DE10221158A1 - Method for the discovery of ligands that bind to a drug target by means of heteronuclear nuclear magnetic resonance spectroscopy - Google Patents

Abstract

The present invention provides a method for finding ligands that bind to a given drug target. This method is based on the measurement method of nuclear magnetic resonance spectroscopy (NMR) and specifically on the possibility of being able to clearly select the test substances using NMR isotope markers. To this end, the present invention also describes methods for setting up corresponding substance banks. The present invention accordingly comprises the following aspects: (1) A method for compiling a substance bank exclusively from those compounds which, for the purpose of their selective NMR spectroscopic observability, contain at least one core of an NMR-observable element or isotope which does not or only in the drug target occurs with a very low frequency; (2) the substance bank that can be produced with this method; (3) a method for identifying active substances which bind to a drug target using this substance bank, which is characterized by the recording of NMR spectra edited or filtered according to the present element or isotope.

Description

1 Introduction

The present invention relates to a method of finding ligands attached to a given drug bind target using heteronuclear nuclear magnetic resonance spectroscopy (NMR), as well as for this suitable substance banks and a process for their compilation.

2. Background and state of the art

The search for drug candidates in the life sciences (pharmaceuticals, animal and plant protection) today by means of diverse biological, chemical and physical Screening process operated. Thereby one becomes as possible extensive and chemically diverse collection of small molecules Connections searched to a given disease-relevant target molecule (drug target; usually a protein, RNA or DNA section) bind and affect its biological function. So found Ligands (initial hits) are starting compounds for development of drug candidates. You need to in the subsequent Phase of drug development in terms of their affinity and selectivity for drug target usually need to be improved before their effectiveness, toxicity and pharmacokinetic properties (such as uptake, distribution, Metabolism and excretion in biological systems) by Variation of chemical structure according to principles of medical Chemistry and first in vivo tests can be adapted. With these drug optimization processes it is vital to have the relevant interactions and characteristics of the binding of a ligand to the drug target exactly to know. Based on this knowledge, further drug development can rational, targeted and accelerated.

• – NMR stellt eine universell einsetzbare physikalische Meßmethode dar, die das Screening nach (bindenden) Liganden ermöglicht. Es müssen also nicht erst zeitaufwändig system- bzw. funktionsspezifische biochemische Nachweisverfahren (bio-assays) entwickelt werden, die zudem das Risiko häufiger falschpositiver Nachweise durch Nebenreaktionen bergen. Ebenso kann bereits ohne Kenntnis der Proteinfunktion unmittelbar im Anschluß an die Identifizierung eines drug targets (z.B. mittels Methoden der Bioinformatik) mit der Wirkstoffsuche per NMR begonnen werden.
• – NMR ist die derzeit empfindlichste Screeningmethode zum Nachweis auch sehr schwach bindender Wirkstoff-Vorläufer. Sie kann daher insbesondere auch neuartige Ansätze der Wirkstoffentwicklung über neue Ausgangsmoleküle erschließen.
• – NMR erlaubt zudem die Klassifizierung aufgefundener Liganden nach ihrer Bindungsstärke (Affinität) über einfache Titrationsexperimente.
• – NMR liefert atomar aufgelöste Spektren. Damit können anhand von NMR-Spektren des drug targets unmittelbar verschiedene Bindungsstellen des Liganden am drug target erkannt und unterschieden werden. Dies wiederum erlaubt es, schnell unspezifische oder irrelevante Wechselwirkungen und damit falschpositive Anzeigen auszuschließen. Mittels NMR-spektroskopischer Methoden (in Verbindung mit geeigneten rechnergestützten Simulationsverfahren wie molecular dynamics etc.) ist es weiterhin möglich, die räumliche Struktur des Komplexes aus Ligand und drug target und damit den Bindungsmodus zu bestimmen (H.-J.Böhm, G.Klebe, H.Kubinyi: "Wirkstoffdesign", Spektrum Akademischer Verlag GmbH, 1996, Heidelberg/Berlin/Oxford, ISBN 3-8274-0174-7). Die Entdeckung neuer Bindungsstellen und -modi eröffnet neuartige Ansätze der Wirkstoffentwicklung.
• – Mittels NMR können molekulare Strukturen, Beweglichkeiten und Wechselwirkungen mit atomarer Auflösung bestimmt und umfassend charakterisiert werden. Damit schafft NMR die essentiellen Grundlagen für die erfolgreiche Optimierung der aufgefundenen ersten bindenden Liganden (inital hits) nach Prinzipien der rationalen strukturbasierten Wirkstoffentwicklung (rational drug design). Die besondere Stärke der NMR-Spektroskopie liegt also in der Kombination und gleichzeitigen Eignung zum Aufspüren und zielgerichteten Optimieren neuer Liganden.

Magnetic nuclear magnetic resonance spectroscopy (NMR), with which molecular structures, mobilities and interactions can be characterized with atomic resolution, offers a combination of particular strengths for the commercial search and development of active substances, which are explained in more detail below. It thus solves many of the problems inherent in alternative methods.

• - NMR represents a universally applicable physical measurement method that enables screening for (binding) ligands. System-specific and function-specific biochemical detection methods (bio-assays) do not have to be developed, which also involve the risk of false positive evidence from side reactions. Likewise, the knowledge of the protein function can be started immediately after the identification of a drug target (for example by means of bioinformatics methods) with the active ingredient search by NMR.
• - NMR is currently the most sensitive screening method for the detection of very weakly binding precursors. In particular, it can therefore also open up new approaches to drug development using new starting molecules.
• - NMR also allows ligands found to be classified according to their binding strength (affinity) using simple titration experiments.
• - NMR provides atomically resolved spectra. Different binding sites of the ligand on the drug target can thus be immediately recognized and differentiated on the basis of NMR spectra of the drug target. This in turn allows non-specific or irrelevant interactions and thus false positives to be excluded quickly. Using NMR spectroscopic methods (in connection with suitable computer-aided simulation methods such as molecular dynamics etc.) it is also possible to determine the spatial structure of the complex of ligand and drug target and thus the binding mode (H.-J.Böhm, G.Klebe , H.Kubinyi: "Active ingredient design", Spektrum Akademischer Verlag GmbH, 1996, Heidelberg / Berlin / Oxford, ISBN 3-8274-0174-7). The discovery of new binding sites and modes opens up new approaches in drug development.
• - Using NMR, molecular structures, mobilities and interactions can be determined and comprehensively characterized with atomic resolution. Thus NMR creates the essential basis for the successful optimization of the first binding ligands (initial hits) according to the principles of rational structure-based drug development (rational drug design). The particular strength of NMR spectroscopy lies in the combination and simultaneous suitability for the detection and targeted optimization of new ligands.

The basis for the suitability of NMR spectroscopy for screening is the sensitivity with which NMR signals react to binding-induced molecular changes - such as the chemical environment, molecular structure and dynamics. This often changes several signal characteristics or indicators at the same time, which primarily include the signal frequency, intensity and shape. The reporter signals on which these binding-induced changes are recorded must be clear can be recognized and separated in the NMR spectrum. Accordingly, two main methodological aspects of all NMR screening methods are the sensitivity of the selected indicator and the resolvability or selective observability of the selected reporter signals.

• – auf den Liganden detektierende Verfahren (ligandendetektierende Verfahren)
• – auf dem drug target detektierende Verfahren (targetdetektierende Verfahren)

The methods of NMR screening can in principle be divided into two classes (Diercks, Coles et al .: Current Opinion in Chemical Biology 2001, 5, 285-291):

• - methods for detecting ligands (methods for detecting ligands)
• - on the drug target-detecting method (target-detecting method)

Most ligand-detecting Procedures in which after changes in binding caused by the NMR spectra of ligands are searched, either direct intermolecular interactions between the drug target and a Ligands over dipolar relaxation phenomena, or the decrease in the free mobility of a ligand apparent increase due to the molecular size the attachment to the (much larger and sluggish) drug target, i.e. slowing down its diffusion and / or self-rotation.

Diffusion becomes NMR spectroscopic directly above the signal decrease when using magnetic field gradient echoes detected. In diffusion-ordered NMR spectroscopy, DOSY (Johnson: Progress in Nuclear Magnetic Resonance Spectroscopy 1999, 34, 203–256), becomes dependency the signal decrease of echo strength or duration to determine the diffusion coefficients. These are then plotted in a second pseudo dimension and thus allow a quick separation of substance mixtures in individual components diffusing at different rates (Barjat, Morris et al .: Journal of Magnetic Resonance 1995, B108, 170-172). On constant field gradient echo acts as a simple diffusion filter, which selectively dampens the ligands that bind to the drug target. On such diffusion filter takes place in that described in WO 98/48264 Method of ligand-detecting NMR screening application.

The intrinsic molecular rotation can only be detected indirectly by influencing NMR relaxation processes. This includes entropic T ₂ relaxation, which leads to the broadening and attenuation of the NMR signals and increases with decreasing molecular mobility. T ₂ relaxation rates can be determined with suitable NMR experiments from the signal decrease depending on the duration of a SpinLock or eg CPMG echo sequence. A constant SpinLock or echo duration acts as a T ₂ relaxation filter, which, like the diffusion filter described, only attenuates binding ligands. This principle of ligand-detecting NMR screening is also described in WO 98/48264.

