Fatty acids, which are first converted to acyl-CoA esters, are completely degraded to acetyl-CoA after repeateldy passing through ß-oxidation reactions. In mammalian cells, thiolase enzymes exist in both the mitochondria and in peroxisomes; whereas in plants, these enzymes appear to occur only in peroxisomes (Kindl 1980). Mammalian peroxisomal ß-oxidation functions as a chain-shortening system and needs the mitochondrial system for the complete oxidation of the fatty acid. Meanwhile, plant peroxisomal ß-oxidation is able to carry out the complete ß-oxidation of fatty acids in glyoxysomes (Mannaerts 1982).

Glyoxysomes are specialized peroxisomes that are mainly found in endosperms and cotyledons of germinating oilseeds; they are characteristically high in ß-oxidation activity. Research in this area has focused primarily on plant fatty tissues, specifically on the mobilization of storage reserves from castor bean endosperms and the mesophyll of fat-storing cotyledons (Cooper and Beevers 1969; Rees 1980).

While the enzymes of the plant peroxisomal beta-oxidation systems have been biochemically well-characterized (Gerhardt 1992), it is unknown whether or not the ß-oxidation system plays as important a role in plants as it does in animals.

Loss of either peroxisomal or mitochondrial beta-oxidation systems in animals causes severe biochemical abnormalities and clinical symptoms. Zellweger cerebro-hepato-renal syndrome, infantile phytanic acid storage disease, and neonatal adrenoleukodystrophy (ALD) are examples of diseases characterized by a decreased number or absence of peroxisomes in liver and other tissues (Watkins et al. 1989; Brown et al. 1993). Loss of various human beta-oxidation enzymes produces sudden, unexplained death in childhood, acute hepatic encephalopathy, skeletal myopathy, hypotonia, or cardiomyopathy (Hale et al. 1985; Treem et al. 1991; Sims et al. 1995; Strauss et al. 1995).

Work ß-oxidation enzyme deficiency in plants has been very limited. ß-oxidation enzymes have been identified in non-fatty tissues (Gerhardt 1981, 1983; Gühnemann-Schäfer and Kindl 1995); although, their function in the development of the plant is not known. Recently, the coexistence of two thiolase isoforms, thiolase I and II, has been shown in sunflower glyoxysomes, the only system in which this is known (Gerhardt 2002).

The two sunflower thiolases are by far the best catalytically and biochemically characterized glyoxysomal thiolases from plant tissue (Gerhardt 2002); however, only one of these two plant thiolases has been cloned (Schiedel 2004). It is the cloning of thiolase II that permits the bioinformatics approach of studying plant thiolase and allows a better understand the fatty acid catabolism in plant growth and development.

Until recently, bioinformatics has often only been used as an alternative method to establish and verify various theories. However, many researchers now use it as a first approach to study proteins as opposed to using experimental analysis. These results and data, which are primarily predictions from computational analysis, have significantly helped protein biologists to design successful experiments by providing a range of limits to work with. This, in turn, has helped reduce trial and error; thus, minimizing the time consumed in designing experiments and acquiring results.

Computational tools and databases like DXS (Souret 2003) have previously been used to analyze proteins in order to gain insight into their function, structure and evolution. Thiolase II from eight plant species have been sequenced and made available on the NCBI database; however, no comparative or phylogenetic studies of these plant thiolases have been found to exist.

This study uses a computational approach to analyze the physico-chemical properties and structural features of sunflower thiolase II (3-oxoacyl CoA thiolase). Furthermore, phylogenetic studies using the sunflower thiolase II and thiolase II from the seven other plants are conducted to compare these enzymes and determine their structural and functional relationship. Data from this research will serve as the basis for a better structural understanding of sunflower thiolase II. Additionally, this information will be useful in both comparative studies and in development of purification protocols for thiolase II.

