In Silico Homology Modeling of Presenilin 2- Therapeutic Target for Alzheimer’s disease

 

Sowmya Hari1*, S. Akilashree2

1,2Department of Bio-Engineering, School of Engineering, Vels Institute of Science, Technology and Advanced Studies (VISTAS), Pallavaram, Chennai-600117, Tamilnadu, India.

*Corresponding Author E-mail: sowmya.se@velsuniv.ac.in

 

ABSTRACT:

Alzheimer’s disease is a neurodegenerative disorder which is caused by several mutations in causative proteins like Amyloid Precursor Protein (APP), Presenilin1 (PSEN1), and Presenilin2 (PSEN2). Mutations in presenilin 2 lead to the overproduction of amyloid beta peptide which gets accumulated in the brain and causes neuron death. This result in Early Onset Alzheimer’s disease (EOAD) and there are around 17 mutations in presenilin 2 results in this disease condition. There are 3 isoforms of PSEN2 which are normally found in the cells. Three dimensional structure of presenilin 2 has not been determined; hence, in this study the structure of presenilin 2 was predicted using homology modeling. The 3dimensional structures are modeled using Modeller9.20, Swiss Model and Geno3D. The structures are validated using PROCHECK. The structure of Swiss Model was found to be more reliable with fewer amino acids in the disallowed region, followed by Modeller9.0 and Geno3D. These structures will be fundamental in determining the crystal structure of presenilin 2 and for drug discovery.

 

KEYWORDS: Homology Modeling, Presenilin 2, Alzheimer’s disease, Modeller9.20, Swiss Model, Geno3D.

 

 


1.    INTRODUCTION:

Alzheimer’s disease (AD) is the neurodegenerative disorder which affects the neurons in the brain. Neurons get affected due to the accumulation of tangles and plaques which is formed by the amyloid beta peptide. These protein plaques occur in the areas of brain like hippocampus, forebrain, amygdala which are responsible for memory, learning, emotion. The neurons which use neurotransmitters like acetylcholine and glutamate are mostly affected[1,2]. AD is categorized into early-onset and late-onset AD conditions where early-onset AD is rare compared to late-onset condition. There are several proteins which are responsible for the disease condition but the real causatives are yet to be discovered. Presenilin 1, amyloid precursor protein (APP), presenilin 2 are some of the proteins which are involved in early-onset Alzheimer’s disease (EAOD)[3]. Among, them the presenilin 2 mutations are reported which involve in over production of amyloid beta peptide.

 

 

 

 

AD patients with PSEN2 mutations have a wide range in the age of onset, from 40 to 80 years[4]. Normally, presenilin 2 functions as the processing protein which processes APP and converts into small peptides like soluble APP (sAPP) and amyloid beta peptide. These mutations mislead the processing of APP by presenilin 2 and leads to the over production of Amyloid beta peptide which occur as clumps in the brain called as amyloid plaques. Amyloid plaques in the brain are responsible for the AD as it leads to neuron death. The structure of the native and mutated Presenilin 2 has yet not been reported. Protein structure can be determined by several methods like crystallography, dual polarization interferometry, nuclear magnetic resonance spectroscopy (NMR) which are highly expensive and tedious compared to Insilico methods. In silico structure prediction methods include Ab-initio methods, homology modeling, protein threading, secondary structure prediction, signal peptide prediction and transmembrane helix[5].Homology modeling/comparative modeling is the best method for the sequences which have the highest template similarity. By using Modeller9.20, Swiss Model and Geno3D, the structure prediction can be carried out[6,7]. These are the structure prediction software and online tools which rely on the reference sequence. This Insilico approach helps to predict the structure of unknown protein which can be used in further analysis. Presenilin2 is a target for diseases like AD, hence, this Insilico study on structural analysis and further docking studies of presenilin 2 can lay a foundation for drug discovery.

