Intraoperative visualization of nerves using a myelin protein-zero specific fluorescent tracer

Surgery, and science therein, has come a long way since Mr. Gunning and Lord Thurlow, respectively, stated to each other in 1796: “there is no more science in surgery”, in reply “than there is in butchery” [1]. Contradictory to these statements, the rapid translation of innovative minimally invasive surgical technologies has initiated the concept of “precision surgery” (2‐5). This concept is mostly driven by engineering efforts in the form of medical devices such as endoscopic cameras, refined instruments, and robotic manipulators that enable modern surgeons to intervene in the human body in ways that were previously not thought possible [3, 4, 6]. Imaging provides an alternative to impact surgical accuracy; the application of minimally invasive procedures is strengthened by the ability to map areas of disease in the context of healthy anatomy (so-called surgical roadmaps) using non-invasive preoperative imaging modalities such as MRI or PET/CT [7‐9]. Unfortunately, intraoperative detection of preoperatively identified lesions/structures can be challenging. For instance, increased distancing between the surgeon and the patient limits the surgeon’s sensory experience in the form of palpation (e.g. when a surgical robot is used). This shortcoming can at least in part be compensated through the use of interventional molecular imaging. To date, this imaging sub-discipline has predominantly focused on intraoperative detection of cancerous lesions using either radio- or fluorescence guidance [10, 11]. Here the main applications have included complex anatomies such as the head-and-neck or pelvic area where image guidance is exploited for both detection of nodal metastases and primary tumour margins [12‐16]. However, in these same anatomies accidental surgical damage to nerves can yield debilitating side effects such as loss of sensory feeling or speech, incontinence and/or erectile dysfunction. This occurrence of surgically induced nerve damage is not uncommon: despite the fact that more than 70% of prostate cancer patients receive nerve-sparing surgery, it is accepted that 30% of patients suffer from loss of erectile function at 1 year post-surgery [17]. Here, it should be noted that the extent of the damage may be hard to predict [18]. In addition, 10–15% of patients suffer from urinary incontinence [19]. In head and neck cancer patients, nerve anatomies are complex [20] and recurrent laryngeal nerve injury and mandibular nerve injury is seen in 14% of patients undergoing, respectively, thyroid surgery or neck dissection [21, 22]. Permanent paralysis is seen in 4–7% of patients [22].

Nerve-specific fluorescence imaging has been poised as a means to allow high-resolution nerve identification in real time [23]. For this application, a number of different targeting strategies have been evaluated, ranging from neuronal tracing to targeting intracellular expressed proteins in myelin sheets such as myelin basic protein [23‐27]. In some cases, the nerve-specific target is unknown [27‐30]. When pursuing conventional receptor-targeted molecular imaging strategies, extracellularly expressed targets are generally sought after. In that sense, myelin protein zero (MPZ, or P0), a 124-amino-acid-residues-large homotypic protein that makes up 80% of the protein content in peripheral myelin (Fig. 1A, [31]), and that is located on the outer membrane of the Schwann cells that form the myelin sheath, would most certainly be a target that is worth exploring. Uniquely, P0 is specific for the peripheral nervous system (PNS) and is not expressed in the central nervous system (CNS). In the CNS myelin sheath formation is facilitated by the adhesive properties of myelin proteolipid protein (PLP; [32]).

Building on the truncation of the extracellular portion of P0 (Fig. 1B) that was previously proposed by Makowska et al. [33] and the homotypic binding properties of P0 (e.g. intrinsic binding between P0 and P0), fluorescently labelled nerve-specific synthetic P0-derived peptides were extrapolated from the crystal structure of P0 (Fig. 1B; peptides in blue). After initial solubility, (photo)physical and in vitro evaluation (affinity and microscopic localization) a lead compound that showed the most ideal properties was selected. This fluorescent tracer was further scrutinized in three-dimensional (3D) dorsal root ganglion (DRG) cell cultures, on ex vivo nerve specimens and in vivo in mice (macroscopic localization, nerve/myelin-specificity, biodistribution). In vivo nerve visualization in a porcine model was performed as a proof of principle for real-time nerve visualization in a robot-assisted surgery setting.

