Humanization of antibodies using a machine learning approach on large-scale repertoire data.

MOTIVATION: Monoclonal antibody (mAb) therapeutics are often produced from non-human sources (typically murine), and can therefore generate immunogenic responses in humans. Humanization procedures aim to produce antibody therapeutics that do not elicit an immune response and are safe for human use, without impacting efficacy. Humanization is normally carried out in a largely trial-and-error experimental process. We have built machine learning classifiers that can discriminate between human and non-human antibody variable domain sequences using the large amount of repertoire data now available. RESULTS: Our classifiers consistently outperform the current best-in-class model for distinguishing human from murine sequences, and our output scores exhibit a negative relationship with the experimental immunogenicity of existing antibody therapeutics. We used our classifiers to develop a novel, computational humanization tool, Hu-mAb, that suggests mutations to an input sequence to reduce its immunogenicity. For a set of therapeutic antibodies with known precursor sequences, the mutations suggested by Hu-mAb show substantial overlap with those deduced experimentally. Hu-mAb is therefore an effective replacement for trial-and-error humanization experiments, producing similar results in a fraction of the time. AVAILABILITY AND IMPLEMENTATION: Hu-mAb (humanness scoring and humanization) is freely available to use at opig.stats.ox.ac.uk/webapps/humab. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The temporal Doppler effect: when the future feels closer than the past.

People routinely remember events that have passed and imagine those that are yet to come. The past and the future are sometimes psychologically close ("just around the corner") and other times psychologically distant ("ages away"). Four studies demonstrate a systematic asymmetry whereby future events are psychologically closer than past events of equivalent objective distance. When considering specific times (e.g., 1 year) or events (e.g., Valentine's Day), people consistently reported that the future was closer than the past. We suggest that this asymmetry arises because the subjective experience of movement through time (whereby future events approach and past events recede) is analogous to the physical experience of movement through space. Consistent with this hypothesis, experimentally reversing the metaphorical arrow of time (by having participants move backward through virtual space) completely eliminated the past-future asymmetry. We discuss how reducing psychological distance to the future may function to prepare people for upcoming action.

Endogenous 3,4-dihydroxyphenylalanine and dopaquinone modifications on protein tyrosine: links to mitochondrially derived oxidative stress via hydroxyl radical.

Oxidative modifications of protein tyrosines have been implicated in multiple human diseases. Among these modifications, elevations in levels of 3,4-dihydroxyphenylalanine (DOPA), a major product of hydroxyl radical addition to tyrosine, has been observed in a number of pathologies. Here we report the first proteome survey of endogenous site-specific modifications, i.e. DOPA and its further oxidation product dopaquinone in mouse brain and heart tissues. Results from LC-MS/MS analyses included 50 and 14 DOPA-modified tyrosine sites identified from brain and heart, respectively, whereas only a few nitrotyrosine-containing peptides, a more commonly studied marker of oxidative stress, were detectable, suggesting the much higher abundance for DOPA modification as compared with tyrosine nitration. Moreover, 20 and 12 dopaquinone-modified peptides were observed from brain and heart, respectively; nearly one-fourth of these peptides were also observed with DOPA modification on the same sites. For both tissues, these modifications are preferentially found in mitochondrial proteins with metal binding properties, consistent with metal-catalyzed hydroxyl radical formation from mitochondrial superoxide and hydrogen peroxide. These modifications also link to a number of mitochondrially associated and other signaling pathways. Furthermore, many of the modification sites were common sites of previously reported tyrosine phosphorylation, suggesting potential disruption of signaling pathways. Collectively, the results suggest that these modifications are linked with mitochondrially derived oxidative stress and may serve as sensitive markers for disease pathologies.

Physical attractiveness predicts endorsement of specific evolutionary psychology principles.