 

1.1 Presenlin 2:

Presenilin 2 mutations are less studied and known than other proteins. PSEN2 is the responsible gene which code for presenilin 2 and is located in the chromosome lq42.13. Normally, it is an unstable holoprotein and undergoes autocatalytic endoproteolysis to form a stable and active protein structure[8]. The native presenilin 2 protein is present in most of the cells which help in processing of proteins that passes the signals from cell membrane to the nucleus. These signals are responsible for the cell growth and the maturation of cells, hence, presenilin 2 work as the processing protein. Presenilin 2 is widely known for the role of regulates the cleaving of APP which is found in the brain and other parts of body. It involves in cleaving of APP protein into smaller peptides like soluble amyloid precursor protein (sAPP) and amyloid beta peptide. These peptides are involved in neurogenesis process both before and after birth. The presenilin 2 has two isoforms, isoform 1 is present mainly in placenta, heart and the isoform 2 is found in brain, kidney, heart, skeletal muscle[9]. There are several mutations which occur in presenilin 2 protein and changes the characteristics of the native presenilin 2. These mutations mislead the processing of APP by presenilin 2 and leads to the over production of Amyloid beta peptide which occur as clumps in the brain called as amyloid plaques. Amyloid plaques in the brain are responsible for the AD as it leads to neuron death[10,11]. It has been found that there are around 200 mutations occur in Presenilin 1 which leads to the disease condition. Presenilin 1 is found to be highly homologous to the presenilin 2 protein. The mutations occur in the protein alters the Gamma secretase enzyme which increases the production of amyloid beta peptide. It is noted that PSEN2 mutations appeared not only in AD patients but with other disorders, including breast cancer, Parkinson’s disease with dementia, dementia with Lewy bodies, front temporal dementia and dilated cardiomyopathy. About 40 mutations in presenilin 2 were identified and the studies are being performed in different aspects. Only 17 of the 38 was expected to be disease-causing mutations. Ten of the mutations are not pathogenic and the others are still unclear. Sixteen mutations are located within transmembrane domains. Cell-based studies suggest that four of these mutations, T122P, N141I, M239I, and M239V, cause an increase in the amount of Aβ peptide[12]. The mutations T122R, S130L, and M239I were found to alter calcium signaling [13,14]. Most of the mentioned mutations are discovered in African and European populations. Until now, only four missense mutations were described in Asian populations: Asn141Tyr was associated with EOAD in a Chinese Han family, Gly34Ser was found in a Japanese patient[17] and Arg62Cys and Val214Leu were described in the Korean patients[15].

 

Major disadvantages of experimental methods are the expensive techniques, time consuming and these methods are not amenable to high throughput techniques. Contrarily, In Silico methods provides the better solutions. In this paper, the Insilico analysis and homology modelling studies on Presenilin 2 – therapeutic target for AD was reported. The structures are very important to understand the function of protein and their structural features.

 

2.    MATERIALS AND METHODS:

2.1 Sequence Retrieval:

The amino acid sequence was retrieved from the UniprotKB, a Protein Database in the FASTA format to pursue with the further analysis. Table 2.1 shows the list of proteins considered in the study.

 

Table 2.1: Target protein considered for study

Target Protein

Accession No

Length

Description

PSEN2

P49810

448

Presenilin 2

 

2.2 Physico-Chemical Characterization:

Table 2.2 shows the results of the physico-chemical characteristics of Presenilin 2. The Expasy’s Prot Param Server was used to identify the properties like Molecular Weight (MW), number of amino acid residues (-R, +R), Theoretical Isoelectric Point (pI), Extinction Coefficient (EC)[18]. Aliphatic Index (AI)[19] Grand Average Hydropathy (GRAVY)[20] and Instability Index (II).

 

2.3 Functional Characterization:

The transmembrane regions are identified by the SOSUI server which tells the length and position of the regions. Disulphide bonds are the important parameter to be analyzed for the functional characterization of the protein. Table 2.3 shows the results from the SOSUI server and the Table 2.4 shows the disulphide bond prediction using the DiANNA web server.


 

Table 2.2: Physico chemical characterization of Presenilin 2

Target protein

Accession No

Length

MW

-R

+R

pI

EC

AI

GRAVY

II

Presenilin 2

P49810

448

50140.39

53

28

4.51

77280

103.75

0.299

47.41

 


 

Table 2.3: Transmembrane region results using SOSUI server

S. No

N-Terminal

Transmembrane Region

C-Terminal

Type

Length

1.

