Composite Organic-Inorganic Nanoparticles –COINs
Composite Organic-Inorganic Nanoparticles (COINs) with Surface Enhanced Raman Scattering (SERS) properties were fabricated as previously described 
. The COIN clusters are silver nanoparticle aggregates initiated with either heat or salts in the presence of organic Raman dyes. In the context of this report, we elaborate on the fabrication and uses of two different COINs, fabricated using either of the methods with the Raman dyes: Acridine Orange, and Basic Fuchsin.
The AOH COIN was fabricated by heating a 12 nm silver seed in the presence of Acridine Orange reduced with sodium citrate. This process generates random clusters of silver particles with a specific Raman signature (). The BFU COIN was generated by first enlarging the silver seed to 24 nm and inducing aggregation in the presence of sodium chloride (NaCl) and Basic Fuchsin. The BFU COIN clusters are similar to the AOH COIN but have a different Raman signature (). The Raman intensity of spectra for COINs was significantly enhanced by the generation of clusters. Mixing of the silver seed particles with the Raman dye generated colloid silver particles with non-detectable Raman shifts. However, the aggregation of the silver particles into COIN clusters, significantly enhanced the Raman signal intensity by approximately 104–105 fold. (). To determine Raman activity related to COIN cluster size we generated COINs of increasing sizes. The nanoparticle size and polydispersity was determined using photon correlation spectroscopy (PCS: Zetasizer, Malvern). The crude COINs were scanned for their Raman spectra using IRBA (see following paragraph). We found that the intensity of the Raman spectra increases with the size of the COIN particles (). The trend is different for the different COINs. The Raman intensity for the AOH COINs increased abruptly when the mean size grew beyond 50 nm, and the intensity decreased when the particle size grew beyond 80 nm. The increase of the Raman signal for the BFU COIN was moderate but reached optimal intensity between 50–60 nm and decreased beyond that size. The COIN size suitable for bioassays was determined to be 60±6 nm for AOH and 52±5 nm for BFU, where the optimal intensity of the Raman peak was observed for each COIN. Thus, we have generated SERS based COIN nanoparticles that have specific and enhanced Raman shifts.
Characteristics of Composite Organic-Inorganic Nanoparticles COINS.
To reliably detect the Raman signal in a format appropriate for cellular analyses, we developed a automated Raman scanner (Intel Raman BioAnalyser – IRBA) that is suitable for detecting Raman signals (). The schema for the IRBA is illustrated in (). The key components of the microscope are the dichroic filter and notch filter. The dichroic filter allows the laser light to reach the sample, and reflect all other wavelengths. The notch filter blocks the laser light, and transmits all other light wavelengths. The Raman scattering is measured as spectral shifts as little as 30 nm from the excitation laser-light source, hence the slope of the notch filter is high (~90 degrees).
The IRBA scans 64-wells in a microtiter plate-like format. Biological specimens are immobilized on aldehyde glass slides, assembled into a FAST Frame slide holder adopting the 64-well footprint. The sample wells were filled with phosphate buffered saline (PBS), covered with cover glass and loaded into the sample tray holder of the IRBA (see arrow in ). During the scan, samples were probed by a continuous wave, diode-pumped, solid-state laser. IRBA custom software is prompted to automatically focus the laser beam onto the sample using an aspheric objective lens with f/0.5 numerical aperture and a 20× magnification. The laser power at the sample stage is 100 mW, with a laser spot size ~1 µm in diameter. A mechanical shutter reduces the sample exposure to laser light. A typical exposure time is 0.1 seconds per spot. The detector is a back-illuminated, thermoelectrically-cooled CCD camera. The IRBA custom software conducts automated data acquisition of the slide using a user-defined raster scan. The IRBA configuration is set up to collect a single Raman spectrum from a 1 micron spot at a distance of 10 microns with an acquisition time of 100 ms. The IRBA performs a raster scan of the sample containing wells, using a scan matrix of 2×2 up to 20×20 with 100 µm intervals. We tested the optimal raster scan using an AOH COIN solution. We found an increase in the Raman intensity signal with the increase in scan parameters. The optimal results were obtained using a scan matrix of 17×17 matrices with 100 µm intervals (). Thus, the Raman scanner is able to scan a sample plated in a well-chamber. This is applicable to further analysis of cells, as detailed below.
