Partitioning large-sample microarray-based gene expression profiles using principal components analysis

A framework for longitudinal latent factor modelling of treatment response in clinical trials with applications to Psoriatic Arthritis and Rheumatoid Arthritis

OBJECTIVE

Clinical trials involve the collection of a wealth of data, comprising multiple diverse measurements performed at baseline and follow-up visits over the course of a trial. The most common primary analysis is restricted to a single, potentially composite endpoint at one time point. While such an analytical focus promotes simple and replicable conclusions, it does not necessarily fully capture the multi-faceted effects of a drug in a complex disease setting. Therefore, to complement existing approaches, we set out here to design a longitudinal multivariate analytical framework that accepts as input an entire clinical trial database, comprising all measurements, patients, and time points across multiple trials.

METHODS

Our framework composes probabilistic principal component analysis with a longitudinal linear mixed effects model, thereby enabling clinical interpretation of multivariate results, while handling data missing at random, and incorporating covariates and covariance structure in a computationally efficient and principled way.

RESULTS

We illustrate our approach by applying it to four phase III clinical trials of secukinumab in Psoriatic Arthritis (PsA) and Rheumatoid Arthritis (RA). We identify three clinically plausible latent factors that collectively explain 74.5% of empirical variation in the longitudinal patient database. We estimate longitudinal trajectories of these factors, thereby enabling joint characterisation of disease progression and drug effect. We perform benchmarking experiments demonstrating our method's competitive performance at estimating average treatment effects compared to existing statistical and machine learning methods, and showing that our modular approach leads to relatively computationally efficient model fitting.

CONCLUSION

Our multivariate longitudinal framework has the potential to illuminate the properties of existing composite endpoint methods, and to enable the development of novel clinical endpoints that provide enhanced and complementary perspectives on treatment response.

Genetic discovery and risk characterization in type 2 diabetes across diverse populations

Genomic discovery and characterization of risk loci for type 2 diabetes (T2D) have been conducted primarily in individuals of European ancestry. We conducted a multiethnic genome-wide association study of T2D among 53,102 cases and 193,679 control subjects from African, Hispanic, Asian, Native Hawaiian, and European population groups in the Population Architecture Genomics and Epidemiology (PAGE) and Diabetes Genetics Replication and Meta-analysis (DIAGRAM) Consortia. In individuals of African ancestry, we discovered a risk variant in the TGFB1 gene (rs11466334, risk allele frequency (RAF) = 6.8%, odds ratio [OR] = 1.27, p = 2.06 × 10⁻⁸), which replicated in independent studies of African ancestry (p = 6.26 × 10⁻²³). We identified a multiethnic risk variant in the BACE2 gene (rs13052926, RAF = 14.1%, OR = 1.08, p = 5.75 × 10⁻⁹), which also replicated in independent studies (p = 3.45 × 10⁻⁴). We also observed a significant difference in the performance of a multiethnic genetic risk score (GRS) across population groups (p_{heterogeneity} = 3.85 × 10⁻²⁰). Comparing individuals in the top GRS risk category (40%–60%), the OR was highest in Asians (OR = 3.08) and European (OR = 2.94) ancestry populations, followed by Hispanic (OR = 2.39), Native Hawaiian (OR = 2.02), and African ancestry (OR = 1.57) populations. These findings underscore the importance of genetic discovery and risk characterization in diverse populations and the urgent need to further increase representation of non-European ancestry individuals in genetics research to improve genetic-based risk prediction across populations.

Mass Spectrometry-Based Comprehensive Analysis of Pancreatic Cyst Fluids

Pancreatic cyst fluids (PCFs) enriched in tumour-derived proteins are considered a potential source of new biomarkers. This study aimed to determine compositional and quantitative differences between the degradome and proteome of PCFs aspirated from different types of pancreatic cyst lesions (PCLs). 91 patients who underwent endoscopic ultrasound-fine needle aspiration under routine clinical diagnosis of PCLs were enrolled. Four cysts were malignant (CAs), and 87 were nonmalignant and consisted of 18 intraductal papillary mucinous neoplasms (IPMNs), 14 mucinous cystic neoplasms (MCNs), nine serous cystic neoplasms (SCNs), 29 pseudocysts (PCs), and 17 unclassified. Profiles of the <5 kDa fraction, the degradome, and the trypsin-digested proteome were analysed using an LTQ-Orbitrap Elite mass spectrometer coupled with a nanoACQUITY LC system. Qualitative analyses identified 796 and 366 proteins in degradome and proteome, respectively, and 689 (77%) and 285 (78%) of them were present in the Plasma Proteome Database. Gene Ontology analysis showed a significant overrepresentation of peptidases and peptidases inhibitors in both datasets. In the degradome fraction, quantitative values were obtained for 6996 peptides originating from 657 proteins. Of these, 2287 peptides were unique to a single type, and 515 peptides, derived from 126 proteins, were shared across cyst types. 32 peptides originating from 12 proteins had differential (adjusted p-value ≤0.05, FC ≥1.5) abundance in at least one of the five cysts types. In proteome, relative expression was measured for 330 proteins. Of them, 33 proteins had significantly (adjusted p-value ≤0.05, FC ≥1.5) altered abundance in at least one of the studied groups and 19 proteins appeared to be unique to a given cyst type. PCFs are dominated by blood proteins and proteolytic enzymes. Although differences in PCF peptide composition and abundance could aid classification of PCLs, the unpredictable inherent PCF proteolytic activity may limit the practical applications of PCF protein profiling.

