Domestication Syndrome Is Investigated by Proteomic Analysis between Cultivated Cassava (Manihot esculenta Crantz) and Its Wild Relatives

Cassava (Manihot esculenta Crantz) wild relatives remain a largely untapped potential for genetic improvement. However, the domestication syndrome phenomena from wild species to cultivated cassava remain poorly understood. The analysis of leaf anatomy and photosynthetic activity showed significantly different between cassava cultivars SC205, SC8 and wild relative M. esculenta ssp. Flabellifolia (W14). The dry matter, starch and amylose contents in the storage roots of cassava cultivars were significantly more than that in wild species. In order to further reveal the differences in photosynthesis and starch accumulation of cultivars and wild species, the globally differential proteins between cassava SC205, SC8 and W14 were analyzed using 2-DE in combination with MALDI-TOF tandem mass spectrometry. A total of 175 and 304 proteins in leaves and storage roots were identified, respectively. Of these, 122 and 127 common proteins in leaves and storage roots were detected in SC205, SC8 and W14, respectively. There were 11, 2 and 2 unique proteins in leaves, as well as 58, 9 and 12 unique proteins in storage roots for W14, SC205 and SC8, respectively, indicating proteomic changes in leaves and storage roots between cultivated cassava and its wild relatives. These proteins and their differential regulation across plants of contrasting leaf morphology, leaf anatomy pattern and photosynthetic related parameters and starch content could contribute to the footprinting of cassava domestication syndrome. We conclude that these global protein data would be of great value to detect the key gene groups related to cassava selection in the domestication syndrome phenomena.


Introduction
Although cassava genome sequence is an information resource, the value of the genome is its annotation, which bridges the gap from the sequence to the biology of cassava. Cassava genome is a multi-step process, including three categories: nucleotide-level, protein-level and processlevel annotation [17]. Despite the recently significant advances on the nucleotide-level annotation, very little is known about the cassava global protein-level annotation, particularly focusing on wild species existing in the world.
Proteomics is a useful tool to compile a definitive catalogue of cassava global proteins, to name them and to assign them putative functions, providing a global protein-level annotation for cassava whole genome. It is applied to all protein expression in a particular organelle or tissue or in response to a particular stress. Proteomic analysis has revealed which proteins are responsible for cell differentiation in Arabidopsis under salt and osmotic stress and drought responsiveness in maritime pine, maize and wild watermelon [18]. In cassava, proteomics was employed to compare proteome patterns between fibrous and storage roots [19] and also used to describe the proteome characteristics of somatic embryos, plantlets and storage roots in cassava SC8 [8]. Owiti et al. (2011) investigated the molecular changes during physiological deterioration of cassava root after harvest using isobaric tags for relative and absolute quantification of proteins in soluble and non-soluble fractions prepared during a 96 h post-harvest time course, establishing a comprehensive proteome map of the cassava root and identified quantitatively regulated proteins [7]. Recently, An et al. (2014) employed a proteomic method to detect the changes of cassava polyploidy genotypes at proteome levels, and provided an insight into understanding the protein regulation mechanism of cassava polyploidy genotype [6]. However, the proteome diversity between cassava cultivars and its wild relatives is poorly understood.
The purpose of the present study was to compare the differences of anatomy, physiology and proteomes in leaves and storage roots between cassava cultivars and wild relative W14. All identified proteins were classified into cohesive groups based on their biochemical functions and indicated proteome diversity. The biological network of protein-protein interaction was set up to describe differential proteins regulations in the photosynthesis and starch accumulation. The proteome differences were supported by cassava anatomic and physiological data. This study will provide important clues on the improvement of cassava breeding through exploring the key gene groups related to the domestication syndrome phenomena.

