Comparative Evaluation of Antibody-Oligonucleotide Conjugation Strategies for Multiplexed Imaging Applications

Abstract

Antibody oligonucleotide conjugates (AOCs) have emerged as versatile tools with applications spanning diagnostics, therapeutics, and high-dimensional imaging. One major application of these is in multiplexed imaging techniques such as Co-detection by imaging (CODEX) that allow for the visualization of tissue networks at the single-cell level. In this study, we evaluated four methods—maleimide-modified, amine-modified, DBCO-modified, and a site-specific enzyme-based method—to optimize the generation of AOCs for multiplexed imaging applications. Our assessment focused on key performance parameters including conjugation efficiency, signal brightness, stability, reproducibility, and cost-effectiveness. Each conjugation chemistry proved effective, though the azide chemistry with DBCO oligonucleotides demonstrated more consistent conjugation success and stable signal retention over time. Compared to other protocols, this method produced reliably bright images and offered a more favorable cost profile, as further confirmed in a full-scale CODEX multiplexed imaging experiment that yielded reproducible spatial data. The observed stability and reproducibility of the DBCO approach suggest that it may help reduce reagent waste and labor costs while facilitating the development of more comprehensive antibody panels. These findings indicate that the DBCO-modified oligonucleotide conjugation method is a valuable option for generating AOCs for multiplexed imaging and target current shortcomings, enabling more consistent, broader, and deeper multiplexed profiling.

Introduction

Antibody oligonucleotide conjugates (AOCs) are powerful, multi-modal entities that have impactful applications ranging from diagnostics and therapeutics to imaging. First established nearly four decades ago¹, AOCs have seen an explosion in recent years based on the number of increasing applications². For example, there are at least six AOC constructs in clinical trials and wide-spread development of AOCs for use in immuno-PCR³, proximity ligation⁴ or extension⁵ assays, protein arrays and multiplexed imaging. In addition to antibody-drug conjugates their antigen recognition power has also been harnessed for imaging, through immunohistochemistry (IHC) and immunofluorescence (IF).

Conjugation with oligonucleotides has allowed antibody-based imaging to advance past the handful of markers that methods like flow cytometry, IF, and IHC have been historically limited to. Cell-specific DNA barcodes have been widely employed in single-cell sequencing^6–10, and barcoding methods are being continuously refined¹¹. Recent developments in multiplexed protein-imaging technologies now allow for spatially resolved single-cell data acquisition across tens to hundreds of different markers^12–16.

Antibodies tagged with oligonucleotide barcodes are the basis of a host of multiplexed imaging technologies, including CO-Detection by indEXing (CODEX)^17–19, immune-SABER²⁰, DNA Exchange Imaging (DEI)²¹, and Exchange-PAINT²² that enable multiplexed imaging through oligonucleotide exchange (Fig. 1A). These technologies have produced multidimensional single-cell spatial datasets, leading to the development of new cell-type annotation techniques^23,24 and revealing complex interactions across intercellular networks and between cells (e.g., cell-cell signaling and motile pathways^25,26). Such data has also provided insight into organ function, by uncovering organ substructures (e.g., immune follicles¹⁷ and intestinal crypts²⁵) and elucidating tissue architecture and organizational motifs²⁷.

Figure 1:

Optimization of different antibody-oligonucleotide (oligo) conjugation protocols for multiplexed protein imaging. (A) Schematic of the CODEX workflow for multiplexed protein imaging using antibody-oligo conjugates and their fluorescent complements. (B) Conjugation schematic for non-specific chemistries of DBCO-, amine-, and maleimide-modified oligos and site-specific chemistry. (C) Reliability of maleimide-modified oligos was tested by performing 10 TCR conjugations and using each conjugate to stain mouse spleen. Graph indicates resulting signal viability across the 10 conjugates. (D) Consistency of signal of maleimide-modified oligos across batches. Three different maleimide-modified oligos with known signal strengths were each conjugated to TCR in three batches and signal strength was compared. (E) Image of splenic stain TCR-oligo 77 conjugation 1 (scale bar = 100 μm). (F) Quantitative comparison of signal strength resulting from different oligo-to-antibody and azide-to-antibody ratios for DBCO-modified oligo protocol. (G) Image of mouse splenic tissue stained with TCR-DBCO-oligo (200:1, 25:1) (scale = 50 μm). (H) Quantitative comparison of signal strength resulting from different oligo sequences and oligo-to-antibody ratios for the amine oligo protocol. (I) Image of mouse splenic tissue stained with TCR-amine-oligo (Sequence 1, 2:1) (scale = 50 μm).

Achieving successful conjugations with maleimide-based oligos is costly and time consuming. For an average multiplexed imaging panel of 60 antibodies, it requires at least $60,000. We and others have observed variation in conjugation of antibody-oligonucleotide conjugates as it can take multiple rounds of conjugation attempts before a successful conjugate is produced. Previously, many labs have used pre-modified maleimide-conjugated oligonucleotides due to ease of conjugation; however, maleimide groups are sensitive to chemical degradation.

The importance of AOCs in multiplexed imaging—as well as in therapeutics and diagnostics—has led to many efforts to optimize and discover novel antibody-oligonucleotide conjugation methods. Modifications made to the oligonucleotide allow attachment to the antibody through lysine or cysteine residues in a nonspecific manner. However, for some IgG isotypes, a site-specific conjugation strategy can be applied due to a conserved sugar on an amino acid sequence that allows for the attachment of the oligonucleotide in the heavy chain²⁸.

