Phenotypic Analysis Combined with Tandem Mass Tags (TMT) Labeling Reveals the Heterogeneity of Strawberry Stolon Buds


 Background: Ramet propagation in strawberry (Fragaria × ananassa) is the most effective way in production. However, the lack of systematically phenotypic observations and high-throughput methods limits our ability to analyze the key factors regulating the heterogeneity in strawberry stolon buds.
Results: From observation, we found that the axillary bud located in the first node quickly stepped into dormancy (DSB), after several bract and leaf buds were differentiated. The stolon apical meristem (SAM) degenerated as the new ramet leaf buds (RLB) and the new active axillary stolon buds (ASB) differentiated continually, after the differentiation of the first leaf. Using tandem mass tags (TMT) labeling method, totally 7,271 strawberry proteins were identified, and were used for further bioinformatics analysis in differentially expressed proteins (DEPs) between the groups of ASB and DSB, RLB and DSB, and RLB and ASB. Between ASB and DSB, the spliceosome DEPs, such as Ser/Arg-rich (SR) and heterogeneous nuclear ribonucleoprotein particle (hnRNP), showed the highest enrichment and high PPI connectivity. This indicated that the differences in DEPs (e.g., SF-3A subunit 2 isoform X1, hnRBP C1827.05c, and PK, cytosolic isozyme) at the transcriptional level may be causing the differences between the physiological statuses of ASB and DSB. As expected, the photosynthetic pre-form RLB mainly differentiated from ASB and DSB judging by the DEP enrichment of photosynthesis. However, there are still other specialized features of DEPs between RLB and DSB and between ASB and DSB. The DEPs relative to DNA duplication [e.g., minichromosome maintenance protein (MCM 2, 3, 4, 7)], provide a strong hint of functional gene duplication leading the bud heterogeneity between RLB and DSB. In addition, the top fold change in DEP LSH 10-like protein might be involved in the degeneration of SAM into RLBs. As for RLB/ASB, the phenylpropanoid biosynthesis pathway probably regulates the ramet axillary bud specialization, and further promotes the differentiation of xylem when ASB develops into a new stolon [e.g., cinnamyl alcohol dehydrogenase 1 (CAD1) and phenylalanine ammonia-lyase 1 (PAL1)].
Conclusions: The definite dormancy phase of DSB, and the biological pathways and gene networks that might be responsible for stolon buds heterogeneity were also revealed.

3 into a new stolon e.g., cinnamyl alcohol dehydrogenase 1 (CAD1) and phenylalanine ammonia-lyase 1 (PAL1). Conclusions: The definite dormancy phase of DSB, and the biological pathways and gene networks that might be responsible for stolon buds heterogeneity were also revealed.

Background
Strawberry is a sequenced member of Rosaceae, characterized by perennial evergreen, diminutive herbaceous plants amenable to genetic transformation. It has a relatively small genome (Fragaria vesca ~240 Mb, F. × ananassa ~700 Mb) and shares substantial sequence identity with the other economically important rosaceous plants [ 1 , 2 ]. It is widely used as a preference model plant by horticultural researchers. The cultivated strawberry, F. × ananassa, originated ~250 years ago, is among the youngest crop species, characterized by a highly nutritive large fruit, ease of vegetative propagation, and high economic value and is wildly and commercially produced by 76 countries in the world [ 3 , 4 ]. Because of the high heterozygosity of strawberry cultivars, it will be caused high variability and genetic segregation in the progeny seedlings. Thus, the sexual reproduction by seeds is not suitable for strawberry production. Instead, stolon (elongated stems) vegetative propagation was usually and efficiently used to produce clonal ramets from an aerial stolon (runner). However, the bud in the first node of cultivated strawberry (F. × ananassa) stolon usually remains in dormancy, and only the buds of the second node have the ability to form ramets[ 4 4 ]. Thus, new insights into the mechanisms underlying the stolon bud decision to produce either dormant or active buds for forming plantlets are crucial for improving strawberry productivity.
Morphologically, a strawberry stolon is a special lateral branch of crown, which is originated from the mother plant's axillary meristems with its subtending nodal ramet structure [ 5 , 6 ]. Anatomically, a strawberry stolon consists of a large proportion of thick cortex and a relatively small proportion of phloem, xylem, and pith for transporting water, ions, and photoassimilates between the mother plant and ramets [ 7 , 8 ]. The process of forming ramets in the second node of strawberry stolon can be summarized as follows: the second node degenerates into the first leaf of a future ramet and is wrapped up by its bracts. The adventitious roots are formed from the bases of the second nodes. Upon the completion of the rooting process, the lateral bud on the second node begins to elongate as the next new stolon. This newly formed stolon is not a continuous part of the mother plant, but is the lateral buds located on the first plantlet's axil. After that, the second or third ramets can be sequentially formed under favorable environmental conditions [ 4 ].
Former studies on strawberry stolon were mostly focused on the mechanism of its formation or on the dictates of the flowering-runnering decision. According to a recent study, the DELLA protein seems to be an important factor controlling runner formation during asexual reproduction in strawberry [ 9 ]. The gibberellic acid (GA) biosynthesis in the axillary meristem is essential for 5 inducing stolon differentiation. The possibility of the FveGA20ox4 gene regulating the runnering-flowering decision in strawberry has been revealed [ 6 ]. The studies on the differences between the first and second nodes of strawberry stolon have always focused on the internode differences instead of the differences between buds. For example, Fang et al. (2011) used two-dimensional gel electrophoresis for comparing the proteomic profiles of the strawberry stolon internodes I-1 and I-2 [ 10 ]. They found that the ubiquitin-proteasome and sugar-hormone pathways might be important during adventitious root formation at the second node of new clonal plants.
By quantifying the movement of resources and their allocation between mother plants and daughter ramets along Fragaria stolons with respect to hierarchy, the results showed that the stolon anatomy develops rapidly at the apical end, facilitating hierarchical ramet development, which is evident as a basipetal increase in hydraulic conductivities. The rapid development of transport tissue functionality enables young unrooted ramets to acquire water, as well as mineral ions disproportionally with respect to older ramets, in order to supply an expanding leaf area [ 8 ]. However, the mechanism by which the first node buds of cultivated strawberry usually remain in dormancy is not clear, and even when this dormancy is released under favorable environment, the first node buds have no ability to form ramets, but how they develop into new stolon branches is still unclear. In addition, the regulation of the mechanism underlying the dormancy of a stolon branch bud in the first node, but activity in the newly formed ramet located in the second node, is not clearly understood.
Thus, we first elucidated the developmental characterizations of the first and second node buds at different developmental stages, by using anatomical observation. Subsequently, we illustrated the definite dormancy phase of the first node bud, and the detailed 6 developmental processes of the ramet formation and ramet axillary bud elongation in the second node of strawberry stolon, using stereomicroscopy and scanning electron microscopy. Considering the proteomics utilization, especially the tremendous advantages in method of TMT in digging the DEPs among multiple groups of plant materials [11][12][13][14][15]. In the last part, we mainly laid the foundation for understanding the mechanisms of strawberry stolon phenotype and bud development at the protein level. By using phenotypic observation combined with proteomic networks with different types of strawberry stolon buds, the definite dormancy phase of DSB was identified, and the biological pathways and gene networks that might be responsible for heterogeneity among different stolon buds in strawberry were also revealed.

