A Hybrid Detection Method Based on Peroxidase-mediated Signal Amplification and Click Chemistry for Highly Sensitive Background-free Immunofluorescent Staining

Abstract

The copper-catalyzed azide-alkyne cycloaddition (CuAAC) reaction is increasingly used for detection of various macromolecules and metabolites in biological samples. Here, we present a detailed analysis of the CuAAC reaction conditions in cells and tissue sections. Using the optimized CuAAC conditions, we have devised a highly sensitive immunostaining technique, based on the tyramide signal amplification/catalyzed reporter deposition (TSA/CARD) method with a novel alkyne tyramide substrate. The described method offers improved detection threshold compared to conventional immunofluorescent staining and produces significantly lower non-specific background than TSA/CARD with fluorescent tyramides.

Introduction

The immunofluorescent staining represents an indispensable tool for studying expression and localization of macromolecules in cells and tissues. The efficiency of immunostaining is determined by epitope availability in the specimen and the sensitivity of the detection method. Since the sample preparation for immunostaining and in situ hybridization inevitably involves fixation, and possibly other deleterious processes (e.g., organic solvent treatment, paraffin embedding), the preservation of proteins and nucleic acids of interest is frequently compromised. The antigen retrieval methods ¹ and the advanced fixatives ² were developed to overcome this issue; however, all these techniques possess their drawbacks and limitations. For instance, the antigen retrieval procedures are associated with the risk of disturbing the tissue integrity, and the employment of non-crosslinking fixatives might produce localization artifacts due to the diffusion of the target molecule. ³ Therefore, a sensitive detection system is of paramount importance to achieve high-quality results in immunostaining.

Click chemistry is a class of organic reactions that are characterized by high rate, selectivity and compatibility with physiological environment. The implementation of Click chemistry-based detection has greatly expanded the scope and improved the sensitivity of multiple in situ analyses, including proliferation/cell cycle assays, ⁴ apoptosis detection ⁵ and studies of protein post-translational modifications. ⁶ In these areas, the copper-catalyzed azide-alkyne cycloaddition (CuAAC) is the most commonly employed Click reaction.

The goal of this study was to explore the possibility of employing CuAAC for increasing the efficiency of immunofluorescent detection. We have carefully investigated the CuAAC reaction conditions in cultured cells and tissue sections, utilizing a model reaction between a 5-ethynyl-2’-deoxyuridine (EdU) incorporated into the cellular DNA, and several fluorescent azides. Using the optimized CuAAC conditions, we have devised a sensitive immunostaining procedure that is based on a tyramide signal amplification/catalyzed reporter deposition (TSA/CARD) method with a novel alkyne tyramide substrate for horseradish peroxidase. The described method facilitates the detection of low abundance epitopes that are undetectable by the conventional staining protocol with fluorescent-labeled secondary antibody.

Materials and Methods

iPS Cell Cultures and Neural Differentiation

Human-induced pluripotent stem (iPS) cell lines IPSHD1.1S and IPSPDSP1S were generated previously by reprogramming the human skin fibroblasts with Oct4, SOX2, Nanog, and Klf4 genes. ⁷ iPS cells were cultured in mTESR medium (Stem Cell Technologies, Vancouver, BC, Canada) on the plastic Petri dishes (SPL Life Sciences, Pocheon, Gyeonggi, South Korea) coated with Matrigel (BD Biosciences, San Jose, CA, USA).

The neural progenitors were derived from iPS cells according to a previously described protocol.^8,9 Briefly, the neural differentiation was induced in DMEM-F12 1:1 medium (Paneco, Moscow, Russia) with 2% KnockOut™ serum replacement (Life Technologies, Grand Island, NY, USA), 1× N-2 supplement (Life Technologies), 80 ng/ml recombinant human noggin (Peprotech, Rocky Hill, NJ, USA), 10 µM SB431542 (Stemgent, Cambridge, MA, USA), 1 mM non-essential amino acids (Paneco), 2 mM L-glutamine (Paneco), and 1× penicillin-streptomycin (Paneco). This medium was changed daily for 7–10 days. The formed neural rosettes were mechanically dissected, transferred to 24-well Ultra Low Attachment plates (Corning Inc., NY, USA) and cultured in the neural differentiation medium with 1× B-27 supplement (Life Technologies) to generate the floating neurospheres. The neurospheres were dissociated with 0.05% trypsin (Paneco), and the obtained cells were seeded on matrigel-coated coverslips. The cells were then passaged for 2–4 weeks in DMEM-F12 containing 1× B-27 (Life Technologies), 1× N-2, and 2 mM L-glutamine.

