Single-Domain Antibodies for Targeting, Detection, and In Vivo Imaging of Human CD4+ Cells

The advancement of new immunotherapies necessitates appropriate probes to monitor the presence and distribution of distinct immune cell populations. Considering the key role of CD4+ cells in regulating immunological processes, we generated novel single-domain antibodies [nanobodies (Nbs)] that specifically recognize human CD4. After in-depth analysis of their binding properties, recognized epitopes, and effects on T-cell proliferation, activation, and cytokine release, we selected CD4-specific Nbs that did not interfere with crucial T-cell processes in vitro and converted them into immune tracers for noninvasive molecular imaging. By optical imaging, we demonstrated the ability of a high-affinity CD4-Nb to specifically visualize CD4+ cells in vivo using a xenograft model. Furthermore, quantitative high-resolution immune positron emission tomography (immunoPET)/MR of a human CD4 knock-in mouse model showed rapid accumulation of 64Cu-radiolabeled CD4-Nb1 in CD4+ T cell-rich tissues. We propose that the CD4-Nbs presented here could serve as versatile probes for stratifying patients and monitoring individual immune responses during personalized immunotherapy in both cancer and inflammatory diseases.


INTRODUCTION
In precision medicine, diagnostic classification of the disease-associated immune status should guide the selection of appropriate therapies. A comprehensive analysis of a patient's specific immune cell composition, activation state, and infiltration of affected tissue has been shown to be highly informative for patient stratification (1)(2)(3). In this context, CD4 is an important marker, as it is found on the surface of immune cells such as monocytes, macrophages, and dendritic cells and most abundant on CD4 + T cells (4,5). CD4 + T cells are a key determinant of the immune status due to their essential role in orchestrating immune responses in autoimmune diseases, immune-mediated inflammatory diseases (IMIDs), cancer, and chronic viral infections (6)(7)(8)(9)(10)(11)(12)(13). Current diagnostic standards such as intra-cytoplasmic flow cytometry analysis (IC-FACS), immunohistochemistry, and ex vivo cytokine assays or RT-PCR analysis are exclusively invasive and therefore limited with respect to repetitive analyses over time (14)(15)(16)(17). Considering the emerging role of infiltrating lymphocytes and the impact of CD4 + T cells on the outcome of immunotherapies, novel approaches are needed to assess CD4 + T cells more holistically (18). In this context, noninvasive imaging approaches offer a significant benefit compared to the current diagnostic standard. To date, radiolabeled antibodies have been applied to image CD4 + cells in preclinical models (10,(19)(20)(21). Due to the recycling effect mediated by the neonatal Fc receptor, full-length antibodies have a long serum half-life, which requires long clearance times of several days before high-contrast images can be acquired (22). Additionally, effector function via the Fc region was shown to induce depletion or functional changes in CD4 + cells including the induction of proliferation or cytokine release (23)(24)(25). Notably, also higher dosages of recombinant antibody fragments like Fab fragments or Cys-diabodies derived from the monoclonal anti-CD4 antibody GK1.5 were recently shown to decrease CD4 expression in vivo and inhibit proliferation and interferon (IFN)-g production in vitro (24)(25)(26). These studies highlight the importance of carefully investigating CD4 + cell-specific immunoprobes for their epitopes, binding properties, and functional effects.
During the last decade, antibody fragments derived from heavy-chain-only antibodies of camelids, referred to as VHHs or nanobodies (Nbs) (27), have emerged as versatile probes for molecular imaging [reviewed in (28)]. In combination with highly sensitive and/or quantitative whole-body molecular imaging techniques such as optical or radionuclide-based modalities, particularly positron emission tomography (PET), Nbs have been shown to bind their targets within several minutes of systemic application (29). Due to their great potential as highly specific imaging probes, numerous Nbs targeting immune-or tumor-specific cellular antigens are currently in preclinical development and even in clinical trials (28,30,31).
Here, we generated a set of human CD4 (hCD4)-specific Nbs. Following in-depth characterization of their binding properties, we selected candidates that did not affect T-cell proliferation, activation, or cytokine release and converted them into immune tracers for noninvasive optical and PET imaging. Using a mouse xenograft model and an hCD4 knock-in mouse model, we successfully demonstrated the capacity of these CD4-Nbs to visualize CD4 + cells in vivo.

Generation of High-Affinity CD4 Nanobodies
To generate Nbs directed against hCD4, we immunized an alpaca (Vicugna pacos) with the recombinant extracellular portion of hCD4 following an 87-day immunization protocol. Subsequently, we generated a Nb phagemid library comprising~4 × 10 7 clones that represent the full repertoire of variable heavy chains of heavychain antibodies (VHHs or Nbs) of the animal. We performed phage display using either passively adsorbed purified hCD4 or CHO and HEK293 cells stably expressing full-length hCD4 (CHO-hCD4 and HEK293-hCD4 cell lines, respectively). Following two cycles of phage display for each condition, we analyzed a total of 612 individual clones by whole-cell phage ELISA and identified 78 positive binders. Sequence analysis revealed 13 unique Nbs representing five different B-cell lineages according to their complementarity determining region (CDR) 3 ( Figure 1A). One representative Nb of each lineage, termed CD4-Nb1-CD4-Nb5, was expressed in bacteria (Escherichia coli) and isolated with high purity using immobilized metal ion affinity chromatography (IMAC) followed by size exclusion chromatography (SEC) ( Figure 1B). To test whether selected Nbs are capable of binding to full-length hCD4 localized at the plasma membrane of mammalian cells, we performed live-cell staining of CHO-hCD4 cells ( Figure 1C, Supplementary Figure 1). Executed at 4°C, images showed a prominent staining of the plasma membrane, whereas at 37°C, the fluorescent signal was mainly localized throughout the cell body, presumably a consequence of endocytotic uptake of receptor-bound Nbs. CHO wild-type (wt) cells were not stained by any of the five CD4-Nbs at both temperatures (data not shown). CD4-Nb1 and CD4-Nb3, both identified by whole-cell panning, displayed strong staining of CHO-hCD4 cells. Of the Nbs derived from panning with recombinant hCD4, CD4-Nb2 also showed strong cellular staining, whereas staining with CD4-Nb4 revealed weak signals. CD4-Nb5 showed no staining under these conditions and was consequently excluded from further analyses ( Figure 1C). To quantitatively assess binding affinities, we performed biolayer interferometry (BLI), measuring serial dilutions of Nbs on the biotinylated extracellular domain of hCD4 immobilized at the sensor tip. For CD4-Nb1 and CD4-Nb2, K D values were determined to be~5 and~7 nM, respectively, while CD4-Nb3 and CD4-Nb4 displayed lower affinities of 75 and 135 nM, respectively ( Figure 1D, Table 1, Supplementary Figure 2A). In addition, we determined corresponding EC 50 values with full-length plasma membranelocated hCD4 on HEK293-hCD4 cells by flow cytometry. In accordance with cellular staining and biochemically determined affinities, these values revealed a strong functional binding for CD4-Nb1 and CD4-Nb2 with EC 50 values in the subnanomolar range (~0.7 nM), whereas CD4-Nb3 and CD4-Nb4 displayed substantially lower cellular affinities ( Figure 1E, Table 1, Supplementary Figure 2B). In summary, we generated four CD4-Nbs that bind isolated and cell-resident hCD4. While CD4-Nb3 and CD4-Nb4 appeared less affine, CD4-Nb1 and CD4-Nb2 displayed high affinities in the low nanomolar range.

