3D Visualization of Human Blood Vascular Networks Using Single-Domain Antibodies Directed against Endothelial Cell-Selective Adhesion Molecule (ESAM)

High-quality three-dimensional (3D) microscopy allows detailed, unrestricted and non-destructive imaging of entire volumetric tissue specimens and can therefore increase the diagnostic accuracy of histopathological tissue analysis. However, commonly used IgG antibodies are oftentimes not applicable to 3D imaging, due to their relatively large size and consequently inadequate tissue penetration and penetration speed. The lack of suitable reagents for 3D histopathology can be overcome by an emerging class of single-domain antibodies, referred to as nanobodies (Nbs), which can facilitate rapid and superior 2D and 3D histological stainings. Here, we report the generation and experimental validation of Nbs directed against the human endothelial cell-selective adhesion molecule (hESAM), which enables spatial visualization of blood vascular networks in whole-mount 3D imaging. After analysis of Nb binding properties and quality, selected Nb clones were validated in 2D and 3D imaging approaches, demonstrating comparable staining qualities to commercially available hESAM antibodies in 2D, as well as rapid and complete staining of entire specimens in 3D. We propose that the presented hESAM-Nbs can serve as novel blood vessel markers in academic research and can potentially improve 3D histopathological diagnostics of entire human tissue specimens, leading to improved treatment and superior patient outcomes.


Introduction
The detailed visualization of tissue samples is the basis to understanding, detecting, and characterizing complex pathologies, and represents an indispensable step in modern disease diagnosis. Classical histopathology, which is the current gold standard routinely used in pathology laboratories, focuses on the analysis of representative two-dimensional (2D) physical sections of samples. Besides the intrinsic limitations of the physical sectioning process, such as tissue loss or artifacts including tissue distortion, the classical 2D histopathology process is severely limited in accuracy, integrity, and spatial information,

Generation and Selection of hESAM-Specific Nanobodies
For the generation of high-affinity Nbs targeting the human ESAM protein (hESAM), two llamas (Lama glama) were subjected to a 42-day immunization protocol using a recombinant hESAM-Fc protein (Supplementary Figure S1A). The hESAM-Fc immunogen consisted of the extracellular domain of the hESAM protein (ECD, amino acids 1-247) fused to the Fc domain of the human IgG1 protein (hIgG1-Fc). Immunization resulted in two Nb phagemid libraries, each containing ≈10 8 independent clones, which were used to perform three consecutive rounds of phage display against the recombinant hESAM-Fc antigen. After enrichment of antigen-specific phages, 285 single clones were randomly selected and analyzed by ELISA screening for the presence of hESAM-specific Nbs in their periplasmic extracts. By comparing binding efficiency to wells coated with the hESAM-Fc immunogen, the hIgG1-Fc domain alone, or empty control wells, the ELISA data identified 141 positive binders for hESAM. Subsequent sequence analysis displayed 33 unique Nb candidates, which represented 6 different B-cell lineages, according to their complementary determining region (CDR) 3. Based on the ELISA as well as the B-cell lineage data, we selected one representative Nb of each lineage for further analysis (Supplementary Figure  S1B,C). However, as two candidates of the same B-cell lineage displayed robust ELISA screening data, both of these clones were included in our study.
In summary, we constructed two hESAM-specific Nb phagemid libraries, which were subjected to antigen-specific selection using phage display and ELISA screening. Out of 33 unique Nbs identified, we selected 7 Nb clones for production and assessment of 3D-histopathological imaging eligibility (Figure 1). immunogen consisted of the extracellular domain of the hESAM protein (ECD, amino acids 1-247) fused to the Fc domain of the human IgG1 protein (hIgG1-Fc). Immunization resulted in two Nb phagemid libraries, each containing ≈10 8 independent clones, which were used to perform three consecutive rounds of phage display against the recombinant hESAM-Fc antigen. After enrichment of antigen-specific phages, 285 single clones were randomly selected and analyzed by ELISA screening for the presence of hESAM-specific Nbs in their periplasmic extracts. By comparing binding efficiency to wells coated with the hESAM-Fc immunogen, the hIgG1-Fc domain alone, or empty control wells, the ELISA data identified 141 positive binders for hESAM. Subsequent sequence analysis displayed 33 unique Nb candidates, which represented 6 different B-cell lineages, according to their complementary determining region (CDR) 3. Based on the ELISA as well as the B-cell lineage data, we selected one representative Nb of each lineage for further analysis (Supplementary Figure S1B,C). However, as two candidates of the same B-cell lineage displayed robust ELISA screening data, both of these clones were included in our study. In summary, we constructed two hESAM-specific Nb phagemid libraries, which were subjected to antigen-specific selection using phage display and ELISA screening. Out of 33 unique Nbs identified, we selected 7 Nb clones for production and assessment of 3D-histopathological imaging eligibility ( Figure 1).

