MXPA00008696A - Ligand screening and design by x-ray crystallography - Google Patents

Abstract

X-ray crystallography can be used to screen compounds that are not known ligands of a target biomolecule for their ability to bind the target biomolecule. The method includes obtaining a crystal of a target biomolecule;exposing the target biomolecule crytal to one or more test samples;and obtaining an X-ray crystal diffraction pattern to determine whether a ligand/receptor complex is formed. The target is exposed to the test samples by either co-crystallizing a biomolecule in the presence of one or more test samples or soaking the biomolecule crystal in a solution of one or more test samples. In another embodiment, structural information from ligand/receptor complexes are used to design ligands that bind tighter, that bind more specifically, that have better biological activity or that have better safety profile. A further embodiment of the invention comprises identifying or designing biologically-active moieties by the instant process. In a further embodiment, a biomolecule crystal having an easily accessible active site is formed by co-crystallizing the biomolecule with a degradable ligand and degrading the ligand.

Description

SELECTION AND DESIGN OF LINKING THROUGH X-RAY CRYSTALGRAPHY TECHNICAL FIELD OF THE INVENTION X-ray crystallography is useful for identifying ligands that bind to receptor molecules, for which binding is intended, and for designing ligands with enhanced biological activity, for the target recipient.

BACKGROUND OF THE INVENTION X-ray crystallography (crystallography) is an established, well-studied technique that provides what can best be described as a three-dimensional image of what a molecule would look like in a crystal. Scientists have used crystallography to solve crystal structures for many biologically important molecules. Many kinds of biological molecules can be studied by crystallography, including, but not limited to: proteins, DNA, RNA and viruses. Even scientists have reported crystalline structures of biological molecules that carry ligands within their receptors (a "ligand-receptor complex"). Given an "image" of a target biological molecule (biomolecule) or a ligand-receptor complex, scientists can search for receptacles or receptors in which biological activity can take place. Then scientists can design experimentally or by computer, ligands (or drugs) of high affinity for the receptors. Alternately, computational methods have been used to select the binding of small molecules. However, these previous attempts have had limited success. There are several problems that afflict the design of ligands by computer methods. The computation methods are based on estimates rather than exact determinations of the binding energies, and are based on simple calculations, compared to the complex interactions that exist within a biomolecule. In addition, computer models require experimental confirmations that frequently expose the models as false positives that do not work in the real destination. In addition, the high-affinity, experimental ligand design, based on an "image" of the ligand-receptor complex, has been limited to biomolecules that already have known ligands. Finally, scientists have only recently reported the crystallographic study of interactions between organic solvents and target biomolecules. Alien and coauthors, J. Phys. Chem., V.100, pp. 2605-1 (1196). However, these studies are limited to mapping solvent sites, rather than to ligand sites. It would be convenient to directly identify the potential ligands and obtain detailed information on how the ligand binds and changes in the target biomolecule. Additionally, methods for identifying and / or designing ligands possessing biological and / or pharmaceutical activity with respect to a given target molecule would be convenient.

BRIEF DESCRIPTION OF THE INVENTION Crystallography can be used to select and identify compounds that are not known ligands, from a target biomolecule, in terms of their ability to bind to the target. The method (hereinafter referred to as "CrystaLEAD" ™) comprises obtaining a biomolecule of interest (of destination); expose the destination to one or more test samples that are potential ligands of the destination; and determining whether a ligand / biomolecule complex is formed. The target is exposed to potential ligands by various methods including, but not limited to: soaking a crystal in a solution of one or more potential ligands or co-crystallizing a biomolecule in the presence of one or more potential ligands. In another embodiment, the structural information of the ligand / receptor complexes is used to design new ligands that bind more strongly and that are more specific binding, have better "biological activity or have a better safety profile, than known ligands. a preferred embodiment is used compound banks "diversity of forms" to allow direct identification of the ligand-receptor complex, even when the ligand is exposed as part of a mixture. Here three steps are simultaneously achieved: The calculated electron density function directly reveals the binding event, identifies the bound compound and provides a detailed three-dimensional structure of the ligand-receptor complex. In a modality, once a success has been found, many analogs or derivatives of the hit could be selected for stronger union or better biological activity, by means of traditional discrimination methods. Another modality uses the success and information about the structure of the destination to develop analogs or derivatives with stronger union or better biological activity. In yet another embodiment, the ligand-receptor complex is exposed to additional iterations of potential ligands, so that two or more hits can be linked together to form a more potent ligand.

BRIEF DESCRIPTION OF THE DRAWINGS , Figure 1 illustrates a structure-based drug design, where an initial header compound is found, and then used as scaffolding to carry additional portions that fit into the subsites surrounding a main site. Figure 2 illustrates a fragment link approach for a biomolecule having two or more adjacent primary receptacles. Figure 3 is a CrystaLEAD ™ scheme where a crystal is soaked in a solution of various potential ligands (l | -l? O) and a series of diffraction data is collected, transformed into electron density map, which it is inspected for the union of the compound. Figure 4 illustrates a typical mixture of composite in two dimensions and in three dimensions. The three-dimensional figures are theoretical electron density maps 2Fo-Fc, which represent the "shape" of the molecules. Figure 5 is a primary sequence of human urokinase. Figure 6 illustrates how a hit was detected and was identified by the shape after urochinasa was soaked in a solution containing a mixture of potential ligands. Figure 6A is the initial map Fo-Fc. Figure 6B shows how the compound binds to the active site of urokinase. Figure 6C illustrates the active site without a bound ligand, when no compound of the mixture has been bound. Figure 7 illustrates a hit for urokinase soaked in a solution containing a mixture of potential ligands. Figure 7A is the initial map Fo-Fc. Figure 7B shows how the compound binds to the active site of urokinase. Figure 8 illustrates a hit for urokinase soaked in a solution containing a mixture of potential ligands. Figure 8A is the initial map Fo-Fc. Figure 8B shows how the compound binds to the active site of urokinase.

Figure 9 illustrates two additional hits for urokinase soaked in a solution containing a mixture of potential ligands. Figure 9A is the Fo-Fc map for a strong ligand within the mixture. Figure 9B is the Fo-Fc map for a weaker ligand within the mixture. The weakest ligand was detected only after the strong ligand was separated from the mixture. - Figure 10 illustrates the comparative crystal structures between a leader compound, found by CrystaLEAD and an optimized follower compound. Figure 11 illustrates successes that were identified for VanX. Figure 12 illustrates a pair of urokinase hits. Figure 13 illustrates the crystal structure of compound 44, with ErmC Figure 14 illustrates the crystal structure of compound 45 with ErmC DETAILED DESCRIPTION OF THE INVENTION.

CrystaLEAD ™ gives an efficient screening method to identify compounds that will bind to a target biomolecule. These compounds can serve as headers or scaffolds to design ligands and / or drugs that have improved biological activity for the target. It should be noted that ligands that bind more strongly do not necessarily provide better biological activity or constitute a better drug, although this usually happens. It is possible that a ligand that binds weaker provides better biological activity due to factors other than strong binding (eg, selectivity, bioavailability). Crystallography has been used extensively to view receptor-ligand complexes in terms of its structure-based drug design. To see such complexes, usually known ligands are soaked in the crystal of the molecule of interest (target), and then the crystallography of the complex is carried out. Sometimes it is necessary to co-crystallize the ligands with the target molecule, to obtain an adequate crystal. Until now, crystallography has not been implemented to select potential ligands, despite the detailed structural information provided. Possible prejudices against the selection of compounds by crystallography include the belief that the method is too complicated or time consuming, and that it is difficult to obtain adequate crystals.; that the available crystals could not tolerate the soaking of more than one compound (much less mixtures of ten or more compounds), that many biomolecules would be needed, that the routine assembly of the crystals would be too slow, and that the crystals would change constantly in the X-ray goniometer would be too tedious. However, the technology currently available has solved many of these perceived barriers. For example, at some time, molecular targets or targets were only obtained from natural sources and were sometimes unsuitable for crystallization due to natural degradation or glycosylation. In addition, often the natural concentration was too low to obtain the amount of highly purified protein necessary for crystallization. With molecular biology, large amounts of protein can be expressed and purified for crystallization. When necessary, you can even engineer the protein to give different or better forms of crystal. Also available are bright light sources (synchrotron radiation) and more sensitive detectors, so that the time required to collect the data has been dramatically reduced from days to hours and even minutes. In addition, existing technologies, which are less routine at this time, will soon become routine, allowing complete collections of data series in the order of seconds, or even fractions of a second (for example, Laue diffraction). J. Hajdu and co-authors, Nature, v. 329, pp. 178-81 (1987). Faster computers and more automatic application programs have gre reduced the time required for data collection and analysis. Finally, the inventors have discovered that it is possible to soak or co-crystallize mixtures of compounds to select potential ligands. In such a way, as described in the following, crystallography is now a practical and feasible method of selection.