Stolon anatomy
The anatomical observation showed that the first node of the stolon (10 cm length) in cultivated strawberry (F. × ananassa Duch.) was extremely small and easily ignorable.
When the bract of the first node was peeled off, a very tiny bud (Fig. 1A) was observed.
On the contrary, for the observation of the second node, two different types of buds were observed underneath the bract-one was a plump bud (Fig. 1B), and the other was a leaf cluster mixed with several developing leaf buds (Fig. 1C). The slice observation showed that the buds in the first node stopped growing and stepped into dormancy at an early stage ( Fig. 2A). The dormancy of the bud located at the first node could be released only under favorable environmental conditions, and it continues to develop into a new stolon branch (Fig. 2B). The leaf buds, located inside of the bract of the second node, have a distinct trifoliolate structure (Fig. 2C), and the vascular bundles of the newly formed 7 stolon, which is laterally located on the leaf buds, are connected inward with the primary stolon (Fig. 2D). The structure of the strawberry stolon was observed by cross-sectional anatomy. The tissues from outside to inside of the stolon included epidermal hair and epidermis, thick cortex, cambium, phloem, xylem, and the pith, which are composed of a large number of parenchymatous cells (Fig. 2E). These two types of lateral buds on the first and second nodes of stolon were inwardly connected with the primary stolon in a same pattern (Figs. 2F-G). At the base of the second node, which is connected to the terminal strawberry stolon buds, there are numerous adventitious root primordia (Figs. 2H-I). Each adventitious root primordium originated from the cambium tissue, which consists of meristematic cells containing dense cytoplasm and swollen nuclei (Figs. 2E, 2I).

