A consensus platform for antibody characterization

Abstract

Antibody-based research applications are critical for biological discovery. Yet there are no industry standards for comparing the performance of antibodies in various applications. We describe a knockout cell line-based antibody characterization platform, developed and approved jointly by industry and academic researchers, that enables the systematic comparison of antibody performance in western blot, immunoprecipitation and immunofluorescence. The scalable protocols, which require minimal technological resources, consist of (1) the identification of appropriate cell lines for antibody characterization studies, (2) development/contribution of isogenic knockout controls, and (3) a series of antibody characterization procedures focused on the most common applications of antibodies in research. We provide examples of expected outcomes to guide antibody users in evaluating antibody performance. Central to our approach is advocating for transparent and open data sharing, enabling a community effort to identify specific antibodies for all human proteins. Mid-level graduate students with training in biochemistry and prior experience in cell culture and microscopy can complete the protocols for a specific protein within 1 month while working part-time on this effort. Antibody characterization is needed to meet standards for resource validation and data reproducibility, which are increasingly required by journals and funding agencies.

Introduction

Antibodies are fundamental tools in biomedical research, yet the absence of standardized performance evaluation methods poses challenges for researchers selecting appropriate reagents. Currently, there is a reliance on published descriptions and commercial quality control data. However, these lack detailed characterization and data inclusion¹, hindering effective antibody selection. Additionally, the proliferation of commercially available antibodies targeting human proteins further complicates the process, often leading to time-consuming and inefficient searches for optimal reagents². Moreover, instances of invalidated top-cited antibodies have tainted scientific literature^3–9, underscoring the need for standardized comparison methods to improve data quality and reproducibility.

To address these challenges, we established ‘Antibody Characterization through Open Science’ (YCharOS), a collaborative effort among academia, leading antibody manufacturers and knockout (KO) cell line providers. YCharOS has developed an antibody characterization platform to screen antibodies for their performance in three common applications—western blot (WB), immunoprecipitation (IP) and immunofluorescence (IF)—using KO cell lines as isogenic controls. YCharOS has access, through its partners, to ~80% of all renewable (recombinant and monoclonal) antibodies currently available in all commercial catalogs, with the remaining 20% from sources who are not part of our partnerships. YCharOS also has access to hundreds of commercial human KO cell lines. This setup facilitates direct comparisons among antibodies targeting a specific protein. Furthermore, given that ~15% of commercial and academic KO cell lines continue to express the targeted gene or protein, the iterative validation of KO cell lines using specific antibodies is another key component of the platform.

A key strength of the YCharOS platform is the support of industrial partners who endorse these protocols and contribute antibodies and KO cell lines to enable comprehensive antibody characterization¹⁰. While these antibody manufacturers provide their own internal KO and knockdown (KD) characterization data for some of their antibodies, especially for WB, integrating these protocols internally for all cases would impose an added financial burden on these companies, mainly due to the costs associated with obtaining and using KO cell lines. Moreover, these costs would probably be passed on to antibody customers. Consequently, there arises a need for a third-party organization, such as YCharOS, backed by diverse funding sources, to spearhead a large antibody characterization effort that can function with an economy of scale by testing all commercial antibodies against any given protein in single experiments.

The antibody characterization data generated for each target protein are shared openly¹⁰, benefiting the global biomedical community and promoting robust and reproducible research. By identifying specific antibodies for human proteins, YCharOS contributes to improving the reliability of scientific data, including the removal by YCharOS’ partners of nonspecific antibodies from commercial catalogs¹¹. The consensus protocols employed by YCharOS thereby facilitate the creation of a publicly accessible database containing trusted antibody characterization data¹², aiding researchers in antibody selection and reviewers in assessing antibody suitability. Sharing our protocol aims to assist researchers in conducting their own characterizations, given the multitude of antibodies of varying quality available for most protein targets¹³. While there is a financial burden associated with testing multiple antibodies, leading antibody manufacturers offer small pack size of antibodies and refund policies, which facilitate investigators in purchasing multiple antibodies. This has enabled various research groups to characterize multiple antibodies against different target proteins^{3,5,7,14–16}.

In the following sections, we detail the development and present an overview of the YCharOS platform, its comparison with other antibody characterization methods, its applications, expertise and equipment requirements, limitations and experimental procedures. Through these efforts, YCharOS aims to address the critical need for standardized antibody characterization methods, ultimately enhancing the reliability and reproducibility of scientific research.

Development of the platform

Our protocols are built on KD- and KO-based protocols, originally developed by Stadler et al.¹⁷, Davies et al.¹⁴, Laflamme et al.⁵ and Ayoubi et al.¹¹. The establishment of the YCharOS platform stems from a collaborative effort between academic and industry partners¹⁸, aimed at addressing the critical need for standardized antibody characterization methods^1,19–22. The coauthors of this article all actively participate in the YCharOS public–private partnership, which forms the foundation of this endeavor.

To ensure robust and comprehensive protocols, senior scientists from leading antibody manufacturers, including Abcam, ABclonal, Addgene, Aviva Systems Biology, Bio-Techne (comprising Novus Biologicals and R&D Systems), Cell Signaling Technology, Developmental Studies Hybridoma Bank, GeneTex, Proteintech and Thermo Fisher Scientific, have collaborated in refining the methodologies and implemented them internally. Furthermore, antibody manufacturers donate antibodies to the project, alongside contributions of KO cell lines from Abcam and Horizon Discovery (part of Revvity).

The optimized platform has enabled the characterization of antibodies against a broad spectrum of human proteins, encompassing soluble, membrane-bound and secreted proteins. Notably, as of October 2024, the YCharOS team has tested 1,142 antibodies that target 108 distinct human proteins using this platform.

Overview of the procedure

The workflow implemented by YCharOS involves three common antibody-based applications: WB (Procedure 1), IP (Procedure 2) and IF (Procedure 3), as illustrated in Fig. 1. Each procedure consists of multiple stages, including the preparation of wild-type (WT) and KO cell lines, sample preparation, antibody staining and detection. Example outcomes and troubleshooting options are provided to assist researchers in evaluating antibody performance and refining experimental approaches as needed. Additionally, an analysis code for IF has been developed and made publicly available, facilitating the segmentation and direct comparison of fluorescence intensity between WT and KO cell lines. These protocols enable rigorous evaluation of antibody specificity, defined as the primary antibody’s ability to bind its target antigen within a protein mixture.

Fig. 1 |. Experimental design of the antibody characterization workflow.

All antibodies are tested in all three applications. The antibodies are first tested in WB to iteratively validate the KO lines and the antibodies (Procedure 1). The antibodies are next tested in IP followed by WB to evaluate their performance to capture their intended target (Procedure 2). The antibody selected for WB in Procedure 2 was previously validated in Procedure 1. The antibodies against intracellular proteins are next screened in IF (Procedure 3). An earlier version of this experimental workflow was published by Laflamme et al.⁵.

The strength of the YCharOS platform lies in its iterative approach involving KO lines and antibodies. The platform starts with Procedure 1 to validate the cellular system for studying a particular target protein (Fig. 1). WB screening typically identifies antibodies that demonstrate specificity toward their intended target, thereby confirming target protein expression in WT cells and the absence of the protein in KO lines. Truncated proteins or reduced signals in KO lines can be readily detected in WB by assessing protein size. In contrast, an antibody targeting an epitope present on a truncated protein expressed in a KO line may yield a signal in IF similar to that of WT cells, potentially leading to the erroneous conclusion that the antibody is nonspecific, when in fact, the antibody is specific. Following Procedure 1, antibody users can proceed with Procedure 2, Procedure 3 or both simultaneously (Fig. 1).

Advantages of using KO cell lines

The use of KO cell lines is effective for testing antibodies because it directly demonstrates target specificity. An antibody that immunodetects its target protein in WB will produce a distinct band (or potentially multiple bands in presence of isoforms or posttranslational modifications) in the WT lysate that is absent in the KO lysate (Fig. 2a, case 1). However, a specific antibody might also recognize the target protein along with other unwanted proteins (Fig. 2a, case 2). Therefore, using a KO lysate is essential to determine whether these bands represent posttranslational modifications, protein truncation or nonspecific binding. Some antibodies recommended in WB by the manufacturers fail to detect the target protein even in a cell line with confirmed target expression (Fig. 2a, case 3). Among the antibodies recommended in WB by their manufacturers that we tested, 44% detected only their intended targets, 35% detected both their intended targets and unwanted proteins, and 21% failed to detect their intended target¹¹.

Fig. 2 |. Interpretation of antibody performance in WB.

a, Three selected antibodies against the CD44 protein (UniProt ID: P16070) are presented to illustrate various types of target specificity in WB. In case 1, the antibody specifically detected only CD44, the intended target, as determined by the presence of a band in the WT lysate and the complete absence of any band in the KO lysate (asterisk). In case 2, the antibody not only detected CD44 as determined by the absence of the main band in the KO lysate (asterisk) but also detected unwanted proteins (bands present in both WT and KO lysates). In case 3, the antibody failed to recognize CD44 as the band detected in the WT lysate is also detected in the KO lysate. The expected molecular mass for CD44 is 81 kDa. b, A specific antibody against the CNN3 protein (UniProt ID: Q15417) was used to characterize two independent commercial CNN3 KO clones generated in the same cell line background. CNN3 was detected at ~40 kDa in the WT lysate (asterisk). In case 1, a truncated ~35 kDa protein was detected in the lysate derived from the putative CNN3 KO clone, defined here as a failed clone (arrowhead). In case 2, the antibody did not detect any form of residual CNN3 protein in the CNN3 KO. The expected molecular mass of CNN3 is 36.4 kDa. The 4–20% TG gels were used. For each presented WB, the antibody-related chemiluminescent signal is shown at the top of its corresponding ponceau S-stained membrane.

Despite the validation of edited gene modifications in KO lines through genomic PCR and DNA sequencing, WB analysis revealed that ~14% of the KO lines tested within the YCharOS platform do not completely lack the target protein. For instance, some KO lines resulted in truncated proteins rather than complete loss (Fig. 2b), emphasizing the necessity of WB antibody screening using both WT and KO lysates to validate target protein expression and antibody specificity, as well as to validate the KO clone.

Selection of antibodies

Each antibody is tested for all three applications regardless of manufacturers’ recommendations, as the use of antibodies can be extended to applications not considered by the manufacturers. YCharOS typically tests 10–15 antibodies per protein provided by its partners¹¹. If fewer than three renewable antibodies are shared by YCharOS partners, renewable antibodies from other manufacturers may be purchased, when available, and tested side-by-side with the donated antibodies. Throughout this document, our protocols describe steps for testing 12 antibodies, a number chosen to simplify the number of gels used in WB and IP experiments. These protocols can be adjusted to accommodate different numbers of antibodies to be tested. When selecting from multiple antibody options for a certain protein, the investigators can follow these guidelines: (1) prioritize recombinant or monoclonal antibodies with specified clone numbers to prevent duplicate purchases from different suppliers and ensure reagent renewability; (2) favor primary manufacturers with stringent internal validation standards, offering antibodies in small pack sizes and refund policies if users demonstrate antibody specificity issues; and (3) opt for antibodies characterized using KO or KD cells, with data provided by manufacturers or referenced in published articles. Typically, commercial antibody vials contain 50–100 μg of purified antibody at concentrations ranging from 0.1 to 1.0 mg/ml, with 10 μg usually sufficient for conducting the described protocols.

Conducting experiments

After careful consideration, it was determined that conducting experiments in replicates does not offer notable benefits for the following reasons. First, the validation of the KO cell lines involves the use of multiple antibodies targeting various epitopes. Once a specific antibody is identified, it validates the protein expression of the intended target in the cell line and supports conclusions regarding the specificity of the other antibodies. Second, we typically have access to the same antibody from two to three different manufacturers (cross-licensed antibodies), which effectively serve as replicates, enabling the validation of test reproducibility. Third, all experiments are performed using master mixes, and meticulous attention is paid to sample preparation and experimental execution. In IF, the use of two different concentrations serves to evaluate antibody specificity and can aid in assessing assay reliability.

Expertise and specialized equipment needed to implement the protocol

The protocols described here can be adapted to most standard molecular and/or cell biology laboratories. However, two pieces of equipment mentioned may not be commonly found in all laboratories: a benchtop ultracentrifuge capable of ultraspeed centrifugation of low volumes (ranging from 0.5 ml to 3 ml) and a high-content microscope. Alternatives for these instruments are provided within the corresponding protocol sections and in the troubleshooting section. For cell culture, WB and IP, most trainees with prior biochemistry knowledge should be able to perform these protocols or could learn them with appropriate training, which would be valuable for exploring or initiating a career in related research areas. IF steps require training in microscopy and fluorescence image analysis. While all analyses can be performed on standard desktop computers with minimal software requirements (including the open access Fiji software) using the provided analysis code, the cellpose (RRID: SCR_021716) segmentation code works best on a system with a Compute Unified Device Architecture-capable graphics card.

