Genomics and metabolomics of early-stage thioacetamide-induced liver injury: An interspecies study between guinea pig and rat.

To study the complex processes involved in liver injuries, researchers rely on animal investigations, using chemically or surgically induced liver injuries, to extrapolate findings and infer human health risks. However, this presents obvious challenges in performing a detailed comparison and validation between the highly controlled animal models and development of liver injuries in humans. Furthermore, it is not clear whether there are species-dependent and -independent molecular initiating events or processes that cause liver injury before they eventually lead to end-stage liver disease. Here, we present a side-by-side study of rats and guinea pigs using thioacetamide to examine the similarities between early molecular initiating events during an acute-phase liver injury. We exposed Sprague Dawley rats and Hartley guinea pigs to a single dose of 25 or 100 mg/kg thioacetamide and collected blood plasma for metabolomic analysis and liver tissue for RNA-sequencing. The subsequent toxicogenomic analysis identified consistent liver injury trends in both genomic and metabolomic data within 24 and 33 h after thioacetamide exposure in rats and guinea pigs, respectively. In particular, we found species similarities in the key injury phenotypes of inflammation and fibrogenesis in our gene module analysis for liver injury phenotypes. We identified expression of several common genes (e.g., SPP1, TNSF18, SERPINE1, CLDN4, TIMP1, CD44, and LGALS3), activation of injury-specific KEGG pathways, and alteration of plasma metabolites involved in amino acid and bile acid metabolism as some of the key molecular processes that changed early upon thioacetamide exposure and could play a major role in the initiation of acute liver injury.

TOXPANEL: A Gene-Set Analysis Tool to Assess Liver and Kidney Injuries.

Gene-set analysis is commonly used to identify trends in gene expression when cells, tissues, organs, or organisms are subjected to conditions that differ from those within the normal physiological range. However, tools for gene-set analysis to assess liver and kidney injury responses are less common. Furthermore, most websites for gene-set analysis lack the option for users to customize their gene-set database. Here, we present the ToxPanel website, which allows users to perform gene-set analysis to assess liver and kidney injuries using activation scores based on gene-expression fold-change values. The results are graphically presented to assess constituent injury phenotypes (histopathology), with interactive result tables that identify the main contributing genes to a given signal. In addition, ToxPanel offers the flexibility to analyze any set of custom genes based on gene fold-change values. ToxPanel is publically available online at https://toxpanel.bhsai.org. ToxPanel allows users to access our previously developed liver and kidney injury gene sets, which we have shown in previous work to yield robust results that correlate with the degree of injury. Users can also test and validate their customized gene sets using the ToxPanel website.

Concordance between Thioacetamide-Induced Liver Injury in Rat and Human In Vitro Gene Expression Data.

The immense resources required and the ethical concerns for animal-based toxicological studies have driven the development of in vitro and in silico approaches. Recently, we validated our approach in which the expression of a set of genes is uniquely associated with an organ-injury phenotype (injury module), by using thioacetamide, a known liver toxicant. Here, we sought to explore whether RNA-seq data obtained from human cells (in vitro) treated with thioacetamide-S-oxide (a toxic intermediate metabolite) would correlate across species with the injury responses found in rat cells (in vitro) after exposure to this metabolite as well as in rats exposed to thioacetamide (in vivo). We treated two human cell types with thioacetamide-S-oxide (primary hepatocytes with 0 (vehicle), 0.125 (low dose), or 0.25 (high dose) mM, and renal tubular epithelial cells with 0 (vehicle), 0.25 (low dose), or 1.00 (high dose) mM) and collected RNA-seq data 9 or 24 h after treatment. We found that the liver-injury modules significantly altered in human hepatocytes 24 h after high-dose treatment involved cellular infiltration and bile duct proliferation, which are linked to fibrosis. For high-dose treatments, our modular approach predicted the rat in vivo and in vitro results from human in vitro RNA-seq data with Pearson correlation coefficients of 0.60 and 0.63, respectively, which was not observed for individual genes or KEGG pathways.

A toxicogenomic approach to assess kidney injury induced by mercuric chloride in rats.

