Dynamic Changes in Reactive Oxygen Species in the Shoot Apex Contribute to Stem Cell Death in Arabidopsis thaliana

In monocarpic plants, stem cells are fated to die. However, the potential mechanism of stem cell death has remained elusive. Here, we reveal that the levels of two forms of reactive oxygen species (ROS), superoxide anion free radical (O2·−) and hydrogen peroxide (H2O2), show dynamic changes in the shoot apex during the plant life cycle of Arabidopsis thaliana. We found that the level of O2·− decreased and disappeared at four weeks after bolting (WAB), while H2O2 appeared at 3 WAB and showed a burst at 5 WAB. The timing of dynamic changes in O2·− and H2O2 was delayed for approximately three weeks in clv3-2, which has a longer lifespan. Moreover, exogenous application of H2O2 inhibited the expression of the stem cell determinant WUSCHEL (WUS) and promoted the expression of the developmentally programmed cell death (dPCD) marker gene ORESARA 1 (ORE1). These results indicate that H2O2 triggers an important signal inducing dPCD in stem cells. Given that O2·− plays roles in maintaining WUS expression and stem cell activity, we speculate that the dynamic shift from O2·− to H2O2 in the shoot apex results in stem cell death. Our findings provide novel insights for understanding ROS-mediated regulation during plant stem cell death.


Introduction
Plant stem cells are harbored in shoot apical meristems (SAMs), which play essential roles in growth and development. Over the past decade, multiple works have shown the molecular mechanisms of stem cell maintenance [1,2]. In Arabidopsis thaliana, WUSCHEL (WUS) and CLAVATA3 (CLV3) constitute a negative feedback loop to maintain the homeostasis of the stem cell population [3,4]. WUS, as a stem cell marker, not only positively controls meristem activity but also maintains the stem cell population during the postembryonic development stage [5,6]. Several factors, such as the peptide CLV3 and receptor kinases CLV1, CLV2, BARELY ANY MERISTEM (BAM), and receptor-like protein kinase 2 (RPK2)/TOADSTOOL 2, are involved in regulating the spatiotemporal expression pattern of WUS in the SAM [7]. As an intracellular signal, the plant hormone cytokinin can also regulate WUS expression via both CLV-dependent and CLV-independent pathways [7][8][9]. Recently, the intracellular molecule reactive oxygen species (ROS) have been reported to be involved in regulating stem cell maintenance during the vegetative stage, indicating that ROS balance plays a key role in stem cell fate regulation [10]. ROS, mainly composed of superoxide anion free radical (O ·− 2 ) and hydrogen peroxide (H 2 O 2 ), are naturally produced in all living cells [10]. ROS are associated with stress stimuli such as high light, salinity, heavy metals, cold, and pathogens [11].
As one of the ROS species, H 2 O 2 has been shown to be a crucial signaling molecule in various growth and developmental processes, including seed germination, leaf development and senescence, development of trichome and male sex organs, development of pollen tubes on the pistil, lateral root and root hair development, aerenchyma formation, and even self-incompatibility [12]. It is worth noting that H 2 O 2 gives rise to programmed cell death (PCD) in most of the biological processes mentioned above [1,12]. In addition, intracellular redox status is connected to meristem activity in both the vegetative SAM and the root apical meristem (RAM) [10,[13][14][15][16]. However, whether ROS influence stem cell status in the senescent SAM during the plant life cycle is still unclear.
We previously observed the senescence phenomenon in shoot stem cells and specified that the final fate of shoot stem cells undergoes a developmental PCD (dPCD) process [2]. In addition, we pointed out that shoot stem cell dPCD is associated with the expression of ROS-related genes [2]. Given that stem cells directly contribute to plant lifespan and that the death of stem cells is an important part of stem cell fate, it is necessary to investigate the ROS-mediated PCD process during the stem cell lifespan more deeply.
In this study, we first monitored the dynamic spatiotemporal patterns of ROS species O ·− 2 and H 2 O 2 in proliferative and senescent SAMs. By applying exogenous H 2 O 2 , we found that the accumulation of H 2 O 2 could not only inhibit WUS expression but also induce dPCD in stem cells. Moreover, we used two scavengers, N,N -dimethylthiourea (DMTU) [10,17] and potassium iodide (KI) [10,13], to reduce the levels of O ·− 2 and H 2 O 2 , respectively, in the inflorescence meristem (IM) domain. The results further explored the function of H 2 O 2 in shutting down stem cell activity, leading to the activation of the dPCD process in the stem cell population. Based on this evidence, we established a regulatory model in which ROS dynamics were thought to be at the core of explaining the ROS-mediated mechanism of stem cell dPCD. The results of this study can enrich our understanding of age-dependent dPCD in stem cells and may help to facilitate the regulatory mechanism of stem cell fate governing the whole plant lifespan.

