AbstractFluorescence in situ hybridization (FISH) is the primary technology used to image and count mRNA in single cells, but applications of the technique are limited by photophysical shortcomings of organic dyes. Inorganic quantum dots (QDs) can overcome these problems but years of development have not yielded viable QD-FISH probes. Here we report that macromolecular size thresholds limit mRNA labeling in cells, and that a new generation of compact QDs produces accurate mRNA counts. Compared with dyes, compact QD probes provide exceptional photostability and more robust transcript quantification聽due to enhanced brightness.聽New spectrally engineered QDs also allow quantification of multiple distinct mRNA transcripts at the single-molecule level in individual cells.聽We expect that QD-FISH will particularly benefit high-resolution gene expression studies in three dimensional biological specimens for聽which quantification and multiplexing are聽major聽challenges. IntroductionFor a half century, in situ hybridization (ISH) has been used to count, localize, and characterize individual nucleic acids within cells and tissues1,2. Originally developed with radioisotopic labels, the advancement of fluorescence ISH (FISH)3 led to broad adoption in cytogenetics and clinical diagnostics4, today comprising a multi-billion USD industry. Over the past 20 years, FISH has extended to single-RNA analysis5, becoming a standardized method for measuring gene expression at the single-cell level, and enabling the discovery of聽regulatory mechanisms of transcription and translation聽driven by subcellular localization6,7. However signals from organic dye labels used in FISH rapidly deteriorate during photoexcitation, particularly when imaging in three dimensions and under high photon flux needed for super-resolution8. In addition, the technique is limited to simultaneous analysis of ~3 RNA targets due to dye emission spectra overlap, unlike high-throughput ex situ techniques like singe-cell whole transcriptome sequencing that simultaneously probe thousands of transcripts from lysed cell extracts. Creative approaches have increased RNA FISH throughput using repeated cycles of labeling, imaging, and label depletion9, but the methodologies are laborious and challenging to adopt for non-specialists.It is widely anticipated that in situ techniques requiring stable, multiplexed probes will substitute dyes with nanocrystalline quantum dots (QDs) due to their extremely stable and intense emission and vastly expanded multiplexing capabilities deriving from narrow emission bands tunable across the ultraviolet, visible, and infrared spectra10. But despite concerted efforts, considerable industry investment, and broad use in solution-based assays, QDs have not been widely used in FISH protocols. Presumably this is due to inaccurate labeling resulting from the聽large sizes (15鈥?5鈥塶m) of commercially available probes11,12, which cannot transport into crowded macromolecular environments of fixed cells to聽densely label targets. To determine the conditions under which QDs can be applied for accurate counting of mRNA transcripts,聽rigorous quantification must be applied using well-controlled cellular expression systems聽together with聽direct comparisons to聽standardized analytical聽techniques.Here we confirm that critical thresholds for cytoplasmic sieving limit RNA FISH and that a new generation of compact and stable QDs can overcome steric hindrance problems to match labeling accuracies of organic dyes. We show that QD-FISH provides improved signal stability, improved fidelity of molecular counting, and the capacity for multiplexed RNA quantification聽at the single-molecule level.ResultsImpact of QD size on mRNA labelingWe generated a series of QDs coated with multidentate polymers that allow the total hydrodynamic diameter of the probe to be as small as ~7鈥塶m. These products are stable as off-the-shelf materials for years and are azide-functional for facile conjugation to proteins and nucleic acids through high-precision click-chemistry13. Fig.聽1a shows representative FISH images of HeLa cells stained for transcripts of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), comparing ~1鈥塶m organic dyes or QDs with compact (13.3鈥塶m) or large (17.4鈥塶m) hydrodynamic diameters, all using the same oligonucleotide probe sequences based on the Raj et al. multiple labeling method8. Similar single-molecule counts are observed for dyes (425) and compact QDs (487), whereas counts for large QDs (75) are much lower. RNA counts per cell for a range of probes are quantified as scatter plots in Fig.聽1b. We synthesized these QDs as HgxCd1鈭?i>xSe/CdyZn1鈭?i>yS core/shell structures with a wide range of diameters of 3.3, 5.7, and 8.7鈥塶m (Fig.聽1c), all with emission in the red spectrum (Supplementary Fig.聽1), tuned by the聽core alloy composition parameter x14. After polymer coating, the hydrodynamic diameters of the respective aqueous QDs were 9.2鈥塶m (QD9.2), 13.3鈥塶m (QD13.3), and 17.4鈥塶m (QD17.4) measured by protein-calibrated gel permeation chromatography (GPC, Fig.