Machine learning was applied for the automated derivation of causal influences in cellular signaling networks. This derivation relied on the simultaneous measurement of multiple phosphorylated protein and phospholipid components in thousands of individual primary human immune system cells. Perturbing these cells with molecular interventions drove the ordering of connections between pathway components, wherein Bayesian network computational methods automatically elucidated most of the traditionally reported signaling relationships and predicted novel interpathway network causalities, which we verified experimentally. Reconstruction of network models from physiologically relevant primary single cells might be applied to understanding native-state tissue signaling biology, complex drug actions, and dysfunctional signaling in diseased cells.

Altered growth factor responses in phospho-protein-driven signaling networks are crucial to cancer cell survival and pathology. Profiles of cancer cell signaling networks might therefore identify mechanisms by which such cells interpret environmental cues for continued growth. Using multiparameter flow cytometry, we monitored phospho-protein responses to environmental cues in acute myeloid leukemia at the single cell level. By exposing cancer cell signaling networks to potentiating inputs, rather than relying upon the basal levels of protein phosphorylation alone, we could discern unique cancer network profiles that correlated with genetics and disease outcome. Strikingly, individual cancers manifested multiple cell subsets with unique network profiles, reflecting cancer heterogeneity at the level of signaling response. The results revealed a dramatic remodeling of signaling networks in cancer cells. Thus, single cell measurements of phospho-protein responses reveal shifts in signaling potential of a phospho-protein network, allowing for categorizing of cell network phenotypes by multidimensional molecular profiles of signaling.

The Fluorescence Activated Cell Sorter (FACS) was invented in the late 1960s by Bonner, Sweet, Hulett, Herzenberg, and others to do flow cytometry and cell sorting of viable cells. Becton Dickinson Immunocytometry Systems introduced the commercial machines in the early 1970s, using the Stanford patent and expertise supplied by the Herzenberg Laboratory and a Becton Dickinson engineering group under Bernie Shoor. Over the years, we have increased the number of measured FACS dimensions (parameters) and the speed of sorting to where we now simultaneously measure 12 fluorescent colors plus 2 scatter parameters. In this history, I illustrate the great utility of this state-of-the-art instrument, which allows us to simultaneously stain, analyze, and then sort cells from small samples of human blood cells from AIDS patients, infants, stem cell transplant patients, and others. I also illustrate analysis and sorting of multiple subpopulations of lymphocytes by use of 8–12 colors. In addition, I review single cell sorting used to clone and analyze hybridomas and discuss other applications of FACS developed over the past 30 years, as well as give our ideas on the future of FACS. These ideas are currently being implemented in new programs using the internet for data storage and analysis as well as developing new fluorochromes, e.g., green fluorescent protein and tandem dyes, with applications in such areas as apoptosis, gene expression, cytokine expression, cell biochemistry, redox regulation, and AIDS. Finally, I describe new FACS methods for measuring activated kinases and phosphatases and redox active enzymes in individual cells simultaneously with cell surface phenotyping. Thus, key functions can be studied in various subsets of cells without the need for prior sorting.

No abstract

Leukocyte functional antigen 1 (LFA-1), with intercellular adhesion molecule ligands, mediates T cell adhesion, but the signaling pathways and functional effects imparted by LFA-1 are unclear. Here, intracellular phosphoprotein staining with 13-dimensional flow cytometry showed that LFA-1 activation induced phosphorylation of the beta(2) integrin chain and release of Jun-activating binding protein 1 (JAB-1), and mediated signaling of kinase Erk1/2 through cytohesin-1. Dominant negatives of both JAB-1 and cytohesin-1 inhibited interleukin 2 production and impaired T helper type 1 differentiation. LFA-1 stimulation lowered the threshold of T cell activation. Thus, LFA-1 signaling contributes to T cell activation and effects T cell differentiation.

Intracellular assays of signaling systems have been limited by an inability to correlate functional subsets of cells in complex populations on the basis of active kinase states. Such correlations could be important in distinguishing changes in signaling status that arise in rare cell subsets during functional activation or in disease manifestation. Here we demonstrate the ability to simultaneously detect activated kinase members of the mitogen-activated protein kinases family (p38 MAPK, p44/42 MAPK, JNK/SAPK), members of cell survival pathways (AKT/PKB), and members of T-cell activation pathways (TYK2), among others, in subpopulations of complex cell populations by multiparameter flow-cytometric analysis. We demonstrate the utility of these probes in identifying distinct signaling cascades for (1) both artificial and physiological stimulatory conditions of peripheral blood mononuclear cells (PBMCs), (2) cytokine stimulation in human memory and naïve lymphocyte subsets as identified by five differentiation markers, and (3) ordering of kinase activation in potential signaling hierarchies. Polychromatic flow-cytometric active kinase measurements demonstrate that multidimensional analysis of signaling pathways can provide functional signaling pathway assessment on a single-cell level and allow for potential correlation with biological and clinical parameters.

Patients with the most common and aggressive form of high-grade glioma, glioblastoma multiforme, have poor prognosis and few treatment options. In 2 immunocompetent mouse brain tumor models (CT26-BALB/c and Tu-2449-B6C3F1), we showed that a nonlytic retroviral replicating vector (Toca 511) stably delivers an optimized cytosine deaminase prodrug activating gene to the tumor lesion and leads to long-term survival after treatment with 5-fluorocytosine (5-FC). Survival benefit is dose dependent for both vector and 5-FC, and as few as 4 cycles of 5-FC dosing after Toca 511 therapy provides significant survival advantage. In the virally permissive CT26-BALB/c model, spread of Toca 511 to other tissues, particularly lymphoid tissues, is detectable by polymerase chain reaction (PCR) over a wide range of levels. In the Tu-2449-B6C3F1 model, Toca 511 PCR signal in nontumor tissues is much lower, spread is not always observed, and when observed, is mainly detected in lymphoid tissues at low levels. The difference in vector genome spread correlates with a more effective antiviral restriction element, APOBEC3, present in the B6C3F1 mice. Despite these differences, neither strain showed signs of treatment-related toxicity. These data support the concept that, in immunocompetent animals, a replicating retroviral vector carrying a prodrug activating gene (Toca 511) can spread through a tumor mass, leading to selective elimination of the tumor after prodrug administration, without local or systemic pathology. This concept is under investigation in an ongoing phase I/II clinical trial of Toca 511 in combination with 5-FC in patients with recurrent high-grade glioma ( NCT01156584).

A new microtubule-binding molecule, myoseverin, was identified from a library of 2,6,9-trisubstituted purines in a morphological differentiation screen. Myoseverin induces the reversible fission of multinucleated myotubes into mononucleated fragments. Myotube fission promotes DNA synthesis and cell proliferation after removal of the compound and transfer of the cells to fresh growth medium. Transcriptional profiling and biochemical analysis indicate that myoseverin alone does not reverse the biochemical differentiation process. Instead, myoseverin affects the expression of a variety of growth factor, immunomodulatory, extracellular matrix-remodeling, and stress response genes, consistent with the activation of pathways involved in wound healing and tissue regeneration.