This article is a US government work and, as such, is in the public domain in the United States of America
Secreted proteins play an important role in intercellular interactions, especially between cells of the immune system. Currently, there is no universal assay that allows a simple noninvasive identification and isolation of cells based on their secretion of various products. We have developed such a method. Our method is based on the targeting, to the cell surface, of heterofunctional nanoparticles coupled to a cell surface-specific antibody and to a secreted protein-specific antibody, which captures the secreted protein on the surface of the producing cell. Importantly, this method does not compromise cellviability and is compatible with further culture and expansion of the secreting cells. Published 2012 Wiley-Periodicals, Inc.
Currently, immunohistochemistry and flow cytometry are the two main techniques allowing the identification of individual cells secreting a particular protein. However, because both techniques identify secretory proteins inside the cell, they do not distinguish between cells that actually secrete these proteins from cells that only store them (1–3). Moreover, identification of cells which harbor potentially secreted proteins by flow cytometry requires additional manipulations that include artificial blocking of the secretory pathway to accumulate the secreted protein inside the cell and the permeabilization of the cell membrane (4, 5), and thus compromises cell viability. However, several secretion-capture assays have been reported earlier. These assays rely on encapsulating living individual cells in a matrix, which captures and concentrates secreted proteins, allowing the detection of these proteins on the surface of the secreting cell. These affinity matrixes were made of gels immobilizing antibodies against a secreted protein of interest (6–8). Another approach to identify the secreting cells was developed by Brosterhus et al. (9). It consists of generating bispecific monoclonal antibodies that bind simultaneously to a cell surface antigen and to the secreted protein of interest by coupling large amounts of pure antibodies of different specificities (9). For the vast majority of investigators, these methods are too cumbersome.
Here, we report on a novel, easy, inexpensive, and versatile method, which allows the identification of living cells actually secreting any protein of interest. Our method is based on the targeting, to the cell surface, of nanoparticles, which capture the secreted protein on the surface of the secreting cell. This method allows further characterization of a secreting cell by multicolor flow cytometry and does not compromise cell viability.
Coupling of Magnetic Nanoparticles
A total of 1 mg of carboxyl terminated magnetic iron oxide nanoparticles (MNPs) (Ocean NanoTech, Springdale, AR) of various diameters (15–25 nm) were coupled to 2 mg (50 μl of a 40 mg/ml solution) of goat anti-mouse (GAM) purified antibodies (SouthernBiotech, Birmingham, AL), using the manufacturer's coupling reagents and buffers kit and following the kit's protocol. Briefly, the MNPs are activated using carbodiimide and N-hydroxysuccinimide followed by conjugation to amino groups that are present on the target protein. One mg of MNPs was activated in 400 μl of activation buffer supplemented with 1.7 mM 1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride (EDC) and 0.76 mM N-hydroxysulfosuccinimide (Sulfo-NHS) for 20 min at room temperature. After activation, 500 μl of coupling buffer was added to the particles, immediately followed by the addition 2 mg of the purified antibody. The coupling was allowed to proceed for 2 h in a thermomixer at room temperature with gentle mixing. The reaction was stopped by adding 10 μl of quenching solution and transferred to a 12 × 75 mm2 tube. Two wash-steps with wash/storage buffer were performed using a SuperMAG-01 magnetic separator (Ocean NanoTech) at 4°C. The coupled MNPs were suspended in 4 ml of storage buffer and stored at 4°C to a final concentration of 0.25 mg/ml of iron oxide. The superparamagnetic property of the MNPs is inversely proportional to the particle size. Therefore, while 25 nm show superparamagnetic properties, the cumulative exposure to the high magnetic fields used in the separation steps of the coupling and the formation of Ab-MNPs complexes preparation (see below) may result in magnetization of the MNPs, leading to particle aggregation. Therefore, we recommend using 15 nm MNPs.