Discovery, fine-mapping, and conditional analyses of genetic variants associated with C-reactive protein in multiethnic populations using the Metabochip in the Population Architecture using Genomics and Epidemiology (PAGE) study

C-reactive protein (CRP) is a circulating biomarker indicative of systemic inflammation. We aimed to evaluate genetic associations with CRP levels among non-European-ancestry populations through discovery, fine-mapping and conditional analyses. A total of 30 503 non-European-ancestry participants from 6 studies participating in the Population Architecture using Genomics and Epidemiology study had serum high-sensitivity CRP measurements and ∼200 000 single nucleotide polymorphisms (SNPs) genotyped on the Metabochip. We evaluated the association between each SNP and log-transformed CRP levels using multivariate linear regression, with additive genetic models adjusted for age, sex, the first four principal components of genetic ancestry, and study-specific factors. Differential linkage disequilibrium patterns between race/ethnicity groups were used to fine-map regions associated with CRP levels. Conditional analyses evaluated for multiple independent signals within genetic regions. One hundred and sixty-three unique variants in 12 loci in overall or race/ethnicity-stratified Metabochip-wide scans reached a Bonferroni-corrected P-value <2.5E-7. Three loci have no (HACL1, OLFML2B) or only limited (PLA2G6) previous associations with CRP levels. Six loci had different top hits in race/ethnicity-specific versus overall analyses. Fine-mapping refined the signal in six loci, particularly in HNF1A. Conditional analyses provided evidence for secondary signals in LEPR, IL1RN and HNF1A, and for multiple independent signals in CRP and APOE. We identified novel variants and loci associated with CRP levels, generalized known CRP associations to a multiethnic study population, refined association signals at several loci and found evidence for multiple independent signals at several well-known loci. This study demonstrates the benefit of conducting inclusive genetic association studies in large multiethnic populations.

Transethnic insight into the genetics of glycaemic traits: fine-mapping results from the Population Architecture using Genomics and Epidemiology (PAGE) consortium

AIMS/HYPOTHESIS

Elevated levels of fasting glucose and fasting insulin in non-diabetic individuals are markers of dysregulation of glucose metabolism and are strong risk factors for type 2 diabetes. Genome-wide association studies have discovered over 50 SNPs associated with these traits. Most of these loci were discovered in European populations and have not been tested in a well-powered multi-ethnic study. We hypothesised that a large, ancestrally diverse, fine-mapping genetic study of glycaemic traits would identify novel and population-specific associations that were previously undetectable by European-centric studies.

METHODS

A multiethnic study of up to 26,760 unrelated individuals without diabetes, of predominantly Hispanic/Latino and African ancestries, were genotyped using the Metabochip. Transethnic meta-analysis of racial/ethnic-specific linear regression analyses were performed for fasting glucose and fasting insulin. We attempted to replicate 39 fasting glucose and 17 fasting insulin loci. Genetic fine-mapping was performed through sequential conditional analyses in 15 regions that included both the initially reported SNP association(s) and denser coverage of SNP markers. In addition, Metabochip-wide analyses were performed to discover novel fasting glucose and fasting insulin loci. The most significant SNP associations were further examined using bioinformatic functional annotation.

RESULTS

Previously reported SNP associations were significantly replicated (p ≤ 0.05) in 31/39 fasting glucose loci and 14/17 fasting insulin loci. Eleven glycaemic trait loci were refined to a smaller list of potentially causal variants through transethnic meta-analysis. Stepwise conditional analysis identified two loci with independent secondary signals (G6PC2-rs477224 and GCK-rs2908290), which had not previously been reported. Population-specific conditional analyses identified an independent signal in G6PC2 tagged by the rare variant rs77719485 in African ancestry. Further Metabochip-wide analysis uncovered one novel fasting insulin locus at SLC17A2-rs75862513.

CONCLUSIONS/INTERPRETATION

These findings suggest that while glycaemic trait loci often have generalisable effects across the studied populations, transethnic genetic studies help to prioritise likely functional SNPs, identify novel associations that may be population-specific and in turn have the potential to influence screening efforts or therapeutic discoveries.

DATA AVAILABILITY

The summary statistics from each of the ancestry-specific and transethnic (combined ancestry) results can be found under the PAGE study on dbGaP here: https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000356.v1.p1.

Variant Discovery and Fine Mapping of Genetic Loci Associated with Blood Pressure Traits in Hispanics and African Americans

Despite the substantial burden of hypertension in US minority populations, few genetic studies of blood pressure have been conducted in Hispanics and African Americans, and it is unclear whether many of the established loci identified in European-descent populations contribute to blood pressure variation in non-European descent populations. Using the Metabochip array, we sought to characterize the genetic architecture of previously identified blood pressure loci, and identify novel cardiometabolic variants related to systolic and diastolic blood pressure in a multi-ethnic US population including Hispanics (n = 19,706) and African Americans (n = 18,744). Several known blood pressure loci replicated in African Americans and Hispanics. Fourteen variants in three loci (KCNK3, FGF5, ATXN2-SH2B3) were significantly associated with blood pressure in Hispanics. The most significant diastolic blood pressure variant identified in our analysis, rs2586886/KCNK3 (P = 5.2 x 10⁻⁹), also replicated in independent Hispanic and European-descent samples. African American and trans-ethnic meta-analysis data identified novel variants in the FGF5, ULK4 and HOXA-EVX1 loci, which have not been previously associated with blood pressure traits. Our identification and independent replication of variants in KCNK3, a gene implicated in primary hyperaldosteronism, as well as a variant in HOTTIP (HOXA-EVX1) suggest that further work to clarify the roles of these genes may be warranted. Overall, our findings suggest that loci identified in European descent populations also contribute to blood pressure variation in diverse populations including Hispanics and African Americans—populations that are understudied for hypertension genetic risk factors.

