Lecture 21 February 2005

These are protein folds in which the domain contains only β-strands. This is a very diverse group, both in terms of structure and of function. It includes enzymes, antibodies, transport proteins, anti-freeze proteins, and viral coat proteins, among many others.

They can be arranged in a circular barrel, where the last strand H-bonds to the first strand,

The jelly roll contains 1 Greek key motif, but has other distinguishing features.

One example of this class is the lipid binding proteins. They come in 2 varieties:
Intracellular-10 β-strands and 2 α-helices between strands 1 and 2.
Extracellular-8 β-strands and 1 α-helix at the C-terminus (Fig. 1-19)

The intracellular retinol binding protein binds retinol (Vitamin A), a lipophilic compound.

As shown in Jmol, it is nearly a perfect barrel. There is a small separation between 2 of the 10 β-strands. All strands are anti-parallel and each is connected to its 2 nearest neighbors (except the first and last, which are only H-bonded together.)

Antibodies (Fig. 1-56) and many other proteins from the immune system have domains with β-sandwich domains. They commonly have 7-9 antiparallel strands, with 3-5 strands per sheet. (Jmol). It is a good example of modular design. Some domains are important for holding the chains together. Others come together to form the antigen binding sites.

Each domain in an antibody, and in many β-sandwich proteins, contains at least one Greek key motif.

The difference between an up-and-down sheet and a Greek key is the connectivity of the strands:

The four strands in a Greek key motif need not be in the same sheet. It is a topology (or connectivity) motif. In the immunoglobulin shown (Jmol) the Greek key has 2 strands in each sheet . In the gamma-crystallin, the Greek keys are split 1 strand to one sheet and 3 to the other. (Jmol)

A more complex connectivity in anti-parallel sheets is the jelly roll motif. It arises from the following pairings of β-strands:

Many viral coat proteins contain jell roll domains, but some enzymes, such as allantoicase do also. (Jmol)

In the previous examples, the anti-parallel β-sheets were always on the surface, and so would tend to be amphipathic. There is one common class of folds in which anti-parallel β-sheets are largely buried. This is the β-propeller.

These structures are generally very rigid, but some of the enzymes use the central channel for transport or catalytic function. Although many β-propeller proteins are enzymes, perhaps the most well-known, the β-subunit of the heterotrimeric G protein, is a protein whose function is binding other proteins.

Unlike the others we have just seen, this is an all β domain that has parallel strands. How is that accomplished?

The chain is wrapped in a helical fashion, and there are 2 or 3 faces composed of parallel β-strands. The strands are typically short and the sheets are flat and long.

α/β Structures

These are protein folds in which there is a regular pattern of repeating α-helices and β-strands, which form a parallel sheet. There are 3 main classes:

1. α/β barrel

2. open twisted sheet

3. horseshoe fold

The first two can be distinguished by considering two different ways of linking 2 β-α-β units;

In the barrel, the two units are linked in the same orientation, so that the helices are all on the same face of the sheet. The strands then form a circular sheet in which the ends are enclosed, like a barrel. The helices are all on the outside of the barrel.

In the open twisted sheet, some of the units are rotated 180˚ with respect to other, such that some helices are on each side of the sheet.

In the horseshoe fold, the first pattern is started, but the ends are not joined.

β-sheet		α-helix

α/β barrel	open twisted sheet	horseshoe fold

The number of possibilities is reduced because 2 β-strands are almost always connected by an α-helix in a right-handed way, rather than a left-handed way. (Fig. 1-60)

To determine the handed-ness of the connection, line up the β-strands with the C-termini up, and the polypeptide chain going away from you. If the α-helix connection is on the right, it is right-handed. Probably because the β-strands tend to twist to the right in this view, the left-handed cross-over connections are extremely rare.

The first 2 folds are illustrated in (Fig. 1-61).

1. α/β barrel

Nagano N, Orengo CA, Thornton JM.
One fold with many functions: the evolutionary relationships between TIM barrel families based on their sequences, structures and functions.
J Mol Biol. 2002 Aug 30;321(5):741-65. Review.

This is a very common fold, mostly found in enzymes. Typically, about 200 amino acids in the domain. It is a rather regular fold. Most examples have 8 parallel β-strands. Viewed from the top, the sheet is circular or elliptical. The β-strands are usually 5 amino acids, with the side chains of number 2 and 4 pointing towards the center of the barrel. For example, Val, Ile, Phe, Leu. Typically the center is filled with such side chains. In some cases, the center is used for substrate binding. Amino acids at positions 1, 3, and 5 may be somewhat polar. They will mostly interact with the helices, but perhaps with water. Overall, these β-strands will be composed of hydrophobic amino acids. In contrast, the α-helices are clearly amphipathic, with one face to the solvent and one to the interior of the protein. (Jmol)

Some proteins have multiple domains with more than one enzyme function. One or more domains might be an α/β barrel. In E. coli, a double barrel protein exist with each domain catalyzing a reaction in the tryptophan synthesis pathway. The active sites are at opposite ends.

Where are the active sites in α/β barrel proteins? They are found in the loops at the C-termini of the β-strands.

Then, are all α/β barrels related? Have they been generated by divergent evolution from a common ancestor, or have unrelated genes found the same fold? Several analyses have suggested that many such proteins are related. Recently it was suggested that this fold might be the result of gene duplication and fusion of a 4-strand domain. Substrates commonly have a phosphate, and a few proteins have symmetrically placed phosphate binding sites.

In a 1990 analysis (Farber GK, Petsko GA.The evolution of α/β barrel enzymes.Trends Biochem Sci. 1990 Jun;15(6):228-34. Review.)

