A Truncated 14-Amino-Acid Myelin Protein-Zero-Targeting Peptide for Fluorescence-Guided Nerve-Preserving Surgery

Background: The occurrence of accidental nerve damage during surgery and the increasing application of image guidance during head-and-neck surgery have highlighted the need for molecular targeted nerve-sparing interventions. The implementation of such interventions relies on the availability of nerve-specific tracers. In this paper, we describe the development of a truncated peptide that has an optimized affinity for protein zero (P0), the most abundant protein in myelin. Methods and Materials: Further C- and N-terminal truncation was performed on the lead peptide Cy5-P0101–125. The resulting nine Cy5-labelled peptides were characterized based on their photophysical properties, P0 affinity, and in vitro staining. These characterizations were combined with evaluation of the crystal structure of P0, which resulted in the selection of the optimized tracer Cy5-P0112–125. A near-infrared Cy7-functionalized derivative (Cy7-P0112–125) was used to perform an initial evaluation of fluorescence-guided surgery in a porcine model. Results: Methodological truncation of the 26-amino-acid lead compound Cy5-P0101–125 resulted in a size reduction of 53.8% for the optimized peptide Cy5-P0112–125. The peptide design and the 1.5-fold affinity gain obtained after truncation could be linked to interactions observed in the crystal structure of the extracellular portion of P0. The near-infrared analogue Cy7-P0112–125 supported nerve illumination during fluorescence-guided surgery in the head-and-neck region in a porcine model. Conclusions: Methodological truncation yielded a second-generation P0-specific peptide. Initial surgical evaluation suggests that the peptide can support molecular targeted nerve imaging.


Introduction
Image-guided applications play an important role in the current progression realized in the field of head-and-neck surgery [1][2][3][4]. A prime example of such an approach is found in the surgical resection of sentinel nodes, a procedure that serves as a secondary means for the identification of (micro)metastatic tumor spread [1,2,5,6]. An upcoming application is found in the visualization of primary tumor margins [7][8][9][10][11]. Unfortunately, the location of both possibly involved lymph nodes and the primary tumor can coincide with the location of nerves within the complex anatomy of the head and neck area (e.g., the hypoglossal, lingual, and vagus nerves and the marginal mandibular trunk of the facial nerve; [12,13]). As a result, the occurrence of accidental surgically induced nerve damage is not uncommon. For instance, injury to the laryngeal nerve or mandibular nerve is seen in 14% of patients undergoing thyroid surgery or neck dissection, respectively [14,15].

Materials and Methods
All chemicals were received from Actu-All Chemical (Oss, The Netherlands), Sigma Aldrich (St. Louis, MO, USA), Tokyo Chemical Industry (Tokyo, Japan), Biosolve BV (Valkenswaard, The Netherlands), and VWR Chemicals (Solon, OH, USA) and used without further purification. DMF and DMSO were dried over 4 Å molecular sieves for at least 24 h prior to use. Preparative high-pressure liquid chromatography (prep-HPLC) was performed on a Waters HPLC system (Waters Chromatography B.V., Etten-Leur, The Netherlands) using a 2545 quaternary gradient module pump and a 2489 UV detector. A Dr. Maisch ReproSil-Pur 120 C18-AQ 10 µM (250 mm × 20 mm) column (Dr. Maisch HPLC GmbH, Ammerbuch-Entringen, Germany) was applied with an operating flow rate of 12 mL/min. Analytical HPLC was performed on a Waters Acquity UPLC-MS system using a Acquity UPLC photodiode array detector, an SQ Detector mass spectrometer, and a flow rate of 0.5 mL/min (Waters BEH C18 130 Å 1.7 mm (100 mm × 2.1 mm) column). Lyophilization was performed using a VaCo 2-II lyophilizer (Zirbus technology GmbH, Bad Grund (Harz), Germany). Absorption spectrometry was performed using a UV1280 UV-vis spectrometer (Shimadzu, Kyoto, Japan), and fluorescence spectrometry was performed using an LS55 (Perkin Elmer, Waltham, MA, USA). Fluorescence confocal imaging was performed using an SP8 WLL confocal microscope (Leica Microsystems, Wetzlar, Germany). The obtained images were analyzed using Leica Confocal Software (Leica Microsystems). The mean fluorescence intensity values per sample were measured using an LSRII flow cytometer (BD Biosciences, Franklin Lakes, NJ, USA) with APC-A settings (635 nm laser and 750 nm long-pass filter) for a Cy5 dye. Image acquisition and processing for the assessment of P0 staining were performed using LASX software (Leica Application Software Suite 4.8). Image analysis (3D) was performed using Image J [26]. For the affinity assessment, fluorescence was measured using a FACSCanto II flow cytometry device (BD Biosciences, Franklin Lakes, NJ, USA) in the APC-A channel, and the accompanying FlowJo TM v5 software (BD Biosciences, Franklin Lakes, NJ, USA) was used for further analysis. Figure 1 was created using BioRender (https://biorender.com/, accessed on 30 May 2023). Feature clustering and Pearson correlation assessment were performed using MATLAB 9.9.0.1495850 (R2020b) software (The MathWorks Inc.: Natick, MA, USA). For crystal structure analysis, RCSB PDB (https://www.rcsb.org/) was used.