Comparison with other methods

The International Working Group for Antibody Validation has recommended five antibody characterization methodologies: (1) genetic strategies utilizing KO or KD cell lines as controls, (2) orthogonal strategies correlating antibody signals to known information about the target protein, in particular RNA sequencing data, (3) overlap of signals of two independent antibodies recognizing different epitopes in the same target, (4) antibody recognition of an overexpressed tagged version of the protein target and (5) employment of mass spectrometry to determine whether the protein target captured by an antibody corresponds to the major signal in the immunoprecipitate²³. Among these strategies, the most frequently used by antibody manufacturers is the orthogonal approach, which accounts for 61% and 83% of antibodies recommended in WB and IF, respectively¹¹. In contrast, 30% and 7% of antibodies recommended in WB and IF, respectively, were characterized by manufacturers using genetic strategies¹¹.

However, our analyses have demonstrated a drawback with a reliance on orthogonal approaches. Specifically, our findings have revealed discrepancies between our evaluation of commercial antibody performance and the recommendations provided by manufacturers. This is particularly true in IF, in which only 38% of antibodies characterized by manufacturers using orthogonal strategies showed the expected specificity in our tests¹¹. This has led to the inclusion of underperforming antibodies used in IF in hundreds of published articles¹¹. This presents a compelling argument to prioritize genetic strategies for antibody characterization.

Applications

Data dissemination and uptake by the research community

YCharOS’s data generation and dissemination are intended to benefit the global life sciences community, but its impact depends on real-world uptake of the data. So far, 1,142 different antibodies targeting 108 human proteins have been tested and characterization data are consolidated in the form of reports, with one report per protein. The reports are uploaded on Zenodo (https://zenodo.org; RRID: SCR_004129), an open repository operated by the European Organization for Nuclear Research, and assigned a digital object identifier. As of October 2024, the YCharOS working group generates and uploads approximately one new report per week on Zenodo. YCharOS reports can be accessed by searching the protein name or its corresponding UniProt ID on the community’s page: zenodo.org/communities/ycharos. To assess the possibility of better outreach by indexation on PubMed, some Zenodo reports are converted into peer-review articles published in F1000 (f1000research/ycharos), accompanied by a guide to help interpret the antibody characterization data¹⁰. The dataset related to each study, which includes raw data from antibody testing for all applications, can also be viewed and downloaded from Zenodo.

To ensure the proper identification of each antibody tested, each YCharOS report presents detailed antibody information, including antibody concentration, lot number and research resource identifier (RRID). The lot number information is particularly important for polyclonal antibodies, which represent a finite resource, and for hybridoma-derived monoclonal antibodies, which can suffer from genetic drift²⁴. An RRID is a unique and persistent tag assigned to an antibody (and other research resources) that integrates the following detailed information in the case of antibodies: the target antigen, antibody clonality, catalogue number and supplier, clone ID, application(s) recommended by the manufacturers, host organism and availability of third-party validation data. Over 2.5 million antibodies are registered with an RRID and listed in the Antibody Registry (https://www.antibodyregistry.org; RRID: SCR_006397) and in the RRID portal (https://scicrunch.org; RRID: SCR_003115). RRIDs represent the gold standard for research reagent identification and are requested by over 1,000 journals^25–27. They facilitate access to third-party characterization data through the RRID portal. By integrating characterization data with RRIDs via Biomed Resource Watch (https://scicrunch.org/ResourceWatch), the RRID portal has the potential to become the primary centralized database for genetically validated antibodies¹².

YCharOS antibody characterization reports are also linked to the CiteAb online reagents database (https://www.citeab.com; RRID: SCR_009653)²⁸. Moreover, YCharOS data will be cross-referenced with the UniProt database in the second quarter of 2025.

Implication of antibody manufacturers

We describe standardized, industry-approved protocols for comparing and evaluating the performance of antibodies in WB, IP and IF. Initial characterization data assessment allows the identification of nonspecific or poorly performing antibodies, facilitating their exclusion from future antibody selection by users. We strongly advocate for users to communicate feedback to antibody suppliers regarding underperforming antibodies, as most suppliers will evaluate user data and take proactive measures to withdraw or amend antibody descriptions accordingly.

The participating antibody manufacturers, who have endorsed these protocols through extensive dialogue and who are represented as coauthors of this article, are also actively using the antibody characterization data in their marketing materials to help scientists select the most appropriate products for their research needs. In addition, these same companies are withdrawing or reevaluating antibodies whose performance in these assays appears substandard¹¹, underlining the importance of informing antibody manufacturers in the latter case. Finally, targets for which better antibodies are needed are identified and designated for the development of new antibodies.

Increasing the adoption of recombinant antibodies

The research market is currently dominated by polyclonal antibodies, and their use contributes to the reproducibility crisis observed in biomedical research. Despite their advantages as renewable products that reduce reliance on animal-based antibody production²⁹, the scientific community has been slow to adopt recombinant antibodies. Our analysis of a subset of antibodies indicates that recombinant antibodies perform comparably to or exceed the performance of polyclonal antibodies or antibodies derived from hybridomas¹¹. These findings suggest a strong rationale for the adoption of recombinant technology.

When testing multiple antibodies, it is inevitable that different antibodies will exhibit varying degrees of specificity toward their intended targets. Whenever possible, we, therefore, advise that researchers should prioritize the use of recombinant antibodies. Drawing from our prior research, we anticipate that widespread adoption of these protocols can facilitate the identification of specific, renewable antibodies for ~50–75% of human proteins, depending on the application¹¹.

Limitations

Beyond the described protocols, antibodies have extensive applications in techniques, such as flow cytometry, ELISA and immunohistochemistry. However, given that antibodies are fit-for-purpose reagents, their performance as demonstrated cannot reliably predict their specificity in applications outside the scope of this platform.

One limitation of this platform is the reliance on a single cell line for evaluating antibody performance, as factors such as target protein abundance impact results. For example, WB data indicate that MDA-MB231 cells display about a tenfold increase in CD44 protein expression compared with HAP1 (Fig. 3a, case 1), leading to variations in antibody performance that are dependent on cell type. For instance, one antibody can detect CD44 in both cell types (Fig. 3a, case 1), while another can only in the higher-expressing MDA-MB231 line (Fig. 3a, case 2). The THP-1 cell line, which has a high PLCG2 expression, allows testing of PLCG2 antibodies after phorbol 12-myristate 13-acetate (PMA) treatment, which induces differentiation toward adherent, macrophage-like cells³⁰, enabling IF experiments. Although PMA treatment slightly reduces PLCG2 protein levels in THP-1 cells, these cells still exhibit approximately a threefold higher PLCG2 protein level compared with HAP1 (Fig. 3b, WB). The same PLCG2 antibody shows specific IF signals in THP-1 cells treated with PMA but not in HAP1 cells (Fig. 3b, IF). This result suggests that the observed specificity is probably due to higher PLCG2 expression in THP1 cells. However, we cannot rule out other possibilities, such as the antibody reacting with unwanted proteins expressed in HAP1 that are not present in THP1.

Fig. 3 |. Protein abundance influences antibody performance.

a, Two antibodies against the CD44 protein (UniProt ID: P16070) were selected to illustrate the effect of protein abundance on antibody performance. Both selected CD44 antibodies are different from those shown in Fig. 2. The RNA levels corresponding to both cell lines were extracted from DepMap. HAP1 and MB231 express CD44 at 2.6 and 9.6 log₂(TPM + 1), respectively. In case 1, the antibody was able to specifically detect CD44 in both cell lines (asterisk). In case 2, the antibody detected CD44 in MB231 but not in HAP1 (asterisk). The expected molecular mass for CD44 is 81 kDa. The 4–20% TG gels were used. MB231, MDA-MB-231. b, The intracellular protein PLCG2 (UniProt ID: P16885) was selected to illustrate the effect of protein abundance on antibody performance in IF. The same PLCG2 antibody was used in WB and in IF. PLCG2 was detected by WB (asterisks) in WT cells after comparing lysates from HAP1 WT and PLCG2 KO, as well as THP-1 WT and PLCG2 KO treated or not with PMA. The RNA levels are 2.4 log₂(TPM + 1) in HAP1 and 5.9 in THP-1. The expected molecular mass for PLCG2 is 148 kDa. The 4–20% TG gels were used. The PLCG2 antibody was tested on HAP1 (left IF) and PMA-treated THP1 (right IF). The WT (green outline) and KO (purple outline) cell lines were plated as a mosaic and were segmented postimage acquisition. The gray-scale antibody channel is shown (top), together with the corresponding DAPI stain (nucleus, bottom). THP-1 are small cells that adopt a round shape. Scale bar, 10 μm. The WBs are presented as in Fig. 2.

The absence of a universal, public collection of KO human cell lines hinders the ability of scientists to immediately use the described protocols. However, a considerable portion of human genes have already been targeted and knocked out in cell lines generated by academic researchers. Cellosaurus (https://www.cellosaurus.org; RRID: SCR_013869) is a knowledge resource that assigns an RRID identifier to cell lines used in biomedical research, including KO cell lines, whether generated by academic laboratories or industry³¹. A search on Cellosaurus (release 48, 1 February 2024) indicates that 13,644 KO cell lines covering 4,873 human genes have been generated with the majority being commercially available. For example, Horizon Discovery has a collection of over 6,000 KO cell lines targeting more than 3,000 genes in the human HAP1 cell line, now part of Revvity’s portfolio. Similarly, Abcam has a catalog of ~5,300 KO cell lines covering 2,915 human genes in various cell line backgrounds. Horizon Discovery and Abcam have the largest off-the-shelf KO cell line selections. Together, these companies cover a total of 2,731 human genes with their commercial KO lines, annotated with an RRID, with the corresponding WT cell line expressing the target RNA level above 2.5 log₂ (transcripts per million (TPM) + 1) (Supplementary Table 1). The target RNA expression value of 2.5 log₂(TPM + 1) has emerged as a minimal threshold that yields detectable protein levels suitable for antibody screening¹¹. While commercial KO cell lines are well documented, those generated by individual researchers are often not registered, highlighting a gap in data accessibility. Adding a cell line to Cellosaurus can be done by writing to cellosaurus.org/contact.

In the absence of readily available KO cell lines, using small interfering RNA (siRNA) can be a suitable alternative^32,33. One of the main advantages of siRNA is its flexibility in choosing the cell line background, in contrast to commercial KO cell lines, in which a specific cell line background has already been selected that might not be optimal for the target. This approach is particularly beneficial for targeting essential^33,34. However, a notable drawback is the unpredictable efficiency of siRNA, necessitating validation of the KD effectiveness, which can be aided by using multiple antibodies. Nevertheless, when KD is inefficient, it becomes challenging to distinguish between an ineffective KD and nonspecific antibody binding.

The use of cancer cell lines containing gene mutations poses a potential challenge, as these mutations may be within the epitope coding sequence or other regions of the gene responsible for the intended target. Such alterations should impact the binding affinity of antibodies. This represents an inherent limitation of any approach that employs cancer cell lines.

These protocols were developed to identify specific antibodies for core, unmodified proteins. Characterizing antibodies for posttranslational modifications requires distinct negative controls, often involving cell treatments to induce the modification of interest and the use of specific inhibitors in lysis buffers. However, the characterization of antibodies targeting these modifications is not included in the platform described.

These protocols focus on antibodies being tested against human targets. Therefore, additional characterization using KO lines or tissues from the relevant species is essential before applying these antibodies to proteins from other species.

Experimental design

Cell line selection

The selection of a suitable cell line for studying a specific protein target involves identifying a cell line that readily accommodates clustered regularly interspaced short palindromic repeats and CRISPR-associated protein 9 technology and can be grown as a single-cell clone, while maintaining appropriate endogenous RNA/protein expression levels. Supplementary Table 2 lists 14,550 human genes with RNA levels exceeding 2.5 log₂(TPM + 1) in one of the 20 preferred CRISPR-Cas9-compatible cancer cell lines. In contrast, Supplementary Table 3 provides RNA levels for all human genes in one of these cell lines. Assessment of target expression in each cell line is conducted using RNA sequencing data accessible via the Cancer Dependency Map Portal (DepMap (https://depmap.org); RRID: SCR_017655), which offers transcriptomic profiles for over a thousand cancer cell lines³⁵. Transcriptomics values can be obtained by searching the gene name and downloading the ‘Expression Public 23Q4’ file from the ‘characterization’ tab.

While cell line selection can be straightforward for proteins with high, ubiquitous expression, this step can be difficult for proteins expressed at lower levels or only in specific and specialized cell types. For this latter situation, we combine RNA sequencing data (orthogonal approach) and the use of a few unique antibodies (independent antibodies approach)²³ to help in selecting an optimal cell line background for generating a KO cell line. In brief, we select four to eight available cancer cell lines with the highest RNA score, as well as one or two lines with an RNA score close to or equal to zero and use at least two unique antibodies (that is, different clone numbers if monoclonal antibodies) to assess protein expression in WB. This has proven useful when at least two antibodies show similar protein expression patterns in the cell lines with high RNA values and no signal in the lines with RNA close to zero, suggesting specificity to the target (Fig. 4a, cases 1 and 2). A nonspecific antibody would produce a different protein expression pattern than putative specific antibodies and would not correlate with RNA levels (Fig. 4a, case 3). The generation of a KO line remains essential to confirm the specificity of the signal and ascertain whether the selected WT cell line exhibits appropriate expression of the target protein (Fig. 4b). For instance, the SYT1 antibody in case 1 (Fig. 4a) specifically detects several SYT1 protein species, while the SYT1 antibody in case 2 (Fig. 4a) detects SYT1 along with undesired proteins, a scenario verifiable only using KO cells (Fig. 4a). KO cell lines are essential to accurately distinguish between undesired nonspecific binding and genuine signals derived from the protein target (isoforms, post-translational modifications or degradation).

