For 56 days following the surgery, two rats per cage were fed Ssniff food pellets (standard formula rat/mouse; D-59454 Soest, Germany), i.e., no post-operative drug application.

For 56 days, two rats per cage received Ssniff pellets containing 1,000 ppm nimodipine (donated by Bayer AG), supplied daily as fresh food because of the light sensitivity of nimodipine. The average daily food intake was 52 g/kg body weight/cage, i.e., 52 mg nimodipine/kg bw/day or about 10 mg nimdipine for a 200 g rat per day. In each of the 3 cages, the parallel weight gain of both rats during the 56 days treatment period did not indicate that one of the rats per cage had taken up more medicated food pellets than the other. This treatment led to a circadian plasma level of nimodipine ranging from 15 ng/ml at 0600 h to 90 ng/ml at 1500 h (light: midnight—noon, darkness: noon—midnight; Figure 2, unpublished data of W.F. Neiss in cooperation with Bayer AG).

As sham treatment in parallel to the FKT group, both PT and NT animals received 28 subcutaneous injections of the same volume of 0.9% saline during 56 days.

Six rats were subcutaneously injected in the posterior cervical region with FK506 (Tacrolimus, Prograf^®; donated by Fujisawa Co., Japan) with a mean dose of 0.5 mg FK506 per kg body weight per day. This dosage has been shown to function ideally in rats, acting as an accelerator of the peripheral nerve axon outgrowth, while simultaneously not inducing an immunosuppressant response (Fansa et al., 1999; Chunasuwankul et al., 2002; Udina et al., 2003; Yang et al., 2003). Each ampule contained 10 mg FK506 in 1 ml ethanol 95% plus vehicle (current packaging 5 mg/ml). This was diluted 1:10 with 9 ml 0.9% NaCl. Approximately 250 μl of this dilution was subcutaneously injected as a single dose every 48 h (cf. Jensen et al., 2005) at exactly 11:00 h (1 mg FK506/kg bw/48 h). Treatment was started on the day of conduit implantation and continued for the next 56 days (28 injections). Precise drug doses were adjusted weekly according to body weight changes.

In a pilot experiment we labeled 1,236 ± 72 motoneurons by application of DiI crystals to the proximal stump of the freshly cut buccal branch of the facial nerve (mean ± SD; n = 6 rats). In several studies, the intramuscular injection of a single tracer into the whisker pad has labeled 1,310 ± 83 motoneurons (weighted mean of means, Table 1). These neurons are localized exclusively in the lateral facial subnucleus (Figure 1A; see for HRP-labeling: Figure 1 in Angelov et al., 1996) and project into the whisker pad through the buccal branch of the facial nerve.

Using preoperative and post-operative double labeling on the unoperated control side, we counted less than 1,300 neurons in all three groups of this experiment (Table 2). In the same rats that had been injected with 100 μl 1% FG, 76 days prior to perfusion and with 100 μl 1% FB at exactly the same site 66 days later, i.e., 10 days prior to perfusion (cf. Table 2), we counted only 667 ± 148 FG-prelabeled neurons (mean ± SD; n = 17), but 1,147 ± 174 FB-post-labeled neurons. In all groups, we found considerably less FG-prelabeled than FB-post-labeled neurons (Table 2 and Figures 1B–D). These data sets are easily explained by the neuron cell bodies containing both FG and FB. The bright blue fluorescence of the latter outshines and masks the red fluorescence of the former, causing a false negative and had decreased the number of FG-labeled cells (Popratiloff et al., 2001). In light of this unavoidable methodological problem, the number of double-labeled neurons—that theoretically should reach 100% on the unoperated side—ranged from 93 to 97% (Table 2) of the number of prelabeled neurons.

