Functional conservation and divergence of arabidopsis VENOSA4 and human SAMHD1 in DNA repair

Abstract

The human deoxyribonucleoside triphosphatase (dNTPase) Sterile alpha motif and histidine-aspartate domain containing protein 1 (SAMHD1) has a dNTPase-independent role in repairing DNA double-strand breaks (DSBs) by homologous recombination (HR). Here, we show that VENOSA4 (VEN4), the probable Arabidopsis thaliana ortholog of SAMHD1, also functions in DSB repair by HR. The ven4 loss-of-function mutants showed increased DNA ploidy and deregulated DNA repair genes, suggesting DNA damage accumulation. Hydroxyurea, which blocks DNA replication and generates DSBs, induced VEN4 expression. The ven4 mutants were hypersensitive to hydroxyurea, with decreased DSB repair by HR. Metabolomic analysis of the strong ven4-0 mutant revealed depletion of metabolites associated with DNA damage responses. In contrast to SAMHD1, VEN4 showed no evident involvement in preventing R-loop accumulation. Our study thus reveals functional conservation in DNA repair by VEN4 and SAMHD1.

One sentence summary: Human SAMHD1 is involved in dNTP metabolism and DNA repair; the latter function is conserved in VEN4, its likely Arabidopsis ortholog.

1. Introduction

In plants and animals, deoxyribonucleoside triphosphate (dNTP) metabolism involves de novo biosynthesis, recycling, and degradation. A balanced dNTP pool is essential for maintaining accurate DNA replication and genome stability [1]; its imbalance during DNA biosynthesis can lead to incorrect dNTP incorporations into the nascent strand, which in turn may reduce replication fidelity if mismatched dNTPs are not eliminated, stall DNA polymerases, and induce replication stress [[2], [3], [4], [5]]. Replication stress affects genome integrity and can be caused not only by dNTP misincorporations but also by the presence of unusual secondary structures in the DNA template strand, or R-loops formed at certain regions by collisions between the transcription and replication machineries (reviewed by Zeman and Cimprich [6]).

Human (Homo sapiens) Sterile alpha motif (SAM) and histidine-aspartate (HD) domain-containing protein 1 (SAMHD1) is a nuclear protein with dNTPase activity: it degrades dNTPs into their constituent deoxyribonucleoside and inorganic triphosphate [7,8]. SAMHD1 acts in regulating dNTP metabolism and suppressing antiviral immune responses (reviewed by Chen et al. [9]). Moreover, loss of SAMHD1 function has been related to Aicardi-Goutières syndrome, a rare congenital autoimmune disorder that leads to neurodegeneration and typically manifests in early childhood, with symptoms resembling those of a chronic viral infection [7,8,[10], [11], [12]].

Other studies have described a dNTPase-independent role for SAMHD1 in DNA double-strand break (DSB) repair by homologous recombination (HR). SAMHD1 colocalizes with tumor protein p53 binding protein 1 (TP53BP1) at DSB foci produced by treatment with the DSB-inducer camptothecin. At these DSB foci, SAMHD1 recruits C-terminal-binding protein 1-interacting protein (CtIP), an endonuclease that promotes DNA end resection. Cells lacking SAMHD1 function show hypersensitivity to camptothecin, reduced ability to repair DSBs by HR, and impaired recruitment of CtIP to DNA damage foci [13,14]. Additionally, in human cell lines, SAMHD1 facilitates the resection of gapped or reversed DNA replication forks by interacting with and stimulating the 3′–5′ exonuclease activity of the DSB repair protein MRE11, which degrades the nascent DNA, thus stabilizing the stalled forks and preventing their collapse into DSBs. Loss of SAMHD1 function also results in the cytoplasmic accumulation of single-stranded DNA fragments from stalled forks, which could induce type-I interferon inflammatory signaling responses [15].

A recent study demonstrated that SAMHD1 participates in preventing the accumulation of the co-transcriptional R-loops generated in chromatin regions where the transcription and replication machineries collide [16]. R-loops are three-stranded DNA–RNA structures that form when a nascent RNA anneals with its template DNA strand, displacing the complementary, non-template strand. R-loops are considered a major source of replication stress, DNA breaks, and genome instability, which are a hallmark of cancer; they also impede DNA replication, transcription, and repair (reviewed by Kumar and Remus [17]). Mutations in SAMHD1 have been identified in different types of cancers and are associated with a poor patient prognosis (reviewed by Li et al. [18], Mauney and Hollis [19], Coggins et al. [20], Chen et al. [21], and Schott et al. [22]).

We previously performed a large-scale screening for ethyl methanesulfonate (EMS)-induced point mutations causing abnormal leaf shape, size, or pigmentation in Arabidopsis (Arabidopsis thaliana). We isolated hundreds of viable mutants, which were classified into 19 phenotypic classes; mutants in the Venosa (Ven) class show reticulation of their rosette leaves [[23], [24], [25], [26]]. We used map-based cloning to identify the ven4 mutant, which we refer to as ven4-0 throughout this paper and in Sarmiento-Mañús et al. [27] Through iterative linkage analysis of the ven4-0 mutation using molecular markers, we determined that VENOSA4 (VEN4) is At5g40270, the most likely Arabidopsis ortholog of human SAMHD1 [27,28]. VEN4 and its rice (Oryza sativa) ortholog, STRIPE3 (ST3), are related to chloroplast and leaf development, stress and immune responses, and dNTP metabolism [[27], [28], [29], [30], [31]]. However, to date, no studies have associated VEN4 or ST3 with DNA damage repair. Here, we analyzed three ven4 allelic mutants to shed light on whether VEN4 functions in DNA DSB repair by HR. Our results, particularly the study of the original ven4-0 point mutation, reveal cross-kingdom functional conservation between Arabidopsis VEN4 and human SAMHD1 in DNA damage repair, in addition to their previously known function in dNTP metabolism.

