Molecular imaging of tumors with nanobodies and antibodies: Timing and dosage are crucial factors for improved in vivo detection

,


INTRODUCTION
In vivo molecular imaging focuses on the non-invasive detection and characterization of target structures with specific probes (1,2).Based on their unmatched binding specificity and affinity, monoclonal antibodies are considered the most specific probes for targeted imaging (3)(4)(5).However, their preclinical and clinical utility is limited due to their relatively poor and slow tissue penetration, slow clearance from circulation, and long retention in non-targeted tissues (6).These characteristics impair their signal-to-background ratio (7).The development of new contrast agents and novel engineered forms of antibodies, such as diabodies, minibodies, single-chain variable fragments, and nanobodies, has triggered a new wave of antibody-based imaging approaches (3,5,6,8).Among these, nanobodies are the smallest available antigen-binding fragments derived from camelid heavy-chain-only antibodies (9,10).With only ~15-18 kDa, these small antibody fragments are soluble, very stable and are renally cleared from the circulation (9,11).These properties make them particularly suited for specific and efficient targeting of tumor antigens in vivo (12)(13)(14)(15)(16)(17)(18)(19)(20).
Recent studies have shown that nanobodies allow higher tumor-to-background (T/B) ratios than conventional antibodies in molecular imaging applications in vivo (17,19).However, the T/B ratio depends on two parameters: specific binding of the antibody-construct leading to accumulation in the tumor as well as clearance of unbound constructs from the body.Both parameters depend on the molecular size of the probe.A small nanobody (~15 kDa) is expected to penetrate tumor tissue more rapidly than a larger conventional antibody (150 kDa).In turn, clearance of a nanobody via the kidney is fast, leading to short circulation times with a half-life of only ~1.5 h, as compared to days or weeks of conventional antibodies (3,11,21).Therefore, the comparison of two differently sized antibody-constructs for specific imaging of targets in vivo has to take into account the different kinetics of tumor accumulation and elimination from the circulation (11).This means that different imaging time points have to be investigated to determine best imaging conditions for each individual antibody construct.Moreover, different doses of the antibodies have to be compared, since higher doses may increase both specific and unspecific signals alike, thus reducing the maximum achievable T/B ratio.
There are only few studies directly comparing nanobodies and conventional antibodies for in vivo molecular imaging.Moreover, these studies did not optimize the doses of the conventional antibody and some used high doses of up to 100 μg antibody per animal (17,19).The high dosage inevitably leads to an excess of free circulating antibodies.When assessing T/B ratios based on the region of interest (ROI) in near-infrared fluorescence (NIRF) imaging experiments, the signal of the normal tissue serving as "background" will increase by the circulating antibodies.This results in lower T/B ratios of antibodies compared to nanobodies, not due to differences in specific signal, but due to higher unspecific background signal.Moreover, these excess antibodies are all prone to non-specific accumulation in target antigen-negative tumors by the enhanced permeability and retention (EPR) effect (22,23).Apart from kinetics and dosage, an intraindividual assessment of antigenpositive and -negative xenografts would enable a direct comparison of specific and unspecific signals due to the EPR effect.In addition, a comparative ex vivo analysis of explanted tumors would further determine the specificity observed in imaging experiments in vivo to optimize imaging conditions of differently sized constructs.
Therefore, we performed a direct NIRF-imaging comparison of a single-domain nanobody (s+16a, 17 kDa) and a monoclonal antibody (Nika102, 150 kDa) directed to the same target to improve specific in vivo NIRF-imaging in a lymphoma xenograft model.S+16 is a nanobody (single variable domain) derived from a heavy-chain-only llama antibody.Nika102 is a conventional monoclonal antibody (rat IgG2a kappa, composed of two heavy chains and two light chains) (Fig. 1A).Compared to mAb Nika102, nanobody s+16a exhibits a lower binding affinity (40 nM vs. 5 nM) and shorter in vivo blood half-life (2 h vs. 8 days) (11,24,25).The model target antigen ADP-ribosyltransferase ARTC2 is expressed on the surface of lymphoma cells (25)(26)(27)(28)(29). Beyond assessment of the advantages and disadvantages of nanobodies and conventional antibodies, this study was designed to determine the specific requirements, such as timing and dosage, for optimum imaging of tumors.

RESULTS
Purity of s+16a and Nika102 before and after conjugation to AF680 was confirmed (Fig. 1B).Assessment of binding affinities showed less than 10% reduction in labeling efficiencies upon overnight incubation in serum (Fig. 1C).Competition studies revealed that nanobody s+16a and antibody Nika102 recognize different epitopes of ARTC2 (Fig. 1D).Internalization studies showed prominent staining of the cell surface upon incubation of cells at 4°C with both constructs.Upon incubation at 37°C, most of the labeled nanobody s+16a 680 and antibody Nika102 680 remained on the cell surface.Patchy cytosolic staining with both constructs after incubation for 24 h at 37°C indicates that a fraction of fluorescent label is internalized during prolonged incubation at 37°C (online Supplementary Fig. 1).

