The cell wall (CW), considered as first line of defense in plants, is composed of polysaccharides (cellulose, hemicellulose, pectin), and proteins. These proteins can either be structural or non-structural depending upon their functionality. Since the first report of cell wall proteins (CWPs) in Hydrodictyon africanumin (Northcote et al., 1960), hundreds of proteins have been identified which serve as an integral structural part (about 10% of wall dry weight) and perform multiple functions in various signaling pathways.

CWPs have key importance in sensing environmental stresses and controlling CW dynamics in response to the growth and development of the plant. However, currently we have a limited understanding of the structure, function and interaction of CWPs, and also very little knowledge of association of cuticle with plants reactive phytochemicals (Ahuja et al., 2016). Proline rich proteins (PRPs), proline rich extensin like proteins (PRExts), hydroxy-proline rich O-glycoproteins (HRGPs), expansins, and formin like proteins are some of the known classes of CW proteins with covalent scaffold and glycosylation as their known interactions (Boron et al., 2014; Suzuki et al., 2015). Arabidopsis thaliana is a model plant comprising of five chromosomes with 33,542 genes, where 1,605 genes are responsible for the CW development. Out of these 1,605 genes, 252 are responsible for cellulose, 10 for hemicellulose, and 317 for pectin regulation (Albenne et al., 2013). In this review, we have mainly described PRPs, which are the pivotal constituent of the CW together with mining of genes behind these proteins. Along with this, we have attempted to build wired networks to see how these genes co-express and interact in regulating PRPs during the biological and physical processes as well as in determining the molecular functions. We have used the gene mining approach to identify 58 genes located on different chromosomes (Figure 1). These genes are either directly or indirectly involved in the regulation of proline related proteins in the CW under various biotic and abiotic stresses. We have classified these genes into five groups based on their expression for the different structural and functional proteins (Figure 2). Moreover, schematic diagrams (using GeneMANIA and GENEVESTIGATOR) have been generated for the networks of gene co-expression, gene co-localization, genetic interaction, physical interaction, shared protein domains, and predicted interaction (Figures 2–8). A heat map, genome array, Pearson's correlation coefficient (PCC) and hierarchical clustering are also presented for the selected genes for an estimation of the genetic interactions and their level of co-expressions at different plant developmental stages and in the various plant anatomical parts.

The CWPs are divided into nine classes based on their signaling events (Albenne et al., 2013). They are linked to the several important pathways including lipid and carbohydrate metabolism, structural components, proteolytic andoxidoreductive activity, cell signaling, molecular interaction, miscellaneous, and the proteins with an unknown activity.

The CWPs involved in cell signaling, in response to abiotic stresses, have been extensively studied. Under such stress conditions, the major classification of CWPs include the salt overly sensitive kinases, phospholipases, transcription factors, dehydration responsive element binding proteins, C-repeat binding factor, mitogen activated proteins, and abscisic acid responsive element binding factors (Vinocur and Altman, 2005). The involvement of CW in different stress reception mechanisms is not surprising. Kinases are perceived as potential candidates for the CW sensor (Steinwand and Kieber, 2010). Activation of various kinases in response to the changing levels of the same stress has already been well-reported (Kacperska, 2004). In A. thaliana, 26 genes related to the CW associated kinases (CWAKs), and similar functions have been reported (Verica and He, 2002). In addition to the abiotic stresses, CWAKs are also involved in the plant defense against pathogens (Bellincampi et al., 2014). Recently, a number of new CWAKs have been reported, which include proline rich extensin like receptor kinases (PERKs), leucine rich repeats receptor like kinases (LRKs), and lectin receptor kinases (LecRKs) (Wolf et al., 2012). The CW plasma membrane interface is hypothesized as a key site for the stress signal perception where the interaction was studied between arabinogalactan proteins (AGPs) and receptor like kinases (Baluška et al., 2003). The production of hydrogen peroxide and downward redox signaling during the stress is an interesting aspect of CWPs (Spasojević and Pristov, 2010). The generation of reactive oxygen species (ROS) in response to CWP signaling (Barceló and Laura, 2009) is an important and interesting phenomenon, because mitochondria and chloroplast are considered as the major players of ROS production (Voothuluru and Sharp, 2013). In response to abiotic stresses, the extracellular ROS accumulation is tightly regulated by the enzymes (Jaspers and Kangasjärvi, 2010) in cell membrane, which in turn are tightly bonded to the CW (Plieth, 2012). The speedy response of the CW (associated with the changes in its composition or structure) has led researchers to make a detailed study of the various functional proteins with an enzymatic activity within the CW. These include the CW formation, reorganization, loosening and carbohydrate metabolism (Brown et al., 2005; Gupta et al., 2005; Sasidharan et al., 2011; Xu et al., 2014).