Kidney injury caused by disease, trauma, environmental exposures, or drugs may result in decreased renal function, chronic kidney disease, or acute kidney failure. Diagnosis of kidney injury using serum creatinine levels, a common clinical test, only identifies renal dysfunction after the kidneys have undergone severe damage. Other indicators sensitive to kidney injury, such as the level of urine kidney injury molecule-1 (KIM-1), lack the ability to differentiate between injury phenotypes. To address early detection as well as detailed categorization of kidney-injury phenotypes in preclinical animal or cellular studies, we previously identified eight sets (modules) of co-expressed genes uniquely associated with kidney histopathology. Here, we used mercuric chloride (HgCl2)-a model nephrotoxicant-to chemically induce kidney injuries as monitored by KIM-1 levels in Sprague Dawley rats at two doses (0.25 or 0.50 mg/kg) and two exposure lengths (10 or 34 h). We collected whole transcriptome RNA-seq data derived from five animals at each dose and time point to perform a toxicogenomics analysis. Consistent with documented injury phenotypes for HgCl2 toxicity, our kidney-injury-module approach identified the onset of necrosis and dilation as early as 10 h after a dose of 0.50 mg/kg that produced only mild injury as judged by urinary KIM-1 excretion. The results of these animal studies highlight the potential of the kidney-injury-module approach to provide a sensitive and histopathology-specific readout of renal toxicity.

Assessing Chemical-Induced Liver Injury In Vivo From In Vitro Gene Expression Data in the Rat: The Case of Thioacetamide Toxicity.

Consumers are exposed to thousands of chemicals with potentially adverse health effects. However, these chemicals will never be tested for toxicity because of the immense resources needed for animal-based (in vivo) toxicological studies. Today, there are no viable in vitro alternatives to these types of animal studies. To develop an in vitro approach, we investigated whether we could predict in vivo organ injuries in rats with the use of RNA-seq data acquired from tissues early in the development of toxicant-induced injury, by comparing gene expression data from RNA isolated from these rat tissues with those obtained from in vitro exposure of primary liver and kidney cells. We collected RNA-seq data from the liver and kidney tissues of Sprague-Dawley rats 8 or 24 h after exposing them to vehicle (control), low (25 mg/kg), or high (100 mg/kg) doses of thioacetamide, a known liver toxicant that promotes fibrosis; we used these doses and exposure times to cause only mild toxicant-induced injury. For the in vitro study, we treated two cell types from Sprague-Dawley rats, primary hepatocytes (vehicle; low, 0.025 mM; or high, 0.125 mM dose), and renal tube epithelial cells (vehicle; low, 0.125 mM; or high, 0.500 mM) dose) with the thioacetamide metabolite, thioacetamide-S-oxide, selecting in vitro doses and exposure times to recreate the early-stage toxicant-induced injury model that we achieved in vivo. RNA-seq data were collected 9 or 24 h after application of vehicle or thioacetamide-S-oxide. We found that our modular approach for the analysis of gene expression data derived from in vivo RNA-seq strongly correlated (R2 > 0.6) with the in vitro results at two different dose levels of thioacetamide/thioacetamide-S-oxide after 24 h of exposure. The top-ranked liver injury modules in vitro correctly identified the ensuing development of liver fibrosis.

Short-term metabolic adjustments in Plasmodium falciparum counter hypoxanthine deprivation at the expense of long-term viability.