Figure 1.
Generation and identification of hESAM-specific Nbs. (A) Generation of hESAM-specific Nb phagemid library by llama immunization with a recombinant hESAM-Fc antigen, followed by collection of llama peripheral blood lymphocyte total RNA and subsequent VHH gene amplification. Resulting VHH gene products were integrated into pMECS phagemid vectors to obtain a VHH gene plasmid library. (B) For selection of hESAM-specific Nbs, the phagemid library was subjected to three consecutive rounds of biopanning on plates coated with the hESAM-Fc immunogen. After each round, bound phages were eluted and amplified in E. coli cells. (C) Binding specificity of enriched candidates was assessed by ELISA screening using bacterial extracts of 285 randomly selected clones. Out of 141 positive binders, sequence analysis identified 33 unique Nbs, and 7 Nb clones were selected for further analysis. Resulting V H H gene products were integrated into pMECS phagemid vectors to obtain a V H H gene plasmid library. (B) For selection of hESAM-specific Nbs, the phagemid library was subjected to three consecutive rounds of biopanning on plates coated with the hESAM-Fc immunogen. After each round, bound phages were eluted and amplified in E. coli cells. (C) Binding specificity of enriched candidates was assessed by ELISA screening using bacterial extracts of 285 randomly selected clones. Out of 141 positive binders, sequence analysis identified 33 unique Nbs, and 7 Nb clones were selected for further analysis.

Expression and Purification of hESAM-Specific Nanobodies
Prior to high-yield production, selected Nb sequences were recloned from the phagemid vector used for phage display (pMECS [22]) into an IPTG-inducible expression vector, which was specifically designed to produce Nbs with a C-terminal 6xHis protein tag [23]. Following overnight production in E. coli bacteria, Nbs were isolated in high purity, using the 6xHis protein tag in immobilized metal ion affinity chromatography (IMAC), followed by dialysis. For quality control, the purity of Nbs was evaluated by Coomassie Blue-stained SDS-PAGE, which confirmed the presence of a single protein band at the expected molecular weight for the respective Nb clones (Figure 2A). Furthermore, western blot analysis verified the identity of the Nbs by specifically detecting the 6xHis protein tag within the single protein band ( Figure 2B).
(IMAC), followed by dialysis. For quality control, the purity of Nbs was evaluated by Coomassie Blue-stained SDS-PAGE, which confirmed the presence of a single protein band at the expected molecular weight for the respective Nb clones (Figure 2A). Furthermore, western blot analysis verified the identity of the Nbs by specifically detecting the 6xHis protein tag within the single protein band ( Figure 2B).
Collectively, we performed high-yield production of the selected Nb clones and specific 6xHis-mediated protein purification in high purity, as demonstrated by Coomassie Blue-stained SDS-PAGE and western blot analysis. There were no unspecific bands detected by either of the methods, indicating that the Nbs were efficiently purified and thus could be tested as suitable imaging reagents.

Establishing 2D Nanobody Stainings in Human Skin Samples
Before assessing whether the Nb candidates are suitable for high-quality 3D imaging approaches, the staining quality of the selected Nbs was first evaluated in standard 2D immunofluorescence stainings using 5 µm thick cryosections of human skin specimens. In order to compare staining properties, sections were not only stained with every Nb clone, but, as means of positive control, also with commercially available antibodies against hESAM and the blood vessel surface marker CD31. As expected, both control stainings visualized small intradermal blood vessels, with CD31 labeling blood endothelial cells only and the hESAM antibody additionally showing unspecific labeling of epidermal layers. Notably, stainings with the Nbs revealed specific binding to CD31positive blood vessels for all seven clones, albeit with highly varying quality in terms of unspecific background signal. While all seven Nb clones detected distinct blood vascular structures in comparable quality to the ESAM antibody, stainings with three clones only featured weak blood vessel staining and substantial unspecific background signals, which resulted in the exclusion of these clones from further experiments (Supplementary Figure  S2). However, stainings with the other four Nb clones presented visualization of blood vascular structures with similar unspecific background signals, as concomitantly detected by the hESAM antibody ( Figure 3). Importantly, Nb clone '2ES41' showed the most Collectively, we performed high-yield production of the selected Nb clones and specific 6xHis-mediated protein purification in high purity, as demonstrated by Coomassie Bluestained SDS-PAGE and western blot analysis. There were no unspecific bands detected by either of the methods, indicating that the Nbs were efficiently purified and thus could be tested as suitable imaging reagents.