Chemical and Photophysical Properties
Determination of the chemical and photophysical properties (LogP, solubility, serum binding, net charge, and brightness) was performed as previously described [27,31,36].

Correlation Assessment and P0-Related Interactions
Cluster analysis was applied based on standardized values for each feature using the clustergram function in MATLAB [37]. An evaluation of P0-related interactions was performed using peptide characteristic and inter-and intra-molecular interactions that were extracted from the crystal structure of the extracellular portion of P0 [38].

In Vivo Imaging of the Optimized Peptide
In vivo experiments in pigs were approved by the ethical board of the University of Ghent (EC2019/79). Pigs were housed at the animal facility at ORSI Academy (Melle, Belgium) until use for nerve imaging experiments during surgical training (weight per animal: approximately 40 kg). Experiments were performed in accordance with the Experiments on Animals Act (Wod, 2014) and the applicable legislation in Belgium, and in accordance with the European guidelines (EU directive no. 2010/63/EU) regarding the protection of animals used for scientific purposes. The pigs were bred and kept in accordance with Belgium law in and by a licensed establishment for the use of experimental animals.
Cy7-P0 112-125 (0.25 mg, 300 µmol) was evaluated in pigs undergoing open surgery in the head and neck area using a clinical-grade handheld fluorescence camera (N = 2, PDE-mod, Hamamatsu Photonics; [39]). Animals received either intravenous injection (in the jugular vein under ultrasound guidance) or injection directly into the vagus nerve (intraneural administration). In both cases, fluorescence imaging of the vagus nerve and surrounding tissues was performed at 1 h after tracer administration. The animals were maintained under isoflurane anesthesia for the complete duration of the surgical training and subsequent nerve imaging experiments and were euthanized before awakening from the anesthesia. After resection, ex vivo imaging was applied using the same camera system. Image processing was performed using in-house custom-developed software [27,40].

Statistical Evaluation
A statistical evaluation to compare the affinity (K D ) was performed using unpaired Student's t-test. Values of p < 0.05 were considered significant. Pearson correlation analysis was applied based on standardized feature values according to previously described methods [37].

P0 Staining by Fluorescence Confocal Microscopy
In line with previous reports, the staining of Cy5-P0 101-125 in P0-expressing RT4 Schwannoma cells [27] was in agreement with the location of P0 on the membrane (P0related staining in red; Figure 2) and, more specifically, on the cellular outgrowths (white arrows) of the Schwannoma cells.

P0 Staining by Fluorescence Confocal Microscopy
In line with previous reports, the staining of Cy5-P0101-125 in P0-expressing RT4 Schwannoma cells [27] was in agreement with the location of P0 on the membrane (P0related staining in red; Figure 2) and, more specifically, on the cellular outgrowths (white arrows) of the Schwannoma cells.

C-Terminal Matrix
A comparison between the compounds within the C-terminal matrix and the lead compound Cy5-P0 101−125 revealed differences in localization, as well as P0-related staining intensity (Figure 2A, in red). The level of staining of the cellular outgrowths was shown to decrease with each truncation step, suggesting that C-terminal truncation resulted in a loss of P0 specificity. The differences in the localization of staining were underlined by surface plot analysis of the fluorescence ( Figure S13A). The loss of staining of the outgrowths was shown to correspond to an increase in the level of co-localization with the intracellular control staining/non-specific staining (Table S1).

