Unless stated otherwise, all chemicals were purchased from Sigma (St Louis, MO, United States). According to previous reports (Li et al., 2015; Zhu et al., 2016), the preparation of β-CD-hemin (β-CDH) was carried out. Hemin (used as an inducer of HO-1) and β-CD with an appropriate molar ratio were mixed by grinding for at least 60 min after adding de-ionized water. After freeze-dried, the brown powder was regarded as β-CDH. Our pilot experiment confirmed that 1 nM β-CDH which contains 1 nM hemin and 500 nM β-CD, exhibited a maximal response in the induction of tomato LR (Li et al., 2015).

Hydrogen peroxide (H₂O₂), as a positive control, was applied at 100 μM. N,N′-dimethylthiourea (DMTU; Ma et al., 2014), a membrane-permeable scavenger of H₂O₂, was used at a final concentration of 500 μM. Ascorbic acid (AsA; another membrane-permeable scavenger of H₂O₂) purchased from Solarbio Life Sciences (Beijing, China), was used at 200 μM. Diphenyleneiodonium (DPI), a NADPH oxidase inhibitor (Xie et al., 2011), was used at 0.1 μM. According to our pilot experiments, the concentrations of above chemicals exhibiting the effective responses were selected.

Tomato (Solanum lycopersicum L.) seeds “baiguoqiangfeng” were obtained from Jiangsu Academy of Agricultural Sciences. Selected seeds were surface-sterilized with 2% NaClO for at least 10 min, and germinated in distilled water at 25 ± 1°C in the dark for 2 days. Afterward, tomato seedlings were transferred to an illuminating incubator (25 ± 1°C) with a light intensity of 200 μmol m^-2 s^-1 at 14-h photoperiod. After growing for 1 day, the selected identical seedlings were transferred to 4 ml solution containing the indicated chemicals for the indicated time points. Afterward, photographs were taken, and the number of emerged LRs (LRs; >1 mm) per seedling and the length of primary root (PR), as well as the emerged LR density (the number of LR per cm PR; LRs/cm) were determined with Image J software. LR primordia (LRP) per seedling were also observed by root squash preparations and quantified by a light microscope (model Stemi 2000-C; Carl Zeiss, Germany; Correa-Aragunde et al., 2006). In our test, at least three independent experiments were carried out for each treatment, and at least 15 seedlings were used for each.

For the subsequent biochemical, molecular and proteomics analyses, only the LR-inducible segments were used. Therefore, the root apical meristems of seedlings at the indicated time points were cut off, and the shoots were removed by cutting below the root-shoot junction (Zhu et al., 2016).

H₂O₂ signals were assessed by a laser confocal scanning microscopy (LCSM) using the ROS fluorescent probe 2′,7′-dichlorofluorescein diacetate (H₂DCF-DA) (Maffei et al., 2006; Li et al., 2011; Xie et al., 2011). Also, DMTU and AsA, two membrane-permeable scavengers of H₂O₂, were used to confirm its specificity. Roots were incubated in HEPES buffer (20 mM, pH 7.5) which contains 20 μM H₂DCF-DA for 30 min in dark (25°C). Then the fresh HEPES buffer was used to wash three times. All images were visualized by using UltraVIEW VoX (Perkin Elmer, Waltham, MA, United States). Thereafter, photographs were representative of identical results obtained after the processing and analysis of seven samples for each condition in three independent experiments. Volocity Demo software was used to quantify the production of H₂O₂ in roots.

Total RNA was isolated using the Trizol reagent (Invitrogen, Gaithersburg, MD, United States) according to the manufacturer’s instructions. The RNA samples were treated with RNAase-free DNase (TaKaRa Bio Inc., Dalian, China) to eliminate traces of DNA, followed by the quantification by using the NanoDrop 2000 (Thermo Fisher Scientific, Wilmington, DE, United States). Afterward, total RNA (2 μg) was reverse-transcribed using an oligo(dT) primer and M-MLV reverse transcriptase (BioTeke, Beijing, China).

