AP endonuclease EXO-3 deficiency causes developmental delay and abnormal vulval organogenesis, Pvl, through DNA glycosylase-initiated checkpoint activation in Caenorhabditis elegans

AP endonuclease deficiency causes cell death and embryonic lethality in mammals. However, the physiological roles of AP endonucleases in multicellular organisms remain unclear, especially after embryogenesis. Here, we report novel physiological roles of the AP endonuclease EXO-3 from larval to adult stages in Caenorhabditis elegans, and elucidated the mechanism of the observed phenotypes due to EXO-3 deficiency. The exo-3 mutants exhibited developmental delay, whereas the apn-1 mutants did not. The delay depended on the DNA glycosylase NTH-1 and checkpoint kinase CHK-2. The exo-3 mutants had further developmental delay when treated with AP site-generating agents such as methyl methane sulfonate and sodium bisulfite. The further delay due to sodium bisulfite was dependent on the DNA glycosylase UNG-1. The exo-3 mutants also demonstrated an increase in dut-1 (RNAi)-induced abnormal vulval organogenesis protruding vulva (Pvl), whereas the apn-1 mutants did not. The increase in Pvl was dependent on UNG-1 and CHK-2. Methyl viologen, ndx-1 (RNAi) and ndx-2 (RNAi) enhanced the incidence of Pvl among exo-3 mutants only when combined with dut-1 (RNAi). This further increase in Pvl incidence was independent of NTH-1. These results indicate that EXO-3 prevents developmental delay and Pvl in C. elegans, which are induced via DNA glycosylase-initiated checkpoint activation.

The exo-3 mutants exhibit developmental delay. To clarify the contribution of AP endonucleases to worm development from the L4 to adult stages, worms deficient in either or both AP endonucleases (EXO-3 and APN-1) were incubated under normal rearing conditions for 3 days from the fertilized egg stage (Fig. 1c). Developmental stages among the L4, young adult and gravid adult stages were distinguished by the state of vulval morphology and brooding of eggs (Fig. 1d,f). Although all of the N2 worms developed into gravid adults, only 14% of the exo-3 mutants were in the gravid adult stage, 85% were in the young adult stage and 1% was in the larval stage (Fig. 1g), suggesting that EXO-3 deficiency causes the cessation of development or developmental delay. In contrast, all of the apn-1 mutants became gravid adults, and the apn-1;exo-3 mutants were at almost the same developmental stages as the exo-3 mutants (Fig. 1g). Twelve hours later, all the exo-3 and apn-1;exo-3 mutants reached the gravid adult stage (data not shown), indicating that EXO-3 deficiency does not cause cessation of development at the young adult stage, only developmental delay. To precisely investigate how long the delay of the exo-3 mutants was, we measured the reaching time to gravid adult of each worm every two hours and calculated the difference of the average time between N2 (N = 8) and the exo-3 mutants (N = 16). The difference was 6 hours.
The developmental delay in the exo-3 mutants is dependent on the DNA glycosylase NTH-1. It is reasonable to infer that the developmental delay phenotype of exo-3 mutants is caused by the accumulation of AP sites or 3′-blocked SSB in DNA because these are substrates for EXO-3 15 . These structures in DNA can be generated by DNA glycosylases. Of the two DNA glycosylases conserved in C. elegans, UNG-1 generates AP sites through monofunctional DNA glycosylase activity on uracil in DNA, and NTH-1 produces 3′-blocked SSB via bifunctional DNA glycosylase activity on oxidative pyrimidine lesions in DNA. Therefore, we examined the dependency of the delay in the exo-3 mutants on UNG-1 and NTH-1. Three days after developing from eggs, 9% of the ung-1;exo-3 mutants were in the gravid adult stage, 86% were in the young adult stage and 5% were in the larval stage, and these proportions were almost the same as those shown by the exo-3 mutants (11% in the gravid adult stage, 85% in the young adult stage and 4% in the larval stage) (Fig. 2) independent of UNG-1. In contrast, 94% of the nth-1;exo-3 mutants were in the gravid adult stage and 6% were in the young adult stage. The nth-1;ung-1;exo-3 mutants exhibited similar results (Fig. 2), suggesting that the delay is dependent on NTH-1.