2. Results

2.1. Arabidopsis VEN4 and human SAMHD1 share functionally crucial residues

The full-length human SAMHD1 and Arabidopsis VEN4 proteins, comprising 626 and 473 amino acids, respectively, show 37% sequence identity. The structure of human SAMHD1 protein has been extensively studied, and several domains and functionally crucial residues have been identified (reviewed by Morris and Taylor [32]). The SAM domain [33], a putative protein–protein and protein–nucleic acid interaction domain (reviewed by Qiao and Bowie [34]), is located in the N-terminal region of SAMHD1 (spanning residues 45 to 110 [35]) and appears to be necessary for protein stabilization after viral infection [36]; however, VEN4 lacks the SAM domain (Fig. 1 and Fig. S1).

Fig. 1.

Arabidopsis VEN4 and human SAMHD1 share 3D structural similarity, with conservation of functionally key residues. (A and B) Cartoon and surface representations of the predicted Arabidopsis VEN4 (A) and human SAMHD1 (B) 3D structures. Helices in the cartoon representations are shown as cylinders/stubs. The HD and SAM domains are colored in blue and pink, respectively. The residue altered by the ven4-0 mutation (E249) and the K248 and R231 residues (which are known to be functionally relevant in SAMHD1) are shown as spheres and highlighted in red, cyan, and orange, respectively (A); the conserved residues in human SAMHD1 are colored similarly (B). Close-up views of the VEN4 and SAMHD1 surfaces have been shaded to highlight protein cavities, which demonstrates that the E249 residue of VEN4 is more accessible to solvent than the equivalent E355 of SAMHD1.

The HD domain [37] (spanning residues 115 to 465 in human SAMHD1 [35]) is highly conserved and required for dNTPase and viral restriction activities [7,8,[38], [39], [40], [41]]. In the HD domain of human SAMHD1, the R451 and L453 residues constitute an RXL motif (where X represents any amino acid). This RXL motif is required for SAMHD1 tetramerization, as well as its dNTPase and retroviral restriction functions, achieved through its interaction with cell cycle regulators such as cyclin A2, cyclin-dependent kinase 1 (CDK1), or CDK2 (Fig. S1) [42]. We identified an RXL motif in an equivalent position in VEN4 (R335 and L337) as well as in its plant orthologs, except for tomato (Solanum lycopersicum), which has an I residue instead of an L (Fig. S2). However, this is a conservative substitution, since I and L are similar aliphatic amino acids. Other key residues of the human SAMHD1 HD domain are conserved in its plant orthologs, including K354 (K248 in VEN4), which is required for DNA DSB repair through HR. Moreover, the deacetylation of K354 by Sirtuin 1 (SIRT1) allows SAMHD1 to function in DSB repair [35].

The negatively charged E249 residue of VEN4, which is mutated to a nonpolar L amino acid in the ven4-0 mutant, corresponds to E355 in human SAMHD1 (Fig. 1 and Fig. S1) [27]. This residue is also E in the tomato ortholog of SAMHD1 and VEN4, but D in the orthologs from the other plant species that we used in our multiple sequence alignment (Fig. S2). However, E and D are similar, negatively charged amino acids. Interestingly, the E249 residue of VEN4 is adjacent to K248, the equivalent residue of K354 in human SAMHD1 (Fig. 1 and Fig. S1) [27], which is involved in DNA DSB repair by HR³⁵, as previously mentioned.

Outside of the HD domain, another conserved residue, T592 (T464 in VEN4), is phosphorylated by the cyclin A2/CDK complex and participates in negative regulation of HIV-1 restriction and MRE11-dependent resection of stalled DNA replication forks [15,43,44]. K484 (K374 in VEN4) is crucial for interaction with CtIP and its recruitment to DSBs to promote DNA end resection by HR [14,45].

As expected from their almost identical primary structures (Fig. S1), the predicted secondary structures of Arabidopsis VEN4 and human SAMHD1 displayed high three-dimensional conservation, except for their N- and C-terminal regions (Fig. 1). In fact, both proteins showed a root mean square deviation (RMSD) of 0.897 Å for 371 pruned (best-matching) atom pairs, while the RMSD across all 457 atom pairs was 7.961 Å.

2.2. ven4 mutants are hypersensitive to hydroxyurea

Ribonucleotide reductase (RNR) catalyzes the limiting step of the de novo dNTP biosynthesis pathway in eukaryotes [46]. In Arabidopsis, the RNR R1 subunit is encoded by the RNR1 gene [47,48], and the RNR R2 subunit by three paralogous genes: RNR2A, RNR2B, and TSO MEANING 'UGLY' IN CHINESE 2 (TSO2), with TSO2 being the primary contributor to RNR activity. The rice thermosensitive conditional mutants virescent3 (v3) and stripe1 (st1) are mutated at the V3 and ST1 genes, which encode the R1 and R2 subunits of RNR, respectively, and display chlorotic leaves with green stripes [49].

Rice v3 and Arabidopsis rnr2a mutants are hypersensitive to hydroxyurea, a genotoxic agent that inhibits DNA replication by blocking RNR activity, hence reducing the available dNTP levels. Hydroxyurea also produces stalled DNA replication forks with single-strand breaks (SSBs) and DSBs, which ultimately activate ATAXIA TELANGIECTASIA-MUTATED AND RAD3-RELATED (ATR) for cell-cycle arrest and DNA repair [[49], [50], [51], [52], [53]]. Indeed, Arabidopsis atr mutants also exhibit hypersensitivity to hydroxyurea and other DNA-replication-blocking agents [51]. Expression of the Arabidopsis RNR1, RNR2A, and RNR2B genes is induced by hydroxyurea and TSO2 expression is induced by the DSB inducer bleomycin, while the tso2-1 rnr2a-1 double mutant accumulates DSBs even in the absence of genotoxins [52,54].

We grew the ven4-0, ven4-2, and ven4-3 mutants (see Methods) on medium supplemented with 1 mM hydroxyurea; all plants were hypersensitive to this DNA-replication-blocking agent, showing severely altered morphology: decreased rosette size and increased leaf reticulation (Fig. 2A–J) as well as inhibited primary root growth (Fig. 2K and Fig. S3) compared with their respective wild types and with their growth in the absence of hydroxyurea. We also observed increased beta-glucuronidase (GUS) enzymatic activity in Col-0 VEN4_pro:GUS transgenic plants when they were grown on medium containing 1 mM hydroxyurea (Fig. 2L–O), indicating that VEN4 expression is responsive to this compound, similar to RNR1, RNR2A, and RNR2B [31,50,52,53,55]. Taken together, these results further support a role for VEN4 in dNTP metabolism, previously proposed by Xu et al. [29], Lu et al. [31], and Sarmiento-Mañús et al. [27], and suggest an additional direct or indirect function in the DNA damage response (DDR).

