Depletion of KNL2 Results in Altered Expression of Genes Involved in Regulation of the Cell Cycle, Transcription, and Development in Arabidopsis

Centromeres contain specialized nucleosomes at which histone H3 is partially replaced by the centromeric histone H3 variant cenH3 that is required for the assembly, maintenance, and proper function of kinetochores during mitotic and meiotic divisions. Previously, we identified a KINETOCHORE NULL 2 (KNL2) of Arabidopsis thaliana that is involved in the licensing of centromeres for the cenH3 recruitment. We also demonstrated that a knockout mutant for KNL2 shows mitotic and meiotic defects, slower development, reduced growth rate, and fertility. To analyze an effect of KNL2 mutation on global gene transcription of Arabidopsis, we performed RNA-sequencing experiments using seedling and flower bud tissues of knl2 and wild-type plants. The transcriptome data analysis revealed a high number of differentially expressed genes (DEGs) in knl2 plants. The set was enriched in genes involved in the regulation of the cell cycle, transcription, development, and DNA damage repair. In addition to comprehensive information regarding the effects of KNL2 mutation on the global gene expression, physiological changes in plants are also presented, which provides an integrated understanding of the critical role played by KNL2 in plant growth and development.


Introduction
The centromeres are specialized chromosomal domains that are required for proper separation of chromosomes during mitosis and meiosis. The centromere is composed of centromeric DNA, often enriched in satellite repeats, and the large protein complex "kinetochore". In centromeric nucleosomes of most eukaryotes, histone H3 is partially replaced by the centromeric histone H3 variant cenH3 (also known as CENP-A in mammals, CID in Drosophila, Cse4 in Saccharomyces cerevisiae, and Cnp1 in Schizosaccharomyces pombe [1]). Deposition of cenH3 at the centromeric region is a prerequisite of the correct assembly and function of the kinetochore complex. It depends on different cenH3 assembly factors and chaperones [2], the transcription of the centromeric repeats [3][4][5], and the epigenetic status of centromeric chromatin [6,7]. In mammals, the Mis18 complex composed of Mis18α, Mis18β, and Mis18-binding protein 1 (also known as KNL2) plays an important role in the licensing of centromeres for cenH3 recruitment [8,9]. The human Mis18 protein complex localizes to centromeres

Loss of KNL2 Function Leads to Massive Transcriptional Changes
To understand the role of KNL2 for the kinetochore function, mitotic and meiotic divisions, and subsequently for plant growth and development, we compared the transcriptomes from two tissue types, namely seedlings and flower buds, of wild-type and the knl2 plants. The experiments in the current study were designed in a way to weaken factors that could introduce experimental noise and would diminish the biological relevance of the data. For the RNA-seq study, we compared control and mutant plants with the same genetic background (Columbia ecotype) and physical age by performing experiments under the same plant growth conditions. This allowed the identification of DEGs and enriched biological processes. At the same time, the information obtained is limited to a single snapshot of gene expression reflecting the physical age of plants. Further comparative experiments including more time points to study temporal gene expression profiles in the knl2 mutant line would be desirable, such as the comparison of the transcriptomes of knl2 and wild-type plants not of the same "physical" but "biological age", since the knl2 mutant plants showed a delay in growth and development [13].
RNA-seq analysis based on DeSeq2 [21] identified 3261 genes in seedlings (Table S1, Supplementary Materials) that were altered in the knl2 mutant line (adjusted p-value < 0.05). Among them, 459 genes ( Figure 1) were highly differentially expressed (fold change (FC) cutoffs of ≥ 2). performing experiments under the same plant growth conditions. This allowed the identification of DEGs and enriched biological processes. At the same time, the information obtained is limited to a single snapshot of gene expression reflecting the physical age of plants. Further comparative experiments including more time points to study temporal gene expression profiles in the knl2 mutant line would be desirable, such as the comparison of the transcriptomes of knl2 and wild-type plants not of the same "physical" but "biological age", since the knl2 mutant plants showed a delay in growth and development [13]. RNA-seq analysis based on DeSeq2 [21] identified 3261 genes in seedlings (Table S1, Supplementary Materials) that were altered in the knl2 mutant line (adjusted p-value < 0.05). Among them, 459 genes ( Figure 1) were highly differentially expressed (fold change (FC) cutoffs of ≥ 2).
In flower buds, more DEGs were identified in the knl2 (4750 in total). Out of 1858 highly expressed genes (FC cutoffs of ≥ 2) (Figure 1), 1194 were highly upregulated and 664 were highly downregulated in the knl2 mutant (adjusted p-value < 0.05). In both tissues, more genes were significantly upregulated than downregulated (Table S1). In agreement with previously published data [13], which showed absence of the full-length KNL2 transcript in the knl2 plants, the mRNA level of KNL2 was greatly reduced (log2 FC of −2.63 and −4.57) in seedling and flower bud tissues, respectively, of knl2 plants, suggesting that the observed transcription responses are the result of KNL2 depletion.
To explore the distribution of DEGs and to identify the biological consequences of the KNL2 inactivation, we used the GO::TermFinder [22]. Gene Ontology (GO) enrichment analysis for the "Biological process" was compared for downregulated and upregulated DEGs (adjusted p-value < 0.05) represented in seedlings and, separately, in flower buds. Detailed information about all overrepresented processes is given in a supporting information dataset (Dataset S1, Supplementary Materials). In this paper, we consider in more detail the categories associated with cell division, chromatin status, and plant growth and development. The results of the GO enrichment analysis can shed new light on how KNL2 depletion leads to plant growth and development alterations.