Real-time qPCR reactions were performed using a Mastercycler^® ep realplex real-time PCR system (Eppendorf, Hamburg, Germany) with SYBR^®Premix Ex Taq^TM (TransGen Biotech, Beijing, China) according to the manufacturer’s instructions. The primer sequence information was listed in Supplementary Table S1. Relative expression levels of corresponding genes were presented as values relative to the control samples at the indicated time points, after normalization with Actin transcript levels (Zhu et al., 2016).

The total proteins in tomato root tissues were extracted by Plant Total Protein Extraction Kit (Sigma-Aldrich, St Louis, MO, United States). Protein samples (200 μg BSA equivalent) were digested using filter-aided sample preparation (FASP) method (Wiśniewski et al., 2009). The protein extraction was reduced with 10 mM dithiothreitol (DTT) for 1 h at 56°C, and then alkylated with 55 mM iodoacetamide (IAA) for 45 min at 25°C in darkness. Afterward, the protein samples were buffer-exchanged with 100 mM NH₄HCO₃ (pH 8.0–8.5) using 10 kDa molecular weight cut-off Amicon Spin Tube (Millipore, Billerica, MA, United States). Subsequently, 4 μg of sequencing-grade modified trypsin (Promega) was added to each sample, and digestion was carried out overnight at 37°C (trypsin: protein ration = 1: 50). Digested peptides were desalted by Ziptip C18 (Milipore) and quantified using a NanoDrop 2000 spectrophotometer (Wilmington, United States).

For LC-MS/MS conditions, a label-free quantitative method was used to detect the relative amount of proteins. Three biological replicates from three independent experiments (about 60 roots for each independent experiment) of β-CDH-treated and control groups were analyzed by nano LC system (Dionex, part of Thermo Fisher Scientific) on-line coupled to LTQ-Orbitrap mass spectrometer (Thermo Electron, Bremen, Germany). The resulting peptides (1.5 μg) were acidified with 0.1% formic acid (FA), and subsequently loaded onto the nano trap column (Acclaim PepMap100 C18, 75 μm × 2 cm, 3 μm, 100 Å, Thermo Scientific) at a flow rate of 4 μL⋅min^-1 in loading buffer (2% acetonitrile, 0.1% FA in HPLC-grade water). Chromatographic separation was carried out on the analytical column (Acclaim PepMap^®RSLC, C18, 75 μm × 15 cm, 3 μm, 100Å, Thermo Scientific) using a linear gradient of 3–55% buffer B (80% acetonitrile and 0.1% FA) at a flow rate of 0.25 μL⋅min^-1 over 112 min. Due to loading and washing steps, the total time for an LC-MS/MS run was 160 min longer. For LTQ-Orbitrap analysis, one scan cycle included an MS1 scan (m/z 300–1800) at a resolution of 60,000, followed by 10 MS2 scans by LTQ, to fragment the 10 most abundant precursor ions at normalized collision energy of 35 eV. The lock mass calibration was activated, and dynamic exclusion time was set to 30 s.

Raw data were analyzed by MaxQuant (version 1.5.2.5) (Tyanova et al., 2016) using standard settings with the additional options match between runs, and LFQ selected. The generated ‘proteingroups.txt’ table was filtered for contaminants, reverse hits, and number of unique peptides (>0) in Perseus (from MaxQuant package).

Where indicated, results were expressed as the mean values ± SE of at least three independent experiments (with at least three replicates for each) with similar results. Statistical analysis was performed using SPSS 17.0 software. For statistical analysis, one-way analysis of variance (ANOVA) followed by Duncan’s multiple range test (P < 0.05) was chosen.