Another NMR relaxation phenomenon is the Nuclear Overhauser Effect (NOE), in which magnetization is transmitted between spatially close nuclear spins. The extent of NOE depends heavily on molecular mobility, apart from the distance between the nuclear spins and the chosen development period. In particular, the sign of the homonuclear NOE (between protons) changes during the transition from small to large molecules. In the bound state, a small ligand behaves like the large drug target and, after its dissociation, takes the temporarily acquired NOE intensity into the free state as so-called transfer NOE (trNOE). A ligand that binds only temporarily can therefore be distinguished by NMR spectroscopy from non-binding small molecules in that, in contrast to these, it also has trNOE with the opposite sign. This principle is the basis of the in DE 19649359 described method and published as Bioaffinity NMR (Meyer, Weimar et al .: European Journal of Biochemistry 1997, 246, 705-709) for ligand-detecting NMR screening.

In addition to the intramolecular described NOE phenomena Magnetization can occur between nuclear spins within the same molecule (here the ligand) also between nuclear spins of the drug target and nuclear spins of its ligand be replaced. Such an intermolecular NOE can in principle be observed both on the ligand and on the drug target and is an immediately accessible, not primarily changes Molecular mobility affected local indicator of Ligand binding. It is the basis of ligand detection, for example NMR screenings using saturation transfer difference = STD (Klein, Meinecke et al .: Journal of the American Chemical Society 1999, 121, 5336-5337; Mayer and Meyer: Angewandte Chemie International Edition in English 1999, 38, 1784-1788), NOE pumping (Chen and Shapiro: Journal of the American Chemical Society 1998, 120, 10258-10259; Chen and Shapiro: Journal of the American Chemical Society 2000, 122, 414-415) and WaterLOGSY (Dalvit, Pevarello et al .: Journal of Biomolecular NMR 2000, 18, 65-68).

The intermolecular interaction between the nuclear spins of a ligand and electron spins of a paramagnetically labeled drug target is the basis of the SLAPSTIC method for ligand-detecting NMR screening (Jahnke, Rüdisser et al .: Journal of the American Chemical Society 2001, 123, 3149–3150). This interaction leads to an extreme acceleration of the T ₂ relaxation and thus to the extinction of NMR signals only of those ligands that bind to the drug target in the vicinity of the paramagnetic label. The detection principle is comparable to the relaxation-filtered methods described. However, the disadvantage is the need for elaborate labeling of the drug target and the restriction of detection to localized binding in the vicinity of the paramagnetic labeling.

The biggest advantage of ligand-detected NMR screening is its potentially high throughput rate because usually only very short measuring times (a few minutes per sample) become

All known ligand-detecting NMR screening methods observe (ie measure the FID) for protons, which, however, are also present with high frequency in all drug target molecules. This results in the general and fundamental problem of being able to clearly, directly and simply distinguish the reporter signals of the substances from the background signals of the drug target within the screening spectra. For this purpose, two basic methods are described in the known methods, but in practice very inadequate. On the one hand, the proton spectrum of the free drug target can be subtracted from the screening spectrum. However, for fundamental reasons, this subtraction is never complete, especially not in the presence of binding ligands, which also induce shifts in the spectrum of the drug target. The only metrological possibility of direct signal suppression is to install a (T ₂ - or T _1ρ -) relaxation _filter , which suppresses the signals of the less mobile drug target much more than the reporter signals of the highly mobile test substances. However, this mobility-dependent selection presupposes that the test substances are considerably more mobile than the drug target molecules and in principle suppresses not only the drug target but also the target-bound state of a ligand. In principle, the detection of the bond can still take place indirectly via the signal decrease for the free form of the ligand. However, the reliable detection of signal intensities requires complete suppression of the background signals by the mobility filter, which can hardly be achieved in practice, which in particular allows the signals of locally flexible molecular regions of the drug target to pass.

The problem of signal suppression Resonances of the drug target can also be solved elegantly, by using ligands which carry NMR-active heteronuclei, that do not occur in the drug target or only occur with very little frequency. Enable these ligands the implementation from NMR experiments, where detected on the hetero nuclei or that are filtered or edited with the hetero core. In the resulting NMR spectra, the nucleus is free drug target invisible.

A disadvantage of this method is of course that the testing ligand must carry the heterocore. This limits the Choosing the right connections drastically and / or requires An additional synthetic effort.

The suitability of the hetero core ¹⁹ F is in this connection in the article "NMR-Based Approaches for Lead Generation in Drug Discovery" by JWPeng in Methods of Enzymology, Vol. 338 (2001), 202-230 and in the publication "Cross-Correlated Relaxation Measurements for the Study of Fluorinated Ligand-Receptor Interactions "also indicated by JW Peng in Journal of Magnetic Resonance 153 (2001) 32-47. However, none of these publications describes specific embodiments of this concept.

The known work around these problems Target-detected methods that detect binding-induced changes point at reporter signals of the drug target and - since these can be assigned to individual atoms of the drug target - immediately Record the location and type of ligand binding on the drug target. The In these methods, the reporter signals are clearly selected using a Magnetic magnetic fields Coupling (correlation) of several nuclear spins within the drug target molecule, like they do for the low molecular weight test substances is not possible. In this way become characteristic of two or more dimensions get fingerprint spectra from correlation signals of drug targets, in which (almost) no NMR signals of the test substances occur. It is crucial that this selective detection of the reporter signals both outdoors and is possible in the ligand-associated state of the drug target, whereby high-affinity ligands can now be observed directly can. A binding-induced shift is usually simply used as an indicator of the correlation signals.

In the target-detected NMR screening, on the one hand, homonuclear correlation spectra with two coupled ¹ H dimensions can be recorded as fingerprint spectra, as described in German patent application 101 60 177.8 (dated December 7, 2001). On the other hand, heteronuclear correlation spectra can also be measured, in which the frequencies of a nucleus different from protons in the indirect dimension are typically correlated with the frequencies of the directly bound protons in the directly measured dimension. The most common representatives of this technique are ¹⁵ N, ¹ H (WO 97/18469, WO 97/18471) and ¹³ C, ¹ H (WO 00/62074) NMR correlation experiments. For this, the drug target must be enriched with the NMR-detectable stable isotopes ¹⁵ N-nitrogen or ¹³ C-carbon, since their natural frequency (0.36% or 1.1%) is insufficient for fast NMR spectroscopy. The drug target is clearly marked by this - genetically simple - isotope exchange and marked in front of the test substances and the solvent. The isotope labeling thus allows a high-resolution specific and exclusive detection of the desired reporter signals of the drug target in free and associated form.

However, the advantages of target-detected NMR screening methods mentioned - atomically resolved direct detection and spatial assignment of both weakly and arbitrarily strongly binding ligands - are essentially offset by two disadvantages. On the one hand, the recorded spectra of the drug targets with increasing molecular size more and more complex and weaker in intensity due to rapid relaxation phenomena; immobilized or insoluble drug targets are hardly accessible to NMR spectroscopy in solution. On the other hand, the effective throughput rate that can be achieved is considerably lower compared to the above-mentioned ligand-detected NMR screening methods because the required multidimensional spectra require more measuring time.

3. Summary the invention

The present invention for ligand-detected NMR screening has the task of methods for NMR-based drug search to provide, which are free from those described above Disadvantages of the known methods. It becomes the task in particular treated to provide a method that has maximum throughput rates allows. Furthermore, the present invention has the task of methods for NMR-based search for active substances to provide an increased Security towards offers false positive and / or false negative results.

For this purpose, the method according to the invention transmits the aforementioned Principle of target-detected screening for the ligand-detected Screening by the conventional Selection of the ligand signals by molecular size through a clear and completely size-independent selection after NMR frequencies is replaced. The ligand is thus both in free (dissociated) as well as bound (target associated) form selectively observable.

The present invention further encompasses a method of assembling a special bank, which especially on the implementation of the method according to the invention is aligned with increased throughput rates. That bank exists exclusively from mixtures of such substances with NMR-observable isotopes those that are not marked in the drug target or only with a very small number are marked Frequency. The essential features of this method are listed in claim 1. The Subclaims 2 to 5 describe preferred embodiments of this method. There will be particular opportunities demonstrated how such substance banks with isotope-labeled Connections simple and inexpensive can be obtained taking advantage of the unique affinity sensitivity of NMR screening open up maximum chemical diversity and as molecular building blocks at the same time have high pharmacological relevance.

The present invention relates continue to refer to the substance bank according to claim 6 by this Process available is. Advantageous embodiments of the substance bank according to the invention are in the subclaims 7 to 10 specified.

Another aspect of the present Invention relates to ligand-detected screening methods to search for active substances using the substance bank described above, that are based on the measurement of elements or isotopes that are not (or only infrequently) are present in the drug target. The main features of this procedure are mentioned in claim 11 while the subclaims 12 to 17 particularly advantageous embodiments of this method describe. An alternative method is set out in claims 18 called to 21. The underlying problem of the invention According to the invention, the throughput rates are increased by the simultaneous measurement a large number of substances in the mixtures of the substance bank solved, which in turn can only be achieved by simplifying the spectra obtained enables the use of hetero cores becomes.

The screening method according to the invention can be carried out using various NMR experiments, which are the ligand-specific reporter signals of the isotope markers detect and thereupon both one or more of the binding indicators mentioned record in the dissociated as well as in the associated state. The obtained heteronuclearly edited or filtered screening spectra do not contain interference signals unmarked components more. This is the first time that the metrologically outstanding problem becomes of water suppression Completely solved. Likewise solves the present invention is superior in spectra evaluation Problem more troublesome Residual signals of the drug target, which were only possible with the previous methods incomplete repressed become. The screening spectra are also of maximum simplicity because they have few reporter signals per test substance (corresponding to the Number of built-in isotope markers). The is preferred Use of substance mixtures in which each test substance only generates a single reporter signal. On the one hand, this simplifies automated spectra evaluation and identification binding Ligands. The clarity on the other hand, the mixture spectra allow the compilation and clear spectral differentiation of larger substance mixtures and is thus in view of that on which the invention is based Classify problem as optimal. The method according to the invention allows the highest so far Throughput of all NMR screening methods with which over 10,000 Tested substances per day and immediately identified binding ligands can be.