Fig. 2.

ven4 mutants are hypersensitive to hydroxyurea but do not show altered chloroplast genome replication. (A–J) Rosettes of the ven4-0, ven4-2, and ven4-3 mutants and their Ler and Col-0 wild types grown on medium not supplemented (A–E) or supplemented with 1 mM hydroxyurea (F–J). Photographs were taken at 19 days after stratification (das). Scale bars: 2 mm. (K) Percentage of root growth inhibition in the ven4-0, ven4-2 and ven4-3 mutants, and their Ler and Col-0 wild types, grown for 19 das on vertically oriented Petri dishes containing medium supplemented with 1 mM hydroxyurea. Error bars indicate standard deviation. Primary root length was measured from at least 10 plants. Asterisks indicate values significantly different from the corresponding wild type in a Student's t-test (∗P < 0.001). (L–O) GUS staining of Col-0 VEN4_pro:GUS transgenic plants grown on medium not supplemented (L and M) or supplemented with 1 mM hydroxyurea (N and O). Photographs were taken at 11 das. Scale bars: 2 mm. (P) Chloroplast-to-nuclear genome copy ratio in Ler, ven4-0, Col-0, and ven4-2, calculated according to Yoo et al. [49] The ratio obtained for the Ler and Col-0 wild types was set to 1. Error bars indicate the interval delimited by $2^{–\Delta \Delta {\mathrm{C}}_{\mathrm{T}}\pm \text{SD}}$, where SD is the standard deviation of the ΔΔC_T values. Three biological replicates were examined per genotype, with three technical replicates in each experiment. The ΔC_T values were not significantly different from those of the corresponding wild type in a Student's t-test.

At restrictive growth conditions, the v3 and st1 rice mutants display similar chloroplast aberrations to those found in Arabidopsis ven4 mutants [27,28], exhibiting low chlorophyll content and a reduced chloroplast-to-nuclear genome copy ratio [49]. An explanation has been proposed for these phenotypes: when the dNTP pool is limited, replication of the nuclear genome is likely to be prioritized over that of the chloroplast genome for plant survival [47,49]. However, when we studied the chloroplast-to-nuclear genome copy ratio by qPCR amplification of unique sequences from the chloroplast and nuclear genomes in DNA extracted from ven4-0 and ven4-2 plants collected 15 days after stratification (das), we did not find a statistically significant reduction compared with their Ler and Col-0 wild types (Fig. 2P).

2.3. DNA DSB repair by HR is impaired in ven4 mutants

We next aimed to determine whether Arabidopsis VEN4 participates in DNA DSB repair by HR, like its human SAMHD1 ortholog. To this end, we used the IU.GUS 8 transgene, which was designed for examining somatic HR by the synthesis-dependent strand annealing (SDSA) mechanism. The IU.GUS 8 transgene contains two overlapping sequences of a non-functional GUS gene: a sequence with an internal region of the GUS gene inverted (IU), and the GUS gene lacking this internal region (GU-US). In this assay, a GUS-stained blue spot appears as a result of a DSB in the GU-US sequence if it is repaired by HR, using the IU sequence in the sister chromatid of plants homozygous for the IU.GUS 8 transgene [56,57].

The GUS reporter assay was performed using Hyg^R ven4-0 and ven4-2 F₃ plants (the IU.GUS 8 construct confers resistance to the hygromycin B antibiotic) derived from crosses of ven4-0, and ven4-2 to the IU.GUS 8 transgenic line (all in the Col-0 background), which were grown in the presence or absence of 1 mM hydroxyurea as an indirect inducer of DNA DSBs. In the presence of hydroxyurea, we observed on average 1.37 and 0.16 blue sectors per plant in ven4-0 and ven4-2 homozygous plants, respectively, and 3.77 and 1.16 in their phenotypically wild-type (VEN4/VEN4 or VEN4/ven4) siblings (Fig. 3A–C). We observed 63.76% fewer blue sectors in ven4-0, compared with the Ler/Col-0 hybrid, and 86.2% fewer blue sectors in ven4-2 compared with their phenotypically wild-type siblings after inducing DNA damage with hydroxyurea. In the absence of hydroxyurea, these mutants also showed fewer blue sectors (34.7% fewer in ven4-0 and 75.0% fewer in ven4-2) (Fig. 3A–C). These observations indicate that repair of both spontaneous and hydroxyurea-induced DSBs is impaired in the ven4 mutants. These results support the hypothesis that VEN4 plays a role in DNA DSB repair by HR.

Fig. 3.

ven4 mutants display signs of DNA damage and impaired DNA repair by HR. (A–C) Spontaneous and hydroxyurea-induced HR events in the ven4-0 and ven4-2 mutants. (A and B) Distribution of blue sectors visualized by GUS staining in ven4 or phenotypically wild-type (VEN4/VEN4 or VEN4/ven4) F₃ plants derived from ven4-0 (in a Ler genetic background) × IU.GUS 8 (Col-0) (A) and ven4-2 (Col-0) × IU.GUS 8 (Col-0) (B) crosses; plants were grown on medium not supplemented or supplemented with 1 mM hydroxyurea. Plants were collected at 12 das. (C) Mean number of blue sectors per F₃ plant. Error bars indicate standard deviation. At least 100 plants from two different F₃ families were analyzed. Asterisks indicate values significantly different from those of the corresponding phenotypically wild-type F₃ plants in a Student's t-test (∗P < 0.001). (D) Nuclear ploidy distribution in third-node leaves of Ler, ven4-0, Col-0, and ven4-2 plants. Error bars indicate standard deviation. Three biological replicates of 100,000 events each were analyzed per genotype. Leaves were collected at 15 das. (E) RT-qPCR analysis of the relative expression of BRCA1, PARP1, RAD51, and PARP2 in Ler, ven4-0, Col-0, and ven4-2. Error bars indicate the interval delimited by $2^{–\Delta \Delta {\mathrm{C}}_{\mathrm{T}}\pm \text{SD}}$, where SD is the standard deviation of the ΔΔC_T values. Three biological replicates were examined per genotype, with three technical replicates in each experiment. The ACT2 gene served as an internal control for normalization and relative quantification of gene expression. Asterisks indicate ΔC_T values significantly different from those of the corresponding wild type in a Student's t-test (∗P < 0.05).