Histology

Mice of 129 strain were sacrificed by cervical dislocation. Embryos were isolated from pregnant mice at 19–20 d.p.c. Adult and embryonic brains were fixed by immersion in 4% neutral buffered formaldehyde (NBF) for 48–72 hr at 4C. Cultured cells and neurospheres were fixed with NBF for 24 hr and pre-embedded in 2% low melting point agarose (Sigma-Aldrich, St. Louis, MO, USA). Fixed brain tissue and agarose cell blocks were dehydrated in isopropanol, cleared in xylene and embedded in Histomix Extra paraffin (Biovitrum, Moscow, Russia). The blocks were sectioned at 6 µm and the sections were mounted on 3-aminopropyltriethoxysilane (Thermo Scientific, Rockford, IL, USA) coated glass slides. Before the staining, sections were deparaffinized with xylene and rehydrated with graded ethanol series.

BrdU and EdU Detection

EdU (Sigma-Aldrich) and BrdU (Abcam, Cambridge, UK) at the final concentration of 1 mM were introduced into the culture medium for labeling of the proliferating cells. The cells and neurospheres were then fixed with 4% NBF or 100% methanol for 30 min at +4C. The adherent cells were permeabilized with 0.5% triton X-100 (Sigma-Aldrich).

Cells and deparaffinized sections were blocked with 10% FBS (Gibco, Camarillo, CA, USA) in PBS for 30 min. For BrdU staining, the cells were treated with 0.5 N HCl for 30 min at 37C and washed three times with PBS. Anti-BrdU antibody (Santa-Cruz, Dallas, TX, USA) was applied at 1:200 dilution overnight at 4C. The samples were then washed three times with PBS and incubated with goat anti mouse IgG antibodies conjugated to either Alexa 488 or Alexa 568 (Invitrogen, Carlsbad, CA, US), at 1:1000 dilution for 2 hr at room temp. After the incubation, the samples were washed three times with PBS.

EdU was detected via CuAAC reaction with BODIPY-FL, Cy3, and Sulfo-Cy3 azides (Lumiprobe, Hunt Valley, MD, USA). Stock 10 mM azide solutions were prepared in DMSO. Reactions were performed in PBS (pH = 7.4) or Tris-HCl buffer (pH = 7.5) containing 1–5 mM CuSO₄ and 50 mM ascorbic acid (Sigma-Aldrich). The incubations with the reaction mix were performed at room temperature in humidified chambers for 10–60 min, and the samples were then washed three times with PBS. DMSO, N, N-dimethylformamide (DMF), and formamide (Sigma-Aldrich) were of molecular biology grade.

Optimized CuAAC Protocol

The use of Sulfo-Cy3 azide is preferred because it provides the highest signal-to-noise (S/N) ratios, and its non-specific binding can be readily washed off from samples. The reaction mix consisted of: 1-2,5 µM Sulfo-Cy3 azide, 5 mM CuSO4, and 50 mM ascorbic acid in PBS. The samples were incubated with the reaction mix for 40 min, then washed three times with PBS. The nuclei were counterstained with DAPI.

Immunofluorescence

The following primary antibodies were used: mouse monoclonal anti-NeuN, 1:1000, (Merck-Millipore, Temecula, CA, US); mouse monoclonal anti-NMDAR1 1:100, (Abcam); and rabbit polyclonal to GRIN2A, 1:300 (Sigma-Aldrich). Samples were processed for either conventional immunofluorescence (IF), TSA/CARD with fluorescent tyramides or CuAAC-mediated immunofluorescent staining.

