Analysis of the Natural Killer Cell Receptors using PDBe Tools

Jawahar Swaminathan, Ph.D.

Introduction

Natural killer (NK) cells are a distinct lineage of lymphocytes that display cytotoxic activity without a previous encounter with an infected cell. NK cells are a major component of the innate immune system and play an important role in the rejection of tumors and cells infected by viruses. The usual mechanism of action of NK cells involves the secretion of cytoplasmic granules of proteins called perforin (a cytolytic protein that inserts into the target cell plasma membranes) and granzyme B (an exogenous serine protease), eventually leading to the death of the target cell by . Historically, these cells were classified as natural killer cells since it was thought that these cells did not require activation to act upon cells that were deficient in MHC I markers on the cell surface. However, it is now known that this is now true and NK cells have precise activation and regulation mechanisms.

In order for NK cells to defend the body against viruses and other pathogens, they require mechanisms that enable the determination of whether a cell is infected or not. The exact mechanisms remain the subject of current investigation, but recognition of an "altered self" state is thought to be involved. To control their cytotoxic activity, NK cells possess two types of surface receptors: activating receptors and inhibitory receptors. Most of these receptors are not unique to NK cells and can be present in other T cells subsets as well. Please see http://en.wikipedia.org/wiki/NK_cell for a detailed explanation of the mechanism of action of NK cells.

Purpose

There are over 1700 structures of proteins involved in the immune system. The majority of structures of immune interest are those of antibodies and similar molecules that contain the immunoglobulin fold. This tutorial will analyse the family of NK cell receptors using a variety of PDBe tools and services that are all available on the internet from the PDBe Portal at http://www.ebi.ac.uk/pdbe. It is hoped that at the end of this exercise, the user will have learned how to use the PDBe services and tools for searching and retrieving information from the PDBe using selected search criteria, exploring /inhibitor binding sites, understanding and evaluating structures on the basis of quality, as well as appreciate the concepts of protein folds and quaternary structure assemblies.

Requirements

a) A computer running any operating system connected to the internet. b) Any modern web browser such as Internet Explorer/Firefox/Mozilla c) Java Run-Time environment 1.5 or higher. d) Rasmol/Raswin with mime-type chemical/x-pdb set in the browser to start rasmol/raswin when requested. e) If running firefox, then the installation of Biobar (http://biobar.mozdev.org/) may help aid searching PDBe and other databases.

Natural Killer Cell Receptors

Natural Killer cell receptors are differentiated by structures and can be divided into the following 4 types (from the Wikipedia).

1. CD94 : NKG2 (heterodimers) — a C-type lectin family , conserved in both rodents and primates and identifies non-classical (also non-polymorphic) MHC I molecules like HLA E. Though indirect, this is a way to survey the levels of classical (polymorphic) HLA molecules, however, because expression of HLA-E at the cell surface is dependent upon the presence of classical MHC class I leader peptides. PDB entry: 1B6E. 2. Ly49 (homodimers) — a relatively ancient, C-type lectin family receptor; are of multigenic presence in mice, while humans have only one pseudogenic Ly49; the receptor for classical (polymorphic) MHC I molecules. PDB entry: 1JA3. 3. KIR (killer cell immunoglobulin-like receptors) — belong to a multigene family of more recently- evolved Ig-like extracellular domain receptors; are present in non-rodent primates; and are the main receptors for both classical MHC I (HLA-A, HLA-B, HLA-C) and also non-classical HLA-G in primates. Some KIRs are specific for certain HLA subtypes. PDB entry: 1B6U. 4. ILT or LIR (leucocyte inhibitory receptors) — are recently-discovered members of the Ig receptor family. PDB entry: 2DYP.

Ly49 NK Cell Receptor

The Ly49 family of NK receptors (Ly49A through W), which includes both inhibitory and activating receptors, are homodimeric type II transmembrane glycoproteins, with each subunit composed of a C- type lectin-like domain tethered to the membrane by a stalk region. Due to the constraints imposed by structure determination methods, all 3-d structures of this receptor in the Protein Data Bank are of the extra-cellular lectin domain only. Lectins are carbohydrate recognition proteins, found in almost all species. They play an important role in cell-cell recognition, cell communication, transport of various molecules, such as sugars, and in the immune system. Lectins are classified according to their structure and the nature of the carbohydrate that they bind. We will be looking here at a particular family of lectins, the C-type lectins. This group is so named because they bind sugars in a calcium-dependent manner. Before they are able to bind sugars, this group of lectins must first bind calcium, which stabilises the conformation of the sugar binding site. TUTORIAL

Initial Search

Let us begin by searching for all structures in the PDBe that contain the keyword “Natural Killer Cell”. In order to do this, navigate to the main page for the PDBe at http://www.ebi.ac.uk/pdbe and click on the PDBeLite link as shown below.

