B. distachyon 21 (Bd21) seeds were surface sterilized with 20% sodium hypochlorite twice for 10 min, rinsed five times with sterile H₂O, incubated in H₂O for 12 h at room temperature, and then transferred to a wet filter paper to germinate in darkness for 24 h at 22–25°C. Uniformly germinated seeds were spread on plastic pots containing 1/2 MS (Murashige and Skoog), and the medium was changed every 2 days. After 2 weeks, the seedlings were transferred to soil and grown in a growth chamber with a daily photo cycle of 16/8 h light/dark, 22°C/18°C and 65–75% air humidity.

Yeast strains used in this study are listed in Supplementary Table S1 and the cell culture conditions were as previously described (Guo et al., 2016). Yeast cells were transformed by a lithium acetate method (Ito et al., 1983). The mms2Δ::HIS3 (Xiao et al., 1999) and ubc13Δ::HUH (Wen et al., 2008) disruption cassettes were as previously described.

To clone B. distachyon UEV1 genes, total RNA was extracted from B. distachyon seedlings using a Trizol reagent (Invitrogen), the first-strand of cDNA was synthesized by a RevertAid First Strand cDNA Synthesis Kit (Thermo). The BdUEV1 ORFs were amplified by PCR from the above cDNAs using gene-specific primers (Supplementary Table S2). The yeast two-hybrid vectors pGAD424Bg and pGBT9Bg were derived from pGAD424 and pGBT9 (Bartel and Fields, 1995).

Constructed Gal4_AD and Gal4_BD vectors were transformed into the yeast two-hybrid strain PJ69-4a (James et al., 1996) in pairs and allowed to grow at 30°C for 2–3 days. Transformants were selected on SD-Leu-Trp plates. Protein interaction was determined on synthetic complete medium lacking Trp, Leu and His, supplemented with 3-amino-1,2,4-triazole (3-AT, Sigma-Aldrich), or on plates lacking Trp, Leu, and Ade. SD-Leu-Trp plates were used as a control.

Full-length BdUEV1 ORFs were cloned in plasmid pGEX-6p and BdUBC13s were cloned in plasmid pET30a as previously described (Guo et al., 2016). The His₆ and GST fusion constructs were transformed into E. coli strain BL21 (DE3) and the recombinant proteins were purified with Ni Sepharose and glutathione (Amersham Pharmacia), respectively. For the pull-down assay, crude cell extracts were loaded on Glutathione Sepharose^TM 4B beads and then 10 μg of purified His₆-BdUbc13 was later added. After incubation and washing, the GST beads were boiled with SDS-PAGE loading buffer for 10 min prior to Western blotting.

Ub conjugation reactions were performed by using purified fusion proteins and Ub thioester conjugation reagents (Boston Biochem). The reaction mixture contained 225 nM E1 enzyme, 200 μM Ub (or recombinant Ub-K63R), 1 mM MgATP, 1 mM Ubc13 and 1 mM Uev1 in 20 μl of reaction buffer. The conjugation reactions were performed at 37°C for 2 h, samples were subjected to 12% SDS-PAGE and Ub-containing molecules were detected by Western blotting using the mouse anti-Ub monoclonal antibody P4D1 (Cell Signaling).

Yeast strain HK578-10D, its isogenic mms2Δ single and ubc13Δ mms2Δ double mutants were transformed with pGAD-BdUEV1s, or co-transformed with pGBT-BdUEV1 and pGAD-BdUBC13. Transformants were selected on SD-Leu or SD-Leu-Trp plates. The gradient plate and serial dilution assays were as described (Xu et al., 2014).

Spontaneous mutagenesis was measured by the reverse mutation rate of the trp1-289 allele in the DBY747 strain (Xiao and Samson, 1993). DBY747 cells with the mms2Δ mutation were transformed with pGAD-BdUEV1s and the transformants were selected on SD-Leu plates. This modified Luria–Delbruck fluctuation test was adapted as described (Blackwell et al., 2014) and each treatment contained five independent cultures.