Plant materials
Two cassava cultivars, M. esculenta cv. SC205 and SC8, and cassava's closest wild relative M. esculenta ssp. Flabellifolia (W14) were selected for the present study. SC205 and SC8 were released from Tropical Crops Genetic Resources Institute (TCGRI), CATAS. W14 originated in Brazil and is currently planted in Cassava Germplasm Bank (CGB), TCGRI, CATAS. The stem cuttings of SC205, SC8 and W14 were grown in the field at CGB on February 2012. The functional leaves of SC205, SC8 and W14 grown for three months and storage roots grown for ten months were taken. Three replicates consisting of three leaf/root slices each were sampled and immediately used for microscopy observation, and also frozen in liquid nitrogen for protein extraction.
Morphological observation under light microscopy and scanning electron microscopy extracted from storage roots between SC205, SC8 and W14, were observed under scanning electron microscopy (SEM). The samples (dried starch powder) were mounted on SEM stubs with double-sided adhesive tape and coated with gold. Scanning electron micrographs were taken using an S-3400N scanning microscope (Hitachi) in Jiangsu University, China [20].

Photosynthetic activity measurement by imaging pulse amplitude modulation
The Maxi-version of the Imaging Pulse Amplitude Modulation (Imaging PAM) and the software Imaging WIN version 2.39 (both Heinz Walz GmbH, Effeltrich, Germany) were used to determine the photosynthetic activities of W14, SC205 and SC8 according to An et al. (2014). For each genotype, three individual plants were used and the results were averaged.

Determination of dry matter content, starch content and starch component
Dry matter content (DMC), starch content, and starch component including Amylose contents (AC) and amylopectin contents (APC) were measured as previously described by Gu et al. (2013) [21].

Protein extraction, 2-DE separation and identification
Proteins from functional leaves and storage roots of SC205, SC8 and W14 were extracted with phenol extraction according to Chen et al. (2009) [18]. Protein separation was conducted following the previous described in An et al. (2014) [6]. Three independent biological replications were carried out. Gel matching for protein quantification was performed using an Image Scanner III (GE healthcare) and Delta 2D (Decodon GmbH, Greifswald, Germany) software, and spot pairs were confirmed visually. The significance of differences was determined by Scheffe's test at P <0.05. The abundance of each protein spot was estimated by the percentage volume (%Vol). Tryptic in-gel digestion and Protein identification were performed by the methods reported in An et al. (2014) [6].

Western blot analyses
Proteins of functional leaves and storage roots of three cassava genotypes were extracted [8]. Western blotting was performed according to the method previously reported [6]. Proteins detected by immuno-staining with anti-Rubisco-polyclonal antibody (AS07218), anti-OEC antibody (AS 05092) and anti-D1 antibody (AS05084) from Agrisera for leaves, anti-GBSS1 antibody and anti-linamarase antibody, produced by GenScript, anti-beta-amylase antibody (AS09379) from Agrisera for storage roots. Western blots were developed according to the method of NBT/BCIP from Roche (11681451001).

Results
Plant morphology, leaf anatomy and photosynthetic capacity between cultivars and wild relatives Fig 1 shows leaf transverse sections of two cassava cultivars SC205 and SC8 versus the wild relative W14. The measurements of all leaf strata, including midrib, cuticle, epidermal and mesophyll layers, revealed significant differences between the cultivars and the wild relative. The most noticeable difference between the species existed in the midrib (Fig 1a1, 1b1 and  1c1). Amplification of midrib showed that SC8 had a small area of primary xylem (PX), primary phloem (PP) and collenchyma (CC) less than those of SC205 and wild relative (Fig 1a2,  1b2 and 1c2). Compared to the wild relative, the cassava cultivars had a more distinctive bundle sheath with small and thin-walled cells. In the cultivars, the vascular bundles occur below the layers of elongated palisade cells. While part of the bundle sheath was in contact with palisade cells, there was not a uniform layer of mesophyll cells around the bundle sheath as it found in C 4 plants (Fig 1a3, 1b3 and 1c3).
In the leaves of 3 month-old cassava plants SC205, SC8 and W14, the photosynthetic abilities of SC205 and SC8 were greater than that of W14 (Fig 2 and Table 1), indicating variations in the efficiency of excitation energy capture by open Fv/Fm, FPSII and NPQ/4. These data imply that an increase in maximal and effective quantum yield and a concomitant increase in NPQ/4 processes are sensitive markers for cassava genotypes.