In CrystaLEAD ligands are identified for a target molecule, which has a crystalline form, by exposing a bank of small molecules, either individually or in mixtures, to the target or target (eg, protein, nucleic acid, etc.). Then the crystallographic data is obtained to compare the electron density map of the putative target-ligand complex with the electron density map of the target biomolecule. The electron density map simultaneously provides direct evidence of ligand binding, identification of the ligand bound and the detailed three-dimensional structure of the ligand-target complex. The junction can also be monitored by changes in individual reflections within the crystallographic diffraction pattern, which are known to be sensitive to ligand binding at the active site. This could serve as a pre-selection, but it would not be the primary method of choice, because it provides less detailed structural information. By observing the changes in the level of electron density of the ligand, or the intensity of certain reflections in the diffraction pattern, as a function of the concentration of ligand, either added to the crystal or in co-crystallization, one can also determine the binding affinities of the ligands for the biomolecules. The binding affinities can also be obtained by means of competitive experiments. Here the new compound or the new compounds are soaked or co-crystallized with a ligand or a series of ligands of various configurations, with known binding affinity. If the known ligand appears on the electron density map, the unknown ligands are weaker linkers. However, if one of the new compounds is found to be competing for the site, it would be the strongest linker. By varying the concentration or identity of the known ligand, a binding constant can be estimated for the CrystaLEAD hit. The number of selected compounds is based on the desired detection limit, the solubility of the compound and the amount of organic cosolvent that the crystals will tolerate. The exact numbers depend on each crystal. For example, for a typical crystal that tolerates 1% organic cosolvent, the sensitivity limit would be Kd < 1.5 mM to select 10 compounds simultaneously. For 20 compounds, the sensitivity limit would be Kd < 0.63mM. However, crystals that tolerate organic cosolvents in high amounts (for example 40%) can select up to 50 compounds within a detection limit Kd < 1.5 mM. In the more general application of CrystaLEAD ™, the hit or lead compound is used to determine which compounds should be tested for biological activity in a structure-based drug design. Derivatives and analogs are then obtained by traditional medical chemistry, to find the best ligand or drug of all. As an alternative, the structure information collected in the selection process can be used directly to suggest analogs or derivatives of the success. This approach is illustrated when the active site is composed of a primary receptacle surrounded by a variety of subsites and small receptacles (Figure 1). The detailed structural information about how a compound is bound by the receptor, it is obtained simultaneously when a hit is detected. This information is useful for the ordinary expert to design better ligands. P. Colman, Curr. Opln. in Struct. Biology, v. 4; pp. 868-74 (1994); J. Greer and coauthors, J. Med. Chem., V. 37, pp. 1035-54 (1994); C. Verlinde and coauthors, Structure, v.15, pp 577-85 (1994). In particular, the success identifies sites for analogue synthesis, which would allow access to surrounding subsites and small receptacles. This suggests the design of new compounds that fit better in the active site. In addition, in cases where there is an existing structure-function relationship, one can directly transfer the activity-enhancing substitution patterns, to the new scaffolding header, at the three-dimensional structural level. Another illustration (figure 2) usually applies to a destination that has two or more separate pockets, which will accommodate fragments. Here, the crystalline destination is selected in terms of ligands that occupy all the sites, either in sequence or simultaneously. Because the binding event is monitored by visualizing co-crystalline structures, the ligand binding site is directly identified, and there is no need for competitive experiments to ensure that the ligands actually occupy different sites in the protein. The selection allows separately that ligands that bind to different receptacles overlap at their binding sites. The selection for the second, in the presence of the first, would detect the cooperative union in a second site. Once a potential header and a structure-activity relationship have been established, links between each of the sites can be designed using the detailed structural information, and the fragment link approach, which was previously described, to produce novel ligands , much more powerful. S. B. Shuker and co-authors, Science, v. 274, pp. 1531-34 (1996); C. Verlinde and coauthors, Structure, v. 15, pp. 577-85 (1994). In a third application, the confluence approach to the scaffolding (not shown), the target active site is composed of two or more subsites. The crystalline protein is selected for one or more ligands that are joined by means of these subsites and the relative binding orientation of the ligand is observed, for multiple experiments. These ligands must be joined by occupying one or more subsites and overlapping the multiple-hit structures; a core can be designed to facilitate access to multiple subsites. This nucleus would then serve as a new, novel and more potent lead compound, which would also serve as a scaffolding leader in the drug design cycle. The fragment approach linked by CrystaLEAD implements experimentally the linked fragment-based structure approach, reported only at the computation level by Verlinde and co-authors in J. Comput. Aided Mol. Des., V. 6, pp. 131-47 81992). Verlinde and coauthors proposed fragments of ligand based on mathematical calculations. The proposed fragments were then analyzed for their binding activity. If the fragments are actually joined, their three-dimensional structures were determined by X-ray crystallography and a designed linker. In contrast, CrystaLEAD ™ concurrently detects the binding event and provides an experimentally designed three-dimensional structure of the ligand-protein complex. The invention also provides a process for determining the association constant between a target molecule and its ligand. The invention does not require special marking of the destination. Accordingly, the target molecule can comprise proteins, polypeptides, nucleic acids, nucleoproteins or any other suitable target molecule, which is isolated from natural sources or by recombinant methods, from any suitable host system, such as those developed and practiced by the ordinary technicians. There are several advantages in the crystallographic selection. An important advantage is that the union event is monitored directly, so that the probability of false positives is reduced to almost zero. The crystallographic data provide a "take" of the three-dimensional electron density of the ligand-receptor complex, which shows that said compound binds and how it binds.

