Probing Oligomerized Conformations of Defensin in the Membrane

Abstract

Computational prediction and design of membrane protein–protein interactions facilitate biomedical engineering and biotechnological applications. Due to their antimicrobial activity, human defensins play an important role in the innate immune system. Human defensins are attractive pharmaceutical targets due to their small size, broad activity spectrum, reduced immunogenicity, and resistance to proteolysis. Protein engineering based modification of defensins can improve their pharmaceutical properties. Here we present an approach to computationally probe defensins’ oligomerization states in the membrane. First, we develop a novel docking and rescoring algorithm. Then, on the basis of the 3D structure of Sapecin, an insect defensin, and a model of its antimicrobial ion-channel, we optimize the parameters of our empirical scoring function. Finally, we apply our docking program and scoring function to the hBD-2 (human β-defensin-2) molecule and obtain structures of four possible oligomers. These results can be used in higher level simulations.

1. Introduction

Prediction and design of membrane protein–protein interactions have the potential to facilitate biomedical engineering for medical and biotechnological applications [1]. Computational study for weakly stable β-structures in membrane is important to engineer the biophysical properties including oligomerization state [2]. Defensins are crucial to innate immunity. They contribute to the antimicrobial action of granulocytes in the mucosa in the small intestine, in the epithelial host defense in the skin and elsewhere [3, 4]. They have antiviral activity against both enveloped and non-enveloped viruses [5], and they are important in HIV infection [6]. The oligomerization of defensins either forms of ion pores in bacterial membranes or aggregate into positively charged patches which disrupt the integrity of the lipid bilayer [7–9].

Humans express two types of defensins, α and β. Three human β-defensins: hβ-defensin−1, −2, and −3, have similar sequences, however, different properties [10]. It has been reported that several molecules can induce or enhance the production of defensins, for example, NOD2/CARD15 [11], TLR2 and TLR4 [12], and IL-12/IL-23/IL-27 [13]. Inducible hBD-2 could play a critical role in the protection of M. pneumoniae infection [14]. Human defensins also have complex roles in tumor growth, tumor monitoring, and cancer treatment [15]. hBD-2 exerts its growth suppression effect toward human melanoma cells via downregulation of B-Raf, cyclin D1, and cyclin E expression, upregulation of p21(WAF1) expression and activation of pRB [16]. hBD-2 may also control cell growth via arrest of G1/S transition and pRB activation [17]. Due to their well-established antimicrobial properties, defensins are also being investigated as therapeutics agents, especially as potential source to combat resistant bacteria. Human defensins are also attractive pharmaceutical targets due to their small size, broad activity spectrum, reduced immunogenicity and resistance to proteolysis [10, 18, 19].

Defensins perform their biological functions through three mechanisms: (1) Direct binding and modulation of host cell surface receptors and disruption of intracellular signaling which can inhibit viral replication [20]; (2) an indirect antiviral mechanism, where they function as chemokines to augment and alter adaptive immune responses; and (3) membrane disruption and pore formation [7–9]. The membrane-bound structure and topology of a human α-defensin indicate membrane pores consisting of dimers [21].

The characteristic folds of defensins are β-sheets stabilized with three disulfide bonds (Fig. 1). Their structural features, such as the helical N-terminal domains and oligomerization at the membrane surface, may modulate the efficiency of membrane insertion and selectivity for microbial or host-cell membranes. Both defensing-2 and −3 can interact with membranes as extended β-sheet platforms that present amphipathic helices for insertion into the lipid bilayer [22]. Nonetheless, many questions regarding the antiviral activities of defensins remain. Although significant mechanistic data are known for α-defensins, molecular details for β-defensins inhibition are mostly lacking [5]. The typical β-defensin action mechanism is not yet established, and one of the main challenges for the activation mechanism of the defense is the assembly in the membrane and the mechanism of membrane disruption.

Fig. 1.

Flowchart for the strategy to investigate the defensin oligomerization in membrane. The ribbon structure in left corner highlights three disulfide bonds in human defensin

Computational approaches have been employed to explore the dimerization of human β-defensin-2 [23], and to design sequences de novo based on flexible templates [24]. Here we present a computational protocol to probe possible oligomerization states of defensin in the membrane. We evaluate candidate states by a multiple protein–protein docking protocol. We focus on two β-defensins, one is the insect defensin Sapecin and another is human β-defensin-2 (hBD-2). The reason for choosing the two systems are (1) experimental information is available for possible protein–protein interactions and protein–membrane interactions for the insect Sapecin [25](Notes 1 and 2); and (2) human β-defensin-2 (hBD-2) is biologically important. Understanding the mechanism is a necessary first step to design novel antimicrobial peptides.

2. Methods

The system-specific docking protocol uses the following strategies:

2.1. Dock Sapecin Using SymmDock to Test Its Trimeric Assembly

SymmDock is a program to dock proteins and generate protein oligomers in C_n symmetries (n≥2) [26, 27]. The program can run through a webserver http://bioinfo3d.cs.tau.ac.il/SymmDock/ or a standalone version. The program can be installed in unix environment by running ./install_SymmDock.pl from the directory with SymmDock program files.

1. Download experimental pdb structures of Sapecin (1l4v) anddefensing-2 (1FD4). Prepare a pdb file with the molecule you want to dock: unit.pdb (remove hydrogen atoms if the pdb structure of the protein is obtained by NMR).

2. Create parameter file by running the script: buildParams.pl n unit.pdb, where n = the number of symmetric units and unit. pdb is the name of the PDB file of one unit. The script will create parameter file named params.txt. All the parameters are explained within the parameter file.