Dynamic Changes in ROS Components in Wild-Type and clv3-2 Mutant Plants
Our previous study revealed that the inflorescence meristem (IM) of clv3-2 mutants possesses a longer longevity. Moreover, clv3-2 produces an increased number of flowers and siliques on primary shoots ( Figure S1) [2]. Thus, we think that CLV3 may be a safeguard that prevents the longer expression time of WUS [2]. However, we found that the expression of WUS was still terminated at 6 weeks after bolting (WAB) in clv3-2 mutants ( Figure S2) [2], indicating that there must be other factors that inhibit WUS transcription. Recently, a study reported that two ROS components, O ·− 2 and H 2 O 2 , are involved in regulating stem cell development and that the balance between them is indispensable to stem cell maintenance and differentiation [10]. Moreover, O ·− 2 can activate WUS expression and maintain stemness, and H 2 O 2 accumulation in the peripheral zone (PZ) negatively regulates O ·− 2 biosynthesis, resulting in stem cell termination [10]. This evidence strongly suggests that ROS homeostasis influences stem cell fate determination and that H 2 O 2 may be a regulator of WUS expression. More significantly, our preceding work also shows that the H 2 O 2 signal may be involved in WUS expression termination [2]. Therefore, it is necessary to understand the spatial-temporal patterns of ROS components in the IM.
Through NBT (4-nitro blue tetrazolium chloride) staining, we examined the distribution and accumulation characteristics of O ·− 2 in the IM domains of the WT and the clv3-2 mutant. In WT, O ·− 2 displayed a strong signal in the IM domain, especially in the stem cell layers, at 1 WAB and 2 WAB. At 3 WAB, the level of O ·− 2 in the stem cells was clearly decreased in the IM (pink dashed lines) but became undetectable from 4 WAB to 6 WAB ( Figure 1). In the clv3-2 mutant, O ·− 2 was detected for 2 weeks longer than in the WT (from 1 WAB to 5 WAB). Clear O ·− 2 signals were obviously displayed in the clv3-2 IM domain from 1 WAB to 4 WAB. At 5 WAB, the level of O ·− 2 in the stem cell layers was reduced and fully disappeared from the stem cells from 6 WAB ( Figure 1). The spatial-temporal distribution of H2O2 in the WT and clv3-2 IM domains was observed using DAB (3,3′-diaminobenzidine) staining. In the WT IM domain, H2O2 initially appeared at 3 WAB and showed a burst at 5 WAB ( Figure 2). There were no detectable signals in clv3-2 IM domains from 1 WAB to 5 WAB. Then, H2O2 signals initially appeared in the clv3-2 IM domain at 6 WAB. Subsequently, clear and increased H2O2 signals were detected in the clv3-2 IM domain until 10 WAB. Notably, the level of H2O2 showed an accumulated peak at 10 WAB. At 11 and 12 WABs, there were no detectable H2O2 signals due to whole IM death (Figures 2 and S1). Compared with the WT, H2O2 signals appeared 3 weeks later in the clv3-2 IM domain ( Figure 2). Based on these findings, we found that the time during which O 2 ·and H2O2 were detected displayed a switch period. In WT, this switch period occurred from 3 WAB to 4 WAB, and the level of O 2 ·decreased from 3 WAB and fully disappeared from the IM domain at 4 WAB. In addition, a very weak H2O2 signal was initially detected in the stem cell layers at 3 WAB and became stronger at 4 WAB. Similarly, there was also a conversion The spatial-temporal distribution of H 2 O 2 in the WT and clv3-2 IM domains was observed using DAB (3,3 -diaminobenzidine) staining. In the WT IM domain, H 2 O 2 initially appeared at 3 WAB and showed a burst at 5 WAB ( Figure 2). There were no detectable signals in clv3-2 IM domains from 1 WAB to 5 WAB. Then, H 2 O 2 signals initially appeared in the clv3-2 IM domain at 6 WAB. Subsequently, clear and increased H 2 O 2 signals were detected in the clv3-2 IM domain until 10 WAB. Notably, the level of H 2 O 2 showed an accumulated peak at 10 WAB. At 11 and 12 WABs, there were no detectable H 2 O 2 signals due to whole IM death (Figures 2 and S1). Compared with the WT, H 2 O 2 signals appeared 3 weeks later in the clv3-2 IM domain ( Figure 2). The spatial-temporal distribution of H2O2 in the WT and clv3-2 IM domains was observed using DAB (3,3′-diaminobenzidine) staining. In the WT IM domain, H2O2 initially appeared at 3 WAB and showed a burst at 5 WAB ( Figure 2). There were no detectable signals in clv3-2 IM domains from 1 WAB to 5 WAB. Then, H2O2 signals initially appeared in the clv3-2 IM domain at 6 WAB. Subsequently, clear and increased H2O2 signals were detected in the clv3-2 IM domain until 10 WAB. Notably, the level of H2O2 showed an accumulated peak at 10 WAB. At 11 and 12 WABs, there were no detectable H2O2 signals due to whole IM death (Figures 2 and S1). Compared with the WT, H2O2 signals appeared 3 weeks later in the clv3-2 IM domain ( Figure 2). Based on these findings, we found that the time during which O 2 ·and H2O2 were detected displayed a switch period. In WT, this switch period occurred from 3 WAB to 4 WAB, and the level of O 2 ·decreased from 3 WAB and fully disappeared from the IM domain at 4 WAB. In addition, a very weak H2O2 signal was initially detected in the stem cell layers at 3 WAB and became stronger at 4 WAB. Similarly, there was also a conversion Based on these findings, we found that the time during which O ·− 2 and H 2 O 2 were detected displayed a switch period. In WT, this switch period occurred from 3 WAB to 4 WAB, and the level of O ·− 2 decreased from 3 WAB and fully disappeared from the IM domain at 4 WAB. In addition, a very weak H 2 O 2 signal was initially detected in the stem cell layers at 3 WAB and became stronger at 4 WAB. Similarly, there was also a conversion period of the time during which O ·− 2 and H 2 O 2 were detected in clv3-2 mutants from 6 WAB to 7 WAB, which represented a two-week delay compared with that in WT (Figures 1 and 2). These results indicated that dynamic changes in ROS components occurred during the lifespan of IM in both WT and clv3-2 mutants, and the deferred conversion period in clv3-2 IM was in line with the prolonged WUS expression time, suggesting that ROS conversion might regulate the expression time of WUS.

Features of Programmed Stem Cell Death in the clv3-2 Mutant
Whereas the stem cell death process has been studied by fluorescein diacetate (FDA) and propidium iodide (PI) staining and the key dPCD marker gene BIFUNCTIONAL NUCLEASE1 (BFN1) in WT Arabidopsis [2], similar studies in clv3-2 stem cells are still lacking. Thus, we examined the details of stem cell death in both WT and clv3-2 mutants in this study. The FDA/PI staining results showed that stem cell death started at 7 WAB in clv3-2 IM, and strong PI signals could be detected up to 10 WAB. Subsequently, all stem cells were stained by PI at 11 WAB and 12 WAB, as the entire clv3-2 IM was completely dead (Figures 3 and S1). In contrast, the PI signal was visible at 5 WAB in WT IM. At 6 WAB, all stem cells were stained by PI because the WT IM was dead (Figures 3 and S1). period of the time during which O 2 ·and H2O2 were detected in clv3-2 mutants from 6 WAB to 7 WAB, which represented a two-week delay compared with that in WT ( Figures  1 and 2). These results indicated that dynamic changes in ROS components occurred during the lifespan of IM in both WT and clv3-2 mutants, and the deferred conversion period in clv3-2 IM was in line with the prolonged WUS expression time, suggesting that ROS conversion might regulate the expression time of WUS.