In vitro NIRF-imaging experiments
Flow cytometry showed specific labeling of ARTC2-transfected DC27.10 lymphoma cells with s+16a 680 and Nika102 680 , but not of parental ARTC2-negative DC27.10 cells (Fig. 2A).Staining with s+16a 680 yielded lower fluorescence intensities (MFI = 15100 ± 1700) as compared to Nika102 680 (MFI = 73300 ± 1500), which reflects the lower labeling efficiency with fluorochrome AF680 of s+16a compared to Nika102 (0.3 dyes/molecule for s+16a as compared to 2.0 dyes/molecule for Nika102).To provide a basis for the subsequent in vivo and ex vivo xenograft experiments, we further analyzed the same cells in vitro by fluorescence microscopy and with the NIRF-imaging system.Labeling of s+16a and Nika102 with AF680 not only allowed detection of ARTC2-positive cells with fluorescence microscopy but also semi-quantitative analyses with the NIRF-imaging system intended for in vivo experiments (Fig. 2B,  C).ARTC2-negative cells showed no detectable signals using either technique.As shown by flow cytometry, the fluorescence signal detected from ARTC2-positive cells with the NIRF-imaging system was lower when labeled with s+16a 680 (radiant efficiency = 5.1 ± 2.1 × 10 7 ) than with Nika102 680 (radiant efficiency = 9.4 ± 2.5 × 10 7 ).Signals from ARTC2-negative cells were more than 10 times lower than those of ARTC2-positive cells for s+16a 680 (radiant efficiency = 3.0 ± 1.5 × 10 6 ) as well as for Nika102 680 (radiant efficiency = 3.5 ± 2.3 × 10 6 ).

NIRF-imaging experiments in vivo
Next, we aimed to evaluate the suitability of Alexa-680conjugated nanobodies and antibodies for in vivo imaging of ARTC2-positive tumors.Therefore, we intravenously injected s+16a 680 and Nika102 680 into mice at four different doses (5, 10, 25, and 50 μg) 7-9 days after subcutaneous injection of ARTC2positive and ARTC2-negative DC27.10 lymphoma cells on the opposite flanks of the same animals.The injected doses correspond to 0.004, 0.009, 0.022, and 0.044 mg of dye/kg for s+16a 680 and to 0.003, 0.006, 0.016, and 0.032 mg of dye/kg for Nika102 680 for a mouse of 24 g.Imaging was performed before and at defined time-points over 24 h after injection of the fluorochrome conjugates (Figs. 3 and 4 and Supplementary Fig. 1).The results showed efficient and specific labeling of ARTC2-positive tumors with both constructs.When injected with Nika102 680 , the specific signal in ARTC2-expressing tumors slowly increased over time, whereas after injection of s+16a 680 , the signal was already strong at early time points and declined over time (Fig. 3).Both constructs showed higher signals of ARTC2-positive tumors with increasing concentrations.At early time points after injection of s+16a 680 , strong signals were observed also in the kidneys, reflecting passage of the nanobodies through the renal filtration barrier.Note the high background signals of the entire animal  after injection of 50 μg of Nika102 680 and the unspecific signals of the negative tumor when using higher doses of Nika102 680 (Figs.3D and 4A).
The intermediate dose of 25 μg showed lower specific signals for both s+16a 680 and Nika102 680 as compared to 50 μg and higher specific signals as compared to 10 μg (Supplementary Fig. 2).The dose of 25 μg also showed higher unspecific signals in ART2-negative tumors for Nika102 680 resulting in a lower T/B-ratio as compared to 10 μg.The lowest dose of 5 μg showed the lowest specific and unspecific signals for both constructs with lowest resulting T/B ratios.In summary, the comparison of different doses of AF680-conjugates showed best imaging results with 50 μg of s+16a 680 and with 10 μg of Nika102 680 by achieving highest specific signal intensities (s+16a 680 ) and by minimizing unspecific signals of ARTC2-negative tumors and background while maintaining sufficient specific signals (Nika102 680 ), thereby allowing for highest T/B-ratios.
Semi-quantitative ROI analyses confirmed a rapidly increasing T/B ratio of ARTC2-positive tumors after injection of s+16a 680 , which reached a maximum of 12.4 ± 4.2 (50 μg) and 4.7 ± 0.1 (10 μg), respectively, already 4-6 h post-injection (Fig. 4B).The T/B ratio of ARTC2-positive tumors detected with s+16a 680 was significantly higher than of ARTC2-negative tumors throughout 2 h to 24 h post-injection.In contrast, the T/B ratio using Nika102 680 increased only slowly and reached its maximum of 8.7 ± 3.9 (10 μg) and 6.1 ± 2.0 (50 μg), respectively, not until 24 h post-injection.In the case of Nika102 680 , the T/B ratio of ARTC2-positive tumors was not significantly higher than of ARTC2-negative tumors until 8 h to 24 h post-injection due to the high unspecific signal of ARTC2-negative tumors.
The need to evaluate different time points and different doses when comparing differently sized antibodies for coherent NIRFimaging experiments is illustrated in Fig. 5.The direct comparison of the T/B ratios obtained by using either 10 μg or 50 μg of each AF680-conjugate at different time points revealed that in the case of s+16a 680 , significantly higher T/B ratios can be obtained with the higher concentration of 50 μg 4 to 6 h post-injection.Interestingly, in Nika102 680 -injected mice, a significant difference between the two dosing regimens was seen only after 24 h, and, as shown above, with a higher T/B ratio obtained by the lower concentration of 10 μg Nika102 680 .Based upon these observations, the following ex vivo experiments were performed at the optimum imaging time points and optimum probe concentrations of 6 h and 50 μg for s+16a 680 and of 24 h and 10 μg for Nika102 680 .