Similarity Search on Sequence Database

PSI-BLAST available at http://www.ncbi.nlm.nih.gov/BLAST is used to find other plant thiolase II proteins related to the sunflower thiolase II using NCBI BLAST. The PSI (Position Specific Iterated) and PHI-BLAST tools are selected for the similarity search. When the sunflower thiolase II sequence is inputted into the textbox of the program, the expect value changes to 0.001. A lower expect value corresponds to higher similarity between the sequences. The sequence is blasted with the remaining parameters unchanged. The results are then carefully inspected to select sequences of interest, and the iteration is repeated several times to find the related sequences.

Prediction of Physico-Chemical Parameters

The physical and chemical properties of the sunflower thiolase II are estimated using ProtParam, a tool available on the ExPASy server (http://www.expasy.ch/tools/#primary). The full length coding sequence of the sunflower thiolase II (AY308827) is entered in the raw format and the parameters are computed. Knowledge of the physical and chemical parameters of a protein is very helpful when designing experiments. These parameters include molecular weight, the isoelectric point, extinction coefficient, half-life, instability index, aliphatic index, average hydropathicity and amino acid composition.

Prediction of Post-Translational Modifications

The protein is searched for patterns in the PROSITE database with ScanProsite (http://www.expasy.ch/tools/scanprosite/), and the function of the protein is predicted. The raw data is entered in the search box in the Scan a protein for PROSITE matches section, and the Patterns check box is checked for the scan. The protein is then analyzed for domains using the two domain databases: InterProScan (http://www.ebi.ac.uk/interpro/scan.html) and NCBI Conserved Domain server database (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). In addition to the sunflower thiolase II protein, the related sequences from the PSI-BLAST are also analyzed for domains, and the results are then compared.

Hydrophobicity Analysis

The hydrophobicity profiles of the eight thiolase sequences are obtained using ProtScale (http://www.expasy.ch/cgi-bin/protscale.pl). Linear regression analysis using MS Excel is then used to study the occurrence of any trend in these profiles.

Secondary Structure Prediction

The PredictProtein Server (http://www.sdsc.edu/predictprotein/) is used to predict the secondary structure and any nother related features of the protein. Some of these features include solvent accessibility of the various residues, identification of transmembrane helices and their topology, coiled/coil regions, and bound cysteines in the protein sequence.

Multiple Sequence Alignment

The raw sequences of the sunflower thiolase II and its related proteins from the PSI BLAST result are inputted into Tcoffee (http://igs-server.cnrs-mrs.fr/Tcoffee) for multiple sequence alignment. The result is then examined manually to identify conserved regions.

Phylogenetic Tree

The aligned sequences from Tcoffee are used to build the Neighbor Joining tree with Phylip (http://bioweb.pasteur.fr/seqanal/phylogeny/phylip-uk.html). The distance matrix for the multiple sequence alignment is calculated using Protdist (http://bioweb.pasteur.fr/seqanal/interfaces/protdist.html), which is provided by Phylip. The data is bootstrapped with the seed set to an odd number before it is analyzed. The odd number in this case is three, and the number of replicates is 100. The resulting distance matrix from Protdist is inputted into another program neighbor (http://bioweb.pasteur.fr/seqanal/interfaces/neighbor.html), which is also provided by Phylip and bootstrapped with the same parameters used for Protdist. The result that is in the Newick format is then displayed with Phylodendron (http://www.es.embnet.org/Doc/phylodendron/treeprint-form.html) and represented as a phenogram.

PSI Blast is used instead of Blast in this study, because the search focus is on sequences that belong to a large protein family. Each new round of blast is called an iteration, and sequences of interest are manually selected from the blast result during each iteration. Based on this selection, the PSI blast retrieves only those sequences from the database that match the common properties found in the sequences selected in the previous iterations. The iteration for this study is repeated ten times, and the following plant thiolase sequences are selected from the final iteration: arabdidopsis thaliana (NM_128874), Japanese rice (XP_468412), mango (CAA53078), rape seed (TO7989), pumpkin (S72532), cucumber (CAA47926) and soybean (AAQ93070).