Stereoscopic and SEM observation of developing stolon buds
Dormancy bud in the first node--In order to acquire more details on the developmental characteristics of the first and the second node buds of the strawberry stolon, different developmental phases of the stolon buds were observed under the stereomicroscope and scanning electron microscope (SEM), respectively. We observed that in the early stage of stolon elongation (when the stolon length was 4-10 cm), the buds on the first node of the stolon grew with the respective development of the stolon at the early stage (Figs. 3 A-C).
For example, when stolon was 4 cm in length, a very tiny growing point located in the center of a trifoliolate bud could be seen after the outermost bract was peeled off from the first node (Fig. 3A). Continually, with the growth of stolon (when the stolon length was 6-7 cm), the top trifoliolate leaf bud on the first node developed further, and the growth point at the central base of the buds also grew (Figs. 3B-C). With the further elongation of the primary stolon, when the stolon length was 8-9 cm, the bud of the original trifoliolate gradually developed into a young trifoliolate bract and was densely covered with trichomes (Figs. 3D-E). When this young trifoliolate was peeled off sequentially, another tightly closed thin trifoliolate bud could be seen (Fig. 3F), and for protecting the underneath growing point. This is the landmark when the bud of the first node in the strawberry stolon ceases to develop and enters into dormancy; therefore, when we observed continually with the primary stolon elongated further, this thin trifoliolate bud structure showed no change. Our conclusion was further confirmed by the magnified observation of SEM, that is, the first trifoliolate bud under the bract of the first node continuously develops into a young trifoliate, with the development of the primary stolon at an early stage of stolon development or elongation (Figs. 3G, 3J). Similarly, when this 9 young trifoliolate leaf was peeled off, the structure of the thin, tightly closed trifoliolate bract was visible (Fig. 3J). With all this, the outermost new trifoliolate leaf bud and the inner growing point ceased to develop, and showed no further development as the primary stolon elongated continually, indicating its stepping into a state of dormancy.  Ramet buds in the second node--Unlike the axillary development of DSBs and ASBs, the developmental process of the strawberry ramet leaf bud (RLB) is relatively simple and rapid (Fig. 5). The apical region of a strawberry stolon contains multiple leaf bud primordia; when one leaf primordium gradually develops into a young trifoliolate, the next leaf primordium is initiated out as a visible developing trifoliolate (Figs. 5A, 5B), and then each leaf primordium develops into a young leaf, orderly, to formed a young leaf cluster (Figs. 5C-F). The developing activity stolon branch is located laterally to these leaf clusters (Figs. 5E, 5F). The SEM observation showed that the trifoliolate bract firstly grown out is located at the top of each trifoliolate primordium to protect the inner part (Fig. 5G).
Each trifoliolate bract was tightly connected with each other in a complementary manner   Table S1, S2). A 1.2-fold-change cut-off with P-value<0.05 was used to indicate significant changes in the abundance of DEPs among different strawberry stolon buds (supplemental Fig. S2). By analyzing the quality control data, we found that the TMT results that were achieved by using high-quality Q Exactive mass spectrometer was reliable. The accuracy and high resolution in this experiment can maintain good quality deviation in the process of data acquisition, and finally obtain the high-quality spectrograms of MS1 and MS2. The quality deviation of all identified peptides was mainly within 10 ppm (supplemental Fig. S3), indicating that the identification results were accurate and reliable. When the rigid analyzing tool of MASCOT (FDR<0.01) was used for judging each MS2 spectrogram, we obtained an ideal score with a median of 34.06, and more than 86.21% peptides scored higher than 20 (supplemental Fig. S4). The protein ratio (approximately 1.0) distribution of the three groups (ASB/DSB, RLB/DSB, and RLB/ASB) are shown in supplemental Fig. S5.
Features of identified proteins--The distribution of unique peptides defining each protein is shown in Figure 6B, with over 61% of them, including at least two unique peptides (supplemental Table S1). The average molecular mass of the identified gene products ranged from 10 to 70 kDa (Fig. 6C). The distribution of t (PI) of the identified proteins was mainly in the area of 5.0-10.0, with most PIs ranging from 6.0 to 7.0 (supplemental Fig.   S6). Comparisons between the DSB, ASB, and RLB groups led to the identification of  Table S3). The variation of the three biological replicates of each group (DSB, ASB, and RLB) was calculated according to their quantitative data, with most proteins exhibiting less than 20% variation (supplemental Fig. S7), indicating the high quality and repeatability of data. Common DEPs between groups--According to the fold changes in each group (supplemental Table S3), we selected out top 10 up-and down-regulated DEPs between groups for further analyzing (Table 1). By further selecting the common DEPs in these 10 up-or down-regulated proteins (supplemental Table S4), we found that all five common proteins between the groups of ASB/DSB and RLB/DSB were down-regulated DEPs. These  Table S5). Table 1

. Top 10 of up-and down-regulated differentially expressed proteins between groups
Additional large-scale analysis of DEPs between groups of co-up and co-down regulation was also carried out, as shown in the Venn diagram of Figure 7 (detailed information in supplemental Table S6, S7, S8). We found that no DEP showed co-up regulation; only one DEP of GDSL esterase/lipase showed co-down regulated mode among all three groups (Fig.   7A, 7B). When the co-up plus co-down regulated proteins were counted at the same time, there are 45 common-DEPs in all three groups (Fig. 7C). Among all statistics, one group of data showed special performance, that is, ASB/DSB and RLB/ASB have almost no co-up or co-down proteins (Fig. 7A, 7B) separately, but when we counted all total co-up plus codown DEPs, 407 common DEPs appeared in both ASB/DSB and RLB/ASB groups, simultaneously (Fig. 7C). These results indicated that each of the 407 DEPs shared common down-regulated or up-regulated expression patterns in the groups of DSB and RLB when compared with ASB, further suggesting that all these proteins have dual functions in stimulating the RLB formation while promoting the stepping of DSBs into dormancy, but have an antagonistic effect on the ASB development in strawberry stolon, simultaneously. Thus, exploring the regulatory mechanisms of these 407 DEPs is of great significance to clarify the dormancy of the first node DSBs and the formation of the second node RLBs.

