A comparative structural analysis of sepiapterin reductase from Drosophila by homology modeling

Multiple alignments and phylogenetic analysis

Homologous sequences were searched against SR proteins using the BLAST program, available from the National Center for Biotechnology Information (NCBI). The sequence alignment of SR proteins from different organisms was carried out using the ClustalW2 program (EMBL-EBI, Hinxton, Cambridgeshire, UK) and the MUSCLE software version 3.8.31 (EMBL-EBI, Hinxton, Cambridgeshire, UK) [16]. Phylogenic analysis was performed using the ClustalW2 program and the Molecular Evolutionary Genetics Analysis (MEGA, v. 6.0) software program [17]. The validity of the various branches of phylogenetic trees was tested by bootstrap analysis, using 1000 replications.

Homology modeling

Three-dimensional structure predictions of Drosophila SRs and zebrafish SR were generated by the I-TASSER server for protein structure and function prediction based on a threading alignment algorithm [18, 19]. All structural images were prepared with the Molegro Molecular Viewer software v. 2.5.0 (Molegro ApS, Aarhus C, Denmark).

Evaluation of model

The best structural model of Drosophila SR was chosen on the basis of the stereochemistry quality report, generated using the RAMPAGE server (Crystallography and Bioinformatics group at the University of Cambridge, Old Addenbrooke’s, Cambridge, UK).

Identification of Drosophila SRs

A homology search against the Drosophila database yielded two highly similar genes: CG12117 and CG12116. Two different SR genes are located on chromosome X of Drosophila at the 7E11 (CG12117) and 7F1 (CG12116) loci; these genes do not contain introns. The molecular weight of CG12117 and CG12116, predicted from the deduced amino acid sequences, is 29.2 and 31.0 kDa, respectively. Recently, the CG12117 gene was named Sptr. Although both genes were predicted to have SR activity, only CG12117 showed SR enzymatic activity in vitro [14]. The deduced amino acid sequences were compared with the SR sequences of other organisms [9–11] (Figure 1A). Drosophila CG12117 and CG12116 respectively exhibited approx. 31% and 23% sequence similarities to the SR proteins of other organisms (mouse, rat, human, zebrafish); in addition, 67 and 114 residues were fully conserved and less conserved in the two Drosophila SRs, respectively. Furthermore, both Drosophila SRs were predicted to share the same folds, including eight α-helices and seven β-sheets (Figure 1A). Most SRs, in other organisms, have a Rossmann fold: a -G-X-X-G-X-G- sequence for NADP(H) binding of the N-terminal cofactor, a strictly conserved serine, and a -Y-X-X-X-K- sequence motif (a Ser-Tyr-Lys triad), which is implicated in the substrate-binding site [20, 21]. In the Drosophila CG12117 and CG12116 sequence alignments, CG12117 was similar to other SRs, sharing a common -G-X-X-G-X-G- sequence in the NADP(H) cofactor-binding site and a highly conserved Ser-Tyr-Lys triad motif in the substrate-binding site. However, the amino acid residues of CG12116, corresponding to glycine in the Rossmann fold and to tyrosine in the Ser-Tyr-Lys triad motif of other SRs, are replaced with proline and leucine residues, respectively. These residues, proposed to be important for substrate binding, are relatively less conserved in CG12116 (Figure 1A, black triangles and asterisks).

Figure 1:

Multiple sequence alignment of SR homologues. (A) The amino acid sequence alignment of SR homologues from five species was performed using the ClustalW2 program. Strictly conserved residues are shaded in black, and similar residues are colored gray. The NADP(H)-binding site in the N-terminus and the catalytic site in the C-terminus are represented by black triangles (▴) and asterisks (*), respectively. The representative position of the α-helix (curved lines) and β-sheet (black arrow) is shown above the alignment. The amino acid sequences are mouse SR (Mus muculus, NP_035597.2), rat SR (Rattus norvegicus, NP_062054.1), human SR (Homo sapiens, NP_003115.1), zebrafish SR (Danio rerio, NP_001019601.2), fly CG12117 (Drosophila melanogaster, NP_727265.1), and fly CG12116 (Drosophila melanogaster, NP_572481.1). The figure was drawn using the BoxShade server. (B) Phylogenetic analysis of six SRs. Protein sequences were aligned using ClustalW2, and the phylogenic tree was constructed by the MEGA 6.0 program, using the maximum-likelihood method. The genetic distances are displayed on each internal branch.

