Machine Learning-Enhanced Quantum Chemistry-Assisted Refinement of the Active Site Structure of Metalloproteins

Understanding the fine structural details of inhibitor binding at the active site of metalloenzymes can have a profound impact on the rational drug design targeted to this broad class of biomolecules. Structural techniques such as NMR, cryo-EM, and X-ray crystallography can provide bond lengths and angles, but the uncertainties in these measurements can be as large as the range of values that have been observed for these quantities in all the published structures. This uncertainty is far too large to allow for reliable calculations at the quantum chemical (QC) levels for developing precise structure-activity relationships or for improving the energetic considerations in protein-inhibitor studies. Therefore, the need arises to rely upon computational methods to refine the active site structures well beyond the resolution obtained with routine application of structural methods. In a recent paper, we have shown that it is possible to refine the active site of cobalt(II)-substituted MMP12, a metalloprotein that is a relevant drug target, by matching to the experimental pseudocontact shifts (PCS) those calculated using multireference ab initio QC methods. The computational cost of this methodology becomes a significant bottleneck when the starting structure is not sufficiently close to the final one, which is often the case with biomolecular structures. To tackle this problem, we have developed an approach based on a neural network (NN) and a support vector regression (SVR) and applied it to the refinement of the active site structure of oxalate-inhibited human carbonic anhydrase 2 (hCAII), another prototypical metalloprotein target. The refined structure gives a remarkably good agreement between the QC-calculated and the experimental PCS. This study not only contributes to the knowledge of CAII but also demonstrates the utility of combining machine learning (ML) algorithms with QC calculations, offering a promising avenue for investigating other drug targets and complex biological systems in general.

Novel Nanogels Loaded with Mn(II) Chelates as Effective and Biologically Stable MRI Probes

Here it is described nanogels (NG) based on a chitosan matrix, which are covalently stabilized by a bisamide derivative of Mn-t-CDTA (t-CDTA = trans-1,2-diaminocyclohexane-N,N,N',N'-tetraacetic acid). the Mn(II) complex acts both as a contrast medium and as a cross-linking agent. These nanogels are proposed as an alternative to the less stable paramagnetic nanogels obtained by electrostatic interactions between the polymeric matrix and paramagnetic Gd(III) chelates. The present novel nanogels show: i) relaxivity values seven times higher than that of typical monohydrated Mn(II) chelates at the clinical fields, thanks to the combination of a restricted mobility of the complex with a fast exchange of the metal-bound water molecule; ii) high stability of the formulation over time at pH 5 and under physiological conditions, thus excluding metal leaking or particles aggregation; iii) good extravasation and accumulation, with a maximum contrast achieved at 24 h post-injection in mice bearing subcutaneous breast cancer tumor; iv) high T1 contrast (1 T) in the tumor 24 h post-injection. These improved properties pave the way for the use of these paramagnetic nanogels as promising magnetic resonance imaging (MRI) probes for in vitro and in vivo preclinical applications.

Combining Solid-State NMR with Structural and Biophysical Techniques to Design Challenging Protein-Drug Conjugates

Several protein-drug conjugates are currently being used in cancer therapy. These conjugates rely on cytotoxic organic compounds that are covalently attached to the carrier proteins or that interact with them via non-covalent interactions. Human transthyretin (TTR), a physiological protein, has already been identified as a possible carrier protein for the delivery of cytotoxic drugs. Here we show the structure-guided development of a new stable cytotoxic molecule based on a known strong binder of TTR and a well-established anticancer drug. This example is used to demonstrate the importance of the integration of multiple biophysical and structural techniques, encompassing microscale thermophoresis, X-ray crystallography and NMR. In particular, we show that solid-state NMR has the ability to reveal effects caused by ligand binding which are more easily relatable to structural and dynamical alterations that impact the stability of macromolecular complexes.