A Transcriptomic Signature of Mouse Liver Progenitor Cells

Liver progenitor cells (LPCs) can proliferate extensively, are able to differentiate into hepatocytes and cholangiocytes, and contribute to liver regeneration. The presence of LPCs, however, often accompanies liver disease and hepatocellular carcinoma (HCC), indicating that they may be a cancer stem cell. Understanding LPC biology and establishing a sensitive, rapid, and reliable method to detect their presence in the liver will assist diagnosis and facilitate monitoring of treatment outcomes in patients with liver pathologies. A transcriptomic meta-analysis of over 400 microarrays was undertaken to compare LPC lines against datasets of muscle and embryonic stem cell lines, embryonic and developed liver (DL), and HCC. Three gene clusters distinguishing LPCs from other liver cell types were identified. Pathways overrepresented in these clusters denote the proliferative nature of LPCs and their association with HCC. Our analysis also revealed 26 novel markers, LPC markers, including Mcm2 and Ltbp3, and eight known LPC markers, including M2pk and Ncam. These markers specified the presence of LPCs in pathological liver tissue by qPCR and correlated with LPC abundance determined using immunohistochemistry. These results showcase the value of global transcript profiling to identify pathways and markers that may be used to detect LPCs in injured or diseased liver.

Immune cell subsets and their gene expression profiles from human PBMC isolated by Vacutainer Cell Preparation Tube (CPT™) and standard density gradient

Background

High quality genetic material is an essential pre-requisite when analyzing gene expression using microarray technology. Peripheral blood mononuclear cells (PBMC) are frequently used for genomic analyses, but several factors can affect the integrity of nucleic acids prior to their extraction, including the methods of PBMC collection and isolation. Due to the lack of the relevant data published, we compared the Ficoll-Paque density gradient centrifugation and BD Vacutainer cell preparation tube (CPT) protocols to determine if either method offered a distinct advantage in preparation of PBMC-derived immune cell subsets for their use in gene expression analysis. We evaluated the yield and purity of immune cell subpopulations isolated from PBMC derived by both methods, the quantity and quality of extracted nucleic acids, and compared gene expression in PBMC and individual immune cell types from Ficoll and CPT isolation protocols using Affymetrix microarrays.

Results

The mean yield and viability of fresh PBMC acquired by the CPT method (1.16 × 10⁶ cells/ml, 93.3 %) were compatible to those obtained with Ficoll (1.34 × 10⁶ cells/ml, 97.2 %). No differences in the mean purity, recovery, and viability of CD19+ (B cells), CD8+ (cytotoxic T cells), CD4+ (helper T cell) and CD14+ (monocytes) positively selected from CPT- or Ficoll-isolated PBMC were found. Similar quantities of high quality RNA and DNA were extracted from PBMC and immune cells obtained by both methods. Finally, the PBMC isolation methods tested did not impact subsequent recovery and purity of individual immune cell subsets and, importantly, their gene expression profiles.

Conclusions

Our findings demonstrate that the CPT and Ficoll PBMC isolation protocols do not differ in their ability to purify high quality immune cell subpopulations. Since there was no difference in the gene expression profiles between immune cells obtained by these two methods, the Ficoll isolation can be substituted by the CPT protocol without conceding phenotypic changes of immune cells and compromising the gene expression studies. Given that the CPT protocol is less elaborate, minimizes cells’ handling and processing time, this method offers a significant operating advantage, especially in large-scale clinical studies aiming at dissecting gene expression in PBMC and PBMC-derived immune cell subpopulations.

Electronic supplementary material

The online version of this article (doi:10.1186/s12865-015-0113-0) contains supplementary material, which is available to authorized users.

Prospective associations of coronary heart disease loci in African Americans using the MetaboChip: the PAGE study

BACKGROUND

Coronary heart disease (CHD) is a leading cause of morbidity and mortality in African Americans. However, there is a paucity of studies assessing genetic determinants of CHD in African Americans. We examined the association of published variants in CHD loci with incident CHD, attempted to fine map these loci, and characterize novel variants influencing CHD risk in African Americans.

METHODS AND RESULTS

Up to 8,201 African Americans (including 546 first CHD events) were genotyped using the MetaboChip array in the Atherosclerosis Risk in Communities (ARIC) study and Women's Health Initiative (WHI). We tested associations using Cox proportional hazard models in sex- and study-stratified analyses and combined results using meta-analysis. Among 44 validated CHD loci available in the array, we replicated and fine-mapped the SORT1 locus, and showed same direction of effects as reported in studies of individuals of European ancestry for SNPs in 22 additional published loci. We also identified a SNP achieving array wide significance (MYC: rs2070583, allele frequency 0.02, P = 8.1 × 10(-8)), but the association did not replicate in an additional 8,059 African Americans (577 events) from the WHI, HealthABC and GeneSTAR studies, and in a meta-analysis of 5 cohort studies of European ancestry (24,024 individuals including 1,570 cases of MI and 2,406 cases of CHD) from the CHARGE Consortium.

CONCLUSIONS

Our findings suggest that some CHD loci previously identified in individuals of European ancestry may be relevant to incident CHD in African Americans.