The method is uniquely sensitive to structural changes, both in the target and in the ligand. The observation of structural changes is critical to design the scaffolds that combine the information of the structures of different ligand-target complexes. One such example occurs when a protein changes structure in order to accommodate a ligand, but the structure change concurrently blocks the binding of a second ligand. Similarly, it is important to detect structural changes because if the primary scaffolding is attached differently, it may not be possible to combine them into larger scaffolding. Since the binding event is monitored directly, CrystaLEAD ™ does not require samples with special labels, probes or target molecules that are indirectly sensitive to the ligand association. As long as a crystalline structure of the target can be obtained, CrystaLEAD can be used to select the ligands. If the compound mixtures are suitably designed to have a variety of shapes, the invention alleviates the need to undo the convolutions of banks that are soaked as a mixture, because the binding event is detected directly by examining the shape of the density of the mixture. electrons in the binding site. Thus, the shape of the electron density identifies both the binding event and the identity of the compound, directly. Alternatively, the mixture can be designed to contain compounds with anomalously diffracting atoms (eg, Br, S) that can be identified by anomalous diffraction techniques. Additionally, because CrystaLEAD directly monitors the junction, it is particularly well suited to study destinations in which there are no known ligands. Because the electron density function calculated in CrystaLEAD shows the "real space" of the crystal, you can focus directly on the region of interest. Thus, the binding can be detected exclusively at the site of interest, although the method is not limited to the active site. The binding in other sites, which complicates the analysis in most binding analyzes, can be totally eliminated. CrystaLEAD ™ also provides a method to concurrently monitor the junction at different sites. That is, for a destination with more than one receptacle, the selection for a second site does not require selection in the presence of the first ligand. However, selection for a second site in the presence of the first ligand can be completed in order to discover cooperating ligands. CrystaLEAD ™ is applicable to any target molecule for which a crystalline structure can be obtained. According to current literature, this includes any soluble macromolecule with molecular weight between about 5,000 and 200,000. However, this scale extends almost daily in response to technological advances. The method is also sensitive to a large variety of binding dissociation constants (&pictolar to molar). Using more sensitive CCD camera detectors, data can be collected in less than about 4 hours to 4 hours, with a rotating anodic source. This allows the selection of thousands of compounds per detector, per day. Using synchrotron sources, the number of selected compounds increases to many thousands per detector; and with Laue data collection methods and test mixes, CrystaLEAD ™ data can be collected in a second or less, allowing thousands of compounds to be tested per day, per line of lightning. Therefore, multiple detectors, or a single synchrotron beam line, facilitate true selection at high production. Figure 3 schematizes the invention. Crystals of the target molecule are exposed to one or more compounds, by soaking the crystal or by co-crystallizing the target in the presence of one or more compounds. Then the crystallographic data is collected, they are processed and converted to electron density maps, which are examined for the evidence of ligand binding. One way to detect ligand binding is to compare the structure of the original crystal with the structure of the exposed crystal. New destinations can be crystallized by published conditions or by other methods well established in the art. Similarly, target structures can be available from databases, such as the Protein Data Bank, or could be determined by well-established methodology. Advances in molecular biology and protein engineering will facilitate the crystallization of the destination, while advances in data collection will help quickly determine the structure for targets or targets of previously unknown structure. Crystals that are exposed to potential ligands by "soaking require an accessible, empty active site." Crystals with an empty active site can be obtained by various methods, including, but not limited to: crystallization in the absence of a ligand; crystallization in the presence of bound ligand at a distal site; or crystallization in the presence of a non-covalent ligand, which is easily diluted or changed from the target, once the biomolecule crystallizes. By a novel method, the inventors have obtained crystals of a biomolecule by crystallizing the biomolecule in the presence of a degradable ligand as an active site, and then degrading the ligand, once a crystal is formed. Alternatively, it is possible to develop the crystals in the presence of the compounds to be selected. The crystals are allowed to equilibrate in the presence of the mixture, at which point the ligands bind as a function of their concentration and their binding affinity. For the soaking method, the sensitivity of the method can be approximated by simple equilibrium ratios because the concentration of the protein in the crystal can be calculated and the concentration of the ligand is a known quantity. For example, the concentration of a 25,000 molecular weight protein (urokinase) in a crystal is calculated as follows: There are four molecules in the orthorhombic unit cell (all angles are 90 °) that have a volume of 55 x 53 x 82 A3, using the Avogadro number the concentration is 28 mM. Accordingly, a mixture of compounds having a concentration of 6 mM for each ligand, will result in a calculated sensitivity limit of Kd < 1.5 mM (assuming a detection limit of around 80% occupancy in the crystal). Soaking mixtures of compounds also brings about the issue of multiple occupancy (more than one ligand that binds to the site of interest). For cases of multiple occupancy, where the ligands are bound in different receptacles (see Figure 2), resolution by CrystaLEAD ™ is easy because the binding at the separate sites can be distinguished individually by electron density maps. For the scenario where different ligands compete for the occupation of the same site, a simple competitive inhibition model can be used to calculate the requirements for said union. From empirical observation, it is believed that crystallography can solve situations in which the occupation of an inhibitor is 80% and another 20%. Therefore, a binding affinity ratio that is greater than four would result in an apparent occupancy for only the highest affinity ligand. In the unlikely event that the rate of binding constants of two compounds in the mixture is less than four, the resulting electron density would be an average weight of the two separate densities, and could be difficult to identify. Consequently, it would be necessary to perform additional soaking experiments to undo the convolution of the mixture (for example, looking at each individual compound, in separate crystals) only when the ratio of binding affinities is less than four. This would still be worthwhile and efficient, since it easily determines that there are at least two hits in the mix. The compounds that are to be selected are formed into banks. For the purposes of this discussion, banks are large mixtures of compounds (100 to 10,000 +) and can be general or directed to the structure. A general bank is random; that is, totally different in size, shape and functionality. A bank directed to the structure is intended for a particular functional mixture or a particular subsite in the active site of the target molecule (for example, a bank in which all compounds contain a carboxylated functionality that is to be directed towards a positive load on the destination active site). In a preferred embodiment, any type of bank is divided into smaller groups of mixtures of different form. Accordingly, a mixture is defined as a subseries of the bank, which can be soaked or developed to the glass. The mixture is determined in terms of shape diversity by visual inspection of two-dimensional chemical structures, or by computer, through programs. The diversity of shape of the mixture allows to directly identify a bound ligand from the resulting electron density map (see Figure 4). This eliminates the need for follow-up experiments to determine which compound of the mixture is a hit (it is bound to the target). If the test compounds are soluble in water, typical regulators and precipitant solutions used in the crystallization can be used to solubilize the mixtures and soak them in the glass. The least water soluble compounds are dissolved individually at a final concentration of 2M in a suitable organic solvent. In one embodiment, they are dissolved in 100% DMSO and stored at 4 ° C, and mixed by mixing the DMSO materials before exposing the crystal. These mixtures would serve most glass systems, when the conditions for the development of crystals do not include organic reagents. The compounds would typically be soaked at a final DMSO concentration of 1-10% and allowed to equilibrate with the crystal protein for a predetermined amount of time (4-24 hours). Under this scenario, each crystal is exposed to multiple compounds by soaking mixture. Some glass growth conditions may include a high concentration of organic solvent (40-50%), which are typical alcohol derivatives. In this case the compound banks can be dissolved in the organic crystallization solvent, which would allow a final cosolvent concentration of 40-50% for the soaking experiment. Here you could increase the number of compounds per soaking mixture. After soaking, each crystal is exposed to a protective cycle, such as 5-20% glycerol in the soaking mixture, mounted on a nylon circuit and placed in the X-ray unit under a stream of nitrogen (160K). It is also possible to carry out crystal studies at room temperature or under other suitable conditions, as necessary for the stability of the crystals, automatic assembly and glass change equipment can be used to accelerate this step of the process. crystallographic data and processed when each reflection (point) in the diffraction pattern is assigned an index (h, k I) and the intensity is measured as a norm in the field X-ray sources can be X-ray generators laboratory or synchrotron sources of great brilliance that allow the collection of diffraction data at very high speed, Specifically, the collection of laboratory data can take from 30 minutes to several hours per crystal, although the time can be reduced using synchrotron sources. could reduce data collection by crystal to fractions of a second using Laue data collection schemes. Then we turn to electron density maps, using methods that are familiar to ordinary experts. Electron density maps are three-dimensional images of the ligands and / or target biomolecules. For Fo-Fc maps, we subtract the calculated factor factor amplitudes (| Fc |) that are obtained from the known crystal structure, without bound ligand, of the observed amplitudes (| Fo |). Thus, this map represents a direct subtraction of the data that arise from the natural protein structure from the data that arise from the crystals soaked in the presence of a bank mixture. The result is an electron density map that has positive and negative peaks. The peaks relevant to CrystaLEAD ™ are positives, which are the direct result of ligand binding at the target site of interest (ie, the addition of the ligand to the target biomolecule). In Figure 3, the Fo-Fc map clearly shows a large positive peak at the active site of urokinase. The shape of the peak corresponds to the ligand 2-amino-8-hydroxyquinoline. The ligand is shown occupying the density of positive difference. The other positive peaks correspond to the bound sulphate portion (indicated by SO 2") and the water molecules attached (indicated by H2O) This type of map is also very sensitive to small structural changes, indicated by?) that, when used in conjunction with the 2Fo-Fc maps, allows the determination of the detailed structure of the entire ligand-protein complex. To calculate 2Fo-Fc maps, subtract (| Fc |) from 2 (| Fo |). Here the map is positive and has density for all the atoms of the molecule. In Figure 3 the inspection of the map indicates the identity and structure of the bound compound. It is preferred that the maps of the exposed crystals be compared with the maps of the target molecule without exposing, to differentiate the positive density that can be found in the Fo-Fc map. Sometimes water molecules occupy the active site in the crystal, in the absence of a bound ligand. This differs easily because the bound water molecules are often oriented in a geometry consistent with hydrogen bonding, and because they are not connected by a network of covalent bonds. Thus, the resulting map tends to be disconnected, which indicates bound solvent, rather than an organic compound. If it is determined that the density in the Fo-Fc or 2Fo-Fc map represents an organic compound, the three-dimensional form of that compound is compared with that of the compounds present in the bank, and an equalization of best fit is made. Alternatively, programs such as QUANTA's XFIT modules (Molecular Simulations Inc., Quanta Generating and Displaying Molecules, San Diego; Molecular Simulations Inc., 1997), can automate this process. As the ability to measure or process diffraction intensities improves, it may not be necessary to make the comparison on electron density maps. The junction can be detected by simply comparing the diffraction patterns of the exposed crystals with the unexposed crystals. Therefore, it is necessary to create an electron density map only if a binding event is detected in this preselection process.

As shown above, it can be applied CrystaLEAD ™ to any biomolecular destination for which the crystallographic structure can be obtained. Due to its broad applicability, this is best illustrated by the examples that follow. The urokinase and VanX examples represent two scenarios for the use of CrystaLEAD ™. For urokinase, microUK (μUK) crystals diffract very well and are from a group of very symmetrical space. In contrast, VanX crystals diffract more weakly and with less symmetry. Thus, VanX requires more data collection time. Additionally, the urokinase crystals have one molecule in the asymmetric unit, while VanX has six. The larger asymmetric unit requires higher resolution data collection and makes map inspection more tedious. However, in the case of VanX, no mimetic binder without substrate was known before those discovered by CrystaLEAD ™. Accordingly, CrystaLEAD ™ provided a novel non-peptide header compound to be fed into the drug discovery cycle. For urokinase, CrystaLEAD ™ provided a novel primary scaffolding. Applicants were able to rapidly increase the potency of the primary scaffold using existing SAR and crystalline structures to design higher affinity derivatives with improved bioavailability relative to known urokinase ligands. However, these examples illustrate the preferred embodiment of the present invention and do not limit the claims or description. Whoever is an ordinary expert will easily appreciate that changes and modifications can be made in the specified modalities, without departing from the spirit and scope of the invention. Finally, all the citations of the present are incorporated in it by means of the reference.

EXAMPLES EXAMPLE 1 UROCINASA Urokinase, a serine protease, is strongly associated with tumor cells. Urokinase activates plasminogen in plasmin which, in turn, activates matrix metalloproteases. Plasmin and metalloproteases degrade the extracellular matrix and promote the development and metastasis of the tumor. Thus, inhibitors that target specifically in urokinase can serve as effective anticancer agents. Human pro-urokinase consists of 411 amino acids (figure 5). Verde and coauthors, Proc. Nat'l. Acad. Sci., V.81 (5), pp. 4727-31 (1984); Nagain and co-authors, Gene, v. 36 (1-2), pp. 183-8 (1985). When activated by proteolytic cleavage in the peptide ligation Lys158-lle159, the enzyme becomes two chains connected by a single disulfide bridge (Cys148-Cys279). Chain A (residues 1-158) contains an EGF-like domain and a "kringle" domain. Chain B (residues 159-411) contains the catalytic domain of serine protease. Further incubation of the urokinase results in an additional proteophytic cleavage in the Lys135-Lys136 peptide ligation to form low molecular weight urokinase. The crystals of this enzyme form, in complex with the covalent inhibitor Glu-Gly-Arg, were obtained by Spraggon and coauthors, Structure, v 3, pp. 681-91 (1995) and they were shown to diffract at a resolution of 2.5A in a high energy synchrotron source. However, the poor diffraction quality of these crystals, together with the presence of a covalent binding inhibitor, makes the application of CrystaLEAD ™ difficult.

