Signal amplification for multiplexed immunofluorescence assays

The SABER Technology

The Signal Amplification By Exchange Reaction (SABER) method is used for amplifying signal from multiplexed in situ fluorescence staining experiments. Developed by the Yin and Cepko labs at Harvard University and the Wyss Institute, the technique uses Primer Exchange Reactions (PERs) to generate three-letter (A, T, C) concatemeric sequences in bulk in vitro reactions. These concatemers can then be in situ hybridized to fixed cells and tissues and act as scaffolds that localize fluorescent 'imager' strands. The method can further be paired with DNA-Exchange Imaging (DEI) to increase multiplexing via rapid stripping of old imager strands and hybridization of new imager strands (imager exchange) and/or cell segmentation with puncta counting on a per cell basis for quantitative analyses. See the following references and resources below for further information. SABER provides a scalable and cost-effective way to amplify multiplexed in situ stainings for protein targets (Immuno-SABER) and RNA/DNA targets (SABER-FISH, website).


Spatial mapping of proteins in tissues is hindered by limitations in multiplexing, sensitivity, and throughput. To facilitate spatial protein profiling, we have developed a DNA-based multiplexed detection and signal amplification method, called Immunostaining with Signal Amplification by Exchange Reaction (Immuno-SABER) that enables achieving high multiplexing and sensitivity simultaneously. Immuno-SABER leverages: i) multiplexing via DNA-Exchange Imaging (DEI) that enables fast and spectrally-unlimited imaging through single-step immunostaining with DNA-barcoded antibodies, and ii) a new in situ signal amplification method, based on Primer Exchange Reactions (PERs) that autonomously synthesizes long single-stranded DNA concatemers.

Immuno-SABER offers independently programmable signal amplification without in situ enzymatic reactions, and intrinsic scalability to rapidly amplify and visualize a large number of targets when combined with fast exchange cycles of fluorescent imager strands. We have performed proof-of-principle studies in diverse samples (cultured cells, and FFPE, cryosectioned or whole mount tissues) and demonstrated 5–180-fold signal amplification. For our initial demonstration, we performed simultaneous signal amplification for 10 different proteins in mouse retina cryosections. To enable higher multiplexing, we provide a library of 50 orthogonal sequences, 30 of which we validated in situ. We are expanding this list and working on to optimize the workflows for simultaneous multiplexing beyond 10plex.

We also combined SABER with expansion microscopy to enable rapid, multiplexed super-resolution tissue imaging. This application also helps to compensate for the reduction of signal due to dilution of the fluorophores in the physically expanded specimen.


Immuno-SABER: S. K. Saka*, Y. Wang*, J. Y. Kishi, A. Zhu, Y. Zeng, W. Xie, K. Kirli, C. Yapp, M. Cicconet, B. J. Beliveau, S. W. Lapan, S. Yin, M. Lin, E. S. Boyden, P. S. Kaeser, G. Pihan, G. M. Church, P. Yin. Immuno-SABER enables highly multiplexed and amplified protein imaging in tissues. Nature Biotechnology  2019.
*** Supplementary Information
*** User-Friendly Protocols
*** Press release (with animation)

SABER-FISH: J. Y. Kishi*, S. W. Lapan*, B. J. Beliveau*, E. R. West*, A. Zhu, H. M. Sasaki, S. K. Saka, Y. Wang, C. L. Cepko, P. Yin. SABER amplifies FISH: enhanced multiplexed imaging of RNA and DNA in cells and tissues. Nature Methods  2019.
*** Supplementary Information
*** User-Friendly Protocols
*** Additional Supplementary Resources
*** Press release (with animation)

PER (Primer Exchange Reaction) synthesis: J. Kishi, T. Schaus, N. Gopalkrishnan, F. Xuan, and P. Yin. Programmable autonomous synthesis of single-stranded DNA. Nature Chemistry  2018.
*** Primer/hairpin sequences
*** Simulated binding parameters
*** PER synthesis protocol
*** Press release (with animation)

OligoMiner (probe design pipeline): B. J. Beliveau, J. Y. Kishi, G. Nir, H. M. Sasaki, S. K. Saka, S. C. Nguyen, C.-t. Wu, and P. Yin. OligoMiner provides a rapid, flexible environment for the design of genome-scale oligonucleotide in situ hybridization probes. PNAS  2018.
*** Genome-wide DNA FISH probes
*** Instructions for RNA FISH probe set design
*** Computational pipeline (software)

Protocols for Immuno-SABER

User-friendly protocols for Immuno-SABER in cells and tissues can be found in the Supplementary Protocols provided with our recent publication. Please also see the FAQ section below and feel free to reach out to us if you have further questions.