Analysis of starch properties from storage roots
The highest dry matter content (DMC) of storage root was cultivar SC205 with a mean of 77.08%, reversely; the lowest DMC was wild relatives W14 with an average of 57.80% (Table 2). Cultivars SC205 and SC8 with starch contents (28.86% and 29.74%, respectively) were at least seven times more than wild relative W14 with 3.75%. Amylose contents (ACs) in SC205 and SC8 (18.81% and 19.15%, respective) were significantly greater than that of W14 (6.72%), whereas amylopectin contents (APCs) in cultivars were significantly less than wild relative ( Table 2). The results regarding starch staining with KI also supported the described above (Fig  3a1-3c1 and 3a2-3c2). No significant differences of DMC, starch content, AC and APC were observed in SC205 and SC8, however, there was a significantly difference between the cultivars and the wild relative. Granule size and shape varied largely between cultivars and wild relative (Fig 3a3-3c3). No significant differences were observed in the starch granules in SC205 and SC8. The transverse (Fig 3a4-3c4) and longitude (Fig 3a5-3c5) sections of storage roots showed greater number of starch granules in SC205 and SC8 than that in the wild relative W14. gel images of W14, SC205 and SC8 leaves respectively. More than 300 protein spots on the image of each cassava genotype were analyzed, among which 148, 157 and 152 protein spots were identified in W14 (Fig 4a), SC205 (Fig 4b) and SC8 (Fig 4c) respectively. A total of 122 spots common to W14, SC205 and SC8 were detected (Fig 5a). They were classified according to gene ontology (Fig 5b), and listed in Table 3 and S2 Fig. As a 1.2-1.5-fold change threshold has been often used [7,18], a 1.5-fold change in pairwise comparison of SC205/W14 and SC8/W14 was used as significance to assess protein profiles. While 36 and 31 proteins were observed to vary differentially within the pairs for SC205/W14 and SC8/W14, respectively, with greater than ±1.5-fold in all triplicate gels. These included 25 up-and 11 down-regulated in SC205, and 21 up-and 10 down-regulated in SC8 compared to W14 (Table 3). Five common proteins were detected in both SC8 and W14 leaves, 10 common proteins between SC205 and W14, and 23 common proteins in both SC205 and SC8 leaves (Fig 5a). Additionally, 11, 2 and 2 spots were unique to W14, SC205 and SC8, respectively (Fig 5a, S3 Fig and Table 4).
respectively. The identified proteins were annotated according to gene ontology and listed in Fig 7b and Table 5. One hundred and twenty-seven protein spots were common to all samples (Fig 7a and S4 Fig), in which 39 stained spots in SC205 and 40 stained spots in SC8 were found to have significant changes (p<0.05) with greater than 1.5-fold altered intensity compared with W14 in all three biological replicates. Of these, 19 up-and 20 down-regulated proteins in the pairwise comparison of SC205/W14, 18 up-and 22 down-regulated in the comparison of SC8/W14 were shown in Table 6. Five common proteins were detected in storage roots between W14 and SC205, 6 common proteins between W14 and SC8, and 87 common proteins between SC205 and SC8 (Fig 7a, Table 6). In addition, 58, 9 and 12 protein spots were unique to W14, SC205 and SC8, respectively (Fig 7a, S5 Fig and Table 6). Imaging pulse amplitude modulation of W14 (a), SC205 (b) and SC8 (c) leaves. Parameters were Fv/Fm [maximal photosystem II (PSII) quantum yield], ΦPSII (effective PSII quantum yield) (at 185μE m -2 s -1 ), and NPQ/4 (nonphotochemical quenching) (at 185μE m -2 s -1 ). The color gradient provided a scale from 0 to 100% for assessing the magnitude of the parameters.

