Institute of Biotechnology

DBDD - Research projects

The  Department  of  Biothermodynamics  and  Drug  Design performs fundamental and applied research focused on protein-ligand  interactions  and  drug  design.  The  state  of  the art  in  today’s  industrial  drug  design  is  still  based  on  highthroughput approaches due to the lack of fundamental understanding  of  physical  forces  underlying  such  processes  as protein folding and protein-ligand interactions. It is still impossible to predict and computer-model the compounds that would exhibit desired affinity and selectivity profiles towards their target proteins.

Carbonic anhydrases as drug targets

Carbonic anhydrases (CAs), a group of zinc containing enzymes,  are  involved  in  numerous  physiological  and  pathological processes, including gluconeogenesis, lipogenesis, ureagenesis, and tumorigenicity. In addition to the established role of CA inhibitors as diuretics and drugs used to treat glaucoma and high-altitude sickness, it has recently emerged that CA inhibitors could have potential as novel anti-obesity, anticancer, and anti-infective drugs (Supuran, 2008, 2012). CAs catalyze the conversion of CO₂ to the bicarbonate ion and protons.
There  are  12  catalytically  active  CA  isoforms  in  humans. CAs I, II, III, VII and XIII are cytosolic, CAs IV, IX, XII and XIV are membrane-bound and located on the outside of the cell, CAs VA and VB are found in mitochondria, and CA VI is the only secreted isoform found in saliva and milk. A number of CA inhibitors, mostly aromatic sulfonamides, have  been  designed  and  developed  into  drugs.  However, most  inhibitors  possess  low  selectivity  towards  the  target
CA  isoforms.  It  is  especially  important  to  develop  highly  selective  inhibitors  towards  the  novel  anticancer  target isoforms, CA IX and XII, that are highly overexpressed in numerous tumors and increase cancerous cell survival and metastatic invasiveness.
We have cloned and purified all human CAs or their catalytic domains in bacterial or mammalian cells. Over 600 novel compounds were designed and synthesized that bound CAs with micromolar to picomolar affinities. Four CA isoforms were crystallized in complex with numerous inhibitors and solved to high resolution thus providing structural insight into compound affinity and selectivity. A series of fluorinat ed CA inhibitors exhibited high affinity and great selectivity towards CA IX isoform [1]. Several other series of compounds were determined to bind various CA isoforms.
However, there are several linked reactions that occur simultaneously with the binding reaction. Such linked reactions greatly  influence  the  observed  thermodynamic  parameters of binding. For example, affinities are greatly dependent on pH, the enthalpies of binding – on the buffer in solution. Therefore,  we  determine  the  intrinsic  thermodynamic  parameters  of  binding  that  are  independent  of  experimental conditions and could be directly correlated with structures.

Discovery and characterization of novel selective inhibitors of carbonic anhydrase IX

 A series of fluorinated benzenesulfonamides with substituents on the benzene ring were designed and synthesized. Several of these exhibited a highly potent and selective inhibition profile against CA IX (Figure 1, Table 1). Three fluorine atoms significantly increased the affinity by withdrawing electrons and lowering the pKa of the benzene sulfonamide group. The bulky ortho substituents such as cyclooctyl or even cyclododecyl groups fit to the hydrophobic pocket in the active site
of CA IX but not CA II, as shown by the compound cocrystal structure with the chimeric CA IX (Figure 2). The strongest inhibitor of recombinant human CA IX catalytic domain produced  in  human  cells  achieved  the  affinity of 50pM.
However,  the  high  affinit y diminished the selectivity . The most selective compound for CA IX exhibited 10 nM affinity. The compound which showed the best balance between affinity and selectivity properties bound with 1 nM affinity. The inhibitors described in [1] provide the basis for novel anticancer therapeutics targeting CA IX.