聽1c). Both QD9.2 and QD13.3 yielded GAPDH mRNA counts that were similar to those of dyes (p鈥?gt;鈥?.05; Student鈥檚 t-test), whereas counts using QD17.4 labels were significantly lower (p鈥?lt;鈥?.001). An alternative large QD variant from a commercial vendor (QDcom) with dissimilar surface chemistry (PEG-coated amphiphilic polymers) likewise under-labeled RNA targets (p鈥?lt;鈥?.001).Fig. 1Fluorescence in situ hybridization (FISH) using dye labels or quantum dot (QD) labels with diverse sizes. Data show HeLa cells stained for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) transcripts. a Schematics show RNA target labeling density; representative 3D deconvolved epifluorescence images show cells in two orthogonal orientations, using dyes, small QDs (13.3鈥塶m), or big QDs (17.4鈥塶m). Scale bar鈥?鈥?鈥壩糾. b FISH transcript counts (2D) using dyes, custom designed QDs with three hydrodynamic diameters (9.2, 13.3, or 17.4鈥塶m), or commercially available QDs (com.). Asterisks indicate: p鈥夆墹鈥?.05 (*), p鈥夆墹鈥?.01 (**), and p鈥夆墹鈥?.001 (***); Student鈥檚 t-test. N鈥?鈥?5. Comprehensive statistical comparisons are provided in Supplementary Table 2.聽c Gel permeation chromatograms and TEM images (with core size) of the four QDs from panel b. Scale bar鈥?鈥?0鈥塶m. d Intensity histograms of FISH spots for dyes, small QDs (13.3鈥塶m), and big QDs (17.4鈥塶m) are shown in black compared with histograms of single-fluorophore intensities in white. e FISH counts after different times of laser excitation, comparing stability of QD13.3 and dyes, including representative images. Scale bars鈥?鈥?0 渭m. N鈥?鈥?5. f Correlation between FISH counts in 2D and 3D images for QD13.3 and dye labels. Comprehensive statistical comparisons are provided in Supplementary Table聽1. g Impact of customized blocking conditions on specific and nonspecific labeling. Nonspecific labeling counts (2D) for QDs were statistically the same as those of background when applying both 5% bovine serum albumin (BSA) and 0.125鈥?dextran sulfate (DS). N鈥?鈥?5. All error bars represent s.dFull size imagePresumably the majority of GAPDH transcripts are located in crowded cytosolic regions inaccessible to QDs larger than 13.3鈥塶m. We specifically chose GAPDH as a target due to high expression with distribution throughout the heterogeneous cytoplasm. We observed that the transcripts that could be labeled by QD17.4 were also bound to fewer fluorophores compared with QD13.3 and dyes. Fig.聽1d shows fluorescence intensities per labeled RNA, compared with single fluorophores, showing that RNAs were labeled with a mean of 8.0 dyes, 10 QD13.3, and 2.3 QD17.4 (Supplementary Table聽3). The low labeling density for QD17.4 shows that steric hindrance is limiting even for the most accessible RNAs, consistent with reports demonstrating substantially greater obstruction of cytosolic diffusion for ~16鈥塶m pentamers of green fluorescence protein (GFP) than ~11鈥塶m GFP trimers15. For the ensuing work below, we exclusively use QD13.3.Photostability comparisonsPhotostability is substantially improved for QD-FISH compared with Dye-FISH. Fig.聽1e shows that counts rapidly diminish during photoexcitation of Dye-FISH labeled cells, reducing by ~12% in 30鈥塻 and to nearly zero counts in 10鈥塵in. This is significant because tens of seconds are needed to acquire a full z-stack for 3D cell imaging. In comparison, QDs exhibit long-term stability with no significant change in GAPDH mRNA counts (p鈥?gt;鈥?.05; Student鈥檚 t-test) after 12鈥塵in of excitation, which is consistent with previous results for QD-based stains measured by net intensity16. QD-FISH yielded a significantly higher 3D count number compared with Dye-FISH (by 15鈥?0%) for cells with the same 2D counts at a single nuclear focal plane (Fig.聽1f). This result likely derives from the rapid decline in dye signal, and is the origin of the lower measured labeling density per transcript for dyes compared with QD13.3 (Fig.聽1d), which could likely be improved with specialized anti-fade media and dyes optimized for photostability.Spot counting fidelityThe ability to identify puncta corresponding to individual molecules through automated algorithms is also significantly improved with QDs compared with dyes. Numerous image analysis algorithms have been developed to recognize individual fluorescently labeled molecules as diffraction-limited spots, each of which invariably applies a hypothesis test to decide whether a spot should be categorized as positive or negative, typically corresponding to a signal-to-noise threshold for a fit to a two-dimensional Gaussian function. The imposed threshold usually requires ad hoc empirical adjustments through human intervention8. Spot counting using a scanning window method with serial image depletion (Multiple Target Tracking algorithm17) is shown in Fig.聽2a, b for Dye-FISH and QD-FISH, respectively, both using the same probe sequences. The x-axis of each plot shows the threshold imposed for spot detection based on the statistical fit of each image spot to a point spread function (described further in Methods). The slope of each positive count curve is plotted in panel c. Compared with Dye-FISH, the curve is much flatter for QD-FISH, indicating a lower sensitivity to threshold selection that is critical for robust automation to eliminate manual selection biases. This outcome derives from the higher brightness of QDs that is far above the spatially variable autofluorescence background (Fig.聽2d), whereas the dye channel is highly convolved with autofluorescence and yields a widely varying spot brightness (Fig.聽1d).Fig. 2Computational identification of mRNA spots. a Spot counts in individual cells using Dye-FISH (blue) or cells without labels (red) for different spot detection thresholds (described further in Methods). Shading indicates s.d. of counts between cells. N鈥?