In this study, we evaluated and optimized three non-specific distinct conjugation chemistries and one site-specific method for attaching oligonucleotides to antibodies for CODEX multiplexed imaging. Our analysis included multiple parameters, including reagent reliability, fluorescence intensity, stability, consistency, and cost-effectiveness. By assessing these variables, we aimed to identify conjugation strategies suited for generating large-scale antibody panels for multiplexed imaging applications. Our approach helps provide data on optimal reagents and protocols for AOC synthesis and also provides a robust framework for selecting methods to create extensive, high-quality antibody panels for complex imaging experiments.

Results

Chemistries evaluated for attaching oligonucleotides to antibodies

To understand trade-offs between sensitivity, specificity, and reliability, we optimized and evaluated four different direct non-specific conjugation chemistries. We chose these methods for their positive track record, accessibility, and application potentials across fields. Two methods take advantage of the functional group maleimide, which connects with thiol groups on the antibody. The first method (Method 1) attaches maleimide to the oligonucleotide via direct modification^18,29–31 (Fig. 1B), while the second (Method 2) attaches maleimide to amine-modified oligonucleotides via sulfosuccinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate) (‘sulfo-SMCC’)^32–34 (Fig. 1B). The third method (Method 3) harnesses the power of copper-free click chemistry, also known as strain-promoted alkyne-azide cycloaddition (SPAAC), a bioorthogonal conjugation technique that has been used for therapeutics, imaging, and analytics^35–38. For this method, we reacted dibenzocyclooctyne (‘DBCO’)-modified oligonucleotides with antibodies whose amine groups we had attached to azidoacetic acid N-Hydroxysuccinimide ester (‘NHS-ester azide’), moieties (Fig. 1B). The fourth method we evaluated was a site-specific enzyme chemistry (Method 4) where azides are generated on the sugar moiety of the Fc portion of the antibody (Fig. 1B).

Benchmark Testing of Maleimide-Modified Oligonucleotide Antibody Conjugation

Motivating this study was our experiment to assess the reliability of pre-activated, stored maleimide-modified oligos (Method 1). We did conjugations of 10 unique oligo sequences that we had ordered and stored for a duration of one month to two years to an anti-mouse TCR antibody. We evaluated these conjugations by using them to stain fresh frozen mouse spleen tissue sections. Only three of the conjugates produced staining with a signal-to-noise ratio appropriate for use in a multiplexed experiment (Fig. 1C). Three produced unusable signal with insufficiently high signal-to-noise ratios, and four failed to produce signal altogether (Fig. 1C).

This was not due to variability in the conjugation process itself or to staining procedures because three distinct conjugations of three distinct oligonucleotide sequences to anti-TCR yielded reproducible intra-oligo signal and the same clone that is known to work was used (Fig. 1D, E). Since the stored maleimide-modified oligos had been previously validated as working, this indicated problems with long-term storage of maleimide activated oligos. This lack of reproducibility is both costly and time-consuming since it requires troubleshooting and ordering new modified oligonucleotides. Moreover, for conjugating to new antibody clones it is not always clear if the problem was the antibody or oligonucleotide barcode.

Optimization of Azide-Mediated Oligonucleotide Antibody Conjugation Reactions

We previously optimized the pre-activated maleimide oligo methodology (Method 1), so we also optimized conjugation protocols for the two new non-specific conjugation methods. To optimize conjugation efficiency, conjugation parameters (including conjugation incubation time and temperature, number of washes, incubation solution composition, storage conditions (data not shown), oligo to antibody ratio, sulfo-SMCC to oligo ratio, and NHS-ester azide to antibody ratio) were varied. For each condition, three different oligo sequences were conjugated to either anti-TCR or anti-B220 antibodies and evaluated by staining on mouse spleen. The greatest signal obtained from DBCO-modified oligo condition (Method 3) was obtained with a 200:1 azide-to-antibody molar ratio and a 25:1 oligo-to-antibody molar ratio (Fig. 1F, G). The greatest signal in the amine-modified oligo condition (Method 2) was obtained with a 60:1 oligo-to-antibody ratio (Fig. 1H, I).

DBCO-Modified Oligonucleotides Produce Higher Staining Signal than Site Specific Chemistry

Since the azide and DBCO conjugation method worked most consistently, and because others have shown site-specific chemistry with azide modified sugars we compared this chemistry to a site-specific enzymatic method²⁸. This fourth method applied the site-specific enzyme chemistry (Method 4) where azides are generated on the sugar moiety of the Fc portion of the antibody and already optimized since we purchased a commercial kit. The DBCO-modified oligonucleotides were then reacted with the reactive azide group, similar to the DBCO-modified oligo chemistry method (Method 3) (Fig. 1D). We conjugated an anti-TCRβ using both the site-specific chemistry and the DBCO-modified oligonucleotide method and observed that the site-specific chemistry at 10 ms resulted in an image that was 2.7 times less bright than the DBCO-modified oligonucleotide method (Supplemental Fig. 1).

Comparative Analysis of Conjugation Methods: Reproducibility, Reliability, and Efficiency

We continued to compare the first three non-specific optimized methods, by evaluating signal reproducibility, signal reliability, and cost (which includes the materials and labor). We defined signal reproducibility in terms of the rate of successful conjugation across different antibodies and oligo sequences. We defined signal reliability in terms of the rate of successful conjugation after different time points of oligo and reagent storage. We calculated the cost based on the prices of the materials and on the time required for each protocol as labor cost.