NIRF-imaging experiments ex vivo
ARTC2-positive and ARTC2-negative tumors were explanted 6 h and 24 h post-injection to quantify tumor-associated fluorescence and T/B ratios in the absence of potentially confounding signals from other tissues.The results of NIRF-imaging ex vivo reflect those of the in vivo experiments (Fig. 6).Both AF680 conjugates showed high signals from ARTC2-positive tumors, which decreased for s+16a 680 and increased for Nika102 680 over time (Fig. 6A).Signals of ARTC2-positive tumors were significantly higher than of ARTC2-negative tumors in the case of s+16a 680 only at 6 h (p < 0.001) and in the case of Nika102 680 only at 24 h (p < 0.05) post-injection.Note the low unspecific signal of ARTC2-negative tumors after injection of s+16a 680 at both time points, as compared to the high unspecific signal of ARTC2negative tumors after injection of Nika102 680 .In the case of s+16a 680 , the T/B ratio of dissected ARTC2-positive tumors was significantly higher than that of ARTC2-negative tumors at both time points, 6 h and 24 h post injection (p < 0.01 and p < 0.05, respectively) (Fig. 6B).As for the absolute signal intensities, Nika102 680 culminated in a significantly higher T/B ratio of positive tumors as compared to negative tumors only 24 h postinjection (p < 0.05).Note the much higher T/B ratio (61.4 ± 30.9) of ARTC2-positive tumors 24 h post-injection of Nika102 680 as compared to the maximum achievable T/B ratio of s+16a 680 Biodistribution analyses of spleen, lungs, liver, kidneys, stomach, ileum, and muscle revealed an overall decline of signal intensities at 24 h as compared to 6 h (Fig. 7A).At both time points, 6 h and 24 h post-injection, the highest unspecific organ-to-background ratios were observed for the liver in the case of Nika102 680 and for the kidneys in the case of s+16a 680 (Fig. 7B).Interestingly, in the case of Nika102 680 , at 6h post-injection the second highest organ-to-background ratio was observed for the ARTC2-negative tumor and at 24 h the ARTC2-negative tumor showed even the highest unspecific signals when compared to other organs.

Flow cytometry and fluorescence microscopy ex vivo
Next, we determined possible causes of the unspecific signals from normal tissue after injection of Nika102 680 in vivo (Figs.3D,  4A) as well as from ARTC2-negative tumors in vivo and ex vivo (Figs.4A and 6).Therefore, we performed flow cytometry ex vivo to quantify levels of unbound AF680-conjugates in serum and of cell bound AF680-conjugates on dispersed cells from the explanted tumors (Fig. 8).Serum samples showed little, if any, detectable circulating s+16a 680 at 6 h and 24 h post-injection (Fig. 8A).In contrast, mice that had been injected with Nika102 680 showed significantly higher levels of unbound circulating Nika102 680 in serum at both time points, which somewhat decreased over time, but were still present in excess 24 h postinjection.Urine analyses revealed high levels of s+16a 680 , particularly 6 h post-injection, but little, if any, Nika102 680 (data not shown).
Flow cytometry of dispersed cells from xenografts dissected 6 h and 24 h post-injection showed specific labeling of ARTC2positive lymphoma cells with both AF680-conjugates and no unspecific labeling of ARTC2-negative tumor cells (Fig. 8B).As seen with NIRF-imaging in vivo and ex vivo, labeling and specific signals from ARTC2-positive tumors decreased over time after injection of s+16a 680 , whereas signals of ARTC2-positive tumors from animals injected with Nika102 680 increased over time.Signals of cells from ARTC2-positive tumors from animals injected with s+16a 680 were higher than signals after injection of Nika102 680 at 6 h post injection, but lower at 24 h after injection.We further analyzed the distribution of injected AF680conjugates within explanted and cryosectioned tumors ex vivo by confocal fluorescence microscopy (Fig. 9).S+16a 680 revealed homogeneous and specific labeling of ARTC2-positive tumor cells 6 h post-injection, similar to the staining of these cells in culture (Fig. 2B).In contrast, Nika102 680 showed only weak staining of cells in ARTC2-positive tumors after 6 h.Moreover, the monoclonal antibody showed staining evidently not associated with tumor cells in both ARTC2-positive and ARTC2-negative tumors (Fig. 9A, arrow).