鈥?. b Spot counts in individual cells using QD-FISH (blue), cells with QDs added but no probe oligonucleotides (purple), or cells without labels (red), for different detection thresholds. Shading indicates s.d. of counts between cells. N鈥?鈥?. c Slopes of positive counts plotted against detection threshold. d Representative 2D images of Dye-FISH and QD-FISH are shown on the left, next to calculated images showing the locations of detected spots in white for each of the threshold values indicated above the images. The higher stability of detection for QD-FISH is evident from the similar numbers of detected spots for each of the threshold values spanning 25鈥?5, compared with a wider range of detected spots for Dye-FISH. Inset numbers, n, indicate the number of detected spotsFull size imageIn the preceding work, we used a 2-step labeling approach in which QDs attached to streptavidin (SAv) label biotinylated nucleic acid probes pre-hybridized to RNA targets. This was necessary for direct comparison of QD with different sizes, as current QD-nucleic acid conjugates have size-dependent valencies. However, the results are statistically the same when comparing this 2-step labeling process using QD13.3 with 1-step direct labeling in which the same QDs are conjugated with oligonucleotides before addition to cells (Supplementary Fig.聽2鈥?a data-track=\"click\" data-track-label=\"link\" data-track-action=\"supplementary material anchor\" href=\"/articles/s41467-018-06740-x#MOESM1\">3). The 2-step process also allowed identification of specialized blocking conditions to eliminate nonspecific binding of QDs, which, as solid-phase colloids, have a propensity to adsorb to cellular structures. We used an iterative optimization process (Fig.聽1g and Supplementary Fig.s聽4鈥?a data-track=\"click\" data-track-label=\"link\" data-track-action=\"supplementary material anchor\" href=\"/articles/s41467-018-06740-x#MOESM1\">6) to find that both bovine serum albumin (BSA) and polyanions (dextran sulfate; DS) reduce nonspecific binding, and their combination has an additive effect, virtually eliminating nonspecific binding when used together. We attribute the success of this blocking cocktail to the elimination of both denatured hydrophobic domains of proteins (by BSA) and polycationic sites (by DS), both of which can adsorb QDs. However QDs could not be mixed directly with DS due to colloidal aggregation, necessitating sequential blocking. Notably DS requires precise concentration control, having diminishing effect at concentrations greater than 0.125%.Validation of labeling accuracyTo measure the extent to which exact levels of mRNA can be measured in single cells with QD-FISH, we modulated transcript numbers聽in cultured cells using short interfering RNA (siRNA). We focus on the transcript of tumor suppressor phosphatase and tensin homolog (PTEN), a key tumor suppressor gene often deleted in prostate cancer in association聽with a poor prognosis18,19. Representative images for benign prostate hyperplasia (BPH-1) epithelial cells are shown in Fig.聽3a, for which an average 75% reduction in PTEN RNA counts was measured after siRNA treatment (Fig.聽3b), a magnitude similar to that measured at the population level by quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) (Fig.聽3c). We performed the same analysis in VCaP prostate cancer cells (Fig.聽3d), which are notably much smaller than BPH-1 cells. We again observed a similar magnitude of siRNA-induced聽transcript knockdown through QD-FISH (Fig.聽3e) as that observed at the population level with qRT-PCR (Fig.聽3f). This outcome聽is important because mRNA FISH results are challenging to correlate across cell types with different sizes due to differing degrees of spatial overlap of fluorescent spots20.Fig. 3QD-FISH analysis of phosphatase and tensin homolog (PTEN) transcripts in prostate cancer cell lines. Representative images show BPH-1 cells (a) and VCap cells (d) with or without treatment by siRNA to knock down PTEN expression, or using a scrambled siRNA sequence. Scale bar鈥?鈥?0 渭m. Single-cell QD-FISH counts are shown for (b) BPH-1 cells and (e) VCaP cells, in comparison with transcript measurements by population qRT-PCR for (c) BPH-1 cells and (f) VCap cells. Significantly reduced mRNA levels are observed for anti-PTEN siRNA treatment, with a similar magnitude between QD-FISH and qRT-PCR. Asterisks indicate: p鈥夆墹鈥?.05 (*), p鈥夆墹鈥?.01 (**), and p鈥夆墹鈥?.001 (***); Student鈥檚 t-test. All error bars represent鈥塻.d. N鈥?鈥?5 for (b) and (e), N鈥?鈥? for (c) and (f). Comprehensive statistical comparisons are provided in Supplementary Tables聽4鈥?a data-track=\"click\" data-track-label=\"link\" data-track-action=\"supplementary material anchor\" href=\"/articles/s41467-018-06740-x#MOESM1\">7. PTEN mRNA FISH probe sequences are shown in Supplementary Table聽8Full size imageMultiplexed QD-FISHFinally, we validated the ability聽to use QDs for multiplexed quantification of multiple mRNA sequences at the single-molecule level. We synthesized three QDs (QD608, QD693, and QD800) with spectrally distinct emission bands (Fig.聽4a) by tuning the composition of the HgxCd1鈭?i>xSe alloy聽domain, which has little impact on QD size but a substantial impact on electronic bandgap21. The three QDs were compact and similar in hydrodynamic size after coating with multidentate polymers by GPC (Supplementary Fig.