Fig. 4 |. Identification of an adequate cell line background for KO generation.

a, The identification of an adequate cell line for the SYT1 protein (UniProt ID: P21579). Seven cancer cell lines were selected with RNA expression spanning from 0.3 to 4.6 log₂(TPM + 1). The RNA levels were extracted from DepMap and presented in blue below the corresponding cell line. The lysates were prepared, processed by SDS–PAGE and probed with three unique primary SYT1 antibodies (cases 1, 2 and 3). In case 1 and case 2, both antibodies putatively identified a main SYT1 band at ~66 kDa in HCT116, as they provide a similar banding pattern with absence of signal in cells with low RNA value. In case 3, the antibody provided a signal that does not correlate with the signal of the other two antibodies. b, An HCT116 SYT1 KO line was generated and used to validate that HCT116 expresses the endogenous SYT1 protein and the specificity of the same SYT1 antibodies used in a. The asterisks indicate the specific SYT1 bands detected with antibody case 1 at ~40, 48, 60 and 66 kDa. The expected molecular mass for SYT1 is 47.5 kDa. The 4–20% TG gels were used. The WBs are presented as in Fig. 2.

Sample preparation

The sample preparation procedure followed for WB and IP varies depending on whether the target of interest is an intracellular (Fig. 5, case 1) or secreted protein (Fig. 5, case 2). A secreted protein is defined as having a signal peptide and no transmembrane domains. It has been predicted that ~3,000 human proteins are secreted (referred to as the secretome), representing ~15% of the human proteome³⁶. These proteins are expected to be primarily identified in the conditioned medium of cell lines (Fig. 5, case 2). However, ~35% of the secretome may remain intracellular awaiting an appropriate stimulus to trigger secretion or are not released to the medium due to retention on the plasma membrane following secretion³⁶. In the latter cases, the target protein can be detected both in the conditioned medium and in the cell lysate (Fig. 5, case 3). Subcellular annotation from UniProt (https://www.uniprot.org; RRID: SCR_002380) can be used to predict whether a protein is secreted, yet the identification of the target protein in the medium using specific antibodies or by mass spectrometry provides the definitive evidence that a protein can be secreted.

Fig. 5 |. Antibody performance correlates with sample preparation.

The proteins were prepared from both cell lysates and conditioned media from a cell line endogenously expressing the corresponding intended target. Protein targets were searched through UniProt to determine whether they are predicted to be secreted or not. In case 1, the antibody targets ECE1 (UniProt ID: P42892), a predicted intracellular protein. ECE1 was detected exclusively in the cell lysate sample (asterisk). The predicted molecular mass for ECE1 is 87 kDa. The 4–20% TG gels were used. In case 2, the antibody targets ANG (Uniprot ID: P03950), a predicted canonical secreted protein. ANG was only detected in the medium (asterisk). The predicted molecular mass for ANG is 16.5 kDa; 10% BT gels with MES running buffer were used. In case 3, the antibody targets the protein QPRT (Uniprot ID: Q15274), predicted to be secreted and to retain an intracellular distribution. QPRT was detected both in cell lysate (asterisk) and medium (asterisk). The predicted molecular mass for QPRT is 31 kDa. The 4–20% TG gels were used. The WBs are presented as in Fig. 2.

For secreted proteins, the protocols involve the use of cell culture medium. Although secretory pathways inhibitors such as brefeldin A have been used to inhibit secretion of various classes of proteins, we found that brefeldin A was not universally effective in inhibiting secretion. Thus, screening antibodies on culture medium ensures a more reliable screening process applicable to secreted proteins.

Cell preparation

The procedures detailed below involve the use of adherent cell lines. The cell lysates from both WT and KO cell lines are initially prepared for antibody screening in WB. One confluent 150 mm Petri dish of the most common cancer cell lines corresponds to ~2 × 10⁷ cells and 2 mg of protein lysate. To test 12 antibodies by WB (between 20 and 50 μg per lane), one confluent 150 mm dish is sufficient, but preparing two confluent dishes is recommended to allow freezing of extra protein lysate for future use. To collect secreted proteins, two confluent 150 mm dishes are sufficient for WB, but the preparation of protein from three dishes is recommended to allow freezing of some extra medium for future use. To test 12 antibodies by IP on lysate (1.0 mg protein per IP), seven confluent 150 mm dishes are sufficient. The IP of secreted proteins (0.5 mg protein per IP) requires 13 dishes for 12 antibodies (an additional dish to secure starting material samples). Finally, 12 antibodies can be tested by IF (8,000 cells per well) using one 150 mm dish.

The IF protocol is not applicable to cells in suspension. For example, in instances in which proteins are expressed exclusively in myeloid-like cells, necessitating the use of nonadherent cells such as THP-1 or U937, we employ 200 ng/ml of PMA for 2 d to induce differentiation of these monocytic cells into macrophage-like cells, which then become adherent^30,37. PMA serves as an analog to the second messenger diacylglycerol and primarily activates protein kinase C, thereby initiating a plethora of cell signaling pathways that may impact the protein of interest. The effect of the PMA treatment on the protein target level is assessed by WB before the antibody screening in IF.

Procedure 1: antibody screening in WB

The use of commercial lysis buffers offers several advantages, including improved reproducibility, throughput and standardization. The radio-IP assay (RIPA) denaturing buffer extracts most intracellular proteins from culture cell lines, including cytoplasmic, nuclear and membrane-bound proteins³⁸. The commercial RIPA buffer used is composed of 25 mM Tris–HCl pH 7.8 (pH measured at 4 °C), 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate and 0.1% sodium dodecyl sulfate (SDS).

The secreted proteins are collected directly from the cell culture medium. To this end, the cells are grown in a serum-free medium for 18 h, the medium is collected and debris in suspension are removed by centrifugation. The cleared medium is then concentrated by filtration. The desired amount of WT or KO lysates or WT or KO cell media is run on SDS–polyacrylamide gel electrophoresis (PAGE). All antibodies are tested in parallel in WB.

Various gel chemistries offer distinct advantages and disadvantages for protein separation. Although larger proteins are generally harder to detect, Tris–glycine (TG) gels are effective for proteins up to 500 kDa. As an example, plectin (PLEC), a large protein with a molecular mass of 532 kDa, was successfully detected using a KO-validated antibody on a TG gel (Fig. 6a, left WB), a Bis–Tris (BT) gel combined with the 3-(N-morpholino)propanesulfonic acid (MOPS) buffer (Fig. 6a, middle WB) and on a Tris–acetate (TA) gel (Fig. 6a, right WB). The BT and TA gels demonstrated better separation and resolution of PLEC than that of TG gels. However, TG gels were chosen for standardization due to their broad molecular weight range. Conversely, TG gels are less effective for transferring smaller proteins, as illustrated by the detection of FCERG1 (Fig. 6b, left WB). To address this, BT gels combined with 2-(N-morpholino)ethanesulfonic acid (MES) buffer are employed, as this gel chemistry offers the best separation and resolution of proteins under 25 kDa (Fig. 6b, right WB).

Fig. 6 |. Choice of SDS–PAGE chemistry in WB.

a, A KO-validated antibody against the large PLEC protein (Uniprot ID: Q15149) was used in WB of WT and PLEC KO lysates ran on three gels with distinct chemistries: 4–20% TG, 8% BT and 3–8% TA using TG–SDS, MOPS–SDS and TA–SDS buffers, respectively. The chemistry of the SDS–PAGE modifies the reading of the antibody signal. TG and BT gels were transferred in 1× Tris/glycine buffer as detailed in Table 3. TA gel was also transferred in 1× Tris/glycine buffer and following the same settings as the BT gel. All transfer conditions were inspired from manufacturers’ recommendations and then further optimized. PLEC has nine putative isoforms produced by alternative splicing, with the canonical PLEC isoform expected at 532 kDa (UniProt). The vertical line followed by a asterisk indicates the region of the gels where the isoforms are identified. Note that the PLEC KO cell line used here expresses residual PLEC protein isoforms. b, A KO-validated antibody against the small FCER1G protein (UniProt ID: P30273) was used in WB of WT and FCER1G KO THP-1 lysates, each PMA treated or not. The samples ran on 4–20% TG with TG–SDS buffer (left) and 10% BT with MES–SDS buffer (right) were compared. PMA treatment was used to differentiate THP-1 into adherent macrophage-like cells. FCER1G is expected at ~10 kDa. The BT gels improved both the antibody-based signal and the resolution. The WBs are presented as in Fig. 2.

Proteins do not necessarily migrate to their expected molecular weight in WB. SDS–PAGE separates proteins mainly on the basis of their mass by using the strong anionic detergent SDS, which denatures proteins and confers a uniform charge-to-mass ratio. Consequently, after the application of an electric current, SDS-bound proteins migrate through the gel toward the positively charged electrode. However, several factors influence SDS binding to proteins, thereby affecting their migration in an SDS–PAGE gel. For instance, glycosylation increases the molecular weight of the target protein and consequently slow down its migration in SDS–PAGE. Phosphorylation adds negative charges to proteins, reducing their affinity for SDS and further slowing down migration. Residual positive charges, particularly in arginine- and/or lysine-rich proteins, increase binding to SDS and enhance their mobility in gels. The proteins with acidic isoelectric points bind to SDS less evenly and migrate slower on gels. These diverse parameters collectively contribute to the observation that ~40% of the proteins in Schizosaccharomyces pombe did not migrate to positions predicted solely from their theoretical mass³⁹.

Procedure 2: antibody screening in IP

Antibodies can be used to immunocapture target proteins from cell extracts or media. To assess if an antibody can IP the target, IPs are performed using cell lysates generated in a nondenaturing buffer (intracellular proteins) or cell media (secreted proteins), followed by WB (IP–WB) with KO-validated antibodies, ideally renewable, identified in the previous WB screening. A successful antibody should enrich its intended target in the IP, as compared with the starting material, and deplete it from the unbound fraction (Fig. 7). The unbound fraction is collected once incubation of the protein sample with the bead/antibody conjugate is complete. The described IP–WB method allows assessment of the antibody’s ability to capture the intended target. However, it does not provide information on binding partners of the intended target captured in the IP nor does it identify proteins that nonspecifically bind to the antibody-beads conjugate used for purification. Antibody performance in IP can be assessed using mass spectrometry approaches⁴⁰, which for most laboratories is costly and time-consuming. Methods using mass spectrometry for characterizing antibody specificity in IP are detailed elsewhere^40,41.

Fig. 7 |. Interpretation of antibody performance by IP.

Three selected antibodies directed against the LRP1 protein (UniProt ID: Q07954) illustrates different degrees of capture efficiency in IP. A specific LRP1 antibody in WB was used to detect the LRP1 protein level between three distinct fractions: the SM, the UB and the IP. The SM sample represents 4% of total lysate, and the UB sample represents 4% of the lysate that failed to bind to the antibody–bead conjugate after incubation. In case 1, the antibody did not capture the target protein as determined by the absence of signal in the IP fraction and unchanged level of the LRP1 protein in the UB. In case 2, the antibody captured the target protein to slightly below the level of the SM and failed to deplete LRP1 from the UB. In case 3, the antibody enriched its intended target in the IP several folds over the SM and mostly depleted LRP1 from the UB. This antibody successfully immunocaptured its intended target in the conditions used. LRP1 is a 600 kDa precursor protein cleaved into a 515 kDa α-chain and an 85 kDa β-chain. For simplicity, only the 85 kDa form of LRP1 is shown here. The 4–20% TG gels were used. The WBs are presented as in Fig. 2. SM, 4% starting material; UB, 4% unbound fraction.

The cell lysates (starting material) are prepared using a nondenaturing commercial lysis buffer (IP buffer) composed of 25 mM Tris–HCl pH 7.8 (pH measured at 4 °C), 150 mM NaCl, 1 mM EDTA, 1% NP-40 and 5% glycerol. This buffer allows the efficient extraction of all targets tested thus far, including cytosolic, nuclear and membrane-bound proteins. The cell lysates are incubated with antibody–bead conjugates. After incubation with the lysate, an aliquot of the unbound fraction is collected. Antibody–bead conjugates are then washed with lysis buffer to remove or minimize unbound and nonspecifically bound proteins. Following the final wash, bound protein(s) is (are) eluted from the beads. Similar volumes from the starting material and unbound fractions are separated by SDS–PAGE side by side with the eluted fraction, followed by WB to detect the target protein. We were able to identify at least one antibody that can capture its intended protein for 73 out of 95 proteins (77% success rate).