Knockout of KNL2 Impairs Expression of Genes Involved in Kinetochore Function
Kinetochores are responsible for the accurate segregation of chromosomes during mitosis and meiosis. Our RNA-seq analysis revealed GO terms related to physical events associated with both types of cell division, such as the "chromosome segregation", "meiosis I", "meiotic chromosome separation", "mitotic metaphase plate congression", "mitotic recombination", and "attachment of mitotic spindle microtubules to kinetochore" in seedlings (Table 1). All DEGs representing these categories were downregulated in the knl2 plants (Dataset S1, Tables S2 and S3). In flower buds, more DEGs were identified in the knl2 (4750 in total). Out of 1858 highly expressed genes (FC cutoffs of ≥ 2) (Figure 1), 1194 were highly upregulated and 664 were highly downregulated in the knl2 mutant (adjusted p-value < 0.05). In both tissues, more genes were significantly upregulated than downregulated (Table S1).
In agreement with previously published data [13], which showed absence of the full-length KNL2 transcript in the knl2 plants, the mRNA level of KNL2 was greatly reduced (log 2 FC of −2.63 and −4.57) in seedling and flower bud tissues, respectively, of knl2 plants, suggesting that the observed transcription responses are the result of KNL2 depletion.
To explore the distribution of DEGs and to identify the biological consequences of the KNL2 inactivation, we used the GO::TermFinder [22]. Gene Ontology (GO) enrichment analysis for the "Biological process" was compared for downregulated and upregulated DEGs (adjusted p-value < 0.05) represented in seedlings and, separately, in flower buds. Detailed information about all overrepresented processes is given in a supporting information dataset (Dataset S1, Supplementary Materials). In this paper, we consider in more detail the categories associated with cell division, chromatin status, and plant growth and development. The results of the GO enrichment analysis can shed new light on how KNL2 depletion leads to plant growth and development alterations.

