Fast and successful association of proteins to protein complexes is essential for many biological processes. It has been shown that yeast proteins interact with an average of eight others, for humans this number is going to be much higher131. In a crowded environment like the cellular medium the proteins exist at the solubility product. The concentration of individual proteins is very small though. For an up-regulated house keeping protein the concentration is in the lower nanomolar range leading to an average inter-protein distance of around 1000 Å. The average time until two binding partners encounter each other purely through diusion is very high compared to the reaction time. Therefore, these interactions are called diusion controlled reactions that might become a (partly) reaction controlled reactions should a subsequent chemical modication signicantly prolongs the binding.
The association constant ka is mainly controlled by the diusion through the medium but also by structural limitations of the binding pocket132. The diusion time is increased by the presence of crowding agents, e.g. other proteins or PEG that hinder diusional progress133135. Salts hinder electrostatic interactions by masking charges necessary for successful association136140. Substances like glycerol or sugars increase viscosity addi-tionally increasing diusion time141,142.
Mutational studies showed that primarily charged amino acids close to or in the binding pocket have a strong inuence on the binding process. Neutral mutations did not change the association constant. Computer based studies indicate the presence of energy funnels near the binding pocket that govern the binding process and can signicantly increase kd.
The association process can be described by a four state model142: A+B −−)k−−1*
k−1 AB∗−−)k−−*2
k−2 AB∗ ∗−−)k−−3*
k−3 AB (3.30)
Figure 3.7: Schematic representa-tion of the energy prole of a protein catalysed reaction. (1) the prod-uct state, (2) the diusion encounter complex, (3) the transition state, (4) the intermediate complex, (5) the educt state.
with A and B being two stable proteins in solution.
They form an initial unstable diusion encounter com-plex that can dissociate very easily as k-1k2. In a diusion controlled reaction the encounter complex is never the dominant species. It only can be observed at very high protein concentrations which lead to an increased observed association rate. This lies at or above the detection rate of stopped ow or BIAcore measurements or other heterogeneous phase detection methods used to perform real time association exper-iments. The encounter complex is more of an arti-fact from the times when protein interactions were re-garded as regular mono-molecular reactions.
The encounter complex then becomes the interme-diate complex which is already close to the nal com-plex. This is the rate limiting step with k3k-2. It is a
fully dehydrated complex, and all orientational rear-rangements and/or chemical modications are already completed. The intermediate complex has to undergo a certain reorganization to become the nal complex.
This can either be very fast as in the reaction of cystatin A and papain (230 s-1)143,144 or very slow as shown by the interaction of hen egg-white lysozyme with the antibodies 10 or 26 (HyHEL-10, HyHEL-26) with a rate constant of 10-3 per second145.
To become the intermediate complex the proteins have to pass the transition state, or transient complex, which is the most unstable one. Here, covalent bonds are in the process of being made or broken. The transition state theory was developed for reaction-controlled mono-molecular reactions but can be used for protein interactions as well, as long as only the relative change in the transition state is measured as a function of mutations or changes in buer conditions. A great number of studies have been reported136,138,146,147. Experi-mentally the most direct results can be achieved using double-mutant cycles. Should the change of the association rate be additive by a pair of mutations, the two amino acids have most likely no direct eect onto each other. Experiments with the model systems barnase-barstar and thrombin-hirundin have shown that charged residues that interact during the transient complex also interact during encounter complex. Mutations with uncharged residues have shown no eects148,149. Increasing the salt concentration lowers the association rate as it masks the electrostatic charges. These salting out eects are also used in protein crystallization to reduce the speed of protein association.
Another popular method for investigating protein interaction is the phi-value analy-sis150. The phi-value is dened as:
Φ = ∆∆G/∆∆GD (3.31)
with ∆∆G being the free energy dierence of the free proteins and the transitions state of wild-type and mutation and ∆∆GD is the free energy dierence between the free proteins and the complex of wild-type and mutation. A value close to one indicates interaction during the transition state while a value around zero indicates interaction after the transition state140.
