Primary cerebellar slice cultures (stages p6–p30) were prepared as described previously (Meller et al., 2005; Wessel et al., 2014). All procedures were conducted under established standards of the German federal state of Northrhine Westphalia, in accordance with the European Communities Council Directive 2010/63/EU on the protection of animals used for scientific purposes.

The studies have been performed under the terms of the German animal protection law. Wistar rat pups at postnatal day 1 and day 9 (p1, p9) were used to obtain primary cerebellar slice cultures according to the roller tube technique (Wessel et al., 2014). Rat pups were decapitated, followed by dissection of the brain. Cerebella were isolated from the forebrain and brainstem, and the pia mater and superficial blood vessels were removed under a binocular microscope. Using the McIlwain tissue chopper, cerebella were cut with a razor blade into parasagittal slices with a thickness of 275 μm. After chopping, these slices were separated carefully and attached to collagen-coated (C7661; Sigma-Aldrich) glass cover slips (32 mm; Kindler, Freiburg, Germany) with 10 μl plasma (P3266; Sigma-Aldrich) and 10 μl thrombin (605157; Calbiochem). After coagulation, each cover slip was placed into a roller tube with an additional 1.3 ml nutrient medium. The medium contained 25% heat-inactivated fetal horse serum (S9135; Biochrom), 15% Hanks-Medium (H1641; Sigma-Aldrich), 6.5 mg/ml glucose, basal medium eagle (BME, B1522; Sigma-Aldrich), 0.01g/ml penicillin (P3032; Sigma-Aldrich), 0.01g/ml L-glutamine (G7513; Sigma-Aldrich), and 25 ng/ml nerve growth factor (NGF-7S, N0513; Sigma-Aldrich). Finally, these slice cultures were placed in a roller-drum incubator at a temperature of 37°C, which rotated slowly at the speed of four rounds per minute.

After 6 days in vitro (div) a mitosis inhibitor mixture (0.33 mM Uridine, 0.33 mM Cytosine-ß-D-arabinofuranoside hydrochloride, 0.33 mM 1-(2-Desoxy-ß-D-ribofuranosyl)-5-fluoruracil; U3003, C6645; Sigma-Aldrich; Serva 21555) was added to the slice cultures for 24 h to inhibit fibroblast proliferation, whereas after the 7th div nutrient medium was used, with a reduced heat-inactivated fetal horse serum concentration (15%). During the incubation period of up to 21 div, the nutrient medium was replaced twice per week.

Experiments were performed with three different cerebellar slice culture groups to examine the impact of VEGF on neonatal (group 1: p1 + 5 div), juvenile (group 2: p1 + 16 div) and mature PCs (group 3: p9 + 21 div). After these specific incubation times, the method of microinjection or immunohistochemistry were performed.

To be able to use the microinjection technique, slice cultures have to flatten for a minimal amount of 2 weeks in vitro. To investigate developing PCs we chose the stage p1 plus 16 div PCs as the juvenile experimental group. For the analysis of mature PCs we had to use p9 rats since it is known to achieve best possible results in regard to cell survival (Krah and Meller, 1999). To analyze the effects on dendritic maturation in these cultures an additional period of 21 div is meaningful, in which the PCs are able to develop their dendritic shape.

The impact of VEGF (V4512, Sigma Aldrich) on the morphology of PC dendrites was investigated in cerebellar slice cultures of p1 rats. In comparison to controls, slice cultures were incubated with VEGF (c = 0.1 μg/ml) nutrient medium as described previously (Jin et al., 2006), with a mixture of VEGF (c = 0.1 μg/ml) and axitinb (c = 0.01 mg/ml), or solely with axitinib (c = 0.01 mg/ml) for 48 h. Afterwards they were fixed with 4% paraformaldehyde (PFA) for 1 h and washed with phosphate-buffered saline (PBS).