3. Create the Connolly surface for the molecule by running the script: runMSPoint.pl.

4. Additional input may include potential binding site residues for the molecule, which reduces running time and improve the docking quality. The format of the active site file is as follows: each line includes residue number and chain id for one residue. For example

   347 A

   348 A

   The name of the file with the binding site residues is specified in the parameter file. Add or uncomment the activeSiteParams line: activeSiteParams siteFile.txt 2 0.7.

   Binding site residues can be used in the matching and scoring stage. The integer parameter of activeSiteParams can control the usage of the binding site in the matching stage: (0) don’t use, (1) use only for first base point, (2) use for both base points. The last parameter (0.7 here) is for the scoring stage, which specifies the minimal ratio of the active site score in the results. Docking solutions with smaller ratios are discarded.

5. Running the symmetry dock program: symm_dock.Linux <params_file><output_file>

   The params_file is the parameter file “params.txt” that was previously created by “buildParams.pl”. output_file is the name of the file that will include the results, which contain the ranking and transformation matrix to create docked pdb structures.

   Each line represents one solution, with the following format:

   # |score | pen. |int. area| as1 | as2 | desolv. | Transformation

   1 | 6967 | 2.72 | 1761.00 | 0 | 0 | 461.34 | 2.04 1.07 2.82 34.36 2.80 19.23

   #—trans number

   score—geometric score

   pen.—maximal surface penetration of surface points

   int. area—buried surface area of the interface

   as1—geometric score based only on residues that were given as potential binding site for one side of the interface

   as2—geometric score based only on residues that were given as potential binding site for other side of the interface

   desolv.—DeLisi desolvation energy [28]

   Transformation—transformation matrix to generate oligomer structure: three rotational angles and three translational parameters

6. Generating docked PDB files by running: transOutput.pl output file n1 n2.

   The output file is the file created by the program earlier. n1 and n2 are the numbers of transformations to generate. For example running: “transOutput.pl output.txt 1 10” will create PDB files with the first ten transformations. The script generates a file named result.transNumber.pdb, where ‘number’ is the transformation number.

2.2. Rescore the Docking Solution Using DFIRE2 to Evaluate the Protein–Protein Interaction Energy

1. DFIRE2 is a program to calculate protein–protein interactions using knowledge-based functions [29, 30], which is available from http://sparks-lab.org/. For each solution generated from Symmetry docking, DFIRE2 energy can be evaluated by running: DFIRE dfire_pair.lib result.transNumber.pdb.

2. Refine the scoring function for defensin assembly in the membrane using experimental information as a guide. Normal docking and scoring functions are designed for interactions in aqueous solution or in the crystal complex. In order to reevaluate the docking solutions specifically for defensin in the membrane environment, we re-designed the scoring function to rank the docked defensin oligomer as:

   $$E_{\text{membrane}}={\text{binding-energy}}^{*}a+{\text{desolvation-energy}}^{*}b+{\text{interface-area}}^{*}c$$

   Where the parameters a, b, and c are to be optimized from docking of the Sapecin trimer in the membrane to fit experimental observations. The binding energy is calculated with the DFIRE2, and the desolvation energy and the interface area are calculated with the SymmDock. Based on extensive docking of the Sapecin and re-ranking of the solution to fit experimental binding modes, we obtained the optimized parameters: a = 1, b = 0.006, and c = 0.003 (Note 3).

2.3. Docking and Ranking the Human Defensin-2 Oligomers

1. Repeat the symmetry docking procedure using human β-defensin-2 (hBD-2) dimer structure as input to construct hBD2 octamers.

2. Using the optimized scoring function to evaluate the hBD-2octamer in the membrane. The flowchart of the computational approaches is in Fig. 1. (Notes 4 and 5)

4. Conclusion

Re-parameterizing symmetry dock for membrane environment can provide insight into the oligomerization structures of the membrane damaging antibacterial defensin in membrane. If combined with high level simulations in further optimization of protein structure and sequence, the integrated approach could be a valuable method in computational protein design (Notes 4 and 5).

Fig. 2.

Two best conformations with the highest ranking in optimized score function, which fit NMR observation, the numbers indicate the ranking from the initial symmetry docking

Fig. 3.

The octamer structures of hBD2 with pore conformation obtained from symmetry docking. The numbers indicate the ranking from the initial symmetry docking. The result 1 is the top ranking structure from the initial symmetry docking, and results 103, 216, 172, 74, and 193 are five top ranking structure from our reoptimized score function in membrane. It can be seen that the top re-ranked structures have larger pore sizes

Table 1.

Top 20 ranked hBD2 octamers from symmdock with new scoring function

Rank	Score (new ranking)	Result (symmetry dock ranking)
1	14.42336	result.103.pdb
2	13.41008	result.216.pdb
3	12.74822	result.172.pdb
4	12.48742	result.74.pdb
5*	12.48012	result.193.pdb
6	12.3413	result.97.pdb
7	12.02262	result.170.pdb
8	11.99962	result.251.pdb
9	11.91628	result.151.pdb
10	11.77326	result.133.pdb
11	11.53318	result.9.pdb
12*	11.4873	result.270.pdb
13	11.33686	result.175.pdb
14	11.27296	result.295.pdb
15	11.23988	result.429.pdb
16	11.14096	result.23.pdb
17	11.09774	result.255.pdb
18*	11.07944	result.87.pdb
19*	11.06372	result.124.pdb
20	10.91142	result.298.pdb

^*The conformers with are pore-forming octamers