was provided twice a day. The isolation and transformation of immature embryos was performed 12–14 days after pollination, following the protocol described in Holme, Dionisio [65]. After the transformation and regeneration process, plants were grown in greenhouse conditions until full maturity, and the seeds of chosen individuals were harvested.

4.2. Construct Design

To knock out Mlo, we targeted the second exon, and a protospacer sequence was created with the CRISPR tool available on benchling.com. The genome sequence of barley cv. ‘Golden Promise’ was used as a reference. Before designing the protospacer, Mlo was sequenced in cv. ‘Maythorpe’ to confirm that the sequence is the same as the reference. Although the CRISPR tool on benchling.com considers off-targets when designing protospacers, to double-check, we also used the online tool Cas-OFFinder, available at http://www.rgenome.net/cas-offinder/ (accessed on 1 May 2022).

We used a binary vector system based on the destination vector pANIC6A [66]. The entry vector pJG85 was used to clone the protospacer using the In-Fusion^®^ Snap Assembly Master Mix (Takara). Vector pJG80 was used as an entry vector for wheat codon-optimized Cas9, and both entry vectors were cloned into the pANIC6A destination vector, according to Holme, Wendt [67]. After the assembly, the final transformation vector was transformed into the A. tumefaciens strain AGL-0 using the freeze-thaw method [68]. Glycerol stocks were made and kept at −80 °C until usage. The protospacer sequence, target site, and final transformation vector map are available in Figure 1.

4.3. Genotyping Generations T0 and T1

Two to three weeks after regenerating T_0_ plantlets, emerged leaf samples (two pieces of around 3 cm each) were collected to extract gDNA and genotype each plantlet. gDNA was extracted following the protocol from Pallotta, Graham [69]. First, we checked if the plant was transformed with PCR amplifying the hygromycin resistance gene using the Herculase II DNA polymerase (Agilent Technologies, Santa Clara, CA, USA) according to the manual provided. Then, positive samples were used in the subsequent PCR to amplify the part of the Mlo gene and screen for the mutation presence. The amplicon was 492 bp long and included the targeted sequence in exon 2. The reaction was performed using the Q5^®^ High-Fidelity DNA Polymerase (NEB, Ipswich, Massachusetts, USA), following the supplier’s manual. Amplicons were purified from 1% agarose gel using a gel purification kit (Macherey-Nagel GmbH & Co. KG, Dueren, Germany) and sequenced by Macrogen (Macrogen Europe, Maastricht, The Netherlands). All primer sets, PCR conditions, and sequencing primer are presented in the Supplementary Materials.

Amplicons of heterozygous samples were TOPO cloned using the pCR^®^4Blunt-TOPO^®^ vector from Invitrogen (Thermo Fisher Scientific, Waltham, Massachusetts, USA). Then, formed plasmids were transformed into STELLAR chemically competent E. coli (Takara Bio Europe SAS, Saint-Germain-en-Laye, France), following the manual. Positive colonies were picked, and plasmids were extracted using NucleoSpin Plasmid kit (Macherey-Nagel GmbH & Co. KG, Dueren, Germany) and sequenced at Macrogen using the T3 universal primer (Supplementary Materials).

Genome-edited T_0_ plants were selected, and their seeds were harvested at full maturity. The seeds were subsequently sown in pots (17 cm wide,14 cm deep) filled with peat moss substrate, and plants were grown in the greenhouse conditions to screen the T_1_ generation. As previously reported, leaf samples were collected 2 to 3 weeks after the emergence to extract gDNA. In T_1_, plants were not screened for the hygromycin resistance gene presence, and genotyping was performed only by amplifying and sequencing Mlo, as already described in Section 2.3. Plants with selected mlo genotypes were grown until full maturity, and their seeds were harvested (Figure 1).

Protein prediction was performed for selected mlo genotypes to assess if mutations led to premature stop codon formation or changes in amino acid sequence. The genotyping results for selected T_1_ plants were used to predict amino acid sequences. The amino acid analysis and sequence alignments were performed using tools available on benchling.com. With the obtained amino acid sequences, a 3D model prediction of the wild-type and mutant proteins were performed using the AlphaFold3 algorithm with default settings on https://alphafoldserver.com/ (accessed on 4 February 2026). Screenshots of the 3D model prediction are in the Supplementary Materials. Additionally, for the mutant lines without premature stop codon, differences in protein stability were assessed using the instability index (II) available at benchling.com.