Reproducibility Testing

To test the reproducibility of the conjugation methods across different oligo sequences and antibodies, we performed test conjugations with five DBCO-modified oligos (Method 3) and five amine-modified oligos (Method 2) on anti-TCR, anti-B220, or both antibodies. These first conjugations were done upon receiving the oligos from the vendor and served to validate the reproducibility of the methods at “Time 0” (i.e. before storage) (Fig. 2A). The conjugations were evaluated by testing the conjugates on mouse spleen (Fig. 2B) and determined each conjugation according to a three-tier classification schema of “good signal” (useable for a CODEX imaging), “dim signal” (detected signal but not useable because of low signal to noise ratio) and “no signal.” Of the 10 different conjugation pairs (5 different sequences, each conjugated to two antibodies) tested on the DBCO-oligo protocol (Method 3), 9 of the conjugations resulted in good signal and 1 in dim signal at Time 0 (Fig. 2A). Of the 5 different conjugation pairs (one antibody tested per oligo sequence) tested on the amine-oligo protocol (Method 2), each resulted in good signal at time 0 (Fig. 2D).

Figure 2:

Comparisons of reliability, reproducibility, and cost across different conjugation methods. (A) Evaluation of signal from different DBCO-modified oligo conjugations over increasing amounts of oligo storage time. Each circle indicates a separate conjugation. (B-C) Images of mouse spleen region stained with a TCR-DBCO-oligo 7 conjugate conjugated after 3 months (B) and 10 months (C) of oligo storage (scale = 100 μm). (right) Magnified view of the regions indicated with the green (B) or pink (C) box in the left image (scale bar = 50 μm). (D) Evaluation of signal from different amine-modified oligo conjugations over increasing amounts of oligo storage time. Each circle indicates a separate conjugation. (E) (left) image of mouse spleen region stained with a TCR-amine-oligo 65 conjugate conjugated after 5 months of oligo storage (scale = 100 μm). (right) Magnified view of the region indicated with the turquoise box in the left image (scale bar = 50 μm). (F) Comparison of the rates of positive signal (proxy for oligo reactivity) across conjugation methods broken down by 1–5 months or 6–11 months of storage. (G) Comparison of the cost, time, and success rate between different conjugation methods by cost of materials, labor time, and success rate indicated by dot size (higher success rate = bigger size). (H) Comparison of total cost (including raw material and labor time) between different conjugation methods.

Reliability Testing

To test the reliability over time of the different conjugation protocols, we carried out new conjugations with the same oligos and antibodies after the oligos had been stored for 1, 3, 5, 7, 10, and 11 months (Fig. 2A, D). The conjugations were tested, as before, by performing stains on mouse spleen and examining the signal (Fig. 2B, C, E). In mouse spleen stains of the DBCO oligo method, the results did not differ much between any of the time points, including the oligos stored for 10 months. However, for the amine-modified oligo method, the staining results at later timepoints were dimmer, and after an 11-month period, only 40% of the conjugations were still reactive (Fig 2F).

In total, 53 conjugation tests were performed, and the reliability rates were calculated across methods. Measured across all time points overall, the DBCO-modified oligos (Method 3) retained an 87.5% signal viability rate (21/24 conjugations) across validated oligo conjugations (Fig. 2F). The amine-modified oligos (Method 2) retained a 45% signal viability rate (5/11 conjugations) across validated conjugations (Fig. 2F). The maleimide oligos (Method 1) were tested during our initial viability testing with our pre-existing maleimide store, which included oligos that were a few months to 2 years old; these oligos retained a 30% signal viability rate (3/10 conjugations) (Fig. 1C).

Of the conjugations performed across validated oligos after 1–5 months in storage, the DBCO-modified oligos (Method 3) again outperformed the other methods, with an 84% signal viability rate (11/13 conjugations) (Fig. 2F). The amine-modified oligos (Method 2) showed a 50% signal viability rate (Fig. 2F). After 5–11 months in storage, the amine oligos (Method 2) dropped to a 40% signal viability rate, in sharp comparison to the 100% signal viability rate of the DBCO-modified oligos (Method 3) (Fig. 2F). We concluded that the DBCO-modified oligos were significantly more reliable than the other two modalities at the time scale of months of reagent storage that is often required by labs building multiplexed imaging panels.

Efficiency Testing

We measured efficiency in terms of cost of materials and the time spent on each protocol. Where appropriate, failure rates of each conjugation method and materials lost (due to washing, etc.) during conjugation were factored into cost calculations for as accurate an appraisal as possible and incorporate raw material costs and raw material and labor costs (Fig. 2G). The DBCO-modified oligo chemistry method (Method 3 - $22 per 50 μg antibody conjugation) was significantly cheaper than the amine-modified oligo (Method 2 - $44 per 50 μg antibody conjugation), maleimide-modified oligo conjugation methods (Method 1 - $60 per 50 μg antibody conjugation), and much higher site-specific costs (Method 4 - $75 per 50 μg antibody conjugation) (Fig. 2H). We have also tested the reversed conjugation method for DBCO-click chemistry where azide-modified oligonucleotides were added to DBCO-modified antibodies (Supplemental Fig. 2). The the efficiency and cost ($20 per 50ug antibody conjugation) of this method is similar to DBCO-modified oligo method but provides an intermediate step where the number of DBCO molecules added to each immunoglobulin can be measured by the absorbance of reactive DBCO at 309 nm (Supplemental Fig. 2B, C).