The discrimination of interaural level difference sensitivity functions: development of a taxonomic data template for modelling

Background

A major cue for the position of a high-frequency sound source in azimuth is the difference in sound pressure levels in the two ears, Interaural Level Differences (ILDs), as a sound is presented from different positions around the head. This study aims to use data classification techniques to build a descriptive model of electro-physiologically determined neuronal sensitivity functions for ILDs. The ILDs were recorded from neurons in the central nucleus of the Inferior Colliculus (ICc), an obligatory midbrain auditory relay nucleus. The majority of ICc neurons (~ 85%) show sensitivity to ILDs but with a variety of different forms that are often difficult to unambiguously separate into different information-bearing types. Thus, this division is often based on laboratory-specific and relatively subjective criteria. Given the subjectivity and non-uniformity of ILD classification methods in use, we examined if objective data classification techniques for this purpose. Our key objectives were to determine if we could find an analytical method (A) to validate the presence of four typical ILD sensitivity functions as is commonly assumed in the field, and (B) whether this method produced classifications that mapped on to the physiologically observed results.

Methods

The three-step data classification procedure forms the basic methodology of this manuscript. In this three-step procedure, several data normalization techniques were first tested to select a suitable normalization technique to our data. This was then followed by PCA to reduce data dimensionality without losing the core characteristics of the data. Finally Cluster Analysis technique was applied to determine the number of clustered data with the aid of the CCC and Inconsistency Coefficient values.

Results

The outcome of a three-step analytical data classification process was the identification of seven distinctive forms of ILD functions. These seven ILD function classes were found to map to the four “known” ideal ILD sensitivity function types, namely: Sigmoidal-EI, Sigmoidal-IE, Peaked, and Insensitive, ILD functions, and variations within these classes. This indicates that these seven templates can be utilized in future modelling studies.

Conclusions

We developed a taxonomy of ILD sensitivity functions using a methodological data classification approach. The number and types of generic ILD function patterns found with this method mapped well on to our electrophysiologically determined ILD sensitivity functions. While a larger data set of the latter functions may bring a more robust outcome, this good mapping is encouraging in providing a principled method for classifying such data sets, and could be well extended to other such neuronal sensitivity functions, such as contrast tuning in vision.

Transcriptomic profiling of human peritumoral neocortex tissues revealed genes possibly involved in tumor-induced epilepsy

The molecular mechanism underlying tumor-induced epileptogenesis is poorly understood. Alterations in the peritumoral microenvironment are believed to play a significant role in inducing epileptogenesis. We hypothesize that the change of gene expression in brain peritumoral tissues may contribute to the increased neuronal excitability and epileptogenesis. To identify the genes possibly involved in tumor-induced epilepsy, a genome-wide gene expression profiling was conducted using Affymetrix HG U133 plus 2.0 arrays and RNAs derived from formalin-fixed paraffin embedded (FFPE) peritumoral cortex tissue slides from 5-seizure vs. 5-non-seizure low grade brain tumor patients. We identified many differentially expressed genes (DEGs). Seven dysregulated genes (i.e., C1QB, CALCRL, CCR1, KAL1, SLC1A2, SSTR1 and TYRO3) were validated by qRT-PCR, which showed a high concordance. Principal Component Analysis (PCA) showed that epilepsy subjects were clustered together tightly (except one sample) and were clearly separated from the non-epilepsy subjects. Molecular functional categorization showed that significant portions of the DEGs functioned as receptor activity, molecular binding including enzyme binding and transcription factor binding. Pathway analysis showed these DEGs were mainly enriched in focal adhesion, ECM-receptor interaction, and cell adhesion molecules pathways. In conclusion, our study showed that dysregulation of gene expression in the peritumoral tissues may be one of the major mechanisms of brain tumor induced-epilepsy. However, due to the small sample size of the present study, further validation study is needed. A deeper characterization on the dysregulated genes involved in brain tumor-induced epilepsy may shed some light on the management of epilepsy due to brain tumors.

Array-based DNA methylation profiling of primary lymphomas of the central nervous system

Background

Although primary lymphomas of the central nervous system (PCNSL) and extracerebral diffuse large B-cell lymphoma (DLBCL) cannot be distinguished histologically, it is still a matter of debate whether PCNSL differ from systemic DLBCL with respect to their molecular features and pathogenesis. Analysis of the DNA methylation pattern might provide further data distinguishing these entities at a molecular level.

Methods

Using an array-based technology we have assessed the DNA methylation status of 1,505 individual CpG loci in five PCNSL and compared the results to DNA methylation profiles of 49 DLBCL and ten hematopoietic controls.

Results

We identified 194 genes differentially methylated between PCNSL and normal controls. Interestingly, Polycomb target genes and genes with promoters showing a high CpG content were significantly enriched in the group of genes hypermethylated in PCNSL. However, PCNSL and systemic DLBCL did not differ in their methylation pattern.

Conclusions

Based on the data presented here, PCNSL and DLBCL do not differ in their DNA methylation pattern. Thus, DNA methylation analysis does not support a separation of PCNSL and DLBCL into individual entities. However, PCNSL and DLBCL differ in their DNA methylation pattern from non- malignant controls.

Candidate protein biodosimeters of human exposure to ionizing radiation

PURPOSE

To conduct a literature review of candidate protein biomarkers for individual radiation biodosimetry of exposure to ionizing radiation.

MATERIALS AND METHODS

Reviewed approximately 300 publications (1973 - April 2006) that reported protein effects in mammalian systems after either in vivo or in vitro radiation exposure.