Iron oxide nanoparticles as positive T1 contrast agents for low-field magnetic resonance imaging at 64 mT

We have investigated the efficacy of superparamagnetic iron oxide nanoparticles (SPIONs) as positive T₁ contrast agents for low-field magnetic resonance imaging (MRI) at 64 millitesla (mT). Iron oxide-based agents, such as the FDA-approved ferumoxytol, were measured using a variety of techniques to evaluate T₁ contrast at 64 mT. Additionally, we characterized monodispersed carboxylic acid-coated SPIONs with a range of diameters (4.9-15.7 nm) in order to understand size-dependent properties of T₁ contrast at low-field. MRI contrast properties were measured using 64 mT MRI, magnetometry, and nuclear magnetic resonance dispersion (NMRD). We also measured MRI contrast at 3 T to provide comparison to a standard clinical field strength. SPIONs have the capacity to perform well as T₁ contrast agents at 64 mT, with measured longitudinal relaxivity (r₁) values of up to 67 L mmol^-1 s^-1, more than an order of magnitude higher than corresponding r₁ values at 3 T. The particles exhibit size-dependent longitudinal relaxivities and outperform a commercial Gd-based agent (gadobenate dimeglumine) by more than eight-fold at physiological temperatures. Additionally, we characterize the ratio of transverse to longitudinal relaxivity, r₂/r₁ and find that it is ~ 1 for the SPION based agents at 64 mT, indicating a favorable balance of relaxivities for T₁-weighted contrast imaging. We also correlate the magnetic and structural properties of the particles with models of nanoparticle relaxivity to understand generation of T₁ contrast. These experiments show that SPIONs, at low fields being targeted for point-of-care low-field MRI systems, have a unique combination of magnetic and structural properties that produce large T₁ relaxivities.

Molecular Engineering of Self-Immolative Bioresponsive MR Probes

Real-time detection of bio-event in whole animals provides essential information for understanding biological and therapeutic processes. Magnetic resonance (MR) imaging represents a non-invasive approach to generating three-dimensional anatomic images with high spatial-temporal resolution and unlimited depth penetration. We have developed several self-immolative enzyme-activatable agents that provide excellent in vivo contrast and function as gene expression reporters. Here, we describe a vast improvement in image contrast over our previous generations of these bioresponsive agents based on a new pyridyl-carbamate Gd(III) complex. The pyridyl-carbamate-based agent has a very low MR relaxivity in the "off-state" (r1 = 1.8 mM-1 s-1 at 1.41 T). However, upon enzymatic processing, it generates a significantly higher relaxivity with a Δr1 = 106% versus Δr1 ∼ 20% reported previously. Single X-ray crystal and nuclear magnetic relaxation dispersion analyses offer mechanistic insights regarding MR signal enhancement at the molecular scale. This work demonstrates a pyridyl-carbamate-based self-immolative molecular platform for the construction of enzymatic bio-responsive MR agents, which can be adapted to a wide range of other targets for exploring stimuli-responsive materials and biomedical applications.

Paramagnetic NMR restraints for the characterization of protein structural rearrangements

Mobility is a common feature of biomacromolecules, often fundamental for their function. Thus, in many cases, biomacromolecules cannot be described by a single conformation, but rather by a conformational ensemble. NMR paramagnetic data demonstrated quite informative to monitor this conformational variability, especially when used in conjunction with data from different sources. Due to their long-range nature, paramagnetic data can, for instance, i) clearly demonstrate the occurrence of conformational rearrangements, ii) reveal the presence of minor conformational states, sampled only for a short time, iii) indicate the most representative conformations within the conformational ensemble sampled by the molecule, iv) provide an upper limit to the weight of each conformation.