Domain Mapping
Next, we applied chemo-enzymatic coupling using sortase A for site-directed functionalization of CD4-Nbs (32,33). We thereby linked peptides conjugated to a single fluorophore to the Cterminus of CD4-Nbs, yielding a defined labeling ratio of 1:1 (34). Live-cell immunofluorescence imaging showed that all sortase-coupled CD4-Nbs retained their capability of binding to cell-resident hCD4 of CHO-hCD4 cells (Supplementary Figure 3A). To localize the binding sites of the selected CD4-Nbs, we generated domain-deletion mutants of hCD4. Expression and correct surface localization of these mutants in CHO cells were confirmed by staining with antibody RPA-T4 binding to domain 1 of CD4. For mutants lacking domain 1, we introduced an N-terminal BC2 tag (35) to allow for live-cell surface detection with a fluorescently labeled bivBC2-Nb (34) (Supplementary Figure 3B). Transiently expressed domaindeletion mutants were then tested for binding of CF568labeled CD4-Nbs by live-cell immunofluorescence imaging, including a non-specific fluorescently labeled green fluorescent protein (GFP)-binding Nb (GFP-Nb) as negative control. Based on these results, we allocated binding of CD4-Nb1 and CD4-Nb3 to domain 1, whereas CD4-Nb2 and CD4-Nb4 bind to domain 3 and/or 4 of hCD4 ( Figure 2A, Supplementary Figure 3C).
To further examine combinatorial binding of the different CD4-Nbs, we performed an epitope binning analysis by BLI. Recombinant full-length hCD4 was immobilized at the sensor tip, and combinations of CD4-Nbs were allowed to bind consecutively (Supplementary Figure 4). Unsurprisingly, CD4-Nbs binding to different domains displayed combinatorial binding. Interestingly, a simultaneous binding was also detected for the combination of CD4-Nb1 and CD4-Nb3, suggesting that both CD4-Nbs bind to different epitopes within domain 1. In contrast, we did not observe simultaneous binding for CD4-Nb2 and CD4-Nb4, which might be due to close-by or overlapping epitopes at domain 3/4 for the latter Nb pair.
For a more precise epitope analysis, we conducted a hydrogen-deuterium exchange mass spectrometry (HDX-MS) analysis of hCD4 bound to CD4-Nb1, CD4-Nb2, or CD4-Nb3 ( Figures 2B-E, Supplementary Figure 5). Due to its low affinity, CD4-Nb4 was not considered for HDX-MS analysis (data not shown). In accordance with our previous findings, binding of CD4-Nb1 and CD4-Nb3 protected sequences of domain 1 from HDX, whereas CD4-Nb2 protected sequences of domains 3 and 4 of hCD4 ( Figure 2B). The results obtained for binding of CD4-Nb1 ( Figure 2C) are similar to those obtained for CD4-Nb3 ( Figure 2D) in that binding of either Nb reduced hydrogen exchange at amino acid (aa) residues from aa T17 to N73, albeit with a different extent of protection at individual sequence segments. For CD4-Nb1, the greatest protection from HDX was observed for the sequence ranging from aa K35 to L44 corresponding to b strand C´and C´´of the immunoglobulin fold of domain 1 and residues aa K46-K75, comprising b strands D and E. In contrast, binding of CD4-Nb3 confers only a minor reduction in HDX within the latter sequence but additionally protects sequence aa C84-E91, which correspond to b strands G and F and their intermediate loop. For CD4-Nb2, we found protection of sequences aa W214-F229 (b strands C and C´) and aa K239-L259 (b strands C´´-E) and to a minor extent sequence aa R293-L296 as part of b strand A of domain 4 ( Figure 2E). In summary, our HDX-MS analysis revealed that all three tested Nbs bind three dimensional epitopes within different parts of hCD4. It further provides an explanation how CD4-Nb1 and CD4-Nb3 can bind simultaneously to domain 1 of hCD4 and confirms that the epitope of CD4-Nb2 is mainly located at domain 3.