RESULTS

We found 261 radiation-responsive proteins including 173 human proteins. Most of the studies used high doses of ionizing radiation (>4 Gy) and had no information on dose- or time-responses. The majority of the proteins showed increased amounts or changes in phosphorylation states within 24 h after exposure (range: 1.5- to 10-fold). Of the 47 proteins that are responsive at doses of 1 Gy and below, 6 showed phosphorylation changes at doses below 10 cGy. Proteins were assigned to 9 groups based on consistency of response across species, dose- and time-response information and known role in the radiation damage response.

CONCLUSIONS

ATM (Ataxia telengiectasia mutated), H2AX (histone 2AX), CDKN1A (Cyclin-dependent kinase inhibitor 1A), and TP53 (tumor protein 53) are top candidate radiation protein biomarkers. Furthermore, we recommend a panel of protein biomarkers, each with different dose and time optima, to improve individual radiation biodosimetry for discriminating between low-, moderate-, and high-dose exposures. Our findings have applications for early triage and follow-up medical assessments.

Proteomic analyses of pancreatic cyst fluids

OBJECTIVES

There are currently no diagnostic indicators that are consistently reliable, obtainable, and conclusive for diagnosing and risk-stratifying pancreatic cysts. Proteomic analyses were performed to explore pancreatic cyst fluids to yield effective diagnostic biomarkers.

METHODS

We have prospectively recruited 20 research participants and prepared their pancreatic cyst fluids specifically for proteomic analyses. Proteomic approaches applied were as follows: (1) matrix-assisted laser-desorption-ionization time-of-flight mass spectrometry peptidomics with LC/MS/MS (HPLC-tandem mass spectrometry) protein identification; (2) 2-dimensional gel electrophoresis; (3) GeLC/MS/MS (tryptic digestion of proteins fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and identified by LC/MS/MS).

RESULTS

Sequencing of more than 350 free peptides showed that exopeptidase activities rendered peptidomics of cyst fluids unreliable; protein nicking by proteases in the cyst fluids produced hundreds of protein spots from the major proteins, making 2-dimensional gel proteomics unmanageable; GeLC/MS/MS revealed a panel of potential biomarker proteins that correlated with carcinoembryonic antigen (CEA).

CONCLUSIONS

Two homologs of amylase, solubilized molecules of 4 mucins, 4 solubilized CEA-related cell adhesion molecules (CEACAMs), and 4 S100 homologs may be candidate biomarkers to facilitate future pancreatic cyst diagnosis and risk-stratification. This approach required less than 40 microL of cyst fluid per sample, offering the possibility to analyze cysts smaller than 1 cm in diameter.

Exometabolic and transcriptional response in relation to phenotype and gene copy number in respiration-related deletion mutants of S. cerevisiae

The transcriptional and metabolic impact of deleting one or both copies of a respiration-related gene has been studied in glucose-limited chemostats. Integration of literature information on phenotype with our exometabolome and transcriptome data enabled the identification of novel relationships between gene copy number, transcriptional regulation and phenotype. We found that the effect of complete respiratory deficiency on transcription was limited to downregulation of genes involved in oxidoreductase activity and iron assimilation. Partial respiratory deficiency had no significant impact on gene transcription. Changes in the copy number of two transcription-factor genes, HAP4 and MIG1, had a major impact on genes involved in mitochondrial function. Regulation of respiratory chain components encoded in the nucleus and mitochondria appears to be divided between Hap4p and Oxa1p, respectively. Similarly, repression of respiration may be imposed by the action of Mig1p and Mba1p on nuclear and mitochondrial gene expression, respectively. However, it is not clear whether Oxa1p and Mba1p regulate mitochondrial gene expression via their interaction with mitochondrial ribosomes or by some indirect means. The phenotype of nuclear petite mutants may not simply be due to the absence of respiration; e.g. Oxa1p or Bcs1p may play a role in the regulation of ribosome assembly in the nucleolus. Integration between respiration and cell growth may also result from the action of a single transcription factor. Thus, Hap4p targets genes that are required for respiration and for fitness in nutrient-limited conditions. This suggests that Hap4p may enable cells to adapt to nutrient limitation as well as diauxy.

One-hit effects in cancer: altered proteome of morphologically normal colon crypts in familial adenomatous polyposis

We studied patients with Familial Adenomatous Polyposis (FAP) because they are virtually certain to develop colon cancer, and because much is known about the causative APC gene. We hypothesized that the inherited heterozygous mutation itself leads to changes in the proteome of morphologically normal crypts and the proteins that changed may represent targets for preventive and therapeutic agents. We determined the differential protein expression of morphologically normal colon crypts of FAP patients versus those of individuals without the mutation, using two-dimensional gel electrophoresis, mass spectrometry, and validation by two-dimensional gel Western blotting. Approximately 13% of 1,695 identified proteins were abnormally expressed in the morphologically normal crypts of APC mutation carriers, indicating that a colon crypt cell under the one-hit state is already abnormal. Many of the expression changes affect pathways consistent with the function of the APC protein, including apoptosis, cell adhesion, cell motility, cytoskeletal organization and biogenesis, mitosis, transcription, and oxidative stress response. Thus, heterozygosity for a mutant APC tumor suppressor gene alters the proteome of normal-appearing crypt cells in a gene-specific manner, consistent with a detectable one-hit event. These changes may represent the earliest biomarkers of colorectal cancer development, potentially leading to the identification of molecular targets for cancer prevention.