This will open up the PDBeLite search system. On the form, add the search term “Natural Killer Cell” under Text Search and check the box above to search only structures determined by X-ray crystallography. The other search options can be used to fine tune results. Click on “Start Search”.

This will execute a search on the PDBe database and come back with a result page of entries that contain the terms Natural Killer Cell.

The first link on each line will take you to a summary “atlas” page for the structure identified by the 4 letter code on the link. The second link will open up the AstexViewer to visualize the structure in detail. From the set of results above, click on 1JA3. This will open up the Atlas page for this entry.

The Summary Pages

The summary pages contains at-a-glance information about the structure including sequence information as well as any bound molecules present in the structure. Let us look at some of the other information contained in the “Atlas” pages for this entry.

Visualisation Page

The visualization pages for every entry in the PDBe database contains links to 3-D viewing software as well as static images of the structure.

Click on the AstexViewer Link on the page as shown above to start up the java-based visualization software to view 1JA3.

The AstexViewer is a complete structure visualization and analysis tool. It has built-in graphs showing structure quality as well as a sequence browser. In order to get an image like the one above, follow the sequence: Color -> Chain, Protein -> Cartoon, Protein -> Line, Graphs -> Rama, Graphs -> Omega. As this is a java applet, clicking on any item on any of the above panels (sequence, graphs or the will cause the structure to center on the chosen residue and other graphs to show the position of the chosen residue).

As can be seen from the structure, there are two independent molecules of the Ly49A receptor in this structure. The overall fold of this protein is that of an alpha-beta proteins, C-type lectin domain. More information about this fold can be found from the Tertiary Structure Section of the Summary Pages.

Cross References and Links

The Cross References Section tab on the summary pages offer to SCOP and CATH databases of protein folds, as well data indexed from InterPro, UniProt and GO. Feel free to explore the SCOP and CATH classification of this protein.

Expand the Links Section on the bottom left of the page and click on the PDBsum link on this page. This will open up the PDBsum entry for this structure. Click on the Yellow Graph on the right side of the new page to open up the Ramachandran Plot (a plot that assesses the overall fold and local geometry of a protein).

According to the graph above, all residues are within acceptable regions of the fold space, and this means that the overall structure has no serious problems. Do note that a few residues are marked in red in this graph. These are residues that are the edge of acceptable criteria and any analysis based on these residues could be suspect. You can also view the validation analysis of this protein by choosing the PDBe Validation link from the same section of the PDBe Summary page for 1JA3.

From the Links Section, click on the PDBe SSM link. This will start PDBeFold (aka SSM(, a service that provides:

. pairwise comparison and 3D alignment of protein structures. . multiple comparison and 3D alignment of protein structures . . examination of a protein structure for similarity with the whole PDB or SCOP archives. . best Ca-alignment of compared structures . This job is run on a computer farm back at the EBI and may take a few minutes to complete. Once the task is finished, you will be presented with a list of entries in the PDB that have a similar 3-D structure to the NK Cell Receptor Ly49a.

Each line of the table has various scores, followed by overall RMSD between the matched structures, number of residues in the 3-D alignment and sequence identity between the matches etc. Choose “Sort by Seq%” at the bottom of the page and when the page updates, jump to around result 244. This residue should be for PDB entry 2MSB.

As can be seen, there is only 15% sequence identity between 2MSB (the Mannose-binding protein) and Ly49a (PDB entry 1JA3). However, structural identity is 88% suggesting that even though the sequences are vastly different, the structures are very similar. Click on the number on the left side for this match to open up the details.

Choose the “View Superposed” button to open up Jmol and see the two structures. Also scroll down the current page to see a residue-by-residue listing of the two structures. The two structures are shown superposed in the Jmol window. You can immediately see that the two structures share the same overall fold (two helices packing in a beta-sheet).

The circled region in the Jmol screenshot above is a loop that corresponds to residues 177 to 194 from the Mannose-binding protein (2MSB). This region is the binding-site for mannose and calcium in this protein. In the residue-by-residue listing between 1JA3 and 2MSB, you will notice that the positions of the Cysteine residues is conserved between the two protein, even though sequence identity is low. As a matter of fact, there are 4 cysteine residues in both the structures forming two di-sulphide bonds. The position of these residues is conserved across the C-lectin family and is probably one of the constraints that keeps the structures similar. The equivalent residues for 177-194 in the Ly49a structure are absent and therefore, it is not possible to comment on the similarity between the binding-loop of Mannose- binding protein and Ly49 using this structure.