Establishing 2D Nanobody Stainings in Human Skin Samples
Before assessing whether the Nb candidates are suitable for high-quality 3D imaging approaches, the staining quality of the selected Nbs was first evaluated in standard 2D immunofluorescence stainings using 5 µm thick cryosections of human skin specimens. In order to compare staining properties, sections were not only stained with every Nb clone, but, as means of positive control, also with commercially available antibodies against hESAM and the blood vessel surface marker CD31. As expected, both control stainings visualized small intradermal blood vessels, with CD31 labeling blood endothelial cells only and the hESAM antibody additionally showing unspecific labeling of epidermal layers. Notably, stainings with the Nbs revealed specific binding to CD31-positive blood vessels for all seven clones, albeit with highly varying quality in terms of unspecific background signal. While all seven Nb clones detected distinct blood vascular structures in comparable quality to the ESAM antibody, stainings with three clones only featured weak blood vessel staining and substantial unspecific background signals, which resulted in the exclusion of these clones from further experiments (Supplementary Figure S2). However, stainings with the other four Nb clones presented visualization of blood vascular structures with similar unspecific background signals, as concomitantly detected by the hESAM antibody ( Figure 3). Importantly, Nb clone '2ES41' showed the most promising staining performance, as it unveiled the same structures as the ESAM antibody control, while showing significantly less unspecific background signals. promising staining performance, as it unveiled the same structures as the ESAM antibody control, while showing significantly less unspecific background signals. The generated Nb clones were raised against the hESAM ECD to visualize blood vascular structures in human samples; however, since the murine ESAM (mESAM) ECD showed a 67% amino acid sequence homology to hESAM ECD (Supplementary Figure  S3A), we further assessed whether the Nb clones were also suitable to visualize the blood vasculature in murine specimens. To test for cross-reactivity in murine samples, we double-stained cryosections of 14.5 day-old wild type mouse embryos (E14.5) with Nb clones and a commercially available CD31 antibody. Congruent to the stainings in human skin cryosections, Nb clone '2ES41' achieved the best staining results, detecting CD31positive endothelial cells and visualizing the morphology of blood vascular structures (Supplementary Figure S3B). Yet, other Nbs showed strong unspecific background signal and were determined to be unsuitable for high-quality microscopic imaging in mice (data not shown).
Collectively, standard 2D immunofluorescence stainings demonstrated specific visualization of blood vascular structures for all selected Nb clones in human skin cryosections. Yet, only four clones exhibited comparably low unspecific background signals, whereas three clones were excluded from further experimentations. Notably, one clone also achieved robust stainings of CD31-positive blood endothelial cells in cryosections of 14.5 day-old wild type mouse embryos.

Directly-Labeled Nanobodies Improve 2D Staining Quality and Incubation Time
Standard 2D immunofluorescence stainings of the selected Nbs validated their capability to specifically stain blood endothelial cell target structures and demonstrated comparable staining qualities as the commercially available hESAM control antibody. Yet, the entire staining procedure was not suitable for high-throughput imaging analysis, since the staining procedure consisted of three time-consuming staining steps: incubation with the respective Nb clone, followed by incubation with a secondary antibody binding the The generated Nb clones were raised against the hESAM ECD to visualize blood vascular structures in human samples; however, since the murine ESAM (mESAM) ECD showed a 67% amino acid sequence homology to hESAM ECD (Supplementary Figure S3A), we further assessed whether the Nb clones were also suitable to visualize the blood vasculature in murine specimens. To test for cross-reactivity in murine samples, we double-stained cryosections of 14.5 day-old wild type mouse embryos (E14.5) with Nb clones and a commercially available CD31 antibody. Congruent to the stainings in human skin cryosections, Nb clone '2ES41' achieved the best staining results, detecting CD31-positive endothelial cells and visualizing the morphology of blood vascular structures (Supplementary Figure  S3B). Yet, other Nbs showed strong unspecific background signal and were determined to be unsuitable for high-quality microscopic imaging in mice (data not shown).
Collectively, standard 2D immunofluorescence stainings demonstrated specific visualization of blood vascular structures for all selected Nb clones in human skin cryosections. Yet, only four clones exhibited comparably low unspecific background signals, whereas three clones were excluded from further experimentations. Notably, one clone also achieved robust stainings of CD31-positive blood endothelial cells in cryosections of 14.5 day-old wild type mouse embryos.

