WO1997014814A1

WO1997014814A1 - A binary coding method for use in combinatorial chemistry

Info

Publication number: WO1997014814A1
Application number: PCT/US1996/016718
Authority: WO
Inventors: Ravi S. Garigipati; Susanta K. Sarkar
Original assignee: Smithkline Beecham Corporation
Priority date: 1995-10-19
Filing date: 1996-10-18
Publication date: 1997-04-24
Also published as: EP0856067A1; EP0856067A4; JP2000500565A

Abstract

The present invention is to a technique for labeling a ligand or compounds covalently bound to solid-phase-synthesis resin with a tag containing 13C and/or 15N. This tagging along with high-resolution isotope filtered 1H NMR, and Magic Angle Spinning probes provides for a technique which is general, nondestructive, and leaves the resin-bound sample intact and available for further chemical transformations. The limit of detection (< 800 pmol) in this system is sufficient to allow NMR to be used in deciphering the contents of a 'one-bead, one-compound' combinatorial chemistry library.

Description

A Binary Coding Method For Use in Combinatorial Chemistry

FIELD OF THE INVENTION

Complex combinatorial libraries require a technique for identification of the contents of the library which has been synthesized. A technique which is non¬ destructive and consistently reliable is the purpose of the present invention.

BACKGROUND OF THE INVENTION

In building a complex combinatorial library, one needs an efficient way to decipher the contents of the library. One needs to develop a molecular tag that is stable to all the reaction conditions required to build the library and the conditions to cleave the individual components of the library.

Binary coding has been established as a viable technique in encoding complex combinatorial libraries (Ohlmeyer, et. al., J. Am. Chem. Soc, 1995, 117, 5588). Even though this tagging technique is versatile, the tags have to be released from the bead before they can be identified and the ligand history deciphered.

By first requiring the release of the tag from the bead before its identification places many constraints on the techniques used and previously available. A need, therefore, exists for methods to identify the ligands of interest prior to their release from the bead.

SUMMARY OF THE INVENTION

The present invention provides for a technique which is an efficent method to decipher the contents of a combinatorial library. This technique is utilized for the selective identification of a chemical compound still bound to a single bead of a solid-phase synthesis resin. The tag, ^C and/or ^$N is stable to all the reaction conditions imposed upon it in building the library, and is also stable to the conditions to cleave the ligand of interest (i.e. the individual components) of the library. This technique therefore, provides a means whereby the identity of a ligand of interest can be decrypted either before or after it's release. The detection of the labeled substrate is preferably by (multidimensional) isotope-filtered ^H NMR (HMQC).

BRIEF DESCRIPTION OF THE DRAWINGS FIGURE 1 shows a ^C labeled substrate bound to a Wang resin.

FIGURE 2 (a) shows a 500-Mhz ¹³C-filtered H NMR spectrum of the compound in figure 1 coupled to the resing, obtained on a single bead using the decoulper coil of a standard commercially-available heteronuclear Nano-NMR probe. FIGURE 2(b) shows the corresponding unfiltered ^H NMR spectrum from Figure 2(a).

FIGURE 3 shows the contour plot of the 2 dimensional IH 13C HMQC spectrum obtained on the same single bead sample as those above.

DETAILED DESCRIPTION OF THE INVENTION An NMR technique has been developed which can generate high resolution

1 H-NMR spectra of compounds covalently bound to a solid phase synthesis (SPS) resins. The present invention provides for the novel application of this high resolution NMR technique to decipher the contents of a combinatorial library, i.e. to detect material bound to a single (lOOum) bead of resin. In this method, ^l3C and/or ^N labeled compound is selectively detected using isotope-filtered *H NMR (HMQC)(Bax, et al., S. J. Magn. Reson. 1986, 67, 565-569) in a two-coil high- resolution magic-angle spinning (MAS) probe (Barbara, T. J. Magn. Reson., Ser. A. 109, 265-269 (1994)). This technique suppresses signals from the unlabeled portions of the resin bead, thus allowing an unambiguous identification of the bound compound.

Currently, samples are cleaved from the resins, either chemically or photolytically (Lloyd-Williams, et al., Tetrahedron, 1993, 49(48), 11065-1 1133), and are analyzed by conventional structure determination techniques. However, there is considerable interest in developing analytical techniques that can analyze samples while they are still bound to the polymer matrix. This is one such technique. The fact that this technique is non-destructive is also particularly useful in the rapidly expanding field of "one bead, one-compound" combinatorial chemistry. (Lam, et al., J. Nature 1991, 354, 82-84; Lebl, et al., Biopolymers 1995, 37, 177-198). In these experiments, a reaction mixture containing a diverse library of possible millions of individual beads (representing millions of different structures) may be tested. From this mixure, only one bead may be isolated which has the desired activity, and the success of this approach critically depends upon determination of the structure of the approximately 0.1-lnmol of sample generally bound to that single bead. (Madden et al., Prespect. Drug. Discovery Des., 2, pp. 269-285 (1994)). Solid phase-synthesis (SPS) techniques (Merrifield, R. B., J. Am. Chem. Soc.