Bioinformatics analysis
All DEPs detected by MS were subjected to a bioinformatics analysis for further classification.
Cluster Analysis--The hierarchical clustering results were expressed as a respective heat map (Fig. 8). The X-and Y-coordinates represented sample and differentially expressed proteins, respectively. As determined by a horizontal comparison, the samples could be classified into three categories: DSB, ASB, and RLB. Such a classification was associated with high accuracy, suggesting that the selected DEPs could effectively distinguish between samples. Furthermore, a vertical comparison indicated that the selected proteins could be classified into two categories with opposite directional variation, which displayed the expression patterns of DEPs in three groups (supplemental Fig. S8), demonstrating the rationality of the selected DEPs. The cluster analysis, thus, supported that the DEPs screened out in this experiment were reasonable and accurate.  Table S9-11). The DEPs were individually analyzed against the Gene Ontology (GO) database using three sets of ontologies: biological process, molecular function, and cellular component (Fig. 9). The analysis showed that among all three groups of ASB/DSB, RLB/DSB, and RLB/ASB, numerous DEPs could be classified in the same GO category (supplemental Table S9-11).
The top two commonbiological processcategories were metabolic process (over 35%) and cellular process (over 25%). The top four common molecular function categories were catalytic activity (over 35%), binding (over 25%), transporter activity, and structural molecule activity. The top four common cellular component categories were cell (over 25%), cell part (over 25%), organelle (over 15%), and membrane proteins (over 15%). A small number of other DEPs existed in cellular component categories, including membrane part, organelle part, and macromolecular complex, with the ratio of approximately 10%.
For further exhibition of top 20 enriched GO terms (supplemental Fig. S9), we know detailed information of functional proteins in biological process ( BP) of ASB/DSB were oxidation-reduction process (~100 DEPs) and regulation of RNA metabolic process (supplemental Fig. S9A), in RLB/DSB, they were DNA metabolic process and photosynthesis, with the same number of DEPs (14), as well as DNA conformation change and replication (supplemental Fig. S9B), whereas for RLB/ASB (supplemental Fig. S9C), there were small numbers of DEPs, which were classified into the secondary metabolic process, one-carbon metabolic process, secondary metabolite biosynthetic process, and phenylpropanoid metabolic process.
Molecular function ( MF) analysis showed that the GO terms in ASB/DSB were catalytic activity (~450 DEPs) and oxidoreductase activity (over 100 DEPs), as well as DNA binding (supplemental Fig. S9A). In RLB/DSB, the MF proteins were DNA binding and helicase activity (supplemental Fig. S9B). In the RLB/ASB MF GO terms, a large numbers DEPs belonged to catalytic activity (~250) and oxidoreductase activity (~60) and protein dimerization activity (supplemental Fig. S9C).
The cellular component ( CC) terms in ASB/DSB were thylakoid and thylakoid part and chromatin. Combining the GO terms identified in the MF and BP analysis above, we found that the differences existed mainly at the RNA level (supplemental Fig. S9A). Between the groups of RLB and DSB, the GO terms were thylakoid (22), thylakoid part (16), plastid thylakoid (14), chloroplast thylakoid (14), photosynthetic membrane (14), and thylakoid membrane (12) (supplemental Fig. S9B). This suggested that their differences between RLB and DSB mainly occurred in their capacity of photosynthesis. In RLB/ASB, all DEPs of CC were functional compartment of chromosome-or DNA-relative proteins, and showed a coincidence trend of low number (supplemental Fig. S9C). we analyzed all DEPs among groups (Fig. 10). The results indicated that the spliceosome (43 DEPs, as shown below) and ribosome (29) had high enrichment between ASB and DSB ( Fig. 10A, supplemental Table S12). This suggested that the differences in transcription or translation are thefundamental reason for the difference between ASB and DSB. As for RLB/DSB (Fig. 10B, supplemental Table S13), the photosynthesis (13) pathway is the highest enrichment of DEPs. This is an additional proof of the fact that the main RLB function is photosynthesis for the next clonal generation of ramets. In addition, two highly enriched pathways of DNA replication and spliceosome still existed, indicating that both genetic and transcriptional level differences existed between RLB and DSB. The participation of DEPs in the phenylpropanoid biosynthesis pathway (18), as well as in carbon metabolism relative pathways, such as starch and sucrose metabolism (15), amino sugar and nucleotide sugar metabolism (14), and glycolysis/gluconeogenesis (11), showed high enrichment between groups RLB and ASB (Fig. 10C, supplemental Table S14). This indicated that phenylpropaniod biosynthesis worked for differentiation between RLBs and ASBs in the second node of strawberry stolon, especially for the formation of vessels during the ASB developmental processes, as discussed below.  Table S15), which was expressed as nodes and links, contributed to extracting effective protein information from various points of view and obtaining comprehensive information that could not be obtained through the analysis of only a single protein (Fig. 11).
According to the results, 20 high-connectivity degree DEPs, with a degree value of more than 30, were selected out between groups ASB and DSB (Table 2). Six and 11 DEPs, with connectivity degrees higher than 10, were selected from the RLB/DSB and RLB/ASB groups, respectively ( Table 2). The results of this part are highly consistent with those of KEGG, which indicated that the difference between ASB and DSB was mainly due to the difference at the transcriptional level, while the difference between RLB and DSB was mainly due to the difference at the genetic level. For further showing the direct proteinprotein relationship, we selected four typical DEPs of NADH-GOGAT & GDH, PK, MCM 2-4 and 6-7 as the PPI core ( Fig. 11A-C). In additional, we drew the PPIs in the phenylpropanoid biosynthesis pathway (Fig. 11D) to further clarify the key protein-protein interactions.  were selected for PRM analysis (Fig. 12). The screening criteria were formulated based on the following two principles: 1) proteins with potential biological functions and peptide fragments greater than 1, as identified by LC-MS/MS; and 2) proteins that were specifically expressed in one group of buds when compared with the other two groups of buds and have not been reported yet.
The results of the LC-PRM/MS analysis performed on 12 peptide fragments of three target proteins from three groups of strawberry samples showed that the quantitative information of target peptide fragments could be obtained in all nine samples.
Subsequently, the relative quantitative analysis was carried out on target peptide fragments and proteins through the incorporation of heavy isotope-labeled peptide fragments. The results indicated that of the three target proteins, the expression quantities of PK and PAL1 in the ASB group were markedly upregulated compared with the DSB and RLB groups; whereas the expression quantity of MCM2 in the RLB group was significantly up-regulated compared with that in the DSB and ASB groups, additionally verifying the facticity and accuracy of the TMT method in this study. Data are means and standard errors, of 3 groups of each type bud, and the experiment was repeated 3 times. Different letters in the same index means the significant difference among buds, separately (P<0.05). Bars represent the standard deviation (n=3).

