Advanced NMR Screening: Unveiling Bioactive Compounds in Complex Molecular Mixtures

Abstract

Nuclear magnetic resonance (NMR) spectroscopy represents an indispensable analytical tool for the structural and functional characterization of complex molecular mixtures, with significant applications in metabolomics, natural product discovery, and pharmaceutical analysis. A primary advantage of NMR is its capacity for the direct analysis of complex matrices, such as synthetic libraries and crude natural extracts, obviating the need for extensive chromatographic fractionation and thereby enabling the direct identification of bioactive constituents. This perspective provides a comprehensive review of ligand-observed NMR methodologies based on the principle that molecular binding is a prerequisite for biological function. Key techniques are systematically described, including Saturation Transfer Difference (STD), Transferred-NOE SpectroscopY (trNOESY), and methods based on relaxation rates (e.g., T ₂-relaxation filtering using CPMG pulse sequences). Further approaches include Diffusion Ordered SpectroscopY (DOSY), WaterLOGSY, and ¹⁹F NMR spectroscopy, a highly sensitive modality for screening fluorinated compound libraries. Collectively, these techniques facilitate the rapid identification of binding hits and provide critical insights into the structural and dynamic features of ligand–receptor interactions, including binding epitopes and bound-state conformations. The paper envisages future perspectives, emphasizing the potential of hyperpolarization methods to overcome the sensitivity limitations of benchtop instruments and the growing importance of in-cell and on-cell NMR applications for investigating molecular interactions within a native physiological context, which constitutes a significant frontier in drug discovery.

1. Introduction

Nuclear magnetic resonance (NMR) spectroscopy is a game changer for studying complex molecular mixtures and has been proven as an invaluable tool in fields like metabolomics, − natural product discovery, , and pharmaceutical analysis. , One of its key benefits is that NMR analysis is nondestructive; thus, samples can be recovered after analysis for future experiments or storage, which is crucial for rare or limited samples. NMR also delivers incredibly detailed information about a molecule’s structure, covering everything from atom connectivity and spatial arrangements to stereochemistry. By examining chemical shifts, coupling constants, and running multidimensional NMR experiments (like COSY, HSQC, HMBC, and NOESY), unknown compounds can be directly identified within a mixture. − Moreover, a major advantage of NMR is its inherently quantitative nature: the signal integral directly correlates to the number of nuclei, allowing for precise quantification of mixture components without needing extensive calibration curves or external standards (though internal standards are often used for absolute quantification). ,, This “standard-free” quantification is particularly beneficial when dealing with complex mixtures where pure standards for every component simply are not available. Furthermore, unlike many other analytical techniques that demand extensive chromatographic separation (such as gas chromatography–mass spectrometry (GC-MS) or liquid chromatography–mass spectrometry (LC-MS)), NMR can analyze complex mixtures directly in solution. − This dramatically reduces sample preparation time, minimizes the risk of sample degradation or loss during separation, and even enables the study of dynamic processes or unstable compounds.

These unique features establish NMR as an indispensable analytical methodology, not merely for the comprehensive characterization of complex molecular matrices but also for enabling the development of expeditious screening protocols targeting the identification of bioactive compounds. This broad applicability extends equally to both purpose-built synthetic libraries and inherently diverse natural product extracts.

Identifying ligands that bind macromolecular targets is a pivotal stage in drug discovery. Historically, researchers have uncovered such compounds by applying high-throughput screening (HTS) to extensive corporate compound libraries. , However, the substantial time and financial investment, coupled with a high incidence of false positives, has encouraged the search for alternative strategies. Advances in synthetic chemistry and screening technologies have directed discovery efforts toward compound and fragment libraries, offering optimized chemical structures tailored to improved target engagement and pharmacological properties. These collections represent additional sources of chemical diversity suitable for high-throughput or biophysical screening, and NMR spectroscopy has been proven to be a very useful tool for hit identification and lead optimization. Fragment-based drug discovery (FBDD) is a particularly compatible approach with NMR-based screening, due to its ability to detect weak (K _d: μM–mM) but specific interactions between low-molecular-weight fragments (≤300 Da) and biological targets. , During the past decade, FBDD has gained recognition as a powerful strategy, producing candidate compounds for a wide spectrum of targets ranging from nucleic acids to kinases, enzymes, membrane proteins, and even intrinsically disordered proteins.