Comparing DBCO-modified oligo conjugation to maleimide for multiplexed antibody panel creation and multiplexed imaging

Given that the DBCO-modified oligo (Method 3) produced strong, reliable, reproducible signal, and was the cheapest conjugation method, we sought to verify that the method could be used in a full-scale CODEX multiplexed protein imaging experiment. We compiled a 15-marker antibody panel consisting primarily of immune and tumor cell markers (Fig. 3A) to conjugate and stain on mouse spleen and tumor tissues (Supp. Table 1). After the antibodies were conjugated to 15 unique DBCO-modified oligo sequences, the staining was performed, and the CODEX assay was run on a mouse spleen and tumor (Fig. 3B, C). We also performed another stain on additional sections from the same mouse spleen and tumor tissues, using a pre-existing validated panel of the same antibody clones conjugated to maleimide-modified oligos (Method 1) (Fig. 3D, E). Both panels provided similar staining and were separately examined for positive staining with good signal to noise ratios for each marker (Fig. 3F).

Figure 3:

CODEX multiplexed imaging of mouse spleen and tumor tissue with DBCO-modified and maleimide-modified oligo antibody panels. (A) A 15-antibody panel was chosen to discriminate adaptive and innate immune cell populations, as well as endothelial and tumor cells. (B) (left) Image of a mouse spleen stained with the DBCO-modified oligo-antibody panel (scale bar = 1000 μm). Six representative markers are highlighted: B220 (green), CD8a (red), CD4 (blue), CD31 (pink), CD11b (yellow). (right) Magnified view of the region indicated with cyan box in the image of the spleen (scale bar = 100 μm). (C) (left) Image of a mouse tumor stained with the DBCO-modified oligo-antibody panel (scale bar = 1000 μm). Five representative markers are highlighted: CD45 (green), 16/32 (blue), CD3 (red), CD31 (pink). (right) Magnified view of the region indicated with cyan box in the image of the tumor (scale bar = 100 μm). (D) A magnified view of mouse spleen stained with the maleimide-modified oligo antibody panel (scale bar = 100 μm). Six representative markers are highlighted: B220 (green), CD8a (red), CD4 (blue), CD31 (pink), CD11b (yellow). (E) A magnified view of mouse tumor stained with the maleimide-modified oligo antibody panel (scale bar = 100 μm). Five representative markers are highlighted: CD45 (green), 16/32 (blue), CD3 (red), CD31 (pink). (F) Magnified images of mouse spleen stained with different DBCO-modified oligo-antibodies (scale bars = 25 μm). Top row from left to right: CD19, 16/32, CD8a, CD90, B220. Bottom row from left to right: CD11b, CD31, CD4, CD3, CD45.

Quantification of DBCO and maleimide chemistries used for multiplexed tissue imaging

Next, we performed cell segmentation, and cell types were each hand-gated for the DBCO-conjugated panel (Fig. 4A) and for the maleimide conjugated panel (Fig. 4B). Immune, tumor, and epithelial cell types were defined and quantified, and cell-type percentages were calculated for both runs. The two methods yielded concordant estimates of cell type proportions across both high-level cell classes (e.g. tumor, CD45+, endothelial, fibroblasts) (Fig. 4C, D) and at higher-resolution sub-types within these categories (Fig. 4E–H). This indicates that the DBCO method could be used to perform CODEX multiplexed imaging with large antibody panels and still achieve high quality data as with our previously published method.

Figure 4:

Gating and cell type population quantification comparisons for DBCO-modified oligo vs. maleimide-modified oligo antibody panels on mouse spleen and tumor tissue. (A-B) Gating of immune cell populations in mouse splenic tissue stained with DBCO-modified oligo conjugates (A) or maleimide oligo conjugates (B). 1 of 2 regions shown. Top row, left to right: i. spatial plot of x, y positions cells gated based on quantified fluorescent signal for DAPI nuclear stain, linear scale. ii. Nucleated cells gated based on cellular size and DAPI signal. iii. Immune cells from nucleated cell population, gated based on CD45 expression. Second row, left to right: iv. T cells from immune cell population, gated based on CD3 and TCRb expression. T cells shown within the red gate. v. T cell sub-populations from T cell population, gated based on CD8 and CD4 expression. CD4-CD8+ cells shown in upper left, CD4+CD8+ (DP) shown in upper right, CD4-CD8+ shown in lower right, CD4-CD8- (DN) shown in lower left. Third row, left to right: vi. B cells from the not T cell population (everything excluded from the gate in iv.), gated based on B220 expression. B cells shown within red gate. vii. CD19hi B cells from the B cell population, gated based on CD19 expression. CD19hi B cells shown within red gate. Fourth row: viii. 16/32hi cells from the not B cell population (everything excluded from the gate in vi.), gated based on 16/32 and H2Db expression. 16/32 cells shown within red gate. Everything excluded from 16/32 population (everything outside red gate) labeled other immune. Axes scale types: i: linear. ii: log. iii-viii: arcsinh. (C-H) Cell type compositions quantified from CODEX images for comparing DBCO-modified oligo panel and maleimide-modified oligo panel for (C) mouse spleen stains for all cells, (D) mouse tumor stains for all cells, (E) mouse spleen stains for immune (CD45+) cells and (F) non-immune (CD45-) cells, (G) mouse tumor stains for immune (CD45+) cells and (H) non-immune (CD45-) cells. (I-K) Stacked bar plot comparisons in mouse spleen stained with the DBCO-modified oligo antibody 1 month after panel conjugation and 3 months after panel conjugation for (I) all cells, (J) immune (CD45+) cells, and (K) non-immune (CD45-) cells.