1963, 85, 2149-2154), combined with the growing field of combinatorial chemistry (Gallop, et al., J. Med. Chem. 1994, 37, 1233-1251 ; Gordon, et al., J. Med. Chem. 1994, 37, 1385-1401) are currently have a very large impact upon drug discovery and chemistry in general. Unfortunately, the analytical techniques for SPS samples are not as highly developed as the synthetic techniques.

In classic peptide combinatorial chemistry, for instance, the peptide would be cleaved from the bead and sequenced by either Edman degradation or mass spectronomy. This strategy fails for non-peptide samples which are not built up from simple repeating units, so "encoding methods" have been developed (Brenner et al., Proc. Natl. Acad. Sci. USA 1992, 89, 5381-5383). To encode a resin library, an additional "tag" compound is added to each bead to document its synthetic history, creating a unique "bar code" (Chabala, et al., Perspect. Drug Discovery Des. 1994, 2, 305-318) which can be cleaved and indentified (by GC or GC MS) to indirectly indentify the primary sample. Examples of reporter tags that have been used include nucleotides (which were cleaved, amplified by PCR, and sequenced), (see Brenner et al., supra; and Needels et al., M. A. Proc. Natl. Acad. Sci. USA 1993, 90, 10700-10704) peptides (sequenced by Edman degradation) (Kerr et al., J. Am. Chem. Soc. 1993, 775, 2529-2531), or a combined series of hydrocarbon homologues and polychlorinated aromatics (analyzed by GC) (Ohlmeyer, et al., Proc. Natl. Acac Sci. USA 1993, 90, 10922-10926; Borchardt, et al., J. Am. Chem. Soc. 1994, 777, 373-374; and Baldwin, et al., J. Am. Chem. Soc. 1995, 777, 5588- 5589). In all cases, the use of reporter tags complicates synthetic strategies, increases the risk of side reaction and byproducts, and yields only indirect evidence of structure. The presence of the encoding structure may also affect the results of the binding assays. While new MS techniques have recently been developed that can analyze single-bead samples of peptides (Youngquist, et al., J. Am. Chem. Soc. 1995, 777, 3900-3906; Egner, et al., J. Org. Chem. 1995, 60, 2652-2653; Brummel et al., Science 1994, 264, 399-402) and peptoids (Zambias, et al., Tetrahedron Lett. 1994, 35, 4283-4286) without the use of encoding ligands, all such methods are currently destructive. Although NMR spectroscopy is routinely used to analyze samples that have been cleaved from SPS resins, the NMR spectra of resin-bound samples consist of very broad lines, even if the resin is swollen by the addition of solvents. While relatively low-resolution *H and ^C NMR spectra have been reported, (See Giralt, E et al., Tetrahedron 1984, 40, 4141-4152; Albericio, et al., J. Org. Chem. 1989, 54, 360-366; and Mazure et al., C. R. Acad. Sci., Ser. 2 1986, 303, 553-556) ¹³C NMR has proven more useful because its larger spectral dispersion at least partially compensates for the broad linewidths. The incoφoration of site-specific ^³C labels in resin-bound samples has been shown to facilitate the use of ^l3C NMR to monitor the progress of synthetic SPS reactions (Look et al., J. Org. Chem. 1994, 59, 7588- 7590).

This invention is based upon a technique of coding either suitable SPS components or the compounds themselves with ^{, 3}C/¹⁵N labels. Preferably to reduce the amount of labels needed for combinatorial work the coding method is binary. It is recognized however, that individual uniquely labeled beads could be detected and identified for a combinatorial library. The detection of the labelled substrate is preferably by (multi-dimensional) isotope-filtered *H NMR (HMQC) with Magic Angle Spinning (MAS). This combination is now known as Nano NMR probe technology. This technique utilizes the ability to selectively observe a proton attached to a ¹³C/¹⁵N label by inverse detection (Muller, L. J. Am. Chem. Soc, 1979, 707 , 4481 ; Bax et al., J. Mag. Reson. , 1986, 67, 565). T₂, T₃ etc.