聽7). Each QD was conjugated to oligonucleotides complementary to either GAPDH (QD608), PTEN (QD693), or A20 (QD800) mRNA and were then mixed and applied simultaneously to LNCaP prostate cancer cells using the 1-step QD-FISH protocol. Expression levels were modulated by treating the cells with either anti-PTEN siRNA to knock down PTEN expression, or tumor necrosis factor alpha (TNF-伪) to selectively induce A20 gene expression22, each validated by qRT-PCR in Fig.聽4b. A representative image from each QD-FISH color channel is shown for individual cells in each experimental group in Fig.聽4c, and corresponding transcript counts are shown in panel d. By QD-FISH, GAPDH levels were similar in all three experimental groups, whereas PTEN transcript level decreased by ~80% with anti-PTEN siRNA treatment, and A20 expression significantly increased with TNF-伪 treatment. These single-cell results correlated well with the population-level qRT-PCR results.Fig. 4Multiplexed QD-FISH quantification of three transcripts in single LNCaP cells. The three transcripts include GAPDH, PTEN, and A20. a Emission spectra of QD608, QD693, and QD800 and corresponding emission bandpass filters used for imaging. b Graph shows the expressions of PTEN and A20 transcripts relative to GAPDH by qRT-PCR after cells were exposed to either anti-PTEN siRNA to knock down PTEN expression or TNF-伪 to induce A20 expression. Asterisks indicate: p鈥夆墹鈥?.05 (*), p鈥夆墹鈥?.01 (**), and p鈥夆墹鈥?.001 (***); Student鈥檚 t-test. N鈥?鈥?. c Representative QD-FISH images of single cells in each color channel corresponding to bandpass filters shown in panel a. QD probes were specific against GAPDH (QD608), PTEN (QD693) or A20 (QD800), and cells were treated with or without anti-PTEN siRNA or TNF-伪. Scale bar鈥?鈥?0 渭m. d Single-cell QD-FISH results are shown for GAPDH, PTEN and A20 transcripts for each of the three experimental conditions corresponding to the images in panel c. N鈥?鈥?5. All error bars represent s.d. A20 mRNA FISH probe sequences are shown in Supplementary Table聽9Full size imageDiscussionIn this work, we identified a critical size threshold limiting the accuracy of RNA labeling in cells and showed that new QDs with compact sizes can label mRNA targets to yield similar counts as those measured with dyes. The ability to tune QD crystalline size independently from fluorescence emission is a key capability of these materials that allowed us to directly measure the impact of probe size on bioanalytical performance without interference from substantial photophysical mismatch. We anticipate that QD-FISH will drastically聽 improve single-molecule FISH studies in thick samples for which repetitive excitation leads to rapid signal deterioration and when autofluorescence limits the accuracy of single-molecule identification. These materials should be well suited for studies requiring high-level multiplexing, as the multispectral tunability of QDs is greater any other current fluorescent probe and can exhibit high efficiency emission in the first and second near-infrared spectra where cellular autofluorescence is negligible and detectors have now become affordable23,24. New QD engineering approaches have recently become available to widely tune emission wavelengths without changing the size, which is necessary to maintain total minimum dimensions for accurate labeling14. The same physicochemical design rules can further be extended to nanoparticle labels such as rare-earth up-conversion materials and responsive plasmonic materials compatible with unique imaging modalities.MethodsReagentsAll chemical reagents were purchased from Sigma-Aldrich or Alfa Aesar unless otherwise specified. QDcom was purchased from Thermo-Fisher Scientific as Qdot鈩?605 Streptavidin Conjugate.QD9.2 synthesisQD9.2 with emission at 645鈥塶m and 3.3鈥塶m diameter by electron microscopy was synthesized with a core/shell HgCdSe/CdZnS structure using methods similar to those previously reported14. CdSe cores (2.3鈥塶m) were synthesized using a high-temperature injection reaction between cadmium oxide (0.6鈥塵mol), diphenylphosphine selenide (0.2鈥塵mol), and trioctylphosphine selenide (3鈥塵mol) in tetradecylphosphonic acid (1.33鈥塵mol), hexadecylamine (7.1鈥塯), trioctylphosphine (7鈥塵L), and 1-octadecene (ODE; 27.6鈥塵L) at 300鈥壜癈 for 30鈥塻. After purification, cadmium was partially exchanged with mercury using mercury octanethiolate in oleylamine to yield a HgCdSe core. After purification, a 2.2-monolayer shell of CdZnS was deposited layer-by-layer in 0.8-monolayer increments using cadmium acetate in oleylamine (0.1鈥塎), zinc acetate in oleylamine (0.1鈥塎), and elemental sulfur in ODE (0.1鈥塎) as shell stock solutions. The shell composition comprised 0.8 monolayers of Cd0.5Zn0.5S, 1.2 monolayers of Cd0.2Zn0.8S, and 0.2 monolayers of ZnS.QD13.3 synthesisQD13.3 with emission at 605鈥塶m and 5.7鈥塶m diameter by electron microscopy was synthesized with a core/shell CdSe/CdZnS structure using methods similar to those previously reported14. CdSe cores (3.2鈥塶m) were synthesized using a heat-up reaction between cadmium behenate (1鈥塵mol), selenium dioxide (1鈥塵mol), and 1,2-hexadecanediol (1鈥塵mol) in ODE (20鈥塵L) at 230鈥壜癈 for 15鈥塵in. After purification, a 4.7-monolayer shell of CdZnS was deposited using the same methodology as that for