Binding of CD4-Nbs to Human Peripheral Blood Mononuclear Cells
Having demonstrated that all selected Nbs bind to recombinant and exogenously overexpressed cellular hCD4, we next examined their capability and specificity of binding to physiologically relevant levels of CD4 + T cells within peripheral blood mononuclear cell (PBMC) samples. We costained human PBMCs from three donors with CD4-Nb1-CD4-Nb4 coupled to CF568 (100 nM for high-affine CD4-Nb1 and CD4-Nb2; 1,000 nM for low-affine CD4-Nb3 and CD4-Nb4) in combination with an anti-CD3 antibody and analyzed the percentage of doublepositive cells (CD3 + CD4 + ) by flow cytometry (Figure 3, Supplementary Figure 6). Compared to staining with an anti-CD4 antibody used as a positive control, all CD4-Nbs stained a similar percentage of CD4 + T cells for all tested donors, while the non-specific GFP-Nb yielded a negligible percentage of doublepositive cells even at the highest concentration (1,000 nM) ( Table 2). Our analysis further revealed that, as observed with a conventional anti-CD4 antibody, the CD4-Nbs stain a substantial proportion of CD3cells, indicating that all selected candidates are also able to recognize cells such as monocytes, macrophages, or dendritic cells that express lower levels of CD4 ( Figure 3).
Impact of CD4-Nbs on Activation, Proliferation, and Cytokine Release of CD4 + T and Immune Cells In view of the envisioned application as clinical imaging tracer, we next evaluated the potential of the Nbs to be further developed into clinically approved binding molecules. Since CD4-Nb2 and CD4-Nb3 contain a number of cysteine residues in their CDR3, we excluded them at this stage because such noncanonical unpaired cysteines are often associated with expression problems and a higher tendency to form aggregates in downstream production (37,38). With CD4-Nb1 and CD4-Nb4, we pursued two candidates that do not contain noncanonical cysteines and also cover a broad affinity spectrum. For these two Nbs and a non-specific GFP-Nb as a control, we then examined their influence on CD4 + T-cell activation, proliferation, and cytokine release. To rule out adverse effects of bacterial endotoxins in the Nb preparations, we first removed endotoxins by depletion chromatography, resulting in Food and Drug Administration (FDA)-acceptable endotoxin levels of <0.25 EU per mg. Typically, Nb-based radiotracers are applied at serum concentrations between 0.01 and 0.2 μM in (pre)clinical imaging (39,40). To investigate the effects of Nbs at the expected, but also at a 10-fold increased concentration and consequently elongated serum retention times that might occur during in vivo (pre)clinical imaging, we treated carboxyfluorescein succinimidyl ester (CFSE)labeled human PBMCs from three preselected healthy donors with three Nbs at concentrations ranging from 0.05 from 5 μM for 1 h at 37°C. Subsequently, cells were washed to remove Nbs and stimulated with an antigenic [cognate major histocompatibility complex (MHC)II peptides] or a non-antigenic stimulus (phytohemagglutinin, PHA-L) and analyzed after 4, 6, and 8 days by flow cytometry with the gating strategy shown in Supplementary Figure 7A. According to the highly similar CFSE intensity profiles observed, the total number of cell divisions was not affected by the different Nb treatments (exemplarily shown for one of three donors on day 6; Supplementary Figure 7A). For samples of the same donor and time point, no substantial differences in the percentage of proliferated cells were observed between mock incubation and individual Nb treatments. For both stimuli, the average percentage of proliferated cells increased over time in all donors tested, with no clear differences between conditions ( Figure 4A). As a quantitative measure of Tcell activation, we also determined the cell surface induction of a very early activation marker (CD69) and of the interleukin (IL)-2 receptor a chain (CD25) on CD4 + T cells ( Figure 4B). Among samples of the same donor and stimulation, we found highly similar activation profiles for all Nb treatments. While the percentage of CD4 + CD25 + cells steadily increased over time for MHCII peptide stimulation, for the PHA-stimulated condition, the percentage of positive cells was similarly high at all times of analysis. Importantly, regardless of the differences between donors, the individual Nb treatments from the same donor did not result in significant differences in the percentage of CD4 + CD25 + or CD4 + CD69 + cells for either stimulation at any point in the analysis.
Next, we analyzed cytokine expression of CD4 + T cells by intracellular cytokine staining after restimulation with cognate MHCII peptides. The corresponding gating strategy is shown in Supplementary Figure 7B. Samples of the same donor treated with different Nbs had highly similar percentages of cytokine [tumor necrosis factor (TNF), IFN-g, or IL-2] or activation marker (CD154)-positive cells without stimulation and upon stimulation with MHCII peptides ( Figure 4C). Overall, exposure to CD4-Nbs did not affect the proliferation, activation, or cytokine production of CD4 + T cells. In addition, we analyzed potential effects of CD4-Nbs on the release of cytokines from full-blood samples of three further donors. Upon stimulation with lipopolysaccharide (LPS) or PHA−L, we determined the serum concentrations with a panel of pro-and antiinflammatory cytokines (Supplementary Table 2). Although there was significant inter-donor variation for some cytokines, Nb treatment did not result in significant differences in either stimulated or unstimulated samples (Supplementary Figure 8).