When the proteins of interest have molecular masses close to those of immunoglobulin chains (~50 kDa for heavy chains and 25 kDa for light chains), the antibodies used in WB following IP may cross-react with these chains in the IP lane, masking the protein signal (Fig. 8a, case 1 and Fig. 8b, case 4). Two options are proposed to avoid these cross-reactivities: (1) use a secondary detection system that should not react with unstructured immunoglobulins resulting from the elution under reducing conditions used here (Fig. 8a, case 2 and Fig. 8b, case 5), or (2) where possible, choose a primary KO-validated antibody raised in a different host than the antibody used in IP, ideally allowing minimal detection of immunoglobulins from different species (Fig. 8a, case 3 and Fig. 8b, case 6).

Fig. 8 |. Selection of secondary detection systems for IP-WB experiments.

A rabbit (a) or mouse (b) antibody targeting human UBQLN2 (UniProt ID: Q9UHD9) was used in IP in combination with different secondary WB detection systems. In case 1 and 2, a rabbit primary antibody was used in WB and detected using either a secondary anti-rabbit:HRP or prot A:HRP, respectively. In case 3, a mouse primary antibody was used in WB coupled with a secondary anti-mouse:HRP. In case 4 and 5, a mouse primary antibody was used in WB and detected using either a secondary anti-mouse:HRP or anti-mouse IgG for IP:HRP, respectively. In case 6, a rabbit primary antibody was used in WB coupled with a secondary anti-rabbit:HRP. The bracket indicates different UBQLN2 protein species detected by WB. HAP1 cell lysates were used for all the IPs. The expected molecular mass of UBQLN2 is 65.7 kDa. The WBs are presented as in Fig. 2. SM, 4% starting material; UB, 4% unbound fraction; HC, heavy chain; LC, light chain.

Procedure 3: antibody screening in IF

In IF studies, fixation and cell permeabilization steps enable antibodies to reach their intracellular targets. Standardization of IF protocols is challenged by the diversity of fixation and permeabilization reagents and concentrations. A study comparing the suitability of six IF protocols with known specific antibodies targeting 18 proteins with distinct subcellular distributions revealed that fixation with paraformaldehyde (PFA), followed by permeabilization with Triton X-100, was adequate for detecting all proteins analyzed in their study, suggesting that a PFA/Triton X-100-based protocol is adequate for most human proteins⁴². The processes described here use 4% PFA with 0.1% Triton X-100 for permeabilization and 0.01% for the later steps. While we recognize that this protocol will not be suitable for all human proteins, we were able to identify at least one specific antibody suitable in IF for 49 out of 82 intracellular proteins (60% success rate). The failure to identify specific antibodies may not be due to the IF protocol itself but rather the quality of the antibodies—particularly for less-studied proteins—or the abundance of the protein target in the selected cell lines.

To avoid imaging or user bias when assessing antibody performance, the IF protocol takes advantage of WT and KO cells plated together as a mosaic, enabling their imaging within a single field of view (Fig. 9a). To distinguish between the two cell populations, WT and KO cells are labeled with fluorescent dyes of different wavelength (Fig. 9b). Staining is performed with primary antibodies and a secondary antibody coupled to a fluorophore that emits at a different wavelength from those of the cell dyes (Fig. 9c).

Fig. 9 |. Semiautomated analysis of antibody performance in IF.

a, Images from all four channels corresponding to DAPI (nucleus), CellTracker Green CMFDA (WT cell mask), the anti-TGM2 staining case 1 (coupled to an Alexa 555 conjugated secondary antibody) and CellTracker Deep Red (KO cell mask) were acquired with an ImageXpress high-content microscope and prepared for analysis. A semiautomated image analysis of a mosaic culture of WT and KO cells was conducted using in-house developed codes that take advantages of the publicly available cellpose algorithm and FIJI (ImageJ) software. This platform images and analyzes at least 500 WT and KO cells. Two antibodies against the TGM2 protein (UniProt ID: P21980) are presented. A TGM2 KO line, validated by WB, was used (data not shown). b, A Python script that executes cell segmentation using Cellpose1.0 was ran on both cell mask channels. c, An ImageJ macro was used to generate cell mask outlines and perform background signal subtraction on the antibody channel using minimum intensity projection. The processed images were overlayed with cell masks outlines in the antibody channel and intensity was quantified in the segmented cells. All scripts are openly available on the YCharOS GitHub page. d, The same processes as in a–c were applied to the anti-TGM2 antibody case 2. Scale bars, 20 μm. e, Plot showing the antibody intensity ratio of WT to KO cells.

The use of a high-content imaging system, designed to image numerous fields of view per well, enables rapid imaging of thousands of cells for all antibodies tested. The goal of the IF approach is not to determine the cellular location of the target protein but to determine whether there is a notable difference in overall signal coming from WT and KO cells (Fig. 9c–e). A larger collection of IF images allows a more robust analysis. We have developed a collection of scripts in Python and in ImageJ (RRID: SCR_003070) or FIJI (RRID: SCR_002285) made openly available on GitHub (https://github.com/ABIF-McGill/YCharOS_IF_characterization) to quantify and compare fluorescence from WT cells and KO cells (Fig. 9b,c,e). Generally, this quantitative analysis pipeline consists of object detection to generate masks of WT and KO cells, followed by background estimation and subtraction and, finally, intensity measurement of antibody labeling intensity in each detected cell in each image. Antibody intensity in WT versus KO cells can be expressed as a ratio for each cell and plotted to compare antibody labeling intensity of several different antibodies for a given target (Fig. 9e). This more detailed analysis of numerous cells improves the comparison of performance between antibodies. Moreover, antibody titration is performed routinely in IF experiments, in which two concentrations (Fig. 9e) are tested, including the concentration recommended by the manufacturers, when available.

A total of 30 wells are required for testing 12 antibodies and for the remaining controls (Table 1). Wells numbers 1–24 are designated for testing primary antibodies with two distinct concentrations. On the basis of our experience from hundreds of IF experiments, we found that 1.0 μg/ml is an appropriate initial concentration, as it generally provides an adequate signal within the detection range of a microscope. When an antibody is recommended in IF by the manufacturer, the recommended concentrated is tested alongside 1.0 μg/ml. If the recommended concentration is 1.0 μg/ml or not specified for IF, the antibody is tested at both 1.0 μg/ml and 2.0 μg/ml. Antibody titration should be performed if the signal obtained falls outside the linear range of detection. Wells numbers 25–26 are dedicated for secondary antibody controls depending on the species of the primary antibody tested (for example, rabbit and mouse). Wells numbers 27–29 address bleed through from the channel 1, 2 and 4 into channel 3. Well number 30 is used to estimate the base image background and contains media but no cells.

Table 1 |.

Suggested conditions for antibody testing in IF

Well	Channel 1	Channel 2	Channel 3	Channel 4
No.	DAPI	CellTracker Green CMFDA	Antibody (Alexa fluor 555)	CellTracker Deep Red
1	+	+	1° Ab no. 1, concentration 1	+
2	+	+	1° Ab no. 1, concentration 2	+
3	+	+	1° Ab no. 2, concentration 1	+
4	+	+	1° Ab no. 2, concentration 2	+
5	+	+	1° Ab no. 3, concentration 1	+
6	+	+	1° Ab no. 3, concentration 2	+
7	+	+	1° Ab no. 4, concentration 1	+
8	+	+	1° Ab no. 4, concentration 2	+
9	+	+	1° Ab no. 5, concentration 1	+
10	+	+	1° Ab no. 5, concentration 2	+
11	+	+	1° Ab no. 6, concentration 1	+
12	+	+	1° Ab no. 6, concentration 2	+
13	+	+	1° Ab no. 7, concentration 1	+
14	+	+	1° Ab no. 7, concentration 2	+
15	+	+	1° Ab no. 8, concentration 1	+
16	+	+	1° Ab no. 8, concentration 2	+
17	+	+	1° Ab no. 9, concentration 1	+
18	+	+	1° Ab no. 9, concentration 2	+
19	+	+	1° Ab no. 10, concentration 1	+
20	+	+	1° Ab no. 10, concentration 2	+
21	+	+	1° Ab no. 11, concentration 1	+
22	+	+	1° Ab no. 11, concentration 2	+
23	+	+	1° Ab no. 12, concentration 1	+
24	+	+	1° Ab no. 12, concentration 2	+
25	+	+	Only mouse 2° Ab	+
26	+	+	Only rabbit 2° Ab	+
27	+	+	No antibodies	−
28	+	−	No antibodies	−
29	−	−	No antibodies	+
30	−	−	No antibodies	−

¹° Ab, primary antibody

²° Ab, secondary antibody.

Imaging a whole 96-well plate with single-plane images and multiple fields-of-view per well typically takes less than 2 h on a high-content imaging system. Acquisition of images for analysis of 12 antibodies usually takes less than 30 min. If manual movement of samples is required, the process may take 3–4 h. Image analysis, including segmentation and intensity measurements for 12 antibodies, can be completed in less than 2 h with preinstalled scripts and packages. Segmentation and intensity measurements may take 20–40 min each on a system with a data-capable graphics processing unit (GPU). Additional tasks, such as estimating cell diameter, generating background images and file management, may take 30–50 min. Installation and testing of packages and scripts may take an initial hour or two but typically only need to be done once.

Segmentation stands as a critical step in any image analysis pipeline, demanding a reliable algorithm. In our experience, cellpose has proven most reliable in this regard. Initial attempts at segmentation using Fiji proved inaccurate, as cytosolic signal-based cell detection often resulted in fused objects. Consequently, cellpose, operable in a GPU-capable Python environment, played a pivotal role in script development. Subsequently, we opted for intensity measurements using cellpose masks in the Fiji macro language for enhanced ease of modification. Given users’ existing familiarity with Fiji, its scripting language facilitates learning and customization compared with other programming languages.

Materials

Biological materials Cell lines

Cell lines we have used are listed in Supplementary Table 4.

▲ CAUTION All cell lines used in your research should be regularly checked to ensure they are authentic and are not infected with mycoplasma.

Reagents

▲ CAUTION Safety data sheets provided by the supplier must be checked for hazards identification and precaution measures to take when handling.

• BLUelf Prestained Protein Ladder (FroggaBio, cat. no. PM007–0500K)

• Boric acid (Fisher Scientific, cat. no. A73–3)

• Bovine serum albumin (Wisent, cat. no. 800–095)

• Bradford reagent (MilliporeSigma, cat. no. B6916)

• CellTracker deep red dye (Thermo Fisher Scientific, cat. no. C34565)

• CellTracker green chloromethyl derivative of fluorescein diacetate (CMFDA) dye (Thermo Fisher Scientific, cat. no. C2925)

• Dako mounting medium (Dako, cat. no. S3023)

• 4′,6-Diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific, cat. no. D3571)

• Dimethyl sulfoxide (DMSO) (Sigma, cat. no. D8418)

• Fetal bovine serum (Wisent, cat. no. 080450)

• Invitrogen HiMark pre-stained protein standard (Thermo Fisher Scientific, cat no. LC5699)

• IP lysis buffer (Thermo Fisher Scientific, cat. no. 87788)

• l-Glutamine (Wisent, cat. no. 609–065-EL)

• Lithium dodecyl sulfate (LDS) sample buffer (4×) (Thermo Fisher Scientific, cat. no. NP0007)

• LSB, Laemmli SDS sample buffer, reducing (6×) (Thermo Fisher Scientific, cat. no. J61337.AD)

• MES–SDS running buffer (20×) (Thermo Fisher Scientific, cat. no. J62138.K2)

• Methanol (MilliporeSigma, cat. no. MX0485)

• Normal goat serum (NGS) (Gibco, cat. no. 16210–064)

• Normal donkey serum (NDS) (MilliporeSigna, cat. no. S30–100ML)

• PFA 16% aqueous solution, electron microscopy (EM) grade (Electron Microscopy Sciences, cat. no. 15710)

• Penicillin–streptomycin, 100× (Wisent, cat. no. 450–201-EL)

• Peroxidase substrate, Femto (Thermo Fisher Scientific, cat. no. PI34096)

• Peroxidase substrate, regular (Thermo Fisher Scientific, cat. no. PI32106)

• Phosphate saline buffer 1× (PBS; Wisent, cat. no. 311010CL)

• PMA (Abcam, cat. no. ab147465)

• Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Fisher Scientific, cat. no. 23225)

• Poly-l-lysine (Sigma Aldrich, cat. no. P9155–5MG)

• Ponceau S powder (Thermo Fisher Scientific, cat. no. BP103–10)

• Prestained molecular weight marker (FroggaBio, cat. no. PM008–0500)

• Protease inhibitor cocktail mix (Millipore Sigma, cat. no. P8340)

• Protein A dynabeads (Thermo Fisher Scientific, cat. no. 10001D)

• Protein A–peroxidase horseradish peroxidase (HRP) (Millipore Sigma, cat. no. P8651)

• Protein G dynabeads (Thermo Fisher Scientific, cat. no. 10004D)

• Protein L magnetic beads (Thermo Fisher Scientific, cat. no. 88850)

• RIPA lysis buffer (Thermo Fisher Scientific, cat. no. 89901)

• Skim milk powder (Bioshop, cat. no. SKI400.1)

• Sodium hydroxide (NaOH 10.0 N) (VWR, cat. no. BDH7247)

• Sucrose (Fisher Scientific, cat. no. S5–500)

• Transfer buffer TG 10× (Bio-Rad, cat. no. 1610771)

• Trichloroacetic acid (Fisher scientific, cat. no. SA433–500)