Discussion
Heterogeneity among the buds of different stolon nodes Stolon is an asexual reproductive organ of strawberry. It is important to study the development process of the buds at different nodes for strawberry production. The stolon of the octoploid cultivated strawberry (F. × ananassa Duch.) consists of two nodes-the first node usually remains in dormancy, and the second node has the ability to form the ramet ( Fig. 1) (4). A new stolon is usually originated from the axillary bud inside the first leaf of the ramet, where the bud is mostly conducive to absorbing water and nutrients, and is, therefore, most likely to develop as a new stolon branch [ 16 ]. The colonizing behavior and functional morphology of stolons (Fig. 2) indicate that ramet survival, prior to rooting, is achieved through the plasticity of intrastolon ramet competition for resources such as water, ions, and photoassimilates [ 7 ].
By observing the buds on the nodes, we found that the buds on the first node developed into a certain stage and then ceased development at the early phase of the primary stolon elongation, indicating their entry into dormancy (Fig. 2, 3). Furthermore, under favorable conditions, the buds of the first node can sprout out as another stolon branches, but these new stolon branches are usually much smaller and thinner than the primary stolon ( Fig.   22   2B). If the tip of the stolon is removed, the first node can develop into a ramet rather than a stolon branch [ 4 ], indicating that the bud in the first node has the binary functions of forming a new stolon branch or an independent ramet simultaneously, given that it might be an undifferentiated bud or has the ability to dedifferentiate again. However, under normal conditions, the factors that regulate the ceasing of the first node bud development and its entry into dormancy are still unknown. We can elaborate the possible factors by comparing the former studies on the key regulators of axillary bud growth and dormancy.
The shoot branching process generally involves two developmental stages: the formation of axillary meristems in the leaf axils and the growth of axillary buds [ 17 ]. In many plant species, the growth of axillary meristems is inhibited by the primary shoot or primary inflorescence [ 18 ]. This phenomenon is generally known as apical dominance. The plant hormones auxin and cytokinin are thought to have a major role in controlling this process [ 19 ]. Auxin has an inhibitory effect on the growth of axillary buds, whereas cytokinin promotes axillary bud outgrowth. The mechanisms of axillary bud outgrowth depend on the ratio of these two hormones rather than the absolute levels of either hormone. In other plants, the axillary meristems might initiate a few leaves and then become developmentally arrested or dormant because the terminal bud inhibits the growth of axillary buds to grow predominantly [ 17 ]. As for strawberry, we suggest that the first node bud development on stolon belong to this type of axillary growth, and we observed that axillary meristems initiated a few trifoliolate bracts, as shown in Figure 2, followed by ceased development and entry into dormancy. Consistent with the previous studies, the arrested development of axillary buds in the first node of strawberry and their stepping into dormancy might be 23 comprehensively caused by environmental factors and a feedback to apical dominance.
Thus, according to our observation, this dormancy could be released under suitable environment and growth could be resumes to develop into a new branch of stolon (Fig.   2B). It is possible that a set of genes or proteins that controls the outgrowth or dormancy of axillary buds acts at different phases of the bud developmental processes. This type of molecular study might provide the basis for understanding the regulation of dormant or outgrowing axillary buds in strawberry stolon nodes.