The phylogenetic relationships between SRs from various organisms were examined by constructing a rooted phylogenetic tree based on the SR protein sequences, using the maximum-likelihood method. The phylogenetic tree shows similarities between the SR protein sequences of the rat, mouse, and human. However, Drosophila CG12116 showed marked differences from other SRs, suggesting a different method of evolution for CG12116 (Figure 1B).

Comparative structural modeling of SRs from various organisms

Although there have been reports on 3D structure analyses of SR from various organisms, i.e., mouse, human, and bacteria, using X-ray crystallography [12, 13], the 3D structure of Drosophila SRs has not been reported yet. To investigate the structural differences and characteristics of Drosophila SRs, structural models of Drosophila SRs were generated, based on the structure of human SR [18, 19, 22]. Five models of CG12117 and four models of CG12116 were generated using the I-TASSER server (Table 1). The most relevant score for models predicted by I-TASSER is the confidence score (C score), with a range of −5 to +2. A model with a C score >−1.5 is considered to be a correct fold and represents higher confidence in the model [18]. Model 1 of both CG12117 and CG12116 with higher C scores was selected as the best structural model and was used for the analyses (Table 1). The model predictions were evaluated using the template modeling score (TM score) and the root mean squared difference (RMSD). The TM score is a measure of the structural similarity between the predicted model and the native structure. A TM score >0.5 indicates a model with correct topology [18]. The TM score of the respective model 1 of CG12117 and CG12116 was 0.75±0.10 and 0.78±0.10, respectively. The expected RMSD was 5.3±3.4 and 5.0±3.2 Å, respectively. Furthermore, the stereochemical quality of model 1 for Drosophila SRs was validated using the RAMPAGE server [23]. A high-quality model would be expected to have >90% of amino acid residues located in the favorable region. RAMPAGE analyses revealed that 90.3% and 86.7% of the residues were located in the favorable region, respectively, and that 6.6% and 9.2% of the residues were present in the allowed region, respectively. Furthermore, only 3.1% and 4.1% of the residues were located in the disallowed region of the Ramachandran plots of the models for CG12117 and CG12116, respectively (Figure 2). These results suggest that our predicted structural models for Drosophila SRs are reliable and of high quality.

Figure 2:

Homology model validation using the RAMPAGE server. (A) Ramachandran plot for Model 1 of CG12117. (B) Ramachandran plot for Model 1 of CG12116. RAMPAGE analyses showed that 90.3% and 86.7% of the residues are located in the favorable region and that 6.6% and 9.2% of the residues are present in the allowed region, respectively. The analyses also showed that 3.1% and 4.1% of the residues are located in the disallowed region of the Ramachandran plots of the models for CG12117 and CG12116, respectively.

A seven-stranded parallel β-sheet in the center of the SR is sandwiched by two arrays of three α-helices, and these structural characteristics are found in other short-chain dehydrogenase/reductases, including mouse SR, rat SR, human SR, and zebrafish SR [12, 24]. Although the number of stranded parallel β-sheet in the center of the molecule is small (one or two), the overall folding of Drosophila SRs CG12117 and CG12116 is similar to those of other SRs, sharing a common tertiary structure with a cofactor-binding site. The entrance to the NADP(H)-binding sites of the SRs from mouse, rat, human, and zebrafish was shaped like a narrow and deep tunnel (Figure 3A–D). However, the cofactor-binding sites of Drosophila SRs were widely shaped, with open crafts (Figure 3E and F).

Figure 3:

Potential electrostatic surfaces of NADP(H)-binding sites. The structures shown are from (A) mouse, (B) rat, (C) human, (D) zebrafish, (E) fly CG12117, and (F) fly CG12116. NADP⁺ is shown as a stick model, with the carbon, oxygen, nitrogen, and phosphate atoms indicated in gray, red, blue, and orange, respectively.