Cryosection of rat cerebella were obtained from p6, p17 and p30 Wistar rats. After decapitation and dissection of the brain, p17 and p30 rats were fixed with 4% PFA. Cerebella were then isolated and post fixated with 4% PFA at 4°C overnight. After PBS washing, cerebella were transferred to 30% sucrose for 2 days. Afterwards, they were deep-frozen in isopentane at −50°C. Samples for cryosectioning were fixed with tissue freezing medium (Leica) and cut on a cryostat (Leica CM 3050 S; chamber and stage temperature of −18°C and −20°C respectively). Cryosections (40 μm thick) were applied on glass slides (J1800AMNZ, Menzel-Gläser, Braunschweig, Germany).

Surgical procedures and cryosections (15 μm) of rat cerebella were obtained from p6 and p30 Wistar rats under sterile RNase-free conditions as described in Pieczora et al. (2016). All solutions were prepared with DEPC-treated water. In situ hybridization was performed according to the instruction manual “FFPE in situ hybridization using double-labeled Fluorescein or DIG miRCURY LNA^TM microRNA Detection probes” (Exiqon). Briefly, the cryosections were incubated in 4% PFA at room temperature for 15 min. After washing 3× 3 min with PBS the slides were incubated with 20 μg/ml proteinase K (microRNA ISH Buffer Set #9000; Exiqon) for 10 min at 37°C. For hybridization the tissue was incubated with 80 nM of double DIG-LNA^TM mRNA probe (VEGFR-2 custom LNA mRNA detection probe, Exiqon) probe diluted in microRNA ISH Buffer (#9000, Exiqon). The tissue was incubated for 2 h at 54°C. Afterwards the tissue was washed once with 5× SSC for 5 min, respectively twice with 1× SSC and 0.2× SSC for 5 min at hybridization temperature and finally with once with 0.2× SSC for 5 min at room temperature. To visualize the probe the slides were incubated for 1 h with anti-Digoxigenin Alkaline Phosphatase, fab fragments (sheep, 1:800, 11093274910; Roche). Meanwhile a NBT-BCIP tablet (11697471001; Roche) was dissolved in DECP-water, adding Levamisol (31742; Fluka) at the end. The AP substrate was incubated for 90 min in the dark at room temperature. Finally the tissue was incubated with 150 μl Nuclear Fast Red (N3020; Sigma-Aldrich) per slide for 1 min, washed, dehydrated and fixed with mounting medium. The tissue was stored over night at room temperature and analyzed by light microscopy (Olympus BX 61) the subsequent day.

With aid of the microinjection technique yellow fluorescent protein-actin plasmids (pEYFP-actin; 0.195 μg/μl; custom made) were injected into single juvenile (group 2) and mature (group 3) PC cell nuclei of cerebellar slice cultures. Either sterile glass capillaries (diameter: 0.2–0.5 μm, Femtotips; Eppendorf, Germany) or capillaries (Hilgenberg, Nr. 1403512, Germany; borosilicate glass: 1.5 mm/0.2 mm; Puller: Sutter Instruments P97, USA) were filled with 2 μl of plasmid solution. Using an inverted microscope equipped with long-distance phase-contrast optics (Axiovert 35, Zeiss, Germany), PCs were imaged in slice cultures. During microinjection cultures were kept at 37°C on a heating stage. Application settings of the pressure injection devices (InjectMan NI2 and FemtoJet; Eppendorf) were set to inject with 80 hPa for 0.5s The constant pressure was defined as 60 hPa. After microinjection, nutrient medium was replaced carefully and slice cultures then further incubated for at least 24 h to receive a sufficient signal.

In our study we examined the effects of VEGF on neonatal, juvenile and mature PCs in cerebellar roller tube slices. The functionality of PCs in slice cultures was proven in former studies by analysis of axonal transport and functional synapses (Krah and Meller, 1999; Wessel et al., 2014). With aid of immunohistochemistry and CLSM our first aim was to analyze cerebellar roller tube slices in regard to their organotypic structure. Therefore, various incubation times of cerebellar slices of p1 and p9 pups were compared with cerebella cryosections of p6, p17 and p30 rats (Figures 1A–F, 2A–F).