P0 Affinity Determined by Flow Cytometry
Peptides within the C-terminal truncation matrix did not show nanomolar binding affinity for P0 (K D ; Figure 3A), which is in line with the lack of specific staining seen in Schwannoma cells (Figure 2A). N-truncation did, however, show a significant increase in P0-related affinity from 175 to 83 (p = 0.04) and 69 nM (p = 0.0001) for Cy5-P0 105-125 , Cy5-P0 108-125 , and Cy5-P0 112-125 , respectively. As an example, a representation of the increase in signal intensity and the resulting saturation binding curve for Cy5-P0 112−125 are provided in Figure 3C,D. Again, the affinity could be related to P0-related staining ( Figure 2B).

Relationships between Chemical and Biological Features
Analysis of the crystal structure of the extracellular portion of P0 [38] allowed an investigation of the inter-and intra-molecular interactions between the parental amino acids in this protein (Table 2). Pearson correlation assessment based on the peptide characteristics and the interactions observed in the crystal structure revealed three different subclasses of compounds (Figure 4, y-axis). In this classification, an intermediate or low number of acceptor and donor molecules and a low number of supramolecular/intermolecular interactions were found to be the main denominators. These structural features showed a direct correlation with the level of affinity (high number of donors/acceptors = high affinity, low number of donors/acceptors = low affinity). This correlation was also seen for the level of solubility (Table 1).  Interestingly, solubility (Table 1) seems to be linked to affinity. Tracers with a low affinity of >1000 nM were soluble, while higher affinity tracers (e.g., Cy5-P0 108-125 and Cy5P0 112-125 ) showed relatively low solubility (Table 1). This unexpected effect might be explained through the intrinsic homotypic interactions between P0 molecules, wherein the levels of inter-and intra-molecular interactions can both affect affinity and solubility.

Relationships between Chemical and Biological Features
Analysis of the crystal structure of the extracellular portion of P0 [38] allowed an investigation of the inter-and intra-molecular interactions between the parental amino acids in this protein ( Table 2). Pearson correlation assessment based on the peptide characteristics and the interactions observed in the crystal structure revealed three different subclasses of compounds (Figure 4, y-axis). In this classification, an intermediate or low number of acceptor and donor molecules and a low number of supramolecular/intermolecular interactions were found to be the main denominators. These structural features showed a direct correlation with the level of affinity (high number of donors/acceptors = high affinity, low number of donors/acceptors = low affinity). This correlation was also seen for the level of solubility (Table 1). Truncation of the original lead compound Cy5-P0 101-125 , a peptide that largely overlaps with the G B-helix in the extracellular portion of the P0 structure ( Figure 5), decreased the number of interactions. This yielded varying effects on the binding affinity of the tracers (Figures 3 and 4). C-terminal truncation of the PTRY 122-125 amino acid sequence resulted in a loss of affinity. This implies that this sequence has a critical function within the peptide. N-terminal truncation of the KNPP 101-104 amino acid sequence yielded a 1.67-fold drop in affinity (p > 0.0003), suggesting that this sequence plays a role in P0 binding, but one that is less critical.
The crystal structure of P0 indicates that N 102 can bind to V 107 , thereby creating a loop within the peptide ( Figure 5). Interestingly, truncation of the DIV 105-107 amino acid sequence again yielded a 2.1-fold rise in affinity (p = 0.0002, Figure 3A). This is especially interesting when one realizes that the P0 crystal structure indicates that the amino acids V 107 and Q 112 facilitate intermolecular van der Waals interactions ( Figure 5; [41]). Further truncation of GKTS1 08-111 only yielded a modest 1.2-fold (p = 0.09) rise in affinity. Removal of the QVTL 112-115 sequence reduced the affinity by nearly 2.13-fold (p = 0.09) and therefore seems to be rather critical for maintaining affinity for P0. Further truncation of YVFE 116-119, a sequence containing a plurality of H-donors and acceptors, again resulted in a complete loss of affinity. Further analysis of the crystal structure suggests that sequence TLYVFE 114-119 provides intramolecular binding for the A' and F B-sheets ( Figure 5 (top right); [41]). the number of interactions. This yielded varying effects on the binding affinity of the tracers (Figures 3 and 4). C-terminal truncation of the PTRY122-125 amino acid sequence resulted in a loss of affinity. This implies that this sequence has a critical function within the peptide. N-terminal truncation of the KNPP101-104 amino acid sequence yielded a 1.67-fold drop in affinity (p > 0.0003), suggesting that this sequence plays a role in P0 binding, but one that is less critical.  (C) Lead and truncated P0-targeting peptides Cy5-P0101-125 and Cy5-P0112-125. Hydrogen donors (blue) and acceptors (red) that support inter-or intra-molecular interactions have been indicated on peptide sequences. The importance of the peptide sequences is color coded, with the most important binders in green and the least important ones in red. F, fluorophore. The crystal structure of P0 indicates that N102 can bind to V107, thereby creating a loop within the peptide ( Figure 5). Interestingly, truncation of the DIV105-107 amino acid sequence again yielded a 2.1-fold rise in affinity (p = 0.0002, Figure 3A). This is especially interesting when one realizes that the P0 crystal structure indicates that the amino acids V107 and Q112 facilitate intermolecular van der Waals interactions ( Figure 5; [41]). Further truncation of GKTS108-111 only yielded a modest 1.2-fold (p = 0.09) rise in affinity. Removal of the QVTL112-115 sequence reduced the affinity by nearly 2.13-fold (p = 0.09) and therefore seems to be rather critical for maintaining affinity for P0. Further truncation of YVFE116-119, a sequence containing a plurality of H-donors and acceptors, again resulted in a complete loss of affinity. Further analysis of the crystal structure suggests that sequence TLYVFE114-119 provides intramolecular binding for the A' and F B-sheets ( Figure 5 (top right); [41]).