BACKGROUND: The malarial parasite Plasmodium falciparum is an auxotroph for purines, which are required for nucleic acid synthesis during the intra-erythrocytic developmental cycle (IDC) of the parasite. The capabilities of the parasite and extent to which it can use compensatory mechanisms to adapt to purine deprivation were studied by examining changes in its metabolism under sub-optimal concentrations of hypoxanthine, the primary precursor utilized by the parasite for purine-based nucleic acid synthesis. METHODS: The concentration of hypoxanthine that caused a moderate growth defect over the course of one IDC was determined. At this concentration of hypoxanthine (0.5 µM), transcriptomic and metabolomic data were collected during one IDC at multiple time points. These data were integrated with a metabolic network model of the parasite embedded in a red blood cell (RBC) to interpret the metabolic adaptation of P. falciparum to hypoxanthine deprivation. RESULTS: At a hypoxanthine concentration of 0.5 µM, vacuole-like structures in the cytosol of many P. falciparum parasites were observed after the 24-h midpoint of the IDC. Parasites grown under these conditions experienced a slowdown in the progression of the IDC. After 72 h of deprivation, the parasite growth could not be recovered despite supplementation with 90 µM hypoxanthine. Simulations of P. falciparum metabolism suggested that alterations in ubiquinone, isoprenoid, shikimate, and mitochondrial metabolism occurred before the appearance of these vacuole-like structures. Alterations were found in metabolic reactions associated with fatty acid synthesis, the pentose phosphate pathway, methionine metabolism, and coenzyme A synthesis in the latter half of the IDC. Furthermore, gene set enrichment analysis revealed that P. falciparum activated genes associated with rosette formation, Maurer's cleft and protein export under two different nutrient-deprivation conditions (hypoxanthine and isoleucine). CONCLUSIONS: The metabolic network analysis presented here suggests that P. falciparum invokes specific purine-recycling pathways to compensate for hypoxanthine deprivation and maintains a hypoxanthine pool for purine-based nucleic acid synthesis. However, this compensatory mechanism is not sufficient to maintain long-term viability of the parasite. Although P. falciparum can complete a full IDC in low hypoxanthine conditions, subsequent cycles are disrupted.

Deep Neural Network Models for Predicting Chemically Induced Liver Toxicity Endpoints From Transcriptomic Responses.

Improving the accuracy of toxicity prediction models for liver injuries is a key element in evaluating the safety of drugs and chemicals. Mechanism-based information derived from expression (transcriptomic) data, in combination with machine-learning methods, promises to improve the accuracy and robustness of current toxicity prediction models. Deep neural networks (DNNs) have the advantage of automatically assembling the relevant features from a large number of input features. This makes them especially suitable for modeling transcriptomic data, which typically contain thousands of features. Here, we gaged gene- and pathway-level feature selection schemes using single- and multi-task DNN approaches in predicting chemically induced liver injuries (biliary hyperplasia, fibrosis, and necrosis) from whole-genome DNA microarray data. The single-task DNN models showed high predictive accuracy and endpoint specificity, with Matthews correlation coefficients for the three endpoints on 10-fold cross validation ranging from 0.56 to 0.89, with an average of 0.74 in the best feature sets. The DNN models outperformed Random Forest models in cross validation and showed better performance than Support Vector Machine models when tested in the external validation datasets. In the cross validation studies, the effect of the feature selection scheme was negligible among the studied feature sets. Further evaluation of the models on their ability to predict the injury phenotype per se for non-chemically induced injuries revealed the robust performance of the DNN models across these additional external testing datasets. Thus, the DNN models learned features specific to the injury phenotype contained in the gene expression data.

Identification of the Toxicity Pathways Associated With Thioacetamide-Induced Injuries in Rat Liver and Kidney.

Ingestion or exposure to chemicals poses a serious health risk. Early detection of cellular changes induced by such events is vital to identify appropriate countermeasures to prevent organ damage. We hypothesize that chemically induced organ injuries are uniquely associated with a set (module) of genes exhibiting significant changes in expression. We have previously identified gene modules specifically associated with organ injuries by analyzing gene expression levels in liver and kidney tissue from rats exposed to diverse chemical insults. Here, we assess and validate our injury-associated gene modules by analyzing gene expression data in liver, kidney, and heart tissues obtained from Sprague-Dawley rats exposed to thioacetamide, a known liver toxicant that promotes fibrosis. The rats were injected intraperitoneally with a low (25 mg/kg) or high (100 mg/kg) dose of thioacetamide for 8 or 24 h, and definite organ injury was diagnosed by histopathology. Injury-associated gene modules indicated organ injury specificity, with the liver being most affected by thioacetamide. The most activated liver gene modules were those associated with inflammatory cell infiltration and fibrosis. Previous studies on thioacetamide toxicity and our histological analyses supported these results, signifying the potential of gene expression data to identify organ injuries.

vNN Web Server for ADMET Predictions.

In drug development, early assessments of pharmacokinetic and toxic properties are important stepping stones to avoid costly and unnecessary failures. Considerable progress has recently been made in the development of computer-based (in silico) models to estimate such properties. Nonetheless, such models can be further improved in terms of their ability to make predictions more rapidly, easily, and with greater reliability. To address this issue, we have used our vNN method to develop 15 absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction models. These models quickly assess some of the most important properties of potential drug candidates, including their cytotoxicity, mutagenicity, cardiotoxicity, drug-drug interactions, microsomal stability, and likelihood of causing drug-induced liver injury. Here we summarize the ability of each of these models to predict such properties and discuss their overall performance. All of these ADMET models are publically available on our website (https://vnnadmet.bhsai.org/), which also offers the capability of using the vNN method to customize and build new models.