Figure 1.
Effects of AP endonuclease deficiency on larval development of worms under normal rearing conditions. (a,b) At each time point after birth, N2 worms were harvested, and the total RNA isolated from the worms was subjected to real-time PCR analysis using specific primer sets for exo-3 (a) and apn-1 (b The developmental delay of the exo-3 mutants is enhanced by MMS and NaHSO 3 . To clarify whether AP site-generating agents can cause developmental delay, we conducted a developmental assay using MMS and sodium bisulfite (NaHSO 3 ) (Fig. 3a). MMS is known to indirectly create AP sites 20,21 and shown to induce DNA lesions in the genome of C. elegans 22 . At 3.5 days after the eggs were laid on plates containing 0.94 mM MMS, 97% of N2 were in the gravid adult stage and 3% were in the larval stage, whereas 61% of the exo-3 mutants were in the gravid adult stage, 34% were in the young adult stage and 5% were in the larval stage (Fig. 3b), suggesting that MMS-induced AP sites cause further developmental delay in the exo-3 mutants than in N2. On the other hand, 93% of the apn-1 mutants were in the gravid adult stage, 6% were in the young adult stage and 1% were in the larval stage, which is similar to the results for N2 (Fig. 3b). NaHSO 3 damages DNA mainly through deamination of cytosine to uracil 23 . Four days after developing from the egg stage, all exo-3 mutants not treated with NaHSO 3 developed into gravid adults, but 33% of those treated with 10 mM NaHSO 3 were in the gravid adult stage, 12% were in the young adult stage and 55% were in the larval stage (Fig. 3c), indicating that NaHSO 3 induced developmental delay in the exo-3 mutants. This delay was alleviated in the ung-1;exo-3 mutants, as 79% of those treated with 10 mM NaHSO 3 were in the gravid adult stage, 14% were in the young adult stage and 7% were in the larval stage (Fig. 3c), suggesting that the NaHSO 3 -induced developmental delay was due to UNG-1 activity. Taken together, AP sites may cause developmental delay as well as 3′-blocked SSB generated by NTH-1.
The developmental delay of the exo-3 mutants is induced by CHK-2. We next hypothesized that the DNA glycosylase-initiated developmental delay of the exo-3 mutants was due to DNA damage checkpoint activation driven by cleavage products produced by DNA glycosylases. DNA damage checkpoint genes, such as chk-2 and clk-2, have been described in C. elegans 24 . To test our hypothesis, the developmental assay was conducted under chk-2 (RNAi) or clk-2 (RNAi) conditions (Fig. 4a). Although 14% of the exo-3;control (RNAi) worms were in the gravid adult stage and 84% were in the young adult stage, and exo-3;clk-2 (RNAi) worms demonstrated similar proportions (18% in the gravid adult stage and 82% in the young adult stage), 93% of the exo-3;chk-2 (RNAi) worms were in the gravid adult stage (Fig. 4b). Thus, the knockdown of chk-2 rescued the developmental delay in the exo-3 mutants, suggesting that CHK-2 induces the developmental delay in the exo-3 mutants.

The increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants is dependent on UNG-1 irrespective of NTH-1.
To examine whether the increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants occurred through UNG-1 activity, we investigated the dependency of the phenotype on UNG-1. The percentage of ung-1;exo-3 mutants with Pvl was 2%, which was the same as that of the ung-1 mutants with Pvl (1%) (Fig. 6a), suggesting that the increase in Pvl in the exo-3 mutants is only due to UNG-1 expression. Next, we examined whether the cleavage products produced by UNG-1 are needed to induce Pvl, i.e., whether AP sites are transformed into 3′-blocked SSB via the AP lyase activity of NTH-1. The percent of exo-3 mutants with Pvl was comparable to that of nth-1;exo-3 mutants (Fig. 7b), suggesting that NTH-1 is not necessary for dut-1 (RNAi)-induced Pvl.

The increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants is driven by CHK-2.
We hypothesized that the UNG-1-dependent increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants can occur via DNA checkpoint activation driven by cleavage products produced by UNG-1 in addition to developmental delay. It has also been reported that dut-1 (RNAi)-induced Pvl is caused by the checkpoint kinase CLK-2 27 . Therefore, we examined whether the increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants was dependent on CHK-2 and CLK-2. Although 49% of exo-3;control (RNAi) worms had dut-1 (RNAi)-induced Pvl, only 10% of the exo-3;chk-2 (RNAi) worms had it (Fig. 6b). On the other hand, the proportion of exo-3;clk-2 (RNAi) worms with Pvl was almost the same (44%) as that of the exo-3;control (RNAi) worms. Therefore, the knockdown of chk-2 rescued the increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants, as observed for developmental delay. These results suggest that the increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants occurs via CHK-2 expression.
Oxidative DNA damaging agents enhanced the proportion of Pvl in the exo-3 mutants only when combined with dut-1 (RNAi). To investigate other DNA damaging agents that cause an increase in Pvl in the exo-3 mutants, we evaluated whether Pvl formation is enhanced by oxidative DNA damaging agents such as ndx-1 (RNAi), ndx-2 (RNAi) and methyl viologen (MV). NDX-1 and NDX-2 hydrolyze 8-oxo-dGDP into 8-oxo-dGMP 24,28 . Accordingly, ndx-1 (RNAi) and ndx-2 (RNAi) lead to an increase in 8-oxo-dGDP in the nucleotide pool. The presence of 8-oxo-dGDP reduces the 8-oxo-dGTPase activity of NDX-4, causing 8-oxo-dGTP to accumulate in the pool 24 . 8-oxo-dGTP is incorporated to DNA during DNA replication, resulting in the accumulation of 8-oxoG in DNA 24,28 . MV generates superoxide radicals, which subsequently cause oxidative lesions in DNA 29 . However, single treatment with ndx-1 (RNAi), ndx-2 (RNAi) or MV had no effect on the incidence of Pvl in the exo-3 mutants. (data not shown). Next, we examined whether, when combined with the dut-1 (RNAi) treatment, each oxidative DNA damaging agent further enhanced the increase in Pvl in the exo-3 mutants (Fig. 7a). Each treatment tested enhanced the increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants by approximately 10% (Fig. 7b), suggesting that oxidative lesions in DNA can also cause an increase in Pvl.
In in vitro experiments, NTH-1 was found to possess weak DNA glycosylase activity toward 8-oxoG in DNA, in addition to its much higher activity toward oxidative pyrimidine lesions 30 . Thus, we suspect that the higher increase in dut-1 (RNAi)-induced Pvl by the additional oxidative agents depends on the activity of NTH-1. Although we confirmed the dependency of the phenotype on NTH-1, NTH-1 deficiency did not alter the proportion of worms exhibiting Pvl (Fig. 7b).