Conventional IF Staining

The cells and sections were blocked with 10% FBS in PBS for 30 min and incubated with primary antibodies in PBS overnight at 4С. After washing with PBS, the secondary antibodies (Alexa 568 goat anti-mouse IgG and Alexa 546 goat anti-rabbit IgG, Invitrogen) were applied for 2 hr at 1:1000 dilution. The samples were washed three times with PBS (5 min) and counterstained with DAPI.

TSA/CARD With Fluorescent Tyramides

TSA was performed as previously described. ¹⁰ Fluorescent tyramides were prepared from Sulfo-Cy3-N-hydroxysuccinimidyl ester (NHS; Lumiprobe), Cy3-NHS (Syntol, Moscow, Russia), and Cy3-NHS (Lumiprobe) by the reaction with tyramine freebase (Sigma-Aldrich) in DMSO according to Hopman et al. ¹¹

The samples were blocked with 10% FBS in PBS for 30 min and incubated with primary antibodies diluted in PBS overnight at 4С. After washing, biotinylated goat anti-mouse IgG (Imtek, Russia, Moscow, 1:1000) or biotinylated goat anti-rabbit IgG (Imtek; 1:1000) were applied for 1 hr at room temp. The samples were washed three times with PBS (5 min) and treated with streptavidin-peroxidase conjugate (Roche, 1:1000) for 1 hr at room temp. After three washes with PBS, the samples were treated with fluorescent tyramides (1–5 ug/ml) in PBS containing 0.015% H2O2 for 20 min. The samples were then washed three times with PBS. Attempts to wash the samples with 70%, 96% ethanol or DMSO to decrease the background after TSA with fluorescent tyramides were unsuccessful.

CuAAC-mediated Detection

N-(4-hydroxyphenethyl)-pent-4-ynamide (alkyne tyramide) was prepared by acylation of tyramine (3.94 mg, 28.4 µmol), with pentynoic acid 4-sulfotetrafluorophenyl ester (10 mg, 28.7 µmol) (Lumiprobe) in 100 µl DMSO. The reaction was carried out overnight at room temperature and the obtained solution was used without further purification. The stock 1 mg/ml solution of this substrate was prepared in DMSO and stored at -20C.

Cells and deparaffinized sections were blocked with 10% FBS in PBS for 30 min and incubated with primary antibodies diluted in PBS overnight at 4С. After washing, biotinylated goat anti-mouse IgG (Imtek; 1:1000) or biotinylated goat anti-rabbit IgG (Imtek; 1:1000) were applied for 1 hr at room temp. The samples were then washed and treated with streptavidin-peroxidase conjugate (Roche; 1:1000) for 1 hr at room temp. After washing with PBS, the reaction with alkyne tyramide (working concentration 1–5 µg/ml) was carried out in PBS containing 0.015% H₂O₂ for 20 min. The samples were then washed three times with PBS. The CuAAC reaction with fluorescent azides was performed in PBS containing 5 mM CuSO₄, 50 mM ascorbic acid (Sigma-Aldrich), and 1 uM Sulfo-Cy3 azide. The samples were incubated at room temperature in humidified chamber for 40 min, and then were washed three times with PBS. The nuclei were counterstained with DAPI hydrochloride (Sigma-Aldrich), 250 ng/ml.

Microscopy

Images were acquired with Imager Z1 microscope equipped with HRm monochrome CCD camera and bundled with AxioVision 4.8 software (all from Zeiss, Jena, Germany). Images were taken at 4164 х 3120 resolution and 16-bit depth. The mercury arc 103 W lamp was used as the reflected light source. The following Zeiss filter sets were used (ex: excitation, bs: beamsplitter, em: emission): #44 (ex 475/40; bs 500; em 530/50) for Alexa 488/BODIPY-FL, #45 (ex 560/40; bs 585; em BP 630/75) for Alexa 568, #20 HE (ex 546/12; bs 560; em 607/80) for Alexa 546, Cy3 and Sulfo-Cy3 and #49 (ex 365; bs 395; em 445/50) for DAPI. The camera exposure time was determined using the oversaturation tool that highlights the pixels that achieve the upper limit of camera’s dynamic range. The exposure was set 1–2 msec below the value at which oversaturated pixels are observed in the specific staining area.