Features of Programmed Stem Cell Death in the clv3-2 Mutant
Whereas the stem cell death process has been studied by fluorescein diacetate (FDA) and propidium iodide (PI) staining and the key dPCD marker gene BIFUNCTIONAL NU-CLEASE1 (BFN1) in WT Arabidopsis [2], similar studies in clv3-2 stem cells are still lacking. Thus, we examined the details of stem cell death in both WT and clv3-2 mutants in this study. The FDA/PI staining results showed that stem cell death started at 7 WAB in clv3-2 IM, and strong PI signals could be detected up to 10 WAB. Subsequently, all stem cells were stained by PI at 11 WAB and 12 WAB, as the entire clv3-2 IM was completely dead (Figures 3 and S1). In contrast, the PI signal was visible at 5 WAB in WT IM. At 6 WAB, all stem cells were stained by PI because the WT IM was dead (Figures 3 and S1). To understand which type of cell death occurred in the dead stem cells, we checked the spatial-temporal expression profiles of dPCD marker genes. The key dPCD marker gene BFN1, which encodes a nuclease, was initially expressed in the bottom regions of the clv3-2 IM domain at 3 WAB. Then, the expression area of BFN1 moved upward but still did not enter into the stem cell layers until 6 WAB, which is the time point at which WUS started to be terminated (Figures 4 and S2). From 7 WAB to 10 WAB, BFN1 was expressed in stem cells, resulting in stem cell death in clv3-2 IM (Figures 3 and 4). Because the whole clv3-2 IM was dead after 11 WAB, there were no BFN1 signals in the IM region ( Figure 4). To understand which type of cell death occurred in the dead stem cells, we checked the spatial-temporal expression profiles of dPCD marker genes. The key dPCD marker gene BFN1, which encodes a nuclease, was initially expressed in the bottom regions of the clv3-2 IM domain at 3 WAB. Then, the expression area of BFN1 moved upward but still did not enter into the stem cell layers until 6 WAB, which is the time point at which WUS started to be terminated (Figures 4 and S2). From 7 WAB to 10 WAB, BFN1 was expressed in stem cells, resulting in stem cell death in clv3-2 IM (Figures 3 and 4). Because the whole clv3-2 IM was dead after 11 WAB, there were no BFN1 signals in the IM region ( Figure 4). It has been reported that the NAC transcription factor ORESARA1 (ORE1) directly regulates BFN1 and that the expression patterns of these two senescence-enhanced genes largely overlap during leaf senescence [18]. However, the spatial-temporal expression profiles of ORE1 during IM senescence are still unknown. As shown in Figure 5, there were no ORE1 expression signals (blue color) in the IM domains in WT and clv3-2 mutants from 1 WAB to 2 WAB. At 3 WAB, ORE1 was expressed in the bottom area of the IM domain in both WT and clv3-2 mutants. In WT, the expression domain of ORE1 moved upward at 4 WAB and showed clear signals in stem cell layers at 5 WAB. The expression signal of ORE1 could not be detected at 6 WAB because the whole WT IM was dead (Figures 5 and S1). In contrast, in clv3-2 mutants, the expression region of ORE1 started to be observed in the upper layers from 4 WAB to 6 WAB. From 7 WAB to 10 WAB, ORE1 expression signals could be detected in stem cell layers. Because the whole clv3-2 IM was dead, no ORE1 expression signals could be detected at 11 WAB or 12 WAB (Figures 5 and S1). These results proved that the spatial-temporal expression profiles of BFN1 and ORE1 were almost consistent during the IM senescent period in WT and clv3-2 mutants and suggested that the ORE1-BFN1 cascade might be involved in programmed stem cell death regulation. It has been reported that the NAC transcription factor ORESARA1 (ORE1) directly regulates BFN1 and that the expression patterns of these two senescence-enhanced genes largely overlap during leaf senescence [18]. However, the spatial-temporal expression profiles of ORE1 during IM senescence are still unknown. As shown in Figure 5, there were no ORE1 expression signals (blue color) in the IM domains in WT and clv3-2 mutants from 1 WAB to 2 WAB. At 3 WAB, ORE1 was expressed in the bottom area of the IM domain in both WT and clv3-2 mutants. In WT, the expression domain of ORE1 moved upward at 4 WAB and showed clear signals in stem cell layers at 5 WAB. The expression signal of ORE1 could not be detected at 6 WAB because the whole WT IM was dead (Figures 5 and S1). In contrast, in clv3-2 mutants, the expression region of ORE1 started to be observed in the upper layers from 4 WAB to 6 WAB. From 7 WAB to 10 WAB, ORE1 expression signals could be detected in stem cell layers. Because the whole clv3-2 IM was dead, no ORE1 expression signals could be detected at 11 WAB or 12 WAB (Figures 5 and S1). These results proved that the spatial-temporal expression profiles of BFN1 and ORE1 were almost consistent during the IM senescent period in WT and clv3-2 mutants and suggested that the ORE1-BFN1 cascade might be involved in programmed stem cell death regulation.

Application of Exogenous H 2 O 2 Influences WUS and ORE1 Expression
Based on our previously published results [2] and the abovementioned evidence ( Figures 1-5), we speculated that the conversion of ROS components in the IM domain governs stem cell activity termination and stem cell senescence and death. To verify this hypothesis, an exogenous H 2 O 2 assay was performed. First, we studied the effects of exogenous H 2 O 2 on WUS expression under different concentrations and found that 5 mM, 10 mM, and 20 mM H 2 O 2 could inhibit WUS expression after one week of continuous treatment (Figures 6 and S3). This result indicated that 5 mM exogenous H 2 O 2 treatment was sufficient to terminate WUS expression. Next, we studied how soon WUS expression responds to 5 mM exogenous H 2 O 2 treatment. As shown in Figure 6, WUS could respond to exogenous H 2 O 2 after only 1 day of treatment. These results revealed that WUS was sensitive to exogenous H 2 O 2 .