4. Description of the drawings

1.a. shows a ¹⁹ F-NMR spectrum of submixture A without target protein (light gray, top), with lysozyme as a blank (gray, center) and with αII-chymotrypsin as target protein (black, bottom).

1.b shows a ¹⁹ F-NMR spectrum of submix B without target protein (light gray, top), with lysozyme as a blank (gray, center) and with αII-chymotrypsin as target protein (black, bottom).

1.c shows a ¹⁹ F-NMR spectrum of the submixture C without target protein (light gray), with lysozyme as a blank (gray) and with αII-chymotrypsin as target protein (black).

5. Detailed Description of the invention

5.1 Definitions

In the following, a drug target is understood to mean a molecule whose biological function and / or activity is to be changed by interaction with a ligand in such a way that a certain desired effect (for example the therapy of a disease) is achieved. Typical drug targets include naturally occurring proteins, DNA and / or RNA molecules. Such drug targets typically have a significantly larger molecular weight than the test substances. However, the present invention is not limited to such drug targets. It is even possible to use (large or immobilized) drug targets which are not accessible to NMR spectroscopy in solution, since these are not themselves detected in the method of ligand-detected NMR screening according to the invention. To carry out the described method, the drug target must be presented in sufficient quantity (approx. 10 ² mg per 1000 substances to be tested) and purity. Genetically modified microbial expression systems are particularly suitable for this. The corresponding biochemical methods are known to the person skilled in the art, in particular since isotope labeling is not required for the method of NMR screening claimed according to the invention.

The term ligand denotes one chemical (test) substance that binds to the drug target. Such ligands are typically low molecular weight chemical compounds that can be isolated as natural substances or are accessible via chemical synthesis; is the present invention for ligand search (screening) however independent from the size of the to testing substances.

Binding of the ligand to the drug target denotes any interaction between the two molecules that is strong enough to characteristically influence the NMR spectra of the ligand. Typically, the present method can detect bonds whose dissociation constant K _{D is} at least in the millimolar range. There is no upper limit on affinity.

As an isotope marker or hetero core an element or isotope is referred to in the context of this invention, which can be detected by NMR spectroscopy and in all molecules the Test substances is included while it does not exist in the corresponding drug target, or only in so little natural frequency (preferably less than 5%, stronger preferably less than 2%) occurs in the screening method according to the invention leads to no observable NMR signals of the drug target. This Requirement can also be met if the drug target also contains isotope markers, these however to NMR signals clearly outside of the frequency range of the isotope markers of the test compounds. The Object of the isotope marker in the context of the present invention is going to exclusively for the Test substances to provide NMR-observable reporter signals that always allow a clear differentiation from the drug target. The term "isotope labeling" does not imply that the Marking in question through a specific chemical reaction introduced into the connection becomes. For execution the present invention it is also possible that the connection is already being produced with the isotope marker.

Under a heteronuclear NMR experiment is understood such an NMR experiment in which either the The ultimately observable NMR coherence is excited on a hetero core, or where at least a coherence or magnetization transfer between two different cores Types (including at least one hetero core). Within the scope of this invention However, protons of the test substances such as heteronuclei also become treated and together with these under the term isotope marker summarized, whose signal frequencies are clearly outside the proton spectrum of the drug targets lie. The excited NMR coherence can be in the form of a heteronuclear edited NMR spectrum recorded in which the frequencies of the hetero core signals (in the present Invention referred to as reporter signals) are recorded directly. In contrast, a heteronuclear filtered NMR experiment the heteronuclear excited NMR coherence coupling protons and on this indirectly in the form of a selected proton spectrum detected.

Under measurement conditions are related of the present invention when recording an NMR spectrum elected Pulse sequence, the temperature at the time of measurement, the solvent and the pH of the solution to be measured, and the NMR spectroscopic parameters used in the measurement (in particular delays, transmission frequencies and pulse powers) understood.

5.2 Compilation procedure the substance bank according to the invention

• (i) Auswahl einer Anzahl von Substanzen, die jeweils eine oder mehrere Isotopenmarker des selben Typs aufweisen;
• (ii) Bestimmung der NMR-Resonanzfrequenzen der Isotopenmarker der ausgewählten Substanzen;
• (iii) Gruppierung der ausgewählten Substanzen in eine oder mehrere Teilmenge(n), so daß innerhalb jeder Teilmenge sich die Signalfrequenz eines jeden Isotopenmarkers mindestens um das Doppelte der jeweiligen Breite des Signals auf halber Höhe (sogenannte Linien- oder Halbwertsbreite) von den Signalfrequenzen aller anderen Isotopenmarker innerhalb der Teilmenge unterscheidet (d.h. daß der Abstand benachbarter Signale größer als die Summe ihrer Halbwertsbreiten ist) und innerhalb jeder Teilmenge alle Signalfrequenzen der Isotopenmarker innerhalb eines spektralen Bereiches von 15 kHz fallen, mit der Maßgabe, daß die Substanzen jeder Teilmenge untereinander nicht reagieren;
• (iv) Herstellung von einem oder mehreren Substanzgemisch(en), wobei für jedes Substanzgemisch alle Substanzen der jeweiligen Teilmenge gemäß Schritt (iii) gemischt werden.

An essential aspect of the present invention is a method for the compilation of substances according to the general and special requirements of the screening method according to the invention. This process includes the following essential steps:

• (i) selecting a number of substances, each having one or more isotope markers of the same type;
• (ii) determination of the NMR resonance frequencies of the isotope markers of the selected substances;
• (iii) Grouping of the selected substances into one or more subset (s), so that within each subset the signal frequency of each isotope marker is at least twice the respective width of the signal at half height (so-called line or half-value width) from the signal frequencies of all distinguishes other isotope markers within the subset (i.e. that the distance between adjacent signals is greater than the sum of their half-value widths) and within each subset all signal frequencies of the isotope markers fall within a spectral range of 15 kHz, with the proviso that the substances in each subset do not react with one another ;
• (iv) Production of one or more substance mixture (s), with all substances of the respective subset according to step (iii) being mixed for each substance mixture.

The substance bank according to the invention is a compilation a big one Number (e.g. over 102, preferably over 103, stronger preferably over 5.103) on substances that are in the form of mixtures. The maximal Number of substances that can be contained in the substance bank in principle not limited. Because of financally reasons and for reasons the practicability the maximum number of connections in most cases is The range is from 104 to 106 substances, with a typical maximum value is 105. The use of mixtures is with regard to the maximum possible Throughput of substances in the sceening process according to the invention essential aspect of the present invention. The number of mixtures should be as low as possible being held. The minimum number of mixtures required according to the quotient of the total number of substances in the Substance bank and the maximum number of substances in one Mixture can be combined without the essential criteria of the above step (iii) hurt. With a size of the substance bank of about 104 connections and a number of 50 to 100 connections for example, there is a number of about 100 per mixture up to 200 mixtures.

• – die Wahrscheinlichkeit erhöhen, daß innerhalb der Gemische unerwünschte Reaktionen zwischen den Substanzen auftreten;
• – Löslichkeitsprobleme in den Stocklösungen (s. Abschnitt 5.2.8 unten) verursachen können, da sie die Gesamtkonzentration an gelösten Substanzen in der Stocklösung erhöhen; und
• – da sie das Risiko bergen, daß stark bindende Liganden innerhalb der Gruppe der zusätzlichen Substanzen schwach bindende Liganden aus der Gruppe der erfindungsgemäßen Substanzen verdrängen, so daß letztere nicht oder nur schlecht als Liganden identifiziert werden.

It is of course not excluded that the mixtures contain, in addition to the substances selected according to the criteria described above, other substances which, for example, carry another isotope marker (eg ² H instead of ¹⁹ F). Such additional substances do not contribute to the concept of increasing throughput according to the invention, but can cause additional problems because they

• - increase the likelihood that undesirable reactions between the substances occur within the mixtures;
• - can cause solubility problems in the stock solutions (see section 5.2.8 below) as they increase the total concentration of dissolved substances in the stock solution; and
• - Since they carry the risk that strongly binding ligands within the group of additional substances displace weakly binding ligands from the group of substances according to the invention, so that the latter are not or only poorly identified as ligands.

Faces these potential problems just a small amount of time saved compared to that more Measurements on the additional Substances carried out can be without in between a sample change is necessary (as far as the sample head is at all allows). at modern equipment With an automatic sample changer or flow probe, this time can be saved however, do not compensate for the above potential problems. The production and use of mixtures in which no other except for substances selected according to the introductory rules containing isotope markers is therefore within the scope of the present Invention preferred.

5.2.1 General requirements

As is generally true for ligand-detected NMR screening methods, the substances in aqueous solution must be chemically stable over the duration of a screening run (> 1 day) and must not react with one another in a mixture. The NMR spectra of mixtures of suitable substances (mixture spectra) should also have no superimpositions in order to be able to further differentiate the individual components directly by spectral means. This condition is met if adjacent signal positions are at least twice the sum of their half-value widths apart. Adequate for inclusion in Intensive NMR spectra within a short time, the test substances should be at least 0.05 mM, more preferably 0.1 mM, particularly preferably 0.2 mM soluble in aqueous solution. However, since the method according to the invention also allows selective NMR observation of the ligands in the bound state for the first time, this solubility condition applies less strictly: now the ligands must meet this minimum solubility either in aqueous solution (ie in the free state) or in complex with the drug target. However, it cannot be ruled out that these minimum concentrations can be further reduced in the future through the development and use of more sensitive probe heads. In this case, of course, the required minimum concentrations in the DMSO stock solution (see section 5.2.8 below) decrease accordingly.