• Tris-buffered saline with Tween 20 (TBST) 10× (Cell Signaling Technology, cat. no. 9997)

• TA SDS Running buffer (20×) (Thermo Fisher Scientific, cat. no. LA0041)

• TG SDS Running buffer (10×) (Bio-Rad, cat. no. 1610772)

• Triton X-100 (Thermo Fisher Scientific, cat. no. BP151–500)

• Trypsin (Wisent, cat. no. 325–542)

Antibodies

• Primary antibodies (those we have used are listed in Supplementary Table 5)

• Secondary antibodies: HRP conjugated for WB and Alexa Fluor 555 conjugated for IF (those we have used are listed in Supplementary Table 6)

Equipment

• 15 ml conical tubes (Thermo Fisher Scientific, cat. no. 339658)

• 16-bit sCMOS camera (Molecular Devices)

• 50 ml conical tubes (Thermo Fisher Scientific, cat. no. AM12501)

• 1.5 ml microtubes (Sarstedt, cat. no. 72.706)

• 5 ml microtubes (Wards Sciences, cat. no. 470225–020)

• 96-well plates, clear flat bottom (Revvity, cat. no. 6055300)

• Aura III light source (Lumencor, cat no. 80–10415)

• Cell culture dishes, 150 mm (Fisher Scientific, cat. no. 08–772-6)

• Cell culture incubator (Forma Scientific, cat. no. 1998–081)

• Cell scraper (Sarstedt, cat. no. 83.1830)

• Centrifugal filter unit, Amicon Ultra-15 (Millipore, cat. no. UFC901096)

• DynaMag-2 magnet (Thermo Fisher Scientific, cat. no. 12321D)

• Heat block (Fisher Scientific, cat. no. 11.718)

• iBright chemiluminescence instrument (Thermo Fisher Scientific, cat. no. CL1500)

• ImageXpress Micro Confocal High-Content Imaging System (Molecular Devices)

• Megafuge 16 centrifuge (Thermo Fisher Scientific, cat. no. 75004270)

• Nitrocellulose membrane (Bio-Rad, cat. no. 1620097)

• Open-Top Thickwall polycarbonate tube, 3.5 ml (Beckman Coulter, cat. no. 349622)

• Optima MAX-XP Ultracentrifuge (Beckman Coulter)

  • (Alternative) Sorvall MTX 150 Benchtop Micro-Ultracentrifuge (Fisher Scientific, cat. no 361020365)

• pH meter (Thermo Fisher Scientific, cat. no. STAR2116)

• PhenoPlate, 96-well, optically clear flat bottom (Perkin Elmer, cat. no. 6055300)

• Plate reader (Thermo Fisher Scientific, cat. no. A51119700C)

• Precast BT polyacrylamide gels, 12 wells, midi (Thermo Fisher Scientific, cat. no. WG1201BOX)

• Precast TA polyacrylamide gels, 12 wells, midi (Thermo Fisher Scientific, cat. no. WG1601BOX)

• Precast TG polyacrylamide gels, 12 wells, midi (Thermo Fisher Scientific, cat. no. WXP42012BOX)

• Rotating mixer (Fisherbrand, cat. no. 88861041)

• Shaker, multiplatform (Fisherbrand, cat. no. 88861021)

• Sonicator (Thermo Fisher Scientific, cat. no. FB120A110

• TLA-100.3 fixed-angle rotor (Beckman Coulter, cat. no. 349490)

• Wet protein transfer system—criterion blotter (Bio-Rad, cat. no. 1704070)

• XCell4 SureLock Midi-Cell electrophoresis system (Thermo Fisher Scientific, cat. no. WR0100)

Software

• Fiji (ImageJ 1.54f, https://imagej.net/software/fiji/)

• MetaXpress High-Content Screening Software version 6.7.1.157 (Molecular Devices, https://www.moleculardevices.com/products/cellular-imaging-systems/high-content-analysis/metaxpress#overview)

• PyCharm 2023.3.4 (professional edition, https://www.jetbrains.com/pycharm/)

• RStudio 2023.09.1 (https://cran.rstudio.com/)

Reagent setup

Labeling culture medium

Supplement the appropriate type of complete medium with only 5% vol/vol serum. Store at 4 °C for up to 6 months.

Borate buffer (0.15 M, pH 8.4)

Weigh 4.64 g of boric acid powder. Add the boric acid to a glass beaker containing 450 ml of distilled water and stir until the powder is completely dissolved. Adjust the pH to 8.4 using 10 N NaOH. Complete to 500 ml with distilled water. Store at room temperature (23–25 °C) for up to a year.

CellTracker deep red dye, 1,000× solution

Dissolve 15 μg of CellTracker deep red dye with 40 μl of DMSO. Aliquot into 5 μl samples and store at −20 °C for up to a year.

CellTracker green CMFDA dye stock, 1,000× solution

Dissolve 50 μg of CellTracker green CMFDA dye with 20 μl of DMSO. Aliquot into 5 μl samples and store at −20°C for up to a year.

Complete IP lysis buffer

Add 10 μl of the protease inhibitor cocktail into 1.0 ml of ice-cold IP lysis buffer. Add immediately before using the buffer and keep on ice.

Complete RIPA lysis buffer

Add 10 μl of the protease inhibitor cocktail into 1.0 ml of ice-cold RIPA lysis buffer. Add immediately before using the buffer and keep on ice.

DAPI stock concentration (5 mg/ml)

Dissolve 10 mg of DAPI in 2.0 ml of deionized water. Aliquot and store at −20 °C (can be stored for years).

DAPI working concentration (5 μg/ml)

Add 5 μl of the DAPI stock concentration (5 mg/ml) to 5 ml of deionized water. Prepare 1 ml aliquot and store at −20 °C (can be stored for years).

IF blocking buffer (1× PBS, 0.01% vol/vol Triton X-100, 5% wt/vol BSA and 5% vol/vol NGS or NDS)

Add 2.5 ml of NGS or NDS to 47.5 ml of IF incubation buffer. Mix gently at 4 °C immediately before use.

IF DAPI solution (1× PBS + 5 ng/ml DAPI)

Add 5 μl of DAPI working concentration (5 μg/ml) to 5 ml of 1× PBS. Dilute before use and discard leftover.

IF fixation buffer (0.5× PBS, 8% vol/vol PFA and 20% wt/vol sucrose)

Combine 5 ml of 1× PBS to 5 ml of PFA 16% aqueous solution. Dissolve 2 g of sucrose into a 10 ml solution made of 5 ml of 1× PBS and 5 ml of PFA 16% in water. Store at 4 °C for up to 3 months.

IF incubation buffer (1× PBS, 0.01% vol/vol Triton X-100 and 5% wt/vol BSA)

Add 10 μl of Triton X-100 and 5 g of BSA to 80 ml of 1× PBS. Rock gently at 4 °C until the BSA is completely dissolved. Complete at 100 ml with 1× PBS. Keep on ice, and it can be stored at 4 °C for up to 1 week.

IF permeabilization buffer (1× PBS, 0.1% vol/vol Triton X-100)

Add 50 μl of Triton X-100 to 50 ml of 1x PBS. Mix gently and keep on ice; it can be stored at 4 °C for up to 1 week.

Labeling culture medium

Supplement the appropriate type of medium with only 5% vol/vol serum. Store at 4 °C for up to 6 months.

Poly-l-lysine stock solution (1.0 mg/ml)

Dissolve 5 mg of poly-l-lysine in 4 ml of sterile distilled water to make a stock at 1.0 mg/ml.

Complete to 5 ml with sterile distilled H₂O. Store at room temperature (can be stored for years).

Poly-l-lysine working solution (10 μg/ml)

Dilute the poly-l-lysine stock to 1:100 with 0.15 M borate buffer (pH 8.4) for a final concentration of 10 μg/ml. Sterilize by filtration using a 0.2 μm filter unit. Store at room temperature for up to a year.

Ponceau S working solution

Dissolve 1 g of Ponceau S powder in 485 ml of deionized water. Add 15 ml of trichloroacetic acid. Protect from light and store at room temperature; it can be reused frequently for up to 3 months.

Running buffer MES–SDS 1×

Add 50 ml of MES–SDS running buffer 20× to 950 ml of distilled water. Dilute fresh before use.

Running buffer TA SDS 1×

Add 50 ml of TA SDS running buffer 20× to 950 ml of distilled water. Dilute fresh before use.

Running buffer TG SDS 1×

Add 100 ml of TG SDS running buffer 10× to 900 ml of distilled water. Dilute fresh before use.

Serum-free medium

Supplement the appropriate type of medium with all components except serum. Store at 4 °C for up to 6 months.

1× TBST

Add 100 ml of TBST 10× to 900 ml of distilled water. Store at room temperature for up to 2 weeks.

Transfer buffer TG 1× (20% vol/vol methanol)

Add 150 ml of TG transfer buffer 10× to 1,050 ml of distilled water. Add 300 ml of methanol before transfer. Prepare just before use.

WB blocking solution, used also for primary and secondary antibody preparation

Dissolve 5 g of nonfat milk powder in 100 ml of 1× TBST. Prepare just before use.

Equipment setup

Imaging setup for IF

We use an ImageXpress micro widefield high-content microscope equipped with 395, 475, 555 and 635 nm solid-state light-emitting diode (LED) lights and bandpass filters to excite and capture separately DAPI, CellTracker Green CMFDA, Alexa568 (Alexa Fluor 555) and CellTracker Deep Red, respectively. The filter cube specifications are the following: (1) excitation (the excitation spectra are based on the emission band of the light sources (Lumencor Aura III): blue, 395/25; green, 475/28; red, 555/28 and far red, 635/22, 2) emission: blue, 432/36; green, 520/35; red, 600/37 and far red, 692/40. The objective used is a water Apo LambdaS LWD with magnification of 20×, NA 0.95. For the camera, we use a 16 Bit sCMOS 1.97 mm field of view.

Procedure 1: antibody screening by WB

• TIMING 2 d

Protein extraction

• 1Perform protein extraction. Follow option A to prepare cell lysates for antibodies against intracellular proteins and option B to collect culture media for antibodies against secreted proteins.

Option A: cell lysate preparation for WB—intracellular protein

• TIMING 1.5 h (day 1)

  1. Grow 2 × 150 mm dishes of WT cells and 2 × 150 mm dishes of KO cells to 80% confluence in complete medium. WT and KO cells may grow at different rates and, therefore, require seeding at different densities.
  2. Place the culture dish on ice, remove and discard medium.
  3. Wash the adherent cells three times with ~10 ml of ice-cold 1× PBS. Ensure total removal of PBS between washes using, for example, a vacuum.
  4. After the last wash, add 1.0 ml of RIPA lysis buffer supplemented with 1× protease inhibitor cocktail mix to each 150 mm dish.
  5. Use cell scrapers to detach adherent cells.
  6. Collect and pool the cell lysates from the same condition together into a 5 ml tube.
  7. Sonicate both cell lysates 3× 5 s at 40% amplitude.
     ◆ TROUBLESHOOTING
  8. Rock for 30 min at 4 °C.
  9. Centrifuge at ~110,000g for 15 min at 4 °C using a refrigerated ultracentrifuge.
     ▲ CRITICAL STEP Ultraspeed centrifugation is an effective method for removing insoluble contaminants that might interfere with protein migration in polyacrylamide gels.
     ◆ TROUBLESHOOTING
  10. Gently remove the tubes from the rotor and place them on ice.
  11. Transfer supernatants to fresh 1.5 ml microtubes kept on ice. Discard pellets.
      ■ PAUSE POINT Aliquots of WT and KO cell lysates can be stored at −20 °C for 6 months and at −80 °C for a year.

Option B: culture medium collection for WB—secreted protein

• TIMING 1.5 h (day 1)

  1. Grow 3× 150 mm dishes of WT cells and 3× 150 mm dishes of KO cells to 80% confluence in complete medium. WT and KO cells may grow at different rates and therefore require seeding at different densities.
  2. Wash all dishes three times with sterile warm 1× PBS under a laminar flow cell culture hood.
  3. Add 20 ml of warm serum-free medium to each 150 mm plate.
     ▲ CRITICAL STEP A serum-free medium is used to avoid contaminating the pool of secreted cellular proteins with highly abundant exogenous proteins present in bovine serum.
  4. Incubate plates in an incubator at 37 °C, 5% CO₂ for 18 h.
  5. Collect media in 50 ml conical tubes on ice.
  6. Centrifuge the 50 ml tubes at 500g for 10 min at 4 °C to eliminate cells and large contaminants.
  7. Transfer the supernatants to new 50 ml conical tubes and centrifuge at 4,000g for 10 min at 4 °C to eliminate small contaminants.
  8. Transfer the supernatant to new 50 ml conical tubes on ice.
  9. Add 15 ml of cleared medium to each 15 ml centrifugal filter unit.
     ▲ CRITICAL STEP Selection of the appropriate nominal molecular weight limit depends on the target protein. For example, the 10 kDa cutoff filter units can be used for proteins with a molecular mass greater than 10 kDa.
  10. Centrifuge at 4,000g for 30 min at 4 °C. In each filter unit, a volume of ~500 μl of medium remains after centrifugation, resulting in a ~30-fold concentrated medium.
      ◆ TROUBLESHOOTING
  11. Collect the concentrated media into 1.5 ml microtubes on ice.
  12. Add the corresponding volume of 100× protease inhibitor cocktail mix for a final concentration of 1×.
      ■ PAUSE POINT Aliquots of concentrated media can be stored at −20 °C for 1 year.