Evolutionary psychology has emerged as a controversial discipline, particularly with regard to its claims concerning the biological basis of sex differences in human mate preferences. Drawing on theories of motivated inference, we hypothesized that those who are most likely to be privileged by specific aspects of the theory would be most likely to support the theory. In particular, we predicted that physical attractiveness would be positively associated with endorsement of predictions of evolutionary psychology concerning mating strategies. Two studies confirmed this hypothesis. In Study 1, participants rated as higher in physical attractiveness were more likely to support specific principles of evolutionary psychology. In Study 2, a manipulation designed to boost self-perceived physical attractiveness increased endorsement of those same principles. Observer-rated physical attractiveness generally predicted individuals' support of the theoretical principles better than did gender, political orientation, or self-esteem. Results suggest that those most likely to benefit according to certain predictions of evolutionary psychology are also those most likely to be sympathetic toward its relevant principles.

Region-specific protein abundance changes in the brain of MPTP-induced Parkinson's disease mouse model.

Parkinson's disease (PD) is characterized by dopaminergic neurodegeneration in the nigrostriatal region of the brain; however, the neurodegeneration extends well beyond dopaminergic neurons. To gain a better understanding of the molecular changes relevant to PD, we applied two-dimensional LC-MS/MS to comparatively analyze the proteome changes in four brain regions (striatum, cerebellum, cortex, and the rest of brain) using a MPTP-induced PD mouse model with the objective to identify potential nigrostriatal-specific and other region-specific protein abundance changes. The combined analyses resulted in the identification of 4,895 nonredundant proteins with at least two unique peptides per protein. The relative abundance changes in each analyzed brain region were estimated based on the spectral count information. A total of 518 proteins were observed with substantial MPTP-induced abundance changes across different brain regions. A total of 270 of these proteins were observed with specific changes occurring either only in the striatum and/or in the rest of the brain region that contains substantia nigra, suggesting that these proteins are associated with the underlying nigrostriatal pathways. Many of the proteins that exhibit changes were associated with dopamine signaling, mitochondrial dysfunction, the ubiquitin system, calcium signaling, the oxidative stress response, and apoptosis. A set of proteins with either consistent change across all brain regions or with changes specific to the cortex and cerebellum regions were also detected. Ubiquitin specific protease (USP9X), a deubiquination enzyme involved in the protection of proteins from degradation and promotion of the TGF-beta pathway, exhibited altered abundance in all brain regions. Western blot validation showed similar spatial changes, suggesting that USP9X is potentially associated with neurodegeneration. Together, this study for the first time presents an overall picture of proteome changes underlying both nigrostriatal pathways and other brain regions potentially involved in MPTP-induced neurodegeneration. The observed molecular changes provide a valuable reference resource for future hypothesis-driven functional studies of PD.

Analysis of multiplex gene expression maps obtained by voxelation.

BACKGROUND: Gene expression signatures in the mammalian brain hold the key to understanding neural development and neurological disease. Researchers have previously used voxelation in combination with microarrays for acquisition of genome-wide atlases of expression patterns in the mouse brain. On the other hand, some work has been performed on studying gene functions, without taking into account the location information of a gene's expression in a mouse brain. In this paper, we present an approach for identifying the relation between gene expression maps obtained by voxelation and gene functions. RESULTS: To analyze the dataset, we chose typical genes as queries and aimed at discovering similar gene groups. Gene similarity was determined by using the wavelet features extracted from the left and right hemispheres averaged gene expression maps, and by the Euclidean distance between each pair of feature vectors. We also performed a multiple clustering approach on the gene expression maps, combined with hierarchical clustering. Among each group of similar genes and clusters, the gene function similarity was measured by calculating the average gene function distances in the gene ontology structure. By applying our methodology to find similar genes to certain target genes we were able to improve our understanding of gene expression patterns and gene functions. By applying the clustering analysis method, we obtained significant clusters, which have both very similar gene expression maps and very similar gene functions respectively to their corresponding gene ontologies. The cellular component ontology resulted in prominent clusters expressed in cortex and corpus callosum. The molecular function ontology gave prominent clusters in cortex, corpus callosum and hypothalamus. The biological process ontology resulted in clusters in cortex, hypothalamus and choroid plexus. Clusters from all three ontologies combined were most prominently expressed in cortex and corpus callosum. CONCLUSION: The experimental results confirm the hypothesis that genes with similar gene expression maps might have similar gene functions. The voxelation data takes into account the location information of gene expression level in mouse brain, which is novel in related research. The proposed approach can potentially be used to predict gene functions and provide helpful suggestions to biologists.