4.4. Inoculation with Powdery Mildew in Generation T2

Seeds of barley cv. ‘Maythorpe’ (susceptible, Mlo) and five homozygous T_1_ CRISPR-made mlo mutant lines (Figure 1) were sown in pots filled with peat moss substrate (Pindstrup) and placed in a growth chamber at 20 °C/15 °C and 16 h light (150 μmol s^−1^ m^−2^)/8 h darkness cycle. About 15 seeds were sown in each pot, and each genotype had two pots. No fertilizer was applied, and watering was optimal.

For inoculation, Bh isolate C15 was used. The isolate was propagated on the susceptible barley line ‘P-02’, a near-isogenic line of cv. ‘Pallas’, placed in the above conditions. For scoring resistance to Bh, primary leaves of seven-day-old seedlings were bent over the horizontal plastic support, with adaxial side up, and secured with two rubber bands (Supplementary Materials Figure S3). Eight leaves were bent for each pot. Inoculation was performed during daylight. Pots were placed in a tray inside a settling tower. A microscope slide was placed in the tray between pots to monitor inoculum density. A few Bh-infected barley plants with freshly produced conidia were held above the settling tower, and conidiospores were blown into the tower. The spores were allowed to settle for 15–20 min, and subsequently, a spore density of four–six spores per square millimeter was estimated on microscope slide. The inoculated plants were kept in the above conditions.

4.5. Macroscopic and Microscopic Resistance Evaluation

The macroscopic resistance evaluation was performed visually 7 days post-inoculation (dpi) when the number of powdery mildew colonies were assessed.

To assess resistance microscopically, we repeated inoculation as previously described. At 4 dpi, leaf pieces (around 4 cm each) were dyed with trypan blue to score fungal penetration success and hypersensitive response (HR). The staining was performed according to Bowling, Clarke [70] with modifications. The leaves were boiled for 10 min and incubated overnight before destaining.

Four leaves per genotype were used for scoring under the light microscope, and 100 randomly selected penetration attempts were counted per leaf. Penetration was considered successful if a haustorium or fungal colony was formed, while HR was registered if cellular uptake of the stain was detected.

Average values were calculated per genotype, and data were statistically analyzed with RStudio using the agricolae 1.3.7, FSA 0.10.1, and ggplot2 4.0.1 packages. The data were visually assessed for normality, and sqrt transformation was applied if necessary. Datasets were normally distributed, so one-way ANOVA followed by Tukey’s HSD test was conducted. Significant differences were denoted using asterisks: “” (p < 0.05), “” (p < 0.01), and “” (p < 0.001). Different letters indicated statistically distinct groups based on Tukey’s HSD test.

4.6. Microbiome Analysis of mlo Mutants in Generation T2

4.6.1. Experimental Setup and Rhizosphere Sampling

To assess whether mlo knockout affects the microbiome, we conducted an experiment in field-like conditions in a fine mesh tunnel at Forsøgsvej 1, Slagelse, Denmark (55.325° N, 11.391° E). Six plots were established to compare two barley genotypes: the mlo mutant and the wild type (cv. ‘Maythorpe’). Each genotype was replicated across three plots, and each plot was further divided into two rows, resulting in six replicates per genotype. We had two sampling time points, 18 and 32 days after sowing (19 September and 2 October 2023). Two plants were collected at each sampling time point in each row and pooled as one sample. Loosely attached soil was removed by shaking the roots, and the roots with the adhering rhizosphere soil were lyophilized, ground, and used for downstream analysis. In total, 24 biological samples were collected, comprising two genotypes across two time points, with six biological replicates per genotype and time point.

4.6.2. DNA Extraction

Total DNA was extracted from 0.25 g of rhizosphere soil using the PowerSoil DNA Isolation Kit (QIAGEN, Germantown, MD, USA), following the manufacturer’s protocol. The extracted DNA samples were stored at −80 °C until further use.