As placebo treatment the control animals were fed standard food pellets, i.e., no treatment in addition to the nerve surgery. Injection of FG into the whisker pad 10 days prior to insertion of the facial nerve conduit and injection of FB 56 days after insertion had prelabeled 782 motoneurons confined within the lateral facial subnucleus and post-labeled 1,308 motoneurons (Table 2, PT). These neurons were spread throughout the lateral, dorsal and intermediate facial subnucleus (Figure 1B), but were not found in the medial facial subnucleus, the axons of which run through the posterior auricular nerve branch that had not been transected during implantation of the nerve conduit. Thus, the distribution of post-labeled neurons and the misdirection of reinnervation were the same after conduit implantation, as well as after direct facial nerve suture (Figure 2 in Angelov et al., 1996, 1999). Only 296 motoneurons were double-labeled, i.e., 38% of the original motoneurons of the whisker pad that had been prelabeled before surgery.

Immediately after operation, oral treatment was started with 1,000 ppm nimodipine in food pellets for 56 days (see Figure 2) as previously reported (Angelov et al., 1996, 1997; Guntinas-Lichius et al., 1996, 1997; Mattsson et al., 1999, 2001, 2005). We counted 691 FG-prelabeled and 1259 FB-post-labeled motoneurons (Table 2, NT), of which only 269 motoneurons were double-labeled, suggesting that 40% of the regenerated motor axons correctly reinnervated the original target. As in the placebo group PT, all FG-prelabeled neurons were exclusively localized in the lateral facial subnucleus indicating correct myotopic distribution, whereas the FB-post-labeled neurons were scattered throughout the entire facial nucleus (except for the medial facial subnucleus) with complete loss of myotopic organization pointing to aberrant reinnervation (Figure 1C).

FK506 was administered by subcutaneous injection in the dose of 1.0 mg/kg body weight per 48 h. On the experimental side, 679 FG-prelabeled and 1223 FB-post-labeled motoneurons were counted after systemic treatment with FK506. 261 motoneurons, i.e., 39% of the FG-prelabeled motoneurons were double-labeled also with FB (Table 2, FKT). The myotopic distribution of FG-prelabeled neurons in the lateral facial subnucleus and of FB-post-labeled neurons were scattered throughout the facial nucleus (Figure 1D) while their distribution was not distinguishable from that of the rats treated with nimodipine or placebo.

Statistical evaluation of the data summarized in Table 2 (two-way MANOVA) proved significance (P < 0.05) of the clear difference in the number of double-labeled neurons on the operated side (261–296, respectively, 38–40%) and the unoperated side (490–700, respectively, 93–97%; Figures 1A–D) in the three groups PT, NT and FKT. There was, however, no difference in the number of pre-, post- or double-labeled neurons between placebo, nimodipine or FK506 treated rats on the operated side (Figures 1B–D). Neither nimodipine nor FK506 improved the accuracy of reinnervation.

During exploratory behavior intact rats that had not underwent surgery or drug treatment (Table 3), showed 6 cycles of spontaneous protraction and retraction per second (frequency 6.0 ± 0.9 Hz; Figure 3A). At maximal protraction the rostral angle between the longitudinal axis of the body and the test whisker reached a minimum of 74° ± 14°. Higher values of this angle in experimental conditions suggest a reduced ability of the whisker pad muscle to pull the whisker forward. The amplitude of whisking was measured as the angle between the most forward and most backward position of the test whisker during both protraction and retraction. The standard amplitude was 46° ± 13°. The speed (460°/s) and acceleration (16,800°/s²) of the angular whisking movement were the maximal measurements that occurred during protraction, i.e., during the phase of active muscle contraction.

On the unoperated left side of placebo-, nimodipine- and FK506-treated rats, with an artificial nerve conduit on the right side (Table 3) the whisking frequency was similar to control conditions (6.1 Hz). However, the whisking amplitude in all groups was elevated to 111% of the normal value (51°), the angle at the maximal protraction was increased by 123% (91°), the maximal velocity by 266% (1,200°/s) and the maximal acceleration amounted to 245% (41,000°/s²) of control values. These elevations on the intact side most likely reflect compensation of the functional loss on the operated side.