GFP-BdUEV1 fusion genes were constructed by cloning BdUEV1 ORFs into the modified binary vector pCAMBIA1300-GFP downstream of GFP at the SacI and KpnI sites so that the fusion gene was driven by a CaMV 35S promoter. The constructed plasmids were introduced into A. tumefaciens PMP90, which was then co-infiltrated with A. tumefaciens P19 into N. benthamiana leaves (Waadt and Kudla, 2008). GFP and DAPI fluorescence was observed 2 days after transformation using a Zeiss 780 confocal microscope.

To determine the expression level of BdUEV1 genes in different development stages and in response to abiotic stress, ddRT-PCR was performed by using a Bio-Rad QX200^TM ddPCR System. All reagents and consumables for the experiments including droplet generator oil, DG8^TM cartridges and gaskets, droplet reader oil, and ddPCR supermix for EvaGreen were purchased from Bio-Rad. Briefly, the Droplet Digital PCR began by partitioning the reaction mix containing EvaGreen Supermix, primers and sample cDNA into aqueous droplets in oil via the QX200 Droplet Generator; after transfer of droplets to a 96-well PCR plate, a thermocycling protocol including 95°C for 5 min; 95°C for 30 s, 40 cycles; 60°C for 60 s, 40 cycles (ramp rate set to 2°C/s); 4°C for 5 min; 90°C for 5 min and 4°C infinite was carried out in a conventional thermal cycler. The PCR plate was then transferred to the QX200 Droplet Reader for automatic reading of samples. QuantaSoft^TM Software was used for data analysis. The primers used to amplify the BdUEV1s are listed in Supplementary Table S3.

To identify B. distachyon UEV1 genes, Arabidopsis UEV1 genes were used to blast the B. distachyon genomic database. Three genes, BRADI1G11410.1, BRADI4G02400.1 and BRADI4G27530.1, were obtained and named BdUEV1A, BdUEV1B and BdUEV1C, respectively. Protein sequence alignment of BdUev1s reveals 81% identity between BdUev1A and BdUev1B, 77% identity between BdUev1A and BdUev1C, and 82% identity between BdUev1B and BdUev1C. The predicted BdUev1A, BdUev1B and BdUev1C proteins contain 160, 147 and 148 amino acids, respectively. It is noted that BdUev1A contains a C-terminal tail but the other two BdUev1s do not (Figure 1A). The sequences of BdUev1 proteins were also aligned with those of Uev1 proteins from six other eukaryotic organisms. As shown in Figure 1A, several critical residues deemed to be necessary for the Uev activity are also conserved in BdUev1s, including Phe13 of hMms2 (red asterisk) required for the physical interaction with Ubc13 (Pastushok et al., 2005), and Ser27 and Ile57 of ScMms2 (blue asterisks) required for the non-covalent interaction with Ub and poly-Ub chain assembly (Tsui et al., 2005; Eddins et al., 2006; Pastushok et al., 2007).

To further investigate the evolution of B. distachyon UEV1s, the CDS sequences of three BdUEV1s were compared with UEVs from six other organisms. The analysis reveals that BdUEV1s and OsUEV1s are closely related (Figure 1B). Hence, BdUEV1s could be another ideal model to study plant UEV1 gene functions, especially to study crop UEV1 gene functions.