At the same time, natural products, frequently occurring as complex molecular mixtures, are among the most valuable sources of bioactive compounds for drug discovery. This is largely because many of the organisms that produce them, such as plants, , fungi, and microbes, , are sessile and constantly exposed to environmental stressors, leading to the production of a wide range of secondary metabolites with defensive, signaling, and metabolic properties. , The long-term therapeutic use of natural products across cultures and centuries, along with their unique structural diversity, constitutes the foundation for their broad spectrum of bioactivities, ,,, to the point that about a quarter of all drugs approved by Food and Drug Administration (FDA) and European Medicines Agency (EMA) are plant-derived, either natural products or their chemically modified derivatives, , thus highlighting how nature has provided key chemical scaffolds that continue to inspire modern pharmacology. ,, The high molecular diversity and complexity, and evolutionary optimization for biological activity, make them excellent candidates for modulating complex disease pathways. , Nevertheless, the common way of identification of natural products relies on an extraction phase, followed by extensive fractionations and time-consuming structural elucidation steps, which can limit the identification of bioactive molecules. , The fractions have to be assessed for the specific biological activity of interest, and then the constituents further separated until the isolation of that or those responsible for the activity, which finally need to be characterized from the structural point of view. Even with significant advancements in analytical techniques and high-throughput screening methodologies, the inherent labor and time demands involved in this process make the identification of potentially active components in complex natural mixtures a persistent and formidable challenge in natural product-based drug discovery. ,

However, recent developments of NMR-based methodologies allowing the detection and characterization of ligand–receptor interactions paved the way for the extensive use of this spectroscopy for the rapid screening of synthetic libraries and complex natural product mixtures, including crude natural extracts, aimed at the identification of bioactive compounds. Such results can be obtained efficiently and reproducibly using straightforward NMR approaches reviewed in the following paragraphs, with a focus on rapidly and easily obtainable NMR parameters that can be observed to detect molecular binding.

2. NMR Methodologies for the Screening of Complex Mixtures

Several advanced NMR approaches have been developed and are based on the paradigm for which molecular binding is the conditio sine qua non for molecular activity; as a consequence, the detection of molecular interactions between a ligand and a specific target of interest can be exploited to guide the identification of bioactive compounds in the framework of several biomedical and biotechnological applications. Given a biomolecule that represents a relevant biomedical or pharmacological target, NMR can support the screening of complex compound mixtures to identify their putative ligands. To this aim, several NMR experiments based on the observation of ligand resonances can be employed. ,

2.1. Saturation Transfer Difference (STD)

STD NMR experiments allow the rapid identification of ligands of a target macromolecule, the ligand-binding epitope characterization, and the dissociation constant determination. − Each experiment requires the acquisition of two spectra: (a) the “on-resonance” spectrum, in which the target is selectively irradiated and this saturation is transferred to ligand via intramolecular ¹H–¹H cross-relaxation pathways at the ligand–receptor interface; the saturation of ligand’s resonances can be detected when it dissociates back in the free state and (b) the “off-resonance” spectrum, in which the target is not irradiated and a normal spectrum of the sample is acquired. The subtraction of the “on-resonance” from the “off-resonance” spectrum yields the STD spectrum, which displays only the signals of the ligand that experienced the saturation transfer from the target (Figure A).

1.

STD NMR experiment. (A) Principle of STD NMR. On-resonance spectrum: a train of selective radiofrequency (RF) pulses selectively saturates some receptor resonances. This saturation spreads throughout the protein via spin diffusion and is transferred to any bound ligands at the binding interface. Off-resonance spectrum (Control): an identical RF pulse train is applied at a frequency where neither the receptor nor the ligand has signals, yielding a reference spectrum. Difference spectrum (STD): the on-resonance spectrum is subtracted from the off-resonance spectrum. The resulting STD difference spectrum isolates signals belonging only to the ligand bound to the receptor, revealing its binding epitope. (B) Identifying ligands in a mixture with STD NMR experiment: an STD NMR experiment is acquired on a mixture of compounds in the presence of a target molecule. Interpreting the spectra. Off-resonance spectrum shows all compounds. STD spectrum: acts as a filter. Only signals from compounds that bind to the protein are visible. Outcome: the appearance of signals of mixture component(s) in the STD spectrum is a direct confirmation of binding, making it a powerful tool for hit discovery from a library.