On the operated right side of all experimental rats (Table 3), treated with either placebo, nimodipine or FK506 (Figures 3B–D), the whisking frequency ranged from 93 to 100% of the standard value (5.6–6.0 Hz). The whisking amplitude was reduced to 33–48% (15–22°) of the standard 46°. The whisking velocity was decreased to 48–58% (223–269°/s) of the preoperated value 460° and the angular acceleration had dropped to 35–42% (5,856–7,080°/s²) of the value of intact rats (16,800°/s²), but the angle at maximal protraction had increased to 153–177% (113–131°) indicating a severe impairment of active movement. Analysis of the data by means of ANOVA did not reveal any significant differences between the rats treated with placebo (PT), nimodipine (NT) or FK506 (FKT) after surgery (Figures 3B–D).

In comparison to placebo, neither nimodipine nor FK506 had beneficial effects on the accuracy of reinnervation and recovery of motor function after facial nerve repair. These findings support our data on retrograde labeling, where the drug application achieved similar results as observed in the control group (Table 2).

Administered after facial nerve reconstruction, has some other—presumably beneficial—effects, i.e., neuroprotective effects on motoneurons, astroglia, microglia, and Schwann cells (Guntinas-Lichius et al., 1996, 1997; Angelov et al., 1998; Mattsson et al., 1999; Jia et al., 2015; Sin et al., 2018; Leisz et al., 2019). Nimodipine treatment in rats, following crush of the recurrent laryngeal nerve (Hydman et al., 2007) or the facial nerve (Mattsson et al., 2001) has been shown to have positive, clinical observations on the time course, as well as amplitude of the compound muscle action potentials, and visual score of vibrissae movement. Treatment with 1,000 ppm nimodipine in food pellets is the standard post-operation dosage used by previous authors mentioned, as well as in our experiments.

In general, crushed nerves regenerate quite well (Lundborg, 2004). Our data on nimodipine therapy after facial nerve transection, do not support the favorable findings of Mattsson et al. (2001) working in a crush model. We measured amplitude of whisking by quantitative motion analysis (data in brackets: amplitude). The amplitude was decreased from 46° as in intact rats, to 15° (-67%) in both groups: facial nerve conduit plus placebo and facial nerve conduit plus nimodipine. This value was also reduced to 22° (-52%) in those rats with facial nerve conduit plus FK506 (Table 3 and Figure 3). A similar finding was reported by Franchi et al. (2006) in newborn rats after facial axotomy and repair, in which the reorganized vibrissal representation was reduced to the medialmost portion of the normal vibrissal representation. Hence, in our transection model the regeneration of motor function presented the same results, as without nimodipine therapy and vibrissae whisking was even significantly worse.

With regard to function, nimodipine was not beneficial for the rats in our experiment. Concerning morphology, nimodipine did not perform better. Following conduit reconstruction of the facial nerve, 38% of the neurons were double-labeled with placebo and 40% with nimodipine (Table 2). In our experiments, nimodipine also did not reduce hyperinnervation. Hyperinnervation was first described in the study of Angelov et al. (1993) on hypoglossal-facial anastomosis describing a higher reinnervation of distinct target muscles after nerve repair and regeneration, e.g., in the whisker pad, than on the contralateral normal side. After direct facial nerve suture, a hyperinnervation of 132.6% in young adults (Angelov et al., 1996) and 160.2% in 2-year-old rats (Streppel et al., 1998) were observed. With nimodipine treatment, the hyperinnervation was reduced to 113% in young adult rats (Angelov et al., 1996, 1997). In the present investigation, however, there was little to no detection of hyperinnervation in the artificial nerve conduit. Hyperinnervation was only 107% in both the placebo and nimodipine groups (Table 2). The data gathered can be interpreted as a lack of excessive sprouting in both groups, rather than a failure in nimodipine’s ability to reduce hyperinnervation.