Application of Exogenous H2O2 Influences WUS and ORE1 Expression
Based on our previously published results [2] and the abovementioned evidence (Figures 1-5), we speculated that the conversion of ROS components in the IM domain governs stem cell activity termination and stem cell senescence and death. To verify this hypothesis, an exogenous H2O2 assay was performed. First, we studied the effects of exogenous H2O2 on WUS expression under different concentrations and found that 5 mM, 10 mM, and 20 mM H2O2 could inhibit WUS expression after one week of continuous treatment (Figures 6 and S3). This result indicated that 5 mM exogenous H2O2 treatment was sufficient to terminate WUS expression. Next, we studied how soon WUS expression responds to 5 mM exogenous H2O2 treatment. As shown in Figure 6, WUS could respond to exogenous H2O2 after only 1 day of treatment. These results revealed that WUS was sensitive to exogenous H2O2.

Application of Exogenous H2O2 Influences WUS and ORE1 Expression
Based on our previously published results [2] and the abovementioned evidence (Figures 1-5), we speculated that the conversion of ROS components in the IM domain governs stem cell activity termination and stem cell senescence and death. To verify this hypothesis, an exogenous H2O2 assay was performed. First, we studied the effects of exogenous H2O2 on WUS expression under different concentrations and found that 5 mM, 10 mM, and 20 mM H2O2 could inhibit WUS expression after one week of continuous treatment (Figures 6 and S3). This result indicated that 5 mM exogenous H2O2 treatment was sufficient to terminate WUS expression. Next, we studied how soon WUS expression responds to 5 mM exogenous H2O2 treatment. As shown in Figure 6, WUS could respond to exogenous H2O2 after only 1 day of treatment. These results revealed that WUS was sensitive to exogenous H2O2.  In addition, the proORE1::GUS line was treated for one week with exogenous H 2 O 2 . After 5, 20, and 40 mM H 2 O 2 treatments, the expression area of ORE1 was similar to that in the control group (mock), indicating that ORE1 did not respond to low concentrations of H 2 O 2 (Figures 7 and S4). Interestingly, the expression signals of ORE1 could be detected in stem cell layers after one week of 50 mM H 2 O 2 treatment (Figure 7). These results revealed that WUS inhibition and ORE1 induction by H 2 O 2 showed different threshold values.
In addition, the proORE1::GUS line was treated for one week with exogenous H2O2. After 5, 20, and 40 mM H2O2 treatments, the expression area of ORE1 was similar to that in the control group (mock), indicating that ORE1 did not respond to low concentrations of H2O2 (Figures 7 and S4). Interestingly, the expression signals of ORE1 could be detected in stem cell layers after one week of 50 mM H2O2 treatment (Figure 7). These results revealed that WUS inhibition and ORE1 induction by H2O2 showed different threshold values.

Effects of Endogenous O 2 ·and H2O2 Scavengers
To confirm whether endogenous ROS can regulate WUS expression during the proliferative stage, we used the ROS scavengers N,N'-dimethylthiourea (DMTU) and potassium iodide (KI) to reduce endogenous O 2 ·and H2O2, respectively. After one week of 20 mM DMTU treatment (the treatment was at 1 WAB), the inflorescence at 2 WAB displayed a clear flowering-arrested phenotype, while several flowers bloomed normally in the control group (mock) ( Figure 8A,B). To determine whether the scavenger DMTU effectively inhibits O 2 ·production in the WT IM domain, we carried out NBT staining of IM at 2 WAB. The results showed that there was no significant O 2 ·signal (blue color) after one week of DMTU treatment. In contrast, a strong O 2 ·signal could be detected in the entire IM domain ( Figure 8C,D). It has been reported that O 2 ·accumulation can maintain WUS expression and that the delicate Arabidopsis seedlings cannot generate SAM [10]. Given this, we tried to understand the role of O 2 ·in regulating WUS expression in proliferative IM. The longitudinal section results of the proWUS::GUS line revealed that after DMTU treatment, the WUS expression signal disappeared in the IM at 2 WAB ( Figure 8F), which should show the WUS expression signal ( Figure 8E, mock), suggesting that endogenous O 2 ·accumulation could also regulate WUS expression in proliferative IM.

Effects of Endogenous O ·− 2 and H 2 O 2 Scavengers
To confirm whether endogenous ROS can regulate WUS expression during the proliferative stage, we used the ROS scavengers N,N -dimethylthiourea (DMTU) and potassium iodide (KI) to reduce endogenous O ·− 2 and H 2 O 2 , respectively. After one week of 20 mM DMTU treatment (the treatment was at 1 WAB), the inflorescence at 2 WAB displayed a clear flowering-arrested phenotype, while several flowers bloomed normally in the control group (mock) (Figure 8A,B). To determine whether the scavenger DMTU effectively inhibits O ·− 2 production in the WT IM domain, we carried out NBT staining of IM at 2 WAB. The results showed that there was no significant O ·− 2 signal (blue color) after one week of DMTU treatment. In contrast, a strong O ·− 2 signal could be detected in the entire IM domain (Figure 8C,D). It has been reported that O ·− 2 accumulation can maintain WUS expression and that the delicate Arabidopsis seedlings cannot generate SAM [10]. Given this, we tried to understand the role of O ·− 2 in regulating WUS expression in proliferative IM. The longitudinal section results of the proWUS::GUS line revealed that after DMTU treatment, the WUS expression signal disappeared in the IM at 2 WAB ( Figure 8F), which should show the WUS expression signal ( Figure 8E, mock), suggesting that endogenous O ·− 2 accumulation could also regulate WUS expression in proliferative IM. It has been reported that H 2 O 2 accumulation can repress WUS expression in vegetative SAM [10] and possibly inhibit WUS expression in proliferative IM [2]. Next, we wondered whether endogenous H 2 O 2 accumulation restrains WUS expression in proliferative IM. At 3 WAB, the inflorescence of the control group (mock) displayed a flowering-arrested phenotype ( Figure 9A). However, IM treated with 5 mM KI for one week showed relatively high activity due to several opening flowers surrounding the IM ( Figure 9B) Figure 9C,D). Furthermore, we found that WUS was certainly transcribed in the KI-treated IM but terminated in the control IM ( Figure 9E,F). These results suggested that endogenous H 2 O 2 accumulation in proliferative IM could repress WUS expression.

