Bioinformatics Approaches for the Molecular Characterization and Structural Elucidation of a Hypothetical Protein of Aedes albopictus  ^†

Abstract

The most critical issues in computational biology are characterizing and predicting uncharacterized proteins’ secondary and tertiary structures from their uploaded amino acid sequences in databases. Aedes albopictus (A. albopictus), sometimes referred to as the Asian tiger mosquito or forest mosquito and the carrier of dengue-like diseases, has many proteins, many of which are still poorly understood. The current work aims at elucidating the physicochemical properties and structures of the as-yet-uncharacterized A. albopictus protein AEW48448.1. ExPASy Protaram, CD Search, SOPMA, PSIPRED, and other advanced computerized tools were used following the standard flowchart for characterizing a hypothetical protein to ascertain the roles and structures of AEW48448.1. After identifying the protein’s secondary and tertiary structures, the structures were evaluated for quality using tools like PROCHECK and the ProSA-web. Later, the active site was also discovered using CASTp v3.0. The protein is more stable because it has a higher aliphatic index value and more negatively charged than positively charged residues. The modeling of the proteins’ 2D and 3D structures using multiple bioinformatics tools confirmed that they had domains, indicating that they were functional proteins involved in the host’s antiviral, cytokine, and interferon production pathways. Additionally, the protein was revealed to have active regions where ligands may bind. This work aims at elucidating the characteristics and structures of an uncharacterized A. albopictus protein that may serve as a therapeutic target for the creation of antiviral candidates and vaccines.

1. Introduction

A. albopictus spreads the arboviruses Dengue (DENV), Zika virus (ZIKV), and Chikungunya (CHIKV) [1]. In temperate and tropical parts of the planet, this mosquito is a strong and very invasive species [2]. Understanding the basic biology of A. albopictus is crucial for preventing its aggressive spread, as its exceptional adaptability has led to its rapid expansion and global public concern [3]. Studies estimate that nearly 3.9 billion people worldwide are at risk of contracting DENV, putting about half of the world’s population at risk. Despite the lack of a specific cure for severe dengue, early discoveries and access to quality medical care significantly reduce its fatality rates [4]. The structure of cells and organisms is one of the many roles that proteins carry out within living things. They also participate in several crucial processes in vivo through interactions with other molecules [5]. Uncovering the biological characteristics and activities of these uncharacterized proteins from various animals is currently a frequent activity in the domain of bioinformatics because there are millions of proteins that are still uncharacterized [6,7]. A. albopictus contains several functional proteins, many of which are still unknown or are poorly understood [8,9]. Developments in computational biology have led to the development of numerous platforms and techniques for predicting protein structure, binding sites, and biological activity [10,11]. Protein studies use bioinformatics techniques to classify novel domains, evaluate 3D structural conformations, and determine function [12]. Furthermore, this comprehensive understanding can provide effective pharmacological tactics for developing potential treatments for a variety of disorders [13].

2. Materials and Methods

2.1. Selection of the Hypothetical Protein and Sequence Retrieval

We obtained the amino acid (aa) sequence for cytochrome c oxidase subunit I, partial mitochondrion protein (A. albopictus), in FASTA format from the NCBI database [14].

2.2. Physicochemical Properties Analysis

Using the ExPASy server ProtParam tool, we found the theoretical isoelectric point (pI), the instability index, the aliphatic index, the GRAVY (which measures how hydrophobic or hydrophilic a protein is), the extinction coefficients, and the amino acid sequence composition [15].

2.3. Identification and Validation of the Secondary Structure

We applied the self-optimized prediction method with alignment (SOPMA) to anticipate secondary structural elements [16].

2.4. Three-Dimensional Structure Prediction and Validation

We predicted the three-dimensional structure of the chosen protein using the SWISS-MODEL [17] and AlphaFold [18]. Furthermore, we used the PROCHECK software of the SAVES program (v.6.1) to structurally validate the modeled 3D protein structure [19].

2.5. Ligand Binding Site or Active Site Determination

We used the DeepSite tool [20], PrankWeb server [21], and CASTp v.3.0 server [22] to predict the active sites of the modeled protein.

2.6. Pathogenecity Prediction

We used CDPred to predict the celiac disease-causing peptides from the protein’s FASTA sequences [23]. We also utilized the FASTA sequence to construct the protein’s 3D structure and detect disease-causing protein mutations. We utilized the FASTA sequence to forecast and scrutinize the location of a protein sequence mutation. Additionally, we made use of a variety of services, including Polyphen-2, SIFT, predictSNP, MAPP, PhD-SNP, and SNAP [24]. The Align-GVGD program also used substitution, predicted by predictSNP, and the FASTA format of the amino acid (AA) sequence to calculate GV and GD [25].