Knockout of KNL2 Impairs Expression of Genes Involved in Kinetochore Function
Kinetochores are responsible for the accurate segregation of chromosomes during mitosis and meiosis. Our RNA-seq analysis revealed GO terms related to physical events associated with both types of cell division, such as the "chromosome segregation", "meiosis I", "meiotic chromosome separation", "mitotic metaphase plate congression", "mitotic recombination", and "attachment of mitotic spindle microtubules to kinetochore" in seedlings (Table 1). All DEGs representing these categories were downregulated in the knl2 plants (Dataset S1, Tables S2 and S3). Table 1. Assignment of differentially expressed downregulated genes in knl2 seedlings to different functional categories related to cell processes. Gene Ontologies were analyzed for term enrichment using the Generic Gene Ontology GO::TermFinder tool (cutoff of 5% false discovery rate (FDR)). GO terms characterizing the functions of kinetochores, like microtubule binding, chromosome movement, and checkpoint signaling were overrepresented in the downregulated gene list of knl2 flower bud samples ( Table 2). Table 2. Assignment of differentially expressed downregulated genes in knl2 flower buds to different functional categories related to cell processes. Gene Ontologies were analyzed for term enrichment using the Generic Gene Ontology GO::TermFinder tool (cutoff of 5% FDR). For instance, reduced transcript levels were found for genes with known microtubule-associated functions [23] such as KINESIN 5 (ATK5; AT4G05190), TITAN1 (TTN1; AT3G60740), and HINKEL (AT1G18370). The CELL DIVISION CYCLE 20.1 (CDC20.1; AT4G33270), known for its critical role in the spindle assembly checkpoint-dependent meiotic chromosome segregation [24], was also significantly suppressed in the flower buds of knl2 plants. Another example is the reduced expression of BUB1 (AT2G20635), a protein kinase containing the Mad3-BUB1-I domain. Being together with CDC20.1 in the same protein-protein interaction network, BUB1 plays an important role in the assembly of the checkpoint proteins at the kinetochore [25]. In our RNA-seq study, it was highly downregulated in both types of tissue. Several other kinetochore associated genes, like GAMMA-TUBULIN (TUBG1; AT3G61650), PLEIADE (PLE; AT5G51600), and AURORA KINASE 1 (AT4G32830) were downregulated in knl2 flower buds. The expression of the inner kinetochore CENP-C (AT1G15660) was decreased in both seedlings and flower buds. Knockout of KNL2 results in a reduced amount of cenH3 transcripts of Arabidopsis [13]. This result was confirmed by RT-qPCR analysis of the samples used for the current RNA-seq study.

Reduced DNA Methylation in the knl2 Mutant Might Be Responsible for the Activation of a High Number of Transposons
Knockout of Arabidopsis KNL2 leads to a reduced level of DNA methylation [13]. In our RNA-seq study, a number of genes related to the DNA and histone methylation showed repressed differential expression in the knl2 line (in both tissues). For example, histone H3 lysine 9 di-methyltransferase SUVH4 (AT5G13960), CHROMOMETHYLASE 2 (CMT2; AT4G19020), and ARGONAUTE 6 (AGO6; AT2G32940) were found among differentially downregulated genes in the flower buds (adjusted p-value < 0.05), whereas DEMETER-LIKE 1 (DML1; AT2G36490) showed reduced expression in both tissues used for analysis.

Altered Expression of the Root-related Genes Explains the Slow Root Growth of knl2
Root length was compared between seven-day-old control and the knl2 mutant seedlings. The average root length of mutant seedlings (0.66 cm) was about 3 times shorter than that of the wild-type (1.82 cm) (Figure 3), and GO terms related to the root development such as "regulation of root development" and "primary root development" were highly overrepresented among the genes downregulated in the knl2 seedlings (Table 1). In contrast, the non-LTR retrotransposons Sadhu 1-3 (AT3G44042), Sadhu 3-1 (AT3G44042), Sadhu 4-1 (AT5G28913), and Sadhu 5-1 (AT4G01525) had decreased expression level in the knl2 flower buds. Depression of Sadhu 3-1 retroelement was also observed in the epigenetic mutant suvh4 [29].

Altered Expression of the Root-related Genes Explains the Slow Root Growth of knl2
Root length was compared between seven-day-old control and the knl2 mutant seedlings. The average root length of mutant seedlings (0.66 cm) was about 3 times shorter than that of the wild-type (1.82 cm) (Figure 3), and GO terms related to the root development such as "regulation of root development" and "primary root development" were highly overrepresented among the genes downregulated in the knl2 seedlings (Table 1). We additionally tested all root-related genes available in TAIR for their overlap with the differentially expressed genes in the knl2 seedlings. As a result, 119 differentially expressed genes were identified (Table S4, Supplementary Materials). Among them, 14 were highly suppressed (FC cutoffs of ≥ 2). The reduced transcript level was found for SHORT AND SWOLLEN ROOT 1 (SSR1; AT5G02130). The ssr1-2 plants show reduced root growth [30]. Other downregulated genes encoding the chloroplast/plastid localized GAPDH isoforms are GAPCp1 (AT1G16300) and GAPCp2 (AT1G79530). gapcp double mutants are characterized by arrested root development, dwarfism, and sterility [31]. The transcription factor AGAMOUS-like 14/XAL2 (AT4G11880) preferentially expressed in roots is highly suppressed in knl2 seedlings (log 2 FC of −1.05, adjusted p-value 0.0001). Several studies demonstrated that xal2 mutants have short roots [32,33]. In addition, we observed the reduced transcript level of EMBRYO DEFECTIVE 2757 (EMB2757; AT4G29860) in knl2 seedlings. Genetic analysis of EMB2757 in Arabidopsis has demonstrated that mutations in this gene cause defects in both embryo and seedling development [34]. Upregulated differentially expressed genes in knl2 seedlings can be further exemplified with RAPID ALKALINIZATION FACTOR 23 (RALF23; AT3G16570). Overexpression of RALF23 leads to slower growing seedlings with roots that have reduced capacity to acidify the rhizosphere [35].