Proteomics for analyzing different stolon buds
Comparative proteomics is a useful approach for identifying functional proteins in illustrating the developmental regulation mechanism of plants[ [20][21][22][23]. Recently, with the tremendous release of the plant reference genome data, more and more comparative proteomics approaches have been applied for studying bud heterogeneity in crops [11][12][13][14]. Previous studies compared the proteomes of the first and second nodes of the strawberry stolon to elucidate the internode differences, and found that the DEPs were mostly related to photosynthesis [ 10 ]. This study was useful for understanding the heterogeneity of stolon buds in strawberry, but it still requires further investigation. On one hand, the second node of strawberry stolon contains not only a single type of bud, but two types of buds, RLB and ASB; thus, it is necessary to anatomically separate the second node into two different types of buds before analyzing them. As mentioned above, DSB, ASB and RLB on the same stolon were separated and categorized into three types of bud groups, with each group having three biological repeats, and each repeat containing 200 buds, before performing the TMT analysis. Approximately 540 fresh stolons, which were in the same growth phase, were anatomically dissected and their DSBs, ASBs, and RLBs were collected as independent samples for experiments in order to fully shield the differences among the experimental individuals and ensure the reliability of the experimental data. In addition, the TMT method is more accurate than the traditional two-dimensional gel electrophoresis in isolating DEPs, and can successfully identify DEPs with low expression levels among groups [ 15 ]. Therefore, we used the TMT method to explore the primary causes leading to the differences between DSB and ASB, which existed as the stolon axillary shoot buds, but under quite different physiological conditions. In addition, we firstly determined the factors regulating bud differentiation between RLB and ASB, which are commonly located on the second node of a strawberry stolon. It further investigation to elucidate the reason underlying the development of an axillary ASB from a newly formed ramet, which mainly originates from an RLB and then further develops into an elongated stolon.