2.4. Endoreduplication is increased and DNA repair genes are downregulated in ven4 mutants

DNA damage has been associated with a reduced division rate in animal and plant cells because cell division halts until specific DNA repair mechanisms are activated (reviewed by Sancar et al. [58] and Toettcher [59]). Cells from certain plants, such as Arabidopsis, use endoreduplication as a response to DNA damage, including DSBs [60], a mechanism that is used less frequently by animal cells (reviewed by Fox and Duronio [61] and Nisa et al. [62]). As DNA DSB repair by HR is impaired in ven4 mutants, we examined their ploidy levels. We found lower 2C and 4C ploidy levels and higher 16C levels in ven4-0 and ven4-2 cells compared with Ler and Col-0 cells, respectively (Fig. 3D), resulting in an endoreduplication index [63,64] that was 19% and 10% greater than their wild types, respectively.

Expression of TSO2, encoding the major contributor to Arabidopsis RNR activity, is induced by bleomycin, a genotoxic agent that causes DNA DSBs, but not by hydroxyurea [52,65]. TSO2 is primarily nuclear in meristematic cells and cytoplasmic in adult cells. However, in quiescent cells, TSO2 is not detected under normal conditions but accumulates in the nucleus after exposure to ultraviolet (UV)-C radiation, suggesting a role for this protein in DNA damage protection and/or repair [65]. Supporting this hypothesis, the POLY (ADP-RIBOSE) POLYMERASE 1 (PARP1), PARP2, and RADIATION SENSITIVE 51 (RAD51, which encodes a recombinase) genes, which play key roles in DNA DSB repair, are up-regulated in tso2-1 rnr2a double mutant plants [54]. BREAST CANCER SUSCEPTIBILITY1 (BRCA1), RAD51, PARP1, and PARP2 are also up-regulated when DNA damage is induced and in several Arabidopsis mutants with impaired DNA repair activity, such as photoperiod-independent early flowering1-3 (pie1-3), actin-related protein6-3 (arp6-3), SWR1 complex6-1 (sw6-1), jing he sheng 1 (jhs1), or retinoblastoma-related 1 (rbr1) [[66], [67], [68]].

As previously mentioned, we observed that ven4 mutants are hypersensitive to hydroxyurea and show reduced ability to repair DSBs by HR. In addition, since human SAMHD1 is involved in the resolution of DSBs by both HR¹⁴ and NHEJ [69], it is likely that VEN4 also has a role in DNA repair. To study whether DDR is altered in ven4 mutants, we analyzed the expression of some of the above-mentioned key DNA repair genes by reverse transcription quantitative PCR (RT-qPCR). Interestingly, the mRNA levels of the DNA damage-induced genes BRCA1, RAD51, PARP1, and PARP2 were significantly lower in ven4-0 and ven4-2 mutants than in their Ler and Col-0 wild types when grown in the absence of genotoxic drugs, which suggests a direct or indirect role for VEN4 in the basal induction of these genes (Fig. 3E).

2.5. R-loops do not accumulate in ven4 mutants

Recent work showed that human SAMHD1 functions as a tumor suppressor by resolving R-loops that often form at regions of the genome where transcription and replication conflict. Indeed, even in the absence of DNA damage inducers, R-loops accumulate in cells lacking SAMHD1 function, increasing replication stress and genome instability, and triggering DDRs [16].

To determine whether this function of human SAMHD1 is conserved in VEN4, we performed a dot blot analysis using the well-established S9.6 monoclonal antibody [70] to detect R-loops in total DNA extracted from ven4-0 and ven4-2 plants grown in the presence or absence of 1 mM hydroxyurea and collected at 19 das (Fig. 2A–J). Given the previously described cross-reactivity of the S9.6 antibody with single-stranded (ss) or double-stranded (ds) RNAs [71], we employed RNase T1 and RNase III to cleave ssRNAs and dsRNAs, respectively. Additionally, we used RNase H, which specifically cleaves RNAs within R-loops, thus ensuring the specificity of the S9.6 antibody signal. Simultaneously, we employed an anti-dsDNA antibody as a loading control. However, unlike in human cells lacking SAMHD1 function, there appeared to be no substantial accumulation of R-loops in the ven4-0 and ven4-2 mutants compared with their Ler and Col-0 wild types, even when grown in the presence of the DNA-replication-blocking agent hydroxyurea (Fig. S4).

2.6. The ven4-0 and tso2-1 mutants have less fumarate and malate

Fumarate, malate, and succinate have been identified as competitive inhibitors of some α-ketoglutarate-dependent enzymes involved in DDRs across human, yeast, and bacterial cells [[72], [73], [74], [75], [76]]. Fumarate has also been identified as a defense-priming agent against pathogens in Arabidopsis [77], but its association with the DDR has not yet been established; nor has that of malate or succinate.

Interestingly, metabolite profiling analysis of the ven4-0 mutant detected lower levels of fumarate and malate than in the Ler wild type. The tso2-1 mutant, carrying a loss-of-function allele of TSO2, also exhibited lower levels of fumarate, malate, and succinate than Ler (Table S2). These findings support the hypothesis of a direct or indirect role for VEN4 in the DDR and additionally point to the conserved functions of fumarate, malate, and/or succinate in DNA repair in plants.