Sample preparation for WB

• TIMING 1.5 h (day 1)

  • 2
    Measure the protein concentration using a BCA protein assay kit for lysates (intracellular protein) or a Bradford reagent for media (secreted protein).
    ▲ CRITICAL STEP Precise monitoring of protein concentration is key to interpret antibody specificity by WB. Measurement of protein concentration in triplicate allows more precise and reproducible quantification for both the BCA and Bradford assays.
  • 3
    Dilute WT and KO samples to identical final concentrations:

    • For lysates, dilute in RIPA buffer
    • For concentrated media, dilute in the same serum-free medium used to incubate the cells or in deionized water
      Depending on the abundance of the target, between 20 μg and 50 μg of protein are loaded per lane. Therefore, the volumes needed will depend on the number of antibodies to test.
      ▲ CRITICAL STEP Antibody signal (band intensity) follows a linear and proportional relationship with the protein concentration used to test every antibody/target pair⁴³. To reach this linear range of detection, we load a defined amount of protein depending on the putative abundance of the target in the chosen WT cell line. To do so, we first search the protein target in PAXdb⁴⁴ (https://pax-db.org; RRID: SCR_018910) to determine protein abundance in the selected cell line. PaxDB provides protein abundance in parts per million (ppm) which represents a consistent expression unit in which each protein entity is presented relative to all other protein molecules in the sample⁴⁵. We use 50 μg of lysate or medium for low abundance proteins between 10 and 1,000 ppm and 20 μg of lysate or medium for high abundance proteins with over 1,000 ppm. Even with these considerations, primary antibody titration may be required later.

Protein electrophoresis

• TIMING 2 h (day 1)

  • 4
    Select the appropriate type of polyacrylamide gels on the basis of the target protein size, as set out in Table 2. The 12-well midi precast gels running under denaturing conditions are used here.
  • 5
    Prepare master mixes of both WT and KO samples at the same final protein concentration by adding the appropriate loading sample buffer to a final concentration of 1×. The sample buffer must be compatible with the gel chemistry chosen above (Table 2). The 12-well midi gels used here can hold up to 45 μl of sample per well.
  • 6
    Prepare a master mix of the molecular weight marker in a similar manner. The volume can be increased with RIPA buffer or deionized water to match the volume of the sample master mix. Choose the appropriate type of molecular weight marker for your protein. Two types of molecular weight marker are listed in Table 2; while the BLUelf prestained protein ladder covers proteins from 3.5 kDa to 245 kDa, the HiMark prestained high-molecular weight protein standard is suggested for proteins above 245 kDa.
    ▲ CRITICAL STEP Molecular weight markers vary between different gel chemistries. Users must refer to the manufacturer’s datasheet to label protein ladders appropriately.
  • 7
    Heat the master mixes of protein samples and molecular weight markers for 10 min at 65 °C in a heat block to help dissolve the SDS and/or glycerol in the loading sample buffer, which facilitates the loading into the gel.
    ▲ CRITICAL STEP The samples must not be boiled. For some proteins, boiling samples can create artifacts. The G-protein-coupled receptor S1PR1 runs as two major bands (~44, 48 kDa) with additional minor bands detected below and above both major bands (Fig. 10, left). Boiling the samples led to an intense artifactual smear above 245 kDa and a reduction of the signal at ~44 and 48 kDa (Fig. 10, right). However, in rare exceptions, protein complexes are partially resistant to SDS at room temperature, and boiling the samples is required to identify monomers in WB⁴⁶.
  • 8
    Pulse spin the samples and molecular weight master mixes using a microcentrifuge, and load samples into a 12-well polyacrylamide gels in the order suggested in Extended Data Fig. 1. A total of four antibodies can be tested on each 12-well gel. To test 12 antibodies, 3× 12-well gels are required.
  • 9
    Follow the conditions recommended by the manufacturer to run the gel using the appropriate running buffer (Table 2), until the dye front reaches ~3 mm from the bottom.

Table 2 |.

Recommended polyacrylamide gels depending on the molecular weight on the target protein

Range of protein molecular masses	Type of SDS–PAGE gels	Polyacrylamide Running percentage buffer	Loading sample buffer	Molecular weight marker
3.5–25 kDa	BT	10% MES SDS	LDS (reducing)	BLUelf Prestained Protein Ladder
26–500 kDa	TG	4–20% TG SDS	LSB (reducing)	BLUelf Prestained Protein Ladder or HiMark Pre-Stained High Molecular Weight protein standard

Fig. 10 |. Boiling protein samples creates an aggregation artifact.

WBs are presented as in Fig. 2. The lysates of a cell line expressing endogenous levels of the transmembrane protein S1PR1 (UniProt ID: P21453) were prepared, and the samples were either heated at 65 °C or 95 °C for 10 min. The single asterisks indicate the major bands representing S1PR1, whereas the double asterisk indicates at the aggregated form. The expected molecular mass for S1PR1 is 42.8 kDa. The 4–20% TG gel was used.

Protein transfer to nitrocellulose membrane

• TIMING 1.5 h (day 1)

  • 10
    Transfer proteins from the gel to a membrane; nitrocellulose membranes are used here. The Bio-Rad criterion blotter is the wet transfer system that we have employed. Transfer conditions are detailed in Table 3.
  • 11
    Take the membrane out of the sandwich and wash twice with deionized water in a container.
  • 12
    Stain all transferred proteins by covering the membranes with a Ponceau S solution.
  • 13
    Incubate for 1 min. Ponceau S staining solution can be reused several times.
  • 14
    Wash off excess Ponceau S with deionized water until the area of membranes not covered by proteins becomes white.
  • 15
    Dry the Ponceau S-stained membranes on Whatman filter paper.
  • 16
    Label membranes properly with a suitable smudge proof pen. For this platform, marking each lane with the catalog number of the antibody, as well as the host species, is mandatory.
  • 17
    Scan membranes using a regular paper scanner. Ponceau S-stained lanes can be later presented as loading controls.
    ◆ TROUBLESHOOTING
  • 18
    Trim membrane strips containing each molecular weight marker and a WT–KO sample pair.
    ■ PAUSE POINT Dried membranes can be stored at room temperature for months.

Table 3 |.

Protein transfer settings

Type of polyacrylamide gels	Transfer buffer	Methanol percentage	Power and time settings	Use of ice pack to reduce overheating
BT	TG	20%	45 V, 1 h	No
TG	TG	20%	85 V, 1 h	Yes

Blocking and primary antibody incubation

• TIMING 1 h and overnight (O/N) (day 1 and 2)

  • 19
    Rehydrate membrane strips and remove Ponceau S staining by incubating the membrane in 1× TBST for 5 min.
  • 20
    Block the membrane with the WB blocking solution for 1 h at room temperature.
  • 21
    During the blocking step, prepare the primary antibody solution in WB blocking solution following manufacturer’s recommendations.
    ◆ TROUBLESHOOTING
  • 22
    Incubate each membrane strip with the corresponding primary antibody dilution.
    A container or a resealable flat plastic bag can be used for the incubation.
  • 23
    Rock overnight at 4 °C.

Secondary antibody incubation and signal detection

• TIMING 3 h (day 2)

  • 24
    Discard antibody solution and wash membrane strips in a container with 1× TBST for 10 min under constant rocking. Repeat twice more for a total of three washes.
  • 25
    During washes, dilute the corresponding HRP-conjugated secondary antibody in WB blocking solution to a concentration of 0.2–0.5 μg/ml.
  • 26
    Rock membrane blot strips with the corresponding HRP-conjugated secondary antibody in a container or a sealed plastic bag for 1 h at room temperature.
  • 27
    Wash three times in 1× TBST for 10 min with constant rocking.
  • 28
    After the last wash, place membrane strips on a clean surface and incubate with the peroxidase substrate for 1 min, then remove excess.
  • 29
    Place membrane strips in a chemiluminescence imaging system and follow the manufacturer’s guidelines for signal detection. The iBright imaging system is used here.
    ▲ CRITICAL STEP Signal strength varies depending on the antibody and its concentration. Several exposures must be typically taken to observe the bands of interest at intensities comparable between the different antibodies.
  • 30
    Name images properly and export the different exposures for later figure preparation.
  • 31
    Analyze the band pattern to assess antibody specificity (Fig. 2a).
    ▲ CRITICAL STEP A specific antibody will produce a signal in the WT sample but not in the KO, confirming the endogenous expression of the target protein in the WT cells. Identification of a cell line with confirmed endogenous expression of the target is essential to proceed with IP–WB and IF procedures.
    ◆ TROUBLESHOOTING
    ■ PAUSE POINT The next procedures can be started at any time after screening antibodies by WB was successful.

Procedure 2: antibody screening by IP

• TIMING 3 d

  ▲ CRITICAL Protein extractions for WB and IP are similar but differ in the greater amount of protein required for the IP protocol. Additionally, for IP, a nondenaturing lysis buffer is necessary, and the lysates need to be freshly prepared. Lysates for IP should not be sonicated to prevent protein denaturation, thereby preserving protein–protein interactions. Concentrated media must be freshly collected from the cells as well.

Protein extraction

• 1Perform protein extraction. Follow option A to prepare cell lysates for antibodies against intracellular proteins and option B to collect culture media for antibodies against secreted proteins.

Option A: cell lysate preparation for IP—intracellular protein

• TIMING 1.5 h (day 1)

  1. Grow 7× 150 mm dishes of WT cells to 80% confluence to generate enough lysate for 12 IPs.
  2. Place each culture dish on ice and discard medium.
  3. Wash the adherent cells three times with ~10 ml of ice-cold 1× PBS. Ensure total removal of PBS between washes with, for example, a vacuum.
  4. After the last wash, add 1.0 ml of ice-cold IP lysis buffer supplemented with 1× protease inhibitor cocktail mix to each 150 mm dish.
  5. Use cell scrapers to gently detach adherent cells from the Petri dish.
  6. Collect and pool WT cell lysates into a 15 ml tube. If fewer dishes are used, the cells can be collected into a 5 ml tube.
  7. Rock cell lysates for 30 min at 4 °C.
  8. Centrifuge at ~110,000g for 15 min at 4 °C (as with WB).
     ▲ CRITICAL STEP Ultraspeed centrifugation pellets insoluble contaminants that could otherwise adhere to the bead–antibody conjugate and, subsequently, interfere with the detection of the captured protein in WB.
     ◆ TROUBLESHOOTING
  9. Gently remove the tubes from the rotor and place them on ice.
  10. Pool the supernatants into the same tube kept on ice. Discard the pellets.
      ▲ CRITICAL STEP Freshly prepared lysates must be used for the IP experiment. Freezing the protein sample might affect the epitope to be recognized by the antibodies tested.

Option B: culture media collection for IP—secreted protein

• TIMING 1.5 h (day 1)

  1. Grow 13× 150 mm dishes of WT cells to 80% confluence for 12 IPs.
  2. As described in the WB procedure (Procedure 1, option B(ii–xii)), wash and grow the cells without serum for 18 h, then collect and concentrate media.
  3. Combine all concentrated media from WT cells into the appropriate tube.
  4. Add the corresponding volume of 100× protease inhibitor cocktail mix for a final concentration of 1× and keep on ice.
     ▲ CRITICAL STEP Freshly collected and concentrated medium must be used for the IP experiment.

Sample preparation for IP

• TIMING 1.5 h (day 1)

  • 2
    Measure protein concentration using a BCA protein assay kit (lysate) or Bradford reagent (medium).
  • 3
    Adjust protein concentrations:

    • For lysates, adjust protein concentration to 2.0 mg/ml with IP lysis buffer. A total of 1 mg (500 μl at 2.0 mg/ml) is used later for each IP
    • For concentrated media, the concentration is usually at ~1 mg/ml. A total of 0.5 mg (500 μl at 1.0 mg/ml) is used later for each IP
  • 4
    Save enough protein lysate for starting material samples to be run side by side with each unbound fraction and immunoprecipitate. The platform consists of using 4% starting material samples. For 12 antibodies, ~300 μl are saved and kept on ice for subsequent immunoblots, allowing side-by-side comparison with each IP.