ProtParam Predictions

The ProtParam program computes the various parameters of a protein without taking into consideration sequence modifications that may occur before protein activation. The sunflower thiolase parameters are in close proximity to the other seven thiolases, which indicates a similarity in the chemical and physical properties of these enzymes (Table 1). The program, however, ignores various factors such as posttranslational modifications or complex maturations. These modifications contribute significantly to the molecular weight of a protein; hence, deviations from experimental results may occur as a result.

Table 1 . Physical and chemical parameters of the sunflower thiolase II as predicted by ProtParam (http://www.expasy.ch/tools/#primary).

The ProtParam program predicts the various physical and chemical properties required for designing experiments for further protein analysis. Thereby, it minimizes the number of experimental trials. For instance, knowledge of the isolelectric point (pI) is useful for developing buffer systems for purification, while determination of protein concentration or extinction coefficients helps in the quantitative study of protein-protein and protein-ligand interactions in solution.

A stable protein’s instability index is smaller than 40. The relative volume of a protein occupied by the aliphatic side chains (Ala, Val, Ile & Leu) is calculated to measure the aliphatic index. The thermostability of the protein increases with increasing aliphatic index. GRAVY, or knowledge of hydropathicity, helps in understanding the solubility characteristics of the protein. Protein solubility helps predict protein structure. The hydropathicity of a protein is found by summing the hydropathicity indices of each amino acid of the protein. The parameters for the seven thiolases are similar to the sunflower thiolase II with exception of mango thiolase, which is shorter (430 aa) (Table 1).

Post translational Modifications

The PROSITE database is a collection of patterns and profiles. Patterns are highly conserved segments. For example, [RK]-x-[ST] describes a small group of amino acids that constitute the signature of a function, a posttranslational modification, a domain, or a protein family. Profiles, on the other hand, describe every position of a protein family and not just a few conserved segments. This makes them more complicated than patterns. Proteins are compared with the list of patterns and profiles contained in the PROSITE database.

InterProScan, a domain database, is used to predict the domains present in a protein. A domain is an independent globular folding unit, and an average protein contains two or three domains. Domains also play an important and specific role in the function of the protein. Unlike other servers or databases, the NCBI Conserved Domain (CD) database scores the reported hits (i.e., results) that help in discriminating good matches from the spurious ones. Though the CD server does not integrate as many databases as others, some domains found in the NCBI database are not found elsewhere.

InterProScan detects three patterns in the sunflower thiolase II sequence. The first one (132-150) is called thiolase 1, “thiolase acyl-enzyme intermediate signature”. The second one (381-397) is called thiolase 2, “thiolase signature”. Finally, the third one (418-431) is called thiolase 3, “thiolase active site”. These three regions between positions 132-150, 381-397, and 418-431 have previously been identified as important in the enzymatic function of sunflower thiolase II (Schiedel et al., 2004). These regions were also shown to be conserved to a large extent in the other seven sequences. The carboxyl-terminal, amine-terminal, and AcCoA-C-Actrans (also known as Fatty oxidation complex ß subunit) domains have also been identified in the sunflower thiolase II. These same domains are also present in all the other thiolases as well. Other patterns that do not belong to any particular class or family (such as ?-ketoacyl synthase) are also identified in all the sequences including sunflower (not shown). Mango and sunflower thiolase II sequences are shown to share the same set of these unclassified patterns (Figure 1).

Figure 1 . InterPro Scan identified three pattern motifs in sunflower thiolase II. The straight line represents the primary sequence of thiolase II. The colored blocks on the line represent different patterns identified in the sequence. The longer blue block represents the amine terminal domain. The shorter blue line represents the carboxyl terminal domain.