Directly-Labeled Nanobodies Improve 2D Staining Quality and Incubation Time
Standard 2D immunofluorescence stainings of the selected Nbs validated their capability to specifically stain blood endothelial cell target structures and demonstrated comparable staining qualities as the commercially available hESAM control antibody. Yet, the entire staining procedure was not suitable for high-throughput imaging analysis, since the staining procedure consisted of three time-consuming staining steps: incubation with the respective Nb clone, followed by incubation with a secondary antibody binding the Nb 6xHis protein tag, and finally, incubation with a fluorescently-labeled antibody visualizing the secondary antibody. Consequently, in order to reduce incubation times, we utilized NHS-Ester chemistry to directly label the four selected Nb clones with a fluorophore, instead of using secondary and tertiary antibodies. In addition to potentially improving staining qualities, costly resources as well as multiple days of incubation periods can be evaded by avoiding secondary and tertiary staining steps.
The staining quality of directly-labeled Nbs was evaluated on 5 µm cryosections of human skin specimens, which were additionally stained with CD34, a commonly used marker for endothelial as well as hematopoetic cells [24]. As previously observed in the first 2D immunofluorescence stainings using uncoupled Nbs, the selected Nb clones, again, presented specific identification of blood vascular target structures, corroborating the preceding assessments of their eligibility for histopathological immunostainings. Of note, the strong labeling of epidermal layers as well as unspecific background signals were significantly reduced in comparison to the first standard 2D immunofluorescence stainings for two Nb clones (Figure 4). However, direct-labeling of the other two clones resulted in increased unspecific background signals, most probably due to provoked conformational changes in the Nb antigen-binding domains by fluorophore linkage (data not shown).
visualizing the secondary antibody. Consequently, in order to reduce incubation times, we utilized NHS-Ester chemistry to directly label the four selected Nb clones with a fluorophore, instead of using secondary and tertiary antibodies. In addition to potentially improving staining qualities, costly resources as well as multiple days of incubation periods can be evaded by avoiding secondary and tertiary staining steps.
The staining quality of directly-labeled Nbs was evaluated on 5 µm cryosections of human skin specimens, which were additionally stained with CD34, a commonly used marker for endothelial as well as hematopoetic cells [24]. As previously observed in the first 2D immunofluorescence stainings using uncoupled Nbs, the selected Nb clones, again, presented specific identification of blood vascular target structures, corroborating the preceding assessments of their eligibility for histopathological immunostainings. Of note, the strong labeling of epidermal layers as well as unspecific background signals were significantly reduced in comparison to the first standard 2D immunofluorescence stainings for two Nb clones (Figure 4). However, direct-labeling of the other two clones resulted in increased unspecific background signals, most probably due to provoked conformational changes in the Nb antigen-binding domains by fluorophore linkage (data not shown).
Taken together, the directly-labeled Nbs manifested superior staining properties, as they not only specifically visualized blood vessel target structures and produced less unspecific background signal, but also significantly reduced incubation times and were, therefore, affirmed to be suitable for rapid 3D histopathological stainings.

3-Dimensional ESAM Staining of Human Skin Shows Detailed Vascular Network
Most antibodies currently used in academic and diagnostic imaging approaches are not suitable for high-quality 3D imaging technologies, since they either do not fully penetrate larger tissue samples or do not properly bind to their target structures in the harsh chemical conditions during 3D imaging sample preparation. For this reason, and owing to the fact that the commercially available hESAM antibody failed to produce highquality 3D imaging results (data not shown), we were especially interested to validate the 3D imaging eligibility of our Nb clones in a clinically-relevant 3D imaging setting. To this end, we utilized the directly-labeled Nbs of the two selected clones in whole-mount stainings of 5 × 5 mm human skin punch biopsies and assessed if the respective Nbs were Taken together, the directly-labeled Nbs manifested superior staining properties, as they not only specifically visualized blood vessel target structures and produced less unspecific background signal, but also significantly reduced incubation times and were, therefore, affirmed to be suitable for rapid 3D histopathological stainings.