DNA Damage Repair Genes are Downregulated in the knl2 Plants
The genome stability is maintained by DNA damage responses. The failure to repair DNA damage leads to negative processes related to plant growth, reproduction, and even lethality. We found that genes representing the GO term "DNA repair" (26 genes) and related categories, such as "double-strand break repair via homologous recombination", "non-recombinational repair", and "DNA ligation involved in DNA repair" were overrepresented among knl2 DEGs (Table 1). All genes of these categories were downregulated in the mutant. Among downregulated genes were the key players participating in the canonical nonhomologous end joining, such as KU70 (AT1G16970), KU80 (AT1G48050), and LIGASE 4 (AT5G57160). It is known that mutations in these genes lead to increased sensitivity to double-strand DNA breaks (DSB)-inducing factors (reviewed in [36]). AT5G20850, encoding a DNA recombination and repair protein RAD51, is another example of a gene with a decreased expression level in knl2 seedlings (log 2 FC of −0.99, adjusted p-value 1.47 × 10 −6 ). Loss of RAD51 function does not affect the vegetative development of Arabidopsis, probably due to functional redundancy with other genes of the RAD51 family, but is essential for meiotic repair of DSBs caused by AtSPO11−1 [37]. In addition, the RAD51-like gene, DISRUPTION OF MEIOTIC CONTROL 1 (DMC1; AT3G22880), known to promote interhomolog recombination was strongly downregulated in the knl2 flower buds (log 2 FC of −1.20, adjusted p-value 1.25 × 10 −8 ). Interestingly, the mammalian kinetochore protein BUB1, playing an important role in chromosome segregation, is also known to participate in the DNA damage response [38]. In human and mouse cells, cenH3/CENP-A accumulates at DSBs together with CENP-N, CENP-T, and CENP-U [39]. Since the induction of DNA damage by radiation resulted in an increased expression of cenH3, the authors proposed a cenH3 function in DNA repair. Further research is needed to unravel the interrelationships between kinetochore genes and the DNA damage response.
To test whether the altered mRNA levels are associated with changed capacity of knl2 plants for DNA damage repair, we exposed wild-type and mutant plants to 10 µM DNA inter-strand crosslink inducer mitomycin C (MMC), 50 nM DNA strand breaker bleomycin, and 10 nM enzymatic DNA-protein crosslinker camptothecin ( Figure S1, Supplementary Materials). This pilot experiment indicated sensitivity of knl2 plants to MMC, but not other tested drugs (Figure 4).
To validate this result, we treated knl2 plants with a series of MMC concentrations (2.5 to 15 µM) and found significant reduction in their root growth compared to wild-type plants ( Figure 4A,B). Next, we performed propidium iodide (PI) staining of root apices from plants treated for 24 h with 0, 10, and 20 µM MMC. Intense PI staining inside the cells indicates damaged plasma membranes, typical for stressed and dead cells. Under mock conditions, wild-type plant root apical meristems were intact, whereas the meristems of knl2 plants showed few intensely stained cells. MMC treatment caused dose-dependent accumulation of intensely stained cells in the meristems of wild-type plants, but the increase in the meristem of knl2 plants was more prominent.
Based on this, we conclude that knl2 plants fail to properly activate specific DNA damage repair factors, have reduced capacity for repair of DNA inter-strand crosslinks, and suffer from increased cellular damage.