Assessing Stability of DBCO AOC Panel Over Time and Across Applications

To ensure the stability of the DBCO-modified oligo panel over time and the reproducibility of our initial results, we repeated the spleen and tumor stain and multicycles three months after the creation of the panel. After qualitative evaluation and conclusion that the staining patterns were the same, we segmented and hand-gated the data as before. We then performed the same cell-type quantifications and compared the results with the original DBCO panel results. Cell type percentages are compared in Fig. 4I–K.

To further assess the stability of the DBCO-modified antibodies, the 15-marker panel was run in three consecutive cycles for a total of 45 cycles. Signal intensities remained consistent across all three rounds and antibodies, indicating that repeated fluorophore introduction and removal did not affect antibody performance (Supplemental Fig. 3). These findings support the robustness of the conjugation method for larger panels and extended multicycle experiments.

We also validated this method across a number of species and antibody targets. The method has performed reliably across diverse tissue types, including mouse heart, human intestine, human brain, and human tonsil (Supplemental Fig. 4). Furthermore, while our earlier panels focused largely on immune markers, the technique also works well for non-immune markers such as podoplanin, CD138, PDGRFb, and PNAd. Together, these findings support the broader applicability of this conjugation approach beyond immune-focused panels, offering a flexible and reliable method for building customized, large-scale CODEX antibody panels across biological system. We conclude that the DBCO-oligo conjugation method (Method 3) produces robust data, while performing well for efficiency (time and cost), reliability, and reproducibility.

Discussion

We have previously noticed variability in antibody conjugation quality and efficiency. Our results here indicate that this is primarily related to maleimide-modified oligonucleotide storage. The maleimide-modified oligos are shipped with a 2,5-dimethylfuran cycloadduct protecting group, which must be removed with toluene and heat to deprotect and activate the oligos for conjugation. The activation process is time-consuming and results in high amounts of oligo loss, necessitating it to be performed on large batches of oligo at a time. The oligos are subsequently stored without the protecting group, rendering the unprotected maleimide-groups vulnerable to degradation. Maintaining large panels and high-quality oligonucleotides consequently can be costly and time-consuming. For laboratories performing routine multiplexed imaging other chemistry approaches are necessary to make data generation more reliable and cost-effective.

In this study, we conducted a comprehensive evaluation of four distinct oligonucleotide-antibody conjugation methods, focusing on their performance, reliability, and efficiency. Our findings demonstrate that the DBCO-modified oligonucleotide method (Method 3) worked well in each of these areas. The DBCO-modified method exhibited reproducibility, with 90% of conjugations resulting in good signal at initial testing. Moreover, it displayed high reliability over time, maintaining an 87.5% signal viability rate across all time points and a 100% viability rate after 5–11 months of storage. The DBCO method also proved to be the most cost-effective, at $22 per 50 μg antibody conjugation where the most expensive site-specific method would cost nearly $75 per 50 μg antibody conjugation (Method 4). In contrast, the maleimide-modified (Method 1) and amine-modified (Method 2) oligonucleotide methods showed lower reliability and intermediate costs.

When applied to a full-scale 15-marker CODEX multiplexed protein imaging experiment, the DBCO method yielded robust and reproducible results comparable with current maleimide-based methods.²² This full-panel testing was significant for two reasons. First, it confirmed that the DBCO method was compatible with an entire panel of antibody and oligo conjugates. This required the ability to produce strong and specific enough signal to characterize differential protein expression to use for cell type identification and quantification. Second, it confirmed that the conjugates were stable enough to undergo the multiple rounds of stripping washes that the CODEX assay entails. This not only verified stability for CODEX staining but suggests the potential for the method to be used in other research contexts. These findings collectively affirm the DBCO-modified oligonucleotide conjugation method as another reliable choice for creating stable, efficient, and high-performing antibody-oligonucleotide conjugates for multiplexed imaging applications.

DBCO-modified antibody conjugation method has also demonstrated versatility across a range of tissue types and cellular targets. Over 50 markers have been successfully conjugated using this approach and incorporated into various CODEX imaging panels beyond data reported here. These antibodies have been applied in human, pig, rat, and mouse tissues, with consistent staining observed across epithelial, immune, neural, and endothelial cell populations.

The optimized DBCO-modified oligonucleotide method offers advantages in antibody-oligonucleotide conjugate (AOC) stability and efficiency. This improved stability reduces the need for frequent re-conjugations, minimizing waste and resource consumption. Conjugated antibodies also remain viable for at least a year, helping with long-term experimental planning and resource management.

By reducing reagent waste and labor costs, more labs can build custom multiplexed imaging panels, further democratizing the technology. Additionally, labs can allocate more resources to add additional markers to their antibody panels. This expansion allows for the inclusion of additional protein markers, potentially leading to more comprehensive multiplexed imaging, increased spatial data, and more insight into tissue networks and cellular interactions. The increased number of markers could be used to identify further cell subtypes that are more rare or functional markers that give insight to how cell-cell positions relate to mechanisms. In addition to increasing the breadth of markers, a reduced cost and increased efficiency also enable expansion of the number of barcodes or total antibodies in a panel. Previous panels largely capped around 58 barcodes could now expand with the design of new oligonucleotide barcodes to higher numbers. These advancements may also facilitate broader applications of AOCs beyond imaging—enabling their use in diagnostic assays, targeted therapies, and emerging biotechnologies.