Protocols for SABER-FISH

User-friendly protocols for FISH probe set design, PER, SABER-FISH in cells, and SABER-FISH in tissues can be found in the Supplementary Protocols provided with our recent publication. Please also see the FAQ section below and feel free to reach out to us if you have further questions.

Additional SABER Resources

Here, we provide links to software we used in our publications for (1) segmenting and analyzing 3D SABER-FISH data in cells, (2) analyzing 2D SABER-FISH data in cells, (3) visualizing SABER-FISH data with multicolor overlays, (4) generating probe libraries, (5) designing FISH probes, (6) identifying nuclear contours in human tonsil tissues using a deep learning algorithm (Immuno-SABER), and (7) segmenting nuclei in 2D Immuno-SABER data.

*** PD3D Matlab package for detecting SABER puncta in 3D and assigning them to cells based on membrane marker-based segmentation.
*** CellProfiler pipelines used for 2D segmentation of SABER puncta in cells
*** Image processing functions used for several image alignment and color overlays in the publication.
*** Generating complex Oligopaint probe pools (e.g. for DNA FISH, large-scale RNA FISH).
*** OligoMiner software for designing FISH probe sequences.
*** Deep learning algorithm for automated identification of nuclear contours in human tonsil tissues.
*** MATLAB code for 2D nuclear segmentation of Immuno-SABER data.

Frequently Asked Questions (FAQs)

Sequence Design, Ordering, and Modeling

Q: How do I get started?
A: First, check out the SABER-FISH and Immuno-SABER papers and associated Supplementary Protocols (FISH, IF) for step-by-step probe design, PER, FISH, antibody conjugation, and immunostaining protocols. Then you can download the table of sequences we used in the manuscript here. We recommend starting with the sequences we have already validated before moving on to testing new sequences.

Q: What are the best PER primer sequences?
A: Each primer sequence has different kinetics, and in general with good storage of reagents (see below) we see consistent concatemers for each primer/hairpin combo once we have found a good condition (salt, time, concentration). We recommend starting with primers 25, 27, 28, 30, or 31.

Q: Where do you order reagents?
A: We typically order PER and probe oligos from IDT and larger probe libraries from Twist. Bst Large Fragment polymerase can be ordered from NEB or McLab. We usually order individual dNTPs from NEB.

Q: How should I model binding interactions between different sequences?
A: We recommend using NUPACK to model binding interactions, as well as secondary structure of potential probe/PER sequences. The online tool has great demos for analysing binding probabilities of strands, evaluating likely secondary structures, and designing oligos. The source code can also be downloaded and compiled.

Q: How can I model melting temperature of my sequences?
A: There are several online tools to measure melting temperature (e.g. OligoAnalyzer). For probe design with the OligoMiner pipeline, we use the Biopython Tm_NN function to calculate melting temperatures of candidate probe sequences.

Q: Can I combine any 42-nt bridge sequence with any PER primer?
A: In most cases, these can be arbitrarily combined. We have found a couple example pairs that don't seem to perform well, possibly due to secondary structure causing a mis-priming reaction. We are working to model this and provide an updated list of recommended 42-nt/primer (as well as branch/primer) sequences - stay tuned! In the meantime, we would recommend starting with the sequence combinations we used in the publications.

PER Concatemerization

Q: How do you recommend storing reagents?
A: We store individual dHTPs (dATP, dCTP, dTTP), at high concentration (100mM) and aliquoted to ~20 microliters at -20C. Bst Large Fragment polymerase, oligos, and heat inactivated PER reactions can also be stored indefinitely in the freezer (-20C). PBS, MgSO4, and water can be stored at room temperature or refrigerated (4C), and we recommend aliquotting these as well to further reduce potential for contamination (particularly for RNA experiments).

Q: How much PER concatemer should I make?
A: We like to make anywhere from 100 microliters of each reaction (1 micromolar primer) at a time, which can be enough material for up to 12 experiments. (For an ISH volume of 125 microliters, we would usually use less than 10 microliters of each ~1 micromolar PER concatemer). Once you have found a good extension condition that produces the concatemer of your desired length (see next question), you can also prepare much more material at once and store this bulk material in the freezer indefinitely.

Q: What concatemer lengths do you recommend?
A: For primary probe concatemers, we recommend total concatemer lengths around ~500-750nt. For branch sequences, we found shorter extensions around ~250-450nt work well. These lengths are often determined based on the gel run with respect to double-stranded DNA ladders and can be taken as more of a proxy than absolute lengths.