Knockout of KNL2 Results in Deregulated Expression of a High Number of Transcription Factors
We hypothesized that a large number of DEGs in the knl2 mutant may be the result of deregulated expression of transcription factors. Therefore, the knl2 mutant transcriptome was further screened for the presence of differentially expressed transcription factors (TFs). TFs were retrieved from The Plant TF database v.4 [40]. In knl2 seedlings, 202 TFs were expressed (43 highly differentially expressed), whereas in knl2 flower buds 330 TFs (137 highly differentially expressed) were detected. The classification of TFs is visualized in the heat map ( Figure 5   TFs with an altered expression in the knl2 mutant are involved in different processes and provide insight into Arabidopsis development changes due to the inactivation of the kinetochore gene KNL2. There was a striking coordinated downregulation of genes representing the MADS-box TFs with an altered expression in the knl2 mutant are involved in different processes and provide insight into Arabidopsis development changes due to the inactivation of the kinetochore gene KNL2. There was a striking coordinated downregulation of genes representing the MADS-box family, which are the key regulators of seed and flower development (detailed description is presented below in 3.7). The upregulated TFs can be exemplified by the genes of the WRKY family. The TFs of this family play an important role in plant development and responses to environmental stress stimuli [41][42][43]. From 74 members of the WRKY TF family, 13 were differentially expressed in seedlings and 26 in the flower buds. The differentially expressed genes WRKY33 (AT2G38470) and WRKY46 (AT2G46400), highly upregulated in seedlings and flower buds, are known to play a role in heat stress responses. The expression of these genes was elevated in MBF1c-overexpressing plants, which showed enhanced tolerance to heat compared with wild-type plants [44]. WRKY46 additionally regulates responses to other stresses in Arabidopsis [45][46][47].
The heat map also revealed genes that are differentially expressed in flower buds but not in seedlings and vice versa. We observed that 94 highly differentially expressed genes were present only in flower buds and 18 highly regulated genes were seedling-specific (Dataset S2). The transcriptional repressor of the RAV family TEMPRANILLO 1 (TEM1), known to postpone floral induction [49], is another example of the highly upregulated gene present in the flower buds (log 2 FC of 1.55).

Late Flowering of the knl2 Plants is Determined by the Altered Expression of Flowering Genes
We observed that flowering of the knl2 plants delays for 10-14 days compared to wild-type ( Figure 6; Figure S2, Supplementary Materials).
From this observation, we could identify a number of enriched GO terms related to the floral development in our RNA-seq data from knl2 flower buds. These represent such GO categories as "reproduction", "regulation of pollen tube growth", "pollen germination", "pollination", and "pollen development" (Table 3). Table 3. Assignment of genes differentially expressed in knl2 flower buds to development-related functional categories based on GO::TermFinder. Genes downregulated in the knl2 flower buds were included into the GO enrichment analysis (cutoff of 5% FDR). Genes representing these and related categories in Table 3 were downregulated in the knl2 flower buds. Some of the genes belong to the MYB transcription factors and regulate the plant microgamete development. For instance, AtMYB103 (AT1G63910), important for pollen development [50], is one of the highly suppressed flower bud genes in knl2. The expression of other prominent transcription factors like AGAMOUS-LIKE 66 (AGL66; AT1G77980) and AGL104 (AT1G22130) involved in pollen maturation and tube growth was significantly reduced. Adamczyk and Fernandez [51] demonstrated that double mutant plants agl66 agl104 could produce pollen but had severely reduced fertility and aberrant pollen tube growth. Along with AGL66 and AGL104, AGL94 (AT1G69540) belongs to the MIKC * factors which are highly active in pollen as major regulators of pollen maturation programs [51,52]. These three genes were significantly downregulated in the knl2 flower buds (adjusted p-value < 0.001). This finding might explain the reduced fertility of knl2 plants in addition to the role of KNL2 in mitosis and meiosis [13]. From this observation, we could identify a number of enriched GO terms related to the floral development in our RNA-seq data from knl2 flower buds. These represent such GO categories as "reproduction", "regulation of pollen tube growth", "pollen germination", "pollination", and "pollen development" (Table 3).  The RNA-seq data analysis and observation of the delayed flowering prompted us to inspect the RNA-seq data of the knl2 seedlings. We asked the questions of whether and, if so, how many flowering-related genes based on the FLOR-ID overlap with the differentially expressed genes in the knl2 seedlings. From 306 genes, 36 genes could be identified (Dataset S3, Supplementary Materials). There was a striking coordinated downregulation of several prominent flowering-related genes. For example, the floral integrator Flowering Locus T (FT; AT1G65480), known to be expressed in leaves, was significantly downregulated in knl2 seedlings. Its reduced expression is in correspondence with the downregulation of SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1; AT2G45660) in the knl2 seedlings (log 2 FC of −2.24, adjusted p-value 7.72 × 10 −61 ). It is known that flowering activator SOC1 acts in a positive-feedback loop with AGL24, which is the important MIKC c -type transcription factor positively regulating flowering in Arabidopsis (reviewed in [53]). In our RNA-seq study, the AGL24 showed reduced expression in both knl2 seedlings and flower buds.