Limitations of the study

While this study evaluated several oligonucleotide conjugation methods, it did not exhaustively compare all available techniques. Another limitation is that while this was used across 15 different antibodies, it was not screened against hundreds of antibodies. With a method that uses non-specific lysine chemistry, this could impact the binding sites on Fab portions of some antibodies. Another limitation was that our DBCO comparison to site-specific chemistry was limited and leveraged protocols derived from the manufacturer. Optimizing the site-specific chemistry could provide a better comparison to DBCO chemistry which constrains direct comparisons between these tests and the other three methods we optimized. However, the two DBCO conditions were evaluated using identical reagent quantities and protocols, allowing for valid comparisons. Additionally, increasing the time monitoring the reagents over a longer period, such as two years, could provide more insights to reagent stability. Our future research will continue to evaluate other antibody chemistry conjugation strategies alongside the methods evaluated here.

Materials and Methods:

Antibody Oligonucleotide Conjugations

Three different conjugation protocols were tested. On average, 8 antibodies were conjugated at a time. Each conjugation iteration included a validated antibody and oligonucleotide pairing from a validated antibody batch and a validated oligonucleotide batch to act as a positive control. If the positive control conjugate failed to produce signal, the entire conjugation batch was discarded. The procedures are briefly described in the following:

Conjugation Materials:

The DBCO-modified and amine-modified single-strand oligonucleotides were purchased from Stanford School of Medicine’s Protein and Nucleic Acid (PAN) Facility. The maleimide-modified single-strand oligonucleotides were purchased from Integrated DNA Technologies (IDT). Commercial antibodies were purchased from BioXCell (#BE0102, anti-mouse TCRB) (#BE0067, anti-mouse B220); full antibody panel details can be found in (Supplementary table 1). Azidoacetic acid NHS ester was purchased from Sigma-Aldrich (#900919). Dulbecco’s Phosphate Buffered Saline (DPBS) was purchased from Thermo Fisher Scientific (#14190144). TWEEN 20 was purchased from Sigma-Aldrich (#P1379). EDTA was purchased from Teknova (#E0308). Dimethyl Sulfoxide (DMSO) was purchased from Sigma-Aldrich (#472301). Candor PBS antibody stabilizer was bought from Thermo Fisher Scientific (#nc0436689).

Maleimide-modified oligonucleotide conjugation:

The maleimide-modified oligonucleotides are first deprotected in an activation step, then stored for subsequent conjugation. Before conjugation, the Amicon centrifuge tube is blocked with 500 μL of PBS-TWEEN. After spinning, the flowthrough and column were decanted. The antibodies’ concentration was determined using a NanoDrop. From the concentration the volume of antibody was calculated. The antibodies were washed in 400 μL of PBS. After washing the antibodies, 360 μL of tris (2-carboxyethyl) phosphine (TCEP) was added and incubated for 30 minutes. The antibodies were spun down and the flow through was decanted. The antibodies were then washed with 400 μL buffer C. In a separate tube, the lyophilized maleimide modified oligonucleotide was combined with buffer c at a ratio of 200 μL of buffer C to 100 μg of oligonucleotide. After resuspension, 200 μL of high salt buffer C was then added to the tube. The entire contents of the tube were then added to the column and the solution incubated for at least 2 hour at room temperature. The ratio of oligonucleotide to antibody was 2 to 1 respectively. After 2 hour incubation, the columns were spun down and then washed three times with 450 μL of high salt PBS. Lastly the antibodies are resuspended in CODEX antibody stabilizer solution at a ratio of 2 μL of CODEX stabilizing solution to 1 μg of antibody before inverting the column into a new tube for collection.

Amine-modified oligonucleotide conjugation:

Amine-modified oligonucleotides are dissolved in 1,000 μL water, and then 25 μL of 1M pH 8.5 bicarbonate buffer solution was added. Sulfo-SMCC was dissolved in DMSO so that the total volume was less than 300 μL per tube. The oligonucleotides and sulfo-SMCC are combined at a 60:1 sulfo-SMCC to oligonucleotide molar ratio and left to incubate for one hour at room temperature. Add 30 uL (1/10^th the volume of the total oligo-sulfo-SMCC solution, adjusted accordingly) of 3M pH 5.2 sodium acetate to a separate tube. After incubation aspirate the clear solution and mix with the sodium acetate, note the total volume of the solution. To the oligonucleotide solution add 3:1 ethanol to the solution by volume. Then centrifuge for 10 minutes at 14,000. Aspirate the clear volume out and do not touch the pellet of oligonucleotide at the bottom. Add 70% ethanol added to the remaining precipitate and centrifuged for four minutes and repeat once more. Buffer C is added to the remaining precipitate at a 50 μL to 1 mg buffer to oligonucleotide ratio. The concentration of the resulting mixture is measured on the nanodrop, and the mixture is divided into known quantities and put into PCR tubes. The PCR tubes are flash frozen in liquid nitrogen and put into the freezer for storage until subsequent conjugation. Subsequent conjugations are carried out as described for the maleimide-modified oligonucleotide protocol, save for the oligonucleotide activation step.

DBCO-modified oligonucleotide conjugation:

DBCO-modified oligonucleotide quality is tested by nanodrop analysis; DBCO to oligonucleotide ratio is quantified by comparison of absorption peaks, with 260 nm and 390 nm corresponding to DNA and DBCO respectively. After quality assessment, oligonucleotides can be stored for subsequent conjugation. Immediately before conjugation, antibodies are washed twice in 400 μL of PBS in a spin column that was blocked with 500 μL of PBS-Tween. After washing, antibodies are incubated with NHS ester azide at a 180:1 azide to antibody molar ratio in DMSO and 100 μL of PBS for four hours. After incubation, antibodies are washed two times in 400 μL PBS to remove excess NHS ester azide. Next antibodies are washed in 400 μL of Buffer C. DBCO-modified oligonucleotide is added to the antibody in 100 μL of high salt Buffer C solution at a 25:1 oligonucleotide to antibody molar ratio and left overnight at room temperature. After overnight incubation, the conjugates are washed three times with 400 μL of high salt PBS. Lastly the antibodies are resuspended in 100 μL of CODEX antibody stabilizer solution before inverting the column for collection.