A genome-scale map of expression for a mouse brain section obtained using voxelation.

Gene expression signatures in the mammalian brain hold the key to understanding neural development and neurological diseases. We have reconstructed two-dimensional images of gene expression for 20,000 genes in a coronal slice of the mouse brain at the level of the striatum by using microarrays in combination with voxelation at a resolution of 1 mm3. Good reliability of the microarray results were confirmed using multiple replicates, subsequent quantitative RT-PCR voxelation, mass spectrometry voxelation, and publicly available in situ hybridization data. Known and novel genes were identified with expression patterns localized to defined substructures within the brain. In addition, genes with unexpected patterns were identified, and cluster analysis identified a set of genes with a gradient of dorsal/ventral expression not restricted to known anatomical boundaries. The genome-scale maps of gene expression obtained using voxelation will be a valuable tool for the neuroscience community.

Neuronal ceroid lipofuscinoses are connected at molecular level: interaction of CLN5 protein with CLN2 and CLN3.

Neuronal ceroid lipofuscinoses (NCLs) are neurodegenerative storage diseases characterized by mental retardation, visual failure, and brain atrophy as well as accumulation of storage material in multiple cell types. The diseases are caused by mutations in the ubiquitously expressed genes, of which six are known. Herein, we report that three NCL disease forms with similar tissue pathology are connected at the molecular level: CLN5 polypeptides directly interact with the CLN2 and CLN3 proteins based on coimmunoprecipitation and in vitro binding assays. Furthermore, disease mutations in CLN5 abolished interaction with CLN2, while not affecting association with CLN3. The molecular characterization of CLN5 revealed that it was synthesized as four precursor forms, due to usage of alternative initiator methionines in translation. All forms were targeted to lysosomes and the longest form, translated from the first potential methionine, was associated with membranes. Interactions between CLN polypeptides were shown to occur with this longest, membrane-bound form of CLN5. Both intracellular targeting and posttranslational glycosylation of the polypeptides carrying human disease mutations were similar to wild-type CLN5.

Isolation and Characterization of a Distinct Subpopulation from the WM115 Cell Line That Resembles In Vitro Properties of Melanoma Cancer Stem Cells.

Despite key advances in cancer therapies, malignant tumors, such as melanoma, continue to be one of the leading causes of mortality. Recent debate on whether cancer can originate from a tumor-initiating subpopulation has permeated oncology and stem cell research. It has been well established that primary and immortalized tumor cells consist of heterogeneous cell populations. The profound effect of tumor heterogeneity on tumor growth and drug resistance remains elusive, but it is highly likely that subpopulations of cancer cells have different capabilities of self-renewal and drug resistance. Discrepancies between excellent in vitro potency and efficacy and poor patient response have been observed on multiple cancer therapeutics. Although this observation can be attributed to many factors, a better understanding of the contribution from subpopulations within a cancer will help bridge the gap between in vitro assay results and patient prognosis. To comprehend this impact, it is critical to isolate and characterize cancer subpopulations that possess higher growth and drug resistance properties so that novel therapeutics can be developed to eventually eradicate all cancer cells. In this article, we describe a method to enrich a subpopulation, CB4, from the melanoma cell line WM115. CB4 exhibited higher anchorage-independent growth, higher survival under serum starvation condition, and lower drug sensitivity to commonly used melanoma treatment compared with WM115. Details of functional properties and gene expression of CB4 compared with WM115 are reported. Our study demonstrates that it is feasible to isolate and enrich a subpopulation that exhibits higher growth capacity and treatment resistance from an immortalized tumor cell line.

The pluripotency factor-bound intron 1 of Xist is dispensable for X chromosome inactivation and reactivation in vitro and in vivo.