3. Discussion

3.1. Arabidopsis VEN4 and human SAMHD1 might play similar roles not only in dNTP metabolism but also in maintenance of genome stability

Here, we examined VEN4, the likely Arabidopsis ortholog of human SAMHD1 [[27], [28], [29]] (Fig. 1 and S1), which is highly conserved in plants, but lacks the SAM domain present in SAMHD1 (Fig. S2). Since conservation of the dNTPase-dependent roles of SAMHD1 and VEN4 has already been stablished [[27], [28], [29]], we aimed to determine whether VEN4 has also a dNTPase-independent role, participating in the maintenance of genome stability, particularly in DNA DSB repair by HR and resolution of R-loops, similar to human SAMHD1 [[13], [14], [15], [16],35,45].

DNA DSBs are considered the most harmful genetic lesions in animal and yeast cells; they can be induced by exposure to exogenous genotoxic agents or chemicals, such as ionizing radiation or hydroxyurea, but can also arise spontaneously due to cellular metabolism (reviewed by Chapman et al. [78]). In animals, DDRs ensure genome integrity by arresting cell cycle progression to aid DNA repair and, when necessary, by inducing apoptosis. In vascular plants, endoreduplication enables cells to duplicate their genomes without division, thereby increasing ploidy levels. This mechanism effectively prevents the transmission of damaged DNA to daughter cells [79]. For instance, in Arabidopsis, endoreduplication is activated in response to DNA DSBs, but not in direct response to replication stress [60]. Indeed, in the absence of DNA damage inducers, we found that ven4 mutants exhibit increased ploidy levels in leaf cells, suggesting an over-accumulation of spontaneous DSBs.

In eukaryotes, transcription of genes encoding RNR subunits is activated in the S phase of the cell cycle and in response to DNA damage (reviewed by Guarino et al. [80]). Furthermore, to overcome the accumulation of DNA lesions, DDR genes are induced under genotoxic conditions [52]. Using a VEN4_pro:GUS transgene, we detected transcriptional induction of VEN4 upon hydroxyurea treatment, supporting the hypothesis that VEN4 acts in DNA repair. Interestingly, the ven4 mutants exhibited reduced expression of key genes involved in DSB repair (BRCA1, RAD51, PARP1, and PARP2) compared with Ler or Col-0, when grown in culture medium not supplemented with hydroxyurea, suggesting a direct or indirect role for VEN4 in the basal transcriptional induction of these genes. Indeed, IU.GUS transgenic lines in the ven4 mutant background showed impaired HR when grown in the absence of genotoxic drugs, which worsened following supplementation with 1 mM hydroxyurea, pointing to a role for VEN4 in spontaneous or induced DSB repair by HR, similar to its human ortholog.

Comparing the amino acid sequences and domain structures, along with the understanding of SAMHD1 function, can help us infer how VEN4 functions, and which of these functions are conserved with SAMHD1. Deacetylation of the K354 residue of human SAMHD1, located in the middle of its HD domain, occurs in response to DSBs. Deacetylation of K354 facilitates SAMHD1 binding to ssDNA at DSBs, which also facilitates CtIP binding to ssDNA in order to promote DNA end resection and HR [35]. Interestingly, the ven4-0 mutation causes a E249L substitution in the residue adjacent to the K248 of VEN4, which is equivalent to the K354 of human SAMHD1 and is highly conserved in other plants. In silico predictions of the effects of the E249L substitution on VEN4 and of the equivalent E355L mutation on human SAMHD1, revealed an increase in conformational stability and a decrease in flexibility [27], which might impair the deacetylation of K248 (K354 in human SAMHD1) and the ability of the mutant proteins to bind ssDNA at DSBs. This hypothesis is consistent with the over-accumulation of both spontaneous and induced DSBs suggested by the impaired HR revealed by our IU.GUS assays, as well as by the reduced expression of genes involved in DNA DSB repair.

3.2. Fumarate and malate depletion in the ven4-0 mutant also supports the implication of VEN4 in the DDR

Fumarase is a highly conserved enzyme best known for catalyzing the hydration of fumarate to L-malate in the tricarboxylic acid (TCA) or Krebs cycle in mitochondria, facilitating a transition step in the production of energy for the cell. However, a cytoplasmic echoform [81] of this enzyme has been found in eukaryotes. Initially proposed to function as a scavenger of fumarate from the urea cycle and amino acid catabolism, this fumarase echoform has also been associated with the cellular DDR to DSBs (reviewed by Leshets et al. [82]). This role requires the movement of cytoplasmic fumarase to the nucleus in response to DNA damage, where it catalyzes the reverse conversion of L-malate to fumarate, leading to a local accumulation of fumarate at DSBs, which promotes their repair via HR or NHEJ by inhibiting some α-ketoglutarate-dependent histone demethylases [75,76].

Indeed, yeast and human cells lacking fumarase are hypersensitive to some DNA DSB and replication-block inducers, such as ionizing radiation or hydroxyurea [74]. The fumarase of the prokaryote Bacillus subtilis also appears to participate in the DDR, achieved in this case by the biosynthesis of L-malate, which affects the translation and subcellular localization of the first protein to be recruited to DSB sites [73]. Similar to fumarate, succinate appears to play a role in the control of epigenetic marks following DNA damage in yeast and human cells, but also in Escherichia coli [72,75]. By acting as structural mimics to α-ketoglutarate, fumarate and succinate competitively inhibit certain α-ketoglutarate-dependent enzymes involved in the successful repair of damaged DNA. Consequently, they have been defined as oncometabolites (reviewed by Liu and Yang [83]).

The ven4-0 point mutation has been particularly useful for predicting the effects of VEN function in DNA DSB repair, strongly reinforcing our other experimental results. In particular, we found a depletion of fumarate and malate in this mutant, which may be partly related to the reduced expression of key DNA repair genes when plants are grown in medium not supplemented with hydroxyurea. This hypothesis warrants further research, especially considering that the loss of function of fumarase has been found to affect the transcriptional induction of many DNA-damage-responsive genes in E. coli [72].

