Necrosis Binding of Ac-Lys0(IRDye800CW)-Tyr3-octreotate: Pros and Cons of Using Cyanine-labeled Small Molecules

There is a growing body of nuclear contrast agents that are repurposed for uorescence guided surgery. New contrast agents are obtained by substituting the radioactive tag with, or adding a uorescent cyanine to the molecular structure. This enables intra-operative uorescent detection of cancerous tissue, leading to more complete tumor resection. However, these uorescent cyanines can have a remarkable inuence on pharmacokinetics and tumor uptake. Here we demonstrate the effect of cyanine-mediated dead cell binding of Ac-Lys 0 (IRDye800CW)-Tyr 3 -octreotate (800CW-TATE) and how this can be used as an advantage for the purpose of uorescence guided surgery. We exposed cultured U2OS cells (alive or dead, with or without SSTR 2 -expression) and frozen human tumor tissue sections of NCI-H69 (SSTR 2 positive) and CH-157MN (SSTR 2 negative) to 800CW-TATE. We then injected 800CW-TATE into NCI-H69-tumor-bearing mice. Blocking experiment were included with DOTA 0 -Tyr 3 -octreotate (DOTA-TATE). Paran sections of the resected tumors were imaged for near infrared uorescence and cell death staining was performed.


Results
Binding of 800CW-TATE could be blocked with DOTA-TATE and was absent in SSTR 2 negative cells.
However, strong binding was observed to dead cells, which could not be blocked with DOTA-TATE and was also present in dead SSTR 2 negative cells. No SSTR 2 -mediated binding was observed in frozen sections, possibly due to disruption of the cells in the process of sectioning the tissue before exposure to the contrast agent. DOTA-TATE blocking resulted in incomplete reduction of 61.5 ± 5.8% uorescence uptake by the tumors in the mice. Para n sections revealed that uorescence uptake persisted in necrotic regions upon blocking.

Conclusion
This study shows that labeling peptides with cyanines can result in dead cell binding. This does not hamper the ultimate purpose of uorescence guided surgery, as necrotic tissue appears in most solid tumors. Hence, the necrosis binding can increase the overall tumor uptake. Moreover, necrotic tissue should be removed as much as possible: it cannot be salvaged, causes in ammation, and is tumorigenic. However, when performing binding experiments to cells with disrupted membrane integrity, which is routinely done with nuclear probes, this dead cell binding can resemble non-speci c binding. This study will bene t development uorescent contrast agents.

Page 4/12
There is a growing interest in development of optical molecular contrast agents for the purpose of uorescence-guided surgery [1,2]. In most cases, successful targeting vectors from nuclear imaging agents are repurposed, either by replacing the radiolabeling part by a uorescent tag for single modal imaging, or by adding both a radioactive and a uorescent chemical moiety for bi-modal imaging. These approaches have already yielded ample probes, with some explored in clinical trials [3,4]. Most of these tracers are based on antibodies, which are relatively big molecules and therefore can carry the molecular burden of both tags without affecting the epitope a nity [1,5]. Currently, a growing body of peptides and small molecules are advanced as targeting vectors for optical imaging. However, uorescent tags can have a drastic in uence on the pharmacokinetic properties of smaller vectors due to their relatively big size, electrostatic properties, solubility, and binding to serum proteins. However, these effects are di cult to generalize [6][7][8].
Tyr 3 -octreotate [10] serves as a targeting vector, able to bind to the somatostatin receptor subtype 2 (SSTR 2 ), which is commonly upregulated in neuroendocrine tumors (NET) [11], such as meningioma's [12]. Tyr 3 -octreotate conjugated to DOTA and labelled with gallium-68 or lutetium-177 is routinely used for diagnosis and radionuclide therapy of NET patients, respectively [13][14][15]. In our previous work we found that cyanine dyes such as IRDye800CW speci cally bind to cells that lost membrane integrity and thus, to necrotic regions in tumors [16][17][18][19]. The presence of necrotic regions is common in fast growing tumors and necrosis binding of the dye can lead to unexpected results when studying optical probes.
The aim of this study is to investigate the effect of the IRDye800CW-mediated dead cell binding on physiological properties of 800CW-TATE. These dead cell binding properties should be kept in mind when developing novel cyanine-labeled contrast agents because these can interfere with results from routine experiments.