X chromosome inactivation (XCI) is a dynamically regulated developmental process with inactivation and reactivation accompanying the loss and gain of pluripotency, respectively. A functional relationship between pluripotency and lack of XCI has been suggested, whereby pluripotency transcription factors repress the master regulator of XCI, the noncoding transcript Xist, by binding to its first intron (intron 1). To test this model, we have generated intron 1 mutant embryonic stem cells (ESCs) and two independent mouse models. We found that Xist's repression in ESCs, its transcriptional upregulation upon differentiation, and its silencing upon reprogramming to pluripotency are not dependent on intron 1. Although we observed subtle effects of intron 1 deletion on the randomness of XCI and in the absence of the antisense transcript Tsix in differentiating ESCs, these have little relevance in vivo because mutant mice do not deviate from Mendelian ratios of allele transmission. Altogether, our findings demonstrate that intron 1 is dispensable for the developmental dynamics of Xist expression.

Molecular analyses of human induced pluripotent stem cells and embryonic stem cells.

Recent work from our group and others has argued that human induced pluripotent stem cells (hiPSCs) generated by the introduction of four viruses bearing reprogramming factors differ from human embryonic stem cells (hESCs) at the level of gene expression (Chin et al., 2009). Many of the differences seen were common across independent labs and, at least to some extent, are thought to be a result of residual expression of donor cell-specific genes (Chin et al., 2009; Ghosh et al., 2010; Marchetto et al., 2009). Two new reports reanalyze similar expression data sets as those used in Chin et al. (2009) and come to different conclusions (Newman and Cooper, 2010; Guenther et al., 2010). We compare various approaches to perform gene expression meta-analysis that all support our original conclusions and present new data to demonstrate that polycistronic delivery of the reprogramming factors and extended culture brings hiPSCs transcriptionally closer to hESCs.

Female human iPSCs retain an inactive X chromosome.

Generating induced pluripotent stem cells (iPSCs) requires massive epigenome reorganization. It is unclear whether reprogramming of female human cells reactivates the inactive X chromosome (Xi), as in mouse. Here we establish that human (h)iPSCs derived from several female fibroblasts under standard culture conditions carry an Xi. Despite the lack of reactivation, the Xi undergoes defined chromatin changes, and expansion of hiPSCs can lead to partial loss of XIST RNA. These results indicate that hiPSCs are epigenetically dynamic and do not display a pristine state of X inactivation with two active Xs as found in some female human embryonic stem cell lines. Furthermore, whereas fibroblasts are mosaic for the Xi, hiPSCs are clonal. This nonrandom pattern of X chromosome inactivation in female hiPSCs, which is maintained upon differentiation, has critical implications for clinical applications and disease modeling, and could be exploited for a unique form of gene therapy for X-linked diseases.

Induced pluripotent stem cells and embryonic stem cells are distinguished by gene expression signatures.

Induced pluripotent stem cells (iPSCs) outwardly appear to be indistinguishable from embryonic stem cells (ESCs). A study of gene expression profiles of mouse and human ESCs and iPSCs suggests that, while iPSCs are quite similar to their embryonic counterparts, a recurrent gene expression signature appears in iPSCs regardless of their origin or the method by which they were generated. Upon extended culture, hiPSCs adopt a gene expression profile more similar to hESCs; however, they still retain a gene expression signature unique from hESCs that extends to miRNA expression. Genome-wide data suggested that the iPSC signature gene expression differences are due to differential promoter binding by the reprogramming factors. High-resolution array profiling demonstrated that there is no common specific subkaryotypic alteration that is required for reprogramming and that reprogramming does not lead to genomic instability. Together, these data suggest that iPSCs should be considered a unique subtype of pluripotent cell.

Mitochondrial dysfunction, oxidative stress, and apoptosis revealed by proteomic and transcriptomic analyses of the striata in two mouse models of Parkinson's disease.