3.3. VEN4 does not appear to prevent R-loop accumulation

Human SAMHD1 resolves R-loops at genomic regions where the transcription and replication machineries collide [15,16]. This role of SAMHD1 is independent of its dNTPase activity. Our results support the hypothesis of a conserved role for VEN4 in maintaining genome integrity, specifically in DNA DSB repair by HR. However, we failed to obtain evidence that VEN4 resolves R-loops, even upon genotoxic treatment with hydroxyurea. This may be the correct result, or the dot blots may only have enough sensitivity to detect major changes in R-loop accumulation and not small and/or localized changes in R-loop formation at transcription–replication conflict regions of the genome. Thus, despite our negative results, we cannot definitively conclude that VEN4 does not participate, directly or indirectly, in the resolution of R-loops at these conflictive genomic regions, as SAMHD1 does in humans.

3.4. Concluding remarks

Therefore, in summary, we show that, like SAMHD1, Arabidopsis VEN4 functions in DNA repair. However, whether it functions in resolving R-loops remains to be determined by additional studies. Examining how the differing domain structures of the two proteins (VEN4 lacks the SAM domain present in SAMHD1 but contains other conserved domains) affect their respective functions also remains an intriguing question for future research. Finally, it will be interesting to decipher the mechanism by which these proteins function in DNA repair, and how the lack of VEN4 affects plant development. In addition to opening these new avenues of inquiry, our study provided insight on the conservation of DNA repair mechanisms and their relationship to dNTP metabolism, in eukaryotes.

5. Star★methods

5.1. Key resources table

5.1.1. Resource availability

5.1.1.1. Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, María Rosa Ponce (mrponce@umh.es).

5.1.1.2. Materials availability

Materials generated in this study will be made available on request. For further details contact the lead contact.

5.1.1.3. Data and code availability

Sequence data from this article can be found at TAIR (http://www.arabidopsis.org) under the following accession numbers: VEN4 (At5g40270), TSO2 (At3g27060), BRCA1 (At4g21070), RAD51 (At5g20850), PARP1 (At2g31320), PARP2 (At4g02390), and ACT2 (At3g18780). This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

6. Experimental model and study participant details

The Arabidopsis thaliana (L.) Heynh. wild-type accessions Col-0 and Ler, and the T-DNA insertional mutants ven4-2 (SALK_077401) and ven4-3 (SALK_131986), in the Col-0 genetic background, were originally obtained from the Nottingham Arabidopsis Stock Centre (NASC; Nottingham, United Kingdom) and then propagated in the laboratory for further analysis. The EMS-induced ven4 mutant, in the Ler genetic background, was isolated in the laboratory of José Luis Micol [23,[84], [85], [86]] and renamed recently as ven4-0 by Sarmiento-Mañús et al. [27] Seeds of the IU.GUS 8 transgenic line [57], in the Col-0 genetic background, were kindly provided by Ortrun Mittelsten Scheid (Gregor Mendel Institute of Molecular Plant Biology, Austrian Academy of Sciences, Vienna BioCenter, Vienna, Austria), and tso2‐1 [54] seeds, in the Ler genetic background, by Zhongchi Liu (University of Maryland, College Park, MD, USA).

7. Method details

7.1. Plant growth conditions and genotyping

Except for metabolite profiling (see below), plants were grown under sterile conditions on half-strength Murashige and Skoog (MS; Duchefa Biochemie) medium containing 1% (w/v) sucrose (Duchefa Biochemie), 0.1% (w/v) 2-Morpholinoethanesulfonic acid (MES) monohydrate (Duchefa Biochemie), and 0.7% (w/v) plant agar (Duchefa Biochemie) at 20°C ± 1°C, 60–70% relative humidity, and continuous fluorescent light of ∼75 μmol/m²/s, as previously described by Ponce et al. [87] and Berná et al. [23] When required, MS medium was supplemented with 15 μg/mL of hygromycin B (Invitrogen). Mutants were genotyped by PCR amplification and/or Sanger sequencing, as previously described by Sarmiento-Mañús et al. [27], using the primers shown in Table S1. Unless specified otherwise, all plants employed in this work were homozygous for the indicated mutations.

7.2. Protein structure visualization and bioinformatic analyses

The 3D structures of the Arabidopsis VEN4 and human SAMHD1 proteins shown in Fig. 1 were obtained from the AlphaFold Protein Structure Database (AlphaFold DB; Jumper et al. [88]; Varadi et al. [89]; https://alphafold.ebi.ac.uk/), where they are identified as AF-Q9FL05-F1 and AF-Q9Y3Z3-F1, respectively, and visualized using UCSF ChimeraX 1.2.5 (Goddard et al. [90]; Pettersen et al. [91]; https://www.rbvi.ucsf.edu/chimerax/). Root mean square deviations (RMSD) for both proteins were determined with the MatchMaker function of UCSF ChimeraX 1.2.5, using default parameters. The multiple sequence alignments shown in Figs. S1 and S2 were obtained using ClustalX 1.5b [92,93]. The identical and similar residues among the analyzed protein sequences were shaded in black and gray, respectively, using BoxShade (https://junli.netlify.app/apps/boxshade/).

7.3. Hydroxyurea treatment

The hydroxyurea hypersensitivity of the ven4 mutants was measured as the percentage of primary root growth inhibition. Plants were grown for 19 das on vertically oriented Petri dishes with half-strength MS medium supplemented or not with 1 mM hydroxyurea. Primary root length was measured from photographs taken of plants grown on those Petri dishes using the NIS Elements AR 3.1 image analysis software (Nikon). To analyze rosette morphological phenotypes, plants were grown for 19 das on horizontally oriented Petri dishes with half-strength MS medium supplemented or not with 1 mM hydroxyurea.

7.4. GUS staining and analysis of IU.GUS transgenic plants

Plants homozygous for the VEN4_pro:GUS transgene in a Col-0 genetic background [27] were collected at 11 das from two different T₂ families grown on half-strength MS medium supplemented or not with 1 mM hydroxyurea. For DNA repair analyses, Col-0, ven4-0, and ven4-2 plants were crossed with the IU.GUS 8 transgenic line [57], and 100 Hyg^R seedlings (50 phenotypically mutant and 50 phenotypically wild type) from two different F₃ families derived from each cross were collected at 12 das. GUS enzymatic activity in VEN4_pro:GUS and IU.GUS transgenic plants was visualized as previously described by Robles et al. [94] Blue sectors in IU.GUS transgenic plants were counted using a MZ6 stereomicroscope (Leica) equipped with a DXM1200 digital camera (Nikon).