Bioactive peptides control receptor for advanced glycated end product-induced elevation of kidney insulin receptor substrate 2 and reduce albuminuria in diabetic mice

BACKGROUND/AIMS

Sixteen-week-old db/db mice exhibit significantly elevated blood glucose and albuminuria. Kidney mesangial cell matrix expansion and collagen IV synthesis correlate with disease progression, but the underlying mechanism is unclear.

METHODS

Adaptive biochemical datasets were generated in cultured 293 kidney cells and in db/db mice.

RESULTS

In animals receiving daily subcutaneous bolus injections (weeks 8-13) of 20 microg/day humanin or 40 microg/day protein kinase C (NPKC) (a PKC-beta2 inhibitor peptide), there was a significant reduction in albuminuria, insulin receptor substrate 2 (IRS-2) and phospho-Akt (Ser473) levels in kidney tissue extracts (p < 0.05 in all cases). Elevated IRS-2 (not IRS-1), altered Akt1 and selective phosphorylation of p-Akt/Ser473 and p-IRS-1/Ser307 (but not p-Akt/Thr308 or p-IRS-2/Ser731) are correlates of the receptor for advanced glycated end product activation and are linked to albuminuria in vivo, whereas in P38 peptide-treated animals, collagen IV synthesis can be uncoupled from albuminuria altogether.

CONCLUSION

Taken together, our results suggest that elevated IRS-2 and altered Akt phosphorylation may be more closely tied to the cause of diabetic kidney disease in db/db mice than mesangial matrix expansion per se, though both may originate from elevated circulatory glucose, and mesangial matrix expansion may independently exacerbate kidney dysfunction.

Comparison of strategies for identification of regulatory quantitative trait loci of transcript expression traits

In order to identify regulatory genes, we determined the heritability of gene transcripts, performed linkage analysis to identify quantitative trait loci (QTLs), and evaluated the evidence for shared genetic effects among transcripts with co-localized QTLs in non-diseased participants from 14 CEPH (Centre d'Etude du Polymorphisme Humain) Utah families. Seventy-six percent of transcripts had a significant heritability and 54% of them had LOD score ≥ 1.8. Bivariate genetic analysis of 15 transcripts that had co-localized QTLs on 4q28.2-q31.1 identified significant genetic correlation among some transcripts although no improvement in the magnitude of LOD scores in this region was noted. Similar results were found in analysis of 12 transcripts, that had co-localized QTLs in the 13q34 region. Principal-component analyses did not improve the ability to identify chromosomal regions of co-localized gene expressions.

Intracellular crowding defines the mode and sequence of substrate uptake by Escherichia coli and constrains its metabolic activity

The influence of the high intracellular concentration of macromolecules on cell physiology is increasingly appreciated, but its impact on system-level cellular functions remains poorly quantified. To assess its potential effect, here we develop a flux balance model of Escherichia coli cell metabolism that takes into account a systems-level constraint for the concentration of enzymes catalyzing the various metabolic reactions in the crowded cytoplasm. We demonstrate that the model's predictions for the relative maximum growth rate of wild-type and mutant E. coli cells in single substrate-limited media, and the sequence and mode of substrate uptake and utilization from a complex medium are in good agreement with subsequent experimental observations. These results suggest that molecular crowding represents a bound on the achievable functional states of a metabolic network, and they indicate that models incorporating this constraint can systematically identify alterations in cellular metabolism activated in response to environmental change.

Multivariate analysis of low-dose radiation-associated changes in cytokine gene expression profiles using microarray technology

OBJECTIVE

The availability of microarray technology, which permits evaluation of the entire cellular transcriptome in a single experiment, has provided new insights on the function of the genome under normal and pathological conditions, as well as in response to genotoxic stimuli, including ionizing radiation. The aims of this study were to: 1) determine whether specific cytokine gene expression profiles can be delineated in individuals exposed to chronic, low-dose radiation; and 2) compare analyses from three multivariate analytic methodologies, hierarchical clustering analysis (HCA), principal component analysis (PCA), and projection pursuit (PP), in evaluating transcriptional responses in human mononuclear cells to low doses of ionizing radiation (IR), as determined by cDNA microarrays.

MATERIALS AND METHODS

Total RNA isolated from mononuclear cells of 19 apparently healthy adult individuals exposed to low doses of IR ranging from 0.18 mSv to 49 mSv over a period of 11 to 13 years, as a result of the Chernobyl Nuclear Power Plant catastrophe, was reverse transcribed in the presence of radioactive dATP to generate radiolabeled complementary cDNA. Target cDNA was hybridized to human cytokine and receptor arrays and mRNA transcriptional patterns were evaluated using HCA, PCA, and PP.

RESULTS

Statistical analyses of the data generated from 19 microarrays revealed distinct gene expression patterns in mononuclear cells of individuals exposed to radiation doses of greater than 10 mSv or less than 10 mSv. Genes encompassed within clusters discerned by HCA, PCA, and PP varied depending on the methodology used to analyze the microarray data. The most frequently expressed genes across all radiation doses were serine/threonine protein kinase receptor (11/19), transforming growth factor (TGF) receptor (11/19), EB13 (10/19), and CD40 ligand.

CONCLUSIONS

Although our findings suggest that it may be possible to assign gene expression profiles to low-dose-irradiated individuals, we show that gene expression profiles vary, depending on the statistical method used to analyze the data. Since there is, as of yet, no consensus regarding the best method to analyze a multivariate dataset, and since discarding the raw data and repeating the experiment at a later date constitutes an unwarranted expenditure, it is important to submit microarray data to public databases where these data can be reevaluated and interpreted by investigators holding expertise in various fields within the scientific community, including radiation biology, statistics, and bioinformatics.