In Vivo Imaging of Nerves in the Head and Neck Region
To provide a proof-of-principle of intraoperative real-time near-infrared nerve visualization in an open surgery setting, Cy7-P0 112-125 was synthesized and evaluated in a porcine model.
Intravenous administration of Cy7-P0 112-125 in the jugular vein yielded a fluorescence signal in the vagus nerve ( Figure 6A, black arrow). However, the signal-to-background ratio was low. Intraneural administration substantially improved the staining ( Figure 6A). Ex vivo assessment of excised nerve tissue followed by fluorescence-intensity-based image processing ( Figure 6B) confirmed tracer uptake in the nerve after intravenous administration (SBR: 1.9) and after intraneural administration (SBR: 4.9). cine model.
Intravenous administration of Cy7-P0112-125 in the jugular vein yielded a fluorescence signal in the vagus nerve ( Figure 6A, black arrow). However, the signal-to-background ratio was low. Intraneural administration substantially improved the staining ( Figure 6A). Ex vivo assessment of excised nerve tissue followed by fluorescence-intensity-based image processing ( Figure 6B) confirmed tracer uptake in the nerve after intravenous administration (SBR: 1.9) and after intraneural administration (SBR: 4.9).

Discussion
Methodological peptide truncation was shown to provide a successful optimization route, resulting in the selection of Cy5-P0112-125. This P0-specific peptide showed a 1.5-fold increase in affinity compared to the lead compound Cy5-P0101-125 ( Figure 3; p = 0.01), while a 47% reduction in the number of incorporated amino acids was realized (from 26 (101-125; KNPPDIVGKTSQVTLYVFEKVPTRY) to 14 (112-125; QVTLYVFEKVPTRY), Figure  1). This route not only allowed the generation of a new lead, but also provided increased mechanistic insight into the molecular requirements of P0 binding.
Truncation combined with affinity studies (Figure 3), chemical analysis (Table 1), and P0 crystal structure observations (Figures 3 and 4, and Table 2) gave interesting insights into the (homotypic) peptide binding of the P0 proteins. P0-related affinity seems to be dominated by two primary binding interactions. The first is the apparent anchoring function of the C-terminal VPTRY121-125 sequence [41], a feature that was shown to be critical in the positioning of the P0 proteins, and, in particular, the GKTSQVTLYVFE108-119 pharmacophore, for binding. (Figure 3). The GKTSQVTL108-115 portion of the original lead sequence was shown to have the most prominent effect on P0-related binding. The presence of Q112 in this sequence seemingly facilitates the targeting of V107, while the remaining amino acids provide a supporting function. More specifically, YVFE116-119 on its own did not show affinity for P0, but one can argue that its main role is to help facilitate the positioning of Q112.
Previous work on P0-derived myelin-targeting peptides indicates that this target is specific for the peripheral nervous system [27], reducing the risk of late-term toxic effects due to accumulation in the CNS. This feature is a common side effect for small-molecule nerve agents derived from, e.g., the neural imaging agent Pittsburgh compound B