86

KHVIMLFVPVTLCMIVVVATIKS

108

Primary

23

2.

138

SVLNTLIMISVIVVMTIFLVVLY

160

Primary

23

3.

167

FIHGWLIMSSLMLLFLFTYIYLG

189

Primary

23

4.

198

AMDYPTLLLTVWNFGAVGMVCIH

220

Secondary

23

5.

238

LMALVFIKYLPEWSAWVILGAIS

260

Primary

23

6.

285

NEPIFPALIYSSAMVWTVGMAKL

 

 


Table 2.4: Predicted Disulphide bonds results from DiANNA web server

Predicted Bonds

14 - 400

SEEEVCDERTS - ILIGLCLTLLL

31 - 269

PTPRSCQEGRQ - LVAVLCPKGPL

65 - 164

PDRYVCSGVPG - LYKYRCYKFIH

218 - 391

AVGMVCIHWKG - NTTLACFVAIL

Predicted Connectivity

1-9, 2-7, 3-5, 6-8

 

2.4 Secondary structure predication:

The secondary structure prediction for Presenilin 2 protein was carried out by SOPMA server[21] which provides the details about the alpha helix, Pi helix, beta strands, and coils in the protein secondary structure. The results from the SOPMA server is shown in Table 2.5

 

Table 2.5: secondary structure prediction results from SOPMA server

Target Protein

Presenilin 2

Secondary structure

Alpha helix

39.51%

310  helix

0.00%

Pi helix

0.00%

Beta bridge

0.00%

Extended strand

14.29%

Beta turn

2.23%

Bend region

0.00%

Random coil

43.97%

Ambiguous states

0.00%

Other states

0.00%

 

2.5    Model Building and Validation:

The modeling of Presenilin 2 structures was carried out by the different homology modeling programs, Modeller9.20[17] Swiss Model[18] and Geno3D[22]. WHATIF web tool[24] was used to obtain the complete structure of the template PDB file which was further used to build the models. The models obtained from the results of these homology programs were further introduced to the Swiss PDB viewer[25] for the energy minimization process. The validation of the obtained models was carried out by PROCHECK which also includes the Ramachandran plot analysis [23] for the position of residues. The Modloop web server was used to remove the residues in the disallowed regions and the finalized models are viewed using the PYMOL software. The results from the Ramachandran plot are shown in Table 2.6.

 

Table 2.6: Ramachandran plot results using PROCHECK comparing Modeller9.20, Swiss Model, Geno3D models

Server

Presenilin 2

Modeller9.20

Residues in most favored regions

91.0%

Residues in additional allowed regions

7.2%

Residues in generously allowed regions

0.8%

Residues in disallowed regions

1.0%

Swiss Model

Residues in most favored regions

83.6%

Residues in additional allowed regions

13.3%

Residues in generously allowed regions

2.8%

Residues in disallowed regions

0.3%

Geno3D

Residues in most favored regions

82.6%

Residues in additional allowed regions

13.2%

Residues in generously allowed regions

2.1%

Residues in disallowed regions

2.1%

 