At birth cerebellar structures are still immature, with a characteristic external granule cell layer (eGCL) above the molecular cell layer (MCL). In cryosections of p6 rat cerebella, and in cerebellar slice cultures of p1 rats cultured for 5 days, the fourth layer (eGCL) still exists (Figures 1A,D). At this early stage PCs are distinguished by multiple short perisomatic dendrites (Figures 2A,D). At this point of time granule cells (GCs) migrate from the eGCL through the MCL and Purkinje cell layer (PCL) to generate the inner granule cell layer (GCL), with its migration peak at p10-11 (Altman, 1972; Kunimoto and Suzuki, 1997), and its end after 16 days of maturation (Mancini et al., 2009). Thus, cryosections of p17 rat cerebella show the three-layered cerebellar cortex with a band of PCs, which is quite similar to the slice cultures of p1 rats incubated for 16 div (Figures 1B,E). At these juvenile stages, PCs contain one or two stem dendrites covered with a moderate number of dendritic spines (Figures 2B,E).

Matured cerebellar slices of p9 rats followed by 21 div also displayed the typical three-layered cerebellar cortex structure, which is comparable to the morphology of cryosections of p30 rats, characterized by a thick MCL with prominent dendritic arbors of the PCs. Cell bodies of PCs are arranged, as a clear band in the PCL, and densely packed GCs are located in the innermost GCL (Figures 1C,F). At this mature stage, the impressive dendritic arbor arises from the stem dendrites ramified with two or three main, and multiple smaller dendritic branches, with numerous dendritic spines attached to them (Figures 2C,F). Thus, the roller tube slices represent a functional and compact organotypic cerebellar cortex, which can be cultivated for several weeks. It is highly suitable for obtaining a deeper insight into the mechanism of VEGF-effects on PCs of the cerebellum when concerning their maturation.

To determine any impact of VEGF on PC morphology, which is likely to be mediated through an interaction with VEGFR-2, expression of this receptor was verified in neonatal, juvenile and mature PCs (p6, p17, p30) in the cerebellum. Therefore, the methods of immunohistochemistry, and additionally laser microdissection with qPCR were used.

An intense signal of the VEGFR-2 antibody was detectable in neonatal PC cell bodies and dendrites as well as in Bergmann glia fibers, along with a diffuse signal in the eGCL, MCL and inner GCL (Figures 3A–C). In the PCL and MCL of juvenile p17 rat cerebella, positive VEGFR-2 labeling degraded in PC dendrites and cell somata (Figures 3D–F). In p30 rat cerebella with almost mature PCs, positive VEGFR-2 signals were hardly observed in the PCL (Figures 3G–I).

In order to examine the expression of VEGFR-2 mRNA in PCs at the age p6 and p30 we performed in situ hybridization using a specific VEGFR-2 double DIG-LNA^TM mRNA probe (Exiqon). We detected a high expression of VEGFR-2 mRNA in very young PCs (Figure 4A). The PCL could be visualized very distinct by labeling the cell bodies and dendrites of neonatal PCs, located between the MCL and inner GCL. In p30 rat cerebella positive VEGFR-2 labeling was hardly detectable, (Figure 4B). Additionally we could reveal a positive signal in the eGCL and inner GCL in p6 rat cerebella as well as in the GCL at the stage p30. These findings demonstrate that VEGFR-2 expression extremely decreases during PC development and correspond with the results presented by immunohistochemistry.

Via the qPCR method, mRNA expression of KDR, that codes for VEGFR-2, was detected in microdissected PC-enriched samples of very young and mature stages (Figure 4C). Gene expression values for p6 were set to 100%. At p30 the expression level of KDR mRNA was decreased to only 28% (Figure 4D). Additionally we analyzed the expression of the VEGFR-1 encoding FLT1 gene in p6 and p30 rat cerebella, setting KDR gene expression values for p6–100%. In p6 and p30 the expression of FLT1 mRNA did not significantly change in young and mature PCs, the ratio amounted less than 5% (Supplementary Figure 1). Thus, laser microdissection followed by qPCR, as well as immunohistochemistry, underline that VEGFR-2 is expressed in PCs during their entire development, albeit with a significant decrease in mature stages.