STD experiments are only applicable to ligands in medium/fast exchange between the free and the bound states, which are characterized by a K _d ranging from 10 nM to 10 mM; in the case of stronger interactions, the saturation transfer is not efficient, and therefore, this interaction is not detectable. ,

Moreover, by analyzing a sample containing the target and a mixture of putative ligands at once, STD NMR allows the prompt discrimination between ligands and nonbinders in a fast experiment, yielding an STD spectrum in which only the resonances of the binding molecules are displayed (Figure B). The subsequent signal assignment will provide the identification of the target ligands, as well as the characterization of their binding epitopes. ,,

STD NMR has been applied to the screening of several complex mixtures. − For instance, we developed an STD NMR-based method to identify antiamyloidogenic compounds in natural extracts as a promising prophylactic strategy against Alzheimer’s disease (AD) and other neurodegenerative pathologies. As an example, we report the dissection of green and roasted coffee extracts. STD experiments were run with an “on-resonance” selective irradiation of the resonances of amyloid-β1–42 (Aβ1–42) oligomers, the most toxic intermediates of the amyloid aggregation pathway in AD, and led to the identification of three interacting compounds, i.e., chlorogenic acid and its regioisomers neochlorogenic and cryptochlorogenic acids, demonstrating the technique applicability to complex natural matrices, also with a high degree of structural similarity. Similarly, the extracts of four Humulus lupulus varieties and their polyphenolic-enriched fractions were characterized for their ability to counteract Aβ1–42 aggregation and prevent its neurotoxicity. The most active polyphenolic-enriched fraction was screened for its interaction with Aβ1–42 by STD NMR. Due to the large/complex chemical nature of these extracts, the severe signal overlapping did not allow one to unequivocally identify the ligands in the mixture, but compound classes such as chlorogenic acids and several flavonoids were proven to bind to Aβ oligomers.

Moreover, STD NMR can be combined with a wide range of pulse sequences, thus unlocking access to a whole set of experiments such as STD TOCSY (TOtal Correlation SpectroscopY), STD HSQC (Heteronuclear Single Quantum Coherence), and STD DOSY (Diffusion Ordered SpectroscopY) ,, to alleviate spectral overlap, even though these approaches are more time-consuming. For instance, STD TOCSY has been applied, in combination with other experiments, to evaluate the binding potential of the natural extract of Stryphnodendron polyphyllum toward human serum albumin (HSA). In this work, Tanoli and colleagues identified myricetin-3-O-rhamnopyranoside, quercetin-3-O-glucopyranoside, quercetin-3-O-xylopyranoside, and quercetin-3-O-rhamnopyranoside as HSA active site blockers. The combination of STD spectra with two-dimensional (2D) approaches has allowed the correct assignment of resonances, thus counteracting spectral overlap. In particular, protons located in the aromatic region and those of the glycone part showed the strongest cross-peaks in the STD TOCSY spectra, thus sustaining their involvement in the molecular binding to the HSA active site. This experiment, also combined with the other approaches, has also provided evidence of flattened conformations during the binding of the recognizable ligands to HSA. ,

Recent advances in STD NMR concern the use of strategies to reduce experimental time, such as reduced data set STD (rd-STD) NMR, which is based on the acquisition of only two experimental data points to determine the initial growth rate of the STD build-up curve. As a consequence, this approach allows the determination of K _D and ligand-binding epitopes in a much faster way compared with traditional STD NMR analysis. An additional recent innovation consists of imaging STD NMR, which has been described by Monaco and colleagues. This approach is based on the combination of STD NMR with chemical shift imaging (CSI), which allows the acquisition of spatially resolved NMR experiments along the z-axis of an NMR tube containing ligand concentration gradients against a homogeneous target. In this way, both K _D and ligand-binding epitopes can be determined and titrations performed in a single experiment, also minimizing biases deriving from sample manipulation. The application of such improvements to the screening of mixtures will be highly advantageous for accelerating hit identification in HTS activities.