Elucidating the concentration-dependent effects of thiocyanate binding to carbonic anhydrase

Many proteins naturally carry metal centers, with a large share of them being in the active sites of several enzymes. Paramagnetic effects are a powerful source of structural information and, therefore, if the native metal is paramagnetic, or it can be functionally substituted with a paramagnetic one, paramagnetic effects can be used to study the metal sites, as well as the overall structure of the protein. One notable example is cobalt(II) substitution for zinc(II) in carbonic anhydrase. In this manuscript we investigate the effects of sodium thiocyanate on the chemical environment of the metal ion of the human carbonic anhydrase II. The electron paramagnetic resonance (EPR) titration of the cobalt(II) protein with thiocyanate shows that the EPR spectrum changes from A-type to C-type on passing from 1:1 to 1:1000-fold ligand excess. This indicates the occurrence of a change in the electronic structure, which may reflect a sizable change in the metal coordination environment in turn caused by a modification of the frozen solvent glass. However, paramagnetic nuclear magnetic resonance (NMR) data indicate that the metal coordination cage remains unperturbed even in 1:1000-fold ligand excess. This result proves that the C-type EPR spectrum observed at large ligand concentration should be ascribed to the low temperature at which EPR measurements are performed, which impacts on the structure of the protein when it is destabilized by a high concentration of a chaotropic agent.

Engineered Nonviral Protein Cages Modified for MR Imaging

Diagnostic medical imaging utilizes magnetic resonance (MR) to provide anatomical, functional, and molecular information in a single scan. Nanoparticles are often labeled with Gd(III) complexes to amplify the MR signal of contrast agents (CAs) with large payloads and high proton relaxation efficiencies (relaxivity, r1). This study examined the MR performance of two structurally unique cages, AaLS-13 and OP, labeled with Gd(III). The cages have characteristics relevant for the development of theranostic platforms, including (i) well-defined structure, symmetry, and size; (ii) the amenability to extensive engineering; (iii) the adjustable loading of therapeutically relevant cargo molecules; (iv) high physical stability; and (v) facile manufacturing by microbial fermentation. The resulting conjugates showed significantly enhanced proton relaxivity (r1 = 11-18 mM-1 s-1 at 1.4 T) compared to the Gd(III) complex alone (r1 = 4 mM-1 s-1). Serum phantom images revealed 107% and 57% contrast enhancements for Gd(III)-labeled AaLS-13 and OP cages, respectively. Moreover, proton nuclear magnetic relaxation dispersion (1H NMRD) profiles showed maximum relaxivity values of 50 mM-1 s-1. Best-fit analyses of the 1H NMRD profiles attributed the high relaxivity of the Gd(III)-labeled cages to the slow molecular tumbling of the conjugates and restricted local motion of the conjugated Gd(III) complex.

Large Protein Assemblies for High-Relaxivity Contrast Agents: The Case of Gadolinium-Labeled Asparaginase

Biologics are emerging as the most important class of drugs and are used to treat a large variety of pathologies. Most of biologics are proteins administered in large amounts, either by intramuscular injection or by intravenous infusion. Asparaginase is a large tetrameric protein assembly, currently used against acute lymphoblastic leukemia. Here, a gadolinium(III)-DOTA derivative has been conjugated to asparaginase, and its relaxation properties have been investigated to assess its efficiency as a possible theranostic agent. The field-dependent ¹H longitudinal relaxation measurements of water solutions of gadolinium(III)-labeled asparaginase indicate a very large increase in the relaxivity of this paramagnetic protein complex with respect to small gadolinium chelates, opening up the possibility of its use as an MRI contrast agent.

Solid-state NMR methods for the characterization of bioconjugations and protein-material interactions

Protein solid-state NMR has evolved dramatically over the last two decades, with the development of new hardware and sample preparation methodologies. This technique is now ripe for complex applications, among which one can count bioconjugation, protein chemistry and functional biomaterials. In this review, we provide our account on this aspect of protein solid-state NMR.

