What to do with waters in the binding pocket of proteins is a conundrum that drug designers frequently face. To address or replace: that is the question. Should a given water be left where it is so that a drug candidate can form H-bonds to it (address). Or should the water be replaced by a drug molecule functional group which mimicks or improves upon the water’s H-bonds (replace). The story of the HIV protease flap water sheds some light on this important topic and the two alternate strategies.

HIV

An estimated 16.5 million AIDS-related deaths have been averted since 2001, thanks in large part to the global roll-out of HIV treatments. In 2020, there were 680 000 deaths from AIDS-related causes, a decline of 58% versus 2001. Protease inhibitors have played a central role in Antiretroviral therapy (ART), a combination of HIV medicines from at least two different HIV drug classes taken every day. In total 10 HIV protease inhibitors have been approved for human use, with Saquinavir being the first in 1995.

proteases – what are they

Proteases (or proteinases) are enzymes which cleave peptide bonds, this process is called proteolysis. Proteolysis occurs in virtually all stages of a cell’s life, in all cell compartments, and in many stages of a protein’s existence. There are 641 protease genes in the human genome (i.e. 3% of the human genome) and viruses like HIV need them too.

Proteases as drug targets have been very fruitful with approved drug in the fields of angiotensin converting enzyme inhibitors for blood pressure, HIV protease inhibitors, proteasome inhibitors for myeloma and dipeptidyl peptidase IV inhibitors for type II diabetes.

There are five classes of proteases defined based on the chemical functionality responsible for peptidic bond cleavage. The five classes are:

• Metalloproteinases (~200 members)
• Aspartic acid proteases (~25 members)
• Serine proteases (~180 members)
• Cysteine proteases (~160 members)
• Threonine proteases (~30 members)

Proteases use two distinct mechanisms to cleave amide bonds. Metallo- and aspartic acid proteases use a highly reactive water as the nucleophile while serine, cysteine and threonine proteases use covalent catalysis to cleave the peptide bond. The HIV protease is an aspartic acid protease.

HIV protease and the flap water

The HIV protease forms a dimer in order to recognize the peptide which is to be cleaved. On one side of the peptide two aspartic acids activate the nucleophilic water molecule. One the other side of the peptide two flexible loops (flaps) lock down on the peptide via a water molecule – the flap water. The flap water binds to the backbone NH of ILE40 from each of the two flaps and two additional H-bonds to the backbone carbonyls of the P1 and P1´ peptides (see Figure below).

the thermodynamics of water

The thermodynamics of water molecules on protein surfaces versus bulk water is a central topic in drug design. All protein bound waters are more entropically constrained than their counterparts in bulk water. Dunitz estimated the entropic cost of transferring a single water molecule to a site in which it is relatively immobile (i.e. in ice or in a crystalline salt) to be about 2 kcal/mol at 300 K. This value can be taken as a an approximate upper limit of entropic gain upon displacement of one water. The entropic gain for displacing the Flap water in HIV protease has been calculated to be even higher at around 3 kcal/mol at 300K.

With regards to enthalpy, some protein bound waters are more stable while other are less stable than bulk water. The Gibbs free energy equation combines the enthalpic and entropic terms resulting in waters which are less stable, the same and more stable than bulk water molecules. Things are further complicated by the fact that waters often exist in networks on the surfaces of proteins and thus the overall changes in the water networks needs to be considered and not just single water molecules. But the flap water is in isolation making its study ideal to introduce water addressing and replacing.

thermodynamic signature grids

The flap water is highly energetically unstable with a calculated ΔG of +2.5 kcal/mol. This free energy is completely entropy driven due to the highly constrained nature of the water with the entropic term -TΔS being 2.6 kcal/mol: The enthalpy of the flap water is bulk water-like (ΔH of -0.1 kcal/mol.). (See Thermodynamic Signature Grids below).

addressing the flap water

Saquinavir addresses the flap water. In fact, the first 9 approved HIV protease inhibitors all address the water. Two H-bonds are formed between amide carbonyls of Saquinavir and the Flap water (se Figure below). In addition, a secondary alcohol group interacts with the two catalytic aspartate residues.

replacing the flap water

Surprisingly, it was already known before the approval of Sasquinavir, that also displacing the flap water was possible and indeed advantageous, but it took until 2008 for the first “flap water displacing” HIV protease drug to be approved. In 1994 Lam et al. discovered cyclic ureas which displace the flap water replacing the H-bonds between the flap water and the backbone NH groups with H-bonds formed with the stronger H-bond accepting cyclic urea.

Tipranavir, the tenth and last HIV protease inhibitor to be approved (in 2008), is the only HIV protease inhibitor to adopt the displace strategy against the flap water. It does this with a cyclic lactone in an analogous manner to the cyclic urea. This highlights that the first strategy applied in drug discovery is not always the only option (especially when it comes to waters).