Variability of DNA Microarray Gene Expression Profiles in Cultured Rat Primary Hepatocytes

DNA microarray is a powerful tool in biomedical research. However, transcriptomic profiling using DNA microarray is subject to many variations including biological variability. To evaluate the different sources of variation in mRNA gene expression profiles, gene expression profiles were monitored using the Affymetrix RatTox U34 arrays in cultured primary hepatocytes derived from six rats over a 26 hour period at 6 time points (0h, 2h, 5h, 8h, 14h and 26h) with two replicate arrays at each time point for each animal. In addition, the impact of sample size on the variability of differentially expressed gene lists and the consistency of biological responses were also investigated. Excellent intra-animal reproducibility was obtained at all time points with 0 out of 370 present probe sets across all time points showing significant difference between the 2 replicate arrays (3-way ANOVA, p ≤ 0.0001). However, large inter-animal biological variation in mRNA expression profiles was observed with 337 out of 370 present probe sets showing significant differences among 6 animals (3-way ANOVA, p ≤ 0.05). Principal Component Analysis (PCA) revealed that time effect (PC1) in this data set accounted for 47.4% of total variance indicating the dynamics of transcriptomics. The second and third largest effects came from animal difference, which accounted for 16.9% (PC2 and PC3) of the total variance. The reproducibility of gene lists and their functional classification was declined considerably when the sample size was decreased. Overall, our results strongly support that there is significant inter-animal variability in the time-course gene expression profiles, which is a confounding factor that must be carefully evaluated to correctly interpret microarray gene expression studies. The consistency of the gene lists and their biological functional classification are also sensitive to sample size with the reproducibility decreasing considerably under small sample size.

A biological question and a balanced (orthogonal) design: the ingredients to efficiently analyze two-color microarrays with Confirmatory Factor Analysis

Background

Factor analysis (FA) has been widely applied in microarray studies as a data-reduction-tool without any a-priori assumption regarding associations between observed data and latent structure (Exploratory Factor Analysis).

A disadvantage is that the representation of data in a reduced set of dimensions can be difficult to interpret, as biological contrasts do not necessarily coincide with single dimensions. However, FA can also be applied as an instrument to confirm what is expected on the basis of pre-established hypotheses (Confirmatory Factor Analysis, CFA). We show that with a hypothesis incorporated in a balanced (orthogonal) design, including 'SelfSelf' hybridizations, dye swaps and independent replications, FA can be used to identify the latent factors underlying the correlation structure among the observed two-color microarray data. An orthogonal design will reflect the principal components associated with each experimental factor. We applied CFA to a microarray study performed to investigate cisplatin resistance in four ovarian cancer cell lines, which only differ in their degree of cisplatin resistance.

Results

Two latent factors, coinciding with principal components, representing the differences in cisplatin resistance between the four ovarian cancer cell lines were easily identified. From these two factors 315 genes associated with cisplatin resistance were selected, 199 genes from the first factor (False Discovery Rate (FDR): 19%) and 152 (FDR: 24%) from the second factor, while both gene sets shared 36. The differential expression of 16 genes was validated with reverse transcription-polymerase chain reaction.

Conclusion

Our results show that FA is an efficient method to analyze two-color microarray data provided that there is a pre-defined hypothesis reflected in an orthogonal design.

Exploratory data analysis of DNA microarrays by multivariate curve resolution

In this work, the application of a multivariate curve resolution procedure based on alternating least squares optimization (MCR-ALS) for the analysis of data from DNA microarrays is proposed. For this purpose, simulated and publicly available experimental data sets have been analyzed. Application of MCR-ALS, a method that operates without the use of any training set, has enabled the resolution of the relevant information about different cancer lines classification using a set of few components; each of these defined by a sample and a pure gene expression profile. From resolved sample profiles, a classification of samples according to their origin is proposed. From the resolved pure gene expression profiles, a set of over- or underexpressed genes that could be related to the development of cancer diseases has been selected. Advantages of the MCR-ALS procedure in relation to other previously proposed procedures such as principal component analysis are discussed.

The transcriptome of HCV replicon expressing cell lines in the presence of alpha interferon

We have used DNA microarray analysis of human hepatoma and epithelial carcinoma cells expressing hepatitis C virus (HCV) subgenomic replicons to test whether HCV replication alters gene expression and influences the alpha interferon (IFN-alpha) response. We directly compared the HCV replicon system with a similar system based on a subgenomic replicon of the West Nile virus (WNV) subtype Kunjin virus. We found that in contrast to WNV replicons, persistent replication of HCV replicons did not significantly alter the transcriptome of infected cells nor did it inhibit the nature of the IFN-stimulated genes (ISGs). Our results also provided evidence for the existence of a small number of ISGs that could play a role in the inhibition of HCV replication by IFN-alpha. Finally, we identified ISGs that are activated by the cytokine in a cell-type specific fashion.

A latent variable model for chemogenomic profiling

MOTIVATION

In haploinsufficiency profiling data, pleiotropic genes are often misclassified by clustering algorithms that impose the constraint that a gene or experiment belong to only one cluster. We have developed a general probabilistic model that clusters genes and experiments without requiring that a given gene or drug only appear in one cluster. The model also incorporates the functional annotation of known genes to guide the clustering procedure.