General Purpose 2D and 3D Similarity Approach to Identify hERG Blockers.

Screening compounds for human ether-à-go-go-related gene (hERG) channel inhibition is an important component of early stage drug development and assessment. In this study, we developed a high-confidence (p-value < 0.01) hERG prediction model based on a combined two-dimensional (2D) and three-dimensional (3D) modeling approach. We developed a 3D similarity conformation approach (SCA) based on examining a limited fixed number of pairwise 3D similarity scores between a query molecule and a set of known hERG blockers. By combining 3D SCA with 2D similarity ensemble approach (SEA) methods, we achieved a maximum sensitivity in hERG inhibition prediction with an accuracy not achieved by either method separately. The combined model achieved 69% sensitivity and 95% specificity on an independent external data set. Further validation showed that the model correctly picked up documented hERG inhibition or interactions among the Food and Drug Administration- approved drugs with the highest similarity scores-with 18 of 20 correctly identified. The combination of ascertaining 2D and 3D similarity of compounds allowed us to synergistically use 2D fingerprint matching with 3D shape and chemical complementarity matching.

Using the Variable-Nearest Neighbor Method To Identify P-Glycoprotein Substrates and Inhibitors.

Permeability glycoprotein (Pgp) is an essential membrane-bound transporter that efficiently extracts compounds from a cell. As such, it is a critical determinant of the pharmacokinetic properties of drugs. Multidrug resistance in cancer is often associated with overexpression of Pgp, which increases the efflux of chemotherapeutic agents from the cell. This, in turn, may prevent an effective treatment by reducing the effective intracellular concentrations of such agents. Consequently, identifying compounds that can either be transported out of the cell by Pgp (substrates) or impair Pgp function (inhibitors) is of great interest. Herein, using publically available data, we developed quantitative structure-activity relationship (QSAR) models of Pgp substrates and inhibitors. These models employed a variable-nearest neighbor (v-NN) method that calculated the structural similarity between molecules and hence possessed an applicability domain, that is, they used all nearest neighbors that met a minimum similarity constraint. The performance characteristics of these v-NN-based models were comparable or at times superior to those of other model constructs. The best v-NN models for identifying either Pgp substrates or inhibitors showed overall accuracies of >80% and κ values of >0.60 when tested on external data sets with candidate Pgp substrates and inhibitors. The v-NN prediction model with a well-defined applicability domain gave accurate and reliable results. The v-NN method is computationally efficient and requires no retraining of the prediction model when new assay information becomes available-an important feature when keeping QSAR models up-to-date and maintaining their performance at high levels.

Critically Assessing the Predictive Power of QSAR Models for Human Liver Microsomal Stability.

To lower the possibility of late-stage failures in the drug development process, an up-front assessment of absorption, distribution, metabolism, elimination, and toxicity is commonly implemented through a battery of in silico and in vitro assays. As in vitro data is accumulated, in silico quantitative structure-activity relationship (QSAR) models can be trained and used to assess compounds even before they are synthesized. Even though it is generally recognized that QSAR model performance deteriorates over time, rigorous independent studies of model performance deterioration is typically hindered by the lack of publicly available large data sets of structurally diverse compounds. Here, we investigated predictive properties of QSAR models derived from an assembly of publicly available human liver microsomal (HLM) stability data using variable nearest neighbor (v-NN) and random forest (RF) methods. In particular, we evaluated the degree of time-dependent model performance deterioration. Our results show that when evaluated by 10-fold cross-validation with all available HLM data randomly distributed among 10 equal-sized validation groups, we achieved high-quality model performance from both machine-learning methods. However, when we developed HLM models based on when the data appeared and tried to predict data published later, we found that neither method produced predictive models and that their applicability was dramatically reduced. On the other hand, when a small percentage of randomly selected compounds from data published later were included in the training set, performance of both machine-learning methods improved significantly. The implication is that 1) QSAR model quality should be analyzed in a time-dependent manner to assess their true predictive power and 2) it is imperative to retrain models with any up-to-date experimental data to ensure maximum applicability.