As an essential feature, the chosen Test substances each contain at least one isotope marker, i.e. a nucleus of an element or isotope, the NMR spectroscopic is observable and not in the drug target or only with a great deal less (preferably less than 5%, more preferably less than 2%) frequency occurs, or an NMR signal of the same kind clearly outside the frequency range Delivers elements or isotopes in the drug target. From NMR spectroscopic Nuclei with high NMR sensitivity and are particularly visible spectral dispersion advantageous to the intensive and well-resolved reporter signals to lead.

Furthermore, all test substances of the Substance banks have the same isotope marker.

It is preferred if there are as few reporter signals as possible per test compound in order to obtain NMR spectra of minimal complexity. The number of reporter signals received is equal to the number of isotope markers or magnetically equivalent isotope markers (eg in CX ₃ units). Test compounds which deliver only a single reporter signal are particularly preferred. In this way, a maximum of extensive substance mixtures (for example with up to 100 or 150 compounds) can be compiled in their mixture spectra without superimpositions, which can then be used directly in the NMR screening method according to the invention and can be distinguished immediately according to the individual components. The throughput rates are increased by at least a power of ten compared to conventional NMR screening methods. A minimal number of isotope markers makes it difficult to detect ligand binding via local effects (such as shifting the signal frequency or direct intermolecular magnetization transfer between ligand and drug target). The resulting risk of false negative display is eliminated by the (simultaneous) detection of ligand binding via global effects (such as diffusion or relaxation-modulated signal intensities), which can be determined at any position in the ligand without major measurement effort. However, if primarily local binding effects are to be recorded, the highest possible distribution of the isotope markers in the ligand is advantageous.

Isotope labeling of the mostly synthetic, non-biogenic test substances is more difficult than the now almost routine genetic labeling of drug targets with the NMR-active nuclei ¹³ C, ¹⁵ N or ² H (Goto and Kay: Current Opinion in Structural Biology 2000, 10 , 55-592). In particular, a general isotope exchange - for example of the predominant, NMR-invisible ¹² C carbon (with 99% natural frequency) against the rare, NMR-spectroscopic ^{1 3} C isotope - is difficult to carry out synthetically. However, there are a number of methods for attaching NMR-active nuclei to the test substances as isotope markers by chemical derivatization.

While the isotope marker as the supplier of the NMR reporter signal is essential for the method according to the invention, further general requirements have to be made of the substance bank for the successful search for active substances. Above all, it should cover the greatest possible chemical diversity in order to increase the likelihood of being hit by the NMR screening process. The sensitivity of the screening process with regard to the minimum detectable binding strength is unsurpassed in NMR screening. For this reason, the ligands to be found do not have to fit into their complementary binding pocket on the drug target as closely as is necessary for a hit display in the less sensitive high-throughput screening methods (HTS). The chemical-structural space must therefore be covered less densely by a substance bank optimized for NMR screening. The substance bank can be limited to a considerably smaller number (10 ³ - 10 ⁵ compounds) of simple chemical frameworks, which are generally less strong, but are more likely to bind to a given drug target. In contrast to more complex compounds, these simple basic motifs can be purchased or can be easily synthesized using certain isotope markers. The problem of accessibility of a chemical diversity sufficient for NMR screening with isotope-labeled compounds can thus be solved easily.

A possible problem is the influence of the isotope marker on the pharmacological or binding chemical properties of a ligand. If the isotope marker is only installed temporarily and is to be removed again in the medical-chemical optimization process, the pharmacological properties (eg toxicity, bioavailability, etc.) of the isotope marker are irrelevant. It is then important, however, that the isotope marker does not significantly influence the ligand binding to the drug target, so that the affinity between the drug target and the ligand is maintained even when the isotope marker is subsequently removed remains. If the isotope marker remains permanently in the ligand, only its pharmacological properties are relevant.

5.2.2 Fluorine as an isotope marker

In a particularly preferred embodiment of the method according to the invention, the element fluorine is used as the isotope marker and, accordingly, a substance bank composed of fluorine-containing compounds is put together for the purpose of fluorine-detected NMR screening. The element fluorine is present with 100% natural abundance in the form of the outstandingly NMR-accessible isotope ¹⁹ F and, moreover, is present in practically no known biogenic substance and thus also in practically no known drug target. At the same time, tens of thousands of fluorinated substances are commercially available, which can be used to image the chemical diversity required for effective ¹⁹ F-NMR screening. Fluorinated compounds also have above-average pharmacological properties, which is why ¹⁹ F can mostly remain as a permanent isotope marker. For example, aromatic active substance candidates are preferably fluorinated in the para-position in medical-chemical optimization in order to block their rapid cytochromatic degradation. ¹⁹ F is also particularly suitable as a temporary isotope marker, since the atom is almost isosteric (of the same size) to hydrogen and the chemical properties of a compound usually change only marginally after ¹⁹ H is replaced by ¹ H. With regard to its NMR spectroscopic properties, the spin-½ core ¹⁹ F with 83% relative sensitivity is almost equivalent to the proton ¹ H. With comparable line widths and signal intensities, the spectral dispersion for ¹⁹ F, on the other hand, is approx. 30 times larger than for Therefore, considerably more extensive substance mixtures (for example with up to 100 to 150 compounds) can be used without overlaps in their mixture spectra in 19F NMR screening and immediately a distinction is made, which increases the throughput rates by at least a power of ten compared to conventional (proton-detecting) NMR screening methods. In addition, the signal frequencies of ¹⁹ F are about 30 times more sensitive to changes in the local chemical environment compared to ¹ H and can therefore register binding-induced effects even at a considerably greater distance. For the small chemical backbones that are preferably used, a single ¹⁹ F core is often sufficient in order to be able to reliably detect a ligand binding even on the basis of the local signal frequency indicator.

5.2.3 Deuterium as an isotope marker

In a further preferred embodiment of the method according to the invention, the hydrogen isotope deuterium ( ² H) is used as the isotope marker and, accordingly, a substance bank composed of deuterium-containing compounds is put together for the purpose of ² H-detected NMR screening. Deuterium is practically undetectable with only 0.015% natural frequency in all biogenic compounds and drug targets. In contrast to fluorinated compounds, however, deuterated compounds are not commercially available in the required variety. However, the targeted incorporation of deuterium using chemical synthesis methods is widely possible and can cover the required chemical diversity well. One example is the accessibility via commercially available halogenated compounds (chlorine, bromine or iodine-containing compounds) by conversion to the corresponding Grignard salts and absorption in heavy water ( ² H ₂ O) (Jerry March: Advanced Organic Chemistry; Wiley -Interscience, ISBN 0-471-85472-7; H. Beyer, W. Walter: Textbook of Organic Chemistry; S. Hirzel Verlag Stuttgart; ISBN 3-7776-0406-2). Binding chemical and pharmacological properties are identical for the ² H and ¹ H isotopes, so that deuterium can be used as a permanent or as a temporary isotope marker that can be exchanged for ¹ H at any time. Deuterium has only about 1% of the NMR sensitivity of ¹ H, which is, however, still sufficient for sufficiently fast ² H-detected NMR screening. Spectral dispersion (in ppm) and chemical shift (in ppm) are the same for the ¹ H and ² H isotopes. In contrast to ¹ H, ² H is a quadrupole nucleus (spin = 1), the relaxation of which is much more sensitive to changes in the local chemical environment than with a spin ½ nucleus such as the proton. In this respect, signal intensity and width become more meaningful or relaxation-filtered ² H-NMR measurements compared to ¹ H as binding indicators.

5.2.4 Tritium as an isotope marker

The methods according to the invention can also be used as an isotope marker using the third hydrogen isotope, tritium ( ³ H). The explanations regarding the hydrogen isotope ² H apply accordingly. However, the NMR properties of this spin ½ core are much more similar to the proton. However, it should be noted that tritium is radioactive, so that its use requires a high level of safety engineering.

5.2.5 ¹³ C as an isotope marker

In a further preferred embodiment of the method the carbon-isotope is used ^{1 3} C and according to the right ^{1 3} C detected a substance Bank ^{1 3} C-containing compounds for the purpose or indirectly, ¹³ C-filtered / ¹ H-detected as isotope labels NMR screenings compiled. With only 1.1% natural frequency, ^{1 3} C is practically undetectable in all biogenic compounds and drug targets, as is ² H. However, ^{1 3} C-labeled compounds are commercially very expensive and only available with a low chemical diversity. However, here too there are numerous possibilities for introducing ^{1 3} C by means of chemical synthesis processes. For example, commercially available ^{1 3} C-methyl alcohol can be installed (Jerry March via standard esterification and etherification reactions: Advanced Organic Chemistry, Wiley-Interscience, ISBN 0-471-85472-7; H. Beyer, W. Walter: Textbook of Organic Chemistry ; S. Hirzel Verlag Stuttgart; ISBN 3-7776-0406-2). Binding chemical and pharmacological properties are identical for the isotopes ¹³ C and ¹² C, so that ¹³ C can be used as a permanent as well as a temporary isotope marker that can be exchanged for ¹² C at any time. ¹³ C is a spin ½ core that is very well measurable by NMR spectroscopy. The relative NMR sensitivity of 1.6% (related. ¹ H) is sufficient for quick ^{1 3} C-detected NMR screening. It can be significantly increased if ^{1 3} C-bound protons are observed, ie ¹³ C-filtered / ¹ H-detected is measured (these indirect isotope-filtered measurement methods are generally feasible if the isotope is directly bound to a proton) , In ^{direct-1 3} C- be detected, NMR screening, however, the opposite ¹ H greater ca. 20x spectral dispersion acts in particular advantageous effect on the maximum possible number of test substances that may be present in the mixture without overlap of their reporter signals. This allows a further increase in throughput rates, as already described for ^{1 9} F-detected NMR screening.