7.5. Estimation of chloroplast-to-nuclear genome copy ratio

Chloroplast-to-nuclear genome copy ratios were estimated by qPCR using genomic DNA from Ler, Col-0, ven4-0, and ven4-2 plants at 15 das and the At4g04930_F/R and ArthCp025_F/R primer pairs (Table S1) for amplification of the nuclear At4g04930 and chloroplast ArthCp025 single-copy genes, respectively, as first described by Garton et al. [47] The relative ratios of chloroplast-to-nuclear genome copies were subsequently calculated according to Yoo et al. [49], normalizing to the wild-type values, which were set to 1.

7.6. Flow cytometry

Nuclei were extracted from plants collected 7 das as previously described by Horiguchi et al. [95] Briefly, first- and second-node leaves were chopped with a razor blade in 500 μL of cold (kept on ice) Galbraith's nuclear extraction buffer [96]. The cell suspension was then filtered through a 30 μm pore size nylon mesh, treated with 100 μg/mL of RNase A (Roche), and stained with 50 μg/mL of propidium iodide (PI; Sigma-Aldrich) for 10 min. Nuclear DNA content was measured by flow cytometry using a SH800S Cell Sorter (Sony). Three biological replicates of 100,000 events each were analyzed per genotype, and the endoreduplication index (EI) was calculated as EI = (0 × %2C) + (1 × %4C) + (2 × %8C) + (3 × %16C) + (4 × %32C), as previously described by Sterken et al. [63].

7.7. Total RNA isolation and RT-qPCR experiments

For gene expression analysis, total RNA from plant rosettes harvested at 10 das was extracted with TRIzol (Invitrogen). Reverse transcription and qPCR amplification were performed as described by Wilson-Sánchez et al. [97] qPCRs were carried out in a Step-One Real-Time PCR System (Applied Biosystems) using three technical replicates per biological replicate (each consisting of three plant rosettes). The primers used are listed in Table S1. The housekeeping gene ACTIN2 (ACT2) served as an internal control for normalization and relative quantification of gene expression. The C_T values obtained, detailed in Table S3, were normalized using the $2^{-\Delta \Delta {\mathrm{C}}_{\mathrm{T}}}$ method [98].

7.8. R-loop recognition by the S9.6 monoclonal antibody

Genomic DNA from ∼200 mg of plant rosettes harvested at 19 das was extracted using a DNeasy Plant Mini Kit (Qiagen) according to the manufacturer's instructions, except that use of the RNase A included in the kit was omitted. The isolated DNA was quantified using a Qubit 2.0 Fluorometer and a Qubit dsDNA High Sensitivity (HS) Assay Kit (Invitrogen). Samples of 1.2 μg of DNA were treated with either 5 U of RNase H (New England Biolabs), 1000 U of RNase T1 (Thermo Scientific), or 1 U of Ambion RNase III (Invitrogen) for 2 h at 37°C to assess the specificity of the anti-DNA–RNA hybrid (S9.6) antibody (AB_2861387; Sigma-Aldrich). An untreated sample was also included. Samples of 25, 50, or 100 ng of DNA (in a volume of 2 μL), either treated or untreated with RNases, were blotted onto two positively charged nylon membranes, one for the S9.6 antibody and the other for the anti-dsDNA antibody (AB_470907; Abcam). The samples were allowed to saturate into the membranes before crosslinking with UV light (1200 μJ × 100). The membranes were blocked with 5% (w/v) skimmed milk in 1× Tris-Buffered Saline solution with 0.05% (v/v) Tween-20 (1× TBST) for 1 h at room temperature on a shaker. After three subsequent washes with 1× TBST, the membranes were incubated overnight at 4°C on a shaker with either the S9.6 or the anti-dsDNA antibody, at 1:1000 or 1:10,000 dilution, respectively, in 1× TBST. Following three washes with 1× TBST, the membranes were incubated with WesternSure goat anti-mouse IgG HRP-conjugated secondary antibody (AB_2721263; LI-COR), diluted at 1:50,000 in 1× TBST, for 2 h at room temperature on a shaker. After rinsing the membranes in 1× TBST, antibodies were detected using WesternSure PREMIUM Chemiluminescent Substrate (LI-COR) and a C-Digit Blot Scanner (LI-COR).

7.9. Metabolite profiling

The first leaves of Ler, ven4-0, and tso2-1 plants grown on rock wool at 22°C under a long-day photoperiod (16-h light/8-h dark) were collected at 14 das from three biological replicates. After harvesting, 20–80 mg of tissue was immediately frozen in liquid nitrogen and stored at −80°C until further use. Metabolite profiling was performed by capillary electrophoresis time-of-flight mass spectrometry (CE‐TOF MS), as described by Oikawa et al. [99,100] and Ferjani et al. [101] The raw CE-TOF MS data were processed, and metabolite peaks were identified, aligned, and annotated using MasterHands [102]. Metabolites with significantly different levels between Ler and ven4-0 or tso2-1 plants were identified using a Student's t-test (Table S2).

Quantification and statistical analysis

Quantification and statistical analyses are described in the Method Details and Results sections as well as in the figure legends.

CRediT authorship contribution statement

Raquel Sarmiento-Mañús: Writing – original draft, Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Sara Fontcuberta-Cervera: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Kensuke Kawade: Writing – review & editing, Methodology, Investigation, Formal analysis. Akira Oikawa: Writing – review & editing, Methodology, Investigation, Formal analysis. Hirokazu Tsukaya: Writing – review & editing, Supervision, Investigation, Formal analysis, Conceptualization. Víctor Quesada: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. José Luis Micol: Writing – review & editing, Writing – original draft, Supervision, Resources, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization. María Rosa Ponce: Writing – review & editing, Writing – original draft, Supervision, Resources, Methodology, Investigation, Funding acquisition, Formal analysis, Conceptualization.