DISCUSSION
The utility for both nanobodies and conventional antibodies for in vivo imaging is well established, but optimum dosing and timing schedules for one versus the other have not been determined so far.Here, we used NIRF-dye AF680-conjugated nanobodies and conventional monoclonal antibodies directed at the same target on lymphoma cells for a direct comparison of in vivo and ex vivo analyses.We showed that nanobodies are well suited as diagnostic tools for rapid and specific in vivo detection of lymphomas, with superior tissue penetration compared to conventional antibodies and significantly higher T/B ratios when performing same-day imaging in vivo.In addition, our experiments revealed that at later time points the T/B ratio of conventional antibodies can be improved by using lower doses of antibody conjugates.
In vitro, both, nanobody s+16a 680 and antibody Nika102 680 bound specifically to ARTC2-positive lymphoma cells with no unspecific labeling of ARTC2-negative cells (Fig. 2).Albeit that Nika102 680 showed stronger signals in vitro, s+16a 680 allowed a faster and more specific detection of ARTC2-positive xenografts in vivo (Figs. 3 and 4).Apart from the different kinetics for best tumor visualization in vivo, the major drawback of Nika102 680 at higher doses (50 μg) was the high unspecific signal originating not only from normal tissues (causing fluorescence signals from the entire mouse), but also from ARTC2-negative tumors (Fig. 4).This apparently insufficient T/B ratio could be improved by using a lower dose (10 μg) of Nika102 680 , resulting in dramatically reduced signals of normal tissues and ARTC2-negative tumors and only slightly decreased specific signals of ARTC2-positive tumors (Fig. 5).In contrast, increasing the dose of s+16a 680 caused stronger signals of ARTC2-positive tumors but not of normal tissue or ARTC2-negative tumors.
NIRF imaging of dissected tumors ex vivo revealed the overall strongest T/B ratio of ARTC2-positive tumors 24 h post-injection of Nika102 680 in the absence of potentially confounding signals from other tissues (Fig. 5).These in vivo and ex vivo NIRF-imaging observations could be well explained by the results of ex vivo flow cytometric analyses for quantification of injected AF680-conjugates in serum (Fig. 8A).These showed high levels of unbound and free circulating Nika102 680 6 h and even 24 h post-injection, even at the lower dose of 10 μg, whereas after the injection of 50 μg of s+16a 680 no unbound nanobodies were  detectable in serum at these time points.These results are in accordance with a recent study in which we determined the pharmacokinetics of s+16a 680 and Nika102 680 (11).It is conceivable, that higher doses would further increase the amount of free circulating Nika102 680 , and thereby further increase the observed unspecific signals in vivo.Ex vivo flow cytometric analyses of dispersed cells from dissected tumors showed specific staining of ARTC2-positive tumor cells with both constructs but no unspecific binding of injected AF680-conjugates to ARTC2-negative lymphoma cells (Fig. 8B).These findings are consistent with in vitro labeling experiments, which show prominent cell-surface staining of ART2-positive tumor cells as well as some internalized fluorescence for both constructs after 24 h incubation at 37°C (Supplementary Fig. 1).The higher staining intensity of ART2positive tumor cells with nanobody s+16a 680 than with mAb Nika102 680 at 6 h post-injection likely reflects the higher injected dose and faster tissue penetration of the nanobody.The lower staining intensity of ART2-positive tumor cells with nanobody s+16a 680 than with mAb Nika102 680 at 24 h post-injection likely reflects the lower affinity and renal elimination of excess nanobody.The increase in staining intensity of mAb Nika102 680 at 24 h versus 6 h likely reflects increasing accumulation in the tumor from excess levels of circulating mAb.Fluorescence microscopy completed the ex vivo imaging approach.In the case of the nanobody s+16a, homogeneous staining of cells in ART2C-positive tumor sections correlated well with the staining of cells in vitro (Figs.9A and 2B), confirming that the nanobody was able to reach even remote areas within the tumor after 6 h.In contrast, the monoclonal antibody showed only weak staining of cells in ARTC2-positive tumors after 6 h.Moreover, the monoclonal antibody showed staining within or close to the tumor vasculature even in ARTC2-negative tumors (Fig. 9A), providing a plausible explanation for the unspecific signals during in vivo imaging.The better tissue penetration of the nanobody s+16a at early time points is likely attributed to its ability to cross the endothelial barrier more easily to reach its target.NIRF imaging of other organs ex vivo revealed high unspecific signals for the liver in the case of Nika102 680 and for kidneys in the case of s+16a 680 .High signals in the kidney at early time points are consistent with renal filtration and partial retention of nanobody s+16a 680 in the kidneys.High signals in the liver for both constructs are consistent with hepatic elimination of fluorochromes.Note that at the time point of ex vivo analyses (6 h and 24 h), the signals from kidneys had already decreased dramatically as compared to the in vivo signals at 1 h and 2 h post-injection, presumably by elimination of most of labeled nanobody s+16a 680 from the body via urination.Signals from the kidneys even after 24 h likely reflect partial tubular resorption of nanobodies and/or fluorochromes.