5.2.6 ³¹ P as an isotope marker

In a further embodiment of the method according to the invention, the element phosphorus is used as the isotope marker and, accordingly, a substance bank composed of phosphorus-containing compounds is put together for the purpose of phosphorus-detected NMR screening. The element phosphorus is present with 100% natural frequency in the form of the isotope ^{3 1} P, which is easily accessible by NMR, but is also frequently found in drug targets (for example in phosphorylated proteins and in DNA and RNA). However, the NMR spectral range is relatively narrow here and can be distinguished from the ^{3 1} P signals of phosphorus-containing test substances. While the NMR properties of the spin ½ core ³¹ P are favorable (6.6% of the NMR sensitivity of ¹ H, but an approximately 40 times greater spectral dispersion), the accessibility of a wide chemical variety of phosphorus-containing compounds is problematic. Because of the often significant influence on the binding properties, phosphorus is mostly only suitable as a permanent isotope marker. A general statement about the pharmacological properties cannot be made in view of the very small number of known phosphorus-containing active ingredients. ^{3 1} P-detected NMR spectroscopy using the method according to the invention is therefore particularly useful in individual cases with specialized substance banks.

5.2.7 Further isotope markers

The method according to the invention can also be used further NMR-observable isotope markers. A list of other observable isotope markers can e.g. the BRUKER-Almanac from the company BRUKER Biospin GmbH (nmr@bruker.de).

5.2.8 Production of stock solutions

The test substances selected in this way are preferably provided in highly concentrated form in dimethyl sulfoxide (DMSO), in which they are generally very soluble and are durable.

5.2.9 Production of mixtures of substances

When carrying out the method according to the invention, the test substances are not provided individually, but already as mixture stock solutions, in which the components should preferably have a concentration of at least 0.05 M (more preferably 0.1 M) so that they can be added to the aqueous solution of the drug target as soon as possible small volume can take place; this is necessary because above about 5 to 10 percent by volume of DMSO, structural changes in the drug target protein can occur. This considerably simplifies the screening routine, the screening of substance mixtures not only for increased sample throughput, but also for reasons which will be explained in more detail below is advantageous in terms of spectra standardization. When compiling these mixed stock solutions, it is of course important to ensure that the components do not react with one another. Furthermore, the reporter signals of the components must not overlap in order to continue to enable their clear and direct spectral differentiation. Due to fundamental NMR methodological problems with homogeneous broadband excitation, the mixtures should finally be compiled in such a way that all reporter signals contained fall within a spectral range that is covered by the maximum excitation range of the NMR experiment used in the screening process ( with the currently available pulse profiles, this is a maximum of approx. 15 kHz; however, it cannot be ruled out that in the course of future further developments of hardware and the development of further, improved pulse profiles, the maximum excitation widths that can be achieved will be over 15 kHz; in this case it is of course possible adapt the composition of the mixture to the available excitation widths within the scope of the present invention). The reporter signals within a substance mixture should therefore be in a spectral range of less than 15 kHz, preferably less than 10 kHz. This condition is particularly important for the isotope marker ^{1 9} F, whose known spectral dispersion is above 150 kHz (corresponding to 300 ppm at 500 MHz field strength). Taking into account the criteria mentioned (sufficient solubility, no signal overlap and spectral dispersion lower than approx. 15 kHz), mixed stock solutions with, for example, up to 100 to 150 components can be put together and used directly in NMR screening. Of course, the requirement for the stability of the stock solutions also means that the compounds within each mixture do not react with one another. A comparison of the NMR spectra of the mixture over a certain period of time can serve as a test for this criterion: the mixtures have sufficient stability if the NMR spectrum of the mixture over a period of at least one day (preferably: one month, more preferred) : one year) does not change significantly (ie that there are no signal reductions of more than 10% (preferably 5%) and no new signals of more than 5% (preferably: 2%) of the intensity of the respective reporter signals). In view of the object according to the invention of increasing the throughput of measured compounds, it is preferred to prepare mixtures which contain at least 10 (preferably at least 20, more preferably at least 50) substances. Limiting factors for the maximum number of compounds in a mixture are signal overlap, spectral width and reactivity of the substances contained.

5.3 The substance bank according to the invention

The substance bank according to the invention relates to a Compilation of substances that meet the above requirements Fulfills and therefore for the implementation of the screening method described below is suitable.

5.4 The screening method according to the invention

• (A) Aufnahme eines nach dem vorliegenden Isotopenmarker editierten oder gefilterten Referenzspektrums einer Substanzmischung aus der Substanzbank gemäß einem der Ansprüche 6 bis 10;
• (B) Herstellung einer Mischung aus dem drug target und der Substanz oder Substanzmischung nach Schritt (A);
• (C) Aufnahme eines Screeningspektrums der in Schritt (B) hergestellten Mischung unter den Meßbedingungen aus Schritt (A);
• (D) Identifizierung bindender Liganden durch Vergleich des Screeningspektrums mit dem Referenzspektrum anhand von Veränderungen der NMR Signale oder des NMR Signals der oder des bindenden Liganden.

In its preferred embodiment, the screening method according to the invention comprises the following steps:

• (A) recording a reference spectrum of a substance mixture from the substance bank edited or filtered according to the present isotope marker according to one of claims 6 to 10;
• (B) preparing a mixture of the drug target and the substance or substance mixture after step (A);
• (C) recording a screening spectrum of the mixture prepared in step (B) under the measurement conditions from step (A);
• (D) Identification of binding ligands by comparing the screening spectrum with the reference spectrum based on changes in the NMR signals or the NMR signal of the binding ligand or ligands.

5.4.1 Recording the reference spectrum (Step A)

The recording of the reference spectrum of a substance mixture in step A of the screening method according to the invention is generally carried out before addition to the drug target, so that a sample change is required between the reference and screening spectrum. The sample conditions - above all pH value, temperature, solvent, sample and salt concentration - should match as exactly as possible, since otherwise they will undesirably influence binding indicators and can therefore lead to a false positive display. This danger threatens in particular for the highly sensitive NMR signal frequencies as an indicator. For the method according to the invention, the test substances are therefore generally measured in dilute, aqueous solution under screening conditions in order to record the reference spectra, ie at the same temperature and concentration in the same buffer solution in which the drug target is also dissolved during the subsequent screening becomes.

According to the invention, the reference spectra are not recorded for individual substances, but for the substance mixtures compiled according to the above criteria. For this purpose, by appropriately adding the DMSO-containing mixture stock solution to the aqueous buffer solution, a substance concentration is set which will later also apply to the screening (generally in the range 10 ⁻¹ to 10 ⁰ mM). A concentration setting that is only of the same order of magnitude is usually sufficient, since this parameter generally has little effect on the NMR indicators. As mentioned, however, the DMSO addition should not exceed 5 percent by volume and precipitation of the substance should be avoided. The effect of concentration changes on the reference spectra can be determined using a titration series. The buffer concentration must be sufficiently high to be able to absorb pH changes due to acidic or basic reaction of the substance mixture and the drug target. The buffer capacity may limit the possible size and addition of the substance mixtures. As a rule, the measurement temperature has a major influence on the NMR indicators, which therefore has to be the same for reference and screening spectra. The temperatures prevailing in the reference measurement and in the screening measurement should therefore not differ by more than 1K.

The particular advantage of the method according to the invention the measurement of substance mixtures is next to the time saving Fact that reference spectra inherent in mixtures of substances are standardized. So can not only absolute but also relative signal frequencies can be determined. The signal of a test substance is thus relative by its position standardized to the signals of the other test substances in the mixture. In particular, however, the standardization of relative signal intensities is also possible are usually more important as binding indicators in practice. The relative Standardization of signal frequencies and intensities will all the more reliable the more extensive the mixtures of substances are.

In a further embodiment of the screening method according to the invention, additional standards for (relative) signal frequencies and intensities can also be added to the substance mixtures. As a standard, any substance is suitable that is sufficiently water-soluble and does not interact with the drug target, so that it is constructed as simply as possible and without functional groups. Suitable standards for ¹⁹ F-detected NMR screening would be TFE (trifluoroethanol), TFA (trifluoroacetic acid) or hexafluorobenzene.

When repeating the screening method according to the invention on different drug targets, it is particularly advantageous to: NMR screening always at the same measuring temperature, pH value and buffer concentration perform, in order not to have to record reference spectra each time. This is indeed for a big Number of different drug targets possible around room temperature (298K) and pH = 7 (preferably in> 50mM Phosphate buffer) are stable and functional. It is still recommended pH and temperature dependence of the reference spectra by varying them accordingly To determine parameters. Furthermore, the long-term stability of the mixture stock solutions should also be ensured latter Follow-up measurements are checked.

The recording of the reference spectrum should use the same NMR experiment and the same measurement parameters done as for the screening spectrum. It is described in detail there.