Ar* has a unique ¹³C / ¹⁵N label

Tags can be identified and ligand identity deciphered before or after the bioassay ligand

tags T,, T₂ and T₃ etc., coding for individual steps is incoφorated into the solid support

Indirect detection of protons attached to carbons and/or nitrogens labeled with ¹³C- and/or ¹⁵N-, by isotope filtered NMR, allows one to detect the proton resonances selectively by suppressing all the signals from the protons attached to unlabeled carbons and/or nitrogens (Bax, et al., S. J. Magn. Reson. 1986, 67, 565- 569; Bax, et al., j. Magn. Reson., 1983, 55, 301-315,). This technique is extensively applied for structure determination of biological macromolecules like proteins and nucleic acids. (G.M.Clore and A. M. Gonenborn, NMR of Proteins, CRC Press, Ann Arbor, 1993). The present invention is the novel application of this theory to use in combinatorial chemistry applications, and even more so its novel use on a single bead.

The amount of material attached to a single bead in combinatorial chemistry libraries are of the order of about 0.1 to 1 nmole. It is a significant challenge to detect such a small amount of material in the presence of large background signals from the resin, impurities and from the solvent. In order to circumvent these problems ^³C and/or ^N-labels are attached as a tag in combinatorial libraries and then using suitable NMR techniques, such as isotope-filtered H NMR (HMQC) under magic angle spinning conditions (MAS) (Barbara et al., J. Magn. Reson., Ser. A, 1994, 109, 265-269; and Garroway, A., J. Magn. Reson., 1982, 49, 168-171) to selectively detect the ^C- and/or ^\ - labeled compound. It has been found that this technique suppresses signals from the unlabeled portions of the resin bead, thus allowing the heretofor unambiguous identification of the bound compound.

As shown in the following scheme (Scheme II- a typical library member in a three step library is shown) the tags can be incoφorated along with the ongoing library synthesis. The tags can be used as a binary code. It is also another aspect of the present invention that the tagging label can be incoφorated into the ligand itself. In the example shown, tags A, B and C will have unique ^!H and ¹³C chemical shifts (Spectroscopic methods in Organic Chemistry, Wiilams & Fleming, McGraw- Hill, 1966) which can be easily monitored using the proposed isotope filtered technique. As noted, this technique can also be used when the label (¹³C / ¹⁵N ) is incoφorated into the ligand itself.

This experiment can be conducted on a 2 channel NMR spectrometer, with inverse detection and magic angle spinning capability, operating at a proton frequency of 300 -750 MHz. A dual probe (IH observe, ¹³C,¹⁵N decoupling) with the capability of magic angle spinning is required for this puφose.

OH

SCHEME II Another aspect of the present invention is the ability to utilize either a number of different 13 C/ ^N labels in a sample, (i.e., to encode the bead in the conventional binary way) or to utilize the site of the Cl N labels to encode additional information. The encoding would occur without adding any additional, and potentially disruptive chemical structures to the resin, without designing any complicated mutually compatible parallel synthetic methodologies, and since the detection would occur nondestructively this allows the sample to be left intact and still bound to the resin. An added feature of this novel technique would then allow the sample to be subject to further chemical transformation or analysis, if so desired. This technique allows for synthesis of uniformly labeled molecules and their subsequent analysis by multidimensional isotope filtered NMR experiments. While demonstrated herein on approximately 800pmol of material in the Experiments, this technique will allow for sensitivity detections of as little as lOOpmol of sample. Additionally it is recognized that while the 1H decoupler coil of the Nano NMR probe was used for these experiments, other more efficient inverse-geometry probe designs may allow for the necessary sensitivity to detect reduced amounts of material bound to single beads.

EXPERIMENTAL SECTION The Η NMR spectrum of such a small amount of material is typically complicated by large signals arising from solvent backgrounds, impurities, fingeφrints on the outside of the sample cell, as well as smaller peaks due to the polystyrene backbone of the bead itself. To circumvent these problems, we synthesized [3,5-dimethoxy-'³C]benzoic acid (compound 1, Figure 1), coupled it to 100 μm diameter Wang resin beads, and utilized isotope-filtered NMR to selectively observe the proton resonances of the [3,5-dimethoxy-^l3C] group. A ID (1 dimensional) ^I3C-filtered Η NMR spectrum of compound 1 is shown in Figure 2(a). The protons of the [3,5-dimethoxy-¹³C] moiety (at 3.7 ppm) and the ¹³C satellites of the solvent (at 5.32 ppm) are clearly observable. The corresponding unfiltered spectrum is shown in Figure 2(b), and it is clear from the spectrum that assignment of the proton resonances is difficult due to the presence of large background signals. These signals, as well as the non-¹³C-labeled peaks from the polystyrene resin itself, are completely suppressed in the ¹³C-filtered spectrum. Since complete structural characterization of chemical compounds relies on through-bond and through-space proton-proton or proton-carbon connectivities, which are generally obtained from using 2D (2 dimensional) NMR data, data is shown from a 2D version of 1H-1 C HMQC in Figure 3. Again, correlations arising from the protons of the [3,5- dimethoxy-¹³C] group and the ¹³C satellites of the solvent are clearly visible. These data verify that the sensitivity and selectivity are sufficient to observe any unique ¹³C labeled substrate still bound to a single resin bead using ID NMR, and that 2D NMR was feasible if mixture analysis is/was required.