Cooperative effects and optimal halogen bonding motifs for self-assembling systems.

Halogen bonding, due to its directionality and tunable strength, is being increasingly utilized in self-assembling materials and crystal engineering. Using density functional theory (DFT) and molecular mechanics (OPLS/CM1Ax) calculations, multiply halogen bonded complexes of brominated imidazole and pyridine are investigated along with their potential in construction of self-assembling architectures. Dimers with 1-10 halogen bonds are considered and reveal maximal binding energies of 3-36 kcal/mol. Cooperative (nonadditive) effects are found in complexes that extend both along and perpendicular to the halogen bonding axes, with interaction energies depending on polarization, secondary interactions, and ring spacers. Four structural motifs were identified to yield optimal halogen bonding. For the largest systems, the excellent agreement found between the DFT and OPLS/CM1Ax results supports the utility of the latter approach for analysis and design of self-assembling supramolecular structures.

Characterization of biaryl torsional energetics and its treatment in OPLS all-atom force fields.

The frequency of biaryl substructures in a database of approved oral drugs has been analyzed. This led to designation of 20 prototypical biaryls plus 10 arylpyridinones for parametrization in the OPLS all-atom force fields. Bond stretching, angle-bending, and torsional parameters were developed to reproduce the MP2 geometries and torsional energy profiles. The transferability of the new parameters was tested through their application to three additional biaryls. The torsional energetics for the 33 biaryl molecules are analyzed and factors leading to preferences for planar and nonplanar geometries are identified. For liquid biphenyl, the computed density and heat of vaporization at the boiling point (255 °C) are also reported.

Exploring Adsorption of Water and Ions on Carbon Surfaces using a Polarizable Force Field.

Graphene, carbon nanotubes, and fullerenes are of great interest due to their unique properties and diverse applications in biology, molecular electronics, and materials science. Therefore, there is demand for methods that can accurately model the interface between carbon surfaces and their environment. In this letter we compare results for complexes of water, potassium ion, and chloride ion with graphene, carbon nanotube, and fullerene surfaces using a standard non-polarizable force field (OPLS-AA), a polarizable force field (OPLS-AAP), DFT, and ab initio theory. For interactions with water, OPLS-AA with the TIP3P or TIP4P water models describes the interactions with benzene (C(6)H(6)) and coronene (C(24)H(12)) well; however, for acenes larger than circumcoronene (C(54)H(18)) and especially for C(60), the interaction energies are somewhat too weak and polarization is needed. For ions interacting with carbon surfaces, inclusion of polarization is essential, and OPLS-AAP is found to perform well in comparison to the highest-level quantum mechanical methods. Overall, OPLS-AAP provides an accurate and computationally efficient force field for modeling condensed-phase systems featuring carbon surfaces.

Intramolecular gas-phase reactions of synthetic nonheme oxoiron(IV) ions: proximity and spin-state reactivity rules.

The intramolecular gas-phase reactivity of four oxoiron(IV) complexes supported by tetradentate N(4) ligands (L) has been studied by means of tandem mass spectrometry measurements in which the gas-phase ions [Fe(IV)(O)(L)(OTf)](+) (OTf = trifluoromethanesulfonate) and [Fe(IV) (O)(L)](2+) were isolated and then allowed to fragment by collision-induced decay (CID). CID fragmentation of cations derived from oxoiron(IV) complexes of 1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane (tmc) and N,N'-bis(2-pyridylmethyl)-1,5-diazacyclooctane (L(8)Py(2)) afforded the same predominant products irrespective of whether they were hexacoordinate or pentacoordinate. These products resulted from the loss of water by dehydrogenation of ethylene or propylene linkers on the tetradentate ligand. In contrast, CID fragmentation of ions derived from oxoiron(IV) complexes of linear tetradentate ligands N,N'-bis(2-pyridylmethyl)-1,2-diaminoethane (bpmen) and N,N'-bis(2-pyridylmethyl)-1,3-diaminopropane (bpmpn) showed predominant oxidative N-dealkylation for the hexacoordinate [Fe(IV)(O)(L)(OTf)](+) cations and predominant dehydrogenation of the diaminoethane/propane backbone for the pentacoordinate [Fe(IV)(O)(L)](2+) cations. DFT calculations on [Fe(IV)(O)(bpmen)] ions showed that the experimentally observed preference for oxidative N-dealkylation versus dehydrogenation of the diaminoethane linker for the hexa- and pentacoordinate ions, respectively, is dictated by the proximity of the target C-H bond to the oxoiron(IV) moiety and the reactive spin state. Therefore, there must be a difference in ligand topology between the two ions. More importantly, despite the constraints on the geometries of the TS that prohibit the usual upright σ trajectory and prevent optimal σ(CH)-σ*(z2) overlap, all the reactions still proceed preferentially on the quintet (S = 2) state surface, which increases the number of exchange interactions in the d block of iron and leads thereby to exchange enhanced reactivity (EER). As such, EER is responsible for the dominance of the S = 2 reactions for both hexa- and pentacoordinate complexes.