Materials And Methods
In vitro dead cell binding Human osteosarcoma cells (U2OS), either transfected with SSTR 2 [20] or wild-type (without SSTR 2 expression) were cultured in DMEM (Gibco Life Technologies, Waltham, MA, USA) with 10% fetal calf serum and 1% penicillin. The cells were seeded in 24 well plates (10 5 cells per well) and grown until con uent. The cells were kept alive or killed with 50 µL EtOH (70%) and incubated with 800CW-TATE (10 nM, piCHEM, Grambach, Austria) in culturing medium for 30 min. A blocking experiment was included by co-incubation of 800CW-TATE (10 nM) with DOTA-TATE (10 µM, Bachem, Bubendorf, Switzerland). Control experiments were included by incubation of the cells with a NIR-uorescent dye that does not bear a cyanine motive: Rhodamine 800 (10 nM, Sigma-Aldrich). Additionally, wells without seeded cells were exposed to PBS or EtOH and then incubated with 800CW-TATE, this to investigate non-speci c binding to plastic. After gentle washing with PBS, NIR-uorescence imaging of the whole plate was performed on an Odyssey atbed scanner (Li-Cor, Lincoln, USA). 800CW-TATE binding was based on uorescent signal ± SD and each experiment was performed in n = 12.
In vitro dead cell binding [ 111 In]In-DOTA-TATE U2OS cells, either transfected with SSTR 2 [20] or wild-type were seeded in 24 well plates (10 5 cells per well) and grown until con uent. The cells were kept alive or killed with 50 µL EtOH (70%) and incubated with [ 111 In]In-DOTA-TATE (10 nM, 50 MBq/nmol) in culturing medium for 30 min. A blocking experiment was included by co-incubation with DOTA-TATE (10 µM). After gentle washing with PBS, autoradiography of the whole plate was performed using a super resolution phosphor screen and a Cyclone® Plus system (Perkin Elmer, Waltham, MA, USA). Then, the cells were detached with NaOH (1 M, 1 mL) and collected in tubes for γ-counting (1480 γ-counter Perkin Elmer, Waltham, MA, USA). Each experiment was performed in n = 6. Microscopy U2OS cells, either transfected with SSTR 2 [20] or wild-type were seeded on 10 mm coverslips in 24 well plates (10 5 cells per well). The cells were kept alive or killed with 50 µL EtOH (70%) and incubated in culturing medium with or without 800CW-TATE (100 nM) for 30 min. After PBS washing, the cells were incubated at 37˚C with Calcein AM (1 µM) and Propidium iodide (1 µM) in PBS for 15 minutes to stain live and dead cells, respectively. The cells were washed with PBS, xated with PFA (4% in PBS, 20 min) and nuclei were stained with DAPI (Sigma-Aldrich, 1:1000 in PBS, 5 min). After nal washing with PBS, the coverslips were removed from the well plate and mounted on a microscope slide for uorescence microscope imaging using a magni cation of 20x.

Ex vivo experiments
Frozen NCI-H69 and CH-157MN xenografts were sectioned in 10 µm thick slices, which were mounted on Starfrost glass slides (Thermo Fisher). To prevent non-speci c binding, fresh sections were incubated for 10 min at room temperature (RT) with washing buffer (167 mM Tris-HCl, 5 mM MgCl 2 ) with bovine serum albumin (BSA, 0.25 gram/L). Slides were drained and incubated for 1 hour at RT with a xed volume of 100 µL washing buffer, containing 800CW-TATE (100 nM or 1 µM) and BSA (1 gram/L). To determine speci city, additional sections were co-incubated with increasing concentrations of DOTA-TATE block (0x, 1x, 5x, 10x, 50x, 100x, 250x, 500x, or 1000x the concentration of 800CW-TATE). The sections were drained, washed and after drying, placed in an Odyssey CLx (Li-cor Biosciences) for NIR-uorescence detection. Subsequently, H&E staining was performed on the sections, based on which regions of interests (ROIs) were drawn on the viable sections of the tumors for quanti cation of the NIR-uorescence signal using ImageJ.
When the tumors reached a size of 400 mm 3 , 800CW-TATE (3 µg, 1.36 nmol, 50 µL saline) was injected retro-orbitally. A blocking study was included, for which a mouse received two injections of DOTA-TATE (3 mg, 2.2 µmol, 2x50 µL saline) followed by one injection with 800CW-TATE (3 µg, 1.36 nmol, 50 µL saline), with 5 min intervals. Four-hour post injection, the mice were sacri ced, after which tumors were collected and embedded in para n for histological analysis. Adjacent para n sections of 4 µm were prepared of the tumors. The sections were imaged for NIR-uorescence on an Odyssey. Immuno uorescence (IF) staining was performed for SSTR 2 (SSTR 2 primary antibody, Abcam; 1:100; secondary antibody goat antirabbit Alexa 596, Abcam, 1:1000). Histochemical TUNEL dead cell staining was performed using the In Situ Cell Death Detection Kit, Fluorescein (Roche Diagnostics GmbH, Mannheim, Germany). Stitched uorescence microscope images of the stained sections were acquired on a uorescence microscope (TriPath-Imaging, Burlington, USA) equipped with an Aziocam MRm (Zeiss, Oberkochen, Germany) and an Axiocam 208 color (Zeiss) using a magni cation of 40x.
The Institutional Review Board of the University of Groningen approved the study and animal care complied with the Guide for the Care and Use of Laboratory Animals.