Since Uevs play an essential role in Ubc13-mediated Lys63-linked polyubiquitination through forming a stable complex with Ubc13 to maintain its unique E2 activity (Hofmann and Pickart, 1999; McKenna et al., 2001, 2003), a yeast two-hybrid assay (Fields and Song, 1989) was used to determine whether BdUev1s interact with BdUbc13s. As seen in Figure 2A, all three BdUev1s interact with both BdUbc13s and the co-transformants are able to grow in the presence of 5 mM 3-AT, while no self-activation is observed in the -His plate without 3-AT. To further confirm the physical interaction between BdUev1s and BdUbc13s in vitro, a GST-affinity pull-down assay was performed. As shown in Figure 2B, recombinant GST-BdUev1 fusion proteins can pull down purified recombinant His₆-BdUbc13 proteins, but the GST alone cannot. In the GST-affinity pull-down assay, the strength of interaction with BdUbc13 has no detectable difference. Based on the above observations, we conclude that BdUev1s can form stable heterodimers with BdUbc13s in vitro.

So far, the only defined Uev activity is to promote Lys63-linked polyubiquitination along with Ubc13 (Hofmann and Pickart, 1999; Deng et al., 2000; Hoege et al., 2002). A standard means of characterizing Uev enzymatic activity is by an in vitro Ub conjugation assay (Hofmann and Pickart, 1999). To avoid Ub conjugation to Ubc13 itself (McKenna et al., 2001), we made the corresponding BdUbc13-K94R mutation that does not affect the free Ub chain formation (Guo et al., 2016). As shown in Figure 3, BdUev1s together with BdUbc13A can indeed generate free Ub chains (lanes 3, 6, and 9), while BdUbc13A (lane 1) or BdUev1s (lanes 2, 5, and 8) alone cannot. More importantly, when the Ub-K63 is mutated to R, there are no poly-Ub chains observed in the same reaction (lanes 4, 7, and 10, asterisks indicate the background of UbK63R, see Supplementary Figure S1, lane 3). Hence, we conclude that BdUev1s are indeed required for Lys63-linked poly-Ub chain assembly. Since it has been previously reported that both BdUbc13A and BdUbc13B have the same ability to promote Lys63-linked poly-Ub chains along with AtUevs (Guo et al., 2016), this study did not repeat the reaction with BdUbc13B.

Yeast Mms2 is a member of the error-free DDT pathway (Broomfield et al., 1998; Brusky et al., 2000). To determine whether BdUev1s also play critical roles in error-free DDT, we assessed the ability of BdUEV1s to rescue the yeast mms2 mutant phenotypes. As shown in Figure 4A, the expression of any one of the three BdUEV1 genes can protect the mms2 null mutant from killing by MMS, whereas the expression of pGAD424 vector cannot. It is interesting that the ability of BdUEV1A to rescue the mms2 null mutant is slightly less than other two BdUEV1s.

DNA damage induced by MMS, UV irradiation or 4-nitroquinoline 1-oxide (4NQO) is largely considered to cause replication blocks and the mms2 null mutant is hypersensitive to these agents (Broomfield et al., 1998; Xu et al., 2013). To further confirm that BdUEV1s function in the DDT pathway, we performed a serial dilution assay in the presence of several representative DNA-damaging agents as mentioned above. A serial dilution assay is a semi-quantitative method to assess the relative sensitivity of cells to treatment (Xu et al., 2014). As shown in Figure 4B, both BdUEV1B and BdUEV1C can fully protect the yeast mms2 null mutant from killing by MMS, UV or 4-NQO, while the protection provided by BdUEV1A appears to be compromised.

Loss of MMS2 dramatically increases spontaneous mutagenesis, which serves as a hallmark indicating that it is involved in the error-free DDT pathway (Xu et al., 2015). Under our experimental conditions, deletion of MMS2 caused a 10-fold increase in the spontaneous mutation rate. When the mutant cells were transformed with BdUEV1s, the spontaneous mutagenesis rate was reduced to less than twofold (Figure 4C). Again BdUEV1B and BdUEV1C appear to be more effective than BdUEV1A in complementing the yeast MMS2 function. Hence, our results indicate that BdUev1B and BdUev1C can fully replace the function of Mms2 in the yeast error-free DDT pathway, but BdUev1A is partially functional.

