Tyrosine residues in different proteins:
Phenol ring rotation angle database

When an electron is subtracted from tyrosine, a cation radical is formed which immediately (at physiological pH) drops the oxygen proton, so that the tyrosyl radical is a neutral species. The ring can rotate around the C1 - C_b bond in a free tyrosine (Fig . 1).

The appearance of the EPR spectrum (its lineshape) is defined by the hyperfine interaction of the unpaired electron with the radical's protons. While such interaction for the four ring protons is independent of the orientation of the ring, on which the most of the spin density resides, the hyperfine splitting constants for the b-protons of the methylene group do depend on the rotation angle q according to the McConnell relation:

{1}

{2}

Thus, a simulation of an EPR spectrum of a tyrosyl radical provides us with the values of the isotropic hyperfine splitting constants for the methylene protons. These two values then can be used in system {2}, and the values of q and r_C1 can be found.

The solutions of system {2} can be found numerically. I have created an Excel file (theta_in_tyr.xls) that can be used for solving system {2} on a routine basis:

Click to download: (macros should be allowed) Theta_in_tyr.xls (950 KB)

As explained above, an EPR spectrum of a tyrosyl radicals in a protein provides very definite information about the rotational angle q of the phenoxyl ring in the radical. Successful simulation of such EPR spectrum yields two values of the isotropic hyperfine splitting constants for the methylene protons, and angle q can be found from these values. This database can then be used to identify the Tyr residues in the protein which are in the conformation close to that found for the radical. Thus this database can help to assign the radical to a particular Tyr site (and there could be dozens of those in one protein!).

The EPR spectrum of the radical formed in the reaction of human haemoglobin with hydrogen peroxide and its simulation are shown in Fig. 3.

Fig. 4 shows that there are always four solutions of the system of two McConnell equations written for the two methylene protons {2}. Two of the four solutions are characterised by a common value of r_C1 and two rotational conformations (two values of q) symmetrical with respect to -30°. The other two solutions are characterised by another value of r_C1, also common for both, and by two rotational conformations symmetrical with respect to 60°. Certainly, only one of the two possible values of r_C1 is correct and corresponds to the spin density on atom C1 in the radical. Analysis of literature data on tyrosyl radicals shows that it must be the value of 0.375, and the value of 0.140 is too low to be true and therefore should be discarded. Thus, we are left with two possible values of rotational angle q, both for r_C1 = 0.375: q = 53.0° and q = 67.0°. These two angles are equivalent from the EPR point of view. There is no way to say from the EPR spectrum alone, which of these two rotational angles is the characteristic of the radical observed.

Here we come to the point when our database should be used. We should find the Tyr residues in the human haemoglobin structure with rotational conformation of the ring close to either one or the other possible angle found by the EPR spectroscopy, i.e. 53.0° or 67.0°.

Human haemoglobin has twelve Tyr residues, three in each subunit (Fig. 5).

We can now go to 'Search', specify haemoglobin's PDB ID code (1HGB) in the pull-down menu and have the table of all twelve Tyr residues in this protein with individual values of angle q. By default, the data in the table are sorted by tyrosine number. You can sort the data also by chain (A, B, C and D in our example), by q or by the difference |q-q_target|. We will be particularly interested in the last option, having specified this target value of q, first, as 53.0° and then as 67.0°. When the data are sorted in this way, the residues with the most close rotational conformation to either 53.0° or 67.0° will be arranged descending from the top.

Chains A and C in haemoglobin are identical (the two a-subunits), so are chains B and D (the b-subunits). However, homologous tyrosine residues not always have identical, or at least very close conformations: while angle q in aTyr140 has close values in A and C chains, and the same is true for bTyr145, the rotational angles in other homologous residues are different by 9-13°.

The speculation on the nature of this effect is out of the scope of this database description. Instead, we would average the values of q for homologous residues and sort the data by the difference between this averaged angle and the two target values. This could be easily done if we copy the table on the web page and paste it into a blank Excel spreadsheet. The result would be as follows:

Thus, a formal use of our database allowed us to produce a short list of tyrosine residues in the human haemoglobin structure which are the most likely candidates to host the observed tyrosyl radical. The smallest difference |q_average - q_target| is found for Tyr42 in the a-subunits when q_target = 53°. Therefore, this tyrosine should be considered as the most probable site of the free radical location. Next probable in the list would be are Tyr130 and Tyr35 in the b-subunits if q_target = 53° and Tyr42 in the a-subunits if q_target = 67°.