Coupling of Carboxyl Quantum Dots
A total of 1 nmol of carboxyl terminated quantum dots with emission 620 nm QD620 (Ocean NanoTech) was conjugated to 1 mg (25 μl of a 40 mg/ml solution) of GAM purified antibodies (Southern Biotech), using the chemicals, buffers, and Qdots provided with the Qdot conjugation kit following the kit's protocol. Briefly, 125 μl QDots was activated by adding 300 μl reaction buffer, after thorough mixing the antibody solution was added, followed by 50 μl of EDC. The reaction was allowed to continue for 2 h with gentle mixing at RT in a thermomixer. The reaction was stopped by addition of 10 μl quenching solution. Excess antibody and salt solution were removed by centrifugation in a 300 KDa molecular weight cutoff concentrator MWCO (PALL, Port Washington, NY), followed by two wash-steps with wash/storage buffer. QDots were resuspended in 1 ml of wash/storage buffer and stored at 4°C
Preparation of Precomplexed MNPs or QDots
Precomplexes were prepared by incubating 100 μl of GAM coupled magnetic particles, or 50 μl GAM coupled QDots, 6 μg total of mouse antibodies (typically 3 μg of anti-cell surface targeting antibody such as anti-CD45 plus 3 μg anti-cytokine antibody) for 30 min at RT with occasional gentle shaking. Antibody-particle complexes were separated from unbound antibody by one of two methods, magnetic columns or MWCO concentrators. Magnetic separation is accomplished by adding particle/antibody solution to equilibrated MACS MS columns in a MACS magnet (Miltenyi Biotec, Auburn, CA).The column is washed three times with MACS buffer (PBS 2 mM EDTA, 0.2% BSA), and particles are collected by removing the column from the magnet and flushing with 200 μl MACS buffer (2× the starting volume of particles). Alternatively particles were separated by centrifugation in 300 KDa MWCO concentrators at 4,000 rpm in microcentrifuge, followed by two wash steps with PBS and then collected in 200 μl of MACS buffer.
Whole blood was obtained from the NIH Blood Bank per their protocol, and peripheral blood monocytes (PBMC) were isolated using lymphocyte separation medium (Lonza, Walkersville, MD). PBMC were placed in culture overnight in RPMI 1640 (Invitrogen) supplemented with gentamicin, fungizone (Invitrogen), and 10% fetal bovine serum (Gemini Bioproducts, West Sacramento, CA). PBMC were either nonactivated or activated for 5 h to 2 days with phorbol 12-myristate 13 acetate-PMA (5 ng/ml) and Ionomycin (500 ng/ml) (Sigma-Aldrich, St Louis, MO). Immediately before use in experiments, cells were stained with Live/Dead Fixable Blue dye (Invitrogen, Life Technology, Grand Island, NY) for 15 min at 4°C and washed with cold PBS.
Particle to Cell Targeting
To demonstrate the ability of the MNPs to bind specifically rather than nonspecifically to cells, we prepared precomplexes of 15 nm GAM-coupled MNPs with an anti-human CD3AlexaFluor 488 (eBioscience, San Diego, CA) Ab, mouse IgG PE (Invitrogen), or both anti-CD3 and msIgG together. A total of 40 μl MNPs-Ab complex was bound to 1 × 106 PBMC in 100 μl PBS with 1% NMS and 1% NGS (Sigma) for 20 min at 4°C, washed with 2 ml cold PBS and spun down at 400g for 5 min to remove any unbound beads, then fixed in 1% paraformaldehyde-PFA (Electron Microscopy Sciences, Hatfield, PA) in PBS.