Funding

This work was supported by the Ministerio de Ciencia e Innovación of Spain (PGC2018-093445-B-I00 and PID2020-117125RB-I00 [MCI/AEI/FEDER, UE] to M.R.P, and PID2021-127725NB-I00 [MCI/AEI/FEDER, UE] to J.L.M.), the Generalitat Valenciana (CIPROM/2022/2 to M.R.P. and J.L.M.), and the Exploratory Research Center on Life and Living Systems of Japan (BIO-NEXT project to K.K. and H.T.).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is/are the supplementary data to this article.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-DNA–RNA hybrid (S9.6) antibody	Sigma-Aldrich	Cat#MABE1095; RRID: AB_2861387
Anti-dsDNA antibody	Abcam	Cat#ab27156; RRID: AB_470907
WesternSure goat anti-mouse IgG HRP-conjugated secondary antibody	LI-COR	Cat#926–80010; RRID: AB_2721263
Chemicals, peptides, and recombinant proteins
Murashige and Skoog (MS) medium	Duchefa Biochemie	Cat#M0221
Sucrose	Duchefa Biochemie	Cat#S0809; CAS: 57-50-1
2-Morpholinoethanesulfonic acid (MES) monohydrate	Duchefa Biochemie	Cat#M1503; CAS: 145224-94-8
Plant agar	Duchefa Biochemie	Cat#P1001; CAS: 9002-18-0
Hygromycin B (50 mg/mL)	Invitrogen	Cat#10687010
Hydroxyurea	Sigma-Aldrich	Cat#H8627; CAS: 127-07-1
Sodium phosphate monobasic monohydrate	USB	Cat#S5206 CAS: 10049-21-5
Sodium phosphate dibasic heptahydrate	Fisher BioReagents	Cat#BP331; CAS: 7782-85-6
Triton X-100	Sigma-Aldrich	Cat#T8787; CAS: 9036-19-5
Potassium hexacyanoferrate(II) trihydrate	Sigma-Aldrich	Cat#P9387; CAS: 14459-95-1
Potassium hexacyanoferrate(III)	Sigma-Aldrich	Cat#P8131; CAS: 13746-66-2
5-Bromo-4-chloro-3-indolyl beta-D-glucuronic acid cyclohexylammonium salt (X-Gluc, X-Glucuronide)	X-Gluc Direct	CAS: 114162-64-0
N,N-Dimethylformamide (DMF)	PanReac AppliChem	Cat#131785; CAS: 68-12-2
Ethanol absolute pure	PanReac AppliChem	Cat#141086; CAS: 64-17-5
RNase A from bovine pancreas	Roche	Cat#10109169001
Magnesium chloride hexahydrate	Supelco	Cat#105833; CAS: 7791-18-6
Trisodium citrate dihydrate	Duchefa Biochemie	Cat#S0521; CAS: 6132-04-3
4-Morpholinepropanesulfonic acid (MOPS)	Duchefa Biochemie	Cat#M1502; CAS: 1132-61-2
Propidium iodide (PI)	Sigma-Aldrich	Cat#81845; CAS: 25535-16-4
TRIzol Reagent	Invitrogen	Cat#15596026
Chloroform	Sigma-Aldrich	Cat#C2432; CAS: 67-66-3
Maxima Reverse Transcriptase (200 U/μL)	Thermo Scientific	Cat#EP0742
RiboLock RNase Inhibitor (40 U/μL)	Thermo Scientific	Cat#EO0381
Maxima SYBR Green/ROX qPCR Master Mix (2×)	Thermo Scientific	Cat#K0221
RNase H (5 U/μL)	New England Biolabs	Cat#M0297S
RNase T1 (1000 U/μL)	Thermo Scientific	Cat#EN0541
Ambion RNase III (1 U/μL)	Invitrogen	Cat#AM2290
Tris ultrapure	Duchefa Biochemie	Cat#T1501; CAS: 77-86-1
Sodium chloride	Duchefa Biochemie	Cat#S0520; CAS: 7647-14-5
Hydrochloric acid 37% (v/v)	PanReac AppliChem	Cat#131020; CAS: 7647-01-0
Tween 20	Sigma-Aldrich	Cat#P7949; CAS: 9005-64-5
WesternSure PREMIUM Chemiluminescent Substrate	LI-COR	Cat#926-95000
Critical commercial assays
TURBO DNA-free Kit	Invitrogen	Cat#AM1907
DNeasy Plant Mini Kit	Qiagen	Cat#69104
Qubit dsDNA High Sensitivity (HS) Assay Kit	Invitrogen	Cat#Q32851
Experimental models: Organisms/strains
Arabidopsis wild-type accession Col-0	NASC	NASC ID: N70000
Arabidopsis wild-type accession Ler	NASC	NASC ID: NW20
Arabidopsis EMS-induced ven4-0 mutant	Berná et al. [23]	N/A
Arabidopsis T-DNA insertional ven4-2 mutant	NASC	NASC ID: N577401
Arabidopsis T-DNA insertional ven4-3 mutant	NASC	NASC ID: N664024
Arabidopsis EMS-induced tso2-1 mutant	Wang and Liu [54]	N/A
Arabidopsis Col-0 IU.GUS transgenic plants	Orel et al. [57]	N/A
Arabidopsis Col-0 VEN4pro:GUS transgenic plants	Sarmiento-Mañús et al. [27]	N/A
Arabidopsis ven4-0 IU.GUS transgenic plants	This work	N/A
Arabidopsis ven4-2 IU.GUS transgenic plants	This work	N/A
Oligonucleotides
Primer random p(dN)6	Roche	Cat#11034731001
For the other primers used in this work, see Table S1	Sigma-Aldrich	N/A
Software and algorithms
UCSF ChimeraX 1.2.5	Goddard et al. [90]; Pettersen et al. [91]	https://www.rbvi.ucsf.edu/chimerax/
ClustalX 1.5b	Thompson et al. [92]; Chenna et al. [93]	N/A
BoxShade	N/A	https://junli.netlify.app/apps/boxshade/
NIS Elements AR 3.1	Nikon	N/A
MasterHands	Sugimoto et al. [102]	N/A
Other
30 μm pore size nylon mesh	Millipore	Cat#NY3004700
Positively charged nylon membrane	Roche	Cat#11417240001