Treatment of Halogen Bonding in the OPLS-AA Force Field; Application to Potent Anti-HIV Agents.

The representation of chlorine, bromine, and iodine in aryl halides has been modified in the OPLS-AA and OPLS/CM1A force fields in order to incorporate halogen bonding. The enhanced force fields, OPLS-AAx and OPLS/CM1Ax, have been tested in calculations on gas-phase complexes of halobenzenes with Lewis bases, and for free energies of hydration, densities, and heats of vaporization of halobenzenes. Comparisons with results of MP2/aug-cc-pVDZ(-PP) calculations for the complexes are included. Implementation in the MCPRO software also allowed computation of relative free energies of binding for a series of HIV reverse transcriptase inhibitors via Monte Carlo/free-energy perturbation calculations. The results support the notion that the activity of an unusually potent chloro analog likely benefits from halogen bonding with the carbonyl group of a proline residue.

The directive of the protein: how does cytochrome P450 select the mechanism of dopamine formation?

Dopamine can be generated from tyramine via arene hydroxylation catalyzed by a cytochrome P450 enzyme (CYP2D6). Our quantum mechanical/molecular mechanical (QM/MM) results reveal the decisive impact of the protein in selecting the 'best' reaction mechanism. Instead of the traditional Meisenheimer-complex mechanism, the study reveals a mechanism involving an initial hydrogen atom transfer from the phenolic hydroxyl group of the tyramine to the iron-oxo of the compound I (Cpd I), followed by a ring-π radical rebound that eventually leads to dopamine by keto-enol rearrangement. This mechanism is not viable in the gas phase since the O-H bond activation by Cpd I is endothermic and the process does not form a stable intermediate. By contrast, the in-protein reaction has a low barrier and is exothermic. It is shown that the local electric field of the protein environment serves as a template that stabilizes the intermediate of the H-abstraction step and thereby mediates the catalysis of dopamine formation at a lower energy cost. Furthermore, it is shown that external electric fields can either catalyze or inhibit the process depending on their directionality.

Trends in Aromatic Oxidation Reactions Catalyzed by Cytochrome P450 Enzymes: A Valence Bond Modeling.

The mixed density functional theory (DFT) and valence bond study described herein focuses on the activation of 17 benzene derivatives by the active species of Cytochrome P450, so-called Compound I (Cpd I), as well as by the methoxy radical, as a potentially simple model of Cpd I (Jones, J. P.; Mysinger, M.; Korzekwa, K. R. Drug Metab. Dispos. 2002, 30, 7-12). Valence bond modeling is employed to rationalize the P450 mechanism and its spin-state selectivity from first principles of electronic structure and to predict activation energies independently, using easily accessible properties of the reactants: the singlet-triplet excitation energies, the ionization potentials of the aromatics, and the electron affinity of Cpd I and/or of the methoxy radical. It is shown that the valence bond model rationalizes all the mechanistic aspects and predicts activation barriers (for 35 reactions) with reasonable accuracy compared to the DFT barriers with an average deviation of ±1.0 kcal·mol(-1) (for DFT barriers, see: Bathelt, C. M.; Ridder, L.; Mulholland, A. J.; Harvey, J. N. Org. Biomol. Chem. 2004, 2, 2998-3005). The valence bond modeling also reveals the mechanistic similarities between the P450 Cpd I and methoxy reactions and enables one to make predictions of barriers for reactions from other studies.