Images were analyzed with Fiji software. ¹² For calculation of signal/noise ratio, we manually selected 30–50 cells per group (signal/noise) on each section or coverslip culture. The signal intensity was measured in the EdU+ nuclei, while the cytoplasmatic regions and the nuclei EdU– cells were used to determine the background level. The area of the nuclei was from 50 (paraffin-embedded neural progenitors (NPs)) to 200 µm² (iPS cells), thus the measurements were made from total area of 2500–10,000 µm² per sample, or 80,000–350,000 pixels for 20× objective. The area for signal and noise measurements was approximately equal.

Statistics

The statistical analyses were performed using Sigma Plot 12.5 (Systat Software Inc, US).

Results

Determination of the Optimal CuAAC Conditions Using the Model Reaction Between EdU and Several Fluorescent Azides

EdU is a thymidine analog that is readily incorporated into the genomic DNA of the proliferating cells. ⁴ The presence of EdU-labeled DNA in the cells is revealed by CuAAC reaction with fluorescent azides.^4,13 Considering the flexibility and robustness of EdU labeling and detection, we have utilized the reaction between EdU and several fluorescent azides as a model to perform a systematic study of CuAAC reaction conditions in cells and tissue sections

First, we have compared the S/N ratios of BODIPY-FL, Cy3, and Sulfo-Cy3 azides in two cell types: human iPS cells and human NP cells. The iPS cells were investigated as adherent cultures grown on the coverslips, and NPs were examined both as coverslip cultures and as the slide-mounted sections of paraffin-embedded cells. The fluorescence intensity of EdU-labeled nuclei was used for estimation of the specific staining efficiency, and the background staining (noise) was measured in the regions outside nuclei (iPS cultures) or the EdU-negative cells (NPs). The early S-phase cells, characterized by uneven punctuate EdU staining pattern were excluded from the measurements.

In all investigated specimen, the S/N ratios (from lowest to highest) were in the following order: BODIPY-FL < Cy3 < Sulfo-Cy3 (Fig. 1A and B). We presume, that lower BDP-FL- and Cy3-azide S/N ratios (Figs. 1 and 2) in adherent cells are explained by non-specific azide adsorption that hindered the reaction with EdU. Since the cells treated with BrdU in equal conditions showed intense selective nuclear staining, it could be ruled out that calculated S/N values are associated with inefficient genomic DNA labeling or incorporation of thymidine analog into extrachromasomal DNA (Fig. 2B).

Figure 1.

EdU detection with different azides. (A) Representative images of EdU-labeled iPS cell colonies stained with BODIPY-FL (BDP-FL), Cy3, and Sulfo-Cy3 fluorescent azides. Images were taken with same exposure settings and demonstrate the differences in S/N ratios between three investigated azides. Scale bar—100 µm. (B) Calculated S/N ratios for EdU detection with different azides in adherent cultures of iPS cells, NPs and sections of paraffin-embedded cells. Abbreviations: EdU, 5-ethynyl-2’-deoxyuridine; iPS, induced pluripotent stem; S/N, signal-to-noise ratio; DAPI, 4,’ 6-diamidino-2-phenylindole; NP, neural progenitor.

Figure 2.

Consequential 1-hour pulse-labeling of adherent cultures of human neural progenitors with equal concentrations of EdU and BrdU. (A) and (D)—EdU staining via CuAAC with BDP-FL and Sulfo-Cy3 azides, respectively. (B) and (E)—BrdU IF. (C) and (F) DAPI counterstaining. (A–C)—Strong background staining develops in CuAAC reaction with BDP-FL, and the EdU+ nuclei are barely detectable. This observation cannot be explained by low level of EdU incorporation, since labeling with BrdU in the same concentration reveals strong background-free nuclear staining in the proliferating cells. (D–F)—CuAAC with Sulfo-Cy3 azide affords significantly improved S/N ratio for EdU in these cells. Scale bar—50 µm. Abbreviations: EdU, 5-ethynyl-2’-deoxyuridine; BrdU, 5-Bromo-2’-deoxyuridine; IF, immunofluorescence; DAPI, 4,’ 6-diamidino-2-phenylindole; CuAAC, copper-catalyzed azide-alkyne cycloaddition; S/N, signal-to-noise ratio.