    • A 4% starting material sample from a 1.0 mg of protein lysate consists of 40 μg (20 μl of the 2.0 mg/ml lysate)
    • A 4% starting material sample from a 0.5 mg of medium consists of 20 μg (20 μl of the 1.0 mg/ml medium)

Antibody–beads conjugation

• TIMING 1.5 h (day 1)

  • 5
    Select the appropriate type of magnetic beads for each antibody⁴⁷ (Table 4).
  • 6
    Resuspend the beads slurry by pipetting repeatedly.
  • 7
    Label enough microtubes (one microtube per antibody to be tested) and add 30 μl of magnetic bead slurry to 1.0 ml of IP lysis buffer.
  • 8
    Add 2.0 μg of the corresponding primary antibody.
    ▲ CRITICAL STEP When screening multiple antibodies, an antibody that fails at capturing the protein of interest usually serves as a negative control. However, when testing fewer than three antibodies, other negative controls must be included to the experiment (corresponding isotype-matched controls and/or bead-only control).
    ◆ TROUBLESHOOTING
  • 9
    Maintain constant agitation on a rotating mixer for 1 h at 4 °C.
  • 10
    Place tubes on the DynaMag-2 magnet and allow 15 s for the beads to attach to the magnet.
  • 11
    Vacuum out the buffer to remove unbound antibodies.
  • 12
    Add 1.0 ml of IP lysis buffer and allow release of the beads by taking the tube off the magnet.
  • 13
    Wash the beads by inverting the tube multiple times to resuspend.
  • 14
    Repeat Steps 10–12 to wash a second time and remove the excess unbound antibodies.
    ▲ CRITICAL STEP Do not let the beads dry out at any step.

Table 4 |.

Types of bead and species reactivity

Species immunoglobulin isotype	Type of magnetic beads
Rabbit all isotypes	Protein A
Mouse IgG	Protein G
Mouse IgM	Protein L
Goat all isotypes	Protein G
Chicken all isotypes	Protein G
Guinea pig all isotypes	Protein A
Sheep all isotypes	Protein G

IP

• TIMING 2 h (day 1)

  • 15
    Remove the buffer from the antibody–bead conjugate using the magnet.
  • 16
    Add the protein sample to each tube of antibody-conjugated beads:

    • For lysates, add 1.0 mg (500 μl at 2.0 mg/ml)
    • For concentrated media, add 0.5 mg (500 μl at 1.0 mg/ml)
  • 17
    Incubate antibody–bead conjugates with the protein sample for 1 h at 4 °C with constant agitation on a rotating mixer.
    ◆ TROUBLESHOOTING
  • 18
    Place each microtube on DynaMag-2 and allow at least 15 s for the beads to converge toward the magnet.
  • 19
    From each tube, collect 20 μl of supernatant, which represents the unbound fraction (proteins that did not bind to the antibody–bead conjugate) and contains the same amount of protein as the starting material.
  • 20
    Pipet each unbound fraction in a labeled microtube, set aside on ice.
  • 21
    Vacuum out any remaining samples from each tube on the DynaMag-2.
  • 22
    Wash the magnetic beads 3× in 1 ml IP lysis buffer supplemented with 1× protease inhibitor cocktail mix.
  • 23
    After the last wash, elute with 30 μl of the appropriate reducing loading sample buffer diluted to 1× in IP lysis buffer. When working on the same target, use the same loading sample buffer selected previously from Table 2 and added in Procedure 1, Step 5.
    ◆ TROUBLESHOOTING
  • 24
    Add reducing loading sample buffer to the starting material and unbound fractions.
  • 25
    Similarly, prepare the molecular weight marker sample. The volumes should be completed with IP buffer to match sample volumes.
    ▲ CRITICAL STEP The final concentration of loading sample buffer must be identical in all samples (usually 1×).
  • 26
    Heat all samples for 10 min at 65 °C in a heat block.
    ■ PAUSE POINT The samples can be left at room temperature for 1 d or stored at −20 °C for several weeks.

WB assessment of antibody performance by IP

• TIMING 4.5 h (day 2), O/N and 3 h (day 3)

  • 27
    Select the same polyacrylamide gel as selected for the WB screening (Table 2).
  • 28
    Load samples on 12-well polyacrylamide gels in the order suggested in Extended Data Fig. 1b. A total of three antibodies can be evaluated from a single 12-well gel. To test 12 antibodies, 4 × 12-well gels are required.
  • 29
    Perform the WB as detailed in Procedure 1, Step 19–30 using KO-validated antibodies selected from the WB screening. One primary antibody is ideally selected and used in WB to assess the performance of all antibodies tested by IP. Ideally, this selected antibody would be renewable with high specificity toward its target.
  • 30
    Analyze the data by comparing the WB signal in the starting material, unbound fraction and immunoprecipitate (Fig. 7). An antibody that efficiently captures its intended target should enrich the target in the IP fraction and deplete it from the unbound fraction.
    ◆ TROUBLESHOOTING

Procedure 3: antibody screening by IF

• TIMING 4 d

Parental/KO mosaic preparation in 96-wells for IF

• TIMING 2 h (day1)

  • 1
    Grow 1× 150 mm dish of WT cells and 1× 150 mm dish of KO cells to 80% confluence for 12 antibodies. WT and KO cells may grow at different rates and, therefore, require seeding at different densities. A total of 30 wells is required to test 12 antibodies at two different concentrations together with necessary controls.
  • 2
    Coat each well of a 96-well, clear flat bottom plate with 100 μl of poly-l-lysine working solution. The suggested 96-well plate (Revvity, cat. no. 6055300) is compatible with cell imaging and most high-content imaging systems.
  • 3
    Incubate at 37 °C, 5% CO₂ for 1 h in a cell culture incubator.
  • 4
    Wash each well twice with 100 μl of sterile water.
  • 5
    Wash 150 mm dishes of WT and KO lines with 10 ml of warm 1× PBS.
  • 6
    Add 5.0 ml of warm trypsin to both the WT and the KO cell dishes.
  • 7
    Incubate dishes at 37°C, 5% CO₂ for 2 min in a cell culture incubator.
  • 8
    Confirm that cells have detached from the plate by using a bright field microscope or visually inspecting the plate. If cells are still attached, continue the incubation in the incubator until they have detached. Incubation time will vary between cell lines.
  • 9
    Inactivate trypsin by adding 5.0 ml of complete culture medium.
  • 10
    Collect WT and KO cells in separate 15 ml canonical tubes.
  • 11
    Centrifuge at 1,500g for 3–5 min to pellet cells.
  • 12
    Discard the supernatant.
  • 13
    Resuspend the WT cell pellet with 2.0 ml of labeling culture media containing 5 μM of CellTracker Green CMFDA Dye. Keep ~100,000 WT cells unlabeled for controls (Table 1, well numbers 28 and 29).
  • 14
    Resuspend the KO cell pellet with 2.0 ml of labeling culture media containing 1 μM of CellTracker Deep Red Dye. Keep ~100,000 KO cells unlabeled for controls (Table 1, well numbers 27 and 28).
  • 15
    Incubate the cell suspensions in a cell culture incubator for 30 min with the lid slightly open.
  • 16
    Gently tap the bottom of each tube every 5 min to put the cells back in suspension.
  • 17
    Centrifuge both 15 ml tubes at 1,500g for 3–5 min.
  • 18
    Discard the supernatant.
  • 19
    Resuspend each labeled cell pellet with complete medium and count the cells.
  • 20
    Prepare a master mix of 100 μl per well in cell culture media, containing labeled WT:KO cells at 1:1 ratio. For most cancer cell lines, use 10,000 WT cells combined with 10,000 KO cells (20,000 cells in total) to achieve 50–60% confluence after an overnight culture in 96-well plate, which is optimal for subsequent imaging. For control wells numbers 27, 28 and 29 (Table 1), combine the corresponding labeled cells with unlabeled cells.
  • 21
    Plate WT–KO mixed cells in 29 wells of a 96-well plate following guidelines provided in Table 1. Add only 100 μl of culture media to well number 30.
  • 22
    Incubate the 96-well plate overnight at 37 °C in the cell incubator.

Fixation in 96-well plates

• 23Add 100 μl of prewarmed (37 °C) IF fixation buffer on top of the culture medium in each well.

The final concentration of PFA is 4%.

• 24Incubate the 96-well plate for 15 min at 37 °C.
• 25Aspirate and wash each well 3× with 100 μl PBS at room temperature.

  ■ PAUSE POINT The plates can be stored at 4 °C for a few days. Protect the plates from light.

  Seal the plate using parafilm to avoid evaporation.

Primary antibody staining

• TIMING 1 h and O/N (day 2)

  • 26
    Incubate wells from the 96-well plate with 100 μl of IF permeabilization buffer for 10 min at room temperature.
  • 27
    Wash wells three times with 100 μl of 1× PBS.
  • 28
    Incubate wells with 100 μl of IF blocking buffer for 30 min at room temperature.
    ▲ CRITICAL STEP To decrease nonspecific binding of the secondary antibody to reactive sites on fixed cells, a practical approach involves preincubation with serum sourced from the same host species as the secondary antibody. Except for when using goat primary antibodies, the secondary antibodies we have used (Supplementary Table 6) have been raised in goat, and IF blocking buffer includes NGS.
  • 29
    During the incubation, prepare a 100 μl of each primary antibody dilution in IF incubation buffer.
  • 30
    Incubate wells with the appropriate antibody dilution overnight at 4 °C or with IF incubation buffer for control conditions.

Secondary antibody labeling

• TIMING 2 h (day 3)

  • 31
    Wash the wells three times for 10 min with 100 μl of IF incubation buffer.
  • 32
    Incubate wells for 1 h at room temperature with the corresponding secondary antibody dilution containing either 0.1 μg/ml of goat secondary antibodies coupled to Alexa 555 in IF incubation buffer or with IF incubation buffer alone for specific control wells (Table 1). Protect the plates from light.
  • 33
    Wash the wells three times for 10 min with 100 μl of 1× PBS.
  • 34
    Incubate the wells with 1× PBS containing 5 ng/ml of DAPI for 3 min, except specific control wells.
  • 35
    Wash once with 100 μl of 1× PBS.

Imaging

• TIMING 2 h (day 3)

  • 36
    Image using the ImageXpress micro widefield high-content microscope:

    • Attenuate or optimize the excitation light intensity on the basis of the signal intensity of different antibodies, taking care to ensure there no saturating pixels
    • On the Camera, set a binning of 2 with a calibration (binned) of 0.6792 × 0.6792 um. Set the target maximum intensity at 33,000
    • Set the Z-series at a two-dimensional projection image only
      ◆ TROUBLESHOOTING
  • 37
    3Image the wells (see Fig. 9a for the raw images). The average number of WT and KO cells imaged per condition is minimally 500.
    ▲ CRITICAL STEP Sequential imaging setup is preferred to avoid any bleed through between channels. To control for bleed through when imaging three or four channels, four different controls are needed for each imaging experiment. Use the same setting for imaging the control samples. Image the single labeled control with all filters sets and carefully analyze potential bleed through in the unlabeled channels. Adjust the emission spectra for each channel so that there is no/minimal bleed through into the unlabeled channels. To control for auto fluorescence, image an unlabeled cell sample that has gone through the staining protocol, in each of the four channels.

Image analysis

• TIMING 2 h (day 4)

  ▲ CRITICAL We describe the image analysis pipeline we use. Alternatively, antibody performance can be assessed visually⁵. Antibodies that can immunolocalize their protein target will generate a specific signal in the WT cells and a signal in KO cells comparable with the image background (area outside a cell). For a larger initiative, the automatic cell segmentation and quantification allow a more robust and reliable comparison of antibody performance between antibodies since antibody signals can be measured from hundreds of cells.

  • 38
    Inspect the cell mask channels using Fiji across multiple wells to make sure the staining is distinct from background noise. Use the ellipse tool to estimate cell diameter in pixels.
  • 39
    Segment the cells by running the cellpose segmentation⁴⁸ pipeline on the cell mask images. It is recommended to set up cellpose in a Conda environment on a Compute Unified Device Architecture-capable GPU-equipped system and to use a script to batch process all images from all wells from a plate. Use the cell diameter estimated in step 37 as an input parameter and choose the cytosol-specific (‘cyto’) model. The output image files after running cellpose are labeled masks of cells detected in each image, saved in the same folder as the raw images, with ‘_cp_masks.png’ as a suffix. These files will be used in Step 40 for antibody staining intensity quantification. (Fig. 9b). For this step, we provide a script written in Python (cellpose_batch_ycharos_IMX_images.py).
  • 40
    Open the antibody channel images from the empty well (control well no. 30) in Fiji and generate a minimum intensity projection of these images to get an estimate of the image background. Save the minimum intensity projection image as ‘bg_baseline.tif’ in the same folder as the raw images and cellpose-generated masks. This ‘bg_baseline.tif’ image is used in the analysis pipeline to estimate and subtract background in the next step.
  • 41
    Execute the tasks outlined in the provided Fiji script (main_ycharos_IMX_images_script_Fiji. ijm). This script automatically opens the images and extracts data from intensity and masks images. Detailed instructions and the script are available on GitHub. The script performs the following general steps:

    • For each antibody image, a thresholded binary image is generated by first calculating a pixel intensity value threshold using the Otsu method and converting the intensity image to a binary masks image
    • For each antibody image, the median intensity for all pixels outside of cellpose and Otsu thresholded objects are calculated, and the intensity is divided by the base background median intensity. The base background image is then multiplied by that ratio and the resulting image subtracted from the antibody image. The resulting image is a background-subtracted antibody staining image, based on the background image obtained from an empty well, scaled to within image background intensities
    • In the background-subtracted antibody images, the intensity and dimensions statistics (mean, standard deviation, median, area, xy coordinates and so on) are measured for each mask.
  • 42
    Calculate the mean antibody intensity ratio using the data table generated from the previous step by dividing the mean antibody signal intensity in WT cells by that in KO cells for each image. Plot these ratios for each antibody tested using the provided R script (calculate_ratios_and_plotting_template.R) (Fig. 9e). Note that the error bars are not included, as independent replicates were not performed.
    ◆ TROUBLESHOOTING
  • 43
    Generate cropped images with cell mask outlines overlaid on both antibody and DAPI images (Fig. 9c,d) using the Fiji helper script (crop_and_make_figure_panels.ijm).