Is an immunosuppressive drug that also dose-dependently accelerates neuronal regeneration (Gold et al., 1995, 1998, 1999; Doolabh and Mackinnon, 1999; Fansa et al., 1999; Gold, 1999; Wang and Gold, 1999; Jost et al., 2000; Lee et al., 2000; Navarro et al., 2001; Chunasuwankul et al., 2002; Sulaiman et al., 2002; Udina et al., 2003; Konofaos and Terzis, 2013; Shahraki et al., 2015; Labroo et al., 2016; Huang et al., 2017; Jo et al., 2019). FK506 can bind immunophilin neuronal receptors, causing activation of heat-shock proteins and, in turn, increased expression of growth-associated genes, i.e., growth-associated protein (GAP)-43, and activation of the mitogen-activated protein kinase pathways (Gold et al., 1995, 2004; Huang et al., 2017; Jo et al., 2019). Further, FK506 protects against neuronal cell death (Winter et al., 2000; Sosa et al., 2005; Konofaos and Terzis, 2013). The immunosuppression is based on inhibition of the calcium/calmodulin-dependent phosphoserine/phosphothreonine protein phosphatase calcineurin by the complex of FK506 and the 12-kDa FK506-binding protein (FKBP-12). As the immunosuppressive Cyclosporin A likewise inhibits calcineurin, but does not improve neural regeneration, FK506’s nerve regenerating property must involve a distinct, calcineurin-independent mechanism (Wang et al., 1997; Gold, 1999). Hence, non-immunosuppressive ligands to FK506-binding proteins were sought to elicit the neuroprotective/neurotrophic effect of FK506. One such compound, GPI-1046 has proved useless in a neuroprotection assay (Winter et al., 2000).

In our experiments, FK506 had no positive functional effects but showed side effects. Despite the subimmunosuppressive dose of 1 mg FK506/kg bw/48 h (Yang et al., 2003), one rat developed an abscess at the injection site in the neck region, then a peritoneal abscess/peritonitis and was lost during treatment. 56 days after implantation of an artificial nerve conduit, whisking was as reduced under FK506 treatment, as well as the placebo and nimodipine therapy (Table 3 and Figure 3). Regarding accuracy of reinnervation, 39% of the neurons that had originally innervated the whisker pad, were double-labeled following FK506 treatment, while only 40% of neurons were double-labeled after nimodipine and 38% following placebo treatment (Table 2). FK506-treated rats only differed from those of the placebo and nimodipine group in that they showed a much higher hyperinnervation (122 vs. 107%, Table 2).

The increase of hyperinnervation, i.e., an increase of axonal branching that physiologically occurs during nerve regeneration (Shawe, 1955; Brushart et al., 1998; Al-Majed et al., 2000; Jiang et al., 2007) is probably due to a neurotrophic effect. FK506 has been reported to raise the GAP-43 expression in axon sprouts of dorsal root ganglion cells (Gold et al., 1998), and to stimulate the proliferation of Schwann cells in vitro (Fansa et al., 1999). In rodents, FK506 was shown to increase the number of axons in the regenerating nerve (Chunasuwankul et al., 2002; Sobol et al., 2003; Yang et al., 2003; Brenner et al., 2004; Huang et al., 2017) while others did not observe such results (Fansa et al., 1999; Navarro et al., 2001; Udina et al., 2003). Neuroregenerative properties of FK506 are in line with hyperinnervation, however, it certainly does not improve the recovery of motor function. Congruent with our unsuccessful result, it has been reported that ultimately no functional recovery was achieved in rats after experimental nerve injury and treatment with FK506 (Doolabh and Mackinnon, 1999; Fansa et al., 1999). Except for the study of Brenner et al. (2004), walking track analysis after sciatic or tibial nerve transection and repair did not reveal noticeable differences between FK506- and placebo-treated rodents at the end of the observation periods (Jost et al., 2000; Navarro et al., 2001; Sobol et al., 2003). Brenner et al. (2008) have reported that benefits of FK506 in sciatic nerve regeneration as observed in comparison to placebo at 40 days post-operation had disappeared at 70 days post-operation. We have used 56 days regeneration time as endpoint of this study, as previous studies up to 112 or 224 days post-suture (Angelov et al., 1993; Streppel et al., 1998) have shown that at 56 days after facial nerve repair the regeneration of motoneurons is almost complete in rats, and obviously only then final function can be judged. In summary, our data suggest that acceleration of regeneration does not necessarily entail improvement of functional recovery.