TMT revealed the heterogeneity of stolon buds in strawberry
The proteome has an important characteristic difference with the genome, that is, proteins have a direct influence on each other [ 24 ]. The realization of the function of a protein usually depends on its interaction with other proteins implying that no independent functional protein exists [25][26][27]. Therefore, through comprehensive analysis and evaluation of GO annotation (BP, MF, CC), enrichment in KEGG, and connective degree in PPI, we can predict the core functional DEPs involved in the key metabolic pathways [ 26 , 28 ].
Between ASB and DSB--According to the comprehensive analysis between GO and KEGG, we know that the difference at the transcriptional level might lead to the differences in their physiological statuses. On combining the PPI analysis, the DEP of splicing factor 3A 25 subunit 2 isoform X1 showed a higher connectivity degree value of 42. As previous studies reported that alternative splicing has a wide influence on the evolution of the complex networks of the regulation of gene expression and variation in contribution to the adaptation of plants to their environment and, therefore, will impact the strategies for improving plant and crop phenotypes, such as entry into dormancy under stress conditions [ 29 , 30 ]. Splicing factor (SF), as a positive contributor in the process of alternative splicing, recruits splicing-related proteins and confirms the splicing position and spliceosome assembly, and then participates in the morphological determination of plant organs [ 31 ]. SF mainly contained two families of proteins-Ser/Arg-rich (SR) and heterogeneous nuclear ribonucleoprotein particle (hnRNP). In our PPI analysis, of the 19 DEPs, which had high connectivity degree values of more than 30, four (21%) were SRs and six (32%) were hnRNPs. Thus, we suggest that these high connectivity degree SRs and hnRNPs might act as crucial factors in regulating the morphological determination of the heterogeneity of the stolon axillary ASBs and DSBs (Fig. 13) Similarly, we should also pay attention to the other two DEPs of pyruvate kinase, cytosolic isozyme (degree 33) and uncharacterized RNA-binding protein C1827.05c (degree 31).
Pyruvate kinase (PK) has been well studied in modulating bud dormancy or bud break in pomology, and the activity of PK has been found to be lower in dormant buds than in nondormant buds and peaked in the green tip stage just before the start of rapid expansion and declined thereafter[ 26 32-34 ]. In our study, PK, cytosolic isozyme degree 33 is also another highly connective DEP with high expression quantity in ASB, but is expressed at relatively low levels in DSB and RLB (Fig. 12), suggesting that this pyruvate kinase, cytosolic isozyme functions mainly in ASB than in DSB and RLB (Fig. 12). This was consistent to a previous study, but elucidation of the detailed functional mechanism still needs further investigation.
In eukaryotes, RNA-binding proteins (RBPs) play crucial roles in all aspects of posttranscriptional gene regulation. They regulate diverse developmental processes by modulating the expression of specific transcripts. Clearly, they function by regulating pre-mRNA splicing, polyadenylation, RNA stability, and RNA export, as well as by influencing chromatin modification [ 29 ]. Uncharacterized RNA-binding protein C1827.05c (degree31), as a special DEP with relatively high fold change between groups ASB and DSB, might co-function with the splicing factors and ribosomal proteins (Fig. 13, supplemental Table S15).
Between RLB/DSB--The differences between RLB and DSB mainly focus on the DEPs involved in photosynthesis ( Fig. 9-10); this might be caused by the difference at gene duplication level. We identified four DNA replication licensing factors minichromosome maintenance (MCM) 2, 3, 4, and 7 from a total of six high-connectivity degree (>10) DEPs (supplemental Table S15 36 ]. Here, we suppose that MCM 2, 3, 4, and 7 might upregulate the expression of photosynthetic genes and indirectly regulate photosynthesis substance allocation and transportation by modulating DNA replication or endoreduplication. This was consistent with the previous studies, which reported that the parenchyma cells that store starch, sugar, and other substances in the fruits or seeds of plants reproduce through DNA replication or endoreduplication [ 37 ]. In order to verify the MCM expression mode in strawberry buds, we selected MCM2 as an identified protein and found that MCM2 showed a significantly high expression in RLB when compared with DSB and ASB (Fig. 12). Our hypothesis might also partially elucidate the findings of Atkinson et al. (2012), who reported that the hydraulic conductivity and polar auxin transport (PAT) pathway could determine hierarchical resource partitioning and ramet formation in Fragaria stolons (Fig. 13) The DEP with the highest fold change in expression between RLB and DSB was LIGHT-DEPENDENT SHORT HYPOCOTYLS (LSH) 10-like, with a fold change of 3.24 (Table I, supplemental Table S15). The LSH protein is an important functional regulator in modulating the plant shoot initiation process and could be used as a shoot marker in presaging the sites of shoot formation [ 38 , 39 ]. LSH can be early expressed at the very early stage during zygotic embryogenesis in Arabidopsis [ 38 ]. As for strawberry, no studies have been conducted on short crown formation.
Previous studies showed that differentiated organs can be converted to the other type of organs by various methods; for example, incubation in cytokinin-rich shoot induction medium converts the premature roots into shoots, particularly in those regions where the cytokinin receptor genes are up-regulated[ 28 40 ]. Flower-meristem-identity gene LEAFY is sufficient to determine the floral fate in lateral shoot meristems of both Arabidopsis and the heterologous species aspen, with the consequence that flower development is induced precociously [ 41 ]. In our study, the LSH 10-like protein was uniquely identified from among a total of 7,271 identified proteins by using high-throughput proteomics analysis between RLB/DSB. We speculate that the LSH 10-like protein should be involved in the degeneration of stolon apical meristem into RLBs (Fig. 13). Further studies are required to investigate on how LSH 10-like regulates the formation of a shorted crown of ramet and how ASB could be initiated from the position of a leaf axil in a newly formed ramet, and why the new-born secondary stolon (ASB formed) keep in continually running on the ground, instead of growing upward to the air.
Between RLB and ASB--The differences between RLB and ASB in the second node are partially week compared to those between RLB and DSB. For example, glutamate synthase 1 [NADH], chloroplastic isoform X1 (connectivity degree 15, fold change 0.74, shortened as NADH-GOGAT 1) and glutamate dehydrogenase 1 (connectivity degree 12, fold change 2.05, shortened as GDH 1). NADH-GOGAT and GDH are important enzymes that participate in nitrogen metabolism by synthesizing glutamate [ 42 ]. The catalytic function of GDH was directly and more energy-efficient when compared to that of GOGAT [ 43 , 44 ], and GDH mainly exists in non-photosynthetic tissues, such as root and early development cotyledons, of plants [ 45 ]. Unanimously, the fold change of NADH-GOGAT was higher than that of GDH between RLB and ASB.
Another group of special-feature DEPs could be found after a comprehensive analysis 29 between RLBs and ASBs; they are the proteins involved in phenylpropanoid biosynthesis (Ko00940). According to previous reports, the multiple roles of phenylpropanoid biosynthesis in plant development are mainly focused on providing anthocyanins for pigmentation, flavonoids, such as flavones, for protection against UV photodamage, various flavonoid and isoflavonoid inducers of Rhizobium nodulation genes, polymeric lignin for structural support and assorted antimicrobial phytoalexins [ 46 ]. In particular, it plays an important role in the differentiation and development of lignin [ 47 ]. After comprehensively analyzing the KEGG pathways and PPIs, as well as fold change values among different groups, 18 DEPs highly related to the phenylpropanoid biosynthesis were selected out for further analysis (Fig. 11, supplemental Table S16).
Among them, 15 DEPs were up-regulated in the group of ASB/DSB, whereas downregulated in the RLB/ASB group simultaneously, and only three DEPs showed an opposite trend. In addition, almost all DEPs in the RLB/DSB group showed no significant difference in expression (Fold change 0.9-1.2). This means that most DEPs in the phenylpropanoid biosynthesis positively function in the formation process of ASB (Fig. 13). The DEPs of cinnamyl alcohol dehydrogenase 1 (CAD1) and phenylalanine ammonia-lyase 1 (PAL1) are only expressed in the RLB/ASB phenylpropanoid biosynthesis pathway. It has been reported that CAD1 and PAL1 are closely related to lignin synthesis [ 48 , 49 ]. In addition, by determining the expression quantity of PAL1, we further confirmed that the target protein PAL1 was highly expressed in ASB, but showed low-level expression in DSB and RLB (Fig. 12). Based on that, we suggest that both of these two proteins might play important roles in the axillary bud specialization of a new ramet leaf into ASB. We also speculated that they might play important roles in xylem differentiation 30 or vascular formation when ASB developed into a new stolon (Fig. 13).