QD9.2. The shell composition comprised 2.4 monolayers of CdS, 0.8 monolayers of Cd0.8Zn0.2S, and 1.5 monolayers of ZnS.QD17.4 synthesisQD17.4 with emission at 680鈥塶m and 8.7鈥塶m diameter by electron microscopy was synthesized with a core/shell CdSe/CdZnS structure. A CdSe core with a first exciton peak at 645鈥塶m was synthesized using a method similar to that for QD13.3 with the substitution of cadmium behenate for cadmium myristate. The shell growth process was similar to that used for QD9.2 to yield a shell聽composition of 4.0 monolayers of Cd0.5Zn0.5S and 1.6 monolayers of ZnS.QD800 synthesisCdSe cores with emission at 549鈥塶m were synthesized in a heat-up reaction mixture of cadmium behenate (0.2鈥塵mol), selenium dioxide (0.2鈥塵mol), and 1,2-hexadecanediol (0.2鈥塵mol) in ODE (5鈥塵L) at 240鈥壜癈 for 60鈥塵in. After purification, cadmium was partially exchanged with mercury to yield a HgCdSe core by mixing the CdSe cores with mercury acetate in oleylamine and chloroform, followed by the addition of octanethiol to quench the reaction. After purification, a shell of CdZnS was deposited using the same methodology as that for QD9.2 to yield a shell聽composition of 2.4 monolayers of CdS, 0.8 monolayers of Cd0.5Zn0.5S, and 0.8 monolayers of ZnS.QD693 synthesisThe synthesis was the same as that used for QD800, except the CdSe core emission wavelength maximum was 558鈥塶m and mercury exchange was performed with mercury octanethiolate in oleylamine to reduce the degree of redshift.QD coating and conjugationQD9.2, QD13.3, and QD17.4 were coated with polyacrylamido(histamine-co-TEG-co-azido-TEG) (P-IM-N3) to generate aqueous azide-functional colloids13. Dibenzocyclooctyne (DBCO)-functionalized streptavidin (SAv) was prepared by mixing SAv (AnaSpec) in phosphate buffered saline (PBS; 0.5鈥塵g鈥塵L-1) with a solution of DBCO-N-hydroxysuccinimidyl ester (DBCO-NHS, Click Chemistry Tools) in DMSO (2.5鈥塵M) at a molar ratio of 1:5, followed by repeated pipetting and incubation on ice for 2鈥塰. The conjugate was purified by centrifugal filtration using a filter with 3鈥塳Da molecular weight cutoff (MWCO) at 4鈥壜癈. Azide-functional QDs in PBS were then mixed 1:1 with DBCO-SAv (and other ratios for optimization) and allowed to react at room temperature overnight. The reaction was quenched by adding a 50-fold molar excess of 2-azidoacetic acid on ice for 15鈥塵in. These conjugates were used directly for 2-step QD-FISH. For 1-step QD-FISH, QD-SAv conjugates were mixed with biotin-labeled probes at a 1:1 molar ratio for 1鈥塰 at room temperature.QD characterizationAbsorption spectra of QD dispersions were acquired using an Agilent Cary 5000 UV鈥揤is鈥揘IR spectrophotometer. Fluorescence spectra of QD dispersions were collected using a Horiba NanoLog spectrofluorometer, with solutions diluted to eliminate self-quenching. Signal acquisition conditions such as scan time, slit widths, and number of scans were adjusted so that the brightest sample was not saturating the detector (photomultiplier tube) and such that all spectra showed sufficiently high signal-to-noise ratios to yield smooth curves. Transmission electron microscopy images of QDs were obtained using a JEOL 2010 LaB6 high-resolution microscope in the Frederick Seitz Materials Research Laboratory Central Research Facilities at the University of Illinois. Samples were prepared by placing a drop of dilute QD solution in hexane or chloroform on an ultrathin carbon film TEM grid (Ted Pella, #01824) and then wicking the solution off with a tissue. QD-SAv conjugates were characterized by agarose gel electrophoresis (Supplementary Fig.聽8) using excess biotin-labeled DNA to confirm SAv conjugation to QDs by a migration shift13. The DNA sequence was 5鈥?Biotin/(T)68 TAGCCA GTG TAT CGC AAT GAC G-3鈥?(Integrated DNA Technologies). QD-SAv was incubated with biotin-DNA at room temperature for 15鈥塵in and electrophoresis was performed in a 2% polyacrylamide, 0.5% agarose gel at 4鈥壜癈.CellsHeLa cells (ATCC) were cultured in Eagle鈥檚 Minimum Essential Medium (EMEM) with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (P/S) at 37鈥壜癈 in 5% CO2. BPH-1 and LNCaP (ATCC) cells were cultured in RPMI Medium 1640 with 10% FBS and 1% P/S at 37鈥壜癈 in 5% CO2. VCaP (ATCC) cells were cultured in Dulbecco鈥檚 Modified Eagle鈥檚 Medium (DMEM) with 10% FBS and 1% P/S at 37鈥壜癈 in 5% CO2. For FISH studies, cells (1鈥壝椻€?05) were seeded on 18鈥塵m round #1 coverglass in each well of a 12-well cell culture plate and cultured until 70% confluent. The cells were then washed with PBS, fixed with 4% paraformaldehyde for 10鈥塵in at room temperature, and permeabilized with 70% (v/v) ethanol for 24鈥塰 at 4鈥壜癈. For qRT-PCR analysis, BPH-1, VCaP and LNCaP cells (3鈥壝椻€?05) were seeded in 6-well plates and cultured until 80% confluent. For PTEN mRNA silencing, cells were transfected with anti-PTEN or scrambled siRNA (Santa Cruz) using Lipofectamine 2000 (Thermo Fisher Scientific) following manufacturer protocols at a siRNA concentration of 1.5 fM for 24鈥塰. The medium was then aspirated and the cells were washed with PBS and fixed in 4% paraformaldehyde in PBS at room temperature for 10鈥塵in. After two washes with PBS, the cells were permeabilized with 70% (v/v) ethanol for at least 1鈥塰 at 2鈥壜癈 to 8鈥壜癈.Nucleic acid probesGAPDH mRNA Dye-FISH nucleic acid probes were synthesized and optimized by LGC Biosearch Technologies, labeled with either biotin or CAL Fluor庐 