GO ID
GIGANTEA (GI; AT1G22770), an important gene in regulating photoperiodic flowering, had a reduced expression in the knl2 mutant line compared to wild-type. Another regulator of transition to flowering FKF1 (ADO3; AT1G68050) was highly downregulated in the knl2 mutant line (log 2 FC of −1.70, adjusted p-value 2.79 × 10 −17 ). A member of the MADS-box family XAANTAL2 (XAL2/AGL14; AT4G11880), which is a necessary and sufficient agent to induce flowering [54], showed decreased expression level in the knl2 seedlings (log2 FC of −1.05, adjusted p-value 0.002).
The reduced activity of flower inductive pathway genes observed in the knl2 seedlings might explain the delay in flower initiation, at least under long-day conditions.

Knockout of KNL2 Results in Altered Expression of Genes Controlling Seed Development
The loss-of-function knl2 mutant showed reduced seed production [13]. We used the SeedGenes Project database including 481 genes in order to examine the overlap between these genes and the 4750 genes that we found to be significantly differentially expressed in response to KNL2 inactivation. From these genes, 42 were differentially expressed in the knl2 flower bud samples (Table S5, Supplementary Materials). Among downregulated genes, there are those whose disruption causes embryo abortion, such as CYL 1 (AT5G13690), EMB 1674 (AT1G58210), EMB 2184 (AT1G75350), EMB 3003 (AT1G34430), EMB 3004 (AT3G06350), and IMPL 2 (AT4G39120). Highly upregulated genes that include HSP 17.4 (AT3G46230), RLP37 (AT3G23110), and NCED3 (AT3G14440) are known for their expression in seeds and response to abiotic stresses or defense.

Real-time Quantitative PCR Confirms RNA-seq Analysis
To validate the quality of the RNA-seq data by qRT-PCR, we selected eight genes differentially expressed in seedlings and 11 genes differentially expressed in flower buds (Figure 7; Figure S3, Supplementary Materials). In the case of seedlings, four genes involved in the regulation of root growth (AT4G11880, AT4G28720, AT3G50060, AT2G43140) and two genes involved in the regulation of flowering time (AT2G45660, AT1G68050) were selected for the analysis. In the case of flower buds, an expression of five genes involved in the regulation of flowering time (AT5G48560, AT2G40080, AT5G37770, AT1G10120, AT1G02580) and four genes involved in the regulation of seed development (AT2G01860, AT3G23110, AT5G13690, AT5G23940) was validated. KNL2 (At5g02520) and cenH3 (At1g01370) genes were included for the analysis in both types of tissues. Our qRT-PCR confirmed the extreme downregulation of KNL2 gene expression in the mutant line. All RNA-seq-based determined up-and downregulated genes showed the same qualitative response in both seedlings and flower buds based on qRT-PCR (correlation coefficients between qRT-PCR and RNA-seq data of 0.996 and 0.974, respectively).

Conclusions
KNL2 of Arabidopsis is involved in regulating the centromeric loading of cenH3, and its knockout results in mitotic and meiotic defects and reduced DNA methylation, followed by reduced growth rate and fertility compared to wild-type [13]. Our comparative RNA-seq analysis revealed that these abnormalities were associated with an altered expression of genes regulating corresponding processes. For instance, the reduced growth rate of roots and late flowering of the knl2 mutant correlate with the altered expression of genes regulating root growth and flowering, respectively. We propose that KNL2 has an essential regulatory function in plant growth and development that is much broader than the regulation of centromeric localization of cenH3 only. development (AT2G01860, AT3G23110, AT5G13690, AT5G23940) was validated. KNL2 (At5g02520) and cenH3 (At1g01370) genes were included for the analysis in both types of tissues. Our qRT-PCR confirmed the extreme downregulation of KNL2 gene expression in the mutant line. All RNA-seq-based determined up-and downregulated genes showed the same qualitative response in both seedlings and flower buds based on qRT-PCR (correlation coefficients between qRT-PCR and RNA-seq data of 0.996 and 0.974, respectively).