First, tomato seedlings were loaded with reactive oxygen species (ROS)-specific fluorescent dye 2′,7′-dichlorofluorescein diacetate (H₂DCF-DA), and LCSM was used to investigate changes in ROS-induced fluorescence. Since the DCF-dependent green fluorescence detected in 100 μM H₂O₂-treated tomato seedlings for 12 h, was obviously impaired following the addition of DMTU and AsA, two membrane-permeable scavengers of H₂O₂ (Figure 1), the visual signal can be mostly ascribed to endogenous H₂O₂ accumulation. Thus, the fluorescence was used to report endogenous H₂O₂ levels subsequently.

Further result showed that, compared to the control sample, the addition of 1 nM β-CDH for 12 h was able to induce endogenous H₂O₂ production in tomato seedlings, mimicking the response of H₂O₂ when was exogenously applied. We also noticed that this time point of H₂O₂ production triggered by β-CDH and exogenous H₂O₂, apparently preceded LR formation, beginning at 48 h of treatments (Li et al., 2015). Above results indicated the possible link between endogenous H₂O₂ production and LR formation triggered by β-CDH.

In order to evaluate the possible role of endogenous H₂O₂ in β-CDH-induced LR development, DMTU and AsA were also used. Similar to the previous reports (Ma et al., 2014; Li et al., 2015; Zhu et al., 2016), both 1 nM β-CDH and 100 μM H₂O₂ increased tomato LR density and number (Figures 2A–C). Meanwhile, no significant difference in PR length was observed (Figure 2D). By contrast, the co-treatment with DMTU and AsA respectively not only blocked endogenous H₂O₂ production (Figure 1), but also arrested the thereafter induction of LR formation (Figure 2), triggered by exogenous β-CDH and H₂O₂. When applied alone, DMTU (in particularly) and AsA could inhibit LR formation respect to the chemical-free control plants. Meanwhile, endogenous H₂O₂ levels were also decreased (Figure 1).

For the origin of endogenous H₂O₂, the plasma-membrane (PM) NADPH oxidase confers important roles in H₂O₂ signaling (Desikan et al., 2006; Xie et al., 2011). Since DPI is an inhibitor of NADPH oxidase responsible for endogenous H₂O₂ production during LR formation (Ma et al., 2014), this chemical was applied together with β-CDH. Similar to the inhibition responses of DMTU and AsA (Figures 1, 2), β-CDH-induced H₂O₂ and LR formation were respectively impaired by 0.1 μM DPI, suggesting the possible role of NADPH oxidase-dependent H₂O₂ in β-CDH action. When applied alone, DPI, similar to DMTU (in particular) and AsA, could inhibit LR formation, compared to the chemical-free control plants (Figure 2). Changes in endogenous H₂O₂ displayed the similar tendencies (Figure 1).

Further microscopical analysis showed that both β-CDH- and H₂O₂-triggered LR primordial (LRP; 3 days) exhibited a similar accelerated anatomic structure, both of which were individually impaired by the cotreatment with DMTU, DPI or AsA (Figure 3). When applied alone, DMTU, DPI, or AsA strongly inhibited the development of LRP. We also noticed that above results were comparable to the phenotypes in the LR formation (Figure 2).

The inhibiting effect of DPI on β-CDH-elicited LR formation suggested the possible role of NADPH oxidase. The following experiments were carried out to test above hypothesis. As shown in Figure 4, the transcript of RBOH1 was rapidly increased after β-CDH or exogenous H₂O₂ treatments for 6 h. Meanwhile, the removal of H₂O₂ (Figure 1) by the scavengers of H₂O₂ (DMTU and AsA) and inhibitor of NADPH oxidase (DPI) completely blocked above responses. Similarly, DMTU, AsA and DPI alone also exhibited the inhibition in RBOH1 expression compared to control samples.

