Vanderbilt Institute of Chemical Biology

 

 

Discovery at the VICB

 

 

 

 

 

 

The Power of Multiplexed Activity Metabolomics

 

By: Carol A. Rouzer, VICB Communications
Published:  January 17, 2018

 

This new approach promises to revolutionize bioactive natural product discovery.      

 

An organism's metabolome is the total of all the small molecules that it produces. Each of these molecules has the potential to interact both intra- and extracellularly with the macromolecular components (DNA, RNA, proteins) of the organism itself or those around it. Thus, the metabolome is the foundation for both the regulation of intracellular homeostasis and cell-to-cell communications. Consequently, an effort to search cellular metabolomes for molecules that exhibit interesting and/or useful biological activities is ongoing. However, this effort is frequently thwarted by the complexity of the metabolomes of even the simplest organisms and the heterogeneous nature of the cell populations or tissues to which a potential effector molecule is to be targeted. Clearly the identification of new bioactive metabolites would greatly benefit from a rapid and efficient screening system that enables the evaluation of entire metabolomes against a physiologically relevant primary cell population. Now, Vanderbilt Institute of Chemical Biology members Jonathan Irish and Brian Bachmann introduce a new approach, multiplexed activity metabolomics (MAM) to do just that [D. C. Earl, et al. Nat. Commun., (2018) published online January 2, DOI: 10.1038/s41467-017-02470-8].

 

MAM begins by the identification of a "stimulus organism" that will provide the metabolome to be searched for new effector molecules. The investigator prepares an extract from this organism and then separates it by high performance liquid chromatography, diverting a small portion of the chromatography eluent to a mass spectrometer for analysis, and passing the remainder through a UV/visible diode-array detector before final distribution to the wells of a microtiter plate. The result of this process is a plate containing an array of metabolites with associated data that can be used to identify the contents in wells of interest (Figure 1). In the next step, the researcher obtains cells from the "response organism", adds those cells to the wells of the microtiter plate containing the stimulus organism's metabolome, and incubates them for a desired time period. After the cells are stained for viability, fixed, and permeabilized, they are then stained again with two fluorescent compounds (in this case NHS-Pacific Orange and NHS-Pacific Blue) in combinations of concentrations that are unique for each plate well. This process, called fluorescent cell barcoding (FCB) enables all the cells in a well to be differentiated from those in other wells by the intensities of the two fluorescent labels. The investigator now mixes the cells together and stains them further using fluorescently labeled antibodies that either recognize cell type-specific markers or detect cellular responses to any biologically active metabolites in the array. Critically, each fluorescent label must possess distinct optical properties, so that they can be differentiated from each other. Analysis by flow cytometry with fluorescence detection then provides the data necessary to distinguish one cell type from another, localize it to the correct well, and explore biological responses to the metabolites in the well. The data enable identification of wells containing active metabolites, and the associated spectral information can then be used to tentatively identify compounds of interest in the well (Figure 2).

 

 

 

FIGURE 1. Diagrammatic representation of a MAM experiment. Metabolites extracted from the stimulus organism are separated chromatographically, analyzed by mass spectrometry and UV/visible spectrophotometry, and then dispensed onto a microtiter plate. Cells from the response organism are added to the plate and incubated for a desired time period. The cells are then labeled by fluorescent markers unique to each well (barcoded), pooled, and then further labeled to distinguish cell type and measure responses to metabolite exposure prior to analysis by flow cytometry. Figure reproduced under the Creative Commons Attribution 4.0 International License 4.0 from D. C. Earl, et al. Nat. Commun., (2018) published online January 2, DOI: 10.1038/s41467-017-02470-8.

 

 

 

FIGURE 2. Analysis of response organism cells by MAM. Cells from the response organism are added to the metabolome-containing microtiter plate, incubated for a desired time period, and then fixed and permeabilized. The cells are then labeled by fluorescent markers unique to each well (barcoded), pooled, and then further labeled with fluorescently labeled antibodies that recognize cell type-specific surface markers or proteins that change in response to metabolite exposure. Analysis by flow cytometry provides the data that enable cells to be distinguished on the basis of well location, cell type, and response to metabolites. Figure reproduced under the Creative Commons Attribution 4.0 International License 4.0 from D. C. Earl, et al. Nat. Commun., (2018) published online January 2, DOI: 10.1038/s41467-017-02470-8.