The Expression Profiles of ROS Metabolism-Related Genes
In our previous work, a total of eight ROS metabolism-related factors, including two catalase (CAT)-encoding genes, three peroxidase (PRX)-encoding genes (PRX51, PRX53, and PRX70), one α-dioxygenase (DOX2)-encoding gene, one glutathione peroxidase 6 (GPX6)-encoding gene, and one acyl-CoA oxidase 1 (ACX1)-encoding gene, were identi-fied by RNA sequencing [2]. In this study, we further submitted these candidate genes to the BAR database to check their expression patterns in shoot apexes at vegetative, converted (i.e., the transition from the vegetative stage to the proliferative stage), and inflorescent (i.e., proliferative stage) stages. The results showed that only CAT3 and ACX1, which have opposite roles in H 2 O 2 metabolism, were highly expressed in all three types of shoot apex tissues ( Figure 10A), indicating that they might play roles in regulating shoot apex development. In addition, we compared our previous RNA-sequencing data (4 WAB vs. 2 WAB, shoot apex) [2] with the published transcriptome data of growing and arrested shoot meristem [19]. Only CAT3 was identified as a common gene in the overlap between 75 and 8 ROS metabolism-related genes ( Figure 10B). To understand the internal expression profiles of CAT3 and ACX1 in WT and clv3-2 IM tissues, we carried out a quantitative real-time PCR (qRT-PCR) assay using IM tissues at 2 WAB and 4 WAB. The results indicated that CAT3 was upregulated at 4 WAB in both WT and clv3-2 IMs compared with that at 2 WAB, but the expression levels in clv3-2 IM were significantly increased at 4 WAB compared with that in WT. In addition, ACX1 was significantly induced at 4 WAB in WT IM, while there was no change in clv3-2 IM at the same time point ( Figure 10C). These results revealed that CAT3 and ACX1 might take part in H 2 O 2 metabolism during shoot apex development and that the clv3-2 mutant might have a higher ability of H 2 O 2 clearance than WT.  It has been reported that H2O2 accumulation can repress WUS expression in vegetative SAM [10] and possibly inhibit WUS expression in proliferative IM [2]. Next, we wondered whether endogenous H2O2 accumulation restrains WUS expression in proliferative IM. At 3 WAB, the inflorescence of the control group (mock) displayed a flowering-arrested phenotype ( Figure 9A). However, IM treated with 5 mM KI for one week showed relatively high activity due to several opening flowers surrounding the IM ( Figure 9B). Then, we monitored H2O2 accumulation in control and KI-treated IMs. Compared with

The Expression Profiles of ROS Metabolism-Related Genes
In our previous work, a total of eight ROS metabolism-related factors, including two catalase (CAT)-encoding genes, three peroxidase (PRX)-encoding genes (PRX51, PRX53, and PRX70), one α-dioxygenase (DOX2)-encoding gene, one glutathione peroxidase 6 (GPX6)-encoding gene, and one acyl-CoA oxidase 1 (ACX1)-encoding gene, were identified by RNA sequencing [2]. In this study, we further submitted these candidate genes to the BAR database to check their expression patterns in shoot apexes at vegetative, converted (i.e., the transition from the vegetative stage to the proliferative stage), and inflorescent (i.e., proliferative stage) stages. The results showed that only CAT3 and ACX1, which have opposite roles in H2O2 metabolism, were highly expressed in all three types of shoot apex tissues ( Figure 10A), indicating that they might play roles in regulating shoot apex development. In addition, we compared our previous RNA-sequencing data (4 WAB vs. 2 WAB, shoot apex) [2] with the published transcriptome data of growing and arrested shoot meristem [19]. Only CAT3 was identified as a common gene in the overlap between 75 and 8 ROS metabolism-related genes ( Figure 10B). To understand the internal expression profiles of CAT3 and ACX1 in WT and clv3-2 IM tissues, we carried out a quantitative real-time PCR (qRT-PCR) assay using IM tissues at 2 WAB and 4 WAB. The results indicated that CAT3 was upregulated at 4 WAB in both WT and clv3-2 IMs compared with that at 2 WAB, but the expression levels in clv3-2 IM were significantly increased at 4 WAB compared with that in WT. In addition, ACX1 was significantly induced at 4 WAB in WT IM, while there was no change in clv3-2 IM at the same time point ( Figure 10C). These results revealed that CAT3 and ACX1 might take part in H2O2 metabolism during shoot  One is the comparative data using growing and arrested meristems, and the other is the comparative data using 2 WAB and 4 WAB shoot apexes. (C) The expression profiles of the ROS clearance-related gene CAT3 and ROS production-related ACX1 at 2 WAB and 4 WAB in WT and clv3-2 IM tissues. qRT-PCR assays were performed. Each experiment was replicated three times, and error bars indicate SD. One-way ANOVA post Tukey's HSD test (p < 0.05) was carried out to calculate the differences among different groups. Different letters indicate significant differences, while the same letters indicate no significant differences.

Discussion
Senescence and death are inevitable processes for tissues, organs, somatic cells, and even stem cells. In Arabidopsis, WUS plays essential roles in governing shoot and floral meristem identity, reproductive organ development, and embryogenesis [20,21]. Therefore, the termination of WUS expression in IM is the precondition for stem cells entering senescence and death processes [2]. Given that WUS directly determines stem cell activity and longevity and thus affects plant lifespan, it is necessary to determine the age-dependent repressor of WUS. Although the preceding evidence gives us some cues [2,10], the relationships among stem cell marker genes, ROS components, and age-dependent dPCD during stem cell senescence and death processes are still largely unknown. In this study, Figure 10. The expression profiles of ROS metabolism-related genes. (A) The expression patterns of eight ROS metabolism-related genes in the BAR database. The gene expression data in vegetative, converted (i.e., the transition from the vegetative stage to the proliferative stage), and inflorescent (i.e., proliferative stage) shoot apex are shown. (B) Common gene isolation between two published RNA-seq datasets. One is the comparative data using growing and arrested meristems, and the other is the comparative data using 2 WAB and 4 WAB shoot apexes. (C) The expression profiles of the ROS clearance-related gene CAT3 and ROS production-related ACX1 at 2 WAB and 4 WAB in WT and clv3-2 IM tissues. qRT-PCR assays were performed. Each experiment was replicated three times, and error bars indicate SD. One-way ANOVA post Tukey's HSD test (p < 0.05) was carried out to calculate the differences among different groups. Different letters indicate significant differences, while the same letters indicate no significant differences.