3-Dimensional ESAM Staining of Human Skin Shows Detailed Vascular Network
Most antibodies currently used in academic and diagnostic imaging approaches are not suitable for high-quality 3D imaging technologies, since they either do not fully penetrate larger tissue samples or do not properly bind to their target structures in the harsh chemical conditions during 3D imaging sample preparation. For this reason, and owing to the fact that the commercially available hESAM antibody failed to produce high-quality 3D imaging results (data not shown), we were especially interested to validate the 3D imaging eligibility of our Nb clones in a clinically-relevant 3D imaging setting. To this end, we utilized the directly-labeled Nbs of the two selected clones in whole-mount stainings of 5 × 5 mm human skin punch biopsies and assessed if the respective Nbs were able to fully penetrate the voluminous tissue samples and specifically visualize blood vascular structures.
Indeed, biopsy specimens stained with the directly-labeled Nbs showed precise visualization of intricate blood vascular structures within the entire tissue sample and considerably little unspecific background signal, demonstrating specific target recognition and complete Nb penetration of the voluminous samples (Supplementary Video S1). Furthermore, the delicate stainings not only allowed macroscopic representation of the entire blood vascular network, but also microscopic analysis of morphological details, such as vessel size and volume, as well as vascular branching points ( Figure 5). Therefore, we confirmed that the hESAM Nb clones '2ES41' and '2ES176' are highly suitable for whole-mount 3D immunostainings of blood vascular networks, which can potentially improve academic and diagnostic histopathological imaging approaches. vascular structures.
Indeed, biopsy specimens stained with the directly-labeled Nbs showed precise visualization of intricate blood vascular structures within the entire tissue sample and considerably little unspecific background signal, demonstrating specific target recognition and complete Nb penetration of the voluminous samples (Supplementary Video S1). Furthermore, the delicate stainings not only allowed macroscopic representation of the entire blood vascular network, but also microscopic analysis of morphological details, such as vessel size and volume, as well as vascular branching points ( Figure 5). Therefore, we confirmed that the hESAM Nb clones '2ES41' and '2ES176' are highly suitable for whole-mount 3D immunostainings of blood vascular networks, which can potentially improve academic and diagnostic histopathological imaging approaches.

Discussion
Thorough understanding of complex pathologies and spatial biological structures necessitates high-quality visualization of histological samples in their 3D morphological context. However, the current gold standard in histopathology still mainly relies on stainings of representative 2D physical sections due to technical limitations for 3D visualization of entire specimens, as well as the absence of suitable labeling reagents for whole-mount sample stainings. Aggravated by the usage of IgG antibodies, which are oftentimes not suitable for 3D imaging due to their relatively large size and consequently limited tissue penetration, the current 2D histopathological approach does not provide sufficient spatial information for appropriate diagnosis.
In this study, we therefore aimed to generate and experimentally validate a novel single-domain antibody (nanobody, Nb) directed against the human endothelial cellselective adhesion molecule (hESAM), which can be used to specifically label blood endothelial cells in whole-mount 3D imaging approaches. To this end, we immunized llamas with a recombinant hESAM-Fc protein, composed of the hESAM extracellular domain (ECD) linked to the Fc domain of human IgG1, which ultimately resulted in the identification of 33 unique Nb clones. Although limited in sequence variety, as represented by only six different sequences for the complementary determining region (CDR) 3, the identified Nb clones demonstrated robust antigen-specific binding affinities in ELISA screenings. Low diversity of detected CDR3 sequences as well as the general