Figure  1.  The  top  panel  shows  the  chemical  structures  of  CA  inhibitors 1-6.  Acetazolamide  (6,  AZM)  is  commonly  used  as  a  control  inhibitor of  CAs.  Panels  A,  B  and  C  show  compound  1  (graphs  on  the  left)  and 3  (graphs  on  the  right)  binding  and  inhibition  of  CAs.  A.  Binding  of compounds  as  determined  by  the  thermal  shift  assay.  Datapoints  show the    Tm s as a function of total added compound concentration while the lines are simulated according to 42. Red filled squares – CA IX, magenta open  squares  –  chCA  IX,  black  filled  triangles  –  CA  II,  and  blue  filled circles – CA I. The largest  Tm shift for similar proteins corresponds to strongest binding Kd. The inset graphs show normalized raw fluorescence data  as  a  function  of  temperature  at  zero  (filled  red  diamonds)  and 50  μM  (open  red  triangles)  total  added  compound  concentrations.  The melting midpoints correspond to the Tm. B. Binding of the compounds as determined by the isothermal titration calorimetry. Colors and symbols for CA isoforms are same as in panel A. The ITC curve fitting Kds are listed in Table 1. Insets show the raw ITC curves of the respective compound binding  to  CA  IX.  C.  The  inhibition  of  CA  isoforms  as  determined  by the stopped-flow kinetic CO2 hydration assay. Colors and symbols for CA isoforms are same as in panel A. Datapoints correspond to % inhibition of a CA as a function of total added compound concentration. The lines
are fit according to the Morrison equation as explained in the materials and methods section. Insets show raw activity curves (drop in absorbance/pH  due  to  acidification  by  the  CA  IX)  at  various  added  compound concentration:  magenta  –  0  nM,  cyan  –  15.6  nM,  violet  –  31.3  nM, and  green  –  spontaneous  CO2  hydration  in  the  absence  of  CA  IX.  The
CA IX concentration was 20 nM. All three methods conclusively indicate that both compounds 1 and 3 bound and inhibited CA IX significantly stronger than CA I and CA II. Furthermore, compound 3 bound tighter to  most  CA  isoforms  than  1.  However,  compound  1  exhibited  greater selectivity ratio towards CA IX than 3.

Figure 2. Compounds 1 (Panel A, PDB ID 4Q06) and 3 (Panel B, PDB ID  4Q07)  bound  to  chCA  IX  as  determined  by  X-ray  crystallography. The  Zn  is  shown  as  a  blue  sphere,  while  the  histidine  residues  holding the Zn atom are transparent. The amino acids of chCA IX are shown in grey.  The  terminal  atoms  of  amino  acids  which  form  the  hydrophobic cavity are shown as CPK (light grey). Several atoms of cyclooctyl group are also shown as CPK (dark grey). Dashed lines connect the atoms that make hydrogen bonds or electron donor-acceptor interaction (with Zn). Water molecule is shown as a red sphere. The compounds are shown in lightsteelblue.

Table 1. Compound dissociation constants for all 12 catalytically active human CA isoforms, determined by FTSA, ITC, and stopped-flow kinetic inhibition assay. X-ray crystallographic structures PDB IDs of available structures are listed.

Observed and intrinsic thermodynamics of binding compounds to CAs

Figure 3 shows the reactions occurring simultaneously upon ligand binding to the active site of CA. Direct experimental observation of the binding will always yield only the observed parameters of binding. However, only the intrinsic parameters are meaningful if we intend to analyze any structure-activity relationships. As described in [10] and [4], the dissection of these linked reactions is a laborious process requiring numerous experiments applying not only the fluorescent thermal shift assay (differential scanning fluorimetry, ThermoFluor), but also isothermal titration calorimetry and requires large amounts of purified protein.
After detailed dissection of linked reactions and estimation of the intrinsic parameters, the maps can be drawn that compare the binding thermodynamics of similar compounds (Figure 4) and correlate the binding thermodynamics with protein-ligand crystal structures (Figure 5).

 Figure 3. The observed and intrinsic binding thermodynamics. The upper panel  shows  the  main  linked  reactions  occurring  upon  ligand  binding to CAs. The lower panel lists the enthalpies of all processes linked to the binding of 3b to CA I. The two left-central reactions show the bindinglinked deprotonation of the inhibitor sulfonamide and the protonation of  the  zinc-bound  hydroxide,  respectively.  Top  and  bottom  lines  show linked  phosphate  buffer  (de)protonation  reactions.  Th e number s give estimates of the enthalpies for each process multiplied by the number of linked protons (n) yielding the observed enthalpic contribution of each reaction at pH 7.0, 37 °C. The intrinsic enthalpy of binding, shown by the rightmost arrow, is equal to −51.98 kJ/mol. The observed enthalpy, estimated for phosphate buffer at pH 7.0, is equal to −27.90 kJ/mol. Zinc atom is shown as grey shaded sphere and the carbonic anhydrase protein is shown as CA.