PREPARATION AND STRUCTURE OF μUK CRYSTAL To implement CrystaLEAD ™, human urokinase was engineered to be consisted of only residues 159-404 of the B chain, where Asn302 was replaced by a glutamine to remove a glycosylation site, and Cys279 was replaced with an alanine to eliminate the free sulfhydryl portion. This form of urokinase (μUK) proved to be fully active and was found to crystallize in a crystal form compatible with CrystaLEAD ™. (See also U.S. Patent No. 5,112,744, issued May 12, 1992 to Heyneker and co-inventors).

PREPARATION OF VECTOR CONSTRUCTION pBC-LMW-UK- Ala279 Human UK mutant was cloned into a dicistronic bacterial expression vector pBCFK12. Pilot-Matias and co-authors, Gene, v.128, pp. 219-25 (1993). The following oligonucleotides were used to generate various UK mutants by PCR: No. Sequence SEQUENCE SENSITIZER RCP 1.5'-2.5'-ATTAATGTCGACTAAGGAGGTGATCTAATGTTAAAATTTCAGTGTGGCCAA-3 ATTAATAAGCTTTCAGAGGGCCAGGCCATTCTCTTCCTTGGTGTGACTCCTGATCCA-3 '3.5'-ATTAATTGCGCAGCCATCCCGGACTATACAGACCATCGCCCTGCCCT-3' was carried out the initial cloning of a low molecular weight UK, hereinafter referred to as LMW-UK (L 44-L411), using as template UK cDNA and as sensitizers SEQ ID NO. 1 and 2, in a normal CPR reaction. The DNA amplified by PCR was purified in gel and digested with restriction enzymes Salí and Hindlll. The digested product was then ligated into a pBCFK12 vector, previously cut with the same two enzymes, to generate the expression vector pBC-LMW-UK. The vector was transformed into DHdalfa cells (Life Technologies, Gaithersburg, MD, E.U.A.), isolated and the sequence confirmed by DNA sequencing. The production of LMW-UK in bacteria was analyzed by SDS-PAGE and zymographically. Granelli-Piperno and co-authors, J. Exp. Med., V 148, pp. 223-34 (1978), which measures plasminogen activation by UK. That LMW-UK was expressed in E. coli and was shown to be active in the zymographic analysis by gel stained with Commassie blue. Successful rapid expression and detection of LMW-UK in E. coli made it possible to perform the UK mutagenesis analysis in order to determine its minimal functional structure. A mutant was made that had replacement of Cys279 to Ala279 with SEQ ID NO. 2 and 3, by means of RCP. The PCR product was cut with Avill and Hindlll and used to replace an Avill and Hind 111 fragment in the construction of pBC-LMW-UK. The resulting construct pBC-LMW-UK-Ala279 was expressed in E. coli and the product proved to be active in zymography.

CLONING AND EXPRESSION OF uUK (UK (I159-K404) A279Q302) IN BACULOVIRUS UUK (UK with amino acids lle159-Lys404, containing Ala279Gln302) was generated by PCR, with the following oligonucleotide sensitizers: SEQ ID # RCP SENSITIZER SEQUENCE 4 5'-ATTAATCAGCTGCTCCGGATAGAGATAGTCGGTAGACTGCTCTTTT-3 5. 5'-ATTAATCAGCTGAAAATGACTGTTGTGA-3 '6 5'-ATTAATGTCGACTAAGGAGGTGATCTAATGTTAAAATTTCAGTGTGGCCAA-3 '7. 5'-ATTAATGCTAGCCTCGAGCCACCATGAGAGCCCTGCT-3 '8. 5'-ATTAATGCTAGCCTCGAGTCACTTGTTGTGACTGCGGATCCA-3' 9. 5'-GGTGGTGAATTCTCCCCCAATAATGCCTTTGGAGTCGCTCACGA-3 'To mutate the only glycosylation site (Asn302) in UK, oligonucleotide is sensitizers used SEQ ID NO: 4 and 6 and SEQ ID NO: 5 and 8, in two PCR reactions with pBC-LMW-UK-Ala279 as template. The two PCR products were cut with the restriction enzyme Pvu II, ligated with T4 DNA ligase and used as a template to generate LMW-UK-A279-Q302. In the meantime, the natural UK header sequence was directly fused by PCR with SEQ ID NO: 7 and 9 using UK cDNA as a template. This PCR product was used as a sensitizer, together with SEQ ID NO: 8 in a new PCR reaction with DNA from LMW-UK-A279-Q302 as template, to generate μUK cDNA. ΜUK was cut with Nhe I and ligated to a baculovirus transfer vector pJVP10z, cut with the same enzyme. Vialard and coauthors, J. Virology, v.64 (1); pp. 37-50 (1990). The resulting construct, pJVP10z-μUK, was confirmed by normal DNA sequencing techniques. The pJVP10z-uUK construct was transfected into Sf9 cells by the calcium phosphate precipitation method, using the BaculoGold equipment from PharMingen (San Diego, CA, E. U. A.). Active uUK activity was detected in the culture medium. The only recombinant virus expressing μUK was purified on plate by common methods and a large deposit of virus was made in storage. Large-scale expression of μUK was made in another insect cell line, High-Five cells (Invitrogen, Carlsbad, CA, USA) in suspension that develops in Ecel 405, serum-free medium (JRH Biosciences, LeneXa, KS), in two-liter flasks, shaking at 80 rpm, 28 ° C . High-Five cells were developed at 2 x 10 6 cells / ml, recombinant μUK virus was added at multiplicity of 0.1 infection, and culture was continued for three days. The culture supernatant was harvested as a starting material for purification. The activity of μUK in the culture supernatant was measured by amidolysis of a chromogenic UK substrate S2444, which was at 6-10 mg / liter. Claeson and coauthors, Haemostasis, v. 7, p. 76 (1978).

EXPRESSION OF uUK IN Pichia pastoris To express μUK in Pichia, an expression vector with a synthetic leader sequence was used. The expression vector in Pichia, pHil-D8, was constructed by modifying the vector pHil-D2 (Invitrogen) to include the synthetic leader sequence for the secretion of a recombinant protein. The header sequence 5'-ATGTTCTCTCCAATTTTGTCCTTGG AAATTATTTTAGCTTTGG CTACTTTGCAATCTGTCTTCGCTCAGCCAG TTATCTGCACTACC GTTGGTTCCGCTGCCGAGGGATCC-3 '(SEQ ID NO: 10) encodes a PHOI secretion signal (indicated by the single underline), operably linked to a propeptide sequence (indicated by the bold letters) for the KEX2 division. To construct pHil-D8, a PCR was performed using p H il-S 1 (Invitrogen) as a template, since this vector contains the sequence encoding PH01, a positive sensitizer (SEQ ID NO: 11) corresponding to nucleotides 509- 530 of pHil-S1, and a reverse sensitizer (SEQ ID NO: 12), having a nucleotide sequence encoding the last portion of secretion signal PH01 (nucleotides 4566 of SEQ ID NO: 10) and a propeptide sequence (nucleotides 67-108 of SEQ ID NO: 10). The sensitizing sequences (obtained from Operon Technologies, Inc., Alameda, CA, USA) were as follows: SEQ ID # SEQUENCE OF CPR SENSITIZER 11. 5'-GAAACTTCCAAAAGTCGCCATA-3 '12. 5'-ATTAATGAATTCCTCGAGCGGTCCGGGATCCCTCGGCAGCGGAACCAACGG TAGTGCAGATAACTGGCTGAGCGAAGACAGATTGCAAAGTA-3' Se carried out the amplification under normal conditions for PCR, the PCR product was purified in gel (approximately 500 base pairs), cut with Blpl and EcoRI and ligated to pHyl-D2 cut with the same enzymes. E. coli HB101 cells and positive clones were identified by restriction enzyme digestion and sequence analysis A clone having the appropriate sequence was designated pHil-D8.

The following two oligonucleotide sensitizers were used to amplify μUK to clone into pHil-D8: SEQ ID # RCP SENSITIZER SEQUENCE 13. 5'-ATTAATGGATCCTTGGACAAGAGGATTATTGGGGGAGAATTCACCA-3 '14. 5'-ATTAATCTCGAGCGGTCCGTCACTTGGTGTGACTGCGAATCCAGGGT-3' The PCR product was obtained with SEQ ID NO-13 and 14 using pJVP10z-μUK as a template. The amplified product was cut with BamHI and Xhol and ligated to pHyl-D8 cut with the same two enzymes. The resulting plasmid, pHiiD8-μUK was confirmed by DNA sequencing, and was used to transform a strain of Pichia GS115 (Invitrogen), according to the supplier's instructions. The activity of μUK was measured with the chromogenic substrate S2444. The level of expression of μUK in Pichia was higher than that observed in baculovirus-High Five cells, varying from 30 to 60 mg / L.