Cytokine Secretion Assays
PBMC were activated with PMA/ionomycin and evaluated for their ability to secrete IL-2 or IFN-γ after 5 h, MIP-1α and MIP-1β after 18 h, and RANTES after 2 days. Precomplexes were prepared with 15 nm MNPs or Quantum dots and equivalent amounts of mouse anti-human CD45 eFluor450 (eBioscience) as a targeting antibody to bind to all leukocytes and mouse anti-human cytokine capture antibody [IL-2, IFN-γ, MIP-1α, MIP-1β, or RANTES (R&D Systems, Minneapolis, MN)]. A total of 40 μl complexed MNPs were incubated with 1 × 106 cells in 100 μl 1% NMS 1% NGS in PBS for 15 min at 4°C with occasional gentle shaking. The cell/bead complexes were washed with 2 ml cold PBS, spun down, and diluted to 250,000 cells/ml in 4 ml of warm culture medium in a FACS tube and incubated at 37°C for 45 min with constant mixing. After incubation, cells were spun down, washed at 4°C with 2 ml cold PBS, and resuspended in 100 μl of cold 1% NMS 1% NGS in PBS. Cells were incubated with antibodies against cell surface markers (CD45, CD3, CD4, CD8) and fluorescently labeled monoclonal anti-cytokine antibodies for 20 min at 4°C. Cells were washed in 2 ml PBS and fixed in 1% PFA in PBS and analyzed by flow cytometry. Flow cytometry gating strategy involves gating on lymphocytes by FSC vs SSC, doublet discrimination, and live cell selection followed by the identification of CD45+ cells and then CD8 versus anti-cytokine-labeled cells. CD8 is used since CD3 and CD4 are known to be downregulated with polyclonal stimulation of lymphocytes.
Intracellular Cytokine Staining
PBMC activated and nonactivated were treated with GolgiPlug (Invitrogen) for the final 4 h of activation for traditional intracellular cytokine staining. Briefly, 1 × 106 cells in 100 μl 1% NMS, 1% NGS in PBS were incubated with cell surface antibodies for 20 min at RT. Cells were washed with 2 ml PBS, spun down, and resuspended in 100 μl Cytofix (Invitrogen) and fixed for 20 min at RT. Cells were washed with 2 ml PBS, spun down, and resuspended in 100 μl CytoPerm (Invitrogen) with cytokine detection antibodies and incubated at 4°C for 20 min. Cells were washed with 2 ml PBS, spun down, and resuspended in 1% PFA in PBS and analyzed by flow cytometry.
Commercially Available Secretion Assay
IL-2 and IFN-γ secretion were also verified by analyzing both activated and nonactivated PBMC using commercial secretion kits (Miltenyi Biotec). Briefly, 1 × 106 cells were incubated with 10 μl capture antibody for 10 min then diluted in 4 ml of culture medium and incubated at 37°C for 45 min with constant mixing. Cells were spun down and resuspended in 100 μl 1% NMS 1% NGS in PBS and incubated with cell surface marker antibodies and 20 μl cytokine detection antibody for 20 min at 4°C. Cells were washed and fixed in 1% PFA in PBS and analyzed by flow cytometry.
Flow Cytometric Analysis
Samples were acquired on a BD LSRII (Becton Dickinson, San Jose, CA) equipped with five LASERs 355, 405, 488, 532, and 638 nm and with 22 detectors. Data were analyzed with FlowJo software (TreeStar, Ashland, OR). Flow cytometry gating strategy involves gating on lymphocytes by FSC vs. SSC, doublet discrimination, and live cell selection followed by identification of cells fluorescently labeled by antibodies on MNPs.
Data normality was verified by the Shapiro-Wilk W test. When the data did not pass the normality test, the nonparametric procedure of Steel-Dwass for multiple comparison was used to compare distribution across the three groups studied (secretion-capture assay, intracellular staining, and commercial secretion assay), when only two groups were compared (secretion-capture assay and intracellular staining) the Wilcoxon procedure was used. When the data were found normally distributed (IL-2 related data), we used an ANOVA with the Tukey-Kramer HSD (honestly significant difference) correction. All statistical analyses were performed with JMP 9.0 (SAS Institute, Cary, NC).
We coupled carboxylated MNPs of sizes varying from 15 to 25 nm (10) with polyclonal anti-mouse IgG (H+L) goat antibodies (GAM). The superparamagnetic nature of MNPs allows their simple separation, in a magnetic field, from uncomplexed antibodies, while maintaining their colloidal properties (Fig. 1a). We reasoned that if an MNP could bind both a cell-targeting antibody and an antibody specific for a secreted protein (Fig. 1b), they would attach to the cell surface and capture the cell-secreted product, which could then be detected on the surface of the secreting cell. For such a strategy to function, GAM-MNPs, must be bound to two antibodies of different specificities, and be attached to the cell surface via one of these antibodies, which recognizes a cell surface antigen, while the second antibody captures the secreted protein (Fig. 1c).