RESULTS

We applied our model to the clustering of 79 chemogenomic experiments in yeast. Known pleiotropic genes PDR5 and MAL11 are more accurately represented by the model than by a clustering procedure that requires genes to belong to a single cluster. Drugs such as miconazole and fenpropimorph that have different targets but similar off-target genes are clustered more accurately by the model-based framework. We show that this model is useful for summarizing the relationship among treatments and genes affected by those treatments in a compendium of microarray profiles.

AVAILABILITY

Supplementary information and computer code at http://genomics.lbl.gov/llda.

Leiomyoma and myometrial gene expression profiles and their responses to gonadotropin-releasing hormone analog therapy

Gene microarray was used to characterize the molecular environment of leiomyoma and matched myometrium during growth and in response to GnRH analog (GnRHa) therapy as well as GnRHa direct action on primary cultures of leiomyoma and myometrial smooth muscle cells (LSMC and MSMC). Unsupervised and supervised analysis of gene expression values and statistical analysis in R programming with a false discovery rate of P < or = 0.02 resulted in identification of 153 and 122 differentially expressed genes in leiomyoma and myometrium in untreated and GnRHa-treated cohorts, respectively. The expression of 170 and 164 genes was affected by GnRHa therapy in these tissues compared with their respective untreated group. GnRHa (0.1 microm), in a time-dependent manner (2, 6, and 12 h), targeted the expression of 281 genes (P < or = 0.005) in LSMC and MSMC, 48 of which genes were found in common with GnRHa-treated tissues. Functional annotations assigned these genes as key regulators of processes involving transcription, translational, signal transduction, structural activities, and apoptosis. We validated the expression of IL-11, early growth response 3, TGF-beta-induced factor, TGF-beta-inducible early gene response, CITED2 (cAMP response element binding protein-binding protein/p300-interacting transactivator with ED-rich tail), Nur77, growth arrest-specific 1, p27, p57, and G protein-coupled receptor kinase 5, representing cytokine, common transcription factors, cell cycle regulators, and signal transduction, at tissue levels and in LSMC and MSMC in response to GnRHa time-dependent action using real-time PCR, Western blotting, and immunohistochemistry. In conclusion, using different, complementary approaches, we characterized leiomyoma and myometrium molecular fingerprints and identified several previously unrecognized genes as targets of GnRHa action, implying that local expression and activation of these genes may represent features differentiating leiomyoma and myometrial environments during growth and GnRHa-induced regression.

Gene expression profiles in the rat streptococcal cell wall-induced arthritis model identified using microarray analysis

Experimental arthritis models are considered valuable tools for delineating mechanisms of inflammation and autoimmune phenomena. Use of microarray-based methods represents a new and challenging approach that allows molecular dissection of complex autoimmune diseases such as arthritis. In order to characterize the temporal gene expression profile in joints from the reactivation model of streptococcal cell wall (SCW)-induced arthritis in Lewis (LEW/N) rats, total RNA was extracted from ankle joints from naïve, SCW injected, or phosphate buffered saline injected animals (time course study) and gene expression was analyzed using Affymetrix oligonucleotide microarray technology (RAE230A). After normalization and statistical analysis of data, 631 differentially expressed genes were sorted into clusters based on their levels and kinetics of expression using Spotfire^® profile search and K-mean cluster analysis. Microarray-based data for a subset of genes were validated using real-time PCR TaqMan^® analysis. Analysis of the microarray data identified 631 genes (441 upregulated and 190 downregulated) that were differentially expressed (Delta > 1.8, P < 0.01), showing specific levels and patterns of gene expression. The genes exhibiting the highest fold increase in expression on days -13.8, -13, or 3 were involved in chemotaxis, inflammatory response, cell adhesion and extracellular matrix remodelling. Transcriptome analysis identified 10 upregulated genes (Delta > 5), which have not previously been associated with arthritis pathology and are located in genomic regions associated with autoimmune disease. The majority of the downregulated genes were associated with metabolism, transport and regulation of muscle development. In conclusion, the present study describes the temporal expression of multiple disease-associated genes with potential pathophysiological roles in the reactivation model of SCW-induced arthritis in Lewis (LEW/N) rat. These findings improve our understanding of the molecular events that underlie the pathology in this animal model, which is potentially a valuable comparator to human rheumatoid arthritis (RA).

The operons, a criterion to compare the reliability of transcriptome analysis tools: ICA is more reliable than ANOVA, PLS and PCA

The number of statistical tools used to analyze transcriptome data is continuously increasing and no one, definitive method has so far emerged. There is a need for comparison and a number of different approaches has been taken to evaluate the effectiveness of the different statistical tools available for microarray analyses. In this paper, we describe a simple and efficient protocol to compare the reliability of different statistical tools available for microarray analyses. It exploits the fact that genes within an operon exhibit the same expression patterns. In order to compare the tools, the genes are ranked according to the most relevant criterion for each tool; for each tool we look at the number of different operons represented within the first twenty genes detected. We then look at the size of the interval within which we find the most significant genes belonging to each operon in question. This allows us to define and estimate the sensitivity and accuracy of each statistical tool. We have compared four statistical tools using Bacillus subtilis expression data: the analysis of variance (ANOVA), the principal component analysis (PCA), the independent component analysis (ICA) and the partial least square regression (PLS). Our results show ICA to be the most sensitive and accurate of the tools tested. In this article, we have used the protocol to compare statistical tools applied to the analysis of differential gene expression. However, it can also be applied without modification to compare the statistical tools developed for other types of transcriptome analyses, like the study of gene co-expression.