Biomarker discovery for arsenic exposure using functional data. Analysis and feature learning of mass spectrometry proteomic data

Plasma biomarkers of exposure to environmental contaminants play an important role in early detection of disease. The emerging field of proteomics presents an attractive opportunity for candidate biomarker discovery, as it simultaneously measures and analyzes a large number of proteins. This article presents a case study for measuring arsenic concentrations in a population residing in an As-endemic region of Bangladesh using plasma protein expressions measured by SELDI-TOF mass spectrometry. We analyze the data using a unified statistical method based on functional learning to preprocess mass spectra and extract mass spectrometry (MS) features and to associate the selected MS features with arsenic exposure measurements. The task is challenging due to several factors, the high dimensionality of mass spectrometry data, complicated error structures, and a multiple comparison problem. We use nonparametric functional regression techniques for MS modeling, peak detection based on the significant zero-downcrossing method, and peak alignment using a warping algorithm. Our results show significant associations of arsenic exposure to either under- or overexpressions of 20 proteins.     

Plasma Gelsolin Depletion and Circulating Actin in Sepsis—A Pilot Study

Background

Depletion of the circulating actin-binding protein, plasma gelsolin (pGSN) has been described in septic patients and animals. We hypothesized that the extent of pGSN reduction correlates with outcomes of septic patients and that circulating actin is a manifestation of sepsis.

Methodology/Principal Findings

We assayed pGSN in plasma samples from non-surgical septic patients identified from a pre-existing database which prospectively enrolled patients admitted to adult intensive care units at an academic hospital. We identified 21 non-surgical septic patients for the study. Actinemia was detected in 17 of the 21 patients, suggesting actin released into circulation from injured tissues is a manifestation of sepsis. Furthermore, we documented the depletion of pGSN in human clinical sepsis, and that the survivors had significantly higher pGSN levels than the non-survivors (163±47 mg/L vs. 89±48 mg/L, p = 0.01). pGSN levels were more strongly predictive of 28-day mortality than APACHE III scores. For every quartile reduction in pGSN, the odds of death increased 3.4-fold.

Conclusion

We conclude that circulating actin and pGSN deficiency are associated with early sepsis. The degree of pGSN deficiency correlates with sepsis mortality. Reversing pGSN deficiency may be an effective treatment for sepsis.     

Intratumor heterogeneity of biomarker expression in breast carcinomas

Small biopsy samples are used increasingly to assess the biomarker expression for prognostic information and for monitoring therapeutic responses prior to and during neoadjuvant therapy. The issue of intratumor heterogeneity of expression of biomarkers, however, has raised questions about the validity of the assessment of biomarker expression based on limited tissue samples. We examined immunohistochemically the expression of HER-2neu (p185^erbB-2), epidermal growth factor receptor (EGFR), Bcl-2, p53, and proliferating cell nuclear antigen (PCNA) in 30 breast carcinomas using archived, paraffin embedded tissue and determined the extent of intratumor heterogeneity. Each section was divided into four randomly oriented discrete regions, each containing a portion of the infiltrating carcinoma. For each tumor, the entire lesion and four regions were analyzed for the expression of these markers. Scores of both membrane and cytoplasmic staining of HER-2neu and EGFR, scores of cytoplasmic staining of Bcl-2, and scores of nuclear staining of both p53 and PCNA were recorded. The intensity of staining and the proportion of immunostained cells were determined. A semiquantitative immunoscore was calculated by determining the sum of the products of the intensity and corresponding proportion of stained tumor cells. We analyzed both invasive (IDC) and in situ (DCIS) carcinomas. The Wilcoxon signed-rank test was used for paired comparisons between overall and regional immunoscores and between overall and regional percentages of stained cells. Spearman's correlation coefficients were used to assess the level of agreement of overall biomarker expression with each of the regions. Generalized linear models were used to assess overall and pair-wise differences in the absolute values of percent changes between overall and regional expression of biomarkers. For IDCs, there were no statistically significant differences in the expression of the biomarkers in terms of either the percentage of cells staining or the immunoscores when comparing the entire tumor with each region except for the lower EGFR expression of arbitrarily selected region 1 and lower p53 expression of region 1 compared to that of the entire tumor section. For DCIS, there were no statistically significant differences in the expression of the biomarkers between the entire tumor and each region except in PCNA of region 2 compared to that of entire tumor section. Positive correlation of immunoscores was observed between the entire tumor and each region as well as across all four regions for IDC. Similar observations were noted with DCIS except for HER-2neu and PCNA. No statistically significant differences were observed in the absolute values of percent changes of biomarker expression between overall and the four regions for both DCIS and IDC. Therefore, no significant intratumor heterogeneity in the expression of HER-2neu, Bcl-2, and PCNA was observed in IDC. Minor regional variations were observed for EGFR and p53 in IDC. Similarly, no significant regional variation in the expression of markers was observed in DCIS except for PCNA.     