CD4-Nbs for In Vivo Imaging
For optical in vivo imaging, we labeled CD4-Nbs with the fluorophore Cy5.5 (CD4-Nb-Cy5.5) by sortase-mediated attachment of an azide group followed by click-chemistry addition of dibenzocyclooctyne (DBCO)-Cy5.5. First, we tested potential cross-reactivity of the four Cy5.5-labeled CD4-Nbs to murine CD4 + lymphocytes. Notably, flow cytometric analysis showed that none of the selected CD4-Nbs bound murine CD4 + TABLE 2 | Percentage of double-positive cells of three donors, stained with CD4-Nb1 or CD4-Nb2 (100 nM), or CD4-Nb3 or CD4-Nb4 (1,000 nM), compared to anti-CD4 antibody and negative control Nb (GFP-Nb, 1,000 nM).  cells, suggesting exclusive binding to hCD4. Moreover, low-affine binding CD4-Nb4 bound neither mouse nor human CD4 + cells at the concentration used here (0.75 μg/ml,~49 nM) (Supplementary Figure 9). Consequently, we focused on CD4-Nb1 as the most promising candidate and CD4-Nb4 as a candidate with a high off-target rate, both of which we further analyzed for their in vivo target specificity and dynamic distribution using a murine xenograft model. To establish hCD4-expressing tumors, NOD SCID gamma (NSG) mice were inoculated subcutaneously with CD4 + T-cell leukemia HPB-ALL cells (41). After 2-3 weeks, mice bearing HPB-ALL xenografts were intravenously (i.v.) injected with 5 μg of CD4-Nb1-Cy5.5, CD4-Nb4-Cy5.5, or a control Nb (GFP-Nb-Cy5.5) and non-invasively in vivo investigated by optical imaging (OI) in intervals over the course of 24 h ( Figure 5A, Supplementary Figure 10A). The Cy5.5 signal intensity (SI) of the control Nb peaked within 10-20 minutes and rapidly declined thereafter to approximately the half and a quarter of maximum level at 2 and 24 h, respectively ( Figure 5B, Supplementary Figure 10B). While the SI of the low-affinity CD4-Nb4-Cy5.5 did not exceed the SI of the control Nb at any time (Supplementary Figure 10B), CD4-Nb1-Cy5.5 reached its maximum SI within the HPB-ALL xenograft of~1.8-fold above the control Nb at 30 min and slowly declined tõ 90% and~80% of maximum after 2 and 4 h, respectively ( Figure 5B). Based on the differences in the SI between CD4-Nb1-Cy5.5 and GFP-Nb-Cy5.5, we observed constant high target accumulation and specificity between 30 and 480 min post injection ( Figure 5B). After 24 h, mice were euthanized, and the presence of fluorophore-labeled CD4-Nbs within the explanted tumors was analyzed by OI ( Figure 5C, Supplementary Figure  10C). Compared to control, tumors from mice injected with CD4 −Nb1−Cy5.5 had~4-fold higher Cy5.5 SI, indicating a good signalto-background ratio for this Nb-derived fluorescently labeled immunoprobe even at later time points. To confirm CD4-specific targeting of CD4-Nb1 within the xenograft, we additionally performed ex vivo immunofluorescence of HPB-ALL tumors at 2 and 24 h post injection (Supplementary Figure S11). At the early time point, when the in vivo OI signal peaked, CD4-Nb1 was widely distributed throughout the whole tumor, whereas no Cy5.5 signal was detected in the GFP-Nb-injected mice (Supplementary Figures 11A, B). Semiquantitative analysis at the single-cell level revealed intense CD4-Nb1 binding at the surface of HBP-ALL cells that correlated with the CD4 antibody signal and internalization of CD4-Nb1 in some cells (Supplementary Figure 11C). In contrast, no binding was observed upon administration of unrelated GFP-Nb (Supplementary Figure 11D). At 24 h post injection, we observed regions of strongly internalized CD4-Nb1 (Supplementary Figures  11E, G), but also regions showing a low residual CD4-Nb1 uptake (Supplementary Figures 11E, H).
The OI data from the xenograft model clearly indicates that the high-affinity CD4-Nb1 but not CD4-Nb4 is suitable to specifically visualize CD4 + cells in vivo within a short period (30-120 min) after administration. Considering that this model does not reflect the natural distribution of CD4 + T cells in an organism, we continued with a model that allowed us to visualize the physiological composition of CD4 + immune cells. Thus, we employed a humanized CD4 murine knock-in model (hCD4KI) in which the extracellular fraction of the mouse CD4 antigen was replaced by the hCD4 while normal immunological function and T-cell distribution is restored (42). 64 Cu-CD4-Nb1 Specifically Accumulates in CD4 + T Cell-Rich Organs To generate immunoPET compatible tracers, CD4-Nb1 and GFP-Nb were labeled with the PET isotope 64 Cu using a copper-chelating BCN-NODAGA group added to our azidecoupled Nbs. Radiolabeling yielded high radiochemical purity (≥95%) and specific binding of 64 Cu-hCD4-Nb1 to CD4-expressing HBP-ALL cells (46.5% ± 5.6%) in vitro that was~30 times higher than the non-specific binding to CD4negative DHL cells or of the radiolabeled 64 Cu-GFP-Nb control (Supplementary Figure 12A).
Subsequently, we injected 64 Cu-CD4-Nb1 i.v. in hCD4KI and wt C57BL/6 mice and performed PET/MRI repetitively over 24 h. In two of the hCD4KI animals, we additionally followed tracer biodistribution over the first 90 min by dynamic PET (Supplementary Figure 12B). As expected for small-sized immunotracers, after an initial uptake peak within the first 10 min, 64 Cu-CD4-Nb1 is rapidly cleared from the blood, lung, and liver via renal elimination. In comparison to wild type, mice carrying the hCD4 antigen on T cells showed an increased tracer accumulation in lymph nodes, thymus, liver, and spleen ( Figure 6A). In these organs, which are known to harbor high numbers of CD4 + T cells (43), discrimination of CD4 + -specific signal from organ background was optimal 3 h post injection ( Figure 6B). Here, lymph nodes yielded a~3-fold, spleen a~2.5fold, and liver a~1.4-fold higher 64 Cu-CD4-Nb1 accumulation in hCD4KI mice compared to wt littermates ( Figure 6B). In contrast, we observed similar uptake levels for blood, muscle, lung, and kidney in both groups (Supplementary Figure 12C). Analyzing ex vivo biodistribution 24 h post tracer injection confirmed persistent accumulation of 64 Cu-CD4-Nb1 in lymph nodes and spleen of hCD4-expressing mice, although the limited number of animals per group did not allow statistical analysis (Supplementary Figure 12D). In summary, these results demonstrate that CD4-Nb1 is capable of visualizing and monitoring CD4 + T cells in both optical and PET-based imaging.

