3 Method
4.1 The trefoil knot
The trefoil knot (or the threefoil or overhand knot) is the simplest knot of all, which is characterized by 3 crossings. It is mathematically denoted as a 31 knot, where the subscript 1 indicates that there is only one prime knot with 3 crossings. The proteins with a trefoil knot are (1) methyltransferase, (2) transcarbamylase, (3) methionine adenosyltransferase, (4) carbonic anhydrase and (5) YMPa supantigen.
4.1.1 Methyltransferase
The S-adenosyl-L-methionine (AdoMet)-dependent methyltransferase (MTase) can be grouped into the knotted type and the unknotted. The majority of MTases, like the classical MTase (1vid and 1mwi), the SET domain1 (1ml9) or others (1nth and 1msk), do not have a knot3. But other MTases like SpoU4,3 and TrmD5,6 (or collectedly called as SPOUT) are known to have a 31 knot at the AdoMet-binding site7,3. We have identified 23 knotted MTases or MTase-like hypothetical proteins in the PDB (Table 1). Interestingly, 43% of them (i.e., 10
-8-
out of 23) are annotated as 'hypothetical protein' in PDB. We clustered them into 8 cluster groups using BLASTClust8 with 30% sequence identity as the similarity threshold. The knot of MTase arises from a local structural motif that is of a βαβ topology. The knot motif is made up of 3 parallel β-strands arranged in the order of β2, β1 and β3. A typical example is shown in Figure 8~11. Because of the right-handedness of the βαβ motif, the helix α1,2and α2,3 occur on the opposite sides of the plane spanned by β1, β2 and β3 strands.
The unique feature of this motif is that the peptide segment comprises β3 and its entailing loop threads through the peptide hoop comprising β1, α1,2 and the loop between them and this results in a 31 knot. We refer to this motif as the βαβ-knot motif. This knot motif is important in contributing most of the AdoMet-binding and catalytic activity6.
CATH9 and SCOP10 define an alpha/beta knot superfamily for MTase. Both data sets currently contain identical sets of proteins (19) in alpha/beta knot superfamily. Interestingly, there is one protein, 1oy511, which does not have a knot. Though 1oy5 belongs to the TrmD MTase family known to have a knot, its structure does not have a knot. 1oy5 has a βαβknot-like as that of a typical knotted MTase 1mxi:A4 Figure 12, but its chain segment composed of β3 and its entailing loop fails to thread through the hoop formed by β1, α1,2 and the loop between them. However, there is concern3 that the absence of a knot in 1oy5 may be due to its poor structural quality or its misfolded inactive structure10. At present, there are 5 knotted MTases (1x7o, 1x7p, 1zjr, 1v6z and 2cx8; see Table 1) that have not yet been included in SCOP or CATH data sets.
4.1.2 Transcarbamylase
There are only 3 transcarbamylases (1yh0, 1yh1 and 1js1) have a 31 knot.
The knot of transcarbamylase arises from a local structural motif similar to that of MTase. The knot motif is also of the βαβ topology, as in the case of methyltransferase. An example (1yh1) is shown in Figure 13. In this particular example, the knot motif comprises 4 parallel β strands arranged in the order of β3β2β1β4. The segment composed of β4 strand and its entailing loop goes through the hoop made up of β1, α12 and the loop between them. Another unqiue feature is that the loop between β1 and α12 contains lots of prolines. It is instructive to compare the knotted region of the knotted transcarbamylase with that of the unknotted one. One of the conspicuous differences between these two structures is that 1yh1 has a much longer, proline-rich loop (TYHPKPLN)12 while 1als has a much shorter loop (GDGN). The other knotted transcarbamylase (1yh0 or 1yh1), which has 31% sequence identity with 1js1, also have a long proline-rich loop (TWAPHPRPKPQ) 13. The proline-rich loop is rather rigid and, due to its longer length, forms a much larger hoop for a peptide segment to thread through it to form a knot. Additionally, the proline-rich loop also functionally important since it forms part of the active site13,12.
4.1.3 Sunperantigen YPMa – the dynamic knot
YPMa14 is a microbial protein belonging to the family of superantigens, which are able to excessively activate T cells by binding to the T cell antigen receptor (TCR). Both the solution structure (1poq) and the crystal structure (1pm4) of YPMa have been solved14. Thought the crystal structure does not contain a knot, we found that one of its solution structures contains a 31 knot. It will be interesting to see how the knot comes about in the solution structures.
-10-
We compare the different NMR structures of YPMa. In the knotted conformation, the N ad C termini thread through the loop (in cyan), creating 3 crossover points characterics of a 31 knot. If any one of them is missing, the knot will not form. This type of knot is different from what we have considered so far and may be termed as the transient knot, since its formation depends on the conformational fluctuation. The knots in MTase, transcarbamylase or proteins in the later section may be termed as the intrinsic knots, since they do not depend on the dynamics of the protein conformation. Since X-ray structure of YPMa does not contain a knot, one may tend to rule out the possibility of a knot in the solution structure. However, since YPMa packs as a trimer in the crystal structure, but behaves as monomer in solution14, the crystal structure may not reflect some distinguished features inherent in the solution structure. On the other hand, though, topologically speaking, this knot is a real one, whether a knot physically exists in YPMa depends on the quality of its solution structures.
Obviously, further experimental work is required to resolve this issue.
4.1.4 Methionine Adenosyltransferase
Methionine Adenosyltransferase (MAT) is the solely enzyme that is responsible for the synthesis of AdoMet, the most important methyl donor present in living organisms. MAT contains a 31 knot15,16. The knot is due to the global fold of MAT, instead of a local knot motif like MTase or transcarbamylase. There are 12 MAT proteins (or 26 chains) in the PDB (see supplementary material).
4.1.5 Carbonic anhydrase
Carbonic anhydrase has long being recognized to contain a 31 knot17 Figure 14. We have identified 230 chains of carbonic anhydrase in PDB. Like MAT, carbonic anhydrase has a knot due to its global fold instead of a isolated knot
motif.