 

To validate the MAM procedure, the Irish and Bachmann labs initially tested the ability of FCB to identify cells in designated wells and to differentiate which cells had been exposed to two known effector molecules. These were etoposide, a topoisomerase inhibitor that causes DNA damage, and staurosporine, an inducer of apoptosis. To carry out this experiment, they placed either etoposide or staurosporine in every other well of a 48 well plate and then added Kasumi-1 cells to all of the wells. Following an overnight incubation, they fixed, permeabilized, barcoded, and then combined the cells. Following additional staining with fluorescent antibodies against γH2AX (a marker of DNA damage) or cleaved caspase-3 (cCasp3, a marker for apoptosis) for cells exposed to etoposide or staurosporine, respectively, they subjected the sample to flow cytometry. The data easily assigned the cells correctly to the wells in which they had been incubated on the basis of the barcoding labels. Furthermore, cells that had been exposed to wells containing etoposide or staurosporine exhibited increased labeling for γH2AX or cCasp3, respectively, as would be expected.
     

To determine if the MAM protocol was sufficiently robust to distinguish cell locations and detect responses to bioeffectors in the presence of a complex metabolome, the investigators next added etoposide and staurosporine to a metabolite extract from a Streptomyces species known to produce no metabolites that induce DNA damage or apoptosis. The Bachmann lab prepared a metabolite array from this extract, and the Irish lab added Kasumi-1 cells and then subjected them to FCB and labeling for γH2AX and cCasp3. Evaluation by flow cytometry not only correctly identified the wells of origin of the cells, it also identified the wells that had contained etoposide and staurosporine on the basis of increased labeling of γH2AX and cCasp3 in the cells from those wells. This was true despite the fact that both compounds were present at concentrations too low to be easily identified in the total ion chromatogram from the metabolome fractionation.
     

Having shown that the presence of the components of a complex metabolome did not interfere with FCB or antibody labeling, the investigators next turned to a search for compounds in microbial microbiomes that are toxic to human tumor cells. For this experiment, they selected as their responder organisms patients suffering from acute myeloid leukemia (AML). This particularly deadly adult cancer carries a poor prognosis, with only 21% of patients surviving for 5 years, so the identification of new therapeutic agents is a serious challenge. AML lends itself particularly well to the MAM approach because cancerous cells from bone marrow biopsies are acquired as suspensions of individual cells that are readily analyzed by flow cytometry. The investigators obtained two such biopsies, one from a 23 year old female bearing an MLL-MLLT3 gene translocation, and the second from a 68 year old male bearing an FMS-like tyrosine kinase 3 internal tandem duplication. The latter genetic mutation is associated with resistance to daunorubicin, an anthracycline chemotherapeutic agent that is often the drug of choice for the initial treatment of AML. For their stimulus organism, the researchers chose Streptomyces specus, a microorganism known to produce anthracycline metabolites similar to but chemically distinct from daunorubicin. After metabolite arrays were prepared from an S. specus extract, the investigators incubated them with cells from their two patients. Following FCB, they stained the cells with Ax700 (to detect viability) and then with antibodies directed against cCasp3 (apoptosis), γH2AX (DNA damage), pS6 (growth, protein synthesis), p-histone H3 (proliferation), and CD45 (general surface marker for leukocytes). After the cells were analyzed by flow cytometry, the investigators used the data to distinguish different cell types, assign cells to the correct microtiter plate well, and assess the effects of the microbial metabolites on cellular function. They found that the active metabolites most strongly affected cell viability, apoptosis, and DNA damage. Some effects were cell type-specific. For example, metabolites in well 17 and 24 of the microtiter plate more strongly affected leukemic blast cells than lymphocytes. Patient-specific effects also occurred. In particular, wells containing anthracycline compounds exerted toxic effects more strongly against cells from the first patient than the second patient, whose leukemic cells carried the anthracycline-resistant mutation. The most bioactive peak in the metabolome array, located in well 21, was the anthracycline specumycin A1 (Figure 3). Strong activity was also found in well 20, and identified as specumycin B1 (Figure 3). An interesting observation was that the levels of activity in the two wells were similar although specumycin B1 was present at a much lower concentration than specumycin A1. This suggests that specumycin B1 may be a more potent anti-tumor agent than its structurally similar analog, specumycin A1. Also of particular interest was the finding that well 24 contained a metabolite with strong activity against patient 2's anthramycin resistant cells. This metabolite's activity was also selective for leukemic over normal cells, suggesting that it could prove to be an interesting new lead for therapeutic discovery.