Q: Should I purify my PER concatemers?
A: We recommend column purification of PER concatemers for tissue experiments (with e.g. MinElute columns from Qiagen). While purification is not usually essential, particularly in cell experiments, it can be helpful to purify out excess buffer/salts from the PER solution (as well as potentially some amount of shorter hairpin sequences). The columns are also an easy way to pool and concentrate concatemers into small elution volumes. See our Supplementary Protocols for our recommended purification protocol. (Importantly, be mindful of the capacity of the column you are using so as not to overload the column and lose a significant amount of material).

Q: How should I optimize the reaction to get concatemers of the right length?
A: There are a number of parameters you can vary (concentration, time) to change the kinetics. Our preferred method is to set a fixed amount of time (e.g. 2 hours or 3 hours) and vary hairpin concentrations accordingly, so that we can run multiple types of PER extension reactions in parallel with the same thermocycler program.

Q: My concatemers look faint on the gel. Did the PER step not work well?
A: This is very normal. PER concatemers are highly single-stranded and therefore can be difficult to image successfully on a gel depending on how powerful your scanner is, so it is very possible your PER step did work fine but you're just having trouble visualizing them. Likely due to their structures, the hairpins (PER hairpin and Clean.G hairpin) are likely to create much stronger bands in the intercalating dye channel than the individual PER concatemer ladder bands. Here is an example Cy5 channel scan of PER run with a Cy5-labeled primer that shows how PER concatemers with very good extension efficiency. Visualizing the same gel in the Sybr Gold Channel shows how strong the hairpin bands can be compared to high efficiency concatemer bands. We have had the best success visualizing the ssDNA with the Sybr Gold dye using our scanner, but if you are having trouble visualizing the concatemers on an E-gel with your imager, you can try pouring a 1% agarose gel with an increased amount of Sybr Gold to increase the signal level, and/or loading more material into each gel lane. It is also good to include a negative control lane without one of the critical PER elements (e.g. without polymerase or hairpin) so that you can see the original probe-primer band does indeed completely disappear in the extension conditions.

Q: My branches are not extending well.
A: The branch sequences tend to be more difficult than standard sequences to extend, likely due to the potential for secondary structure between the branch binding domain (30-nt of A/T/G) and the primer (A/T/C code) resulting in stalled extensions. Our recommendation if you are trying just a few colors is to start with the branch sequences we validated experimentally in the publication. We also provide a list of recommended branch sequences that have gone through some filtering to check for unwanted secondary structure and priming interactions (see Sequences table in our additional Supplementary material). There may still be some unfavorable interactions in that set of branches, and we are starting to look at potentially filtering that list further.


Q: Where can I get oligo-conjugated antibodies from? Do I need to conjugate them myself?
A: Unfortunately, currently commercial antibody vendors do not typically provide oligo-conjugated antibodies for custom sequences and for your preferred antibodies. There are some companies (including but not limited to Expedion, Biolegend, Biosynthesis) that would provide custom oligo conjugation services, but these would typically be pricey and require use of large amounts of antibody. For more practical smaller antibody batches, you can use commercial antibody conjugation kits (that may be available from Thermo Fisher, Expedion, Abcam, Biotium, Biosyn among many other) or use any custom antibody-oligo conjugation protocol. Conjugation kits might be a good option for beginners, but might get expensive when conjugating multiple antibodies. We typically create our own conjugated by utilizing thiol-modified DNA oligonucleotides (bridges) to lysine residues on antibodies in a non-sequence-specific way. But in principle, other conjugation methods (like targeting cysteines and glycosyl groups) should also be fine to use (depending on the antibody). The oligos can be ordered from IDT either in desalted format (economical) or with purification (higher quality) and resuspended in nuclease-free water at 1 mM, and stored at -20°C. We typically order standard desalted format for our own experiments. We design the bridge oligos as follows: /5ThioMC6-D/ tt Bridge sequence (42mer) The step by step protocol for conjugation is available here.

Q: Doesn’t the antibody conjugation interfere with the antigen binding?
A: When we perform non-specific conjugation, there is a possibility that the antigen binding site (paratope) may be modified and become less or non-functional. We try to optimize the crosslinking conditions (such as the crosslinker and oligo concentration) such that we obtain conjugates with 1-3 oligos/antibody to minimize the chances that the lysines in the paratope get modified. We find that most antibodies work well after conjugation, but we nevertheless recommend to perform a control to test the specificity of the conjugated antibody in situ. This can be done by performing conventional immunostaining with fluorophore-conjugated secondaries for both the conjugated and unconjugated primary antibody and compare the staining patterns. Note that during conjugation a portion of the antibody may be lost, so the concentration need to be re-adjusted accordingly. For antibodies that show sensitivity to conjugation, site-specific conjugation methods such as click-modification of the glycosyl groups (e.g. Siteclick kits from ThermoFisher) can be employed.