Red 590 Dye. Probes targeting human PTEN mRNA (NM_000314.6) and A20 mRNA (NM_001270508.1) were designed using Stellaris庐 Probe Designer (version 4.2, LGC Biosearch Technologies) and are provided in Supplementary Tables聽8 and 9.Dye-FISHGAPDH mRNA Dye-FISH was performed following standard procedures8 using Wash Buffers A and B and Hybridization Buffer supplied by the probe manufacturer. Probe incubation was performed for 16鈥塰 in the dark at 37鈥壜癈, nuclei were stained with Hoechst 33342 (Thermo Fisher Scientific), and each coverglass was mounted on a slide with 90% glycerol in PBS, sealed using nail polish.Two-step QD-FISHBiotin-labeled FISH probes with the same sequences used for Dye-FISH (LGC Biosearch Technologies) were hybridized with聽fixed and permeabilized cells on coverglass using identical protocols for Dye-FISH and nuclei were stained with Hoechst 33342. The cells were then blocked for 2鈥塰 with blocking conditions as indicated. The optimized mixture contained BSA and DS in 2脳 saline-sodium citrate (SSC) buffer at pH 7.2. After aspirating the blocking buffer, cells were incubated with 10鈥塶M QD-SAv in 1% (w/v) BSA in 2脳SSC buffer at room temperature for 2鈥塰. The ratio between QD:SAv and the time of incubation were independently optimized (Supplementary Fig.聽9 and 10). Cells on coverglass were then washed three times with Wash Buffer B before mounting on slides with 90% glycerol in PBS, sealed using nail polish.One-step QD-FISHFixed and permeabilized cells on coverglass were washed with Wash Buffer A for 5鈥塵in and incubated with Hybridization Buffer for 30鈥塵in, followed by the optimized blocking buffer from 2-step QD-FISH for 2鈥塰. The cells were then incubated in a mixture of biotin-probe conjugates of QD-SAv (8鈥塶M QD) in 10% formamide, 0.33鈥塵g鈥塵l-1 yeast RNA, 10鈥塵M ribonucleoside vanadyl complex, 0.1% BSA, and 2XSSC for 16鈥塰 in the dark at 37鈥壜癈 in a sealed humidified chamber.聽The coverglass was then聽washed with Wash Buffer A at 37鈥壜癈 for 30鈥塵in聽and nuclei were stained with Hoechst 33342 for 30鈥塵in. The coverglass was then washed with Wash Buffer B for 5鈥塵in before mounting on slides with 90% glycerol in PBS, sealed with nail polish.Multiplexed QD-FISHLNCaP cells were transfected with anti-PTEN siRNA as described above. A20 gene expression was induced by treatment with TNF-伪 (100鈥塶g/mL) in complete medium for 1鈥塰 at 37鈥壜癈. The cells were then processed following the 1-step QD-FISH procedure and incubated in a mixture of the three QD-SAv conjugates, each conjugated to biotinylated oligonucleotides complementary to mRNA sequences of GAPDH, PTEN, or A20. QD-FISH signals of QD608, QD693 and QD800 were collected using 488鈥塶m laser excitation and 600/37鈥塶m, 698/70鈥塶m or 809/81鈥塶m bandpass emission filters, respectively.ImagingImmediately after preparation, cells were imaged on a Zeiss Axio Observer Z1 inverted microscope with an EC Plan-Neofluar 100(times) 1.45鈥塏.A. oil-immersion objective. Images were collected with a Photometrics eXcelon Evolve 512 EMCCD camera controlled through Zeiss Zen software. Hoechst was imaged using 100鈥塛 halogen lamp excitation with a 365鈥塶m excitation filter and 445/50鈥塶m emission filter; CAL Fluor庐 Red 590 Dye was imaged using 561鈥塶m laser excitation and a 600/37鈥塶m bandpass emission filter. QD13.3 and commercial QDs were imaged using 488鈥塶m laser excitation and a 600/37鈥塶m bandpass emission filter. QD17.4 and QD9.2 were imaged using 488鈥塶m laser excitation and a 585鈥塶m long-pass emission filter. Z-stack images of entire cells were collected in 0.22 渭m increments. For each sample, 20 areas on the coverglass were selected at random for imaging. To obtain single-molecule fluorescence intensity values, dye-probes and QD-SAv conjugates dispersed in PBS were adsorbed on glass coverslips and imaged via epifluorescence microscopy using identical conditions to those used for FISH images. For each sample, videos during continuous excitation were acquired to identify single molecules by their distinct intensity time-traces using MATLAB algorithms14.Image analysisImages were exported as 8-bit uncompressed TIFF files. For 2D image analysis, spot counting in individual cells was performed using the Multiple Target Tracking (MTT) Algorithm based in MATLAB17 to determine the location and intensity of each spot. For 3D z-stacks, files were deconvolved using AUTOQUANT X3 (Media Cybernetics, Inc.) and analyzed using IMARIS (Bitplane) for 3D distribution reconstruction and signal spot counting. To ensure that deconvolution did not alter spot numbers, MTT analysis of 2D images was performed before and after deconvolution (Supplementary Fig.聽11). In the MTT algorithm, spot detection in images is performed by evaluating each 7鈥壝椻€? window in the image using a generalized likelihood ratio test to decide if a spot fits a point spread function, assuming Gaussian noise. Thus the analysis accounts for local background values, rather than global intensities, which is beneficial to account for the nonuniformity of autofluorescence across a cell. The algorithm also subtracts each spot and repeats the analysis until all spots are detected, which is beneficial when a high spatial density of spots is present. The threshold applied for the hypothesis test is normalized as the probability of false positives per 512 (times) 512 image (false positives per ~250,000 windows). In Fig.