The evolution of paramagnetic NMR as a tool in structural biology

Paramagnetic NMR data contain extremely accurate long-range information on metalloprotein structures and, when used in the frame of integrative structural biology approaches, they allow for the retrieval of structural details to a resolution that is not achievable using other techniques. Paramagnetic data thus represent an extremely powerful tool to refine protein models in solution, especially when coupled to X-ray or cryoelectron microscopy data, to monitor the formation of complexes and determine the relative arrangements of their components, and to highlight the presence of conformational heterogeneity. More recently, theoretical and computational advancements in quantum chemical calculations of paramagnetic NMR observables are progressively opening new routes in structural biology, because they allow for the determination of the structure within the coordination sphere of the metal center, thus acting as a loupe on sites that are difficult to observe but very important for protein function.

Theoretical analysis of the long-distance limit of NMR chemical shieldings

After some years of controversy, it was recently demonstrated how to obtain the correct long-distance limit [point-dipole approximation (PDA)] of pseudo-contact nuclear magnetic resonance chemical shifts from rigorous first-principles quantum mechanics [Lang et al., J. Phys. Chem. Lett. 11, 8735 (2020)]. This result confirmed the classical Kurland-McGarvey theory. In the present contribution, we elaborate on these results. In particular, we provide a detailed derivation of the PDA both from the Van den Heuvel-Soncini equation for the chemical shielding tensor and from a spin Hamiltonian approximation. Furthermore, we discuss in detail the PDA within the approximate density functional theory and Hartree-Fock theories. In our previous work, we assumed a relatively crude effective nuclear charge approximation for the spin-orbit coupling operator. Here, we overcome this assumption by demonstrating that the derivation is also possible within the fully relativistic Dirac equation and even without the assumption of a specific form for the Hamiltonian. Crucial ingredients for the general derivation are a Hamiltonian that respects gauge invariance, the multipolar gauge, and functional derivatives of the Hamiltonian, where it is possible to identify the first functional derivative with the electron number current density operator. The present work forms an important foundation for future extensions of the Kurland-McGarvey theory beyond the PDA, including induced magnetic quadrupole and higher moments to describe the magnetic hyperfine field.

High Relaxivity with No Coordinated Waters: A Seemingly Paradoxical Behavior of [Gd(DOTP)]5- Embedded in Nanogels

Nanogels (NGs) obtained by electrostatic interactions between chitosan and hyaluronic acid and comprising paramagnetic Gd chelates are gaining increasing attention for their potential application in magnetic resonance bioimaging. Herein, the macrocyclic complexes [Gd(DOTP)]⁵⁻, lacking metal-bound water molecules (q = 0), were confined or used as a cross-linker in this type of NG. Unlike the typical behavior of Gd complexes with q = 0, a remarkable relaxivity value of 78.0 mM^–1 s^–1 was measured at 20 MHz and 298 K, nearly 20 times greater than that found for the free complex. A careful analysis of the relaxation data emphasizes the fundamental role of second sphere water molecules with strong and long-lived hydrogen bonding interactions with the complex. Finally, PEGylated derivatives of nanoparticles were used for the first in vivo magnetic resonance imaging study of this type of NG, revealing a fast renal excretion of paramagnetic complexes after their release from the NGs.

Nanogels incorporating [Gd(DOTP)]⁵⁻ complexes (q = 0) exhibit remarkable relaxivity values, thanks to structured water molecules in the second coordination shell of the metal ion involved in strong H-bonding interactions with the phosphonate groups.

Not only manganese, but fruit component effects dictate the efficiency of fruit juice as an oral magnetic resonance imaging contrast agent

Several fruit juices are used as oral contrast agents to improve the quality of images in magnetic resonance cholangiopancreatography. They are often preferred to conventional synthetic contrast agents because of their very low cost, natural origin, intrinsic safety, and comparable image qualities. Pineapple and blueberry juices are the most employed in clinical practice due to their higher content of manganese(II) ions. The interest of pharmaceutical companies in these products is testified by the appearance in the market of fruit juice derivatives with improved contrast efficacy. Here, we investigate the origin of the contrast of blueberry juice, analyze the parameters that can effect it, and elucidate the differences with pineapple juice and manganese(II) solutions. It appears that, although manganese(II) is the paramagnetic ion responsible for the contrast, it is the interaction of manganese(II) with other juice components that modulates the efficiency of the juice as a magnetic resonance contrast agent. On these grounds, we conclude that blueberry juice concentrated to the same manganese concentration of pineapple juice would prove a more efficient contrast agent than pineapple juice.