To ensure that the VEGFR-2 expression is highly unique for PCs, we performed a qPCR with specific marker genes for PCs, Bergmann glia, microglia and GCs. The amount of contamination from cells other than PCs is extreme low (Figures 4E,F). At the age of p6, calbindin (a specific marker for PCs) amounts to 93.6 ± 1.2% and 90.3 ± 1.1% in p30 rat cerebella. Nevertheless, a slight contamination with neuronal and glial marker can be observed. In rat cerebella at stage p6 the relative amount of GFAP expressed mRNA is 5.3 ± 0.08%, 3.41 ± 0.05% in p30. Furthermore the contamination of gene markers for NeuN is 0.46 ± 0.01% in young rat cerebella and 0.83 ± 0.01% in p30. For TNFα the contamination is in the range of 0.1 ± 0.01% in p6 and 0.14 ± 0.01% in p30, for α6GABAa-R it is 0.54 ± 0.01% and VGlutI is 1 ± 0.01% at the age of p6. At the age of p30 the contamination for α6GABAa-R is 3.7 ± 0.02% and VGlut1 0.02 ± 0.01%.

The impact of VEGF in the nervous system is more and more becoming an investigative focus, so that VEGF-receptor expression in the peripheral nervous system (PNS) and CNS has been investigated in several studies. Many VEGF effects are mediated through VEGFR-2 (Zachary, 2005; Carmeliet and Ruiz de Almodovar, 2013) and this receptor expression was detected in neuronal tissues of the forebrain, in the hippocampus (Yang et al., 2003; Fournier et al., 2012), but also in the subventricular zone and olfactory bulb (Maurer et al., 2003), and in the temporal cortex (Boer et al., 2008). Licht et al. (2011) assumed the possibility of low levels of VEGFR-2 expression in non-endothelial cells, confirmed by the verification of VEGFR-2 expression in for example glia cells as astrocytes or Bergmann glia (Ruiz de Almodovar et al., 2010; Wuestefeld et al., 2012). Here, low levels of VEGFR-2 expression were detected in adulthood (Kremer et al., 1997; Matsuzaki et al., 2001; Cui et al., 2010). In line with this, a decrease of VEGF and VEGFR-2 levels in the cerebellum was underlined in a former study, which demonstrated that perikarya of PCs had a positive VEGFR-2 signal during gestation, but 13 months postnatal no VEGFR-2 signal was detectable anymore (Sentilhes et al., 2010). Another study showed VEGFR-2 expression in PCs of very young primary cerebellar slice cultures of mice by immunohistochemistry, and additionally they also detected VEGFR-2 expression in cryosections of mice cerebella with aid of quantitative RT-PCR (Cvetanovic et al., 2011).