Results
In vitro binding of 800CW-TATE on plated cells was demonstrated (Fig. 2). The four wells on the left contain U2OS cells with stable expression of SSTR 2 whereas the four wells on the right contain the wildtype U2OS cells which do not express SSTR 2 . The cells were either killed or kept alive before incubation with either 800CW-TATE or 800CW-TATE in presence of an excess of DOTA-TATE. It is apparent that there was SSTR 2 -mediated binding of 800CW-TATE, as it was effectively blocked by DOTA-TATE. Striking was the strength of the signal observed from the dead cells. It was not mediated by SSTR 2 since the signal was not blocked by the excess DOTA-TATE. Similarly, uorescence was observed from the dead cells without SSTR 2 -expression. The dead cell binding was signi cant as it was double as high as SSTR 2mediated binding (206%, P < 0.001). In contrast, [ 111 In]In-DOTA-TATE did not showcase dead cell mediated binding (Supplementary data 1.1). No 800CW-TATE binding was observed to ethanol exposed wells without cells (Supplementary data 1.2), therefore excluding binding to the ethanol-exposed plastic of the wells. Moreover, dead and alive cells with and without SSTR 2 -expression were exposed to Rhodamine 800, a NIR-uorescent dye which does not bear a cyanine motive. There was no binding observed of Rhodamine 800, indicating that this dead cell binding is mediated via the cyanine motive of IRDye800CW (Supplementary data 1.3). Further con rmation of 800CW-TATE binding to dead cells was obtained from microscopy images. Cells incubated in culturing medium without 800CW-TATE showcased no uorescence, therefore indicating that the NIR-uorescent signal is not caused by auto uorescence from dead cells (Supplementary data 2).
The NIR-uorescence uptake in the frozen tumor sections exposed to 800CW-TATE was heterogeneous.
On the basis of H&E staining viable regions of the tumors were selected to draw ROIs (Supplementary data 3.1). It was observed that uorescent signal was highest in disrupted regions of the tissue sections. Nonetheless, the results from ex vivo experiments on NCI-H69 and CH-157MN cryosections were not conclusive. NCI-H69 tumors have well established expression of SST 2 whereas CH-157MN are SST 2negative. However, co-incubation with DOTA-TATE did not block the uorescence signal from 800CW-TATE in the NCI-H69 sections. Moreover, the CH-157MN sections showcased similar uptake intensities, thereby further con rming that the uorescence binding to the sections is not SST 2 -mediated (Fig. 3).
This nding is not in line with the results from the in vitro experiment in Fig. 2, where SST 2 -mediated binding is effectively blocked from live, SST 2 -expressing cells with DOTA-TATE. Quanti cation of the mean uorescent uptake is shown in Supplementary data 3.
800CW-TATE was studied as a novel contrast agent for molecular uorescence-guided surgery (MFGS) on NCI-H69 tumor xenograft bearing mice. In this model the tracer was able to detect tumor tissue for MFGS [9]. To verify SSTR 2 speci c binding to the tumor, a blocking experiment was performed by preadministration of DOTA-TATE, thereby outcompeting 800CW-TATE for binding to SSTR 2 . It was observed, however, that uorescence was only partially reduced by DOTA-TATE blocking (reduction of 57.8 ± 5.2%).
Moreover, after dissection of the tumors, a heterogeneous uptake of the uorescent compound was observed in para n embedded sections (Fig. 4, red). Adjacent sections were subjected to H&E staining (Fig. 4), uorescently stained for cell death using TUNEL staining (Fig. 4, green), or for SSTR 2 -expression (Supplementary data 4). The top and bottom sections in Fig. 4 represent a tumor from a mouse injected with 800CW-TATE and from the blocking experiment, respectively. A strong overlap of the NIRuorescence uptake and the necrotic regions was observed in these tumors. Similar to what we observed with cultured cells in Fig. 2, DOTA-TATE blocked the signal from viable cells, as the NIR-uorescence signal in the viable regions of the tumors was reduced by DOTA-TATE blocking.