Collectively, 2,170 CWPs have been identified on the basis of their distinct gene expression in various plants (San Clemente and Jamet, 2015). The glycoside hydrolases, lyases, esterases, and hydrolases come under the umbrella of proteins acting on polysaccharides (Jamet et al., 2008). The other class of oxidoreductases contains blue copper binding proteins, multi-copper oxidases and peroxidases, while the proteases consist of Aspartate (Asp) proteases, ser carboxy peptidases, and cysteine (Cys) proteases. The oxidoreductases are an important class of enzymes that transfer OH group at critical physiological stages of plant development and affect the structure of the CW (Fry, 1998). Lipid transfer proteins are involved in lipid metabolism (Lev, 2010), and the AGPs in stress signaling (Shen et al., 2001). While, extensins (EXTs) and glycine rich proteins (GRPs) are the structural proteins. Some proteins, grouped as miscellaneous class like germin and germin like proteins, phosphatases and the phosphate inducible proteins, are still unclassified (Shahzad et al., 2013).

The in silico analyses showed 58 genes responsible for the regulation of proline based proteins in the CW of A. thaliana (Figure 3). The gene interaction and co-expression in the form of a wired network has been constructed for the cellular component and biological and molecular process. A significant variability has been observed in the degree of physical interactions, predicted interactions, co-expressions, genetic interactions, shared protein domains, and co-localization.

The GRPs located within the CW of vascular tissues are regulated during the developmental stages of plants (Ye et al., 1991) by forming a third major group of CW proteins. The manifestation of many GRPs takes place under environmental stresses like water deficiency, high light, ABA, and pathogenic infestations. Although, it is presumed that GRPs were a part of the plant defense system, their mechanism of action is not yet known. It is possible that GRPs presenting other functional domains were necessary to determine how protein activity get affected under different conditions (Mangeon et al., 2010).

In plants, GRPs related genes are regulated during developmental stages and their expression varies in plant tissues (Yan et al., 2015). In different genera of plants, the expression of such genes is controlled by biotic and abiotic stresses (Ahmad et al., 2014; Alghabari et al., 2015, 2016; Ihsan et al., 2015; Yan et al., 2015). In plants, the categorization of GRPs is based on the semi-repetitive glycine rich motifs (Sachetto-Martins et al., 2000). According to a report, the French bean PvGRP1.8, a class 1 GRP, located in un-lignified primary CW, played a structural role in protoxylem through the CW buttressing (Ryser et al., 2004). Reverse genetics approaches have fortified the concept of the involvement of GRPs gene from Arabidopsis in the deposition of the secondary CW and maintenance of proto-xylem structure (Yokoyama et al., 2007).

Undoubtedly, the diversified structure, intonation, expression, prototype, and subcellular localization of GRPs are strongly witnessed for their prominent role and functionality in plants (Sachetto-Martins et al., 2000). Observations have disclosed the involvement and modulation of GRPs in the defense mechanism under pathogenic attack (de Souza Cândido et al., 2011). The NtClG1 gene was induced in turnip whose level was altered by tobacco virus, as shown by increased deposition of cellulose, which restricted the viral movement. This implicate structural roles of GRPs in conferring defense mechanisms in plants (Ueki and Citovsky, 2002). The GRPs constitute almost 70% glycine (Kar et al., 2012). Analysis done by immunocytochemistry has revealed their direct alliance with proto-xylem, where they play a prominent role in repair and stretching phase (Sachetto-Martins et al., 2000). It is perceived that the continuity of glycine rich domains produced beta pleated hydrophobic structure. An in vitro cross-linking experiment in the presence of peroxidases has explored the configuration of networks in Tyr-containing GRPs. Nonetheless, there is further need to do experimentation to generate data to support the characterization of intra and inter molecular networks involving GRPs.