In the present study, these results could be confirmed with the detection of VEGFR-2 expression in rat PCs by immunohistochemistry and in situ hybridization. Here a strong positive receptor signal was detected in the young developmental stages, with a significant reduction of VEGFR-2 protein. Using laser microdissection, followed by qPCR we demonstrated the reduced expression level of VEGFR-2 mRNA in mature stages and an unchangeable low expression level of VEGFR-1. Based on the hardly detectable VEGFR-1 amounts it might be assumed that the impact of VEGF on young cerebellar PCs is most likely mediated by VEGFR-2; no visible effects on mature PCs would go along with the results of IHC and qPCR. To ensure the observed effects ascribe to age-dependent expression levels of VEGFR-2 in PCs, the mRNA level of neuronal and glial markers were analyzed in microdissected PCs. The cerebellar cortical tissue is organized very densely. PCs are surrounded by several cell types such as Bergmann glia (Verkhratsky and Reichenbach, 2010). A slight contamination of microdissected PCs with these cells is inevitable. We were able to detect calbindin, a specific marker for PCs, as the vast majority component in our microdissected samples (Baimbridge et al., 1992). Furthermore, we could exclude contamination with immunoactive microglial cells as the mRNA expression level of TNFα was hardly detectable, indicated the absence of microglial cells (Deng et al., 2008; Kaur et al., 2014). Besides this, contamination with other cerebral cell types could be excluded. The expression level of NeuN, a marker for neuronal cells of the CNS (Mullen et al., 1992), was extreme low as well as the expression levels of α6GABAa-R and VGlut1 mRNA. These findings underlined former studies which disproved the presence of α6GABAa-R and VGlut1 within the PCL in rodents. (Fremeau et al., 2001; Wojcik et al., 2004; Hörtnagl et al., 2013). The samples of microdissected PCs showed a very low amount of contamination with cells surrounding PCs. Additionally, these results reveal another important fact: the extremely low levels of GFAP and NeuN mRNA in p6 and p30 PCs, indicating a negligible impact of VEGFR-3. VEGFR-3 is upregulated in glial cells and strongly associated with the expression of GFAP and NeuN (Shin et al., 2008; Ward and Cunningham, 2015). Recently it could be shown that GFAP was highly expressed at the age of p15 in rats, accompanied with high expression of VEGFR-3. In the neonatal period the expression of NeuN and VEGFR-3 showed a comparable expression pattern (Ward and Cunningham, 2015). In our study the expression of glial and neuronal markers in microdissected PCs could be disproved in both examined stages. The purity of the microdissected samples as well as the closely connected expression of glial and neuronal markers with VEGFR-3 expression allows us to discuss the higher levels of VEGFR-2 in p6 compared to p30 in a developmental context (Brown et al., 2013; Sugino et al., 2015; Ward and Cunningham, 2015).

Two to three days before birth, after PC-migration has stopped, PCs exhibit a bipolar fusiform shape that changes into a stellate shape with perisomatic predendrites. After a short retraction phase, main branches are established and the definitive polarity of these neurons is established (Armengol and Sotelo, 1991; Sotelo and Dusart, 2009). Our slice cultures of group 1 and group 2 were examined as representatives of an early developmental stage. Through the method of microinjection and immunohistochemistry it was possible to label PCs in organotypic cerebellar slice cultures and measure the total dendritic length, circumference of the dendritic arbor and soma size in different developmental stages. As it is impossible to microinject PCs in the cerebellar slices at the neonatal stage, here, solely the method of immunohistochemistry was used to investigate dendritogenesis. The present results demonstrate for the first time a significant increase in the total dendritic length of very young PCs after VEGF addition compared to control slice cultures.

Remarkably, in the prolonged maturation process of dendritogenesis and somatogenesis in PCs, VEGF also seems to be a growth factor for dendrites and cell somata. At this point in time, direct quantification of VEGF-induced changes or changes due to VEGF receptor inhibition on PC morphology was performed through the precise method of microinjection and immunohistochemical analysis. Significant additive effects in all growth parameters were measured under VEGF exposure, so that PC dendritogenesis is most likely enhanced in neonatal and juvenile stages. Therefore, together with NT3, BNDF, NGF, BMP7, Reelin or T₃ VEGF is another highly potential stimulating and enhancing factor of dendritogenesis (Miller and Kaplan, 2003; Boukhtouche et al., 2010; Castagna et al., 2014). The effects of direct exogenous VEGF application on developing PC dendritogenesis were not yet analyzed. However, Cvetanovic et al. (2011) examined whether replenishing VEGF improves the SCA1 phenotype in mice by using clinically affected animals, which overexpressed human VEGF. Thereby, an enhancement of Rotarod performance was measured and a generally better outcome of SCA1 mice was detected, which underlined promoting and protective effects of VEGF on PCs.

Earlier studies revealed that after the initial 30 days development period, PCs reached their mature dendritic shape with tertiary dendritic branches, with no further quantitative changes detected (Altman, 1972; Berry and Bradley, 1976; Sotelo and Dusart, 2009). Our data support this, as PCs of control slice cultures did not reveal any significant changes in dendritic length or cell soma size after 48 h of incubation, indicating that PC growth apparently finished in this mature phase. For therapeutic targets, our study examined PCs of this age to gain new insight into the impact of VEGF after the maturation process. Here, no significant increase on PC dendritogenesis and somatogenesis was detected after VEGF administration. Thus, our data indicate that VEGF effects on PC dendritogenesis and somatogenesis are age-dependent and are restricted to immature PCs, whereas mature PC dendrites and somata are not susceptible to VEGF anymore. This downregulation of VEGF sensitivity is in line with different other recent studies such as glia cell development in CNS and growth cone development in PNS (Foehring et al., 2012; Wuestefeld et al., 2012).