Discussion
Senescence and death are inevitable processes for tissues, organs, somatic cells, and even stem cells. In Arabidopsis, WUS plays essential roles in governing shoot and floral meristem identity, reproductive organ development, and embryogenesis [20,21]. Therefore, the termination of WUS expression in IM is the precondition for stem cells entering senescence and death processes [2]. Given that WUS directly determines stem cell activity and longevity and thus affects plant lifespan, it is necessary to determine the age-dependent repressor of WUS. Although the preceding evidence gives us some cues [2,10], the relationships among stem cell marker genes, ROS components, and age-dependent dPCD during stem cell senescence and death processes are still largely unknown. In this study, our results mainly reveal the possible linkage between stem cell marker genes and ROS metabolism in aging stem cells undergoing dPCD.

Dynamic Changes in ROS Components May Play Key Roles in Controlling Stem Cell Longevity in Arabidopsis thaliana
It is well known that the WUS-CLV3 negative feedback loop regulates stem cell activity and maintenance [22,23]. To date, several studies have proven that the WUS-CLV3 negative feedback loop can be modulated by different factors [7]. Among these factors, the ROS signaling pathway plays an essential role in regulating the stem cell population. For instance, loss of function of the mitochondrial protease AtFTSH4 results in the abnormal accumulation of ROS in the shoot apex and therefore induces SAM termination [24]. These findings indicate the regulatory link between the WUS-CLV3 feedback loop and ROS homeostasis in controlling stem cell development during the vegetative and proliferative stages of Arabidopsis.
During the proliferative, senescent, and dying stages of stem cells, we found that a dynamic conversion of ROS components from O ·− 2 to H 2 O 2 was carried out in both WT and clv3-2 mutants but in a different time series (Figures 1 and 2). The WUS expression period fully overlapped with the time window of detection of O ·− 2 in WT and clv3-2 mutants (Figures 1 and S2). More importantly, the WT IM showed a precocious flowering-arrested phenotype and WUS expression termination after endogenous O ·− 2 was removed by the scavenger DMTU (Figure 8). These results revealed that O ·− 2 can promote WUS expression during stem cell aging processes. It has been reported that another ROS component, H 2 O 2 , accumulates in PZ and inhibits the production of O ·− 2 during stem cell development [10]. Our data indicate that the accumulation of H 2 O 2 in the SAM domain inhibited the production of O ·− 2 during the conversion stage. Thus, the spatial balance between O ·− 2 and H 2 O 2 in the shoot apex is applicable to temporal ROS dynamics in stem cells. In addition, the exogenous H 2 O 2 assay demonstrated that H 2 O 2 can rapidly repress WUS expression ( Figure 6). In contrast, the KI-treated IM maintained prolonged WUS activity, leading to prolonged longevity of stem cell populations (Figure 9). These results suggested that endogenous H 2 O 2 accumulation in aging IM suppressed WUS expression via an unknown straightforward pathway. Moreover, we previously found that the expression time of CLV3 was 1 week longer than WUS [2] and partially overlapped the H 2 O 2 detection time (Figure 2), suggesting that H 2 O 2 accumulation might inhibit CLV3 expression after WUS termination. In contrast, we found that the higher levels of exogenous H 2 O 2 enhanced the expression of the dPCD marker genes ORE1 and BFN1 in stem cell layers, indicating that the accumulation and burst of endogenous H 2 O 2 may directly induce dPCD in the stem cell population in a WUS-independent pathway (Figures 2-5). Based on these results, we hypothesized that dynamic changes in endogenous O ·− 2 and H 2 O 2 not only shut down the proliferative activity of IM but also activated dPCD in the stem cell population. The dynamic shift from O ·− 2 to H 2 O 2 might play a key role in governing stem cell longevity ( Figure 11). ·production and terminate WUS directly. Furthermore, the H2O2 burst may repress CLV3, leading to the full termination of the WUS-CLV3 feedback loop. Moreover, H2O2 directly triggers the dPCD process by activating the ORE1-BFN1 cascade in stem cells. Lines and arrows indicate the known processes. Dotted lines and arrow lines denote the putative processes.

The Essential Roles of Stem Cell Marker Genes in Regulating Stem Cell Fate in Arabidopsis during the Proliferative Stage
Clearly, ACX1 is involved in the production of H2O2 [25]. In contrast, CAT3 catalyzes the decomposition of H2O2 and plays a critical role in controlling the homeostasis of ROS in Arabidopsis [26]. When WUS was expressed in the IM domain, there were no detectable H2O2 signals in the IM domain, especially in stem cell layers in WT and clv3-2 mutants. Interestingly, weak H2O2 signals started to be observed in the stem layers of the WT and IM domains of clv3-2 mutants when WUS expression was terminated at 4 WAB in WT and 7 WAB in clv3-2 mutants ( Figure S2). This evidence indicates that WUS may inhibit H2O2 production through the regulation of the H2O2 biosynthetic gene ACX1. Consistent with this, the relative expression level of ACX1 at 4 WAB, when WUS was not expressed, was maximally 6-fold compared to that at 2 WAB, when WUS was expressed (Figures 10C and  S2). In addition, WUS may also promote CAT3 to eliminate H2O2 as a homeostatic regulation, resulting in a low concentration of H2O2 in stem cells. Consistently, the relative expression level of CAT3 at 4 WAB was significantly higher than that at 2 WAB due to the prolonged WUS expression period in clv3-2 mutants (Figures 10C and S2). However, we found that the relative expression intensity of CAT3 at 4 WAB was also notably higher than that at 2 WAB in WT but lower than that at 4 WAB in clv3-2 mutants ( Figure 10C), indicating that there may be some unknown signals inducing CAT3 expression. Thus, we thought that WUS might regulate the dynamic changes in ROS via the transcriptional regulatory pathway. Moreover, we thought that the dynamic changes in ROS might control WUS expression (see 'Discussion' Section 3.1). Based on these findings, we speculated that WUS-, CAT3-, and ACX1-mediated ROS changes formed a feedback pathway to regulate stem cell fate during the proliferative stage ( Figure 11).
In addition, the spatial-temporal expression levels of WUS and the dPCD marker genes ORE1 and BFN1 also reflect their potential regulatory relationship. In WT, WUS