Discussion
Thorough understanding of complex pathologies and spatial biological structures necessitates high-quality visualization of histological samples in their 3D morphological context. However, the current gold standard in histopathology still mainly relies on stainings of representative 2D physical sections due to technical limitations for 3D visualization of entire specimens, as well as the absence of suitable labeling reagents for whole-mount sample stainings. Aggravated by the usage of IgG antibodies, which are oftentimes not suitable for 3D imaging due to their relatively large size and consequently limited tissue penetration, the current 2D histopathological approach does not provide sufficient spatial information for appropriate diagnosis.
In this study, we therefore aimed to generate and experimentally validate a novel single-domain antibody (nanobody, Nb) directed against the human endothelial cellselective adhesion molecule (hESAM), which can be used to specifically label blood endothelial cells in whole-mount 3D imaging approaches. To this end, we immunized llamas with a recombinant hESAM-Fc protein, composed of the hESAM extracellular domain (ECD) linked to the Fc domain of human IgG1, which ultimately resulted in the identification of 33 unique Nb clones. Although limited in sequence variety, as represented by only six different sequences for the complementary determining region (CDR) 3, the identified Nb clones demonstrated robust antigen-specific binding affinities in ELISA screenings. Low diversity of detected CDR3 sequences as well as the general moderate number of unique Nb clones may be caused by high glycosylation of the hESAM ECD [25,26], which might sterically hinder potential epitope recognition [27].
Human skin 2D stainings of selected Nbs reflected heterogenous results, as some Nb clones depicted strong unspecific background signals and only faint antigen-specific labeling. This finding was not unexpected, as antigen-specific ELISA screenings do not reflect the vast plethora of potential biological antigens, which may additionally be bound by the Nbs, and ultimately underlines the importance of experimental validation of every single Nb clone. Of interest, a single hESAM-Nb also exhibited moderate staining performance in 2D immunofluorescence stainings of murine embryonic tissue, highlighting the broad variety in target recognition among Nb clones with otherwise close sequence homology. Unexpectedly, however, Nb clones with sufficient staining quality in human samples also exhibited additional unspecific labeling of the epidermis, which, although also shown by the commercially available hESAM antibody control, prevents high-quality imaging. This phenomenon was bypassed through direct-labeling of selected Nbs with fluorescent dyes using NHS ester chemistry, additionally reducing incubation times and unspecific background signals. Yet, NHS ester-mediated crosslinking of fluorophores and lysine amino acids may elicit conformational changes within the inherent Nb structure, possibly resulting in altered binding affinities of specific as well as unspecific targets. Notably, we also observed changes in Nb binding specificity, since undesirable binding events were strongly enhanced in some Nb clones. Consequently, due to its unpredictable nature, NHS ester chemistry was not applicable to every Nb, so other labeling strategies may be more favorable, such as sortase-or maleimide-mediated protein labeling approaches [28][29][30]. Nonetheless, for other clones, imaging quality was considerably improved, which was demonstrated in high-quality 2D and whole-mount 3D immunofluorescence stainings. In 3D stainings of human skin biopsies, the hESAM-Nbs allowed unprecedented visualization of ESAM-positive target structures, displaying the delicate blood vascular network of the entire tissue sample.
In conclusion, in this study, we generated and experimentally validated hESAM-Nbs, which not only featured similar 2D staining properties to a commercially available IgG antibody, but also unparalleled 3D imaging quality. Previously impossible by using commercial ESAM antibodies, for the first time, the hESAM-Nbs enabled unrestricted and detailed spatial representations of ESAM-expressing blood endothelial cells in 3D imaging approaches. Owing to the animal-free, high-yield, and low-cost production, the hESAM-Nbs represent a novel and cost-efficient imaging tool for 3D histopathological analysis in academic research and potentially clinical diagnostics.

Llama Immunization and Construction of Nanobody Library
For the generation of Nbs targeting the hESAM protein, llamas (Lama glama) were immunized with the extracellular part of the hESAM protein (amino acids 1-247) fused to the Fc domain of human IgG1 (hESAM-Fc, Supplementary Figure S1A). The recombinant hESAM-Fc protein was expressed in CHO cells and purified from the CHO cell culture supernatant using Protein G Sepharose ® columns (ab193260, Abcam, Cambridge, UK), followed by dialysis against PBS. The immunoadjuvant used during immunizations was Adjuvant P (3111, GERBU Biotechnik GmbH, Heidelberg, Germany). In detail, two llamas were subcutaneously injected with 100 µg carrier-free, recombinant hESAM-Fc protein on days 0, 10, 16, 23, 30, and 37, followed by collection of 100 mL anti-coagulated blood on day 42 for isolation of peripheral blood lymphocytes (PBLs). For each llama, separate VHH gene libraries were constructed by using total RNA from PBLs as a template for first strand cDNA synthesis with oligo(dT) primers. Using the PBL cDNA, VHH encoding sequences were amplified via a multi-step PCR, digested with SapI (R0569, New England Biolabs, Ipswich, MA, USA), and cloned into the SapI restriction site of the pMECS phagemid vector (Prof. Serge Muyldermans, Laboratory of Cellular and Molecular Immunology, Vrije Universiteit, Brussel, Belgium). After ligation, electrocompetent E. coli TG1 cells (60502, Lucigen, Middleton, WI, USA) were transformed with the recombinant phagemid vectors, resulting in two VHH libraries of about 10 8 independent transformants each. A detailed description of the workflow has been published elsewhere [22].