3. Results and Discussion

3.1. Protein Sequence Retrieval

The amino acid (aa) sequence of the cytochrome c oxidase subunit I, partial mitochondrial protein of A. albopictus (AEW48448.1), was obtained from the NCBI database. We modeled the tertiary structure of the protein using the 152 amino acid-long protein sequence (Figure 1).

3.2. Identification of the Physicochemical Properties

We used ProtParam ExPASy to measure physicochemical parameters using the amino acid sequence of the protein AEW48448.1 found in A. albopictus. We retrieved the sequence in FASTA format. The instability index, at 30.29, indicates the protein’s stability with an index smaller than 40 [26]. The protein’s hypothetical pI (pH 6.10) indicates that it is in a weak acidic form [5]. There is 56,800.02 Dalton in the molecular weight, 110.08 in the aliphatic index, 30.29 in the instability index, and 0.731 in the GRAVY, respectively.

The protein has a higher aliphatic index value of 110.08, which is a positive indicator of greater thermostability over a broad temperature range [27]. A GRAVY index value of −0.731 revealed the protein’s hydrophilic nature and its potential for increased water interaction [19]. In vitro, Escherichia coli had an estimated half-life of >10 h; yeast had an estimated half-life of >20 h; and mammalian reticulocytes had an estimated half-life of 1.9 h.

3.3. Identification and Validation of the Secondary Structure

The cytochrome c oxidase subunit I partial (mitochondrion) protein (A. albopictus) predicted secondary structural components using the SOPMA program’s default parameters (window width of 17, number of states of 4, and similarity threshold of 8) (Figure 2).

3.4. The Three-Dimensional Protein Structure Anticipation and Assessment

We used the SAVESv6.1 program’s PROCHECK tool to assess the modeled structures from AlphaFold and SWISS-MODEL servers [5]. We used the Ramachandran plot to assess the quality of the predicted model (by SWISS-MODEL), finding 404 (92.7%) of the total 510 residues in the core, 6.9% in the additional allowed regions, and 0.2% in the generously allowed regions (a, b, l, and p). There was a total of 436 residues that were not glycine or proline. There were three end-residues that were not Gly or Pro, and 44 and 27 glycine and proline residues out of a total of 510 residues (Figure 3). We also examined the protein residues (structure predicted by AlphaFold) in the most favored regions (94.5%), additional allowed regions (5.2%), generously allowed regions (0.0%), and disallowed regions (0.2%) (Figure 3). The quality assessment analysis showed that AlphaFold’s predicted structure outperformed the SWISS-MODEL in terms of quality.

3.5. Ligand Binding Site or Active Site Determination

Ligand binding cavity prediction is particularly significant for drug design because it cleverly utilizes the protein’s geometric, chemical, and evolutionary properties. One prediction tool, DeepSite, can predict the protein’s active site where a drug or protein can bind to reduce protein toxicity [28]. DeepSite predicted that the AEW48448.1 protein contained three cavities (Figure 4A). Another ligand-binding tool, PrankWeb, predicted 12 pockets in the given protein (Figure 4C). CASTp could detect regions on proteins, outline them, identify their dimensions, and compute their areas [5]. The highest active sites were located between the modeled protein’s 800.210 and 1784.976 areas (Figure 4B).

3.6. Pathogenecity Prediction

CDPred generates a tabular output of the peptides from the proteins that cause celiac disease, making it possible to predict disease-associated peptides. The table shows the amino acid sequence, the start position, the end position, and the physicochemical properties of the celiac disease-causing peptide from the protein sequence [23].

Table 1. Prediction of disease associated peptides and its properties.

Sequence	Start	End	Score	Hydrophobicity	Hydropathicity	Hydrophilicity	Charge	Mol wt.
MIFFMVMPIMIGGFGNWLVP	63	82	0.34	0.34	1.65	−1.31	0	2301.25
FFMVMPIMIGGFGNWLVPLM	65	84	0.32	0.33	1.61	−1.31	0	2301.25
MPIMIGGFGNWLVPLMLGAP	69	88	0.32	0.27	1.2	−1.03	0	2114.98
FPRMNNMSFWMLPPSLTLLL	92	111	0.32	0.05	0.54	−0.89	1	2409.27
MNNMSFWMLPPSLTLLLSSS	95	114	0.32	0.08	0.58	−0.86	0	2270.02
NNMSFWMLPPSLTLLLSSSM	96	115	0.32	0.08	0.58	−0.86	0	2270.02
NMSFWMLPPSLTLLLSSSMV	97	116	0.4	0.13	0.97	−0.95	0	2255.05
MSFWMLPPSLTLLLSSSMVE	98	117	0.33	0.14	0.97	−0.81	−1	2270.06
SFWMLPPSLTLLLSSSMVEN	99	118	0.36	0.09	0.7	−0.73	−1	2252.97
FWMLPPSLTLLLSSSMVENG	100	119	0.33	0.11	0.72	−0.75	−1	2222.95
FIGVNLTFFPQHFLGLAGMP	416	435	0.32	0.21	0.98	−1.05	0.5	2206.95
TPSFPMQLSSSIEWYHTLPP	479	498	0.42	−0.02	−0.33	−0.59	−0.5	2318.92

summarizes the protein’s results (sequence id: AEW48448.1).