Now go “back to the match list” (at the top) and choose any 4 or 5 structures at random for multiple alignment by checking the square box. You can choose structures from different pages if you so wish.

In the example below, I randomly chose the following structures from the match list: Lung surfactant protein D (1r14), Protein CEL-I From Cucumaria Echinata (1wmy) , toxin convulxin (1uos) etc. Click on “submit for multiple alignment”. Once the results are available, you can view the superposed structures as previously and as shown below. All the structures appear to have the same basic fold and although they perform very different functions and have low sequence identity, they all share the C-lectin fold.

Now go back to the summary assembly page for 1JA3. (Click here).

The Assembly Page

The Quaternary Structure page contains information about the probably quaternary structure of the protein as determined by the PDBePISA service. The calculated quaternary structure of this protein is dimeric. This means that in solution, two molecules of Ly49a are likely to form the functional form of the protein. In this case, this is also proven from functional studies.

PDBePisa: Protein Interfaces, Surfaces and Assemblies

PISA is an interactive tool for the exploration of macromolecular (protein, DNA/RNA and ligand) interfaces, prediction of probable quaternary structures (assemblies), database searches of structurally similar interfaces and assemblies, as well as searches on various assembly and PDB entry parameters. PDBePISA can also be used to upload one’s own PDB file for the sort of analysis described below. Start PISA by going to: http://www.ebi.ac.uk/msd-srv/prot_int/pistart.html. Click on “Start PISA” button and on the next page, type in 1JA3 on the submission form. Wait for the page to update then click on the Assemblies button.

This page provides information about the quaternary structure of the protein and gives statistics and energy criteria which go into the prediction of this complex. Feel free to explore further details of the assembly by choosing the appropriate button.

Now choose the interfaces link (as shown in red in the above screenshot). View the dimer interface as shown below.

The interface details between the two Ly49a molecules are given in the details section. You can also search the whole PDB for other structures that have similar interfaces. Lets do that. Choose the button (shown in red above).

The calculations can take a little while but you should soon see a page that looks like below:

Find 2MSB (our Mannose-binding protein example from the structure similarity match) in this list. Click on the number to see the interface similarity results.

View the interface of 2MSB dimer from the bottom of the results page. The interface formed between the 2MSB (Mannose-binding protein) dimer (left) and the 1JA3 (Ly49a) dimer (right) are quite similar.

So this means that not only are Ly49a and Mannose-binding protein similar structures, but they also tend to associate in a similar way in solution and have similar inter-molecular interfaces. This is despite a low sequence identity and different biological functions.

The role of Mannose-binding protein is to bind mannose-rich cell walls of bacteria and cause death of the invading organism via opsonization. The role of Ly49a is to bind to MHC Class I molecules on the surface of infected cells. Are there any similarities in the binding region and are the binding site somehow related ?

Go to the summary page for PDB entry 1P1Z (Ly49a in complex with MHC I). (Click here).

From the Chain details of this structure, note that Ly49 is chain D in this structure and MHC I is designated chain A. Keep these details in mind.

Go to the PDBePISA service. (Click here). Start PISA and fill in 1P1Z for PDB entry and wait for page to update. Then click on the Interfaces button to see the interfaces for this structure.

On the interfaces page, choose the interface between Chain A (MHC I) and Chain D (Ly49 receptor). You can also choose to view this particular interface in the structure. Choose details of selected interface as shown below.

Open up the details of the chains D and A interface.

The interface between chain D and A is shown above. For Chain D (Structure 1), residues 226 to 244 appear to form salt bridges with the MHC I molecules. So this region in Ly49 receptor would appear to be the region that binds MHC I and trigger a subsequent cascade of events. So where is this region from Ly49 (residues 226 to 244) in this protein ? And is this region similar to that found in Mannose-binding proteins ?

Go to the PDBeFold structure comparison service. (Click here). Fill in the form as shown below. Essentially we are going to compare chain D (Ly49) from 1P1Z and chain B from 2MSB. Submit the job and wait a few minutes while the alignment is carried out.

See the results of this match.

The overall sequence identity is 16% and the overall RMSD between the two structures is only 1.97A. See the results in detail by clicking on the link on the left of this page.

View the superposed structures.

The equivalent loop regions are shown below. It would appear that residues 226 to 244 from Ly49 are in a loop that is analogous to the mannose-binding loop for Mannose-binding protein (residues 180-195). However, the sequence identity between the two structures in this particular region is very low.

It would therefore, be interesting to speculate that even though the origins for both proteins lie in a similar gene many millions of years ago, the functions and sequences of the proteins have diverged to a large extent to different biological functions, while still retaining the same overall fold.