Q: I already have an optimized immunostaining protocol for my tissues? Do I need to change that to be able to perform Immuno-SABER?
A: Although we were not able to test our current protocols with all the possible types of tissues and preparations, we do not expect a huge problem if your samples are normally compatible with conventional immunofluorescence and FISH. We expect you should be able to use your common protocol with some modifications and appending them with the Immuno-SABER protocol. These modifications include adding additional blockers (e.g. sheared salmon sperm DNA and dextran sulfate) in the antibody incubation buffer to avoid nonspecific interactions that may come from the oligo. After washing the excess of antibodies we recommend a post-fixation step to make sure that the antibodies are not washed later during the concatemer hybridization, washing and imager displacement steps. After post-fixation, you can continue with the SABER protocol we provide, and make further adjustments if required for the particular tissue.

Q: What are the common problems I may encounter? How do I troubleshoot?
A: Here are a list of issues that might come up when starting to apply the method with new antibodies/ sequences or on different sample preparations, and recommended steps to troubleshoot:

Issue Potential reason Potential solution
No specific signals in the conjugated-antibody testing experiments 1. The antibody might have lost binding capability after DNA conjugation
2. The antibody concentration after conjugation is too low
3. The antibody penetration is not as good.
1. We noticed that the antigen-binding capability may be lost after DNA-conjugation for some antibodies. Without knowing the protein sequences of the antibodies, it is difficult to find the exact cause for the loss of antibody activity. It is always a good idea to check DNA-antibody conjugation products using SDS-PAGE gel electrophoresis. Antibodies conjugated with too many DNA oligonucleotides may lose binding activities. Decreased amount of crosslinkers could be used to reduce the number of DNA oligonucleotides per antibody. Site-specific conjugation (e.g. Siteclick from ThermoFisher) could also be tested to see if it solves the problem. Alternative antibodies binding to the same target can also be tested.
2. The antibody concentration can be estimated by comparing the gel band intensity with a control with known antibody concentration. The typical recovery rate of our DNA-conjugation protocol is ~50%. A fraction of the antibodies may be lost during the Amicon column filtering steps and washes. Make sure that the columns are of high quality and not broken.
3. Might be a potential concern for some intracellular targets, increasing the detergent concentration for better permeabilization effect might be helpful.
I see nuclear background in addition to my expected staining There could be a few different steps that may cause this:
1. Antibody-oligo conjugation
2. Concatemer hyb
3. Branch hyb (if applicable)
4. Imager hyb
First, try to identify where the background is coming from, by omitting different components and seeing under what condition the unspecificity persists. You can perform one or more of these suggestions depending on the source of the problem.

If it is due to reason 1:
- Decrease the antibody concentration
- Clean up/purify your conjugate to get rid of the unbound strands better
- Perform more/higher stringency washes before post-fixation
- Use higher concentration of blockers (especially dextran sulfate) during the antibody incubation
- Perform the antibody incubation at higher temperature for a shorter time.

If it is due to reason 2 or 3:
- Decrease the concatemer concentration
- Purify your concatemer to get rid of the leftover reaction components
- Increase the stringency of the concatemer incubation and wash steps (e.g. higher temperature)

If it is due to reason 4:
- Decrease the imager concentration
- Apply additional blockers (such as short single stranded DNA mixtures) before or during the imager incubation
- Increase the stringency of the imager incubation and wash steps (higher temperature, small amounts of formamide)
- Consider using a different fluorophore (some fluorophores are my hydrophobic and can lead to unspecific membrane stainings)
DNA-conjugated antibodies work well at the beginning but stop working after a while DNA-conjugated antibodies are improperly stored Long-term storage of DNA-conjugated antibodies is challenging. We noticed that some antibodies lose their binding activities when stored at 4 degree for several months. Storing antibodies with glycerol at -20 degree could also be used. However, we had mixed experience with this storage condition with some antibodies showed decreased binding activities after a short-term storage. We recommend to add sodium azide, EDTA and BSA in the storage buffer. Sodium azide can prevent growth of bacteria; EDTA can inhibit enzymatic degradation of the oligo by DNAses and BSA can reduce the absorption of antibodies to the storage tube. Reducing antibody storage time and preparing new batches of DNA-conjugated antibodies every 2-3 months is recommended.