The GRPs are presumed to be involved in promoting expression of genes in plants, exemplified through the involvement of RNA binding GRP gene AtCSG2 and their regulation during flower development (Sachetto-Martins et al., 2000). The plants in which AtCSG2 was silenced due to biotic or abiotic stress showed premature flowering with reduced stamens and abnormal embryo development (Fusaro et al., 2007). The GRPs yet isolated from plants are categorized as CW-GRPs, RNA-GRPs, and cytokeratin like GRPs (Sharma et al., 2012). Analysis conducted through bioinformatics tools has explored 12 genes controlling different types of GRPs in the CW of A. thaliana. These genes also perform salient molecular functions in different kinds of cells and plant parts (Table 4). The study has also revealed eight genes, which could not be categorized to any kind of CWPs and the functions of these genes are indicated in Table 5.

In A. thaliana, 2,700 proteins express 6,200 highly reliable interactions. The interactive maps provided a dynamic approach in better understanding of the plant biological systems and a base for future crop improvement (Braun et al., 2011). The exploitation of co-expression networks in Arabidopsis provides a dimension to mine genes involved in the synthesis of CW and to unravel the structural hierarchy of CW in systematic progression (Obayashi and Kinoshita, 2009). Here, we have made queries through PubMed to understand the genes expression and co-expression in a wired way in the CW through different kinds of structural proteins. Through bioinformatics tools, 58 genes involved in proline based CW regulation were found. The established wired networks showed the genes co-expression, and interaction in structural components, biological processes and molecular functions by regulating the synthesis of proline based CW proteins. In regulation of biological processes, these genes have shown physical interaction 55.7%, co-expression 15.6%, genetic interaction 6.9%, shared proteins domains 3.6%, and co-localization 0.6% (Figure 3). However, in determining cellular components, these genes have revealed physical interaction 7.4%, co-expression 55.9%, genetic interaction 1.1%, shared proteins domains 2.9%, and co-localization 1.8%. Moreover, these proteins helped in signal transduction by regulating the molecular functions for which physical interaction 39.0%, co-expression 11.8%, genetic interaction 3.3%, shared proteins domains 14.6%, and co-localization 0.4% were calculated. Similarly, gene networking and interaction, based on biological process, molecular process and cellular component for protein rich protein extensins, GRPs, arabinogalactan proteins, and PRPs indicated the genes co-expression and interaction even when they were considered separately for a particular family of proteins (Figure 4). These wired networks as a whole, are the clusters of genes repertoire, interacting and co-expressing for different kinds of proteins present in the CW. However, out of this treasury of genes, we mined the genes interacting for a particular class of proteins. Therefore, in this complete wired network, we have showed genes, which are interacting for a specific family of proteins as dark (black) spots. However, light spots represented those genes whose interactions were not considered (Figure 4).

The dynamic signals of the environment, triggering the genetic regulatory network, fluctuate continuously in plant's life cycle with varying extent of predictability. Plants are sessile beings, so to cope with such fluctuating environment, they do not show behavioral adaptations but they respond by modulating their development and physiology (Leakey et al., 2009). The percent of expression of these mined genes was found variable (0–100%) in different anatomical parts of A. thaliana, as it was revealed by heat map constructed through META analysis (Figure 5A). Likewise, fluctuating manifestation of these genes was noticed for different developmental stages (Figure 5B).

A genome array map for 10 representative genes of A. thaliana, in 27 anatomical parts at different developmental stages (germination to senescence) was constructed that represented an alteration in gene expression level with changing developmental stages of the plant (Figure 6). A total of 93 proteins (Boudart et al., 2005) have been identified using proteomic and bioinformatics approaches. A comparison of rosette plants revealed the highly cell type specific involvement of CW proteomes regulated by multigenic families. The plant allocates 10% of their genome (~2,500 genes in Arabidopsis) to synthesize and rearrange the CW (McCann et al., 2007) indicated by the study of protein sequential annotation. Nevertheless, the number of mined genes in the CW of A. thaliana are limited but have been justified experimentally for their involvement in the CW formation. However, cis- regulatory DNA and proteins interact to synthesize the organized CW structure, assured by the synthesis and modifications of these basic constituents in a well-maintained and coordinated fashion, by specific enzymatic activities as revealed by transcriptional coordination of their related genes (Brown et al., 2005).

Xylan is involved in the development of secondary CW in plants and its synthesis network possesses a set of highly expressed genes. A comparative co-expression analysis between rice and Arabidopsis disclosed the absence of some gene families that were present in other species, indicating a clear difference in their CWs (Oikawa et al., 2010). Previous research has published numerous studies about gene interactions, gene expression and protein interactions, but now there is a major need to integrate this knowledge to understand the basic features of living organisms that are proceeding in an organized way. The different interactions including gene expression, gene interactions and protein interactions were used to assemble a biological network that has defined the basic principles to wire a network in a chronological way (Beyer et al., 2007).