Protein interaction networks
A protein interaction map was generated with 19 differential proteins involved with photosynthesis (Fig 10a) and 11 differential proteins related with starch accumulation (Fig 10b). The interactional relationships between the 30 differential proteins included regulation, chemical reaction, molecular transport, expression and binding, responding to photosynthesis, sugar metabolism, and starch metabolism (S1 and S2 Tables). There were direct interactions between 17 up-and 4 down-regulated proteins, associated with photosynthesis and sugar metabolism, in SC205 and SC8 compared with W14 (Fig 10a). Of these, ribulose-bisphosphate carboxylase, phosphoribulokinase,     ribulose-phosphate-3-epimerase, ribose-5-phosphate isomerase, RCA, transketolase, ATP synthase subunit beta, phosphoglycerate kinase, malate dehydrogenase, alcohol dehydrogenase and enoyl-ACP reductase would be directly involved with photosynthesis and carbohydrate and energy metabolism (Tables 3 and 4). The analysis of differential proteins in storage roots showed that there were direct interactions between 10 up-and 3 down-regulated proteins involved in starch accumulation in SC205 and SC8 compared with W14 (Fig 10b), in which succinate dehydrogenase, dihydrolipoyllysine-residue succinyltransferase, UDP-glucosyltrans-ferase, transaldolase, uroporphyrinogen decarboxylase, pectinesterase, triosephosphate isomerase, N-acetyltransferase were related with carbohydrate and energy metabolism (S1 and S2 Tables).

Discussion
In the present study it is the first time to investigate the differences of anatomy and physiology associated with leaf photosynthesis and starch accumulation of storage roots in combination with proteomic technique between cassava cultivars and the wild relative. In this study, we also focused on understanding the relation between the pathways of photosynthesis and starch accumulation, and then indicated the regulated mechanism from photosynthesis to starch synthesis demonstrated by the phenotype of W14, which is a cassava wild species with low starch content.
Proteome changes in photosynthetic activity in leaves M. esculenta ssp. Flabellifolia, the potential wild progenitor of M. esculenta, exhibits typical traits of C 3 photosynthesis, indicating that cultivated cassava, despite its peculiar photosynthetic characteristics, is not derived from wild C 4 species [22]. The mesophyll surrounds the bundle sheath cells, where CO 2 is enriched around Rubisco and the reduction of carbon takes place. The chloroplasts of mesophyll and bundle sheath tissues are adapted to their respective   Rubisco represents about 50% of the total protein content in leaves, and may be among the controlled keys of the photosynthetic pathways. Oxygen-evolving enhancer protein 1 (OEE1), are involved in photosynthesis and may be synthesized in the shoots and then transported to the roots [8]. In the C 3 pathway, CO 2 is fixed by Rubisco and is incorporated into carbohydrate. This metabolic pathway operates only in the mesophyll cells. Leaves of C 4 plants display Kranz anatomy, in which vascular bundles are surrounded by an outer layer of the mesophyll cells and an inner layer of bundle sheath cells [24,25]. As showed in Fig 1, cassava leaves have distinct green bundle-sheath cells, with small, thin-walled cells, spatially separated below the palisade cells (different from Karanz-type leaf anatomy) [26], suggesting cassava is intermediate between C 3 and C 4 species [27].     In the present study, 13 (11 up-regulated and 2 down-regulated) and 11 (10 up-regulated and 1 down-regulated) differentially expressed proteins are directly involved with photosynthesis metabolism in leaves by pairwise comparison of SC205/W14 and SC8/W14, respectively (Table 3). Nice unique proteins related with photosynthesis were also detected in cultivars compared with W14 (Table 4). Therefore, the cultivars may have a higher photosynthetic rate than its wild relative W14. The proteome data imply that up-regulated protein patterns may be related with the increased photosynthetic activities. In two cultivars, the expressions of 9 differential proteins associated with photosynthesis in SC8 were higher than that in SC205, indicating SC8 has higher photosynthetic activity than SC205 (Table 3). These results were consistent with the data provided from measurement of photosynthetic activities using Imaging PAM (Fig 2 and Table 1) and the leaf anatomy (Fig 1). The photosynthesis performance, the expression of C 3 photosynthetic enzymes Rubisco (Fig 8a) and higher resource use efficiency indicate that M. esculanta is likely to be a C 3 and C 4 species. Relative to its wild relative M. esculenta ssp. Flabellifolia, the higher carboxylation efficiency and greater resource use efficiency of M. esculenta are due to its markedly higher C 3 photosynthetic enzyme activities. The high expression of OEC and D1 related with photosynthesis detected by Western blot also supported the result described above (Fig 8b and 8c).