The molecular mechanisms underlying the changes in the nigrostriatal pathway in Parkinson's disease (PD) are not completely understood. Here, we use mass spectrometry and microarrays to study the proteomic and transcriptomic changes in the striatum of two mouse models of PD, induced by the distinct neurotoxins 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and methamphetamine (METH). Proteomic analyses resulted in the identification and relative quantification of 912 proteins with two or more unique peptides and 86 proteins with significant abundance changes following neurotoxin treatment. Similarly, microarray analyses revealed 181 genes with significant changes in mRNA, following neurotoxin treatment. The combined protein and gene list provides a clearer picture of the potential mechanisms underlying neurodegeneration observed in PD. Functional analysis of this combined list revealed a number of significant categories, including mitochondrial dysfunction, oxidative stress response, and apoptosis. These results constitute one of the largest descriptive data sets integrating protein and transcript changes for these neurotoxin models with many similar end point phenotypes but distinct mechanisms.

A method for selective enrichment and analysis of nitrotyrosine-containing peptides in complex proteome samples.

Elevated levels of protein tyrosine nitration have been found in various neurodegenerative diseases and age-related pathologies. Until recently, however, the lack of an efficient enrichment method has prevented the analysis of this important low-level protein modification. We have developed a method that specifically enriches nitrotyrosine-containing peptides so that both nitrotyrosine peptides and specific nitration sites can be unambiguously identified with LC-MS/MS. The procedure consists of the derivatization of nitrotyrosine into free sulfhydryl groups followed by high efficiency enrichment of sulfhydryl-containing peptides with thiopropyl sepharose beads. The derivatization process includes: (1) acetylation with acetic anhydride to block all primary amines, (2) reduction of nitrotyrosine to aminotyrosine, (3) derivatization of aminotyrosine with N-Succinimidyl S-Acetylthioacetate (SATA), and (4) deprotection of S-acetyl on SATA to form free sulfhydryl groups. The high specificity of this method is demonstrated by the contrasting percentage of nitrotyrosine-derivatized peptides in the identified tandem mass spectra between enriched and unenriched samples. Global analysis of unenriched in vitro nitrated human histone H1.2, bovine serum albumin (BSA), and mouse brain homogenate samples had 9%, 9%, and 5.9% of identified nitrotyrosine-containing peptides, while the enriched samples had 91% , 62%, and 35%, respectively. Duplicate LC-MS/MS analyses of the enriched mouse brain homogenate identified 150 unique nitrated peptides covering 102 proteins with an estimated 3.3% false discovery rate.

Spatial mapping of protein abundances in the mouse brain by voxelation integrated with high-throughput liquid chromatography-mass spectrometry.

Temporally and spatially resolved mapping of protein abundance patterns within the mammalian brain is of significant interest for understanding brain function and molecular etiologies of neurodegenerative diseases; however, such imaging efforts have been greatly challenged by complexity of the proteome, throughput and sensitivity of applied analytical methodologies, and accurate quantitation of protein abundances across the brain. Here, we describe a methodology for comprehensive spatial proteome mapping that addresses these challenges by employing voxelation integrated with automated microscale sample processing, high-throughput liquid chromatography (LC) system coupled with high-resolution Fourier transform ion cyclotron resonance (FTICR) mass spectrometer, and a "universal" stable isotope labeled reference sample approach for robust quantitation. We applied this methodology as a proof-of-concept trial for the analysis of protein distribution within a single coronal slice of a C57BL/6J mouse brain. For relative quantitation of the protein abundances across the slice, an 18O-isotopically labeled reference sample, derived from a whole control coronal slice from another mouse, was spiked into each voxel sample, and stable isotopic intensity ratios were used to obtain measures of relative protein abundances. In total, we generated maps of protein abundance patterns for 1028 proteins. The significant agreement of the protein distributions with previously reported data supports the validity of this methodology, which opens new opportunities for studying the spatial brain proteome and its dynamics during the course of disease progression and other important biological and associated health aspects in a discovery-driven fashion.

Endogenously nitrated proteins in mouse brain: links to neurodegenerative disease.