5.4.2 Addition of the test substances to the drug target (step B)

The addition of the test substances to the solution the drug target (step B) is preferably in the form of a corresponding one Amount of mixture stock solution up to a maximum of about 5 percent by volume of DMSO. In this screening Mixture is said to be the concentration of each substance below that of the drug target because there is an excess of ligand the ratio of bound to free Species and thus usually the rash of the binding indicator reduced. Moreover remains only in the deficit too enough free drug target in the presence of a strongly binding ligand for the detection of competing weaker ligands. Many test substances are in aqueous solution anyway less soluble as the drug target. The substance concentration is therefore up through solubility or solubility of the drug target, sufficient for the exposure downwards through the required measuring time limited intense NMR spectra. A minimum concentration is preferred of 0.1 mM, which makes the measurement sufficiently intense one-dimensional Spectra of the heteronuclear isotope markers within 5 minutes allows. The use of highly sensitive cryo measuring heads can lower the concentration limit to below Lower 0.05 mM.

5.4.3 Recording the screening spectrum (Step C)

5.4.3.1 General

To record the screening spectrum (step C), an NMR experiment edited or filtered according to the present isotope marker is carried out on the screening mixture, which is suitable for recording the NMR indicators signal frequency and (bond-modulated) signal intensity on the reporter signals to draw. While the signal frequency can in principle be obtained from any conceivable NMR experiment, the signal intensity can be modulated by means of various NMR experiments of different bond-dependent phenomena (such as diffusion, relaxation or magnetization transfer).

Like signal frequencies, signal intensities are subject to the influence of several parameters, but, in contrast to the latter, are less dependent on sample conditions than on measurement parameters. In this respect, it must primarily be ensured by measurement technology that the modulation of the signal intensities is dominated by the desired binding-dependent effects. In the NMR screening techniques of the method according to the invention, the excitation takes place on the isotope markers, which can sometimes have long enthalpic T ₁ relaxation times. With a rapid sequence of the scans of an NMR experiment, the saturation of the reporter signals can vary, depending on the extent of the excitation and the inter-scan waiting time. Partial saturation falsifies the signal intensities and must be minimized by small excitation pulse angles or sufficiently long inter-scan waiting times, which should be at least three times the longest T ₁ relaxation time. T ₁ relaxation times can quickly be determined by a person skilled in the art using inversion recovery experiments (Bodenhausen: Progress in Nuclear Magnetic Resonance Spectroscopy 1981, 14, 137ff) on the pure substance mixture. In practice, an Interscan waiting time of 2-3 seconds is usually sufficient.

5.4.3.2 Use of inverse, heteronuclear filtered experiments

In an alternative embodiment of the heteronuclearly edited experiments described below, with direct observation and frequency recording of the isotope markers, it is also possible to select for these if the isotope marker strongly couples with protons, such as, for example, ^{1 3} C to directly bound ¹ H. In this case, indirect reference is made to the scalar coupled protons detected. The NMR modules for selecting hetero nuclei (so-called isotope filters) are familiar to the person skilled in the art (Breeze: Progress in Nuclear Magnetic Resonance Spectroscopy 2000, 36, 323-372). The advantage of this method is the exploitation of the usually considerably higher NMR sensitivity and the possibly more favorable relaxation properties of the protons compared to the isotope marker. However, the necessary suppression of the intensive water signal as well as the lower dispersion and high superposition in the homonuclear (mixture) proton spectrum are problematic. With strong coupling, the unobserved core should always be decoupled.

5.4.3.3 Basic experiment

In a preferred embodiment of the method according to the invention, the simplest conceivable NMR experiment, consisting of only one excitation pulse for the isotope markers, is carried out to generate a one-dimensional spectrum of the heteronuclear reporter signals. This experiment has naturally the largest excitation width and is therefore suitable especially for the C ^{1 3 19} F isotope marker and. The resulting spectrum not only records the signal frequencies, but also the relaxation-dependent and thus bond-modulated signal intensities that are subject to entropic T ₂ relaxation during the duration of the signal recording.

This basic experiment shows - due to the minimum number of just one excitation pulse - the largest homogeneous Bandwidth and is therefore particularly suitable when substance mixtures with very high spectral dispersion of the reporter signals Need to become. However, the experiment also points out because of its simplicity the least certainty of display because the NMR-observable binding-induced Effects not intensified here become.

5.4.3.4 Use of relaxation filters

In a particularly preferred embodiment of the method according to the invention, the extent of the bond-modulated signal attenuation due to entropic relaxation is further increased by the installation of a relaxation filter. This T ₂ or T _{1 ρ} relaxation is particularly suitable for clearly indicating the binding-related decrease in mobility of a ligand. This applies in particular to isotope markers with quadrupole nuclei (eg ² H) which are particularly sensitive to relaxation when their chemical environment changes. To emphasize the relaxation effect, only the runtime of the NMR coherence has to be extended in the experiment. In the case of T ₂ filtering, this is done by installing one or more centric 180 ° rephasing pulses on the isotope marker in order to achieve spin echoes. In the case of T _1ρ filtering, the NMR coherence is not periodically rephased by pulses, but is frozen by continuous irradiation of a weak radio frequency field (so-called SpinLock). Both methods are well known to the person skilled in the art (Meiboom S., Gill D., Rev. Sci. Instrum. 1958, 29, 688; Hajduk PJ, Olejniczak ET, Fesik SW, J.Am.Chem.Soc. 1997, 119, 12257 –12261: "One-dimensional relaxation- and diffusion-edited NMR methods for screening compounds that bind to macromolecules"). The additional emphasis on the binding indu decorated relaxation losses through the relaxation filter represents a significant improvement compared to the basic experiment mentioned in 5.4.3.3. The effectiveness of the module decreases sharply with increasing spectral dispersion of the reporter signals in mixtures; above about 10 kHz spectral dispersion, this embodiment should no longer be used.

5.4.3.5 Use of diffusion filters

In a further embodiment of the claimed method, a diffusion filter for binding-induced modulation of the signal intensities is implemented. This is done by incorporating a field gradient echo (Johnson: Progress in Nuclear Magnetic Resonance Spectroscopy 1999, 34, 203-256; Wu, Chen and Johnson: Journal of Magnetic Resonance 1995, A115, 260). The diffusion filter dampens the mobile free form more than the immobile target-bound form of the ligands; the (weaker) T ₁ relaxation damping acts in the same direction during the diffusion mixing time. The entropic T ₂ relaxation attenuation, which is of particular importance for the immobile target-bound form and which inevitably acts during the gradient echoes and the acquisition time, is opposite to the effect of the diffusion filter and thus undesirably reduces the amplitude of the bond-induced signal modulation. To avoid false-positive display, in which the diffusion-related signal amplification is compensated by T ₂ -related signal attenuation, the drug target itself should still be observable by NMR, ie not so large that its rapid relaxation leads to signal cancellation.

Otherwise, the diffusion filter module a higher one Bandwidth than the relaxation filter module and can therefore also be used for mixture spectra with more than 10 kHz can be used well.

5.4.3.6 Use and measurement of the heteronuclear ¹ H, X-NOE as a binding indicator

In a further preferred embodiment of the method according to the invention, the reporter signal intensity is modulated by the bond-dependent heteronuclear ¹ H → X-NOE. As is known to the person skilled in the art, proton magnetization is disturbed for this purpose, for example by inversion or preferably by continuous irradiation (saturation). The magnetization transferred to the isotope marker by dipolar coupling is read off after a certain NOE development time or saturation period (from 10 ⁻¹ to 10 ¹ seconds) as intensity modulation of the reporter signal. It is obtained by comparison with an identical spectrum without disturbing the proton magnetization (for this purpose, the proton transmission frequency is preferably simply shifted outside the resonance range by at least 20 kHz (preferably by at least 50 kHz) in the same experiment). Division of both (one-dimensional) spectra gives the value of the ¹ H → X-NOE; Division of the difference between the two spectra by the undisturbed signal intensities gives the value of the ¹ H → X-NOE gain, η.

The measurable ¹ H → X-NOE is made up of several contributions. The intermolecular ¹ H → X-NOE between protons of the drug target and the isotope marker of the ligand is an immediate and local binding indicator, so that if possible, several isotope markers should be distributed over the ligand molecule to ensure that they can be detected. The intramolecular ¹ H → X-NOE between protons and isotope markers of the same ligand molecule, however, indicates binding as a relaxation phenomenon indirectly and globally via a decrease in mobility. With the exception of the heteronuclear ¹ H → ¹⁹ F-NOE, which is comparable to the homonuclear ¹ H → ¹ H-NOE, both the amplitude and the build-up rate of the ¹ H → X-NOE generally decrease with mobility. For strong binding ligands, two separate sets of signals are observed for free and bound forms. Naturally, only the latter can indicate the bond. For weakly binding ligands, an averaged set of signals is observed in which the ¹ H → X-NOE of the free and bound forms overlap. Again, only the contribution of the intramolecular ¹ H → X-NOE for the bound ligand, known as transfer NOE (trNOE), is modulated by the decrease in mobility due to binding.

In a preferred embodiment of this embodiment, the distinction between intermolecular and intramolecular ¹ H → X-NOE is dispensed with, since any changes in the reporter signal intensities modulated by the ¹ H → X-NOE after addition of the substance mixture to the drug target are obviously attributed to ligand binding can be. This very simple experiment has a range comparable to the basic experiment mentioned in 5.4.3.3, with the main areas of application described there. However, intensity-modulating intra- and intermolecular NOE can be compensated for, with the risk of false negative display.