PURIFICATION OF uUK The culture supernatant of High Five or Pichia cells was collected in a 20 liter vessel. Protease inhibitors, iodoacetamide, benzamidine and EDTA were added at a final concentration of about 10 mM, 5 mM and 1 mM, respectively. The supernatant was diluted five times by adding 5 mM Hepes buffer, pH 7.5 and placed through 1.2 and 0.2 μ filter membranes. The uUK was captured in a Sartorius S100 membrane adsorber (Sartorius, Edgewood, NY, E.U.A.), passing through the membrane at a flow rate of 50-100 ml / min. After extensive washing with 10 mM Hepes buffer, pH 7.5, 10 mM iodoacetamide, 5 mM benzamidine, 1 mM EDTA, the μUK of the S100 membrane was eluted with a gradient of NaCl (20 mM to 500 mM, 200 ml) in 10 mM Hepes buffer, pH 7.5, 10 mM iodoacetamide, 5 mM benzamidine, 1 mM EDTA. The eluate (about 100 ml), ten times in 10 mM Hepes buffer containing inhibitors was diluted, and loaded onto an S20 column (BioRad, Hercules, CA, E. U. A.). The uUK was eluted with a NaCl gradient of 20 times the column volume (20 mM to 500 mM). No inhibitors were used in the elution regulators. Then the eluate was diluted five times with 10 mM of Hepes regulator, pH 7.5, and loaded onto a heparin-agarose column (Sigma). The uUK was eluted with a gradient of NaCl from 10 mM to 250 mM. The eluate of the heparin column, from μUK (about 50 ml) was applied to a benzamidine-agarose column (Sigma, St. Louis, MO, USA) (40 ml), equilibrated with 10 mM Hepes buffer, pH 7.5, 200 mM NaCl. The column was then washed with the equilibrium regulator and eluted with 50 mM NaOAc, pH 4.5, 500 mM NaCl. The elution of μUK (about 30 ml) was concentrated to 4 ml by ultrafiltration and applied to a Sephadex G-75 column (2.5 x 48 cm); Pharmacia® Biotech, Uppsala, Sweden), equilibrated with 20 mM NaOAc, pH 4.5, 100 mM NaCl. The only major peak containing μUK was collected and lyophilized as final product. The purified material appeared by SDS-PAGE as a single major band. The high quality μUK crystals facilitated the determination of its apo-three-dimensional structure by X-ray crystallography, at a resolution of 1.0A. The crystals were obtained by the pendant drop vapor diffusion method. Typical solutions consisted of 0.15M Li2SO4, 20% polyethylene glycol of molecular weight 4,000 and succinate buffer, pH 4.8-6.0. In the coverslip, 2 μl of this solution was mixed with 2 μl of protein solution and the coverslip was sealed over the concavity. The crystallization occurred at approximately 18-24 ° C within 24 hours. The protein solution contained 6 mg / ml (0.214 mM) of μUK in 10 mM citrate, pH 4.0, 3 mM p-carbethoxyphenyl ester of e-aminocaproic acid (inhibitor) with 1% > of DMSO as cosolvent. The inhibitor used in the co-crystallization is believed to acylate the active site serine 195 and is subsequently deacylated in an enzymatic manner because the three-dimensional X-ray structure of the crystals developed in the presence of this compound shows no inhibitor remaining in the active site of enzyme . Menegatti and coauthors, J. Enzyme Inhibition, v. 2, pp. 249-59 (1989). The only density present is that due to the molecules. of solvent attached. Because μUK does not crystallize in the absence of the inhibitor, it is believed that the metastable inhimer: UK complex is the crystallization entity. It is important that the resulting μUK crystals are composed of enzyme with an empty active site, which is the ideal case for the implementation of CrystaLEAD ™ - The crystals obtained under these conditions belong to the spatial group P2Í2-21, with unit cell dimensions from a = 55.16 A, b = 53.00 A, c = 82.30A and a = ß =? = 90 °. They diffract to beyond 1.5 A in a rotating anode source. Additionally, a series of natural data was collected, at a resolution of 1.0 A, at the Cornell High Energy Synchrotron Source, Ithaca, New York, USA. The crystal structure was determined by the molecular replacement method, using the AMORE program, Navaz, J. Acta Cryst., A50: 157-163 (1994), with the structure of low resolution urokinase as a probe, Spraggon and coauthors, Structure, v. 3, pp. 681-691 (1995); PDB, annotation ILMW. The structure was refined using the XPLOR program package, A. Brunger, X-PLOR, (version 2.1) Manual, Yale University, New Haven, CT, E. U. A., (1990).

SELECTION FOR WEAK BASES ΜUK was selected against a bank directed to the structure, in order to find a novel primary scaffold that had favorable pharmacokinetic properties. Since the active site of urokinase is composed of a primary receptacle containing a free carboxylate moiety in the form of an aspartic acid (Asp189), the most well-known scaffolds are strongly basic and contain amidine or guanidine moieties. It has been found that the basic group is bound by hydrogen with salt bond, with Asp189. This can be a pharmacologically speaking problem, since it is known that strong bases decrease oral bioavailability. Consequently, a weakly basic bank containing compounds that had not previously been known as uroquinase linkers was selected. A weak base bank containing 61 compounds with pKa between about 1 and 9 was located in the directory of available chemical substances (ACD). The bank was decomposed to 9 mixtures of about 6 to 7 compounds of different form, as determined by visual inspection of the two-dimensional chemical structure. Compound mixtures were selected by the method previously written. Specifically, each compound was dissolved in 100% DMSO at a final concentration of about 2M (or saturation for the least soluble). Equal volumes of each of the six or seven compounds comprising the mixture were mixed at a final individual compound concentration of 0.33M. Individual μUK crystals were placed in 50 ml of 27% PEG4000, 15.6 mM succinate, pH 5.4, 0.17 M Li2S04 and 0.5-0.8 ml of the compound mixture added to give 1 to 1.6% DMSO and 3.3 to 5.2 mM final concentration of the individual compound. Under these conditions, it is expected that the sensitivity of the experiment will detect the linkers with Kd < 1.0mM. The crystals were allowed to equilibrate for about 8 to 24 hours. Data was collected in a Rigaku RTP 300 RC rotary anode source, with an RAXISII or MAR image plate detector. Typical data consisted of 45-50 oscillations of 2 ° with exposures of 2-5 minutes. The typical data were 70-90% complete at a resolution of 2.0-3.0 A, with coincident R factors of 13-26%. Therefore, the data quality varied from regular to poor, due to the protocol of rapid data collection. However, it was shown that this data quality was adequate for the detection of primary linkers, due to the high quality of the starting model that had been refined at a resolution of 1.5A (R = 20.7% Rh re = 25.3%). The data were processed by the programming package DENZO, Otwinowski and co-authors, Methods in Enzymology, 276 (1996) and the electron density maps were calculated by the XPLOR package. The electron density maps were inspected at a Silicon Graphics INDIGO2 workstation, using the QUANTA 97 program package (Molecular Simulations Inc., Qua? Ta Generating and Displaying Molecules, San Diego: Molecular Simulations, Inc., 1997). The shape of the density in the active site was visually identified, as a result of one or more of the compounds present in the mixture, indicating a positive success, or from the ordered water molecules, which indicate the absence of binding. For the experiments that resulted in a positive hit, the appropriate compound was visually moved to the electron density. The electron density maps were also checked for any changes in the protein structure and, if observed, the appropriate modifications were made. Therefore, after the map inspection step / composite adjustment, the three-dimensional structure of the compound complex: protein is known. The urokinase example used the visual movement of the compound towards density, because the selection was still on a small scale. When expanded to a larger scale of compound selection, commercial programs such as QUANTA's XFIT module will facilitate automatic adjustment of the compound to density. 2456-81-7 123333-56-2 62298-43-5 1 2 3 Figure 6 shows an example of a positive hit.

The selected compounds have the numbers 1 to 6, and the electron density map Fo-Fc in the active site is shown in Figure 6A. The shape of the density identified the linker as compound 5. Figure 6B shows the detailed binding mode of the compound in the primary specificity receptacle, as obtained directly by interpretation of the electron density map in CrystaLEAD ™. The amino nitrogen is linked with hydrogen bond with the carboxyl of Asp189 and the pyrimidyl nitrogen is linked with hydrogen bonding with a skeletal carbonyl (Gly218). The structure also shows that the ideal site for the modification would be in the pyridyl methyl. 51-78-5 2198-58-5 22013-33-8 7 8 9 104-94-9 89-57-6 1072-97-5 10 11 12 364-13-6 13 Another mixture of compounds (compounds 7 to 13) did not produce any success. The resulting electron density map, after soaking this group, did not correspond to that of any of the compounds tested in this mixture. Rather they correspond to the solvent molecules attached. See figure 6C. 3167-49-5 14268-66-7 70125-16-5 17 18 19 Figure 7 shows another example of a positive hit.

Of the seven selected compounds (14-20), the Fo-Fc map shown in Figure 7A indicates that the compound 19 is bound. The binding mode illustrated in Figure 7B shows that the 2-amino is linked with hydrogen bonding with the side chain of Asp189, and that the dhydroxyl is an ideal site for substitution, in order to access the adjacent hydrophobic sub-receptacle (denoted as S1ß in Figure 7B). Figure 8 depicts another hit when the compound 22,5-aminoindole (Figure 8A) was found to bind urokinase with the amino group by hydrogen bonding with Asp189 (Figure 8B). The compounds selected were compounds 21-27. 21 22 23 59323-10-4 • 2359-60-6 74784-70-6 24 25 26 71026-66-9 27 Figure 9 shows an example when it was found that compounds of the same mixture (compounds 28-34) bind in multiple occupancy problems. In the initial experiment, when the crystal was soaked in the presence of the whole compound mixture, compound 28 was found to bind (Figure 9A). In addition, when the weakest binding compound 31 was individually soaked, (based on the previous structural activity ratios, established through CrystaLEAD ™) it was also found to bind (Figure 9B). In a more typical application of the method, a bank was re-soaked in the absence of a stronger linker, in order to detect the weakest linkers in the mixture, if desired. 1603-41-4 137-09-7 24313-88-0 31 32 33 2491-72-6 34 Table 1 summarizes the inhibition constants for each of the CrystaLEAD ™ hits, as determined by chromogenic activity of piroGlu-Gly-Arg-pNa / HCl (S-2444, Chromogenix), The analyzes were completed at both pH 6.5 (0.1 M NaP04) and 7.4 (50 mM Tris.) The other conditions of the analysis were: 150 mM NaCl, 0.5% Pluronic F-68 detergent, 200 mM S-2444, with a final DMSO concentration of 2.5 % It was determined that the Km of the substrate was 55 μM.