Increased abundance of nitrotyrosine modifications of proteins have been documented in multiple pathologies in a variety of tissue types and play a role in the redox regulation of normal metabolism. To identify proteins sensitive to nitrating conditions in vivo, a comprehensive proteomic data set identifying 7792 proteins from a whole mouse brain, generated by LC/LC-MS/MS analyses, was used to identify nitrated proteins. This analysis resulted in the identification of 31 unique nitrotyrosine sites within 29 different proteins. More than half of the nitrated proteins that have been identified are involved in Parkinson's disease, Alzheimer's disease, or other neurodegenerative disorders. Similarly, nitrotyrosine immunoblots of whole brain homogenates show that treatment of mice with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), an experimental model of Parkinson's disease, induces an increased level of nitration of the same protein bands observed to be nitrated in brains of untreated animals. Comparing sequences and available high-resolution structures around nitrated tyrosines with those of unmodified sites indicates a preference of nitration in vivo for surface accessible tyrosines in loops, a characteristic consistent with peroxynitrite-induced tyrosine modification. In addition, most sequences contain cysteines or methionines proximal to nitrotyrosines, contrary to suggestions that these amino acid side chains prevent tyrosine nitration. More striking is the presence of a positively charged moiety near the sites of nitration, which is not observed for non-nitrated tyrosines. Together, these observations suggest a predictive tool of functionally important sites of nitration and that cellular nitrating conditions play a role in neurodegenerative changes in the brain.

Characterization of the mouse brain proteome using global proteomic analysis complemented with cysteinyl-peptide enrichment.

We report a global proteomic approach for analyzing brain tissue and for the first time a comprehensive characterization of the whole mouse brain proteome. Preparation of the whole brain sample incorporated a highly efficient cysteinyl-peptide enrichment (CPE) technique to complement a global enzymatic digestion method. Both the global and the cysteinyl-enriched peptide samples were analyzed by SCX fractionation coupled with reversed phase LC-MS/MS analysis. A total of 48,328 different peptides were confidently identified (>98% confidence level), covering 7792 nonredundant proteins ( approximately 34% of the predicted mouse proteome). A total of 1564 and 1859 proteins were identified exclusively from the cysteinyl-peptide and the global peptide samples, respectively, corresponding to 25% and 31% improvements in proteome coverage compared to analysis of only the global peptide or cysteinyl-peptide samples. The identified proteins provide a broad representation of the mouse proteome with little bias evident due to protein pI, molecular weight, and/or cellular localization. Approximately 26% of the identified proteins with gene ontology (GO) annotations were membrane proteins, with 1447 proteins predicted to have transmembrane domains, and many of the membrane proteins were found to be involved in transport and cell signaling. The MS/MS spectrum count information for the identified proteins was used to provide a measure of relative protein abundances. The mouse brain peptide/protein database generated from this study represents the most comprehensive proteome coverage for the mammalian brain to date, and the basis for future quantitative brain proteomic studies using mouse models. The proteomic approach presented here may have broad applications for rapid proteomic analyses of various mouse models of human brain diseases.

DNA methylation controls the timing of astrogliogenesis through regulation of JAK-STAT signaling.

DNA methylation is a major epigenetic factor that has been postulated to regulate cell lineage differentiation. We report here that conditional gene deletion of the maintenance DNA methyltransferase I (Dnmt1) in neural progenitor cells (NPCs) results in DNA hypomethylation and precocious astroglial differentiation. The developmentally regulated demethylation of astrocyte marker genes as well as genes encoding the crucial components of the gliogenic JAK-STAT pathway is accelerated in Dnmt1-/- NPCs. Through a chromatin remodeling process, demethylation of genes in the JAK-STAT pathway leads to an enhanced activation of STATs, which in turn triggers astrocyte differentiation. Our study suggests that during the neurogenic period, DNA methylation inhibits not only astroglial marker genes but also genes that are essential for JAK-STAT signaling. Thus, demethylation of these two groups of genes and subsequent elevation of STAT activity are key mechanisms that control the timing and magnitude of astroglial differentiation.