As in previous studies, the results reported here emphasize that molecular imaging with labeled nanobodies in vivo allows rapid and specific same-day tumor imaging, including the possibility of increasing the dose for optimized T/B ratios without compromising specificity of marked tissues (12)(13)(14)(15)(17)(18)(19).Our study also reveals, that in vivo molecular imaging using conventional antibodies can be considerably improved by determining the optimum dose of the injected large (and therefore not renally cleared) antibodies.A fivefold lower dose of conventional antibody can almost double the T/B ratio.Recent studies, comparing nanobodies and conventional antibodies for in vivo molecular imaging neglected to optimize the dose of the investigated conventional antibody and used very high doses of up to 100 μg per animal (17,19).This leads to an excess of free-circulating antibodies, as we were able to show for mAb Nika102 680 in the present study.These excess antibodies are prone to nonspecific accumulation in target-negative tumors by the enhanced permeability and retention effect (EPR) described earlier for conventional antibodies (22,23).It should be noted, that the EPR effect also leads to an increased uptake of antibodies in the target-positive tumors and may thereby contribute to increase the specific signals at low antibody doses.Taken together, thorough dose optimization of conventional antibodies should be performed for in vivo imaging applications.
For nanobodies, in contrast, dose optimization seems to be less of an issue, since any excess is renally cleared within 2 h and thus unspecific accumulation in non-targeted tissue is minimized.This allows same-day imaging and could be translatable to the clinical setting.However, next-day imaging might be more convenient in a clinical setting.Further, the cost efficiency might be better for the antibody due to the lower dose required, even though nanobodies can be produced at low costs in bacteria and yeast (10).
A limitation of our study is that we did not optimize the amount of fluorescent dyes per antibody construct, which might affect the maximum achievable signal for imaging.Another improvement of the labeling strategy would have been the site-specific conjugation of the NIRF dye, as recently described by Kijanka et al., instead of random conjugation to primary amine groups, which might affect binding affinity (17).Another limitation is that we injected fixed doses of 5, 10, 25 and 50 μg of our constructs regardless of the weight of the animals, instead of performing weight-adapted injections.The differences in weight of individual mice (range 22.5-26.1 g) may explain some of the observed signal variations within different dose groups.
The region chosen for "background" signal estimation also influences the calculated T/B ratios.However, nanobody s+16a and mAb Nika102 show different accumulations in different abdominal organs and blood, precluding a fair and comparable estimation of T/B ratios with these tissues.As an alternative, we chose the hind limb and muscle, respectively, because this tissue introduced less variation, even though it might underestimate the systemic background.
In addition, further studies should investigate time points later than 24 h post-injection (e.g.48 h, 72 h).These might show a further improvement of the maximum achievable T/B ratio using the conventional antibody, as recently shown for Trastuzumab by Kijanka et al. (17).A limitation of nanobodies in imaging is the high confounding signal of the kidneys due to their renal elimination, which is particularly prominent at early time points.This confounding signal might limit the ability of nanobodies to detect tumors located close to the kidneys.However, the renal retention of nanobodies can be reduced by 45% upon co-injection of gelofusine and lysine without reduced tumor uptake, as reported by Gainkam et al. (30).
An intrinsic limitation when comparing llama-derived nanobodies and conventional antibodies is the fact that they bind to different epitopes of the target antigen, which might influence antibody internalization, unspecific binding to other sites or uptake by the reticular endothelial system, thereby affecting blood half-life.However, the differences observed here in terms of dose and timing for optimal imaging can be mainly explained by the size difference of the nanobody and the conventional antibody.wileyonlinelibrary.com/journal/cmmiAn inherent technical limitation of NIRF imaging is its low penetration depth of 7-10 mm, which is particularly suited for imaging of subcutaneous tumors but does not allow for imaging of orthotopic tumors.However, even for subcutaneous tumors, some spillover from abdominal organs such as kidneys or liver cannot be excluded.Therefore, thorough ex vivo validation experiments of explanted tumors and organs are mandatory.Another limitation of NIRF imaging is the only semi-quantitative assessment of biodistribution as compared to radionuclidemediated imaging.However, this limitation is compensated in part by the suitability of NIRF-labeled probes for ex vivo validation experiments, that is, quantitative flow cytometric assessment of antibodies bound to tumor cells.Moreover, for the principle aim of this study, that is, optimizing timing and dosing of nanobodies and antibodies for imaging in vivo, the NIRF technique is well suited.Our results are in accord with a recent study by Oliveira et al. using NIRF-labeled nanobodies (19).If desired, nanobodies can also be radiolabeled for positron emission tomography (PET) imaging of xenograft models.A recent study that compared nanobodies and conventional antibodies for PET imaging also came to the conclusion that nanobodies allow same-day imaging with high tumor-to-background ratios (15).