2.2. Transferred-NOE SpectroscopY (trNOESY)

When ligand molecules bind to a specific target, the occurring intramolecular NOE effects (NOEs) are subject to major alterations, leading to the observation of transferred NOE effects (trNOEs). These phenomena can be observed by a trNOESY experiment, which is based on the observation that low-molecular-weight molecules are characterized by shorter correlation times and positive NOE signals, while large macromolecules, having longer correlation times, show negative NOEs. When molecular binding occurs, the small ligand adopts the NOE behavior of the large macromolecule, thus showing negative NOEs (Figure A). As a consequence, ligands can be promptly discriminated from nonbinding molecules, also present in a mixture, based on the shift in trNOEs sign when compared to NOEs (Figure B).

2.

trNOESY experiment. (A) Principle of trNOESY for detecting binding. Free ligand: A small molecule tumbles rapidly (τc is short) → positive NOE. Free receptor: a large macromolecule tumbles slowly (τc is long) → negative NOE. Bound ligand: tumbles slowly as part of the complex, adopting the receptor properties → negative “transferred” NOE. Conclusion: a switch in a ligand’s NOE sign from positive (gray) to negative (black) is a definitive signature of its binding to a macromolecular target. (B) Identifying ligands in a mixture with trNOESY. Experiment: a single 2D trNOESY spectrum is recorded on a mixture of compounds in the presence of a target. Visual binding assay: the sign (and color) of the NOE cross-peaks acts as a binding filter: positive cross-peak (gray): compound does not bind, remains small and free. Negative cross-peak (black): compound binds to the protein and adopts large-molecule behavior. Outcome: this technique allows for the simultaneous testing of multiple compounds, with the color of the cross-peaks providing an immediate readout of the ’hits’.

Since this technique has been widely used for hit confirmation and structure determination, it is quite common to see it coupled with STD experiments when analyzing natural extracts. An intriguing example is the work of Marcelo and colleagues, in which the crude acetone, butanol, and aqueous extracts of Salvia sclareoides were screened to search for an inhibitor of acetylcholinesterase (AChE), supporting the currently available therapy for the management of AD symptoms. While STD experiments confirmed that the only ligand of AChE is rosmarinic acid, the trNOESY spectra highlighted novel information about the still undetermined mode of binding of rosmarinic acid to the enzyme. The comparison of the NOE patterns suggested a dominant conformation in the bound state, whose strong cross-peaks are not present in the free form. This finding suggests that the flexibility between the allylic (H7′ and H8′) and the aromatic protons (H2′ and H6′) is reduced upon binding to AChE, confirmed both by the NOE contacts between H8′ and H6′/H2′ and H7′and H2′/H6′, and subsequently from docking calculations. In the same context, the butanol extract of S. sclareoides was screened to identify a compound able to interact with Aβ1–42 oligomers. The trNOESY spectrum of a mixture containing Aβ1–42 and the butanol extract of S. sclareoides offered a confirmation of the interaction previously seen in STD experiments and allowed us to identify rosmarinic acid as a ligand of Aβ1–42. A comparison between the NOESY spectrum of the free ligand and the trNOESY spectrum in the presence of Aβ1–42 oligomers revealed the appearance of novel cross-peaks, attributed to close contacts between olefinic protons and protons at both aromatic rings, suggesting a folded bound conformation of rosmarinic acid.

2.3. Relaxation Rates

Also, relaxation rates can be employed to monitor molecular binding. This approach is based on the difference in relaxation times between small molecules, which are characterized by small relaxation rates, and large molecules, which, on the other hand, are characterized by larger relaxation rates. As a consequence, when small ligands bind to the molecular target, their relaxation rates change significantly and become larger. A useful approach in NMR screening is T ₂-relaxation filtering, which can be obtained by employing the Carr–Purcell–Meiboom–Gill (CPMG) pulse train (Figure ).

3.

Principle of T ₂ -relaxation filtering with CPMG for ligand identification. The filter: the CPMG pulse sequence acts as a T ₂ filter, preserving signals from molecules that tumble rapidly (long T ₂) and suppressing signals from those that tumble slowly (short T ₂). Without target: the CPMG spectrum shows signals for all small-molecule compounds in the mixture, similarly to the corresponding ¹H spectrum (on the left). With target: ligands that bind to the macromolecular receptor adopt shorter T ₂ properties. Their signals are subsequently filtered out by the CPMG sequence (on the right). Outcome: by comparison of the two spectra, binding compounds are identified by the reduction/disappearance of their signals.