聽2, the indicated threshold is the probability of false positive of detected spots in logarithmic scaling, in units of decibels.qRT-PCRTotal RNA from cells in 6 well plates was extracted using a RNeasy Mini Kit (QIAGEN) and reverse-transcribed using a High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific). Primer sequences for human GAPDH were 5鈥?AGG GCT GCT TTT AAC TCT GGT-3鈥?and 5鈥?CCC CAC TTG ATT TTG GAG GGA-3鈥? Primer sequences for human PTEN were 5鈥?CAA GAT GAT GTT TGA AAC TAT TCC AAT G-3鈥?and 5鈥?CCT TTA GCT GGC AGA CCA CAA-3鈥? Primer sequences for human A20 were 5鈥?GAC CAT GGC ACA ACT CAT CTC A-3鈥?and 5鈥?GTT AGC TTC ATC CAA CTT TGC GGC ATT G-3鈥?sup>25,26,27,28. All primers were obtained from Integrated DNA Technologies. Real-time qPCR was performed on a Mastercycler庐 RealPlex2 (Eppendorf).Statistical analysisData are presented as mean鈥壜扁€塻.d. Statistical significance was determined using Student鈥檚 t-test and analysis of variance (one-way ANOVA) using GraphPad Instat 3 software. After comparing the overall difference between groups, the Tukey鈥檚 honestly significant difference (HSD) post-hoc test was used to specify where the differences occurred between groups. The data that support the findings of this study are available from the corresponding author upon reasonable request. References1.Gall, J. G. Pardue, M. L. Formation and detection of RNA-DNA hybrid molecules in cytological preparations. Proc. Natl. Acad. Sci. USA 63, 378鈥?83 (1969).ADS聽 CAS聽 Article聽Google Scholar聽 2.John, H. A., Birnstiel, M. L. Jones, K. W. RNA-DNA hybrids at the cytological level. Nature 223, 582鈥?87 (1969).ADS聽 CAS聽 Article聽Google Scholar聽 3.Langer-Safer, P. R., Levine, M. Ward, D. C. Immunological method for mapping genes on Drosophila polytene chromosomes. Proc. Natl. Acad. Sci. USA 79, 4381鈥?385 (1982).ADS聽 CAS聽 Article聽Google Scholar聽 4.Levsky, J. M. Singer, R. H. Fluorescence in situ hybridization: past, present and future. J. Cell. Sci. 116, 2833鈥?838 (2003).CAS聽 Article聽Google Scholar聽 5.Femino, A. M., Fay, F. S., Fogarty, K. Singer, R. H. Visualization of single RNA transcripts in situ. Science 280, 585鈥?90 (1998).ADS聽 CAS聽 Article聽Google Scholar聽 6.Crosetto, N., Bienko, M. van Oudenaarden, A. Spatially resolved transcriptomics and beyond. Nat. Rev. Genet. 16, 57鈥?6 (2015).CAS聽 Article聽Google Scholar聽 7.Buxbaum, A. R., Haimovich, G. Singer, R. H. In the right place at the right time: visualizing and understanding mRNA localization. Nat. Rev. Mol. Cell Biol. 16, 95鈥?09 (2015).CAS聽 Article聽Google Scholar聽 8.Raj, A., van den Bogaard, P., Rifkin, S. A., van Oudenaarden, A. Tyagi, S. Imaging individual mRNA molecules using multiple singly labeled probes. Nat. Methods 5, 877鈥?79 (2008).CAS聽 Article聽Google Scholar聽 9.Lubeck, E., Coskun, A. F., Zhiyentayev, T., Ahmad, M. Cai, L. Single-cell in situ RNA profiling by sequential hybridization. Nat. Methods 11, 360鈥?61 (2014).CAS聽 Article聽Google Scholar聽 10.Kairdolf, B. A. et al. Semiconductor quantum dots for bioimaging and biodiagnostic applications. Ann. Rev. Anal. Chem. 6, 143鈥?62 (2013).CAS聽 Article聽Google Scholar聽 11.Ioannou, D. et al. Quantum dots as new-generation fluorochromes for FISH: an appraisal. Chromosome Res. 17, 519鈥?30 (2009).CAS聽 Article聽Google Scholar聽 12.Ioannou, D. Griffin, D. K. Nanotechnology and molecular cytogenetics: the future has not yet arrived. Nano Rev. 1, 5117鈥?130 (2010).Article聽Google Scholar聽 13.Ma, L. et al. Multidentate polymer coatings for compact and homogeneous quantum dots with efficient bioconjugation. J. Am. Chem. Soc. 138, 3382鈥?394 (2016).CAS聽 Article聽Google Scholar聽 14.Lim, S. J. et al. Brightness-equalized quantum dots. Nat. Comm. 6, 8210鈥?219 (2015).CAS聽 Article聽Google Scholar聽 15.Baum, M., Erdel, F., Wachsmuth, M. Rippe, K. Retrieving the intracellular topology from multi-scale protein mobility mapping in living cells. Nat. Comm. 5, 4494鈥?505 (2014).ADS聽 CAS聽 Article聽Google Scholar聽 16.Wang, L. et al. Single molecule localization imaging of telomeres and centromeres using fluorescence in situ hybridization and semiconductor quantum dots. Nanotechnology 29, 285602鈥?85610 (2018).Article聽Google Scholar聽 17.Serge, A., Bertaux, N., Rigneault, H. Marguet, D. Dynamic multiple-target tracing to probe spatiotemporal cartography of cell membranes. Nat. Methods 5, 687鈥?94 (2008).CAS聽 Article聽Google Scholar聽 18.Kwabi-Addo, B. et al. Haploinsufficiency of the PTEN tumor suppressor gene promotes prostate cancer progression. Proc. Natl. Acad. Sci. USA 98, 11563鈥?1568 (2001).ADS聽 CAS聽 Article聽Google Scholar聽 19.Murphy, S. J. et al. Integrated analysis of the genomic instability of PTEN in clinically insignificant and significant prostate cancer. Mod. Pathol. 29, 143鈥?56 (2016).CAS聽 Article聽Google Scholar聽 20.Coskun, A. F. Cai, L. Dense transcript profiling in single cells by image correlation decoding. Nat. Methods 13, 657鈥?60 (2016).CAS聽 Article聽Google Scholar聽 21.Smith, A. M. Nie, S. Bright and compact alloyed quantum dots with broadly tunable near-infrared absorption and fluorescence spectra through mercury cation exchange. J. Am. Chem. Soc. 133, 24鈥?6 (2011).CAS聽 Article聽Google Scholar聽 22.Golovko, O., Nazarova, N. Tuohimaa, P. A20 gene expression is regulated by TNF, Vitamin D and androgen in prostate cancer cells. J. Steroid Biochem. Mol. Biol. 94, 197鈥?02 (2005).CAS聽 Article聽Google Scholar聽 23.Smith, A. M., Mancini, M. C. Nie, S. M. Second window for in vivo imaging. Nat. Nanotech. 