CONCLUSIONS
Labeling of nanobody s+16a and antibody Nika102 with AF680 allowed validation of in vivo NIRF-imaging results with ex vivo flow-cytometry and fluorescence-microscopy.For this reason, and because it is nonradioactive, highly sensitive, inexpensive, and uses comparatively easy-to-produce targeted probes, we advocate the use of the NIRF-imaging technique for evaluation of new antibody constructs in preclinical molecular imaging experiments.
Our comparative in vivo and ex vivo analyses revealed that the specific contrast between tumor and normal tissue is more important than the absolute amount of conjugate that reaches the tumor.Moreover, our study confirmed that timing and dosage significantly influence specific and unspecific in vivo NIRF-imaging signals of s+16a 680 and Nika102 680 .Nanobody s+16a allowed same-day imaging with high tumor-tobackground ratio, whereas antibody Nika102 gave optimal imaging results only 24 h post-injection.Nanobody s+16 required a high dose while antibody Nika102 had the best tumor-tobackground ratio at a low dose.Therefore, timing and dosage should be addressed when using nanobodies and conventional antibodies for molecular imaging purposes.

Cell lines and mice
ARTC2-transfected and untransfected DC27.10 murine lymphoma cells were cultured as described previously (26).The closely related ARTC2.1 and ARTC2.2T cell GPI-anchored cell surface ecto-enzymes are encoded by tandem genes.The study was performed with reagents specific for ARTC2.2, for better legibility we use the term ARTC2.Tumor xenograft experiments were conducted using athymic nude mice (NMRI-Foxn1 nu ) weighting 24.0 ± 1.4 g (range 22.5 to 26.1 g).Mice were obtained from Charles River Laboratories (Sulzfeld, Germany).Experiments were performed in accordance with international guidelines on the ethical use of animals and were approved by the local animal welfare commission.

Generation of AF680-conjugates
Generation and purification of ARTC2.2-specificnanobody s+16a (~17 kDa) and of monoclonal antibody Nika102 (150 kDa) were described previously (24,25).S+16a carries C-terminal His6x and c-Myc epitope tags and has a calculated size of 17.4 kDa (25).Nanobody s+16a and antibody Nika102 were conjugated to the fluorescent dye AlexaFluor-680 (AF680) (Molecular Probes, Carlsbad, CA, USA) (excitation wavelength = 679 nm, emission wavelength = 702 nm) and number of dye molecules per probe were calculated using molar extinction coefficients of 15720 cm À1 M À1 and 203000 cm À1 M À1 , respectively.Purity of antibody constructs before and after conjugation to AF680 was assessed by SDS-PAGE size fractionation and Coomassie brilliant blue gel stain as described previously (11).Binding affinities as well as stability during overnight incubation at 37°C in serum were assessed by serial dilution of probes and flow cytometric analyses of labeled DC27.10 cells.Competition studies were performed to evaluate whether s+16a and Nika102 recognize distinct or overlapping epitopes.ARTC2-transfected DC27.10 cells were pretreated with phosphate buffered saline (PBS), unlabeled s+16a (5 μg/100 μL) or unlabeled Nika102 (5 μg/100 μL) for 20 min at 4°C before exposure to AF680-conjugated Nika102 (0.2 μg/100 μL), s+16a (0.2 μg/100 μL) or isotype control antibodies for 20 min and analysed by flow cytometry.Internalization studies were performed by staining of DC27.10 ARTC2 cells with s+16a 680 Nika102 680 for 30 min on ice and washed.Cells were further incubated in cell culture medium at 4°C for 2 h or at 37°C for 2 and 24 h before fixation in 2% PFA, counterstaining with Hoechst 33428, and analysis by fluorescence microscopy.

In vitro analyses
For in vitro flow-cytometric analyses, 1 × 10 6 untransfected or ARTC2-transfected DC27.10 cells were stained with s+16a 680 or Nika102 680 (1 μg/μL) or control antibodies for 30 min at 4°C.Cells were washed twice and analyzed by flow cytometry on a FACS Canto II (BD Biosciences, Becton Dickinson, Franklin Lakes, USA).Dead cells were excluded after staining with propidium iodide.Flow cytometry data was analyzed with FlowJo 9.3 software (Tree Star Inc, Ashland, OR, USA).
For in vitro fluorescence microscopy, 1 × 10 5 untransfected or ARTC2-transfected DC27.10 cells were stained with either s+16a 680 , Nika102 680 or control antibodies as described above.Cells were suspended in a volume of 0.2 mL of PBS and centrifuged (CytoSpin, Shandon, Pittsburgh, PA, USA) onto microscope slides at 800 rpm for 5 min.Cells were fixed in acetone for 10 min, washed twice in PBS, and mounted with Mowiol-DAPI (Molecular Probes, Carlsbad, CA, USA).After air-drying for 24 h, slides were analysed with an inverted microscope (Axiovert 200 m, Zeiss, Goettingen, Germany) with excitation of 665 nm and emission of 725 nm.Images were analyzed with ImageJ software (NIH, Bethesda, Maryland, USA).

NIRF imaging in vivo
Prior to NIRF imaging in vivo, 8-10-week-old mice were kept on an alfalfa-free diet for 7 days to reduce autofluorescence of the intestine.For generation of tumor xenografts of comparable size, mice were subcutaneously injected at their back on the right side with 1.5 × 10 6 ARTC2-transfected cells and on the left side with 0.5 × 10 6 untransfected cells in a mix of 0. For NIRF imaging in vivo, mice were anesthetized with isofluorane and positioned in the imaging chamber of the small-animal NIRF-imaging system using the filter settings as described above.After qualitative imaging in vivo, quantitative analyses were performed by placing ROIs around the ARTC2-positive tumors, the ARTC2-negative tumors (negative control) and the hind limb (background signal).Even though the signal of the hind limb might underestimate the systemic background, we chose the hind limb because this tissue introduces less variation than abdominal organs or blood.
Total radiant efficiency was determined with Living Image 4.2 software (Caliper Life Sciences).Tumor-to-background ratio was calculated by dividing the tumor uptake value by the background value determined from the hind limb.