Sulfo-Cy3 azide ensured the lowest background in all investigated specimen (Figs. 1 and 2). Importantly, the EdU staining with Sulfo-Cy3 was not affected by BrdU retrieval procedure (Fig. 2). It was found that the non-specific binding of fluorescent azides inversely correlates with cell culture and tissue density. Notably, the background staining is resistant to detergent solutions (0.1–0.5% triton X-100).

The S/N ratios were mildly improved by increasing the fluorescent azide working concentrations. However, the high azide concentrations (10–50 µM) employed in the study of Salic and Mitchison, ⁴ in our hands, caused significant fluorescence emission bleed-through to the adjacent fluorescence channels (e.g., red fluorescence of Cy3 to FITC/ green fluorescent protein (GFP) channel). Thus, we opted for using azide concentrations below 5 µM.

We found that EdU staining with fluorescent azides is not affected by buffer composition. Equal signal intensity was observed with PBS (pH = 7.4) and Tris-HCl (pH = 7.5) buffers. Moreover, same results were obtained when no buffer salts were added to the reaction mix (data not shown).

Next, we have investigated the impact of fixative on the S/N ratios of Cy3 azide. It was found that both methanol and NBF fixation are suitable for EdU staining in iPS cells, but S/N ratios were marginally better with NBF fixation (Fig. 3A).

Figure 3.

(A) Comparison of EdU S/N ratios with Cy3 azide in iPS cells fixed by neutral buffered formaldehyde (NBF) and methanol (MeOH). *p<0.001 by Mann-Whitney U-test, n=3. (B) Impact of organic solvents on EdU staining. EdU fluorescence intensity is calculated relative to control level. No solvents were added in controls, except 1% DMSO introduced from stock azide solutions. *p<0.01 by Kruskal–Wallis one-way analysis, n=3. (C) Influence of copper sulfate concentration in the reaction mix on EdU fluorescence intensity. The relative to intensity at 1 mM CuSO₄ is taken as one unit *p<0.001 by Mann-Whitney U-test, n=3. (D) Relationship between EdU fluorescence, background level (given in arbitrary units, A.U.) and the reaction time using BODIPY-FL azide. Abbreviations: EdU, 5-ethynyl-2’-deoxyuridine; S/N, signal-to-noise ratio; iPS, induced pluripotent stem; ND, no EdU signal detectable; DMSO, dimethyl sulfoxide; DMF, N, N-dimethylformamide.

Since both BODIPY-FL and Cy3 azides are hydrophobic, we assumed that increasing their solubility in the reaction medium can reduce their non-specific adsorption by lipophilic cellular components. To evaluate this hypothesis, we have investigated how the content of dipolar aprotic solvents DMSO and DMF in the reaction medium would affect the EdU staining. These solvents are traditionally used in CuAAC reactions between water-soluble and hydrophobic compounds. ¹⁴ Contrary to our expectations, even low (10%) solvent concentrations significantly attenuated EdU staining. With 50% DMSO or DMF in the reaction mixture the EdU staining became undetectable (Fig. 3B). We have also found that the addition of formamide, a solvent commonly employed in in situ hybridization for destabilizing the hydrogen bonds between nucleic acids, ¹⁵ that was expected to improve the accessibility of EdU for the reaction with azide, also suppressed EdU staining. Its effect was, however, less prominent compared to DMF and DMSO (Fig. 3B). The detrimental effect of the listed organic solvents was observed both with PBS and Tris-HCl buffers (data not shown).