Axitinb is a potent and selective inhibitor of VEGF receptor tyrosine kinases 1–3 (Hu-Lowe et al., 2008). A current study could show that axitinib has a high specificity to VEGFR-2 (Giddabasappa et al., 2016). Through immunohistochemistry, qPCR and in situ hybridization we decisively proved VEGFR-2 expression in PCs. By blocking through axitinib treatment, it is very likely that VEGFR-2 and its corresponding ligand VEGF mainly mediate crucial effects in developing PCs. During this time, the growth of PC dendrites was strongly reduced under axitinib incubation. The qPCR results show a clear reduction in VEGFR-2 mRNA, and are in line with the lack of effects on mature PCs. In the present study, we observed a significant reduction in dendritic length after axitinib administration, so endogenous VEGF apparently plays an important role for proper dendritogenesis in neonatal PCs. Only slight reductions of dendritogenesis parameters between control and axitinib treated slice cultures were perceived in the juvenile stage. Thus, our presented data indicate for the first time that in this phase of development, endogenous VEGF probably plays a more subordinate role in the physiological maturation progress of the dendritic arbor. We cannot exclude the impact of other members of the VEGF familiy in the maturation of PCs. Focusing on the most active biological form of the VEGF family, VEGF-A, the presented results verify the prominent significance of VEGF/VEGFR-2 interaction during this process. The data are in accordance with other inhibition studies, which investigated the impact of VEGF inhibitors on the survival and dendritic outgrowth of neonatal PCs using primary cultures of p1 mice cerebella (Cvetanovic et al., 2011). These experiments point out a significant reduction in the longest PC dendrite, so that low levels of VEGF negatively influenced survival and dendritogenesis in PCs. In summary, our present study demonstrates that low VEGF levels compromise, yet additional VEGF accelerates PC dendritogenesis and somatogenesis.

VEGF and their inhibiting antibodies and drugs already have a high relevance in clinical practice, as it is used as a therapeutical drug for tumor treatment (Fernando and Hurwitz, 2004; Hutson et al., 2013), or in diabetic retinopathy (Das et al., 2015), age-related macular degeneration (Solomon et al., 2014) and intracranial injuries (Talwar and Srivastava, 2014). Storkebaum et al. (2005) demonstrated that VEGF administration improves motor performance and halts motor neuron degeneration in ALS rats. Besides this, in the case of Parkinson’s disease, VEGF is applied for its neuroprotective function (Cui et al., 2011). Altered PC miRNA expression plays an important role in SCA1 pathogenesis. Because of an increase in miR-150 levels, the expression of VEGF in PCs is down regulated, which modulates disease pathogenesis (Rodriguez-Lebron et al., 2013). SCA1 is primarily characterized by degeneration of the spinocerebellar tracts and the loss of cerebellar PCs (Durr, 2010; Seidel et al., 2012). One recent study analyzed SCA1 and demonstrated the supportive function of VEGF during development (Cvetanovic et al., 2011). Recently, it was of great interest to discover the VEGF that acts as a stimulating growth factor in dendritogenesis and spinogenesis during development, as well as in adulthood. Profound knowledge of VEGF could allow therapeutic use of VEGF for a variety of human diseases that affect the neuronal dendrites such as Down syndrome, Rett syndrome and fragile X syndrome (Kaufmann and Moser, 2000). In addition, for patients with focal ischemic infarcts and their resulting neuronal destruction and degeneration, VEGF could be a promising agent. Our data strongly indicate that VEGF administration is able to promote dendrite maturation during development, but unfortunately, there is a loss of VEGF susceptibility on matured neurons.