Detection of cell surface antigens using Raman COINs
We tested COINs in immunoassays. We first determined the ability to use COIN nanoparticles to detect surface antigens on single cells stained in suspension. Antibodies were conjugated to COINs. The ability for an antibody-conjugated COIN to function in a bioassay was first determined in an IL-8 ELISA sandwich assay (Figure S1A)
. The IL-8 antibody-COIN conjugate that shows a linear reactivity to IL-8 antigen concentration with a linear slope (r2
>0.8) is considered suitable for use in additional bioassays. Both the AOH and BFU COINs, representing two different fabrication processes, passed the initial control and were considered suitable for use in other biological assays.
To further determine the utility of the COINs as detectors, we performed measurements of surface proteins expressed in the U937 cell line. The U937 cell line is a monocytic leukemia with high ICAM-1 (CD-54 adhesion molecule) expression on the cell surface (,
left & center). The AOH and BFU COINs were conjugated with CD54 antibodies and used to detect the CD54 antigen in an ELISA (Figure S1B)
. Linear regression analysis of COIN signal versus antigen concentration in the ELISA yielded correlation coefficients (r2
) of 0.8–0.99. Thus both the AOH and BFU COINs were found suitable to be used in cell staining to analyze the antigen on the cell surface.
Specificity of COIN based Raman spectroscopy for detection of Surface antigens.
We determined the optimal concentration of the COIN in the surface staining protocol. Increasing concentrations (0.1, 0.25, 0.5 mM) of COIN+αCD54 were incubated with U-937 cells. Excess unbound COIN was washed off and 0.5×106
cells were spun down in the scanning chamber wells. The chamber-containing cells were scanned using IRBA and the 17×17 scan protocol, previously determined as optimal. The average spectrum was calculated for the spectra acquired for each well (). Raman peaks for the COIN signal were defined, and peak heights were calculated (Figure S2)
. The peak heights are displayed as histograms (). We found an increase in BFU-COIN specific peak height with an increase in concentration from 0.1 to 0.25 mM, and a decrease in peak height when COIN concentration increased to 0.5 mM. A similar trend was observed for the AOH COIN (Figure S3)
. We determined that the optimal concentration for COIN staining for further experiments is 0.25 mM.
To determine the accuracy of COINs for detecting specific surface antigens, we first tested the ability of the COINs to bind to CD54 antigen expressed on U937 cells compared to CD8 antigen that is not expressed on U937 cells (). We acquired the spectra for cells stained with antibody conjugated COIN and non-conjugated COIN (,
left). The peak heights for each spectrum were quantitated and represented as histograms (,
right). The Raman peak ratios were determined for the relative Raman peak heights of antibody-conjugated COIN compared to non-conjugated COIN. We detected a specific reactivity of the αCD54-antibody conjugated COIN in U937 cells. Both the AOH and BFU COINs showed similar detection reactivity to CD54 on the surface of U937 cells (Figure S4A)
. To determine the cell-specific binding of the COIN we also stained H82 small cell lung cancer (SCLC) cells that do not express CD54 (,
middle panel). We found specific binding of the αCD54-COIN to CD54 expressing U937 cells but not to H82 cells (). The results using the BFU COIN are comparable to the AOH COIN (Figure S4B)
To visualize the localization of CD54-COIN on the cell surface, we analyzed U937 cells by Scanning Electron Microscopy (SEM). We imaged the cells stained with COIN without additional processing, which is usually required for SEM, by using Quantomix capsules. Using SEM on native samples, we detected clusters of COINs at the apex of U937 cells, which is characteristic for the expression of CD54 ().
To determine the utility of COINs for staining primary human cells, we stained human peripheral blood mononuclear cells (PBMC) with αCD8-conjugated COINs. A subset (~7%) of the total hPBMCs is CD8+ T-cells, as measured by flow cytometry (,
right). We detected a αCD8-COIN signal in PBMC but not in either U937 or H82 cells ()
. To determine if only a subset of the cells reacted to the αCD8-conjugated COIN, we examined each scan for Raman spectra. Approximately 10% of the scans yield Raman spectra correlating with specific COIN signals. This percentage of positive signals compares to the range of cells positive by FLOW cytometry. We repeated the stain, now using AOH COIN. The results received with the BFU COIN were comparable to the AOH COIN (Figure S4C)
We conclude that antibody-conjugated COINs bind specifically to antigens when used for immunostaining of single cells. The intensity of the Raman peak height may vary for each COIN. However, the calculated Raman peak height ratio of the antibody-conjugated COIN compared to non-conjugated COIN was similar for both AOH and BFU. Thus, we have demonstrated the utility of COIN for staining cell lines as well as primary human samples.