We demonstrated the feasibility of this strategy by attaching to the cell surface, an isotype control antibody complexed together with a cell surface-specific Ab on GAM-MNPs. Upon incubation with PBMCs, PE-labeled mouse IgG isotype control complexed to 15 nm GAM-MNPs together with an AlexaFluor488-labeled mouse anti-human CD3 mAb was detected on 92 % of CD3 T cells (Fig. 2a). In contrast, the same isotype control IgG complexed with MNPs without anti-human CD3 was found only on 2% of CD3+ cells. This demonstrates that GAM-MNPs can target an irrelevant antibody to the cell surface when it is combined with a cell surface-specific Ab.
Next, we verified that this strategy was suitable for the detection of secreted cytokines on the cell surface. We simulated the secretion of IFN-γ by adding this cytokine into the culture medium bathing PBMCs. A total of 15 nm GAM-MNPs were complexed with anti-IFN-γ and anti-CD45 antibodies. Specifically, we incubated human PBMCs with the purified MNP complexes, and after washing, added recombinant IFN-γ to the cells and washed them 15 min later. IFN-γ was detected with a labeled anti-IFN-γ detection antibody belonging to a different complementation group. We revealed the presence of IFN-γ on the surface of 94.1% of cells. Without MNPs, only 2.9% of cells were IFN-γ positive (Fig. 2b).The latter may represent either cells that adsorbed exogenous IFN-γ or constitutively secrete this cytokine. This experiment confirmed that GAM-MNPs carrying Ab against a cell surface protein and an Ab specific for a secreted product attach to the cell surface and can capture cell-secreted products.
Finally, we verified that complexed GAM-MNPs could indeed capture cell-secreted products. We chose IL-2 and IFN-γ as these two cytokines are produced de novo rather than being prestored inside the cells and thus the results of our assay for evaluating the fraction of secreting cells can be compared with standard intracellular staining. We prepared complexes of GAM-MNPs with anti-CD45 Ab and either anti-IL-2 or anti IFN-γ Ab. These complexes were incubated with human PBMCs, which were activated with a polyclonal activator to stimulate secretion. Using our assay, we found that 16.9 ± 4% and 19.3 ± 3.6% of T cells were secreting IL-2 and IFN-γ, respectively (n = 6) (Fig. 3a). Conventional intracellular staining revealed IL-2 and IFN-γ, respectively, on 16.3 ± 1.4% and on 16.7 ± 6.2% of T cells, showing that for revealing cells that secrete these cytokines the two assays were not statistically different (n = 6, P > 0.8 for both cytokines). We compared our capture assay to commercial IL-2 and IFN-γ capture assays based on the use of hetero-bispecific antibodies, which detected IL-2 on 14.7 ± 1.8% and IFN-γ on 9.3 ± 2.05%, of cells, which were not statistically different from those detected with our assay (P = 0.57 for IL-2 and P = 0.053 for IFN-γ, n = 6) (Fig 3a). These results encouraged us to apply our assay to identify secreting cells that cannot be identified adequately by intracellular staining because they constitutively store the secreted protein. We aimed to identify cells that secrete MIP-1α, MIP-1β, or RANTES. We prepared and incubated with PBMCs MNPs complexed with anti-CD45 antibodies for cell surface targeting and either anti-MIP-1α, anti-MIP-1β, or anti-RANTES antibodies for cytokine capture. The capture assay detected the secretion of each cytokine. MIP-1α was secreted by 19.95 ± 3.5% (n = 6), MIP-1β was secreted by 10 ± 1.8% of cells (n = 6), and RANTES was secreted by 10.94 ± 6.45% of cells (Fig. 3b). Thus, our assay allows the identification of cells that secrete both de novo synthesized or stored proteins.