In a further preferred embodiment, intermolecular and intramolecular ¹ H → X-NOE are distinguished, for which purpose it is necessary to differentiate the protons of the substances and the drug target based on their frequencies. The assignment of the ¹ H → X-NOE to a specific proton can be checked by selective interference of its frequencies and observation of the effect on the reporter signal intensity. As an alternative to these selective one-dimensional ¹ H → X-NOE spectra, a two-dimensional ¹ H, X-NOESY spectrum can be recorded. It contains only correlation signals that indicate the frequencies of the isotope markers (in the direct dimension) and all dipolar coupling protons (in the second, indirect dimension) correlate.

In a particularly preferred embodiment of this embodiment, only proton frequencies which can be clearly assigned to the drug target are selectively disturbed (Klein, Meinecke et al .: Journal of the American Chemical Society 1999, 121, 5336-5337; Mayer and Meyer: Angewandte Chemie International Edition in English 1999, 38, 1784-1788). The larger the drug target molecule, the faster the phenomenon of spin diffusion helps in the desired distribution of the perturbation over the entire molecule. The one-dimensional spectrum of the isotope marker then reveals only the intermolecular transfer of magnetization or saturation from the drug target to the ligand. As a particular advantage, this immediate and unambiguous binding indicator does not require a reference measurement on the free substance (mixture), and therefore no sample change - only a comparison spectrum without proton saturation has to be carried out. The broad spectrum of the experiment is comparable to the basic experiment mentioned in 5.4.3.3. It should only be used if very strong binding ligands (with KD <10 ^–8 M) are sought which cannot be discovered with this method. The method can only be used if there is at least one clearly different proton spectral band for the drug target with respect to the test substances, which allows the required selective saturation of the target. In a special embodiment, the intermolecular magnetization or

Saturation transfer inversely measured as X → ¹ H-NOE on the drug target, particularly preferred in the case of X = ¹⁹ F. This has the advantage that, due to the absence of the isotope marker in the drug target, selective saturation, in this case on the test substances, is always is possible and therefore false positive indication cannot occur due to partial intramolecular saturation. However, this advantage is bought with the necessity to be able to observe the drug target by NMR spectroscopy, which limits its size to a maximum of approximately 40-50 kDa. Similarly, the signals of all isotope markers within substance mixtures must be within 10 kHz, since otherwise homogeneous saturation is no longer possible.

In a further particularly preferred embodiment of this embodiment, a two-dimensional ¹ H, X-NOESY spectrum is recorded, in which intra- and intermolecular ¹ H → X-NOE can be differentiated using the proton frequencies. The observation of intermolecular correlation signals between protons clearly assignable to the drug target and the reporter signal of a ligand is again direct evidence of the binding. If the signal positions of the intramolecular protons of the individual ligands are known, a reference measurement can be omitted. In the case of the ¹ H, ¹⁹ F-NOESY, an exception is also immediately recognizable from the intramolecular correlation signals of the ligand, since here as in the homonuclear ¹ H, ¹ H-NOESY there is a sign reversal in the transition between highly mobile free (negative) and immobile bound form (positive) results. In contrast to the homonuclear case - as mentioned at the beginning - both weak and strongly binding ligands can now be clearly detected due to the isotope marker, since the bound state of the ligand can also be observed selectively. These experiments are particularly preferred when, due to a lack of adequate spectral differentiation of the proton signals from drug target and test substances, selective saturation of only one species (usually the drug target) is not possible.

• (A') Herstellung einer Mischung aus dem drug target und der Substanzmischung aus der Substanzbank gemäß einem der Ansprüche 6 bis 10;
• (B') Aufnahme eines NMR-Screeningspektrums der Mischung aus Schritt (A), mit dem selektiv ein intermolekularer heteronuklearer Magnetisierungstransfer zwischen Protonen des drug targets und Isotopenmarkern der Liganden erfaßt werden kann;
• (C') Aufnahme eines Referenzspektrums anhand derselben Mischung und mit demselben NMR-Experiment wie in Schritt (B), wobei nunmehr jedoch die Störung der Protonenmagnetisierung unterbleiben muß; im Falle des zweidimensionalen ¹H,X-NOESY-Experimentes kann die Aufnahme eines Referenzspektrums entfallen;
• (D') Identifizierung bindender Liganden anhand von Veränderungen ihrer NMR-Signale zwischen Referenz- und Screeningspektrum bzw.

The selective detection of the intermolecular ¹ H → X-NOE as a direct binding indicator by means of one of these experiments thus results in a modified sequence of the method according to the invention, in which the recording of the reference spectrum is facilitated or even made unnecessary. It includes the following steps:

• (A ') Preparation of a mixture of the drug target and the substance mixture from the substance bank according to one of claims 6 to 10;
• (B ') Recording an NMR screening spectrum of the mixture from step (A), with which an intermolecular heteronuclear magnetization transfer between protons of the drug target and isotope markers of the ligands can be detected selectively;
• (C ') recording a reference spectrum using the same mixture and with the same NMR experiment as in step (B), but the interference with the proton magnetization must now be avoided; in the case of the two-dimensional ¹ H, X-NOESY experiment, the inclusion of a reference spectrum can be omitted;
• (D ') Identification of binding ligands based on changes in their NMR signals between the reference and screening spectrum or

It should also be noted about the experiments described in this section that, unlike the other experiments described, they are unable to detect any strongly binding ligands. On the other hand, these experiments have the advantage that they can be carried out with much lower concentrations of the drug target (for example in the range from 10 μm to 100 μm). It is therefore advisable to work with as high a ligand (or substance) excess as possible in these experiments, since the binding indicator is not due to the proportion of dissociating ligands is characterized by the proportion of bound ligands.

5.4.4 Identification binding ligand (step D)

Binding ligands are identified (step D) by comparing the reporter signals of the isotope markers in the reference and screening spectrum; When observing the heteronuclear ¹ H → X-NOE as a binding indicator, the special features described there apply. In any case, both the signal frequency and the relative signal intensity, if necessary also the signal width, are compared and significant changes suggest ligand binding. In a particularly preferred embodiment of the claimed method, the frequencies and relative intensities of all reporter signals of the reference spectrum are first recorded for each substance mixture. Such reference signal lists can be easily and automatically generated by commercial NMR software. Subsequently, the (local) intensity maximum is searched for in the associated screening spectrum within a certain tolerance range around the frequencies specified in the reference signal list; this detects frequency and intensity changes equally. The quotients of corresponding signal intensities and their mean are then calculated. A significantly different intensity quotient indicates binding for the corresponding substance. For example, divergence by more than the standard deviation can be considered significant. With this method it is immaterial whether binding ligands experience a signal increase (eg due to slow diffusion) or a signal attenuation (eg due to accelerated relaxation) through the NMR experiment.

6. Embodiments

The inventive method is exemplified a set of 12 fluorinated compounds for ^{1 9} F-detected NMR screening. Bovine all-chymotrypsin is used as the target protein (Sigma-Aldrich No. C4129). Measurements were taken on a BRUKER DMX600 (with 600 MHz proton frequency) with a ^{1 9} F, ¹ H dual measuring head. At this field strength, the ¹⁹ F frequency is approximately 564.6 MHz.

The fluorinated substances are listed in Table 1 together with their ¹⁹ F signal frequencies. The measurement conditions for this (0.25mM aqueous solutions at pH = 7.50mM phosphate buffer, 298K) correspond to the sample conditions of the target protein, which are decisive for the screening. The substances were provided as 0.5M mixture stock solution in D ₆ -DMSO.

The ^{1 9} F chemical shift was determined for each of the test substances under the above-mentioned standard conditions (cf. Table 1). In order to be able to guarantee a largely homogeneous reporter signal excitation, the test substances were then put together in three submixtures, so that the ¹⁹ F signals contained in each submixture have a maximum dispersion of approximately 15 kHz (at 600 MHz field strength). These three sub-mixes were provided as mixed stock solutions and also measured subsequently. The fixed stoichiometric ratios of the individual test substances to one another thus allow a relative referencing of both the frequencies and the intensities of the reporter signals.

1a) - c ) shows the very clear, overlay free ^{1 9} F-spectra of the Subgemische (oa under standard conditions) that contain only a single reporter signal per test substance. In contrast, the associated ¹ H spectra are very complex due to the larger number of proton signals, some of which are highly superimposed, and can no longer be separated and are dominated by the water signal, which cannot be completely suppressed.

The ¹⁹ F spectra shown were recorded with the simplest and broadest-band basic experiment, consisting of only a single pulse on the ¹⁹ F frequency. The pulse had a duration of 15 μs at maximum transmitter power (approx. 50 W), corresponding to a pulse angle of approx. 75 °. During the recording of the fluorine signal, broadband decoupling from the protons (using the WALTZ16 sequence (Shaka AJ, Keeler, J., Frenkiel T., Freeman R., J.Mag.Reson. 1983, 52, 335-338: "An improved sequence for broadband decoupling: WALTZ16 ") and a 90 ° decoupling pulse of 74 μs length). Since there are no dominant ¹⁹ F background signals, the receive preamplifier could be set to the maximum available device value of 32K. The spectra were recorded with 32 scans and each lasted approximately 2 minutes. The waiting time between individual scans was s 3.2, is within the largely completed the enthalpic relaxation of ^{1 9} F spins (responsible for this is primarily the rapid relaxation at 600 MHz field strength due to the sometimes very large chemical shift of ¹⁹ F-core). Differences in signal intensities are primarily due to slightly different concentrations of the individual test substances in the mixture. Because of its very low solubility in water, flurbiprofen exhibits the most significant deviation in signal intensity. However, these differences in intensity are insignificant, since they are determined by the mixture stock solution and are therefore reproducible in the course of the screening.

