NGF - twenty years a-growing

NGF - twenty years a-growing

A molecule vital to brain growth

It is twenty years since the structure of nerve growth factor (NGF) was determined ref. 1. This molecule is more than 'quite interesting' as it is required for all of us to develop a brain that can take any interest in anything! During the embryonic development of the brain and nerves, cells signal specifically to each other in a variety of ways so that growth, contacts, and changes in structure can be coordinated. One way is through the release of growth factors from a secreting cell and the display of receptors for these factors on the cells that need to respond. NGF is one of a family of neurotrophins which are growth factors involved in the formation of the nervous system. These are also used in adults to maintain and regenerate nerves and also in some pain responses.

The NGF molecule is built from β-strands and with disulfide bonds

It was immediately clear from the crystal structure (PDB entry 1bet) that NGF was different from any previously determined proteins or domains. It had an elongated shape built from a new arrangement of β-strands twisted around each other (see view-1). Each strand is an extended polypeptide chain that hydrogen-bonds through its backbone nitrogen and carbonyl oxygen atoms to a neighbouring strand in the structure. These non-covalent H-bond interactions are shown here for several residue pairs through the structure showing how their backbone atoms hold each other in line.

Additionally, special covalent bonds help to stabilise the NGF structure. These are the three disulfide bonds from one side of the structure to the other. Disulfide bonds are covalent links between cysteines commonly observed in extracellular proteins. Cysteines linked in this way have been historically called 'cystine' - dropping the middle 'e'.

The NGF structure threads disulfides to give a cystine knot

Disulfide bonds had been observed many times before - for example in immunoglobulin structures - and were anticipated in the NGF structure following its sequencing in 1971 by Ralph Bradshaw (ref. 2). He showed that cystines could be isolated from the protein which contained six cysteines along its length as highlighted in view-2.

Strikingly, the NGF fold contains a neat interlocking between these cysteines (see view-3). The cysteines cluster towards one end of the molecule and are arranged so that one S-S bond threads through two others to give an arrangement called the 'cystine knot'.

At the time of its discovery this was such an unexpected feature that the authors had to double check the experimental data to be sure of what they had observed. Subsequently, this neat arrangement was found to be the basis for a complete family of 'cystine knot' containing growth factors that includes other nerve-directed 'neurotrophins' and other important signalling proteins. Interestingly, it is clear from this view that the cysteine residues are close enough that with small adjustments in the register of the residues completely different connections could be made.

Making the NGF dimer involves symmetry in the crystal

NGF is known to be a very tight dimeric molecule in solution and to act by binding specifically to a number of different receptors at susceptible cell surfaces. The binding of NGF produces a change in the structure of these dimeric receptors which in turn activates signalling pathways at the inner surface of the cell membrane.

So why is a dimer not observed in the PDB entry for the NGF structure? In fact the authors did report a dimer in the crystal structure. However, it is customary in X-ray crystallography to deposit only the smallest building block of the crystal - the so-called asymmetric unit. To build the complete crystal, rotational symmetry operations need to be applied to this unit. The operations are specified by the space group definition of how the proteins pack together to form the crystal. In this case the dimer is produced by a so called '2-fold' rotational symmetry operation corresponding to a 180 degree rotation around an axis parallel to the long axis of the protein (see view-4).

Application of this symmetry operation is required to generate a second copy of the NGF molecule (shown here appearing in grey). This ghostly 'copy' is in fact present in the crystal and, significantly, is needed to complete the biological assembly of NGF - shown as a dimer of red and green molecules. The rotational symmetry is more obvious if you rotate the molecule to look down along the axis of symmetry which is how this view ends - but feel free to rotate the molecule for yourself to see how the dimer is constructed.

PDBePISA serves up the NGF dimeric assembly

Many crystal structures in the PDB archive require a similar use of symmetry operations to give the biologically-relevant assembly. For this reason, PDBe provides a sophisticated service to apply crystal symmetry operations to deposited structures in order to identify the most likely quaternary structure. This PDBePISA service automatically examines all contacts between macromolecules in the crystal - including any copies generated by symmetry. These contacts are then scrutinized to determine whether they are likely to persist in solution and stabilise the so-called quaternary structure of the molecules. You can learn about applying PDBePISA to the 1bet NGF structure in this mini-tutorial. The PDBePISA analysis is pre-calculated for most crystal structures in the PDB. You can see a picture of the predicted biological assembly by visiting the PDBe summary page for the entry you are interested in. It will be shown as part of the PDBportfolio of images.

The NGF dimer interacts with cell receptors

In the case of NGF, the construction of the elongated dimer sets up the faces of the protein that interact with similarly elongated multidomain receptors and produce the various biological responses in target cells. NGF family neurotrophins bind to two classes of receptor. One is the Trk tyrosine kinase class which has distinct subtypes each specific to one type of neurotrophin. Binding activates the kinase activity of the intracellular part of these receptors. The other class is the p75 neurotrophin receptor (p75NTR) which binds all the neurotrophins and has an intracellular domain that is structurally similar to death domains which are involved in apoptotic cell death. Despite this homology, NGF binding to p75NTR can promote cell survival. Here (see view-5) is one example (PDB entry 3buk) of an NGF family member in complex with the p75NTR receptor showing how the NGF molecule binds to the receptor domains. This receptor uses cysteine-linked (although unknotted!) and glycosylated extracellular domains to recognise the dimeric NGF-like neurotrophin-3. The transmembrane and intracellular death domain of this receptor were removed before crystallisation. The structure of the death domain is known from a separate NMR structure determination that is deposited in the PDB as entry 1ngr.

Further exploration

Some other growth factors have been crystallised with the dimeric form of the growth factor as the crystal building block. One example of this is the neurotrophin-3 in the PDB entry 3buk, as shown in view-5 where it is bound to its receptor. A dimer was also observed for NGF in a different crystal form in entry 1btg and the related neurotrophin-4 deposited as entry 1b98.

Remember that, although sophisticated, the PDBePISA calculations have been calibrated for only the most common interactions between macromolecules. As a result it can be incorrect in particular cases. You should always use your own scientific judgement and use other information to assess for yourself its predicted assemblies.

The PDB archive contains several structures of NGF family members bound to their receptors. In addition to PDB entry 3buk described above there are examples of NGF family members bound to its tyrosine kinase class of receptors. For example PDB entry 1www which is a crystallised complex of NGF with the ligand-binding domain of the TrkA receptor. Some of these studies have taken advantage of the conservation in structure among mammalian neurotrophin components. When inspecting the source organism PDBprints logo you will see many are MULTI source structures - typically a receptor from one species is binding an NGF family member from another.

The question of whether these receptors follow the dimeric symmetry of the NGF, as in the example shown in view-5, has turned out to be a complicated one. The best way to read the papers on this is with the structures in front of you. To look in detail at how the proteins are interacting you can browse the PDBePISA output for the complexes. This mini-tutorial will help you get started on running PISA analyses and interpreting the output. The output from the analyses of the NGF complexed with the receptors is more complicated but should allow you to answer the following quite interesting question: are the binding interfaces of NGF with p75NTR and TrkA likely to be mutually exclusive? A special storage form of NGF, which is a complex with accessory binding proteins was also crystallised and this structure is available as PDB entry 1sgf.


PDBe would like to thank Prof. Tom L. Blundell for advice on the subject of this QUIPS article.