You will have to register (follow the link on the top of this page) in order to be able to submit data to this database. Once your registration is confirmed by an e-mail message, you can log in (the link on the top of this page) and select from the three possible options: update your personal data, add records to the database, edit your previous record(s) to the database.

Please note that you are not expected to submit actual value of q. Angle q cannot be measured directly because protons are not resolved in the crystal structure. Instead, we have to assume that each Cb-Hb bond is forming a 120° angle with the R-Cb bond, when using view A defined in Fig. 1., and, of course, the angle between Cb-Hb₁ and Cb-Hb₂ is also 120° (Fig. 6.) Therefore, we have to measure the dihedral angle between the ring plane and the R-C_b-C₁ plane, i.e. either angle f₂ or angle f₆ (Fig. 6). Ideally, the sum of these two angles should be equal to 180°. In reality it is never the case since the six carbon atoms of the ring never form a perfect plane. Therefore, the server of our database takes the values of both angle f₂ and f₆ as an input, calculates an average value for two angles, f₂ and (180°-f₆), and uses this average value for calculation of angle q (see Fig. 6).

Fig. 6. The ring rotation angle q can be found from directly measurable dihedral angles f2 and f6. From this diagram: q° = 90° - (120° - f2) = f2 - 30°. Angle f2 is only approximately equals to (180° - f6) since the ring is never ideally flat in the real situations, so we use an average value instead of f2: q = ((180° - f6) - f2)/2 - 30° (note that f2 and f6 have opposite sign).

How can we measure f₂ and f₆? Swiss PdbViewer is a free program that allows viewing and analysing 3-dimensional structure of biomolecules. It has a tool for measuring dihedral angles. Once you have a tyrosine zoomed-in on you screen, you can measure angles f₂ and f₆. First, press Ctrl then click on this button :

Then click consecutively on atoms R, Cb, C1 and C2 for f₂ or on atoms R, Cb, C1 and C6 for f₆ (Fig. 7, note that atom nomenclature will not be displayed).

Fig. 7. The sequence of atoms to click on when measuring the values of the angles f2 and f6 with Swiss PdbViewer.

The value of the dihedral angle will be displayed on the fourth click. Having measured f₂ and f₆ for a tyrosine residue, you can go straight to the submission page and submit these two values; angle q will be calculated automatically and added to the database.

I am very determined to change this. I plan to create a program that would append a structure file from the Protein Databank, would calculate angle q for all tyrosine residues using the atoms coordinates and would add the result to the database. This would make the database completely comprehensive: for every protein structure present in PDB, angle q could be found and ranked for all tyrosine residues. When this task is accomplished, the database will become a very powerful tool in assigning a tyrosyl radical to a particular residue in proteins.

There are also other things which form the following 'to do' list:

• To make available the option of averaging of angle q in homologous tyrosine residues in identical chains (within the same structure).

• To make available the option of viewing (and averaging if necessary) the data on angle q in the same tyrosine residues present in different structure files (by 'different structure files' I mean either the same protein reported at different times or two or more modified structures of the same protein, e.g. by mutation, ligation, pH etc.).

• To make available the option of viewing the data on angle q in conserved tyrosine residues in different (related) proteins.

• To replace all Excel driven fragments of this web site with JavaScript.

• To set up a discussion group.

• To add material about computer simulation of the tyrosyl radical EPR spectra (once my recently submitted paper is out).

• To make the appearance of this web site browser insensitive (originally it was composed for best viewing with Internet Explorer).

Dr Dimitri Svistunenko Department of Biological Sciences University of Essex Wivenhoe Park Colchester Essex CO4 3SQ United Kingdom Phone: 01206 873149 FAX: 01206 872592 E-mail: I am available for web cam conferencing (e-mail me for details)

© Dimitri Svistunenko 2003-2004