DISCUSSION
Given the important role of CD4 as a marker for a variety of immune cells including monocytes, macrophages, dendritic cells, and CD4 + T cells, detailed monitoring of this marker is proving to be extremely important for the diagnosis and concomitant therapeutic monitoring of a variety of diseases. Several mouse studies and early clinical trials have already indicated the value of noninvasive imaging of CD4 + cells in rheumatoid arthritis (20), colitis (21), allogeneic stem cell transplantation (44), organ transplant rejection (45), acquired immunodeficiency disease (10), and in the context of cancer immunotherapies (46), using radiolabeled full-length antibodies or fragments thereof. However, biological activity, particularly CD4 + T-cell depletion, and long-term systemic retention of full-length antibodies limit their development into clinically applied immunoprobes (20,24,47,48).  The aim of this study was to develop hCD4-specific Nbs as novel in vivo imaging probes to overcome the limitations of previous noninvasive imaging approaches. To identify binders that recognize the cellular exposed CD4, we employed two screening strategies where we selected Nbs either against adsorbed recombinant CD4 or against hCD4-expressing cells. Interestingly, both panning strategies proved successful, as demonstrated by the selection of two Nbs each that efficiently bind cell-resident CD4. Combining different biochemical analyses including epitope binning, cellular imaging, and HDX-MS, we were able to elucidate in detail the detected domains, as well as the three-dimensional epitopes addressed by the individual Nbs, and thus identified two candidates, CD4-Nb1 and CD4-Nb3, that can simultaneously bind to different segments within domain 1, while CD4-Nb2 has been shown to bind to domain 3 of CD4.
Notably, for most Nbs currently being developed for in vivo imaging purposes, detailed information about their epitopes is not available (49)(50)(51)(52)(53)(54)(55). There are only few examples such as the anti-HER2-specific Nb 2Rs15d, where the precise epitope was elucidated by complex crystallization (56) and that was successfully applied in a phase I study for clinical imaging (39). However, for CD4-specific Nbs, this knowledge is all the more important because epitope-specific targeting of CD4 + T-cell functions has far-reaching implications. This is true especially for cancer treatment, as CD4 + T cells have opposing effects on tumor growth and response to immunotherapies, crucially depending on the CD4 effector cell differentiation and tumor entity (57,58). In this context, it was shown that domain 1 of CD4 mediates transient interaction of the CD4 receptor and the MHCII complex (59)(60)(61), while T-cell activation is abrogated when T-cell receptor (TCR) and CD4 colocalization is blocked via domain 3 (62). To further elucidate a possible impact on immunomodulation, we analyzed the effect of CD4-Nb1 and CD4-Nb4 targeting two different domains on CD4 + Tcell proliferation and cytokine expression. Notably, neither CD4-Nb1 nor CD4-Nb4 affected the behavior of endogenous CD4 + T-cells in vitro or induced increased cytokine levels in whole blood samples when employed at concentrations that are intended for molecular imaging purposes in patients. From these data, it can be concluded that these Nbs are mostly biologically inert and thus might be beneficial compared to full-length antibodies (24) or other antibody fragments such as the anti-CD4 Cys-diabody, which was recently reported to inhibit the proliferation of CD4 + cells and IFN-g production in vitro (26).
Following our initial intention to generate immune tracer for in vivo imaging, we performed a site-directed labeling approach employing C-terminal sortagging to conjugate an azide group, which can be universally used to attach a multitude of detectable moieties by straightforward DBCO-mediated click chemistry (49). For the fluorescent and radiolabeled CD4-Nb1, we observed rapid recruitment and sustained targeting of CD4 + cells in a xenograft and hCD4 knock-in mouse model. Using a quantitative high-resolution PET/MR imaging approach, our radiolabeled 64 Cu-CD4-Nb1 allowed visualizing T cell-rich organs with high sensitivity. Beyond immune organs including lymph nodes, thymus, and spleen, we could detect enhanced CD4-Nb1 uptake in liver. At this point, we cannot distinguish whether this is due to the presence of CD4 + cells or non-specific elimination occurring through this organ. Consequently, further experiments are needed to analyze whether CD4-Nbs lacking the Fc region are advantageous compared to larger antibody formats, which have a higher tendency to accumulate non-specifically in the spleen and liver due to Fcg receptor-mediated uptake. However, to further modify serum retention times in order to improve specific tissue targeting, CD4-Nbs could easily be modified, as shown by the addition of an albumin-binding fragment (63) or PEGylation (49).
Considering the translation of the CD4-Nb1 for clinical imaging, additional aspects such as potential immunogenicity have to be assessed. Due to their high homology to the human type 3 VH domain, Nbs were described as weakly immunogenic in humans (64), and several strategies are available to humanize Nbs by exchanging a small number of aa residues in the framework regions (65). Moreover, a recent study of two Nbs in phase II clinical trials for PET imaging reported that very few patients developed low levels of anti-drug antibodies after prolonged administration of Nbs (66), indicating that monomeric Nbs present a low immunogenicity risk profile. In addition, long-term kidney retention of radiolabeled Nbs, mediated primarily by the endocytic receptor megalin (67), can cause undesirable nephrotoxicity and interfere with imaging of molecular targets near the kidneys. However, this can be overcome by targeted engineering of Nbs, e.g., removal of charged aa tags or simultaneous administration of positively charged components that interact with megalin receptors (68,69). Of note, compared to other radiolabeled Nbs used for preclinical imaging of similar targets (49), CD4-Nb1 showed relatively low renal accumulation. However, as the molecular reasons for this are unclear at this stage, further studies are needed to gain deeper insight into this phenomenon.
In summary, this study demonstrates for the first time the generation and detailed characterization of Nbs specific for hCD4 and their comprehensive experimental evaluation in vitro and in vivo. In particular, CD4-Nb1 turned out as a promising candidate for a noninvasive whole-body study of CD4 + cells in mice. Considering the increasing importance of advanced molecular imaging in clinical practice, we anticipate that this Nb-based immunotracer could become a highly versatile tool as a novel theranostic to accompany the clinical translation of emerging immunotherapies.

Nanobody Screening
For the selection of CD4-specific Nbs, two consecutive phage enrichment rounds were performed, both with immobilized recombinant antigen and CHO-hCD4 cells. E. coli TG1 cells comprising the hCD4-Nb library in pHEN4 were infected with the M13K07 helper phage to generate Nb-presenting phages. For each round, 1 × 10 11 phages of the hCD4-Nb library were applied on immunotubes coated with hCD4 (10 μg/ml). In each selection round, extensive blocking of antigen and phages was performed with 5% milk or bovine serum albumin (BSA) in phosphatebuffered saline containing 0.05% (v/v) Tween 20 PBS-T, and with increasing panning rounds, PBS-T washing stringency was increased. Bound phages were eluted in 100 mM triethylamine (TEA; pH 10.0), followed by immediate neutralization with 1 M Tris/HCl pH 7.4. For cell-based panning, 2 × 10 6 CHO-hCD4 or HEK293-hCD4 cells were non-enzymatically detached using dissociation buffer (Gibco) and suspended in 5% fetal bovine serum (FBS) in PBS. Antigen-expressing cells were incubated with 1 × 10 11 phages under constant mixing at 4°C for 3 h. Cells were washed 3× with 5% FBS in PBS. Cell lines were alternated between panning rounds. Phages were eluted with 75 mM citric acid buffer at pH 2.3 for 5 min. To deplete non-CD4-specific phages, eluted phages were incubated 3× with 1 × 10 7 wt cells. Exponentially growing E. coli TG1 cells were infected with eluted phages from both panning strategies and spread on selection plates for the following panning rounds. Antigen-specific enrichment for each round was monitored by counting colonyforming units (CFUs).

Whole-Cell Phage ELISA
Polystyrene Costar 96-well cell culture plates (Corning) were coated with poly-L-lysine (Sigma Aldrich) and washed once with H 2 O. CHO-wt and CHO-hCD4 were plated at 2 × 10 4 cells per well in 100 μl and grown to confluency overnight. Next day, 70 μl of phage supernatant was added to the culture medium of each cell type and incubated at 4°C for 3 h. Cells were washed 5× with 5% FBS in PBS. M13-horseradish peroxidase (HRP)-labeled antibody (Progen) was added at a concentration of 0.5 ng/ml for 1 h, washed 3× with 5% FBS in PBS. One-Step Ultra TMB 32048 ELISA Substrate [Thermo Fisher Scientific (TFS)] was added and incubated until color change was visible, and the reaction was stopped by addition of 100 μl of 1 M H 2 SO 4 . Detection occurred at 450 nm at a Pherastar plate reader, and phage ELISA-positive clones were defined by a 2-fold signal above wt control cells.