The above functional complementation studies require that BdUev1s form stable heterodimers with yeast Ubc13. To further confirm that BdUev1 and BdUbc13 can form a stable heterodimer to replace the function of both yeast Mms2 and Ubc13, we co-transformed yeast ubc13 mms2 double mutant cells with the combination of BdUEV1s and BdUBC13s and assessed their relative tolerance to DNA-damaging agents. As shown in Figure 5 and Supplementary Figure S2, expression of BdUEV1 or BdUBC13 alone cannot rescue the mms2 ubc13 double mutant, whereas co-expression of BdUEV1B or BdUEV1C with BdUBC13 can rescue the ubc13 mms2 double mutant to nearly the wild-type level, regardless of the source of DNA damage. In sharp contrast, BdUBC13A or BdUBC13B co-transformed with BdUEV1A did not effectively rescue the ubc13 mms2 double mutant from killing induced by MMS, UV or 4NQO.

Among the three BdUev1s, an obvious sequence difference is the additional 18-amino acid tail found at the C terminus of BdUev1A (Figure 1A). We hypothesized that the lack of complementation of the yeast mms2 mutant by BdUEV1A is due to its encoded C-terminal tail. To test this hypothesis, we created a BdUEV1A-ΔCT construct in which the C-terminal 18 amino-acid coding sequence is removed from BdUEV1A. Figure 6 shows that BdUEV1A-ΔCT provides stronger protection against DNA damage than BdUEV1A and is comparable with that of BdUEV1B. Hence, the C-terminal tail of BdUev1A determines its efficacy in complementing the yeast DDT defect.

To further address the difference between BdUEV1A and BdUEV1B/C in complementing the yeast mms2 mutation, we entertained a hypothesis that their encoded proteins may have different subcellular distributions in plants. To test this hypothesis, three GFP-BdUEV1 fusion constructs were made, transiently transfected to tobacco (N. benthamiana) leaves by A. tumefaciens and their subcellular localization was monitored by fluorescence microscopy. As shown in Figure 7, both GFP-BdUev1B and GFP-BdUev1C were primarily found in the nucleus. In sharp contrast, GFP-BdUev1A was excluded from the nucleus, while truncation of its C-terminal 18 amino acids (BdUev1A-ΔCT) results in its redistribution to the nucleus, reminiscent of the subcellular localization patterns of GFP-BdUev1B and GFP-BdUev1C. Since the DDT function must occur in the nucleus, the above observations further suggest that BdUev1B and BdUev1C are involved in plant DNA-damage response while BdUev1A is not. Meanwhile, this observation indicates that the C-terminal tail of BdUev1A plays a critical role in its subcellular localization and physiological functions.

In a recent study, we found no obvious fluctuation in the expression of BdUBC13 genes in different tissues or during plant development (Guo et al., 2016). This expression pattern is consistent with UBC13 genes from Arabidopsis (Wen et al., 2006) and rice (Zang et al., 2012). In contrast, the expression of Arabidopsis UEV1 genes appears to fluctuate in different tissues and during development (Wen et al., 2008). To investigate the BdUEV1 gene expression during development, total RNA was extracted from 7-day seedlings as well as different tissues and the BdUEV1 transcript levels were determined by ddPCR. As shown in Figure 8A, the three BdUEV1 genes have a relatively uniform expression in different tissues except for BdUEV1C in L2 (leaves from transition phase) and Sp8 (spikes from 8-week old) phases. The BdUEV1B expression also has a significant but moderate increase during L2.

We also analyzed BdUEV1 expression in response to abiotic stress using ddRT-PCR (Figure 8B). It is of great interest to note that cold stress induces the expression of all three BdUEV1s reminiscent of cold-induced BdUBC13 expression (Guo et al., 2016). The transcription level of all BdUEV1s also significantly increased after heat and PEG treatments. On the other hand, only BdUEV1B expression has a twofold increase after MMS treatment, suggesting the involvement of BdUEV1B in DNA-damage response.