Based upon available proteomics and genomics data sets of A. thaliana, geneMANIA can be used to estimate their gene function. To study co-expression and positive as well as negative correlation of gene expression profiles, we applied Plant Array Nets, which works equally good for Oryza, Brassica, and Arabidopsis (Warde-Farley et al., 2010) The immense knowledge as an outcome of genome sequencing paves the way to understand the working philosophy of genes in an integrated way on genome sequencing, expression analysis and protein interaction. The transcriptional coordination has been estimated using PCC. By using this method, co-expression relationships between many genes can be estimated, and visualized as a network in which nodes indicate genes whereas connection between nodes represents the transcriptional coordination of genes (Aoki et al., 2007). The PCC method sometimes becomes defective when some biological processes are strongly transcriptionally co-regulated, while other processes are not. In addition to this, a lower value of PCC results in excessively large gene clusters, possessing thousands of genes (Mao et al., 2009).

The PCC for 50 proline based CW regulating selected genes presented a positive and negative co-expression of +0.969 and −0.827, respectively (Figure 7). Hierarchical clustering of genes based on PCC indicated the co-expression of some genes with same intensity at particular developmental stage with altered expression level changing in the developmental stage (Figure 8). The genes that showed co-expression also represented a high degree of functional correlation. Co-expressions studies of genes can be used to identify other genes. For example, in cellulose synthesis, the co-expression approach can be used to identify the genes involved in the synthesis of hemicellulose (Cocuron et al., 2007). Many genes that are transcriptionally associated with the synthesis of the CW have been already studied (Ruprecht et al., 2011). The genetic redundancy needs mutant combinations or knocks down approaches that will focus upon several homologous genes to generate informative phenotypes. In addition to this, detailed comparative transcriptional studies are still required to mine candidate genes for the CW synthesis.

The co-expression analysis gives one possible caveat of “false positives,” which means some genes are co-expressed by chance rather than being functionally related. However, it has been reported that co-expression relationships are often conserved across species (Obayashi and Kinoshita, 2009). Hence, co-expression analysis across species can improve the reliance of co-expression based functional annotation.

Through computational methodologies, we have generated figures highlighting the expression of genes at different plant specific stages. The module-based predictions provide an approach to formulate hypothesis for functionally unknown genes (1,701) in Arabidopsis and other plant species. It also provides a new imminent into the conservation of co-expression and co-regulation (Heyndrickx et al., 2014). Through proteins architecture studies of the CW, we have identified several genes directly and indirectly involved in proteome manufacturing (Yang et al., 2011). In response to heat stress, P5CR launches its oxido-reductase activity by producing pyrroline-5-carboxylate reductase enzyme at the vicinity of the cytoplasm and CW. Under the conditions of biotic and abiotic stresses, the gene express itself in CW compartment by enhancing proline transport and increasing sensitivity against pathogenic stimuli (Bosch et al., 2011). Hence, modifications in CW proteins and proline transport are an indicator of regulation of genes expression under biotic stresses.

For co-expression studies, the bioinformatics tools have been focused on the model plant Arabidopsis by including the major bulk of publicly available microarray datasets. The candidate genes forming the foundation of the existing A. thaliana CW regulatory network, have been identified by gene expression profiling (Handakumbura and Hazen, 2012). Genes with similar functionality and overlapping effects, such as expression and regulation of floral developmental and defense related genes in response to biotic stress (Ahmad et al., 2013, 2014), can also be coordinated as indicated by global transcript analysis based upon publicly available microarray datasets. Certainly, through co-expression analysis in A. thaliana, many transcriptionally coordinated genes involved in the formation of CW proteins, cellulose, hemicelluloses and lignin have been identified. To facilitate this co-expression analysis, several helpful web based tools have been developed for the researchers to investigate transcriptional co-ordinations as well as to mine the candidate genes involved in the CW integrity. In addition, several tools paved the foundation to make comparative transcriptional analysis across many species, which will potentially increase predictive power about gene functionality.

ZA, JNA, and MI came up with the ideas and reviewed all the literature; MI, SJNA, ZS, and HR took part in writing the manuscript. IA, JNA, and AMB reviewed, critically analyzed and edited the manuscript. All authors discussed and commented on the manuscript.