Detection of Metabolite-Protein Interactions in Complex Biological Samples by High-Resolution Relaxometry: Toward Interactomics by NMR

Metabolomics, the systematic investigation of metabolites in biological fluids, cells, or tissues, reveals essential information about metabolism and diseases. Metabolites have functional roles in a myriad of biological processes, as substrates and products of enzymatic reactions but also as cofactors and regulators of large numbers of biochemical mechanisms. These functions involve interactions of metabolites with macromolecules. Yet, methods to systematically investigate these interactions are still scarce to date. In particular, there is a need for techniques suited to identify and characterize weak metabolite-macromolecule interactions directly in complex media such as biological fluids. Here, we introduce a method to investigate weak interactions between metabolites and macromolecules in biological fluids. Our approach is based on high-resolution NMR relaxometry and does not require any invasive procedure or separation step. We show that we can detect interactions between small and large molecules in human blood serum and quantify the size of the complex. Our work opens the way for investigations of metabolite (or other small molecules)-protein interactions in biological fluids for interactomics or pharmaceutical applications.

Unveiling protein dynamics in solution with field-cycling NMR relaxometry

Field-cycling NMR relaxometry is a well-established technique that can give information on molecular structure and dynamics of biological systems. It provides the nuclear relaxation rates as a function of the applied magnetic field, starting from fields as low as ~ 10^-4 T up to about 1-3 T. The profiles so collected, called nuclear magnetic relaxation dispersion (NMRD) profiles, can be extended to include the relaxation rates at the largest fields achievable with high resolution NMR spectrometers. By exploiting this wide range of frequencies, the NMRD profiles can provide information on motions occurring on time scales from 10^-6 to 10^-9 s. ¹H NMRD measurements have proved very useful also for the characterization of paramagnetic proteins, because they can help characterise a number of parameters including the number, distance and residence time of water molecules coordinated to the paramagnetic center, the reorientation correlation times and the electron spin relaxation time, and the electronic structure at the metal site.

A High-Resolution View of the Coordination Environment in a Paramagnetic Metalloprotein from its Magnetic Properties

Metalloproteins constitute a significant fraction of the proteome of all organisms and their characterization is critical for both basic sciences and biomedical applications. A large portion of metalloproteins bind paramagnetic metal ions, and paramagnetic NMR spectroscopy has been widely used in their structural characterization. However, the signals of nuclei in the immediate vicinity of the metal center are often broadened beyond detection. In this work, we show that it is possible to determine the coordination environment of the paramagnetic metal in the protein at a resolution inaccessible to other techniques. Taking the structure of a diamagnetic analogue as a starting point, a geometry optimization is carried out by fitting the pseudocontact shifts obtained from first principles quantum chemical calculations to the experimental ones.

Revisiting paramagnetic relaxation enhancements in slowly rotating systems: how long is the long range?

Cross-relaxation terms in paramagnetic systems that reorient rigidly with slow tumbling times can increase the effective longitudinal relaxation rates of protons of more than 1 order of magnitude. This is evaluated by simulating the time evolution of the nuclear magnetization using a complete relaxation rate-matrix approach. The calculations show that the Solomon dependence of the paramagnetic relaxation rates on the metal-proton distance (as r - 6 ) can be incorrect for protons farther than 15 Å from the metal and thus can cause sizable errors in R 1 -derived distance restraints used, for instance, for protein structure determination. Furthermore, the chemical exchange of these protons with bulk water protons can enhance the relaxation rate of the solvent protons by far more than expected from the paramagnetic Solomon equation. Therefore, it may contribute significantly to the water proton relaxation rates measured at magnetic resonance imaging (MRI) magnetic fields in the presence of slow-rotating nanoparticles containing paramagnetic ions and a large number of exchangeable surface protons.