The C-lectin fold in found in many seemingly unrelated proteins. And Ly49 is just one of a few C-lectin containing fold that do not appear to have any Calcium or mannose-binding activity. Other example of such proteins are a. Eosinophil major basic protein: Involved in hyper-eosinophilia and binds heparin in the same loop without calcium. PDB entry 2BRS. b. Type II anti-freeze protein: PDB entry 2AFP. c. Lithostathine: PDB entry 1LIT.

Feel free to explore these structures to learn more about the C-lectin family.

Chemistry and Binding Site Analysis

The structure we chose for this tutorial did not have any bound drug, substrate, inhibitor or other inorganic or organic molecules. However, more than 70% of all structures in the PDB are in complex with some molecule or the other. These include drugs, inhibitors, co-factors etc. The PDBe has designed search tools to explore binding sites and chemistry of these ligands. These services are called PDBeChem (Chemistry Database) and PDBeMotif (Binding Site database). We will explore PDBeMotif in details using 2MSB (Mannose-binding site) as an example.

Go to the Summary Ligands Page for 2MSB (Click here). The structure contains many different ligands (MAN – alpha-D-mannose, BMA – beta-D-mannose etc). The alpha and beta forms of mannose differ in the orientation of just one C-O bond in the sugar ring but have exactly the same formula.

Click on the interactions link as shown above to see all interactions of MAN with the protein. This will take you to the PDBeMotif service.

The PDBeMotif page shows the residues that are involved in the binding of MAN and are color codes with the nature of the interaction. See the last entry. This tells us that Mannose binds to the protein in the presence of Calcium, Glutamic Acid, Aspartic Acid and Asparagine.

More detail of the interaction can be seen by clicking on the “MAN” on the line. The binding site can also be viewed with Rasmol, AstexViewer by choosing the appropriate icon from the result line.

Ligand Discrimitation between proteins.

PDBeSite is a currently deprecated service which is very handy for the comparison of binding sites of two different ligands and show the results in an intuitive manner. Let us compare the binding sites of MAN (alpha-D-mannose)(below left) and BMA (beta-D-mannose)(below right). The only difference between the sugars is that in MAN the C1-O1 bond is normal to the plane of the ring and in BMA the bond is almost parallel to the ring.

Go to PDBeSite service and then choose Ligand Bonds from the left hand side menu. (Click here). Input MAN in one box and BMA in the other as shown. Then click on Search Statistics.

The residues that predominantly interact with alpha-D-mannose (MAN) are shown in red bars whilst those proteins that prefer beta-D-mannose (BMA) tend to have residues shown in green. It is immediately obvious that the two sugars have distinctly different environments. Proteins that bind MAN generally have residues such as ASP, THR and SER in the binding pockets. On the other hand, proteins that prefer to bind BMA generally have TRP, PHE or HIS. This difference can also be explained on the basis of the overall structure of the sugars. BMA sugar is usually planar with the C1-O1 bond in line with the rest of the ring. This conformation means that BMA can have stacking interactions with amino acids that have rings (PHE, TYR, TRP or HIS). You can see this yourself by clicking on the green PHE bar on the graph. This will open up a new graph where all interactions between PHE and BMA are classified according to interaction type.

You can also search the PDBeSite service by protein environment. This is useful to find out what ligands could bind in a given environment by analysis of the whole PDB archive. For more help and tutorials on the PDBeSite service, please see the Help section on the PDBeSite page. You can also upload your own PDB file for similar analysis.

Motifs and Sequence Signatures

The PDBeMotif service is an integrated resource, which provides information about ligands, sequence and structure motifs, their relative position and the neighbour environment. This service can provide instant analysis of the protein structure in terms of small molecular motifs and other sequence signatures that may be present in the structure. Let us see what answers we get for our initial structure (1JA3).

Go to PDBeMotif. (Click here).

Choose the PDB Header Search from the top panel as shown and enter 1JA3 in the PDB idcode box presented. Click on Search and then the PDB code from the results shown. This will open up the sequences section for this entry.

The various prosite motifs found in this structure are shown. The most prominent one is the C-Type Lectin motif along with the residues that form part of the signature. Click on any of the motifs to get more information about the protein. Click on the 3D motifs at the bottom of the page to see all the small 3D motifs that could be identified in this structure.

Conclusion

We hope this tutorial has introduced you to the various PDBe services and tools that are available for the search, analysis and retrieval of protein structures. Detailed help is available for every service on the top right side of the page. There are other tutorials which delve in greater detail into our services. These tutorials can be accessed from http://www.ebi.ac.uk/pdbe/docs/education/Tutorial.html. In case you wish need to contact the PDBe, please write to [email protected] with your query and we will be happy to assist you in every way possible.