TABLE 1 CONSTANTS OF INHIBITION AND pKa FOR DETECTED ACIDS THROUGH CrystaLEAD ™ indicates estimated pKa Based on activity and structural information, compound 19 was selected as the lead compound. The crystallographic information indicated that substitution at the position must allow access to the adjacent hydrophobic receptacle (S1ß) and, thereby, result in an increase in potency. Based on the crystallographic and binding information, from a series based on amidine, compound 35 (the 8-aminopyrimidinyl analog of compound 19) was synthesized. This modification resulted in a 200-fold increase in the binding potency at pH 6.5 (Ki pH 7.4 = 2.5 μM, Ki pH 6.5 = 0.32 μM). The experiment indicates that CrystaLEAD ™ can provide both a scaffolding header and the detailed structural information needed to make that scaffolding, by means of the drug design, based on the structure, to a more potent compound.

In figure 10 is shown an overlay view of the crystalline compound 35: urokinase and the original compound 19. The superimposed view shows that the aminopyridine ring is bound in the hydrophobic capstan (S1ß) as predicted and that this substitution results in the movement of the quinoline ring to this site. Compound 35, 8-aminopyrimidinyl-2-aminoquinoline was also tested for oral bioavailability. It was detned that compound 35 was orally available in 30 to 40% in rats, when administered at a dose of 10 mg / kg. Accordingly, the successful implementation of CrystaLEAD ™ resulted in novel scaffolding scaffolding that, through a structure-based drug design cycle, produced a compound that is 200 times more potent, and was found to be bioavailable by oral route EXAMPLE 2 VanX Vancomycin is the drug selected for infections caused by streptococcal or staphylococcal bacterial strains, which are resistant to beta-lactam antibiotics. However, strains of bacteria resistant to vancomycin have been found for this last-resort drug. Some researchers have associated VanX, a metalloproteinase with resistance to vancomycin. VanX is part of a cascade that results in the replacement of the D-Ala-D-Ala tnal portion of the bacterial peptidoglycan chain (the binding site for vancomycin) with a D-ala-D-lactate. This results in a 1000-fold decrease in vancomycin binding. The only known VanX inhibitors are peptides or peptide derivatives, such as phosphonate or phosphinate analogs of the D-Ala-D-Ala substrate. Thus, they are not suitable drugs because they are metabolized and / or degraded in vivo. Initial attempts to find suitable drugs by normal screening methods did not discover an adequate ligand. Subsequently, the applicants turned to CrystaLEAD ™ to find a non-peptidic lead compound for the development of the drug with a view to a treatment for those resistant strains.

PREPARATION OF VanX The E. coli W3110 containing the plasmid pGW1 was developed at 37 ° C, in which the vanX gene is under control of the IPTG-inducible tac promoter, in LB medium containing 100 μg / ml of ampicillin, at an approximate absorbance of 1.3. -1.5 to 595 nm. Then IPTG was added to a final concentration of 0.8 mM and the cells were developed for an additional 1.5 hours. The cells were harvested by centrifugation at 6,000 r.p.m. for 10 minutes. The pellet was then resuspended in 20 mM Tris-HCl (pH 8.0 cooled with ice, containing 0.01% NaN3, 1 mM MgCl2, 1 mM PMSF, 1 mM DTT (regulator A) and 25 units / ml of benzonase (Nicomed Pharma, Copenhagen, Denmark) The cells were lysed by adding ceramic granules to the zirconium of 0.1 micras, to the lisato mixture (1: 1 in volume: volume) with a 1-3 minute operation In a Bead Beater (Biospec), a ball mill with ultrasound, the Bead Beater was operated with a tank packed in ice to keep the lysate cooled, then the lysate was decanted from the pelleted glass granules. 1-2 volumes of the lysed buffer, and the washings were combined with the original lysate.The lysate was centrifuged at 25,000 g for 30 minutes to settle the cell detritus.The supernatant was dialyzed overnight at 4 ° C in 50 mM Tris-HCl, pH 7.6, 1 mm EDTA and 1 mM DTT (regulator B) Posteriorme The dialysate lysate was loaded into a Q-Sepharose fast flow column, previously equilibrated in regulator A, at a rate of four millimeters per minute. The column was thoroughly washed with regulator A, followed by a linear gradient from regulator B to regulator B + 0.5 M NaCl. The active VanX fractions of this step were pooled, concentrated and then applied to a Superose-75 column in regulator B. The vanX fractions from the Superóse column operation were then applied., to a Source-Q column in regulator A, at a flow rate of 2 ml / minute. The column was washed with starting buffer for several column volumes. The VanX protein was then eluted to separate it, with a small gradient from regulator A to regulator A + 25 mM NaCl. The active VanX fractions of this final step were concentrated to a final concentration of approximately 15 mg / ml in regulator A with Amicon filters. Unless otherwise specified, the above procedure was operated at 4 ° C. When purified, the VanX protein had an approximate purity of 95% and crystallized easily.

THE CRYSTAL STRUCTURE OF VanX The crystal structure of VanX at 2.2A resolution was determined by multiple isomorphic replacement. Bussiere and coauthors, Molecular Cell, vol. 2 pp. 75-84 (1998). The recombinant protein obtained above was crystallized in the space group P2 | by the vaporization method of drop that settles. The typical crystals had the unit cell dimensions of: a = 83.4 A, b = 45.5 A, c = 171.4 A, a =? = 90 °, ß = 104 °, with six molecules in the asymmetric unit. Typical concavity solutions consist of 0.1 M of Month, pH 6.4; 0.24 M of ammonium sulfate and 20% of PMME 5000. In the settling droplet / Hampton, E. U. A.), 2 ml was mixed with 2 ml of concavity solution and the chamber was sealed with a sliding cover. Crystallization occurs at 18 ° C and the crystals grow to full size in about 2 to 3 days. The protein solution is composed of 12-15 mg / ml (0.5-0.6 mM) of VanX in 10 mM Tris, 15 mM DTT, pH 7.2. The three-dimensional structure for the crystals developed under these conditions shows an empty active site, which makes it an extremely suitable system for the application of CrystaLEAD ™. The active site of VanX has an extended receptacle, capable of accommodating the substrate D-Ala-D-Ala. The receptacle also contains a catalytic zinc. Thus, for this case, VanX was initially selected against banks targeting zinc, in order to find multiple link scaffolds, which could converge to a single lead compound. Three banks using amino acid, thiol, hydroxamic acid or carboxylate portions, targeted to zinc, were selected.

THE SELECTION The amino acid library consisted of 102 naturally occurring, non-occurring, optically pure, commercially obtainable amino acid compounds. The bank was divided into 12 mixtures of 8 to 10 compounds in various ways, and were selected by the method described above. Specifically, each compound was dissolved in 100% DMSO at a final concentration of 2M (or saturation for the least soluble). Equal volumes of each compound of each mixture were mixed at a final individual compound concentration of 0.33M. Individual VanX crystals were placed in 50 ml of 0.1M of Month, pH 6.4, 0.24M of ammonium sulfate, 20% of PMME 5000 and 0.5-0.8 ml of the compound mixture, added to give 1 to 1.6% of DMSO and 3.3 to 5.2 mM final concentration of the individual compound. The crystals were allowed to equilibrate for 3-4 hours. The thiol, hydroamic and carboxylate banks were prepared and similarly selected. The data collected in a rotary anodic source Rigaku RTP 300 RC with an image formation plate RAXISII, MAR or a MAR CCD detector. For systems with imaging plate, typical data consisted of 90 oscillations of 1.25 ° with exposures of 15 minutes, while for CCD 100 it was exposed to oscillations of 1.0 ° for two minutes. Typical usable data were > 90% complete at 2.6-2.8A resolution, with convergent R factors of 10-20%. This was necessary to adequately visualize and identify inhibitors in the Fo-Fc or 2Fo-Fc maps. For those maps, the starting model had been refined at 2.1A resolution (R = 25%, R | lbre = 28%). The data was processed through the DENZO program package and the electron density maps were calculated using the XPLOR package. In the presence of some compounds of the carboxylate bank, it was shown that the space group deviated from P2-? a C2 (a = 170.6 A, b = 47.5 A, c = 83.6, 143, a =? = 90 °, ß = 104 °). For this form, the asymmetric unit contained a trimer, thus reducing the number of degrees of freedom, so that the lower resolution data (3.0 A) were suitable for visualizing the joint. Electron density maps were inspected at the INDIGO2 Silicon Graphics workstation, using QUANTA 97. The shape of the density in the active site was visually identified, by the shape of one or more of the compounds in the mixture, to indicate a positive success, or by the ordered water molecules, which indicates the absence of union. For the experiments that resulted in a positive hit, the appropriate compound was visually moved to the electron density. The electron density maps were also checked for any changes in the protein structure and, if observed, the corresponding modifications were made in the structure. Therefore, after the map inspection step / composite adjustment, the detailed three-dimensional structure of the compound complex: protein was known. 580-22-3 -89-5 132-32-1 36 37 38 1603-41-4 137-09-7 24313-88-0 39 40 41 Currently six hits have been detected in the VanX selections (compounds 36-41). Figure 11 shows the union mode of the representative hits. In all cases the electron density identified in the binding compound form. Figure 11A shows compound 39 bound to the carboxylate that coordinates the zin for the active site. Figure 11B shows compound 36 bound with the carboxylate that targets active site zinc. In Figure 11C it was also found that compound 37 bound through the carboxylate. The binding of compound 39 and compound 41 (not shown) suggests that active site zinc prefers the coordination of a carboxylate to that of the free thiol. This leads to the selection of a carboxylate bank, in which additional hits were found. In all cases the compounds were selected in mixtures of 7 to 10, and the correctness was directly identified by the shape of the electron density map. These hits are fed directly into the drug design cycle based on the structure, in a manner similar to that described for the urokinase example.