Ex vivo analyses
For ex vivo validation of in vivo measurements, animals were sacrificed 6 h or 24 h post-injection.ARTC2-positive and ARTC2negative tumors and organs (spleen, lungs, liver, kidneys, stomach, ileum and muscle) were dissected.Biodistribution analysis was performed using NIRF imaging ex vivo as described by Bannas et al. (11).Total radiant efficiency of organs and tumors was determined and organ-and tumor-to-background ratio was calculated by dividing the tumor uptake value by the background value determined from explanted muscle tissue.Even though the signal of muscle might underestimate the systemic background, we chose muscle because this tissue introduces less variation than abdominal organs or blood.
For ex vivo fluorescence microscopy, one half of each harvested tumor was fixed in 4% paraformaldehyde over night, placed in 30% sucrose for 24 h and frozen on dry ice in Tissue-Tek® OCT™ (Sakura Finetek, Alphen, The Netherlands).Sections of 8 μm were prepared using a Reichert-Jung Ultracut microtome (Reichert-Jung, Wien, Austria).Tumor cryosections were stained with DAPI (Molecular Probes, Carlsbad, CA, USA) to visualize nuclei and CD31 (M-20, Santa Cruz, Heidelberg, Germany) to visualize vessels.Fluorescence microscopy analysis was performed using a Leica TCS SP5 confocal microscope (Leica Camera AG, Solms, Germany).A HeNe 633 nm laser was used for excitation of AlexaFluor680.Image analysis was performed using Leica LAS (Leica) and ImageJ software (NIH, USA).
For flow cytometry ex vivo, the other half of each tumor was dissected and passed through a 70 μm cell strainer to obtain single-cell suspensions.Dispersed cells were counterstained with pan leukocyte marker anti-CD45 (Clone 30-F1, eBioscience, San Diego, CA, USA) and Pacific Orange-NHS dye (Invitrogen, Grand Island, NY, USA) for live/dead staining.Labeling efficiency of ARTC2 by the injected AF680-conjugates was quantified by flow cytometry.For quantification of free circulating AF680conjugates, serum was collected and used in a volume of 100 μL (1:100 dilution) for labeling 1 × 10 6 ARTC2-expressing DC27.10 cells with subsequent flow cytometry.

Statistical analysis
Data are presented as mean ± SD.Statistical analysis was performed using two-way ANOVA with Bonferroni's post-test analyses to evaluate differences between the two independent variables from experiments presented in Figs. 4 and 5. One-way ANOVA with Bonferroni's post-test analyses was performed to evaluate the significance of differences between four groups shown in Figs. 6 and 7A.P < 0.05 indicates statistical significance.Statistical analysis was performed using Prism 5, Graph Pad Software and Excel, Microsoft.

Figure 1 .
Figure 1.Structure, purity, stability and competition study of ARTC2-specific AF680-conjugates.(A) Scheme of nanobody s+16a and mAb Nika102.(B) Coomassie-stained gel overlaid by a corresponding NIRF image of unconjugated s+16a and Nika102 (lanes 1 and 3) and respective AF680-conjugates (lanes 2 and 4) (1 μg per lane of any given conjugate).(C) To asses the stability of the AF680-conjugates, conjugates were serially diluted and incubated for 24 h either at 4°C in PBS or at 37°C in serum.Conjugates were then used to stain ARTC2-expressing lymphoma cells before analysis by flow cytometry.Fluorescence intensities of the mean ± SD from three independent experiments are plotted.(D) ARTC2-expressing lymphoma cells were pretreated with PBS, unlabeled s+16a or Nika102 before staining with fluorochrome-conjugated Nika102, s+16a or isotype control antibodies and analysis by flow cytometry.

Figure 2 .
Figure 2. Imaging of AF680-conjugates in vitro.Untransfected (À) and ARTC2-transfected (+) DC27.10 lymphoma cells were incubated with s+16a 680 and Nika102 680 .(A) One aliquot of labeled cells (1 × 10 6 ) was subjected to flow cytometry to quantify cell-bound AF680 conjugates.Mean fluorescence intensity of ARTC2 expression on lymphoma cells is plotted.Numbers indicate mean ± SD fluorescence intensity of ARTC2-positive cells (grey histograms) from three independent experiments.Unfilled histograms show isotype controls.(B) Another aliquot of labeled cells (1 × 10 5 ) was used for fluorescence microscopic analysis of specific ARTC2 labeling (red).Nuclei were counter-stained with DAPI (blue).The size bar indicates 10 μm.(C) A third aliquot of cells (1 × 10 7 ) was transferred onto a 96-well plate for NIRF imaging in vitro.Results are representative of three independent experiments.