3.    RESULTS AND DISCUSSION:

RCSB-PDB, protein structural database was used for confirming that the Presenilin 2 protein didn’t have any hits to carry out the further levels of project. For the In Silico analysis of the target protein, the FASTA sequence of Presenilin 2, Accession no: P49810 was retrieved from the Uniprot KB and been used as the input sequence for the structural and functional analysis of protein. Table 2.2 shows the results for the physic chemical characterization of the protein which includes the calculated values for Molecular Weight (MW), number of amino acid residues (-R, +R), Theoretical Isoelectric Point (pI), Extinction Coefficient (EC), Aliphatic Index (AI), Grand Average Hydropathy (GRAVY) and Instability Index (II). The molecular weight of the protein was found to be 50.1KDa with the chemical formula C2302H3554N540O655S27 and the number of positive and negative residues present in the protein sequence was 28 and 53 respectively. The pI can be defined as the point at which the overall charge of protein is zero and attains the neutral stage. The isoelectric point is the specific pH value at which protein carries no charge. This also very helpful in formulizing the buffers during isoelectric focusing process and to know the nature of the protein. According to the results, the pI value of Presenilin 2 was 4.51 (<7) and considered as the acidic protein. The extinction coefficient is referred as the measure of absorption of light and can be measured in different wavelengths. The proteins are normally measured in 280nm as they absorb the light stronger than other molecules. The higher values indicate the presence of high concentration of amino acids like tyrosine, tryptophan, and cysteine. The instability index is the measure of the stability of protein in the test tube and normally the range is less than 40 for the stable protein and the higher values describe the instability of the particular protein. The instability can be explained as the occurrence of certain dipeptides which decreases the overall stability of the protein. Here, the instability index of Presenilin 2 was 47.41 which makes the protein to be unstable. The aliphatic index explains about the thermal stability of the protein for a wide range of temperature. The aliphatic index is the measure of aliphatic chains of amino acids like valine, leucine, isoleucine, alanine in the protein which influences the thermal stability of protein. The higher the AI, higher the thermal stability. The AI of Presenilin 2 was 103.75 which is considered higher and the protein is highly inflexible. The interaction with the water can be computed by the GRAVY value which is calculated as the sum of hydropathy values of amino acids divided by total number of amino acid residues in the whole protein sequence. The GRAVY value for Presenilin 2 was found to be 0.299 which explains the slightly hydrophobic nature of protein.

 

Table 2.3 provides the results from the SOSUI server which was used for the functional characterization of target protein. The SOSUI server predicts the transmembrane regions in the protein sequence which explains the presence of hydrophobic residues in the specific regions. The Presenilin 2 sequence has 6 trasmembrane regions which validates the results of computed GRAVY value in the physic chemical characterization. The disulphide bonds are predicted by the DiANNA web server which provided the results with the position of linkage and the specific sequence. The server predicted 5disulphide bonds and their connectivity in the amino acid sequence. Table 2.4 results conclude that the Presenilin 2 structure have disulphide bonds.

 

The secondary structure prediction was carried out using the SOPMA web server which provided the results of number of helixes, strands and coils in the protein structure. This explains the position of residues in the structure of protein where the alpha helix contributes to 39.51%, extended strand 14.29%, beta turn 2.23% and coil contribute to 43.97%. This explains that the secondary structure is dominated by the presence of random coils followed by alpha helix and extended strands. The parameters used in the analysis was similarity threshold: 8, window width: 17, number of states: 4. The results are shown in the Table 2.5.

 

Modelling of Presenilin 2 was carried out by three programs, Modeller 9.20, Swiss Model, Geno3D which generated different models of the protein and are validated by PROCHECK. Modeller9.20 is used for homology or comparative modeling of protein three-dimensional structures[16].Modeller implements comparative protein structure modeling by satisfaction of spatial restraints[17]. The Modeller 9.20 runs using the Python language and the commands are provided to get the structures of query protein (Presenilin 2). NCBI-BLAST was performed using the target protein sequence to obtain the similar sequence to use the same as the template for Modeller models. Modeller provides the DOPE scores which are used in identifying the higher accurate predictions where lower the DOPE value higher the quality of models. Presenilin 2 structures were also be predicted by using Swiss Model and Geno3D web interface which have the auto template search and provide the templates with high identity scores and develop the models accordingly. Swiss Model uses the QMEAN score which indicates the good agreement between the model structure and experimental structures of similar size. 0 to -4 score provides with the good quality models and below that indicates the low quality models[18]. Ramachandran plots were used to locate the residues in the protein sequence and the residues which are positioned in the disallowed regions are removed and relocated through the Modloop server. The obtained models were viewed using the PYMOL software. Table 2.6 shows the difference between the Ramachandran plot results using 3 different modeling programs. The models from Swiss Model were found to be more reliable with fewer amino acids in the disallowed region, followed by Modeller9.0 and Geno3D based on the validation results using PROCHECK. The models which are provided are compared with the template sequence where blue represents the template (Presenilin 1) and the red represents the Presenilin 2. Figure 3.1 provide the Swiss Model results and the Figure 3.2 represents the corresponding Ramachandran plot for the model.