ScanProsite also identified the three patterns discussed above in the sunflower and other thiolase sequences. The NCBI CD server returned a set of domains for acetyl-CoA acetyl transferase, which is involved in lipid transport and metabolism (Figure 2). In addition to the acetyl-CoA acetyl transferase, the same amine-terminal and carboxyl-terminal domains found in ScanProsite are also identified. These amine and carboxyl terminal domains are also identified in the other seven plant thiolase sequences.

Figure 2. NCBI CD-identified domains for lipid transport and metabolism in sunflower thiolase II. All eight proteins have the same identified domains. The black line represents the primary sequence of the sunflower thiolase II. Domains are represented in red and blue. The blue regions on the sequences are masked out due to low complexity. KOG1389 represents 3-oxoacyl CoA thiolase. PaaJ represents the acetyl-CoA acetyl transferase domain. Thiolase represents the amine terminal domain. Thiolase_C represents the carboxyl terminal domain.

Hydrophobicity Analysis

Kyte-Doolittle hydrophobicity analysis of the sunflower thiolase II predicts structure. Regions of the protein below the 0.0 line are hydrophilic, while regions above the line are hydrophobic. In the case of a membranous protein, a region corresponding to a peak above two indicates the region’s potential of being a transmembrane domain. Previous studies have used linear regression analysis to find the presence of any trend in the protein’s primary structure (Beevers 1961). A consistent trend of increasing hydrophobicity along the length of the protein sequence may be seen using this linear regression analysis. It can be predicted from this trend that the thiolase enzyme is predominantly hydrophobic. Also, it is evident from the plot that there are no transmembrane domains in the thiolase enzyme. This is expected as it is known that thiolase is not located in cellular membranes.

The hydrophobicity plot of sunflower thiolase II is compared to the seven plant thiolase sequences; a linear regression analysis of the profiles is then done to find trends (Figure 3). These plots give an indication of whether or not the protein is predominantly hydrophobic or hydrophilic. Some positions in the enzyme have peaks identical to those in all other sequences (e.g., position 412.) Regions corresponding to these positions are highly conserved and are important for both the structure and function of the enzyme.

Figure 3. Kyte and Doolittle hydrophobicity analysis of sunflower thiolase II, soybean thiolase, and Japanese rice thiolase. Linear regression analysis shows a consistent hydrophobic trend in all the eight thiolase sequences. Regions of profile with similar peaks correspond to conserved regions of the proteins.

Secondary Structure Prediction

A graphical output of the predicted secondary structure of the sunflower thiolase II is shown in Figure 4. The structural composition of the protein is predicted to be 39% Helix, 12% strands, and 49% coils. The average structural composition of the other seven proteins are predicted to be 37% Helix, 13% strands and 50% coils. The solvent accessibility composition (i.e., the core/surface ratio) of the sunflower protein is also predicted. It is found that 44% of the residues of the sunflower thiolase II have 16% or more of their surface exposed to the outside. These predictions are similar for the other proteins as well. An average of 43% of their residues have 16% or more of their surface exposed to the outside.

Figure 4. The predicted secondary structure of a region of the sunflower thiolase II from PSIPRED (http://bioinf.cs.ucl.ac.uk/psipred). The H stands for helix, C for coil, and E for strand. The blue bars for each amino acid represents the confidence of each prediction. The taller the bar, the higher the confidence.

Multiple Sequence Alignment

Multiple alignments provide valuable information for prediction of the structure and function of a protein and for generation of phylogenetic trees. In short, multiple alignment puts amino acids in the same column, as they are homologous for certain criterion. For instance, if the criterion is structural similarity, then amino acids that play the same role in each structure are put in the same column. Tcoffee is a multiple sequence alignment generating software that uses a progressive alignment. In other words, it adds sequences little by little until the complete multiple sequence alignment is finished. All the sequences are then compared two by two; finally, they are clustered based on their shared ability to generate a dendrogram resembling a phylogenetic tree. Tcoffee then uses this dendrogram to align the two sequences in each clustered set. The two sequences in each clustered set are now considered a single alignment, and the single alignment is aligned with the clustered set that is most similar to it. This process continues until all the sequences have been aligned. Results from Tcoffee are found to be more reliable, but take longer to achieve, when compared to other available tools such as ClustalW.