Figure  4.    Inhibitor  structure  correlations  with  the  thermodynamics  of binding. Intrinsic parameters of compound binding to five investigated CA isoforms are given within the shapes. Different colors represent different CA  isoforms.  Numbers  next  to  arrows  show  the  Gibbs  free  energy  (top number, bold), enthalpy (middle number), and entropy (TD_bS, bottom number) of binding differences between two neighboring compounds (in kJ/mol at 37 °C). Numbers to the top and right of the map are averages between same heads and tails of the compounds. The standard deviations indicate the presence and absence of the energetic additivity of compound functional groups.

Figure 5. Compound chemical structure and the thermodynamics of binding correlations with the crystal structures of some compound binding mode in the active site of CAs (1a, 1d, 3a, 3b, 3c, 4a, 4b, and 4c with CAII; 4a, 4b, and 4c with CA XIII; 2b and 4b with CA XII). The thermodynamic parameters of binding and the colors of the shapes are same as in Figure 4 and indicate the CA isoform. Colors in the crystal structures are: yellow shows the pyrimidine ring that is not fixed in the crystal structure and has multiple conformations with low occupancies; blue shows the alternative conformation of the pyrimidine ring when both conformations are visible in the electron density maps.

Ligand binding to proteins at high pressure

The volume changes accompanying ligand binding to proteins  are  thermodynamically  important  and  potentially could be used in the design of compounds with specific  binding  properties.  Measuring  the  volumetric  properties could yield as much information as the enthalpic properties of binding. Pressure-based methods are significantly more laborious than temperature methods and are underused. The pressure shift assay (PressureFluor, analogous to the ThermoFluor, thermal shift assay, differential scanning fluorimetry)  uses  high  pressure  to  denature  proteins.  The PressureFluor  method  was  used  to  study  the  ligand  binding thermodynamics of Hsp90 and human serum albumin. Ligands stabilize the protein against pressure denaturation, similar to the stabilization against temperature denaturation
(Figure 6).

Figure 6. The Gibbs free energy dependence on pressure and temperature. Inner  surface  represents  the  ligand-free  Hsp90N  stability  region,  while outer surface shows stability region of protein-ligand system with 200 μM of added ligand [13].

The Team of amyloid research

We  are  especially  interested  in  amyloid-like  nature  of  prions and  prion-like  nature  of  amyloids.  Protein  aggregation  and amyloidogenesis  are  involved  in  a  number  of  diseases,  including  such  neurodegenerative  disorders  as Alzheimer’s  and Parkinson’s, many systemic amyloidoses and even some localized diseases such as type II diabetes or cataracts. There is an increasing evidence of amyloid nature of proteinaceous infectious particles – prions. One of the most possible ways of abnormal protein spreading is elongation of amyloid-like fibrils, thus there is a chance of all amyloid associated diseases to be potentially infective. The same prion protein may express distinct strains. The strains are enciphered by different misfolded conformations. Strain-like phenomena have also been reported in a number of other amyloid-forming proteins. One of the features of amyloid strains is the ability to self-propagate, maintaining a constant set of physical properties despite being propagated under conditions different from those that allowed initial formation of the strain.
Our most important findings are summarized in Figure 7. We did cross-seeding experiment using strains formed under different conditions. Using high concentrations of seeds results in rapid elongation and new fibrils preserve the properties of the seeding fibrils. At low seed concentrations secondary nucleation  plays  the  major  role  and  new  fibrils  gain  properties predicted by the environment rather than the structure of the seeds. Our findings could explain infectious prion evolution in vivo and conformational switching between amyloid strains observed in a wide variety of in vivo and in vitro experiments.

Figure 7. Conformational switching between amyloid fibrils.

Services

 The DBDD is seeking to license out the compounds described in patents and patent applications. The DBDD is interested in collaborations where our expertise in recombinant protein production and the determination of compound – protein binding thermodynamics and recombinant protein stability characterization could be applied. Protein – ligand binding constants and protein thermal stability profiles at hundreds of conditions may be determined in a single experiment by consuming microgram quantities of protein.