Figure 3 .Figure 4 .
Figure 3.Comparison of different doses and time points for specific NIRF imaging in vivo.Seven to nine days after subcutaneous implantation of ARTC2-positive (+) and ARTC2-negative (-) lymphomas, s+16a 680 (A, B) and Nika102 680 (C, D) were injected intravenously at a dose of 10 μg (A, C) or

Figure 5 .
Figure 5. Optimization of probe concentration and NIRF-imaging time point for maximal T/B ratio of s+16a 680 and Nika102 680 in vivo.T/B ratios obtained from NIRF imaging of ARTC2-positive tumors are plotted as a function of time for mice injected with either 10 μg (white columns) or 50 μg (grey columns) of (A) s+16a 680 and (B) Nika102 680 , respectively.Fifty micrograms of s+16a 680 showed a higher T/B ratio than 10 μg, which was significant 4-6 h post-injection.Nika102 680 showed the opposite behavior, the lower dose yielded a higher T/B ratio than the higher dose, which became significant after 24 h.Data are presented as mean ± SD from three to seven independent experiments.Levels of statistical significance are indicated by asterisks (ns = p > 0.05, * = p < 0.05, ** = p < 0.01, *** = p < 0.001).

Figure 6 .
Figure 6.NIRF imaging ex vivo.ARTC2-positive (+) and ARTC2-negative (À) tumors of mice injected with 50 μg s+16a 680 or with 10 μg Nika102 680 were explanted 6 h and 24 h post-injection in order to quantitate tumor-associated fluorescence and T/B ratios using NIRF imaging in the absence of potentially confounding signals from other tissues.(A) Radiant efficiencies of tumors and (B) T/B ratios are presented as mean ± SD from five independent experiments.

Figure 7 .
Figure 7. Biodistribution analyses.At 6 h and 24 h post-injection of 50 μg s+16a 680 or 10 μg Nika102 680 mice were sacrificed and tumors and organs were explanted.(A) Organ-and tumor-associated fluorescence was quantified and (B) organ-and tumor-to-background ratios were calculated using NIRF imaging in the absence of confounding signals from other tissues.Signals from explanted muscle tissue served as background value to calculate ratios.Radiant efficiencies of organs and tumors and organ-and tumor-to-background ratios are presented as mean ± SD from five independent experiments.

Figure 8 .
Figure 8. Flow cytometric analyses of circulating and cell bound s+16a 680 and Nika102 680 .(A) In order to determine a cause for the high background signals and the unspecific signal of ARTC2-negative tumors post-injection of Nika102 680 , we analyzed serum to monitor levels of circulating unbound AF680-conjugates.Mice bearing ARTC2-negative and ARTC2-positive tumors were injected intravenously with 50 μg s+16a 680 and 10 μg Nika102 680 and sacrificed 6 h and 24 h after injection.Serum at a dilution of 1:100 was used to label ARTC2-transfected lymphoma cells for flow cytometric quantification of circulating intact AF680 conjugates.Fluorescence intensities of the mean ± SD from three independent experiments are plotted.Levels of statistical significance are indicated by asterisks (*** = p < 0.001).(B) To determine the level of injected conjugates specifically bound to tumor cells both ARTC2-positive and ARTC2-negative tumors were dissected from the same animals.Single-cell suspensions were counterstained with anti-CD45 and analyzed by flow cytometry to quantify the amount of cell-bound AF680 conjugates.Means and standard deviations of fluorescence intensities (MFI) from three independent experiments are plotted.

Figure 9 .
Figure 9. Fluorescence microscopy ex vivo of ARTC2-positive and ARTC2-negative tumors.Confocal fluorescence microscopy (40×) of tumor cryosections (A) 6 h and (B) 24 h post-injection of 50 μg s+16a 680 and 10 μg Nika102 680 , respectively.Signal intensities of AF680-conjugates are displayed in red and demonstrate the distribution within the tumor 6 h and 24 h after injection.Nuclei were counter-stained ex vivo with DAPI (blue) and blood vessels were stained with anti-CD31 488 (green).Nanobody s+16a 680 revealed homogeneous and specific labeling of ARTC2-positive tumors readily within 6 h, whereas Nika102 680 showed weak staining after 6 h, which increased only after 24 h.At both imaging time points Nika102 680 also showed unspecific staining within or close to the tumor vasculature (arrow) of both ARTC2-negative and positive tumors.The size bar indicates 50 μm.These results are representative of three independent experiments.
P. BANNAS ET AL.Images were analysed semi-quantitatively by placing a region of interest (ROI) on individual wells.Total radiant efficiency was calculated with Living Image 4.2 software (Caliper Life Sciences).
1 mL RPMI medium and 0.1 mL Matrigel (BD Biosciences, Becton Dickinson, Franklin Lakes, USA).The different number of cells has been used to take into account the different growth rates of transfected and untransfected cells.After 7-9 days, when tumors reached ~8 mm in diameter, nanobody s+16a 680 or conventional antibody Nika102 680 was injected intravenously via the tail vein in a volume of 200 μL.NIRF imaging was performed before injection and 1, 2, 4, 6, 8, 12, and 24 h after injection.Initial studies showed that earlier time points did not yield useful diagnostic information due to high unspecific signals and were therefore not assessed.Experiments were performed with four doses of each AF680-conjugate: 5 μg, 10 μg, 25 μg, and 50 μg.This corresponded to 0.1 μg, 0.2 μg, 0.45 μg and 0.9 μg of injected dye for nanobody s+16a and 0.06 μg, 0.12 μg, 0.3 μg, and 0.6 μg of injected dye for monoclonal antibody Nika102.