The Essential Roles of Stem Cell Marker Genes in Regulating Stem Cell Fate in Arabidopsis during the Proliferative Stage
Clearly, ACX1 is involved in the production of H 2 O 2 [25]. In contrast, CAT3 catalyzes the decomposition of H 2 O 2 and plays a critical role in controlling the homeostasis of ROS in Arabidopsis [26]. When WUS was expressed in the IM domain, there were no detectable H 2 O 2 signals in the IM domain, especially in stem cell layers in WT and clv3-2 mutants. Interestingly, weak H 2 O 2 signals started to be observed in the stem layers of the WT and IM domains of clv3-2 mutants when WUS expression was terminated at 4 WAB in WT and 7 WAB in clv3-2 mutants ( Figure S2). This evidence indicates that WUS may inhibit H 2 O 2 production through the regulation of the H 2 O 2 biosynthetic gene ACX1. Consistent with this, the relative expression level of ACX1 at 4 WAB, when WUS was not expressed, was maximally 6-fold compared to that at 2 WAB, when WUS was expressed ( Figures 10C and S2). In addition, WUS may also promote CAT3 to eliminate H 2 O 2 as a homeostatic regulation, resulting in a low concentration of H 2 O 2 in stem cells. Consistently, the relative expression level of CAT3 at 4 WAB was significantly higher than that at 2 WAB due to the prolonged WUS expression period in clv3-2 mutants (Figures 10C and S2). However, we found that the relative expression intensity of CAT3 at 4 WAB was also notably higher than that at 2 WAB in WT but lower than that at 4 WAB in clv3-2 mutants ( Figure 10C), indicating that there may be some unknown signals inducing CAT3 expression. Thus, we thought that WUS might regulate the dynamic changes in ROS via the transcriptional regulatory pathway. Moreover, we thought that the dynamic changes in ROS might control WUS expression (see 'Discussion' Section 3.1). Based on these findings, we speculated that WUS-, CAT3-, and ACX1-mediated ROS changes formed a feedback pathway to regulate stem cell fate during the proliferative stage ( Figure 11).
In addition, the spatial-temporal expression levels of WUS and the dPCD marker genes ORE1 and BFN1 also reflect their potential regulatory relationship. In WT, WUS was expressed in the organizing center (OC) at 1 WAB and 2 WAB and was terminated at 3 WAB ( Figure S2) [2]. The dPCD marker genes ORE1 and BFN1 displayed quite similar expression profiles. Both genes were initially expressed in the Rib zone in WT and clv3-2 mutants. The appearance of dPCD marker genes in stem cell layers was 1 week later than WUS termination in both WT and clv3-2 mutants (Figures 4, 5 and S2). These results suggested that the expression of WUS might indirectly but strictly restrict the activities of dPCD marker genes in the IM domain, thereby protecting the stem cell population to prevent the dPCD process from untimely activation during the proliferative stage in Arabidopsis. Furthermore, the results indicated that the possible functions of another stem cell marker gene, CLV3, may restrict the production of O ·− 2 . Compared with WT, the existence time of O ·− 2 is prolonged in clv3-2 mutants (Figure 1). In addition, CLV3 can repress WUS expression, but how CLV3 signaling leads to WUS repression is still unclear. Our previous study revealed that CLV3 expression time is sustained until 4 WAB, while the expression of WUS was terminated at 3 WAB [2], and that loss-of-function of CLV3 prolonged the expression time of WUS ( Figure S2) [2]. Combining these findings, it is highly possible that CLV3 may prevent WUS reactivation via the inhibition of the O ·− 2 metabolic pathway ( Figure 11).

Plant Materials and Growth Conditions
In this study, Arabidopsis thaliana in the Landsberg erecta (Ler) background (wild-type) was used as the plant material. In addition, clv3-2 mutant (Ler) was also used as the plant material and has already been described in the previous study [27]. Besides, the WUS promotor reporter lines proWUS::GFP-ER and proWUS::GUS were used to detect the expression pattern of WUS and have been reported previously [2]. The proBFN1::GUS-GFP [2] and proORE1::GUS-GFP (see Section 4.6) in the clv3-2 background were generated by crossing with the clv3-2 mutant line. Arabidopsis seeds were sown in pots containing vermiculite and nutrition soil (Baseconnect, Osaka, Japan) and incubated at 4 • C in the dark for 3 days to promote germination. All plants were cultured in an illumination incubator (BiOTRON, LPH-411SP, Osaka, Japan) under a 16-h light (100 µmol m −2 s −1 )/8-h dark light cycle with 60% humidity at 22 • C.

Phenotypic Definitions and Measurements
In our previous study, we had established a time system to observe the development of each Arabidopsis plant precisely. The term "weeks after bolting" (WAB) was used as the temporal unit [28]. The stem length was measured every day until it approached 1 cm. We defined this time point as the beginning of bolting [29]. In order to count the number of flowers on the main shoot, siliques and flowers (older than stage 7) were counted at the same time. The developmental stage of flower was described previously [30]. Morphological observations of inflorescences on main shoots were photographed using a Canon EOS 600D optical camera (Canon, Tokyo, Japan).

GUS Staining and Tissue Sectioning
For tissues sectioning, inflorescences of main shoots of promotor reporter lines were cut off without any dissection and fixed in 90% acetone solution for 15 min at room temperature. Subsequently, the tissues were rinsed with double-distilled water, and then stained using GUS staining solution. The GUS staining method was previously described by Shirakawa et al. [31]. Tissue sectioning method was described by Yamaguchi et al. previously [32]. The slides were stained with 0.05% neutral red (Wako Chemicals, Kyoto, Japan) for 5 min and rinsed with double-distilled water.