Detection of intracellular phosphorylation signaling using Raman COINs
Next, we tested the potential of COIN nanoparticles for the detection of intracellular phosphorylation events. U937 cells activate intracellular signal transduction pathways when treated with IL-4 and IFNγ. Treatment with IL-4 induces the phosphorylation of Stat6, while treatment with IFNγ induces the phosphorylation of Stat1. We first confirmed the increase in phosphorylation of Stat1 and Stat6 by PhosphoFlow analysis (). We measured a 5.9 fold increase of the phosphorylation of pStat1 following IFNγ treatment and 3.3 fold increase in phosphorylation of pStat6 following IL-4 treatment. BFU and AOH COINs were conjugated to antibodies that recognize the Y701 phosphorylated epitope of the Stat1, and the Y641 epitope of the Stat6 proteins. The cells were then fixed and permeabilized as previously described 
Detection of intracellular phosphorylation signaling using COINs.
To prevent non-specific binding of COIN to intracellular proteins, an additional fixation step was carried out. Non-treated and treated cells were stained with antibody-conjugated and non-conjugated COIN washed and scanned using IRBA. The average spectra for IFNγ and IL-4 treated and non-treated cells are shown for AOH-pStat6 () and BFU-pStat1 (). To determine if the COIN itself affects the binding ability, the antibodies were alternated on each COIN. The changes in peak height were determined and the ratio of the Raman signal in treated cells was compared to non-treated cells (). We detected a 5.9 fold change in pStat1 phosphorylation using αpStat1-BFU COIN and a 6.7 fold change using αpStat1-AOH COIN. We detected a 2.9 fold change in pStat6 phosphorylation using αpStat6-BFU COIN and a 2.7 fold change in using αpStat6-AOH COIN. The detected changes in phosphorylation of the Stat1 and Stat6 molecules using the AOH or the BFU COINs was similar to what was observed by PhosphoFlow.
Thus we have demonstrated the utility of COINs for measuring intracellular phosphorylation events in single cells.
Detection of two simultaneous Raman signals using COINs
Ultimately, we determined the ability to conduct intracellular multiplex assays using COINs. A multi-parameter analysis was designed and simultaneous stained cells with AOH and BFU COINs, for detecting two phosphorylation events in a single cell. We co-treated U937 cells with IFNγ and IL-4. We conducted a simultaneous staining of the cells using BFU conjugated to pStat1 and AOH conjugated to pStat6 antibody. We also stained cells with BFU-pStat1, AOH-pStat6 and non-conjugated BFU and AOH COINs as controls. The cells were then scanned by IRBA and the Raman signal intensities detected from the samples are displayed (). We used the “MultiPlex” program (© Intel Corporation) to deconvolute the two Raman spectra detected simultaneously from the BFU and AOH COINs. The representative Raman spectra for each COIN were identified and deconvoluted. We then extracted spectra for untreated cells, treated cells for the pStat1-BFU and pStat6-AOH COINs (). Peak heights representative for each spectrum, were measured and the changes in ratio of the antibody-conjugated COIN peaks were compared to non-conjugated COIN, in treated and non-treated cells (). The results from the double assay were also compared to the single assay in the experimental setup. We measured a 5.4 fold increase in pStat1 in the double stain compared to 5.7 fold change in the single stain experiment. We measured a 3.1 fold increase in pStat6 in the double stain compared to a 2.9 fold change in the single stain experiment. The calculated changes in peak height ratio were statistically similar when we used two COINs simultaneously compared to using a single COIN in a staining assay.
Detection of two intracellular phosphorylation events using two different COINs simultaneously.
To illustrate the robustness of simultaneous staining procedures for phospho-epitopes with COINs, we treated cells with IFNγ (pStat1) or IL-4 (pStat6) or IFNγ/IL-4 (pStat1/pStat6). We stained the cell samples simultaneously with both COINs; the BFU COIN conjugated to pStat1 and the AOH COIN conjugated to pStat6 antibody. The samples were scanned using IRBA. The Raman spectra were deconvoluted using a software package designed for the dataset, termed MultiPlex (©Intel Corporation). The Raman peak heights were calculated and represented as histograms (Figure S5A)
. The peak heights from cells stained with antibody conjugated COINs were normalized to the Raman signal from cells stained with non-conjugated COINs (Figure S5B)
. The Raman signal from cells treated with either IFNγ or IL-4 cytokine is statistically similar to the signal from cells stained with both cytokines simultaneously (p>0.2).
The data demonstrate it is possible to use COINs for the measurements of two simultaneous phosphorylation events in a cell staining assay.