Conclusions
KNL2 of Arabidopsis is involved in regulating the centromeric loading of cenH3, and its knockout results in mitotic and meiotic defects and reduced DNA methylation, followed by reduced growth rate and fertility compared to wild-type [13]. Our comparative RNA-seq analysis revealed that these abnormalities were associated with an altered expression of genes regulating corresponding processes. For instance, the reduced growth rate of roots and late flowering of the knl2 mutant correlate with the altered expression of genes regulating root growth and flowering, respectively. We propose that KNL2 has an essential regulatory function in plant growth and development that is much broader than the regulation of centromeric localization of cenH3 only.

Plant Materials and Growth Conditions
The A. thaliana knl2 mutant (SALK_039482) in Col-0 background was described previously [13]. Seeds of A. thaliana wild-type and knl2 were germinated in Petri dishes on half strength Murashige

Plant Materials and Growth Conditions
The A. thaliana knl2 mutant (SALK_039482) in Col-0 background was described previously [13].

DNA Damage Sensitivity Assays
For the pilot root length assay experiment, plants were germinated and grown for seven days on 1 2 strength Murashige and Skoog medium containing 0.01% DMSO and different DNA damage inducers-mitomycin C (MMC) (Duchefa, Haarlem, The Netherlands), bleomycin (Calbiochem, San Diego, CA, USA), and camptothecin (Sigma-Aldrich, Saint Louis, MO, USA) with concentrations specified in the text. For the subsequent analysis, plants were germinated and grown for 14 days on media containing different concentrations of MMC (2.5, 5, 7.5, 10, and 15 µM from Sigma-Aldrich). For PI staining assay, plants were grown on solid 1 2 Murashige and Skoog medium for four days, then transferred to liquid 1 2 Murashige and Skoog medium containing 0, 10, and 20 µM MMC and grown for 24 h. Subsequently, the plants were stained with 10 µg/mL propidium iodide solution (Sigma-Aldrich) for 3 min, rinsed with tap water and analyzed using an AxioImager Z2 (Zeiss, Jena, Germany) microscope equipped with the DSD2 confocal module (Andor Technology, Belfast, Great Britain). Plants for all procedures were grown in a Percival growth chamber under long-day (16 h light) conditions and 21 • C.

RNA Isolation and Illumina Sequencing
Total RNA was extracted from the eight-day-old seedlings of A. thaliana. At least 60 seedlings (100 mg) were pooled to produce a biological replicate. For the RNA-seq analysis of flower buds, inflorescences were harvested from three individual plants for each sample (30 mg). Flower buds older than stage 12 and flowers [55] were removed. All tests were performed on three biological replicates per condition and genotype. Total RNA was isolated from seedlings and flower bud samples using the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's manual. The RNA preparations were checked for quality using a NanoDrop spectrophotometer and a 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA). In total, 12 RNA samples (1 µg each) were provided to the IPK-Sequencing-Service (IPK, Gatersleben, Germany) for construction of cDNA libraries with the TruSeq RNA Sample Preparation Kit (Illumina, San Diego, CA, USA). The libraries were sequenced in a HiSeq2500 rapid 100-bp single-read run system After sequencing, the adapter sequences and the barcodes were removed.

RNA-seq Data Processing
Sequencing quality of the reads was examined by using FastQC Read Quality reports, Galaxy Version 0 0.72 [56]. At least 94% of the bases of each read in all samples possessed Illumina Quality >30 and no sequence flagged as poor quality was detected.
Sequences were aligned against the Arabidopsis TAIR 10 genome assembly using HISAT2 Galaxy tool version 2.0.3. [57] with default settings. Read counts for each gene were quantified based on the BAM files produced with the HISAT2 by using the tool feature Counts, Galaxy Version 1.4.6.p5 [58]. The advanced setting parameters were strand specificity = "no", GFF feature type filter = "exon", and GFF gene identifier = "Parent". Differentially expressed features were determined based on the feature Counts tables by applying the tool DESeq2, Galaxy Version 2.11.38 [21] with the setting parameter fit type = "parametric". DESeq2 tested for differential expression based on a model using the negative binomial distribution. Differentially expressed genes were identified by comparison of two groups, namely, (1) a mutant line and the wild-type control, condition seedlings and (2) a mutant line and the wild-type control, condition flower buds.
The results of all statistical tests were adjusted for the multiple testing false discovery rate (FDR) with the Benjamini and Hochberg procedure [59]. A cutoff value of adjusted p-values equal to 0.05 was chosen as a threshold to identify significant differentially expressed genes.