4.6.3. Metabarcoding

To determine the composition of the bacterial and fungal communities, we performed Illumina Miseq amplicon sequencing. First, we generated the libraries. The libraries of bacterial and fungal DNA were generated by a two-step dual-indexing PCR strategy, as previously described [71]. For bacteria, the V3/V4 region of the 16S ribosomal RNA (rRNA) gene was amplified by primers 341F (S-DBact-0341-b-S-17: 5′-CCTACGGGNGGCWGCAG-3′) and 785R (S-D-Bact-0785-a-A-21: 5′-GACTACHVGGGTATCTAATCC-3′) [72]. For fungi, the internal transcribed spacer 2 (ITS2) region of the fungal rRNA gene was amplified by primers fITS7 (5′-GTGARTCATCGAATCTTTG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) [73]. PCR conditions are available in Supplementary Materials. Subsequently, a 15-cycle indexing PCR was performed with a similar PCR mixture and program as described above. A dual-indexing approach was used for index combinations [74,75], using primer sequences with internal barcodes and index previously reported by [71]. SYBR staining was used in a 1.5% agarose gel for visualization to confirm the PCR product size. The barcoded PCR products were equimolarly pooled for equal representation, followed by ethanol precipitation and resolution in 50 μL of Tris-EDTA buffer. Subsequently, the pooled DNA was separated on a 1.5% agarose gel, and the DNA band of the expected size (300 to 450 bp) was extracted using a QIAquick Gel Extraction Kit (Qiagen, USA). Finally, the amplicon sequencing was performed on a MiSeq platform by Eurofins MWG (Ebersberg, Germany).

4.6.4. Bioinformatic Analysis

The obtained sequencing data were analyzed using Quantitative Insights into Microbial Ecology 2 (QIIME2, version 2023.7). Briefly, raw paired-end reads were input, and the quality of the reads was checked by the ‘qiime demux summarize’ function. Accordingly, amplicon sequence variants (ASVs) were determined by DADA2 from raw paired-end reads using the ‘qiime dada2 denoised-paired’ function. The parameters were as follows: for bacteria, forward and reverse reads were trimmed at the 5′ end until 17 and 21 bp, respectively, to remove the primers and truncated at the 3′ end until 285 and 200 bp, respectively, to remove the low-quality base pairs; for fungi, forward and reverse reads were trimmed at the 5′ end until 19 and 20 bp, respectively, to remove the primers and truncated at the 3′ end until 300 and 220 bp, respectively, to remove the low-quality base pairs. To classify the taxonomy of bacterial ASVs, a naïve Bayes classifier was trained by the ‘feature-classifier’ function against the Silva Ref NR99 [release 132] database. The ASVs were then classified using the ‘feature-classifier classify-sklearn’ function, and non-bacteria reads such as chloroplasts and mitochondria were moved from the ASVs table. For fungi, ITS2 regions were further extracted by ITSx and clustered into Operational Taxonomic Units (OTUs) at 98.5% similarity by vsearch, as these OTUs are closer to the fungal species level. To classify the taxonomy of fungal OTUs, the ‘taxid’ of R package ‘DECIPHER’ was employed to identify the taxonomy against the UNITE database.

4.6.5. Statistical Analysis

Microbiome downstream analyses were conducted in R version 4.3.1, using ASV or OTU tables of bacterial and fungal communities, respectively. The vegan package was used for analyses of alpha and beta diversity. For alpha diversity, ASV/OTU tables were rarefied to the minimum sequencing depth across all samples using the rrarefy function. Subsequently, observed ASVs/OTUs, Pielou’s evenness index, and the Shannon diversity index were calculated to estimate community richness, evenness, and diversity. For beta diversity analysis, raw ASV/OTU counts were log_2_-transformed to reduce amplification bias, followed by total sum scaling (TSS) to normalize data into relative abundances. Based on the relative abundance table, Bray–Curtis dissimilarities were calculated, and principal coordinates analysis (PCoA) was performed using the cmdscale function. To test for significant differences between the mlo mutant and wild type, PERMANOVA and ANOSIM analyses were carried out using the adonis2 and anosim functions, respectively.

Finally, to identify differentially enriched bacterial and fungal ASVs/OTUs between the two treatments, DESeq2 analysis was performed on the raw count tables using the DESeq2 package (v1.47.5) in R.

5. Conclusions

In summary, we developed novel CRISPR-induced mlo alleles which confer complete resistance to powdery mildew in barley. This research also produced the first mlo barley mutant with a mutation in the extracellular region of the protein. While mlo knockout did not significantly alter overall root-associated microbiome diversity, we observed differences in the relative abundances of a few specific bacterial and fungal taxa, indicating differentially enriched microbes at various time points. Overall, our study suggests that the powdery mildew resistance acquired by the knockout of mlo did not disrupt or compromise the composition of the root microbiome.