Azide-modified oligonucleotide conjugation:

To minimize nonspecific IgG binding, 400 μL of 0.05% Tween-20 in PBS was added to a 50 kDa MWCO centrifugal filter (Amicon), followed by centrifugation at 12,000 g for 8 min. 50 μg of IgG was added to the column with 400 μL of PBS and washed at 12,000 g for 8 min. Antibodies were resuspended in 100 μL of PBS, and Sulfo-DBCO-PEG4-TFP Ester (50 mM stock in DMSO, Click Chemistry Tools) was added at 150:1 molar ratio of DBCO to antibody and incubated for 4 h at RT for copper-free click conjugation. After incubation, the column was washed twice with 400 μL of PBS at 12,000 g for 8 min.

The average number of DBCO molecules attached per antibody was estimated by calculating the ratio of DBCO to IgG using absorbance values at 309 nm and 280 nm measured by NanoDrop UV-Vis spectrophotometry. Corrections for the extinction co-efficient of DBCO and antibodies and DBCO contribution to A₂₈₀ was used to estimate the ratio of DBCO molecules per IgG^39,40 as follows:

$$\mathit{\text{DBCO}}\mathit{\text{molecule}}/\mathit{\text{antibody}}=\frac{A_{309}}{12000M^{-1}{\mathit{\text{cm}}}^{-1}}/\frac{A_{280}-(1.089*A_{309})}{204000M^{-1}{\mathit{\text{cm}}}^{-1}}$$

An optimized range of ~10 DBCO molecule per antibody showed the best downstream application.

Azide-modified oligonucleotides (500 μM stock in TE, GeneLink) were then added at a 25:1 molar ratio of oligo to antibody in 50 μL TE buffer and incubated overnight at RT. The conjugated antibodies were next washed at 12,000g for 8 min sequentially with 400 μL TE buffer, followed by two washes with 400 μL high-salt PBS (1M NaCl in PBS) and one PBS wash. Antibodies were finally resuspended in 200 μL of stabilization solution (5mM EDTA, 0.5% BSA and 0.02% NaN₃ in PBS), transferred to a clean tube, and stored at 4 °C until use.

Site specific enzyme chemistry conjugation:

Before conjugation, the Amicon 50 kDa centrifuge tube is blocked with 500 μL of PBS-TWEEN. After blocking, 50 μg of antibody is washed three times with PBS to remove any azides from the storage solution. To 100 μL of PBS with antibody, 0.3 U of Peptide-N-Glycosidase F (PNGase), 0.6 U of microbial transglutaminase (mTG), and 80x molar excess of H2N-PEG2-N3 reagents are added. The solution is then left to incubate overnight at 37°C. After overnight incubation, 300 μL of PBS is added to the column to wash excess reagents out of the column. Washing is repeated three more times with 400 μL of PBS. After washing, to 100 μL of PBS, 10x excess DBCO-modified oligonucleotide is added to the antibody and left to incubate overnight at room temperature. After overnight incubation, the antibodies are washed with 300 μL of PBS and washed two more times with 400 μL of PBS. 100 μL of PBS is added to the column, and the column is inverted into a new collection tube. The final modified antibody solution is stored in PBS.

Fluorescent Imaging:

Antibody-oligonucleotide conjugates were tested by staining, rendering, and imaging mouse splenic tissue. Each mouse tissue was stained with one or two antibody conjugates being tested. Staining was done by hand or by robot. To control against batch effects, every round of staining included a previously validated conjugate as a positive control and quality comparison point. If the positive control failed to give signal or gave weak or noisy signal, that batch of staining was discarded. Positive controls were either a-B220 or a-TCR. The detailed staining and rendering protocol can be found in Nature Protocols¹⁸. For evaluation, “dim signal” was defined as adequately visible at a 0.5 millisecond exposure time on high sensitivity settings, but qualitatively less bright than what is usually seen in positive signal used for CODEX multiplexed imaging assays. “No signal” was either not visible under those parameters or had too high of a signal-to-noise ratio to be usable. “Good signal” applied to anything that would be typically used for a CODEX multiplexed imaging assay. Any oligo antibody pairing that produced a good signal was considered validated for purposes of future testing and calculations.