Discussion
In this study, we investigated the in vivo contribution of AP endonucleases from the larval to adult stages in C. elegans by evaluating development and vulval organogenesis in AP endonuclease gene mutants, and clarified that AP endonuclease EXO-3 deficiency causes developmental delay and an increased incidence of dut-1 (RNAi)-induced Pvl via DNA glycosylase-initiated checkpoint activation (Fig. 8).
The exo-3 mutants demonstrated developmental delay, whereas the apn-1 mutants did not (Fig. 1e), suggesting that EXO-3 has a more important role than APN-1 during development from the larval to adult stages.
The delay in the exo-3 mutants was completely dependent on NTH-1 (Fig. 2), suggesting that it is caused by 3′-blocked SSB generated by NTH-1. We therefore hypothesized that AP sites cannot cause developmental delay in the exo-3 mutants. However, MMS and NaHSO 3 induced further developmental delay in the exo-3 mutants (Fig. 3b,c), and we found that the delay by NaHSO 3 was dependent on UNG-1 (Fig. 3c), suggesting that AP sites also caused developmental delay in the exo-3 mutants, although the further delay may be caused by a mixture of both AP sites and cleaved AP sites with 3′-blocked SSB by NTH-1.
The developmental delay in the exo-3 mutants was also due to CHK-2 (Fig. 4b). CHK-2 is an ortholog of mammalian Chk2, which is activated in response to several DNA damaging agents that cause DSB 31 , but there is no direct evidence that C. elegans CHK-2 is involved in a checkpoint mechanism driven by SSB. As 3′-blocked SSB can generate DSB during DNA replication 6 , as can AP sites 32 , it is possible that the resulting DSB activated the CHK-2 response, thereby leading to developmental delay. It was recently reported that ATM, which positively regulates Chk2 activity in response to DSB-generating agents 33,34 , is also activated by SSBs in human cells 35 . Thus, it is possible that 3′-blocked SSB directly activate CHK-2 via an ATM-1/CHK-2 pathway. The exo-3 mutants exhibited an increased incidence of dut-1 (RNAi)-induced Pvl (Fig. 5d), but the apn-1 mutants did not, suggesting that EXO-3 has a more important role than APN-1 in vulval organogenesis and development. The increase in Pvl in the exo-3 mutants was completely dependent on UNG-1 (Fig. 6a). This suggests that Pvl is caused by UNG-1-generating AP sites. The increase in Pvl occurred irrespective of NTH-1 (Fig. 7b), suggesting that the transformation of AP sites into 3′-blocked SSB by NTH-1 is not needed to induce the increase in Pvl.
Dengg et al. previously reported that dut-1 (RNAi)-induced Pvl occurs via the checkpoint kinase CLK-2 27 . CLK-2 may be activated by replication fork collapse-mediated DSB because they found that dut-1(RNAi) enhances the accumulation of RPA-1, ATL-1 (ATR ortholog in C. elegans) and RAD-51 in mitotic germ cells based on UNG-1 activity 27 . In this study, we tried to clarify whether the increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants was dependent on CLK-2, but knockdown of CLK-2 had no effect on the increased incidence of Pvl (Fig. 6b). This discrepancy may result from the methods used to compromise CLK-2 function. Dengg et al. used clk-2 mutant worms, whereas we used clk-2 (RNAi) worms, as reported previously 24 . Instead of CLK-2, another checkpoint kinase, CHK-2, was found to be a causal factor of the increase in Pvl (Fig. 6b). It is possible that a mechanism similar to CLK-2 activation causes Pvl formation through CHK-2 in the exo-3 mutants.
The increase in dut-1 (RNAi)-induced Pvl in the exo-3 mutants was further enhanced by oxidative DNA damaging agents such as ndx-1 (RNAi), ndx-2 (RNAi) and MV (Fig. 7b). The damaging agents may have only caused Pvl in the exo-3 mutants when combined with dut-1 (RNAi) because of a threshold of DNA lesions needed for Pvl. Combining each oxidative damaging agent with dut-1 (RNAi) results in more DNA lesions containing 8-oxoG than single dut-1 (RNAi) treatment. However, a direct homolog of MutM or OGG1 cannot be detected in C. elegans. As a candidate protein to incise 8-oxoG in DNA in vivo in C. elegans, we examined NTH-1. However, the effects of ndx-1 and ndx-2 knockdown on Pvl were independent of NTH-1 (Fig. 7b). A novel DNA glycosylase that can incise 8-oxoG may be responsible for this phenotype, but further studies are needed.
Pvl is formed by prevention of ras/notch/wnt signaling pathway [36][37][38] . We demonstrated that the checkpoint activation caused by UNG-1 results in the induction of Pvl, while the downstream pathway to induce Pvl still remains unclear. Thus, it is possible that checkpoint activation affects Pvl induction through modulation of other pathways such as ras/notch/wnt signaling pathways. However, there is no evidence of the link between checkpoint activation and such signaling pathways. It is reasonable that checkpoint activation causes developmental delay, while it seems paradoxical that the activation also causes Pvl induction. The reason for the point is that developmental delay induced by checkpoint activation is predicted to be a mechanism for preventing mutagenesis, which provides worms beneficial effects. In contrast, Pvl caused by the activation seems to have no valuable effects. However, it is probable that the induction of Pvl, resulting in egg-laying-defective (Egl) worms 27 , is a mechanism for preventing the accumulation of mutations in the next generation.
This study demonstrated that EXO-3 prevents DNA glycosylase-initiated checkpoint activation in order for worms to grow at a normal speed and with normal vulva. Although there have been many studies reported a correlation between BER and biological phenomena, such as carcinogenesis and aging, few studies demonstrating  causation have been conducted 39,40 . Due to the availability of C. elegans mutants and the characteristics of C. elegans mutants lacking AP endonucleases enabling them to mature past embryonic stages and produce the next generation, we were able to examine causal relationships between BER and many biological phenomena throughout life. Further studies on AP endonucleases using C. elegans will provide more detailed information about the in vivo roles of AP endonucleases in multicellular organisms.

Measurement of the proportion of Pvl worms.
To assay the effects of AP endonuclease deficiency on organogenesis, synchronized eggs were placed on dut-1 (RNAi) plates (plus additional RNAi and 0.1 mM methyl viologen (MV), if necessary). After incubation at 20 °C for 4 days, the numbers of total adult worms and Pvl worms were counted, and the proportion of Pvl worms among adult worms was calculated. Significance was determined using one-way ANOVA with Tukey's test for multiple comparisons.
Microscopy. Observation and imaging of C. elegans were performed using an OLYMPUS SZX16 microscope (OLYMPUS, Japan).

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.