Troubleshooting

Procedure 1, option A(vii). Problem: ineffective or excessive sonication

During protein extraction, sonication is used to shear DNA and rupture cellular membranes, facilitating the successful release of proteins. It is imperative that sonication is performed on ice or in a cold room to maintain the integrity of the proteins. In this protocol, sonication is set at 40% amplitude for three cycles of 5 s pulses for the specified 1 ml volume. However, it is crucial to adjust the sonication pulses according to your working volume to prevent excessive heating, which may result in protein denaturation and aggregation.

Procedure 1, option A(ix) and Procedure 2, option A(viii). Problem: ultracentrifuge is inaccessible

In situations in which an ultracentrifuge is unavailable, high-speed centrifugation at 15,000–20,000g can serve as an alternative, and the centrifugation time must be optimized by the user. We suggest starting with 1 h as a baseline.

Procedure 1, option B(x). Problem: low concentration of secreted proteins in the medium

If the protein concentration resulting from centrifugation of the 15 ml centrifugal filter unit is low, 0.5 ml centrifugal filter units can be used to concentrate further.

If the resulting concentration is still too low to work with, for antibody characterization purposes, the incubation time in option B(iv) can be extended to more than 18 h to increase the amount of target protein found in the medium. Depending on the cell type and target protein, 18 h of incubation in serum-free medium may not result in detectable levels of the target protein in medium by WB. It is important to note that serum starvation induces considerable cellular stress and thus should not exceed 36 h.

Procedure 1, Step 17. Problem: inconsistent evaluation of protein transfer

Routine staining of membranes with Ponceau S staining provides a valuable insight into the quality of protein transfer from the gel to the membrane. Incomplete or failed protein transfer can be readily identified through Ponceau staining, prompting evaluation and avoiding potential repetition of experiments. Ponceau S-stained membranes are consistently included in the YCharOS antibody characterization reports as both loading and quality controls.

Procedure 1, Step 21. Problem: undetectable or saturated signal in WB

The initial concentration of primary antibody tested follows the manufacturer’s recommendations. However, when the manufacturer does not recommend the antibody for WB, it is first tested at a concentration of 1 μg/ml. For any tested antibody, titration is required if the resulting signal falls outside the linear dynamic range. Specifically, the concentration is increased fivefold when the signal is absent after a prolonged exposure time of 20 min. Conversely, if a saturated signal is detected at a minimal exposure time of 1 s, as determined by the iBright analysis software, the concentration is reduced fivefold.

Procedure 1, Step 31. Problem: inconclusive WB screening

When WB screening yields inconclusive results, the interpretation should focus on potential antibody or cell line failures. If multiple antibodies, each recognizing distinct epitopes, detect a band at the expected molecular weight that remains in the KO lysate, it suggests a failure in the KO line, and testing additional KO clones would be advisable. Conversely, if no antibody shows a band at the approximate molecular weight, it indicates a lack of protein expression in the WT cell line or that the antibodies are not specific to the target. In this scenario, identifying an appropriate cell line background, as outlined in Fig. 4, becomes necessary, as well as testing additional antibodies when available.

Procedure 2, Step 8. Problem: unavailable initial antibody concentration

If the antibody concentration is not provided by the manufacturer, follow their volume recommendation for IP. In the absence of a volume recommendation and antibody concentration, test a certain volume of the antibody and keep a record of the volume used. In such a scenario, between 5 and 10 μl are typically tested at this platform.

Procedure 2, Step 17. Problem: insufficient IP

The antibody–bead conjugate is incubated with the protein sample for 1 h for convenience, which typically results in at least one antibody effectively capturing the target in the IP.

However, if none of the tested antibodies efficiently capture the target protein, the antibody– protein extract incubation can be extended to up to 18 h at 4 °C.

Procedure 2, Step 23. Problem: ineffective IP elution

Reducing agents in sample buffers are crucial for eluting antibody–protein complexes from the beads by breaking down the disulfide bonds in proteins. To prevent inconsistencies in IP results, it is important to always freshly add reducing agent to the sample buffer.

Procedure 2, Step 30. Problem: inconclusive IP screening

If none of the antibodies enrich the target protein, consider increasing the amount of primary antibody used in the IP, as well as the incubation time, as in troubleshooting for Step 17 of this procedure. The target protein must be detectable by WB in the starting material sample to confirm the successful extraction of the intracellular protein in nondenaturing IP lysis buffer. If the protein is not detectable in the total lysate, target extraction in IP buffer should be monitored by performing a WB on both the solubilized supernatant and the pellet resulting from centrifugation of the lysate (Procedure 2, option A(viii)).

Procedure 3, Step 35. Problem: high-content microscope is inaccessible

The advantage of a high-content microscope is the speed and automation of image acquisition. However, regular fluorescence widefield microscopes and point-scanner confocal microscopes are compatible with the protocols described here. As mentioned in Step 36, a minimum of 500 WT and KO cells should be imaged per antibody. This is still feasible with a manual microscope and typically results in imaging six fields of view with a 20× objective. The authors of the initial paper describing these protocols used a point-scanner confocal microscope⁵.

Procedure 3, Step 41. Problem: inconclusive IF screening

In cases in which no antibodies yield specific signal for the intended protein target, alternatives to fixative reagents and detergents may be explored. The YCharOS team has observed that using a different cell line background with higher endogenous protein levels of the target has enabled the identification of specific antibodies, following the same described protocols (Fig. 3b).

Timing

These timelines do not include troubleshooting or cell preparation.

Procedure 1, WB: 2 d

Day 1

Step 1, option A(i–xi) or option B(i–xii), protein extraction: 1.5 h

Step 2–3, sample preparation for WB: 1.5 h

Step 4–9, protein electrophoresis: 2 h

Step 10–18, protein transfer to nitrocellulose membrane: 1.5 h

Step 19–23, blocking and primary antibody incubation: 1 h and O/N

Day 2

Step 24–31, washing, secondary antibody incubation and signal detection: 3 h

Procedure 2, IP: 3 d

Day 1

Step 1, option A(i–x) or option B(i–iv), protein extraction: 1.5 h

Step 2–4, sample preparation for IP: 1.5 h

Step 5–14, antibody–beads conjugation: 1.5 h

Step 15–26, IP: 2 h

Days 2 and 3

Step 27–29, protein electrophoresis: 2 h; protein transfer to nitrocellulose membrane: 1.5 h; blocking and primary antibody incubation: 1 h and O/N

Day 3

Step 29, washing, secondary antibody incubation and signal detection: 3 h

Procedure 3, IF: 4 d

Day 1

Step 1–24, cell mosaic preparation: 2 h

Day 2

Step 25–29, blocking and primary antibody staining: 1 h and O/N

Day 3

Step 30–34, washing and secondary antibody labeling: 2 h

Step 35–36, cell imaging: 2 h

Day 4

Step 37–42, image analysis: 2 h

Anticipated results

Several key steps are necessary before carrying out this antibody characterization platform. First, it is crucial to determine, using databases and/or published literature, whether the protein of interest is secreted from the cells, as sample preparation differs for secreted versus intracellular proteins (Fig. 5). Additionally, selecting a cell line with appropriate endogenous protein expression is essential for effective antibody testing (Fig. 3). If commercial KO lines are unavailable, it may be necessary to compare protein expression across various cell line using transcriptomic databases (Fig. 4a) before generating a KO line (Fig. 4b). A key aspect of these protocols is the iterative validation process, which requires the confirmation of target expression in the cell line and antibody specificity. This is initially determined by the absence of the target protein in the KO line using WB (Procedure 1). To interpret the initial WB results and proceed with the remaining procedures, at least one specific antibody must be identified. In WB, a specific antibody will provide a detectable signal in the WT cells but not in the KO cells (Fig. 2).

For sample preparation, WB uses a denaturing lysis buffer to ensure complete protein solubilization, while IP employs a nondenaturing buffer to maintain compatibility with downstream mass spectrometry applications. The most appropriate gel type used for WB should be selected depending on the molecular weight of the target protein (Fig. 6). Despite the robustness of the WB, IP and IF protocols described here, differences in buffers, blocking reagents, antibody dilutions and other parameters can influence antibody performance. For instance, boiling samples before WB can impact results (Fig. 10). To ensure data reproducibility in publications in which antibodies are employed, it is imperative to meticulously document experimental setups, including all specific conditions and protocols.

While the platform presented is suitable for most proteins, it may require optimization for certain cases to achieve the desired signal-to-noise ratio. Notably, a KO-based methodology may not be applicable for evaluating antibodies targeting post-translational modifications or essential genes. Hence, end users are strongly encouraged to conduct validation experiments in their own laboratories, as disparities in protocols and cell lines can influence results. Nevertheless, using these protocols offers a productive strategy for assessing the application-specific performance of antibodies and guiding their selection.

Once adequate target expression in WT cells and absence of expression in KO cell lines have been validated, antibodies can be screened in IP (Procedure 2) and in IF (Procedure 3) simultaneously or separately. In IP, effective antibodies will capture and enrich the target protein (Fig. 7). In IF, using a WT–KO mosaic strategy will minimize user and imaging biases during antibody screening (Fig. 9a). A segmentation and analysis tool for IF is made available (Fig. 9). A specific antibody in IF will immunolocalize the target protein by producing a signal in the WT that is absent in the KO cells, in which the signal should be comparable with the background noise, defined as the signal measured outside the cells.

Under the finite conditions tested, researchers will identify antibodies with varying specificities (Fig. 2). Recombinant antibodies should be prioritized to ensure reproducibility and reduce reliance on animal-derived products. Although these protocols enable the identification of high-performing antibodies, it is essential to confirm their specificity in different cellular or tissue contexts.

We anticipate that these consensus protocols will streamline early-stage antibody characterization by researchers, meeting the resource validation and data reproducibility standards increasingly required by journals and funding agencies. Broad and open dissemination of characterization data by multiple antibody users will facilitate the rapid identification of renewable and specific antibodies for human proteins, ultimately benefiting the global life sciences community.

Extended Data

Extended Data Fig. 1 |. Order of sample loading for antibody screening in WB and IP-WB.

a) A scanned Ponceau S-stained membrane used for antibody screening in WB. Master mixes of MWM, WT and KO lysates were prepared, and samples were loaded in the following order on a 12-well SDS–PAGE gel: MWM, WT and KO lysates. Commercial 12-well gels have two side wells. The right side well was used here for MWM (+). Up to 4 antibodies can be tested in WB on a single 12-well gel. b) Scanned Ponceau S-stained membrane used for antibody screening in IP. Samples are loaded in the following order on a 12-wells SDS–PAGE gel. Up to 3 antibodies can be tested in IP on a single 12-well gel. Transfers from 4–20% TG gels are shown as examples in a) and b). MWM=molecular weight marker, SM=starting material, UB=unbound fraction, IP=immunoprecipitate.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41596-024-01095-8.

Key points.

• YCharOS is a knockout-based consensus platform for antibody characterization developed through a collaboration between academia and industry. The platform enables direct comparisons among research antibodies that target a specific protein in three common applications: western blot, immunoprecipitation and immunofluorescence.

• The use of the YCharOS consensus protocols and open data dissemination facilitates a community-driven approach to identifying one, or ideally two, renewable and specific antibodies for each human protein.

Data availability

Underlying data can be found at Zenodo via the following links: https://doi.org/10.5281/zenodo.8356134 (ref. 49), https://doi.org/10.5281/zenodo.10844512 (ref. 50), https://doi.org/10.5281/zenodo.10108291 (ref. 51), https://doi.org/10.5281/zenodo.10149969 (ref. 52), https://doi.org/10.5281/zenodo.12666747 (ref. 53), https://doi.org/10.5281/zenodo.12636746 (ref. 54), https://doi.org/10.5281/zenodo.7459248 (ref. 55), https://doi.org/10.5281/zenodo.10835290 (ref. 56), https://doi.org/10.5281/zenodo.7671286 (ref. 57), https://doi.org/10.5281/zenodo.10835327 (ref. 58), https://doi.org/10.5281/zenodo.7459387 (ref. 59), https://doi.org/10.5281/zenodo.10838677 (ref. 60), https://doi.org/10.5281/zenodo.4724176 (ref. 61), https://doi.org/10.5281/zenodo.10845536 (ref. 62), https://doi.org/10.5281/zenodo.13151151 (ref. 63), https://doi.org/10.5281/zenodo.7971951 (ref. 64), https://doi.org/10.5281/zenodo.10839338 (ref. 65), https://doi.org/10.5281/zenodo.7459541 (ref. 66), https://doi.org/10.5281/zenodo.7671135 (ref. 67), https://doi.org/10.5281/zenodo.10819348 (ref. 68), https://doi.org/10.5281/zenodo.10927535 (ref. 69), https://doi.org/10.5281/zenodo.10819189 (ref. 70) and https://doi.org/10.5281/zenodo.10839647 (ref. 71).