Although CPMG can suffer from nonspecific interactions, such as compounds adsorbing to macromolecules, running competition experiments helps mitigate these artifacts and prevents false positive hits. This approach was used by Shang and collaborators in association with complementary NMR experiments: in the relaxation-edited NMR spectra of mulberry extract in the presence of α-glucosidase, the intensities of some resonances in the spectrum were attenuated compared to the proton spectrum of the sole extract, suggesting the presence of an α-glucosidase ligand. Employing this approach, T ₂ relaxation changes were also used to indirectly measure the dissociation constant for ligand/target complexes, knowing the number of binding sites.

In a recent study, Huang and colleagues highlight how, in order to mitigate phase and multiplet distortions, the novel relaxation-weighted “perfect echo” Carr–Purcell–Meiboom–Gill (peCPMG) sequence offered higher sensitivity than other ligand-based techniques. Briefly, the analysis was conducted on extracts from 3 different plants, Sinomenii caulis, Celastrus orbiculatus, and Stephania tetrandra, traditionally used and known for their inhibitory potential against choline kinase α (ChoKα1), a hallmark of oncogenesis and tumor progression. A significant diminution in peak intensity was observed for one putative ligand in both S. caulis and C. orbiculatus extracts, while for S. tetrandra, the spectra revealed a wider range of active compounds, each marked by reduced peak intensities. This study also showcased how monitoring relaxation rates can identify high-affinity components, while other approaches, such as STD, can detect ligands with weak to moderate binding affinity with their target, as shown by running competition experiments in the presence of RSM-932A, a specific ChoKα1 inhibitor drug.

2.4. Diffusion Ordered SpectroscopY (DOSY)

In order to screen for binding activity, changes in the diffusion rate can also be measured. Diffusion coefficient measurement approaches are based on the study of the changes in the diffusion rate of a compound upon binding with a specific target; this approach is defined as “diffusion editing” or “diffusion filtering”. Although there is no high-molecular-weight limit for the target, diffusion editing is a valuable tool in the characterization of the molecular interactions between small- and intermediate-sized targets, requiring, moreover, a large amount of target. Nonetheless, diffusion editing has been applied to characterize complex compound mixtures, as well as molecular association processes. In particular, the difference in diffusion rate is exploited by Diffusion Ordered SpectroscopY (DOSY), which is a pseudo-bidimensional approach where the resonances of a simple ¹H spectrum are separated in the second dimension based on their diffusion coefficient. − In the screening of complex mixtures, molecular binding between ligand(s) and a target can be detected by the change in the diffusion rate of ligand(s) resonances, which in the bound state assumes the same diffusional behavior of the target (Figure ).

4.

Principle of DOSY screening for ligand identification. The method: DOSY is a pseudo-2D NMR experiment that separates molecules by their diffusion coefficient (D), which is dependent on molecular size (spectrum on the left). The states: nonbinders: remain small and diffuse quickly (high D); binders: associate with the large receptor, causing them to diffuse slowly (low D). The signature: in the 2D plot, the signals of binding ligands drop to a lower D value (spectrum on the right). This provides a direct visual filter for identifying interactions.

DOSY has been proven to be a valuable tool to extract information regarding intermolecular interactions and compound structure. − , For instance, Shang and collaborators combined DOSY and relaxation-edited NMR to identify ligands of α-glucosidase in mulberry leaf extracts. When α-glucosidase was added to the tested extract, some resonances underwent a change in diffusion rates and assumed a value similar to that of α-glucosidase. By employing 2D approaches to counteract spectral overlap, deoxynojirimycin (DNJ) was identified as an α-glucosidase and further characterized for its binding and biological activity. This approach allowed to successfully and directly screen a complex mixture for bioactive molecules in a fast and labor-saving manner, also providing the identification of minor components that could be lost during further fractionation. Moreover, DOSY can also be combined with an STD experiment, thus reducing spectral overlapping in the screening of ligand mixtures and identification of binding epitopes, as first demonstrated by Kramer and Kleinpeter. Another recent approach to reduce spectral overlapping in the analysis of the complex mixture by DOSY is represented by the application of pure-shift techniques; , by applying broadband homodecoupling, a simplified DOSY spectrum is obtained in which all of the multiplet signals are turned into singlets and can therefore be assigned unambiguously to the correct species, even though they share similar diffusion properties.