In greater detail Figure 2a shows a 500- MHz 13c-filtered iH NMR spectrum of compound 1 coupled to Wang resin, which was obtained on a single bead using the decoupler coil of a standard commercially-available Heteronuclear Nano«NMR™ probe. The single bead was suspended in 30 μL of dichloromethane-^ in a standard (40 μL) Nanoprobe cell and spun at the magic angle (54.7°) at ca. 2 KHz. This proton spectrum was acquired using a one-dimensional version of the HMQC sequence --j-- without 13c decoupling but with presaturation ( 11 Hz field strength for 1 s) of the residual protonated dichloromethane resonance at 5.32 ppm. Total experiment time was 170 min (5000 transients, 1.02 s acquisition time, 1.0 s presaturation delay). Figure 2b shows the iH NMR spectrum obtained using a one- pulse experiment. Total experiment time was 250 min (5000 transients, 2.05 s acquisition time, 1.0 s presaturation delay). Both spectra were processed with 4 Hz of exponential weighting (the dichloromethane c satellites have 2 Hz linewidths while the less mobile [3,5-dimethoxy-13c] resonances have 20 Hz linewidths). As noted above Figure 3 demonstrates a contour plot of the 2D 1H- 3C

HMQC spectrum obtained on the same single-bead sample. An f i cross section at the f2 frequency of the 13CH30- protons is also shown, indicating both the high sensitivity and the excellent suppression of the protons attached to l^C. The 17.5 hr experiment used broadband (3.5 kHz) 3c GARP-1 decoupling3 during the 128 msec acquisition time, a 1.13 sec presaturation delay, 512 scans for each of the 49 hypercomplex ti-datapoints, and 5 kHz and 8 kHz spectral widths, and cosine and 10 Hz exponential weightings, in ti and -2, respectively.

Three different single-bead samples were tested to verify the reproducibility of this technique. As noted above it was also found that ID 13c NMR spectra could be obtained on the 13c-labeled single-bead samples used in this study. The

Nanoprobe sample cell restricts 100% of the sample (40 μL maximum) within the active region of the receiver coil to ensure that even a single bead would always be detected.

Compound 1 was coupled to 100-200 mesh Merrifield resin obtained from Calbiochem was functionalized according tothe procedure noted in Wang, S. S. J. Am. Chem. Soc. 1973, 95, 1328-1333. The beads selected for NMR were measured by a light microscope to have a 100 μm diameter.

The disclosures of patents, patent applications and publications cited herein are all incoφorated by reference in their entireties.

The above description fully discloses the invention including preferred embodiments thereof. Modifications and improvements of the embodiments specifically disclosed herein are within the scope of the following claims. Without further elaboration, it is believed that one skilled in the are can, using the preceding description, utilize the present invention to its fullest extent. Therefore the Examples herein are to be construed as merely illustrative and not a limitation of the scope of the present invention in any way.

Claims

What is claimed is:

1. A method for deciphering the contents of a combinatorial library which method comprises: a) tagging compounds contained in a combinatorial library with coded 13c and/or l^N labels; b) detecting the labels by NMR.

2. The method according to Claim 1 wherein the coding is binary coding.

3. The method according to Claim 1 or 2 wherein the NMR technique is isotope-filtered 1H NMR.

4. The method according to Claim 3 wherein the NMR technique is used with Magic Angle Spinning.

5. The method according to Claim 1 wherein the NMR technique uses inverse-geometry probe design.

6. The method according to any of Claims 1 to 3 wherein the compounds are bound to a polymer matrix.

7. The method according to Claim 1 or 2 wherein the compounds and/or resin are labeled with different 13c labels in a sample.

8. The method according to Claim 1 or 2 wherein the compounds and/ or resin are labeled with different l^N labels in a sample.

9. The method according to Claim 1 or 2 wherein the compounds compounds and/ or resin are labeled with different 1 c /l^N labels in a sample.

10. The method according to any of Claims 1, 2, 7, 8, or 9 wherein the coding is by a ratio of c to *N labels.

11. The method according to Claim 1 wherein the compound is attached to a single bead in a combinatorial chemistry library.

12. The method according to Claim 11 wherein the bead is solid or swollen.

13. The method according to Claim 11 wherein the bead is in solution.