To search for binding ligands among the test substances, the mixture stock solutions were added to the 1 mM aqueous solution of the target protein all-chymotrypsin until a fourfold substance was reached (ie 0.25mM substance concentration). This inverse concentration ratio (in which the target protein is present in excess) was chosen for the illustrative example described, since the substance bank was enriched with suspected or known ligand motifs, for the simultaneous detection of which sufficient amounts of protein should be present. The same basic experiment was then repeated under the same sample and measurement conditions as described above for recording the reference spectra.

The ^{1 9} F-screening spectra obtained are in 1a) - c ) underlay and clearly indicate the binding of several ligands in different strengths, mostly via a decrease in their reporter signal intensities. The spectra were referenced, as described in the text, for the same intensity of the most intense reporter signals. For 5-fluoro-1- (tetrahydro-2-furfuryl) uracil, only a shift in the frequency of the reporter signal is observed, but not its decrease in intensity. However, since a frequency shift due to signal averaging (between free and bound form of the substance) should also result in a decrease in intensity, no binding is assumed for this substance. The signal intensity is evaluated as the primary binding indicator and the signal frequency as the optional binding indicator. It is also striking that the known weak ligand 3-fluoro-d1-tyrosine is not indicated, but probably only because it is competitively displaced in the presence of the other, probably more binding ligands.

The screening with lysozyme as the target protein was repeated as a blank sample to verify the results for all-chymotrypsin (under identical sample and measurement conditions). All test substances were expected not to bind to this enzyme. The ¹⁹ F spectra confirm this ( 1a) - c) ). Any minimal changes in the intensity and frequency of the reporter signals compared to the target-free reference spectrum indicate the statistical fluctuation or accuracy of the binding indicators, which must be taken into account when interpreting the screening results (in the described case, the statistically irrelevant deviation of the signal intensity was approx 5-10%).

Table 1 Overview of the fluorinated test substances used along with their particular at 298K ^{1 9} F NMR frequencies (in ppm, measured in aqueous solution with 50 mM phosphate buffer, pH = 7.2, at 25 ° C). The numbers given are used for identification in the spectra shown. After analysis of the frequencies F ^{1 9,} the substances were collected in three submixing such that the displacement dispersion max. approx. 15 kHz (at 600 MHz field strength). The substances in bold were clearly identified as binding ligands of all-chymotrypsin based on a decrease in the intensity of their reporter signals.

1.a. ⁹ shows the ¹ F-spectrum of Submixture A without target protein (light gray, above), (gray, center) with lysozyme as a blank and mitαII-chymotrypsin as the target protein (black, below). The reporter signals are assigned to the test substances from Table 1 using the numbers given. The signal intensity of substance 2 (triflurpromazine) is greatly reduced due to its very low solubility. It becomes clear it nevertheless appears as a binding ligand for all-chymotrypsin, but not for lysozyme, primarily via a decrease in the intensity of its ¹⁹ F reporter signal (see arrow). The slight decrease in intensity of the reporter signal of substance 1 (by approx. -10%) is considered to be statistically irrelevant.

1.b represents the ^{1 9} F-spectrum of Submixture B without target protein (light gray, supra) with lysozyme as a blank (gray, center) and mitαII-chymotrypsin as the target protein (black, below). The reporter signals using given Numbers assigned to the test substances from Table 1. Substances 4, 5, 6 and 7 are indicated as binding ligands for all-chymotrypsin, but not for lysozyme, primarily via a significant decrease in the intensity of their ¹⁹ F reporter signal (see arrows).

The ^{1 9} F spectrum of submix C without target protein (light gray), with lysozyme as a blank (gray) as well as all-chymotrypsin as a target protein (black) is in 1.c shown. The reporter signals are assigned to the test substances from Table 1 using the numbers given. A slight shift in the reporter signal frequency is observed for substances 10 and 11, but only for substance 10 also a simultaneous halving of the signal intensity. Substance 10 is thus clearly indicated as a ligand for all-chymotrypsin, but not for lysozyme. Substance 11 is considered false positive because the primary binding indicator does not indicate signal intensity.

It has thus been shown that the method according to the invention is suitable, several in some cases strong binding ligands of non-binding substances that are in the same mixture, to distinguish. It follows that the invention Basic task of speeding up the screening process is also solved by the use of mixtures without the reliability the results thus obtained is impaired.

Claims (22)

Procedure for compiling a substance bank for the purpose the search for active substances by means of heteronuclear NMR spectroscopy, comprising the following steps: (i) selection of a number of substances, each containing one or more nuclei of an element or isotope (so-called isotope marker), which is not or only to a lesser extent naturally frequency occurs and the measurement is accessible by means of NMR spectroscopy; (Ii) Determination of the NMR resonance frequencies of the isotope markers in the chosen substances; (iii) grouping the selected substances into one or several subset (s), so that within each subset the signal frequency of each isotope marker at least twice the respective width of the signal half height (so-called line width) from the signal frequencies of all others Isotope markers differ within the subset and within each subset all signal frequencies of the isotope markers within fall within a spectral range of 15 kHz, with the proviso that the substances each subset does not react with each other; (iv) Manufacturing of one or more substance mixture (s), whereby for each Substance mixture all substances of the respective subset according to step (iii) are mixed. Method according to claim 1, wherein the substance mixtures in DMSO at a concentration of solved at least 0.05M each become. The method of claim 1 or 2, wherein the isotope marker ^{19 is} F. A method according to claim 1 or 2, wherein the isotope label is ^2H. The method of claim 1 or 2, wherein the isotope marker is ¹³ C. Substance bank, comprising a number of substances, all of which Substances each have one or more isotope labels of the same Have type and wherein the substances in one or more substance mixtures) are present in which the signal frequency is within each substance mixture of each isotope marker at least twice the line width from the signal frequencies of all other isotope markers within of the mixture of substances and within each mixture of substances all signal frequencies of the isotope markers within a spectral 15 kHz range, with the proviso that the substances of each mixture of substances do not react with each other. Substance bank according to claim 6, wherein the substance mixtures in DMSO at a concentration of solved at least 0.05M are. Substance bank according to claim 6 or 7, wherein the isotope ^{marker is 19} F. Substance bank according to claim 6 or 7, wherein the isotope marker is ² H. Substance bank according to claim 6 or 7, wherein the isotope marker ¹³ C is Methods for identifying or screening active substances, that bind to a drug target, include the following steps: (A) Recording a reference spectrum of a mixture of substances from the Substance bank according to one of claims 6 to 10, which is a filtered or edited according to the isotope label NMR spectrum, or an NMR spectrum in which the signals of the Isotope labeling (s) can be measured directly; (B) Manufacturing a mixture of the drug target and the substance mixture Step (A); (C) Recording a screening spectrum of the step (B) mixture prepared under the measurement conditions from step (A); (D) Identification of binding ligands by comparison of the screening spectrum with the reference spectrum based on changes in the NMR signals or of the NMR signal of the binding ligand. Method according to claim 11, wherein in steps (A) and (C) a simple one-dimensional NMR spectrum of the isotope marker is recorded. A method according to claim 11, wherein in steps (A) and (C) a one-dimensional NMR spectrum of the isotope _{marker is} recorded with a built-in T ₂ or T _1ρ relaxation _filter . Method according to claim 11, wherein in steps (A) and (C) a one-dimensional NMR spectrum of the Isotope markers added with a built-in diffusion filter becomes. Method according to claim 11, wherein in steps (A) and (C) a one-dimensional NMR spectrum of the Isotope marker with a built-in heteronuclear NOE mixing sequence for transmission of polarization of the protons on the isotope marker becomes. The method of claim 11, wherein in steps (A) and (C) a one-dimensional spectrum is recorded, in which a coherence transfer from isotope labeling to scalar coupling protons and detection is done on these protons. The method of claim 11, wherein in steps (A) and (C) a one-dimensional spectrum is recorded, in which a heteronuclear X → ¹ H magnetization transfer takes place from the isotope labeling to neighboring protons and the detection is carried out on these protons. A method for identifying or screening active substances which bind to a drug target, comprising the following steps: (A ') producing a mixture of the drug target and the substance mixture from the substance bank according to one of claims 6 to 10; (B ') Recording of a screening spectrum of the mixture from step (A'), which is an NMR spectrum in which heteronuclear magnetization transfer of protons to the isotope marker ( ¹ H, X-NOE) or from the isotope marker to protons (X, ¹ H-NOE); (C ') recording a reference spectrum of the mixture produced in step (A') under the measurement conditions from step (B), but the transmission frequency for protons being shifted by at least 50 kHz compared to the measurement from step (B '); (D ') Identification of binding ligands by comparing the screening spectrum with the reference spectrum based on changes in the NMR signals or the NMR signal of the binding ligand or ligands. A method according to claim 18, wherein only the polarization or saturation of protons of the drug target is selectively disturbed to distinguish between intra- and intermolecular ¹ H, X-NOE. The method of claim 18, wherein a two-dimensional heteronuclear ¹ H, X-NOESY is recorded to distinguish between intra- and intermolecular ¹ H, X-NOE. The method of claim 18, wherein a two-dimensional heteronuclear X, ¹ H-NOESY is recorded. Method according to one of claims 11 to 21, wherein binding ligands iden based on relative frequency and / or intensity changes of the substance mixtures between reference and screening spectrum be certified.