Genetic Susceptible Locus in NOTCH2 Interacts with Arsenic in Drinking Water on Risk of Type 2 Diabetes

Background

Chronic exposure to arsenic in drinking water is associated with increased risk of type 2 diabetes mellitus (T2DM) but the underlying molecular mechanism remains unclear.

Objectives

This study evaluated the interaction between single nucleotide polymorphisms (SNPs) in genes associated with diabetes and arsenic exposure in drinking water on the risk of developing T2DM.

Methods

In 2009–2011, we conducted a follow up study of 957 Bangladeshi adults who participated in a case-control study of arsenic-induced skin lesions in 2001–2003. Logistic regression models were used to evaluate the association between 38 SNPs in 18 genes and risk of T2DM measured at follow up. T2DM was defined as having a blood hemoglobin A1C level greater than or equal to 6.5% at follow-up. Arsenic exposure was characterized by drinking water samples collected from participants'' tubewells. False discovery rates were applied in the analysis to control for multiple comparisons.

Results

Median arsenic levels in 2001–2003 were higher among diabetic participants compared with non-diabetic ones (71.6 µg/L vs. 12.5 µg/L, p-value <0.001). Three SNPs in ADAMTS9 were nominally associated with increased risk of T2DM (rs17070905, Odds Ratio (OR)  = 2.30, 95% confidence interval (CI) 1.17–4.50; rs17070967, OR = 2.02, 95%CI 1.00–4.06; rs6766801, OR = 2.33, 95%CI 1.18–4.60), but these associations did not reach the statistical significance after adjusting for multiple comparisons. A significant interaction between arsenic and NOTCH2 (rs699780) was observed which significantly increased the risk of T2DM (p for interaction = 0.003; q-value = 0.021). Further restricted analysis among participants exposed to water arsenic of less than 148 µg/L showed consistent results for interaction between the NOTCH2 variant and arsenic exposure on T2DM (p for interaction  = 0.048; q-value = 0.004).

Conclusions

These findings suggest that genetic variation in NOTCH2 increased susceptibility to T2DM among people exposed to inorganic arsenic. Additionally, genetic variants in ADAMTS9 may increase the risk of T2DM.     

Effect of prenatal arsenic exposure on DNA methylation and leukocyte subpopulations in cord blood

Prenatal arsenic exposure is associated with increased risk of disease in adulthood. This has led to considerable interest in arsenic’s ability to disrupt fetal programming. Many studies report that arsenic exposure alters DNA methylation in whole blood but these studies did not adjust for cell mixture. In this study, we examined the relationship between arsenic in maternal drinking water collected ≤ 16 weeks gestational age and DNA methylation in cord blood (n = 44) adjusting for leukocyte-tagged differentially methylated regions. DNA methylation was quantified using the Infinium HumanMethylation 450 BeadChip array. Recursively partitioned mixture modeling examined the relationship between arsenic and methylation at 473,844 CpG sites. Median arsenic concentration in water was 12 µg/L (range < 1- 510 µg/L). Log₁₀ arsenic was associated with altered DNA methylation across the epigenome (P = 0.002); however, adjusting for leukocyte distributions attenuated this association (P = 0.013). We also observed that arsenic had a strong effect on the distribution of leukocytes in cord blood. In adjusted models, every log₁₀ increase in maternal drinking water arsenic exposure was estimated to increase CD8+ T cells by 7.4% (P = 0.0004) and decrease in CD4+ T cells by 9.2% (P = 0.0002). These results show that prenatal exposure to arsenic had an exposure-dependent effect on specific T cell subpopulations in cord blood and altered DNA methylation in cord blood. Future research is needed to determine if these small changes in DNA methylation alter gene expression or are associated with adverse health effects.     