2.5. Water–Ligand Observed via Gradient SpectroscopY (WaterLOGSY)

WaterLOGSY − is based on the observation that water molecules are often found bridging the target–ligand interaction, thus forming a ternary complex. After selective irradiation of bulk water, followed by NOE mixing, the magnetization is therefore transferred to the ligand at the ligand–target interface, mediated by labile protons of the target, which can exchange with the solvent. As a result, if the ligand is bound to the target, the intermolecular water–ligand NOE becomes negative due to longer correlation time; on the other hand, the signals of nonbinding compounds are characterized by positive and less intense NOE effects (Figure A). By analyzing a sample containing a complex mixture and the target, this effect can be promptly observed in the NMR spectra as a change in the NOE sign of specific cross-peaks of the ligands (Figure B). The assignment of these resonances will allow both to identify ligand molecules and also their binding epitopes. This approach is characterized by high sensitivity due to the large number of exchangeable protons and bulk water molecules, and also allows the analysis in the presence of a low target concentration.

5.

WaterLOGSY experiment. (A) Principle of WaterLOGSY. Method: measures the Nuclear Overhauser Effect (NOE) transferred from solvent water to binders. Competing pathways: nonbinder: receives magnetization from bulk water → positive signal; binder: receives magnetization primarily via the receptor′s hydration shell → negative signal. Binding signature: the strong negative NOE pathway dominates for binders; therefore, a negative peak in the spectrum is a direct confirmation of the interaction with the target. (B) Identifying ligands in a mixture with WaterLOGSY. Experiment: a single WaterLOGSY spectrum is recorded on a mixture of compounds in the presence of the target. Interpreting the spectrum: the phase of each peak in the spectrum acts as a binding filter: positive signal → compound does not bind; negative signal → compound binds to the target. Outcome: the experiment is a one-shot binding assay, allowing for the rapid visual identification of “ligands” directly from a complex mixture.

WaterLOGSY is very effective in the NMR-based screening of mixtures characterized by medium complexity. ,, For instance, Nikolaev and colleagues selected four well-characterized bacterial and mammalian proteins (AroG, Eno, PfkA, and bovine serum albumin) and identified ligands in complex mixtures of up to 33 metabolites by comparing different ligand-detected NMR methods, such as WaterLOGSY and T ₁ρ.

Ligands were identified with similar correlation when screening a 7-metabolite mixture against the four selected proteins; when increasing the mixture complexity up to 15 and 33 metabolites, they found that most of the previously identified interactions were unambiguously recovered for both ligand-detected NMR methods, thus proving the applicability of these approaches in the screening of bioactive metabolite mixtures and in the mapping of their interactions with the target.

2.6. Ligand-Observed ¹⁹F NMR

When the screening of synthetic compound libraries is the main goal, another very useful approach is the ¹⁹F-based ligand screening. Fluorine can act as a bioisostere, particularly for the hydroxyl group and hydrogen. As a bioisostere, fluorine can be strategically used in drug design to enhance biological activity, potency, and selectivity. Furthermore, this nuclide is present in trace amounts in natural compounds, thus granting no signal background when working with ¹⁹F-labeled synthetic molecules. It is characterized also by high sensitivity in NMR analysis, as well as by a wide chemical shift range, which ensures a much higher signal resolution than ¹H NMR as well as a hyper-responsiveness to changes in chemical environment. Moreover, since changes in ligand ¹⁹F nuclei relaxation times after target binding are typically far greater than changes in their ¹H relaxation times, this system can be employed to screen for molecular binding between small fluorinated ligands and specific hotspots of the target. , Most commonly, T ₂-relaxation filters, such as CPMG pulse sequence, are employed in direct binding fluorine chemical shift anisotropy and exchange for screening (FAXS) experiments, as well as in competitive inhibition FAXS (Figure ). −

6.