Biopanning and Screening for hESAM-Specific Nanobodies
Phage enrichment and biopanning was performed as previously described [22], with the following specifications: For the selection of hESAM-specific Nbs, three consecutive panning rounds were performed on 96-well Microtiter™ microplates (2205, Thermo Fisher Scientific, Waltham, MA, USA) coated with different concentrations of the recombinant hESAM-Fc protein (1st round: 100 µg/mL, 2nd + 3rd round: 50 µg/mL). The coating buffer was 100 mM NaHCO 3 , with pH 8.2. To reduce unspecific human IgG1-Fc binding events, the solution was supplemented with 1 µM recombinant human IgG Fc protein (110-HG, R&D Systems, Minneapolis, MN, USA). Enrichment of antigen-specific phages was about 10 2 -fold and 2 × 10 3 -fold after the 2nd and 3rd round, respectively. After each round of biopanning, E. coli TG1 cells were transfected with the phage output and used to amplify phages for the next round of biopanning and/or subsequent ELISA analysis. After enrichment, 285 clones were randomly selected and subjected to ELISA analysis. In detail, crude periplasmic extracts of bacterial clones were incubated on 96-well Microtiter™ microplates coated with 1 µg/mL hESAM immunogen in blocking buffer ('hESAM'), 2 µg/mL hIgG1-Fc in blocking buffer ('Fc'), or blocking buffer only ('control'). The blocking buffer was 100 mM NaHCO 3 , with pH 8.2. ELISA measurements were performed as described elsewhere [22].

Nanobody Expression and Purification
For production of Nbs, WK6 E. coli harboring the pHEN6c-Nb plasmid of interest were grown in 1 L of 'Terrific Broth' medium (2.3 g/L KH 2 PO 4 , 16.4 g/L K 2 HPO 4 -3H 2 O, 12 g/L tryptone, 24 g/L yeast extract, 0.4% (v/v) glycerol) supplemented with 100 µg/mL ampicillin, 2 mM MgCl 2 , and 0.1% (w/v) glucose. Bacterial cultures were incubated at 37 • C with constant shaking until an OD 600 of 0.6-0.9 was reached. Then, Nb expression was induced by adding isopropyl ß-D-1-thiogalactopyranoside (IPTG) to a final concentration of 1 mM, and cultures were incubated at 28 • C with constant shaking for 16 h. For extraction of Nbs, bacterial cells were centrifuged (8000× g, 8 min, RT), resuspended in 12 mL TES buffer (0.2 M Tris [pH 8.0], 0.5 mM EDTA, 0.5 M sucrose), and shaken for 1 h on ice. Subsequently, 18 mL TES/4 buffer (0.05 M Tris [pH 8.0], 0.125 mM EDTA, 0.125 M sucrose) was added to the cell solution, followed by shaking for 1 h on ice and centrifugation (8000× g, 30 min, 4 • C). The supernatants containing the periplasmic proteins were collected and 6xHistagged Nbs were purified using HIS-Select ® nickel affinity gel (P6611, Sigma-Aldrich, Darmstadt, Germany) according to the manufacturer's instructions. Briefly, 2 mL HIS-Select ® solution was equilibrated in 48 mL of phosphate buffered saline (PBS, pH 7.4), mixed and centrifuged (2000× g, 2 min), followed by removal of the supernatant. The equilibration procedure was repeated for two additional times, after which bacterial periplasmic extracts were added to the equilibrated HIS-Select ® resin and incubated for 1 h at RT with gentle