 

 

 

FIGURE 3. Structure of specumycin A1 and B1. Figure reproduced under the Creative Commons Attribution 4.0 International License 4.0 from D. C. Earl, et al. Nat. Commun., (2018) published online January 2, DOI: 10.1038/s41467-017-02470-8.

 

 

Having successfully applied MAM in a search for natural products from S. specus, the investigators repeated the experiment using Nocardiopsis sp. FU40 as the stimulus organism. They chose this organism because of its ability to produce two classes of biologically active natural products, apoptolidins A-H, and ciromicins A and B (Figure 4). MAM, using cells from the same two leukemic patients described above as the responder organisms, revealed that the apoptolidins primarily induced apoptosis in normal lymphocytes, whereas the ciromicins selectively targeted the leukemic cells, even in the case of those exhibiting anthracycline resistance.

 

 

FIGURE 4. Structure of apoptolidin A, ciromicin A, and ciromicin B. Figure reproduced under the Creative Commons Attribution 4.0 International License 4.0 from D. C. Earl, et al. Nat. Commun., (2018) published online January 2, DOI: 10.1038/s41467-017-02470-8.

The ciromicins are interesting in that ciromicin B is the product of a photochemical reaction of ciromicin A (Figure 4). The researchers isolated larger quantities of these two compounds and tested them directly against AML and normal bone marrow populations using mass cytometry to distinguish individual cell populations with greater granularity. For mass cytometry, the cells were stained using antibodies labeled with different isotopically pure metal chelates. Inductively coupled plasma mass spectrometry, carried out in Vanderbilt's Mass Cytometry Center of Excellence (MCCE, https://my.vanderbilt.edu/mcce/) was then used to detect the presence of the metals. This approach enabled a wider range of different antibodies than fluorescence detection due to the need to avoid fluorophores that have interfering spectral properties. In these experiments, the researchers stained their cells with 29 different antibodies recognizing cell surface markers that differentiate distinct populations of bone marrow and leukemic cells.  In order to cope with the resultant vast amount of data, the Irish and Bachmann labs adapted algorithms that were developed for machine learning.  In particular, to analyze the mass cytometry data, they employed Marker Enrichment Modeling (MEM), which was originally developed as a way to automatically identify different types of cells, and to characterize how cells change over time or in response to a varying condition in the laboratory. In this case, MEM automatically identified 12 distinct cell populations and described how the types of cells changed following exposure to ciromicin A or B. They found that ciromicin B reduced the abundance of leukemic cells, hematopoietic stem cells, and small blast cell subsets. In contrast, ciromicin A caused significant reductions in large blast subsets. Thus, a simple light-mediated chemical reaction caused significant differences in biological activity through the transformation of ciromicin A to ciromicin B. The finding that neither compound had a significant effect on lymphocytes, and that both were active against anthracycline resistant leukemic cells suggests that ciromicins may be interesting molecules for further chemotherapeutic exploration.
     

In summary, the results reveal the power of MAM to identify biologically active natural products in complex metabolomes using heterogeneous mixtures of primary cells and evaluating multiple effects in a single rapid assay. We look forward to seeing what this exciting new tool will yield in the near future!


 

View Nature Communications article: Discovery of human cell selective effector molecules using single cell multiplexed activity metabolomics

 

 

 

 

 

 

TwitterYouTube

 

 

The Vanderbilt Institute of Chemical Biology, 896 Preston Building, Nashville, TN 37232-6304, phone 866.303 VICB (8422), fax 615 936 3884
Vanderbilt University is committed to principles of equal opportunity and affirmative action. Copyright © 2014 by Vanderbilt University Medical Center