Principle of FAXS NMR screening. Direct format (left panel). Binder (S): signal broadens upon addition of the receptor. Nonbinder (C): signal is unaffected. Outcome: changes in the chemical shaft and/or broadening of the corresponding signals allow the identification of receptor binder(s). Competition format (right panel). Setup: molecule S is used as a spy/reporter, showing a broad signal while being bound to the receptor; a competitive ligand is added to the mixture. Outcome: the spy molecule is displaced, and its broad signal sharpens and returns to its original position, signaling a “hit”.

For competitive assays, an appropriate spy molecule can be selected by two main strategies: (1) screening a library of fluorinated molecules against the target or (2) fluorinating an already known ligand. An ideal spy molecule will exhibit a notable decrease in signal upon the application of a CPMG pulse sequence and the introduction of protein. This signal reduction establishes a wide detection window, making it easier to observe displacement by a competitive molecule.

The direct experiment can be easily applied to the screening of mixtures with a setup similar to that adopted for ¹H relaxation-based experiments (Figure ). For example, ¹⁹F-based methodologies have been extensively exploited to characterize libraries of fluorinated carbohydrates as lectin ligands. For example, Martinez and co-workers studied the binding of a library of synthetic fluorinated monosaccharides to DC-SIGN, a lectin playing a pivotal role in various viral infections, including HIV and Ebola, through the recognition of self- and nonself glycans. The developed strategy allowed to screen a compound mixture and, at the same time, obtain crucial information on the specific sugar–protein interactions. The same authors synthesized a library of fluorinated glucose, mannose, and galactose derived by systematically exchanging every hydroxyl group with a fluorine atom, with which they developed a strategy combining chemical mapping and ¹⁹F NMR T ₂ filtering-based screening. They screened the library against three different carbohydrate receptors, i.e., the human macrophage galactose-type lectin, a plant lectin, Pisum sativum agglutinin, and the bacterial Gal-/Glc-binding protein from Escherichia coli, simultaneously defining their monosaccharide selectivity and identifying the hydroxyl groups essential for interaction. A competitive displacement assay, based on ¹⁹F T ₂-relaxation NMR and on the design, synthesis, and use of a strategic spy molecule, was also employed to assess and quantify sialoside ligand binding to Siglecs. The innovative approach developed by Atxabal et al. enables the precise quantification of binding for a diverse range of compounds in solution, specifically natural and modified sialosides, multivalent sialosides, and sialylated glycoproteins. Its utility extends to interactions exhibiting binding affinities that vary by more than 2 orders of magnitude.

Finally, the existence of the ¹⁹F-STD experiment is worthy of mention. It combines the resolution and specificity of ¹⁹F-detected NMR spectroscopy with the sensitivity of the STD technique. , 2D-STD-TOCSYreF experiments are also described in the literature; this new approach has been employed for the comprehensive epitope mapping of fluorinated glycans by combining the spectral resolution of ¹⁹F with the wide spatial coverage of ¹H. In this work, the 2deoxy-2-fluoro derivative of the N-glycan core di-(2-F-Man₂) and trimannoside (2-F-Man₃) was synthesized and characterized for the recognition of P. sativum agglutinin (PSA) by either of the two terminal mannose residues. In fact, crystallographic data suggested two different interaction modes between the trimannoside and PSA; therefore, the authors decided to apply 2D-STD-TOCSYreF to characterize this process in solution and managed to identify the binding epitope and decode this complex recognition process, overcoming the limitations of other previous approaches (i.e., 1D STDreF).

2.7. Possible Limitations of Ligand-Based Technique

Ligand-based NMR is based on powerful techniques for identifying which molecules in a complex mixture bind to a biological target. Their effectiveness, however, is confined to a specific “sweet spot” defined by the interplay between a ligand’s binding affinity and its exchange kinetics.