Staining:

The tissue is first sectioned into 7 μm slices onto poly-L-lysine coated coverslips, and stored in a −80 C freezer. For CODEX assay staining, the tissue is sectioned onto a glass poly-L-lysine slide. Immediately before staining, sectioned tissue is removed from the freezer and let sit on a bed of dry-rite for 2 minutes. The tissue is then incubated in room temperature acetone for ten minutes. The tissue is set to dry for 2 minutes. A 6-well plate with 3 wells of S1 buffer is prepared (detailed buffer compositions can be found in CODEX paper). The tissue is hydrated by sequential incubation in each well of the buffer for 2 minutes. The tissue is then fixed via a 10-minute incubation in a 1.6% PFA buffer mix. Afterwards, the tissue is washed by sequential submersion into the 3 wells of S1 buffer. The tissue is then equilibrated in a staining buffer (which includes blocking components such as rat IgG, mouse IgG, and salmon sperm DNA) for up to 30 minutes before staining. Staining solution is prepared by adding the conjugated antibodies to the staining buffer. A starting point for most antibodies is 1 μL of antibody per 100 μL of staining buffer. Each coverslip will require 100–200 μL of antibody staining solution. The staining solution is then pipetted onto the coverslip, and the tissue is left to stain for 1.5–3 hours at room temperature in a humidity chamber. After the stain is complete, the tissue is washed in 3 wells of high salt buffer S2. The tissue is then fixed for 10 minutes in a 1.6% PFA and high salt buffer solution. After fixing, the tissue is washed in 3 wells of PBS. The tissue is then incubated in ice-cold methanol for 5 minutes; the incubation should be done on ice. The tissue is then washed again in 3 wells of PBS. The tissue is then fixed in a BS3 PBS (20 μL BS3 to 1 mL PBS) solution for 20 minutes. The tissue is then washed in 3 wells of PBS. At this point, the tissue can either be rendered or stored in S4 buffer at 4 C for up to 2 weeks.

Rendering:

The tissue is rendered with complementary fluorescent oligonucleotides. Rendering buffer is prepared from 80% CODEX hybridization buffer and 20% dimethyl sulfoxide and poured into 3 wells. The tissue is set to equilibrate in the rendering buffer while the detection solution is prepared. 100 μL of detection solution is required per coverslip. 100 μL of detection solution contains 92 μL of rendering solution, 7 μL of salmon sperm DNA, 0.2 μL of Hoescht, and 1 μL of each corresponding complimentary fluorescent oligonucleotide. The tissue is then placed in a humidity chamber and covered in 100 μL of detection solution for 10 minutes at room temperature. After 10 minutes, the tissue is washed 3 times in the rendering buffer. The tissue is then washed 2 times in CODEX hybridization buffer. The coverslip is then mounted onto a coverslip for visualization.

Imaging:

All images were taken on the Keyence microscope, with a 20x magnification. Images were taken either at high sensitivity or high resolution (relevant settings are indicated in figures).

Assessing variability within the conjugation and staining processes

To test for variability within the performance of the conjugation and staining processes themselves, we conducted experiments using three different sequences (sequence 7, 65, and 77) of maleimide-modified oligos. We used oligo batches that had been previously tested and chose sequences with three different signal strengths (oligo 77 ‘good’, oligo 65 ‘fair’, oligo 7 ‘bad’). We then conjugated each of these three oligos to mouse anti-TCR three separate times, and stained mouse spleens with each of these nine conjugates (one spleen per conjugate). We then compared signal strength quantitatively and qualitatively across the conjugates from different oligo sequences and between stains from the three conjugates of the same oligo sequence. We did not find significant variability in signal strength within the same oligo sequence.

Signal Quantification and Analysis

Qualitative assessments of the conjugate stains were made immediately after rendering. If antibody signal was too weak to visualize by eye at a two second exposure time under high sensitivity, the conjugate was recorded as a failure, with the note “no signal.” If the signal was visible but very dim, the conjugate was recorded as a failure with the note “dim signal.” If the signal to noise ratio was too low, the conjugate was recorded as a failure. For ease of comparison, these failed conjugates are also listed as “dim signal” or “no signal” for the purposes of this paper, although more detailed notes were also recorded. If the conjugate gave sufficiently bright signal and sufficiently high signal-to-noise ratio, images were taken for quantitative analysis.

Image analysis was executed in ImageJ, which is publicly available at https://imagej.nih.gov/ij/. Signal strength was quantified by comparing average pixel intensity across tissues. For each tissue, the three densest areas of cells were chosen to analyze from three different parts of the tissue. A uniformly sized rectangular square of 75 μm × 75 μm was drawn within each of the three different areas. The average pixel intensity was calculated for the cells inside each square. The mean of these three averages was then calculated. This meta-average was used as the measure of pixel strength of conjugate, which was used as a proxy for strength of signal.

The site-specific chemistry was captured at two exposure times to determine at what exposure time it was most comparable to the DBCO-modified oligonucleotide chemistry. The average brightness signal over a 75 × 75 μm area in the a-TCRβ was captured and averaged for this comparison.

CODEX Multiplexed Imaging Assay and Analysis

Two conjugation methods were tested with a 15-antibody panel on mouse spleen and tumor arrays. This was done with co-detection by indexing (CODEX) multiplexed imaging technology. Detailed staining and imaging protocols can be found in (CODEX paper). The staining protocol is the same as the one summarized above. Antibody information is summarized in supplementary table 1. Imaging data was done with the CODEX fusion machine, which performed stitching, deconvolution, drift compensation, and cycle concatenation. The images were evaluated for specific signal; any markers that produced an untenable pattern or low signal-to-noise ratio were excluded from ensuing analysis. Cell segmentation was performed using CellVisionSegmenter, a neural network R-CNN-based single-cell segmentation algorithm available at https://github.com/bmyury/CellVisionSegmenter or https://github.com/michaellee1/CellSeg. Segmented data was put into CellEngine, and hand-gating was performed to hierarchically gate out distinct cell types. Cell type quantification was carried out in CellEngine. R-squared regressions were done on cell type population quantifications between the two conjugation methods. Further analysis was performed in R statistical software (https://www.r-project.org) in R studio (https://posit.co/download/rstudio-desktop/)

Data Availability Statement:

Data be requested from authors.