The copper (I) catalyst in CuAAC reactions is usually generated in situ from copper (II) salts by the mild reducing agents. ¹⁴ We have investigated the influence of copper (II) salt concentration in the reaction mix on EdU staining intensity. It was found, that with 0.2 mM CuSO₄ the specific staining intensity was attenuated below the detection threshold. Increasing CuSO₄ concentration in the reaction mix from 1 to 5 mM caused small but significant improvement of EdU signal (Fig. 3C). Further increasing CuSO₄ concentrations caused precipitate formation.

Commonly, the copper catalysts are utilized as the complexes with various triazole ligands, such as TBTA, to improve the rate and efficiency of CuAAC reactions. ¹⁴ Our attempts to employ the Cu catalyst as the TBTA complex in the reaction of EdU with fluorescent azides were unsuccessful. Since TBTA is insoluble in water, we used the minimum amount of DMSO to keep this ligand in solution, considering the previously demonstrated negative impact of DMSO on the reaction efficiency. However, in our experiments, the EdU staining intensity with Cu-TBTA complex was inferior to the free copper catalyst in presence of equal amount of DMSO (data not shown).

Next, we have studied the relation between EdU signal intensity, background staining and the reaction time. We present the data for BDP-FL azide, because at all time points its fluorescence fitted within the dynamic range of the camera at the fixed exposure time (Fig. 3D). In line with the data of Salic and Mitchison ⁴ the EdU signal could be detected even after 10 min reaction. However, extending the reaction time progressively increased EdU fluorescence intensity, that was close to the maximum level after 40 min reaction with 2 µM fluorescent azide. Longer reaction times (60–80 min) caused further modest enhancement of specific EdU signal, but also increased the background fluorescence (Fig. 3D).

The Use of CuAAC Reaction for Enhancement of Immunofluorescent Staining

In order to increase the sensitivity of immunofluorescent staining we have employed the Click chemistry–mediated detection instead of the fluorescent-labeled secondary antibodies. To our knowledge, there are currently no reports of using CuAAC for that purpose. We have devised a staining protocol using TSA/CARD technique^16,17 with our custom tyramide derivative, N-(4-hydroxyphenethyl)-pent-4-ynamide (hereinafter referred as alkyne tyramide) (Fig. 4). TSA/CARD is a robust histochemical method that relies on the conversion of tyramine conjugates to the highly reactive radical species by horseradish peroxidase. These radicals become instantly anchored to proteins in the vicinity of enzymatic reaction site, marking it with tyramine-conjugated haptens or fluorescent labels.^16,17 We examined if this technique is suitable for deposition of tyramine-conjugated alkyne moieties, which could be then detected using CuAAC reaction with fluorescent azides.

Figure 4.

The flowchart illustrating side-by-side comparison of the staining protocols for conventional indirect IF, TSA/CARD with fluorescent tyramides, and CuAAC-mediated detection. After each step the samples are washed three times with PBS (not shown on the chart). Due to an additional step in the CuAAC-mediated detection protocol, it is about 30–60 min longer than standard TSA with fluorescent tyramides. Abbreviations: IF, immunofluorescence; TSA/CARD, tyramide signal amplification/catalyzed reporter deposition; CuAAC, copper-catalyzed azide-alkyne cycloaddition.

To evaluate this approach, we performed the immunostaining of mouse embryonic brain serial sections with the NeuN antibody, that labels the nuclear protein Fox-3 specifically expressed in post-mitotic neurons. ¹⁸ The detection was performed either with a fluorescent-dye-conjugated secondary antibody (control), or through sequential treatment with biotinylated secondary antibody, streptavidin-peroxidase conjugate, incubation with alkyne tyramide in presence of hydrogen peroxide, and finally, the reaction with Sulfo-Cy3 azide and the copper catalyst. The peroxidase reaction with alkyne tyramide was carried out as described for other tyramide conjugates^11,19 (see Methods section), and the CuAAC reaction was performed in the optimized conditions described above.