 

 

Figure 3.1: Swiss Model result for Presenilin 2 (Blue: template, Red: Presenilin 2)

 

 

Figure 3.2: Ramachandran plot results after the Modloop submission for the above model

4.    CONCLUSION:

The 3 dimensional structure of presenilin 2 was generated which is useful to study protein-ligand interactions to develop drugs to target the mutated protein- a target in AD. The template 5FN2_B (Presenilin 1) was chosen which has 86% identity score with query sequence. The structure obtained was validated using Procheck. The physico-chemical characterization was analyzed using Expasy’s PratParam server which provided the results for the instability index, aliphatic index, hydropathicity and extinction coefficient. The functional analysis was carried out by SOSUI server which provided the transmembrane regions and DiANNA server provided the disulphide bond predictions. The secondary structure predictions by SOPMA server provide that the structure is dominated by the coils followed by alpha helix and strands. The models of Presenilin 2 was generated using the Modeller9.20, Swiss Model and Geno3D.  Most of the models which are predicted by Modeller and Swiss Model have fewer residues in favored and allowed regions and models from Swiss Model showed less than 1% of residues in the disallowed regions which clearly shows the good quality of the obtained models. Future perspectives will be to study the dynamics of each atoms present in the structure and the other stimulations can also be considered. The mode of inhibitions can be studied with the predicted models using In silico approach as it be the therapeutic target for AD. Few other mutated models can also be predicted specifically and docked with the ligands to check the activity of certain drugs.

 

5.       ACKNOWLEDGMENTS:

I sincerely thank the Vels Institute of Science Technology and Advanced Studies management for their support towards the successful completion of the research work.

 

6.       CONFLICTS OF INTEREST:

The author declare that no conflict of interest.

 

7.       REFERENCES:

1.      Dickson DW. Neuropathological diagnosis of Alzheimer’s disease: a perspective from longitudinal clinicopathological studies. Neurobiology of aging. 1997 Jul 1, 18(4): S21-6.

2.      Braak H, Braak E. Evolution of neuronal changes in the course of Alzheimer’s disease. In Ageing and dementia 1998 (pp. 127-140). Springer, Vienna.

3.      Żekanowski C, Styczyńska M, Pepłońska B, Gabryelewicz T, Religa D, Ilkowski J, Kijanowska-Haładyna B, Kotapka-Minc S, Mikkelsen S, Pfeffer A, Barczak A. Mutations in presenilin 1, presenilin 2 and amyloid precursor protein genes in patients with early-onset Alzheimer's disease in Poland. Experimental neurology. 2003 Dec 1, 184(2): 991-6.

4.      Youn YC, Bagyinszky E, Kim H, Choi BO, An SS, Kim S. Probable novel PSEN2 Val214Leu mutation in Alzheimer’s disease supported by structural prediction. BMC neurology. 2014 Dec, 14(1):105.

5.      Hardy J. A hundred years of Alzheimer's disease research. Neuron. 2006 Oct 5;52(1):3-13.

6.      Eswar N, Webb B, Marti‐Renom MA, Madhusudhan MS, Eramian D, Shen MY, Pieper U, Sali A. Comparative protein structure modeling using Modeller. Current protocols in bioinformatics. 2006 Sep 1, 15(1): 5-6.

7.      Arnold K, Bordoli L, Kopp J, Schwede T. The SWISS-MODEL workspace: a web-based environment for protein structure homology modelling. Bioinformatics. 2006 Jan 15, 22(2):195-201.

8.      Rademakers R, Cruts M, Van Broeckhoven C. Genetics of early-onset Alzheimer dementia. The Scientific World Journal. 2003, 3:497-519.

9.      Annaert WG, Levesque L, Craessaerts K, Dierinck I, Snellings G, Westaway D, George-Hyslop PS, Cordell B, Fraser P, De Strooper B. Presenilin 1 controls γ-secretase processing of amyloid precursor protein in pre-Golgi compartments of hippocampal neurons. The Journal of cell biology. 1999 Oct 18, 147(2):277-94.