Expression Constructs
The cDNA of hCD4 (UniProtKB-P01730) was amplified from hCD4-mOrange plasmid DNA (hCD4-mOrange was a gift from Sergi Padilla Parra; addgene plasmid #110192; http://n2t.net/ addgene:110192; RRID : Addgene_110192) by PCR using forward primer hCD4 fwd and reverse primer hCD4 rev and introduced into BamHI and XhoI sites of a pcDNA3.1 vector variant [pcDNA3.1(+)IRES GFP, a gift from Kathleen_L Collins; addgene plasmid #51406; http://n2t.net/addgene:51406; RRID : Addgene_51406]. We replaced the neomycin resistance gene (NeoR) with the cDNA for Blasticidin S deaminase (bsd), amplified with forward primer bsd fwd and reverse primer bsd rev, by integration into the XmaI and BssHII sites of the vector. CD4 domain deletion mutants were generated using the Q5 Site-Directed Mutagenesis Kit (NEB) according to the manufacturer's protocol. For mutants lacking domain 1 of hCD4, we introduced an Nterminal BC2-tag (35). For the generation of plasmid pcDNA3.1_CD4_DD1_IRES-eGFP, we used forward primer DD1 f w d a n d r e v e r s e p r i m e r D D 1 r e v ; f o r pcDNA3.1_CD4_DD1DD2_IRES-eGFP, forward primer DD1DD2 fwd and reverse primer DD1DD2 rev; for pcDNA3.1_CD4_DD3DD4_IRES-EGFP, forward primer DD3DD4 fwd and reverse primer DD3DD4 rev. For bacterial expression of Nbs, sequences were cloned into the pHEN6 vector (73), thereby adding a C-terminal sortase tag LPETG followed by 6× His-tag for IMAC purification as described previously (34). For protein production of the extracellular domains 1-4 of hCD4 in Expi293 cells, corresponding cDNA was amplified from plasmid DNA containing full-length hCD4 cDNA (addgene plasmid #110192) using forward primer CD4-D1-4 f and reverse primer CD4-D1-4 r. A 6× His tag was introduced by the reverse primer. Esp3I and EcoRI restriction sites were used to introduce the cDNA into a pcDNA3.4 expression vector with the signal peptide MGWTLVFLFLLSVTAGVHS from the antibody JF5 (74).

Cell Culture, Transfection, Stable Cell Line Generation
HEK293T and CHO-K1 cells were obtained from ATCC (CCL-61, LGC Standards GmbH, Germany). As this study does not include cell line-specific analysis, cells were used without additional authentication. Cells were cultivated according to standard protocols. Briefly, growth media containing Dulbecco's modified Eagle's medium (DMEM) (HEK293) or DMEM/F12 (CHO) [both high glucose, pyruvate (TFS)] supplemented with 10% (v/v) FBS, L-glutamine, and penicillin/ streptomycin (P/S; all from TFS) were used for cultivation. Cells were passaged using 0.05% trypsin-EDTA (TFS) and were cultivated at 37°C and 5% CO 2 atmosphere in a humidified chamber. Plasmid DNA was transfected using Lipofectamine 2000 (TFS) according to the manufacturer's protocol. For the generation of the stable HEK293-hCD4 and CHO-hCD4 cell line, 24 h post transfection, cells were subjected to a 2-week selection period using 5 μg/ml Blasticidin S (Sigma Aldrich) followed by single cell separation. Individual clones were analyzed by live-cell fluorescence microscopy regarding their level and uniformity of GFP and CD4 expression.

Biolayer Interferometry
To determine the binding affinity of purified Nbs to recombinant hCD4, biolayer interferometry (BLItz, ForteBio) was performed. First, CD4 was biotinylated by 3-fold molar excess of biotin-Nhydroxysuccinimide ester. CD4 was then immobilized at singleuse streptavidin biosensors (SA) according to manufacturer's protocols. For each Nb, we executed four association/ dissociation runs with concentrations appropriate for the affinities of the respective Nbs (overall between 15.6 nM and 1 μM). As a reference run, PBS was used instead of Nb in the association step. As a negative control, the GFP-Nb (500 nM) was applied in the binding studies. Recorded sensograms were analyzed using the BLItzPro software, and dissociation constants (K D ) were calculated based on global fits. For the epitope competition analysis, two consecutive application steps were performed, with a short dissociation period of 30 s after the first association.

Peripheral Blood Mononuclear Cell Isolation, Freezing, and Thawing
Fresh blood, buffy coats, or mononuclear blood cell concentrates were obtained from healthy volunteers at the Department of Immunology or from the ZKT Tübingen gGmbH. Participants gave informed consent, and the studies were approved by the ethical review committee of the University of Tübingen, projects 156/2012B01 and 713/2018BO2. Blood products were diluted with PBS 1× (homemade from 10× stock solution, Lonza, Switzerland), and PBMCs were isolated by density gradient centrifugation with Biocoll separation solution (Biochrom, Germany). PBMCs were washed twice with PBS 1×, counted with an NC-250 cell counter (Chemometec, Denmark), and resuspended in heat-inactivated (h.i.) FBS (Capricorn Scientific, Germany) containing 10% dimethyl sulfoxide (DMSO; Merck). Cells were immediately transferred into a -80°C freezer in a freezing container (Mr. Frosty; TFS). After at least 24 h, frozen cells were transferred into a liquid nitrogen tank and were kept frozen until use. For the experiments, cells were thawed in Iscove´s Modified Dulbecco´s Medium (IMDM) (+L-Glutamin + 25 mM HEPES; Life Technologies) supplemented with 2.5% h.i. human serum (HS; PanBiotech, Germany), 1× P/S (Sigma-Aldrich), and 50 μm bmercaptoethanol (b-ME; Merck), washed once, counted, and used for downstream assays.