We found that our alkyne tyramide substrate can be effectively deposited by the immuno-bound peroxidase, as the following CuAAC reaction with fluorescent azide revealed the expected staining pattern in the nuclei of hippocampal neurons (Fig. 5). The novel method offered superior sensitivity compared to conventional indirect immunofluorescent staining. For instance, a limited number of NeuN⁺ cells were detectable exclusively in the pyramidal layer of hippocampus, when the detection was performed with Alexa 568-conjugated secondary antibody. A camera exposure of 150 msec was needed to acquire the images of immunopositive cells. In contrast, the CuAAC-mediated detection revealed various degrees of NeuN expression in a significantly higher number of cells in all layers of hippocampus. Importantly, the intensity of NeuN signal achieved saturation at 6 msec exposure. We opted for using the saturation camera exposure instead of fluorescence intensity acquired at the given exposure value, since the former approach ensures more accurate comparison for signals with significantly different intensity.

Figure 5.

The comparison of NeuN staining in the mouse embryo hippocampus using the conventional IF protocol with Alexa 568 conjugated secondary antibody and the CuAAC-mediated detection with alkyne tyramide, followed by reaction with Sulfo-Cy3 azide. The camera exposure time used to acquire the image is indicated in the upper left corner. Scale bar − 100 µm. Abbreviations: IF, immunofluorescence; CuAAC, copper-catalyzed azide-alkyne cycloaddition; DAPI, 4,’ 6-diamidino-2-phenylindole.

The CuAAC-mediated detection was then compared with the original TSA/CARD method. In this experiment we used the same primary NeuN antibodies for staining embryonic mouse brain sections. For the more accurate result we employed Sulfo-Cy3 dye in both methods, as a tyramide derivative in TSA/CARD and in the form of azide in CuAAC-mediated detection system. Intense nuclear staining was obtained with both detection methods, but the exposure time to acquire the saturated signal was slightly higher for TSA/CARD (30 msec) compared to CuAAC-mediated detection (6 msec). Furthermore, Sulfo-Cy3 azide showed strikingly lower propensity for background staining in contrast to its tyramide counterpart (Fig. 6). We observed such “sticky” behavior causing background staining with several other fluorescent tyramides, especially with cyanine dye derivatives. For instance, we encountered strong background binding with different working concentrations (2–20 ug/ml) of Cy3-tyramides, made from corresponding NHS-esters that were acquired from two independent vendors (Lumiprobe and Syntol, data not shown). This background binding issue with bulky tyramides was previously mentioned by several authors,^20,21 and paradoxically, is only seen with the dye-tyramine conjugates, but not the free dyes. ²¹

Figure 6.

The comparison of NeuN staining in mouse embryonic cortex using TSA with Sulfo-Cy3 tyramide, and CuAAC-mediated detection with alkyne tyramide and Sulfo-Cy3 azide. The exposure time used to acquire the image is indicated in the upper left corner. Scale bar − 100 µm. Abbreviations: TSA, tyramide signal amplification; CuAAC, copper-catalyzed azide-alkyne cycloaddition; DAPI, 4,’ 6-diamidino-2-phenylindole.

Furthermore, we have assessed the sensitivity of CuAAC-mediated immunostaining using two different primary antibodies against NMDA receptor subunits, Glun1 and Glun2A. Without antigen retrieval, these antibodies show no specific staining in adult mouse brain paraffin sections using the conventional IF protocol (Fig. 7). At the same time, the employment of CuAAC-mediated detection with Sulfo-Cy3 azide afforded strong specific labeling of both proteins in these sections (Fig. 7). This result indicates that compared to conventional protocol, our new method provides much lower detection threshold, facilitating the detection of epitopes, whose preservation may be compromised during fixation and/or paraffin processing of specimen.

Figure 7.

The comparison of Glun1 and Glun2A staining in the adult mouse cortex and hippocampus using conventional IF protocol with Alexa 568 (Glun1) and Alexa 546 (Glun2A) conjugated secondary antibodies and the CuAAC-mediated detection with Sulfo-Cy3 azide. Note the absence of background staining in the corpus callosum above the hippocampus in the bottom right image. Scale bar − 100 µm. Abbreviations: CuAAC, copper-catalyzed azide-alkyne cycloaddition; DAPI, 4,’ 6-diamidino-2-phenylindole.