By using the P0-derived synthetic peptide Cy5-P0_101–125, we were able to explore the homotypic P0 protein as molecular target that is widely expressed in myelin in the PNS. Specific binding was demonstrated in vitro, in 3D DRG nerve cultures, ex vivo, as well as in vivo.

Truncation of the homotypical P0-protein into peptides yielded Cy5-P0_101–125 as lead compound. In vitro and in vivo studies indicate that this compound is able to provide nerve-specific staining with nanomolar affinity, which is in the same range as reported affinities for other fluorescently labelled targeted tracers [36, 47, 48]. Specificity of Cy5-P0_101–125 for P0 was shown both in vitro (Fig. 2 and Additional file 1: Figure SI3) and in vivo (Fig. 4). An approximate 90% decrease in fluorescence intensity after pre-incubation with a P0-specific antibody was seen (Fig. 2CIII), while no staining was observed for the non-P0-specific control (Cy5-NP41) and the free dye (Cy5-Maleimide). As the Cy5-Maleimide dye variant was used for functionalization of both P0_101–125 and NP41, these results also exclude a targeting effect of the dye itself. This was corroborated by a markedly different staining pattern using free dye alone [30]. An additional beneficial factor for Cy5-P0_101–125 is that the production is scalable and can be done at reasonably low cost. Moreover, the peptide benefits from the superior pharmacokinetics that have been claimed for peptides over proteins such as antibodies [49, 50].

In line with P0 expression, biodistribution studies performed with the radiolabelled analogue [¹¹¹In]In-DTPA-P0_101–125 helped rule out accumulation of P0_101–125 (MW = 3024 Da) in the CNS. Obviously, the chance of toxic side effects is also impacted by dosing and uptake in non-target organs. In nuclear medicine, disease-specific tracers are therefore applied using a micro-dosing regimen (< 100 µg/patient). Although the ability of using fluorescence at micro-dosing levels looks promising [51], this topic remains a subject of debate [52], and many studies still use high dosing regimens to realize in vivo functionalization of molecular targets [16, 53‐55]. Local tracer deposition (an image-guided surgery concept that has proven valuable in e.g. lymphatic mapping and during occult lesion localization) limits dosing to 100 µg/patient [12, 51]. Recent experimental studies underscore that local deposition may also be valid when targeting peripheral nerves within a certain surgical anatomy [24, 37]. Substantiated by previous reports [28] Fig. 4 AIII and B indicate that local tracer administration is feasible for nerve imaging applications.

While most image-guided surgery studies promote the use of near-infrared (NIR) Cy7 analogues [53, 54, 56], far-red Cy5-labels are also increasingly being applied in clinical trials [39, 57‐59]. In fact, fluorescence-guided surgery trials have been reported for the full fluorescent light spectrum [60]. Uniquely, in a head-to-head comparison, Cy5 analogues even were shown to outperform Cy7-analogies in terms of signal intensity [42] and impact on tracer kinetics [36]. Moreover, Cy5 analogues, both a free dye and conjugated to a peptide, have shown to have a high fluorescence brightness in the presence of human serum albumin (Additional file 1: Table SI2; [36, 41, 61]). These factors combined with the fact that most groups are creating tumour-receptor-targeted tracers using NIR Cy7 analogues support the future implementation of multi-wavelength imaging applications, a concept that is gaining traction in the clinic [60].

Although intraneural injection provides a perfect proof of principle in both mice and pigs (Figs. 4AII, C and Additional file 1: Figure SI6), administration into a blood vessel near the target organ, such as the femoral vein (Fig. 4A), still provides further optimization before it can be used in a human setting. Hence, the exact route for local tracer administration in large animals, minimization of the dose and the specificity of targeting remain the subject of ongoing studies. Another potential limitation of the approach presented is that myelinization of nerves in the PNS may vary, where sensory nerves are highly myelinated (and conductive), the level of myelination decreases substantially in autonomous nerves. Despite the fact that myelin is one of the most widely explored targets for nerve imaging, it is therefore not clear if myelin-specific tracers will help address all surgical nerve imaging demands.

By truncating the P0 protein, we have been able to successfully create a nerve-specific fluorescent tracer that is able to specifically stain P0/myelin expression in the PNS.