Multiple sequence alignment of the seven plant thiolases and sunflower thiolase II by Tcoffee is shown in Figure 5 . Sequence alignment shows that many regions are highly conserved. Some conserved regions are short and span only 4 amino acids (e.g., region 55-58), while others are longer and span 9 amino acids (e.g., region 164-173.) Due to their importance in enzyme function, selective pressures slowed the mutation rates in these regions so that they are conserved. The two cysteine residues important for the catalytic mechanism of the thiolase II reaction occur at 143 and 430 in the aligned sequences (Figure 5; Anderson et al. 1990). Another example of a conserved residue important for the enzymatic activity of thiolase II is His, which is at position 398 (Williams et al. 1992).

Phylogenetic Tree

When two protein sequences are being compared and their similarity is statistically significant, it is highly likely that the two proteins are evolutionarily related. The purpose of the phylogenetic tree in the context of this study is to compare proteins from several species based on their amino acid composition. They proteins are the grouped according to their level of similarity. This comparative analysis will provide information about any relationship (e.g., structural) that may exist among these plant thiolases. This result may be used to provide an insight into the function and evolution of these eight enzymes.

Various methods exist for the construction of a phylogenetic tree; “distance methods” are most preferred due to their speed and simplicity. Examples of the distance-based methods include UPGMA, Neighbor-Joining method, Least-square method, and Transformed Distance method. Though each method has its advantages and disadvantages, the Neighbor-Joining method is found to be the preferred choice of most molecular evolutionists as it is faster and more suitable to handle larger datasets. The distance-based methods require a distance matrix to be calculated from the multiple sequence alignment. The Portdist distance matrix is then used to construct the phylogenetic tree. Based on the distance matrix from the multiple sequence alignment, and the similarity of the sequences, the distance-based Neighbor-Joining tree places the eight thiolases into different groups (Figure 6).

Figure 6. Phylogenetic tree showing the evolutionary relationship between the eight thiolase hsequences. Sunflower is grouped with Japanese rice, soybean with mango, and pumpkin with cucumber. The farther a sequence is from the root, the greater the possibility for evolutionary change.

All of the eight plant thiolase sequences belong to the division magnoliophyta. Of these eight flowering plants, Japanese rice is the only monocot and thus expected to be grouped alone. From a taxonomical perspective, a monocot is supposed to be different from the other 7 sequences that are dicot; hence, it is placed in a separate group in the phylogenetic tree. However, this is not done for Japanese rice as it is placed with sunflower thiolase II (Figure 6).

The phylogenetic tree thus has two main classes. The first class has the rape seed sequence as its only member. The second class is further divided into two groups. One group has Arabidopsis Thaliana. The second group, which is further divided, has sunflower, Japanese rice, mango, soybean, pumpkin and cucumber. From the tree grouping, it is observed that sunflower thiolase II (a dicot) is closest to Japanese Rice (a monocot). Mango, soybean, sunflower, and Japanese rice have also been placed farthest from the root in the phylogenetic tree; this implies that they have evolved the most out of all the eight thiolase sequences (Figure 6).

Beevers H. (1961). Metabolic production of sucrose from fat. Nature. 191:433-436.

Brown FR et al. (1993). Peroxisomal Disorders: Neurodevelopmental and Biochemical Aspects. American Journal of Diseases of Children. 147:617-626.

Cooper TG and Beevers HJ. (1969). Beta oxidation in glyoxysomes from castor bean endosperm. J Biol Chem. 224:3514-3520.

Graham IA and PJ Eastmond. (2002). Pathways of straight and branched chain fatty acid catabolism in higher plants. Progress in Lipid Research. 41:156-181.