Proteome changes in starch accumulation in storage roots
The cassava storage root, a vegetative structure, accumulates starch as a reserve compound [28] and has no reproductive properties such as for potato tubers. It develops from fibrous roots through massive cell division and differentiation of parenchyma cells of the secondary xylem [19]. However, not all fibrous roots are designated for storage root formation. Little is known about the mechanism involved in the transition from fibrous roots to storage roots. Li et al. The total protein number 127 a . The numbers corresponded to the 2-DE gel in Fig 6; b , NCBI accession number; c , Fold changes of protein spots between SC205 and W14 (Values were means ± SE); d , Fold changes of protein spots between SC8 and W14 (Values were means ± SE); (+) means up-regulated compare with W14, while (-) means down-regulated compare with W14; * indicates differential protein spots in pairwise comparison of SC205/W14 or SC8/W14; ** indicates differential protein spots in pairwise comparison of SC205/W14 and SC8/W14.        Starch is the main form in which plants store carbon. In the present study, 10 (8 up-regulated and 2 down-regulated) and 9 (7 up-regulated and 2 down-regulated) differential expressed proteins were related with carbohydrate and energy metabolism in storage roots by pairwise comparison of SC205/W14 and SC8/W14, respectively (Table 5). Twenty one unique proteins associated with carbohydrate and energy metabolism were also found in cultivars compared with W14 (Table 6). These up-regulated proteins (AGPase, enolase, and aconitase) in cultivars are associated with starch synthesis, glycolysis and TCA cycle, implying cultivars have a higher starch accumulation than its wild relatives. Starch occurs as semi-crystalline granules composed of two polymers of glucose, called amylose and amylopectin. Starch granules are characterized by internal growth rings. There is enormous variation in granule size and shape between plant organs, and between species [19]. The western blot showed that GBSSI, a key enzyme of amylose synthesis, had higher expression in cultivars SC205 and SC8 more than that in the wild relative W14 (Fig 9b). These data were consistent with the measurement of starch and amylose content in the storage roots of SC205, SC8 and W14 (Table 2), indicating storage root enlargement will coincide with the strong regulation of proteins associated with starch biosynthesis. In addition, activation of ADP-glucose pyrophosphorylase (AGPase), a key enzyme in starch synthesis, resulted in a stimulation of starch synthesis and decreased levels of glycolytic intermediates [29]. In the present study we observed that the AGPase (spot 255) was 3.6 and 2.5-fold more highly expressed in storage roots by pairwise comparison of SC205/W14 and SC8/W14, respectively (Table 5). This result was supported by the starch content analysis and amount calculation of starch granules using light microscope between SC205, SC8 and W14 (Fig 3 and Table 2).