Furthermore, the transcripts of four cell cycle regulatory genes, CYCA3;1, CYCA2;1, CYCD3;1, and CDKA1, were analyzed by qPCR as molecular probes to further investigate the role of H₂O₂ in β-CDH-induced LR formation. After 12 h of β-CDH treatment, above transcripts were up-regulated (Figure 5). Similar results appeared in H₂O₂-treated seedlings. However, the addition with DMTU, AsA, or DPI could significantly prevent β-CDH- and H₂O₂-induced cell cycle regulatory gene expression, all of which were well matched with the LPR number, LR number and density (Figures 2, 3).

Subsequent experiment revealed that β-CDH and H₂O₂ were able to up-regulate the transcripts of auxin signaling genes (ARF1 and RSI-1), and an auxin influx carrier gene (LAX3), together with the down-regulation of IAA14 (encoding a member of the Aux/IAA protein family) and two auxin efflux carriers genes (PIN3 and PIN7; Figure 6). As we expected, the co-treatment with DMTU, AsA, or DPI differently blocked the above mentioned effects. Combined with corresponding endogenous H₂O₂ production (Figure 1) and phenotypes (Figures 2, 3), these findings suggested that above genes might be the targets of H₂O₂ signaling in β-CDH-induced tomato LR formation.

To well address molecular mechanism of β-CDH-induced LR formation, comparative proteomic analysis from tomato seedling roots in the presence or absence of β-CDH was performed with LC-MS/MS. In this study, a total of 86 proteins were identified significantly regulated (fold change > 1.5 or < 0.667) after β-CDH treatment under P value < 0.05 (Supplementary Table S2). Some ROS metabolism related proteins were modulated by β-CDH treatment, such as Isocitrate dehydrogenase [NADP] (decreased), Catalase, Succinic semialdehyde reductase isofom1 (SSR1), NADH-cytochrome b₅ reductase 1 (increased), etc (Table 1). Meanwhile, proteins related to LR formation, like L-ascorbate oxidase (increased), Protein ROOT HAIR DEFECTIVE 3, DWARF1/DIMINUTO, Phenylalanine ammonia-lyase, and Glycine rich RNA binding protein 1a, were found be regulated after β-CDH treatment. Furthermore, 20 proteins like Malic enzyme, Coatomer subunit alpha, PR10 protein, and 40S ribosomal protein S8, etc. were identified to be working in other biological process, such as metabolic process, intracellular transport, response to stress, and protein metabolic process, etc. Additionally, we also noticed that membrane-associated NADPH oxidase protein was not found in our experimental conditions.

To confirm above results, we further tested the effects of H₂O₂ scavengers and inhibitor on the transcripts of three representative genes (Table 1), isocitrate dehydrogenase [NADP], NADH-cytochrome b₅ reductase, and L-ascorbate oxidase homolog (Figure 7). Results showed that, the added H₂O₂ scavengers (DMTU and AsA) and synthetic inhibitor (DPI) could effectively prevent the down-regulation of isocitrate dehydrogenase [NADP] gene expression elicited by β-CDH and H₂O₂. Whereas, the up-regulated NADH-cytochrome b₅ reductase and L-ascorbate oxidase homolog (in particular) transcripts were blocked. These results could be well consistent with the data form LC-MS/MS (Table 1).

It was well-known that H₂O₂ functions as a signaling molecule in regulating stress responses, development, and other cell processes (Neill et al., 2002; Cuypers et al., 2016; Niu and Liao, 2016). As expected, in our experimental condition, an increased endogenous H₂O₂ was induced by β-CDH in tomato seedlings (Figure 1). Evidences showed that, there are several enzymes that can produce H₂O₂ in plants, such as cell wall peroxidase, amine oxidase, flavin-containing enzymes, and NADPH oxidase in particularly (Cona et al., 2006; Xie et al., 2011; Francoz et al., 2015; Niu and Liao, 2016). We further identified that β-CDH-elicited H₂O₂ production resulted from the up-regulation of RBOH1 gene expression (Figure 4). The involvement of NADPH oxidase in β-CDH-triggered LR formation was further corroborated by the findings that NADPH oxidase inhibitor DPI (Bai et al., 2012; Ma et al., 2014) not only inhibited H₂O₂ production (Figure 1), but also caused a significant reduction of LR formation in β-CDH-treated seedlings (Figures 2, 3). Meanwhile, the removal of endogenous H₂O₂ by DMTU and AsA exhibited the similar blocking tendencies, further confirming that LR formation elicited by β-CDH is closely related to endogenous H₂O₂ concentration. Although we can not exclude the possibility that these chemical agents used in the present study may not specifically target H₂O₂, these results clearly revealed that β-CDH-stimulated H₂O₂-mediated LR formation was partly dependent on NADPH oxidase. Consistently, a recent genetic result revealed that RBOH-mediated ROS production facilitated LR emergence in Arabidopsis (Orman-Ligeza et al., 2016).