Discussion

Our analysis of CuAAC reaction conditions has revealed that several marked differences exist between the liquid phase reaction that is described in detail in the literature ¹⁴ and the reaction in fixed cells and tissues that proceeds inside the complex environment of cross-linked protein network (formally, a solid phase reaction). The properties of this network strongly depend on cell type, fixation method and sample processing, as evidenced by the differences in S/N ratios of EdU staining. Basically, the sections of paraffin-embedded cells and tissues exert lower background staining, that is presumably associated with extensive extraction of hydrophobic components during the sample processing. Out of three investigated fluorescent azides, Sulfo-Cy3 displayed the lowest non-specific binding in all specimen, indicating that its amphoteric water soluble character makes it more suitable for CuAAC-based in situ analyses in cells and tissues. In sum, we have found that optimal staining results are obtained with lower azide concentrations (2 µm) and extended reaction time in organic solvent-free aqueous media containing 5 mM free copper catalyst.

For the first time, we have demonstrated, that CuAAC reaction can be used with TSA/CARD in a hybrid detection system as a sensitive immunofluorescent staining method. The oxidation of alkyne tyramide by the immuno-bound peroxidase affords the expected deposition pattern, and shows no evident non-specific binding or signal diffusion. The deposited alkyne tyramide can be efficiently detected with the fluorescent azide using the optimized CuAAC procedure.

Our data show that the novel detection protocol offers several advantages over conventional IF staining and TSA with the fluorescent tyramides. As exemplified by the detection of NeuN protein in the mouse brain, with the same dilution of primary antibodies, the CuAAC-mediated detection affords lower background and stronger fluorescent signals, revealing significantly more NeuN-positive cells than conventional IF (Fig. 5). According to camera exposure time needed to achieve the saturation of NeuN staining, it can be roughly estimated that the sensitivity of CuAAC-mediated detection with Sulfo-Cy3 azide is at least 30 times higher compared to conventional IF protocol with fluorescent-labeled secondary antibody, and 2–5 times higher compared to TSA with Sulfo-Cy3 tyramide (Figs. 5 and 6).

It is well known that TSA/CARD greatly increases the sensitivity of IF staining, but it is also frequently associated with the background issues.^20,21 While some of the background can be eliminated by careful optimization of antibody dilutions, no solutions are currently available to overcome the non-specific tyramide binding. ²¹ In this regard, the CuAAC-mediated detection with alkyne tyramide/Sulfo-Cy3 azide produces significantly less background staining than TSA/CARD reaction with Sulfo-Cy3 tyramide (Fig 6).

The employment of fluorescent-labeled tyramide as the peroxidase substrate implicates a probability of several side reactions: the oxidation of the fluorescent moiety (instead of tyramine residue), as it was demonstrated for several cyanine dyes, ²² and the reaction between the tyramide radical with the fluorescent moiety of another tyramide molecule, leading to its precipitation or quenching. These side reactions are not possible with the CuAAC-mediated detection since the oxidative tyramide deposition and the reaction with the fluorescent dye are performed as separate steps. The advantage of alkyne tyramide over fluorescent tyramides also lies in the smaller size of the former, that is associated with lower steric hindrances between adjacent tyramide molecules during their deposition, and favors the higher density of deposition around the enzymatic reaction site.

The potential applications of CuAAC-mediated detection are obviously not limited to immunocytochemistry/immunohistochemistry, but can also include in situ hybridization and other techniques where peroxidase conjugates with antibodies or streptavidin may be applicable. The CuAAC-mediated detection can be also combined with conventional IF or TSA/CARD with fluorescent or haptenized tyramides in multiplex analyses. Moreover, it can be assumed that with efficient blocking of unreacted alkyne moieties (using a non-fluorescent azide, as described by Liboska et al. ²³ ), it may be possible to employ more than one round of CuAAC-mediated detection for staining of multiple antigens. Taking all the aforesaid into consideration, we believe that the described detection method may be useful for investigators working in the different fields of modern molecular biology.