Confocal Microscopy
The IMs on proWUS::GFP-ER primary shoots were used to observe the GFP signals. The floral buds on inflorescences older than stage 7 were cut off with tweezers under a light microscope. The IMs were embedded into 5% agar (Wako Chemicals, Kyoto, Japan) and sliced with a Liner Slicer PRO7 vibratome (Dosaka, Kyoto, Japan) [32]. The resulting tissue sections were immersed in moderate volumes of 1/10 Murashige and Skoog (MS) [33] solution on glass slides. The GFP signal was immediately observed under an FV 1000 (Olympus, Tokyo, Japan) microscope with FV10-ASW software (https://www.olympus-lifescience.com.cn/en/support/ downloads/#!dlOpen=%23detail847249651 (accessed on 7 March 2022); see Laser Scanning Microscopes option). GFP was excited using the 488-nm argon laser and its emission signal was monitored between 495 and 545 nm.

Reverse-Transcription PCR and Quantitative Real-Time PCR (qRT-PCR)
For the extraction of total RNA, the RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) was used. To prevent the DNA contamination, the RNase-Free DNase Set (Qiagen, Hilden, Germany) was used to digest the genomic DNA in the RNA samples. Reverse-transcription PCR was carried out using PrimeScript™ RT Master Mix (Takara, Shiga, Japan). The method of qRT-PCR was described previously [32]. Arabidopsis ACTIN2 (AT3G18780) [34] was used as the internal reference. The relative expression levels of genes were showed the ratio of gene to ACTIN2. Each qRT-PCR experiment was performed three times repeatedly. The primers used are listed in Table S1.

Plasmid Construction and Plant Transformation
The proBFN1::GUS-GFP transgenic line was reported previously [2]. The proORE1::GUS-GFP line (2.5 kb promoter) was generated using the same method mentioned previously [2]. The Agrobacterium-mediated floral dip method [35] was carried out to generate transgenic plants. T1 seeds were collected and screened using the chemical Basta (Wako Chemicals, Kyoto, Japan). More than 20 T1 plants were obtained, and the representative lines were chosen for further study. The primers used are listed in Table S1.

FDA and PI Staining
FDA and PI staining methods were described previously [2]. Confocal microscope observation was performed using previously published methods [2].

Exogenous H 2 O 2 and ROS Scavenger Treatments
For exogenous H 2 O 2 treatment in the proWUS::GUS line, 1 WAB-old primary shoot inflorescences were immersed in a 5, 10, or 20 mM H 2 O 2 (WAKENYAKU, Shiga, Japan) solution containing 0.01% Silwet L-77 for 10 s each day. The treatment was maintained for one week. After confirmation of the optimal concentration, primary shoot inflorescences of the proWUS::GUS line at 1 WAB were immersed in a 5 mM H 2 O 2 solution to treat for 1, 2, 3, 4, 5, and 6 days to confirm the response speed of WUS. For exogenous H 2 O 2 treatment in proORE1::GUS-GFP and proBFN1::GUS-GFP lines, primary shoot inflorescences at 2 WAB were immersed in a 5, 20, 40, or 50 mM H 2 O 2 solution for 10 s each day. The treatment was maintained for one week. For the control group, double distilled water containing 0.01% Silwet L-77 was used. For each experiment, at least 5 individual inflorescences were treated.
For ROS scavenger treatment in the WT and proWUS::GUS lines, primary shoot inflorescences at 1 WAB were immersed in a 20 mM DMTU (Sigma-Aldrich, Tokyo, Japan) solution containing 0.01% Silwet L-77 for 10 s each day, and primary shoot inflorescences at 2 WAB were immersed in a 5 mM KI (Sigma-Aldrich, Tokyo, Japan) solution containing 0.01% Silwet L-77 for 10 s each day. The treatment was maintained for one week. For the control group, double distilled water containing 0.01% Silwet L-77 was used. For each experiment, at least 5 individual inflorescences were treated.

Data Statistics and Availability
One-way ANOVA post Tukey's HSD test (p < 0.05) was carried out to calculate the differences among different groups. Different letters indicate significant differences, while the same letters indicate no significant differences. RNA-seq datasets were downloaded from the DNA Data Bank of Japan (DDBJ) with the accession number DRA010789 [2] and from the National Center for Biotechnology Information GEO with the accession numbers GSE74386 and GSE79287 [19]. The expression patterns of 8 ROS-related genes were obtained from the BAR database (http://bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi) (accessed on 7 March 2022).

Conclusions
In this study, we examined the dynamic changes in the ROS components O ·− 2 and H 2 O 2 in the stem cell population and revealed their potential regulatory relationship between stem cell marker genes using morphological and physiological methods. Based on the results of this study, we established an ROS-mediated regulatory model of the age-dependent dPCD process during stem cell death. In our putative model, the WUS-CLV3 feedback loop was the regulatory core: at the proliferative stage, the high level of O ·− 2 maintained WUS expression, while CLV3 repressed WUS expression at the same time. The expression of WUS not only prevented the activation of the ORE1-BFN1 cascade, thereby avoiding dPCD in the stem cell population, but also might promote CAT3 and repress ACX1 to maintain a low level of H 2 O 2 . In addition, it was possible that CLV3 inhibited O ·− 2 production. Along with the growth of Arabidopsis, the shoot stem cells started to enter the senescent and dead stages, and H 2 O 2 inevitably and significantly accumulated in IM. A high level of H 2 O 2 further suppressed O ·− 2 production, resulting in O ·− 2 becoming undetectable in IM. Because the WUS-CLV3 loop was ultimately collapsed by high levels of H 2 O 2 , the ORE1-BNF1 cascade was activated, leading to dPCD in the stem cell population. Notably, the promotion of ACX1 might be a key factor in H 2 O 2 accumulation in IM ( Figure 11). More molecular and genetic studies mainly focusing on CAT3 and ACX1 should be performed in the future to reveal the ROS-mediated mechanism of stem cell death.