Morphological evidence for the existence of a more complex cytoskeleton in Amoeba proteus

Summary Various stabilization and extraction procedures were tested to demonstrate the ultrastructural organization of the cytoskeleton in normal, locomoting Amoeba proteus. Most reliable results were obtained after careful fixation in glutaraldehyde/lysine followed by prolonged extraction in a polyethylene glycol/Triton X-100 solution. Before dehydration in a graded series of ethanol and critical-point drying, the amoebae were split by the sandwich-technique, i.e., by mechanical cleavage of cells mounted between two poly-L-lysine-coated glass slides. Platinum-carbon replicas as well as thin sections prepared from such cell fragments revealed a cytoskeleton composed of at least four different types of filaments: (1) 5–7-nm filaments organized as a more or less ordered cortical network at the internal face of the plasma membrane and probably representing F-actin; (2) 10–12-nm filaments running separately or slightly aggregated through the cytoplasm and probably representing intermediate filaments; (3) 24–26-nm filaments forming a loose network and probably representing microtubules; and (4) 2–4-nm filaments as connecting elements between the other cytoskeleton constituents. Whereas microfilaments are responsible for protoplasmic streaming and other motile phenomena, the function of intermediate filaments and cytoplasmic microtubules in amoebae is still obscure.     

Correlation of global and gene-specific DNA methylation in maternal-infant pairs

The inheritance of DNA methylation patterns is a popular theory to explain the influence of parental genetic and environmental factors on the phenotype of their offspring but few studies have examined this relationship in humans. Using 120 paired maternal-umbilical cord blood samples randomly selected from a prospective birth cohort in Bangladesh, we quantified DNA methylation by pyrosequencing seven CpG positions in the promoter region of p16, four CpG positions in the promoter region of p53, LINE-1 and Alu. Positive correlations were observed between maternal and umbilical cord blood at p16, LINE-1, and Alu but not p53. Multiple linear regression models observed a significant association between maternal and umbilical cord blood at LINE-1 and Alu (LINE-1: β = 0.63, p<0.0001; Alu: β = 0.28, p = 0.009). After adjusting for multiple comparisons, maternal methylation of p16 at position 4 significantly predicted methylation at the same position in umbilical cord blood (β = 0.43, p = <0.0001). These models explained 48%, 5% and 16% of the observed variability in umbilical cord %5mC for LINE-1, Alu and p16 at position 4, respectively. These results suggest that DNA methylation in maternal blood was correlated with her offspring at LINE-1, Alu, and p16 but not p53. Additional studies are needed to confirm whether these observed associations were due to the inheritance of epigenetic events or the shared environment between mother and fetus. Future studies should also use a multi-generational family-based design that would quantify both maternal and paternal contributions to DNA methylation in offspring across more than one generation.     

Expression, Purification, and Inhibitory Properties of Human Proteinase Inhibitor 8

In a previous report, the cDNA for human proteinase inhibitor 8 (PI8) was first identified, isolated, and subcloned into a mammalian expression vector and expressed in baby hamster kidney cells. Initial studies indicated that PI8 was able to inhibit the amidolytic activity of trypsin and form an SDS-stable approximately 67-kDa complex with human thrombin [Sprecher, C. A., et al. (1995) J. Biol Chem. 270, 29854-29861]. In the present study, we have expressed recombinant PI8 in the methylotropic yeast Pichia pastoris, purified the inhibitor to homogeneity, and investigated its ability to inhibit a variety of proteinases. PI8 inhibited the amidolytic activities of porcine trypsin, human thrombin, human coagulation factor Xa, and the Bacillus subtilis dibasic endoproteinase subtilisin A through different mechanisms but failed to inhibit the Staphylococcus aureus endoproteinase Glu-C. PI8 inhibited trypsin in a purely competitive manner, with an equilibrium inhibition constant (Ki) of less than 3.8 nM. The interaction between PI8 and thrombin occurred with a second-order association rate constant (kassoc) of 1.0 x 10(5) M-1 s-1 and a Ki of 350 pM. A slow-binding kinetics approach was used to determine the kinetic constants for the interactions of PI8 with factor Xa and subtilisin A. PI8 inhibited factor Xa via a two-step mechanism with a kassoc of 7.5 x 10(4) M-1 s-1 and an overall Ki of 272 pM. PI8 was a potent inhibitor of subtilisin A via a single-step mechanism with a kassoc of 1.16 x 10(6) M-1 s-1 and an overall Ki of 8.4 pM. The interaction between PI8 and subtilisin A may be of physiological significance, since subtilisin A is an evolutionary precursor to the intracellular mammalian dibasic processing endoproteinases.