The transition state is in general a really interesting state as this is the condition where the protein actually does its job. The transition state, however, cannot be described structurally. Atomisticly detailed structures of the single proteins and the complex can be derived using NMR or X-ray crystallography fairly easily, and the encounter complex can be trapped using non-processable ligands, like the non-hydrolisable ATP-analogons AMP-PNP or AlF3. But the actual occurrences during the transient complex, the large conformational movements, the side chain reorientations and the changes in the covalent bonds that lead to ligand processing or complex formation and the signicant conforma-tional changes of the whole protein related to this cannot be determined experimentally.
Force eld based all-atom simulation methods on the other hand are precise enough and fast enough to simulate protein behavior on a 100 ns scale. Since protein reactions are diusion controlled, actual interactions are rare events. To have the protein perform
the desired actions within said timescale, articially biasing potentials have to be applied.
In the past this has been successfully done by applying SMD and TMD on protein ligand interaction and ligand processing and also to some degree on protein complex formation.
4.1 Iron
Iron is an essential micronutrient for almost all organisms, and all organisms require approximately the same concentration of 0.3 - 10µM iron for growth151. It plays a vital role for synthesis of DNA, RNA and chlorophyll, in electron transport, oxygen metabolism and nitrogen xation, and also as a cofactor for many enzymes such as catalase.
Despite its abundance on earth, iron is scarcely available to living organisms152. In anaerobic environments, it is present as Fe2+, which is soluble under physiological pH, and microorganisms can acquire it easily. In aerobic conditions, however, it is oxidized to Fe3+, which forms insoluble iron hydroxides, leaving the available iron concentration below 10−18M153. This concentration is much lower than the absolute minimum require-ment for microorganisms (0.3 - 10µM151) and, therefore, too low for passive uptake154.
Figure 4.1: Structure of heme b
The concentration of free iron within animal or plant host or-ganisms is even lower. Iron is toxic to cells as it promotes the formation of reactive oxygen species, most importantly in the Fenton reaction: F e2++H2O2 → F e3++OH−+OH·. There-fore, iron is bound to ferritin and myoglobin or as a cofactor of enzymes, such as catalases.
In extracellular uids, all iron is bound to transferrin or lacto-ferrin. Both proteins are present in such high concentrations that they are iron-loaded to only 30 %. This ensures that free iron is bound by the respective proteins immediately. Keeping the concentration of free iron so low is also thought to be one of the host's mechanisms to ght infections155.
Because of these limiting iron supplies, bacteria had to develop sophisticated mechanisms to acquire iron. Virtually all aerobic
bacteria synthesize and secrete low molecular weight compounds, termed siderophores, that bind Fe3+ with very high anity and in turn are taken up by specic receptors. In case of Gram negative bacteria, the transport of the iron-loaded siderophores through the outer membrane is performed by TonB dependent receptors (described in section4.3) and energized by the proton motive force of the inner membrane.
Many of these siderophores of host living organisms are able to extract the iron from transferrins and lactoferrins directly.
A second way for bacteria to capture the iron from transferrin and lactoferrin is by direct
interaction of the outer membrane receptor with the iron carrier protein and subsequent uptake of the iron alone.
A third iron source present in mammalian hosts is iron in heme and bound to heme carrier proteins. Heme is an aromatic molecule composed of a porphyrin ring system with an iron atom in the middle. Porphyrins are composed of four pyrroles connected by methine bridges. The pyrrol subunits can be substituted in dierent ways, resulting in several naturally occurring heme types. The most common type is heme b, which is shown in Figure 4.1. Heme receptors will be described in section4.5.
The only organisms known so far that do not need iron are some Lactobacillus species156 and Borellia burgdorferi157. In these organisms the role of iron as a cofactor is lled for example by manganese or cobalt.