Gerhardt B. (1992). Fatty acid degradation in plants. Progress in Lipid Research. 31:417-446.

Gerhardt B et al. (2002). Glyoxysomal acetoacetyl-CoA thiolase and 3-oxoacyl-CoA thiolase from sunflower cotyledons. Planta. 214:597-607.

Gerhardt B. (1981). Enzyme activities of the beta-oxidation pathway in spinach leaf peroxisomes. FEBS Letters. 126:71-73.

Gühnemann-Schäfer K and H Kindl. (1995). The leaf peroxisomal form (MFP IV) of multifunctional protein functioning in fatty acid beta-oxidation. Planta. 196:642-646.

Hale DE et al. (1985). Long-chain acyl coenzyme A dehydrogenase deficiency: an
inherited cause of nonketotic hypoglycemia. Pediat Res. 19:666-671.

Hiltunen JK and Y Qin. (2000). Beta-Oxidation – strategies for the metabolism of a wide variety of acyl-CoA esters. Biochim Biophys Acta. 1484:117-128.

Kindl H and J Frevert. (1980). Purification of glyoxysomal acetyl-CoA acyltransferase. Physiol Chem. 361:537-542.

Mannaerts GP and LG Debeer. (1982). Mitochondrial and peroxisomal beta-oxidation of fatty acids in rat liver. Ann N Y Acad Sci. 386:30-9.

Mittendorf VV et al. (1998). Synthesis of medium-chain-length polyhydroxyalkanoates in Arabidopsis Thaliana using intermediates of peroxisomal fatty acid beta-oxidation. Proc Natl Acad Sci. 95(23):13397-13402.

Nawrath C et al. (1994). Targeting of the polyhydroxybutyrate biosynthetic pathway to the plastids of Arabidopsis Thaliana results in high levels of polymer accumulation. Proc Natl Acad Sci. 91(26):12760-12764.

Poirier Y et al. (1999). Increased flow of fatty acids toward beta-oxidation in developing seeds of Arabidopsis deficient in diacylglycerol acyltransferase activity or synthesizing medium-chain-length fatty acids. Plant Physiol. 121:1359-1366.

Poirier Y et al. (1992). Perspectives on the production of polyhydroxyalkanoates in plants. FEMS Microbiol Rev. 103:237-246.

Poirer Y et al. (1995). Production of polyhydroxyalkanoates, a family of biodegradable plastics and elastomers, in bacteria and plants. Biotechnology. 13(2):142-150.

Schiedel AC et al. (2004). Cloning, expression and purification of glyoxysomal 3-oxoacyl-CoA thiolase from sunflower cotyledons. Prot Exp Purif. 33:25-33.

Sims HF et al. (1995). The molecular basis of pediatric long chain 3-hydroxyacyl-CoA dehydrogenase deficiency associated with maternal acute fatty liver of pregnancy. Proc Natl Acad Sci. 92:841-845.

Souret FF et al. (2003). Computational analysis of the evolution of the structure and function of 1-deoxy-D-xylulose-5-phosphate synthase, a key regulator of the mevalonate-independent pathway in plants. Gene. 313:127-138.

Strauss AW et al. (1995). Molecular basis of human mitochondrial very-long-chain acyl-CoA dehydrogenase deficiency causing cardiomyopathy and sudden death in childhood. Proc Natl Acad Sci. 92:10496-10500.

Treem WR et al. (1991). Hypoglycemia, hypotonia, and cardiomyopathy: the evolving clinical picture of long-chain acyl-CoA dehydrogenase deficiency. Pediatrics. 87:328-333.

Watkins PA et al. (1989). Peroxisomal bifunctional enzyme deficiency. J Clin Invest. 83:771-777.

Williams S et al. (1992). Biosynthetic thiolase from Zoogloea ramigera. Mutagenesis of the putative active-site base Cys-378 to Ser-378 changes the partitioning of the acetyl S-enzyme intermediate. J Biol Chem. 267:16041–16043.