Cue-Signal-Response Analysis of TNF-Induced Apoptosis by Partial Least Squares Regression of Dynamic Multivariate Data

Abstract

Biological signaling networks process extracellular cues to control important cell decisions such as death-survival, growth-quiescence, and proliferation-differentiation. After receptor activation, intracellular signaling proteins change in abundance, modification state, and enzymatic activity. Many of the proteins in signaling networks have been identified, but it is not known how signaling molecules work together to control cell decisions. To begin to address this issue, we report the use of partial least squares regression as an analytical method to glean signal-response relationships from heterogeneous multivariate signaling data collected from HT-29 human colon carcinoma cells stimulated to undergo programmed cell death. By partial least squares modeling, we relate dynamic and quantitative measurements of 20-30 intracellular signals to cell survival after treatment with tumor necrosis factor alpha (a death factor) and insulin (a survival factor). We find that partial least squares models can distinguish highly informative signals from redundant uninformative signals to generate a reduced model that retains key signaling features and signal-response relationships. In these models, measurements of biochemical characteristics, based on very different techniques (Western blots, kinase assays, etc.), are grouped together as covariates, showing that heterogenous data have been effectively fused. Importantly, informative protein predictors of cell responses are always multivariate, demonstrating the multicomponent nature of the decision process.

Substrates Elicit Different Patterns of Intracellular Signaling Which in Turn Cause Differences in Cell Adhesion

“…14,16 The matrix used for PCA (see above) was inserted into a matrix of data (x-block). An output matrix (y-block) containing the cell behavior (cell adhesion or b 1 -integrin expression) as the one column and the 10 different ECM substrates as rows are also included.…”

“…5) was correlated with activities of ERK, JNK, IKK and Akt using PLS regression analysis. 7,16 As discussed in the ''Methods'', PLS uses the kinase activity matrix in which the columns are the activities of the 4 kinases at each time-point plus calculated values (mean, maximum, and AUC) and the rows are the 10 substrate compositions. This ''x-block'' is joined with a ''y-block'' representing some output with which the x-block is to be correlated.…”

“…This matrix was analyzed using principal component analysis (PCA), a matrix analysis technique which uses singular value decomposition to isolate Principal Components (PCs) providing information about the correlation of signal activation with ECM substrate. 14,16 Briefly, PCA uses singular value decomposition to decompose a matrix of data into three matrices: scores, eigenvalues, and loadings. 7 The eigenvalue matrix only has values along the diagonal, and these eigenvalues are ordered such that the largest is in the first row/column, second largest in the second row/column, and so forth.…”

Translating Biomaterial Properties to Intracellular Signaling

“…Janes et al [27] provide an excellent example of moving from statistical modeling to classification of groups of signaling responses using PCA and Partial Least Squares Regression Analysis (PLS). Prior to this study it was known that tumor necrosis factor alpha (TNFa), EGF, and insulin exert various influences on cell proliferation and apoptosis through many of the important signaling pathways.…”

Bioengineering and Systems Biology

“…This task raises exceptional difficulties because the connections between regulatory signaling pathways and downstream functional mechanisms are poorly identified at this point in time. Accordingly, in the near-term relational models are likely to be especially productive in relating molecular-level signals to cell-level behavioral responses (e.g., Janes et al 25 ; Sachs et al 38 ).…”