Different structural characteristics of Drosophila SRs

To find the different structural characteristics of Drosophila SRs, we analyzed the surface electrostatic distributions of the cofactor-binding sites in both Drosophila SRs and compared them to the SRs of other organisms. Interestingly, despite the high structural similarity between CG12117 and CG12116, we found some differences in the NADP(H)-binding site. Consistent with SRs from the other organisms, positively charged amino acid residues in the cofactor-binding sites were observed in CG12117 (Figure 3A–E), whereas only negatively charged amino acid residues were found in CG12116 (Figure 3F). The NADP(H)-binding site of CG12117 contains Gly13, Ser15, Arg16, Gly17, Ile18, Gly41, Arg42, Leu70, Glu71, Thr72, Asn99, Ser125, Leu151, Tyr166, Lys170, Pro196, Gly197, Val198, and Ile199. The NADP(H)-binding site of CG12116 contains Gly18, Ser20, Asn21, Ile22, Gly23, Leu45, Asp46, Leu75, Asp76, Asn110, Glu111, Gly112, Val162, Thr163, Ser164, Leu177, Lys181, Pro207, Gly208, Met210, Thr212, and Cys217. The structural model of Drosophila SRs in the ternary complex (with NADP⁺ and sepiapterin) revealed the binding site of sepiapterin to the active site of CG12117, but not to that of CG12116 (Figure 4A and B). The entrance and the substrate-binding pocket of CG12116 (Figure 4B) were smaller and narrower than those of CG12117 SR (Figure 4A, white circle). The structure makes it difficult for the substrate, sepiapterin, to gain access to the appropriate location in the substrate-binding pocket of CG12116. This result is also consistent with the sequence alignment data that showed less conserved active site residues in CG12116. The substrate-binding site of CG12117 is formed by the hydrophobic and polar residues Thr153, Leu154, Met163, Tyr166, Ala195, Gly197, and Val198. The residues that play a role in substrate and cofactor binding were confirmed by I-TASSER using the predicted structure of the ternary complex of CG12117 with NADP⁺ and sepiapterin (Figure 4C). Hydrogen bonds and electrostatic and steric interactions play a crucial role in protein-ligand interaction by increasing the binding affinity between these molecules. Therefore, the putative interactions between sepiapterin and NADP⁺, and the substrate-binding site of CG12117 were analyzed. Thr153, Leu154, and Val198 were observed to form hydrogen bonds and steric interactions with sepiapterin (Figure 4D). In addition, Thr72, Thr102, Tyr166, and Lys170 were observed to form hydrogen bonds (of lengths ranging from 2.00 to 3.43 Å) with NADP⁺. Arg16, Ile18, Leu47, Thr102, Thr153, Tyr166, Lys170, Gly97, and Arg42 formed steric and electrostatic interactions (of lengths ranging from 2.00 to 4.25 Å) with NADP⁺ (Figure 4E). These predicted structural differences could explain why one of the two Drosophila SRs, CG12117, is the only functional sepiapterin reductase in vivo. However, the enzymatic activities of the two SRs must be further investigated and compared. These results suggest the possibility that CG12116 is a pseudogene that exhibits a different substrate selectivity and a different function.

Figure 4:

Structural comparison of substrate-binding sites in Drosophila SRs. (A and B) Potential surface electrostatic distributions of the substrate-binding site of two Drosophila SRs: CG12117 (A) and CG12116 (B). Positively charged surface distributions are colored in blue and the negatively charged distributions are colored in red. NADP⁺ and biopterin are shown as stick models, with the carbon, oxygen, nitrogen, and phosphate atoms indicated in gray, red, blue, and orange, respectively. The substrate-binding pocket is indicated by the white dotted circle. (C) Model displaying the binding of the substrate, NADP⁺, and biopterin at the substrate-binding site of CG12117. Blue dashed lines indicate the hydrogen bonds. (D and E) Schematic 2D diagram of the predicted interactions between biopterin and NADP⁺, and of the substrate-binding site of CG12117. Blue, green, and red dashed lines indicate the hydrogen bonds and the electrostatic and steric interactions, respectively. The corresponding bond distances (Å) are displayed according to the color coding of the interactions. The figures were generated using the Molegro Molecular Viewer program.