A Quantum Chemistry View on Two Archetypical Paramagnetic Pentacoordinate Nickel(II) Complexes Offers a Fresh Look on Their NMR Spectra

Quantum chemical methods for calculating paramagnetic NMR observables are becoming increasingly accessible and are being included in the inorganic chemistry practice. Here, we test the performance of these methods in the prediction of proton hyperfine shifts of two archetypical high-spin pentacoordinate nickel(II) complexes (NiSAL-MeDPT and NiSAL-HDPT), which, for a variety of reasons, turned out to be perfectly suited to challenge the predictions to the finest level of detail. For NiSAL-MeDPT, new NMR experiments yield an assignment that perfectly matches the calculations. The slightly different hyperfine shifts from the two “halves” of the molecules related by a pseudo-C₂ axis, which are experimentally divided into two well-defined spin systems, are also straightforwardly distinguished by the calculations. In the case of NiSAL-HDPT, for which no X-ray structure is available, the quality of the calculations allowed us to refine its structure using as a starting template the structure of NiSAL-MeDPT.

State-of-the-art quantum chemical methods and paramagnetism-tailored NMR experiments provide a deep insight on the relation between the spectra and the electronic structure for two paramagnetic pentacoordinate nickel(II) complexes.

1H NMR Relaxometric Study of Chitosan-Based Nanogels Containing Mono- and Bis-Hydrated Gd(III) Chelates: Clues for MRI Probes of Improved Sensitivity

Hydrogel nanoparticles composed of chitosan and hyaluronate and incorporating Gd-based MRI contrast agents with different hydration number (e.g., [Gd(DOTA)(H₂O)]^- and [Gd(AAZTA)(H₂O)₂]^-) were prepared and fully characterized. In particular, ¹H NMR relaxometric data, acquired as a function of temperature and applied magnetic field strength, were for the first time thoroughly analyzed using a theoretical model that includes the effects of a static zero-field splitting and an anisotropic molecular tumbling. The paramagnetic nanoparticles show excellent stability in aqueous solution for over 150 h and do not release the load of Gd(III) chelates. These nanoparticles exhibit enhanced efficacy (relaxivity) as relaxation agents, over 6 times that of the free complexes, thanks to the combination of a restricted molecular dynamics in the presence of a fast exchange of metal-bound water molecule(s) and between the water inside the nanogel and the bulk water. The knowledge of the molecular parameters that control the effectiveness of these MRI nanoprobes and those that limit their further increase will be crucial for the development of optimized systems with high sensitivity and stability.

Solution of a Puzzle: High-Level Quantum-Chemical Treatment of Pseudocontact Chemical Shifts Confirms Classic Semiempirical Theory

A recently popularized approach for the calculation of pseudocontact shifts (PCSs) based on first-principles quantum chemistry (QC) leads to different results than the classic "semiempirical" equation involving the susceptibility tensor. Studies that attempted a comparison of theory and experiment led to conflicting conclusions with respect to the preferred theoretical approach. In this Letter, we show that after inclusion of previously neglected terms in the full Hamiltonian, one can deduce the semiempirical equations from a rigorous QC-based treatment. It also turns out that in the long-distance limit, one can approximate the complete A tensor in terms of the g tensor. By means of Kohn-Sham density functional theory calculations, we numerically confirm the long-distance expression for the A tensor and the theoretically predicted scaling behavior of the different terms. Our derivation suggests a computational strategy in which one calculates the susceptibility tensor and inserts it into the classic equation for the PCS.