Cell cycle activation is an important event during LR formation (Himanen et al., 2002, 2004; Casimiro et al., 2003). In the pervious reports, cell cycle genes CYCA3;1, CYCA2;1, CYCD3;1, and CDKA1 were used as the molecular markers in tomato LR formation (Correa-Aragunde et al., 2006; Xu et al., 2011). Similar to the previous results (Li et al., 2015; Zhu et al., 2016), β-CDH treatment up-regulated CYCA3;1, CYCA2;1, CYCD3;1, and CDKA1 gene expression, and these responses mimicked the effects of H₂O₂ (Figure 5; Ma et al., 2014). By contrast, above β-CDH- and H₂O₂-induced expression of cell cycle regulatory genes were prevented or delayed when H₂O₂ was scavenged by DMTU or AsA, or its synthesis was inhibited with DPI (Figure 1). Combined with the impaired LRP and LR formation by the removal of H₂O₂ levels (Figures 1–3), our findings gave further evidence, suggesting that β-CDH-triggered H₂O₂ production was able to modulate the expression of cell cycle regulatory genes and this event is also required for LR formation in tomato seedlings.

Auxin controls cell cycle progression and asymmetric divisions during LR formation (Casimiro et al., 2001; Bhalerao et al., 2002; Lavenus et al., 2013). In our experimental conditions, it was further confirmed that, similar to the responses elicited by H₂O₂, β-CDH up-regulated two auxin signaling genes (ARF7 and RSI-1; Zhu et al., 2016) and an auxin influx carrier gene (LAX3), together with the down-regulation of IAA14 (encoding a member of the Aux/IAA protein family) and two auxin efflux carriers genes (PIN3 and PIN7; Figure 6). Comparatively, the removal of endogenous H₂O₂ drastically impaired corresponding changes conferred by β-CDH and H₂O₂. Previous results showed that slr-1, a gain-of-function mutant of IAA14 exhibited a crucial defect in LR formation in Arabidopsis (Fukaki et al., 2002). Three Arabidopsis thaliana mutants, lax3, pin3 and pin7, which are defective in auxin influx and efflux proteins, showed reduced or increased LR formation (Swarup et al., 2008; Lewis et al., 2011). Combined with previous genetic results, our molecular and pharmacologic evidence further indicated a possible link between β-CDH-induced H₂O₂-mediated LR formation and auxin signaling. This deduction should be investigated at genetic levels in the near future.