4, 710鈥?11 (2009).ADS聽 CAS聽 Article聽Google Scholar聽 24.Bruns, O. T. et al. Next-generation in vivo optical imaging with short-wave infrared quantum dots. Nat. Biomed. Engr. 1, 0056鈥?066 (2017).Article聽Google Scholar聽 25.Shen, Y. H. et al. Up-regulation of PTEN (phosphatase and tensin homolog deleted on chromosome ten) mediates p38 MAPK stress signal-induced inhibition of insulin signaling. A cross-talk between stress signaling and insulin signaling in resistin-treated human endothelial cells. J. Biol. Chem. 281, 7727鈥?736 (2006).CAS聽 Article聽Google Scholar聽 26.Kim, S. et al. Down-regulation of the tumor suppressor PTEN by the tumor necrosis factor-alpha/nuclear factor-kappaB (NF-kappaB)-inducing kinase/NF-kappaB pathway is linked to a default IkappaB-alpha autoregulatory loop. J. Biol. Chem. 279, 4285鈥?291 (2004).CAS聽 Article聽Google Scholar聽 27.Gupta, A. Dey, C. S. PTEN, a widely known negative regulator of insulin/PI3K signaling, positively regulates neuronal insulin resistance. Mol. Biol. Cell. 23, 3882鈥?898 (2012).CAS聽 Article聽Google Scholar聽 28.Gunzl, P. et al. Anti-inflammatory properties of the PI3K pathway are mediated by IL-10/DUSP regulation. J. Leukoc. Biol. 88, 1259鈥?269 (2010).Article聽Google Scholar聽 Download referencesAcknowledgementsThis work was supported by funds from the Mayo-Illinois Alliance, the National Institutes of Health (R33CA196458聽to J.C.C., S.J.M., F.K., and A.M.S.) and the National Science Foundation I-Corps Program (1745812聽to A.M.S.). P.L. was supported by the National Institute of Biomedical Imaging and Bioengineering of the National Institutes of Health under Award Number T32EB019944 and the National Science Foundation Grant 0965918 IGERT: Training the Next Generation of Researchers in Cellular and Molecular Mechanics and BioNanotechnology. L.M. was supported by the聽2017-2018 UIUC聽Graduate College Dissertation Completion Fellowship.聽S.J.L. acknowledges support from the Start-up Fund and Basic Research Programme (18-NT-01) of DGIST funded by the Ministry of Science and ICT of Korea.Author informationAffiliationsDepartment of Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL, 61801, USAYang Liu,聽Phuong Le,聽Sung Jun Lim,聽Suresh Sarkar聽 聽Andrew M. SmithMicro and Nanotechnology Laboratory, University of Illinois at Urbana-Champaign, Urbana, IL, 61801, USAYang Liu,聽Phuong Le,聽Sung Jun Lim,聽Liang Ma,聽Suresh Sarkar,聽Zhiyuan Han聽 聽Andrew M. SmithIntelligent Devices and Systems Research Group, DGIST, Hyeonpung, Daegu, 42988, Republic of KoreaSung Jun LimDepartment of Materials Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, IL, 61801, USALiang Ma,聽Zhiyuan Han聽 聽Andrew M. SmithBiomarker Discovery Program, Center of Individualized Medicine, Mayo Clinic, Rochester, MN, 55905, USAStephen J. Murphy,聽Farhad Kosari,聽George Vasmatzis聽 聽John C. ChevilleDepartment of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, MN, 55905, USAJohn C. ChevilleCarle Illinois College of Medicine, Urbana, IL, 61801, USAAndrew M. SmithAuthorsYang LiuView author publicationsYou can also search for this author in PubMed聽Google ScholarPhuong LeView author publicationsYou can also search for this author in PubMed聽Google ScholarSung Jun LimView author publicationsYou can also search for this author in PubMed聽Google ScholarLiang MaView author publicationsYou can also search for this author in PubMed聽Google ScholarSuresh SarkarView author publicationsYou can also search for this author in PubMed聽Google ScholarZhiyuan HanView author publicationsYou can also search for this author in PubMed聽Google ScholarStephen J. MurphyView author publicationsYou can also search for this author in PubMed聽Google ScholarFarhad KosariView author publicationsYou can also search for this author in PubMed聽Google ScholarGeorge VasmatzisView author publicationsYou can also search for this author in PubMed聽Google ScholarJohn C. ChevilleView author publicationsYou can also search for this author in PubMed聽Google ScholarAndrew M. SmithView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsY.L. performed quantum dot conjugation, cell studies, and microscopy. L.M. and Z.H. performed quantum dot polymer coating and chromatographic analysis. Y.L. and P.L. performed image and data analysis. S.J.L. and S.S. prepared quantum dot cores and performed electron microscopy. A.M.S., Y.L., S.J.M., F.K., G.V., and J.C.C. designed and supervised the experiments. A.M.S. and Y.L. wrote the manuscript.Corresponding authorCorrespondence to Andrew M. Smith.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPublisher\'s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Electronic supplementary material Supplementary InformationRights and permissions Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article鈥檚 Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article鈥檚 Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and PermissionsAbout this articleCite this articleLiu, Y., Le, P., Lim, S.J. et al. Enhanced mRNA FISH with compact quantum dots. Nat Commun 9, 4461 (2018). https://doi.org/10.1038/s41467-018-06740-xDownload citationReceived: 27 April 2018Accepted: 21 September 2018Published: 26 October 2018DOI: https://doi.org/10.1038/s41467-018-06740-x Tapas Kumar Mandal, Nargish Parvin, Kanchan Mishra, Sonaimuthu Mohandoss Yong Rok Lee Microchimica Acta (2019) CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.