Linkages of photosynthesis and storage roots
The photosynthetic rate of cassava is very high and photosynthesis has a broad temperature optimum ranging from 20°C to 45°C [30]. It is cultivated worldwide for the high yield of its storage root containing high amount of starch. However, while our understanding of what is considered their primary function, i.e. starch accumulation and high photosynthetic rate, has increased dramatically in the recent years, relatively little is known about metabolic changes mediated by leaf-root interactions. To help fill in this gap, the anatomic and physiological analysis in combination with proteomics and bioinformatics between cultivated cassava and its wild relative with a low starch content has been employed to indicate the changes of enzyme Biological networks generated from a combination of 19 differential proteins involved with photosynthesis (a) in cassava leaves and 11 differential proteins related with starch accumulation (b) in storage roots. Nineteen differentially up-(a red and upward arrow) and down-(a blue and downward arrow) regulated proteins including ribulose-bisphosphate carboxylase, phosphoribulokinase, ribulose-phosphate-3-epimerase, ribose-5-phosphate isomerase, RCA, transketolase, ATP synthase subunit beta, phosphoglycerate kinase, malate dehydrogenase, alcohol dehydrogenase and enoyl-ACP reductase, ethylene receptor, peroxiredoxin, heat shock protein, glucokinase, glutaredoxin, superoxide dismutase, beta-glucosidase and APX2 in cassava cultivars were used to generate a protein-protein interaction network about photosynthesis through Pathway Studio analysis. Eleven differentially up-(a red and upward arrow) and down-(a blue and downward arrow) regulated proteins including succinate dehydrogenase, dihydrolipoyllysine-residue succinyltransferase, UDP-glucosyltrans-ferase, transaldolase, uroporphyrinogen decarboxylase, pectinesterase, triosephosphate isomerase, Nacetyltransferase, aldo-keto reductase, annexin and pyruvate dehydrogenase in cassava cultivars were used to generate a protein-protein interaction network regarding starch accumulation through Pathway Studio analysis. Regulation is marked as an arrow with R, Chemical Reaction as an arrow with C, MolTransport as an arrow with M, Expression as an arrow with E and Binding as an arrow without any marks. The entity table and relation table were presented in S1 and S2 Tables. activities and the expression levels of global proteins and their linkages between photosynthesis and starch accumulation in the present study. Tables 1and 2 revealed that low photosynthetic activity of W14 leaves resulted in the low dry matter and starch contents in the storage roots compared with cassava cultivars. As shown in Fig 10a, the processes of photosynthesis, starch and sugar metabolism (glycolysis, TCA cycle and pentose phosphate pathway) in cassava leaves produced many intermediate products for starch synthesis, i.e. ADP glucose, 1,4-alpha-D-glucan and Pi (Fig 10a). There are direct interactions between 17 up-and 4 down-regulated proteins, associated with photosynthesis and sugar metabolism, in SC205 and SC8 compared with W14, creating a strong source in cultivated cassava was more than that in its wild relatives (S1 and S2 Tables). A similar association between starch accumulation and the changes of global proteins was also evident in storage roots. Ten up-and 3 down-regulated proteins were detected to involve with starch and sucrose metabolism in cultivated cassava SC205 and SC8 compared with W14, suggesting a strong sink in cultivated cassava was more than that in its wild relatives (Fig 10b and S1 and S2 Tables). In the present study the biological network was established to explain the metabolic changes mediated by leaf-root interactions between cultivated cassava and its wild relatives in global-protein levels. It indicated that the positive crosstalk between the pathways of photosynthesis and starch accumulation in cultivated cassava resulted in the increase of starch content more than that in its wild relatives ( Table 2).

Conclusions
Overall, this study has generated the first comprehensive cassava protein data to show the proteome differences in leave and storage roots between cultivated cassava and its wild relatives due to the genetic differentiation. We detected 148, 157 and 152 leaf-protein spots, as well as 196, 228 and 232 storage-root-protein spots from 2-DE gels of W14, SC205 and SC8, respectively. A total of 175 proteins in leaves and 304 proteins in storage roots were identified, and classified into 12 functional groups annotated via the survey of gene banks. We also developed a biological network to indicate that the positive crosstalk between photosynthesis and starch accumulation may result in the increase of starch content in the storage roots. This implied that both photosynthesis and starch accumulation are equally important for increasing yield of cassava storage roots. We suggested that the divergence in proteome between cultivated cassava and its wild relative was caused by their genetic differentiation and the different number of genes encoding the differential proteins. The detailed analysis in a comprehensive dataset of global proteins and the large-scale bioinformatics will provide a clue for understanding the mechanism of leaf-root interactions and be helpful to choose the key protein markers involved with high starch content in cassava breeding.  Table. Entity table views of protein-protein interactions in biological networks generated for cassava photosynthesis (a) and starch accumulation (b). (XLS) S2 Table. Relation table views of protein-protein interactions in biological networks generated for cassava photosynthesis (a) and starch accumulation (b). (XLS)