Conclusions
By combining the anatomical observation with the phenotypic observation and using proteomic networks with different types of strawberry stolon buds, we identified the definite dormancy phase of DSB and compared to the developmental differences among DSB, ASB, and RLB, as well as identified numerous protein signatures that translated to biological pathways and gene networks that might underlie the real reason of heterogeneity among different stolon buds in strawberry. The possible mechanisms for differentially expressed proteins in regulating the heterogeneity of stolon buds in strawberry were achieved (Fig. 13). The current study provides further information for understanding the heterogeneity of stolon buds in strawberry, as well as other fruit trees. of buds in the first node (dormancy shoot bud) and the second node (including activity shoot bud and stolon apices) of a strawberry stolon were selected as materials (Fig. 1).

31
The collected strawberry stolons had a length and diameter of 10 cm and 3 mm, respectively. Each type of bud sample was set as three replicates, with each replicate containing approximately 200 mg, which was collected from 60 buds. More than 540 fresh uniform stolons should be prepared for these three types of stolon buds as TMT samples.
All material samples were collected as the youngest stolon buds (Figs. 1 A-C), immediately frozen in liquid nitrogen, and then stored at −80 °C for protein extraction. All three types of buds were used in triplicate samples for proteomics.

Phenotypic observation
In order to observe the morphological differences between the buds of the first and second nodes of the strawberry stolon, the bract of the young buds on the nodes should be peeled off using anatomical needles under the stereo microscope (Nikon SMZ 1500). Thereafter, all redundant sequences were removed from this combined dataset.

Gene Ontology and KEGG pathway annotation
The process of GO annotation by Blast2GO[ 53 ] can be roughly divided into four steps: sequence alignment (BLAST), GO entry extraction (mapping), GO annotation (annotation), and annotation augmentation (annotation). Firstly, the National Center for Biotechnology Information (NCBI) basic local alignment search tool BLAST+ (ncbi-blast-2.2.28+-win32.exe) was used to align the target protein set with the appropriate protein sequence database, and the top 10 alignment sequences satisfying E-value less than 1E-3 were retained for subsequent analysis.
Secondly, the mapping process was carried out by using the Blast2GO Command Line to select the relative GO items among the target protein set and qualified items in the first step (Data version: go_201504.obo; download address: www.geneontology.org). Thirdly, in the GO annotation process, the Blast2GO Command Line takes into account the similarity of the target protein sequences and alignment sequences and source reliability of the GO item entries, and evaluates the structure of the GO graph. Subsequently, the GO item information was annotated to the target protein, which was selected in the mapping process. Fourthly, after annotation, in order to further improve the annotation efficiency, we searched the European Bioinformatics Institute (EBI) database to identify the target proteins by matching conserved motifs using InterProScan [ 54 ]. Thereafter, the motif-related functional information was annotated to the target protein, and then ANNEX was run to further supplement the annotation information and build the connections among different kinds of GO items for improving the accuracy of annotations. In summary, the GO project described the roles of proteins in three functional categories: biological process (BP), cellular component (CC), and molecular function (MF).
The KEGG pathway annotation was used to search and compare genes in the database of KEGG GENES using the KAAS (KEGG Automatic Annotation Server) software [ 55 ], followed by the preliminary KO classification of target protein sequences.
Thereafter, the information on the target proteins involved in the metabolic pathways was automatically obtained according to the KO classification. Finally, the target protein set 37 was comprehensively analyzed using GO ontology or KEGG pathway annotation. To evaluate the protein richness of the GO ontology or KEGG pathway, the Fisher's exact test was used to compare the distribution of each GO classification or KEGG pathway in the target protein set, followed by the calculation of the significance level.

Protein clustering
In thermographic clustering analysis, the quantitative information of the target protein set was normalized to ±1 interval. Secondly, the Cluster 3.0 software

Parallel reaction monitoring (PRM) validation
To further check the levels of protein expression determined through TMT analysis, additional quantification was applied through LC-PRM MS analysis [ 56 ]. Briefly, the TMT protocol was used for peptide preparation. The stable isotope AQUA peptide was spiked in each sample and used as a standard internal reference. The tryptic peptides were loaded on stage tips of C18 for desalting prior to reversed-phase chromatography on one of the nLC-1200 easy systems (Thermo Scientific). Subsequently, 1-h liquid chromatography gradients were performed with 5-35% acetonitrile for 45 min.
The Q Exactive Plus MS was applied for PRM analysis. The optimized methods for measuring the energy of collision, state of charge, and retention time of the most crucial peptides were determined by the experiments involving unique peptides with high intensities, and, therefore, each targeted protein could be handled properly. The analysis of raw data was realized via Skyline (MacCoss Lab, University of Washington) [ 57 ], wherein the intensity of signal produced by a certain peptide sequence could be quantified with respect to each sample and referenced to standards via normalization for each protein with important denatured protein samples.

Statistical analysis of data
Data were analyzed using Excel and SPSS by ANOVA followed by Tukey's significant difference test at p≤0.05. All data had three biological repeats.