1. All examples are enzymes (i.e. they are not clearly unrelated)

2. Active sites are always at the same end.

3. All barrels have the same handed-ness and similar tilt. (Further analyzed by Chou KC, Carlacci L., Energetic approach to the folding of α/β barrels. Proteins 1991;9(4):280-95

Handedness is right or left. Tilt is right , left, or none. There is a nominal chance of one in six to get the observed right handedness, and right tilt. This argument is weakened by energetic analysis that demonstrates that the observed situation seems to be the most favorable.

4. Sequence similarity is low to undetectable, except when comparing the same enzyme from various species.

5. Sequence motifs have been detected at the first and fifth positions of the β-strands.

Hocker B, Beismann-Driemeyer S, Hettwer S, Lustig A, Sterner R.
Dissection of a (beta/alpha)8-barrel enzyme into two folded halves.
Nat Struct Biol. 2001 Jan;8(1):32-6.

This study provided the insight that this fold might have come from a four-stranded precursor, and so it becomes more reasonable to think that all α/β barrels might be related.

Open-Twisted Sheets

This fold is also commonly found in enzymes. While the α/β barrels are quite regular, this fold is extremely diverse. For example, the number of β-strands is variable, and sometimes a few are not parallel strands.

Since the crossover connections are virtually always right-handed, it is the order of the β-strands in the sheet that determines whether the helices are found on one face or the other face of the β-sheet. Similar to the barrels, these β-strands are nearly completely buried, and so tend to be composed of hydrophobic amino acids.

Where are the active sites? Again they are found in the loops at the C-terminal ends of the β-strands. Furthermore, they are found at the so-called "switch-points". This is between two β-strands that are going in opposite directions.

The first strand is to the right of the arrow (in red). It comes up and connects at the top (black line) to the helix (blue). The helix connects at the bottom (dotted line) to the second strand. The chain now reverses direction and connect to the left of the arrow. Looking at the C-termini (top) of the β-strands the switch occurs at the center position, indicated by the arrow. A substrate will usually bind between these two loops.

Some proteins with larger sheets will have 2 switch points, perhaps for two substrates.

In flavodoxin there are 5 parallel strands and one switch point where its substrate FMN binds.

See (Jmol)

A nucleotide binding motif was discovered by Michael Rossmann in the 1970s by analyzing several NAD binding enzymes. This has proven to be a valid generalization, and most nucleotide binding proteins have the twisted open sheet fold. e.g. NAD, ATP. This is now called the "Rossmann fold" . It has a characteristic sequence from a β-strand-loop-α-helix. In the ATP synthase, which binds ATP and ADP at such a position, two sequence motifs were identified by John Walker and Matti Saraste. They are called the Walker A and B sequence motifs. The Walker A motif is called the "P loop", since it wraps around the phosphates of the adenine nucleotide. (Jmol)

Walker JE, Saraste M, Runswick MJ, Gay NJ.
Distantly related sequences in the α- and β-subunits of ATP synthase, myosin, kinases and other ATP-requiring enzymes and a common nucleotide binding fold.
EMBO J. 1982;1(8):945-51.

Rossmann MG, Argos P.
The taxonomy of protein structure.
J Mol Biol. 1977 Jan 5;109(1):99-129.

The Horseshoe Fold contains from 4 to 17 parallel β-strands in a curved β-sheet.

Each β-strand is associated with a loop and an α-helix. This unit is also known as a Leucine Rich Repeat (LRR).

A ribonuclease inhibitor protein was found to have the horse shoe fold. (Jmol)

Enkhbayar P, Kamiya M, Osaki M, Matsumoto T, Matsushima N.
Structural principles of leucine-rich repeat (LRR) proteins.
Proteins. 2004 Feb 15;54(3):394-403.

The other two classes of folds are much less regular than the others we have considered.

α + β

Generally an anti-parallel or mixed β-sheet is packed against a group of α-helices on one side. An example is the TATA-Binding protein. (Fig. 1-62) (Jmol)

Irregular domains

These domains have little regular secondary structure, and are often stabilized by disulfide cross-links (Fig. 1-63) (Jmol) or by bound Metals. (Fig. 1-64) (Jmol)

These domains are usually small (e.g. less than 100 amino acids).

Comments/questions: Email me


Up and Down	Greek key	Jelly roll
Simple connectivity. In all 3 cases, this is the top view. The odd strands go up and the even strands go down.	This is a symmetrical, double Greek key. Many similar examples exist with one or two motifs.	This is the commonly found jelly roll motif. It can be a barrel or a sandwich.

Number of blades	Name	Function	Sequence motif	"Velcro" closing(N+C)
4	Haemopexin (Jmol)	Haem binding	Hydrophobic	-S-S-
4	Collagenase	Collagen cleavage	Hydrophobic	-S-S-
5	Tachylectin-2	N-acetylglucosamine binding	DNWL IGxGGW	3 + 1
6	Neuraminidase	Sialic acid cleavage	Aspartate box	1 + 3
6	Glucose dehydrogenase	Oxidation of sugars	2 H-bonding motifs	1 + 3
7	Methylamine dehydrogenase	Methylamine to ammonia + aldehyde	None	3 + 1
7	G protein β-subunit (Jmol)	Signaling	WD repeat	1 + 3
7	Galactose oxidase	Oxidation of alcohols	kelch motif	1 + 3
8	Methanol dehydrogenase	Oxidation of methanol	W-docking motif	1 + 3


Barrel: H-bonds (dots) are between all strands	Sandwich: Strands at the 2 ends are not H-bonded


Up-and -down (1-2-3-4)	Greek key (4-1-2-3)

	Sheet 1 Sheet 2
	Topologically, 2 Greek keys