Discussion
Methodological peptide truncation was shown to provide a successful optimization route, resulting in the selection of Cy5-P0 112-125 . This P0-specific peptide showed a 1.5-fold increase in affinity compared to the lead compound Cy5-P0 101-125 (Figure 3; p = 0.01), while a 47% reduction in the number of incorporated amino acids was realized (from 26 (101-125; KNPPDIVGKTSQVTLYVFEKVPTRY) to 14 (112-125; QVTLYVFEKVPTRY), Figure 1). This route not only allowed the generation of a new lead, but also provided increased mechanistic insight into the molecular requirements of P0 binding.
Truncation combined with affinity studies (Figure 3), chemical analysis (Table 1), and P0 crystal structure observations (Figures 3 and 4, and Table 2) gave interesting insights into the (homotypic) peptide binding of the P0 proteins. P0-related affinity seems to be dominated by two primary binding interactions. The first is the apparent anchoring function of the C-terminal VPTRY 121-125 sequence [41], a feature that was shown to be critical in the positioning of the P0 proteins, and, in particular, the GKTSQVTLYVFE 108-119 pharmacophore, for binding. (Figure 3). The GKTSQVTL 108-115 portion of the original lead sequence was shown to have the most prominent effect on P0-related binding. The presence of Q 112 in this sequence seemingly facilitates the targeting of V 107 , while the remaining amino acids provide a supporting function. More specifically, YVFE 116-119 on its own did not show affinity for P0, but one can argue that its main role is to help facilitate the positioning of Q 112 .
Previous work on P0-derived myelin-targeting peptides indicates that this target is specific for the peripheral nervous system [27], reducing the risk of late-term toxic effects due to accumulation in the CNS. This feature is a common side effect for smallmolecule nerve agents derived from, e.g., the neural imaging agent Pittsburgh compound B [20,22,42,43]. Various studies have put forward local tracer administration as a means to decrease uptake beyond the surgical field [19,27,44]. This strategy is generally employed to help boost the effective local concentration and, as such, facilitate a positive effect on local targeting. In line with our previous work on Cy5-P0 101-125 [27], the porcine studies in Figure 6 indicate that intravenous injection of Cy7-P0 112-125 in the vessel leading to the organ of choice did indeed induce staining of the vagus nerve. As previously shown for other targeted tracers, optimization of the dosing and timing is expected to provide higher SBR values, further adding to the utility of this administration strategy [29]. While intraneural administration yielded explicit nerve staining with a high SBR, longitudinal studies are needed to determine whether this administration route could have negative effects on nerve viability. Surgical visualization of nerves may not only serve head-and-neck surgery, but also help promote nerve sparing in, e.g., neuro-, orthopedic, colorectal, bladder, and prostate surgery. As surgery is moving increasingly towards minimally invasive (robotic) approaches, it is critical that the nerve-specific tracers are compatible with clinical-grade endoscopic camera systems. To this end, the far-red Cy5 dye (l ex = 650 nm, l em = 667 nm) has demonstrated compatibility with a KARL STORZ prototype [40,45], while the nearinfrared Cy7 dye (l ex = 750 nm, l em = 777 nm; [16]) is compatible with the Image 1 S Rubina (KARL STORZ) and the Firefly endoscope+ (Intuitive) set-up. These features will promote further dissemination.

Conclusions
Methodological truncation of the lead compound Cy5-P0 101-125 , combined with assessment of the crystal structure of the extracellular portion of P0 and inter-and intra-molecular binding interactions, resulted in the selection of the optimized tracer Cy5-P0 112-125 . This high-affinity tracer allowed effective P0-related staining in vitro, and its Cy7 derivative was shown to be compatible with clinical-grade fluorescence imaging devices during nerve imaging in a porcine model.

Patents
European patent application No.16180535.3.  Table S1: Semi-quatitative assessment of co-localization staining P0 tracers and lysosomal and nuclear control staining.  Institutional Review Board Statement: The animal study protocol was approved by the Institutional Review Board of the University of Ghent (EC2019/79).

Informed Consent Statement: Not applicable.
Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to patent restrictions.