EXAMPLE 3 SELECTION WITH MIXES OF ONE HUNDRED COMPOUNDS In order to increase the number of compounds that can be selected per unit time by the CrystaLEAD ™ method, a preferred embodiment of the method would be to select mixtures of 100 compounds instead of mixtures of 10. The advantage of this method is a higher production of compound , with a concurrent decrease in the sensitivity of the detection of hits. Additionally, since only the most potent compound in a mixture will join, the weakest hits may be missing. When a general bank is selected, for example, one that is totally diverse in size, shape and functionality, using CrystaLEAD ™, it is expected that the hit rate will be low. Therefore, a thicker selection is guaranteed. Additionally, since the successes from this selection would be the most potent linkers, they could serve as starting scaffolds for structure-based drug design. Since the compound mixture will be composed of 100 compounds, the mixture must be carefully designed to ensure that all members have a sufficiently diverse form to eliminate the need to undo the convolution. Therefore, when detecting a hit, it may be necessary to undo the convolution to some extent to identify the success. To test this particular method, a compound known to bind to μUK was added to a group of 100 compounds. This known linker, compound 19, was originally discovered by means of the CrystaLEAD ™ method, and was shown to bind to μUK with a Ki of 56 μM at pH 6.5 and 137 μM at pH 7.4. The mixture of 100 compounds was constructed by mixing 10 mixtures of 10 compounds. Specifically, each dry mix of 10 was dissolved in 100% DMSO at a final concentration of about 80-240 mM (or at saturation for the least soluble). Equal volumes of each of the mixtures of 10 compounds were mixed at a final concentration of individual compound of 8.0-24.0 mM and the mixture was minced with a material in 100% DMSO of compound 19, so that the final concentration outside 18.0 mM. Individual crystals of μUK were placed in 50 μl of 27% PEG4000, 15.6 mM succinate, pH 5.4, 0.17 M Li2S04 and 0.5 μl of the mixture of compounds, to give 1% DMSO. The final concentration of each compound in the soaking experiment varied from 80 to 240 μM, and the concentration of compound 19 was 180 μM. Under these conditions it is expected that the sensitivity of the experiment detects linkers (linkers) with Kd < 20-60 μM. The crystals were allowed to equilibrate for four hours and 15 minutes. The data was collected on the ID ray line IMCA, from the Argonne National Labs advanced photon source synchrotron, equipped with a MarCCD camera. The data consisted of 100 oscillations of 1o with exposures of 7 seconds. The data was completed in 87.4% at a resolution of 1.6 A, with an R factor of general convergence of 5.4%. The data were processed by the program package DENZO, Atwinowski and co-authors, Methods in Enzymology, 276 (1996) and the electron density map was calculated by the XPLOR package. The electron density map was inspected at an INDIG02 Silicon Graphics workstation, using the QUANTA 97 program package (Molecular Simulations Inc., Quanta Generating and Displaying Molecules, San Diego: Molecular Simulations Inc., 1997). The shape of the density at the active site was visually identified, as a result of which one of the compounds in the mixture indicates a positive hit, which was identified as compound 19 and is illustrated in figure 12. This method is preferable for discover leading compounds. The header compounds would typically have the characteristics of being stronger linkers (for example, within the sensitivity scale of the method). This method also allows to select from a bank of 10,000 compounds, not directed, within a time frame of 1 to 2 weeks. This method would be used in conjunction with other methods to select 10-20 compounds at a time, where the weakest linkers would be identified. These linkers are less likely to serve as lead compounds, but could be attached to a header scaffold in order to increase power.

EXAMPLE 4 SELECTING ErmC 'WITH CrystaLEAD TM ErmC 'is a rRNA methyltransferase, which transfers a methyl group of S-adenosyl-L-methionine to the N6 of adenine, within the peptidyltransferase loop of 23S rRNA. This methylation confers resistance to the antibiotic against numerous macrolide antibiotics, such as popularly prescribed erythromycin. It would be expected that the inhibition of ErmC 'would reverse the resistance. In order to design a specific and potent ErmC 'inhibitor, the cofactor or S-adenosyl-L-methionine binding site has been selected as the target. The S-adenosyl-L-methionine is illustrated below as compound 42 .: 42 The crystal structure of ErmC 'shows that the S-adenosyl-L-methionine site is composed of two primary receptacles that accommodate the adenine ring and methionine. Additionally there is a third receptacle that can accommodate the rRNA adenine that undergoes methylation. In order to establish a SAR at that site, a bank of adenosine analogs substituted at the hydroxyl of N6 and / or 5 'was generated. The variation sites in the bank are represented below, as compound 43. 43 EXPRESSION AND PURIFICATION OF ErmC The expression vector pTERM31 was constructed by amplification by polymerase chain reaction (PCR) of the ermC 'gene and the kdsB cistron upstream of pERM-1. Subcloning of the PCR product into pET24 + (Novagen, Madison, Wl, E.U.A.), was performed using the BamHI and Hindlll sites, included in the "tail" PCR sensitisers. This new construction allowed the expression of ErmC 'by translational coupling to kdsB, under the control of the T7lac promoter. The plasmid pTERM31 was transformed into E. coli strain BL219 (DE3) / pLysS (Novagen) and the resulting strain was used for the production of ErmC '. Cells transformed at 27.5 ° C were developed in a New Brunswick Scientific (Edison, NJ, E.U.A.) Micros fermenter containing 10 liters of Superbroth (BIO 101, La Jolla, CA, E.U.A.), supplemented with kanamycin, chloramphenicol and glucose. When the optical density of the culture reached 1.10, expression of Er C 'was induced by the addition of 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG). The cells were harvested 400 minutes after induction. 200-250 g of frozen cell paste was thawed to room temperature and resuspended in 5-10 volumes of cold lysis buffer (50 mM Tris, 5 _mM 1,4-dithiothreitol (DTT), 1 mM phenylmethyl sulfonate fluoride, 2 mM ethylenediaminetetraacetic acid (EDTA), 0.2% Triton X-100, pH 7.8). The cells were lysed with a French press and the cellular debris was removed by centrifugation. The supernatant was dialyzed overnight against 20 liters of Tris-DTT-glycerol-magnesium buffer (TDGM), pH 7.8 (50 mM Tris, 5 mM DTT, 10% glycerol, 10 mM MgCl 2). The dialysate was then applied to a Sepharose Fast column (Pharmacia) which had been previously equilibrated in the TDGM regulator. Fractions were analyzed for their methyltransferase activity and those containing ErmC 'were pooled, applied to a TSK SP5PW column (TosoHaas, Montgomeryville, PA, E.U.A.) and eluted with a NaCl gradient. The purified protein was then concentrated on a YM-10 membrane (Amicon).

THE CRYSTAL STRUCTURE OF ErmC ErmC 'crystals were developed by the vapor diffusion method of hanging droplet. Balances were obtained containing 5-8 mg / ml ErmC 'in 25 mM Tris / Cl, 100 mM NaCl, 2 mM DTT, 10% (volume / volume) glycerol, pH 7.5, against a deposit that it contained 100 mM Tris, 500 mM NH 4 (SO) 4, 15% PEG 8000, pH 7.8. The crystals appeared within a day and grew to full size within a week. The crystals belonged to the space group P43212. The structure of ErmC 'in this space group was determined by molecular replacement at a resolution of 2.2 A using the crystal structure of ErmC' in the space group P6 (Bussiere and coauthors, Biochemistry, v.37, pp. 7103-7112). The three-dimensional structure for the crystals developed under these conditions shows an empty active site which makes it an extremely suitable system for the application of CrystaLEAD ™.