Proteomic analysis showed the presence of 86 proteins which were significantly regulated by β-CDH treatment for 48 h (Supplementary Table S2). Among these proteins, some were concerned with ROS signaling (Table 1). For example, Isocitrate dehydrogenase [NADP] (NADP-ICDH; EC 1.1.1.42; K4ASC2) catalyzes oxidative decarboxylation of isocitrate to 2-oxoglutarate using NADP⁺ to form NADPH, and the latter is an important cofactor in many biosynthesis pathways and important in cellular defense against oxidative damage (Lee et al., 2002). It was reported that, NADP-ICDH can be damaged by ROS, and the inactivation of ICDH may lead to the perturbation of the antioxidant defense system in many cell process (Lee S.M. et al., 2001). In our experimental conditions, the amount of NADP-ICDH protein was decreased after β-CDH treatment (Table 1), which was in line with the increased ROS in seedling roots (Figure 1). Similarly, the protein level of the major H₂O₂ scavenging enzyme, catalase (CAT; EC 1.11.1.6; K4BAE6), was also decreased by β-CDH. This was consistent with a higher concentration of H₂O₂ in tomato seedling roots after β-CDH treatment, because the altering of CAT level can modulate H₂O₂ levels in plant cells (Vandenabeele et al., 2004). NADH-cytochrome b₅ reductase (K4D331) is found to play a key role in the NADH-dependent reduction of D-erythroascorbyl free radical, and can be active in the oxidative stress response of Saccharomyces cerevisiae (Lee J.S. et al., 2001). In HeLa cells, H₂O₂ -regulated expression of NADH-cytochrome b₅ reductase was previously reported (Bello et al., 2003). In this study, the level of NADH-cytochrome b₅ reductase protein was increased when β-CDH was supplied, also confirming that a rapid H₂O₂ production appeared in seedling roots (Figure 1).

Besides the changes of H₂O₂ and redox related proteins, the β-CDH could regulate some proteins related to LR formation, for example, _L-ascorbate oxidase (AO; EC 1.10.3.3; K4DH18), Protein ROOT HAIR DEFECTIVE 3 homolog (RHD3; K4BMV8), DWARF1/DIMINUTO (Q66YT8), Phenylalanine ammonia-lyase (PAL; EC 4.3.1.5; K4D451), and Glycine rich RNA binding protein 1a (atRZ-1a; L7Q568). The activity and expression of AO are closely correlated with cell expansion, which implies a role in cell wall loosening, cell division, and cell elongation (Kato and Esaka, 1999; Sanmartin et al., 2003). In this study, the level of AO protein was induced by β-CDH (Table 1), and there might be a link between AO expression and β-CDH-induced lateral rooting. Protein ROOT HAIR DEFECTIVE 3 (RHD3; K4BMV8) encodes an 89 kD polypeptide with putative GTP-binding motifs, with a common function in eukaryotic cell enlargement (Wang et al., 1997). The rhd3 mutation alters the size of roots and root hairs. Here, our results showed that RHD3 homolog protein level was increased by β-CDH. DWARF1/DIMINUTO (Q66YT8) gene encodes a protein involved in steroid as well as brassinosteroid (BR) synthesis (Klahre et al., 1998). BRs are known interacting with auxin to promote LR development in Arabidopsis (Bao et al., 2004). Thus, the elevated DWARF1/DIMINUTO protein level can help to lateral rooting. Phenylalanine ammonia-lyase (PAL; EC 4.3.1.5; K4D451) is reported to be highly regulated during development and xylogenesis with the cell wall polymer lignin (Elkind et al., 1990). An increased PAL protein level was found during β-CDH-induced lateral rooting (Table 1). Glycine rich RNA binding protein 1a (atRZ-1a; L7Q568) over-expression plants showed delayed germination and seedling growth under salt and drought stresses (Kim et al., 2007). Since abiotic stress could induce LR formation, an important phenomenon of the stress-induced morphogenic response (SIMR) in plants (Potters et al., 2007), we further deduced that the decreased level in atRZ-1a protein by β-CDH might lead to a positive influence in LR formation. Additionally, changes in three representative genes related to ROS metabolism and LR formation, including isocitrate dehydrogenase [NADP] (A), NADH-cytochrome b₅ reductase 1 (B), and L-ascorbate oxidase homolog (Figure 7), approximately matched with corresponding proteomic data.

In summary, our results showed a vital role of H₂O₂ in the β-CDH-induced tomato LR formation, and β-CDH-elicited H₂O₂-related target proteins might be, at least partially, regulated at transcriptional and translational levels.