Analysis of Differentially Expressed Genes (DEGs)
Functional characterization of the DEGs showing significant expression changes in response to KNL2 depletion was done based on the TAIR10 annotation (https://www.arabidopsis.org/). Gene Ontologies were analyzed for term enrichment using the Generic Gene Ontology GO::TermFinder tool ( [22]. The analysis was carried out using the Benjamini-Hochberg FDR with a filter p-value of < 0.05. Information about flowering-related genes was extracted from the Flowering Interactive Database [60]. The appearance of the transcription factors in the analyzed gene sets was confirmed with the Plant Transcription Factor Database v.4.0 [40]. Information about the genes essential for the Arabidopsis development from the SeedGenes Project [61] was used to find the corresponding genes among differentially expressed genes.

Gene Expression Validation by Reverse Transcription Quantitative PCR (RT-qPCR)
Total RNA extraction was performed as described above. The RNA was treated with DNase I (Ambion, Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's protocol to eliminate any residual genomic DNA. Reverse transcription was performed using a first-strand cDNA synthesis kit, oligo dT (18-mer) primer (both Fermentas, Thermo Fisher Scientific, Waltham, MA, USA), and 2 µg of total RNA as a starting material.
The gene-specific primers (Table S6, Supplementary Materials) were designed using the fully automated QuantPrime tool [62]. The amplification of the UBQ10 (AT4G05320) reference gene [63] was used as an internal control to normalize the data.
Quantitative real-time measurements were performed using POWER SYBR Green Master Mix reagent in a QuantStudio 6 Flex system (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's instructions. The cDNA equivalent to 40 ng of total RNA was used in a 10 µL PCR reaction.
The cycling conditions comprised 10 min polymerase activation at 95 • C and 40 cycles at 95 • C for 3 s and 60 • C for 30 s. Three biological replicates per genotype (wild-type and knl2 mutant line) in both conditions (seedlings and flower buds) were tested. Each biological replicate was represented with three technical replicates, which were analyzed during the same run. Relative gene expression was calculated using the comparative method 2 −∆∆CT [64].

Data Availability
Raw reads are available under the accession number PRJEB32230 at http://www.ebi.ac.uk/ena/ data/view/PRJEB32230.  Figure S1. knl2 sensitivity to DNA damaging agents. The plants were grown for seven days on media containing specified drugs. Figure S2. Effect of KNL2 depletion on the flowering time.The flowering time of the knl2 mutant delayed by 10-14 days compared to Arabidopsis wild-type. Seeds of the knl2 mutant and wild-type were germinated under short-day conditions, 8 h light/20 • C and 16 h dark/18 • C, for two weeks and then plants were transferred to the cultivation room with long-day conditions, 16 h light/20 • C and 8 h dark/18 • C. To demonstrate differences in flowering time between the knl2 mutant and wild-type, images of plant populations were taken with different time intervals. At early growth stages (14-day-old plants), no obvious phenotypical differences between the Col wild-type and the knl2 mutant were observed. Figure S3. qRT-PCR validation of RNA-seq results. Eleven genes were selected from the genes differentially expressed in flower buds according to the RNA-seq data analysis. Detailed annotation of the selected genes is presented in Table S2. *At5g02520 corresponds to KNL2. Error bars indicate the standard error of three biological replicates in qRT-PCR. Table S1. Results of the RNA-seq study based on the DeSeq2 in Col and the knl2 mutant line. Conditions: seedlings and flower buds. Table S2. Differentially expressed genes in knl2 seedlings and their annotation. Table S3. Differentially expressed genes in knl2 flower buds and their annotation. Table S4. Identification of root-related genes in knl2 seedlings. Table S5. Overlap between seed-related genes from the SeedGenes Project DB and differentially expressed genes in knl2 flower buds. Table S6. Selected genes and their primers used for qRT-PCR. Dataset S1. GO terms overrepresented in the knl2