SELECTION The adenosine bank consisted of 59 compounds. The bank was divided into 7 mixtures of 8-9 compounds of variety of shapes, and was selected by the CrystaLEAD ™ method. Specifically, each compound was dissolved in 100% DMSO at a final concentration of 1M (or saturation, for the least soluble). Equal volumes of each compound were mixed to assemble the mixture of 10. Individual crystals of ErmC were placed in 50 μl of 20% PEG 8000, 0.3M of ammonium sulfate, 10% glycerol, pH 7.7 and 0.5-0.8 μl of the mixture of compounds, added to give 1 to 1.5% DMSO and 3.3 to 5.2 μM final individual compound concentration. The crystals were allowed to equilibrate for 3-4 hours. The data was collected in a rotating anode source Rigaku RTP 300 RC, with an image formation plate RAXIXII, MAR, or a MAR CCD detector. For the image forming plate systems, the typical data consisted of 15-20 oscillations of 2 ° with 20-30 minutes of exposure, while for CCD 15-20 oscillations of 2.0 ° were exposed for 8-15 minutes. Typical usable data was complete at 80-90% > at a resolution of 3.4-3.6 A, with R factors of confluence of 7-16%. This was necessary to adequately visualize and identify the inhibitors in the Fo-Fc or 2Fo-Fc maps. For these maps, the departure model had been refined at a resolution of 2.2 A (R = 22% R?, Bre = 25%). The data was processed through the DENZO program package and the electron density maps were calculated using the XPLOR package. The electron density maps were inspected at an INDIG02 workstation of Silicon Graphics, using QUANTA 97. The shape of the density at the active site was visually identified by the shape of one or more of the compounds in the mixture, to indicate a positive success or by means of ordered water molecules, which indicate the absence of union. For the experiments that resulted in a positive hit, the appropriate compound was visually moved to the electron density. The electron density maps were also checked for any change in the protein structure and, sT was observed, the corresponding modifications were made in the structure. Accordingly, after the map inspection step / composite adjustment, the detailed three-dimensional structure of the compound complex: protein was known. Two hits were detected in the selection of adenosine analogs of ErmC '(compounds 44 and 45). Figures 13 and 14 show the crystal structure of the complexes of compounds 44 and 45 with ErmC '. 44 45 In all cases, the electron density form identified the compound that binds. It was found that hydrophobic substitution binds along a partially exposed hydrophobic surface, suggesting a preferred interaction that may have contributed to the binding of these compounds, allowing them to be labeled as hits. No hit was detected that contained a substitution at the 5'OH position. A tracking compound for compounds 44 and 45 contained an optimized indane substituent at that hydrophobic site.

LIST OF SEQUENCES < 110 > Nienaber, Vicki Greer Jonathan Abad-Zapatero, Celerino Norbeck, Daniel < 120 > SELECTION AND DESIGN OF LINKING THROUGH X-RAY CRYSTALGRAPHY < 130 > 6308. us. p1 < 150 > 09 / 036,184 < 151 > 1998-03-06 < 160 > 14 < 170 > FastSEQ for Windows, version 3.0 < 210 > 1 < 211 > 51 < 212 > DNA < 213 > Synthetic < 400 > 1 attaatgtcg actaaggagg tgatctaatg ttaaaatttc agtgtggcca to 51 < 210 > 2 < 211 > 57 < 212 > DNA < 213 > Synthetic < 400 > 2 attaataagc tttcagaggg ccaggccatt ctcttccttg gtgtgactcc tgatcca 57 < 210 > 3 < 211 > 47 < 212 > DNA < 213 > Synthetic < 400 > 3 attaatgcg cagccatccc ggactataca gaccatcgcc ctgccct 47 < 210 > 4 < 211 > 46 < 212 > DNA < 213 > Synthetic < 400 > 4 attaatcagc tgctccggat agagatagtc ggtagactgc tctttt 46 < 210 > 5 < 211 > 28 < 212 > DNA < 213 > Synthetic < 400 > 5 attaa? Cagc tgaaaatgac tgttgtga 28 < 210 > 6 < 211 > 51 < 212 > DNA < 213 > Synthetic < 400 > 6 attaatgtcg actaaggagg tgatctaatg ttaaaatttc agtgtggcca to 51 < 210 > 7 < 211 > 37 < 212 > DNA < 213 > Synthetic < 400 > 7 attaatgcta gcctcgagcc accatgagag ccctgct 37 < 210 > 8 < 211 > 42 < 212 > DNA < 213 > Synthetic < 400 > 8 attaatgcta gcctcgagtc acttgttgtg actgcggatc ca 42 < 210 > 9 < 211 > 44 < 212 > DNA < 213 > Synthetic < 400 > 9 ggtggtgaat tctcccccaa taatgccttt ggagtcgctc acga 44 < 210 > 10 < 211 > 111 < 212 > DNA < 213 > Yeast Pichia pastoría < 400 > 10 atgttctctc caattttgtc cttggaaatt attttagctt tggctacttt gcaatctgtc 60 ttcgctcagc cagttatctg cactaccgtt ggttccgctg ccgagggatc c 111 < 210 > 11 < 211 > 22 < 212 > DNA < 213 > Synthetic < 400 > 11 gaaacttcca aaagtcgcca ta 22 < 210 > 12 < 211 > 92 < 212 > DNA < 213 > Synthetic < 400 > 12 attaatgaat tcctcgagcg gtccgggatc cctcggcagc ggaaccaacg gtagtgcaga 60 taactgctg agcgaagaca gattgcaaag ta 92 < 210 > 13 < 211 > 46 < 212 > DNA < 213 > Synthetic < 400 > 13 attaatggat ccttggacaa gaggattatt gggggagaat tcacca 46 < 210 > 14 < 211 > 47 < 212 > DNA < 213 > Synthetic < 400 > 14 attaatctcg agcggtccgt cacttggtgt gactgcgaat ccagggt 47

Claims (35)

1. - A process for identifying a ligand to a biomolecule of interest, characterized in that it comprises: a) obtaining a biomolecule crystal of interest; b) exposing the crystal of the biomolecule of interest to one or more test samples; and c) obtaining a X-ray diffraction pattern in the crystal, to determine if a ligand / receptor complex is formed.

2. The process according to claim 1, further characterized by additionally comprising the steps of: obtaining a crystal X-ray diffraction pattern of the biomolecule crystal of interest, before exposing it to the test samples, and comparing X-ray diffraction pattern of the molecule of interest before and after exposure.

3. The process according to claim 1, further characterized in that it comprises the step of transforming the diffraction pattern to an electron density map. - -

4. The process according to claim 3, further characterized in that it comprises the step of converting the electron density map into a structure.

5. The process according to claim 1, further characterized in that the biomolecule of interest is exposed to a test sample by soaking the crystal of the biomolecule of interest in a solution containing the test sample.

6. - The process according to claim 1, further characterized in that the biomolecule of interest is exposed to the test samples by soaking the biomolecule crystal of interest in a solution containing a mixture of test samples.

7. The process according to claim 1, further characterized in that the biomolecule of interest is exposed to the test sample by co-crystallizing the biomolecule crystal of interest with a test sample.

8. The process according to claim 1, further characterized in that the biomolecule of interest is exposed to the test samples by co-crystallizing the biomolecule crystal of interest with a mixture of test samples.

9. The process according to claim 6, further characterized in that the samples of the test sample mixture are of a variety of shapes.

10. The process according to claim 8, further characterized in that the samples of the test sample mixture are of a variety of shapes.

11. The process according to claim 1, further characterized in that the ligand is a biologically active portion.

12. The process according to claim 1, further characterized in that the biomolecule of interest is a polypeptide.

13. - The process according to claim 1, further characterized in that the biomolecule of interest is an engineered polypeptide.

14. A biologically active portion, identified by the process according to claim 11.

15. The process according to claim 1, further characterized in that the ligand is a leader compound.

16. A process for designing a ligand for a biomolecule of interest, characterized in that it comprises: a) obtaining a crystal of the biomolecule of interest; b) identifying at least two ligands for the biomolecule of interest by X-ray crystallographic selection; c) determining the spatial orientation of the ligands when they are bound to the biomolecule of interest; and d) linking the ligands together, according to the spatial orientation, to form the ligand.

17. The process according to claim 16, further characterized in that the spatial orientation of bound ligands is determined by forming a multiligand / molecule of interest and generating a crystal structure for X-rays of the multiligand / molecule of interest complex.

18. The process according to claim 16, further characterized in that a ligand is bound to the molecule of interest before attaching another ligand to the molecule of interest.

19. - The process according to claim 16, further characterized in that the ligand is a biologically active portion.

20. The process according to claim 16, further characterized in that the biomolecule of interest is a polypeptide.

21. The process according to claim 16, further characterized in that the biomolecule of interest is an engineered polypeptide.

22. A biologically active portion, designed by means of the process according to claim 19.

23. The process according to claim 16, further characterized in that the ligand is a leader compound.

24. A process for designing a ligand for a biomolecule of interest, characterized in that it comprises: a) obtaining a crystal of the biomolecule of interest; b) identify a ligand for the biomolecule of interest, by X-ray crystallographic selection; c) forming derivatives of the ligand.

25. The process according to claim 24, further characterized in that the ligand is a leader compound.

26. The process according to claim 24, further characterized in that the ligand is a biologically active compound.

27. The process according to claim 24, further characterized in that the molecule of interest is a polypeptide.

28. The process according to claim 24, further characterized in that the molecule of interest is an engineered polypeptide.

29. A heading compound, identified by means of the process of claim 25. 30.- A biologically active compound designed by means of the process according to claim 25. 31.- A biologically active compound, designed by means of the process of according to claim 26. 32.- A process for forming a crystal having an easily accessible active site, from a biomolecule, characterized in that it comprises: a) co-crystallizing the biomolecule with a degradable ligand; and b) degrading the ligand once the crystal is formed. 33. The process according to claim 32, further characterized in that the active site of the biomolecule degrades the ligand. 34. The process according to claim 32, further characterized in that it comprises adding degradation agents to degrade the ligand. 35.- The process according to claim 32, further characterized in that the ligand is spontaneously degraded.