Affinity Determination by Flow Cytometry
For cell-based affinity determination, HEK293-hCD4 cells were detached using enzyme-free cell dissociation buffer (Gibco) and resuspended in FACS buffer (PBS containing 5% FBS). For each staining condition, 200,000 cells were incubated with suitable dilution series of CD4-Nbs at 4°C for 30 min. Cells were washed two times, and for detection of Cy5 AffiniPure Goat Anti-Alpaca IgG, VHH domain (Jackson ImmunoResearch) was applied for 15 min. PBMCs (Department of Immunology/ZKT Tübingen gGmbH, Germany) were freshly thawed and resuspended in FACS buffer. For each sample, 200,000 cells were incubated with suitable concentrations of CD4-Nbs coupled to CF568 in combination with 1:500 dilution of anti-CD3-FITC (BD Biosciences) at 4°C for 30 min. For control staining, PE/Cy5labeled anti-hCD4 antibody (RPA-T4, BioLegend) was used. After two washing steps, samples were resuspended in 200 μl FACS buffer and analyzed with a BD FACSMelody Cell Sorter. Final data analysis was performed via FlowJo10 ® software (BD Biosciences).

CD4 Deuteration Kinetics and Epitope Elucidation
On the basis of the affinity constants of 5.1 nM (CD4-Nb1), 6.5 nM (CD4-Nb2), and 75.3 nM (CD4-Nb3) (predetermined by BLI analysis), the molar ratio of antigen to Nb was calculated, ensuring 90% complex formation according to Kochert  Immobilized pepsin (TFS) was prepared using 60 μl of 50% slurry (in ammonium formate solution pH 2.5) that was then centrifuged (1,000 × g for 3 min at 0°C). The supernatant was discarded. Prior to each analysis, samples were thawed and added to the pepsin beads. After digestion for 2 min in a water ice bath, samples were separated from the beads by centrifugation at 1,000 × g for 30 s at 0°C using a 0.22-μm filter (Merck, Millipore) and immediately analyzed by LC-MS. Undeuterated control samples for each complex and CD4 alone were prepared under the same conditions using H 2 O instead of D 2 O. Additionally, each Nb was digested without addition of CD4 to generate a list of peptic peptides deriving from the Nb. The HDX experiments of the CD4-Nb complexes were performed in triplicate. The back-exchange of the method as determined using a standard peptide mixture of 14 synthetic peptides was 24%.

Chromatography and Mass Spectrometry
HDX samples were analyzed as described previously (75).

HDX Data Analysis
A peptic peptide list was generated in a preliminary LC-MS/MS experiment as described previously (75). For data-based search, no enzyme selectivity was applied; furthermore, identified peptides were manually evaluated to exclude peptides originated through cleavage after arginine, histidine, lysine, proline, and the residue after proline (79). Additionally, a separate list of peptides for each Nb was generated, and peptides overlapping in mass, retention time, and charge with the antigen digest were manually removed. Analysis of the deuterated samples was performed in MS mode only, and HDExaminer v2.5.0 (Sierra Analytics, USA) was used to calculate the deuterium uptake (centroid mass shift). HDX could be determined for peptides covering 87%-88% of the CD4 sequence (Supplementary Figure 11). The calculated percentages of deuterium uptake of each peptide between CD4-Nb and CD4-only were compared. Any peptide with uptake reduction of 5% or greater upon Nb binding was considered protected. All relevant HDX parameters are shown in Supplementary Table S3 as recommended (80).

Endotoxin Determination and Removal
The concentration of bacterial endotoxins was determined with Pierce LAL Chromogenic Endotoxin Quantitation Kit (TFS), and removal occurred using EndoTrap HD 1 ml (Lionex) according to the manufacturer's protocols.

Synthetic Peptides
The YQEFFWDANDIYRIF. All peptides were synthesized and dissolved in water 10% DMSO as previously described (purity ≥80%) and were kindly provided by S. Stevanović (81).

Stimulation and Cultivation of Peripheral Blood Mononuclear Cells
PBMCs from donors previously screened for ex vivo CD4 + T-cell reactivities against MHCII peptides were thawed and rested in Tcell medium (TCM; IMDM + 1× P/S + 50 mM b-ME + 10% h.i. HS) containing 1 mg/ml DNase I (Sigma-Aldrich) at a concentration of 2-3 × 10 6 cells/ml for 3 h at 37°C and 7.5% CO 2 . After resting, cells were washed once and counted, and up to 1 × 10 8 cells were labeled with 1.5-2 mM carboxyfluorescein succinimidyl ester (CFSE; BioLegend, USA) in 1 ml PBS 1× for 20 min according to the manufacturer's protocol. The cells were washed twice in medium containing 10% FBS after CFSE labeling and incubated with 5, 0.5, or 0.05 mM of CD4-Nb1, CD4-Nb4, or a control Nb for 1 h at 37°C in serum-free IMDM medium. Concentrations and duration were chosen to mimic the expected approximate concentration and serum retention time during clinical application. After incubation, cells were washed twice and counted, and each condition was separated into three parts and seeded in a 48-well cell culture plate (1.6-2.5 × 10 6 cells/well in triplicate). Cells were stimulated with either 10 mg/ml PHA-L (Sigma-Aldrich) or 5 mg/ml MHCII peptide(s) or left unstimulated and cultured at 37°C and 7.5% CO 2 . Then, 2 ng/ ml recombinant human IL-2 (R&D, USA) were added on days 3, 5, and 7. One-third of the culture on day 4, one half of the culture on days 6 and 8, and the remaining cells on day 12 were harvested, counted, and stained for flow cytometry analyses. For donor 1, the proliferation/activation status and cytokine production were analyzed in two different experiments, whereas for donors 2 and 3, cells from a single experiment were used for the three assays.

Full Blood Stimulation and Cytokine Release Assay
Here, 100 ml of lithium-heparin blood was incubated for 1 h at 37°C and 7.5% CO 2 . The blood was stimulated with 5 mM Nb (CD4-Nb1, CD4-Nb4, or control Nb), 100 ng/ml LPS (Invivogen, USA), or 2 mg/ml PHA-L in a final volume of 250 ml (serum-free IMDM medium) or left unstimulated for 24 h at 37°C and 7.5% CO 2 . After two centrifugations, supernatant was collected without transferring erythrocytes. The supernatants were frozen at -80°C until cytokine measurements. Levels of IL-1b, IL-1Ra, IL-4, IL-6, IL-8, IL-10, IL-12p70, IL-13, granulocyte-macrophage colony-stimulating factor (GM-CSF), IFN-g, macrophage chemotactic protein (MCP)-1, macrophage inflammatory protein (MIP)-1b, TNFa, and vascular endothelial growth factor (VEGF) were determined using a set of in-housedeveloped Luminex-based sandwich immunoassays each consisting of commercially available capture and detection antibodies and calibrator proteins. All assays were thoroughly validated ahead of the study with respect to accuracy, precision, parallelism, robustness, specificity, and sensitivity (83,84). Samples were diluted at least 1:4 or higher. After incubation of the prediluted samples or calibrator protein with the capture coated microspheres, beads were washed and incubated with biotinylated detection antibodies. Streptavidin-phycoerythrin was added after an additional washing step for visualization. For control purposes, calibrators and quality control samples were included on each microtiter plate. All measurements were performed on a Luminex FlexMap ® 3D analyzer system using Luminex xPONENT ® 4.2 software (Luminex, USA). For data analysis, MasterPlex QT, version 5.0, was employed. Standard curve and quality control samples were evaluated according to internal criteria adapted to the Westgard Rules (85) to ensure proper assay performance.