Discussion
In the current paper we have shown that the necrosis avidity of cyanine-dyes such as IRDye800CW, can drastically contribute to the binding behavior of targeting vectors conjugated to cyanines, intended for MFGS. This has some disadvantages and advantages. One disadvantage is that these probes cannot be used in routine protocols frequently used for radionuclide probes to determine speci c binding of the probe directly on tissue sections. We showed in this study that these experiments resulted in high amounts of unspeci c binding, which can be explained by the fact that in the process of cutting the tissue sections, the cellular membranes in the tissue are compromised. We showed in an earlier publication that cyanine dyes, such as IRDye800CW, bind to endoplasmic proteins that become available when cells loose membrane integrity [16].
When 800CW-TATE is injected systemically in SSTR 2 -expressing tumor bearing mice, it will bind to SSTR 2 via the targeting octreotate peptide, but also to necrotic tissue via IRDye800CW. This was also observed by us in an earlier study using 800CW-2DG and 800CW-EGF, where the targeting molecules bind to their speci c target and IRDye800CW to necrosis [16]. 800CW-EGF has a molecular weight of around 7 kD and could still bind to necrosis. IRDye800CW conjugated to a 25 kD polyethyleneglycol (PEG)-chain on the other hand, could not bind to necrosis, most probably due to steric hindering. This suggests that small molecules or peptides conjugated to IRDye800CW with a molecular weight < 7 kD still retain necrosis avidity [16]. This also explains in vitro the presence of speci c 800CW-TATE signal in live SSTR 2 expressing U2OS cells and a non-speci c signal in dead cells (Fig. 2). The latter was con rmed by the fact that the signal could not be blocked by excess DOTA-TATE and was present in both SSTR 2expressing and non-expressing dead cells. Moreover, in tissue sections of SSTR 2 -positive NCI-H69 tumorbearing mice treated with 800CW-TATE (Fig. 4) the NIR-uorescence signal (red) co-localized (yelloworange) in a great part with the TUNEL signal (green) indicating dead cells. When an excess of DOTA-TATE was co-administered, the speci c signal disappeared but the necrosis speci c signal remained.
These ndings could also explain retrospectively our previous unexpected results in vivo using a dual labeled octreotate peptide. This peptide was attached to both DTPA as a chelator for indium-111 labeling and the uorescent cyanine dye Cy5 [21]. We observed that the in vitro a nity of the dual labeled octreotate dropped by more than a logarithmic unit compared with the peptide only labeled with indium-111. Unexpectedly though, there was no signi cant difference in overall tumor-uptake in vivo between the dual labeled peptide and the radioactive peptide. This might be due to necrosis binding of the cyanine moiety, which added up for the uptake in the tumor and therefore compensated for the reduced a nity for the target protein.
Most solid tumors develop necrotic tissue, especially in the center of the tumor appearing as a necrotic core [22,23]. This is due to improper functioning of immature tumor blood vessels and rapid tumor growth leading to insu cient oxygen supply in the center followed by tissue necrosis [24]. This is furthermore correlated with poor prognosis of the disease [25]. Moreover, we have shown that imaging of necrosis can be used for early determination of therapy e cacy of irradiation therapy [17] or chemotherapy [18]. For this reason, we recently developed and validated [ 111 In]In-DOTA-PEG 4 -800CW, as a speci c probe to image tumor necrosis [19].
What could be the advantage for MFGS of using a small molecule or peptide probe that is conjugated with a cyanine dye like IRDye800CW that also targets necrosis? As mentioned before, most solid tumors will develop a necrotic core or necrotic regions at a certain time point depending on the growth rate [24]. The ultimate goal of MFGS is to identify tumor tissue and distinguish it from normal heathy tissue during surgery. Apart from binding to a speci c tumor target, binding to necrosis which is mostly found in the center of the tumor, will not interfere with the speci c binding and will actually strongly contribute to the total tumor to background ratio and the intensity of the signal. Moreover, any necrotic tissue should be removed as much as possible as it cannot be salvaged, causes in ammatory responses, and even releases tumorigenic factors [22,26].

Conclusion
In conclusion, one has to be aware that tumor targeting molecules or peptides conjugated to a cyanine dye like IRDye800CW will also bind to necrosis, that is, intracellular proteins of necrotic cells that become available when cells have lost membrane integrity. This will prevent its use to determine speci c binding of the tumor targeted probe directly on tissue sections because intracellular proteins will be ubiquitously exposed. When using these tumor targeted probes in vivo to determine the presence of the speci c target, e.g. using NIR-uorescence imaging or afterwards in histological sections, one needs to be aware of possible binding to necrotic tissue. Despite these disadvantages, a great advantage for MFGS could be that binding to its speci c target and necrosis will increase the tumor uptake, contributing to e ciency of MFGS.