10.   De Strooper B, Iwatsubo T, Wolfe MS. Presenilins and γ-secretase: structure, function, and role in Alzheimer disease. Cold Spring Harbor perspectives in medicine. 2012 Jan 1, 2(1):a006304.

11.   Citron M, Westaway D, Xia W, Carlson G, Diehl T, Levesque G, Johnson-Wood K, Lee M, Seubert P, Davis A, Kholodenko D. Mutant presenilins of Alzheimer's disease increase production of 42-residue amyloid β-protein in both transfected cells and transgenic mice. Nature medicine. 1997 Jan;3(1):67.

12.   Walker ES, Martinez M, Brunkan AL, Goate A. Presenilin 2 familial Alzheimer's disease mutations result in partial loss of function and dramatic changes in Aβ 42/40 ratios. Journal of neurochemistry. 2005 Jan;92(2):294-301.

13.   Zatti G, Burgo A, Giacomello M, Barbiero L, Ghidoni R, Sinigaglia G, Florean C, Bagnoli S, Binetti G, Sorbi S, Pizzo P. Presenilin mutations linked to familial Alzheimer's disease reduce endoplasmic reticulum and Golgi apparatus calcium levels. Cell calcium. 2006 Jun 1, 39(6): 539-50.

14.   Zatti G, Ghidoni R, Barbiero L, Binetti G, Pozzan T, Fasolato C, Pizzo P. The presenilin 2 M239I mutation associated with familial Alzheimer's disease reduces Ca2+ release from intracellular stores. Neurobiology of disease. 2004 Mar, 15(2):269-78.

15.   Bai YF, Tian J, Quan WX, Maeda K. Association of Mutations of Presenilin-2 Gene and Sporadic Alzheimer's Disease. Journal of China Medical University. 2011 Apr, 40(4): 357-9.

16.   Martí-Renom MA, Stuart AC, Fiser A, Sánchez R, Melo F, Šali A. Comparative protein structure modeling of genes and genomes. Annual review of biophysics and biomolecular structure. 2000 Jun, 29(1): 291-325.

17.   Šali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. Journal of molecular biology. 1993 Dec 5, 234(3):779-815.

18.   Gill SC, Von Hippel PH. Calculation of protein extinction coefficients from amino acid sequence data. Analytical biochemistry. 1989 Nov 1, 182(2): 319-26.

19.   Guruprasad K, Reddy BB, Pandit MW. Correlation between stability of a protein and its dipeptide composition: a novel approach for predicting in vivo stability of a protein from its primary sequence. Protein Engineering, Design and Selection. 1990 Dec 1, 4(2): 155-61.

20.   Gasteiger E, Hoogland C, Gattiker A, Wilkins MR, Appel RD, Bairoch A. Protein identification and analysis tools on the ExPASy server. In The proteomics protocols handbook 2005 (pp. 571-607). Humana press.

21.   Geourjon C, Deleage G. SOPMA: significant improvements in protein secondary structure prediction by consensus prediction from multiple alignments. Bioinformatics. 1995 Dec 1, 11(6): 681-4.

22.   Combet C, Jambon M, Deleage G, Geourjon C. Geno3D: automatic comparative molecular modelling of protein. Bioinformatics. 2002 Jan 1, 18(1): 213-4.

23.   Ramachandran GT, Sasisekharan V. Conformation of polypeptides and proteins. In Advances in protein chemistry 1968 Jan 1 (Vol. 23, pp. 283-437). Academic Press.

24.   Vriend G. WHAT IF: a molecular modeling and drug design program. Journal of molecular graphics. 1990 Mar 1;8(1):52-6.

25.   Guex N, Peitsch MC. SWISS‐MODEL and the Swiss‐Pdb Viewer: an environment for comparative protein modeling. electrophoresis. 1997;18(15):2714-23.

 

 

 

 

 

 

Received on 08.02.2019         Modified on 02.03.2019

Accepted on 30.03.2019         © RJPT All right reserved

Research J. Pharm. and Tech. 2019; 12(7):3405-3409.

DOI: 10.5958/0974-360X.2019.00575.4