Optical Imaging of CD4-Expressing HPB-ALL Tumors
Human T-cell leukemia HPB-ALL cells (German Collection of Microorganisms and Cell Cultures GmbH, DSMZ, Braunschweig, Germany) were cultured in RPMI-1640 supplemented with 10% FBS and 1% P/S. Here, 10 7 HPB-ALL cells were injected subcutaneously in the right upper flank of 7-week-old NSG (NOD.Cg-Prkdc scid Il2rg tm1WjI /SzJ; Charles River Laboratories, Sulzfeld, Germany) mice, and tumor growth was monitored for 2-3 weeks. When tumors reached a diameter of~7 mm, 5 μg of CD4-Nbs-Cy5.5 or control Nb (GFP-Nb-Cy5.5) were administered into the tail vein of two mice each. Optical imaging (OI) was performed repetitively in short-term isoflurane anesthesia over a period of 24 h using the IVIS Spectrum In Vivo Imaging System (PerkinElmer, Waltham, MA, USA). Four days after the first Nb administration, the CD4-Nbs-Cy5.5 groups received the GFP-Nb-Cy5.5 (and vice versa), and tumor biodistribution was analyzed identically by OI over 24 h. After the last imaging time point, animals were sacrificed and tumors were explanted for ex vivo OI analysis. Data were analyzed with Living Image 4.4 software (PerkinElmer). The fluorescence intensities were quantified by drawing regions of interest around the tumor borders and were expressed as average radiant efficiency (photons/s)/(mW/cm 2 ) subtracted by the background fluorescence signal before Nb injection to eliminate potential residual signal from the previous Nb application. All mouse experiments were performed according to the German Animal Protection Law and were approved by the local authorities (Regierungspräsidium Tübingen).

Radiolabeling With NODAGA and 64 Cu
All procedures for conjugation and radiolabeling with 64 Cu were performed using metal-free equipment and Chelex 100 (Sigma-Aldrich) pretreated buffers. Azide-modified Nbs (100 μg) were treated with 4 μl of 5 mM EDTA in 250 mM sodium acetate buffer (pH 6) for 30 min at room temperature (RT). The protein was reacted with 15 molar equivalents of BCN-NODAGA (CheMatech, Dijon, France) in 250 mM sodium acetate pH 6 for 30 min at RT followed by incubation at 4°C for 18 h. Excess of chelator was removed by ultrafiltration (Amicon Ultra 0.5 ml, 3 kDa MWCO, Merck Millipore) using the same buffer. [ 64 Cu] CuCl 2 (150 MBq in 0.1 M HCl) was neutralized by addition of 1.5 volumes of 0.5 M ammonium acetate solution (pH 6), resulting in a pH of 5.5. To this solution, 50 μg of conjugate was added and incubated at 42°C for 60 min. Then, 1 μl of 20% diethylenetriaminepentaacetic acid (DTPA) solution was added to quench the labeling reaction. Complete incorporation of the radioisotope was confirmed after each radiosynthesis by thinlayer chromatography (iTLC-SA; Agilent Technologies; mobile phase 0.1 M citric acid pH 5) and high-performance size exclusion chromatography (HPSEC; BioSep SEC-s2000, 300 × 7.8 mm, Phenomenex; mobile phase DPBS with 0.5 mM EDTA). All radiolabeled preparations used for in vivo PET imaging had radiochemical purities of ≥97% (iTLC) and ≥94% (HPSEC).

In Vitro Radioimmunoassay
Here, 10 7 HPB-ALL or DHL cells were incubated in triplicate with 1 ng (3 MBq/μg) of radiolabeled 64 Cu-CD4-Nb1 or 64 Cu-GFP-Nb for 1 h at 37°C and washed twice with PBS/2% FCS. The remaining cell-bound radioactivity was measured using a Wizard² 2480 gamma counter (PerkinElmer Inc., Waltham, MA, USA) and quantified as percentage of total added activity.

PET/MRI
Human CD4 knock-in (hCD4KI, genOway, Lyon, France) and wt C57BL/6J mice (Charles River) were injected intravenously with 5 μg (~15 MBq) of 64 Cu-CD4-Nb1. During the scans, mice were anesthetized with 1.5% isoflurane in 100% oxygen and warmed by water-filled heating mats. Ten-minute static PET scans were performed after 1.5, 3, 6, and 24 h in a dedicated small-animal Inveon microPET scanner (Siemens Healthineers, Knoxville, TN, USA; acquisition time: 600 s). For anatomical information, sequential T2 TurboRARE MR images were acquired immediately after the PET scans on a small animal 7 T ClinScan magnetic resonance scanner (Bruker BioSpin GmbH, Rheinstetten, Germany). Dynamic PET data of the first 90 min post injection were gained in two mice and divided into 10-min frames. After attenuation correction by a cobalt-57 point source, PET images were reconstructed using an ordered subset expectation maximization (OSEM3D) algorithm and analyzed with Inveon Research Workplace (Siemens Preclinical Solutions). The volumes of interest of each organ were drawn based on the anatomical MRI to acquire corresponding PET tracer uptake. The resulting values were decay-corrected and presented as percentage of injected dose per volume (%ID/ml). Ex vivo g-counting was conducted after the last imaging time point by measuring the weight and radioactivity of each organ. For quantification, standardized aliquots of the injected radiotracer were added to the measurement.

Analyses and Statistics
Data analysis of the flow cytometry data was performed with the FlowJo Software Version 10.6.2 (FlowJo LLT, USA), and graph preparation and statistical analysis were performed using the GraphPad Prism Software (Version 8.3.0 or higher).

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The animal study was reviewed and approved by Regierungspräsidium Tübingen Konrad-Adenauer-Straße 20 72072 Tübingen.