The core principle relies on detecting a “memory” of the binding event. For this to work, a ligand must bind and unbind rapidly (fast-to-intermediate exchange), allowing the binding information to be observed in the larger pool of free molecules. This makes the techniques ideal for detecting the weak-to-moderate affinity interactions (μM to mM range, sporadically nM range) common in the early stage of drug discovery. This operational window is a double-edged sword, creating two primary failure modes: false positives (deceptive signals) and false negatives (hidden binders). False positives occur when nonspecific events mimic a true binding interaction. The most common cause is compound aggregation, where small molecules clump together and behave like a large receptor, generating a misleading signal; other sources include nonspecific surface binding or technical artifacts. False negatives happen when a genuine binding event goes undetected. Paradoxically, very high affinity ligands are often the culprits. They bind so tightly and dissociate so slowly (K _off) that the binding signal cannot be effectively transferred or observed, rendering them invisible. Similarly, extremely weak binders may not occupy the target enough to produce a detectable signal.

To overcome these challenges, a robust experimental design is crucial. This process includes using proper buffers to prevent aggregation and employing clever strategies like competition-based assays, which use a known weak binder as a “spy” to reveal the presence of otherwise hidden high-affinity compounds. Ultimately, while potent, ligand-based NMR requires a critical understanding of these limitations to ensure the accurate and efficient discovery of novel molecular interactions. In this scenario, the combination of more ligand-based approaches is fundamental to thoroughly characterize these complex systems.

3. Outlook and Perspectives

The NMR methodologies described here enable the screening of even very large compound libraries for ligands of biomedically relevant receptors in very short times. In fact, aside from 2D-STD, trNOESY, or DOSY, which involve 2D or pseudo-2D spectrum acquisition and can sometimes take many hours, the other experiments can be run quite quickly, typically from tens of minutes to just 1–2 h.

Moreover, when combined with activity assays, essential for assessing the biological activity of any mixture components, these approaches can identify the bioactive compound(s) in the mixture and precisely determine their chemical structure. This is accomplished by linking the observed evidence of binding to the target with the measurement of specific NMR parameters and correlations that are crucial for structural elucidation. The importance of this aspect is paramount in analyzing mixtures with predominantly unknown constituents. STD NMR is particularly powerful here, as it can “filter out” the resonances of noninteracting components, bringing to the forefront the signals of ligands, even when they constitute a minor part of the mixture and their resonance are masked in the ¹H NMR spectrum. This drastically resolves spectral complexity by spectroscopically segmenting ligand resonances from other molecules, allowing for their detection even at minute percentage quantities, all without requiring physical separation.

With such a robust foundation, what does the future hold for NMR-based screening?

In terms of instrumentation, a significant challenge is the reliance on high-field NMR spectrometers, as the sensitivity required for screening is currently beyond the reach of low-field benchtop instruments. Hyperpolarization methods offer a promising path to overcome this limitation, potentially enabling simpler, faster, and more cost-effective analyses. Although its application to complex mixtures is still an emerging area, this approach holds considerable potential as a future screening tool. Cala and co-workers are already underway to adapt this method for larger ligand libraries and competitive studies.

From the methodological viewpoint, it is crucial to note that some of these techniques are also well-suited for studying molecular recognition within cells (in-cell NMR) or on their surfaces (on-cell NMR). , In-cell and on-cell methods provide knowledge about the binding efficacy directly in the native cellular environment. This is absolutely key, given that a large number of the most important drug targets either are located within the cell or are essential components of its plasma membrane.

Studying drug–target interactions within living cells is notoriously challenging with standard ¹H NMR; the target often remains undetected, and cellular noise obscures the ligand observation. Luchinat and co-workers developed a real-time in-cell ¹⁹F NMR approach based on the use of fluorinated ligands that allows to directly observe their binding to intracellular targets and even measure binding affinities in human cells via competition binding with a fluorinated reference compound. In particular, they applied this assay to the binding of a set of compounds toward the N-terminal ATP-binding domain of the human stress-inducible 90 kDa heat shock protein α (Hsp90α), a homodimeric molecular chaperone that binds and folds other proteins into their functional 3-dimensional structures. However, this approach, requiring ¹⁹F-labeling, can only be applied to synthetic molecules.

Nevertheless, STD NMR experiments, although based on the observation of proton resonances, have been successfully applied to on-cell studies involving different membrane receptors ,− and also to intracellular binding studies. However, to the best of our knowledge, this approach has not yet been used for simultaneous screening of potential ligand libraries. Given that the STD experiment can not only detect ligands but also allow for their identification without molecular labeling (making it perfect for natural compounds too), we believe this represents one of the most significant and desirable advancements in this field.

The authors declare no competing financial interest.