Predicting SNP, MAPP, PhD-SNP, Polyphen-2, SIFT, and SNAP makes it simple to identify pathogenic SNVs. Compared to other mutation prediction servers, Mutation Assessor has the highest sensitivity for all substitutions evaluated by the predict SNP server (except H41Q).

Table 2. Prediction of disease-related mutation.

Mutants	GV	GD	Prediction 1
M63P	0.00	86.59	Class C65
F65M	0.00	28.53	Class C25
M69P	0.00	86.59	Class C65
F92L	0.00	21.82	Class C15
M95S	0.00	134.86	Class C65
N96M	0.00	141.15	Class C65
N97V	0.00	132.88	Class C65
M98E	0.00	126.08	Class C65
S99N	0.00	46.24	Class C45
F100G	0.00	153.13	Class C65
F416P	0.00	113.73	Class C65
T479P	0.00	37.56	Class C35

¹ GD >= 65 + Tan(10) × (GV^2.5) => Class C65 <=> most likely, GD >= 55 + Tan(10) × (GV^2.0) => Class C55, GD >= 45 + Tan(15) × (GV^1.7) => Class C45. GD >= 35 + Tan(50) × (GV^1.1) => Class C35. GD >= 25 + Tan(55) × (GV^0.95) => Class C25, GD >= 15 + Tan(75) × (GV^0.6) => Class C15, else (GD < 15 + Tan(75) × (GV^0.6)) => Class C0 <=> less likely.

lists the values for the measured parameters. All parameters indicate that M63P, F65M, M69P, M95S, N96M, N97V, M98E, S99N, F100G, and F416P have high predictive accuracy, while F92L is likely to have high variable apparent accuracy when compared to T479P. We ranked all parameters from highly deleterious to less predicted, as shown in Table 2.

The Align-GVGD program’s GD and GV values are crucial as they indicate the distance between the mutant amino acid and the permitted variation, as indicated by GV [29]. We predicted all missense mutants in the protein to be neutral, harmful, or unclassified using the following Align-GVGD criteria: If the substitute falls into the box, then GD = 0. If not, then GD is higher than 0. Therefore, according to the MSA, GD is a measure of the biochemical difference between the mutant and the observed variation at that location [30]. Out of the 12 mutants we know of, M63P, M69P, M95S, N96M, N97V, M98E, F100G, and F416P are very likely to go through missense substitutions, while S99N, T479P, F65M, and F92L have very little chance (

Table 3. Prediction of missense substitutions.

Mutants	predictSNP	MAPP	PhD-SNP	Polyphen-2	SIFT	SNAP	PANTHER	Ranging
M63P	86.91%	87.71%	85.82%	56.23%	79.28%	88.52%	76.65%	1
F65M	86.91%	78.34%	81.73%	63.43%	79.28%	72.04%	76.00%	1
M69P	86.91%	87.71%	81.73%	81.14%	79.28%	84.85%	65.27%	1
F92L	60.55%	75.11%	73.26%	67.64%	79.28%	62.21%	61.08%	2
M95S	86.91%	84.18%	77.34%	55.08%	79.28%	72.04%	69.00%	1
N96M	86.91%	84.18%	60.80%	81.14%	79.28%	84.85%	69.86%	1
N97V	86.91%	77.12%	67.62%	81.14%	79.28%	84.85%	66.07%	1
M98E	86.91%	91.38%	87.52%	67.52%	79.28%	84.85%	68.66%	1
S99N	86.91%	76.54%	60.80%	43.12%	79.28%	80.51%	71.86%	1
F100G	86.91%	81.93%	60.80%	81.14%	79.28%	84.85%	78.00%	1
F416P	86.91%	87.71%	85.82%	64.97%	79.28%	80.51%	74.45%	1
T479P	74%	68%	51%	87%	46%	83%	63%	3

).

4. Conclusions

Using advanced bioinformatics methods to characterize a protein is another novel challenge, similar to various system biology activities. Our research aims to shed light on the physicochemical properties, chemical structures, and biological roles of the putative protein AEW48448.1 from A. albopictus. The protein contains amino acids that have more negatively charged residues, and it is more temperature-stable due to its high aliphatic index and low instability index. Using a variety of bioinformatics techniques, we modeled the protein’s secondary structure and confirmed that it included a domain, indicating its functional nature. Numerous servers verified the protein’s precise 3D structure through tertiary structure prediction. Furthermore, the study discovered an active site in the protein where ligands could bind. More research is needed on the protein to discover innovative treatment options for dengue, which is a result of A. albopictus’s targeting of the protein.