The capture assay described above is not limited to the use of MNPs. Although MNPs offer an undeniable ease of purification of complexed particles from free antibodies in a magnetic field, their use is not compatible with the magnetic separation of the secreting cells, as every cell is magnetically labeled. As an alternative, we have used carboxylated quantum dots whose size is about 15 nm and we have coupled them to GAM-Ig (H+L). These nanoparticles can be separated from uncoupled GAM antibodies and later from uncomplexed Ab by centrifugation on a Nanosep 300 KDa membrane whose pore opening is 32 nm. Using this approach, the targeted cells are also labeled by the Qdot used to form the capture layer on the cell surface, dispensing the need of using labeled targeting antibody to visualize the cellular targeting. We have performed such an assay for detecting IL-2 secretion (Fig. 3c) and have found that the assay was as good as the MNPs assay.
To overcome the limitations of intracellular cytokine staining, several secretion capture assays have been developed. These assays rely on encapsulating living individual cells in a gel matrix, which captures and concentrates the secreted proteins, allowing their detection on the secreting cell (6). The disadvantages of such gel-based affinity matrixes are multiple. They require a dedicated instrumentation to encapsulate the cells in the gel (7, 8), and they assume that a single cell is encapsulated per gel droplet, an assumption that is not always verifiable. The limitations of these original gel affinity matrixes were overcome by Brosterhus et al. (9), who used bispecific monoclonal antibodies to simultaneously bind to a cell surface antigen and to the secreted protein of interest, resolving the issue of single cell encapsulation. However, for each protein assayed, new hetero-bispecific antibodies have to be generated by coupling large amounts of pure antibodies of different specificities (9), or by recombinant DNA technologies, or by the formation and selection of proper hybridoma heterocaryons (11, 12). For the vast majority of investigators, these methods are too cumbersome and impractical. The paucity of commercially available assays based on hetero-bispecific antibodies is a testimony to the difficulties involved in preparing such reagents. Also, the majority of the assays that have been developed and commercialized by biotechnology companies are limited to proteins of immunological interest.
Although, similar to our approach for forming an affinity matrix to capture secreted proteins to the cell surface, none of the previously developed assays affords the flexibility of our universal matrix. Indeed, the technique we describe above is virtually limitless provided that pairs of antibodies (or ligands) specific for the secreted protein(s) of interest are available even in small quantity. Moreover, the origin of the antibodies used has little influence on the assay, because the nanoparticles can be tailored to capture any type of antibody by coupling the relevant xenotype, isotype, or allotype specific polyclonal antibodies. As is it the case for MHC-peptide tetramers (13), our secretion capture assay can be adapted to the use of streptavidin as a bridging agent between biotinylated targeting and capture antibodies, provided these antibodies bear only one biotin to avoid concatenation. The assay, whose principle is reported here, offers the same degree of freedom in targeting a specific cellular population as it does in measuring the secretion of any protein. By choosing an antibody specific for a given cellular marker, one can study the secretion of proteins by the cell population expressing this cellular marker. However, unlike the commercial assays, the use of MNPs to form the affinity matrix, precludes magnetic sorting of the secreting cells, since every cell, whether it secretes or not the protein of interest, is magnetically labeled. By adapting the principle of the assay to the use of non-magnetic labeled nanoparticle, such as Qdots, secreting cells can be readily identified by the expression of the targeting marker and the secretion of the protein studied. In this case, secreting cells can be magnetically sorted as they are in the commercial secretion kits, by using magnetic beads specific for the label (fluorescent or not) of the detection antibody. Using Qdots as an affinity matrix, offers the additional advantage of performing ratio-metric measurements of the two channels used to monitor cell binding and cell secretion, which may allow the normalization and comparison of protein secretion at the single cell level. However, such application is out of the scope of this report. The versatility of the technique we describe here allows the identification of virtually any cells based on the proteins they secrete. The simplicity of the technique and the ease of its application bring this ability outside of the traditional realm of well-equipped laboratories focused on molecules of immunological interest.
The authors thank L. Margolis for his support and stimulating discussions.